1
Formation of protonic defects in perovskite-type oxides with redox-active acceptors: case study on Fe-doped SrTiO3

2
The thermodynamics of water incorporation into Fe-doped SrTiO3 was investigated by thermogravimetric measurements.

3
Changes in valence states of redox-active dopant ions (Fe3+/Fe4+) with water vapor pressure were taken into account in the defect chemical analysis.

4
The proton solubility was significantly enhanced by the presence of the redox centers.

5
The hydration enthalpies and entropies were −60 kJ mol−1 and −122 J mol−1 K−1.

6
The defect chemical model was applied to describe the water vapor dependence of the electrical conductivity in mixed ionic and electronic conducting Fe-doped SrTiO3 single crystals.

Introduction

7
Many perovskite-type oxides (ABO3, A = Sr, Ba, B = Ce, Zr, Ti) doped with acceptors, e.g., Gd, Y, and Sc, have been reported to exhibit high proton conductivity in humid atmospheres.1

8
The water incorporation into these vacancy-dominated oxides introduces protons as hydroxide defects as follows:2where KH2O is the mass action constant for hydration.

9
Besides proton electrolytes with high and prevailing protonic conductivity, mixed conductors exhibiting both electronic and protonic conductivity are of considerable interest in view of potential applications as hydrogen separation membranes or electrodes.1,3–5

10
The presence of proton, oxygen ion and electronic carriers is interesting for catalysis and also favorable for a partially self-regulated fuel cell concept.6

11
The equilibrium concentrations of defects responsible for the transport properties in the mixed conducting perovskites depend on both oxygen and water vapor partial pressures as well as on temperatures.

12
Fe-doped SrTiO3 may be used as a model mixed conducting perovskite system to develop an accurate description of the thermodynamics and kinetics of water incorporation, since the thermodynamic parameters in oxygen incorporation and in intrinsic redox reactions are well established7–10 (see Table 1).

13
In Fe-doped SrTiO3 the dopant is redox-activeThe ‘association’ between negatively charged Fe3+ defects and the holes to form Fe4+ becomes increasingly thermodynamically favorable as temperature falls (exothermic association) and as oxygen partial pressure increases (oxidation).11

14
The electron transition in eqn (2) is the predominant optical excitation process in the visible range and the ‘color’ of the specimens represents the distribution of redox states (Fe3+/Fe4+).8,11–13

15
In previous studies of water incorporation in Fe-doped SrTiO3, the iron ions on Ti sites were assumed to have a fixed +3 valency.14–16

16
In the present study we performed a defect chemical analysis considering the possible change of the ratio Fe3+/Fe4+ with water vapor pressure as well as with temperature and oxygen partial pressure.

Defect chemistry of Fe-doped SrTiO3 in water vapor

17
The defect chemical analysis in this study refers to oxidizing atmosphere, in which possible reduced hydrogen species14–18 are not concerned, and OH˙O according to eqn (1) is the only hydrogen species to be considered.

18
The concentration of excess electrons [e′] (=KB[h˙]−1) is negligible under these conditions and oxygen incorporation into the vacancy-dominated oxides introducing electronic holes is considered:The valence states of Fe dopants introduced by eqn (2), neglecting Fe2+ observed in strongly reducing atmosphere,19 give[Fe′Ti] + [FeTi×] = mFe = Constant,where mFe is the total Fe concentration.

19
In the temperature range of concern (>400 °C) associates of acceptors [Fe′Ti] with oxygen vacancies such as {Fe′Ti V˙˙O} 20 or with protonic defects {Fe′Ti OH˙O}16 are negligible.

20
Then using the electroneutrality condition:[Fe′Ti] = 2[V˙˙O] + [OH˙O] + [h˙].the defect concentrations can be computed from eqns (1)–(3) as a function of PO2, PH2O, and T.

21
The effective acceptor concentration, e.g.[Fe′Ti], follows from[Fe′Ti]3 + (B − 2A)[Fe′Ti]2 + (4mFeA − mFeB)[Fe′Ti] − 2m2FeA = 0where and .

22
Fig. 1b shows the simulated defect concentrations as a function of PH2O in the logarithmic scale (top) and in the linear scale (bottom) for mFe = 1020 cm−3, T = 400 °C and PO2 = 1 atm, using the mass action constants from the literature and the hydration constant obtained in this study (which will be described later) (Table 1).

