Solid-State Chemistry with Nonmetal Nitrides
By Wolfgang Schnick*
Among the nonmetal nitrides, the polymeric binary compounds BN and Si3N4 are of partic ular interest for the development of materials for high-performance applications. The out standing features of both substances are their thermal, mechanical, and chemical stability, coupled with their low density. Because of their extremely low reactivity, boron and silicon nitride are hardly ever used as starting materials for the preparation of ternary nitrides, but are used primarily in the manufacture of crucibles or other vessels or as insulation materials. The chemistry of ternary and higher nonmetal nitrides that contain electropositive elements and are thus analogous with the oxo compounds such as borates, silicates, phosphates, or sulfates
was neglected for many years. Starting from the recent successful preparation of pure P3N5,
a further binary nonmetal nitride which shows similarities with Si3N4 with regard to both its structure and properties, this review deals systematically with the solid-state chemistry of ternary and higher phosphorus(v) nitrides and the relationship between the various types of structure found in this class of substance and the resulting properties and possible applications. From the point of view of preparative solid-state chemistry the syntheses, structures, and
properties of the binary nonmetal nitrides BN, Si3N4, and P3N5 will be compared and con trasted. The chemistry of the phosphorus(v) nitrides leads us to expect that other nonmetals such as boron, silicon, sulfur, and carbon will also participate in a rich nitride chemistry, as initial reports indeed indicate.
1. Introduction tors: extremely strong bonds between the elements and the presence of highly crosslinked covalent structures. Nitrogen, the main component of the atmosphere, is om Besides these two criteria, the extremely high bond energy
nipresent. The lightest element in the fifth main group plays of N2 (941 kJ mol"*) as a possible decomposition product of an important role in chemical compounds, in particular in the nitrogen compounds is also of importance when dis• the oxidation states ν and πι (N03 and NO J, respectively) cussing the thermal stability of nitridic materials. Compared as well as — HI (NH3, —NH2, -NH-, and "N-). The with the corresponding oxides, the thermal dissociation of
oxidation state of nitrogen in the nitrides is also — in; only many nitrides with evolution of N2 (for the oxides 02, bond a few hundred nitrides have so far been characterized, al energy: 499 kJmol"1) occurs at much lower temperatures.
though for example the neighboring element oxygen has Thus, the elimination of N2 from Si3N4 occurs at atmospher•
been shown to form more than ten thousand oxides. In spite ic pressure at about 1900 °C, while Si02 can be heated to of this relatively small number, the nitrides include some over 2000 °C without any noticeable decomposition. Alu• extremely useful compounds: silicon nitride, Si3N4, has be minum nitride decomposes above about 1800°C, while the
come an important nonoxidic material, whose applications extrapolated boiling point of A1203 is about 3000 °C. range from ceramic turbochargers to integrated semiconduc The affinity of most elements for oxygen is larger than that 11,21 tor modules. Because of its unusually high thermal con for nitrogen, thus the bond energies for element-oxygen _1 ll3] ductivity (285 Wm K" ), aluminum nitride is predes bonds are generally higher than those of the corresponding tined for use as a substrate material in semiconductor element-nitrogen bonds (single bond energies: Si-O = 444, manufacture. Boron nitride is used as a high-temperature Si-N = 335; P-O = 407, P-N = 290 kJmol"1[41). Similarly crucible material, as a lubricant (hexagonal (Λ)-ΒΝ), and in the bond enthalpies of the oxides are significantly higher the abrasives sector (cubic (c)-BN). In recent years Λ-ΒΝ has than those of the corresponding nitrides (AH?(S\02) = also become increasingly important in the manufacture of -911, V3[A/fJ>(Si3N4)] = - 248; V2[A//?(B203)] = - 637, composite materials. Δ//?(ΒΝ) = - 254; ι/2[Δ//?(Α1203)] = - 838, Δ//?(ΑΙΝ) = The extreme stability of the substances, which are used as — 318 kJ mol"1151), although a quantitative comparison is high-performance ceramics, is due in part to the strengths of difficult because of their differing compositions. the bonds joining the constituent elements; a second impor The formation of oxides is thus an important side reaction tant factor is the presence of highly crosslinked structures in in the syntheses of nitrides. Thus, the preparation of nitrides the solid state. When the electronegativity difference is only in a pure state requires the complete exclusion of oxygen and small, heteronuclear bonds with a high degree of covalent water. This precondition has certainly previously played an character are formed. The high chemical, thermal, and me important role in hindering a detailed investigation of ni chanical stability of nitridic materials such as silicon nitride trides. or boron nitride results from the interplay of these two fac- Among the theoretically possible binary main group ele ment nitrides with nitrogen in the oxidation state — πι and [*] Prof. Dr. W. Schnick the electropositive elements in the maximum oxidation state Laboratorium für Anorganische Chemie der Universität Postfach 101251, D-W-8580 Bayreuth (FRG) corresponding to their group number, many are either Telefax: Int. code (921)55-2535 nonexistent or have until now not been obtained in a pure
806 (φ VCH Verlagsgesellschaft mbH, W-6940 Weinheim, 1993 0570-0833/93/0606-0806 $ 10.00 + .25/0 Angew. Chem. Int. Ed. Engl. 1993, 52, 806-818 and well-defined form because of their low stability (Fig. 1). gen atoms is decreased by electron-withdrawing groups or
173 Li3N exhibits an unusually high tendency for formation; the by mesomeric effects, a requirement which is hardly realiz reaction between lithium metal and molecular nitrogen able in the speculative compound Sb3N5. In the case of other starts at room temperature and atmospheric pressure with main group elements (e.g. carbon and sulfur) the reasons for out any additional activation. In contrast, there is no reliable the hitherto nonexistence of corresponding binary nitrides evidence for the existence and stability of analogous com (C3N4 and SN2) appear to be more complex or to be due to pounds of the heavier alkali metals. Apparently in the ni preparative problems.
trides M3N (M = Na, K, Rb, Cs) the high formal charge of The binary nonmetal nitrides BN, Si3N4, and P3N5 will be the nitride ion (N3~) and the unfavorable molar ratio of main subjects of the following discussion. These are the only cations to anions (3:1) make it impossible to form a stable nonmetal nitrides so far studied in detail in which the elec ionic structure in which the electrostatic, coordinative, and tropositive elements have the maximum possible oxidation lattice-energetic requirements of a stable solid material are state corresponding to their group number. The syntheses,
fulfilled. For the alkaline earth metals, binary nitrides with structure, and properties of BN and Si3N4 were described
the composition M3N2 are, however, known for all the ele many years ago. However, both compounds had been stud ments. ied because of their application in the area of high-perfor mance materials. Because of preparative difficulties, pure phosphorus(v) nitride has only recently been obtained. Al
though P3N5 is thermodynamically appreciably less stable Li Be Β than BN or Si3N4, it has still been possible to develop a Mg AI Si Ρ multifaceted solid-state chemistry of the phosphorus(v) ni
Ca Ga Ge trides. This success has provided the impetus for a further systematic search for new ternary and higher nonmetal ni Sr In trides. Ba
Fig. 1. Binary nitrides are only known for a fraction of the main group elements in the maximum oxidation state corresponding to their group numbers. 2. The Binary Nonmetal Nitrides
BN, Si3N4, and P3N5
In contrast to the ionic structures of the nitrides of lithium Binary nitrides form the basis for the syntheses of ternary and the alkaline earth metals, the decreasing electronegativ or higher nonmetal nitrides containing electropositive ele ities from the third group onwards lead to the formation of ments. Innumerable methods for the preparation of these compounds with more covalent character (e.g. BN, A1N, compounds are known; however, only a limited number of
Si3N4). As the group number increases, the heavier homo- these procedures which afford pure, well-defined products logues in their highest oxidation state show a clearly decreas can be readily carried out under laboratory conditions. Such ing tendency to form stable binary nitrides: In the third main methods must be applied when the binary nitrides are to be group TIN is "missing", while in the neighboring group tin used in the syntheses of new compounds. and lead form no stable nitrogen compounds with the ex
[61 pected stoichiometry M3N4. This trend continues up to the sixth main group; here no compounds with the composi 2.1. Boron Nitride BN VI tion M N2 (M = S, Se, Te) are known. In some cases the instability of the element-nitrogen bond appears to be re Within the family of known ceramic materials boron ni sponsible for the nonexistence of the corresponding binary tride has the lowest density (ρ — 2.27 gem"3). It is colorless nitrides. Thus stable molecular compounds of antimony(v) in the pure state and sublimes at about 2330 °C under a and nitrogen are only formed when the basicity of the nitro- nitrogen pressure of one atmosphere.181 Its decomposition
Wolfgang Schnick, born in 1957 in Hannover, studied chemistry at the Universität Hannover and received his doctorate there in 1986, having worked with Martin Jansen on alkali metal ozonides. After a postdoctoral year with Albrecht Rabenau at the Μ ax-Planck-Institut für Festkörperfor• schung in Stuttgart he moved to the Universität Bonn. His habilitation in the field of inorganic chemistry was completed at the beginning of 1992. His particular interest within the area of preparative solid-state chemistry lies with the nitrides; his work is mainly concerned with their preparation and characterization, and the determination of the relationship between structures and properties. A further main area of interest involves the preparation of new types of compounds which can potentially be used as ceramic materials, ionic conductors, pigments, and catalysts. In 1989 Wolfgang Schnick was awarded the Benningsen-Foerder Prize of Nordrhein- Westfalen and in 1992 he received α Heisenberg scholarship from the Deutsche Forschungsgemeinschaft. In addition, he obtained α "Dozentenstipendium" from the Fonds der Chemischen Industrie and the Academy Prize of the Göttinger Akademie der Wissenschaften. Since 1993 he has been a Profes• sor of Inorganic Chemistry at the Universität Bayreuth.
