Probing Long- and Short-Range Disorder in Y2Ti2–xHfxO7 by

Nov 1, 2016 - We studied the long-range average and short-range local structures in ..... entry indicates the estimated standard deviation (ESD) refer...
0 downloads 0 Views 4MB Size
Article pubs.acs.org/JPCC

Probing Long- and Short-Range Disorder in Y2Ti2−xHfxO7 by Diffraction and Spectroscopy Techniques Zhaoming Zhang,*,† Maxim Avdeev,† Massey de los Reyes,†,§ Gregory R. Lumpkin,† Brendan J. Kennedy,‡ Peter E. R. Blanchard,‡,∥ Samuel Liu,‡ Anton Tadich,⊥ and Bruce C. C. Cowie⊥ †

Australian Nuclear Science and Technology Organisation, Lucas Heights, New South Wales 2234, Australia School of Chemistry, The University of Sydney, Sydney, New South Wales 2006, Australia ⊥ Australian Synchrotron, 800 Blackburn Road, Clayton, Victoria 3168, Australia ‡

S Supporting Information *

ABSTRACT: We studied the long-range average and short-range local structures in Y2Ti2−xHfxO7 (x = 0−2.0) using diffraction and spectroscopy techniques, respectively. Both neutron and synchrotron X-ray powder diffraction data show a clear phase transition of the average structure from ordered pyrochlore to disordered defect-fluorite at x ≈ 1.6; the long-range anion disorder appears to develop gradually throughout the entire pyrochlore region in contrast to the rapid loss of cation ordering from x = 1.4 to 1.6. The commonly observed two-phase region around the pyrochlore/defect-fluorite phase boundary is absent in this system, demonstrating high sample quality. X-ray absorption near-edge structure (XANES) results at the Y L2-, Ti K- and L3,2-, Hf L3-, and O K-edges indicate a gradual local structural evolution across the whole compositional range; the Y coordination number (CN) decreases and the CN around Ti and Hf increases with increasing Hf content (x). The spectroscopic results suggest that the local disorder occurs long before the pyrochlore to defect-fluorite phase boundary as determined by diffraction, and this disorder evolves continuously from short- to medium- and eventually to long-range detectable by diffraction. This study highlights the complex disordering process in pyrochlore oxides and the importance of a multitechnique approach to tackle disorder over different length scales and in the anion and cation sublattices, respectively. The results are important in the context of potential applications of these oxides such as ionic conductors and radiation-resistant nuclear waste forms.



INTRODUCTION There has been a great deal of interest in the A2B2O7 pyrochlore materials due to their chemical and structural flexibility, resulting in diverse properties.1 The most common pyrochlore oxides are the III/IV pyrochlores containing trivalent A-site and tetravalent B-site cations, because there are many A3+ and B4+ cations with suitable ionic radii to form the pyrochlore structure under ambient conditions (rA/rB = 1.46−1.78). Typical A3+-site cations include lanthanoids and Y3+, while B4+-site cations can be 3d-, 4d-, or 5d-transition metals (such as Ti4+, Zr4+, or Hf4+) or any of the IVB group metals (e.g., Sn4+ or Pb4+). Zirconate and titanate pyrochlore oxides are of particular interest as ionic and electronic conductors in solid oxide fuel cells2,3 and host matrices for the immobilization of actinide-rich nuclear wastes.4−6 Certain lanthanide zirconate pyrochlores have also been considered as potential inert matrix fuel materials.7,8 Many of the properties of pyrochlore are sensitive to the degree of disorder and related changes in the structure,4,9−12 and it was pointed out recently that the disordering process may be more complex than previously thought.13 The ideal (or © 2016 American Chemical Society

fully ordered) pyrochlore structure, A2B2O(1)6O(2), is cubic in space group Fd3̅m. It is closely related to the fluorite structure (A/B)O2 (space group Fm3̅m) and is often considered as an ordered fluorite with 1/8 (12.5%) oxygen vacancies. In the ideal pyrochlore structure, the A and B cations occupy the 16d (1/2, 1/2, 1/2) and 16c (0, 0, 0) sites (in origin choice 2) and the O(1) oxygen ions are located at the 48f (x48f, 1/8, 1/8) site and the O(2) at the 8b (3/8, 3/8, 3/8) site while the 8a (1/8, 1/8, 1/8) sites are unoccupied (Figure 1a). In contrast, both A and B cations occupy the same 4a (0, 0, 0) sites, and all seven oxygens occupy the 8c (1/4, 1/4, 1/4) sites in the defectfluorite structure (Figure 1b). The ordered arrangement of both cations and anions in pyrochlore leads to the doubling of its cell parameter compared to that of the fluorite (ap ≈ 2af). The larger A cation in pyrochlore is located in a scalenohedron (i.e., distorted cube) formed by six O48f and two O8b anions, while the smaller B cation is coordinated to six O48f anions Received: July 15, 2016 Revised: October 28, 2016 Published: November 1, 2016 26465

DOI: 10.1021/acs.jpcc.6b07076 J. Phys. Chem. C 2016, 120, 26465−26479

Article

The Journal of Physical Chemistry C

Figure 1. (a) Pyrochlore structure of formula A2B2O(1)6O(2), showing the A-site (blue) and B-site (green) polyhedra with the O(1) and O(2) oxygen atoms (located at the 48f and 8b sites) depicted as gold and red balls, respectively. (b) Defect-fluorite structure of formula A0.5B0.5O1.75□0.25. For comparison, the relative positions of the A-site (blue) and B-site (green) cations in b are the same as those in a; however, the A- and B-site cations are randomly distributed in the defect-fluorite structure. Similarly, 6/7 of the oxygens are still depicted as gold balls and the rest (1/7) as red balls, but they are located at equivalent positions. In addition, oxygen vacancies (□) are also included as black balls in (b).

(using synchrotron X-ray diffraction and XANES spectroscopy) and atomic-scale simulations. Such a multitechnique approach has enabled us to gain a more detailed understanding of order− disorder phenomena in the Y2Sn2−xZrxO7 pyrochlore/defectfluorite system. However, a recent study of the same materials, using NMR spectroscopy and DFT calculations, reached very different and conflicting conclusions.32 In contrast to the previous work, a very broad two-phase region (with Zr content ranging from 0.2 to 1.2) was proposed by Ashbrook et al.32 for the Y2Sn2−xZrxO7 series, and furthermore, different compositions were suggested for the two coexisting phases with the pyrochlore composition changing very little in the two-phase region (reaching a maximum of 13% Zr only, i.e., Y2Zr0.26Sn1.74O7), whereas the defect-fluorite composition varied throughout. The disorder in the pyrochlore phase, as revealed by X-ray absorption and Raman spectroscopy,25,33 was attributed solely to the presence of the coexisting defect-fluorite phase. In view of this new NMR study of Y2Sn2−xZrxO7, we revisited the order−disorder phenomena in pyrochlore/defectfluorite using a closely related Y2Ti2−xHfxO7 series. The choice of Hf, which displays similar crystal chemistry to Zr based on their nearly identical ionic radius, is to increase the ability to distinguish between Y3+ and Hf4+ using X-rays (although partial replacement of Ti by Hf reduces this contrast). The choice of Ti is to maximize the difference to neutrons between Y3+ and Ti4+ as the latter has a negative neutron scattering length. In other words, these samples are designed to circumvent the drawback of the previously studied Y2Sn2−xZrxO7 series, where all three cations have a similar number of electrons (Y3+ and Zr4+ are in fact isoelectronic) and similar neutron scattering length (bY = 7.75, bZr = 7.16, and bSn = 6.225 fm).34 Hence, it is difficult to experimentally quantify the degree of cation disorder in Y2Sn2−xZrxO7 using either X-ray or neutron diffraction. In the current study, we employed both synchrotron X-ray and neutron powder diffraction methods to determine the longrange average structure, with neutrons and X-rays being more sensitive to the anion and cation sublattices, respectively. The use of synchrotron XRD, which has superior peak shape resolution over laboratory XRD or constant-wavelength (CW)

within a trigonal antiprism (i.e., distorted octahedron), as shown in Figure 1a. The III/IV pyrochlore structure is stabilized when rA/rB = 1.46−1.78, whereas the defect-fluorite structure (or anion deficient fluorite A0.5B0.5O1.75) is formed when rA/rB < 1.46.1 The ordered pyrochlore structure can be transformed to the disordered defect-fluorite structure by a random distribution of both cations and anions onto their respective sublattices, with the A- and B-site coordination number (CN) changing from 8 and 6, respectively, to an average of 7 for both. Such a phase transition can be induced by temperature,14,15 pressure,16,17 composition,3,18−20 or ion irradiation.5,21,22 In addition to the fully ordered pyrochlore and completely disordered defect-fluorite, there is plenty of evidence for partially ordered pyrochlore and partially disordered defectfluorite.2,10,18,23−25 A recent study by Shamblin et al.26 has also highlighted the importance of disorder on different length scales. Using pair distribution function (PDF) analysis of neutron total scattering data, they revealed that although Ho2Zr2O7 has a disordered defect-fluorite structure on a large scale, it consists of ordered weberite type of units locally. However, it can be challenging to experimentally quantify the partially disordered state depending on the degree and length scale of the disorder.27 As such, different studies may reach apparently contradictory conclusions, depending on whether the technique employed is more sensitive to the cation or anion sublattice disorder and its probing length. On the basis of neutron powder diffraction (NPD) results, it was concluded that anion disorder precedes the cation disorder in Y2Ti2−xZrxO7 and Ho2−xNdxZr2O7.3,19 This was supported by Norberg et al.10 in their study of Y2Ti2−xZrxO7 using neutron total scattering. In contrast, first-principles calculations of defect-formation energies in Y2B2O7 pyrochlore (B = Sn, Ti, or Zr) predicted that cation antisite disorder causes oxygen disorder.28 X-ray absorption spectroscopic studies by Nachimuthu et al.,29−31 on the other hand, concluded that cation disorder occurs simultaneously with anion migration in Gd 2 Ti 2−x Zr x O 7 . This was supported by our study of Y2Sn2−xZrxO7,25 which combined both experimental work 26466

DOI: 10.1021/acs.jpcc.6b07076 J. Phys. Chem. C 2016, 120, 26465−26479

Article

The Journal of Physical Chemistry C

Figure 2. S-XRD (LHS) and NPD (RHS) patterns recorded from (a) Y2Ti2O7, (b) Y2Ti0.6Hf1.4O7, (c) Y2Ti0.4Hf1.6O7, and (d) Y2Hf2O7. Superlattice reflections indicative of the pyrochlore structure are indicated by an asterisk “*” and/or the subscript “p”. Black crosses represent the observed data, and solid red line is the fit obtained by the Rietveld method using the cubic structure in either Fd3m ̅ (pyrochlore) or Fm3m ̅ (defect-fluorite). Blue vertical markers show the peak positions expected in the structure, and green line beneath the pattern records the difference between the observed and the calculated patterns. Insets (LHS) enlarge the low-angle region to highlight the difference between the pyrochlore and the defect-fluorite structures. Note that due to the doubling of the pyrochlore cell parameter compared to that of the defect-fluorite (ap ≈ 2af), the (h k l) reflection in defect-fluorite corresponds to (2h 2k 2l) in pyrochlore, i.e., (111) and (002) peaks in defect-fluorite correspond to (222) and (004) in pyrochlore.

