摘要:

Synthesis and characterization of Mg–Co catalytic oxide materials for low-temperature N2O decomposition Min Qian and Hua C. Zeng* Department of Chemical Engineering, Faculty of Engineering, National University of Singapore, 10 Kent Ridge Crescent, Singapore 119260 Binary metal Mg–Co oxide materials have been synthesized from Mg–Co hydroxide precursors by a coprecipitation-thencalcination method.The oxide system shows high catalytic activity for low-temperature decomposition of N2O (27 mol%). Using FTIR, XRD, SEM, EA, DSC, BET and GC techniques, the hydrothermal synthesis and chemistry of the double-metal hydroxides have been studied in detail. In anion exchange and XRD studies, a hydrotalcite-like phase is also found to be present in the hydroxides owing to a partial oxidation of Co2+ to Co3+ in air.The precursor subjected to hydrothermal treatment has a higher Mg content, higher crystallinity and is more stable compared to the one aged at room temperature. However, they all give amorphous Mg–Co oxides after calcination. The Mg–Co oxide prepared from the hydrothermally treated precursor has a higher surface area and is more active for N2O decomposition. With this material system, ca. 6 moles of N2O per kg of the precursor materials can be decomposed at 350 °C within 1 h. Recently, catalytic decomposition of nitrous oxide (N2O) has chemistry studies on these layered materials, including on become an important sub-field of de-NOx research,1–18 owing other intercalation compounds, will help us to design and to the increasing concern about the earth’s atmospheric control.develop active catalysts, since the precursor materials deter- N2O is generally considered as a greenhouse gas, because it is mine the ultimate performance of the catalyst. 320 times as effective in capturing atmospheric heat as an In this paper, we report a systematic study on another type equal volume of carbon dioxide. Furthermore, N2O gas also of double-metal (Mg–Co) hydroxide/hydrotalcite precursor for contributes to catalytic stratospheric ozone destruction.low-temperature N2O decomposition. The study also looks Industrial emissions of N2O are increasing mainly from the into the chemical composition, anion exchangeability, surface use of circulating fluidized beds for combustion, automotive morphology, and chemical and thermal stabilities of these nonexhaust emission, nitric acid production and the global manu- Al-containing materials upon hydrothermal processing.It is facture of large amount of adipic acid for nylon-66.1,2,4 Besides important to mention that, unlike most N2O decomposition these chemical processes, agricultural sources due to anthro- studies carried out at low concentrations of a few hundred ppm, pogenic nitrogen fixation also make a noticeable contribution the current work used a concentrated N2O gas (27 mol%) to to the overall increase in atmospheric N2O.3 As far as the account for the off-gas stream of adipic acid production. above sources are concerned, the N2O concentration ranges from several hundred ppm in a fluidized bed combustion of Experimental coals (<500 ppm) to a few tens of mol% in a gas stream of adipic acid manufacturing.4,11 Materials preparation To tackle increasing N2O emissions, a great variety of Double-metal hydroxides of Mg–Co were prepared using the catalytic materials and processes have been developed over coprecipitation method.Briefly, a 60 ml mixed aqueous nitrate the last 10–15 years.1–18 Very recently, new oxide catalysts solution of Mg–Co (total cation concentration=1.0 mol dm-3; prepared from the thermal decomposition of hydrotalcite-like Mg(NO3)2 6H2O, >99.0%, Merck; Co(NO3)2 6H2O,>99.0%, compounds (HTlc), M-Al-CO3-HT (M=Ni, Co, Cu), have Fluka) with a molar ratio of 151 was added to 300 ml of a shown high catalytic activities in low-temperature N2O 0.5 mol dm-3 ammonia solution (pH=9.5) in a 500 ml two- decomposition,17 where the catalyst can be operated at a necked round-bottomed flask.The addition of metal cations temperature as low as 150–400 °C. This class of materials was carried out under stirring at room temperature, after promises a new generation of catalysts for future low-temperawhich, the flask was sealed to prevent inter-diffusion between ture N2O decomposition. Chemically, the HTlc precursors19,20 ambient air and water and ammonia vapours in the reaction of the above catalysts are the mixed-metal hydroxides with flask. oxidation state 2+ for transition metals and 3+ for Al.It is Two types of ageing treatment were then performed. (i) The believed that the catalytic activity achieved is due to a correct above-prepared precipitate (MC;M=Mg, C=Co) in the sealed combination of the chemical composition, oxidation states of flask was stirred at room temperature for another 5 h, followed the metals and the structural properties of the precursor by filtering and washing with deionized water.The resultant materials. precipitate was then dried in air at 60°C overnight (18 h); the In the hydrotalcite compound, Mg-Al-CO3-HT, both sample is referred to hereafter as MC1.