• Aucun résultat trouvé

Dynamics of a bouncing droplet onto a vertically vibrated interface

N/A
N/A
Protected

Academic year: 2021

Partager "Dynamics of a bouncing droplet onto a vertically vibrated interface"

Copied!
4
0
0

Texte intégral

(1)

T. Gilet, 1,

D. Terwagne, 1 N. Vandewalle, 1 and S. Dorbolo 1,

1

GRASP, Physi s Department B5a,

University of Liège, B-4000 Liège, Belgium

(Dated: April28, 2008)

Low vis osity (

<

100 St)sili on oil droplets are pla ed ona highvis osity (1000 St)oil bath that vibrates verti ally. The vis osity dieren e ensures that the droplet is more deformedthan

thebathinterfa e. Dropletsboun eperiodi allyonthebathwhenthea elerationofitssinusoidal

motion is larger thana threshold value. Thethreshold is minimumfor aparti ular frequen y of

ex itation:dropletandbathmotionsareinresonan e. Theboun ingdroplethasbeenmodelledby

onsideringthedeformationofthedropletandthelubri ationfor eexertedbytheairlayerbetween

thedropletandthebath. Thresholdvaluesarepredi tedandfoundtobeingoodagreementwith

ourmeasurements.

PACSnumbers: 68.15.+e,47.55.Dz,68.03.Cd

Keywords: Dropletphysi s,Vibratedinterfa e,Boun ing

Themanipulationofindividualdropletsbe omes

pro-gressively more important in mi rouidi s, as it is a

promising alternative to uid displa ement in

mi ro- hannels[1,2℄. Adropletmaybe onsideredasa

mi ro-s ale hemi al rea tor with a high e ien y [3℄ or as a

variablefo usopti allens[4℄. Whenadropletislaidon

aliquid bath, its oales en e with the bath oftentakes

a short time sin e the air layer separating the droplet

from the bath has to be drained out. This drainage

maybedelayedbyverti allyvibratingthebath [5℄: the

droplet boun es periodi ally without oales ing. With

this experiment, adropletmaybemanipulated without

any onta t with a solid element, whi h minimizes the

hemi al ontamination. Themanipulation ofboun ing

dropletsisstraightforward: dropletsmovespontaneously

by intera ting with thewavethey produ eon the bath

at ea h impa t[6℄. By using this wave, theyprobethe

surroundingsanddete tthepresen eofotherdropletsor

solidobsta les,theymaybeguided[7℄. Severaldroplets

on the same bath intera t together and experien e

or-bitalmotions[6℄, orform 2D rystallinelatti es[8℄.

Fi-nally,partial oales en eallowslowvis ositydropletsto

be emptied step-by-step. This emptying as ade stops

whendropletsareabletoboun eperiodi ally[9℄.

Couderetal.[5℄investigatedtheboun ingofan

homo-geneoussystem: thedropletandthebatharemadewith

the samevis ous oil(500 St). The verti al position of

thevibratedbathis givenby

A cos(2πf t)

,where

A

and

f

are the for ingamplitude and frequen yrespe tively. Theredu eda eleration

Γ

is denedas

Γ = 4π

2

Af

2

/g

.

Periodi boun ing is observed when

Γ

is higher than a riti alvalue

Γ

C

,thethresholdforboun ing. Couderet al. observedthat

Γ

C

1 ∼ f

2

, andexplained this

s al-ing by balan ing thegravity, theinertialfor es and the

lubri ationfor eexertedonthedropletbythesqueezed

Ele troni address: Tristan. Giletulg.a .be

URL:http://www.grasp.ulg.a .be

airlayer. Deformationsofboththedropletandthebath

werenot onsidered.

