Professional Documents
Culture Documents
Investigations On The Defect Chemistry and The Sintering of Barium Titanate Ceramics by Oxygen Coulometry
Investigations On The Defect Chemistry and The Sintering of Barium Titanate Ceramics by Oxygen Coulometry
www.elsevier.com/locate/ssscie
Abstract
The application of the oxygen coulometry method on the investigation of the defect chemistry and the sintering process of donor- and
acceptor-doped barium titanate ceramics is described. The measuring principle is based on two identical solid electrolyte circonia cells with
two electrode pairs both for measuring of the oxygen partial pressure and for oxygen pumping. Three different oxygen exchange processes of
the samples could be detected. (1) The oxygen exchange due to the change of the equilibrium temperature at a fixed oxygen partial pressure,
which is caused by the change of the concentration of oxygen vacancies. (2) The oxygen exchange which is related to the valence change
of MnTi during the change of the ambient atmosphere. The data are consistent with the assumption that manganese occurs as Mn3+ at an
oxygen partial pressure of 2.4 Pa and a temperature of about 1400 ◦ C. (3) During the sintering of donor-doped barium titanate ceramics,
oxygen expelling processes were detected which are related to the incorporation of the donor into the lattice. In the case of a doping level of
0.5 mol% La, the estimated amount of expelled oxygen is less than expected for exclusive electronic donor compensation. 2002 Éditions
scientifiques et médicales Elsevier SAS. All rights reserved.
Keywords: Barium titanate; BaTiO3 ; Defect chemistry; Sintering; Oxygen coulometry
Fig. 1. Scheme of the measuring equipment of the oxygen coulometry device. See Section 2.2 for details.
We report both on the development of a suitable measur- of the oxygen partial pressure, pO2 , of the flowing gas stream
ing system, on the measuring of the change of the oxygen by measuring the voltage between the electrodes, U cell , ac-
vacancy concentration due to the change of the thermal equi- cording to the Nernst equation (1)
librium, on the change of the valency state of MnTi defects
RT pO2
in dependence on the oxygen partial pressure and on the sin- U cell = ln ref , (1)
tering of undoped and donor-doped barium titanate ceramics 4F pO2
with varying Ba/Ti stoichiometry. R molar gas constant, T absolute temperature, F Faraday
ref
constant, pO 2
oxygen partial pressure of air as the reference
gas.
2. Experimental procedure The second pair of electrodes is used for the transport of
a well-defined amount of oxygen between the flowing gas
2.1. Preparation of specimens and characterisation inside the cell and the ambient air by an applied electric
current, I , according to the following Eq. (2)
Ceramic powders with a nominal composition of Ba1−x -
Lax TiO3 + y TiO2 (0 x 0.05, 0 y 0.05) and I
nO2 = t. (2)
BaTi1−x Mnx O3 (0 x 0.05) were prepared by the clas- 4F
sical mixed-oxide powder technique. After mixing (agate nO2 is the mole number of oxygen transported during the
balls, water) and calcination (1100 ◦ C, 2 h in air) of time t. The sign of I determines the transport direction.
appropriate amounts of BaCO3 (Solvay, VL600), TiO2 The maximum current amounts to 100 mA with a stability
(Merck, no. 808) and La2 (C2 O4 )3 ·9 H2 O (SKW Piesteritz) better than 1 µA.
or MnCO3 (Riedel de Haen, p.a.) the powder was milled Both electrode pairs are connected by an electronic
again and densified to disks with a diameter of 12 mm and a control circuit with the set point U set and the control variable
height of nearly 3 mm. The samples were sintered both in a I in that sense, that a constant output pO2 (U cell ) of the cell
special gas-tight furnace of the coulometry system (see be- is maintained by an oxygen transport into or out of the ZrO2
low) and in air at temperatures between 1300 and 1400 ◦ C tube (see Fig. 1).
