|
11. |
Studies on the electrochemical deposition of niobium oxide |
|
Journal of Materials Chemistry,
Volume 6,
Issue 2,
1996,
Page 187-192
G. Robert Lee,
Preview
|
PDF (880KB)
|
|
摘要:
Studies on the electrochemical deposition of niobium oxide G.Robert Leet and Joe A. Crayston" Department of Chemistry, University of St. Andrews, St. Andrews, Ffe, Scotland, UK K Y16 9ST Various strategies for the electrodeposition of niobium oxide from both non-aqueous and aqueous solution are explored. The methods used were based on the electrochemical manipulation of the pH which leads to hydrolysis of soluble, well defined niobium precursors, leading to oxide deposition at the electrode. The reactions which proceed are believed to be closely related to those involved in conventional sol-gel processing. In non-aqueous systems containing niobium alkoxides, hydroxide ions are generated by the two-electron reduction of tertiary alcohols (at c -2.5 V) and react with the niobium precursor to give a film of niobium oxide on the dropping mercury electrode.Hydroxide generation via the reduction of oxygen to superoxide in 0,-saturated, non-aqueous solvents containing NbOC1, did not, however, lead to niobium oxide films. In acidic aqueous systems containing [NbOCl5I2- species, raising the local pH around the electrode by hydrogen evolution does give rise to niobium oxide deposition, but not at the electrode. In aqueous alkaline solutions of the niobates K,H [Nb,O,,] or (TMA), [Nblo028] (TMA=tetramethylammonium), protons generated by the electrochemical oxidation of water at > +1.5V cause the formation of mixed niobate and niobium oxide films on the electrode after >30 min. Electrochemical deposition of ceramic metal oxide films can, like sol-gel processing (SGP),' offer the advantages of low- temperature processing, high purity and controlled microstruc- ture.However, unlike SGP it also offers the possibility of forming crystalline rather than amorphous low-temperature processed films., Much work has been devoted to developing techniques for the electrodeposition of metal oxide films, mainly from aqueous solution using a redox change technique. Here, a metal is oxidised or reduced at the electrode surface by a fixed applied potential. The pH of the solution is adjusted so that although the initial oxidation state is stable, the new oxidation state readily undergoes hydrolysis to a metal hydrox- ide or oxide. This metal hydroxide/oxide is insoluble and forms a film on the electrode.Some of the first research performed using this technique was that by Tench and Warren3 who successfully electrodepos- ited oxide and hydroxide films (up to 1pm thick) of copper, nickel, cobalt, iron and manganese from aqueous metal salt solutions (pH 6.5-7.5). These films were shown to have some interesting properties: thus, the nickel oxy-hydroxide films showed electrochromic properties although the crystalline Cu,O films exhibited p-type semiconducting behaviour when under illumination. Since then the technique has blossomed with much work being directed at the formation of the YBaCuO superconductors4 and lead/thallium oxide film^.^,^-' Upon co-deposition of T1203and PbO,, Sakai et al.5 obtained a new compound, Pb8T15024, which had a cubic fluorite-type structure, whilst Thomas and Wabner' were able to grow lead dioxide anodes (for use as dimensionally stable electrodes), and by varying the electrodeposition temperature vary the content of P-PbO, relative to a-PbO,.Switzer2*6 prepared films of T1203 on n-silicon to give a heterojunction photoelec- trochemical solar cell of 11% efficiency with 75.3 mW cm-2 natural sunlight, as well as mixed Pb/Tl oxide films with a superlat tice st ruc t Recently, attention has turned to second and third row transition-metal oxide electrodeposition because of their elec- trochromic and catalytic properties. Here, potential cycling, rather than the earlier fixed applied potential method, is used as this favours the formation of porous, hydrated metal oxide coatings.' IrO,," W03, MOO, and mixed W03/Mo03 films" have been prepared using this method and their electrochromic t Present address: Shell Research, Thornton Research Centre, Chester, UK.properties have been investigated. Kulesza and Faulkner12 prepared mixed-valent tungsten(vi ,v) oxide films and examined their electrocatalytic properties. The use of organic solvents for electrodeposition of metal oxides has not received as much attention as that of the aqueous systems and, in general, has been limited to the electrodeposition of pure metals which could subsequently be converted to oxides. YBaCu metal films have been deposited on cubic zirconia substrates from dimethyl sulfoxide, Me,S0,13*14 at extreme negative potentials (-2.5 to -5 V us.Ag/Ag+) and sintered at 900 "Cto form YBaCuO superconduc- tors, with zero resistance at 91 K. However, use of a mixed H20-Me,SO solvent allows for the direct electrodeposition of YBaCuO films" at less negative potentials (-1.4 to -2.0 V us. Ag/Ag+) than those required for the deposition of YBaCu metal films. This favourable shift in the deposition potentials is attributed to the deposition of the oxide/hydroxide with the subsequent evolution of H,. Although hydrogen evolution in this solvent system is observed at -2.0 V, the metal ions in solution act as hydrogen evolution promoters and react as follows [eqn. (l)]: M"++xH,O+xe- +M(OH),+"/,H, (1) We have recently reported on spin-coating and dipping methods for niobium oxide film formation from viscous solu- tions of niobium chloride in alcohol partially hydrolysed via the SGP technique.',-'' Niobium oxide films have also been prepared from niobium alkoxides/acetic acid2' and a niobium oxalate complex in the presence of citric acid and ethylene glycol.21 Electrodeposition of metal oxides, specifically niobium oxide, seemed a natural extension of our work.There is no precedent for niobium oxide electrodeposition, although niobium metal has been electrodeposited from molten salts in alkali-metal~~~-~~or pure liquid halogens,25 and also from propylene carbonate and acetonitrile using LiNbF, .26 Once deposited, the niobium metal may be anodised to niobium oxide.27 There are several difficulties in electrodepositing Nb (and Nb205), including the extreme negative reduction poten- tial required (which leads to interference by H, evolution in aqueous systems), the tendency of niobium species to form electrochemically inactive species when reacted with 0, or H20 and the ability to form stable cluster compounds in low oxidation states.,, We decided to attempt to circumvent these problems by J.Mater. Chem., 1996,6(2), 187-192 187 combining SGP with electrodeposition. The first set of experi- ments consisted of deposition from organic solutions, the second from aqueous solutions. Regarding the former experi- ments, the readily available niobium precursor compounds which are soluble in organic solvents (such as chloride or alkoxide) are very moisture sensitive, readily forming insoluble Nb205.Therefore, in order to electrodeposit Nb205 from organic solvents containing NbX,, H,O or OH- must be generated electrochemically. Two methods for the electro- chemical generation of H20 or OH- were investigated. One uses superoxide in the presence of residual ~ater:,~.,~ 20,-+H,O $ 0,+H0,-+OH-(2) The other method involves the reduction of tertiary alcohols to generate OH- ion. Lund and coworker^^^,^^ have shown that tertiary alcohols can undergo a two-electron reduction of the following type: ROH +2e-+H+ +RH +OH- (3) Turning to electrodeposition under aqueous conditions, we took advantage of the fact that certain chloroaquaniobium species are known to be stable in aqueous acidic conditions, and pH manipulation should cause the formation of hydrated Nb205 gel.Therefore, electrochemical generation of either OH-or H+ in these solutions by electrochemical methods similar to those mentioned above should promote the electrodeposition of niobium oxide films. Experimenta1 All preparative reactions were performed under N, and in dry glassware (at 293 K except where noted), and unless specifically noted $1 solvents, regardless of grade, were redistilled over Na, 3 A molecular sieves or CaH,. 93Nb NMR (coupled) spectra were run on a Bruker MSL 500 spectrometer (zo M 500 ms, z, =5-20 ms), and 'H NMR spectra on an AM300 or WP80 spectrometer. FTIR spectra were obtained on a Perkin-Elmer 1710 spectrometer, and Raman spectra on a Spex 1403 instrument interfaced to a DMlB computer with +Coherent Radiation Innova 90-6 Ar laser excitation at 514.5 nm, power -=100 mW at the sample (spectroscopic data are within f1 cm-I).Unfortunately, good quality Raman spectra of the electrodeposited films could not be obtained owing to weak scattering and strong fluorescence from an unidentified trace impurity. X-Ray powder diffraction (XRPD) studies were run on a Stoe Stadi-P instrument using Cu-Ka, radiation. SEM (scanning electron microscopy) was performed on a JEOL JSM 35CF instrument (15 keV) with a scanning EDAX (energy dispersive analysis by X-rays) attachment at the Gatty Marine Laboratory. Chemicals Triphenylmethanol (TPM) (KochLight, puriss grade) and 9-fluorenemethanol (FM) (Aldrich, 99"/), were recrystallised from dimethylformamide (DMF) and vacuum-dried over P205 for 3 days.The electrolyte was either tetrabutylammonium hexafluorophosphate (TBAPF,, Aldrich) or tetrabutyl-ammonium dihydrogenphosphate (TBADHP) (Aldrich, 97% recrystallised from acetonitrile, vacuum-dried over P205). TBADHP is not soluble in DMF. (NbCl,), (vacuum-dried over P205) was laboratory grade (Aldrich). General procedures for preparing solutions in acetonitrile, dimethyl sulfoxide (Me,SO) and DMF were as described NbOCl, species were prepared in situ by the addition of 2 equivalents of demineralised water and identified by 93Nb NMR spectroscopy:NbOCl, 2CH,CN has a peak at -498 ppm (FWHM M 850 Hz), and NbOC1, in DMF has a peak at -500 ppm (FWHM M 1200 Hz).,, Before carrying out the electrochemical study of (NbCl,), in DMF we investigated whether (NbCl,), reacts with this solvent. However, in our electrochemical experiments there was no 93Nb NMR evidence for either NbOCl, (-500 ppm'8330) or NbC1,-(0 ppm), implying that the solution is anhydrous.'Amorphous' K,H [Nb,O,,] 13H20 K,H [Nb6019] 13H20 was prepared by the method of Edlund et ~1.~~"Thus, 10 g of Nb205 (Aldrich, 99.9%) was fused with 20g KOH (Aldrich, 99.99%) in a nickel crucible to form a clear melt. This was then dissolved in 200 ml of distilled water and filtered. Equal amounts of ethanol (Aldrich, laboratory grade) were added, the solution stirred vigorously, the precipitate filtered and washed with cold ethanol.The pre- cipitate was then recrystallised from water twice to give K,H [Nb6Ol9] 13H,O. This was characterized by XRD stud- ies, and was found to match the theoretical powder diffraction pattern for K,H[Nb,O,,] 13H,O calculated from the single- crystal data3, using a Stoe-modified version of the Lazy Pulverix program. Unfortunately, this product, in common with commercial samples of K,H [Nb6Ol9] * 13H20, is only sparingly soluble in aqueous solution, even 1.0 mol dmP3 KOH. However, the first material to be precipitated during the synthesis is far more soluble than subsequent precipitates. The FTIR spectrum of the first precipitate [Fig.2(c); later] shows three strong bands at 850, 700(br) and 520cm-', characteristic of [Nb6019]8- but without the strong peak at 1450cm-I and the weak peak at 880cm-' characteristic of crystalline K,H[Nb6Ol9] 13H20 [Fig. 2(!1)].~~ Powder diffraction data show a set of peaks in the 28 =9.0-14.0' region characteristic of [Nb601g]'- species. That these peaks do not match a single compound is indicative of a solid solution of cations (K+, H') present in a different ratio to the K,H [Nb6019]-13H,O reference sample. This implies the for- mation of [Nb6019]8- units but no long-range crystallinity, and this first precipitate is referred to as 'amorphous' K,H[Nb6Ol9]*13H2O. The FTIR and Raman spectra of this material were consistent with literature reports [FTIR: v(Nb=D) 852, 700(br) and 520 (TMA), "b100281*6H2O This niobate was prepared by the published method3' of dissolving niobium ethoxide in methanol and precipitation with tetramethylammonium hydroxide-water.The XRD powder pattern was consistent with the literature data35 and the measured IR and Raman spectra were identical to those reported [FTIR: v(Nb=O) 896, 800 ~m-'].~, (TMA), "bOCl5I Niobium chloride (6.2g) was dissolved (accompanied by some effervescence) in ethanol (11.5 cm3) to give a 2 mol dm-, solution. To this was added 4.96 g of tetramethylammonium hydroxide in methanol (20%) followed by water, whereupon white fumes were evolved and a white crystalline precipitate was formed. The precipitate was filtered, washed with ethanol and dried (yield 4 g, 40%).FTIR [v(Nb=O) 925 cm-'1, 'H NMR, CHN microanalysis and Cl and Nb analysis were consistent with the formulation and with literature reports of salts containing this anion.,, Electrochemistry Cyclic voltammetry (CV) was carried out using a combined pulse/scan generator and potentiostat (Pine Instruments RDE4) and a Graphtec xy recorder with a standard three- electrode cell. The working electrode was a home-built glassy carbon electrode (4 mm diameter), the counter electrode was a platinum wire and the reference electrode was a SCE separated by a Luggin capillary. Working and counter elec- trodes and glassware were vacuum-dried over P205and all 188 J. Muter. Chem., 1996, 6(2), 187-192 the electrochemistry was performed with degassed solutions under N,.Background CVs of the Nb-free solvents showed no electrochemistry in the regions being examined. For the aqueous electrochemistry, concentrated HCl or 0.1 mol dm-3 KOH solution were used as the solvent and electrolyte systems. A platinum disc electrode (diameter 6 mm) working electrode was also used in conjunction with the carbon electrode. Solid working electrodes were polished consecutively with 5, 1 and 0.25 pm diamond polish (Engis Ltd, Hyprez Five Star grade), washed in acetone, sonicated in an acetone bath for 30min, washed again with demineralised water and acetone and finally dried at 80°C. The hanging mercury drop electrode was prepared from a 1 mm diameter Pt electrode and triply distilled mercury. During electrodeposition, film formation was checked visu- ally, and then electrochemically by cycling the electrodes in 1.0mol dm-3 H2S04 between 0.2 and -1.OV in order to observe the characteristic Nb205 CVs.“ Films were then analysed using FTIR, X-ray powder diffraction (XRPD) and SEM-EDAX.Superoxide generation Solutions (0.0025 mol dm-3) of (NbCl,), or [Nb(OR),], (where R=Et or Pr) in acetonitrile, DMF or Me2S0 with 0.1 mol dm-3 supporting electrolyte were prepared (Table 1). 0, was bubbled into the solutions at 1 cm3 s-l through a sintered glass frit. Results and Discussion Non-aqueous systems Superoxide generation. Constant potential electrolysis of non-aqueous solutions (CH3CN, DMF and Me,SO) containing 0.0025 mol dm-3 niobium precursor ((NbC15)2, [Nb(OEt),],, or [Nb(OPr‘),],), 0, and electrolyte (0.1 rnol dm-3 TBAPF6 or TBADHP) were undertaken in an attempt to form films of Nb205.The reasoning behind these experiments was as follows. Oxygen readily undergoes a one electron reduction to form the superoxide ion:,’ O,+e-$0,-(4) This can then undergo a second irreversible reduction (Epc= -1.8 V), in which the product species reacts with solvent or electr~lyte:~’~~~ 0,-+e-+OZ2-(5) 02’-+solv-H+HO,-+sOlv (6) HO2-+solv-H+H,Oz +Solv (7) where solv-H is a proton source (often the tetrabutyl-ammonium ion). H202 is known to react with NbC1, to give Table 1 Summary experiments of superoxide generation electrodeposition electrolysis electrode Nb precursor electrolyte solvent‘ time/h C DMF 4-16 Pt DMF 1-16 AUb DMF 1-16 C DMF 14 AUb DMF 1-6 C CH3CN 1 C CH3CN 2 Aub CH,CN 1-16 C Me,SO 1-10 -1.8 to -2.0 V (vs.SCE) was applied in CH3CN or DMF solutions, -1.1V (vs.SCE) in Me,SO solutions. 6 mm diameter Au disc electrode. Nb,O, gels, so it was hoped that the production of H,O, at the electrode surface (viaH+ abstraction from either TBA or H,PO,-) might give rise to Nb205 film formation. Likewise, the production of H02- might lead to Nb205 via the following type of reaction: 10H02-+2NbX5 +Nb(OH)S + 1OX-+50, (8) Cyclic voltammetry in 0,-saturated CH3CN or DMF revealed the 02+e- e0,-redox reaction at -1.0 V (Epc= -1.05, Epa=-0.95 V, vs.SCE) and the second reduction of oxygen, 0,-+e-+02,-at -1.8 V.26*27 When a constant potential of -1.8 to -2.0 V was applied with continuous 0,-bubbling, regardless of solvent (CH,CN, DMF), working electrode (Au, Pt, C), niobium precursor or period of electroly- sis, no Nb205 film was ever observed. The reason for this failure is not clear, but it is probable that the superoxide and its products were reacting with the solvent and electrolyte rather than the niobium precursors. Evidence for this belief is found when the electrolysis was performed using the Au electrode. Upon electrolysis, small dendritic needles of a white compound were formed. SEM-EDAX analysis revealed that this compound contained no metals, indicating the formation of an insoluble organic material.That this was only observed on the Au electrode is ascribed to the fact that Au is the best generator of the superoxide ion.28929 Using Me,SO as solvent, despite being redistilled over CaH,, upon the addition of (NbCl,), the solutions in Me,SO gave voltammograms characteristic of NbOC13 formation’’ arising from the reaction with trace water in the solvent. Superoxide reacts in the following manner with water:29 202-+Hz0 +02+HOz-+OH-(K=0.91 x lo9) (9) By electrolysing the solutions at -1.1 V (us. SCE) it was hoped to generate OH- which would react with the NbOC13 to give Nb205. The failure to form Nb,05 can be ascribed to the neutralisation of OH-by Hf generated in the formation of NbOCl, and the inherent stability of NbOC1, towards hydrolysis.Reduction of tertiary alcohols. This reaction occurs in anhy- drous DMF (0.1 mol dm-3 TBAPF, electrolyte), on a mercury electrode and at very negative potentials, usually beyond -2.4 V us. SCE. The generation of OH- by this reduction [eqn. (lo)] in a solution containing niobium precursor species should instigate a sol-gel reaction and formation of a Nb205 film. Two tertiary alcohols were investigated: triphenylmethanol (TPM, El/, = -2.81 V30) and 9-fluorenemethanol (FM, El/,= -2.47 V3’). Solutions of these alcohols (0.1 mol dm-3) in anhydrous DMF containing 0.1 mol dm-3 TBAPF, and 0.0025 mol dm-3 [Nb(OEt),], were electrolysed at -2.90 and -2.50 V, respectively, for 3-16 h. This produced yellow/white air-stable films around the mercury drop.SEM-EDAX analysis indicated the sole presence of niobium, and the absence of mercury. FTIR spectra confirmed that these films were Nb205.19*38Fig. l(u) and (b) show SEM pictures of one of these coatings. Fig. l(u) shows how the film followed the circular contours of the Hg drop. Although this was a successful technique for the electro- deposition of Nb205 films, the extreme negative potentials excluded the possibility of forming films on more convenient electrodes (e.g. C) and the cost of the alcohols would prove prohibitive for the use of this technique to generate large area Nb205 films. Because of these problems no further investi- gations of this approach were undertaken. J. Muter.Chem., 1996, 6(2), 187-192 189 10 P.1 Fig. 1 (a) $EM images of Nb,O, film produced by the reduction of triphenylmethanol; (b)at higher magnification Aqueous systems Niobium(v) precursor compounds, e.g. Nb(OR), , are notoriously moisture-sensitive and readily form insoluble hydrated Nb205 upon addition of water. A soluble aqueous system may be formed by the addition of (NbC15)2 to conc. HCl.39 Such solutions contain species such as [NbOCl5I2-. Similarly, most niobate compounds are insoluble except for K,H[Nb,Ol9] 13H,O which is sparingly soluble in strong alkaline sol~tion.~~~~~ Both these systems are pH-sensitive: increasing the pH of the HC1 solution from -1 to -0.85 results in the formation of an Nb205 gel. Likewise, lowering the pH of the alkaline solution causes K7H [Nb6O19] 13H,O to undergo hydrolysis to form H6Nb6018 which dissociates irreversibly to form Nb205 and ~ater.~'-~~ As both H+ and OH-can be generated electrochemically from H20, we attempted the electrodeposition of niobium oxide films from these aqueous systems. (NbC1,)2 in conc.HCI. Electrolysis of a 0.1 mol dmP3 solution of (NbCl,), in conc. (10.8 mol dm-3) HC1 at -0.5 V (Pt electrode) or -1.0 V (C electrode) reduces water by the following reaction: e- +H,O+ -+OH- +$H2 (11) The production of OH- increases the pH of the solution locally near the electrode and precipitates Nb205 beneath the electrode, rather than forming a film on the working electrode. The lack of adhesion is surprising in view of the success in forming a film of W03 upon pH manipulation.12 During the electrolysis we noted that H, evolution was accompanied by a large and very fast formation of small H2 bubbles on the electrode surface which were rapidly lost to the solution.Thus it appears that any film formed is immediately pulled off by the buoyancy action of the H, bubbles, giving the observed 190 J. Mater. Chem., 1996, 6(2), 187-192 precipitate. The Nb205, upon washing and subsequent drying at 150°C, gave FTIR spectra similar to those found for the niobium chloroalkoxide gels described previo~sly.'~ (TMA),[NbOCI,]. Since the above solutions are known to contain the [NbOCl5I2- anion, which may be precipitated from solution using NH4+ or tetraalkylammonium ions, we next investigated the electrolysis of solutions made up directly from (TMA), [NbOCl,]. This material dissolves with momen- tary effervescence in water to produce an acidic solution: a 0.2 mol CM-~solution has a measured pH of 1.0.Electrolysis of this solution at -0.5 V (designed to locally increase the pH) led to hydrogen evolution and a cloudy solution but no electrode coating. Control experiments showed that when the pH of this solution is raised to >1.5 an Nb,05 gel begins to form. Prolonged electrolysis (> 1 h) in an undivided cell caused the entire solution to gel, and chlorine evolution at the anode was detected. Mixed organic-aqueous systems. Owing to the unexpected and favourable shift in deposition potentials for the deposition of YBaCuO films in mixed organic-aqueous systems,15 we felt we should examine this type of system for the case of niobium.Addition of 1.08 ml of conc. HCl to 10 ml of 0.1 mol dm-3 (NbCl,), in CH,CN or DMF (Nb:H,O=1:20) yielded a solution whose cyclic voltammogram indicated the formation of NbOC1,. Potential cycling around the NbV/NbIV redox potential gave a blue solution but no film formation. Electrolysis at -1.20 V (us. SCE) using either Pt or C electrodes caused H, gas evolution and an Nb205 precipitate similar to that found using the conc. HCl conditions, but no adherent film. K,H[Nb,O,,] in 0.1 rnol dm-3 KOH. Cyclic voltammetry of 0.002 mol dm-3 solutions of 'amorphous' (see Experimental section) K,H[Nb6019]*13H20 in 0.1 rnol dmP3 KOH revealed no obvious redox processes within the available potential window.Electrolysis of a solution of 'amorphous' K,H[Nb,Ol9] (0.0019mol drn-,) in 0.1 rnol dm-3 KOH between + 1.60 and + 1.95 V for 1-23 h was found to lead to electrode deposits. At these voltages water undergoes the following oxidation: H20+2H+ ++02+2e-( 12) The protons generated neutralise the base, lowering the pH and electrodepositing a non-uniformly thick (ca.0.1 mm) white film on the platinum electrode. XRPD analysis of this film confirmed that it was amorphous. FTIR analysis shows that there is a pronounced shoulder at 850cm-' on the broad Nb-0 absorption band (Fig. 2). This indicates a mixture of [Nb,019]8-groups and Nb205.34338 The voltammograms of these films, obtained in 1.0mol dm-3 H,SO,, are characteristic of Nb205 behaviour (see Fig.3).16 Unfortunately, films could not be electrodeposited on transparent electrodes, thus precluding the use of spectro- electrochemical techniques. The non-uniformity of the film is believed to be due to the slow formation of large 0, bubbles during its formation, which nucleate at the film surface and cause parts of the film to be detached from the electrode. In order to further characterise these films a method of sup- pressing 0, bubble formation would be needed. (TMA),[Nb,,O,] in 0.1 rnol dm-3 KOH. Similar films were obtained using aqueous alkaline solutions of the decaniobate, (TMA)6[Nblo028] .6H20 (0.012 mol dm-3).The FTIR spec- tra of these films indicated complete loss of TMA bands and the high frequency v(Nb=O) band at 896 cm-' but retention of the 802cm-' v(Nb=O) found in the starting material (Fig. 4).This suggests34 that the highly distorted NbO, groups in the starting material are not present in the coating, perhaps Fig.2 FTIR spectra of K7H[Nb6Ol9].13H20 and products: (a) film formed by electrodeposition on Pt from K7H [Nb6019].13H20-KOH solution (0.1 mol dmP3); (b) K7H[Nb,019].13H20 (crystals, as KBr disc); and (c)K7H[Nb6OI9] 13H20 (first precipitate, i.e. 'amorphous' K7H [Nb,o,,]* 13H20 as KBr disc) 11111 -0.4 -0.2 0.0 0.2 0.4 EN vs. SCE Fig. 3 CV (scan rate 100 mV s-') of electrodeposited K,H[Nb,ol9] derived film in 1.0 mol dm-3 H,SO, on a Pt electrode having condensed, but the less distorted NbO, octahedra are incorporated unchanged.Conclusions Electrodeposition of niobium oxide from a number of aqueous and non-aqueous solvent systems, via sol-gel processing techniques, has been attempted by the electrochemical pH manipulation of the solutions. These techniques rely upon the 1: 0 800 400 vlcm-1 Fig. 4 FTIR spectra of: (a) (TMA),[Nbl0O2,] (KBr disc); (b) film formed by electrodeposition on Pt from (TMA)6[Nblo028] in 0.1 mol dm-3 KOH solution electrochemical generation of protons and hydroxide ions. In non-aqueous systems containing niobium alkoxides, hydroxide ions are generated by the two-electron reduction of tertiary alcohols and react with the niobium precursor to give a film of niobium oxide on the electrode.In aqueous alkaline systems containing niobate, protons are generated by the electrochemi- cal oxidation of water, accompanied by a lowering of the pH. This destabilises the niobate sol and forms an electrode coating of mixed niobate and niobium oxide. Further work needs to be targeted at producing reproducible and uniform films. References 1 G. R. Lee and J. A. Crayston, Adv. Mater., 1993,5,434. 2 J. A. Switzer, Am. Ceram. SOC. Bull., 1987,66, 1521. 3 D. Tench and L. F. Warren, J. Electrochem. SOC., 1983,130,869. 4 P. Slezak and A. Wieckowski, J.Electrochem. SOC., 1991,138,1039. 5 M. Sakai, T. Sekine and Y. Yamazaki, J. Electrochem. SOC., 1983, 130,1631. 6 J. A. Switzer, J.Electrochem.SOC., 1986, 133,722. 7 J. A. Switzer, Science, 1990,247,444. 8 J. C. G. Thomas and D. W. Wabner, J. Electroanal. Chem., 1985, 182,25. 9 L. D. Burke and E. J. M. O'Sullivan, J. Electroanal. Chem., 1981, 117, 155. 10 T. Yoshino, N. Baba, H. Masuda and K. Arai, Proceedings of the Symposium on Electrochromic Materials, ed. M. K. Carpenter and D. A. Corrigan, The Electrochemical Society, Philadelphia, VO~.90-2, 1990, p. 89. 11 L. H. Dao, A. Guerfi, M. T. Nguyen and K. Arai, Proceedings of the Symposium on Electrochromic Materials, ed. M. K. Carpenter and D. A. Corrigan, The Electrochemical Society, Philadelphia, VO~.90-2, 1990, p. 30. 12 P. J. Kulesza and L. R. Faulkner, J. Am. Chem. SOC., 1988, 110, 4905. J. Muter.Chem., 1996, 6(2), 187-192 191 13 M. Maxfield, H. Eckhardt, Z. Iqbal, F. Reidinger and 31 M. A. Michel, G. Mousset, J. Simonet and H. Lund, Electrochim 14 R. H. Baughman, Appl. Phys. Lett., 1989,54,1932. R. N. Bhattacharya, R. Noufi, L. L. Roybal and R. K. Ahrenkiel, 32 Acta, 1974,19,629. G. R. Lee and J. A. Crayston, J. Chem. SOC., Dalton Trans., 1991, J.Electrochem. SOC., 1991, 138, 1643. 3073. 15 H. Minoura, K. Naruto, H. Takano, E. Haseo, T. Suguira, Y. Ueno and T. Endo, Chem. Lett., 1991,379. 33 (a) D. J. Edlund, R. J. Saxton, D. K. Lyon and R. G. Finke, Organometallics, 1988, 7, 1692; (b) A. Goiffon, E. Philippot and 16 17 G. R. Lee and J. A. Crayston, J. Mater. Chem., 1991,1,381. S. A. Campbell, G. R. Lee and J. A. Crayston, Topical Issues Glass, 34 M.Maurin, Rev. Chim. Miner., 1980, 17,466. (a) Von R. Mattes, H. Bierbusse and J. Fuchs, 2. Anorg. Allg. 18 19 20 21 22 23 24 25 26 27 28 29 1991, 1,64. G. R. Lee, Ph. D. Thesis, University of St. Andrews, 1992. G. R. Lee and J. A. Crayston, J. Mater. Chem., 1994,4,1093. P. Griesmar, G. Papin, C. Sanchez and J. Livage, Chem. Mater., 1991,3,335. R. C. Faria and L.-0. de S. Bulhoes, J. Electrochem. SOC., 1994, 141, L29. U. Cohen, J.Electrochem. Soc., 1981, 128,731. G. P. Capsimalis, E. S. Chen, R. E. Paterson and I. Ahmad, J.Appl. Electrochem., 1987,17,253. Q.Zhiyu and T. Pierre, J.Appl. Electrochem., 1985,15,259. R. M. Rose and D. R. Sodoway, US Pat., 4517253,1985. N. S. Chong, J. L. Anderson and M. L. Norton, J. Electrochem.SOC.,1989,136, 1245. M. A. Biason Gomes, S. Onofre, S. Juanto and L. 0.de S. Bulhoes, J. Appl. Electrochem., 1991,21, 1023. D. T. Sawyer and E. J. Narni, Oxygen and Oxy Radicals in Chemistry and Biology, Academic Press, 1981,pp. 15-44. D. T. Sawyer and J. S. Valentine, Acc. Chem. Res., 1981,14,393. 35 36 37 38 39 40 41 42 43 Chem., 1971, 385, 230; (b) J-M. Jehng and I. E. Wachs, Chem. Mater., 1991, 3, 100; (c)C. Rocchioccioli-Deltcheff,R. Thouvenot and M. Dabbabi, Spectrochim. Acta, Part A, 1977,33,143. E. J. Graeber and B. Morosin, Acta Crystallogr., Sect. B, 1977, 33,2137. B. Morosin and P. S. Peercy, Chem. Phys. Lett., 1976,40,263. J. Pottras and A. L. Beauchamp, Can. J. Chem., 1994,72,1675. (a) J-M. Jehng and I. E. Wachs, J. Phys. Chem., 1991, 95, 7373; (b) F. D. Hardcastle and I. E. Wachs, Solid State Ionics, 1991, 45,201. L. G. Hubert-Pfalzgraf, M. Postel and J. G. Riess, Comprehensive Coordination Chemistry, ed. G. Wilkinson, R. D. Gillard and J. A. McCleverty, Pergamon, Oxford, 1987, vol. 3, ch. 34, p. 627. F. Fairbrother, The Chemistry of Niobium and Tantalum, Elsevier, London, 1967, p. 37. G. Jander and D. Ertel, J. Inorg. Nucl. Chem., 1960, 14,71. I. Lindqvist, Ark. Kemi, 1952,5,47;Chem. Abstr., 1953,47,7284. J-M. Jehng and I. E. Wachs, J. Raman Spectrosc., 1991,22,83. 30 H. Lund, H. Doupeux, M. A. Michel, G. Mousset and J. Simonet, Electrochim Acta, 1974,19,629. Paper 51055546;Received 21st August, 1995 192 J. Mater. Chem., 1996, 6(2), 187-192
ISSN:0959-9428
DOI:10.1039/JM9960600187
出版商:RSC
年代:1996
数据来源: RSC
|
12. |
LiNi1–yCoyO2positive electrode materials: relationships between the structure, physical properties and electrochemical behaviour |
|
Journal of Materials Chemistry,
Volume 6,
Issue 2,
1996,
Page 193-199
Ismael Saadoune,
Preview
|
PDF (641KB)
|
|
摘要:
