A new borazine-type single source precursor for Si/B/N/C ceramics

Abstract

The new borazine derivative [B{CH(CH₃)(SiCl₃)}NH]₃ (TSEB) was prepared by reacting Cl₃Si–CH(CH₃)–BCl₂ (TSDE) with hexamethyldisilazane (hmds) at ambient conditions. TSEB was characterized by infrared spectroscopy (IR), nuclear magnetic resonance (NMR), and by mass spectrometry (MS). The specially designed molecule serves as a single source precursor for the synthesis of a highly durable Si/B/N/C ceramic material via the well known ‘polymer route’. Pursuing this philosophy, polymerization of TSEB with methylamine as the cross-linking reagent has lead to a highly homogeneous pre-ceramic polymer, which was further characterized by spectroscopic methods. Subsequent thermal degradation of the as-obtained polyborocarbosilazane was monitored by means of DTA/TG/MS up to 1400 °C. The resulting amorphous silicon boron carbonitride of the approximate elemental composition Si₃B₃N₅C₄ exhibits an outstanding thermal durability at 2000 °C under inert conditions and is also stable in pure oxygen up to at least 1300 °C. The ceramic material still contains borazine rings from the TSEB precursor molecule embedded in the covalent Si/B/N/C network, acting as rigid structural units which reinforce the ceramic material on an atomic scale.

---

1. Introduction

Highly durable amorphous ceramics based on the quaternary system Si/B/N/C are extremely promising candidates for applications under extreme conditions, such as in heat engines (e.g. turbines).^1,2 One of the most attractive features of this still young class of materials is their superior high-temperature durability, even under oxidizing conditions. Due to their low density, mechanical strength and chemical resistance at elevated temperatures, these inorganic random networks are indubitably well suited for many applications in aviation and aerospace.³ In contrast to classical ceramics, such as binary carbides or nitrides, often only having a single forte, Si/B/N/C ceramics ideally combine the above mentioned properties in one material.

In order to achieve a homogeneous Si/B/N/C material the only feasible synthetic approach is the polymerization of special metal organic monomers and subsequent pyrolysis of the resulting pre-ceramic polymer (the polymer route^4–6). Suitable precursors for the preparation of such multinary non-oxide compounds are the so-called single source precursors,⁷ which already contain the constituent elements in one and the same molecule. Since every element is fixed to its neighbouring atom by strong covalent bonds, the precursor generally survives the whole sequence of polymerization and pyrolysis. As a particular advantage of single source precursors, spatial inhomogeneities during polymer preparation are avoided (in contrast to co-polymerization procedures), and a highly homogeneous distribution of the elements in the final random network is achieved.⁸

In a number of examples it has been shown that, with increasing carbon content, the thermal durability of Si/B/N random networks rises significantly.^9–12 In addition, the hardness, stiffness and thermal durability of Si/B/N/C ceramics rises when rigid structural elements, e.g. borazine rings, are incorporated into the random network.¹³ This can be explained by the highly inflexible borazine rings, which strengthen the covalent Si/B/N/C network as a rigid backbone. Consequently, the whole network gets stiffer and stronger so that reorientation of atoms (e.g. on upcoming decomposition) is significantly suppressed. In order to develop new Si/B/N/C ceramics with improved material properties we follow our strategy of implementing both a high carbon content together with borazine rings as reinforcing structural units.

The preparation of a silicon boron carbonitride with the above mentioned features requires a precursor molecule containing the elements Si, B, N, and C, as well as a cyclic unit and a functionalized periphery allowing polymerization. Thus, a suitable single source precursor would have a borazine ring as the cyclic unit and would exhibit a carbon linkage between boron and silicon ensuring an efficient incorporation of carbon into the random network. The periphery of the molecule is functionalized with chlorine atoms in order to enable fast and exhaustive crosslinking with methylamine via dehydrohalogenation reactions.

In this contribution, we present the synthesis and characterization of [B{CH(CH₃)(SiCl₃)}NH]₃ (TSEB, Fig. 1), which fulfils the above-mentioned conditions for yielding a high performance Si/B/N/C ceramic material. Furthermore, we describe the preparation and investigation of the corresponding pre-ceramic polymer derived from TSEB followed by a short study of the thermal polymer-to-ceramic conversion into a new amorphous material.

Fig. 1 The borazine derived single source precursor TSEB.

2. Experimental

2.1. General procedures

All reactions were carried out under an inert atmosphere (argon) in rigorously dried reaction apparatus and solvents. Hexane was distilled from CaH₂ (Merck). Commercially available methylamine (Messer Griesheim) was directly used without further purification.

