Structure elucidation of BaSi2O2N2 – A host lattice for rare-earth doped luminescent materials in phosphor-converted (pc)-LEDs

Structure elucidation of BaSi2O2N2 – A host lattice for rare-earth doped luminescent materials in phosphor-converted (pc)-LEDs

Solid State Sciences 11 (2009) 537–543 Contents lists available at ScienceDirect Solid State Sciences journal homepage: www.elsevier.com/locate/sssc...

547KB Sizes 0 Downloads 3 Views

Solid State Sciences 11 (2009) 537–543

Contents lists available at ScienceDirect

Solid State Sciences journal homepage: www.elsevier.com/locate/ssscie

Structure elucidation of BaSi2O2N2 – A host lattice for rare-earth doped luminescent materials in phosphor-converted (pc)-LEDs Juliane A. Kechele, Oliver Oeckler, Florian Stadler, Wolfgang Schnick* ¨ t Mu ¨ nchen, Butenandtstrasse 5–13 (Haus D), 81377 Mu ¨ nchen, Germany Department Chemie und Biochemie, Ludwig–Maximilians–Universita

a r t i c l e i n f o

a b s t r a c t

Article history: Received 30 April 2008 Received in revised form 17 June 2008 Accepted 18 June 2008 Available online 26 June 2008

BaSi2O2N2 is a promising host lattice for rare-earth doped luminescent materials in phosphor-converted (pc)-LEDs. Applying a combined approach, its orthorhombic average structure (space group Cmcm (no. 63), a ¼ 14.3902(3) Å, b ¼ 5.3433(1) Å, c ¼ 4.83256(7) Å and V ¼ 371.58(2) Å3, Z ¼ 4) has been elucidated by electron diffraction and structure solution from X-ray and neutron powder diffraction data with subsequent Rietveld refinement (wRp ¼ 0.0491 for X-ray data). The structure contains layers of highly condensed SiON3 tetrahedra with O terminally bound to Si. The Ba2þ ions are situated between the layers and are surrounded by a cuboid of O atoms capped by two N atoms. In the structure, there is only one Ba site and one Si site, respectively, which is in accordance with a single sharp 29Si NMR signal observed at 52.8 ppm typical for SiON3 tetrahedra in MSi2O2N2 type oxonitridosilicates. Lattice energy calculations support the results of the structure determination. Ó 2008 Elsevier Masson SAS. All rights reserved.

Keywords: Oxonitridosilicate Electron diffraction Structure elucidation Neutron diffraction Powder diffraction

1. Introduction Owing to their chemical and thermal stabilities, oxonitridoalumosilicates (sialons) and oxonitridosilicates (sions) have received remarkable attention in materials science. Recently, oxonitridosilicates have emerged as host lattices for highly efficient rare-earth doped luminescent materials applied in phosphorconverted (pc)-LEDs [1]. In this context, several investigations of the luminescence properties of MSi2O2N2 (M ¼ Ca, Sr, Ba) [2–16] have been carried out and particular promising properties pointed out. Recently, the new efficient phosphor SrSi2O2N2:Eu2þ for application in phosphor-converted white light LEDs has been described. Typically, (pc)-LEDs consist of an InGaN-based blue LED chip coated with a yellow–green phosphor (e.g. SrSi2O2N2:Eu2þ) and a red phosphor (e.g. Sr2Si5N8:Eu2þ). In this context, the luminescence properties of BaSi2O2N2:Eu2þ emerged out to be specifically promising. This material stands out due to its small Stokes’ shift and its narrow emission band. However, these distinguished luminescence properties can merely be understood with the knowledge of the crystal structure of this compound. In the previous years we have already elucidated the structures of MSi2O2N2 (M ¼ Ca, Sr, Eu) and revealed the isotypism of EuSi2O2N2 and the structure of domains in disordered SrSi2O2N2. The Ca and Eu compounds have been analysed in detail by means of

