Article pubs.acs.org/IC
Interlayer Communication in Aurivillius Vanadate to Enable Defect Structures and Charge Ordering Yaoqing Zhang,*,†,‡,§ Takafumi Yamamoto,§ Mark A. Green,∥ Hiroshi Kageyama,§ and Yutaka Ueda*,‡,⊥ ‡
Institute for Solid State Physics, University of Tokyo, Kashiwa, Chiba 277-8581, Japan Department of Energy & Hydrocarbon Chemistry, Kyoto University, Kyoto 615-8510, Japan ∥ National Institute of Standards and Technology, Gaithersburg, Maryland 20899-6100, United States ⊥ Toyota Physical and Chemical Research Institute, Nagakute, Aichi 480-1192, Japan §
S Supporting Information *
ABSTRACT: The fluorite-like [Bi2O2]2+ layer is a fundamental building unit in a great variety of layered compounds. Here in this contribution, we presented a comprehensive study on an unusual Aurivillius phase Bi3.6V2O10 with respect to its defect chemistry and polymorphism control as well as implications for fast oxide ion transport at lower temperatures. The bismuth oxide layer in Bi4V2O11 is found to tolerate a large number of Bi vacancies without breaking the high temperature prototype I4/mmm structure (γ-phase). On cooling, an orthorhombic distortion occurs to the γ-phase, giving rise to a different type of phase (B-phase) in the intermediate temperature region. Cooling to room temperature causes a further transition to an oxygen-vacancy ordered A-phase, which is accompanied by the charge ordering of V4+ and V5+ cations, providing magnetic (d1) and nonmagnetic (d0) chains along the a axis. This is a novel charge ordering transition in terms of the concomitant change of oxygen coordination. Interestingly, upon quenching, both the γ- and B-phase can be kinetically trapped, enabling the structural probing of the two phases at ambient temperature. Driven by the thermodynamic forces, the oxide anion in the γ-phase undergoes an interlayer diffusion process to reshuffle the compositions of both Bi−O and V−O layers.
■
different connectivity with the perovskite layer.12−15 On the other hand, the oxygen stoichiometry in the perovskite slabs is strongly flexible. In correspondence to the simplest member of the Aurivillius family, Bi2BO6 (n = 1), Bi2VO5.5 hosts a large population of disordered oxygen vacancies in the perovskite layer (V−O layer), resulting in exceptionally high oxide ion conductivity in the temperature regime where the discussed structure can be retained.16 Cooling this high-temperature form of Bi2VO5.5 (γ-phase) gives rise to successive endothermic transitions that are accompanied by discontinuous changes in the unit cell parameters. Namely, two oxygen-ordered polymorphs, β and α, are formed at 553 and 386 °C, respectively.16 Although oxygen order/disorder transitions tend to be of second order, γ → β and β → α transitions in Bi2VO5.5 clearly demonstrate a first-order type transition. The V−O layer in the β-phase is composed of zigzag arrays of corner-shared trigonal bipyramids and tetrahedra.17 In the α-phase, a dimeric unit is formed with two edged-shared trigonal bipyramids that are connected to two VO4 tetrahedra.18 In general, a first-order phase transition involves atomic diffusion, and hence, the hightemperature disordered phase can be quenched by rapid cooling. However, the γ-phase discussed here is not quenchable, possibly owing to the fast oxide ion conduction of this material.
INTRODUCTION Layered transition metal oxides provide a rich structural source for a broad spectrum of intriguing properties, from superconductivity1−3 to ferroelectricity,4 energy storage,5,6 and optical activity,7 and they continue to be an active subject of considerable interest in current physics and chemistry. Of these interesting compounds, Aurivillius phases8 represent one important subgroup, formulated as Bi2An−1BnO3n+3, where A is an electropositive atom while the B site typically accepts the early transition metal. The prototype Aurivillius structure consists of alternate stacking of fluorite-like (Bi2O2)2+ layers and perovskite (An−1BnO3n+1)2− layers, where n is the number of perovskite blocks. The Bi−O layer is both structurally and electronically close to tetragonal PbO.9 Bi is pyramidally coordinated to four oxygen atoms which are arranged in a planar square array, and its chemically inert but stereochemically active lone pairs of electrons point toward the position equivalent to the large A cations, such as alkali or alkali earth in the perovskite block.10,11 One important empirical rule received from the prior studies with respect to the crystal chemistry of Aurivillius phases is that the Bi−O layer is rather rigid and nearly inviolate in terms of its nonstochiometry.12 For example, the bismuth atoms in the perovskite layers allow a diversity of substitution possibilities. As opposed to the perovskite block, the Bi−O layer accepts only a partial substitution by some specific A-type cations such as Sr2+, while a substitution with, e.g., lanthanide ions leads to a © XXXX American Chemical Society
Received: August 25, 2015
A
DOI: 10.1021/acs.inorgchem.5b01964 Inorg. Chem. XXXX, XXX, XXX−XXX
Article
Inorganic Chemistry
to ambient conditions could open a new avenue of accessing the oxygen disordered state and designing new materials for fast oxide ion diffusion at unusually low temperature range.
