• Aucun résultat trouvé

Study of RbDP phase transition in the paraelectric (upper) phase by phonon echoes

N/A
N/A
Protected

Academic year: 2021

Partager "Study of RbDP phase transition in the paraelectric (upper) phase by phonon echoes"

Copied!
6
0
0

Texte intégral

(1)

HAL Id: jpa-00208874

https://hal.archives-ouvertes.fr/jpa-00208874

Submitted on 1 Jan 1978

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.

Study of RbDP phase transition in the paraelectric (upper) phase by phonon echoes

A. Billmann, G. Guillot

To cite this version:

A. Billmann, G. Guillot. Study of RbDP phase transition in the paraelectric (upper) phase by phonon echoes. Journal de Physique, 1978, 39 (12), pp.1317-1321. �10.1051/jphys:0197800390120131700�.

�jpa-00208874�

(2)

STUDY OF RbDP PHASE TRANSITION

IN THE PARAELECTRIC (UPPER) PHASE BY PHONON ECHOES

A. BILLMANN and G. GUILLOT

Laboratoire d’Ultrasons (*), Université Pierre-et-Marie-Curie, Tour 13, 4 place Jussieu, 75230 Paris Cedex 05, France

(Reçu le 22 mars 1978, révisé les 19 mai et 17 juillet 1978, accepté le 23 août 1978)

Résumé.

-

Nous avons étudié la transition de phase du RbH2PO4, en mesurant le temps de vie T2

et l’amplitude A du signal d’écho de phonons à 2 03C4, dit dynamique, dans des poudres résonnantes.

L’atténuation suit la loi de variation obtenue par d’autres techniques ultrasonores dans le même type de composés. L’allure de A(T) donne de plus quelques indications sur le comportement critique

des non-linéarités.

Abstract.

2014

A ferroelectric phase transition (RbH2PO4) has been investigated by measuring

the damping constant T-12(T) and the amplitude A(T) of the dynamic 2 03C4-phonon echoes in resonant powders. Our results on the attenuation confirm those obtained with other ultrasonic methods

on the same type of compounds. We also get from the behaviour of A(T) more information about non-linear effects near the transition.

Classification Physics Abstracts

62.25201364.70D201377.80B

Introduction. - The phonon echo method is a

basic acoustic technique, which gives results compa- rable to those obtained by usual ultrasonic techniques.

With it, the study of surface defects [1] or of acoustic properties of some breakable [2] or microsize crystals [3] is much easier.

We have studied the phase transition of RbH2P04 by means of phonon echoes. The static properties of

this compound are known to be very much the same

as those of KDP (§ 1), but its dynamic properties had

not been investigated till recently [4]. We will shortly present the 2 t-phonon echo in powders (§ 2), and interpret our results on the lifetime T2 (§ 3) and amplitude (§ 4) considering our specific case of reso-

nant powders.

1. Static properties of RbDP : analogies with KDP.

-

The rubidium diacid phosphate (RbDP) has similar

structural and critical properties to KDP, a well

known ferroelectric crystal. Its inverse longitudinal

dielectric susceptiblity, measured at constant stress X, varies linearly above the transition temperature Tc ~ 145 K (122 K for KDP), in the upper 4 2 m

paraelectric and piezoelectric phase : X "DIx = (T - Tc)/C [5, 6]. The lower phase has the 2 mm symmetry and is ferroelectric for these compounds.

The elastic constants of RbDP have been measured

through the transition [7], and have the same charac-

teristic behaviour as the constants of KDP [5, 8] in

the upper paraelectric phase. C66 (measured with E kept constant) vanishes at the transition temperature, whereas CP66 (measured with P kept constant) remains

constant as well as all the other elastic constants.

These variations are fully explained if in the expres- sion for the Landau free energy ço(q), the order para- meter ~ is identified with P 3 (third component of the spontaneous polarization) and if the coupling between P 3 and only one elastic mode S6 via the piezoelectric

modulus a36 is considered [9]. Other elastic modes

cannot be coupled to P3 because of the zero values of the relevant piezoelectric moduli in the upper

phase,

The dynamic elastic properties of KDP had been

well investigated near the transition [8 and references therein, 3]. Our results on RbDP will confirm the

analogy with KDP in the paraelectric phase near the

transition.

