• Aucun résultat trouvé

Ultrasonic study of the nematic-isotropic phase transition in PAA

N/A
N/A
Protected

Academic year: 2021

Partager "Ultrasonic study of the nematic-isotropic phase transition in PAA"

Copied!
10
0
0

Texte intégral

(1)

HAL Id: jpa-00209164

https://hal.archives-ouvertes.fr/jpa-00209164

Submitted on 1 Jan 1979

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.

Ultrasonic study of the nematic-isotropic phase transition in PAA

Y. Thiriet, P. Martinoty

To cite this version:

Y. Thiriet, P. Martinoty. Ultrasonic study of the nematic-isotropic phase transition in PAA. Journal

de Physique, 1979, 40 (8), pp.789-797. �10.1051/jphys:01979004008078900�. �jpa-00209164�

(2)

Ultrasonic study of the nematic-isotropic phase transition in PAA (*)

Y. Thiriet and P. Martinoty

Laboratoire d’Acoustique Moléculaire (**), Université Louis-Pasteur, 4,

rue

Blaise-Pascal, Strasbourg, France (Reçu le Il décembre 1978, révisé le 19 avril 1979, accepté le 26 avril 1979)

Résumé.

2014

Nous

avons

étudié la variation thermique de l’absorption et de la vitesse ultrasonore dans

un

échantil- lon de para-azoxyanisole (PAA) orienté par

un

champ magnétique, pour des fréquences comprises entre 0,8 et

5 MHz. Dans le domaine des températures étudiées,

nos

mesures correspondent

au

régime

03C903C4

~ 1 où 03C4 est le temps de relaxation acoustique. Du côté isotrope de la transition,

nos

résultats peuvent être interprétés par la théorie de la chaleur spécifique dynamique qui prédit que le coefficient d’absorption 03B1(T) diverge

avec un

exposant 1,5. Du côté nématique, 039403B1(T), l’anisotropie ultrasonore et a(T) divergent avec

un

exposant 1. Ce résultat semble montrer que la partie critique de l’absorption résulte de la relaxation du paramètre d’ordre lui-même (mécanisme de Landau- Khalatnikov) et que les fluctuations jouent dans cette phase

un

rôle négligeable, tout

au

moins dans le domaine des températures étudiées (de 1 °C à 20 °C de Tc). Nous comparons

ces

résultats à ceux que nous avions obtenus antérieurement dans le p-n-pentyl p’-cyanobiphenyle (PCB).

Abstract.

2014

We present

an

ultrasonic investigation of the nematic-isotropic phase transition in p-azoxyanisole (PAA). Our measurements were performed at several frequencies ranging from 0.8 to 5 MHz as

a

function of temperature and, in the nematic phase,

as a

function of the orientation of the liquid crystal with respect to the ultrasonic wave vector. In both phases, our results are within the

03C903C4

~ 1 regime where 03C4 is the acoustical relaxation time. In the isotropic phase, the results may be quantitatively interpreted using the dynamic heat capacity theory

which predicts for 03B1, the ultrasonic absorption,

a

critical exponent of 1.5. In the nematic phase, we find

a

critical exponent of ~ 1 for 039403B1(T), the attenuation anisotropy, and 03B1(T). This result

seems

to show that the Landau- Khalatnikov mechanism is the dominant contribution in the temperature range investigated which corresponds

to Tc2014 T values from 1 °C to 20 °C. For comparison purposes

we

also include some data for PCB that

we

published

some

time ago.

Classification

Physics Abstracts

61.30 - 64.70E - 62.80

1. Introduction.

-

Orientational order fluctuations

near the nematic-isotropic phase transition have been

widely studied by ultrasonic absorption, and the

occurrence of a pronounced maximum in the atte- nuation and a minimum in the velocity is well

known [1-6]. However, some questions related to the temperature dependence of the attenuation para- meters are still unresolved on either side of the transi- tion. The situation is particularly complex on the

nematic side, where unexpected exponents of 0.4 to 0.5 have been found for the absorption coefficient

a( T) and for the relaxation frequency r-1(T) [2, 5].

However, the experiments were performed on

MBBA [2-4] and PCB [5], compounds which present an

(*) Presented at the 7th International Liquid Crystal Conference Bordeaux, July 1-5 1978.

