Combining 27Al Solid-State NMR and First-Principles Simulations To

Dec 6, 2016 - Synopsis. The nuclear magnetic resonance (NMR) technique gives insight into the local information in a crystal structure, while Rietveld...
4 downloads 37 Views 3MB Size
Download PDF
Article pubs.acs.org/IC

Combining 27Al Solid-State NMR and First-Principles Simulations To Explore Crystal Structure in Disordered Aluminum Oxynitride Bingtian Tu, Xin Liu, Hao Wang,* Weimin Wang, Pengcheng Zhai, and Zhengyi Fu State Key Lab of Advanced Technology for Materials Synthesis and Processing, Wuhan University of Technology, Wuhan 430070, China S Supporting Information *

ABSTRACT: The nuclear magnetic resonance (NMR) technique gives insight into the local information in a crystal structure, while Rietveld refinement of powder X-ray diffraction (PXRD) sketches out the framework of a crystal lattice. In this work, first-principles calculations were combined with the solid-state NMR technique and Rietveld refinement to explore the crystal structure of a disordered aluminum oxynitride (γ-alon). The theoretical NMR parameters (chemical shift, δiso, quadrupolar coupling constants, CQ, and asymmetry parameter, η) of Al22.5O28.5N3.5, predicted by the gaugeincluding projector augmented wave (GIPAW) algorithm, were used to facilitate the analytical investigation of the 27Al magic-angle spinning (MAS) NMR spectra of the as-prepared sample, whose formula was confirmed to be Al2.811O3.565N0.435 by quantitative analysis. The experimental δiso, CQ, and η of 27Al showed a small discrepancy compared with theoretical models. The ratio of aluminum located at the 8a to 16d sites was calculated to be 0.531 from the relative integration of peaks in the 27Al NMR spectra. The occupancies of aluminum at the 8a and 16d positions were determined through NMR investigations to be 0.9755 and 0.9178, respectively, and were used in the Rietveld refinement to obtain the lattice parameter and anion parameter of Al2.811O3.565N0.435. The results from 27Al NMR investigations and PXRD structural refinement complemented each other. This work provides a powerful and accessible strategy to precisely understand the crystal structure of novel oxynitride materials with multiple disorder.

1. INTRODUCTION Owing to fascinating electronic, optical, and mechanical properties, oxynitride materials have found applications in photocatalysts, phosphors, and structural materials.1,2 Existing in the AlN−Al2O3 pseudobinary system, spinel-type aluminum oxynitride (γ-alon) solid solution has already exhibited versatile applications on both the military and commercial sides as an important transparent structural ceramic.3,4 Besides, functionalized γ-alon materials by doping rare earth or transition metal ions showed interesting photoluminescence properties, which have attracted lots of material scientists in recent years.5 The formula of γ-alon solid solution could be described as Al(8+x)/3O4−xNx.6,7 The face-centered cubic crystal structure of γ-alon with space group Fd3m has already been studied by powder X-ray diffraction (PXRD) and neutron diffraction techniques,8 even though the local structure of cationic vacancies and nitrogen atoms can hardly be fully specified by those diffraction approaches. Because disorder always plays an important role in many physical and chemical properties of materials, it is definitely crucial to carry out a detailed investigation of the nature of the disorders in γ-alon. Fang et al. and our previous theoretical investigations have concluded that the cationic vacancies tend to locate at octahedral sites, and the nitrogen atoms prefer to distribute far away from each other.9,10 So far, there is still a lack of experimental discussion © XXXX American Chemical Society

on this issue. In this paper, a combination strategy was proposed to further illuminate the local and long-range crystal structure of γ-alon. Considering that such a problem also exists and is not tackled in other disorder oxynitride crystals, including SiAlON,11 LaMgxTa1−xO1+3xN2−3x,12 GaON,13,14 and TaON solid solutions,15 this study presents a general and flexible scheme for understanding the crystal configuration of oxynitride crystals. Of the few available techniques which can provide information about the local environment of atoms in crystals, X-ray absorption spectroscopy (XAS) and nuclear magnetic resonance (NMR) are the two main choices. In contrast with XAS characterization, which requires the use of limited synchrotron facilities equipped with a soft X-ray beamline, NMR spectroscopy is a more common, feasible, and accurate method. As a local probe technique, NMR possesses the distinctly advantageous ability to investigate the positional and compositional nature of disorder in a solid. Generally, the changes of coordination number and the nearest atomic species will significantly result in a different chemical shift, which can be observed and resolved in NMR spectra.16 In disordered γalon, the partial occupancy of oxygen and nitrogen at anion Received: September 29, 2016

A

DOI: 10.1021/acs.inorgchem.6b02360 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry

Table 1. Simulated (PBEsol and LDA) and Experimental (Exp) 27Al NMR Parameters of Al2O3 (α, θ), AlVO4, LiAlO2 (α, γ), SiAl2O5 (andalusite, kyanite, and sillimanite), YAlO3, Y4Al2O9, and Y3Al5O12 σiso (ppm)

δiso (ppm)

η

CQ (MHz)

Crystal

CN

PBEsol

LDA

Exp

PBEsol

LDA

Exp

PBEsol

LDA

Exp

α-Al2O319 θ-Al2O320

6 4 6 6 6 6 6 4

534.88 470.99 540.57 563.8 528.33 556.92 531.39 469.05

524.57 455.54 530.51 552.8 516.16 546.51 522.04 454.16

18.8 80 10.5 −8.9 27.2 −1.1 17 83

2.281 6.494 3.479 1.001 7.162 −4.919 2.999 3.608

2.375 6.705 3.608 1.146 7.988 −4.711 3.039 3.863

2.4 6.4 3.5 1.64 6.73 5.88 2.8 3.2

0 0.4 0.18 0.77 0.58 0.48 0 0.76

0 0.47 0.12 0.26 0.48 0.4 0 0.77

0.01 0.65 0 0.3 0.42 0.58 0.05 0.70

6 5 6 6 6 6 4 6 6 4 4 6 4

540.12 515.58 539.38 547.39 544.41 540.12 489.45 547.96 542.74 475.63 477.09 550.36 476.68