Angew. Chem. Int. Ed. Engl. 1993. J2, 806-818 807 vapor pressure (N2) is 1.6 Pa at 1900°C and 573 Pa at sharing SiN4 tetrahedra ((
membered [B3N3] rings (
tained from the reactions of boron trichloride BC13 with
NH3 or N2/H2 at temperatures below 1300 °C, or by mi crowave discharge.110,111 Oxygen-free boron nitride can also
be prepared from the reaction of K[BH4] with NH4C1 at temperatures up to 1050 °C.I10] A number of industrial meth ods for the synthesis of boron nitride are also known, al though they do not always afford pure products. These in clude the reactions of oxygen-containing boron compounds
such as B203 or B(OH)3 with nitriding compounds such as
urea, biuret (NH(CONH2)2), dicyandiamide, or melamine
and the carbothermal reduction of B203 with carbon and
ri0J Fig. 2. Crystal structures of Ä-Si N (a) and //-Si N (b) (stereoscopic represen• nitrogen at 1800 to 1900 °C. The reaction of alkaline earth 3 4 3 4 tation looking along [001]). The SiN4 tetrahedra are shown as closed polyhedra. metal cyanamides (CaCN2, SrCN2, BaCN2) with boric acid
B(OH)3 leads to mixtures of Ä-BN and the corresponding
110 121 borates and cyanides. ' The direct nitridation of elemen• Three methods for the preparation of pure silicon nitride tal boron can also be carried out at temperatures above are of particular importance and are also used industrially: 1200°C; however, this process is of neither industrial nor the direct nitridation of elemental silicon,115,18-211 the car• 110,111 preparative importance. Chemical vapor deposition bothermal reduction of silicon dioxide under a nitrogen or (CVD) of boron nitride has been attempted starting from a ammonia atmosphere,115*22"261 and the ammonolyses of large number of volatile, molecular boron compounds, u 5 27 3 SiCl4 or SiH4. · ~ *i The ammonolysis reactions are which are particularly suitable for the formation of amor• particularly suitable for preparative purposes; they lead ini• 181 phous or crystalline thin films or fine powders. Several tially to amorphous and relatively undefined silicon diimide boron-containing polymers have also been used as precur• Si(NH)2, which is converted at temperatures above about 1131 sors for BN. Under typical laboratory conditions most C 900 °C to amorphous Si3N4 and above about 1300 C to processes afford Λ-ΒΝ, which is either amorphous or has a 115,27 29,321 a-Si3N4. " strongly disordered crystalline structure; these can, however, be converted to a regular crystalline state by suitable thermal aftertreatment.18,101 The conversion of hexagonal (a-BN) to 2.3. Phosphorus(v) Nitride P3N5 cubic (/?-BN) boron nitride at high pressures and tempera
1141 tures is favored when Li3BN2 or Mg3BN3 are used. In contrast to the well-studied nitrides BN and Si3N4, very little reliable information on the synthesis, structure, and
properties of P3N5 was available for a long time. Thus, phos•
2.2. Silicon Nitride Si3N4 phorus^) nitride is only mentioned in passing in Allcock's monograph on phosphorus-nitrogen compounds,1331 while Silicon nitride is at present the most important nitridic it takes up only a few lines in Corbridge's thousand-page material in the area of high-performance applications1151 be book on phosphorus compounds.1341
cause of its great hardness (Vickers hardness = 1400 - In principle it should be possible to prepare P3N5 by meth•
2 1700 MNm" ), its high mechanical strength (up to about ods analogous to those used for BN and Si3N4. However,
3 1300°C), and its low density (ρ = 3.2 gem" ). Pure Si3N4 neither the direct nitridation of elemental phosphorus in decomposes above about 1900°C. Crystalline silicon nitride low-pressure plasma1351 nor the simple ammonolysis of
exists in two polymorphous modifications (a-Si3N4, β- molecular phosphorus compounds such as PC15, P4S10,
[36 421 Si3N4); in both cases the structures consist of topologically (PNC12)3, (PN(NH2)2)3, or SP(NH2)3 " lead to the for• similar three-dimensional networks composed of corner- mation of pure, crystalline, and well-defined phosphorus(v)
808 Angew. Chem. Int. Ed. Engl. 1993, 52, 806-818 nitride. Such attempts lead in fact to amorphous products ranging systematic investigations of the chemistry of molec that in some cases still contain chlorine, sulfur, or hydrogen; ular P-N compounds have led to the discovery of a very these generally have a very large surface area and cannot be large number of well-characterized monomeric, oligomeric, characterized further. The main problem in the synthesis of and polymeric phosphazenes; these can contain diverse sub-
pure crystalline P3N5 is that this nitride is much less thermal stituents R (such as F, CI, Br, NH2, NR2, CF3, N3, NCS,
ly stable than BN or Si3N4; thus, decomposition with evolu NCO). It has proved possible not only to synthesize chain tion of nitrogen occurs above about 850 °C [Eq. (1)]. like and cyclic phosphazenes of widely varying molecular size but also to cross-link such units to give polymeric mate
133,34,461 2 P3N5 — 6 PN + 2 N2 > 3 P2 + 5 N2 (1) rials with exactly tailored properties. In contrast to the phosphazenes, the siloxanes are, however, much less Brown undefined amorphous phosphorus(m) nitride is readily accessible; thus, the systematic investigation of this formed in this reaction; thus, in contrast to the situation for class of compounds has taken place only relatively recent- boron or silicon nitride it is not possible to remove impurities (such as H, CI or S) simply by raising the temperature and The situation is completely opposite in the case of the thereby obtaining monophasic P3N5. The preparation of corresponding Si-O and P-N solid-state compounds. While pure crystalline P3N5 starting from molecular phosphorus a large group of preparatively readily accessible Si-O com compounds is thus a tightrope walk between incipient ther pounds (silicates, silicon dioxide)1471 is known which also mal decomposition and sufficient activation of the P-N includes a number of naturally occurring species (e.g. quartz, bonds (cleavage and reformation) for the construction of an pyroxene, amphibole, kaolinite, pyrophillite, mica, feld ordered crystalline solid via amorphous polyphosphazene spar), there were only a few indications of the possible exis 1431 intermediates. Since thermal activation alone is insuffi tence of analogous P-N compounds.148,491 The analogy be cient, the necessary P-N bond breaking and formation must tween silicon oxides (silicates) and phosphorus(v) nitrides is, be supported by chemical means. however, not so close as the isostericity between molecular
The ammonolysis of PCI 5 or (PNC12)3 using NH4C1 rather siloxanes and phosphazenes discussed previously, in which than NH3 at temperatures below 800 °C indeed leads to the in each case a molar ratio of Si:Ο = P:N = 1:1 is involved. formation of colorless, microcrystalline compact P3N5 that The consideration of typical silicate building blocks and the is hydrogen- and chlorine-free [Eqs. (2), (3)]. It seems likely corresponding phosphorus-nitrogen isosteres shows the for that the HCl present, which in pure form decomposes P3N5 mal charge to be different in each case: Si0j~/PN4", at higher temperatures with the formation of volatile prod 2 Si3Of~/P3N£ ~, Si02/PN2". When these charges are com ucts, leads to the reversible and reconstructive formation of pensated in ternary compounds by cations of electropositive [431 crystalline P3N5. elements (e.g. alkali metals or alkaline earth metals), we can expect that mainly covalent P-N substructures as well as (PNC1 ) + 2 NH C1 _^Lh^_ P N + 8 HCl (2) 2 3 4 3 5 contacts between nitrogen and the cations with clearly ionic character will be formed. Thus, in spite of their differing
780 C 2d 3 PC15 + 5 NH4C1 ° > P3N5 + 20 HCl (3) overall composition the phosphorus(v) nitrides should con tain P-N substructures which have isosteric analogues with According to IR, EXAFS, ED, HRTEM, and 15N and in the silicate family. 31P solid-state MAS NMR spectroscopic studies,1*1 phos- For a long time it proved impossible to prepare phospho- phorus(v) nitride P3N5 has a three-dimensional network, rus(v) nitrides in a pure crystalline form, and no reliable consisting of corner-sharing PN4 tetrahedra (d(P-N) information existed on the structure and properties of partic
[43] 3 41 31 2J = 160 pm). The solid, formulated as [P3 N2 N3 ], con ular compounds. The greatest problem was that pure P3N5 tains two types of nitrogen atom in a molar ratio of 2:3, was not available as a starting material for the preparation which are linked to three and two phosphorus atoms, respec of such compounds. Thus, the development of a route to
431 tively/ Because of a stacking disorder in P3N5 demon well-defined phosphorus(v) nitride (see Section 2.3) was a strated by HRTEM,1431 it has so far not been possible to basic precondition for the systematic study of ternary and carry out a complete single-crystal X-ray structural analysis. higher phosphorus(v) nitrides containing electropositive ele However, recently, completely ordered single crystals of ments.
