Water dynamics in a planar lithium hydrate in the interlayer space of a swelling clay. A neutron scattering study

HAL Id: jpa-00209874

https://hal.archives-ouvertes.fr/jpa-00209874

Submitted on 1 Jan 1984

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.

Water dynamics in a planar lithium hydrate in the interlayer space of a swelling clay. A neutron scattering

study

J. Conard, H. Estrade-Szwarckopf, A.J. Dianoux, C. Poinsignon

To cite this version:

J. Conard, H. Estrade-Szwarckopf, A.J. Dianoux, C. Poinsignon. Water dynamics in a planar lithium

hydrate in the interlayer space of a swelling clay. A neutron scattering study. Journal de Physique,

1984, 45 (8), pp.1361-1371. �10.1051/jphys:019840045080136100�. �jpa-00209874�

Water dynamics ^{in a} planar ^lithium hydrate ^{in the} interlayer space of a swelling clay. ^{A neutron} scattering study

J. Conard (*), ^H. Estrade-Szwarckopf (*), A. J. Dianoux (**) ^{and C.} Poinsignon (**)

(*) C.R.S.O.C.I., C.N.R.S. Orléans, ^France

(**) ^Institut Laue-Langevin, 156X, ³⁸⁰⁴² Grenoble Cedex, France

(Rep ^le 7 juillet 1983, ^{révisé le} 24 fevrier ^1984, accepti ^{le 21}

1984)

Résumé. 2014 La structure dynamique ^{de la} première ^{couche d’eau} adsorbée dans les argiles

^{été étudiée} par diffusion

quasiélastique ^{des neutrons. Les} données de la RMN du 1H et de 7Li ont permis ^d’établir

^modèle d’hydrate plat

où les protons ^{de l’eau sont} impliqués ^{dans deux} mouvements de rotation. L’étude par diffusion quasiélastique ^des

neutrons réalisée

l’hectorite Li

confirmé la validité du modèle et précisé ^les temps ^de corrélation des deux rotations.

Les temps caractéristiques de la rotation lente de l’ensemble de l’hydrate par rapport à ^l’axe

^{du feuillet} argileux

et de la rotation rapide de la molécule d’eau _par rapport ^à

C2 ^ont été mesurés. La dépendance

tempé-

rature de

mouvements est faible entre 300 et 240 K, ^{forte entre 240 et 190 K où ils}

bloquent simultanément.

La structure bidimensionnelle de l’hydrate plan apparaît à ^travers l’anisotropie ^de l’amplitude ^{de la} partie quasiélastique ^des spectres ^obtenus pour des échantillons orientés.

Abstract.

Dynamical ^structure of the first water layer

clay ^{has been} investigated by quasielastic ^neutron scattering (Q.N.S.). Using

dynamical model built from N.M.R. data of 1H and 7Li, Q.N.S. studies have been

performed

Li-hectorite to give correlation times of rotations.

Characteristic times for

slow rotation of the whole hydrate ^with respect ^to the c-axis of the clay platelet, ^and

a

fast rotation of water molecules around their C2 ^{axis have been measured.} Both motions stop ^{at the}

tempe-

rature (190 K). Bidimensional structure of the flat hydrate ^{is also} proved through ^the anisotropy ^{of the} quasi-

elastic broadenings ⁱⁿ clay ^{oriented films} of monovalent cations.

Classification

Physics ^Abstracts

61.50E - 61.50K - 61.12

64.70K

Introduction

The pioneering ^neutron scattering ^{works on inter-}

lamellar water dynamics ^{in Li+ and Na+} exchanged

montmorillonite and vermiculite clays [1, 2] ^were mainly concerned with high water contents and the data were only interpreted ^{in terms of} translational fickian diffusion. Hall et al. [3] performed Q.N.S.

measurements on clay ^water systems ^at both medium and high energy resolution and these studies per-

mitted, for the first time, ^a separation of the rotational and translational motions in clay ^water systems, without nevertheless exhibiting any anisotropy ^{in the} proton motion. Our _purpose was to determine the

dynamical ^nature of the first water layer ^{on hectorite,}

before studying ^more hydrated ^{states such as the} gel.

The starting point ^{has been a} 1 Hand ’Li N.M.R.

study ^{of the} (Li+, 3H20) [4] hydrate ^on the hectorite surface supported by I.R., X-ray ^and thermodynamics

measurements [5-7]. ^This study ^{leads to a model}

which was confirmed and precised by ^the reported

neutron quasielastic scattering experiments.

This _paper is organized along four sections. We first describe the hectorite clay used for this microscopic study ^and report ^{what was known} about ion and water structure in this clay ^from thermodynamic, ^{I.R. and}

N.M.R. studies. The methods used for ^a water content estimation are then presented ^The Q.N.S. method, together ^{with the} spectra analysis ^for powder ^and

oriented-films measurements are given ⁱⁿ chapter ^2.