* Corresponding author. Tel.: þ49 89 2180 77436; fax: þ49 89 2180 77440. E-mail address: [email protected] (W. Schnick). 1293-2558/$ – see front matter Ó 2008 Elsevier Masson SAS. All rights reserved. doi:10.1016/j.solidstatesciences.2008.06.014

single-crystal diffraction. As the crystal structure of SrSi2O2N2 turned out to exhibit numerous real structure effects, combined approaches to structure determination using X-ray and electron diffraction as well as HRTEM and other techniques have been employed. All three compounds are characterized by similar highly condensed silicate layers with three-connected nitrogen atoms. The oxygen atoms are terminally bound to the silicon atoms. These layers are separated by pseudohexagonal cation layers [17–19]. However, there is insufficient information about the crystal structure of BaSi2O2N2 so far [4,5,7]. In this contribution we report about the structure elucidation of this orthorhombic compound, applying a combined approach using electron diffraction, solidstate NMR, and both X-ray and neutron powder diffraction data. 2. Experimental section 2.1. Synthesis According to Eq. (1), 0.5 mmol (98.7 mg) of BaCO3 (Alfa Aesar, 99.95%) and 0.33 mmol (47.8 mg) of a-Si3N4 (UBE Industries Ltd., Tokyo, 98%) were mixed in an agate mortar and filled into a tungsten crucible under argon atmosphere in a glovebox (Unilab, MBRAUN; O2 < 1 ppm, H2O < 1 ppm).

(1)

538

J.A. Kechele et al. / Solid State Sciences 11 (2009) 537–543

Fig. 1. Electron diffraction patterns of BaSi2O2N2 with various zone axes, some show diffuse streaking. For cell determination, only patterns with no or weak streaks were used.

The crucible was heated to 1150  C with a rate of about under purified N2 in the reactor of a radio-frequency 20 furnace [20] before the temperature was increased to 1350  C during 5 h. This maximum temperature was kept for 1 h. Subsequently, the crucible was cooled down to 650  C with a rate of about 70  C h1 before quenching to room temperature by switching off the furnace. By this reaction, we obtained BaSi2O2N2 contaminated with different barium oxosilicates and Si3N4 as by-products. The oxosilicates could be removed by treatment with diluted hot HCl while silicon nitride remained as a minority phase during the diffraction measurements.  C min1

Samples were finely ground and images were taken from thin regions near crystallite edges. Images were processed and in some cases Fourier filtered using the program Digital Micrograph 3.8.1 (Gatan). The analysis and interpretation of the diffraction images and the computation of simulations were conducted by means of the EMS program package [21]. 2.4. Powder diffraction experiments

Elemental analyses of bulk samples were performed by energy dispersive X-ray spectroscopy (EDX) using a JSM-6500F scanning electron microscope (Jeol) equipped with a Si/Li EDX detector (Oxford Instruments, model 7418), whereas crystallites characterized by SAD (selected area diffraction) patterns were directly analysed using TEM in the nanoprobe mode with a Si/Li detector (Noran Vantage).

X-ray powder diffraction data were collected on a STOE STADI P diffractometer (Cu Ka1 radiation, Ge(111) monochromator). Timeof-flight (TOF) method neutron powder diffraction measurements have been carried out using GEM diffractometer at ISIS/Rutherford Appleton Laboratory, Chilton/UK. This instrument is specifically useful for the diffraction measurements of small sample sizes. An amount of 150 mg of BaSi2N2O2 contaminated with Si3N4 was enclosed in a vanadium tube for the neutron diffraction experiment. For the analysis of the data, the back-scattering bank at 2q ¼ 154.4 has been used as it is characterized by good resolution and a large d-spacing range. In both cases the GSAS program package [22] was chosen for pattern fitting (LeBail algorithm) and Rietveld refinements.

2.3. Electron diffraction

2.5. Solid-state NMR

The electron diffraction patterns (SAD) were recorded using a transmission electron microscope (TEM, Philips CM30/ST). The acceleration voltage was 300 kV, the spherical aberration constant, Cs, of the device was 1.15 mm and the point resolution was 0.19 nm.

Solid-state NMR measurements have been performed on a Bruker Avance DSX 500 spectrometer with an external magnetic field of 11.75 T. The 29Si 1D-MAS spectra were recorded with direct excitation using a commercial triple resonance MAS probe, which

2.2. Elemental analysis

Fig. 2. Crystallographic relationships between the different structure models for BaSi2O2N2. Cmc21 is a translationengleiche subgroup of Cmcm, Pbcn is a klassengleiche subgroup. In the space group Cmcm the nitrogen atoms are disordered. Thereby, the silicon atoms are coordinated by seven N atoms as a consequence of the overlap of two SiON3 tetrahedra with identical O atoms. Whole structure: view along [001], silicate layers: view along [100] (Ba2þ dark gray, O2 light gray, N3 black). The SiON3 tetrahedra as well as the unit cells are shown.