An earlier effort to examine the tolerated range of oxygen deficiency in Bi2VO5.5 phases (labeled as Bi4V2O11−δ hereafter) led to an interesting finding19 that challenges the contemporary understanding of the structural chemistry in the n = 1 system and perhaps the whole Aurivillius-type bismuth vanadium oxides. The monophasic α-phase was found not realizable for the composition of Bi4V2O11 (δ = 0), whose solid state synthesis is accompanied by an impurity phase of BiVO4. In contrast, slightly decreasing the oxygen stoichiometry to δ = 0.1 enables the formation of a single α-phase. If the oxygen deficiency is further increased to the range of 0.1 < δ < 0.4, a mixture of both α- and A-phases is identified. Interestingly, at exactly δ = 0.4, the phase diagram shows the pure A-phase and, therefore, Bi4V2O10.6 defines a phase boundary between A and (A + α) phases. According to our studies, the A-phase is structurally quite close to a previously studied Bi4V2O10.66 compound,19−22 an assumed product by reducing exactly 1/3 oxygen from Bi4V2O11, but was difficult to form a single phase. Upon deeper reduction than δ = 0.4, there was found a bismuth metal exsolution,23 leaving bismuth vacancies in the Bi−O layer which serve to aid the accommodation of more oxygen loss. For instance, the stoichiometric Aurivillius phase Bi4V2O10 proved difficult to realize, but its single phase is achieved in Bi3.6V2O10. This implies the removal of a Bi2O3 unit from the Bi−O layer of Bi4V2O10.6. Namely, the oxygen extraction in Bi4V2O11−δ changes from the V−O layer (δ < 0.4) to the Bi−O layer (0.4 ≤ δ ≤ 1). We defined this behavior as interlayer switching of reduction. The bismuth exsolution in Bi4V2O11−δ (0.4 < δ ≤ 1) enables the construction of a phase diagram against temperature which is different from that for Bi4V2O11.19 On heating, the A-phase for 0.4 < δ ≤ 1 successively transforms to the B-phase and further to the γ-phase. In the higher temperature region, the γphase applies to the whole composition range (0 ≤ δ ≤ 1). Nevertheless, the precise structures of these three phases have not been resolved yet. Such a structural characterization is of critical significance in clarifying the oxygen-vacancy distribution and the charge states of both bismuth and vanadium cations. We addressed that the apparent average vanadium valence of +4.6 (V4+/V5+ = 2/3) in Bi4V2O10.6 is kept throughout the examined bismuth deficient compositions of Bi4−xV2O11−δ (0.4 ≤ δ ≤ 1), which gives a simple relation x = 2δ/3 − 4/15.19 However, it remains unclear in origin why there is a small gap in oxygen composition and vanadium nominal charge state between Bi4V2O10.66 and Bi4V2O10.60/Bi3.6V2O10. A prior structural investigation on A-Bi3.6V2O10 revealed a similar V− O layer to that of Bi4V2O10.66, but the poor resolving power of the laboratory X-ray diffraction (XRD) did not provide accurate information, in particular, regarding the oxygen stoichiometry and coordination environments in both layers and hence the actual charge state of cations. In addition, the phase identification of Bi3.6V2O10 at elevated temperatures (B and γ) was done by in situ laboratory XRD with powder samples sealed in evacuated capillaries to avoid sample oxidation, where only reflections with relatively strong intensity were observable. Further structural studies are hence necessary in order to illustrate the structural features of Bi3.6V2O10 and compare with the Bi4V2O11−δ phase (0.4 ≤ δ ≤ 1). Here in this contribution, we focus on Bi3.6V2O10 with large amounts of bismuth and oxygen vacancies so as to demonstrate the interplay between the bismuth and vanadium oxide layers and to reveal the structural consequence in all polymorphic phases. In particular, the success in stabilizing the γ-type phase
■
EXPERIMENTAL SECTION
Materials Synthesis. Polycrystalline samples of Bi3.6V2O10 were synthesized via the conventional solid state reaction method using stoichiometric Bi2O3 (Rare Metallic, 5N), V2O3 (Rare Metallic, 2N) and V2O5 (Rare Metallic, 4N) as starting materials. After being thoroughly ground with mortar and pestle, the precursor mixtures were pelletized and sealed into a quartz tube under vacuum of some 10−3 mbar. Pellets were then heated at 600 and 800 °C for 12 and 72 h, respectively, before cooled down to room temperature (RT) within 5 h. The A-type Bi3.6V2O10 samples for both synchrotron and neutron diffraction underwent an annealing treatment that was to dwell the sample powders at 300 °C for 30 days. The B-type Bi3.6V2O10 sample of around 0.7 g can be stabilized by cooling sample in an evacuated silica tube from 800 °C to RT within 2.5 h. For the preparation of the γ-type Bi3.6V2O10 at RT, a pellet-loaded silica tube was taken out of the furnace at 800−850 °C, and the sample was quickly cooled down to RT inside the tube before further characterization. X-ray and Neutron Powder Diffraction. Laboratory RT powder X-ray diffraction data of all as-prepared polycrystalline samples were collected on a Rigaku Smartlab diffractometer with Cu Kα radiation source and a position-sensitive detector. The step size was set to be 0.01° in a 2θ range from 10° to 70°. High resolution synchrotron Xray diffraction experiments were performed on Bi3.6V2O10 including the A-, B- and γ-phases at RT on a Debye−Scherrer camera installed at beamline BL02B2, SPring-8, Japan. The incident beam from a bending magnet was monochromatized to 0.35479(1) Å. The powder samples were loaded into a glass capillary (0.1 mm inner diameter) and rotated during measurements to reduce preferential orientation. The diffraction data were recorded in a 2θ range from 0° to 60° with a step interval of 0.01°. Powder neutron diffraction measurements on the γ-type Bi3.6V2O10 were carried out at RT on an approximately 2 g sample at BT-1 (λ = 1.5403 Å), National Institute of Standards and Technology (NIST), US. For the A-type Bi3.6V2O10 neutron diffraction, data were recorded at RT at ANSTO Echidna (λ = 1.6220 Å), Australia. The Rietveld refinements were conducted using RIETAN-FP24 program and a GSAS-EXPGUI package.25,26 AC Impedance Measurements. For impedance data collection, both surfaces of a pellet sample were deposited with Pt pastes and fired at 300 °C in vacuum conditions (∼10−2 mbar). An AC impedance measurement was carried out over the frequency range from 100 mHz to 1 MHz using a Solartron 1255 Frequency Response Analyzer coupled with a Solartron 1287 Electrochemical Interface controlled by ZPlot electrochemical impedance software and a perturbation voltage of 50 mV was applied. During the whole measurement process, the sample was kept in vacuum to prevent any possible oxidation. The sample after the conductivity measurement was further examined by powder X-ray diffraction. Magnetic Property Characterization. Magnetic susceptibilities of the γ-, B- and A-type Bi3.6V2O10 were measured by a superconducting quantum interface device (SQUID, Quantum Design MPMS-5S) magnetometer in the temperature range of 1.8−300 K under an applied external magnetic field of 5 T.