Boun ingme hanismsofliquidobje tshavebeen

stud-iedinawiderangeof ongurations: adropletboun ing

ona hydrophobi solid surfa e[10℄ or on an horizontal

wall immersed into an immis ible liquid [11℄, a bubble

boun ing on a water/air interfa e [12℄... As in elasti

solids, the deformation of those liquid obje ts is often

the key ingredient that ensures the boun ing property,

due to surfa e tension. Deformations may be

signi- antly damped by vis ous ee ts when the Ohnesorge

number

Oh = νpρ/σR

is largerthanunity, where

R

is thedropletradiusand

ρ

,

ν

and

σ

arethedensity,the vis- osityand thesurfa etensionof theliquid respe tively.

IntheexperimentofCouder,

Oh ∼ 4

,whi hmeansthat deformationsmaybenegle ted.

In this letter, we investigate theboun ing oflow

vis- osity droplets, for whi h the deformationis important

(

Oh ≪ 1

). The s aling law proposed by Couder [5℄ for thethresholda eleration is notvalidanymore : values

of

Γ

C

belowunityhavebeenobserved[7,9℄. Inorderto fo usonthedropletdeformation,weanalyzethe

boun -ing of low vis osity droplets (1.5 to 100 St) on a high

vis ositybath(1000 St) : thesystemisinhomogeneous

andthedeformationofthebathsurfa eismu h smaller

thanthedropletdeformation. First,wemeasure

Γ

C

for various

ν

and

f

with

R

xed. Then,amodelthat in or-poratesboththedropletdeformationandthelubri ation

for eisdeveloped.

A ontainer lled with 1000 Stsili onoil is xedon

averti allyvibratingele tromagneti shaker. Byusinga

syringe,dropletsofradius

R =

0.765mmwithvis osities of1.5, 10,50and 100 Stare pla edonthebath.

Mea-suredthresholda elerations

Γ

C

areshowninFig.1. The thresholdisdeterminedbyusingtwodierentproto ols.

First,thedropletis reatedwhenthea elerationis

su- ientlyhighforboun ingtoo ur.Thefor ingamplitude

isthenslowlyde reased(

f

xed)untilthedropletstops boun ingand oales eswiththebath(

inFig.1). Then, startingfromzero,

Γ

isin reased. Afterea hin rement,

(2)

adroplet is laid onthe surfa eof the bath. Whenthis

droplet boun es, the threshold is rea hed (

N

in Fig.1). An hysteresis, i.e. a dieren e in

Γ

C

dedu ed by both methods,isobservedfor

ν = 1.5

St(Fig.1(a)). Forhigh vis ositydroplets(Fig.1(d)),

Γ

C

> 1

,and

∂Γ

C

/∂f > 0

as predi tedin[5℄. Forlowerdropletvis osities,the

thresh-old may be lower than 1, and there is a minimum in

the

Γ

C

(f )

urve. Athigh frequen y, this urvestrongly in reaseswith

f

.

The following model is proposed in order to des ribe

the

Γ

C

dependan e on the for ingfrequen y

f

and the dropletvis osity

ν

. Theow is assumedto be axisym-metri andthemotionof thedropletmass enter(mass

M

) onned to a verti al axis. The droplet boun ing is modelled with two s alar ordinary dierential

equa-tionsdes ribingtheverti alposition

x

c

ofthemass en-terandtheverti aldeformation

η

ofthedroplet(Fig.2) respe tively. The bath deformation is negle ted.

Dur-ingitsight,thedropletexperien esanapparentgravity

M g(Γ cos 2πf t − 1)

in a frame moving with the bath. Moreover,thedropletisstressedbythesurroundingair,

resulting in a verti al for e

F

. The inuen e of air is negligibleonthedropletmovement,ex eptwhenthereis

athinairlayerbetweenthedropletandthebathsurfa e.