1431 P3N5 have been obtained.
3.1. Alkali Metal and Alkaline Earth Metal 3. Ternary and Higher Phosphorus(v) Nitrides Phosphorus(v) Nitrides
The combination of the two elements phosphorus and Ternary phosphorus(v) nitrides derived from metals nitrogen is isosteric with a corresponding combination should be accessible from the corresponding binary nitrides. of silicon and oxygen. This fact stimulated the systematic The analogy with oxo chemistry suggests that a reaction
study of the siloxanes (typical building block: between an "acidic" nonmetal nitride (P3N5) and a "basic" 144,451 -R2Si-0—SiR2—) in analogy to the isosteric phos- metal nitride should be successful. For various reasons the phazenes (typical building block: -R2P=N-PR2=). Wide- quasibinary Li3N/P3N5 system appeared to be particularly suitable for the systematic study of the ternary phospho [*] EXAFS = extended X-ray absorption fine structure, ED = electron dif fraction, HRTEM = high-resolution transmission electron microscopy, rus^) nitrides: Among the alkali metals only lithium forms MAS = magic angle spinning. a binary nitride with the composition M3N (M = Li, Na, K,
Angew. Chem. Int. Ed. Engl. 1993, 52, 806-818 809 Rb, Cs). Lithium nitride is readily available from its con• P-N bond length =171 pm), which are isoelectronic with
1501 4 3 stituent elements; in addition its thermodynamic stability orthosilicate [SiOJ " and orthophosphate [P04] " building
is sufficient to permit reactions with phosphorus(v) nitride to blocks. In the cubic unit cell of Li7PN4, the complex anions be carried out in the temperature range 600-850 °C. are arranged in a manner analogous to the ß-tungsten type
+
The quasibinary Li3N/P3N5 system has so far afforded (A15) (Fig. 3), while the Li ions are tetrahedrally coordi•
four lithium phosphorus(v) nitrides, which have been pre• nated by the nitrogen atoms of the PN4 tetrahedra. The
pared in a pure form and characterized both structurally and crystal structure of Li7PN4 can be considered as an anti-flu-
1511 152 533 with respect to their properties: Li7PN4, Li12P3N9, - orite type. Thus, the nitrogen atoms adopt a distorted cubic
i54J i55J + Lil0P4N10, and LiPN2 can in each case be prepared in close packing, in which Li ions and phosphorus atoms solid-state reactions between stoichiometric amounts of the occupy all the tetrahedral holes in an ordered manner.1511
binary nitrides Li3N and P3N5 [Eqs. (4)-(7)]. A crucial factor
7 Li3N + P3N5 — — 3 Li7PN4 (4)
W crucible. N2 aim.
4 Li3N 4- P3N5 — —• Li12P3N9 (5)
W crucible. N2 aim.
720 C 5d 10 Li3N -f- 4 P3N5 3 Li10P4N10 (6)
W crucible. N2 atm.
800 C, 4d Li3N + P3N5 - 3 LiPN, (7)
W crucible, N2 aim.
in these reactions is the choice of the crucible material, since Fig. 3. Unit cell of Li7PN4 (stereoscopic representation). The PN4 tetrahedra + at the temperatures used lithium nitride reacts with all the are shown as closed polyhedra; for the sake of clarity the Li ions are not shown [51 J. standard materials. Pure tungsten metal is particularly suit• able for the crucibles for these reactions, since under the reaction conditions used the interior of the crucible is passi- At the quasibinary Li3N/P3N5 intersection, LiPN2 is vated by a layer of tungsten nitride, which, when all the found as the compound with the lowest lithium content. PN4 experimental parameters (temperature, reaction time, par• tetrahedra are again the characteristic building blocks of the ticle size of the Li N used) are optimized, prevents a further 3 P-N substructure. However, because of the molar ratio attack on the crucible to give Li WN . 6 4 P:N = 1:2 they are not isolated but form a three-dimension•
Besides reactions between the binary nitrides, it is also 3 al infinitely linked network [PN4/2], which is topologically possible to react the lithium phosphorus(v) nitrides them• equivalent to and isoelectronic with ß-cristobalite (Si02) selves with Li N or P N to obtain the corresponding i55] 3 3 5 (d(P-N) =164.5(7) pm, * P-N-P =123.6(8)° ). Compared ternary phases [Eqs. (8)-(l 1)]. with the C9 type (the idealized /?-cristobalite structure),
LiPN2 is clearly distorted; all the PN4 tetrahedra are rotated 2 L, N + LiPN _.™'£i*L_^ Li PN 3 2 7 4 (8) by an angle φ = 34.2° about their axes of inversion. Accord W crucible, N2 atm. ing to O'Keeffe and Hyde,1561 a distortion of this type start
7 C 5d ing from a filled variant of the C9 type (φ = 0°) leads, by a 2 Li3N + 4 LiPN2 °° " • Li10P4N10 (9) W crucible, N atm. 2 continuous transformation, to the chalcopyrite type (φ ~ 45 °), a superstructural variant of zinc blende. Thus, the 6P3N5+10Li7PN4 · $*L mal stability decreases. While LiPN2 is stable up to about 900 °C and is hardly affected by moist air or water, Li7PN4 decomposes above 650 °C and is hydrolyzed by water in a violent reaction. The two other ternary compounds LiI2P3N9 and Li10P4N10 take up an intermediate position in both respects, decomposing above about 780 °C and being somewhat sensitive towards hydrolysis. All the lithium phosphorus(v) nitrides referred to have an ionic structure consisting of Li+ ions and complex P-N an• ions. The common structural features are the PN4 tetrahe• Fig. 4. Crystal structure of LiPN2 (stereoscopic representation). The PN4 tetra dra, which can be linked in different ways through common hedra, which are condensed at all their vertices, are shown as closed polyhedra, 7 + vertices: Li7PN4 contains "isolated" [PN4] ~ ions (mean the Li ions as open circles [55]. 810 Angew. Chem. int. Ed. Engl. 1993, 32, 806-818 i541 atoms in LiPN2 and Li7PN4 appear not to be influenced by P4010 (Fig. 6). As in molecular P4O10, and in agreement the significantly different bonding situations of Ρ and Ν on with the assumption of higher double bond or polar bond the one hand or of Li and Ν (mainly covalent and ionic, character, the bonds to the terminal atoms are clearly shorter respectively), on the other; thus, it is surprising to find such than those to the bridging atoms (P4Ol0: d(P-Oterm) = 141 - 591 10 extreme structural similarities. 151, I54] Within the quasibinary Li3N/P3N5 system, Li7PN4 and 158, d(P-Nhr)=168 pm ). 10 LiPN2 are the two ternary compounds with the highest and There is a close relationship between the [P4N10] ~ cage lowest lithium content, respectively. Their anionic P-N sub and the cagelike double rings in the silicates.1581 Thus both 7 3 10 6 structures [PN4] " and [PN4/2] are isosteric analogues of the [P4N j 0] ~ ions and the [Si6015] ' double rings are com 1 1 4 orthosilicate and silicon dioxide, respectively; these two Si- posed of "dreier" * rings. An [Si40lo] ~ building unit iso 10 O compounds have the lowest and highest degree of conden steric with the [P4Nl0] ~ ion, which represents the smallest sation of Si04 tetrahedra, respectively. In analogy with the possible cage built up of corner-sharing Si04 tetrahedra, has large family of silicate structures, it appeared appropriate to so far not been observed. search for phosphorus(v) nitrides with an intermediate de In the solid state, the complex anions in 2(Li5P2N5) = gree of condensation of corner-sharing PN4 tetrahedra, Li10P4N10 have ideal 7^ symmetry; the ten nitrogen atoms which are perhaps similar to the chain or layer silicates. form an almost undistorted section from cubic closest pack Two further ternary compounds within the Li3N/P3N5 ing. In comparison with the situation in molecular phospho- system, Li4PN3 and Li5P2N5, have both been prepared in a rus(v) oxide, a much more favorable packing of the complex pure state and structurally characterized, in comparison building units is attained in the solid state; molecular P4O10, with Li7PN4 and LiPN2, both of these lithium phospho like urotropine, has a distorted body-centered structure rus^) nitrides have an intermediate lithium content. We thus (with respect to the center of gravity of the molecule),1601 10 expected to find an intermediate degree of condensation of while the packing of the [P4N10] " units is derived from the 1541 the corner-sharing PN4 tetrahedra. In fact Li4PN3, in analo cubic face-centered zinc blende-type structure. In the 6 10 gy to the cyclotrisilicate [Si309] ", contains complex anions solid the neighboring [P4N10] " ions, which themselves 152,531 with three corner-sharing PN4 tetrahedra, and thus have an almost completely regular tetrahedral structure, are the formula for this lithium phosphorus(v) nitride is arranged in a manner such that they face each other with ü 3(Li4PN3) = Li12P3N9. The cyclic anions exist in a chair their triangular surfaces parallel and rotated by 60 (Fig. 7). conformation (Fig. 5). 