for the purpose of comparison. To avoid interfering contam- The measuring principle is described as follows. After
ination during sintering they were contained in ZrO2 powder the preparation of the carrier gas (gas mixing and flow
covered Pt /Al2 O3 dishes. control of total 200 ml min−1 by a device manufactured by
The microstructure of the polished and chemically etched Horiba Ltd., Japan) cell 2 is used to produce a working
specimens was examined by optical microscopy and by gas stream with well-defined pO2 for the furnace with the
scanning electron microscopy (SEM). sample to be investigated. After passing the furnace, the gas
streams through cell 1, which set point U set is chosen to
2.2. Oxygen coulometry get a appropriate current I pump . As long as no change of
the oxygen partial pressure between the two cells occurs,
The measuring principle is described by means of the this current is constant and exhibites the base line of the
schematic diagram in Fig. 1. The mean components of the measurement I0 . If an oxygen exchange between the sample
equipment are two identical, electronically controlled, oxy- and the ambient atmosphere occurs, e.g., during change
pump
gen conducting ZrO2 Nernst cells (SensoTech, Magdeburg- of the furnace temperature, a changing current I1 will
Barleben, Germany, model Oxylyt 10-21), which exhibit two compensate the changed oxygen partial pressure of the
pairs of electrodes. One of them is used for the determining output gas stream of the furnace. Thus, the total amount
H.T. Langhammer et al. / Solid State Sciences 4 (2002) 197–203 199
of exchanged oxygen during a time interval t2 − t1 can be Fig. 3. Oxygen pumping current I1 and temperature T in dependence on
calculated by the following expression the measuring time during a two-step temperature change between 1140
and 1350 ◦ C of “undoped” BaTiO3 , pO2 = 2.4 Pa.
t2
1 pump
nO2 = I1 (t) − I0 dt. (3)
4F Table 1
t1 Change of the total concentration of oxygen vacancies [VOtot ] (cm−3 ) after
pump the change of the equilibrium temperature from T1 to T2 at an oxygen
Minimum peaks of I1 indicate an oxygen release or a partial pressure of 2.4 Pa, both determined by oxygen coulometry and
reduction of the sample, while maximum peaks point to an calculated by means of the defect model of Daniels et al. [1]
oxygen consumption or oxidation process.
T1 → T2 (◦ C) Experimental data Model calculation
A quantitative check of the procedure was performed by
1245 → 1140 9.3 × 1017 5.0 × 1017
the decomposition reaction
1140 → 1245 8.8 × 1017 5.0 × 1017
+Q 1245 → 1347 26.2 × 1017 14.4 × 1017
BaO2 −→ BaO + 1
O2 ↑, (4)
2 1347 → 1245 26.7 × 1017 14.4 × 1017
and resulted in errors 5% within a mass range of BaO2
of between 0.4 and 4.0 mg which corresponds to about 10−6
and 10−5 mol O2 ↑. The interesting oxygen exchange effects about 100 ppm (see [2], e.g.) caused by the raw materials.
of the BaTiO3 samples investigated (mass ≈ 1 g) amount The oxygen exchange of a specimen during the temperature
to between 10−7 and 10−5 mol. The sensitivity limit of change between 1140 and 1350 ◦ C in two intervals at an oxy-
the apparatus can be estimated by Eq. (2) to 10−9 mol gen partial pressure of 2.4 Pa is shown in Fig. 3. Assuming
or 0.032 µg O2 if we assume a pumping current of 10 µA that all exchanged oxygen is caused by the formation (oxy-
during a measuring time of 60 s. Hence, the sensitivity is at gen release) or by the annihilation (oxygen consumption) of
least one order better than the sensitivity of a commercial oxygen vacancies after changing the equilibrium tempera-
thermobalance (≈ 1 µg at 1400 ◦ C). ture, the change of their total concentration, [VOtot], can be
To separate the influence of alumina (furnace tube and determined by the measured oxygen exchange nO2 accord-
sample holder) and zirconia (sample pad, ≈ 300 mg) on the ing to
oxygen exchange, empty tests were performed. As shown NA ρBaTiO3
in Fig. 2, only at temperatures higher than about 1300 ◦ C a VOtot = 2nO2 , (5)
msample
noticeable release of oxygen is observed. The total expelled
oxygen amounts to maximum 1.0 × 10−7 mol during the where NA is the Avogadro constant, ρBaTiO3 is the sample
heating at 10 K min−1 up to 1400 ◦ C. Probably, the oxygen density and m is the sample mass. The good correspondence
loss is caused by the formation of Schottky-type oxygen of the data between the heating and cooling steps (see Fig. 3
vacancies in ZrO2 and Al2 O3 . All further quantitative results or Table 1) proves the reversibility of that processes and
were corrected by subtracting of that apparatus effect. supports the assumption above.