Then,

F

anbeestimatedbylubri ationtheory[13℄,and dependsonthethi kness

h

ofthelmanditsrateof de- rease

˙h

. Movementsinsidethedropletalsohavea signif-i antinuen e on theairlmdrainage. This latter an

bemodelledtoleadingorderbyaPoiseuilleowbetween

twoparallelplanarinterfa es. Thebottominterfa eisat

rest (the bath is stati ), while the upper moves with a

verti al velo ity equal to

˙h

and an horizontal velo ity proportional to

˙ηr/R

, where

r

is the radial ylindri al oordinate, and

R

theradius of theunstrained droplet. Therefore

F = c

1

µ

a

R

4



c

2

˙η

h

2

R

˙h

h

3



(1)

where

c

1

,

c

2

arepositive onstants,and

µ

a

isthedynami vis osityof theair. A ordingtothelubri ationtheory,

c

1

= 3π/2

. The parameter

c

2

represents the inuen e oftheowinside thedropletontheowin theairlm.

It annot beestimated by simplearguments. Newton's

se ondlawappliedto thedropletiswrittenas

M

d

2

x

c

dt

2

= M g(Γ cos 2πf t − 1) + F

(2)

For pra ti alpurposes,weuse

h = x

c

R − η

insteadof

x

c

. Theevolutionof

η

ispres ribedbyanenergybalan e intheframeofthemass enterofthedroplet:

d(K + E)

dt

= −P

d

P

f

(3)

where

K

is the kineti energy of the motion inside the droplet,

E

istheinterfa ialenergyand

P

d

isthevis ous dissipative power inside the droplet. The power

devel-oped by

F

, alled

P

f

, is supposed to be equalto

c

6

˙ηF

.

0

20

40

60

80

100

0

1

2

3

4

5

f [Hz]

Γ

C

(f,

ν

=1.5cSt)

(a)

0

20

40

60

80

100

0

1

2

3

4

5

f [Hz]

Γ

C

(f,

ν

=10cSt)

(b)

0

20

40

60

80

100

0

1

2

3

4

5

6

7

f [Hz]

Γ

C

(f,

ν

=50cSt)

( )

0

20

40

60

80

100

0

1

2

3

4

5

6

7

f [Hz]

Γ

C

(f,

ν

=100cSt)

(d)

FIG.1: A eleration threshold

Γ

C

for various vis osities of the droplet: (a)

ν = 1.5

St, (b)

ν = 10

St, ( )

ν = 50

St and(d)

ν = 100

St. For

ν = 1.5

St, thresholdsare dier-enta ordingtowhetherthea elerationisin reased (

N

)or de reased(

). Thesolidline orrespondstothemodel predi -tion(Eq.9), with oe ientsgiveninTable I. Thedash-dot

line is a t by the s aling of Couder (

Γ

C

− 1 ∼ f

2

). The

verti al dashed line enhan es the resonan e frequen y

ω

res

des ribedinthemodel. Error bars orrespondtothesize of

(3)

FIG. 2: Geometri al variables neededto model a boun ing

dropletof undeformedradius

R

:

x

c

is thedistan e between thedroplet enterofmassandthebath,

h

istheminimumair lmthi knessand

η

istheverti aldropletdeformationabout theaxisofsymmetry.

Law Mode2 Mode3 Fit

K = c

3

M ˙η

2

/2

c

3

= 3

/10

c

3

= 1

/7

c

3

= 0

.1

E = c

4

ση

2

/2

c

4

= 16

π/5 c

4

= 40

π/7 c

4

= 10

P

d

=

c

5

νM ˙η

2

/R

2

c

5

= 3

c

5

= 4

c

5

= 3

.3

TABLE I: Constitutive laws for the energy balan e of the

dropletdeformation. Se ond andthird olumnsare

theoret-i al oe ients

c

i

for modes 2 (spheroid)and 3, while the fourth olumn orrespondstothebesttofEq.(9)on

exper-imentaldata.

