• Aucun résultat trouvé

New critical points in frustrated smectics

N/A
N/A
Protected

Academic year: 2021

Partager "New critical points in frustrated smectics"

Copied!
10
0
0

Texte intégral

(1)

HAL Id: jpa-00209978

https://hal.archives-ouvertes.fr/jpa-00209978

Submitted on 1 Jan 1985

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.

New critical points in frustrated smectics

P. Barois, Jacques Prost, T.C. Lubensky

To cite this version:

P. Barois, Jacques Prost, T.C. Lubensky. New critical points in frustrated smectics. Journal de

Physique, 1985, 46 (3), pp.391-399. �10.1051/jphys:01985004603039100�. �jpa-00209978�

(2)

New critical points in frustrated smectics

P. Barois, J. Prost

Centre Paul Pascal, Domaine Universitaire, 33405 Talence, France

and T. C. Lubensky

University of Pennsylvania, Department of Physics, Philadelphia, PA 19104-3859, U.S.A.

(Reçu le 23 juillet 1984, accepté le 5 novembre 1984)

Résumé.

2014

Nous étudions un modèle phénoménologique de Landau de systèmes smectiques frustrés dans lequel

sont en compétition deux vecteurs d’onde de modulation k1 et k2 incommensurables. Outre les phases nématique

et smectique monocouche (SA1), les diagrammes de phase calculés montrent l’existence de deux phases bicouches

distinctes (SA2) séparées par une ligne de transition du premier ordre se terminant en un point critique. Les inten-

sités de diffusion des rayons X sont également calculées : elles conduisent à identifier l’une des phases SA2, présen-

tant une très faible modulation harmonique au vecteur d’onde (0, 0, 2 qo), à la phase partiellement bicouche SAd

connue expérimentalement Un nouveau point bicritique N-SA1-SA2 est mis en évidence en cas de forte incom- mensurabilité.

Abstract.

2014

A phenomenological model for frustrated smectics with competition for order at incommensurate wavevectors k1 and k2 is used to study phase diagrams and x-ray scattering intensity within mean field theory.

It is shown that in addition to the nematic (N) and monolayer (SA1) phases, the phase diagram may involve two distinct bilayer (SA2) phases separated by a first order line terminating in a critical point One of the SA2 phases

exhibits very little harmonic modulation and can be identified with the experimentally characterized partial bilayer

SAd phase. In addition for larger incommensurability, an unusual bicritical point where N, SA1 and SA2 phases

meet is identified.

Classification Physics Abstracts

61.30E - 64. 70M

1. Introduction.

In 1978 G. Sigaud and coworkers [1] challenged the

traditional classification of liquid crystals with the discovery of a transition between two phases macro- scopically classified as smectic A phases. Since then

no less than five distinct smectic A phases have been

identified in polar smectogenic compounds [2]. These phases are most easily distinguished by their x-ray

scattering intensities (Fig. 1). The classical mono-

layer smectic A exhibits a Bragg peak at 2 qo =(2 nj l) n

where I is the molecular length and n the nematic director. In addition, there may be diffuse scattering

at wavevectors I q I 2 qo I. This phase is the

SA 1 phase. The bilayer SA2 phase is characterized

by two Bragg peaks at qo = (2 x/2 1) n and 2 qo.

Finally the experimental signature [3] of the SAd

phase is a Bragg peak at qo = (2 x/1’) n where

Fig. 1.

-

x-ray scattering intensities for the (a) nematic, (b) SAI, (c) SA2 and (d) SAd phases. The dark dots represent

quasi-Bragg peaks and the dotted ellipses diffuse scattering.

The vertical axis is along qo.

Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphys:01985004603039100

(3)

392

1 l’ 2 1, generally with diffuse scattering for q I > qa 1. In addition, there are anti-phases [2, 4]

(A and C), which will not concern us here, with Bragg peaks off of the n axis.

