|
11. |
Growth of PbS thin films from novel precursors by atomic layer epitaxy |
|
Journal of Materials Chemistry,
Volume 4,
Issue 9,
1994,
Page 1409-1412
Erja Nykänen,
Preview
|
PDF (518KB)
|
|
摘要:
J. MATER. CHEM., 1994,4(9),1409-1412 Growth of PbS Thin Films from Novel Precursors by Atomic Layer Epitaxy Erja Nykanen,*"Jutta Laine-Ylijoki," Pekka Soininen," Lauri Niinisto," Markku Leskelaband Liliane G. Hubert-Pfalzgraf" Laboratory of Inorganic and Analytical Chemistry, Helsinki University of Technology, FIN-02750 Espoo, Finland b Department of Chemistry, University of Helsinki, FIN-00074 Helsinki, Finland Laboratoire de Chimie Moleculaire, Universite de Nice, F-06034 Nice, France Two lead tert-butoxide complexes (1) [Pb(OBu')& (rn=3 or 2 in the solid and gaseous phase, respectively) and (2) Pb,0(OBu')6 were used as precursors for atomic layer epitaxy (ALE) deposition of PbS thin films. The growth on soda lime glass, with and without an alumina coating, was studied by varying the source furnace and the substrate temperatures as well as the total number of cycles.Pb(thd), (3)and Pb(dedtc), (4) were used for comparison while H2S served in all experiments as the sulfur source.? The films obtained were smooth and generally highly crystalline. The substrate temperature had a strong effect on the growth rate of PbS thin films. Nevertheless, in the self-controlled region of ALE growth the tert-butfxide complexes gave a significantly higher growth rate than the other source chemicals, with a maximum of 0.9 A per cycle at 150 "C. Upon sublimation 1 is converted to 2, which contains four Pb atoms in a tetrahedral arrangement; this may cause the higher growth rate. Thermogravimetry/differential thermal analysis curves and mass spectrometric data were measured for all precursors. As the butoxide and thd complexes (1-3)are thermally unstable the useful ALE prosessing windows (temperature/pressure) are narrow compared to the much more stable dedtc complex (4).Lead chalcogenides have interesting photoconducting proper-ties and PbS especially has widely been utilized as a sensor material because of its high response in the near infrared (1-3 pm). PbS detectors are of low cost and they are relatively easy to prepare. Most of the commercially applied PbS films are chemically deposited but vacuum evaporation has also been used.' The photoconductivity properties, mainly carrier concentration and lifetime, are affected by the stoichiometry of the films and their doping (by oxygen, for example).It appears that the precursor has a significant effect not only on the growth rate but also on the properties of the films. The recently synthesized lead complexes (1) [Pb(OBu'),], (m=3 or 2 in the solid and gaseous phase, respectively) and (2) Pb,o(oB~')~are potential source candidates for gas-phase deposition of PbS because of their ~olatility.~,~ ALE is a novel thin-film deposition process which exploits the exchange reactions between chemisorbed species on the ALE growth takes place over a relatively narrow~urface.~ temperature range, the so-called self-controlled region, where a monolayer or fraction thereof is chemisorbed on the surface. It has been successfullyapplied to prepare a number of 11-VI compounds, even on a commercial scale.5Previously, we have reported the preparation of PbS thin films by the ALE process using various starting materials such as lead halides and acetate as well as the lead chelates of thd and dedtc.6 Here we report the growth and structure characteristics of PbS films from novel precursors and compare them with new data for the thd and dedtc chelates; the photoconductivity proper-ties will be reported in a subsequent paper.' Experimental The PbS films were deposited by the ALE method in a hot-wall flow-type reactor F-120 manufactured by Microchemistry Ltd, Espoo, Finland.Soda lime glass (5x 5cm2), with and ~ t thd =2,2,6,6-tetramethylheptane-3,5-dione, dedtc =diethyldithio-carbamate.without an alumina coating, was used as substrate. Two additional volatile lead precursors, uiz. Pb(thd), (3) and Pb(dedtc), (4) were used for comparison. The butoxide complexes, 1 and 2 were synthesized at the Laboratoire de Chimie MolCculaire, UniversitC de while Pb(thd), and Pb(dedtc), were prepared in Espoo. The sulfur source was in all experiments H2S (purity 99.999%) and the flow rate was 5x dm3 min-'. Nitrogen (purity 99.999%) was used as an inert purge gas between reactant pulses. The primary N2 flow was 0.7 dm3 min-' and the reactor pressure measured from the pumping line was around 3 mbar. Thermal analysis of the starting materials was used to give an estimate of the temperature sufficient for volatilization and especially to test the stability of tert-butoxide complexes.Simultaneous thermogravimetry (TG) and differential thermal analysis (DTA) curves were recorded in a Seiko instrument (TG/DTA 320, series SSC5200) both in a vacuum (5-10 mbar) and in flowing nitrogen. Mass spectrometry (MS)studies were carried out in order to obtain additional information on the volatilized species using a JEOL MS DX-303 DA5000 and VG 7070E high-resolution mass spectrometers and ionization with 70 eV electron bombardment. The JEOL instrument had a sample inlet with programmable heating which facilitated comparisons with the TG runs. In order to optimize the sublimation temperature of the starting materials they were studied under process contlitions by varying the source furnace temperature in the range 100-140 and 200-240 "Cfor Pb(thd), and Pb(dedtc),, respect-ively.The substrate (reactor) temperature in these experiments was 330 "C.For the butoxide complexes, a reactor temperature of 300°C was used and the source temperature range tested was 100-139°C. PbS thin-film growth was then studied by varying the temperature of the substrate (reactor) between 130 and 390°C in the case of Pb(thd), and tert-butoxides. The source temperature for Pb(thd), was 120°C tert-Butoxides were vaporized at 115"C(1)and 130 or 139 'C (2). The growth from Pb(dedtc), [T(source) =210 "C] was studied at 220-340 "C. The pulse times for lead chemicals and hydrogen mlfide 1410 were typically 0.4 s. The effect of purge time after the Pb(thd), pulse was studied by varying the purge times from 0.2 to 3.0 s at substrate temperatures of 135 and 330°C.The purge time normally used was 0.4 s. Typically the number of growth cycles was 10000, but in the case of Pb(thd), the film growth was also studied as a function of the number of cycles. The film thicknesses were determined by X-ray fluorescence and profilometry. X-Ray diffraction (Philips MPD 1880, Cu-Kor radiation) was used to determine the crystallinity and preferential orientation. Results and Discussion Thermal Analysis Both butoxide compounds were studied by TG/DTA under different experimental conditions in order to establish their thermal stability and optimum conditions for volatilization.The main emphasis was on experiments which simulated the conditions inside the ALE reactor (pressure 5-10 mbar). Because 1 first condenses to 2 by the reaction 2[Pb(OB~')~],+Pb,0(0Bu'), +But20 its thermograms are more complicated than those of 2 (Fig. 1 and 2). The DTA curve of 1 especially shows a more complex behaviour and contains an endothermic peak at 380 "C, possibly indicating a phase transition. The final decomposition product at 400 "C was, according to XRD analysis, PbO (massicot) for 2 and a mixture of various lead oxides (mainly massicot) for 1. The sublimation of 2 is more nearly complete when a slow heating rate is applied (cf.Fig. 2). When too high a heating rate in an I 1100 90 h8 W$ 5.5 ?0 80 0 -0.5 -0c a 70 I I I I 0 100 200 300 400 500 TI'C Fig.1 (a) TG and (b) DTA curves in vacuum (7 mbar) for [Pb(OBut),12 (1). The sample size was 42mg and heating rate 10"C min-'. I100 90 8 v 0 80 I-70 700 200 300 400 500 TI'C Fig. 2 (a) DTA curve in vacuum (7 mbar) for Pb,O(OBut), (2). (b); (c) TG curves for sample sizes and heating rates of 12 mg, 25 "C min-(b)and 15 mg, 2.5 "C min-' (c). J. MATER. CHEM., 1994, VOL. 4 intermediate pressure (5-10mbar) is used there is a danger of the complex decomposing in the solid state before subli- mation. Compound 3 is thermally more stable than 1 and 2, but the heating rate affects the TG curves. With a faster rate (25 "C min-') the residue is small (< lo%), but when heated slowly (2.5"C min-l) the decomposition of the 3 becomes significant and the amount remaining is c~i.30% (Fig.3). Compound 4 was thermally the most stable complex studied and complete volatilization was achieved (Fig. 4). Growth Experiments Fig. 5 summarizes the growth experiments with three starting materials and two types of substrate. The thicknesses of the PbS thin films varied between 70 and 830nm or 20 and 900 nm depending on whether pure soda lime glass or alum-ina-coated glass was used as the substrate. Because 1 trans-forms into 2 upon sublimation both complexes were used in growth experiments. The chemical nature of the starting material and the substrate temperature had a strong effect on the growth rate of PbS films. At all temperatures the tert-butoxide complex containing four Pb atoms gave a signifi- cantly higher growth rate than the other source chemicals.Also with this compound the growth on both substrate types was very similar. The self-controlled growth windowo was 130-180 "C but although a growth rate as high as 0.89 A per cycle was reached, three or four ALE cycles are needed to produce one complete monolayer of PbS. The linear decrease of growth rate when the substrate (reactor) temperature roo3$ 10 n g -1c C a 1 I I I I' -301 100 200 300 400 500I 0 T/"C Fig. 3 (a) DTA curves in vacuum (7 mbar) for Pb(thd),. (b),(c) TG curves for sample sizes and heating rates of 17 mg. 25 "C min-' (b) and 15 mg, 2.5 "C min-' (c). 100 60 -8 U P 20 I I I I 0 100 200 300 400 500 T/"C Fig.4 (a) DTA curve in vacuum (7 mbar) for Pb(dedtc)2. (b),(c)TG curves for sample sizes and heating rates of 11 mg, 25 "C min-' (b) and 5 mg, 2.5 "C min-' (c). J. MATER. CHEM., 1994, VOL. 4 1411 hU 5 Y n n. 391 I 100 200 300 400 500 6bO m Iz Fig. 7 Mass spectrum of Pb(thd), showing the different lead-containing species -substrate T/OC 160-Fig. 5 Growth rates of PbS thin films us. substrate temperature from E 120-the tert-butoxide complex (2) (triangles), Pb(dedtc), (circles) and \ c Pb(thd), (squares). An open mark indicates alumina-coated glass and v) filled one soda lime glass substrate. The source temperatures used for!!?80-r Pb40(OBut)6,Pb(dedtc), and Pb(thd), were 115-130 "C,210 "C and 120"C, respectively.exceeds 180°C is probably due to the decomposition of the precursor (cf. TG/DTA data) before the substrate chamber which leads to a diminished supply of molecules onto the substrate surface. MS results corroborated the decomposition of Pb,0(OBut)6 molecules into several species of lower mol-ecular weight and lead content (Fig. 6). The thermal instability of the precursor is further demonstrated by the thickness profile of the films. The ALE window for Pb(thd), is alsoo very narrow (130-150nC) and the growth rate of 0.60A per cycle is cu. 30% lower than with the tert-butoxide complex. For reactor temperatures of 160-270 "C the growth rate decreases on both substrates linearly and this corresponds to the TG decomposition temperature (Fig.3). Etching of the PbS film by the free ligand or its fragments is another factor which contributes to the dramatic decrease in growth rate. It has recently been established that the free ligand Hthd exhibits a strong etching action on metal-containing thin films such as oxides or sulfide^.^^^ At substrate temperatures >270 "C there appears to exist another the:modynamical equilibrium, yield: ing a growth rate of 0.10A per cycle on glass and 0.05 A per cycle on alumina-coated glass. The second ALE window is probably caused by the fractionation of the precursor to another reactive species. The fractionation was verified by an MS study, and the results indicate that under relatively extreme conditions (uhv, high volatilization temperature) the predominant gaseous species is not Pb(thd), but Pb(thd) (Fig.7). 1001 -7 1241 ' 8 "'.I "".II'''. . 1000 1100 1200 1300 m Iz Fig. 6 High-resolution mass spectrum of Pb,0(OBu')6 (M= 1282 g mol-l). Numerous different Pb,-, Pb,-, Pb,-and Pb-containing dissociation fragments can be seen. 0 0 5000 10000 15000 20000 no.of cycles Fig. 8 Growth of PbS from Pb(thd), (TSOUrCe=120"C) as a function of ALE cycles at a substrate (reactor) temperature of 330 "C Pb(dedtc),, which has a Pb-S bond, differs chemically and structurally both from Pb(thd), and lead tert-butoxide mol-ecules and consequently the growth behaviour us. substrate temperature differs from that found for the other source materials.A sufficient dose of Pb(dedtc), needed to cover the whole substrate area is reached first at source temperature >210 "C with a pulse time of 0.4 s. For this reason the growth mechanism was studied above the substrate temperature of 220°C and constant growth was reached at 240°F on soda lime glass (Fig. 7). The growth rate was 0.43-0.46 4 per cycle. At 260-300°C the growth rate decreased to 0.32 A pcr cycle and again it seems that there is another equilibrium jtate at this level. MS results indicate the partial fractionation of Pb(dedtc), to Pb(dedtc), but not as completely as in the case of Pb(thd),. On alumina-coated soda lime glass, PbS was successfully grown only at 230-260 "C when Pb(dedtc), was used.At higher temperatures no growth was observed on alumina-coated substrate and on soda lime glass at these temperatures the crystallinity was also poor. The stability of surface saturation was studied by changing the purge time after the Pb(thd), pulse from 0.2 to 3.0s on both the self-controlled growth levels at reactor temperatures of 135 and 330°C. No decrease in growth rate was observed. The nucleation mechanism was studied at a reactor tempera-ture of 330°C using Pb(thd), as a starting material and varying the number of ALE growth cycles. The thickness grew linearly (Fig. 8) above 3000 cycles, indicating two-dimensional nucleation, which is typical for the ALE process. XRD Results XRD curves indicated in all cases the polycrystalline cubic PbS structure and they resembled the PbS powder diffraction data, although the highest reflection was not always (200).Fig. 9 shows typical XRD diagrams of PbS thin films in the self-controlled growth region prepared from Pb(tiedtc),, Pb(thd), and teut-butoxides. In the case of Pb(thd), and tert-butoxide complexes, the IIIIII 20 50 60 70 26/degrees Fig.9 X-Ray diffraction curves of PbS thin films grown from (a) Pb(dedtc),, (b) Pb(thd), and (c) Pb,O(OBu'), on soda lime glass. The substrate temperature during the growth was 240°C for (a) and 150 "C for (b)and (c). Fig. 10 Schematic comparison of the arrangement of metal atoms in the structures of (a)Pb,O(OBu'), and (b)Zn40(CH3C0,),3.'o growth rate began to decline above the self-saturated tempera- ture range and the orientation of the films was different.At reactor temperatures 180-240 "C the intensity of the (200) reflection decreased while (220) became the strongest, the (111) and (311) reflections being nearly as high. At substrate temperatures >240-270 "C the rather strong (311) reflection was always present together with either the (111) or in a few cases with the (200) reflection; the last reflection was most clearly seen with Pb(thd),. The films grown on alumina-coated glass substrates were highly oriented towards (200) at all reactor temperatures when Pb(thd), and Pb(dedtc), were used as the starting chemicals. The (200) peak was also very intense, although the films were thin.On the other hand, the growth of PbS films from lead tert-butoxide was similar on both substrates. The temperature dependence of the orientation of PbS thin films grown from Pb(dedtc), on glass differs from that of J. MATER. CHEM., 1994, VOL. 4 Pb(thd), and tert-butoxides. The (111) reflection is slightly stronger at reactor temperatures of 240-280 "C. while (200)is the strongest reflection below and above this temperature range. Thinner films (d < 100 nm) favoured the ( 200) direction more strongly than thicker films. Conclusions Polycrystalline PbS thin films were obtained using volatile lead tert-butoxide, Pb( thd), and Pb(dedtc), complexes as precursors together with H,S. The substrate temperature and the starting chemical caused the strongest effects on the growth rate andoorientation of the thin films.Growth rates as high as 0.9 A per cycle were achieved with lead tert-butoxide at relatively low substrate temperatures. This growth rate was significantly higher than those for the other precur- sors and is probably connected with the oligomeric structure of the tert-butoxide complex, a similar case of enhanced ALE growth rate has been observed earlier with Zn,O(CH,CO,), us. ZnC1, (Fig. 10).lo,llThe presence of Pb-0 bonds together with the low thermal stability of the butoxides (1, 2) make the optimization of the process conditions difficult and may lead to thin films with a thickness profile. Also the thd complex (3) displays incomplete volatilization but to a lesser extent.In this respect, the more stable dedtc complex (4), which already has the Pb-S moiety, behaves in the most controllable way of the four precursors studied. The authors thank Mr. J. Rautanen for the profilometry measurements and Ms. Anita Pirhonen for helping with the XRD measurements. Mr. P. Sarkio and Ms. Kirsti Wiinamaki are thanked for recording the mass spectra. References 1 T. H. Johnson, Proc. SPIE, 1984,443,60. 2 R. Papiernik, L. G. Hubert-Pfalzgraf and M-C. Massiani, Inorg. Chim. Actu, 1989, 165, 1. 3 R. Papiernik, L. G. Hubert-Pfalzgraf and M-C. Massiani, Polyhedron, 1991, 10, 1657. 4 T. Suntola, Muter. Sci. Rep., 1989,4, 261. 5 L. Niinisto and M. Leskela, Thin Solid Films, 1993,225, 130. 6 M. Leskela, L. Niinisto, P. Niemela, E. Nykiinen, P. Soininen, M. Tiitta and J. Vahakangas, Vacuum, 1990,41. 1457. 7 M. Leskela, L. Niinisto, P. Niemela, E. Nykhen, J. Rautanen, P. Soininen and J. Vahakangas, to be published. 8 F. Rosseau, A. Jain, T. T. Kodas, M. Hampden-Smith, J. D. Fair and R. Munchausen, J. Muter. Chem., 1992, 2, 893. 9 J. Aarik, A. Aidla, A. Jaek, M. Leskela and L. Uiinisto, J. Muter. Chern.in the press. 10 L. Hiltunen, M. Leskela, M. Makela and L. Niinisto, Acta Chem. Scund. Ser. A, 1987,41,548. 11 M. Tammenmaa, T. Koskinen, L. Hiltunen and L. Niinisto, Thin Solid Films, 1985, 124, 125. Paper 4/00704B; Received 4th February,1994
ISSN:0959-9428
DOI:10.1039/JM9940401409
出版商:RSC
年代:1994
数据来源: RSC
|
12. |
Thermodynamic and kinetic properties of lithium insertion into titanium misfit layer sulfides |
|
Journal of Materials Chemistry,
Volume 4,
Issue 9,
1994,
Page 1413-1417
Pedro Lavela,
Preview
|
PDF (609KB)
|
|
摘要:
J. MATER. CHEM., 1994, 4(9), 1413-1417 Thermodynamic and Kinetic Properties of Lithium Insertion into Titanium Misfit Layer Sulfides Pedro Lavela, Julian Morales and Jose L. Tirado Departamento de Quimica lnorganica e lngenieria Qurinica, Facultad de Ciencias. Universidad de Cordoba, San Alberto Magno s/n, E-14004 Cordoba, Spain The thermodynamics and kinetics of the electrochemical intercalation of lithium into polycrystalline (PbS),., 8(TiS2)2, (BiS)1.14(TiS2)2,(PbS),.,8TiS2 and (SnS)1.20TiS2 have been studied by coulometric titration techniques. The highest values of the Gibbs energy of intercalation were obtained with (MS), +6(TiS2)2 stoichiometries, particularly in (PbS)1,,8(TiS2)2. The chemical diffusion coefficients of lithium for all compounds hardly varied with the amount of intercalated lithium and the maximum value, 6.5 x lo-’’ cm2s-’, was found in (PbS)l.18(TiS,),.Micropolarization tests also showed that fast kinetics and better reversibility occurred in sulfides with (MS), +&TiS,), stoichiometry, probably due to the presence of true van der Waals gaps created at the TiS,/TiS, interface. Titanium ternary sulfides with (MX), +6(TiS2), (M =Sn, Pb, Bi) stoichiometry belong to a wide family of inorganic misfit layer compounds which are characterized by the stacking of two different types of layers, a triple layer of TiS, composition and a double layer of MS composition.’ The 1+6 subindex expresses the misfit between both sublattices and n the stacking sequence along the c axis (n= 1 MS-TiS2-MS-TiS2-and n =2 MS-TiS2-TiS2-MS-TiS,-TiS2-).Sublattices are held together by weak interactions, and several empty sites of different geometry are created at the interface which are ready to be occupied by several guest species by means of a topotactic reaction mechanism.2 The atoms that define an MS/TiS, interface are shown schematically in Fig.I@). This is the unique interlayer in sulfides with n=l. As observed, three sulfur atoms of the TiS, layer (open circles) and one sulfur atom of the MS layer (shaded circles) give rise to a distorted tetrahedral site. Hence, 1+6 vacant pseudotetra- hedral sites are found at the MS-TiS2 interlayer. In addition, a true van der Waals gap exists between two adjacent TiS, layers in misfit layer sulfides with n =2.An idealized projection along [OOl] and the disposition of the octahedral and tetra- hedral sites are shown in Fig. l(h). One octahedral and two tetrahedral sites appear in the TiS,-TiS2 interlayer per formula unit. Thus, the total number of vacant sites is 4 +6 per formula unit in the sulfides with this stacking sequence. However, the geometry and size of the empty positions are not the only factors to take into account in these processes. Charge-transfer phenomena have been suggested between the adjacent MS and TiS2 layers, In this way, changes in the formal oxidation states have been found in the metal atoms of both sublattices. This is particularly evident when M is a lanthanide element. An excellent example is provided by ‘PbNb2S5’ and ‘SmNb,S,’ compound^.^ Thus, while ‘PbNb,S,’ easily intercalates lithium in contact with n-butyllithium, SmNb,S, has a low affinity to intercalate lithium under similar conditions.A controversy exists for misfit layer compounds with SnS, PbS and BiS as monosulfide double layers. According to Ohn~,~photoelectron spectra gave evidence for the occurrence of charge transfer in (PbS)l +,TiS, and (SnS), +6NbS2. However, Ettema et al., concluded from X-ray photoelectron spectra that no significant charge transfer took place from MS to the TiS2 part. Structural and compositional aspects of the chemical and electrochemical lithium insertion reaction in misfit layer com- pounds have been reported Different intercalation behaviour was shown to be due to the specific geometry of the empty sites of the sulfides corresponding to the different stacking sequences of the MS and TiS2 layers. .....MS und cell (a 0sulfur atom in TiS2 slab -sulfur atom in MS slab TiS2 unit cell 6B M atom in MS slab diffusion path (b 1 sulfur atom in upper TiS, slab A tetrahedral site 0sulfur atom in lower TiS2 slab 0oc&hedral sit@ Fig. 1 Schematic projection along [OOl] of a misfit compound: (a) MS/TiS, interface, sublattice ratio used, uMS/uTiS2=5/3. (b) TiS,/TiS, interface. Arrows indicate the diffusion paths for lithium ions. In this paper, we report the compositional variation of lithium chemical potential in four titanium misfit layer sulfides and the measurement of lithium diffusivity using electrochemi- cal techniques.Experimental (PbS)l .18(TiS2)2 7 (BiS)l.14(TiS2)2 3 (PbS)l.1gTiS2 and (SnS), ,20TiS2 misfit layer sulfides were prepared by heating stoichiometric mixtures of the constituent elements (Strem Chem) in evacuated silica ampoules under conditions described elsewhere.'-l' The phase purity and the structure of the original phases were checked by X-ray powder diffraction and electron microscopy. Equilibrium open-circuit voltage (OCV) composition curves were obtained by coulometric titration. A two-electrode cell was used. The cell consisted of a cathode pellet (ca. 20mg of the material pressed at 200 MPa) of 7 mm in diameter sup- ported on a copper substrate of porous glass-paper discs soaked in the electrolyte and of a lithium disc anode (Strem Chem).The electrolyte was 1mol dm-3 LiC104 in propylene carbonate (PC). Constant currents of 1pA to several hundred pA were supplied from an Amel 549 in the galvanostatic mode. The lithium was inserted stepwise and the cell was allowed to equilibrate. Equilibrium was considered to have been reached when the OCV remained approximately con- stant (1 mV h-'). Chemical diffusion coefficients, D, were determined by using the current pulse relaxation technique introduced by Basu and Worrell.12 The relaxation voltage was measured with a Yokogawa 7561 digital multimeter. A three-electrode cell (using lithium as the reference elec- trode) was subjected to a cathodic scan with increasing current from 0 to 400 pA cmP2 at a sweep rate of 3 pA s-l and then decreasing to zero.The anodic scan was recorded under the same experimental conditions. These micropolarization tests were carried out in a programmable Schlumberger SI1286 electrochemical interface galvanostat controlled by a Hewlett Packard 286 computer. The handling of all materials, cell fabrication and sub- sequent measurements were carried out in an MBraun dry- box under an argon atmosphere. Results and Discussion OCV curves of the Li/misfit layer sulfide cells are shown in Fig. 2. The discharge curve profiles of ( SnS),.,,TiS2 and (PbS),.,, TiS, are similar and show a continuous decrease in voltage with the degree of intercalation, x. This region extends up to a lithium molar concentration of ca.0.5 and the voltage reaches a value of 1.5 and 1.6 V for the Sn and Pb compounds, respectively. This behaviour indicates that a homogeneous single phase exists in the electrode over this insertion range. t 1 0.0 0.8 1.6 0.0 0.8 1.6 x in Li,(MS), +s(TiS,), Fig. 2 Coulometric titration curves for lithium intercalation in several titanium misfit layer sulfides: (a) (PbS),,,,TiS,,(b) (SnS),,,,TiS,, (4 (PbS)1.18(TiS2)2,(4(BiS)1.14(TiS2)2. J. MATER. CHEM., 1994, VOL. 4 For higher degrees of intercalation, the voltage was practically independent of x. For (SnS),.2,TiS2 and (PbS),.,,TiS, at x= 0.25 and 0.30, the X-ray diffraction patterns showed double lines in the position where multiple-order reflections of the basal spacings would be expected. The first set of lines was consistent with the presence of unlithiated phases, while th? second led to an increase of periodic length of 0.12 and 0.08 A for the tin and lead compounds, respectively.These lithiated phases, isostructural with the parent compound, were the only product present for lithium contents close to x=0.8. However, the intensity of these reflections decreased markedly for lithium contents >0.5, indicating a gradual amorphization of the lithiated phase. Note that no trace of diffraction lines corresponding to decomposition products such as SnS, PbS, Sn, Pb or Li,S was observed. The OCV curve of the Li,( PbS)1.18(TiS2)2 electrode showed a more complex profile and was characterized by the occurrence of different regions with variable length at 0.1<x <0.3 and 0.5 <x <0.8, where the voltage undergoes little change with composition.The profile complexity is a consequence of the occurrence of new metastable phases, which are associated with the formation of superstructures due to lithium-ion ordering. A detailed description of the X-ray diffraction data of these intermediates has been reported el~ewhere.~ The OCV curve of (BiS)1.14(TiS2)2 gradually decreased down to a lithium content close to 0.1 and then a pseudo- plateau in the composition range 0.1 <x <0.6 was observed. Two lithiated phases were detected but in one of them the expansion of the lattice is difficult to explain by means of single geometrical factors.( BiS)1.14(TiS2)2has a more complex structure than the remaining misfit layer compounds studied here. (001)Zone electron diffraction patterns of this compound reveal extensive twinning and preferred relative orientations of BiS slabs." Moreover, bismuth-containing misfit layer sulfides have Bi atoms in 2+ formal oxidation state, an unusual valence in Bi compounds. These structural and elec- tronic properties could be the origin of the sudden change in slope of the OCV slope for low lithium contents. Nagelberg and Worrell13 proposed a method for calculating the standard Gibbs energy of intercalation, AintGo,the Gibbs energy change associated with the insertion of an alkali metal into the vacant sites of the host lattice.This parameter can be obtained from equilibrium voltages of the open-circuit curves, according to eqn. (1): where F is the Faraday constant and the integral represents the area under the OCV curve at a specific intercalation degree. Integration of the E against x data gives the AintGo values shown in Fig. 3. Significant differences appear for a lithium content >0.2 F mol-', in particular for (PbS)1.18(TiS2)2. In fact at 1 F mol-', the intercalated Li( PbS)1,18(TiS2)2 compound has a more negative standard Gibbs energy of intercalation than the rest of the intercalated sulfides. A higher value of AintGo means a higher average value of the lithium chemical potential. Although the values obtained for Li,( BiS)1,14(TiS2)2 are lower than those of Li,(PbS)l.18(TiS2)2, they are still higher than those of intercalated sulfides with stoichiometry (MS), + ,TiS, with a stacking sequence MS-TiS,-MS-TiS,.The peculiar structural complexity of bismuth sulfide" can explain the difficulty of having low-energy sites available for alkali-metal ions. The subsequent formation of a metastable phase would require a major structural disturbance of the parent host matrix. The standard Gibbs intercalation energy for lithium- J. MATER. CHEM., 1994, VOL. 4 300r 0 v -L 2001 0 0V b V 7 V 0 0" 04'1001 vHb 50 8 I' Ol I 0.0 0.2 0.4 0.6 0.8 1.o x in Li,(MS), + &(TiSp), Fig. 3 Composition dependence of lithium intercalation Gibbs energy for several titanium misfit layer sulfides: 0, (SnS),,,,TiS,; @, (PbS)1.18TiS2;v, (BiS)1.14(TiS2)2;v?