reduce the effects of preferred orientation and improve powder averaging. Neutron powder diffraction (NPD) measurements were performed on the high-resolution powder diffractometer Echidna at the Open Pool Australian Lightwater (OPAL) reactor, operated by the Australian Nuclear Science and Technology Organisation (ANSTO).41 The finely ground powder samples (∼2.5 g each) were loaded into a small (6 mm diameter) vanadium can to minimize the effect of absorption by Hf, and data were collected at room temperature over the range 5° < 2θ < 160° with a step size of 0.125° (2θ) with the wavelength of the incident neutrons fixed at 1.300 Å. This wavelength was obtained using a Ge (337) monochromator and calibrated against an Al2O3 standard. Structures were refined using the Rietveld method implemented in the program Rietica42 against the NPD and S-XRD data separately. Joint refinements were attempted initially; however, we were not able to achieve a consistent trend for the cation and anion disorder (presumably due to false local minima). The peak shape was modeled using a pseudo-Voigt function, and the background was estimated using a linear interpolation between a set of up to 50 background points. The scale factor, detector zero point, lattice parameters, atomic coordinates, and atomic displacement parameters (ADPs) were refined together with the peak profile parameters. ADPs were all taken to be isotropic and constrained to be equal for cations mixed on the same crystallographic site and for oxygen atoms in the similar 8a and 8b sites to reduce instability in the refinement process due to correlation of ADPs with the occupancies and positional parameter. While it was necessary to apply an absorption correction to the S-XRD data (to ensure positive ADPs), the

neutron diffraction, should enable us to observe coexisting pyrochlore and defect-fluorite phases (if present) and detect any composition inhomogeneity, as suggested to occur in the recent NMR study of Y2Sn2−xZrxO7.32 The short-range local structure is probed by X-ray absorption near-edge structure (XANES) spectroscopy, which has been extensively used to reveal the progressive nature of the structural transformation on the local scale in various pyrochlore and defect-fluorite materials.24,25,31,35−39



EXPERIMENTAL SECTION Samples with the composition of Y2Ti2−xHfxO7 (x = 0, 0.2, 0.4, 0.7, 0.9, 1.0, 1.1, 1.2, 1.3, 1.4, 1.6, 1.8, 2.0) were prepared using the conventional mixed metal−oxide process. All starting materials, commercially available yttrium oxide, titanium oxide, and hafnium oxide (all Aldrich, 99%), were heated to 850 °C for 10 h to completely remove adsorbed H2O and CO2. Stoichiometric mixtures were then attrition milled for 12 h at room temperature using zirconia balls and cyclohexane as the milling media. After drying at 250 °C, the finely ground powders were pressed into 10 mm diameter pellets using a cold uniaxial press at 200 bar. Pellets were then sintered in air at 1500 °C for 168 h without intermediate grinding. The pellets were crushed and ground to fine powder for diffraction and XANES measurements. Synchrotron X-ray diffraction (S-XRD) data were collected at room temperature in the angular range 5° < 2θ < 85° using X-rays of wavelength 0.8255 Å (based on calibration with a LaB6 standard) on the powder diffractometer at BL-10 of the Australian Synchrotron.40 Each sample was placed in a 0.3 mm glass capillary that was rotated during the measurements to 26467

DOI: 10.1021/acs.jpcc.6b07076 J. Phys. Chem. C 2016, 120, 26465−26479

Article

The Journal of Physical Chemistry C

scattering power of oxygen. The X-ray scattering power is proportional to the number of electrons: Z(Y3+) = 36, Z(Ti4+) = 18, Z(Hf4+) = 68, and Z(O2−) = 10, making X-rays less sensitive to the anion sublattice. In contrast, the neutron scattering length of oxygen is comparable to that of cations (bY = 7.75, bTi = −3.438, bHf = 7.77, and bO = 5.803 fm),34 resulting in greater sensitivity to the oxygen sublattice. Furthermore, the negative scattering length of Ti leads to additional sensitivity of neutrons to the cation ordering in pyrochlores at the Ti-rich end, as evidenced by the much stronger superstructure peaks in the NPD pattern of Y2Ti2O7 than the S-XRD pattern (see Figure 2a1 and 2a2). In fact, the intensity of some superlattice reflections in the NPD pattern of Y2Ti2O7 exceeds those of the substructure, which is quite unusual. As the Hf content (x) increases in the Y2Ti2−xHfxO7 series, the superlattice reflections generally decrease in intensity but the long-range average structure is still clearly pyrochlore up to and including x = 1.4 based on both S-XRD and NPD patterns (Figure 2b1 and 2b2). For the Hf content x = 1.6 sample, however, the S-XRD pattern shows no trace of any superlattice reflections belonging to the pyrochlore structure (Figure 2c1), while the NPD pattern still has some intensity for the (133)p and (115)p superstructure peaks (Figure 2c2). These peaks are quite weak and broadened, suggesting small pyrochlore-like domains in the Y2Ti0.4Hf1.6O7 sample. The effect of neutron absorption by Hf on the data was minimized through the use of small diameter sample holders. As the superstructure peaks were only detected in Y2Ti0.4Hf1.6O7 by neutrons not X-rays, this pyrochlore-like ordering is attributed to the oxygen sublattice only, similar to that reported in the Gd2−xTbxZr2O7 series39 and the stuffed pyrochlore Ho2TiO5.48,49 For the last two samples in the series (x = 1.8 and 2.0), both S-XRD and NPD patterns reveal a defect-fluorite structure (Figure 2d1 and 2d2). The S-XRD and NPD patterns corresponding to samples with the Hf content below 1.6 (0 ≤ x ≤ 1.4) were refined separately to the pyrochlore structure in space group Fd3̅m, and those obtained from samples with x equal to 1.8 and 2.0 were refined separately to the defect-fluorite structure in space group Fm3m ̅ . For the Hf content x = 1.6 sample, the defect-fluorite model was refined against the synchrotron X-ray data due to the absence of any superlattice reflections, whereas the pyrochlore model (with ordered anion sublattice only, see reasons given in the previous paragraph) was used in the analysis of the neutron data. Partially decoupled cation and anion sublattice ordering in pyrochlore systems has been reported previously as a function of composition50 or temperature.51 In both cases, the ordering in the anion sublattice precedes that of the cation sublattice. During the refinements against the neutron data in the pyrochlore region the oxygen occupancies were initially allowed to vary at all three oxygen sites (48f, 8a, and 8b) with the overall oxygen occupancy constrained. Within the experimental uncertainty the 8b site remained fully occupied, and hence, its occupancy was fixed to 1 in the final refinement. This indicates that the oxygen ions partially filling the nominally vacant 8a sites are solely from their next nearest oxygen neighbors at 48f sites, which is consistent with the proposed anion migration pathway in pyrochlore based on atomic-scale simulations.52,53 Attempts were also made to model the cation antisite disorder during the refinements against the X-ray and neutron data, with the total occupancy of Y, Ti, and Hf constrained by the chemical composition and both cation sites (16c and 16d) fully occupied. However, despite the fact that convergence could be achieved

neutron data were not corrected for absorption due to the negligible effect of this on the measured data. The Y L2-edge and Ti K-edge XANES spectra were collected on beamline 16A1 at the National Synchrotron Radiation Research Center (NSRRC) in Hsinchu, Taiwan.43 Finely ground powder samples were dispersed onto Kapton tape and placed in front of the X-ray beam at a 45° angle. Spectra were collected in bulk-sensitive fluorescence yield mode using a Lytle detector. Energy steps as small as 0.2 eV were employed near the absorption edge with a counting time of 2 s per step. The energy scale of the Y L2-edge was calibrated using the L3-edge of a pure Zr foil (with the maximum in the first derivative set to 2222.3 eV), and the Ti K-edge was calibrated against a metallic Ti foil (with the maximum of its first derivative set to 4966.4 eV). The Hf L3-edge XANES spectra were obtained on beamline 17C1 at NSRRC.44 Finely ground powder samples were mixed with an appropriate amount of boron nitride and pressed into pellets. Spectra were collected in transmission mode with an energy step of 0.4 eV near the absorption edge and a dwell time of 2 s. The monochromator was detuned by 50% to reject higher harmonics. For energy calibration purpose, a Zn reference foil was placed downstream and a Zn K-edge spectrum was taken simultaneously with each Hf L3-edge spectrum. The maximum in the first derivative of the Zn Kedge was set to 9658.4 eV. Ti L- and O K-edge XANES spectra were collected on the soft X-ray spectroscopy beamline at the Australian Synchrotron.45 Powder samples were thinly dusted onto double-sided carbon tape (SPI Supplies) and inserted into the vacuum chamber via a load lock. The pressure inside the analysis chamber was maintained at better than ∼10−9 Torr. Spectra were collected in X-ray fluorescence yield (FLY) mode, with a step size of 0.05 and 0.1 eV for the Ti L- and O K-edge, respectively. For energy calibration purpose, all spectra were taken simultaneously with a total electron yield (TEY) signal measured from a Ti metal (at Ti L-edge) or oxidized Al (at O K-edge) reference foil positioned upstream in the beamline. The reference foil removed approximately 10% of the beam intensity. This allowed for a precise energy alignment of the spectra obtained from different samples. The absolute energy scale was established based on the X-ray photoelectron spectrum of a gold foil by setting the binding energy of the Au 4f7/2 peak at 84.0 eV. Data analysis of all XANES spectra was carried out using the Athena software program.46 Additional peak fitting was performed using the CasaXPS software package.47



RESULTS AND DISCUSSION Long-Range Average Structure Determined by Synchrotron X-ray and Neutron Powder Diffraction. Both synchrotron X-ray and neutron powder diffraction patterns were obtained from all 13 samples in the Y2Ti2−xHfxO7 series, and examples of the S-XRD and NPD patterns are shown in the left and right panels in Figure 2, respectively. As mentioned in the Introduction, the pyrochlore structure can be described as a cation and anion/vacancy ordered derivative of the fluorite structure. Therefore, diffraction patterns of the pyrochlore phase contain extra superlattice reflections in addition to the main reflections observed from the defectfluorite phase. The intensity of these superstructure peaks depends on the difference between the average scattering length of the cation A and B sites as well as the relative 26468

DOI: 10.1021/acs.jpcc.6b07076 J. Phys. Chem. C 2016, 120, 26465−26479

Article

The Journal of Physical Chemistry C

Table 1. Crystallographic Parameters for Selected Y2Ti2−xHfxO7 Samples Obtained from Rietveld Refinements against the Neutron Diffraction Patterns Onlya x rA/rB space group a (Å) x(O1) A Uiso (10−2 Å2) B Uiso (10−2 Å2) O1 Uiso (10−2 Å2) O2/O3 Uiso (10−2 Å2) occupancy (O3) (%) Rp (%) Rwp (%) χ2 A−O1 (Å) (×6) A−O2 (Å) (×2) B−O1 (Å) (×6)

0 1.684 Fd3̅m 10.1035(1) 0.3289(1) 0.84(3) 0.62(6) 0.83(2) 0.54(4) 1.2(10) 4.83 6.61 8.54 2.4857(10) 2.18748(3) 1.9559(6)

1.0 1.550 Fd3̅m 10.2675(2) 0.3376(2) 1.33(3) 3.22(21) 2.39(4) 1.00(6) 14.4(12) 3.60 4.82 4.65 2.4647(12) 2.22298(5) 2.0257(8)

