You are on page 1of 12

17548 J. Phys. Chem.

C 2009, 113, 17548–17559

An Investigation of Distortions of the Dion-Jacobson Phase RbSr2Nb3O10 and Its


Acid-Exchanged Form with 93Nb Solid State NMR and DFT Calculations

Xuefeng Wang, Jhashanath Adhikari, and Luis J. Smith*


Carlson School of Chemistry and Biochemistry, Clark UniVersity, 950 Main Street,
Worcester, Massachusetts 01610
ReceiVed: March 15, 2009; ReVised Manuscript ReceiVed: July 19, 2009

The different niobium sites in a Dion-Jacobson triple-layered perovskite, RbSr2Nb3O10, and its acid-exchanged
version, HSr2Nb3O10, were investigated by using solid state NMR and DFT methods. 93Nb electric field gradient
(EFG) and chemical shift anisotropy (CSA) tensor values were extracted and site assignments made by using
DFS, VOCS, MQMAS, and {1H}-93Nb CP NMR techniques. The exterior niobium site exhibited a quadrupolar
coupling (CQ) of 45 MHz and a CSA span of 820 ppm while the interior site exhibited a reverse trend with
a CQ of 93 MHz and a CSA span of 530 ppm. Both EFG and CSA tensors for the sites exhibited near axial
symmetry although the CSA tensors had a greater deviation. The symmetry of the tensors is in conflict with
the previously proposed structure for RbSr2Nb3O10 and implies a lower symmetry space group and a possible
tilting of the octahedra. Acid exchange altered the EFG tensors for both sites reducing their CQ values to 39
and 86 MHz, respectively, while not substantially changing the tensor symmetry. The effect on the CSA
tensor occurred only for the surface site with the span reduced to 560 ppm. The effect of the tilting of the
octahedra and alteration of bond angles and bond lengths on the EFG tensor was investigated through periodic
DFT calculations. Interior and exterior sites of the rubidium form exhibited different tilt along the c-axis of
14° and -3°, respectively, based on the 93Nb EFG values. Calculations on the acid-exchanged composition
implied that structural alterations must occur with both sites. The exterior octahedra experience changes in
both bond length and bond angles, while only a subtle elongation of the interior site is found.

Introduction
Lamellar oxides have a large variety of compositions and
possible structures that leads to diversity in the properties of
these materials. Many systems of interest have niobium as a
six-coordinate metal in the oxide lattice. In several of these
layered niobates, photocatalysis,1,2 superconductivity,3,4 and
proton conduction5 have been observed. Exfoliation of these
layered compounds further expands the potential uses. Several
exfoliated layered niobates have been explored as photocatalysts,6,7
solid acid catalysts,8-11 and nanosheet building blocks for more
complex oxide thin films.12,13 In many cases, the details of
structural perturbations or variations that occur with differences
in composition or postsynthetic treatment are not well-known.
X-ray diffraction methods lose resolution when the particle size Figure 1. Crystal structure of RbSr2Nb3O10, at left (niobia octahedra,
or domain size decreases after exfoliation or exfoliation followed green; strontium, blue; rubidium, pink), and subsection, at right, of a
by restacking.14-16 However, through element-selective spec- single triple-perovskite layer with the interior, Nb1, and exterior, Nb2,
troscopic methods such as NMR, the distortions that are present sites labeled.
in the local environment of the metal sites can be determined
even when sufficient long-range order of these distortions is series of these layered perovskites follow the general formula
not present. As acid exchange is the first step in the exfoliation of A[A′k-1BkO3k+1], where k equals the number of octahedra
process used to produce nanosheets from these layered niobates, stacked within a slab. The flexibility of the properties of these
understanding what structural changes occur in this step is materials comes with the wide range of elements that can serve
important for understanding any properties derived from these as the A and B cations. As with three-dimensional perovskites,
nanosheets or the novel thin films built from them. often the A and B cations are not of a sufficient size to fill the
One family of lamellar oxides is based on the perovskites octahedral holes present in the oxide lattice. With smaller B
structure. Corner sharing BO6 octahedra are interleaved with A cations, the result is often a displacement from the center of
cations and form a two-dimensional pervoskite slab (Figure 1). the octahedron. Small interstitial A cations can distort the oxide
The negatively charged layers are interleaved with a second lattice resulting in a pattern of tilting for the BO6 octahedra.
set of cations that are ion-exchangeable.17-20 The Dion-Jacobson The Dion-Jacobson layered perovskites can have these distor-
tions. The exterior octahedra at the interface between layers in
* To whom correspondence should be addressed. E-mail: lusmith@ KCa2Nb3O10 and CsCa2Nb3O10 have displaced niobium atoms
clarku.edu. resulting in an effectively five-coordinated site and are canted
10.1021/jp902318v CCC: $40.75  2009 American Chemical Society
Published on Web 09/11/2009
Distortions of the Dion-Jacobson Phase RbSr2Nb3O10 J. Phys. Chem. C, Vol. 113, No. 40, 2009 17549