(PbS)1.18(TiS2)2.intercalated (SnS),.,,TiS, is greater than that for Li,(PbS),,,,TiS, and less negative than those of Li,( MS), +,(TiS2), compounds. These results are basically in agreement with the above description concerning the size and shape of the vacant sites of both families of compounds. The standard Gibbs energies for intercalated misfit layer compounds at 1 F mol-' are compared in Table 1 with that obtained for LiTiS, .14 Sulfides with (MS), +,TiS, stoichi-ometry have lower values than LiTiS,. On the other hand, the differences are less significant in those sulfides, (MS)1+6(TiS2)2, with true van der Waals gaps.In fact, Li( PbS)1.18(TiS2)2 has a more negative standard Gibbs energy than the intercalated titanium disulfide, as expected from the greater voltage recorded in its OCV curve. Lithium ion diffusivities, D, were determined from the following equation: where AE is the deviation of the voltage from the equilibrium, rn the slope of the OCV curve, V, the molar volume, I the applied current, zthe pulse time and A the electrode/electrolyte contact surface area. Since no measurements of particle size and shape were performed, the calculations were based on the geometrical surface area of the electrode, 0.385 mm2. A current pulse of 500 pA cm-, was applied for 10s. Since D and m are directly related, the chemical diffusivity will tend to zero in those regions of the OCV curve which define a constant-voltage plateau.This is equivalent to having a two-phase region whose kinetics consists of a coupled process of phase transformation and diffusion. Thus, lithium chemical diffusion coefficients were determined in those regions characterized by a continuous decrease in voltage in the coulometric titrations curves (Fig. 2). The compositional ranges selected were O<x<O.l and O<x G0.4 F mol-' for Table 1 Gibbs energy of intercalation for lithium-intercalated titanium disulfide and titanium misfit layer sulfides compound A,,G"/kJ mol-' -239.9 -176.5 -189.2 -275.3 -226.2 1415 -9.0 j-cu U E9-10.0 Qv m-0 -11.01 I I t, 0.0 0.1 0.2 0.3 0.4 x in Li,(MS), + s(TiS2), Fig.4 Composition dependence of the chemical diffusion coefficient of Li (D) in polycrystalline (PbS),,,,TiS, (@), (SnS),,,,TiS, (0), (PbS)1.18(TiS2)2 ('1 and (Bis)1.14(Tis2)2 1-(MS), + TiS,), and (MS), + ,TiS, compounds, respectively. Fig. 3 shows the composition dependence of the chemical diffusion coefficients of the different intercalates. For these ranges of composition, the chemical diffusion coefficients barely changed with the degree of intercalation. According to the geometry and atomic environment of the vacant sites created at the TiS,/TiS, interface [Fig. l(b)] the diffusion paths determined by the interconnected octahedral and tetra- hedral sites of the close sulfur packing provide a more adequate path for lithium ion mobility than those of the MS/TiS, interface.In this latter case [Fig. l(a)] lithium ions can occupy pseudotetrahedral sites formed by three sulfur atoms of the TiS, slab and one sulfur atom of the MS layer. The translational movement of lithium ions through the interlayer space would take place by means of a jump from one pseudotetrahedral site to any other of the next-nearest- neighbour vacant sites. This involves passing through the vicinity of the surrounding M atoms that protrude from the exposed side of the MS double layer. This also means that M atoms should have some blocking effect on the lithium diffusion, thus increasing the energy barrier for lithium jump- ing as a consequence of the repulsive component of the electrostatic interactions.In addition, true van der Waals forces which join the TiS, slabs in the (MS)l+s(TiS2)2 Sulfides allow a greater expansion of the interlayer space as described elsewhere.8 Thus, according to this diffusion model, lithium should be less mobile in misfit layer sulfides with (MS),+BTiSZstoich-iometry than in those with (MS), +6(TiS2)2 stoichiometry. Most data obtained so far [(PbS)1.13TaS2 ,I5 (SnS)1.,6 TaS2,15 ,16(PbS)1.14(NbS2)2 (BiS)1.17(NbS2)217 and some of' those included in Fig. 41 are in agreement with this model. However, the chemical diffusion coefficient found in ( SnS),.,,TiS2 com- pound can be considered too high and deviates from the general behaviour observed. This suggests that, besides the geometry of the interstitial sites, other factors should affect the lithium ion diffusivity.Although there is some controversy concerning the occurrence of charge-transfer phenomena between MS and TiS, layers, a modification of the formal Table 2 Chemical diffusivity of lithium in titanium disulfide and titanium misfit layer sulfides D/cm2 sC1 compound x =0.05 x =0.10 Li,TiS2" 5 10-9 5 x 10-9 Li, (PbS)l, ,TiS, 1.9 x lo-" 1.9 x lo-" Li, (SnS)l .zoTiS2 1.9 x 10-l' 2.9 x lo-'' Li,( PbS)1.18(TiS2)2 6.5 x lo-'' 4.7 x 10-l0 Lix(BiS)l.l4(TiS2)2 3.3 x 10-l0 3.2 x lo-" "Ref. 14. Ref. 12. 1416 J. MATER. CHEM., 1994, VOL. 4 (a) (c1 0 F mol-' 0.05 f rnol-' 0.25 F mol-' 0.5 F rnor' 1.0 F rnor' 0 F mol-' 0.05 f rnol-' 0.25 F mol-' 0.5 F rnol-' 1.0 F rnol-' 400 -200 N ~-400~' \>-.-2 (b) (d) 0, o F m01-l 0.05 F rnol-' 0.25 F m01-l 0.5 F mar' 1.0 F mar' 0 F mol-' 0.05 F rnol-' 0.25 F ml-' 0.5 F mar' 1.0 F rnor' .>*6_Y1r -400 overvoltageN Fig.5 Micropolarization tests recorded at different discharge depths: (a)(PbS),.,gTiS2, (b)(SnS),,,,TiS2, (c) (PbS)1,1g(TiS2)2,((1)(BiS)l,l.,(TiS2)2. oxidation state of M atoms has been deduced from measure- ments of the Hall effect.' The values reported are Pb+2.36 and Snf2.17 for (PbS),.,,TiS, and (SnS),.,,TiS,, respectively." It is worth noting how the (SnS),,,,TiS, compound has the lower value of formal oxidation state. This would originate smaller repulsive forces and a more favourable path for lithium mobility.The comparison of the chemical diffusion coefficients of the misfit layer compounds studied in this paper with the corre- sponding to the Li/TiS2 system has some difficulty due to the remarkable variation of the values obtained by different authors." For this comparison, we have chosen the data reported by Basu and Worrell,12 obtained by applying the same method and which also showed a limited change with the degree of intercalation (Table 2). We can conclude that in the compounds where the lithium ions have a higher chemical diffusion coefficient, the measured value is at least ten times lower than that of the binary dichalcogenide. To have a point knowledge of the reversibility of the system Li/( MS)l +a(TiS,), against the intercalation degree, cyclic micropolarization tests were performed by studying the corn- position range 0 <x <1.Some polarization cycles are shown in Fig. 5. A common behaviour is the observation of a hysteresis loop between the cathodic and anodic micropolariz- ation scan whose magnitude depends on the compound and the degree of intercalation. Fast kinetics and good reversibility are characterized by a limited hysteresis and a quasilinearity of the current-voltage At the beginning of the intercalation process, the hysteresis between the anodic and cathodic scans is high. Apparently, lithium inserted in this composition range is not readily available and may only be accommodated if a structural rearrangement of the cathodic material occurs.Another common feature is a decrease in the width of the hysteresis as the degree of intercalation increases and, there- fore, a better reversibility of the intercalation process. In fact a galvanostatic method has been used and under these con- ditions the lower the scan rate, the larger will be the current passing through the cell and the diffusion of lithium ion into the cathode. This means that the degree of hysteresis can be related to the diffusion of lithium and the micropolarization curves can be used to obtain information on the kinetic properties of the intercalation electrode for those composition ranges in which the current pulse method cannot be prop- erly used. Worth noting is the acceptable agreement between the polarization values, the curve hysteresis included in Fig.5 and the chemical diffusion coefficients shown in Fig. 4. The largest hysteresis is shown by Li,( PbS),.,,TiS, intercalates, which have the lower chemical diffusion coefficients. The three remaining sulfides with greater chemical diffusion coefficients were also characterized by a decrease in both the polarization values and the curve hysteresis. On the other hand, the (MS), +a(TiS2)2 compounds showed lower hysteresis at high intercalation degrees than those of (MS),+6TiS2 stoichiometry. We gratefully acknowledge the support of CICYT (MAT93-1204) and Junta de Andalucia (Grupo 6034). References 1 G. A. Wiegers and A. Meerschaut, Incommensurated Sandwiched Layered Compounds, ed. A.Meerschaut, Trans. Tech. Publications, Zurich, 1992. p. 1. 2 G. A. Wiegers, A. Meetsma, R. J. Haange and J. L. de Boer, Solid State Ionics, 1989,32133, 183. 3 P. Bonneau, J. L. Mansot and J. Rouxel, Mater. Res. Bull., 1993, 28, 757. 4 Y. Ohno, Solid State Commun., 1991,79, 1081. J. MATER. CHEM., 1994, VOL. 4 1417 5 6 7 8 9 10 11 12 A. R. H. F. Ettema, G. A. Wiegers, C. Haas and T. S. Turner, Physica Scr., 1992, T41,265. C. Auriel, A. Meerschaut, P. Deniard and J. Rouxel, C.R. Acad. Sci. Paris Ser. ll,1991,313, 1255. C. Barriga, P. Lavela, J. Morales, J. Pattanayak and J. L. Tirado, Chem. Muter., 1992, 4, 1021. L. Hernan, J. Morales, J. Pattanayak and J. L. Tirado, J. Solid State Chem., 1992,100,262. A. Meerschaut, C. Auriel and J. Rouxel, J. Alloys Compounds, 1992,183,129. C. Barriga, P. Lavela, J. Morales and J. L. Tirado, Solid State lonics, 1993,63-65,450. G. A. Wiegers, A. Meetsma, S. Van Smaalen, J. L. de Boer and R. J. Haange, J. Phys. Condens. Matter., 1991,3, 2603. S. Basu and W. L. Worrell, in Fast Ion Transport in Solids, ed. P. Vashistita, J. N. Hondy and G. K. Shenoy, North Holland, Amsterdam, 1979,p. 149. 13 14 15 16 17 18 19 20 A. S. Nagelberg and W. L. Worrell, J. Solid State Chem., 1979, 29, 345. M. S. Whittingham, J. Electrochem SOC., 1976, 123, 315. P. Lavela, Ph.D. Thesis, Cordoba University, 1993. L. Hernan, J. Morales, L. Sanchez and J. L. Tirado, Chein. Mater., 1993,5, 1167. L. Hernan, J. Morales, L. Sanchez and J. L. Tirado, Sdid State lonics, in the press. K. Kanehori, F. Kirino, T. Kudo and K. Miyauchi, J. Electrochem. Soc., 1991,138, 2216. L. P. Bicelli, S. Maffi, P. Tavighiani and L. Zanotti, Solid State lonics, 1987, 24, 297. M. Patriarca, M. A. Voso, B. Scrosati, F. Bonino and H Lazzari, Solid State Ionics, 1982,6, 15. Paper 4/00385C; Received 21st January, 1994
ISSN:0959-9428
DOI:10.1039/JM9940401413
出版商:RSC
年代:1994
数据来源: RSC
|
13. |
Intercalation of large cluster cations in TaS2 |
|
Journal of Materials Chemistry,
Volume 4,
Issue 9,
1994,
Page 1419-1425
Linda F. Nazar,
Preview
|
PDF (1028KB)
|
|
摘要:
J. MATER. CHEM., 1994,4(9), 1419-1425 Intercalation of Large Cluster Cations in TaS, Linda F. Nazar" and Allan J. Jacobsonb a Department of Chemistry, University of Waterloo, Waterloo, Ontario, Canada N2L 3G 1 Department of Chemistry, University of Houston, Houston, TX 77204, USA Optimum conditions for the formation of stable dispersions of Na,TaS, in N-methylformamide-water mixtures have been determined. Under appropriate conditions, the addition of large guest cations to these dispersions results in instantaneous flocculation and the formation of ordered intercalation compounds. The method has been used to prepare intercalation compounds with cobaltocenium cations, A1 and Ga polyoxycations, and a large iron-sulfur cluster cation. Powder X-ray diffraction data from oriented films of the samples were used to obtain one-dimensional projr, actions of the electron density perpendicular to the layers.These electron-density maps provided information on the orientation of the guest species in the interlamellar space. The intercalation chemistry of layered compounds has been extensively studied over the past two decades, in part due to the properties of such materials as ionic conductors, reversible redox hosts, low-T, superconductors and electrochromic devices.' Intercalation often refers to a redox reaction in which a small inorganic or organometallic guest cation is inserted into a host lattice., The term 'pillaring', on the other hand, has come to imply an ion-exchange process in which large guest ions are incorporated between the layers, thus propping or pillaring the layers apart.After removal of solvent molecules, permanent microporosity often results. Although the pillaring of layered aluminosilicate clays was first reported two decades ago, these reactions have been much less well studied in the case of layered metal sulfides, oxides and hydroxides. Part of the reason for this is that reaction kinetics are generally slow. Rates are limited by the activation energy associated with separating layers far enough apart to allow entry of a large guest species (defined here as one whose dimensions well exceed the layer thickness). In contrast, clays swell in water to give large interlayer separations and react more rapidly. Only a few cases of the insertion of large clusters into layered metal sulfides have been reported, although the inter- calation of organic molecules in layered metal sulfides has been well do~urnented."~ The first example was the intercal- ation of a large iron sulfide cluster into TaS, by ion exchange., The layers of Na,TaS, were first exfoliated, and then reassembled with the guest species between the layers.A modified version of this method was used to incorporate ferrocene and various organic compounds, including c60, in MoS, by 'inclusion'.' The well known pillaring agent, A1,304(OH)24(H20)127+('All,'), has also been inserted in Na,TaS, using aluminum chlorhydrol as the aluminum source, although exchange with hydrolysed AlCl, solutions containing All, was not observed.6 We report here the details of the exfoliation/reassembly method, exemplified by the pillaring of TaS, with Fe6S8 (PEt,)$+, Cp,Co+ , and polyoxycations of aluminum and gallium prepared directly from hydrolysed solutions of the corresponding salts.One-dimensional electron-density map- ping has been used to derive structural information on the orientation of the guest species in these materials. A prelimi-nary account of the intercalation of TaS, with the [Fe6s8] cluster has been reported previ~usly.~ Experimental TaS, Dispersions The 2H phase of TaS, was prepared by heating stoichiometric quantities of tantalum and sulfur in a sealed quartz tube at 900°C for 2 weeks, followed by annealing at 450°C for 2 weeks.7 TaS, was reduced with Na,S,O,, either in water,8 or in a solution of 50: 50 H,O-NMF to form Na,,,,TaS,~(solv),.The latter method was preferred, as it resulted in Na,TaS, that was more readily dispersed in solution. The crystals were washed with distilled water, dried in air and placed in a 1 : 1 mixture of deionized water and distilled NMF. Immediate swelling of the crystals was observed, as evidenced by an approximately threefold increase in volume of the solid. As noted previously for swollen H,NbS2 in H20, the interior of the crystals appeared dark brown, while the exterior retained a silver metallic sheen.g The Na,TaS, could be partially exfoliated by vigorous stirring of the suspension at this stage, but more complete exfoliation was achieved by joint son-ification of the mixture for ca.1-5 min. Dispersions were prepared in the range 0.01-0.5 wt.%. Guest Cations The All, cation was prepared by slow hydrolysis of 0.2 mol dmP3 AlCl, with 0.2mol dmV3 NaOH, after which the solution was allowed to age at room temperature for 7-28 days. As reported by Lahav et a!." this method results in a solution containing more than 80% All,. The composition of the solution was confirmed by 27Al nuclear magnetic reso- nance (NMR) spectroscopy. The other component of the solution visible by NMR was primarily Al(H20)63+. The Ga-polyox ycations were prepared similarly, by hydrolysis of a 0.2 mol dmP3 Ga(NO,), solution. This system was especially susceptible to precipitation of oxyhydroxides when the local concentration of OH-was high, and therefore base was added very slowly (over a period of 2 days).The pH was monitored continuously throughout this reaction, and aliquots were removed at selected intervals, ranging from pH 3.0 to 4.4. Solution NMR spectroscopy was carried out on a Bruker AM500 operating at 130.12 MHz (27Al) or 152.5 MHz ("Ga), as described previously." [Fe6(p3-s)8( PEt,),](BPh,), was prepared following the literature procedure.12 [Cp,Co] PF, (Alfa) was uscd as received. The salts were dissolved in N-methylformanide (NMFbacetone or NMF-H20 to make ca. 0.1 mol dm-, solutions for ion exchange with the Na,TaS, dispersion. NMF was dried over molecular sieves and distilled prior to use.Flocculation and Oriented Film Deposition The colloidal solutions of Na,TaS, were flocculated by the rapid addition of the solutions of the appropriate cation to the stirred dispersions. Flocculation was almost instantaneous J. MATER. CHEM., 1994, VOL. 4 in all cases and yielded an agglomerated, metallic green-black precipitate. In different experiments, the guest cations were added in molar ratios (cation:TaS,) ranging from 0.005 to 0.6, to determine the optimum conditions for reassembly of a completely exchanged, well ordered intercalation compound to take place. The flocculated solid was then centrifuged, washed with a small amount of 50 :50 H,O-NMF, recentri-fuged and most of the supernatant was decanted. The solid phase and remaining supernatant were mixed and the slurry was deposited on a glass slide.Thin films of the intercalation compounds were allowed to dry slowly with gentle heating (50°C) over a period of about 3 h. This method of sample preparation results in highly oriented films with the TaS, layers oriented parallel to the plane of the slide. Therefore only the 001 X-ray reflections are detected. XRD: Calculation of One-dimensional Electron-density Maps The powder X-ray diffraction (XRD) pattern (001 reflections) of the oriented films were recorded using a Siemens D500 diffractometer. Intensity measurements were obtained by scan- ning in the lower-angle region (3" <28 <45") with a slit width of 0.1". The slit width was then changed to 1" and higher angles were scanned with overlap of a portion of the lower- angle region (i.e.35" <28 <110")to calibrate the intensity (I) measurements. The peak areas for each of the 001 reflections were determined by computer integration. I,,, were converted to the structure factors, Fool, by means of the following: where L and p are the Lorentz and polarization corrections, respectively. No other corrections were applied to the intensity values. The structure factors were phased by calculating an approximate value for each Fool from the atomic scattering factors (see below) and assumed position of each atom. In the case of the A1 polyoxycation, for example, the calculation was made on the basis of an All, cation residing between the TaS, layers in one of three possible orientations.The ratio of All, :TaS, =0.027 was determined by chemical analysis of the solid. The z-fractional coordinates of the A1 and 0 atoms were derived from the crystal structure of the All, cation,13 and it was assumed that the cation could sit with the face containing three octahedra pointing either 'up' or 'down'. These calculations showed that the scattering of the tantalum dominated the structure factor terms at these guest species concentrations. The Foolvalues were therefore phased accord- ing to the TaS, contributions. The one-dimensional electron-density maps were generated with the phased structure factors by carrying out the inverse Fourier transform: To minimize problems associated with series termination errors, an exponential damping factor, ePM [where M= B(sin2 8)/i2] was applied in some cases to the phased Fool.The observed and calculated intensities (Ioo,)were compared for the Fe,S,( PEt,)62+ and cp,Co intercalation compounds + by a least-squares method.? The starting atomic positions of Ta, and S; Fe, S, and P, and C; or Co and C, were taken from the crystal structures of TaS,14 and the corresponding cations [Fe6S,(PEt3)62+, Or cp"O+].'2''5 t The intensity data were analysed using a program written by P. J. Wiseman, Inorganic Chemistry Laboratory, Oxford. Results and Discussion Exfoliation of the TaS, Cation intercalation compounds form solvated phases in coordinating solvent systems according to the rea~tion:~ A,MS2 +solv+A, t(solv),MS2" -The extent of solvation is determined by the properties of the solvent, the solvation energy of the ternary element and the ionic strength of the medium.Good solvents are typically those which have a high relative permittivity and dipole moment. With solvents of high relative permittivity and where the layer charge density is low, multilayers of solvent molecules can be incorporated to give 'swelled' systems with very large interlayer separations., The limiting case of swelling is where exfoliation occurs to give single layers. Complete exfoliation is assisted by solvation of the layers themselves: A,MS2 +solv+xA+(solv)+MS,"-(solv) The phenomenon is well known in smectite clays, which spontaneously exfoliate in H,0.I6 Exfoliation is strongly inhibited by the addition of electrolytes, or by any impurities in solution, in accordance with the Derjaguin-Landau-Verwey-Overbeck (DLVO) theory of colloid formation and ~tabi1ity.l~ In general, the extent of swelling is determined by the combination of the interlayer cation and the specific solvent system.For example, H,TaS, appears to exfoliate readily in water to give unstable but not in other polar media, even those with high relative permittivities such as NMF (E= 177). In contrast, for A,TaS, (A=Li, Na), the solvation process with non-aqueous polar solvents such as formamide, NMF, N,N-dimethylformamide (DMF) and dimethyl sulfox- ide is more favourable. Mono- or bi-layer solvates are formed by A,TaS2 in NMF.The application of weak shear forces to these solvates has been reported to give colloidal solution^.^ The degree to which the TaS, exfoliates into its component layers determines the effectiveness of exchange with large guest cations. We have found that if Na,TaS, crystals are placed in NMF and H20 is added (or uice versa), the solid begins to swell rapidly. When these solvent mixtures are sonicated for brief periods of time, the crystals readily disperse. The effect of the H20-NMF ratio on the extent of exfoliation of the Na,TaS2 crystals is shown in Fig. 1. An approximate measure of the degree of dispersion was determined by sonicating 0.020 g of Na,TaS, in 10 cm3 of a solvent mixture for lOmin, and then allowing the resultant sol to stand for 3 h.The crystals that did not substantially exfoliate settled to the bottom of the vial during this time. The sol was decanted, H,&NMF ratio x 100 Fig. 1 Degree of dispersion of Na,.,,(H,O),TaS, in NMF-H,O as a function of the water content J. MATER. CHEM., 1994, VOL. 4 and the remaining solid dried and weighed. The difference between the mass of Na,TaS, added to the vial and that which settles out gives a good relative estimate of the influence of the solvent composition on the degree of dispersion. A decrease in the TaS, particle size may result from sonication but should be independent of the solvent composition. As the plot indicates, the maximum in dispersion stability occurs at a 1: 1 ratio of H,O to NMF.These metallic blue-green sols were extremely stable, and formed readily after as little as 1min of sonication. In transmitted light, the sols were almost transparent, and gold in colour. Transmission electron microscopy (TEM) analysis, carried out on 0.2% dispersions deposited on a holey carbon support, suggests that they consist of exfoliated TaS, layers. Extremely thin sheets were just barely discernable on the support. These sheets were quite flexible, and many were folded over at the corners, or noticeably wrinkled. Some sheets were clearly composed of more than one TaS, layer and showed 'leafing' at the edges, like the pages of an open book. Multiple layers were more visible due to their enhanced contrast with the background.The presence of single layers in these sols was further supported by their flocculation behaviour (videinfra). Sols could also be obtained at other H,O:NMF ratios (<6O%, >20%), but the dispersions that resulted were dis- tinctly different in appearance, being slightly opaque and less intense in colour. These sols settled within minutes to hours of preparation, and are most probably best described as colloidal dispersions. Flocculationof the TaSz Layers Flocculation of the solvated TaS,"- (x=0.33) layers is induced by increasing the ionic strength, or decreasing the relative permittivity. In the former case, on the addition of an electro- lyte to the exfoliated layers, different outcomes are possible which can be summarized as: B" +A, +TaSZx-(dispersion)+A, -,B,YaS2 (solid) One limit of this reaction, termed 'salting out' (y=O), is common in colloidal systems.The added electrolyte induces flocculation by compression of the electric double layer around each TaS, sheet, but does not itself exchange with the cations (A') in the double layer. Compression results in the layers approaching one another at sufficiently close distances that attractive forces overcome the repulsive forces between the negatively charged layers and the structure reassembles. In the other limit, addition of the cation, B"', induces floccu- lation accompanied by ion-exchange for A+ cations (y/n =x) in the double layer, to give the pillared product B,TaS,. The major factors that determine the outcome of the process are the dispersion concentration, the electrolyte (intercalating cation) concentration, and the ratio of the cation concentration to the dispersion concentration. A known intercalation system, [Cp,Co+],TaS,, was used to examine the effect of varying these parameters.The results are summarized in Fig. 2. At very low concentrations of Cp,Co (region D), flocculation+ does not occur irrespective of the concentration of TaSzX-colloid. Conversely, at relatively high concentrations of TaSZx- and Cp,Co+ (region I), flocculation and ion exchange occur to give either a very well ordered, completely ion- exchanged product (at lower [Cp,Co+]/[TaS,"-] ratios), or a poorly ordered product (at higher [C~,CO+]/[T~S,~-] ratios).The XRD pattern (Fig. 3)00f the well ordered product gives an interlayer spacing (11.5 A) identical to that reported for direct intercalation of Cp,Co into TaS2,I8 or by ion exchange with Cp,Co +I-.I9 In region 11, diffraction patterns indicated two phases with distinctly different interlayer spac- 1421 i I I I I I 1 1 1.1- s 1 0.1-I 1 1.0 2.0 3.0 4.0 5.0 6.0 [Cp2Co+]/lO-* mot dm4 Fig. 2 Flocculation diagram for Cp,Co+-TaS,"-. The concentrations are calculated with reference to the total combined solution volume. In region D no flocculation is observed. Regions I and I1 are delineated by the nature of the phase as determined by powder X-ray diffraction: I, single phase, R <0.15; 11, two phases, R >0.15; R= NMF' /Cp,Co+ intercalated).1 hi 25 45 65 85 105 2Hdegrees Fig.3 Powder X-ray diffraction data for an oriented tilm of (N~+),,,,(C~,CO+)~,~,T~S~using Cu-Ka radiation ings. One phase corresponds to [Cp,Co],TaS,, ?nd the other to a phase with an interlayer spacing of 9.6A. The latter probably corresponds to a monolayer of NMF- or H20-solvated Na' residing between the TaS, layers (a monolayer is expected since the data were obtained for thin films dried at 60 "C). It is likely that flocculation of [Cp,Co],TaS, occurs until a minimum critical concentration for ion exchange of Cp,Co+ is reached, at which point 'salting out' of the solvated material commences. The cobaltocenium intercalation compound prepared by flocculation was investigated in more detail.Chemical analysis of the solid gave the composition [Na],.,, [Cp,Co],.,,TaS, .t The XRD pattern of an oriented film gave 14 001 reflections t Chemical analysis for [Na]o.19[Cp2Co]o.13TaS2-H,01,0(Galbraith Laboratories): wt.%: C, 4.4;H, 0.90; N, 0.05; Co, 1.90; Na, 1.54; Ta, 45.04; S, 15.12. 1422 I I Fig. 4 One-dimensional projection of the electron density along the c axis for (Na+),,,,(Cp,Cof),,,,TaS,determined from the data in Fig. 3 out to d= 0.8233 A,indicative of a highly ordered interc$ation compound with an interlayer separation of 11.52A. The projection of the electron density along the c axis, obtained from the X-ray data is shown in Fig. 4. The 1D map shows contribution from the TaS, layers and clearly shows a peak corresponding to the Co atom at the centre of the interlayer gap.Additional features corresponding to the carbon atoms in the cyclopentadienyl rings are also observed. An attempt was made to establish the orientation of the metallocene cation by analysis of the X-ray intensity data. Better agreement was obtained for a model with the C5 axis oriented parallel to the layers. The observed and calculated intensity data assuming the parallel orientation are shown in Table 1. 2H NMR studies, in comparison, have indicated that a dynamic equilibrium between the parallel and perpendicular orientations exists above room temperature.,' The flocculation behaviour established for cobaltocenium with TaS,"- dispersions served as a good model for other large cations.The 'flocculation diagram' is roughly applicable to a wide range of electrolytes, although the size and charge of the flocculant can significantly shift the relative positions of the indicated regions. DLVO theory, for example, indicates that the more highly charged the cation, the more readily it will undergo ion-exchange and induce floc~ulation.'~ Thus, the dispersion region, 'D' is very narrow and almost non- Table 1 XRD data from an oriented film of (Na),,,,(Cp2Co),,,,TaS2" Iobs Llc 001 d0bA (arb. units) (arb. units) 001 11.42 22 700 22 642 002 5.746 5840 5771 003 3.834 457 494 004 2.877 255 328 005 2.303 213 258 006 1.919 260 258 007 1.645 177 157 008 1.440 129 96 009 1.2802 42 27 00,lO 1.1519 20 16 00,ll 1.0485 6 5 00,12 0.9602 7 6 00,13 0.8623 7 5 00,14 0.8229 7 8 "The calculated intensities were determined by fitting the data to a model with the C5 axis of the cobaltocenium cation parallel to the TaS2 layers.J. M.4TER. CHEM., 1994, VOL. 4 existent for (see below). In general, well ordered XRD patterns are obtained at relatively high TaS," -concentrations (0.2-0.4 wt.%), with a 2-3 fold excess of the intercalating cation. Intercalation of Al,, and Gallium Polyoxycations Low Loading The addition of solutions containing the All, polyoxycation to 0.2-0.4 wt.