J.A. Kechele et al. / Solid State Sciences 11 (2009) 537–543

539

a detailed analysis revealed that all of the BaSi2O2N2 samples synthesized so far were contaminated with different kinds and amounts of by-products (see above). Accordingly, straightforward indexing of the powder pattern was not possible and reported lattice parameters of BaSi2O2N2 [4,5] turned out to be wrong. In order to circumvent these problems we have unequivocally determined the lattice parameters by electron diffraction. A multitude of diffraction patterns with different orientations of the single crystals have been recorded (cf. Fig. 1). The analysis of these data yielded an orthorhombic crystal system with the cell parameters a ¼ 14.44 Å, b ¼ 5.34 Å, c ¼ 4.83 Å, which do not agree with the lattice parameters of BaSi2O2N2 mentioned in the literature (monoclinic P lattice with a ¼ 14.070(4) Å, b ¼ 7.276(2) Å, c ¼ 13.181(3) Å, b ¼ 107.74(6) and V ¼ 1285.23 Å3) [4,5]. However, based on our smaller orthorhombic unit cell, all reflections that cannot be assigned to by-products can be indexed. The chemical composition of both the crystallites investigated by electron diffraction as well as parts of bulk samples studied in a scanning electron microscope has been confirmed by EDX measurements within the typical error ranges.

Fig. 3. Significant cut-outs of the calculated X-ray diffraction patterns, which demonstrate the exiguous differences between the patterns for the different structure models (Pbcn: light gray, Cmc21: dark gray, Cmcm: black). Cu Ka1 radiation.

was equipped with a rotor made of ZrO2 (diameter: 4 mm). A repetition delay of 16 384 s (>3 T1) and a rotation frequency of 9 kHz were chosen. The 90 impulse length was 2.5 ms. The given chemical shift values refer to TMS as an external chemical shift reference. 3. Results and discussion 3.1. Structure determination 3.1.1. Determination of the lattice parameters The recorded powder diffraction patterns of our products were quite similar to those reported in the literature [4,5]. However, Table 1 Crystallographic data of BaSi2O2N2 derived from Rietveld refinement of the X-ray data (at 293 K) Formula Formula mass (in g mol1) Crystal system Space group Lattice parameters (in Å)

Cell volume (in Å3) Formula unit (cell) X-ray density (in g cm3) F(000) Radiation Profile range (in  ) No. of data points Observed reflections Structural parameters Profile parameters Background parameters Other parameters Restraints Background function Profile function Rp/wRp Rp/wRp (without backgr.) RjFj2 GOF

BaSi2O2N2 253.51 Orthorhombic Cmcm (no. 63) Pbcn (no. 60) a ¼ 14.3902(3) a ¼ 14.3902(3) b ¼ 5.3433(1) b ¼ 5.34330(10) c ¼ 4.83256(7) c ¼ 4.83254(6) 371.58(1) 371.58(1) 4 4.53 456 Cu Ka1 (l ¼ 1.54056 Å) 5  2q  90 8500 178 237 20 (4 for Si3N4) 18 (4 for Si3N4) 13 (3 for Si3N4) 13 (3 for Si3N4) 36 36 4 4 7 4 Shifted Chebyshev Pseudo-Voigt 0.0354/0.0491 0.0331/0.0452 0.0471/0.0608 0.0417/0.0520 0.0583 0.0460 1.76 1.62

