摘要:

Rates of attachment of fibroblasts to self-assembled monolayers formed by the adsorption of alkylthiols onto gold surfaces Elaine Cooper,a Robin Wiggs,b David A. Hutt,a Lesley Parker,b Graham J. Leggett*a† and Terry L. Parkerb Departments of aMaterials Engineering and Materials Design and bHumanMorphology, T he University of Nottingham, University Park, Nottingham, UK NG7 2RD Self-assembled monolayers (SAMs) of alkylthiols on gold surfaces have been prepared and characterised by contact angle goniometry and X-ray photoelectron spectroscopy. Murine 3T3 fibroblasts were cultured on single-component monolayers of adsorbates with carboxylic acid and methyl termini, and the rates of cellular attachment measured over 90 minutes.Cell attachment was found to be approximately twice as rapid for carboxylic acid-terminated monolayers as for methyl-terminated monolayers.After 24 hours, substantially greater numbers of 3T3 cells were found to be attached to the hydrophilic monolayers than were attached to the hydrophobic ones. Murine 3T3 cells were also cultured on patterned SAMs which were prepared using UV photochemistry and characterised using scanning electron microscopy.The cells attached exclusively to the acidfunctionalised areas of the patterned surface, leaving the methyl-functionalised regions bare. The patterns also strongly influenced the morphology of the attached cells. The phenomena of cellular attachment and adhesion onto controls over cellular attachment and adhesion on solid surfaces, and to explore the potential utility of SAMs as templates solid surfaces are of importance both for fundamental studies for the fabrication of novel cellular devices and structures.of cell growth and function and for technological applications Prime and Whitesides31 have shown that self-assembled which utilise cellular structures, including the development of monolayers formed by the coadsorption of mixed monolayers novel prosthetic materials, nerve regeneration and bioelectronof v-terminated alkylthiols on gold may be employed as model ics and biosensors.Cell–surface interactions remain complex substrates in studies of protein adsorption. The feasibility of and poorly understood, largely because of the vast diversity of culturingcells on SAMs formed by the adsorption of alkylthiols processes and parameters which control them.Both the microon gold has also been demonstrated.13,14 Rat basophilic leu- topography and the surface chemistry of the substratum are kaemia cells were found to attach to SAMs with a variety of known to exert an influence over cell–surface interactions. different surface chemistries, includingmonolayers with tertiary Surface chemistry is known to influence cellular adhesion, but amine, carboxylic acid, methyl and perfluorinated tail groups, in a serum-containing medium, the substratum is likely to although the numbers of attached cells varied to some (un- become covered with adsorbed proteins prior to cell attachquantified) extent with surface chemistry.13 In contrast, SAMs ment.The influence of surface chemistry is thus convoluted by terminated with oligo(ethylene glycol) groups proved non- the structure of the adsorbed protein layer which forms upon adhesive to cells.MG63 osteosarcoma cells exhibited different exposure to the cell culture medium. Indirectly, cellular behaviour, and were observed to attach preferentially to adhesion is influenced via the conformational changes which hydroxy-terminated SAMs, but not to adjacent methyl-termin- occur in the proteins as they adsorb to the substratum and ated SAMs indicating a strong dependence on SAM surface which are determined by the surface chemistry of the energy.14 Predominantly, however, published studies have substratum.employed silane monolayers on silicon substrates. Stenger and The design of suitable experiments by which to investigate co-workers have studied neuronal attachment to amino and the processes of cellular attachment and adhesion is therefore perfluorinated alkylsilanes, and by utilising UV photochemis- a formidable scientific challenge, and no single experimental try have prepared patterned materials for the spatial definition system will enable all of the relevant parameters to be defined of cell growth.10,11 In a systematic study, Lewandowska et al.and examined. Rather, it is necessary to devise a specific studied cell and protein interactions with alkylsilanes with a experimental milieu in which the influence of particular param- variety of tail functionalities, utilising radiolabelling techniques eters may be explored. Recent examples include the utilisation to assay amounts of adsorbed protein, and examining actin of lithographically engineered micro-morphologies to examine stress fibre organisation by phase contrast microscopy.7,8 cellular guidance by ultrafine topography1–3 and the explo- In the present study the influence of surface chemistry on ration of specific chemistries, including polymeric films,4,5 the rate of attachment of substrate-dependent 3T3 murine adsorbed layers of cell-adhesive proteins6 and monolayers of fibroblasts has been explored.Single-component alkylthiol-on- adsorbed organic molecules,7–14 on solid supports. Recent gold SAMs have been prepared from adsorbates with methyl years have seen substantial interest in the utilisation of self- [octanethiol (OT)] and carboxylic acid [3-mercaptopropanoic assembled monolayers (SAMs), formed typically by the acid (MPA)] tail groups and rates of attachments of cells have chemisorption of alkylthiols onto gold surfaces15–17 or by the been investigated over 90 min.In addition, cell morphology chemisorption of alkylsilanes onto silica surfaces,18 to model for the 3T3s cultured on the SAMs with different surface a variety of interfacial phenomena including wetting,19–23 chemistries has been examined by optical microscopy.adhesion,24–26 molecular recognition27–30 and biological inter- Patterned monolayers of OT and MPA have also been created actions.7,8,12–14,31–33 We are interested in using SAMs as model using UV photochemistry and murine 3T3 fibroblasts have materials in systematic studies of cell–surface interactions.The been cultured on these monolayers. A particular advantage of objective of our research is to determine the surface chemical this photolithographic method of patterning SAMs is that possible topographical influences on cell attachment are avoided. † Email: Graham.Leggett@Nottingham.ac.uk J. Mater. Chem., 1997, 7(3), 435–441 435method (after Bain et al.17) on a Rame-Hart model 100-00 Experimental contact angle goniometer.A 2 ml droplet of water was sus- Preparation of SAMs pended from the tip of a microlitre syringe supported above the sample stage. The syringe tip was advanced towards the Two different glass substrates were used to prepare the mono- sample until the droplet made contact with the sample surface.layers. The single-component monolayers were prepared on The syringe was then retracted leaving the droplet on the 19 mm diameter glass coverslips which were cleaned by immer- sample. A sample was said to be wetting if the contact angle sion in 5 mol dm-3 nitric acid for 2 h followed by sonication was less than 10° and the drop perimeter was observed to be in distilled water and drying by heating at 70°C for several deformed.The patterned monolayers were characterised using hours. The staining racks used to support the coverslips were scanning electron microscopy (SEM). The SEM images were cleaned in a similarmanner. Rectangular (Chance 22×64 mm2, acquired in the secondary electron detection mode using a no. 2 thickness) cover slips were used for the patterned mono- JEOL JSM-6400 scanning microscope (chamber pressure layers.In this instance all glassware used in the sample 5×10-6 Torr), with a 35 kV primary electron beam and a preparation was cleaned by soaking in hot (ca. 90°C) ‘Piranha’ current of 3 nA. The electron detector was operated with a solution (mixture of 30% H2O2 solution and conc. H2SO4 in collection voltage of +300 V.a ratio of 357) then rinsed with copious amounts of distilled water and dried in an oven. Care was exercised in the use of Cell culture both nitric acid and ‘Piranha’ solution, which may, under certain circumstances, react violently with organic material. An anchorage-dependent murine fibroblastic cell line (3T3-L1) After they had been cleaned, the different glass substrates were was maintained in minimum essential medium supplemented treated in the same way.with 10% newborn calf serum, 1 mg ml-1 penicillin, 1 mg ml-1 Monolayers of MPA and OT were prepared on evaporated streptomycin, 0.2% NaHCO3 and 2 mmol dm-3 L-glutamine, gold (Goodfellow, 99.99+%) films supported on chromium and routinely subcultured by trypsinisation (0.25% trypsin in (Goodfellow, 99.99+%) primed glass substrates.The thickness EDTA in Ca2+/Mg2+-free phosphate-buffered saline). of the gold films was in the range 20–40 nm and the gold was Substrates were seeded with fibroblasts in 3 ml full culture medium (FCM) at a concentration of 105 cells ml-1, and deposited at a rate of ca. 0.5 A° s-1. Samples prepared within cultured in humidified air with 5% CO2 at 37°C and pH 7.2.this range of thickness were optically transparent and examin- For cell attachment assays, coverslips were dipped briefly in ation of cells by optical microscopy was not impeded by the Earle’s balanced salt solution (EBSS; pH 7.2–7.4), fixed in situ presence of the gold film. Following deposition of the gold with 5% acetic acid in ethanol, and counted by means of film, samples were immersed in 1 mmol dm-3 solutions of the systematic random sampling using a computer-based image thiols in degassed ethanol for 12–18 h.The time interval analysis system connected to a video microscope. between removal of the samples from the evaporator and their immersion in the thiol solution was typically less than 4 min. Following removal of samples from the adsorbate solution, Results they were rinsed with degassed ethanol and dried under a Characterisation of SAMs stream of nitrogen.OT (97%) was obtained from Fluka, and was used as received. MPA (99%) was obtained from Aldrich Single-component self-assembled monolayers of OT and MPA and also used as received. on gold were characterised by X-ray photoelectron spec- The patterned monolayers were created using a modified troscopy and contact angle goniometry.version of the photolithographic process of Tarlov and co- Fig. 1 shows the relevant regions of the XP spectra recorded workers,34,35 employing transmission electron microscope grids for the as-prepared SAMs of OT and MPA. With the exception (Agar) as masks to create micron-scale features on the surface.of the fitted C 1s region of the MPA spectrum, the spectra are UV irradiation of alkylthiol-on-gold SAMs in the presence of shown as recorded with no binding energy or background air results in the photo-oxidation of the alkylthiolate adsorbate corrections made. For the OT SAM the C 1s peak at 285 eV complex to yield an alkylsulfonate. The weakness of the is symmetrical with a full width at half maximum height SO3–Au interaction facilitates the displacement of the alkylsul- (FWHM) of ca. 1.4 eV, indicating a single environment for the fonate by a second thiol adsorbed from solution, resulting in C atoms in the SAM [Fig. 1(a)]. The O 1s/Au 4p3/2 region of the formation of a new SAM on the UV-exposed areas. In this the OT spectrum is shown in Fig. 1(b). The Au 4p3/2 peak at study several electron microscope grids were placed directly 546 eV can be seen, but there is no contribution to the spectrum on a freshly prepared MPA SAM which was then placed under from O indicating that the monolayers are oxygen free. There a medium-pressure mercury arc UV lamp for 60 min. The is a broad peak at 162–163 eV in the S 2p region of the grids were removed and the samples were then immersed in a spectrum which consists of two unresolved components, the freshly prepared 1 mmol dm-3 OT solution for 7 min after 2p3/2 and 2p1/2 [Fig. 1(c)]. which they were taken out of solution, rinsed with degassed For the MPA monolayer three peaks are observed in the C ethanol and dried with N2 gas. The resulting monolayers were 1s region of the XP spectrum [Fig. 1(a)]. The peak at 285 eV hydrophobic in the exposed regions while the masked regions is assigned to the methylene carbon atom attached to the remained hydrophilic. Grids with a variety of geometries and sulfur, the peak at 287 eV to the other methylenic carbon atom dimensions were used to investigatewhether feature dimensions and the peak at 289 eV to the carboxylic acid carbon atom.had a role to play in influencing the growth of cells on the The observed peak shifts of 2 and 4 eV, compared to the patterned surface. standard methylene peak at 285 eV, for the middle and car- The purity of the single component monolayers was assessed boxylic carbon atoms respectively, are similar to those reported by X-ray photoelectron spectroscopy (XPS) and the wettability in the literature for the different carbon environments.17 was measured by contact angle goniometry.XPS measure- However, the observed peak area ratios for the three different ments were performed using a VG ESCALab system equipped carbon atoms are unexpected; all three C 1s peaks for MPA with a twin anode X-ray source and a 100 mm radius hemi- should have the same area.Clearly this is not the case as the spherical electron energy analyser. The sampled area was peak assigned to the methylenic carbon attached to the sulfur ca. 9 mm in diameter and the take-off angle was 70°. Mg-Ka has a much larger area than the other two C 1s peaks which radiation was employed, with the sample held at maximum have approximately the expected peak area ratio of 151.The separation from the X-ray source in order to minimise damage. discrepancy may be explained by considering that the high surface energy of the acid-terminated SAM makes it very Advancing contact angles were measured using the sessile drop 436 J. Mater. Chem., 1997, 7(3), 435–441Fig. 2 Variation in the number of attached fibroblasts with time for MPA and OT monolayers Fig. 1 XP spectra of OT and MPA SAMs: (a) C 1s; (b) O 1s/Au 4p3/2; (c) S 2p Fig. 3 Variation in the number of attached fibroblasts after 24 h for susceptible to contamination from airborne molecules giving MPA and OT monolayers rise to the enhanced area of the peak at 285 eV. This enhanced hydrocarbon signal has been observed previously in the highresolution XP spectra of carboxylic acid polymers where it cells were observed.Fig. 4(a) shows a light micrograph of 3T3 cells attached to a region of an OT SAM adjoining an exposed was similarly attributed to adventitious hydrocarbon contamination resulting from the high surface energy of the polymers.36 region of the glass cover slip masked during evaporation of the gold film. On the glass surface, the cells exhibit a more The oxygen and sulfur regions of the XP spectra of MPA are shown in Fig. 1(b) and (c), respectively. even distribution across the surface and there appears to be substantial cytoplasmic spreading. The contrast with the mor- Advancing contact angles of <10° were recorded for the single-component MPA SAMs and angles of 109±2° were phology of attached cells on the OT monolayers is marked.Fig. 4(b) shows a light micrograph of an MPA monolayer recorded for the single-component OT SAMs. These values agree well with the published literature.17 following cell culture. The morphology of the attached cells is very similar to that observed for the glass substratum; cells are evenly distributed with no close-packed clusters. There is Cell culture on single-component monolayers also considerable cytoplasmic spreading.Rates of attachment of 3T3 murine fibroblasts for singlecomponent monolayers of both MPA and OT were measured Photopatterning over 90 min (Fig. 2). Initially, the number of cells attached to the COOH-terminated monolayer was found to rise more The patterned SAMs were characterised using SEM and Fig. 5 shows typical SEM images recorded for different patterned rapidly than the number attached to the CH3-terminated monolayer, with approximately twice as many cells attaching monolayers. Light contrast is observed in the OT regions of the surface and dark in the MPA regions.The dark contrast to the wetting surface after 20 min. Thereafter, the number of 3T3 cells attached to the MPA monolayer remained approxi- in the wetting regions is thought to arise from the higher energy hydrophilic surface spontaneously adsorbing airborne mately twice that attached to the OT monolayer. After 24 h there are four times as many cells attached to the MPA molecules which attenuate the secondary electron signal arising from the gold.37,38 The SEM images highlight the clear trans- monolayer as the OT monolayer (Fig. 3).The 3T3 cells thus exhibited a clear preference for the COOH-terminated mono- ition between hydrophobic and hydrophilic regions on the surface. Advancing water droplet contact angle measurements layer as compared to the CH3-terminated monolayer. Examination of cell morphology by light microscopy after were also performed to determine differences in wettability between photopatterned regions.The masked MPA areas of 24 h emphasised the extent of this preference. Not only were the numbers of cells greater for the wetting surface, but the the surface remained highly hydrophilic (ha<15°), while the refunctionalised OT areas gave a contact angle of ca. 100°. morphology was strongly indicative of successful adhesion and growth. Micrographs of OT monolayers following 24 h cell This slightly lower than expected OT contact angle (literature value 110°)17 is a result of incomplete ordering in the SAM, culture revealed clusters of closely packed cells.Predominantly, the cells were found to be spread poorly, even where single due to the necessarily short exposure time of the photo- J. Mater. Chem., 1997, 7(3), 435–441 437and 110 mm, the Robertson bar width is 15 mm with 85 mm wide slots and the bars of the grid are 11 mm wide. The overall diameter of each of the grids is 3 mm.The images observed demonstrate that the spatial separation of the MPA and OT regions corresponds to the dimensions of the TEM grids. Cell culture on patterned substrates Fig. 6 shows a series of light micrographs of a Sjostrand patterned substrate following 24 h of 3T3 cell culture.Comparing the light micrographs of the cells to the SEM images obtained for the same patterned surfaces [Fig. 5(a) and (b)], it is apparent that the cells enjoy a significant interaction with the hydrophilic areas of the surface, but not with the hydrophobic regions. Some cells have attached to the OT areas, but they are vastly reduced in number and also appear to be attached to cells anchored to the hydrophilic regions of the pattern.After 24 h cells start to extend in large numbers from the MPA regions onto the OT regions within the pattern due to lack of surface area available within the MPA areas. The pattern is masked after 36 h cell culture, i.e. the cells are confluent across the patterned region giving rise to a ‘disc’ of cells.However, cells on OT regions appear to be stabilised by cell–cell contacts with cells on the MPA areas. The patterned substrates demonstrate clearly that given the choice between Fig. 4 Light micrographs of fibroblasts attached to (a) an OT monolayer (left) and glass (right) and (b) an MPA monolayer Fig. 5 SEM images of different patterned monolayers: (a) and (b) ‘Sjostrand’ patterned substrate; (c) and (d) ‘Robertson’ patterned substrate; and (e) ‘400 grid’ patterned substrate oxidised surface to the OT solution to prevent replacement of the unoxidised MPA by octanethiol.Three different masks are shown in Fig. 5: the ‘Sjostrand’, Fig. 5(a) and (b), ‘Robertson’, Fig. 5(c) and (d), and ‘400 grid’, Fig. 5(e), TEM grids.The Fig. 6 Light micrographs of a Sjostrand patterned substrate following 24 h cell culture Sjostrand grid has two different bar (and slot) widths of 50 mm 438 J. Mater. Chem., 1997, 7(3), 435–441two surfaces there is a strong preference for the 3T3 cells to interact with the carboxylic acid chemistry. The patterned substrate can also influence strongly the morphology of the attached cells as demonstrated by the optical micrograph obtained following cell culture on patterns produced using the Robertson mask (Fig. 7). On this surface the cells conform precisely to the geometry of the patterned region and have only extended in the direction of the MPA lines on the surface, forming wire-like structures. The preference of the fibroblasts for the MPA-functionalised regions (as compared to the hydrophobic regions) has thus forced them to spread in a fashion not observed on unpatterned MPA surfaces.Cell culture on the 400 grid patterned substrate emphasises the influence that the pattern dimensions and geometry have on the morphology of the attached cells (Fig. 8). While the complete pattern has not been exposed as it was for the other two patterns, Fig. 8 shows that a grid pattern is beginning to form. Significantly, the cell nuclei are mainly attached to the point of intersection of the grid lines with cell extension occurring only in the direction of the bars. The intersection Fig. 8 Light micrographs of a ‘400 grid’ patterned substrate following 24 h cell culture presents the largest surface area for cell body attachment and the cell is imobilised at this point by tension in the four directions of the grid lines.Discussion 3T3 murine fibroblasts were cultured successfully on SAMs formed by the adsorption of acid- and methyl-functionalised alkylthiols on gold surfaces. In addition to attaching to the COOH-terminated monolayers in greater numbers, the cells also exhibiteda more well spreadmorphology on these surfaces.In contrast, on the CH3-terminated monolayer, the cells were poorly spread and clumped together. During the first 90 min, twice as many cells were attached to the MPA monolayers as were attached to the OT monolayers; however, after 24 h, four times as many cells were found to have attached to the COOHfunctionalised monolayers. It is possible that this difference is in part a consequence of mitotic cell division on the MPAfunctionalised regions and/or inhibition of cell division on the OT-functionalised regions.For the patterned monolayers the preference for the 3T3 cells to grow on the carboxylic acid-functionalised regions of the surface was very marked.This is attributed to migration of the cells to the preferred areas of Fig. 7 Light micrographs of a Robertson patterned substrate following 24 h cell culture the monolayer with time; an extracellular matrix (ECM) is J. Mater. Chem., 1997, 7(3), 435–441 439produced which enables the cells to migrate towards the MPA Our results indicate that the alkylthiol–gold system offers surface. Importantly, the materials also exhibited great stability important advantages over other systems through the ease over long exposures to biological media.with which clean, precisely defined surface chemistries may be The influence of surface energy (or wettability) on cell prepared. This system also provides an ideal basis for a study attachment has been studied by a number of workers.7,8,10,11,39 of the influence of various surface chemistries on cell attach- For example, Lewandowska and co-workers prepared mono- ment.Initial studies on other alkylthiol-on-gold systems sug- layers of v-functionalised alkylsilanes on glass, and compared gest that surface energy may not be the dominating factor in the attachment of 3T3 fibroblasts to monolayers with a variety determining cell adhesion. Other parameters may have an of tail functionalities, including CH3 and COOH groups.7,8 important role to play.For example, Sun et al. have demon- After 1 h, they observed little difference in numbers of fibro- strated using FTIR–external reflectance spectroscopy that blasts attached to COOH- and CH3-terminated monolayers. chain length is important in the interaction between self- This result is at variance with our observations.However, assembled monolayers containing carboxylic acid terminal Lewandowska and co-workers7,8 recorded an advancing con- groups and vapour-phase n-alkylamine molecules.44 The tact angle of 52° for the COOH-terminated monolayers com- difference is attributed to the degree of structural order in the pared to our value of <10° for MPA monolayers. Research in surface-confined acid group and resulting monolayer acidities.this laboratory using mixed monolayers of MPA and OT For example, mercaptoundecanoic acid displays significant suggests that substantial differences in cellular interactions hydrogen bonding within its COOH terminal groups whereas may be expected for materials that exhibit a difference in MPA has a disordered array of surface-confined COOH groups surface energy of this magnitude.40 The high contact angles making it more susceptible to reaction with incoming mol- reported by Lewandowska and co-workers7,8 may be indicative ecules. Consequently, the less strongly hydrogen-bonded acid of surface contamination; however, the lowest reported advan- tail functionalities in the short-chain monolayer may undergo cing contact angle for the acid-terminated alkylsilane-on-silica dissociation in aqueous media to yield the negatively charged system is ca. 30° (for permanganate periodate oxidised vinyl- carboxylate ion. Surface charge is thought to have a significant terminated monolayers),18 indicating surface energies signifi- influence on cellular attachment and acid dissociation may cantly lower than those attainable for v-functionalised alkylthi- therefore play a role in explaining the observed preference of ols on gold.The measured wettability depends on adsorbate 3T3 cells for MPA compared to OT. structure and ordering and also on the method used to prepare The present study suffers from one main weakness: no the terminal functional group. Wasserman et al. attributed the attempt has been made to regulate the composition of the raised contact angles they measured for acid-functionalised protein layer which adsorbs onto the SAM from the culture alkylsilane SAMs (when compared to acid-terminated alkyl- medium.A variety of proteins may be involved in regulating thiol SAMs on gold) to the presence of unreacted vinyl tail the cellular interactions with the SAM including, in particular, groups.18 The general difficulties associated with the introduc- vitronectin and fibronectin.The sample surface chemistry tion of polar tail functionalities to alkylsilane-on-silica SAMs influences the conformation of the adsorbed proteins, which is therefore mean that these monolayers do not represent an the critical factor in determining the cellular interactions of ideal basis for performing fundamental studies of cellular the SAM, and thus only indirectly influences cellular attach- interactions with polar surfaces.Formation of alkylthiol-on- ment. Systematic studies must attempt to examine the influence gold SAMs allows the preparation of pure, single-component of surface chemistry on protein adsorption phenomena as the polar surfaces and the preparation of mixed chemistries in a key to understanding cellular interactions.We are presently more precisely defined fashion, leading to a clearer understand- attempting to address this problem. ing of differential cell–substratum responses. Surface topography has also been used to direct cell growth. Clark and co-workers have examined the alignment of cells, on patterned fused quartz surfaces which were created using Conclusions microelectronic fabrication techniques.3,41 The results indicated that surface topography can be used to direct cell growth, but The rates of attachment of 3T3 fibroblasts to self-assembled overall cellular response is strongly dependent on cell type, monolayers formed by the adsorption of carboxylic acid- and cell–cell interactions and the geometry of the patterns.The methyl-terminated alkylthiols on gold have been studied. complicated fabrication procedure involved in creating these Approximately twice as many cells were found to attach to surfaces is a significant drawback to this approach to cell the acid-terminated monolayer than attached to the methyl- guidance.A study of growth cone guidance and neuron terminated monolayer. The fibroblasts exhibited a well spread morphology on micropatterned laminin surfaces also required morphology on the acid-terminated monolayer but were the use of complicated microfabrication procedures.6 rounded and clustered together on the methyl-terminated Alkylthiol-on-gold SAM surfaces have been used previously monolayer.Differences were noted between our results and to direct the growth of cells on a solid substratum,42 but, in those of other authors who employed alkylsilane-on-silica contrast to the present study, previous studies have utilised SAMs, because of the attainment of higher surface energies for long-chain and complex alkanethiols, including oligo(ethylene acid-terminated alkylthiol-on-gold SAMs.The importance of glycol)-terminated alkylthiols. This study demonstrates that preparing well characterised model surfaces in studies of cell– short-chain and readily available alkylthiols can be used in surface interactions is emphasised. Patterned monolayers have directing cell growth. A particular advantage is that the short- also been produced and used to direct cell growth by creating chain thiols require shorter processing times to create the spatially defined regions functionalised with MPA and sur- patterns by photolithography.Work in this laboratory has rounded by OT. Fibroblasts attached almost exclusively to highlighted that SAMs formed from short-chain adsorbates on regions functionalised by MPA. The morphologies of attached gold are rapidly photo-oxidised while SAMs formed from cells were found to be defined precisely by the geometry of the adsorbates HS(CH2)n-1CH3 (n>10) are photo-oxidised more MPA-functionalised regions. slowly.43 However, shorter chain thiols exhibit poorer order in the monolayers and it was not known whether using shortchain adsorbates to create the patterned substrates would The authors are grateful to the Royal Society, the Leverhulme result in adequate spatial definition for selectivity of biological Trust (grant F114AY) and the EPSRC (grant GR/K28671) for response.The results of the present study clearly indicate that their financial support, and to Jennifer C. Bussey for assistance adequate differences in surface chemistry/cell response can be achieved using short-chain thiol compounds.with the SEM experiments. 440 J. Mater. Chem., 1997, 7(3), 435–44122 H. A. Biebuyck and G. M. Whitesides, L angmuir, 1994, 10, 4581. References 23 D. A. Offord and J. H. Griffin, L angmuir, 1993, 9, 3015. 1 D. M. Brunette, Exp. Cell. Res., 1986, 164, 11. 24 M. J. Tarlov, L angmuir, 1992, 8, 80. 2 J. Meyle, A. F. von Recum, B. Gibbesch, W.Hutteman, 25 D. R. Jung and A. W. Czanderna, Crit. Rev. Solid StateMater. Sci., U. Schlagenhauf and W. Schulte, J. Appl. Biomater., 1991, 2, 273. 1994, 19, 1. 3 P. Clark, P. Connolly, A. S. G. Curtis, J. A. T. Dow and 26 D. J. Dunaway and R. L. McCarley, L angmuir, 1994, 10, 3598. C. D. W. Wilkinson, J. Cell. Sci., 1991, 99, 73. 27 T. A. Jones, G. P. Perez, B. J.Johnson and R. M. Crooks, 4 T. Matsuda and T. Sugawara, J. Biomed. Mater. Res., 1995,29, 749. L angmuir, 1995, 11, 1318. 5 J. P. Ranieri, R. Bellamkonda, J. Jacob, T. G. Vargo, J. A. Gardella 28 O. Chailapakul and R. M. Crooks, L angmuir, 1995, 11, 1329. and P. Aebischer, J. Biomed.Mater. Res., 1993, 27, 917. 29 K. D. Schierbaum, T. Weiss, E. U. Thoden van Velzen, 6 P. Clark, S. Britland and P.Connolly, J. Cell Sci., 1993, 105, 203. J. F. J. Engbersen, D. N. Reinhoudt and W. Gopel, Science, 1994, 7 K. Lewandowska, N. Balachander, C. N. Sukenik and L. A. Culp, 265, 1413. J. Cell. Physiol., 1989, 141, 334. 30 R. Reiter, H. Motschmann and W. Knoll, L angmuir, 1993, 9, 2430. 8 K. Lewandowska, E. Pergament, C. N. Sukenik and L. A. Culp, 31 K. L. Prime and G. M.Whitesides, Science, 1991, 252, 1164. J. Biomed. Mater. Res., 1992, 26, 1343. 32 P. Tengvall, M.Lestelius, B. Liedberg and I. Lundstrom, L angmuir, 9 S. Britland, P. Clark, P. Connolly and G. Moores, Exp. Cell. Res., 1992, 8, 1236. 1992, 198, 124. 33 M. Lestelius, B. Liedberg, I. Lundstromand P. Tengvall, J. Biomed. 10 D. A. Stenger, J. H. Georger, C. S. Dulcey, J. J. Hickman, Mater.Res., 1994, 28, 871. A. S. Rudolph, T. B. Nielsen, S. M. McCort and J. M. Clavert, 34 M. J. Tarlov, D. R. F. Burgess and G. Gillen, J. Am. Chem. Soc., J. Am. Chem. Soc., 1992, 114, 8435. 1993, 115, 5305. 11 J. J. Hickman, S. K. Bhatia, J. N. Quong, P. Schoen, D. A. Stenger, 35 G. Gillen, J. Bennett, M. J. Tarlov and D. R. F. Burgess, Anal. C. J. Pike and C. W. Cotman, J. Vac.Sci. T echnol. A, 1994, 12, 607. Chem., 1994, 66, 2170. 12 S. Margel, E. A. Vogler, L. Firment, T. Watt, S. Haynie and 36 M. R. Alexander, P. V. Wright and B. D. Ratner, Surf. Interface D. Y. Sogah, J. Biomed.Mater. Res., 1993, 27, 1463. Anal., 1996, 24, 217. 13 G. P. Lopez, M. W. Albers, S. L. Schreiber, R. Carroll, E. Peralta 37 E. W.Wollman, C. D. Frisbie and M. S.Wrighton, L angmuir, 1995, and G. M. Whitesides, J. Am. Chem. Soc., 1993, 115, 5877. 9, 1517. 14 P. A. DiMilla, J. P. Folkers, H. A. Biebuyck, R. Harter, G. P. Lopez 38 G. P. Lopez, H. A. Biebuyck and G. M. Whitesides, L angmuir, and G. M. Whitesides, J. Am. Chem. Soc., 1994, 116, 2225. 1993, 9, 1513. 15 M. D. Porter, T. B. Bright, D. L. Allara and C. E. D. Chidsey, 39 K. E. Healy, C. H. Thomas, A. Rezania, J. E. Kim, P. J. McKeown, J. Am. Chem. Soc., 1987, 109, 3559. B. Lom and P. E. Hockberger, Biomaterials, 1996, 17, 195. 16 R. G. Nuzzo, B. R. Zegarski and L. H. Dubois, J. Am. Chem. Soc., 40 E. Cussen, R. Wiggs, D. A. Hutt, G. J. Leggett and T. L. Parker, 1987, 109, 733. unpublished results. 17 C. D. Bain, E. B. Troughton, Y-T. Tao, J. Evall, G. M. Whitesides 41 P. Clark, P. Connolly and G. R. Moores, J. Cell Sci., 1992, 103, 287. and R. G. Nuzzo, J. Am. Chem. Soc., 1989, 111, 321. 42 R. Singhvi, A. Kumar, G. P. Lopez, G. N. Stephanopoulos, 18 S. R. Wasserman, Y-T. Tao and G. M. Whitesides, L angmuir, 1989, D. I. C. Wang, G. M. Whitesides and D. E. Ingber, Science, 1994, 5, 1074. 264, 696. 19 C. D. Bain and G. M. Whitesides, J. Am. Chem. Soc., 1989, 111, 43 D. A. Hutt and G. J. Leggett, J. Phys. Chem., 1996, 100, 6657. 7164. 20 P. E. Laibinis, R. G. Nuzzo and G. M. Whitesides, J. Phys. Chem., 44 L. Sun, R. M. Crooks and A. J. Ricco, L angmuir, 1993, 9, 1775. 1992, 96, 5097. 21 J. P. Folkers, P. E. Laibinis, G. M. Whitesides and J. Deutch, J. Phys. Chem., 1994, 98, 563. Paper 6/07204F; Received 22nd October, 1996 J. Mater. Chem., 1997, 7(3), 435–441 441