LiNil-yCoy02positive electrode materials: rela tionships bet ween the structure, physical properties and electrochemical behaviour Ismael Saadoune" and Claude Delmas*b "Laboratoire de Chimie du Solide Minkral, Universitk Cadi-Ayyad, Facultk des Sciences-Semlalia, B.P S15, Marrackech, Maroc bInstitut de Chime de la Matisre Condenske de Bordeaux and Ecole Nationale Supkrieure de Chimie et Physique de Bordeaux, Av. Dr. A. Schweitzer, 33608 Pessac Cedex, France The LiNi, -,Co,Oz oxides crystallize in the rhombohedra1 system with a-NaFeO, type structure. A structural study of these materials shows the presence of a few nickel ions in the Li plane for y <0.2. This phenomenon has been confirmed by a study of the magnetic and electrical properties. The LiNil -,Co,O2 lamellar oxides are used as positive electrode materials in Li batteries. The LilILi,Ni, -,Co,Oz cells exhibit electrochemical behaviour strongly related to the existence of structural defects.A high specific energy (close to 500 W h kg-' of active material) is obtained for y=O.3. These performances are attractive from an applications point of view. Considerable attention has been paid, in recent years, to the chemical and physical properties of insertion compounds, especially to the layered oxides, because of their potential applications as positive electrode materials in high-energy- density lithium batteries.'-, LiNiO, and LiCoO, oxides have been studied more intensively since these materials (and the LiMn20, spinel) are the only phases presently known to intercalate lithium reversibly at high cell voltage (in the 3.5-4.0 V potential range).5.6 Attempting to select a very attractive positive electrode, we have investigated the LiNiO,-LiCoO, system.Nevertheless, the structural charac- terization of the LiNil-,Co,02 solid solution has shown that some cationic distribution rearrangements can occur when the cobalt amount varies in the 0 <y <1 composition domain. This paper details the relationship between the crys-tallographic data and the physico-chemical properties of these materials and shows the importance of this study in understanding the electrochemical behaviour of the LilILi,Ni, ~,Co,O, cells. Experimental The LiNil-,Co,O, solid solution is prepared, as outlined in ref.7, by solid-state reaction between stoichiometric amounts of Li2C03, NiO and Co304 oxides. The pellets are first heated for 20 h at 500 "C and 24 h at 800 "C under dry oxygen. After reaction, the samples are allowed to cool slowly to room temperature, still under oxygen. Particular attention must be paid to these syntheses, especially to the nickel-rich phases (O<y<O.2), since the products are rather prone to present a lithium deficiency compensated by an excess of nickel, as previously shown for lithium ni~kelate.~" Chemical analyses of Li, Ni and Co have been performed by atomic absorption spectrometry. The magnetic susceptibility measurements are performed according to the Faraday method, using a DSM 5 (Manics) susceptometer, between 800 and 4K.The electrical conduc- tivity measurements were carried out using the four-probes technique developed by Laplume." The LiNil-,Coy0, phases are tested by using an electro- chemical cell which includes a compressed mixture of LiNi, -,Co,Oz active materials (80 wt.%) and carbon black (20 wt.%) as the positive electrode, lithium metal as the negative electrode and a glass paper as the separator soaked in a solution of 1mol 1-l LiClO, in propylene carbonate which plays the r6le of electrolyte. The cells are cycled galvano- statically in the 3-4 V potential range (the upper-limit voltage was chosen in order to avoid any electrolyte degradation). Results and Discussion A pure LiNil-,Co,O, phase is obtained over the whole composition range (0<y <1).The X-ray data could be indexed with the a-NaFeO, type structure in the trigonal space group Rjm.Lithium ions are in octahedral sites between (Nil -,CoyOZ),, infinite slabs formed by edge-sharing Nil -,C0,06 octahedra. The evolution of the hexagonal cell parameters and the c/u ratio of the LiNi, -,Co,O, oxides are plotted us. y in Fig. 1. The decrease of the metal-metal intrasheet distance, given by the ah,, value, with y is due to the difference iq size betweenothe trivalent nickel and cobalt ions (rNilll=0.56 A, rcom =0.53 A)." Nevertheless, the increase of the (Nil -,CO,O~)~ sheets covalency as y increases explains the decrease of the metal-metal interlayer distance which is equal to chex/3.In the case of the unsubstituted lithium nickel oxide the non-stoichoimetry, due to the excess of nickel (Lil-zNil+z02) is well known after the work of Goodenough et u1.' Note that the substitution of cobalt for nickel increases the two-dimensional (2D) character of the structure (c/u=2& for a cubic lattice). We have recently shown" that for ycO.3, the true formula of the LiNil-,Co,O, phases is (Li, -ZNi,)(Nil -,Co,)O, with a small amount of extra nickel ions (z) in the interslab space. This phenomenon is also related to the synthesis conditions (temperature and nature of the starting materials). In that study, where the experimental conditions were optimized to obtain 2D materials, the Rietveld refinements show that for y=O and 0.05, z=O.O4, whereas it becomes equal to zero for y=0.3," in good agreement with chemical analyses [in all cases the 0/(Li +Ni +Co) atomic ratio has been found to be equal to 11.Thus, the structure of the L~N~,,CO,~~O,lamellar oxide is completely ordered with only lithium ions in the interslab space. Therefore, an important modification of the physical and/or electrochemical properties of the LiNi1-,Co,02 solid solution can be expected when the cobalt amount (y) increases. Magnetic characterization The magnetic properties of layered A,NiO, materials are difficult to describe from a general viewpoint. For NaNi024 J. Muter. Chem., 1996, 6(2), 193-199 193 Table 1 Experimental and calculated Curie constants for the Li,Ni, -,Co,O, phases Y cexp.Ccalc. s-o 4. 9 e 14-4 F . . . 14.0 I 2. 95 2. 9 a: a * b .2. 85 2-2-e7s e 0.0 0.2 0.4 0.6 0.9 1.0 y in UNi,-yCoy02 Fig. 1 Hexagonal cell-parameter variation vs. y of the LiNil -,Co,O, phases (OGyGl) and Ni(OH)2,13 the spins are ferromagnetically coupled within the slabs, leading to a metamagnetic behaviour. In the case of lithium nickelate, a frustrated antiferromagnetic coupling within the triangular lattice has been rep~rted.'~.'~ Moreover, the presence of extra nickel ions in the lithium plane leads to new magnetic interactions which can be monitored thanks to the partial substitution of cobalt for nickel. In this study, as all materials have been obtained under experimental conditions that are not as 'soft' as those utilized for the structural analysis reported elsewhere,12 we can expect a larger amount of nickel in the lithium plane.According to the magnetic behaviour of the LiNi, -,Co,O2 phases, two important domains can be distinguished. 0.3 <y 6 1.0. Plots of the inverse molar magnetic susceptibil- ity, xrn-', DS. temperature of the LiNi1-,Co,02 (y=0.3, 0.5, 0.6, 0.7) oxides are shown in Fig. 2. For each composition, the data could be fitted according to the Curie-Weiss law; positive Curie temperatures show the predominance of the 90" Ni-0-Ni nearest-neighbour ferromagnetic interactions within the (Nil -,Coy02),, layers. The fit of the xrn-'=f(T) curves in the linear domains shows clearly that the Ni"' and Co"' ions adopt (t2)6(e)1 and (t2)'(e)' low-spin configurations, respectively.' The experimental Curie constants agree with the calculated ones as shown in Table 1.Note that the slight deviation from linearity of the xrn-' =f(T) curve for the LiNi0.3C00.702 phase is due to the TIP contri- bution since the xrn values are slightly weaker than for other compositions. For the LiNi0.2C00.802 and L~N~,.,CO~.~O~ phases, the 194 J. Muter. Chem., 1996, 6(2), 193-199 0.05 0.85 0.36 0.10 0.84 0.34 0.15 0.86 0.32 0.30 0.39 0.26 0.50 0.32 0.23 0.60 0.18 0.15 0.70 0.14 0.11 magnetic behaviour (not reported here) is very complex and not yet understood. 0 <y <0.2. The magnetic behaviour of the LiNi02 nickelate is similar to that reported by Bongerd6 and Hirakawa et In the high-temperature domain, Curie-Weiss behaviour is followed by a negative Curie temperature (O,= -70 K).At decreasing temperature (280 K), the magnetic susceptibility becomes infinite. As shown in Fig. 3, similar behaviour has been observed for the other Ni-rich phases (y =0.05, 0.10, 0.15). However, the temperature for which the molar suscepti- bility takes an infinite value decreases as the amount of cobalt ions increases and becomes close to zero for the LiNio&o,.,Oz composition. Furthermore, the shape of the reciprocal suscepti- biltiy us. temperature curve, for yG0.15, is typical of ferrimag- netic behaviour. The extrapolation of the linear part of the x-' us.T curves leads to Curie contants significantly higher than the expected values (Table 1). Nevertheless, note that the temperature domains employed are certainly too narrow to allow a convenient fit. The magnetic behaviour can be discussed on the basis of the crystallographic results. We have shown, for the nickel- rich phases (y<0.2), the existence of extra nickel ions in the interslab space. A nickel ion occupying a lithium site interacts with six nickel ions in each of the (Nil-yC~y02)n slabs. Three of these interactions are antiferromagnetic ( 180" Ni-0-Ni), while the three others are ferromagnetic (90" Ni-0-Ni). As the orbital overlap is greater for the 180" Ni-0-Ni interactions than for the 90" Ni-0-Ni ones, the spin of the extra nickel ion tends to be antiferromagnetically coupled with those 'of the (Nil -,Co,O2), slabs, as shown in Fig.4.The antiferromag- netic character of this coupling is independent of the central nickel oxidation state. However, its magnitude is more import- ant in the case of Ni2+ ions" and tends to cause ferromagnetic ordering of the nickel ions in the two adjacent (Nil-,Coy02), slabs. This leads to the formation of small ferromagnetic islands in the vicinity of the extra nickel ions, giving rise to the apparent ferrimagnetic behaviour. The substitution of cobalt ions for nickel, which decreases the concentration of extra nickel ions in the Li plane, causes a reduction in the ferrimagnetic behaviour, which disappears for y >0.2. Electrical conductivity Thermal variations of the logarithm of the electrical conduc- tivity for some LiNil-,Coy02 phases are plotted in Fig.5 and 6. Over the entire composition range, conduction is thermally activated owing to the semi-conducting character of these oxides. For the nickel-rich phases (y <0.2), the activation energy varies between 0.12 and 0.14eV. These values are typical for small polaron conduction in a mixed-valent system. In fact, the (Li, -zNiz)(Nil -,Co,)O2 formula imposes 2zNi2 and+ ( 1-y -z) Ni"' per formula unit for 0 <y <0.2, which allows for facile transfer of electrons uia hopping. As the concentration of diamagnetic Co"' ions increases, the electrical conductivity decreases and the activation energy lsoo.o 1500.0 b 1000.0 1000.0 500.0 500.0 L -......... -'E r H -lSOO.0 1soo. 0 -1000.0 1000.0 .-500.0 ..-*I.......I......... 500.000 0 i 0.0 100.0 200.0 300.0 0.0 100.0 200.0 300.0 TIK Fig. 2 Thermal evolution of the inverse molar magnetic susceptibility of LiNi, -,Co,Oz phases. y=: (a)0.3; (b)0.5; (c) 0.6;(d)0.7. --lsoo.o 1soo. 0 ..*a-........... -.......1000.0 1000.0 -500.0 500.0 7 ....,-..: ... 1 .... I .... JL 0.0 0.0 ' 'I--0.0 200.0 400.0 600.0 800.0 0.0 200.0 400.0 eoo.0 8~0.05P 7'EH lS00,O 1500.0I E (d) 1000.0 1000.011soo.0 500.0 0.0 ..... , .l....l.... 0.0 0.0 200.0 400.0 eoao eoo.0 0-0 200.0 400.0 600.0 800.0 TIK Fig. 3 Thermal evolution of the inverse molar magnetic susceptibility of LiNi,-,Co,O, phases.y=: (a)0.05; (b)0.10; (c) 0.15; (d)0.20. J. Mater. Chem., 1996, 6(2),193-199 195 Lithium plane 3+ ( N4-,,Coy In slab Ni Fig. 4 Representation of the magnetic couplings between one nickel ion present in an Li plane with nickel ions belonging to the (Nil -,,Co,02), slab. The spin orientation is given to illustrate the ferro- and antiferro-magnetic coupling; it has no absolute meaning. 0.0 r -1.0 h F IG fe -2.0 r b Y -t? -3.0 : --4.0 . -5. 0 3.0 4. 0 5.0 6. 0 7. 0 8-0 lo3 WT Fig.5 Thermal variation of the log (electrical conductivity) for LiNil-,Co,02 phases. y=: (a) 0; (b)0.05;(c) 0.15.-0.0 -..'.-2.0 .. h -I ...-. -... --... *5 e -b--4.0Y 0,0 Fig. 6 Thermal variation of log (electrical conductivity) for LiNi,-,Co,02 phases. y=: (a) 0.15; (b)0.3; (c) 0.6; (d)0.7; (e)0.9. increases (E,=0.16 eV). This evolution is due essentially to the decrease in concentration of Ni"' which is the only active ion in the conduction process. Moreover, as only trivalent nickel ions are present in these materials, the hopping phenomenon is impeded. In lithium-deficient material, the presence of extra nickel ions in the interslab space must also play an important role, as the electrical conductivity is not restricted to the slabs but can occur parallel to the c axis. Electrochemistry Fig. 7 and 8 show the room-temperature cycling behaviour of the LiIILiNil -,Co,O, electrochemical cells under constant current.These experiments have been performed at low rates in order to emphasize the relationship between the cationic distribution and the reversibility. Ni-rich phases (0 Gy G0.5). As shown in Fig. 7, the shape of the charge-discharge curves, during the first cycle, shows very good reversibility of the intercalation-deintercalation process in these lamellar oxides. Furthermore, the polarization (differ- ence in voltage between the charge and discharge curves) is very low in this composition domain (lOmV). Moreover, it decreases continuously for 0 < y < 0.3 as y increases as a result of the fast Li' diffusion in the interslab space. Nevertheless, the increase of the polarization for y =0.4 and 0.5 is due mainly to the increasingly insulating character of the LiNil -,Co,O2 materials when y increases.Note that the potential-composition curves are very similar for O<y<O.5 showing that when cobalt is substituted for nickel, the Ni4+ /Ni3 couple remains the only active redox + couple in the electrochemical process. This result is in accord with our previous workg which shows that the Ni"' ions are preferentially oxidized to the tetravalent state in comparision with the Co"' ions. In good accord with the results of Ohzuku et ~l.,~'note that the LillLi,Ni, -,Co,O, (0< y G 0.5) cells exhibit very promising cycling properties. For the L~,N~,~,CO,~~O,system, 500 W h kg-l of active material have been recovered at the 50th cycle under 400~Acm-~which corresponds in this experiment to the C/35 regime.,' More recently, cycling has been performed, with thin electrodes, at the C/5 rate without significant loss of capacity.These performances explain why several battery companies (Moli Energy Ltd., Sony Energytec Inc., Saft) have recently announced the future commer-cialization of 'rocking chair' batteries using some of the LiNi, -,Co,O2 phases as positive electrode rnaterial~.~l-~~ Co-rich phases (0.6Gy 6 1.0). In this composition domain, the voltage of the Li~~Li,Nil-,Co,02 cells increases as y increases according to the high oxidizing power of the Co4 +-Co3 redox couple. Thus, the high-voltage limit imposed + in our experiments (4 V) is rapidly reached, and an important decrease of the cell capacity follows.At the beginning of the charge (deintercalation process), the LiIILi,Ni, -,Co,O2 ( y =0.9 and 0.8) electrochemical curves are very similar to those of the nickel-rich phases. However, for x<0.9 and 0.8, respectively, the curves become very close to that of the LillLi,CoO, cell. This evolution confirms that the Ni"' ions are oxidized before the cobalt ions. Relationship between the cationic distribution and the electrochemical behaviour The aptitude of Li generators to deliver a high current rate is related essentially to the ease of Li' diffusion within the host matrix, which acts as a positive electrode. Previous crystallographic characterization of the LiNi1-,Co,O2 solid solution has shown the presence of a few nickel ions in the Li planes for yGO.2, whereas for y>O.2, these lamellar oxides exhibit an ideal a-NaFe0,-type structure with only lithium ions in the interslab space.The presence of the nickel ions in the Li' planes for yG0.2 increases the energetic barrier necessary for Li migration from one crystal- + lographic site to another as the nickel-lithium electrostatic repulsion is strong. In order to elucidate the relationship between the electrochemical behaviour of Lil ILi,Ni, -,Co,O2 cells and the cationic distribution in the interslab space, the Li~~Li,Nil-,Co,02(y=O, 0.05, 0.30) cells were tested: (1) by continuous charging up to a composition of L&Ni1 -,CoYO2, under a constant current density of 70 pA cmP2, followed by (2) discharging for various periods, under the same current rate, interrupted with periods of relaxation.The open-circuit 196 J. Muter. Chem., 1996, 6(2), 193-199 -4.5 Li I I LixNiO, -Li I I L~Ni,,Co,, 0,1 4. 0 4.0 -3.s 3.9 t -3.0 3.0 i 2. 5 2.5 l 0 0.4 0.5 0.6 0.7 0.8 0.9 1.00 Li I I LixNi,,Coo.,024-J F4. 0 g 3.5 3.0 ?2.5 1. 4 0.5 0.8 0.7 0.8 0.a 1.0 0 4.5 4.5 Li I I LixNio~,Coo,,02 --4.0 4.0 --3. 5 3.5 L -3.0 3-0 Fig. 7 Cycling behaviour of the LillLi,Ni, -,Co,O, (0< y < 0.5) electrochemical cells (j= 70 yA cm-'; C/200) voltage is considered to be reached when the relaxation voltage is stable within 0.1 mV h-l.The voltage-composition curves of the three batteries are given in Fig. 9. For y=O and 0.05, a voltage quasi-plateau is observed at the end of the discharge (0.8 < x d 1.0). This would indicate the existence of a two-phase system. In contrast, the voltage+omposition curve of the LiJJLi,Nio.,Coo.302 generator exhibits a monotonic variation which is typical of the presence of a solid solution in the 0.6 < x < 1.0 composition range. The difference in behaviour of these batteries can be dis- cussed in terms of the crystallography of Li,Ni1-,Co,02 (y= 0,0.05,0.3) positive electrode materials. For the very deinterca- lated phases (x < 0.8), the high concentration of Li' vacancies coupled with the wide interslab distance (due to the loss of the slab lattice cohesion) are favourable for fast Li' diffusion in the interslab space.Thus, for x<O.8, the polarization, gven in this case by the height of the vertical lines in the V=f(x) curves, remains very small for the three batteries. As the Li concentration in Li,Ni, -,Co,02 ( y = 0, 0.05) increases, the probability for one intercalated lithium ion to be present near the nickel atom in the interslab space, increases. Thus, for O.S<x<l, the increase in the polarization of the LiJJLi,Ni, -,Co,O, ( y = 0, 0.05) cells results from the nickel- lithium electrostatic repulsion. In this lithium composition range, the lithium ions have two environments in the interslab plane: one, formed only by lithium ions, and another formed by lithium and nickel ions.Therefore, the intercalation voltage reflects the nature of the site occupied by the lithium ions during the intercalation process. Although the material is a solid solution from a general structural point of view (X-ray), from an electrochemical point of view, it behaves as a two- phase domain where lithium ions occupy the sites close to an J. Mater. Chem., 1996, 6(2), 193-199 197 -4.5 Li II Li,CoO, -4.0 - 3.5 - 3.0 2.5 I...,l.I..lI,,lllllII.1II1 - 4.0 $3.5 - - 3.0 0-42. 5 0. 6 0. 8 1. 0 LiIIL~Ni,,Coo,02 2. 50.4 0.5 0.6 0.7 0.8 0.9 1.0 X Fig. 8 Cycling behaviour of the LillLixNil -,Co,Oz (0.8<y <1) electro-chemical cells (j=70 pA cm-2; C/200) extra nickel ion.The absence of extra nickel ions in the interslab space explains the monotonic variation of the L~~~L~,N~,~,CO~~~O~cell voltage over the entire composition domain. Conclusion The substitution of cobalt for nickel in lithium nickelate gradually increases the 2D character of the structure, which becomes strictly layered for the LiNi,~,Co,.,02 composition. The presence of extra nickel ions in the interslab space, which is particularly evident in the magnetic study, makes difficult the lithium reintercalation at the end of cell discharge. The substitution of diamagnetic cobalt ions [( t2)6] for nickel ones [( t2)6(e)'] tends to decrease the electronic conductivity. Nevertheless, this effect is important only for y 20.6. Therefore, it seems that the L~N~,.,CO,.~O~ phase is the best positive 198 J.Mater. Chew., 1996, 6(2),193-199 4-0 f 2. 5 0.5 0.6 0. 7 0.R 0.9 1.0 x in Li,NiO, J.o F -3.5 !s -3. 0 L 11 I2. 5 AL.l.-.l&, 1 I I1 L I. 1 I J 0.5 0.6 0-7 0.8 0.9 1.CI x in Li~NiO~95Coo.o,02 2. 50.5 0.6 0. 7 0-8 0. 9 1.0 x in Li,Ni,,Co,~,O, Fig. 9 Potential-composition curves of the LillLi,Ni, -?Co,O, gener-ators (discharge with relaxation; the relaxation period is interrrupted when (dV/dt)<O.l mV h-'. y=: (a)0.0; (b) 0.05;(c)0.30 electrode material from an electrochemical cycling point of view. Moreover, as only nickel ions are involved in the electrochemical process, the cell voltage is very close to that of LiNi02 avoiding the electrolyte decomposition problems which appear for cobalt-rich materials like LiCo02.The authors thank A. Rougier and M. MCnCtrier for fruitful discussions and CNES for its financial support. References 1 C. Delmas, in Chemical Physics of Intercalation, ed. A. P. Legrand and S. Flandrois, NATO AS1 Series, 1987, vol. 172, p. 209. 2 3 4 5 6 7 8 9 10 11 12 13 14 T. A. Hewston and B. L. Chamberland, J. Phys. Chem. Solids, 1987, 48,97. J. P. Kemp and P. A. Cox, J. Phys.: Condens. Matter, 1990,2,9653. P. F. Bongers and U. Enz, J. Solid State Commun., 1966,4, 153. J. R. Dahn, U. Von Sacken and C. A. Michael, Solid State Ionics, 1990,44, 87. J. B. Goodenough, K.Mizushima and T. Takeda, Jpn. J. Appl. Phys., 1990,19,305 (Suppl. 19-3). C. Delmas and I. Saadoune, Solid State Ionics, 1992,53-56, 370. L. D. Dyer, B. S. Borie, jr. and G. P. Smith, J. Am. Chem. Soc., 1954,76,1499. J. B. Goodenough, D. G. Wickham and W. J. Croft, J. Phys. Chem. Solids, 1958, 5, 107. L. Laplume, LOnde Electrique, 1955,335, 113. R. D. Shannon and G. T. Prewitt, Acta Crystallogr., Sect. B, 1969, 25, 925. A. Rougier, I. Saadoune, P. Gravereau, P. Willmann and C. Delmas, Solid State Ionics, in press. H. Miyamoto, Bull. Inst. Chem. Res., Kyoto Univ., 1990,44, 87. J. N. Reimers, J. R. Dahn, J. E. Greedan, C. V. Stager, G. Liu, I. Davidson and U Von Sacken, J. Solid State Chem., 1993, 102, 542. 15 P. Ganguly, V. Ramaswamy, I. S. Mulla, R. F. Shinde, P. P. Bakare, S. Ganapathy, P. R. Rajamohanan and N. V. K. Prakash, Phys. Rev. B, 1992,46,11595. 16 P. F. Bongers, PhD Dissertation, The University of Leiden, The Netherlands, 1957. 17 K. Hirakawa, H. Kadowaki and K. Ubukoshi, J. Phys. SOC.Jpn., 1985,54,3526. 18 G. Dutta, A. Manthiram, J. C. Grenier and J. B. Goodenough, J. Solid State Chem., 1992,96, 123. 19 T. Ohzuku, A. Ueda, M. Nagayama, Y. Iwakoshi and H. Komori, Electrochim. Acta, 1993,38, 1159. 20 C. Delmas, I. Saadoune and A. Rougier, J. Power Sources, 1993, 43-44,595. 21 A. Lecerf, M. Broussely and J. P. Gabano, Eur. Pat. Appl., 89110158,1989. 22 T. Nagaura, M. Broussely and J. P. Gabano, 4th International Rechargeable Battery Seminar, Deerfield Beach, FL, 1990. 23 Y. Nishi, H. Azuma and A. Omaru, US Pat., 4 959 281,1990. Paper 51026671; Received 26th April, 1995 J. Muter. Chem., 1996, 6(2), 193-199 199
ISSN:0959-9428
DOI:10.1039/JM9960600193
出版商:RSC
年代:1996
数据来源: RSC
|
13. |
Structural and electronic properties of the compounds Ba4Ir3MO12(M = Li, Na, Mg, Y, Lu, Zr and Ce) |
|
Journal of Materials Chemistry,
Volume 6,
Issue 2,
1996,
Page 201-206
Jonathan G. Gore,
Preview
|
PDF (741KB)
|
|
摘要:
Structural and electronic properties of the compounds Ba4Ir3MOI2 (M =Li, Na, Mg, Y, Lu, Zr and Ce) Jonathan G. Gore and Peter D. Battle* Inorganic Chemistry Laboratory, South Parks Road, Oxford, UK OX1 3QR Polycrystalline samples of the Irlv/Irv compounds Ba41r3MO12 (M =Li, Na, Mg, Y, Lu, Zr and Ce) have been prepared using standard ceramic techniques. The products have been characterized by X-ray powder diffraction, SQUID magnetometry (6< T/K <296) and electrical conductivity measurements 80< T/K <296). They all adopt the 6H perovskite structure with the cation M occupying only the vertex-sharing six-coordinate sites and with iridium occupying the face-sharing sites and the remaining 25% of the vertex-sharing sites. These compounds are non-metallic and have low magnetic susceptibilities with only a relatively small temperature dependence, suggesting that extensive spin-pairing occurs throughout the measured temperature range.Recently, the mixed metal oxides of iridium have attracted much intere~t.'~~ Many of these compounds are related struc- turally to the mineral perovskite. They have the general formula A21rB'06 and the cations Ir and B' occupy the six- coordinate sites of the perovskite structure in an ordered, 1: 1 manner. The iridium cations can be formally described as Ir1v,4,8*9IrV or Irv1,5,7 depending on the valencies of A and B'. Compounds having a 2 : 1 Ir :B' ratio (A31r2B'09) are also known to contain IrIV and IrV.10 Their structure is usually that of an ordered 6H hexagonal perovskite with Ir in the face- sharing octahedra and B' in the corner-sharing octahedra.Perovskite-like materials with a fully ordered arrangement of two different elements in a 3: 1 ratio on the six-coordinate sites have been reported previously, but none of them contained iridium. The structure of Ba4Sb,M'O12, where M'= Li, Na," is an example of 3 :1ordering in a cubic perovskite [Fig. l(u)], whereas Ba4Ru3NaO12 adopts the hexagonal 8H perovskite structure [Fig. l(b)] in which the sodium and ruthenium ions are fully ordered over the face-sharing and corner-sharing octahedral sites.I2 These two 3 : 1structure types are interesting for different reasons. The sublattice containing the majority- cation in the ordered cubic structure comprises an infinite network of corner-sharing octahedra, and we believe that if this sublattice is occupied by an appropriate cation, i.e.one having a partially filled d shell, then metallic conductivity may result. The 8H structure is interesting, firstly because it occurs so rarely compared to other perovskite polytypes, and secondly because the degree of cation ordering is variable, as can be seen by comparing the structure of Ba4Ru3NaO12 with that of Ba4Ta3LiO12.13 In this paper we describe the synthesis and characterisation of a number of compounds with the general formula Ba41r3MO12 (M =Li, Na, Mg, Y, Lu, Zr and Ce). We hoped that the introduction of Ir in a 3: 1 ratio with some other cation might stabilise the fully ordered 8H structure, and that a consideration of the relative size and charge of Ir and the other species would then enable us to identify the factors necessary for the formation of this structure.We hoped that those compositions that did not adopt the 8H structure would adopt an ordered cubic structure, and that they would show a metallic conductivity. The synthesis of an extensive series of compounds Ba41r3MO12, in which the mean oxidation state of Ir varies from +5 for M=Li and Na to +4.66 for M= Mg, +4.33 for M=Y and Lu, and +4 for M=Zr and Ce, enabled us to determine the electronic and structural properties of these materials as a function of d-electron concentration. The results described below will show that, although all our hopes were not realised, these compounds do behave in an interesting manner.Experimenta1 All materials were synthesized by firing, in air, stoichiometric amounts of iridium metal and the appropiate oxides and carbonates which had been ground together in an agate mortar. During the reactions, which took several days to complete, the reactants were contained in alumina crucibles. Starting mate- rials were spectroscopic grade (Johnson Matthey) BaCO,, iridium metal, Li,CO, ,Na,C03, MgO, Y203, La20,, Nd203, Lu203, Zr02 and CeO,. The final firing temperatures are shown in Table 1. Reaction progress was monitored by period- ically recording the X-ray powder diffraction pattern of the mixture, which was then reground, pelletised and returned to the furnace.Reactions were deemed to be complete when successive powder patterns were unchanged. The X-ray diffrac- tion data used in the structural analyses of the oxides were collected using a Siemens D5000 diffractometer operating in Bragg-Brentano geometry with a 28 step size of 0.02'. Where possible, the oxidation state of iridium was determined by iodometric titration and, if complete reduction under hydrogen at 900 'C was feasible, thermogravimetry was used to determine the oxygen content of the compounds, thus giving an indepen- dent determination of the oxidation state of iridium. Magnetic susceptibility measurements on powdered samples were made using a Cryogenic Consultants SCU500 SQUID suscepto- meter. Samples were loaded at room temperature and measure- ments were taken in the temperature range 6-296K, both after cooling the sample in zero-field (zfc) and after cooling in the measuring field (fc).Field strengths in the range 5<H/kG < 10were used. Electrical conductivity measurements were made using the four-contact dc technique on ingots (approximately 5 mm x 3 mm x 1.5 mm) cut from sintered pel- lets of the oxides. Colloidal silver paint was used to make Table 1 Synthesis temperatures for Ba41r3M0,, ~~ compound T/"C Ba,Ir,LiO 12 650 Ba,Ir,NaO 725 Ba41r3Mg012 1025 Ba4Ir,YOI2 1180 Ba41r,LuO12 1180 Ba,Ir3ZrO12 1260 Ba,Ir3CeO12 1260 J. Muter. Chem., 1996,6(2), 201-206 201 w --Fig. 1 3 : 1 ordering in (a) the cubic perovskite Ba4Sb3MOlz, M =Li, Na (open octahedra contain M atoms and hatched octahedra contain Sb atoms) and (b) the 8H perovskite Ba4Ru3NaOl, (lightly hatched octahedra contain Na atoms and hatched octahedra contain Ru atoms).In both cases open circles are Ba atoms. electrical contact with the ingots, the separation of the voltage contacts being approximately 2 mm. The dc resistivities of the samples were measured in the temperature range 77-298K using an HP3478A autoranging multimeter. Results Our attempts to prepare Ba41r,M01, (M =Li, Na, Mg, Y, Lu, Zr and Ce) resulted in black powders, the X-ray diffraction patterns of which could all be indexed in the hexagonal space group P63/rnrnc, with unit-cell parameters characteristic of the 6H perovskite structure (Fig.2). The compositions determined by chemical analysis and the refined unit-cell parameters are summarised in Table 2. We were unable to prepare monophasic samples for M =La or Nd. Profile analy~is,'~ using the GSAS pa~kage,'~of the X-ray data collected on our monophasic products confirmed the crystal structure assignment and sug- gested that in all cases the face-sharing octahedra are occupied solely by Ir, and that the corner-sharing octahedra are occupied statistically by Ir and M (25 :75 ratio of Ir :M). The atomic positions in the 6H structure are listed in Table 3. The param- 202 J. Mater. Chern., 1996,6(2), 201-206 Fig. 2 6H Perovskite structure of Ba41r3MOlz, M =Li, Na, Mg, Y, Lu, Zr and Ce. The face-sharing octahedra are occupied solely by Ir atoms.The vertex-sharing octahedra are occupied statistically by the remaining Ir and the M atoms in a 25 :75 ratio. Circles are Ba atoms. Table 2 Analytical data and unit-cell parameters for Ba41r3MO12 unit-cell parameters average oxidation state compound of iridium a/A CIA Ba41r3Li0,, + 5.00( 5)" 5.8046( 1) 14.3986(4) Ba41r3NaOlz + 5.03( 5)" 5.8702(2) 14.6079(5) Ba41r3Mg012 +4.66( l)b 5.7847(2) 14.3587(5) Ba41r3YOl, + 4.35( 2)" 5.8829(6) 14.260( 1) Ba41r3Lu0, +4.33(3)" 5.8582(3) 14.5471(8)Ba41r3ZrOlz -C 5.80808(8) 14.4318(2) Ba41r3CeOlz -C 5.89968( 5) 14.7256(2) a By iodometric titration. 'By thermogravimetry. Neither was possible. Table 3 General atomic positions of Ba41r3MO12 (space group P6,fmmc) fractional atom site X Y Z occupancy eters varied in the refinements were thus a scale factor, six variables to describe the width of the pseudo-Voigt Bragg peaks, eight coefficients of the Chebyshev polynomial used to model the background level, two unit-cell parameters, five atomic coordinates and an overall isotropic temperature factor.For certain refinements it was necessary to fix the temperature factor at a value of zero to avoid physically meaningless values. It became apparent that a small impurity phase (ca. 2%) was present in the sample Ba41r3Na0,,. This was identified as BaIrO, and taken into account in the final structural analysis. The refined structural parameters for each individual com- pound are shown in Table4. Selected bond lengths derived from the refinements are given in Table 5.The results of our resistivity measurements on Ba41r,M0,, are shown in Fig. 3. The room-temperature resistivities are Table 4 Structural parameters for Ba,Ir,MO,, (space group P63/mm~) Ba,Ir,LiO,, Ba,Ir,NaO,, Ba41r3Mg012Ba,Ir,YO,, Ba41r3LuO12 Ba41r3ZrOlz Ba,Ir,CeOlz 14.7 14.9 13.4 14.1 15.5 16.1 13.4 6.7 7.4 6.3 7.0 5.2 5.8 6.7 0.9146( 1) 0.9135( 1) 0.9128( 1) 0.9076(2) 0.9091( 1) 0.9067( 1) 0.9012( 1) 0.1549( 1) 0.1566( 1) 0.1548(1) 0.1621(1) 0.1613( 1) 0.16OO( 1) 0.1638(1) 0.490( 1) 0.492(2) 0.487( 1) 0.501(2) 0.496( 1) 0.499( 1) 0.499( 1) 0.191( 1) 0.186( 1) 0.1838(9) 0.179( 1) 0.184( 1) 0.1801(9) 0.180( 1) 0.414( 6) 0.4141(9) 0.4246(6) 0.4167(8) 0.4146(7) 0.4164(5) 0.4230(6) 0" 0" 0" 0" 0" 0.13( 3) 0.05(2) " Fixed at zero, see text.Table 5 Selected bond lengths for Ba,Ir,MO,, Ba,Ir,LiO,, Ba,Ir,NaO 12 2.740( 3) 2.729( 4) 2.09( 1) 2.11(1) Ba41r3Mg012Ba,Ir,Y O,, Ba,Ir3LuOlz Ba,Ir,ZrO12 Ba4Ir,CeOlz 2.733( 3) 2.571(3) 2.580( 3) 2.598( 2) 2.539(2) 2.060( 9) 2.14( 1) 2.09( 1) 2.117( 9) 2.119( 9) a-v-0 """ I I I ?%A&I I, I I, I n 50 100 150 200 250 300 TIK Fig.3 Temperature dependence of the natural logarithm of the resistivity of Ba,Ir,MO,,, M=Li (+), Na (0),Mg (A), Y (x), Lu (+), Zr (0)and Ce (0) given in Table6. Both the temperature dependence and the magnitude of the resistivities indicate that all of the compounds considered in this study are semiconductors or insulators. Attempts to fit the data to a simple Arrhenius model were unsuccessful.Conductivity measurements made on polycrystal- line samples, albeit sintered, will be complicated by grain boundary effects and we were unable to make the thermopower measurements that would have enabled us to eliminate this contribution. We therefore do not feel justified in using a more sophisticated model to fit our data. In view of the absence of metallic conductivity we shall attempt to explain our magnetic data in terms of a localized-electron model. The results of our magnetic susceptibility measurements are shown in Fig. 4 and 5. The molar magnetic susceptibilities of Ba,Ir,MgO,,, Ba,Ir,YO,,, Ba41r3LuO12, Ba41r3ZrO12 and Table 6 Room-temperature (296 K) resistivities for Ba41r3MOlz compound resistivity/Q cm Ba,Ir3LiOlz 0.2 Ba,Ir,NaO,, 1.1 Ba41r3Mg012 25 Ba41r3Y012 49 Ba41r3LuOlz 323 Ba,Ir,ZrO,, 3185 Ba,Ir,CeO,, 874 1.79( 1) 1.82( 1) 1.883(9) 1.94( 1) 1.97( 1) 1.972( 9) 2.24( 1) 2.27( 1) 2.1 36( 9) 1.95( 1) 1.87( 1) 1.95(1) 1.961(9) 2.04( 1) 1.98(1) 2.03(1) 2.040( 9) 2.19(1) 2.25( 1) 2.13( 1) 2.214( 9) Ba41r3CeO12 [Fig.4(u)-(e)] all show qualitatively similar behaviour, with the zfc and fc curves overlying each other throughout the measured temperature range. These com-pounds were initially assumed to be paramagnetic, and the high-temperature susceptibility data were fitted using a Curie-Weiss model with an additional term, a, to allow for temperature-independent paramagnetism (TIP): C xm=-+aT-8 The high values of a and the low values of C which this model produces (Table 7) indicate that our initial assumption is likely to be invalid, and we shall discuss alternative interpretations below.The temperature dependence of the inverse molar magnetic susceptibility of Ba,Ir,LiO,,, measured in a field of 10 kG, is shown in Fig. 5. It can be seen that the zfc and fc curves diverge below 57 K. A Curie-Weiss fit to the high- temperature data (>60 K) was made but the parameters obtained appear to have meaningless values (C=3.94, 0 = -680 K). The presence of a small amount of BaIrO,, which is ferromagnetic below 170 Kf6made it impossible to interpret the magnetic susceptibility data collected on our sample of Ba,Ir,NaO12.Discussion The results of the X-ray diffraction experiments described above show that we have successfully synthesized monophasic samples of Ba41r,MO12 for M=Li, Mg, Y, Lu, Zr and Ce. Our sample of Ba,Ir3NaO12 was contaminated by a small quantity of BaIrO,, possibly as a result of sodium volatilisation during the heating process. We were able to perform a chemical analysis on all but two of the compounds (M =Zr, Ce) and the results show that our samples can be considered as fully stoichiometric. The successful X-ray analysis, coupled with the strong preference of Zr for the oxidation state +4, and the reluctance of Ir to form Ir"' suggests that Ba41r,ZrOl, is also fully stoichiometric. The unit-cell parameters of Ba,Ir,CeO,, (Table 2) are consistent with this being a stoichiometric com- pound of Ce"' and Ir", rather than Ce"' and mixed-valence Ir.Stronger evidence supporting this assignment is provided by the magnetic data on this compound, as will be discussed below. Despite the changes in the formal oxidation state of iridium and the relatively large changes in the radii of the cation M, the 6H perovskite structure is adopted by all of these J. Muter. Chem., 1996, 6(2), 201-206 203 700 600 500 ,9! 1-400 300 200 ( 200---’ 1 I I J 0 50 100 150 200 250 300 TIK Fig. 4 Inverse molar magnetic susceptibility of (a) Ba41r3MgOl,, (b) Ba,Ir3YOl,, (c) Ba,Ir3LuOl,, (d) Ba41r3ZrO12and (e) Ba41r3CeOl, as a function of temperature. The fields used were 10 kG for (a),(b),(c) and (e),and 5 kG for (d).0,zfc; +, fc. Table 7 Magnetic parameters for Ba41r3M01, (M = Mg, Y, Lu, Zr and Ce)300 compound C/emu K mol-l alemut Ba41r3Mg012 0.051 -18.9 1.8 x 10-3 Ba41r3YOl, 0.034 -3.7 1.4 x 10-3 Ba41r3LuOl, 0.056 -7.3 1.1 x 10-3 Ba41r3ZrOl, 0.034 -4.2 0.7 x Ba41r3Ce0 , 0.036 -6.0 0.4 x compounds. The preferred structure type will depend in part on the size of the cations, including A and B’. For example, Sr,Ru,MO,, (M = Li or Na) adopts a pseudo-cubic perovskite structure with a disordered arrangement of cations over the 1 ooo 50 5 100 150 00200 250 300 six-coordinate sited7 whereas Ba4Ru3Li0,, adopts the 6H perovskite structure and Ba,Ru,NaO,, adopts the 8H perovsk-TIK ite structure.I2 The stability of the hexagonal polytypes is Fig.5 Inverse molar magnetic susceptibility of Ba41r3LiO12 as a enhanced for those compounds which contain transition-metal function of temperature in a field of 10 kG. 0,zfc; +, fc. ions with partially filled d orbitals, especially second-and 204 J. Muter. Chem., 1996, 6(2), 201-206 third-row transition-metal ions, because of the possibility of metal-metal bonding in the face-sharing octahedra. In the compounds under discussion, it can be seen that the Ir-Ir distance within the Ir209 units (Table 5) is comparable to and, in some cases, shorter than the separation of 2.72 A found in elemental iridium. However, as was mentioned above, it is more difficult to predict which particular polymorph (for example, 8H or 6H) will form.In the 6H polymorph of Ba41r3MO12, 89% of the Ir cations are in face-sharing octahedra and can thus interact strongly with another Ir cation. The fraction falls to 67% on transforming to an 8H structure. However, in the 6H structure the remaining 11 % of the iridium cations have to occupy the same crystallographic site as the cations M. The stability of this disordered arrange- ment will depend on the differences in size and charge between iridium and M. We conclude that the larger number of Ir-Ir interactions ensures that the 6H structure is adopted unless the size/charge difference between M and Ir is unacceptably large. This explanation is consistent with the structural chemis- try of Ba,LiRu,012 and Ba4NaRu3012, and it would lead us to predict Ba4NaIr3012, the composition for which we were unable to prepare a pure sample, to have the least stable 6H phase in the family Ba,Ir3MO12.During the course of our work, Jung et aL6 described the results of their work on Ba,NaIr301z and Ba,LiIr,O,,. They did not observe any BaIrO, in their sample of the former compound, although careful scrutiny of their susceptibility data suggests that some might have been present. They did, however, report the forma- tion, under high oxygen pressures, of a disordered cubic phase of Ba4Ir3NaOl2 and it would be interesting to see if an 8H phase can be stabilized by treatment under intermediate oxygen pressures. It is perhaps surprising that an 8H phase did not form for M =La or Nd, both relatively large cations compared to Ir.Our failure to prepare a cubic 3 :1 ordered phase of any of the compounds studied is consistent with the idea, stated by Alonso et ul." when discussing Ba4MSb,0,, and supported by Subramanian" when discussing Ba4MBi30,,, that such a structure is most likely to be stabilized in the presence of polarizable cations. Although subject to the low precision inherent in locating the position of oxygen atoms by the analysis of X-ray powder data, the values obtained for the Ir-0 bond lengths (Table 5) seem to be consistent with previous values obtained from neutron powder diffraction experiments on iscyalent comb pounds. For example, the mean values of 1.94 A and 1.97 A determined for the Ir( 1)-0 bond length in the Ir209 dimers of Ba$Ir3LiO12 and Ba,Ir,NaO,, agree well with the value of 1.97 A for the IrV-0 bon! length in B?LaCoIrO,.' Similarly, the mean values of 2.03 A and 2.04 A for Ba,Ir,Zr0,20 and Ba,Ir,CeO,,, respectively, are close to the value of 2.05 A for the Ir'v-O bond length in ST,I~O,,'~ thus providing further evidence that Ba,Ir3CeOl, is an Ir'v/Ce'v compound.Note that the Ir-Ir distance within the Ir209 units (Table 5) shows a general (but not smooth) increase as the oxidation state of iridium increases. This may be due to an increase in the repulsive force between the two cations, which is only partially screened out by displacements of the oxide ions in the common face of the two octahedra. The increase is particularly large on passing from M=Lu, Y to M=Mg. The smaller change on passing from M =Ce, Zr to M =Y, Lu is likely to be due to a disordered distribution of the relatively small number (33%) of IrV cations over the face-sharing and corner-sharing sites.The magnetic susceptibilities of these compounds are all relatively low, and they show a very weak temperature depen- dence, as is evidenced by the low values of the Curie constant, C, listed in Table 7. This implies that the concentration of free, paramagnetic spins is very low. For example, the value of C= 0.034 found in the case of Ba,Ir3ZrOl, corresponds to the presence of only 0.1 S= 1/2 cations per formula unit in a compound which would be expected to contain three IrIV (S=1/2) cations per formula unit.This shows that these materials cannot be regarded as simple paramagnets and suggests that significant Ir-Ir spin-pairing occurs at tempera- tures as high as 296 K. This coupling could occur via direct Ir-Ir interaction or via ca. 90" Ir-0-Ir superexchange (or both) in the Ir209 units, and the value of C is low enough to suggest that some of the Ir cations in the corner-sharing dimers are coupled to the dimers viu a ca. 180" Ir-0-Ir superexchange pathway. Similar high-temperature coupling of magnetic moments in M209 dimers has been observed previously in B~,C~RU,O,,~'in which RuV ions occupy the face-sharing octatedra of the 6H perovskite structure (Ru-Ru distance is 2.65 A21). The magnetic susceptibility of this compound shows a maximum at approxim$tely 450 K.We propose that with an Ir-Ir distance of cu. 2.6A and the involvement of 5d rather than 4d orbitals, the spin coupling in Ba,Ir3ZrOl, occurs at a higher temperature and the magnetic susceptibility is therefore small and essentially temperature-independent at 296 K. Magnetic measurements at temperatures above 300 K are needed in order to confirm this proposition. The magnitude and temperature dependence of the magnetic susceptibility of Ba,Ir,CeO,, are very similar to those of Ba4Ir3ZrOl2, thus indicating that cerium is present as CeIV (non-magnetic) rather than CeIII (paramagnetic, peff~2.5pB). The behaviour of Ba,Ir3LuOI2, Ba,Ir3YO12 and Ba,Ir,MgO,, qualitatively resembles that of Ba,Ir,ZrO,, and Ba,Ir,CeO,,, and it is not unreasonable to suggest a similar cause for the apparently low moments derived using eqn.