Structural elucidation of Β-(Y,Sc) 2Si 2O 7: Combined use of 89Y MAS NMR and powder diffraction

1 Structural Elucidation of β-(Y,Sc)2Si2O7: Combined use of 45Sc and 89Y MAS-NMR and powder diffraction Allix M.,1,2 Alba M.D.,3 Florian P.,1,2 Fernández A.J.,3 Suchomel M.,4 Escudero A.,3 Suard E.5 and Becerro A.I.3 * 1CNRS, UPR3079 CEMHTI, 1D avenue de la Recherche Scientifique, 45071 Orléans cedex2, France. 2Université d'Orléans, Avenue du Parc Floral, BP 6749, 45067 Orléans cedex 2, France. 3Instituto de Ciencia de Materiales de Sevilla (CSIC-Universidad de Sevilla). c/ Américo Vespucio, 49. 41092 Sevilla (Spain). 4Argonne National Laboratory, Advanced Photon Source, Argonne, IL 60439 USA. 5Institut Max Von Laue Paul Langevin, F-38042 Grenoble 9, France. Abstract Although the structures of pure Sc2Si2O7 and β-Y2Si2O7 have been described in the literature using the C2/m space group, 29Si MAS NMR measurements of the intermediate members of the Sc2Si2O7 - β-Y2Si2O7 system indicate a lowering of the symmetry to the C2 space group. Indeed, these compositions exhibit a unique Si crystallographic site and a SiOSi angle lower than 180° non compatible with the C2/m space group. C2 is the only alternative possible space group as Cm can be discarded regarding to its two different Si sites per unit cell. Moreover, 89Y and 45Sc MAS NMR data have revealed the existence of two different RE (RE= Sc, Y) sites in the structure of the intermediate members of the Sc2Si2O7 - β-Y2Si2O7 system, confirming the lowering of the symmetry to the C2 space group. The viability of the C2 model has then been tested and confirmed by refinement of synchrotron and neutron powder diffraction data for the different members of the system. The structural evolutions across the Sc2Si2O7 - β-Y2Si2O7 system are discussed. Keywords Y2Si2O7, Sc2Si2O7, rare earth silicates, 45Sc and 89Y MAS NMR spectroscopy, synchrotron, neutrons diffraction. * Corresponding author: [email protected] Tel no: +34 954489545; Fax no: +34 954460665 2 1.- Introduction Rare Earth disilicates (RE2Si2O7) exhibit different polymorphic forms depending on the RE ionic radius, temperature and pressure (Ito and Johnson, 1968; Felsche, 1973). Y2Si2O7 shows, in particular, up to five polymorphs versus temperature at room pressure (y, , β,  and , also called respectively y, B, C, D and E,) while Sc2Si2O7, or thortveitite, exhibits a unique polymorph ( also called C) up to the melting point of the compound. A remarkable feature of the thortveitite structure is the unusual SiOSi angle, reported as 180º (Cruickshank et al. 1962; Smolin et al. 1973; Bianchi et al. 1988). There has been a considerable amount of discussion about the correctness of this structural model as the value of 180º is rather unusual for sorosilicates. Such angles are usually much smaller (130-140º; Liebau, 1986). The preference for a nonlinear structure with regard to the Si- O-Si bridge is supported by quantummechanical calculations (Meagher et al., 1979; Gibbs et al. 1981): the energy of a bent Si-O-Si link is shown to be smaller than for the linear (180°) counterpart. This discussion was also dealing with the question of whether the correct space group for thortveitite is C2/m or C2, both of them are possible on the basis of the diffraction symmetry. However, it was concluded that the correct space group for Sc2Si2O7 is the centrosymmetric C2/m, as it provides the most consistent bond lengths and angles, in spite of a SiOSi angle of 180º (Cruickshank et al. 1962; Smolin et al. 1973; Bianchi et al. 1988). The structure of β-Y2Si2O7 was studied by Redhammer and Roth (2003), who initially refined it using the C2 space group with acceptable residual values. However, the thermal parameters values for two of the four O atoms were definitely