To compare the experimental data with existing de-
fect models of BaTiO3 , we used the data of Daniels and
3. Results and discussion Härdtl [1], who published a complete set of constants of the
mass action laws of their defect model, enabling the calcu-
3.1. Equilibrium measurements in “undoped” BaTiO3 lation of the total oxygen vacancy concentration in depen-
dence on temperature and oxygen partial pressure. Table 1
To investigate the response of BaTiO3 ceramics to a shows the comparison between the experimental and the cal-
change of the thermal equilibrium, nominally undoped ma- culated data of [VOtot ]. Although the model does not in-
terial was chosen, having an accidental acceptor impurity of clude all defect types which are established by now (e.g.,
200 H.T. Langhammer et al. / Solid State Sciences 4 (2002) 197–203
For Mn-doped BaTiO3 it is known that Mn substitutes Fig. 4. Oxygen pumping current I1 and temperature T in dependence
for titanium and can adopt the valency states 4+, 3+ and on the measuring time during the sintering of BaTi1−x Mnx O3 ceramics
2+ [13–17]. While in air-sintered and otherwise undoped (0.005 x 0.04), pO2 = 2.4 Pa.
samples the isovalent defect Mn4+Ti dominates, the sintering
or annealing in reducing atmosphere reduces manganese to Table 2
Mn3+ 2+
Ti and MnTi , respectively [13,16,17]. In the valence Amount of the expelled oxygen (mole fraction xO ) due to the change
states 3+ and 2+, respectively, Ti-site occupying manganese of the valency of Mn from 4+ to 3+ or 2+, respectively, during the
is negatively charged (Mn
Ti and Mn
Ti, respectively) and sintering of BaTi1−x Mnx O3 in reducing atmosphere (pO2 = 2.4 Pa). The
data were determined both coulometrically (roughly estimated maximum
must be compensated either electronically by holes or
values) and calculated, assuming exclusive oxygen vacancy compensation
by positively charged oxygen vacancies VO•• according to of either Mn3+ or Mn2+ as the dominating valence state of manganese
following equations:1 x Roughly estimated Compensation Compensation
experimental data of Mn3+ of Mn2+
MnxTi ↔ Mn
Ti + h• ↔ Mn
Ti + 2h• , (6)
0.005 0.0024 0.0025 0.005
2 MnxTi + OxO ↔ 2 Mn
Ti + VO•• + 12 O2 ↑, (7) 0.02 0.011 0.010 0.020
0.03 0.016 0.015 0.030
MnxTi + OxO ↔ Mn
Ti + VO•• + 12 O2 ↑. (8) 0.04 0.019 0.020 0.040
the coulometry measurement, compared to air (pO2 = 2.1 × Schottky-type equilibrium oxygen vacancies. To attribute
104 Pa) as the ambient atmosphere during the calcination peaks (II) and (III) to particular processes during the
of the powder. Hence, the sample releases oxygen for as sintering, additional measurements at soaking temperatures
long as the new equilibrium oxygen activity is reached. of 1200 and 1300 ◦ C were performed in the case of La-doped
To check this, the sample was annealed at 1200 ◦ C for material. The results together with the related microstructure
two hours in air to reoxidize and then measured again data are shown in Fig. 7b,c. While a sintering temperature
at a pO2 of 2.4 Pa. The results, also shown in Fig. 5, of 1200 ◦ C is too low to initiate the anormal grain growth
exhibit the expected behaviour. Due to the circumstance of BaTiO3 , the microstructure of the specimens sintered at
that the restoration of the thermodynamic equilibrium in the 1300 and 1430 ◦ C exhibits the typical bimodal intermediate
sintered, coarse-grained sample needs much more time than and final stage, respectively. The onset temperatures of
in the fine-grained power compact, the onset temperature of the peaks (II) and (III) correspond very well with the
the peak is shifted for about 70 K to higher temperatures starting temperatures of the exaggerated grain growth in
as well as the expelled oxygen amounts to approximately the presence of a SiO2 and TiO2 containing liquid phase,
50% of the corresponding value of the former sintering step. respectively [20–22]. In the case of La-doped material, the
Whereas the oxygen exchange between sample and ambient SiO2 impurities result from the abrasion of the agate balls
atmosphere is determined mainly by bulk diffusion in or during milling. Their small amount causes an incomplete
out of the large grains of the sintered specimen, the much grain growth. The Nb-doped samples exhibit an analogous
faster grain barrier diffusion is the rate-controlling process behaviour with respect to the microstructure development at
for the oxygen release of the sintering green compact. This different sintering temperatures.