2.2. Characterization techniques

NMR spectra of the dissolved samples (C₆D₆) were recorded using a Bruker Avance DPX-300 SB, operating at 7.05 T. IR spectra were recorded between 400 and 4000 cm⁻¹ on a Bruker IFS 113v FT-IR spectrometer, using KBr pellets. Scanning electron microscopy pictures of the ceramics were obtained from a Philips XL30 TMP scanning electron microscope. Thermal degradation of the pre-ceramic polymer was investigated by means of simultaneous differential thermal analysis (DTA) and thermogravimetric analysis (TGA) in a Netzsch STA409 (heating rate 10 K min⁻¹, argon flow) equipped with a Balzers QMS421 quadrupole mass spectrometer. The oxidation behaviour of the as-synthesized Si/B/N/C ceramic was also checked in a Netzsch STA409 by heating the samples in a flow of pure oxygen up to 1300 °C at a rate of 10 K min⁻¹. Quantitative analyses of nitrogen and oxygen were carried out simultaneously in a Leco TC-436 hot gas extraction analyzer, whereas carbon was measured in a Leco C-200 hot gas extraction analyzer. Silicon and boron were quantified in a Varian Vista-Pro ICP-OES spectrometer after digestion of the ceramic samples using a mixture of HF–HNO₃–HCl. The macroscopic density of the as-obtained ceramic material was measured using an automatic Micromeritics Accu-Pyc 1330 helium pycnometer. Powder X-ray diffraction patterns were recorded on a STOE Stadi-P diffractometer in transmission mode with Debye–Scherrer geometry, using germanium-monochromated Cu Kα₁ radiation. The ceramic powder samples were placed in a glass capillary (ø 0.5 mm) and diffraction patterns were collected between 4 and 76° in 2θ using a position sensitive detector (PSD).

2.3. Synthesis of [B{CH(CH₃)(SiCl₃)}NH]₃ (TSEB)

The borazine compound TSEB was synthesized by reacting 1-(trichlorosilyl)-1-(dichloroboryl)ethane⁹ (TSDE) with hexamethyldisilazane, according to a previously-described procedure.¹⁴ Under stirring at −60 °C, 25.0 g (102 mmol) of TSDE is added dropwise to 16.5 g (102 mmol) of hexamethyldisilazane. The reaction mixture is allowed to reach room temperature while stirring is continued for 4 h. The by-product chlorotrimethylsilane is removed in vacuum at 10 hPa, and the viscous residue is finally distilled (bp 94 °C/10⁻³ hPa) to yield 17.5 g of pure [B{CH(CH₃)(SiCl₃)}NH]₃ (yield 91%).

2.4. Preparation of the pre-ceramic polymer

A solution of 4.5 g (8 mmol) TSEB in 150 ml hexane is added dropwise to a solution of 100 ml methylamine (69.4 g, 2.234 mol) in 150 ml hexane at −78 °C. The stirred mixture is allowed to reach room temperature, whereupon excessive methylamine evaporates and methylamine hydrochloride precipitates. After filtration of the accumulated methyl ammonium chloride, the residue is washed two times with 50 ml hexane. When the solvent is removed from the filtrate, the pure polymer remains as a viscous colorless liquid, which is further annealed for solidification (24 h at 80 °C).

2.5. Pyrolysis and calcination of the pre-ceramic polymer

The polymer is filled into an inerted boron nitride crucible, which is placed in a quartz tube. Under slightly flowing argon, the sample is heated in a horizontal furnace up to 300 °C at a rate of 100 K h⁻¹ followed by a dwell of 3 h to complete crosslinking of the polymer. Subsequently, the temperature is raised to 900 °C at 100 K h⁻¹ and maintained for another 3 h. In a final step, the pyrolyzed sample is calcinated for 3 h at 1500 °C in an alumina crucible under argon flow.

3. Results and discussion

3.1. Characterization of the single source precursor TSEB

The structure-determining unit of TSEB (Fig. 1) is clearly the borazine ring, which is supposed to become a fixed part of the final covalent Si/B/N/C network after polymerization and pyrolysis. Due to chlorine atoms bonded to silicon, crosslinking of the precursor with nitrogen bases, such as methylamine or ammonia, is enabled and can proceed very rapidly and quantitatively via salt elimination processes (dehydrohalogenation reactions). Another important feature of TSEB is the carbon linkage between boron and silicon. This permits a very efficient incorporation of carbon into the final ceramic, because carbon is tightly integrated via two strong covalent bonds, thus becoming part of the network backbone. In contrast, terminal methyl groups are split-off during pyrolysis much more easily since they are linked to the backbone by only one single bond.