(**) (E.R.A.

au

C.N.R.S.).

intramolecular relaxation (in the same frequency

range as the critical relaxation) that can affect the critical parameters. Furthermore, most of the measu-

rements were made in unaligned samples or in a fre-

quency range not wide enough to obtain accurate

relaxational data. On the isotropic side of the transi-

tion, the situation is also confusing, since exponents of 1 [2] and 1.5 [3, 5] for a(T) have been reported.

The purpose of the present work is to make a detailed study of the nematic-isotropic phase transi-

tion in p-azoxyanisole (PAA), a compound in which

no rotational isomerism can occur in the end groups.

Therefore, in spite of the experimental problems

associated with the high transition temperature, PAA appears to be a promising compound for such a study. Another advantage is the low value of its shear

viscosity, which leads to a very small background absorption.

Our measurements were performed at frequencies

from 0.8 to 5 MHz as a function of temperature and,

Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphys:01979004008078900

(3)

790

in the nematic phase, as a function of the orientation, 0, relative to q, the ultrasonic wave vector. In the temperature range investigated, the data are within

the wi 1 regime.

In the isotropic phase, our results may be quantita- tively interpreted using the dynamic heat capacity theory [7], which predicts for a(T) a critical exponent of 1.5. In the nematic phase, we find a critical expo- nent of - 1 for both oc(T) and Aot(T). This result seems to show that, in the temperature range investigated,

which corresponds to Tc - T values from 20,DC to 1 OC, the dominant contribution to the sound absorp-

tion arises from the Landau-Khalatnikov mecha- nism [8].

The plan of thé paper is as follows : In section 2,

we describe our experimental technique. The theore- tical background is reviewed in section 3. Our results

are presented in section 4 and compared with those

we obtained earlier in PCB.[5], and are analysed in

section 5.

2. Experimental.

-

The ultrasonic measurements were made in the frequency range 0.8-5 MHz, using the

acoustic resonator. This technique, which employs standing sound waves in a cylindrical cavity, allows

the simultaneous determination ofvelocity and absorp-

tion values with high precision from the frequency position and from the 3-dB bandwith of the resonance

peaks of the cavity. On the other hand, this technique requires only small liquid samples of about 3 ml.

A detailed description of the apparatus and of the experimental procedure has been given in ref. [9], so

we shall not discuss them further. However, because of the problems associated with the high transition temperature, we have made a more elaborate version of our previous cell. A sketch of this new cell is

given in figure 1. The temperature during the experi-

ments was regulated to within ± 0.02 OC by oil circulating from a constant temperature bath through

the annular space of the double-walled cell. Tempera-

ture fluctuations in the cell were controlled by measur- ing the position of a resonance peak over a long period of time. One sensitive way to do this is to keep

the frequency of one 3-dB point of a resonance peak

constant and to observe the relative variations in the output amplitude, which are related to the tempera-

ture variations. The quartz transducers are 3 MHz, X-cut, optically polished plates, 30 mm in diameter.

The transducer spacing was 2.04 mm. With this value,

our cell shows resonances at approximately 330 kHz

intervals.

PAA, which is known to be more stable than a

Schiff’ s base, such as MBBA, was obtained from

Merck, and used without further purification. The

transition temperature Tc was 135 °C. To prevent oxidation, the sample inside the resonant cavity had

no contact with the atmosphere-i.e., the sample completely filled the cavity. The transition tempera-

ultrasonic resonator cell

Fig. 1.

-

High-temperature ultrasonic resonator. The following

elements

are

indicated : (1) viton rings and metal rings pressing the

viton rings against the X-cut quartz 61 and Q2 ; (2) parallelism adjustment viton ring ; (3) filling hole ; (4) parallelism adjustment blocks ; (5) annular spaces for thermostat liquid ; (6) thermostat mantles ; (7) BNC connectors with contact wires ; (8) threaded ring to fix

one

half of the cell to the holder ; (9) holder ; (10) adjust-

ment screws; (11) parallelism adjustment

screws.

ture and the quality of the compound were verified by differential thermal analysis before and after the

experiment and a slight shift, of 0.1 OC, was found

in Tc. Because of the existence of a two-phase region,

we have not analysed the temperatures closest to T,.

For the other temperatures, our data plotted as a

function of ! T - rj 1 are not very sensitive to the shift in Tc.

To protect ourselves from hysteresis effects, we performed the experiments by heating the sample.

