Solid-State 91Zr NMR Characterization of Layered and Three

Article pubs.acs.org/JPCC

Solid-State 91Zr NMR Characterization of Layered and ThreeDimensional Framework Zirconium Phosphates Andre Sutrisno,† Lei Liu,‡ Jinxiang Dong,*,‡ and Yining Huang*,† †

Department of Chemistry, The University of Western Ontario, London, Ontario, Canada, N6A 5B7 Research Institute of Special Chemicals, Taiyuan University of Technology, Taiyuan 030024, Shanxi, People's Republic of China

‡

S Supporting Information *

ABSTRACT: Layered and open framework zirconium phosphates (ZrPs) have many current and potential applications in the areas of catalysis, sorption, protonic conductors, solar energy storage, crystal engineering, and ion exchange. Characterization of ZrP-based materials is important because understanding the relationship between the properties of these materials and their structures is crucial for developing new uses and for improving their performances in current applications. However, local Zr environments in many ZrPs have not been characterized directly by 91Zr solid-state NMR (SSNMR). This is because 91Zn (I = 5/2) is an unreceptive nucleus with many NMR unfavorable characteristics, leading to low sensitivity. In this work, the local environments of the zirconium centers in several ion-exchanged derivatives of layered α-ZrP (K+-, Li+-, Co(NH3)63+-ZrP) have been probed directly using 91Zr MAS, static quadrupolar echo, and/or quadrupolar Carr− Purcell−Meiboom−Gill NMR. Several layered and three-dimensional framework zirconium phosphates (ZrPO4-DES8, ZrPO4DES1, ZrPO4-DES2, ZrPOF-pyr, ZrPOF-Q1, ZrPOF-EA, and ZrPOF-DEA) with novel structures were also examined. Theoretical calculations using the CASTEP and Gaussian model cluster approaches were also performed in order to provide insights into the observed spectra. In addition to 91Zr SSNMR, 31P, 13C, and 19F SSNMR spectroscopy was also utilized to characterize the above-mentioned materials.

■

INTRODUCTION Layered zirconium phosphates (ZrPs) and their derivatives have many applications in the areas of catalysis, sorption, protonic conductors, solar energy storage, crystal engineering, and, in particular, ion exchange and intercalation.1,2 Recent years have seen growing interest in the preparation of two- and three-dimensional zirconium phosphates with novel properties through a new synthetic approach, such as ionothermal synthesis. Characterization is important because understanding the relationship between the novel properties of these materials and their structures is crucial for developing new uses and for improving their performances in current applications. Unfortunately, characterization of ZrP-based materials is not always straightforward because it is often difficult to obtain suitable single crystals for X-ray diffraction. Therefore, structures of many zirconium phosphates have been determined from more limited powder XRD data. In addition, the ion exchange and intercalation often result in reduced crystallinity. Solid-state NMR (SSNMR) spectroscopy is a complementary technique to X-ray diffraction. However, in the past, ZrP-based materials have mainly been characterized by 1H and 31P MAS NMR. As shown previously, while the NMR spectra of the metal centers in some related layered metal phosphates exhibit significant changes, the 31P MAS spectra, however, are relatively insensitive to structural modification.3,4 Thus, it is crucial to © XXXX American Chemical Society

directly examine the metal center environments. Unfortunately, the local Zr environments in ZrPs have rarely been probed directly by 91Zr SSNMR. This is because 91Zr, the only NMRactive isotope of zirconium, is a quadrupolar nucleus (I = 5/2) with inherent low sensitivity due to its low gyromagnetic ratio (γ = −2.4975 × 107 rad T−1 s−1), low natural abundance (11.23%), and a moderately sized quadrupole moment (Q = −0.176 × 10−28 m2),5 yielding broad patterns that are difficult to be detected. Consequently, the number of reported 91Zr SSNMR studies of solids is relatively small.6−16 In recent years, the development of sensitivity enhancement techniques for low-γ quadrupolar nuclei, such as quadrupolar Carr−Purcell− Meiboom−Gill (QCPMG) and related sequences,10,17−19 and the availability of magnets with very high field strength make it possible to acquire useful 91Zr SSNMR spectra. The measured central transition (CT) powder patterns and NMR tensor calculations can yield important information regarding the local bonding and geometry at the Zr centers in various materials.20−22 Our previous studies have shown that 91Zr SSNMR is a sensitive tool to probe the local Zr environments in two- and three-dimensional ZrPs.11,12 In the present work, we systemReceived: May 15, 2012 Revised: July 19, 2012

A

dx.doi.org/10.1021/jp304729s | J. Phys. Chem. C XXXX, XXX, XXX−XXX

The Journal of Physical Chemistry C

Article

Figure 1. (a) The layered structure of α-ZrP. (b) (i) The layered structure of K-ZrP (the big gray balls are K+ cations and (ii) the local environment around Zr in K-ZrP [showing Zr(OPO3)614− fragment].

presence of fluorine in these zirconium phosphate fluoride (ZrPOF) materials led to unique properties, such as fluorescence, adsorption, molecular or chiral recognition, and catalysis. Further, introducing terminal fluorine during synthesis promotes the formation of three-dimensional open framework zirconium phosphates. One aim of this work is to characterize the ZrO5F, ZrO4F2, and ZrO2F4 coordination environments.

atically characterize the local structure of (i) several ionexchanged derivatives of layered α-zirconium hydrogen phosphate and (ii) a number of novel two- and threedimensional framework zirconium phosphates that were prepared by ionothermal syntheses. Layered zirconium hydrogen phosphate monohydrate, Zr(HPO4)2·H2O (hereafter referred to as α-ZrP), is the most well-known example of layered metal phosphates. The layered structure of α-ZrP is shown in Figure 1a. Each layer contains a single sheet of Zr atoms. Each Zr atom is octahedrally coordinated to six O atoms belonging to six [PO4] tetrahedra with each P atom being tetrahedrally bound to one hydroxyl group and three other O atoms shared with three different ZrO6 octahedral units. The proton in the hydroxyl group is acidic and, therefore, can be exchanged for other cations. The ion-exchange behavior of α-ZrP has been widely studied, including ion exchange toward alkali, alkaline earth, and transition-metal ions.23−27 In our previous study, we examined α-ZrP and its Na+-exchanged phase by 91Zr SSNMR and demonstrated that 91Zr NMR spectra are sensitive not only to the relatively small distortion in the ZrO6 octahedron but also to the spatial arrangement of the P atoms in the second coordination sphere (i.e., the configuration or geometry of Zr(OP)6 units).11 In the first part of the paper, we extended the study to include K+-, Li+-, and Co(NH3)63+-exchanged α-ZrP. Recently, several novel two- and three-dimensional framework zirconium phosphates (namely, ZrPO4-DES8, ZrPO4-DES1, ZrPO4-DES2, ZrPOF-pyr, ZrPOF-Q1, ZrPOF-EA, and ZrPOFDEA) have been papered, for the first time, through ionothermal syntheses by one of us (J. Dong). These materials have many potential industrial applications, such as selective oxidation of cyclohexane in ZrPO4-DES1 and ZrPO4-DES2,28 hydrogen storage in ZrPOF-pyr,29 photoluminescence in ZrPOF-Q1,30 and selective gas adsorption of CO2/CH4 in ZrPOF-EA.31 In the second part, the Zr centers in these materials were characterized by 91Zr SSNMR. It should be pointed out that six of the above-mentioned materials (ZrPO4DES1, ZrPO4-DES2, ZrPOF-pyr, ZrPOF-Q1, ZrPOF-EA, and ZrPOF-DEA) have terminal and/or bridging fluoride. The

