You may also like

- Biomimetic inner helicoidal microfluidics

Inertial capillarity with enhanced capillary rise for step liquid
lifting mimicking transpiration
Zhaolong Wang, Yinfeng Li, Ziheng Zhan
To cite this article: D. Quéré 1997 EPL 39 533 et al.

- Study on the influence of composition

parameters of magnetorheological fluid on
its vibration transmission characteristics
Zebang Sun, Shaogang Liu, Dan Zhao et
View the article online for updates and enhancements. al.

- Magnetoelastic-Sensor Integrated
Microfluidic Chip for the Measurement of
Blood Plasma Viscosity
Ping Chen, Qiushi Jiang, Shin Horikawa et
al.

This content was downloaded from IP address 203.110.242.40 on 10/01/2025 at 10:31

EUROPHYSICS LETTERS 1 September 1997
Europhys. Lett., 39 (5), pp. 533-538 (1997)

Inertial capillarity

D. Quéré
Laboratoire de Physique de la Matière Condensée, URA 792 du CNRS
Collège de France, 75231 Paris Cedex 05, France

(received 28 April 1997; accepted 14 July 1997)

PACS. 68.10−m– Fluid surfaces and fluid-fluid interfaces.

PACS. 68.15+e – Liquid thin films.

Abstract. – We present new results on an old problem (capillary rise). A small tube is put in
contact with a liquid of low viscosity. It is found that 1) the early stage of the rise is described
by a linear law for the position of the meniscus vs. time; 2) oscillations around the equilibrium
occur if the liquid viscosity is low enough. A simple analytical model allows us to understand
these behaviours.

Usual capillary rise. – If a small tube (radius r smaller than the capillary length) is put
into contact with a bath of a wetting liquid (of surface tension γ, density ρ and viscosity η),
liquid rises in the tube up to a height which is generally of order 1 cm for a tube of inner
diameter 1 mm. This venerable effect was experimentally studied by Newton (1704) and Jurin
(1712) and explained by Laplace (1805). The final static height z0 is obtained by balancing
the capillary force (F = 2πrγ) with the weight of the liquid column (M g = πr2 z0 ρg):
2γ
z0 = , (1)
ρgr
where it is supposed that the liquid perfectly wets the inner of the tube. If it were not the
case, the capillary pressure and thus the final height would be multiplied by cos θ, where θ is
the contact angle between the liquid and the tube.
The dynamics of the rise was described in 1921 by Washburn [1]. Since it concerns a confined
medium, it is generally appropriate to balance the viscous force with the capillary force minus
the weight, which writes
◦
8πrηz z = F − M g , (2)
◦
where z is the position of the meniscus in the tube and z its derivative vs. time. Equation (2)
can be easily solved, but two asymptotic regimes may be stressed: i) at short time, the mass of
c Les Editions de Physique
534 EUROPHYSICS LETTERS

Fig. 1. – Height of the meniscus (in millimetres) as a function of the square root of time (t in seconds),
for a vertical glass tube of radius r = 421 mm put into contact at t = 0 with an infinite bath of viscous
silicone oil (γ = 21.1 mN/m, ρ = 980 kg/m3 and η = 0.5 Pa s). The squares are experimental data
obtained by filming the rise and the line is the analytical expression obtained by integrating eq. (2)
once. Jurin’s height is 10.4 mm.