1.4 1.502 Fd3̅m 10.3227(3) 0.3435(2) 1.68(5) 2.46(11) 3.06(7) 2.05(10) 30.0(19) 3.51 4.72 3.50 2.4370(16) 2.23493(7) 2.0644(11)

1.6b 1.479 Fd3̅m 10.3447(3) 0.3592(7) 1.88(13) 2.21(15) 5.10(25) 4.43(65) 44.6(46) 3.59 4.90 3.59 2.3358(47) 2.23753(7) 2.1473(40)

1.8 1.457 Fm3̅m 5.1840(2)

2.0 1.435 Fm3̅m 5.1995(2)

2.33(3)

2.24(3)

5.32(7)

4.97(7)

3.25 4.55 5.25 2.24474(8)

3.88 5.09 2.15 2.25145(7)

a Fd3̅m: A at 16d, B at 16c, O1 at 48f, O2 at 8b, and O3 at 8a. Fm3̅m: A = B at 4a, O1 = O2 = O3 at 8c. The number in parentheses beside each entry indicates the estimated standard deviation (ESD) referred to the last digit(s) shown. Note that the ESDs here are underestimated as they do not take into account the errors in data collection. bSince there is no evidence for any ordering in the S-XRD pattern for the x = 1.6 sample, cations were distributed uniformly across A and B sites in the pyrochlore structure (i.e., only the anion sublattice is ordered).

for each individual refinement, the partitioning of the three cations seemed to depend on the initial cation occupancies selected for each refinement, suggesting the refinement converged to a local minimum instead of a global minimum. As a result, the obtained cation antisite disorder appeared to vary erratically with increasing Hf content (not to increase systematically as would be expected). Further attempts were made to assess the likelihood of cation antisite disorder in Y2Ti1.0Hf1.0O7 by manually fixing the cation disorder and refining the pyrochlore structure accordingly (see Table S1 in the Supporting Information for details of the 52 different combinations of cation disorder). By comparing the goodnessof-fit of all 52 refinements (Figure S1), it was concluded that cation antisite disorder is not likely present in Y2Ti1.0Hf1.0O7. Therefore, cation occupancies were not refined in the final refinements of any pyrochlore samples. The crystallographic parameters obtained from Rietveld refinements against the NPD data are listed in Table 1 for a selected number of samples (x = 0, 1.0, 1.4, 1.6, 1.8, and 2.0). The results show that the Ti end member (Y2Ti2O7) is essentially a fully ordered pyrochlore. However, anion disorder appears to develop as soon as Hf (x = 0.2) is introduced on the B site to replace Ti. The positional parameter of O(1) oxygen, x(48f), is observed to increase with increasing Hf content (x) as shown in Figure 3, ranging from 0.3289 at x = 0 to 0.3592 at x = 1.6 (recall that when x = 0.375 the anions are at the equivalent position for the defect-fluorite structure). The oxygen 8a site occupancy, which provides a direct measure of anion disorder in the pyrochlore structure, appears to scale linearly with x(48f) (see inset of Figure 3) similar to the trend reported previously in Nd2−xHoxZr2O719 and Nd2−xYxZr2O7.38 The disordering process is also reflected by increasing atomic displacement parameters (Uiso) of both cations and anions. The observed pyrochlore−defect-fluorite phase boundary at x ≈ 1.6 agrees well with that reported by Kong et al.54 and the predicted phase boundary of rA/rB = 1.46.1 Although neutrons are more sensitive to the anion sublattice (i.e., NPD provides more accurate oxygen positions and bond distances), more accurate lattice parameters are determined by

Figure 3. Compositional dependence of the oxygen 48f positional parameter, x(48f), as determined by Rietveld refinements against the NPD data only. (Inset) Approximately linear relationship between the oxygen 8a site occupancy and x(48f) in the pyrochlore structure. Where not apparent the estimated errors are smaller than the symbols.

S-XRD due to its superior resolution. Figure 4 illustrates the compositional dependence of the unit cell parameter (ap or 2af) for Y2Ti2−xHfxO7 as obtained by Rietveld refinements against the S-XRD data only (the results from NPD are similar; see Figure S2). The lattice parameter increases with the increase of Hf content, as a result of replacing smaller Ti4+ ions (IR = 0.605 Å) with larger Hf4+ ions (IR = 0.71 Å) on the octahedral B site.55 As shown in Figure 4, the lattice parameters in each phase region were fitted with a linear function, in excellent agreement with Vegard’s law. It is also interesting to note that the lattice parameter of the pyrochlore phase is noticeably larger than twice that of the defect-fluorite phase; this decreased volume of the defect-fluorite phase was also observed for La 2−x Y xZr2 O 7 , 18 Ho 2−x Nd x Zr 2 O 7 ,19 Y 2 Ti 2−x Zr xO 7 , 20 and Y2Sn2−xZrxO7.25 Although the exact reason for this phenomenological observation is unclear, the decreased volume of the defect-fluorite phase (i.e., more efficient packing) is believed to be related to the disorder in the defect-fluorite structure.19 26469

DOI: 10.1021/acs.jpcc.6b07076 J. Phys. Chem. C 2016, 120, 26465−26479

Article

The Journal of Physical Chemistry C

Figure 4. Compositional dependence of the appropriately scaled unit cell parameter of Y2Ti2−xHfxO7 as determined by Rietveld refinements against the S-XRD data only. Black squares correspond to the unit cell parameter of the pyrochlore phase (ap), and red circles denote twice the unit cell parameter of the defect-fluorite phase (2af). Note the estimated errors are smaller than the symbols.

A common feature observed in pyrochlore/defect-fluorite solid solutions is the presence of a two-phase region near the phase boundary.19,20,25,38 Due to the fact that the lattice parameters of the two phases are very close (i.e., ap ≈ 2af), it is not easy to detect the two-phase region unless the diffractometer peak shape resolution is sufficiently high to resolve the overlapping peaks. For example, the two-phase region in the Y2Sn2−xZrxO7 samples was only detected by synchrotron XRD,25 not by the lower resolution laboratory XRD.33 The current Y2Ti2−xHfxO7 system, however, shows no evidence for a two-phase region (within the composition step of Δx = 0.2), as the S-XRD patterns show no asymmetric peak broadening caused by peak splitting of the coexisting pyrochlore and defect-fluorite phases. We examined the peak shapes closely not only at low 2θ (i.e., high intensity peaks) but also at high 2θ where, although the intensities are less, such an asymmetric peak broadening would be more obvious, see Figure S3 for plots of the (222)p and (488)p reflections for the x = 1.3 sample showing the same symmetric peak shape. Note that the slight broadening of the (222)p reflection for the x = 0.4 sample evident in Figure 5a is believed to be a consequence of local strains, as the (488)p reflection has the same peak shape as that of (222)p, Figure S4. Examples of the main and superlattice reflection peaks in the S-XRD patterns are shown in Figure 5a and 5b, respectively, with the peak positions plotted as a function of the Hf content in Figure 5c. For ease of comparison, the (222)p/(111)f peak height in Figure 5a is fixed to 106, and the (111)p peak height in Figure 5b is normalized accordingly. Figure 5a−c illustrates the gradual (approximately linear) peak shift with the substitution of Ti by Hf for both the (222)p/(111)f main and the (111)p superstructure peaks, without any clear peak splitting over the entire composition range. These results not only demonstrate the absence of a twophase region in the current Y2Ti2−xHfxO7 system but more importantly also provide clear evidence to discount the model proposed by Ashbrook et al.32 to explain the disorder behavior observed in the closely related Y2Sn2−xZrxO7 series. On the basis of their NMR spectroscopic study and DFT calculations, it was concluded that a very broad two-phase region exists for 0.2 ≤ x ≤ 1.2 in Y2Sn2−xZrxO7, and moreover, the pyrochlore phase can only incorporate a maximum of 13% Zr (i.e., x = 0.26), whereas the amount of Zr contained in the defect-fluorite

Figure 5. Plot of the (a) main pyrochlore (222)p or defect-fluorite (111)f peak and (b) superlattice pyrochlore (111)p peak from S-XRD patterns of Y2Ti2−xHfxO7, with the (222)p/(111)f peak height normalized to 106. Note the sudden disappearance of the pyrochlore (111)p peak for x > 1.4 in b. Peak position of both is displayed in c as a function of the Hf content (x).

phase increases with the total Zr content in the sample. This means that the composition of the pyrochlore phase would be mostly fixed in the proposed very wide two-phase region, whereas that of the defect-fluorite phase would contain more Zr than the nominal composition. Therefore, the diffraction peak positions from the pyrochlore structure (both the main and superstructure peaks) would stop shifting with x, while the defect-fluorite peaks would keep shifting toward lower angle (i.e., larger cell parameter) with the progressive replacement of smaller Sn4+ by larger Zr4+. As a result the main peaks in the diffraction pattern, e.g., the overlapped (222)p/(111)f peak from both phases, would get broader (and eventually split into two resolved peaks) with increasing difference in the composition of the two phases, and the superstructure peaks 26470

DOI: 10.1021/acs.jpcc.6b07076 J. Phys. Chem. C 2016, 120, 26465−26479

Article

The Journal of Physical Chemistry C

F(222) is independent of how the three cations are distributed over the A and B sites. To further simplify the expressions above, two cases of cation antisite disorder are considered where z is the amount of Hf or Ti on the A site and Y on the B site of pyrochlore: (1) Y ↔ Hf (i.e., A = Y1−zHfz and B = Hf0.7−zYzTi0.3) → F(111) ≈ −161.581 + 522.08z (0 < z ≤ 0.7) and (2) Y ↔ Ti (i.e., A = Y1−zTiz and B = Hf0.7YzTi0.3−z) → F(111) ≈ −161.581 − 259.04z (0 < z ≤ 0.3). Therefore, the ratio of |F(111)|2/ |F(222)|2 at x = 1.4 would be either (−0.1243 + 0.4017z)2 for Y ↔ Hf disorder or (0.1243 + 0.1993z)2 for Y ↔ Ti, which means that antisite disorder between Y and Hf would reduce the ratio of |F(111)|2/|F(222)|2 whereas disorder between Y and Ti would increase the ratio. The fact that the I(111)/ I(222) intensity ratio at x = 1.4 falls below the predicted curve in Figure 6 suggests that antisite disorder between Y and Hf would be favored over that between Y and Ti; this seems to be logical as Hf4+ is closer in size to Y3+ than Ti4+. This also agrees with the results reported by Norberg et al.10 for the closely related Y2Ti2−xZrxO7 series using neutron total scattering that antisite disorder between Y and Zr precedes that between Y and Ti. However, we cannot completely rule out the possibility of a smaller number of Y ↔ Ti antisite defects in Y2Ti0.6Hf1.4O7, as it is possible, mathematically at least, to have a larger number of Y ↔ Hf antisite defects to cancel out the effect of Y ↔ Ti antisite disorder on |F(111)|2/|F(222)|2 (i.e., with the same net effect). In summary, our synchrotron diffraction data indicate an abrupt transition from pyrochlore at x = 1.4 to defect-fluorite at x = 1.6, without any evidence for significant cation disordering in pyrochlore samples with x ≤ 1.3. A comparison of the calculated X-ray structure factors with the observed intensity ratio of I(111)/I(222) suggests antisite disorder between Y and Hf at x = 1.4. Neutron diffraction results are somewhat different, indicating the gradual development of anion disorder in Y2Ti2−xHfxO7 throughout the entire pyrochlore region (i.e., preceding cation disorder). Although the S-XRD pattern for the x = 1.6 sample shows no sign of any pyrochlore ordering, the NPD pattern still provides evidence for pyrochlore-like ordering in the anion sublattice. This highlights the importance of employing both synchrotron X-ray and neutron powder diffraction for their respective sensitivity to the cation and anion sublattices. Finally, the superb peak-shape resolution of S-XRD has allowed us to demonstrate the absence of a twophase region near the pyrochlore/defect-fluorite phase boundary (within the composition step of Δx = 0.2), as commonly reported for similar systems. Short-Range Local Structure Determined by XANES Spectroscopy. While diffraction is sensitive to long-range order and gives the structure averaged over thousands of unit cells, XANES probes the short-range local chemical environment including valence state, coordination, and site geometry. The latter technique is particularly useful to study doped samples (or solid solutions) as diffraction methods only determine the average location of atoms occupying the same crystallographic site (i.e., essentially the location of the major components in the crystal rather than the location of dopants), and local distortions may be averaged out in diffraction measurements. For example, Bragg diffraction results do not distinguish between Ti and Hf in individual sites in the Y2Ti2−xHfxO7 pyrochlore samples. The combination of diffraction and spectroscopy techniques is required in order to provide a full description of systems with localized disorder.