non-positive. Validation tests of the final C2 structural model clearly showed the presence of additional symmetry (mirror plane), suggesting that C2/m was the correct space group. Thus, a new structural resolution was undertaken, revealing the known thortveitite structure type of β-Y2Si2O7 and yielding better final residual values (with fewer refined parameters) and positive definite displacement parameters for all atoms. The Sc2Si2O7-β-Y2Si2O7 system has been studied previously as a function of temperature and composition (Escudero et al. 2007). The particular behaviour of this system annealed at 1300ºC was analysed, showing a complete solid solubility of one member into the 3 other, with the β-(Sc,Y)2Si2O7 structure type. Structural refinements for every member of the system were performed using the C2/m space group, as established in the literature for the end-members. All unit cell parameters were linear with composition, but the evolution of the c and β unit cell parameters did show a slope change at x~1.0. Likewise, SiOSi angles were obtained for the intermediate compositions of the system from 29Si MAS NMR data (Escudero et al. 2007). These angles varied with composition, showing a minimum of 170.8º at x~ 1.0. This result implied that C2/m is not the space group of the intermediate members, as it forces the SiOSi angle to be equal to 180º. Alternative possible space groups for the intermediate members are C2 and Cm. Both of them allow SiOSi angles to differ from 180º, as suggested by the 29Si MAS NMR data. However, Cm generates two different Si crystallographic sites, in opposition with the single Si site observed on the 29Si MAS NMR spectra of the intermediate members (Escudero et al. 2007). On the other hand, the differences between the C2/m and C2 structural models are very small. Both of them generate a unique Si crystallographic site. However, the C2/m model has a unique RE (RE= Sc, Y) site while in C2, two different RE sites exist, permitting a higher degree of distortion in the C2 structure for the SiO4 tetrahedron. In addition, there is a supplementary oxygen site in the C2 model. The aim of the present study is to analyse the structures across the Sc2Si2O7 – β-Y2Si2O7 system using complementary techniques that allow studying the structure at long and short range ordering. 45Sc and 89Y MAS NMR spectroscopy has been used to analyse the local environments of both nuclei. The MAS-NMR spectra inform on the number of RE crystallographic sites of the structures. In a second stage, and based on the NMR results, synchrotron and neutron powder diffraction refinement have been used to propose a structural model for the intermediate members of the system. 2.- Experimental 2.a.- Synthesis: The sol-gel route used for this study was derived from the synthesis of a well-homogenized gel of Y2Si2O7 (Díaz et al. 2001). The starting materials were Y(NO3)3·6H2O (99.9% Sigma), Sc(NO3)3·3H2O (99.999% Sigma), Si(OC2H5)4 (TEOS, 98% solution Sigma) and 96% ethanol. A TEOS solution in ethanol (1:3 in volume) was added over appropriate amounts of Y(NO3)3·6 H2O and Sc(NO3)3·3H2O previously 4 dissolved in 5 mL ethanol for the preparation of Sc2-xYxSi2O7 members with x = 0, 0.50, 1.00, 1.50, and 2.00. The mixtures were stirred at 40ºC until transparent gels were obtained. The gels were dried at 60ºC for 24 hours in air. Nitrates were eliminated by calcination at 500ºC for 1 hour at a heating rate of 1ºC·min-1. The white powder obtained was subsequently calcined at a heating rate of 5°C·min-1 up to 1300ºC for 24 hours and slowly cooled down to room temperature. These conditions allowed isolating the β- (Sc,Y)2Si2O7 polymorph for all the solid solution compositions (Escudero et al. 2007). (Y,Sc)2SiO5 and SiO2 were obtained in some compositions as subproducts. 