12_ Fig. 5. Cyclotrisilicatc-like [P3Ng] ions (chair form) in Li,,P3N9 152,53]. 0 Fig. 7. Packing of the [Ρ4Ν10]' "" ions in the solid state (stereoscopic represen tation). Since the molar ratio of phosphorus to nitrogen in Li5P2N5 is 2:5, the analogy with the silicates leads us to The extension of this packing principle to a three-dimension expect either layer-type arrangements (silicate example: al infinite solid leads to the formation of "free" layers, which i571 Ba[Si205]) or double rings (example: [Ni(H2N- extend in all directions in space because of the cubic symme I58] (CH2)2NH2)3]3[Si6015]-26H20 ) consisting of corner- try of the crystal (Fig. 7): The lithium ions occupy these sharing PN4 tetrahedra. In fact this lithium phosphorus(v) layers. Because of the topology of the complex anions 10 nitride contains complex [P4N10] " ions, which are thus the in Li10P4N10, the packing described for the solid does first nitrido analogue of molecular phosphorus(v) oxide not permit the relatively high number of cations + 10 (Li :[P4N10] ~ =10:1) to be coordinated in a uniform manner by the nitrogen atoms. The Li+ ions are coordinated either in a trigonal planar manner, tetrahedrally, or with a distorted octahedral nitrogen environment; the molar ratio of these arrangements is 6:1:2. The remaining ten per cent of the lithium ions are distributed with disorder on a site with [*] The term "dreier" ring was coined by Liebau [47] and is derived from the German word drei, which means three; however, a dreier ring is not a three-membered ring, but a six-membered ring comprising three tetrahedral centers and three electronegative atoms (cf. Figs. 5 and 6). Similar terms exist for rings comprising four, five, and six tetrahedral centers (and the 10 Fig. 6. Structure of the [Ρ4Ν,0] " ion in Lil0P4N10. The complex anion has Td corresponding number of electronegative atoms), namely '"vierer", symmetry in the solid and a regular tetrahedral structure [54]. "fünfer", and "sechser" rings, respectively. Angew. Chem. Int. Ed. Engl. 1993, 32, 806 818 811 + higher multiplicity. Lattice-energy and point-potential cal Tablet. Specific Li ion conductivities and activation energies of Li3N, Li PN ,and LiPN . culations ί6ΙΪ confirm that in the highly symmetric packing of 7 4 2 σ the complex anions in Li10P4N10 only a proportion of the 400 Κ Ref. cations can be accommodated in positions with deep poten [Ω-1 cm-1] [kJmoP1] + tial wells (tetrahedrally and octahedrally coordinated Li 3 Li3N 4.0 χ 10" 24 [50] ions). Only flat potential wells are available for the remain 5 Li7PN, 1.7xl0" 47 [51,63] 7 59 [55, 63] ing cations, these leading in part to intrinsic disorder. LiPN2 6.9X10- In all these lithium phosphorus(v) nitrides the Li-N con tact distances determined d(Li-N) = 192-224 pm) are ap proximately equal to the sum of the ionic radii and increase (Li PN : 209pm; LiPN : 209pm).151'55' The number of as expected with the increasing coordination number of the 7 4 2 charge carriers available for ionic conductivity is, however, cations. When the electronegativity differences Αχ between much higher in Li PN . Thus, Li PN , because of its com lithium and nitrogen (Αχ = 2.0) and between phosphorus 7 4 7 4 position, contains much more lithium; in addition the and nitrogen (Αχ = 1.0) are taken into account, a simple anti-fluorite crystal structure (identical with a defect CsCl Pauling-type estimatef 6 21 leads us to expect P-N substruc type) has a large number of interstitial sites which are avail tures with predominantly covalent bond character (78 % co able for ionic conduction. LiPN , in contrast, has a closely valent). The interactions between lithium and nitrogen 2 packed structure (chalcopyrite type, identical to a zinc should in contrast be predominantly (63%) ionic. A system blende modification). In this case no interstitial sites of com atic study of the lithium-phosphorus(v) nitrides is also of parable geometry are available. A diffusion of the Li+ ions interest because the covalent and polarizable P-N substruc in the solid state is thus considerably hindered in LiPN , tures in these compounds, in combination with their ionic 2 which results in a higher activation energy and a conductiv Li-N contacts, should lead to a considerable mobility of the 1631 ity lower than that of Li7PN4. Li+ ions in the solid state, so that they could form a new class Li P N contains structural features which lead us to of ionic conductors. Impedance-spectroscopic measure 10 4 10 + t63} expect a high mobility of the Li ions in the solid: the sym ments on LiPN2 and Li7PN4 (Fig. 8) confirm this predic 10 + metrical packing of the [P N ] " ions leads to the forma tion. The solid state Li ion conductivities determined are, 4 10 tion of free layers which permeate the solid in all directions. The intrinsic disorder of the cations in this compound, the fact that the Li+ ions most probably occupy broad, shallow <— n°q potential wells, and the coordination modes of a fraction of 400 200 100 30 the Li+ ions which are particularly favorable for ionic con duction (trigonal planar) are all factors which would favor a high Li+ ionic conductivity in the solid.153'541 The lithium phosphorus(v) nitrides discussed above con tain either discrete complex P-N anions (Li7PN4, Li12P3N9, Li10P4N10) or a three-dimensional network of PN4 tetrahe- dra(LiPN2). In the silicate family, the cyclosilicates are less stable than the corresponding chainlike compounds. Hard cations (e.g. Li + , Mg2+) increase this effect, while soft cations (e.g. Ca2 + , K + ) stabilize the rings.1471 The replace ment of the monovalent Li+ ions by bivalent alkaline earth metal ions while the Ρ: Ν ratio is kept at 1:3 leads, in contrast to the observations made for the silicates, to a surprising result: A solid-state reaction between the corresponding 1.5 2.0 3.0 3.5 3 _1 10 χΓ[Κ ] • amounts of the binary nitrides Ca3N2 and P3N5 [Eq. (12)] + affords a ternary alkaline earth phosphorus(v) nitride of the Fig. 8. Temperature dependence of the Li ion conductivities in Li3N, LiPN2, and Li7PN4 [63]. composition Ca2PN3. 80 CC 4d 2 Ca3N2 + P3N5 ° ' > 3 Ca2PN3 (12) W crucible, N2 atm. however, lower than the extremely high conductivities of [501 binary Li3N, which arise on the one hand from an appre While "dreier" rings are present in Li12(PN3)3 = ciable doping with hydrogen according to the formula Li12P3N9, the alkaline earth phosphorus(v) nitride contains Li3_xHxN, and on the other from the unusual crystal struc infinite chains LPN2N2/2] of corner-sharing PN4 tetrahedra 1641 ture of lithium nitride. Li7PN4 has a higher conductivity and (Fig. 9). The "zweier" chain found here has an extreme a lower activation energy than LiPN2 (Table 1). This differ stretching factor^ = 1.0, as found, for example, in the chain 147,651 ence between the two lithium phosphorus(v) nitrides can be silicate CaMn[Si206]. As well as the calcium com understood on the basis of the crystal structures, the coordi pound, a magnesium phosphorus(v) nitride Mg2PN3 is also nation of the Li+ ions, and the number of charge carriers known1661 whose crystal structure is an ordered wurtzite [6 71 available in the solid state: In both Li7PN4 and LiPN2 all variant. More recent studies indicate that this phospho- cations are coordinated tetrahedrally by nitrogen, the ob rus(v) nitride also contains infinite "zweier" chains made up served contact distances Li-N being identical on average of corner-sharing PN4 tetrahedra. 