23
The change in the effective dopant concentration [Fe′Ti] with PH2O for given PO2 and T according to eqn (6) should be noted.

24
[Fe′Ti], which is only a few tenths of mFe at low PH2O, increases with PH2O until saturated to mFe.

25
Figs. 1a and c represent the cases of fixed valence dopants A′Ti in amounts corresponding to [Fe′Ti] values of low PH2O and at high PH2O, respectively.

26
Under ‘dry’ conditions (PH2O < 10 Pa), the acceptors [Fe′Ti] (or [A′Ti]) are largely compensated by oxygen vacancies viz., [Fe′Ti] ≅ 2[V˙˙O], and the concentration of the minority protonic defects [OH˙O] increases as according to eqn (1).

27
At very high water vapor pressure, most of the oxygen vacancies are occupied by hydroxide ions so that the dopants [Fe′Ti] (or [A′Ti]) are now largely compensated by the protonic defects, [Fe′Ti] ≅ [OH˙O], whereby the concentrations of the minority defects [V˙˙O] and [h˙] decrease as and , respectively.

28
The presence of the two PH2O regimes described above is a general defect situation but the consequence of the change in the effective dopant concentration [Fe′Ti] between the two regimes should be noted.

29
Along with increasing [Fe′Ti], the concentration of the major counter-charge defects, [OH˙O] increases.

30
The increase beyond the point at which most of the V˙˙O present in the low PH2O have been occupied occurs by the redox reaction:which is facilitated by the valence change of iron dopant Fe4+ to Fe3+.

31
The reaction (7) represents proton incorporation but, in contrast to reaction (1), not water incorporation.

32
Consequently the protonic defects OH˙O incorporated by water vapor exceed twice the consumed vacancies, as represented more generally by Δ[OH˙O] ≅ −2Δ[V˙˙O] + Δ[Fe′Ti], according to the electroneutrality equation (see the bottom graphs of Fig. 1).

33
It should be noted that the enhanced proton solubility due to the redox-active Fe dopants is experimentally observable in the viable water pressure range e.g. 0 to 20 mbar PH2O.

34
The change in the ratio Fe3+/Fe4+ with water incorporation was also monitored by optical spectroscopy.21

35
In the following sections, we will show some consequences of this on the thermodynamic and kinetic properties of Fe-doped SrTiO3 in water vapor.

Experimental

36
For the thermogravimetric measurements 1 mol% Fe-doped SrTiO3 specimens were prepared by conventional ceramic processing.

37
A mixture of SrCO3 (99.996%, ABCR, Germany), TiO2 (99.9%, Aldrich, USA) and Fe2O3 (99.9%, Ventron, India) was wet-milled and calcined twice (at 1250 and 1300 °C for 1 h).

38
The X-ray diffraction pattern confirmed the formation of the single SrTiO3 phase.

39
The specimens for thermogravimetry were powder compacts (prepared by cold isostatic pressing) and sintered pellets (1450 °C for 8 h in N2) for respective batches of ca.

40
2.3 g.

41
Thermogravimetric analysis was carried out with a magnetic suspension balance (Rubotherm, Germany, resolution 2 μg).

42
The samples were loaded into a small alumina crucible, which was connected to the balance via platinum wire and a quartz rod.

43
The specimens were first equilibrated in a dry atmosphere (102 or 103 ppm O2 balanced with Ar) and then water vapor is introduced using mixture of dry gas and wet gas (20 mbar H2O) High vapor pressures above 20 mbar were obtained by employing a tubing pump (Isamatec, Switzerland) and a custom-made evaporator.

44
The oxygen and water partial pressure were monitored by a zirconia sensor (Cambridge Sensotec, UK) and a hygrometer (Panametrics, UK).

45
The buoyancy effect was corrected by dummy measurements.

46
Fe-doped SrTiO3 single crystal (100) plates (6 × 6 × 1 mm3) with doping contents (mFe) of 5 × 1018 and 5 × 1019 cm−3 (Frank & Schulte GmbH, Essen, Germany) were used to measure the conductivity variation with water incorporation at 475 °C at 1 atm PO2.

47
For each doping content, two specimens with different electrode configurations were prepared: YBa2Cu3O6+δ films on the large area surfaces (6 × 6 mm2); and Pt paste on the small area surfaces (6 × 1 mm2).