A onvenientwaytoestimate

K

,

E

and

P

d

asafun tion of

η

refers to the potential ow related to innitesimal apillarywavesatthesurfa eofadroplet[14℄(seeTable

I). The deformation

η

measured experimentally is less than 10% of the initial radius, whi h validates the

lin-ear approa h [15℄. Wesupposethat only themode2 is

ex ited by the boun ing sin ehigher modeshave mu h

higherresonan efrequen ies. Thewhole systemis

writ-tenindimensionlessformbyusing

R

asalengths aleand the apillary time

τ

σ

=

pM/σ

as a time s ale. More-over,Eq.(3)isrepla edby

c

6

timesEq.(2)plus

1/ ˙η

times Eq.(3),in ordertoremovethelubri ationterm.

¨

h + ¨

η = Bo(Γ cos ωt − 1) + c

1

4πνρ

a

Oh



c

2

h

η

˙

2

˙h

h

3



(c

3

+ c

6

η + c

5

Oh ˙η + c

4

η = c

6

Bo(Γ cos ωt − 1) − c

6

h

¨

(4) where

Bo =

M g

σR

is theBond numberand

ω = 2πf τ

σ

is theredu edfrequen y.

Terwagne etal. [16℄observedthedynami s oftheair

lm lo ated between the droplet and the bath using a

mono hromati light: on entri fringes of interferen e

appearwhentheairlmissqueezed. Whenthedroplet

boun es,the motionof thefringes is perfe tly periodi :

noattenuationorphasedrifttakepla eandtheboun ing

is stationary. Ontheother hand,thenumberof fringes

de reases when the droplet does not boun e: the lm

thins. Theperiodi ityofthefringesmotionsuggests

pe-riodi solutions from Eq.(4). Conditions for su h

solu-T = 2π/ω

. Under theassumption of periodi ity, many termsvanish,giving

(

R

T

0

ηdt =

c

6

c

4

BoT

R

T

0

˙

η

h

2

dt =

3c

1

c

2

νρ

µ

a

BoT

Oh

(5)

Termsontheright-handsidearealwaysstri tlypositive.

A ordingtotherstrelation,ame hanismofpotential

energystorage(here,thedropletdeformation)shouldbe

taken into a ount (

η 6= 0

). The droplet has to spend moretime in anoblate state (

η < 0

) thanin aprolate state(

η > 0

). A ordingtothese ondequation,internal movements in the liquid phase, relatedto the

deforma-tion rate, must have a signi ant inuen e on the lm

drainageandtheresultinglubri ationfor e. Moreover,a

signi antphaseshiftbetweentheminimumlm

thi k-ness and the maximum ompression must be observed.

Indeed,

R

T

0

˙ηdt = 0

, while

1/h

2

is stri tly positive and

vanisheswhenthelm thi kens. Tohaveapositiveleft

handside in these ond equation, weexpe tthelm to

be the thinnest when the droplet begins to re over its

spheri alshape (

˙η > 0

). All these required onditions show us that this model is minimal: if the model does

nottakeintoa ountallabovelisted onditions,its

pre-di tion fails and no periodi boun ingsolutions an be

found.

The a eleration threshold

Γ

C

required for periodi boun ingmaybeestimatedstartingfromEq.(4). When

Γ < Γ

C

,thedropletdoesnotboun e,theairlmremains thinand

¨

h ≪ Bo

. These ondequationinEq.(4)doesnot depend on

h

anymore. Thedropletbehavesas asimple for edos illator,i.e.

η = c

6

BoB(ω)Γ cos(ωt + φ) −

c

6

Bo

c

4

,

where

B(ω)

and

φ

aretriviallyobtained. Theresonan e frequen yrelatedtothisos illatoris givenby:

ω

res

2

=

c

4

c

3

+ c

6



1 −

c

4

c

3

+ c

6

Oh

2

2



(6)

Tond

h

withtherstequationofEq.(4),itis onvenient to dene the amplitude

H(t)

of the thi knessvariation as

h(t) = H(t)e

c

2

c

6

BoB(ω)Γ cos(ωt+φ)

. Cal ulationsyield

3c

1

µ

a

Oh

4πρνBo

˙

H

H

3

=





(c

4

c

3

ω

2

) cos(ωt + φ)

