Ordering of nitrogen and oxygen in nitrogen-containing melilites Y2Si3O3N4 and Nd2Si2.5Al0.5O3.5N3.5 Pei-LingWang,*a Per-Eric Werner,b Lian Gao,a Robin K. Harrisc and Derek P. Thompsond aState Key L ab of High Performance Ceramics and SuperfineMicrostructure, Shanghai Institute of Ceramics, Chinese Academy of Sciences, Shanghai 200050, China bDepartment of Structural Chemistry, Arrhenius L aboratory, University of Stockholm, S-106 91 Stockholm, Sweden cDepartment of Chemistry, University of Durham, South Road, Durham, UK DH1 3L E dMaterials Division, Department of Mechanical, Materials andManufacturing Engineering, University of Newcastle, Newcastle upon T yne, UK NE1 7RU Ordering of N and O atoms in nitrogen-containing melilite Y2Si3O3N4 (Y-M) was investigated by Rietveld refinement in the space group P4 : 21m using neutron powder diVraction data. The results show that, contrary to previous work using X-ray data, the occupancies by N and O atoms can be refined to show that about 1.7 N and 0.3 O atoms occupy 2c (inWyckoV notation, space group no. 113) sites in the Y-M unit cell, while the remaining N atoms in the unit cell are disordered at 8f (inWyckoV notation) sites.The similar ordering of N and O atoms in Nd2Si2.5Al0.5O3.5N3.5 (Nd-M¾) was also confirmed, by refinement from neutron data, which means that similar distributions of N and O atoms occurred at 2c sites in spite of the fact that the total numbers of N atoms per unit cell were eight and seven in Y-M and Nd-M¾ respectively. The present results give a more detailed picture of O,N ordering in these structures than was provided by the previous work ofWang andWerner based on X-ray data alone.Nitrogen-containing melilite and solid-solution phases, of gen- eight N atoms were located at 8f sites (eight-fold) of the unit cell4 (Fig. 1). The O,N ordering in Y-M was studied5 with eral formula R2Si3-xAlxO3+xN4-x (x1, R=Y, Nd, Sm, Gd, Dy, etc.), frequently occur at grain boundaries when yttria and MAS NMR by two of the authors and colleagues recently.Readers are referred to that paper for further details of the rare-earth oxides are used as sintering aids in the preparation of a-sialon and mixed a,b-sialon ceramics.1,2 The suggested3 structure. It was found that N atoms probably occupy the 2c (bridging, two-fold) sites and O atoms occupy 4e (terminal, structure of Y-M has space group P4 : 21m and is derived from akermanite (Ca2MgSi2O7) by substituting Y for Ca, Si for Mg four-fold) sites in the Si2(O,N)7 units, while the remaining six N and two O atoms are distributed at 8f sites.This is in and N for eight of the fourteen O in the unit cell (Z=2) for N. Further structure refinements based on the Rietveld whole- accordance with Pauling’s rules for charge distribution that O atoms are located at the 4e sites as these are the only ones in pattern fitting technique for Y2Si3O3N4 (Y-M) and Nd2Si2.5Al0.5O3.5N3.5(Nd-M¾) confirmed the model, in which the structure not coordinated to two silicon atoms and they have the lowest total positive valencies from surrounding cations (ca. 2). In order to determine the N,O ordering at the 2c and 8f sites of the Y-M unit, it is clearly more powerful if two techniques (Rietveld refinement and NMR) are used in combination.Since the scattering factors of N and O atoms for neutrons show a greater contrast than those for X-rays, neutron diVraction data were used to investigate the N,O ordering in Y-M to supplement the X-ray results.This paper describes the conclusions regarding N,O ordering from the refinement of the occupancies by N and O atoms together with other structural and profile parameters based on the distributions of N,O atoms of Y-M obtained by NMR.5 The N,O ordering in the Nd-M¾ unit cell was also derived and compared in this work. Experimental Samples of the Y-M and Nd-M¾ phases were prepared