oxygen release effect strongly varies with different powder Hence, the oxygen release related to the peaks (II) and
qualities and with different chemical compositions, and it is (III) originates from the incorporation reaction of La at Ba
significantly influenced by the Ba/Ti stoichiometry. sites
The oxygen exchange during the sintering of donor-
doped BaTiO3 is shown in Figs. 6 and 7a with NbTi and La2 O3 + 2 TiO2 → 2 La•Ba + 2 Ti
Ti + 6 O×
O + 2 O2 ↑,
1
(9)
LaBa as the dopant elements, respectively. In both cases, four which is in accordance to the findings of Drofenik [23], who
oxygen release peaks can be observed during the heating stated that in BaTiO3 the endothermic donor incorporation
cycle, denoted with increasing temperature by numbers I– according to Eq. (9) is only possible as long as a constant,
IV. Peak (I), which is strongly developed only in the case of sufficient high energy is provided by the stored free surface
La-doping, results from the different oxygen partial pressure energy during the exaggerated grain growth. It should be
of the ambient atmospheres during calcination and sintering noted that this finding is in contradiction to numerous
which was described above. Only peak (IV), for La-doped publications where the authors assumed that the donor
BaTiO3 strongly overlapping with peak (III), corresponds dopant is already incorporated after the calcination of
to a reversible process since it has a counterpart during the ceramic powder at about 1150 ◦ C. To confirm the
the cooling cycle as in the case of undoped BaTiO3 (see coincidence of the beginning of the liquid phase promoted
Fig. 5) and obviously originates from the formation of anomalous grain growth with a distinct oxygen release
202 H.T. Langhammer et al. / Solid State Sciences 4 (2002) 197–203
Fig. 7. Oxygen pumping current I1 in dependence on time during the sintering of specimens with a composition of Ba0.995 La0.005 TiO3 + 1 mol% TiO2 at three
different maximum temperatures 1430 ◦ C (a), 1200 ◦ C (b) and 1300 ◦ C (c). Ambient atmosphere: pO2 = 2.4 Pa. Optical micrographs of the corresponding
microstructure are also presented (bar = 500 µm). Marks (I)–(IV) are explained in the text.
0.1 ( cm)−1 or more are reached, all donors are compen- charge compensation is assumed. Hence, a portion of
sated by titanium vacancies. They explain the observed elec- lanthanum is already incorporated during calcination
trical conductivity and its well-known anomaly in BaTiO3 and/or the donors are, at least partially, compensated by
by the exclusive influence of oxygen vacancies as intrinsic cation vacancies. To disentangle this question and other
donors which should be inhomogeneously distributed be- problems related to the donor incorporation in BaTiO3 ,
tween the core and the outer regions of the grains of the ce- a systematic, quantitative analysis of the grain growth
ramic samples. This hypothesis is a quite new variant in the related oxygen release in a wide range of the doping
long-standing discussion of the doping anomaly of BaTiO3 , level and with different donor dopants is the subject of
but, nevertheless, it seems to be a somewhat doubtful one. In our further work.
our opinion, the observed coincidence between the oxygen
release and the start of the rapid exaggerated grain growth
during sintering is a good argument for the occurence of the Acknowledgements
electronic donor compensation, if Drofenik’s model of the
grain growth anomaly is accepted. The authors would like to thank Dr. Helfried Näfe from
the Max Planck Institute für Metallforschung Stuttgart for
valuable discussion. Financial support from the Kultusmin-
4. Summary
isterium des Landes Sachsen-Anhalt and from the Fonds der
Chemischen Industrie is gratefully acknowledged.
The qualitative and quantitative determination of the
oxygen exchange between the barium titanate ceramic and
its ambient atmosphere both during the restoration of the
References
thermodynamic equilibrium and during non-equilibrium
processes like sintering as well as the determination of the
[1] J. Daniels, K.H. Härdtl, Philips Res. Rep. 31 (1976) 489.
onset temperatures of the corresponding defect reactions [2] N.-H. Chan, R.K. Sharma, D.M. Smyth, J. Am. Ceram. Soc. 64 (1981)
by means of the oxygen coulometry method were shown. 556.