The sample was aligned by a magnetic field of 10 kG,

and for each temperature, the measurements were

made for e equal to 00, 45°, and 90°. With increasing temperature, the attenuation peaks became broader

(as a result of the increase of the ultrasonic absorption) and, in the 1 °C around Tc, those for frequencies higher than 2 MHz disappeared. The transition tem-

perature Tc was defined, for our measurements, as the

highest temperature at which a non-zero value of Da

was obtained.

Near Tc there was an intermediate temperature

range of - 0.3 OC to 0.4 OC in which a typical reso-

nance peak became a double-peak. With a slight

increase in the temperature, the magnitude of one of

(4)

the peaks increased while the other vanished. Since the frequency shift between the two peaks corresponds approximately to the difference in the velocity between

the nematic phase and the isotropic phase, we believe

that this effect was due to gravitational separation

into a two-phase region separated by a horizontal

interface (1). No quantitative measurements were

made in this region. The assumption of a two-phase region is also supported by the fact that a change in

one of the peaks occurred when the magnetic field was rotated, whereas the other peak was unchanged.

Such a coexistence region is presumably due to small

amounts of impurity which must be difficult to elimi- nate, since results in MBBA have shown that the two

phases coexist even after repeated distillations [10].

3. Theoretical background.

-

3.1 ISOTROPIC

PHASE.

-

The anomalous behaviour of the ultrasonic

absorption and velocity on the isotropic side of the N-I transition have been interpreted by Imura and

Okano [7] in terms of a frequency-dependent specific

heat on the basis of de Gennes’ statistical continuum

theory [11]. This approach considers the interaction of the temperature variation of the sound wave with the thermal fluctuations of the tensor order parameter

Qrz/l’ which are described by a correlation function

here, k is Boltzmann’s constant, q is the wave number, and A and L are coefficients of the Landau expansion

of the free energy. The temperature dependence of A

is assumed to be given by A(T)

=

a(T - Ti) where Ti is the virtual second-order transition temperature.

The presence of the sound wave induces periodic changes in A(T) and G(q). Near the transition, the fluctuations of the order parameter have a strong spatial correlation, and G(q) cannot follow the tempe-

rature variations induced by the sound wave. This phase-lag produces a frequency-dependent heat capa-

city. Since the sound velocity depends on the specific

heat ratio, one obtains a complex frequency-depen-

dent sound velocity whose imaginary part gives rise

to the sound absorption. According to this theory,

the absorption per wavelength aÂ, the ultrasonic

absorption aJf 2, and the velocity V are written as

follows (’) :

(1) Two-phase separation effects have also been reported in

ref. [6].

e) In the derivation of eq. (2), the temperature and the frequency dependences of the ultrasonic

wave

velocity

are

assumed to be negligible.

where

cp and are the heat capacity at constant pressure and constant volume, respectively, in the absence of the

fluctuations of the order parameter, and Acp is the

excess specific heat due to the fluctuations.

fi(x) and f2(x) are the functions which provide

the theoretical curves for the frequency dispersion :

where

cvo

=

A(T)IM is the relaxation frequency of the longest wavelength mode (q

=

0) of G(q) and u is the

transport coefficient appearing in the relaxation

equation of G(q). Since y is regular at the transition Wo goes to zero as T - Tc*.

Eqs. (1)-(3) are valid for x > 1 (i.e., w « wo). For

x « 1 (w > wo), where fluctuations with wave num- bers much greater than ,- 1 are predominant, one expects a breakdown in the theory due to the inade-

quacy of the Omstein-Zemike form for G(q) [12].

In the low-frequency limit (x » 1), eq. (2) reduces

to a simple relaxation law of the form :

where r -1 is the acoustical relaxation frequency.

This quantity is related to the relaxation frequency coo of the longest wavelength mode of G(q) by :

Since Acp -(T- T:)-0.5 and Wo "-1 (T - Tc*),

it follows that rx/f2 "-1 (T - T:)-1.5. On the other hand, the relaxation frequency of the longest wave- length mode of Qas is wm

=

A(T)Iil, where Pl is the

transport coefficient appearing in the relaxation

equation for QaP. To a first approximation y - yy/2,

and therefore c-’ - 8 im 1.

Recently, another formulation has been proposed by Matsushita [13] using Mori’s statistical mecha- nical theory of sound attenuation and applying

Kawasaki’s mode-coupling theory to the order para- meter correlation functions. This approach leads to

temperature and frequency dependences of the sound

attenuation and of the velocity which are essentially equivalent to those obtained by Imura and Okano.