Most of the above smectic polymorphism has been successfully explained in terms of a phenomeno- logical model introduced by Prost [5-7]. The nature

of the SAd phase and its relation to the SAI and SA2 phases have not, until now, been understood

completely. In this paper, we will use the Prost model and mean field theory to address this problem. Our

results can be summarized as follows. The SA1 phase

is characterized by the non-vanishing amplitude I l I of the Bragg peak at 2 qo whereas the SA2

phase is characterized by the two non-zero ampli- tudes, I ql2 and I t/J 1 I, of the Bragg peaks at qo and 2 qo. For certain ranges of model parameters, we find that there can be two bilayer phases, which we label

SA2 and SA2’ characterized by different values of the ratio i Ijl t/J 2 1 and of qo. In the SA2 phase, 102 1/1 4(l I is not small, and qo is of order 2 x/2 1

whereas in the SA2’ phase, I t/J 2 1/101 I is small and

qo is of order (2 x/1’) > (2 n/2l). We calculate the x-ray intensities for the two phases and find that the

Bragg peak at 2 qo in the SA2’ phase can be comple- tely masked by diffuse scattering. Thus, we identify

the SA2’ phase with the experimentally characterized SAd phase. The symmetries of the SA2 and SA2’

phases are identical so there cannot be a second

order phase transition between them. We find that there is a first order phase boundary along which

the SA2 and SA2’ phases can coexist that terminates in a critical point C similar to the liquid gas critical

point. The prediction of this critical point is new, and it would be of some interest to find a physical

system in which it is exhibited. The cyano series of reference [8] seem to be good candidates for revealing

its existence. We note, however, that other phases

such as the A and C phases may intervene in real systems and eliminate the critical point C. In addi-

tion we find a mean field bicritical point where the N, SA1 and SA2’ phases meet in systems with large enough mismatch between I and I’. This bicritical

point is unusual in that it arises from a model in which there is a trilinear coupling (of the form I t/J 1 12

I t/J 2 I) between the two order parameters. Such a coupling usually does not allow t/J 1 and 02 to be simultaneously critical but rather causes a first order transition to occur before criticality.

The outline of this paper is as follows. In section 2,

we define the model and discuss mean field theory.

In section 3, we discuss the phase diagrams predicted by mean field theory. In section 4, we calculate x-ray

scattering intensities for the various phases and

show that the predicted intensities for the SA2’ phase

are in agreement with the experimentally observed

intensities in the SAd phase. In section 5, we review

our results and make some comments about critical corrections to mean field theory. Finally in an appen-

dix we consider in more detail the bicritical point

where both qll and tf¡ 2 are critical.

2. Landau theory.

As discussed in reference [7], two order parameters

are needed to describe the properties of frustrated smectics. The first p(r) is the centre of mass density

of the constituent molecules. The second Pz(r) des-

cribes long range head-to-tail correlations of polar

molecules along the z-axis (normal to the smectic

layers). In the absence of coupling between p and Pz,

p would develop spatial modulations along the z-axis

at wavenumber k2 = 2 7r/l where I is of the order of

a molecular length whereas Pz(r) would develop

modulations at wavenumber k1 = 2 7r//’ where l’ > 1 is a length associated with two molecules. To describe the appearance of modulated order, we write

where qll(r) and V12(r) are complex fields :

In terms of these fields, the Landau-Ginzburg free

energy of the Prost model in d-dimensions is

where rl = al(T - T1c) and r2 = a2(T - T2c) mea-

sure the temperature from the noninteracting mean

field transition temperatures T lc and T 2c of the fields 01 and t/12’ Ol is a derivative in the plane perpendi-

cular to the z-axis, parallel to n. The I (d + k2)01 12 and (d + ki) t/12 12 in equation (2.3) favour qp=kl and qp = k2 whereas the I V _Lol 12 and 1 V.L t/12 12

terms favour qp and qp parallel to the z-axis. The Re 02 t/1! term favours lockin at qp = 2 qp.