Chapter ^{3 describes} experimental arrangement ^and results whereas discussion is presented ⁱⁿ chapter ^4.

1.1 HECTORITE.

Hectorite is ^a lamellar silicate, precisely ^a trioctahedral clay [8], ^its formula is deduced from that of talc, by substituting ^{one OH} by ^F

and some Mg2 + by ^{Li+ in the} octahedral sheets

Although ^{the talc} layer ^{is neutral,} isomorphic ^substitu-

tion produces ⁱⁿ hectorite, ^{an electric} deficiency

balanced at the clay layer ^surface by exchangeable

cations (See Fig. 1).

Talc: Si8Mg6(OH)402o.

Hectorite : IS’13(Mg6 -.Li.,) (F, OH)2 0201 [Li+, nH20]x ^with

^0.58.

Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphys:019840045080136100

Fig. ^1.

^Talc and hectorite structure, ^after Hofmann, W., Endell, ^{K. and} Will, D., ^Z. Kristallogr. ⁸⁶ (1933) ^340.

By comparing the behaviour of hydrated ^{talc and} hectorite, ^{it is} clear that the water adsorption pro-

perties of hectorite result, ^{as in other} clays, ^{from the,} balancing ^cations.

Hectorite is a well micro-crystallized clay ^whose

structure and morphology ^have been described in electron microscopy ^studies [8] : ^{the small} crystals

appear like needles (with ^{small coherent} domains,

diameter - 2 000

500 A, giving ^{rise to} difficulties for X-rays studies). The similar ionic radius of Li+

and Mg2 + ^{and the} perfect continuity of octahedral and tetrahedral sheets explain ^the very low distortion in the network, ^and give ^{rise to the} very special ^local properties ^observed by ^I.R., E.P.R. and N.M.R.

Moreover, ^{it must} be noticed that, among all the natural minerals, ^{hectorite has a} very low iron content

( ⁴ ppm). ^In contrast to montmorillonite and ver-

miculite the N.M.R. and E.P.R. lines ^{are narrow} with

a well ^{resolved structure due to} high ^local symmetry.

1.2 WATER IN LI-HECTORITE.

The study ^{of water}

in the interlamellar _space of clays ^{is well documented} [5-7, 9-13]. This space is limited by ^two negative charged oxygen surfaces [12]. It is clear that a study ^of

water adsorbed in clays ^must begin ^{with the} study ^of

the hydration nucleus formed by the cation hydration

shell. Thermodynamic and I.R. studies [5,13,14]

show the first hydration shell around the Li+ balan-

cing cation, ^{to be} composed ^{of three water molecules} firmly ^{bound to the ion} [15]. From theoretical calcu-

lations in the free hydrate [16], ^{the bond} length ^and

orientation of water molecules are known.

Using 7 Li ^and 1 H N.M.R., ^Conard [17] ^confirmed

the axially symmetric ^{site for the Li+} cation and assumed that the water protons undergo ^two rapid

rotations around perpendicular axis : the firts one, related to the whole hydrate, ^{around the} c-axis of the

clay platelets and the second one related to the water molecule (Fig. 2). ^Both orthogonal ^rotations stop around 190 K. However due to very different moments of inertia, their correlation times must be _very different.

Taking ^{into account all the} experimental results,

model has been proposed ^{for the} system. ^{The tri-}

hydrate is centred above and/or ^under

hexagonal

oxygen cavity ^{of the} phyllosilicate layer (Fig. 3). ^Thanks

to the _very peculiar symmetry and the small distances between the different possible sites, ^the protons ^are

Figs. ^2-3.

^First proposed ^{model for} (Li-3H20) ^hectorite

(from [4]).

allowed to jump among 12 equivalent ^{sites on a circle}

of radius 2.13 A centred on the Li + cation : this motion is equivalent ^to

reorientation of the whole

hydrate around its symmetry ^axis parallel ^{to the c-axis}

of the layers. Moreover, ^the orientation of each water molecule towards the neighbouring oxygens permits

pseudo-rotation ^{for the} protons

^{circle of radius}

1.23 A around an axis parallel ^{to the a-b} planes. ^This

relation can occur among 4 ^or 6 equivalent positions according ^{to whether} hexagonal ^{holes in} neighbouring layers superimpose ^{or not. So we} expect ^{two rotational} motions for the protons, ^with quite different charac- teristic ’times. The slow motion involves an oxygen atom motion whereas the fast one involves only hydrogen ^atoms hopping ^{between sites. However}

both motions are probably correlated since they stop

at the same temperature.

1.3 SAMPLE PREPARATION.

Hectorite is a natural mineral extracted from Hector’s (California) ^{vein as a}

calcite-hectorite mixture.

The pH equilibrium ^of hectorite in water is 9 and thus it is very sensitive ^to any chemical treatment :

purification by ^acid leaching ^{must not be} performed.

It is only ^involved by gravimetric ^treatment [18].

Even cation exchange ^{must be} performed very

smoothly ^to preserve mineral ordering ^{of the local} configuration.