3.2 THE NEMATIC PHASE.

-

Two different critical contributions are expected in the nematic phase : one

from the relaxation of the fluctuations of the order

parameter (as discussed above) (this is symmetric

(5)

792

with respect to Tc) and another from the relaxation of S, the mean value of the order parameter (this is

due to the Landau-Khalatnikov mechanism). In the following, we are only concemed with the relaxation of S.

The hydrodynamic theory of nematics gives the following for the attenuation [14]

where 0 is the angle between the ultrasonic wave

vector and the director and V is the velocity. Vl’ V2, and V3 are friction coefficients and v4 - V2 and vs

are volume viscosities.

It follows that the anisotropy in rx/.f2, i.e., the

difference between the 00 and 900 values, is given by :

As shown by Jâhnig [15, 16], the hydrodynamic theory can be generalized to extend outside the

hydrodynamic frequency regime by retaining the

structure of the hydrodynamic equations and intro- ducing a frequency dependence of the elastic and

dissipative parameters of the system.

Because of the anisotropic properties of the elastic tensor, the strength of the coupling between the mechanical variables and the order parameter S

depends on the different tensor components. As a consequence, the relaxation of S appears in the attenuation anisotropy. Assuming that the volume viscosities are the only relaxing quantities, Jâhnig

showed that :

where â.E1 and AE2 are certain elastic parameters.

For co-r. « 1, eq. (9) reduces to :

which is eq. (8) with V4

=

’tm I1E1 and vs

=

’tm M2

and v 1

=

0.

According to mean-field theory, Tm-1 (Tl - T).

In fact, a calculation by Kawamura et al. [3] based on

the mean-field theory but modified for the weak first- order nature of the transitions predicts for alf 2 and im 1 an essentially temperature independent behaviour

in a narrow range near Tc, changing to a power law

(Tl - T) at temperatures farther below Tc. The temperature Tl is defined by

A coupling between the sound waves and the fluc- tuations of the director via Frank’s elastic constants has been proposed by Nagai et al. [5], extending

Imura and Okano’s theory for the isotropic phase

to the nematic phase. The theory of Nagai et al.,

which does not explain our results (see section 4),

will not be reviewed here. However, it should be noted that an error in their calculation of the complex specific heat suggested an anisotropy in the critical

damping. The exact calculation shows that the contri- bution arising from the fluctuations of the director is in fact isotropic.

4. Results.

-

Figure 2 shows typical data of a/12, the ultrasonic absorption as a function of f, the frequency. The separate curves are for temperatures which differ by the amounts indicated from the

transition temperature Tc. The figure shows that the values of a/f2 are frequency independent in the tem- perature range investigated. Therefore, these values

are those for the low-frequency limit wr « 1.

Fig. 2.

-

The absorption coefficient divided by the square of the

frequency,

as a

function of frequency. Individual

curves are

for various temperatures above and below the transition. The data indicate that rx/12 is frequency independent

over

the range of frequency investigated in contrast to the results for PCB which

are

shown in the insert for comparison.

In contrast, nematics at room temperature, like MBBA or PCB, show for the isotropic phase a strong dispersion in the same frequency range (see for example

the insert), which shows that the frequency relaxation

of the order parameter for these compounds is lower

than that for PAA. This difference is easily explained

since the relaxation frequency is inversely propor- tional to the shear viscosity and the latter is smaller for PAA than for PCB.

Figure 3 shows the sharp maximum of the ultra- sonic absorption at the transition. In the nematic

phase, the data are for 0

=

90°. They are in good

agreement with the limited attenuation results (the

(6)

Fig. 3.

-

The attenuation peak in PAA. Below Tc the data

are

for 0

=

90°. The results for PCB at 0.5 MHz

are

shown for compa- rison.

triangles) of Kempf and Letcher [17]. For comparison,

we have also reported, in the same figure, the PCB

data at a frequency of 0.5 MHz. Although the results for the two compounds are qualitatively similar, they differ quantitatively, as follows.

In the isotropic phase, the slow relaxation of the order parameter in PCB leads to an increase of the ultrasonic absorption, which is therefore larger than

that in PAA. However, within a given range of T - Tc,

the ultrasonic absorption increases by the same factor

for PAA and PCB. Thus, the temperature dependence

of the ultrasonic absorption obeys a law which should be identical for the two compounds.

