HAL Id: jpa-00210808
https://hal.archives-ouvertes.fr/jpa-00210808
Submitted on 1 Jan 1988
HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.
Viscoelastic relaxation of insoluble monomolecular films
J.C. Earnshaw, R.C. Mcgivern, P.J. Winch
To cite this version:
J.C. Earnshaw, R.C. Mcgivern, P.J. Winch. Viscoelastic relaxation of insoluble monomolecular films.
Journal de Physique, 1988, 49 (7), pp.1271-1293. �10.1051/jphys:019880049070127100�. �jpa-00210808�
Viscoelastic relaxation of insoluble monomolecular films
J. C. Earnshaw, R. C. McGivern and P. J. Winch
The Department of Pure and Applied Physics, The Queen’s University of Belfast, Belfast BT7 1NN, Northern
Ireland
(Requ le 8 d6cembre 1987, révisé le 7 mars 1988, accepté le 11 mars 1988)
Résumé.
2014On étudie les monocouches du mono-oléate de glycérol à l’interface eau-air en utilisant les ondes
capillaires excitées thermiquement qu’on observe dans une grande plage de nombres d’onde par diffusion
quasi élastique de la lumière. Par une nouvelle procédure d’analyse des résultats on obtient ab initio 4 propriétés viscoélastiques de la surface : les modules élastiques de surface et les viscosités qui gouvement les cisaillements normaux à la monocouche (~ tension) ainsi que la dilatation dans le plan de la couche. Ces
mesures permettent la première comparaison rigoureuse des modules de tension et de dilatation avec leurs valeurs d’équilibre classiques. Plusieurs effets suggèrent que ces deux modules subissent des processus de relaxation différents : pour les modules élastiques on trouve des différences entre les valeurs dynamiques et statiques dans différents états de la monocouche, et pour les viscosités, on trouve des comportements différents suivant l’état de compression de la monocouche. Ces effets dépendent également de la fréquence.
Dans la monocouche complètement comprimée, le module de cisaillement transverse est caractérisé par une relaxation exponentielle, avec un temps ~ 9 03BCs. Ce temps de relaxation décroît exponentiellement lorsque la
monocouche est dilatée, et atteint 100 ns pour des aires par molécule de 60 Å2. On peut exclure rigoureusement la présence de processus plus lents. Le module de dilatation est généralement moins bien
déterminé que celui de cisaillement transverse ; toutefois, dans l’ état dilaté de la monocouche, les données démontrent l’existence d’une relaxation beaucoup plus lente, vers 03C4 ~ 290 03BCs. On discute brièvement les mécanismes moléculaires associés à ces relaxations.
Abstract.
2014Glycerol mono-oleate monolayers at the air-water interface have been investigated by quasi-
elastic light scattering from thermally excited capillary waves over a wide range of wave numbers. Using a relatively novel data analysis procedure four surface viscoelastic properties were deduced ab initio from the
light scattering data : surface elastic moduli and viscosities governing shear normal to the monolayer (~ tension) and dilation in the film plane. The tension and dilational modulus were compared with classical, equilibrium values in the first rigorous comparison of its kind. Various effects suggested that the two moduli
were affected by rather different relaxation processes : discrepancies between the light scattering and equilibrium values of the two elastic moduli occurred in different states of the monolayer, and the two surface
viscosities (both zero for the clean subphase) behaved very differently on monolayer compression. These
effects were observed to be frequency dependent. In the fully compressed monolayer state the transverse shear modulus was characterised by an exponential relaxation, of time scale ~ 9 03BCs. This relaxation time fell
exponentially on monolayer expansion, reaching 100 ns for molecular areas ~ 60 Å2. Slower processes than these were rigorously excluded. The dilational modulus was generally less well determined than that affecting
transverse shear. However in the expanded monolayer state, the data sufficed to demonstrate much slower relaxation, 03C4 ~ 290 03BCs. Possible molecular mechanisms are briefly discussed.