812 Angew. Chem. Int. Ed. Engl. 1993, 32, 806-818 thermal decomposition of HPN2 cannot be halted at the stage of HP4N7. There appears to be a close structural relationship between the two phosphorus(v) nitride imides. Thus, HP4N7 can be regarded as a shear structure variant of HPN2. As shown above, the structure of HP4N7 results from the elimination of one nitrogen atom (as NH3) from four formula units of 1 Fig. 9. Infinite "zweier' chains of corner-sharing PN4 tetrahedra in Ca2PN> HPN2; two of the remaining nitrogen atoms of the P-N I64J. skeleton then saturate the valences at the phosphorus atom. With respect to the 0-cristobalite-like P-N substructure 3 t41 2 [P N4 ]"] in HPN2, a fraction (two sevenths) of the nitro gen atoms in HP4N7 must form three P-N bonds according 3.2. Phosphorus(v) Nitride Imides and P-N Sodalites to the formula ^P^N^N'31'].1733 The ternary phosphorus(v) nitrides so far discussed that The first intermediate in the ammonolysis of phosphorus incorporate electropositive elements (hydrogen, alkali pentachloride could in principle be the corresponding pen- metals, or alkaline earth metals) mainly contain P-N struc taamide P(NH ) . In fact, however, condensation reactions 2 5 tural elements with isosteric analogues in the silicate family. occur when only a fraction of the chlorine atoms have been It thus seemed appropriate to treat Si-O compounds of par replaced by NH groups; these lead to the formation of 2 ticular interest, such as framework silicates and zeolites, as 1681 oligocyclo- and polyphosphazenes. Product formation is structural models for the preparation of new phosphorus(v) influenced both by the reaction temperature and the ratio of nitrides. NH to PC1 . so that either chlorine-rich compounds, such 3 5 The importance of zeolites as catalysts, molecular sieves, as [NPCINHJ^, or completely substituted compounds, such adsorbents, and ion exchangers has increased considerably as [NP(NH ) ] , are obtained. It was postulated that the 2 2 X in recent years. The properties that render them so useful are final product of substitution and condensation reactions in based particularly on the characteristic topology of their the ammonolysis of PC1 was a polymeric compound 5 tetrahedral skeletal structures, which have the general com 3 169-711 l74) JPN2/2(NH2)2;2] ~HPN2. However, it is in fact position T02 (T = Si, Al). By exchanging aluminum or found that the reaction of phosphorus pentachloride and silicon for other elements such as Β, P, Ga, Ge, As, Sb, Ti, ammonia leads to a vast number of different oligomeric and Zr, Hf, Fe, Cr, it proved possible to tailor the catalytic prop polymeric phosphazenes; thus, a homogeneous reaction erties of zeolites in a manner favorable for certain applica- product HPN is not formed. The compound HPN can, 74,751 2 2 tions.* Substitution in the anion substructure of the however, be obtained in a pure crystalline form from the framework, for example by replacing oxygen by other elec heterogeneous ammonolysis of pure phosphorus(v) nitride tronegative elements, has, in contrast, been almost complete 1721 under pressure [Eq. (13)]. A particularly useful procedure ly neglected. It appeared to us that the preparation of nitrido involves the in situ preparation of the ammonia required, zeolites should be particularly interesting in view of the pos starting from the corresponding amounts of ammonium sibility of obtaining desirable material properties (stability) chloride and magnesium nitride [Eq. (14)]. and modifying the chemical properties of the zeolites (basic ity). C P3NS + NH3 -^- :i^ 3 HPN2 (13) The synthesis of a zeolite-like framework structure 3 1763 C 400 C [PN4/2] is possible when, for the in situ preparation of Mg3N, + 6 NH4C1 > 8 NH3 + 3 MgCl2 (14) ammonia in the high-pressure ammonolysis of P3N5, Zn3N2 rather than Mg3N2 is treated with ammonium chloride Like LiPN , phosphorus(v) nitride imide HPN has a net 2 2 [Eq. (17)]. The reaction then proceeds quantitatively to af 3 work structure [PN4/2] consisting of PN4 tetrahedra linked ford Zn5H4[P12N24]Cl2 [Eq. (18)]. Analogously to HPN2, a through all four vertices by corner-sharing. This structure can be derived from the isosteric /?-cristobalite-type (d(P- N) = 160 pm, £ P-N-P = 130o[721); the hydrogen atoms are 400 3C Zn N + 6 NH C1 > 8 NH + 3 ZnCl (17) covalently bonded to one half of the nitrogen atoms of the 3 2 4 3 2 1721 P-N skeleton. 640 C 4 P3N5 + 4 NH3 + ZnCl2 ° > Zn5H4[P12N24]Cl2 + 8 HCl (18) 1731 A second phosphorus(v) nitride imide, HP4N7, can be obtained by reacting the required amounts of P3N5 and am monium chloride in sealed quartz ampoules [Eq. (15)]. phosphorus(v) nitride is formed, with a molar ratio P:N =1:2, while at the same time zinc and chlorine are 2Q Cl4d incorporated into the solid through gaseous ZnCl2, which is 4 P3N5 + NH4C1 - _ 3 HP4N7 + HCl (15) volatile under the experimental conditions. A complete ex Equation (16) shows that the removal of one molecule of change of the hydrogen atoms in the product obtained is possible in a subsequent reaction with additional ZnCl2 in ammonia from four formula units of HPN2 leads mathemat which HCl is liberated [Eq. (19)]. ically to the formation of HP4N7. It is unfortunately not possible to carry out this reaction preparatively, since the 7 0<>C 3d Zn5HJPt2N24]Cl2 + 2ZnCl2 ° > 4HPN2 > HP4N7 (16) Zn7[P12N24]Cl2 +4 HCl (19) Angew. Chem. Int. Ed. Engl. 1993, 32, 806 -818 813 In Zn7[P12N24]Cl2 phosphorus and nitrogen form a so- ing amounts of the metal chloride MC12, hexachlorocyclo- 3 dalite-like three-dimensional network [PN4/2] of PN4 tetra triphosphazene (PNC12)3, and ammonium chloride hedra which are linked through all four vertices by corner- [Eq. (21)]. This reaction is carried out in sealed ampoules, sharing. (Λ(Ρ-Ν)= 163.7 pm, *P-N-P=126°; Fig, 10). 700 c 5 MC12 + 4 (PNCl,)3 +12 NH4C1 .?l M5H4[P12N24]C1, + 44 HCl (21) and the batch size is limited by the amount of HCl formed. An alternative procedure involves the use of a molecular phosphorus component in which the chlorine atoms are completely replaced by amino groups [{PN(NH2)2}3] [Eq. (22)]; in this case the product is the hydrogen-free P-N 1771 sodalite M7[P12N24]C12. 7 MCI2 + 4(PN(NH2)2)3 —M7[P12N24]CI2 + 12NH4C1 (22) A particularly elegant method for the preparation of P-N sodalites modified in various ways is the simple reaction between phosphorus(v) nitride imide HPN2 and the corre Fig. 10. Section of the crystal structure of Zn7[P12N24]CK. The zeolite-like sponding metal halide MX2 [Eq. (23)], which affords com- ß-cage made up of condensed IP4NJ and fP6NJ rings is shown. P: black. N: white, CI: gray, Zn: striped. The size of Zn2+ and CI" corresponds to their respective ionic radii [76]. 70Q C 2( 5 MX, +12 HPN, - U M5H4[P12N24]X2 + 8 HX (23) pounds with a large number of different metal cations and While in HPN2 and LiPN2, as in ß-cristobalite, only three- halide ions (e.g. Μ = Mg, Cr, Μη, Fe, Co, Ni, Cu, Zn, Pb; dimensionally bonded [P6N6] rings are found, the sodalite- X = C1, Br).i73<771 like skeleton also contains [P4N4] rings. The two types of By using the methods described above it has been possible rings together form truncated octahedra (β cages), which are to obtain a wide variety of P-N sodalites. As well as divalent typical building units of sodalites and zeolites. Situated at cations such as Mg2 + , Zn2 + , Mn2 + , Co2 + , Ni2 + , Cu2 + , the center of each ß-cage is a chloride ion, in a tetrahedral 2 + 3 + 3 + 2+ Pb , trivalent cations such as Cr , Fe , and even mono environment of Zn ions. Besides the Zn-Cl contact valent cations such as Cu+ can be incorporated. In all cases (260 pm), the metal cations have contact with three nitrogen atoms of the P-N skeleton (d(Zn-N) =196 pm). According phase widths are observed in which a fraction of the metal ions can be replaced by the corresponding number of hydro to the formula Zn7[P12N24]Cl2, there is a statistical defect occupancy (occupancy factor 7/8) at the Zn position. A frac• gen atoms, which are then covalently bonded to nitrogen 173 771 tion of the Zn2 + ions can be replaced by two protons each, atoms of the P-N skeleton. - which in turn are covalently bonded to nitrogen atoms of the The P-N sodalites exhibit remarkable properties: They P-N skeleton. The P-N sodalite has a phase width of are thermally stable up to about 800 °C (in a nonoxidizing atmosphere) and are inert towards all common solvents as Zn(7-x,H2x[P12N24]Cl2 (0<.Y<2). By starting from a well as hot acids and alkalis. Of particular interest is the fact material with a lower metal content Zn6H2[P12N24]Cl2, it is possible to prepare a chlorine-free phosphorus(v) nitride that, depending on the metal cation present, some P-N so dalites are intensely colored (blue (Co, Ni), brown (Fe), dark Zn(PN2)2 by elimination of HCl [Eq. (20)]; the structure of the product is highly distorted and it is no longer crystal- green (Cr)), which suggests that they may find a use as pig ments. 3c Zn6H2[P,2N24]CI2 —°—- L 6 Ζη(ΡΝ,), + 2 HCl (20) 3.3. Silicon Phosphorus(v) Nitride SiPN3 Iine.