Part of the so-obtained Li-hectorite has been used to prepare oriented self-supporting films from _very dilute suspension : ^{the films are 3 to 4} hundredth of

mm

thick. The distribution of layer orientations is

± 15 deg. ^{as revealed} by X-ray diffraction and in derived Cu-hectorite by ^E.P.R. [23].

The water coverage, equivalent ^to half

monolayer

or 3 water molecules _per Li cation, ^{was obtained} by leaving ^the sample, powder ^{or films to} equilibrate ^{at a}

very ^rY low relative humidity k = Yp 0 ^{10-2 [6]. The water}

content was controlled by the N.M.R. spectra : ^{a Pake} doublet is observed whose distance between peaks ^is A ²

Gauss for a powder ^and A(3 ^{cos’ 6 -} 1) x - for

films, ⁶ being ^the angle between the magnetic ^field

and the c-axis of the layers. ^{This distance A has been}

found to depend upon ^{water content} and is 2.7 Gauss for 3 H20/Li+ water content : the theoretical one

for a powder ^of protons undergoing only ^{two ortho-} gonal rapid ^rotations (Fig. 4). ^For higher water contents

the doublet separation tends towards zero.

For Q.N.S. experiments, sample containers are thin walled aluminium cylindrical ^{chambers with an air-} tight 0-ring ^{of 50}

diameter and 1.5

thickness.

About _{1.5 g} of powder ^was used, ^while only ^0.6 g of oriented films were available, accumulating

great number of small rectangular (20

⁶ mm) ^{thin film} pieces ^{in the same} cell. This stacking ^{did not increase} noticeably ^the angular dispersion ^of the total c-axis orientations.

Fig. ^4.

^{NMR doublet} splitting ^{observed at 300K in}

hectorite

water coverage. The ^water coverage has been estimated using the isotherms given ⁱⁿ [6].

2.1 Q.N.S. METHODS. - Neutron scattering by H20 ,

adsorbed on clays ^is essentially incoherent Since

can neglect, ^{in the} quasielastic domain, the coherent

contribution, ^{the double} differential scattering ^cross-

section

can be written :

where k, ko ^{are the} incident and scattered wavevectors, k - ko

Q, ^{the momentum} transfer, E - Eo

^hm

the energy transfer and ar.h3l (fine the coherent and incoherent cross-sections.

The incoherent scattering ^law S.(Q, w) ^{is related} by ^Fourier transformation in space and time ^to the Van Hove [19] self correlation function G.(r, t) ^where G.(r, t) represents ^{the ensemble} average probability

of finding ^a particle ^{at a} position

^at time t, given

that at t

0 the same particle ^{was at the} origin

It is useful to define the incoherent intermediate

scattering ^function

The protons ⁱⁿ (Li+, ³ H20) hydrate ^can perform rotations, vibrations, translation. By assuming ^no

correlation between these different motions, ^the intermediate scattering ^{function can} be written as

the product :

In the quasielastic region ^the phonon contribution

to a Debye-Waller ^factor, Iv"(Q, ^{t) may be written}

where U2 > ^{is the mean square} amplitude ^{of the}

vibrations. After Fourier transformation, ^{the inco-}

herent scattering function is thus :

where S.(Q, w) describing each reorientation motion has been individualized as srotl and srot2 and where ®

symbols stand for convolution product

2.2 Q.N.S. POWDERED SAMPLE ROTATION SPECTRA. -

The best _way of interpreting ^a Q.N.S. spectrum ^{is to} calculate the self scattering ^{law for a} given ^motional model, ^{to fold it} by the instrument resolution function,

and to compare it with experimental spectra. ^We started from the established model [4] ^{of the} (Li+,

3 H20) hydrate assuming ^{that the} protons undergo

two uniaxial planar reorientations. For random

jumps among N equidistant ^{sites on a circle of radius a} [20], ^the scattering ^function S(Q, co) ^is given by ^the following ^{formula :}

where

The scattering function is then a sum of an elastic

peak le ^{and a} quasielastic part IQ ^made up of (N -1 )

Lorentzian curves. The

quantity ie -- ^{e + Q called elastic}

incoherent structure factor (EISF) ^{is a} function of Q ;

its experimental ^{variation versus} Q ^{can be used to}

check the geometry ^{of the motion} [22].

The in ^values determining ^the reciprocal ^half-

width at half maximum of the nth Lorentzian curve are given by :

It must be noticed [20] ^{that for a} large ^{number of} jumps ^N, t 1 tends ^to the reciprocal 1 /DR ^{of the rota-}

tional correlation time, ^{which can be measured} by

other techniques, ^{such as NMR.} (In ^{our model each} planar ^rotation taking place ^{in the} Li-hydrate ^is

assumed to be completely independent ^{from the}

other ^one and may be described by ^{such a} formula.)

Each rotation is defined by its values for a, N and

(Table I).

