PAPER
www.rsc.org/dalton
| Dalton Transactions
A new family of oxide ion conductors based on tricalcium oxy-silicate
J. Manuel Porras-V´azquez,
aAngeles G. De la Torre,
´
aDavid Marrero-L´opez,
bEnrique R. Losilla
aand
Miguel A. G. Aranda*
aReceived 18th November 2005, Accepted 23rd February 2006 First published as an Advance Article on the web 10th March 2006
DOI: 10.1039/b516324b
Tricalcium oxy-silicates, Ca3(SiO4)O and Ca2.93Mg0.07(Si0.98Al0.02O4)O0.990.01, have been prepared as
crystalline single phases. Ca3(SiO4)O and Ca2.93Mg0.07(Si0.98Al0.02O4)O0.990.01have triclinic and
monoclinic structures, respectively. The samples show oxide anion conductivity with a small p-type electronic contribution under oxidizing conditions. At 1023 K, the oxide transport numbers range between 0.97 and 0.85 from reducing (dry 5%-H2–Ar/air gradient) to oxidizing (O2/air gradient)
conditions in the 1023–1173 K interval. The thermal analyses showed a large weight loss on heating due to the presence of water in the materials. The monoclinic compound has ionic conductivities higher than those of the triclinic stoichiometric oxy-silicate, as expected due to the introduction of oxide vacancies. Typical total conductivities for these un-optimised solids are 10−5–10−4S cm−1at
1100 K. These compounds may contain a small amount of water,∼0.05 H2O moles per chemical
formula, and they display an important proton contribution under a humidified atmosphere.
Introduction
Oxide ion conductors are an important group of ceramic materials that are involved in an increasing number of technological areas, for example, as electrolytes in solid oxide fuel cells (SOFCs), oxygen sensors or electrochemical oxygen pumps. SOFCs are considered as alternative electrical power generation systems with low emission of pollutants and high energy conversion efficiency.1–3Yttria stabilized zirconia, YSZ, is the electrolyte most
widely used in the commercial systems due to its high oxide ion conductivity at elevated temperatures (1173–1273 K). However, there is a great interest in the development of devices with lower operational temperatures to overcome collateral problems like difficulties in cell sealing or low lifetime of the components caused by the high operational temperature of YSZ. Several families of oxide ion conductors are being actively investigated for intermediate-temperature SOFCs. These materials will ideally work in the range 873–973 K and allow the use of cheaper construction materials and more reliable seals. We can highlight some families of oxide conductors which are being extensively studied: i) fluorite-type oxides such as Ce0.8Gd0.2O1.90.14 and
Bi0.75RE0.25O1.50.55 (RE= rare-earth); ii) perovskite-type oxides
such as La0.9Sr0.1Ga0.8Mg0.2O2.850.156 and BaCe0.8Gd0.2O2.90.17;
and iii) BIMEVOX-type oxides such as Bi2V0.86Ni0.14O5.290.21.8
There is also a lot of interest in the oxide conductivity properties of oxy-silicates, such as rare earth oxy-apatite La9.330.67(SiO4)6O2,9
with conductivity values larger than those of YSZ for some selected compositions.
On the other hand, tricalcium silicate, Ca3SiO5(C3S in cement
nomenclature), is an oxy-orthosilicate present in ordinary Port-land cements. This compound has been extensively studied due to its important hydraulic properties. Stoichiometric C3S exhibits
aDept. Qu´ımica Inorg´anica, Universidad de M´alaga, Campus Teatinos,
29071, M´alaga, Spain
bDept. Qu´ımica Inorg´anica, Universidad La Laguna, 38200, La Laguna,
Tenerife, Spain. E-mail: [email protected]
seven polymorphs when is heated10,11: three triclinic forms (T 1, T2,
T3), three monoclinic forms (M1, M2, M3) and one rhombohedral
(R). The phase transformations are T1↔T2↔T3↔M1↔M2↔
M3↔R at 893, 1193, 1253, 1263, 1333 and 1343 K, respectively.
The transformations at 893, 1253 and 1263 K are detectable by both differential thermal analysis (DTA) and laboratory X-ray powder diffraction (LXRPD),12 the T
2 ↔ T3 transformation is
only detectable by DTA, and M3 ↔ R transformation is only
confirmed by LXRPD. All these polymorphic transformations are of the displacive type and the change in the orientation of silicate tetrahedra plays a key role for determining the symmetry of these compounds.13–15Tricalcium silicate is better formulated as
Ca3(SiO4)O to show its oxy-silicate nature. The room-temperature
(RT) structure of stoichiometric triclinic Ca3(SiO4)O is known
from single crystal data13 and it is displayed in Fig. 1. As it is
shown in the structural chemical formula, some oxide anions are only bonded to the calcium atoms forming rows, see Fig. 1, which may display oxide conductivity properties.