Classification Physics Abstracts
68.10
-62.30
-47.35
1. Introduction.
-The viscoelastic properties of a molecular film at a
liquid surface are of considerable interest, both intrinsically and for the light which they may shed upon the phase behaviour of these systems. They are macroscopic manifestations of the intermolecular interactions within the film, and perhaps between
the film and the adjoining fluid. Even for a surface which is isotropic within its equilibrium plane there exist, in principle, five separate independent surface
moduli [1], each of which may comprise elastic and
viscous components [2]. In practice this variety has
not been exploited in experimental studies of am- phiphilic monolayers. The commonest surface properties observed have been the surface tension
Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphys:019880049070127100
1272
and that surface viscosity relating to shear strain within the monolayer plane [3]. This paper concerns the determination from a single experimental obser-
vation of four different surface properties, compris- ing two viscoelastic moduli. These correspond to the
surface tension and to the dilational modulus of the
interface, together with the corresponding surface
viscosities.
The surface of any liquid is continually roughened by molecular agitation. The random surface can be Fourier decomposed into a complete set of capillary
modes. Such thermally excited waves, whilst of microscopic amplitude, scatter light appreciably.
The spectrum of the light scattered by a surface
mode of defined wavenumber reflects the temporal
evolution of that wave, and thus carries information upon the relevant liquid or surface properties [4].
Systems studied include those in which the tension of the surface or interface and the viscosity of the bulk
fluid or fluids were inferred from the light scattering
data (see [4, 5] for recent reviews). Here we will be
concerned with specifically surface effects such as
arise in molecular films.
Capillary waves are governed by the restoring
force of surface tension and are damped by viscous dissipation within the bulk liquid. However other, specifically surface viscoelastic properties influence capillary waves. In particular Goodrich [6] postu- lated that the surface tension and the dilational elastic modulus of a molecular film must be ex-
panded as viscoelastic moduli. This has been verified
on the basis of microscopic arguments [7, 2]. The temporal evolution of capillary waves is thus govern- ed by four specific surface properties, two being the
tension (yo) and dilational modulus (so) and two being viscosities (y’ and E’) associated with these
quantities.
Monolayer covered liquid surfaces (including fatty
acid [8-11], lipid [10-13] and polymer [9,14] mono- layers) have been studied by light scattering from capillary waves. In nearly all cases these studies have involved spreading a very expanded monolayer and subsequently monitoring the changes in the light scattering spectrum as the monolayer was compres- sed. Both correlation techniques [12, 8, 10] and spectrum analysis [9, 11] have been used to estimate the capillary wave frequencies (wo) and damping
constants (r). From these two observables up to four surface properties have been extracted. Such
analyses have incorporated essentially arbitrary as- sumptions or data from other experiments:
(1) information from other experiments concern- ing some of the properties (e.g. yo, eo) has been
used to constrain the interpretation process [8-10].
Unfortunately it appears that the values of yo and Eo affecting the capillary waves probed by light scattering may not be identical with those found in classical experiments. For example molecular relax-
ation processes may involve a yo(w) which differs from the equilibrium tension value [15]. Similarly,
diffusive interchange of surfactant molecules be- tween surface and subphase will cause e to be frequency dependent [16] ;
(2) similarly, whilst a particular surface property
(e.g. y’) may be zero for the case of a specific amphiphile, this cannot be assumed to be true in general ;
(3) the various surface properties affect the dis-
persion behaviour of w with q. Measurement of wo and T as functions of q should thus, in principle,
determine these properties better than an obser- vation at a single q [17]. Unfortunately, frequency dependence of these surface properties may modify
the dispersion behaviour [15]. The characteristic effects of the various properties will thus not be easily discernible.
Unfortunately there can in general be no unique interpretation of two experimental observables in terms of four physical properties, except perhaps in
very special circumstances.