[73*771 It is, however, probable that this compound pos sesses a three-dimensional network structure of corner- With respect to the development of new high-performance sharing PN4 tetrahedra containing [P4N4] and [P6N6] phosphorus(v) nitride ceramic materials it appears attractive i73,771 ringSi to look at purely covalent ternary compounds containing The synthetic method described previously is not suitable both phosphorus and a second electropositive element, the for the preparation of modified P-N sodalites containing latter being able to form a stable nitride which is a known other metal cations, (e.g. alkaline earth metals, transition ceramic material (BN, A1N, Si3N4). Earlier attempts to pre metals, lanthanides). On the one hand, it requires that the pare ternary nitrides in the system Si-P-N or B-P-N starting corresponding binary metal nitride M3N2 be both existent from the binary nitrides were unsuccessful because of the low and stable, while on the other, the metal chloride MC12 self-diffusion coefficient of these substances and the fact that formed in the reaction with NH4C1 must have a certain min the binary nitrides do not melt congruently. imum volatility. Silicon phosphorus(v) nitride SiPN3 is, however, available P-N Sodalites M5H4[P12N24]C12 (M = Zn, Co, Ni) can from the molecular precursor D in which the required struc also be obtained remarkably easily by reacting correspond tural element =Si-N=P= is already preformed (Scheme 1). 814 Angew. Chem. Int. Ed. Engl. 1993, 32, 806-818 Compound D can be obtained in a three-step synthesis start in detail and structurally characterized. Magnesium boron [851 ing from bis(trimethylsilyl) azane ("hexamethyldisilazane") nitride Mg3BN3 is also known; like Li3BN2 it has a cata A and proceeding via the intermediates Β and C. Low tem lytic effect on the conversion of hexagonal (Λ-ΒΝ) to cubic perature ammonolysis of D, followed by removal of the am boron nitride (c-BN) under high-temperature/high-pressure monium chloride formed and pyrolysis in a stream of ammo conditions.1141 A more complex cerium boron nitride [781 1861 nia, leads to SiPN3. Ce15B8N25 has also been described. Besides Na3BN2, these ternary compounds are obtained by reacting the corre sponding binary nitrides at temperatures between 800 i;C (a- SiCL Cl2, -40"C (CH3)3Si-NH-Si(CH3)3 » Cl3Si-NH-Si(CH3)j > Li3BN2) and 1480°C (Ce15B8N25). Since binary sodium ni Λ Β tride as a starting material for the preparation of Na3BN2 does not exist, a procedure for the preparation of this com PCL 1.) NH3, - 70 C CLSi-N-Si(CHOi ^ CLSi-N=PCL * SiPN, pound had to be devised in which Na3N is a formal interme I 3 3 2.)NH , 800C 3 3 diate. The reaction of a mixture of elemental sodium and CI D C sodium azide has proved suitable [Eq. (24)]. Under the reac tion conditions the otherwise unstable alkali metal nitride Scheme 1. appears to react instantaneously to give the required product.1841 Silicon phosphorus(v) nitride has a three-dimensional net ~ ντ ^v, Belt apparatus, 4 GPa work structure of corner-sharing alternating PN4 and SiN4 2 Na + NaN, + BN -> Na3BN, + N, (24) tetrahedra. Analogous to the isosteric compounds 1000'C. 15 min Si N 01791 and Si N NH,i80J the crystal structure of SiPN 2 2 2 2 3 The ternary boron nitrides so far characterized contain is derived from a defect wurtzite modification. It contains only "isolated" B-N anions rather than condensed struc two-dimensional infinitely linked layers of condensed six- tural units. Thus, Li3BN2 and Na3BN2 are constructed of membered [Si N ] rings in a boat form; half of the silicon 3 3 3 alkali metal cations and linear, symmetrical units NBN ". atoms in the rings are replaced by phosphorus atoms. The The complex anion has 16 valence electrons and is thus 2 [(Si/P)3N3] layers, which are arranged parallel to each oth _ isoelectronic with C02, NCO", CNO", N20, N3 , CN?", er in the crystal, are linked through bridging nitrogen atoms 4 [891 C4-[87,88] and CBN ". The relatively short B-N bond which saturate the remaining free valences at phosphorus [82 84] length (134 pm ~ ) can be explained on the basis of ei and silicon (Fig. II).1781 SiPN decomposes above about 3 ther a considerable degree of double bond character or a polar bond; this result is in agreement with spectroscopic 1811 studies. The two dimorphic modifications of Li3BN2 dif 3 fer in particular in a reorientation of the linear [BN2] ~ ions; + all the Li ions in /?-Li3BN2 (isostructural with Na3BN2) are almost tetrahedrally coordinated by nitrogen atoms, while in 182,831 a-Li3BN2 they are in part linearly coordinated. 3 Unexpectedly, Mg3BN3 also contains linear [BN2] ~ ions, 2 + 3 3 the structure in fact is [(Mg )3(BN ")(N -)]; the "isolated" N3" ions have no direct contact with boron 1851 6 atoms. A trigonal planar anion [BN3] " (d(B-N) = 3 146 pm), isoelectronic with orthoborate [B03] ~, was found 3 in Ce15B8N25; again "isolated" N " ions are also present, which are octahedrally surrounded by cerium atoms.[861 The Fig. 11. Crystal structure of SiPN (stereoscopic representation). The SiN and s 4 remarkably short Ce-Ce distances ( 363 pm) in PN4 tetrahedra are shown as closed polyhedra [78]. dicate the presence of metal-metal bonds as well as the presence of a mixed-valency compound according to 4+ 3 + 3 [86] [(Ce )6(Ce )9(BNr)8(N -)]- 1000 C to give Si3N4 and gaseous phosphorus (P4), which acts as an oxygen scavenger. After 3 h at 1400 °C the decom position product is pure a-Si3N4, which acts as a nucleus for crystallization. Calcination of commercially available amor 5. Ternary Silicon Nitrides phous Si3N4 by the addition of small amounts of SiPN3 gives pure crystalline Si3N4 with a low oxygen content and a high In spite of many attempts it has so far been possible to content of a-Si3N4, which is preferred for sintering process prepare only a few ternary silicon nitrides containing elec es.1781 tropositive elements in a pure form and to characterize them both structurally and with respect to their properties. The isotypic compounds MSiN2 (M = Be, Mg, Mn, Zn) with the 4. Ternary Boron Nitrides same valence electron concentration (VEC) of 41901 can be considered as ternary substitution variants of aluminum Very few ternary boron nitrides have so far been prepared. nitride (AIN).191 "9?1 Their preparation in pure form is pos [81 831 The dimorphic lithium boron nitride Li3BN2 ~ and the sible by solid-state reactions of the corresponding binary I84] analogous sodium compound Na3BN2 have been studied nitrides [Eq. (25)] or, in the case of the manganese com- Angew. Chem. Int. Ed. Engl. 1993, 52, 806- 818 815 pound, by reacting Si3N4 with elemental manganese in an pounds or to determine their composition unequivocal 1943 1101 1041 ammonia atmosphere [Eq. (26)]. ly. " Studies of the quasibinary Li3N/Si3N4 system have also produced some evidence for the existence of ternary lithium silicon nitrides with ionic conducting proper Mg N, + Si N > 3 MgSiN (25) 1051 3 3 4 2 ties in the solid state.* A number of different phases was N7 [106 1071 also identified in the Be3N2/Si3N4 system. - C 3Mn + Si3N4 + 2 NH3 -°— - 3 MnSiN2 + 3 H2 (26) A number of germanium compounds isotypic with the corresponding silicon nitrides mentioned above are also known ι49·93-94-97- ιοι. ιο3. ιοβ] In the solid state these compounds contain three-dimen• sional infinite network structures with SiN4 tetrahedra linked through all four vertices by corner-sharing, which forms condensed [Si6N6] twelve-membered rings ( c 1961 174-180pm, * Si-N-Si = 122 in MgSiN2 ). Together with the metal cations these lead to a wurtzite-like structure. In agreement with the quantification of the Van-Arkel- 1981 1109,1101 LiSi2N3 and the structurally very similar silicon nitride Ketelaar triangle as described by Allen, the non- 1801 imide Si2N2NH have defect wurtzite structures. The lithi• metals carbon, sulfur, or selenium (which are more elec um compound can be obtained by reacting stoichiometric tronegative than boron, silicon, or phosphorus) form binary amounts of the binary nitrides (100 h, 1000 °C). Crystalline nitrides which exhibit a clear preference for molecular struc silicon nitride imide1801 is obtained from silicon and ammo• tures. These binary nitrogen compounds in low oxidation nia under ammonothermal conditions by using potassium states are often discrete molecules with homonuclear bonds [111] amide as a mineralizer [Eq. (27)]. (e.g. (CN)2, S4N4, S4N2, S5N6 ), though some polymeric compounds are also known (e.g. (CN)X and (SN)X). Carbon and sulfur in the maximum oxidation state corresponding to 2 + NH Si N NH + 4 H, (27) Si 3 3 2 2 their group number (iv and vi, respectively) have so far not 6kbarNH, provided any indication of the existence of the binary ni trides C3N4 and SN2, respectively. Although ab initio calcu Analogous to Si N 0 and SiPN (see Section 3.3), 2 2 3 lations indicate that a hypothetical binary carbon nitride LiSi N and Si N NH consist of two-dimensional infinite, 2 3 2 2 with a /?