by the hot-pressing technique.The starting powders used were silicon nitride (LC12, H.C. Starck, Berlin), aluminium oxide (Wu Song Chemical Works, China, 99.9%) and aluminium nitride (prepared at Shanghai Institute of Ceramics), while Y2O3 and Nd2O3 were the products of Yaolung Chemical Works, China. Fig. 1 N-Melilite crystal structure, (001) projection. Large circles Powder mixtures were prepared with the overall nominal represent Y/Nd cations and small filled circles, at tetrahedron centres, composition R2Si3-xAlxO3+xN4-x (x=0 for R=Y and x=0.5 correspond to Si (and Al) in 4e and 2a sites.The remaining circles for R=Nd), taking into account surface oxygen on the particles correspond to 8f (filled) and 4e (open) non-metal sites, while open squares indicate 2c sites.of both Si3N4 (1.8 mass%) and AlN (2.0 mass%). The powders J. Mater. Chem., 1997, 7(10), 2127–2130 2127were mixed in absolute alcohol and milled in an agate mortar cell dimensions (a,c), one peak shape parameter and five background parameters (so-called profile parameters). No for 1.5 h. The compacted samples were fired by the hot-pressing technique (20 MPa) in a flowing nitrogen atmosphere for 1 h absorption correction was applied, and a common isotropic temperature factor was used for all the atoms. However, at the at 1750 and 1675 °C for Y-M and Nd-M¾ phases, respectively.6 The synthesised samples were characterised by powder X-ray beginning of the refinement procedure, an attempt was made to refine thermal parameters together with occupancies (30 diVraction (XRD) and it was found that the major phase was melilite, with a very small amount of the J-phase R4Si2O7N2 variables in total ).Although the error parameters of the fitting were lowered by this procedure, some of the thermal param- (R=Y, Nd) also being present in both samples. The unit cells of the Y-M and Nd-M¾ phases were refined, using Si powder eters were found to be negative, which is unacceptable.For this reason, we prefer to quote results obtained by constraining as an internal standard, from X-ray Guinier-Ha�gg camera diVraction patterns (Cu-Ka1 radiation, l=1.540 5981 A ° ) evalu- all thermal parameters to be equal (see Table 3). The refinement was terminated when all shifts in the parameters were <10% ated with a computer-controlled film scanner and associated programs.7,8 The neutron data collections (l=1.470 A ° ) of Y- of the corresponding standard deviations.At the final stage, the occupancies of N and O atoms for Y-M at 2c and 8f sites M and and Nd-M¾ were performed at 275 K at the Swedish Studsvik R2 reactor. Results were obtained in steps of 0.08° in were refined together with 24 parameters, keeping the total number of N and O atoms constant.The final R values for the ranges for 2h of 15–130 and 10–109.4° for Y-M and Nd- M¾ respectively, with a measuring time of 3 min per step. 195 reflections, together with some essential data, are shown in Table 1. For comparison, the refinement results of Y-M from neutron and X-ray data,4 in which two O and eight N Results atoms were located at 2c and 8f sites respectively, are also listed in Table 1.Fig. 2 shows the observed and computer-fitted neutron diVraction pattern of Y-M. The unit-cell dimensions obtained after Because of the higher scattering amplitudes of N,O atoms in the neutron case, the distributions of N,O atoms were much least-squares refinements are: a=b=7.6137(2), c=4.9147(2) A ° for Y-M and a=b=7.7462(5), c=5.0390(4) A ° for Nd-M¾.The more sensitive indicators than in the X-ray case. It is shown in Table 1 that all reliability index R values were obviously Rietveld refinements were performed with a version of the refinement program written by Wiles et al.9 The background reduced by fixing two N atoms at 2c sites, together with six N and two O atoms at 8f sites.The results were even better when