It should be noticed that TSEB is synthesized from the chiral precursor 1-(trichlorosilyl)-1-(dichloroboryl)ethane⁹ (TSDE) with one asymmetric centre located at the methine group (Scheme 1). Consequently, the TSEB molecule itself contains three asymmetric centres located at each methine group. Since the synthesis of TSEB starts from a racemic mixture of TSDE and proceeds without any control of stereochemistry, four different stereoisomers of TSEB form in equal amounts (RRR, SSS, RRS, and SSR). Within this mixture, two chemically different diastereomers exist, both of which are observed in the well resolved ¹H-NMR spectrum (Table 1). Here, the quartet of the methine group as well as the doublet of the methyl group is split with a small but significant difference of 0.004 and 0.007 ppm in the chemical shift, respectively. In contrast, no shifts in the resonances could be resolved in the ¹³C-, ¹¹B-, and ²⁹Si-NMR spectra for the detected diastereomeric isomers (Table 1).

Scheme 1 Synthesis of TSEB (stars mark asymmetric centers).

Table 1 NMR spectroscopic data of the TSEB molecule

Nucleus	Observed multiplicity	δ/ppm	Assignment
a The two chemical shifts apply to two chemically different diastereomers (for further explanation see Section 3.1.).
^1H	Quartet	0.788/0.792^a	CH
^1H	Doublet	1.045/1.052^a	CH3
^1H	Singlet	4.90	NH
^13C	Singlet	20.6	CH
^13C	Singlet	9.4	CH3
^11B	Singlet	34.5
^29Si	Singlet	13.1

---

The FT-IR spectrum of the TSEB molecule is shown in Fig. 2a. Beside the strong stretching mode of the NH group at 3440 cm⁻¹ and the typical bands for alkyl groups in the range 2800–2980 cm⁻¹, TSEB shows a very intense absorption at 1472 cm⁻¹, which clearly originates from the antisymmetric stretching vibration of the six-membered borazine ring. Likewise, the other functional groups can be identified by their characteristic stretching modes, e.g. the B–C bond at 1133 cm⁻¹, the Si–C bond in the range 740–770 cm⁻¹, and the SiCl₃ group at 574 cm⁻¹. An overview of the most important IR absorptions of TSEB together with the corresponding assignments is given in Table 2.

Fig. 2 FT-IR spectra of (a) TSEB, (b) polymer derived from TSEB, (c) ceramic derived from TSEB.

Table 2 IR absorptions of the TSEB molecule, TSEB polymer, and TSEB ceramic

Vibrational mode	TSEB molecule wave number/cm^−1	TSEB polymer wave number/cm^−1	TSEB ceramic wave number/cm^−1
ν (NH)	3440	3427
ν (CH3)	2966	2945
ν (CH)	2879	2806
ν as (B3N3)	1472	1470	1377
δ as (CH3)	1352	1375/1334
δ s (CH3)	1263	1263/1197
ν (BC)	1133	1100
δ (NH)	996	1021/989
ν (SiN)		903	889
ν as (B3N3)	838	803	792
ν (SiC)	764/745
γ (NH)	622
ν as (SiCl3)	574
ν s ( SiCl3)	466

---

The mass spectrum of the TSEB molecule (Fig. 3.) shows the expected molecular ion at m/z 565. Its further fragmentation is mainly characterized by the loss of some end groups like Cl, SiCl₃, and CH(CH₃)SiCl₃ leading to signals with m/z 530, 430, and 402 (base peak), respectively. In several successive steps the base peak loses HCl, as can be deduced from the signals occurring with m/z differences of 36.

Fig. 3 Mass spectrum of the molecular borazine derivative TSEB.