To study the mean field phase diagram of linearly

modulated phases, we choose qp and qp parallel to

n = êz and seek spatially independent fields

which minimize ð.Fs’ We expect the following phases :

1) the nematic phase (N) with I t/J 1 I = I t/J2 1 = 0,

(4)

2) the monolayer smectic phase (SAl) with I t/J 1 I = 0,

and

3) the bilayer anti-ferroelectric smectic phase (SA2)

with and

Note that an anti-ferroelectric smectic A phase

with no long range mass density modulation (I fj 1 I =1= 0, I fj 2 I = 0) is never stable (see Ref. [5]) because a non- zero I fj 1 1 always generates a non-zero I fj 2 1 via the

third order term in ð.Fs’ In addition" there are anti- phases, which will not concern us here, in which qp is not collinear with qp [2, 7].

The mean field free energy density of the N phase

is zero :

The free energy density of the SAI phase is obtained

by minimizing

with respect to 11/12 I to yield

The free energy density of the bilayer SA2 phase is

obtained by minimizing

with respect to I t/J 1 1, I t/J 2 1 and qo. The minimization with respect to qo is straightforward, and we obtain

To proceed with the algebraically complex minimiza-

tion with respect to [ gli [ and I t/J 2 I, we introduce

rescaled variables

so that

x, therefore, measures the degree of order and 0 the relative amplitude of t/J 1 and ql2 with small 0 cor-

responding to larger t/J l’ In terms of these variables,

the free energy density becomes

where

yl and Y2 are the temperature variables, and z2 is

the incommensurability parameter measuring the degree of mismatch between k1 and (1/2) k2. The

usual stability requirements on fourth order terms require ðUl,2 > - 1. We now write

Minimizing f with respect to x, we obtain

For a given 0, f is a minimum either at

(The smaller root x = [1/(2 a)] (- b corresponds to a local maximum of the free energy). We, therefore, obtain

Finally, we performed a numerical minimization of

equation (2.17b) with respect to 0 to locate the lowest free energy solutions. Plots of f (x(9), 0) vs. 0 for

different values of y, at constant y2 are shown in

figure 2. The various phase diagrams we obtained are

shown in figure 3 and will be discussed in the next section.

3. Phase diagrams.

For small incommensurability parameter Z2 and

symmetric elastic and fourth order terms (bul =ðU2)’

the phase diagram (Fig. 3a) looks like the theoretical N-SA1-SA2 diagram of reference [5]. There is a

second order N-SA 1 line terminating at a mean field

(5)

394

Fig. 2.

-

Free energy as a function of sin 0, (a) in the strongly

first order region and (b) in the vicinity of the critical point.

In each figure, there are three curves at constant y2, but

differing Y1’ In figure 2(a) the absolute minimum of curve 1

corresponds to the SA2’ phase and that of curve 3 to the SA2 phase. Curve 2 with two energetically equivalent minima corresponds to the phase boundary where the SA2 and SA2’ phases co-exist. In figure 2a, y1 = - 0.200, and Y2 =

-

0.569 x 10-2, - 1.069 x 10-2 and - 1.569 x 10-2 in

curves 1, 2 and 3. In figure 2b, Y1 = - 0.375, and Y2 = + 1.520 x 10- 3, - 3.480 x 10- 3 and - 8.480 x 10- 3 for

curves 1, 2 and 3.

critical end point [9] Q where the N, SA 1 and SA2 phases meet. There is a second order N-SA2 line

terminating at a tricritical point P. The line QP is

a first order N-SA2 line which continues into the smectic region as a first order SAI-SA2 line termi-

nating at a tricritical point R. Beyond R, there is a second order SAI-SA2 line that is expected to be in

the Ising universality class [5, 10]. The wavevector qo

depends on yl and Y2’

If the elastic constants D1 and D2 are different and favour an easy compression of t/11 (i.e., bul increases

and bu2 decreases), then, at small values of Z2 , a new

first order phase boundary separating two SA2 phases appears. The two SA2 phases are distinguished by different values of 0 and from equation (2.11)

different values of qo. For 0 small, qo is of order kl,

Fig. 3.