ISSN:0959-9428    

DOI:10.1039/a607204f

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

Self-assembly of aluminium-pillared clay on a gold support Pegie Cool,*a Abraham Clearfield,b Vimala Mariagnanam,b Laurel J. Mc. Ellistrem,b Richard M. Crooksb and Etienne F. Vansanta aL aboratory of Inorganic Chemistry, Department of Chemistry, University of Antwerp, Universiteitsplein 1, 2610 Wilrijk, Belgium bDepartment of Chemistry, T exas A&M University, College Station, T exas 77843-32565, USA Multilayer films of self-assembled aluminium Keggin ion pillared laponite and saponite have been grown layer-by-layer on a gold support. 4-Aminothiophenol (4-ATP) was used for anchoring the first clay layer on the gold surface. Subsequently layers of clay and Al pillar were adsorbed by dipping the substrate into the appropriate solutions. Characterization of the films was performed by X-ray photoelectron spectroscopy (XPS), IR spectroscopy (FTIR–ERS), ellipsometric thickness measurements and X-ray diffraction (XRD).XPS elemental analysis showed that both clay and pillar were present on the gold substrate. From the atomic film compositions, the charges on the Keggin ions were determined to be +3 and+5 for laponite and saponite, respectively. Ellipsometry and IR spectroscopy indicated gradual and regular growth for both films.Structural order in the synthesized films was demonstrated by X-ray diffraction. The interlayer free spacing (IFS) was found to be 6 A° by XRD, which corresponds to the increase in ellipsometric film thickness after each aluminium layer adsorption. Aluminium-pillared laponite films have been used as chemically sensitive films on surface acoustic wave (SAW) devices to measure the adsorption capacity of six volatile organic compounds (VOCs).The influences of the different terminal film layers and calcination-induced chemical changes on the extent of adsorption of the VOCs were investigated. Pillared clays (PILCs) are a well known class of porous on a solid support (e.g. Si or Au).Laponite clay, a-zirconium materials useful for catalytic and adsorption applications.1–3 phosphate and layered metal-oxide semiconductors (LMOS) In this study multilayer films of aluminium-pillared laponite such as K2Nb6O172- have been used as negative layers in and saponite were grown on a gold wafer by means of the previous studies.9–11 Polymeric or oligomeric cations utilised alternate adsorption of clay and pillar on the support.The as positive species are bonded electrostatically to the negative first step involved the reaction of a coupling agent between layers. The ionic attraction between different layers forms the the first clay layer and the gold support. The gradual and basis and stability of the film growth. The grown films can regular growth of the films was evidenced by ellipsometry and potentially be used as optical elements or sensors.12 FTIR–external reflectance spectroscopy (FTIR–ERS).X-Ray diffraction was utilised to show the structural order of the Experimental grown films. Atomic concentrations as determined by XPS allow the calculation of the positive charge on the Al13 Keggin Materials and material preparation ions in the multilayer films.Gas adsorption experiments per- The laponite RD was supplied by Laporte Inorganics. The formed using surface acoustic wave (SAW) mass transducers idealised unit cell formula is: (Na0.5 nH2O)(Mg5.5Li0.5)- clarify the gas adsorption properties of the films as a function Si8(OH)4O20 (MW=761) with a cation-exchange capacity of their top layer, thickness and calcination step.(CEC) of 0.733 mequiv. g-1.13 Following conventional techniques, PILCs are synthesized Calcium saponite was obtained from the Source Clay by an exchange reaction of a pillaring solution containing large oligomeric hydroxy–metal cations (e.g. Al,4 Zr,5 Ti6) with Minerals Repository of the Clay Minerals Society, Missouri, a smectite clay. The charge-compensating cations between the USA.The fraction <2 mm was obtained by wet sedimentation. clay sheets are replaced by the oligomers. Upon calcination The clay was converted into the Na form by treatment with the polyoxycations are converted to the oxide form, creating 1 mol dm-3 NaCl (3×), followed by washing the mixture until a two-dimensional porous system. However, the porosity of it was free of chlorine.The unit cell formula is as follows: these PILCs is still not what was expected because of the Na0.88Mg6(Si7.1Al0.72Fe0.14)O20(OH)4 (MW=780) with a CEC incomplete swelling of clay layers resulting in the interstratifi- of 1.13 mequiv. g-1. cation of pillared and unpillared layers. Therefore, some repro- Gold substrates were obtained by thermal evaporation of ducibility problems are often encountered.Theoretical studies 2000 A° Au over a 50 A° Ti adhesion layer on Si(100) wafers. based on geometrical models indicate aggregates of only fifteen After dicing the substrate to 2.5×1.3 cm2 pieces, the substrates and six parallel layers for Al-intercalated montmorillonite and were cleaned in an argon ion plasma (Harrick, model PDClaponite, respectively.7,8 Using these conventional methods the 32G) for 1.5 min and rinsed in ethanol.Al-pillaring of laponite results in an X-ray amorphous material, 4-Aminothiophenol (4-ATP) was obtained from Aldrich and owing to non-parallel or random stacking of the small clay purified by vacuum sublimation prior to use. plates. Also for the larger sized clays, like saponite, the nonparallel stacking of plates is still important and results in poor Characterization techniques crystallinity.A Seifert-Scintag PADII Powder diffractometer (Cu-Ka radi- In this study a self-assembly method has been utilised in ation; 40 kV; 30 mA) was used to determine the order in the order to improve the pillaring of clays. More controllable grown multilayer on the Au wafer.pillaring can be achieved by a layer-by-layer deposition of clay FTIR–ERS measurements were made using a Digilab FTS- and pillar. By means of the alternate adsorption of oppositely charged species, it becomes possible to grow multilayer films 40 spectrometer equipped with a Harrick Scientific Seagull J. Mater. Chem., 1997, 7(3), 443–448 443reflection accessory and a liquid N2-cooled MCT detector. An ATP in absolute ethanol for 24 h.The wafer was washed with EtOH and blown dry with N2. Subsequently (step 2), the wafer Au wafer immersed in a KI solution (2 mmol dm-3 in EtOH) for 10 min was used as the blank. All spectra are the sum of was dipped for 15 min in a 0.2% clay suspension of laponite or saponite. Prior to use, the suspensions were sonicated 256 individual scans (resolution 2 cm-1) using p-polarized light at an 84° angle of incidence with respect to the Au overnight in a water-bath in order to obtain complete exfoliation of the clay.In the case of the saponite clay some substrate. All spectra were baseline corrected. Multilayer thickness measurements were made using a flocculation was still observed. Therefore the centrifugate remaining after 2 min of centrifugation was used for layered- Gaertner Scientific ellipsometer (model L116C).Data were collected using the 633 nm He–Ne laser line at a 70° incident film preparation. For laponite, complete exfoliation of the clay was achieved by sonication and suspensions prepared in this angle, and assuming a refractive index of 1.46. XP spectra were acquired using a Perkin-Elmer (PHI) way were used after acidification to a pH of 4 using 1 mol dm-3 HCl.At this pH, 4-ATP becomes protonated and attracts Model 5500 spectrometer. In a typical XPS data acquisition, a pass energy of 29.3 eV, a step increment of 0.125 eV and a the negative clay plates. After reaction, the wafer was washed with deionised water and blown dry with N2.In a following Mg anode power of 400 W were employed. Atomic compositions were calculated from peak areas using sensitivity factors reaction (step 3), the wafer was dipped for 15 min in a 0.02 mol dm-3 Al pillar solution, followed by washing with deionised provided in the software of the instrument. Solution NMR spectra for 27Al were recorded on a Varian water and drying with N2.The Al13 pillar solution was prepared by dissolving 12 g of AlCl3 6H2O (0.05 mol) in XL-200 with a BB80S3 probe at 52.1 MHz. Chemical shifts were obtained in ppm referenced to external Al(H2O)63+ in 200 ml deionised water. To this solution 250 ml of 0.5 mol dm-3 NaOH (0.125 mol, molar ratio OH/Al=2.5) was added D2O as derived from 0.1 mol dm-3 aqueous AlCl3 solution. SAW device measurements were made at 25°C using two dropwise under vigorous stirring while heating at 80°C for 3 h.After cooling, the solution was diluted and adjusted to a (98 MHz) ST-cut quartz oscillators housed in a custom-made flow system.14 Modified SAW devices were dosed with vapour- pH of 5.0. A 27Al NMR spectrum showed a single peak at d phase probe molecules mixed in N2 at 25% of saturation 63.5 indicating the presence of the Al13 Keggin ion.In order (flow-rate=0.5 dm3 min-1). The change in SAW device fre- to deposit the next clay layer (step 4), 0.2% clay suspensions quency (Df ) due to the adsorption of vapour-phase molecules adjusted to pH 6 with 1 mol dm-3 HCl were employed. The is related to the mass loading per unit area (ma) through the immersion time of the wafer with the clay suspension was equation Df/f0=-kcmf 0ma, where, f 0 is the SAW resonance limited to 15 min, following which the wafer was washed and frequency (98 MHz), k is the fraction of the distance between dried with N2.The last two steps (3 and 4) were repeated to the centres of the transducers covered by the Au film (0.65) grow multilayer films on the Au substrate by the alternate and cm is the mass sensitivity coefficient of the device adsorption of oppositely charged species.(1.33 cm2 g-1 MHz-1 for ST-cut quartz).15,16 Procedures Results and Discussion The procedure used to prepare the layered clay materials is X-Ray photoelectron spectroscopy represented in Scheme 1. Each cycle consists of two different The atomic compositions of the substrate (Au+ATP) and the steps, in which clay and pillar are adsorbed successively. grown multilayers of Al-pillared laponite and saponite after 30 Initially (step 1), the Au wafer is reacted with a coupling agent, cycles were determined by XPS analysis and the results are 4-aminothiophenol (4-ATP).The reaction was carried out by immersing the gold wafer in a 1 mmol dm-3 solution of 4- presented in Table 1.Scheme 1 444 J. Mater. Chem., 1997, 7(3), 443–448Table 1 Atomic compositions (%) as determined by XPS analysis and laponite are structurally identical and have a smaller charge density on their layers, compared to saponite. element ATPa laponiteb saponitec IR spectroscopy Au 36.55 0.19 0.19 C 55.19 21.49 20.95 The growth of the film at initial stages of the cycling process Si 10.81 9.79 has been investigated by FTIR–ERS measurements.O 52.40 52.13 Spectrum A in Fig. 1 shows a 4-ATP monolayer confined to Mg 6.43 7.01 the surface through a strong interaction between the Au and Al 8.58 9.82 Na 0.10 0.11 the sulfur atom of the thiol.20 N 4.19 One can distinguish the peaks of the MNH2 deformation S 4.07 (1625 cm-1), the aromatic CNC stretch (1589, 1491 cm-1) and the CMN stretching (1292 cm-1).During the first step of aATP monolayer on gold. bAl-pillared laponite film on gold after 30 the cycling process a pH of 4 for the clay suspensions was cycles. cAl-pillared saponite film on gold after 30 cycles. used in order to protonate the amino groups. In this way a positive layer charge is created on the Au substrate, which can serve to attract the anionic clay plates.In spectrum B it is After the cycling process, the atomic compositions of Au seen that the attachment of a laponite layer on the organic and C detected on the surface decrease significantly owing to ATP monolayer results in the disappearance of the MNH2 the presence of a thick clay multilayer on top of the substrate.peak, together with the appearance of the silicate peak (SiMO The elements present in the saponite silicate layers (Si, O, Mg, vibration at 1071 cm-1). Al) were detected. In the case of laponite,atomic concentrations The growth of the silicate peak at 1071 cm-1 for the first of all elements (Si, O, Mg) except Li are determined. The eight cycles is clear for both laponite [Fig. 2(a)] and saponite reason is the low concentration and the low XPS sensitivity [Fig. 2(b)]. The gradual increase in the peak height, which factor for Li. The presence of very small amounts of Na ions correlates with the number of clay layers, is shown in the and large amounts of aluminium indicates that almost total insets. In the case of saponite it is almost linear. exchange of the Na+ for Al polyoxycations has been attained.Calculations can be performed knowing the structural formu- Ellipsometry lae, molecular masses (MW) and CECs of both clays. The theoretically obtained values for the Si/Mg and Si/Al ratios The ellipsometrically determined thicknesses of both the lapon- are given in Table 2 and are compared to the experimental ite and saponite films indicate gradual and regular growth for results obtained by XPS. For the Si/Al calculation, a total the first eight cycles (Fig. 3). The thickness of the ATP mono- exchange of the CEC for the [Al13O4(OH)24+x(H2O)12-x]n+ layer on Au was determined to be 7 A° and this value was (n=7-x; MW=1038.7) is assumed, while the charge n on the subtracted from the thicknesses shown in this plot. Each point Keggin ion varies (n=3–7).In the case of saponite, the is the result of three measurements obtained at different spots contribution of Al present in the clay sheets was taken into on the substrate, and they vary by a maximum of ±3 A° . This account. proves that the thicknesses of the multilayer films for both Good correlation between theory and experiment was clay types are very homogeneous.It becomes apparent that obtained for Al13 charges of +3 and +5 in the cases of the small plate size of laponite (300 A° ), compared to that of laponite and saponite, respectively. The ideal literature formula saponite (10000 A° ), has no adverse effect on the homogeneity for the Keggin ion requires a charge of +7; this value changes of the synthesized films. It is possible to calculate the amount upon hydrolysis of the ion.17,18 For Al13 intercalated in mont- of adsorbed clay and pillar per cycle by assuming complete morillonite, pillar charges varying from +2.33 to +4 are coverage of the substrate by the clay, the plate sizes given obtained by performing theoretical calculations on the pillared above and the dimensions of the substrate.The numbers of material.7 A charge of +3.15 rather than the formal value of clay sheets deposited on the substrate per cycle are then +7 has been reported by Jones and Purnell.19 For adsorption determined to be 3.61×1011 for laponite and 3.25×108 for of Al13 on saponite and hectorite, the average charges per Al saponite. If we further consider the number of plates per gram are +0.4 and +0.2, respectively, as described by Schoonheydt for laponite (4.28×1017) and saponite (3.78×1014),8 the above et al.4 These latter values correspond to Al135.2+ on saponite results indicate that 0.85 mg is adsorbed per cycle, irrespective and the lower charged Al132.6+ on hectorite, in close approximation to our data derived from XPS analysis (+5 and +3 respectively).A hydrolysis reaction on Al13 involves a sequence of deprotonation steps occurring in solution and after adsorption on the clay layers.A higher hydrolysis degree upon exchange, forming lower charged Keggin ions is more likely to occur on hectorite and laponite than on saponite; hectorite Table 2 Theoretical and experimental values as obtained by XPS for Si/Mg and Si/Al ratios for Al-pillared laponite and saponite theoretical experimental laponite Si/Mg 1.68 1.68 Al13n+, n=3; Si/Al 1.22 1.25 n=7; Si/Al 2.84 saponite Si/Mg 1.37 1.40 Al13n+, n=3; Si/Al 0.62 n=5; Si/Al 0.99 0.99 Fig. 1 FTIR–ER spectra of 4-aminothiophenol on Au (A) and after n=7; Si/Al 1.33 protonation and subsequent deposition of a laponite layer (B) J. Mater. Chem., 1997, 7(3), 443–448 445of the Keggin ion is stated as a prolate spheroid with long and short axes of 9.5 and 7 A° , respectively.21 The theoretical curve in Fig. 3 is calculated based on the clay layer thickness of 9.6 A° and the smallest dimension of the Keggin ion (7 A° axis). For both laponite and saponite the experimental curves closely approach the theoretical, and therefore we infer a flat orientation of Al13 between the clay sheets. More precisely, the mean increment in film thickness after each adsorption of Al pillars, as determined by ellipsometry, is ca. 6A° for both clays. X-Ray diffraction The ellipsometrically determined average interlayer free spacing (IFS) of 6 A° can be compared to that derived from the XRD pattern (Fig. 4). The first broad peak in both patterns at a small 2h value is due to the synthesized films on the substrate.All other diffractions at higher 2h originate from the Au substrate itself (as determined from control experiments) and are of minor importance here. XRD does confirm the existence of structural order in the eight-layer films, although the diffraction peaks are very broad. By applying the Scherrer equation it becomes possible to estimate the number (N) of coherently diffracting clay layers along the c axis.The equation expresses the correlation between the number of parallel layers, N, and the 001 peak width at half height as determined by XRD.17,22 N is found to be 5 for both the Al-pillared laponite and the saponite film after eight cycles. The d001 values vary between 14.5 and 16.5 A° for both clays. The average d001 value is 15.5 A° , which corresponds to an IFS of 6 A° after subtraction of the clay layer thickness. This indicates a small axis of only 6 A° for the adsorbed Al13 pillars.Literature data suggest that the dimension of the Keggin ion should be larger: IFS values of 7–10 A° have been obtained after Al pillaring.21,23,24 However, on the basis of theoretical calculations we conclude a cylindrical symmetry for the Keggin ion with a height of 9 A°and a diameter of 6.32 A° .7 This latter dimension is a better approximation to our experimentally observed spacing of 6 A° .Fig. 2 FTIR–ER spectra of (a) pillared laponite on Au after 1, 5 and The washing procedure is also very important and crucial to 8 deposition cycles, and (b) pillared saponite on Au after 2, 5 and 8 deposition cycles.The insets show the peak area vs. number of obtain the typical 18–19 A° d001 value conforming to the deposition cycles. dimension of the Keggin ion.4,13 The Keggin ions might not be the only intercalated Al species while washing causes chemical changes of the intercalated Al products. During the washing the pH is raised slightly, allowing polymerization and reorganization of Al species in the interlayer space.For Alhectorite and Al-saponite, d001 spacings of 14.6 and 13 A° , respectively, have been reported before washing. These values increased by 3.2 and 5.5 A°respectively after four separate washes.4 The short washing procedure as part of our cycling process accounts for the limited polymerization of Al13 on the clay.Fig. 3 Ellipsometric thickness as a function of the number of cycles for (a) laponite and (b) saponite; the calculated (theoretical) thickness is also shown (c) of the clay type. After a total exchange of the CEC for Al133+ on laponite, and Al135+ on saponite, we calculate loadings of 0.21 mg Al13 per cycle. Consistent with our calculations, Fig. 3 shows a nearly identical evolution of film thickness as a function of layer number for both clays. Each curve consists of two different slopes that alternate depending on whether the clay or pillar is adsorbed.In the first step of each cycle a laponite or saponite monolayer with a thickness of 9.6 A° adsorbs. As a following step in each cycle, Keggin ions are adsorbed onto the clay. Therefore, the increment in thickness Fig. 4 X-Ray diffractograms for pillared laponite (A) and pillared saponite (B) on Au after 8 deposition cycles must correspond to the Al13 dimensions. The original symmetry 446 J. Mater. Chem., 1997, 7(3), 443–448Fig. 5 Results of gas-dosing experiments performed on a four-layer pillared laponite film having either clay or Al pillar on top. Data were obtained using a mass-sensitive SAW device.Parts A and C represent the frequency shifts as a function of time for films terminating in clay and pillar, respectively. Part B gives the adsorption capacities in nmol cm-2 for the different gases. The order of gas dosing in parts A and C is the same as shown in B. Vapours were present at 25% of saturation. VOC adsorption To determine the Al-pillared laponite film porosity, adsorption measurements of six organic vapours were performed using surface acoustic wave (SAW) mass sensors.15,16 Prior to vapour adsorption all films were degassed at 120°C for 4 h in a vacuum oven.We determined the dependence of the extent of vapour adsorption as a function of the chemical composition of the top layer of the film, the film thickness and the calcination step.The SAWdevice frequency shifts as a function of time for adsorption of the six volatile organic compounds (VOCs) shown in Fig. 5B on Al-PILC films terminated in either laponite or pillar are shown in Fig. 5A and C, Fig. 6 Adsorption capacities of organic vapours on pillared laponite respectively. films after 1, 6 and 12 cycles with clay on top Negative frequency shifts correlate to an increase in adsorbed mass.It can be seen that most of the adsorptions are reversible internal surface area as well. Moreover, this penetration will processes since the frequency returns to baseline upon purging be easier for the thicker twelve-cycled film compared to the of the films with pure N2 (except for a temperature-induced well ordered thinner films because of a higher degree of drift, which is especially apparent in Fig. 5C). Only methanol disorder in the upper film layers. remains irreversibly bound to the film to an appreciable extent. The influence of a calcination step on VOC adsorption has This effect is more significant when the film is terminated in also been investigated. A pillared clay film after four cycles Al pillars (Fig. 5C). This result is consistent with chemical with Al pillars on top was heated at a rate of 5° min-1 to intuition that suggests that alcohols will interact strongly via 300°C and kept for 2 h at this temperature. The results of this hydrogen bonding with the hydroxy groups of the Al13 Keggin experiment are shown in Fig. 7. For the apolar molecules ions. In Fig. 5B the frequency shifts have been converted to (C7H16, C6H6 ,CCl4 ) an increase in mass loading upon calci- mass loading per unit area (in units of nmol cm-2) to remove nation is observed, while for the polar gases (CH2Cl2, CHCl3, the molecular mass bias.The clay-terminated film nearly CH3OH) the reverse occurs. We interpret this result in terms consistently adsorbs more VOC than the Al-terminated films.of chemical transformations that occur in the films during This effect is most apparent in the cases of polar gas molecules like CH2Cl2, CHCl3 and CH3OH. The enhanced adsorption of the polar Cl-containing molecules, CH2Cl2 and CHCl3, on the laponite layer can be explained by interaction with Na+ ions adsorbed to the film, imparting electrical neutrality to the system. The higher amount of adsorbed methanol in the case of a clay top layer is not only due to its polarity but also due to an induced swellingof the Al-pillared film on the Au support.The swelling of films upon methanol adsorption becomes clear when films with different thicknesses are used as substrates. Fig. 6 shows the amount of VOC adsorbed as a function of thickness for laponite-terminated films.The increasing level of CH3OH adsorption with the number of cycles can be explained by swelling effects of the pillared clay films. Adsorption takes place on the external surface area, and Fig. 7 Adsorption capacities of organic vapours on a pillared laponite film with the Al pillars as the top layer, before and after calcination penetration into the porous system allows adsorption on the J.Mater. Chem., 1997, 7(3), 443–448 447calcination; specifically, a change in charge on the Al13 pillars. P. C. acknowledges the NFWO/FNRS for financial support as research assistant and thanks the Fulbright Program for That is, during calcination, the Keggin ions are converted into the educational exchange between the University of Antwerp the Al2O3 form, decreasing the positive charge on the Al and Texas A&M University.A. C. wishes to acknowledge overlayer. A weaker electronic interaction of this layer with support of this study by the National Science Foundation the polar Cl-containing molecules leads to a decrease in through Grant No. DMR-9407899. R. M. C., V. M. and adsorption. There will be less H2O adsorbed onto the Al layer L.J. Mc. E. acknowledge support from the US National after calcination, resulting in a higher adsorption of the apolar Science Foundation (CHE-9313441). molecules onto the film. References Summary and Conclusion 1 T. J. Pinnavaia, Science, 1983, 220, 365. 2 Catal. T oday, Special Issue on Pillared Clays, 1988, vol. 2. This study shows that by a self-assembly method it becomes 3 A.Molinard and E. F. Vansant, Adsorption, 1995, 1, 49. possible to grow homogeneous films of Al-pillared laponite 4 R. A. Schoonheydt, H. Leeman, A. Scorpion, I. Lenotte and and saponite on a Au support. Protonated 4-aminothiophenol P. Grobet, Clays ClayMiner., 1994, 42, 518. serves as the anchoring molecule between the Au substrate 5 K. Ohtsuka, Y. Hayashi and M. Suda, Chem. Mater., 1993, 5, 1823.and the first clay layer. Cycling is performed by adsorbing clay 6 J. Sterte, Clays Clay Miner., 1986, 34, 658. 7 N. Maes, I. Heylen, P. Cool, M. De Bock, C. Vanhoof and and pillar layers successively during a 15 min contact time of E. F. Vansant, J. PorousMater., 1996, 3, 47. the respective solutions with the substrate. After sonication 8 P. Cool, N. Maes, I.Heylen, M. De Bock and E. F. Vansant, of the clay suspensions to give totally exfoliated solutions, clay J. PorousMater., 1996, 3, 157. monolayers can be adsorbed with each cyclisation step. The 9 E. R. Kleinfeld and G. S. Ferguson, Science, 1994, 265, 370. films grow layer-by-layer as evidenced by FTIR–ERS and 10 S. W. Keller, K. Hyuk-Nyun and T. E. Mallouk, J. Am. Chem. Soc., ellipsometry studies.XPS revealed the atomic compositions of 1994, 116, 8817. 11 G. S. Ferguson and E. R. Kleinfeld, Adv. Mater., 1995, 7, 414. the films. A comparison of the calculated and experimentally 12 G. Cao, H-G. Hong and T. E. Mallouk, Acc. Chem. Res., 1992, determined Si/Mg and Si/Al ratios allows us to estimate the 25, 420. Al13 pillar charge. For laponite the charge on the Keggin ion 13 R.A. Schoonheydt, J. Van Den Eynde, H. Tubbax, H. Leeman, is +3, while for saponite a charge of +5 results. These results M. Stuyckens, I. Lenotte and W. E. E. Stone, Clays Clay Miner., are in close agreement with literature data. The average for 1993, 41, 598. the interlayer free spacing of the Al-pillared laponite and 14 H. C. Yang, D. L. Dermody, C. Xu, A. J. Ricco and R. M. Crooks, L angmuir, 1996, 12, 726. saponite is determined to be 6 A° by XRD and corresponds to 15 R. C. Thomas, L. Sun, R. M. Crooks and A. J. Ricco, L angmuir, the increment in film thickness after each Al layer adsorption, 1991, 7, 620. as measured by ellipsometry. We interpret this result in terms 16 H. Wohltjen, Sens. Actuators, 1984, 5, 307. of a flat orientation of the spheroidal Al13 pillars between the 17 A. Molinard, PhD Thesis, University of Antwerp, 1994. layers. The short washing procedure is a possible explanation 18 D. E. W. Vaughan, Catal. T oday, 1988, 2, 187. 19 J. R. Jones and J. H. Purnell, Catal. L ett., 1993, 18, 137. for the reduced polymerization of Al13 on the clay layers, 20 R. G. Nuzzo and D. L. Allare, J. Am. Chem. Soc., 1983, 105, 4481. resulting in a small diameter of 6 A° . Finally, VOC adsorption 21 A. Clearfield and B. D. Roberts, Inorg. Chem., 1988, 27, 3237. data measured by SAW devices are consistent with the top 22 O. Braddell, R. C. Barklie and D.H. Doff, ClayMiner., 1990, 25,15. layer of the film and the film thickness. Calcination also 23 M. Tokarz and J. Shabtai, Clays Clay Miner., 1985, 33, 89. influences the adsorption capacity. In the case of methanol the 24 R. A. Schoonheydt and H. Leeman, ClayMiner., 1992, 27, 249. adsorption process becomes irreversible owing to swelling effects of the pillared clay. Paper 6/06129J; Received 5th September, 1996 448 J. Mater. Chem., 1997, 7(3), 443–448