The monolayers used in this work are formed of the neutral lipid glycerol monooleate (GMO). Both
mono- and bimolecular films of this lipid have previously been studied by light scattering. Black lipid films of GMO provided the first unambiguous
evidence for the existence of an interfacial viscosity
associated with the modulus governing shear normal to the interface plane (tension) [18]. Subsequent
studies of such films have shown viscoelastic relax- ation of this modulus [15]. The phase transitions of GMO (-- 16° C) in both bi- and monomolecular films have been studied [19, 13]. The present exper- iments are restricted to room temperature, well above this transition, where the classical surface pressure-area isotherm of GMO is well established
[20-23].
This paper concerns two major points :
(1) a relatively novel approach to data analysis for monolayer covered surfaces [24, 25], an extension of
a method successfully applied to interfaces [26, 27]
for which only the interfacial tension and the viscosi- ties of the bulk fluids affect the capillary waves. For
the monolayer case the physically interesting par- ameters of the fit are the two surface moduli mentioned above (each comprising elastic and visc-
ous portions). Here we demonstrate the practicabili- ty of extracting all four surface properties ab initio
from a single observation of the spectrum of the scattered light. Where possible, the surface proper- ties thus deduced are systematically compared with
data obtained by classical methods ;
(2) the frequency dependence of the surface mod-
uli. Viscoelastic relaxation is clearly demonstrable
for the transverse shear case, and can be inferred for
the dilational case. For transverse shear the variation of the relaxation rate with molecular packing has
also been measured. These observations, the first of their kind, open a new field of interfacial viscoelas-
ticity.
2. Theoretical background.
Various fluctuations, both transverse (or capillary)
and longitudinal (or dilational), in a monolayer may scatter light. However, the intensity scattered by longitudinal waves is, in general, very much lower than that due to the transverse waves [9]. We thus only consider scattering by the capillary waves. The theory is well established [4] and only a brief
summary of the relevant points is necessary.
2.1 CAPILLARY WAVES. - Capillary modes are
characterised by a surface wave number q ( = 2 1T /A). Such a surface fluctuation can be described by the departure of the surface from its
equilibrium plane :
Experimentally waves of real q propagating in the
x direction are selected for observation, their tem- poral evolution being characterised by the frequency
w (= w 0 + i F ). This is related to q via the dispersion equation, which is [4, 28] :
where y is the surface tension, q the viscosity and
p the density of the liquid, and
In equation (2), E is the dilational modulus of the
monolayer,
A being the molecular area in the film.
The spectrum of the light scattered by the ther- mally excited capillary waves reflects their temporal
evolution. For a monolayer covered surface the spectrum is an explicit function of the surface
properties y and e [4] :
While the spectrum is approximately Lorentzian, the
deviations of P (w ) from that form are well estab- lished [29, 26]. The several surface properties affect
P ( w ) in different and characteristic fashions. As will be seen below, this fact permits a single experimental
observation of the spectrum to be analysed in terms
of the four properties affecting P ( w ). In our exper- iments we measure the field correlation function of the scattered light, which is the Fourier transform of
P(w).
2.2 SURFACE PROPERTIES.
-By analogy with the
three-dimensional case, e corresponds, for isotropic strain, to the modulus of hydrostatic compression in
the plane of the film (K), whereas for uniaxial strain it also involves [30] the corresponding shear modulus
(S ) :
Capillary waves restricted to the x - z plane involve
uniaxial compression and dilation, so that the appro-
priate value of E will be Buniaxial. The value of
e appropriate to the measurement of a monolayer
isotherm will depend upon the exact nature of the strain applied [31]. In practice, apart perhaps from highly compressed, « solid » mono layers, Swill usually be negligible. Despite these complications,
8 will here be referred to as the « dilational modu- lus » for brevity.
As pointed out by Goodrich [6], both Band
y may be extended to incorporate viscous .effects.
Following the usual rheological convention we may write (there is no established notation in this field) :
Here yo and Eo are elastic moduli describing the
response of the system to shear normal to the interface (tension) and dilation within the plane of
the interface respectively, y’ and E’ being the corresponding surface viscosities. These viscosities
are not the conventionally measured « surface vis-
cosity » which refers to shear within the plane of the
interface [3], but are distinct quantities appropriate
to well-defined strains applied to the monolayer.