Nominally undoped, acceptor- and donor-doped material [3] N.-H. Chan, R.K. Sharma, D.M. Smyth, J. Am. Ceram. Soc. 65 (1982)
was investigated. 167.
[4] N.-H. Chan, D.M. Smyth, J. Am. Ceram. Soc. 67 (1984) 285.
[5] R.J. Panlener, R.N. Blumenthal, J. Am. Ceram. Soc. 54 (1971) 610.
(1) In the case of “undoped” BaTiO3 , the change of the
[6] D.F.K. Hennings, Philips Res. Rep. 31 (1976) 516.
equilibrium oxygen vacancy concentration due to a [7] H.-J. Hagemann, D.F.K. Hennings, J. Am. Ceram. Soc. 64 (1981) 590.
changed equilibrium temperature between 1140 and [8] M.M. Nasrallah, H.U. Anderson, A.K. Agarwal, B.F. Flandermeyer,
1350 ◦ C at pO2 = 2.4 Pa was determined. Correspond- J. Mater. Sci. 19 (1984) 3159.
ing data calculated by a defect model [1] are in the [9] M. Drofenik, A. Popovic, L. Irmancnik, D. Kolar, V. Krasevec, J. Am.
same order. Systematic investigations of the oxygen ex- Ceram. Soc. 65 (1982) C203.
[10] K. Teske, W. Gläser, Microchim. Acta 1 (1975) 653.
change after the change of equilibrium parameters like [11] K. Beetz, Die geschlossene Festelektrolyt-Sauerstoffpumpe, Ph.D.
temperature and oxygen partial pressure of the ambient Thesis, Universität Karlsruhe, Fortschr.-Ber. VDI Reihe 8 Nr. 358,
atmosphere shall be done in the future to check existing VDI-Verlag, Düsseldorf, 1993.
defect models with respect of their validity as a whole or [12] H.T. Langhammer, Q.M. Song, H.-P. Abicht, J. Eur. Ceram. Soc. 21
for the correction of the values of the model parameters. (2001) 1893.
[13] H.-J. Hagemann, H. Ihrig, Phys. Rev. B 20 (1979) 3871.
(2) The oxygen exchange of Mn-doped BaTiO3 due to the
[14] S.B. Desu, E.C. Subbarao, Ferroelectrics 37 (1981) 665.
formation of oxygen vacancies caused by the change of [15] P. Moretti, F.M. Michel-Calendini, Phys. Rev. B 36 (1987) 3522.
the valence state of MnTi was detected. The amount of [16] B. Milsch, Phys. Status Solidi A 133 (1992) 455.
the expelled oxygen during the sintering of air-calcined [17] F. Ren, S. Ishida, S. Mineta, J. Ceram. Soc. Jpn. 102 (1994) 106.
specimens in a reducing atmosphere of pO2 = 2.4 Pa can [18] F.A. Kröger, H.J. Vink, Solid State Phys. 3 (1956) 307.
be explained by the complete transition from Mn4+ to [19] J.-Y. Kim, C.-R. Song, H.-I. Yoo, J. Electroceram. 1 (1997) 27.
[20] K.W. Kirby, B.A. Wechsler, J. Am. Ceram. Soc. 74 (1991) 1841.
Mn3+ in the doping range of 0.5 [MnTi ] 4.0 mol%.
[21] H.-P. Abicht, H.T. Langhammer, K.-H. Felgner, J. Mater. Sci. 26
(3) The oxygen release caused by the incorporation of the (1991) 2337.
donor into the BaTiO3 lattice related to the anomalous [22] Y.-S. Yoo, H. Kim, D.-Y. Kim, J. Eur. Ceram. Soc. 17 (1997) 805.
grain growth during sintering was proved for La and [23] M. Drofenik, J. Am. Ceram. Soc. 76 (1993) 123.
Nb which substitute for Ba and Ti, respectively. In the [24] D. Makovec, Z. Samardžija, U. Delalut, D. Kolar, J. Am. Ceram.
case of 0.5 mol% La-doped material about the half of Soc. 78 (1995) 2193.
[25] F.D. Morrison, D.C. Sinclair, A.R. West, J. Am. Ceram. Soc. 84 (2001)
the expected amount of oxygen release was detected 474.
if total donor incorporation and exclusive electronic