(*) Associated with the Centre National de la Recherche Scien-

tifique.

Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphys:0197800390120131700

(3)

1318

2. 2 T-phonon écho in piezoelectric média.

-

The

echo phenomena in piezoelectric crystals has been widely studied for some years, either in single crystals [10, 11] or in powders [12-14]. It is essentially the

re-radiation of a coherent signal at time 2 r by a sample which was excited by two radio-frequency pulses of electric field at times 0 and -r. T2 is the life- time of this signal when s varies and is directly related

to an acoustical lifetime [15].

We measure the 2 T-echo decay constant T2, and

the echo amplitude A (extrapolation at s

=

0 of the amplitude A(,r», as functions of the temperature T, from 165 K to 145 K (Fig. 1), in RbDP powders (size range : 20-40 03BC) at 110 MHz. The powder is in

a capacitor, part of a matched electrical circuit, under

a low pressure of gaseous helium.

FIG. 1.

-

2-r dynamic echo. Lifetime (T2(+)) and amplitude (A (0)) measured in RbDP powders. The continuous curve repre- sents the expected theoretical law for T2(T), where the parameters

T2P1,

pl.

and Tc have been fitted to the values : T2pl = 15.2 ps ; Tc = 143.5 K.

3. Study of the damping constant T2 ’(T).

-

3. 1 CONTRIBUTIONS TO T1- l(T). - The attenuation processes involved in T2- 1 have three main origins : a) The acoustic transmission loss to the external medium must not show any variation with the tem-

perature [3].

03B2) The internal attenuation processes, intrinsic to the properties of the solid medium (such as diffusion by defects, impurities, or thermal phonons) may be

supposed to give a contribution constant with the temperature over the temperature range ; and the

coupling between the order parameter P3 and the

elastic mode S6 is the only source of critical attenuation

(which we will examine in detail further on, in § 3.2).

y), The third process is due to mode conversion and is specific to the experimental condition of microsize

resonant powders : in every particle, there is a complex stationary vibration mode, which is a mixture of pure

resonance modes with different damping constants

and different non-linear efficiencies to build the 2 T-

echo. Every particle has already a complex damping

constant related to the pure mode damping constants.

Further, we measure an average value of T2- 1 taken

on all the particles. Consequently, the critical part we

extract from the measured attenuation should not

coincide numerically with the theoretical critical atte-

nuation of the S6 mode.

3.2 THEORETICAL CRITICAL ATTENUATION.

-

Gar- land [8] made a thorough study of the critical attenua-

tion in the upper phase of KDP. But he worked with

single crystals of much larger sizes than the half acoustic wavelength at his frequency (at 15 MHz, Â/2 - 150 03BC) : the crystal was clamped by the use of

a frequency much higher than the acoustic resonance frequency, or the surface of the crystal could not follow the excitation.

This is different from our case : we have studied

mechanically resonant powders, and we are in the

case of a free crystal.

In a clamped crystal, Garland found a critical atte- nuation of the mode S6 at constant strain (S) :

where u is the ultrasonic velocity of the mode S6 (pu’

=

CE66) and rs is the polarization relaxation

time at constant strain. The Landau model gives this

relaxation time for the order parameter above the transition température ; it is related to the suscepti- bility and to the kinetic coefficient L describing the

order parameter fluctuations by Ts

=

1/2 LX331s. Some

thermodynamic considerations [9, 3] give :

with

and

But in our case of the free crystal, we must suppose

we measure the critical attenuation of the mode S6

at constant stress (X) :

where

So :

(4)

3.3 EXPERIMENTAL RESULT.

-

We extract from

T2- ’(T) the part which is constant with temperature

over the temperature range T2-1plateau, and keep only

the temperature dependent part of the attenuation

8i l(T) :

As we expect the variation of 021(T) to result only

from the coupling between the mode S6 and the order parameter P3, we try to describe 02(T) with a law :

The experimental data fit quite well with such a law (Fig. 1). We find :

3.4 DISCUSSION.

--

Thus we estimate Tc from our

measurement of T2(T) ; we find 143.5 K. From the

literature, the value of T, should vary from one

sample to another.