■
RESULTS AND DISCUSSION Phase Relations. The Bi3.6V2O10 phase is indeed immune from the Bi exsolution; see the scanning electron microscopy images in Figure S1. As temperature rises, it exhibits three allotropic forms: A, B, and γ.19 The phase transitions occur at 370 and 570 °C for A → B and B → γ, respectively. A close examination revealed that both γ- and B-phases can be stabilized to RT, indicating that they can be kinetically trapped, in a marked contrast to Bi4V2O10.6, where the γ- and B-phases are not quenchable. In fact, quenching the pellet sample inside a sealed silica glass ampule from some 800 °C stabilized the γB
DOI: 10.1021/acs.inorgchem.5b01964 Inorg. Chem. XXXX, XXX, XXX−XXX
Article
Inorganic Chemistry
structures (Figure 1) with a larger unit cell of √2ap × 3√2bp × c (a = 5.4700, b = 17.2394 and c = 14.9288 Å). Here it is interesting to note that ceramic pellets (pressed at ∼1.8 × 107 kg/m2) of either γ- or B-phase of Bi3.6V2O10 going through the respective transitions readily transform to powders of the A-phase on relatively slow cooling. Such an observation reminds us of a recently reported phase transition in Nd0.35La1.45Pr0.2CuO4 from pellet T-phase to fine powder T″phase due to the change of unit cell volume by ∼5%.27 The phase transition from the γ- and B-phases to the A-phase in the present case also involves a large cell contraction along the c axis (3.511−3.486%), which can explain the formation of powder. Varying the cooling conditions, the high temperature γ-phase follows different paths and produces phase mixtures. It is also worth noting that Abraham et al.28 attempted to synthesize Bi1.33V2O6 using Bi2O3, V2O5, and V2O3 in the proportion of 4:3:3. The main phase received was hollanditestructured Bi1.7V8O16 while an unknown compound was also identified which was indexed in the body-centered tetragonal system with a = 3.8767 and c = 15.337 Å. In comparison with our results, it is very likely this accompanying phase is the γtype Bi3.6V2O10. High Temperature γ-Phase. In order to examine the structural response of the Aurivillius phase to the introduction of vacancies into O (and Bi) sites, both synchrotron and neutron diffraction data were collected over RT-stabilized γBi3.6V2O10 powder samples. A joint synchrotron X-ray and neutron Rietveld analysis is known to provide accurate structural parameters such as atomic positions, occupancy and thermal motions. As a starting model for the Rietveld analysis, we assumed a structure with a V−O layer formed by cornershared VO6 octahedra obtained from the in situ diffraction study on Bi4V2O11 (I4/mmm) at high temperature.17 For the refinement of the synchrotron data, the atomic coordinates of both Bi and V atoms including occupancy factors were allowed to vary, while for the neutron data the atomic coordinates of the O atoms were refined. During the neutron analysis, a constraint on the occupancy factors (g) of the Bi site was posed at 0.9, whereas the total amount oxygen was fixed to 10 in a chemical formula. When the g values of the oxygen sites were freed, they became less than 1, i.e. g(O1) = 0.93(1), g(O2) = 0.72(1) and g(O3) = 0.85(2), indicating that vacancies are randomly distributed at the oxygen sites in the V−O and Bi−O layers. Since the isotropic displacement parameter (Biso) of the apical oxygen site (O3) of VO6 octahedra was unusually high (∼12 Å2), we used anisotropic displacement parameters βij for the oxygen sites. The alternate refinements of the synchrotron and neutron were repeated several times, yielding the final fitting results as shown in Figure 2 and Table 1. The small agreement factors, Rwp = 5.60%, Rp = 4.00% and χ2 = 5.06 for the synchrotron and Rwp = 4.36%, Rp = 3.47% and χ2 = 1.02 for the neutron, suggest that the refined structure is reliable. Figure 3a illustrates the crystal structure of the RT-stabilized γ-type Bi3.6V2O10 obtained from the joint refinement. It is confirmed to be layered-structured with the space group of I4/ mmm. The Bi−O layer accommodates 10% vacancies at the single Bi (4e) site. All four Bi−O bond lengths are equal to each other, ∼ 2.31 Å. Neither neutron data nor electron diffraction patterns (see Figure S2) provide any evidence to suggest the formation of superstructures. We found that the oxygen in the Bi−O layer shows slightly anisotropic thermal motion. This could be of thermal origin as the reported Uiso (equivalent isotropic atomic displacement parameter) value of γ-Bi4V2O11
phase (RT-stabilized γ-phase hereafter) at ambient temperature, while cooling at a moderate rate (see Experimental Section) gave the RT-stabilized B-phase, in each case without yielding the A-phase. This means that relatively fast cooling does not allow the necessary diffusion of oxide ions and therefore completely suppresses the phase transitions. As a result, it becomes feasible to probe the crystal structures of the two high temperature phases at RT, which was impossible in Bi4V2O10.6. The population of oxygen vacancies in the V−O layer of the RT-stabilized γ (B)-phase of Bi3.6V2O10 is expected to be larger than that of Bi4V2O10.6. As will be discussed later, the interlayer diffusion of oxide ions from the bismuth oxide layer to the vanadium oxide layer is a key to understand the stabilization of the high temperature phase to RT. According to the RT X-ray powder diffraction patterns collected over these quenched samples (Figure 1), the RT-
Figure 1. Room temperature X-ray diffraction patterns of Bi3.6V2O10 including tetragonal γ-phase, orthorhombic B-phase, and oxygenvacancy ordered A-phase (from top to bottom). Despite the fact that both γ- and B-phases represent high temperature phases, they can be retained at ambient temperature by a fast cooling rate in a manner that can suppress the phase transition. Compared with the lattice parameters (ap × bp × c) of the simple perovskite-related structure, the relationship between the unit cells of each phase is illustrated with arrows to indicate the peak splitting.
stabilized γ-phase of Bi3.6V2O10 crystallizes in a face-centered tetragonal (FCT) structure (a = 3.8934 Å, c = 15.4721 Å) with the space group of I4/mmm which is identical to that of the high temperature γ-phase of Bi4V2O11. The RT-stabilized Bphase is of the body-centered orthorhombic (BCO) symmetry (a = 5.4849 Å, b = 5.5227 Å, c = 15.4680 Å) and no additional reflections are observed, indicating the absence of oxygenvacancy ordering. In light of the extinction rule for reflections, the space group for the B-phase is consistent with F222 or Fmm2. The transition from the γ-phase involves a peak splitting of, e.g., (110) to (200) and (020), which, as indicated by the arrows in Figure 1, is of orthorhombic distortion, leading to the in-plane superlattice of √2ap × √2bp ( × c), where ap and bp represent the simple perovskite unit cell constant. Apparently, this transition is different from the one from the γ-phase to the β-phase in Bi4V2O11 because the β-phase has a unit cell of 2√2ap × √2bp × c, resulting from oxygen-vacancy ordering.17,18 The A-phase of Bi3.6V2O10 is assigned as oxygenvacancy ordered as demonstrated by the presence of superC
DOI: 10.1021/acs.inorgchem.5b01964 Inorg. Chem. XXXX, XXX, XXX−XXX
Article
Inorganic Chemistry
Figure 3. Representation of the refined crystal structure (a) and atom anisotropic thermal motion (b) for RT-stabilized γ-Bi3.6V2O10. Bi atoms are shown as violet spheres, vanadium atoms are shown as green spheres, and oxygen atoms are orange colored ellipsoids with strong anisotropic thermal displacement along the a-b plane. Figure 2. Rietveld refinement of (above) synchrotron and (bottom) neutron powder diffraction data of RT-stabilized γ-Bi3.6V2O10 collected at room temperature. Observed data, the calculated pattern, and the difference between observed and calculated data are shown as plus sign (red), solid line (green) and continuous line (blue), respectively. The positions of Bragg reflections are indicated by vertical tick marks (green).