−c

5

Ohω sin(ωt + φ)



1



e

2c

2

c

6

BoBΓ cos(ωt+φ)

(7)

Byintegratingthisequationover

n

periods(

T = 2π/ω

), weobtain:

H

nT

=



H

0

−2

8πρνBo

3c

1

µ

a

Oh

CnT



−1/2

(8) where

 C = (c

4

c

3

ω

2

)BΓI

1

(2c

2

c

6

BoBΓ) − I

0

(2c

2

c

6

BoBΓ)

I

k

(x) =

1

π

R

π

0

e

(4)

When

C < 0

, the averaged lm thi kness

H

de reases with timeand the dropletnally oales es. Conversely,

when

C > 0

,

H

divergesand the solutionis not longer valid. Thedroplettakeso,

¨

h

annotbenegle ted any-morein Eq.(4) andboun ingo urs. Thethreshold

a - eleration for boun ing

Γ

C

an thus be dened as the valueof

Γ

su h that

C = 0

. Thisequation hasone pos-itive solution when

c

4

c

3

ω

2

> 0

, and no solution in

the other ase. There is a ut-o frequen y

ω

2

c

=

c

4

c

3

above whi h the model annot predi t boun ing (

C

is alwaysnegative). Thisfrequen y orrespondstothe

nat-ural resonan e of mode 2, when the droplet is dire tly

ex ited (i.e. not through the air lm dynami s). It is

always higherthan

ω

res

, related to the for ingthrough theairlmdynami s. Su h afrequen ywasalready

ob-served in [9℄. The urve

Γ

C

(ω)

tends asymptoti allyto a onstantvalue

> 1

when

ω → 0

. Moreover,when

Oh

is su ientlysmall, aminimum in

Γ

C

isobserved fora nite valueof

ω

, lowerthan

ω

res

sin e

∂(BΓ

C

)/∂ω > 0

when

ω < ω

c

. Therefore, nominimumisobservedwhen

ω

res

is omplex,i.e. when

Oh

2

> 2(c

3

+ c

6

)/c

4

.

Inorder to ompare themodelpredi tions to the

ex-perimental data shown in Fig.1(a) to 1(d), a single t

hasbeenmadeon oe ients

c

2

,

c

3

,

c

5

and

c

6

(

c

1

isnot presentin Eq.(9) and

c

4

isxed to

10

). Obtainedvalue

(c

2

, c

3

, c

5

, c

6

) ≃ (25, 0.1, 3.3, 1)

are similar to the values estimated theoreti ally (TableI). The omparison with

experimentsis a eptable, bothqualitativelyand

quan-titatively. In parti ular, the minima for low vis osities

and thedivergen e for high frequen iesare reprodu ed.

Quantitativedis repan ies maybe due to the fa t that

only mode 2 is onsidered in the modelling. The

reso-nan e o urs for a redu ed frequen y

ω

res

< 3

as long as

Oh . 0.47

. The ut-o redu ed frequen y for the mode2is

ω

c

10

. Foranoildropletwith

R =

0.765mm boun inginmode2,resonan eisobservedatamaximum

frequen yof51Hzwhenthevis osityislessthan32 St,

andthe ut-o frequen yofthis mode isabout165Hz.

Thisis onsistentwithourexperimentalobservations.

In on lusion, we have measureda eleration

thresh-oldsfor boun ing dropletson averti ally vibratedhigh

vis osity bath. The for ing frequen y and the droplet

vis ositywerevaried. Thereisa ut-ofrequen yabove

whi h the droplet annot use a mode 2deformation to

boun e. For low vis osity droplets, a minimum in the

Γ

C

(f )

urveisobserved,whi h orresponds tothe reso-nan eofthemode2. Inorder toexplainthese features,

a theoreti al model was developed, that in ludes both

thedeformation of the droplet and its inuen e on the

lmdrainage. Theseee tsarene essaryinorderto

ob-tainperiodi boun ingsolutionssu h as those observed

experimentally: themodel isthus minimal. Byvarying

fourofthesix onstitutive oe ientsofthemodel,itis

possibletottheexperimentaldata.