For acoustic studies of RbDP, Pierre et al. [7] found Tc ~ 147 K, and in a recent paper, Singh and Basu [4]

found in two different samples Tc = (145.2 ± 0.1) K

and (143.5 ± 0.1) K, studying then particularly from

the lower phase. A numerical evaluation of Mlh. can

be made with the results on the elastic and piezo-

electric constants of RbDP, obtained by Pierre et

al. [7] and on the order parameter relaxation rate

made in KDP by Garland [8].

With

we find

The theoretical temperature dependence is experi- mentally well found but the measured temperature

dependent relaxation time is higher than the theore- tical one of the mode S6 [8] by about two orders of magnitude. We attribute this discrepancy to the

effect of mode conversion.

Comparable results were obtained in the study of

the K.D.P. crystal by phonon echoes in the same phase : the temperature dependence found was

correct, but the measured temperature dependent

relaxation time was too high by two orders of magni-

tude [3].

4. Study of the écho amplitude A(T).

-

The echo amplitude can also give information on some thermo-

dynamic coefficients playing a part in its formation.

But this amplitude is only an average value on all

the modes, the anisotropy effects are not measurable by such a method.

4.1 TWO MODELS, TWO INTERPRETATIONS OF A(T).

- We will first summarize the features of two main models for the 2 T-echo with which we could interpret

the variation of A(T).

a) First model.

-

The first model was elaborated for phonon echoes in single crystals [10] and is easily

also applied to the case of powders where the particle

sizes are much larger than the acoustic half wave-

length. A first pulse of frequency v generates phonons

at the surfaces of the crystals at time 0. They propagate freely in the crystals till the instant r of application

of a second pulse at the same frequency. They are

then reversed by the double harmonic of the electric field, during the application of the second pulse : the

free evolution of their phases afterwards is the oppo- site to their evolution between 0 and T. At time 2 T has been achieved a complete phase reversal and the initial coherent state of phase is recovered. As T increases the echo amplitude should decrease as

e

Such a model gives generally a good account of

the experimental observations in the out of resonance cases.

foi) Second model.

-

New characteristics of echoes in resonant powders cannot be understood with such

an interpretation. We restrict here ourselves to the so-called dynamic 2 T-echo [12, 14]. Its amplitude

varies no longer simply exponentially with T, as expected from a signal due to a parametric interaction of the type above suggested. The signal observed in

resonant powders first increases with T instead of

being maximum for small -r; then it reaches a maximum at Tmax and decreases as e - 2t/T2 for large T(r ~ 2 !max).

This behaviour had been previously seen in single crystals of CdS at 9 GHz and 4 K, for very high amplitudes of the excitation fields [10].

Therefore, when the excitation fields are stronger

as in CdS single crystals or more efficient (as in reso-

nant powders) a new type of non-linearity must be

invoked. This process should permit a build-up

period of the echo amplitude for small T. Any type of anharmonic interaction between the excited modes

can reverse the phase conditions of the modes pro- duced by the first pulse, mixing them with the modes

issued of the second pulse from the time T. So, a coherent signal can appear at time 2 r ; its amplitude

is roughly described [12] by a law : e - 2t/T2( 1 - e - 2t/T 2),

which shows a build-up period for small r.

y) RbDP case : short T2.

-

When T2, measuring

the average mode lifetime, is short enough, the initial increase of the echo amplitude for small T is not

visible. In the limit T > Irmax (Tmax ~ 0.2 T2), a para- metric process or an anharmonic interaction would both give an echo amplitude proportional to e- 2t/T2.

For our experiments, T2 is too short to choose bet-

ween the two models. But the relevant non-linear

(5)

1320

coefficients would not be the same for these two types of echoes. We shall present both cases with the res-

pective relevant coefficients.

4.2 INVOLVED LINEAR AND NON-LINEAR COEFFI- CIENTS.

-

a) The electromechanical coupling factor k2(T)

=

e2(T)/CE(T) XI(T) [5] (where e is the piezo-

electric coefficient) measures the conversion, either

from electric to elastic, or from elastic to electric energy. It must be taken into account at least twice, for the elastic modes generated by the first electric

pulse, and for the echo detection at 2 r. Therefore,

the detected echo amplitude varies as A(T) oc k2(T).