Bi−O layer. This tendency becomes particularly noticeable at high temperature. Another noticeable feature in the refined structural parameters for RT-stabilized γ-Bi3.6V2O10 is that the oxygen atom at the apical site (O3) of the V−O layer shows strong anisotropic thermal factors in the a-b plane as shown in Figure 3b. This points to a possible high mobility of the apical oxygen and a conductivity measurement will be shortly discussed. Here it is worth noting that one previous in situ neutron diffraction study on the γ-type Bi4V2O11 at 670 °C did not recognize anisotropic atomic displacement parameters for either O2 or O3. Such difference is mostly because the previous structural model attributed the large anisotropic spread of charge density for O2 and O3 to the lowered symmetry by placing O2 and O3 not at the 4c and 4e sites, respectively, but at a 16n position. Intermediate Temperature B-Phase. Assuming that the composition in the vanadate layer still holds true for the B-type structure, the Rietveld analysis over synchrotron data of RTstabilized B-Bi3.6V2O10 was performed as shown in Figure 4 and Table 2. Of two possible orthorhombic space groups (F222 and Fmm2), the F222 space group resulted in a better fit with reasonable R-factors (Rwp = 4.42%, Rp = 2.96%, and χ2 = 8.358) as well as structural parameters. It can be seen that the γ → B phase transition involves a displacive distortion in the V−O layer, and a coherent rotation of the VO6 octahedra is necessary to stabilize the structure when temperature is lowered as in some other Aurivillius phases. The inset of Figure 4 is an illustration of the derived crystal structure. Bi is 2.866 Å away
derived from an in situ high temperature study is much higher.17 The refined compositions are [Bi1.80O1.86(2)] and [VO3.14(6)] for the respective Bi−O and V−O layers. The amount of oxygen vacancies in the Bi−O layer is smaller than the expected composition of [Bi1.80O1.70] assuming that both Bi and oxygen are removed as a Bi2O3 unit from the Bi−O layer so as to retain the vanadium valence. Complementarily, the V−O layer bears higher oxygen deficiency [VO3.14(6)] than in the Bi4V2O10.6 analogue where the corresponding layer is [VO3.30], giving the mean oxidation state of V4.6+. These results mean the original [Bi2O2]2+ layers structurally tolerate a large number of vacancies with flexibility in terms of its composition as well as charge in the form of [Bi1.80O1.86(2)]1.68+. Such a defect chemistry is unknown in any other compounds containing [Bi2O2]2+ layers. We define these features as sorts of interlayer communication in Aurivillius vanadate. The observed lower (higher) oxygen deficiency of the Bi−O layer (V−O layer) might be understood in light of statistical thermodynamics. Namely, more oxygen vacancies in the V−O layer gain entropy and lower the free energy because the number of oxygen sites is larger in the V−O layer than in the
Table 1. Parameters from the Rietveld Refinement of Neutron Powder Diffraction Data for γ-Type Bi3.6V2O10a Atom
site
g
x
y
z
B
Bi V O1 O2 O3
4e 8i 4d 4c 4e
0.9 0.25 0.93(1) 0.72(1) 0.85(2)
0 0.088(2) 0 0 0
0 0 0.5 0.5 0
0.33157(4) 0 0.25 0.5 0. 1011(5)
1.74(1) 1.6(1)
β11
β22
β33
0.023(1) 0.057(5) 0.29(1)
0.023(1) 0.099(7) 0.29(1)
0.0022(2) 0.020(1) 0.0033(5)
a
Space group: I4/mmm (139) tetragonal. Lattice parameter: a = 3.88034(4) Å, c = 15.4201(2) Å for synchrotron X-ray diffraction (SXRD); a = 3.8957(1) Å, c = 15.4738(6) Å for neutron diffraction (ND). Composition: Bi3.6V2O10 (Z = 1). D
DOI: 10.1021/acs.inorgchem.5b01964 Inorg. Chem. XXXX, XXX, XXX−XXX
Article
Inorganic Chemistry
Figure 4. Rietveld refinement of synchrotron powder diffraction data of RT-stabilized B-Bi3.6V2O10 collected at room temperature. Observed data, calculated pattern, and the difference between observed and calculated data are shown as plus sign (red), solid line (green), and continuous line (blue), respectively. The positions of Bragg reflections are indicated by vertical tick marks (green). The inset is a polyhedral illustration of the refined structure viewed along the ⟨110⟩ direction where octahedral VO6 and all oxygen sites (O1−O4) are indicated.