Fig. 1 Crystal structure of triclinic Ca3(SiO4)O with the SiO4 groups shown as tetrahedra, calcium cations and oxide anions shown as small filled and large open spheres, respectively.
C3S forms a solid solution with an appreciable number of
elements and some polymorphs may be stabilised at room temperature by the presence of foreign ions in the structure. The R-form structure has been studied at room temperature using Sr2+as stabilizer16 and at high temperature using an aluminium
doped C3S.17Monoclinic polymorphs may be stabilised at room
temperature by the incorporation of Mg2+ and Al3+ into the
structure. In fact, ordinary Portland cements contain C3S with
the M3and/or M1polymorphs18–20due to the presence, among
others, of Mg2+ into the C
3S structure. There is interest in
knowing the crystal structures of these polymorphs in order to obtain accurate quantitative phase analyses of these important materials.15,21,22The M
3polymorph, also called alite, crystallises at
RT in a monoclinic subcell23but the true cell is much larger with
dimensions:a=33.1 A˚ ,b=7.0 A˚ ,c=18.5 A˚ andb=94.1◦(3× 1×2 superstructure).21
The aim of this work was to investigate the electrical properties of tricalcium oxy-silicate as part of our research in oxy-silicates and oxy-aluminates. To do so, we have prepared Ca3(SiO4)O
and Ca2.93Mg0.07(Si0.98Al0.02O4)O0.990.01. This last material was
synthesized to introduce oxide vacancies and to avoid the phase transitions on heating that may complicate this first study. The two compounds have been characterized by a set of electrochemical techniques and they behave as complex oxide anion conductors.
Experimental
Ca3(SiO4)O and Ca2.93Mg0.07(Si0.98Al0.02O4)O0.990.01were prepared
by the ceramic method, using the appropriate amounts of the high purity oxides and carbonates: SiO2(ABCR, quartz powder
99.31%), c-Al2O3 (Alfa, 99.997%), CaCO3 (Aldrich, 99%) and
4MgCO3·Mg(OH)2·5H2O (Aldrich, 99%). Precursors were ground
for 30 min in a Fritsch ball mill (model Pulverisette 7, 45 cm3
agate vessel containing 7 agate balls with a diameter of 15 mm) at 200 rpm with reverse rotation each 5 min and heated at 1273 K for 6 h in order to prepare∼3 g of sample by the following overall reaction:
(3−x)CaCO3+(x/5)4MgCO3·Mg(OH)2·5H2O+(1−2y)SiO2+(y/2)Al2O3
D
−−−−−→(Ca3−xMgx)O3(SiO2)1−2y(Al2O3)y+(15−x)/5CO2+(6x)/5 H2O
The resulting powders were reground in the planetary ball mill, pelletized (400 MPa,∼20 mm diameter and∼3 mm thickness) and heated in air at 1773 K for 6 h three times with intermediate grind-ing. Hereafter, Ca3(SiO4)O and Ca2.93Mg0.07(Si0.98Al0.02O4)O0.990.01
samples are labelled C3S-t and C3S-m, respectively.
Powder diffraction
The compounds were characterized by laboratory X-ray powder diffraction (LXRPD) at room temperature. The powder patterns were collected on a Siemens D5000 automated diffractometer, using graphite-monochromated Cu-Ka1,2 radiation. The samples
were loaded in flat aluminium holders and scanned between 20 and 60◦(2h) in 0.03◦steps, counting 20 s per step. All Rietveld24
analyses were done using the GSAS suite of programs.25
Thermal analysis
Thermogravimetric–differential thermal analyses (TG-DTA) were performed on a Pyris-Diamond Perkin-Elmer apparatus. The temperature was varied from RT up to 1273 K at a heating/cooling rate of 10 K min−1with a flux of atmospheric air of 80 cm3min−1
(provided by an air compressor), using a mass of around 60– 70 mg. The TG-DTA instrument was calibrated using standard samples and the equipment baseline, measured with two empty crucibles, was subtracted. Different heating/cooling cycles were carried out to study the thermal reversibility and reproducibility of the measurements.
Microstructural characterization
The sintered pellets morphology was studied using a JEOL SM 840 scanning electron microscope. The surfaces were polished with diamond spray (3lm) and thermally etched at 50 K below the sintering temperature for 15 min. Finally, the samples were metallised by gold sputtering for better image definition.