-

Phase diagrams in the Y 1 - Y 2 plane for different values of the incommensurability parameter z2 and the

potentials ðU1 and bu2. Solid (dashed) lines correspond to

first (second) order transitions.

a) z’ = 0.25, ðU1 = 0, 6U2 = 0. This diagram shows N,

SAl 1 and SA2 phases and is almost identical to the phase diagram for z2 = 0 calculated in reference [5].

b) z2 = 0.25, bul = 6, du2 = - 0.85 +. There are now two SA2 phases, characterized by different values of qo and

I t/J 1 II I t/J 2 1, separated by the first order co-existence line PC.

The point C is a critical point similar to the liquid gas critical

point and the point P is a mean field critical end point.

c) z2 = 0.45, 6ui = 6, 6U2 = - 0.85 +. A mean field bicri- tical point B and a triple point T appear in this diagram.

The first order SA2-SA2’ co-existence line and critical point

C remain.

(6)

and t/J 1 is much larger than t/J 2’ Thus a small 0 or

SA2’ phase can be identified with the experimentally

observed [3] SAd phase even though t/J 2 is strictly speaking non-zero. We shall strengthen this identi- fication in the next section when we consider x-ray

scattering. For 0 intermediate between 0 and 7r/2,

the amplitudes of t/J 1 and 4/2 are comparable leading

to the traditional SA2 phase. We will label the tra- ditional bilayer phase SA2 and the small 0, SAd-like phase SA2’. The dicontinuities in W02 and sin2 0 across

the new phase boundary are related from equa- tion (2 .10) via

The new phase boundary is tangent to the first order N-SA2 line at P (Fig. 3b) and terminates at a critical

point C where Oqo goes to zero. Critical fluctuations in the vicinity of C similar to those encountered at the

liquid gas critical point are expected. The presence of the critical point C indicates as expected that there is no

difference in symmetry between the SA2 and SA2’

phase with different relative amplitudes of 0, and t/J 2’

The experimental phase diagram [11] for the binary

mixture DB7-TBBA is very similar to the central part of figure 3b provided one identifies the small 0 SA2’

phase with the experimental SAd phase. As just discussed, this is a reasonable identification that will be further justified in the next section. In addition

the experimental phase diagrams for the DBn and

DBnO series [12] are similar to the region around P

in figure 3b provided one assumes that the points Q

and C are outside the range of experimental obser-

vation.

When the incommensurability parameter Z2 is further increased or bul increased and bu2 decreased,

a new first order SAl-SAd line (Fig. 3c) appears

terminating at a mean field bicritical point B where

the N, SAI and SAd phases meet, and a triple point T

where the SAI, SA2 and SAd phases co-exist The critical point C marking the end of the line distin-

guishing the SA2 and SAd phases continues to exists

The existence of the mean field bicritical point B is surprising in that it arises from a theory with a trilinear

t/Ji t/Ji coupling that normally leads to a first order transition as in figure 3b. In appendix A, we give an

alternate derivation of this result This phase dia-

gram in the vicinity of B is in good agreement with the experimental phase diagram [ 1 3] of the bindery mixture DB7-NO2-DB8NO2. Note, however, that the experi-

mental diagram does not show an SAd-SA2 transition.

Rather there is an SAd-C transition, the possibility of

which we do not consider in this paper. This is an

indication, that the new critical point C may be pre-

empted by non-collinear anti-phases. Finally, we note

that there is a small re-entrant nematic region in the experimental phase diagram which is probably due

to fluctuations beyond mean field theory not included

in the present theory [6, 7, 14].

Fig. 4.

-

qo as a function of y2 for different yi. Note the

similarity between this diagram and the PV liquid-vapour diagram.

In figure 4, we plot qo versus the temperature variable y2 for different yi in the vicinity of C. Note

the close resemblance of the behaviour of this variable to that of the liquid-gas density difference in the

vicinity of the liquid gas critical point.

4. x-ray scattering.

The most definitive experimental probe of the structure

of smectic liquid crystals is x-ray scattering. In this

section we will calculate the x-ray structure factors for the SA2 and SA2’ phases discussed in the previous

section by considering Gaussian fluctuations about

mean field theory.