(1) (Table 7). However, it is difficult to give a more detailed or quantitative interpretation of the data on these mixed-valence compounds because the orbital contribution to the magnetic moment of IrV will be sensitive to the local symmetry of the crystallographic site it occupies, and the susceptibility will also depend on whether the Ir-Ir interaction is dominated by superexchange or metal- metal bonding. The magnetic susceptibility of Ba,Ir,LiOlz (Fig. 5) differs markedly from those of the compounds discussed above, and also from those of other compounds of IrV. For example, the ordered perovskites BaLaMgIrO6 and La,LiIrO,, which con- tain isolated IrV in cubic symmetry, have been found to have a temperature-independent susceptibility of ca.7 x lo-, emu rnol-',* a result which is consistent with IrV having a t2: electron configuration and hence a non-magnetic (j =0) ground state. The observation of hysteresis in the susceptibility of Ba,Ir3LiO12 is inconsistent with such a description, suggesting as it does that some form of magnetic transition takes place at 56 K, and that frustration is present in the system. We do not have an adequate explanation for these data, but we note that our field-cooled data are in reasonable agreement with those reported by Jung et aL6 for the same compound. It is not clear whether their data were collected after field-cooling or zero-field-cooling.We have seen evidence of magnetically active IrV previously in the 6H perovskite Ba,C01r,0~.~ This compound orders as a weak ferromagnet at ca. 110 K, and we argued, by making a comparison with the behaviour of Ba3CoSb209, that the IrV cations must be involved in the magnetic coupling. It appears that Ba,Ir3Li0,, is a second example of a 6H perovskite containing magnetic IrV. As mentioned previously, the limitations of our electrical conductivity measurements make it difficult to interpret these results in detail. However, as can be seen from Fig. 5 and Table 6 there is a systematic variation of the electrical conduc- tivity of Ba,Ir,MO,, with the average oxidation state of iridium.The compounds Ba41r3LiO12 and Ba41r3NaO12, containing IrV only, have higher conductivities than the mixed-valence compounds and the IrIv-containing compounds, Ba,Ir3ZrOl, and Ba,Ir,CeO,, , have the lowest conductivity of all. This contrasts with the observed behaviour of the 6H perovskites Ba,Ru,LiO,, and B~,RU~M~O~,,~~where the latter, a mixed-valence compound, has a higher conductivity J. Muter. Chem., 1996, 6(2), 201-206 205 than the former, a compound containing RuV only. In semicon- ducting materials there are many competing factors that will determine the electrical conductivity. In Ba4Ru3LiOI2 we believe that the high correlation energy of the d3 configuration is the dominating factor. In the Ir compounds it is possible that the principle factor determining the conductivity is the density of states at the Fermi energy in the 5d band, which is expected to be higher for the IrV (d4)compounds than for the Ir'" (d5) compounds. In conclusion, we have shown once again that the interplay of a large number of factors (size, charge, electron configur- ation, etc.)makes it difficult to prepare new mixed-metal oxides having specified structural or electronic properties.The 8H perovskite structure remains particularly elusive, despite our use of the appropriate 3: 1 Ir: M stoichiometry for a wide range of cations, M. 4 5 6 7 8 9 10 11 12 13 14 15 R. C. Currie, J. F. Vente, E. Frikkee and D. J. W. Ijdo, J. Solid State Chem., 1995,116,199. D. Y. Jung, P. Gravereau and G. Demazeau, Eur.J. Solid State Chem., 1993,30,1025. D. Y.Jung, G. Demazeau, J. Etourneau and M. A. Subramanian, Muter. Res. Bull., 1995,30, 113. D. Y. Jung and G. Demazeau, J. Solid State Chem., 1995,115,447. A. V. Powell, J. G. Gore and P. D. Battle, J. Alloys Comp., 1993, E. M. Ramos, I. Alvarez, M. L. Veiga and C. Pico, Muter. Res. Bull., 1994,29, 881. R. C. Byrne and C. W. Moller, J. Solid State Chem., 1970,2,228. J. A. Alonso, E. Mzayek and I. Rashes, Muter. Res. Bull., 1987, 22, 69. P. D. Battle and S. H. Kim, J. Solid State Chem., 1992, 101, 161. B. M. Collins, A. J. Jacobson and B. E. F. Fender, J. Solid State Chem., 1974,10,29. H. M. Rietveld, J. Appl. Crystallogr., 1969,2, 65. A. C. Larson and R. B. von-Dreele, General Structure Analysis System, Los Alamos National Laboratories, Report LAUR 201, 73. 86-748,1990. We are grateful to the EPSRC for financial support. 16 17 A. V. Powell and P. D. Battle, J. Alloys Comp., 1993,191, 313. I. S. Kim, T. Nakamura, M. Itoh and Y. Inaguma, Muter. Res. Bull., 1993,28, 1029. 18 M. A. Subramanian, J. Solid State Chem., 1994,111, 134. References 19 A. V. Powell, P. D. Battle and J. G. Gore, Acta Crystallogr., Sect. G. Demazeau, D. Y. Jung, J. P. Sanchez, E. Colineau, A. Blaisse and L. Fournes, Solid State Commun., 1993,85,479. 20 C, 1993,49,189. J. Darriet, M. Drillon, G. Villeneuve and P. Hagenmuller, J. Solid State Chem., 1976, 19,213. G. Demazeau, B. Sicherchicot, S. Matar, C. Gayet and 21 J. Wilkens and H. Muller-Buschbaum, J. Alloys Comp., 1991,177, A. Largeteau, J. Appl. Phys., 1994,75,4617. L31. P. D. Battle, J. G. Gore, R. C. Hollyman and A. V. Powell, J. Alloys Comp., 1995,218, 110. Paper 51057291; Received 30th August, 1995 206 J. Muter. Chem., 1996, 6(2), 201-206
ISSN:0959-9428
DOI:10.1039/JM9960600201
出版商:RSC
年代:1996
数据来源: RSC
|
14. |
Study of the formation mechanism of complex oxides obtained by the sol–gel method: influence of the structure of iron, aluminium and yttrium acetylacetonate precursors on the phase composition of the ZrO2ceramics |
|
Journal of Materials Chemistry,
Volume 6,
Issue 2,
1996,
Page 207-212
O. G. Ellert,
Preview
|
PDF (672KB)
|
|
摘要:
Study of the formation mechanism of complex oxides obtained by the sol-gel method: influence of the structure of iron, aluminium and yttrium acetylacetonate precursors on the phase composition of the ZrO, ceramics 0.G. Ellert,"I. A. Petrunenko,"M. V. Tsodikov,*b0.V. Bukhtenko,b D. I. Kochubey,' Yu. V. Maksimovd and A. Dominguez-Rodriguez" "Institute of General and Inorganic Chemistry of Russian Academy of Sciences (RAS), Leninskii Prospekt 31, Moscow, 117907, Russia bInstituteof Petrochemical Synthesis of RAS, Leninskii Prospekt 29, Moscow, 11 7912, Russia 'Institute of Catalysis of Academy of Sciences, Siberian Branch, Acad. Lavrent 'ev Prospekt, Novosibirsk, 630090, Russia dN.N. Semenov Institute of Chemical Physics of RAS, Kosygina Street 4, Moscow, 11 7977, Russia eDepartamento De Fisica de la Materia Condensada, Universidad de Sevilla, Apartado 1065-41 080, Sevilla, Spain To obtain new effective basic reagents for single-phase ceramic production with the use of alkoxosynthesis methods, three heterometallic iron-yttrium acetylacetonate complexes were originally synthesized by electrochemical anodic dissolution in acetylacetone.By means of X-ray diffraction (XRD), extended X-ray absorption fine structure (EXAFS), magnetic susceptibility and Mossbauer methods, the structure and properties of the new complex oxides were studied. It was shown that for the iron- containing gels and oxides, Fe3+/Y3+ ions participate in the oxide formation mechanism. The sets of interatomic distances found by means of EXAFS in the new iron-yttrium precursor and in the zirconium gel are practically equal.The presence of structure fragments formed by these bonds and especially direct Fe-Y in the precursors allow the formation of single-phase triple ceramic oxides. There is growing interest in studies of the development and structure of single-phase ceramic materials based on zirconium oxide.'-4 One of the promising advances in this field is sol-gel processing using alkoxides and other metal-organic precursors, predominantly heteronuclear alkoxide~.~~~ Unlike alcoholates, compounds of organic acid salts, P-diketonates, metallic com- plexes containing p-diketonate and alcoholate fragments are used as the starting reagents in alkoxo~yntheses.~-'' The harnessing of P-diketonate complexes is of great interest as the partial exchange of alkoxy groups for chelate groups significantly improves the stability of compounds towards hydrolysis and is favourable for the formation of the products, which are homogeneous in composition, structure and mor- ph~logy.~-'' In this connection, the idea of using of heteronu- clear acetyl acetonates with known structures as the basic reagents seems to be promising.In this paper we discuss the structure and magnetic proper- ties of the newly synthesized iron-yttrium acetylacetonates for the production of ceramic zirconium oxide. We also attempt to analyse the influence of the nature and the structure of the mononuclear, in comparison with heteronuclear, acetylaceton- ates of yttrium, iron and alumina as precursors on the forma- tion mechanism and structure of the complex oxides obtained by the heat treatment of the products of alkoxosyntheses with zirconium isopropylate.Experimenta1 Synthesis of heteronuclear B-diketonate precursors The syntheses of the organometallic precursors were carried out in a 500cm3 electrolyser, equipped with a reversed con- denser with a calcium chloride tube and two cylindric iron- yttrium electrodes (d= 10 nm). Alloy electrodes of various compositions were used: Y :Fe = 1:1 (synthesis 1, precursor 1) and Y:Fe= 1:2 (synthesis 2, precursors 2, 3). The latter electrode is an intermetallide of known composition.12 The electrochemical anodic dissolution was conducted in absolute acetylacetone (Hacac) mixed with background solvent aceto- nitrile (AcN) (Hacac-AcN =1:1).The dissolution was per- formed using an ac current (I =0.1 A, V= 120V) over 18 h. As the current passed through, tetraethylammonium bromide (Et4NBr) was added. The current was then switched off, and the organic solution was filtered and thermostatted at 4-5 "C for 48 h. The mixture of complex 1, Fe(acac), and the resin product was precipitated from this solution. Most of the Fe(acac), and resin was flushed out from complex 1 with warm hexane (50 "C) until decolourization of the mixture was observed. The remaining trace amount of Fe(acac), was flushed out from the solid product with a warm mixture of hexane-acetone (100:1).Then the solid complex 1 was recrystallized from hexane-acetone (6: 1). The mother liquor was evaporated off and complex 1 was precipitated from the remaining solid according to ref. 13. The precipitated compounds were dried in vacuum at a pressure of 1Torr. Elemental analysis indicated that the obtained polycrystal- line powder consisted of Fe, 3.6; Y, 22.2; C, 43.5; H, 6.0; and 0,24.5 mass%. This composition corresponds to the empirical formula FeY4C60Hlo@25. Two iron-yttrium compounds were separated from the products of the synthesis 2. Complex 2 was obtained by precipitation from the organic solution at 4-5°C for 48 h. Complex 3 was obtained by evaporation of the mother liquor. Fe(acac), and resin were removed from both complexes in the same way as described for complex 1.According to the elemental analysis data, complex 2 consists of Fe, 18.4; Y, 8.4; C, 42.3; H, 5.3; 0, 25.6mass%, and complex 3 consists of Fe, 2.4; Y, 31.6; C, 35.0; H, 4.4; 0,26.6 mass%. The empirical formulae of the complexes are con- sidered to be YFe&&3016 (2) and FeY9C72H110040 (3). Mononuclear Fe"' and Y'" acetyl acetonates were also prepared by the electrochemical dissolution of yttrium and iron electrodes of high purity in acetylacetone solution with J. Muter. Chew., 1996, 6(2), 207-212 207 further multiple recrystallization, as in ref. 5. Commercial Al(acac), and zirconium(1v) isopropylate [Zr(OPr',)] were used. Alkoxosyntheses of the gels and thermal preparation of the oxides To prepare solution 1, an equimolar amount of acetylacetonate was added to Zr(OR), (0.0082 mol) dissolved in 50cm3 absolute benzene, and the solution was stirred for 30 min.Solution 2 was prepared by dissolving 0.00086 mol of hetero- or mono-nuclear acetyl acetonates of Fe"', Y"', Al"' or their mixtures in 20cm3 absolute benzene. The total amount of metal acetyl acetonate was calculated as 5 mol% with respect to ZrO,. Solution 2 was added to solution 1 while stirring and was kept for 24 h in a closed vessel. After this time, 80% ethanol containing a stoichiometric amount of water was slowly added dropwise to the benzene solution of the organometallic precursors. After 10 days the resulting product was evaporated slowly at room temperature, yielding the glass- like gel.Complex oxides were obtained by thermal treatment of these gels at 500 and 1000°C over 8 h. Techniques The compositions of the gels and the metal oxides were determined by atomic absorption spectroscopy (AAS) using a Perkin-Elmer 400 spectrometer. The static magnetic susceptibility was measured by means of a home-made Faraday-type magnetic balance', in the temperature range 77-300 K. X-Ray diffraction (XRD) analysis was performed with a Philips PW-1700 powder diffractometer using Cu-Ka radi- ation. The lattice-type unit-cell parameters were determined by the procedure described in ref. 15. 27Al magic-angle spinning (MAS) NMR powder spectra were recorded by means of an MSL-400 spectrometer.Room-temperature EXAFS spectra were recorded and ana- lysed as in ref. 8 with the use of the VEPP-3 electron-storage rings in the Institute of Nuclear Physics, Novosibirsk.16 IR spectra of the gels were recorded by means of a Specord- 8000 instrument using the KBr disk method. Results and Discussion Three new organometallic iron-yttrium complexes (1-3) were synthesized by the anodic electrochemical dissolution of Fe-Y alloys [Fe2Y/Y, Fe,Y; Y: Fe =1 :1, 2: 1 mol%, respectively] in acetyl acetone solution. IR spectra of complexes 1-3 are characteristic of metal p-diketonates. In these spectra, intense absorption bands at v= 1520 and 1530 cm-', characteristic of chelate fragments of the carbonyl group vibrations are observed. Structure and properties of the iron-yttrium precursors XRD analysis of complex 1 gives evidence of a monoclinic structure resembling monoclinic Y (acac),, and differing com- pletely from the rhombohedra1 Fe(acac), structure. The Mossbauer spectrum of this complex shows a narrow doublet with the following parameters: isomer shift (IS) = 0.49 mm s-', quadrupole splitting (QS)=0.37 mm s-' and linewidth (r)=0.36 mm s-'.The parameters differ completely from the relaxation spectrum of Fe(acac),. In the absence of XRD results for monocrystals of 1, Table 1 EXAFS data on the local coordination around Fe and Y atoms in complex 1 and in the gel obtained after reaction of 1 with zirconium isopropylate complex 1 gel relative relative bond intensity Rlnm intensity Rlnm Fe-0 0.37 0.203 0.35 0.203 Fe-Y 0.16 0.282 0.13 0.280 Fe -Fe 0.11 0.402 0.10 0.410 Y-0 1.18 0.220 1.46 0.220 dence on the magnetic field [x(H)] in the temperature range 79-300 K and at magnetic fields from 1.5 to 7.8 kOe, respect- ively, are shown.The curve of [x-'(T)] is nearly linear. An extrapolated 0value is small and positive. In addition, x exhibits a significant dependence on the applied magnetic field. Above all, these results suggest that complex 1 is magnetically inhomogeneous, which may be due to the existence of several iron-containing phases. We may expect one such phase to be highly dispersive ferromagnetic clusters with the general properties of superparamagnetic parti~1es.I~ Precursors 2 and 3 were also obtained as a result of electrochemical dissol~tion.'~ XRD, magnetic susceptibility, Mossbauer and X-ray microanalysis data show that the monocrystalline complex 2 appeared to be hydrated iron@) acetylacetonate, Fe(CH3COCHCOCH3),~2H,0. It was shown that such monocrystals are covered by an X-ray amorphous 1 1 I 1 I 0 50 100 150 200 250 300 T/K Fig.1 z-l(T)for the iron-yttrium organometallic precursor 1 (a) and its gel (b) 300K \ =: 2.5 2.01.5 0'I\-lI?&I complementary data on the local coordination around the Fe atoms in complex 1 were obtained from EXAFS measurements 12345678 (Table 1). H/kOe In Fig. l(u) and 2(a), the temperature dependence of the Fig. 2 x(H) for the iron-yttrium organometallic precursor 1 (a) and magnetic susceptibility [x-'( T)]and the susceptibility depen- its gel (b) at 300 K 208 J.Muter. Chem., 1996, 6(2), 207-212 film, consisting of yttrium ions and superparamagnetic clusters (Fe, .-xYx). It was found that the polycrystalline oligomeric acetylacetonate complex 313 contained Fe" and YIIl ions and a small amount of a ferromagnetic admixture which is supposed to be superparamagnetic clusters. Influence of the acetyl acetonate precursors on the formation mechanism of complex oxides To study the influence of different heteronuclear iron-yttrium acetyl acetonates and aluminium acetyl acetonate on the formation mechanism of complex oxides, the alkoxosyntheses of precursors 1-3 with zirconium isopropylate were performed and the products were heat treated.Table 2 gives the phase composition and structural charac- teristics of the complex oxides obtained at 500 and 1000°C. We have previously studied the formation mechanism of the double oxides Fe,( Zr,Ti), -0,75x02-based on alkoxosynthesis. As basic reagents Fe(acac), and Zr(OPr'), were On the basis of EXAFS and magnetic susceptibility data it was shown that small clusters with the ZrO, structure are formed in iron-containing gels. These clusters are 0.1-0.12 nm long and are bound together by iron ions in an octahedral environment. According to the EXAFS data these structure elements are characterized by the following interatomic dis- tances: Fe-0, 0.194; Fe-Fe, 0.312; Fe-Zr, 0.330; Zr-0, 0.215; Zr-Zr, 0.335 nm.After heat treatment at 500 "C this gel readily transforms to a single-phase oxide. The interatomic distances in the oxide are practically equivalent to those found in the gel. We have therefore shown that oxide structure formation take place just at the gel stage. Crystallization contributes to the contraction of cubic polyhedra in the gel, and extraction of organic fragments occurs. This results in the appearance of the cubic metal oxide Feo.13Zro.902-6 with small structure distortions. On the basis of EXAFS data, as well as the magnetic susceptibility and XRD data, a model of the iron-containing zirconium oxide was proposed.' Our proposed model consists of crystal blocks of cubic Zr02, while Fe3+ ions produce an epitaxial film on the surface of the ZrO, cluster.Different ZrO, clusters are connected to each other by the Fe3+ film in which a metal-metal bond (Fe-Fe, 0.312 nm) is identified. Therefore, the main advantage of the proposed mechanism is the formation of intermediates with clearly defined structural phase characteristics in gels. This phenomenon enables us to prepare single-phase double oxides at considerably lower tem- peratures, possessing the structural characteristics of the formed intermediates. As seen in Table2, the mechanism described above, is supposed to occur in the syntheses of different acetylacetonate precursors ( 1-9) with zirconium isopropylate. The further heat treatment at 500°C of such reaction products gives rise to single-phase oxides with lattice parameters (0.5-1.2 nm) close to those of cubic ZrO,.By analogy with ref. 8 we analysed the structure and properties of some gels and polycrystalline treated oxides by a number of methods. In Fig. 1, 2(b), 3, 4 and 5(u), magnetic susceptibility data and Mossbauer spectra for gels and oxides based on precursor 1 are presented. The temperature dependences of the magnetic susceptibility for the gel [Fig. l(b)] based on precursor 1 reveal , that iron is incorporated into the gel. It should be noted that curves (a) and (b)in Fig. 1 are very similar. The absolute values of x at room temperature and at liquid nitrogene temperature are closely related. Both curves have very small Nee1 temperatures, especially x(T)for the gel (Fig.1). In other words, a significant contribution of isolated Fe"' or Fe" ions in these dependences can be expected. However, in Fig. 2(u), (b)we can see the dependence of the susceptibility on the magnetic field, which is predominantly due to the presence of small ferromagnetic clusters revealed in the initial complex. Considering that x-'(T) for an ensemble of such superparamagnetic particles is of the Couri-low typeI7 the contributions of the isolated iron ions and ferromagnetic, highly dispersive clusters cannot be distinguished in the x-'(T) dependence. A narrow doublet in the gel spectrum (Fig. 3b; IS=0.24f0.03 mm s-l; QS=O.19+0.03 mm s-lis most likely due to these clusters. The main part (60%) of the resonance absorption [also a doublet (Fig.3a)l belongs to paramagnetic high-spin Fe3+ ions (IS=0.34+0.03 mm s-', QS= 1.07+ 0.03 mm s-l) which predominantely contribute to x-'(T) [Fig. l(b)]. Taking into account that yttrium ions can strongly stabilize the cubic structure of ZrO,,, and according to the similar data for iron ions,8 Fe3+ ions are supposed to form simultaneously with Y3+ ions the cubic ZrO, lattice in the reaction with zirconium isopropylate. One can suppose that Y3+ and Fe3+ might occupy equivalent positions and might be bonded directly to each other or via the oxygen bridges. Indeed, the data on the local coordination around the Fe atoms in complex 1, determined from the radial distribution function (RDF) curves of the Fe K edge EXAFS results and presented in Table 1, show that three maxima occur in the RDF curves at R =0.203, 0.282 and 0.402 nm.These could be assigned to the Fe-0, Fe-Y and FeoFe bonds, respectively. Therefore, the Fe-Y distance of 2.82 A can be attributed to a direct metal-metal bond. Note that such interatomic distances appeared to be similar to those found in the gel (Table 1). These data confirm the mechanism suggested above, accord- ing to which we suggest the possibility of the existence of some heterometallic polyhedra which are retained during the chemi- cal reaction with Zr(OPr'), and in the following annealing process. To substantiate the conclusions reached above, we can compare Mossbauer and magnetic data for the complex oxides (500"C) and ceramics ( 1000 "C) based on complexes 1, 2 and the mixture 7 (see Table 2).Fe3+ ions, which appear after the reaction with Zr(OPr'), in gel 1 and occur in the oxide structure formation, define the absorption spectrum of the cubic oxide (Fig. 3) and form the oxygen octahedra with practically the same distortions in the ceramic oxide (see Table 2). According to these data, we suggest that the interatomic distances found in complex 1, and in particular the distance Fe-Y =0.282 nm, which are not changed in the gel are retained in the oxides at both temperature. It is likely that the set of chemical b,onds with interatomic distances outlined in Table 1 form the firm basis of the single-phase ceramic. As is shown in Fig. 5(4, Fe3+ ions in this oxide as well as for the ceramic based on precursor 2 are antiferromagnetically coupled.It is therefore evident that one of the structure fragments of the triple Fe-Y-Zr oxide is the following unit [-Fe3+-0-Fe3+ -3 I I -y3+ -y3+ I I 0 0 By analogy, the presence of the same strong bonds in the iron-yttrium acetylacetonate complex 2 allows the formation of a single-phase ceramic. In fact, attempts to obtain-single phase oxides at 1000"C with mixture 7 failed (Table 2), despite the mixture containing the same concentration of iron and the excess of Y3+ ions which c.an stabilize the cubic structure. This may be accounted for to some extent by the absence of -Fe-Y-0- fragments in the gel, and therefore in the oxides. In this case the iron ions slide over the faces of the zirconium polyhedra and precipitate as an individual phase, a-Fe,03, at 1000"C.The lack of direct iron-yttrium bonds in precursor 313, together with the high J. Muter. Chern., 1996, 6(2), 207-212 209 Y Table 2 Nature of the organometallic precursors, Mossbauer data, structure and phase composition of zirconia-based complex oxides Q\ XRD data Mossbauer data n w =-' lattice h, phase composition parame ters/nm paramagnetic a-Fez03s treatment relative relative relative sample molecular formula of temperature/ crystal content content IS/ QS/ content number precursors single-phase metal oxides "C system ("/) a C (Yo) mms-l mms-' (%) Hi,/kOe 1 heteronuclear: 500 cubic 100 0.513 100 0.25 0.97 Y: Fe=4: 1 1000 cubic 100 0.514 100 0.24 0.95 2 Y :Fe =0.3 :1 Y0.02Fe0.08Zr0.9302 + d 500 cubic 100 0.510 100 0.27 0.98 non-single phase lo00 cubic 15 0.5 15 -monoclinic 76 -a-Fe,O, 9 3 Y:Fe=9:1 Y0.09Fe0.01Zr0.9302 + 6 500 cubic 100 0.512 100 0.30 0.97 lo00 cubic 100 0.514 100 0.35 0.95 4 mononuclear: Fe0.10Zr0.9302 +6 500 cubic 100 0.509 100 0.35 1.05 Fe(acac), non-single phase 1000 monoclinic 93 100 515; 500 a-Fe,O, 7 5 Al(acac), A10.10zr0.9302 +6 500 cubic 100 0.509 non-single phase 1000 tetragonal 70 0.508 0.512 monoclonic 30 6 Y (acac), 500 cubic 100 0.512 1000 cubic 100 0.515 7 Mixture: y0.08Fe0.0Zzr0.9302+ 6 500 cubic 100 0.512 100 0.29 0.97 3.5Y(acac),+ non-single phase 1000 cubic 70 0.514 32 0.31 1.07 68 515; 500 Fe(acac), monoclinic 25 a-Fe,O, 5 8 3.5Y(acac), + y0.08A10.02zr0.9302+ 6 500 cubic 100 0.512 A1 (acac), 1000 tetragonal 100 0.509 0.515 1.oo \I \I 1.10.99E a 1; I II\ I 1.oo 1.oo 3 0.995.-C cnc 0 e .-c ii I' Q, ti .-5 1.00i?? d ii II 0.996 1.oo 0.996 II ir? II II 1.oo f I *I0.995 Ihr 1111.Ij I1 I uu I -10 -5 0 5 10 vetocity/rnrns-' Fig.3 Mossbauer spectra for the gels obtained in the reaction with zirconium isopropylate, oxides and ceramic materials on the basis of mixture 3SY(acac),+Fe(acac), (a, c, e) and complex 1 (b, d, f). Bars: 1, Fe3+ octahedra; 2, [YFe]; 3, cr-Fe,O,. 1.0.-0.5 -0.0w 012345670 H/kOe 5.0 4.51 4.01 6 3.0 2.5 -72.0 -I /-,/'0.5 Fig. 5 x-'(T) for the ceramic oxides based on precursor 1 (a) and precursor 2 (b) iron concentration may cause the destruction of any single- phase oxide at 1000"C (Table 2).We also compared the influence of the nature of an ion- modifier which was very different from Fe3+ ion on the mechanism of the oxide formation, described above. Aluminium acetylacetonate (Table 2, sample 5) was chosen as a basic precursor. A mixture of 3.5Y(acac), +Al(acac), (Table 2, sample 8) was also tested. As a result, cubic single- phase oxides with lattice parameters almost identical to those obtained for the iron-containing precursors (Table 2) were obtained both for precursor 5 and for mixture 8 at 500 "C. We could not obtain a single-phase ceramic at 1000°C also, as in the case of the iron-containing precursors, except for precursors 1 and 2.Double-phase ceramic oxide 5 consists of 70% tetragonal phase and 30% monoclinic phase. The ceramic oxide based on mixture 8 differs from the iron-containing ceramic double-phase oxide ,7. In the case of oxide 7, we probably obtained a triple-phase oxide with a phase composi- tion analogous to the iron-yttrium precursors (Table 2). Ceramic oxide based on the aluminium-containing precursor In the case of mixture 8 the ceramic oxide appeared to be single phase and is the only distinguishing feature of the synthesis based on the aluminium-containing precursor observed in the presence of excess yttrium. It may be reasonabl? to suggest that aluminium(m) ions, with ionic radius 0.57 A, which is smaller than that for zir- conium(1v) and iron(m), can, unlike iron ions, penetrate into the interior of the ZrO, lattice and substitute yttrium ions in tetragonal positions.According to the 27Al NMR data the line position corresponds to aluminium ions in an octahedral surrounding, therefore the line is broaded. This broadening is greater than that for the known aluminium oxides and confirms the high extent of asymmetry of the aluminium surroundings. This may be the case when aluminium ions penetrate into the tetragonal ZrO, lattice since oxygen forms a distorted cubic lattice in which the A1 ions are stabilized. Conclusions We have found that homo- and hetero-nuclear acetylacetonates of Y''', Fe"' and Al"' interact with zirconium isopropylate in the mother liquor and form cubic heterometallic intermediates based on Zr02 polyhedra.It has been shown that for all iron-containing gels and oxides described in this paper, Fe3+ ions as well as Y3+ ions Fig. 4 x(H) for the ceramic oxide based on precursor 1 at (a) 300 participate in the oxide formation mechanism described in and (b)77 K ref. 8. J. Muter. Chem., 1996,6(2), 207-212 211 The sets of interatomic distances found by means of EXAFS in complex 1and in the zirconium gel based on 1 are practically equal, and are probably retained in the oxides. Moreover, the presence of structure fragments formed by these bonds, and in particular the direct iron-yttrium bond, in the precursors 1 and 2 allows the formation of single-phase triple ceramic oxides.We managed to obtain another triple single-phase ceramic oxide based on the mixture 3.5Y(a~ac)~ +Al(acac), (sample 8). In contrast to Fe3+ ions in the ceramic oxide based on an analogous mixture with Fe(acac), (sample 7), A13+ ions pen- etrate into the lattice volume of the tetragonal ZrOz. Thus, aluminium ions presumably substitute Y3+ ions in tetragonal positions. The support of the International Science Foundation (MI 8000, MI 8300) and NATO grant (HTECH.CRG 931258) is acknowledged. References J. Livage, K. Doi and C. Maziere, J.Am. Ceram. SOC., 1968,51,349. E. Tani, M. Yoshimura and S. Somiya, J. Am. Ceram. SOC., 1983, 66, 11. K. Tsukama, K.Veda and M. Shimada, J. Am. Ceram. SOC., 1985, 68,4. T. L. Wen, V. Hebert, S. Vilminot and J. C. Bernier, J. Muter. Sci., 1991,26,3787. M. V. Tsodikov, 0. V. Bukhtenko, 0. G. Ellert, D. I. Kochubey and S. M. Loktev, Structural Ceramics Processing, Microstructure and Properties, ed. J.J. Bentzen, J.B. Bilde-Snrrensen, N. Christiansen, A. Horsewell and B. Ralph, Proceedings ofthe 11th International Symposium on Metallurgy and Materials Science, 1990, Roskild, Denmark, p. 505. 6 R. Schmid, H. Ahamdane and A. Mosset, Inorg. Chim. Acta, 1991, 190,237. 7 N. M. Kotova, S. G. Prutchenko and M. I. Yanovskaya, Inorg. Muter., 1994,30, N3,387. 8 M. V. Tsodikov, 0. V. Bukhtenko, 0. G. Ellert, V. M. Shcherbakov and D. I. Kochubey, J. Muter. Sci., 1995,30, 1087. 9 D. C. Bradley, Chem. Rev., 1989,89,1317. 10 K. G. Cautton and L. G. Hubert-Pfalzgraf, Chem. Rev., 1990, 90, 969. 11 W. Bidell, H. W. Bosch, D. Veghini, H. V. Hund, J. Doring and H. Berke, Helv. Chim. Acta, 1993,76, 596. 12 E. N. Savitskii and V. F. Terehova, Metallovedenie redkozemelnyh metallov, Nauka, Moscow, 1975, p. 271. 13 M. V. Tsodikov, 0.V. Bukhtenko, 0.G. Ellert, I. A. Petrunenko, A. S. Antsyshkina, G. G. Sadikov, Yu. V. Maksimov, Yu. V. Titov and V. M. Novotortsev, Russ. Acad. Sci. Bull. Dev. Chem., 1995, 8, 1396. 14 V. T. Kalinnikov and Yu. V. Rakitin, Vvedenie v magnetochimiyu, Nauka, Moscow, 1980, p. 53. 15 S. S. Gorelyk, L. N. Rastorguev and Yu. A. Skakov, Rentgenographicheskii i electronographicheskii analys, Metallurgia, Moscow, 1970, p. 450. 16 D. I. Kochubey, EXAFS spectroscopy of catalysts, Science, Novosibirsk, 1992, p. 145. 17 Yu. V. Maksimov, M. V. Tsodikov, 0. G. Ellert and V. M. Shcherbakov, J. Catal., 1994,148, 119. Paper 5/04609B; Received 13th July, 1995 212 J. Muter. Chem., 1996, 6(2), 207-212
ISSN:0959-9428
DOI:10.1039/JM9960600207
出版商:RSC
年代:1996
数据来源: RSC
|
15. |
Preparation of ceria-coated composite silica particles |
|
Journal of Materials Chemistry,
Volume 6,
Issue 2,
1996,
Page 213-219
Anders Törncrona,
Preview
|
PDF (1269KB)
|
|
摘要:
Preparation of ceria-coated composite silica particles Anders Torncrona,"* Lars Lowendahl," Jan-Erik Otterstedt" and Kjell Janssonb aDepartment of Engineering Chemistry, Chalrners University of Technology S-412 96 Gothenburg, Sweden bDepartment of Inorganic Chemistry, Arrhenius Laboratory, Stockholm University S-106 91 Stockholm, Sweden A method for preparing particles with highly dispersed ceria by a stepwise procedure is described. The surface of silica particles was modified by aluminate and coated with an interlayer of titania. CeOSO, was deposited on the titania coated silica particles. The sulfur in the dispersed ceria was removed by calcining the particles in air at 800 "C for 30 min. A highly dispersed ceria was obtained when using titania-coated silica particles with diameters of 80,240 and 500 nm.The prepared products were characterised by scanning electron microscopy (SEM) and transmission electron microscopy (TEM) with energy dispersive spectroscopy (EDS), X-ray diffraction (XRD), UV spectroscopy and electrophoresis. Ceria is used in various applications such as ceramic opacifiers, phosphors, capacitors and solar cells because of its unique In addition, it has been used in car exhaust catalysts for many years because of its oxygen storage capacity and promoting effects on supported noble metal^.^-^ Furthermore, ceria has proved to be an active catalyst in the oxidation of gaseous hydrocarbons as well as solid carbon particulates, e.g. diesel soot particulate^.^.^ Ceria particles can be prepared by several procedures.Sols consisting of cerium(II1) basic carbonate have been synthesised by ageing cerium(n1) nitrate solutions at elevated temperatures in the presence of urea.' Colloidal particles of cerium(1v) sulfate have been made by forced hydrolysis of solutions containing cerium(rv) sulfate and sulfuric acid." In the latter investigation it was shown that the ceriium(Iv) sulfate concentration as well as the mole ratio of cerium@) sulfate to sulfuric acid in the stock solutions determined the formation of the particles, i.e. their size, shape and composition at a specific temperature. Lower concen- trations of cerium@) sulfate and sulfuric acid gave spherical particles of cerium oxide, CeO,, whereas rod-shaped particles consisting of cerium(1v) oxysulfate, CeOSO,, were formed at higher reactant concentrations. In a recent investigation, the hydrolysis of cerium(1v) sulfate in dilute sulfuric acid was studied in detail using EXAFS and SAXS" and it was found that 3 nm particles consisting of cerium@) oxysulfate were formed under the conditions used.A temperature rise first gave colloidal particles and finally precipitates of cerium(iv) oxysulfate. Besides aqueous ceria dispersions, colloidal disper- sions of ceria in organic media have also been formed.', When used as an active component in catalysts, ceria is usually incorporated as a solution of cerium(i11) nitrate added to the I Silica sol I -Aluminorilicate modified sol silica sol t t t Cena coating Cena coating Ceria coating 6 6 p+q Fig.1 Schematic route for making metal-oxide coated silica sols. Ceria coat successfully applied only to titania-coated silica sol. slurry containing the catalyst support particle^.^-^ The ceria precipitates when the catalyst is dried and forms rather large particles, in their turn built up from small crystallites, in the catalyst support resulting in ceria of quite low surface area, < 10 m2 g-', i.e. a material with a low degree of dispersion of ceria. It is therefore of great interest to study preparation methods for catalytic materials yielding highly dispersed ceria. The preparation of small particles with highly dispersed ceria on the particle surface is described in this paper and the design of the experimental programme is shown in Fig.1. The particles prepared in this work are primarily intended for use in catalysts for the combustion of diesel soot particulates and large organic solvent molecules. Experimenta1 Materials Cerium(1v) sulfate [Ce( *4H20,' Merck, Germany, pro analysi], urea [(NH2),C0, Riedel de Haen, Germany, pro analysi], titanium(rv) chloride (TiCl,, Tioxide Specialities Ltd, UK), zirconium(rv) chloride [ZrCl,, Merck, Germany], sodium aluminate, (NaAlO,, 55% A1203, Merck, Germany, purum) and tetraethylorthosilicate (TEOS, 98%, Aldrich Chemie, Germany). Cation-exchange resin [Dowex HCRS( E)] and anion-exchange resin (Dowex SBR) were from Dow Chemicals, USA. Buffer solutions, glycine-hydrochloric acid (pH 1) and citrate-hydrochloric acid (pH 2 and pH 4), were from Merck, Germany.Methods Preparation of colloidal particles of silica. Silica sols with particle diameters of 80nm and 240nm were prepared by a build-up pr~cedure.l~-'~ The sodium content was reduced by slowly passing the sols through a column filled with a strong cation-exchange resin in the H+ Silica particles with a particle diameter of 500 nm were prepared by base-catalysed hydrolysis of tetraethylorthosilicate in ethanol.16 Introduction of aluminosilicate ions onto the surface of the silica particles. Sodium aluminate (11.6 g) was added to 24.1 g distilled water under stirring and heating to dissolve the powder. The resulting solution was filtered to remove insoluble impurities.The pH of each silica sol was adjusted to pH9 with dilute hydrochloric acid in order to lower the solubility of silica. Small amounts, corresponding to 1.5 aluminate ions per square nanometre of the silica surface, of the sodium aluminate solution were then slowly added to three different silica sols, with particle diameters of 80, 240 or 500 nm, while J. Muter. Chern., 1996, 6(2), 213-219 213 stirring vigorously. The sols were then heated at 80 "C for 2 h whereby the aluminate ions were exchanged into the silica surface. Finally, the electrophoretic mobility of the sols was measured as a function of pH to verify that aluminosilicate sites actually had been formed on the silica surface. Coating aluminate modified silica particles with titania or zirconia. A solution of titanium@) chloride was prepared by dissolving TiCl, in concentrated hydrochloric acid (12 mol dm-3) and diluting the solution with distilled water, yielding a clear solution of 0.79 mol dm-3 in TiCl, and 1.37 mol dm-3 in HCl.Solutions of titanium(1v) chloride must be used within a few days since titania precipitates when stored for about a week at room temperature. Zirconium(1v) chloride was dissolved in distilled water, giving a clear and stable solution of 0.64 mol dmP3 in ZrCl,. Three silica sols with particle diameters of 80, 240 or 500nm were slowly added (ca. 20min) to a solution of either titanium(1v) chloride or zirconium(1v) chloride (see Table 1 ). The silica sols all contained aluminosilicate groups on the particle surface.The sols were then heated at 90°C for 1 h with moderate stirring to complete the coating of the silica particles. The coated sols were then slowly cooled to room temperature, and finally the pH was raised from ca. pH 1.5 to ca. pH 2 by using an anion-exchange resin in the OH-form. Samples from each sol were taken and centrifuged at 3000 rpm for 15 min and the supernatant was analysed with respect to titanium or zirconium content. Deposition of ceria on precoated silica particles A cerium(1v) sulfate stock solution, 10 mmol dm-3 in Ce(SO,), and 50 mmol dm-3 in H,SO,, was made by dissolving cerium(1v) sulfate in dilute sulfuric acid at room temperature. The solution was then filtered to remove any inorganic insoluble particles.To promote the hydrolysis of cerium(1v) ions and hence the formation of polycations, the cerium(1v) sulfate solution was preheated at 60 "C for 30 min (stock solution 1). One portion each of four sols with different surface composition was then either slowly (20 min) or rapidly (2 s) added to a preheated cerium@) sulfate solution (see Table 2). After adding the sols to the cerium(1v) sulfate solution, the sols were heated at 80 "C for 20 h. Another cerium(1v) sulfate stock solution was prepared in Table 1 Aluminate-modified silica sols of different sizes used in the coating experiments with either titania or zirconia silica sol' particle diameter/nm added MO,/mg mT2 1:l 500 1.5' 1:2 240 1.5' 1:3 80 1.5' 1:4 500 1.5' 1:5 240 1.5' 1:6 80 1.5' a Aluminate-modified silica sols with 1.5 A1 nmP2 (particle surface area).'MO, =TiO, (mg m-' particle surface area). MO, =ZrO, (mg m-' particle surface area). Table 2 Composition of the sols added to cerium(rv) sulfate solutions. When the sols were coated with ceria they were either slowly (20 min) or rapidly (ca. 2 s) added to the cerium@) sulfate solutions sol sol composition (500 nm) 2: 1 SiO, 2:2 SiO," 2:3 Ti02/Si02"2:4 ZrO,/SiO," 'Silica sol modified with aluminate. 214 J. Mater. Chern., 1996, 6(2), 213-219 the same way whereupon urea was added to the solution. The resulting solution (stock solution 2) was 10mmol dm-3 in Ce(SO,),, 50mmol dm-, in H2S04 and 100mmol dmV3 in (NH,),CO.Sols identical with those previously used with solution were then either slowly or rapidly added to the urea- containing solution (see Table 2). A third cerium(1v) sulfate stock solution was prepared in the same way as solution 1, but without the preheating step, whereupon ammonium hydroxide was added to promote the formation of cerium (~v) polycations. Urea was finally added to the solution. The resulting stock solution (3) was 10 mmol dm-3 in Ce(SO&, 50 mmol dmP3 in H,S04, 100 mmol dm-3 in NH, and 100mmol dm-3 in (NH,),CO. Four 500 nm sols with different surface modifications (see Table2) were then each rapidly added to portions of stock solution 3, and heated at 80 "C for 20 h. Three titania-coated silica sols with particles of 80, 240 and 500 nm in diameter were rapidly added to a portion of stock solution 3, and then heated at 80 "C for 20 h.The amount of ceria, as CeO,, used in all experiments was about 5 mass% of the final particle mass. The entire experimental schedule is shown in Fig. 1. A pure cerium(1v) oxysulfate sol was prepared by heating a portion of solution 1 at 80 "C for 20 h. Material characterisation The titanium, zirconium and cerium contents were all deter- mined by colorimetric method^,'^-'^ (Shimadzu UV 160 A). The size of the sol particles was determined by dynamic light scattering (Brookhaven BI-90 particle sizer). The sol particles were dispersed in a 10mmol dm-3 NaCl solution in which the pH was adjusted with dilute HCl or NaOH solution.The electrophoretic mobility of the particles was then measured as a function of pH by using a Zetasizer IIc (Malvern Inc.) instrument. Measurements of pH were made with a combi- nation electrode with Ag/AgCl reference from the Broadley- James Corporation (USA). Surface studies of the materials were carried out using scanning electron microscopy (SEM; JEOL JSM-5200), and by transmission electron microscopy (TEM; JEOL 2000FX) equipped with a detector for energy dispersive spectroscopy (EDS; LINK AN 1000). Samples of the coated particles were obtained by drying droplets of the sol solution on an aluminium foil at 110 "C for 30 min. In the case of TEM studies the particles were dispersed in butanol and transferred onto a hollow carbon film supported by a copper grid.Thermogravimetry (TG; Perkin Elmer TGA 7) was used to study the decomposition of the ceria coatings in air with a heating rate of 5°C min-'. The crystalline phases which appear due to the calcination of the ceria coatings was determined by powder X-ray diffraction (XRD) using a Guiner-Hagg camera working with Cu-Ka, radiation. A1 was used as internal standard. Specific surface areas of ceria coated titania/silica particles of different sizes were determined by nitrogen adsorption according to the BET method on a Digisorb 2600 instrument (Micromeritics). The surface area analyses were made after calcination of the particles at 800°C for 30 min in air. Results and Discussion Surface modification of silica particles Silica sol particles have a low negative surface charge below pH 6 due to a low degree of ionisation of the surface silanol groups.In order to increase the negative surface charge of the silica particles at low pH values and thereby facilitate the adsorption of positively charged titania or zirconia species on the surface, aluminosilicate sites were introduced into the particle surface layer. Fig 2 shows the electrophoretic mobility of unmodified and aluminate-modified silica sols, with a Fig.2 Electrophoretic mobility as a function of pH for silica sols (-, 240 nm in diameter) and aluminate-modified silica sol with 1.5 A1 nmP2 (---, 240 nm in diameter) particle diameter of 240 nm, as a function of pH.The increase in electrophoretic mobility in the pH range 3-7, due to the aluminosilicate sites, is evident. The increased negative electro- phoretic mobility of the silica particles containing aluminosilic- ate sites (1.5 A1 nm-2 of the particle surface) is caused by a fixed negative charge on each aluminosilicate site. The fixed negative charge is believed to originate from the aluminate ion, Al(OH)4-, which is exchanged into the silica surface owing to its geometrical similarity to Si(OH)4.20 Since solutions containing polycations of titania or zirconia are very acidic, the silica particles modified with aluminate can be expected to be more appropriate to use for the deposition of titania or zirconia than unmodified silica particles, provided that the polycations are supposed to be electrostatically attached to the particle surface.It was also noticed that secondary particles were sometimes formed during the modification of the silica particles with aluminate ions. Initial pH and sodium content of the silica sol used were factors which were found to affect the formation of secondary particles which evidently results from a reaction between dissolved silica and aluminate during the heating of the sol. For this reason, the initial pH of the silica sol must not exceed pH 9 since the solubility of silica is markedly increased above this pH value. It was found that a high sodium content in the silica sol also leads to the formation of secondary particles during the heating of the sol with the aluminate.If the pH of the sodium-containing silica sols was adjusted to pH 9, and the sodium content of the sols had been reduced by repeated cation exchanges, no secondary particles were formed. The sodium-free silica sol, prepared by hydrolysis of TEOS, did not yield secondary particles when modifying the surface with aluminate if the pH of the sol was kept at pH 9 when adding the sodium aluminate solution. Titania and zirconia coatings on modified silica particles Aluminate modified silica particles of three different sizes (SO, 240 and 500 nm) were coated with titania or zirconia and the results are shown in Table 3. The yield of deposited material was calculated with the assumption that all the titania or zirconia had been deposited on the particle surface except for the equilibrium concentrations of titanium or zirconium in the solutions. The solubility of zirconia is at least an order of magnitude higher than that of titania at a pH of about 2, which explains the lower fraction of added zirconia adsorbed on the surface of the silica particles.The electrophoretic mobilities of the titania- and zirconia-coated silica particles with a particle diameter of 240nm, as a function of pH, are shown in Fig. 3. The isoelectric point (iep) of titania-coated silica particles was at about pH 4.5 and at pH 6.5 for zirconia coated silica particles. The iep of titania-coated silica particles is approximately one pH unit lower than that reported for pure titania particles.21 It has been shown that the occurrence Table 3 Equilibrium concentrations of the coatings (expressed as titania or zirconia) and the final pH of each sol after coating and anion exchange of the sols ppm MO, in mass% of added MO, on silica solu supernatant particle surface PH 1:l 79’ 81.0’ 2.0 1:2 58’ 93.5’ 2.4 1:3 67’ 97.4’ 2.1 1:4 455‘ 62.1‘ 2.3 1:5 500‘ 80.8‘ 2.2 1:6 465‘ 93.8‘ 2.3 Aluminate-modified silica sols with 1.5 A1 nm-2 (particle surface area).’MO, =TiO,. MO, =ZrO,. Fig.3 Electrophoretic mobility as a function of pH for silica sols (240 nm in diameter) coated with titania (-) and zirconia (---) as a function of pH of fluoride or chloride ions in alumina lowers the iep of alumina22 and it is therefore possible that the decrease in iep is caused by the presence of chloride ions.During the coating process chloride ions were released when the titanium(1v) chloride or zirconium(1v) chloride was deposited on the silica particle surface as titania or zirconia, respectively. Owing to the high chloride ion concentration, the sols flocculated when the coating process was completed by heating the sols. The sols were therefore anion-exchanged after they had been cooled to room temperature. During the anion exchange, most of the chloride ions were replaced by hydroxide ions, which resulted in deflocculation and stable sols were obtained. Fig. 4(u) and (b)show TEM images of a silica particle, 240 nm in diameter, coated with titania and zirconia, respectively.The fuzzy edges of the particles in Fig. 4(u) and (b)contrast with the smooth edges observed for uncoated silica particles. Elemental analysis by EDS verified that titania as well as zirconia had been deposited on the surface of the silica particles (about 2.0 atom% and in the range 0.5 to 3.5 atom%, respectively). However, the morphologies of the titania and zirconia coatings were found to be different. Fig. 4(a) shows that the titania forms small particles well distributed on the particle surface, whereas zirconia yielded homogeneous layers, [see Fig. 4(b)]. In the latter case a few uncovered particles were found in the material. One of the motives for precoating the silica particles with titania or zirconia was to stabilize the silica against reaction with steam at elevated temperatures.Another import- ant reason for precoating the silica particles was to facilitate the dispersion of ceria on the surface of the particles. Ceria coating on particles of different composition The surface of four ceria treated sols was examined with SEM (see Table 2). From the results of stock solution 1 (no urea), it could be concluded that ceria was not deposited on any of the four different sols regardless of the speed of addition to the cerium(1v) sulfate stock solution. Instead, the ceria formed rod- shaped particles as seen in Fig. 5. One possible explanation is J. Muter. Chem., 1996, 6(2), 213-219 215 means that polycations are not electrostatically attracted to Fig.4 TEM images of silica particles coated with a thin layer of titania (a) and zirconia (b) Fig.5 SEM image of uncoated silica particles together with rod-shaped particles consisting of cerium@) oxysulfate, CeOSO, that pH decreases during the heating of the different sols together with the cerium(1v) sulfate solution since H+ is one of the hydrolysis products.The pH of the cerium(1v) sulfate stock solutions (1 and 2, Table 2) was PH 1.2. At this low PH pure silica particles are slightly positively charged23 which 216 J. Mater. Chem., 1996, 6(2), 213-219 these particles. When silica particles modified by aluminate were used, the aluminium was probably dissolved into the solution owing to the low pH.24 As a consequence, these particles lost their negative surface charge, which resulted in the same condition as for pure silica particles (see above).Concerning the particles coated with either titania or zir- conia, the thin layers of these compounds probably dissolved when the sol particles were heated together with the acidic cerium(1v) sulfate solution since the solubility of titania and zirconia increases markedly below pH 2.25 When using stock solution 2, to which urea had been added, the ceria precipitated during the heating of the stock solution together with the particles owing to the decomposition of the urea and subsequent rise in pH. The purpose of using urea in the stock solution was to achieve a controlled precipitation of the ceria onto the surface of the particles as a result of a uniform rise in pH throughout the solution owing to decompo- sition of urea at 80°C.However, SEM analyses revealed that the precipitates consisted of pm-sized particles of irregular shape. Such precipitates were found in the samples containing pure silica particles, aluminosilicate-modified silica particles and zirconia-coated silica particles. It was thus not possible to coat these particles with a uniform layer of ceria, probably for the same reasons as when stock solution 1 contained no urea. However, ceria was deposited, but in the form of irregular particles, on the surface of titania-coated silica particles. Fig. 6 shows a TEM image of ceria on titania-coated silica particles. The ceria seems to be concentrated in certain areas on the surface of the particles, which indicates that only a fraction of the total surface area in the system consists of ceria.The difference between titania- and zirconia-coated silica particles observed with respect to deposition of ceria can be explained by the marked difference in solubility between the two metal oxides at low pH values. It was also found that the speed of addition of the different sols to stock solution 2 did not affect the deposition of the ceria. The use of the stock solution containing ammonium hydrox- ide (3) resulted in satisfactory deposition of ceria on the particle surface when using titania-coated silica particles. Fig. 7(a) and (b)show SEM and TEM images, respectively, of a 500 nm titania-coated silica particle, which has been almost completely covered with ceria.TEM-EDS analysis showed that the covered particles contained Si, Ti, Ce and S. The sulfur was present in the ceria crystallites with a sulfur :cerium mole ratio close to 2: 1. The pH of stock solution 3 was pH 1.9 compared to pH 1.2 in the previously used stock solution Fig. 6 TEM image of dispersed cerium(1v) oxysulfate from stock solution 2 on the surface of titania-coated silica particles Fig. 7 SEM (a) and TEM (b)images of titania-coated silica particles after deposition of cerium@) oxysulfate from stock solution 3 without ammonium hydroxide (2). When using titania/silica particles the pH of the stock solution is important since the solubility of titania increases with decreasing pH.It was not possible to obtain a dispersion of ceria on the surface of particles of other compositions than titania/silica. The results obtained when using other particles than titania/silica were the same as when using the previous cerium(1v) sulfate solution containing no ammonia (2). Fig. 8 shows the electrophoretic mobility of a titania/silica sol on which ceria has been deposited. The mobility of these particles is slightly negative between pH 3 and pH 8. The corresponding mobility for particles of pure cerium(1v) oxysulf- ate is shown in the same figure. The iep of cerium(1v) oxysulfate is at pH 4.5 which indicates that the ceria does not cover the entire surface of the titania-coated silica particles.When coat- ing silica particles with titania, a distribution of the thickness of the titania layer from one particle to another is obtained owing to the coating method used in this investigation. It is reasonable to assume that the silica particles which were among the last added to the titanium@) chloride solution, were coated with a thinner layer of titania compared to the ones first added. One possible explanation for the negative mobility of the ceria containing particles below pH 4.5 may thus be that the titania layers on some of the particles were too thin, which led to dissolution of the titania layers during the heating of the sol particles together with the acidic cerium(1v) sulfate solution. The dissolution of the thin titania layers then resulted in uncoated silica particles with a negative mobility.The fact that some of the particles were naked silica particles was also verified by TEM-EDS. No ceria was deposited on the surface of uncoated silica particles whereas a significant amount of ceria was deposited on the surface of titania coated silica particles. Effect of calcination temperature on the ceria coating The particles which had been successfully coated with ceria were examined by TG and Fig. 9 shows thermograms for pure cerium(1v) sulfate and ceria-coated titania/silica particles. Ce(SO,), *4H@ undergoes three different steps in the overall transformation to form Ce02 including the intermediates Ce( and Ce20(S0,),.26 The transformation into pure ceria in such a TG analysis is complete at ca.840 "C. The mass loss curve of the dried sol containing ceria-coated titania/silica particles comprises many different steps in the temperature range 100-500°C. These steps include loss of structural water and hydroxy groups on the metal oxides as well as loss of anions, especially chlorides and sulfates. It is of special interest to find out at which temperature all the sulfur has been removed from the particles, since sulfur inhibits catalytic As seen in Fig. 9, the coated particles lose no more mass after about 780 "C, suggesting that the transformation into the pure oxides has been completed. The ceria coating on the surface of the particles was also studied by TEM-EDS. After calcination up to 550°C, the fine-grained ceria coatings observed before calcination [Fig.7(b)] were transformed into nm-sized grains firmly anchored on the titania-coated silica particle surface (see Fig. 10(a)]. EDS measurements 100 200 300 400 500 600 700 800 900 TPC Fig. 8 Electrophoretic mobility as a function of pH for cerium(1v) oxysulfate sol (-) and for titania-coated silica sol on which cerium(rv) oxysulfate has been deposited (---) Fig.9 TG curves of cerium@) sulfate, Ce(SO,),-4H2O (-) and of titania-coated silica particles on which cerium@) oxysulfate has been deposited (~ ~--) J. Muter. Chem., 1996, 6(2), 213-219 217 Fig. 10 TEM images of titania-coated silica particles on which cerium@) oxysulfate has been deposited after calcination at 550 "C (a) and 800 "C(b)for 30 min 1 4000 1 t1 1 11 0 20 30 40 50 60 70 80 2Wdegrees Fig.11 Powder XRD pattern of ceria-coated titania/silica particles after calcination at 950°C for 1 h. Diffraction peaks of crystalline ceria (*) and the internal standard aluminium (Al) are labelled. revealed that sulfur still remained in the ceria coatings. However, after calcination at 800 "C for 30 min no sulfur signal was detected (detection limit 0.3 atom%), and the correspond- ing TEM image shows the occurrence of crystallites on the titania/silica surface (see Fig. 10(b)]. As seen in Fig. 10, the dispersed ceria grains on the particle surface are smaller on the particles calcined at 800°C [Fig. lo@)], compared to the 218 J. Muter.Chem., 1996, 6(2), 213-219 Fig. 12 TEM images of ceria-coated titania/silica particles of different sizes after calcination at 800"Cfor 30 min. (a)500, (b)240 and (c)80 nm. ceria grains on the particles calcined at 550°C. The obtained XRD powder patterns show that only CeO, was present as a crystalline phase in the samples which had been calcined at 800°C for 30 min or at 950°C for 1h (see Fig. 11). To successfully prepare catalyst-support particles with highly dis- persed ceria, the titania path in Fig. 1 using stock solution 3, should be followed. Coating particles of different size with ceria To investigate the usefulness of the developed method, silica particles with diameters of 80, 240 and 500 nm were coated with ceria by using the titania path in Fig.1. After the deposition of ceria, the pH of the sols was about 3.5. At this pH the equilibrium concentration of Ce4+ is low. The corre- sponding cerium content in the supernatant of the sols used was 40, 35 and 47 ppm, respectively, which means that almost all of the added ceria was deposited on the surface of the titania coated silica particles. Fig. 12(u)-(c) shows TEM images of ceria-coated titania/silica particles of 500 (a), 240 (b) and 80 nm (c) after calcination in air at 800 "C for 30 min. As seen in the micrographs, the ceria (dark particles) is well dispersed on the surface of the particles regardless of particle size. This shows that the method for dispersion of ceria developed in this investigation is applicable to particles in a wide size range (80-500 nm).The size of the ceria crystallites was found to decrease with decreasing silica particle size. These results clearly show that the new preparation method is applicable to the preparation of ceria-coated particles suitable for use in catalysts. The surface areas of the particles in Fig. 12(u)-(c) determined by the BET method were 5.2, 10.9 and 32.6 m2 g-', respectively. Conclusions A method for dispersing ceria on the surface of silica particles has been developed. The method comprises three separate steps. In the first step the surface of the silica particles was modified by aluminate to facilitate adsorption of titania poly- cations, which were adsorbed on the modified silica particles in the second step.Ceria was deposited on the surface of the titania coated silica particles in the third step. Using this method we have shown that it is possible to disperse ceria on titania coated particles regardless of the silica particle size. A high dispersion of ceria on silica particles requires an interlayer consisting of a metal oxide. The investigation showed that it was possible to disperse ceria on titania but not on zirconia, aluminate-modified silica particles or pure silica particles. The dispersed material probably consisted of cerium(1v) oxysulfate. The sulfur was removed by calcining the particles in air at 800°C for 30min, whereby the cerium@) oxysulfate was transformed into cerium dioxide. Particles with highly dis- persed ceria are of potential interest as starting materials for different catalysts such as diesel exhaust catalysts or catalysts for oxidation of large solvents. By combining these particles with noble-metals, catalysts of special interest for cleaning of diesel exhausts can be prepared.This work was financially supported by the Swedish National Research Council and the Swedish Board for Industrial and Technical Development. References 1 P. Maestro, J. Less-Common Met., 1985, 111,43. 2 U. Lavrencic Stangar, B. Orel, I. Grabec, B. Ogorevc and K. Kalcher, Sol. Energy Muter., 1993,31, 171. 3 J. G. Nunan, H. J. Robota, M. J. Cohn and S. A. Bradley, J. Catal., 1992,133,309. 4 S. H. Oh and C. C. Eickel, J. Catal., 1988, 112, 543.5 E. C. Su, C. N. Montreuil and W. G. Rotchild, Appl. Catal., 1985, 17,75. 6 K. C. Taylor, Catal. Rev. Sci. Eng., 1993,35,457. 7 J. van Doorn, J. Varloud, P. Meriaudeau and V. Perrichon, Appl. Catal. B, 1992, 1, 117. 8 U. Hoffmann and T. Rieckmann, Chem. Eng. Technol., 1994, 17, 149. 9 E. Matijevic and W. P. Hsu, J. Colloid Interface Sci., 1987,118,506. 10 W. P. Hsu, L. Ronnquist and E. Matijevic, Langmuir, 1988,4,31. 11 V. Briois, C. E. Williams, H. Dexpert, F. Villain, B. Cabane, F. Denevue and C. Magnier, J. Muter. Sci., 1993,28, 5019. 12 H. G. Brittain and P. S. Gradeff, J. Less-Common Met., 1983, 94, 277. 13 W. L. Albrecht, US Pat., 3 440 174,1969. 14 M. Mindick and L. E. Reven, US Pat., 3 139406,1964. 15 I-M. Axelsson, L. Lowendahl and J-E. Otterstedt, Appl. Catal., 1988,44, 251. 16 W. Stober, A. Fink and E. Bohn, J. Colloid Interface Sci., 1968, 26,62. 17 G. Charlot, in Colorimetric Determination of Elements, Elsevier, Amsterdam, 1964, pp. 408-410. 18 I. M. Kohlthoff and P. J. Elving, in Treatise on Analytical Chemistry, Wiley, New York, 1961, Part 2, vol. 5, p.130. 19 W. Westwood and A. Mayer, Analyst, 1948,73,275. 20 R. K. Iler, J. Colloid Interface Sci., 1976,55,25. 21 E. A. Barringer and H. Bowen, Langmuir, 1985,1,420. 22 L. Vordonis, A. Akratopulu, P. G. Koutsoukos and A. Lycourghiotis, in Preparation of Catalysts IV, ed. P. Delmon, P. Grange, P. Jacobs and G. Poncelet, Elsevier, Amsterdam, 1987, p. 309. 23 R. K. Iler, in The Chemistry of Silica, Wiley, New York, 1979, p. 185. 24 R. K. Iler, in The Chemistry of Silica, Wiley, New York, 1979, p. 408. 25 C. F. Baes and R. E. Mesmer, in The Hydrolysis of Cations, Wiley, New York, 1976, pp. 150,159. 26 Gmelin Handbuch der anorganischen Chemie, Sc, Y, La-Lu, ed. H. Hein, I. Hinz, P. Merlet and U. Vetter, Springer-Verlag, Berlin, Heidelberg, New York, 1981, vol. C8, p. 122. 27 C. H. Bartholomew, Chem. Eng., 1984,12,96. Paper 5/05708F; Received 29th August, 1995 J. Muter. Chern., 1996, 6(2), 213-219 219
ISSN:0959-9428
DOI:10.1039/JM9960600213
出版商:RSC
年代:1996
数据来源: RSC
|
16. |
Iron-doped zirconium silicate. Part 1.—The location of iron |
|
Journal of Materials Chemistry,
Volume 6,
Issue 2,
1996,
Page 221-225
Frank J. Berry,
Preview
|
PDF (628KB)
|
|
摘要:
Iron-doped zirconium silicate Part 1.-The location of iron Frank J. Berry,*" Desmond Eadon,b Joanne Holloway" and Lesley E. Smarta aDepartmentof Chemistry, The Open University, Walton Hall, Milton Keynes, UK MK7 6AA bJohnson Matthey Technology Centre, BEount's Court, Sonning Common, Reading, UK RG4 9NH Iron-doped zirconium silicate (zircon) has been shown by X-ray powder diffraction (XRD), energy dispersive X-ray analysis (EDX) and 57Fe Mossbauer spectroscopy to consist of inclusions of a-Fe203 together with discrete paramagnetic Fe3+ species within the zircon structure. Electron paramagnetic resonance (EPR) spectra show that the Fe3 species in the samples containing + less than 0.2 mass% Fe occupies low-symmetry rhombic sites, but that in materials with higher concentrations of iron the Fe3+ also occupies other locations, including sites with axial symmetry.The colour of the iron-doped zircon is due to both the low symmetry Fe3+ species and the inclusions of a-Fe203. The compound ZrSiO, is found as the naturally occurring mineral zircon. The structure of zircon consists of a garnet- related structure with edge-sharing chains of ZrOS4+ dodeca- hedra and Si04,- tetrahedra.' Zircon can be made directly by calcining a mixture of zirconia (ZrO,) and silica (SO2) above 1500 "C. However, calcining zirconia and silica in the presence of salts such as alkali-metal halides (known as mineralisers) significantly reduces the temperature of the preparation2-, to about 1000 "C. The lower-temperature reaction, when per- formed with metal oxides or sulfates, is used commercially to produce metal-doped zircon pigments which are stable at the high temperatures commonly used in the ceramics industry.There are three well known metal-doped zircon pigment^:^,^ vanadium imparts a strong blue colouration, praseodymium gives yellow and the iron-doped zircon gives coral colours. The structural properties of the and praseody- mium-d~ped~*~?~zircons have been the subject of considerable investigation, but little is known of the properties of the iron- doped zircon.2.6 Our aim has been to elucidate both the mechanism of formation of the iron-doped zircon using mineralisers (which will be reported elsewhere7) and the fundamental properties of the iron-doped zircon.We report here on the nature and location of iron within zircon and its influence on the colour of the pigment. Experimental Iron-doped zircons were prepared, with iron concentrations ranging from 0.19 to 1.82 mass%. Monoclinic ZrO, (1 mol), SiO, (1 mol), FeS0,-7H20 (0.1-0.25 mol), NaCl(O.1-0.3 mol), NaF (0.25-0.67 mol) and MN03 (0.03-0.1 mol; M=alkali metal) were mechanically mixed (5 min). Water (15 cm3) was added and mixing continued (5 min). This reactant mixture was heated in a covered, unglazed ceramic crucible to 1060°C with a ramp rate of 100°C h-l, held at 1060°C (6 h) and cooled in the furnace. All samples were acid-washed to remove unreacted mineral- isers and excess a-Fe203 from the product. A ground sample (20 g) was refluxed in 50 cm3 aliquots of 5 mol dmP3 HC1 until the washings were colourless, and the solid product was washed with water and dried at 120°C (12 h). Iron-impregnated zirconia and iron-impregnated zircon samples containing 200 ppm iron were prepared for EPR spectroscopy from an aqueous solution of iron(II1) nitrate nonahydrate.The slurries were dried in air at 120°C (12 h). Samples were calcined at 300, 500, 700, 1060 and 1450°C. EPR spectra were obtained before and after acid washing. X-Ray powder diffraction (XRD) patterns were recorded with a Siemens D5000 X-ray diffractometer using Cu-Ka radiation. Inductively coupled plasma-atomic emission spectroscopy analyses (ICP-AES) were recorded on a Thermo Jarrel Ash Atom Scan G1E ICP-AES instrument. Samples (0.5-0.1 g) were fused with sodium peroxide and dissolved in concentrated HCl.The samples were diluted to 100ml with 10% HCI. X-Ray fluorescence (XRF) analysis was performed on a ARL 8420 + dual goliiometer wavelength-dispersive XRF spec- trometer, using a 3 kW Rh tube, with flow proportional and scintillation counters. Glass discs were produced by 20 min fusion of 1 part dried sample and 5 parts dried lithium metaborate/tetraborate flux (Johnson Matthey Spectroflux 100B) in Pt-S%Au crucibles at 1100 "C. The melt was swirled repeatedly to ensure complete dissolution and homogenisation and then moulded and pressed to form a 1.5 mm thick disc. 57Fe Mossbauer spectra were recorded with a micropro- cessor-controlled Mossbauer spectrometer using a 25 mCi 57Co/Rh source at 298 K.The drive velocity was calibrated with a 57Co/Rh source and a natural iron absorber. The spectra were computer fitted. All chemical isomer shift data are quoted relative to metallic iron. EPR spectra were recorded at 298 K, using a Varian E-line Century Series spectrometer operating in the X-band region over a field range of 5000 Gauss at a frequency of cu. 9.17 GHz. Energy-dispersive X-ray (EDX) analysis was performed on a JEOL FX2000 microscope under an accelerating voltage of 200 kV and using a beam of diameter 14 nm. X-Ray photoelectron spectra (XPS) were recorded with a Leybold-Heraeus LH-10 spectrometer operating under a vacuum of (2 x 10-6)-(2 x Pa using Mg-Ka radiation.The spectra were accumulated over 25 scans and binding energies were measured relative to carbon 1s at 284.8 eV. Temperature-programmed reduction (TPR) profiles were recorded using a Varian Series 300 katharometer connected to a programmable tube furnace. A reducing gas mixture of 10% hydrogen-90% nitrogen (20 cm3 min-l) was passed over the samples (ca. 30mg) which were heated at-a-rate of 10 "C min-' from 25 to 1000 "C. Spectral reflectance measurements were recorded from ground materials using a Micro-match ICS QMM 200 reflec-tance spectrophotometer which was calibrated against stan- dard white (total reflectance) and black (total absorbance) tiles. The reflectance curves were obtained by andysing the light J. Muter.Chem., 1996,6(2), 221-225 221 reflected at an angle of 10" from the surface of the coloured samples. Measurements were recorded with a small aperture and a standard D65 light source. The colour coordinates, L*, a*, b*, C* [C* = + b*2)+] and H,b (Hub=tan-'b*/a*),were measured according to the CIELAB (CIE 1978) trichromatic system of colour measurement. Results and Discussion The XRD results showed that the acid-washed materials contain a zircon-related phase with lattice parameters the same, within experimental error, as those quoted in the literature' for pure zircon. No trend in lattice parameters with different concen- trations of iron was observed. The XRD patterns for all samples showed the presence of a small amount of unreacted zirconia.No unreacted silica was observed, although evidence for the presence of a glassy silicon-containing phase was obtained (vide infra). The XRD patterns showed no evidence for the presence of mineralisers. The presence of a-Fe203 was detected in samples containing 0.51 mass% Fe and above. The concentration of iron in the samples was measured using ICP-AES analysis and XRF analysis; the results are summarized in Table 1. The values show that only a small proportion of the reactant iron is incorporated into the zircon, and that this increases with increasing iron content of the reaction mixture. All subsequent discussion refers to the con- centration of iron determined by ICP. The 57Fe Mossbauer spectra were best-fitted to a quadrupole split absorption characteristic of high-spin Fe3 + and a sextet pattern characteristic of a-Fe203 (Fig.1). The Mossbauer parameters are collected in Table 2. The results confirm the XRD evidence for the occurrence of a-Fe203 in samples shown by ICP analysis to contain more than 0.51 mass% Fe, but also reveal the presence of the magnetically ordered a-Fe203 phase, which is presumably below the limit of detectability in XRD, in the sample with a lower concentration of iron [Fig. l(a)]. The dominance of the doublet in the spectrum recorded from the sample containing 0.22 mass% iron [Fig. l(a)] suggests that low concentrations of iron are prefe- rentially accommodated within the zircon structure as para- magnetic Fe3+ species, the site symmetries of which have been examined by EPR spectroscopy (vide infra).The results also show that the concentration of a-Fe203 increases in materials with increasing concentrations of iron. EDX analysis of all the acid-washed iron-doped zircon samples showed that zirconium and silicon are evenly distrib- uted. Materials containing less than 0.22 mass% Fe showed an even distribution of iron throughout the crystals examined, whereas samples with higher concentrations of iron contain additional iron-rich regions. This observation is in agreement with the Mossbauer spectra and XRD results, which both indicated the presence of a-Fe203 in the samples with higher concentrations of iron, and suggests that the iron-rich areas are inclusions of a-Fe,03 while the paramagnetic Fe3 + species are distributed throughout the zircon structure.Given that the lattice parameters of the iron-containing zircon-related phase do not vary with iron content, we associate Table 1 Nominal and measured concentrations of iron in iron-doped zircon concentration of iron/mass% nominal concentration/mass% determined by ICP determined by XRF 0.77 0.19 0.42 3.01 0.22 0.43 3.73 0.51 0.82 7.2 0.96 1.42 10.42 1.55 2.30 13.43 1.82 2.56 100.05 99.851 .-E 100.00 5b 99.50 98.5099.001 I (b) + 98.00.j-14 I -10 I -6 I -2 I 2 6 Ib velocity/mm s-1 Fig. 1 57Fe Mossbauer spectra recorded at 298 K of (a) 0.22 mass% Fe and (b) 1.55 mass% Fe iron-doped zircon samples Table 2 "Fe Mossbauer parameters recorded from acid-washed iron- doped zircon at 298 K iron content/ 6 (*0.05)/ d (f0.0:)/ H (f4)/ spectral area mass% Fe mm s-mm s-kG ("/.I 0.22 0.41 521 43 0.36 1.14 57 1.55 0.38 522 89 0.36 1.14 11 1.82 0.40 509 93 0.34 1.16 the paramagnetic Fe3 species as being accommodated within + interstitial sites in the garnet-related structure.Furthermore, since a-Fe203 forms at a lower temperature than zircon, we envisage the initial formation of particles of a-Fe20, which are trapped within the developing zircon matrix and protected from leaching by acid. This is supported by the TPR profiles recorded from all the acid-washed iron-doped zircon samples which showed no consumption of hydrogen at temperatures up to 1000 "C [Fig.2(a)].This is in stark contrast to the profiles observed for the unwashed samples [Fig. 2(b)] which produced a single sharp peak due to the reduction of a-Fe203 on the surface of the material. The results are consist- ent with the iron in iron-doped zircon being embedded within the structure and thus protected from reaction with hydrogen by the surrounding zircon matrix. XPS showed both an enhanced level of silicon and oxygen at the surface and the presence of potassium and fluorine. The amount of iron at the surface after acid washing was very small (<0.03 mass%) indicating that iron, on and near to, the surface of zircon is leached out by dilute acid. We associate the high silicon content and the presence of fluorine with a glassy silicon-containing phase, which is not detected by XRD, but which originates during the formation of zircon through a silicon-fluorine intermediate.We will report on the mechanism of formation of iron-doped zircon in a separate p~blication.~ 222 J. Muter. Chem., 1996, 6(2), 221-225 797'C Fig. 2 TPR profiles: (a) 1.55 mass% Fe, acid-washed iron-doped zircon; (b) 1.82 mass% Fe, unwashed iron-doped zircon The EPR spectra of acid-washed iron-doped zircon samples were found to be very complex, showing many overlapping resonances (Fig. 3 and Table 3). To aid assignment of surface and sub-surface species, EPR spectra were also recorded from unwashed and acid-washed iron-impregnated zirconia and iron-impregnated zircon samples following calcination at vari- ous temperatures. The spectrum of pure zirconia, ZrOz, showed a small broad resonance at g=4.16 which we assign to high- spin Fe3+ in rhombic sites: it is predicted that a rhombic site with asymmetry ratio A=E/D= 1/3 gives similar g,, g,, and g, values coinciding at about g=4.2.899 The result indicates that reactant ZrO, contains a small amount of iron impurity.Since ZrO, is commonly oxygen-deficient, we associate the sharp signal observed at g= 2.07, which decreased significantly in intensity after calcination at 500°C for 6 h, with electrons trapped at oxygen vacancies. The EPR spectra of the iron-impregnated zirconia samples both before and after calcination and before and after acid washing (Table4) are shown in Fig.4.The spectra of the unwashed materials are similar to those observed in analogous materials and reported previously," and show that as the calcination temperature increases the sharp peak at gw2 diminishes as a result of the annealing of anionic vacancies in the ZrO, structure. No spectra have previously been reported for acid-washed iron-impregnated zirconia. The results in Fig. 4 show that acid- washing results in an increase in the relative intensity of the gz2 peak. We envisage that leaching of the Fe3+ in surface sites generates additional anionic vacancies to balance the increasing charge. The broad peak at gz2 is assigned' to low- Table 3 Measured g-values of iron-doped zircons concentration of iron/mass% resonance 0.19 0.22 0.51 0.96 1.55 1.82 a 7.93 8.02 7.82 7.59 7.81 8.07 b 5.63 5.58 5.69 5.63 5.78 C 5.29 5.29 5.29 d 4.93 4.93 4.92 4.91 4.98 e 4.22 4.22 4.22 4.23 4.20 4.35 f 3.52 3.55 3.54 3.58 3.54 gh 2.01 3.51 2.02 3.52 2.02 3.52 2.06 3.52 2.02 3.50 2.04 i 1.98 1.99 1.99 1.99 1.99 1.99 kj 1.84 1.64 1.84 1.63 1.84 1.64 1.64 1.64 1.66 1 1.48 1.48 1.48 1.48 1.48 k b Fig.3 EPR spectra of acid-washed iron-doped zircon at 298 K: (a) 0.19; (b) 0.22; (c) 0.96; and (d) 1.55 mass% Fe Table 4 Measured g-values of iron-impregnated zirconia, before (u) and after (w) acid washing iron-impregnated zirconia: calcination temp/"C resonance pure ZrO, not calcined 300 500 700 1000 1450 a u 8.98 8.91 8.92 8.93 8.70 8.81 W 9.30 9.30 9.30 9.30 b u 4.32 4.32 4.32 4.32 4.33 4.27 w 4.34 4.35 4.35 4.35 4.35 4.35 C U 4.22 w 4.27 4.29 4.27 4.27 4.27 4.29 d 4.16 u 4.19 4.20 4.19 4.20 4.19 4.14 w 4.19 4.19 4.19 4.18 4.18 4.17 e u 2.03 2.03 2.03 2.03 2.03 W 2.04 2.04 f 1.98 u 1.99 1.99 1.99 1.99 1.99 W spin Fe3+ ;this gradually decreases in intensity with increasing calcination temperature and disappears completely from the spectra of materials heated at 1450 "C.The broad peak at g z2 is also absent in the spectra of acid-washed materials which had been calcined at lower temperatures [Fig.4(u) and (b)].The J. Muter. Chem., 1996, 6(2), 221-225 223 Fig. 4 EPR spectra of iron-impregnated zirconia at 298 K: (a)prior to calcination, and calcined for 2 h at: (b) 300; (c) 500, (d) 700, (e) 1060;and (f)1450°C. u, unwashed and w, acid-washed samples. results suggest that Fe3+ species which are weakly bound to the surface can be removed by acid washing. It also appears that these species gradually move into the ZrOz structure under the influence of heating and that this begins at about 500"C, a significantly lower temperature than was previously thought." As the calcination temperature increases, both the number of peaks and the relative intensities of the peaks in the region around gz 4.2 increases and, for materials calcined at tempera- tures above 500"C, these peaks remain in the spectra recorded after acid washing.The results indicate that the peaks at gz4.2 are initially due to high-spin Fe3+ species in low-symmetry sites on the s~rface,*~~ and that as the temperature is increased, these Fe3+ species are gradually incorporated into the structure thus creating more than one type of site and, because these species occupy subsurface sites, they are resistant to acid leaching. The gradual appearance of a weak line at gz 8.9 in materials calcined at higher temperatures could be an indication of the Fe3+ moving into higher symmetry, and thus possibly, substitutional sites in the structure. The EPR spectrum of pure zircon formed by the solid-state method without mineralizers, showed resonances in two areas.The broad unresolved doublet at gz4.3 and the broad reson- ance superimposed by a sharp resonance at gz 2, indicate the presence of small quantities of iron impurities; these could originate from the reactant ZrO,. The EPR spectrum of the zircon impregnated with iron(1n) nitrate and dried at 120°C showed a very similar spectrum, although the peaks in the gz 2 region decreased in intensity for the acid-washed mate- rials, indicating that the surface iron species formed during the drying process can be easily removed. After calcining the iron- impregnated zircon samples at increasing temperatures, it was observed that both peaks in the gz2 region decreased in intensity, suggesting that the Fe3+ responsible for these signals had gradually migrated into the structure.This was endorsed by the observation that acid washing further reduced the intensity of the broad gz2 peak for the samples calcined at temperatures less than 700 "C but had no effect on the samples calcined above 700°C. This suggests that under the influence of heat, iron(II1) species adsorbed on the surface gradually move into surface sites where they cannot be removed by acid. The gz 4.3 resonance indicative of high-spin Fe3 species in + low-symmetry sites' did not change significantly in the spectra recorded from any of the samples, and was not affected by acid washing. On the basis of the results reported above, some distinct trends in the more complex spectra recorded for the iron- doped zircons can be distinguished. The spectrum of the iron- doped zircon containing 0.19 mass% Fe showed two broad resonances centred around gz 4.2 and gz 2 which are assigned, as before, to Fe3+ in rhombic sites in the structure and to Fe203.As the concentration of iron in zircon increases, these two resonances can still be distinguished but the spectra also show other resonances, in particular at gz 5.7, and at gz4.9 and gz3.5. Fe3+ in low symmetry sites with A=O.25 is calculated to have g,, g,,, and g, values of 4.86, 4.28 and 3.53, respectively. The results indicate that in the materials with iron concentrations above 0.19 mass% Fey the Fe3+ occupies more than one low-symmetry site in the structure. When the Fe3+ has axial symmetry (A =0 for tetragonal symmetry), then the values of glland g, are predicted to be 2 and 6, respectively; 224 J. Muter.Chem., 1996,6(2), 221-225 other g values can be accounted for by 1 with values between 0 and 1/3." Our spectra indicate, therefore, that as the concentration of iron increases, then the number of different sites occupied by the Fe3+ ions also increases, with some of the iron(111) species being located in sites with axial symmetry. 1OC (a) I 80 h z.60 E8 40 e 20 . II9 1 400 500 600 700 800 wavelengthlnm 100 90 80 3 uig 70 6 c .o,60-50 1 I I I 0.0 0.5 1.o 1.5 2.0 44n 0.0 0.5 1.O 1.5 2.0 40 c2-20 Ei? c 10 0: 1 I I I 0.0 0.5 1 .o 1.5 2 .o iron concentration/mass% Fe Fig.5 Iron-doped zircon with iron concentrations of: A, 0; A, 0.19; 0,0.22; H, 0.51; 0,0.96; x, 1.55 and El, 1.82 mass%: (a) spectral reflectance curves; (b)lightness, L*; (c)hue angle, Hab;(d)chroma, C* The location of the Fe3+ species may be associated with the lattice parameter data, which was insensitive to the amount of iron incorporated in the zircon structure. We suggest that the majority of the Fe3 species are accommodated in interstitial + rhombic and axial sites. The colour of the iron-doped zircons varied from almost white (<0.19 mass% Fe) through pale peach (0.19-0.51 mass% Fe) to dark red-brown (> 0.96 mass% Fe) similar to the colour of pure a-Fez03.It is quite striking that the colour of the iron- doped zircon containing 1.82 mass% Fe is very similar to that of macroscopic a-Fe,O, . The commercially produced iron- doped zircon produces coral colours in ceramic glazes, whereas pure a-Fe,03 gives an almost colourless glaze. The spectral reflectance curves for all the acid-washed iron-doped zircon samples are shown in Fig. 5(u). The results show that pure zircon reflects almost all of the light resulting in a blue-white colour. The amount of light reflected decreases sharply for the samples containing less than 0.51 mass% Fe, but the changes thereafter are much less marked. The reflectance curves for materials with concentrations of iron greater than 0.96 mass% are very similar to each other and to that of a-Fe,03.Fig. 5(b)-(d) show the values of lightness, L*, hue angle, Hob, and chroma, C*, plotted against the total iron concentration. It is evident that for each of the four plots in Fig. 5, a limiting value is reached at an iron concentration of about 1%. The results from XRD, Mossbauer spectroscopy and EDX all indicate that the proportion of a-Fe,O, increases signifi- cantly in the zircons containing higher concentrations of iron. Hence the results suggest that at concentrations of iron greater than 1 mass%, the colour properties are all dominated by the presence of the a-Fez03. However, at lower concentrations it is evident that the colour also originates from the presence of paramagnetic Fe3 ; at these concentrations the EPR spectra + indicate that Fe3+ is in low-symmetry rhombic sites.Conclusion Iron-doped zirconium silicate containing low concentrations of iron (<cu. 1.0 mass%) formed by the reaction of zirconia, silica, iron@) sulfate and mineralizers at 1060 "C and sub- sequently washed in 5mol dmP3 HC1, contains Fe3+ in low- symmetry rhombic sites together with inclusions of a-Fez03. In materials with higher iron content (>cu. 1.0 mass%) the Fe3 also occupies higher-symmetry sites and the amount of + included a-Fe203 increases. We associate the colour of these materials both with Fe3+ in low-symmetry sites and with the presence of inclusions of a-Fe203, the latter dominating the colour properties at higher Fe concentrations.We thank Mr. Trevor Bell for helpful discussions, Dr. JosC F. Marco for measuring some of the Mossbauer and XP spectra, Drs. Peter Webb and John Watson for the XRF data and Johnson Matthey plc for the award of a studentship (to J.H.). References 1 G. V. Gibbs and K. Robinson, Am. Mineral., 1971,56,782. 2 B. T. Bell, Rev. Prog. Coloration, 1978,9,48 and references therein. 3 R. A. Eppler, Am. Ceram. SOC. Bull., 1977, 56, 213 and references therein. 4 N. C. Wildblood, Trans.J. Brit. Ceram. SOC.,1973,72, 3119. 5 L. Bragg and G. F. Claringbull, Crystal Structure of Minerals Volume IV, ed. L. Bragg, G. Bell and Sons Ltd, London, 1965. 6 M. S. Bibilashvili, 0. S. Grum-Grzhimailo and N. S. Belostotskaya,Glass Ceram., 1982,39,606, 7 F. J. Berry, D. Eadon, J. Holloway and L. E. Smart, to be published. 8 J. R. Pilbrow and M. R. Lowrey, Rep. Prog. Phys., 1980,43,28. 9 R. Aasa, J. Chem. Phys., 1970,52,3919. 10 J. C. Evans, C. R. Owen and C. C. Rowlands, J. Chem.SOC.FaradayTrans. 1,1989,85,4039. 11 B. Henderson, J. E. Wertz, T. P. P. Hall and R. D. Dowsing, J. Phys. C: Solid State Phys., 1971,4, 107. Paper 5/04431F; Received 6th July, 1995 J. Muter. Chem., 1996, 6(2), 221-225 225
ISSN:0959-9428
DOI:10.1039/JM9960600221
出版商:RSC
年代:1996
数据来源: RSC
|
17. |
Chemomechanical polishing of silica and silicon by fluoride- and oxide-based reagents: identification of a reaction intermediate |
|
Journal of Materials Chemistry,
Volume 6,
Issue 2,
1996,
Page 227-232
David S. Boyle,
Preview
|
PDF (1259KB)
|
|
摘要:
Chemomechanical polishing of silica and silicon by fluoride- and oxide-based reagents: identification of a reaction intermediate David S. Boyle and John M. Winfield" Department of Chemistry, University of Glasgow, Glasgow, UK G12 8QQ The effect of added hydrogendifluoride anion on the chemomechanical polishing of silica and silicon wafers by aqueous suspensions of cerium(1v) or silicon(1v) oxides has been investigated over a range of solution pH. The effect on silica is marked at very low pH; an intermediate under these conditions has been identified as K2SiF6 coated with a thin silica-like layer. The [HF,] -anion is ineffective for silica polishing above pH 7 and for silicon under all conditions examined. The most effective reagent for the latter substrate is a mixture of cerium(1v) oxide and alkaline silica sol.An explanation for the role of [HFJ is offered. The phenomenon of polishing, where the aim is to produce a specular reflective surface which is free of damage and artifacts, has been known for many centuries. The earliest characteris- ation of optical surfaces was by visual inspection of the surface finish by those who ground lenses to shape, then polished the component to remove the matt finish. Today the science of polishing occurs beyond the resolution of the human eye, as subnanometre finishes are required to reduce light scattering in high-performance optical systems.' The mechanisms involved in the polishing of optical glass have been the focus of interdisciplinary attention and as a result, it is accepted by most authorities that the chemomechanical pathway is correct., Chemomechanical polishing is the term given to the process in which a controlled chemical etch of a substrate is allied with the mechanical action of a polishing pad and where appropriate, polishing particles. In a preliminary communication3 we reported results which provided direct evidence for the formation of a reaction intermediate, formed when silica glass (Spectrosil B) is polished using a mixture of potassium hydrogendifluoride, cerium(1v) oxide and sucrose in aqueous hydrochloric acid.The intermedi- ate behaves as a passivating layer,4 an insoluble or sparingly soluble compound produced at the solid-solution interface which may impart a degree of selectivity in the chemomechan- ical polishing process.The concept of a passivating layer is an important development in understanding the process of chemo- mechanical polishing, since it brings together the chemical, for example the surface reaction occurring, and the mechanical aspects, for example in changes in surface morphology and the formation of small particles, of the process. It was developed Experimenta1 General methods Silica glass (ca. 2 x 1 x 0.5 cm3; Spectrosil B, Multilab Ltd) and silicon (ca. 2 x 1x 0.1 cm3; p-type [1001; MCP Wafer Technology Ltd.) wafers were lapped following a standard procedure. Polishing was accomplished using Logitech PM2A and PP5 vacuum jig equipment; where chemically aggressive reagents (fluoride-based) were investigated, a Logitech CP2000 precision polishing unit was employed. As many of the mate- rials in this work were intractable to IR spectroscopy in the transmission mode, spectra were recorded using a Nicolet 5DX C spectrometer equipped with a SpectraTech collector for diffuse reflectance (DRIFTS) and a MTEC 100 cell for photo- acoustic (FTIR-PAS) analysis.A Nicolet Nic-Plan FTIR microscope with an ATR (ZnSe crystal) accessory was used to obtain microspectroscopic and ATR data. 29Si MAS NMR spectra were recorded using a Bruker AM 300/WB spec- trometer equipped with a 7 T superconducting magnet and 9cm bore. Powder XRD measurements were made using a Philips PW 1050-35 diffractometer with a vertical goniometer and Co-Ka radiation.Determinations of the surface roughness and planarity of the substrates were achieved using a Rank Taylor Hobson Talystep stylus instrument (STD conical dia- mond stylus, 0.155 mm radius; loading 10-30 pN) which allowed reproducible measurements to be taken over an ambi- ent temperature range of 278-3 13 K, Logitech optical interfer- ometry testing systems and a Zeiss Nomarski microscope. Scanning electron microscopy (SEM) of specimens was per- to explain the polishing of silica by aqueous oxide s~spensions~~ formed using but its application to situations that involve fluoride-based reagents is attractive in view of the well known effect of aqueous hydrogen fluoride or the hydrogendifluoride anion on the rates of silica dissolution in etching and related situ- ation~,~coupled with the relatively low solubility of hexa- fluorosilicates such as K2SiF6.5b Particularly intriguing is the suggestion that etching can be inhibited by the application of an electric field owing to the formation of an insoluble layer.5" As part of a general investigation of polishing and etching of optical, electronic and opto-electronic materials,6 this paper describes the development of the silica glass polishing work3 and its extension to studies of silicon substrates.Under typical experimental conditions, in the open atmosphere or in aqueous solution containing dissolved oxygen, a silicon substrate is covered with an oxide layer. As a consequence, any polishing process for silicon will involve the removal of the oxide layer in the initial stages.A rationale of the behaviour of fluoride- based reagents and conventional oxide-based reagents for the chemomechanical polishing of silica and silicon is presented. a Philips 500 scanning electron microscope. Samples were gold coated using a Polaron SC 515 SEM coating unit to prevent irregular charging of the specimen surface in the SEM. ['gF]-labelled CsF (18F; tt =110 min, p+), used to prepare aqueous solutions of ["F]-labelled hydrogen-difluoride anion, was made by acid distillation of [18F]-HF into a solution of caesium hydroxide at 273 K, after neutron irradiation (3.6 x 10'' neutron cm-, s-' for 50 min) of lithium carbonate using the sequence 6Li(n,a)3H, '60(3H,n)'8F.Radiochemical purity was determined by half-life measurement. Count rates were determined using a NaI scintillation counter with a scaler ratemeter (Ecko and NE), background and decay corrections being applied. Preparation of polishing reagents The oxide-based reagents were an aqueous suspension of cerium(1v) oxide (Opaline, 0.5 pm; RhBne Poulenc), an alkaline silica sol (Syton, pH 10-10.5; Monsanto Ltd), the fluoride- modified silica sol (pH % 8) used previously to polish LiNb036a J. Muter. Chem., 1996, 6(2), 227-232 227 and a reagent described by Pilkington Ltd (pH z 1)for finishing glass surface^.^ Owing to their corrosive nature, plastic appar- atus was employed for the preparation of the fluoride-based reagents.The composition of fluoride-modified silica sol was potassium hydrogendifluoride ( 1.00 g, GPR; 95% acidimetric assay; BDH) in Syton (125 cm3). The Pilkington reagent was prepared as f01lows.~ Sucrose (450 g, GPR; BDH) was added slowly, with stirring, to a solution of hydrochloric acid (120 cm3 HC1, 36% m/m; M&B Ltd; in 400cm3 distilled water). Potassium hydrogendifluoride (270 g) was added to the solu- tion, stirring until dissolution was complete, then cerium(1v) oxide (50 g, Opaline, 0.5 pm; Rh8ne Poulenc) was added to produce a thick suspension. The neutraliser solution was composed of an aqueous solution of sodium carbonate (100 g dmV3, GPR; M&B Ltd) with cerium(1v) oxide (10 g dm-3) and was applied (to the polishing plate and sample) immedi- ately at the end of an experiment.When effervescence due to the evolution of CO, ceased, the sample and polishing plate were washed with distilled water. Solution pH values for polishing and etching reagents were determined to a precision of & 0.1 pH unit with antimony M45-PP/R44 2-EG electrodes (Unicam) and an EIL 7050 pH meter. Polishing experiments Typical polishing conditions employed were a plate (polyure- thane grade LP87) rotating at 30-40 rpm under minimal loading and a reagent feed rate of ca. 500 cm3 h-l. When the surface finish was acceptable by visual inspection, the sample was cleaned thoroughly under running water then examined using a Zeiss Nomarski microscope. When lapping damage was no longer visible, the surface roughness was determined using a Talystep stylus instrument.Polishing reagents were assessed by their ability to produce rapidly a surface finish which approached the subnanometre level. Surfaces were com- pared using the parameter RA, the arithmetic mean of the departures of the roughness profile from the mean line. In all cases the RA value was the result of a trace obtained at the centre of a wafer over a 500 pm horizontal displacement. Dip-etch experiments The identification of the active species in the acidic [HF,]-, CeO, reagent7 was determined by a series of dip-etch experi- ments. The behaviour of single and binary combinations of each component towards silica and silicon was examined. Solutions of each component were prepared using concen-trations appropriate to those in the polishing reagent.Binary combinations of HCl-KHF, were brought to the desired pH by dropwise addition of concentrated aqueous ammonia (0.88 specific gravity) to solutions of KHF, (1.0 or 2.0 mol dm-3) in aqueous HC1 (1.04 mol dm-3). Solutions were made up to 250 cm3 with distilled water. Polished silica or silicon wafers (ca. 2 x 1 cm2), degreased and cleaned by ultrasonic agitation for 5 min in concentrated nitric acid (GPR, 68% m/m; Rhane Poulenc plc), Genklene (stabilised CH3CCl,; ICI plc) and isopropyl alcohol (GPR, M&B Ltd), were added to solutions (ca. 100cm3) in plastic beakers and agitated by magnetic stirring for 0.25, 0.5, 1 and 24 h at room temperature. In order to minimise experimental artifacts, studies were repeated using Perspex wafer holders.Wafers were removed using plastic forceps, rinsed under running water then distilled water, oven- dried and allowed to cool in a desiccator. The wafers were reweighed (precision & 0.0001 g; the handling and weighing procedures were standardised to ensure reproducibility) and changes in mass recorded. Radiotracer experiments During the course of both polishing and etching experiments employing the hydrogendifluoride anion at low pH, a layer 228 J. Muter. Chem., 1996,6(2), 227-232 often developed on the surface of a silica wafer, and spectro- scopic and other methods indicated that fluorine had been incorporated. ["F]-Labelled [HF,] -was used to determine both the degree of fluorine incorporation (over a wide pH range) and the lability of the fluorine species with respect to radiochemical exchange. Fresh silica, silica which had been etched and 'layer material' (isolated in separate experiments) of similar mass and geometry were investigated.Similar strategies were employed for silicon substrates. Two experimental procedures were employed. Preliminary investigations involved the addition of measured masses of ["FI-labelled CsF to FEP (fluorinated ethylene-propylene co-polymer) counting vessels containing aqueous solutions of hydrogendifluoride anion (of known concentration and of different pH values) and the substrate. The mass of CsF in each FEP vessel was not identical and therefore the ["F] retained on the substrate was expressed as a percentage of the solution activity at the beginning of the experiment.In sub- sequent experiments, substrates were exposed to aliquots of a standard solution containing C1'F]-labelled [HF,] -(of known concentration and at a measured pH). The specific count rate of the standard solution was determined, typical values being 106-107 count s-l (atom F)-' and hence the incorporation of fluorine could be determined. In all experiments, a series of counts was determined for FEP vessels containing substrates and aliquots of the ['8F]-labelled solution. The substrates were removed, dried and counted in fresh FEP vessels. Counts were corrected for decay and background and count rates (count s-l) determined. It was assumed that the efficiency of counting the solid and solution were equal; previous work at Glasgow had indicated this was a reasonable assumption. Results Polishing of silica and silicon Samples of silica were polished using the four reagents described above: an aqueous suspension of cerium(w) oxide, an alkaline silica sol, fluoride-modified silica and an aqueous acidic fluoride-based reagent7 Talystep profiles of surface roughness were obtained and the results plotted as envelopes of surface finish us.time (Fig. 1) using ca. 10-12 data points for each reagent to define the polishing envelope. All four polishing compositions were capable of producing surfaces which approached the subnanometre level, although the times required to achieve this were reagent-dependent.The acidic fluoride-based7 and cerium(1v) oxide suspension reagents enabled satisfactory finishes to be achieved very rapidly, often within 20min; the spread of results for the former were in contrast to the narrow range of values obtained with cerium(1v) oxide alone. Although rates of stock removal from the sub- strates were not determined during polishing, the RA values determined after various polishing times indicated that alkaline silica sol and fluoride-modified silica sol reagents required considerably longer polishing times, and even then, RA values were inferior to those obtained with cerium(rv) oxide or the acidic fluoride-based reagent (Fig. 1). It was concluded that, under polishing conditions, the [HF,] -anion was effective only at low solution pH.There was evidence from interfero- metric studies that a compressive surface stress, a Twyman effect,' developed on substrates polished with the acidic fluor- ide-based reagent. Talystep profiles for silicon substrates polished under similar conditions were recorded and the results (ca. 8-10 data points to determine the polishing envelope for each reagent) plotted (Fig. 2). The acidic fluoride-based reagent was unsuccessful for the removal of lapping damage and so no Talystep measure- ments were made. Cerium(1v) oxide, alkaline silica sol and fluoride-modified silica sol were successful for the chemomech- Fig. 1 Variation of surface roughness, R,, with polishing time for silica wafers: (a) acidic fluoride-based reagent;7 (b) cerium(1v) oxide; (c) alkaline silica sol; (d) fluoride-modified silica sol.Polishing conditions as described in Experimental section. Fig. 2 Variation of surface roughness, R,, with polishing time for silicon wafers: (a) alkaline silica sol; (b) fluoride-modified silica sol; (c) cerium(rv) oxide; (d) alkaline silica sol-cerium(1v) oxide. Polishing conditions as described in Experimental section. anical polishing of silicon; however, only a binary combination of alkaline silica sol and cerium(rv) oxide was capable of producing a subnanometre finish within 1 h. The spread of results obtained for the addition of [HF,]-to alkaline silica sol indicated that this reagent was inferior to alkaline silica sol alone.Etching experiments Dip-etch experiments with silica substrates revealed that no single component in the acidic fluoride-based reagent had a dominant effect, as determined by mass loss. Binary combi- nations were more informative. A KHF,-HCl combination produced significant etching and mass loss in silica and silicon substrates. Mass loss from silica wafers occurred preferentially at low pH and was of the order of 0.001-0.002 g hpl. The optimum concentration of [HF,] was 2.0-2.5 mol dm-3 in aqueous solutions of HCI (2.6 mol dm-3). Pronounced mass changes occurred over 24 h. During the course of both pol- ishing and dip-etch experiments employing the [HF,] anion at low pH (1.0-5.0) an opaque surface layer, which fractured and so was removed easily, often developed on silica.Optical J. Muter. Chem., 1996, 6(2), 227-232 229 microscopy under plane-polarised light revealed the composite nature of the layer, with areas of crystalline and spongy aggregates. Examination under cross polars provided evidence for the isotropic crystalline (or amorphous) nature of the layer, which appeared mainly dark, while peripheral areas were visibly light and consistent with a material exhibiting stress birefringence. Examination by SEM of material isolated (over the pH range 1-3) indicated it to be composed of two crystalline components, one of which was water soluble, in an amorphous matrix with twinning between crystals. The mor- phology of the layer was pH-dependent; material formed at pH 1.0, in contrast to pH 3.0, had a rippled surface [Fig.3(u) and (b)].Euhedral crystals with perfect cleavage were most evident, some of which exhibited hexagonal etch pits; the remaining crystalline material was composed of thin base1 platelets of poor cleavage, often displaying a conchoidal frac- ture. There was a wide distribution in particle sizes of 1-600 pm. It was not possible to conduct elemental analysis of the materials in situ owing to experimental limitations. Related behaviour was observed for silicon wafers. However, the mass changes were of an order of magnitude smaller than those found for silica under similar experimental conditions, and in contrast to silica, wafers of silicon exhibited small mass increases upon exposure to HCl-KHF2 solutions for 24 h.The Fig. 3 SEM images of layer materials formed at (a) pH =1.0, (b) pH = 3.0. Both samples were water-rinsed before examination. pattern of results suggested that in the absence of mechanical effects, the early stages of dip-etch reactions for silicon exposed to aqueous HCl-KHF, resembled those observed for silica. Radiotracer experiments Fresh silica, silica etched prior to radiochemical work and opaque layer material isolated from earlier experiments and stored in a vacuum desiccator, were exposed to aqueous ["F]-labelled [HF2] -solutions over the pH range 1.0-9.0 (samples of similar mass and geometry). Dip-etching in solution pro- duced measurable ['8F] activity from the materials.The pattern of results indicated that fluorine was incorporated preferentially at low pH (Fig. 4). Under comparable conditions, etched silica displayed significantly less incorporation of fluor- ine than the virgin material (Table 1). The fluorine species incorporated was water soluble as there was no detectable ["F] activity on materials after they had been rinsed with distilled water. Consistent with this observation, [I8F] activity from layer material, grown deliberately on the surface of silica wafers and exposed to ['8F]-[HF2] -,could not be determined with any degree of precision, as the bulk of the material was water-soluble. The pattern of results obtained for fresh and prior-etched silicon samples was similar to that found for silica, with [I8F] count rates greatest at low pH (Fig.5 and Table 2). Under comparable conditions, the levels of incorporation (defined as a percentage of the activity of the original solution) were of an order of magnitude less than those observed for silica, cu. 1%, (cf:11% at pH 1.0 for virgin silica wafers). In order to determine quantitatively the extent to which fluorine was incorporated at a given pH, silica and silicon (fresh and prior-etched in both cases), silica exposed overnight to an aqueous suspension of cerium(rv) oxide and opaque layer material isolated from silica dip-etch experiments were compared. The levels of ['*F] uptake on fresh and ceria-treated silica were similar (Fig. 6), and were approximately twice those found for etched silica.In agreement with the earlier results, fresh and etched silicon displayed levels of [I8F] uptake of an order of magnitude less than silica under comparable conditions and the fluorine incorporated was inert to radiochemical exchange. Fluorine 0' A ' ,I 1 0 2 4 6 8 10 pH of solution Fig. 4 ['8F] count rate from silica wafers us. pH of [I8F]-labelled [HFJ-solution. Fresh silica (W), etched silica (+) and layer material (A)all investigated under conditions described in the text. Table 1 Radiochemical uptake" of [I8F] on silica from aqueous solutions containing ['8F]-labelled [HFJ - fresh silica etched silica layer material fresh silica fresh silica fresh silica fresh silica PHYOuptake 1.0 11.1 1.o 2.6 1.0 0.6 3.0 4.7 5.0 3.0 7.0 1.9 9.0 2.0 " Detectable 18Factivity on dried substrates expressed as a percentage of solution activity as described in Experimental.230 J. Muter. Chem., 1996, 6(2), 227-232 120 9 0 2 4 8 8 10 12 pH of solution Fig. 5 [18F] count rate from silicon wafers us. pH of ['8F]-labelled [HF,] -solution. Fresh silicon (a)and etched silicon (B)investigated under conditions described in text. 500 I A51 v-c$400 c \z300 91 2042c 200 a 0:100 s n. etched fresh ceria etched fresh silica silica silica silicon silicon Fig.6 ['*F] count rates from silica and silicon wafers after 2 h exposure to an [18F]-radiolabelled [HF,] -solution of pH = 1.0 atom coverage was of the order (2.5 x 10'8)-1019 F atom cm-' for silica wafers and 2.5 x lOI7 F atom cm-2 for silicon wafers.Identification of the reaction intermediate formed during polishing and etching In order to identify its chemical components, the layer material was examined by several techniques. Powder XRD produced diffraction lines and intensities which were consistent with those of cubic and hexagonal modifications of K2SiF6;9 under all conditions the cubic form was predominant. The pattern of results indicated that the formation of hexagonal K,SiF6 was dependent on the pH and was greatest at pH 1.0. Transmission IR spectra of the material prepared as a KBr disc (vmax 740 and 482cm-') and Nujol mull (v,,, 742, 660 and 482cm-l) were characteristic of [SiF612- anion in K2SiF6.10 In contrast, the surface of the material exhibited features consistent with those of an hydrated silica species.'l Diffuse reflectance IR and photoacoustic spectra, although differing in detail, shared common features in the 1000-1200 cm-' region assigned to siloxane (Si-0-Si) stretching modes.There was little evidence for the presence of the [SiF612- anion. FTIR micro- scopy (in transmission and reflectance modes) confirmed these observations. Representative spectra were reported in the preliminary comm~nication.