2.b.- Characterization: Magic Angle Spinning Nuclear Magnetic Spectroscopy: Magic Angle Spinning Nuclear Magnetic Spectroscopy: 45Sc MAS experiments were carried on a Bruker Avance WB 750 MHz (17.6 T) spectrometer operating at 182.22 MHz. Powder samples were packed into 2.5mm zirconia rotors and spun at 30 kHz in a double resonance Bruker probe. The 45Sc chemical shifts were referenced relative to a 1 M Scandium chloride solution with 3% HCl added. MAS experiments were obtained at a radio-frequency field rf of 25 kHz using 0.5 µs pulse width (i.e. smaller than /24 for quantitativity (Samoson et al. 1981)). The spin-lattice relaxation time T1 was estimated to be 7 s with a saturation-recovery experiment and the recycle delays were set accordingly (Ernst et al. 1987). A spectral width of 2.5 MHz was used to avoid folding the spinning sidebands back into the spectrum, baseline correction being applied manually afterwards to remove the baseline oscillation induced by the 4.5 µs dead-time. The 45Sc 3Q MQMAS (Frydman et al. 1995, Medek et al. 1995) NMR experiments were performed spinning at 30.03 kHz in a 2.5 mm rotor and using a shifted-echo pulse sequence (Massiot et al. 1996). Excitation and mixing were done at rf =150 kHz with pulse lengths of 1.2 µs and 0.4 µs respectively, the selective  pulse being applied at rf = 7.5 kHz, the echo shift was 45 rotor periods (i.e. 1.5 ms and full-echo acquisition) and the recycle delay 5 s. In order to avoid sidebands in the indirect dimension the t1 time increment was set to the rotor period 33.3 µs (Massiot 1996) and typically 30 t1 increments were needed. 5 The 89Y MAS experiments were performed on a Bruker Avance WB 300 MHz (7.0 T) operating at 14.7 MHz. Powdered samples were packed into 7mm silicon nitride rotors and spun at 4 kHz in a double resonance “low-” Bruker probe specifically design to work at low frequency. The 89Y chemical shifts were referenced relative to a 1 M yttrium chloride solution. The spin-lattice relaxation time T1 of Y2Si2O7 was estimated to be 1300 s with a saturation-recovery experiment and hence for this compound a Hahn-echo experiment was used for a /2 pulse of 39 µs, a recycle delay of 1650 s (Ernst et al. 1987) and 96 scans accumulated. We took advantage of the line broadening observed for Sc2- xYxSi2O7 x = 1.5, 1.0, 0.5 and used a rotor-synchronized CPMG (Carr et al. 1954, Cheng et al. 1989, Cowans et al. 1993, Hung et al. 2004) sequence where trains of whole spin echoes are recorded to obtain an increase in signal-to-noise ratio with respect to a regular Hahn-echo sequence. We used /2 pulses of 39 µs, acquiring between 30 and 60 echoes separated by 28 rotor periods (7.0 ms) with a recycle delay of 700 s and summing between 100 and 400 transients. The echo trains were not directly Fourier-transformed but rather the echoes were separated, convoluted with an adequate Gaussian line broadening, Fourier-transformed and summed to obtain the final spectra. Diffraction techniques: High-intensity and high-resolution synchrotron powder XRD data have been recorded on the 11-BM diffractometer at the Advanced Photon Source (APS), Argonne National Laboratory. Data were collected over the 0.5-55° 2θ range with a 0.001°step size at room temperature using a wavelength of =0.413581Å. The sample was contained in a 0.5 mm capillary and was spun at 600 Hz during data collection. Neutron powder diffraction patterns were recorded on the D2B high-resolution/high-flux powder diffractometer at the Institut Laüe-Langevin in Grenoble, France. The sample was packed in a vanadium can. Data were acquired with a =1.594Å wavelength at 2θ intervals of 0.05° over the 10º< 2θ < 160° angular range. Diffraction patterns were analysed with the Rietveld method using the GSAS software (Larson and