3.2. Characterization of the pre-ceramic polymer

TSEB was polymerized with methylamine rather than with ammonia, thus taking advantage of two important features. Firstly, due to the methyl group present in methylamine, the carbon content of the resulting silicon boron carbonitride is increased, which causes in turn a significant enhancement in high-temperature performance. Secondly, since the degree of cross-linking is lower when using methyl amine instead of ammonia, the polymer remains soluble in organic solvents (e.g. hexane) and can be simply filtered off from the hydrochloride. This facilitates considerably the separation of the synthesized polyborocarbosilazane from precipitated methylamonium chloride, which forms during the polymerization process due to dehydrohalogenation reactions. The FT-IR spectrum of the as-obtained polymer is shown in Fig. 2b. As one would expect, the absorptions resemble the bands observed in the IR spectrum of the TSEB molecule (Table 2), particularly with regard to the strong bands at 3427, 2800–2950, and 1470 cm⁻¹, which are correlated to NH groups, alkyl groups, and borazine rings, respectively. The significant increase in intensity of the bands at 3427 and 2806–2945 cm⁻¹ compared to the TSEB monomer clearly demonstrate the presence of new terminal Si–N(H)–CH₃ and bridging Si–N(CH₃)–Si groups, which were introduced during polycondensation with methyl amine.

The NMR spectra of the pre-ceramic polymer derived from TSEB (Table 3) document two important facts. Firstly, the ²⁹Si-NMR spectrum confirms that the coordination sphere for silicon has changed from SiCl₃C in the molecule (δ_29Si 13.1 ppm) to SiN₃C (δ_29Si −19.3 ppm) in the polymer.¹⁵ Secondly, from the ¹¹B chemical shift of 36.6 ppm (TSEB molecule: 34.5 ppm) it can be seen that boron is still coordinated by two nitrogen atoms and one carbon atom.^16,17 Accordingly, it can be stated that the borazine ring and the B–C bond introduced via the molecular precursor both survive the polymerization procedure.

Table 3 NMR spectroscopic data of the pre-ceramic polymer derived from TSEB

Nucleus	Observed multiplicity	δ/ppm	Assignment
^1H	Quartet	0.39	CH
^1H	Doublet	1.10	CH3
^1H	Singlet	2.49	N–CH3
^1H	Singlet	4.94–5.00	NH
^13C	Singlet	27.9	N–CH3
^13C	Singlet	10.6	CH
^13C	Singlet	10.4	CH3
^11B	Singlet	36.6	BN2C
^29Si	Singlet	−19.3	SiN3C

---

3.3. Pyrolytic conversion of the pre-ceramic polymer

The pyrolytic conversion of the as-obtained pre-ceramic polymer into an amorphous silicon boron carbonitride was monitored simultaneously by differential thermal analysis, thermogravimetric analysis, and mass spectrometry (DTA/TG/MS). In analogy to the thermal degradation of earlier reported polyborocarbosilazanes,^11,12,18 the pyrolysis proceeds in two well-resolved stages (Fig. 4). In the first step between 200 and 450 °C, the polymer loses 14.4% of its initial mass under evolution of mainly methylamine (m/z 30), which indicates further polycondensation and crosslinking of the polyborocarbosilazane through transamination reactions. The temperature range between 450 and 600 °C marks the second step of the pyrolysis and is mainly characterized by the evolution of methane, most probably originating from the terminal methyl groups N–CH₃ and/or CH–CH₃. During this phase, a mass loss of 19.8% is recorded. Finally, above 1000 °C a slight decrease in mass of 2.7% is observed, which marks the evolution of some nitrogen and hydrogen. Thus, an overall ceramic yield of 63.1% results for the thermal conversion of the TSEB derived pre-ceramic polymer into the corresponding Si/B/N/C ceramic.

Fig. 4 DTA/TG curves for the pyrolytic conversion of the TSEB polymer.

3.4. Characterization of the Si/B/N/C ceramic

Pyrolysis of the pre-ceramic polymer and subsequent calcination at 1500 °C under an argon atmosphere yields black grains of the Si/B/N/C ceramic (Fig. 5) with a macroscopic density of 1.9371(8) g cm⁻³. The irregular surface and morphology of the as-obtained ceramic grains and their sharp edges resemble somewhat pieces of broken glass, which already reflects the non-crystallinity of the silicon boron carbonitride. In addition, the amorphous character of the ceramic was tested by means of X-ray powder diffraction, which clearly confirms a random network structure (Fig. 6).

Fig. 5 Scanning electron micrograph of the TSEB derived ceramic.

Fig. 6 XRD powder pattern of the Si/B/N/C ceramic after pyrolysis and annealing at 1500 °C.

The chemical composition of the as-synthesized Si/B/N/C material (Table 4) shows, that the Si : B ratio of 1 : 1 in the precursor molecule is precisely transferred through the polymer stage into the final ceramic. As requested, carbon was incorporated into the random network to a high degree. This can clearly be assigned to the high carbon content of the TSEB molecular precursor together with its bridging methine group. Again, this demonstrates the powerful concept of the single source precursor approach, which offers a very good control over the composition and homogeneity of a projected compound.