■

EXPERIMENTAL METHODS Sample Preparation. The ion-exchanged phases of α-ZrP were prepared according to the previously reported procedures for K-ZrP,25 Li-ZrP,24,32 and Co-ZrP.27 The novel zirconium phosphate and zirconium phosphate-fluoride phases (ZrPO4DES8, ZrPO4-DES1, ZrPO4-DES2, ZrPOF-pyr, ZrPOF-Q1, ZrPOF-EA, and ZrPOF-DEA) were synthesized according to the previously reported methods.28,30,31,33,34 For experimental details, see the Supporting Information. Sample Purity. 31P MAS NMR and powder X-ray diffraction were performed to confirm sample identity and purity. Additional 13C, 19F, and 6/7Li MAS NMR experiments were also conducted (see the Supporting Information for details) wherever necessary. Powder X-ray diffraction experiments were recorded on a Rigaku diffractometer equipped with a graphite monochromator using Co Kα radiation (λ = 1.7902 Å). 91 Zr Solid-State NMR Spectroscopy. All the 91Zr NMR spectra were recorded at room temperature. Most 91Zr solidstate NMR experiments were conducted at 21.1 T (ν0 (91Zr) = 83.72 MHz) on a Bruker Avance II spectrometer at the National Ultrahigh-Field NMR Facility for Solids in Ottawa, Canada. All the MAS spectra were acquired using either a single-pulse or a quadrupolar echo sequence with proton decoupling on a Bruker 3.2 or a 4 mm H/X MAS probe. For the 4 mm MAS probe, silicon nitride rotors were utilized to avoid the Zr background signal. However, for the 3.2 mm probe, only ZrO2 rotors were available and background subtraction was performed. All static experiments were carried out using WURST-QCPMG10 and/or a quadrupolar echo sequence on a home-built 7 mm H/X low-γ probe with the B



Article

quadrupole moments.5 The CS tensor components are described by three principal components (δ11, δ22, and δ33) with Herzfeld−Berger convention: δiso = (δ11 + δ22 + δ33)/3; Ω = δ11 − δ33; κ = 3(δ22 − δiso)/Ω. One of the three Euler angles, β, describes the angle between the two largest components of the EFG and CS tensors (VZZ and δ33) and ranges from 0 to 180°. Gaussian 09. Ab initio calculations on model clusters were also conducted using the Gaussian 09 program44 running on SHARCNET clusters (www.sharcnet.ca). All model clusters used in the calculations were truncated from the layered and framework structures.28,30,31,33,34,41 This model cluster approach is computationally less expensive in general and has been shown to predict NMR tensors reasonably well.45−49 It is also suitable for exploring the dependence of calculated interactions on local structural parameters, such as bond lengths and angles. The 91Zr NMR tensors were calculated using hybrid density functional theory (DFT) at the B3LYP level of theory using the GIAO method. The basis sets used were Horn’s (17s11p8d) contracted to the [12s7p4d] allelectron basis set50 for Zr atoms and 6-311G** on the other atoms. The NMR tensor parameters were then extracted from the Gaussian output using the EFGShield program. 51 Calculated 91Zr isotropic chemical shielding (σiso) values for all model clusters were converted to the corresponding chemical shift (δiso) values by referencing it to α-ZrP: δiso = 1545 − σiso (all in ppm), with 1545 ppm corresponding to the sum of the experimental shift value (−385 ppm) and calculated shielding value (1930 ppm) of α-ZrP.

samples packed in a 7 mm glass tube. One additional 91Zr static NMR spectrum of ZrPO4-DES1 was also acquired on a Varian Inova 600 (ν0 (91Zr) = 55.73 MHz) spectrometer. A 3.2 mm HXY MAS probe was utilized, and the tightly packed powdered sample was sealed in a 3 mm (outer diameter) glass tube. The 91Zr chemical shifts were referenced to either a solid BaZrO3 (δiso = 0 ppm) or a concentrated solution of Cp2ZrCl2 in CH2Cl2 (as a secondary standard sample, δiso = −317.2 ppm, relative to BaZrO3). The selective π/2 pulse lengths for CT were determined on the above-mentioned compound, ranging from 1.3 to 7.5 μs (with the exception of a 50 μs pulse length in WURST-QCPMG experiments), depending on the spectrometer, probe, and pulse sequence used. A recycle delay of 1 s was used for all the experiments. For the QCPMG type of experiments, the acquisition time (τa) for each echo was adjusted to obtain a spikelet separation (1/τa) of 1000−5000 Hz. Detailed experimental conditions are listed in Table S1 in the Supporting Information. 91 Zr NMR Spectral Simulations. All NMR parameters, including CQ, ηQ, δiso, Ω, and κ (for definitions of these parameters, see the Theoretical Calculations section), were determined by analytical simulations of NMR spectra using the WSOLIDS simulation package.35 The simulation error for each measured parameter was determined by visual comparison of the experimental spectrum with the simulated one. The parameter of concern was varied bidirectionally starting from the best fit value, and all other parameters were kept constant, until noticeable differences between the spectra were observed. Theoretical Calculations. CASTEP. First-principles (ab initio) calculations based on plane-wave pseudopotential density functional theory (DFT) were conducted using the CASTEP36,37 program setup of the Materials Studio graphical user interface. The NMR module38−40 was used to calculate the electric field gradient (EFG) and chemical shielding (CS) tensors. This program separates periodic structures into two regions designated as atomic spheres and interstitial regions, using element-specific pseudopotentials to describe the former. The gauge-including projector augmented wave (GIPAW) method, which uses pseudopotentials and plane-wave basis sets to describe three-dimensional lattices in crystalline materials, was utilized. Unit cell parameters and atomic coordinates were taken from corresponding crystal structures.28,33,41 The calculations were performed using ultrasoft pseudopotentials generated from the “on-the-fly” method implemented within the CASTEP. The generalized gradient approximation (GGA) with the Perdew, Burke, and Ernzerhof (PBE) functional42,43 was used, and a plane-wave cutoff energy of 500 eV (medium basis set accuracy) was applied to all calculations. Whenever appropriate, geometries were optimized either on hydrogen positions or on all atoms using the same GGA approximation, PBE exchange-functional, Monkhorst−Pack k-point grid spacings, and cutoff energies as in the corresponding singlepoint energy calculations. The calculated EFGs (VXX, VYY, VZZ) were converted to the quadrupolar coupling constant (CQ) and asymmetry parameter (ηQ) according to the following definitions: |VXX| ≤ |VYY| ≤ |VZZ|; CQ = (eVZZQ/h) × 9.7177 × 1021 (Hz); ηQ = (VXX − VYY)/VZZ, where e is the electric charge, Q is the nuclear quadrupole moment [Q(91Zr) = −0.206 × 10−28 m2], and h is Planck’s constant. A conversion factor of 9.7177 × 1021 V m−2 was needed to convert eQVZZ to CQ (in Hz) due to VZZ being calculated in atomic units. The CQ values were calculated automatically from the EFG tensor by CASTEP and adjusted accordingly using the appropriate

■

RESULTS AND DISCUSSION Ion-Exchanged α-ZrP. Three ion-exchanged derivatives of α-ZrP, namely, K-ZrP, Li-ZrP, and Co-ZrP, were investigated by 91Zr SSNMR. Figure S1 (Supporting Information) shows powder X-ray diffraction patterns of α-ZrP and its ionexchanged derivatives. The interlayer spacing (d) increases with increasing the size of the cation (Table S2, Supporting Information). K-ZrP. K-ZrP is the potassium-exchanged form of α-ZrP. The crystal structure of K-ZrP was first determined from powder Xray diffraction52 and later refined by neutron diffraction data (shown in Figure 1b).41 It crystallizes in space group P2/c with one Zr, one K, and two P sites. The powder XRD pattern (Figure S1, Supporting Information) confirms that the K+ cations fully replaced the H+ ions of the α-ZrP.25 The 31P MAS spectrum (Figure S2, spectrum a, Supporting Information) indicates two signals with a 2:1 ratio, consistent with the proposed structure. Figure 2 illustrates the 91Zr SSNMR spectra of K-ZrP at 21.1 T. Spinning the sample at the magic angle averages out the 91Zr chemical shift anisotropy (CSA). Thus, the line shape seen in the MAS spectrum is due to the residual second-order quadrupolar interaction. The spectrum can be well simulated with a single Zr site, yielding the following EFG parameters: CQ = 7.0(2) MHz; ηQ = 0.60(10); δiso = −385(5) ppm. The 91Zr static spectra were acquired by using both WURST-QCPMG and quadrupolar echo sequences (Figure 2). The advantage of the WURST-QCPMG sequence can be seen clearly as the spectrum acquired in about 1 h is comparable to that obtained by utilizing the conventional echo sequence with an acquisition time of 18 h. The static spectra cannot be well simulated if only the EFG parameters are used, indicating that the chemical shielding (CS) interaction also contributes to the observed C