Von Dreele, 1994). Refined parameters were: background coefficients, unit cell and profile parameters, atomic coordinates, site occupation factors, isotropic temperature factors and phase fractions. Anisotropic temperature factors were used for the refinement of the neutron data. Structural refinements have been performed using both 6 C2 and C2/m space groups. In the C2 case, starting parameters were those given by Smolin et al. (1971) for β-Yb2Si2O7. The Y/Sc ratios have been refined on the two sites, using constraints on the composition: Occ(Y1+Y2) (and Occ(Sc1+Sc2)) have been constrained to their nominal values in each sample. In the C2/m case, the starting parameters for the refinements of the YxSc2-xSi2O7 structures with x ≤ 1 were taken from those reported for pure β-Sc2Si2O7 (Smolin et al. 1972) while the refinement of the YxSc2- xSi2O7 structure with x = 1.5 was carried out from the parameters reported for pure β- Y2Si2O7 (Redhammer and Roth, 2003). (Y,Sc)2SiO5 and SiO2 have been added as secondary phases (no SiO2 has been detected for the YScSi2O7 composition). 3.- Results 3.1.- Nuclear Magnetic Resonance results 3.1.1.- 89Y MAS NMR results Figure 1 (left) shows the experimental 89Y MAS NMR spectra of β-YxSc2-xSi2O7 samples with x = 0.5, 1.0, 1.5 and 2.0, which have been registered using CPMG technique to get a high S/N ratio. The spectrum of the Y-rich end-member, β-Y2Si2O7 (Figure 1a), displays a narrow peak at 207.2 ppm with FWHM = 2.3 Hz, in good agreement with the literature (Becerro et al., 2004). The presence of a unique narrow 89Y signal of β-Y2Si2O7 indicates the existence of a unique Y crystallographic site in the structure, compatible with space group C2/m and a SiOSi angle of 180º, in agreement with previous structural studies on this compound (Redhammer and Ross, 2003). The spectra of the rest of compositions (Figures 1b-d) exhibit broad and asymmetric bands that point to the presence of more than one Y resonance. This result implies the existence of more than one Y crystallographic site in the unit cell of the intermediate members and points to a C2 space group, which contains two different RE crystallographic sites in the unit cell. The maximum of the band shifts towards higher frequency as the Y content decreases. The spectrum of β-Y0.5Sc1.5Si2O7 shows, in addition, two low intensity signals at 155.6 and 240.7 ppm corresponding to Y in the X2-(Sc,Y)2SiO5 phase, which is a subproduct of the synthesis, as commented in the Experimental section. These signals are also present in the rest of the intermediate compositions spectra, but are not clearly observed because of the high FWHM of the resonances. 7 The 89Y MAS CPMG spectra have been simulated (Figure 1, right) with the resonances corresponding to the expected Y sites in the structure (two sites for the intermediate compositions and a unique site for the end-member). The results are displayed in Table 1. The simulation included two additional resonances accounting for two Y sites in X2- (Sc,Y)2SiO5 (Becerro et al. 2004). The 89Y chemical shift of one of the Y sites of β- (Y,Sc)2Si2O7 does vary linearly with composition (open circles in Figure 2), and the chemical shift of the single Y site in pure β-Y2Si2O7 is located in the extrapolation of that line. The second Y site does not show a linear dependence with composition (solid circles in Figure 2). On the other hand, the area under the curve of each resonance indicate that Sc2-xYxSi2O7 members with x=1.0 and x= 1.5 show a homogeneous distribution of Y between both sites, while the x= 0.5 composition shows a preferential occupation of Y for one of the crystallographic sites, in good agreement with diffraction results, as will be shown below. 3.1.2.