Table 2. Parameters from the Rietveld Refinement of Neutron Powder Diffraction Data for B-Type Bi3.6V2O10a Atom
site
g
x
y
z
B
Bi V O1 O2 O3 O4
8a 4a 4c 4d 8h 8g
0.9 1.0 0.955(1) 0.945(1) 0.951(2) 0.600(3)
0 0 1/4 1/4 1/4 0
0 0 1/4 1/4 1/4 0
0.33164(5) 0 1/4 3/4 0.0379(7) 0.1037(6)
0.0216(3) 0.0582(19) 0.0120(1) 0.0571(6) 0.2217(7) 0.0158(3)
a
Space group: F222 (22) orthorhombic. Lattice parameters: a = 5.4826(6) Å, b = 5.5253(2) Å, c = 15.4697(9) Å, V = 468.6332(4) Å3. Composition: Bi3.6V2O10 (Z = 2).
from the nearest O in the V−O layer, and therefore, there is no bonding between the adjacent layers (the Bi−O bond length is typically in the range from 2.2 to 2.7 Å).12 Moreover, the γ → B transition does not bring noticeable structural change to the Bi−O layer. Compared to the β-Bi4V2O11 phase, the B-type Bi3.6V2O10 here differs significantly in structure, particularly for the V−O layer. In the β case, V occupies two crystallographic sites, with more complex V−O coordination environments, including tetrahedra and trigonal pyramids. Another difference is the formation of the interlayer bond of around 2.497 Å between O in the V−O layer and Bi. Room Temperature A-Phase. The prior Rietveld refinement of laboratory X-ray diffraction data for A-type Bi3.6V2O10 suggested a similar V−O layer structure to the idealized Bi4V2O10.66 phase. Namely, the octahedrally coordinated V4+ spin chain running along the a axis links with every two parallel V5+O4 tetrahedral chains forming a basic building unit for the V−O layer (refer to Figure 5c).19,20 This structure is partly supported by the magnetization behavior as to be discussed later. As mentioned earlier, however, the idealized phase Bi4V2O10.66 whose V4+ and V5+ cations are strictly in the molar ratio of 1/2 is not realized in practice. Instead, Bi4V2O10.60 can be isolated, whose composition in the V−O layer (VO3.30) is slightly oxygen deficient compared to the stoichiometric composition (VO3.33) and supposed to be identical to that of
Figure 5. Rietveld refinement of (a) synchrotron and (b) neutron powder diffraction data of A-Bi3.6V2O10 collected at room temperature. A polyhedral illustration of the refined structure is shown in (c) while (d) details the spatial arrangement of vanadium with respect to the coordinating oxygen, of which O8 sites are partially vacant to give rise to VO5 pyramids.
A-Bi3.6V2O10. Apparently, the nominal V4+/V5+ ratio in Bi4V2O10.60 and Bi3.6V2O10 is slightly larger than 1/2. The compound of the current discussion, Bi3.6V2O10, contains Bi vacancies and much more oxygen vacancies, which effectively E
DOI: 10.1021/acs.inorgchem.5b01964 Inorg. Chem. XXXX, XXX, XXX−XXX
Article
Inorganic Chemistry Table 3. Parameters from the Rietveld Refinement of Neutron Powder Diffraction Data for A-Type Bi3.6V2O10a Atom
site
g
x
y
z
B (Å2)
Bi1 Bi2 Bi3 Bi4 V1 V2 O1 O2 O3 O4 O5 O6 O7 O8 O9
4h 4h 8i 8i 4h 8i 8i 8i 8i 8i 8i 8i 8i 4h 4h
0.883(6) 0.885(6) 0.905(4) 0.911(7) 1 1 0.88(1) 0.76(1) 0.91(1) 1 1 1 1 0.90(4) 1
0.1991(6) 0.2501(5) 0.2344(3) 0.2835(4) 0.251(2) 0.266(2) 0.993(1) 0.086(1) −0.003(1) 0.002(2) 0.004(2) 0.296(1) 0.243(1) 0.301(2) 0.220(3)
0.25 0.25 0.4120(1) 0.4277(1) 0.25 0.3875(3) 0.9838(3) 0.3673(4) 0.3257(4) 0.3316(5) 0.3335(5) 0.4032(4) 0.4718(3) 0.25 0.25
0.31668(2) 0.6706(2) 0.8341(2) 0.1678(2) 0.0279(8) 0.4867(5) 0.2475(5) 0.2706(5) 0.7494(5) 0.0528(3) 0.4659(3) 0.5999(4) 0.4281(3) 0.1703(7) 0.9149(6)
0.68(6) 0.57(7) 0.66(6) 0.80(4) 2.6(3) 1.8(1) 0.57(5) 0.57(5) 0.57(5) 1.75(5) 1.75(5) 1.75(5) 1.75(5) 1.75(5) 1.75(5)
a Space group: Pnma (62) orthorhombic. Lattice parameter: a = 5.44901(4) Å, b = 17.1726(2) Å, c = 14.8702(1)Å for SXRD; a = 5.4668(2) Å, b = 17.2323(7) Å, c = 14.9198(6)Å for ND. Composition: Bi3.6V2O10 (Z = 6).
compositions for the Bi−O and V−O layers are [Bi1.80(1)O1.70(2)] and [V1.00O3.30(1)], respectively. This means that the composition of the V−O layer is identical in composition to that of Bi4V2O10.6 without Bi- and O-vacancies in the Bi−O layer. Also, the composition of the Bi−O layer corresponds to the value when one assumes the removal of a Bi2O3 unit from the Bi−O layer. The refined structure is shown in a polyhedral representation in Figures 5(c) and (d), where 10% of V1O6 octahadra have pyramidal coordination (namely, 10% of O8 sites are deficient) and Bi- and O-vacancies are randomly distributed in the Bi−O layers. Most interestingly, the comparison of the structure between the γ- and A-type Bi3.6V2O10 phases sheds light on the nature of the phase transition as well as the thermal stability. According to the Rietveld refinement, the A-phase has a composition of [Bi1.80(1)O1.70(2)] and [VO3.30(1)], which is different from the RT-stabilized high temperature γ-phase that possesses the composition of [Bi1.80O1.86(2)] and [VO3.14(6)]. It follows that when the γ-phase is cooled, some oxide ions have to diffuse from the Bi−O layer to the V−O counterpart. That is to say the high degree of oxygen deficiency in the V−O layer in the high temperature range is no longer allowed at lower temperatures where the A-phase is stable. The interlayer diffusion is accompanied by the change of the vanadium coordination geometry from the deficient octahedra to a mixed type including tetrahedra, pyramids and octahedra. This interlayer diffusion from Bi−O to V−O layers is not fast enough, which allows the high temperature γ(B)-phase to quench. In contrast, other compositions such as Bi4V2O10.6 do not involve such an interlayer oxygen migration between the γ-type and A-type phases as a result of few oxygen vacancies in the Bi−O layer and the high temperature γ-phase is not quenchable. In the V−O layer of A-phase, there is a structural moiety consisting of one V4+ (V1 in Table 3) chain and two V5+ (V2 in Table 3) chains, all of which run parallel to the a axis. Therefore, the transition from the γ- to A-phase is driven by oxygen-vacancy ordering and accompanied by a charge ordering with the different oxygen coordination. At a glance, the complete separation between the structural moieties is seen as if a chemical scissor is applied. However, these oxygen atoms in VO4 tetrahedra which are not shared with octahedral VO6 are bonded to bismuth (Figure 5c). These Bi3−O7 bonds of