from the pyrochlore phase alone, e.g., the (111)p peak, would have fixed positions (corresponding to a maximum of 13% Zr in the pyrochlore phase) irrespective of the change in nominal composition. Clearly the current Y2Ti2−xHfxO7 series does not show any of the behavior described above as evidenced in Figure 5a−c. It should be mentioned that Ashbrook et al.32 acknowledged the fact that their disorder model contradicts results from published diffraction studies of Y2Sn2−xZrxO7,25,33 and they attributed this discrepancy to the assumptions used in the structural refinement process. However, our results shown in Figure 5 are based on experimentally observed diffraction data without involving any structural refinements, and these results clearly demonstrate that the assumptions used to model the NMR spectra in ref 32 are invalid. The peak height of the S-XRD (111)p peak (Figure 5b) is normalized to that of the main (222)p peak (Figure 5a). The sudden disappearance of the (111)p superlattice reflection at x = 1.6 is striking, which is attributed to the rapid loss of cation ordering from x = 1.4 to 1.6. In order to confirm this, X-ray structure factors were calculated for the pyrochlore (111)p and (222)p peaks at x = 0−1.4 assuming perfect cation ordering (i.e., no cation antisite disorder); see the Supporting Information for the formula of F(111) and F(222). Since the intensity of the (hkl) reflection is proportional to the square of the amplitude of the corresponding structure factor, |F(hkl)|2, the intensity ratio of I(111)/I(222) should be proportional to |F(111)|2/|F(222)|2 if the assumption of no cation antisite disorder is valid. Figure 6a shows the plot of |F(111)|2/

Figure 6. Plot of (a) calculated |F(111)|2/|F(222)|2 assuming no cation disorder, and (b) observed I(111)/I(222) from the S-XRD patterns as a function of the Hf content (x) in Y2Ti2−xHfxO7.

|F(222)|2 as a function of the Hf content (x) in Y2Ti2−xHfxO7, utilizing the O48f positional parameter and oxygen occupancies obtained from the refinements against the NPD data. As seen, the observed relationship of I(111)/I(222) with x (Figure 6b) follows the trend of |F(111)|2/|F(222)|2 reasonably well until x = 1.3, suggesting minimal cation antisite disorder for pyrochlore samples with x ≤ 1.3. The measured intensity ratio I(111)/ I(222) levels off with a further increase of the Hf content to 1.4, in contrast to the calculated |F(111)|2/|F(222)|2 ratio, which continues to increase with x, signaling the onset of substantial amount of cation disorder. Taking cation antisite disorder into consideration, the structure factors of F(111) and F(222) for the x = 1.4 sample can be derived as F(111) = −8f(B) − 2.365f(O) + 8f(A), where f(A), f(B), and f(O) correspond to scattering from the A- and B-site cations and oxygen, respectively, and F(222) = 16f(B) + 16f(A) ≈ 1299.55, i.e., 26471

DOI: 10.1021/acs.jpcc.6b07076 J. Phys. Chem. C 2016, 120, 26465−26479

Article

The Journal of Physical Chemistry C In the fully ordered pyrochlore structure, as shown in Figure 1a, the smaller B cation is coordinated to six equally distant O48f anions within a trigonal antiprism (distorted octahedron), while the larger A cation is in a scalenohedron (distorted cube) formed by six O48f anions situated at equal distances and two O8b anions at slightly shorter distances. The shape of the coordination polyhedra changes with the positional parameter of 48f oxygen anions, which can vary within the range of 0.3125 and 0.375.1 The B-site cation has a perfect octahedral coordination when x(48f) = 0.3125, while the A-site cation sits in a perfect cubic environment when x(48f) = 0.375 (corresponding to the ideal fluorite arrangement). For convenience, octahedral and cubic symmetry are used to approximate the actual (lower) local symmetry in pyrochlore around the B- and A-site cations, respectively, and the XANES results are discussed within the framework of crystal field theory (CFT). For an octahedral MX6 unit (where M is at the origin and X ligands are positioned on the Cartesian axes, e.g., Ti in SrTiO356), the triply degenerate t2g orbitals (dxy, dxz, and dyz) are located at lower energy than the doubly degenerate eg orbitals (dx2−y2 and dz2), as the latter point toward the ligand orbitals and hence are subject to a stronger electrostatic repulsion from the ligands. In contrast, the three d-t2g orbitals lie higher in energy with respect to the two d-eg orbitals for a cubic MX8 unit (where M is at the origin and X ligands are situated at the cube corners, e.g., Ca in CaF256), as the t2g orbitals are now oriented with their lobes much closer to the ligands than the eg orbitals. Figure 7a shows the Y L2-edge XANES spectra for the Y2Ti2−xHfxO7 series, resulting primarily from the Y 2p1/2 → 4d dipole-allowed transitions, providing direct information on the occupancy and energy distribution of the final Y(4d) states. Since these Y(4d) orbitals are directly involved in bonding, the local coordination environment around Y can be inferred from the line shape of the spectra.25,38 As shown, each Y L2-edge spectrum displays a bimodal feature (labeled A and B), which corresponds to the splitting of the unoccupied Y(4d) orbitals under the local crystal field exerted by the surrounding oxygen ligands. At the pyrochlore end, the two observed features can be attributed to the splitting of the Y(4d) orbitals into the eglike and t2g-like orbitals at a distorted cubic site. The energy of feature A remains unchanged throughout the series, whereas that of feature B increases with increasing Hf content (x). The energy gap between A and B reflects the crystal field splitting (CFS), which increases with increased interaction between the Y(4d) and the ligand 2p orbitals. In order to quantify the splitting, the spectra were fitted by a least-squares algorithm. An example of the fitted XANES spectrum is shown in Figure 7b for Y2Hf2O7, where electronic transitions to the vacant Y(4d) orbitals are represented by Voigt-shaped peaks and the absorption edge jump by an arctangent function. While feature A can be fitted by a single peak (A), two peaks (B1 and B2) are necessary to fit feature B due to the shoulder on the high energy side. A similar high-energy shoulder was observed at the Zr L3,2-edge in several tetravalent zirconium oxides with CN ranging from 6 to 8, and its origin was ascribed to solid state effects.57 The intensity of B2 in our samples decreases with increasing Hf content (i.e., as the structure progresses from ordered pyrochlore to disordered defect-fluorite), the same trend as that reported for Nd2−xYxZr2O7.38 Initially no restrictions were imposed in the peak-fitting process. However, it became evident that, due to significant peak overlap, constraints had to be applied in order to achieve stable and

Figure 7. (a) Normalized Y L2-edge XANES spectra of the Y2Ti2−xHfxO7 series (spectra have been offset vertically to enhance visibility, and two solid lines are drawn to indicate the peak movement). Note that the labeling of eg (peak A) and t2g (peak B) orbitals is only applied to samples with the pyrochlore structure, not those with the defect-fluorite structure. (b) Fitted XANES spectrum of Y2Hf2O7. (c) Energy difference between the fitted peaks of B1 and A, and peak intensity ratio of IA/IB (where IB = IB1 + IB2) as a function of Hf content (x).

consistent fitting results. In the final fitting process, the full width at half-maximum (fwhm) of each peak was fixed at 1.65, 2.55, and 3.00 eV for peaks A, B1, and B2 respectively; these peak widths gave rise to satisfactory fits for all samples in the series. The energy difference between the fitted peaks of B1 and A (ΔE) and the peak intensity ratio of IA/IB (where IB = IB1 + IB2) are plotted as a function of Hf content (x) in Figure 7c. Both ΔE and IA/IB increases gradually with increasing Hf content (i.e., as the average structure evolves from ordered pyrochlore toward disordered defect-fluorite), similar to those reported for 26472

DOI: 10.1021/acs.jpcc.6b07076 J. Phys. Chem. C 2016, 120, 26465−26479

Article

The Journal of Physical Chemistry C Nd2−xYxZr2O738 and Y2Sn2−xZrxO7.25 The increase in ΔE, which is directly related to CFS, is attributed to the increased interaction between the Y(4d) and the O(2p) states, as the Y coordination number (CN) decreases from 8 in pyrochlore to an average of 7 in defect-fluorite (with concurrent decrease in Y−O bond lengths). It is evident that changes seen in the XANES data evolve continuously throughout the whole series starting with the smallest amount of Hf dopant (Y2Ti1.8Hf0.2O7), well before the pyrochlore to defect-fluorite phase boundary at x ≈ 1.6 as revealed by diffraction methods. The fact that there is no abrupt change at the phase boundary in either ΔE or IA/IB suggests that on the local scale the order to disorder transformation develops gradually, and the abrupt phase transformation seen by diffraction simply reflects the length scale of the disorder reaching a critical value. Ti L3,2-edge XANES spectra result predominantly from the excitation of Ti 2p3/2,1/2 electrons into the unoccupied 3d states and are very sensitive to the local structure and, importantly, the medium-range order as well.58−60 Figure 8a displays the Ti L3,2-edge XANES spectra from Y2Ti2−xHfxO7, comprising the L3-edge (with main peaks C and D plus pre-edge peaks A and B) and L2-edge (with peaks E and F), respectively. The separation between the L3 and L2 edges results from the spin− orbit coupling of the 2p core hole, which splits the Ti 2p orbitals into the Ti 2p3/2 (L3) and Ti 2p1/2 (L2) levels. Each edge splits further into sub-bands due to strong crystal field from the surrounding oxygen ligands. The L2-edge features are significantly broadened compared to those of the L3-edge (owing to the much shorter lifetime of 2p1/2 core holes via Coster−Kronig decay)58 and hence do not show as fine details in the peak shape. As mentioned earlier, the B-site Ti is in a distorted octahedron in the ideal pyrochlore structure, and the approximate octahedral ligand field splits each edge further into the t2g-like (peak C) and eg-like (peak D) states. Although no peak shifts are observed at the L2-edge (i.e., peaks E and F), there are significant changes at the L3-edge across the series especially in the line shape of the eg-like peak (D). At the pyrochlore end, the eg-like component is comprised of two separate peaks (D1 and D2, see Figure 8b). As the Hf content (x) increases, the two well-resolved peaks of the eg-like component gradually merge together and the relative intensity of the two peaks changes from a slightly lower intensity peak at the higher energy side to a shoulder on the lower energy side. The same trend was observed in Gd2Ti2−xZrxO731 and Yb2Ti2−xFexO7‑δ.61 The pre-edge peaks (A and B) broaden significantly with the substitution of Hf for Ti and get weaker with increasing Hf content. These pre-edge peaks are believed to be an atomic multiplet feature due to the strong interaction between poorly screened 3d electrons and the 2p core hole.58 The position of the t2g-like peak (C) does not change with x, but its peak height decreases relative to that of D with increasing x. It should be mentioned that the B-site cation is no longer located in a distorted octahedron in the defect-fluorite structured sample (x = 1.8), so it is not correct to call peaks C and D as the t2g-like and eg-like components for this sample. However, we will describe peak C as “L3-t2g” and D as “L3-eg” for all samples in the following discussion for simplicity. The splitting of the L3-eg peak into an asymmetric doublet was initially attributed to the local distortion of the TiO6 octahedron.58,62 Although this explanation was questioned by Crocombetter and Jollet59 in 1994, it was widely accepted until 2010 when an alternative explanation was proposed by Krüger60 based on first-principles multichannel multiple-