48
Impedance response was recorded using a LCR meter (4248A, Hewlett-Packard, USA).

Results and discussion

49
Fig. 2 shows an example of thermogravimetric measurements.

50
The weight change of the strontium titanate powder compact was monitored as the water vapor pressure changed stepwise from ∼0 mbar (dry gas) to 4 mbar and to 20 mbar and then back to 0 mbar at 450 °C.

51
Considering the buoyancy effect of ∼15 μg from the dummy measurement, the weight change of the specimen was ∼1.1 × 102 μg at 20 mbar PH2O.

52
In the case of fixed-valence acceptors (A), 2[V˙˙O] + [OH˙O] ≅ [A′Ti], the increase of proton concentration corresponds to the decrease of oxygen vacancy concentration, i.e. Δ[OH˙O] ≅ −2Δ[V˙˙O].

53
Neglecting proton concentration in nominally dry gas (PH2O < 10 Pa) [OH˙O]wet ≅ Δ[OH˙O].

54
This quantity can be obtained from the thermogravimetric results according towith w and Δw being sample weight and weight change, MH2O molecular mass of water and MSTO molar mass of SrTiO3.

55
The mass action constant of hydration in eqn (1) can be written asWhen the water vapor pressure change leads to a significant change in effective dopant concentration [Fe′Ti], as shown in Fig. 1, Δ[OH˙O] ≅ [OH˙O]wet = [Fe′Ti]wet − 2[V˙˙O]wet according to the electroneutrality condition.

56
Thus the mass action constant of hydration can be expressed as

57
When the effective dopant concentration significantly increases with water incorporation, [Fe′Ti]wet > [Fe′Ti]dry ≅ 2[V˙˙O]dry, eqn (10), rather than eqn (9), gives more accurate estimates.

58
The weight change measured by thermogravimetry is used to obtain [V˙˙O]wet:where MO atomic mass of oxygen and hydrogen mass is neglected (cf. eqn (8)).

59
[Fe′Ti]wet can be estimated from [V˙˙O]wet, the experimental parameters (mFe, PO2), and known parameters of KO2 and KFe:The symbols in Fig. 3 thus represent defect concentrations that are estimated from the thermogravimetric measurements in Fig. 2 using known mass action constants KO2 and KFe (Table 1).

60
The averaged KH2O value was found to be 2.83 × 1015 Pa−1cm−3.

61
The lines then represent the calculated concentration using the complete set of mass action constants including KH2O.

62
Similarly KH2O values at different temperatures were obtained as represented in Fig. 4.

63
As well as the powder compact specimens in Fig. 2, sintered pellets were also investigated.

64
The weight change decreases with increasing temperatures; high water vapor pressures of 20 and 100 mbar were used for measurements at 600 °C in order to obtain measurable mass changes.

65
The measurement of the sintered pellet showed that there is not a significant amount of adsorbed water at the surface of powder specimens.

66
Fig. 4 showed that all of the data fall roughly on a line regardless of the oxygen partial pressures, the water vapor pressure ranges, and the sample types.

67
From an Arrhenius fit, a hydration enthalpy, ΔHH2O°, of −60 ± 12 kJ mol−1 (−0.63 ± 0.13 eV) and a hydration entropy, ΔSH2O°, of −122 ± 9 J mol−1 K−1 (−0.0013 ± 0.0001 eV K−1), were estimated.

68
Waser16 found ΔH°H2O values of −46 kJ mol−1 for undoped SrTiO3 and −29 kJ mol−1 for a Fe-doped SrTiO3 (0.3 mol%) by IR spectroscopy.

69
Kreuer et al22. obtained −23 kJ mol−1 for Sc-doped SrTiO3 (2 mol% Sc) from a hydration isobar at 1 atm PH2O by thermogravimetry.

70
Cherry et al23. calculated −69 kJ mol−1 for pure SrTiO3 by atomistic simulation methods.

71
Some of these are comparable to our value (−60 ± 12 kJ mol−1) and some smaller (in absolute magnitude).

72
In Fig. 4, the results of Sc-doped SrTiO322 are compared.

73
Not only the difference in the hydration enthalpy, but, more significantly, a difference by more than one order of magnitude in the absolute magnitude of hydration constants are indicated.