% TaS, dispersions resulted in immediate flocculation of the metal sulfide.When the ratio of Al,,:TaS, was very low (of the order of 0.005), the XRD patterns obtained for oriented films of the flocculated material showed no evidence for intercalation of the Al13 catio?. In these cases, the observed interlayer spacing of d =11.92A indicated that reassembly of the original layered material, Na,, 33( H20),TaS2, occurred ('salting out'). At slightly higher All, :TaS, ratios of ca. 0.01, a well ordered diffraction pattern comprised of more than 11 001 reflections was occz$onally obtained, in which the interlayer spacing was 14.4A. This result, however, was difficult to reproduce consistently. The intercalated material was also quite unstable at these loadings: gentle heating/ drying at 35 "C caused collapse of the diffraction pattern to a disordered set of broad reflections.Chemical analysis of this compound gave a composition Nao~07A10~o,TaS,~,, , indicating incomplete exchange of Na' . Proton incorporation probably accounts for !he residual cationic charge balance. The 14.4A d-spacing may represent the average inter- layer spacing of a randomly 1:1 interstFatified intercalation phase, where the All, cation (doo,=16 A) is present in one gallery, and Na(H,O); (do,, =12 A) is present in another.,' Alternatively, the observed d-spacing could result from All, cations and Na(H,O),f co-intercalated in every gallery, in an approximately 1: 1 ratio. Intermediate interlayer separations are observed in cases where two cations of dissimilar size are intercalated in one gallery.22 The exact value of the interlayer separation in such cases depends on the layer rigidity and the relative amounts of the two interlayer cations.23 High Loading As the amount of All, added to flocculate the TaS, layers increases, the lines in the diffraction pattern at first broaden, indicating a high degree of disorder, and then sharpen at a ratio of ca.0.04 (AlI3: TaS,). Just below this ratio, two or more phases are apparent in the diffraction pattern. For example, at a ratio of,0.03, the major phase has a interlayer separation of ca. 20A, but additional diffraction peaks are also visible with interlayer separations of 5.5 and 10.5 A. For All, :TaS2 ratios between 0.04 and 0.06, a single-phase product is obtained with a sharp diffraction pattern [Fig.S(u)].Fifteen very sharp 001 reflections are presept, corresponding to an interlayer separation of 20.3 A (20 A phase). The number of 001 reflFctions (13-17) and the interlayer separation (20.0-20.4 A) depends on the exact flocculation and drying conditions. Oriented films that were still slightly wet showed even higher interlayer !pacings, but these values rapidly decreased to about 20.3 A as the films dried. Prolonged aging of the films under ambient conditions showed a slower, but gradual decrease of the interlayer spacing. The peak widths initially broadened during this decrease, and then sharpened again. The change in layer spacing is essentially complete after 2 h, and the final XRD pattern [Fig.5(b)], shows no further change at ambient conditions after 2 days. A well orcered phase is evident, witk a interlayer separation of 16.2 A (14 001 reflections, '16 A phase'). The 20 A phase could be regenerated from the 16 A phase by the addition of a few drops of 50: 50 NMF-H20 followed J. MATER. CHEM., 1994, VOL. 4 1423 x c..-u,L Q,c .-c 1 II t-A -I 20 40 60 80 2Bldegrees Fig. 5 Powder X-ray diffraction data for oriented films of All, cation intercalated oTaS2 phases: (a) 20.3 A phase obtained in NMF-H,O; (b) the 16.2 A phase obtained on drying by equilibration for 30min. The 20A phase again slowly reconverted into the 16A phase at ambient conditions. I? contrast, when a few drops ofoH,O were added to the 20A phase, conversion to the 16A phase was immediate.This process could be repeated many times. The reversible changes in the interlayer separation suggest that some, or all of the H20 molecules hydrating the All, cation are being exchanged with NMF. The diffFrence in van der Waals diameters of NMF and H20 (3.8 A) agrees well with the observed change in interlayer distance between the two phases. In addition, 13C and ‘H NMR studies have shown that amides such as DMF and NMF readily coordinate to A13+ .24 The amount of Al13 incorporated into the TaS,, and the amount of residual Na+ cations were chemically analysed as a function of different Al13:TaS2 ratios, and the different sonication times used for preparation of the TaS, dis-persions.Samples prepared from a 0.2% dispersion and sonicated for either 15 or 45min were flocculated with All, at a molar ratio of 0.06 All, :TaS,. Analysis gave compositions “a10.005 [A113 10.023Tas1.96t; ild “alo.003 [A11310.022TaS1.86,respectively, showing that a 15min sonication period is sufficient to obtain a dispersion which effectively exchanges the sodium cation for All,. The results were confirmed by a separate experiment using a different dispersion and a 15 min sonication period, that gave essentially the same composition of [Na],.,,, [A113],,o,7TaSl~88. The analytical data also sug- t Chemical analysis for [Na],,,,, [A1,3],~,2,TaS,.,,: wt.%: Al, 2.38; Na, 0.031; S, 17.50; Ta, 52.96. gests that some hydrolysis (<7%) of the TaS, layers may occur during the dispersion/reassembly process.1D Electron-density Projections The X-ray data for two samples of All, cation-intercalated TaS,; with interlayer separations corresponding to 16.2 and 20.3 A, were analysed to obtain the one-dimensional projec- tions of the electron density along the c axis by the method describedo above. The projection of the electron density for the 16.2 A phase is shown in Fig. 6 together with a drawing of the structure of the A11304(0H),,(H20)l,7+ cation with coordinates taken from ref. 13. Two unit cells are plotted. In addition to the scattering from the TaS, layers, four prominent peaks are apparent between the layers. Comparison with the structure of the All, cation [Fig.6(b)] suggests that these peaks correspond to the four layers of oxygen atoms in the cation oriented as shown (or in the inverse configuration). A similar arrangement has been previously proposed to occur in All, pillared clays,25 but is different from that observed in MOO, where the C, axis of the cluster is perpendicular to the layers.26 The difference may arise from the puckered topology of tbe MOO, sheets. The distance between the maxima is 2.5 A, consistent with the presence of four close-packed oxygen layers. No attempt was made to refine the observed intensity data to a model because of the complexity of the interlayer molecular arrangement. A similar projection of the electron density was generated for the 20.3 A phase, which contains both water and NMF molecules of solvation.The peaks in the electron density between the layers were less well defined, presumably reflecting contributions from the different possible orientations and locations of the NMF molecules. The results reported here differ from those reported pre- viously6 in two respects. An apparently higher degree of order is observed for the intercalation compound prepared using /oo[ t> -1 .o< electron density -Fig. 6 (a) One-dimensipnal projection of the electron density along the c axis for the 16.2 A phase determined from the data in Fig. 5(b); (b)the structure of the All, cation J. MATER. CHEM., 1994, VOL. 4 -I1 I I Ithe dispersion/flocculation preparation technique described 2730 ' ------I here compared with the direct exchange of Na,.,( H20),TaS2 with aluminum chlorhydrol solutions.As pointed out in ref. 6 the chlorhydrol solution may contain lower aluminum oligo-mers in addition to the Al13 cation. Co-intercalation of more than one species may result in increased disorder. Increased disorder may also result from the method used to prepare the sodium intercalation compound. Indirect reduction of TaS2 via partial hydrolysis of the layers with NaOH is known to give more disordered intercalation compounds than when sodium dithionite is used as a reducing agent.27The disper-sion/flocculation method also appears to permit exchange reactions to occur under conditions where direct ion exchange is inhibited, for example, at high sodium ion concentrations in the exchange Intercalation of Ga Polyoxycations Addition of Ga(NO,), solutions, hydrolysed with NaOH to a pH of 3.2, 3.4, 3.6 and 3.8 all resulted in immediate flocculation of the TaS, dispersion.Flocculation was more complete as the solution pH increased. XRD patterns of dry, oriented films of these materials showed that a reasonably well ordFred product was obtained with an interlayer spacing of 20.5 A (8 001 reflections). The similarity :f these results to those obtained for the AIl3 cation (d = 20.3 A) under identical conditions suggests that an NMF-solvated Ga polyoxytation of size and structure similar to that of All, (d=20.3 A, see above) is intercalated between the TaS, layers.Evidence for the existence of large polyoxogallium clusters has previously been reported.28 In pillaring montmorillonite with gallium solutions hydrolysed at an OH : Ga = 2.5 (at pH 3-4), Kydd and cooworkersobserved a phase with an interlayer expansion of 9.3 A. Flocculated products obtained in the pH range 4.0-4.2 exhibited highly disordered diffraction patterns. At pH 4.6, the diffraction pattern was again reasonably well ordertd, and showed a decrease in the interlayer spacing to 17.1 A. The origin of the lower interlayer separation and the nature of this phase is not yet well understood, but may correspond to intercalation of an unidentified gallium cluster species reported to form in solution at higher hydrolysis ratios.28 InterCU1UtiOn of the Fe,(P3-s)8(PEt3)62+Cluster Flocculation of the TaSZx-dispersion with a 50: 1 molar excess of Fe&3-S)g(PEt3)62+ (as the BF4-salt) resulted in incorporatiFn of the cluster to give an interlayer spacing cf d = 17.491A.This corresponds to a lattice expansion of 11.5 A, consistent with the size of this cluster estimated by molecular modelling. The composition of the compound determined by chemical analysis, [Fe6S8(PEt3)3],-,048TaS2,t indicated that loss of three phosphine groups from the cluster occurred on intercalation, possibly due to steric crowding of the molecule in the interlamellar gap. The stoichiometry agrees with that predicted using space-filling models (0.05). A higher molar excess of the iron salt resulted in decomposition of the product; significantly lower molar ratios of the iron salt gave two-phase mixtures, or disordered diffraction patterns.The XRD pattern (Fig. 7) of an oriented film of the compound prepared using a 50 : 1 molar ratio displayed an unusually high degrFe of order along the c axis (19 001 reflections, to d=0.9212 A). The Fourier transform of the structure factors derived from the X-ray intensities gave the one-dimensional electron-density map along the E axis shown in Fig. 8. The two electron-density maxima centred about the middle ? Chemical analysis for [Fe,S,( PEt3)3]0.048TaS2.(NMF),.,(H20)1.4,wt.%: C, 8.10 H, 2.41; N, 3.05; Fe, 4.39; P, 1.20; Ta, 49.08; S, 22.40. -2340 -1950 04 0 1560 -v)c.-5 1170 0 -L390 -780 20 1 20ldegrees Fig.7 Powder X-ray diffraction data for an oriented film of L Fe,SsP (C2H513 10.043TaS2 -5-s FeS sqP d= 2.15 AZ 0.5 PFe'S 0.0 electron density -Fig. 8 One-dimensional projection of the electron density along the c axis for [Fe,S~P(C,H,)3]o,04,TaS2determined froni the data in Fig. 7 of the interlamellar space indicate that the cluster is oriented with the C3 axis of the Fe, octahedron perpendicular to the layers. The two other possible orientations of the molecule, i.e. with the C2 or C, axes perpendicular to the layers, would both result in more than two peaks. The distance between the two maxima (which each correspond to a superposition of scattering from thtee iron atoms and three sulfur atoms in the cluster) is 2.16 A.This is in good agreement with the value determine$ from the crystal structure of the tetraphenylborate salt, 2.15 A.12 The calculated 001 X-ray intensities, using the above model to represent the cluster orientation, were in excellent agreement with the observed data (Table 2). To compute the 001 intensities, the cluster occupancy in the gap was fixed at the above composition. Chemical analysis and thermogravimetric (TG) analysis gave ( H20)1.4(NMF),,, and (H,O),.,(NMF),., respectively, indicating that about two solvent molecules per formula weight were incorporated in the solid. These were included in the refinement. Oriented thin films of [Fe,S,(PEt,),],,,,,TaS, heated at 200 "C in N, showed a gradual decrease in interlayer spacing with prolonged thermal treatment, and progressive broaden-ing of the XRD lines.Atter 24 h the interlayer spacing stabilized at a value of 12.4A, but only the first three 001 lines were visible in the XRD pattern. The decrease in interlayer J. MATER. CHEM., 1994, VOL. 4 Table 2 XRD data from an oriented film of [Fe,S8(PEt,),]o,o,8TaS2" lobs Icalc ~~~~ 001 dobs/A (arb. units) (arb. units) 001 117.18 4298 4302 002 8.785 2591 2596 003 5.764 852 856 004 4.384 217 217 005 3.4807 63 80 006 2.914 28 21 007 2.496 65 68 008 2.180 24 21 009 1.9442 56 38 00,lO 1.7446 28 20 00,ll 1.5915 26 30 00,12 1.4558 21 16 00,13 1.3461 10 7 00,14 1.2494 8 9 00,15 1.1659 2 2 "The calculated intensities were determined by fitting the data to a model with the C, axis of the intercalated ion perpendicular to the TaS, layers. spacing is consistent with loss of interlamellar solvent mol- ecules and triethylphosphine ligands from the intercalated cluster cation.An estimate of the interlayer expansion arising from pillarigg of the TaS, layers by the resultant bare Fe6s8 'core' i: 6 A. This would give rise to an interlayer spacing of 12A, consistent with the observed value. Loss of the triethylphosphine ligands from the iron-sulfur core under these conditions was confirmed by thermal analysis of [Fe6s,(PEt,)6][BPh,]2. The TG curve showed a well defined 40% weight loss at 205°C corresponding to loss of the six PEt, groups (expected, 36%).The TG trace to 650°C of Fe6S8-pillared TaS2 in flowing helium showed that no further weight loss occurred after the loss of solvent and phosphine, suggesting that the cluster-pillared compound was thermally stable at this temperature. Conclusions Stable dispersions of TaSzx- layers can readily be prepared by sonication of NaxTaS2 in NMF-H,O solutions. The maxi- mum degree of exfoliation is obtained at a 1:1 ratio of the two solvents. Intercalation of large guest cations can be achieved by flocculation of the TaSz layers accompanied by ion exchange of the sodium ions. Intercalation of cobaltocen-ium cations by this method was used to define the optimum conditions for flocculation and to demonstrate that similar intercalation compounds were formed by comparison to the direct exchange or ion-exchange methods used previously.Two types of large cluster cations, Al,,/Ga polyoxycation cluster cations and an Fe6s8 cluster cation, have been interca- lated by this technique. In both cases, XRD patterns indicative of a high degree of order perpendicular to the layers were obtained. One-dimensional projections of the electron density along the c axis were derived from the X-ray data and used to obtain information on the orientation of the guest species in the interlamellar gap. In the case of the Fe& cluster, the results unambiguously show that the C3 axis of the molecule is oriented perpendicular to the TaSz layers.The results for the two phases obtaine; on intercalation of Al13 are less definitive. For the 16 A phase, the electron-density data strongly suggest that the cluster is oriented in a similar manner to that observed in clays. The electron density map of the 20A phase that contains NMF in addition to water is more complex and cannot easily be interpreted in terms of a simple model of cluster orientation. These two phases can be intercon- verted by treatment with the appropriate solvent. Thermal treatment of the Fe6S8-intercalated compound results in formation of a stable, iron-sulfur pillared layered sulfide. The dispersion/flocculation method requires careful control of the interlayer cation/solvent combination in order to pro- duce a highly exfoliated, stable dispersion.Once formed, such dispersions provide an effective route for the synthesis of novel intercalation compounds where the kinetics of the direct ion exchange are very slow. L.F.N. acknowledges the technical help of Brian Sayer in obtaining the NMR data, and the NSERC for partial support. A.J.J. thanks the Robert A. Welch Foundation for partial support of this work. References 1 A. J. Jacobson and M. S. Whittingham, Intercalation Chemistry, Academic Press, New York, 1982. 2 R. Schollhorn, Angew. Chem., Int. Ed. Engl., 1980,19,983 3 A. Lerf and R. Schollhorn, Inorg. Chem., 1977,16,2950. 4 L. F. Nazar and A. J. Jacobson, J. Chem. Soc., Chem. Commun., 1986, 570. 5 W. M. R.Divigalpitiya, R. F. Frindt and S. R. Morrison, Science, 1989,246,369;P. Joensen, R. F. Frindt and S. R. Morrison, Mater. Res. Bull., 1986, 21,457. 6 A. Lerf, E. Lalik, W. Kolodziejski and J. Klinowski, 1. Phys. Chem., 1992,96,7389. 7 F. R. Gamble, J. H. Osiecki and F. J. DiSalvo, J. Cheni. Phys., 1971,55,3525. 8 R. Schollhorn, E. Sick and A. Lerf, Muter. Res. Bull., 1975, 16, 1005. 9 D. W. Murphy and G.W. Hull, J. Chem. Phys., 1975,62,973. 10 N. Lahav, U. Shani and J. Shabtai, Clays Clay Minera., 1978, 26, 107. 11 G. Fu, L. F. Nazar and A. D. Bain, Chem. Muter., 1991,3, 603. 12 A. Agresti, M. Bacci, F. Cecconi, C. A. Ghilardi and S. Midollini, Znorg. Chem., 1985,24,689. 13 G.Johannson, Acta Chem. Scund., 1962,16,403. 14 F. Jellinek, J. Less Common Metals, 1965,4,9. 15 E. Frasson, G. Bombieri and C. Panattoni, Actu Crysrallogr., 1963, 16, A68. 16 H. Suquet, C. de la Calle and H. Pezerat, Clays Clay Rfineral., 1975,23, 1; T. Pinnavaia, Science, 1983, 220, 365. 17 See, e.g. Colloid Science, ed. H. R. Kruyt, Elsevier, Amslerdam, 1952,vol. 1. 18 M. B. Dines, Science, 1975,188, 1210. 19 R. P. Clement, W. B. Davies, K. A. Ford, M. L. H. Green and A. J. Jacobson, Inorg. Chem., 1978,17,2754. 20 S. J. Heyes, N. J. Clayen, C. M. Dobson, M. L. H. Green and P. J. Wiseman, J. Chem. SOC., Chem. Commun., 1987,1560. 21 D. C. Johnston, J. Less Common Met., 1982,84,327. 22 H. X. Jiang, S. A. Solin, H. Kim and T. J. Pinnavaia, Muttr. Res. Soc. Proc., 1988, 111,225. 23 M. F. Thorpe, Phys. Reu. B, 1989,39,10370. 24 W. G. Movins and N. A. Natwiyoff, Inorg. Chem., 1967, 6, 847; L. F. Nazar, L. C. Klein and D. Napier, Muter. Res. Soc Proc., 1988,121,133. 25 D. Plee, F. Borg, L. Gatineau and J. J. Fripiat, J.Am. Chem. SOC., 1985,107,2362. 26 L. F. Nazar, X. Yin and S. Liblong, J. Am. Chem. Soc.. 1991, 113,5899. 27 W. Biberacher, A. Led, F. Buheitel, T. Biitz and A. Hiibler, Mater. Res. Bull., 1982,17,633. 28 S, M. Bradley, R. A. Kydd and R. Yamdagni, J. Chem. Soc., Dalton Trans., 1990,413. Paper 3/06087J; Received 12th October 1993
ISSN:0959-9428
DOI:10.1039/JM9940401419
出版商:RSC
年代:1994
数据来源: RSC
|
14. |
Selectivity and composition dependence of response of gas-sensitive resistors. Part 1.—Propane–carbon monoxide selectivity of Ba6FexNb10 –xO30(1 ⩽x⩽ 2) |
|
Journal of Materials Chemistry,
Volume 4,
Issue 9,
1994,
Page 1427-1431
G. S. Henshaw,
Preview
|
PDF (1545KB)
|
|
摘要:
J. MATER. CHEM., 1994, 4(9), 1427-1431 Selectivity and Composition Dependence of Response of Gas-sensitive Resistors Part 1.-Propane-Carbon Monoxide Selectivity of Ba,Fe,Nb,,-,O,, (1 <x <2) G. S. Henshaw, L. J. Gellman and D. E. Williams* Chemistry Department, University College London, 20 Gordon Street, London, UK WCIH OAJ A series of tetragonal tungsten bronze structure compounds of formula B~Fe,Nb,o-,O, (1 <x<2) have been made and characterised by X-ray powder diffraction and electron microprobe analysis. The compounds were ostensibly single-phase materials, but since their electrical conductivity did not vary with composition, their structure was interpreted as being Ba,FeNb,O, with intergrowths of, possibly, BaFe,O,. The gas-sensing behaviour to propane and carbon monoxide was investigated.The compounds were n-type semiconductors, exhibiting a resistance decrease in the presence of ppm levels of the gases in air. Pellets were found to show some selectivity to propane over carbon monoxide which increased with temperature in the range 400-485°C. The selectivity was shown to be dependent on the porosity of the pellets. There was no systematic variation of either sensitivity or selectivity with stoichiometry, x. Temperature-programmed mass spectrometric experiments showed that the onset of carbon monoxide oxidation occurred at 310 "Cover these compounds, while propane combustion started at 420 "C. The propane-carbon monoxide selectivity was interpreted as being due to differences in the combustion kinetics of the two gases which resulted in different gas concentration gradients within the pellet.This was confirmed by using a novel electrode geometry, which enabled the resistance to be measured simultaneously at both the centre and the edge of a pellet. Many metal oxides have been shown to produce an electrical resistance change upon exposure to oxidising and reducing gases present at low concentration in air. The type of response (increase or decrease in resistance) depends on the oxide (n-type or p-type) and the gas (oxidising or reducing).lP5 It has been demonstrated that this phenomenon occurs for a large number of metal oxides in response to a wide array of gases.' The gas response has been described by a mechanism in which the bulk resistance in a porous solid is determined, primarily, by the resistance at junctions between the oxide particles, which in turn is controlled by adsorbed oxygen species acting as surface electron traps.5 If a reducing gas reacts with these surface oxygen species on an n-type oxide the resistance decreases due to the removal of electron traps and a decrease in the potential barrier for conduction.In order for the surface oxygen coverage (and hence conductivity) to be correlated with the concentration of the reactive gas, the rates of adsorption and desorption of oxygen need to be slower than the rate of combustion.' A general model has been developed5 which rationalises the observed pattern of response behaviour to different gases by different materials: a systematic change in response pattern with change in the volume density of electron-donor states in the solid is predicted.The reactant gas concentration at a sensor surface is different from the bulk gas composition due to relative rates of gas-phase diffusion and surface reaction. For a porous solid, the gas composition, and hence electrical conductivity, within the interior will be a function of position. Differences in the surface-catalysed reaction rates of gases might, therefore, be exploited to give a selective gas response.' It has been reported that the compound Ba6FeNb,030, an n-type semiconductor, exhibits 'anomalous' behaviour to some gases;' that is, in some gases it shows n-type behaviour whilst in others it exhibits p-type behaviour.According to the general model of re~ponse,~ it should be possible to modify this behaviour, perhaps to the extent of achieving some selective gas response, by altering the volume density of electron-donor states. In principle this could be achieved by alteration of the Fe :Nb ratio. A series of compounds Ba6FexNb10-n030 (where 16x62) have, therefore, been made and their response to propane and carbon monoxide examined. The feasibility of such substitution ?as inferred !ram the similarity of i2nic radii [NbS+, 0.64 A; Fe3+, 0.65 A (high spin); Fe4+, 0.59 A].7 It was expected, from consideration of the charge balance, that substitution of Fe for Nb should result in the oxidation of some of the Fe3+ to Fe4+.For x=2, all of the iron should be in the Fe4+ state. As a consequence the material was expected to change from n-type to p-type with increasing Fe:Nb ratio and furthermore to show a marked increase in conductivity. For compositions with x <1, charge balance woul! require the formation of either Fe2+ [ionic radius, 0.78 A (high spin)] or Nb4+ (ionic radius, 0.68 A),' with the material remaining n-type and the conductivity increasing with decreasing x. The parent compound for the srries is Ba6Nblo030, with Nb in both +4 and +5 valence states. Effects of combustion-induced composition gradients have been explored by studying pellets of different microstructure and by using different electrode geometries. Experimental The barium-iron niobate compounds were made by wet mixing in acetone the correct stoichiometric amounts of BaCO,, Nb,O, and Fe,O, (Fluka, >99% pure) in a ball mill.The acetone was then evaporated and the resultant pinkish powder fired at 1200°C for 12 h. The resulting compounds were all light brown in colour. The powder X-ray diffraction pattern of Ba,FeNbloO,o confirmed a tetragonal tungsten bronze (TTB) structure. Comparison with a computer-simulated pattern8 showed excellent agreement. Compounds of the other stoichiometries also exhibited the same TTB structure. An impurity phase, identified as BaFe, -xNb,O, (O<x<O.5) was evident in small amounts in some of these materials. Rietveld refinement showed lattice parameters to be independent of stoichiometry (Table 1).The compounds were pressed, at a pressure of 1 ton, into 13 mm diameter pellets and sintered at temperatures of between 1050 and 1200"C for 12 h prior to testing. The pellets were characterised by scanning electron microscopy using a Hitachi s-570 SEM. Simultaneous electron-probe microanalysis (EPMA) linescans of Fe and Nb were carried out on a JEOL Superprobe Table 1 Rietveld-refined lattice parameters for three Ba6FexNb,0-x030 compounds with x =Oh, 1.0 and 1.6 this work ref. 9 a/A CIA U/A CIA 6% 4030 12.58 3.99 Ba6FeNb,030 12.59 3.99 12.60 3.99 -hiFe1 6Nb8 4O30 12.59 3.99 -JXA-8600 equipped with twin-wavelength dispersive X-ray detectors. This appeared to show that the compounds were homogeneous and solid-solution substituted over the whole composition range.The pellet test rig consisted of a ceramic tube holder into which the pellets were sandwiched between gold electrodes. Alumina spacers separated the gold electrodes which enabled up to four pellets to be tested simultaneously. The ceramic holder was housed inside a quartz tube which fitted inside a tube furnace. Mass-flow controllers were used to control the gas flow and concentration through the quartz tube. The resistances of the pellets were measured every 60s using a Keithley 175A autoranging multimeter .The furnace, resistance measurements and gas-flow controllers were computer con- trolled to enable sophisticated gas and temperature variations to be performed.The sensitivity and selectivity experiments were carried out after the pellets had been stabilised in dry air at 470°C for 2 h. Propane or carbon monoxide was then introduced into the rig, and the pellet resistance measured at gas concen- trations of 160, 250, 400 and 1000ppmv. Greater than 80% of the total resistance change occurred within 2min of the gas supply change. The sensitivity (Sgas)was defined as Sgas= Ao/oalr=(Ralr-Rgas)/Rgas,where o denotes conductivity, R,,, the resistance in air and Rgas the resistance in the gas. The selectivity was defined as the ratio Spropane/SCO.Experiments which measured the pellet resistance as a function of tempera- ture were carried out with a temperature ramp of 100°C h-' for both heating and cooling stages.Resistance measurements at different places within a pellet were performed within the test rig using the electrode geometry described in Fig. 1. The pellet was gold-coated with a centre disc and a outer ring on one side and complete coverage on the other. This enabled the resistance changes to be detected at both the pellet centre and at the pellet edge. The propane to carbon monoxide selectivity at two distances into the pellet could be calculated. The combustion of propane and carbon monoxide over the barium-iron niobate was studied by mass spectrometry. The powder (1.0g) was loaded onto a glass frit inside an 8 mm id glass tube. It was initially heated to 550 "C in flowing air to remove contaminants and then cooled.Fig. 1 Schematic diagram of the electrode arrangement to measure resistances at the centre and the edge of a pellet. (a) 1, Macor ceramic spacer with hole for gold electrode lead; 2, gold disk electrode; 3, pellet; 4,Macor ceramic with central gold electrode ( 1 mm diameter) and outside gold ring electrode (0.25 mm diameter wire). (b)Head-on view of Macor ceramic (4) showing gold central electrode and gold ring. J. MATER. CHEM., 1994, VOL. 4 1OOOppm propane or carbon monoxide in 20% 02-Ar mix- ture was passed through the tube at a flow rate of 50 cm3 min-' while the tube was heated at 10°C min-'. The gas composition, downstream of the tube, was sampled using a stainless-steel leak valve and analysed by a VG 7070F mass spectrometer. A mass spectrum was acquired every 10 s in the range 10-140 u.Results Although the expected stoichiometric variation within the TTB structure appeared to have been achieved (Fig. 2), there was no significant variation in electrical conductivity, or activation energy for conduction, with composition (Table 2). The activation energy increased slightly with temperature, from around 1 eV in the temperature range 400-500°C, to around 1.5 eV in the temperature range 550-750 "C. Resistivity for different preparations of a given composition could vary by a factor of up to 10. It is shown later that the porosity and grain size could vary markedly between preparations, and it is considered that such variations were the origin of the variability in resistivity.A plot of gas response us. gas concentration is given in Fig. 3 for propane and carbon monoxide. The response law was found to take the form: Szp*, where p denotes gas partial pressure (concentration). Carbon monoxide exhibited a positive interception for many preparations. The propane- carbon monoxide selectivity (Spropane/SCO)plotted as a function of the theoretical iron stoichiometry in the barium-iron niobate system for pellets sintered at 1200°C is shown in Fig. 4.The selectivity ranged in value from 1 to ca. 7, with no clear trend with iron concentration. Six pellets with the same composition, Ba,Fe, .6Nb8.4030, and identical fabrication, all gave different values. Clearly, there was a parameter, other than the stoichiometry, which caused this variation in selectivity.1.6 1 a 0 x- m C c 0 .- 1.4 t@Q, a a i 0 1. 