2.3 ORIENTED SAMPLE Q.N.S. ^{SPECTRA. - The for-}

mulae (8-9) ^{have been} established for powder, ^that

is for randomly distributed rotational axis. For a

crystalline sample, ^the scattering ^function depends

on the angle 9 between the displacement ^{vector and}

the neutron transferred momentum Q. ^{For instance in}

our oriented samples, ^{assumed to be ideal} platelets,

there would be no broadening ^with Q perpendicular

to the rotation planes, (Q1.) from the slow motion.

On the contrary broadening ^{would take} place ^with Q parallel ^{to the} layers (Qll) ^{(Fig. 5). A complete calcu-}

lation of the quasielastic spectra lineshape ^and intensity ^{for all} Q orientations is difficult to achieve but ^{we can} obtain ^an estimation of the anisotropy

factor.

This factor is defined by the ratio A ||A.L ^{of the}

quasielastic intensities of the spectra ^obtained respec-

tively ^for Q jj ^{and Q 1. to the film planes. In} crystalline samples, ^{this factor} would be infinite. However,

we know that our films are not perfectly ^{oriented :}

the c-axis of the individual particles ^{are tilted around}

the mean axis of the film. Such ^an orientation distri- bution has been measured by E.P.R. in Cu hectorite to have 1 5° half width at half maximum. Therefore,

the anisotropy factor would be no*longer infinite but

strongly ^reduced

In order to calculate this factor, ^{we made use of}

the formula (14) given by A. J. Dianoux et ale [20]

for the case of uniaxial rotational diffusion in liquid crystals. Indeed the site number N = 12 is great, ^the approximation of continuous rotation is quite ^valuable.

Moreover, instead of keeping ^{the whole sum of} Lorentzians, ^we only consider the first one giving

the ^main contribution in the explored Q-range. ^Its intensity ^{is then} given by the first order Bessel function

with a

rotation radius and 0 the angle ^between

the rotation axis and the Q ^direction.

To take account of the c-axis distribution of the

sample, ^{we introduce an} axially symmetric distribution function p(B) ^{of the} individual particles ^{c-axis on the} unity ^half sphere (see Fig. 5c).

0 being ^{now the} angle between the mean axis Oz of the sample ^and Q, ^the intensity A(O) ^{of the} quasi-

Table I.

Proposed ^model.

N

number of sites,

rotational radius.

Fig. ^5.

Estimation of the anisotropy ^factors expected ^from Q.N.S. measurements

(Li’3H,O) ^{hectorite self supporting}

films, ^{related to} a) ^{the slow rotation of the whole} hydrate, b) ^the fast rotation of protons ^{in each water molecule.}

elastic part ^{of the} spectra ^{is then} given by

(Q, c) ^{is a function} g/(0, P, cp) ^{of 0} (experimentally fixed) ^{and the} spherical coordinates (p, cp) ^{of the c}

unit vector.

So

This double integration ^{is not in} general ^cases easy but we look only ^{for an} estimation of the anisotropy

factor A(n/2)/A(0)

AIIIA, in the ^{case of the slow}

rotation of the whole hydrate. This rotation has been,

as we will see later, observed at A

10.05 A ^for

which the transferred momentum Q ^{is 0:85} A-1

when observing in reflexion geometry (Fig. 6). ^The

rotation radius a is equal ^{to 2.18} A.

The orientation distribution function p(p) ^has

been written as either a square function

Fig. ^6.

Q.N.S. experimental configuration

^{IN5 TOF}

spectrometer ^{at ILL.}

either, ^as suggested ⁱⁿ [20],

Maier-Saupe (M.S.)

continuous function

In both descriptions, ^{Z is the} normalization factor and 30, ^{the HWHM} of the distribution will determine

m in the second case. The anisotropy factor values so-obtained are given ^{in table II} for different HWHM flo ^{and for both} distribution shapes. ^{In this}

table II, ^{we see that the} calculated values are indeed

quite ^{sensitive to} flo ^{but also to its} shape.

For the fast rotation, anisotropy ^{is not so} easy ^to evidence ^as the rotation axis are distributed ^on

circle in the _a, b plane (Figs. ^{3 and} 5b). Therefore the calculation becomes much ^more complicated, ^with

an expected anisotropy smaller than for the slow motion. The only way ^we found to estimate easily

this anisotropy ^{factor was to assume a} simple Doppler

effect between the moving protons ^{and the neutrons :} the quasielastic intensity ^{is then} proportional ^{to the} projection ^{of the} proton displacement ^vector upon the Q direction. In that _way, with

square angular

distribution of the platelets ^{we find}

with flo

^150. (With ^{the same} description, A 11 IA,

would be 12 for the slow rotation, ^{cf. Table} II.) Thus, ^{for the} quick ^{rotation, we} expect only

^small anisotropy, opposite ^{to that of the slow rotation,}

with All ^Aol.

3.1 EXPERIMENT. - The experiment ^was performed

at the high ^{flux reactor} of the Institut Laue-Langevin

with the Time-of Flight multichopper spectrometer IN5.