The nature of the quantities here called y’ and
£’ is not as yet fully understood. They may be
regarded as macroscopic hydrodynamic variables [6]
or as surface excess properties arising from the microscopic inhomogeneity of the fluid interfacial
region [7, 2]. Theoretical predictions of their magni-
tudes are not available.
The dilational modulus affects the capillary waves
because these waves couple to longitudinal fluctu-
ations of the film [28]. This coupling has been
treated elsewhere [4, 32] and we simply present
some results useful for the discussion of experimental
data. As Eo increases (relative to a fixed yo) both the
wave frequency wo and the damping r vary (Fig. 1), passing through maxima when Eo/,yo = 0.16. For
£o ? yo both wo and r vary slowly with Eo. The
surface viscosity s’ reduces the magnitudes of these
1274
Fig. 1.
-The variation of capillary wave frequency and damping as functions of Ea/ yo, computed for three
different wave numbers and assuming yo = 72.75 mN/m.
The variations are shown for these different values of e’ : 0 (- ), 10- 4 (- -) and 5- mN s/m (- - -).
variations. In contrast the surface viscosity y’ sys-
tematically increases r and reduces Wo.
2.3 VISCOELASTIC RELAXATION. - In conventional
rheological notation [33], an oscillatory stress (T(t) = T * e‘ w‘) and strain (u (t ) = u* eiwt ) are re-
lated via a complex dynamic modulus G*(w):
In the present context the storage modulus G’(co)
can be identified with yo (or 80) while the loss
modulus G"(w) corresponds to w y’ (or toE’).
There are no microscopic theories of interfacial
rheological relaxation. However for linear viscoelas-
ticity, combinations of simple rheological models
can be found which exhibit arbitrary frequency dependences of G*. We mention only two such
models : the Voigt viscoelastic solid and the Maxwell fluid. For the Voigt model both G’(w ) and G " (to )/ w are constant. The Maxwell fluid, corre- sponding to a single exponential relaxation (of time
constant T), is described by
where Ge is the equilibrium (w 0 ) elastic modulus,
and G the strength of the relaxation process. The Maxwell model interpolates reasonably between the
low frequency viscous and high frequency elastic
behaviour of a viscoelastic fluid [34].
3. Experimental methods.
3.1 LIGHT SCATTERING.
-Our heterodyne spec- trometer has been described in detail elsewhere [35].
Briefly, light from an Ar+ laser (A = 488 nm) was spatially filtered to ensure a Gaussian intensity profile (TEMO. mode) and illuminated the liquid
surface. Light scattered by thermally excited capil- lary waves was detected using heterodyne methods
to measure the small frequency shifts of the scattered
light (- kHz ). The train of photodetections from the
detector was processed by a multi-bit photon cor-
relator (Malvern K7025), interfaced to a PDP11/34A
minicomputer.
Autocorrelation of a heterodyne mixture of two
optical fields, corresponding to intensities Is (scat-
tered field) and Ir (reference field), yields [36]
where g (’) (,r ) and g(2)(T ) are the first and second order correlation functions of the scattered field
respectively. Provided that Iris sufficiently greater than I, the third term of this equation dominates the time-dependence. In practice, our experiments in-
volved ratios of IS/Ir 10- 3, so that the self-beat
term was much smaller than the random noise on the observed correlation functions. Such 7s//r ratios are
much lower than those usually advocated [37] for the
rather different case of monotonically decaying
correlation functions. The low Is/Ir ratios proved entirely satisfactory here, probably because our experiments employed methods recently described
for rapid data acquisition [38], involving amplifi-
cation of the small modulation of G(,r).
The wave number q of the capillary mode ob-
served is used as an input to the direct fitting
method : P ( w ) is evaluated as a function of the four surface properties at a specific q value. A good
estimate of q is thus central to the success of this
approach to data analysis. In the present work q was determined to 0.25 % [35].