with respect to each other and the interior octahedra.21,22 The transition, -1/2 T 1/2, as the other transitions are too broad for
interior octahedra do not have displaced B cations but are tilted convenient observation. Magic angle spinning (MAS) can reduce
with respect to each other. The small size of the Nb5+ and Ca2+ the impact of the anisotropies on the observed line shape by
cations leads to these distortions. Previous X-ray powder completely averaging the effect of the CSA and partially
diffraction studies of RbSr2Nb3O10 suggest that the larger averaging the effect of the EFG if the spinning speed exceeds
strontium cation does not perturb the oxide lattice leading to a the magnitude of the CSA. As a result, line shapes observed
highly symmetric structure.23 While no tilting is present, the under static conditions contain both CSA and EFG tensor
displacement of the niobium atoms in the exterior octahedra is information while MAS based experiments can be made to yield
still observed. Calculation of the polyhedral volume ratio for only EFG tensor values and the isotropic chemical shift.38
Sr2+ and Nb5+ and the tilt angle derived from it shows that a Assisting with the extraction of the tensor values is the different
tilt would be expected in a three-dimensional perovskite with field dependencies of the interactions. The CSA interaction is
these cations.24 Recent Raman spectroscopy studies have proportional to the magnetic field strength, and the EFG
suggested that the layered structure does indeed have a tilt interaction is inversely proportional, when second-order interac-
present.25 However, the interpretation of Raman spectra is tions are present. Thus a combination of multiple magnetic fields
difficult as the technique observes the whole structure and does and static and MAS experiments can extract the values of the
not easily permit the interpretation of the effect of distortions two tensors and their relative orientations. With 93Nb nuclei,
on the individual octahedra.26,27 both the EFG and CSA can have large magnitudes that require
NMR is well suited for the examination of the local this multiple experiment approach. The CSA and EFG are
environment of the niobium atom at the center of the octahedron. sufficiently large enough that normal MAS methods do not
With this specific focus, the changes in coordination and bond suffice and two-dimensional rotor-synchronized multiple quan-
length that come with the distortions possible in perovskites tum MAS (MQMAS) experiments are required.39,40 Rotor-
can be quantified. The NMR parameters determined can be synchronization along with an alternate shearing scheme for
directly linked to the crystal structure through the use of processing the two-dimensional data (vide infra) make it possible
quantum mechanical calculations. Furthermore, density func- to extract the EFG values without the influence of CSA and
tional theory (DFT) calculations that incorporate periodic are crucial to this study.
boundary conditions enable a more thorough determination of In this study, the structure of the Dion-Jacobson layered
interactions that have long-range contributions such as the perovskites, RbSr2Nb3O10, is examined via 93Nb NMR to
electric field gradient (EFG) experienced at the metal site.28-33 examine the extent of structural distortion present. Furthermore,
The EFG is a particularly relevant descriptor of metal sites as the effect of acid exchange, a precursor to exfoliation, is
many inorganic materials incorporate quadrupolar nuclei. These measured. The EFG and CSA tensor information for the niobium
are nuclei with nuclear spins greater than 1/2 and constitute about sites is determined by using multiple magnetic field, wide-line
75% of the NMR observable nuclei. As a result, methods applied experiments and MQMAS experiments at moderate field
to these nuclei can be useful in numerous oxide systems. strengths. The approach utilized was able to extract EFG and
Additional interactions with the field created by the local CSA information despite the substantially large values for both.
bonding electrons are manifested in the chemical shift anisotropy The NMR tensor information is related to the possible structures
(CSA), which is dependent on the local symmetry of the metal for the compounds based on symmetry arguments and DFT
site. CSA is potentially present for all nuclei, quadrupolar or calculations. Both possible tilt angles and structural rearrange-
not, but can vary in magnitude depending on the element. ments are predicted based on the NMR and computational
Both EFG and CSA interactions lead to anisotropies that results.
affect the shape of the powder distribution observed in NMR
experiments. The EFG tensor can be described by three Experimental Section
components VXX, VYY, and VZZ in its own principal axis system 1. Synthesis. The RbSr2Nb3O10 sample was prepared via
(PAS). The terms are typically defined such that |VXX| e |VYY| microwave heating. A ground mixture consisting of RbNO3,
e |VZZ| and VXX + VYY + VZZ ) 0. Two independent terms result: SrCO3, and Nb2O5 with a molar ratio of 1.25:2:3 for Rb:Sr:Nb
the quadrupolar coupling, CQ ) eQVZZ/h, where Q is the was used. The sample was heated with use of silicon carbide
quadrupolar moment, and the asymmetry parameter, η ) (VXX susceptors in an insulated thermal box (EPL Ceramic Materials)
- VYY)/VZZ with values ranging from 0 to 1.34 The CSA tensor in conjunction with a 1200 W domestic microwave oven at 90%
is described by three components in its respective PAS: δ11, power for 5-10 min followed by 50% power for 10-15 min.
δ22, and δ33. The tensor components are combined together to The product was then washed with distilled water and dried.
yield three descriptive parameters: the isotropic chemical shift, To produce the acid forms, the parent compound was treated
δISO, the span, Ω, and the skew, κ. The span of the CSA is with 6 M HNO3 for 3 days at 80 °C. Each day the acid solution
defined by Ω ) δ11 - δ33, and represents the magnitude of the was replaced. After acid exchange, the product was filtered,
anisotropy. The skew is defined by κ ) 3(δISO - δ22)/Ω, and washed with distilled water, and dried at 130 °C.
represents the symmetry of the tensor with values ranging from 2. X-ray Diffraction. All powder diffraction data were
-1 to 1.35 As each tensor is defined within its own PAS, a set collected with a Bruker AXS D8 Focus diffractometer, using
of Euler angles, R, β, and γ, which define their relative Cu KR radiation. Powder diffraction was used to determine the
orientation, is needed to relate the direction of one tensor in phase identity of the samples after synthesis or postsynthetic
terms of the other. processing. Whole pattern fitting was conducted by using the
The NMR active isotope for niobium, 93Nb, has a NMR spin LeBail method in conjunction with the TOPAS software
of 9/2 and has 100% natural abundance making it extremely suite.41,42
useful for these studies. The size of the quadrupole moment of 3. Solid State NMR Spectroscopy. The 87Rb MAS spectrum
93
Nb is large enough that the EFG interaction in most materials was collected at a field strength of 9.4 T, using a resonance
is best described by a second-order quadrupolar interaction.36,37 frequency of 130.81 MHz. Data were collected with use of a
As a result, most NMR investigations focus on the central 2.5 mm double resonance MAS probe with a spinning rate of
17550 J. Phys. Chem. C, Vol. 113, No. 40, 2009 Wang et al.

20 kHz. A Hahn echo pulse sequence using π/2 and π pulse quadrupolar shift information divided among isotropic sidebands
lengths of 2 and 4 µs with a recycle delay of 0.8 s was used. if the original anisotropies were too large to be averaged by
All 87Rb chemical shifts are reported relative to a 0.5 M aqueous MAS. In the new scheme, the following relations (for I ) 9/2)
RbNO3 solution (0 ppm). describe the frequencies in each dimension:
All 93Nb NMR data were collected at magnetic field strengths
of 9.4 and 14.1 T on Unity INOVA spectrometers with resonance 648 0
frequencies of 97.74 and 146.75 MHz, respectively. A 2.5 mm δ1(θ, φ) ) δ + 432 · δQ4(θ, φ) · P4(cos β) (1)
17 Q
double resonance MAS probe was used for all experiments on the
9.4 T instrument. Experiments on the 14.1 T instrument utilized a
648 0
5 mm triple resonance liquid state probe. Variable offset cumulative 〈δ1〉 ) δ (2)
spectra (VOCS)43 echo data were constructed from 15 Hahn echo 17 Q
spectra collected with an offset difference of 78.125 and 62.5
kHz, using a radio frequency field strength of 20 and 11 kHz 10
for the 9.4 and 14.1 T spectrometers, respectively. Incomplete δ2 ) δCS
0
- 〈δ 〉 (3)
27 1
double frequency sweeps (DFS) in conjunction with the VOCS
method were used on both instruments for selective enhance-
ment of small EFG environments for spectral editing purposes.44 where δQ0 is the quadrupolar shift defined as δQ0 ) -[CQ2(3 +
The frequency sweep was placed before the Hahn echo pulses. η2)]/[10νL2{2I(2I - 1)}2], δQ4(θ,φ) · P4(cos β) describes the
The DFS pulse had a start frequency of 650 kHz and final quadrupolar line shape, νL is the Larmor frequency, and 〈δ1〉 is
frequency of 250 kHz with a radio frequency field strength of the first moment of the quadrupolar line shape in the first
20 and 10 kHz for the sweep for the 9.4 and 14.1 T dimension. More information and a more general derivation for
spectrometers, respectively. The start frequency was selected all half-integer quadrupolar spins are presented in the Supporting
Information.
based on the size of the quadrupolar coupling of the site of
All data were processed with the program RMN.48 Single
interest. The final frequency was selected so that the sweep
field 1D spectra were simulated with the program DMFIT.49
ended before covering the central transition line shape. The Hahn
Simulations of the multiple field static echo patterns and
echo pulses had the radio frequency field strengths as the normal
difference spectra from the incomplete DFS experiments were
VOCS echo experiment. An adiabaticity parameter45 of 0.32
achieved by using an iterative fitting method with the program
was used for both sweeps. The resulting difference spectrum
WSOLIDS.50 Errors for the NMR parameters determined with
resulted from the subtraction of the normal VOCS spectrum
the WSOLIDS programs were estimated based both on visual
from the VOCS spectrum created by the incomplete DFS
inspection and on the change in the parameter that yielded a
experiment. The static cross-polarization (CP) experiments were
doubling in the sum of the squares of the deviation of the
collected at 9.4 T, using a 1H resonance frequency of 399.76
observed data from the model.
MHz, and the VOCS method was used to construct the final
4. Ab Initio Calculations. The first-principle periodic cal-
spectrum with an offset difference of 39.0625 kHz and utilized
culations were carried out utilizing DFT with a generalized
radio frequency strengths of 13 kHz for 93Nb and 65 kHz for
1 gradient approximation (GGA) in the parametrization of Perdew,
H with a contact time of 2 ms. All spectra were referenced to
Burke, and Ernzerhof51 as implemented in the WIEN2k code.52
NbCl5 in acetonitrile (0 ppm).
The electrons of the atoms were considered without any core
3QMAS experiments were conducted with use of a FAM-II approximation. Full potential (linearized) augmented plane
conversion pulse46 and a selective π pulse to generate a shifted waves plus local orbitals (L/APW+lo)53,54 were used. Better
echo to ensure whole echo data collection. A 96-step phase cycle convergence was ensured with a combination of basis functions,
following the +3Q f +1Q f -1Q pathway was used to with s and p states in LAPW mode and the d state in APW+lo
acquire the echo coherence transfer pathway only. An excitation mode. The core states were treated fully relativistically and for
pulse of 1.7 µs and a four-pulse FAM-II sequence with pulse the valence states a scalar, relativistic scheme was employed
lengths of 0.5 and 0.3 µs were used with a radio frequency field with -6.0 Ry as the cutoff energy. Temperature smearing of
strength of 85 kHz. The selective π pulse had a length of 2.0 0.002 Ry was used throughout the calculation to help in the
µs, using a radio frequency field strength of 50 kHz. The dwell elimination of virtual states emerging from the possible non-
time in the indirect dimension was set to ensure rotor- linearization of the wave functions. The charge density mixing
synchronized conditions. A MAS rate of 28.5 kHz was used. scheme used was the multisecant method55 with 0.1 as the
Eight data points were collected in the indirect dimension. A mixing parameter. All calculations were performed in single
spectral window of 400 kHz was used in the direct dimension. processor mode with the Clark University Computing Cluster,
An alternative shearing scheme was used to represent the 2D which is comprised of two integrated Linux clusters with 48
rotor synchronized data in a fashion more amenable to the large dual processor nodes.
chemical shift range of 93Nb. The method used is based on an The muffin-tin radii used were Rmt(Nb) ) 1.7 au, Rmt(O) )
approach outlined by Gan.47 An initial shearing transformation 1.3 au, Rmt(Rb) ) 2.4 au, Rmt(Sr) ) 2.2 au, and Rmt(H) )
using a coefficient of 3 as opposed to 91/36 was used to 0.99 au. The number of plane waves describing the system is
eliminate the chemical shift contribution in the indirect dimen- defined by a cutoff parameter RmtKmax, where Rmt (muffin-tin
sion. An additional shearing transformation was then applied radius) is the minimum radius of all the atoms present and Kmax
to the data to rotate the quadrupolar anisotropy to lie along the is the largest K vector in the plane wave expansion. To ensure
indirect dimension only. As a result, the first dimension only proper convergence, calculations up to RmtKmax ) 7.00 for the
contains anisotropy information related to the second order rubidium form and up to RmtKmax ) 5.00 for the acid-exchanged
quadrupolar interaction under MAS conditions. Since the form were carried out and showed no appreciable changes in
indirect dimension is rotor synchronized, all chemical shift the observed quantities compared to those determined with
anisotropy information in this dimension has been removed. The RmtKmax ) 6.50 or 6.00 for the rubidium form and with RmtKmax
direct dimension now only contains isotropic chemical shift and ) 4.50 or 4.00 for the acid-exchanged form. The cutoff of
Distortions of the Dion-Jacobson Phase RbSr2Nb3O10 J. Phys. Chem. C, Vol. 113, No. 40, 2009 17551

