A novel high temperature dielectric ceramic has been demonstrated based on
strontium sodium niobate, Sr2NaNb5O15 (SNN1.0), which has the tungsten bronze
crystal structure and shows relaxor-like weak frequency dispersion. Initial
experiments involved co-doping with low levels of, Ca2+, Y3+ and Zr4+ to introduce
chemical inhomogeneity to broaden the twin dielectric peaks (Curie peak at 305 ᵒC,
and a low temperature peak at -15 ᵒC in SNN1.0). The resulting materials were
observed to exhibit high and stable relative permittivity across the technologically
important -55 to 300 ᵒC temperature range, with the aim of restricting the variation in
permittivity to within the Electronics Industries Alliance R-type ±15 % value. Unlike
previously reported materials with similar dielectric properties, the new dielectrics do
not contain Bi or Pb. In principle this makes them compatible with base metal
electrode multilayer ceramic manufacturing processing.
For tri-doped compositions, Sr2-x-yCaxYxNaNb5-yZryO15 (SCNN-YZ) dielectric
peaks became very diffuse. At x = y = 0.05, median εr values were 1310 +/-10 % from
-65 to 300 °C, with dielectric loss tangents, tan δ, ≤ 0.035 from -34 to 378 °C.
Microstructural analyses excluded core-shell mechanisms being responsible for the
flattening of the εr –T response. A further series of experiments with individual dopants
were performed to examine the reasons for peak suppression and attainment of Rtype
stability in relative permittivity. It was also realised that SNN was a solid solution
Sr2+xNa1-2xNb5O15 (SNN). On cooling from sintering temperatures the SNN1.0
composition, x = 0, lies outside the single phase solid solution region at room
temperature and exists as a two phase mixture of the limiting solid solution
composition, (x ~ 0.9), and a Sr modified NaNbO3 phase. Sodium deficient variants of
SNN (Sr2.1Na0.8Nb5O15, SNN0.8) were therefore investigated to produce a single
phase composition. Nevertheless, because the early tri-doped results were based on
x = 0 (Sr2NaNb5O15) it was decided to continue to use this as a starting point from
which to examine the effects of single-dopants. The single doping experiments
indicated that A-site vacancies bring about a severe suppression of the Curie peak.
However the presence of all three dopants is required to produce the near flat
response which was observed for SCNN-YZ (x = y = 0.05).
iv
Although the dielectric properties of the SCNN-YZ ceramics appear favourable
in terms of developing new high temperature capacitor materials, the presence of Na
vacancies in the SNN solid compositions is a drawback. A greater problem was
discovered in the latter months of the project. It had been reported in the literature that
SNN was metastable below ~1200 °C. The significance of this for a capacitor
application would depend on the kinetics of decomposition at capacitor operating
temperature. Annealing experiments at 300 to 600 °C were initiated. At 600 °C after
only 2 to 3 hours additional phases of SrNb2O6 and Sr2Nb10O27 form from the SNN1.0,
indicating partial decomposition. At 400 °C there is evidence of decomposition in
SNN0.8 after 24 weeks. Even at 300 °C, the upper working temperature of capacitors
for some power electronics applications, there is tentative evidence of structural
changes.