Article

smaller in the optimized structure. The shortest and longest Zr−O bond distances in the neutron structure are 1.912 and 2.213 Å, and these values became 2.035 and 2.099 Å in the optimized structure. The existence of very short (1.912 Å) and long (2.213 Å) Zr−O bonds in the neutron structure is likely the reason why the CQ calculated by using the neutron structure is overestimated remarkably. Perhaps, the optimized data better describe the true structure. As an alternative to CASTEP, one can also use ab initio methods, such as hybrid DFT calculations (using the Gaussian program) on model clusters truncated from the framework, in order to calculate the NMR tensors of a particular nucleus. This approach is computationally more expedient and has been successfully applied to various materials with two- and threedimensional structures.45,47,49,57 Previous studies have shown that the EFG tensor of a metal ion can be calculated reasonably well when the second and third coordination spheres around the metal center are considered.45−49 To test if the cluster approach can be used as a viable alternative when CASTEP calculations are practically not feasible, we built a cluster (Figure 1b-ii), Zr(OPO3)614−, constructed base on the K-ZrP structure. In this cluster, the center Zr atom is connected to six tetrahedral PO43− groups, and all the geometrical parameters are taken from the structure refined from neutron diffraction data. The calculated CQ value was 27.9 MHz, which is in good agreement with the overestimated value (30.2 MHz) predicted by the CASTEP calculation using the same structure. When the coordinates from the CASTEP-optimized structure were used, the Gaussian calculation on the cluster yielded a CQ value of 8.16 MHz. This result confirms that the optimized structure is indeed a better representation of the true structure. The fact that both the CASTEP and the Gaussian cluster calculations yield similar EFG tensor parameters suggests that, for the ZrP system, when the CASTEP calculation cannot be performed due to the constraint of computational resources, the cluster model approach may provide comparable results. Li-ZrP and Co-ZrP. The powder XRD pattern of Li-ZrP (lithium-exchanged form of α-ZrP) confirms that the H+ ions have been fully replaced by the Li+ cations32 (Figure S1, Supporting Information). The structure of Li-ZrP is unknown. Thus, SSNMR experiments were carried out to obtain structural information. 31P MAS and 6/7Li MAS spectra (Figures S2b and S3, Supporting Information) show one P and one Li signal, indicating only one unique crystallographic P and Li site. To characterize the local Zr environment, we acquired 91Zr SSNMR spectra at 21.1 T (Figure 3). The MAS spectrum

Figure 2. 91Zr NMR spectra of K-ZrP at 21.1 T.

static spectra. Indeed, when the CS tensor parameters [Ω = 200(10) ppm; κ = −0.9(1); β = 80(5)°; α = γ = 0] were included, the powder pattern can be fitted very well. The attempt in acquiring 91Zr static spectra with good quality for simulation at 9.4 T was not successful. The values of both asymmetry parameters (ηQ) and skew (κ) suggest that the EFG and CS tensors are not axial symmetrical, which is consistent with the Zr sitting at the general position.41 The isotropic chemical shift value of K-ZrP is almost identical to that of α-ZrP.11 However, the value of CQ(91Zr) for K-ZrP is significantly larger compared with that of the parent αZrP (5.8 MHz), indicating that there is a larger distortion on the ZrO6 octahedral unit. A quick inspection of the structural data (Table S3, Supporting Information) reveals that the degrees of ZrO6 distortions in K-ZrP are indeed much larger than those in α-ZrP.53 For example, the ranges of Zr−O bond distance are 1.912−2.213 and 2.049−2.075 Å for K-ZrP and αZrP, respectively. The variations in the O−Zr−O bond angle are 81.0−102.8 and 88.9−91.2° for K-ZrP and α-ZrP, respectively. To better understand the effect of distortion of local symmetry on the EFG at the Zr sites, computational studies were also carried out. The quadrupolar interaction is a groundstate property, which is proportional to the inverse cube of the separation between the nucleus of interest and charge density contributing to the EFG, therefore, reflecting the local structure and symmetry at the nucleus probed. Recently, the gaugeincluding projector augmented wave (GIPAW) method has been implemented in the CASTEP code.36 It employs periodic boundary conditions to fully account for the effects of the crystal lattice and can be used to calculate the EFG tensors in periodical solids. As shown in recent years, the GIPAW method is indeed a powerful tool to predict NMR properties of solids and should be utilized wherever possible.54−63 The CASTEP calculations were performed by using the proposed crystal structure.41 Surprisingly, the calculated CQ value (30.2 MHz) is drastically larger than the experimental one (7.0 MHz). To examine if the discrepancy might result from the inaccuracy in the crystal structure, which was determined from neutron rather than single-crystal X-ray diffraction data, the geometry of all the atoms in the unit cell is optimized, and the results are listed in Table S3 (Supporting Information). We then used the optimized structure to perform the CASTEP calculation, which predicts a CQ (=8.40 MHz) and a ηQ (=0.51 MHz). These values are now in a much better agreement with the measured EFG parameters. A comparison of the optimized ZrO6 geometry with that obtained from neutron diffraction (Table S3, Supporting Information) reveals that the Zr−O bond length and O−Zr−O bond angle distributions are much

Figure 3. 91Zr NMR spectra of Li-ZrP at 21.1 T. The number sign (#) indicates a small of impurity observed in the MAS spectrum. D



Article

Figure 4. (a) 91Zr NMR spectra of Co-ZrP at 21.1 T (see text for the assignment of Zr1 and Zr2 sites). (b) A diagram showing the proposed scheme in the intercalation of Co(NH3)63+ ions within the layers of α-ZrP, which simply expands the interlayer spacing.

The structure of hexaamminecobalt(III)-exchanged α-ZrP (Co-ZrP) is also unknown. For the large ions, the extent of ion exchange depends not only on the size of the ions but also on their heat of hydration.64,65 The larger ions replace the protons of α-ZrP at very slow rates due to the high activation energy required for the expansion of the interlayer region.2,64,65 The powder XRD pattern of Co-ZrP is consistent with that reported in the literature27 and indicates that the sample is a mixture of Co-ZrP and the parent α-ZrP phase (Figure S1, Supporting Information). As reported in the literature,27 within the Co-ZrP phase, only about ∼30% of H+ ions were replaced by Co(NH3)63+ and there are unexchanged 3(Zr−O−)P−OH units. The 31P MAS spectrum in Figure S2c (Supporting Information) shows four distinct resonances. The peak at −20.5 ppm with a shoulder at −21.1 ppm is assigned to 3(Zr− O−)P−OH units with the −20.5 ppm resonance belonging to the parent α-ZrP and the shoulder at −21.1 ppm originating from the unexchanged 3(Zr−O−)P−OH unit in the new CoZrP phase. The other two signals at −22.1 and −22.8 ppm are assigned to the unequivalent 3(Zr−O−)P−O− in the Co-ZrP phase. Figure 4a displays 91Zr SSNMR spectra of Co-ZrP at 21.1 T. The MAS spectrum can be fitted using two Zr sites with the following EFG parameters. Zr1 site: CQ = 5.8(2) MHz; ηQ = 0.27(10); δiso = −385(5) ppm. Zr2 site: CQ = 6.0(2) MHz, ηQ = 0.20(10), δiso = −432(5) ppm. On the basis of the relative intensity and chemical shift, the Zr1 site (labeled with #) is assigned to the Zr site in the parent α-ZrP phase as well as the Zr centers associated with P−OH groups in the Co-ZrP where the acidic protons are not replaced by Co(NH3)63+ [i.e., the 3(Zr−O)P−OH environment]. The Zr2 site is due to the − environment in the Co-ZrP. No CSA was 3(Zr−O)P−O necessary for the fitting of the static spectra, suggesting that the line shape is dominated by the quadrupolar interaction. It is interesting to note that the CQ and ηQ for both sites are very similar, indicating that the geometry of the ZrO6 unit in Co-ZrP is very similar to that in α-ZrP. It appears that the large Co(NH3)63+ ions simply expand the α-ZrP galleries without significantly distorting the Zr local environments (Figure 4b). Figure S4 (Supporting Information) compares all the static 91 Zr SSNMR spectra of α-ZrP and its ion-exchanged derivatives studied in this work and reported in the literature at 21.1 T.11