Figure 3. 87Rb MAS one-pulse NMR spectrum (top) and simulated


spectrum (bottom), using the parameters listed in Table 2.

Figure 2. X-ray powder diffraction patterns for RbSr2Nb3O10 (top) but was not clearly identified. The values for the acid-exchange
and HSr2Nb3O10. (bottom). Possible impurity reflections are designated form show the expected decrease in the c axis as the space
by asterisks. between layers decreases due to the smaller cation. The decrease
of 0.44 Å is similar to the change observed in other
TABLE 1: Lattice Constants for the Rubidium and Dion-Jacobson materials upon acid exchange.58 The lateral
Acid-Exchanged Formsa
dimensions of the layers do not appear to change as the a axis
a (Å) c (Å) length is the same for the acid-exchanged form.
Rb form P4/mmm 3.8988(1) 15.2831(5) The unit cell described by the literature represents the smallest
P4/mbm 5.5136(1) 15.2834(5) unit cell possible for the triple-layer perovskites. A unit cell with
Pbamb 5.5127(4) 15.2835(5) double the volume is also a possible structure. The lattice constants
5.5148(5) for the a axis differ by the square root of two, which is achievable
H form P4/mbm 5.5077(3) 14.8357(18) if the doubled structure has octahedra at both the corner and the
Pbamb 5.5057(18) 14.8437(28)
center of the ab plane. The space groups P4/mbm, P4jb2, P4bm,
5.5093(19)
and P4/m will accommodate such a structure. The increased size
a
Error on the last digit(s) is listed in parentheses. b For the of the unit cell allows for tilting of the octahedra, which can
orthorhombic space group, the b lattice constant is listed below the lengthen the Nb-O bonds (vide infra). Lower symmetry space
a lattice constant. groups with an expanded unit cell such as Pbam were also fit
by using the LeBail approach. The values for the lattice constants
charge density expansion in the interstitial region is limited to are reported in Table 1.
Gmax ) 12 (20 for the acid-exchanged form). For full conver- 2. Solid State NMR. 87Rb MAS experiments yielded a well-
gence in terms of the irreducible wedge of the Brillouin zone defined quadrupolar powder pattern that could be fit by using
(IBZ), the sampling was done up to 15 special k-points, which one rubidium environment (Figure 3). The isotropic chemical
corresponds to a 9 × 9 × 3 Monkhorst-Pack mesh56 in the shift for the site was -4.1 ( 0.1 ppm, the value of CQ was 5.1
full Brillouin zone. Higher k-point sampling did not alter the ( 0.1 MHz, and the value of η was 0.04 ( 0.01 (Table 2). The
results significantly showing the acceptable achievement of the sharp features of the signal allude to a well-organized sample
saturation point. For the relaxation of the structures, RmtKmax ) that most likely has only one crystallographic site for the
5.00 was used with 10 k-points in the IBZ. The NEW1 rubidium cation.
93
minimization method was used, and a calculation with final Nb MAS spectra were not well-defined due to the large
forces less than 5 mRy/au was considered relaxed. sideband pattern as the maximum spinning rate was not
The EFG tensor was determined from the expansion of the sufficient to average the different anisotropies. However, rotor
converged self-consistent electron density over the spherical synchronized MQMAS was able to resolve the different 93Nb
harmonics of the wave. The periodic calculation takes into environments of low to moderate quadrupolar couplings due to
account the representative unit cell with infinite repetition in the shearing method applied (Figure 4). Different environments
all three dimensions. As the calculation is self-consistent and with very different quadrupolar couplings can be clearly
deals with all electrons, no Sternheimer shielding factor was observed due to the different locations of the powder patterns
needed. The quadrupolar moments used for 87Rb and 93Nb were along the anisotropic axis. The more the powder pattern is
13.35 and -32.0 fm2, respectively.57 While the sign of the VZZ shifted to larger negative values, the larger the value of CQ.
component could be determined in the calculation, it was ignored The separation of the line shapes along with the alignment of
as NMR is not sensitive to the sign. the sidebands makes the determination of CQ, η, and the
isotropic chemical shift possible even with experiments in which
the MAS rate is not large enough to minimize the presence of
Results sidebands. The extent of the sideband pattern along the isotropic
1. X-ray Powder Diffraction. The powder diffraction data dimension also gives an indication of the size of the static
(Figure 2) for the rubidium sample and the acid-exchanged form powder pattern for the particular site and thus an indication if
were analyzed by using LeBail fits to determine the lattice the site has a large CSA in addition to quadrupolar anisotropy.
parameters. The parameters derived from the fits with use of a Due to incomplete excitation of the total sideband manifold,
P4/mmm space group are listed in Table 1. The lattice the intensities of all the sidebands for the broad patterns are
parameters for the rubidium form match the literature values.23 not reproduced properly, hence the discrepancies between the
A small impurity phase was observed in the diffraction pattern summed spectra and the simulated spectra. However, using the
17552 J. Phys. Chem. C, Vol. 113, No. 40, 2009 Wang et al.