~ Several samples of layer material isolated in different experi- ments displayed a single peak in their 29Si MAS NMR spectra (6 -186 to -188 relative to Me,Si) characteristic of octa- hedrally coordinated silicon.12 Spectra of silica glass (Spectrosil B) and silicic acid, recorded under similar conditions, exhibited signals attributable to tetrahedrally coordinated sili- con (6 -105 to -110).Attempts to reduce line broadening in spectra of the layer material using high-power proton decoup- ling (HPPD 29Si MAS NMR) were unsuccessful and cross- polarisation (CP 29Si:1H MAS NMR) did not produce any signal enhancement; in contrast, the effect was seen clearly for silicic acid. In both CP and HPPD experiments, a search for signals arising from amorphous silica was negative for samples of the layer material. In each case elemental analysis of the solids confirmed that the material contained potassium and fluorine; however, the K: F ratios were not consistent with pure KzSiF6, but rather suggested material in which some hydrolysis of [SiF6I2-had occurred.Discussion and Conclusions The utility of a reagent for chemomechanical polishing of an optical or electronic substrate material must be assessed ulti- mately by its ability to reduce surface roughness to subnanome- tre levels. In the present study R, values have been determined using a stylus instrument; as the stylus dimensions exceed greatly those of the departures from surface planarity and as the concept of a 'subnanometre surface' is not reconciled readily with the range of interatomic distances, absolute values of RA can be challenged. They should be used therefore only for comparative purposes.Satisfactory surface finishes are obtained when silica is polished by an aqueous cerium@) oxide suspension or by CeO, suspended in a [HF2] -,sucrose solution at low pH. Silicon is polished most effectively using cerium@) oxide added to an alkaline silica sol, pH = 10.0-10.5. Comparisons between the results obtained for polishing and dip-etching experiments have enabled two important consider- ations to be identified: the formation of relatively insoluble surface reaction intermediates, and the effects of solution pH which can determine the nature of the substrate surface. In the present investigation, conclusive evidence for a sparingly sol- uble intermediate is provided by the behaviour of the acidic- based fluoride reagent towards silica.Radiotracer experiments using [18F]-labelled [HF,] -indicate that interactions between [HFJ and silica or silicon substrates are most extensive at low pH, incorporation of ["F] radioactivity being greatest at pH = 1.0 and undetectable above pH = 10.0. For a given pH, incorporation of ["F] is greater for virgin wafers compared with those that have been etched previously. In all cases [18F] is removed completely by rinsing with water. Isolation of the surface layer formed by chemomechanical polishing or dip- etching of silica with the acidic-based fluoride reagent indicates that it is a mixture of species and that the morphology is pH- dependent. Elemental analysis of the solids indicates that they contain potassium and fluorine. The pattern of results obtained from examination by XRD and 29Si MAS NMR, coupled with FTIR spectroscopy indicates that while K2SiF6 is the major component of the bulk material, Si-OH and Si-0-Si groups predominate on the surface.There was no evidence for the formation of cerium-fluorine compounds. Results from SEM and optical microscopy are consistent with a material composed of a mixture of isotropic cubic crystals and some hexagonal material, and provide evidence for the presence of crystalline quartz. On the basis of all the available information, Table 2 Radiochemical uptake" of ["F] on silicon from aqueous solutions containing ['*F]-labelled [HF,] -fresh silicon etched silicon fresh silicon etched silicon fresh silicon etched silicon fresh silicon etched silicon PH 1.o 1.o 3.0 3.0 7.0 7.0 10.0 10.0 % uptake 1.2 0.8 0.4 0.1 0.07 0.05 0.05 0.05 ~ ~ ~~ " Detectable '*F activity on dried substrates expressed as a percentage of solution activity as described in Experimental. J.Muter. Chem., 1996,6( 2), 227-232 231 it is concluded that the reaction intermediate formed on silica surfaces, in both polishing and etching processes, is K,SiF, coated with a thin, silica-like layer. The latter may result from hydrolysis of Si-F species at the surface, as authentic K2SiF6 exposed to atmospheric moisture over extended periods, in contrast to samples stored in a vacuum desiccator, often produced reflectance IR spectra similar to layer material. Alternatively, the data may reflect the incorporation of silica particles on K,SiF,.The .pattern of results in this work provides evidence that kinetic, for example, removal of reaction products from the solid-solution interface, rather than thermodynamic factors, for example, relative stabilities and concentrations of reactive species, are more likely to be important in chemomechanical polishing processes for silica with the acidic fluoride-based reagents, in agreement with previous studies conducted for ceria, silica system^.^ The solubilities of the primary reaction products formed rapidly at the interface are likely to be limited, so redeposition effects may be important; incomplete removal of these reaction products by mechanical means may result in the formation of a passivating layer, whose function is to protect those 'troughs' on the surface at which reaction has already occurred so that further reaction occurs preferentially at the 'peaks'.The opaque layer material formed when silica is polished or etched with acidic [HF,]- appears to function in this way. The situation when silica is polished by CeO, alone, nominally at neutral pH, is less clear. Other studies13' involving CeO, have focussed on its abrasive effects and it is clear that the size and shape of individual crystallites are im~0rtant.l~'Complex formation between ceria and a hydrated glass surface has been invoked previously4' to account for the removal and redistribution of silica and we speculate further that complexation could involve Lewis acid-base interactions between CeIV sites on ceria particles and deprotonated surface silanol groups.Subsequent reaction will involve removal of silicon as %(OH),, or a CeO, complex; such species are expected to undergo condensation and polymerisation reactions readily.14 Although removal of monomeric silicon-containing species from aqueous solution has been reported to be accelerated by the presence of F-anion at a concentration as low as 1 ppm,15 we suggest that a different pathway becomes possible using [HF,] -at low pH. The acidic nature of silicon, pK, =ca. 6.5,'' and its isoelectric point (2.8)16 will result in a significant concentration of surface [Si-OH,]+ groups only when the pH of the contacting solution is very low.The marked pH dependence (Fig.4) of the uptake of ["F] by silicon from [18F]-[HF2]-is consistent with the initial step in the etching reaction [eqn. (l)] [Si-OH2lsUf+ +[HF,]-+[Si-FlSurf+HF +H,O ( 1) Subsequent reactions lead to the formation of [SiF,I2- par-ticles, the surfaces of which become covered with a thin, silica- like layer, due either to surface hydrolysis or to the incorpor- ation of silica directly from the surface by a mechanism similar to that outlined4' when CeO, alone is used. The two surface processes are summarised in Scheme 1. Silica can be polished using an alkaline silica sol, albeit very slowly [Fig. 1 (c)], but addition of [HF,] -has no accelerating effect, which is consistent with the rationalization presented above.The small, negative effect observed [Fig. l(d)] may be a consequence of the change in pH (from ca. 10 to 8) occurring on addition of [HF,]- with concomitant sol gelation and increase in average size of the parti~1es.I~ Incorporation of ["F] from [18F]-[HF,]- by silicon has a similar pH dependence to that by silica, but its extent is far smaller. Presumably the initial reaction involves the oxide layer on silicon and a process identical to eqn. (1); however, the reagent is ineffective for polishing, possibly because the surface layer is too thin and under these conditions is not 232 J. Muter. Chem., 1996, 6(2), 227-232 Scheme 1 Representation of processes that occur when silica is etched or polished by aqueous [HF,] -or an aqueous suspension of CeO, regenerated.The most effective polishing reagent for silicon under these conditions used is CeO, added to alkaline silica sol, a pH region where termination of silicon surface atoms by hydrogen has been shown recently to play a key role in the mechanism of polishing.18 We thank Dr. J. A. Chudek and Prof. G. Hunter, University of Dundee, Drs. T. Baird (Department of Chemistry) and A. Hall and D. Turner (Department of Geology) all of the University of Glasgow, Drs. R. Richardson and D. Hepburn, Glasgow Caledonian University, the staff of the Scottish Universities Research Reactor Centre, East Kilbride, and Nicolet Instruments Ltd., Warwick, for the provision of facili-ties. We thank DTI, EPSRC and Logitech Ltd. for support of this work, partially via the LINK Nanotechnology programme.References 1 J. M. Bennett, Meas. Sci. Technol., 1992,3, 1119. 2 D. C. Cornish, The Mechanism of Glass Polishing, Taylor and Francis Ltd, London, 1961; L. Holland, The Properties of Glass Surfaces, Chapman and Hall, London, 1964; D. Cornish and L. Watt, Br. Sci. Instr. Res. Assoc. Rep., 1963, R295. 3 D. S. Boyle, J. A. Chudek, G. Hunter, D. James, M. 1. Littlewood, L. McGhee, M. I. Robertson and J. M. Winfield, J. Muter. Chem., 1993,3,903. 4 (a)L. M. Cook, J. Non-Cryst. Solids, 1990,120, 152; (b)N. J. Brown, Precis. Eng., 1987,9, 129. 5 (a)H. Nielsen and D. Hackleman, J. Electrochem. SOC., 1983, 130, 708; (b) A. C. M. Spierings, J. Muter. Sci., 1993, 28, 6261 and refs.therein. 6 (a) M. Beveridge, L. McGhee, S. G. McMeekin, M. I. Robertson, A. Ross and J. M. Winfield, J. Muter. Chem., 1994, 4, 119; (b) L. McGhee, S. G. McMeekin, I. Nicol, M. I. Robertson and J. M. Winfield, J. Muter. Chem., 1994,4,29. 7 N. A. Murphy, J. G. Banner, E. Fletcher and A. Brown, UK Pat. Appl., 1980,2055792A; US Pat., 1980,4 343116. 8 A. J. Dalladay, Trans. Opt. SOC. London, 1921, 23, 170; E. G. Nikolova, J. Muter. Sci., 1985,20, 1. 9 Joint Committee for Diffraction Standards, International Centre for Diffraction Data: card no. 7-217, cubic K,SiF,; 16-578, hexagonal K,SiF,. 10 R. B. Badachhape, G. Hunter, L. D. McCory and J. L. Margrave, Inorg. Chem., 1966,5,929. 11 N. L. Rockley, M. K. Woodard and M. G. Rockley, Appl. Spectrosc., 1984,38,329; R. L. White and A. Nair, Appl. Spectrosc., 1990,44,69. 12 J. F. Stebbins and M. Kanzaki, Science, 1991,251,294. 13 (a) e.g. N. B. Kirk and J. V. Wood, Br. Ceram. Trans. J., 1994, 93, 25; (b) N. B. Kirk and J. V. Wood, J. Muter. Sci., 1995,30,2171. 14 e.g. N. N. Greenwood and A. Earnshaw, Chemistry of the Elements, Pergamon, Oxford, 1984, pp. 393-399. 15 R. K. Iler, The Chemistry of Silica, Wiley, New York, 1979. 16 C. D. Fung, P. W. Cheung and W. H. KO,IEEE Trans., Electron Devices, 1986,33,8. 17 R. K. Iler, Surface and Colloid Science, ed. E. Matijevic, Wiley, New York, 1973, vol. 4, p. 11; E. M. Rabinovich, D. M. Krol, N. A. Kopylov and P. K. Gallagher, J. Am. Ceram. SOC., 1989, 72, 1229. 18 G. J. Pietsch, G. S. Higashi and Y. J. Chabal, Appl. Phys. Lett., 1994,64,3115. Paper 5/04661K; Received 17th July, 1995
ISSN:0959-9428
DOI:10.1039/JM9960600227
出版商:RSC
年代:1996
数据来源: RSC
|
18. |
Alumina purification by carbothermal reduction |
|
Journal of Materials Chemistry,
Volume 6,
Issue 2,
1996,
Page 233-238
Dulcina P. F. de Souza,
Preview
|
PDF (918KB)
|
|
摘要:
Alumina purification by carbothermal reduction Dulcina P. F. de Souza* and Milton F. de Souzab "Federal University of Sao Carlos, Department of Materials Engineering-DEMa, 13.565-905, SP, Brazil bSao Paulo University, Department of Physics and Materials Science, IFSC, 13.560-250, Sao Carlos, SP, Brazil Sodium, silicon, iron, calcium and gallium as impurities in alumina were removed by carbothermal reduction from alumina particles between 1100 and 1400 "C. Reduction of calcium and silicon oxides starts as low as 1200 and 1100 "C, respectively. Thermodynamics and rate of the alumina cleaning process are discussed. Carbothermal reduction of oxides is a chemical reaction cur- rently been used to prepare metals, carbides and nitrides. It can also be used to prepare zirconium oxide by carbothermal reduction of zircon.' The synthesis of aluminium nitride, AlN, by carbothermal reduction of alumina in a nitrogen flux has been of interest in recent The chemical reactions that can occur between an oxide and carbon are a direct solid-state reaction: MO,+nC+nCO+M (1) or an indirect reaction via an intermediate gas phase:4 MO,+nCO+nCO,+M (2) co,+c+2co (3) Although the literature on the carbothermal reduction of oxides is extensive, the reduction of oxides as impurities in a host matrix has not yet been investigated.In this work we present and discuss the results of the purification of alumina by the carbothermal reduction process (CRP). The starting material was a commercial aluminium hydroxide. After conver- sion to a-alumina the impurities are located mainly in the bulk of the alumina particles.When compared to the CRP of pure oxides the impurity-oxide reduction has different features. Impurities must diffuse to the alumina surface in order to react with carbon. The possibility of the formation of a new structure for the impurity oxide on the alumina surface before its carbothermal reduction must also be considered. Experimental Gibbsite powder from the Bayer process (Alcoa, Brazil) and carbon black having 0.2 mass% of ash content and an average particle size of 0.07 pm were used as the starting materials for the CRP. The gibbsite was vibratory milled to reach an average particle size of 0.3 pm as measured using a Micromeritics Sedigraph (see Fig.1). Carbon black and poly- (vinyl alcohol) (both 2 mass% relative to AlOH) were then added. The amount of impurities introduced by the ash content of the carbon black was 40ppm, a value well below the total of impurities in the samples after the carbothermal reduction. The mixture was ball-milled over 10 h. The foamy slurry was oven-dried at 150 "C and formed granules of ca. 30 mm3. The density of these porous granules was 1.8g~m-~.The CRP experiments were performed in a 99.9% argon flow of 3.0 dm3 min-'. The argon flow produced a carbon monoxide partial pressure Pco m 5 x atm as a consequence of the presence of some oxygen in the argon gas. The CRP was performed inside an alumina tube with one sealed end and with the necessary fittings in the other end for the inlet and outlet of gases and for the protected platinum thermocouple.Each batch used 20 g of granules inside an alumina crucible. The argon was fluxed in the bottom of the crucible by an alumina tube. The vertical furnace provided a heating rate of 800°C h-'. The experiments were performed at 1100, 1200, 1300 and 1400 "C.An argon flow of 3.0 dm3 min-' (measured at room temperature) was chosen for the experiments at 1400 "C. Experiments with argon flows higher and lower than this value were performed and the samples chemically analysed. It was found that rates lower than 2.8 dm3 min-' decreased the rate of removal of iron and calcium. Sodium and silicon removal is decreased at lower argon flow rates. In order to determine the kinetics of the CRP, experiments of 80, 180 and 360 min duration were performed for each temperature.Other experiments were conducted on milled gibbsite with and without carbon in air, without argon flow, in the same furnace and conditions. CRP experiments were also performed on pure alumina with calcium oxide particles on its surface. The calcium oxide particles were obtained from calcium carbonate pre- viously milled together with the alumina powder. The powders prepared by the CRP are described in Table 1. They were characterized by their chemical compositions, particle-size 100 (0 1 equivalent spherical diameter/pm Fig. 1 Particle size distribution before and after carbothermal reduction as described in Table 1 Table 1 Powder sample compositions and calcination conditions sample composition" G G G+C G+C G+C G+C G+C G+C G+C atmosphere T/"C air 25 air 1300 air 1300 air 1400 Ar flowb 1100 Ar flowb 1200 Ar flowb 1300 Ar flowb 1400 Ar flow' 1400 G =Gibbsite; C =carbon.3.0 dm3 min- '. 0.5 dm3 min- '. J. Muter. Chem., 1996,6(2), 233-238 233 distribution and particle shape. Chemical analysis was per- formed by first completely dissolving the alumina powders in concentrated acid. The concentration of each element was then measured by plasma spectroscopy. The excess carbon was eliminated by heating the samples in air at 800°C. Dense alumina bodies were prepared by isopressing at 270 MPa and sintering at 1600 "C over 2 h.Density measure- ments were conducted using the Archimedes method. Samples of the dense alumina bodies for SEM were polished and thermally etched at 1550 "C for 5 min. Results Particle size The results of the particle-size measurements of the powders prepared as described in Table 1 are shown in Fig. 1. As expected, the particle size increased with the temperature of the CRP, but this increase was smaller for the alumina powders obtained by calcination with carbon black and in an argon flow. Only the carbon black particles provided the physical separation between the alumina particles during the CRP; the poly(viny1 alcohol) binder leaves very little carbon after its decomp~sition.~Considering the particle size, and the mass ratio of carbon black and aluminium hydroxide, it is found that for each aluminium hydroxide particle there are two carbon black particles.This ratio is not sufficient for complete insulation of the alumina particles from each other, but is enough to decrease their physical contacts, as was observed. Alumina particles from sample 8 are almost spherical, while the particles of sample 3 have large particle aggregates due to the burning of the protective carbon particles, as shown in Fig. 2. These results show that aluminium hydroxide particles imbedded in carbon black can be converted to nearly spherical alumina particles during calcination in their conversion to a-alumina, if enough carbon is present and protected from reaction with air. Alumina purification The degree of purification achieved in the alumina powder after 6 h of CRP is shown in Table 2.The sodium concentration reached the lowest level, at the limit of our chemical analysis. Iron oxide and silicon oxide were removed in proportions of 90 and 93%, respectively. A large amount (81%) of the calcium oxide was also removed. This result was unexpected consider- ing that pure calcium oxide needs a higher temperature than alumina to be reduced by carbon. The low concentration of gallium oxide present in the samples was also reduced to 20% of its initial value of 120 ppm. The gallium concentration was measured only for 6 h of CRP at 1400°C. Another aspect of the alumina powder purification is shown in Fig.3. The microstructure of the sintered compacts made of purified alumina is shown in Fig. 3A, while in Fig. 3B the microstructure of a sample made of impure alumina following the same procedure can be seen. The purification reduces the grain growth and the number of plate-like grains. It is known that impurities, especially silicon and calcium, are responsible for this behaviour of the impure alumina corn pact^.^*^ For comparison, a sintered compact made using CR-30 Baikowski pure alumina is shown in Fig. 3D. The sintered body made from sample 9 shows the microstructure of Fig. 3C. The EDS analysis of its grain boundaries shows the presence of a calcium-rich phase. It can be concluded from Fig.3C and its EDS analysis that the impurity oxides were reduced by the CRP but the reaction products were not efficiently removed by the reduced argon flow. In order to better understand the calcium oxide behaviour during CRP, an additional experiment was performed. Calcium carbonate, CaCO,, was added to a pure alumina and ball- milled. The dried mixture was calcined at 1100 "C in order to convert CaCO, into CaO. This powder, containing 1500 ppm of CaO was prepared for CRP as usual. The CRP was performed under the same conditions as sample 8. Chemical analysis before and after CRP gave the same results, indicating that calcium oxide was not reduced. The reaction kinetics of the CRP for each studied impurity are shown in Fig. 4. These graphs were plotted assuming that at a fixed temperature the alumina cleaning rate would be described by the equation dC/dt= -aC, where C is the impurity concentration and a the cleaning rate.It is found that this relationship applies well to the cases studied but deviations occur when the impurity concentration reaches low values. Fig. 5 shows the temperature dependence of the rate constant a. For the iron oxide impurity the rate constant, aFe, shows a monotonic increase with temperature, which allowed us to determine its activation energy, E,, using Fig. 2 SEM images of powders prepared as described in Table 1; A, sample 8; B, sample 3 Table 2 Chemical analysis after 6 h of carbothermal reduction of the samples defined in Table 1 NazO Si02 b03 CaO Ga203 sample CRP temp./"C ppm" YOoutb ppm" YOoutb ppm" % outb ppm" YOoutb ppm" YOoutb 1 -5000 0.0 1300 0.0 500 0.0 1500 0.0 120 0.0 5 1100 3240 35.0 1092 16.0 315 37.0 1500 0.0 6 1200 40 99.2 830 36.0 186 63.0 1140 24.0 --7 1300 20 99.6 170 87.0 115 77.0 280 81.0 8 1400 10 99.8 86 93.0 50 90.0 280 81.0 24 80.0 " Impurity concentration remaining in the sample.Percentage of impurity removed. 234 J. Muter. Chem., 1996,6(2), 233-238 Fig. 3. Microstructure of compacts after sintering at 1600 "C: A, sample 8 (purified alumina); B, sample 1 (impure alumina); C, sample 9 (reduced alumina under low argon flow); D, Baikowski CR-30 alumina. All compacts were prepared in the same conditions. -1.5 --Calcium -2.0 ! 1 1 I 11 1 0 100 200 4ocdo -1.5 - -1.57 -2D- "7 Silicon -23 f 1 I 1 I 1 I 1 1 -so : I I .1 8 I 0 100 200 ux) 430 0 1m 200 300 c I time/min Fig. 4 Time dependence of the reduction of the impurities for each operation temperature. It is assumed that dC/dt = -C. aFe=aoFe exp(-E,/kT). For the other impurities the plot a us. Discussion T shows a sharp increase in a values at 1200°C for sodium and at 1300°C for silicon and calcium. These increases in a Thermodynamics indicate the onset of an additional reduction process making Reduction of oxide impurities can be accomplished by direct it impossible to determine values for the activation processes reaction with carbon [eqn. (l)]or carbon monoxide [eqn. (2)]. that were taking place. Carbon dioxide, a product of eqn.(2) will eventually react J. Mater. Chem., 1996, 6(2), 233-238 235 1x0 1300 TPC Fig. 5 Temperature dependence of the initial reaction rate, ai,of the carbothermal reduction of impurities in alumina with the excess of carbon, producing carbon monoxide (eqn. (3)]. Therefore, both reactions will give the same final products. Eqn. (2) is sometimes called the carbothermal reduction reaction via an intermediate gas phase.7 However, it must be considered that the standard free energy values, AGO, for each reaction are different. Therefore, the temperatures where the condition AG'=O holds for eqns. (1) and (2) are different. This condition will be satisfied for eqn. (1) at a lower temperature than for eqn.(2). Consider the CRP for the sodium oxide impurity, Na20. The chemical reaction of type 1 1s 2Na,0 +2C+2CO +4Na (4) which can be represented by the combination of the following reactions: 4Na +O2+2Na20 (5) 2c +02 +2co (6) Thermodynamic data for basic equations like (5)and (6) are found in specialized data banks or in Richardson-Ellingham charts.* The chemical reaction (2) for sodium oxide is Na,O +C04C02+2Na (7) which can also be represented by: 4Na +O2+2Na20 (5) 2co +02-2c02 (8) On the basis of the published data on oxide decomposition in reducing atmosphere^,^ we will assume that metallic sodium, calcium and iron are the final products of CRP of their respective oxides. Silicon monoxide, SiO, is the product of the silicon dioxide reduction.Fig. 6 shows the temperature depen- dence of the standard free energy AGO for the basic chemical reactions for the reduction of sodium, silicon and calcium oxides. For iron oxide, Fe20,, the condition AG'>O starts to be valid at ca. 800°C for reactions (1) and (2), below the temperature where the CRP of this work was performed. Therefore, for the iron oxide reduction, both reactions (1) and (2) were operating simultaneously in our experiments. For sodium oxide the condition AGO >0 is valid above 1000 "C for reaction (l), while for reaction (2) it is valid for temperatures 236 J. Muter. Chem., 1996, 6(2), 233-238 -fd'' ' I / ' ' ' 0 5Qo to I500 2Ooo 2930 TIT Fig. 6 Richardson-Ellingham plot for the carbothermal reactions (ref. 8).A, 2C +0,=2CO; B, 2CO +0,=2C02; C, 2Ca +0,=2Ca0; D, 2SiO +0,=2Si0,; E, 4Na+ 0,=2Na20; C', adjusted C curve for ca. 50% higher enthalpy of formation in order to cross B curve at 1300°C (CaO+CO+Ca+CO,); D', adjusted D curve for ca. 50% higher enthalpy of formation in order to cross B curve at 1300°C (SiO, 4iO +CO,). above 1200 "C. The equilibrium constants of the chemical reactions (4),(5)and (6)are: K4= [Na]4[C0]2/[Na20][C]2, K,=[Na20]2/[Na]4[02] and &= [CO]2/[C]2[02], which are related by &=&I&, or by the equivalent relation between the standard free energy of the reactions: AGO4= AGo6-AGo5. The variations with temperature of AGO, and AGO6 are shown in Fig. 6 as curves E and A, respectively.For K, >1 the thermodynamic equilibrium of eqn. (4) is displaced towards the reaction products. If the chemical kinetics are fast enough, equilibrium can be attained and sodium vapour will be produced to be transported by the argon flow. For K4<<l the equilibrium concentration of sodium will be low. In this paper we consider the condition K, 2 1 as necessary for enough sodium vapour to be produced in order to be detected by the chemical analysis. The condition K, =1, (AG', =AGO,) is shown in Fig. 6 as the point where lines A and E cross, i.e. T=98O0C. Similarly for reaction (7), reduction of Na20 by CO, the crossing of curves E and B is at T=1200 "C. Following the data of Fig. 6, reduction of bulk silica, calcia and alumina in the temperature range studied would produce a very low concentration of the reaction products.In fact, alumina was not reduced in our experiments; a much higher temperature would be needed for this to occur." On the other hand, the reduction of silica, as shown by several starts at lower temperatures than predicted by the Richardson-Ellingham charts. In our experiments it began at 1100 "C. Pure calcium oxide would be reduced at a higher temperature than alumina, but we found that its reduction as an impurity in alumina started at 1200°C and proceeded at a high rate at 1300°C. One way to start the reduction of the oxides at lower temperatures is through the formation of intermediate sub- stances, as occurs with alumina." Another possibility is the formation of structures with the same chemical composition but a higher enthalpy of formation, as in thin films. In fact, this type of structure has been found for calcium oxide on the alumina surface by Baik and Whitei4 and McCune and Ku.'~ Baik and White found that calcium segregates preferentially to the (101 0) face of aluminium oxide single crystals, and that at 1300 "C an ordered CaO monolayer grows quickly on that face.We speculate that CRP of the calcium oxide impurity would be allowed by the formation of such monolayers. The enthalpy of formation of such monolayers will be higher than for a three-dimensional structure. A rough estimate of the necessary increase in the enthalpy of formation of the pure oxide can be performed by drawing a line C' parallel to line C and crossing line B at 1300 "C where the chemical reduction of CaO by CO is supposed to be occurring, as shown in Fig.6. The simple procedure outlined above gives an enthalpy of formation for the CaO monolayers 50% higher than that of the pure CaO. The segregation of calcium to the Al,O, crystallographic surfaces is non-isotropic, decreasing the rate of reactions (1) and (2). For silicon dioxide we assume that silica thin layers are formed on the alumina surface allowing their reduction by the CRP. Line D' was set to cross line B at 1300°C because we found that at this temperature the rate of CRP of silica showed an increase. The values found for the enthalpy of formation of these silica layers was 54% higher than for the pure silica as predicted by the Richardson- Ellinghan chart.Reaction rates The observed a-rates shown in Fig. 4 and 5 were measured by chemical analysis of the impurities remaining in the alumina powder after the CRP. These cleaning rates are dependent on the impurity diffusion coefficient (D), the vapour pressure of the reaction products (P"), the argon flow rate (%), the porosity of the granules and the kinetics of the chemical reactions (7). Granule porosity and argon flow are constants in our experi- ments. We found by experiment that the maximum uNa values (aNaat 1400°C) do not change if 5 increases. The effect of y, D and P, on the cleaning rate a will be discussed for each impurity. Firstly, let us assume that the alumina particles can be describe as dense spheres of radius 0.5 pm.The impurities will diffuse continuously from the bulk of these spheres to their surfaces if the chemical reaction rate, y, is high enough to maintain the necessary concentration gradient. The average impurity concentration, c,inside the grain will fall to half its initial value in a time, tllz,given by t1/2=(5 x lO-"/D) s if D is expressed in cm2 sP1.16 Sodium and calcium have large diffusion coefficients in alumina.17 At 1100 "C, DNa= lo-'' cm2 s-', which gives tl,2= 0.5 s. Increasing the particle size to 20 pm, t?,2becomes 13 min. Therefore we can conclude that diffusion inside the alumina grains is not a limiting factor for aNa.The same conclusion is valid for aCa.Metallic sodium and calcium both have high vapour pressures,l8 P, for sodium being several atmospheres at the temperature of our experiments.If enough sodium or calcium were present they would produce a values much higher than those observed. To estimate these values, we assumed that sodium and calcium equilibrium vapour pressures were continuously mixed with the argon gas. We can conclude that aNaand a,, are determined by their chemical reaction rates, yNa and y,,, respectively. The small aNaobserved at 1100°C (Fig. 5) is attributed to reaction (l), which is thermo-dynamically more favourable from 1000 "C. The observed higher rates above 1200°C are attributed to reaction (2). This model for explaining the chemical reaction rates suggests that carbon monoxide finds and reacts with the impurity oxide faster than the direct reaction between solid carbon and the impurity oxide.In fact, the contact points between carbon and alumina particles are few. While the sodium cleaning rates, aNa,above 1200 "C are the highest observed in our experiments, a,, below 1200 "C is the lowest. The chemical reaction between calcium oxide and carbon or carbon monoxide is apparently only allowed after CaO monolayers become present on special alumina crystalline surface^.'