摘要:

A density-functional based tight-binding approach to 111-V semiconductor clusters Joachim Elsner Michael Haugk Gerd Jungnickel and. Thomas Frauenheim Technzsche Unzversztat lnstitut fur Physzk,. Theoretzsche Physzk Ill D-09107 Chemnztz Germany We determine here the Hamiltonian and overlap matrix elements necessary in a non-orthogonal two-centre tight-binding (TB) scheme for gallium aluminium arsenic gallium arsenide and aluminium arsenide using a local density-functional method. The repulsive energy required by such methods is described by radial short-range repulsive pair contributions. These are determined with respect to self-consistent cohesive energy calculations performed on diatomic molecules and bulk crystalline forms Without inclusion of any parameters we show that the method can be applied successfully to simulations of small homo- and hetero- nuclear clusters giving an accuracy comparable to ub initzo calculations In addition we prove the transferability of the scheme to the crystalline modifications of As Al Ga GaAs and AlAs by determining the equilibrium structural vibrational and electronic band-structure properties There is great technological interest in 111-V compound semi- conductors in particular in the study of epitaxial growth mechanisms and interface formation which has highlighted a need for an understanding of the properties of such materials at a fundamental level.Theoretical models have been developed to characterize such diverse problems as small homo- and hetero-nuclear clusters' and crystalline surfaces.These methods range from empirically constructed potentials derived from fitting parameters to equilibrium structures to fully self- consistent ub inrtzo schemes ' Although the former are very efficient and thus capable of dealing with extended (amorphous or Crystalline) systems they suffer from a transferability prob- lem. They generally only work well within the regime in which they were fitted and thus are not predictive in structural simulations In contrast the problem of transferability is solved in general by using fully self-consistent ab initio schemes on the basis of density-functional (DF) theory '. These methods can charac- terize very accurately the equilibrium structures and properties Although a lot of progress has been made in applying these methods to ever-larger systems they are still far too slow to examine many interesting problems for example unconstrained surface reconstructions epitaxial growth and nucleation phenomena Therefore more approximate schemes combining the advan- tage of the efficiency of the empirical potentials with the transferability and accuracy of the self-consistent field (SCF) methods are highly desirable In this context tight-binding (TB) models have become very popular recently,' lo providing one of the most accurate alternatives in the determination of the total energy and equilibrium geometry of various systems In particular two-centre-oriented schemes considering only two-centre integrals in the Hamiltonian give results that deviate only slightly from those of more sophisticated methods However the usual procedure of fitting the matrix elements necessary to calculate the band-structure energy is to an arbitrary set of input data It is thus rather complicated and does not guarantee a general transferability to all scale systems Nevertheless good results have been achieved by using empiri- cal TB models for the description of GaAs surface reconstruc- tions" and a-GaAsI2 However it is not clear whether the same TB formulations apply to small GaAs clusters too Our method based on density-functional theory using a minimal basis linear combination of atomic orbitals ( LCAO) framew~rk,'~avoids difficulties arising from an empirical parametrisation Instead we calculate the two-centre Hamiltonian and overlap matrix elements from atom-centred valence-electron orbitals and atomic potentials derived from SCF single-atom calculations within the local-density approxi- mation (LDA). This yields exactly the same energy expression as common parametrised TB schemes but with the important difference of having a well defined procedure for determining the desired matrix elements Recently this method has been applied successfully to the description of the equilibrium structure and properties of various scale C Si and BN struc-tures including interactions with hydrogen,14 l6 as well In addressing possible applications to other 111-V compound semiconductors we now apply the same DF-TB method to Ga As Al GaAs and AlAs clusters and verify the transfer- ability to crystalline forms This paper is organized as follows First we outline briefly the theoretical background of the method followed by a description of how the two-centre Hamiltonian and overlap matrix elements and the short-range repulsive potential are determined In the next section we apply the method to small Ga As Al GaAs and AlAs clusters and present results of the ground-state configurations To verify the transferability of the DF-TB scheme to extended crystalline forms we then calculate the equilibrium bulk properties (cohesive energy bulk modulus lattice constant) and compare our derived band structure with the results of more sophisticated methods Finally we summar- ize our results and give suggestions for further research Method Our method based on the work of Seifert Eschrig and Bieger,13l7 applies the formalism of optimized linear combi- nation of atomic orbitals (0-LCAO) as introduced by Eschrig and Bergert for band-structure calculations '' In this approxi- mation the Kohn-Sham orbitals $ of the system containing 1 atoms are expanded in terms of atom-centred localized basis functions q5v $49= 1Cl"Z4JY-4) (1) lv where I= 1 2 etc is the number of atoms and z= 1 2 etc is the number of Kohn-Sham orbitals By restricting to a minimal basis the q5y represent only valence-electron orbitals which are expanded in terms of Slater-type functions and spherical harmonics As many tests have shown,lg five different values of a and n = 0 1 2 3 form a sufficiently accurate basis set for all elements up to the third row in the periodic table J Muter Chem 1996,6(lo) 1649-1656 1649 Using eqn (2) we perform a self-consistent solution of the modified all-electron single-atom Kohn-Sham equations [T+ vpsat(r)]4v(r)=E,psat(bv(r) (3) vpsat(r)vnucleus(r) + JGartreeCn(r)I + vxcLDACn(r)I += tJ (4) and determine the single-atom electron wavefunctions and potentials. The exchange correlation potential Vx is expressed in terms of the local density approximation as parameterized by Perdew and Zunger 2o.The additional term (r/r,)Nappearing in V(r)in eqn (4) was first introduced by Eschrig et a1 l9 in order to improve the band-structure calculations performed within LCAO It forces the wavefunctions to avoid areas far from the nucleus thus resulting in an electron density that is compressed in comparison to the free atom.The radius r may be optimized to yield the best results however we have found that ro z 2r,, where r, is the covalent radius of the element and N = 2 is an overall good choice Using eqn (1) the Kohn-Sham equations of the many-atom structure are solved within a non-self-consistent treatment in the next step fi$,(V) = &,$,W fi= T+ Kff(4 (5) As an approximation we express the effective one-electron potential V,,Ar) of the many-atom structure as a sum of spherical atomic contributions where V is the Kohn-Sham potential of a neutral pseudo- atom due to its compressed electron density but no longer containing the additional term (~/r~)~(For more details of eqn (6) the reader is referred to the paper of Porezag et a! 14) As a result the Kohn-Sham eqn (5)are transformed into a set of algebraic equations 1CI vr(Hflv'l-E,SpvI11 =0 vl I( = 1,2 etc (number of valence orbitals) (7) 1 = 1,2 etc (number of atoms) where HflVl1= (4jh -R,)IfiI4,(r -RID = (4p(r -R')I4,'k -R' 1) (8) Eqn (7) may also be written as a generalized eigenvalue problem HC = &,SC (9) with the hermitian Hamilton matrix H and the positive definite overlap matrix S are defined by eqn (8) The overlap matrix consists only of two-centre elements which can be calculated in a straightforward manner Consistent with eqn (6) one can neglect several contributions to the Hamiltonian matrix elements HpV,l7 yielding ifp=v and I=I v,' + v,' ) If 1 # 1' otherwise (10) This actually reduces the problem to a two-centre approach As has been shown already by several a~thors,~l-~~ the total energy of the system within this approach can be written with the usual tight-binding equation as a sum over the 'band- structure' energy (sum of the occupied Kohn-Sham orbital energies) and a short-range repulsive two-particle potential 1650 J Mater Chem 1996 6(lo) 1649-1656 Etot({Rk >)=EBS({Rk 1) + Erep({ IRk -R,1 1) = 2 niEt({Rk))+ 2 kp(lR1-R/rl) (11) I k <I where n is the occupation number of Kohn-Sham orbital The short-range repulsive contributions V,,,(R) can be deter- mined easily as the difference of the cohesive energy resulting from self-consistent total-energy calculations on molecular and crystalline reference systems and the related band-structure energy E, for different values of interatomic distances R %p(R) = ELDASCF(R) -E13S(R) (12) From eqn (11) we can now derive the interatomic forces FtJ= -2=1,2 N J=l,2,3w acting on the nuclei As one can show these derivatives can be expressed in the form Here the c are the eigenvectors of the single particle states with energies E Eqn (13) simply involves derivatives of the Hamiltonian and overlap matrices and the repulsive potential Hence for MD simulations one has to diagonalise eqn (9) to calculate the forces at each time step Determination of the tight-binding matrix elements As an example we discuss the construction of interatomic Hamiltonian and overlap matrix elements for GaAs In determining the atomic wavefunctions and potentials as described in eqn (3) and (6) we choose ro = 4 7a0 for gallium and ro=45ao for arsenic as the confinement radius of the additional potential term All matrix elements of eqn (10) to be substituted into the general eigenvalue problem [eqn (9)] are evaluated for differ- ent internuclear distances r ranging from 2 0 to 10 Oao using a distance step of 0 02a0 and tabulated as Slater-Koster (SK) integrals.The corresponding curves for GaAs As and Ga are shown in Fig 1-3 t From these SK integrals which are determined only once the eigenvalue problem (9) for any geometry can be constructed by a uniform coordinate trans- formation 25 Hence we can calculate the required integrals (and their derivatives) from their pre-determined atomic orbitals for every geometry rather than fitting a simpler functional form Finally the repulsive potential is obtained by eqn (12) including SCF cohesive energy data of the diatomic molecule and experimental values for the bulk as refeiences [Fig (4)] Clusters Since we have included the SCF dimer data into the fit of our repulsive energy the diatomic properties of the SCF calculation are reproduced Aluminium and gallium clusters A13 Ga,.Configuration interaction (CT) calculations for the aluminium trimer find that the ground state is an isosceles very close to an equilateral triangle with a bond angle x of ?The related SK tables are available upon request for the authors Fig. 1 (u)Hamiltonian and (b)overlap matrix elements vs interatomic separation H, S for gallium arsenide 562' (ref 26) and 605" (ref 27) and bond lengths r=5u Jones' performed SCF LDA calculations and obtained an equilateral triangle with a bond length of 4 65a We also find that an isosceles triangle (a= 62") is the most stable structure but we obtain a slightly enlarged bond length (r = 5 394,).The linear chain (r = 5 16a,) is higher in energy by 0 3 eV Here again Jones' found a shorter bond length of r =4 86~0 For Ga various authors report many low-lying states with very different geometries Balasubramanian and Feng" find an isosceles triangle (a= 61 2" r = 4 88u,) for the ground state which has been confirmed by Meier et a12' but with slightly larger bond lengths In contrast Jones' predicts an equilateral triangle having considerably shorter bond lengths (r = 4 39u,) to be the ground state Further he reports that the most stable linear form (r = 4 56a0) is about 0 9 eV higher in energy In our calculations we determine a variety of geometries of nearly degenerdte energies also.There is a flat isosceles triangle (3 = 129" r = 9 la,) an equilateral triangle (r = 5 31Uo) the linear chain and a second isosceles triangle (a= 50 5" r = 4 42u0) all of them separated by only 0 04 eV Since the very small energy difference is surely within the error margin of the present method we are therefore not able to predict any configuration to be the most stable one A14 Ga,.For Al we have performed SCF LDA calculations indicating a very soft energy surface with respect to geometrical changes As a consequence any rhombus with angles between 72 and 90 IS stable to reasonable accuracy While Jones indeed determines the rhombus (D,,,bond angle a= 56 5") to be the most stable configuration,' we obtain a square for the ground state (D,,,5 14Uo) Pacchioni and K~utecky,~'and Meier et a129 also predict a rhombic ground state but the square Fig. 2 (a)Hamiltonian and (b)overlap matrix elements us interatomic separation H,, S for arsenic (E = -0 5193Eh,E* = -0 1954Eh) (D4h)is very close in energy (0005 and 0 04 eV higher respect-ively) In accord with Jones,' there are also a lot of higher lying local minima in the energy surface The most stable Ga cluster in our calculations is again a square (r = 5 06a,).This is in between the isoenergetically stable square (r =4 632a0) and rhombus (a= 71 6" r = 4 49Uo) found by Jones' and a square with significantly longer bond lengths (r= 5 31a0)obtained by CI 29 A slightly smaller second square is obtained to be metastable by the DF-TB method with a side length of 4 744 and an energy decrease of 0 19 eV In addition a rhombus (a= 50" AE = 1 eV) forms another metastable structure A] Ga,.. There have been several studies of the aluminium pentamer Whereas CI calculation^^^ and semi-empirical LCAO calculation^^^ found that pyramidal structures [Jahn-Teller distorted CZL:(ref 27) and C4c(ref 31)] were the most stable configurations Petterson et al 32 report a planar (C ) form (with bond lengths constrained to be equal) to be 0 2 eV more stable than the pyramid In contrast Jones' calculations' yield two low-lying structures with almost identical energies a substantially deformed pyramid and a planar structure (C,t) This result is also confirmed by the present calculations where we find that a slightly deformed Al pyramid is the most favourable structure but the planar structure [C Plate 1(a)] is higher in energy by only 0 25 eV In the case of Ga a square capped by an atom on one side [see Plate l(b)] is obtained as the most stable geometry in agreement with SCF-LDA results A planar structure [Plate l(a)] and a pyramid are metastable at slightly higher energies of 0 27 and 0 14 eV respectively Fig. 4 Short-range repulsive-pair potentials for Ga (rcUtoff= 5 2a,) As (rCutoff= 5 14ao)and GaAs (rcutoff= 5 1 8ao) Al Ga,.In our calculations for Al the regular octahedron Jones,' we find that the most stable planar structure consisting 0 (r = 5 07no)forms the most stable structure in accordance of two aligned rhombi with bond angle a M 60" In Plate 1(d) with the ah znztzo predictions of Petterson3' for symmetric is significantly higher (AE = 0 9 eV) in energy structures. Jones' and Jug et found that the octahedron. The situation In Ga IS quite similar to that in A16.We again undergoes a Jahn-Teller distortion to a more stable trigonal identify the regular octahedron 0 (r= 5 13ao) as the most antiprism stable structure almost degenerate with a prism structure. The A metastable prism structure [Plate 1 (c)] in the present most stable planar structure formed by two aligned squares is calculations is obtained at AE = 0.26 eV. In agreement with clearly higher in energy (0.54 eV) 1652 J. Muter Chem. 1996 6( lo) 1649-1656 Al Ga,. As in the calculations of Jug et Raghavachari and Jones,' we confirm the Al ground state to be a C3c structure which can be viewed as a 'capped' antiprism shown in Plate l(e). In the energetic order it is followed by a planar structure [see Plate l(f)] at AE = 1.4 eV.For Ga the ground state is the same as for Al and the planar structure lies 1.33 eV above the ground state. Al Gag Al Ga,. Results on larger clusters Al Al and Ga Gag have been reported by SCF-LDA' and semi-empiri- cal LCA031 calculations. In accordance with Jones we find an Al ground state shown in Plate l(g).. This has a shape similar to a face-centred cubic (fcc) lattice primitive cell but is quite distorted bond angles and bond lengths vary from 50 to SO" and 4.91 to 5.05~~~respectively whereas they are 60" and 5.41~1,in the fcc lattice primitive cell. For Ga our calculations yield a very similar structure but with slightly different bond lengths. Finally the lowest lying Al structure shown in Plate l(h) is in good agreement with the structure predicted by Jones.' Again a similar geometry is stable for Ga confirming the result of Jones.' In Fig.5 we have plotted the dependence of the cohesive energies per atom on the cluster size. As can be seen the DF-TB scheme yields energies which are in a good agreement with the SCF-LDA calculations of Jones.'. The energy devi- ations for some of the medium-sized clusters are probably due to the fact that these energy surfaces are very complex thus permitting a variety of local minima. Arsenic clusters As,.. The arsenic trimer has been identified in gas-phase charge-transfer reaction^.,^. Theoretical studies at the SCF-LDA and HF-CI levels have been performed by Jones,2 Igel-Mann et a/.,' and Balasubramanian et uL3 In agreement with Jones,2 we determine an isosceles triangle (C21; CI z 58" r = 4.40~~)and a nearly degenerate D structure (r= 4.49~~) to be the most stable geometries..The linear chain (r = 4.29~~) lies at significantly (2.1 eV) higher energy. As,. Since As4 is the most prominent component of arsenic vapour between 400 and 850 K,37*38the tetramer has received considerable attention. Gas-phase diffraction measurements of .~~Morino et ~ 1find a tetrahedron with an interatomic separa- tion of r = 4.602 & 0.008~~as the most stable configuration. This is supported by local spin density (LSD) calculations which give 4.613 (ref.40) and 4.56~~~~while CI calculations lead to slightly enlarged bond lengths 4.73 (ref. 41) and 4.67~1,~~~respectively.. The present calculations also yield a tetrahedron with r = 4.652 followed by a 'roof' structure Plate 2(u) at AE = 1.5 eV and a square (r= 4.49~~)which however lies 2.0 eV above the ground state.As,. In accordance with Jones,2 we find that the As ground state is a C2" structure as shown in Plate 2(b).. The slightly distorted pentagon a planar CZ0 structure with bond lengths between 4.40 and 4.49~1,~and the regular pentagon (D,,,r= 4.55~~)are almost isoenergetic metastable configurations at 0.8 eV above the ground state. As,.. The present calculations give several local minima in the energy surface of As,.. The lowest lying structure is a D configuration (trigonal prism).. The regular planar hexagon (D6h) lies 2.74 eV higher.Other local minima are described by a buckled structure shown in Plate 2(c) and a distorted octahedron (bond lengths 4.52-5.12~~)found at AE = 2.1 and 4.5 eV respectively. A%. In agreement with Jones,2 our calculations yield a C structure [Plate 2(d)] corresponding to a local minimum in the energy surface. However a capped prism [C2,; Plate 2(e)] is found to be 0.2 eV more stable. As,. For As we report a cubic oh structure (bond length 4.58~~)to be the most stable geometry. Jones2 predicts that although the cubic structure corresponds to a local minimum a structure of the form Plate 2(f) is clearly more stable. In our calculations this geometry is at 0.40eV above the cubic ground state. 3 4 5 6.7 8 9 cluster size Fig.5Energies of small clusters for Ga (a) and A1 (b). 0,from ref. 1 - this work. Plate 2 (u)-(g) Equilibrium structures for arsenic clusters J. Muter. Chem. 1996 6(lo) 1649-1656 1653 As,. As usual there is a variety of local minima on the energy surface of clusters of this size.. The geometry correspond- ing to the lowest energy determined for As is shown in Plate 2(g). It can be derived from the cubic structure of As by adding a bridging atom with two-fold coordination. Some examples of Ga,As and Al,As clusters There are of course a large number of possible isomers for the binary clusters. We see however that chemical bonding constraints reduce significantly the number of favourable geo- metries. Our calculations were performed with starting geo- metries suggested by Andreoni and some highly symmetric structures.Ga2As2 Al,As,. For the smallest reported heteronuclear GaAs cluster we confirm the ab initio result of Andreoni in obtaining a planar rhombus as the most stable isomer. In this cluster the As-As (4.44~~)bond is clearly favoured and the As-Ga-As bond angle is ca. 51" (ca. 52O in ref. 3). Another planar structure [Plate 3(a)],is a local minimum for Ga,As but the lowest lying geometry of A1,As2 lies 0.21 eV above. The energetically lowest three-dimensional cluster for Ga As is a roof structure at 1.4 eV above the ground state. Our calculations confirm that the same structures also correspond to local minima on the Al,As energy surface. However the energetic ordering is changed..The minimum energy configuration has the geometry shown in Plate 3(a). The Al-1 Al-As and As-As bond lengths are 5.04 5.24 Plate 3 (a)-(c) Stable structures for GaAs blue spheres represent gallium atoms the red spheres arsenic 1654 J. Muter. Chem. 1996 6( lo) 1649-1656 and 4.40a0 respectively.. The planar rhombus the most stable structure for Ga,As is now found to be 0.2eV higher in energy. Ga,As Al,As,. In accordance with Andreoni3 we report the hexamer to have two low lying states (AEz 0.35 eV).. The lower one can be viewed as an edge-capped trigonal bipyramid [Plate 3(h)].. The Ga-Ga bond (I' = 5.83~~)is much longer than the As-As and the Ga-As bonds which are 4.5 and 5.1a0 respectively.. The second structure is shown in Plate 3(c).Here the same large differences between the bond lengths occur. In both structures the Ga-As arrangement is not alternate one As atom having two As nearest neighbours which is consistent with indications from experiments on the singly ionized clusters. For the AI,As clusters we find that the same geometries are stable and the energetic ordering of the isomers is identical. Ga,As Al,As,.. The present calculations predict a slightly distorted octahedron for the ground-state geometry.. The bond lengths vary from 4.7 to 5.1U0 the shorter one corresponding to the As-As bond. A regular octahedron with bond length 5.13~~is 1.47 eV higher in energy. Here as in the previous structures we observe that the Ga-Ga and the Ga-As bonds are of the same length.In our most stable planar structure a slightly distorted hexagon however the Ga-Ga and Ga- As bond lengths differ considerably Ga -As being 0.6~~~ shorter. As in the case of other stoichiometric relations the same isomers in the same energetic ordering were obtained for Al,As also. Ga4As Al,As,. As for Ga,As we find a slightly distorted octahedron to be the ground state of Ga,As,.. The As-As bonds are again shorter than the Ga-As bonds by 0.6~'. Again a variety of local minima exist on the energy surface. The most stable linear structure a chain is determined at 8 eV above the ground state. The lowest lying geometry for Al,As is almost identical to that of Ga,As2. Most of the isomers tested for Ga,As were also found to have the same geometry for Al,As,.. There is however a slight change in their energetic ordering.Bulk properties and band-structure calculations In order to verify the transferability of the density-functional tight-binding approach to bulk structures and properties we have calculated the cohesive energy per GaAs and AlAs dimer in the zinc blende and rocksalt structures as a function of the interatomic distance.. The equilibria for the zinc blende struc- ture determined at an interatomic distance I' = 4.6~~for both compounds confirm very well the experimental results.42. The bulk moduli are calculated to be 7.87 x 10" Pa (GaAs) and 7.37 x 10" Pa (AlAs).. They are in good agreement with the experimental values of 7.56 x 10"Pa (GaAs) and 7.80 x 10'' Pa (A~As).~~ For the rocksalt structure we find the equilibrium interatomic distance at 5.12~~ (AIAs) and 5.13~~ (GaAs) with corresponding decreases of the cohesive energy of 0.72 and 0.66 eV per dimer relative to the zinc blende structure.While the cohesive energy differences between the two com- pounds as well as the energetic difference between the zinc blende and the rocksalt structure are described qualitatively well the absolute energies show the usual over-binding which is known to occur in LDA calculations.. The crystalline phases of aluminium gallium and arsenic are described at a compar- able level of accuracy. Furthermore we make use of the solutions of eqn. (2) as atomic basis functions in a usual tight-binding equation to calculate the band structures of Al As GaAs and AIAs..The resulting valence bands are in good agreement with those calculated with more sophisticated methods. However as can be expected from any sp3 tight-binding scheme our method produces poor results concerning the conduction bands As an example we show the derived band structures of AlAs in Fig 6 and As in Fig 7 In addition we have calculated the phonon densities of states (DOS) for GaAs and AlAs. The results are shown in Fig 8. The eigenvalues of both spectra were broadened with Lorentzians of 15 cm-' width A rough analysis reveals the following properties GaAs.. The vibrational density of states of zinc blende GaAs can be split into two main bands.The high wavenumber region between 200 and 300cm-' is clearly occupied by Ga-As stretching vibrations. The dominating peak is centred at 258 cm-' Bending vibrations can be assigned to the DOS at wavenumbers between 75 and 200cm-' Here we obtain two maxima at 116 and 181 cm-' At these low wavenumbers there are apart from bending vibrations also strong translational components of atomic groups occupying the modes Fig. 8 Vibrational density of states for AlAs (a) and GaAs (h) AIAs. For AlAs we obtain qualitatively the same DOS as -03 for GaAs Here the high wavenumber region arising from Al-As stretching vibrations varies from 250 to 350 cm-' with one principal peak lying at 329cm-' and a smaller peak at -04 289 cm-'. The wavenumbers due to bending vibrations range from 100 to 250cm-' with two maxima determined at 121 and 204 cm-' Again at these low frequencies we observe -05 translational modes of atomic groups 4-06 L G X U.K G Summary We have presented here a density-functional based non-ortho- gonal tight-binding scheme for Al Ga As AlAs and GaAs Fig.6 Calculated band structure for AlAs (E,= -0 13Eh) Fig. 7 Calculated band structure for As (E,= -0 07Eh) Though only two-centre integrals are considered the derived interatomic potential is highly transferable without additional changes to the total energy expression We have applied the method to small clusters of gallium aluminium and arsenic and their binary compounds Apart from a few cluster con- figurations we confirm the most stable geometries and the energetic order of metastable clusters as predicted by self- consistent LDA and ab rnztro quantum chemical calculations In order to verify the transferability to bulk systems we have calculated bulk crystalline properties such as the binding energies bulk moduli and phonon densities of states for GaAs and AlAs.These results are also in a good agreement with experimental data We have therefore demonstrated that despite its simplicity above all in its complete neglect of three- centre integrals the method is highly transferable giving reliable results for geometries cohesive energies phonon fre- quencies and band structures for clusters and solids and hence may be used in predictive MD simulations We are actually applying the method to perform a detailed study of the various surface reconstructions of differently oriented GaAs and AlAs In near future this will serve as the basis for molecular growth J Muter Chem 1996 6(10) 1649-1656 1655 simulations and investigations of interface formation in coupling with other semiconducting materials 19 20 H Eschrig Optimized LCAO Method and the Electronic Structure of Extended Systems Akademie-Verlag Berlin 1988 J P Perdew and A Zunger Phys Rev B 1981,23,5048 We would like to thank all of our group for useful discussions and in particular Dirk Porezag who supported our investi- gations on Al and Ga using an SCF full potential LDA code 21 22 23 24 G Seifert and R 0 Jones Z Phys D 1991,20,77 P Blaudeck T Frauenheim D Porezag G Seifert and E Fromm J Phys Condens Matter 1992,4,6389 W M C Foulkes and R Haydock Phys Rev B 1989,39,12521 D Tomanek and M A Schluter Phys Rev B 1987,36 1208.25 J C Slater and G F Koster Phys Rev 1954,94 1498 References 26 27 H Basch Chem Phys Lett 1987,136,289 T H Upton J Chem Phys 1987,86,7054 1 2 3 4 5 6 7 8 9 10 11 12 13 14 R 0 Jones J Chem Phys 1993,99,1194 P Ballone and R 0 Jones J Chem Phys ,1994,100,4941 W Andreoni Phys Rev B 1992,45,4203 G-X Qian R M Martin and D J Chadi Phys Rev B 1988 38,7649 T Ohno Phys Rev Lett 1992,70,631 E Kaxiras Y Bar-Yam J D Joannopoulos and K C Paudey Phys Rev B 1987,359625 R Car and M Parinello Phys Rev Lett ,1985,55,2471 K Laasonen and R M Nieminen J Phys Condens Matter 1990 2,1509 C H Xu C Z Wang C T Chan and K M Ho J Phys Condens Matter 1992,4,6047 M Menon and K R Subbaswamy Phys Rev B 1993,47,12754 C Mailhiot C B Duke and D J Chadi Surf Sci ,1985,149,366 C Molteni L Colombo and L Miglio Phys Rev B 1994 50 4371 G Seifert H Eschrig and W Bieger Z Phys Chem (Leipzig) 1986,267,529 D Porezag,.Th Frauenheim,. Th Kohler R Kaschner and 28 29 30 31 32 33 34 35 36 37 38 39 40 K Balasubramanian and P Y Feng Chem Phys Lett 1988 146,155 U Meier S D Peyerimhoff and F Grein Z Phys D 1990,17,209 G Pacchioni and J Koutecky Ber Bunsenges Phys Chem 1984 88,242 K Jug H P Schluff. H Kupka and R Iffert J Comput Chem 1988,9,803 L G M Petterson C W Bauschlicher Jr and T Halicioglu J Chem Phys 1987,87,2205 K Raghavachari Bull Am Phys Soc 1990,35,606 G Bachelet D R Hamann and M Schluter Phjs Rev B 1982 26,4 199 G Igel-Mann H Stoll and H Preuss Mol Phys 1993,80,325 K Balasubramanian K Sumathi and D Dai J Chem PIzys 1991 95,3494 J S Kane and J H Reynolds J Chem Phys 1956,25,342 J M Dyke,S Elbe1,A Morrisand J C H Stevens J Chem SOC Faradny Trans 2,1986,82,637 Y Morino T Ukaji and T Ito Bull Chem SOCJpn 1966,39,64 J Andzelm N Russo and D R Salahub Chem Phys Lett 1987 142,169 G Seifert Phys Rev B 1995,51 12947 41 U Meier S D Peyerimhoff and F Grein Chem Phys 1991 15.Th Frauenheim F Welch,. Th Kohler D Porezag G Seifert and 150,331. 16 S Uhlmann Phys Rev B 1995,52,11492 J Widany,. Th Frauenheim,. Th Kohler M Sternberg D Porezag and G Jungnickel Phys Rev B 1996,52,4443 42 43 J Ihm and J D Joannopoulos Phys Rev B 1981,24,4191 Landolt-Bornstein Elastic Piezoelectric and Related Constants vol III/l 1 of New Series Springer-Verlag Berlin 1979 17 18 G Seifert and H Eschrig Phys Status Solidi B 1985,127 573 H Eschrig and I Bergert Phys Status Solidi B 1978,90,621 Paper 6/00703A Received 30th January 1996 1656 J Muter Chem 1996 6(lo) 1649-1656

ISSN:0959-9428    

DOI:10.1039/JM9960601649

出版商:RSC    

年代:1996

数据来源: RSC

摘要:

Stability of silicon carbide structures: from clusters to solid surfaces Rafael Gutierrez," Thomas Frauenheim," Thomas Kohler" and Gothard Seifertb "Theoretische Physik III, Institut fur Physik, Technische Universitat, D-09107 Chemnitz, Germany bTechnische Universitat, Institut fur Theoretische Physik, Mommenstrasse 13, D-01062 Dresden, Germany We present a density-functional based non-orthogonal tight-binding (DF-TB) Hamiltonian in application to silicon carbide. The Kohn-Sham orbitals of the system are represented by a linear combination of atomic orbital (LCAO) equation with respect to a minimal basis of the localized valence electron orbitals of all atoms. Within a two-centre approach all Hamiltonian and overlap matrix elements are derived in a parameter-free way via the construction of pseudo-atomic orbitals and potentials by self- consistent single-atom calculations using the local-density approximation (LDA).This is in favour of a tabulation of the corresponding Slater-Koster integrals us. distance. Making use of a non-self-consistent solution of the Kohn-Sham equations for the many-atom structure and an adjustment of the universal short-range repulsive two-particle potentials with respect to self- consistent field (SCF)-LDA results in the method becoming sufficiently accurate to obtain the total energy of all-scale silicon carbide structures and this is transferable and efficient for predictive molecular-dynamics simulations. We present results for the energetic stability and properties of various microclusters and molecules including interactions with hydrogen.We give proof of the stability of the solid-state modifications and calculate the vibrational density of states for the most stable zinc blende structure. In addressing further applications to surface properties, we discuss the (1 x 1) reconstruction of the (1 10)Sic surface. For a theoretical description of the ground-state properties of clusters, molecules and solid-state modifications, different approaches have been developed. Self-consistent field (SCF) Hartree-Fock (HF)lq2 and density-functional theory (DFT) calculation^^.^ including correlation effects at different levels of approximation [many-body perturbation theory, local-density approximation (LDA)] yield very accurate results. However, owing to the increasing computing time necessary, they become unpracticable at larger particle numbers.In particular, for implementation into molecular-dynamics (MD) simulations investigating structure formation at surfaces and interfaces, more efficient approaches are highly desirable. In contrast to the aforementioned ab initio methods the empirical potential approaches are, unfortunately, in many cases not transferable to a wider class of systems than they have been derived for. They become inappropriate for most structural simulations of predicting character. The present method belongs to a third class of structural simulation methods, which has received increasing attention in recent years: a hybrid between an ab initio method and a parametrized tight-binding (TB) It differs from the empirical TB approaches, where the Hamiltonian and overlap matrix elements are derived from fitting to an equilibrium structure data base, in that they are calculated straightfor- wardly, using a basis of atomic-like orbitals and potentials derived from DFT-LDA calculations on contrasted pseudo- atoms.Similar to the parametrized TB schemes, only two- centre integrals are retained and the total energy is composed of a 'band-structure energy' (sum of occupied Kohn-Sham orbitals) and a universal short-range repulsive two-particle potential, which additionally has to be adjusted to the differ- ence of the band-structure energy and SCF-LDA cohesive energies of proper reference systems.In two recent publication^^*^ this method has been proven to be highly transferable for complex C(H) and Si(H) struc- tures. In an extension of this work, the method will be firstly applied to various scale silicon carbide structures (Sic), ranging from small clusters to crystalline solids and solid surfaces. We will benchmark our method by comparison with recent studies on small Si,C, clusters and molecules within HF theory including correlation effects through Moller-Plesset (MP) perturbation theory and configuration interaction (CI) Furthermore, the properties of solid-state modifi- cations are compared with ab initio pseudo-potentials21 and SCF-TB calculations.22 The considered surface reconstruction on Sic( 110) is related to recent results of Bechstedt et Sabisch et and Mehandru and Anderson.25 The paper is organized as follows.First, we give a short outline of the theoretical method. Next, we present the results on small clusters, molecules and solids, emphasising the dis- cussion of low-energy configurations and relative stabilities. As first application to the study of the surface properties, we determine the stable surface reconstruction of Sic (110) and finally give a short summary. Density-functional tight-binding method Since the method will be addressed to calculations of total energy and interatomic forces in MD structure simulations, it will be based on first principle concepts. Instead of solving the Kohn-Sham equations self-consistently, we construct a non- orthogonal TB Hamiltonian within the LCAO-LDA frame-work of DFT.We will give only a brief outline of the main ideas and refer the reader for more detail to refs. 7, 26 and 27. Contracted atomic-like valence orbitals #p (v-Rk) and potentials are generated by an SCF solution of the Kohn-Sham equations within LDA for a single pseudo-atom. The pseudo- atoms are constructed by incorporating an additional con-finement potential (~/r~)~in the effective single-particle poten- tials. The confinement radius ro is related to the covalent radius of the considered atom type and does not involve further parametrization. The minimal valence basis thus generated is used in an LCAO equation for representing the wavefunctions of the extended system: $i(v)=C c;#p(v-Rk) (1) pc.k Within a two-centre approximation all Hamiltonian and overlap matrix elements HpV,S,, are then calculated straight- forwardly within this basis.Since the three-centre and crystal- field integrals are neglected, all matrix elements depend only on the interatomic separation. They are calculated once for each atom type combination in favour of the tabulation of J. Muter. Chem., 1996, 6( lo), 1657-1663 1657 Slater-Koster integrals,? which may be used for many-atom calculations of a large variety of different-scale structures. Within our non-self-consistent treatment for the many-atomic structure the Kohn-Sham equations can be trans-formed into an algebraic system leading to a general eigenvalue problem: Using simple matrix diagonalization techniques we determine the single particle energies E~ and the eigenstate expansion coefficients.As is commonly accepted,28 the total energy can be written as the sum of a 'band-structure energy' Ebs and a repulsive energy Erep: occ Etot=EbsfErep=xEZ+ 1U(IRZ-RkI) 1 l<k The latter term as a universal short-range repulsive two- particle interaction is determined as the difference of the band- structure energy and the SCF cohesive energies of the diatomic molecule and the crystalline zinc blende Sic modification. In Fig. 1 the dependency of the matrix elements and the repulsive potential for Sic on the interatomic separation is shown. Corresponding results for carbon and silicon have been pub- lished re~ently.~?' Results Molecules As a first test of our method we have investigated several carbon-substituted silane molecules.In general, the structural parameters and symmetries of the investigated systems are in good agreement with ab initio calculations. The Si-Co bond length in all cases 1,s slightly overestimated (0.01-0.03 A) Fnd ranges from 1.663 A in monosubstituted silenes to 1.946 A in cyclopolysilane. The bond lengths and angles of the considered molecules are summarized in Table 1; the molecules are shown in Plate 1. HSiCH, H,CSiH,. There is no experimental evidence for the existence of the silene molecules HSiCH. CI calculation^'^ give a bent structure as the most stable form, followed by a linear chain about 9 kcal mo1-I higher in energy (1 cal= 4.184 J).The DF-TB method yields the same energetic order but a slightly reduced energy djfference of AE ~7 kc$ mol-I. The Si-C bond length (1.663 A) is enlarged by 0.03 A relative to the ab initioovalue. Compared with H2CSiH2 it is shortened by about 0.08 A, probably owing to the formation of a partial double bond in the latter compound. The bond angles in bent HSiCH, however, show larger deviations from the ab initio values, AZZ 15". In Table 2 the normal mode vibrational frequencies for H2CSiH2 are shown and compared with CI(d,+p) re~u1ts.l~In the last column we give the relative error taking as reference the experimental frequencies. The largest deviations are found for the high-frequency stretching modes, while lower lying modes are better reproduced.H,SiCH,. The structural parameters for this molecule agree quite well with SCF/3-21G16 calculations. The bond angles at the silicon atom are closer to the typical value of 109" for sp3 hybridization than found by ub initio calculations and the Si-C bond is lengthened slightly. Si3C3H6. This molecule forms a ring structure of D3,, symmetry, with alternating Si and C atoms, see Plate 1. The calculated bond lengths and angles agree very well with the t The Slater-Koster tables for Si-C, C-C, C-H, Si-H, H-H and Si-Si are available upon request from the authors. 1658 J. Muter. Chem., 1996, 6(10), 1657-1663 -0.4 h ---' -0.6 t 0.8 -(b) ___. s, s,,\ \ s, \ \ s,\ -\0.3 \ __ s, \ \ l .Y' 1 -0.7 2 4 6 8 0 bond length/a, Fig. 1 (a)Hamiltonian (HMv)and (b)overlap S,,, matrix elements, and (c) repulsive potential, Erep,us. interatomic separation ab initio re~u1ts.I~ Only the C-H bond is found to be slightly longer in the present approach. Silabenzene, SiC,H,. Our results are compared with SCF/3- 21G*18 and ST02G19 calculations, see Table 1. The bond lengths and angles are described reasonably well. The C-Si-C bond angle is underestimated by about 6%. Cyclopolysilane,(SiH,),CH, and 1,3-substituted cyclotetrasil- ane, (SiH2)* (CH,),. These compounds form isoscale and square configurations, respectively, with hydrogen pairs bonded out-of-plane to each atom.The Si-Si bond length in cyclopolysilan~ (see Table 1) is shorter compared with that of disilane (2.34 A). The structural parameters of these molecules agree well with the ab initio result^.^^^^^ The 1,3-substituted tetrasilane molecule is stretched slightly along the Si-Si direction according to our method, yieldigg an overestimation of the Si-C bond length by about 0.02 A. Table 1 Geometric properties of different C-substituted silane molecules (see Plate 1) compared with ab initlo calculations bond length/A, bond angleldegrees molecule symmetry HSiCH (silene) cm S1-Hs1-c C-H bent-H Sic H c, s1-c S1-H C-H H-Si-C H-C-H H,CSiH, C21 si-c (silaethylene) Si-H C-H H-Si-H H-C-H H3SlCW3 si-c S1-H C-H H-C-H H-SI-H S13C3Hh s1-c Si-H C-Hc-s1-c s1-c-Sl SiC5H6 s1-c (silabenzene) Si-H C-H c-c c-c-c c-s1-c S1-C-H si-c-c (SlH,),CH, s1-c (cyclopolysilane) Si-H C-H s1-Sl H-Si-H s1-c-Sl ( S1Hz)2(CHd2 s1-c ( 1,3-~ubstltuted) S1-H cyclotetrasilane) C-Hsi-c-s1c-s1-c H-C-H H-Si-H Small silicon carbide clusters In this section we present results for small silicon carbide clusters Si,C, We will discuss the minimal energy configur- ations and the energetic differences to different low-energy metastable configurations For comparison we choose as refer- ence recent high-level ab znitzo Hartree-Fock calculations including correlation effects on the MP2 level and by CI of refs 9-14 Conjugate gradient relaxation techniques have been per- formed to obtain the minimal energy clusters In Table 3, characteristic bond lengths and angles of the ground-state geometries found with the DF-TB method are given, the Sic clusters are shown in Plate 2 For comparison the ab znztzo data of refs 9-14 have been included In general, the symmetry of the investigated clusters and their geometric parameters are in good agreement with the ab znztio calculations The energetic differences of various metastable structures yield the same trends as obtained on the ab znztzo level (see Fig 2) DF-TB ab initto ref 1459 1451 1614 1587 1108 1075 1663 1635 1471 1467 1127 1082 144 9 128 8 135 17 150 1 1 74 1703 1475 1459 1127 1083 118 1 114 5 109 115 1946 1917 16 1484 1 49 112 1082 104 6 108 3 109 56 107 8 1777 1766 17 1482 1483 1140 1094 119 8 118 7 120 1 121 3 1775 176 19 178 1472 1465 1477 111 1073 1092 1394 1391 1 400 124 9 101 3 125 2 107 4 116 6 107 5 126 6 125 4 124 1 118 2 125 9 118 2 1946 1901 15 1482 1464 1127 1084 2 266 2 251 114 1 112 6 71 2 72 6 1946 1923 15 1484 113 91 5 85 9 88 53 94 0 103 7 111 1 Si,C.Ab initio calculations9 using double-zeta plus double polarisation basis set show that the minimum on the potential- energy surface corresponds to a bent structure wit! Czv symmetry and geometrical parameters r(Si-C)= 1 670 A and Si--S-Si=133" With our method we obtain r(Si-C)= 171 A and Si-C-Si=136", in good agreement with the results of the more sophisticated calculations However, the consideration of electron correlation within CI gives a smaller Si-C-Si bond angleJ120") and a slightly increased Si-C bond length of 1686 A The bent structure is energetically followed by a linear chain at a slightly higher energy (AEw 0 95 kcal mol-I), in reasonable agreement with the ab znrtzo results (AEwO6-2 1 kcal mol-l) Sicy The energies of the different Sic3 isomers obtained with our method are displayed in Fig 2(u) in comparison with ab inztio results lo We obtain a good agreement in the relative stabilities of the different configurations However, while the J Mater Chew, 1996, 6(lo), 1657-1663 1659 Table 3 Geometric properties of the SIC clusters (shown in Plate 2) in comparison with ab initio data" (only the structural parameters of the minimal energy configurations found with the present method are given) bond length/A, bond angleldegrees molecule symmetry DF-TB ab initio" CSi2 c2, Si-C 1.71 1.67 Si-C-Si 130 120 Sic, (IV) CZc C(2)-C(3) 1.3 10 1.298 C(3)--C(4) 1.275 1.307 Si( 1)-C(2) 1.751 1.722 Si,C, (11) C, Si(1)-C(2) 2.11 I .92 C(2)--C(4) 1.24 1.23 C(4)-Si(6) 1.86 1.88 C(2)-Si( l)-C(3) 101.4 101.8 C(4)-Si( 6)-C( 5) 81.99 82.2 Si( 1)-Si(2) 2.55 3.1 1 Si(2)-C( 5) 1.84 1.67 C(5)-C(6) 1.33 1.32 Si( 3)-C( 5) 1.807 1.83 C(5)-C(4) 1.30 1.28 Si( 3)- Si( 1) 2.56 2.47 C(S)-Si( 1) 2.32 2.19 -~Si(6)-Si(3)-Si( 1) 59.69 -Si(6)-Si(2)-Si( 1) 59.69 C( 1 )-C(2) 1.335 1.38 C(l)-Si(3) 2.16 1.92 C(1)-Si(5) 1.727 1.738 Si( 3)- Si(5) 2.56 2.563 "Refs 9-14.Plate 1 Structures of C-substituted silane molecules in Table 1. Si, yellow circles; C, grey circles; H, green circles. Table 2 Vibrational frequencies of silaethylene; the corresponding ab initio results [CI (d, +p)] and experimental values are given as reference molecule rep. DF-TB CI(d, +p)" exp." error (%)b ~ CH2SiH, a, 958 989 927 3 (7) 1191 1041 985 20 (6) 1526 1476 1350 13 (10) 2733 2388 2229 18 (4) 2994 3254 ----a2 640 750 --bl 495 438 72 1 801 741 -2 (8) Plate 2 Structures of the Sic clusters in Table 3. Si, yellow circles; C, b2 286 501 green circles.834 876 817 2 (7) 2757 2404 2239 23 (7) 3059 3352 corresponding to isomer IV (all roman numerals correspond a Ref. 15.b Relative errors with respect to the experimental results to the notations in ref. 10). This strong competition between (values in parentheses correspond to the ab inito calculations). the linear and ring structures has also been observed in the ab initio calculations." The ground state obtained at the ah initio ab initio calculations give a strong stabilization of a four- level," within the present calculation scheme, is determined to membered ring, having Czvsymmetry, with a transannular be 30 kcal mol-' less stable than the linear chain coming carbon-carbon bond (111), we find the linear structure with fourth in the energetic order [see Fig.2(a)]. Interestingly, the the Si atom at the end of the chain as the energetically lowest- authors report that a linear arrangement would be the ground lying isomer, 1. This is followed by a ring structure without state using only a double-zeta basis, while including polariz- C-C cross bonding at about 2 kcal mol-' higher energy, ation function the ring structure is favoured. This indicates 1660 J. Muter. Chem., 1996, 6(lo), 1657-1663 150 100 50 0 -100 1 t-200 8 Y 12 150 100 50 0 'd X xv -50 0 Fig. 2 Relative stabilities, AE/kcal mol-' (1 cal=4.184 J) of Sic clusters calculated within the DF-TB method (---); (-) correspond to the ab initio calculations. (a) Sic,; (b) Si,C,; (c) Si,C,; (d) Si,C,; (e) Si,C,.The energy differences of the Sic isomers refer to the ground states found within ab initio calculations. that the energetic order can be affected strongly and even reversed by changing the size of the basis set. Si2C4. The ab initio calculations'' including correlation effects at the MP2 level yield a linear chain as the most stable configuration followed by a chair-like ring structure with C, symmetry. Within our calculations, the latter is not a stable point on the potential-energy surface, but it relaxes into a planar ground-state configuration with C, symmetry (see Table 3). However, the bond lengths and angles of this structure correspond quite well with those of Miihlhauser et al." with the exception of the Si( 1)-C(3) bond length, see Table 3.Note that this structure is relatively 'soft' with respect to out- of-plane distortions. A planar structure with D,,symmetry (I) is determined with almost the same energy, AEz0.16 kcal mol-I whereas the linear chain [I11 in Fig. 2(b)], is about 25 kcal mol less stable than structure 11. Three-dimensional Si,C, configurations IV and V are found to be highly unstable and confirm the ab initio results. Si3C3' This cluster shows a very complex potential-energy surface with a plenty of stable isomers, see Fig. 2(c). Since many of these structures have almost the same energy, the energetic stability is expected to be very sensitive to the size of the basis set and approximation used.According to our results, the structure with C, symmetry, X, displayed in Table 3, is determined as the ground state, followed by an edge-capped trigonal bipyramid, XII, also having C, symmetry. The ground state derived from ab initio calculations," a tetrahedral Sic, system faced-capped by two Si atoms, I, comes third in our calculations, about 12 kcal mol-' higher in energy. Si4C,. As expected, by increasing the Si/C ratio in the clusters, three-dimensional structures become favourable over linear or planar arrangements. A non-planar structure with C,, symmetry [I1 in Table 3 and Fig. 2(d)],is clearly the most stable structure within our treatment as well as in the ab initio calculations." Also the energetics of the other isomers agree quite well with the ab initio results.Only the extreme high energy of structure IV could not be confirmed in our calcu- lations. The bond angles and bond lengths determined by the DF-TB calculations agree well with those of the ab initio calculations for nearly all the structures considered. In some cases the Si-Si and the Si-C bond lengths are overestimated by a few percent in the present scheme, see Table 3. Si,C2. For this cluster we again can confirm the ab initio prediction~.'~The ground state, displayed in Table 3, is a J. Muter. Chem., 1996,6(lo), 1657-1663 1661 planar structure with CZVsymmetry Isomer I1 is higher in energy by about 20 kcal mol-', and structure 111 is even less stable, in good agreement with the ub initio calculations, see Fig 2(e) Solid modifications and surface reconstruction To address applications of the present method to bulk systems we have performed total energy calculations as a function of the Si-C bond length for the zinc blende (zb) and rocksalt (rs) structures The calculations make use of a r-point Brillouin-zone sampling, which is a valid approximation for the considered large periodic supercells including 216 atoms The cohesive energy curver per atom $re shown in Fig 3(u) The equilibrium bond length o,f 1 887 A agrees very well with the experimental value of 1 89 A 29 The cohesive energy Ecoh= 6 93 eV atom -matches quite well the experimental (6 34 eV atom ') data and values obtained with an ah znztzo pseudo-potential (666eV atom 1)21 and SCF-TB (634eV atom-') calculations 22 In calculating the cohesive energies we have chosen the spin-polarised atoms as reference, ze the spin- polarisation energies of 0 64 eV for Si and 1 12 eV for C have been subtracted from the LDA values The rocksalt modifi- cation is obtained to be metastable and about 1 2 eV atom higher in energy than the zb structure This may be compared to the very high value of ca 4 eV atom reported by SCF-TB calculations 22 In addition, the phonon spectrum within the harmonic approximation of the zinc blende modification has been obtained by solving the eigenvalue problem HY=02MY, where H is the dynamical matrix, M denotes a matrix with -45 rocksan 0 ' .0.0 zinc blende vlcm-' Fig.3 (u) Cohesive energy per atom for the zinc blende and rocksalt silicon carbide structures us bond length obtained with respect to the spin-polarized Si and C atoms (b) Vibrational density of states for zinc blende (3C) silicon carbide The insert is the ab inztio density of states taken from ref 31 1662 J Muter Chem, 1996, 6(10), 1657-1663 the atomic masses on the main diagonal and CL) and Y are the eigenvalues and the corresponding eigenvectors, respectively The vibrational density of states for a 216-atom supercell of the zinc blende modification was then calculated The obtained spectrum of eigenvalues, which was broadened by a Lorentzian of constant width 30 cm , is plotted in Fig 3(h) The comparisons with recent ab ~izztiodata obtained by Bechtedt et a1 30 yields a good qualitative and partly quantitative agreement The frequency region around 750cm contains two dominant peaks which correspond to SirC stretching vibrations The peak positions and relative intensities are quite well described The gap in the spectrum between about 600 and 730 cm-' is slightly widened within the DF-TB calculations compared with the ah znztzo results,30 see also the insert in Fig 3(b) Finally, as a first application to surface structures, we have investigated the reconstruction of the clean (110) surface in the zinc blende modification of Sic using commonly accepted periodic surface-slab techniques 31 The model consists of 7 (110) layers including 9 C and 9 Si atoms each The bottom layer during the relaxation was held fixed in order to simulate the effect of an infinite crystal substrate and dangling bonds were removed by hydrogen saturation The results are com- pared with ah rnztto calculation^^^ 24 as well as with atomic superposition and electron delocalization (ASED) band tech- niques 25 The most relevant structural parameters are displayed in Table4 and the (110) surface layer geometry is shown in Plate 3(u) and (b) In accord with the predicitons of the aforementioned studies we find a buckling of the first surface layer The top layer Si-C bond is contracted in tomparison with the unrelaxed (I x 1) surface, 1 82 and 1 89 A, respectively The determined bond length is slightly larger than that predicted in refs 23 Table 4 Structural parameters of the reconstructed Sic( 110) surface (see Plate 3 for notation of atoms) bond lengt h/A, bond angleldegrees sic ( 110) DF TB u6 initio method r( 1)-(2) 1826 1767 DFT LDA" 41)-(4) 2 09 1761 186 189 ASEDb r(2)-(3) 1832 1823 1 84 (2)-(1)-(4) 9651 99 8 106 C-Si-c 1153 1202 122 (1)-(2) -(3) 119 1 1164 112 "Ref 24 bRef 25 Plate 3 1 x 1 ( 110) surface reconstruction of zinc blende silicon carbide (a)side view (b)top view and 24 [r(Si-C)= 1 76 A] The silicon and carbon atoms in the first surface layer relax parallel as well as perpendicular to the surface While both atom types are displaced outwards, the C atoms are more strongly pulled out of the surface, yielding the observed buckling Considering the substrate, we find a relatively strong relaxation of the subsurface layer, too Silicon atoms of the second layer relax slightly inwards and the carbon atoms move slightly outwards Aso a consequence, a quite long C,,, -Si,,b bond length of 2 07 A is established When compared with the ab znztzo results, this effect causes an enlarged twist-angle and an increased byckling height, +z, normal to the surface, AzDF TB =O 397 A (AZSCF =O 234 A), which is responsible for the shear distortion in the surface layer The Csub -Si,,, -C,,, and C,,, -Si,,, -C,,, bond angles of about 119 and 115", respectively, indicates sp2 hybridization of the Si atoms in the surface layer In general, the bond angles agree well (within 2-4%) with the ab znztzo results The energy gain of 0 72 eV dimer-' after the relaxation is only slightly larger than the values obtained by Bechstedt et al 23 (063 eV dimer '), Sabisch et (064 eV dimer-') and the ASED result of Mehandru and Anderson25 (064 eV dimer -') Summary We have presented a scheme for the determination of non- orthogonal tight-binding Hamiltonian and overlap matrix elements for SIC on the basis of density-functional theory Despite its simplicity the method has been proven to be transferable to all-scale silicon carbide modifications without addiitonal parametrization of the total energy By applying the method we obtain reliable results for the ground-state configurations of small clusters, molecules and bulk crystalline modifications The energetic differences between different meta- stable states of small clusters as well as their structural parameters and symmetries are described well In a first application to the study of Sic surface reconstruc- tions we confirm the buckled dimer 1 x 1 reconstruction on the Sic( 110) surface found in ab znztzo calculations Various other applications to the study of crystalline defect structures and the ( loo), ( 111) surface reconstructions of silicon carbide as well as amorphous a-SIC H modifications are now in progress and the results will be published elsewhere We gratefully acknowledge support from the Deutsche Forschungsgemeinschaft References 1 K Raghavachari, J Chem Phys ,1986,84,5672 2 K Raghavachari and C M Rohlfing, J Cheni Phys 1988, 89, 2219 3 M R Pederson and K A Jackson, Phys Rev B, 1990,41 74.53 4 K Laasonen and R M Nieminen, J Phys Condenv Matter, 1990, 2,1509 5 0 F Sankey and D J Niklewski, Phys Rev B, 1989,40,3979 6 M Menon and K R Subbaswamy, J Chem Phys 1993,47,12 754 7 D Porezag, Th Frauenheim, Th Kohler, G Seifert and R Kaschner, Phys Rev B, 1995,51 12 947 8 Th Frauenheim, F Welch, Th Kohler, D Porezag and G Seifert, Phys Rev B, 1995,11492 9 R S GrevandH F SchaeferII1,J Chem Phys 1985,82 4126 10 I L Alberts, R S Grev and H F Schaefer 111, J Chem Phys 1990, 93,5046 11 M Muhlhauser, G Froudakis, A Zdetsis, B Engels and S D Peyerimhoff, J Chem Phys 1994,101,6790 12 M Muhlhauser, G Froudakis, A Zdetsis, B Engels, N Flytzanis and S D Peyerimhoff, 2 Phys D,1994,32,113 13 M Muhlhauser, G Froudakis, A Zdetsis and S D Peyerimhoff, Chem Phys Lett, 1993,204,617 14 G Froudakis, M Muhlhauser and A Zdetsis, preprint 15 Y Apeloig, in The Chemistry of Organic Silicon Compounds, ed Patai and Z Rappoport, Wiley, Chichester, 1989 16 W J Hehre, L Radom, P v R Schleyer and J A Pople Ah Initio Molecular Orbital Theory, Wiley, Chichester, 1986 17 K B Wiberg and K E Laidig J Org Chem, 1992,57,5092 18 K K Baldridge and M S Gordon, J Am Chem Soc 1988 110,4204 19 K K Baldridge and M S Gordon, Organometallics, 1988, 144, 7 20 R S Grev and H S Schaefer 111, J Am Chem Soc 1987, 109, 6577 21 K J Chang and M Cohen, Phys Rev B, 1987,35,8196 22 M Kohyama, S Kose, M Kinoshita and R Yamamoto, J Plzys Condens Matter 2, 1990,2,7791 23 B Wenzien, P Kackell and F Bechstedt, Surf Sci , 1994 307,989 24 M Sabisch, P Kruger and J Pollmann, Phys Rev B 1995, 51, 13367 25 S P Mehandru and A B Anderson, Phis Rev B, 1990 42,9040 26 G Seifert and R 0 Jones, 2 Phys D,1991,20,77 27 P Blaudeck, T Frauenheim, D Porezag, G Seifert and E Fromm, J Phys Condens Matter, 1992,4,6389 28 W M C Foulkes and R Haydock, Phys Re0 B, 1989 39, 12520 29 Physics of Group IV Elements and 111-V Compounds vol 17a of Landolt-Bornstein Tables, ed 0 Madelung, M Schulz and H Weiss, Springer-Verlag, Berlin, 1982 30 M Hoffmann, A Zywietz, K Karch and F Bechstedt Phys Re0 B, 1994,50,13401 31 Th Frauenheim, U Stephan, P Blaudeck, D Porezag, H-G Busmann, W Zimmerling-Edling and S Lauer, Phys Rer B, 1993,48,18189 Paper 6/00593D, Recezued 25th January, 1996 J Mater Chem, 1996, 6(10), 1657-1663 1663

ISSN:0959-9428    

DOI:10.1039/JM9960601657

出版商:RSC    

年代:1996

数据来源: RSC

摘要:

Mixed oxides Si02-Zr02 and Si02-Ti02 by a non-hydrolytic sol-gel route Mahandrimanana Andrianainarivelo, Robert Corriu, Dominique Leclercq, P. Hubert Mutin" and AndrC Vioux Laboratoire des Prkcurseurs Molkculaires de Matkriaux, UMR CNRS 44, Universitk Montpellier 2, case 007, Place E.Bataillon, 34095 Montpellier Ckdex 5, France SO,-ZrO, and SiOz-TiOz mixed oxides with various metal contents have been prepared by a non-hydrolytic sol-gel route involving the condensation between chloride and isopropoxide functions at 110"C. Well condensed, monolithic gels were obtained in one step, without the use of additives. The Si/M ratio of the oxide may be controlled easily by the composition of the starting mixture. The Si/Zr oxides remained amorphous after calcination for 5 h at 600 "C; IR and 29Si NMR spectroscopy showed a large amount of Si-0 -Zr bonds, indicating a homogeneous distribution of the components on the atomic scale.The crystallization of tetragonal zirconia took place at higher temperature; the transformation of tetragonal to monoclinic zirconia was strongly retarded and did not take place after 2 h at 1300"C. The crystallization of zircon (for a sample containing 50 mol% Zr) started at 1500"C and was completed after 20 h at 1500"C. IR spectroscopy indicated the presence of a limited number of Si-0-Ti bonds in all the Si/Ti oxides after calcination for 5 h at 500 "C. The sample within the stable glass region (5 mol% Ti) appeared perfectly homogeneous: it crystallized at 900 "C as single-phase cristobalite oxide, with Ti4+ ions substituting Si4+ ions at random.On the other hand, the precipitation of anatase was observed for the %/Ti oxides with a high Ti content (20-50 mol% Ti), which are outside the stable glass region. The transformation of anatase to rutile was not observed even after 2 h at 1300"C. The technological interest in silica-titania and silica-zirconia mixed oxides arises from their chemical resistance and their thermomechanical or optical properties: Si0,-TiO, glasses and zircon, SiZrO,, are characterized by very low thermal expansion, which confer them a high thermal shock resistance; Si0,-TiO, and SiO,-ZrO, glasses have high refractive indices, and they are also of great importance as catalysts or as cata!yst supports.Owing to their refractoriness, these oxides are difficult to produce by conventional melting techniques. Accordingly, the sol-gel processing of alkoxides,' which allows the prep- aration of glasses at low temperature, has been used widely for the preparation of Si02-Ti022-7 and Si0,-Zr028-10 glasses or zircon."-'4 The homogeneity of binary oxide gels has a great influence on the structural evolution of the gels during the heat treat- ment. However, obtaining homogeneous Si0,-Ti02 or Si0,-ZrO, gels by hydrolysis/condensation is not straightfor- ward, owing to the difference in reactivity between silicon alkoxides and transition-metal alkoxides, leading to the fast formation of M-0-M bonds and to the precipitation of the metal oxide. To overcome this limitation several procedures have been elaborated, such as the pre-hydrolysis of the less reactive silicon alk~xide~.~.' or the stabilization of the metal alkoxide by comple~ation.~~'~ The aim of these procedures is to promote the formation of mixed Si-0-M bonds, i.e.homogeneity on the atomic scale. In recent years, we have been developing a new sol-gel route, based on non-hydrolytic reaction^.'^-'^ In this paper we wished to investigate the application of these reactions to the preparation of SiO,-ZrO, and SiOz-TiOz mixed oxides with various metal contents. We used IR and 29Si solid-state NMR spectroscopy to evidence mixed Si-0-M bonds. The crys- tallization behaviour of the mixed oxides as a function of the metal content and the calcination temperature was studied using powder X-ray diffraction.Background In conventional sol-gel processes, the formation of sols or gels results from the formation of M-0-M bridges through hydrolysis and polycondensation reactions. In the non-hydrolytic sol-gel process reported here, M -0-M bridges are obtained by condensation between halide and alkoxide groups, with elimination of alkyl halide: M-X+M-OR-M-0-M+R-X In most cases these reactions are thermally activated and, depending on the reactives involved, temperatures ranging from room temperature to about 100"C are required. Actually, the condensation competes with the redistribution of the ligands, which usually takes place rapidly at room temperature and leads to a complicated mixture of halogenoalkoxides:20-22 MX,+M(OR),+MX,(OR),-, with Odxdn To avoid the use of alkoxides, which are often quite expensive, the alkoxide groups may be formed in situ by the etherolysis of the metal halide by an organic ether (such as diethyl or diisopropyl ether): l6,l9 M-X+R-OR-M-OR+R-X The stoichiometry of these reactions requires an equal number of alkyl groups (borne by the alkoxide or by the ether) and halide groups.Thus, the idealized equations of reaction corre- sponding to the preparation of mixed oxide gels with different M/Si ratios are: halide-alkoxide route: x MX4 + y M(OR), + z M'X, + t M'(OR), -, M(x+y)M)1=+t~02(x+y+z+t)+4(X+Z) R-X with x+z=y+t, M=Si, M'=Ti or Zr halide-ther route: x MX4+y M'X4+2(x+y) ROR+M,M'yO~(,+y)+ 4(x +y)R-X with M = Si, M' =Ti or Zr Experiment a1 Si/Zr and Si/Ti gels with varied compositions were prepared from chlorides and isopropoxides (halide-alkoxide route) or chloride and diisopropyl ether (halide-ther route) (Table 1). J.Mater.Chern.,1996, 6(lo), 1665-1671 1665 Table 1 Preparation of the gels sample moles of reactantsa moles of solvent 1 Si/ZrA 1 ZrCl,+ 1 Si(OPr'), 10 CH,Cl, 1Si/ZrB 1 ZrCl, +1 SiCl, +4Pr,'0 10 CH,Cl, 5Si/Zr 1 ZrC1, +2 SiC1, +3 Si(OPr'), 4 CHZC1, 10Si/Zr 1 ZrC1, +4.5 SiCI, + 5.5 Si(OPr'), 30 CH,C1, 99Si/Zr 49 SiC1, +50 Si(OPr'), +1 ZrC1, 110 CH,Cl, 1 Si/TiA 1 Ti(OPf),+l SiC1, 2 CHCI, 1 Si/TiB 1 TiC1, +.l SiC1, +4Pr,'0 10 CHC1, -4Si/Ti 1 Ti(OPf),+ 2.5 SiCl, + 1.5 Si(OPf), 4Si/Ti 1.5 Si(OPr'),+ 2.5 SiCl, + 1 Ti(OPr'), 6 CHCI, -20Si/Ti 19 Si(OPr'),+20 SiC14+ 1 Ti(OPr'), "In the order of addition.Starting materials Silicon tetrachloride (Aldrich), titanium tetrachloride (Aldrich), zirconium tetrachloride (Jansen) and titanium tetraisopropox- ide (Jansen) were used as received. Silicon tetraisopropoxide was prepared from silicon tetrachloride and isopropyl alcohol according to the method described by Bradley and Hill.23 Diisopropyl ether, dichloromethane and chloroform (Aldrich) were distilled before use. Preparation of the samples The starting solutions were obtained by adding the silicon compound(s) to the metallic compound in a Schlenk tube (Table 1). Solvent was added when needed to obtain a clear solution.After mixing at room temperature, the starting solu- tions were transferred into sealed tubes and heated for 4 days at 110°C. In all cases monolithic gels formed in less than 1 day. The tubes were opened in a glove box under argon, the gels washed with CHCI, and dried for 4 h at 110°C under vacuum, leading to fine, brown powders. The oxides were obtained by the calcination in air of the dried gels, leading to white powders. The overall oxide yield was better than 90% in all cases. Characterization techniques X-Ray powder diffraction was used with Cu-Ka radiation to identify the oxide phases after thermal treatment (Siefert MZ IV diffractometer). Thermal analysis was performed in dry air, at a heating rate of 10 K min-l, on a Netzsch STA 409 thermobalance.Specific surface areas were determined by N, adsorption-desorption experiments using the BET method, using a Micromeritics ASAP 2400. Elemental analyses were performed by the 'Service Central d'Analyses du CNRS' at Vernaison, France: Si, Ti and Zr contents were determined by inductively coupled plasma (ICP) from aqueous solutions; C and H contents by high-temperature combustion and IR spectroscopy; chlorine content by potentiometric titration. The Si/M atomic ratios of the oxides were also determined by EDXA using an energy dispersive X-ray analyser Link AN 1000 fitted to an SEM Cambridge Stereoscan 360. The IR spectra of the solids in Nujol were recorded on a Perkin Elmer 1600 series FTIR spectrophotometer.The 29Si NMR spectra were collected on a Bruker ASX 200 spectrometer at 39.73 MHz, with a 7 mm MAS NMR probe (spinning fre- quency: 3.5 kHz), using proton decoupling, 46 pulses, 60 s recycle delay and adding 500- 1000 scans. Exponential multipli- cation using a line broadening of 100 Hz was performed prior to Fourier transformation. Results and Discussion Formation of the gels As mentioned in the introduction, the easy redistribution of the C1 and OPr' ligands around the silicon and the titanium 1666 J. Muter. Chem., 1996,6( lo), 1665-1671 or zirconium centres has to be kept in mind. This reaction takes place at room temperature during the mixing of the reactants.22 For example, after stirring a mixture of SiCl, and Ti(OPr'), for 3 h at room temperature, 29Si NMR shows the formation of SiCl,(OPr') at 6 -18.9, SiCl,(OPr'), at 6 -41.4, SiCl(OPr'), at 6 -74.3 and a small amount of Si(OPr'), at 6 -87.2.Owing to the occurrence of these redistribution reac- tions, Si-Cl and Si-OPr' as well as M-Cl and M-OPr' (M=Ti or Zr) functions are present. Thus, Si-0-Si, M-0-M and Si-0-M bridges (M=Ti or Zr) may result from the condensation between these functions; in this case, homogeneous samples (on the atomic scale) will be formed only if the rate of the heterocondensation reactions (formation of Si-0-M bridges) is comparable to the rate of the homo- condensations (formation of Si -0-Si or M -0-M bridges). The condensations take place during the heat treatment, leading to the formation of light brown, monolithic gels.The liquid expelled by the spontaneous shrinkage of the gels (syneresis liquid) was analysed using gas chromatography (GC) and 'H NMR spectroscopy. No isopropyl alcohol arising from hydrolysis was detected. The only byproduct detected was isopropyl chloride, indicating that the gelation arose only from non-hydrolytic condensations between M -C1 and M -OPr' groups. Elemental analysis of the Si/Zr and %/Ti xerogels indicated the presence of residual isopropoxide and chloride groups. The empirical formulae of the gels are given in Table 2. The oxygen content was calculated assuming that our samples contained no hydroxy groups or adsorbed water. The mass loss obtained by thermogravimetry in air agreed well with the theoretical mass loss calculated from the empirical formula, assuming formation of the oxide: This result indicates that the empirical formulae and the condensation degrees are reliable.The condensation degrees of the %/Ti xerogels were quite high, >85%; the condensation degrees appeared significantly lower for the Si/Zr xerogels, except for the sample with the lowest Zr content (1 mol%). The formation of well condensed, monolithic Si/Zr and Si/Ti gels suggests that the condensations around the silicon atoms and the zirconium or titanium atoms have comparable rates. From other studies, we know that Ti-Cl/TiO-Pr' and Zr -Cl/ZrO-Pr' condensations effectively takes place at 110 0C,15924 whereas Si-Cl/SiO-Pr' condensations are very slow in the absence of titanium or zirconium.Accordingly, fast precipitation of titania or zirconia should occur. However, Si -Cl/SiO -Pi condensations are efficiently catalysed by Lewis acids (FeCl,, AlCI,, TiCl,, et~.).,~,,~ZrC1, also catalyses these condensations, as shown by the rapid gelation of the sample 99Si/Zr. Thus, the fact that no macroscopic precipi- tation of titania or zirconia occurred and the formation of monolithic gels may be ascribed to a catalysis of the conden- sations around the Si atoms by the transition-metal species, leading to a levelling of the condensation rates around the silicon and the transition-metal atoms. Characterizationof the oxides obtained after calcination The mixed oxides obtained after calcination in air at 500 or 600°C of the dried gels were characterized by elemental analysis, energy dispersive X-ray analysis (EDXA) and X-ray diffraction (XRD). Considering that one of the potential appli- cations of mixed Si/Ti and Si/Zr oxides is heterogeneous catalysis, we have also measured the specific surface area of some samples.The presence of mixed Si-0-M bonds, which prove that mixing at the molecular level occurred during the gel formation, was sought using Fourier-transform IR (FTIR) spectroscopy and 29Si NMR spectroscopy. The crystallization behaviour of the mixed oxides upon heat-treatment in air was studied using XRD. Table 2 Empirical formulae of the gels after drying and mass loss during calcination condensation YO mass loss degree (YO) expl.' (calc.') 57 38 (42) 70 36 (36) 75 33 (37) 80 27 (34) 89 15 (24) 87 21 (23) 86 23 (27) 88 25 (24) 94 20 (17) sa mple empirical formula" 1 lS5Si/Zr 191 Si/ZrA i/ZrB 0Si/Zr 9Si/Zr SiITiA ~iZr0.94(sizr0.79~~izr0.2SiZrO. SiZr0.01(siTi0.92~ 0Pri)1.51C11.8102.22 0p~~1.26c10.8602.52 0~0P~~0.70c10.S1~1.80 )0.56c10.3 lo 1.77 0P~)0.29C10.1601.80 0P2~0.S2c10.4503.36 lS i/TiB ~iTi0.86( 0P2)0.76C10.2903.20 4Si/Ti 2~~~ ~ 0Si/Ti ~ SiTi0.25(0SiTiO.06 Pf)0.40C10. (Opt 10.24c10. 18O2.21 03O 1.99 "From elemental analysis. 'Experimental mass loss, from thermogravimetry, 20-1000 "C in air at + 10 deg. min ~ '. 'Theoretical mass loss derived from the empirical formula, assuming complete oxidation.Composition of the oxides and adsorption-desorption data. According to elemental analysis, the oxides obtained after calcination at 500 or 600°C contained less than 0.3 mass% carbon and chlorine. The metal contents of the glasses (deter- mined either by EDXA or by elemental analysis) were close to the expected values (Table 3). EDXA was also used to verify that no macroscopic phase separation took place during the synthesis of the oxides. For this purpose, 10-point measure- ments were performed, indicating practically constant Si/M ratios in the samples, which is consistent with homogeneity at the micrometre level. The surface areas of our oxides were relatively high, varying between 300 and 1000 m2 g-' (Table 3). All the samples were mesoporous, except the 20Si/Ti sample, which exhibited a type I1 isotherm without desorption hysteresis, indicative of a dominating contribution of microporosity.Microporosity for %/Ti oxides with a low Ti content has been reported by several a~thors*~,~~ Powder X-ray diffraction. All the Si/Zr oxides obtained by calcination at 600°C for 5 h were amorphous. On the other hand, anatase was detected in all the Si/Ti samples calcined for 5 h at 500"C, except the sample with the lowest concen- tration of titanium (20Si/Ti) which was amorphous. FTIR spectroscopy. The FTIR spectra of the Si/Zr oxides are reported in Fig. 1. The spectrum obtained for the 99Si/Zr sample showed the typical absorptions of silica at ca. 1080, 1220 cm-' (Si-0-Si asymmetric stretching vibrations) and 800 cm-' (network Si-0-Si symmetric bond stretching vibration).The presence of hydroxy groups led to a shoulder at 970cm-' (Si-OH bond stretching) and broad bands between 3000 and 3700 cm-' arising from isolated and hydro- gen-bonded SiO-H stretching vibrations and hydrogen-bonded water. When the zirconium content increased, the absorption at 800 cm-' decreased and the shoulder at Table 3 Characterization of the Si/Zr and %/Ti oxides: metal contents of the glasses [from elemental analysis (EA) and EDXA results] and specific surface areas (from BET analysis of N2 adsorptionaesorp- tion data) sample mol% M (nominal) mol% M (EA) mol% M (EDXA) surface area/ m2 g-l 1 Si/ZrA" 50 50 53 334 1 SilZrB' 50 44 48 359 5 Si/Z ra 10Si/Zr" 16.7 8.3 17 10 17 10 651 - 99Si/Zr" 1 0.9 1 - 1 Si/TiA' 1 Si/TiB' 50 50 46 48 47 47 640 528 4Si/Tic 20Si/Ti' 20 4.8 19 6 20 - 790 1000 "Calcined at 600°C for 5 'Calcined at 500°C for 5 h.h; 'Calcined at 600"C, n o holding time. J 8c (de 8n (d 5siIzrs"" A"\,A I I I I I I 4000 3000 2000 1500 1000 500 wavenumberkm -l Fig. 1 IR spectra of the Si/Zr samples calcined for 5 h at 500°C. Samples diluted in a Nujol emulsion, bands due to the Nujol are indicated by asterisks. 970cm-' increased. The decrease of the absorption at 800 cm-' indicates that the silicate network is broken down by the introduction of Zr4+ ions.' In the case of Si0,-ZrO, gels and oxides, the band at 970cm-' has been related to vibrational modes involving Si-0-Zr linkage^.^^,^' This band might also arise from the stretching of Si-OH groups, as in silica.However, the intensity of the broad absorption at ca. 3400 cm-I (SiO-H stretching vibrations) does not signifi- cantly increase with the zirconium content. Therefore, the increase in the intensity of the band at 970 cm-' should arise mainly from the formation of Si-0-Zr linkages. The FTIR spectra of the Si/Ti oxides are reported in Fig. 2. Besides the typical absorptions of silica, a strong band was found at 950 cm-' for all the samples calcined at 500 "C for 5 h. According to IR31-33 and Raman34 spectroscopic studies on mixed Si0,-TiO, gels and oxides, this absorption at about 950 cm-' is associated with a vibrational mode involving Si04 tetrahedra bonded to a titanium in four-fold coordination.In J. Mater. Chem., 1996,6(lo), 1665-1671 1667 I I I I 4000 3000 2000 1500 1000 500 wavenumberkm-l Fig. 2 Comparison between the IR spectra of the 99Si/Zr sample and the &/Ti samples calcined for 5 h at 500°C Samples diluted in a NUJO~ are indicated by asterisks emulsion bands due to the NUJO~ our samples, the intensity ratio of the band at 950cm ' to that at 1080 cm-' did not significantly vary with the Ti content of the oxide (5-50 mol% Ti), which suggests that the number of Si-0-Ti bonds is limited in &/Ti mixed oxides A similar behaviour has already been observed for mixed oxides prepared by hydrolysis of alkoxides thus, Best and C~ndrate~~ reported that the number of Si-0-Ti bonds reached a maximum between 6 3 and 11 2 mol% T10, These observations are in good agreement with the fact that stable S102-T102 glasses in which tetrahedrally coordinated Ti atoms replace Si atoms at random exist up to ca 8 5 mol% TiO, only 35 29Si MAS NMR spectroscopy.The Si/Zr and %/Ti oxides calcined for 5 h at 500 or 600 "C in air have been charactenzed by 29S1 MAS NMR spectroscopy In all cases, the spectra displayed single, broad peaks, as expected for these highly disordered materials The corresponding chemical shifts and the widths at half-height (FWHH) are reported in Table 4 The 29S1 spectra of the 99Si/Zr, lSi/ZrA and lSi/TiA samples, Table4 NMR data for the mixed oxide samples obtained by calcination in air at 500 "C for 5 h (heating rate 5 "C min ') mol% M starting gel (nominal) 6 FWHH (PPd 1Si/ZrA 50 -98 5 23 4 1Si/ZrB 50 -101 5 23 1 5Si/Zra 17 -106 1 16 3 10Si/Zr" 10 -106 1 16 2 99Si/Zr 1 -107 4 17 6 1Si/TiA 50 -107 6 17 0 1 Si/TiC 50 -107 4 17 3 4Si/Tl 20 -107 9 16 7 "Calcined for 5 h at 600°C 1668 J Muter Chem , 1996,6(10), 1665-1671 which are representative of all the spectra obtained, are com- pared in Fig 3 The 99Si/Zr sample contains only about 1 mol% Zr, and its spectrum is typical of the spectra obtained for silica gels calcined under the same conditions, with a peak (FWHH= 17 5 ppm) centred at 6 ca -107 5 The incorpor- ation of large amounts of zirconium in the silica network leads to a significant low-field shift of the peak maximum and to an increase of the linewidth Thus, the spectra of the lSi/ZrA and lSi/ZrB oxides, which contain about 50 mol% Zr, display much broader peaks (FWHH =23 ppm) that are centred at 6 -98 5 and -101 5, respectively On the other hand, after heat treatment at 800 "C, leading to the crystallization of tetragonal zirconia, the peak maximum shifted to 6 -107 and the FWHH decreased to 17 ppm In the Si/Ti samples, the chemical shifts and the linewidth appear independent of the Ti content, and they are close to the values observed for the 99Si/Zr sample and for calcined silica The sensitivity of 29S1 NMR spectroscopy to the nature of the second-nearest neighbours is well documented Thus, the 29S1 chemical shifts of natural aluminosilicates and zeolites depend mainly on the number of A1 atoms in the second coordination sphere of silicon atoms The five possible Si(OSi),(OAI), ,tetrahedra (Q4 ,,At)have characteristic 29S1 shift ranges, each additional aluminium substitution leading to a low-field shift of ca 5-7 ppm 36 Unfortunately, there are much less unambiguous data concerning the effect of the substitution of silicon atoms by titanium or zirconium atoms In the case of zirconium, the 29S1 chemical shift reported for Q4Z= units in zircon is very close to those reported for Q4 units in crystalline aluminosilicates, andalusite and kyanite (Table 5) In the case of titanium, no crystalline binary oxide with silicon exists, however, a low-field shift of 6-8 ppm per titanium substituent may be derived from the chemical shifts of the Q2Ba units in fresnoite3' and of the QIBa units in benitoite (assuming a 10 ppm shift per barium substituent) In addition, the 29S1 NMR spectrum of the ETS-10 titanosilicalite shows a 6 2-9 2 low-field shift between Q4and QlTI3tetrahedral sites 38 Thus, the effect of zirconium or titanium second-nearest neighbours on the 29S1 chemical shift is most probably similar to the effect of aluminium However, it is not straightforward to identify Si-0-Ti or I.I.I.I.1 -60 -80 -100 -120 -140 s Fig 3 29S1MAS NMR spectra of the 99Si/Zr (full line) 1Si/Zr (dashed line) and 1Si/Ti (dotted line) samples calcined for 5 h at 500 "C Table 5 29Si NMR chemical shifts in some silicates and silica gels (from ref.35) tetrahedral site 6 compound -78.8 andalusite (A12Si05) -82.9 kyanite (A12Si05) -81.6 zircon (ZrSiO,) -94.2 benitoite (BaTiSi,O,) -82.0 fresnoite (Ba,TiSi,O,)" -70.3 barium silicate (Ba,SiO,) -109.3 -99.8 silica gels -90.6 'From ref. 36. Si-0-Zr bonds in mixed gels and glasses using ,'Si NMR spectroscopy. Indeed, these samples usually contain numerous surface hydroxy groups bonded to silicon atoms. The low-field shift caused by these hydroxy groups on the ,'Si chemical shift, about 9-10 ppm per OH group, is significantly larger than the effect of aluminium (Table 5). The same silicon atom may be bonded simultaneously to OM and OH groups, in QxOH,(4-n-x)Mnsites.The total number of such sites is 15. In addition, variations in the Si-0-Si angles also lead to a variation in the chemical shift. In the amorphous samples we are dealing with, the different Q signals are not resolved, leading to a single, broad peak. As the chemical shifts and linewidths of the different Q species are not known precisely, the deconvolution of such peaks is not reliable, and the number of Si-0-M bonds cannot be determined. Nevertheless, in SO,-MO, oxides, an increasing number of Si-0-M bonds should lead to an increase of the intensity of QlM3, QlM2, Q2M2,etc. resonances at the expense of the Q" resonance, which will lead to a low-field shift of the peak maximum and to a significant increase of the linewidth.This is what we observe in the lSi/Zr oxides. There is no doubt that these samples are highly homogeneous and contain a large number of Si-0-Zr bonds. Similar spectra were recently reported for SO,-ZrO, calcined gel particles obtained by hydrolysis of aerosols of tetraethoxysilane and zirconium n-propoxide.'" In the same way, a peak centred at 6 -102 with a half-width of 13ppm was reported recently for an Si0,-ZrO, oxide prepared by another non-hydrolytic sol-gel process.39 In the Si/Zr oxides with lower Zr contents (10 and 17 mol% ZrO,), the presence of Si-0-Zr bonds cannot be ascertained, showing the limitations of 29Si NMR spectroscopy in the detection of Si-0-M bonds in mixed oxides.In these cases, "0 NMR spectroscopy would be certainly more sensi- tive4', but it would imply the use of 170-enriched samples, and thus the synthesis of I70-enriched alkoxides or diisopropyl ether, which is not straightforward. In the case of our Si/Ti oxides, Si-0-Ti bonds are present according to the FTIR spectra and the XRD results. However, the 29Si spectra showed no noticeable influence of the Ti content on the chemical shift and linewidth, whatever the Ti content of the glass. This shows that the amount of Si-0-Ti Crystallization behaviour of the oxides The powder XRD results concerning the mixed oxides calcined at different temperatures are summarized in Tables 6 and 7. As mentioned above, all the Si/Zr oxides obtained by calcination at 600 "C for 5 h were amorphous. Crystallization of tetragonal zirconia took place between 600 and 13OO"C, depending on the Zr02 content of the samples; however, the 99Si/Zr sample remained amorphous after heating for 2 h at 1300"C.The distinction between tetragonal and cubic zirconia polymorphs from the XRD spectra only is not straightforward owing to the broadness of the lines.On the other hand, the Raman spectra are readily di~tinguishable.~' In our case, the tetragonal phase was identified clearly by Raman spectroscopy. The temperatures of crystallization of t-ZrO, reported for Si0,-ZrO, gels prepared by hydrolytic sol-gel processes vary between about 450 and 900 0C,10714942p45whereas pure ZrO, crystallizes to t-ZrO, at 330 0C.46 A strong retarding effect has often been ascribed to a good chemical homogeneity of the starting gels, i.e.to a high degree of Si-0-Zr bonding.14 If this criterion was used, the homogeneity of non-hydrolytic Si/Zr gels would be the average of that of hydrolytic gels. The transformation of tetragonal to monoclinic zirconia was retarded strongly in our samples and did not take place after 2 h at 1300 "C,instead of 600-700 "C in pure zirc~nia~~?"~ or 1000-1300 "C in Si0,-ZrO, gels prepared by hydrolytic sol-gel proces~es.~~,~~*~~-"~This retarding effect has been ascribed to blocking by the low-expansivity silica matrix of the tetragonal to monoclinic transformation which involves a positive volume change,45 and the hindering of particle growth by silica, which maintains the size of zirconia particles below the critical size (ca.30 nrn),,, where they transform to the stable monoclinic f~rm.'~,~~ The 1Si/Zr samples were heated at higher temperature to study the formation of zircon, ZrSiO, (Fig. 4). The first traces of ZrSiO, are detected after 2 h at 1500"C; the samples were completely crystallized to zircon after 20 h at 1500 "C. According to the literature, the conversion of Si/Zr gels (without additives) to zircon requires rather high temperatures, between about 1200 and 1600 0C.11,14,43,44*47348 It is noteworthy that in all cases the crystallization of zircon takes place after the crystallization of zirconia. Thus, a high chemical homogeneity of the starting gel is not necessarily favourable for the crystallization of zircon at low temperature.Indeed, Vilmin et al. reported that the formation of zircon in nanoheterogeneous gels (obtained from silica and zirconia sols) I m bonds is too low to be detected by ,'Si NMR spectro~copy.~~ According to Evans, stable %/Ti glasses exist up to ca. 8.5 mol% TiO, only.35 In these glasses, tetrahedrally coordinated Ti atoms replace Si atoms at random. In such cases, the percentage of Si-0-Ti bonds would not exceed 8.5% what- ever the Ti content; their detection by ,'Si NMR spectroscopy would be hindered by the presence of hydroxy groups. This interpretation is consistent with the FTIR spectra, which show a nearly constant intensity of the 950 cm- band whatever the TiO, content of the samples.Here too, 170NMR spectroscopy would be more appropriate for the study of these glasses.37 5 10 15 20 25 30 35 28ldegrees Fig.4 X-Ray diffraction patterns of the lSi/ZrA sample calcined at different temperatures. The most intense reflection of each phase is indicated as follows: m, monoclinic zirconia; t, tetragonal zirconia; z, zircon; c, cristobalite. J. Muter. Chem., 1996, 6(lo), 1665-1671 1669 Table 6 Crystalline phases formed in Si0,-Zr02 samples heat treated at several temperatures (from X-ray powder diffraction) sample Zr mol% 6OO"C, 5 h 1Si/ZrA 50 Am 1Si/ZrB 50 Am 5Si/Zr 17 Am 10Si/Zr 8 Am 99Si/Zr 1 Am "Am, amorphous, t-ZrO,, tetragonal zirconia Table 7 Crystalline phases formed in S1O2-TiO2 samples heat-treated at several temperatures (from X-ray powder diffraction)" heat treatment ~~~~~~~~ ~ sample mol%Ti 500 "C, 5 h 900 "C,2 h 1300"C, 2 h 1Si/TiA 50 Am +An An An 1Si/TiB 50 Am+An An An 4Si/Ti 20 Am+An AnfCr An+Cr 20Si/Ti 5 Am Cr Cr aAm, amorphous, An, anatase, Cr, cnstobalite occurred at a lower temperature than in homogeneous, mono-phasic gels 48 In the case of the Si/Ti oxides, only the 20Si/Ti sample remained amorphous after calcination for 5 h at 500°C (Table 7) It is noteworthy that this sample, containing ca 5 mol% TiO,, is within the 'stable' glass region, which exists up to about 8 5 mol% TiO, according to Evans,35 whereas the other samples are outside this stable glass region Interestingly, the 20Si/Ti sample crystallized as single phase cristobalite at 900°C (note that the 99Si/Zr sample was still amorphous at 1300"C), neither anatase nor rutile was detected, even after annealing for 1h at 1540"C (Fig 5) In addition, the positions of the diffraction lines of cristobalite were modified slightly for instance, the peaks corresponding to the [1131 and [212] reflections were observed at 28 =46 70 and 48 28", instead of 47 05 and 48 63" in pure SiO, cnstobalite The same behaviour was observed for ultra-low expansion (ULE) glasses prepared by flame hydrolysis of chlorides,35 it was ascribed to the formation of a solid solution of T10, in SiO, the substitution of Sl4+ ions by larger Ti4+ ions leads to a linear increase of the tetragonal unit-cell constants The above 28 yalues corre-spond to cell parameters a= 5 00 A and c= 6 98 A, which are in excellent agreement with the values reported by Evans for h l Ih v)c.c V' .A ..M'Ls 1300 "C, 2 h V' > c v) C al + E 900"C,2h 600°C, 5 h 5 10 15 20 25 30 35 40 2Wdegrees Fig.5 X-Ray diffraction patterns of the 20Si/Ti sample calcined at different temperatures 1670 J Muter Chem, 1996, 6(10), 1665-1671 heat treatment 800"C, 5 h lOOO"C, 2 h 13OO0C,2 h t-ZrO, t-ZrO, t-ZrO, - - - t-ZrO, t-ZrO, t ZrO, Am t-ZrO, t-ZrO, Am Am Am an oxide containing 5 mol% Ti35 Accordingly, all the Ti atoms in the 20Si/Ti sample are incorporated in the silica network, suggesting that the starting gel was perfectly homogeneous In the 4Si/Ti sample, which is outside the stable glass region, both anatase and cristobalite crystallized on heating This behaviour shows that part of the T10, formed a solid solution with SiO,, and the remainder crystallized as anatase A similar behaviour was reported for S1O,-T1O2 glasses containing 10 and 30 mol% TiO,, prepared by hydrolysis of alkoxides 49 In the other samples, anatase remained the only crystalline phase detected up to 1300 "C The crystallization of anatase in Si/Ti oxides prepared by hydrolytic sol-gel processes occurs between about 500 and lOOO"C, depending on the conditions of synthesis and the titania content 24 Accordingly, the crystallization temperature of anatase in our samples other than 20Si/Ti is rather low This suggests that the starting gels are inhomogeneous, which is consistent with the small number of Si-0-Ti bonds detected by FTIR spectroscopy, and the 29S1NMR spectra Although anatase crystallized at a rather low temperature in our &/Ti samples, the transformation of anatase to rutile was not observed, even after heating for 2 h at 1300°C The transformation of anatase to rutile is usually detrimental to the applications of T10, oxides as catalysts, owing to the related decrease of surface area In pure titania gels, this transformation takes place between 600 and 1000"C 46 The same suppressive effect on the transformation of anatase to rutile has been reported for mixed SiO,-TiO, oxides prepared by a hydrolytic sol-gel process,so provided that the gels are relatively homogeneous Conversely, the formation of rutile at temperatures as low as 600 "C is reported for samples prepared by physical mixing of silica and titania gels5' or for alkoxide gels prepared in excess water 51 Conclusions The non-hydrolytic sol-gel route described here, involving chlorides and isopropoxides (or diisopropyl ether), applies well to the preparation of Si/Zr and Si/Ti mixed oxides Monolithic Si/Zr and &/Ti gels are readily obtained in one step, without the use of additives The Si/M ratio of the gel and of the oxide may be controlled easily by the composition of the starting mixture The surface areas of the oxides obtained by calcination of the gels are quite high In both Si/Zr and &/Ti oxides, IR and/or 29S1 NMR spectroscopies showed the presence of Si-0-M bonds, indi-cating some degree of homogeneity on the atomic scale However, the number of such bonds remained relatively low in Si/Ti mixed oxides, compared to Si/Zr oxides, which agrees well with the behaviour reported recently for mixed oxides prepared by a hydrolytic sol-gel process 40 The Si/Zr oxides (up to 50 mol% Zr) remained amorphous after 5 h at 600°C The transformation of tetragonal to mono-clinic zirconia was strongly retarded in our samples and did not take place after 2 h at 1300"C The crystallization of zircon (for a sample containing 50 mol% Zr) started at 1500°C and was completed after 20 h at 1500°C The precipitation of anatase was observed after 5 h at 500°C for the Si/Ti oxides with high Ti contents (20-50 mol% Ti), which are outside the stable glass region On the other hand, the transformation of anatase to rutile was not observed even after 2 h at 1300°C 23 24 25 26 D C Bradley and D A W Hill, J Chem SOC, 1963,2101 P Arnal, These, Universite de Montpellier 2, 1995 K A Andnanov,T N GaninaandN N Sokolov J Gen Chem USSR, 1956,26,1897 W Noll, Chemistry and Technology of Szlzcones, 2nd edn ,Wiley, New York, 1951 The sample within the stable glass region (<8 5 mol% Ti) 27 D C M Dutoit, M Schneider and A Balker, J Catal 1995, crystallized at 900 "C as single phase cristobalite oxide, with Ti4+ ions substituting Si4+ ions at random, which suggests that a highly homogeneous gel was formed in this case 28 29 153,165 B E Hardy, M Maciejewski, A Balker and A Wokaun, J Muter Chem ,1992,6,673 S W Lee and R A Condrate, Sr ,J Muter Sci ,1988,23,2951 30 J A Navio, M Macias, G Colon, P J Sanchez-Soto, References 31 V Augugliaro and L Palmisano, Appl Surf Sci , 1994,81,325 C F Smith, Jr, RA Condrate, Sr and W E Votava, Appl 1 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 C J Brinker and G W Scherer, SoI-Gel Science The Physics and Chemistry of Sol-Gel Processing, Academic Press, San Diego, 1990 H Hosaka and K Meguro, Bull Chem SOC Jpn, 1971,44,1252 S Sakka and K Kamiya, J Non-Cryst Solids, 1980,42,403 B E Yoldas, J Non-Cryst Solids, 1980,38&39,81 Y Abe, N Sugimoto, Y Nagao and T Misono, J Non-Cryst Solids, 1988, 104, 164 M Emih, L Incoccia, S Mobilio, G Fagherazzi and M Guglielmi, J Non-Cryst Solids, 1985,74, 129 I M Miranda Salvado and J M Fernandez Navarro, J Non-Cryst Solids, 1992,1471148,256 M Nogami, J Non-Cryst Solids, 1985,69,415 M Nogami, J Non-Cryst Solzds, 1994,178,320 I M Miranda Salvado, C J Serna and J M Fernandez Navarro, J Non-Cryst Solids, 1988,100, 330 Y Kanno, J Muter Sci ,1989,24,2415 T Mory, H Yamamura, H Kobayashi and T Mitamura, J Am Ceram SOC,1992,75,2420 R Oheim, H Paulus and C Russel, J Muter Scz Lett, 1991, 10,1171 P Tartaj, J Sanz, C J Serna and M Ocana, J Muter Sci, 1994, 29,6533 R J P Corriu, D Leclercq, P Lefevre, P H Mutin and A Vioux, J Muter Chem ,1992,6,673 R J P Corriu, D Leclercq, P Lefevre, P H Mutin and A Vioux, J Non-Cryst Solids, 1992,146,301 R J P Cornu, D Leclercq, P Lefevre, P H Mutin and A Vioux, Chem Muter, 1992,4,961 S Acosta, R J P Corriu, D Leclercq, P Lefevre, P H Mutin and A Vioux, J Non-Cryst Solids, 1994,170,234 S Acosta, R J P Cornu, D Leclercq, P Lefevre, P H Mutin and A Vioux, J Sol-Gel Sci Technol ,1994,2,25 K Moedntzer, in Organometallzc Reactzons, ed EI Becker and M Tsutsui, Wiley Interscience, 1971, vol 2, pp 1-116 H Weingarten and J R Van Wazer, J Am Chem SOC, 1965, 87,724 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 Spectrosc ,1975,29,79 Z Dend, E Breval and C G Pantano, J Non-Cryst Solids, 1988, 100,364 A Kasgoz, K Yoshimura, T Misono and Y Abe, J Sol-Gel Sci Technol, 1994,1, 185 M F Best and R A Condrate, Sr ,J Muter Scz Lett, 1985,4,994 D L Evans, J Non-Cryst Solids, 1982,52, 11 5 G Engelhardt and D Michel, High-Resolution Solzd-State NMR of Silicates and Zeolites, Wiley, Chichester, 1987 P J Dirken, M E Smith and H J Whitfield, J Phys Chem, 1995, 99,395 M W Anderson, 0 Terasaki, T Oshuna, A Philippou, S P MacKay, A Ferreira, J Rocha and S Lidin, Nature (London), 1994,367,347 M Jansen and E Guenther, Chem Muter, 1995,7,2110 P J Dirken, R Dupree and M E Smith, J Muter Chem, 1995, 5,1261 C M Phillippi and K S Mazdiyasni, J Am Ceram SOC, 1971, 54,254 S K Saha and P Pramanik, J Non-Cryst Solids, 1993,159,37 J Campaniello, E M Rabinovich, P Berthet, A Revcolevschi and N A Kopylov, in Better Ceramics through Chemistry IV Muter Res SOC Symp Proc 1990, vol 180, p 541 G Monros, M C Marti, J Carda, M A Tena, P Escribano and M Anglada, J Muter Sci ,1993,28, 5852 V S Nagarajan and K J Rao, J Muter Sci ,1989,24,2140 M Ocaiia, V Fornes and C J Serna, Ceram Inter, 1992,18,99 A B Hardy and W E Rhine, Chemical Processing of Advanced Materials, ed L L Hench and J K West, Wiley, 1992, p 577 G Vilmin, S Komarneni and R Roy, J Muter Sci , 1987,22,3556 I M Miranda Salvado and J M Fernandez Navarro, J Non-Cryst Solids, 1992,1471148,256 H Nakabayashi, K Nishiwaki and A Ueno, Muter Res Bull, 1988,23,555 B E Handy, M Maciejewski, A Balker and A Wokaun, J Muter Chem , 1992,2,833 22 D C Bradley, D C Hancock and H Wardlaw, J Chem SOC, 1952,2773 Paper 6/02404A, Received 9th April, 1996 J Muter Chem, 1996, 6(10), 1665-1671 1671

ISSN:0959-9428    

DOI:10.1039/JM9960601665

出版商:RSC    

年代:1996

数据来源: RSC

摘要:

29SiMAS NMR investigation of the pyrolysis process of cross-linked polysiloxanes prepared from polymethylhydrosiloxane Rafik Kalfat," Florence Babonneau,b NCji Gharbi*" and HCdi Zarrouk" "Laboratoire Chimie des Matiriaux, Dipartement de Chimie, Faculti des Sciences, 1060 Tunis, Tunisia bChimie de la Mutiire Condensie, URA CNRS 1466, Uniuersiti Pierre et Marie Curie, 4 Place Jussieu, 75252 Puris, France Polymethylhydrosiloxanes (PMHSs) crosslinked with hexa-1,5-diene or hexane- 1,6-diol have been considered as potential precursors for Si- C- 0 systems. Their pyrolysis under an argon flow has been investigated mainly by 29Si MAS NMR spectroscopy. The two systems, with similar C/Si but different O/Si ratios, show different pyrolytic behaviour. This is essentially due to different graftings between the siloxane backbone and the alkyl chains, Si-C us.Si-0-C bonds. In both cases, degradation of the organic chains occurred at 600 "C; but in the case of PMHS cross-linked with hexanediol (D06), all the C groups are lost, while for PMHS cross-linked with hexadiene (DE6), the C groups bonded to Si remain. As a consequence, the oxycarbide phase obtained at 1000"C is much richer in C for DE6, compared to D06, but the amount of free C is also higher. These differences in composition strongly influence the nature of the samples obtained at 1500"C: crystalline silicon carbide for DE6 and mixture of amorphous silica and silicon carbide for D06. Silicon oxycarbide glasses have received increasing attention in the last ten years due to their high-temperature stability, good crystallization and oxidation resistance, and high mechanical strength.Ip4 Introduction of carbon into a silica network by direct solid-state reactions5 is very difficult, but it is quite easy by pyrolysing appropriate polysil~xanes.~ The use of polymers has also allowed the development of shaped objects such as silicon oxycarbide fibres7p9 and composites.lO," Several synthetic strategies have been followed to prepare the starting polymers: hydrolysis of chloro- or alkoxy-sil-anes,12-18 catalytic redistribution of oligo- and poly-sil~xanes,~~ and modification of polycarbosilanes.20 They have generated a variety of polysiloxane architectures containing Si- R groups (R=CH,, C6H5, CH2=CH, etc.), and also Si-H21-23 and Si- Si Commercial silicone resins have also been used.26 Pyrolysis of these polymeric precursors around 900-1000°C leads to a silicon oxycarbide network based on mixed SiC,04 -,units with 0 <x<4, as demonstrated clearly in most of the studies by 29Si MAS NMR spectroscopy. Usually a free carbon phase is also present, from the decompo- sition of the organic moieties.The composition of the oxycar- bide phase as well as the free carbon content is clearly dependent on the nature of the R gro~p,~,~',~~ and on the O/Si and C/Si molar ratios21,29,30 in the starting precursors. The O/Si molar ratios mainly dictate the composition of the oxycarbide phase, and the C/Si ratio determines the amount of free carbon.Pyrolysis at T> 1500"C will lead to a possible transformation of the silicon oxycarbide phase into silicon carbide, through carbothermal reactions depending on the amount of free carbon. This route to Sic has been explored by several All these studies show that it is possible to tailor the architecture of the starting polysiloxane to influence its poly- mer-to-ceramic transformation and thus the structure of the final materials. Following this idea, polymethylhydrosiloxane (PMHS), has been cross-linked via hydrosilylation reactions with bifunctional hydrocarbon chains of various lengths, to produce polysiloxanes with various C and 0 content^.^' PMHSs reacted with hexa-1,Sdiene and hexane-1,6-diol have been pyrolysed to 1500°C: X-ray diffraction results show a crystalline silicon carbide phase in the case of the diene, and a mixture of amorphous silica and silicon carbide in the case of the di~l.,~ The objective of this paper is to present a MAS NMR spectroscopic characterization of the pyrolysed products obtained from the two different cross-linked PMHSs in order to understand how they transform with temperature into two different inorganic structures.Experimental Samples were prepared by mixing precursors in stoichiometric proportions as shown in Table 1. Hexachloroplatinic acid was used as a catalyst [4 x mol (g PMHS)-l, in THF]. The pyrolyses were performed under an argon flow in a tubular furnace with a heating rate of 5°C min-' and a holding time of 5 h at the maximum temperature. Solid-state NMR studies were carried out on an MSL 400 Bruker spectrometer using the MAS technique (spinning rate: 4 kHz).29Si MAS NMR spectra were obtained at 79.5 MHz with a 50 kHz spectral width, a pulse width of 2 ps (0 30") and a relaxation delay of 60 s. Short flip angles along with relatively long recycle delays were used to try to overcome the problem of long 29Si relaxation times. The 13C CP MAS NMR experiments (100.62 MHz) were recorded with a contact time of 1 ms, a recycle delay of 6 s and a spectral width of 50 kHz. Tetramethylsilane was used as the external reference for all MAS NMR spectra. Peaks were labelled using the X, M,, D, and T, notation. X, M, D, T and Q refer to SiC,-,O, units with x=O, 1, 2, 3 and 4, respectively, where n is the number of bridging 0 atoms between two neighbouring Si.The simu- lations of the spectra were produced with the WIN-FIT program.37 Thermogravimetry (TG) studies were carried out with a Netzsch STA 409 instrument under an argon flow with a heating rate of 10°C min-l. The chemical analysis was performed by the 'Service Central d'Analyses du CNRS', Vernaison, France, for Si, C and H. Oxygen was estimated by difference. Results Characterizationof the precursors DE6 and DO6 samples were obtained by coupling PMHS with hexa- 1,5-diene and hexane- 1,6-diol, respectively, through hydrosilylation reactions. The schematic structures of these J. Muter. Chem., 1996, 6(lo), 1673-1678 1673 Table 1 Experimental conditions and nature of obtained products molar ratio reaction reagent (reagent)/(Si-H) time/min hexa- 1,5-diene 05 15 hexane- 1,6-d1ol 05 90 compounds are shown in Fig 1 The structural characterization of such samples has already been published36 and the NMR results will be summarized here for clarity The 29S1 MAS NMR spectrum of DE6 [Fig 2(u)] shows a main peak at 6 -22 representing 290% of the total Si sites and due to D, clearly indicating as expected the replacement of Si-H bonds by Si-C bonds The peak at 6 6 6 (cu 5%) is due to M endgroups of the PMHS chains Two other small peaks are present at 6 -40 due to residual D2H sites from PMHS,39 and at 6 -57 due to T, sites, MeSi(O,,),(OH), related to hydrolysis of the DZHsites The 13C CP MAS NMR spectrum of DE6 [Fig 3(u)] is dominated by four peaks due to CH3 -Si (6 0), C, ,, (6 18), C, ,, (6 23) and C, a, (6 33) sites Four peaks corresponding to C sp2 sites are present at 6 115, 125, 131 and 139, suggesting the existence of two types of residual vinylic groups those due to hexadiene that have reacted only on one side (6 115 and 139) and those resulting from the formation of unreactive double bonds in an isomerisation process promoted by the Pt SI SI I I0 0 SI SIDE6 SI I ? 0 0 I I CH3-St-O-CH2-CH2-CH2-CH2-CH2-CH2-0 -SI-CH~ I a p a a'p'a' I0 0 I I SI SIDO6 Fig.1 Schematic structures for DE6 and DO6 samples Fig. 2 "S1 MAS NMR spectra of DE6 (a) and DO6 (b) 1674 J Muter Chem, 1996,6( lo), 1673-1678 nature of final product dried gels at 100°C transparent gel white powder, DE6 transparent gel white powder, DO6 -1 1 150 100 50 0 -50 b Fig.3 13C CP MAS NMR spectra of DE6 (a) and after pyrolysis at 600°C (b) catalyst 40 41 All these characterizations show that hydrosilyl- ation occurred to a large extent The 29S1 spectrum of DO6 [Fig 2(b)] shows the presence of a main peak at 6 -58 corresponding to T, sites4, resulting from the reaction of alcohol functions with PMHS The small peak at 68 is due to end M sites No peak due to sites containing Si-H bonds are present It should be noted that, unlike DE6, the IR spectrum of DO6 shows the presence of OH groups36 Some of the T, sites could thus contain Si-OH groups Solid-state 13C CP MAS NMR of DO6 [Fig 4(u)] shows one peak at 6 -3 6 assigned to the carbon atoms of methyl groups in T units belonging to the siloxane chain" and three peaks at 626, 33 and 62 due to Caa,, C,,, and C,,, carbon atoms of the aliphatic chains, respectively 43 The chemical shift values of C, ,, cannot differentiate assignment to HO-C, ,, or Si-OC, ,, sites Pyrolysis of DE6 samples TG analyses were performed up to 1400°C under an argon flow to study the thermal behaviour of the samples (Fig 5) DE6 is thermally resistant up to 300°C Further heating to 600°C causes a major mass loss (44%) Indeed, elemental analysis (Table 2) shows that at 600 "C the C/Si ratio has decreased from 4 to 2 This strongly suggests a degradation of the alkyl chains with just two C from the alkyl chains remaining in the pyrolysed sample It can be anticipated that they are the C,,, groups, and this will be confirmed by the following NMR study The other C groups, which have been transformed into volatile species, are responsible for the large mass loss The small mass loss observed between 620 and 800 "C (5%) should be due essentially to a loss of hydrogen the elemental analysis shows a sharp decrease in the H/Si ratio from 4 5 to 0 5 between 600 and 1000°C Fig.4 13C CP MAS NMR spectra of DO6 (a) and after pyrolysis at 600°C (b) -50 (b) -70 0 200 400 600 800 *-I--LLUX(-.XI-CLL1000 1200 1400 TI'C Fig. 5 TG traces of DE6 (a) and DO6 (b) Table 2 Chemical analyses (mass%) of pyrolysed DE6 at different temperatures molar T/"C Si (YO) C (YO) H (YO) 0 (YO.)" composition 25 26.49 48.91 8.88 15.72 SiC4.32H9.4001.04 600 36.51 30.90 5.89 26.7 SiC1.98H4.5301.28 1000 40.75 26.48 0.80 31.97 SiCl,52Ho.5501,38 1200 39.26 26.50 <0.20 34.24 SiCl.5701.53 1400 40.15 25.56 <0.20 34.29 SiC1.4801.51 1500 67.76 27.76 <0.20 4.28 SiCo.9600.11 "Oxygen content determined by difference, 0 (YO)=100-[Si (%)+C (Yo)+H (Yo)].The 29Si MAS NMR spectra were recorded on the samples pyrolysed up to 1500°C [Fig. 6(u)]. The 13C MAS NMR spectra were recorded using the cross-polarization technique only on the 600 "C sample, which contains appreciable amounts of protonated C sites (Fig. 3). The I3C MAS NMR spectrum of the 600°C sample [Fig.3(b)] is quite different to that of the starting DE6 sample [Fig. 3(u)]. In the aliphatic region, the spectrum is now domi- nated by one broader peak at 6 1 due to CH, groups bonded to Si. The peaks due the C atoms of the alkyl chains have disappeared in agreement with the elemental analysis and TG 1500 "C 1400°C I200 "C 600°C .-n MK) "C J1, Fig. 6 29SiMAS NMR spectra of DE6 (a) and DO6 (b) pyrolysed at different temperatures analysis, which suggests decomposition of these chains. The increase in the linewidth of the signal due to the siloxane sites can be related to a decrease in mobility of the siloxane chains in a network that becomes more rigid. In the C sp2 region, the sharp peaks due to residual vinyl groups have disappeared, but now a broad peak is present centred at 6 140 associated with spinning sidebands, which is certainly characteristic of aromatic carbons present in a free carbon phase.16 This phase is related to the decomposition of the organic moieties.The peaks due to D (6 -22) and M (6 6) units are still present in the 29Si MAS NMR spectrum of the 600°C sample [Fig. 6(u)], showing clearly a retention of the siloxane back- bone. The linewidths have increased slightly, certainly because of an increasing rigidity of the network, responsible for a larger distribution of sites. The peak due to D2Hunits at 6 -39 has disappeared completely. Consumption of these units could be due to condensation reactions in the presence of traces of H20, such as physisorbed water.This reaction proceeds rapidly at moderate temperature (around 290 "C). Thus intra- or inter- molecular cross-linking of PMHS can occur according to the following successive reactions as reported by Hetem et =Si-H +H20+ SSi-OH +H2r -=Si- OH +H-SiE +ESi-0-Sif +H2 Small peaks are observed at 6 -57 and -66 due to T, and T, units respectively. The presence of these units may be due to hydrolysis of D,Hunits, eventually followed by condensation, according to the above equations. It could also be due to redistribution reactions involving Si-C and Si- 0 bonds that are known to take place at these temperatures in similar system^.^^.^^ The 29Si NMR spectrum of DE6 pyrolysed at 1000°C [Fig.6(a)] shows a great change in the structure of the material. Broad peaks are observed in the spectrum, indicating an increase in disorder, associated with the polymer-to-ceramic transformation. The proportion of T units (29%) characterized by the peak at 6 -70 is greatly increased. A new peak appeared at 6 -110 (23%) due to Q units. The region between 6 20 and -50 exhibits several broad components which could be simu- lated with three peaks at 6 7, -13 and -32 corresponding to M (4%), X (16%) and D (28%) units, respectively (Fig. 7). The redistribution reactions mentioned for the 600 "C sample have clearly proceeded to a larger extent at 1000°C. The spectrum at this temperature is characteristic of an oxycarbide network with a complete distribution of SiC,04-, units.J. Muter. Chern., 1996, 6(lo), 1673-1678 1675 A 0 0I1 I ' ' I-----I -20 -40 -60 -80 -100 I -120 -140 -040 20 I x6 ?? 0 200 400 600 800 10001200140016001Fig. 7 Simulation of the 29S1 MAS NMR spectrum of DE6 pyrolysed 55 at 1OOO"C-1 (a) Between 1000 and 1400 "C, the proportions of mixed SiC,O,-, 0units (O<x<4) decrease strongly in favour of SiO, and SIC, units Indeed, at 1400°C the material is mainly composed of SIC, (6 -15) and SiO, (6 -110) units, with respective percent- ages of 35% and 44% Such structural rearrangement of silicon oxycarbide at high temperature leading to a phase separation 't between silica and silicon carbide-rich phases has already been P I I reported 23 34 In contrast, the spectrum at 1500"C shows only one sharp peak at 6 -18 corresponding to silicon carbide This result is in perfect agreement with the X-ray diffraction data which show only peaks due to crystalline Sic phases, mainly p with some traces of a (Fig 8) All the 29S1 MAS NMR spectra have been simulated to extract the percentages of the various Si sites From these results, it is possible to evaluate the number of Si-0 and Si-C bonds per Si as has already been done in several papers (Fig 9) l7 The number of Si-0 and Si-C bonds per Si do not vary greatly until 1400 "C, showing that in this temperature range the changes in Si sites shown in the spectra are related mainly to redistribution reactions between Sir 0 and Si- C bonds Such structural rearrangements known for siloxanes in the 200-600°C range have already been found at higher temperatures for silicon oxycarbide phases 34 The slight increase in the number of Sir0 bonds between 600 and 1000°C is certainly caused in this temperature range by the polymer-to-ceramic transformation which might occur with some Si-C bond cleavages An interesting feature is the total disappearance of Si-0 bonds at 1500 "C, due to carbothermal reduction of those bonds by the free carbon phase, the forma- tion of which was predicted from the I3C CP MAS NMR spectrum of the 600°C sample An evaluation of the free carbon content can be determined by a comparison of the 10 20 30 40 50 60 70 80 90 2Bldegrees Fig.8 X-Ray diffraction patterns of (a) DE6 and (b) DO6 pyrolysed at 1500"C under argon 1676 J Muter Chew , 1996,6(10), 1673-1678 0 200 400 600 800 loo0 1200 1400 1600 T/"C Fig.9 Variation of Si-X/Si ratios for DE6 (a) and DO6 (b) as a function of temperature NMR and chemical analysis results assuming that the C atoms in the oxycarbide phase are bonded to 4 Si atoms Around 70-75% of the total C content is in a free C phase in the samples pyrolysed from 1000 to 1400"C, and this C reacts completely at 1500°C with the oxycarbide phase to give crystalline silicon carbide Pyrolysis of DO6 samples DO6 exhibits a mass loss (175%) at low temperature (180-420°C) that was not observed for DE6 [Fig 5(b)] As mentioned already, the IR spectrum shows the presence of OH groups and water in this sample, and thus this mass loss may be ascribed to the evaporation of adsorbed solvent In a way similar to that observed for DE6, a large mass loss (38%) is observed between 440 and 560°C due to the decomposition of the aliphatic chains This is confirmed by chemical analysis results, which show that only one carbon per silicon remains after pyrolysis at 600°C indeed, it corresponds to a total disappearance of the organic chain in agreement with NMR data (to be presented later) Then the small mass loss between 600 and 840 "C (3 5%) is due essentially to a loss of hydrogen, as confirmed by chemical analysis which shows a sharp decrease of H content from 600 to lOOO"C, from 40 to 0 6 mass% The I3C NMR spectrum of DO6 has changed greatly between room temperature and 600 "C (Fig 4) Only one peak at 6 -3 7 remains in the spectrum after treatment at 600"C, corresponding to CH, -Si groups in T units This agrees with the drastic decrease in C and H contents as revealed by Table 3 Chemical analyses (mass%) of pyrolysed DO6 at different temperatures molar T/"C Si (YO) C (YO) H (Yo) 0 (%)o composition 25 2301 4064 882 2753 S1Cq13H10760210 600 4561 1645 404 3390 SiCo84H2490130 1000 4396 1366 063 41 75 SiCo,3H0400,67 1500 4585 1201 to20 4194 SIC0610161 "Oxygen content determined by difference, 0 (%)= 100-[Si (%)+C (%)+H (Yo)] Table 4 Composition of the silicon oxycarbide phases obtained at lo00 "C sample C/Sl" O/Sl" sio, (%)b sico, (%)b sic,o, (%)b sic30 (%)b SIC, ( %)b DE6 152 138 23 29 28 4 16 -DO6 0 73 167 44 34 19 "From chemical analysis bFrom 29S1 NMR results chemical analysis (Table 3) and corresponds to the high mass loss observed in the TG curve Note that no evidence for aromatic carbon at this temperature is present in this spectrum, compared to the spectrum of DE6 pyrolysed at the same temperature [Fig 3(a)] The 29S1 spectrum at 600°C [Fig 6(b)] shows significant changes in the chemical structure of the starting polymer The peak corresponding to T, sites [H3C-Si(0, 5)2-OR] observed at 6 -58 in the spectrum before pyrolysis, has decreased strongly and is present only as a shoulder of a main peak at 6 -67 characteristic of T, sites [H,C-Si(O,,),] Almost all of the T, sites have thus been transformed into T3 sites This indicates clearly that a large number of the Si-0-C bonds have been cleaved, leading to the formation of new rSi-O-Si= linkages This is in agreement with the previous results (chemical analysis, TG analysis and I3C NMR spectra) which show a complete degradation of the organic chains the siloxane network has indeed been transformed into a polymethylsilsesquioxane The peaks at 6 -21 and -110, due to new D, and 4, units, result from redistribution reactions as already mentioned for DE6 The same phenomenon of peak broadening mentioned for DE6 is also observed in the case of DO6 by pyrolysis up to 1000 "C [Fig 6(b)] The number of D (6 -37) and Q (6 -106) units has increased to the detriment of that of T (6 -70) units Distribution reactions between Si-0 and Si-C bonds are still occurring The composition of the oxycarbide phase is as follows 44% Q units, 35% T units, 19% D units and 3% M units, which is quite close to what was observed previously for silicon oxycarbide phases derived from gels based on T units 30 At 1500°C, the 29S1 NMR spectrum [Fig 6(b)] shows a broad peak at 6 -17 due to SIC, units and an intense one at 6 -110 due to SiO, units showing that the material is mainly made of a mixture of SiO, (73%) and SIC (22%) phases This result is in agreement with the X-ray diffraction pattern [Fig 8(b)] Some oxycarbide sites are still present, charac- terized by a broad peak around 6 -35 (5%) The evolution of the number of Si-C and Si-0 bonds per Si does not show any remarkable changes us pyrolysis tempera- ture, even at T= 1500°C [Fig 9(b)] Discussion The difference in pyrolytic behaviour of the two cross-linked PMHSs is clearly related to the difference in their composition C/Siz4 for both but O/Siz1 for DE6 and z2 for D06, as well as in their architecture the grafting of the organic chains is made with Si-C bonds for DE6 and Si-0-C bonds for DO6 At 600"C, degradation of the organic chains has occurred, leading to a complete loss of the C groups in the case of D06, while for DE6 the C groups bonded to the siloxane backbone remain, so that C/Si z2 for DE6 and z1 for DO6 This large difference in composition clearly influences the nature of the oxycarbide phases obtained at 1000°C and reported in Table4 Mutin et have already shown for Si -C-0 precursors with different compositions that the environment of the Si sites in the derived oxycarbide phase can be described as a purely random distribution of Si-0 and Si-C bonds, and thus that the structure depends on the O/Si molar ratio In the present examples, O/Si= 1 38 for DE6 and 1 67 for D06, and indeed the compositions of the oxycar- bide phases show trends very similar to that predicted by Mutin, much richer in C for DE6 (Coxy/Si = 0 40) than for DO6 (Cox,,/Si =0 22) Comparison between the 29S1 NMR results and elemental analysis shows the presence of a free carbon phase, Cfree/Si= 1 12 for DE6 and 0 51 for DO6 The high-temperature (1000< T < 1500) behaviour of the silicon oxycarbide phase can usually be divided into two steps at T < 1500 "C, structural rearrangement involving Si -C and Si-0 bonds occurs, leading to a strong decrease in the mixed Si units, SiC,O, -,(0< x < 4) A clear phase separation occurs between silica and silicon carbide-rich regions This has been observed in the present samples At T3 1500"C, the Sir-C-0 system is known to be unstable46 48 Carbothermal reaction between silica and carbon to produce Sic is well known The extent of such reactions is governed by the amount of C and 0 present in the pyrolysed samples In the case of DE6, this reaction clearly has occurred at 1500°C and leads to the crystallization of silicon carbide, with an almost total consump- tion of Si-0 bonds In contrast, silica is still present in DO6 pyrolysed at 1500"C, due to a much lower amount of free carbon, compared to the 0 content that certainly prevents carbothermal reactions from occurring to a large extent References 1 G M Renlund, S Prochazka and R H Doremus, J Muter Res, 1991,6,2716 2 H Zhang and C G Pantano, Muter Res Soc Symp Proc, 1992, 271,783 3 G D Soraru, V M Sglavo, S Dire, G DAndrea and F Babonneau, in Third Euroceramics, ed P Duran and G F Fernandez, Faenza Editrice Iberica S L , 1993, vol 2 p 1157 4 L Bois, J Maquet, F Babonneau and D Bahloul, Chem Muter, 1995,7,975 5 J Homeny, G G Nebon and S H Risbud, J Am Ceram Soc, 1988,71,386 6 G D Soraru, J Sol-Gel Sci Techno1, 1994,2,843 7 J Lipowitz, J Znorg Organornet Polym , 1991,1,277 8 K Kamiya, A Katayama, J Matsuoka and H Nasu, New Glass, 1994,3,4 9 F I Hurwitz, S C Farmer, F M Terepka and T A Leonhardt, J Muter Sci , 1991,26, 1247 10 H Zhang and C G Pantano, in Ultrastructure Processing of Advanced Materzuls, ed D R Uhlmann and D R Ulrich, Wiley, Chichester, 1992,pp 223-232 11 F I Hurwitz, L Hyatt, J Gorecki and L D'Amore, Cerum Eng Sci Proc , 1987,8,732 12 F K Chi, Ceram Eng Sci Proc , 1983,4,704 13 H Zhang and C G Pantano, J Am Ceram Soc , 1990,73 958 14 F Babonneau, K Thorne and J D Mackenzie, Chem Muter, 1989,1,554 15 K Kamiya, T Yoko, T Sano and K Tanaka, J Non-Cryst Solids, 1990,119,14 16 F Babonneau, L Bois and J Livage, J Non-Cryst Solids, 1992, 147 & 148,280 17 L BOB, J Maquet, F Babonneau, H Mutin and D Bahloul, Chem Mater , 1994,6,796 18 V Belot, R J P Corriu, D Leclercq, P H Mutin and A Vioux, J Non-Cryst Solids, 1992, 147 & 148, 52 19 R M Laine, J A Rahn, K A Yougdahl, F Babonneau, M L Hoppe, Z-F Zhang and J F Harrod, Chem Muter, 1990, 2,464 20 F Babonneau, L Bois, C-Y Yang and L V Interrante Chem Muter, 1994,6, 51 21 F Babonneau, G D Soraru, G D'Andrea, S Dire and L Bois, Muter Res SOC Symp Proc , 1992,271,789 22 A K Singh and C G Pantano, Mater Res Soc Symp Proc , 1992, 271,795 J Muter Chem, 1996, 6(10), 1673-1678 1677 23 G D Soraru, G DAndrea, R Campostrini, F Babonneau and G Mariotto, J Am Ceram SOC,1995,78,379 35 R Kalfat, N Gharbi, H Zarrouk and F Babonneau, J Soc Chirn Tunis, 1994,3, 533 24 V Belot, R J P Corm, D Leclercq, P Lefevre, P H Mutin and 36 R Kalfat and N Gharbi, J Mater Synth Process 1994,2,379 A Vioux, J Non-Cryst Solids, 1991,127,207 37 D Massiot, H Thiele and A Germanus, Bruker Rep, 1994,140,43 25 26 V Belot, R J P Cornu, D Leclercq, P H Mutin and A Vioux, J Non-Cryst Solids, 1992,144,287 G M Renlund, S Prochazka and R H Doremus, J Muter SOC, 1991,6,2716 38 39 40 J M Yu, D Teyssie and S Boileau, Polym Bull, 1992,28,435 M Hetem, G Rutten, B Vermeer, J Rijks, L Van De Ven, J De Haan and C Cramers, J Chromatogr, 1989,477,3 J F Harrod and A J Chalk, J Am Chem Soc ,1964,86,1776 27 28 J R Fox,D A White,S M Oleff, R D BoyerandP A Budinger, Muter Res Soc Symp Proc ,1986,73,395 D A White,S M Oleff,R D Boyer,P A Budingerand J R Fox, Adv Ceram Muter ,1987,2,45 41 42 43 A J Chalk and J F Harrod, J Am Chem Soc, 1965,87, 16 R Spindler and D F Shriver, Macromolecules, 1988,21, 648 J B Stothers, in Carbon-13 NMR Spectroscopy, Academic Press, New York and London, 1972 29 30 31 V Belot, R J P Corriu, D Leclercq, P H Mutin and A Vioux, J Non-Cryst Solids, 1994,176,33 R J P Corriu, D Leclercq, P H Mutin and A Vioux, Muter Res Soc Symp Proc ,1994,346,351 G C Wei, C R Kennedy and L A Harris, Ceram Bull 1984, 63,1054 44 45 46 V Belot, R J P Corriu, D Leclercq, P H Mutin and A Vioux, J Polym Scz Part A Polym Chem , 1992,30,613 V Belot, R J P Corriu, D Leclercq, P H Mutin and A Vioux, J Muter Sci Lett 1990,9, 1052 T Mah, N L Hecht, D E McCullum, J R Hoenigman, H M Kim, A P Katz and H A Lipsitt, J Muter Sci, 1989, 32 D A White, S M Oleff and J R Fox, Adv Ceram Muter, 1987, 24,271 2, 53 47 S M Jonhson, R D Brittain, R H Lamoreaux and D R 33 K C Chen, K J Thorne, A Chemseddine, F Babonneau and J D Mackenzie, Muter Res Soc Symp Proc ,1988,121,571 48 Rowcliffe, J Am Ceram SOC,1988,71, C132 K L Luhtra, J Am Ceram Soc ,1986,69, C231 34 G T Burns,R B Taylor,Y Xu,A ZangvilandG A Zank, Chem Muter, 1992,4, 1313 Paper 6/022351, Received 1st April, 1996 1678 J Muter Chem, 1996, 6(10), 1673-1678

ISSN:0959-9428    

DOI:10.1039/JM9960601673

出版商:RSC    

年代:1996

数据来源: RSC

摘要:

Chemically modified kaolinite. Grafting of methoxy groups on the interlamellar aluminol surface of kaolinite James J. Tunney and Christian Detellier" Ottawa-Carleton Chemistry Institute Department of Chemistry University of Ottawa Ottawa Ontario Canada Kl N 6N5 The interlayer aluminol surface of kaolinite has been modified by the reaction of methanol at temperatures between 200 and 270 "C with both the dimethyl sulfoxide intercalate of kaolinite (Kao-DMSO) and the N-methylformamide intercalate of kaolinite (Kao-NMF). The product was a methoxy-functionalized organomineral material which was resistant to thermal decomposition in both air and N2atmospheres up to temperatures >350 "C and also to water hydrolysis. Based on results from thermal analysis IR analysis NMR spectroscopy (I3C CP MAS 29Si CP MAS and 27Al MAS) and elemental analysis a structural model has been proposed in which every third interlayer surface hydroxy group on the aluminol surface of kaolinite has been replaced by a methoxide group.The methyl groups which point away from this surface are keyed into the macro-rings of the adjacent silicate surface resulting in a non-centrosymmetric two-dimensionally ordered organomineral assembly. Organomineral derivatives are of interest because they combine the structural physical and chemical properties of both the inorganic host material and the organic guest species at a nanometer scale. Methods based on the modification of pre- existing inorganic structures offer considerable flexibility in the design of new hybrid materials.2 Kaolinite despite its wide abundance in nature3 and interesting intercalation properties,has been less studied as a precursor mineral for new organomineral materials when compared to the smectite clays for presumably due to the difficulty in expanding the interlayers of kaolinite compared to other expandable layered materials. This is unfortunate since kaolinite offers certain structural advantages over other layered materials. Kaolinite Al2Si,O5(OH) is a dioctahedral 1:1 layered aluminosilicate made up of two types of interlayer surfaceThe first is derived from the tetrahedrally coordinated silicate sheet and is made up of basal oxygens interconnected to form (SiO) macro-rings. The second surface is an aluminol surface (A1-OH) derived from the octahedrally coordinated aluminate sheet of kaolinite.The net effect of the asymmetry caused by the presence of two surface types is to impart a dipolar character to the interlayer spaces of kaolinite. Indeed it is this dipolar interaction which is to a large extent responsible for the relatively large cohesive energy in ka linite,which often makes intercalation difficult. Despite this it has been shown recently that it is possible to intercalate large polyethy- lene glycol molecules of molecular mass 3400 into kaolinite albeit very s10wly. Importantly this interlayer structural asymmetry also enables kaolinite to act as a two-dimensional non-centrosymmetric supramolecular organizing medium for guest molecules. For example it has been demonstrated that when dimethyl sulfoxide (DMSO) is intercalated into the dipolar interlayer environment of kaolinite the two methyl groups become inequivalent thus making the DMSO molecule chiral. "- In most intercalation complexes of kaolinite the guest species are held in place by relatively weak attractive forces between the guest species and the inorganic kaolinite host These forces include hydrogen-bonding inter-actions dipolar interactions and van der Waals forces.Previously we reported that it was possible to graft a number of diols and alcohol ethers onto the interlayers of kaolinite. 20v21 Especially studied was the ethylene glycol grafted product which was proposed to have one end grafted to the aluminol surface of kaolinite via an A1-0-C bond with the free alcohol end keyed into the (SiO)6 macro-ring of the adjacent silicate surface.21 The surface modification of inorganic materials through the formation of surface ether linkages (=Surf-0-R) is not new and has been reported for a number of materials besides kaolinite. Reactions in which surface ether linkages are formed have been reported for silica,22-26 al mina " gibbsite and boehmi te,34-40 Ni(OH)2,4 H-magadiite (H4Si14030),42 zeolite^,^^-^^ MOO H20,49950vanadium pentoxide hydrate," and layered HLaNb207 . xH20 perov kite. . Many of these reactions are formed through the surface condensation reaction of alcohols whereby an alcohol group (R-OH) reacts with a surface hydroxy group (=Surf-OH) to give the condensation products =Surf- 0-R and H20.Initial attempts to graft the simplest alcohol methanol onto the interlayers of kaolinite in a similar way had been unsuccess- ful presumably due to the mild reaction conditions employed (refluxing Kao-DMSO in methanol at 60 "C). Subsequently it was found that the methoxy functionalization of kaolinite is indeed possible if one employs harsher reaction conditions. These results are described in this paper. Experimental Materials All chemicals used were of reagent-grade quality and were not further purified unless otherwise specified. KGa-1 well crys- tallized kaolinite was obtained from the Source Clay Repository of the Clay Minerals Society (Department of Geology University of Missouri USA).Additional purifi- cation primarily by standard sedimentation procedures was required to remove coarse impurities in most of the clays. 54 Even after purification small amounts of anatase (TiO,) impurities cpdd be detected by XRD as a sharp weak reflection at d=3. 52 A corresponding to the (101) reflection of anatase. Characterization XRD powder patterns were produced using a Philips PW 3710 a$omated diffractometer using Cu-Ka radiation (A=1. 5416 A). All measurements were taken using a generator voltage of 45 kV and a generator current of 40 mA. A step size of 28=0. 04" was used with a dwell time of 0. 5 s per step. The sample was spun during pattern acquisition and an automatic J.Mater. Chern. 1996 6( lo) 1679-1685 1679 divergent slit and a 0 1 mm receiving slit was used without employing any mask Samples were mounted on a circular glass disk by first dispersing 20-30 mg of sample with 1 ml of methanol sonicating for 10 s and allowing the suspension to dry on the glass disk. The intercalation ratio (I R) which is an indication of the extent of modification was calculated from the relative intensities of the (001) reflections of the modified organokaolinite phase and of the unexpanded kaolin- ite phase4. The c-spacing or basal spacing is calculated based on the indexed (001) reflections Reflections due to the presence of residual unreacted kaolinite are identified with the symbol (K)IR spectra were obtained on a Bomem Michelson MB 100 FTIR spectrometer using 30-50 averaged scans at 4cm resolution.The samples were prepared as KBr pellets (025-0 50% by mass in KBr) X-Ray fluorescence (XRF) measurements to measure the Si/Al ratio were performed on powder samples on a Philips PW 2400 fluorescence spec- trometer equipped with a Philips PW 2510 sample holder with the scans referenced to a mixture of SiO and A1203. The data were analysed with the PSA software giving an accuracy on the atomic ratios of 4% Ceramic yields were determined by measuring the residual mass (as a percentage of the starting mass) of the product after calcination at 1100"C for 3 h in air atmosphere. Thermal analyses (TG and DSC) runs were performed on a Polymer Labs 1500H instrument under either flowing nitrogen or air (20-90 cm3 min ) at a heating rate of 10-20 "C min- Approximately 10 mg of sample was used for each run using alumina sample and reference pans Water contents of the reaction media were measured on a Mettler DL18 Karl Fischer Titrator 13C CP MAS (50 33 MHz) 29S1 CP MAS (39 76 MHz) and 27Al MAS NMR (52 15 MHz) spectroscopies were all per- formed on a Bruker ASX-200 instrument with magic angle spinning rates ranging between 3 and 6 kHz.The 13C CP MAS spectra were referenced to hexamethylbenzene at 6 149 the 29S1 CP MAS spectra were referenced to tetramethylsilane (Me,&) at 6 00 and the 27Al MAS NMR spectra were referenced to an aqueous AI(N03)3 solution at 6 0 0 Solution-state NMR (lH and 13C) were performed on either a Varian XL-300 or a Bruker AMX-500 and both H and 13C chemical shifts are reported relative to Me,Si in CDC1 Kao-DMSO and Kao-NMF starting materials The preparations of both Kao-DMSO and Kao-NMF start-ing materials have been described elsewhere and are based on standard literature procedures 2o 21 Briefly this involved allowing 20g of the unexpanded purified kaolinite to remain in contact with DMSO or NMF in an enclosed jar for at least two months at room temperature followed by washing with 1,4-dioxane to remove excess DMSO or NMF and air drying After this time both Kao-DMSO and Kao-NMF were found by XRD to be formed with an intercalation ratio (I R ) exceeding 90% XRD patterns IR spectra and TG analyses of the products were all consistent with literature data Reactions with methanol Reaction 1.Kao-DMSO (I R =O 97 5 0 g) was mixed with 160 ml methanol (003% water by Karl-Fischer titration) in a 250 ml capacity PTFE-lined autoclave sealed and placed in a silicone oil bath. This was maintained at an oil bath tempera- ture ranging from 200 to 270 "C for 89 h. The reaction vessel was then allowed to cool to ambient temperature opened and the reaction mixture was filtered and washed with 40 ml methanol. This was dried at 100"C for 1 h. The final yield was 3 3 g of an off-white powder (some product was lost during the filtration step) Reaction 5 (control). Kaolinite (10 g) was mixed with 20 g dry methanol (<0 05% water by Karl-Fischer titration) in a 30 ml capacity glass-lined autoclave sealed and heated 111 a silicone oil bath between 200 and 215°C for 120h After filtering washing with methanol and air-drying this afforded an off-white powder XRD and FTIR confirmed this to be unreacted kaolinite Water treatment of Kao-OMe Kao-OMe (0 3 g reaction 1)was mixed with 10 ml water and stirred for 14 days at room temperature Upon centrifugation and washing (once) with water tbe sample was collected and characterized XRD Table 1 Summary of the reactions of kaolinite with methanol showing XRD basal spacing and intercalation ratio characterizations reaction conditions XRD characterization starting material reaction T/"C duration/h d spacing/A IR Kao -DMSO 1 200-270 Kao -DMSO 2 190-230 Kao -DMSO 3 155-160 Kao-NMF 4 190-200 Kaolinite 5 200-2 15 "No reaction Results and Discussion Reaction of methanol with kaolinite A number of attempts were made to react the interlayer hydroxide surface of kaolinite with methanol at high tempera- tures (Table 1) It was found that when the kaolinite was first intercalated with either DMSO or NMF and then exposed to methanol at sufficiently high temperatures in an autoclave a new product was formed whiFh is characterized by ,a basal spacing ranging between 8 17 A (reaction 1) and 8 63 A (reac-tion 2) depending on the reaction conditions When the reaction temperature was too low (reaction 3) one obtained the unmodified Kao-DMSO startigg material as the dominant phase characterized by an 112 A basal spacing A control experiment (reaction 5) where unexpanded kaolinite was used instead of either Kao-DMSO or Kao-NMF showed that no reaction with methanol occurs unless access is first provided to the interlayer spaces of kaolinite Only then can the conden- sation of methanol molecules on the interlayer aluminol surface of kaolinite proceed to form the methoxy-modified kaolinite product (Kao-OMe) Water is formed as a byproduct and the intercalated guest species (DMSO or NMF) are expelled during the course of reaction An idealized schematic structural model of the Kao-OMe reaction product is given in Fig 1 illustrating its proposed non-centrosymmetric structure The water contents of the solvent before (003%) and after reaction (2 2%) for reaction 1 were monitored by Karl-Fischer titration. This increase in water content is greater than one expects if the sole source of water production is assumed to be through the covalent grafting of methoxy groups onto the hydroxy surface of kaolinite It appears that while the forma- tion of water is consistent with the proposed surface conden- sation grafting mechanism it cannot account for the formation of all of the water It is probable that dimethyl ether and other possible condensation products were formed during the course of this high-temperature sealed reaction.The formation of organic condensation products from methanol using alurnina3l and pillared as catalysts is known In fact the presence of secondary condensation reaction products was qualitatively confirmed by GC-MS analysis of the mother-liquor which showed the presence of dimethyl ether and 2-methoxyethanol The basal spacing of the mFthoxy-modified kaolinite (Kao:OMe) from reaction 1 is 8 2 A (Fig 2),owhich represents a 1 0 A layer expansion from the parent 7 2 A kaolinite mate- rial Residual unexpanded kaolinite is observed but approxi- mately 90% of the layer modification was achieved judging from the intercalation ratio A number of higher order basal reflections could be indexed Since the minimum clearance space for a methanol molecyle trapped between two layers has been estimated to be 3 7 A,56 it is unlikely that methanol is simply loosely intercalated in the interlayer spaces of kaolinite. The very small basal spacing observed is attributed to the keying-in of the methyl group into the (SiO) macro- ring of the silicate sheet in the adjacent kaolinite layer Precedents for this keying-in phenomenon for kaolinite are known 4 15 21 57 58 Interestingly it owas recently reported that a similar small expansion of 13 A occurred for the layered perovskite HLaNb,07 -xH,O upon methoxy modification with methanol 52 Halloysite which is a polytype of kaolinite has been observed to form a m$hanol intercalation compound with a bas:l spacing of 106A5960 Carr and Chih later reported a 9 5 A basal spacing for a methanol treated hydrated halloysite which they attributed to a partial dehydration of the complex rather than to a true halloysite-methanol complex More recently Costanzo and Giese,61 on the basis of IR results have reported the formFtion of a disordered methanol intercalate derived from an 8 4 A synthetically hydrated kaolin- ite In this case the XRD pattern of the complex could not be measured because deintercalation of the methanol proceeded too rapidly under ambient ccyditions Similarly the intercal- ation of methanol into an 8 6 A synthetically hydrated kaolinite was reported to yield an intercalate with a basal spacing of 10 95 A 62 In all cases methanol-intercalated kaolinite or hal- loysite exhibit much larger basal spacings than that observed for the 8 2A phase and were unstable under ambient conditions.The layer expansions reported for other methoxy- functionalized layered materials such as b eh mite FeOC163 and N1(OH)241 are much larger than the 10 A observed here due in part to the fact that there are methoxy groups on both the bottom and top interlamellar surfaces of these materials and not just on one of the surfaces The relationship between theo Kao-OMe 8 2 A phase (reac- tion 1)and the Kao-OMe 8 6 A phase (reaction 2) is thought to be related tc the presence of co-intercalated water in the Kao-OMe 8 6 A phase Support for thisohypothesis is provided from the observation that when the 8 2 A phase is washed with water for 72 h !t room temperature a product with a dool reflection at 8 6 A is observed by XRD (Fig 3) IR (see below) also confirms that the wate$-washed sample is very similar to that of the Kao-OMe 8 6 A product of reaction 2 as well as the Kao-OMe 8 5 A product of reaction 4 An 84A bydrated ehase of kaolinite has been reported,64 as have 8 4 A and 8 6 A synthetically hydrated kaolinites 65 The synthetically hydrated kaolinites were prepared following the partial replacement of some of the interlayer aluminol hydroxy groups with fluoride ion in order to reduce the interlayer cohesive forces enough to allow for a monolayer of water to be remain trapped between the kaolinite layers By replacing a number of these hydroxy groups with methoxy functionalities this same reduction in cohesive energy may have resulted in the intercalation of water to form an 8 6 A partially hydrated Kao-OMe-H,O product where the spacing between layers is now determined by a monolayer of interlayer water Upoq heating the 8 6 A phase at 150"C in an oven for 4 h the 8 2 A phase was regenerated albeit at a slightly smaller intercalation ratio than before (Table 2).These treatments suggest that some of the methoxy functionalities were hydrolysed upon prolonged exposure to intercalated water Comparison of the relative intensities of the v(CH) stretching 2O/degrees Fig. 3 XRD patterns (20=6-16 ") of (a) Kao-DMSO starting mate- rial (b)Kao-OMe (reaction l) (c) B +water washing for 72 h at room temperature (K) indicates residual kaolinite Table 2 Summary of XRD data showing the effects of water washing on the methoxy-functionahzed kaolinite product of reaction 1 intercalation sample sample description dool/A ratio a Kao-DMSO starting material 11 14 0 98 b Kao-OMe (reaction 1) 8 20 0 89 c b +water 8 62 0 86 d c +heating at 150"C for 4 h 8 10 0 82 1682 J Muter Chem 1996 6(10) 1679-1685 intensities of the products before and after washing also suggests that some hydrolysis of the methoxy functionalities had occurred (Fig 4) A chemical formula for Kao-OMe (reaction 1) of Si,Al,0,(OH)3 13(OCH3)0 87 can be assigned based on the ceramic yield of the product Within the limits of error this formula matches the hydrogen content (calc 2 12% found 2 00%)as well as the Si/Al ratio (calc 1 0 found 1 1) determined from XRF.The discrepancy in the carbon content (calc 3 86% found 2 17%) is plausibly due to the difficulty in combusting all of the carbonaceous material during elemental analysis Carbonaceous material may be trapped in the interlayers of kaolinite forming a carbon-aluminosilicate matrix.This phenomenon of carbon-mineral nanocomposite formation was observed for other thermally robust organokaolinites l4 21 Only through prolonged combustion in air at temperatures >1050"C is all of the organic material combusted (see thermal analysis results below) This formula implies that between a quarter and a third of the surface hydroxy groups have been replaced with methoxide groups Since a significant amount of the product is in the form of residual unexpanded kaolinite (approximately 11 YOas estimated from the intercalation ratio) replacement of about one third of the surface hydroxy groups by methoxi<e appears to be the better estimate for the Kao-OMe 82A product phase For comparison approximately one half of the hydroxy groups were replaced with methoxide for the methoxy-func- tionalized boehmite der vative and one third of the surface hydroxy groups were replaced in Ni [(OH), (OCH3)l/3]241 Since the distance between adjatent surface hydroxy groups in kaolinite is approximately 2 8 A," 66 and the van der Waals diameter of a methyl group is 3 9 a completely meth- oxylated surface would be sterically forbidden Assuming ide$ hexagonal symmetry with interhydroxy distances of 2 8 A replacement of one third of the interlayer hydroxy groups with methoxy groups in an idealized hexagonal two-dimensional ordcring would give intermethoxy distances of approximately 4 9 A which allows for a good separation of methyl groups The replacement of one third of the hydroxy groups for methoxide groups would also correspond to a ratio of one methoxide group per (SiO) macro-ring.This would allow each methoxy group to be keyed into one (SiO) macro-ring 3800 3600 3400 3200 3000 2800 v/crn- Fig. 4 FTIR spectra (3800-2800 cm ) of (a) kaolinite (b)Kao-DMSO (c) Kao-OMe (reaction l),(d) (c)+water washing for 72 h at room temperature of the silicate surface of tht adjacent layer and could account for the very low 1. 0A layer expansion which was observed. On this basis the theoretical upper limit for the methoxy functionalization of kaolinite should be Si2A1205(OH)3. 0(OCH3)l.0. IR analysis The IR OH stretching region of kaolinite is very sensitive to the effects of interlayer modificati n. * . * Consequently interlayer methoxy functionalization of the hydroxy surface should have a major influence on the diagnostic OH stretching pattern of kaolinite. This is indeed shown to be the case in Fig. 4 where it is readily observed that the OH stretching pattern of Kao-OMe [Fig. 4(c)] is very different from that of either the parent kaolinite material [Fig. 4(a)] or Kao-DMSO [Fig. 4(b)]. The band at 3696 cm- which is attributable to one of the stretching modes of the surface hydroxy groups of kaolinite remains at roughly the same wavenumber albeit at somewhat weaker relative intensity.The inner hydroxy stretch- ing vibration at 3620cm- is also unshifted compared to kaolinite although once again the relative intensity of this band appears to be diminished somewhat. The major difference between the IR spectra of kaolinite [Fig. 4(a)] and Kao-OMe [Fig. 4(c)] isothe appearance of a band at 3646 cm- for Kao-MeOH 8. 2 A. This is assigned to the OH stretching motion of weakly perturbed surface A1-OH groups. It is noteworthy that no low energy bands due to guest molecules hydrogen bonded to the surface hydroxy groups are observed as is the case for Kao-DMSO [Fig. 4(b)] where hydrogen-bonded surface hydroxy bands are observed at 3538 and 3505 cm- .Nor is there any conclusive evidence for C -0-H stretching bands due to hydrogen-bonded intercalated methanol where one would expect to find a fairly intense band near 3340 cm- where the v(0H) band for neat methanol is found. The weak broad band which one observes between 3350 and 3600cm- is of the same intensity as for kaolinite and can be attributed to externally adsorbed water in either the clay or the KBr matrix. The absence of an alcohol v(0H) stretching band is consistent with the condensation of methanol on the hydroxy surface to yield a surface methoxide group. Further evidence for grafting is also seen from the C-H stretching region of Kao-MeOH [Fig 4(c)] where one observes three bands at 2957 (w) 2927 (w) and 2849 (w) cm- corresponding to the symmetric and antisymmetric C- H stretching bands of the surface methoxide group.Neat meth- anol by contrast has two C-H stretching bands at 2942 and 2832cm- whereas methanol in dilute CCl solution has v(CH) bands at 2977,2938 and 2830 cm-1. 32 Other methoxy- modified materials show similar bands. Both methoxyboehmite [A10(OH)o 5(OCH3)o 5]34 haveand aluminium meth xide CH stretching bands at 2940 and 2840cm-l. Methanol adsorbed on alumina activated at 500 "C exhibits at least three surface species each with a characteristic v(CH) IR pattern. 32 Bands attributed to the presence of a bridging methoxide group on alumina have been assigned at 2970 2955 and 2844cm- whereas bands due to a coordinated form have been assigned at 2960 and 2850 m- .These latter values are similar to those originally reported by Greenler who assigned C-H stretching bands at 2950 and 2840 cm- for an alumina methoxide species. 27 All this inditates that the position of the v(CH) bands for Kao-OMe 8. 2 A is consistent with methoxide species. Moreover the appearance of an ill- defined 6(CH3) band at 1460 cm- (vw) is also consistent with the presence of methoxide species. 32 Other regions of the IR spectra show that the kaolinite host mineral has been perturbed. The very strong Si-0 lattice vibration bands of kaolinite (1010 and 1032cm- ) were replaced by bands at 1033 and 1050 cm- for Kao-OMe and the b(A1-OH) in-plane bending vibration of the inner hydroxy of kaolinite6* was also perturbed from 915 cm- in kaolinite to 910 cm- in Kao-OMe.The 6(Al-OH) in-plane bending vibration of the surface hydroxy groups of kaolinite at 938cm- was no longer observed in the case of Kao-OMe. Kaolinite bands at 790 (w) 755 (w) and 694 (m)cm- have been replaced by new bands at 818 (vw) 802 (w) 743 (w) and 678 (m) cm- . Thermal analysis The thermal stability of Kao-OMe (reaction 1) was quite remarkable for an organokaolinite material. As shown in Fig. 5 the material did not begin to decompose in air until above 350 "C with the decomposition centred at 515 "C. Both FTIR spectra and XRD patterns of the material were identical before and after heating at 210°C for 1h showing the persist- ence of the presence of these intercalated methoxy groups after heating.This thermal stability strongly supports methoxy formation since when methanol was intercalated in the inter- layer spaces of hall ysite or hydrated kaolinite,61 methanol was deintercalated at ambient temperatures. In contrast when methoxide is grafted directly to the interlayer surface as is the case for HLaNb207-xH,O molybdenum trioxide dihy-drate49 and vanadium pent xide the methoxy groups were lost at temperatures above 200"C and in the case of the methoxy-modified layered perovskite HLaNb,O xH,O above 300 0C. 52 The decomposition of Kao-OMe (reaction 1) in air was characterized by an endothermic peak temperature of 515 "C which is 23 "C lower than the peak dehydroxylation tempera- ture of the kaolinite starting material (538 "C).Between 920 and 1050"C there was an exothermic event with a peak temperature of 1009 "C assigned to a structural reorganization of the metakaolinite-like aluminosilicate mineral matri . . In the pure kaolinite parent material this exothermic event was found to occur at 1003 "C and no mass loss was associated with this event. For Kao-OMe in air a 1. 4% mass loss was observed between 920 and 1050"C which is associated with this structural reorganization. It is also in this temperature range that the product regains its off-white colour. All this suggests that the carbonaceous material which was trapped in the metakaolinite matrix during the dehydroxylation step (ca. 400-550 "C) was released during this structural reorganization (ca.lOOO"C) and then combusted in air leading to the observed 1. 4% mass loss in the TG trace (Fig. 5). Fig. 5 TG of Kao-OMe (reaction 1) under air flow (85 cm3min- ) ramping at a rate of 20 "C min- J. Mater. Chem. 1996 6(lo) 1679-1685 1683 NMR analysis A 13C CP MAS NMR spectra of the Kao-OMe (reaction 1) product was recorded in order to obtain more information regarding the nature of the organic component of this organo- kaolinite Fig 6 shows that only one resonance is present at 6 51 1 as one might expect for a methoxy-functionalized kaolinite. This represents a slight downfield chemical shift compared to liquid methanol (6 49 3) 43. The difference between the 13C chemical shifts of aluminium alkoxides and the corre- sponding free alcohols is quite 71 For example it was noted by Inoue et u13 that the chemical shifts of the methyl and methine carbons of Al(OPr ) are similar to those of propan-2-01 (6 25 1 and 63 4 for methyl and methine carbons respectively) In following by I3C solution-state NMR the sol-gel synthesis of aluminium oxide from mixtures containing sec-butyl alcohol Rezgui et uZ71 found that the methine I3C resonance of sec-butyl alcohol (CH,CH,13CHOHCH,) at S 69 was shifted only slightly to 6 73 when in the form of an aluminium alkoxide ligand (= A1-0-I3CHCH3-CH,CH3) This suggests that it is difficult to differentiate between a methoxide group and a methanol group solely on the basis of chemical shift especially if the linewidth is fairly broad as is the case for Kao-OMe (vl =75 Hz) The I3C chemical shifts of Al(OCH,) methoxylated alum- ina and methoxylated boehmite are unreported but the 13C CP MAS NMR spectra of methanol adsorbed onto zeolites have been reported 43 44 46 Methanol adsorbed on zeolite H-Y exhibits two resonances 46.The resonance at 6 50 1 was attri- buted to two types of species a mobile species due to surface- adsorbed methanol and a rigidly bound methoxy species. The peak at 6 55 7 was also attributed to a rigidly fixed methoxy group on a different site For methanol adsorbed onto zeolite ZSM-5 a peak at 6 508 was attributed to surface-adsorbed methanol and not to methoxide species 43 From these zeolite results it appears that a reliable assignment for the formation of a surface methoxide group is unlikely solely on the basis of the 13C NMR chemical shift A dipolar dephasing experiment (I3C DD MAS),72 where the I3C signal was allowed to dephase for 40 ps cross polariz- ation showed that 55% of the signal intensity was maintained compared to the CP MAS conditions.This confirms that despite the keying in of the methyl group into the (SiO) macro-ring of the adjacent silicate surface the methyl groups are still fluctional presumably from rapid rotation about the C3 axis Recently Hayashi reported that in Kao-DMSO the methyl group that is keyed into the silicate surface is actually more fluctional (E,= 130 kJ mol-l) than the methyl group which is parallel to the kaolinite layer (E,= 16 5 kJ mol- ) Deuterium NMR studies on methoxy-modified HLaNb,O -xH,O indicated that the CD30 group was rigidly bound to the interlayer surface with almost all motion coming Fig.6 13C CP MAS NMR spectrum of Kao-OMe (reaction 1) (Bruker ASX-200 50 32 MHz spinning rate =4 kHz contact time = 1 ms) 1684 J Muter Chem 1996 6(lo) 1679-1685 from the rotation of the methyl group about the C axis (along the internuclear C-0 axis) 52 29S1 CP MAS NMR (39 76 MHz) studies of the Kao-MeOH product showed a single resonance at 6 -92 6 (v = 100 Hz) which compares to a resonance at 6 -91 2 (vl =80 Hz) for kaolinite. This 14 ppm upfield shift is diagnostic of a pertur- bation of the silicate surface to a more shielded environment but no major structural changes have taken place.The silicon atoms have remained in tetrahedral Q3(OAl) sites 73. These results are in accord with the 29S1 chemical shifts observed for other organokaolinite intercalates l5 I6 I9 2o Finally a 27Al MAS NMR (52 15 MHz spinning at 6 kHz) spectrum of Kao-OMe showed only one broad resonance centred at 6 -11 Using the same acquisition parameters kaolinite also showed a single broad resonance at 6 -18. These data are consistent with the aluminium in the Kao-OMe product being solely in octahedral coordination as it is for kaolinite Both 29S1 and 27Al NMR results indicate that the basic structural integrity of the kaolinite layers has been maintained in accord with the model of a methoxy-modified kaolinite Conclusions It has been shown that it is possible to modify the interlayers of kaolinite with methanol by first expanding the interlayers with DMSO or NMF and then treating at high-temperatures (>2OO"C).The methanol units are attached to the hydroxy surface of kaolinite through an alcohol condensation reaction whereby Al-0-C bonds are formed Kao-OMe exhibited thermal stability in excess of that expected for classical organo- kaolinite intercalates and did not decompose in nitrogen or air below 350°C. The observed d spacing was less than that expected based on the van der Waals dimensions of a methoxy group but can be rationalized in terms of the keying in of the methyl group into the (SiO) macro-ring of the adjacent silicate surface All data are consistent with the structural integrity of kaolinite being maintained upon methoxide modification Kao-OMe can be described as a non-centrosymmetric array of rigidly fixed methyl groups which have been grafted onto repeating kaolinite units The Natural Sciences and Engineering Research Council of Canada (NSERC) is gratefully acknowledged for its continuous financial support We thank Dr Glenn Facey (University of Ottawa) for recording the solid-state NMR spectra and Dr C Bensimon (University of Ottawa) for performing the XRF measurements References 1 J A Rausell-Colom and J M Serratosa in Chemistry ofclays and Clay Minerals ed A C D Newman Mineralogical Society London 1987 p 371 2 G A Om Adv Muter 1992,4,612 3 H H Murray in Hydrous Phyllosilicates ed S W Bailey Mineralogical Society of America Washington DC 1988 p 67 4 B K G.Theng,. The Chemistry of Clay-Organo Reactions Adam Hilger London 1974 ch 6 pp 239-260. 5 D M C MacEwen and M J Wilson in Crystal Structures of Claj Minerals and. Their X-Ray Identzjcatron ed G W Brindley and G Brown Mineralogical Society London 1984 p 197 6 R M Barrer Zeolites and Clay Minerals Academic Press London 1978 p 407 7 A Okada A Usuki T Kurauchi and 0 Kamigaito in Hybrid Organic Inorganic Composites ed J E Mark C Y-C Lee and P A Bianconi American Chemical Society Washington DC 1995 p 55 8 T J Pinnavaia Science 1983,220 365 . 9 R F Giese in Hydrous Phyllosilicates ed S W Bailey Mineralogical Society of America Washington DC 1988 p 29 10 D L Bish Clays Clay Miner 1993,41,738 11 R F Giese Clays Clay Miner 1978 26 51 12 R F Giese Bull Mineral 1982 105 417.13 M Cruz H Jacobs and J J Fripiat Proc Int Clay Conf 1973 46 C E Bronnimann and G E Maciel J Am Chem Soc 1986 P 35 108,7 154 14 J J Tunney and C Detellier Chem Muter 1996,8,927 47 C Tsiao D R Corbin and C Dybowski J Am Chem Soc 1990 15 S Hayashi J Phys Chem 1995,99,7120 112,7140 16 J G. Thompson Clays Clay Mzner 1985,33,173 48 M T Aronson R J Gorte W E Farneth and D White J Am 17 J G. Thompson and C Cuff Clays Clay Mzner 1985,33,490 Chem SOC 1989,111,840 18 M Raupach P F Barron and J G. Thompson Clays Clay Miner 49 E M McCarron 111 R H Staley and A W Sleight Inorg Chem 1987,35,208 1984,23,1043 19 M J J Duer J Rocha and J Klinowski J Am Chem Soc 1992 50 W E Farneth R H Staley and A W Sleight J Am Chem SOC 114,6867 1986,108,2327 .20 J J Tunney and C Detellier Chem Mater 1993,5,747 51 S Kittaka N Fukuhara and H Sumida J Chem Soc Furaduy 21 J J Tunney and C Detellier Clays Clay Mzner 1994,5 552 Trans 1993,89,3827 22 S Kitahara Bull Chem SOC Jpn 1976,49,3389 52 S Takahashi T Nakato S Hayashi Y Sugahara and K Kuroda 23 24 H Balard M Srdqi E Papirer J B Donnet A Tuel H Hommel and A P Legrand Chromatographia 1988,25,707 J R Sohn S G Ryu and J H Song J Mol Catul 1990,62 L1 53 Inorg Chem 1995,34,5065 T Matsuda N Miyamae and M Takeuchi Bull Chem Soc Jpn 1993,66,1551 25 J Chmielowiec and B A Morrow J Colloid Interf Sci 1983 54 R K Schofield and H R Samson Discuss Faraday Soc 1954 94,319 18,135.26 K Y Blohowiak D R Treadwell B L Mueller M L Hoppe 55 Y Morikawa Adv Catal 1993,39,303 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 S Jouppi P Kansal K W Chew C L S Scotto F Babonneau J Kampf and R M Laine Chem Muter 1994,6,2177 R G Greenler J Chem Phys 1962,37,2094 G K Boreskov M Yu K Shchekochikhin A D Makarov and V N Filimonov Dokl Akad Nauk SSSR Engl Trans1 Phys Chem Sect 1964,156,564 H Knozinger and B Stubner J Phys Chem 1978,82,1526 H Knozinger A Scheglila and A M Watson J Phys Chem 1968,72,2770 R S Schiffino and R P Merrill J Phys Chem 1993,97,6425 G Busca P F Rossi V Lorenzelli M Benaissa J Travert and J- C Lavalley J Phys Chem 1985,89,5433 Y-H Chin and P D Ellis J Am Chem SOC,1989,111,7653 T Kubo and K Uchida Kogyo Kagaku Zasshz 1970,73,70 M Inoue K Kitamura H Tanino H Nakayama and T Inui Clajs Claj Mzner 1989,37,71 M Inoue H Kominami and T Inui J Am Ceram Soc 1990 73,1100 M Inoue H Kominami and T Inui J Chem SOC Dalton Trans 1991 3331 M Inoue Y Kondo and T Inui Chem Lett 1986 1421 M Inoue Y Kondo and T Inui Inorg Chem 1988,27,215 M Inoue H Tanino and Y Kondo Clays Clay Miner 1991 39,151 S Le Bihan J Guenot and M Figlarz J Solid State Chem 1976 17 15 L Mercier G A Facey and C Detellier J Chem Soc Chem Commun 1994 21 11 M W Anderson and J Klinowski J Am Chem Soc 1990 112 10 M W Anderson and J Klinowski Nature (London) 1989 339 200 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 R M Carr and H W A Chrh Clay Miner 1971,9,153 C T Johnston and D A Stone Clays Clay Mzner 1990,38 121 Y Sugahara S Satokawa K Yoshioka K Kuroda and C Kato Clays Clay Mzner 1989,37 143 D M C MacEwan J Chem Soc Faraday Trans 1948,44,349 D M C MacEwan Nature (London) 1946,157,159 P M Costanzo and R F Giese Jr Clays Clay Miner 1990 38 160 N Wada R Raythatha and S Minomura Solid State Commun 198 7,63,78 3 S Kikkawa F Kanamaru and M Koizumi Inorg Chem 1976 15,2195 J .ITunney and C Detellier Clays Clay Mzner 1994,42,473 P M Costanzo R F Giese and M Lipsicas Clays Cluy Miner 1984,32,419 D L Bish and R B Von Dreele Clays Clay Miner 1989,37,289 D L Guertin S E Wiberly W H Bauer and J Goldeyson J Phys Chem 1956,60,1018 V C Farmer in. The Infra-Red Spectra of Minerals ed V C Farmer Mineralogical Society London 1974 p 33 1 P J Sanchez-Soto A Justo and J L Perez-Rodriquez Mater Scz 1994,29,1276 L A Perez-Maqueda J L Perez-Rodriguez G W Scheiffele A Justo and P J Sanchez-Soto J. Therm Anal 1993 39 S Rezgui B C Gates S L Burkett and M E Davis Chew Muter 1994,6,2390 L B Alemany,D M Grant,T D Algerand R J Pugmire J Am Chem Soc 1983,105,6687 S Hayashi T Ueda K Hayamizu and E Akiba J Phvs Chem 1992,96,10 922 1055-1 067 45 M T Aronson R J Gorte and W E Farneth J Catal 1987 105,455 Paper 6/01976E Received 21st Murch 1996. J Mater Chem 1996,6(10); 1679-1685 1685