The light scattering was significantly affected by
instrumental effects which arose from the finite extent of the laser beam on the surface [39]. The
observed spectrum was a convolution of P (w ) with
an instrumental function. For a laser beam of Gaussian profile the observed correlation functions
can be written as
where j3 is the standard deviation of the instrumental function in the frequency domain and f ( T ) is the
time dependence of the correlation function ex-
pected from waves of the selected, central q value.
3.1.1 Data analysis.
-Two different analyses were applied to our correlation data. Both involved the
use of non-linear least-squares fitting with appropri-
ate mathematical forms. The wave frequency wo and
damping r were determined using equation (13)
with an exponentially damped cosine time depen-
dence :
where the phase term 0 (small and negative for
these data) accounts for most of the deviations of
P(w) from an exact Lorentzian form [12]. Such
estimates of wo and T are known to be unbiassed
[40].
The exact spectral form of equation (5) was also
used in fitting experimental data. The applicability
of such analysis to experimental data is as yet in its infancy : systems involving only yo and q have been investigated [26, 27], but the extension to film- covered surfaces or interfaces is relatively novel [13].
We refer to this approach as « direct fitting ». The
observed correlation functions were analysed in
terms of the four surface properties (yo, y’,
EO, e’) by fitting with equation (13) using a time dependence defined by the Fourier transform of P (w ) formulated as a function of these four proper- ties. The viscosity and density of the subphase were
assumed to have their accepted values. In detail we set
and use [26]
in equation (13). Here the parameter A allows for the possibility of a self-beat or second-order contri- bution to the observed correlation functions. Fitted values of A were always very small, if not zero, as
expected from our low 7s//r ratios. In analysing the
data presented below, we therefore fixed A = 0.
Detailed discussion of the sensitivity of the direct
fitting to the various surface parameters is deferred
to section 5. Tests with simulated data [24, 25]
suggested that this analysis was stable and robust.
For data of reasonable quality (random errors on G (,r ) - 1 % of amplitude A) the surface properties
used to generate the data were adequately recov-
ered. For only one combination of surface properties
were erroneous solutions found in these tests [25],
and even then the majority of fits gave accurate estimates of the properties. This situation involved
small values of y’ and zero e : in the poor fits
yo and E’ were excessive (the latter greatly so), while
the instrumental linewidth {3 was underestimated.
Constraining {3 to be of physically reasonable magni-
tude (> half of value found for good fits) avoided the
apparent secondary minimum in the sum-of-squares hyperspace quite efficiently. In our experiments,
several correlation functions were recorded under identical conditions at each molecular area studied.
Application of appropriate constraints to those few functions which gave surface properties differing
from the rest always yielded internal consistency in
the experimental results. Such ambiguities only
occurred in cases where both the surface pressure and y’ were small.
3.1.2 Instrumental effects.
-The form of the instru-
mental function is basic to the data analysis ; we
have therefore measured it for our apparatus. The time dependence of the observed correlation func- tion is of the form
where f (T ) is the exact theoretical form (Eq. (16)
with A = 0) for waves of the selected q value.
Experimental correlation functions for light scattered by capillary waves on the free surface of water were analysed (Fig. 2). G ( T ) was divided by f(T),
evaluated for the known properties of water, yielding h ( T ). These estimates of h ( T ) were somewhat noisy, particularly around the zero-crossing points of G(,r). However h ( T ) was clearly linear in T 2, the slope giving j8 = 4 480 s-1 (within 2 % of that from the direct fitting procedure). This Gaussian h ( 7- ) is
shown in figure 2, together with [/(T).A(T)],
which agrees excellently with the observed data.
This comparison involves no fitting of f(,r). We
conclude that a Gaussian function provides an acceptable description of the instrumental effects upon G ( T ).
It has been suggested [39] that f3 can be calculated
a priori, given certain experimental parameters. The measured value just quoted agrees reasonably with
that computed for our apparatus. However because various other effects (e.g. vibrations, electronic distortion) might affect f3 we prefer to keep it as a
free parameter of the data fitting. We have exper-
imentally confirmed the theoretically predicted [39]
dependences of /3 upon q, yo and the laser beam diameter.
3.2 MONOLAYERS. - Our Langmuir trough was
’