ISSN:0959-9428    

DOI:10.1039/a606129j

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

Deposition of LaNiO3 thin films in an atomic layer epitaxy reactor Helene Seim,a† Heini Mo�lsa�,a Minna Nieminen,a Helmer Fjellva°g*b and Lauri Niinisto� a aHelsinki University of T echnology, L aboratory of Inorganic and Analytical Chemistry, FIN-02150 Espoo, Finland bUniversity of Oslo, Department of Chemistry, P.O. Box 1033 Blindern, N-0315 Oslo, Norway LaNiO3 thin films have been deposited in an atomic layer epitaxy (ALE) reactor, using La(thd)3, Ni(thd)2 and ozone as reactants, thereby proving the feasibility of the ALE technique to produce films of ternary oxides.Depositions were made on Corning glass in the temperature range 150–450 °C. The growth conditions were studied and the growth rate showed a linear dependence on the number of cycles. At 400 °C the growth rate was 0.24–0.26 A° per cycle.The growth rate of the LaNiO3 thin films was greatly influenced by the deposition temperature but in the temperature range 215–250 °C the growth saturated at 0.08 A° cycle-1 independent of the deposition temperature, thus indicating an ALE window. As-deposited thin films were amorphous but crystallized when heated at 600 °C. Simultaneously the colour of the films changed from yellow–brown to black.Possible reasons for the colour changes are discussed. Resistivity measurements showed that the crystalline thin films were metallic, r= (5–20)×10-6 V m. The amorphous thin films had resistivity values five orders of magnitude larger, r>3 V m. According to scanning electron microscopy (SEM) and atomic force microscopy (AFM), the films were homogeneous and dense.The surface roughness increased on crystallisation. X-Ray photoelectron spectroscopy (XPS) and magnetic susceptibility measurements were employed in order to further characterize the amorphous and crystalline thin films. Perovskite-type related oxides have found a variety of appli- In the case of laser deposition7,8 LaNiO3 was grown on LaAlO3, SrTiO3 and yttrium-stabilised zirconia (YSZ) sub- cations such as ferroelectrics, sensors, superconductors, electrodes and catalysts.1,2 Some of these oxides, e.g.PbTiO3 and strates. Depositions were carried out at substrate temperatures between 440 and 700 °C under an oxygen partial pressure of LaNiO3, are simple compounds from a chemical point of view, whereas the related high-temperature superconducting mate- (1.6–4.0)×10-4 bar.The method was successful in providing films with low resistivity, ca. 1.5×10-6 V m at 15 K. The films rials have a very complex chemistry with three or more different cations in the unit cell. Our interest is mainly focused obtained at 500 °C showed expected Pauli paramagnetic properties, whereas those obtained at 700°C showed a larger on LaNiO3, which is metallic, exhibits Pauli paramagnetic behaviour and has a slightly rhombohedrally distorted per- Curie–Weiss contribution to the magnetic susceptibility, thereby indicating the presence of local moments, probably ovskite-type structure. The metallic conductivity of LaNiO3 makes the material interesting for electrode applications.3,4 NiII species.8 In the spray-ICP technique at atmospheric pressure, the Quite often, metallic conductivity is first achieved for related oxides when turning to aliovalently substituted systems, such LaNiO3 films were deposited on MgO, sintered high-purity alumina, Si or sapphire substrates at temperatures between as e.g.La1-tSrtMnO35 and La1-tCatCrO3,6 where the chemical complexity is larger. 350 and 800 °C.9 (111)- and (100)-oriented LaNiO3 films were The metallic properties and a good lattice match with other obtainedon sapphire (001) and MgO(100) substrates,9 respectperovskite- type oxides like PbTiO3 and YBa2Cu3O7-d ively. The resistivity of films deposited above 600 °C was found (YBCO), make LaNiO3 a very interesting candidate for thin to be ca. 4×10-6 V m. LaNiO3 synthesized during such experi- film applications.Thin film depositions of LaNiO3 have been ments was found to have better characteristics as the bottom reported by various techniques including laser deposition, electrode for PbTiO3 ferroelectric films than the conventional spray combustion flame technique, rf magnetron sputtering, Pt electrodes. spray-ICP and spin coating followed by pyrolysis7–13 where Multilayer thin films of BaTiO3–LaNiO3 and PbTiO3– the motivation has been the use of LaNiO3 as an electrode, as LaNiO3 have been deposited on MgO substrates by using the a substrate for YBCO or as the metallic part in SNS junctions.spray combustion flame technique.10 In similar studies LaNiO3 The crystal structure of the thin films of LaNiO3 is reported was deposited on sintered alumina, sapphire and MgO.11 The to be close to cubic,9–11 possibly corresponding to a high- lowest resistivity, 6×10-6 V m, was measured on a LaNiO3 temperature modification of LaNiO3, which is reported to film deposited on MgO at a temperature above 700°C.exist above 940 °C.14 This temperature is, however, above the As a bulk material, LaNiO3 can be synthesized in a rather thermal decomposition limit.15 Recent studies indicate that the straightforward way, provided that care is taken to avoid cubic modification occurs for stoichiometric samples and that decomposition owing to the reduction of NiIII at high temperaslight deviations in the oxygen content or the incorporation tures (above 800 °C in pure oxygen) and/or under low oxygen of alkali-metal impurities give a rhombohedral distortion.16 partial pressures. Decomposition will give phase mixtures consisting of La2Ni2O5, La4Ni3O10, La2NiO4 , LaNiO2, NiO, La2O3 and Ni depending on the conditions.14–16 The synthesis † Present address: University of Oslo, Department of Chemistry, P.O.Box 1033 Blindern, N-0315 Oslo, Norway. is best performed by using sol–gel precursors, e.g.citrate gels, 449 J. Mater. Chem., 1997, 7(3), 449–454450 or in basic carbonate melts. At low temperatures where struc- Characterization tural reconstructions are kinetically hindered, LaNiO3 can be Crystal structure data and crystallite orientation of the films reduced continuously in a topotactic reduction from LaNiO3 were determined by X-ray diffraction (XRD) measurements to LaNiO2.5=La2Ni2O5 .17 The latter type of reduction can with a Philips MPD 1880 powder diffractometer using Cu-Ka usually be performed for perovskite-type oxides with transition- radiation.The average thickness of a deposited thin film was metal cations in high oxidation states. During the reductions, estimated by measuring the thickness at three different points the properties of the materials may change dramatically, e.g.using a profilometer (Veeco Instruments Dektak 3030). The the conductivity may be changed by several orders of magni- steps were etched by 5 mol dm-3 hydrochloric acid. A four- tude. Hence, tuning the physical properties of the bulk and point probe method was used to measure the sheet resistances. presumably also thin films is partly possible via monitoring The measurements were performed in air at room temperature.the oxygen content of the material during thermal treatment. The resistances were measured at several places and an average The motivation for pursuing thin film deposition of LaNiO3 value was calculated. is manifold. Besides the applicational aspects of LaNiO3, it is A Seiko TG–DTA instrument of the SSC 5200 series was of fundamental interest prior to further studies of perovskite- used to study the thermal behaviour of the precursors, La(thd)3 type oxide thin films to demonstrate the feasibility of CVD and Ni(thd)2.TG and DTA curves were recorded simul- and ALE techniques for producing LaNiO3 thin films. Thin taneously in a nitrogen atmosphere under 7 mbar pressure.film growth of binary metal oxides by ALE has been demon- The sample masses were ca. 7 mg. strated for several compounds but the more complex ternary X-Ray photoelectron spectroscopy measurements were per- compounds have not yet been studied.18 The chemical com- formed with a Kratos Analytical Axis 165 instrument using plef LaNiO3 is fortunately considerably smaller than that monochromated Al-Ka radiation, 0.5 eV step and 80 eV ana- encountered, for example, in materials like La1-tSrtMnO3 and lyser pass energy.As many films were insulating, sample YBa2Cu3O7-d. Secondly, it is of interest to produce a LaNiO3 surfaces were flooded with slow electrons during the acqui- thin film for which the physical properties subsequently can sition.The deviation in the binding energies, due to the charge be adjusted by subjecting it to temperature/oxygen partial neutralisation, was corrected using the C 1s contamination pressure conditions which are selected independently on the peak referenced at 284.8 eV. A scanning electron microscope basis of studies of bulk samples. of type Philips XL30 was used to study the morphology and quality of the thin films.AFM measurements were made in a Nanoscope III AFM instrument. Magnetization measurements were carried out between 5 and 320 K in a magnetic field of Experimental 1 kOe with an MPMS system (Quantum Design). Film growth Films were grown in a flow-type ALE reactor which has been Results and Discussion described elsewhere.19 La(thd)3 and Ni(thd)2 (thd=2,2,6,6- tetramethylheptane-3,5-dionate), synthesized in Espoo, were Growth conditions for LaNiO3 films used as precursors and ozone as the oxidizer.Ozone was produced by feeding oxygen gas into the reactor through an A major motivation for this work was to demonstrate the feasibility of making LaNiO3 as a thin film in an ALE reactor. ozone generator (Fischer model 502). The concentration of ozone was ca. 10% (60 g m-3). The gas flow-rate during the Hence, the work was concentrated on studying the growth rate as a function of deposition temperature and number of pulse was ca. 60 cm3 min-1, measured for the oxygen gas. The reactant pulses were separated by nitrogen gas purging. After cycles, and to characterize the LaNiO3 thin films thereby obtained. Atomic layer epitaxy (ALE) is a technique used for lanthanum and nickel pulses the purging time was 2.5 s and after ozone it was 3 s.Typical reactant pulse durations were growing single crystals and thin films.20 In this method the reactants are alternately pulsed into the reactor chamber, 1.8 s for La(thd)3 and Ni(thd)2 and 1.0 s for ozone. Pulsing sequences used for obtaining various thin film thicknesses were where the substrates are located.Between the reactant pulses any excess of the reactants is purged out with an inert gas, chosen as amultiplet of a basic cycle consisting of 15 alternating cycles of La and O precursors followed by 15 cycles of Ni and leaving ideally one monolayer of the reactant chemisorbed on the substrate. In practice, however, the growth per cycle is a O precursors. The films were grown on Corning 7059 glass substrates fraction of a monolayer owing to steric and other effects.20–22 La–Ni–O thin films were deposited from La(thd)3 and under a pressure of 0.4–1 mbar (measured before the reaction chamber).The substrate size used was ca. 5×5 cm2. Ni(thd)2 using ozone as an oxygen source. By varying the source temperatures the optimum sublimation behaviour of Depositions were made in the temperature range 150–500 °C.The growth rate was studied, both as function of temperature the La and Ni precursors was found at source temperatures of 190 and 145 °C, respectively. The precursor materials were and by varying the number of cycles (30–272)×30 at a selected temperature, 400 °C. pulsed separately, since the difference in sublimation temperature is too large to mix the solid precursors before they are Several films were subsequently heated in a tube furnace in flowing oxygen at 600 °C for 12 h.Other thin films were heat- sublimed. The TG–DTG–DTA recordings of the precursor materials, La(thd)3 and Ni(thd)2, show almost complete subli- treated in a tube furnace which was connected to a gas mixing system and an oxygen sensor (yttrium-stabilised zirconia, mation.La(thd)3 sublimes at around 200 °C and Ni(thd)2 at around 150 °C, see Fig. 1. The pressure used in these thermo- Dansensor). Reduced oxygen partial pressure was obtained by using N2 [ p(O2)#3.8×10-4 bar] or a CO2–Ar mixture with analytical studies (7 mbar) is somewhat higher than the pressure in the reactor during the depositions (0.4–1 mbar). After 2% H2 [ p(O2)#10-24 bar].J .Mate r . Chem., 1997, 7(3), 449–454 451 Fig. 2 Dependence of LaNiO3 film thickness on the deposition temperature Fig. 3 Dependence of film thickness on the number of basic cycles Fig. 1 TG–DTG–DTA curves for the precursors recorded with a (deposition temperature 400°C) heating rate of 10 °C min-1 in flowing nitrogen at 7 mbar pressure: (a) La(thd)3 6.5 mg, (b) Ni(thd)2 7.1 mg basic cycle).This means that the thickness of the thin film can be controlled easily by the number of deposition cycles. depositions in the ALE reactor, however, there were residues in the source crucibles indicating partial decomposition during Characterization of as-deposited and annealed La–Ni–O films prolonged heating.The residue in the Ni(thd)2 crucible was partly white–grey, different from the pink source material. In The thin films deposited at temperatures below 450 °C are transparent and X-ray amorphous, as shown by X-ray diffrac- the case of La(thd)3, however, there were no visible changes between the residue and the white La(thd)3 source material. tion patterns, see Fig. 4(a). They exhibit a range of colours, mostly from yellow to brown, but sometimes the leading edge The deposition temperature has a great influence on the film growth. The dependence of the film thickness on the growth of the substrate is almost black. The size of this black area tends to increase when the deposition temperature is lowered temperature for depositions consisting of 272 basic cycles (272×30 cycles) is shown in Fig. 2. The growth rate increases and thin films deposited at 150–215 °C are totally black–grey. The possible reasons for these colour variations are discussed as a function of temperature from 150°C to 215 °C. In the temperature region from 215 to 250°C the growth rate seems later. Some of the thin films deposited at 450 °C are partly crystalline and all the XRD peaks can be indexed as LaNiO3 to be constant at 0.08 A° cycle-1 (2.4 A° per basic cycle).This plateau shows saturation of the growth independent of the reflections. The crystalline LaNiO3 thin films are black and mirror-like. At higher reactor temperatures (500 °C) the films deposition temperature, which is an indication of a possible ALE window.21 At temperatures between 250 and 400°C the are black and much less mirror-like, and the XRD pattern indicates a phase mixture of La2NiO4 and NiO.This means growth rate increases again indicating another reaction mechanism. For thin films deposited above 400 °C the thicknesses that 450 °C is close to the upper temperature limit for successful deposition of LaNiO3. This temperature is far below the were not measured because of the poor film quality.The dependence of the film thickness on the number of observed decomposition temperature of polycrystalline LaNiO3 in an oxygen atmosphere.14,15 cycles was studied at 400°C, using the pulse durations described above. The growth rate shows an almost linear The amorphous films become crystalline when heated in flowing oxygen at 600 °C for 12 h either in the reactor or in a dependence on the number of cycles, see Fig. 3. The growth rate saturates to a value of 0.24–0.26 A° cycle-1 (7.2–7.8 A° per separate tube furnace. Some crystallinity develops also for thin452 Fig. 5 SEM picture of an LaNiO3 thin film deposited at 400 °C on Corning glass (pressure 0.4 mbar) and heated at 600 °C for 12 h in O2.Film thickness ca. 2000 A° . orientation and the intensities match well those of polycrystalline LaNiO3. Since Corning glass was used as the substrate, epitaxial or strongly textured films were not expected. The quality and morphology of the thin films were examined by scanning electron and atomic force microscopies. SEM pictures indicate that the thin films are dense and no phase inhomogenities are evident, but various amounts of depressions Fig. 4 Powder X-ray diffractogram (Cu-Ka radiation) of (a) an and other thickness variations in the thin films are seen in amorphous thin film, deposited at 400 °C, 0.4 mbar on Corning glass some samples. The surface of the LaNiO3 thin films obtained substrate, (b) an LaNiO3 thin film after annealing at 600 °C for 12 h in O2.Miller indices are given. Film thickness ca. 2000 A° . appears to be more smooth than that of LaNiO3 films deposited by other methods.9,11 Generally the SEM pictures are contourless. For some films, small amounts of spherical films heated at 500 °C in air for 12 h; however, when treated particles are observed, see Fig. 5. It was not possible to at 400 °C in air or oxygen they remain amorphous.distinguish between the composition of the thin film and the Simultaneously with the crystallization a colour change from particles. Similar particles have been reported for thin LaNiO3 yellow–brown to black occurs. The XRD patterns from the films made by spray combustion flame and spray-ICP tech- heated samples agree with that for LaNiO3, see Fig. 4(b). The niques.9,11 Such particles are observed for the crystalline as d-values obtained for a LaNiO3 film are listed in Table 1.well as for the amorphous thin films. No obvious relation Compared with the powder X-ray patterns of the polycrystal- between the number of particles and the deposition parameters line bulk materials, no splitting of reflections owing to a could be found. The SEM examinations show no significant rhombohedral distortion (aRH=5.395 A° and a=60.77° for difference between the amorphous, crystalline, yellow–brown LaNiO317) is observed.The crystalline film is therefore cubic. and black films. This is consistent with earlier reports on LaNiO3 thin films AFM measurements (Fig. 6) show that the as-deposited deposited by other techniques.9–11 For the LaNiO3 film heated/ amorphous thin films contain small crystallites.The rough- crystallized in oxygen at 600 °C [Fig. 4(b)], the unit-cell param- nesses of the as-deposited films are the same or even less than eter a=3.804 A° was calculated. As seen from the Miller indices the roughness of the substrate, which is a sign of smooth in Fig. 4(b) for LaNiO3, there is no indication of a preferred growth of the amorphous thin film. No variations in microstructure due to thickness variations or different deposition Table 1 Positions of Bragg reflections given as d-values for LaNiO3 temperatures were found.However, owing to annealing the thin film crystallized at 600 °C; data from refs. 11 and 23 are included crystallite size and the roughness of the films increases, see for comparison Fig. 6. hkl d/A° a d/A°b d/A° c Crystallinity, colour and physical properties 100 3.809 3.816 3.850 110 2.690 2.692 2.730 The La–Ni–O films show a large range of characteristics, from 111 2.187 2.192 2.229 amorphous to crystalline, from yellow via brown to black, and 200 1.905 1.907 1.932 a five orders of magnitude variation in resistivity. In order to 210 1.703 1.708 1.728 understand the cause of these variations a number of annealing 211 1.548 1.578 experiments, XPS, magnetic susceptibility and resistivity stud- 220 1.345 1.348 1.365 ies were undertaken.First the colour will be considered. The thin films annealed aThis study. bRef. 11. cRef. 23.J . Mate r . Chem., 1997, 7(3), 449–454 453 described for deposition of the mixed oxide.The number of cycles used was 5000. Depositions of the Ni–O thin films were made at 250, 350 and 400 °C. XRD measurements show that the as-deposited Ni–O thin films were crystalline. The XRD patterns for films grown at 250°C (1 mbar) and 350 °C (0.8 mbar) were indexed as NiO. At 400 °C (1 mbar) the thin film consisted of a mixture of NiO and Ni. The La–O thin film growth was studied in the temperature range 200–450 °C, depositions being made every 50°C.XRD measurements show that the as-deposited thin films were crystalline when grown at temperatures between 350 and 450°C.24 In order to study further the possible reasons for the changes in colour, some annealing experiments were carried out under a reducing atmosphere. In a very reducing atmosphere [CO2–N2–2% H2 ; p(O2)=10-24 bar, T=600 °C], complete decomposition occurs for both amorphous and crystalline La–Ni–O films, giving in both cases dark brown films containing crystalline Ni+La2O2CO3 (type I or IA).Under modestly reducing conditions [N2; p(O2)=10-4 bar], the yellow and brown amorphous films turn black. However, they still remain X-ray amorphous. These experiments show that the atmosphere is of great importance for obtaining a crystalline LaNiO3 thin film, and further that the colour is not connected directly with the crystallinity. For La–Ni–O films converted to crystalline LaNiO3 by heat treatment in oxygen, exposure to N2 does not change the colour or the diffraction pattern. The latter experiments were conducted under normal atmospheric conditions, under which the thin film may reoxidize from a reduced state, see below.If the amorphous phase is less stable than the crystalline LaNiO3 under an N2 atmosphere, the observed changes in colour may possibly be due to decomposition of Fig. 6 Atomic force microscopy (AFM) pictures of LaNiO3 thin films the amorphous LaNiO3 phase into X-ray amorphous forms of deposited on Corning glass at 250°C (film thickness ca. 670 A° ). (a) Asdeposited, (b) after annealing at 600 °C for 12 h in O2 . one or more of the phases La2Ni2O5, La4Ni3O10, La2NiO4, LaNiO2, NiO, La2O3 and Ni. The range of oxygen non-stoichiometry in LaNiO3 is large in O2 at 600 °C contain LaNiO3 . They are black, independent at low temperatures where topotactic reduction will remove of the state of the intermediate amorphous film.The reason oxygen from alternating sheets in the crystal structure, resulting for the yellow–brown–black colouring of the amorphous films in La2Ni2O5 with octahedral and square-planar coordinated appears unclear. The colour variations may have different nickel.25 At higher temperatures than, say, 500 °C, such causes: (i) film thickness; (ii) carbon impurities introduced reduction is no longer possible, the kinetically stable perovsk- from decomposed precursor; (iii) degree of crystallinity; ite-related structure is broken up and the thermodynamically (iv) multiphase nature of the thin films; or (v) oxygen non- stable two-phase mixture is obtained.For topotactically stoichiometry. The first possibility can be excluded since the reduced bulk specimens, complete reoxidation occurs at room colour variations are found over films where the measured temperature.It is therefore not very probable that the thin thickness is constant. Furthermore, some of the blackest films film should contain regions with significantly different oxygen are actually very thin. Possibility (ii) is also unlikely, since non-stoichiometries.enhanced carbon formation owing to thermal decomposition As regards the electrical resistivity, it was found to vary of La(thd)3 and Ni(thd)2 precursors is expected at the higher from 3.2 V m for an amorphous as-deposited film to deposition temperatures. This is not the case since films 1.8×10-5 V m for the same LaNiO3 film after annealing and deposited at 400 °C still have colour variations, whereas some crystallization. The lowest resistivity observed was films deposited at 200 °C are virtually black.All films deposited 5.4×10-6 V m. The latter value is slightly larger than found at temperatures below 450°C are X-ray amorphous and turn for spray-ICP deposited films9 and for laser deposited LaNiO3 crystalline only after thermal treatment. Hence, the colour films,8 but probably equal within reasonable uncertainty limits.differences are not just connected with the X-ray crystallinity On the other hand, the resistivity is significantly lower and the of the films; however, they may still be connected with micro- minimum is obtained at lower temperature than for LaNiO3 crystallinity. prepared by pyrolysis of spin-coated organometallic films.13 The colour variations in the amorphous thin films could There is no difference in resistivity between the yellow, brown possibly be caused by a multiphase situation (iv), being due to and black amorphous films.either a mixture of La2O3 and NiO or other La–Ni–O phases. In order to evaluate this possibility, films of the single oxides Characterization by XPS and magnetic susceptibility La2O3 and NiO were deposited, using La(thd)3 and Ni(thd)2 measurements as precursors and the same pulsing and purging times as Characterization of the electronic state of nickel in La–Ni–O films was attempted via XPS and magnetic susceptibility454 studies.The surface chemistry of two films with and without annealing was studied with XPS. The films were deposited at 250 and 400 °C and the annealing was carried out in O2 at 600 °C for 12 h.As the samples were measured without any in situ cleaning in UHV, both carbon and oxygen contamination species were present on the surface. No other impurities were detected from the spectra. The surface concentration of nickel in the nonannealed specimen was roughly three times that of lanthanum whereas for annealed samples the amounts of lanthanum and nickel at the surface were nearly equal.As the last sequence in the growth of the films was 15 cycles of nickel and ozone precursors, the XPS results indicate that the outermost layers reacted only during the annealing when the films also changed from amorphous to crystalline. The O 1s line shape changed markedly upon annealing (Fig. 7). The as-deposited films had two resolvable components. The broad high binding energy component at 531.3±0.1 eV probably originates from the chemisorbed O- and OH- groups and the nickel-bound oxygen.26–28 The low energy component at 528.7 eV was assigned to the La2O3 component.29 However, the annealed films were composed of three resolvable peaks. The positions of the low and high binding energy components Fig. 8 XP spectra of the Ni LMM Auger level of (a) an NiO reference film grown at 250 °C; (b) as-deposited and (c) annealed LaNiO3 films remained nearly unchanged (at 531.6 and 528.7 eV). The grown at 400 °C on a Corning glass substrate additional component at 530.0±0.1 eV is in good accordance with the reported values for LaNiO3.27,28 ted by the deposition temperature and annealing.Nickel Since the La 3d3/2 emission interferes severely with the Ni 2p interferes with the La 3d3/2 signal, but a subtle change in the emission, both the weak Ni 3p and Ni LMM Auger emissions splitting ratio of the La 3d5/2 component was detected after were also studied. In all films analysed the Ni 2p peak was annealing the films. This could indicate changes in the conduc- very similar in shape and position to the reference spectrum tion band electron configuration which could be due to changes of NiO film, verifying that no Ni2O3 was present on the surface.However, in the Ni LMM Auger signal of the annealed in the La coordination number before and after annealing, as films (Fig. 8) a slight change in the peak shape was observed.discussed by Burroughs et al.30 This could be related to structural changes in the matrix. Metallic LaNiO3 is reported to be Pauli paramagnetic.8 Any The La 3d5/2 core-level spectra show split lines with maxima presence of reduced NiII species will add an additional Curie– at 834.1±0.3 and 838.1±0.3 in agreement with the literature Weiss contribution to the susceptibility.Sagoi et al.8 reported, values for LaNiO3.26 The positions of the lines remain unaffec- for LaNiO3 deposited at 700 °C, the presence of 16% NiII, whereas in the low-resistivity films they observed no Curie– Weiss contribution from NiII. The present susceptibility measurements were performed on small pieces of the substrate coated with the La–Ni–O film. Two approaches were used: Fig. 7 XP spectra of the O 1s level of (a) as-deposited and (b) annealed LaNiO3 films grown at 400°C on a Corning glass substrateJ . Mate r . Chem., 1997, 7(3), 449–454 455 the first involved measurement of the substrate plus film, then measurements and aiding their interpretation as well as Mr. Mikko Utriainen, MSc, for the AFM measurements. removal of film by treatment with hydrochloric acid and subtraction of the measured signal for the pure substrate; the second method was a simple comparison between as-measured References susceptibilities without subtraction of the contribution from the glass substrate.The magnetic susceptibility curves for the 1 N. Q. Minh, J. Am. Ceram. Soc., 1993, 76, 563. 2 Properties and Applications of Perovskite-type Oxides, ed.as-deposited film at 400 °C are very similar to those observed L. G. Tejuca and J. L. G. Fierro, Marcel Dekker, New York, for a film annealed subsequently at 600°C in O2 . All curves 1992, ch. 10–17. show a small Curie–Weiss contribution. This indicates that the 3 U.Ko� nig, O. Blum, R. Christ and I. Reeh, J. Phys. Chem., 1993, as-deposited films contain the same amount of reduced nickel 97, 488.species. For all samples a more or less clear cusp appear in 4 R. N. Singh, L. Bahadur, J. P. Pandey and S. P. Singh, J. Appl. Electrochem., 1994, 24, 149. x(T ) around 50 K. It is possible that a similar, but less 5 A. Mackor, T. P. M. Koster, J. G. Kraaijkamp, J. Gerretsen and pronounced, peak is also present in the data of Xu et al.31 In J. P. G. M. van Eijk, Proc. 2nd Int. Symp. Solid Oxide Fuel Cells, the study by Sreedhar et al.32 it is not possible to observe this The Electrochemical Society, Pennington, NJ, 1991, p. 463. peak, but the reason for this might be few measuring points 6 J. Mizusaki, S. Yamauchi, K. Fueki and A. Ishikawa, Solid State in the relevant temperature range. The feature is hardly present Ionics, 1984, 12, 119. in the as-deposited La–Ni–O at 400 °C, however, becoming 7 K.M. Satyalakshmi, R. M. Mallya, K. V. Ramanathan, X. D. Wu, B. Brainard, D. C. Gautier, N. Y. Vasanthacharya and very pronounced after crystallization and oxidation. It is M. S. Hedge, Appl. Phys. L ett., 1993, 62, 1233. believed that the feature is due to adsorbed oxygen.33 8 M. Sagoi, T. Kinno, T. Hushimoto, J. Yoshida and K.Mizushima, Appl. Phys. L ett., 1993, 62, 1833. 9 H. Ichinose, M. Nagano, H. Katsuki and H. J. Takagi, J. Mater. Conclusions Sci., 1994, 29, 5115. 10 H. Ichinose, Y. Shiwa and M. Nagano, Jpn. J. Appl. Phys., 1994, Thin film growth of LaNiO3 by an ALE process has been 33, 5903. demonstrated from b-diketonate type precursors and ozone. 11 H. Ichinose, Y. Shiwa and M. Nagano, Jpn.J. Appl. Phys., 1994, 33, 5907. Even a very low deposition temperature of 250 °C can be used 12 C. C. Yang, M. S. Chen, T. J. Hong, C. M. Wu and J. M. Wu, Appl. but crystallization requires annealing at 600 °C. The annealed Phys. L ett., 1995, 66, 2643. films show metallic behaviour and are black, in contrast to 13 A. Li, C. Ge and P. Lu�, Appl. Phys. L ett., 1996, 68, 1347. the yellow colour and high resistivity of the as-deposited 14 H.Obayashi and T. Kudo, Jpn. J. Appl. Phys., 1975, 14, 330. amorphous films. The XP spectra indicate that the amorphous 15 H. Fjellva°g, O. H. Hansteen, B. Gilbu, A. Olafsen, N. Sakai and as-deposited thin films consist of separate oxide layers with H. Seim, T hermochim. Acta, 1995, 256, 75. 16 S. Rakshit and P. S. Gopalakrishnan, J.Solid State Chem., 1994, La–O and Ni–O character. These separate layers will react 110, 28. during annealing and form the crystalline perovskite structure. 17 M. Crespin, P. Levitz and L. J. Gatineau, J. Chem. Soc., Faraday T rans. 2, 1983, 79, 1181. Financial support from Norsk Hydro A/S and NorFa and 18 L. Niinisto�, M. Ritala and M. Leskela�, J. Eng. Mater., in press. the Academy of Finland is gratefully acknowledged. We also 19 T. Suntola, A. Pakkala and S. Lindfors, US Pat., 4 3983. thank Dr. Leena-Sisko Johansson for performing the XPS 20 M. Leskela� and L. Niinisto�, in Atomic L ayer Epitaxy, ed. T. Suntola and M. Simpson, Blackie and Son Ltd., Glasgow, 1990, p. 1. 21 T. Suntola, in Handbook of Crystal Growth, ed. D. T. J. Hurle, Elsevier, Amsterdam, 1994, vol. 3, p. 601. 22 H. Mo�lsa� and L. Niinisto�, Mater. Res. Soc. Symp. Proc., 1994, 335, 341. 23 JCPDS file 33-710; H. Wustenberg, Hahn Inst. fu�r Kristallogr., Technische Hochschule Aachen, Germany. 24 H. Seim, M. Nieminen, H. Fjellva°g and L. Niinisto�, unpublished work. 25 T. Moriga, O. Usaka, T. Imamura, I. Nakabayashi, I. Matsubara, T. Kinouchi, S. Kikkawa and F. Kanamaru, Bull. Chem. Soc. Jpn., 1994, 67, 687. 26 J. Choisnet, N. Abadzhieva, P. Stefanov, D. Klissurski, J. M. Bassat, V. Rives and L. Minchev, J. Chem. Soc., Faraday T rans., 1994, 90, 1987. 27 J. L. G. Fierro and G. Tejuca, Appl. Surf. Sci., 1987, 27, 453. 28 L. G. Tejuka and J. L. G. Fierro, T hermochim. Acta, 1989, 147, 361. 29 J. F. Moulder, W. F. Stickle, P. E. Sobol and K. D. Bomben, in Handbook of X-Ray Photoelectron Spectroscopy, ed. J. Chastain and R. C. King, Eden Prairie, Minnesota, 1995, p. 261. 30 P. Burroughs, A. Hamnett, A. F. Orchard and G. Thornton, J. Chem. Soc., Dalton T rans., 1976, 1686. 31 X. Q. Xu, J. L. Peng, Z. Y. Li, H. L. Ju and R. L. Greene, Phys. Rev. B, 1993, 48, 1112. 32 K. Sreedhar, J. M. Honig, M. Darwin, M. McElfresh, P. M. Shand, J. Xu, B. C. Crooker and J. Spalek, Phys. Rev. B, 1992, 46, 6382. 33 Z. Zhang and M. Greenblatt, J. Solid State Chem., 1995, 117, 236. Paper 6/06316K; Received 13th S