shows a major resonance due to Li-ZrP and a small amount of impurity. Simulation of the MAS spectrum led to the following EFG parameters: CQ = 7.6(2) MHz; ηQ = 0.80(10); and δiso = −380(5) ppm. The static spectra were also acquired using quadrupolar echo and WURST-QCPMG sequences, and the presence of the minor impurity does not affect the appearance of the static spectra as the signal belonging to Li-ZrP is much broader. The static spectra can also be well fitted by a single Zr site using the same set of EFG parameters without the inclusion of the CSA. 91Zr NMR data suggest that Li-ZrP only has a single Zr site. The CQ of Li-ZrP is slightly larger than that of KZrP, implying a more distorted ZrO6 octahedron in the lithiumexchanged phase. The fact that the CQ of Li-ZrP is much larger than that of the parent α-ZrP (5.8 MHz) indicates that the Zr local environment (and, therefore, the layer) undergoes significant distortion upon ion exchange. Previous studies have shown that small alkali cations, such as Na+, K+, and Li+, are able to exchange with the protons of α-ZrP at acidic pH and a high rate.23−25 The mechanism is thought to occur in two stages:26 (i) At the surface of the crystal, the cations, such as Li+, Na+, and K+, initially undergo dehydration or partial dehydration in order to enter the space between the layers. They then displace the protons from the P−OH groups, which are hydrogen-bonded to the water molecules occluded between the layers. The H+ ions then bind to H2O molecules, forming H3O+, and diffuse out of the lattice. (ii) The H2O molecules diffuse back into the crystal lattice, resulting in subsequent rehydration of the cations. It is at this stage that an expansion in the interlayer distance occurs and the distortion in the layer often leads to the formation of a new phase. In the present case, the small Li+ ions are highly polarizing and, therefore, interact with 3(Zr−O)P− O− groups strongly. Such interaction is evident from the 31P MAS NMR spectrum. Upon ion exchange, the 31P isotropic chemical shift changed from −21 ppm to the more deshielded direction by 6 ppm. It appears that the strong 3(Zr−O−)P− O−···Li+ interaction leads to the distortion of the ZrO6 unit. Furthermore, the small Li+ ion also has a strong ability of hydration. Thus, the crystal lattice must undergo further distortion to accommodate highly hydrated Li+ ions, resulting in a larger CQ. E



Article

The spectra are mostly dominated by the second-order quadrupolar interaction and sensitive to the Zr local geometry. The observed CQ(91Zr) values are in the following order: α-ZrP < Co-ZrP < Na-ZrP(site 2) < K-ZrP < Li-ZrP < Na-ZrP(site 1) < NH4-ZrP. Although there appears to be no obvious relationship between the CQ(91Zr) and the ionic radius, both the CQ(91Zr) and the isotropic shift values do empirically correlate with the structural parameters describing the distortion of ZrO 6 octahedra (Figure S5, Supporting Information). These results indicate that the different cations affect Zr NMR spectra via distortion of the ZrO6 octahedra. Zirconium Phosphates Prepared by Ionothermal Synthesis. In the second part of this work, several novel two- and three-dimensional zirconium phosphate and phosphate fluoride (ZrPOF) systems were examined. Most zirconium phosphate materials in the early days were synthesized hydrothermally or solvothermally. Recently, ZrPs have been synthesized ionothermally, and some novel two- and three-dimensional materials have been reported.28,30,31,33,34 These new materials have shown important applications. A few recent examples include catalysis for selective oxidation of cyclohexane,28 hydrogen storage,29 photoluminescence,30 and selective gas adsorption of CO2/CH4.31 The local Zr environments in seven novel zirconium phosphates and phosphate fluorides are probed directly by 91 Zr SSNMR. The crystal structures of these materials are known, except for ZrPOF-DEA (Table S4, Supporting Information). Theoretical calculations were also carried out to assist in spectral interpretation and assignments. ZrPO4-DES8. [C6H16N2]0.5Zr(H0.5PO4)2·H2O (designated as ZrPO4-DES8) is a novel layered zirconium phosphate prepared ionothermally in a deep eutectic mixture of oxalic acid and tetrapropyl ammonium bromide with 1,4-dimethylpiperazine (DMPIP) as a structure-directing agent.33 The material exhibits excellent tribological properties.31 Its inorganic layer is similar to that found in α-ZrP. Each Zr atom is coordinated to six oxygen atoms from six phosphate groups. For each [PO4] group, three oxygen atoms bridge three Zr atoms and the fourth oxygen exists in the form of P−O− and P−OH, interacting with the protonated DMPIP and water molecules, respectively. The structure can be described as pillared α-ZrP (Figure 5a) with DMPIP groups acting as the guest molecules. The crystal structure determined by single-crystal X-ray diffraction belongs to a space group of P21/c with one Zr and two P sites.33 The powder XRD pattern (Figure S6a, Supporting Information) matches that reported in the literature. The 31P MAS spectrum in Figure S7a (Supporting Information) shows two resonances with a 1:1 ratio at −22 and −25 ppm, consistent with the proposed structure. The 13C CP MAS spectrum (Figure S8a, Supporting Information) confirms the identity of the guest molecule and suggests that there is only one unique DMPIP molecule in the unit cell. Figure 6 shows 91Zr SSNMR spectra of ZrPO4-DES8 acquired at 21.1 T. The 91Zr MAS spectrum at 21.1 T was acquired by spinning the sample at 20 kHz in a commercial 3.2 mm ZrO2 rotor. The experiment was challenging due to the presence of the signal from the ZrO2 rotor (see Figure S9, Supporting Information). After background subtraction, the MAS spectrum (Figure 6a) shows a signal with a typical quadrupolar line shape. The resonance can be well fitted with the following EFG parameters: CQ = 7.5(2) MHz; ηQ = 1.0(1); and δiso = −407(5) ppm. Static quadrupolar echo and WURSTQCPMG spectra were also acquired in order to extract the CS

Figure 5. Framework structures of all the ZrPO/ZrPOF materials studied: (a) ZrPO4-DES8, viewed along b axis, (b) ZrPO4-DES1, (c) ZrPO4-DES2, viewed along b axis, (d) ZrPOF-pyr, viewed along c axis, (e) ZrPOF-Q1, viewed along b axis, (f) ZrPOF-EA. For (d) and (f), the SDA molecules are not shown for clarity. Blue balls are N atoms, gray ones are C atoms, and white ones are H atoms.

Figure 6. 91Zr NMR spectra of ZrPO4-DES8 at 21.1 T.

tensor parameters. Even though the static spectra are mainly dominated by the second-order quadrupolar interaction, the relatively small contribution of the CSA (Ω = 90(10) ppm; κ = 0.9(1)) cannot be neglected as it is needed to fit the experimental spectra. An attempt in acquiring a static spectrum at a lower field of 9.4 T was not successful. The CASTEP calculations based on the reported crystal structure33 yielded the following EFG parameters: CQ = 5.70 MHz and ηQ = 0.67 (Table 1). The predicted EFG parameters are in relatively good agreement with the experimental values, although CQ is slightly underestimated. The positions of the hydrogen atoms in ZrPO4-DES8 were further optimized, but the optimization did not improve the agreement between experimental and calculated values. DFT calculation on the model cluster Zr(OPO3)614− constructed by using the coordinates from the crystal structure (shown in Figure 7a) yields a CQ value of 7.07 MHz, agreeing well with the experimental value. F



Article

in the space group of P1.̅ 28 The unit cell only has one crystallographically nonequivalent site for Zr, P, and F. The powder XRD pattern (Figure S6b, Supporting Information) indicates a highly crystalline sample. The 31P MAS spectrum (Figure S7b, Supporting Information) illustrates only one major P signal, consistent with the proposed structure. The 19F MAS spectrum (Figure S10a, Supporting Information) also displays one isotropic signal at −23 ppm belonging to the terminal F−Zr site.15,66 Figure 8 displays 91Zr MAS and static spectra of ZrPO4DES1 acquired at 14.1 and 21.1 T. The MAS spectrum

Table 1. Summary of the Calculated and Experimental NMR Parameters for the ZrPs with One Zr Sitea,b,c,d,e

a

Basis set used for Gaussian cluster calculations: (17s11p8d)[12s7p4d] for Zr atoms and 6-311G* for other atoms. The EFG tensor is described by three principal components VJJ (J = X − Z), ordered such that |VZZ| ≥ |VYY| ≥ |VXX|. bCQ = eQVZZ/h. cηQ = (VXX − VYY)/VZZ. dThe equation used was δiso = (1545 − σiso) ppm, with 1545 ppm corresponding to the difference between the experimental shift value (−385 ppm) and the calculated shielding value (1930 ppm) of α-ZrP. eThe values in red indicate the ones in best agreement with the experimental values.