In the nematic phase, the contribution from rota- tional isomerism in the end chain of the PCB moleçule

causes the ultrasonic absorption to be larger with this compound and the critical increase to be less sharp

than in PAA. Moreover, the ultrasonic absorption in

PCB is so high at the T farthest below Tc that there is

probably another relaxation process. In fact, it has

been recently shown [18] that PCB is a rather peculiar compound which in the nematic phase has a pro- nounced local order of the smectic type, which could contribute to the attenuation.

Figure 4 shows the temperature dependence of the

sound velocity. Within our resolution, no frequency dispersion was observed between 1 and 5 MHz.

Given the results of the absorption measurements, these values of the velocity are those at zero frequency.

Fig. 4.

-

Temperature dependence of the sound velocity. Between

1 and 5 MHz

no

angular dependence and

no

frequency dispersion

were

observed within

our

resolution.

5. Comparison with theory and discussion. - 5.1 THE ISOTROPIC PHASE.

-

To compare the data with the theory, one must account for the second- order transition temperature Ti and substract the contribution of the shear viscosity and of the non-

relaxing volume viscosity. The contribution of the shear viscosity to rx,f¡2 is about - 20 x 10-17 cm-1 S2 without significant temperature change. In this esti- mate, we used the capillary measurements of ref. [19].

For the non-relaxing volume viscosity ’1v, we have

assumed ’1v ’" 4/3 ’1s, as for a conventional liquid.

Fig. 5.

-

Temperature dependence of the ultrasonic absorption

in the isotropic phase of PAA. The solid line has

a

slope of 1.5.

The result for PCB is shown for comparison (from ref. [5]).

(7)

794

Anticipating a power law dependence on (T - Tc*)

we made a log-log plot of the î.l/f2 values obtained

versus T - Tc*, varying Tc* within a reasonable range.

With Tc*

=

Tr one obtains a distinctly bent curve to

which one cannot fit a straight line passing through

the points. Taking Tc - Ty

=

3 OC, one obtains a

curve with the opposite bend. For Tc - Tc* ~ 1.1 °C,

one obtains the 1.52 power law shown in figure 5, which confirms the 1.5 power law that we obtained earlier for PCB [5] at 0.5 MHz for T > Tc + 2 °C.

Using the result of the fit in the nematic phase (see

next section) we find Tl - Tc* "-1 1 °C. Therefore Tl

is about 0.1 °C above Tc while T* is about 1.1 °C

below Tr.

According to eq. (5) we may deduce the ratio of the relaxation frequencies from attenuation measurements in PAA and PCB provided the ratio Acplc’ is known.

Calorimetric measurements showed that àcplco is of

the same order of magnitude for the two compounds,

and from the data of figure 5 we find that

Since the temperature variation of -rPC1B is known

we may estimate

and therefore at

This high value of the relaxation frequency explains

that our values of aJf2 are frequency independent

between 1 and 5 MHz in the temperature range

investigated. However, since to deduce -r;¿ we have

used the results of experiments from many sources, the above value must be considered only an estimate.

We tum now to the relationship between the relaxa- tion frequency of the order parameter and the relaxa- tion frequency which is measured by ultrasonic expe- riments. Both quantities have been measured for PCB and are reported in figure 6. Curve (a) shows

the temperature dependence of the acoustical relaxa- tion frequency -r-t (from ref. [5]) and curve (b) shows

that of the relaxation frequency of the order para- meter im 1 (from ref. [20]). It is clear that the acoustical relaxation frequency is one order of magnitude higher

than the relaxation frequency of the order parameter.

A similar observation can be made for MBBA [2].

From measurements shown in figure 6 we find

which is consistent with the theoretical expectation,

and therefore y - il/2. The same conclusion was

obtained at the A-N phase transition in CBOOA [9]

Fig. 6.

-

(a) Temperature dependence of the acoustical relaxa- tion frequency -r-l/2

7r

for PCB (from ref. [5]). (b) Temperature dependence of the relaxation frequency of the order parameter

T.’/2 n for PCB (from ref. [23]). The data indicate thatr-’ - 8 -r;

1

(see text).

and appears to be a general feature of phase transitions in liquid crystals.

Finally, from the data in figure 4 we estimate the

velocity dispersion, v(oo) - V(0)

=

180 m. In this estimate, V(oo) was obtained by extrapolating the velocity in the isotropic phase to Tc. Thus

and, using eq. (3), àcpIco _ 1.5. This value is consis- tent with that found by differential scanning calori-

metry (D.S.C.), àcplco - 2.2, considering that our

estimate of V(oo) - V(O) is an underestimate and that D.S.C. gives only an order of magnitude.