530.56 502.51 529.38 537.26 534.43 530.18 474.37 538.19 533.29 461.95 463.45 541.56 462.1

13 35.5 14.9 4 7.7 11 63.9 4.7 10.7 78.2 76.2 2.1 77.5

15.24 5.442 10.58 4.14 −6.958 −10.3 −6.418 −8.652 1.809 10.71 10.31 −0.8775 −6.863

15.23 5.96 10.76 4.159 −6.727 −10.09 −6.615 −8.537 1.84 11.25 10.71 −0.9054 −6.964

15.3 5.8 10.1 3.6 6.6 9.2 6.74 8.83 1.61 10.81 10.36 1.13 6.21

0.19 0.69 0.31 0.94 0.52 0.33 0.6 0.57 0.96 0.49 0.83 0 0

0.19 0.69 0.32 0.92 0.5 0.33 0.61 0.51 0.98 0.5 0.85 0 0

0.08 0.69 0.27 0.85 0.59 0.38 0.51 0.49

AlVO421

α-LiAlO222 γ-LiAlO222 SiAl2O5: Andalusite23 Kyanite24

Sillimanite25 YAlO326 Y4Al2O926 Y3Al5O1226 a

a

0.48 0.77 0.05

CN denotes the coordination number of Al.

the GIPAW method, where the generalized gradient approximation (GGA) was used for the exchange-correlation term in DFT calculations.25 A chemical shielding reference of 555.19 ppm was used to calibrate the theoretical results. However, the universality of such a reference needs to be further clarified. In this work, we first developed first-principles simulations on 11 aluminum-bearing crystal compounds (as listed in Table 1) whose NMR parameters (chemical shift, δiso, quadrupolar coupling constants, CQ, and asymmetry parameter, η) were well documented. The optimized structures were used to estimate the theoretical NMR parameters of 27Al at different local environments by the GIPAW algorithm. The relationship between theoretical NMR parameters and experimental values was deduced. Based on the previous study on the crystal structure of γ-alon, the theoretical 27Al NMR parameters of Al22.5O28.5N3.5 were simulated and predicted. Further, the theoretical results were used as the initial values in analytical fitting of 27Al MAS NMR spectra for γ-alon. A series of techniques, such as PXRD, ICP, etc., were also introduced to understand the crystal structure of the as-prepared sample. Finally, the comprehensive approach, which is based on 27Al MAS NMR analysis and PXRD Rietveld crystal structure refinement, to the understanding of the crystal structure of oxynitride was summarized.

sites can lead to the change of chemical shift in composite resonances of 27Al nuclei NMR spectra, where the contribution of different coordinated environments could be revealed by analytical fitting. Nevertheless, the chemical shift resolved from the NMR line shape can be affected by the variation of internal factors (bond length, bond angle, etc.) and external dynamics in an experiment. For this reason, even though the NMR spectra have successfully presented the complex disordered distribution in series materials,17−19 accurate resolution and assignment of chemical shifts in analytical fitting of NMR spectra are still challenges. Nowadays, with the development of the gauge-including projector augmented wave (GIPAW) algorithm based on density functional theory (DFT), first-principles calculation is certainly an efficient approach to interpret, assign, and even predict the NMR spectra of disordered crystals.20,21 For disordered materials, the random distribution or partial occupancy of atoms can hardly be maintained in the construction of a supercell for first-principles calculations. The atomic position should be precisely located by tuning the atomic configurations, which provides a flexible way to get insight into the local environment. Combining with highresolution magic-angle spinning (MAS) NMR, it offers a powerful tool for understanding the disordered crystal structure. A series of remarkable investigations based on GIPAW calculation and advanced NMR experiments have ever been successfully developed for understanding the crystal structure, and electronic and optical properties of tris(8hydroxyquinoline) aluminum(III) (Alq3).22−24 Reader et al. reported first-principles calculations alongside experiments to interpret the 89Y MAS NMR spectra of Y2Ti2−xSnxO7, and confirmed a random distribution of Sn/Ti in the crystal.19 For 27 Al nuclei, Choi et al. presented a detailed relationship between 27Al NMR parameters and the local structures using

2. COMPUTATIONAL AND EXPERIMENTAL DETAILS 2.1. Calculations. All theoretical simulations were carried out with the CASTEP package and were done using well determined atomic positions and the lattice parameter in the literature.26−33 The supercell model of γ-alon with the formula Al22.5O28.5N3.5 was constructed according to the previously optimized model in terms of the sitepreference of N and cationic vacancies.10 The GGA with PBE functional for solids (PBEsol) and the local density approximation (LDA) based on the data of Ceperley and Alder34,35 were used to B