ISSN:0959-9428    

DOI:10.1039/a606316k

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

Effect of catalyst on in situ silica reinforcement of styrene–butadiene rubber vulcanizate by the sol–gel reaction of tetraethoxysilane Yuko Ikeda,*a Akira Tanakaa and Shinzo Kohjiyab aFaculty of Engineering and Design, Kyoto Institute of T echnology, Matsugasaki, Sakyo, Kyoto 606, Japan bInstitute for Chemical Research, Kyoto University, Gokasyo, Uji 611, Japan In situ silica reinforcement of styrene–butadiene rubber (SBR) vulcanizate has been achieved by the sol–gel reaction of tetraethoxysilane (TEOS) using n-butylamine as a catalyst.SBR was vulcanized with sulfur and soaked in TEOS and in an aqueous solution of the catalyst. When hydrochloric acid was used as a catalyst for the sol–gel reaction, silica particles were not introduced into the SBR matrix in this study. The increase of the dynamic modulus and tensile strength at break was considered to be due to the interaction between the rubber and the in situ silica filler in the SBR vulcanizate.Rubber is an elastomer and the characteristic of rubber elas- moisture or silane curing,11,12 and it has been carried out ticity for practical applications is provided by vulcanization already on silicone rubber,13–15 polyisobutene16 and epoxidized and the formation of compounds with fillers.1,2 The most natural rubber.17 However, no report has yet been made on reinforcing filler is carbon black.3,4 Silica is known as a ‘white the effect of the catalyst of the sol–gel reaction in the generalcarbon’ and has been developed as a reinforcing filler for purpose grade rubbers from the viewpoint of the morphology synthetic rubbers.3–5 Conventionally, filler-reinforced rubber is of in situ silica in the rubber matrix.prepared by mechanical mixing of the filler followed by compression moulding for curing. However, the incorporation of silica into rubbers by normal mixing techniques increases the viscosity because of the small size and large specific surface area of the filler. Silica also deactivates curing agents and accelerators which are commonly used in rubber technology, thus resulting in an insufficient cure.These problems are partially overcome or minimized by specific additives and mixing procedures.6 The other important difficulty is the incompatibility of inorganic silica with the organic rubber. As a method for overcoming these difficulties, the in situ polymerization of tetraethoxysilane (TEOS) by the sol–gel process in the conventional diene rubber matrix has been developed and described in our earlier communications.7,8 It is an application of the sol–gel process, for the preparation of inorganic glasses at low temperatures,9,10 to rubber chemistry. 11 The reaction of TEOS takes place in two steps, hydrolysis and condensation, and results in SiO2 as shown in Fig. 1. This application of the sol–gel process to rubber chemistry is somewhat related to the use of silane coupling agents and Si OC2H5 Si OH + H2O + C2H5OH Si OC2H5 Si O + HO + C2H5OH Si Si (2) (1) Si OH Si O + HO + H2O Si Si (3) hydrolysis condensation overall reaction Si(OC2H5)4 + 2H2O SiO2 + 4C2H5OH (4) Fig. 1 Hydrolysis and condensation reactions of TEOS 455 J.Mater. Chem., 1997, 7(3), 455–458456 In this study, the reinforcing effect of in situ formed silica Thermogravimetry. Thermogravimetry (TG) was carried out using a Rigaku TG Instrument. A sample (ca. 100 mg) was particles in the styrene–butadiene rubber (SBR) vulcanizate was elucidated. In particular, the effect of the catalyst of the placed in a platinum pan and heated in air to 1000 °C at a rate of 20°C min-1.The silica contents of the in situ silica- sol–gel reaction on the in situ silica filling in the SBR matrix is described. filled vulcanizates were determined by TG, i.e. calculated using eqn. (6) Experimental silica content (%)=100 (M3/M4) (6) Materials where M3 is the mass of in situ formed silica and M4 is the Styrene–butadiene rubber (SBR 1502) was supplied by the mass of the silica-filled vulcanizate.The value was based on Japan Synthetic Rubber (JSR) Co. and its properties are the residual mass at 800 °C, which was an appropriate tempera- summarized in Table 1. Tetraethoxysilane (TEOS) was ture since the mass was observed to be constant at ca. 600 °C. obtained from Shin-etu Chemical Ind.Co. The catalysts were It was corrected by using the residual mass of non-filled hydrochloric acid and n-butylamine, which were reagent grade. vulcanizate. All reagents were used as received unless otherwise noted. Preparation of the rubber vulcanizates Tensile test. Tensile properties of the silica-filled vulcanizates were measured at room temperature at a strain rate of 100 mm SBR and reagents were mixed in a two-roll mill; the pro- min-1 using ring-shaped specimens.The values reported here portions of reagents are shown in Table 2. Rubber vulcanizates are based on an average of five measurements for each sample. were prepared by curing at 150 °C for a curing time of 50 min, which was determined from the cure curve obtained with a JSR Curelastometer III.Dynamic mechanical analysis. Dynamic mechanical analysis (DMA) was carried out with a Rheospectoler DVE-4 instru- Preparation of the in situ silica-filled vulcanizates by the sol–gel ment (Rheology Co., Kyoto) at a frequency of 10 Hz and a method heating rate of 2°C min-1. SBR vulcanizates were swollen in TEOS at 30°C for 48 h and soaked in an aqueous solution of 1 mol dm-3 hydrochloric Optical microscopy.The optical microscopic observation acid or in a 10 mass% aqueous solution of n-butylamine at was carried out using a Nikon Polarizing Microscope (Model 30°C for 24 h. Then, the samples were heated at 50°C for 72 h POH 3). and dried for several days at 50°C under reduced pressure. The amounts of TEOS and the catalyst solution were ten times the mass of the sample film.The in situ silica-filled vulcanizate Transmission electron microscopy. Ultrathin films of the prepared using the acid catalyst and that prepared using the samples were prepared using a microtome (KLB 4800A base catalyst are hereafter abbreviated as SBR-a and SBR-b, Ultrotome) in liquid nitrogen of KLB 14800 Cryokit. The respectively. specimen was placed on a copper grid, which was coated with FolmbarB and evaporated carbon in advance. Then, trans- Characterization of the silica-filled vulcanizates mission electron microscopy (TEM) observation was carried Swelling.The sample was soaked in TEOS at 30°C for 48 h, out with a JEOL TEM-100U instrument, without staining. and the degree of swelling was calculated using eqn.(5) The accelerating voltage was 80 kV. degree of swelling=100 [(M1-M2)/M2] (5) where M1 is the mass of film after swelling and M2 is the mass Results and Discussion of film before swelling. Conversion of TEOS into silica in the SBR matrix Table 1 Properties of SBR 1502 Table 3 shows the results of swelling of the vulcanizates in type cold TEOS and the silica content after the sol–gel reaction.The bound styrene (mass%) 23.5 conversion of TEOS into silica in the SBR vulcanizates was Mooney viscosity (ML1+4, 100 °C) 52 product stain non-staining Table 2 Reagents used in the preparation of SBR vulcanizate (phra) Table 3 Results of swelling and the sol–gel reaction SBR 1502 100 residue degree of degree of ZnO 5.0 at 800 °C swelling SiO2 sol–gel stearic acid 1.5 sample in TG (%) in TEOS (%) content (%) reaction (%) MSA-Gb 0.5 sulfur 0.5 SBR-g 3.4 145.9 — — SBR-a 22.2 — 19.5 59.6 aParts per hundred rubber in terms of mass.bN-Oxydiethylene-2- SBR-b 25.9 — 23.3 74.8 benzothiazole sulfenamide.J . Mate r . Chem., 1997, 7(3), 455–458 457 evaluated from the degree of swelling in TEOS and from the silica content by using eqn. (7) degree of sol–gel reaction (%)=100 (M3/M5) (7) where M5 is the mass of silica calculated from the incorporated TEOS in the vulcanizate.The conversion of TEOS of SBR-b was ca. 75%, while that of SBR-a was ca. 60%. The conversion of TEOS is controlled by the temperature and time of the sol–gel reaction. Effect of catalyst on the reinforcement of the SBR vulcanizates filled with in situ formed silica Fig. 3 Optical photographs of SBR-g, SBR-a and SBR-b Tensile stress–strain curves of SBR-a and SBR-b vulcanizates are illustrated in Fig. 2, together with that of SBR gum (SBRg) vulcanizate. The effect of in situ silica reinforcement was significant in SBR-b, i.e. its modulus and its tensile strength at break increased greatly compared to those of SBR-g. However, the mechanical properties of SBR-a were only slightly improved, although the silica content of SBR-a was ca. 20%. In order to evaluate the difference of the tensile properties of SBR-a and SBR-b, the morphology of silica in these SBR vulcanizates was elucidated. First, SBR-a and SBR-b were examined by optical microscopy. The sections of the samples, which were prepared by cutting the sheets, are shown in Fig. 3 with that of SBR-g. The state of SBR-b was almost similar to that of SBR-g. However, the picture of SBR-a clearly shows the presence of silica on the surface layer of the SBR vulcanizate. Fig. 4 TEM photographs of SBR-g, SBR-a and SBR-b (SBR-a-1, the inside; SBR-a-2, the interface between the silica layer and the rubber) TEM photograph (SBR-a-2) of the interface between the silica layer and the rubber is also shown. These observations suggest that hydrochloric acid is not an adequate catalyst for the Fig. 2 Stress–strain curves of SBR vulcanizates with and without in reinforcement of the SBR vulcanizate by the technique situ formed silica. ——, SBR-g; - - - -, SBR-a; — - —, SBR-b. described in the Experimental section. This is due to the nonsolubility of hydrochloric acid into the swollen rubber with Next, the samples were subjected to TEM analysis.Fig. 4 is TEOS. the result of TEM observation for SBR-g, SBR-a and SBR-b. Silica was formed and filled in the vulcanizate when The dark portions in the photograph of SBR-b are silica n-butylamine was used as a catalyst for the sol–gel reaction of particles. Interestingly, the diameter of the in situ silica particles TEOS in the SBR matrix. The morphology of the in situ filled was ca. 20–35 nm and they were dispersed homogeneously in silica in SBR-b is considered to bring excellent reinforcement the matrix. On the other hand, silica particles were not detected within the SBR-a sample as shown in Fig. 4 (SBR-a-1). The to the SBR vulcanizate.458 Dynamic mechanical properties of SBR-b References 1 T he Chemistry and Physics of Rubber-L ike Substances, ed.The temperature dependence of the dynamic modulus (E¾) and L. Bateman, Maclaren & Sons, London, 1962. tan d for SBR-b and SBR-g are illustrated in Fig. 5. The E¾ of 2 Natural Rubber Science and T echnology, ed. A. D. Roberts, Oxford the in situ silica filled SBR vulcanizate was larger than that of University Press, 1988.the non-filled SBR vulcanizate. This result agrees with that 3 G. Kraus, Reinforcement of Elastomers, Interscience, New York, of the tensile test. The tan d peak of SBR-b, which is attributed 1965. to the glass transition temperature, was detected as well as 4 Science and T echnology of Rubber, ed. F. R. Eirich, Academic SBR-g, but its peak height was decreased compared to that of Press, Orlando, 1978. 5 R. K. Iler, T he Chemistry of Silica, John Wiley & Sons, New SBR-g. This observation suggests the presence of the inter- York, 1979 action between the rubber and the in situ formed silica in the 6 M. P. Wagner, in Rubber T echnology, ed. M. Morton and V. N. SBR vulcanizate. Interestingly, the tan d peak of SBR-b was Reinhold, New York, 1987, p.86.shifted to the lower temperature region after the sol–gel 7 S. Kohjiya and Y. Ikeda, in New Functionality Materials, ed. reaction. It is considered that the swelling in TEOS may T. Tsuruta, M. Doyama and M. Seno, Elsevier, Amsterdam, 1993, contribute to the disentanglement of the polymer chains in the vol. C, p. 443. 8 S. Kohjiya, A. Yajima, Y. Ikeda and J.R. Yoon, Nippon Gomu vulcanizate. Moreover, the plasticization of the rubber by Kyokaishi (J. Soc. Rubber Industry, Jpn.), 1994, 67, 859. 9 S. Sakka, T he Science of the Sol–Gel Process, Agune-Shofusha, Tokyo, 1988 (in Japanese). Fig. 5 Temperature dependence of E¾ (a) and tan d (b) for SBR vulcanizates with and without in situ formed silica. ——, SBR-g; — - —, SBR-b. residual oligomers from the sol–gel reaction might contribute to this phenomenon.Conclusions In situ silica reinforcement of SBR vulcanizates was carried out by the sol–gel process using TEOS. n-Butylamine was a good catalyst for in situ silica filling of the SBR vulcanizate. The silica particles formed in the SBR vulcanizate were dispersed homogeneously in the matrix and the diameter of the particles was ca. 20–35 nm. This study represents a simple and effective method for compounding silica particles into rubber vulcanizates for its reinforcement.J . Mate r . Chem., 1997, 7(3), 455–458 459 10 C. J. Brinker and G. W. Scherer, Sol–Gel Science, Academic Press, 14 J. E. Mark, Science of Ceramic Chemical Processing, ed. L. L. Hench and R. D. Ulrich, John Wiley & Sons, New York, 1985. New York, 1982. 11 Y. Ikeda, A. S. Hashim and S. Kohjiya, Bull. Inst. Chem. Res., 15 J. E. Mark., Chemtech, 1989, 19, 230. 16 C-C. Sun and J. E. Mark, J. Polym. Sci., Part B: Polym. Phys., 1987, Kyoto Univ., 1995, 72, 406. 12 S. Kohjiya and S. Yamashita, J. Appl. Polym. Sci., Appl. Polym. 25, 1561. 17 A. S. Hashim. Y. Ikeda and S. Kohjiya, Polym. Int., 1995, 38, 111. Symp., 1992, 50, 213. 13 J. E. Mark and S-J. Pan, Makromol. Chem., Rapid Commun., 1982, 3, 681. Paper 6/06395K; Received 17th September, 1996

ISSN:0959-9428    

DOI:10.1039/a606395k

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

Neutron diffraction study of the influence of structural disorder on the magnetic properties of Sr2FeMO6 (M=Ta, Sb) Edmund J. Cussen,a Jaap F. Vente,a Peter D. Battle*a and Terence C. Gibbb aInorganic Chemistry L aboratory, University of Oxford, South Parks Road, Oxford, UK OX1 3QR bSchool of Chemistry, L eeds University, L eeds, UK L S2 9JT The crystal structure of the perovskite Sr2FeTaO6 has been refined by simultaneous analysis of X-ray and neutron powder diffraction data collected at 280 K; space group Pbnm, a=5.6204(3), b=5.6161(3), c=7.9266(3) A° .The structure is of the GdFeO3 type, with a disordered distribution of Fe and Ta over the six-coordinate cation sites. The structure of Sr2FeSbO6 has been refined in a similar manner; space group P21/n, a=5.6132(5), b=5.5973(5), c=7.9036(7) A° , b=90.01(1)°.The two crystallographically distinct six-coordinate sites in Sr2FeSbO6 are occupied in a partially ordered manner [0.795(6)50.205(6)] by Fe and Sb atoms. Neutron diffraction data collected from Sr2FeTaO6 at 1.5 K show no evidence of long-range magnetic ordering and, in the light of previous susceptibility and Mo�ssbauer measurements, it is concluded that Sr2FeTaO6 is a spin glass below 23 K.Neutron diffraction data collected from Sr2FeSbO6 at 1.5 K include magnetic Bragg peaks characteristic of a type I magnetic structure with an average ordered moment of 3.06(9) mB per Fe atom on the Fe-dominated octahedral site, and no significant ordered moment on the second site. The magnetic Bragg scattering decreases to zero in the temperature interval 1.5T /K37(2).It is concluded that the partial cation ordering leads to the coexistence of a magnetically ordered spin system and a spin-glass system. The magnetic susceptibility of many transition-metal oxides observed magnetic hysteresis and thermal remanent magnetispasses through a maximum value on cooling. This is often an ation, we suggested that Sr2FeTaO6 is a spin glass below 23 K.indication that the compound has an antiferromagnetic low- In contrast, Sr2FeSbO6 crystallises in the monoclinic space temperature phase, but it is unwise to assume this on the basis group P21/n and shows a partial ordering of the Fe3+ and of susceptibility data alone. It has become clear in recent years Sb5+ cations over the six-coordinate sites.8 The relatively low that, for a number of oxides, the spin system is static below effective magnetic moment and the negative Weiss constant the temperature of the susceptibility maximum but, in contrast derived from magnetic susceptibility data (mFe=5.0 mB, h= to the situation in a true antiferromagnet, there is no long- -221 K) suggested that strong, short-range, antiferromagnetic range orderingof the atomic magnetic moments.The formation ordering is present in Sr2FeSbO6 in the temperature range of this so-called spin-glass phase is most likely to occur when 50T/K280, although no difference between the field-cooled the compound contains more than one magnetic species1 or (FC) and zero-field-cooled (ZFC) magnetic susceptibility data when the particular combination of chemical composition was observed8 in this temperature regime.However, such a and crystal structure leads to a degree of atomic disorder. difference was apparent below the susceptibility maximum at Sr2FeNbO6,2 in which paramagnetic Fe3+ and diamagnetic 36 K, suggesting that a spin-glass phase might exist at low Nb5+ cations are distributed in a disordered manner over the temperatures, although the possibility of long-range magnetic six-coordinate sites of a pseudo-cubic perovskite structure, order could not be ruled out.In order to establish unambigubelongs to the latter category of compound. The concentration ously the nature of any long-range magnetic ordering in these of magnetic cations (50%) on the sites of what is essentially a two compounds, we have now collected neutron diffraction simple cubic cation sublattice is considerably greater than the data on Sr2FeMO6 (M=Ta, Sb) over the temperature percolation threshold (30.7%) for this non-frustrated system,3 range 1.5T /K300.The results of these experiments are but susceptibility and hysteresis data suggest that Sr2FeNbO6 described below. behaves as a spin glass below 32.5 K,4 and there is no evidence of a transition to a phase showing long-range magnetic order.This behaviour is surprising in view of the fact that magneti- Experimental cally concentrated LaFeO3 shows long-range order at tempera- Polycrystalline samples (ca.10 g) of Sr2FeMO6 (M=Ta, Sb) tures as high as 750 K,5 and theoretical models6 predict a fall were prepared as described previously.8,9 Rietveld analyses10 in TN of only ca. 550 K when the Fe3+ concentration on the of X-ray powder diffraction patterns collected at room tem- magnetic sublattice is reduced to 50%.Rodriguez et al.4 perature in Bragg–Brentano geometry on a Siemens D5000 explained this behaviour by postulating the existence of short- diffractometer (Cu-Ka1 radiation, 52h/degrees120, D2h= range structural ordering between Fe3+ and Nb5+, but sub- 0.02°) established that the samples were monophasic.Constant- sequent EXAFS and Mo�ssbauer studies7 failed to find any wavelength neutron powder diffraction data were collected evidence for such an effect. using the instruments D1b and D2b at the ILL, Grenoble. The Following the work of Rodriguez et al., we have reported diffractometer D1b is equipped with a position-sensitive detec- behaviour similar to that of Sr2FeNbO6 in a wide range of tor (PSD), having 400 detectors 0.2° apart in 2h, making it mixed iron oxides with the general formula A2FeMO6 (A= possible to measure a low-resolution but complete diffraction Ca, Sr, Ba;M=Nb, Ta, Sb).8,9 Of these compounds, Sr2FeTaO6 pattern over the angular range 102h/degrees90 without was studied in most detail.The X-ray diffraction pattern moving the detector. The instrument operates at a wavelength indicated that this compound is a cubic perovskite [a= of 2.522 A° . During the course of our experiments on 3.9664(9) A° ] at room temperature with no (structural) ordering Sr2FeSbO6, a computer-controlled cryostat was used to ramp of the cations on the transition-metal sublattice.9 On the basis the sample temperature through the range 1.5T/K100 at of the non-Brillouin temperature dependence of the internal hyperfine field observed in the Mo�ssbauer spectrum, and the a constant rate of 9 K h-1.The data were stored, and the J. Mater. Chem., 1997, 7(3), 459–463 459detectors reset to zero, every 15 min, giving a set of diffraction by performing a refinement using neutron and X-ray diffraction data simultaneously, thus making use of the complementarity patterns with a temperature resolution of 2.25 K.The conventional diffractometer D2b operates at a neutron wavelength of of the two techniques. Thirty-nine variables were refined; seven atomic coordinates, four isotropic temperature factors, three 1.5938 A° (calibrated using X-ray data from our samples).During the course of each experiment (ca. 6 h), the bank of 64 lattice parameters and a set of profile parameters for each of the two data sets. The resulting atomic coordinates are listed detectors was used to collect data over the angular range 52h/degrees140 with a 2h stepsize of 0.05°.Data were in Table 1 and the most interesting bond lengths and bond angles are presented in Table 2. The values of the standard collected on a weighed amount of sample contained in a cylindrical vanadium can (diameter=10 mm) and mounted in deviations on the coordinates are approximately half those obtained in the analysis based on the neutron data alone. The a variable-temperature cryostat.All the neutron powder diffraction profiles collected in this program of experiments final agreement factors for the neutron profile used in the simultaneous refinement were Rwp=8.10; Rp=5.91; Rl=5.62; were analysed by the Rietveld profile analysis technique10 using the GSAS suite of programs.11 The data were not DW-d 0.217; the value of xred2 based on both data sets was 4.55.corrected for absorption. The following neutron scattering lengths were used: bSr=0.702, bFe=0.954, bTa=0.Sb= A survey of Sr2FeTaO6 at 1.5 K on D1b suggested that there were no magnetic Bragg peaks in the low-angle region of the 0.564 and bO=0.5805×10-14 m. The background was fitted using a shifted Chebyshev polynomial, and the shape of the diffraction pattern, and this was confirmed by a high-resolution data set collected on D2b at 1.5 K.All the observed Bragg Bragg peaks was described by a pseudo-Voigt function. peaks could again be accounted for by an orthorhombic GdFeO3 structure. The results of a profile analysis involving Results a total of 27 variables, including 11 atomic parameters, 5 backgroundparameters and 4 peak-shape parameters are given Sr2FeTaO6 in Table 3.Fig. 1 shows the final observed and calculated Inspection of a high-resolution neutron powder diffraction diffraction patterns. The resulting agreement parameters were pattern collected from Sr2FeTaO6 at 280 K on D2b showed as follows: Rwp=8.78; Rp=6.59; Rl=3.21; DW-d 0.210 (lower that, contrary to previous reports,9 this compound is not a limit of 90% confidence level=1.901 13), xred2=9.06.simple cubic perovskite at room temperature. The presence of additional Bragg peaks throughout the measured angular Sr2FeSbO6 range suggested a larger unit cell, and the diffraction pattern was indexed in the orthorhombic space group Pbnm with unit- The neutron powder diffraction profile of Sr2FeSbO6 collected on D2b at 290 K was consistent with the monoclinic space cell parameters ca.Ó2 ap×ca.Ó2 ap×ca.2 ap.This setting, often referred to as the GdFeO3 structure type,12 is consistent with our X-ray data in that it requires a disordered arrange- Table 3 Structural parameters of Sr2FeTaO6 at 1.5 K (space group ment of Fe3+ and Ta5+ cations over a single crystallographic Pbnm) six-coordinate site. The refined values of the transformed unitatom site x y z Uiso/A° 2 cell parameters confirmed that the magnitude of the orthorhombic strain was very small, and, because of this, there was Sr 4c 0.0008(9) 0.0097(8) 1/4 0.0067(3) a high correlation between certain structural parameters in Fe/Ta 4a 1/2 0 0 0.0028(2) our refinements.The agreement parameters resulting from the O(1) 8d 0.2417(9) 0.244(1) 0.0201(3) 0.0077(4) analysis of our neutron diffraction data [Rwp=9.07; Rp=6.59; O(2) 4c 0.9526(6) 0.505(1) 1/4 0.0045(6) Rl=5.55; DW-d 0.179 (lower limit of 90% confidence level= 1.89913) xrec2=11.66] were all higher than might have been a=5.6153(2) A° ; b=5.6039(2) A° ; c=7.9124(3) A°; V=248.98(2) A° 3 .expected and the standard deviations on many of the atomic parameters were large.Furthermore, it was necessary to work with an overall temperature factor in order to stabilize the refinement, an unusual and undesirable occurrence in analysing neutron powder diffraction data collected at 280 K. An improved description of the crystal structure was achieved Table 1 Structural parameters of Sr2FeTaO6 at 280 K (space group Pbnm) atom site x y z Uiso/A°2 Sr 4c 0.0004(1) 0.0026(5) 1/4 0.0122(2) Fe/Ta 4a 1/2 0 0 0.0048(2) O(1) 8d 0.2530(9) 0.2522(9) 0.0168(3) 0.0109(3) Fig. 1 Observed (dots), calculated (full line) and difference neutron O(2) 4c 0.9569(3) 0.5160(3) 1/4 0.0079(5) powder diffraction patterns of Sr2FeTaO6 at 1.5 K (D2b data). Reflection positions are marked. a=5.6204(3) A° ; b=5.6161(3) A° ; c=7.9266(3) A° ; V=250.20(2) A° 3.Table 2 Selected bond distances (A° ) and angles (degrees) in Sr2FeTaO6 at 280 K SrMO(1) 2.705(5) (2×) SrMO(1) 2.881(5) (2×) SrMO(1) 2.720(5) (2×) SrMO(1) 2.924(5) (2×) SrMO(2) 2.571(6) (1×) SrMO(2) 2.744(6) (1×) SrMO(2) 2.894(6) (1×) SrMO(2) 3.051(6) (1×) Fe/TaMO(1) 1.988(5) (2×) Fe/TaMO(2) 1.9984(4) (2×) Fe/TaMO(1) 1.994(5) (2×) shortest OMO 2.749(4) O(1)MFe/TaMO(1) 90.30(1) Fe/TaMO(1)MFe/Ta 172.26(13) O(1)MFe/TaMO(2) 92.97(13) Fe/TaMO(2)MFe/Ta 165.13(18) O(1)MFe/TaMO(2) 90.77(13) 460 J. Mater.Chem., 1997, 7(3), 459–463group P21/n which permits (partial) ordering of Fe3+ and magnetic ordering (Fig. 3). A magnetic structure of this kind can be considered to consist of an antiferromagnetic stacking Sb5+ over the octahedrally coordinated cation sites.During the course of the structure refinement, the Fe3+5Sb5+ distri- of ferromagnetic cation sheets along the z axis of the unit cell. In order to elucidate the details of the magnetic structure, we bution over these sites was refined within the constraints that the sites remained fully occupied and that the overall performed a Rietveld analysis using the free-ion form factor14 of Fe3+ to describe the angular dependence of the magnetic Fe3+5Sb5+ ratio remained at 151.The isotropic temperature factors of these two sites were constrained to be equal, as were scattering amplitude. There was no evidence of any major change in the crystal structure between 290 and 1.5 K, and, in those associated with the three crystallographically distinct oxygen atoms.This is another pseudo-symmetric structure, view of the limited number of Bragg peaks in the D1b data set, the atomic parameters were held constant at the values and we again chose to take advantage of the complementarity of neutron and X-ray diffraction by performing a simultaneous determined at 290 K. Refinement of the appropriate profile parameters and the cation magnetic moment resulted in aver- refinement of two data sets.In the last cycles of refinement, 48 variables (including 16 atomic parameters, 4 lattice parameters age values for the latter of 3.06(9) mB per Fe3+ on the B(1) site (79.5% Fe3+) and 0.0(3) mB per Fe3+ on the B(2) site (20.5% and 2 sets of profile parameters) were refined. The final agreement factors for the neutron data were as follows: Rwp= Fe3+).The ordered moment was constrained to lie along the x axis, but alignment along y is equally likely given the 5.28; Rp=4.10; Rl=6.27; DW-d 0.587 (lower limit of 90% confidence level=1.90513); the value of xred2 based on both resolution of our data. The magnitudes of the magnetic moments are correlated with the chosen form factor, and may data sets was 1.56.The values of the refined parameters are given in Table 4, with the corresponding bond lengths and thus be in error by more than the statistical error quoted. Fig. 4 shows the final observed and calculated diffraction angles in Table 5. Despite the use of the simultaneous refinement technique, the pseudo-symmetry present in the structure patterns at 1.5 K. The broad feature in the range 70< 2h/degrees<75 which the model does not account for is resulted in relatively large standard deviations on the bond lengths.The observed and calculated neutron diffraction patterns are shown in Fig. 2. The diffraction data collected on the diffractometer D1b at 1.5 K contained additional Bragg reflections indicative of the presence of long-range, type I antiferro- Table 4 Structural parameters of Sr2FeSbO6 at 290 K (space group P21/n) atom site x y z Uiso/A°2 Sr 4e -0.0007(8) 0.0019(8) 0.252(1) 0.0092(2) B(1) 2d 1/2 0 0 0.0039(2) B(2) 2c 0 1/2 0 0.0039(2) O(1) 4e 0.249(2) 0.251(2) 0.019(4) 0.0096(2) O(2) 4e 0.259(2) 0.254(2) 0.480(4) 0.0096(2) O(3) 4e 0.9578(6) 0.515(1) 0.250(2) 0.0096(2) Fig. 3 The type I magnetic structure proposed for Sr2FeSbO6.Only Fractional occupancies on B(1): 0.795(6) Fe, 0.205(6) Sb; on the octahedrally coordinated cations are drawn: open circles B(1), B(2): 0.205(6) Fe, 0.795(6) Sb. a=5.6132(5) A° ; b=5.5973(5) A° ; shaded circles B(2). c=7.9036(7) A° ; b=90.01(1); V=248.32(6) A° 3 . Fig. 4 Observed (dots), calculated (full line) and difference neutron Fig. 2 Observed (dots), calculated (full line) and difference neutron powder diffraction patterns of Sr2FeSbO6 at 290 K (D2b data).powder diffraction patterns of Sr2FeSbO6 at 1.5 K (D1b data). Reflection positions are marked. Reflection positions are marked. Table 5 Selected bond distances (A° ) and angles (degrees) of Sr2FeSbO6 at 290 K SrMO(1) 2.697(27) SrMO(2) 2.669(25) SrMO(1) 2.699(25) SrMO(2) 2.714(27) SrMO(1) 2.891(27) SrMO(2) 2.881(28) SrMO(1) 2.916(27) SrMO(2) 2.940(28) SrMO(3) 2.566(5) SrMO(3) 2.880(9) SrMO(3) 2.737(9) SrMO(3) 3.049(5) B(1)MO(1) 1.995(10) (2×) B(2)MO(1) 1.979(9) (2×) B(1)MO(2) 2.006(9) (2×) B(2)MO(2) 1.972(9) (2×) B(1)MO(3) 1.994(14) (2×) B(2)MO(3) 1.989(14) (2×) B(1)MO(1)MB(2) 171(2) B(1)MO(3)MB(2) 165.5(2) B(1)MO(2)MB(2) 170(2) J. Mater.Chem., 1997, 7(3), 459–463 461one crystallographic site demonstrates that their cations do not differ sufficiently in size or charge for an ordered arrangement to be significantly more stable.The situation is different in Sr2FeSbO6 , where ordering is present to a considerable extent, presumably because of the smaller size of Sb5+. This reduction is apparent in the bond lengths listed in Table 5. The actual degree of ordering calculated in the simultaneous refinement of neutron and X-ray diffraction data [0.795(2)] is in excellent agreement with that deduced previously8 from Mo�ssbauer data [0.78(2)]. The ordering of the cations over the six-coordinate B sites, and the consequent displacements of the oxide ions, lowers the symmetry of the structure to monoclinic, but Sr2FeSbO6 can still be thought of as a pseudo-cubic perovskite. We previously described Sr2FeTaO6 as a spin glass on the basis of susceptibility and Mo�ssbauer data.The absence of magnetic Bragg peaks in the low-temperature neutron diffrac- Fig. 5 Average magnetic moment of Fe3+ on the B(1) site in tion pattern proves that there is no long-range magnetic order Sr2FeSbO6 as a function of the temperature in this compound below the susceptibility maximum at 23 K, despite the presence of a hyperfine field in the Mo�ssbauer data. This confirms that the magnetic moments of the Fe3+ cations are disordered, despite being static (on the experimental timescale), and justifies the classification of this compound as a spin glass.The origin of this behaviour is still not clear.The transition-metal sublattice in the perovskite structure is not frustrated if only nearest-neighbour (NN) interactions are considered. One possibility is that frustration is introduced by competition between NN and next-nearest-neighbour (NNN) interactions. The latter will be weaker than the former, but they are more numerous (1256). However, estimates of the ratio of the exchange constants, JNNN/JNN, in perovskite fluorides are only of the order 1/200,16 and even allowing for the substantial increase in covalent overlap in a corresponding oxide, this argument is not entirely convincing.It is also possible that the answer lies in the presence of antiferromag- Fig. 6 Neutron diffraction patterns (D1b) of Sr2FeSbO6 as a function netic domains which are small when measured on the length- of temperature.Magnetic reflections are labelled ‘M’. scale of a diffraction experiment. We are presently carrying out experiments and calculations to explore this possibility further. thought to be instrumental in origin; it was present in all the profiles collected on D1b, but absent from the data collected The sensitivity of the magnetic properties to the degree of (chemical) cation ordering on the six-coordinate sites is demon- on the same sample on D2b.Diffraction patterns collected during computer-controlled warming of the sample in the strated by the markedly different magnetic behaviour of Sr2FeSbO6. The susceptibility and Mo�ssbauer data8 enabled temperature range 1.5T /K100 indicated that the magnetic structure remained type I until long-range magnetic order was us to determine that magnetic frustration was present in this compound, but the presence of partial cation ordering led us lost altogether.These data were therefore analysed in a similar manner to that described above, although the ordered magnetic to draw parallels with SrLaNiSbO6,17 and we suggested that a backbone of magnetically ordered spins might establish itself moment of the cations on the B(2) site was fixed at zero. In all cases the agreement factor Rwp fell to less than 4%.The in Sr2FeSbO6, with clusters of frustrated spins, not connected to the backbone, also being present. The results of our neutron temperature dependence of the resulting average moments on the B(1) site is plotted in Fig. 5. The value of the ordered diffraction experiments support this explanation. The refined value of the average ordered magnetic moment per Fe3+ cation moment decreases smoothly in the temperature range 1.5T /K37(2), above which it takes an approximately [3.06(9) mB] on the B(1) site is considerably lower than that observed in an ideal antiferromagnetic oxide of Fe3+ constant value which is within 3s of zero.Visual inspection of the diffraction patterns presented in Fig. 6 suggests that the (ca. 4.4 mB18), even after allowance has been made for possible errors in the form factor, thus suggesting that only ca. 70% of non-zero value refined at T39 K results from an attempt to fit background noise, thus defining the sensitivity limit of our the Fe3+ cations on these sites belong to the magnetic backbone. Alternatively, a higher percentage could be coupled, but measurements at ca. 1 mB . with an imperfect alignment of the spin vectors. A third explanation is that the antiferromagnetic domain size is becom- Discussion ing small enough to reduce the observed moment. In the first two cases an unaligned spin component might contribute to Neutron diffraction has shown that our previous description9 of Sr2FeTaO6 as a simple cubic perovskite was wrong.The the hysteresis observed in the magnetic susceptibility, while domain growth in a field would produce the same effect. The symmetry was previously overestimated because the deviation from high symmetry is caused almost entirely by a relatively adoption of a type I magnetic structure implies that the most significant antiferromagnetic superexchange coupling takes small displacement of oxide ions from their ideal positions, and X-rays are therefore insensitive to the change. The revised place between cations separated by Ó2a0; that is, between NNN transition-metal sites.The magnetic structure is thus crystal structure described above is common among perovskites, including those which contain more than one transition- consistent with the crystal structure in that Ó2a0 is the distance between pairs of B(1) sites, which are both likely (63%) to be metal species,15 and the bond lengths in Table 2 are unexceptional.The disordered distribution of Fe and Ta atoms over occupied by Fe; the B(1)MB(2) distance is shorter (a0), but 462 J.Mater. Chem., 1997, 7(3), 459–463the probability of adjacent B(1) and B(2) sites both being Sr2FeTaO6 and Sr2FeSbO6 to the differing degrees of structural order on their transition-metal sublattices, an observation that occupied by Fe is only 16%. In other words, each Fe3+ cation demonstrates once again how structural chemistry can control on a B(1) site is likely to be surrounded by (12×0.795) Fe3+ solid-state physics.cations on NNN sites, but only (6×0.205) Fe3+ cations on NN sites. Our neutron diffraction results show that the concen- We are grateful to J. P. Hodges, P. G. Radaelli and B. Ouladdiaf tration (20%) of Fe atoms present on the B(2) sites is too low for experimental assistance, and to EPSRC for financial for them to take part in any long-range magnetic ordering.support. However, the Mo�ssbauer data do not show any evidence for a paramagnetic component at 4.2 K. It is not uncommon for frustrated spin systems to show spin freezing at low tempera- References tures and the apparent lack of a finite moment on the B(2) 1 S. H. Kim and P. D. Battle, J. Solid State Chem., 1995, 114, 174. site prompts us to assume that these atoms are indeed frus- 2 M.F. Kupriyanov and E. G. Fesenko, Sov. Phys. Crystallogr., trated and contribute to the spin-glass component which the 1962, 6, 639. susceptibility data show to be present in the sample (a true 3 M. F. Sykes and J. W. Essam, Phys. Rev. A, 1964, 133, 310. antiferromagnet cannot show the observed divergence of the 4 R. Rodriguez, A. Fernandez, A. Isalgue, J.Rodriguez, A. Labarta, J. Tejada and X. Obradors, J. Phys. C: Solid State Phys., 1985, FC and ZFC susceptibilities). The temperature dependence of 18, L401. the magnetic Bragg (Fig. 6) shows that the long- 5 W. C. Koehler and E. O. Wollan, J. Phys. Chem. Solids, 1957, range magnetic ordering is lost at 37(2) K. Consideration of 2, 100. these data along with the magnetic susceptibility data8 then 6 D.J. Breed, K. Gilijamse, J. W. E. Sterkenburg and A. R. Miedema, shows that the glass-transition temperature and the Ne�el point J. Appl. Phys., 1970, 41, 1267. 7 T. C. Gibb, J.Mater. Chem., 1993, 3, 441. are coincident, within experimental error. 8 P. D. Battle, T. C. Gibb, A. J. Herold and J. P. Hodges, J. Mater. The neutron diffraction experiments described above, con- Chem., 1995, 5, 75.sidered in conjunction with Mo�ssbauer and susceptibility data, 9 P. D. Battle, T. C. Gibb, A. J. Herod, S-H. Kim and P. H. Munns, have allowed us to build up a self-consistent description of the J. Mater. Chem., 1995, 5, 865. magnetic properties of Sr2FeTaO6 and Sr2FeSbO6. The former 10 H. M. Rietveld, J. Appl. Crystallogr., 1969, 2, 65. is a spin glass, with no long-range magnetic order present 11 A. C. Larson and R. B. von Dreele, General Structure Analysis System (GSAS), Los Alamos National Laboratories, Report above 1.5 K, whereas an ordered backbone of spins and a LAUR 86-748, 1990. spin-glass fraction coexist in the latter. In both cases there is 12 S. Geller, J. Chem. Phys., 1956, 24, 1236. evidence to suggest that the magnetic interactions between 13 R. J. Hill and H. D. Flack, J. Appl. Crystallogr., 1987, 20, 356. cations on NNN sites may be strong enough to influence the 14 R. E.Watson and A. J. Freeman, Acta Crystallogr., 1961, 14, 27. properties of the compound, an observation which urges 15 P. D. Battle and C. W. Jones,Mater. Res. Bull., 1987, 22, 1623. 16 L. J. D. Jongh and A. R. Miedema, Adv. Phys., 1974, 23, 1. caution in the use of models which rely on NN interactions to 17 M. P. Attfield, P. D. Battle, S. K. Bollen, T. C. Gibb and explain the properties of simple perovskite structures. However, R. J. Whitehead, J. Solid State Chem., 1992, 100, 37. we emphasise that the importance of NNN interactions is not 18 B. C. Tofield and B. E. F. Fender, J. Phys. Chem. Solids, 1970, proven in Sr2FeTaO6 , and that alternative explanations, for 31, 2741. example the presence of small antiferromagnetic domains, must be explored. We ascribe the variation in behaviour between Paper 6/07083C; Received 17th October, 1996 J. Mater. Chem., 1997, 7(3), 459–463