intensity Ybi at the ith step was described by the polynomial the occupancies of N,O atoms at 2c and 8f sites were refined Ybi=SBm[2hi/BKPOS-1]m together. An unreasonable occupancy of N atoms at 2c sites was obtained for a similar refinement from X-ray data. where Bm are parameters to be refined and BKPOS is the origin of the polynomial for the background.e peak shape For refinements of Nd-M¾ the models were more complicated, because there are three kinds of possible substitutions used was a Pearson VII function. The extent of a peak was taken to be 3.0 times the FWHM (full-width at half-maximum), of Al for Si, i.e. one Al atom distributed in either 4e or 2a or in both of these sites, in addition to the diVerent distributions Hk, on either side of the peak centre.Hk was given by of anions. Similar refinement procedures as used for Y-M were Hk2=Utan2hk+Vtanhk+W performed on Nd-M¾ neutron data under the diVerent distributions of Si, Al atoms. Table 2 gives the final R values for where U, V, W are the width parameters and k is the reflection index. The atomic coordinates of Y-M and Nd-M¾ obtained the diVerent distributions of N,O,Si,Al in Nd-M¾.The atomic coordinates, isotropic thermal and occupancy parameters for in our previous work1 were used as the starting parameters. In the first trial for Y-M, two N atoms were located at 2c sites both Y-M and Nd-M¾ structures are shown in Table 3. Some selected interatomic distances for Y-M and Nd-M¾ are summar- of the space group P4 : 21m, and the remaining six N atoms were taken to occupy 8f sites of the unit cell together with two ised in Table 4.Similar refinement levels to those for Y-M were obtained O atoms. The convergent refinement involved twelve structural parameters and twelve profile parameters, i.e. one scale, ten for the Nd-M¾ phase. The R values were much lower when N,O atoms were refined at (2c,8f ) sites or were fixed at those atomic coordinates, and one isotropic temperature factor (socalled structural parameters), together with the zero-point sites (see Table 2).The results shown in Table 3 confirm that the N atoms occupy the two sites 2c and 8f in both Y-M and position, three peak half-width parameters (U, V, W ), the unit Nd-M¾ cases. The occupancies of N, O atoms further indicate Table 1 Final R values and refinement details for the diVerent distributions of O,N in Y-M N,O O(2c), N(8f ) N(2c) refined at N,O(8f ) (2c,8f ) X-ray4 neutron neutron neutron RF (%)a 5.25 6.37 4.00 3.94 RB (%) 6.55 11.91 6.81 6.67 RP (%) 6.76 6.88 4.97 4.89 RWP (%) 8.45 9.06 6.77 6.69 U 0.047(0) 1.54(8) 1.41(6) 1.41(6) V -0.014(1) -0.83(7) -0.74(5) -0.74(5) W 0.024(1) 0.20(1) 0.18(1) 0.18(1) no.of structural parameters 17 12 12 13 no. of profile parameters 11 12 12 12 Fig. 2 The neutron diVraction pattern for Y-M. Top: profile calculated by least-squares Rietveld refinement. Middle: experimental pattern. aReliability index R can be defined as RF, RB, RP, RWP. RF: R value for structure amplitudes; RB: R value for Bragg intensities; RP: the Bottom: diVerence pattern.The asterisks indicate peaks assigned to Y4Si2O7N2, which is the principal impurity in Y-M. pattern R factor; RWP: the weighted pattern R factor. 2128 J. Mater. Chem., 1997, 7(10), 2127–2130Table 2 Final R values and refinement details for the diVerent distributions of N,O,Si,Al in Nd-M¾ (x=0.5) (l=1.470 A° ) 1 Al and 3 Si at (4e) sites 4 Si at (4e) sites 1 A1, 5 Si refined at 2 Si at (2a) sites 1 A1 and 1 Si at (2a) sites (4e, 2a) sites 0(2c) N(2c) N,O O(2c) N(2c) N,O O(2c) N(2c) N,O N(8f ) N,O(8f ) refined N(8f ) N,O(8f ) refined N(8f ) N,O(8f ) refined RF (%) 6.52 3.46 3.52 6.59 3.50 3.52 6.54 3.44 3.49 RB (%) 11.26 5.65 5.53 11.24 5.69 5.55 11.24 5.60 5.49 RP (%) 7.24 5.16 5.06 7.25 5.20 5.11 7.24 5.16 5.06 RWP (%) 9.27 6.81 6.71 9.28 6.84 6.74 9.26 6.81 6.70 Table 3 Final refinement