- 45Sc MAS NMR results Figure 3a and 3b show the two-dimensional 45Sc (3Q) MAS NMR spectra of samples YxSc2-xSi2O7 with x = 0.0 and x = 1.0, respectively. The number of the different Sc sites has been extracted from the isotropic dimension of these 45Sc MQMAS spectra. The isotropic dimension of the MQMAS spectrum of pure β-Sc2Si2O7 (Figure 3c) shows a single contribution compatible with a unique Sc site in the unit cell and a C2/m space group, in agreement with the structural studies of this compound (Cruickshank et al. 1962; Smolin et al. 1973; Bianchi et al. 1988). However, two Gaussian lines were necessary to fit the isotropic spectrum of the β-ScYSi2O7 composition (Figure 3d), which are likely due to two different Sc sites. This result implies a lowering of the symmetry for this intermediate composition which allows the existence of two Sc sites, in agreement with the 89Y NMR results. The unique possibility conserving the unit cell metric is the C2 space group. Figure 4 shows the single pulse 45Sc MAS NMR spectra of the YxSc2-xSi2O7 samples. Adding Y to Sc2Si2O7 progressively smoothen the characteristic discontinuities of the Sc NMR lineshapes, indicating a progressive increase of disorder. The SP 45Sc MAS NMR spectra have been simulated with the resonances corresponding to the expected Sc sites 8 in the structure (two sites for the intermediate compositions and a unique site for the endmember), and the results (isotropic chemical shifts, quadrupolar coupling constants, asymmetry parameter and area under the curve of each resonance) are displayed in Table 2. The areas under the curve show a homogeneous distribution of Sc in both sites for x ≥ 1.0, in agreement with the 89Y MAS NMR data, and the diffraction data, as shown below. On the other hand, Figures 5a and 5b show the 45Sc weighted mean isotropic chemical shift and quadrupolar coupling constants values, respectively, versus composition, obtained from the simulation of the MQMAS spectra. Both parameters decrease linearly with increasing Y content, as expected in the case of a solid solution of β-Y2Si2O7 in Sc2Si2O7. 3.2.- Diffraction results Based on the 89Y and 45Sc NMR data described above which suggest the existence of two different RE crystallographic sites, as well as on the 29Si MAS NMR data published in Escudero et al. (2007), which indicate the existence of a unique Si crystallographic site coupled to a non linear SiOSi angle, C2 appears to be the only alternative possible space group. In fact, Cm, giving the same reflection conditions, can be discarded regarding to its two different Si sites per unit cell. The viability of the C2 space group, compatible with these NMR data, will now be demonstrated by refinement of the structure under this space group from synchrotron and neutron powder diffraction patterns. Refinements using C2/m have also been included for a comparison purpose. 3.2.1.- Study of the intermediate members of the system 3.2.1.1.- Synchrotron powder diffraction Structural refinements have been performed for YxSc2-xSi2O7 compositions with x = 0.5, 1.0 and 1.5, using both C2 and C2/m space groups models. Figure 6 shows the experimental and fitted patterns for Y0.5Sc1.5Si2O7, using space groups C2 (fig. 1a) and C2/m (fig. 1b). The fittings of the two other compositions are very similar and are not shown. Tables 3, 4 and 5 show the atomic parameters obtained for the three compositions using space groups C2 and C2/m. Refinements using C2 space group indicate a preferential occupation of Y and Sc for one of the RE sites in the Sc2-xYxSi2O7 with x= 0.5 composition, while the two other samples show a homogeneous distribution of Sc and Y between both crystallographic sites, in good agreement with the NMR data (asterisks 9 in Tables 3, 4 and 5 indicate the site occupation factors expected in case of homogeneous distribution