DOI: 10.1021/acs.inorgchem.6b02360 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry describe the electronic exchange-correlation (XC) potential.36 The ultrasoft pseudopotential was generated using an “on-the-fly” method. The plane-wave cutoff energy of 420 eV was used. The Brillouin zone was sampled with a grids spacing of approximately 0.03 Å−1 according to the Monkhorst−Pack scheme. Crystal lattice and atomic positions in the supercell were fully relaxed using the Broyden−Fletcher− Goldfarb−Shanno (BFGS) method with the following convergence tolerances: 5.0 × 10−7 eV for energy change, 0.01 eV/Å for maximum force, and 5.0 × 10−4 Å for maximum displacement. The σiso, CQ, and η of 27Al nuclei in optimized structures were simulated using the GIPAW method, which has been widely adopted to predict the NMR parameters. 2.2. Sample Preparation and Characterization. The polycrystalline γ-alon sample was prepared by a conventional solid-state method. A stoichiometric amount of commercial AlN (∼99.9%) and α-Al2O3 (∼99.99%) powders were used to synthesize γ-alon with the designed formula of Al22.5O28.5N3.5. The detailed preparation process of the powder was consistent with the method of MgAlON in our previous report.37 The Al content was quantitatively determined to be 54.58 wt% by ICP-AES analysis (model Optima 4300DV; PerkinElmer Instruments, Boston, MA), which is very close to the expected nominal proportion. Then, the contents of O and N can be extrapolated from the stoichiometric formula of γ-alon, Al(8+x)/3O4−xNx, which are 41.04 wt% and 4.38 wt%, respectively. So, the chemical formula of the obtained compound can be expressed as Al2.811O3.565N0.435. By the well-established hot gas extraction method (model TC-600; Leco, St. Joseph, MI), where SiO2 and TaN were used as calibration standards for O and N, respectively, the concentrations of O and N were directly detected to be 41.18 wt% and 4.53 wt%, which agreed well with the ICP-AES results. The PXRD patterns were recorded on a PANalytical X’Pert Pro X-ray diffractometer equipped with monochromatized Cu Kα radiation (1.540598 Å, 40 kV, 40 mA). The Rietveld crystal structure refinements were performed using the Fullprof program. The solid-state 27Al NMR experiments were performed on a Bruker Avance III 500 WB (11.7 T) spectrometer operating at a frequency of 130.43 MHz. A DVT triple resonance 2.5 mm o.d. Bruker CP/MAS probe was used with a spinning frequency of 30.0 kHz. To obtain quantitative results, 27Al MAS NMR spectra were recorded with a single pulse excitation using a short pulse length (π/12) and a recycle delay of 1 s (the RF field strength was 166.67 kHz, the tip angle was π/ 12, and the spin−lattice relaxation time, T1, is 150 ms). 27Al chemical shifts were referenced using 1 M Al(NO3)3 aqueous solution as an external reference (0 ppm). The NMR parameters δiso, CQ, and η were then extracted from fitting the curve using the CzSimple model in the Dmfit package.38,39

calculations. In ab initio calculations where the GGA method was used for the exchange-correlation functional, σref of 555.19 ppm and unity slope were adopted to calculate the theoretical chemical shift of 27Al nuclei.25 However, this value cannot be used for theoretical results obtained from PBEsol or LDA methods. As shown in Figure 1a, to establish the relationship between simulated σiso and experimental δiso, the data were fitted by linear equations as follows:

Figure 1. (a) Linear relationship between calculated the isotropic chemical shielding tensor, σiso, and the experimental chemical shift, δiso, where σiso was simulated using both PBEsol and LDA methods, and (b) experimental δiso as a function of the coordination number of the 27 Al nuclei.

PBEsol: δisoexp = 550.52 − 0.996σisocalc

3. RESULTS AND DISCUSSION 3.1. Theoretical Prediction of 27Al NMR Parameters. The isotropic chemical shielding tensor (σiso) can be simulated as well as the electric field gradient (EFG) tensor from firstprinciples calculations. In order to compare the calculated shielding constants with the experimentally measured chemical shifts, the calculated σiso needs to be converted into the chemical shift scale. The δiso is obtained from the σiso through eq 1: δiso = −(kσiso − σref )/(1 − σref ) ≈ −(kσiso − σref )

(2)

LDA: δisoexp = 508.72 − 0.936σisocalc

(3)

.When fixing k to 1, the coefficient of determination, R2, decreased from 0.997 to 0.993 for LDA, but almost did not change (R2 ≈ 0.997) for the PBEsol method. Such a high quality linear relationship observed in a broad chemical shift range is a good indication of the reliability of the calculated shielding constants. Compared with LDA, the chemical shielding tensor of 27Al can be more accurately predicted by the PBEsol method. As the coordination number increases, the chemical shift of 27Al presents an obvious decreasing trend, as indicated in Figure 1b, which could help to reveal the coordination environment of Al cations. The tensor components, VXX, VYY, and VZZ, in the principal axis system are used to compute the CQ and η. Considering the fact that the largest component VZZ can be positive or negative, the CQ, which is defined by eQVzz/h (quadrupolar moment, Q, is 146.6 mb for 27Al), can be negative. Since the CQ value determined experimentally is often the absolute value, only the

(1)

where k is equal to 1 for experimental and fully accurate theoretical results, and σref is an isotropic chemical shielding reference. In theoretical simulations, σref is reproduced by linearly fitting the simulated σiso of a series of compounds to experimental values of δiso. In Table 1, the calculated and experimental NMR parameters of Al nuclei in selected compounds are listed. The simulated σiso from LDA is obviously different from the value calculated by the PBEsol calculation, which demonstrates that the simulated chemical shielding shift is sensitive to the exchange-correlation functional in DFT C

DOI: 10.1021/acs.inorgchem.6b02360 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry

Al22.5O28.5N3.5, Al23O27N5, Al23.5O25.5N6.5, and Al3O3N, have been investigated in our previous report.7,43 Owing to the fact that the stoichiometry of Al22.5O28.5N3.5 is very close to that of our prepared sample, the NMR parameters of the Al22.5O28.5N3.5 supercell were calculated. In crystallography, all Al at the tetrahedral site are identical, and all octahedrally coordinated Al are identical. However, the symmetry of the disordered crystal structure has to be ignored in supercell calculations. There are four types of coordination environments of Al in the structural model of Al22.5O28.5N3.5 according to the site preference of nitrogen,10 which are AlO3N and AlO4 for tetrahedral coordination, and AlO5N and AlO6 for octahedral coordination. After fully optimization, the average values of σiso, CQ, and η were calculated, and the chemical shift δiso of 27Al nuclei were further computed using eq 2 and eq 3. All theoretical NMR parameters as well as the relative populations of Al at tetrahedral and octahedral sites were included in Table 2. One may notice that tetrahedrally coordinated Al environments showed a much larger chemical shift than Al located at octahedrally coordinated positions, which is in good agreement with the previous result.44 The substitution of O by N in both tetrahedral AlO4 coordination and octahedral AlO6 coordination leads to the increase of chemical shift. It was notable that the results obtained from the LDA method are consistent with that from PBEsol calculations. The theoretically predicted values of δiso and CQ were then used as the initial parameters in the analytical resolution of 27Al MAS NMR experimental spectra. The qualitative solid-state NMR study on γ-alon has been developed to identify the possible coordination environments of Al,44−46 but there is still no report about NMR parameters extracted from MAS NMR spectra thus far. The 27Al NMR peaks of γ-alon with 16.7 mol% AlN were found at 65 and 14 ppm, attributed to AlO4 and AlO6, respectively,44 which are similar to those in γ-Al2O3.47 As shown in Figure 3, two little broad asymmetric peaks around 63.5 and 9.7 ppm are shown in our 27Al MAS NMR spectra at 11.7 T, which obviously originate from tetrahedral and octahedral Al, respectively. Compared with reported values, the peak positions present an overall downshift. Figure 3 also presents the curves obtained from fitting the 27Al MAS NMR spectra. This two-step strategy was developed in our fitting: first, the δiso and CQ were fixed at the values predicted from DFT-GIPAW simulations to optimize the asymmetry parameter; when the convergence was reached, the δiso and CQ were then released to fully fit the curve. Table 2 shows the parameters obtained by analytical fitting. The parameters obtained from experiments are in agreement with the theoretical structural model of Al22.5O28.5N3.5. NMR spectral integration can provide invaluable quantitative information, as the integral of a spectra indicates the populations of nuclei in different coordinations which

absolute calculated values versus experimental results were plotted in Figure 2a. It was found that the simulated results are

Figure 2. Comparison between the calculated and experimental (a) quadrupolar coupling constant, CQ, and (b) asymmetry parameter, η, of 27Al nuclei. The absolute values of the calculated CQ were plotted. The experimental and simulated results were distinguished by superscripts of exp and calc, respectively.

in good agreement with experiments, which leads to the successful theoretical prediction of the experimental CQ. On the contrary, as shown in Figure 2b, the theoretical η, which is defined by (VYY − VXX)/VZZ, presents a relative divergence from experiments. However, as the δiso and CQ are being predicted, the η can be extracted more efficiently and accurately from the analytical fitting of 27Al MAS NMR experimental spectra. The theoretical prediction of NMR parameters using the DFT-GIPAW method has been successfully developed in several systems.19,40−42 It could be concluded that the experimental chemical shift and quadrupolar coupling constant can be well reproduced by first-principles calculations, even though, depending upon the calculation methodology, the reference isotropic shielding should be carefully determined. Based on the accurate prediction of δiso and CQ, the assignment and resolution of NMR parameters from experimental data processing will be precisely and efficiently carried out. 3.2. 27Al MAS NMR Investigation of γ-alon. Five models of the γ-alon solid solution, which were denoted as Al22O30N2,

Table 2. Calculated (PBEsol and LDA) Chemical Shift, δiso, Quadrupolar Coupling Constant, CQ, and Asymmetry Parameter, η, of Al22.5O28.5N3.5, and Experimental (Exp) δiso, CQ, and η of Al2.811O3.565N0.435 δiso (ppm)

η

CQ (MHz)

Population (%)

Site

Coor

PBEsol

LDA

Exp

PBEsol

LDA

Exp

PBEsol

LDA

Exp

PBEsol & LDA

Exp

Tetra.

AlO4 AlO3N AlO6 AlO5N

67.41 86.87 18.25 21.49

68.55 87.45 17.71 20.49

69.64 73.84 15.91 21.07

6.66 12.54 5.23 8.41

6.91 13.11 5.48 8.74

5.05 6.96 4.76 7.98

0.75 0.40 0.54 0.51

0.75 0.40 0.54 0.50

0.612 0.609 0.611 0.609

36.36

34.70

63.64

65.30

Octa.

D

DOI: 10.1021/acs.inorgchem.6b02360 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry

Figure 3. 27Al MAS NMR spectra (500 MHz) of Al2.811O3.565N0.435, and the individual contributions of different coordinations: AlO4, AlO4N, AlO6, and AlO5N, and spinning sidebands (ssb), with population ratios of 19.18%, 12.58%, 33.91%, 25.89%, and 8.4%, respectively. The δiso, CQ, and η of 27Al at different sites were resolved and included in Table 2.