Figure 8. 91Zr NMR spectra of ZrPO4-DES1 at 14.1 and 21.1 T.

acquired at 21.1 T exhibits a sharp single resonance. This peak is very symmetric and narrow with a full width at half-height (fwhh) of about 650 Hz, indicating a very small quadrupolar interaction experienced by the Zr. To better determine the CQ value, the static echo spectra were recorded at 14.1 and 21.1 T. Figure 8 shows that the total width of the static powder pattern approximately scales inversely with the magnetic field, indicating that the pattern is not from a distribution of chemical shift or quadrupolar interaction and that, although the CQ is small (2.5 MHz), the spectrum is still dominated by the second-order quadrupolar interaction. The simulations yielded one set of the EFG parameters: CQ = 2.5(2) MHz; ηQ = 0.60(10); δiso = −371(5) ppm. The small CQ implies that the EFG at the Zr must be very small. However, the CASTEP calculations based on the known crystal structure consistently predict a significantly larger CQ value (10.53 MHz). Optimizing the structure did not improve the agreement in CQ with the experimental value (see Table 1). It should be pointed out that the crystal structure used for calculations was determined at 93 K.28 Therefore, the discrepancy between the measured and the calculated CQ suggests that the crystal structure at room temperature might be different from that at 93 K. To further probe the difference in the local Zr environment between the room temperature and the low-temperature structures, we carried out the DFT calculations of the 91Zr NMR tensors on a series of ZrF2(OPO3)410− clusters with slightly different geometries. The geometry of the initial cluster (Figure 7b) was constructed based on the low-temperature structure with the following parameters (Table S3, Supporting Information): Zr−O bond distances, 2.052 (×2), 2.068 (×2) Å; Zr−F bond distances, 2.009 (×2) Å; O−Zr−O bond angles, 88.5(×2)°, 91.5(×2)°; and O−Zr−F bond angles, 87.9(×2),

Figure 7. Different model clusters used for Gaussian calculations: (a) Zr(OPO3)614−, (b) trans-ZrF2(OPO3)410−, (c) ZrF(OPO3)512−, (d) cisZrF2(OPO3)410−, and (e) ZrF4(OPO3)26−.

ZrPO4-DES1. [NH4]4Zr(PO4)2F2 (denoted as ZrPO4-DES1) is a novel zirconium phosphate fluoride that is synthesized in a deep eutectic mixture of urea and tetramethyl ammonium chloride. It has a chain framework structure made up of alternating octahedral [ZrO4F2] and tetrahedral [PO4] units with a F/Zr ratio of 2. The four equatorial O atoms in each [ZrO4F2] octahedron connect to four P atoms from four different [PO4] units. The two F atoms are trans to each other (Figure 5b). For each [PO4] unit, two oxygen atoms are bound to P in the form of a terminal PO bond and the other two oxygen atoms in the P−O bonds bridge Zr atoms. It crystallizes G



Article

Figure 9. Calculated 91Zr CQ values as a function of (a) two Zr−F1 and (b) two Zr−O1 bond distances and (c) two F1−Zr−O1 and (d) two O2− Zr−O1 bond angles. Dotted lines indicate crystallographic value.

in F1−Zr−O1 angle from the initial value of 92.1° to 90.3° leads to a change in CQ(91Zr) by 1.33 MHz, and similarly, a 2% decrease in the O2−Zr−O1 angle from 88.5° to 86.7° results in a very small increase in CQ(91Zr) by only 0.60 MHz. It appears that the discrepancy between the measured and the calculated CQ is likely due to the differences in the Zr−F and Zr−O bond lengths between the room and the low-temperature structure. We further suggest that the smaller CQ(91Zr) measured at room temperature results mainly from slightly shorter Zr−O distances at room temperature since the CQ (3.28 MHz) calculated with a shorter Zr−O bond distance of 2.027 Å is comparable to the measured value (2.5 MHz) at room temperature. ZrPO4-DES2. Using the same eutectic mixture, but a different F/Zr ratio of 0.61, another layered zirconium phosphate fluoride, [NH4]3Zr(PO4)2F (herein referred to as ZrPO4DES2), can also be synthesized. The crystal structure determined at 93 K shows that, within the layer, there are eight- and four-membered rings formed from alternating [ZrO5F] octahedra and [PO4] tetrahedra (Figure 5c). For each [ZrO5F], zirconium is bound to one terminal F and five bridging O atoms. The P in each [PO4] unit forms two terminal PO bonds and is connected to Zr atoms via P−O−Zr linkages. The ammonium ions located between the layers result from the decomposition of urea. ZrPO4-DES2 crystallizes in the space group of C2 with one Zr and two P sites.28 The powder XRD (Figure S6c, Supporting Information), 31P MAS NMR (Figure S7c, Supporting Information), and 19F MAS NMR (Figure S10b, Supporting Information) are in agreement with the crystal structure. The solid-state 91Zr NMR spectra of ZrPO4-DES2 acquired at 21.1 T are shown in Figure 10. The line shapes of the spectra suggest a slight disordering in the local Zr environment. Such disordering might be due to that, at room temperature, the distribution of NH4+ ions is disordered. Since the EFG is a long-range effect, such disordering will contribute to the overall observed line shape. Nonetheless, both the MAS and the static spectra can readily be fitted with a single Zr site with the following EFG parameters: CQ = 7.8(2) MHz; ηQ = 0.65(10); and δiso = −350(5) ppm. The isotropic chemical shift value of

89.5(×2), 90.5(×2), 92.1(×2)°. Similar to the CASTEP calculation, the Gaussian calculation on the cluster with the initial geometry of the crystal structure predicted a much larger CQ (8.48 MHz). The reported crystal structure exhibits an inversion center at the Zr atom, which is confirmed by the 31P and 19F NMR spectra acquired at room temperature. Therefore, the initial cluster (Figure 9) was modified in such a way that the inversion center at Zr is preserved. Specifically, four structural parameters were varied separately by simultaneously changing (1) two Zr−F1 bonds, (2) two Zr−O1 bonds trans to each other, (3) two F1−Zr−O1 bond angles, and (4) O2−Zr−O1 bond angles by the same amount. Figure 9a shows the variation of the magnitude of CQ as a function of Zr−F1 bond length (see Table S5, Supporting Information, for the data shown in Figure 9). The plot exhibits a V-shaped curve with its minimum occurring at 1.998 Å (0.011 Å shorter than the crystallographic distance at 2.009 Å). The CQ is highly sensitive to the Zr−F distance. For instance, if the Zr−F1 bond lengths are decreased by only about 0.011 Å from their crystallographic value of 2.009 to 1.998 Å, the CQ decreases from 8.48 to 6.34 MHz. When two Zr−O1 bonds are stretched or compressed simultaneously by the same degree, a similar V-shaped curve was also obtained (Figure 9b). The CQ value corresponding to the crystallographic Zr−O1 distance (2.068 Å) also does not coincide with the minimum point on the curve. Figure 9c,d shows the calculated CQ(91Zr) values as a function of F1−Zr−O1 and O−Zr−O bond angles, but the angles corresponding to the crystallographic values coincide with the minima of the curves. Overall, the calculations indicate that Zr−F and Zr−O distances as well as F−Zr−O and O−Zr−O angles all contribute to the observed CQ(91Zr). However, the effect of both the Zr−F and the Z−O bond lengths is much larger than that of the bond angles. For instance, a 2% decrease in the Zr− F1 distance from its initial value of 2.009 to 1.969 Å leads to a significant increase in the CQ(91Zr) by 4.62 MHz (from 8.48 to 13.10 MHz). Similarly, changing the Zr−O1 distance by 2% from the crystallographic value of 2.068 to 2.027 Å results in a comparably large change in CQ(91Zr) from 8.48 to 3.28 MHz (ΔCQ = 5.20 MHz). The influences of F−Zr−O and O−Zr−O bond angles, on the other hand, are much smaller: a 2% change H



Article

Figure 10. 91Zr NMR spectra of ZrPO4-DES2 at 21.1 T.