of Sc and Y between both RE sites). Figures 7a and 7b are views of the structures obtained for YScSi2O7 using space groups C2 and C2/m. SiOSi angles obtained from the refinements in C2 were respectively 164.4(3)º, 167.2(3)º and 166.4(2)º for Y0.5Sc1.5Si2O7, YScSi2O7 and Y1.5Sc0.5Si2O7 compositions. The three values are lower than 180º, as calculated from 29Si NMR data (Escudero et al. 2007). Figure 7c is a Fourier map obtained by summing the Fobs from 0.35 to 0.65 along z of the YScSi2O7 pattern. The deviation of the SiOSi angle from 180º is clearly observed. These results demonstrate that the refinements using the C2 space group are possible and stable for the intermediate compositions. Reliability factors are quite good in both space groups (Table 6) although, in terms of fit quality, there is no real difference between the two space group refinements. Thus, synchrotron XRD data refinements indicate that the C2 model is possible for the intermediate members of the β-(Sc,Y)2Si2O7 system, as suggested by 89Y and 45Sc MAS NMR data. 3.2.1.2.- Neutron powder diffraction Neutron diffraction data refinements performed on YScSi2O7 composition were also possible in both C2 and C2/m space groups. Figure 8 shows the experimental and fitted patterns for YScSi2O7, using space groups C2 (fig. 8a) and C2/m (fig. 8b). Reliability factors (Table 6) are quite good in both space groups but, again, in terms of fit quality, there is no real difference between the two space group refinements. Table 7 shows the atomic parameters obtained for this composition. The SiOSi angle obtained from the refinements in C2 was 169.2(17)º, in good agreement with the synchrotron data. 3.2.2. Study of the end-members of the system Although the set of 89Y and 45Sc MAS NMR and diffraction data on the intermediate members strongly suggest a C2 model, an extra argument to ensure that the non 180º values of the SiOSi angles of the intermediate compositions are real and probably linked to the C2 space group would be to obtain angle values close to 180º for the Sc2Si2O7 and β-Y2Si2O7 end members when refined in C2, even if it has been proved that this is not the 16 Figure 6: Experimental (dots) and fitted (solid line) synchrotron patterns recorded on the 11BM station (APS, USA), for Y0.5Sc1.5Si2O7, using a) space group C2 and b) space group C2/m. The difference curves are also included. 17 Figure 7: a) and b): views of the structures of YScSi2O7 using, respectively, space groups C2 and C2/m. Data obtained from structure refinement of synchrotron data. c) Fourier map of YScSi2O7 along z; deviation of the SiOSi angle from 180º is clearly observed. 18 19 Figure 8: Experimental (dots) and fitted (solid line) neutron patterns recorded at the D2B station (ILL, Grenoble, France), for YScSi2O7, using a) space group C2 and b) space group C2/m. The difference curves are also included. Table 2. 45Sc SP MAS NMR fit parameters for β- YxSc2-xSi2O7 x β-YxSc2-xSi2O7 δiso (ppm) CQ (MHz) η Area under the curve (%) 0.0 77.8 20.70 0.45 100 0.2 81.2 16.00 0.01 7.9 75.3 20.95 0.43 92.1 0.4 77.1 16.47 0.23 16.9 72.3 20.90 0.44 83.1 0.6 74.1 16.32 0.34 20.7 70.1 20.26 0.47 79.3 1.0 69.6 16.22 0.60 56.9 66.0 20.97 0.60 43.1 1.5 63.8 14.94 0.60 42.4 60.3 18.54 0.60 57.6 1.7 62.5 14.08 0.60 19.2 58.9 16.92 0.60 80.9 Table 1. 