1.0 1.2 o 1.4 1.6 o 1.8 theoretical x Fig. 2 Theoretical Fe stoichiometry (x) in the Ba6FexNb,o~x0,, pel- lets 0s. the experimental value calculated from Fe :Nb ratios obtained from EPMA elemental linescans. The compound Ba,FeNb,O,,, was used as an external standard. Table 2 Resistivities and activation energies for conduction in dry air for Ba6Fe,Nb,o-x0,0 pellets composition activation energy resistivity (400-530°C)/eV at 4SO°C/108 R cm Ba6Fel lNb8.9O3O 1.o 1.68 Ba6Fel 3Nb8 '7030 1.1 2.35 Ba6Fel 5Nb8.5030 1.1 2.76 Ba,Fe, .,Nb8.,O,O 1.o 2.77 Ba6Fel.9Nb8 1°30 1.1 0.865 J. MATER. CHEM., 1994, VOL.4 0 10 20 30 Illgas concentration(ppm)~ Fig. 3 Gas response (Sgas) us. (gas concentration)+ for Ba,Fe,,,Nb,,,O,, at 480°C: 0,propane; 0,CO rn mm .m m rn I 2[;,:$, 0 1.0 1.2 1.4 1.6 1.8 2.0 X Fig. 4 Selectivity of Ba6FexNb10-s030 pellets, sintered at 1200 "C, as a function of the stoichiometry x Scanning electron microscopy images (Fig. 5) of the pellets indicated there was a correlation between the selectivity and the apparent porosity. The porosity, in turn, appeared to be very sensitive to the pellet firing temperature. Pellets fired at temperatures within a 1100-1200 "C window exhibited differ- ent average crystallite sizes. Highly sintered pellets which exhibited large crystallite size (5-10 pm) and low porosity [Fig.5(a)J showed poor gas sensitivity and selectivity. Pellets with small crystallite size (< 1 pm) and high porosity [IFig. 5(b)]exhibited moderate sensitivities but poor selectivit- ies. The pellets which had shown selectivities of around 7, however, had grain diameters and porosities that were between the two extremes [Fig. 5(c)].That the variation in selectivity was so dramatic, even for pellets of the same nominal composi- tion, suggested that there was a critical level of porosity for maximum selectivity. It appeared, therefore, the selectivity of response to propane and carbon monoxide was strongly dependent on the intercon- nected porosity and gas access of the pellets. These properties would be influenced by the degree of sintering and initial particle size.The propane<arbon monoxide selectivity of the:,e com-pounds as a function of temperature is shown in Fig. 6. The selectivity to propane increased, by a factor of three, as the temperature was stepped from 395 "Cto a maximum at 483 "C. At 500 "C the selectivity decreased. This temperature depen- dence of the selectivity suggested that the reaction kinetics of carbon monoxide oxidation and propane corn bustion were different on these compounds. The resistance 11s. tem-perature response (plotted in the form In R us. 1/T) for Ba6Fe1,3Nb8.7030in dry air, carbon monoxide and propane (Fig. 7) corroborated the results from the selectivity /is. tem-perature experiment. The curve in dry air decreased mono- tonically and represented an activation energy for conduction of 1.0eV.The 250ppm propane curve exhibited sigmoidal 8r 6-0 CIY2 a3 4-I v' 400 440 480 TI'C Fig. 6 Selectivity(Spropane/SCO)us. temperature for pellets of composi-tion Ba,FexNb,,-,030 fired at 1200°C: 0,Fe,.,; 0,Fe1,9; y7, Fe,,,; v7 Fe1.5 15pm Fig. 5 Scanning electron microscopy images of Ba6Fe1,6Nb8,4030 pellets with, (a)poor sensitivity to propane and poor selectivity against CO (firing temp. = 1200"C);(b)moderate sensitivity to propane but poor selectivity against CO (firing temp. = 1100"C); and (c) good sensitivity to propane and high selectivity against CO (firing temp. = 1200"C) 1430 2o r 18 -h 16-v -c 14 -121 I I 1.o 1.5 2.0 lo3 WT Fig.7 Resistance us. temperature relationships for Ba&,Nb,,,O3o in dry air, 1000ppm CO and 250 ppm propane. The vertical dashed line corresponds to a temperature of 483 "C. behaviour when the pellet was initially heated from room temperature. This was the first measurement in the entire sequence and the effect was ascribed to the effect of pellet moisture loss, as has been reported for ZnOl' and SnO,," since it was not found when the pellet was cooled. The cooling curve was essentially a straight line with a slight upward curve at temperatures above 485 "C, which represented a decrease in the activation energy. The 1OOOppm carbon monoxide curve was flatter than that for propane except for temperatures above 485 "C whereupon the slope of the curve increased until the resistance matched that for dry air.This suggested complete reaction of carbon monoxide had occurred on the metal oxide surface by this temperature. Fig. 7 implied a large response at low temperature. However, constant-temperature studies showed that the response time, short at elevated temperatures, became long at temperatures below 400'C. This suggested that the apparently large low-temperature response was probably the result of the freezing-out of the high-temperature condition. A common reaction product of both propane and carbon monoxide over BaFe,.,Nb,,,O,, was carbon dioxide. The reaction of each gas over Ba,Fe,,,Nb,,,O,, was monitored by inspection of the mass 44 intensity, as a function of tempera-ture (Fig.8). Note that propane also exhibited a C3H,+ peak of mass 44; however, the major propane peak was mass 43 (C,H,+), the intensity of which decreased at the same tempera-ture as the mass 44 signal increased. The mass 44 peak intensity would be expected to be more sensitive to CO, concentration since three molecules of CO, were produced v) v)CI c t3 7 t 0 0 u -8 UJ0 6 -65 -rd v) 4 -4(I!t I I I I 100 300 500 T/OC Fig.8 Mass 44 (CO;) intensity us. reactor temperature, for the reaction of 1OOOppm CO and l000ppm propane over Ba6Fe1.6Nb8.4030 J. MATER. CHEM., 1994, VOL. 4 from the complete combustion of one molecule of propane. Fig. 8 shows the onset of carbon monoxide oxidation to CO, occurred at a temperature around 310"C, while propane combustion started at ca.420 "C.The effect of gas composition gradients within a pellet was explored using the electrode arrangement shown in Fig. 1. The pellet resistance was meas-ured at both the pellet centre and pellet edge simultaneously during exposure to sequential concentrations of propane and carbon monoxide at 480 "C (Fig. 9). The central disc electrode exhibited a selectivity Spropane/SCO=9.9, compared with a value of Spropane/SCO=4.8for the ring electrode at the pellet edge. As a qualitative indication of the difference in surface-catalysed combustion rate of the two gases, at this temperature the mass spectrometry data indicated ca. 30% of the propane and 80% of the carbon monoxide had reacted upon passage through the short packed bed of the solid.Discussion Compositional measurements indicated that Fe had been successfully substituted for Nb. However, the lack of variation of the conductance with composition implied that the expected oxidation to Fe4+had not occurred. One possible explanation was that the charge balance was established by oxygen vacancies. Another was that the TTB, Ba,FeNb,O,,, remained the major phase with the excess of Fe and Ba in the substituted compositions accommodated partly by the separation (as observed) of a small amount of the impurity perovskite phase BaFe, -xNb,03 (0<x<0.5) and (speculatively) partly by intergrowths of a phase such as BaFe,O,. The intergrowths would be too small to be detected by X-ray diffraction or electron-probe microanalysis.It is interesting that any effect on the gas response of these compositional variations was completely swamped by the effects of varying porosity and surface area (see below). The gas response law was the same as that found on tin dioxide,, a result which has been rationalised in terms of the reaction of the gas with 0,-species on the surface, changing the surface coverage of these species. That the high conduc-tance in the presence of the combustible gas at high tempera-ture can apparently be frozen-in at low temperature (see the discussion of Fig. 7, above) is consistent with the notion that the ionosorption of oxygen is a process which is activated and slow., The positive intercept on the concentration axis for many preparations in response to carbon monoxide (Fig.3) can be interpreted as being due to the effect of combustion on the concentration profile of carbon monoxide through the pellet. Detailed theory will be given elsewhere.I2 The temperature dependence of the selectivity for propane t/104 s Fig.9 Resistance measurements at the centre disk and outside ring electrodes of a pellet of Ba6Fe,,,Nb8 upon exposure to pulses of 160, 250 and 400 ppm of propane and then CO in dry air J. MATER. CHEM., 1994, VOL. 4 over carbon monoxide, and the variation of both the selectivity and sensitivity with degree of sintering, can be explained in terms of a gas composition gradient established inside the porous pellet.At high temperatures (around 480 “C),carbon monoxide would be oxidised at the physical extremities of the pellet and not penetrate into the interior. The resultant change in pellet resistance would be small. Propane, on the other hand, would diffuse into the pellet interior, producing a large change in resistance. That the concentration gradient is steeper for carbon monoxide was simply a consequence of the faster combustion reaction of this gas, as demonstrated by its lower temperature for the onset of combustion (Fig. 8). The existence of this concentration gradient, different for carbon monoxide and propane, was demonstrated directly by the result shown in Fig. 9. The effects of pellet microstructure had the same basic explanation but with some subtleties.An outline of the theory has been given previously.‘ There are three factors. Firstly, the sensitivity depends upon the surface area exposed to the gas: sensitivity should increase with decreasing crystallite size at a given porosity. Secondly, the rate constant k’, for combus- tion of the gas increases with increasing surface area. Thirdly, the gas concentration found at the centre of the pellet decreases as the ratio k’r2/D’increases, where r is the pellet radius, k‘ is the pseudo-first-order rate constant for surface-catalysed com- bustion (dependent on the surface area exposed to gas per unit volume of the porous solid) and D’is the diffusivity of the gas within the pellet, D’=(d/z)D, where D denotes the free-space diffusion coefficient, E the porosity (volume fraction of free space in the pellet), 6 the constrictivity (a measure of the area of the smallest part of the largest channels) and zthe tortuosity (a measure of the pathlength through the pores).This ratio is the ratio of the characteristic time for gas diffusion into the centre of the pellet (r2/D’)to the character- istic time for combustion of the gas on the surface of the pores (l/k’). When it is large the gas does not penetrate into the pellet and the sensitivity is correspondingly low. Increasing the degree of sintering would decrease the surface area, porosity and constrictivity but perhaps increase the tortuosity.This would have the effect of decreasing the sensitivity because of the decrease in surface area. The decrease in D’could result in an increase in the ratio k’/D’. If, as a consequence, a substantial gas concentration gradient were to be established into the centre of the pellet, the sensitivity to the particular gas would be decreased and the selectivity might, as a consequence, increase. With small crystallites but large open porosity, the sensitivity would be high but the ratio k’/D’ could be relatively low for both gases, giving poor selectivity. With small crystallites, better compacted, the basic sensitivity would be high and it would be conceivable that with increasing temperature a range could be found where k’/D’ was small for propane and large for carbon monoxide, leading to good selectivity.These qualitative ideas were borne out by the present study, as Fig. 6 illustrates. One could imagine that the temperature of maximum selectivity would be dependent on the porosity of the pellet due to the competing processes of diffusion and combustion, and the dissimilar combustion kinetics of propane and carbon monoxide. Since the porosity was determined by the pellet firing temperature (Fig. 5) good control of this parameter should yield reproducible porosities and gas selectivities. Conclusions The effect on the response to propane and carbon monoxide of iron substitution in Ba6FexNb10-x030 pellets and the consequent crystallographic and compositional changes was negligible, compared with the effects of variation in the physical properties of the sensor porosity and particle size.The selectivity of Ba,Fe,Nb,,~,O,, pellets to propane over carbon monoxide was shown to be a function of temperature (higher temperatures, up to 480 “C, produced increased selec- tivity) and pellet porosity (too high or too low porosity led to lower selectivities). This variation in selectivity was due to different concentration gradients of propane and carbon mon- oxide inside the porous pellet. These gradients were iffected by both the pellet porosity and the operating temperature. Propane was able to diffuse further into the pellet inter- ior before reaction and therefore produce a greater gas response than carbon monoxide.Maximum gas selectivity in Ba,Fe,Nb,,~,~,, pellets would arise by careful choice of both the porosity (by accurately controlling the sintering temperature) and the operating temperature. This work was funded by SERC, the European Community BRITE/EURAM programme through Capteur Sensors and Analysers Ltd., and the Ford Motor Company. The authors thank Dr. D. H. Dawson and Dr. I. G. Wood for their assistance. References 1 P. T. Moseley, A. M. Stoneham and D. E. Williams, in Techiques and Mechanisms in Gas Sensing, ed. P. T. Moseley, J. 0.W, Norris and D. E. Williams, Adam Hilger, Bristol, 1991. 2 P. T. Moseley and D. E. Williams, Polyhedron, 1989,8, 1615. 3 P. T. Moseley and D. E. Williams, Sensors Actuators R, 1990, 1, 113. 4 D. E. Williams, Anal. Proc., 1991, 28, 366. 5 D. E. Williams and P. T. Moseley, J.Muter. Chem., 1991, 1,809. 6 D. E. Williams, in Solid State Gas Sensors, ed. P. T. Mosc ley and B. C. Tofield, Adam Hilger, Bristol, 1987. 7 R. D. Shannon, Acta Crystallogr., Sect. A 1976,32,751. 8 LAZY-PULVERIX program, R. H. Jones, The Royal Institution, London. 9 ICSD (Inorganic Crystal Structure Database), Daresbur y, entry no. 26095. 10 M. Nakagawa and H. Mitsudo, Sug. Sci., 1986,175,157. 11 J. F. McAleer, P. T. Moseley, J. 0.W. Norris and D. E. I4 illiams, J. Chem. SOC., Faraday Trans. I, 1987,83,1323. 12 G. S. Henshaw and D. F. Williams, in preparation. Paper 4/00885E; Received 14th Februar,li, 1994
ISSN:0959-9428
DOI:10.1039/JM9940401427
出版商:RSC
年代:1994
数据来源: RSC
|
15. |
Influence of chlorine–oxygen substitution on the electrical properties of some oxychloride tellurite glasses |
|
Journal of Materials Chemistry,
Volume 4,
Issue 9,
1994,
Page 1433-1436
José M. Rojo,
Preview
|
PDF (963KB)
|
|
摘要:
J. MATER. CHEM., 1994, 4(9), 1433-1436 Influence of Chlorine-Oxygen Substitution on the Electrical Properties of some Oxychloride Tellurite Glasses Jose M. Rojo,a Pilar Herrero,a Rosa M. Rojas,a Jesus Sanqa Jean M. Reau,*b Sylvie Rossignol,b Bernard Tanguyb and Josik Portierb a lnstituto Ciencia de Materiales (CSIC), c/Serrano 7 75 dpdo 28006, Madrid, Spain Laboratoire de Chimie du Solide du CNRS, 357, cours de la Liberation, 33405 Talence, Cedex, France The ionic transport properties of oxychloride tellurite glasses with (47 -x)LiO,~,-xLiCI-53TeO, (0 <x,< 22) composition have been studied as a function of temperature and composition. The influence of the distribution of Li' ions deduced from S,(7Li) NMR analysis on the electrical properties has been discussed and compared with that reported for the homologous oxyfluoride glasses.A large range of glassy compositions has been found within the ternary LiF-Li20-Te02 system. The ionic conductivity due to Li' ions has been studied as a function of the fluorine- oxygen substitution for the (50-x)LiOo.,-xLiF-50Te O2 (Odx<50) series which has a constant lithium content. A significant decrease in conductivity associated with an increase in activation energy is observed when the F :Te ratio is higher than 0.5. This has been explained on the basis of relative ionic distribution and the microstructure which was deduced from the NMR and transmission electron microscopy (TEM) The decrease in conductivity with increasing x has been explained by a progressive decrease of the F-F associ-ations around the Te4' ions and a simultaneous clustering of the Li' and F-ions.This aggregation produces small ordered domains which hinder the Lif diffusion in the amorphous matrix, in agreement with the cluster by-pass modeL4 LiF and TeO, clusters have been detected by TEM.2,3 The existence of a large glassy domain inside the LiC1-Li,O-TeO, system5y6 has led us to undertake an investi- gation on oxychloride tellurite glasses, similar to that made with the oxyfluoride glasses: investigation of the influence of the chlorine-oxygen substitution and the microstructure on the ionic conduction properties of tellurite glasses of composi- tion close to 50LiOo~,-50TeOz. A comparative study of oxyfluoride and oxychloride glasses could be performed in so far as these both series of materials have analogous lithium contents.Moreover, an analysis by TEM of the oxychloride glasses has been performed to investigate whether LiCl clus- ters, similar to the LiF clusters already observed, are present. Experimenta1 Glasses belonging to the LiCl-Li,O-TeO, system have been prepared in a nitrogen-filled glove box, in which the water content was <5 ppm. The starting compounds were a-TeO, (Ventron), Li2Te03 and LiCl (Merck). Li,Te03 was prepared by reaction of Li2C03 (Merck) and a-TeO, at 430 "C under a nitrogen atmosphere to avoid the oxidation of Te4+ to Te6+. Anhydrous LiCl halide, used for glass preparation, was dried at 500°C in vacuum for some hours and kept in a dry glove box.The mixtures were introduced into a silica crucible and melted at 650-800 "C for about 10 min. The melts were poured on a brass plate and pressed with a second plate. The samples were transparent and their amorphous nature was confirmed by X-ray diffraction. In all cases, the composition of the samples was deduced from chemical analysis of chlorine, lithium and tellurium, provided by the Analysis Central Service of the CNRS. Electrical conductivity measurements were carried out by the complex impedance method using a 1260 Solartron fre- quency response analyser. The samples were pellets of cu. 0.8 cm diameter and ca. 0.2 cm thickness, on which gold electrodes were deposited by vacuum evaporation.The fre- quency range used was 10-,-1O6 Hz and measurements were made between room temperature and 20°C below Tg for several temperature cycles under a nitrogen atmosphere. 7Li NMR spectra were recorded with an SXP 4/100 Bruker spectrometer. The frequency used for lithium was 34.9 MHz, which corresponds to an external magnetic field of 2.11 T. The number of accumulations was 400, for which the signal- to-noise ratio was >20. A 71/2 pulse of 3 ps and a period of 5-15 s between successive accumulations was chosen to optim- ize the signal intensity. Transmission electron micrographs were obtained on a JEOL 2000 FX electron microscope, working at 200 kV and equipped with a k45" goniometer stage. The samples were crushed in an agate mortar, suspended in acetone and trans- Table 1 Analytical compositions of some glasszs of (47-x)Li0,,,-xLiC1-53Te02 formulation sample composition A 47.5LiOo.,-53.5Te02 B 46.4LiOo,,-0.8LiC1-52.8Te~l2 C D 44.2LiO0,,-2.5LiC1-53.3Te(l2 40.4LiOo,,-6.8LiC1-52.STe~l2 E 32.9Li00..- 14.3LiC1-52.8Te02 F 25.2LiOo,,-22.0LiC1-52.8Te0, -'T I; 4 1.5 2.0 2.5 3.0 lo3 WT Fig.1 Variation of log oT us. T' for some oxychloride glisses of different composition (Table 1) 1434 ferred to carbon-coated copper grids. A carbon film was then deposited by vacuum evaporation. The film minimizes damage effects by reducing sample heating. Results Oxychloride Tellurite Glasses The glasses studied were prepared from various mixtures of starting compounds with ca.50% Te02. During glass syn- thesis, TeC1, sediments, which are more prevalent for starting mixtures richer in LiC1, were observed on the colder surfaces of the containers; their presence can be explained by the following reaction: TeO, +4LiC1+TeCI4(? ) +2Li20. There- fore the chemical analysis of glasses is absolutely necessary. Table 1 gives the analytical compositions of the glasses studied with the formula (47 -x)LiOo,,-xLiC1-53Te02 (0<x2< 22). The maximum value of xLiclobtained under our experimental conditions, (xLiCIlmax=22, is clearly lower than the limit reached (ca. 50) when the oxychloride glasses are prepared in Ionic Conductivity Properties Conductivity results are plotted in Fig. 1 as a function of the reciprocal of temperature for the glasses studied.In the temperature range under investigation, the data were fitted Table 2 Electrical characteristics of some oxychloride (47 -x)LiOo,,-xLiC1-53Te0, and oxyfluoride (50-x)LiO,,,-xLiF-SOTeO, glasses X=CI A 0 -6.92 0.80 5.76 B 0.8 -6.65 0.80 6.03 C 2.5 -6.35 0.79 6.21 this D 6.8 -6.10 0.79 6.46 work E 14.3 -5.95 0.80 6.73 F 22.0 -5.75 0.78 6.68 X=F -0 -6.99 0.78 5.44 3 -10 -6.85 0.76 5.33 3 -23 -6.63 0.74 5.29 3 -47 -8.49 0.84 4.70 3 'Values are f0.02. 'Values are +0.01. 'Values are +0.02. 0.7t ~~ 0 0.5 1.o 0 J. MATER. CHEW 1994, VOL. 4 (R>0.98) by a root-mean-square method to an Arrhenius law: oT= ooexp(-E,,/kT). The conductivity parameters are gathered in Table 2.We also reported in Table 2 the conduc- tivity parameters of the oxyfluoride (50-.u)LiO,,,-xLiF-50Teo2 (0<x ;L< 1) glasses obtained previ~usly.~ Fig. 2(a) shows the variation of log 0400 as a function of the X:Te ratio (X=F,Cl) for both series of materials. It is found that: (a) In the composition range (0<A':Te <0.42), an increase of conductivity is observed for both series of materials with increasing X:Te ratio. However, for the same X:Te value, oxychloride glasses exhibit a higher conductivity than oxyfluoride ones. The variation observed for the oxychloride glasses is different from that found when these glasses were prepared in air. In fact, a decrease in conductivity for increas- ing C1:Te ratio has been rep~rted.~,~ (b) The presence of a maximum in the conductivity for the oxyfluoride glasses is not observed in the oxychloride glasses which clearly have a smaller glassy domain.Variations of activation energy (EaT)and pre-exponential factor (log go) as a function of the X :Te ratio for both series of materials are given in Fig. 2(b) and 2(c), respectively. It is observed that (a) EaT decreases with increasing C1:Te for X :Te,< 0.42. The EaTvalues are slightly higher for X =Cl than for X=F. (b)Whereas a decrease of log go is apparent with increasing F:Te, an increase of this factor is observed with increasing C1:Te. Consequently, log crO is higher for the oxychloride glasses and the difference between (log o~)~=cl and (log oO)X=Fbecomes more significant when X:Te increases.'Li NMR Investigation 7Li NMR spectra (I=3/2) show only one line centred at the resonance frequency [Fig. 3(a)]. The line, which is assigned to the 1/2+-1/2 transition, is not appreciably affected by the 3/2+ 1/2 and -1/2+ -3/2 quadrupole transitions. The same conclusions have already been deduced from the 7Li and 6Li NMR investigations of Li,O-TeO, and LiF-TeO; glasses.2 Therefore, dipolar interactions (lithi um-lithium and lithium-halogen) are mostly responsible for the shape and width of the observed spectra. Second moment (S,) values of NMR lines give information about the distribution of the nuclei studied in the glasses."" S2(7Li) values were deduced by the same procedure as in the case of oxyfluoride gla~ses.~-~ The S2(7Li) variation as a ..9 0.5 1.o 0 0.5 1.o I X:Te Fig.2 Variation of (a) ionic conductivity at 400 K, ~400K, (b) activation energy, EgT,(c) pre-exponential factor, log G", as a function of the X :Te ratio (X=F,Cl) for (47-x) LiOo.,-xLiC1-53Te0, (x) and (50-x)LiOo.,-xLiF-50 TeO,(O). The solid and dashed lines are drawn to guide the eye. J. MATER. CHEM., 1994, VOL. 4 crystalline LiF 2f crystalline LiCl t0 0.5 1.o 0 0.5 1.o CI:Te F:Te Fig. 3 (a) 7Li NMR spectra of some oxychloride glasses. (b)Variation of S2(7Li) as a function of the C1:Te ratio for the oxychloride series of glasses. (c) Variation of S2(’Li) as a function of the F:Te ratio for the oxyfluoride series of glasses.function of the C1: Te ratio is given in Fig. 3(b),the S2(7Li) value corresponding to crystalline LiCl is also included. By way of comparison, analogous results previously obtained in oxyfluoride glasses3 are shown in Fig. 3(c). S2(7Li) decreases when the C1:Te ratio increases and a plateau seems to be attained for C1: Te NN 0.25. The S2(7Li) variation for the oxy- chloride glasses is opposite to that observed in the oxyfluoride ones, in which the S2(7Li) increases with F :Te ratio. TEM Analysis A transmission electron microscopy investigation of oxychlor- ide glasses was undertaken to detect the eventual presence of clusters inside the amorphous matrix. The stability of samples under the electron beam decreases with increase in LiCl content.Consequently, the TEM study has been performed on sample D which corresponds to an intermediate composition. Despite the carbon film covering the sample, rudiation damage cannot be completely avoided and some alteration of the sample texture is observed. Fig. 4(u) shows a micrograph recorded immediately after the sample was placed under the beam. Only an amorphous matrix is observed. The image [Fig. 4(b)], corresponding to the sample after 5-10 min under the beam, indicates the onset of crystallization. Such a result distinguishes these materials from oxyfluoride glasses, in which crystalline domains have been observed immediately inside the amorphous matri~.~ HlOOA H70a Fig. 4 TEM micrographs corresponding to sample D registered (a) immediately, (b)after 5-10 min J.MATER. CHEM., 1994, VOL. 4 Discussion The comparative study of electrical properties carried out on (47-x)LiOo,,-xLiC1-53Te02 and (50-x)Li0,,5-xLiF-5OTeO2 glasses with a quasi-constant lithium content has shown differences in behaviour for these series of materials. For low values of the X :Te ratio (X :Te,<0.42), conductivity increases in both cases, but it is more important for the oxychloride glasses. The value X :Te equal to 0.42 is the upper limit of the oxychloride glassy domain. For higher values of the F:Te ratio, a maximum of conductivity is observed for F :Te z 0.50, and above this value a rapid decrease of conduc- tivity is shown. The 7Li NMR investigation of both series of materials also indicates a different behaviour of the second moment against halogen-oxygen substitution. S2(7Li) increases with the X :Te ratio in oxyfluoride glasses.This result has been explained3 by a progressive increase in Li-F association. On the contrary, a decrease of S,(7Li) with X :Te ratio for oxychloride glasses is observed. This indicates that lithium and chlorine are dispersed in the tellurite amorphous matrix. This observation agrees with the fact that chlorine is associated with tellurium as deduced from IR data.7 The TEM micrographs of oxychloride glasses show an amorphous matrix only; no crystalline domains are detected. Therefore, the oxychloride glasses are more homogeneous than the oxyfluoride ones. The EaT values determined for the oxyhalogenide glasses are slightly higher for the oxychloride glasses [Fig.2(b)]. Nevertheless, the (Eu,)x=cl and (EuT)x=Fvalues are relatively close, the small difference could be attributed to the slightly higher TeO, content in the oxychloride series. The larger dispersion of Li' ions in oxychloride glasses associated with the homogeneous character of these samples explains the higher values of the conductivity pre-exponential factor of these glasses as compared with the oxyfluorine ones [Fig.2(c)J. The differences of the uo values are more pro- nounced as the lithium halide content increases [Fig. 2(c)]. As conductivity depends on both E,, and go, for a given LiX content (X =F,Cl), the oxychloride glasses show higher con- ductivity values than the oxyfluoride ones [Fig.2(a)]. Conclusions The substitution of oxygen by chlorine in lithium tellurite glasses led to glasses with (47-x)LiOo.5- xLiC1-53Te02 (0<x 622) compositions. For increasing Cl :Te ratios an increase of conductivity was found. A progressive dispersion of Li+ and C1 ions in the amorphous matrix was deduced from the S2(7Li)NMR analy-sis carried out on samples with increasing LiCl content. A TEM investigation has shown that the oxychloride tellurite glasses are homogeneous and no segregation of crystalline domains was detected. The oxychloride glasses differ from the homologous oxy- fluoride glasses, in which association of Li' and F- ions in crystalline domains hinders diffusion of Li+ ions in the amorphous matrix.The authors thank the French-Spanish Acciones Integradas programme for grant 96 B. They also thank R. Ropero for electron microscopy technical assistance. References 1 T. Yoko, K. Kamiya, K. Tanaka, H. Yamada and S. Sakka, Nippon Seramikkusu Kyokai Gakujutsu Ronbunshi, 1989,97, 289. 2 J. M. Rojo, J. Sanz, J. M. RCau and B. Tanguy, J. Non-Cryst. Solids, 1990,116, 167. 3 J. M. Rojo, P. Herrero, J. Sanz, B. Tangu!, J. Portier and J. M. Reau, J. Non-Cryst. Solids, 1992,146, 50. 4 M. D. Ingram, M. A. Mackenzie, W. Muller and M. Torge, Solid State Ionics, 1988,28-30,677. 5 K. Tanaka, T. Yoko, H. Yamada and K. Kamlya, J. Non-Cryst. Solids, 1988,103,250. 6 K. Tanaka, T. Yoko, K. Kamiya, H. Yamada and S. Sakka, J. Non-Cryst. Solids,1991,135, 211. 7 T. Yoko, K. Kamiya, H. Yamada and K. Tanaka, J. Am. Cerum. SOC.,1988,71, C 70. 8 P. J. Bray, S. J. Gravina, D. H. Hintenlang and R. V. Mulkern, Magn. Reson. Rev.,1988, 13,263. 9 P. J. Bray, J. Non-Cryst. Solids, 1987, 95-96,45. 10 J. F. Stebbins, J. Non-Cryst. Solids, 1988, 106, 359. 11 D. R. Macfarlane, J. 0.Browne, T. J. Baston and G. W. West, J. Non-Cryst. Solids, 1989,108, 289. Paper 4/017288; Received 23rd March, 1994
ISSN:0959-9428
DOI:10.1039/JM9940401433
出版商:RSC
年代:1994
数据来源: RSC
|
16. |
Electrode kinetic behaviour of (U0.4Pr0.6)O2±x/YSZ/(U0.4Pr0.6)O2±xcells |
|
Journal of Materials Chemistry,
Volume 4,
Issue 9,
1994,
Page 1437-1440
S. P. S. Badwal,
Preview
|
PDF (552KB)
|
|
摘要:
J. MATER. CHEM., 1994, 4(9), 1437-1440 Electrode Kinetic Behaviour of (UOm4Prom6)O2 fx_+x/YSZ/(Uom4Pr,m,)0, Cells S. P. S. Badwal CSIRO, Division of Materials Science and Technology, Private Bag 33, Rosebank MDC, Clayton 3169, Victoria, Australia Results of kinetic studies on (Uo~4Pro,6)0,+x/stabilised cells as a function of oxygen partial zirconia/(Uo,4Pro,6)02kx pressure (1 -1 OW4atm), temperature (500-900 "C) and electrode heat treatment have been described. Impedance spectroscopy was used for kinetic studies. At all temperatures and in various oxygen partial pressures a single arc, skewed on the high-frequency end of the impedance spectrum, was observed. The data were deconvoluted into two electrode processes with relaxation times differing by about an order of magnitude.The electrode process relaxing in the lower frequency range had an activation energy of 165f10 kJ mol -', a small angle of depression for the impedance arc, [p(O,)] dependence for the time constant and has been attributed to an adsorption/dissociation of oxygen at the gas/electrode interface. The activation energy associated with the high-frequency arc was lower by ca. 35 kJ rnol-' than that for the low-frequency arc and this arc appears to be associated with the charge-transfer reaction. Stabilised urania materials fulfil several of the criteria required of electrodes for solid oxide fuel cells and oxygen sensors. Pure uranium oxide (UO,,,) has a fluorite-type structure but the material undergoes phase transformations when heated to high temperatures in oxidising atmospheres.However, the fluorite-type structure can be stabilised by the addition of a number of other metal oxides such as Y20,, Sc203, Dy203, La203 and Pro, over wide ranges of temperatures and oxygen partial pressures and stabiliser-metal/U Such mate- rials have high electronic conductivity and show excellent catalytic activity to the oxygen-transfer These materials mixed with a small amount of platinum have been used successfully in the low-temperature (300-600 "C) oper-ation of potentiometric (zirconia-based) oxygen sensors.6 In this paper results of kinetic studies on (U,.,Pr,,,)O,~,/YSZ/( U,.4Pro~6)02kxcells as a function of temperature, oxygen partial pressure and electrode heat treat- ment temperature have been described.Experimenta1 An electrode powder of (Uo.4Pro.6)02+x composition was prepared by the coprecipitation method (by addition of metal nitrate solution to ammonium hydroxide). The powder was dried and calcined at four different temperatures (600, 700, 800 and 900 "C). The electrode powders were characterised by X-ray diffraction (XRD), surface area determination (BET method) and scanning electron microscopy. The electrolyte used in these studies was 10mol% Y203-Zr02 (YSZ). Discs of this composition were prepared by mixing oxide powders, pressing and sintering at 1900°C in a gas-fired furnace. The electrolyte discs were ca. 2.2 mm thick, with diameter ca. 9.0 mm and density >90% of the theoretical.The electrode paste (material calcined at 600 "C) in ethanol was applied to the ground surfaces of the electrolyte discs and sandwiched between porous electrode discs of the same com- position. The connections to the cell were made with gold annular rings pressed onto porous electrode discs. The electrode kinetic behaviour was studied with impedance spectroscopy using a Solartron 1174 frequency response ana- lyser over the frequency range 0.5 mHz-l MHz in oxygen partial pressures of between 1 and atm. Premixed, pre- analysed O,-N, mixtures were used to define the gas atmos- phere around the cell. Normally cells were heated to 600 or 800°C in air and data recorded during the cooling cycle at several temperatures down to 500 "C. In more exhaustive tests three types of experiments were performed on the (Uo.4Pro.6)02+JYSZ/( Uo.4Pro.6)02+xcell. In the first set, the cell was heat treated at 600 "C for ca. 60 h and impedance measurements were performed between 600 and 500°C in 100% oxygen and then at 600°C as a function of oxygen partial pressure ( 1-10-4 atm).In the second type of measure- ments, the same cell was heated to 900 "C in pure oxygen and left to equilibrate for 40 h. In a given gas composition, impedance data were recorded as a function of temperature between 900 and 500°C. The gas atmosphere was changed at 900°C after taking the cell back to this temperature and checking for reproducibility of the data. These measurements were made in five different gas atmospheres (loo%, 21%, 1.19%, 0.1% and 0.012% 0,).In the third type of measure- ments, the temperature of the cell was kept constant at 800, 700 or 600°C and the effect of oxygen partial pressure (five different oxygen concentrations) on the electrode kinetic behaviour was studied.Several hours were allowed before commencing data recording after a change in oxygen partial pressure around the cell. Impedance data were analysed with non-linear least-squares programs (modified for ease of data display and manipulation) originally developed by Macdonald.' Results and Discussion After the 600 "C calcination, XRD of the coprecipitated elec- trode powder showed the presence of a single fluorite-type phase. All powders heated at higher temperatures had the same structure.The powders calcined at 600 and 900 ' C had a surface area of 39 & 5 and 9.3 & 1 m2 g-' and a grain size of ca. 0.02-0.03 and 0.1-0.3 pm, respectively. The results of surface area measurements are given in Table 1. At all temperatures and in various oxygen partial pressures mainly a single arc, skewed at the high-frequency end of the impedance spectrum, was observed (Fig. 1). The data were deconvoluted into two electrode processes with relaxation times differing by about an order of magnitude. Several different plausible circuit models were tried. However, the majority of the data showed much better fit to two distributed relaxation in series (Fig. 1 and 2). The angle of depression for the low-frequency arc was, in general, small (between 5 and 15 "), whereas for the high-frequency arc it was usually between 25 and 40 O (Fig.1).After the 900 "C heat treatment, the low- J. MATER. CHEM., 1994, VOL. 4 Table 1 Results of surface area measurements on (Uo,4Pro,6)02kx as treatment, are a decrease in the electrode surface area and the a function of the calcination temperature" likely improved contact between the electrode/electrolyte interface. At 900 "C,no noticeable interdiffusion of cations or calcination conditions surface area reaction between the electrode and electrolyte phases was T/'C t/h /m2 g-l detected. Fig. 3 shows Arrhenius plots of the electrode resistance for 600 24 39+5 the high- and low-frequency arcs in air and Table 2 gives the 700 24 17+2 activation energies for both high- and low-frequency arcs in 800 24 9.5* 1 various oxygen partial pressures.The electrode process900 24 9.3?C 1 relaxing in the lower frequency range had an average acti- ~ 'Calcination atmosphere: 21YO02. vation energy of 165+ 10 kJ mol-'. The activation energy associated with the high-frequency arc was (on average) lower by ca. 30-35 kJ mol-' than that for the low-frequency arc. The electrode resistance at a constant temperature showed 1000\I1 cu$ 0 4. 1000 2000 3000 4000 5000 9 60 i 0 50 110 170 230 290 350 Z/R cm2 Fig. 1 Impedance plane plots for a (U,,,Pr,,,)O,~,/YSZ/ ( Uo,4Pro.6)02~xcell showing a large angle of depression for the high- frequency arc and a high angle of depression for the low-frequency arc.Data recorded after the 900 'C heat treatment. @, Experimental data; arcs drawn as a solid line are fitted curves. (a)501 "C; 1.2% 0, in N2; (b)601 "C; 21% 0, in N2. 80 120 160 200 240 280 ZJR cm2 Fig. 2 Impedance plots for a ( Uo.,PrO.,)O2+,/YSZ/ (Uo.4Pro.6)02k~ cell after 600 "C (a)and 900 OC (b)heat treatments. @, Experimental data; arcs drawn as a solid line are fitted curves. Impedances recorded at 601 'C in 100% 02. frequency arc was much larger than the high-frequency arc at all measurement temperatures and oxygen partial pressures. Heat treatment of the electrode had a significant effect on the shape of the electrode arc and the contribution of the high- and low-frequency arcs.The contribution of each arc to the total electrode resistance was similar after the 600 "Cheat treatment (Fig. 2). After the 900 "C heat treatment, a dramatic reduction in the size of the high-frequency arc (by a factor of ca. 5 at 600 "C in pure oxygen) was observed, but only a small change in the capacitance. The low-frequency arc showed a slight increase (by a factor of ca. lS), but a large decrease in the capacitance (by a factor of ca. 6). The only possible changes occurring to the electrode, as a result of 900T heat only a small dependence on the oxygen partial pressure. However, the time constant of electrode processes associated with the high- and low-frequency arcs exhibited rather large dependence on the oxygen partial pressure (Fig.4). The pressure exponent for the low-frequency arc determined from 3.51 //"/ ../ I 1O~WT Fig. 3 Typical Arrhenius plots for the (0) low-and (A) high-frequency arcs after 900 "C heat treatment of (Uo.4Pro.,)02+,/YSZ/ (Uo.4Pro.,)02+x cell. The data were recorded in air. Table 2 Activation energy for electrode processes relaxing in the low- and high-frequency range oxygen E,(lfIS)/kJ mol -' concentration ("/o) low-frequency arc after 600 "C heat treatment (500-600 "C) 100 165 after 900 "C heat treatment (500-800 "C) 100 152 21 169 1.19 167 0.104 172 0.012 167 -1.5 I -2.01. \ 8 -2.5 0, 52 -3.0 -3.5 -4.0 1 high-frequency arc 154 113 128 140 139 144 -7 Fig. 4 Relaxation times for the (A)high-and (a)low-frequency arcs us.oxygen partial pressure at 700 "C J. MATER. CHEM., 1994, VOL. 4 these plots was cu. 0.5 for the low-frequency arc and cu. 0.4 for the high-frequency arc. The capacitive component of the low-frequency arc showed rather large dependence on oxygen chemical potential and (to a lesser degree) on the temperature. The capacitance was higher at low oxygen partial pressures and at higher tempera- tures. In (U,M)02,, fluorite-type solid solutions (M =Y, Pr, Gd, Dy, La etc.), the 0/(U +M) ratio in oxygen or air for a U/M ratio of 0.4/0.6 is slightly larger than 2.0 but it decreases towards stoichiometric ratio with decreasing oxygen partial pressure, although only slightly.8 Also, at a given oxygen partial pressure the 0/(U +M) ratio decreases slightly with increasing temperature.' Thus it appears that the electrode process becomes slower (significant increase in the capaci- tance) as the electrode composition moves towards a stoichio- metric 0/(U +M) ratio.At low oxygen partial pressures and at higher temperatures another arc, in addition to the main electrode arcs discussed above, was observed on the low-frequency end of the impedance spectrum. The size of this arc increased rapidly with decreasing oxygen concentration. The data for this arc were limited but it seemed to obey a [p(02)]-' law and therefore must be associated with molecular diffusion of oxygen. Discussion At the (Uo.4Pro~6)02~,/electrolyteinterface, the molecular diffusion of oxygen became rate limiting but only in low oxygen partial pressures and was observed at higher tempera- tures in the form of an additional arc on the low-frequency end of the impedance spectra. For most of the data, the overall electrode arc (shown in impedance diagrams) consisted of a contribution from two processes differing in time constant by about an order of magnitude.These processes had different activation energies (165& 10 kJ mol-' for the slower and 130f15 kJ mol-' for the faster process) and a different dependence on the oxygen partial pressure. With a decrease in the surface area of the electrode and the resulting grain growth due to the high temperature heat treatment, the low- frequency arc or the slower process became more dominant whereas the rate of the electrode process associated with the high-frequency arc increased.In order to explain the observed behaviour, the most likely or basic reaction steps for oxygen-exchange reaction at the non-stoichiometric oxide (Uo.4Pro,6)02,,/electrolyte interface to be considered are as follows. Step 1: dissociation/adsorption of oxygen at the (Uo~,Pro,,)02kx/gasinterface, leading to incorporation of oxygen in the lattice. The fluorite solid solutions of the type used in this study are known to undergo rapid oxidation and reduction with varying oxygen partial pressure and tempera- ture while retaining the fluorite structure over wide oxygen partial pressure and temperature range^.'-^^^^'^ Thus it is justifiable to assume that the oxidation kinetics of these electrades play a major role in the reaction mechanism for the oxygen-transfer reaction. Possible reaction steps for oxygen adsorption/dissociation reaction at the surface of (U0,4Pr0.6)02fxelectrode materials are: 02(g)=20,d (electrode surface) Oad=02-(lattice site) +2h or alternatively Oad+U4+ =O2-(lattice site) +U6+ Step 2: diffusion of oxygen ions within the (Uo.4Pro,,)02+, grains to the electrode/electrolyte interface.Step 3: transfer of oxygen ions across electrode/electrolyte interface. Alternative mechanisms such as the oxygen dissociation at the electrolyte surface appears less likely as the burk of the evidence points to a strong role for the electrode material. The Slow Process (the Larger, Low-frequency Arc) Two processes can possibly contribute to the slower reaction step or the low-frequency arc.These are either adsorption/ dissociation reaction at the gas/electrode interface or a diffusion-limited process. If the slow process is associated with step 1 then oxygen exchange kinetics at the (Uo.4Pro.,)02~,/gasinterface will play a dominant role and the origin of electrode capacitance for this process may be chemical (e.g. oxidation/reduction of cations at the gas/elec- trode interface) and may be the reason for the large dependence of the time constant rather than the electrode resistance on the oxygen partial pressure and temperature. No data are available on the oxygen surface exchange kinetics for (Uo,,Pro.,)02kx.However, for other trivalent metal oxide (M203)doped uranium oxides and for a U/M ratio of 0.4/0.6, the 0/(U +M) ratio is close to stoichiometry (2.0) and little change in stoichiometry occurs over a wide oxygen partial pressure and temperature ranges.8.'0,'1 Similar behaviour is expected for the (Uo.4Pro,6)02kxelectrode mate- rials used in this study. If oxygen exchange at the gas/electrode interface were responsible for this arc then the small oxygen partial pressure dependence of the electrode resistance associ- ated with the low-frequency electrode arc would be consistent with the small variation in the O/(U+M) ratio over the oxygen partial pressure range. Also, the activation energy is too large for a diffusion-controlled rate process through the defect structure of the electrode.For oxygen-ion migration in a similar system (urania-yttria), an activation energy of 110-115 kJ mol-' has been reported from oxygen self-diffusion measurements.12 Also the shape of the impedance arc [symmetric with a small angle of depression (5-15")] is inconsistent with a Warburg-type diffusion-limited rate pro- cess.13 Furthermore, the results of heat treatment at 900 "C, that is a decrease in the electrode surface area leading to an increase in the size of the low-frequency arc, support an oxygen-exchange reaction dependent on the surface area of the electrode. Based on most of the evidence, the low-frequency arc appears to be associated with the oxygen dissociation/ adsorption reaction at the (Uo,4Pro~,)0,tx/gas interface and it dominates the reaction kinetics.The proposed mechanism is not unique and has been reported for a number of perovskite electrodes used in solid oxide fuel cell^.'^^^^ For example, for LaCo0,-based perovskite electrodes (doped at the A site or at both A and B sites), which are known to have fast axygen diffusion rates through the lattice, an oxygen-exchange reac- tion similar to that proposed for the electrode materials of this study has been suggested to be the rate-limiting step.', Faster Process (the Smaller, High-frequency Arc) If the high-frequency arc with the large angle of depression were associated with the diffusion of oxygen ions through the electrode grains (a process in series with step 1) then it would be inconceivable that the electrode resistance would decrease as a result of grain growth (longer diffusion paths in larger grains) due to the 900 "C heat treatment compared with the 600 "C heat treatment.Again the activation energy tor the process relaxing at the high frequency is too high coripared with that previously reported for diffusion of oxygen ons in urania-yttria fluorite-type solid so1utions.l2 The mort likely cause for the high-frequency arc appears to be a charge- J. MATER. CHEM., 1994, VOL. 4 40 mo AA -60 -70-80 -50 -60-70 -40 -50 -60 -30 -40 -50 -20 -30 -40 -10 -20 -30 290 370 450 530 610 TI% Fig. 5 Evaluation of (Uo.4Pr,,6)02~x+Pt composite electrodes in potentiometric oxygen sensors over the temperature range 300-600 "C.The (Uo,4Pro,6)02kxelectrode surface area (m2 g-'): C1, 39; C2, 17; C3, 9.5. Air us. 1.0% O2 in N,. Open symbols, heating cycle; filled symbols, cooling cycle. transfer process. Strong evidence for this is presented by the decrease in the size of the high-frequency arc (and increase in the rate of reaction) resulting from high-temperature heat treatment which is expected to increase both the adhesion and the contact area between electrode and electrolyte phases. It is the charge-transfer process at the electrode/electrolyte interface which is likely to be facilitated with increase in the contact area between the two phases.Diffusion of oxygen ions within the electrolyte in a small layer near the electrode/ electrolyte interface cannot be considered as the rate-limiting process for this arc as the time constant for oxygen-ion migration within zirconia-yttria electrolytes is independent of oxygen partial pressure over a wide oxygen partial pressure range ( 1-10-22 atm at 800 "C). Moreover, the activation energy for oxygen-ion migration in the electrolyte in the temperature range of this study is <100kJ mol-l, i.e. much lower than that observed for the high-frequency arc. Based on the available evidence, this arc can be attributed to a charge-transfer process (step 3) at the electrode/electrolyte interface. Thus the oxygen exchange reaction for (U0,4Pr0.6)02fx electrodes is controlled mainly by the slow oxygen exchange at the gas/electrode interface with charge transfer at the electrode/electrolyte interface making a minor contribution.Conclusions For (Uo.4Pr,~6)02+xelectrodes two processes mainly contrib- ute to the overall electrode reaction. The electrode process relaxing in the lower frequency range appears to be associated with adsorption/dissociation of oxygen at the gas/electrode interface. The high-frequency arc appears to be associated with the charge-transfer reaction. For these electrodes the overall kinetics for oxygen exchange are controlled by the slow reaction at the gas/electrode interface. These electrode materials can be prepared in fine form with a large surface area.Fig. 5 shows the performance of zirconia potentiometric sensors, incorporating (Uo.4Pro.6)02_fxmixed with ca. 20 wt.% Pt electrodes, as a function of temperature and electrode surface area. In all cases the sensor performance was good down to 300-320 "C; however, the best behaviour was observed for electrodes prepared with large surface area (Uo.4Pro.,)02+, powder. Such electrode materials show potential for use in zirconia solid electrolyte devices. The author thanks Dr. S. P. Jiang for reading this manuscript. References 1 I. B. de Alleluia, M. Hoshi, W. G. Jocher and ('. Keller, J. Znorg. Nucl. Chem., 1981,43, 1831. 2 V. F. Hund and U. Peetz, 2.Anorg. Allg. Chem.. 1952, 271,6. 3 E. Stadlbauer, U. Wichmann, U. Lott and C. Keller, J. Solid State Chem., 1974,10,341. 4 S. P. S. Badwal and D. J. M. Bevan, J. Muter. Sci.,1979, 14, 2353, 5 S. P. S. Badwal, D. J. M. Bevan and J. O'M. Bockris, Electrochim. Acta, 1980,25, 1115. 6 S. P. S. Badwal and F. T. Ciacchi, J. Appl. Electrochem., 1986, 16, 28. 7 J. R. Macdonald, personal communication. 8 E. A. Aitkin and R. A. Joseph, J. Phys. Chem., 1966,70, 1090. 9 S. P. S. Badwal, Ph.D. Thesis, The Flinders University of South Australia, 1977. 10 E. A. Aitkin, J. Nucl. Muter., 1966, 19,248. 11 J. F. Wadier, CAE-R4507, Atomic Energy Commission, C.E.N., Fontenay-aux-Roses, France, 1973. 12 S. P. S. Badwal and D. J. M. Bevan, J. Aust. Ceram. Soc., 1978, 14, 1. 13 Impedance Spectroscopy, ed. J. R. Macdonald. John Wiley, New York, 1987. 14 F. Grosz, S. C. Singhal and 0.Yamamoto, Proc.. 2nd Int. Symp. on Solid Oxide Fuel Cells, Commission of European Communities, Brussels, EUR 13564 en, 1991. 15 S. C. Singhal and H. Iwahara, Proc. 3rd Int. Symp. on Solid Oxide Fuel Cells, The Electrochemical Society, Princeton, NJ, 1993. 16 B. C. H. Steele, Mater. Sci. Eng. B, 1992, 13,79. Paper 4/02253J; Receiced 15th April 1994
ISSN:0959-9428
DOI:10.1039/JM9940401437
出版商:RSC
年代:1994
数据来源: RSC
|
17. |
Phase diagrams and stoichiometries of the solid electrolytes, Bi4V2O11: M, M = Co, Cu, Zn, Ca, Sr |
|
Journal of Materials Chemistry,
Volume 4,
Issue 9,
1994,
Page 1441-1444
Chnoong Kheng Lee,
Preview
|
PDF (514KB)
|
|
摘要:
J. MATER. CHEM., 1994, 4(9), 1441-1444 Phase Diagrams and Stoichiometries of the Solid Electrolytes, Bi,V,Ol1: M, M =Co, Cu, Zn, Ca, Sr Chnoong Kheng Lee: Goik See Limaand Anthony R. We& a Chemistry Department, Universiti Pertanian Malaysia, 43400 Serdang, Selangor, Malaysia Chemistry Department, University of Aberdeen, Meston Walk, Aberdeen, UK AB9 2UE The compositional ranges of Bi,V,O,, solid solutions containing Co, Cu, Zn, Ca and Sr have been determined by means of a phase diagram study at solidus temperatures. Depending on the size of the divalent cation, at least three and probably four mechanisms for accommodating variable cation content have been identified. These are: V Z$ Bi, V $ M, Bi +M and probably, formation of interstitial M. Most previous studies have focused on the single mechanism V $-M whereas, in fact, the solid solutions are shown to be far more varied and extensive than previously supposed.A new family of oxide ion conducting solid electrolytes, known as BIMEVOX, has been prepared by doping Bi,V2011.1-3 The highest conductivities have been reported for the partial substitution of V by Cu or Ni, with conductivities as high as 11 10-3 ~-cm-' at 240 "C. Bi4V2011 has three crystallo- graphic polymorphs: a (stable below ca. 460 "C), fi (stable between cu. 460 and 560 "C)and y (stable above 560 "C).High levels of oxide ion conductivity occur in the y-form, which is stabilised to lower temperatures by the partial substitution of V by Cu or Ni. The explanation usually given for the high conductivity is that y-Bi,V2011 has a layer structure, con- sisting of alternating sheets of constitution Bi2022f and V03,5@o.52-, where 0 refers to oxide ion vacancies which are randomly distributed and are responsible for the ionic conductivity.Doped Bi,V201 materials are currently under intensive study in several laboratories because of their attractive electri- cal properties. They are, however, complex materials both structurally and compositionally. Thus, although the parent formula is given as Bi4V2011, and most doping studies start with this formula, the phase has, in fact, a variable Bi:V A phase diagram study of the relevant region of the join Bi203-V2054 showed that Bi,V,O,, solid solutions form over the range ca.66.8-68.5 mol% Bi203 at 400 "C (a-poly-morph). Solid solutions extend further with increasing tem- perature and at solidus temperatures of ca. 860°C cover the range ca. 66.6-70.3Oh Bi,O, (y-polymorph). With increasing Bi20, content, melting temperatures increase gradually from cu. 860 to 890°C and both the cr~pand /?$y transition temperatures decrease slightly. Also with increasing Bi203 content, it becomes possible to preserve first the fi-and then the y-polymorph to room temperature by rapid quenching., The existence of this range of Bi4V2011 solid solutions is now well e~tablished;~,~ however, as yet no satisfactory mechanism for the incorporation of excess Bi (or deficiency in V) has been given. Several structural studies of cr-Bi4V2011 have been rep~rted.~~'Although there is a clear similarity to Aurivillius phases, such as Bi2W06,6q8 closer examination has highlighted certain difference^.^,^ Aurivillius phases contain layers of con-stitution Bi,022 : square-pyramidal units may be identified + in which Bi at the apex is coordinated to four oxygens at the corners of the base.These units link by sharing edges to form infinite Bi2O2 sheets.' By high-resolution electron microscopy these Bi202 layers were not seen in Bi4V2011;5 single-crystal X-ray diffraction showed that layers are present but are formed instead of very irregular, corner-sharing BiO, tetra-+hedra, giving an overall formula BiZ02,750.5. Aurivillius phases also contain layers of corner-sharing X06 octahedra, of overall constitution XO,"-.In a-Bi,V201,, however, and unlike Aurivillius phases, some oxygens are common to both the V-0 and Bi-0 layers since they belong to the coordi- nation sphere of both Bi and V; consequently, the V-0 layers have the effective formula -.' The oxygen atoms O(3) which link Bi and V have an occupancy factor of 0.75. The structural relationships between a-,/?-and y-poly- morphs have been summarised inoterms of an y-thorhombic 'mean' cell, a, =5.53 A, b, z 5.61 A, c, =15.28 A.139 In the a-polymorph a supercell is present with a=3a, whereas in the p-polymorph, the supercell is ca. 2a,. The y-polymorph is tetragonal with atet=a,/J2. Numerous reports have been given of success in stabilising the y-polymorph to room temperature by doping.These mainly involve the partial substitution of V by lower-valent cations such as Co, Ni, Cu, Zn'-3 and Ge." Doping with Mo leads to stabilisation of the p-polymorph and its supercell was found to be incommensur- ate with a periodicity that varied smoothly with Mo content.' The possibility of other substitution mechanisms involving Bi sites as well as V sites has been less studied. Significantly, however, a range of solid solutions formed by substitiition of Pb onto Bi sites has been reported, together with possible substitution of Pb onto V sites." Following on from the phase diagram study of the Bi,O,-V,O, join showing a range of Bi-rich solid solution^,^ similar results were found in Ge-doped materials" and Pb-doped materials," demonstrating that at least two solid-solution mechanisms are required to describe properly the structures and stoichiometries of such doped materials. Many conductivity studies on pure and doped BI,V~O~, have been reported, but the electrical behaviour is not simple.In Bi4V2011, three approximately linear regions in conduc- tivity Arrhenius plots are seen, corresponding to the a-, p-and y-polymorphs, but only data for the y-polymorph appear to be fully reversible on heat/cool cycles. Data for the a-polymorph especially appear to depend on sample history and are sensitive to moi~ture.~The nature of the mobile species in the a-polymorph is not known, but in the p-and y-forms, transport-number measurements indicate the trans- port number for oxide ions to be close to unity.' A detailed study of the preparation and properties of the highly con- ducting BICUVOX.10 composition showed non-linear Arrhenius behaviour for the tetragonal (y) phase, with a smooth change in slope at 486-508"C7depending on the sample.12 The change in slope was attributed to a certain orderdisorder transition in the oxide ion arrangement, although it is not clear whether this transition is also seen as a thermal effect by differential thermal analysis (DTA 1." In parallel with the development of doped Bi,V,Oll as oxide ion conductors, they have also been shown to exhibit ferroelectric and pyroelectric proper tie^.^,'^-'^ The properties of Ge-doped single crystals (V CGe doping mechanism) and Nd, Gd-doped ceramics (Bi +Nd, Gd) have been studied.'s319 Our approach to the investigation of doped Bi vanadates has been to use phase diagrams as the first step in synthesizing and determining the stoichiometry ranges of doped materials. This enables different possible doping mechanisms to be evaluated and yields single-phase doped materials suitable for systematic structural studies and property measurements.Here we report the effect of a range of divalent dopant cations on the stoichiometry of Bi,V,O,, and show that the substi- tution mechanism(s) depend on cation size. Experimental Reagents used were Bi203 (99.9%, Aldrich), V205 (99.8%, Aldrich), CuO Aldrich), ZnO (99 + YO, Merck), ST(NO~)~(99 +YO,Aldrich), CaO (99.90/0, Aldrich), C0(C204) (Johnson Matthey).Some were dried prior to weighing at: 300'C (Bi,O,, V2OS), 600°C (CuO), 700°C (CaO), 800'C (ZnO). Co and Sr salts were used undried. Compositions were weighed (ca. 3 g total), mixed with acetone in an agate mortar, dried and fired in Au foil boats at temperatures in the range 150-900 "C, depending on composition. Typically, final firing was at 820-880°C 20 h. Samples were analysed by X-ray powder diffraction (XRD, Philips diffractometer, Cu-Ka, radiation). For DTA a DuPont 991 instrument with a 1200 "C cell and a heating rate of 10 "C min-' was used. Results For each divalent dopant cation, the relevant region of the phase diagram Bi2O3-V2OS-MO was investigated in order to establish the compositional extent of the single-phase Bi,V20,, solid solutions.In each case, the effects of reaction temperature have to be considered since the solid solutions may be more extensive at high temperatures, as found on the Bi203-V205 join.4 The cooling rate is also important since sometimes, the various -y+/?