Electrical measurements
Electrical characterization was carried out on cylindrical pellets (∼10 mm in diameter and∼1 mm thick) obtained by pressing
∼0.1 g of sample with small average particle size [d50 ∼7 lm]
at 125 MPa, for 2 min. The pellets were sintered at 1873 K for 6 h at a heating rate of 10 K min−1. Fine powders were
obtained by grinding the initial powdered samples in a vibratory ball mill (Retsch, 2 zircona balls with a diameter of 12.25 mm and a vessel volume of 8 cm3) at 15 Hz for 40 min. After
that, the fine powders were deagglomerated with acetone in an agate mortar for 45 min. The optimum milling time was selected measuring the particle/agglomerate size distribution after different grinding times on a Malver Mastersizer laser diffraction analyzer. Electrodes were made by coating opposite pellet faces with METALORR 6082 platinum paste and heating to 1323 K
at a rate of 10 K min−1for 15 min in air to decompose the paste and to harden the Pt residue. Successive treatments were made to achieve an electrical resistance on both pellet faces lower than 1X.
Impedance spectroscopy data in four different flowing atmo-spheres (dry 5%H2–N2, wet 5%H2–N2, dry synthetic air and
wet synthetic air) were collected using a HP4284A impedance analyser over the frequency range from 20 Hz to 1 MHz with an applied voltage of 1 V. Electrical measurements were taken on heating and cooling in the temperature range of 673–1273 K every 50 K (accuracy of±1 K) at 10 K min−1with a delay time
of 45 min at each temperature to ensure thermal stabilization. Measurements were electronically controlled by the winDETA package of programs.26
High-temperature conductivity measurements as a function of oxygen partial pressure [p(O2) from air to∼10−20 atm] were
performed in a closed tube furnace cell. The p(O2) values were
monitored by using a YSZ oxygen sensor, placed next to the pellet in the cell. The conductivity was continuously recorded as a function ofp(O2). The process consists of flushing the system with
a dry 5%H2–N2gas mixture for 12 h at 1173 K to reach a minimum
in oxygen activity inside the furnace and to ensure that the sample is close to equilibrium at those conditions. Then, the flushing was
switched off and the oxygen partial pressure slowly raised back to atmospheric pressure by free diffusion, since the system is not airtight. Each isothermal cycle took over 84 h to complete.
The electronic conductivity was determined by a modified electromotive force (emf) method, taking into account electrode polarization.27,28This modification of the classical emf technique
eliminates possible errors in the determination of ion transference numbers, arising due to electrode polarization. These errors are not negligible for electrolyte materials which have relatively low electronic conductivity.28,29 The ionic transport numbers t
o
were measured under a p(O2) gradient of pure O2/air and dry
5%H2–Ar/air, using a continuous flux of these gases in the 973–
1173 K temperature range. A YSZ tube was used to measure the theoretical emf under these conditions,Eth. The emf observed in
the sample,Eobs, was measured with an external variable resistance
RM, in parallel to the measuring cell, varying from 100Xto 20 kX.
Experimental data were fitted with the equivalent circuit (eqn (1)) proposed by Gorelov.27 In this case, the ionic resistanceR
o, the
electronic resistanceRe and the polarization resistance Rg are
related to the emf values (EthandEobs) from the relation:
Eth
Eobs
−1=(Ro+Rg)
1
Re
+ 1
RM
(1)
The dependence of (Eth/Eobs−1)vs1/RMis a linear plot with
slope (Ro + Rg) and the intercept of the (1/RM)-axis is equal
to (−1/Re). Total resistanceRT was determined independently
by impedance spectra data. Oxide ion transport numbers were calculated as:to=1−RT/Re.
Results and discussion
Synthesis and crystal structures
Ca3(SiO4)O and Ca2.93Mg0.07(Si0.98Al0.02O4)O0.990.01have been
pre-pared as highly crystalline compounds. Their powder patterns were analysed by the Rietveld method using the structural descriptions previously reported.13,21Only the overall parameters
were refined: histogram scale factor, background coefficients, unit cell parameters, zero-shift error and peak shape pseudo-Voigt coefficients. Atomic parameters were not refined. The final unit cell parameters werea=11.633(1) A˚ ,b=14.210(1) A˚ ,c=13.653(1) A˚ ,
a = 104.835(3)◦,b = 94.479(3)◦,c = 90.163(4)◦ and V/Z =
120.81(2) A˚3for C
3S-t; anda=33.089(3) A˚ ,b=7.0338(2) A˚ ,c=
18.514(2) A˚ ,b=94.169(5)◦andV/Z=119.33(3) A˚3for C 3S-m. As
expected, the normalised volume for C3S-m is smaller than that of
C3S-t due to the magnesium substitution. The Rietveld refinement
plots for C3S-t and C3S-m are shown in Fig. 2a and b, respectively.