ISSN:0959-9428    

DOI:10.1039/a607083c

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

Electrochemical solid–solid conversion of bismuth oxide to bismuth metal Oliver Pyper, Brigitte Hahn and Robert Scho� llhorn Institut fu� r Anorganische und Analytische Chemie, T echnische Universita�t Berlin, Strasse des 17. Juli 135, D-10623 Berlin, Germany It is shown that binary and ternary bismuth oxides can be reduced quantitatively in aqueous electrolytes at ambient temperature via a two-phase solid–solid transition by electron/proton transfer to metastable porous metal and alloy systems.Single-crystal studies demonstrate the pseudomorphous character of the transition; topochemical correlations between educt and product phase could not be established, however. The solid-state reaction may proceed even at rather low temperatures of -40°C. In a recent publication a concept has been discussed that For structural investigations by powder X-ray diffractometry concerns the controlled formation of metastable porous homo- the Guinier technique and a powder diffractometer (Siemens geneous and heterogeneous systems by the electrochemical D5000, linear counter) were used with Cu-Ka radiation.reduction of transition-metal oxides, chalcogenides and halides Further characterization was achieved by transmission electron to the corresponding metal systems via electron/proton transfer microscopy (JEOL JSEM 200 B) and scanning electron at ambient temperature in a solid–solid conversion.1 The basic microscopy with an EDX analytical probe (Hitachi Svalidity of the reaction scheme has been demonstrated with 2700/Kevex).Magnetic susceptibility data were obtained using copper compounds as the example.Prerequisites for this type a Faraday balance. of reaction are a low enthalpy of formation of the metal compound and a thermodynamically or kinetically accessible redox potential of the metal cation–metal couple. Many oxides Results and Discussion of the p-block main-group metals are known to exhibit moder- Reduction of bismuth oxide ate enthalpies of formation, e.g.the oxides of the heavier elements thallium, lead and bismuth. We have performed a Preliminary investigations on the cathodic reduction of sintered study on the reactivity of bismuth(III ) oxide, which has been Bi2O3 pellets in aqueous neutral electrolyte showed the forma- proposed in earlier investigations as electrode material in tion of a metallic grey zone around the platinum contact site primary batteries with aprotic electrolytes.2–4 We report here which spread rapidly across the entire pellet surface; the end on the solid–solid conversion of Bi2O3 in aqueous electrolytes point of the reduction was indicated by the sudden formation and on similar reactions of some related binary bismuth of hydrogen gas.Analytical investigations confirmed the quan- compounds. Since the potential formation of new metastable titative conversion of the oxide to bismuth metal according alloys is a principal aspect in our investigations of these solid– to eqn. (1). solid reactions we also included studies on appropriate ternary bismuth phases. Bi2O3+6H++6e-�2Bi0+3H2O (1) The potential vs.charge transfer diagram for the galvanostatic Experimental cathodic reduction of Bi2O3 working electrodes in 0.1 mol dm-3 aqueous KOH in air atmosphere is given in Fig. 1. The Bismuth oxide, Bi2O3, was prepared by sintering pressed pellets initial strong overpotential is due to the small original reaction (diameter 12 mm, thickness ca. 1 mm) of the carbonate zone at the platinum point contact.Bi2O3 is a wide bandgap (BiO)2CO3 for 20 h at 750 °C. Ternary bismuth phases with semiconductor and the reaction can start only at the small Cu, Pb, W, Yb were obtained by sintering of the binary oxides metalliclead (Pt)/Bi2O3/electrolyte triphase boundary. Further in the appropriate stoichiometric ratio according to the progress of the reaction occurs subsequently at the bismuth literature: Bi2CuO4,5 Pb2Bi6O11,6 (Bi2O3)0.75(WO3)0.25,7 metal/Bi2O3 moving boundary zone.Since the Bi2O3/Bi Bi1.714Yb0.286O3.8,9 The oxyhalides BiOCl,10 BiOl11 and the interface area increases rapidly in the initial reaction period oxyacetate BiO(CH3CO2)12 were prepared by crystallization the overvoltage decreases strongly and is followed by a poten- from aqueous solutions.Large natural single crystals were tial plateau. Towards the end of the reaction the interface used in the case of bismuth sulfide, Bi2S3. region diminishes again which results in an increasing overpot- Electrochemical reactions were performed in three-electrode ential. The end of the solid–solid conversion is indicated by a cells with commercial potentiostat equipment, a Hg/HgO potential step and the formation of molecular hydrogen at the reference electrode and 0.1 mol dm-3 KOH as the standard working electrode.If the reduction is carried out in contact electrolyte. Working electrodes were sintered pellets, contacted with air the experimental charge transfer observed (as deter- by platinum clamps, or, in the case of oxyhalides and bismuth mined by the potential step) is usually somewhat higher oxyacetate, in platinum grid-pressed powder electrodes with 1 compared to the calculated value of 6 e- per Bi2O3.Under mass% PTFE. Bismuth sulfide crystals were contacted by platinum clamps. an inert gas atmosphere (nitrogen, argon) charge-transfer data 465 J. Mater. Chem., 1997, 7(3), 465–469466 Fig. 1 Potential vs. charge transfer diagram for the cathodic reduction Fig. 3 Low-temperature galvanostatic reduction of Bi2O3 at T= of Bi2O3 working electrodes in 0.1 mol dm-3 KOH at different current -40°C; (electrolyte 7 mol dm-3 aqueous KOH; sintered pellet values (300 K); #, air; &, N2 atmosphere working electrode, mass 392 mg, current 5 mA) close to the theoretical value are found owing to the absence The potential plateau observed in the galvanostatic of oxygen.The influence of the current density is illustrated in reduction of Bi2O3 to Bi suggests a two-phase process, in Fig. 2. In contact with air low current densities (i.e. long agreement with the reaction assumed by eqn. (1). This is reaction times) lead to strongly increased values for the charge confirmed by the X-ray diffraction data given in Fig. 4, which transfer due to the partial oxidation of Bi at the cathode by confirm the coexistence of the oxide educt Bi2O3 and the metal O2. High current densities result in a strong overpotential product Bi; no intermediate phase can be detected. leading to slow discharge of hydrogen as a competing process. In order to determine the reduction potential in potentio- No significant influence of the variation of the electrolyte static mode a step-screening experiment was performed (poten- concentration (0.1–10 mol dm-3) and of the electrolyte cation tial step chronocoulometry).At a potential of -750 mV vs. (Li, Na, K) on the charge transfer could be observed. the Hg/HgO reference electrode, a charge transfer of 5.9 e- We were interested to find out whether this solid-state f.u.-1 could be measured.The current vs. time diagram is given reaction would proceed also at temperatures below 25°C with reasonable kinetics. The experiment was carried out in a cryostat at -40 °C with concentrated aqueous KOH (7 mol dm-3) as the electrolyte in order to avoid solidification. The galvanostatic curve is given in Fig. 3. The initial region is characterized by a strong overpotential followed by a potential plateau.The sharp potential step indicates a charge transfer of 6 e- f.u.-1 (f.u.=formula unit) in very good agreement with the calculated value. It is surprising that this non-topotactic solid-state reaction can proceed with reasonable reaction rates and quantitative conversion at rather low temperatures.Fig. 4 Galvanostatic reduction of Bi2O3. Upper section: change of Fig. 2 Cathodic reduction of Bi2O3 in aqueous KOH: dependence relative intensities of selected educt and product reflections with the charge transfer; lower section: observed lattice parameters of the of the nominal charge-transfer values upon the cell current; #, air; &, N2 atmosphere coexisting phases vs.charge transfer.J . Mate r . Chem., 1997, 7(3), 465&ndash the reduced samples were determined by dc susceptibility measurements in the temperature range 200–300 K; in agreement with literature data, diamagnetism was observed. Reduction of single-crystal material For a demonstration of the pseudomorphous character of the solid-state transformation under investigation and for the control of potential structural correlations between educt and product, single-crystal material is required.Bi2O3 crystals were not available but single crystals of bismuth oxychloride, BiOCl, and of bismuth sulfide, Bi2S3, could be obtained. We were able to show that both phases can be reduced quantitatively to bismuth metal according to Fig. 6 X-Ray diffractogram of bismuth metal obtained by cathodic BiOCl+H2O+3e-�Bi0+2OH-+Cl- (2) reduction of Bi2O3 at -40 °C (cf.data given in Fig. 3) Bi2S3+3H2O+6e-�2Bi0+3OH-+3SH- (3) in Fig. 5. The shape of the curve can be explained by the BiOCl single crystals were obtained as platelets with quad- Bi2O3/Bi interface area change as discussed above; irregularit- ratic shape. The crystals were rather small (edge length ies are assumed to be due to crack regions in the sintered 5–10 mm) and very thin, which turned out to be most favourable sample.The quantitative charge transfer at a defined potential for TEM investigations. Fig. 7 shows a TEM image, Fig. 8 an suggests again the conclusion that no potential intermediate electron diffraction image of the single crystal used. The phases appear.electrochemical reduction was performed in aqueous suspen- The product obtained by cathodic reduction is brittle and sion in platinum vessels. TEMimages after reductionconfirmed exhibits a metallic grey colour. After washing in water and that the morphology of the platelets was retained. Dark-field acetone and drying in vacuum the electrode mass was identical images displayed a system of interconnected metal clusters to the calculated value within <1%; i.e.no residual electrolyte with an average particle diameter of 500–1000 A° . Electron or water was retained in the pore system. The lattice parameters diffraction studies showed that all lines of the pattern obtained of the products prepared at 20°C and -40°C were close to could be attributed to crystalline bismuth metal (Fig. 9). There those reported in the literature. For samples prepared at was no indication, however, for a preferential orientation of ambient temperature the increase in Bragg linewidth was the metal particles identified. rather small as compared to well ordered polycrystalline In the case of the sulfide, Bi2S3, large metallic grey crystal bismuth metal. From particle size calculations13 it can be assumed that the primary particles have a nominal diameter >500 A° .The linewidth broading increases with increasing diffraction angle, which suggests specificlattice defects.Bismuth metal prepared from Bi2O3 at -40°C showed, as expected, rather broad reflections, however, indicating strong lattice disorder (Fig. 6). A calculation of the nominal particle size (domains with coherent diffraction) yielded a value of<400 A° .SEM images showed crack regions at the surface (10 mm range); the resolution limit of the instrument used did not allow detection of the pore structure. Magnetic properties of Fig. 5 Potential step chronocoulometry of a Bi2O3 electrode: current Fig. 7 TEM image of two BiOCl crystals. Dimensions of the smaller vs. time diagram for the reaction at-750 mV (0.1 mol dm-3 aqueous KOH, pressed sintered pellets of mass 100 mg).Area=-37.2 mA h crystal ca. 5×5 mm.468 Fig. 9 Electron diffraction pattern of bismuth metal obtained by Fig. 8 Electron diffraction patternof the BiOCl crystals shown in Fig. 7 cathodic reduction of BiOCl. Beam diameter ca. 0.5 mm. agglomerates from a natural source (bismuthinite, Namibia, transfer curve for the reduction of sintered Bi2CuO4 working Helikon II, East) were available.It was possible to isolate electrodes in 0.1 mol dm-3 KOH under an inert gas atmos- needles with 0.3×0.3×0.8 mm3 size by cleaving. Galvanostatic phere is given in Fig. 10. The electrochemical charge-transfer reduction yielded a charge transfer equivalent to 5.9 e- f.u.-1; value of 8.2 e- f.u.-1 correspond reasonably well with the X-ray data confirmed that the product was bismuth metal.value calculated according to eqn. (4). EDX–SEM studies demonstrated that, independent of the location of the current lead contact point, the reaction started Bi2CuO4+4H2O+8e-�(Bi2/Cu)+8(OH)- (4) synchronously at all crystal edges and at local crystal defects, The X-ray diagram of the product exhibits a strong back- which is a consequence of the high conductivity of the narrow ground with broadened reflections corresponding to bismuth bandgap semiconductor sulfide.X-Ray rotation photographs metal while no other lines, in particular no reflections that can of Bi2S3 crystal needles (c axis perpendicular to X-ray beam, be attributed to copper metal, can be detected. This may be observation of 0k0 reflections) were made before and after understood in terms of a mixture of an amorphous metastable reduction.In the latter case only faintly structured closed copper–bismuth alloy along with a fraction of crystalline diffraction cone cross-sections could be observed. It must be bismuth metal. An alternative interpretation of the X-ray data concluded again that there is no correlation between the initial could be based, however, on a model that involves microdo- orientation of the Bi2S3 crystal and the orientation of the metal mains of copper metal, too small to give appreciable diffraction crystallites present after reduction, although the transition is intensity, in a bismuth matrix.EXAFS measurements are in clearly pseudomorphous. progress to decide between the two models. If this material is heated to 650 °C for 12 h in closed evacuated quartz ampoules Reduction of ternary bismuth oxides then X-ray data reveal the presence of the diffraction lines of bismuth as well as of copper metal (along with some weak The principal idea here was to prove the possibility of the preparation of a metastable porous alloy and heterogeneous reflections belonging to Bi2O3, indicating surface oxidation of the product).Similar results have been obtained in our earlier phases via low-temperature electrochemical conversion of appropriate oxide systems;1 an overview is given in Table 1. investigations on the electrochemical reduction of superconducting Bi2Sr2CaCu2O8+x.15 Bismuth and copper are not miscible in the solid state,14 the reduction of bismuth oxocuprate, Bi2CuO4, at 300 K could Yellow–orange sintered pellets of the ternary bismuth–lead oxide Pb2Bi6O11 could be reduced electrochemically to a grey therefore be expected to lead to the formation of a metastable amorphous alloy since the related binary oxides Bi2O3 and product with a metallic appearance; the charge transfer value of 21.9 e- f.u.-1 found experimentally is in agreement with the CuO can both be reduced to metal.The potential vs. chargeJ . Mate r . Chem., 1997, 7(3), 465–469 469 Table 1 List of bismuth compounds investigated material prep.a symmetry ncalc.b nobs.b EDc/mV Bi2O3 (BiO)2CO3 (750°C) monoclinic 6 6.2 (20°C) -755 6.0 (-40 °C) Bi2S3 mineral (xx) orthorhombic 6 5.9 -1050 BiOCl ref. 10 (xx) tetragonal 3 3.15 -750 BiOI ref. 11 tetragonal 3 3.15 -660 BiO(ac)d ref. 12 tetragonal 3 3.15 -720 Bi2CuO4 ref. 5 tetragonal 6 (Bi)+2 (Pb) 8.2 -750 Pb2Bi6O11 ref. 6 monoclinic 4 (Pb)+18 (Bi) 21.9 -750/-1050 Bi14W2O27 ref. 7, 18 tetragonal 42 (Bi) 43 -1100 Bi1.72Yb0.28O3 ref. 9 cubic 5.14 (Bi) 5.3 -720 axx=Single crystal. bn=Charge transfer (e- f.u.-1).cED=Potential for reduction vs. Hg/HgO reference electrode measured under current. dac=CH3CO2. ytterbium oxide should remain as dispersed phases in the metal matrix. X-Ray investigations of sintered pellets of the yellow–green bismuth–tungsten oxide found mainly single-phase Bi14W2O27 (tetragonal).18 They could be reduced galvanostatically with a charge-transfer value of 43 e- f.u.-1 as compared with the calculated value of 42 e- f.u.ed on eqn. 6. Bi14W2O27+42H2O+42e-�14Bi+2WO3+42OH- (6) The X-ray diffractogram of the product exhibit only the lines for bismuth metal except for a few additional reflections that could not be indexed reliably owing to the low line intensities. Obviously tungsten oxide remains as an amorphous oxide–hydrate fraction in the product.Bi1.72Yb0.28O3 can be considered as a rare-earth-metal stabilized d-Bi2O3. The X-ray diagram of the reduction product shows the lines for bismuth metal along with a strong back- Fig. 10 Potential vs. charge transfer diagram for the cathodic reduction ground in the lower 2h range. Amorphous ytterbium oxide– of Bi2CuO4 (sintered pellet contacted with platinium clamp, mass hydroxide is assumed to be dispersed in the metal matrix. 650 mg, current 3 mA, 0.1 mol dm-3 aqueous KOH as electrolyte, N2 An attempt was made to reduce a bismuth compound with atmosphere) rather large anions (i.e. a strong spatial ‘dilution’ of Bi3+) in order to find out whether the line width and particle size of calculated value of 22 e- if both lead and bismuth are reduced the product would increase and decrease respectively.The to the metallic state [eqn (5)]. compound selected was bismuth oxyacetate; the X-ray data of the reduction product did not however exhibit line broadening Pb2Bi6O11+22H2O+22e-�(Pb2/Bi6)+22OH- (5) superior to that found for the cathodic reduction of bismuth Upon exposure to air the grey material changes to grey– oxide.black over a few hours owing to oxidation processes. The Pb/Bi phase diagram exhibits only one intermediate e- Reoxidation reactions phase around the composition Pb7Bi3.16 X-Ray diagrams of Samples prepared by reduction of Bi2O3 undergo slow visible the reduction product indicate the presence of bismuth metal oxidation upon storage in air at 300 K: after several days and Pb7Bi317 as the major crystalline components (along with changes from metallic grey to golden yellow and blue–black traces of Pb3O4, probably due to oxidation).Only two weak could be observed while no measurable change in mass or X- lines corresponding to Pb metal could be detected. After ray diffraction pattern was detected. Upon electrochemical exposure of reduced samples to air for 3 weeks at 300 K oxidation of bismuth metal (prepared from Bi2O3) similar continued oxidation takes place: besides the reflection of changes in colour are observed.If samples are first oxidized bismuth metal and Bi2O3 a series of lines corresponding to a- until the potential of O2 gas formation is reached and sub- PbO were found, while the lines associated with Pb7Bi3 and sequently reduced, then a charge transfer of ca. 0.3 e- f.u.-1 Pb had virtually disappeared. We conclude that the freshly at -740 mV can be measured. Quantitative reoxidation is reduced product does contain also a component of amorphous thus not possible; the high stability of thin oxide layers on lead–bismuth alloy. bulk bismuth metal upon electrochemical oxidation has been For studies on the preparation of heterogeneous metal–non- described earlier.19 reducible oxide porous systems the two ternary phases (Bi2O3)0.75(WO3)0.25 (described as cubic)7 and Bi1.72Yb0.28O38,9 (cubic) were selected (Table 1).We assumed that in both cases Conclusions only Bi3+ could be reduced to the metal while tungsten and The processes investigated belong to an interesting specific type of low-temperature solid-state reaction. The exact mor-470 6 JCPDS, International Centre for Diffraction Data, Newtown phology of the microscopic pore structure has not been estab- Square, PA, 1990, PDF card no. 41-0404. lished so far, it may depend on several parameters.1 The data 7 Y. J. Lee, C. O. Park, H. D. Baek and J. S. Hwang, Solid State of the present study on bismuth compounds favour the model Ionics, 1995, 76, 1.of a matrix system built up from approximately isometric 8 H. Iwahara, T. Esaka, T. Sato and T. Takahashi, J. Solid State metal cluster units rather than a sponge-like structure similar Chem., 1981, 39, 173. 9 JCPDS, International Centre For Diffraction Data, Newtown to, e.g., reticulated glassy carbon materials. The results of a Square, PA, 1990, PDF card no. 41-0288. preliminary scanning tunnelling electron microscopy study on 10 G. Brauer, Handbuch der pra� parativen anorganischen Chemie, the reduction of a copper oxide single crystal also appear to Ferdinand Enke Verlag, Stuttgart, 1975, p. 596. support the cluster model;20 it seems problematic, however, to 11 Gmelin, Handbuch der anorganischen Chemie, vol. 19 (Wismut), conclude from structures found in the surface region on the VCH, Weinheim, 1964, pp. 722ff. 12 B. Aurivillius, Acta Chem. Scand., 1955, 9, 1213. architecture of the bulk material. X-Ray diffraction, electron 13 H. P. Klug and L. E. Alexander, X-ray diffraction procedures, diffraction and TEM studies do not exhibit texture effects, i.e. Wiley, New York, 1974, 2nd edn., p. 695. there seems to be no measurable correlation between the 14 M. Hansen, Constitution of binary alloys, McGraw-Hill,New York, starting lattice and the product phase. 1958, p. 308. 15 R. Bezzenberger, E. Gocke and R. Scho�llhorn, Fiz. Nizk. T emp. (Kharkov), 1990, 16, 572. We acknowledge the support of the work by the DFG Deutsche 16 M. Hansen, Constitution of binary alloys, McGraw-Hill,New York, Forschungsgemeinschaft, Bonn, Germany. 1958, p. 325. 17 S. Rasmussen and B. Lundtoft, Powder Diffraction, 1987, 2, 28; PDF card no. 39-1087. 18 A. Watanabe, N. Ishizawa and M. Kato, J. Solid State Chem., 1985, References 60, 252; PDF card no. 39-0061. 19 Gmelin, Handbuch der anorganischen Chemie, vol. 19 (Wismut), 1 G. Pfletschinger, B. Hahn and R. Scho�llhorn, Solid State Ionics, Verlag Chemie, Berlin, 1927, pp. 23, 81. 1996, 84, 151. 20 G. Kru�ger, N. Breuer and R. Scho�llhorn, unpublished work. 2 J. O. Besenhard and H. P. Fritz, Electrochim. Acta, 1975, 20, 513. 21 JCPDS, International Centre for Diffraction Data, Newtown 3 P. Fiordiponti, G. Pistoia and C. Temperoni, J. Electrochem. Soc., Square, PA, 1992, PDF card no. 44-1246. 1978, 125, 14. 4 G. Pistoia, J. Power Sources, 1985, 16, 263. 5 K. Sreedhar and P. Ganguly, Inorg. Chem., 1988, 27, 2261. Paper 6/05939B; Received 28t