5.2 THE NEMATIC PHASE.

-

Figure 7 shows the temperature dependence of the critical increase of the volume viscosities V4 - V2 and vs. These viscosity

coefficients were determined using the following

equations :

(8)

Fig. 7.

-

Our best estimate of the temperature dependence of the

critical part of the volume viscosities V4 - V2 and vs. The solid

straight line has

a

slope of 1.05.

from which the non-critical absorption must be

removed.

To this end, we took for v2 and V3 the values in ref. [14] and, assuming as for a conventional liquid

that the residual part of the volume viscosities v4 and v5 are of the same order of magnitude as the shear viscosity, we evaluated the non-critical absorption as

-

70 x 10-" cm-1 s2. Using this value we obtained the straight lines shown in figure 7, with a slope of

1.05 ± 0.05 and with uncertainties determined by the x2 test. In fact, the non-critical absorption is so small compared to the critical absorption that a log-log plot of the raw values of (11) and (12) appears linear

with a slope of - 1. As mentioned previously, two

contributions to the critical sound attenuation are

Fig. 8.

-

Log-log plot of the attenuation anisotropy Aa in the

nematic phase of PAA. Aa - (v5 - v4 + vl). The upper

curve

is the

raw

data. The lower

curve

with

a

slope of N 1 corresponds to the

critical part and the contribution of the coefficient vl has been removed.

expected in the nematic phase, one from the relaxa- tion of the order parameter fluctuations and another from the relaxation of the mean value of the order parameter (the Landau-Khalatnikov mechanism).

These two contributions do not have the same theo- retical temperature dependence. Since the critical exponent is 1.5 for the fluctuation mechanism and 1 for the Landau-Khalatnikov mechanism, the straight

line with a slope of 1.05 shown in figure 7 seems to

prove that the Landau-Khalatnikov mechanism is the dominant contribution in the temperature range

investigated, that is, Tc - T Jit 1 °C.

We tum now to the temperature dependence of Aa/f2 , the attenuation anisotropy, which has the form given by eq. (8). Note that this quantity does not

contain the contribution of the specific heat which, according to eq. (5), is isotropic. To obtain the critical part of Aalf ’, one must substract the shear viscosity

vi. In principle, this coefficient may be deduced from the effective shear viscosity vl + v2 - 2 V3, which is given by the following combination :

However, this shear contribution is so small com-

pared to the critical attenuation that it can be deduced

only for the T farthest below T., say, Tc - T > 15 °C.

In this temperature range, Vt + v2 - 2 V3 ~ 0.18 P.

Taking v2

=

0.034 P and V3

=

0.024 P [14], we may estimate v1. We find v1 ~ 0.21 P. Assuming for Vl

the same temperature dependence as the capillary viscosity and assuming vl of the same order of magni-

tude as v5, we obtain from our measurements of the attenuation anisotropy the linear variation of (vs - v4)

with a slope of 1, shown in figure 8. Although the experimental uncertainties are large for the T farthest

from Tc, the result shows the effectiveness of the Landau-Khalatnikov mechanism. According to this mechanism, the position of the attenuation peak

occurs at cor.

=

1, and it shifts to lower temperature with increasing frequency. From the relative position

of the peaks at two different frequencies, it should be

possible to deduce the temperature dependence of Irm 1 . Using a pulse technique with a fixed-path cell,

which is the resonant cell itself, we measured the

absorption coefficient across the transition at a fre- quency of 10 MHz (the highest frequency available

with our apparatus) and compared these measure-

ments with ones made simultaneously at 1 MHz.

Unfortunately it was not possible to deduce the law

for zm 1 because the data show that the peak at 10 MHz

still lies within the coexistence region. Since the region

is about 0.4 OC we can say only that

(9)

796

This high value for m 1 confirms the fact that our

measurements between 1 and 5 MHz correspond to the

co,r « 1 regime. On the other hand, this value for ’tm-1

appears to be slightly higher than that deduced from

the data in the isotropic phase

It is clear that it would be of interest to make measu- rements at frequencies higher than 10 MHz, but such

measurements are difficult because of the enormous

absorption at the transition.