These plots indicate that both compounds are single phase. The three thermo-mechanical treatments at high temperature ensure the absence of free lime, CaO, and dicalcium silicate, Ca2SiO4, as
side-phases.
At this point, two issues must be clarified. Firstly, it is necessary to highlight that the oxide vacancies, shown in the structural chemical formula of C3S-m, are only indirectly established from
the absence of aluminium compounds as secondary crystalline phases. The very complex superstructure of C3S-m, with a unit
cell volume of 4297.6 A˚3, precludes the direct determination of
the oxide vacancies even from a joint refinement of neutron and synchrotron powder diffraction data.21Secondly, the Mg2+doping
Fig. 2 Observed (crosses), calculated (full line) and difference (bottom) Rietveld refined LXRPD patterns for a) triclinic Ca3(SiO4)O and b) for monoclinic Ca2.93Mg0.07(Si0.98Al0.02O4)O0.990.01. The vertical bars are the allowed Bragg reflections.
is unlikely to help the possible oxide conductivity as its ionic radius is smaller than that of Ca2+. However, this doping was chosen to
ensure the formation of the M3form which does not undergo any
phase transition up to 1300 K. This has been confirmed by a DTA study. Finally, Ca3(SiO4)O doping with Al3+is known.30
Electrical characterization
The sintering conditions for the pellets led to dense specimens with compactions ranging between 95 and 98% of the theoretical value (taking into account the mass, volume and the crystallographic density of the pellets). No weight losses were detected during the sintering process. Fig. 3a and b show SEM micrographs for C3S-t
and C3S-m, respectively. The microstructure is well developed with
an average grain size close to 2lm for C3S-t and 10lm for C3S-m.
The porosities of the pellets were low but not negligible as can be observed in Fig. 3. No indication of liquid phase formation or phase segregation at the grain boundary were found in any of the studied samples. Additionally, no contamination due to the ball-milling process was detected in the sintered pellets.
Fig. 3 SEM micrograph of polished and thermally etched surface of sintered pellets for a) Ca3(SiO4)O and b) for Ca2.93Mg0.07(Si0.98Al0.02O4 )-O0.990.01compositions.
Representative impedance data for C3S-m at two temperatures
are shown as impedance complex plane plots in Fig. 4a and b. Similar plots were obtained for C3S-t. At low temperatures,i.e.
673 K, the impedance spectrum displays an almost non-deformed semicircle with an associated capacitance at the maximum of 1.8 pF cm−1. At high temperatures, i.e. 973 K, a broad arc is
observed where, at least, two components can be distinguished. These components are due to the presence of several contributions to the electrical response of the pellets.
In order to investigate the electrical microstructure of the pellets and, in particular, to determine whether the overall pellet resistances represented the bulk resistance of the grains or whether they were influenced by grain boundary effects, the experimental data were replotted as the imaginary parts of the impedance,−Z, and as electric modulus,M, against log(frequency), see Fig. 4c
and d. At low temperature, the maxima of both curves are almost coincident, which indicates that the impedance peak is associated with the same RC element responsible for the modulus peak. The associated capacitances (∼1.8 pF cm−1forMand−Z) are
characteristic for the bulk response. Hence, the semicircle of the complex impedance plane plot in Fig. 4a mainly corresponds to the bulk contribution and the intersection at low frequency with theZ
axis is the bulk impedance, apparently free from grain boundary contributions. At higher temperature, although the maxima of both curves are almost coincident, the−Zplots shows a broad peak with a smooth shoulder at lower frequencies (marked with an asterisk in Fig. 4d). These maxima (∼2.0 pF cm−1for−Zand
M) are due to the bulk/intrinsic response of the sample, and the
shoulder (with an associated capacitance of∼6.6 pF cm−1) is due
to a second contribution. This relaxation is difficult to assign and can be due to a thick grain–grain boundary contribution or to a second minor conducting phase.