results of atomic coordinates, isotropic thermal (B) and occupancy (N) parameters in Y-M and Nd-M¾ (x=0.5) (l=1.470 A ° ) Nd-M¾ (N, O refined at (2c, 8f ) sites) Y-M WyckoV N, O refined 1 Al, 3 Si at 1 Al, 1 Si at 1 Al, 5 Si refined atom notation x,y,z, B, N at (2c, 8f ) sites (4e) sites (2a) sites at (4e, 2a) sites Y/Nd 4e x 0.3365(4) 0.3361(4) 0.3359(4) 0.3360(4) y 0.1635(4) 0.1639(4) 0.1641(4) 0.1640(4) z 0.5038(8) 0.5019(9) 0.5022(9) 0.5020(9) N 0.5 0.5 0.5 0.5 Si(1)/Al(1) 4e x 0.1456(8) 0.1429(9) 0.1423(8) 0.1427(9) y 0.3544(8) 0.3571(9) 0.3577(8) 0.3573(9) z 0.9464(13) 0.9473(14) 0.9474(14) 0.9473(14) Si(1) N 0.5 0.375 0.5 0.407(26) Al(1) N 0.125 0.093(26) Si(2)/Al(2) 2a x 0.0 0.0 0.0 0.0 y 0.0 0.0 0.0 0.0 z 0.0 0.0 0.0 0.0 Si(2) N 0.25 0.25 0.125 0.218(26) Al(2) N 0.125 0.032(26) N(1)O(1) 2c x 0.5 0.5 0.5 0.5 y 0.0 0.0 0.0 0.0 z 0.1882(12) 0.1785(11) 0.1783(11) 0.1784(11) N(1) N 0.213(6) 0.211(6) 0.212(6) 0.212(6) O(1) N 0.037(6) 0.039(6) 0.038(6) 0.038(6) N(2)/O(2) 8f x 0.0852(3) 0.0847(3) 0.0849(4) 0.0848(3) y 0.1600(4) 0.1641(4) 0.1641(4) 0.1641(4) z 0.7976(6) 0.8047(6) 0.8049(6) 0.8047(6) N(2) N 0.787(6) 0.664(6) 0.663(6) 0.663(6) O(2) N 0.213(6) 0.336(6) 0.337(6) 0.337(6) O(3) 4e x 0.1416(4) 0.1397(5) 0.1396(5) 0.1397(5) y 0.3584(4) 0.3603(5) 0.3604(5) 0.3603(5) z 0.2783(11) 0.2842(11) 0.2786(11) 0.2784(11) N 0.5 0.5 0.5 0.5 Ba 0.83(4) 0.24(4) 0.24(4) 0.24(4) a Isotropic thermal parameter, see text that ca. 1.7 N and 0.3 O atoms occupy 2c sites in both Y-M and Nd-M¾ units in spite of the fact that the total numbers of Table 4 Some selected interatomic distances in Y-M and Nd-M¾ N atoms are eight and seven in Y-M and Nd-M¾ respectively; the remaining N atoms in the unit cells are disordered to Y-M (N, O refinedNd-M¾ (N, O, Al, occupy 8f sites.at 2c, 8f sites) Si refined) It was diYcult to distinguish between Al and Si atoms in M(Y/Nd)M[N( l ) O(1)] 2.346(5) 2.426(5) Nd-M¾ because of the small diVerence in scattering factors, as MO(3) 2.360(5) 2.428(6) stated in our previous work.4 However, the atomic ratio of M[N(2), O(2)] (×2) 2.397(4) 2.472(5) Al5Si at 4e sites (0.1950.81) appears to be marginally higher MO(3) (×2) 2.551(5) 2.607(5) than the one (0.1350.87) at 2a sites.M[N(2), O(2)](×2) 2.754(4) 2.806(5) average 2.514(5) 2.578(5) This work was partly supported by a Royal Society Joint (Si/Si,Al)MO(3) 1.661(8) 1.669(9) M[N(1), O(1)] 1.702(7) 1.687(7) project between Shanghai Institute of Ceramics and the M[N(2), O(2)] (×2) 1.714(7) 1.720(8) Nitrogen Ceramics Research Group in the University of average 1.698(7) 1.699(7) Newcastle, UK. We also acknowledge financial support from the National Natural Science Foundation of (Si/Si,Al)M[N(2), O(2)] (×4) 1.701(3) 1.737(3) China.J. Mater. Chem., 1997, 7(10), 2127–2130 21296 P. L. Wang, H. Y. Tu, W. Y. Sun, D. S. Yan, M. Nygren and References T. Ekstro�m, J. Eur. Ceram. Soc., 1995, 15, 689. 1 S. Slasor, K. Liddell and D. P. Thompson, Br. Ceram. Proc., 1986, 7 K. E. Johansson, T. Palm and P.-E. Werner, J. Phys. E.: Sci. 37, 51. Instrum., 1980, 13, 1289. 2 P. L. Wang, W. Y. Sun and T. S. Yen, (D. S. Yan,), Mater. Res. Soc. 8 P.-E.Werner, Ark. Kemi, 1969, 31, 513. Symp. Proc., ed. I.-W. Chen, P. F. Becher, M. Mitomo, G. Petzow 9 D. B. Wiles, A. Sakthivel and R. A. Young, User’s Guide to Program and T. S. Yen, MRS Pittsburgh, PA, 1993, vol. 287, p. 387. DBW32s for Rietveld Analysis of X-Ray and Neutron Powder 3 A. W. J. M. Rae, D. P. Thompson, N. J. Pipkin and K. H. Jack, DiVraction Pattern, Version 8804, School of Physics, Institute of Special Ceram., 1975 6, 347. Technology, Atlanta, USA, 1988. 4 P. L. Wang and P.-E. Werner, J.Mater. Sci., 1997, 32, 1925. 5 A. Koroglu, D. C. Apperley, R. K. Harris and D. P Thompson, Paper 7/02842C; Received 25th April, 1997 J.Mater. Chem., 1996, 6, 1031. 2130 J. Mater. Chem., 1997, 7(10), 2127–21