TABLE 2: NMR Fit Parametersa


δISO (ppm) CQ (MHz) η Ω (ppm) κ Rb (deg) β (deg) γb (deg)
RbSr2Nb3O10
Rb -4.1(1) 5.1(1) 0.04(1)
Nb1 -1060(9) 93.0(4) 0.01(2) 530(30) 0.80(11) 0(80) 1(1) 0(80)
Nb2 -971(9) 44.6(6) 0.05(3) 820(25) 0.85(6) u 0(1) u
Nbi -1008(15) 29.5(6) 0.09(6) 120(35) -0.6(3) 80(10) 75(15) 70(20)
HSr2Nb3O10
Nb1 -1069(11) 86.0(9) 0.02(2) 540(30) 0.66(10) 0(30) 1(2) 0(30)
Nb2 -1148(9) 38.5(5) 0.06(3) 560(25) 0.75(11) u 0(1) u
a
Error on the last digit(s) is listed in parentheses. b Angle undefined (u) if β ) 0.

Figure 4. 93Nb triple quantum MQMAS 2D spectrum of RbSr2Nb3O10 with alternate shearing scheme (left); sum spectrum of side bands (black
line) for different sites with the accompanying simulation (red line) based on parameters listed in Table 2 (right). The two sum spectra are offset
from each other due to different sizes of the EFG tensors. The origins of the different sum spectra are linked to the final spectrum via the black
arrows.

Figure 5. 93Nb triple quantum MQMAS 2D spectrum of HSr2Nb3O10 with alternate shearing scheme (left); sum spectrum of side bands (black
line) with the accompanying simulation (red line) based on parameters listed in Table 2 (right). No small EFG site was observed with the acid-
exchanged form.

position of the singularities in the powder pattern, the values location of the singularity, further justifying the EFG values
for CQ and η can be determined. In addition, only CQ and η for assigned for this environment (Supporting Information, Figure
the site control the effective isotropic shift for the powder pattern S1). It appears that the environment with the larger EFG
along the anisotropic axis and serve as an additional measure anisotropy also has a significant CSA as such an EFG alone
of the size of the anisotropy (see the Supporting Information would only have a sideband manifold of approximately 1500
for more details). ppm. The values associated with the two environments are listed
For the rubidium form, two environments of different CQ in Table 2.
values are clearly distinguished (Figure 4). From the location The MQMAS experiment on the acid-exchanged form only
of the singularities for the larger CQ environment and the pattern showed one environment in the low to moderate EFG range
for the smaller CQ environment (referred to as Nb2 and Nbi), (Figure 5). The singularities in the powder pattern had better
the CQ values were extracted yielding 44.6 and 29.5 MHz and intensities, as the sideband manifold was not as extensive as in
asymmetry parameters of 0.05 and 0.09, respectively. To better the Rb form due to the smaller EFG and a smaller CSA. The
distinguish the more negative singularity for the large CQ environment had a CQ value of 38.5 MHz and an asymmetry
environment, a second MQMAS experiment was conducted with parameter of 0.07. The values for this environment are listed
a shift in the transmitter offset to better excite the region on the in Table 2.
more negative side of the sideband envelope. The improved To determine the CSA and the relative populations for the
intensity of the sideband pattern on this side confirmed the various niobium environments, since MQMAS is not inherently
Distortions of the Dion-Jacobson Phase RbSr2Nb3O10 J. Phys. Chem. C, Vol. 113, No. 40, 2009 17553

Figure 6. 93Nb static VOCS spectra for RbSr2Nb3O10 at 9.4 (a) and 14.1 T (b). The original spectra are shown at teh top and the final simulations
using the parameters listed in Table 2 are shown second from the top. The bottom three simulated spectra, in order from top to bottom, are for the
external site, Nb2, the interior site, Nb1, and the impurity, Nbi, using the parameters from Table 2.

Figure 7. 93Nb static VOCS spectra for HSr2Nb3O10 at 9.4 (a) and 14.1 T (b). The original spectra are shown at the top and the final simulations
using the parameters listed in Table 2 are shown second from the top. The bottom two simulated spectra, in order from top to bottom, are for the
external site, Nb2, and the interior site, Nb1, using the parameters from Table 2.

quantitative, wide-line experiments were conducted with use parameters of nearly zero. The CSA for the larger EFG
of the VOCS method to extend the frequency range covered environments were similar with spans of 530-540 ppm and
by the experiment (Figures 6 and 7). Initial experiments revealed skews that were nearly similar with values of 0.8 and 0.7 for
the static powder patterns for the environments observed in the the rubidium and acid-exchanged forms, respectively. The tensor
MQMAS experiments. When the frequency offsets were values for the EFG and CSA for the environments are listed in
expanded such that the window covered was (600 kHz, it Table 2. For both samples, the EFG tensors are effectively
became clear that an additional niobium environment (Nb1) was axially symmetric. However, the CSA tensors for both samples
present with a significantly larger EFG and CSA. While some do not have this symmetry. The EFG and CSA tensors are
reduction in the width was observed with the acid-exchanged aligned in both samples for the large EFG environments.
form, the large EFG environment was present in both samples. For the smaller EFG environments in the two samples, the
The lower population of this more anisotropic environment also values for the CSA tensors and the angles between the EFG
contributed to the initial difficulty in its detection. Using the and CSA tensors were determined by using the multiple field
two magnetic fields, the EFG and CSA for the two major data; the EFG tensor values were set according to the MQMAS
niobium environments were determined. In both the rubidium results. For this environment in the rubidium form, the CSA
and acid-exchanged forms, the ratio of the larger EFG to the was found to be significantly larger with a span of 820 ppm
smaller EFG environment was 1:2, which matches the ratio of and a skew of 0.85. In the acid-exchanged form, the CSA tensor
interior sites to exterior sites found in the crystallographic was similar to that for the larger EFG environment with a span
structure for triple-layer perovskites (Figure 1). of 560 ppm and a skew of 0.75. For both samples, the primary
The larger EFG environments (Nb1) had CQ values of 93 axes of both tensors were aligned, as β is effectively zero. The
MHz for the rubidium form and 86 MHz for the acid-exchanged values for the simulations are listed in Table 2.
form, which explains why these environments were not observed The smallest EFG environment (Nbi) is probably due to an
in the MQMAS experiments. Both environments had asymmetry impurity in the sample. While it was observed in the MQMAS
17554 J. Phys. Chem. C, Vol. 113, No. 40, 2009 Wang et al.

Figure 8. 93Nb difference spectra for RbSr2Nb3O10 created from incomplete DFS experiments at 9.4 (a) and 14.1 T (b) are shown at the top.
Converging frequency sweeps from 650 to 250 kHz were used. The simulated spectra shown at the bottom use the parameters for the impurity, Nbi,
listed in Table 2. An artifact from the external site, Nb2, is observed at the more negative frequencies and was not included in the simulation.