means some local changes to this structural model are necessary. Due to the poor resolving power of the laboratory X-ray source, the structural details were not fully revealed and discussed previously. To bridge the gap between the hypothetical structure and the real one, we performed both synchrotron and neutron Rietveld refinements of A-Bi3.6V2O10 powder samples that were annealed at 300 °C for 30 days in order to best reflect the phase equilibrium. We conducted the synchrotron and neutron refinements starting with a reported structural model for Bi4V2O10.66.20 During the analysis, we fixed the total oxygen amount of 10 (as in the γ-phase) and the g of the Bi site at 0.9. Note that the V2 with a 4-fold coordination by O4, O5, O6, and O7 (see Figure 5d) is assumed nonoxygendeficient (i.e., g = 1) because, otherwise, the V2 atom would have three-coordination, which is hardly acceptable for V5+. It is clearly seen that the apical oxygen sites O8 and O9 coordinated with V1 became deficient, leading to oxygen deficient V1O6 octahedra, with the occupancy factors of 0.84(2) and 0.64(4), respectively. Here, the V1−O9 bond length (1.69 Å) became unexpectedly shorter than the V1−O8 bond length (2.12 Å). The bond valence sum (BVS) calculation for V1O6 octahedra gave +3.95. In contrast, the BVS for V1O5 pyramid with the apical oxygen O8 (i.e., O9 is vacant) yielded +2.4, which is unrealistic given the expected valence of V4+. On the other hand, the BVS value is calculated to be +3.5 when the apical O9 site is filled (i.e., O8 is vacant). In the original A-type structure of Bi4V2O10.66, the V1−O9 bond length (1.74 Å) is shorter than the V1−O8 bond length (1.84 Å), suggesting a pseudopyramidal coordination of V1O6 octahedron.22 Hence, when the apical oxygen site becomes deficient to form a pyramidal coordination, the O8 site with longer V−O bond length should be preferentially deficient, which is exactly what has been observed here in A-type Bi 3.6 V 2 O 10 . These considerations led us to fix the g value of the O9 site to unity in the following refinement analysis. After several rounds of alternate refinements of the synchrotron and neutron profiles, we reached the final fitting results as shown in Figure 5 and Table 3. The small agreement factors (χ2 = 4.19, Rp = 3.11% and Rwp = 4.48% for synchrotron and χ2 = 1.17, Rwp = 4.88%, and Rp = 3.79% for neutron) suggest that the refined structure is reasonable. The refined F
DOI: 10.1021/acs.inorgchem.5b01964 Inorg. Chem. XXXX, XXX, XXX−XXX
Article
Inorganic Chemistry ∼2.44 Å help stabilize the A-type structure, although the formation of these bonds is not unique because similar bonds exist in the α-phase. Here it is worth noting that a similar structure with segregation of octhahedra and tetrahedral was realized in Bi6TiP2O16 and Bi6(M, Bi)1P2(O,F)16‑x Aurivillius derivatives. Particularly, in these phosphates the oxygen deficiency was also identified around the MO6‑x polyhedra.29,30 Interestingly, the distance along the c axis between Bi and the nearest O in A-type Bi3.6V2O10 is compressed to 0.013 Å from 0.067 Å for the γ-phase and 0.054 Å for the B-type equivalent. This is because, in addition to Bi3−O7, Bi3+ interacts with apical oxygens (O8) to form extra Bi1−O8 bonds, further stabilizing the structure. Such an extra bonding only applies to the A-phase whereas it is absent in the γ- and B-phase. As a consequence of this extra bonding, the Bi−O layer in the Aphase has been distorted noticeably from the square-pyramidal coordination. The Bi−O bonds in the layer are no longer equal to each other; Bi maintains one short bond and three longer bonds with oxygen atoms in the Bi−O layer, for instance, one Bi3−O1 bond at 2.14 Å and three Bi3−O3 bonds at 2.321, 2.35, and 2.39 Å. For the Bi−O layer, if the notion of antistructure is applied, it can be described as a combination of OBi4 and OBi3.31 As mentioned above, assuming tetravalent V1 with octahedral (or pyramidal) coordination and a pentavalent V2 with tetrahedral coordination the structure gives a V4+/V5+ ratio of 1/2, namely an average valence of +4.66 for vanadium, which is different from the apparent value of V4.60+ in Bi3.6V2O10. To account for this discrepancy of vanadium valence, one may consider simply replacing 10% of tetrahedral V5+ sites with V4+. However, the oxidation state of tetrahedral vanadium is known only in favor of +5 and any attempts to change it would risk dramatically distorting the host structure.32−34 Evidently our refinement recognized no such significant structural changes. One may also consider introducing lower oxidation state than 4+ into the V4+ chains. In other words, in addition to V5+ two other oxidation states including V4+ and V3+ are supposed to coexist in the A-phase of Bi3.6V2O10. However, this possibility is simply inconsistent with the chemistry of the Bi4V2O11−δ system which defines a very narrow valence range for vanadium where beyond Bi4V2O10.6 no more V4+ ions are tolerated in Bi4V2O11−δ, let alone V3+. The valence gap may be balanced by a different charge compensation mechanism in the Bi−O layer, for example, minor Bi1+ species might be present to form a possible charge configuration [Bi1.8O1.7]1.94+·[VO3.30]1.94−. Magnetic and Conducting Properties. The temperature dependence of the magnetic susceptibility of the RT-stabilized γ- and B-phases of Bi3.6V2O10 (Figure 6a) exhibits a Curie− Weiss response. The difference between the two data is nearly negligible, suggesting that magnetic vanadium cations (and thereby oxygen vacancies) in the γ- and B-phases could be distributed in a disordered manner, which is consistent with the result of the structural refinement. The analysis of the paramagnetic susceptibility of the γ-phase in the temperature range of 2−300 K by the Curie−Weiss law with a constant term, χ = χ0 + C/(T − θ), yielded χ0 = 3.43 × 10−5 emu mol−1 Oe−1, a Curie constant of C = 0.0417 emu K mol−1 and a Weiss constant of θ = −2.94 K, implying a weak magnetic interaction. The effective magnetic moment is calculated to be μeff = 0.577 μB (where μB is the Bohr magneton) for each Bi1.8VO5 formula unit, which means that one-third of vanadium cations are in the oxidation state of V4+(3d1) with spin S = 1/2 if a complete quenching of the orbital momentum is assumed. This is in a
Figure 6. (a) Temperature dependence of magnetic susceptibility for γ-Bi3.6V2O10 (blue triangles), B-Bi3.6V2O10 (black circles), and ABi3.6V2O10 (yellow squares) powder samples in both zero field cooling (ZFC) and field cooling (FC) modes with an applied external magnetic field of 5 T. The red solid lines represent the fitting of the susceptibility data for each phase according to the Curie−Weiss law or an antiferromagnetic spin chain model. It should be noted that the result for A-Bi3.6V2O10 was reproduced from ref 19 to compare the magnetization behavior between different phases. (b) Nyquist impedance plot for a RT-stabilized γ-Bi3.6V2O10 pellet measured at 200 °C in vacuum.