ISSN:0959-9428    

DOI:10.1039/a605939b

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

Electrochemical insertion of magnesium in a mixed manganese–cobalt oxide Luis Sa�nchez and Jean-Pierre Pereira-Ramos L .E.C.S.O., C.N.R.S., UMR 28, 2 rue Henri-Dunant, 94320 T hiais, France The electrochemical insertion of Mg2+ into the cation-deficient mixed oxide Mn2.15Co0.37O4 is studied in regard to its possible use as a positive electrode in magnesium ion-transfer batteries. The cation vacancies of the mixed oxide offer a pathway for Mg2+ insertion with a maximum uptake of 0.23 Mg per mole of oxide.Voltammetric and chronopotentiometric measurements indicate a single insertion step located near 2.9 V vs Li/Li+. The strong polarizing effect of Mg2+ ions induces a more pronounced disordering process than for Li accommodation. This results in a limited stable specific capacity of 30 A h kg-1 when discharge– charge cycles are performed at a C/6 rate within the cycling limits 4.05–1.85 V vs.Li/Li+. The intercalation of divalent cations into transition-metal were determined by the following procedure. The sample (ca. 50 g) was dissolved in 5 ml of 0.1 mol dm-3 Fe2+, 5 ml of oxides has received little attention to date. However, the insertion of magnesium is of particular theoretical and practical conc.H2SO4, 5 ml of 35% HCl and 10 ml of H2O under a continuous flow of argon and heated until complete dissolution. interest both because of its similar size to lithium and because of its prospective application in ion-transfer battery systems. After cooling to 20–25 °C, 10 ml of H3PO4, 25ml of H2O and 1.5 ml of indicator (0.3% diphenylamine in ethanol) were For ion-transfer batteries, magnesium exhibits attractive properties in comparison with Li, owing to its natural abun- added.The solution was titrated with standard 0.05 mol dm-3 K2CrO4 . Previously, a blank titration was carried out under dance, relatively low price and higher expected safety. The relevant literature dealing with Mg2+ insertion reactions the same conditions. The difference between the titrations was assigned to the total content of oxidizing species and was used mainly focuses on oxides.Our laboratory studied Mg2+ insertion into V2O5 at 150 °C in molten dimethyl sulfone.1 Gregory to calculate the oxygen to metal (O/M) ratios. XRD studies were performed on an INEL X-ray system et al.2 screened about 15 compounds for their ability to insert Mg2+ ions.More recent works have shown the possibility of using Cu-Ka radiation. The oxide Mn2.15Co0.37O4 obtained from thermal decomposition of the corresponding mixed car- inserting Mg2+ species into V2O5 at room temperature,3 and also into hydrated vanadium bronzes4,5 as well as in ortho- bonate at 400 °C exhibits a tetragonal spinel structure with the following cell parameters: at=5.805 A° , ct=9.490 A° (space rhombic molybdenum oxide.6 Chemical insertion of magnesium ions2,7 has been carried group I41/amd).The chemical composition corresponds to the simultaneous presence of 0.43 MnIV and 1.72 MnIII ions (mean out in various manganese oxides such as Mn2O3, Mn3O4 and c-MnO2, whereas electrochemical Mg insertion has not yet oxidation state of metallic ions, Z=3.17) and a ratio O/M= 1.587 which indicates the presence of 0.48 vacancies per mole been reported.We sought to extend the Mg2+ insertion process to a related host lattice which belongs to a group of new of oxide.The working electrode consisted of a stainless-steel grid on cation-deficient mixed Mn–Co spinel oxides MnyCozO4 (2.5<y+z2.62), which are prepared via a solution technique. which the cathodic material was pressed.The cathode was made of a mixture of active material (80 mass%) with graphite The carbonate precursor method used results in ultrafine materials and the simultaneous presence of manganese ions (7.5 mass%), acetylene black (7.5 mass%) and PTFE (5 mass%). Anhydrous magnesium perchlorate was dried under with valences of +4 and +3.A previous electrochemical investigation on Li intercalation,8,9 has shown interesting vacuum at 190 °C for 15 h. Magnesium trifluoromethanesulfonate was dried under vacuum at 80°C for 5 h. Either an properties with one reversible process located near 2.8 V involving a faradaic yield of 0.7–0.8 F (mol oxide)-1. Moreover, Mg(ClO4)2 solution (0.1 mol dm-3) in propylene carbonate (PC) (twice distilled, Fluka) or an Mg(CF3SO3)2 solution because lithium ions are preferentially incorporated in vacant sites usually occupied by transition-metal ions, the magnitude (0.05 mol dm-3) was used as the electrolyte while a magnesium rod served as the counter electrode.An Li/Li+ couple in a of the Jahn–Teller effect is minimized as reduction proceeds.The high oxidation state of Mn (z=3.17) combined with the separated compartment acted as the reference. A thin porosity frit was used to prevent any diffusion of lithium ions. high content of cation vacancies (0.48) should also make this compound an attractive material for Mg intercalation. This paper reports on Mg insertion into the mixed oxide Mn2.15Co0.37O4 .Results and Discussion Typical voltammetric curves performed between the cycling limits 4.05–1.85 V vs. Li/Li+ for the reduction–oxidation of Experimental Mn2.15Co0.37O4 are reported in Fig. 1. Only one redox process is observed. The magnesium insertion process appears to be The mixed manganese–cobalt carbonate was prepared by the addition of a 1 mol dm-3 solution of NaHCO3 to a 0.5 mol reversible over a few cycles with cathodic and anodic peaks located at 3.05 and 3.83 V, respectively.dm-3 solution of the divalent ions MnII and CoII under a continuous flow of CO2. The composition homogeneity in The corresponding chronopotentiometric curves allow the clear quantification of the Mg insertion process (Fig. 2). A carbonate and oxide samples was determined by electron microprobe energy dispersive X-ray analysis (EDXA) with a single reduction step centred near 2.9 V leads to a faradaic yield of 0.44 F (mol oxide)-1 at a 1.8 V cut-off voltage.A more Philips SEM 501 B apparatus. The stoichiometry was con- firmed by atomic absorption spectrometry. important hysteresis appears between the reduction and oxidation curves than in the case of Li insertion.8,9 The average oxidation states of the metal ions in the sample J.Mater. Chem., 1997, 7(3), 471–473 471From chemical tests in a solution of dibutylmagnesium in heptane as insertion reagent, specific charges of ca. 150 and 220 A h kg-1 corresponding to stoichiometries of Mg0.66Mn3O4 and Mg0.8Co3O4 were estimated2 without any data on reversibility. Our results on the cation-deficient Mn–Co mixed oxide in terms of maximum Mg content are lower than those obtained from chemical experiments with stoichiometric Mn and Co spinel oxides.For host lattices, higher Mg contents are found from electrochemical experiments: 0.45 Mg2+ in V2O5,1,3 0.38/0.5 Mg2+ in MoO3,6 1.8 Mg2+ in hydrated Mg(V3O8)(H2O)y.3 The XRD patterns obtained for the electrochemically formed Mg0.23Mn2.15Co0.37O4 show that the structure of the oxide is maintained, (Fig. 3). However, even when the main diffraction lines are well identified, a disordering process, more pro- Fig. 1 Cyclic voltammetric curves of an Mn2.15Co0.37O4 electrode in nounced than that for Li0.3Mn2.15Co0.37O4 , takes place.8 This a 0.1 mol dm3 Mg(ClO4 )2 solution in propylene carbonate at 20 °C; is probably related to the divalent charge, the strong polarizing sweep rate=40 mV s-1 effect and the larger size of Mg2+ radius (0.65 A° , vs.ca. 0.6 A° for Li ions). In order to provide some information on the localization of Mg ions and their content in the cation-deficient mixed oxide Mn2.15Co0.37O4, we investigate the chronopotentiometric properties of electrochemically formed samples MgxMn2.15 Co0.37O4 in a lithium-based electrolyte.As shown in Fig. 4, the Mg-containing electrodes exhibit the typical behaviour expected for Li insertion in the parent oxide: two well defined insertion steps centred in the potential windows 3.5–2.0 and 2.0–1.0 V. However, the higher the Mg content, the shorter the transition time of the first step while the length of the second one remas unchanged.For Li and Fig. 2 Discharge–charge curves for Mn2.15Co0.37O4 in 0.1 mol dm-3 Mg(ClO4)2/PC electrolyte ( j=100 mA cm-2) According to the following equation: Mn2.15Co0.37O4+2xe-+xMg2+ =MgxMn2.15Co0.37O4 the experimental faradaic balance, consistent with both the vacancies content (0.48) and the number of reducible MnIV ions (0.43), shows that ca. 0.22 Mg2+ ions reversibly enter the mixed oxide host lattice. Lower current densities (50–20 mA) did not enable this value to be exceeded significantly. However, this maximum uptake is lower than that found for Li insertion Fig. 3 X-Ray diffraction pattern of Mg0.23Mn2.15Co0.37O4 in reflec- (ca. 0.62). In the latter case, Li ions were found to be incorpor- tion geometry ated preferentially in vacant octahedral sites, which are usually occupied by transition-metal oxides in cubic spinels of stoichiometric M3O4 but partially filled in the deficient oxide (8d sites of the space group I41/amd). Additional Li insertion of 0.14 ions in interstitial 8c octahedral sites associated with the reduction of some Mn3+ ions was considered to account for the Li content of 0.62.8 Conversely, owing to its size and charge, the maximum content for magnesium insertion can be explained by the quantitative reduction of Mn4+ ions and the filling of only half of the available vacant 8d cationic sites.Strong coulometric interactions owing to the valence of the magnesium ions probably hinder further magnesium insertion and the correlative reduction of Mn3+.In other respects, probably owing to the small size of magnesium ions, the voltage of the discharge curve is very close to that found for Li intercalation8,9 with a coulombic capacity of 75% of the Li capacity. This indicates a high diffusivity of Mg2+ ions which easily enter the pathway offered by cation vacancies. Indeed, larger cation species such as sodium and potassium were Fig. 4 Discharge–charge curves for MgxMn2.15Co0.37O4 in 1 mol dm-3 shown to enter the same host lattice at around 2.5 and 2.2 V LiClO4 /PC electrolyte (j=100 mA cm-2). (a) x=0.27; (b) x=0.23; (c) x=0.1. respectively.8 472 J. Mater. Chem., 1997, 7(3), 471–473divalent cation. In all cases, the charging process does not reveal any reversibility, as is known for stoichiometric oxides. When a deeper discharge is performed to obtain Mg0.27Mn2.15Co0.37O4, its further reduction in 1 mol dm-3 LiClO4/PC gives rise to a single voltage plateau at 1.5 V.In order to further characterize the properties relevant to the prospective use of the mixed Mn–Co oxide for secondary battery electrodes, we investigated the cycling behaviour of Mn2.15Co0.37O4. Results obtained for two cycling limits are reported in Fig. 5 and 6. During the first five cycles, it can be seen that a significant fraction of the inserted Mg2+ ions remains bonded in the crystal lattice of the cation-deficient mixed oxide and cannot be extracted from the material (Fig. 5). The specific charge decreases from 60 to 40 A h kg-1 by the fifth cycle and slowly decreases thereafter to reach 30 A h kg-1 by the 30th cycle (Fig. 6). This may be explained by the disordering process induced by the magnesium insertion as Fig. 5 Discharge–charge cycles of Mn2.15Co0.37O4 within the potential indicated by XRD experiments. It is possible to obtain a more range 4.05–1.85 V (C/6 rate) stable specific capacity by cycling within the limited voltage range 4.05–2.8 V (Fig. 5). The initial Coulombic capacity of 40 A h kg-1 decreases in the early cycles to stabilize at ca. 30 A h kg-1 after 20 cycles. These first results are noteworthy as compared with the fairly irreversible behaviour exhibited by the stoichiometric Co3O4 oxide with low Coulombic capacities (5 A h kg-1). It is the first time that reversible magnesium insertion into a manganese oxide has been reported. For the sake of comparison, only a few cathodic materials have been proved to reversibly accommodate Mg ions.For instance, vanadium pentoxide3 can be utilized after 20 cycles with an appreciable Coulombic capacity (50 A h kg-1) while ca. 60–80 and 50 A h kg-1 can be attained with hydrated vanadium bronzes5 and MoO3 in a liquid inorganic molten salt electrolyte at 80°C, respectively.6 We express our gratitude towards DGICYT, Spain, for financial support.Fig. 6 Evolution of the specific capacity for Mn2.15Co0.37O4 during cycling galvanostatic experiments in 0.1 mol dm-3 Mg(ClO4)2/PC References electrolyte. Voltage range: (#) 4.05–2.8 V; (%) 4.05–1.85 V (j= 100 mA cm-2). 1 J. P. Pereira-Ramos, R. Messina and J. Perichon, J. Electroanal. Chem., 1987, 2–8, 241. Mg insertion, the overall faradaic yield never exceeds the 2 T.D. Gregory, R. J. Hoffman and R. C. Winterton, J. Electrochem. Soc., 1990, 137, 775. expected value of ca. 0.6–0.65 F mol-1 at a 2.0 V cut-off 3 P. Novak and J. Desilvestro, J. Electrochem. Soc., 1993, 140, 140. voltage. These results seem to indicate the partial occupancy 4 P. Novak, W. Scheifele and O. Hass, J. Power Sources, 1995, 54, 479.of 8d octahedral vacant sites by Mg2+ ions with retention of 5 P. Novak, W. Scheifele, F. Joho and O. Hass, J. Electrochem. Soc., the structural integrity.8 Because there is no change in voltage, 1995, 142, 2544. it can be suggested that subsequent Li insertion occurs in a 6 M. E. Spahr, P. Novak, O. Hass and R. Nesper, J. Power Sources, similar manner by filling the remaining 8d sites and also some 1995, 54, 346. 7 P. G. Bruce, F. Krok, J. Nowinski, V. C. Gibson and K. Tavakkoli, interstitial 8c sites, as suggested previously for an Li insertion J. Mater. Chem., 1991, 1, 705. process.8 This requires the additional reduction of Mn3+ ions. 8 J. Farcy, J. P. Pereira-Ramos, L. Herna�n, J. Morales and Assuming each Mg2+ ion occupies one site, these experiments J. L. Tirado, Electrochim. Acta, 1994, 39, 339. show that the faradaic yield for Mg insertion is mainly 9 L. Sa� nchez, J. Farcy, J. P. Pereira-Ramos, L. Herna�n, J. Morales controlled by the initial content of Mn4+, but also by high and J. L. Tirado, J. Mater. Chem., 1996, 6, 37. Coulombic repulsive interactions which prevent achievement of the maximum value of 0.3 for the degree of insertion of a Paper 6/05210J; Received 25th July, 1996 J. Mater. Chem., 1997, 7(3), 471–4