−350 ppm is the most deshielded of all previously investigated systems in this work. The coordinates of the protons from the guest ammonium ions were not resolved in the structure. Hence, they are not available for calculations. To perform the CASTEP calculation, we added H atoms to the N atom to form ammonium ions and subsequently optimized the geometry. CASTEP calculations on this structure (Table 1) yielded a CQ(91Zr) of 8.4 MHz, which is in a reasonably good agreement with the measured value. The Gaussian calculation on the ZrF(OPO3)512− model cluster (Figure 7c) slightly overestimates the CQ (10.5 MHz). Like ZrPO4-DES1, the difference in CQ may be attributed to the fact that experimentally obtained CQ is measured at room temperature and the calculated CQ is based on the crystal structure determined at low temperature (93 K). Unlike ZrPO4-DES1, this difference is fairly small, indicating that the overall Zr local structure at room temperature is close to that at 93 K. ZrPOF-pyr. [(C5H6N)4(H2O)2]Zr12P16O60(OH)4F8 (designated as ZrPOF-pyr; pyr = pyridine) has a novel threedimensional open framework structure with 10-ring channels (Figure 5d), made up from alternating octahedral [ZrO6] and tetrahedral [PO4] units. Pyridine serves as a structure-directing agent and is occluded inside the channels of the as-synthesized material. The refinement of the high-resolution synchrotron powder XRD data reveals a Pnnm space group with three different Zr-centered octahedra (ZrO6, ZrO5F, and ZrO4F2), four unique P, and two F sites.34 One F atom is shared between ZrO5F and ZrO4F2 octahedra, while the other terminal F atom is pointing toward the 10-ring channel. The powder XRD pattern (Figure S6d, Supporting Information) confirms the identity of the sample used; 31P, 13C, and 19F MAS spectra (Figures S7d, S8b, and S10c, Supporting Information) show that there are four P sites, one nonequivalent pyridine molecule, and two F sites. Figure 11a displays static 91Zr WURST-QCPMG and static echo NMR spectra of ZrPOF-pyr at 21.1 T. The breadth of the powder pattern spans more than 200 kHz, the widest one observed in this study. Such a broad pattern precludes us from acquiring any meaningful MAS spectrum. The broad pattern likely results from the overlapping of three chemically nonequivalent Zr sites. Fitting the spectra with three Zr sites was impossible due to the large number of NMR parameters required for simulation. For the computational resource available to us, the size of the unit cell of ZrPOF-pyr (1920 Å3) is too large to perform CASTEP calculations. Alternatively, DFT calculations on the model cluster constructed from the crystal structure were carried out. The model clusters used were [Zr(OPO3)614−], [ZrF2(OPO3)410−], and [ZrF(OPO3)512−] for Zr1, Zr2, and Zr3

Figure 11. 91Zr static NMR spectra of (a) ZrPOF-pyr and (b) ZrPOFQ1 at 21.1 T.

sites, respectively (Figure 7a,d,c). The clusters for Zr1 and Zr3 are similar to those used earlier for ZrPO4-DES1 and ZrPO4DES2, and the two F atoms in the cluster for Zr2 are cis to each other. The calculated CQ values for Zr1, Zr2, and Zr3 are 36.00, 12.09, and 22.53 MHz, respectively (Table 2). As demonstrated earlier, the CQ(91Zr) is highly sensitive to the Zr−F and Zr−O distances. The presence of a very short Zr1−O bond (1.923 Å) is likely the main reason why the calculated CQ value of Zr1 is the largest. The calculated spectrum with all three sites in a 1:1:1 intensity ratio is shown in Figure 11a. It seems that the Table 2. Summary of the Calculated NMR Parameters for All Systems Having Multiple Zr Sitesa,b,c,d,e

a

Basis set used for Gaussian cluster calculations: (17s11p8d)[12s7p4d] for Zr atoms and 6-311G* for other atoms. The EFG tensor is described by three principal components VJJ (J = X − Z), ordered such that |VZZ| ≥ |VYY| ≥ |VXX|. bCQ = eQVZZ/h. cηQ = (VXX − VYY)/VZZ. dThe equation used was δiso = (1545 − σiso) ppm, with 1545 ppm corresponding to the difference between the experimental shift value (−385 ppm) and the calculated shielding value (1930 ppm) of α-ZrP. eThe numbers in purple indicate the Zr sites that were observed experimentally. I



Article

belonging to bridging F−Zr−F sites), consistent with the proposed structure. Figure 12a shows MAS, static WURST-QCPMG, and quadrupolar echo NMR spectra of ZrPOF-EA acquired at

observed pattern mainly contains the signals due to Zr2 and Zr3. The pattern of Zr1 is likely too broad to be observed as it spreads over in a very wide frequency range. The line shape also indicates a certain degree of disordering in the Zr environments. Such disordering likely reflects the fact that the OH groups connected to the neighboring P atoms in the second coordination sphere are disordered.34 ZrPOF-Q1. [(C9H8N)4(H2O)4]Zr8P12O40(OH)8F8 (denoted as ZrPOF-Q1; Q = quinoline) has a novel layered structure. The crystal structure is resolved by powder XRD data.30 The layer is somewhat similar to that of α-ZrP; that is, the [ZrO6] units are linked to one another via corner sharing [PO4] tetrahedra to form a sheet with terminal P−OH on both sides. However, the unique feature of this material is that there is an additional isolated [ZrO2F4] octahedral unit anchored to the phosphate outer layers (Figure 5e). The closest distance between two F atoms in two adjacent layers is only 3.1 Å, making it a “pseudo” three-dimensional framework.30 The inorganic zirconium phosphate layers are separated by layers of quinolinium ions and water molecules. There are four unique Zr sites. The static 91Zr WURST-QCPMG and quadrupolar echo NMR spectra of ZrPOF-Q1 at 21.1 T are illustrated in Figure 11b. The spectrum does not display a distinct quadrupolar line shape, resulting from overlapping of four Zr sites. Similar to ZrPOF-pyr’s case, the CASTEP calculation was not possible due to its large unit cell. Gaussian model cluster calculations were then performed on isolated Zr octahedral units for each site. The model cluster used for calculation of Zr1, Zr2, and Zr3 sites is shown in Figure 7a, while Figure 7e displays the [ZrO2F4] octahedral unit for Zr4. The geometry for each cluster is based on the crystal structure. Table 2 summarizes all the calculated CQ values, which reveals that they are relatively similar to one another. Figure 11b compares the calculated spectra based on three different scenarios: (i) only two sites, Zr1 and Zr4; (ii) three sites, Zr1, Zr4, and Zr2; and (iii) all four Zr sites were observed. It seems that case (ii) is the most likely scenario as it gives a better agreement with the measured spectrum; that is, the observed pattern is due to Zr1, Zr2, and Zr4 sites. The absence of Zr3 is likely due to the fact that its pattern is too broad (CQ = 16.34 MHz). ZrPOF-EA and ZrPOF-DEA. Using a deep eutectic mixture of oxalic acid and ethyl ammonium chloride, a novel threedimensional open framework zirconium phosphate, [(C2H8N)8(H2O)8]Zr32P48O176(OH)16F8 (herein referred to as ZrPOF-EA, EA = ethyl ammonium), was prepared.31 This porous material is a promising size-selective molecular sieve for CO2/CH4 separation compared to other conventional eightring pore materials due to its pore structure and polar −OH group directed into the pore channels.31 The structure, solved by high-resolution powder XRD data, consists of [41482] units arranged in a rectangular array connected by [ZrO6], [ZrO5F], and [PO4] units.31 There are eight unique crystallographic Zr sites: four from [ZrO6] octahedra and another four from [ZrO5F] units in a very large unit cell (4895 Å3). The powder XRD pattern (Figure S6f, Supporting Information) matches the one reported in the literature well, and the 13C MAS spectrum (Figure S8d, Supporting Information) shows that there are two unique ethyl ammonium ions in the unit cell. The 31P MAS spectrum shows several overlapping signals, and the observed broad profile is due to the distribution of all 12 P sites (Figure S7f, Supporting Information). The 19F spectrum clearly shows two isotropic peaks (Figure S10e, Supporting Information, both

Figure 12. 91Zr NMR spectra of (a) ZrPOF-EA and (b) ZrPOF-DEA at 21.1 T.