the liquid column is negligible, so that z increases as the square root of t (t = 0 is the moment
when the tube touches the bath); ii) when approaching the final height, z can be taken as a
constant equal to z0 in the left member of eq. (2), which gives an exponential relaxation of
time constant τη = 16γη/ρ2 g 2 r3 ). τ is 25 s for a tube of radius 400 mm wetted by a silicone
oil of viscosity 0.5 Pa s and increases sharply when decreasing the tube radius.
The Washburn solution fits fairly well the experimental data, as seen in fig. 1, where
it is drawn and compared with results obtained by filming the meniscus for a glass tube
(r = 421 mm) put into contact with a silicone oil (γ = 21.1 mN/m, ρ = 980 kg/m3 and
η = 0.5 Pa s). The height of the meniscus is plotted in millimetres as a function of the square
root√of time (expressed in seconds). It can be seen in particular that, at short time, z increases
as t.
Close to the moment when the tube touches the liquid, Washburn’s law leads to an unphys-
ical infinite velocity of imbibition, as noticed for example by Bosanquet [2]. Of course, inertia
must be considered at that particular moment, and to stress its effect, a similar experiment
was performed with ethanol, a liquid of much lower viscosity, while keeping the density and
the surface tension roughly constant.

Early stage of the rise: the linear regime

O b s e r v a t i o n s. – Pictures were taken with a high-speed video-camera (1 image each
5 ms), for a glass tube (r = 689 mm) put into contact with a bath of ethanol (γ = 21.6 mN/m,
ρ = 780 kg/m3 and η = 1.17 mPa s). The results are drawn in fig. 2, where the height of the
meniscus is plotted as a function of time expressed in milliseconds.
√
Figure 2 looks like fig. 1, providing that the abscissa is t instead of t: the early stage of
the rise now follows a linear law in time. Then, because of gravity and viscosity, the height
of the meniscus relaxes towards the Jurin value (the theoretical height calculated from eq. (1)
is 8.1 mm, finally reached after about 1 s), which shows that the liquid completely wets the
tube.
d. quere: inertial capillarity 535

Fig. 2. – Height of the meniscus (in millimetres) as a function of time (in milliseconds), for a vertical
glass tube of radius r = 689 mm put into contact at t = 0 with an infinite bath of ethanol (γ =
21.6 mN/m, ρ = 780 kg/m3 and η = 1.17 mPa s). The data are obtained by filming the rise with a
high-speed camera. Jurin’s height is 8.1 mm.

A n a s y m p t o t i c m o d e l. – Ignoring the influence of viscosity close to the contact

moment and taking into account inertia modify eq. (2). By noting V the meniscus velocity
◦
(V = z), it becomes

d(M V )
=F, (3)
dt
where gravity is also neglected, since we are interested in the early stage of the rise. Equa-
tion (3) has a solution of constant velocity c, which writes
 2γ 1/2
c= . (4)
ρr
This velocity has the typical form resulting from a balance between inertia and capillary forces:
it is, for example, the bursting velocity of a soap bubble of thickness r [3]. Numerically, for
typical values of the different parameters, c lies between 20 cm/s and 2 m/s when r decreases
from 1 mm to 10 mm.

D i s c u s s i o n. – At first sight, experimental data are found to agree with this simple model:
at short time, fig. 2 indeed shows that the height of the liquid column increases linearly with
time. But a question remains: the velocity deduced from fig. 2 is c = 17 ± 1 cm/s, to be
compared with the value calculated from eq. (4) for the parameters corresponding to the
experiment. Then, we find: c = 28 cm/s.
We propose as a main cause for this quantitative disagreement the existence of a dynamic
contact angle at the liquid front. A contact line advancing on a dry solid makes with this
solid a contact angle which depends on the line velocity, because of the viscous dissipation in
the liquid wedge [3], [4]. Note that eq. (3) ignores the influence of viscosity: at short time,
the viscous boundary layer has not yet invaded the whole tube, but as it develops from the
point where liquid and solid meet, i.e. the contact line, viscosity must be considered at that
place. Hoffman studied extensively the value of the dynamic contact angle as a function of
the capillary number (Ca = ηV /γ), for a wetting liquid inside a tube [5].
536 EUROPHYSICS LETTERS