Because of the unèxpected exponents found in

MBBA and PCB, it was generally thought that the

critical behaviour in the nematic side of the transition

was more complicated than that observed here, and

tentative interpretations have been proposed [3, 5].

For example, the theory discussed in ref. [5], which

assumes a coupling between the sound wave and the fluctuations of the director through the temperature variation of Frank’s elastic constants, predicts for the

relaxation frequency a critical exponent of 0.5 in agreement with the results in MBBA (0.4) and PCB (0.5). Obviously, this theory fails to explain the critical exponent of 1 in PAA. The unexpected exponents in MBBA and PCB are probably because the intramole- cular component has not been separated from the

critical component and the experiments were perform-

ed in unoriented samples. In this respect, it should be noted that a study of the attenuation anisotropy

in MBBA has been recently published by Castro

et al. [6]. In that paper the authors separated the

contribution due to rotational isomerism in the butyl

end groups of the molecule, from the contribution due to the critical relaxation of the order parameter.

From this analysis they deduced an expopent of - 1 for Da(T ), which is consistent with our direct determi- nation in PAA. However, on the basis of a reported exponent of 1 in the isotropic phase, they concluded

that the critical relaxation observed in Da is due to a

fluctuation mechanism similar to that observed above

T,r , that conclusion contrasts with ours, which states that da is mainly govemed by the Landau-Khalat- nikov mechanism. As for PAA, the value for tm-1

in the nematic phase of MBBA

appears to be one order of magnitude higher than

the corresponding value in the isotropic phase

Irm 1 - 108 àT s - 1 from the li gh t-scatterin g data of

ref. [21 ]).

6. Conclusion.

-

We find that our ultrasonic mea- surements in PAA are those for the wr « 1 regime in

the temperature range investigated, which corresponds

to T - Tc values from - 0.5 °C to - 15 OC and to Tc - T values from - 1 °C to - 20 °C. We estimate to within - 0.3 °C-0.4 °C the width of the transition

(i.e., the coexistence region of the two phases) and

we have not made quantitative measurements in this region.

In the isotropic phase, the critical component of a(T) is due to fluctuations in the order parameter, and the dynamic heat capacity theory, with an exponent of 1.5 for a(T), describes our experimental data well.

In the nematic phase, we observe a divergence of Da(T) with a critical exponent of - 1, which we attribute to the relaxation of the order parameter itselr (Landau-Khalatnikov mechanism). Since within

experimental error rx(T) diverges in the same way as

Aoc(r), this mechanism seems to be the dominant one

at least in the temperature range investigated (up to

1 °C from Tc). Of course very near Tc both the relaxa- tions of the fluctuations and the relaxation of the order parameter should contribute to the ultrasonic

absorption and, in this temperature range, it is likely

that a(T) behaves differently with respect to tempera-

ture (and frequency) than Aa(T).,

In both phases we find an exponent of 1 for the relaxation frequency, and in the nematic phase this

result eliminates one of the suggested models for ultrasonic absorption in this phase.

There is a différence of one order of magnitude

between the relaxation frequency of the order para- meter and the relaxation frequency associated with the fluctuations of the order parameter.

A self-consistency is beginning to emerge within a

subset of ultrasonic results in the nematic and the

isotropic phases. In the isotropic phase, studies on

PAA and PCB seem to be consistent with a diverging

as (T - Tc*)-1.5. In the nematic phase, this work and ref. [6] are consistent with Aoe diverging with an exponent of 1, and the results given in refs. [6], [22]

and [23] Support the assumption that rotational

isomerism in the butyl end groups is responsible for

the normal (as opposed to critical) relaxation process.

Liquid crystals present a great variety of more or

less second-order phase transitions. In our view, the dynamic heat capacity theory is of great interest in the

study of these phase transitions since it allows the

magnitude of the anomalies in the velocity and in the absorption per wavelength to be predicted from the

measurement of the specific heat. This point is illus-

trated in ref. [24] for the A-N transition in CBOOA and TBBA. Of course, in the vicinity of the transition,

more complicated behaviour is expected due, for example, to mode-coupling between the order para- meter and the hydrodynamic shear mode [25], or to

fluctuation effects whose wave numbers are much greater than ç -1 [12].

Acknowledgments.

-

Part of this work was done

(10)

at the Laboratoire de Physique des Solides of the Université Louis-Pasteur. It is a pleasure to thank

everyone there for their kind hospitality. We are also

grateful to Mme Pouyet, of the Centre de Recherches

sur les Macromolécules, Strasbourg, for doing the

calorimetric experiments.