ISSN:0959-9428    

DOI:10.1039/a605210j

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

The titanium(IV) salt of N,N-(diphosphonomethyl )glycine: synthesis, characterisation, porosity and proton conduction Enrique Jaimez, Gary B. Hix and Robert C. T. Slade* Department of Chemistry, University of Exeter, Exeter, UK EX4 4QD A new diphosphonic acid, N,N-(diphosphonomethyl)glycine, has been prepared. The titanium(IV) salt [Ti(dpmg)] of this acid has been characterised by X-ray powder diffraction, thermogravimetry, 31P MAS NMR spectroscopy, isothermal N2 adsorption– desorption and ac conductivity measurements.Phosphorus is present in a mixture of bonded phosphonate and free phosphonic acid groups. The material is both amorphous and porous (BET specific surface area=119 m2 g-1), and its water content is relative humidity (RH)-dependent. Ti(dpmg) is a protonic conductor (at 90% RH and 90°C, s=3×10-2 S cm-1), with conductivity exhibiting Arrhenius-type behaviour at constant RH.Conductivity and Arrhenius parameters are strongly dependent on the water content. Several recent studies have focused on the synthesis and crystal equal volume of ethanol was then added. After crystallisation for 1 day the product was removed by filtration and washed structure determination of layered phosphonates of di-, triand tetra-valent metals, e.g.ref. 1–3. In the case of tetravalent with ethanol prior to drying/storage over saturated NaCl(aq) (relative humidity, RH=75%). metals, the phosphonates usually have a layered structure, similar to that of their purely inorganic analogue a-zirconium Elemental microanalysis results obtained for 1 are in excellent agreement with the formulation (HOOCCH2)N- phosphate [Zr(HPO4)2·H2O],4 but with pendant organic groups (R) replacing the -OH groups in the interlayer region.(CH2PO3H2 )2 (calc. C, 18.25; H, 4.22; N, 5.32; P, 23.55%; obs. C, 18.22; H, 4.12; N, 5.51; P, 23.47%); thermogravimetry Such solids can be synthesised with a wide range of R,5 and combinations of two or more phosphonic acids may also be showed no evidence of hydration.Mass spectrometry showed the molecular ion at m/z=265, and intense peaks at m/z=81 used.6–8 These synthetic aspects can be exploited in designing new materials with potential applications such as sorption, ion (PO3H2) and 96 (CH2PO3H2),fully consistentwith the molecular formula. exchange, chromatography, chiral molecular recognition, photochemistry, catalysis and ionic conduction,9–15 and the reactivities of the pendant organic functions have also been Titanium(IV) derivative of N,N-(diphosphonomethyl)glycine, studied.1,16–19 In a recent review Kreuer highlighted the proton Ti(dpmg) 2.The synthesis of phosphonates of tetravalent conductivities of layered acidic phosphates and phosphonates metals shows kinetic control.7,8 The precipitation of phosphon- of zirconium(IV) in their hydrated and dry forms.20 Layered ates usually occurs immediately, generating very small particles. phosphonates containing sulfonic acid groups in the organic This suggests a high nucleation rate and high insolubility.The moiety have conductivities approaching that of the sulfonated use of diphosphonic acids further increases the rate of precipi- fluorocarbon polymer Nafion,21,22 which is the electrolyte tation, yielding materials of very low crystallinity.membrane commonly used in developing solid polymer fuel The synthetic method used a reaction mixture containing cell technology. TiIII which was oxidised slowly to TiIV by contact with air (this An important problem in the synthesis of such phosphonates method has been used in the synthesis of crystalline titanium is the limited commercial availability of phosphonic acids with phosphonates with other phosphonic acids).11,23 N,N- suitable functional groups.It is, therefore, necessary to prepare (Diphosphonomethyl)glycine (1, 1 g) was dissolved in 75 cm3 phosphonic acids as the first step in the preparation of a new of deionized water.The solution was mixed with an aqueous proton-conducting phosphonate. In this paper, we report syn- solution of TiCl3 (2.3 mol dm-3 in 20 mass% HCl, 0.8 cm3, thesis and characterisation of (i) the new diphosphonic acid Aldrich) using a phosphorus to titanium ratio of 4 (phosphonic N,N-(diphosphonomethyl)glycine (such a compound offers the acid in excess of the stoichiometric requirement).The white prospect of chiral phosphonates, e.g. for use as chiral stationary solid precipitate formed immediately was heated in contact phases in chromatography); and (ii) the TiIV salt of the new with the mother liquor for 10 days at 90°C under reflux acid. This acidic material is a porous proton conductor. conditions, and then recovered by filtration and washed thoroughly with deionized water. Experimental Instrumental Synthesis Samples were characterised by thermogravimetry (TG) and differential thermal analysis (DTA) in the range 25–900 °C N,N-(Diphosphonomethyl)glycine, 1.Phosphorus trichloride (13.1 cm3, BDH Laboratory Reagent) was added dropwise to (heating rate 10°C min-1) in air using a Stanton Redcroft STA 781 instrument.Elemental microanalysis (C, H, N) used a stirred, ice-cold solution of phosphorous acid (12.3 g, Fluka) in 25 cm3 of deionized water. To this was added a solution a Perkin-Elmer 240 elemental analyser. IR spectra were recorded as KBr disks on a Nicolet Magna 550 FTIR spec- containing a stoichiometric equivalent of glycine (Aldrich). The resulting solution was heated to 90°C and maintained at this trometer. 1H and 31P solution-state NMR spectra (D2O solvent) were obtained on a Bruker DRX 400 spectrometer. X- temperature during slow addition of paraformaldehyde (9.9 g, Fisons) over a period of 3 h. The resultant solution was then Ray powder diffraction (Philips PW 1050 instrument, modified for computer-driven step-scanning and data acquisition; Ni heated at reflux for 1 h, after which the volume of solution was reduced by a factor of 3 by vacuum distillation and an filtered Cu-Ka radiation, l=1.54178 A° ) revealed Ti(dpmg) 2 J.Mater. Chem., 1997, 7(3), 475–479 475to be amorphous to X-rays. Proton-decoupled 31P MAS NMR spectrum of 1 exhibits a very strong doublet due to the stretching vibration of PNO (hydrogen bonded) at spectra of solids were recorded on a Varian Unity 300 instrument at a frequency of 121 MHz and employing cross-polaris- 1244–1150 cm-1 and a broad, medium intensity band at 933 cm-1 arising from the PMO stretching vibration. The ation with flip-back; chemical shifts are reported with respect to analytical grade 85% H3PO4 .Solid samples were spun at phosphonate Ti(dpmg) 2 shows features typical of such materials: a broad band located at 3500 cm-1 associated with the 2.5 kHz and 3.5 kHz to identify spinning satellites.N2 adsorption –desorption isotherms were measured volumetrically at OH bond vibration of interlayer water, a very intense band centred at 1050 cm-1 and another at 1615 cm-1, the latter 77 K using a Micromeritics Gemini 2375 instrument equipped with a computer-controlled measurement system.two associated with the PMO bond.25 The structure of 1 in solution was confirmed by NMR Electrical impedance measurements utilised a Hewlett Packard 4192A LF impedance analyser programmed via an spectroscopy: dH (400 MHz, D2O): 3.5 (4 H, s, NCH2PO3H2 ), 4.0 (2 H, s, NCH2COOH) (COOH and PO3H2 are not IBM-compatible computer for data collection and analysis (using a program embedding EQUIVCRT modelling observable owing to chemical exchange); dP (161 MHz, D2O): 9.1 (one phosphorus environment in solution). The 31P MAS software24).Impedance spectra were recorded at 10°C intervals in the range 20T /°C90, using frequencies of 5 Hz to NMR spectrum of solid 1 (Fig. 2) showed two peaks (dP 13.6, 15.5) with equal intensities (taking account of spinning side- 13 MHz and an oscillating voltage of 300 mV.Pellets (13 mm diameter, 0.7–1.2 mm thickness) were prepared by pressing ca. bands). Two phosphorus environments in the solid-state structure might be explained in terms of intramolecular 200 mg of material at 40 kN cm-2 in a pellet die. The two flat surfaces were painted with conductive silver paint (Electrodag condensation (Scheme 1).The observed peaks correspond, however, to similar chemical shifts and are therefore likely to 915, Acheson Colloids) to give blocking electrodes. The pellet was placed between two copper foils and the sample was arise from crystallographically inequivalent phosphonic acid groups. The absence of condensation is consistent with elemen- spring-loaded to ensure good electrode–electrolyte contact.This assembly was placed inside a cell and maintained at tal analysis (see earlier). controlled relative humidity [RH: 0% above anhydrous CaCl2, 100% above H2O(1)]. Prior to measurements, the cell was equilibrated for 1 h at each temperature (this time being found by experience to be twice the minimum necessary for temporal stability in impedance spectra).Experiments were repeated at least three times to ensure reproducible conductivity values and averages are given. Scheme 1 Results and Discussion The possibility of Ti–O–Ti-type aggregates in 2 was exam- Spectra ined by calcination (6 h at 900°C in air); the X-ray diffraction profile for the resulting solid corresponded to that of Fig. 1 shows the IR spectra of N,N-di(phosphonomethyl)gly- titanium(IV) pyrophosphate, with no evidence for any form of cine 1 and its TiIV derivative, Ti(dpmg) 2. The spectra are seen titanium oxide. Compound 2, therefore, results from reactions to have similarities: a strong signal due to the CNO stretching between the diphosphonic acid and the tetravalent metal, with vibration of the carbonyl group of COOH at 1733 cm-1 (as no side reactions resulting in the generation of impurity phases.in a-amino acids), a doublet in the region 1230–1030 cm-1 Fig. 3 shows the 31P MAS NMR spectrum of Ti(dpmg) 2. due to the CMN stretching vibration of the tertiary amine A single broad feature with a shoulder was observed, which (signals for protonated amine are not detected), a medium can be resolved into two peaks (dP 8.9,-1.2) and their spinning intensity signal due to the CH2 deformation mode of the PCH2 sidebands. These peaks do not correspond to signals typical group at 1421 cm-1 and bands at 3000–2800 and of layered titanium(IV) phosphate (TiP) phases (a-TiP, d-18.4; 1400–1300 cm-1 arising from CMH stretching and defor- c-TiP, d -10.9, -32.8; titanium(IV) dihydroxyphosphate, d mation modes respectively.In addition to these bands, the Fig. 2 31P MAS NMR spectra for N,N-(diphosphonomethyl)glycine Fig. 1 FTIR spectra of (a) N,N-(diphosphonomethyl)glycine 1 and 1, presented as (top-to-bottom) experimental spectrum, fitted spectrum, Gaussian components and difference plot (b) Ti(dpmg) 2 476 J. Mater. Chem., 1997, 7(3), 475–479Fig. 5 shows TG and DTA traces for a sample dried at 50°C for 2 days and stored over anhydrous CaCl2 .The TG trace shows no mass change between 20 and 80°C, indicating the absence of adsorbed water. The endotherm in the DTA trace at 80–160 °C is assigned to dehydration. The TG trace in the range 80–650 °C indicates a complex dehydration and decomposition sequence (several processes overlap), including the loss of interlayer water, oxidation of the organic portion and condensation of phosphoric groups.The DTA trace shows two exothermic process, assigned as arising from the oxidation of the organic moiety and the phosphate-to-pyrophosphate transformation. The ‘free’ (non-layer) phosphonic acid groups are transformed in air to water, carbon dioxide and phosphorus pentoxide. The final mass loss takes place at 750°C and is marked by a sharp peak in the DTA trace, corresponding to a final transformation to cubic pyrophosphate.26 On the basis of the total mass loss occurring up to 900°C (final product TiP2O7) and elemental microanalysis, Ti(dpmg) 2 (dried at 50°C for 2 days, stored over anhydrous CaCl2) can be formulated as Ti[HO2CCH2N(CH2PO3 )2 ]0.28 Fig. 3 31P MAS NMR spectra for Ti(dpmg) 2, presented as (top-to- [HO2CCH2N(CH2PO3H2)CH2PO3]1.43·0.6H2O bottom) experimental spectrum, fitted spectrum, Gaussian components and difference plot [calc. C, 16.27; H, 3.17; N, 4.74%; obs. C, 16.05; H, 3.32; N, 5.11%; TG loss (calc.)=55.1, TG loss (obs.)=54.2%]. The ratio of P(phosphonate)5P(phosphonic acid) in this formu- -6.4).25 The solid does not contain significant aggregations lation is 1.3951, in good agreement with the NMR result of of titanium phosphate (hydrolysis of 1 would yield phosphoric the previous section.acid in solution; the degree of hydrolysis at the working pH is shown to be negligible). The synthesis used an excess of Porosity diphosphonic acid (see earlier) and, owing to the fast reaction rate, the TiIV reacts predominantly with only one of the The nitrogen adsorption–desorption isotherm of Ti(dpmg) 2 phosphonic acid groups of diphosphonic acid.This is in good at 77 K (Fig. 6) corresponds to type IV of the BDDT classifi- agreement with two peaks in the 31P spectrum; a high-field cation,27 the form usually associated with titanium(IV) phos- signal for the ‘free’ phosphonic acid group (dP -1.2), and a phonates.28 The material has a BET specific surface area of low-field signal (dP 8.9) for the phosphonate group bonded to 119 m2 g-1, a value ten times greater than that typical of TiIV.The peak assigned to the phosphonate site is, however, layered phosphates. The isotherm has a narrow hysteresis loop more intense than that assigned to the phosphonic acid site of the H3 type, characteristic of aggregates of plate-like par- (integral ratio 1.451.0, including sidebands), which points to a ticles giving rise to slit-shaped pores.29 small fraction of diphosphonate bridging.Fig. 7 shows the cumulative and differential pore volume curves for Ti(dpmg). In the analysis of the desorption isotherm Thermal properties at a relative pressure of ca. 0.42 a strong N2 desorption is observed which gives rise to a peak in the pore radius Thermal analyses of Ti(dpmg) 2 showed the water content to distribution. The presence of such a peak in a porosity distri- depend on the drying and storage conditions. Fig. 4 shows the bution has been considered by the IUPAC Commission,30 who equilibrated water content as a function of relative humidity consider that this effect, typical of many materials, is illusory.29 (RH) for a sample pre-dried at 50°C for 2 days.The analysis of the desorption isotherm by the slit-shape model Fig. 4 Equilibrium water content at 25 °C of as-prepared Ti(dpmg) 2 Fig. 5 TG and DTA traces for Ti(dpmg) 2 dried at 50 °C and stored as a function of relative humidity (RH) (analytical chemical formulation as presented in the text refers to material dried at 50 °C) over CaCl2(s) J.Mater. Chem., 1997, 7(3), 475–479 477Fig. 8 Impedance spectrum obtained for Ti(dpmg) 2 at 50°C and Fig. 6 N2 adsorption–desorption isotherm at 77 K for Ti(dpmg) 2 RH=100% degassed at 100°C Fig. 9 Temperature dependence of conductivity, s, for Ti(dpmg) 2 at RH=0 (a) and 100% (b) Fig. 7 Cumulative and differential pore volume curves for Ti(dpmg) 2 Table 1 Room-temperature conductivities and Arrhenius parameters for Ti(dpmg) 2 at RH=0 and 100% gives a porosity distribution with a maximum at 160 A° (DP is the distance between slit walls). RH (%) s/S cm-1 Ea/kJ mol-1 log10 (A/S cm-1 K) 0 1.9×10-7 85 (1) 24.5 (0.3) Impedance spectra and proton conductivity 100 3.7×10-4 63 (2) 22.7 (0.2) Characterisation of the electrical conductivity of the phosphonate 2 was carried out by ac impedance measurements.In all cases, the impedance plane plots obtained consisted of a single decreasing RH (and level of hydration). At constant temperature, observed values of s are 2–3 orders of magnitude greater depressed semicircle (due to the bulk electrolyte resistance) with a low-frequency tail (due to electrode–pellet interfacial when measurements are carried out at 100% RH atmosphere than those under 0% RH conditions.Water molecules clearly impedance); an example is shown in Fig. 8. Such impedance spectra are typical of studies of solid electrolytes with electrodes play an important ro�le in the conduction processConductivity increases by two orders of magnitude temperature is increased blocking the ionic charge carriers.The sample resistance was calculated by extrapolation of the high-frequency arc to the [at 0% RH s(25°C)=1.9×10-7 S cm-1 and s(90°C)= 6.9×10-5 S cm-1; at 100% RH s(25°C)=3.7×10-4 S cm-1 real axis. Application of a dc voltage to the sample cell led to very rapid exponential decay in the consequent dc current, and s(90°C)=3.5×10-2 S cm-1].The strong dependence of conductivity on the water content this behaviour also being fully consistent with ionic, rather than electronic, conduction in Ti(dpmg). indicates a vehicular mechanism, where protons are carried by pore water (at low relative humidities loss of water reduces In the temperature range 20–90 °C the plots log10(sT ) vs. 1/T are linear and can be modelled with an Arrhenius-type the number of vehicles present and thus conductivity is lower). We could not detect protonic conductivity at T130 °C, and equation [sT=A exp(Ea/RT )]. The Arrhenius plots for 2 at 0% and 100% RH are shown in Fig. 9. Ea and A values are a water-free mechanism involving proton jumps between layer sites or between pendant groups is, therefore, not involved.listed in Table 1; errors in these parameters assume an average relative error of 3% on the pellet resistance ( justified by the The observed conductivities at room temperature are comparable to that of amorphous zirconium phosphate.31 The reproducibility of experimental values). Ea increases with 478 J. Mater. Chem., 1997, 7(3), 475–4796 G. Alberti, U.Costantino, S. Allulli and N. Tomassini, J. Inorg. values at elevated temperature are, however, higher than Nucl. Chem., 1978, 40, 1113. anticipated by that comparison. This increase in conductivity 7 K. J. Scott, Y. Zhang, R-C. Wang and A. Clearfield, Chem. Mater., arises from two sources: (1) proton transport in layered 1995, 7, 1095. phosphates is dominated by surface transport; surfaces 8 E.Jaimez, A. Bortun, G. Hix, J. R. Garcý�a, J. Rodrý�guez and areas>10–20 m2 g-1 have not previously been obtained, and R. C. T. Slade, J. Chem. Soc., Dalton T rans., 1996, 2285. 9 G. L. Rosenthal and J. Caruso, Inorg. Chem., 1992, 31, 3104. only moderate proton conductivities have hitherto been 10 L. Maya and P. O. Danis, J.Chromatogr., 1980, 190, 145. reported at elevated temperatures;32,33 (2) the phosphonate 11 A. Bortun, A. Clearfield, L. Bortun, E. Jaimez, M. A. Villa-Garcý�a, contains both free carboxylic and free phosphonic acid groups, J. R. Garcý�a and J. Rodrý�guez, J. Mater. Chem., submitted. hence creating a high density of charge carriers for the conduc- 12 G. Cao, M. E. Garcý�a, M. Alcala�, L.F. Burgess and T. E. Mallouk, tion process. J. Am. Chem. Soc., 1992, 114, 7574. 13 M. Ogawa and K. Kurada, Chem. Rev., 1995, 95, 399. 14 A. Clearfield, in Surface Organometallic Chemistry: Molecular Approaches to Surface Catalysis, ed. J. M. Basset, Kluwer Conclusions Academic Publishers: Norwell MA, 1988; pp. 271–298. 15 M. Casciola, U. Costantino, A. Peraio and T. Rega, Solid State Ti(dpmg) is a porous proton-conducting material,with conduc- Ionics, 1995, 77, 229.tivity strongly dependenton relative humidity and temperature. 16 D. A. Burwell and M. E. Thompson, Chem.Mater., 1991, 3, 91. At elevated temperature and high relative humidity, conduc- 17 D. A. Burwell and M. E. Thompson, Chem.Mater., 1991, 3, 730. tivities are comparable to that of Nafion under similar 18 C.Bhardwaj, H. Hu and A. Clearfield, Inorg. Chem., 1993, 32, 4294. conditions [s(90°C)=3.5×10-2 S cm-1 for RH=100%]. 19 C. Y. Ortiz-Avila, C. Bhardwaj and A. Clearfield, Inorg. Chem., 1994, 33, 2499. Conductivities are, however, much lower at near-ambient 20 K. D. Kreuer, Chem.Mater., 1996, 8, 610. temperatures. 21 G. Alberti, M. Casciola, U. Costantino, A.Peraio and R. Vivani, The porosity of Ti(dpmg) rules out its use in an electrolyte Solid State Ionics, 1991, 46, 53. separator (e.g. in a fuel cell). The strong dependence of s on 22 G. Alberti, M. Casciola, U. Costantino, A. Peraio and relative humidity does, however, point to potential use in a E. Monterini, Solid State Ionics, 1992, 50, 315. 23 A. Bortun, E. Jaimez, R. Llavona, J.R. Garcý�a and J. Rodrý�guez, conductimetric humidity sensor. Mater. Res. Bull., 1995, 30, 413. 24 B. A. Boukamp, Solid State Ionics, 1986, 20, 31. We thank the Commission of the European Communities for 25 B. Bujoli, P. Palvadeau and M. Queignec, Eur. J. Solid State Inorg. funding (BRITE-Euram II contract BRE2-CT93-0535) and for Chem., 1992, 29, 141. 26 K. Segawa, S. Nakata and S.Asaoka, Mater. Chem. Phys., 1987, a personal bursary for E.J. (contract BRE2-CT94-3096). We 17, 181. thank the EPSRC Solid-state NMR Service (Durham) for 27 S. Brunauer, L. S. Deming, W. S. Deming and E. Teller, J. Am. recording and fitting of MAS NMR spectra. Chem. Soc., 1940, 62, 1723. 28 M. A. Villa-Garcý�a, E. Jaimez, A. Bortun, J. R. Garcý�a and J. Rodrý�guez, J. PorousMater., 1995, 2, 293. 29 S. J. Gregg and K. S. W. Sing, Adsorption Surface Area and References Porosity, Academic Press, London, 1982. 1 S. Drumel, P. Janvier, P. Barboux, M. Bujoli-Doeuff and B. Bujoli, 30 K. S. W. Sing, D. H. Everett, R. A. W. Haul, L. Moscou, Inorg. Chem., 1995, 34, 148. R. A. Pierotti, J. Rouquerol and T. Simieniewska, Pure Appl. 2 D. M. Poojary, Z. Baolong and A. Clearfield, Angew. Chem., 1994, Chem., 1985, 57, 603. 106, 2422. 31 G. Alberti and M. Casciola, in Proton Conductors: Solids, mem- 3 S. Bruque, M. A. Aranda, E. R. Losilla, P. Olivera-Pastor and branes and gels—materials and devices, ed. Ph. Colomban, P. Maireles-Torres, Inorg. Chem., 1995, 34, 893. Cambridge University Press, Cambridge, 1992, p. 238. 4 A. Clearfield and G. D. Smith, Inorg. Chem., 1969, 8, 431. 32 R. C. T. Slade and J. A. Knowles, Solid State Ionics, 1991, 46, 45. 33 R. C. T. Slade, H. Jinku and J. A. Knowles, Solid State Ionics, 1992, 5 M. B. Dines, P. DiGiacomo, K. P. Callahan, P. C. Griffith, R. Lane 50, 287. and R. E. Cooksey, in Chemically Modified Surfaces in Catalysis and Electrocatalysis, ed.J. S. Miller, ACS Symp. Ser. 192, American Chemical Society,Washington, DC, 1982. Paper 6/06024B; Received 2nd September, 1996 J. Mater. Chem., 1997,

ISSN:0959-9428    

DOI:10.1039/a606024b

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

Chemical stability study of BaCe0.9Nd0.1O3-a high-temperature proton-conducting ceramic Fanglin Chen,a,b O. Toft Sørensen,*b Guangyao Menga and Dingkun Penga aDepartment ofMaterials Science and Engineering, University of Science and T echnology of China, Hefei, Anhui 230026, P. R. China bMaterials Department, Risø National L aboratory, DK-4000 Roskilde, Denmark BaCe0.9Nd0.1O3-a (BCN) ceramic is known to be an excellent high-temperature proton conductor and is a candidate electrolyte for use in solid oxide fuel cells, hydrogen or steam sensors and steam electrolysers.In this work, the chemical stability of BCN was investigated systematically by combining XRD and DTA–TG techniques to study its processing compatibility and its feasibility in potential applications.It was found that above 1200 °C, BCN reacted with alumina or zirconia, leading to the loss of barium and an excess of cerium. In cold water, both sintered BCN disks and powder samples had very low solubility and did not hydrolyse, but they were soluble in some mineral acids, especially in HCl with the liberation of Cl2. In boiling water, BCN pellets dissolved readily with decomposition into CeO2 and Ba(OH)2 .In 1 atm CO2 , BCN decomposed to form CeO2 and BaCO3 below 1200°C during heating, but during cooling it was stable above 1000 °C, possibly because BCN has different crystal structures at low and high temperatures. At 600–1000 °C, BCN showed a slight mass loss when exposed to a reducing atmosphere, and a slight mass gain in an oxidizing atmosphere. XRD results revealed that BCN demonstrated chemical and structural stability in both reducing and oxidizing atmospheres.BaCeO3 doped with rare-earth-metal oxides such as Nd2O3 is Following the above observations, the purpose of this using known to exhibit significant protonic conduction in hydrogen work is to systematically examine the chemical stability of Ndor water vapour containing atmospheres at elevated tempera- doped BaCeO3 in the preparation process and in different tures.1–6 Their ability to conduct protons makes these systems environments to investigate the possibility of using this material potential candidates for applications in many novel electro- for practical applications.chemical devices such as solid oxide fuel cells, hydrogen or steam sensors, electrolysers for hydrogen production and hightemperature membrane reactors.BaCeO3-based perovskite oxides are usually formed by a Experimental conventional ceramic process involving calcining mixtures Sample preparation of the respective oxides and carbonates at elevated temperatures(1100°C) followed by sintering powder com- BaCe0.9Nd0.1O3-a (BCN) was prepared through a conven- pacts at 1400–1600 °C.1,6 The sample holders used are usually tional ceramic route.The starting materials were BaCO3, CeO2 alumina or zirconia. At high temperatures, there is probably and Nd2O3 (all from Alfa). The chemicals, mixed in stoichio- reaction between the sample and the sample holder, and this metric ratio, were ball-milled in ethanol for 36 h. The resulting will undoubtedly contaminate the product and affect its electri- slurry was then dried in air and calcined at 1000 °C, which cal properties.resulted in a single perovskite phase. After calcination, the When these materials are used as the electrolytes for solid powder was crushed and ball-milled in ethanol, after which it oxide fuel cells, wet fuels (dilute hydrogen) are supplied. In the was pressed into pellets without the addition of binders and cases of steam sensors and electrolysers, water vapour is sintered in air by two different methods at 1000, 1200 and present in the working environments. So, the stability of these 1400°C for 10 h and at 1500 °C for 2 h.In the first method, materials in water vapour containing atmospheres is of particu- green pellets were placed on alumina or zirconia plates, while lar importance. Tanner and Virkar8 found that both pure and in the second method, pellets were placed on a thick layer of rare-earth-metal doped BaCeO3 were thermodynamically calcined powders of the same composition which was laid unstable in the temperature range 500–900 °C in an atmosphere between the pellets and the alumina or zirconia plates.The of ca. 430 Torr H2O (through a 90°C water bubbler), and contact surface of the sintered specimens, alumina and zirconia decomposed to form CeO2 and Ba(OH)2. However, the water plates was characterized by X-ray diffractometer (Philips solubility of BaCeO3-based material at lower temperatures is PW1078/10). still unreported. In another promising application of solid oxide fuel cells using readily available city gas as the fuel (of which CH4 is the main component), the stability of the electrolyte in CO2 Stability in water containing atmospheres is vital, since by reforming city gas at Some 1500 °C sintered pellets prepared by the second method 800 °C, a quasi-fuel gas atmosphere containing ca. 8% CO2 is (>97% theoretical density, no open porosity) were immersed produced.The stability of undoped BaCeO3 in a CO2 atmos- in cold water, while others were immersed in water and heated phere has been studied by Scholten et al.9 and Gopalan and to boiling point. The pH values in both cases were measured Virkar,10 but there is some discrepancy between their results. using a pH meter. In cold water, the pellets were intact after Scholten et al.found that BaCeO3 decomposed below 1185 °C, one week, while those in boiling water broke into pieces and while Gopalan and Virkar showed that BaCeO3 were thermo- finally became powders after 6 h. The resulting pellets and dynamically unstable below 1090°C. Therefore, the stability of BaCeO3 in a CO2 atmosphere needs to be reinvestigated. powders were dried and examined by XRD.J. Mater. Chem., 1997, 7(3), 481–485 481Exposure to different atmospheres Some of the 1500 °C sintered pellets from the second method were ground into powders. XRD revealed them to be a single perovskite phase. The powders were examined by differential thermal analysis (DTA) and thermogravimetry (TG), using a Netzsch STA 429 thermal analyser. Samples were heated from 25 to 1400 °C at a heating rate of 2°C min-1 and a cooling rate of 5°C min-1 in 7% H2–93% Ar, air and CO2, respectively.After DTA–TG measurements, the powders were characterized by XRD. Results and Discussion Reaction between BCN and alumina or zirconia When pellets were sintered in direct contact with alumina, the contact surface of the sintered pellet turned yellow; when laid directly on zirconia, it was grey, while Nd-doped BaCeO3 was dark brown.With an increase in the sintering temperatures, the change in colour of the contact surface became more obvious. XRD traces of the contact surface of the sintered samples are shown in Fig. 1 and 2. The diffraction peaks belonging to BaCeO3 were identified easily, but additional peaks due to CeO2 were also present.Fig. 1 and 2 clearly show that at high sintering temperatures, BCN reacted with the contact substrates and a considerable amount of CeO2 was formed. XRD patterns of the contact substrates showed that BaAl2O4 was formed on alumina and BaZrO3 was formed on zirconia. Accordingly, it can be concluded that BCN reacted with Al2O3 or ZrO2 and decomposed through the following Fig. 2 XRD patterns of the contact surface of the sintered BCN pellet reactions: directly on zirconia at (a) 1000, (b) 1200, (c) 1400 and (d) 1500°C. #, BaCeO3+Al2O3�BaAl2O4+CeO2 (1) BaCeO3; +, CeO2. BaCeO3+ZrO2�BaZrO3+CeO2 (2) heights of the CeO2 and BaCeO3 peaks. When the specimen The extent of the reaction between the sample and the was sintered at high temperatures, the reaction between the substrate could be determined by comparison of the relative sample and alumina was greater than that between the sample and zirconia.The Gibbs energies of the individual compounds and reactions (1) and (2) are listed in Table 1. It can be seen that BaCeO3 is thermodynamically unstable in contact with alumina or zirconia and the reaction with Al2O3 is more thermodynamically ured than that with ZrO2.When BCN was sintered at high temperature on alumina or zirconia, the contact surface of the sintered specimen showed mainly CeO2 peaks, and no trace of barium aluminate or barium zirconate was identified. This implies that reactions (1) and (2) took place, and barium reacted with the substrate, leaving CeO2 on the contact surface of the sintered specimen.This will lead to an excess of CeO2 in the product. Since the reaction between BCN and Al2O3 proceeded more smoothly than that of BCN and ZrO2, the contact surface of the sintered specimen was greyish when sintered on zirconia and yellowish (a typical colour of CeO2) on alumina. When a pellet of BCN was placed on a thick layer of calcined powders of the same composition and sintered, both the surface and the inner part of the sintered pellet had the same dark brown colour.XRD showed that the pellet had a single perovskite phase, indicating that no reaction had taken place between the sample and the substrate. XRD patterns revealed that the calcined powder layer was a mixture of CeO2 and perovskite phase, indicating its reaction with the substrate.So the thick layer of calcined powders acted as a buffer and prevented the reaction between the pellet and the substrate. Therefore, using a thick layer of calcined powders in the sintering process seems to be an effective way to obtain sintered samples with desired BCN compositions. At low sintering temperatures (<1000 °C), BaCeO3 exhibited Fig. 1 XRD patterns of the contact surface of the sintered BCN pellet kinetic stability over alumina or zirconia, and BaCeO3-based directly on alumina at (a) 1000, (b) 1200, (c) 1400 and (d) 1500°C.#, BaCeO3 ; +, CeO2 . thin films have been successfully fabricated on alumina 482 J. Mater. Chem., 1997, 7(3), 481–485Table 1 Gibbs energy values (in kJ mol-1) of some compounds and reactions (1) and (2) T /K BaCeO3a BaZrO3a BaAl2O4b CeO2b Al2O3b ZrO2b DGf (1) DGf (2) 298 -1623.86 -1664.43 -2368.84 -1105.95 -1687.20 -1111.42 -163.73 -35.10 1000 -1411.77 -1465.58 -2550.17 -1183.52 -1774.62 -1180.04 -547.30 -57.29 1300 -1317.35 -1379.47 -2659.12 -1229.97 -1834.42 -1222.96 -737.32 -69.13 1500 -1254.32 -1322.07 -2739.12 -1263.94 -1882.51 -1254.76 -866.23 -76.93 1700 -1191.46 -1264.92 -2824.22 -1300.06 -1934.11 -1289.32 -998.71 -84.20 1800 -1160.09 -1236.44 -2868.53 -1318.77 -1961.14 -1307.27 -1066.07 -87.85 aRef. 11. bRef. 12. substrates by Kelder et al.13 and Jiang et al.14 using lowtemperature synthesis techniques. Stability in water When the sintered pellet from the second method was immersed in cold water (298 K), the pH value was unchanged after one week. The pellet was intact, had good mechanical strength and exhibited no mass change.XRD revealed that the pellet remained a single perovskite phase. When HNO3 or H2SO4 was added dropwise to the cold water, the pellet dissolved gradually and a yellowish powder was finally obtained, which was identified as CeO2 by XRD. When HCl was added to the cold water, many bubbles appeared on the surface of the pellet, the pellet dissolved readily and a small amount of light blue powder was obtained, which was identified as Nd2O3 by XRD.When BCN sintered specimens were ground into powders and then placed in cold water, no apparent dissolution was observed, and the pH value was only slightly increased after one week. Yokokawa et al.15 constructed an electrochemical potential diagram to analyse the solid–liquid equilibria between SrCeO3 and the aqueous species.They found that SrCeO3 was in equilibrium wirh Sr2+ and Ce3+ at P(O2)=1 bar. Analogously, it can be assumed that BaCeO3 has the following dissolution reaction: BaCeO3+3H2O+e-�Ba2++Ce3++6OH- (3) and the standard Gibbs energy for this reaction is Fig. 3 XRD patterns of the sintered BCN sample after exposure to 142.2 kJ mol-1, indicating that BaCeO3 cannot dissolve in different environments.(a) BCN pellet after 6 h in boiling water, (b) cold water according to this reaction. In the presence of HNO3 BCN powders after DTA–TG measurement in CO2 atmosphere and or H2SO4, BaCeO3 dissolved in cold water and CeO2 was (c) BCN powders after DTA–TG measurement in H2 atmosphere. $, BaCO3; +, CeO2; #, BaCeO3.formed, so the following reaction took place: BaCeO3+H2O�Ba2++2OH-+CeO2 (4) of BaCO3 arises from the following reaction with the existence The standard Gibbs energy for this reaction can be calculated of CO2 in ambient atmosphere: to be -242.9 kJ mol-1, indicating that BaCeO3 is thermo- Ba(OH)2+CO2�BaCO3+H2O (6) dynamically unstable in the presence of water. This reaction may be kinetically inert in cold water, since hydrolysis of Tanner and Virkar8 found that both pure and rare-earth- BaCeO3 leads to an increase of the pH value, while no apparent metal doped BaCeO3 were thermodynamically unstable in the increase of the pH value was detected in our work.But in the temperature range 500–900 °C in an atmosphere of ca. 430 Torr presence of acid, reaction (4) can proceed much more readily H2O and decomposed to form CeO2 and Ba(OH)2.They and this is consistent with our experimental observations. concluded that the perovskite decomposed through a bulk- In hydrochloric acid, the bubbles on the surface of the pellet decomposition mechanism which involved dissolved H2O are due to the evolution of Cl2 through the following redox within the BaCeO3 lattice.As shown in this work, in cold reaction: water, BaCeO3 had a very low solubility and consequently negligible hydrolysis. But in boiling water, it showed significant BaCeO3+3H2O+Cl-�Ba2++Ce3++6OH-+cCl2(g) hydrolysis, and this fast hydrolysis may be accelerated by high (5) solubility. Therefore, our work confirms the bulk-decompo- Consequently, the perovskite structure was destroyed and sition mechanism.The instability of doped BaCeO3 in H2O- Nd2O3 precipitated from the solution. Uchida et al.18 also containing environments may restrict its use as an electrolyte found that SrCeO3 dissolved in hydrochloric acid to evolve Cl2. material, especially as electrolytes for electrolysers. In boiling water, the pellet began to chip and finally broke into small pieces.The pH increased gradually up to 14, Reaction in CO2 atmosphere indicating that the hydrolysis reaction according to eqn. (4) took place. After the pellet was boiled in water for 6 h, it Fig. 4 shows the DTA–TG diagrams of the solid-state reaction of the mixture BaCO3–CeO2–Nd2O3 in a CO2 atmosphere. At disintegrated completely into a powder. XRD patterns of the resulting powder, as shown in Fig. 3(a), revealed a mixture of 1200°C, there is a large mass loss accompanied by an apparent endothermic peak, which corresponds to the decomposition of CeO2, BaCO3 and a small amount of BaCeO3. The presence J. Mater. Chem., 1997, 7(3), 481–485 483chemically stable at 1000 °C in 1 atm CO2. Below 1000°C, BCN may transform from cubic to orthorhombic and hence loose its stability.From the above observations, it seems that BCN reacts with CO2 through an equilibrium reaction as expressed in eqn. (7), which depends on the temperature, CO2 partial pressure and structure of the perovskite phase. Despite the fact that BaCeO3-based materials decompose to form BaCO3 and CeO2 in CO2-containing atmospheres, it is expected that sintered BaCeO3-based ceramics may be kinetically stable due to the interfacial nature of the decomposition reaction with CO2.This is because CO2 cannot dissolve in BaCeO3 and the reaction must occur at the solid/gas interface. Taniguchi et al.17 studied the operating properties of solid Fig. 4 DTA–TG results of 50% BaCO3–45% CeO2–5% Nd2O3 in oxide fuel cells using BaCe0.8Gd0.2O3-a electrolytes with 80% CO2 atmosphere H2–20% CO2 as fuel gas and found that BaCe0.8Gd0.2O3-a could still be applied very well as a practical electrolyte for low-temperature fuel cells, although there was a larger cell BaCO3.The subsequent slow mass loss is due to the decompovoltage degradation rate (due to the reation of the electrolyte sition of BaCO3 and the formation of BaCeO3. During cooling, with CO2) compared with that for pureuel gas.at 1000°C, BaCeO3 began to decompose. Surprisingly, the decomposition of BaCeO3 did not go to completion and no Stability in reducing and oxidizing atmospheres mass loss occurred below 900°C. The powders obtained after DTA–TG measurements were examined by XRD, which, as Fig. 6 shows the DTA–TG results for BCN powders in H2.It shown in Fig. 3(b), showed a mixture of BaCO3, CeO2 and can be seen that, below 400 °C, there is some mass gain, which perovskite phase. This means that there is only partial may be caused by the adsorption of gas on the surface of the decomposition of BCN. The DTA–TG trace of air-sintered sample. With an increase in temperature, there is a gradual, BCN powder in a CO2 atmosphere is shown in Fig. 5. With small mass loss. Before the DTA–TG measurements, the increasing temperature there is a gradual mass gain, indicating powder was dark brown, but it became grey afterwards. When that BCN reacts with CO2 through the following reaction: the grey powder was subsequently heated in air, it changed back to dark brown. When a sintered dark brown BCN pellet BaCe0.9Nd0.1O3-a+CO2�BaCO3+0.9CeO2+0.1Nd2O3 was fired at 900 °C in an H2-containing atmosphere, it showed (7) agreen colour, but it maintained integrity and good mechanical Fig. 5 also shows that when the temperature increased up strength. When it was refired at 900 °C in air, it became dark to 1000 °C, BCN decomposed almost completely. Above brown and still demonstrated integrity and good mechanical 1100 °C, BaCO3 began to decompose and the formation of strength.The colour change from dark brown to grey may be BaCeO3 initiated. At 1400°C, a single perovskite phase was caused by the partial reduction of Ce4+ to Ce3+, and when obtained. During the subsequent cooling process, at 1000 °C, subsequently fired in air, Ce3+ was oxidized to Ce4+, and it BaCeO3 started to react with CO2 , and there was no mass changed back to dark brown.Uchida et al.18 also found that change below 800 °C. The XRD pattern of the powder after in SrCeO3-based materials, when heat-treated in H2, some of DTA–TG measurement was the same as that shown in the Ce4+ was reduced to Ce3+. Fig. 3(b). Fig. 7 shows the DTA–TG results of BCN powders in air. Taniguchi and Gamo16 studied Gd-doped BaCeO3 in air It can be seen that with increasing temperature, there is a and CO2 atmospheres and found that this material reacted slight mass gain.There is no colour change before and after with CO2 when the CO2 partial pressure was more than DTA–TG measurements. XRD patterns of the powders from 0.17 atm (1.8×104 Pa) and decomposed to form BaCO3 and the DTA–TG measurements in both H2 and air revealed a a fluorite-type oxide based on ceria.They found that the single perovskite phase, as shown in Fig. 3(c), indicating that crystal structure changed gradually from orthorhombic to BCN was chemically and structurally stable in both idealised tetragonal above 550 °C and finally to cubic at 800°C as the reducing and oxidizing atmospheres. temperature increased in air.They claimed that the cubic In Nd-doped BaCeO3 material, the following defect reactions phase was very stable physically and chemically and was (Kro�ger–Vink notation) exist. hardly affected by CO2. Our result is in partial agreement with defect caused by doping: theirs. As the temperature was increased, BCN probably started from an orthorhombic structure, which was readily attacked Nd2O3�NdCe¾+VO (9) by CO2 and was unstable below 1200 °C.During subsequent cooling, BCN may start from a cubic structure and was Fig. 6 DTA–TG diagrams of BCN in H2 atmosphere Fig. 5 DTA–TG diagrams of BCN in CO2 atmosphere 484 J. Mater. Chem., 1997, 7(3), 481–485has a different crystal structure at low and high temperatures. Despite the decomposition reaction of BaCeO3 with CO2, BaCeO3-based materials could still be applied very well as practical electrolytes for low-temperature fuel cells.At 600–1000 °C, although BCN exhibited a slight mass loss when exposed to a reducing atmosphere, and a slight mass gain in an oxidizing atmosphere, it demonstrated chemical and structural stability, integrity and good mechanical strength which fulfil the requirements of materials as electrolytes for solid oxide fuel cells.This work is supported by the National Natural Science Foundation of China under Grant No. 59372103. F. C. is Fig. 7 DTA–TG diagrams of BCN in air grateful to the Danish International Development Assistance (Danida) and State Science and Technology Commission in a reducing atmosphere: (SSTC) of China for offering his PhD study at the Materials Department, Risø National Laboratory.Dr. N. Bonanos is OOx�VO +O2+2e¾ (10) acknowledged for his critical reading of the manuscript, invalu- Ce4++e¾�Ce3+ (11) able discussions and helpful advice. Special thanks are given to P. Jensen and T. Strauss for their experimental assistance. in an oxidizing atmosphere: VO +O2�OOx+2h (12) In H2 atmosphere, the slight mass loss was caused by the References formation of the oxide vacancies.When BCN was heated in 1 H. Iwahara, H. Uchida, K. Ogaki and K. Ono, J. Electrochem. air, the slight mass gain was the result of the annihilation of Soc., 1988, 135, 529. the oxide vacancies and formation of the oxide lattice ions. 2 J. F. Liu and A. S. Nowick, Solid State Ionics, 1992, 50, 131.The slight mass change does not affect the integrity and 3 H. Iwahara, H. Uchida and K. Morimoto, J. Electrochem. Soc., mechanical strength of the electrolyte ceramic, which satisfies 1990, 137, 462. 4 H. Iwahara, H. Uchida, K. Ogaki and H. Nagato, J Electrochem. the requirements of practical fuel cell applications, in which Soc., 1991, 138, 195. the anode is the reducing atmosphere and the cathode is the 5 H.Iwahara, Chemical Sensor T echnology, 1991, 3, 117. oxidizing atmosphere. Moreover, Iwahara et al.19 found that 6 C-H. Lu and L. C. De Jonghe, J. Am. Ceram. Soc., 1994, 77, 2523. combustion products such as CO and C2H4 produced by 7 D. A. Stevensen, N. Jiang, R. M. Buchanan and F. E. G. Henn, reforming city gas in the anode compartment did not affect Solid State Ionics, 1992, 62, 279.the performance of the solid oxide fuel cell using 8 C. W. Tanner and A. V. Virkar, J. Electrochem. Soc., 1996, 143, 1386. BaCe0.9Nd0.1O3-a as the electrolyte. 9 M. J. Scholten, J. Schoonman, J. C. van Miltenburg and H. A. J. Oonk, Solid State Ionics, 1993, 61, 83. Conclusions 10 S. Gopalan and A. V. Virkar, J. Electrochem.Soc., 1993, 140, 1060. 11 T. Matsui, T hermochim. Acta, 1995, 253, 155. At temperatures 1200 °C, BCN reacted with alumina or 12 I. Barin and O. Knacke, T hermochemical properties of inorganic zirconia to form BaAl2O4 or BaZrO3, leading to a loss of substances, Springer-Verlag, Berlin, 1973. 13 E. M. Kelder, O. C. J. Nijs and J. Schoonman, Solid State Ionics, barium and an excess of cerium in the product. At temperatures 1994, 68, 5. below 1200 °C, the reaction between BaCeO3 and alumina or 14 N. Jiang, R. M. Buchanan, Z. Lu, D. A. Stevenson, R. Hiskes and zirconia was kinetically unfavourable. S. A. DiCarolis, Appl. Phys. L ett., 1994, 64, 3104. In cold water, BCN had a very low solubility, but it dissolved 15 H. Yokokawa, N. Sakai, T. Kawada and M. Dokiya, Denki readily in some mineral acids. In boiling water, BCN had a Kagaku, 1990, 58, 561. high solubility and a strong hydrolysis reaction which led to 16 N. Taniguchi and T. Gamo, Denki Kagaku, 1994, 62, 327. 17 N. Taniguchi, E. Yasumoto and T. Gamo, J. Electrochem. Soc. the decomposition of the ceramic into CeO2 and Ba(OH)2. 1996, 143, 1886. The instability of doped BaCeO3 in H2O-containing environ- 18 H. Uchida, A. Yasuda and H. Iwahara, Denki Kagaku, 1989, 57, ments at elevated temperatures may restrict its use as an 153. electrolyte material, especially in electrolyser applications. 19 H. Iwahara, H. Uchida and K. Morimoto, J. Electrochem. Soc., In 1 atm CO2, BCN decomposed to form CeO2 and BaCO3 1990, 137, 462. below 1200°C with an increase in temperature, but durit was stable above 1000°C, possibly because BCN Paper 6/05377G; Received 1st August, 1996 J. Mater. Chem., 1997, 7(3), 481–485 485