good agreement with the chemical states in the ideal composition Bi4(V4+1/3V5+2/3)2O10.66 where 33.3% of vanadium is reduced in valence from +5 to +4 (relative to Bi4V2O11) if all bismuth cations are in +3. The refinement result is also supportive of this value. Similarly, the effective moment for the B-phase was calculated to be 0.581 μB (corresponding to the V4+ concentration of 33.5%), essentially confirming the identical oxidation state of vanadium in both γ- and B-phases. For comparison, the magnetic susceptibility of the A-phase is also plotted in Figure 6a, where a broad maximum appears near the critical temperature of 55 K, indicative of the lowdimensional magnetism that can be well explained in terms of an S = 1/2 one-dimensional antiferromagnetic Heisenberg chain model.19 Such a spin structure is in good agreement with the V4+ chain structure determined by the diffraction data. The disorder of oxygen vacancies in the γ-phase of Bi4V2O11 and its derivatives is known as a key factor for the observed fast oxide ion conduction based on the vacancy governing mechanism in ionic conductors.35 In light of this consideration together with the large thermal displacement parameter G
DOI: 10.1021/acs.inorgchem.5b01964 Inorg. Chem. XXXX, XXX, XXX−XXX
Article
Inorganic Chemistry observed in the γ-Bi3.6V2O10, this phase deserves further investigation with respect to its conducting behavior. Attempts to measure the electronic conductivity at ambient conditions failed, as the resistivity of RT-stabilized γ-Bi3.6V2O10 has exceeded the measuring range allowed by our Physical Property Measurement System (up to 10 MΩ in 4-wire mode). An AC impedance measurement was then carried out on the pellet sample with Pt covered on both surfaces as electrodes. The sample is sensitive to oxygen at elevated temperatures, and there are clear oxidation behaviors above 300 °C. A vacuum atmosphere (about 10−2 mbar) generated from a pump was employed in the current characterization up to 300 °C, and Xray diffraction patterns showed no phase change after the measurement (Supplementary Figure S3). Shown in Figure 6b is the complex impedance plane plot for a RT-stabilized γBi3.6V2O10 pellet measured at 200 °C in vacuum. The high frequency semicircle with an associated capacitance in the range of 10−10 F can be attributed to the grain boundary component, while at frequencies below 50 Hz a nonblocking electrode response is observed. According to the spectra, the overall conductivity reached 10−5 S/cm, a conducting level that is comparable with that for the oxide ion conductor Ce0.9Gd0.1O1.95 or the Cu substituted Bi2V1−xCuxO5.5−3x/2 (0 ≤ x ≤ 0.07) although the bulk conductivity is expected to be higher and the electronic contribution is not determined.36,37 The conducting process in the traditional Aurivillius oxide conductors is confined to the oxygen substoichiometric V−O layers. Here for the current compound the Bi−O layers also host oxygen vacancies and it might be reasonable to speculate both layers contribute to the total conductivity. More importantly, the stabilization of the γ-phase at ambient temperature provides a new pathway toward high oxygen mobility at low temperatures.
simply by a quenching process, facilitating good conductivity of 10−5 S/cm at around 200 °C. The two different layers involve reduction switching, bond formation, oxygen diffusion, and charge compensation, defining a rich interlayer communication behavior.
■
ASSOCIATED CONTENT
S Supporting Information *
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.inorgchem.5b01964. Scanning electron microscopy (SEM) images, and Electron diffraction and XRD patterns in Figures S1− S3 (PDF)
■
AUTHOR INFORMATION
Corresponding Authors
*E-mail:
[email protected] (Y.Z.). *E-mail:
[email protected] (Y.U.). Present Address †
(Y.Z.) Materials Research Center for Element Strategy, Tokyo Institute of Technology, Yokohama 226-8501, Japan. Funding
This work is financially supported by a Grant-in-Aid for Scientific Research from Japan Society for the Promotion of Science (JSPS). The sponsorship to Y.Z. from MOE Key Laboratory of Physics and Technology for Advanced Batteries affiliated with Jilin University is acknowledged. Notes
The authors declare no competing financial interest.
■
■
ACKNOWLEDGMENTS C. Tassel and Y. Kobayashi at Kyoto University are acknowledged for the fruitful discussion of Rietveld refinement and the help in collecting synchrotron diffraction results, respectively. The authors are grateful to D. Nishio-Hamane at Institute for Solid State Physics of University of Tokyo for the experimental assistance with scanning and transmission electron microscopy.