3.1.2. X-ray powder diffraction Space group determination from systematic absences in the powder pattern in combination with electron diffraction patterns was not straightforward, as overlap with reflections from impurity phases and sections through diffuse streaks, respectively, can suggest the violation of systematic absences. Therefore, a P lattice seemed likely but C centering could not be excluded. Furthermore, n glide planes parallel (010) and (001) seemed very likely, however, b parallel (010) could also not be excluded. The structure solution was attempted in various space groups consistent with these findings by direct methods using Sirware EXPO [23]. Reasonable positions for Ba, Si and O atoms were only obtained in space group Pnn2 (no. 34). At this level of structure determination, it turned out that BaSi2O2N2 has a layered structure. A comparison of the Si and O positions of BaSi2O2N2 with those of SrSi2O2N2 exhibited the presence of quite similar silicate layers. Thus, the positions of the bridging N atoms, which were not obvious from the structure solution, could be geometrically calculated. The resulting structure exhibited disordered nitrogen atoms and could as well be described in the space group Cmcm (no. 63). Due to the multiplicity of the single N site in this space group, possible ordered structures can only be obtained by lowering the symmetry. This results in two structure models with fully ordered N positions in the space groups Pbcn (no. 60) and Cmc21 (no. 36), respectively. The crystallographic

Fig. 4. Observed (crosses) and calculated (line) X-ray powder diffraction patterns as well as difference profile for the Rietveld refinements of BaSi2O2N2 and a-Si3N4 (Cu Ka1, l ¼ 1.5406 Å) (space group Cmcm).

540

J.A. Kechele et al. / Solid State Sciences 11 (2009) 537–543

Table 2 Crystallographic data of BaSi2O2N2 derived from Rietveld refinement of the neutron data (at 293 K) Cmcm Lattice parameters Detector position (2q) ( ) Observed reflections No. of refined parameters Atomic parameters Profile parameters Other parameters Restraints Structure refinement Background function Profile function Rp/wRp Rp/wRp (without backgr.) RjFj2 GOF

1637

Pbcn The lattice parameters determined from X-ray data were fixed 154.4 2322

9 11 2 2 (linked with Si3N4 parameters) 5 5 7 4 Rietveld refinement, GSAS [22] Shifted Chebyshev with 10 terms W.I.F. David (convolution of the Ikeda–Carpenter and Pseudo-Voigt functions) 0.0331/0.0461 0.0298/0.0413 0.0356/0.0528 0.0319/0.0486 0.05255 0.02080 2.63 2.36

relationships between these different structure models are shown in Fig. 2. These two structure models were refined by Rietveld methods and yielded approximately the same residuals as the disordered structure model. Although differences between the patterns are expected as a consequence of different lattice centering, they are very small (Fig. 3 represents significant cut-outs of the calculated patterns for all three structure models) as the structure models differ only concerning the position of the light atom N and cannot be detected unambiguously due to peak overlap. The final refinements turned out a slight tendency favouring the model in Pbcn. This is corroborated by R values (Rp(Cmcm) ¼ 0.0354; Rp(Pbcn) ¼ 0.0331; Rp(Cmc21) ¼ 0.0340) as well as difference Fourier syntheses that yielded one reasonable position for N in Pbcn but showed almost an equal disorder in Cmcm and Cmc21, although ordering would be possible in the latter space group. However, diffuse streaks in electron diffraction patterns indicate stacking disorder. Therefore, the disordered model probably represents the average structure of a disordered compound. The crystal data and structure refinements for the models in Cmcm and in Pbcn are summarized in Table 1. For both models the positional and isotropic displacement parameters are listed in Tables 3 and 4. The observed and calculated X-ray powder diffraction patterns for the average model in Cmcm (cf. Section 3.2.3), as well as the corresponding difference profile of the Rietveld refinement, are shown in Fig. 4. The isotropic displacement parameters for the oxygen and nitrogen atoms were constrained to be equal. Furthermore, the distances of Si–N were restrained to 1.72 Å and the distances of Si–O to 1.66 Å, allowing for variations of 0.01 Å. Further details of the crystal structure investigations are available from the Fachinformationszentrum Karlsruhe, D-76344

Table 3 Atomic coordinates and isotropic displacement parameters (in Å2) for the average structure of BaSi2O2N2 in Cmcm Atom

Wyckoff position

x

y

z

Ueq

Occupancy

Ba

4c 8g

O

8g

N

16h

0.2494(4) 0.280(4) 0.3372(7) 0.318(1) 0.235(2) 0.247(3) 0.1285(7) 0.106(2)

1/4 1/4 3/4 3/4 3/4 3/4 0.8969(4) 0.9060(9)

0.0176(4) 0.0212(8) 0.045(1) 0.0212(8) 0.012(1) 0.0212(8) 0.012(1) 0.0212(8)

1

Si

0 0 0.7957(2) 0.7930(4) 0.9045(3) 0.9054(5) 0.7192(4) 0.7201(6)

1 1 0.5

The standard deviations are given in parentheses (for each atom: first row X-ray, second row neutron data).