Table 4 Chemical composition of the TSEB derived Si/B/N/C ceramic

Element	Mass (%)	Empirical formulae
Si	34.3
B	13.5	exact: Si1.0B1.0N1.8C1.4O0.1
N	29.7
C	20.4	approx.: Si3B3N5C4
O	1.3

---

The broad but strong IR absorption of the ceramic material at 1377 cm⁻¹ (Fig. 2c) can undoubtedly be assigned to the antisymmetric stretching mode of borazine rings, which are known to occur between 1367 and 1400 cm⁻¹ in the solid state.¹⁹ Thus, the borazine rings introduced via the single source precursor TSEB have become part of the final random network. Other typical structural increments like Si–N, Si–C (both 700–900 cm⁻¹) and B–C (900–1100 cm⁻¹) can also be identified from their corresponding broad bands.

The high-temperature behaviour of the TSEB ceramic was tested with a DTA/TG thermal analysis (Fig. 7). Since for silicon boron carbonitrides the onset of weight loss is commonly observed around 1850 °C, this value is rather inappropriate for a reliable assessment of high-temperature stability.² In order to use a more precise criterion for quantifying and comparing the thermal durability of Si/B/N/C ceramic materials, the weight loss at 2000 °C has been taken into account. On heating the as-prepared ceramic sample in a helium flow at a rate of 10 K min⁻¹ up to 2000 °C, a final weight loss of 7% was observed. This indicates a significant increase in temperature stability as compared to several other pre-described Si/B/N/C ceramics, which were derived from acyclic single source precursors.^7,9,11

Fig. 7 DTA/TG curves of the as-prepared Si/B/N/C ceramic upon heating up to 2000 °C in a helium flow at 10 K min⁻¹.

An X-ray powder diffraction analysis (XRD) of the ceramic after heating up to 2000 °C indicates the formation of small crystallites of Si₃N₄ and SiC (Fig. 8). The observed composition of phases is typical for the general crystallization behaviour of silicon boron carbonitrides.² Again, silicon nitride is still present far beyond its real decomposition temperature of ca. 1600 °C at ambient pressure. This special feature has already been discussed²⁰ and can most probably be ascribed to the encapsulation of Si₃N₄ crystallites, which lead to an increase of the local nitrogen pressure, and thus to an increase of the Si₃N₄ decomposition temperature.

Fig. 8 XRD powder patterns of the TSEB derived Si/B/N/C ceramic after heating up to 2000 °C at 10 K min⁻¹ in a helium flow.

The oxidation behaviour of the new ceramic compound plays a crucial role with respect to possible applications under atmospheric conditions. Therefore, a thermal analysis of the TSEB derived ceramic up to 1300 °C was carried out using a rate of 10 K min⁻¹ in a flow of pure oxygen (Fig. 9). On heating up the sample, no significant thermal effects are observed in the DTA curve, indicating that no severe oxidation or decomposition of the investigated material took place. Only a small increase in weight of 3.2% at 1300 °C occurs, which can be attributed to the formation of a passivating surface double layer preventing the ceramic from further being oxidized.³

Fig. 9 DTA/TG curves of the as-prepared Si/B/N/C ceramic upon heating up to 1300 °C in a flow of pure oxygen at 10 K min⁻¹.

4. Conclusions

The new borazine derivative TSEB was used as a single source precursor for the preparation of the carbon-rich silicon boron carbonitride Si₃B₃N₅C₄via the well-known polymer route. TSEB was polymerized with methylamine to yield a pre-ceramic polymer, which, after pyrolysis up to 1500 °C, provided an amorphous ceramic material. The new compound shows a remarkable high-temperature resistance with a mass loss of only 7% at 2000 °C, as well as stability against oxidation in pure oxygen up to at least 1300 °C.

The molecular precursor TSEB comprises two important structural features, both of which lead to a significant enhancement of the high-temperature durability of the derived material. Firstly, the carbon bridge between silicon and boron provides a very efficient incorporation of carbon into the resulting carbonitride. Since carbon preferably bonds to four neighbours (instead of three neighbours as in the case of nitrogen), the covalent network is further strengthened. Secondly, through TSEB, borazine rings are tightly integrated into the random network serving as rigid structural elements. They reinforce the network and make it stiffer and stronger with regard to conformational changes of the structure, thus retarding the decomposition and/or crystallization of the amorphous Si/B/N/C ceramic.