ISSN:0959-9428    

DOI:10.1039/JM9960601679

出版商:RSC    

年代:1996

数据来源: RSC

摘要:

Growth of emerald crystals by evaporation of a K,O-MOO, flux Shuji Oishi" and Hirofumi Yamamoto Department of Chemistry and Material Engineering, Faculty of Engineering, Shinshu University, Wakasato, Nagano 380, Japan The growth of emerald (Be,Al,Si,O,, :Cr) crystals by the flux evaporation method in the system K20-Mo03 is reported. The crystal growth was conducted by heating a mixture of solute and flux at 1100 "C, followed by holding at this temperature for 1-10 days. Emerald crystals of lengths up to 4.5 mm and widths of 2.9 mm were readily grown isothermally. The crystal sizes were dependent on the evaporation loss of the flux and scaling up the mass of the solution. The obtained crystals were transparent and exhibited the typical emerald-green colour. The form of the emerald crystals was a twelve-sided prism bounded by well developed faces.The aspect ratios were in the region of 1.2 to 2.3. The density was 2.66 f0.02 g ~m-~. The IR absorption bands were in good agreement with literature data. The chromium was preferentially incorporated in the centre part of the crystals. Emerald (Be3A12Si,01, :Cr) is a beryl (beryllium aluminium silicate) doped with chromium. The chromium gives emerald its characteristic green colour. Emerald has been regarded as a beautiful gemstone for centuries. Like diamond, emerald has been the object of many attempted syntheses. Because emerald melts incongruently,' the crystals have been grown mainly by flux and hydrothermal method^.^.^ The growth of emerald crystals from fluxes is particularly attractive because it readily allows growth.A number of investigations into the flux growth of emerald crystals have been conducted. About 10 kinds of flux have been used for the growth of emerald crystal^.^.^ In particular, a Li,O-MOO, flux has been used successfully with slow cooling and temperature gradient techniques. We have found that emerald crystals can be grown easily by the evaporation of an Li,O-MoO, flux.' The oxide MOO, has been used as a flux, but it is very volatile. Addition of LizO made the high-temperature solution relatively non-volatile. The Li,O acted to control the amount of flux evaporation. Further, emerald crystals were also grown from a K20-MOO, flux;6 a combination of slow cooling with evaporation of the K20-MOO, flux has produced crystals of emerald., Little work has been reported on the growth of emerald crystals by the flux evaporation method apart from our Because flux evaporation may be carried out isothermally, this tech- nique offers the advantages connected with growth at constant temperature; for example, the temperature control is very easy.We have grown emerald crystals from a PbO-V205 flux by the slow cooling Further, emerald crystals were readily grown from MOO,, Moo3-B20, and Li,O-MoO, fluxes by the evaporation meth~d.~.~-" The present paper describes the growth of emerald crystals from a K20-MOO, flux by an isothermal technique involving evaporation of the flux. The effects of the addition of K20 to the MOO, flux, and of the holding time on the crystal growth were investigated.After scaling up the mass of the solution by a factor of 6, the growth of large emerald crystals was attempted. The mor- phology, density, IR spectra, homogeneity of the dopant and imperfections of the resulting crystals were also examined. In addition, crystals of cristobalite, phenacite and chrysoberyl which occurred as byproducts are described. Experimental The beryl mixture (Be,A1,Si,Ol8) was prepared from reagent- grade BeO, A1203 and Si02 powders. The oxide dopant (Cr,O,) was added at a concentration of 1.0 mass% of the beryl mixture. This mixture was used as the solute for the flux growth runs. Reagent-grade K2C03 and MOO, powders were used as the flux.The starting compositions and holding times are given in Table 1. The solute and flux powders were weighed, mixed together and placed in platinum crucibles of capacity 30cm3 (run 1-18: 36mm diameterx40mm high) and 240cm3 (run 19: 60 mm diameter x 80 mm high). The lids were loosely fitted and the crucibles were placed in an electric furnace with silicon carbide heating elements. The furnace was heated at a rate of ca. 45 "C h- to 1100 "C and was held at this temperature for 1-10 days. Afterwards, the crucible was removed and allowed to cool rapidly to room temperature. The crystal products were then separated from the flux in warm water. The crystals obtained were examined using an optical micro- scope and a scanning electron microscope (SEM).The crystal phases were identified by X-ray diffraction (XRD). The length, 1 (parallel to the c axis), and width, w (perpendicular to the c axis), of the emerald crystals grown were measured. The average length, la", and width, wav, of the crystals were calcu- lated for each growth run. The numbers of the crystals >O.lO mm in size were counted. The density of the crystals Table 1 Growth conditions of the emerald crystals from K,O-MOO, flux flux solute K,O run content/g content/g 1 6.00 0.00 2 6.00 0.75 3 6.00 1.50 4 6.00 2.25 5 6.00 3.OO 6 6.00 3.75 7 6.00 4.50 8 6.00 6.00 9 3.00 1.50 10 3.50 1.50 11 3.70 1.50 12 3.80 1.50 13 3.90 1.50 14 4.00 1.50 15 5.00 1.50 16 7.00 1.50 17 3.80 1.50 18 3.80 1.50 19 22.80 9.00 MOO, holding content/g time"/days 30.00 1 29.25 28.50 27.75 27.00 26.25 25.50 24.00 28.50 1 28.50 1 28.50 1 28.50 1 28.50 1 28.50 1 28.50 1 28.50 1 28.50 5 28.50 10 171.00 10 "Time held at the maximum temperature of 1100"C.J. Muter. Chem., 1996, 6(lo), 1687-1691 1687 was determined pycnometrically. IR spectra of the crystals were measured with a Perkin-Elmer Model 1650 FTIR spec-trometer. The samples were prepared as KBr disks. An electron probe microanalyser (EPMA) was used to study variations in the concentration of the major constituents in the grown crystals. The presence of impurities from the K,O-MOO, flux and Pt crucible was also checked.In addition, aluminium in beryl crystals grown from the K,O-MOO, flux was determined using electron spectroscopy for chemical analysis (ESCA) in order to study the possibility of chromium substitution of aluminium. Results and Discussion Growth of emerald crystals by flux evaporation Emerald crystals of lengths up to 4.51 mm and widths of 2.89 mm (ca. 0.04 g=O.2 carat) were grown isothermally from the K20-MOO, flux in runs 1-7 and 9-19. The crystals grown were prismatic and exhibited the typical emerald-green colour. They were identified as emerald by their XRD patterns, using literature data.I4 Typical crystals of emerald are shown in Plate 1. Emerald crystals could be grown at 1100"Cby evapor- ation of the K20-MOO, flux in analogy with the case of the Li,O-MOO, flux.5 The numbers and average mass of the emerald crystals >0.10 mm in size and the evaporation rate of flux are shown in Table2.No growth of emerald crystals was observed in run 8. The solubility of emerald in the K20-Mo0, flux was considered to increase with the MOO, concentration. The solution used in run 8 did not become supersaturated even after all of the flux had evaporated. Plate 1 Optical micrograph showing emerald crystals grown from K20-MOO, flux Table2 Number and average mass of the emerald crystals and evaporation rate of the flux crystal average evaporation run number mass/g rate/g h-' mm-2 1 84 4.8 10-4 1.2 x 10-3 2 6131 5.4 x 10-4 6.7 x 10-4 3 4443 5.6 x 10-4 4.2 x 10-4 4 436 1 3.8 x 10-~ 1.3 x 5 1025 2.0 x 10-4 6.3 x 10-5 6 62 0.5 x 10-4 5.6 x lo-' 7 51 0.4 x 10-4 3.3 x lo-' 8 - - 3.3 x 10-~ 9 4 2.5 x 10-4 4.5 x 10-4 10 579 1.0 x 10-3 4.2 x 10-4 11 1140 1.4x 10-3 4.7 x 12 1133 1.7 x 10-3 4.5 x 13 14 1181 1024 1.1 x 10-3 1.4 x lop3 4.6 x 5.2 x 10-4 15 1202 5.9 x 10-4 4.6 x 16 42400 8.0 x 10-~ 5.0 x 10-~ 17 785 2.6 x 10-3 1.4 10-4 18 714 2.9 x 10-3 7.9 x 10-5 19 550 1.4 x lo-* 1.5x 10-4 1688 J.Mater. Chem., 1996, 6(lo), 1687-1691 The effect of the flux compositions on the growth of the emerald crystals was investigated in runs 1-8. The relationship between the loss of flux upon evaporation and the amount of K,O in 30.00g flux is shown in Fig.1. About 100 massoh of the MOO, flux evaporated over a period of up to 1 day in the absence of K20. The evaporation loss decreased gradually with increasing amount of K20. About 34 mass% of the flux containing 1.50 g K,O evaporated. The rate of evaporation was calculated to be ca. 4.2 x g h-' mm-'. It was con- sidered that the evaporation loss from the K,O-MOO, fluxes consisted mainly of MOO,. The evidence for this view came from the precipitation of MOO, on the crucible lid and furnace brick. Owing to the preferential evaporation of MOO,, the flux composition changes with time. When 4.50 or 6.00g of K20 were present in the flux, the evaporation loss was only 2.7 mass%.The rate of evaporation was c~i.3.3 x g h-' rnrn-,. Low evaporation loss was considered to be due to the formation of K,O-MOO, complexes. The compositions of low-volatility fluxes used in runs 6-8 were regarded as K,O :4.6 MOO,, K20:3.7Mo0, and K,O :2.6Mo0,, respectively. These are close to the compound compositions, K2M040,,, K,Mo,O,, and K2Mo2O7, present in the system K,O-MOO,.'~ Addition of K20 to MOO, makes the high- temperature solution relatively non-volatile owing to the inter- action between a basic oxide, K,O, and an acidic oxide, MOO,. It was found that the flux evaporation loss could be determined by controlling the amount of K20. Emerald crystals grew from the MOO, fluxes containing up to 4.50g of K,O (runs 1-7).In run 3, large crystals up to I= 1.40 mm and tv=0.99 mm were grown. The I,, and w,, values were 0.74 and 0.56 mm, respectively. The number of crystals was 4443 and the yield was 2.49 g. The average mass of the crystals was 5.6 x g. The crystal sizes decreased with either decreasing or increasing amounts of K20. The crystal sizes and numbers appeared to be related to the solubility of emerald in the flux and the evaporation loss of the flux. The aspect ratios (/av/\vav) of the emerald crystals grown in runs 1-7 were 1.4, 1.2, 1.3, 1.5, 2.3, 2.3 and 2.2, respectively. The values were in the region of 1.2-2.3 and were dependent on the amount of K20.Generally speaking, the crystals with high aspect ratios were grown from the flux containing large amounts of K,O.No emerald crystals were grown in run 8. Taking the flux evaporation and crystal sizes into account, K,O (1.50 g)-MOO, (28.50 g) was found to be the most suitable flux for growing emerald crystals. The effect of solute content on the growth of the emerald crystals was investigated in runs 3 and 9-16. The evaporation losses of the fluxes were in the region of 34-42 mass% because the composition of the fluxes used was the same. The evapor- 100 h 8 80 v)2 E 60 -8 C0= 40?! 8. a5 20 0 0 1 2345 K20/9 Fig. 1 Relationship between evaporation loss of flux and amount of K20 in 30.0 g of K,O-MOO, flux ation rates were calculated to be (4.2-5.2) x g h-’ mm-2. All experiments gave emerald crystals up to I= 1.75 mm and w = 1.30 mm.The !,, and w,, are plotted against solute content in Fig. 2. Large crystals of la”= 1.18 mm and w,, =0.88 mm were grown from the mixture containing 3.80 g of solute (run 12). The number of crystals was 1133 and the yield was 1.93 g. The average mass of the crystals was 1.7 x lo-, g. The crystal size decreased gradually with either increasing or decreasing solute content. When the solute amount was 7.00 g, 42400 small crystals (3.39 g) of I,, =0.43 mm and w,, =0.31 mm were grown. The average mass of the crystals was 8.0 x lop5g. The numbers of crystals obtained increased with the amounts of solute. This indicates that the solute was consumed for the formation of nuclei in preference to the crystal growth.The use of dilute solutions favours the growth of large crystals. The aspect ratio of the grown crystals was 1.3-1.4 because the compositions of the fluxes used were the same. The optimum solute content in the flux for growth of emerald crystals is 3.80g. The effect of holding time on the growth of emerald crystals was investigated in runs 12, 17 and 18. In these runs, the evaporation losses were 34, 56 and 64 mass%, respectively. The evaporation loss increased with increasing holding time. As described above, crystals of I,, = 1.18 mm and w,, = 0.88 mm were obtained in run 12. The crystal sizes increased gradually with holding time. When the holding time was 10 days, crystals of I,, = 1.40 mm and w,,= 0.90 mm were obtained.Regardless of the holding times, the number of crystals produced was in the region of 714-1133. The yields of the crystals grown in the 1 day and 10 days growth runs were 1.93 and 2.05 g, respect-ively. There was no marked differences in the number and yield of crystals. The aspect ratios of the crystals grown in these runs were 1.3, 1.5 and 1.6, respectively. Despite the fact that the starting compositions of the fluxes used were the same, the aspect ratios of the grown crystals increased with increasing holding time. This is due to the changes in flux composition, i.e. the preferential evaporation of MOO,. Long holding times were found to be effective for the growth of large emerald crystals. An attempt to grow large crystals by scaling up the mass of the mixture by a factor of 6 was investigated in run 20.The composition of the mixture was the same as those of the mixtures used in runs 12, 17 and 18. The evaporation loss of the flux was 58 mass%. Emerald crystals of lengths up to 4.51 mm and widths of 2.89 mm were grown. The mass of the largest crystal was about 0.04 g (0.2 carat). Typical crystals are shown in Plate 2. 550 crystals were obtained. The average sizes were I,, = 2.55 mm and w,,= 1.2 1.o <E 0.8 $.-b ,m 0.6 0.4 U ‘0 0.2 3 4 5 6 7 solute contentlg Fig. 2 Variation in average length, I,, (O), and width, byav (St),of emerald crystals grown with solute content (runs 3 and 9-16) Plate 2 Photograph showing large emerald crystals grown from K,O-MOO, flux in run 19 Fig.3 SEM photograph showing a cyclic twin of chrysoberyl crystals grown in run 10 1.51 mm, respectively. The yield of the crystals was 7.70 g. The average mass of the crystals was 1.4 x g. The aspect ratio was 1.7. The use of a concentrated mixture was found to be effective for the growth of large emerald crystals. Byproducts such as a-cristobalite ( SiO,) and phenacite (Be,SiO,) crystals formed in almost all runs. The former crystals were up to 0.04 mrn in size, colourless and transparent. Their form was a hexagonal thin plate. The latter crystals were up to 1 mm in size, colourless and transparent. Their form was a hexagonal rod. In addition, chrysoberyl ( BeA1204) crystals, which were up to 0.5 mm in size, were formed from the solutions containing small amounts of solute (runs 9.- I 1).The colour of the crystals was dark green. Typical chrysoberyl crystals are shown in Fig. 3. The grown crystals were always twinned. The twin crystals had pseudo-hexagonal symmetry owing to cyclic twinning as reported by Tabata et Owing to the formation of 2 or 3 kinds of byproduct crystals, the system used for growing emerald crystals cannot rigorously be reduced to a pseudo-binary system of solute and flux. The crystallization process of emerald crystals from the high-temperature solutions of the K,O-MOO, flux is very complex. Characteristicsof the emerald crystals Many emerald crystals of good quality were obtained up to a size of 4.51 mm in length and 2.89 mm in width, which were transparent, having the typical emerald colour.Inclusions were rarely found in the emerald crystals grown from K,O-MOO, flux. The crystals were twelve-sided prisms with very flat surfaces. A typical example is shown in Fig. 4. On the basis of the XRD data and interfacial angle measurements, it was found that the crystals were bounded by the c{OOOl}, m{lOiO} and a(ll?O} faces. The c and rn faces were always well developed. The basic form and colour were not related to the growth conditions. The aspect ratios ( 1.2-2.3) were dependent on the flux composition as described above. This morphology was similar to those of the crystals grown from Moo,’ and Li,O-MoO,’ fluxes. Natural emerald crystals are also pris- J.Muter. Chem., 1996, 6( lo), 1687-1691 1689 Fig. 4 SEM photograph showing a twclvc-sided prism of emerald crystal. The base has the two c faccs. Thc prism has the six large 171 faces and the six small (I faces. 4000 3000 2000 1500 1000 500 wavenumberkm -1 Fig. 5 IR spectrum of emerald crystals grown from K,O-MOO, flux matic." On the other hand, hexagonal thin plate-like crystals of emerald were grown from a Moo,-B,O, flux.* The density of the crystals was determined pycnometrically to be 2.66f0.02 gcm-3 in good agreement with literature ~values (2.64,14 2.65 0.025 and 2.69 g cm "). In addition, IR spectra recorded in the range 450-4000 cm-' were obtained for the emerald crystals grown. An example is shown in Fig.5. Absorption bands at 493, 523, 590, 650, 681, 740, 806, 961, 1020 and 1204+3 cm-' were observed. These values were in good agreement with literature data for emerald crystals grown from L~,O-MOO~~ and V20519 fluxes. The observed bands were independent of the crystal growth conditions. A broad band at 3450cm-1 was also observed, which is believed to be extraneous to the sample and related to water pick-up by the KBr during sample preparation. Of course, sharp OH bands were not observed because the flux-grown emerald crystals contained no water. Natural and synthetic hydrothermal emer- ald exhibited sharp OH bands.20 The variations in the concentration of the major constituents in the grown emerald crystals were investigated by the use of electron probe microanalysis ( EPMA).Aluminium, silicon and oxygen were distributed almost homogeneously in the crystals. The distribution of beryllium could not be determined due to the low atomic mass of the element. The EPMA data showing the distribution of the dopant (chromium) in emerald crystals grown are shown in Plate 3. Plate 3 (a) and (b) show the chromium distributed in the faces cut perpendicular and parallel to the c axis, respectively. Obviously, the chromium was incorporated preferentially in the centre parts of the crystals. A small amount of chromium existed in the outer parts of the crystals. In addition, the chromium was not equally distributed in all directions so the amount was anisotropic. There was a preferential incorporation into prismatic faces compared to basal ones.Because some aluminium ions are substitutionally replaced by chromium ions in emerald crys- tal~,~.~'the amount of aluminium existing in the c and m faces of beryl crystals grown from the K,O-MOO, flux was deter- mined by ESCA. Consequently, no difference in the concen- tration of aluminium between the c and m faces was found. Therefore, the orientation-dependent chromium incorporation is probably due to the difference in adsorption on specific faces or in the crystallographic direction. The emerald crystals were contaminated by a very small amount of potassium from the flux. The potassium was also incorporated in the central parts of the crystals. However.molybdenum from the flux and platinum from the crucible was not detected. Conclusions Emerald crystals were grown by an isothermal technique involving the evaporation of a K,0-Mo03 flux. Emerald Plate 3 EPMA data showing the distribution of chromium in the faces cut perpendicular (u) and parallel (b)to the c axis 1690 J. Mtrter. ChPtTl., 1996, 6(lo), 1687-1691 crystals of lengths up to 4.5 mm and widths of 2.9 mm were readily obtained. The crystal sizes were dependent on the evaporation losses of the flux and scaling up the mass of the solution. The crystals were transparent and exhibited the typical emerald-green colour. The form of the emerald crystals 9 10 11 12 13 S. Oishi, Seramikkusu, 1994,29,417. S. Oishi and Y. Sumiyoshi, Kagaku To Kyoiku, 1995,43,531. S.Oishi and M. Hirao, J. Muter. Sci.,1991,26,6401. S. Oishi, N. Nishizawa and M. Hirao, Nippon Kessho Seicho Gakkaishi, 1991, 18,268. S. Oishi, N. Nishizawa and M. Hirao, Hoseki Gakkaishi, 1993, was a twelve-sided prism. The characteristics of the emerald crystals obtained were investigated. 14 15 18,11. JCPDS card 9-430. E. M. Levin and H. F. McMurdie, Phase Diagramsfor Ceramists, 1975 Supplement, The American Ceramic Society, Columbus, Ohio, 1975, p. 84 (Fig. 4273). References 16 H. Tabata, E. Ishii and H. Okuda, J. Cryst. Growth, 1974, 24/25, 656. 1 2 3 R. P. Miller and R. A. Mercer, Mineral. Mag., 1965,35,250. K. Nassau, J. Cryst. Growth, 1976,35,211. D. Elwell, Man-Made Gemstones, Ellis Horwood, Chichester, 1979, 17 18 E. S. Dana and W. E. Ford, A Textbook of Mineralogy, Wiley, New York, 1972, pp. 579-581. D. Elwell, Man-Made Gemstones, Ellis Horwood, Chichester, 4 D. Elwell and H. J. Scheel, Crystal Growth from High-Temperature pp. 15-30,58-69. 19 1979, p. 178. M. Ushio and Y. Sumiyoshi, Nippon Kagaku Kaishi, 1972, 1648. Solutions, Academic Press, London, 1975, pp. 20-58, 558-622. 20 For example, K. Kodaira, Y. Iwase, A. Tsunashima and S. Oishi and K. Mochizuki, J. Mater. Chem., 1995,5, 1257. T. Matsushita, J. Cryst. Growth, 1982,60, 172. C. Sakamoto, Jpn. Pat., 23278,1973. 21 C. Aurisicchio, 0.Grubessi and P. Zecchini, Can. Mineral., 1994, S. Oishi and K. Mochizuki, Br. Ceram. Trans., 1993,92,214. 32, 55. S. Oishi, K. Mochizuki and S. Hirano, J. Ceram. Soc. Jpn., 1994, 102, 502. Paper 6/01410K; Received 27th February, 1996 J. Muter. Chem., 1996, 6(lo), 1687-1691 1691

ISSN:0959-9428    

DOI:10.1039/JM9960601687

出版商:RSC    

年代:1996

数据来源: RSC

摘要:

Nucleation and crystal growth of analcime from clear aluminosilicate solutions Geoffrey S. Wiersema and Robert W. Thompson Department of Chemical Engineering, Worcester Polytechnic Institute, Worcester, Massachusetts 01609, USA Hydrothermal syntheses of the zeolitic mineral analcime have been carried out in clear aluminosilicate solutions with batch composition A1,0, 84S10, 87Na20 2560H20 in the temperature range 130-160 "C and under autogenous pressures Observation of the changes of the crystal sizes with time indicated that the analcime crystals grew at a constant rate, dependent on the synthesis temperature, and that the rate decreased when the particles settled to the bottom of the autoclave At that point in the synthesis, a second population of nuclei was observed to form and grow in the solution above the settled crystals It was demonstrated that when pure forms of silica were used in these syntheses, fewer crystals were formed These results supported the concept that nucleus formation is promoted by impurities in the silica source Molecular sieve zeolites are crystalline aluminosilicates which have numerous industrial uses owing to their very regular pore openings on the molecular level The term 'molecular sieve' stems from their ability to separate molecules based on their physical size in relation to the pore openings in the zeolites Catalytic properties of zeolites are due to acid sites in the crystal framework, which arise from the aluminate ions in the lattice when the charge is compensated by a proton While there are significant deposits of natural zeolites which have some commercial importance, especially in ion-exchange pro- cesses, most zeolites used commercially are synthetic, having the advantages of purity and uniformity Zeolites generally are synthesised in hydrothermal systems at elevated temperatures at solution pHs in the range 9-13 5, depending on the system, although some low-pH syntheses have been reported Typical crystallizations from solution involve formation of nuclei of the crystalline phase by any of several well researched mechanisms,2 followed by the growth of those crystals by the incorporation of solute from the solution phase The relevant issues pertaining to molecular sieve zeolite synthesis are the mechanism by which zeolites nucleate to form the smallest fragment of crystalline material, and whether the rate of crystal growth is limited by diffusion from the bulk solution to the crystal surface or by the kinetics of incorporation at the crystal surface In spite of over four decades of research and progress in zeolite synthesis techniques and experience, there is still much uncertainty regarding the relevant mechanisms of zeolite nucleation and crystal growth While most syntheses of impor- tance occur in the presence of an amorphous aluminosilicate gel which forms soon after mixing the ingredients, there have been several reports of synthesising various zeolites from clear aluminosilicate solutions 21 Clear solutions provide a means of evaluating these crystallisation systems in situ by optical microscopy9 lo or quasi-elastic light scattering spectroscopy (QELSS)," l8 r e, without having to invade or interrupt the process Additionally, since most clear solution syntheses pro- duce a single burst of nuclei in a short time, the crystal populations from such systems are monodisperse, making crystal size and crystal growth-rate determinations easier than in polydisperse systems Analcime itself is not a commercially used zeolite, however, it is a common by-product from sodium aluminosilicate zeolite synthesis systems, and is a stable end-member, since it does not redissolve in its synthesis solution to form more stable phases, as some other more useful zeolites do Consequently, it was found to be a convenient crystal phase to use to investigate the fundamental nucleation and crystal growth aspects of zeolite syntheses In this work, individual autoclaves were removed from the synthesis ovens, the syntheses were quenched, and the products were analysed ex sztu to study the features of the zeolite synthesis process which are reported here In recent investigations using the same batch composition and a synthesis temperature of 150OC19 it has been reported that bimodal crystal-size distributions formed, and were the result of extended ageing at ambient temperature The size of the two populations of crystals formed could be regulated by altering the ageing time of the mixed solution prior to synthesis at elevated temperature, with increasingly smaller crystals forming with increasingly longer ageing times In the current study, the ageing time was kept as short as possible (hours rather than days) and experiments were conducted at various temperatures to note the effect of changing that synthesis variable Experimenta1 Analcime was synthesised in this study nominally from the batch composition reported previously,19 which was given by A1,0, 84S10, 87Na20 2560H20 The reagents used in these experiments were aluminium wire (Aldrich, 99 999%), sodium hydroxide (Aldrich, 99 99%), Cab-0-Si1 (Eastech, 99 + YO), deionized water (Barnstead NANOpure 11, > 17 0 MR cm-'), puratronic silica (Johnson Matthey, 99 999%), tetraethyl orthosilicate (Aldrich, 99 999%), sodium silicate pentahydrate (Sigma) and sodium silicate non- ahydrate (J T Baker) Elemental analyses were not available for these reagents, but they were all research-grade materials and, therefore, were expected to have impurities present in trace amounts, as noted previously 22 The aluminium wire was dissolved in gently boiling sodium hydroxide and water in a HDPE Nalgene flask with a reflux condenser, a portion of the water and sodium hydroxide in the batch were used This step took approximately 1 h The remaining portions of the water and sodium hydroxide solution were used to dissolve the silica source, which was allowed to mix unheated overnight Each of the silicate solu-tions were prepared with identical compositions, neglecting any impurities which might have been present The two solutions were mixed at room temperature, followed by stirring for 15 min The entire mixture was then filtered through a 0 20 pm polysulfone membrane to remove particu- late matter A 'clear' aluminosilicate solution was produced following this procedure The solution thus prepared was distributed among 8 cm3 Teflon-lined steel autoclaves, which were sealed and placed in a convection oven at the desired temperature (130-160 "C) J Muter Chem, 1996, 6( lo), 1693-1699 1693 Individual vessels were removed from the oven at predeter- mined times and quenched in cold water to effectively stop the synthesis reactions The contents of the autoclaves were filtered through 0 20 pm polysulfone membranes, rinsed several times with deionized water, and dried at 80 "Cfor at least 4 h Products were weighed to determine a solid yield from each autoclave Powder X-ray diffraction was used to determine the crystalline phase present in each sample, with Cu-Ka as the X-ray source Observations were made on a JSM 840 scanning electron microscope, from which particle-size distributions were determined Results and Discussion The first aspect of this study was to determine the effects of changing the synthesis temperature on the analcime crystal growth rate This task was accomplished by monitoring the evolution of crystal sizes with time at synthesis temperature, using M-5 Cab-0-Si1 as the silica source In these experiments the water and sodium hydroxide were divided approximately equally between the aluminate solution and silicate solution during preparation The results of this part of the study are shown in Fig 1, in which the change in the linear dimension of the crystals is shown to have increased at a constant rate at each temperature The best lines drawn through the data points at each tempera- ture shown indicated that each crystal growth process began quite early in the synthesis, I e, each line passes through the origin, with the exception of the line passing through the 150°C data The crystal growth rates computed from the slopes of the respective lines are reported in Table 1 It was observed that the crystal growth rates were essentially constant during the times reported for the experiments in Fig 1 Zeolite crystal growth rates can be expected to depend on the batch composition, the solution pH and the reaction temperature While the temperature was constant, one might expect reagent concentrations to change during crystal growth owing to consumption of material from the clear solution However, the conversion of aluminium was only approximately 10% during these experiments, while all other reagents were present in significant excess Therefore, the driving force for crystal growth could be assumed to be essentially constant during the synthesis times noted in Fig 1 If the crystal growth rate can be assumed to depend on 80 0 5 10 15 20 25 synthesis time/h Fig.1 Change in the linear dimension of analcime crystals at various synthesis temperatures (A)130°C, (A) 140"C, (0)150°C and (B) 160 "C, using M5 Cab-0-Si1 in the standard batch composition Table 1 Temperature dependence of the linear growth rate of analcime synthesis temperature/"(= linear growth rate/pm h ~~ 130 23 140 50 150 72 160 110 synthesis conditions in the following way G=k(T)g(C,,C2,C3, , pH, ) (C, =composition n) that is, if the temperature effect can be isolated in a separate function, then the temperature dependence can be evaluated by computing an Arrhenius-type activation energy by plotting the natural logarithm of the crystal growth rate against the reciprocal of the absolute temperature A straight line fit of the data from the experiments at different temperatures was obtained, as shown in Fig 2 An activation energy of 75 kJmol-' was computed from the data obtained, which suggests a surface kinetics controlled rate-limiting process rather than a diffusion-limited process 23 Table 2 shows the value obtained from this work compared to values of the activation energy for zeolite crystal growth reported in the literature for several other systems While the value obtained here for analcime is slightly higher than some of the other reported values, it is certainly of the same order of magnitude, and clearly points to surface kinetics being the rate-determining step rather than diffusion Schoeman and co-workers16 l7 used the results of a chronomal analysis to demonstrate that the growth-limiting step in silicalite synthesis from a clear solution was a first-order surface reaction step Effects of the silica source In the following experiments with different silica sources, the solutions were divided in such a way that approximately 12% of the water and 3 5% of the sodium hydroxide were used to prepare the aluminate solutions, while the remaining water and sodium hydroxide were used to prepare the silicate solu-tion The final batch composition was the same as in previous experiments, and the combined solution mixing and filtration were conducted in the same way as before Syntheses were conducted at 160 "C Fig 3 shows the linear crystal growth curve for the first generation sf analcime crystals formed in this synthesis when M-5 Cab-0-Sil was used as the silica source While the evolution of the largest crystals occurred at a constant rate during the synthesis time shown, the linear growth rate was ' -21.50 I 2.30 2.35 2.40 2 45 2 50 103 KIT Fig.2 Plot of the logarithm of the analcime crystal growth rate us reciprocal absolute temperature to determine the activation energy for crystal growth Table 2 Comparison of activation energies for several zeolites zeolite system activation energy/kJ mol ref NaA 46 24 NaX 59 NaY 63 mordenite 46 25 NaX 63 26 NaA 44 27 silicalite 96 11 silicali te 45 16 analcime 75 this work 1694 J Mater Chew, 1996, 6(10), 1693-1699 0A-t 0 2 4 6 synthesis time/h Fig.3 Change in the linear dimension of analcime crystals at 160°C using four different silica sources in syntheses using the standard batch composition. Symbols represent the silica sources: (0)Cab-0-Sil, (A) puratronic silica, (0)sodium silicate nonahydrate and (+) sodium silicate pentahydrate. computed to be 13.4 pm h-l, i.e., slightly higher than in the previous experiments. It is conjectured that the preparation technique used in this series of experiments resulted in a slightly higher silica concentration in solution, increasing the driving force for growth by a small degree. Fig. 3 also shows the linear crystal growth curves for synth- eses carried out the same way as the Cab-0-Sil experiment, and with the same batch composition, but using different silica sources. It is obvious that one line drawn through the Cab-0- Sil data adequately fits all the data in the figure, and that the linear growth rate was the same for the crystals in all four synthesis batches.These results indicate that the driving force for zeolite crystal growth was the same in all four experiments, and was, therefore, a function of the material in the clear aluminosilicate solutions. Fig. 4 shows the increase of the yield of analcime with time at constant temperature in these experiments. Note that the mass of analcime increased rather rapidly at early times, and slowed down after about 20 h. The final yield was close to 5 g of analcime per kg of synthesis solution in all of the experi- ments, although there is some scatter around that value.For reasons that will be discussed below this scatter is to be expected, and it is more informative to examine the yield during the first few hours of synthesis. Note that the time scales in Fig. 3 and 4 are different, and that most of the mass of analcime added to the solid product occurred well beyond the time scale of the linear crystal growth results shown in Fig. 3. Between about 5 and 100 h of synthesis, the first generation of analcime crystals, which had grown to approximately 50 pm in dimension, settled to the bottom of the autoclave, and began to grow into a coalesced mass, losing their individual crystal identity, and another generation of nuclei formed in the solution phase above this settled popu- lation.Fig. 5 shows scanning electron photomicrographs of several samples taken from this experiment, some at early .” 0 20 40 60 80 100 120 synthesis time/h Fig. 4 Yield of analcime us. time using the same four silica sources, designated by the same symbols, as noted in Fig. 3 times, and some much later. Clearly the early samples contain one population of individual crystals, while the later samples showed evidence of a bimodal size distribution and subsequent agglomeration of larger crystals from the first generation. Fig. 5(d) shows that several of the smaller second generation of analcime crystals settled on top of the mass of agglomerated first generation crystals and began to become overgrown by the subsequent growth of the agglomerates, by a mechanism described elsewhere.28 Therefore, it might be expected that at later times, when the growing crystals have agglomerated randomly and lost their identity, the growth of analcime mass would be difficult to describe quantitatively and might proceed differently from one experiment to the next as noted in the previous paragraph.One would not expect the thermodynamic yield to be different in these experiments. However, focusing on the growth of the first population should provide infor- mation about the first nucleation event, that is the nucleation event governed by the constituents in the original synthesis solution. Fig. 6 shows the analcime yield during the first 6 h of synthesis for the same experiments reported in Fig.3 and 4. While the yield was rather difficult to determine experimentally, and especially so during the early periods owing to the small amounts formed, it is demonstrated that the yield using puratronic silica was lower than that for Cab-0-Sil, while the yield using sodium silicate pentahydrate was higher. The sodium silicate nonahydrate data are somewhat more difficult to evaluate, but that yield also became higher than the Cab- 0-Sil yield as time progressed. The individual crystals grew at the same rate, as shown in Fig. 3, but the yields were different during that time period. These results suggest that the number of crystals formed in each case was different. Table 3 shows values for the number concentration of crys- tals formed in several of these experiments, computed from the following relationship: (sample mass) =(analcime density) x (crystal volume) x (number of crystals) where the sample mass was that measured prior to 5 h of synthesis to account for only the first generation of crystals which formed.The uniformity of the crystal sizes made the use of a single average size to represent the whole population a very good assumption. In the control experiment using Cab- 0-Sil, for example, 5.19 x lo5 crystals (kg reaction solution)-’ nucleated in the first generation. This crystal number concen- tration is about six orders of magnitude less than the number concentration of silicalite crystals reported by Twomey et d.” in their clear solution syntheses, and their final crystal size was of the order of 1pm in dimension.Analcime and silicalite are fundamentally different zeolites, and their syntheses cannot be compared to any great extent, other than to note that these limited results indicate that, in these respective clear solutions, analcime nucleated far fewer crystals than silicalite. The number concentration of crystals nucleated using the other silica sources was slightly different than when using Cab- 0-Sil, as shown in Table 3. Thus, when using puratronic silica, the absolute number of crystals formed by nucleation was somewhat smaller (approximately one order of magnitude), but the driving force for crystal growth of individual crystals, i.e., the driving force for the surface kinetics step, was essentially the same in both cases.These results suggest that there is something inherent in the silica sources which affected zeolite nucleation in clear aluminosilicate solutions, but not zeolite crystal growth, since all other reagents were common to these experiments. These ‘things,’ which are unidentified at the moment, may be impurities or colloidal matter small enough to pass through the 0.20 pm membrane filter, and may promote nucleation in some manner as yet not completely determined. .~~Hamilton et ~1 reached similar conclusions in their study of J. Muter. Chem., 1996, 6(lo), 1693-1699 1695 Fig. 5 SEM photomicrographs of analcime crystals synthesised at 160 C from the standard batch using M5 Cab-0-Si1, at (a) 1 h and (b) 3 h showing one uniform sized population (c) Sample after 5 h at temperature shows a second population (d) Samples from the bottom of the autoclave with crystals from the second population overgrown by the agglomerated first population 1696 J Muter Chem , 1996, 6(lo), 1693-1699 -L 1.00 t .-0 c2 0.80 $ 0) 0.60 0)\U5 0.40.-)r Q) .-E 0.20 -0 az 0.00 0 2 4 6 a synthesis time/h Fig.6 Analcime yield from Fig. 4, but at early synthesis times, with the same symbols as in Fig. 4 Table 3 Nuclei concentration from various silica sources" number of nuclei formed/ silica source (kg reaction solution)-' Cab-0-Sil 5.19 x 105 puratronic 3.75 x 104 Na2Si03 .9H20 5.01 x 105 Na2Si0, .5H,O 1.04 x lo6 TEOSb 9.1 x 105 Cab-0-Silc 2.9 x 107 "Measurements made after 3 h of synthesis; only one population evident.bTEOShydrolysed for 10 days at room temperature. 'Ethanol added to the solution to mimic the equivalent amount generated owing to the hydrolysis reaction when using TEOS. zeolite NaX synthesis, and correlated their results to metal impurities in the silica source, but the variation in their results was greater with different silica sources. The fact that there were differences noted in this study, while the only parameter varied was the silica source, is at least consistent with their conclusions. It is unlikely that nucleation of analcime in these systems was by the classical nucleation mechanism.It is improbable that the second population of crystals was nucleated by that mechanism once the first population had grown to such a large size and settled to the bottom of the autoclave, thereby removing crystal mass from the solution and reducing the supersaturation. Of the order of 10% of the A1 was converted by this time, and far less of the silica was converted on a fractional basis (quantitatively, if 0.10 moles of A1 were con- verted, leaving 0.90 moles, then 0.20 moles of Si were converted, leaving 83.80 moles, based on the starting composition.) Twomey et al." physically removed the first population of silicalite crystals from the synthesis solution by filtration, and also observed that a second generation of nuclei formed in the remaining solution. It is more likely that impurities were present in sufficient concentration to catalyse the nucleation of the second population of analcime crystals.Effects of using tetraethylorthosilicate Another source of pure silica is tetraethyl orthosilicate (TEOS), carried out at room temperature, with stirring, for 10 days, after which time the synthesis mixture was prepared as before. A modified Cab-0-Sil synthesis solution was prepared by adding an equivalent amount of ethanol to simulate the by- product of the hydrolysis reaction with TEOS, and to serve as a control. It should be noted that in each of these four examples, two separate liquid phases persisted, only one of which supported the zeolite precipitation reactions.Fig. 7 shows the evolution of the maximum analcime crystal size with time during the syntheses described above. The maximum crystal size was measured in these experiments, since there was a rather broad crystal size distribution formed in these systems. The linear growth rate in the first 6 h in the Cab-0-Sil-ethanol system was 4.5 pm h-l, while the crystal growth rate in the TEOS system (considering the first four data points) was 1.4 pm h-l. These initial growth rates were appreciably slower than the analcime growth rates noted in the experiments without ethanol. The growth in the Cab-O- Sil-ethanol system slowed considerably after the first 10 h owing to settling, and a second population formed. The growth rate in the TEOS system appeared to remain essentially constant for up to 50 h, even though settling had occurred, and a second population formed.Fig. 8 shows the yield from the two systems just described. Several features are noted in comparison to Fig. 4 and 6. First, the two systems containing ethanol both exhibited a longer induction time than in the absence of ethanol. This result could be simply a dilution effect due to the partitioning of silica between the two liquid phases, and would not be expected to be due to lowered solubility of aluminosilicates, since there was no turbidity observed owing to gel formation, or zeolite, since that would have increased the yield. It appears from the data in Fig. 8 that the yield may have continued to increase in the Cab-0-Sil-thanol system at longer synthesis times, but we have no way of predicting the final yield in that system.100 1 1 80 c + +I+El 0.-+Otv) 0' 1 0 20 40 60 80 100 synthesis time/h Fig. 7 Change in the linear dimension of analcime crystals synthesised at 160 "C with Cab-0-Sil and added ethanol and tetraethylorthosilic- ate, using the standard batch composition. Symbols designate silica sources: (0)Cab-0-Sil with added ethanol; (+) TEOS. 7 I h c .-0 c3-3-28a liquid silica source which can be hydrolysed to yield Si02, but which then contains ethanol as a by-product. When the hydrolysis reaction was allowed to occur in situ with the synthesis reaction, i.e., simultaneously with the synthesis, the final zeolite synthesis product contained predominantly unidentified impurities, suggesting that the hydrolysis was incomplete, and that the solution composition was not the same as in the previous cases.When the hydrolysis was carried out in a separate vessel overnight at room temperature and used in the synthesis, rather slow conversion to analcime was observed using the standard synthesis conditions. There was virtually no difference noted when the hydrolysis reaction was 0,Y Y 0 20 40 60 80 100 120 140 synthesis time/h Fig. 8 Analcime yield from the experiments noted in Fig. 7, with the same symbols as in Fig. 7 J. Mater. Chem., 1996,6( lo), 1693-1699 1697 The yield from the TEOS system appeared to have stopped increasing after about 40 h, and to have produced only slightly more than 0 4g of analcime per kg of reaction solution, about an order of magnitude less than in the systems described in Fig 4 and 6 In spite of the reduced yield from the TEOS system, it was rather dramatic that the particle size from that system was the largest noted in all the expenments conducted This statement must be qualified by indicating that a much broader crystal- size distribution was observed in these experiments than in the ethanol-free systems, so these particle dimensions clearly do not represent the average crystal size The reason for the broader size distribution was probably that the silica in these expenments was distributed between two liquid phases, and that silica was transferred to the aqueous phase as the synthesis proceeded Thus, this system behaved more like a gel system in which nucleation typically occurs over a longer time as the gel particles dissolve, leading to a broader crystal-size distri- bution The identity of the first population was lost when settling and overgrowth occurred, i e, it is very likely that in the ethanol-free system, crystals would have grown to a much larger size had they not formed an agglomerated mass at the bottom of the autoclave Table 3 also shows the crystal number concentration for these experiments, computed as before from the yield and the average crystal size before the formation of the second popu- lation It appears that the systems with ethanol present nucleated more crystals than the ethanol-free systems Among the various explanations for this result (e g ,impurities, silicate concentration, reduced zeolite solubility) the most reasonable might be to suggest that the number of nuclei formed is related to the Al/Si ratio, and that that ratio would be higher in the liquid phase in which the precipitation occurred compared to the ethanol-free system, owing to partitioning of the silica between the liquid phases This result was not expected, and will be the subject of continued investigation While it may seem inconsistent that the TEOS system, which exhibited the lowest yield and produced the largest crystals, also nucleated a rather large number of crystals, it is not fair to compare the results of the ethanol-free systems to these, especially at long synthesis times The broad crystal-size distri- bution compared to the unimodal crystal population in the ethanol-free systems, and the settling, overgrowth and sub- sequent nucleation of new populations cloud the comparison It is only realistic to compare these systems at very early times, when the yields are low and only one population existed which is, in fact, how the crystal number concentrations were deter- mined Therefore, one must conclude that the nucleation levels reported in Table 3 are representative of the earliest stages of these processes Conclusions Synthesis studies of the molecular sieve analcime have been conducted Syntheses carried out at various temperatures demonstrated that analcime crystal growth was constant in the early stages of synthesis The temperature dependence of the linear crystal growth rate in the range 130-160°C was calculated to be 75 kJ mol-l, a value consistent with prior values computed for other zeolite systems, and one which suggests that the limiting step in the growth process is surface kinetics rather than diffusional transport limitations Syntheses carried out with four different silica sources at 160 "C showed that the linear crystal growth rate was the same for all silica sources tested These results indicated that the driving force for growth was the same in these experiments, and probably had to do with the silicate anion oligomer distribution in the solution, which has been reported previously not to vary with the silica source22 The yields of analcime from these expenments were different, however, and pointed 1698 J Muter Chem, 1996, 6(10), 1693-1699 to different nucleation rates in each system These differences were suggested to stem from materials inherent in the silica sources, since the water, sodium hydroxide and alumina used in all of these experiments were the same Syntheses with the same batch composition, but using a hydrolysed silicate in which ethanol was a by-product, showed that the linear crystal growth rates were lowered, as were the final yields Nucleation rates in these systems, however, were reported to be greater, most likely owing to the partitioning of silica between two liquid phases and the increased Al/Si ratio This matter is still being investigated, however Typical of this system was the observation that a population of crystals was nucleated very early in the process and in a very short time period, resulting in a unimodal size distribution As these crystals grew, they settled to the bottom of the reaction vessel, where overgrowth occurred, as reported pre- viously,28 forming a membrane-like mass of agglomerated crystals Subsequently, a second population of crystals formed in the solution above the agglomerating mass These, too, eventually settled to the bottom and became incorporated in the agglomerating analcime mass This is perhaps the second report, the other being in ref 11, of formation of a second distinct population of zeolite crystals nucleated late in the synthesis in an unstirred system A similar observation was made in a stirred system," but was attnbuted to a possible secondary nucleation mechanism involving collisions References 1 D W Breck, Zeolite Molecular Sieves Structure Chemistry and Use, John Wiley & Sons, New York, 1974 2 A D Randolph and M A Larson, Theory of Particulate Processes, Academic Press, San Diego, 2nd edn ,1988 3 S Ueda and M Koizumi, Am Mineral, 1979,64,172 4 S Ueda and M Koizumi, Am Mineral, 1980,65, 1012 5 S Ueda and M Koizumi, Proc First Int Symp Hydroth React, 1983,p 695 6 S Ueda, T Sera, Y Tsuzuki, M Koizumi and S Takahashi,J Clay Scz , 1983,23,60 7 S Ueda, N Kageyama and M Koizumi, Proc 6th Int Zeol Conf, ed D Olsen and A Burio, Butterworth, Guildford, 1984, p 905 8 P Wenqin, S Ueda and M Koizumi, Proc 7th Int Zeol Conf, ed Y Murakawi, A Iijima and J W Ward, Kodansha/Elsevier, Tokyo/Amsterdam, 1986, p 177 9 J C Jansen, C W R Engelen and H van Bekkum, ACS Symp Ser ,ed M L Ocelli and H E Robson, American Chemical Society, Washington, DC, 1989,398,257 10 C S Cundy, B Lowe and D Sinclair, J Cryst Growth, 1990, 100,189 11 T A M Twomey, M Mackay, H P C E Kuipers and R W Thompson, Zeolites, 1994,14, 162 12 B J Schoeman, J Sterte and J-E Otterstedt, J Chem SOC Chem Commun , 1993,994 13 B J Schoeman, PhD Dissertation, University of Goteborg, Sweden, 1994 14 B J Schoeman, J Sterte and J-E Otterstedt, Zeolites, 1994, 14, 110 15 B J Schoeman, J Sterte and J-E Otterstedt, Zeolites, 1994, 14, 208 16 A E Persson, B J Schoeman, J Sterte and J-E Otterstedt, Zeolites, 1994,14, 557 17 B J Schoeman, J Sterte and J-E Otterstedt, Zeolites, 1994, 14, 568 18 B J Schoeman, J Sterte and J-E Otterstedt, Stud Surf Sci Catal, No 83, 1994, Elsevier, Tokyo 19 F DiRenzo, R Dutartre, P Espiau, F Fajula and M-A Nicolle, Crystallization of Zeolites in the Microgravity Environment of Space 8th Euro Symp on Muter & Fluid Sci in Space, Brussels, April 12-16,1992, ESA SP-33, p 691 20 A A Brock, G N Link, P S Poitras and R W Thompson, J Muter Chem ,1993,3,907 21 F DiRenzo, F Fajula, P Espiau, M-A Nicolle and R Dutartre, Zeolites, 1994,14,256 22 K E Hamilton, E N Coker, A Sacco, Jr, A G Dixon and R W Thompson, Zeolites, 1993,13,645 23 R W Thompson, in Molecular Sieves-Science and Technology, 24 25 ed.H. Karge and J. Weitkamp, Springer-Verlag, Heidelberg, 1996, ch. 1, p 23. D. W. Breck and E. M. Flanigen, Molecular Sieves, 1968, Society of Chemical Industry, London, p. 47. D. Domine and J. Quobex, Molecular Sieves, 1968, Society of 27 28 S. P. Zhdanov and N. N. Samulevich, Proc. 5th Conf. on Zeolites, 1981, Heyden, London, p. 75. S. Gonthier and R. W. Thompson, in Advanced Zeolite Science and Applications, ed. J. C.Jansen, Elsevier, Amsterdam, 1994, p. 43. 26 Chemical Industry, London, p. 78. S. P. Zhdanov, Ado. Chem. Ser., 1971,20,101. Paper 6/02429G; Received 9th April, 1996 J. Muter. Chem., 1996,6( lo), 1693-1699 1699

ISSN:0959-9428    

DOI:10.1039/JM9960601693

出版商:RSC    

年代:1996

数据来源: RSC

摘要:

Hydrothermal conversion of amorphous NiFez -,Al, (OH)* into crystalline phases Emilia Wolska," Wlodzimierz Wolski, Janusz Kaczmarek, Pawel Piszora and Wojciech Szajda Department of Magnetochemistry, Adam Mickiewicz University, Poznari, Poland Mixtures of amorphous hydroxides prepared from Fe"', Al"' and Nil' nitrate solutions in proportion to obtain a series of compounds with 0.1 (if necessary 0.05) increments for x in NiFe,-,Al,(OH),, have been subjected to hydrothermal ageing at 150"C in mother-liquor for one or six months. X-Ray, IR, TG and magnetic studies revealed no influence of ageing time on the properties of the crystalline phases obtained. It was proved that for certain values of x, it is possible to obtain hydrothermally a pure phase of nickel ferrite-aluminate solid solutions, displaying magnetic characteristics similar to those of solid solutions produced by standard ceramic methods.At x =0.5 the spinel phase deteriorates, the haematite structure appears and the nickel ions engage (till x =2) in the formation of layered double hydroxides (LDHs), accompanied from x =0.8 by the boehmite phase. At x = 1.8 the spinel phase appears again, having a lattice parameter corresponding to 0.72 A13+ in the ferrite structure. Among the large variety of divalent nickel hydroxides, two modifications can be distinguished clearly: the unhydrated hexagonal brucite-type P-Ni(OH), with the lattice parameters a =3.12 and c =4.60 A, and !he hydrated hexagonal a-Ni(OH), with a=3.08 and c=8.09 A.' Different routes to a-Ni(OH), synthesis2-' and its stabilization are known., The hydrated hydroxide may be considered as the basic framework of a number of layered compounds having, in connection with iron and aluminium ions, mineralogical representatives: re- evesite, Nio ,'Fe, 25(OH)2(C03)0 125( H2O)O', and eardleyite, Nio 25(OH)2(C03)0125(H,O)o '.They both belong to a large family of compounds, the so-called layered double hydroxides (LDHs), wbich are related by similar lattice param- eters (a=3.1, c =23.4 A appr~ximately)~and are characterized by a great tolerance to a wide range of radii of the substituting me@ ions, accepting even an astonishing difference of about 0.5 A.' In this respect they exceed by ten times the tolerance encountered in hydroxy double salts (HDSs).' Cations with dissimilar valences are incorporated isomorphously, and the hydrated nickel hydroxide balances the electrostatic equilib- rium by the insertion of a variety of organic and inorganic anions, as well as water molecules in its van der Waals gallery.'*1° In nature these anions are most commonly carbon- ates, as for example in the case of Ni2+/A13+, in takovite, Ni,A12(0H),,(C0,)(OH).4Hzo,a synthetic counter-part of which has been obtained hydrothermal1y.ll Minerals of the LDH family with layers of [M2+1 -yM3+y(OH)2]"+ macrocations usually have the y =0.25.For the Ni2+ /A13 couple, a monophasic LDH system + can be obtained in the laboratory, according to the published procedure, for 0.2 dy d0.33,12 or for 0 <y <0.4,13 and there are reports of the preparation of an LDH structure with y= l.14 The Ni2+/Fe3+ couple in turn yields LDH phases with carbonates and water in the interlayer region in a range as large as 0.75 3 y2 0.25." In our investigations into the conversion of hydrothermally treated, originally amorphous NiFe, -,Al,(OH), precursors with Odxd2, (i.e.with a constant M3+/Ni2+ ratio of 2), we created conditions which were rather unfavourable for the formation of the LDH structure as a simple crystalline phase, which instead promoted the transformation of the system into the spinel structure, this is a very successful low- temperature route to defect, cation-deficient ferrites from 2Fe"'M"(OH), .',"' However, as the results show, when A13+ ions are present in these hydroxide precursors, the conversion pathways are somewhat unpredictable.Experimental In order to obtain the NiFe,-,Al,(OH), precursors with x increments of 0.1 (up to x=O.5 with 0.05),sodium hydroxide was added dropwise with constant stirring to water solutions of 0.1 mol dm-3 Ni"-0.2 mol dm-3Fe'1'-0.2 mol dm-3 Al"' nitrates until pH 10 was reached. Teflon vessels containing the suspensions in the mother-liquor were placed in autoclaves and maintained at 150 "C for one to six months. The products, washed after ageing till a negative reaction for Na+ and NO3- was obtained, were dried at 30 "C for 24 h. After being crushed and ground, they were kept in phials for further examination. To avoid carbon dioxide all operations were performed in an N2 chamber.X-Ray powder diffraction (XRD) studies were carried out on a TUR-61 diffractometer (Co-Ka radiation), equipped with a proportional counter spectrometer joined to the counting components and impulse printing unit. Magnetic measure-ments were made on a magnetic balance for ferromagnetics designed in our laboratory, thermomagnetograms were regis- tered on a Gear X-Y recorder. IR spectra were obtained on Perkin-Elmer 180 spectrometer, thermal analyses were carried out on a Shimadzu TGA 50H instrument, in air, at a heating rate of 10°C min-l. Metal ion contents were determined by atomic absorption spectrometry, unwashable nitrates were determined by the colorimetric method with phenoldisulfonic acid at ,I=410 nm.Results and Discussion As the X-ray analysis showed, one month and six months storage of the originally amorphous NiFe, -,Al,(OH), series of compounds with increasing A13 concentration led, under + the applied conditions, to identical crystalline products of similar crystallinities and similar relative intensities of reflec- tions. Therefore, we refer hereafter to the samples aged for one month and, unless stated otherwise to 0.2mol increments of x only. The XRD patterns shown in Fig. 1 indicate that for x=O and for low A13 concentrations the hydrothermal treatment + results in the formation of a pure spinel phase. This phase remains dominant up to x=O.4; for x>O.4 the nickel ions begin to construct a hexagonal layered double hydroxide framework structure, which is most noticeable by its prominent reflection at 28~13.45".An increase in the amount of A13+ ions by 0.1 mol A13+, i.e. to x=O.5, drastically hampers the further development of the spinel phase, while iron ions form simultaneously the hexagonal haematite lattice. Judging by the d-spacings of this lattice, aluminium ions are not incorporated in it. By maintaining a constant primary mole ratio, A3+/(Ni2+ +A3+)=2/3 (A3+ =Fe3+, A13+), with steadily increasing A13 + concentration, the LDH phase structure is perfected. Note that up to x=O.6, no traces of a separate aluminium phase can be observed (this is also true for x =0.7). J. Muter. Chern., 1996, 6(lo), 1701-1707 1701 8.366o L Fig.1 XRD patterns of crystalline phases built up during hydrother- mal storage of amorphous NiFe2-,Al,(OH),, x=0.6 (a), 0.5 (b),0.4 (c), 0.2 (d),0 (e). LDH, layered double hydroxide structure; S, spinel phase; H, haematite structure. The question of whether the A13+ ions are isomorphously incorporated into the nickel ferrite or not, is settled in Fig. 2 by the peaks in the XRD patterns, which indicate the positions of the spinel structure triad with Miller indices (422), (511) and (440). As the A13+ concentration increases, the peaks are displaced progressively to higher angles, revealing a decrease of the unit-cell parameter. The variation of this parameter together with the changes in saturation magnetization, as diamagnetic A13+ ions replace Fe3+ ions in the octahedral sublattice, is shown in Fig.3. The slope of u=f(x) up to I I I I 11 I I I 111 I I I I 1 111 I I I 62 64 66 68 70 72 74 76 78 2Bldegrees Fig. 2 Sections of X-ray diffraction patterns illustrating the limit of spinel phase formation: x=0.50 (a), 0.40 (b),0.30 (c),0.20 (d), 0.15 (e), 0.10 (A,0.05 (g),0 (h) 1702 J. Muter. Chem., 1996,6( lo), 1701-1707 5--10 5-0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 x in NiFe,,AI,(OH), Fig. 3 Lattice constant a of spinel phase, and saturation magnetization 0(20"), vs. x in the NiFe2-,Al,(OH), precursor. Dotted line: a=f(x) for NiFe2-xAl,04 [a(NiA1204)= 8.05 A], after ref. 19. x=0.4 almost corresponds to the Vegard line joining the lat- tice parameter of nickel ferrite with that of nickel aluminate, NiAl,O,, obtained at high temperature (1400 OC).19 This sug- gests that nearly all of the A1 is engaged in the formation of a solid solution of nickel ferrite-aluminate.The larger unit-cell parameters of ferrites, shown in Fig. 3, have been observed also in other ferrites prepared hydrothermally.20 A plausible explanation of this anomaly is offered by the TG curves (see later). The slope of the 0=f(x) curve in Fig. 3 fits nearly exactly that observed for the sintered NiFe,-,Al,O, series21 with A13+ in B sites. It should be added that for 0 dx ,<0.8 such a cation distribution in NiFe2-,A1,0, has been determined by X-ray diffraction, neutron diffraction and Mossbauer spectroscopy,22 and confirmed using the equilibrium conditions of ref.23 and the cation distributions for pure NiFe,O, and NiA1,04.24 A plot of the Curie points us. increasing x gives almost a straight line as shown in Fig. 4. This is another manifestation of nearly total incorporation of the available A1 into the nickel ferrite lattice, provided that x does not exceed 0.4. The inset in Fig. 4 explains the question mark posed by x=0.5: during the recording of the thermomagnetogram for this composition, the magnetization suddenly increased at about 430 "C, instead of decreasing, as the temperature rose. This might seemingly 3/"1560 470 460 ? I I T Fig.4 Dependence of the Curie point on x in NiFe,-,Al,(OH),. Inset: anomaly for x=0.5 reflect the P-type of curve forseen by Nee125 when one cationic sublattice loses its magnetization faster than the other.In our case the reason is much more simple: as the temperature increases, the thermally unstable LDH phase decomposes and finally divided nickel oxide26 supplies the material for the easy formation of an additional portion of ferrimagnetic spinel phase. Presenting the results of the IR investigation on ferrites, it should be remembered that, although the first spectrum of nickel ferrite with its two characteristic absorption bands at ca. 600 and ca. 400 cm-' was published more than four decades the full lattice vibration set of the four modes forseen by the factor-group analysis,28 was demonstrated only recently.29 The other two vibrations lie at 186 and 146cm-' and are scarcely visible29 even when the spectrum is carried out on a sintered sample.For the hydrothermally obtained preparations, these bands are not visible at all. Nonetheless, Fig. 5 shows an (admittedly small) shift of both bands towards higher wavenumbers, as should occur in the case of alum in ate^.^' An explanation of what kind of swelled spinel lattice is built hydrothermally is offered by the traces of the TG curves of Fig. 6. In the region of O<x G0.4 the plots of mass loss us. temperature are similar. Using the x=O curve as an example, three sections can be distinguished. The first one belongs to the physically adsorbed H20, the second to the remainder of the loosely bound water and water held partially by hydrogen bonding on the surface of crystallites. The last stage, responsible for the swelling of the cell, belongs to OH groups (evolved as H,O), randomly substituting 0,-ions in the anionic sublattice, engendering the defect ferrite structure.In the cases of x=OS and 0.6, the TG curves show steep slopes for the first two sections, reflecting a sudden enrichment of the preparations with evolving matter (ca. 15 and ca. 19%, respectively). The first section represents the loss of physisorbed H20, the second, up to ca. 400"C, the loss of interlayer water, followed at the higher-temperature end of this section by the loss of interlamel- lar nitrates of the LDH phase and by the dehydroxylation of hydroxide slabs.The steady loss of mass above 500°C corre-sponds to the loss of OH groups both from nickel oxide and from the haematite phase, which under the applied conditions is in fact a hydrohaematite pha~e.~'?~~ When the A13+/Fe3+ ratio is ca. unity, strongly antagonistic relationship between these cations is found. Such behaviour was noticed previously in the AlFe(OH), system,33734 and a similar case is shown in Fig. 7. For x=O.8, besides a satisfac- torily expanding LDH structure, there is a scarcely developed haematite phase and only minor traces of the boehmite phase, identifiable by the prominent (020) reflection of that structure, v1crn-l Fig. 5 Main IR absorption bands of spinel phases for x=0.0, 0.2 and 0.4 in the NiFe,-,Al,(OH), precursor [no v,(NO,-) detectable] 0.00--1.00 -\r: am.c2 -2.00-0 v)' 3. EF -3.00-4.00. 0.00 500.00 1OOO.00 TI'C Fig.6 TG curves during sample heating for given x in NiFe2-,A1,(OH), (see text) B 320)I LDH 10 20 30 40 50 60 70 80 2Bldegrees Fig. 7 XRD patterns of phases formed for x= 1.4 (a), 1.2 (b), 1.0 (c) and 0.8 (d) in NiFe,-,Al,(OH),. LDH, layered double hydroxide structure; B, boehmite; H, haematite structure. here at very low intensity. The least crystalline system, apart from the LDH phase was recorded for x= 1.2, whereupon the boehmite phase begins to appear. Unexpectedly, for x =1.8 (Fig. 8), when Al/Fe=9, the reflections of the spinel phase reappear at (lll), (220), (311), (400) and (422).0These all fit very well into the Fd3m unit cell with a =8.23 A.This value corresponds to x =0.72 in the nickel ferrite-aluminate NiFe, -xAlx04 obtained by standard ceramic technique^.^^ The cation distribution between the tetrahedral and octahedral positions at this A1 concentration (0.72) in the lattice should cause a magnetization of nearly zero,22 as was found in this J. Muter. Chew., 1996,6(lo), 1701-1707 1703 B (020)/ i LDH I I LDH 1,1,1,1,1,1,1,~ 10 20 30 40 50 60 70 80 2Udegrees Fig. 8 XRD patterns of phases formed for x =2 0 (a), 18 (b)and 16 (c) in NiFe,-,Al,(OH), Astensks denote spinel phase, other notation as in Fig 7 case It is necessary to mention that in the compounds which underwent six months storage, spinel structure reflections were alsp detectable for x= 18, mplying an a parameter of ca 8 23 A Comparing Fig 7 and 8, one cannot overlook that coinci- dently with the disappearance of haematite phase at x= 18 and the engagement of iron ions in the rebuilding of the spinel structure, the developing (021), (130) and ( 150) reflections of the boehmite phase appear suddenly Comparison of the a, b and c parameters of the boehmite phases formed in the experiments presented, with some reference data (Fig 9), shows that our preparations display somewhat elevated a and c parameters (this will be discussed later) The a parameter of LDH phases existing in the ageing products from x=O4 (traces) too x=20, changes slightly but steadily with x (3 03-3 04 A) The c parameter, however, seems to increase more markedly with increasing x (Fig 10) Except for x =O 8, where the boehmite phase is just only detectable by the appearance of the (020) line, the TG curves pore boehmne after references[ I12l225261 $122319 [391 --12 201 5 0370 -13711381 --3 69 -1361--I3711381 1391 1'0 ' 1'2 ' 1'4 ' 1'6 ' 1'8 ' $0 x in NiFe,,Al,(OH), Fig.9 Unit-cell parameters of boehmite phase as a function of x in ongnal NiFe, -,Al,(OH), 1704 J Muter Chem, 1996, 6(10), 1701-1707 23 23 23 23 23o_a 22 228' 31 a I-%-YY 30 1104 0'6 I 0'8 I 1I 0 1'2 1'4 ' 1'6 ' 1'8 I 2'0 x in NiFe, ,Al,(OH), Fig.10 Vanations of a and c parameters of LDHs with changing x in NiFe, ,Al,(OH), for the region 1,<x ,<2 (Fig 11) may be divided into four sections up to ca 240 "C, between 240 and 390 "C, 390-570 "C and above 570°C (the maxima of transformation rates on DTG curves, e g for x = 1, are 64, 274, 371 and 543 "C, while for x=2, which is an exception, there are only three 231, 381 and 561°C) In the first section, as well as the evolution of physisorbed H,O from all phases present, there is partial loss of the LDH interlayer species, in the second section the total destruction of the hydroxide layers of the LDH structure occurs, the third section represents the decomposition of the boehmite phase, and in the fourth region the steady disappear- ance of the temperature-resistant OH groups from nickel oxide, \ --Fig.11 TG curves of samples formed during storage of N1Fe2-,A1,(OH), for x=O 8 (a), 1 0 (b), 1 2 (c), 14 (d), 1 6 (e), 18 (f)and 2 0 (g) iron oxide and alumina occurs. The mass loss increases from about one fifth for the sample with x =1 to one quarter of the total mass heated up to 1000"Cfor x =2. As mentioned previously, the washing of preparations was repeated until the supernatant tested negative for Na+ and NO3-. However, after dissolution in sulfuric acid samples within the region of 0.4<x <2 did display unwashable nitrates (found for 0.4 <x <2 increasing by 0.2: 0.01, 0.20, 0.23, 0.62, 0.72, 0.70, 0.63, 0.45 and 0.40 mass%, respectively) and only negligible traces of carbonates. Thus the distinctive IR band at ca.1360cm-I in Fig. 12 and 13 belongs to nitrates (v3). Because of the partial overlapping of the vibration bands and the poor crystallinity of some phases, the spectra of 0.5<x <2 look complicated. To make them more readable the vibrations belonging to the best crystalline and the most rich in bands belonging to the boehmite phase stemming from the sample with x=2 are compiled separately in Table 1. None-theless, the presence of either the haematite phase, the absorp- tion bands of which change their positions with x in a-Fe, -x,3(OH)x03-x,43.44 or the ferrite aluminate phase, results in the deformation of the bands of the LDH and boehmite phases present within the 300-8OOcm-' region. Thus, in addition, the positions of the peaks belonging to the LDH structure, determined with the help of ref.5, 10, 45, have been marked on Fig. 12 and 13 with dotted lines. In the IR spectra, NO3- ions with D,, symmetry start to 4000 3000 2000 1600 1200 800 400 vlcrn-' Fig. 12 IR spectra of products formed during hydrothermal storage of NiFe,-,Al,(OH), for x=0.5 (a), 0.6 (b),0.8 (c), 1.0 (d) and 1.2 (e). LDH, layered double hydroxide structure; B, boehmite phase; H, haematite phase. 4000 3000 (ea)2000 1600 1200 800 400 vlcm-' Fig. 13 IR spectra of products obtained after storage of the NiFe2-,A1,(OH)I, precursor for x= 1.4 (a), 1.6 (b),1.8 (c) and 2.0 (d). Notation as in Fig. 12 and Table 1. become detectable from x =0.5 and reach highest abundance when x= 1,1.2 and 1.4 (Fig.12 and 13).Then, with the increase of the ratio of boehmite phase to LDH (Fig. 14), their presence gradually decreases (Fig. 13). Whether the signals at ca. 1100 cm-' (x=O.6) and ca. 1080 cm-' (x=O.8) in Fig. 12 are indicative of boehmite phase formation or if it is a weak vibration belonging to vl(NO3-) is hard to answer, since in the system under study there is no selective reagent to dissolve one phase without causing a destructive effect on the other. This fact creates, unfortunately, an unsurmountable stumbling block in defining the LDH phases chemically and in giving an unambiguous answer as to if the LDH phase is the only one containing nitrates. The co-produced boehmite phase, as a consequence of an unfavourable Ni2+ /A13+ ratio for the formation of the pure hydrotalcite-type structure, has been noted alread~,~~.~~ but without comment about any anomalies in that phase, such as for example the elevated a and c parameters encountered here. Except for the last sample (x=2), it might be the iron and/or nickel ions entering the octahedral sheets of the boehmite structure that elongate these two parameters.If this were to happen, we would rather expect an influence on the b parameter as well. Secondly, although A13+ ions may be accepted in a limited mole fraction into the isomorphous lepidocro~ite,~~,~~ there are a number of report^^'-^, showing that boehmite does not fom solid solutions with iron ions. The substitution of Ni2+ for A13+ would implicate the compensation of charge, which is not excluded but is less probable.Fortunately, it has been established in ref. 53 and 54 on a large experimental study supported by a computer simulation, that not only the (020) reflection (as was believed), but also the (021) line and the (150)/(002) doublet may be displaced towards lower 28 values with decreasing crystallite size, causing larger apparent d values. The crystallites of our boehmite phases are well developed in the y direction, giving the literature-like b param-eters, but are poorly developed in the xz-sheets, causing larger apparent a and c parameters, as in Fig. 9. Although it is not possible to define our LDHs by chemical formulae, there are some hints that help to describe the obtained species. The large number of experimental points for the a and c parameters, consequently demonstrating the trends with the variable x, corroborate the IR results, allow the following conclusions to be drawn, after remembering first, very briefly, the history of claims and counter-claims about Ni-Ni distances responsible for the a parameter in the well crystallized, unhydrated P-Ni(OH), and poorly crystallized a-Ni(OH), with intercalated water molecules in the van der Waals g!p.Bode et al.' found a small Ni-Ni contraction (ca. 0.05 A) within the Ni(OH), layers of the hydrated a-modification. McEwen's3 results Eontradicted this and EXAFS results55 showed that the ca. 0.05 A contraction is not apparent. The supposed contraction is caused by the hydrogen bonding between the hydroxide groups of the layers and the interlamel- lar woater.Thus taking the data from Bode, a= 3.1? and c= 4.60 A for the p-form and a= 3.08 and c/3 =KO9 A for the cl-form, and comparing theom with Braconnier'~~ a:-hydrated form with a =3.08 f0.01 A and c =23.41 &0.05 A, we see (Fig. 10) t$at our a parameter is still contracted by about 0.04-0.05 A. Feitkne~ht~~described a layered species obtained from Ni2+/A13+ chlorides with a stoichiometry Ni2+ :A13+= 4: 1, which had the parameters a =3.02 and c/3 =7.8 A.Jn our last sample (x=2) we have a=3.04 A and c/3=7.81 A. It is then evident from our XRD results that the supplementary contraction of a parameters in the LDH phases is caused by the presence of smaller, statistically distributed (Fe,A1)3 ions+ or A13+ ions only in the octahedral sites of the cationic layers in materials in somewhat higher A13+/Ni2+ ratios than in ref.56. Charge neutrality is preserved by the intercalation of nitrates in the interplanar spacing, distributed among the molecules of gallery water. As the Fe3+ ion concentration J. Muter. Chem., 1996, 6(lo), 1701-1707 1705 Table 1 IR absorption bands of boehmite phase for x=2 wavenumber/ cm-' symmetry species 3300 Bzu 3100 \ B3u 2150 1970 1155 1063 755 tJ B2U B3U BIU 6, ?OH } 650 410 365 323 B3u Bzu B3U B1" 1 J "After ref 40 bAfter ref 41 (with the help of ref 42) "15 -30-8 v r s 5-(0 g 20-2 m s 5-s -10-5-c.0-U 1'0 1'2 1'4 1'6 1'8 2'0 xin NiFe, xAIx(OH), Fig.14 Boehmite to LDH phase ratio us x in NiFe, xAlx(OH), determined from integrated intensities of X-ray reflections decreases and the aluminium ions have greater chance to build their own phase, y-AlO(OH), the A13+/Ni2+ ratio diminishes and the a parameter in LDH increases slightly This process is accompanied by a diminishing requirement for the charge- compensating anions in the gallery The electrostatic forces between the positive macrocationic layers and the negatively charged interslab layers weaken, so that the distances between them increase, leading to an increase of the c parameter The linkage between OH groups of positively charged host layers and the accumulated H,O molecules in the interlamellar space weakens with increasing x Fig 12 and 13 show a steady displacement of vOH from 3440 (x=O 6) to 3550 cm-' (x=2) But it is still at a lower wavenumber (by 100 cm-') than that of P-Ni(OH),, in which the sharp OH stretching band is not deformed by hydrogen bonding In the system studied, in the light of earlier reasons, we are able to define our LDH species only by a generic for- mula { [Ni2+1-z( Fe,A1)3fZ(OH)2]z+(zN03- yH,O)} mH,O, for 0 4 <x <2 and -ZA13+z(OH)2]Z+(~N03-yH,O)) rnH,O, for x=2 in the amorphous NiFe, ,Al,(OH),, where the entities in square brackets represent the host layers as macrocations and the round brackets encompass the guest constituents The authors gratefully appreciate the grant (no 3 T09A 064 08) from the State Committee for Scientific Research (KBN), which allowed the realization of this project References 1 H Bode, K Dehmelt and J Witte, Electrochim Acta, 1966, 11, 1079 2 S Le Bihan, J Guenot and M Figlarz, C R Acad Sci C (Paris), 1970,270,2131 1706 J Muter Chem, 1996, 6(10), 1701-1707 assignment" assignmentb mode s3} OH stretching modes hydrogenic mode S, (probably coupled modes of as yet unspecified participants) OH deformation hydrogenic mode S, hydrogenic mode S2 torsion hydrogenic mode S6 -one of v3 A10, (Flu) v4 '410, (Flu) skeletal modes of A1-0 layers A1-OH 3 R S McEwen, J Phys Chem, 1971,75,1782 4 J J Braconnier, C Delmas, C Fouassier, M Figlarz, B Beaudouin and P Hagenmuller, Rev Chim Miner, 1984,21,496 5 P V Kamath and G N Subbanna, J Appl Electrochem, 1992, 22,478 6 P V Kamath, J Ismail, M F Ahmed, G N Subbanna and J Gopalakrishnan, J Muter Chem ,1993,3,1285 7 H F W Taylor, Mineral Mag, 1973,39,377 8 M Meyn, K Beneke and G Lagaly, Inorg Chem ,1993,32,1209 9 R Schollhorn and B Otto, J Chem SOC, Chem Commun, 1987, 1559 10 C Delmas and Y Borthomieu, J Solid State Chem ,1993,104,345 11 D L Bish and G W Bnndley, Am Mineral, 1977,62,458 12 G W Brindley and S Kikkawa, Am Mineral, 1979,64,836 13 S A Solin, D Hines, S K Yun, T J Pinnavaia and M F Thorpe, J Non Cryst Solids, 1995,182,212 14 G Busca, V Lorenzelli and V Escribano, Chem Muter, 1992, 4,595 15 E Uzunova, D Klissurski and S Kassabov, J Muter Chem ,1994, 4,153 16 W Wolski, E Wolska and J Kaczmarek, J Muter Sci Lett, 1993, 12,1011,1994,13,1206 17 W Wolski, E Wolska and J Kaczmarek, J Solid State Chem, 1994,110,70 18 W Wolski, E Wolska, J Kaczmarek and P Piszora, Phys Status Solidi A, 1995, 152, K19 19 T Yao, 0 Imafuji and H Jinno, J Am Cerum SOC , 1991,74,314 20 W Wolski, E Wolska and J Kaczmarek, Phys Status Solidi A, 1993,139, K51 21 L R Maxwell and S J Pickart, Phys Rev, 1953,92,1120 22 J J Bara, A T Pedziwiatr, Z M Stadnik, A Szytula, J Todorovic, Z Tomkowicz and W Zarek, Phys Status Solidi A, 1977,44,325 23 J S Smart, Phys Rev, 1954,94,847 24 J J Bara, Phys Status Solidi A, 1977,44, 737 25 L Neel, Ann Phys, 1948,3,137 26 A Delahaye-Vidal, K T Ehlsissen, P Genin and M Figlarz, Eur J Solid State Inorg Chem ,1994,31, 823 27 R D Waldron, Phys Rev, 1955,99,1727 28 W B White and B A DeAngelis, Spectrochim Acta, Part A, 1967, 23,985 29 H D Lutz, B Muller and H J Steiner, J Solid State Chem ,1991, 90,54 30 P Tarte, Spectrochim Acta, Part A, 1967,23,2127 31 E Wolska, Z Kristallogr ,1981, 154, 69 32 E Wolska and W Szajda, J Muter Sci , 1985,20,4407 33 E Wolska, Monatsch Chem , 1975,106,905 34 E Wolska, J Muter Sci Lett, 1984,3, 817 35 S Niziol, Phys Status Solidi A, 1973, 17, 555 36 P P Reichertz and W J Yost, J Chem Phys, 1946,14,495 37 R C T Slade and T K Halstead, J Solid State Chem, 1980, 32,119 38 G G Christoph, C E Corbato, D A Hofmann and R T Tettenhorst, Clays Clay Miner, 1979,27,81 39 G Brown, in Crystal Structures of Clay Minerals and Their X-Ray Ident$cation, ed G W Brindley and G Brown, Mineralogcal Society, London, 1984, p 364 40 A B Kiss, G Keresztury and L Farkas, Spectrochim Acta, Part A, 1980,36,653 41 J J Fripiat, H Bosmans and P G Rouxhet, J Phys Chem, 1967, 71,1097 42 K A Wickersheim and G K Korpi, J Chem Phys ,1965,42,579 43 E Wolska and U Schwertmann, Z Kristallogr ,1989,189 223 44 45 E.Wolska, Solid State lonics, 1988,28-30, 1349. M. Figlarz and S. Le Bihan, C.R. Acad. Sci. C (Paris), 1971, 272, 580. 50 51 52 P. Maurel, C.R. Acad. Sci. D (Paris), 1966,263, 1925. E. Wolska, Monatsch. Chem., 1976,107,349. T. Tsuchida, R. Furuichi, T. Ishii and K. Itoh, Thermochim. Acta, 46 0. Clause, B. Rebours, E. Merlen, F. Trifiro and A. Vaccari, 1983,64,337. 47 48 49 J. Catal., 1992, 133,231. T. Sato, H. Fujita, T. Endo, M. Shimada and A. Tsunashima, React. Solids, 1988,5,219. U.Schwertmann and E. Wolska, Clays Clay Miner., 1990,38,209. E. Wolska, J. Subrt, Z. Haba, J. Tlaskal and U. Schwertmann, J. Muter. Sci.,1994,29, 3269. 53 54 55 56 R. T. Tettenhorst and D. A. Hofmann, Clays Clay Miner., 1980, 28, 373. R. T. Tettenhorst and C. E. Corbato, Clays Clay Miner., 1988, 36,181. K. J. Pandya, W. E. O’Grady, D. A. Corrigan, J. McBreen and R. W. Hofman, J. Phys. Chem., 1990,94,21. W. Feitknecht, Helv. Chim. Acta, 1942,25, 555. Paper 6/00869K; Received 6th February, 1996 J. Muter. Chern., 1996, 6(lo), 1701-1707 1707

ISSN:0959-9428    

DOI:10.1039/JM9960601701

出版商:RSC    

年代:1996

数据来源: RSC

摘要:

Structural study by energy dispersive X-ray diffraction of amorphous mixed hydroxycarbonates containing Co, Cu, Zn, A1 Marilena Carbone,' Ruggero Caminiti*yb and C. Sadud 'Dipartimento di Chimica, Universita' di Roma La Sapienza, P.le A. Moro, 5, 00185 Roma, Italy bIstituto Nazionale per la Fisica della Materia, Universita di Roma 'La Sapienza', P.le A. Moro, 5, 00185 Roma, Italy A non commercial 8-8 energy dispersive X-ray diffractometer equipped with a solid state detector has been used for the determination of the amorphous structures of nine mixed hydroxycarbonates containing Co, Cu, Zn and Al. Energy dispersive X-ray diffraction (EDXD) provides information about the local order around the divalent metal ions and Al"'. An octahedral configuration around all the cations was found.Debye functions were calculated and compared with the experimental curves. Large angle X-ray scattering (LAXS) is a powerful technique for determining the structural parameters of liquid and amorphous systems, since it can provide information about the short-range order.'.' In particular, energy dispersive X-ray diffraction (EDXD)394 has been found to be a suitable tool for the investigation of such systems owing to its speed and reliability compared to a traditional angular scanning diffractometer. We have used the EDXD technique to determine the struc- tural properties of amorphous mixed Cu-Zn-Co-A1 hydroxy-carbonates, and the parameters which can influence the degree of crystallinity of such samples. It is important to establish which parameters, during the preparation, influence the struc- tural characteristics of copper mixed hydroxycarbonates, since these compounds are precursors for alcohol synthesis catalysts as well as for CO oxidation, namely the corresponding mixed oxides generally obtained by thermal a~fivation.~-'' A great deal of work has been devoted to the structural characterization of crystalline mixed hydroxycarbonates con- taining copper and one or more metal ions such as Zn, Co and Al,l2-I5 but no attempt to characterise the corresponding amorphous samples has been reported.Providing an investi- gation method for amorphous materials is even more important because of the growing interest in copper mixed amorphous materials for use as methanol synthesis catalysts.Wright et al." investigated the structural properties and the catalytic activity for CO oxidation of amorphous Cu-Co-Mn mixed oxides obtained by keeping the activation temperature of the corresponding carbonates low. Coteron and HayhurstI6 performed structural characteriz- ation and catalytic activity tests in methanol synthesis of Cu-Zn, Cu-Zr and Cu-Zn-A1 amorphous systems obtained by the spark erosion technique. Because of the important role played by the precursor structures in the catalytic activity, we were interested in determining the critical cation concentrations in some ternary and quaternary mixed hydroxycarbonates which lead to the formation of amorphous mixed hydroxycarbonates. It is also important to determine the structural parameters of the synthe- sized samples, namely the coordination number and the first coordination shell of the cations, and the influence of some preparation parameters, such as the drying method, on the sample structures.Coprecipitation from nitrate or acetate mixed-cation solu- tions with NaHCO, is a common synthesis method for crystal- line copper-based binary hydroxycarbonate~.~~.~~Binary systems such as Cu-Zn, Cu-Co and Cu-Mn have been obtained by coprecipitation and well charaterized from a structural point of view.12914 Different coprecipitation pro-cedures are reported as well, which lead to different crystalline phases. One important parameter is the addition sequence; different phases are obtained if the nitrate (or acetate) solution is added to the hydroxycarbonate or vice versa. If we refer to a Cu-Zn binary system, a malachite-like phase [Cu2(OH),C03] is obtained in the first case, while gherardite [Cu2(0H),N03] is obtained in the second." The cation ratio is also a critical parameter.If we consider the same Cu-Zn system, a Cu/Zn atomic ratio of 5.7 is the critical value above which a malachite-like phase forms. At higher Zn concen-trations an aurichalcite phase [(Cu,Zn), (OH),( CO,),] forms together with the malachite-like phase." The presence of a cation with higher oxidation number can also influence the crystalline phase of the sample. A hydrotal-cite-like phase [M116M111Z(OH)16C03.4Hz0]forms'3 when the ratio of divalent and trivalent cations is 3 :1.However, the preparation procedures reported in the litera- ture" do not always lead to the formation of crystalline samples. Marchi et aLZ0 studied the parameters that can influence the crystallinity of Cu-Co-A1 samples obtained by coprecipitation. Among the important parameters they quoted the ratio between divalent and trivalent cations, the pH of the solution and the supersaturation ratio. The structural properties of the mixed oxides, their homogen- eity and cation interdispersion, which contribute to the deter- mination of the efficiency of the catalytic process, are greatly affected by the structural properties of the corresponding precursor^.^-^ We considered two series, one of ternary and one of quatern- ary samples, with varying Co concentrations (3-20%), and the effects of this variation on the physical state of the samples was investigated.Moreover, A1 was present in the series of quaternary samples at a constant percentage (9%0), a value lower than that at which crystalline hydrotalcite forms (25%0)." Materials The compounds studied were prepared by coprecipitation, according to the procedure reported in refs. 12-15. Atomic ratios are reported in Table 1. Surprisingly, although we adopted the same preparation procedure for the synthesis of crystalline hydroxycarbonates, we obtained almost always amorphous samples. In a typical preparation 0.6 1 of an 0.7 mol I-' aqueous solution of the cation nitrates in suitable proportions was added to 0.9 1 of NaHCO, (1.1 moll-') at 323 K under vigorous stirring.The slurry was digested for 4 h under the same conditions. The pH just after the precipitation was 6.5-7.0 and it increased to 8.5-9.0 at the end. The precipitates J. Muter. Chem., 1996, 6(lo), 1709-1716 1709 Table 1 Atomic ratios in the studied compounds sample cu Zn co A1 czco 68 29 3 --CZCl 63 27 10 NOAL 63 18 19 NOAL1 56 24 20 CZCAO 63 26 3 9 CZCAl 60 25 6 9 CZCA2 57 25 9 9 CZCA4 55 23 13 9 CZCA3 51 22 18 9 MACA 50 22 19 9 MARUL 50 22 19 9 were pale blue; they turned green after 2 h. The samples were washed with 8 1 of distilled cold water (to eliminate Na' and NO,-ions), dried at 363 K and finally ground in an agate mortar.In order to determine the effect of the preparation conditions, the coprecipitation of the sample containing Cu :Zn :Co :A1 in atomic ratio 51 :22 :18:9 (CZCA3) was carried out at a higher temperature, (373 K). Moreover the sample containing Cu :Zn :Co :A1 in atomic ratio 50 :22 :19:9 was prepared twice. In the first preparation the conventional procedure was followed (MARUL sample), in the second the drying of the washed precipitate was performed over CaC1, (MACA sample). Techniques Elemental Cu, Zn, Co, A1 analyses were performed by atomic absorption with a Varian Spectra AA30 instrument and Table 2 reports the analytical results referring to the nine amorphous samples studied. The anion contents of the hydroxycarbonates were determined by evaluating the amount of H,O and CO, produced in a thermal decomposition in a flowing system connected to a traditional BET vacuum line.The X-ray diffraction experiments described here were car- ried out by employing a non-commercial X-ray energy scanning diffractometer,' equipped with an X-ray generator (water cooled, W target with 3.0 kW maximum power), a solid-state detector (SSD) connected to a multichannel by means of an electronic chain collimator system, step motors and a sample holder. The X-ray source is a standard Seifert tube, used at 50kV and 40mA or 45kV and 35mA, whose white Bremsstrahlung component was used. The primary beam inten- sity Io(E)was measured directly using the same voltage (50 or 45 kV), by reducing the tube current to 2 mA at the zero scattering angle without the sample.The transmission of the samples was measured under the same conditions; from the following equation we obtain the experimental value exp[-p(E)]t that it is used in eqn. (6) and (7) for the absorption corrections.22 The fluorescence lines present in the 5-11 keV range due to Table 2 Element concentration in the prepared hydroxycarbonates ~~~ sample Co/moll-' Cu/mol 1-1 Zn/moll-' Zn/Cu Al/moll-' CZCl 3.61 1 17.399 6.894 0.396 -NOAL 3.978 13.290 3.978 0.288 -NOAL1 5.593 14.221 5.622 0.395 -CZCAl 1.604 12.03 1 5.615 0.467 1.872 CZCA2 2.466 12.054 4.657 0.386 2.192 CZCA4 3.397 11.237 4.181 0.372 2.091 MACA 3.647 9.899 4.428 0.447 2.084 MARUL 3.647 9.899 4.428 0.447 2.084 CZCA3 5.590 11.740 4.752 0.405 2.516 1710 J. Muter.Chew., 1996,6( lo), 1709-1716 Fig. 1 The energy scanning diffractometer used for the measurements Table 3 Scattering parameters associated with the minimum and maxi- mum values of the used energy for each measurement angle 17-38 keV 17-42 keV angle/degrees smm Smax Snun Srnax 21.0 6.18 13.80 6.18 15.26 15.5 4 61 10.30 4.61 11.38 10.5 3 14 7.02 3 14 7.39 8.0 2.40 5 36 2.40 5 92 5.0 1.50 3.36 1.50 3.71 3.5 1.05 2.35 1.05 2.60 3.0 0.90 2.01 0.90 2.22 2.0 0.60 1.34 0.60 1.48 1.5 0.45 1.oo 0.45 1.11 1.0 0.30 0 67 0 30 0.74 W, Co, Cu, Zn and, to a lesser extent, Al, do not disturb our measurements, since they are outside the region of our interest.A Seifert and Rich high-voltage power supply (stability -=0.1%) was used. The detecting system consisted of a EG&G liquid-nitrogen cooled ultrapure Ge SSD (ORTEC, model 92X) connected to a PC 286 via ADCAM hardware. The current pulse from the detector was converted into a digital signal, which was visualized on a computer screen through a multichannel analyser (MCA). The collimating system was composed of four adjustable-width W slits purposely placed to reduce the X-ray beam angular divergence. The X-ray tube and detector holding arms rotated in the vertical plane around a common centre in order to reach the desired 26 scattering angle; the movement was accomplished by step motors which allowed reproducibility within 0.001 O for the scattering angles.The diffractometer is shown in Fig. 1. The working conditions used were as follows. Supply: high voltage =50 kV, current 40 mA, total power =2000 W for the samples CZC1, CZCA1, CZCA2, CZCA3 and CZCA4, or 45 kV, 35 mA, total power =1575 W for the samples MACA, MARUL, NOAL and NOAL1. Measurement angles and the energy ranges used are reported in Table 3. Using the formula s= 1.014 E sin I!? it is possible calculate the s range of the angles used. The powder samples were formed into pellets in order to perform the measurements. The total intensity 1scattered by a sample and observed by an energy dispersive detector in approximation of single scat- tering and transmission geometry4,,, can be expressed as: with where 8 is the scattering angle, E is the photon energy revealed by the detector and E‘ is the initial energy of a photon scattered inelastically at the observed energy E.From Compton’s incoherent scattering we have the expression r 1 1E’=E K is the scale factor between the intensity reaching the detector and the intensity scattered by a stoichiometric unit of the sample. Io(E) is the energy spectrum of the primary beam measured at 8= 0” (as described previously). P(E,8) is the polarization factor by a scattering of a primary radiation with polarization @(E)that is: P(E,8)= ( 1+ COS’ 28)/2+ sin228 @ (E)/2 (4) with @ (E)= C(lp,n(E) -Ip.p(E )l/CIp,n(E)+ Ip,p(E)l ( 5) where lp,nand Ip,pare the intensities of the normal and parallel polarization components, respectively, with respect to the scattering plane.Acoh(E,8) is the X-ray elastic absorption coefficient Ac0h(E,8)= exp [ -p(E) t sece] (6) A,,,(E,E’,B) is the X-ray inelastic absorption: exp [ -p(E) t sece] -exp [ -p(E’) t sed?]A,nc(E,E’,o) = -p (E) t sec8-p (E’)t sec8 (7) where p is the absorption coefficient and t the thickness of the pellets. Icoh(E,8) is the total elastic scattered intensity: 1Coh(E*8)=c cnfn2(s)+i(s) (8) n IInc(E,8)is the incoherent contribution to the total scattered intensity.1c,f: (s) is the self scattering intensity; c, is the concen- n tration of the different species, i(s) is the intensity of interfering waves scattered by atom pairs, s is the scattering parameter and is defined by 471 sin 8 s= A = 1.014 E sin 8 (9)~ when E is expressed in keV and s in A-’.Data treatment After correction of the experimental data for the escape peak suppression, the intensity data were handled by means of our DIFl program written in FORTRAN IV. This program also made the necessary absorption corrections to combine the -----21.r-159 10s * 8.W---5.0. 3.5’ 3.0’ . 2.w .. . . . . . -1.5’ . 1.0’ 0 2 4 6 8 10 12 14 s1A-l Fig. 2 Picture of the scattered intensity (e.u.) for the consecutive measurements angles various angular data sets, as described in the paper by Nishikawa and Iijima.” The rescaled intensity, in electron units (e.u.), for each scattering angle used is shown in Fig. 2 for the sample NOAL.Normalization to a stoichiometric unit of volume containing one Co atom was performed. Table 4 reports the elemental normalized concentrations. Radial distribution functions, D(r), were calculated from the static structure functions i(s): i(s)= ICoh(E,8)-1CL2(s) ( 10) n according to the expression: D(r)= 4nr2p0+ 2rn-’ s-i(s)-M(s)sin(rs)ds (11) In this equation po= [1nif,(0)]2V-’,where V is the stoichio- 1 metric unit of the chosen volume, niis the number of atoms i per unit volume, andfi is the scattering factor per atom i. M(s)= ~f2~o~~~~zco~~~~~~~P~-0.010 s2) ( 12) is the sharpe$ng factor. For the upper integration limit s,, we used 14.5 A-’. Theoretical peaks were calculated by a corresponding Fourier transform of the theoretical intensities for pairs (p, q) of interactions (Debye functions): using the same sharpening factor and the same s,,, value as for the experimental data and assuming the rms variation in the interatomic distance to be opq.Data analysis The observed structure functions, in the form s-i(s)-M(s),for both series of samples are reported in Fig. 3(u)and (b).In the figures, the Al-containing samples are ordered by decreasing Table 4 Element concentrations, density (d) and stoichiometric volume (V)[the concentrations are given as number (n,)per stoichiometric unit of volume. I/ was chosen, in all cases, to correspond to a value containing one Co atom] sample co cu Zn A1 Zn/Cu 0 C H d/g cm-3 v/A3 CZCl 1.o 4.8 183 1.9092 -0.369 22.3653 3.6367 11.4556 3.2830 495.8541 NOAL 1.o 3.4667 1.oOOo -0.299 15.5334 2.3333 8.5333 3.3000 335.5612 NOALl 1.o 2.3889 0.9444 -0.395 14.7778 2.3333 7.7778 3.3072 278.9758 CZCAl 1.o 7.500 3.500 1.1667 0.467 45.500 1 7.1667 24.000 1 2.6737 1035.1 160 CZCA2 1.o 4.888 1.8889 0.8889 0.386 3 1.7778 4.8889 17.1 1 11 2.7397 673.4508 CZCA4 1.o 3.3077 1.2308 0.6154 0.372 21.9231 3.4615 11.5385 2.6134 488.7652 MACA 1.0 2.7143 1.2 143 0.5714 0.447 18.2143 2.5714 10.5000 2.6051 455.3012 MARUL 1.0 2.7143 1.2143 0.5714 0.447 18.2143 2.5714 10.5000 2.4848 477.3485 CZCA3 1.o 2.1000 0.8500 0.4500 0.405 12.0500 1.2500 8.300 2.7953 297.0258 J.Muter. Chew., 1996,6(10), 1709-1716 1711 2 x lo4 1.5 x lo4 5000+ ? J0.tNOAL 1 0 3 6 91215 0 3 6 91215 1.5 x lo42x1044 ikZCA 1 CZCA 4 -1.5 x 1O4 ' f ' ! ' 1 0 3 6 91215 0 3 6 91215 0 3 6 91215 t -. -l""i.-. J,,, 'b' ,,, '4',, ,~~,,,l 0 MARUL CZCA 3 I,,,, ,,*I, I,,,. It,,, I,,,,-1.5 x lo4 I I I I 0 3 6 91215 sIA-' Fig. 3 Observed structure functions s.i(s).M(s)(in em. A-')for (a) samples not containing A1 and for (b) samples containing A1 A1 content calculated per unit-cell volume of Co. The samples without A1 are ordered, instead, by increasing Co content. The radial distribution functions in the DIFF(r) =[D(r)-4zr2p,] form are reported in Fig. 4(u) and (b). The main feature which emerges from the s-i(s)M(s)analysis of the samples containing no A1 is the loss of structure of the oscillations at increasing Co content.The amorphous sample at low Co concentration (CZC1) shows structured oscillations over the whole s range examined, while these structures gradually decrease in intensity in the samples with higher Co contents. The quaternary samples show a similar behaviour, namely a loss of structure is displayed in the si(s)-M(s)functions at increasing Co content for samples CZCA1, CZCA2 and CZCA4. This behaviour is not followed by the MACA, MARUL and CZCA3 samples, since the s.i(s)-M(s)functions of these samples seem to be more structured. We assume, therefore, that other factors can influence the local structure, as will be discussed in the following section.1712 J. Muter. Chem., 1996, 6(lo), 1709-1716 The radial distribution functions in the DIFF(r) form, obtained from the scattered intensities with the previously described proFedure, show the prFsence in all the samples of a peak at 2.0A and one at 3.2A. The first peak has been attributed to the metal-oxygen interaction, the second one to the interaction between metal ions. Peaks at higher r values are expected to be produced in the DIFF curve by the interaction between cations bridged by an oxygen atom. Therefore the peaks at 5.4, 6.2 and 8.4A, that appear in the DIFF curves have been mainly attributed to interaction of metal atoms belonging to adjacent coordination polyhedrons. More detailed information on the studied systems can be obtained by the analysis of the radial distribution function D(r), which is the area of each peak which is proportional to the number of scattering atoms and to their scattering factors.We have focused our attention on the peak at 2.0A, which provides information about the cations first coordination shell. Therefore, by comparing the experimental peak with the theoretically calculated peak shape, the coordination numbers 401-i20 10 -1 0 -3 0 0 2 -4 6 3 8 1 1 0 -20/NoA;ll I ' I I ~ -30 0 2 4 6 I ' ~, ' I 8 1 (b) 403 40: 40)30 30120 I h L Y Q 10 .'20 CZCA ? -30; 0 2 4 6 8 1 0 0 2 4 6 8 1 0 0 2 4 6 8 10 40 c 301 1 40q30 4 ..401 20'Ol : * *5: -* 0 .... f : ' .* */ :: ... 0 . .. ,. ... .* .. 0.... .* r' 0... i -30 0246810 0246810 024681rlA-' Fig. 4 Radial distribution functions, in the form D(r)-4471p2p, (in e2 k3x lop3)referring to (a)the series of samples containing A1 and (b) the series of samples not containing A1 and the bond distances of all the cations have been derived for each sample. The experimental D(r)peak shapes compared to the theoretical peak shapes are reported in Fig. 5(u) and (b). Results and Discussion The first interesting feature emerging from the analysis carried out on the prepared hydroxycarbonates is the possibility of obtaining either crystalline or amorphous samples under similar preparation conditions. The conventional X-ray powder diffraction patterns for all prepared samples showed that some samples had showed that some samples have an amorphous or low crystallinity pattern, without evident Bragg peaks, while other samples have a crystalline structure (CZCO, CZCAO).All the diffraction pat- terns, obtained with Cu-Ka radiation, are reported in the thesis of M. Carb~ne;~~we have studied by EDXD only those with no evident Bragg peaks. If the samples do not contain Co, or if they contain Co at low concentrations (3%, CZCO, CZCAO), their structures are completely crystalline. In samples without Al, a malachite-like and an aurichalcite phase are formed. The presence of A1 in these hydroxycarbonates favours the formation of hydrotalcite together with the usual malachite-like phase.The degree of crystallinity changes when a higher concentration of Co is present in the samples. We varied the Co concentration in the samples in a discrete way, so we do not know exactly the critical Co concentration. However, the samples containing 8 ?lo Co (series without Al) or 6% Co (series with Al) are completely amorphous, suggesting that Co concentration is a fundamental parameter affecting the crystallinity of the samples. The Co critical value must be in the range 3-6% when both Zn and Cu are present in the hydroxycarbonates. This effect is even more surprising if we consider that binary Cu-Zn or Cu-Co as well as quaternary Cu-Zn-Co-A1 hydroxycarbonates with similar cation proportions are ~rystalline.l~-~~ J.Muter. Chem., 1996,6(lo), 1709-1716 1713 60. 60 . ... 60 . 50-r 50-7 50-r ..-40-7 40-r 40-r 30-1 30-: L 20-I 20-: lo-: lo-: :NOAL : NOAL1 0-0. -1 0 I1I1IItl II I II III I a ILILlII I II IIt -10 lI1lI1lllI1lllI1llII1lllIII1fIL1ll I I1II I I I I "I I -1 0--I II I ' I I' I 'III I I ; I 40:.i; 30-: 30f 20-r lo-: CZCA2 CZCA 4 -1 0~ -1 0i 60. 60 60. ...* SO-: '* SO-: 2.... 50-r .. 40-r 40-: 40-: 30-r 30-: 30-: * * 20-: 20-: lo-; MACA lo--CZCA 3 0. 0. -1 0 lI1l~ILIIILI1l~llllIfI1lI1lll;llllo-=' I I f ' I" I' I I I II1' I " I---10-,' I I j fI I 1 I I I I I I I I I II I i t I I I I-1 rlA Fig. 5 Theoretical (solid line) and experimental (dotted line) radial distribution functions D(r) (in e A-3 x of (a) samples containing A1 and (b)samples not containing A1 A further increase of the Co content favours an increase of MACA and MARUL, which have similar cation concen-the amorphous character as shown by the loss of structure trations but were dried by different methods, show similar in the si(s).M(s)functions for the samples without Al.Therefore s-i(s).M(s)functions, suggesting that this step of the preparation the Co content is a parameter that controls not only the procedure does not influence the final sample structures. formation of amorphous samples but also the extent of their Quite different is the case of the CZCA3 sample, in which amorphous character. the relatively high temperature of preparation might have Even A1 seems to have an effect on the amorphous state of favoured the formation of a more crystalline sample.the samples. In comparing the structure functions of two The metal-oxygen distances, the coordination numbers and samples with similar cation proportions but different A1 con- the standard deviations used in reproducing the first pfak of tents [CZCl and CZCA2 in Fig. 3(u) and (b)respectively] we the experimental radial distribution function (at 2.0 A) are see that the former, which shows less structured oscillation, is reported in Table 5 for all the samples, and the values for therefore more amorphous. sample CZCA3 are given in Table 6. The chosen distances are LongestAnother parameter which can influence the degree of crystal- averaged over the values reported in the literat~re.~~-~* linity of the mixed hydroxycarbonates is the Zn/Cu ratio.We and shortest literature values are collected in Table 7. can see this effect, by comparing the s.i(s)-M(s)of two similar We assumed a coordination number of 6 for both Co and samples, in this case, two samples with similar Co concen- Cu on the basis of diffuse reflectance spectra, whose absorption trations: CZCA4 and MARUL. The second one, with a higher bands clearly indicate the presence of the cations in octahedral Zn/Cu ratio, shows a more structured s.i(s).M(s).The samples and distorted octahedral symmetry, respectively. 1714 J. Muter. Chem., 1996, 6(lo), 1709-1716 Table 5 Metal-oxygen distances (r), number of nearest neighbours (n) and standard deviation (a)used for all the samples r/A n CIA 1.98 0.1 2.50 0.1 2.07 0.1 2.12 0.1 1.88 0.1 Table 6 Metal-oxygen distances (r),number of nearest neighbours (n) and standard deviation (a)used for sample CZCA3 r/A n CUP 1.95 4 0.1 CUa 2.80 2 0.1 Table 7 Lowest (rmln) and highest (rmax) metal-oxygen distances reported in the literature 1.88 2.04 2.23 3.97 2.07 2.15 2.10 2.20 1.85 1.98 We performed two different peak shape calculations, using 4 and 6 as the coordination number of the Zn cation.The peak area is, in all the cases, lower when coordination 4 is used [Fig. 5(a) and (b)],thus indicating that 6 is the actual coordination number for Zn.Fig. 6(a) and (b) compare the experimental curves and the theoretical peak shapes using 4 and 6 as the coordination number for zinc for the samples CZCAl and CZCl respectively. Obviously the coordination 50 -1 A -1 ot-s 0 0.5 1 1.5 2 2.5 3 3.5 Q (blL I 4050f -1o>l 0 0.5 1 1.5 2 2.5 3 3.5 r /A Fig. 6 Radial distribution functions D(r) (in e2 k3x of (a) the sample CZCAl and of (b) sample CZC1: experimental curve (dotted line); theoretical curve using 6 as coordination number for the zinc (solid line); theoretical curve using 4 as coordination number for zinc (dashed line). numbers 4 and 6 used are given with an error between 5 and lo%, as reported in the literature for this technique. It is worthwhile noting that the cation coordination could be deduced even without the aid of the reflectance spectra.A good agreement between theoretical and experimental peaks is achieved only when 6 is used as the coordination number for all of the cations. The use of coordination number 4 for any of them always gives a theoretical peak shape lower than the experimental one. When calculating the theoretical peak shapes, we obtained the best agreement with the experimental curves using two different copEer-oxygen bond lengths. Indeed, the experimental peak at 2.0A shows a small asymmetry at higher r values, which can be easily attributed to a greater Cu-0 distance due, in six-fold coordination of Cu, to two oxygen atoms Jahn-Teller distorted away from the central metal.The coordination distances we found for Co and Cu are very close to those reported by Wright et al." referring to Cu-Mn and Co-Mn crystalline carbonates with the rho- docrosite (MnCO,) structure, investigated by EXAFS. The two different Cu-0 distances, both planar and apical, reported for malachite12 are not detected in the amorphous mixed hydroxycarbonates. Two different sites are present in malachite,oone having planar and apical bond distances of 1.98 and 2.71 A re$pectively, the other having bond distances of 2.01 and 2.41 A. The presence of Co or Zn in a solid solution with malachite, although varying the bond distances does not remove the duplicity of the ~ite.'~,'~ Only one kind of site is present, instead, in the amorphous samples studied, with a bond length which is nearly the average of the malachite sites.We checked that only one site contributes to the Cu-0 coordination, by calculating a peak shape with the Cu-0 distances reported for malachite, but the agreement with the experimental D(r) was worse. Different values of Cu-0 coordination distances had to be used to reproduce the peak of the CZCA3 sample, which, as explained before, was prepared at a higher temperature. In particular, this sample has shorter planar and longer apical distances compared to the other samples. This might indicate the presence of small copper oxide particles, whose formation was favoured by the high temperature during the preparation. This hypothesis is confirmed by the reflectance spectrum, which shows the typical CuO absorption at 12500cm-l, due to the transition between the valence and conduction bands.Finally, it is interesting to note that identical parameter values were used for all the samples, irrespective of the cation ratios. This implies that the first coordination shell of the amorphous mixed hydroxycarbonates is not significantly affec- ted by the cation concentrations. On the basis of literature findings2' and of the experiments carried out, we can conclude that the coprecipitation method for hydroxycarbonate preparation can be used in the synthesis of amorphous samples as well, the determining parameter being the cation ratios. EDXD has been employed in order to study some Cu-Co-Zn-A1 mixed hydroxycarbonates with fixed A1 content and variable Co concentration, prepared by coprecipitation. Co has a primary role in determining the physical state of the samples. The presence of Co in the ternary and quaternary mixed hydroxycarbonates can reduce long-range order.The critical content of Co inducing the formation of amorphous samples lies in the range 3-6%. Moreover, increasing Co content together with the presence of A1 contribute to a lowering of the degree of order in the amorphous samples. References 1 R. Caminiti, C. Munoz Roca, D. Beltran Porter, and A. Rossi, Z. Naturforsch. Ted A, 1988,43, 591. J. Mater. Chem., 1996, 6(lo), 1709-1716 1715 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 D Atzei, R Caminiti, C Sadun, R Bucci and A Corrias, Phosphorus, Sulfur, Silicon Relat Elem, 1993,79,13 T Egami, H J Guenterhodt and H Beck Glassy Metals I, Springer Verlag, Berlin, 1981, p 25 G Fritsch and D A Keimel, J Muter Sci Eng A, 1991,134,888 K Klier, Adv Catal, 1982,31,243 K Klier, Appl Surf Sci ,1984,19,267 K Klier, Inorg Chem ,1989,28,3868 A Suger and E Freund, US Put, 4 122 110,1978 A Sugier and E Freund, US Pat, 4 291 126,1981 J E Baker, R Burch and S E Golunski, Appl Cat, 1984,53,279 P A Wright, S Natarajan, J M Thomas and P L Gai-Boyes, Chem Muter, 1992,41053 P Porta, S De Rossi, G Ferrans, M Lo Jacono, G Minelli and G Moretti, J Catal, 1988,109,367 P Porta, R Dragone, G Fierro, M Inversi, M Lo Jacono and G Moretti, J Muter Chem, 1991, 1, 531 P Porta, G Moretti, M Lo Jacono, M Musicanti and A Nardella, J Muter Chem, 1991,1, 129 S Morpurgo, M Lo Jacono and P Porta, J Muter Chem, 1994, 4,197 A Coteron and A N Hayhurst, Appl Catal A, 1993,101,151 P Courty and C Marcilly, Preparation of Catalysts I, Elsevier, Amsterdam, 1976,p 119 18 P Courty and C Marally, Preparation of Catalysts IZI, Elsevier, Amsterdam, 1983, p 485 19 J S Campbell, Ind Eng Chem Process Des Dev, 1970,9,588 20 A J Marchi, A G Sedran and C R Apesteguia, Proc IVth Int Symp on The Scienttjic Bases for the Preparation of Heterogeneous Catalysts Louvazne La Neuve 1986, ed B Delmon, Elsevier, Amsterdam, 1987, p 529 21 R Caminiti, C Sadun, V Rossi, F Cilloco and R Fehci, XXVth Italian Congress of Physical Chemistry, Cagliari, 1991, p 4, 138, It Pat, RM/93 A000410,1993 22 K Nishikawa and T Iijima, Bull Chew SOC Jpn ,1984,57,1750 23 M Carbone, Graduational dissertation, Univ ‘La Sapienza’, Rome, 1992 24 J N Van Niekerk and F R L Schoening, Acta Crystallogr, 1953, 6,609 25 H Von Riffel, F Zettler, H Bokern and H Hess, Z Anorg Allg Chem, 1979,454,175 26 G B Johansson and 0 Lundquist, Acta Crystallogr Sect B, 1976,32,407 27 D T M Cromer, J Kay and A C Larson, Acta Crystallogr, 1966, 21,383 28 R F Zarhrobsky and W H Baur, Acta Crystallogr Sect B, 1968, 24,508 Paper 6/02815B, Received 23rd April, 1996 1716 J Muter Chem, 1996,6(10), 1709-1716