Table 4 Atomic coordinates and isotropic displacement parameters (in Å2) for the BaSi2O2N2 polytype in Pbcn Atom

Wyckoff position

x

y

z

Ueq

Occupancy

Ba

4c 8d

O

8d

N

8d

0.2497(4) 0.268(2) 0.3360(6) 0.3223(8) 0.231(2) 0.251(2) 0.1224(8) 0.1058(8)

1/4 1/4 0.709(1) 0.657(1) 0.740(2) 0.739(1) 0.860(1) 0.8152(10)

0.0171(4) 0.0103(3) 0.033(1) 0.0103(3) 0.003(1) 0.0103(3) 0.003(1) 0.0103(3)

1

Si

0 0 0.7979(2) 0.7955(3) 0.9057(2) 0.9044(3) 0.7244(4) 0.7245(3)

1 1 1

The standard deviations are given in parentheses (for each atom: first row X-ray, second row neutron data).

Eggenstein-Leopoldshafen (Germany), on quoting the depository number CSD-419451 (Cmcm) and CSD-419450 (Pbcn), the name of the authors, and citation of the paper. 3.1.3. Neutron powder diffraction The aforementioned structural models of BaSi2O2N2 in space groups Pbcn, Cmcm and Cmc21, respectively, mainly differ concerning the layer stacking sequences which are directly related to the N positions. Therefore, neutron powder diffraction patterns of BaSi2O2N2 were recorded. The analysis of the data was difficult due to the contamination with Si3N4, whose reflections exhibit a much higher relative intensity than in X-ray powder diagrams. As the peak overlap between Si3N4 and BaSi2O2N2 is severe, the atomic coordinates of Si3N4 as well as lattice parameters were not refined. The different models of BaSi2O2N2 were evaluated through Rietveld methods, whereas the reflections of Si3N4 were LeBail fitted using F (calc) weighted intensities. The refinements indicated a small amount of a further by-product, which has not been detectable with X-rays. However, reflection overlap impeded the characterization of this by-product. Nevertheless, the analysis of these data favours the structure model with the space group Pbcn. As certain degree of the nitrogen atom disorder, i.e. layer stacking disorder as indicated by diffuse streaks in electron diffraction patterns, cannot be excluded by the analysis of the neutron data, the structure refinements of the neutron data are summarized in Table 2 for the models in Cmcm and in Pbcn. The positional and isotropic displacement parameters for both models are summarized in Tables 3 and 4 for the neutron and X-ray diffraction data. For the refinement based on neutron data, displacement parameters of all atoms were set equal. Similar to the analysis of the X-ray data, the distances Si–N and Si–O atoms were restrained. 3.1.4. Solid-state NMR As the obtained samples differed concerning the kinds and amounts of by-products, solid-state NMR measurements were carried out on different samples. In addition to the signals of the different by-products, all spectra exhibit one strong and sharp signal at 52.8 ppm. This chemical shift is in the same range as the

Table 5 Results of the MAPLE calculations (in kJ/mol) for BaSi2O2N2 polytypes and increment calculations: partial MAPLE values, total MAPLE sums, difference

Cmc21 Pbcn

Ba2þ

Si4þ

O2

N3

Total MAPLE

D (%)

1855 1843

9483 9513

2172 2175

6243 6237

37 652 37 695

0.03 0.14

Total MAPLE (0.5 Ba2SiO4 þ 0.5 Si3N4): 37 642 kJ/mol Typical MAPLE values (in kJ/mol): Ba2þ: 1500–2000; Si4þ: 9000–10 200; N[3]3: 5200–6300; O[1]2: 2050–2800 [24].