Figure 8. (a) Normalized Ti L3,2-edge XANES spectra of the Y2Ti2−xHfxO7 series (spectra have been offset vertically to enhance visibility, and vertical lines are drawn to guide the eyes). (b) Fitted Ti L3-edge XANES spectra of Y2Ti0.2Hf1.8O7 and Y2Ti2O7 after linear background removal. (c) Energy separation between peaks C and D in a as well as the peak intensity ratio (IC/ID) as a function of Hf content (x).

scattering (MCMS) calculations involving large numbers of TiO6 clusters. The experimentally observed splitting of the L3-eg band in TiO2 could only be reproduced by considering large clusters containing at least 60 atoms (on a length scale of ∼1 nm) and thus was concluded as a “nonlocal” effect.60 Therefore, the line shape of the L3-eg band is very sensitive to the mediumrange order, which was supported by studies of Pb1−xLaxTiO3,63 titania aerogels64 as well as ultrathin films of TiO2 on F-doped SnO2.65 This explanation is also in excellent agreement with our results (Figure 8a) that disorder around Ti atoms is not limited to the first nearest neighbor, and it 26473

DOI: 10.1021/acs.jpcc.6b07076 J. Phys. Chem. C 2016, 120, 26465−26479

Article

The Journal of Physical Chemistry C develops gradually and continuously throughout the entire Y2Ti2−xHfxO7 series and eventually reaches the length scale detectable by diffraction at the pyrochlore to defect-fluorite phase boundary. As in the case of the Y L2-edge XANES spectra, the crystal field splitting of the Ti(3d) orbitals can also be used as a measure of the local disorder around Ti. However, it is not as straightforward to define CFS of the Ti(3d) orbitals (i.e., ΔE between peaks C and D) due to the considerable changes in the line shape of peak D. Again, peak deconvolution was employed in order to quantify the changes; the Ti L3-edge spectra were fitted by five Voigt-shaped peaks after linear background removal (see examples in Figure 8b for fitted spectra of Y2Ti0.2Hf1.8O7 and Y2Ti2O7). Note that the choice of a linear background was purely for simplicity; ideally an arctangent background would be used; however, the presence of the adjacent L2-edge means that the “true” L3-edge background cannot be determined accurately from the data. As shown, peak D consists of two peaks (D1 and D2); we have taken the peak position ED as the weighted average of ED1 and ED2 (energies of peaks D1 and D2, respectively), i.e., ED = (ED1ID1 + ED2ID2)/ (ID1 + ID2). The CFS-related ΔE is then taken as the energy difference between EC (energy of peak C) and ED, which is seen to decrease with increasing Hf content (Figure 8c). The decrease of crystal field splitting, associated with Ti(3d) orbitals, with the increase of Hf content implies decreased interaction between the Ti(3d) and the ligand 2p orbitals as a result of increased Ti−O bond distance. Note that this inferred increase in the local Ti−O bond distance is attributed to increased CN with oxygen; hence, it is different from the increased average B−O bond distance derived from diffraction measurements (due to the replacement of smaller Ti4+ by larger Hf4+). Also plotted in Figure 8c is the relationship between the peak intensity ratio (IC/ID) and the Hf content (x); this ratio is also seen to decrease with increasing Hf content (x). The easiest way to understand the decrease in the peak intensity ratio at the Ti L3-edge with increasing CN (or increase at the Y L2-edge with decreasing CN) is to approximate the intensity ratio to that of the number of unoccupied states (i.e., proportional to t2g/eg = 3/2 at the octahedral site and eg/t2g = 2/3 at the cubic site), and then the ratio would decrease with increasing CN, from 3/2 at CN = 6 to 2/3 at CN = 8.25,57 Although this simplistic picture explains the trend in the peak intensity ratio, the “real” values deviate significantly from the statistical values of 3/2 or 2/3 due to ligand field multiplet effects.56 Further, the values plotted in Figures 7c and 8c are significantly impacted by the peak-fitting process (e.g., how overlap peaks and the background are treated). Similarly, the peak intensity ratio was reported to increase at the Zr L3-edge for Ln2Zr2O7 (Ln = lanthanoid),24 Nd2−xYxZr2O7,38 and Gd2−xTbxZr2O739 with increasing rA/rB (i.e., from defectfluorite to pyrochlore, or decreasing CN). XANES spectra were also obtained from Y2Ti2−xHfxO7 at the Ti K-edge, and the results are similar to those reported for Gd2Ti2−xSnxO7,35 Gd2Ti2−xZrxO7,36 and Y2Ti2−xHfxO7 synthesized by a soft chemistry route.54 The main-edge feature (labeled B in Figure 9) corresponds to a dipole transition from the Ti(1s) core level to the unoccupied Ti(4p) states. As shown, the main edge does not shift with the substitution of Hf for Ti, confirming that Ti remains tetravalent throughout the Y2Ti2−xHfxO7 series. However, the intensity of the main absorption peak decreases with increasing Hf content (x), which signifies increasing disorder in the samples.66,67 On the

Figure 9. Normalized Ti K-edge XANES spectra of the Y2Ti2−xHfxO7 series, with the pre-edge region enlarged in the inset.

basis of simulated Ti K-edge XANES spectra of Gd2Ti2O7 by Sanjuán et al.,68 the decrease in the intensity of the white line is consistent with the increase in (local) cation antisite defects. This seems to suggest that although diffraction methods failed to detect any cation antisite defects until x approaches the pyrochlore−defect-fluorite phase boundary, such disorder exists locally at much smaller x and continues to evolve from short- to medium- and eventually to long-range (i.e., hundreds of nanometers). Comparing the XANES spectra at the Ti Kedge (Figure 9) with those at the L-edge (Figure 8a), it is obvious that the changes at the Ti K-edge are more subtle than those observed at the Ti L3-edge. This is because the K-edge and L-edge probe the unoccupied density of states of the Ti(4p) and Ti(3d) orbitals, respectively, and, unlike the Ti(3d) orbitals, the Ti(4p) orbitals are not directly involved in bonding. The pre-edge region of the Ti K-edge (inset of Figure 9) is, however, quite sensitive to the local coordination environment around the Ti atom. Although the Ti(1s) to Ti(3d) transition is forbidden by dipole selection rules, the pre-edge probes the unoccupied Ti(3d) states via the 1s to 3d quadrupole transition as well as the 4p components in 3d−4p hybridized orbitals.69 The 3d−4p mixing can be either local (i.e., 3d and 4p states of the same Ti atom) if the absorbing atom site is noncentrosymmetric (e.g., tetrahedral site) or nonlocal where the empty 4p states of the absorbing Ti atom are hybridized with the empty 3d states of the Ti next nearest neighbors via O 2p states.70−72 Because the cation site symmetry is D3d in pyrochlore and Oh in defect-fluorite (both are centrosymmetric), there is no localized mixing between 3d and 4p states (i.e., no local dipole transition). Figure 9 shows three features in the pre-edge region of the Ti K-edge, which result from either local quadrupole or nonlocal dipole transitions. Feature A1 is attributed to the 1s → 3d-t2g local transitions, and A3 originates from the nonlocal 1s → 4p transitions. Feature A2 is comprised of both local and nonlocal contributions: the lower energy region (A2a) results from local 1s → 3d-eg transitions, and the higher energy region (A2b) arises from nonlocal intersite hybridization.35,71 The spectrum of the Ti end member, Y2Ti2O7, is similar to that of Gd2Ti2O7.35,36 With the introduction of Hf to replace Ti, the most obvious change in the pre-edge region is the decrease in intensity of the two “nonlocal” peaks (A2b and A3), indicating that the central Ti has fewer unoccupied next-nearest neighbor Ti(3d) states. This is expected as some of the Ti next-nearest neighbors are now replaced by Hf, leading to decreased intersite hybridization.35 26474

DOI: 10.1021/acs.jpcc.6b07076 J. Phys. Chem. C 2016, 120, 26465−26479

Article

The Journal of Physical Chemistry C The A1 and A2a peaks are broadened gradually with the increase of Hf content (x), becoming significantly overlapped for the defect-fluorite sample (x = 1.8). The peak position of A2a also shifts toward lower energy with increasing Hf content; this decreased energy gap between A1 and A2a indicates smaller CFS of the Ti(3d) orbitals as these two peaks represent quadrupole excitations from the Ti(1s) electrons to the crystal field split 3d-t2g and 3d-eg states, respectively. The trend observed here agrees well with the relationship between CFS and the Hf content as determined from the Ti L3-edge spectra (see Figure 8c). XANES spectra were also recorded at the Hf L3-edge. These reflect the transition of Hf 2p3/2 electrons to the unoccupied 5d states, and three representative spectra are shown in Figure 10a for Y2Ti1.8Hf0.2O7, Y2Ti1.0Hf1.0O7, and Y2Hf2O7. In contrast to the resolved peaks observed at the Ti L3,2- (Figure 8a) and Y L2-edges (Figure 7a), due to the crystal field splitting of the Ti(3d) and Y(4d) orbitals, respectively, the Hf L3-edge consists of a single broad white line. This is primarily caused by larger natural widths associated with higher energy atomic levels.73 However, the splitting of the Hf(5d) orbitals by the ligand field can still be seen clearly in the second-derivative spectra (inset of Figure 10a). The crystal field splitting (CFS) of the Hf(5d) states can be estimated by deconvoluting the seemingly singlepeak white line at the Hf L3-edge, as demonstrated by a recent study of Ln2Hf2O7 (Ln = lanthanoid).37 The Hf L3-edge spectra were fitted with an arctangent background and two Voigtshaped peaks with no constraints on the peak widths; an example of the fitted spectrum is shown in Figure 10b for Y2Ti1.8Hf0.2O7. The energy gap between peaks A and B (ΔE) is associated with the CFS of the Hf(5d) orbitals, and this is plotted in the inset of Figure 10b. As shown, ΔE decreases gradually with the increase of Hf content (x) in the pyrochlore region, mirroring the behavior at the Ti L3-edge (Figure 8c). In order to confirm the validity of our peak-fitting procedure (given the large uncertainty in fitting a broad single peak with two component peaks), changes in the peak width of the Hf L3edge white line were examined, as a wider peak should correspond to a larger crystal field splitting. Figure 10c shows the background-removed XANES spectra at the Hf L3-edge, clearly demonstrating the gradual decrease in the peak width with increasing Hf content (x). The fwhm was obtained using the computer software OriginPro and is plotted in the inset of Figure 10c as a function of x. Although the value of fwhm is much larger than ΔE and cannot be used as a direct measure of the CFS, its change with respect to x is essentially identical to that of ΔE (inset of Figure 10b). Since it is much easier and, more importantly, less subjective to determine the value of fwhm than to fit two peaks into a broad single peak, fwhm can be used as a less biased alternative to monitor changes in CFS. The decrease in the crystal field split d states with increasing Hf content (insets of Figure 10b and 10c) can be attributed to the increase in the local Hf−O bond distance and coordination number,37 similar to that observed at the Ti L3-edge (Figure 8c). The O K-edge XANES spectra of Y2Ti2−xHfxO7 are shown in Figure 11, along with those of the TiO2 (rutile), HfO2, and Y2O3 oxides. The cation coordination number is 6 in TiO274 and Y2O375 and 7 in HfO2.76 These spectra correspond to transitions from the O(1s) core level to the unoccupied O(2p) states in the conduction band. If these compounds are purely ionic, no absorption peak would be observed as the O(2p) orbitals would be fully occupied. The large intensity seen at the