The x-ray scattering intensity at wavenumber q is proportional to

where pel(r) is the electronic density which can be expressed as a linear combination of the smectic order parameters :

Thus, we have

with

for A, B = Pz, p. To discuss correlation functions of Pz(r) and p(r), we write

where t/11 ) = I t/11 I and t/12) = I t/12 I are the

mean field order parameters calculated in section 2.

(7)

396

The qJ-fields represent deviations of t/1I and t/12 from

their equilibrium values and by definition have

expectation value zero, (i.e. qJ Ix ) = 0, etc.). After

some simple algebra, we obtain

where

and

Equation (4. 6a) follows from the fact that fluctuations in Pz and p occur in different Brillouin zones.

The maximum intensities of the two Bragg peaks at wavevectors qo and 2 qo arising from the first terms in equation (4. 7) and (4. 8) are proportional respectively to ai V2 I t/11) 12 and CX2 2 V21 V’2 > t2 corresponding

to long range anti-ferroelectric and mass density ordering. Thermal averages of the diffuse scattering terms can

be calculated in the harmonic approximation as follows. Let 45. = ((Plxl (Ply, (P2x, (P2y) be the four component

vector describing fluctuations out of equilibrium. The free energy AF, is second order in ø (X is then

where ðFMF is the volume integral of the free energy density of equation (2.8). There is no linear term in ø since øa ) = 0 and ( VØa) = 0. The stability matrix Maa(q) is

Thermal averages of 0 .. (q) Øp( - q) are obtained by

inverting the matrix Map(q) :

can now be obtained from equations (4.6)-(4.8) and equation (4.12). Different profiles for I(q_L = 0, qz) are

shown in figure 5 for different points of interest in the

phase diagram. The finite scattering volume V,

as well as the coefficients al and a2 (whiqh are taken

to be equal for simplicity) are maintained constant in all profiles so that relative intensity domparisons

can be made among the different curves.

We note the following behaviour shown in figures 5a

to 5d.

The nematic phase exhibits two diffuse spots

(Fig. 5a) centred at the incommensurate wavenumbers

k1 and k2.

A strong quasi-Bragg peak condenses at wavevector

k2 = 2 qo in the SA1 phase while the intensity of the

diffuse scattering is lowered. The diffuse spot at wave

vector ki is still incommensurate with 2 qo as observed

experimentally [3, 15]. Note that figure 5b is in

agreement with other computations of x-ray diffrac- tion patterns [16].

A second quasi-Bragg peak appears in the bilayer

SA2 phase at wavevector qo(lj2) k2 qo kl) com-

(8)

Fig. 5.

-

Typical x-ray intensities /(0, 0, qz) (q2 scans) for

different temperatures. The other parameters are the same

as in figure 3c. Reduced co-ordinates are used as explained

in the text (Eq. (2.13)). Arbitrary units are used in the figures

but relative intensities can be compared provided one mul- tiplies intensities by factors of 10 and 5 in figures 5b and 5c.

a) N phase, the maxima of the diffuse spots are at the natural

(incommensurate) wavenumbers k1 and k2.

b) SA1 phase, there is a single quasi-Bragg peak at k2 and

a diffuse maximum at k 1.

c) SA2 phase near the SA1-SA2 phase boundary for large 0,

there are two commensurate quasi-Bragg peaks at qo and 2 qo with qo k1 and 2 qo very slightly greater than k2

so that kl 1 > qo > k2/2 qo.

d) SA2 phase for small 0, there are two quasi-Bragg peaks

at qo - k1 and 2 qo and a large diffuse scattering peak at

qz ’" k2. Note that the Bragg peak at 2 qo is almost entirely

hidden by the diffuse scattering peak. The intensity of the Bragg peak grows as V2 and should always dominate

diffuse peaks for large enough scattering volume V. V is, however, always bounded by mosaicity and other experi-

mental limitations.

mensurate with that of the monolayer order of 2 qo

(Fig. 5c). The intensity of this new peak at (0, 0, qo)

grows continuously from zero away from the SA 1- SA2 phase boundary when this transition is second order [10] whereas the intensity of the (0, 0, 2 qo) peak does not change significantly. Deep in the SA2 phase the intensities of the (0, 0, qo) and (0, 0, 2 qo) peaks are comparable. Note that there remains a

small diffuse scattering amplitude (Fig. 5c) centred

around the wavevector k2 incommensurate with the established order at qo and 2 qo.