References

[1] HOYER, W. A. and NOLLE, A. W., J. Chem. Phys. 24 (1956) 803.

[2] EDEN, D., GARLAND, C. W. and WILLIAMSON, R. C., J. Chem.

Phys. 58 (1973) 1861.

[3] KAWAMURA, Y., MAEDA, T., OKANO, K. and IWAYANAGI, S., Jpn. J. Appl. Phys. 12 (1973) 1510.

[4] NATALE, G. G. and COMMINS, D. E., Phys. Rev. Lett. 28 (1972) 1439.

[5] NAGAI, S., MARTINOTY, P. and CANDAU, S., J. Physique 37 (1976) 769.

[6] CASTRO, C. A., HIKATA, A. and ELBAUM, C., Phys. Rev. A

17 (1978) 353.

[7] IMURA, H. and OKANO, K., Chem. Phys. Lett. 19 (1973) 387.

[8] LANDAU, L. D., Collected Papers edited by D. Ter Haar (Gordon and Breach, New York) 1967, p. 626.

[9] KIRY, F. and MARTINOTY, P., J. Physique 39 (1978) 1019.

[10] GULARI, E. and CHU, B., J. Chem. Phys. 62 (1975) 795.

[11]

DE

GENNES, P. G., Phys. Lett. A 30 (1969) 454 ; Mol. Cryst.

Liq. Cryst. 12 (1971) 193.

[12] KAWASAKI, K., in : Phase transitions and critical phenomena,

vol. 5 A, eds. C. Domb and M. S. Green (Academic Press, New York) 1976.

[13] MATSUSHITA, M., Phys. Lett. 65A (1978) 149.

[14] FORSTER, D., LUBENSKY, T. C., MARTIN, P. C., SWIFT, J. and PERSHAN, P. S., Phys. Rev. Lett. 26 (1971) 17.

[15] JÄHNIG, F., Z. Phys. 258 (1973) 199-208.

[16] JÄHNIG, F., Chem. Phys. Lett. 23 (1973) 262-264.

[17] KEMPF, K. A. and LETCHER, S. V., Phys. Rev. Lett. 27 (1971)

1634.

[18] LEADBETTER, A. J., RICHARDSON, R. M. and COLLING, C. N., J. Physique Colloq. 36 (1975) C1-37.

[19] CHISTIAKOV, I. G., Sov. Phys. Usp. 9 (1967) 551.

[20] COLES, H. J. and JENNINGS, B. R., to be published in the Proceed- ings of the Brunel 1978 Meeting

on

Molecular Electro- optics. See also H. J. Coles, Proceedings of the 7th Inter- national Liquid Crystal Conference (Bordeaux) 1978.

[21] STINSON, T. W., LITSTER, J. D., Phys. Rev. Lett. 25 (1970) 503.

[22] CANDAU, S., MARTINOTY, P. and ZANA, R., J. Physique Lett.

36 (1975) L-13.

[23] NAGAI, S., MARTINOTY, P., CANDAU, S. and ZANA, R., Mol.

Cryst. Liq. Cryst. 31 (1975) 243.

[24] MARTINOTY, P., J. Physique Lett. 40 (1979) L-291.

[25] MATSUSHITA, M., Phys. Lett. 66A (1978) 507.

Références

Documents relatifs

Precise measurements of the Kerr effect [6, 9] and magnetic birefringence [6, 7] in the isotropic phase of. some nematic substances have been

In figure 6 we show a full logarithmic plot of the transi- tion densities for all four liquids, each scaled with respect to the value of its density p, at

of mixed rods and disks, and in the context of an entropy model for hard cylinders, the data indicate the relative fraction of rods and disks as the KL

Effect of micellar flexibility on the isotropic-nematic phase transition in solutions of linear

In conjunction with estimates of the volume discontinuity obtained from our previously reported PVT data, these turbidity data are used to calculate the values of

Measurement of the birefringence induced in liquids by ultrasonic waves : application to the study of the isotropic phase of PAA near the transition

on elastic interaction, recalling that its average energy is proportional to 1 /r per unit length of straight and parallel edge dislocations. Simple considerations

2014 Nous étudions le diagramme de phase d’un système dense de particules allongées avec une interaction Coulombienne, et montrons l’existence d’une transition