J.A. Kechele et al. / Solid State Sciences 11 (2009) 537–543

541

Fig. 5. Structure and silicate layers of MSi2O2N2 (M ¼ Ca, Sr, Ba). For clarification the zigzag lines, tetrahedra with vertices up are depicted in dark gray, with vertices down in light gray (N3: black, O2: light gray).

shifts for the SiON3 tetrahedra of the Q3 type in MSi2O2N2 (M ¼ Ca, Sr) [17,24]. Furthermore, this result corroborates the presence of only one Si position in the structure of BaSi2O2N2. 3.1.5. Lattice energy calculations In order to confirm the rather obvious crystallographic differentiation of O and N, which could be determined neither with X-ray nor with neutron data, and to evaluate the different structure models based on their lattice energies, lattice energy calculations (MAPLE, Madelung part of lattice energy) [25,26] were performed based on the structure models derived from X-ray and neutron data. No significant differences were found for the possible polytypes. The Cmc21 model yielded a slightly better result, but its significance cannot be claimed as the exact MAPLE values strongly depend on the exact O and N positions, which could not be determined with high precision. In Table 5, the calculations based on the X-ray data are summarized as an example for the small differences between both the structure models.

During the structure determination it was assumed that the oxygen atoms are terminally bound to the Si atoms (O[1]) and the nitrogen atoms are three-connected (N[3]). These assumptions, which were made according to the Pauling rules and based on the similarity to MSi2O2N2 (M ¼ Ca, Sr, Eu) [17–19], are confirmed by the MAPLE calculations. By exchanging the O and N positions, the partial MAPLE values differ considerably from the typical MAPLE values of these atoms and also the differences between the calculated total MAPLE value and the reference value (0.5 Ba2SiO4 þ 0.5 Si3N4) increase.

3.2. Structure description 3.2.1. Ideal structure As the structure determination turned out a tendency toward the Pbcn polytype, this stacking variant will be described first. The layered structure contains only one Ba and Si site, respectively.

Fig. 6. Crystal structure of sinoite Si2N2O (N3: black, O2: light gray). Left: view along [001], cutting the structure in layers is adumbrated; right: view along [100] (tetrahedra with vertices up: dark gray; tetrahedra with vertices down: light gray).

542

J.A. Kechele et al. / Solid State Sciences 11 (2009) 537–543

Table 6 Selected interatomic distances (in Å) and angles (in  ) in BaSi2O2N2 derived from the refinement of X-ray data

Si–O Si–N Si–N Si–N Ba–O Ba–O Ba–O Ba–O Ba–N

Cmcm

Pbcn

1.658(6) 1.721(1) 1.724(1) 1.729(1) 2.781(2)

1.655(5) 1.720(6) 1.726(6) 1.730(7) 2.73(1) (2) 2.82(1) (2) 2.906(7) (2) 3.088(9) (2) 3.343(6) (2)

(2) (2) (2) (4)

2.930(8) (2) 3.078(10) (2) 3.298(6) (4)

O–Si–N O–Si–N O–Si–N N–Si–N N–Si–N N–Si–N Si–N–Si Si–N–Si Si–N–Si

Cmcm

Pbcn

113.0(4) 114.5(5) 98.7(3) 109.6(2) 113.3(2) 107.2(2) 119.6(2) 114.7(3) 119.1(3)

108.3(4) 115.9(5) 107.7(6) 107.1(2) 112.6(2) 105.3(2) 121.5(1) 116.3(2) 119.5(3)

Standard deviations in parentheses.

The highly condensed silicate layers of BaSi2O2N2 consist exclusively of vertex-sharing SiON3 tetrahedra of Q3 type. In accordance with Pauling’s rules the nitrogen atoms connect three silicon tetrahedral centers (N[3]), while the oxygen atoms are terminally bound (O[1]). The up–down sequence of the tetrahedra in a layer is shown in Fig. 5. This layer configuration resembles those in sinoite Si2N2O [27,28], where the layers are directly interconnected via common oxygen atoms (cf. Fig. 6). In space group Pbcn, consecutive silicate layers are rotated against each other by an angle of 180 and are separated by Ba2þ ions. As a consequence, O and N atoms of the consecutive layers form a dicapped cuboid around Ba2þ with coordination number 8 þ 2 (cf. Fig. 7). The determined interatomic distances and angles are within the typical range, a selection is given in Table 6. As the bond lengths of Si–N and Si–O were restrained during the structure refinements, they will not be discussed in detail. 3.2.2. Comparison with MSi2O2N2 (M ¼ Ca, Sr, Eu) The structure of BaSi2O2N2 is closely related to the layered structures of the corresponding Ca, Sr and Eu compounds. As the EuSi2O2N2 and SrSi2O2N2 are isotypic, only the Ca and Sr compounds are considered in the following discussion. In all MSi2O2N2 compounds (M ¼ Sr, Ca, Ba), the silicate layers are separated by M2þ layers. The configuration of the Q3 type SiON3 tetrahedra in SrSi2O2N2 (up–down sequence) is quite analogous to BaSi2O2N2. However, the silicate layers are shifted against each other in the Sr compound (cf. Fig. 5). In BaSi2O2N2 the terminally bound oxygen atoms are directly facing each other. Thus, a larger coordination sphere for the Ba2þ ions is generated (coordination number 8 þ 2). As already described, the Ba2þ ions are coordinated by eight O atoms of consecutive layers. The resulting cuboid is additionally capped by