The total pellet conductivities were obtained from the intercept of the low frequency end on theZaxis. These conductivity values,
obtained in air, are plotted in a traditional Arrhenius format in Fig. 5 for C3S-t and C3S-m. A plot of logrvs1000/Tshould give a
straight line if the activation energyEais temperature-independent
and only one conduction mechanism takes place. The Arrhenius plots clearly show two regions with different activation energies. Below ∼900 K, the Arrhenius plots for both compositions are linear with activation energies of 0.59 and 0.53 eV for C3S-t and
C3S-m, respectively. Above∼900 K, the activation energy increases
to 1.03 and 0.93 eV for C3S-t and C3S-m, respectively. Some
possible explanations for these changes in the activation energy
Fig. 5 Arrhenius plots of log(rT) for Ca3(SiO4)O (♦) and Ca2.93Mg0.07 -(Si0.98Al0.02O4)O0.990.01(), and of log(rbulk) for C3S-m ().
Fig. 4 Complex impedance plane plot for Ca2.93Mg0.07(Si0.98Al0.02O4)O0.990.01at a) 673 K and b) 973 K in air. The full line is the data fit using the equivalent circuits described in the text. Spectroscopic plots of−ZandMvslog(f) for C
3S-m at c) 673 K and d) 973 K in air.
are: i) a phase transition, but this is ruled out by the LXRPD and DTA studies for C3S-m; ii) an enhanced grain boundary
conductivity at high temperature, although this is not expected since, in general, the grain boundary resistance is small at high temperatures compared to that of the bulk; and iii) a change in the conduction mechanism, expected for cases when different charge carriers predominate for low and high temperatures, respectively.
To separate the two contributions (bulk and grain boundary), complex impedance spectra were analysed by a non-linear least squares fitting method, using equivalent circuits with the Zview program.31 The impedance data at low temperature (673–823 K
range) have been fitted using an equivalent circuit formed by one (RQ) element, corresponding to the bulk responses, where
Ris the resistance in parallel with the pseudocapacitance,Q. At higher temperatures, the impedance data have been fitted using an equivalent circuit formed by the association of two (RQ) elements in series: (RbQb)(RgbQgb), corresponding to the bulk and grain
boundary contributions. These parameters have been determined in the 873–1173 K temperature range where both responses are detectable. The calculated spectrum at 673 and 973 K is shown (solid line) in Fig. 4a and b as an example of the goodness of the fits. These calculations gave the bulk conductivities which are also plotted in Fig. 5 (solid squares) for C3S-m. As can be
seen, there is not a typical Arrhenius behaviour forrbulk which indicates a possible mixed conduction mechanism. In order to clarify the electrical behaviour, the samples were further studied under different atmospheres.
Mixed ionic (oxide-proton) conductivity. Tricalcium
oxy-silicates were studied by thermogravimetric analysis under air flow. Fig. 6 shows the TGA curve for C3S-m (a similar curve
was obtained for C3S-t which is not shown). As can be seen, the
first cycle of the TGA study shows a weight loss that takes place in two steps centred close to 750 and 890 K. The overall mass loss is 0.42% for C3S-m (the value for C3S-t was 0.24%). The
Fig. 6 TG curves for (top) Ca2.93Mg0.07(Si0.98Al0.02O4)O0.990.01(two heat-ing and coolheat-ing cycles) and (bottom) Ca2SiO4(a heating and cooling cycle) as a standard. The measurements were carried out under atmospheric air flow.
weight loss can be due to water absorbed in the sample which is characteristic of proton conductors. Under this reasonable assumption, the overall water content of these samples must be close to 0.052 and 0.030 moles of H2O per chemical formula
for C3S-m and C3S-t, respectively. A very small weight gain is
observed near 800 K which may be due to oxygen uptake by the defects present in C3S-m. During the cooling process, the initial
state is not exactly recovered as the thermal analyses were carried out under atmospheric air flows at a rapid heating/cooling rate of 10 K min−1, but some water uptake around 700 K is clearly observed. A second cycle for the TGA analysis was carried out to check the dynamic nature of the water uptake/release. Under these non-equilibrium conditions, the absorbed water is again released close to 860 K. These small mass changes are not due to the equipment or to an improper correction of the baseline. A TGA curve for dicalcium silicate, Ca2SiO4, was recorded, as a standard
similar sample, under exactly the same conditions as those for tricalcium oxy-silicates. Ca2SiO4does not have oxide groups and
it does not show the weight changes displayed by C3S-m.