Figure 10. (a) Normalized Hf L 3 -edge XANES spectra of Y2Ti1.8Hf0.2O7, Y2Ti1.0Hf1.0O7, and Y2Hf2O7 (spectra have been offset vertically to enhance visibility) with their second derivatives plotted in the inset. (b) Fitted XANES spectrum of Y2Ti1.8Hf0.2O7, with the inset showing the energy difference between the fitted peaks B and A as a function of Hf content (x). (c) Background-removed Hf L3-edge XANES spectra of Y2Ti2−xHfxO7, normalized to the white line instead of the edge jump to highlight the change in the peak width (fwhm) across the series (inset).

O K-edge clearly demonstrates the covalent nature of M−O bonding in these materials, reflecting unoccupied density of states of the various cation orbitals hybridized with the O(2p) states. Hence, the O K-edge line shape is also quite sensitive to the local structure.37 The spectra shown in Figure 11 can be divided into two regions. The first region (up to 9 eV above the absorption threshold) is attributed to the O(2p) orbitals hybridized with the metal d states (Ti 3d, Y 4d, and Hf 5d), and this region is quite sensitive to the local symmetry and ligand coordination.77 The second region above ∼540 eV is due to the larger overlap of the O(2p) orbitals with the metal sp states (Ti 26475

DOI: 10.1021/acs.jpcc.6b07076 J. Phys. Chem. C 2016, 120, 26465−26479

Article

The Journal of Physical Chemistry C

the CN of Ti increasing from 6 to 7.31 The features appearing at higher energies (i.e., in the region with mixed O 2p and metal sp orbitals) become broadened with the substitution of Hf for Ti. This suggests that the local disorder develops over a larger length scale with increasing Hf content, as spectral line broadening is associated with the loss of long-range order.37,78 The most significant conclusion from our XANES measurements is that the local structure evolves continuously throughout the whole composition range, in contrast to the diffraction results which show a clear phase transition at x ≈ 1.6. The crystal field splitting of the metal d orbitals (Ti 3d, Y 4d, and Hf 5d) can be derived from their respective L-edge XANES spectrum and used as a parameter to directly measure the local disorder in Y2Ti2−xHfxO7. Moreover, parts of the XANES spectra, e.g., the eg-like feature at the Ti L3-edge, also provide information on the medium-range order, suggesting that the length scale of disorder develops progressively with the increase of Hf content (and eventually reaches the threshold detectable by diffraction at x ≈ 1.6).



CONCLUSIONS Although both neutron and synchrotron X-ray powder diffraction results reveal a phase transition of the long-range average structure from pyrochlore to defect-fluorite at x ≈ 1.6 in Y2Ti2−xHfxO7, there are subtle differences between the two data sets reflecting their sensitivity to the anion and cation sublattices, respectively. The Y2Ti0.4Hf1.6O7 sample appears to have completely disordered cations but still retains some pyrochlore-like ordering in the oxygen sublattice. NPD results also demonstrate the gradual development of long-range anion disorder in the pyrochlore samples starting at the lowest Hf content (x = 0.2), primarily by filling the originally vacant 8a sites with oxygens from the nearest-neighbor anion shell at the 48f sites. In contrast, S-XRD results do not show much evidence for the presence of long-range cation antisite disorder until x ≈ 1.4, just before the phase transformation at x ≈ 1.6. The combination of both neutron and synchrotron X-ray diffraction results suggests that the anion disorder precedes that of the cations in Y2Ti2−xHfxO7. The use of high-resolution SXRD also allows us to exclude the presence of a two-phase region in Y2Ti2−xHfxO7 (within the composition step of Δx = 0.2), demonstrating that the observed disorder in pyrochlore cannot be attributed to the coexisting defect-fluorite phase as proposed in a recent NMR study.32 In contrast to the abrupt phase transition from ordered pyrochlore to disordered defect-fluorite as revealed by diffraction, XANES analysis clearly demonstrates that the short-range local structure evolves continuously throughout the entire composition range in Y2Ti2−xHfxO7. The XANES spectra not only are very sensitive to structural changes in the short range but also provide some indication to changes in the medium range. Our spectroscopic results indicate that the local disorder develops gradually with the substitution of Hf for Ti, and the length scale of the disorder increases with increasing Hf content (x) in Y2Ti2−xHfxO7. These results are consistent with those reported in recent studies of similar materials using neutron total scattering method.10,26 Although XANES is not as quantitative as pair distribution function (PDF) analysis of total scattering data, it can provide direct information on the variation of CN without the need for extensive data analysis. In addition, XANES analysis can be applied to study local disorder in thin films, which are used in many applications of pyrochlore materials.

Figure 11. Normalized O K-edge XANES spectra of the Y2Ti2−xHfxO7 series along with those of the TiO2 (rutile), HfO2, and Y2O3 oxides (spectra have been offset vertically to enhance visibility, and vertical lines are drawn to guide the eyes).

4sp, Y 5sp, and Hf 6sp) and is more sensitive to the longer range order in the sample.77 Note that the energy of the metal nd states increases with n (i.e., Ti 3d < Y 4d < Hf 5d) because inner shell electrons are bound more tightly to the nucleus. Pyrochlore-structured Y2Ti2O7 exhibits two strong peaks, A and C, at ∼530.6 and 533.4 eV, respectively, coinciding with those of Ti 3d-t2g and 3d-eg orbitals in rutile. This is not surprising as Ti in rutile74 is also in a slightly distorted octahedral site with similar Ti−O bond distances as those in Y2Ti2O7 (Table 1). In contrast, the O K-edge spectrum of Y2Ti2O7 does not resemble that of the Y2O3 binary oxide due to different local environments: the Y CN is 8 in Y2Ti2O7 with six long and two short Y−O bond distances (Table 1), whereas the two Y sites in Y2O3 are both coordinated to 6 oxygens.75 Thus, the O K-edge XANES spectrum of Y2Ti2O7 suggests that there is no cation antisite disorder locally in the pyrochlore end member, similar to the diffraction results over a much longer length scale (i.e., the local structure is similar to the average structure for Y2Ti2O7). The defect-fluorite end member of Y2Hf2O7 displays two main features B and D at ∼532.9 and 537.9 eV, which are associated with the Hf(5d) orbitals with added contributions from Y(4d) states. The peak positions of B and D in Y2Hf2O7 match those in HfO2 where the CN of Hf is 7,76 supporting the completely disordered defect-fluorite structure on the local scale as well. The substitution of Hf for Ti in the Y2Ti2−xHfxO7 series causes significant changes in the O K-edge XANES spectra, and these changes occur gradually throughout the entire series rather than abruptly at the diffraction derived pyrochlore/defect-fluorite phase boundary (x ≈ 1.6). The intensity of peak A decreases gradually with increasing Hf content (x), showing a positive Ti concentration dependence (i.e., peak A is attributed to Ti 3d-t2g states alone). The position of peak A does not change with composition, agreeing well with that observed at the Ti L3-edge (Figure 8a). The second peak position decreases gradually from 533.4 (C) to 533.0 eV (B) with the Ti content decreasing from 2.0 to 0.2, consistent with 26476