Figure 4. Atomic ratio of aluminum at the 8a site to the 16d site in the crystal structure of γ-alon solid solution (Al(8+x)/3O4−xNx), where the position of the solid star indicates most of the cationic vacancies residing at the 16d site in the Al2.811O3.565N0.435 crystal, and the hollow star originates from the theoretical model of Al22.5O28.5N3.5.

contributed to a given line. The proportions of different coordinated Al were computed from the integrated intensities of the central transition signatures (discarding the outer transitions). As indicated in Table 2, the populations of Al at tetrahedral and octahedral sites are 34.70% and 65.30%, respectively, which are slightly different from the theoretical model. The relationship between the molar ratio of Al at the tetrahedral site and those at the octahedral site (n[Al8a]/ n[Al16d]) and the composition x in the formula of γ-alon, Al(8+x)/3O4−xNx, described as follows, Al vacancies reside at 16d site: n[Al 8a]/n[Al16d] = 3/(5 + x)

(4)

Al vacancies reside at 8a site: n[Al 8a]/n[Al16d] = (2 + x)/6

(5)

Figure 5. Observed, calculated, and difference patterns obtained by Rietveld structural refinement of the PXRD data of Al2.811O3.565N0.435. The crystallographic information and final reliability factors RB, RF, Rwp, and Gof were obtained; see Table 3.

relates to the distribution of the cationic vacancy. The relationships defined by eq 4 and eq 5 were plotted in Figure 4. By studying the Al populations using MAS NMR, information on the vacancy distribution can be obtained. In this work, n[Al8a]/n[Al16d] is calculated to be 0.531 from deconvolution of the 27 Al MAS NMR spectra of Al2.811O3.565N0.435, as indicated in Figure 4, which is very close to the value of 0.552, for which all vacancies locate in the octahedral coordination. This means that the aluminum vacancies in the crystal structure of γ-alon prefer to occupying the 16d site, but a small part of them can exist at the 8a site to maintain the stabilization, which is consistent with our previous theoretical results.10 Combined with the quantitative determination of the Al composition, the site occupancies of Al at the 8a and 16d sites were calculated to be 0.9755 and 0.9178, respectively, which will be introduced in the PXRD structural refinement in the following section. 3.3. Structure Refinement of γ-alon. The XRD pattern of prepared Al2.811O3.565N0.435 powder was collected and used to refine the crystal structure. As plotted in Figure 5, all Bragg reflections appearing in the diffraction pattern were indexed and assigned to the cubic crystal system with space group Fd3m (No. 227). The crystal structure of Al2.811O3.565N0.435 was further refined by the Rietveld method. As it is known that the

site fractions of nitrogen and oxygen atoms at anion sites could not be resolved by the PXRD technique due to the close atomic scattering factor of them, the powder neutron diffraction pattern had been collected to refine the occupancy of O and N.8 In our refinement, the atomic occupancies of O and N at the corresponding crystallographic sites were fixed according to the anterior part of the Results and Discussion, and the values are 0.8912 and 0.1088, respectively. As obtained from NMR investigations, the occupancies of 0.9755 and 0.9178 for Al at the 8a and 16d sites, respectively, were used in the Rietveld structural refinement. From Rietveld analysis, as illustrated in Figure 5, the calculated pattern fitted well to the experimental data. The final reliability factors RB, RF, Rwp, and Gof were calculated to be 0.058, 0.050, 0.163, and 1.3, respectively. These factors normally preclude a reliable crystal structure refinement. The crystallographic parameters and atomic positions were obtained and listed in Table 3. Resulting from the severe disorder atomic distribution of Al at the 16d site, the isotropic thermal parameter (Biso) of Al at the 16d site is larger than that of O and N at the 32e site. Such a result agrees well with the E

DOI: 10.1021/acs.inorgchem.6b02360 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry Table 3. Crystallographic Data and Atomic Position of Al2.811O3.565N0.435 Formula Space group Radiation Lattice parameter, a (Å) Z Density, ρcalc (g/cm3) Profile range (deg) Step size (deg) Profile function Number of varied parameters Reliability factors

a

Al2.811O3.565N0.435 Fd3m Cu Kα, λ = 1.5406 Å 7.9471 8 3.6785 15.072 ≤ 2θ ≤ 139.994 0.0131 pseudo-Voigt 18 RB = 0.058, RF = 0.050, Rwp = 0.163, Gof = 1.3 Atomic positions:

Atom

Wyckoff position

x

y

z

Bisoa

Occupancy

Al Al O N

8a 16d 32e 32e

0.125 0.5 0.2564(7) 0.2564(7)

0.125 0.5 0.2564(7) 0.2564(7)

0.125 0.5 0.2564(7) 0.2564(7)

0.486(2) 1.098(2) 0.198(3) 0.198(3)

0.9755 0.9178 0.8912 0.1088

Biso is the isotropic temperature factor.

Rietveld crystal structure refinement of Mn-doped γ-alon from synchrotron XRD data combined with X-ray diffraction absorption fine structure (XAFS) measurements.48 Additionally, the high Biso of the cation at the 16d site has also been reported in disordered crystal structures of spinel-type GaON and MgAlON.37,49,50 As it is known that X-ray diffraction may be incapable of distinguishing the occupancies of O and N at the 32e site, but works well to identify the partial occupation of Al at the 8a or 16d site, to further evaluate our strategy to determine the crystal structure, the occupancies of Al at 8a and 16d were refined with charge balance restraint in the Rietveld structural refinement. After the best fitting was achieved, the extracted lattice parameters, anion parameters, and reliability factors were identical with those obtained from our NMR plus XRD strategy. Obviously, the Rietveld structural refinement is accurate to represent the framework of the crystal lattice. The convergence led to slightly diverse disorder cationic occupancies (0.9745 for the 8a site, 0.9183 for the 16d site) in γ-alon. The n[Al8a]/n[Al16d] was calculated to be 0.5306. It is notable that the refined occupancies of Al located at 8a and 16d were almost identical with the initial values from NMR investigations. However, the reliability of the Rietveld refinement will be inevitably lowered in more complicated oxynitride, with more than three types of disordered atomic configurations, for instance MgAlON and MnAlON.37,48 To this point, our study developed an effective and reliable strategy to explore the complicated disordered crystalline system. XRD structural refinement is no doubt one of the most important and general techniques to explore the crystal structures of novel materials, even though, owing to the limit of analytical resolution and similar scattering factors of disordered atoms, this approach reveals a slight powerlessness to describe the oxynitride crystal with multiple disorder. There are several techniques can be utilized combined with XRD to understand the crystal structure, such as neutron diffraction, Xray photoelectron spectroscopy, energy dispersive spectroscopy, etc. Here, we present a competitive strategy to reveal the disordered crystal structure by integrating the advantages of DFT calculations, NMR, and XRD techniques. In our case, the first-principles calculations provided predictions of NMR parameters and the local environment of disorder atoms.