The water content in tricalcium oxy-silicate samples and the anomalous electrical behaviour suggest a proton conductivity contribution to the total conductivity which was further studied by impedance spectroscopy under dry and wet atmospheres. High temperature proton conductivity has been reported many times in some oxide materials. For example, complex oxides with the perovskite structure, such as BaCeO3-based materials,32SrZrO333
or La4(Ga2−xTixO7+x/21−x/2)O234cuspidines present variable
pro-ton conductivity in humidified atmospheres. Fig. 7a shows the
Fig. 7 (a) Complex impedance plane plots at 973 K for Ca2.93Mg0.07 -(Si0.98Al0.02O4)O0.990.01under different flowing atmospheres. (b) Arrhenius plots of log(rT) for the same sample and atmospheres.
impedance spectra at 973 K for C3S-m under a constant flow
of dry 5%H2–N2, wet 5%H2–N2, dry synthetic air and wet
synthetic air. It is obvious from that figure that the sample has the highest conductivity under wet synthetic air and the lowest conductivity under dry 5%H2–N2. The Arrhenius plots of the total
conductivities are shown in Fig. 7b. The conductivity in wet air is almost two orders of magnitude higher than in dry air in the low temperature range. Furthermore, this proton contribution is very important even at high temperatures. This indicates that a minor fraction of water remains in the structure even at very high temperature under a humidified atmosphere in agreement with thermogravimetric measurements.
Therefore, the curvatures in the Arrhenius plots, Fig. 7, may be explained as a consequence of the change in the main conduction mechanism from predominantly proton conduction below∼900 K to predominantly oxide conduction above that temperature.
The ionic conductivity for a pure ionic conductor is independent of the oxygen partial pressure in a wide range ofp(O2) values.
However, the conductivity increases for a mixed ionic–electronic conductor as p(O2) increases or decreases, depending on the
predominant electronic contribution (p- and n-type, respectively). The conductivity data for C3S-m at 1073 and 1173 K as a
function ofp(O2) are shown in Fig. 8a. This curve shows a p-type
contribution at high oxygen partial pressure, but an anomalous behaviour is observed at very low reducing conditions with an unexpected drop in conductivity. This drop of conductivity can not be associated to a phase transition or material degrada-tion, because samples reduced under 5%H2–N2were studied by
LXRPD and the patterns were similar to those of the pristine materials.
Fig. 8 (a) Overall conductivity data at 1073 and 1173 K as a function of oxygen partial pressure for Ca2.93Mg0.07(Si0.98Al0.02O4)O0.990.01. The solid lines are the fitted results obtained by applying eqn (8) to the data. (b) Dependence of the ion oxide transference number,to, for the same sample under oxidizing (dry O2/air gradient) and reducing (dry 5%H2–Ar/air gradient) atmospheres. The inset shows the thermal dependence of overall ionic resistance (Ro), polarization resistance (Rg) and electronic resistance (Re) under oxidizing conditions.
This drop in conductivity at lowp(O2) values has been already
reported in other oxide electrolytes which also display proton con-ductivity. For instance La2Zr2O7 with the pyrochlore structure35
and SrZrO3with the perovskite structure36,37have this behaviour.
This effect is due to a drop in the water partial pressure as decreasing the oxygen partial pressure gives rise to a decrease in proton conductivity.
In oxide proton conductors the water uptake occurs as follows, taking into account the defect chemistry and the Kr ¨oger–Vink notation where V¨o is an oxide anion vacancy and Oxo is an
interstitial oxide anion:38
V¨o+H2O→Oxo+2H
+ (2)
The equilibrium constant of reaction is:
Kw=[H•i]2[V¨o]−1P−1w (3)
This last equation indicates that the proton concentration varies with the water vapour partial pressurePwfollowing a dependence
of (Pw)1/2.
In addition, the contributions to the conductivity of the different charge carriers can be expressed as:
ri=zieli (4)
where e is the electron charge,zithe valence andlithe mobility of
speciesi. The total conductivity for the different contributions is expressed as a function of the water and oxygen partial pressure as:
rt=rion+rp+rH=rion+r*pPO2 1/4+r*
HPw1/2 (5)
whererionis the anion-oxide conductivity and it is supposed to be constant with oxygen partial pressure,rp* the p-type conductivity that it is assumed to present a typical power law dependence of the form (PO2)
1/4andrH* is the proton conductivity at an oxygen
partial pressure of 1 atm and it varies with water vapour partial pressure of the form (Pw)1/2.