DOI: 10.1021/acs.jpcc.6b07076 J. Phys. Chem. C 2016, 120, 26465−26479

Article

The Journal of Physical Chemistry C

(4) Sickafus, K. E.; Minervini, L.; Grimes, R. W.; Valdez, J. A.; Ishimaru, M.; Li, F.; McClellan, K. J.; Hartmann, T. Radiation Tolerance of Complex Oxides. Science 2000, 289, 748−751. (5) Ewing, R. C.; Weber, W. J.; Lian, J. Nuclear Waste Disposal Pyrochlore (A2B2O7): Nuclear Waste Form for the Immobilization of Plutonium and ’Minor’ Actinides. J. Appl. Phys. 2004, 95, 5949−5971. (6) Weber, W. J.; Ewing, R. C. Plutonium Immobilization and Radiation Effects. Science 2000, 289, 2051−2052. (7) Lutique, S.; Staicu, D.; Konings, R. J. M.; Rondinella, V. V.; Somers, J.; Wiss, T. Zirconate Pyrochlore as a Transmutation Target: Thermal Behaviour and Radiation Resistance against Fission Fragment Impact. J. Nucl. Mater. 2003, 319, 59−64. (8) Raison, P. E.; Haire, R. G. Zirconia-Based Materials for Transmutation of Americium and Curium: Cubic Stabilized Zirconia and Zirconium Oxide Pyrochlores. Prog. Nucl. Energy 2001, 38, 251− 254. (9) Yamamura, H.; Nishino, H.; Kakinuma, K.; Nomura, K. Electrical Conductivity Anomaly around Fluorite−Pyrochlore Phase Boundary. Solid State Ionics 2003, 158, 359−365. (10) Norberg, S. T.; Hull, S.; Eriksson, S. G.; Ahmed, I.; Kinyanjui, F.; Biendicho, J. J. Pyrochlore to Fluorite Transition: The Y2(Ti1−xZrx)2O7 (0.0 ≤ x ≤ 1.0) System. Chem. Mater. 2012, 24, 4294−4300. (11) Radhakrishnan, A. N.; Rao, P. P.; Linsa, K. S. M.; Deepa, M.; Koshy, P. Influence of Disorder-to-Order Transition on Lattice Thermal Expansion and Oxide Ion Conductivity in (CaxGd1‑x)2(Zr1‑xMx)2O7 Pyrochlore Solid Solutions. Dalton Trans. 2011, 40, 3839−3848. (12) Perriot, R.; Uberuaga, B. P. Structural vs. Intrinsic Carriers: Contrasting Effects of Cation Chemistry and Disorder on Ionic Conductivity in Pyrochlores. J. Mater. Chem. A 2015, 3, 11554−11565. (13) Uberuaga, B. P. Complex Oxides: Intricate Disorder. Nat. Mater. 2016, 15, 496−497. (14) Michel, D.; Perez y Jorba, M.; Collongues, R. Etude de la Transformation Ordre-Desordre de la Structure Fluorite a la Structure Pyrochlore pour des Phases (1−x) ZrO2 - x Ln2O3. Mater. Res. Bull. 1974, 9, 1457−1468. (15) Rushton, M. J. D.; Grimes, R. W.; Stanek, C. R.; Owens, S. Predicted Pyrochlore to Fluorite Disorder Temperature for A2Zr2O7 Compositions. J. Mater. Res. 2004, 19, 1603−1604. (16) Zhang, F. X.; Lian, J.; Becker, U.; Ewing, R. C.; Hu, J. Z.; Saxena, S. K. High-Pressure Structural Changes in the Gd2Zr2O7 Pyrochlore. Phys. Rev. B: Condens. Matter Mater. Phys. 2007, 76, 214104. (17) Xiao, H. Y.; Zhang, F. X.; Gao, F.; Lang, M.; Ewing, R. C.; Weber, W. J. Zirconate Pyrochlores under High Pressure. Phys. Chem. Chem. Phys. 2010, 12, 12472−12477. (18) Whittle, K. R.; Cranswick, L. M. D.; Redfern, S. A. T.; Swainson, I. P.; Lumpkin, G. R. Lanthanum Pyrochlores and the Effect of Yttrium Addition in the Systems La2‑xYxZr2O7 and La2‑xYxHf2O7. J. Solid State Chem. 2009, 182, 442−450. (19) Clements, R.; Hester, J. R.; Kennedy, B. J.; Ling, C. D.; Stampfl, A. P. J. The Fluorite-Pyrochlore Transformation of Ho2‑yNdyZr2O7. J. Solid State Chem. 2011, 184, 2108−2113. (20) Liu, Y.; Withers, R. L.; Noren, L. The Pyrochlore to ’Defect Fluorite’ Transition in the Y2(ZryTi1‑y)2O7 System and its Underlying Crystal Chemistry. J. Solid State Chem. 2004, 177, 4404−4412. (21) Lian, J.; Zu, X. T.; Kutty, K. V. G.; Chen, J.; Wang, L. M.; Ewing, R. C. Ion-Irradiation-Induced Amorphization of La2Zr2O7 Pyrochlore. Phys. Rev. B: Condens. Matter Mater. Phys. 2002, 66, 054108. (22) Lumpkin, G. R.; Smith, K. L.; Blackford, M. G.; Whittle, K. R.; Harvey, E. J.; Redfern, S. A. T.; Zaluzec, N. J. Ion Irradiation of Ternary Pyrochlore Oxides. Chem. Mater. 2009, 21, 2746−2754. (23) Glerup, M.; Nielsen, O. F.; Poulsen, F. W. The Structural Transformation from the Pyrochlore Structure, A2B2O7, to the Fluorite Structure, AO2, Studied by Raman Spectroscopy and Defect Chemistry Modeling. J. Solid State Chem. 2001, 160, 25−32. (24) Blanchard, P. E. R.; Clements, R.; Kennedy, B. J.; Ling, C. D.; Reynolds, E.; Avdeev, M.; Stampfl, A. P. J.; Zhang, Z.; Jang, L.-Y. Does

This work highlights the challenges faced when studying partially disordered systems. In order to provide a full description of such systems it is imperative to choose appropriate experimental techniques according to the length scale of the disorder. An obvious, but important, conclusion from this study is that different experimental studies may reach apparently different conclusions, depending on whether the techniques employed are more sensitive to the cation or anion sublattice and the probing length of the technique.



ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.jpcc.6b07076. List of 52 ways of distributing three cations of Y, Ti, and Hf in Y2Ti1.0Hf1.0O7 over two cation sites A and B and the resulting goodness-of-fit (χ2) associated with each cation arrangement; appropriately scaled unit cell parameter of Y2Ti2−xHfxO7 as a function of composition (determined by Rietveld refinements against the NPD data); plots of the main pyrochlore (222)p and (488)p peaks for the Y2Ti1.6Hf0.4O7 and Y2Ti0.7Hf1.3O7 samples; formula of the X-ray structure factors for the pyrochlore (111)p and (222)p peaks, F(111) and F(222) (PDF)



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Phone: +61 2 9717 9091. Present Addresses

§ M.d.l.R.: Environment Protection Authority, 250 Victoria Square, Adelaide, SA 5000, Australia. ∥ P.E.R.B.: Canadian Light Source, 44 Innovation Boulevard, Saskatoon, SK S7N 2 V3, Canada.

Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This work was, in part, performed on the powder diffraction (PD) and soft X-ray spectroscopy (SXR) beamlines at the Australian Synchrotron. The assistance of Dr. Justin Kimpton on PD is much appreciated. The work carried out at the NSRRC was supported by the International Synchrotron Access Program (ISAP), funded by the Australian Synchrotron. We thank Drs. Ling-Yun Jang, Jeng-Lung Chen, Jyh-Fu Lee, and Chih-Wen Pao at NSRRC for their assistance in Taiwan. B.J.K. also acknowledges the support of the Australian Research Council.



REFERENCES

(1) Subramanian, M. A.; Aravamudan, G.; Subba Rao, G. V. Oxide Pyrochlores  A Review. Prog. Solid State Chem. 1983, 15, 55−143. (2) Wuensch, B. J.; Eberman, K. W.; Heremans, C.; Ku, E. M.; Onnerud, P.; Yeo, E. M. E.; Haile, S. M.; Stalick, J. K.; Jorgensen, J. D. Connection between Oxygen-Ion Conductivity of Pyrochlore FuelCell Materials and Structural Change with Composition and Temperature. Solid State Ionics 2000, 129, 111−133. (3) Heremans, C.; Wuensch, B. J.; Stalick, J. K.; Prince, E. Fast-Ion Conducting Y2(ZryTi1‑y)2O7 Pyrochlores: Neutron Rietveld Analysis of Disorder-Induced by Zr Substitution. J. Solid State Chem. 1995, 117, 108−121. 26477

DOI: 10.1021/acs.jpcc.6b07076 J. Phys. Chem. C 2016, 120, 26465−26479

Article

The Journal of Physical Chemistry C Local Disorder Occur in the Pyrochlore Zirconates? Inorg. Chem. 2012, 51, 13237−13244. (25) Zhang, Z. M.; Middleburgh, S. C.; de los Reyes, M.; Lumpkin, G. R.; Kennedy, B. J.; Blanchard, P. E. R.; Reynolds, E.; Jang, L. Y. Gradual Structural Evolution from Pyrochlore to Defect-Fluorite in Y2Sn2‑xZrxO7: Average vs Local Structure. J. Phys. Chem. C 2013, 117, 26740−26749. (26) Shamblin, J.; Feygenson, M.; Neuefeind, J.; Tracy, C. L.; Zhang, F.; Finkeldei, S.; Bosbach, D.; Zhou, H.; Ewing, R. C.; Lang, M. Probing Disorder in Isometric Pyrochlore and Related Complex Oxides. Nat. Mater. 2016, 15, 507−511. (27) Keen, D. A.; Goodwin, A. L. The Crystallography of Correlated Disorder. Nature 2015, 521, 303−309. (28) Panero, W. R.; Stixrude, L.; Ewing, R. C. First-Principles Calculation of Defect-Formation Energies in the Y2(Ti,Sn,Zr)2O7 Pyrochlore. Phys. Rev. B: Condens. Matter Mater. Phys. 2004, 70, 054110. (29) Nachimuthu, P.; Thevuthasan, S.; Engelhard, M. H.; Weber, W. J.; Shuh, D. K.; Hamdan, N. M.; Mun, B. S.; Adams, E. M.; McCready, D. E.; Shutthanandan, V.; et al. Probing Cation Antisite Disorder in Gd2Ti2O7 Pyrochlore by Site-Specific Near-Edge X-ray-Absorption Fine Structure and X-ray Photoelectron Spectroscopy. Phys. Rev. B: Condens. Matter Mater. Phys. 2004, 70, 100101. (30) Nachimuthu, P.; Thevuthasan, S.; Shutthanandan, V.; Adams, E. M.; Weber, W. J.; Begg, B. D.; Shuh, D. K.; Lindle, D. W.; Gullikson, E. M.; Perera, R. C. C. Near-Edge X-ray Absorption Fine-Structure Study of Ion-Beam-Induced Phase Tr ansformation in Gd2(Ti1‑yZry)2O7. J. Appl. Phys. 2005, 97, 033518. (31) Nachimuthu, P.; Thevuthasan, S.; Adams, E. M.; Weber, W. J.; Begg, B. D.; Mun, B. S.; Shuh, D. K.; Lindle, D. W.; Gullikson, E. M.; Perera, R. C. C. Near-Edge X-ray Absorption Fine Structure Study of Disordering in Gd2(Ti1‑yZry)2O7 Pyrochlores. J. Phys. Chem. B 2005, 109, 1337−1339. (32) Ashbrook, S. E.; Mitchell, M. R.; Sneddon, S.; Moran, R. F.; de los Reyes, M.; Lumpkin, G. R.; Whittle, K. R. New Insights into Phase Distribution, Phase Composition and Disorder in Y2(Zr,Sn)2O7 Ceramics from NMR Spectroscopy. Phys. Chem. Chem. Phys. 2015, 17, 9049−9059. (33) de los Reyes, M.; Whittle, K. R.; Zhang, Z.; Ashbrook, S. E.; Mitchell, M. R.; Jang, L.-Y.; Lumpkin, G. R. The Pyrochlore to Defect Fluorite Phase Transition in Y2Sn2‑xZrxO7. RSC Adv. 2013, 3, 5090− 5099. (34) Sears, V. F. Neutron Scattering Lengths and Cross Sections. Neutron News 1992, 3, 26−37. (35) Aluri, E. R.; Grosvenor, A. P. An X-ray Absorption Spectroscopic Study of the Effect of Bond Covalency on the Electronic Structure of Gd2Ti2‑xSnxO7. Phys. Chem. Chem. Phys. 2013, 15, 10477−10486. (36) Walker, J. D. S.; Hayes, J. R.; Gaultois, M. W.; Aluri, E. R.; Grosvenor, A. P. A Case for Oxygen Deficiency in Gd2Ti2−xZrxO7 Pyrochlore-Type Oxides. J. Alloys Compd. 2013, 565, 44−49. (37) Blanchard, P. E. R.; Liu, S.; Kennedy, B. J.; Ling, C. D.; Avdeev, M.; Aitken, J. B.; Cowie, B. C. C.; Tadich, A. Investigating the Local Structure of Lanthanoid Hafnates Ln2Hf2O7 via Diffraction and Spectroscopy. J. Phys. Chem. C 2013, 117, 2266−2273. (38) Blanchard, P. E. R.; Liu, S.; Kennedy, B. J.; Ling, C. D.; Zhang, Z.; Avdeev, M.; Cowie, B. C. C.; Thomsen, L.; Jang, L.-Y. Investigating the Order-Disorder Phase Transition in Nd2‑xYxZr2O7 via Diffraction and Spectroscopy. Dalton Trans. 2013, 42, 14875−14882. (39) Reynolds, E.; Blanchard, P. E. R.; Kennedy, B. J.; Ling, C. D.; Liu, S.; Avdeev, M.; Zhang, Z.; Cuello, G. J.; Tadich, A.; Jang, L.-Y. Anion Disorder in Lanthanoid Zirconates Gd2‑xTbxZr2O7. Inorg. Chem. 2013, 52, 8409−8415. (40) Wallwork, K. S.; Kennedy, B. J.; Wang, D. The High Resolution Powder Diffraction Beamline for the Australian Synchrotron. AIP Conf. Proc. 2006, 879, 879−882. (41) Liss, K.-D.; Hunter, B.; Hagen, M.; Noakes, T.; Kennedy, S. Echidnathe New High-Resolution Powder Diffractometer Being Built at OPAL. Phys. B 2006, 385−386, 1010−1012.

