Barium sodium niobate (Ba2NaNb5O15)
is a tungsten bronze structure that exhibits a complicated sequence of
six structural phase transitions, including three incommensurate (IC)
phases. The phases are unusual in that all but the highest temperature
P4/mbm
structure are ferroelectric. Unlike the situation for most incommensurate insulators, in
which ferroelectricity develops at low temperatures along the modulation
direction, the polarization direction in barium sodium niobate is orthogonal to
the modulation(s), permitting some unusual phenomena. In the present study
we analyse the thermal and dielectric behaviour at the Curie temperature
TC near 830 K as well
as that at the Ccm 21–IC(1q) transition
near 543 K, the IC(1q)–IC(2q) transition near
565 K and the IC(2q)–P4bm
transition at 582 K. The entropy change at 565 K is related to the wall
roughening model of Rice et al (1981 Phys. Rev. B 24 2751). Data near
TC = 830 K
indicate close proximity to a tricritical point, and discussions of critical
exponents are presented, all of which are found to be mean field. Because of
Na vacancies, transition temperature variation is found among specimens
Ba2Na1−xNb5O15
(830 K<
TC(x)<
865 K), and the system appears to be describable by the disordered exclusion model as a slightly
first-order intrinsic system whose dynamics are suppressed by weak disorder. Near
TC the specific
heat C(T)
is compared with the random bond prediction of Harris (1974 J. Phys. C: Solid State Phys. 7 1671):
C(T) = C0(T)/[1+bx2C0(T)],
where C0(T)
is the intrinsic specific heat of the vacancy-free crystal varying as
(TC−T)−1/2
and x
is the sodium vacancy concentration. In agreement with Harris’s model, the shifts in
TC(x) are to lower
T with increasing
x and scale as
x; the broadening
scales as x2; and the effective critical exponent remains unchanged at
α = 1/2.