Although the scattering profiles of the SA2 and

SA2’ (SAd or small angle SA2) phases are qualitatively equivalent, they are quantitatively distinct. As already discussed, the intensity of the (0, 0, qo) peak in the

SA2’ phase is greater than that of the (0, 0, 2 qo) peak

whereas the converse is generally true of the SA2

phase. This difference is particularly visible as the

first order SA2-SA2’ line is crossed in the vicinity of

the triple or pseudo-triple points T and P in figures 3b

and 3c. In fact, the intensity into the (0, 0, 2 qo) peak

goes to zero as the second order N-SA2’ phase boun- dary is approached in these figures. In the SA2 phase,

the intensity of the (0, 0, 2 qo) peak is significantly

greater than that of the (0, 0, qo) peak (Fig. 5c). In the

SA2’ phase at the order side of the boundary (Fig. 5d)

the intensity of the (0, 0, 2 qo) peak is strongly lowered

and almost completely overshadowed by an intense

diffuse spot centred around (0, 0, k2) and the first order peak at (0, 0, qo) is dominant with the lockin wavevector qo close to k1 (Fig. 5d). If the (0, 0, 2 qo) peak were completely hidden by the diffuse spot at (0, 0, k2), in an experiment of standard resolution,

then the profile of figure 5d would be identical to the

experimental SAd profile of reference [15]. The signa-

ture of the SA2-SA2’ (SA2-SAd) transition is thus

a discontinuous jump from an x-ray pattern with

a dominant Bragg peaks at wavevector (0, 0, qo - k2/2) and (0, 0, 2 qo) with diffuse scattering at k1

to one with a Bragg peak at (0, 0, qo - kl) with

diffuse scattering at k2 as is commonly observed [15].

Recently the weak (0, 0, 2 qo) Bragg peak has been

observed in high resolution x-ray experiments on

DB6 [17] in excellent agreement with the present

theory. As the critical point C is approached, the

intensities of the two Bragg peaks in each phase

become more nearly equal. Finally at the point C, the distinction between the SA2 and SA2’ phases disappears. Stated differently, the classical SAd phase changes continuously into the bilayer SA2 phase as C

is approached. The difference between the lockin wavevector qo of the SA2 and SA2’ phases also approaches zero as C is approached.

5. Discussion.

In this paper, we have used the Prost model to study phase transitions in frustrated smectics in mean field

theory. We find nematics (N), monolayer (SA1) and

two bilayer phases (SA2) and (SA2’). The general types of mean field phase diagrams are shown in figure 3.

Of particular interest are figures 3b nd 3c where

there appears a new critical point te inating a first

order phase boundary separating the A2 and SA2’

phases. We identify the SA2’ phase with the experi-

mental SAd phase. With this identification, the

calculated phase diagrams of figure 3 and are in good agreement with the experimental phase diagrams of

references [ 11-13].

This paper was devoted exclusively to mean field theory. There remain the questions of whether critical fluctuations will lead to qualitative changes in any of the mean field phase diagrams and what universality

class any surviving critical points will fall into. The

critical point C would appear to have Ising symmetry

in analogy with the liquid-gas transition. Coupling

(9)

398

of the order parameter to elastic degrees of freedom could, however, lead to a different universality class

and is currently being investigated. The tricritical

point R and critical end points Q and P probably will

survive the inclusion of fluctuation. The situation for the bicritical point B is, however, less clear because of the third order coupling between 01 and 02, which

would normally drive a transition first order. It seems

plausible, however, that the incommensurability which

makes the mean field bicritical point possible will also, make the bicritical point possible in a complete theory. However, even if the bicritical point survives,

its universality class is uncertain. The second order N-SA 1 and N-SA2’ lines are believed to be in the inverted xy universality class [18] with negative specific heat exponent a. Normally when two second

order lines with negative a meet, the resulting multi-

critical point is tetracritical [9] rather than bicri- tical. Thus, either the mean field bicritical point

becomes a tetracritical point when fluctuations are

included, or it is in a new universality class. These

points are under current investigation.