^,^^ Baik and White14 found that CaO monolayers are present at 1200"C, but they grow more readily at 1300 "C. The observed calcium segregation only to special surfaces observed by these authors contributes to a further decrease in the direct contact between carbon and the impurity oxide. This will give a very low rate for reaction (1).On the other hand, the fast growth of the CaO monolayers will favour the rate of eqn. (2) from 1300 "C, as we observed. Iron oxide reduction has a different behaviour than that previously discussed for sodium and calcium oxide. Reactions (1) and (2) are both thermodynamically favourable above 800 "C, and therefore no jump in a is expected, as was observed (Fig. 5). Also, aFe<< aNa.The Fe3+ diffusion coefficient and Fe vapour pressure are several orders of magnitude lower than those of sodium: &3+ DNa+and P,(Na)>>P,(Fe). The diffusion coefficient for Fe3+ in alumina was studied by Lloyd and Bowen,Ig and was found to be lowered by reducing pressure and by the presence of other impurities such as silicon and calcium.The diffusion of iron in alumina, in the tempera- ture range of their work, was found to be DFe=9.2 x cm2 s-' exp[-112 kJ mo1-l (RT)-l].17 Assuming that the observed decrease in D also applies for the temperature range studied in of our work, a value of DFe= cm2 s-l is found, giving tl,2 = 90 min. If sample irregularities are considered, this time will increase, causing Fe3+ diffusion to be the limiting factor for a. In addition, we calculated the activation energy for aFe and found E= 117 kJ mol-', which agrees with the published Fe3+ diffusion energy.17 The P,(Fe) ranges from 5 x lop7atm at 1100 "C to 4 x atm at 1400 "C.18 Considering the vapour pressure, the calculated value for the rate of removal of iron vapour, dC/dt, at 1100 "C is 7.0 x loP6g min-l, while the observed rate in this work was 5.6 x lop6g min-' .For higher temperatures, calculated values turn out to be much higher than those observed experimentally. The iron removed by the argon flow was partially deposited in the cold parts of the furnace alumina tube. Therefore we concluded that the Fe3+ diffusion in the alumina particles is the limiting factor on aFe. The silicon diffusion coefficient in alumina was not found in the literature. The silicon vapour pressure does not influence asi because silicon monoxide, the reaction product, is a gas in the temperature range of our experiments.The values for asi are low compared with aNaand a,, indicating that diffusion may be responsible in the limiting of asi. Reaction (1) is occurring already at 1100 "C,and produces low asi values. At 1300 "C, reaction (2) causes a limited increase in asi. This may be due to a low diffusion coefficient for silicon, or another cause related to the arrangement of the silica surface layers. The low temperature for the onset of the CRP of silica indicates that a new structure for the Si02 arrangements on the alumina surface may be present. We estimate in Fig. 6 the enthalpy of formation of this silica layer. Microstructures The effect of the carbothermal purification process on the purity of the alumina powders can be clearly seen in the microstructures of their compacts. It is well known from the literat~re~,~.~~ that calcium and silicate impurities promote abnormal grain growth, resulting in elongated grains as opposed to equiaxial normal grains.Fig. 3B shows a character- istic microstructure, with large elongated grains, produced by the alumina of sample 1 which was not purified by the carbothermal process. The smaller equiaxial grains of the microstructure of Fig. 3D are from a pure commercial alumina. Fig. 3A shows the effect of the carbothermal purification on sample 8: decrease of grain size and abnormal growth can be seen. However, owing to the remaining calcium impurity (Table 2), the abnormal growth was not completely suppressed. The microstructure of Fig.3C was generated from sample 9, which was purified under a reduced argon flow. The grain sizes show a strong improvement as compared with sample 1 J. Mater. Chem., 1996,6(2), 233-238 237 (Fig. 3B), but its grain boundary shows the presence of another phase. This phase was analysed using EDS, and was found to be rich in calcium and iron, as was expected from their lower vapour pressures as compared with sodium and silicon monox- ide. This result shows that the impurities have diffused to the surface and have been reduced but not removed. During the sintering of the compacts they have been segregated in the grain boundaries. It was not possible at this stage of our work to determine the amount of each impurity present in the grain boundary.Conclusions Carbothermal reduction of impurities in alumina in argon flow has been shown to be an efficient process for the removal of alumina of sodium, calcium, iron, silicon and gallium impurit- ies. It is proposed that two chemical reactions are involved in the process, both having the same final products: a direct reaction between carbon and impurity oxides and an indirect reaction via intermediate CO. The direct reaction starts at a lower temperature but has a lower reaction rate. In order to explain why CaO and SiOz impurities were already reduced at 1200 "C and 1100 "C,respectively, the assumption is made that calcium oxide and silicon dioxide have a new structural arrangement on the alumina surface. These structures must have a higher enthalpy of formation than their respective pure substances by ca.50%. The CaO structure is identified with the CaO monolayer found to develop on special alumina surfaces already mentioned in the literature. In the case of silicon, no similar evidence was found in the literature, but carbothermal reduction of SiOz has been mentioned to occur over a large temperature range. The cleaning rate for the sodium impurity is limited by the rate of the chemical reduction reactions. For iron, the low diffusion coefficient of the Fe3+ ion is the limiting factor. Although no diffusion coefficient was found in the literature for silicon, in alumina asi is apparently also limited by diffusion. For calcium, once the CaO monolayers are formed the reduction rate is fast.Nearly 20% of the calcium impurity remained unreacted. The cause of this behaviour was not identified. The presence of carbon between the alumina particles was found to noticeably decrease their aggregation. During calci- nation, carbon-protected alumina particles were found, after calcination at 1400 "C, to have a spheroidal shape. Financial support from the FAPESP and CNPq are gratefully acknowledged. References 1 S. De Souza and B. S. Terry, J. Muter. Sci., 1994,29, 3329. 2 P. Le Fort and M. J. Billy, J.Am. Ceram. SOC., 1993,76,2295. 3 H. K. Chen, C. Lin and C. Lee, J. Am. Ceram. SOC., 1994,77,1753. 4 R. J. Higgins, E. W. Rhine, J. M. Cima, H. K. Bowen and E. W. Farneth, J. Am. Ceram.SOC., 1994,77,2243. 5 H. Song and A. L. Coble, J.Am. Ceram. SOC., 1990,73,2077. 6 S. I. Bae and S. Baik, J. Am. Ceram. SOC., 1993,76, 1065. 7 J. Szekely, J. W. Evans and H. Y. Sohn, in Gas-Solid reactions, Academic Press, London, 1976, p. 177. 8 R. T. De Hoff, in Thermodynamics in Materials Science, McGraw Hill, New York, 1993. 9 R. H. Lamoreaux, D. L. Hildebrand and L. Brewer, J. Phys. Chem. Ref. Data, 1987,16,419. 10 E. Grunet and J. Mercier, US Pat. 2,974.032, 1961. 11 N. Kilnger, E. L. Strauss and K. L. Komarek, J. Am. Ceram. SOC., 1966,49, 369. 12 P. D. Miller, J. G. Lee and I. B. Cutler, J. Am. Ceram. SOC.,1979, 62, 147. 13 B. C. Bechtold and I. B. Cutler, J.Am. Ceram. SOC., 1989,63,271. 14 S. Baik and C. L. White, J. Am. Ceram. SOC.,1987,70,682. 15 R. C. McCune and R. C. Ku, in Advances in Ceramics, vol. 10, ed. W. D. Kingery, Am. Ceram. SOC., Columbus, OH, 1984, p. 217. 16 W. Jost, in Digusion in Solids, Liquids, Gases, Academic Press, New York, 1960, p. 45. 17 R. Freer, J. Muter. Sci., 1980, 15, 803. 18 Handbook of Chemistry and Physics, 72nd edn., ed. D. R. Lide, CRC Press, Boca Raton, FL, 1991/92, pp. 5-70. 19 I. K. Lloyd and H. K. Bowen, J. Am. Ceram. SOC., 1981,64,774. 20 A. Atkinson and R. I. Taylor, J. Muter. Sci., 1978, 13,427. Paper 5/03465E; Received 31st May, 1995 238 J. Mater. Chem., 1996, 6(2), 233-238
ISSN:0959-9428
DOI:10.1039/JM9960600233
出版商:RSC
年代:1996
数据来源: RSC
|
19. |
Intercalation ability of hydrolysed aluminium species inn-alkylmonoamine α-zirconium phosphate compounds |
|
Journal of Materials Chemistry,
Volume 6,
Issue 2,
1996,
Page 239-246
Kris Peeters,
Preview
|
PDF (1073KB)
|
|
摘要:
Intercalation ability of hydrolysed aluminium species in n-alkylmonoamine a-zirconium phosphate compounds Kris Peeters,"" Piet Grobetb and Etienne F. Vansant' "University of Antwerp (UIA), Department of Chemistry, Laboratory of Inorganic Chemistry, Universiteitsplein 1, B 2610 Wilrijk, Belgium bCatholic University of Leuven, Centre of Surface Science and Catalysis, Kardinaal Mercierlaan 92, B 3001 Heverlee, Belgium The intercalation of hydrolysed A1 species in the n-alkylmonoamine a-zirconium phosphate (a-ZrP) structure was performed. A similar synthesis procedure and the same A1 solutions as for the preparation of A1-PILCs (pillared interlayered clays) were used. However, no intercalation of the A1 Keggin ion [A104A11,(OH)24( H20)12]7+in the a-ZrP structure occurred.It was demonstrated that the nature of the intercalated ions depends strongly on the competitive ions present in the A1 solution and on the reaction or exchange temperature. We could prove that the lack of the Keggin ion intercalation is due to its inaccessibility in the interlayer space and not to its decomposition upon intercalation. Alternative synthesis methods using different amine-ZrP intercalates did not promote the Keggin ion intercalation. The versatility of the pillaring of layered substrates with to interpillar void spaces in the interlayer region, can be really inorganic polyoxycations has invoked a lot of interest from prepared. academics and industry.'-3 This research is mainly focused on It seemed, therefore, to be of interest to carry out a systematic smectite clays because of their ability to swell in aqueous investigation on the Al-pillaring of a-ZrP, using similar syn- thesis methods and A1 solutions as for clays.This study allows solutions, their moderate exchange capacity and low ~ost.~,~ Among the different pillaring solutions, the A1 solution gained a lot of attention and besides the synthesis of these A1-PILCs6-8 (pillared interlayered clays) a lot of work has been performed on the identification of the nature of the pillaring species, both in solution and intercalated in the clay.'-'' The A1 polyoxy-cation that is believed to be responsible for the pillaring is called the Keggin ion, [A104A112(OH)24(H20)1217+. More recently the pillaring of layered metal@) phosphates has also been taken into c~nsideration.'~-'~ Because of their higher layer charge density (8-10 times higher), the metal(1v) phosphates do not swell as readily as clays.Therefore, in order to facilitate the diffusion of the large Keggin ions, the original interlayer distance of these materials was preliminarily enlarged by the intercalation of long-chain alkylamines" or by using colloidal s~spensions,'~ which were obtained by intercalation of propylamine.20 From a structural point of view, there is another consequence of the high charge density in layered metal(1v) phosphates. As reported previously,21y22 straight calculations show that the aluminium Keggin ion pillars should fill the interlayer regions of these compounds, leaving no micropores.This expectation has generally been confirmed for a-zirconium phosphate (a-ZrP).22 However, porous materials with a large surface area have been surprisingly obtained in the case of very hydrolysable MIv phosphates, such as a-tin and a-titanium phosphate^.'^-'^,^^ This fact could be related to a decreased charge density due to hydrolytic loss of acid phosphate groups.2'*22 Note, however, that the loss of phos- phates essentially produces disordered structures in which a large degree of mesoporosity has been found.,, Some porosity has also been obtained for a-ZrP when colloidal suspensions are used for the ~illaring.''g~~ While the interaction of the A1 species with the a-zirconium phosphate layers has been put in evidence by 27Al MAS NMR spe~tra,'~.'~.~~ the nature of the A1 species present in the final solids is, however, still contro- versial.Thus, in our opinion, further investigation seems to be necessary in order to clarify if Al-pillared a-MIV phosphates, in which the microporosity can be unambiguously attributed us to obtain a better insight into the pillaring procedure and the nature of the intercalated ions. To improve the intercalation of the large Keggin ion, amines with different alkyl chain lengths are inserted in the a-ZrP structure prior to the pillaring. Experimental Materials Synthesis of butylamine-ZrP (BuAZrP). Crystalline a-ZrP [Zr( HP04)2.H,0] was synthesized according to the method of Clearfield and Styne~,'~ by refluxing a gel in a 12 mol dm-3 phosphoric acid solution for 72 h.Exposing a-ZrP to an aqueous solution of butylamine (0.1mol dm-3) for 1 week resulted in the completely exchanged Zr( PO,), .2C4H,NH3-H,0 (BuAZrP). Following the same method, other amines (CnH2n+1NH2,n =1-3,5,6) were intercalated into a-ZrP. Synthesis of the A1 solution. The A1 solutions were prepared by adding dropwise at room temperature NaOH (0.2 mol dm-3) to A1C13.6H,0 (0.2 mol dm-3) with vigorous stirring until the desired hydrolysis degree (OH :Al) was reached. OH :A1 ratios of 1.85 and 2.3 were chosen and correspond to A1 concentrations of 0.07 and 0.06 mol dm-3, respectively. After the addition, the solutions were heated at reflux for 24 h. In a second series of experiments, the solutions were used without prior reflux.Two other A1 solutions were prepared. First, KOH and secondly, H20 instead of NaOH were added to AlC136H,0 according to the same OH: A1 ratios or A1 concentrations. A1 intercalation. Under continuous stirring, the BuAZrP was added to the pillaring solution according to a ratio of 1 g per 100 ml at three different temperatures (0"C, room temperature and 70 "C). After 4 h, the solid was filtered off, washed until C1- -free and air-dried. J. Mater. Chem., 1996, 6(2), 239-246 239 pH titration curves. A sample of 0.2 g a-ZrP was contacted with 20 ml of a 0.1 mol dm-3 MCl (M =Na, K) solution. Afterwards an amount of butylamine (0.1 mol dm-3) was added and the suspensions were equilibrated for 1 day under continuous stirring.Finally the pH of the solutions were measured. Measurements X-Ray diffraction (XRD) measurements were recorded on a Philips PW 1840 powder diffractometer (45 kV, 30 mA), using Ni-filtered Cu-Ka radiation. Thermal analyses were performed on a Mettler TG50 ther- mobalance, equipped with an M3 microbalance and connected to a TClOA processor. The thermograms were recorded from ambient temperature to 800°C at 10°C min-' in a dry nitrogen flow. A Nicolet 510 FTIR spectrometer, loaded with a Spectra Tech diffuse reflectance cell, was used for the collection of the IR spectra. 200 scans were collected in the range 400-4000 cm-' with a resolution of 4 cm-'. The samples were finely ground, diluted with dry KBr and mounted in a micro- cup.All the spectra were converted to Kubelka-Munk units. The concentrations of A13+, Na+ and K' were determined with a Perkin Elmer 5100 Zeeman atomic absorption spec- trometer (AAS). Na' and K+ were atomized in an air-acetylene flame and measured at wavelengths of 589 and 766.5nm, respectively. A13+ was measured in a HGA600 graphite furnace at a wavelength of 394.4 nm. The ion content in the samples was calculated by subtracting the remaining ion concentration in the filtrate from its content in the initial solution. 27Al NMR spectra of the solutions were measured with a Bruker AMX 300 apparatus at 78.208 MHz. 128 scans were collected using a repetition time of 5 s and a pulse time of 9 ps.The reference was Al(OH),- in D20, showing an NMR signal at 6 80. The integration values of the A1 solutions can be compared with each other because the intensity of the reference line is always set at a constant value. The same instrument was used for the solid-state 27Al MAS NMR spectra, operating at a magnetic field strength of 7T. The standard settings were 1600 scans, a pulse time of 0.3 ps, a pulse angle of 10 ",a repetition time of 2 s and a spinning rate of 12 kHz. Signal positions are referenced with respect to a 0.5 mol dm-3 A1(N03)3 solution at 6 0. A comparison between the integration values of the different samples became possible by a correction for their mass. Surface area measurements were obtained with an Autosorb- 1 from Quantachrome.However, the surface areas were very low (SBET=10 m2 g-') and outgassing the samples at 200 "C resulted in amorphous products. Results and Discussion In the first series of experiments the influence of the reaction temperature (BuAZrP +A1 solution) and the amount of ah- minium added was examined. A solution with an OH :A1 ratio of 1.85, heated at reflux for 24 h and a final pH of 3.95 was used throughout these experiments. The amount of A1 was changed by diluting the A1 solution with demineralized water until a constant ratio of 100 ml (g BuAZrP)-' was reached. Influence of the temperature The XRD patterns of the AlZrP powders, synthesized in an ice-bath, at room temperature (RT) and at 70 "Care illustrated in Fig.1. Increasing the temperature corresponds to a decreas- ing dOo2spacing. The reactions carried out in an ice-bath an! at RT both result in samQ1es with a dOo2 spacing of 11.9 A (28=7.45 "). A peak at 9.9 A (28=8.96 ") is observed a! 70 "C. Based on the dimensions of the Keggin ion (9.5 x 7 A2) and 240 J. Muter. Chem., 1996, 6(2), 239-246 2eldegrees Fig. I XRD patterns of the AlZrq samples prepared at: (a) 7q"C (9.9 A); (b) room temperature (11.9A); and (c) 0 "C (ice-bath; 11.9 A) the thickness of an a-ZrP layer (6.5 A),the insertio; of this ion should give a dOo2spacing between 13.5 and 16A. Thus, for all the reaction temperatures the basal spacings are too small to be due to the Keggin ion. OthFr a~thors'~,~~ reported similar XRD phases and found a 9.4 A phase at 70 "C and a 10 A phase at RT for a-titanium phosphate.They ascribed this expansion to the intercalation of the A1 monomer, [A1(H20)6]3'. However, the occurrence of different dOo2spac-ings at the examined temperatures points to the presence of different pillaring species. The AlZrP samples synthesized both at RT and 70°C were studied in detail. Influence of the amount of A1 In Fig. 2 the XRD patterns as a function of the amount of Al, added at RT [Fig. 2(u)] and at 70°C [Fig. 2(b)], are shown. Increasing the amqunt of A1 results in a progressive decrease of the peak at 17.6 A (28=5.03 ") correspondjng to the butylam- ine intercalated phase [second order; 8.9 A (28=9.28 ")I. At the same time, a new peak at 11.9 A (RT) or 9.9 A (70 "C) appears.The amine peaks disappear at a loading of 3.5 mmol A1 g-' for the RT reaction and at 2.5 mmol 8-l at 70°C. The percentage of butylamine retained by the solid, determined by the mass loss in the temperature range 150-400 "C,is reported in Table 1. Even at high A1 loadings a small mass loss due to the amine removal is detected. Furthermore, Table 1 shows that the amine deintercalation at 70°C is faster than at RT. This is probably due to the higher mobility of the amine molecules and pillaring ions, giving rise to a faster exchange. Identificationof the 11.9 A and 9.9 A peaks Caution needs to be taken in ascribing the 9.9 A and 11.9 A peaks, because these may be due to the A13' species or to Na' ions.The pillaring solution, which is prepared by a controlled hydrolysis of AlC1, -6H20 with NaOH, contains both A1 species and Na' ions. AAS measurements revealed that both ions are indeed taken up during the reaction. For comparison, the half-exchanged and completely exchanged NaZrP samples were prepared. According to the literature data,26 the former [ZrNaH(P04),.5H20] has a d0020spacing of 11.8 A and the latter [Zr(NaP0,),.3H20] of 9.8 A. These XRD reflections correspond very well with the experimental dOo2spacings of the AlZrP samples. .$ 0 2 4 6 8 10 12 14 16 C42L.-0 2 4 6 8 10 12 14 16 28/degrees Fig.2 XRD patterns of the AlZrP samples as a function of the amount of A1 added (mmol g-'), (a) at RT: A, 1.5; B, 2.5; C, 3; D, 3.5; E, 7 mmol, and (b) at 70 "C: A, 1.5; B, 2; C, 2.5; D, 3; E, 7 mmol Table 1 Deintercalation of the butylamine as a function of the amount of A1 added amine retained (mass%) A13+ added/mmol (g BuAZrP)- RT 70 "C 0 32.6 32.6 1.5 17.9 14.9 2 12.9 10.2 2.5 6.3 3.5 3 4.8 1.2 3.5 1.6 1.3 4 1.2 1.2 7 1.1 1.25 To evaluate the impact of the interfering ions present in the A1 solution on the intercalation, other pillaring solutions were prepared: (1) instead of NaOH, KOH was used for the hydrolysis of A1C13-6H20; (2) the A1 solution was directly prepared by the addition of water to the A1Cl,.6H20 salt.This results in an acidic A1 solution free from any competitive cations. The acidic nature of the solution prevents hydrolysis and polymerization of the A1 species and Al(H20)63+ is essentially present in the solution." Irrespective of the reaction temperature (RT or 70°C) the samples prepared with the KOH hydrolysed A1 solution exhibit a dOo2spacing at 8.1 A, consistent with the formation of ZrKH(P04)2.H20,26while the samples prepared with the H20 hydrolysed A1 solution exhibit a do,, spacing of 9.8 A.A summary of the basal spacings of the samples prepared with the different A1 solutions are listed in Table 2. The samples prepared with NaOH, H20 and KOH hydrolysed A1 solutions are referred to as AlZrPl, AlZrP2 and AlZrP3, respectively. For comparison, the basal spacings of the ZrNaH( P04),.5H20 (NaHZrP), Zr( NaP04)2.3H,0 (NaNaZrP) and ZrKH( P04),.H20 (KHZrP) phases are included.Because the 8.1 A phase does not correspond to any interlayer distance caused by the A1 intercalation and this distance is exactly the same as the KHZrP phase, it is concluded that this particular XRD peak is a result of the K+ ion exchange and not of the A1 intercalation. The same conclusion can be drawn for the AlZrPl samples prepy-ed with the NaOH-hydrolysed A1 solu-tion at RT. The 11.9 A phase is ascribed to Na+ uptake with the formation of !he NaHZrP phase. However, based on the XRD peak at 9.9 A occurring for both the AlZrPl and AlZrP2 samples synthesized at 70 "C, one cannot distinguish between the types of species [Na+ or Al(H20)63'] that may be responsible for this spacing.Therefore, AAS measurements, the features related to the pH titration curves and FTIR spectra were consulted. In Table3 the A13+ and Na+ uptake for the samples prepared with the NaOH and H20 hydrolysed Al-solutions are summarized. The same terminology as in Table 2 is main- tained. Owing to the absence of interfering ions only A13+ is detected for the AlZrP2 samples. For both the RT and 70°C reaction, an uptake of 1.4mmol A1 (g BuAZrP)-l is found. This is in good agreement with the maximum theoretical value, which can be calculated as follows. The cation exchange capacity (CEC) of a-ZrP is 6.64 mequiv g-' and reduces to 4.47 mequiv 8-l for BuAZrP. The latter value is divided by three and becomes 1.49 mmolg-l for a trivalent ion like A13+.The uptake of both A13+ and Na+ is different for the AlZrPl samples at RT and 70°C. At 70°C a smaller amount of Na+ and a larger amount of A13+ is intercalated. Because both ions (A13+, Na+) are present at the examined temperatures no direct evidence is provided for assigning the 9.9 A phase to the intercalation of A13+ or Na+ . Nevertheless, note that the amount of intercalated Na+ is smaller than half the exchange capacity (2.235 mmol g-'). Ion exchange studies26 of a-ZrP with Na' show that only if the Na+ uptake is higher than half the exchange capacity, the completely exchanged NaNaZrP phase with a dOo2spacing of 9.9 A occurs. A,t lower loadings the half-exchanged phase (NaHZrP, 11.9 A) is present. Thp if, despite the presence of the XRD peaks (9.9 A and 11.9 A), the stated concentrations of Na+ are not reached, then the interlayer distance is not determined by the Na' uptake.This is true for the AlZrPl sample prepared at 70"C, which has, despiie its low Na+ uptake (0.55 mmol g-'), a dOo2spacing of 9.9 A. Therefore, we suggest that the interlayer distance of this sample, which contains even less Na+ than the RT sample, is governed by the intercalation of Al( H20)63+ species. In addition, the pH values observed during the intercalation process are too low for the formation of the completely exchanged Na+ form of a-ZrP. These pH values were com- J. Matev. Chem., 1996, 6(2), 239-246 241 Table 2 Basal spacing (A)of the AlZrP and NaHZrP, NaNaZrP and KHZrP samples temperature AlZrP 1 AlZrP2 AlZrP3 NaHZrP NaNaZrP KHZrP RT 11.9 9.8 8.1 11.9 9.9 8.1 70 "C 9.9 9.8 8.1 11.9 9.9 8.1 Table 3 Na' and A13+ uptake (AAS) in the BuAZrP sample AlZrPl/ AlZrP2/ mmol (g BuAZrP)-' mmol (g BuAZrP)-' Mn+a RT 70 "C RT 70 "C Na+ 1.55 0.55 +~13 1 1.35 1.4 1.4 a M, ion; n, charge of the ion.pared with the pH titration curves of a-ZrP suspended in NaCl or KCl solution and titrated with butylamine. Butylamine, instead of the corresponding hydroxide (NaOH or KOH), is used to mimic the reaction conditions of BuAZrP with the A1 solution as closely as possible. The obtained titration curves are similar to the traditional pH curves (MC1+ MOH, M =Na, K) reported in the literature.26 During the intercalation process, butylamine is removed for the A1 species and Na+ or K+ ions and gives rise to a pH increase of the pillaring solution. For the A1 solution with OH :A1 = 1.85, the initial pH is 3.95 and increases to 4.15 after the addition of BuAZrP.Consulting the titration curves, this pH value corresponds to the formation of the half-exchanged Na' phase (NaHZrP). It is only at higher pH values (pH=6) that the completely exchanged Na' phase (N!NaZrP) will occur. This is a further indication that the 9.9 A phase, formed by the reaction of BuAZrP with the A1 solution at 70"C, is due to the intercalation of Al(H20)63+ and not to the additional uptake of Na+ with the formation of the NaNaZrP phase. Finally, FTIR spectra reveal additional information about the nature of the AlZrP samples.In Fig. 3 the IR spectra of the NaHZrP, NaNaZrP, AlZrPl and AlZrP2 samples are plotted. The spectrum of the AlZrPl sample prepared at RT (Fig. 3B) shows a good resemblance to that of NaHZrP (Fig. 3A). The peaks at 3646,3597 and 3496 cm-' are ascribed to the 0-H stretching vibration of the remaining free POH group and the asymmetrical and symmetrical 0-H stretching vibrations of water, re~pectively.~~ A new peak at 1320 cm--', which has not yet been assigned, appears only for these two samples. The peaks in the region of 1250-400cm-' are due to the PO-H bending vibrations and the P-0 stretching and bending vibrations. Based on the XRD patterns, the IR spectrum of the AlZrPl sample prepared at 70 "C (Fig.3D) should resemble the spectrum of NaNaZrP (Fig. 3C) or that of the AlZrP2 sample (Fig. 3E). However, the spectrum of AlZrPl at 70°C shows very broad bands which do not correspond to the sharp peaks of the spectrum of NaNaZrP. In the latter sample two types of water, which differ in their position between the layers, are present and give rise to four sharp OH stretching vibrations at 3540, 3475, 3250 and 3190 cm-1.27 The AlZrPl sample at 70 "C seems to be similar to the AlZrP2 sample. Although FTIR spectroscopy does not provide direct infor- mation about the nature of the intercalated species, it gives, together with XRD, AAS and pH titration data, further evidence for the presence of the postulated intercalated species at the examined temperatures.In summary, we can conclude that the temperature and the type of competitive ion (Na' ,K') are crucial in the intercal- ation step. Although both the competitive ion and A13' ions are present in the AlZrP samples, the question of which ion is responsible for the dOo2spacings has been raised. A solution is proposed based on the evaluation of the affinity differences between the competing ions for the examined reaction conditions. The uptake differences (Na', A13+) at RT and 70°C can be explained in terms of the charge of the exchanging ion and the layer charge density of the substrate. The ion-exchange process starts from the external part of the crystal and moves on to the centre of the crystal.If the charge of the ion is high a strong interaction with the exchange sites of the substrate occurs and the diffusion of the ion to the more inner sites is U I 3500 3000 2500 2Ooo 1500 1000 400 wavenurnber/cm-1 Fig.3 FTIR spectra of A, NaHZrP; B, AlZrPl at RT; C, NaNaZrP; D, AlZrPl at 70°C; E, AlZrP2 hindered. For a-ZrP the layer charge is very high (6.64 mequiv g-') and therefore the exchanging ions with a high charge have a strong interaction with the exchange sites or POH groups. For a monovalent ion this interaction is weaker and it can move freely to the interior of the crystal. Thus, although the affinity of a multivalent ion is higher, its diffusion in the interlayer space is hindered and as a result the ion exchange of the competitive monovalent ion will be preferred.This mechanism is believed to occur at RT for the system Na'-A13+. Increasing the temperature enhances the diffusion of the ions in the interlayer space, and now ions with a higher charge can also be inserted. This is observed at 70°C, where A13+ uptake is preferred to Na' . For the K+-A13' system, K' is exchanged preferentially to A13+ at both temperatures. The doo2spacing (8.1 A) consistent with the formation of the KHZrP phase, is found at RT and 70°C. Increasing the temperature cannot reverse the uptake of the ions, as in the case of the Na'-A13' system. The difference between both systems is related to the affinity of the competitive ions towards the POH groups of a-ZrP. From the potentiometric titration curves previously men- tioned, it is concluded that K+ is preferred to Na'.This higher preference or affinity suggests, therefore, that even at higher reaction temperatures the monovalent ion K+ is exchanged and not the A13' ions. Besides this mechanism, a higher temperature can improve 242 J. Mater. Chem., 1996,6(2), 239-246 the A13+ uptake in a second way. Increasing the temperature can result in decomposition of the large polymerized A1 species (Keggin ions) into the smaller monomers Al(H20)63+. Because of their smaller dimensions these monomers are more easily intercalated. A very instructive tool for evaluating the influence of the temperature and other parameters on the pillaring solution, and more specifically on the type of A1 species, is 27Al NMR spectroscopy.Secondly, the inability of the A1 Keggin ion intercalation can be screened with this technique. The intercalation failure may be due to the initial absence of the Keggin ions in the solution, to a structural breakdown upon intercalation or to its inaccessibility in the ct-ZrP struc- ture. A distinction between these possibilities can be made by comparing the solution 27Al NMR spectra with the correspond- ing solid-state 27Al MAS NMR spectra and XRD patterns of the AlZrP samples. "A1 NMR of the pillaring solutions The 27Al NMR spectra usually shows three signals due to: (1) the octahedral coordination of the monomeric [A1(H20)6]3+ at 6 0; (2) the octahedral coordination of the dimeric to pentameric A13+ at around 6 5, and (3) the tetrahedral coordination of the central A13+ atom of the Keggin ion at 6 63.' This central A1 atom is surrounded by twelve A106 octahedra. The resonance signal of the octahedra is not visible in the NMR spectra because of their non-symmetrical environment.A strong electrical field gradient occurs, giving rise to a very broad band.g The influence of the hydrolysis degree (OH:Al), the ageing temperature (24 h reflux or RT ageing), the dilution and the reaction temperature on the occurrence of the Keggin ions in the pillaring solution was examined. The former two param- eters are summarized in Table 4 and the latter two in Tables 5 and 6, respectively. The following observations and con- clusions can be made.