experiment on the rubidium sample, it is difficult to observe in


the wide-line spectra and could not be properly simulated by
using the two-field data collected. To properly describe the two
sites, it is necessary to determine the contribution of the impurity
to the overall powder pattern. To resolve this environment,
incomplete DFS experiments44 were conducted at both 9.4 and
14.1 T (Supporting Information, Figure S2) and the difference
spectra were used to resolve the contribution from this third
niobium environment (Figure 8). Since the effectiveness of the
technique is dependent on the inherent difference in the width
of the satellite transition patterns of the different niobium
environments, some excitation of the other environments is
possible. However, the values from the MQMAS experiment
can help determine the appropriate start frequency, and the
sideband pattern observed indicates that the extent of the static Figure 9. {1H}-93Nb cross-polarization spectrum for HSr2Nb3O10
powder pattern should be around 300 ppm or less based on the recorded at 9.4 T (top) and the simulated spectra (bottom) using the
parameters for the exterior site, Nb2, listed in Table 2.
MAS rate of 28.5 kHz. An isolated line shape was found for
the difference spectra at both magnetic fields that corresponded TABLE 3: DFT Calculation Results for RbSr2Nb3O10
with the isotropic chemical shift determined through MQMAS.
CQ (MHz) η
Due to the selection of the start and final frequencies for the
sweep, a small part of the line shape for the other smaller EFG Nb1 Nb2 Rb Nb1 Nb2 Rb
environment was also enhanced. The fragment from the Nb2 literature structure -21.057 37.174 3.033 0.00 0.00 0.00
site corresponds to the edge of the powder pattern (as seen in (P4/mmm)
the original DFS spectrum in Figure S2 in the Supporting relaxed structure -3.314 9.773 5.823 0.00 0.00 0.00
Information) and thus was not included in the simulated powder (P4/mmm)
proposed structure -93.499 44.529 5.206 0.00 0.00 0.01
pattern for the impurity. With the two-field data, the CSA tensor (P4/mbm)
information was found through simulation although the param-
eters are less accurate due to the lack of significant resolvable
features. The values for the simulation of this environment are the rubidium site in the structure. The structure based on the
listed in Table 2. The incorporation of this environment into literature yielded results that were radically different from the
the wide-line simulation of the rubidium form accounted for observed experimental NMR values (Table 3). Often better
the lack of the deep valley expected in the axially symmetric agreement between theory and experiment has been achieved
powder patterns and explained the presence of a narrow valley through relaxation of the crystallographic structure. Relaxation
in the 14.1 T wide-line spectrum. of the atom positions for the P4/mmm structure did not yield a
While the assignment of the smaller EFG environment to the significant improvement for the values of the EFG for the
exterior niobium site would be appropriate based on the relative niobium sites but did bring the rubidium site EFG closer to the
populations of the large and small EFG environments, CP experimental values.
experiments were conducted to further determine which niobium With the failure of the P4/mmm structure to match the
environment was in close proximity to the proton in the acid- experimental NMR data, alternate structures were explored to
exchanged form. The static CP experiment further confirms the find an improved match with experimental values. The use of
assignment that the smaller EFG site is associated with the a P4/mbm structure afforded the opportunity to vary the Nb-O
proton as it was the only pattern that had polarization transfer bond lengths for the interior site in particular. The lower
(Figure 9). Furthermore, the pattern that was observed does symmetry space group permits the tilting of the octahedra with
match the shape of the powder pattern simulated by using the the rotation axis parallel to the c axis. The lack of a structure
data derived from the MQMAS and VOCS experiments. necessitated that the values of the tilt angle be varied to search
3. Ab Initio Calculations. The literature structure of the for a value of |VZZ| that was in the vicinity of the experimental
rubidium form23 was used in DFT calculations to determine the values. The change in tilt angle for the interior site generally
CQ and η values for the two crystallographic niobium sites and had only a small effect on the size of the VZZ value for the
Distortions of the Dion-Jacobson Phase RbSr2Nb3O10 J. Phys. Chem. C, Vol. 113, No. 40, 2009 17555

Figure 12. Plot of the total energy calculated with WIEN2k for the
RbSr2Nb3O10 structure versus the tilt angle of the interior site (Nb1)
octahedron.

Figure 10. Plot of the quadrupolar coupling of the interior site (Nb1)
and the exterior site (Nb2) niobium atoms in RbSr2Nb3O10 versus the
tilt angle of the interior octahedron is shown at the bottom. A fit (black
line) of the interior site CQ values using eq 4 is included. The fit
equation, with error in parentheses, was -663(11) + 625(12) cos3 θ.
The nature of the rotation (yellow arrow) is shown with the depiction
of the interior site (niobium, green; oxygen, red; strontium, blue) at
the top.

Figure 13. Plot of the total energy calculated with WIEN2k for the
RbSr2Nb3O10 structure versus the tilt angle of the exterior site (Nb2)
octahedron.

structure was found that matched the experimental values found


with 93Nb NMR (Table 3). The total energies for the different
structures were calculated and are plotted as a function of tilt
angle (Figures 12 and 13). For both sites, the lowest energy
structure correlates well with the structures that are closest to
the observed EFG values. In the case of the exterior site, a sign
relative to the tilt direction of the interior site could be established.
The tilt angle for the two niobium sites were 14((1)° and -3((1)°
for the interior and exterior sites, respectively. The axial bond length
for the interior site was 1.95 Å. The axial bond lengths for the
exterior site were 1.75 Å for the terminal bond and 2.45 Å for the
bridging oxygen.
The utility of the DFT calculations is limited for the acid-
exchanged material (HSr2Nb3O10) without a clear structure with
a known hydrogen atom position. Difficulties associated with
Figure 11. Plot of the quadrupolar coupling of the interior site (Nb1) the appearance of spurious eigenvalues, or ghost bands, limited
and the exterior site (Nb2) niobium atoms in RbSr2Nb3O10 versus the the possible structures as an Rmt of the hydrogen atom could
tilt angle of the exterior octahedron is shown at the bottom. A fit (black
line) of the exterior site CQ values using eq 4 is included. The fit not be made small enough to create short O-H bonds. A
equation, with error in parentheses, was -465(7) + 512(7) cos3 θ. The structure with a P1 space group was used to facilitate the
nature of the rotation (yellow arrow) is shown with the depiction of placement of the hydrogen atom at different positions between
the exterior site (niobium, green; oxygen, red) at the top. the layers. Starting with the structure derived for the rubidium
form, the change in the interlayer spacing was introduced while
exterior site, and vice versa (Figures 10 and 11). Thus the tilt the bond distances within the layers were maintained. To attempt
angles could be optimized independently. The observed CQ for to match the observed CQ values, the lengths of the axial Nb-O
the rubidium site was sensitive to the terminal Nb-O bond bonds were varied (Table 4). Calculations were run for structures
length of the exterior site and served as an additional constraint that had the hydrogen atom equally shared by the layers through
for the modification of the structure used in the calculations. A hydrogen bonding and that had the hydrogen more tightly held
Rb-O bond length of 3.13 Å for the eight Rb-O bonds yielded by one of the terminal oxygen atoms. The lower symmetry space
an appropriate CQ value for the rubidium site. A potential group and different relative positions for the hydrogen atoms
17556 J. Phys. Chem. C, Vol. 113, No. 40, 2009 Wang et al.

TABLE 4: Effect on the Electric Field Gradient and Quadrupolar Coupling of Nb upon Variation in the Nb1-O2 (bridging
O-atom) Bond Length in HSr2Nb3O10a
Nb1 Nb2a Nb2b
Nb1-O2 bond length (Å) EFG (1021 V/m2) CQ (MHz) EFG (1021 V/m2) CQ (MHz) EFG (1021 V/m2) CQ (MHz)
1.938 -13.806 -106.821 4.212 32.596 4.363 33.762
1.945 -12.869 -99.575 4.070 31.496 4.248 32.874
1.952 -12.007 -92.906 3.928 30.400 4.132 31.975
1.960 -11.092 -85.827 3.776 29.215 4.008 31.009
1.967 -10.309 -79.765 3.640 28.162 3.896 30.149
a
P1 space group with the same tilt angle Nb1-O-Nb1 as in the parent compound RbSr2Nb3O10.