In our experiment, the capillary number is a constant in the linear regime, of the order
10−2 , which implies a dynamic contact angle θ of order 55◦ [5]. Thus, the capillary force is less
efficient than expected when considering a zero contact angle:√F in eq. (3) must be replaced
by F cos θ, so that the velocity in eq. (4) is reduced by a factor cos θ . In our experiment, this
factor is of order 0.75, so that the expected value for the velocity is rather 21 cm/s, much closer
to the actual one. The role of the dynamic contact angle was finally confirmed by doing the
same experiment after prewetting the tube with ethanol: the rise in a wet tube was also found
to be linear at the beginning, but quicker than in the dry case: we measured c = 24 ± 1 cm/s,
closer to the expected value.
Other causes can explain the observation of a velocity of imbibition lower than expected,
among which: i) the role of the reservoir: some liquid is put in movement inside the reservoir,
for the alimentation of the liquid column. Actually, the region of space concerned by this flow
is rather small (it is of order the radius of the tube [6]), which allows us to neglect in first
approximation the role of the reservoir; ii) the design of the boundary between the reservoir
and the tube [7] in our geometry, which implies sharp edges at the point where the tube touches
the liquid, which provokes vortices at this place. By comparing the horizontal imbibition in a
tube and in a pipette of same diameter (in a pipette, the transition between the reservoir and
the tube is smooth), we found that the imbibition velocity was indeed increased of about 20%
in the latter case.
Let us stress finally the limits in time of the linear regime. The transition to the viscous
regime (Washburn equation) occurs when the linear regime crosses the parabolic solution: it
occurs at a time τ ∗ = ρr2 /4η, which can be physically understood as the time necessary for the
viscous boundary layer to diffuse on a length of order the radius of the tube. Note that even if
this time is short (of order 0.1 s in our experiment), the corresponding length is rather large,
because of the high speed of the liquid front in the inertial regime. Conversely, the inertial
regime needs some time to set up: the meniscus must develop inside the tube, which implies
a time τ0 of order r/c (where c is defined in eq. (4)). τ0 dimensionally writes (ρr3 /γ)1/2 ,
typically of order 1 ms (thus too short to be observed with our camera). In this regime of pure
acceleration (before τ0 ), the height should vary as t2 , which ensures continuity of the velocity
around t = 0.

Capillary oscillations
O b s e r v a t i o n s. – As stressed above, the inertial regime is short but implies macroscopic
lengths. Thus, an interesting question concerns the case where the time necessary to reach
Jurin’s height (eq. (1)) with the inertial celerity (eq. (4)) is much smaller than the time scale
τ ∗ associated with viscosity. Written for example on the liquid viscosity, the latter condition
provides: η  gρ3/2 r5/2 γ −1/2 , which practically implies a viscosity smaller than 1 mPas for a
tube of diameter 1 mm. So we performed a last experiment with the same tube, now put in
contact with a bath of ether (γ = 16.6 mN/m, ρ = 710 kg/m3 and η = 0.3 mPa s), the liquid
of lowest viscosity among the usual ones. Once again, pictures are taken with a high-speed
camera and the height of the meniscus is plotted as a function of time in fig. 3.
The new fact is the occurrence of oscillations. The oscillations are due to inertia, whose role
was emphasised by choosing a liquid of very low viscosity. Then, they are damped and Jurin’s
height is reached (its theoretical value calculated from eq. (1) is 7.1 mm). Viscous damping
occurs after a time τ ∗ (expressed above), which is 250 ms for this experiment, in agreement
with the data. We now try to analyse the special shape of these oscillations (in particular the
first one).