This factor also measured the amplitude of the signal

radiated after a single pulse, signal which is a noise

in our measurement.

The electromechanical coupling factor k 36 2 between

the critical mode P3 and its unique coupled mode S6

is the only one varying with the temperature as k2 T - Te - To

from an ex erimental estima-

k36Crit(T) = T - T ; crit from an experimental estima-

T -T 0

tion of (Tc - TO) - 4.2 K [7], k36crit(T) would not

be doubled over our experiment range. The other

coupling factors are quite constant with the tempe-

rature near the transition [7].

We also find through the attenuation study that

the critical experimental lifetime is much greater than the S6 mode lifetime, and we attribute this to mode conversion effects (§ 3.4); it shows that the S6 mode

excitation is not prevailing as compared to those of

the other pure elastic modes.

03B2) We call ~(T) the factor measuring the efficiency

of the non-linearity building the echo.

-

First model. - If the echo is due to a term as

03B2ESS in the Hamiltonian (E is the electric field, S is

the strain) [10], the electric field acts via its second

XSN B

.

XNL 2

harmonie E2W (E2w=

=

XNL E2w), XL ) and the echo amplitude

should finally vary as :

-

Second model.

-

If the anharmonic elastic interaction of a term 41 ! Ce S4 in the Hamiltonian

produces the echo [ 12], the electromechanical coupling

factor must still be considered twice, as the elastic

modes from the second electric pulse act quadratically,

to reverse the phases of the elastic modes from the first pulse. In this case, the echo amplitude should

vary as : :

4.3 EXPERIMENTAL RESULT. - Our experimental

data are well aligned on a straight line of slope n = - 1

when plotted with 20 Logo (T - Tc) taken as abscis-

sa, for Tc

=

143.5 K. The relative variation of n does

not exceed 10 % if Tc varies of ± 0.5 K (Fig. 2).

We obtain :

FIG. 2.

-

2,r echo amplitude (dB) plotted with 20 Log10 (T - Tc)

takén as abscissa. When Tc

=

143.5 K, the experimental points

are well aligned on a straight line of slope - 1.

We also notice that near the transition, the signal (echo) to noise ratio is greatly enhanced (the noise

in our measurement is the ringing consequent on a

single pulse). From this observation, we attribute the

divergence of A(T) above Tc uniquely to ~(T), the efficiency of the non-linearity, and we neglect the

influence of k2(T). Anyway, as we discuss it in sec-

tion 4.2 a, only one electromechanical coupling factor

.

2 Tc - To

varies above varies above T Tc as K36 crit = as k2

=

Tx - TO T - T° ; and the and the atte- atte-

nuation study indicates that the S6-mode is not mainly excited. We thus conclude that only q(T) diverges as ( T - Te)-l.

4.4 DISCUSSION.

-

1) In the case of a parametric

effect with the electric field at the doubled frequency,

~(T) . P(T) XSNL(T)

~(T) is proportional to A

-CE (T) XSL( XSL(T) . A precise

knowledge of X’NL(T)IXI(T) (efficiency of the electric

field double harmonie conversion) at our frequencies

should give some insight about an eventual critical behaviour of ’1(T).

2) In the case of an elastic anharmonicity, ~(T) is

.

2 CNL(T )

proportional to k (T) CE (T ). As we deduce from

propotional to k2(T)

CE(T)

°

As we deduce from the signal to noise ratio behaviour that ~(T) diverges

near the transition as (T - Tc)-l (the electromecha-

nical coupling factor k2(T) varies slowly near T,,

anyway), we should conclude :

(6)

Some examples of critical divergences of the third order elastic constant are known, even if the second order elastic constant vanishes at the transition tem-

perature in perovskite structures [16]. But no compa- rable result on the fourth order elastic constant is known to us.

As a conclusive remark, it may be noticed that in both models, the divergence of C 6(T) could explain

the divergence of q(T) we observe.

So, complementary studies of dielectric and elastic non-linearities of this type of compound in our fre-

quency range would be necessary for a complete interpretation of our result on the echo amplitude.

But it gives already some information on non-linear coefficient products.

Concerning the model one could finally retain,

Laïkhtmann [17] has shown that acoustic non-

linearities should be expected to have a greater impor-

tance than other non-linearities.