TABLE 5: Effect on the Electric Field Gradient of the


External Site upon Variation in the Differences of the of 4/mmm. The P4/mbm structure can account for the nonaxial
Terminal O-H Bond in HSr2Nb3O10a symmetry of the atom, as it would be located at a 2c position
with site symmetry of mmm. The translation to the P4/mbm
Nb2a Nb2b
(long O-H bond)b (short O-H bond)b
structure would also maintain the restriction of one crystal-
lographic rubidium atom in the asymmetric unit, which appears
difference in O-H EFG CQ EFG CQ to be the case in the 87Rb MAS spectrum (Figure 2). The
bond lengths (Å) (1021 V/m2) (MHz) (1021 V/m2) (MHz)
possibility of the structure corresponding to the P4/mbm space
0.352 3.029 23.436 4.430 34.274 group has been suggested based on Raman spectroscopy.25
0.246 3.338 25.828 4.300 33.269
0.166 3.565 27.583 4.169 32.261
From the 93Nb NMR, it is clear that the local site symmetry
0.086 3.776 29.215 4.008 31.009 for the niobium positions in the rubidium form does not contain
a 4-fold axis. The exterior site clearly has a nonaxial EFG tensor
a
P1 space group with Nb1-O ) 1.960 Å, Nb2a-O ) 1.755 Å, while the asymmetry parameter for the interior niobium site is
and Nb2b-O )1.762 Å. b (Nb2a) O-H ) 1.599 Å and (Nb2b)
O-H ) 1.247 Å for the first set of calculations.
zero within the error. However, the CSA tensors for both sites
clearly point to a nonaxial tensor environment. As with the
TABLE 6: Effect on the Electric Field Gradient of External position of the rubidium atom, this would eliminate the
Site upon Variation in the Differences of the Terminal O-H possibility of the niobium atoms sitting at positions with 4-fold
Bond in HSr2Nb3O10 When the Terminal Bond Is Longera symmetry and thus prevent the structure from being described
Nb2a Nb2b by any tetragonal space group. Nevertheless, the actual structure
(long O-H bond)b (short O-H bond)b would not deviate extremely from the tetragonal structure as
difference in O-H EFG CQ EFG CQ the near-axial nature of the tensors attests to the similarity to
bond lengths (Å) (1021 V/m2) (MHz) (1021 V/m2) (MHz) the idealized higher symmetry structure. As the Euler angle β
is essentially zero, the alignment of the EFG and CSA tensors
0.305 4.631 35.833 4.176 32.311
0.199 4.770 36.909 4.071 31.503 for both environments further points to the relatively high
0.119 5.292 40.946 4.067 31.471 symmetry nature of the structure that is present. A lower
0.039 5.362 41.485 3.683 28.495 symmetry group along the lines of Pbam or Pba2 would satisfy
the symmetry requirements for both the rubidium and niobium
a
P1 space group with Nb1-O ) 1.960 Å, Nb2a-O ) 1.806 Å,
and Nb2b-O )1.762 Å. b (Nb2a) O-H ) 1.552 Å and (Nb2b) sites.
O-H ) 1.247 Å for the first set of calculations. It is interesting to note that the size of the EFG tensor values
for the two sites is not necessarily what would be expected based
on the appearance of the niobium sites. In particular, the interior
effectively split the exterior sites and are referred to as Nb2a niobium site would appear to be relatively symmetric, yet the CQ
and Nb2b. The results of the various structures are shown in value is one of the largest recorded for 93Nb.36 The conventional
Tables 5 and 6. expectation has been that the EFG would increase as the
octahedral site became increasingly rhombic;60 yet the interior
Discussion site does not drastically deviate from tetragonal symmetry. It
Since the utility of X-ray powder diffraction methods for is clear that the long-range lattice with the influence of the
crystal structure solutions is limited due to the overlapping of positively charged alkaline earth cations, along with the oxygen
reflections, additional information from NMR experiments is atoms further out, add to the EFG, although in this case, the
necessary to constrain the list of possible structures. The simple effect of the increase in bond length caused by the tilt
crystallographic symmetry of the structure can place restrictions of the octahedra plays an even greater role in determining the
on the symmetry of the NMR tensor for that site. The local magnitude of the gradient. Symmetry in these materials does
symmetry established by the space group can limit the directions not imply an effective cancellation or reduction of the gradient.
of the tensor components and their relative magnitudes, es- The DFT calculations further support the presence of a tilt
sentially dictating whether or not axial symmetry must be in the octahedra and the resulting deviation from P4/mmm
present.59 From the EFG and CSA tensors for the two quadru- symmetry. The calculated EFG values for the literature structure
polar nuclei in the samples, it is possible to determine if the are in line with the expectation that the high symmetry interior
proposed structures are appropriate. The 87Rb NMR yields an site would have a small EFG. As no nearly isotropic EFG
EFG tensor with nonaxial symmetry as η ) 0.04. In this case, environments were observed experimentally, the structural
the rubidium atom cannot be located at a position with site model used was clearly not appropriate. Since previous Raman
symmetry higher than orthorhombic. This constraint thus studies have alluded to a tilt in the octahedra, the P4/mbm
invalidates the P4/mmm structure for the rubidium form as the structure with an expanded unit cell was utilized in order to
rubidium atom would sit at the 1d position with site symmetry obtain the flexibility needed in the calculations to vary the EFG
Distortions of the Dion-Jacobson Phase RbSr2Nb3O10 J. Phys. Chem. C, Vol. 113, No. 40, 2009 17557

at the two sites. The adjustments in the oxygen atom positions the coefficient will become smaller. For the bond angle of 100.9°
were further controlled by the need to match the 87Rb EFG used in the DFT calculations at the exterior site in the structure,
values. The primary modification was in the tilt of the octahedra the coefficient is predicted to decrease by a factor of 0.85. The
as the assumption was that the laboratory powder X-ray observed ratio between the coefficients derived from the DFT
diffraction data were unable to determine the small shift in the based data is 0.82 ( 0.02. A bond angle of 101.8° would give
ab plane that created the tilt. The modification resulted in a this ratio. While the point charge model is not exact in describing
structure that better matched the NMR results. the differences observed in the structure, it is clear that it can
The tilt of the interior niobium site had a drastic effect on serve as a reasonable qualitative model for trends in these
the size of the EFG making possible the determination of the Dion-Jacobson niobates and possibly other niobates.
size of the tilt (Figure 10). A tilt of 14° yields a CQ value within The acid exchange of the material produced changes at both
the error of the experimentally determined value and is niobium sites implying that protonation does not only affect
consistent with the magnitude of the tilt angles found in other the structure of the surface site. Both sites have a decrease in
perovskites.24,61 The exterior site seemed to have a smaller tilt
CQ value of approximately 7 MHz but very little change in
angle (Figure 11) and its orientation relative to the interior site
symmetry is observed. Changes related to the isotropic chemical
could only be determined based on the total calculated energy
shift and the CSA are limited to the exterior site only.
(Figure 13). The coordination of the strontium atoms would
Protonation of the terminal oxygen atom results in a change in
appear to perturb the orientation of the octahedra. A tilt of the
the isotropic chemical shift of 177 ppm and a decrease in the
interior site octahedra brings two of the four equatorial oxygen
span of the CSA of 260 ppm as observed from both the static
atoms into closer proximity with the strontium atom with a
change to 2.66 Å from the original distance of 2.96 Å. A slight echo and static CP experiments. The bonding of the proton with
tilt of the exterior site octahedra would result in two oxygen the terminal oxygen atom would be expected to change its
atoms moving even closer to the strontium atom. Without the bonding with the niobium atom. The lack of change in symmetry
tilt at the exterior site, the strontium cation has six oxygen atoms implies that the effect is mostly limited to the change in the
at a close distance of approximately 2.66 Å. These oxygen atoms length of the terminal Nb-O bond or at least a change in the
lie closer than the axial oxygen atoms at the interior site, which niobium atom position along the c axis relative to the oxygen
sit at a distance of 2.77 Å from the strontium. The improved atoms in the polyhedron. The net effect must reduce the
coordination would explain why significant tilting only occurs asymmetry of the polyhedron to account for the reduction in
at the interior site. both anisotropic interactions.
The value of CQ for both sites appears to have a similar The DFT calculations help to clarify the possible changes in
dependence on the tilt angle. A simple point charge model of the structure that might occur upon protonation. The variations
an octahedron in which the change in the equatorial Nb-O bond in the bond length of the axial Nb-O bond of the interior site
length depends on the tilt angle predicts a cos3 θ dependence: show that a small change in length can drastically affect the
EFG at the interior site when the same tilt angle as in the