A n i d e a l m o d e l. – We still ignore the influence of viscosity, since we focus on the first

stages of the rise (times shorter than τ ∗ ). Equation (3) is used again, but we have now to
d. quere: inertial capillarity 537

Fig. 3. – Same experiment as in fig. 2, except that the bath is now ether, comparable with ethanol for
its tension and density (γ = 16.6 mN/m and ρ = 710 kg/m3 ) but of lower viscosity (η = 0.3 mPa s).
Jurin’s height is 7.1 mm.

introduce the weight of the column, since it rises up to Jurin’s height (and even above). It
becomes
d(M V )
= F − Mg . (5)
dt
If z is scaled by z0 and t by τ = (z0 /g)1/2 (τ = 27 ms in the experiment), eq. (5) is reduced to

1 d2 z 2
=1−z. (6)
2 dt2
Equation (6) can be analytically integrated twice, provided that the solution passes by zero
(which is the case in our experiments). Then we find a solution of unexpected simplicity,
 t
z(t) = t 1 − . (7)
6
The first oscillation is a parabola, of maximum height 1.5z0 (50% above Jurin’s height), and
duration 6τ . Then, if there were no viscosity, a similar parabola would be followed, and so on:
if the liquid were inviscid, the static height would not be reached [8].

D i s c u s s i o n. – In spite of its simplicity, eq. (7) gives a good description of the first part
of the rise, in fig. 3. The beginning of the rise is linear, as shown in eqs. (4) and (7), and
the measured velocity is slightly smaller than the calculated one, as in the second section: we
measure: c = 23 cm/s and calculate c = 26 cm/s. Then, a parabolic curve is followed during
slightly more than one half-oscillation. The maximum of the height, reached at t = 84 ms (in
good agreement with t = 3τ predicted in eq. (7), which is 81 ms), is zmax = 1.3z0 , clearly
higher than Jurin’s height. The maximum is logically lower than expected (eq. (7) predicts
zmax = 1.5z0 ) since the starting velocity c is smaller: c is related to the driving force F by
eq. (4) (c ∼ F 1/2 ). Thus if c is smaller than expected of a quantity ∆c, the shift in the
maximum height is ∆z = 2z∆c/c. The whole behaviour (including the effect of viscosity)
observed in fig. 3 cannot be expressed analytically, but it can be solved numerically: the
solution is found to fit the data fairly well [8].
538 EUROPHYSICS LETTERS

Conclusion. – We have found two different effects due to inertia in the classical problem of
capillary rise:
1) At short time, the height increases linearly with time: the rising velocity results from
the superposition of a global inertial behaviour (an increasing mass of liquid is entrained) and
a local effect of viscosity (forcing the liquid close to the contact line to make a dynamic angle
with the solid, and thus reducing the driving capillary force).
2) At longer time, the height can pass Jurin’s height and oscillates. A criterion for such
a behaviour is proposed, and experimental data are reported. As long as viscosity can be
ignored, oscillations are found to be non-linear (they are parabolic), because of the variable
mass of the oscillator.
In both cases, asymptotic models were proposed and found to describe the data in a
satisfactory way, even if we restricted to the cases where analytical solutions could be derived.
Inertial effects in imbibition processes might be relevant in the case where the contact time
between a liquid and a porous medium is very small, as in printing. Further investigations are
under way, focusing on the case of liquids of ultra-low viscosity (liquid gases).

***
Experiments were done with the high-speed camera of Rhône-Poulenc: the author thanks
J. Y. Martin and L. Vovelle for their constant help. It is also a pleasure to thank P. G.
de Gennes, J. Y. Ollitrault and E. ´ Raphaël for precious discussions.

REFERENCES

[1] Washburn E. W., Phys. Rev., 17 (1921) 273.

[2] Bosanquet C. H., Philos. Mag., 45 (1923) 525.
[3] de Gennes P. G., Faraday Discuss., 104 (1996) 1.
[4] Thompson P. A. and Robbins M. O., Phys. Rev. Lett., 63 (1989) 766 and references therein.
[5] Hoffman R. L., J. Colloid Interface Sci., 50 (1975) 228.
[6] Bird R. B., Stewart W. E. and Lightfoot E. N., Transport Phenomena (John Wiley) 1960.
[7] Szekeley J., Neumann A. W. and Chuang Y. K., J. Colloid Interface Sci., 35 (1971) 273.
[8] Quéré D., Ollitrault J.Y. and Raphael E., ´ in preparation.