Furthermore, the theoretical results guided the analytical investigation of 27Al MAS NMR spectra, which led to the quantitative determination of the occupancy of disordered atoms. Finally, based on the results from NMR investigations, the Rietveld structural refinement was efficiently and accurately performed and led to the full understanding of the crystal lattice of Al2.811O3.565N0.435. The NMR technique and PXRD structural refinement complement each other to contribute a powerful and accessible approach for deep understanding of disordered crystal structures.

4. CONCLUSIONS By combining first-principles simulations and the 27Al MAS NMR technique, the disordered local crystal structure of γ-alon with formula of Al2.811O3.565N0.435 was comprehensively studied. The DFT based GIPAW method was employed to simulate the 27 Al NMR parameters (δiso, CQ, and η) in several Al-bearing oxides. The δiso can be predicted from simulated chemical shielding tensor by empirical linear equations whose parameters depend on the selected exchange-correlation functional in DFT simulations. The CQ was quantitatively reproduced by simulated results, while η presented a slight divergence from theoretical values. The calculated 27Al NMR parameters of Al22.5O28.5N3.5 were then used to estimate the experimental δiso, CQ, and η of an as-prepared sample whose formula was determined to be Al2.811O3.565N0.435. The reliable deconvolution of the 27Al MAS NMR spectra was conducted by theoretical prediction, which revealed the site preferences of cationic vacancies as described in a previous study. The site occupancies of Al (8a), Al (16d), O, and N obtained from theoretical and experimental NMR investigations were used in PXRD Rietveld structural refinement, which finally leads to the systematic understanding of the crystal structure of Al2.811O3.565N0.435. The advantages of DFT calculations, NMR, and PXRD Rietveld refinement were integrated into the rational refinement of the crystal structure. This work provided an effective and practical approach to investigate the crystal structure of novel oxynitride materials with multiple disorder conditions. F

DOI: 10.1021/acs.inorgchem.6b02360 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry

■

Zinc Oxynitride Photocatalyst (Ga1‑xZnxN1‑xOx): origin of visible light absorption. Chem. Commun. 2010, 46, 2379−81. (15) Lumey, M. W.; Dronskowski, R. The Electronic Structure of Tantalum Oxynitride and the Falsification of α-TaON. Z. Anorg. Allg. Chem. 2003, 629, 2173−2179. (16) MacKenzie, K. J. D.; Smith, M. E., Multinuclear Solid-State Nuclear Magnetic Resonance of Inorganic Materials; Elsevier Science Ltd.: Oxford, 2002. (17) Dupree, R.; Lewis, M. H.; Smith, M. E. A Study of the Vacancy Distribution in Non-Stoichiometric Spinels by Magic-Angle Spinning NMR. Philos. Mag. A 1986, 53, L17−L20. (18) Pedone, A.; Charpentier, T.; Malavasi, G.; Menziani, M. C. New Insights into the Atomic Structure of 45S5 Bioglass by Means of SolidState NMR Spectroscopy and Accurate First-Principles Simulations. Chem. Mater. 2010, 22, 5644−5652. (19) Reader, S. W.; Mitchell, M. R.; Johnston, K. E.; Pickard, C. J.; Whittle, K. R.; Ashbrook, S. E. Cation Disorder in Pyrochlore Ceramics: 89Y MAS NMR and First-Principles Calculations. J. Phys. Chem. C 2009, 113, 18874−18883. (20) Yates, J. R.; Pickard, C. J., Computation of Magnetic Resonance Parameters for Crystalline Systems: Principles. In NMR Crystallography; Harris, R. K., Wasylishen, R. E., Eds.; Wiley: Chichester, U.K., 2009. (21) Pickard, C. J.; Mauri, F. All-Electron Magnetic Response with Pseudopotentials: NMR Chemical Shifts. Phys. Rev. B: Condens. Matter Mater. Phys. 2001, 63, 245101. (22) Suzuki, F.; Nishiyama, Y.; Kaji, H. Clarification of Isomeric Structures and the Effect of Intermolecular Interactions in BlueEmitting Aluminum Complex Alq3 Using First-Principles 27Al NMR Calculations. Chem. Phys. Lett. 2014, 605−606, 1−4. (23) Kaji, H.; Kusaka, Y.; Onoyama, G.; Horii, F. CP/MAS 13C NMR Characterization of the Isomeric States and Intermolecular Packing in Tris(8-hydroxyquinoline) Aluminum(III) (Alq3). J. Am. Chem. Soc. 2006, 128, 4292−4297. (24) Suzuki, F.; Fukushima, T.; Fukuchi, M.; Kaji, H. Refined Structure Determination of Blue-Emitting Tris(8-hydroxyquinoline) Aluminum(III) (Alq3) by the Combined Use of Cross-Polarization/ Magic-Angle Spinning13C Solid-State NMR and First-Principles Calculation. J. Phys. Chem. C 2013, 117, 18809−18817. (25) Choi, M.; Matsunaga, K.; Oba, F.; Tanaka, I. 27Al NMR Chemical Shifts in Oxide Crystals: A First-Principles Study. J. Phys. Chem. C 2009, 113, 3869−3873. (26) Vosegaard, T.; Jakobsen, H. J. 27Al Chemical Shielding Anisotropy. J. Magn. Reson. 1997, 128, 135−137. (27) O’Dell, L. A.; Savin, S. L.; Chadwick, A. V.; Smith, M. E. A 27Al MAS NMR study of a sol-gel produced alumina: Identification of the NMR parameters of the θ-Al2O3 transition alumina phase. Solid State Nucl. Magn. Reson. 2007, 31, 169−73. (28) Nielsen, U. G.; Boisen, A.; Brorson, M.; Jacobsen, C. J.; Jakobsen, H. J.; Skibsted, J. Aluminum orthovanadate (AlVO4): Synthesis and characterization by 27Al and 51V MAS and MQMAS NMR spectroscopy. Inorg. Chem. 2002, 41, 6432−6439. (29) Müller, D.; Gessner, W.; Scheler, G. Chemical shift and quadrupole coupling of the 27Al NMR spectra of LiAIO2 polymorphs. Polyhedron 1983, 2, 1195−1198. (30) Rocha, J. Direct observation of highly distorted hexacoordinated aluminium in andalusite by very fast 27Al MAS NMR. Chem. Commun. 1998, 2489−2490. (31) Smith, M. E.; Jaeger, C.; Schoenhofer, R.; Steuernagel, S. Structural characterisation of the aluminium sites in kyanite by 27Al magic angle spinning centreband, satellite transition and double rotation NMR. Chem. Phys. Lett. 1994, 219, 75−80. (32) Massiot, D. Sensitivity and Lineshape Improvements of MQMAS by Rotor-Synchronized Data Acquisition. J. Magn. Reson., Ser. A 1996, 122, 240−244. (33) Florian, P.; Gervais, M.; Douy, A.; Massiot, D.; Coutures, J.-P. A Multi-Nuclear Multiple-field Nuclear Magnetic Resonance Study of the Y2O3-Al2O3 Phase Diagram. J. Phys. Chem. B 2001, 105, 379−391.

