same order, σz ∝ 1/h. Thus β = 1/(3λp) and the stress increases at the same rate as the radius decreases: h = h0 exp(−t/(3λp)), σz = σ0 exp(t/(3λp)). (318) As shown in figure 98, (318) works reasonably well. The polymer relaxation time was determined by fitting small amplitude oscillatory shear data to a Zimm spectrum [546], and identifying λp with the longest time scale [550]. However to determine the so-called extensional viscosity νe = σz/
˙, the prefactor σ0 in (318) is needed, which according to (317) is determined by the tension T (t)in the thread. To this end one has to understand how a thread, as seen in figure 97, is matched to the drops between which it is strung. As is evident from figure 97, the transition region where the thinning thread is attached to a drop develops into an increasingly sharp corner. Recently [316] it has been shown experimentally and theoretically that the profile in the corner region is self-similar, with a typical size that is set by the thread radius. Using this self-similar structure, and within the confines of the onedimensional model equations (314)–(316), the tension was found to be T = 3γ h/ρ. The extensional viscosity νe = 3λpγ /(ρh) (319) can thus be determined from a measurement of the thread thickness alone. However, the slenderness assumptions underlying the one-dimensional description is no longer valid inside the drop, so the precise value of the prefactor in (319) is expected to be different if the full axisymmetric flow profile is used. Thus, a fully quantitative calculation of the extensional viscosity of a thinning polymeric thread remains an unsolved problem. A particularly simple description of a polymeric thread is achieved for observation over a time short compared with λp, formally derived from (315), (316) in the limit λp → ∞. In particular, one finds [551, 552] that for negligible initial deformation σz − σr = G(1/h2 − h4), where G = νp/λp is the ‘elasto-capillary’ number [533]. In this limit in which there is no polymer relaxation, the fluid behaves like an elastic solid. As a result, a stationary thread thickness is reached, corresponding to a balance of surface tension and elastic stresses [551]. A linear analysis of this stationary state shows that it is stable, with non-dispersive elastic waves running on it. Figure 99 illustrates the effect of flexible polymers on the dripping of a low-viscosity solvent (water), (a) showing the case of pure water for reference. The images in (b) and (c) are split to demonstrate the considerable power of the onedimensional model (314)–(316) in reproducing experimental data. In this comparison, only the polymer timescale λp was used as an adjustable parameter [542]. If polymers are added to the water, a thin thread forms between the main drop and the nozzle, analogous to figure 97. However, if the nozzle diameter is sufficiently large, a smaller ‘satellite’ bead is trapped in the middle of the thread. In particular, if the polymer timescale λp is greater than the Rayleigh time (2), polymer stretching sets in right after the initial linear disturbance growth on the fluid neck, resulting in a symmetric profile, cf figure 99(c). In this case, a uniform thread is formed. In the opposite case, polymer stretching is negligible at first, Figure 99. (a) A drop of water falling from a faucet, h0 = 3 mm. (b), (c): closeup of the pinch region, with 100 ppm of PEO solution added. (left: numerical simulations, right: experiment) (b): h0 = 3 mm, tc − t = 6, 2, 0, −3, −5 ms; (c): h0 = 0.4 mm tc − t = 1, 0 ms. Model parameters: ηp = 3.7 × 10−4 Pa s, λp = 1.2 × 10−2 s, b = 2.5 × 104, ηs = 1 × 10−3 Pa s, γ = 6 × 10−2 N m−1. and a highly asymmetric Euler solution (212) develops [521], see also figure 14. This normally leads to the formation of a satellite drop, but since polymer stretching eventually becomes important, this satellite drop is connected by thin threads cf figure 99(b). In [521], the thinning rate at the abrupt transition between the Euler solution and a thread was proposed as a measure for the polymer time scale. This rate was found to be similar to, yet significantly different from the subsequent thinning rate of the thread. Indeed, while the FENE-P model (314)–(316) is successful in describing the early stages of pinching shown in figure 99, it fails to satisfactorily describe the subsequent thinning of the thread. In [541] a strong dependence of β on the polymer concentration c was found, which persists to concentrations far below overlap concentration c∗, and which is attributed to interactions between polymers. Namely, polymers become highly deformed in the early stages of pinching, which greatly increases their interaction radius. However, two observations made in [542] point to even more fundamental problems of classical descriptions like FENE-P. Firstly, in [542] λp was fitted to match the transition from the Euler solution to a polymeric thread, but the same value does not predict the correct thinning rate of the thread. Thus more than one time scale seems to be necessary. Even more worryingly, the value of β was also found to depend on the radius of the capillary, so β cannot be an intrinsic property of the model. In [542] it is argued that for small solvent viscosities, polymer deformation is already large at the onset of thread formation, so finite size effects may play a significant role. Still another phenomenon occurs for somewhat more concentrated (above c∗) aqueous solutions, as first described 69