+a transitions may be suppressed by rapid cooling. The phase diagrams presented here refer to final reaction temperatures in the range 850-900 "C, depending on composition (these temperatures were found by trial and error to be close to the solidus) after which samples were removed from the furnace and allowed to cool naturally in air. In order to facilitate discussion of the results, and in particular to assess the likely substitution mechanism(s), the directions of possible solid-solution formation for Bi4V2OI1 doped in various ways are shown in Fig.1. Since the divalent dopant M is of different charge to both Bi and V, additional compensation mechanisms are required. The various schemes shown in Fig. 1 cover the range of possible ionic compensation mechanisms, including formation of both cation interstitials and anion vacancies. It is important to recognise that proper crystallographic studies are required to determine the precise structural details of each substitution; for instance, in cases where the overall mechanism is substitution of M for V, direction(2), the precise location of M, site symmetry, bond distances etc., cannot be inferred from the phase diagram.In Cu-doped materials (Fig. 2) the solid solutions extend parallel to direction (2), Fig. 1, indicating the principal substi- tution mechanism to be V +Cu. The general formula may be written as: Bi4+yV2-y-xMxOll -y-3xj2; from Fig. 2, the solid solutions cover most of the compositional range 0 <y <0.2; 0<x <0.3 (for y =0, the maximum x is 0.27). On the 'stoichio- metric join', y=O, the a-polymorph is obtained for ~(0.15; J. MATER. CHEM., 1994, VOL. 4 M Fig. 1 Loci of solid solutions for different substitution mechanisms Bi4v2-xc~x011 -3x12 00 -0.2 -0.1 0 0.1 0.2 0.3 0.4 Bi4+yv2-y011-y Fig. 2 Compositional extent of Cu-containing Bi4\'2011: A, a; 0,1'; and 0,mixture of phases all other compositions studied, including those for y>O, gave the 1'-polymorph.Literature reports on Cu-doped materials have been con- fined exclusively to the stoichiometric join, and are similar to those found here, both in terms of the ato y changeover and the maximum x.',~' In Co-doped materials (Fig. 3) the maximum solid solution extent is displaced to be somewhere between directions 2 and 3; it is also much more extensive than for Cu substitution, with, for instance, a limit of x=O.66 at y=O compared with x=0.27 at y=O for Cu. This value is also significantly greater than the value of x=0.55 reported in ref. 21. Our firing temperature was 860 "C whereas that used in ref. 21 was only 810°C and this may account for the discrepancy.The Co solid solutions extend to negative values of y, unlike the Cu solid solutions and the V-rich limit of the solid solution area is roughly parallel to direction (3). The most plausible com- pensation mechanism for such compositions may ideally be that represented by direction (3), with 2V +5co; such a mechanism would clearly necessitate the formation of inter-stitial Co2+ ions. Any other mechanism to account for the Co-rich solid solutions would involve a disruption in occu- pancy of the Bi sites by substituting V and/or Co for Bi. As with the Cu system, only compositions with y=O and low values of x gave the a-phase; all others gave ;'. In Zn-doped materials (Fig. 4)a large solid-solution area is again formed, broadly similar to that with Co.The main differences from Co are that the solid solutions extend to yet more negative values of y, making it impossible to account for these using mechanism (3) as a limiting mechanism (dotted J. MATER. CHEM., 1994, VOL. 4 Bi4V2 xcoxoll -3x/2 0 foo Fig. 3 Compositional extent of Co-containing Bi,V,O,,: A, a; @, y;A,%+other phases; and 0,]'+other phases 0 0 mechanism (3),' Bi4V2 -xznxoll -3x12 0 mechanism 0 0 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 Y Fig. 4 Compositional extent of Zn-containing Bi4V2OI1:A, a; W, p; and @, 1' line, Fig. 4). Instead, we must consider the possible partial occupancy of Bi sites by Zn, according to mechanism (5) (dashed line, Fig. 4), together with creation of interstitial Zn2+ ions.Most of the Zn-containing solid solutions are also y,but some compositions close to the CI to y boundary give p. An additional difference is that the Zn solid solutions have a lower limiting x value, ca. 0.6, than the Co solid solutions, ca. 0.8. In Ca-doped materials (Fig. 5) the solid solutions are most extensive in direction (5), but extend to yet more negative y values than for Zn substitution. In Sr-doped materials (Fig. 6) the solid solution is much smaller than for Ca but is again located in the same general area of the phase diagram; such compositions for both Ca and Sr require mechanism (7) with the simple substitution Bi sCa, Sr, as the limiting mechanism. For these it is no longer necessary to require interstitial cations since the entire area of negative y values can be accounted for using the combination of mechanisms (2) and (7).Fig. 5 Compositional extent of Ca-containing Bi,V,O,,: A. r; @, y; A, %+other phases; and 0,"Jother phases 0 /mechanism (7) 0 -0.2 -0.1 0 0.1 0.2 Y Bi4+yv2-yoll -y Fig. 6 Compositional extent of Sr-containing Bi4V20,1:A, 3; a, 7; A, a+other phases; and 0,g+other phases Discussion The phase diagrams of the five divalent-doped systems reported here demonstrate unequivocally, for the first time, that several mechanisms are required to account for the wide range of single-phase compositions that form. With the excep- tion of the Pb-doped materials," all previous reports have focused only on the substitution mechanism V$ M in stoichiometric Bi,V,O,, .At least three mechanisms are required to account for the various solid solutions. These are: V+Bi (1) V+M (2) Bi+M (7) There is also a strong possibility that mechanisms inkolving creation of interstitial M are needed, such as 2V+5M (3) and Bi+V-+4M (5) These later mechanisms are required, as an alternative to (7), to account for the extension of the Co-and Zn-containing solid solution areas to Bi-deficient compositions (negative values of y). Thus for Ca- and Sr-doping, as for Pb-doping (Fig. 7) the limiting locus of the solid solution area for such Bi-deficient compositions corresponds to mechanism ( 7), i.e. Bi4V2 -x Pbxol 1 -3x12 \ \ \ A /\A .OoO X0.2 A0 \A \ 'A, o.lAl-A' 2.1 / ',fI / -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 Bi4+yV2-y011-y Y Fig.7 Compositional extent of Pb-containing Bi4V2OI1 solid solu- tions, adapted from ref. 11: A,a; and 0,y Table 1 Octahedral ionic radiilpm" coz+ 65-74 Ca2+ 100 Zn2+ 74 Bi3+ 103v5 54 Sr2 118++ Pb2+ 119 aFrom R.D. Shannon, Acta Crystallogr., Sect. A, 1976, 32, 751. Bi +M and there is crystallographic support for this mechan- ism in the case of Pb-doped materials." All intermediate single-phase compositions could therefore be regarded as arising from mechanisms (7) and (2) (negative y values) or (2) and (1) (positive y values) in combination. This is logical since Pb, Ca and Sr are large enough to readily substitute for Bi (Table 1) and would appear also to be not too large to occupy V sites.Zn and especially Co are considerably smaller than Pb, Sr, Ca. They can readily substitute into V sites therefore but may also occupy (unidentified) interstitial sites rather than substitute onto Bi sites. Crystallographic studies are required to elucidate the mechanism for Zn-and Co-containing materials in the region of negative y values. There appears to be a clear correlation between dopant M2+ size and the locus of the solid-solution area. With increasing size in the sequence Co <Zn <Ca <Sr <Pb, the solid solutions extend to progressively more Bi-deficient com- positions. This is interpreted as a greater readiness for the larger ions to substitute directly onto Bi sites, mechanism (7).These phase diagram results raise several important questions. (i) What is the detailed crystallography of mechanism (l)? This point has already been raised in the Introduction. Does Bi substitute onto V sites, with oxygen vacancy creation or are larger, extended defects involved? (ii) Do M2+ ions of intermediate size occupy interstitial sites in the region of negative y values? J. MATER. CHEM., 1994, VOL. 4 (iii) How does the oxygen vacancy concentration vary with composition x,y? (iv) How do electrical properties vary with x, y? Almost all studies to date have looked only at compositions with y= 0 and variable x. Further work is in progress to address these questions. We thank Mr. Azali Md. Sab, Soil Science Department, UPM, for assistance with the X-ray diffraction analyses.C. K. L. is grateful to the Majlis Penyelidikan Kemajnan Sains Negara, Malaysia for financial support, grant no. 2-07-05-009. References 1 F. Abraham, J. C. Boivin, G. Mairesse and G. Nowogrocki, Solid State Ionics, 1990,40141,934. 2 T. Iharada, A. Hammouche, J. Fouletier, M. Kleitz, J. C. Boivin and G. Mairesse, Solid State Ionics, 1991,48,257. 3 F. Abraham, M. F. Debreuille-Gresse, G. Mairesse and G. Nowogrocki, Solid State Ionics, 1988,28-30. 529. 4 C. K. Lee, D. C. Sinclair and A. R. West, Solid State Ionics, 1993, 62, 193. 5 W. Zhou, J. Solid State Chem., 1988,76,290. 6 K. D. R. Varma, G. N. Subbanna, T. N. Guru Row and C. N. R. Rao, J. Muter. Res., 1990,5, 2718. 7 M.Touboul, J. Lokaj, L. Tessier, V. Kettman and V. Vrabel, Acta Crystallogr., Sect. C, 1992,48, 1176. 8 B. Aurivillius, Ark. Kemi, 1950, 2, 519. 9 R. N. Vannier, G. Mairesse, F. Abraham and G. Nowogrocki, J. Solid State Chem., 1993, 103,441. 10 C. K. Lee, M. P. Tam and A. R. West, J. Muter. Chem., 1994, 4, 525. 11 R. N. Vannier, G. Mairesse, G. Nowogrocki, F. Abraham and J. C. Boivin, Solid State Ionics, 1992, 53-56, 713. 12 F. Krok, W. Bogusz, P. Kurek, M. Wasiucionek, W. Jakubowski and J. Dygas, Muter. Sci.Eng., 1993, B21, 70. 13 V. G. Osipyan, L. M. Savchenko, V. L. Elbakyan and P. B. Avakyan, Russ. J. Inorg. Muter., 1987,23,467. 14 V. N. Borisov, Yu. M. Poplavko, P. B. Avakyan and V. G. Osipyan, Sov.Phys-Solid State, 1988,30,904. 15 A. A. Bush and Yu. N. Venevtsev, Russ. J. Inorg. Chem., 1986, 31, 769. 16 K. V. R. Prasad and K. B. R. Varma, J. Phys. D, 1991,24, 1858. 17 K. V. R. Prasad, K. B. R. Varma. A. R. Raju, K. M. Satyalakshmi, R. M. Mallya and M. S. Hegde, .4ppl. Phys. Left.,1993,63, 1898. 18 K. V. R. Prasad and K. B. R. Varma, Ferroelectrics, preprint. 19 K. V. R. Prasad, G. N. Subbanna and K. B. R. Varma, Bismuth Instit Ute, preprint. 20 J. B. Goodenough, A. Manthiram, M. Paranthaman and Y. S. Zhen, Solid State Ionics, MRS Special Publication, Elsevier, 1992, p. 79. 21 R. Essalim, B. Tanouti, J.-P. Bonnet and J. M. Reau, Muter. Lett., 1992,13, 382. Paper 4/01943A; Received 30th March, 1994
ISSN:0959-9428
DOI:10.1039/JM9940401441
出版商:RSC
年代:1994
数据来源: RSC
|
18. |
Reinterpretation of the magnetic structures of the perovskites SrFeO2.710and Sr2LaFe3O8.417 |
|
Journal of Materials Chemistry,
Volume 4,
Issue 9,
1994,
Page 1445-1449
Terence C. Gibb,
Preview
|
PDF (759KB)
|
|
摘要:
J. MATER. CHEM., 1994, 4(9), 1445-1449 Reinterpretation of the Magnetic Structures of the Perovskites SrFe02m,lo and Sr2LaFe,0,~,l, Terence C. Gibb School of Chemistry, The University, Leeds, UK LS2 9JT The mixed-valence oxygen-deficient perovskite SrFeO,,, annealing of rapidly quenched material. The Mossbauer spectra confirm the essential features of two distinct types of iron which can be assigned as Fe3+ and Fe4+ high-spin cations. The unusual characteristics of the spectra from the antiferromagnetic phase, and their similarity to those from Sr2LaFe308.417, have been interpreted in terms of a structural model with layers of Fe3+ in octahedral coordination separated by layers of Fe4+ in square-pyramidal coordination. Ferromagnetic and antiferromagnetic superexchange interactions within the Fe4+ layers can lead to significant frustration and can explain the absence of magnetic hyperfine splitting from these cations below the ordering temperature where the Fe3+ cations show long-range coupling.The magnetism is therefore approximately two-dimensional in character. has been prepared essentially free of other phases by lengthy ,, The perovskite SrFeO, and related compounds have attracted considerable interest. Stoichiometric SrFeO, was first pre- pared under high pressures of oxygen by MacChesney and co-workers'.2 and is one of the few oxide phases to contain iron in the $4 oxidation state. It loses oxygen readily to produce SrFe02.5, which is more commonly referred to as Sr2Fe20S by analogy with the structure of Ca2Fe205 known as brownmillerite.This compound is derived from the cubic perovskite lattice of SrFeO, by removing oxygen atoms from chains along the [llO] direction, and so generating equal numbers of iron sites in distorted octahedral and tetrahedral some degree of contamination with the neighbouring phases. Although much can be deduced from the Mossbauer spectra, there are a number of curious aspects to the magnetic inter- actions at low temperatures which have not been explained. This paper seeks to demonstrate a method for przparing SrFe0,.,5 of much higher quality, and to suggest ;I novel interpretation for the magnetism of the antiferromagnetic phase at low temperatures. Experimental coordination in alternating (001) planes.The large ~upercell~,~ The sample of SrFeO,.,,, has dimensions of J2a, x 4a, x $a, where a, is the cell-size of the cubic perovskite. Mossbauer spectroscopy was found to be useful' in characterising the oxidation states and coordi- nations of the iron cations, but although this early work showed evidence for oxygen deficiency in the region SrFeO,-y (O<y<O.25), it was thought that only a simple perovskite lattice was involved. Later work found evidence for a supercell at SrFe02.,5 which was as 2,/2a, x 2a, x J2ac. It was suggested that a different arrangement of the remaining oxygen vacancies along the [1101 directions would produce equal numbers of octahedral and five-coordinate sites within the (001) planes. Work from this laboratory7 using mainly Mossbauer spectroscopy on phases with (0.15<y <0.25) found evidence for a thermally activated and reversible elec- tron-transfer process above 550 K in STF~O,.~~~.Furthermore, this electron transfer was maintained at higher temperatures while oxygen loss took place until phase separation and the formation of SrFe0,,5 began. It was therefore believed (erroneously) that the composition range was a single phase. The existence of two intermediate phases was established from high-quality X-ray powder diffraction which showed evidence for phases with the nominal composi-tions SrFe0,.75 as orthorhombic 2J2uC x 2a, x J2a, and SrFeO,.,,, as tetragonal 2J2aC x 2J2ac x 2a,. The presence of thermally activated electron-hopping processes in both phases has been confirmed." It has been suggested that these superstructures are generated upon cooling a vacancy-disordered high- temperature phase.The end-member Sr,Fe,O, adopts a microdomain-textured structure9*" above 850"C, although it is impossible to preserve this form by quenching. However, such a matrix has been obtained', more recently by quenching a chromium-substituted material, SrCr0.1Fe0.,02.5.The structures of the two intermediate phases remain unknown. There preparation by quenching the disordered state from an appropriate temperature inevitably leads to was made as follows. Accurately weighed amounts of Fe203 and SrC0, were ground together in a ball mill and fired in a platinum crucible to 1200°C for a total of 6 days with intermediate grinding.Finally, the material was annealed at 600°C for 6 days before quenching it into liquid nitrogen to achieve an oxygen content close to STF~O,,~~. The sample was then sealed under vacuum in a silica tube, and annealed for long periods at 500 (7 days), 450 (7 days), 400 (7 days), 350 ( 11 days) and 300 "C (7 days) before turning off the furnace and allowing it to cool to room temperature. Under these conditions the normal oxidation by uptake of oxygen is prevented while allowing time for equilibration of the defect structure to take place. Chemical analysis by a redox titration gave a final composition of SrFeO,,,,o, 5).. X-Ray powder diffraction data were recorded with a Philips diffractometer using nickel-filtered Cu-Ka radiation.Mossbauer data were collected in the range 78 <T/K <290 using a 57Co/Rh source matrix held at room temperature; isomer shifts were determined relative to the spectrum of metallic iron. Results and Discussion The normal method of preparing samples of SrFe0,-in the intermediate composition range is to quench material which has been equilibrated in air at a chosen temperature. Very slow cooling in air from 1100 K results" in a steady uptake of oxygen until ca. 650K and a composition of about SrFe02.83. However, the latter composition undergoes ii trans-ition on heating from tetragonal to cubic symmetry at 570 K (300 "C), and STF~O,.~~ similarly shows an orthorhonibic to cubic transition at 670 K (400°C).Thus the formation of STF~O,.~~by quenching from ca. 900 K involves a fast lattice rearrangement from an initially disordered state, so that inhomogeneity and phase separation occur and the product 1446 is contaminated with small amounts of other materials. This prevents complete structural analysis on a bulk sample by diffraction techniques. Furthermore, the comparative ease with which these phases lose oxygen on heating and in uucuo reduces the value of electron diffraction studies. In the present work a large sample (log) nominally of was made by quenching rapidly from 600°C into liquid nitrogen. The X-ray powder pattern of this material was consistent with the published data for the phase.The sample was then sealed into a silica ampoule under vacuum and annealed at 500°C for an extended period, followed by similar periods at decreasing temperature intervals of 50 "C until a final anneal at 300°C. In this way it was hoped to allow equilibration of the oxygen defects, although no attempt was made to assess the effectiveness of each particular tem-perature stage. The cubic to orthorhombic transition takes place in the vicinity of 400 "C. The final product was analysed and gave a composition of SrFe02.710(5).The X-ray powder diffraction pattern was similar to that of the starting material, but the lines were considerably sharper, and it was now possible to resolve many more of the multiplets and to obtain a convincing analysis of the unit cell.The d-spacings are give? in Table 1. The cell dimensions obtained were a = 10.980(8) A, Table 1 X-R?y data for SrFe02.710: a = 10.980(8) A,b = 7.704(6) A, c =5.474(3) A h k 1 dobs /A dCdA intensity 0 1 0 7.704 7.769 vw 1 1 0 6.304 6.306 vw 1 1 1 4.135 4.134 W 2 0 1 3.877 3.876 W 2 2 2 3.149 vw0 4 00:i0 0 2 2.734 2.737 vs 2 2 1J 2.732 1 1 2 2.509 2.511 W 1 3 0 2.500 2.500 W 4 2 0 2.235 2.235 m 0 2 2 2.230 2.231 m 5 1 07 3 2.112 w'1 { ;:;:]3 30 4 0 2 1.9380 1.9382 S 0 4 1.9248 1.9260 S 2 1.7335 vw4 ; -I.>0 {::::::} V w2 4 1 1.7238 1.7248 6 2 1 1.5816 1.5824 S 2 2 3 1.5789 1.5793 S 1.5754 S0 4 : :}0 { ::z}vvw7 1 1.5351 1.5371 5 3 2 1.4250 1.4249 W 8 0 0 1.3727 1.3725 m 0 0 4 1.3692 1.3685 m 4 1.3663 m 6 2 :}0 4 3 1.2885 vw 2 1.2876 5 4 7 1.2798 1.2799 vw 8 0 2 1.2266 1.2269 mw 6 1.2250 m4 ;6 :}2 1 1.2188 1.2188 mw 1 1.1694 W8 ; i} {::::;:} mw8 4 0 1.1179 1.1177 0 4 4 1.1155 1.1155 mw The d-spacings given in italics were not used in the cell refinement because they represent very weak lines or ambiguous assignments.J. MATER. CHEM., 1994, VOL. 4 b =7.704(6) A and c =5.474(3) A, corresponding to a cell of 242a, x 2a, x J~u,. These figures $gree well yith the earlier result^^*^ of a = 10.972A, b =7:700 A, c =5,47 I A for a sample of SrFe02.73and a = 10.981 A, b =7.705 A, I =5.474 A for" SrFe02.75.It is not clear whether the phase has a significant composition range. Several groups appear to have prepared 'good' samples with a marginally low oxygen content.The Mossbauer spectra of the present sample described later establish that it contains less contaminating phases than in previous work, but these and the X-ray patterns are not very sensitive to small amounts of other perovskitc-related phases. There is also the possibility of phase intergrowth on (001) planes, which would result in compositional variation. It is also feasible that there are additional oxygen vacancies in an otherwise homogeneous phase. Previous work' has suggested a composition range of 2.68 < 3 -y 6 2.73, and it is possible that a stoichiometric phase with y =0.25 does not exist as such. The Mossbauer spectrum for the annealed sample at 290 K is shown in Fig.1. The lines appear to be much sharper than in previously published spectra and are also sharper than for the initial quenched material; there is no evidence for contami-nation by SrFe02,5or STF~O~,~~~.The spectrum has been fitted with two symmetrical doublets and the values for the isomer shift 6, quadrupole splitting A and linewidth r are shown in Table 2. These doublets may be assigned on the basis of different isomer shifts to Fe3+ (more positive 6) and Fe4+cations in approximately equal numbers (as required for STF~O~.~~).The model used is not unique in the sense that the overlapping components can be assigned in other ways while retaining a comparatively good fit. However, in such cases the isomer shift for the Fe3+component is then unrealist-ically high when compared to other perovskite phases and is also inconsistent with the isomer shift observed at 78 K and the expected second-order Doppler shift.However, the quad-rupole splitting of the Fe4+ component in particular can be regarded as less well defined than the statistical errors would imply. The spectrum at 78 K is shown in Fig. 2. As described previow~ly,~-~~the phase achieves antiferromagnetic ordering below ca. 220K. The Fe3+ spins show a large magnetic hyperfine splitting, which saturates at the rather low value of 46.1 T. The linewidth remains broader than in the paramag-netic state, suggesting that there is some underlying structural disorder which can affect the magnetic field slightly. The behaviour of the Fe4+ component is rather unusual.The expected resolved hyperfine splitting does not appear even at 4.2 K, although with decreasing temperature there is eventu-ally some degree of broadening. Previous work has failed to explain this observation in a convincing way. Various opinions have been expressed as to the structure -4 -3 -2 -1 0 1 2 3 4 vetocity/rnm s-' Fig. 1 57FeMossbauer spectrum of SrFeO, at 290 K J. MATER. CHEM., 1994, VOL. 4 Table 2 Mossbauer parameters for SrFe02,710 at 290 and 4.2 K compound T/K 6/mm s-' A/mm s-' T/mm s-' BIT 1raction SrFeO,.,,, 290 0.353( 2) 1.323 (2) 0.247( 2) -0.500 -O.OSO(2) 0.316(2) 0.231(2) -0.500 78 0.455(5) -0.673( 5) 0.523(5) 43.95( 1) 0.495 -0.053( 5) -0.912(5) -0.505 Sr,Fe,O, 295 0.37 -0.35 50.1 0.5 0.7 0.30 42.2 0.5 Sr2LaFe308.417 290 0.327(5) -0.388( 5) 0.70(2) 43.46( 4) 0.654 -0.165(10) -1.27(3) -0.346 The quadrupole splitting in cases where there is also a non-zero magnetic field B is in fact a quadrupole perturbation parameter, E.I .-.. Fig.2 ,'Fe Mossbauer spectra of (a) SrFe02,710 at 78 K and (b) Sr2LaFe308,417at 290 K. The two compounds have different ordering temperatures and the temperatures are chosen to represent approximately the equivalent point on the Brillouin curve. Note the strong similarity and the stronger intensity of the magnetic sextet in SrzLaFe308.417,which has an increased Fe3+ content. of the idealized SrFeO,,,, lattice. The first model proposed5 for a cell of 242~1,x 2a, x 2$a, embodied strings of vacancies along [I1lo], in alternate (Ool), planes as in brownmillerite, but containing only half the number of vacancies so as to generate equal numbers of five and six-coordinate sites in every (OOl), plane.The two types of polyhedra alternate along [OOl],, and alternate (OOl), planes are identical but displaced by 42a,. The revised cell' of 242a, x 24 x ,/2a, which is also compatible with the new results was formulated on the basis of improved X-ray data supported by electron diffraction patterns. An almost identical vacancy model was proposed' but with a reduced c axis. It was also suggested that the cell showed an extinction rule of h+k=2n, and thus could be assumed to be base-centred.The new observation of a very weak 010 reflection in the X-ray powder diffraction data from the improved preparation suggests that this is not the case, and that perhaps the dominant scattering from the heavy metals imposes a pseudo-symmetry. Note that the overall pattern closely resembles a cubic perovskite but with partial splitting of the high-angle reflections and only very weak intermediate reflections. This explains why the supercell was initially mis-identified, and why it is difficult to determine sample purity from X-ray data alone. The hyperfine interactions observed in the Mossbauer spectra have led to differing interpretations. The present work establishes that there are only two iron sites which can be described as Fe3+ and Fe4+, respectively, in equal proportions (other species reported in previously recorded spectra are thus due to impurity phases).It can therefore be concluded that these sites relate to five and six coordination only. The isomer shift of the Fe3+ component appears to be compatible with a high-spin d5 cation in six-coordination and is similar to the isomer shift for six-coordination in Sr2Fe205 (Table l), although there are so few examples of five-coordination in iron oxides that a degree of ambiguity does exist. The magnetic hyperfine field has a flux density at saturation of only 46.1 T, which is unusually low for six-coordination and can be compared to 53.9 and 46.4T at the octahedral and tetrahedral sites, respectively, in Sr2Fe,0,.It is generally agreed'.' from examination of the quadrupole perturbation of the magnetic spectrum that the magnetic field is effectively parallel to the principal axis of the electric field gradient. The large quadru- pole interaction has prompted speculation' that the Fe3 site+ is five-coordinate, but an even larger field gradient is observed at the octahedral sites in Sr2Fe205 despite an apparently regular coordination geometry, and thus the deduction is unsubstantiated. The Fe4+ component is clearly designated by the lower isomer shift, but is unusual in that there is no significant magnetic splitting, although the broadening of this component which gradually takes place with decreasing temperature may well be magnetic in origin.It has been suggested' that this is due to the presence of low-spin d4 Fe4+ which has a 3T,g ground state with a J=O state lying lowest. This is consistent with the rather low magnetic moment recorded7 above 220 K, but does not explain why it should be strongly field-sensitive below the critical temperature. However, an alternative description" in terms of Fe3+ in octahedral sites and high- spin Fe4+ in five-coordination has been promulgated. A high-spin configuration seems more likely in view of the thermally activated electron-transfer processes, and it has been recently demonstrated that high-spin Fe4+ is present'3714 in the related compound Sr,FeO,. Further evidence15 to support octahedral Fe3+ comes from the similarity of SrFeO,.,,, with the compound sr2LaFe308,417.This material orders magnetically below ca.500 K and shows very similar Mossbauer spectra. Here also an averaged electron state is seen above 500K. A typical spectrum at 290 K is shown in Fig. 2 together with that from SrFeO,.,,, at 78 K. The sign of the quadrupole perturbation of the Fe3+ component is the same, although the magnitude is different, and the Fe4+ component remains comparatively sharp until about 150 K, whereupon it broadens soniewhat but never achieves a static hyperfine pattern. The Fe3+ component comprises some 62% of the total, compared to an estimated 61% of octahedral sites for a perovskite-related structure with corner-shared polyhedra. This leads to a model for the idealized composition Sr,LaFe30,.5 in which (001), layers of Fe4+ in five-coordinate sites are separated by two layers of Fe3+ in octahedral sites.It is also significant that J. MATER. CHEM., 1994, VOL. 4 the structure16 of Sr,LaFe308 (the Sr and La are disordered) is related to the brownmillerite Sr,Fe,O, by replacing the octahedral layers with double-octahedral layers. Layers of +Mn3 ( high-spin d4 and isoelectronic with Fe4+ ) in square- pyramidal coordination exclusively in Ca,Mn,O,, rather than mixed coordinations. This compound is anti- ferromagnetic below ca. 350 K (precise magnetic structure unknown) and the identical layers generate chains of corner- sharing pyramids perpendicular to the layers. This paper seeks to establish that a model embodying alternate layers of six-coordinate Fe3 and five-coordinate + Fe4+ can be used to give an explanation for the unusual magnetic properties of SrFeO,.,,, .A possible arrangement of the atoms in the five-coordinate (OOl), layers is illustrated schematically in Fig. 3. The unit-cell dimensions in the ac plane are indicated by the dashed lines. Support for this model comes directly from the structure of Ca,Mn,O, which adopt^'^,'^ essentially this arrangement. The octahedral layers above and below share vertices in the conventional manner. The nominal square-based pyramids will ensure a large ligand- field effect on the d4 configuration, such that the singly occupied eg state in octahedral symmetry has now split to give orthogonal singlet orbitals aligned along (occupied) and perpendicular (empty) to the axis of the pyramid.The align- ments of the occupied orbitals are drawn in Fig. 3. The magnetic structure will be determined by the superexchange interactions between the cations. The two-dimensional superexchange within the Fe3 layers+ will be unexceptional with strong antiferromagnetic coupling between nearest neighbours. The superexchange between Fe4+ cations requires more careful examination, although the pri- mary factors have been reviewed" in depth. The o-super- exchange transfer from a half-filled to an empty orbital can lead to the possibility of 180" Fe-0-Fe ferromagnetic coupling. The o-superexchange between half-filled orbitals and the weaker n-transfer from the half-filled t,, orbitals will always be antiferromagnetic. In the present instance the oxygen vacancies will exert a controlling influence on the Fe4+ superexchange.Thus from Fig. 3 the superexchange coupling between sites labelled A and B can be predicted to be ferromagnetic, and between B and C antiferromagnetic. However, all the a-superexchange Fig. 3 A schematic representation of the square-pyramidal Fe4+ layer in SrFeO,,,,,. The square-pyramids share three vertices and are also joined to the octahedral Fe3+ sites in the adjacent layers. The d4 electron configuration ensures that the two d-orbitals, which can take part in a-transfer super-exchange are either half-filled (as shown) or empty.The result is a ferromagnetic coupling between atoms A-B, C-D, D-E and F-A, but antiferromagnetic coupling between B-C and E-F, leading to overall conflict and frustration in the spin system. interactions between Fe3+ and Fe4+ in adjacent layers will be equivalent because they feature ferromagnetic a-transfer by overlap of a half-filled and an empty orbital, together with an antiferromagnetic n-transfer. If atom A couples ferromag- netically to Fe3+ atoms in the layers above and below, then the spin of atom B is in conflict with the strong antiferromag- netic exchange within the Fe3+ layers. Furthermore, the extreme anisotropy of the square-pyramidal sites may try to impose spin alignments which are in conflict with the superexchange. As a result of these considerations, it seenis probable that the Fe3+ spins within a layer which are subject to strong superexchange interactions will be antiferromagnetic.It is also possible that weak long-range coupling between these layers may be able to support three-dimensional antiferomagnetism. The Fe4+ cations within a layer are subject to conflicting superexchange interactions which may result in a high degree of frustration and a rapid relaxation of the spins. The sensi- tivity of the effective magnetic moment to the measuring field below the critical temperature would support this possibility. Thus the large magnetic flux densities obser~ed'~*~~ for Fe4+ in Sr,FeO, are not observed in SrFeO,.,,,. One of the unusual features of the Mossbauer spectra of SrFe02.710 is the comparatively low value of the flux density, B, of the hyperfine field of the Fe3+ component which is only some 46.1 T at saturation.This figure is compared in Table 3 with values of the flux densities in other perovskite-related compounds containing Fe3 + in octahedral coordination. The value of B in the orthorhombic perovskites2'such as LaFeO, and LuFeO, is in the range 54-56 T. A small contribution to B comes from a transferred hyperfine field from nearest- neighbour spins, and this has been shown2' to be of the order of 1 T per Fe3+ neighbour. Thus generating an isolated octahedral layer would only reduce the field from transferred hyperfine effects by some 2 T and cannot explain the present observation.Octahedral bilayers are fo~nd~'.