21.1 T. The static spectra clearly show a distribution of quadrupolar coupling constants across a relatively small frequency range of ∼60 kHz due to the overlap of multiple sites. No attempt was made to simulate the observed spectra as there are too many parameters. Table 2 lists all calculated EFG parameters for all eight Zr sites. Similar to the case of ZrPOFQ1, calculated spectra resulting from overlapping of different numbers of Zr sites are presented in Figure 12a. It seems that only Zr3, Zr7, and Zr8 sites were observed experimentally as the others were too broad and, therefore, less visible. When diethyl ammonium chloride was used instead of ethyl ammonium chloride in the eutectic mixture for synthesis, another framework structure, ZrPOF-DEA (DEA = diethyl ammonium), was produced. The structure of this material has not been solved yet. The powder XRD pattern and 31P MAS spectrum of ZrPOF-DEA (Figures S6g and S7g, Supporting Information) are similar to those of ZrPOF-EA, indicating that their framework structures are similar. The 13C MAS spectrum of ZrPOF-DEA in Figure S8e (Supporting Information) shows the presence of diethyl ammonium ions inside the framework. The 19F MAS spectrum of ZrPOF-DEA (Figure S10f, Supporting Information) also shows two isotropic chemical shifts, but their values (−26 and −36 ppm) are slightly different than those of ZrPOF-EA (−29 and −39 ppm). Although the 91 Zr static WURST-QCPMG NMR spectrum of ZrPOF-DEA at 21.1 T shows a line shape very similar to that of ZrPOF-EA, the breadth of the pattern of ZrPOF-DEA is clearly broader (Figure 12b). The spectral profile indicates the overlapping of multiple resonances. Both PXRD and NMR data suggest that the structure of ZrPOF-DEA might be very similar to that of ZrPOF-EA. On the basis of the solid-state NMR data, we further suggest that (i) there are multiple Zr sites present, and they are in a more distorted environment compared to those of ZrPOF-EA; (ii) there are at least four unique crystallographic P J



Article

nmr900.ca). We thank Drs. Victor Terskikh and Eric Ye for technical assistance and performing CASTEP calculations. This work was made possible by the facilities of the Shared Hierarchical Academic Research Computing Network (SHARCNET: www.sharcnet.ca). We also thank The University of Western Ontario (UWO) for a small ADF grant and The Centre for Advanced Materials and Biomaterials Research at UWO for a small grant for international collaboration. The Chinese group was financially supported by the National Natural Science Foundation (Grant Nos. 20825623, 20801039), the Program for the Top Young Academic Leaders and for the Top Science and Technology Innovation Teams of Higher Learning Institutions of Shanxi, and the Program for New Century Excellent Talents in University (NCET-111038).

sites; and (iii) there is only one nonequivalent DEA molecule and two F sites in the unit cell.

■

CONCLUSIONS We have directly characterized the local environment around the Zr center in several representative layered and threedimensional microporous zirconium phosphates by 91Zr solidstate NMR at a very high magnetic field of 21.1 T. The observed spectra are sensitive to the local Zr environments and mostly dominated by the second-order quadrupolar interaction. Computational studies of model clusters have shown that the geometric parameters around Zr centers, such as Zr−F and Zr− O bond distances and F−Zr−O and O−Zr−O bond angles, all contribute to the observed CQ, but the Zr−F and Zr−O bond lengths are the more dominant factors. The results of SSNMR and CASTEP calculations have shown that the local Zr environment in K-ZrP is quite different from that in the neutron structure reported in the literature. For ZrPO4-DES1 and ZrPO4-DES2, 91Zr SSNMR data suggest that the crystal structures of these two materials at room temperature are different from those determined at 93 K. This is particularly true for ZrPO4-DES1 as the calculations on model clusters clearly indicate that the Zr−O and Zr−F bond distances in ZrPO4-DES1 are shorter at room temperature. For ZrPOF-pyr, ZrPOF-Q1, and ZrPOF-EA, where multiple Zr sites exist, the Gaussian calculations on model clusters were performed to assist in spectral assignments and interpretation. The crystal structures of several materials, including Li-ZrP, CoZrP, and ZrPOF-DEA, are unknown. 91Zr, 6/7Li, 31P, 13C, and 19 F NMR spectra were obtained to gain partial structural information.

■

■

(1) Murugavel, R.; Choudhury, A.; Walawalkar, M. G.; Pothiraja, R.; Rao, C. N. R. Chem. Rev. 2008, 108, 3549. (2) Auerbach, S. M.; Carrado, K. A.; Dutta, P. K. Handbook of Layered Materials; CRC Press: New York, 2004. (3) Zhu, J.; Huang, Y. Langmuir 2010, 26, 10115. (4) Zhu, J.; Huang, Y. Inorg. Chem. 2009, 48, 10186. (5) Pyykko, P. Mol. Phys. 2008, 106, 1965. (6) Bastow, T. J.; Smith, M. E. Solid State Nucl. Magn. Reson. 1992, 1, 165. (7) Bastow, T. J.; Smith, M. E.; Stuart, S. N. Chem. Phys. Lett. 1992, 191, 125. (8) Bastow, T. J. Z. Naturforsch., A: Phys. Sci. 1994, 49, 320. (9) Hung, I.; Schurko, R. W. J. Phys. Chem. B 2004, 108, 9060. (10) O’Dell, L. A.; Schurko, R. W. Chem. Phys. Lett. 2008, 464, 97. (11) Yan, Z.; Kirby, C. W.; Huang, Y. J. Phys. Chem. C 2008, 112, 8575. (12) Zhu, J.; Lin, Z.; Yan, Z.; Huang, Y. Chem. Phys. Lett. 2008, 461, 260. (13) Pauvert, O.; Fayon, F.; Rakhmatullin, A.; Kramer, S.; Horvatic, M.; Avignant, D.; Berthier, C.; Deschamps, M.; Massiot, D.; Bessada, C. Inorg. Chem. 2009, 48, 8709. (14) Rossini, A. J.; Hung, I.; Johnson, S. A.; Slebodnick, C.; Mensch, M.; Deck, P. A.; Schurko, R. W. J. Am. Chem. Soc. 2010, 132, 18301. (15) Fedotov, M.; Belyaev, A. J. Struct. Chem. 2011, 52, 69. (16) Lapina, O. B.; Khabibulin, D. F.; Terskikh, V. V. Solid State Nucl. Magn. Reson. 2011, 39, 47. (17) Larsen, F. H.; Jakobsen, H. J.; Ellis, P. D.; Nielsen, N. C. J. Phys. Chem. A 1997, 101, 8597. (18) Schurko, R. W.; Hung, I.; Widdifield, C. M. Chem. Phys. Lett. 2003, 379, 1. (19) Siegel, R.; Nakashima, T. T.; Wasylishen, R. E. Chem. Phys. Lett. 2004, 388, 441. (20) Harris, R. K.; Wasylishen, R. E.; Duer, M. J. NMR Crystallography; John Wiley & Sons Ltd: Chichester, U.K., 2009. (21) Ashbrook, S. E. Phys. Chem. Chem. Phys. 2009, 11, 6892. (22) Hanna, J. V.; Smith, M. E. Solid State Nucl. Magn. Reson. 2010, 38, 1. (23) Clearfield, A.; Duax, W. L.; Medina, A. S.; Smith, G. D.; Thomas, J. R. J. Phys. Chem. 1969, 73, 3424. (24) Clearfield, A.; Troup, J. J. Phys. Chem. 1970, 74, 314. (25) Clearfield, A.; Duax, W. L.; Garces, J. M.; Medina, A. S. J. Inorg. Nucl. Chem. 1972, 34, 329. (26) Kullberg, L.; Clearfield, A. J. Phys. Chem. 1981, 85, 1585. (27) Hasegawa, Y.; Kizaki, S.; Amekura, H. Bull. Chem. Soc. Jpn. 1983, 56, 734. (28) Liu, L.; Li, Y.; Wei, H.; Dong, M.; Wang, J.; Slawin, A. M. Z.; Li, J.; Dong, J.; Morris, R. E. Angew. Chem., Int. Ed. 2009, 48, 2206. (29) Dong, J.; Ban, G.; Zhao, Q.; Liu, L.; Li, J. AIChE J. 2008, 54, 3017.