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.inorgchem.6b02360. Crystallographic data of Al2.811O3.565N0.435 (CIF)

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Hao Wang: 0000-0002-3511-3895 Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The authors would like to acknowledge Pengyu Xu and Xiao Zong for preparation and characterizations of Al2.811O3.565N0.435 powder. This work was supported by National Natural Science Foundation of China (Nos. 51472195 and 51502219) and China Postdoctoral Science Foundation (No. 2015M582286).

■

REFERENCES

(1) Kudo, A.; Miseki, Y. Heterogeneous Photocatalyst Materials for Water Splitting. Chem. Soc. Rev. 2009, 38, 253−278. (2) Patel, P. J.; Swab, J. J.; Gilde, G., Fracture Properties and Behavior of Transparent Ceramics. In Proc. SPIE, Marker, A. J., Arthurs, E. G., Eds.; SPIE Digital Library: San Diego, 2000; Vol. 4102, pp 15−24. (3) Tustison, R. W.; Nag, N.; Goldman, L. M.; Balasubramanian, S.; Sastri, S.; Zelinski, B. J., Multi-Functional Windows. In Window and Dome Technologies and Materials XIII; Prof. SPIE: 2013; Vol. 8708. (4) Salem, J. A.; Sglavo, V. Transparent Armor Ceramics as Spacecraft Windows. J. Am. Ceram. Soc. 2013, 96, 281−289. (5) Xie, R.; Hintzen, H. T. Optical Properties of (Oxy)Nitride Materials: A Review. J. Am. Ceram. Soc. 2013, 96, 665−687. (6) Lejus, A. M. Formation at High Temperature of Nonstoichiometric Spinel and of Derived Phaes in Several Oxide Systems Based on Alumina and in the System Alumina-Aluminum Nitride. Rev. Int. Hautes Temp. Refract. 1964, 1, 53−95. (7) Tu, B.; Wang, H.; Liu, X.; Wang, W.; Fu, Z. First-Principles Insight into the Composition-Dependent Structure and Properties of γ-Alon. J. Am. Ceram. Soc. 2014, 97, 2996−3003. (8) Willems, H. X.; de With, G.; Metselaar, R.; Helmholdt, R. B.; Petersen, K. K. Neutron Diffraction of γ-Aluminium Oxynitride. J. Mater. Sci. Lett. 1993, 12, 1470−1472. (9) Fang, C. M.; Metselaar, R.; Hintzen, H. T.; de With, G. Structure Models for Gamma-Aluminum Oxynitride from Ab Initio Calculations. J. Am. Ceram. Soc. 2001, 84, 2633−2637. (10) Tu, B.; Wang, H.; Liu, X.; Wang, W.; Fu, Z. First-Principles Study on Site Preference of Aluminum Vacancy and Nitrogen Atoms in γ-Alon. J. Am. Ceram. Soc. 2013, 96, 1937−1943. (11) Schwarz, M.; Zerr, A.; Kroke, E.; Miehe, G.; Chen, I. W.; Heck, M.; Thybusch, B.; Poe, B. T.; Riedel, R. Spinel Sialons. Angew. Chem., Int. Ed. 2002, 41, 789−793. (12) Pan, C.; Takata, T.; Nakabayashi, M.; Matsumoto, T.; Shibata, N.; Ikuhara, Y.; Domen, K. A Complex Perovskite-Type Oxynitride: The First Photocatalyst for Water Splitting Operable at up to 600 nm. Angew. Chem., Int. Ed. 2015, 54, 2955−9. (13) Dharmagunawardhane, H. A. N.; Woerner, W. R.; Wu, Q.; Huang, H.; Chen, X.; Orlov, A.; Khalifah, P. G.; Parise, J. B. Photocatalytic Hydrogen Evolution Using Nanocrystalline Gallium Oxynitride Spinel. J. Mater. Chem. A 2014, 2, 19247−19252. (14) Yashima, M.; Yamada, H.; Maeda, K.; Domen, K. Experimental Visualization of Covalent Bonds and Structural Disorder in a Gallium G