CONCLUSIONS This work addresses fundamental problems regarding bismuth and oxygen nonstoichiometries, the polymorphs, and their phase relation in an Aurivillius phase, Bi3.6V2O10. With reduced temperature, the two-dimensional framework allows interesting interlayer oxygen communication (diffusion) to hold the highly distorted Bi−O and V−O layers together. The [Bi2O2] sheets are found to be more flexible structurally, compositionally, and even in the charge. All the room temperature A-type Bi4−xV2O10.6−1.5x phases across a wide range of stoichiometries (0 ≤ x ≤ 0.4) maintain the compositions of [Bi2−xO2−1.5x] and [VO3.30] for the corresponding layers by switching the oxygen reduction from the V−O layer to the Bi−O layer (i.e., the precipitation of Bi metal). In the former layer, oxygen vacancies and charge are both ordered between the octahedrally (pyramidally) coordinated V4+ and the tetrahedrally coordinated V5+. Most surprisingly, the distribution of oxygenvacancies is different between the RT-stabilized high temperature γ-phase and the low temperature A-phase for Bi3.6V2O10. The Bi−O layer (V−O layer) is less (more) oxygen-deficient in the RT-stabilized γ-phase than in the A-phase, with vacancy order and charge order, which does not happen in the bismuth stoichiometric Bi4V2O11−x phase. This implies the necessity of oxygen diffusion from the Bi−O layer to the V−O layer to achieve the A-phase, which accounts for the observation that high temperature phases become quenchable from the high temperature of around 800 °C. Without resorting to foreign cation substitutions as traditionally done, the mobile oxygen in the current material can be stabilized to ambient temperature
■
REFERENCES
(1) Takada, K.; Sakurai, H.; Takayama-Muromachi, E.; Izumi, F.; Dilanian, R. A.; Sasaki, T. Nature 2003, 422, 53−55. (2) Bednorz, J. G.; Müller, K. A. Z. Phys. B: Condens. Matter 1986, 64, 189−193. (3) Kamihara, Y.; Watanabe, T.; Hirano, M.; Hosono, H. J. Am. Chem. Soc. 2008, 130, 3296−3297. (4) Noguchi, Y.; Miyayama, M.; Kudo, T. Appl. Phys. Lett. 2000, 77, 3639−3941. (5) Mizushima, K.; Jones, P. C.; Wiseman, P. J.; Goodenough, J. B. Mater. Res. Bull. 1980, 15, 783−789. (6) Padhi, A. K.; Nanjundaswamy, K. S.; Goodenough, J. B. J. Electrochem. Soc. 1997, 144, 1188−1194. (7) Miah, M. I. Mater. Chem. Phys. 2010, 119, 402−405. (8) Fang, P. H.; Robbins, C. R.; Aurivillius, B. Phys. Rev. 1962, 126, 892. Aurivillius, B.; Fang, P. H. Phys. Rev. 1962, 126, 893−896. (9) Leciejewicz, J. Acta Crystallogr. 1961, 14, 1304. (10) Goodenough, J. B. Annu. Rev. Mater. Res. 2003, 33, 91−128. (11) Galy, J.; Enjalbert, R. J. Solid State Chem. 1982, 44, 1−23. (12) Newnham, R. E.; Wolfe, R. W.; Dorrian, J. F. Mater. Res. Bull. 1971, 6, 1029−1039. (13) Hervoches, C. H.; Lightfoot, P. J. Solid State Chem. 2000, 153, 66−73. H
DOI: 10.1021/acs.inorgchem.5b01964 Inorg. Chem. XXXX, XXX, XXX−XXX
Article
Inorganic Chemistry (14) Haluska, M. S.; Misture, S. T. J. Solid State Chem. 2004, 177, 1965−1975. (15) Berdonosov, P. S.; Charkin, D. O.; Knight, K. S.; Johnston, K. E.; Goff, R. J.; Dolgikh, V. A.; Lightfoot, P. J. Solid State Chem. 2004, 177, 2632−2634. (16) Abraham, F.; Debreuille-Gresse, M. F.; Mairesse, G.; Nowogrocki, G. Solid State Ionics 1988, 28−30, 529−532. (17) Mairesse, G.; Roussel, P.; Vannier, R. N.; Anne, M.; Pirovano, C.; Nowogrocki, G. Solid State Sci. 2003, 5, 851−859. (18) Mairesse, G.; Roussel, P.; Vannier, R. N.; Anne, M.; Nowogrocki, G. Solid State Sci. 2003, 5, 861−869. (19) Zhang, Y.; Ueda, Y. Inorg. Chem. 2013, 52, 5206−5213. (20) Joubert, O.; Jouanneaus, A.; Ganne, M. Nucl. Instrum. Methods Phys. Res., Sect. B 1995, 97, 119−122. (21) Huve, M.; Vannier, R.-N.; Nowogrocki, G.; Mairesse, G.; Tendeloo, G. V. J. Mater. Chem. 1996, 6, 1339−1345. (22) Garcia-Gonzalez, E.; Arribas, M.; Gonzalez-Calbet, J. M. J. Mater. Chem. 2001, 11, 2320−2323. (23) Neagu, D.; Tsekouras, G.; Miller, D. N.; Menard, H.; Irvine, J. T. S. Nat. Chem. 2013, 5, 916−923. (24) Izumi, F.; Momma, K. Solid State Phenom. 2007, 130, 15−20. (25) Larson, A. C.; Von Dreele, R. B. GSAS General Structure Analysis System; Los Alamos National Laboratory: Los Alamos, NM, 1994. (26) Toby, B. H. J. Appl. Crystallogr. 2001, 34, 210−213. (27) Patabendige, C. N. K.; Azad, A. K.; Connor, P. A.; Rolle, A.; Irvine, J. T. S. Sci. Rep. 2013, 3, 1504. (28) Abraham, F.; Mentre, O. J. Solid State Chem. 1994, 109, 127− 133. (29) Colmont, M.; Huve, M.; Ketatni, E. M.; Mentre, O. Solid State Sci. 2008, 10, 533−543. (30) Steinfink, H.; Lynch, V. J. Solid State Chem. 2004, 177, 1412− 1415. (31) Krivovichev, S. V.; Mentre, O.; Siidra, O. I.; Colmont, M.; Filatov, S. K. Chem. Rev. 2013, 113, 6459−6535. (32) Zavalij, P. Y.; Whittingham, M. S. Acta Crystallogr., Sect. B: Struct. Sci. 1999, 55, 627−663. (33) Schindler, M.; Hawthorne, F. C.; Baur, W. H. Chem. Mater. 2000, 12, 1248−1259. (34) Nordlinder, S. Nanotubes for Battery Applications. Ph.D. Thesis, Uppsala University, Uppsala, Sweden, 2005. (35) Kendall, K. R.; Navas, C.; Thomas, J. K.; zur Loye, H.-C. Chem. Mater. 1996, 8, 642−649. (36) Abraham, F.; Boivin, J. C.; Mairesse, G.; Nowogrocki, G. Solid State Ionics 1990, 40−41, 934−937. (37) Steele, B. C. Solid State Ionics 2000, 129, 95−110.
I
DOI: 10.1021/acs.inorgchem.5b01964 Inorg. Chem. XXXX, XXX, XXX−XXX