Our model led to a double rotational motion for the protons ^{in the flat} hydrate (Li+ ³ H20). ^From

the different order of magnitude ^{of both} correlation times, ^we expected, by using ^different resolutions,

to be able to observe independently ^both scattering

functions. The TOF spectra ^{were measured with}

two incident wavelengths, respectively A

^{10.05 A} (0.020 ^meV resolution) ^{and A}

5.14 A (0.138 ^meV

resolution full width at half maximum).

The scattered neutrons are detected at twelve

scattering angles ^with Q values between 0.136 and

Table II. - Calculated anisotropy factor All IA,

A(x/2) A(0).

1.14 A-1 for low incident energy and between 0.27 and 2.28 A-1 for high ^incident energy. For the

powder sample study, the container was in reflexion geometry ^{with an} angle ^{of 450 with} respect ^{to the} incident beam. The oriented self-supported samples

were examined successively ⁱⁿ reflexion and trans- mission geometry ^{at an} angle ^{of 45°}

^{1 35°} relatively

to the beam (Fig. 6).

All measured spectra ^were corrected for sample

holder scattering, absorption, self-shielding ^{and nor-}

malized to each other by comparison ^{with a Vanadium}

standard and with the frozen sample spectra (for

T 180 K all proton ^{motions are} stopped [4]).

Comparing spectra in both orientations (transmission

and reflexion geometry) ^{is thus valid.}

3.2 POWDER SAMPLE RESULTS.

Figure ^{7 shows}

energy spectra ^{obtained with the} powdered ^hectorite (Li+ ³ H20), ^{at room} temperature. They ^{are all} composed ^{of a} ð(ro) ^function superimposed ^to

quasielastic part, convoluted by ^the apparatus ^reso- lution.

The full widths at half maximum of the quasi-

elastic broadening ^{are a few tens of} pev ^{at A}

^10.05 A

and hundreds of pev ^{at 5.15} A. They ^{are almost} Q-independent, ^{in the} experimental Q-range, ^however

the quasielastic intensities of the spectra ^increase with increasing Q-values.

All those spectra ^{are almost} temperature ^inde-

pendent ^{down to 200} K ; ^{at lower} temperature ^the quasielastic parts disappear relatively abruptly.

Those features agreeing ^{with the} previously ^des-

cribed model,

conclude that both Q.N.S. spectra

are

due to a double rotational motion for the protons of the hydrate. ^{Thanks to their} quite ^different rates, the rotations could be observed separately by ^the

mean of two different resolutions and in ^a preliminary approximation, ^{we tried to fit both} experimental ^sets

of spectra, ^{as due to} independent planar ^rotations.

Thus we used the scattering functions described in

§ 2.2. ^{For each} wavelength ^{the fits were obtained}

in two steps :

a) The elastic part ^{of the} spectra ^{is cut off and} only quasielastic wings ^are analysed and fitted for all Q-values together. These fits yielded, ^{for both} rotations, ^{the residence} times Tr

(slow one) ^and Tf

(fast one) ^and corresponding radii a. and af, with

respectively ^N

^{12 and 6} equivalent positions ^on

a circle. The best fitted values are given ^{in table III}

in which we see th1it they ^{are not} far from the proposed

ones.

b) ^An elastic contribution (1) ^is introduced as

(1) ^{Note : As}

^{matter of fact, in this elastic} contribution,

we

have introduced not only the structural 2014 OH groups but also

number taking ^{into account the} scattering ^which

arises from the disorder. 70 % of this elastic intensity

from the structural OH-groups, ^the remaining being ^this

disorder contribution.

Table III.

Proposed ^values deduced.from ^NMR experiment ^and adjusted ^{ones obtained} by fitting ^the quasi-

elastic part of ’ the experimental

with the assumed scattering junction (i

correlation time).

constant parameter and the whole experimental spectra (elastic ⁺ quasielastic parts) ^{are fitted, for}

all the Q-values together (Fig. 7).

Similar fits were performed for the different tempe-

ratures. The resulting Tr and _if are given ⁱⁿ figure ^8.

We observe that both rotations have a similar small temperature dependence, ^{but their rates slow down} abruptly ^{near a} temperature of about 190 K, ^where both motions freeze.

3.3 ORIENTED FILMS RESULTS.

As we saw in § 2.3, by using ^oriented films, ^{it is} possible ^{to observe} selectively ^{the neutrons scattered} with transferred moments parallel ^or perpendicular ^{to the} layer plane ^{and thus to the assumed motion} planes. ^These

observations are achieved by using spectra ^obtained

at 20

900 for samples in reflexion and trans- mission geometry (Q II and Q ) (Fig. 6), ^{at both wave-} lengths (A

10.05 and 5.15 A and ^at different tempe-

ratures.

At room temperature, the slow rotation anisotropy

was evidenced through ^the intensity difference of the

quasielastic part ^{of the} spectra ^for Q and ^{Ql (Fig. 9),}

normalized to the frozen samples spectra.