Appendix.

In this appendix, we will review why a t/Jî t/J! coupling usually prevents t/J 1 and 02 from being simultaneously

critical and show how the incommensurability of the present model allows them to be. Consider first fSA2

in equation (2. 8) with k21 = qo and k2 = 4 q2

When r 1 r2l I t/J 1 1 is critical before I t/J 2 1, and we

can seek an effective free energy in terms of I t/J 1 I only.

Minimizing fsA2 with respect to 1 t/J 2 I, we find

Inserting this into equation (A. 1) we find

The coefficient ueff = u1 - W2/r2 of I ol 14 becomes negative for small r2 leading to a first order transition.

Thus, a second order transition with rl 1 = r2 is

impossible. The first order lines QP in figures 3a

and 3b are a result of this effect.

Now consider the full free energy equations (2.14)

and (2.15) in terms of the reduced variables. This free energy has already been minimized with respect to qo. Here we minimize first with respect to 0 rather

than with respect to x. Seeting q = sin 0, we find

where

Thus f is a minimum either for cos 0 = 0 (i.e. t/J 1 = 0)

or for g(q) = 0. The former case corresponds to the

N-SAI transition with ordered phase free energy

g(n) is a cubic polynomial in q so there are three

solutions to g(n) = 0. When x = 0, these are trivially

For x small, there are corrections to the above solu- tions that can be expressed as a power series in x.

The first solution (Eq. A. 7a) becomes

Leading to an effective free energy

This predicts a second order transition at yl = 0 if 1 + bul > (z’ + y2)-1 with ordered phase free energy

The SA2’ subscript indicates that the ordered phase

is an SA2’ phase with 1 2 1 0 1 1 0 0’ We have

thus established that there are two second order extrema leading to N-SA 1 and N-SA2’ transitions

meeting at a multi-critical point at yl = Y2 = 0. If this multi-critical point is not eliminated by the

alternate solution to f(~) = 0 (Eq. A. 6b), then there

will be a first order transition for Yl 0, y2 0 between the SAI and SA2’ phases when fSAl = ,fSA2’

or along the line

for small I y, I and y2 I. This gives a topology in

agreement with that in the vicinity of B in figure 3c.

(10)

We now consider the alternate solution to f( q). It

leads to the existence of a first order N-SA2 transition line, which can be calculated in closed and exact form

by using the fact that the f(x, 0) = 0 equation has a

double root in x other than zero. The parametric representation of this phase boundary reads :

Varying 0 in the [0, n/2] interval defines the curve, which has a physical meaning only as long as y, and Y2 are positive. Let us first remark that for vanishingly

small incommensurability (Z2 _ 0), yl is always positive so that this first order N-SA2 transition

occurs before the second order N-SA2 and N-SA2’

transitions can be reached Increasing Z2 shifts the

curve toward negative values of both y1 and Y2’

It will cross the bicritical point when equation (A. .12),

admits the yl = Y2 = 0 solution, in a physically acceptable domain. This requires :

for z’ zf, the situation is that of figures 3a and 3b,

as just discussed. For Z2 > Z2c, the first order line no

longer passes through the first quadrant; the alternate solutions to f(~) take over leading to the second order N-SA2 and N-SA2’ and first order SA2-SA2’ transi- tions meeting in a bicritical point at B shown in figure 3c. The existence of this point requires further 6U2 > - 1 (stability of the fourth order term) and bul > 1 (cos2 e 1).

Acknowledgments.