two N atoms. The shift of the silicate layers in SrSi2O2N2 leads to a distorted trigonal prism of O atoms around Sr2þ, which is capped by one N atom (cf. Fig. 7). In CaSi2O2N2, the SiON3 tetrahedra are condensed with a different up–down sequence (cf. Fig. 5). Nevertheless, the coordination sphere of the cations is comparable to that in the Sr compound (coordination number 6 þ 1). 3.2.3. Disorder phenomena Many electron diffraction patterns showed diffuse streaks along a* (Fig. 1), indicating stacking disorder in the structure of BaSi2O2N2. Annealing the samples did not alter these phenomena. A Eu2þ-doped sample was also investigated and it showed the same effects to the same extent. The structure determination of the Sr and Eu compounds yielded remarkably disordered structure models as well [18,19]. For BaSi2O2N2, the models in the space groups Cmc21 and Pbcn correspond to different stacking sequences of the silicate layers. The lattice energy calculations (MAPLE [25,26]) prove that the lattice energies of both models are almost equal. This corroborates the tendency to form domains of both types. Assuming a high degree of disorder, the average structure of a disordered crystal corresponds to the structure model in Cmcm (cf. Fig. 2), however, our investigations show that the Pbcn polytype is more dominant. In the real structure of MSi2O2N2 phases additional local disorder may concern the up–down sequence in individual layers. However, this aspect was not analysed so far.

4. Conclusion This contribution demonstrates the possibility of reliable structure determination of a microcrystalline disordered compound in a contaminated sample, applying a combined approach using electron diffraction, X-ray and neutron powder diffraction, solid-state NMR and lattice energy calculations. The level of contamination with Si3N4 was up to 50%. So far, it was not possible to synthesize a pure sample of BaSi2O2N2. The layered structure is similar, but not isotypic to any of the hitherto known structures of MSi2O2N2 (M ¼ Ca, Sr, Eu) and exhibits disorder phenomena. The knowledge of the structure is essential to understand the remarkable luminescence properties of Eu2þ-doped BaSi2O2N2. The Stokes shift of the Eu2þ emission in BaSi2O2N2:Eu2þ is the smallest one in the series MSi2O2N2:Eu2þ (M ¼ Ca, Sr, Ba). In addition, the width of the emission band is quite small [4,7]. These important materials properties are directly related to the fact that BaSi2O2N2 contains only one Ba2þ site with a rather highly symmetric environment.

Fig. 7. Coordination sphere of the cations in MSi2O2N2 (M ¼ Ca, Sr, Ba) (left: Ca2þ, middle: Sr2þ, right: Ba2þ; N3: black, O2: light gray).

J.A. Kechele et al. / Solid State Sciences 11 (2009) 537–543

Acknowledgements The authors would like to thank Prof. Dr. Dr. h. c. mult. A. Simon for enabling the TEM measurements, V. Duppel for the practical TEM work and T. Rosenthal for the evaluation of electron diffraction data. EDX and NMR measurements by C. Minke are gratefully acknowledged. Furthermore, we would like to thank Dr. W. Kockelmann for his help during the neutron measurements at the GEM diffractometer at ISIS/Rutherford Appleton Laboratory. This work has been financially supported by the Fonds der Chemischen Industrie and Philips Lumileds Lighting Company.