In order to extract the intensity ^{of the} quasielastic

part ^{of the} spectra, it is necessary ^to fit them with an

analytical ^formula. However, as we saw in § 2.3,

the line-shape ^{could not} be calculated and for simpli- city, ^{we assume} the lines to be single Lorentzians.

Their linewidths are imposed ^{in both} configurations

(Q II and Q1), and fitted with the same shape assump- tions than for the 900 powder spectra.

The HWHM used values were thus : 40 J.1eV for slow rotation 150 J.1e V for the fast one.

The only parameter ^{to be} adjusted ^{was then the} amplitude ^{A of the} quasielastic ^line.

The ratio obtained of Q II and Ql configuration amplitudes ^{are :}

A _II

=

2.9 for slow motion All ~ 1 for fast rotation .

2013 = 2.9 for slow motion, A ^for fast rotation.

Thus, the slow rotation _appears quite anisotropic

with A _II > A1. Comparing with table II, ^{we note} that the square-law distribution would give ^an agreement ^{with the} experimental A II / A ol ^{only for}

Po > 300 while the Maier-Sampe shape fits it for

Bo

^{190. Of course the} square-law ^{cannot be}

realistic model for our oriented sample ^{and this}

solution must be eliminated. The second one with

Bo

^{190 seems more} realistic and _agrees fairly

with previous E.P.R.-Po measurements (15 °) ^{on similar} samples (Cu-hectorite [23]).

However, ^{we have to} keep in mind that the experi-

mental as well as the theoretical values are obtained

by dividing ^a measurable term by ^{a small one and}

thus the ratio cannot be _very accurate. Moreover

we saw the theoretical factor to be sensitive to the

shape ^{and to} the width of the assumed distribution function. In such conditions, ^{it seems} quite ^satis- factory ^{to obtain}

good agreement ^between experi-

mental and theoretical anisotropy ^{factors. For the} quick rotation, ^no experimental anisotropy ^{could be}

detected and in spite ^{of the} rough theoretical predic-

tion we will look for possible ^reasons leading ^{to an}

apparent isotropy of the motion.

4.1 GENERAL DISCUSSION.

The motion deduced from N.M.R. analysis ^is equivalent ^{to two rotations}

with orthogonal ^axis (Figs. 2-3).

This model has been built using the theoretical

proposal ^{of a flat} trihydrate [17] in which the Li atom lies in the oxygen mid-plane. ^{However, the} experi- mentally measured electric field gradient (EFG)

suffered by ^{the 7Li} nuclei, ^{has been found smaller}

than the ^one calculated from a completely flat confi-

guration ^{of the 3 water} dipoles [4]. ^{In order to} explain

such

discrepancy, ^{it was} assumed that those dipoles flip ^rather rapidly (t ^{10- 5} s), ^the Sp3 ^orbitals

remaining pointing ^towards the Li-atom. The _appa-

rently planar hydrate configuration will thus result from

time _average.

But at

short time scale (10-14 s) ^{I. R. measure-}

ments demonstrated definitely ^{that no} H-bonding

took place and that the Li-atom appeared ^{to be out}

Fig. ^7.

Energy ^{spectra obtained at 300 K for different} Q ^{values in} powder. a) slow motion observed at 10.05 A, b) ^fast

motion observed at 5.14 A.

Fig. ^8.

Q.N.S. correlation times T. and Tf

tempe-

rature.

of the mid-position in the interlamellar _space, the

dipolar ^{moment of water molecule} being ^{550 tilted} relatively ^{to c-axis} [6, 14]. ^{At 10-’o}

^time scale, powder Q.N.S. measurements has confirmed such rotations and their radii and allowed the determina- tion of their correlation times (Table II). Q.N.S.

by oriented films proved ^{also the} anisotropy ^{of the} planar slow rotation concerning ^{the whole} (Li +,

3 H20) hydrate about its c-axis. However the quick

rotation of the water-molecule about their Li-O bond did not appear ^as anisotropic ^as predicted by

the flat model and we have to modify it somewhat In our case, the trihydrate ^{is no} longer ^{free but}

lies in the interlamellar space of the hectorite-clay,

between two oxygen planes ^{and more} precisely,

between two oxygen hexagons (2 ).

The situation may be quite different from the free state. For Li+ in the interlamellar space ^some kind of equilibrium position ^is produced by the lattice oxygen ^atoms and the water molecules (Fig. 10).

The six lattice oxygen ^atoms (bound ^to silicium)

are pointing ^one of their free spl ^{orbital towards}

the cation, ^and provide ^{a 12 electrons nest for the Li}

cation. Indeed, ^{such a} configuration may give ^{rise to}

two equivalent potential ^{wells on} either side of the

hydrate symmetry plane.

Fig. ^9.

Quasielastic intensity anisotropy ^{observed at}

300K

(Li+, 3H,O) hectorite self supporting ^{films at}

i

10.05 A.

Fig. ^10.