5.

-

Conclusion.

-

Our experiment gives a correct temperature dependence of the critical ultrasonic

lifetime, though its order of magnitude is largely

overestimated by mode conversion effects. We can

also attribute ~(T), the efficiency of the non-linearity building the echo, the divergence in (T - Tr 1 we

observe on the echo amplitude ; this could be a pre- cious piece of information to study non-linear effects.

This confirms the value of the phonon echo method for investigation of phase transitions with very light experimental devices.

Acknowledgments.

-

We are very grateful to

M. Remoissenet who provided us with the samples

we studied and had previously investigated their

elastic constants [7].

References

[1] JOFFRIN, J., LEVELUT, A., SALIN, D., J. Physique 37 (1976) 271.

[2] BILLMANN, A. et LEVELUT, A., J. Physique 38 (1977) 1307 (Transition de phase de l’A.D.P.).

[3] JOFFRIN, J., LEVELUT, A. et FRÉNOIS, Ch., J. Physique 37 (1976) 275 (Echos de phonons et transitions de phase).

[4] SINGH, G. P. and BASU, B. K., Phys. Lett. 64A (1978) 425.

[5] MASON, W. P., Piezoelectric crystals and their application to

ultrasonics (Van Nostrand Cy, Inc.) 1950.

[6] MATTHIAS, B. T., Phys. Rev. 85 (1952) 723.

[7] PIERRE, C., DUFOUR, J. P. and REMOISSENET, M., Solid State Commun. 9 (1971) 1453.

[8] GARLAND, C. W., Phase transition and critical points in Physical Acoustics, Vol. VII (Academic Press) 1970.

[9] SAWADA, A., UDAGAWA, M. and NAKAMURA, T., Phys. Rev.

Lett. 39 (1977) 829.

[10] BILLMANN, A., FRÉNOIS, Ch., JOFFRIN, J., LEVELUT, A. and ZIOLKIEWICZ, S., J. Physique 34 (1973) 453 ;

FRÉNOIS, Ch., JOFFRIN, J. et LEVELUT, A., J. Physique 34 (1973) 747.

[11] SHIREN, N. S., MELCHER, R. L., GARROD, D. K. and KAZYAKA, T. G., Phys. Rev. Lett. 31 (1973) 819.

[12] FOSSHEIM, K., KAJIMURA, K., KAZYAKA, T. G., MELCHER, R. L. and SHIREN, N. S., Dynamic polarization echoes in piezoelectric powders in Phys. Rev. B (Feb. 78).

[13] MELCHER, R. L. and SHIREN, N. S., Phys. Rev. Lett. 36 (1976) 888.

[14] SMOLENSKI, G. A., POPOV, S. N., KRAINIK, N. N. and LAIKHT- MANN, B. D., Zh. Eksp. Teor. Fiz. 72 (1977) 1427.

[15] FRÉNOIS, Ch., JOFFRIN, J. et LEVELUT, A., J. Physique Lett.

35 (1974) L-221.

[16] ARNOLD, W., Phys. Rev. B 1 (1970) 982.

[17] LAÏKHTMANN, B. D., Sov. Phys. Solid State 18 (1976) 357.

Références

Documents relatifs

sodium nitrate structure... For model I, it is well known [15] in case of large rotational temperature factors that the fitting procedure gives too small a value

The measured field cooled magnetization Mf., ,.(T) is found stable over our measuring window but does exhibit a slight temperature dependence. 1) and-can be measured

amplitude modes in the biaxial phase mix the components of biaxiality and birefringence of the order parameter, the critical (quasi-biaxial) mode presenting

Gennes [1] to calculate the critical behaviour of the elastic constants : i) The fluctuations of the smectic order parameter are coupled to the director by means.

The data (two runs) belowlabove To are specified by full/open symbols. In X-ray single crystal experiments we have studied the influence of domains in the distor- ded low

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des

Due to the presence of a large amount of ordered mate- rial, nucleation of the ordered phase was not neces- sary, and growth could proceed on the boundaries of the not

The vibration frequency was high enough (KHz) to eliminate the transitory internal friction. A peak of internal friction accompanied by an anomaly of the elastic modulus