89Y MAS NMR fit parameters for β-YxSc2-xSi2O7 x (β-YxSc2-xSi2O7) δ (ppm) ∆1/2 (ppm) Area under the curve (%) 0.5 199.9 64.0 34.6 225.0 38.5 65.4 1.0 196.9 65.9 50.2 220.9 34.7 49.8 1.5 207.8 35.3 51.6 217.1 23.5 48.4 100 207.2 2.3 100 20 Table 3a. Refined atomic coordinates for Y0.5Sc1.5Si2O7 from synchrotron powder diffraction (11BM-APS) data collected at RT (space group C2; a=6.62050(6)Å; b=8.61152(7)Å; c=4.67462(3)Å and =102.366(1)°). Asterisk indicates site occupation factors expected in case of homogeneous distribution of Sc and Y between both RE sites. Atom Site x y Z Uiso (*100) Occ. Y1/Sc1 2b 0.5 0.8055 0 0.44(6) 0.335(4)/0.665(4) (0.25/0.75)* Y2/Sc2 2b 0.5 0.1935(1) 0 0.66(7) 0.165(4)/0.835(4) (0.25/0.75)* Si 4c 0.7205(1) 0.5034(4) 0.4109(1) 0.51(2) 1 O1 2a 0.5 0.4778(4) 0.5 1.23(2) 1 O2 4c 0.8854(2) 0.4992(7) 0.7188(2) 1.23(2) 1 O3 4c 0.7297(6) 0.6473(4) 0.2276(8) 1.23(2) 1 O4 4c 0.7428(6) 0.3392(4) 0.2183(8) 1.23(2) 1 Table 3b. Refined atomic coordinates for Y0.5Sc1.5Si2O7 from synchrotron powder diffraction (11BM-APS) data collected at RT (space group C2/m; a=6.62047(6)Å; b=8.61153(7)Å; c=4.67465(3)Å and =102.366(1)°). Atom Site x y Z Uiso (*100) Occ. Y/Sc 4g 0 0.30614(4) 0 0.52(1) 0.25/0.75 Si 4i 0.2212(1) 0 0.4107(1) 0.52(2) 1 O1 2c 0 0 0.5 1.48(2) 1 O2 4i 0.3847(2) 0 0.7192(2) 1.48(2) 1 O3 8j 0.2367(1) 0.1539(1) 0.2233(1) 1.48(2) 1 21 Table 4a. Refined atomic coordinates for YScSi2O7 from synchrotron powder diffraction (11BM-APS) data collected at RT (space group C2; a=6.71363(3)Å; b=8.72253(5)Å; c=4.67973(3)Å and =102.084(1)°). Asterisk indicates site occupation factors expected in case of homogeneous distribution of Sc and Y between both RE sites. Atom Site x y Z Uiso (*100) Occ. Y1/Sc1 2b 0.5 0.76960 0 0.66(6) 0.50(1)/0.50(1) (0.50/0.50)* Y2/Sc2 2b 0.5 0.15906(5) 0 0.40(6) 0.50(1)/0.50(1) (0.50/0.50)* Si 4c 0.72062(7) 0.4663(4) 0.4131(1) 0.64(1) 1 O1 2a 0.5 0.4871(3) 0.5 1.30(2) 1 O2 4c 0.8809(1) 0.4659(7) 0.7163(2) 1.30(2) 1 O3 4c 0.7534(3) 0.6204(3) 0.2227(6) 1.30(2) 1 O4 4c 0.7161(3) 0.3164(3) 0.2213(6) 1.30(2) 1 Table 4b. Refined atomic coordinates for YScSi2O7 from synchrotron powder diffraction (11BM-APS) data collected at RT (space group C2/m; a=6.71352(3)Å; b=8.72241(5)Å; c=4.67975(3)Å and =102.084(1)°). Atom Site x y Z Uiso (*100) Occ. Y/Sc 4g 0 0.30527(2) 0 0.51(1) 0.5/0.5 Si 4i 0.22097(6) 0 0.41264(9) 0.71(1) 1 O1 2c 0 0 0.5 1.81(2) 1 O2 4i 0.3817(1) 0 0.7165(2) 1.81(2) 1 O3 8j 0.2345(1) 0.1523(1) 0.2225(1) 1.81(2) 1 22 Table 5a. Refined atomic coordinates for Y1.5Sc0.5Si2O7 from synchrotron powder diffraction (11BM-APS) data collected at RT (space group C2; a=6.80085(7)Å; b=8.85057(9)Å; c=4.70109(4)Å and =101.854(1)°). Asterisk indicates site occupation factors expected in case of homogeneous distribution of Sc and Y between both RE sites. Atom Site x y Z Uiso (*100) Occ. Y1/Sc1 2b 0.5 0.8244 0 0.68(10) 0.71(1)/0.29(1) (0.75/0.25)* Y2/Sc2 2b 0.5 0.2143(1) 0 0.37(9) 0.79(1)/0.21(1) (0.75/0.25)* Si 4c 0.7204(1) 0.5223(10) 0.4144(2) 0.53(3) 1 O1 2a 0.5 0.5005(6) 0.5 1.21(4) 1 O2 4c 0.8726(2) 0.5185(21) 0.7094(3) 1.21(4) 1 O3 4c 0.7477(8) 0.6717(8) 0.2279(15) 1.21(4) 1 O4 4c 0.7222(8) 0.3714(8) 0.2204(15) 1.21(4) 1 Table 5b. Refined atomic coordinates for Y1.5Sc0.5Si2O7 from synchrotron powder diffraction (11BM-APS) data collected at RT (space group C2/m; a=6.80082(7)Å; b=8.85048(9)Å; c=4.70108(4)Å and =101.854(1)°). Atom Site x y Z Uiso (*100) Occ. Y/Sc 4g 0 0.30519(4) 0 0.51(1) 0.75/0.25 Si 4i 0.2203(1) 0 0.4142(2) 0.54(2) 1 O1 2c 0 0 0.5 1.49(3) 1 O2 4i 0.3735(2) 0 0.7105(3) 1.49(3) 1 O3 8j 0.2358(2) 0.1504(1) 0.2252(2) 1.49(3) 1 Table 6. Reliability factors obtained from structural refinements of powder synchrotron and neutron diffraction patterns. 23 Synchrotron diffraction (11BM) Neutron diffraction (D2B) Rwp Rp RF2 2 Rwp Rp RF2 2 Sc2Si2O7 C2 8.22% 6.42% 2.32% 7.57 Non recorded C2/m 8.27% 6.46% 2.53% 7.66 Y0.5Sc1.5Si2O7 C2 7.20% 5.32% 2.65% 6.49 Non recorded C2/m 7.34% 5.46% 2.85% 6.75 YScSi2O7 C2 4.25% 3.39% 1.43% 1.82 3.55% 2.84% 3.03% 1.75 C2/m 4.41% 3.55% 1.65% 1.97 3.59% 2.85% 3.09% 1.80 Y1.5Sc0.5Si2O7 C2 7.23% 5.37% 1.63% 4.90 Non recorded C2/m 7.30% 5.46% 1.76% 4.99 Y2Si2O7 C2 6.60% 5.16% 1.61% 4.01 3.64% 2.97% 2.43% 1.94 C2/m 6.65% 5.19% 1.64% 4.05 3.66% 2.99% 2.03% 1.95