One of us (T.C.L.) received partial support for this work from the National Science Foundation under grant no. DMR82-19216. J. P. is also grateful to NSF

under the above contract for partial support for a visit

to the University of Pennsylvania where this work

was initiated. P. B. acknowledges the assistance of P.

Richetti in computer calculations.

References

[1] SIGAUD, G., HARDOUIN, F., ACHARD, M. F., Phys. Lett.

72A (1979) 24.

[2] For a review, see HARDOUIN, F., LEVELUT, A. M., ACHARD, M. F., SIGAUD, G., J. Chim. Phys. 80 (1983) 53.

[3] HARDOUIN, F., LEVELUT, A. M., SIGAUD, G., J. Physi-

que 42 (1981) 71.

[4] BAROIS, P., COULON, C., PROST, J., J. Physique Lett.

42 (1981) L-107.

[5] PROST, J., J. Physique 40 (1979) 581.

[6] PROST, J., Proceedings of a Conference on Liq. Crystals

of One and Two Dimensional Order-Garmisch Partenkirchen (Springer Verlag, Berlin, Heidel- berg, New York) 1980, p. 125.

[7] PROST, J. and BAROIS, P., J. Chim. Phys. 80 (1983) 65.

[8] HARDOUIN, F., ACHARD, M. F., DESTRADE, C., NGUYEN

HUU TINN, J. Physique 45 (1984) 765.

[9] See, for example, ZIMAN, T. A. L., AMIT, D. J., GRINS-

TEIN and JAYAPRAKASH, C., Phys. Rev. B 25 (1982) 319.

[10] WANG JIANG and LUBENSKY, T. C., Phys. Rev. A 29 (1984) 2210.

[11] Reference [2], figure 16.

[12] Reference [2], figure 2.

[13] Reference [2], figure 22.

[14] COULON, C. and PROST, J., J. Physique Lett. 42 (1981)

L-241.

[15] LEVELUT, A. M., TARENTO, R. J., HARDOUIN, F., ACHARD, M. F. and SIGAUD, G., Phys. Rev. A 24 (1981) 2180.

[16] WANG, J. and LUBENSKY, T. C., to be published in

J. Physique.

[17] KELBY CHAN, SORENSEN, P. and PERSHAN, P. (unpu- blished) ;

KELBY CHAN (unpublished thesis, Harvard University).

[18] CHANDAN DASGUPTA and HALPERIN, B. F., Phys. Rev.

Lett. 47 (1981) 1556;

TONER, J., Phys. Rev. B 26 (1982) 462 ; see also LUBENSKY, T. C., J. Chim. Phys. 80 (1983) 31.

[19] FISHER, M. E. and NELSON, D. R., Phys. Rev. Lett.

32 (1974) 1350 ;

KOSTERLITZ, J. M., NELSON, D. R. and FISHER, M. E.,

Phys. Rev. B 13 (1976) 412.

Références

Documents relatifs

structure in odd-membered ones was identified as a new type of smectic phase, smectic CA [8], in which the tilt direction of mesogenic groups is invariable in every second

2014 A theory of static deformations of smectics A is presented, taking into account the large curvature deformations of the layers, but assuming that the variation in

2014 The existence of ferroelectricity in the chiral smectic C phase has been established from polarization measurements in the presence of an electric field [1].. In

- Proton magnetic resonance experiments [1] parallel and perpendicular self-diffu- sion coefficients were measured in smectics A-TBBA and EBBA (D~~ is related to

Abstract. frustration in a restricted number of dimensions ; they are investigated in a typical model, the stacked triangular antiferromagnetic Ising model. For

- Phase diagrams of binary systems in which a significant lowering of the smectic phase stability is observed : (a) and (b) mixture of compounds yielding smectic

We have also observed important fluctuations at room tempe- rature, in the form of diffuse spots of wave-vector qo = 0.18 C*, which may be associated with a

Abstract 2014 The effect is tested of the smectic layer spacing ratio, r, on the phase diagram for the binary systems consisting of 80CB (smectic Ad) and one of