References [1] R. Mueller-Mach, G. Mueller, M.R. Krames, H.A. Ho¨ppe, F. Stadler, W. Schnick, T. Juestel, P. Schmidt, Phys. Status Solidi A 202 (2005) 1727. [2] C.-C. Yang, C.-M. Lin, Y.-J. Chen, Y.-T. Wu, S.-R. Chuang, R.-S. Liu, S.-F. Hu, Appl. Phys. Lett. 90 (2007) 123503/1. [3] R.-S. Liu, Y.-H. Liu, N.C. Bagkar, S.-F. Hu, Appl. Phys. Lett. 91 (2007) 061119/1. [4] Y.Q. Li, A.C.A. Delsing, G. de With, H.T. Hintzen, Chem. Mater. 17 (2005) 3242. [5] Y.Q. Li, G. de With, H.T. Hintzen, J. Mater. Chem. 15 (2005) 4492. [6] B.-G. Yun, Y. Miyamoto, H. Yamamoto, J. Electrochem. Soc. 154 (2007) J320. [7] V. Bachmann, C. Ronda, O. Oeckler, W. Schnick, A. Meijerink, submitted for publication. [8] D. Becker, T. Fiedler, W. Hempel, F. Jermann, PCT Int. Appl., 2007, WO 2007096333 A1. [9] B.-G. Yun, K. Machida, H. Yamamoto, J. Ceram. Soc. Jpn. 115 (2007) 619.

543

[10] V. Bachmann, T. Juestel, A. Meijerink, C. Ronda, P.J. Schmidt, J. Lumin. 121 (2006) 441. [11] A.C.A. Delsing, H.T. Hintzen, Y.-Q. Li, PCT Int. Appl., 2004, WO 2004030109 A1. [12] T. Fiedler, F. Jermann, Ger. Offen, 2006, DE 102004051395 A1. [13] H. Brunner, T. Fiedler, F. Jermann, J. Strauss, M. Zachau, PCT Int. Appl., 2005, WO 2005031797 A2. [14] T. Fiedler, F. Jermann, PCT Int. Appl., 2005, WO 2005030905 A1. [15] H. Brunner, T. Fiedler, F. Jermann, M. Zachau, B. Braune, PCT Int. Appl., 2005, WO 2005030904 A1. [16] H. Brunner, T. Fiedler, F. Jermann, M. Zachau, PCT Int. Appl., 2005, WO 2005030903 A1. [17] H.A. Ho¨ppe, F. Stadler, O. Oeckler, W. Schnick, Angew. Chem. 116 (2004) 5656; Angew. Chem., Int. Ed. 43 (2004) 5540. [18] F. Stadler, O. Oeckler, H.A. Ho¨ppe, M.H. Mo¨ller, R. Po¨ttgen, B.D. Mosel, P. Schmidt, V. Duppel, A. Simon, W. Schnick, Chem. Eur. J. 12 (2006) 6984. [19] O. Oeckler, F. Stadler, T. Rosenthal, W. Schnick, Solid State Sci. 9 (2007) 205. [20] W. Schnick, H. Huppertz, R. Lauterbach, J. Mater. Chem. 9 (1999) 289. [21] P. Stadelmann, Ultramicroscopy 21 (1987) 131. [22] A.C. Larson, R.B. Von Dreele, GSAS – General Structure Analysis System, Los Alamos, USA, 1998. [23] A. Altomare, M.C. Burla, G. Cascarano, C. Giacovazzo, A. Guagliardi, A.G.G. Moliterni, G. Polidori, J. Appl. Crystallogr. 28 (1995) 842. [24] H.A. Ho¨ppe, Optische, magnetische und strukturelle Eigenschaften von Nitridosilicaten, Oxonitridosilicaten und Carbidonitridosilicaten, dissertation, Ludwig-Maximilians-Universita¨t Mu¨nchen, 2003. [25] R. Hoppe, Angew. Chem. 78 (1966) 52; Angew. Chem., Int. Ed. Engl. 5 (1966) 95. [26] R. Hoppe, Angew. Chem. 82 (1970) 7; Angew. Chem., Int. Ed. Engl. 9 (1970) 25. [27] J. Sjoeberg, G. Helgesson, I. Idrestedt, Acta Crystallogr., Sect. C 47 (1991) 2438. [28] A. Bischoff, T. Grund, T. Jording, B. Heying, R.-D. Hoffmann, U.C. Rodewald, R. Po¨ttgen, Z. Naturforsch. B 60 (2005) 1231.