ISSN:0959-9428    

DOI:10.1039/JM9960601709

出版商:RSC    

年代:1996

数据来源: RSC

摘要:

The structures of strontium tellurite and strontium telluride aluminate sodalites studied by powder neutron diffraction, EXAFS, IR and MAS NMR spectroscopies Sandra E. Dann and Mark T. Weller Department of Chemistry, The University of Southampton, Highfeld, Southampton, UK SO1 7 1BJ Aluminate sodalites containing the telluride and tellurite ions have been synthesised for the first time. The structures of Sr, [A10,],,( TeO,), and its reduced product Sr, [AlO,] 12Te, have been determined using Rietveld refinement of powder neutron diffraction data. In Sr, [A102]i2Te2 the anion occupies a position at the sodalite cage centre, whilst in Sr, [A102]12(Te03)2 the larger tellurite ion is displaced to allow reasonable coordination to the strontium ions. The reduction of the occluded TeO,,- to Te2- has also been investigated using IR spectroscopy, 27Al and 125Te MAS NMR spectroscopies and Te L,,,-edge EXAFS.The sodalites are a well known class of anion-containing framework consisting of P-cages, formed from TO4 (T =Be, Si, Ge, Al, Si, Ga, etc.) tetrahedra, directly linked through the six- membered rings and containing a centrally placed anion tetrahedrally coordinated to four cations (Fig. 1). The composi- tion of sodalite structures, with general formula [MnI8+ [{ TTO, ],I6- -[X,I2-,is very diverse;'-, many different cat- ions and anions (M=Na+, Ca2+, Kf, Ag+, Sr2+, etc.; X= C1-, Br-, S2-, Se2-, Mn04-, NO3-, etc.) can be accommo- dated in the sodalite cage through partial collapse of the framework, by tilting the tetrahedra out of their normal planes and by deviation from perfect tetrahedral geometry around T( T').The guest-host interactions of the framework structure with the occluded species are exploited in pigments where the framework stabilises unstable or transient species. Of note, as members of this class, are the well known ultramarines which contain and stabilise the polysulfide radicals S2-and S3-and the corresponding selenium analogues. Recently attention has centred on the occlusion of semiconducting particles in frame- work structures by encapsulation of S2-and Se2- and divalent cations; however, the telluride (Te2-) ion has been less well studied. Sodalite synthesis has been achieved in several ways includ- ing hydrothermal methods,'~~ high-temperature sintering5 and structural conversion.6 After phase synthesis, exchange and Fig.1 Depiction of sodalite structure showing the central anion (large sphere) tetrahedrally coordinated by the cations (dark spheres) occluded within the sodalite framework reaction of the anions and cations entrapped in the sodalite cage is also possible without decomposition of the surrounding framework, e.g. SCN- decomposition to form S3-and S2-in the synthesis of ~ltramarine.~ The group of materials where T=T'=Al are the aluminate sodalites with general formula M, [AlO,],,X, (M =Ca2+, Sr2+, Cd2+; X=S04,-, W042-, Cr04,-, MOO^^-, S2-). Many of these materials have been extensively studied by Depmeier,,-lo particularly with respect to the orientation of the tetrahedral anions located at the centre of the P-cage unit. These materials are also of note as they disobey Lowenstein's rule of aluminium avoidance'' and form a group of aluminium- rich structures along with bicchulite, Ca, [A1,Si06],(OH), .Owing to the high aluminium content, highly acidic frame- works are produced, which may have potential uses in catalysis and ion exchange. In recent studiesi2*13 it was found that sulfate- and selenite-bearing sodalites could be reduced to yield sodalites containing sulfide and selenide by reaction under hydrogen. This reaction proceeds by an intracage process, i.e. the framework remains intact during modification of the anion. In this paper we describe the synthesis of the first aluminate sodalites bearing tellurium-containing anions and their struc- ture determination using a combination of neutron diffraction, EXAFS, IR and MAS NMR spectroscopies.Experimental The parent compound Sr, [AlO,] 12( Te03), was prepared using solid-state synthesis at high temperature as explored initially by Kondo14 and Depmeier." SrO (99.95%), TeO, (99.9%) and A1203 (99.5%) were ground homogeneously using a pestle and mortar in a glove box. The reaction mixture was then sealed in a silica tube (to prevent volatisation of TeO,, which would occur at 900 "C, below the reaction temperature of alumina) and heated to 1200 "C in a tube furnace for 48 h. The tube was then slow-cooled to room temperature. The product crystallised as a fine white powder whic! was shown to be a single cubic phase with a=9.425( 1) A by powder X-ray diffraction.Sr8[A10,],,Te2 was prepared by the reduction of Sr8[A10,],,(Te0,), in a stream of hydrogen at 850 "C for 8 h. The mass loss of 5.3% was in good agreement with the expected value of 5.45% for the reaction: Sr, [A10,],,(Te0,), +ST, [A1O,],,Te2 +6H20 The product recrystallised as a light-brown powder which was shown to be a single cubic phase by powder X-ray diffraction with a =9.379(1)A. IR spectra were obtained for both compounds on a Perkin- J. Muter. Chem., 1996,6(10), 1717-1721 1717 67.341 (a) A I EXAFS *$ 1100 900 700 500.-c c 0.33 Fig. 2 Part of the IR spectra of Sr,[AlO,],,Te, (a) and Sr8[A102]12(Te03)2(b) showing loss of the Te032- v1 stretch at 758 cm-' on reduction to Tez- sodalite in hydrogen.The aluminate cage bands are also assigned as vBs, vsl, vSz. Elmer FTIR 1710 spectrometer with a 3600 data station on pressed KBr discs. Reduction of the parent sodalite showed loss of the v1 stretch of Te03,- at 758 cm-', as seen in Fig. 2. The other absorptions in the IR spectra were assigned to the four IR-active bands expected for aluminate sodalites, which will be discussed in detail later (uide infra). Te L,,,-edge EXAFS were collected in transmission mode on station 7.1 at the SRS laboratory at Daresbury on a sample of Sr, [A102]12(Te03)2 diluted to 5% in mass with boron nitride. Data reduction and background subtraction was undertaken using PAXAS16 and data analysis using EXCURV92.I7 Fourier transform of the Te-edge EXAFS data cn the TeO,,--containing sodalite gave a single peak at 1.845 A.Best-fit parameters were achieved for three nearest neighbours at this distance and gave an R, value of 24.2%. The fit to the data is shown in Fig. 3. Solid-state 125Te and 27Al MAS NMR spectra were collected using a spin rate of 12 kHz at room temperature referenced to dimethyl tellurium and 1 mol dm-, AlCl, solution, respect- ively, on a Varian VXR300 spectrometer fitted with a 7T superconducting magnet at the University of Durham. 27Al NMR spectra were recorded using a field of 78.158 MHz and a 1 ps pulse corresponding to a 10" pulse in the reference solution of 1 mol dm-, AlCl, over 2000 repetitions.lZ5Te NMR spectra were recorded using a field of 94.45 MHz with a 1 ps pulse width and a relaxation delay of 30 s. Typically 1000 scans were performed to achieve good signal-to-noise ratios. The 125Te and 27Al NMR patterns for Sr8[A102]12Te, are shown in Fig.4. A sharp resonance is observed in the 125Te NMR pattern at 6-1372 which is characteristic of the Te2- anion. In contrast the tellurite ion resonates at 6 + 1742. Although there are very few lZ5Te data available for compari- son, this shift appears in good agreement with that of K2Te03 which has a lZ5Te resonance at 6-t 1732.18 The 27Al NMR spectra showed the characteristic shapes expected for a quadru- polar nucleus in a slightly distorted environment with incom- plete averaging of quadrupolar coupling. However, the distortion of the tetrahedral geometry in the aluminate soda- lites is relatively small and the isotropic chemical shift can be obtained by fitting the spectrum for various quadrupolar coupling constants.19 Simulation of the experimental spectra allowed calculation of the isotropic chemical shift, giving values of 6 75.7 and 75.2, respectively, for the telluride- and tellurite- containing sodalites.1718 J. Mater. Chern., 1996, 6(lo), 1717-1721 rlA Fig. 3 Fit to the Fourier transform of the Te L,,,-edge data for best fit to three oxygens at 1.845 A: (-) experimental data; (---)theoretical data 250 150 50 -50 -150 , -1000 -1400 -1800 -2200 -2600 6 Fig. 4 (a) 27Al and (b) 125Te MAS NMR spectra obtained from Sr8[A102]12Te2.Details of data collection are given in the text Structure determination Time-of-flight neutron diffraction data were obtained for Sr, [A10,],,( TeO,), and Sr, [AlO,] ,,Te2 at room temperature over the d-spacing range 3.08-0.568 A using the medium- resolution high-flux instrument, POLARIS, at the Rutherford Appleton Laboratory. Both patterns could be indexed on a body-centred cubic structure with systematic absences consist- ent with the space group 143~2,as expected for a sodalite with a single tetrahedral framework species. Rietveld refinement analysis was performed in this space group using the pro- gram GSAS.20 Starting models were taken from Brenchley and Weller13 and used as the basis for both refinements with Te on the 2a (O,O,O) site, Sr on 8c (x,x,x) x=O.24, aluminium on 12d (1/4,1/2,0) and the cage oxygen on the 24g (x,x,z) x=O.15, z= 0.47.For the tellurite the oxygen of the tellurite group was initially positioned on 24g (x,x,z) x=0.38 and z=0.51, which allowed good coordination to the cage and was chosen by Depmeier for chromite sodalite' and Brenchley and Weller', for selenite sodalite. Neutron scattering lengths were taken as Te 0.580, Sr 0.702, A1 0.3449 and 0 0.5805 x cm. Data were normalised and corrected for sample absorption. Initial stages of the refinement proceeded well in both cases including scale factor, cell constant and a five-point polynomial back- ground. For the tellurite sodalite a more multi-term complex cosine Fourier background was required to fit modulations in the background, probably associated with thermal diffuse scattering from the reorienting Te032- group.Four small excluded regions were also included in the background to remove reflections from poorly crystalline SiO,, from the quartz reaction tube, which were not apparent in the X-ray diffraction pattern. In the case of the telluride sodalite, all variable atomic positions and peak-shape parameters varied smoothly and finally the isotropic temperature factors were added. In the case of the tellurite, although the parameters defining framework geometry refined smoothly the coordinates delineating the occluded species were unstable suggesting the initial model to be incorrect.A better structural model for sodalites containing large asymmetric anions has been proposed by Mead and Weller21 for the Br0,2--containing sodalites and by Felsche and co- workers for acetate sodalite.22 In this description, unrealistically short interactions between the anions and the non-framework cation, M, are alleviated through the introduction of a second type of site for M. This effectively allows displacement of M through a six-ring into a neighbouring cage such that M coordinates only to one side of the anion, Fig. 5. This structural model involves placing a portion of the strontium atoms on (x,x,x) and the remainder on (-xl,-xl,-xl) and this model was therefore adopted in this refinement with ~~0.25 and XI 250.2.Refinement of the strontium atoms on these sites gave a vastly improved fit to the profile with significantly better R \ 'i // Fig.5 Proposed model for the cation displacement caused by the introduction of a large pyramidal anion into the sodalite structure factors. Refinement of the occupancy of the two sites gave 0.504(9) and 0.496(9), respectively. These values are somewhat further from the 0.75/0.25 ratio expected in the simple model proposed initially, where only one quarter of the strontium ions need to be displaced to accommodate the Te0,2- ion. However, the refined values indicate that further strontium ions are displaced to positions that coordinate to the oxygen side of the tellurite ion. Further improvement of the structural model involved dis- placement of the tellurium atom of the TeO, group to an (x,x,x)site with x =0.05, which produces four-fold disordering.This also produced a significant improvement in the fit and is also expected in order to allow room for oxygen atoms within the beta cage. The expected geometry of the tellurite anion is pyramidal; using the information gleaned from the tellurium-edee EXAFS, the three oxygens were expected to be located 1.84 A from the refined tellurium site. Evidence for this rigid body was then searched for in a difference Fourier map calculated using the structure factors obtained from fitting the data with just the framework atoms, strontium and tellurium. Fig. 6 shows one calculated section and ,a possible site for oxygen at (0.12, -0.01, 0.125) about 2A from tellurium and similar to that used by Brenchley and Weller in their model for seo3,12313 that is (x,x,z) [equivalent to (x,z,x)] with x 250.12, z =0.05.Oxygen was introduced into the refinement of this site and a slack cqnstraint introduced setting the Te- 0 distance to 1.84( 1) A as required by the EXAFS data. Refinement of the oxygens on this site, again with four-fold disordering, produced an excellent fit to the d$a though a high temperature factor on the oxygen of 7.4A2; this high temperature factor is probably due to librations of the tellurite group. Similarly, higher temperature factors are generally observed for the oxygens of other tetrahedral and pyramidal occluded species.23 This choice of site for the oxygen of the tellurite group produces 12 possible sites with the unit cell; of these six produce reasonable Te- 0 distances and these six sites may be assigned to two orientations of the TeO, anion.Of note is the 0-Te-0 angle of 102" produced for these two choices, which is in very good agreement with data from other tellurite structures; in Na2T?O3-5H2024 the tellurite group has three oxygens at 1.86 A and an 0-Te-0 angle of 99.5". Final refined param- eters are given in Table 1 and 2 for the telluride and tellurite, 0.5 -0,5 -0.5 0 0.5 X Fig. 6 Section, z =0.125, of the neutron diffraction difference Fourier map of Sr,[A10,],,(Te0,)2 showing a possible site near 0.11, 0.00, 0.125 for the tellurite oxygen J.Mater. Chew., 1996, 6(lo), 1717-1721 1719 Table 1 Refined parameters for Sr, [A102]12Te,; e.s.d.s are given in parentheses site X Y Z B,,, 12d 24g 8c 2a 0.25 0.15627(8) 0.20665(8) 0 0.5 0.15627(8) 0.20665(8) 0 0 0.46618(7) 0.20665(8) 0 0.18 (4) 0.70(2) 0.94(2) 1.54(6) 1111 I I I I I I I 0.4 016 018 1.0 1.2 1.4 1.6 1:s TOF/ms Fig. 7 Final fit to neutron diffraction data for Sr8[A10,],,(Te0,),. atom A1 Sr Te e .-ci’-ol0.0 Fig. 8 Refined orientation of the Te03,- anion in Sr8[A102],2(Te03), showing its interaction with the strontium ions. Only one possible position for the tellurite ion is shown. Table 3 Derived bond distances and angles for Sr, [AlO,],,Te, and Sr, [A10,],,(Te03), sodalites bond distance/& The crosses represent the experimental data points and the upper bond angle/degrees Sr, [AlO,] 12Te, ,,( Sr, [AlO,] TeO,), continuous line the calculated pattern.The lower continuous line represents the difference. Tick marks for the reflection positions are also shown. respectively, and the final fit to the experimental profile is shown in Fig. 7 for the tellurite sodalite. Results and Discussion Bond distances and angles calculated from the refined atomic positions are given in Table 3. In Sr,[AlO,],,Te, the central telluride is tetrahedrally coordjnated to four strontium atoms. The bond distance of 3.355 A is qslightly shorter than that observed for bulk SrSe of 3.401 A, which has the rock-salt structure and hence octahedral coordination to the metal.This behaviour has also been observed in calcium sulfide, strontium sulfide and strontium selenide-bearing sodalites and is presum- ably related to the smaller effective ionic radius of the anion in four-fold, as opposed to the more normal six-fold, coord- ination. The strontium coordination approximates to tetra- hedral with three strong bonds to the oxide ions in the six- ring and a single bond to the central tellurium. There are also three much weaker bonds to the other three oxygens in the ring. The 3+3 coordination to the ring is caused by partial collapse of the cage around the occluded anion. The structure of the tellurite sodalite is more complex due to the large anion which has lower symmetry and is disordered.Placement of the pyramidal TeO, with the tellurium at the cage centre (O,O,O) would make it impossible to site four strontium ions within the unit cell to produce chemically reasonable Sr-0 and Sr-Te distances. For this reason the ~ Al-0 Sr-0 Sr-Te Te-0(Te) Sr-0(Te) Sr(2)-0 0-O(Te) 0-Al-0 x 2 0-Al-0 x 4 A1-0-A1 tilt angle, 4 0(Te)-Te-0(Te) ~~ ~ ~~ ~ 1.737( 1) 2.523( 1) x 3 2.896( 1) x 3 3.355( 1) ----119.30(8) 104.83(8) 145.09(8) 12.21(8) -~~~~ 1.723(4) 2.581 (4) x 3 2.740( 5) x 3 -1.842(2) 3.026( 6) 3 x 0.25 2.791(6) 3 x 0.25 2.794( 6) x 3 2.716(4) x 3 2.979( 5) 119.46( 12) 104.72( 5) 150.22( 12) 2.75( 12) 102.00( 2) tellurite ion is displaced off the cage centre towards one of the six-rings. Obviously a strontium ion located at (x,x,x), ~~0.2,woul4 be unreasonably close to the Te4+ centre (Sr-Te M 2.3 A).Therefore this ion is displaced into the neigh- bouring cage, where presumably it is located on the oxygen side of a TeO, group. Hence in the tellurite sodalite there are two strontium environments to be considered. The strontium centred at (0.25, 0.25, 0.25) has a very similar coordination to that observed for the selenite-containing aluminate sodalite. There are three short and three long distances to the oxides in the six-ring plus two contact distances to the oxygens of the tellurite group. The other strontium atom near (-0.2, -0.2, -0.2) has only six bonds to the ring oxygens It is noteworthy that the two strontium positions and their refined occupan- Table 2 Refined atomic parameters for Sr, [AIO,],,( TeO,),; e.s.d.s are given in parentheses ~ ~~ ~ atom site X Y Z B,,, occupancy A1 12d 0.25 0.5 0 0.80( 11) 1 0.15795( 14) 0 24gSr 8c 0.2426( 3) 0.15795( 14) 0.2426( 3) 0.4924( 5) 0.2426(3) 2.45(6) 1.09( 12) 1 0.504(9) Te 8c 0.0599(2) 0.0599( 2) 0.0599( 2) 2.45(4) 0.25 Sr 8c -0.2027(4) -0.2027(4) -0.2027( 4) 2.54( 16) 0.496( 9) O(Te) 24g 0.1228( 2) 0.1228( 2) 0.0302(3) 7.40( 44) 0.25 ~~ Q =9.42061(8)A; R,, =2.76%; Rexp=0.83%; x2 =11.18.1720 J. Muter. Chem., 1996, 6(10), 1717-1721 cies are very similar to those observed by Engelhardt and co- workers22 for the single-crystal structure determination of acetate sodalite. The cell collapse is almost totally performed by variations in the Al-0-A1 bond angle which can take values between 120 and 160" and is generally larger than that for aluminosilic- ate sodalites where bond ionicites are lower. The degree of collapse can be described by several values as investigated by De~meier.~~Of importance are a and a' which are the tetra- hedral distortion angles for the 0-Al-0 angles in the AlO, tetrahedron, which can distort considerably from the perfect 109.48".The more important term in defining cell collapse is the tilt angle, 4, which describes how much the tetrahedra have tilted out of their 4 axis. 4 can hold values between 0 and 35", where 0 equates to a fully expanded centrosymmetric sodalite with symmetry lm3rn.It can be seen that on changing the cell contents from tellurite to telluride the tilt angle increases from 2.75" to 12.21" to accommodate the smaller anion. The IR spectra of aluminosilicate sodalites have been mod- elled extensively by Creighton et These authors showed that of the fourteen expected IR-active modes of aluminate sodalites, only eight had detectable intensity and only four of these were experimentally observed over the range 300-1000 cm-' with moderate intensity. These four bands have been denoted v,,, vS1, vS2 and 6 for the asymmetric, two symmetric and deformation modes, respectively. The frequency of these bands in aluminate sodalites have been shown to linearly correlate with the cell parameter13 and the IR absorp- tion bands at 878,676,620 and 397 cm-I for the telluride and 884, 718, 622 and 395 cm-I for the tellurite are fully consistent with the previously determined data.Previous studies on the correlations between 27Al MAS NMR spectra and the coordination geometry of framework al~minates~~'~~,~~have shown a linear relationship between the quadrupolar coupling constant (C,) and the tetrahedral distortion, $, measured according to the equation =Itanlax- 109.481 X where ax is one of the six tetrahedral angles (0-A1-0) around the aluminium. Values of C, were determined as C,= 5.91 and 5.96 MHz for the telluride and tellurite sodalites, respectively, from the fitting of the spectra.$ was calculated as 0.672 and 0.685 for the telluride and tellurite sodalites from the structure refinements. These results are in very good agreement with values expected by interpolation of the pre- viously determined re1ati0nships.l~ In addition a linear corre- lation between the 27Al isotropic chemical shift (hiso) and the A1-0-A1 bond angle has been determined" and the values reported in this work [S 75.7, 145.1" (Te2-); 6 75.2, 150.2" (Te032-)] fit this behaviour well. Conclusions The tellurite ions has been incorporated into the sodalite structure for the first time. Previous reports of oxotellurium anions in sodalite structures are limited to the compound Ca2Na, [AlSiO,],( TeO,),, a hauyne derivative reported by Neurgaonkar and H~mmel.~~ However, as the free Te042- ion is not known it is likely that this compound was incorrectly described and is in fact a tellurite.Indeed, our attempts to produce sodalites containing the Te042- ion have been unsuc- cessful, producing tellurites even under very strongly oxidising conditions. Structure determination of sodalites containing complex anions requires the application of a variety of techniques in order to define the species present. In this work a combination of MAS NMR, EXAFS and neutron diffraction was required in order to define the nature and position of the oxoanion. Reduction of tellurite ion to telluride can be accomplished inside the sodalite cage, showing that intracage reactions are accomplished easily.Attempts to oxidise tellurite to tellurate by heating in a stream of oxygen at various temperatures were unsuccessful. As the sodalite structure often stabilises tetra- hedral species, e.g. permanganate is stable to 500°C inside a sodalite cage,30 it might be expected that the tellurate ion could be generated inside the aluminate sodalite. However, the framework of Sr, [A102]12(Te03)2 is rather expanded (Al-0-A1 angle 150.2", tilt angle 2.75") and further expansion to accommodate the tellurate ion is impossible. We thank the EPSRC for a grant in support of this work, David Apperley at the solid-state NMR service at Durham for running the 27Al and lzsTe NMR spectra and Dr. Richard Oldroyd for collection and refinement of the Te Lm-edge data.References 1 R. M. Barrer and J. F. Cole J. Chem. SOC. A, 1970, 1516. 2 D. Taylor and C. M. B. Henderson, Phys. Chem. Miner., 1978, 2, 325. 3 M. T. Weller and G. Wong, J. Chem. SOC., Chem. Commun., 1988, 1103. 4 S. E. Dann and M. T. Weller, Inorg. Chem., 1996,35, 555. 5 J. S. Prener and R. J. Ward, J. Am. Chem. Soc., 1988,72,2780. 6 I. Chang, J. Electrochem. Soc., 1974, 121, 815. 7 F. Hund, Z. Anorg. Allg. Chem., 1984,511,225. 8 W. Depmeier, Acta Crystallogr., Sect. B, 1988,44,201. 9 W. Depmeier, Acta Crystallogr., Sect. C, 1987,43,2251. 10 W. Depmeier, Acta Crystallogr., Sect. C, 1984,40,226. 11 W. Lowenstein, Am. Mineral, 1954,39,92. 12 M. E. Brenchley and M. T. Weller, J. Muter. Chem., 1992,2, 1003.13 M. E. Brenchley and M. T. Weller, Chem. Muter., 1993,5,970. 14 R. Kondo, Yogyo Kyokai Shi., 1965,71, 1. 15 W. Depmeier, Kristall. Technik., 1972, 7, 229. 16 N. Binsted, PAXAS: EXAFS Analysis Program, 1988. 17 N. Binsted, S. J. Gurman and I. Ross, J. Phys. C.: Solid State Phys., 1984, 17, 143. 18 R. K. Harris and B. E. Mann, NMR and the Periodic Table, Academic Press, London, 1978. 19 M. T. Weller, M. E. Brenchley, D. C. Apperley and N. A. Davies, Solid State Nucl. Magn. Reson., 1994,3, xxx. 20 A. C. Larson and R. B. Von Dreele, GSAS: Generalised Structural Analysis System, Los Alamos, NM, 1990. 21 P. J. Mead and M. T. Weller, Zeolites, 1995, 15, 561. 22 P. Sieger, A. M. Scheider, M. Wiebcke, P. Behrens, J. Felsche and G. Englehardt, Chem. Muter., 1995,7, 163. 23 M. E. Brenchley and M. T. Weller, Zeolites, 1994, 14, 682. 24 E. Philippot, M. Maurin and J. Moret, Acta Crystallogr, Sect. B, 1979,35,1337. 25 W. Depmeier, Acta Crystallogr., Sect. B, 1984,40, 185. 26 J. A. Creighton, H. W. Deckman and J. M. Newsam, J. Phys. Chem., 1991,95,2099. 27 G. Engelhardt, H. Koeller, P. Sieger, W. Depmeier and A. Samoson, Solid State. Nucl. Magn. Reson., 1992, 1, 127. 28 G. Engelhardt and W. J. Veeman, J. Chem. SOC., Chem. Commun., 1993,622. 29 R. R. Neurgaonkar and F. A. Hummel, Muter. Res. Bull., 1976, 11,61. 30 M. T. Weller and K. E. Haworth, J. Chem. SOC.,Chem. Commun., 1991,734. Paper 6/03606F; Received 23rd May, 1996 J. Mater. Chem., 1996,6(lo), 1717-1721 1721

ISSN:0959-9428    

DOI:10.1039/JM9960601717

出版商:RSC    

年代:1996

数据来源: RSC