The water vapour/H2ratio increases during the re-oxidation in
the electrochemical cell, as described by the mass action law shown in the following reaction:
2H2+O2
Kw −−−→
←−−−2H2O (6)
and an additional relation can be obtained between the water and the oxygen partial pressure for nearly constant hydrogen content containing gas species in a closed atmosphere:Pw+PH2 =Pw,m,
wherePH2andPware hydrogen and water partial pressure andPw,m
is the maximum water vapour attained in oxidizing conditions. In this case, one easily obtains the following relation35:
Pw=Pw,m/[1+(KwPO2)
−1/2] (7)
This last equation allows the calculation of the water partial pressure as a function of the oxygen partial pressure and it is suitable for studying proton conductivity in oxides. The total conductivity can be expressed in terms of the oxygen partial pressure substituting eqn (7) into eqn (5) which yields.
rt=rion+rp*PO21/4+rH*Pw,m1/2[1+(KwPO2)−1/2]−1/2 (8)
Experimental data obtained from Fig. 7 were fitted with eqn (8) and it provides a satisfactory fitting with the model proposed. This allows us to explain the anomalous behaviour at lowp(O2)
values as a consequence of the drop in the water partial pressures and the decreasing proton conductivity.
The dependence of the oxide ion transport number (to) under
oxidizing (dry O2/air gradient) and reducing (dry 5%H2–Ar/air
gradient) conditions as a function of temperature is shown in Fig. 8b for C3S-m. As can be observed,toshows a slight decrease
under both atmospheres: from 0.89 at 1023 K to 0.86 at 1173 K in dry O2/air gradient and from 0.97 at 1023 K to 0.93 at
1173 K in dry 5%H2–Ar/air gradient. This behaviour can be
mainly ascribed to a small p-type contribution predominant under oxidizing conditions. The values ofto under dry 5%H2–Ar/air
are larger than under a dry O2/air gradient, which confirms the
results obtained from the conductivity study as being a function of
p(O2) (Fig. 8a). It must be underlined that the p-type contribution
is minimised in a dry 5%H2–Ar/air gradient environment and
therefore theto values are higher under dry 5%H2–Ar/air than
under a dry O2/air gradient.
An example of the thermal dependence ofRo,RgandReis given
in the inset of Fig. 8b for dry O2/air gradient. The polarization
resistance has the same order of magnitude as the electronic one in the studied temperature range. Hence, the electrode polarization has a significant effect and therefore the transport numbers determined by the classical emf method are negatively influenced by the electrode polarization. For example, thetovalue obtained
from the classical emf method under a dry 5%H2–Ar/air gradient
at 1073 K was 0.81, after the correction of the polarization effect by the modified emf method took the value of 0.96.
This emf study demonstrate that C3S materials present nearly
pure ion oxide conductivity under very reducing conditions in a dry atmosphere. A p-type electronic contribution appears as the
p(O2) increases, rendering an ionic transport number of 0.85 under
oxidizing conditions. It must be underlined that thetovalues under
wet atmospheres must be much smaller than those reported here due to a proton conductivity contribution.
Conclusions
Tricalcium oxy-silicates, Ca3(SiO4)O and Ca2.93Mg0.07(Si0.98Al0.02O4
)-O0.990.01, are complex oxide ion conductors. Monoclinic trical-cium silicate has an ion conductivity higher than that of the triclinic stoichiometric silicate. The measured oxide ion transport numbers ranged between 0.86 and 0.97 for temperatures in the 1023–1173 K interval. In addition to the oxide ion conduction, there is an electronic p-type contribution at high oxygen partial pressure values. Finally, a significant proton contribution to the overall conductivity is also taking place as shown in the impedance study under wet and dry atmospheres.
Acknowledgements
The authors thank Dr M.A. Rodr´ıguez and Dr P. Pena (ICV-CSIC, Madrid), for support using the scanning electron microscope and for their help in the sintering process. Financial support from MAT2003-7483-C2-1 research grant is gratefully acknowledged.
References
1 R. M. Ormerod,Chem. Soc. Rev., 2003,32, 17.
2 A. Boudghene Stambouli and E. Traversa, Renewable Sustainable Energy Rev., 2002,6, 433.
3 B. C. H. Steele and A. Heinzel,Nature, 2001,414, 345.
4 S. Torrens, N. M. Sammes and G. A. Tompsett,Solid State Ionics, 1998,
111, 9.
5 N. M. Sammes, G. A. Tompsett, H. N¨afe and F. Aldinger,J. Eur. Ceram. Soc., 1999,29, 1801.
6 T. Ishihara, H. Matsuda and Y. Takita,J. Am. Chem. Soc., 1994,116, 3801.
7 N. Taniguchi, E. Yasumoto, Y. Nakagiri and T. Gamo,J. Electrochem. Soc., 1998,145, 1744.
8 F. Krok, I. Abrahama, D. Bango, W. Bogusz and J. A. G. Nelstrop,
Solid State Ionics, 1998,111, 37.