-Si N -like structure should have an unusual me parallel layers of condensed [Si N ] twelve-mem bered rings 3 4 6 6 chanical stability,1112,1131 no compound of the composition in a boat form, which are linked together by bridging nitro• C N has until now been prepared in a pure defined gen atoms ( + The cyanamides can, however, formally be considered as atoms, while in LiSi2N3 the Li ions occupy free tetrahedral ternary carbon(iv) nitrides (M CN , Μ = Li, Na, K, Ag, Tl; sites in the defect wurtzite structure. Silicon nitride imide 2 2 H 11 M CN2, Μ = Ca, Sr, Ba, Zn, Pb) which contain the linear (like silicon diimide Si(NH2)2, which has so far only been anion CN2" with 16 valence electrons.1117"1201 Lithium obtained in an undefined amorphous form) is an intermedi• cyanamide can be prepared from lithium carbide and lithium ate in the industrial preparation of Si3N4 from the am• nitride in molten lithium [Eq. (28)], though the separation of monolysis of silicon tetrachloride. It decomposes above the product from lithium metal is preparatively diffi about 1050 °C with elimination of ammonia to give 1117,1181 1801 cult. Pure Li2CN2 can be obtained easily on a Si3N4. preparative scale from the reaction between lithium nitride Lanthanum silicon nitride LaSi3N5 is obtained by reacting [1211 C 1991 and melamine in a molar ratio of 1:2 [Eq. (29)]. the binary nitrides under pressure (1830 C, 270 bar N2); the reaction of Si3N4 with La203 (2000 "C, 50 bar N2) also 11001 affords the compound as single crystals. As for all C 0 Li2C2 + 4 Li3N - — Κ 2 Li2CN2 +10 Li (28) ternary silicon nitrides which have so far been characterized, Li melt LaSi3N5 has a three-dimensional network structure with (NCNH2)3 + 2 Li3N — 3 Li2CN2 + 2 NH3 (29) SiN4 tetrahedra linked through all four vertices by corner- sharing. The solid contains "dreier", "vierer", "fünfer", and "sechser" (six-, eight-, ten-, and twelve-mem bered) rings Alkali metal hydrogen cyanamides (NaHCN2, with alternating silicon and nitrogen atoms. According to 1122 1231 Na4H2(CN2)3 ' ) and crystalline cyanamide iI24,125i the molar ratio Si: Ν = 3:5, two-fifths of the nitrogen atoms (H2CN2 ) have been prepared and structurally char are bonded to three silicon atoms (i/(Si-N) =173-181 pm), acterized. The reaction of dicyandiamide with cesium car while the remainder (three-fifths) have only two silicon bonate leads to cesium dicyanamide Cs[(CN)2N], which in atoms as directly bonded neighbors (d(Si-N) = 162- the solid state contains the bent pentaatomic anion 173 pm). Lanthanum is coordinated by a total of nine nitro N=C-N-C=N".11261 gen atoms of the Si-N substructure ( Besides the compounds mentioned above, the existence of here is the sulfur(iv) compound K2SN2, which in the solid 2 [127] further ternary silicon nitrides such as Li2SiN2, Li5SiN3, state contains the bent anion SN " isosteric with S02. and Li8SiN4 has been postulated, although it has so far not Exact structural data are, however, not yet available for this been possible to obtain exact structural data for these com compound. 816 Angew. Chem. Int. Ed. EngS. 1993, 32. 806 818 7. Outlook and Future Prospects [14] a) C. Hohlfeld, J. Mater. Sei. Lett. 1989, 8, 1082; b) M. Kagamida, H. Kanda, M. Akaishi, A. Nukui, T. Osawa, S. Yamaoka. /. Crystal Growth 1989,94,261; c) S. Nakano, H. Ikawa, O. Fukunaga, J. Am. Ceram. Soc. Numerous new publications, not only in the area of non- 1992, 75, 240. metal nitrides, make it clear that nitride chemistry is still only [15] H. Lange, G. Wötting,G. Winter, Angew. Chem. 1991,103,1606; Angew. Chem. Int. Ed. Engl. 1991, 30, 1579. r 1281 in its infancy and that many fascinating results can be [16] M. Billy, J.-C. Labbe, A. Selvaray, Mater. Res. Bull. 1983. 18, 921. expected. As shown by the results of systematic studies on [17] R. Grün, Acta Crystallogr. Sect. Β 1979, 35, 800. [18] G. Schwier, G. Nietfeld, Werkstoffe und Konstruktion 1988, 2. 149. the phosphorus nitrides, nonmetal nitrides have the poten• [19] G. Schwier, G. Nietfeld, Sprechsaal, 1988, 121, 174. tial for the development of novel solids with interesting [20] G. Boden, S. Klemm, K. Quaritsch, Silikattechnik 1987. 38, 161. properties; these can be either related to the technically, in• [21] G. Boden, R. Irmisch, G. Himpel, T. Reetz, Sprechsaal 1989, 122, 224. [22] Y. Tuohino, R. Laitinen, K. Torkell in Proc. 3rd Int. Conf. Powder Proc. dustrially, and economically important sulfates, phosphates, Sei. (Eds.: G. L. Messing, S. Hirano, H. Hausner). American Ceramic and silicates, or may be completely novel. In contrast to Society, Westerville, USA, 1989, p. 337 ff. [23] K. Komeya, H. Inoue. J. Mater. Sei. 1975, 10, 1243. oxygen, nitrogen affords an extension of the structural [24] H. Inoue, K. Komeya, A. Tsuge, J. Am. Ceram. Soc. 1982, 65, 205. scope, since covalent bonds can be formed not only to two, [25] M. Mori, Η. Inoue, Τ. Chirai in Progress in Nitrogen Ceramics (Eds.: F. L. but also to three or even four neighboring atoms.11291 The Riley), Nijhoff, Den Haag, 1983, p. 149 ff. [26] T. lshii, A. Sano, I. Imai in Silicon Nitride-1 (Eds.: S. Somiya. M. synthesis of nitrido zeolites according to the structural model Mitomo, M. Yoshimura), Elsevier, London, 1990, p. 59ff. provided by the zeolites appears particularly attractive with [27] M. Blix, W. Wirbelauer, Ber. Dtsch. Chem. Ges. 1903, 36, 4220. respect to desirable material properties and the modification [28] O. Glemser, P. Naumann, Z. Anorg. Allg. Chem. 1959, 298, 134. [29] K. S. Mazdiyasni, C. M. Cooke, J. Am. Ceram. Soc. 1973, 56, 628. of known compounds. However, the extended structural [30] S. Prochazka, C. Greskovich, Am. Ceram. Soc. Bull. 1987, 57, 579. possibilities offered by nitrogen lead us to expect that com• [31] W.M. Shen (Union Carbide), EP-B 365295, 1989, Chem. Abstr. 1990, 113, P11030n. pletely new types of structures should also be possible. [32] Y. Kohtoku in Silicon Nitride-1 (Eds.: S. Somiya, M. Mitomo, M. Regardless of the possible applications of such com• Yoshimura), Elsevier, London, 1989, p. 71 ff. pounds as ceramic materials, ionic conductors, catalysts, or [33] H. R. Allcock, Phosphorus-Nilrogen Compounds, Academic Press, New York, 1972, p. 288. pigments, the systematic exploration of new nonmetal ni• [34] D. E. C. Corbridge, Phosphorus - An Outline of its Chemistry. Biochem trides will close a large gap in the chemistry of the main istry, and Technology, Elsevier, Amsterdam, 1990, p. 120. [35] S. Vepfek. Z. Iqbal, J. Brunner, Μ. Schärli, Philosoph. Mag. Β 1981, 43 group elements. s 527. [36] J. Cremer, H. Harnisch (Hoechst AG), DE-A 2608018, 1977; Chem. My grateful thanks are due to my co-workers Ute Berger, Abstr. 1978,55, 9160 b. Verena Schultz-Coulon, Jan Lücke, and Martin Volkmann for [37] J. Cremer, E. Joerchel, H. Harnisch (Hoechst AG), DE-A 2516915,1976; Chem. Abstr. 1977, 86, 57574v. their tenacity, their inspiration, and their enthusiasm during [38] A. Stock, Ber. Dtsch. Chem. Ges. 1906, 39, 1967. our pleasant and fruitful cooperation in the area of the phos• [39] A. Stock, B. Hoffman, Ber. Dtsch. Chem. Ges. 1903, 36. 314. phorus nitrides. I should like to thank the Deutsche [40] Α. M. Ficquelmont, C. R. Hebd. Seances Acad. Sei. 1935, 200, 1045. Forschungsgemeinschaft, the Fonds der Chemischen Industrie, 141] Η. Jacobs, R. Kirchgässner. Z. Anorg. Allg. Chem. 1990, 581. 125. [42] W Schnick, Z. Naturforsch. Β 1989, 44, 942. and the Ministerium fur Wissenschaft und Forschung des [43] W. Schnick, J. Lücke, unpublished results. Landes Nordrhein- Westfalen for their generous financial sup• [44] U. Wannagat, G. Bogedain, A. Schervan, H. C. Marsmann, D. J. Brauer, port. I am particularly indebted to Monika Schmitt for her H. Bürger, F. Dörrenbach, G. Pawelke, C. Krüger, Κ. H. Claus, Z. understanding, her support, and for many stimulating sugges• Naturforsch. Β 1991, 46, 931. [45] U. Wannagat, D. Burgdorf, H. Bürger, G. Pawelke, Z. Naturforsch. Β tions. 