TG andSDthankFRIA/FNRS fornan ialsupport.

Ex hanges between laboratories have been nan ially

helped by the COST a tion P21. J.M.Aristo(MIT),

J.W.Bush(MIT),H.Caps(ULg),J.P.Le omte(Dow

Corn-ing)andA.Bourlioux(U.Montreal)area knowledgedfor

fruitfuldis ussions.

[1℄ H.A. Stone, A.D. Stroo k, and A. Ajdari, Annu. Rev.

FluidMe h.36 ,381(2004).

[2℄ T. Squires and S.R. Quake, Rev. Mod. Phys. 77 , 977

(2005).

[3℄ D.L. Chen, L. Li, S. Reyes, D.N. Adamson, and R.F.

Ismagilov,Langmuir23 ,2255 (2007).

[4℄ C.C. Cheng,C.A.Chang,andJ.A.Yeh,Opti sExpress

14 ,4101(2006).

[5℄ Y. Couder, E. Fort, C.H. Gautier, and A. Boudaoud,

Phys.Rev.Lett.94 ,177801(2005).

[6℄ Y. Couder,S.Protière, E.Fort, and A.Boudaoud,

Na-ture437 ,208(2005).

[7℄ S.Protiere,A.Boudaoud,andY.Couder,J.FluidMe h.

554 ,85(2006).

[8℄ S. Lieber, M. Hendershott, A. Pattanaporkratana, and

J.Ma lenna,Phys.Rev.E75 ,056308(2007).

76 ,035302(2007).

[10℄ D. Ri hard, C.Clanet,and D. Quéré, Nature417 ,811

(2002).

[11℄ D. Legendre, C. Daniel, and P. Guiraud, Phys. Fluids

17 ,097105(2005).

[12℄ M. Kranz,K.Lunkenheimer,and K.Malysa,Langmuir

19 ,6586 (2003).

[13℄ O.Reynolds,Philos.Trans.R.So .LondonSer.A177 ,

157(1886).

[14℄ L.LandauandE.Lif hitz,Courseontheoreti alphysi s,

6-Fluidme hani s (AddisonWesley,1959).

[15℄ E.Be ker,W.Hiller,andT.Kowalewski,J.FluidMe h.

231 ,189(1991).

[16℄ D.Terwagne,N.Vandewalle,andS.Dorbolo,Phys.Rev.

Figure

FIG. 1: Aeleration threshold Γ C for various visosities of the droplet: (a) ν = 1 . 5
FIG. 2: Geometrial variables needed to model a bouning

Références

Documents relatifs

The evolution of Gauss and mean curvatures is analyzed in detail in order to show how the breakup process from a liquid jet instability to the final droplet can be easily described

For the U.S.-Cuba policy case and portion of the thesis, I have conducted twelve interviews with leading activists and thinkers in the movement to further open U.S.-Cuba

Mauricette Fournier et Annick Stoehr-Monjou, « « Cartographie géo-littéraire et géo-historique de la mobilité aristocratique au V e siècle d’après la correspondance de

However, cruise ship traffic through the Canadian Arctic is typically between the months of July to late September or early October, where the environmental conditions are

We reviewed the susceptibilities to CMX of bacteria isolated from blood and stool specimens in a reference hospital in Abidjan, Ivory Coast, from 1994 to 1996.. The Centre

Notre contribution se propose d’alimenter la réflexion sur les fonctions et la place des manuels dans l’enseignement de la rédaction à travers la comparaison de

Currently, it is unknown whether only specific phospholipids are recruited to the surface of nucleated and/or nascent lipid droplets, whether different lipid droplets are nucleated

triacylglycerol-specific or cholesteryl-ester-specific intracellular lipid storage droplets. Wodlei, and H.H. von Grunberg, Phospholipid demixing and the birth of a lipid