DOI: 10.1021/acs.inorgchem.6b02360 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry (34) Hedin, L.; Lundqvist, B. I. Explicit Local Exchange-Correlation Potentials. J. Phys. C: Solid State Phys. 1971, 4, 2064−83. (35) Ceperley, D. M.; Alder, B. J. Ground State of the Electron Gas by a Stochastic Method. Phys. Rev. Lett. 1980, 45, 566−569. (36) Perdew, J. P.; Burke, K.; Ernzerhof, M. Generalized Gradient Approximation Made Simple. Phys. Rev. Lett. 1996, 77, 3865−3868. (37) Liu, X.; Wang, H.; Lavina, B.; Tu, B.; Wang, W.; Fu, Z. Chemical composition, crystal structure, and their relationships with the intrinsic properties of spinel-type crystals based on bond valences. Inorg. Chem. 2014, 53, 5986−92. (38) Neuville, D. R.; Cormier, L.; Massiot, D. Al Environment in Tectosilicate and Peraluminous Glasses: A 27Al MQ-MAS NMR, Raman, and XANES Investigation. Geochim. Cosmochim. Acta 2004, 68, 5071−5079. (39) Massiot, D.; Fayon, F.; Capron, M.; King, I.; Le Calvé, S.; Alonso, B.; Durand, J.-O.; Bujoli, B.; Gan, Z.; Hoatson, G. Modelling One- and Two-Dimensional Solid-State NMR Spectra. Magn. Reson. Chem. 2002, 40, 70−76. (40) Pallister, P. J.; Moudrakovski, I. L.; Ripmeester, J. A. Mg-25 Ultra-High Field Solid State NMR Spectroscopy and First Principles Calculations of Magnesium Compounds. Phys. Chem. Chem. Phys. 2009, 11, 11487−500. (41) Johnston, K. E.; Mitchell, M. R.; Blanc, F.; Lightfoot, P.; Ashbrook, S. E. Structural Study of La1−xYxScO3, Combining Neutron Diffraction, Solid-State NMR, and First-Principles DFT Calculations. J. Phys. Chem. C 2013, 117, 2252−2265. (42) Mitchell, M. R.; Carnevale, D.; Orr, R.; Whittle, K. R.; Ashbrook, S. E. Exploiting the Chemical Shielding Anisotropy to Probe Structure and Disorder in Ceramics: 89Y MAS NMR and FirstPrinciples Calculations. J. Phys. Chem. C 2012, 116, 4273−4286. (43) Tu, B.; Wang, H.; Liu, X.; Khan, S. A.; Wang, W.; Fu, Z. Composition-dependent bonding and hardness of γ-aluminum oxynitride: A first-principles investigation. J. Appl. Phys. 2014, 115, 223511. (44) Fitzgerald, J. J.; Kohl, S. D.; Piedra, G.; Dec, S. F.; Maciel, G. E. Observation of four-coordinate aluminum oxynitride (AlO4‑xNx) environments in AlON solids by MAS 27Al NMR at 14 T. Chem. Mater. 1994, 6, 1915−1917. (45) Ren, H.; Yue, Y.; Ye, C.; Ying, D.; Cewen, N. The Use of Solid State NMR in Structural Studies of AlON Spinel. Chem. J. Chin. Univ. 1999, 20, 297−298. (46) Smith, M. E. Observation of mixed aluminum oxynitride Al(O,N)4 structural units by 27Al magic angle spinning NMR. J. Phys. Chem. 1992, 96, 1444−1448. (47) Düvel, A.; Romanova, E.; Sharifi, M.; Freude, D.; Wark, M.; Heitjans, P.; Wilkening, M. Mechanically Induced Phase Transformation of γ-Al2O3 into α-Al2O3. Access to Structurally Disordered γ-Al2O3 with a Controllable Amount of Pentacoordinated Al Sites. J. Phys. Chem. C 2011, 115, 22770−22780. (48) Takeda, T.; Xie, R.; Hirosaki, N.; Matsushita, Y.; Honma, T. Manganese valence and coordination structure in Mn,Mg-codoped γAlON green phosphor. J. Solid State Chem. 2012, 194, 71−75. (49) Soignard, E.; Machon, D.; McMillan, P. F.; Dong, J.; Xu, B.; Leinenweber, K. Spinel-Structured Gallium Oxynitride (Ga3O3N) Synthesis and Characterization: An Experimental and Theoretical Study. Chem. Mater. 2005, 17, 5465−5472. (50) Huppertz, H.; Hering, S. A.; Zvoriste, C. E.; Lauterbach, S.; Oeckler, O.; Riedel, R.; Kinski, I. High-Pressure Synthesis, Electron Energy-Loss Spectroscopy Investigations, and Single Crystal Structure De t e r m i na ti o n o f a S p i n e l - T y p e G a l l iu m O x o n i tr i d e Ga2.79□0.21(O3.05N0.76□0.19). Chem. Mater. 2009, 21, 2101−2107.

H

DOI: 10.1021/acs.inorgchem.6b02360 Inorg. Chem. XXXX, XXX, XXX−XXX