74
The difference may be due to the different experimental parameters in the respective studies (isobaric thermogravimetry and high vapor pressure of 1 atm in ref. 22) and/or may indicate a significant difference in the thermodynamics in water incorporation between Sc-doped SrTiO3 and Fe-doped SrTiO3.

75
Kreuer et al22. showed that the formation of protonic defects is very sensitive to the choice of acceptor dopant in BaZrO3.

76
Fig. 5 shows the electrical conductivity change in single crystals with different Fe concentrations, 5 × 1018 (in open circles) and 5 × 1019 cm−3 (in solid squares), respectively, upon a stepwise introduction of water vapor of 20 mbar at 475 °C under 1 atm PO2.

77
The conductivity of the specimen with the lower Fe content decreased upon water incorporation by ∼22%, while that with the higher Fe content slightly increased (∼3%).

78
The data in Fig. 5 were obtained using highly reversible YBa2Cu3O6+δ electrodes on large surfaces of the plate specimens; the spectra exhibited only bulk semicircular responses from which the conductivity was estimated.

79
The specimens with Pt electrodes on the small surface sides showed perceptible electrode contributions in the low frequency range separated from the high frequency response from which the bulk conductivity was obtained.

80
Besides quite different relaxation behaviors between YBa2Cu3O6+δ and Pt electrodes, which will be detailed elsewhere, the absolute magnitudes of the conductivity of the same Fe content were not in exact agreement with each other (Fig. 6).

81
Nevertheless, the specimens with Pt electrodes also exhibited a conductivity increase for Fe content of 5 × 1018 cm−3 and decrease for 5 × 1019 cm−3.

82
In Fig. 6 the calculated (total) conductivity (σT) is compared with the experimental values.

83
The total conductivity is the sum of partial conductivities of holes (σh), oxygen ions (σO), and protons(σH).

84
The defect concentrations are estimated from mass action constants including KH2O of the present study.

85
Kinetic parameters are taken from the literature: hole mobility 7.5 × 103(T/K)−1.5 cm2 V−1 s−1 and vacancy mobility 3.0 × 103 (T/K)−1 cm2 V−1 s−1 e−0.86 eV/kT in Fe-doped SrTiO3 from .ref. 24

86
They are improved parameters from those reported in refs. 7 and 11.

87
Proton diffusivity in ref. 25, 7.9 × 10−4 cm2 s−1 e−0.41 eV/kT, is used which was obtained in Sc-doped SrTiO3.

88
The proton mobility may be sensitive to the acceptor dopants as shown in BaZrO3,22 however, similarly with the proton formation energetics.

89
The calculations are not exactly consistent with the measurements but the difference between the two specimens is semi-quantitatively explained.

90
Despite the much smaller concentration of holes compared to the ionic defects, the conductivity is mainly electronic at 475 °C and 1 atm dry oxygen atmosphere due to the high hole mobility.

91
With increasing water vapor pressures the proton conductivity becomes an appreciable part of the total conductivity.

92
According to the simulation, both specimens possess a significant proton contribution of ∼6% in the total conductivity at 20 mbar PH2O.

93
The bottom graphs in Fig. 6 represent defect concentrations in linear scale in which variations in [Fe′Ti] can be clearly seen (the dashed lines indicate PH2O region in the upper graphs).

94
It should be noted that the degree of trapping and thus also the Fe4+ to Fe3+ concentration ratio is higher for the higher Fe concentration in SrTiO3.

95
The detrapping induced by water incorporation and corresponding increase in [OH˙O] and [h˙] is thus more pronounced in the higher Fe contents and a more distinct transition between two PH2O regimes results in SrTiO3 with lower Fe content as similarly seen in the case of fixed-valence dopants (see also Fig. 1).

96
The slight increase in the conductivity of the higher dopant specimen in Fig. 6 can thus be ascribed to the stronger increase in proton conductivity overcompensating the moderated decrease in hole conductivity due to the increase in Fe3+/Fe4+ ratio.

Conclusion

97
The thermogravimetric measurements on the water incorporation in Fe-doped SrTiO3 were analyzed considering the redistribution of the valence states of Fe dopants.

98
The hydration enthalpy and entropy were estimated to be −60 ± 12 kJ mol−1 and −122 ± 9 J mol−1 K−1, respectively.

99
The defect model that also takes account of redox changes with water incorporation satisfactorily described the conductivity behavior of SrTiO3 single crystals with different Fe content in water incorporation.