~~ in Sr,LaFe,O, and in SrEu,Fe,O, and also show large values of B. Single octahedral layers are found in Sr,Fe,O, (53.9 T) where there is strong coupling via tetrahedral layers to produce 3D magnetism, and in SrLaFeO, (50.0 T),,, where an intervening SrO layer in the K2NiF4 structure weakens the superexchange and the magnetism approaches that of a 2D magnet. The decrease of ca. 4T in the saturation field is noteworthy, and there is clearly a very significant reduction over and above the transferred hyperfine effect upon lowering the dimensional- ity of the magnetism. The low flux densities in Sr,LaFe,O,~,,, (51.9 T) and SrFeO,.,,, (46.1 T) which are now postulated to contain octahedral bilayers and isolated layers with at best weak interlayer coupling are now understandable.It is also interesting to note the low flux density,, in CaFe,O, (48.5 T) where there is a strong coupling within irregular chains of octahedra, with weaker coupling between chains. If the Fe3 + spins in SrFeO,.,,, are aligned along the b axis (which seems the logical choice for the principal value of the electric field gradient tensor) then the octahedral layers may well have the characteristics of a two-dimensional Ising magnet. Is it feasible to have a partial magnetic order in which magnetically ordered layers are separated by layers of frus- trated cations? Evidence to suggest that it is possible has come from work on an unrelated oxide system.The structure of Lu2Fe30, can be de~cribed~"'~ in terms of three cation layers, U, V and W. The Lu3+ occupies the U (Lu203) layers. The V layers (FeO1.,) contain Fe3+ in trigonal-bipyramidal coordination in a triangular net, and the W double-layers (Fe2O2.5) feature edge-sharing bipyramids containing equal numbers of Fe2+ and Fe3+. The overall stacking sequence is U-V-U-W. Neutron diffraction establishes strong antiferro- magnetic interaction in two dimensions within the layers, but J. MATER. CHEM., 1994, VOL. 4 Table 3 Saturation flux density in perovskite-related compounds compound coord. BIT structural features dimension LaFeO, LuFeO, Sr,LaFe,O, SrEu,Fe,O, Sr,Fe,O, SrLaFeO, Sr2LaFe308.417 CaFe20, Sr2Fe0, 6 6 6 4 6 6 4 6 6 5 6 5 6 6 56.4 54.5 55.5 44.3 54.5 53.9 46.4 50.0 51.9 ? 46.1 3 48.5 32.9 orthorhombic perovskite orthorhombic perovskite octahedralbilayer tetrahedral layer octahedral bilayer octahedral layer tetrahedral layer octahedral layer octahedral bilayer pyramidal Fe4+ octahedral layer pyramidal Fe4+ irregular chains octahedral layer (Fe4+) 3D 3D 3D 3D 3D 2D 2D 2D 1D 2D 30.3 (spin arrangement complex) 27.2 24.6 with a much weaker interlayer interaction. The Mossbauer spectra show that the Fe cations in the W layer develop a magnetic hyperfine pattern below 260K, but those in the V layer remain as a collapsed doublet until below 70 K.The long-range order is strongly two-dimensional in nature. The 5 6 7 B. C. Tofield, C. Greaves and B. E. F. Fender, Muter. RCY.Bull., 1975, 10, 737. B. C. Tofield, Proc. 8th Int. Symp. on Reactivity of Solids, ed. J. Wood, 0. Lindqvist and C. Helgesson, Plenum Press, New York, 1977. T. C. Gibb, J. Chem. SOC., Dalton Trans., 1985,1455. triangular net of atoms in the V layer will be inherently frustrated, and coupling to the W layers becomes important only below 70 K. With specific reference to SrFeO,,,,, and Sr,LaFe,08.41,, it seems possible that the structures are based on layered 10 8 9 Y. Takeda, K. Kanno, T. Takada, 0. Yamamoto, M. Takano, N. Nakayama and Y. Bando, J. Solid State Chem., 1986,63,237. M. Takano, T. Okita, N. Nakayama, Y. Bando, Y. Takeda, 0.Yamamoto and J. B. Goodenough, J. Solid State Chem., 1988, 73, 140. L. Fournes, Y. Potin, J. C. Grenier, G. Demazeau and arrangements of cations as suggested above with a strong two-dimensional character to the magnetic behaviour which is produced by an inherent degree of frustration along the b axis.The apparently low value of the effective magnetic moment just above the ordering temperature is consistent 11 12 13 M. Pouchard, Solid State Commun., 1987,62,239. J. C. Grenier, N. Ea, M. Pouchard and P. Hagenmuller, I. Solid State Chem., 1985,58,243. T. C. Gibb, J. Muter. Chem., 1991,1,23. S. E. Dann, M. T. Weller, D. B. Currie, M. F. Thomds and A. D. Al-Rawwas, J. Muter. Chem., 1993,3, 1231. with strong two-dimensional an tiferromagnetic interactions within the octahedral layers. It is also clear that some form of parasitic ferrimagnetism develops immediately below the critical temperature, which complicates any analysis of the magnetic susceptibility.It does not seem possible to explain 14 15 16 17 P. Adler, J. Solid State Chem., 1994, 108,275. P. D. Battle, T. C. Gibb and S. Nixon, J. Solid State Chem., 1989, 79, 75. P. D. Battle, T. C. Gibb and P. Lightfoot, J. Solid State Chem., 1990,84,237. K. R. Poepplmeier, M. E. Leonowicz and J. M. Longo, .r. Solid the features of these compounds by a non-layered arrangement of the Fe4+ cations. It is hoped that the improved preparation technique will ultimately lead to a full determination of the vacancy superlattice from neutron diffraction data. 18 19 State Chem., 1982,44,89. K. R. Poepplmeier, M. E. Leonowicz, J. C. Scanlon, J. M. Longo and W. B. Yelon, J. Solid State Chem., 1982,45,71. J. B. Goodenough, Magnetism and the Chemical Bond, Interscience, New York, 1963. 20 M. Eibschutz, D. Shtrikman and D. Treves, Phys. Rev. 1967, Acknowledgement is made to SERC for financial support. 21 156, 562. T. C. Gibb, J. Chem. SOC., Dalton Trans., 1984,667. 22 T. C. Gibb, J. Phys. C, 1981,14,1985. References 23 24 T. C. Gibb, unpublished data. H. Yamamoto, T. Okada, H. Watanabe and M. Fukase, J Phys. 1 P. K. Gallagher, J. B. MacChesney and D. N. E. Buchanan, J. Chem. Phys., 1964, 41, 2429. 2 J. B. MacChesney, R. C. Sherwood and J. F. Potter, J. Chem. Phys., 1965,43, 1907. 3 C. Greaves, A. J. Jacobson, B. C. Tofield and B. E. F. Fender, Acta Crystallogr., Sect. B, 1975,31, 641. 25 26 27 SOC.Jpn., 1968,24,275. M. Tanaka, N. Kimizuka, J. Akimitsu, S. Funahashi and K. Siratori, J. Magn. Magn. Muter., 1983,31-34, 769. T. Sugihara, K. Siratori, N. Kimizuka, J. Iida, H. Hiriyoshi and Y. Nakagawa, J. Phys. SOC. Jpn., 1985,3,1139. J. Iida, M. Tanaka and S. Funahashi, J. Magn. Magn. .Muter., 1992,104-107,827. 4 M. Harder and H. Miiller-Buschbaum, Z. Anorg. Allg. Chem., 1980,464,169. Paper 4/01594K; Received 17th March, 1994
ISSN:0959-9428
DOI:10.1039/JM9940401445
出版商:RSC
年代:1994
数据来源: RSC
|
19. |
Synthesis under high pressure and characterisation by Mössbauer spectroscopy of non-stoichiometric Ca2Fe2O5.12 |
|
Journal of Materials Chemistry,
Volume 4,
Issue 9,
1994,
Page 1451-1455
Terence C. Gibb,
Preview
|
PDF (730KB)
|
|
摘要:
J. MATER. CHEM., 1994, 4(9), 1451-1455 Synthesis under High Pressure and Characterisation by Mossbauer Spectroscopy of Non-stoichiometric Ca,Fe,O,.,, Terence C. Gibb," Andrew J. Herod, Duncan C. Munro and Nianhua Peng School of Chemistry, The Universify, Leeds, UK LS2 9JT ~ ~ ~ ~~~ The calcium iron oxide Ca,Fe,O, which contains alternate layers of iron-oxygen octahedra and tetrahedra has been found to take up additional oxygen under high pressure and elevated temperature. The Mossbauer spectra in the range 4.2-290 K establish that all the original material has been transformed into a new phase which is still structurally related to the original lattice, but features a disordered introduction of oxygen into the tetrahedral layers and oxidation of some of the iron to the +4 oxidation state.At least six distinct iron sites can be identified, and the coordination numbers established. The structural relationship of these sites and the magnetic characteristics are discussed. Following the discovery of copper superconductors, there has been a renewal of interest in perovskite-related oxides contain- ing Fe4+ because, like Cu2+, the high-spin d4 configuration can be expected to undergo a Jahn-Teller distortion. This is a comparatively rare oxidation state in iron compounds, and its existence is facilitated by preparations carried out under high pressures of oxygen. This has led us to investigate the oxidation of the Fe3+ compound Ca2Fe205, which is one of a number of materials related to the mineral 'brownmillerite' and has been extensively studied by many techniques including Mossbauer spectroscopy.The structure was first determined' by Bertaut et al., and later refined by Colgille' and Ber~gren., The ophorhombic cell with a =5.425 A, b =14.769 A, c = 5.598 A in space group Pnma (no. 62, DiE) contains alternate layers of iron-oxygen octahedra and of tetrahedra normal to the b axis (some descriptions of the structure are given in the alternative setting Pcmn, but for consistency only the standard descriptions for this and other structures will be used through- out). The structure can be derived from the parent cubic perovskite by inserting chains of vacancies along [1101, axes in alternate layers. The individual corner-sharing polyhedra are considerably distorted with a comparatively low sym- metry, although the octahedra are effectively elongated along the b axis.As a result the Mossbauer magnetic hyperfine patterns are quite distinct because of large quadrupole pertur- bations. Mossbauer spectra recorded using linearly polarized prays were used to establish4 that the antiferromagnetically coupled spins in the G-type magnetic structure lie along the a axis, and the same structure has been confirmed by three independent neutron diffraction ~tudies.~-~ The related com- pound Ca2FeA105, known as brownmillerite, which is in fact part of a solid solution in which the aluminium shows a strong preference for the tetrahedral sites, adopts' the different space group Ima2.The magnetic structure shows a subtle variationg in that the spins now lie along the c axis. Thus the compound demonstrates that different spin orientations in the ac plane are possible, although it is still an antiferromagnet because the octahedral and tetrahedral sublattices are also antiferromagnetic within themselves. The similar compound Sr2Fe205 has been refinedloT1' in the space groups Imma and Ima2, and Sr2CoFe05 refinedi2 in Imma. The variations in these various structures can be attributed to small differences in the stacking of the tetrahedral layers and in some cases a degree of stacking disorder has been suggested. However, the Mossbauer spectra are generally insensitive to these effects which have little effect on the individual polyhedra.Although both the parent perovskite phases CaFeO, and SrFe03 can be prepared under high pressures of oxygen, it is not easy to introduce extra oxygen into the tetrahedral layers of the brownmillerite-related phases. The strontium phase does oxidise readily on heating in air, but rapid quenching from different temperatures has been shown13 to produce a mixture of Sr2Fe20, with STF~O~.,~. The structure of the latter is unknown, but the evidence suggests14 that it IS also a layered structure with the tetrahedral layers of Sr2Fe20, replaced by five-coordinate Fe4 sites in square-pyramidal + geometry. There is also a second intermediate phase at STF~O~.',~,and there is considerable evidence15-'* to support the existence of a cubic phase at high temperatures over the whole composition range.The high-temperature cubic phase of Sr,Fe205 is belie~ed",'~ to be a microdomain-textured brownmillerite, which cannot be preserved by quenching. More recently', a microdomain-textured sample of SrCro.lFeo,902.5has been prepared in uucuu by cooling it rapidly. A similar preparation in flowing argon of composition SrCro.1Feo.902,578was found to have the X-ray pattern of a brownmillfrite-related pha!e. The b axis had increased from 15.509(5) A to 15.652(5)A with the introduction of extra oxygen into the tetrahedral layers. This is believed to be the first example of the introduction of a significant excess of oxygen into the brownmillerite structure without forming a second phase or a microdomain-textured intergrowth.Although both CaFeO, and Ca,Fe205 are known, there has been no report of any intermediate phases. Ca21?e205 does not oxidise in air like Sr2Fe205. This paper reports the preparation of a partially oxidized derivative of Ca2Fe205 under high pressures of oxygen, and its characterization by Mossbauer spectroscopy. Experimenta1 Calcium peroxide was obtained in hydrated form by precipi- tation from aqueous solutions, as described20 in the literdture. It was dehydrated over desiccants and by heating towards 200 "C. The material used for reactions contained a somcwhat lower proportion (35.0%)of peroxide oxygen than the formula requirement (44.4%) for CaO,, although this was essentially in agreement with the original reference.Ca2Fe205was prepared by the standard ceramic technique. A mixture of calcium peroxide and iron oxide powders in the appropriate cation ratio was ball-milled and fired at 1200 "C in air, followed by cooling in the furnace. The integrity of the material was verified by X-ray powder diffraction and M ossbauer spectroscopy. Reactions were carried out in small gold tubes (diameter 4mm, length 20m, capacity 0.25 cm3) with the calcium peroxide packed at one end, and separated from the other reactant by a loosely fitting partition of gold foil. The central region of the pressure apparatus was an assembly of concentric steel rings having an overall diameter of 250 mm, height 65 mm, with a central bore of diameter 31 mm.The reaction assembly consisted of a pyrophyllite cylinder (diameter 31 mm, height 36 mm) with the gold sample tube in the centre, and surrounded by a ring of six carbon rods of 3 mm diameter as heating elements. The heating current (up to 380 A at 5.5 V) gave a power input of up to 2100 W. Reaction temperatures were estimated from the relationship, determined in separate experiments, between power input and central temperature. In high-pressure preparative runs, the sample assembly was compressed to a pressure of ca. 1.0 GPa (10 kbar), and the temperature raised by stages over a period of about 30min, and then maintained at a steady temperature of towards 800-850 "C for 40-60 min. The assembly was cooled under pressure.X-Ray powder diffraction data were recorded with a Philips diffractometer using nickel-filtered Cu-Ka radiation. Mossbauer data were collected in the range 4.2<T/K<290 using a 57Co/Rh source matrix held at room temperature; isomer shifts were determined relative to the spectrum of metallic iron. Results and Discussion A sample of Ca2Fe,05 was prepared by standard ceramic techniques, and its composition verified by X-ray powder diffraction and Mossbauer spectroscopy. The Mossbauer spec- trum at 290 K is shown in Fig. l(c) and is in full agreement with previous data. The Fe3+ cations in equal numbers of octahedral and tetrahedral sites generate two magnetic hyper- fine splittings, which are well resolved. This material was placed into a gold tube with calcium peroxide as a source of oxygen, and then compressed to a pressure of ca.1.0 GPa, before it was heated to 800°C to decompose the peroxide. After cooling and pressure release, the sample was removed and examined by X-ray powder diffraction and Mossbauer spectroscopy. In anticipation of the discussion which follows, this material will be referred to as Ca2Fe205+y. The X-ray powder pattern is very different from that of the .. .. .....-. *-I..-. (c)i 1 ;< -12 -8 -4 0 4 8 12 vetocity/mm s-' Fig. 1 57Fe Mossbauer spectra at 290 K of (a)Ca,Fe,O,.,, obtained under high pressure of oxygen at 8OO0C,(b)a repeated preparation with a residue of unreacted starting material and (c) the starting material Ca,Fe,O, J.MATER. CHEM., 1994, VOL. 4 starting material, and in particular the characteristic 020 reflection of the large orthorhombic cell of Ca2FezO5 is absent, confirming that a complete transformation had taken place. A repeat preparation under nominally identical conditions yielded a sample in which both the X-ray pattern and the Mossbauer spectrum clearly showed the presence of some residual Ca2Fe2O5. However, the reflections listed in Table 1 were still clearly present as the major component in the X-ray pattern. The high-angle reflections are broadened and gener- ally weak, perhaps due to poor crystallinity, and it is not possible to give a unique interpretation. However, the Mossbauer data described below establish that the material is almost certainly single-phase and structurally related to the original lattice but with the insertion of extra oxygen atoms.It is not unreasonable to suggest that this insertion creates a significant degree of structural disorder, which may include for example disordered layer stacking or microdomain tex- tures, and as a result produces a loss of detail in the X-ray powder pattern. There is no evidence for a supercell in the y direction, and this may be an indication of disorder in the layer stacking. However, the albeit limited evidence is not inconsistent with the existence of a perovskite-related material with structural disorder. The Mossbauer spectrum at 290 K is shown in Fig. l(a), and the spectrum of the repeat preparation containing some Ca2Fe205 is shown in Fig.l(b). The magnetic pattern of the starting material is clearly visible in Fig. l(b), but is absent in (a), confirming that all the material in the reaction tube has been transformed. There are at least two large hyperfine fields visible in spectrum (a), and these are consistent with a substantial proportion of Fe3+ in the material. The degree of oxidation achieved is clearly not large. However, an under- standing of the structural changes involved could only be obtained by examining the spectrum as a function of tempera-ture. Five spectra at different temperatures are shown in Fig. 2. The spectrum at 4.2K in Fig. 2(u) provides the key to understanding the problem, because at low temperatures any dynamic processes are frozen out and the individual hyperfine fields are close to the saturation limit.The resolution of three clear lines at the extreme left of the spectrum and a weak component at just below -3 mm s-' establishes that there are at least four distinct magnetic sites. These are indicated as A, B, C and D in the computed fit which is shown in Fig. 3, and the relevant parameters are given in Table 2. For compari- son, the figures for Ca,Fe205 are also given. Various con- straints on parameters such as the linewidths have very little effect on the results. The isomer shifts and magnetic flux densities for components C and D are typical figures for Fe3+ in octahedral coordination to oxygen. In particular, the large negative quadrupole perturbation E for component D estab-lishes that this component comes from octahedral Fe3 sites+ very similar to those in Ca2Fe20S with the principal axis of the electric field gradient near to b and normal to the spin axis in the uc plane (from the near similarity of the published spectra for Ca2Fe,05 and Ca2A1Fe05 it is evident that the Table 1 X-Ray data for Ca,Fe,O,.,, intensity dobslA intensity 3.858 12 1.929 26 3.73 1 6 1.866 12 2.763 24 1.745 2 2.681 100 1.681 2 2.637 3 1.583 10 2.543 4 1.554 8 2.458 3 1.535 5 2.206 3 1.342 11 2.181 3 J. MATER.CHEM., 1994, VOL. 4 t .-0 u? .-E E c -12 -8 -4 0 4 8 12 velocity/mm s-l Fig. 2 57Fe Mossbauer spectra of Ca,Fe,O,.,, at (a)4.2, (b)78, (c) 150, (d) 200 and (e) 290 K C .-0 u?u?.-EC !!"I I I I I I I I I I C D -1 2 -8 -4 0 4 a 12 velocity/mm s-' for example21,23-25 in the related Sr,Fe20, and in CaFeO, and other perovskite-derived phases.The charge separation to Fe3+ and Fe5+ is not necessarily complete, and the reason for this unusual behaviour remains obscure. From the rather limited data available, the flux densities from Fe4+ and Fe5+ are generally in the same range, and a more significant difference appears to lie in the lower isomer shift for Fe5+. Moreover, the magnetic flux density observed for a high-spin d4 configuration will include substantial orbital and dipolar terms, so that the net value will be more variable than is the case for the high-spin d3 and d5 configurations of FP and Fe3+.In the present instance, the combined area of compo- nents A and B is some 51%, which would correspond to the tetrahedral layers (50%)with an injection of extra oxygen to give Fe4+ ions. However, the relative intensities of A and B are poorly defined because of the considerable overlap. Nevertheless, from a close examination of the spectra for Ca,Fe,05 +y we believe there is no evidence to suggest that a charge disproportionation is taking place, and it is perhaps unlikely that this could take place at such low concentrations of Fe4+ cations: taking into account the chemical isomer shift value, component A is therefore assigned to Fe4+ ions.There is no evidence for any additional absorption in the central region of the spectrum, i.e. there is no significant contami- nation by paramagnetic compounds or relaxational collapse within the spectrum at 4.2 K. The evidence of the Mossbauer spectrum at 4.2 K thus leads to an approximate Fe4+ concen- tration of 11-13Oh and a composition of about Ca,Fe,05~,,. Although the true figure could in principle be determined independently by wet chemical analysis, this has not been attempted as the very small quantity remaining was judged to be insufficient for a reliable estimation, and the preparation is not entirely reproducible. The spectrum at 78 K is shown in Fig. 2(b). Component A has weakened significantly, but there is an increase in the central absorption region corresponding to some 14% of the total area which cannot be accounted for by the inner lines of the main magnetic components.It appears that primarily Fig. 3 57Fe Mossbauer spectrum of Ca2Fe,05,,, at 4.2 K showing an analysis in terms of four hyperfine fields Mossbauer spectrum is not sensitive to the direction of the spin within that plane). Component C has a similar isomer shift and flux density, but a very different value of E, and therefore can be assigned to an octahedral site which has been perturbed by a change in its immediate environment. Component B shows parameters which are very similar to the tetrahedral Fe3+ sites in Ca,Fe,O,, and can also be assigned to tetrahedral sites. Component A shows a very low flux density (23.88 T) and a low isomer shift (0.057mm s-'), which are similar to the values observed recently21722 for Fe4+ sites in Sr,Fe04 (in particular the d site listed in Table 2). Although this compound has an undistorted K2NiF4 structure at 4.2 K, the Mossbauer spectrum shows four different magnetic iron sites, suggesting a complex spin arrangement which has not been determined as yet.Although Fe4+ is the obvious assignment for component A, it is also necessary to consider the possibility that this is the Fe5 component of a charge-disproportionation as shown + the Fe4+ spins are involved in a relaxational collapse. The octahedral D sites have now separated into two groups which will be referred to as D1 and D2 in order of increasing field, while the emergence of shoulders show that the tetrahedral sites have separated into two groups B1 and B2, again in order of increasing field.This is most evident in the group of lines around -8 mm s-'. It is difficult to refine all the parameters involved to high accuracy, but typical figures obtained from a fit in which parameters such as the isomer shift and linewidths were constrained in groups are included in Table 2 as an illustration. With further increase in temperature to 150 K and with reference to the group of lines at -8 mm s-', the C component begins to move under B2, and B1 becomes more obvious. At 200K, C and B2 have effectively merged, and there is a noticeable collapse into the centre of the spectrum in which the B1 component is believed to be involved, and this trend is continuing at 290K.The linewidths are somewhat broad even at 4.2 K, but increase in temperature accentuates this for reasons which will now be discussed. In an ideal antiferromagnet where all cation sites are identical the magnetic hyperfine splitting, which approaches the saturation limit at 4.2 K, decreases with increase in temperature but remains sharp until the flux density falls to zero at the NCel temperature. On the other hand, in a substituted solid solution where there is a wide variation in the local site environment there is usually a significant degree of additional line broadening even at the saturation limit due to small variations in the flux density and the quadrupole perturbation.The initial temperature dependence of the spec- J. MATER. CHEM., 1994, VOL. 4 Table 2 Mossbauer parameters for Ca,Fe,O,.,, compound site 6/mm s-' E/mm s-' r/mm s-' BIT fraction 4.2 KCa2Fe205.12 A B C 0.057(7) 0.329( 2) 0.503( 3) 0.229( 7) 0.264( 2) 0.093(4) 0.542(5) 0.542( 5) 0.542( 5) 23.88( 5) 46.72( 2) 51.20( 3) 0.11 0.40 0.29 78 K D A 0.507( 5) 0.195 -0.307( 6) collapsed 0.542( 5) 53.30( 5) 0.20 0.14 B1 0.309 0.235 0.459 43.6 0.12 B2 0.309 0.300 0.459 45.6 0.29 C 0.485 0.124 0.459 49.3 0.22 D1 0.485 -0.169 0.459 50.7 0.09 D2 0.485 -0.275 0.459 52.7 0.14 Ca,Fe,05 4.2 K oct tet 0.46 0.29 -0.28 0.35 54.7 42.2 0.5 0.5 Sr,FeO, 4.2 K a b C 0.07 0.07 0.07 -0.33 -0.20 0.06 32.9 30.3 27.2 0.25 0.25 0.25 d 0.07 0.36 24.6 0.25 Sr,Fe,O,4.2 K a b 0.32 -0.06 -0.07 - 41.7 28.3 0.50 0.50 CaFeO, a b 0.00 0.34 -- 27.9 41.6 0.50 0.50 trum can be fully explained24 using molecular field theory.The flux density at a given site shows a temperature depen- dence which deviates substantially from the Brillouin curve and is a function of the near-neighbour environment; in general a substituted environment leads to a decrease in the flux density. In addition there may be differences in the supertransferred hyperfine field produced by the near-neighbour spins which will alter the value of the saturation field. The overall result is a broadening of the spectrum in a solid solution at low temperatures, which becomes more pronounced with rise in temperature and may lead to the resolution of individual near-neighbour components in the middle of the temperature range.As the ordering temperature is approached, the molecular field approximation is less appropriate, and it is common to find some degree of relax- ational collapse as spins which are less strongly coupled to the ensemble begin to relax on a timescale faster than the observation time of the Mossbauer observation. All these features are evident in the work from this laboratory13 on the compound SrCro.1Feo,902,5, where substitution of chromium onto octahedral sites results in the spectrum showing compo- nents from three distinct octahedral and two tetrahedral iron sites at 290 K.In the particular case of Ca2Fe205,,2 a similar situation will exist but with the additional complication that the substitution of Fe4+ also involves the insertion of extra oxygen into the tetrahedral layers. A schematic diagram of a part of a tetrahedral layer with an additional oxygen anion is shown in Fig. 4. It can be assumed that this oxygen will form a defect cluster with two Fe4+ cations in approximately square-pyramidal coordination (there is evidence to suggest14 that this coordination occurs in STF~O~.~~). The adjacent layers above and below that shown in Fig. 4 contain octahedra which share corners with the tetrahedra. Using this model, one can develop an interpret- ation for the multiple site environments which are produced.At low concentrations, each Fe4+ cation will be a nearest neighbour to two Fe3+ cations in octahedral sites, and two Fe3+ cations in tetrahedral sites. On the other hand, an octahedral site will have 0, 1 or 2 (in different layers) nearest- neighbour Fe4+ cations, and a tetrahedral site will also have 0, 1 or 2 neighbours. Assuming that component D at 4.2 K, which closely resembles the site in the parent Ca2Fe,0,, is produced by Fe3+ sites with no Fe4+ neighbours, then a binomial site distribution would give a fraction (of the total iron) of 0.20 at a value of y=O.18, which is somewhat higher n n W WT oxygen iron0 Fig. 4 Schematic representation of a tetrahedral layer in the Ca,Fe,O, structure (distortions are ignored) showing the effect of inserting an additional oxygen anion.The a and c axes of the original structure are shown. It seems likely that each oxygen will form a defect cluster with two Fe4+ cations in square-pyramidal coordination. than the estimate above. On the other hand, if the tetrahedral sites with no Fe4+ neighbours correspond to the B2 compo-nent at 78 K, then this fraction of 0.29 would give a value for y of about 0.12, which agrees with the fraction of Fe4+ observed directly. The relative intensities of the three different octahedral sites observed at 78 K do not appear to correlate with this simple model (with random substitution and taking into account only the cation charges). An adaptation of this model would be to assume that it is the inserted oxygen anion which perturbs the lattice by displacing its immediate pair of neighbours and thereby influencing the distortion and tilt of some four octahedra in each of the planes above and below.This would effectively double the number of octahedra affected, and could explain the discrepancy noted above. A third model would be to assume that only some of the tetrahedral layers contain inserted oxygen. This would explain the apparent loss of the supercell along y, but would not explain the existence of B1 and B2 sites so conveniently. If J. MATER. CHEM., 1994, VOL. 4 the D component comes from octahedral layers which only have normal tetrahedral layers on either side, then once again a value of about y=0.18 can be deduced.All these simple models fail to give a clear explanation for the relative pro- portions of the three kinds of octahedral site. The parent Ca2Fe205 structure is antiferromagnetic with antiparallel coupling of the Fe3+ spins in each Fe-0-Fe linkage to give a G-type structure with all the spins aligned along a.The insertion of extra oxygen and some Fe4+ cations will perturb this situation. The magnetic coupling of Fe4+ in perovskites has already been discussed14 in connection with SrFe02,75. All the Fe -0-Fe couplings will remain strongly antiferromagnetic. If the Fe4 cation is in square-pyramidal + coordination, it can be seen from Fig. 4 that there will be a strong ligand field acting along the axis of the pyramid at an angle of some 45" to a.Thus of the two d-electron orbitals which can give a o-superexchange overlap only that which lies along the axis of the pyramid is likely to be occupied. The result is that the o-superexchange between Fe3+ and Fe4+ in adjacent layers by overlap of a half-filled and an empty orbital will be ferromagnetic, which is in conflict with the weaker antiferromagnetic n-transfer. Furthermore, the extreme anisotropy of the square-pyramidal geometry will try to impose a spin alignment in conflict with the superexchange of the bulk matrix which causes alignment along a. As a result, the Fe4+ cations will be only weakly coupled to the spin system, and therefore it is easy to understand why they should decouple from the Fe3+ spins so readily.The flux density of the D2 sites at 290 K is some 44.7 T, which with the value at 4.2 K and the S=5/2 Brillouin function leads to an estimate of the ordering temperature of ca. 500 K, com-pared to 725 K in Ca,Fe,O,, and this is in accord with the reduction in the number of Fe3+ -0-Fe3 superexchange interactions. Therefore it can be seen that magnetic relaxation is taking place substantially below the critical temperature as might be expected in a disordered material with conflicting exchange interactions. In conclusion, the preparation of Ca,Fe,O,.,, is the first reported material between stoichiometric CaFeO, and Ca,Fe,O,. The Mossbauer spectra have established the struc- tural relationship to the parent Ca2Fe,05 structure, and are fully consistent with a single-phase material with an ordering temperature of ca.500K. The additional oxygen enters the tetrahedral-site layers. The considerable disorder in the mate- rial results in at least six distinct lattice sites for iron of which five contain Fe3+ cations in both octahedral (3) and tetra- hedral (2) coordination, and the sixth contains Fe4+ cations. The latter are only weakly coupled to the lattice, and there is an increasing degree of magnetic relaxation in the Mbssbauer spectrum with increasing temperature. Acknowledgement is made to the SERC for a Fellowship (to N.P.), for a studentship (to A.J.H.) and for financial support. References 1 E. F. Bertaut, P. Blum and A.Sagnieres, Acta Crystallogr., 1959, 12, 149. 2 A. A. Colville,Acta Crystallogr., Sect. B, 1970,26, 1469. 3 J. Berggren, Acta Chem. Scand., 1971,25, 3616. 4 U. Gonser, R. W. Grant, H. Wiedersich and S. Geller, Ai'pl.Phys. Lett., 1966, 9, 18. 5 L. M. Corliss, J. M. Hastings, W. Kunnmann and E. Banks, Acta Crystallogr., Sect. A, Suppl., 1966,21,95. 6 Z. Friedman, H. Shakad, and S. Shtrikman, Phys. Left., 1967, 25A, 9. 7 T. Takeda, Y. Yamaguchi, S. Tomiyoshi, M. Fukase, M. Sugimotoand H. Watanabe, J. Phys. SOC.Jpn., 1968,24,446. 8 A. A. Colville and S. Geller, Acta Crystallogr., Sect. B, 1971, 27,2311. 9 R. W. Grant, S. Geller, H. Wiedersich, U. GonGer and L. D. Fullmer, J. Appl. Phys., 1968,39, 1122. 10 C. Greaves, A. J. Jacobson, B. C.Tofield and B.E. F. Fender, Acta Crystallogr., Sect. B, 1975,31, 641. 11 M. Harder and H. Muller-Buschbaum, Z. Anorg. Allg. Chem., 1980,464,169. 12 P. D. Battle, T. C. Gibb and P. Lightfoot, J. Solid Statc! Chem., 1988, 76, 334. 13 T. C. Gibb, J. Muter. Chem., 1991,1,23. 14 T. C. Gibb, J. Muter. Chem., submitted for publication. 15 T. C. Gibb, J. Chem. SOC.,Dalton Trans., 1985, 1455. 16 Y. Takeda, K. Kanno, T. Takada, 0. Yamamoto, M. Takano, N. Nakayama and Y. Bando, J. Solid State Chem., 1986,63,237. 17 M. Takano, T. Okita, N. Nakayama, Y. Bando, Y. Takeda, 0.Yamamoto and J. B. Goodenough, J. Solid State Chein., 1988, 73, 140. 18 L. Fournes, Y. Potin, J. C. Grenier, G. Demazeau and M. Pouchard, Solid State Commun., 1987,62,239. 19 J. C. Grenier, N. Ea, M. Pouchard and P. Hagenmuller, J. Solid State Chem., 1985,58,243. 20 E. H. Riesenfeld and W. Nottebohm, 2.Anorg. Allg. Chern., 1914, 89,405. 21 S. E. Dann, M. T. Weller, D. B. Currie, M. F. Thomas and A. D. Al-Rawwas,J. Muter. Chem., 1993,3, 1231. 22 P. Adler, J. Solid State Chem., 1994,108,275. 23 Y. Takeda, S. Naka, M. Takano, T. Shinjo, T. Takada and M. Shimada, Mater. Res. Bull., 1978, 13, 61. 24 P. D. Battle, T. C. Gibb and S. Nixon, J. Solid State Chela., 1988, 77, 124. 25 T. C. Gibb and M. Matsuo, J. Solid State Chern., 1989,81. 83. 26 J. M. D. Coey and G. A. Sawatzky, Phys. Stat. Solidi R, 1971, 44, 673. Paper 4/02605E; Receiued 3rd Ma v, 1994