The influence of the rigid units can easily be seen by comparing the mass loss of the ceramics derived from the cyclic single source precursor TSEB and the similar but acyclic precursor TSDE (Table 5). While the ceramic derived from the acyclic TSDE loses 12% of its mass between 1200 and 2000 °C,⁹ the ceramic derived form the cyclic TSEB exhibits a mass loss of only 7% at the same conditions. A similar phenomenon was recently observed in silicon boron carbonitrides, which were derived from pairs of comparable acyclic and cyclic precursor molecules (Table 5). It was shown there that the ceramics derived from acyclic precursors lose 8–12% more of their mass than it is the case for the appropriate ceramics derived from cyclic borazine precursors. This demonstrates very clearly the impact and the importance of the structure of a single source precursor when material properties of Si/B/N/C ceramics are to be tailored.

Table 5 Comparison of weight losses at 2000 °C of Si/B/N/C ceramics derived from cyclic and acyclic single source precursors

Cyclic single source precursor
	TSEB	TSMB^12	DSMB^12
Chemical composition	Si3B3N5C4	Si3B3N7C5	Si3B3N5C7
Weight loss at 2000 °C	7%	11%	0%
Comparable acyclic single source precursor
	TSDE^9	TSDM^11	DSDM^11
Chemical composition	Si3B3N5C4	Si3B3N7C5	Si3B3N5C7
Weight loss at 2000 °C	12%	19%	12%

---

Of course, the production of a precursor derived material (like the TSEB ceramic) in technically feasible dimensions requires a very cost efficient synthesis of the precursor. In this respect, TSEB is extremely promising, because the acyclic precursor TSDE,⁹ from which TSEB is prepared, is accessible in an elegant one-pot hydroboration reaction in almost quantitative yield. In contrast, the synthesis of the pre-described borazine based single source precursors TSMB and DSMB (c.f. Table 5) is more expensive, since the preparation of the corresponding acyclic precursors required for the synthesis of TSMB and DSMB are more laborious, comprising a Grignard reaction step with a relatively low yield.¹¹

References

1. H. P. Baldus and M. Jansen, Angew. Chem., Int. Ed. Engl., 1997, 36, 328.
2. M. Jansen, B. Jäschke and T. Jäschke, Struct. Bonding, 2002, 101, 137.
3. H. P. Baldus, M. Jansen and D. Sporn, Science, 1999, 285, 699.
4. P. Chantrell and E. P. Popper, in Special Ceramics, ed. E. P. Popper, Academic Press, NY, 1964, p. 87.
5. H. Lange, G. Wötting and G. Winter, Angew. Chem., Int. Ed. Engl., 1991, 30, 1579.
6. D. Seyferth, Adv. Chem. Ser., 1995, 245, 131.
7. H. P. Baldus, O. Wagner and M. Jansen, Mater. Res. Soc. Symp. Proc., 1992, 271, 821.
8. D. Heinemann, W. Assenmacher, W. Mader, M. Kroschel and M. Jansen, J. Mater. Res., 1999, 14, 3746.
9. H. Jüngermann and M. Jansen, Mater. Res. Innovations, 1999, 2, 200.
10. T. Jäschke, PhD thesis, University of Bonn, 2003, Shaker Verlag, Aachen.
11. T. Jäschke and M. Jansen, C. R. Chim., 2004, 7, 471.
12. J. Haberecht, F. Krumeich, H. Grützmacher and R. Nesper, Chem. Mater., 2004, 16, 418.
13. T. Jäschke and M. Jansen, J. Eur. Ceram. Soc., 2005, 25, 211.
14. T. Jäschke and M. Jansen, Z. Anorg. Allg. Chem., 2004, 630, 239.
15. C. Gerardin, M. Henry and F. Taulelle, Mater. Res. Soc. Symp. Proc., 1992, 271, 777.
16. B. E. Mann, in NMR and the Periodic Table, ed. R. K. Harris and B. E. Mann, Academic Press, London, New York, San Francisco, 1978, p. 87.
17. H. Nöth and H. Vahrenkamp, Chem. Ber., 1966, 99, 1049.
18. M. Kroschel and M. Jansen, Z. Anorg. Allg. Chem., 2000, 626, 1634.
19. A. R. Phani, J. Mater. Res., 1999, 14, 829.
20. M. Weinmann, J. Schuhmacher, H. Kummer, S. Prinz, J. Q. Peng, H. J. Seifert, M. Christ, K. Müller, J. Bill and F. Aldinger, Chem. Mater., 2000, 12, 623.

---