ASSOCIATED CONTENT

* Supporting Information S

Detailed sample preparation and experimental SSNMR conditions, list of experimental interlayer spacings in ZrP and its derivatives, summary of relevant bond distances and angles, structural data for all ZrP/ZrPOF materials, calculated 91Zr CQ values of a zirconium phosphate model cluster for ZrPO4DES1, powder XRD spectra, and additional multinuclear (6/7Li, 13 C, 19F, and 31P) MAS NMR spectra at 9.4 T. This material is available free of charge via the Internet at http://pubs.acs.org.

■

REFERENCES

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected] (Y.H.). Tel: 519-661-2111, ext. 86384 (Y.H.). Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS Y.H. thanks the Natural Science and Engineering Research Council of Canada for a Discovery Grant and a Discovery Accelerator Grant, and the Canada Foundation for Innovation for an equipment grant. Funding from the Canada Research Chair program is also gratefully acknowledged. Access to the 900 MHz NMR spectrometer was provided by the National Ultrahigh Field NMR Facility for Solids (Ottawa, Canada), a national research facility funded by the Canada Foundation for Innovation, the Ontario Innovation Trust, Recherche Québec, the National Research Council Canada, and Bruker BioSpin and managed by the University of Ottawa (http://www. K



Article

(64) Alberti, G.; Casciola, M.; Costantino, U.; Vivani, R. Adv. Mater. 1996, 8, 291. (65) Clearfield, A.; Costantino, U. Compr. Supramol. Chem. 1996, 7, 107. (66) Ben Ali, A.; Body, M.; Leblanc, M.; Maisonneuve, V. Solid State Sci. 2011, 13, 394.

(30) Liu, L.; Li, J.; Dong, J.; Sisak, D.; Baerlocher, C.; McCusker, L. B. Inorg. Chem. 2009, 48, 8947. (31) Liu, L.; Yang, J.; Li, J.; Dong, J.; Sisak, D.; Luzzatto, M.; McCusker, L. B. Angew. Chem., Int. Ed. 2011, 50, 8139. (32) Alberti, G.; Constantino, U.; Allulli, S.; Massucci, M. A.; Pelliccioni, M. J. Inorg. Nucl. Chem. 1973, 35, 1347. (33) Liu, L.; Chen, Z.-F.; Wei, H.-B.; Li, Y.; Fu, Y.-C.; Xu, H.; Li, J.P.; Slawin, A. M. Z.; Dong, J. Inorg. Chem. 2010, 49, 8270. (34) Dong, J.; Liu, L.; Li, J.; Li, Y.; Baerlocher, C.; McCusker, L. B. Microporous Mesoporous Mater. 2007, 104, 185. (35) Eichele, K.; Wasylishen, R. E. WSOLIDS1: Solid-State NMR Spectrum Simulation, version 1.17.30; Universität Tübingen: Tübingen, Germany, 2001. (36) Clark, S. J.; Segall, M. D.; Pickard, C. J.; Hasnip, P. J.; Probert, M. I. J.; Refson, K.; Payne, M. C. Z. Kristallogr. 2005, 220, 567. (37) Segall, M. D.; Lindan, P. J. D.; Probert, M. J.; Pickard, C. J.; Hasnip, P. J.; Clark, S. J.; Payne, M. C. J. Phys.: Condens. Matter 2002, 14, 2717. (38) Pickard, C. J.; Mauri, F. Phys. Rev. B: Condens. Matter Mater. Phys. 2001, 63, 245101. (39) Profeta, M.; Mauri, F.; Pickard, C. J. J. Am. Chem. Soc. 2003, 125, 541. (40) Yates, J. R.; Pickard, C. J.; Mauri, F. Phys. Rev. B: Condens. Matter Mater. Phys. 2007, 76, 024401. (41) Rudolf, P. R.; Clearfield, A. Inorg. Chem. 1985, 24, 3714. (42) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996, 77, 3865. (43) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1997, 78, 1396. (44) Frisch, M. J.; et al. Gaussian 09, revision A.1; Gaussian, Inc.: Wallingford, CT, 2009. (45) Peng, L.; Liu, Y.; Kim, N.; Readman, J. E.; Grey, C. P. Nat. Mater. 2005, 4, 216. (46) Chapman, R. P.; Bryce, D. L. Phys. Chem. Chem. Phys. 2007, 9, 6219. (47) Cuny, J.; Messaoudi, S.; Alonzo, V.; Furet, E.; Halet, J.-F.; Le Fur, E.; Ashbrook, S. E.; Pickard, C. J.; Gautier, R.; Le Polles, L. J. Comput. Chem. 2008, 29, 2279. (48) Cadars, S.; Brouwer, D. H.; Chmelka, B. F. Phys. Chem. Chem. Phys. 2009, 11, 1825. (49) Dogan, F.; Hammond, K. D.; Tompsett, G. A.; Huo, H.; Conner, W. C.; Auerbach, S. M.; Grey, C. P. J. Am. Chem. Soc. 2009, 131, 11062. (50) Schaefer, A.; Horn, H.; Ahlrichs, R. J. Chem. Phys. 1992, 97, 2571. (51) Adiga, S.; Aebi, D.; Bryce, D. L. Can. J. Chem. 2007, 85, 496. (52) Clearfield, A.; McCusker, L. B.; Rudolf, P. R. Inorg. Chem. 1984, 23, 4679. (53) Troup, J. M.; Clearfield, A. Inorg. Chem. 1977, 16, 3311. (54) Griffin, J. M.; Wimperis, S.; Berry, A. J.; Pickard, C. J.; Ashbrook, S. E. J. Phys. Chem. C 2009, 113, 465. (55) O’Dell, L. A.; Schurko, R. W. Phys. Chem. Chem. Phys. 2009, 11, 7069. (56) Pallister, P. J.; Moudrakovski, I. L.; Ripmeester, J. A. Phys. Chem. Chem. Phys. 2009, 11, 11487. (57) Brouwer, D. H.; Moudrakovski, I. L.; Darton, R. J.; Morris, R. E. Magn. Reson. Chem. 2010, 48, S113. (58) Hanna, J. V.; Pike, K. J.; Charpentier, T.; Kemp, T. F.; Smith, M. E.; Lucier, B. E. G.; Schurko, R. W.; Cahill, L. S. Chem.Eur. J. 2010, 16, 3222. (59) Michaelis, V. K.; Kroeker, S. J. Phys. Chem. C 2010, 114, 21736. (60) Widdifield, C. M.; Bryce, D. L. J. Phys. Chem. A 2010, 114, 2102. (61) O’Dell, L. A.; Schurko, R. W.; Harris, K. J.; Autschbach, J.; Ratcliffe, C. I. J. Am. Chem. Soc. 2011, 133, 527. (62) Griffin, J. M.; Berry, A. J.; Ashbrook, S. E. Solid State Nucl. Magn. Reson. 2011, 40, 91. (63) Johnston, K. E.; Griffin, J. M.; Walton, R. I.; Dawson, D. M.; Lightfoot, P.; Ashbrook, S. E. Phys. Chem. Chem. Phys. 2011, 13, 7565. L


Solid-State 91Zr NMR Characterization of Layered and Three

Recommend Documents