[ ]
rubidium form is used (Table 4). A change of 7 MHz could
2e2Q 2Zax 2Zeq
CQ ∝ - 3 cos3 θt (4) easily be achieved by a simple increase of approximately
h 3
rax req,0 0.01 Å. The change in the position of this bridging oxygen
significantly alters the EFG at the exterior by 1-2 MHz. So
where Z is the charge for the axial and equatorial oxygen atoms, the rearrangement of the bridging oxygen atoms can explain
rax is the axial Nb-O bond length, req,0 is the equatorial Nb-O some of the decrease at the exterior site, but some changes
bond length if no tilt was present, and θt is the tilt angle. Both involving the other oxygen atoms of the site must occur.
data sets follow this functional form reasonably well (Figures The various models with hydrogen atoms being equally shared
10 and 11). The influence of the lattice outside the octahedron across the gap or preferentially bound to one oxygen atom
should have an effect on the final values but this is most likely demonstrate the effect of the increasing strength of the O-H
manifested as an offset for the function. The coefficients for bond on the niobium EFG (Table 5). A tightly held hydrogen
both terms are different for the two sites. As the sites have atom increases the magnitude of the EFG and a longer O-H
different charged cations in close proximity, it is not surprising bond decreases it. A difference in O-H bond length of 0.35 Å
that the offset differs. However, the change in the coefficient leads to a difference of 11 MHz in CQ. Similarly, the longer
for the cosine term can be explained by the change in the the terminal Nb-O bond, the larger the size of the EFG. A
geometry of the octahedron. If a non-90° bond angle is allowed change in the bond length of 0.051 Å increased the CQ value
for the equatorial oxygen atoms, then the influence on the Nb-O by 12 MHz if the oxygen is not bonded to hydrogen as seen in
bond length can be introduced. The modified point charge model the change for the Nb2a site. As the expectation is that the
equation is now: terminal Nb-O bond should increase in length upon protonation
due to loss in electron density in the bond, additional rearrange-
ments must occur that lead to the net decrease in the EFG. One
2e2Q
CQ ∝ × candidate would be an increase in the O-Nb-O bond angle
h

[ ]
resulting in the niobium atom sitting further away from the
Zax1 Zax2 2Zeq(3 cos2 θONbO - 1)sin3 θONbO 3 equatorial oxygen atoms. On the basis of the calculations on
+ + cos θt
r3ax1 r3ax2 3
req,0 the rubidium form, the increase in the O-Nb-O bond angle
will decrease the size of the EFG. In a sense, an elongation of
(5) the five-coordinate polyhedron would occur with the protonation.
Without further information to constrain the structure, such as
17
where rax1 and rax2 are the short and long axial Nb-O bond O EFG information, it is not possible to conclusively determine
lengths and θONbO is the bond angle between axial and equatorial the changes responsible for the decrease in the EFG at the
oxygen atoms. When the O-Nb-O angle is greater than 90°, exterior site, although it is clear that some combination of
17558 J. Phys. Chem. C, Vol. 113, No. 40, 2009 Wang et al.