Cross sections of [Li+, 3H20] ^{between two}

oxygen hexagons by

[Li-0 ^water, A plane. ^L1

^axis.

L1 is ,the slow rotation axis of the whole hydrate. ^{I is the} guide rotation axis of the water molecule. When Li atom is in the average hydrate plane, ^{then the H rotate} along ^the circle, cutting it in A and B. There is

negative nest, ^out of the average hydrated plane ^{of the} Li +, ^{due to the six} sp’ ^orbitals

of the lattice _oxygens 01, 02

pointing ^towards the L1 axis.

If Li+ stays ^{in the} nest, ^{drawn out} of the plane ^at 1,4 A,

water molecules rotate around the I’ axis. Then the protons

move

along

^circle cutting ^the plane in A’ and B’. The proton motion is thus intermediate between circular and

spherical.

(Z) ^Those assumptions may appear

simplificated

However they may be justified ^Indeed,

pointed ^out in § 1.1,

in poorly hydrated hectorite, ^all spectroscopic techniques

confirm the high symmetry ^of the cation site. However, ^this high symmetry ^{is not obtained} immediately ^after

pertur- bation of the system, ^{for instance of the} water content :

very long waiting ^time (several months) is necessary ^to reach the equilibrium signal. ^Such

long ^{time cannot be correlated}

to the local organization ^{of the sole water} molecules. Heavier

systems like the whole clay layers ^{must be} involved, ^{and the}

equilibrium time may be considered

the time necessary

for the oxygen planes ^to superimpose

^{to create the}

highest symmetry ^{at the cation site.}

With the three water molecules of the Li hydrate

each Li cation is surrounded by 3 oxygen ^{atoms 2} A apart ^{and 6 ones at about 3} A, ^{structure which is}

very similar to the calculated free system (Li+,10 H20) (5 ^{water molecules at 2} A, ^{3 ones at 3} A, and the last two further) [16].

4.2 DYNAMICS oF THE SYSTEM (Li, ³ H20) ^{IN HEC-}

TORITE. - Thus two equivalent ^sites separated by ^a

distance of the order of 1 A _may be found for the balanc-

ing cation in the interlamellar space of hectorite.

If the Li atom is able to jump ^{from one to the other} site, ^the characteristic time of this oscillation must be such that 10-6

10 -14 s. With the IR time scale (10-14 s), the Li appears ^to be in one of those sites. At much longer ^{time scale,} by ^{7Li NMR, Li}

atom will appear in ^an axially symmetric ^average position.

This Li-oscillation will induce an oscillatory ^mo-

tion of the symmetry ^{axis I of 3 water} molecules and thus of their protons. Therefore, ^the quick proton motion ( ~ ^10-12 s) will appear ^as almost isotropic

as observed by Q.N.S. ^{ratio. Moreover this} implies

the Li displacement ( ~ ¹ A) ^{to be small} relatively

to the radius of slow rotation ( ~ ² A). ^On figure 10,

we see a representation ^{of the described model.}

Finally the Li-oscillation will hinder H bonding

to take place between the water molecules and the lattice oxygens ^as observed by ^I.R. Thus, ^{the Li+}

oscillation on either equivalent position, ^{can account}

for all the room temperature experimental ^results (NMR, Q.N.S. I.R.). However, ^is this Li oscillation between two wells real ? It assumes the cation site to be perfectly symmetric, ^{i.e. an exact} superposition

of the oxygen hexagons. But this has not been cris-

tallographically demonstrated We only develop-

ed here above, ^{a lot of} experimental ^{results in agree-}

ment with its

If the double well ^were not real, the Li-cation would then stay ^{on one side of the} free-hydrate ^mid plane.

Depending ^{on the} shape ^of potential well, ^{we could} find as well ^an agreement ^with I.R. and NMR results.

But the question ^of whether oscillations of the Li- cation in a single potential ^{well are able to account}

for the observed Q.N.S. isotropy, remains still open.

4.3 THERMAL DEPENDENCE.

On figure ^{8, correla-}

tion times _T, and _Tf are almost temperature indepen-

dent from 300 to 200 K and both increase abruptly

at lower temperature.

This thermal law _appears to be unusual. Indeed,

one obtains frequently ^an Arrhenius law at ligh

temperature. At ^low temperature,

competing ^tunnell- ing ^{mechanism can take} place, helping ^the preceding

process and giving ^{an almost} temperature indepen-

dent law : both mechanisms act in parallel, ^the qui-

ckest one governing the thermal behaviour.

In our case, ^we find just ^the opposite. ^{The motion}

is temperature independent ^at ligh temperature. ^At low temperature ^the slopes ^{of r}

f (1/T) ^{are conti-} nuously increasing ^{up to the} freezing point ^{of the}

motion. Thus the motion rate is fixed by the slowest mechanism and both seem to act in series.