1991, 46, 1039. [46] Η. R. Allcock, Chem. Rev. 1972, 72, 315. Received: October 6,1992 [A 907 IE] [47] F. Liebau, Structural Chemistry of Silicates, Springer, Berlin, 1985. German version: Angew. Chem. 1993, 105, 846 [48] Ε. V. Borisov, Ε. E. Nifant'ev, Russ. Chem. Rev. (Engl. Transl.) 1977,46. 842. [49] Ν. E. Brese, M. O'Keeffe, Struct. Bonding (Berlin) 1992, 79, 307. [1] C. Boberski, R. Hamminger, Μ. Peuckert, F. Aldinger, R. Dillinger, J. [50] A. Rabenau, Solid State Ionics 1982, 6, 277. Heinrich, J. Huber, Angew. Chem. Adv. Maier. 1989, 101, 1592; Angew. [51] W. Schnick, J. Lücke, J. Solid State Chem. 1990, 87, 101. Chem. Int. Ed. Engl. Adv. Mater. 1989,28,1560; Adv. Mater. 1989, /, 378. [52] W. Schnick, Phosphorus Sulfur Silicon Relat. Elem., in press. [2] G. Petzow. Ber. Bunsenges. Phys. Chem. 1989, 93, 1173. [53] W. Schnick, U. Berger, unpublished results. [3] G. A. Slack, J. Phys. Chem. Solids 1987, 48, 641. [54] W. Schnick, U. Berger, Angew. Chem. 1991,103, 857; Angew. Chem. Int. [4] Α. F. Hollemann, E. Wiberg, Lehrbuch der Anorganischen Chemie, de Ed. Engl. 1991, 30, 830. Gruyter, Berlin, 1985, p. 139. [55] W. Schnick, J. Lücke, Z. Anorg. Allg. Chem. 1990, 588, 19. [5] G. H. Aylward, T. J. Findlay, Datensammlung Chemie, Verlag Chemie, [56] M. O'Keeffe, B. G. Hyde, Acta Crystallogr. Sect. Β 1976, 32, 2923. Weinheim, 1975. p. 80 fT. [57] Κ. F. Hesse, F. Liebau, Z. Kristallogr. 1980, 153, 33. [6] A few reports of the preparation of tin nitride have been published, but [58] Y. Smolin. Sov. Phys. Crystallogr. (Engl. Transl.) 1970, 15, 23. in all cases the samples obtained could not be clearly characterized: [59] M. Jansen. B. Lüer, Z. Kristallogr. 1986, 177, 149. a) R. S. Lima, P. H. Dionisio, W. H. Schreiner, Solid State Commun. [60] M. Jansen, M. Mobs, Z. Kristallogr. 1982, 159, 283. 1991, 79, 395; b) R. G. Gordon, D. M. Hoffman, U. Riaz, Chem. Mater. [61] R. Nesper, (ΕΤΗ-Zürich), personal communication. 1992, 4, 68; c) L. Maya, Inorg. Chem. 1992, 31, 1958. [62] L. Pauling, The Nature of the Chemical Bond, Cornell University Press, [7] W. Neubert, Η. Pritzkow, Η. P. Latscha, Angew. Chem. 1988. 100, 298; Ithaca New York, 1960, p. 97 ff. Angew. Chem. Int. Ed. Engl. 1988, 27, 287. [63] W. Schnick, J. Lücke, Solid State Ionics 1990, 38, 271. [8] R. T. Paine, C. K. Narula, Chem. Rev. 1990, 90, 73. [64] W. Schnick, V. Schultz-Couion, Angew. Chem. 1993. 105, 308; Angew. [9] V. L. Vinogradov, Α. V. Kostanovskii, High Temp. (Engl. Transl.J 1991, Chem. Int. Ed. Engl. 1993, 32, 280. 29, 901. [65] R. L. Freed, D. R. Peacor, Am. Mineral. 1967, 52, 709. [10] a) A. Meiler, Gmelin Handbuch der Anorganischen Chemie, Boron Com [66] R. Marchand, Y. Laurent, Mater. Res. Bull. 1982, 17, 399. pounds, 1st Suppl. Vol. 2, Springer, Berlin, 1980, p. 1; b) ibid. 2nd Suppl. [67] W. Schnick, V. Schultz-Coulon, unpublished results. Vol. 1, 1983, p. 20; c) ibid. 3rd Suppl. Vol. 3, 1988, p. 1, and references [68] A. Schmidpeter, C. Weingand, F. Hafner-Roll, Z. Naturforsch. Β 1969, therein. 24, 799. [11] Yu. I. Krasnokutskii, S. N. Ganz, V. D. Parkhomenko. J. Appl. Chem. [69] Η. Schiff, Justus Liebigs Ann. Chem. 1857, 101, 299. USSR (Engl. Transl.) 1976, 49, 307. [70] W. Couldridge, J. Chem. Soc. 1888, 53, 398. [12] S. Podsiadlo, J. Orzel, Pol. J. Chem. 1984, 58, 323. [71] C. Gerhardt, Ann. Chim. Phys. 1846, 18. 188. [13] a) M. G. L. Mirabelli, L. G. Sneddon, Inorg. Chem. 1988,27, 3271; b) D. [72] W. Schnick, J. Lücke, Z. Anorg. Allg. Chem. 1992, 610, 121. Seyferth. W. S. Rees, Jr., Chem. Mater. 1991, i, 1106; c) L. Maya, Η. L. [73] J. Lücke, dissertation Universität Bonn, 1993. Richards, J. Am. Ceram. Soc. 1991, 74, 406. [74] J. V. Smith, Chem. Rev. 1988, 88, 149. Angew. Chem. Int. Ed. Engl. 1993, 32, 806-818 817 [75] W. Hölderich, Μ. Hesse, F. Naumann, Angew. Chem. 1988, 100, 232; [100] Z. Inoue, M. Mitomo, N. Ii, J. Mater. Sei. 1980, 15, 2915. Angew. Chem. Int. Ed. Engl. 1988, 27, 226. [101] R. Juza, H. H. Weber, E. Meyer-Simon, Z. Anorg. Allg. Chem. 1953,273, [76] W. Schnick, J. Lücke, Angew. Chem. 1992, 104. 208; Angew. Chem. Int. 48. Ed. Engl. 1992, 5/, 213. [102] J. Lang, J. P. Chariot, Rev. Chim. Miner. 1970, 7, 121. [77] W. Schnick, J. Lücke, M. Volkmann, unpublished results. [103] J. David, J. P. Chariot, J. Lang, Rev. Chim. Miner. 1974, //, 405. [78] Η. P. Baldus, W. Schnick, J. Lücke, U. Wannagat, G. Bogedain, Chem. [104] A. T. Dadd, P. Hubberstey, J. Chem. Soc. Dalton Trans. 1982, 2175. Mater, in press. [105] H. Yamane, S. Kikkawa, M. Koizumi. Solid State Ionics 1987, 25, 183. [79] I. Wrestedt, C. Brosset, Acta Chem. Scand. 1964, 18, 1879. [106] I. C. Huseby, H. L. Lukas, G. Petzow, J. Am. Ceram. Soc. 1975, 58, 377. [80] D. Peters, H. Jacobs, /. Less-Common Met. 1989, 146, 2A\. [107] Τ. M. Shaw, G. Thomas, J. Solid State Chem. 1980, 33, 63. [81 j J. Goubeau, W. Anselment, Z. Anorg. Allg. Chem. 1961, 310, 248. [108] R. Marchand. Y. Laurent, J. Guyader, P. L'Haridon, P. Verdier, J. Eur. [82] H. Yamane, S. Kikkawa, H. Horiuchi, M. Koizumi, J. Solid State Chem. Ceram. Soc. 1991, 8, 197. 1986, 65, 6. [109] L. C. Allen, J. Am. Chem. Soc. 1989, ///, 9003. [831 H. Yamane, S. Kikkawa, M. Koizumi, / Solid State Chem. 1987, 71, 1. [110] L. C. Alien, J. Am. Chem. Soc. 1992, 114, 1510. [84] J. Evers, M. Münsterkötter, G. Oehlinger, K. Polborn, B. Sendlinger. / [Ill] J. D. Woollins, Non-Metal Rings, Cages and Clusters, Wiley, New York, Less-Common Met. 1990, 162, LI 7. 1988, p. 96 IT. [85] H. Hiraguchi, H. Hashizume, O. Fukunaga, A. Takenaka, M. Sakata, J. [112] Α. Y. Liu, M. L. Cohen, Science 1989, 245, 841. Appl. Crystallogr. 1991, 24, 286. [113] A. Y. Liu, M. L. Cohen, Phys. Rev. B. 1990, 41, 10727. [86] J. Gaude, P. L'Haridon, J. Guyader, J. Lang, J. Solid State Chem. 1985, [114] Η. X. Han, B. J. Feldman, Solid State Commun. 1988, 65, 921. 59, 143. [115] L. Maya, L. A. Harris, J. Am. Ceram. Soc. 1990, 73, 1912. [87] H. J. Meyer, Z. Anorg. Allg. Chem. 1991, 593, 185. [116] L. Maya, D. R. Cole, E. W. Hagaman, J. Am. Ceram. Soc. 1991, 74,1686. [88] W. Jeitschko, R. Pöttgen, Inorg. Chem. 1991, 30, 427. [117] M. G. Down, M. J. Haley, P. Hubberstey, R. J. Pulham, A. E. Thunder, [89] H. J. Meyer, Z. Anorg. Allg. Chem. 1991, 594, 113. J. Chem. Soc. Chem. Commun. 1978, 52. [90] E. Parthe, Crystal Chemistry of Tetrahedral Structures, Gordon and [118] M. G. Down, M. J. Haley, P. Hubberstey, R. J. Pulham, A. E. Thunder, Breach, New York, 1964, p. 3ff. / Chem. Soc. Dalton Trans. 1978, 1407. [91] P. Eckerlin, A. Rabenau, Η. Nortmann, Ζ. Anorg. Allg. Chem. 1967,353, [119] Κ. M. Adams, M. J. Cooper, M. J. Sole, Acta Crystallogr. 1964,17,1449. 113. [120] M. J. Cooper, Acta Crystallogr. 1964, 17, 1452. [92] P. Eckerlin, Z. Anorg. Allg. Chem. 1967, 353, 225. [121] U. Berger, W. Schnick, unpublished results. [93] J. David, Y. Laurent, J. Lang, Bull. Soc. Fr. Mineral. Cristallogr. 1970, 93, [122] M. G. Barker, A. Harper, P. Hubberstey, J. Chem. Res. Synop. 1978,432. 153. [123] A. Harper, P. Hubberstey, J. Chem. Synop. 1989, 194. [94] M. Maunaye, R. Marchand, J. Guyader, Y. Laurent, J. Lang, Bull. Soc. [124] Β. H. Torrie, J. Raman Spectrosc. 1992, 23, 465. Fr. Mineral. Cristallogr. 1971, 94, 561. [125] Β. H. Torrie, R. Von Dreele, A. C. Larson, Mol Phys. 1992, 76, 405. [95] M. Wintenberger, R. Marchand, Μ. Maunaye, Solid State Commun. [126] P. Starynowicz. Acta Crystallogr. Sect. C 1991, 47, 2198. 1977. 21, 733. [127] M. Herberhold, W. Ehrenreich, Angew. Chem. 1982, 94, 637; Angew. [96] M. Wintenberger, F. Tcheou, J. David, J. Lang, Z. Natur forsch. Β 1980, Chem. Int. Ed. Engl. 1982, 21, 633; Angew. Chem. Suppl. 1982, 35, 604. 1346. [97] Τ. Endo, Υ. Sato, Η. Takizawa, Μ. Shimada, / Mater. Sei. Lett. 1992, //, [128] Compare for example a) K. Dehnicke, J. Strähle, Angew. Chem. 1992, 424. 104, 978; Angew. Chem. Int. Ed. Engl. 1992, 31, 955; b) F. J. DiSalvo, [98] J. David, Y. Laurent, J. P. Chariot, J. Lang, Bull. Soc. Fr. Mineral. Cristal Science 1990, 247, 649. logr. 1973. 96, 21. [129] See for example the coordination and bonding situation in Si02 and [99] C. E. Holcombe, L. Kovach, Am. Ceram. Soc. Bull. 1981, 60, 546. Si3N4, P2Os and P3NS or B203 and BN. 818 Angew. Chem. Int. Ed. Engl. 1993, 32, 806-818