(1) Ageing at reflux temperature results in a strong decrease of the 6 63 signal and a more intense 6 0 signal. (2) A higher hydrolysis degree (2.3 us. 1.85) favours the 6 63 signal slightly, but this influence is more pronounced for the 6 0 signal. Increasing the OH :A1 ratio reduces this signal significantly. These observations are in good agreement with the results of Schoonheydt et uL9 (3) A dilution of the solution results in a higher pH that can enhance the hydrolysis process with the formation of more and larger polymerized species. At an 0H:Al ratio of 1.85, no signal at 6 63 occurs for both the non-diluted and the diluted solutions (Table 5). The dilution ratio (4.7) is in good agreement with the ratio of the integration values of the monomeric signals (4.8) at 6 0 for the solutions containing 7 mmol and 1.5 mmol A13+ 8-l.Therefore the dilution and the subsequent increase of the pH (3.95 to 4.3) did not influence the polymerization of the A1 species. A good Table 4 27Al NMR integration values of the A1 solutions (influence of the ageing temperature and the hydrolysis degree) OH :A1 = 1.85 OH :Al= 2.3 integrated peak, 6 RT reflux RT reflux 0 1.13 1.18 0.22 0.38 63 0.2 0.01 0.27 0.02 Table 6 27Al NMR integration values of the A1 solutions (influence of the reaction temperature) OH :A1=2.3 T OH :A1= 1.5 mmol A160 1.85 3+ g-l 6 mmol A13+ g-' 60 6 63 RT 0.68 0.8 0.74 70 "C 0.71 0.8 0.76 correspondence between the dilution ratio (4) and the ratio of the integration values of the 6 0 and 6 63 signals is also observed for a solution with OH:Al=2.3. Again the pH variation (4.3 us.4.6) did not initiate a further polymerization. (4) The reaction temperature does not influence the A1 species present in the solution (Table 6). Increasing the temperature to 70 "C does not generate new Keggin ions (compare OH :A1= 1.85; 1.5 mmol A13+ g-' at RT and at 70°C). On the other hand, the Keggin ions are stable at 70 "C and no decomposition to smaller species occurs (compare OH :A1 =2.3; 6 mmol A13+ 8-l at RT and at 70°C). Influence of the A1 solution on the pillaring According to the NMR data, the non-diluted A1 solutions with the optimal (OH:Al=2.3, no reflux) and least preferable (OH:Al= 1.85, reflux) parameters were used for the A1 pil- laring.In Fig. 4 the XRD patterns of these two AlZrP samples are shown. The pillaring reaction was performed at 70°C. No obvious difference *between these samples was observed. The dooz spacings (9.9 A) are in agreement with our earlier resul!s and a pillaring at RT gives an interlayer distance of 11.9 A. This means that changing the synthesis parameters of the A n I I I I -re-2 4 6 8 10 12 14 16 2eidegrees Fig. 4 XRD patterns of the AlZrP samples prepared at 70°C with A1 solutions: A, OH :A1 = 1.85, reflux; B, OH :A1 =2.3, no reflux Table 5 27Al NMR integration values of the A1 solutions (influence of the dilution) OH :A1=2.3 mmol AP + (g BuAZrP) - OH :A1= 1.85 mmol A13+ (g BuAZrP)-' added 60 added 60 6 63 1.5 0.7 1.5 0.2 0.18 7 3.3 6 0.8 0.74 ratio: 4.7 4.8 4 4 4.1 J.Muter. Chem., 1996,6(2), 239-246 243 pillaring solution does not influence the final intercalated AlZrP. Although Keggin ions are present in the solution (OH :A1=2.3), no intercalation of these ions occurs. The absFnce of the Keggin ion in the ZrP structure (dOo2=13.5 or 16A) points to the inability of this ion to intercalate in the host structure. This is further investigated by 27Al MAS NMR spectroscopy. "Al MAS NMR spectroscopy of the AlZrP samples In order to obtain a better idea about the processes taking place during the intercalation, the solid-state 27Al MAS NMR spectra were compared with the solution 27Al NMR spectra.For the solid-state NMR, all the A1 is visible, and the octahedral signal of the Keggin ion can now be detected (6 0). Because the A1 Keggin ion consists of a tetrahedral A1 atom surrounded by twelve A1 octahedra, the ratio of the octahedral to tetra- hedral signal (0:T) is 12. Table 7 shows the integration values of the 6 63 and 6 0 signals as a function of several parameters (ageing temperature, hydrolysis degree, reaction temperature and dilution). Owing to the initial absence of Keggin ions in the refluxed A1 solutions with OH :A1 =1.85, no Keggin ions are observed in the solid samples. Nevertheless, the spectra of the diluted samples do exhibit a tetrahedral signal, giving an O:T ratio of 17, meaning that besides the octahedra of the +Keggin ion, Al( H20),3 monomers are also present.This tetrahedral signal was not observed in the 27Al NMR spectra of the diluted solutions. However, these spectra were recorded before the addition of BuAZrP and therefore no additional pH increase due to the desorption of butylamine was possible. pH measurements revealed that the desorption strongly affects the pH of the diluted solutions. Before the BuAZrP addition, the pH of the diluted A1 solution with OH :Al= 1.85 is 4.3; after the addition, a pH of 6.7 is observed. For the diluted A1 solution with OH :A1=2.3, pH values of 4.6 and 7, respectively, are measured. This large pH increase initiates a further hydroly- sis and polymerization of the A1 species. The latter might also explain the occurrence of a tetrahedral signal for the AlZrP sample prepared with a refluxed, non-diluted A1 solution (OH :A1=2.3).Refluxing the A1 solution destroys the Keggin ions (cf: Table4), but after contact with the BuAZrP sample, new Keggin ions can be formed. The initial pH of the solution is 4.3 and increases to 5.6 after desorption of the butylamine. No such effect is observed for the AlZrP sample prepared under the same conditions with a refluxed A1 solution with OH: Al= 1.85. Here, the pH increase is too small (3.95 to 4.15) to generate new Keggin ions. The reaction temperature influ- ences the 0:T ratio and a higher ratio is observed at 70 "C. This might point to a higher uptake of octahedrally coordi- nated A1 monomers with respect to the RT reaction.A comparison of the solution 27Al NMR and the solid-state 27Al MAS NMR spectra shows that the absence of Keggin ions in the initial A1 solution is not the determining factor for the uptake of these ions in the AlZrP samples. The conditions under which the exchange occurs are more important. If the exchange of the 'pillaring' ions (Na+ or A1 species) for the butylamine present in the sample is accompanied by a large pH increase, then new Keggin ions can be formed and adsorbed. Therefore, AlZrP samples prepared from A1 solutions without Keggin ions (e.g.refluxed solutions) can still exhibit a tetrahedral signal in the 27Al MAS NMR spectra. If the pH increase is small, the uptake of Keggin ions will depend on their presence in the initial A1 solution.Based on XRD and 27Al MAS NMR data, we can conclude that for the samples which exhibit a tetrahedral signal, Keggin ions are adsorbed on the external part of the crystal and not in the interlayer spacp. With respect to the dimensions of the Keggin ion (7 x 9.5 A2) an interlayer distance between 13.5 and 16A should be found. Its large dimensions, high charge and the high CEC of a-ZrP are probably the major drawbacks for a smooth intercalation. Influence of the type of amine Because no Keggin ions had been inserted so far and the reason for this lack was attributed to accessibility problems, the alkyl chain-length of the amine was changed. It was presumed that larger amines might allow the intercalation of larger A1 species and under certain conditions of the type of amine and the loading, a-ZrP becomes completely delaminated, thereby enhancing the interaction with the exchange sites and the intercalation of large molecules or ions considerably.28 Completely intercalated a-ZrP compounds with different types of n-alkylmonoamines (C,H2,+ ,NH2; n =1-6) have been prepared and added to the pillaring solution (OH :A1=2.3, RT ageing and reaction at 70 "C).The amineZrP samples were added as powders or colloidal dispersions to the A1 solution. Fig. 5 shows the X-ray diffraction patterns of the obtained samples. Amines with short chains gave rise to completely amorphous samples with no XRD reflections (methylamine and ethylamine) or broad peaks (propylamine), while in the case of longer-chain amines (e.g.hexylamine), the peaks of the Table 7 27Al MAS NMR integration values of the AlZrP samples (influence of ageing temperature, hydrolysis degree, reaction temperature and dilution)" OH :A1=1.85 T(ageing) 0H:Al ratio T(reaction) dilutionb integrated peaks, 6 RT reflux 1.85 2.3 RT 70 "C 2.5 7 1408 1336 345 1408 2270 1408 746 375 75 71 - 75 146 75 43 - 18.8 18.8 18.8 15.5 18.8 17 a Standard reaction conditions: non-diluted A1 solution (OH :A1=2.3, RT ageing) and T(reaction)=70 "C. Dilution expressed in mmol (g BuAZrP)-'. Table 8 Influence of the alkyl chain-length on the Na+, A13+ uptake and determination of the % residual amine at RT and 70 "C (non-diluted A1 solution; OH :A1=2.3) PrAZrP BuAZrP PeAZrP HeAZrP RT 70 "C RT 70 "C RT 70 "C RT 70 "C Na+/mmol g-' 2.7 1.7 2.1 1.85 0.4 1.6 0.7 0.9 AP+/mmol g-' 1.4 1.7 1.5 1.8 1.1 1 0.8 1.5 amine/mass% 1.6 1.3 1.8 1.6 11.5 2 24.5 19.8 244 J.Muter. Chem., 1996, 6(2), 239-246 0 2 4 6 8 10 12 14 16 wdegr~s Fig. 5 XRD patterns of the AlZrP samples, prepared with amineZrP samples of different alkyl chain lengths C,H2,+ ,NH, (n =1-6): A, MeAZrP and EtAZrP; B, PrAZrP; C, BuAZrP; D, PeAZrP; E, HeAZrP original amine intercalated compounds could still be observed. The chemical and thermal analysis of the obtained samples (Table 8) indeed confirm that a complete replacement of long alkyl chain amines with A1 species was not obtained.At RT pentylamine is not completely exchanged for the Na+ and A1 species. Consequently, the highest ion uptakes are measured for PrAZrP and BuAZrP. In order to obtain the favourable conditions of a delaminated structure, Alberti et aL2' established a method consisting of the intercalation of propylamine at 50% of the exchange capacity. A stable, colloidal suspension is formed. In Fig. 6 the A A I I I I 1 I I I 0 2 4 6 8 10 12 14 16 28/degrees Fig. 6 XRDpatterns of the AlZrP samples prepared with PrAZrP: A, colloidal suspension (50% exchange); B, crystalline powder ( 100% exchange) XRD patterns of the AlZrP samples, prepared by the addition of the PrAZrP suspension (A) and the 100% intercalated PrAZrP powder sample (B), respectively, to the A1 solution, are compared.After the reaction of the colloidal suspension with the A1 solution a very amorphous material occurs. The same amorphous XRD patterns are observed for the MeAZrP (methylamineZrP) and EtAZrP (ethylamineZrP) samples. The latter samples also form a colloidal suspension, but at 100% intercalation (cf: Fig. 5). Thus, although the contact of the exchange sites or POH groups with the A1 species is enhanced and the replacement of the amine molecules for the A1 Keggin ions is improved, no regeneration of the crystalline structure is obtained. This amorphous pattern again does not imply a true intercalation of the Keggin ions. In summary, we can state that although no Keggin ions are inserted, the most suitable precursors for the exchange with A1 species are the BuAZrP and PrAZrP samples, which are reacted at 70°C.For shorter chain amines, very broad XRD peaks or even amorphous patterns are observed, while amines with a longer alkyl chain are not completely exchanged for the pillaring ions. The latter samples do not promote the intercalation of larger A1 species. Conclusions The A1 pillaring of a-ZrP, previously expanded with n-alkyl- monoamines, has been studied. A synthesis procedure and pillaring solutions (AlCl, +NaOH) similar to those used for the preparation of PILCs, did not result in a thermally stable or a porous AlZrP. Changing the reaction or exchange tem- perature gave rise to different interlayer distances, consistent with the intercalation of different ions.The preference of these ions is strongly related to the affinity of the respective ions for the exchange sites or POH groups of a-ZrP and the exchange or reaction temperature. For the NaOH-hydrolysed A1 solu-tion, the combined results of XRD, AAS, FTIR and pH titration curves showed that th? interlayer distance of the samples prepared at RT (b1.9 A) is characterized by Na+ exchange, and at 70 "C (9.9 A) by an Al(H20),3+ intercalation. If the A1 solution is synthesized by a controlled hydrolysis with KOH instead of NaOH, K+ will be preferred tq A13+ at both temperatures. An interlayer distance of 8.1 A, which corresponds to the formation of ZrKH( P04)2.H20, was found.The small interlayer distances illustrate that no A1 Keggin ions are taken up. The reason for this failure was monitored, using 27Al NMR spectroscopy. The initial A1 solutions and the solid AlZrP samples were investigated. Although the synthesis parameters of the A1 solution have a strong influence on the occurrence of the A1 Keggin ions, the AlZrP samples always exhibit the same small interlayer distances. Based on XRD and 27Al MAS NMR data, it was demonstrated that the Keggin ions, which are present in the AlZrP samples, are situated on the external surface. Therefore, the lack of the Keggin ion intercalation is not due to the decomposition of the ion, but to its inaccessibility in the a-ZrP structure.In order to improve the accessibility, the influence of the type of amine used for the expansion of a-ZrP was determined. The intercalation of small amines (methyl- and ethyl-amine) as well as propylamine at 50% of the exchange capacity give rise to a completely delaminated structure. Although this might improve the contact with the Keggin ions, an XRD pattern characteristic of an amorphous substrate was found. Larger amines (pentyl- and hexyl-amine) are more difficult to exchange for the ions present in the pillaring solution and part of the amine is retained in the sample. These attempts did not lead to a better pillaring and the addition of BuAZrP to the A1 solution gave the best-resolved XRD peaks and complete exchange of the amine.Nevertheless, it should be stressed that a different synthesis method and different A1 solutions might result in a true A1 J. Mater. Chem., 1996, 6(2),239-246 245 Keggin ion intercalation into a-ZrP. In this context successful 12 D. J. MacLachlan and D. M. Bibby, J. Chem. SOC., Dalton Trans., and 1989,895.attempts have already been presented previo~sly,~~~~~~~~ 13 P. Maireles-Torres, P. Olivera-Pastor, E. Rodriguez-Castellon, we are currently undertaking research in this field. My sincere thanks and appreciation are expressed to Prof. G. Alberti and Dr. F. Marmottini for their valuable discussions and comments. The authors also wish to thank Mrs. H. Geerts kindly for performing the 27Al MAS NMR measurements and the IUAP for their financial support.K. P. is financed by the Flemish Institute promoting scientific and technological research for industry (IWT). References 1 D. E. W. Vaughan, in Catalysis Today, ed. R. Burch, Elsevier, Amsterdam, 1988, vol. 2, p. 187. 2 J. H. Purnell, in Pillared Layered Structures: Current Trends and Applications, ed. I. V. Mitchell, Elsevier, London, 1990, p. 107. 3 A. Clearfield, M. Kuchenmeiska, J. Wang and K. Wade, in Zeolite Chemistry and Catalysis, ed. P. A. Jacobs, N. I. Jaegar, L. Kubelkova and B. Wichterlova, Elsevier, Amsterdam, 1991, p. 485. 4 A. Clearfield, in Surface Organometallic Chemistry: Molecular Approaches to Surface Catalysis, ed. J. M. Basset, B. C. Gates, J. P. Candy, A. Choplin, M. Leconte, F. Quignard and C.Santini, Nato AS1 Series, vol. 231, Kluwer, Dordrecht, 1988, p. 271. 5 C. J. B. Mott, in Catalysis Today, ed. R. Burch, Elsevier, Amsterdam, 1988, vol. 2, p. 199. 6 N. Lahav, U. Shani and I. Shabtai, Clays Clay Miner., 1978, 26, 107. 7 T. J. Pinnavaia, M. S. Tzhou, S. D. Landau and R. H. Raythatha, J. Mol. Catal., 1984,27, 195. 8 A. Molinard, N. Maes, K. Peeters and E. F. Vansant, in Separation Technology, ed. E. F. Vansant, Elsevier, Amsterdam, 1994, p. 437. 9 R. A. Schoonheydt, J. Van Den Eynde, H. Tubbax, H. Leeman, M. Stuyckens, I. Lenotte and W. E. E. Stone, Clays Clay Miner., 1993,41, 598. 10 G. Fu, L. F. Nazar and A. D. Bain, Chem. Muter., 1991,3,602. 11 J. F. Lambert, S. Chevalier, R. Franck, H. Suquet and D. Barthomeuf, J.Chem. SOC., Faraday Trans., 1994,90,675. A. Jimenez-Lopez, L. Alagna and A. A. G. Tomlinson, J. Muter. Chem., 1991,1,319. 14 A. Espina, J. B. Parra, J. R. Garcia, J. A. Pajares and J. Rodriguez, Muter. Chem. Phys., 1993,35,250. 15 P. Maireles-Torres, P. Olivera-Pastor, E. Rodriguez-Castellon, A. Jimenez-Lopez and A. A. G. Tomlinson, J. Solid State Chem., 1991, 94, 368. 16 P. Maireles-Torres, P. Olivera-Pastor, E. Rodriguez-Castellon, A. Jimenez-Lopez and A, A. G. Tomlinson, J. Muter. Chem., 1991, 1, 739. 17 P. Olivera-Pastor, J. Maza-Rodriguez, P. Maireles-Torres, E. Rodriguez-Castellon and A. Jimenez-Lopez, J. Muter. Chem., 1994,4, 179. 18 A. Clearfield and M. Kuchenmeister, in Supramolecular Architecture, ed. T. Bein, ACS Symp. Ser., Washington, 1992, p. 128. 19 E. Rodriguez-Castellon, P. Olivera-Pastor, P. Maireles-Torres, A. Jimenez-Lopez, J. Sanz and J. L. G. Fierro, J.Phys. Chem., 1995, 99,1491. 20 G. Alberti, M. Casciola and U. Costantino, J. Colloid Interface Sci., 1985,107,256. 21 G. Alberti, in Multifunctional Mesoporous Inorganic Solids, ed. C. A. C. Sequeira and M. J. Hudson, Kluwer, Dordrecht, 1993, p. 179. 22 A. Clearfield, in Multifunctional Mesoporous Inorganic Solids, ed. C. A. C. Sequeira and M. J. Hudson, Kluwer, Dordrecht, 1993, p. 159. 23 A. Clearfield and B. D. Roberts, Inorg. Chem., 1988,27,3237. 24 D. J. Jones, J. M. Leloup, Y.Ding and J. Roziere, Solid State lonics, 1993,61, 117. 25 A. Clearfield and J. A. Stynes, J.Inorg. Nucl. Chem., 1964,26,117. 26 A. Clearfield, in Inorganic Ion Exchange Materials, ed. A. Clearfield CRC Press, Boca Raton, FL, 1982, pp. 20-23. 27 U. Costantino, Thesis Giampaolo Fioro, 1974. 28 A. J. Jacobson, in Soft Chemistry Routes to New Materials, ed. J. Rouxel, M. Tournoux and R. Brec, Trans Tech Publications, Switzerland, 1994, p. 1. Paper 5/050661; Received 31st July, 1995 246 J. Muter. Chem., 1996,6(2), 239-246
ISSN:0959-9428
DOI:10.1039/JM9960600239
出版商:RSC
年代:1996
数据来源: RSC
|
20. |
New porous Ga/Cr mixed oxide pillaredα-zirconium phosphate materials |
|
Journal of Materials Chemistry,
Volume 6,
Issue 2,
1996,
Page 247-252
Manuel Alcántara-Rodríguez,
Preview
|
PDF (745KB)
|
|
摘要:
New porous Ga/ Cr mixed oxide pillared a-zirconium phosphate materials Manuel Alchntara-Rodriguez, Pascual Olivera-Pastor, Enrique Rodriguez-Castell6n and Antonio Jim&nez-Lbpez* Departamento de Quimica Inorgbnica, Facultad de Ciencias, Universidad de Malaga, 29071-Mcilaga, Spain Mixed oligomeric solutions of gallium(II1) and chromium(m), with Ga :Cr mole ratios between 10 :90 and 90 :10, have been prepared by adding n-propylammonium acetate and n-propylamine to the mixed nitrate solutions at pH 4.4-4.5. The mixed oxyhydroxyacetate oligomers thus obtained were intercalated into colloidal a-zirconium phosphate or precipitated with an excess of n-propylamine. Whilst the mixed precipitates show X-ray diffraction (XRD) reflections typical of mixed GaxCr2 -x03oxides after calcination above 400 "C, most of the mixed intercalates were amorphous between room temperature and 800 "C, and only after calcination at 1000 "C were XRD reflections corresponding to zirconium pyrophosphate and Ga/Cr mixed oxides observed.These results suggest that mixed Ga/Cr oxide pillared a-zirconium phosphate materials are formed at 400-800 "C. The mixed pillared materials with Ga :Cr ratios <60 :40 have high BET surface areas (200-300 m2 g-') and micropore volumes (0.08-0.11 cm3 g-').' Higher Ga: Cr ratios lead to materials with BET surface areas lower than 100 m2 g-' and low micropore volumes. The mixed oxide powders also present high BET surface areas (184-236 m2 g-') and micropore volumes (0.08-0.10 cm3 g-'), probably due to a templating effect of the acetate ion.A comparison of the bulk and surface compositions of pillared materials, obtained by chemical analysis and X-ray photoelectron spectroscopy (XPS), respectively, reveal that, at least up to a Ga :Cr ratio of 60 :40, the intercalated species into the phosphate are mixed oligomers. It is known that the surface area and acidity of layered compounds, such as clays, can be considerably enhanced by pillaring with metal oxides.' Like zeolites, pillared clays have been used as selective adsorbents,, ion exchangers3 and cata- lyst~.~Many organic reactions, such as hydration of alkenes, oxidative dehydrogenation of alkanes or deep oxidation of chlorinated organic compound^,^ are catalysed by pillared clays.Metal@) layered phosphates are another family of hosts able to intercalate metal oligomers by ion exchange6 or by acid-base reaction^.^ In previous work we have reported the synthesis and characterization of alumina-and chromia-pillared a-zirconium phosphate materials. The latter were active for the decomposition of isopropyl alcohol and, interes- tingly, the selectivity of the reaction could be controlled by the chromium content.8 However, the number of cations capable of forming oligomers in solution is scarce, and in some cases their existence is restricted to a short pH range or to a short lifetime, as occurs with the gallium(m) oligomers.' This problem is overcome by using mixed oligomeric solutions of cations capable of forming solid solutions of mixed oxides.In this way, a considerable number of metal ions can be interca- lated into layered hosts." The aim of this paper is the preparation and characterization of new Ga/Cr mixed oxide pillared a-zirconium phosphate materials in order to dispose of bifunctional catalysts, with acid sites and oxidative centres for applications in important catalytic reactions. The materials were characterized by X-ray diffraction (XRD), IR spectroscopy, diffuse reflectance UV-VIS spectroscopy, thermogravimetry-differential thermal analysis (TG-DTA), scanning electron microscopy (SEM), X-ray photoelectron spectroscopy (XPS) and N, adsorption. Experimenta1 The host material, a-zirconium phosphate (a-ZrP), was synthe- sized using the sol-gel rnethod.l3 The colloidal a-ZrP was prepared by neutralization (up to 60%) of the cation-exchange capacity with n-propylamine (0.1 mol dm-3).The oligomeric cationic solutions were prepared by dissolv- ing together Ga(NO,), and Cr(N03),.9H20 in water. The pH was controlled at 4.4-4.5 (to prevent hydrolysis of a-ZrP) by adding n-propylammonium acetate (0.1 mol dm-3) and n-propylamine. The mole ratio of OAc-:Cr3+ added was 2.8. The mixed Ga/Cr pillaring solutions, containing a total ([Ga+3]+[Cr"]) equal to ten times the cation-exchange capacity of the a-ZrP (6.64 mequiv. g-'), were contacted with a colloidal suspension of a-ZrP (1g) and heated under reflux for 2 days. After reaction, the solids were separated by centrifu- gation, washed with deionized water up to a conductivity of the washing water of <50 pS, and air-dried.Gallium was analysed by atomic absorption spectroscopy (AAS), and chromium colorimetrically as chromate (A= 372 nm), after dissolving the samples in NaOH-H202. XRD patterns of cast films were recorded using a Siemens D501 diffractometer (Cu-Ka source) and graphite mono-chromator. TG-DTA curves were obtained with a Rigaku Thermoflex instrument (calcined Al,03 as reference, 10 K min -' heating rate). Electronic spectra (diffuse reflectance) were carried out in a Shimadzu MPC 3100 spectrophotometer, and IR spectra were performed with a Perkin-Elmer 883 spectrometer. SEM images were obtained using a JEOL SM 840 instrument. XP spectra were measured with a Fisons ESCALAB 200 R spectrometer with an Mg-Ko! excitation source (hv = 1253.6 eV) and a hemispherical electron analyser.The pressure in the analysis chamber was maintained below lO-'Torr. The binding energies (Eb)were obtained with an accuracy of kO.2 eV, and by charge referencing with the adventitious C 1s peak at 284.9 eV. Adsorption-desorption isotherms were determined in a con- ventional volumetric apparatus at 77 K, degassing the samples at 473 K ( Torr overnight). Results and Discussion The Ga :Cr ratio of the mixed oligomers intercalated into the phosphate uersus the added Ga :Cr ratio in solution is plotted in Fig. 1A. The relationship is linear with a slope of 1.24, which means a higher affinity of the phosphate for gallium, which begins to be apparent from a Ga :Cr ratio of 0.75.Therefore, mixed Ga/Cr oligomers with a higher gallium content with respect to the solution, are intercalated in zirconium phosphate. The overall uptake curve of Ga3+ and Cr3+ is represented in Fig. 1B. A retention in excess of the cation exchange capacity J. Muter. Chem., 1996,6(2), 247-252 247 14 12 73 10 .-E d 8 L2 6<3 4 2 0 2 4 6 8 1 Oi90 50150 9011 0 Ga:Cr added Fig. 1 A, Variation of the Ga :Cr ratio in the intercalated mixed oligomers as a function of the Ga :Cr ratio in solution. B, Overall uptake curve ofGa3+ and Cr3+. (1.8-2.6 times) is observed over the whole composition range. Initially, the uptake of Ga3+ and Cr3+ increases with the Ga :Cr ratio in solution until a plateau approximately between Ga: Cr ratios of 40 :60 and 80: 20 is reached.For the sample with a ratio Ga/Cr of 90 :10, uptake increases again. The empirical formulae of the intercalate compounds are listed in Table 1. From chemical analysis it is not possible to rationalise the formulae of the intercalated mixed oligomers, probably due to the presence of more than one species in all the intercalation compounds. The intercalated mixed oligomers contain a small proportion of acetate. The Ac- :Cr3+ ratio, which is initially 2.8 in solution, decreases slightly in the intercalates from 0.47 to 0.21 up to a Ga: Cr ratio of 70: 30, from which it increases up to 0.66. The IR spectra of the intercalates (Fig.2) confirm the presence of acetate in the mixed oligomers, by the appearance of two bands at 1550 and 1450 cm-'assigned to v,,(CO,-) and v,r" (C0,-), respectively, with a difference (v,, -vsym= 100 cm- ) typical of bidentate Ac-ligands.14 In addition, the band corresponding to v3(PO4,-) is significantly modified after intercalation of the mixed oligomers; it is shifted to lower frequency and is split at 1139 and 1009 cm-' as a consequence of the strong interaction of the phosphate layer with the oligomers. After calcination at 400°C the acetate bands disappear and only a broad band of v3(PO4,-) is now observed at 1040 cm-'. The electronic spectra (diffuse reflectance) of the intercalates (Fig. 3) show the two typical transitions of octahedral Cr3+, 4T2,+4A2e and 4T,,+4A2g (Table 2).Taking into account that the retention ratio Ac- :Cr3+ is close to 0.4, the almost constant v1 and v2 values confirm that Ac- ions are preferen- tially coordinated to this cation. On the other hand, the Al :A, ratio is always close to 1.1, which precludes the formation of clusters of chromium with high connecti~ity.'~ The TG-DTA curves of the intercalates are shown in Fig. 4. I I I I I 4000 3200 2000 1400 800 200 Wavenumber/cm-l Fig. 2 IR spectra of: (a) a-ZrP; (b) 50 :50 sample at 25 "C and (c) 50 :50 sample at 400 "C Samples with Ga:Cr ratios between 10:90 and 70: 30 exhibit similar thermal behaviour, i.e. the mass loss occurs in three stages: the first corresponds to the removed of hydration water (100-120°C), the second to acetate combustion which is accompanied by a strong exothermic effect (300-330 "C), and Table 1 Chemical composition and empirical formulae of mixed oligomeric Ga/Cr a-zirconium phosphate intercalates Ga :Cr Ga :Cr Ac :Cr added incorporated incorporated empirical formulae" 10:90 0.13 0.47 20: 80 0.32 0.41 30: 70 0.55 0.41 40: 60 0.74 0.37 50: 50 1.28 0.43 60 :40 1.85 0.2 1 70: 30 2.69 0.30 80 :20 4.96 0.57 90: 10 12.8 0.66 " Superficial XPS analysis indicated PO4 :Zr k:2 :1 in all samples.248 J. Muter. Chem., 1996, 6(2), 247-252 I 1 I I 400 600 800 hlnm Fig. 3 Diffuse reflectance UV-VIS spectra of the 50 :50 intercalate Table 2 Electronic spectra (diffuse reflectance) sample Ga :Cr added uJnm u,/nm At :A2" B/cm - 10:90 591 424 1.10 663 20 :80 591 424 1.09 663 30 :70 593 424 1.11 669 40 :60 587 424 1.01 646 50: 50 589 425 1.14 649 60 :40 591 424 1.17 663 70: 30 590 425 1.24 648 80 :20 601 431 1.18 653 90: 10 601 431 1.06 653 " Relative intensity. the third, between 350-700 "C,is attributed to dehydroxylation of the mixed oligomers to form interlayer mixed oxides.Samples with the highest gallium contents also show a mass loss at different stages; these DTA curves are more complicated, presenting several endothermic and exothermic effects in the range 300-550 "C. The endothermic effects at 360 and 415 "C and the exotherm at 525°C are characteristic of gallium oxyhydroxides, which in this temperature range are trans- formed to the corresponding gallium oxide.16 Samples with Ga :Cr ratios lower than 80:20 are amorphous at room temperature, which may indicate the presence of more than one single species in the interlayer regiqn.Only the 10 :90 sample presents a diffraction peak at 21 A. This solid also becomes amorphous upon heating at 300 "C. After calcination at 600 "C, these compounds still remain amorphous (Fig. 5). When the intercalated materials are calcine$ at 1000 "C in air, XRD reflections appear at 4.43 and 4.12 A, characteristic of zirconivm pyrophosphate, and a series of peaks at 3.63,2.67 and 2.49 A corresponding to mixed a-Ga203, a-Cr203 oxides are also seen (Table 3).On the other hand, the 80:20 and 90: 10 samples show XRD patterns at room temperFture with a set of peaks at 4.91, 4.114, 3.317, 2.644 and 2.403 A of low intensity, typical of CI-GaOOH, which indicates that small amounts of this substance are precipitated on the external surface of the intercalated phosphates. At 400°C the formation of amorphous Ga20, from a-GaOOH make these samples completely amorphous, even when they are calcined in air (Fig. 5). Since all the other materials are amorphous, in order to corroborate that no precipitation of hydroxides takes place during the intercalation process of oligomers into a-ZrP, three mixed hydroxides have been prepared with Ga :Cr mole ratios A-TI% Fig.4 Thermal analysis of the mixed intercalates: (a) 10:90; (b) 20 :80; (c) 30:70; (d) 40:60; (e) 50:50; (f) 60:40; (g) 70:30; (h) 80:20; and (i) 90: 10 Ib Fig. 5 XRD powder patterns of 10:90 samples: (a) room temperature; (b) 300°C; (c) 1OOO"C; and 90: 10 samples: (d) room temperature; (e) 400°C; and (f) 1000°C J. Muter. Chem., 1996, 6(2), 247-252 249 Table 3 XRD reflection data of mixed Ga/Cr oxide pillared materials, synthetic mixed Ga/Cr oxides and reference compounds d-spacing/A a-Ga203" 3.630 2.651 cr-Cr,O,b 3.631 2.665 pillared mixed oxides ( 1000"C) Ga: Cr = 10: 90 4.431 4.1 12 50: 50 4.441 4.112 90: 10 4.1 12 synthetic mixed oxides (800 "C) Ga:Cr=10:90 3.636 2.666 50: 50 3.638 2.664 90: 10 3.628 2.651 a JCPDS file 6503.JCPDS file 38 1479 (synthetic eskolaite). of 10:90, 50: 50 and 90: 10 under the same experimental conditions, i.e., maintaining an Ac- : Cr3+ mole ratio of 2.8 and adding n-propylamine in excess up to total precipitation of hydroxides. In Fig. 6 it is observed that at room temperature the 10: 90 precipitated hydroxide is amorphous, but after calcining at 600°C under N,, or at 400°C in air, the reflections of solid-solution Ga203 : Cr,03 powders can be clearly seen. This result is in contrast with the pillared 10: 90 material, which remains amorphous even at 800 "C.The 50: 50 pillared material has a similar behaviour. The 90 : 10 hydroxide precipitate at room temperature shows peaks typical of a-GaOOH and is already crystalline at 600 "C (under N, or air) while the pillared material with this nominal ratio is amorphous at this temperature.Table 3 compiles the reflections of the hydroxide precipitates, pillared materials and those of a-Ga203 and a-Cr203 as references. All these data indicate the different thermal behaviour of intercalates with respect to the equivalent precipitates, and confirm that the formation of intercalates of mixed acetatohydroxides oligomers takes place under the described experimental conditions and that only in the case of samples with higher amounts of Ga3+ (80:20 and 90: 10) does a little coprecipitation of a-GaOOH occur. Furthermore, we infer that mixed Ga203/Cr,03 pillared layered a-ZrP materials are formed between 400 and 600°C and that no segregation of the oxides occurs up to at least 800 "C.The SEM images of the pillared materials clearly show r-. --1 I Fig. 6 XRD patterns of the 10: 90 precipitates: (a) room temperature; (b) 400 "C; and 90 : 10 precipitates: (c) room temperature; (d) 400 "C; and (e) 600°C 2.491 2.177 1.184 1.665 2.479 2.175 1.186 1.672 3.633 2.672 2.488 3.643 2.671 2.490 2.671 2.489 2.484 2.177 1.817 2.489 2.179 1.819 2.489 2.175 1.814 a layered structure, in contrast to those of mixed oxide powders which exhibit a granular aspect (Fig. 7). XPS data for the mixed Ga/Cr oxide pillared materials (obtained by calcination of the intercalates at 400 "C) are given in Table4. The binding energies of P 2p and Zr 3d,,, are, in all cases, very similar to those in a-ZrP, suggesting that the +-I 1pm H10pm Fig.7 SEM images of the pillared 50: 50 material (a) and the 50: 50 precipitate (b).Magnification: 6500 (a), 800 (b). Table 4 Binding energies (eV) of mixed Ga/Cr oxide pillared zirconium phosphates Ga :Cr Cr 2P3/2 10:90 182.3 132.9 576.5 11 17.9 20 : 80 182.2 133.0 576.4 1117.8 30: 70 182.3 133.0 576.5 11 17.9 40 :60 182.3 133.0 576.4 11 17.8 50: 50 182.3 132.9 576.5 1 117.9 60 :40 182.2 133.0 576.6 1 117.9 70: 30 182.2 132.9 576.5 1118.0 80 : 20 182.3 132.9 576.7 1118.1 90: 10 182.2 132.9 576.6 1118.0 250 J. Mater. Chem., 1996, 6(2), 247-252 14 1 12 10 y O6 4 2 100 Ga:Cr added Fig.8 Variation of the bulk and surface Ga :Cr ratios with the Ga :Cr ratio added: (A) from chemical analysis (Ga :Cr, bulk); (0)from XPS (Ga :Cr, surface) layered structure is preserved after pillaring. Calcination of the solid was carried out under N, in order to avoid oxidation of Cr"' to Crvl, which is segregated from the interlayer region as CrO3.I7XPS data reveal that only Cr"' is present in the mixed oxide pillars. The binding energies of Cr 2p,,, are nearly constant for the whole range of compositions of the intercalated mixed oxides. Fig. 8 shows the plot of bulk and surface (XPS) Ga :Cr ratio versus the Ga :Cr ratio in solution. Up to a Ga: Cr ratio of 40: 60, the bulk and surface compositions are coincident, but at higher ratios the Ga: Cr ratio on the phosphate surface is lower than in the bulk materials, and this divergence increases with increasing Ga :Cr ratios in solution, indicating that 00 0 0.2 0.4 0.6 0.8 I zirconium phosphate preferentially adsorbs chromium on the external surface.Therefore, in addition to the mixed oligomers, other species enriched with gallium or chromium are also present in samples with high Ga: Cr ratios. This is especially true in the case of the materials with the highest Ga :Cr ratios (80:20 and 90: lo), in which the presence of gallium oxy-hydroxides is detected by XRD. Textural parameters for the mixed Ga/Cr oxide pillared materials are listed in Table 5. Up to a Ga :Cr ratio of 60 :40, the materials present high BET surface areas, between 193 and 297m2 g-I, and are mesoporous solids with a significant amount of micropores.For these samples, micropore volumes of ca. 0.1 cm3 g-' are found (Table 5). This microporosity is originated in the mixed oxide pillars formed inside the inter- layer region, which hold the phosphate layers apart, and therefore provides strong evidence for the formation of pillared structures. The pore-size distributions for these pillared mate- rials are very narroow, with most of the pores having radii in the range 7.5-1 1.5 A (Fig. 9). In contrast, samples with nominal Ga: Cr ratios >60: 40 present low BET surface areas and low micropore volumes (<0.05 cm3g-'). These samples have a high gallium content in the interlayer region, giving rise to the formation of closely packed structures, which impedes the access of the N, mol-ecules to the internal pores.The pore-size distributions in the mesopore regions are wider than in the case of the pillared materials with lower Ga :Cr ratios. This behaviour is similar to that observed in mixed A1:Cr oxide pillared zirconium phosphate materials." When the pillared materials (first calcined under N2) are subsequently calcined in air at 400 "C, the pore-size distribution is shifted towards a region of larger mesopores (Fig. 10).This 50 40 30 20 10 f 0.0 0.20 0.40 0.60 0.80 1.o PlPo 2.5 t 2 1.5 I 1 o2 Fig. 9 Adsorption-desorption isotherms and pore-size distributions of two pillared materials: 20 :80 sample, (a)and (c);SO :20 sample, (b)and (d) J.Muter. Chem., 1996, 6(2), 247-252 251 Tabie 5 Textural parameters for mixed Ga/Cr oxide pillared a-zirconium phosphates and mixed oxide powders Ga :Cr sEIm/mz g-' XSa/m2 g-' XVpa/cm3 g-' ~-micropores~/cm~g-la pillared materials 10:90 25 1 223 0.2366 0.101 20:80 297 273 0.1945 0.105 30: 70 265 298 0.2088 0.108 40: 60 257 232 0.1859 0.114 50:50 239 194 0.1177 0.092 60:40 193 170 0.1176 0.078 70 :30 99 84 0.065 0.046 80 :20 59 47 0.0467 0.026 90: 10 74 63 0.0679 0.033 mixed oxides 10:90 234 149 0.21 0.105 50: 50 184 116 0.06 0.079 90: 10 236 265 0.18 0.100 CS and XV' are the total area and volume, respectively, of all pores from the Cranston-Inkley method.Calculated from the Dubinin equation. " 0.0 50 1.0 Id 1.5 10' TplA Fig. 10 Pore-size distributions of: (a) pillared 20 :80 material; (b) sample (a) after calcination in air is attributed to a partial oxidation of Cr"' to CrV'; the CrO, formed is unstable in the interlayer region and is segregated outside, leading to more open structures.? Significantly, after this treatment the BET surface area is still >200 m2 g-'. By this simple method, it is possible to change the pore structure of these materials, which may prove useful for catalysis appli- cations. The mixed precipitates calcined at 400 "C exhibit high BET surface areas (184-236 m2 g-') and micropore volumes (0.08-0.1 cm3 g-') (Table 5).This is attributed to the presence of acetate coordinated to the metal ions, which acts as a template during the thermal transformation to the metal oxide. Therefore, the use of acetate solutions seems to be an interesting and useful method for the formation of mixed oxides with high surface areas. This research was supported by the CICYT (Spain) Projet MAT94-0678 and by the C.E. Programme BRITE-EURAM, contract BRE2-CT93-0450. The authors thank Dr. J. L. G. Fierro for acquiring the XPS data. t At T> 197 "C, CrO, loses O2 to give amorphous Cr203.'* During this decomposition process, the amorphous Cr20, is eliminated from the interlayer region ('segregation'), giving rise to collapsed structures, as verified already in the formation of chromia-pillared montmorillonite." References 1 Pillared Layered Structures, Current Trends and Applications, ed.I. V. Mitchell, Elsevier, London, 1990. 2 R. T. Yang and M. S. A. Baksh, AIChE J., 1991,37,5. 3 A. Dyer and T. Gallardo, Cation and anion exchange properties of pillared clays, in Recent Developments in Ion Exchange, ed. P. A. Williams and M. J. Hudson, Elsevier, London, 1990, p. 75. 4 Catalysis Today, Pillared Clays, ed. R. Burch, Elsevier, Amsterdam, 1998, p. 187. 5 L. Storaro, R. Ganzerla, M. Lenarda and R. Zanoni, J. Mol. Catal. A, 1995,97,139. 6 P. Maireles Torres, P. Olivera Pastor, E. Rodriguez Castellon, A. JimCnez Lopez and A. A. G. Tomlinson, J. Muter. Chem., 1991, 1,739. 7 M. Martinez Lara, E. M. Farfan Torres, J.Santamaria Gonzalez and A. JimCnez Lopez, J. Solid State Chem., 1994,73,189. 8 A. Guerrero Ruiz, I. Rodriguez Ramos, J. L. G. Fierro, A. JimCnez Lopez, P. Olivera Pastor and P. Maireles Torres, Appl. Catal. A., 1992,92, 81. 9 S. M. Bradley, R. A. Kydd and R. Yamdagni, J. Chem. SOC.,Dalton Trans., 1990,413. 10 A. Vieira Coelho and G. Poncelet, Appl. Catal., 1991, 77, 303; W. Y. Lee, R. H. Ray Thatha and B. J. Tatarchuck, J. Catal., 1989, 115,159;P. Olivera Pastor, J. Maza Rodriguez, P. Maireles Torres, E. Rodriguez Castellon and A. JimCnez Lopez, J. Mater. Chem., 1994,4, 179. 11 P. Olivera Pastor, J. Maza Rodriguez, A. JimCnez Ldpez, I. Rodriguez Ramos, A. Guerrero Ruiz and J. L. G. Fierro, in New Developments in Selective Oxidation II, ed. V. Cortb Corberan and Vic Bellon, Elsevier, Amsterdam, 1994, p. 103. 12 C. R. Bayense, A. J. H. P. Van der Pol and J. H. C. van Hooff, Appl. Catal., 1991,72, 81. 13 H. Benhamza, A. Bouhaous and J. Livage, J. Chim. Phys., 1991, 88, 1875. 14 K. Nakamoto, The Infrared Spectra of Inorganic and Coordination Compounds, Wiley Interscience, New York, 1963. 15 H. Stunzi, F. P. Rontzinger and W. Marty, Inorg. Chem., 1984, 23,2160. 16 R. Roy, V. G. Hill and E. F. Osborn, J. Am. Chem. Soc., 1952, 74,719. 17 A. JimCnez Lbpez, J. Maza Rodriguez, P. Olivera Pastor, P. Maireles Torres and E. Rodriguez Castellon, Clays Clay Miner., 1993,41,3,328. 18 F. A. Cotton and G. Wilkinson, Advanced Inorganic Chemistry, Wiley, New York, 5th edn., 1988, p. 683. Paper 5/03708E; Received 9th June, 1995 252 J. Muter. Chem., 1996,6(2), 247-252
ISSN:0959-9428
DOI:10.1039/JM9960600247
出版商:RSC
年代:1996
数据来源: RSC
|
|