increase in bond length and bond angle must be responsible (13) Sasaki, T. J. Ceram. Soc. Jpn. 2007, 115, 9–16.
for the observed change in 93Nb EFG values. (14) Ebina, Y.; Sasaki, I.; Watanabe, A. Solid State Ionics 2002, 151,
177–182.
(15) Paek, S. M.; Jung, H.; Lee, Y. J.; Park, M.; Hwang, S. J.; Choy,
Conclusion J. H. Chem. Mater. 2006, 18, 1134–1140.
(16) Fukuda, K.; Akatsuka, K.; Ebina, Y.; Ma, R.; Takada, K.; Nakai,
Solid-state NMR and DFT methods were able to clarify the I.; Sasaki, T. ACS Nano 2008, 2, 1689–1695.
structure of RbSr2Nb3O10 and elucidate the structural changes (17) Ebitani, K.; Ohmatsuzawa, T.; Matsunami, E.; Tanaka, T.; Hattori,
that must occur upon acid exchange of this compound. 93Nb H. Catal. Today 1993, 16, 447–454.
(18) Jacobson, A. J.; Lewandowski, J. T.; Johnson, J. W. Mater. Res.
MQMAS and multiple field VOCS-based wide-line NMR Bull. 1990, 25, 679–686.
experiments were able to determine the EFG and CSA tensor (19) Jacobson, A. J.; Lewandowski, J. T.; Johnson, J. W. J. Less-
information for the different niobium environments in a triple- Common Met. 1986, 116, 137–146.
perovskite layered niobate. The spectra demonstrate that the (20) Dion, M.; Ganne, M.; Tournoux, M. Mater. Res. Bull. 1981, 16,
1429–1435.
niobium site symmetry is lower than what is expected based (21) Fukuoka, H.; Isami, T.; Yamanaka, S. J. Solid State Chem. 2000,
on the known literature structure and that octahedral tilting must 151, 40–45.
be present. Periodic all-electron DFT calculations were capable (22) Dion, M.; Ganne, M.; Tournoux, M.; Ravez, J. ReV. Chim. Miner.
of reproducing the EFG values observed if tilting is incorporated 1984, 21, 92–103.
(23) Thangadurai, V.; Schmid-Beurmann, P.; Weppner, W. J. Solid State
into the structure. The tilting does not significantly distort the Chem. 2001, 158, 279–289.
structure from a tetragonal lattice and occurs along only one (24) Thomas, N. W. Acta Crystallogr. B 1998, 54, 585–599.
direction. The magnitude and direction of the rotation differs (25) Kurzman, J. A.; Geselbracht, M. J. In Solid-State Chemistry of
for the two niobium sites with the interior site having a larger Inorganic Materials VI, Proceedings of the 2006 MRS Fall Meeting, Boston,
MA, 2006; Seshadri, R., Kolis, J. W., Mitzi, D. B., Rosseinsky, M. J., Eds.;
deviation of 14° and the exterior site having a tilt of -3°. Acid Materials Research Society: Warrendale, PA, 2007; no. 0988-QQ08-06.
exchange of the interlayer cation resulted in changes in the (26) Jehng, J. M.; Wachs, I. E. Chem. Mater. 1991, 3, 100–107.
structure for both sites. DFT calculations of the trends associated (27) Byeon, S. H.; Nam, H. J. Chem. Mater. 2000, 12, 1771–1778.
(28) de Lacaillerie, J. B. D.; Barberon, F.; Romanenko, K. V.; Lapina,
with different Nb-O bond lengths point to only a change in O. B.; Le Polles, L.; Gautier, R.; Gan, Z. H. J. Phys. Chem. B 2005, 109,
axial bond lengths for the interior site. However, changes in 14033–14042.
the exterior site must be more complex with changes associated (29) Ooms, K. J.; Feindel, K. W.; Willans, M. J.; Wasylishen, R. E.;
with not only the axial Nb-O bonds but also the bond angles Hanna, J. V.; Pike, K. J.; Smith, M. E. Solid State Nucl. Magn. Reson.
2005, 28, 125–134.
associated with the equatorial oxygen atoms. The use of 93Nb (30) Ashbrook, S. E.; Le Polles, L.; Gautier, R.; Pickard, C. J.; Walton,
NMR in conjunction with quantum mechanical calculations can R. I. Phys. Chem. Chem. Phys. 2006, 8, 3423–3431.
yield insights on the subtle changes in structure that occur with (31) Body, M.; Legein, C.; Buzare, J. Y.; Silly, G.; Blaha, P.; Martineau,
postsynthetic processing and could prove to be useful for the C.; Calvayrac, F. J. Phys. Chem. A 2007, 111, 11873–11884.
(32) Martineau, C.; Body, M.; Legein, C.; Silly, G.; Buzare, J. Y.; Fayon,
study of these layered niobates under conditions of exfoliation F. Inorg. Chem. 2006, 45, 10215–10223.
or restacking when X-ray diffraction may not be informative. (33) Bryce, D. L.; Bultz, E. B.; Aebi, D. J. Am. Chem. Soc. 2008, 130,
9282–9292.
Acknowledgment. The work was supported in part by the (34) Ashbrook, S. E.; Duer, M. J. Concepts Magn. Reson., Part A 2006,
28, 183–248.
Petroleum Research Fund (PRF 46923-AC5) and NSF Grant (35) Herzfeld, J.; Berger, A. E. J. Chem. Phys. 1980, 73, 6021–6030.
No. DMR-0748399. The computer cluster work was supported (36) Lapina, O. B.; Khabibulin, D. F.; Romanenko, K. V.; Gan, Z. H.;
in part by NSF Grant No. DBI-0320875. The X-ray powder Zuev, M. G.; Krasil’nikov, V. N.; Fedorov, V. E. Solid State Nucl. Magn.
Reson. 2005, 28, 204–224.
diffraction equipment was supported by the Kresge Foundation (37) Lapina, O. B.; Khabibulin, D. F.; Shubin, A. A.; Terskikh, V. V.
Science Initiative. Prog. Nucl. Magn. Reson. Spectrosc. 2008, 53, 128–191.
(38) Massiot, D.; Montouillout, V.; Fayon, F.; Florian, P.; Bessada, C.
Supporting Information Available: Derivation of alternate Chem. Phys. Lett. 1997, 272, 295–300.
shearing scheme, 93Nb MQMAS spectrum for offset experiment, (39) Medek, A.; Harwood, J. S.; Frydman, L. J. Am. Chem. Soc. 1995,
117, 12779–12787.
additional DFS spectra, listing of EFG values, and crystal (40) Massiot, D. J. Magn. Reson. Ser. A 1996, 122, 240–244.
structures used for WIEN2k calculations. This material is (41) Lebail, A.; Duroy, H.; Fourquet, J. L. Mater. Res. Bull. 1988, 23,
available free of charge via the Internet at http://pubs.acs.org. 447–452.
(42) Bruker AXS TOPAS, V2.1: General profile and structure analysis
software for powder diffraction data; Bruker AXS, Karlsruhe, 2003.
References and Notes (43) Massiot, D.; Farnan, I.; Gautier, N.; Trumeau, D.; Trokiner, A.;
Coutures, J. P. Solid State Nucl. Magn. Reson. 1995, 4, 241–248.
(1) Yoshimura, J.; Ebina, Y.; Kondo, J.; Domen, K.; Tanaka, A. J.
Phys. Chem. 1993, 97, 1970–1973. (44) Wang, X.; Smith, L. J. Solid State Nucl. Magn. Reson., 2009.
(2) Ikeda, S.; Tanaka, A.; Shinohara, K.; Hara, M.; Kondo, J. N.; Submitted for publication.
Maruya, K. I.; Domen, K. Microporous Mater. 1997, 9, 253–258. (45) Kentgens, A. P. M.; Verhagen, R. Chem. Phys. Lett. 1999, 300,
(3) Fukuoka, H.; Isami, T.; Yamanaka, S. Chem. Lett. 1997, 703–704. 435–443.
(4) Takezawa, S.; Teranishi, T.; Ishikawa, H.; Tokumitsu, T.; Toda, (46) Morais, C. M.; Lopes, M.; Fernandez, C.; Rocha, J. Magn. Reson.
K.; Uematsu, K.; Sato, M. J. Ceram. Soc. Jpn. 2006, 114, 861–865. Chem. 2003, 41, 679–688.
(5) Kobayashi, Y.; Schottenfeld, J. A.; Macdonald, D. D.; Mallouk, (47) Gan, Z. H. Presented at 45th Rocky Mountain Conference on
T. E. J. Phys. Chem. C 2007, 111, 3185–3191. Analytical Chemistry, Denver, CO, July 2003; Paper no. 179.
(6) Sarahan, M. C.; Carroll, E. C.; Allen, M.; Larsen, D. S.; Browning, (48) Grandinetti, P. J. RMN, Version 1.3.0; The Ohio State University:
N. D.; Osterloh, F. E. J. Solid State Chem. 2008, 181, 1678–1683. Columbus, OH, 2005.
(7) Izawa, K.; Yamada, T.; Unal, U.; Ida, S.; Altuntasoglu, O.; (49) Massiot, D.; Fayon, F.; Capron, M.; King, I.; Le Calve, S.; Alonso,
Koinuma, M.; Matsumoto, Y. J. Phys. Chem. B 2006, 110, 4645–4650. B.; Durand, J. O.; Bujoli, B.; Gan, Z. H.; Hoatson, G. Magn. Reson. Chem.
(8) Dias, A. S.; Lima, S.; Carriazo, D.; Rives, V.; Pillinger, M.; Valente, 2002, 40, 70–76.
A. A. J. Catal. 2006, 244, 230–237. (50) Eichele, K.; Wasylishen, R. E. WSOLIDS NMR Simulation Package,
(9) Takagaki, A.; Lu, D. L.; Kondo, J. N.; Hara, M.; Hayashi, S.; Version 1.17.30; University of Tuebingen, Tuebingen, Germany, 2001.
Domen, K. Chem. Mater. 2005, 17, 2487–2489. (51) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. ReV. Lett. 1996, 77,
(10) Takagaki, A.; Yoshida, T.; Lu, D. L.; Kondo, J. N.; Hara, M.; 3865–3868.
Domen, K.; Hayashi, S. J. Phys. Chem. B 2004, 108, 11549–11555. (52) Schwarz, K.; Blaha, P.; Madsen, G. K. H. Comput. Phys. Commun.
(11) Takagaki, A.; Sugisawa, M.; Lu, D. L.; Kondo, J. N.; Hara, M.; 2002, 147, 71–76.
Domen, K.; Hayashi, S. J. Am. Chem. Soc. 2003, 125, 5479–5485. (53) Sjostedt, E.; Nordstrom, L.; Singh, D. J. Solid State Commun. 2000,
(12) Schaak, R. E.; Mallouk, T. E. Chem. Mater. 2002, 14, 1455–1471. 114, 15–20.
Distortions of the Dion-Jacobson Phase RbSr2Nb3O10 J. Phys. Chem. C, Vol. 113, No. 40, 2009 17559

(54) Madsen, G. K. H.; Blaha, P.; Schwarz, K.; Sjostedt, E.; Nordstrom, (59) Weil, J. A.; Buch, T.; Clapp, J. E. AdV. Magn. Reson. 1973, 6,
L. Phys. ReV. B 2001, 64, 195134. 183–257.
(55) Marks, L. D.; Luke, D. R. Phys. ReV. B 2008, 78, 075114. (60) Fitzgerald, J. J.; Prasad, S.; Huang, J.; Shore, J. S. J. Am. Chem.
(56) Monkhorst, H. J.; Pack, J. D. Phys. ReV. B 1976, 13, 5188–5192. Soc. 2000, 122, 2556–2566.
(57) Pyykko, P. Mol. Phys. 2001, 99, 1617–1629. (61) Thomas, N. W. Acta Crystallogr., Sect. B 1996, 52, 16–31.
(58) Jacobson, A. J.; Lewandowski, J. T.; Johnson, J. W. J. Less-
Common Met. 1986, 116, 137–146. JP902318V

You might also like