ISSN:0959-9428    

DOI:10.1039/a605377g

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

The influence of the milling liquid on the properties of barium titanate powders and ceramics Hans-Peter Abicht,*a Dieter Vo� ltzke,a Andreas Ro�der,b Reinhard Schneiderc and Jo�rg Woltersdorfc aFachbereich Chemie, Martin-L uther-Universita�t Halle-W ittenberg, D-06120 Halle/S, Germany bFachbereich Physik, Martin-L uther-Universita�t Halle-W ittenberg, D-06120 Halle/S, Germany cMax-Planck-Institut fu� r Mikrostrukturphysik, Martin-L uther-Universita�t Halle-W ittenberg, D-06120 Halle/S, Germany The influence of the milling liquid on the properties of donor-doped (La3+) semiconducting barium titanate (BaTiO3) ceramics, generated by the mixed oxide technique, was investigated.Distilled water and propan-2-ol were used as milling liquids. Water was found to have two essential effects.First, it dissolves Ba2+ ions out of BaTiO3 grains, thus creating core–shell structures which were confirmed by high-resolution electron microscopy (HREM) and electron energy loss spectroscopy (EELS). They consist of a 3–5 nm thick TiOx-rich layer followed by a layer (ca. 10 nm thick) with a molar Ba/Ti ratio increasing from 0 to 1. These core–shell structures of the BaTiO3 powder positively affect the sintering behaviour of the greens by the high reactivity of the Ti-rich interlayer.Secondly, water cleans the BaTiO3 powder of acceptor contaminants, producing ceramics with a low electrical resistivity at room temperature. Propan-2-ol-milled ceramics of a comparable composition show an electrical resistivity up to six orders of magnitude higher, owing to the compensation of La3+-doping by acceptor contaminants.Barium titanate (BaTiO3) based ceramics represent an import- microchemistry of the water-leached powders in detail, and to ant area within the growing market of functional ceramics. explain the leaching mechanism. Furthermore, the differences This is true for undoped BaTiO3 ceramics used as dielectrics in the sintering behaviour and the electrical resistivity at room as well as for doped BaTiO3 used as PTCR (positive tempera- temperature of the ceramics prepared using either water or ture coefficient of resistivity) materials.propan-2-ol-milled powders are reported. In recent years, for highly pure BaTiO3 a variety of chemical synthesis methods have been developed (oxalate precipitation, sol–gel procedures, hydrothermal syntheses), but because of Experimental high costs they are used for specific applications only.BaCO3 (Merck 1711), TiO2 (Merck 808) and La2(C2O4)3 Therefore, the conventional ceramic preparation method 9H2O (SKW Stickstoffwerke Piesteritz) were used as starting (mixed oxide technique, see ref. 1) is still the method of choice. materials, and distilled water and propan-2-ol (Merck 109634) This method includes two milling–mixing steps.as milling media. In a PVC container, the batches were milled The influence of the milling liquid on the properties of the for 24 h using PTFE-coated steel balls. The mass ratio final ceramics using identical methods and starting materials mpowder5mballs5mliquid was 15154. After filtering and drying the has not been investigated in detail.Most of the papers premixture was calcined in air at 1100 °C for 4 h before it was suppose that the stoichiometry, i.e. the Ba/Ti molar ratio, is fine-milled under the same conditions as mentioned above. determined by quantity of the starting materials used. Distilled The Ba contents of the aqueous milling liquid were analysed water is the commonly used milling medium.The leaching gravimetrically as BaSO4. Trace elements and the Ba content behaviour during the milling–mixing step has been investigated in propan-2-ol were determined by atomic absorption spec- by Adair et al.2 They reported the dissolution of Ba2+ ions troscopy (AAS) using a Varian Spectra 20 instrument. The from undoped, slightly TiO2-rich BaTiO3 powders as a function analyses were carried out after both the milling steps.of the pH of the aqueous milling liquid. The microstructure of After fine-milling, filtering and drying the powder was mixed the leached powders was not studied. The amount of Ba2+ with 5 mass% polyvinyl alcohol solution as a pressing aid. leaching was found to affect the amount of exaggerated grain After 24 h it was pressed into pellets (10 mm in diameter and growth during sintering.Adair et al.3 also analysed a large ca. 2 mm thick) with a density of 3.1 g cm-3. The greens were number of organic solvents (with different relative permittivities sintered at three different temperatures (Ts=1300, 1350, and electron-donating abilities) with respect to their ability to 1400°C, dwelling time 1 h).dissolve Ba2+ ions from BaTiO3 particles. No direct compari- For testing the electrical resistivity at room temperature the son between water- and organic solvent-milled BaTiO3 powsinteredpellets were first polished to flatten the surface followed ders was carried out. Investigations concerning the solubility by electroding with gallium–indium eutectic to provide an of Ba salts in ethanol, propan-2-ol and alcohol–water mixohmic contact. The electrical resistivity was determined by tures4,5 proved that they have low solubility in pure organic measuring the voltage U and the current intensity I at 22°C, solvents.with U=0.1 V for low-resistivity samples and with U=10 V Heywang and Bauer6 studied wet- and dry-milled Sb2O3- for high-resistivity ones.doped BaTiO6 powders, revealing essential differences in the Dilatometric investigations were carried out using a electrical properties of the final ceramics. The ceramics of the SETARAM TMA92 dilatometer (1.0 g loading, Al2O3 equip- dry-milled batch exhibited properties of a second-order PTCR ment, 10 K min-1 heating rate from 20–1400°C). element, such as the high electrical resistivity at room For electron microscope investigations the specimens were temperature, which was attributed to the formation of prepared by dispersing a small amount of the powder in pure Ba(Ti0.965Sb0.035)O3 shells.The aim of this study was to analyse the microstructure and alcohol, mixing it in an ultrasonic generator, and pipetting a J. Mater. Chem., 1997, 7(3), 487–492 487drop of this dispersion on a copper mesh covered with a holey Formvar film.Electron energy loss spectroscopy (EELS) was applied using a parallel-recording spectrometer (PEELS Gatan model 666) attached to a TEM/STEM Philips CM 20 FEG instrument, run at 200 keV (ca. 0.8–1.2 eV energy resolution). Either point analyses were made, or a series of EEL spectra along a line were recorded in the STEM mode, with the electron probe (diameter ca. 2 nm) digitally scanned by the Gatan Digiscan model 688. For spectrum processing the software package EL/P from Gatan was used. To minimize or even to avoid contamination during small probe analyses, which can lead to a drastic carbon build-up, the specimen was kept at liquidnitrogen temperature using a cooling holder (Gatan model 636).High-resolution electron microscopy (HREM) was also applied using a CM 20 FEG microscope. Results Milling in water Batch V1, with 0.001La2(C2O4)3 9H2O50.998BaCO351.00 TiO2 as the starting molar composition, was prepared using distilled water as the milling liquid. After the first mixing– milling process a Ba2+ content corresponding to 0.05 mol% of the starting BaCO3 was determined in the water.The finemilling liquid showed a Ba2+ content of 1.46 mol% of the Ba in the calcined powder. The liquid showed a pH value of ca. 10, which means that after the ceramic preparation procedure the originally stoichiometric starting mixture had a molar Fig. 1 Shrinkage behaviour of (a) greens of batch V1 (starting composi- Ba/Ti ratio of 0.985.The results of the AAS analysis of the tion 0.001La2(C2O4)3 9H2O, 0.998BaCO3, 1.00TiO2, milled in distilled milling liquids of the first milling and of the fine-milling water), and of (b) greens of batch V2 (starting composition procedures are summarized in Table 1. The Ba contents are in 0.001La2(C2O4)3 9H2O, 0.998BaCO3, 1.02TiO2, milled in propan- 2-ol). good agreement with those determined gravimetrically.Ti was not detected, noective trace. The presence of the trace elements Na, K and Al indicated strong leaching by the preceding aqueous milling, which resulted in a cleaning effect. According to the manufacturer’s certificate7 the BaCO3 used had a contamination level of <50 ppm Al, <50 ppm K and <20 ppm Na. These values were experimentally confirmed to be 25 ppm Na and 30 ppm K.The TiO2 used (4 g TiO2+56 ml distilled water, 2 h planetary ball mill) was found to contain 830 ppm K and 259 ppm Na as dissolvable contaminants. Time-dependent leaching experiments and a modified batch V1 with an excess of BaCO3 showed that the dissolved Ba content in V1 is not due to unreacted BaO, but to the leaching of the surface of BaTiO3 grains.The results of dilatometric investigations of the greens of V1 are presented in Fig. 1(a). The shrinkage behaviour is characterized by a pre-densification step beginning at 880 °C, a main densification starting at 995 °C and a shrinkage maximum at Fig. 2 Microstructure of ceramic V1 sintered at 1350 °C 1229 °C. The densification is completed at 1318 °C.After sintering, the ceramics of batch V1 are dark blue–grey Milling in propan-2-ol bodies which, when sintered at 1300 °C, show a bimodal grain size distribution with ungrown starting powder (ca. 1 mm) and Ceramics of composition La0.002Ba0.998Ti1+xO3+2x (x=0, grown grains with diameters of 30–100 mm. Fig. 2 shows the 0.005, 0.01, 0.02) were prepared using propan-2-ol as the microstructure of a ceramic sintered at 1350 °C.The micro- milling medium to enable with the direct comparison to water- structure is fully grown, consisting of 20–80 mm grains. There milled samples of similar effective composition. The analysis is a similar microstructure for the sample sintered at 1400 °C. of the milling liquid of the first and the fine-milling step is exemplary for the propan-2-ol-milled samples V2 given in Table 1.Sample V2 has a composition La0.002Ba0.998Ti1.02O3.04. Table 1 AAS analysis of the milling liquids after the first mixing– The Ba2+ content dissolved in the milling liquid was <0.005 milling step and after fine-millinga mol%, the trace element concentrations were 3 ppm Na and Na (ppm) K (ppm) Al (ppm) Ba (ppm) 2.6 ppm K referring to BaCO3. Al was below the detection limit in the milling liquid.V1 (1st milling) 15.9 79.6 6.2 51 The dilatometric investigations of the green body V2 are V1 (fine-milling) 1.6 15.6 8.2 1440 illustrated in Fig. 1(b). The shrinkage behaviour is charac- V2 (1st milling) <0.1 <0.1 <4 <5 terized by a beginning of densification at 1020 °C and maxima V2 (fine-milling) 0.4 0.34 <4 <5 at 1250 and 1317 °C.The shrinkage is complete at 1338 °C. Although the effective chemical composition of batches V1 aV1-milling in distilled water; V2=milling in propan-2-ol. 488 J. Mater. Chem., 1997, 7(3), 487–492and V2 are nearly the same, there are distinct differences orthotitanate (Ba2TiO4). Ba2TiO4 is known to inhibit the grain growth of BaTiO3.8 On the other hand, only 30 mm from the between the two sintered bodies.The colour of the V2 sintered bodies is light-brown, unlike the V1 samples which are blue– Ba-rich zone, there is Ti enrichment [at the lower left of Fig. 4(b)] with a molar Ti/Ba ratio of 2.0. grey. The V2 samples, sintered at 1300 °C, show a bimodal grain size distribution of anomalously grown grains and an ungrown matrix [see Fig. 3(a)]. The portion of the grown grains in the matrix is smaller than that in V1. The differences between batches V1 and V2 are more pronounced at sintering temperatures of 1350 and 1400 °C. The propan-2-ol-milled samples are not fully grown. There are still areas of finegrained material (1–2 mm) beside 10–50 mm grains [see Fig. 3(b), (c)]. Electron probe microanalysis (EPMA; see Fig. 4) revealed that a strongly inhomogeneous element distribution is characteristic of the propan-2-ol-milled batches. The fine-grained areas in the middle of the secondary-electron micrograph of Fig. (4a) exhibit a molar Ti/Ba ratio of 0.5 [see line scan in Fig. 4(b), (c)], corresponding to the composition of barium Fig. 4 SE micrograph of sinter V2 (Ts=1400°C) (a), corresponding Ti Fig. 3 Microstructure of ceramics V2 sintered at 1300 °C (a), 1350°C (b) and 1400°C (c). mapping (b) and element concentrations (c) along the line in (b). J. Mater. Chem., 1997, 7(3), 487–492 489amorphous layer the degree of inelastic scattering is strongly Discussion reduced relative to the inner particle regions, and hence the The influence of distilled water as the milling medium on the recording time had to be fitted appropriately to produce a properties of both the intermediately prepared BaTiO3 powder useful signal.and the final ceramics can be summarized as follows. A series of EEL spectra recorded along a line of ca. 20 nm First, during fine-milling Ba2+ ions are leached out of the length perpendicular to the particle surface (cf.marked line in surface of the BaTiO3 grains. This dissolution is facilitated by Fig. 5) revealed the element composition of the BaTiO3 particle the mechanical activation of the powder, i.e. by the milling shown in Fig. 6(a). Each spectrum was taken in the energy loss balls. It is not prevented by a longer calcination period or range ca. 400–900 eV at an integration time of 4 s with four higher calcination temperatures.read-outs accumulated. In Fig. 6(a), the lower curves with no The dissolution of Ba2+ ions from the BaTiO3 structure is notable signal are from the amorphous surface layer. They are supposed to take place according to eqn. (1), using followed by spectra with marked Ti L23, O K and Ba M45 Kro�ger–Vink notation. edges. Appropriate selection of the vertical scale shows that spectra 1 to ca. 4, representing the surface layer, primarily (BaBa+TiTi+3OO)solid+2H2O� exhibit titanium and oxygen with traces of barium [cf. (TiTi+VBa+OO+2OHO°)solid+(Ba2++2OH- )aq (1) Fig. 6(b)]. Thus, it can be concluded that this layer is mainly composed of TiOx. As the spectra are extremely noisy, near- Barium vacancies in the perovskite structure are electrically edge structures cannot be resolved, thus no information about compensated by hydroxy groups on oxygen sites.The occu- the chemical bond state can be deduced. pation of oxygen sites by hydroxy groups was detected in both The remaining spectra (lines 5, 6, 7 etc.) show Ti L23 edges single crystals9,10 and in wet-chemically prepared BaTiO3 which, relative to the Ba M45 edges, have higher jump ratios powders.11 In the IR spectra the band (nOH) of these ions is (edge maximum-to-background signal) than those in the inner between 3477 and 3485 cm-1.In the DRIFT spectra of the regions of the particle [see upper curves in Fig. 6(a)]. This calcined and fine-milled powder V1 the absorption maximum suggests titanium enrichment in the outer periphery of the is at 3609 cm-1.This band lies between the positions of OH particle extending from the thin TiOx-rich surface layer to ca. groups built into the lattice and surface OH groups 10 nm into the inner regions, which was proved by quantifi- (3680–3720 cm-1).11 cation. The element concentrations along the line profile of The experimental results can be explained by the formation Fig. 6 are given in the plot of Fig. 7. Both the TiOx -rich layer of core–shell structures consisting of a TiO2-rich surface layer, and the adjoining zone depleted in barium are clearly visible followed by a ca. 10 nm thick zone depleted in barium (defect in this diagram. The average composition of the Ti-rich zone perovskite) and a BaTiO3 core. Considering the dissolved Ba is 37±4.9 atom% Ti, 45±6 atom% O and 18±2.4 atom% content of 1–2 mol% and assuming spherical BaTiO3 grains Ba, whereas in the inner regions of the particle the chemical of 1 mm in diameter a thickness of 1.7–3.3 nm of the Ti-rich shell can be estimated, corresponding to a thickness of 4–8 unit cells of BaTiO3.The existence of such structures was proved directly by TEM and EELS. However, note that theseervations do not apply to all particles in the same way.In particular, it was difficult to find individual particles randomly lying freely over a hole of the supporting film and, in addition, thin enough for EELS analysis (thinner than 50 nm). Part of a typical BaTiO3 particle, tempered at 500 °C, is shown in the HREM image of Fig. 5. The particle is a single crystal, which can be concluded from the atomic rows occurring in its thin regions.However, the particle is partially covered with a thin amorphous layer of ca. 3–5 nm thickness. Measurements in the spot mode (electron probe of ca. 2 nm in diameter) showed a particular beam sensitivity of this layer, which resulted in damage of the layer after a certain time of illumination. Thus, the conditions for EELS analyses had to be chosen carefully so that information on its composition was not lost.At the same time, on the other hand, owing to the very small thickness of the Fig. 6 Element composition of the BaTiO3 particle, shown in Fig. 5, along a line in perpendicular arrangement to the surface, series of Fig. 5 HREM image of a single BaTiO3 particle partially exhibiting EEL spectra (a); detailed drawing of those spectra taken from the outermost zone (b) an amorphous surface layer of ca. 3–5 nm in thickness 490 J. Mater. Chem., 1997, 7(3), 487–492Table 2 Electrical resistivity at room temperature of La3+-doped BaTiO3 ceramics prepared with distilled water (V1) and propan-2-ol (V2) as milling media r/V cm sintering temperature/°C V1 V2 1300 41.7 35.5×106 1350 7.7 39.0×106 1400 9.2 8.3×106 values of the electrical resistivity shows clearly the differences between water-milled and propan-2-ol-milled ceramics with a Fig. 7 Element concentrations taken from the line profile of Fig. 6 comparable composition of La0.002Ba0.998Ti1.02O3.04. For the propan-2-ol-milled ceramics (despite the remarkable grain growth implying grains of diameter 20–50 mm), the electrical composition is 23±3.0 atom% Ti, 57±7.6 atom% O and resistivity of up to six orders of magnitude higher results 20±2.6 atom% Ba, which is in good agreement with the from the compensation of the La3+-doping by acceptor stoichiometry of BaTiO3.contaminants. Takeuchi et al.12 investigated the phase content of BaTiO3 This is our explanation for the differences in the electrical powders prepared conventionally, and showed that a layer of properties of wet and dry-milled batches described by Heywang cubic BaTiO3 had formed at the surface of the grains.The and Bauer.4 thickness of this surface layer was estimated to be ca. 5 nm, irrespective of the grain size of the BaTiO3 powder. If we consider vacancies (VBa, VO°°) producing microstress to be Summary responsible for the stabilization of the cubic phase at room Two ceramic batches were prepared using the same starting temperature13,14 the leaching behaviour is easy to explain.The materials and the same technology but different milling media cubic BaTiO3 structure on the surface of the grains, already (distilled water, propan-2-ol). The final ceramics showed dis- disturbed by vacancies, is mechanically activated by milling in tinct differences in their microstructure and electrical proper- water.The dissolution of Ba2+ ions, creating further VBa sites ties. The differences are attributed to the interaction of water in the cubic layer [eqn. (1)], results in the formation of a with the starting materials and the calcined powder.The distinctive defect structure. interaction takes place in two ways: This surface layer of the grains of batch V1, rich in TiO2 (1) Water dissolves Ba2+ ions out of the BaTiO3 grains thus and defects, is also the reason for the better shrinkage behav- creating core–shell structures, which were proved directly by iour of the green bodies. Densification processes are causally HREM and EELS.Next to a 3–4 nm thick TiOx-rich surface connected with material transport. If there is no liquid, as region a ca. 10 nm thick defect perovskite structure is detected, occurs with the BaTiO3 system at temperatures below 1320 °C, in which Ba vacancies are electrically compensated by hydroxy transport processes occur by solid-state diffusion (grain bound- groups at oxygen sites.The molar Ba/Ti ratio increases from ary diffusion). Diffusion processes are essentially influenced by 0 to 1. These core–shell structures are typical of the water- the defect chemistry. The localization of the defects at the milled BaTiO3 grains and positively affect the sintering surface of the grains is advantageous as the surface diffusion behaviour. is the mechanism dominant at the initial stage of sintering.15 (2) Water has a purifying effect on the acceptor contami- According to Valdivieso et al.,16 the presence of hydroxy nants of the starting materials.If not eliminated, these contami- groups at the surface of BaTiO3 grains has a positive effect on nants will counteract the semiconductivity arising from donor the diffusion and shrinkage processes at 850 °C.doping, thus causing high electrical resistivity. At higher temperatures (>1320°C), there are optimum conditions for the formation of Ba6Ti17O40 on the TiO2-rich surfaces of the BaTiO3 grains, and therefore for the formation The authors gratefully acknowledge the financial support of of Ba6Ti17O40–BaTiO3 eutectic, and consequently for liquid- this work by the Deutsche Forschungsgemeinschaft and the phase sintering.The result is a fully grown microstructure of Fonds der Chemischen Industrie. the sinters. The resulting excess of TiO2 (batch V1) is distributed homogeneously over the surface of the grains, whereas in case References of batch V2 (propan-2-ol milled) the excess TiO2 initially added is distributed randomly in the mixture, resulting in 1 A. Bauer, D.Bu�hling, H-J. Gesemann, G. Helke and W. Schreckenbach, in T echnologie und Anwendungen von locally different Ba/Ti ratios and an inhomogeneous micro- Ferroelektrika, Akademische Verlagsgesellschaft Geest & Portig structure after sintering. K.-G., Leipzig, 1976, pp. 199 ff. Secondly, water as the milling liquid purifies the starting 2 D. A. Anderson, J.H. Adair, D. V. Miller, J. V. Biggers and materials (BaCO3, TiO2 ) and the BaTiO3 powder from T. R. Shrout, Ceram. T rans., 1988, 485, 485. acceptor impurities. Chemical analyses reveal the contents of 3 D. V. Miller, J. H. Adair and R. E. Newnham, Ceram. T rans., 1988, dissolved impurities to be 0.0845 mol% Na, 0.264 mol% K 485, 493. 4 W. Heywang and H. Bauer, Solid State Electronics, 1965, 8, 129.and 0.054 mol% Al (in total 0.4025 mol% of impurities). These 5 G. W. Ferner and M. G. Mellon, Ind. Eng. Chem., 1943, 6, 345. ions act as acceptors, having a counteracting effect on the 6 Landolt-Bo�rnstein, Gleichgewichte-L o� sungsgleichgewichte, semiconductivity of La3+-doped BaTiO3 . These impurity con- Springer-Verlag, Berlin, 1962–1964, vol. 2, part 2b, pp. 3–618 ff. tents remain within the propan-2-ol-milled batch V2 so that 7 E. Merck, Dept. SFC/V, Tl, Data sheet Barium carbonate the effect of La3+-doping should be eliminated. Selectipur, 1711. In Table 2 the electrical resistivities of ceramics V1 and V2 8 R. K. Sharma, N. H. Chan and D. M. Smyth, J. Am. Ceram. Soc., 1985, 68, 372. at room temperature are summarized. Comparison of the J. Mater. Chem., 1997, 7(3), 487–492 4919 A. Jovanovic, M. Wo� hlecke, S. Kapphan, A. Maillard and 14 J. M. Criado, M. J. Dianez, F. Gotor, C. Real, M. Mundi and S. Ramos, Ferroelectr., L ett. Sect., 1992, 14, 79. G. Godefroy, J. Phys. Chem. Solids, 1989, 50, 623. 10 R. Waser, Ber. Bunsen-Ges. Phys. Chem., 1986, 90, 1223. 15 J-L. Hebrard, P. Nortier, M. Pijolat and M. Soustelle, J. Am. Ceram. Soc., 1990, 73, 79. 11 G. Busca, V. Buscaglia, M. Leoni and P. Nanni, Chem. Mater., 1994, 6, 955. 16 F. Valdivieso, M. Pijolat, C. Magnier and M. Soustelle, Solid State Ionics, 1996, 83, 283. 12 T. Takeuchi, K. Ado, H. Kagegawa, Y. Saito, C. Masquelier and O. Nakamura, J. Am. Ceram. Soc., 1994, 77, 1665. 13 D. Vo� ltzke and H-P. Abicht, J.Mater. Sci., 1995, 30, 4896. Paper 6/04730K; th July, 1996 492 J. Mater. Chem., 1997, 7(3), 487–492

ISSN:0959-9428    

DOI:10.1039/a604730k

出版商:RSC    

年代:1997

数据来源: RSC