(42) Hunter, B. A.; Howard, C. J. LHPM-A Computer Program for Rietveld Analysis of X-ray and Neutron Powder Diffraction Patterns; Lucas Heights Research Laboratories: Sydney, Australia, 1998. (43) Dann, T.-E.; Chung, S.-C.; Huang, L.-J.; Juang, J.-M.; Chen, C.I.; Tsang, K.-L. A High-Performance Double-Crystal Monochromator Soft X-ray Beamline. J. Synchrotron Radiat. 1998, 5, 664−666. (44) Tsang, K. L.; Lee, C. H.; Jean, Y. C.; Dann, T. E.; Chen, J. R.; D’Amico, K. L.; Oversluizen, T. Wiggler X-ray Beamlines at Synchrotron Radiation Research Center. Rev. Sci. Instrum. 1995, 66, 1812−1814. (45) Cowie, B. C. C.; Tadich, A.; Thomsen, L. The Current Performance of the Wide Range (90−2500 eV) Soft X-ray Beamline at the Australian Synchrotron. AIP Conf. Proc. 2010, 1234, 307−310. (46) Ravel, B.; Newville, M. ATHENA, ARTEMIS, HEPHAESTUS: Data Analysis for X-ray Absorption Spectroscopy using IFEFFIT. J. Synchrotron Radiat. 2005, 12, 537−541. (47) Fairley, N. CasaXPS, 2.3.17dev6.4k; Casa Software Ltd.: Teignmouth, Devon, UK, 2014. (48) Lau, G. C.; Freitas, R. S.; Ueland, B. G.; Dahlberg, M. L.; Huang, Q.; Zandbergen, H. W.; Schiffer, P.; Cava, R. J. Structural Disorder and Properties of the Stuffed Pyrochlore Ho2TiO5. Phys. Rev. B: Condens. Matter Mater. Phys. 2007, 76, 054430. (49) Lau, G. C.; McQueen, T. M.; Huang, Q.; Zandbergen, H. W.; Cava, R. J. Long- and Short-Range Order in Stuffed Titanate Pyrochlores. J. Solid State Chem. 2008, 181, 45−50. (50) Wuensch, B. J.; Eberman, K. W. Order-Disorder Phenomena in A2B2O7 Pyrochlore Oxides. JOM 2000, 52, 19−21. (51) Popov, V. V.; Menushenkov, A. P.; Yaroslavtsev, A. A.; Zubavichus, Y. V.; Gaynanov, B. R.; Yastrebtsev, A. A.; Leshchev, D. S.; Chernikov, R. V. Fluorite-Pyrochlore Phase Transition in Nanostructured Ln2Hf2O7 (Ln = La-Lu). J. Alloys Compd. 2016, 689, 669− 679. (52) Wilde, P. J.; Catlow, C. R. A. Defects and Diffusion in Pyrochlore Structured Oxides. Solid State Ionics 1998, 112, 173−183. (53) Pirzada, M.; Grimes, R. W.; Minervini, L.; Maguire, J. F.; Sickafus, K. E. Oxygen Migration in A2B2O7 Pyrochlores. Solid State Ionics 2001, 140, 201−208. (54) Kong, L.; Zhang, Z.; de los Reyes, M.; Karatchevtseva, I.; Lumpkin, G. R.; Triani, G.; Aughterson, R. D. Soft Chemical Synthesis and Structural Characterization of Y2HfxTi2−xO7. Ceram. Int. 2015, 41, 5309−5317. (55) Shannon, R. D. Revised Effective Ionic-Radii and Systematic Studies of Interatomic Distances in Halides and Chalcogenides. Acta Crystallogr., Sect. A: Cryst. Phys., Diffr., Theor. Gen. Crystallogr. 1976, 32, 751−767. (56) Ogasawara, K.; Iwata, T.; Koyama, Y.; Ishii, T.; Tanaka, I.; Adachi, H. Relativistic Cluster Calculation of Ligand-Field Multiplet Effects on Cation L2,3 X-Ray-Absorption Edges of SrTiO3, NiO, and CaF2. Phys. Rev. B: Condens. Matter Mater. Phys. 2001, 64, 115413. (57) Ikeno, H.; Krause, M.; Höche, T.; Patzig, C.; Hu, Y.; Gawronski, A.; Tanaka, I.; Rüssel, C. Variation of Zr-L 2,3 XANES in Tetravalent Zirconium Oxides. J. Phys.: Condens. Matter 2013, 25, 165505. (58) de Groot, F. M. F.; Fuggle, J. C.; Thole, B. T.; Sawatzky, G. A. L2,3 X-Ray-Absorption Edges of d0 Compounds: K+, Ca2+, Sc3+, and Ti4+ in Oh (Octahedral) Symmetry. Phys. Rev. B: Condens. Matter Mater. Phys. 1990, 41, 928−937. (59) Crocombette, J. P.; Jollet, F. Ti 2p X-Ray Absorption in Titanium Dioxides (TiO2): the Influence of the Cation Site Environment. J. Phys.: Condens. Matter 1994, 6, 10811−10821. (60) Krüger, P. Multichannel Multiple Scattering Calculation of L2,3Edge Spectra of TiO2 and SrTiO3: Importance of Multiplet Coupling and Band Structure. Phys. Rev. B: Condens. Matter Mater. Phys. 2010, 81, 125121. (61) Aluri, E. R.; Grosvenor, A. P. An Investigation of PyrochloreType Oxides (Yb2Ti2−xFexO7‑δ) by XANES. J. Phys. Chem. Solids 2013, 74, 830−836. (62) de groot, F. M. F.; Figueiredo, M. O.; Basto, M. J.; Abbate, M.; Petersen, H.; Fuggle, J. C. 2P X-ray Absorption of Titanium in Minerals. Phys. Chem. Miner. 1992, 19, 140−147. 26478

DOI: 10.1021/acs.jpcc.6b07076 J. Phys. Chem. C 2016, 120, 26465−26479

Article

The Journal of Physical Chemistry C (63) Mastelaro, V. R.; Neves, P. P.; de Lazaro, S. R.; Longo, E.; Michalowicz, A.; Eiras, J. A. Electronic Structure of Pb1−xLaxTiO3 Ferroelectric Materials from Ti 2p and O 1s Soft X-ray Absorption Spectroscopy. J. Appl. Phys. 2006, 99, 044104. (64) Kucheyev, S. O.; van Buuren, T.; Baumann, T. F.; Satcher, J. H.; Willey, T. M.; Meulenberg, R. W.; Felter, T. E.; Poco, J. F.; Gammon, S. A.; Terminello, L. J. Electronic Structure of Titania Aerogels from Soft X-ray Absorption Spectroscopy. Phys. Rev. B: Condens. Matter Mater. Phys. 2004, 69, 245102. (65) Kronawitter, C. X.; Kapilashrami, M.; Bakke, J. R.; Bent, S. F.; Chuang, C.-H.; Pong, W.-F.; Guo, J.; Vayssieres, L.; Mao, S. S. TiO2SnO2:F Interfacial Electronic Structure Investigated by Soft X-ray Absorption Spectroscopy. Phys. Rev. B: Condens. Matter Mater. Phys. 2012, 85, 125109. (66) Farges, F.; Brown, G. E., Jr.; Rehr, J. J. Coordination Chemistry of Ti(IV) in Silicate Glasses and Melts: I. XAFS Study of Titanium Coordination in Oxide Model Compounds. Geochim. Cosmochim. Acta 1996, 60, 3023−3038. (67) Farges, F.; Brown, G. E.; Rehr, J. J. Ti K-Edge XANES Studies of Ti Coordination and Disorder in Oxide Compounds: Comparison Between Theory and Experiment. Phys. Rev. B: Condens. Matter Mater. Phys. 1997, 56, 1809−1819. (68) Sanjuán, M. L.; Guglieri, C.; Díaz-Moreno, S.; Aquilanti, G.; Fuentes, A. F.; Olivi, L.; Chaboy, J. Raman and X-ray Absorption Spectroscopy Study of the Phase Evolution Induced by Mechanical Milling and Thermal Treatments in R2Ti2O7 Pyrochlores. Phys. Rev. B: Condens. Matter Mater. Phys. 2011, 84, 104207. (69) Yamamoto, T. Assignment of Pre-edge Peaks in K-edge X-ray Absorption Spectra of 3d Transition Metal Compounds: Electric Dipole or Quadrupole? X-Ray Spectrom. 2008, 37, 572−584. (70) de Groot, F. M. F. Novel Techniques and Approaches to Unravel the Nature of X-Ray Absorption Spectra. AIP Conf. Proc. 2006, 882, 37−43. (71) Cabaret, D.; Bordage, A.; Juhin, A.; Arfaoui, M.; Gaudry, E. First-Principles Calculations of X-ray Absorption Spectra at the K-edge of 3d Transition Metals: an Electronic Structure Analysis of the Preedge. Phys. Chem. Chem. Phys. 2010, 12, 5619−5633. (72) Cabaret, D.; Joly, Y.; Renevier, H.; Natoli, C. R. Pre-edge Structure Analysis of Ti K-edge Polarized X-ray Absorption Spectra in TiO2 by Full-Potential XANES Calculations. J. Synchrotron Radiat. 1999, 6, 258−260. (73) Krause, M. O.; Oliver, J. H. Natural widths of atomic K and L levels, Kα X-ray lines and several KLL Auger lines. J. Phys. Chem. Ref. Data 1979, 8, 329−338. (74) Sabine, T. M.; Howard, C. J. Determination of the Oxygen x Parameter in Rutile by Neutron Powder Methods. Acta Crystallogr., Sect. B: Struct. Crystallogr. Cryst. Chem. 1982, 38, 701−702. (75) Paton, M. G.; Maslen, E. N. A Refinement of the Crystal Structure of Yttria. Acta Crystallogr. 1965, 19, 307−310. (76) Ruh, R.; Corfield, P. W. R. Crystal Structure of Monoclinic Hafnia and Comparison with Monoclinic Zirconia. J. Am. Ceram. Soc. 1970, 53, 126−129. (77) Soriano, L.; Abbate, M.; Fernández, A.; González-Elipe, A.; Sanz, J. Chemical Analysis of Ternary Ti Oxides Using Soft X-ray Absorption Spectroscopy. Surf. Interface Anal. 1997, 25, 804−808. (78) Hill, D. H.; Bartynski, R. A.; Nguyen, N. V.; Davydov, A. C.; Chandler-Horowitz, D.; Frank, M. M. The Relationship Between Local Order, Long Range Order, and Sub-Band-Gap Defects in Hafnium Oxide and Hafnium Silicate Films. J. Appl. Phys. 2008, 103, 093712.

26479

DOI: 10.1021/acs.jpcc.6b07076 J. Phys. Chem. C 2016, 120, 26465−26479