This conjecture ^is supported by ^the simultaneous stop ^{of both} motions around - 80 °C : the complete

motion needs a combination of the two elementary

ones.

The slower _one, related to the rotation of the whole

hydrate requires evidently ^a thermal activation of the water oxygen ^atoms in order to pass ^over the lattice oxygen ^atoms. That ^means the quickest ^{motion to}

be temperature independent ^{Beside the} H-bonding,

which we saw above not to take place, ^the only possible temperature independent ^{motion is a} weakly

hindered rotation. This is to be constrated with the

case of free water, ^{in which a} correlation time of same

order of magnitude ^{has been} observed, but which is temperature dependent [24, 25].

5. Conclusion.

With these ^{Q.N.S. results at low water coverage of}

hectorite clay mineral, ^we have evidenced two motions for the water protons. ^The quickest ^{one, has the same}

order of magnitude ^and frequency ^{as that observed}

in normal water but is almost temperature indepen-

dent The slower one is related to the rotation of the whole hydrate around its c-axis. The anisotropy ^{of the}

slow motions is demonstrated This is, ^{to our know-} ledge, the first time that cation hydration ^{shell rota-}

tion has been observed by Q.N.S. ⁱⁿ clay ^minerals.

It seems to be quite general ^{and to} persist ^{even at} higher ^water coverage ^as will be reported ^{in a further} publication where the correlation between the two rotations will be expressed ^{in the} scattering ^law.

Acknowledgments.

P. Lauginie ^is sincerely acknowledged ^{for his kind}

help ⁱⁿ « tricky ». numerical calculations.

References

[1] HUNTER, ^R. J., STIRLING, ^G. C., WHITE, ^J. W., ^Nat.

Phys. ^{Sci. 230} (1971) ^92-4.

[2] CEBULA, D., THOMAS, ^R. K., WHITE, ^J. W., Clays Clay

Minerals 28 (1980) ^19.

[3] HALL, ^P. L., Ross, ^D. K., TUCK, ^J. J., HAYES, ^{M. H.} B.,

Proc. Int. Clay ^{Conf., Oxford 1978} (1979) ^121-130.

[4] CONARD, J., Proc. Int. Clay ^Conf. (1975) p. 221.

[5] FARMER, ^V. C., RUSSEL, ^J. D., ^{Trans. Farad. Soc. 67} (1971) ^1737.

[6] PROST, R., Thesis, Paris Ann. Agron. ²⁶ (1975) ^463.

[7] CALVET, R., ^Ann. Agron. ²⁴ (1973) ^{77 et 133.}

[8] OBERLIN, A., MERING, J. Bull. Soc. Fr. Miner. Cristall.

LXXXIX (1966) ^29-40.

[9] WOESNER, ^D. E., ^J. Magn. ^{Res. 16} (1974) ^483.

[10] HECHT, ^A. M., GEISSLER, E., J. Coll. Int. Sci. 44 (1973) ^1.

[11] HOUGARDY, J., STONE, W., FRIPIAT, J. J., ^{J. Chem.} Phys.

69 (1976) ^9.

[12] ^KADI HANIFI, N., ^Thesis (1976) ^Paris.

[13] POINSIGNON, C., CASES, ^J. M., Bull. Soc. Fr. Mineral.

Cristall. 101 (1978) ^469-477.

[14] POINSIGNON, C., FRIPIAT, ^J. J., CASES, ^J. M., ^{J. Chem.}

Phys. ⁸² (1978) ^16-1855.

[15] POINSIGNON, C., YVON, J., MERCIER, R., ^{Israel J.}

Chem. 22 (1982).

[16] CLEMENTI, E., POPKIE, H., ^{J. Chem.} Phys. ⁵⁷ (1972)

1077.

[17] CONARD, J., Magn. Res. Coll. Interf. ^{Sci. ACS} Symp.

Series T (1976) ^34-85.

[18] GLAESER, R., MERING, J., ^{C.R. Hebd. Séan. Acad. Sci.}

267 (1968) 463-466.

[19] ^VAN HOVE, L., Phys. ^{Rev. 95} (1954) ^249.

[20] ^{DIANOUX, A. J., VOLINO, F. and HERVET, H., Mol.}

Phys. ³⁰ (1975) ^1181-1194.

[21] MCBRIDE, ^M. B., PINNAVAIA, ^T. J., MORTLAND, ^M. M.,

J. Phys. ^{Chem. 79} (1975a) ^2430-2435.

[22] LECHNER, R., LEADBETTER, A., ^{in the} Plastically Crystallite ^state (Sherwood, ^Ed. Wiley ^{J. and} Son) (1979).

[23] GUTIERREZ-LEBRUN, M., GAITE, ^J. M., ^J. Magn. ^Res.

40 (1980) ^105.

[24] TEXEIRA, ^{J. and} BELLISSENT, ^M. C., private ^communi-

cation.

[25] SPOSITO, G., ^{J. Chem.} Phys. ⁷⁴ (1981) ^6943.