9 S. Nakayama, T. Kageyama, H. Aono and Y. Sadaoka,J. Mater. Chem., 1995,5, 1801; P. R. Slater, J. E. H. Sansom and J. R. Tolchard,Chem. Rec., 2004,4, 373 and references therein.
10 H. F. W. Taylor,Cement Chemistry, Thomas Telford, London, 1997, vol. 5.
11 F. Dunstetter, M.-N. de Noirfontaine and M. Courtial,Cem. Concr. Res., 2006,36, 39.
12 M. Bigare, A. Guinier, C. Mazieres, M. Regourd, N. Yannaquis, W. Eysel, TH. Hahn and E. Woermann,J. Am. Ceram. Soc., 1967,50, 609.
13 N. I. Golovastikov, R. G. Matveeva and N. V. Belov, Sov. Phys. Crystallogr. (Engl. Transl.), 1975,20, 441.
14 M. Bigare, A. Guinier, C. Mazieres, M. Regourd, N. Yannaquis, W. Eysel, Th. Hahn and E. Woermann,J. Am. Ceram. Soc., 1967,11, 609. 15 F. Nishi, Y. Takeuchi and I. Maki, Z. Kristallogr., 1985, 172,
297.
16 A. M. Il’inets, Y. A. Malinovskii and N. N. Nevskii,Sov. Phys. Dokl. (Engl. Transl.), 1985,30, 191.
17 F. Nishi and Y. Takeuchi,Z. Kristallogr., 1984,168, 197.
18 J. C. Taylor, I. Hinczak and C. E. Matulis,Powder Diffr., 2000,15, 7. 19 A. G. De La Torre, A. Cabeza, A. Calvente, S. Bruque and M. A. G.
Aranda,Anal. Chem., 2001,73, 151.
20 M.-N. de Noirfontaine, F. Dunstetter, M. Courtial, G. Gasecki and M. Signes-Frehel,Cem. Concr. Res., 2006,36, 54.
21 A. G. de la Torre, S. Bruque, J. Campo and M. A. G. Aranda,Cem. Concr. Res., 2002,32, 1347.
22 A. G. de la Torre and M. A. G. Aranda,J. Appl. Crystallogr., 2003,36, 1169.
23 W. G. Mumme,Neues Jahrb. Mineral., 1995,4, 145. 24 H. M. Rietveld,J. Appl. Crystallogr., 1969,2, 65.
25 A. C. Larson and R. B. von Dreele,GSAS program, Rep. No. LA-UR-86748, Los Alamos National Laboratory, Los Alamos, CA, 1994. 26winDETA, Novocontrol GmbH, Hundsangen, Germany, 1995. 27 V. P. Gorelov,Elektrokhimiya, 1988,24, 1380.
28 I. P. Mazorau, D. Marrero-L ´opez, A. L. Azula, V. V. Kharton, E. Tsipis, P. N ´u ˜nez and J. Frade,Electrochim. Acta, 2004,49, 3517.
29 V. V. Kharton, A. P. Viskup, F. M. Figueiredo, E. N. Naumovich, A. A. Yaremchenko and F. M. B. Marques,Electrochim. Acta, 2001,46, 2879. 30 D. Stephan and S. Wistuba,J. Eur. Ceram. Soc., 2005,26, 141 and
references therein.
31 D. Johnson,ZView, a software program for IES analysis, Version 2.16, Scribner Associates, Inc., Southern Pines, NC, 2002.
32 H. Iwahara, H. Uchida, K. Ono and K. J. Ogaki,Electrochem. Soc., 1988,135, 529.
33 T. Yamija, H. Susuki, T. Yogo and H. Iwahara,Solid State Ionics, 1992,
51, 101.
34 M. C. Mart´ın-Sede ˜no, D. Marrero-L ´opez, E. R. Losilla, L. Le ´on-Reina, S. Bruque, P. N ´u ˜nez and M. A. G. Aranda,Chem. Mater., 2005,17, 5989.
35 J. A. Labrincha, J. R. Frade and F. M. B. Marques,Solid State Ionics, 1997,99, 33.
36 J. A. Labrincha, J. R. Frade and F. M. B. Marques,Solid State Ionics, 1993,61, 71.
37 J. A. Labrincha, F. M. B. Marques and J. R. Frade,J. Mater. Sci., 1995,
77, 2785.