ISSN:0959-9428
DOI:10.1039/JM9940401451
出版商:RSC
年代:1994
数据来源: RSC
|
20. |
YMoO4revisited: the crystal structure of Y5Mo4O18 |
|
Journal of Materials Chemistry,
Volume 4,
Issue 9,
1994,
Page 1457-1461
N. J. Stedman,
Preview
|
PDF (724KB)
|
|
摘要:
J. MATER. CHEM., 1994, 4(9), 1457-1461 YMoO, Revisited: The Crystal Structure of Y,Mo,O,~ N. J. Stedman,"A. K. Cheethamb and P. D. Battle*" a lnorganic Chemistry Laboratory, South Parks Road, Oxford, UK OX7 30R Materials Department, University of California, Santa Barbara, CA 93706, USA Analytical electron microscopy and X-ray powder diffraction have been used to show that the 'compound' YMoO, is a mixture of two oxides having the approximate compositions Y5M04018 and Y2M030,0. Furthermore, thc phase previously described as a high-temperature form of YMoO, is shown to be a silicate, the silicon being leached out of the silica tubes used as reaction vessels during the synthesis of the compound. The crystal strutture of the yttrium- rich phase has been refined from neutron powder diffraction data: space group /4,/a, a =5.1 76 67(2) A, c =11.004 51 (7) A.The structure is that of a defect scheelite with extensive disorder on both the molybdenum and oxygen subiattices. The stoichiometry refined to the formula Y5M03.&17.3. The original aim of this work was to map the Y-Mo-0 phase system with a view to finding new yttrium molybdenum oxides in which the formal oxidation state of molybdenum was less than six. Only three such phases were known prior to this study, namely Y5M02012, YMoO, and Y2M0207.The first was mistakenly formulated as Y2Mo05for a number of years'-6 until single crystals were obtained by molten-salt electroly~is,~and the structure, which contains pairs of metal- metal bonded molybdenum atoms, was solved.* YMoO, has been reported to give a powder X-ray diffraction (XRD) pattern characteristic of a ~cheelite,~ and Y,Mo,O,, which is only formed at temperatures above ca.1400"C, adopts the pyrochlore structure and shows spin-glass behaviour." Two more reduced Y-Mo oxides, Y4Mo18032and Y4M0401111'12 were reported during the course of this work, but the number of examples known remains small in comparison with the La-Mo-0 system, which contains a large number of com- pounds in which the formal oxidation state of molybdenum is less than six and many cluster compounds in which it is lower than four. These include LaMo50,,l3 which contains a novel Mo,, cluster formed from two Mo, octahedra sharing a common edge, and LaMo,.,0,,,14 which contains capped Mo octahedra; it was hoped that a thorough investigation of the Y-Mo-0 system might reveal a comparable wealth of cluster chemistry. We undertook such a study, intending to use YMoO, as a standard composition in analytical electron microscopy experiments designed to identify new phases.However, it soon became clear that a reinvestigation of this compound itself was necessary, and the ensuing work is described below. Experimental All the syntheses described below were carried out by standard solid-state techniques using evacuated, sealed silica tubes to contain the reaction mixture. In each case, the appropriate amounts of Y203(Aldrich, 99.99%, dried before use), MOO, (Aldrich, 99.5%) and Mo metal (BDH, oxygen content estab- lished by thermogravimetric analysis as MOO^,^,) were mixed well, ground in an agate mortar and pressed into a pellet.Each pellet was sealed in an evacuated silica capsule, and heated at 1100"C for a suitable length of time. More specific details are given below when appropriate. The products of each reaction were established by powder XRD and analytical electron microscopy. The latter was carried out in a JEOL 2000FX microscope operating in transmission mode with a Tracor Northern TN5500 energy-dispersive X-ray detector. Thin crystallites (partly transparent to the electron beam) were analysed as the effects of fluorescence and absorption are negligible in this case.15 A large sample of Y5Mo4018 was prepared for powder neutron diffraction by mixing the products from four separate 2 g syntheses, each of which had been carried out in it sealed silica tube with the pellets wrapped in Pt foil (see below).Powder neutron diffraction data were collected at room temperature on the combined sample using the instrument POLARIS at the Rutherford Appleton L!boratory. Data in the range 4800-19 100 ps (d=0.778-3.096 A) from the highest resolution (backscattering) bank of detectors were focused, normalised and corrected for attenuation prior to Rietveld analysis, which was carried out using the Rut herford Laboratory in-house software.', Results and Discussion Constitution of 'YMoO,' YMoO, was originally preparedg by heating stoichiometric amounts of Y203, MOO, and molybdenum metal in an evacuated, sealed silica capsule at 1200°C for 1-2 h.The X-ray powder diffraction pattern of the product obtFined was indexed 0; a tetragonal unit cell with a =5.170(1)A and c = ll.OOO(4) A, isomorphous with that of the mineral scheelite, CaWO,. We prepared a sample of YMoO, using the same method and the diffraction pattern of our product was clearly characteristic of a scheelite, although the peaks in the pattern were quite broad. Close examination of the data revcaled a few weak, broad peaks that could not be indexed on a scheelite cell. These were found to correspond to the strongest peaks of the monoclinic, high-temperature form of YMoO,, pre-viously reported by M~Carroll.'~ The presence of small amounts of a second form of the same composition in the sample was not expected to affect the performance of the sample as a standard in electron microscopy.However, when the sample was analysed in the microscope the results obtained were inconsistent with the presence of just one comptosition. Even after repeating the synthesis, but heating the reaction mixture for longer times (typically for 3 days at 1100 'C), the material still did not appear to be monophasic when studied in the electron microscope. For this reason, a sample of Y5M02012,prepared by molten-salt electroly~is,~ was chosen as the standard for further analytical microscopy. The analytical electron microscopy results for 'Y MOO,' were then reviewed, using the new standard.The analyses were found to fall into two distinct groups, as shown in Fig. 1, corresponding to phases containing yttrium and molybdenum in the approximate ratios 2 :3 and 5 :4. No particles containing 1458 5-4->0 yttrium: molybdenum ratio Fig. 1 Analytical electron microscopy results for YMo04 yttrium and molybdenum in a ratio of 1:l were observed. Two possible explanations for these observations were evident: either one composition adopted a scheelite structure, and the other a structure that had been mistakenly attributed to a high-temperature form of ‘YMoO,‘, or one composition adopted a scheelite-related structure and the second was amorphous to X-rays. The broad peaks in the powder diffrac- tion patterns of ‘YMoO,‘ that could not be indexed on an ideal scheelite cell appeared to be too weak to correspond to a phase which comprised a large proportion of the sample, so the second explanation appeared to be the more likely.Furthermore, it has been noted by Gopalakrishnan and Manthiram that hydrogen reduction of the normal molybdates Ln,Mo,O,, to the oxygen-deficient phases Ln2M03012-leads to X-ray amorphous products for the heavier lanthanides and yttrium,18 suggesting that the phase with the yttrium : molybdenum ratio of 2 :3 in the mixture ‘YMoO,’ could also be amorphous. The composition ‘YMoO,‘ was reheated in an attempt to bring more long-range order to any possible amorphous component, but no change in the powder pattern was observed.In order to progress beyond this point it was necessary to determine the oxygen content of the two phases, Y,Mo,O, and Y&fO,O,,, which appear to make up our sample of overall stoichiometry YMoO,. The experimental procedure described above was used to prepare numerous samples having Y: Mo ratios of 2 :3 and 5 :4,each sample having a different oxygen content. Our choice of the most appropriate samples to study was guided by the results of our wider investigation into the Y-Mo-0 system, which is described elsewhere.” The cleanest powder X-ray diffraction pattern for the series ‘Y&fO@,’ was obtained from the composition Y,Mo,O,,, although the pattern still contained the weak, broad peaks that had been assigned to a high-temperature form of ‘YMoO,’ by McCarroll.Compositions containing less oxygen than Y5Mo,018 gave powder X-ray diffraction pat- terns which showed peaks due to Y~Mo~O,, and MOO, in addition to the peaks characteristic of a scheelite-like phase. Analytical electron microscopy of these relatively reduced mixtures showed the presence of particles containing molyb- denum only (MOO,), and yttrium and molybdenum in the ratios 5 :2 (Y5M02012) and 5 :4 (the scheelite-like phase). For the stoichiometry of the system to balance, the second phase present in the mixture ‘YMoO,‘ along with the scheelite Y5M0401, was required to have the composition Y,Mo,Olo. Several attempts at the synthesis of samples of Y2M03010 and other members of the series Y2M030x were carried out.None gave X-ray amorphous products. Each powder pattern could be assigned as a mixture of fully oxidised Y2Mo3Ol2, J. MATER. CHEM., 1994, VOL. 4 MOO, and a scheelite-like phase initially thought to be Y5M04018, together with the broad peaks supposedly charac- teristic of monoclinic YMoO,. The presence of four crystalline phases (and possibly one amorphous phase). more than the three permitted by the phase rule, suggested that the system was not at equilibrium, but reheating the samples did not simplify the powder patterns. In order to gain further insight into the nature of the phases present in these mixtures, a sample of overall composition ‘Y2Mo,01< was studied in the electron microscope. Initially it appeared that only three different compositions were present in the sample: a phase containing molybdenum only (MOO,), a phase containing yttrium and molybdenum in the ratio 5:4 and a phase containing yttrium and molybdenum in the ratio 2:3.However, close examination of the X-ray emission spectra for crystallites from this sample revealed a rather subtle detail; some of the spectra of particles containing both yttrium and molybdenum featured a shoulder on the low-energy side of the Y-La peak, and some did not (Fig. 2). The position of this shoulder corresponded exactly to that expected for the silicon Ka emission line. Therefore the system did not contain three components, but four, the silicon having been leached out of the silica tubes used for the reactions.That this had been overlooked in previous experiments was not surprising, as the proximity of the Si-Ka and Y-La peaks had led to the two being treated as just one peak (the Y-La peak) by the automatic spectra analysis software of the Tracor system. Several particles containing silicon (as inferred from the presence of a shoulder on the Y-La peak) were then analysed for yttrium and molybdenum in the usual way, revealing an Y: Mo ratio of ca. 4:3. It thus appears that this new yttrium molybdenum silicon oxide is not related to the lanthanum molybdenum silicon oxide La3M04SiO14 .20 x Mo-La energylev Fig. 2 X-Ray emission spectra for two different particles within a sample of overall composition Y,Mo,O,,. Only spectrum (b) shows an Si-Ka shoulder on the Y-La peak.J. MATER. CHEM., 1994, VOL. 4 The incorporation of silicon into the products of the reactions carried out in this region of the phase diagram had clearly led to the formation of products which did not represent true equilibrium in the system. Each reaction was therefore repeated, wrapping each pellet of reagents in plati- num foil before sealing the pellet in an evacuated silica capsule. The synthesis of YsM04018 by this method resulted in a powder XRD pattern free from the peaks which had been assigned to a high-temperature, monoclinic form of ‘YMoO,’, which now appear more likely to be due to the new yttrium molybdepum silicon oxide. Apart from one very weak peak at 3.51 A, possibly the strongest peak of Y2M03012, the pattern was clean.The analysis of particles of this product in the electron microscope confirmed that this material was now virtually monophasic (Fig. 3). The synthesis of ‘YMoO,‘ was also repeated using platinum foil to prevent the incorporation of silica in the products. The powder XRD pattern of the product could be indexed cleanly on a scheelite-like tetragonal unit cell, like that of Y5Mo,018, but analytical electron microscopy again confirmed the presence of two phases (Y :Mo 5 :4 and 2 :3) in the products. A number of compositions Y2Mo,0, were also prepared again, using platinum foil to prevent reaction with the silica. These reactions were not all successful and in many cases it was obvious from the range of products that equilibrium had not been attained, leaving this area of the phase diagram as yet undetermined. However, interesting results were obtained for the composition ‘Y2M03010.5’which suggest that the second phase present in the mixture ‘YMoO,‘ may not be amorphous.‘Y2M~3010.5’appeared to contain Y2Mo,012, MOO, and a great deal of a scheelite-like phase by powder XRD, but in the electron microscope no particles containing yttrium and molybdenum in the ratio 5 :4could be found. All the particles analysed contained yttrium and molybdenum in the ratio 2: 3 or molybdenum only. While this result is not conclusive, it implies that the phase Y2Mo3OlOis not amorph- ous to X-rays, but that, like Y5Mo,018, it too adopts the scheelite structure.The unit-cell parameters of the scheelite present in the powder pattern of the mixture ‘Y2M03010.5’ and those of the scheelite Y5Mo4018 were compared, and were found to differ by a small but significant amount; for the schee1i:e present in ‘Y2M03010~S’a=5.1683(4) A and cz 10.993(2) A while for !he scheelite Y5Mo4OI8a= 5.176 67(2) A and c =11.00451(7) A (determined in the neutron diffraction experiments described below). The difference in these param- eters is small enough that the rather broad Bragg peaks in the X-ray powder diffraction pattern of the mixture ‘YMoO,’ could easily be the result of a superposition of the patterns of 5-4-x $3-0-Lc 2-1 n., I 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5 yttrium: molybdenum ratio Fig.3 Analytical electron microscopy results for Y5Mo,OI8 prepared in platinum foil 1459 the two phases. The positions of the peaks in ‘YMoO,’ were also used to dttermine a ‘unitecell’, the parameters of which [a=5.1698(9) A, c= 10.997(3) A] were found to be intermedi- ate between those values found for the scheelites present in the other two compositions. Thus, although the details of the phase diagram around ‘YMoO,‘ are still unclear, it is possible to account for the data presented above by assuniing that ‘YMoO,’ is actually a binary mixture of the moljbdenum- deficient scheelite Y5M04018 and the yttrium-deficient scheel- ite Y2Mo3OlO. Structure of Y,Mo,O,, The work described above had led to the isolation of the molybdenum- and oxygen-deficient scheelite believed to be Y5M040,8, in which, according to the formula, both the molybdenum and the oxygen sites within the structure are only partially occupied.Rietveld refinement of this model against powder diffraction data was then attempted. Preliminary calculations showed that powder XRD would be remarkably insensitive to any deviation in the true stoichi- ometry of the phase from ‘YMoO,‘, whereas neutron diffrac- tion was predicted to be more sensitive. The background level of the neutron powder diffraction pattern was uneven and it was therefore fitted by linear interpolation between fixed points selected by eye. Bragg peaks due to the Y2Mo3OI2impurity, although few in number, were more readily apparent than they had been in our previous X-ray studies; they were excluded from the refine- ment.The complex peak shape was modelled using a Voigt function convoluted with two exponential functions. Terms describing the Lorentzian and Gaussian contributions to the peak shape were refined, but the sample-independent param- eters were fixed at the values obtained from Rietveld analysis of data collected on a silicon standard. The starting model comprised molybdenum on the 4(a) site and yttrium on the 4(b) site within the scheelite structure (space group 14,/a, 21) together with an oxygen site, the initial coordinates of which were taken as those for the oxygen site in the structure of NaY (MOO,), described elsewhere.22 Each site was at first assumed to be fully occupied, corresponding to a forornula ‘YMoO,‘, and isotropic temperature factgrs of B =0.5 A2 for yttrium and molybdenum, and B == 0.8 A2 for oxygen were assumed.In all cycles of the refinement the unit- cell dimensions, the scale factor and the peak-shape param- eters were varied. Following refinement of the coordinates of the oxygen atom, the occupancies of the oxygen md molyb- denum sites were allowed to vary while keeping the tempera- ture factors fixed at their starting values. The occupancy of the molybdenum site decreased to 0.69, and that of the oxygen site decreased to 0.76, confirming that the phase was deficient in both molybdenum and oxygen with respect ta ‘YMoO,’. At this stage the model corresponded to a simple scheelite in which the molybdenum site (partially occupied) was coordi- nated by four oxygen sites (also partially occupied) with an approximately regul$r tetrahedral geometry, with Mo-0 bond lengths of 1.88 A.An isotropic temperature factor for each atom was then refined, holding the occupancy of the molybdenum and yttrium sites fixed at the refined values. Although the overall agreement between the observed and calculated diffraction profiles improved dramatically at this stage, the fit was still unsatisfactory and, furthermore, the temperature Factors became uvreasonably large (BY=1.52 A2, BMo= 1.20 A2 and Bo= 1.81 A2). The temperature factors were reset to reason- able values and a difference Fourier map was calculated. In this way, a new partially occupied site of scattering density, 0(2), was located close to the original oxygen site, and its J.MATER. CHEM., 1994, VOL. 4 inclusion in the model caused the profile fit to improve, although not to a satisfactory level. Two further sites of scattering density were located in subsequent difference Fourier maps; Mo(2), close to the molybdenum 4(a) site, and 0(3), close to the original oxygen site. Inclusion of these sites, partially occupied, in the model greatly improved the profile fit. No further scattering density was located. In reaching this stage, the occupancy of each of the new sites had been allowed to vary without any constraint on the overall stoichiometry of the phase. In subsequent refinements the overall Y: Mo ratio was constrained to the most accurate value (1:0.77) available from analytical electron microscopy.The use of this value, rather than the ideal 1:0.8, admitted the presence of the Mo-rich impurity Y,Mo,OI2. Further refinement of the model had to be carried out cautiously as the level of correlation between certain variables was very high. In particular, neither the positions of all of the three oxygen sites nor the occupancy and position of the second molybdenum site could be refined in the same cycle without leading to instability. The position of each of Mo(2), O(l), O(2) and O(3) was refined in turn until no further significant changes occurred, then the occupancy of each of the molyb- denum sites (constrained as above) and the oxygen sites were allowed to vary.In the last cycles of the refinement, the scale, the profile parameters, the relative occupancies of the two molybdenum sites, the occupancies of the three oxygen sites and isotropic temperature factors for the yttrium site, the molybdenum sites (constrained to be equal) and the oxygen sites (again constrained to be equal) were refined. While the occupancies of the three oxygen sites could not be constrained, it was encouraging to note that the final sum of the fractional occupancies (0.87) was close to that expected (0.9) for Y,Mo,O,, in the absence of any Y2Mo3OI2impurity. The final agreement factors were RI=9.61%, R, =4.10%, and R,, =3.94% (Rexp=0.76%). The observed, calculated and difference profiles are shown in Fig.4,and the atomic param- eters are listed in Table 1. Where esds are quoted in Table 1 the values are those calculated during the last cycle of refinement, and are likely to be underestimates as the effect of any correlation with variables not refined during the last cycle is not included; the value of Rex, will be affected in a similar way. No esds are available for the atomic coordinates, as these could not be refined freely. Table 1 Final atomic parameters for Y,Mo,Ol8 atom site X Y z occupancy B/A2 ~~~ ~ Y 4(b) 0.5 0.75 0.125 1.000 1.73(5) Mo(1) 4(a) 0.0 0.25 0.125 0.628(7) 0.52(6) Mo(2) 8(e) 0.0 0.25 0.080 0.071(3) 0.52(6) 0(1) 16(f) 0.185 0.496 0.213 0.562(7) 0.61(4) O(2) 16(f) 0.240 0.354 0.448 0.206(4) 0.61(4) O(3) 16(f) 0.259 0.487 0.230 0.097(3) 0.61(4) Unit-cell parameters a=5.17667(2) A,c= 11.00451(7)A.As a full least-squares refinement of all of the variables describing this structure was not possible the final model obtained cannot be described as definitive. However, it is a good description of a complex material. The refined stoichi- ometry (Y5h!f03&17.3) differs from the ideal in a way which is consistent with the presence of a Y2Mo3OI2impurity. The two possible molybdenum sites are only 0.49 A apart, and differ only in their z coordinate. The oxygen sites are also close together. The m?iety O(2)-O( 1)-O( 3) is approximately linear with O(2)0.46 A from O(1) and O(3) only 0.43 A from O(1).The scattering density revealed in the difference Fourier maps thus reflects not new atomic sites, but the presence of very considerable disorder in the original sites.The scattering density around the molybdenum site suggests that the molyb- denum atoms are disordered around the 4(a) site, but in particular that many are displaced away froni the special position along the z axis in either direction. The scattering density found around the oxygen site also reflects extensive disorder in a direction roughly perpendicular to the Mo-0 and Y-0 bonds, as one might expect of extreme thermal motion. This disorder is represented in Fig. 5, where the size of each of the molybdenum and oxygen ‘atoms’ has been adjusted to give some impression of the relative occupancy of the sites.The rather high final intensity R factor would suggest that still more disorder may be present that is not accounted for by the model given in Table 1. In particular, the high temperature factor found for yttrium could indicate that disorder is present around the 4(b) site as well, but as no significant peaks near the yttrium site were found in the final 10 15 time of flight/ms Fig. 4 Observed (data points), calculated (line) and difference neutron powder diffraction profiles for Y,Mo,O,, . Reflection markers are shown. J. MATER. CHEM., 1994, VOL. 4 "0 on noOWI IxV 1A vA l xI no on I I 1Ob a Fig. 5 A representation of the structure of Y,Mo,O,,, viewed along y. Molybdenum atoms are shown in black, yttrium atoms are shaded and oxygen sites are unshaded.Table 2 Selected bond lengths and angles for Y5Mo4OI8 Y-O(1) 4 x 2.41 4, 4 x 2.31 a Mo( 1)-O( 1) 4 x 1.87 A O(l)-Mo( 1)-0( 1) 4 x 105.6", 2 x 117.4' Y-O(2) 4 x 2.37 A, 4 x 2.53 A Y-O(3) 4 x 2.18 A, 4 x 2.49 A Mo( 1)-0(2) 4 x 1.664 Mo( 1)-0(3) 4x 2.15 4 Mo( 1)-Mo(2) 2 x 0.49 A 0.46 A 0.43 A difference Fourier maps this could not be modelled in the same way as that around the molybdenum and oxygen sites. Selected bond lengths and angles are given in Table 2. For the main part these reflect the geometry of the basic model, comprising Y( l), Mo( 1) and O(l), as it is not possible to interpret the distances between Mo(2), O(2) and O(3) in terms of a realistic model.Although esds are not available, we note that the Y-0 bond lengths are comparable with those found in other compounds, and that the Mo-0 bond length is somewhat longer than that found in, many com-pounds containing tetrahedral MeV' (ca. 1.75 A), consistent with the lower formal oxidation state of Mo (+5.25) in Y,Mo,O,, [or Mo(+ 5.09) in Y5M03.85017.3]. Conclusions Analytical electron microscopy has shown that the scheelite form of 'YMoO,' reported by earlier workers was, in fact, biphasic, and that a phase which had been reported as a monoclinic, high-temperature form of 'YMoO,' contains sili- con, leached from the silica tubes used as reaction vessels. One component of the mixture 'YMoO,', having a composi- tion close to Y5M04018,was isolated and studied by powder neutron diffraction. Rietveld analysis of the neutron diffraction data showed that this phase adopts a scheelite-related struc- ture in which considerable disorder is found around the partially occupied molybdenum and oxygen sites.The other component, having a composition close to Y,Mo3O1,, is also believed to adopt a scheelite-related structure. We are grateful to SERC for the provision of equipment and a studentship for N.J.S. References 1 H. Kerner-Czeskleba and G. Tourne, Bull. SOC. Chim. Fr., 1976, 5-6, 729. 2 P-H. Hubert, C.R. Acad. Sci. Paris, Ser. C, 1977,285, 1507. 3 P-H. Hubert, C.R. Acad. Sci. Paris, Ser. C, 1978,286,191. 4 H. Kerner-Czeskleba and G. Tourne, Muter. Res. Bull., 1978, 13, 271.5 A. Manthiram and J. Gopalakrishnan, J. Proc. Znd. Acad. Sci., 1978,87A, 267. 6 A. Manthiram and J. Gopalakrishnan, J. Less-Common Met., 1979,68, 167. 7 W. H. McCarroll, C. Darling and G. Jakubicki, J. Solid State Chem., 1983,48,189. 8 C. C. Torardi, C. Fecketter, W. H. McCarroll and F. J. DiSalvo, J. Solid State Chem., 1985,60, 332. 9 E. Banks and M. Nemiroff, Znorg. Chem., 1974,13,2715. 10 J. E. Greedan, M. Sato, Xu Yan and F. S. Razavi, Solid State Commun., 1986,59,895. 11 P. Gougeon, P. Gall and R. E. McCarley, Acta Crystallogr., Sect. C, 1991,47,2026. 12 P. Gougeon, P. Gall and R. E. McCarley, Acta Crystallogr., Sect. C, 1991,47,1585. 13 S. J. Hibble, A. K. Cheetham, A. R. L. Bogle, H. R. Wakerley and D. E. Cox, J. Am. Chem. Soc., 1988,110,3295. 14 H. Leligny, M. Ledesert, Ph. Labbe, B. Raveau and W. H. McCarroll, J. Solid State Chem., 1990,87,35. 15 A. K. Cheetham and A. J. Skarnulis, Anal. Chem., 1981,53,1060. 16 W. I. F. David, M. W. Johnson, K. J. Knowles, C. M. Moreton- Smith, G. D. Crosbie, E. P. Campbell, S. P. Graham and J. S. Lyal, Rutherford-Appleton Laboratory Report RAL-89-118, 1989. 17 W. H. McCarroll, personal communication to JCPDS Powder Diffraction File (card no. 35-1477), JCPDS International Centre for Diffraction Data, Swarthmore, PA, 1987. 18 J. Gopalakrishnan and A. Manthiram, J. Chem. Soc., Dalton Trans., 1981, 668. 19 N. J. Stedman, D. Phil. Thesis, Oxford University, 1994. 20 P. W. Betteridge, A. K. Cheetham, J. A. K. Howard, G. Jakubicki and W. H. McCarroll, Znorg. Chem., 1984,23,737. 21 International Tables for Crystallography, ed. T. Hahn Kluwer, Dordrecht, 1989, vol. A, no. 88. 22 N. J. Stedman, A. K. Cheetham and P. D. Battle, J. Muter. Chem., 1994,4,707. Paper 4/02309I; Received 19th April, 1994
ISSN:0959-9428
DOI:10.1039/JM9940401457
出版商:RSC
年代:1994
数据来源: RSC
|
|