(ii) The precipitate divalent Mg2+ and trivalent Al3+ cation elements are located (MC) was treated hydrothermally at 65°C for 18 h with in brucite sheets while the anion species (CO32-) is in the continuous stirring. Filtering, washing and drying (60°C, over- inter-sheet space.21 Likewise, the above-mentioned HTlc matenight) then gave a sample referred to hereafter as MC2.rials have a similar structure, except that the alkaline-earth- The final pH value for the system was ca. 9.0 after the above metal Mg2+ in the brucite sheets is now substituted by the ageing treatments. In the thermal stability study, the above- divalent transition-metal cations.17 From the materials viewprepared MC1 and MC2 samples were further heat-treated at point, it would be of interest to explore the catalytic activity 150, 200, 250, 300 and 350°C, respectively, for 2 h with static of the calcined HTlc materials when the trivalent Al3+ is replaced by transition metals, instead of Mg2+.Materials air in an electric furnace (Carbolite). J. Mater. Chem., 1997, 7(3), 493–499 493The anion-exchange properties of the MC1 and MC2 (20 ml min-1). Other experimental arrangements have been described in detail in our previous publications.4,18 samples were investigated using the impregnation method.This involved impregnating the finely ground MC1 and MC2 The catalyst activity was evaluated in terms of the conversion rate (X) of the N2O gas.18 N2O decomposition is a mole- powders in a sodium carbonate solution (0.5 mol dm-3, Na2CO3, 99.9%, BDH) for different times (5 min to 10 h) number-increasing reaction (N2+1/2 O2).Eqn. (1) considers this fact and has been used in evaluation of catalytic activity under magnetic stirring at room temperature. Filtering, washing and drying (details as above) were then carried out for of N2O decomposition,18 noting that the partial pressures have the same values as their respective molar fractions since the these CO32--treated materials.To conduct comparative studies, pure forms of the monohy- total pressure is 1 atm: droxides of Mg and Co were also prepared using the procedure X=(pin,N2O-pout,N2O)/(pin,N2O+0.5pin,N2Opout,N2O) (1) described above. The cation solutions for Mg and Co were 1.0 mol dm-3 in these cases; the other parameters and treat- where pin,N2O and pout,N2O are the partial pressures of N2O in the inlet and outlet gases respectively. ments were unaltered.Materials characterization Results and Discussion Formation of double-hydroxide compounds Crystallographic details for the prepared materials were established using the X-ray diffraction (XRD) method.The diffrac- Fig. 1 displays the FTIR spectra of the as-prepared MC1 and tion intensity vs. 2h spectra were measured in a Philips PW MC2, along with those of the monohydroxides Mg(OH)2 and 1729 instrument with Cu-Ka radiation (l=1.5418 A° ) with a Co(OH)2. In all samples, the sharp strong peaks located at 2h range of 5–70° at a scanning rate of 4° min-1.The 3700–3631 cm-1 are assigned to OH vibrations of the metal crystallinity and morphology of the samples were examined hydroxides whereas the broad absorption bands at by scanning electron microscopy (SEM, JEOL JSM-T330A). 3434–3488 cm-1 can be attributed to OH vibrational modes To improve the electric conductivity, the samples were surface- of water molecules which are inter-connected with other H2O coated with a gold layer of thickness ca. 10 nm in a vacuum molecules and with anions (NO3- in the current case) encapsu- evaporator (JEOL-JEE-4X) prior to the SEM analyses. The chemical compositions of the as-prepared and CO32-- treated samples were investigated with elemental analysis (EA) in a Labtam Plasmascan F10 instrument using inductively coupled plasma–atomic emission spectroscopy.Metal–oxygen bonds and included functional groups (anions) were studied by FTIR spectroscopy (Shimadzu FTIR-8101) using the potassium bromide (KBr) pellet technique. Forty scans were performed for each spectrum to ensure a good signal to noise ratio. To investigate the thermal behaviour of the above materials, differential scanning calorimetry (DSC, Netzsch DSC200) studies were conducted.Samples for DSC measurements were heated from 40 to 500 °C at a rate of 10°C min-1 under a nitrogen atmosphere with a gas flow-rate of 15 ml min-1. Surface areas of the heat-treated samples were determined by N2 adsorption–desorption in a NOVA1000 instrument, using the BET equation. Prior to the BET measurement, each calcined sample received a degas treatment under vacuum for 1 h at a temperature lower than the calcination temperature used.Catalytic activity evaluation For catalytic activity evaluation, the as-prepared MC1 and MC2 powders were vacuum-pressed under an external pressure of 6 tonnes cm-2. The pressed MC1 and MC2 plates were then divided and screened with sizes of 355 to 1000 mm. The catalytic activities of MC1 and MC2 were evaluated in a quartz-tube microreactor with a constant diameter of 1.0 cm.The above-prepared MC1 and MC2 (mass 0.605 g, volume, V 0.5×10-3 dm3) were added to the quartz reactor. Prior to the catalytic test the MC1 or MC2 was decomposed in situ under an N2O–He flow at 350°C. After the above heat-treatment, MC1 and MC2 had been converted to metal oxides owing to the depletion of OH groups.The N2O decomposition experiment was then carried out on the resulting oxide catalyst. In these experiments, the N2O gas was supplied continuously from a gas cylinder (27 mol% N2O+73 mol% He) at a flowrate (F) of 1.5 dm3 h-1 and the GHSV (F/V ) for the feed gas Fig. 1 FTIR spectra of the as-prepared precursor materials: thus corresponds to 3000 h-1.The decomposed gases were (a) Mg(OH)2 aged at 25 °C for 5 h, then dried at 60°C overnight cooled in a cooling coil and then vented through a scrubber. (25+60); (b) Mg(OH)2 hydrothermally aged at 65°C for 18 h, then The inlet and outlet gas compositions were analysed by gas dried at 60°C for 18 h (65+60); (c) Co(OH)2 aged at 25 °C for 5 h, chromatography on a Shimadzu GC-14A (TCD detector) using then air-dried at 25°C for 18 h (25+25); (d) Co(OH)2 (25+60); (e) Co(OH)2 (65+60); ( f ) MC1 (25+60); (g) MC2 (65+60) a Porapak Q column (4 m length) with He as the carrier gas 494 J.Mater. Chem., 1997, 7(3), 493–499lated via hydrogen bonding.19–21 Water bending vibrations are materials are well maintained before and after NO3-–CO32- exchange, especially for the anion-rich case MC1.The obser- also observable between 1611 and 1640 cm-1. Anionic species such as NO3- and CO32- (resulting from atmospheric CO2 vation here suggests the presence of a hydrotalcite-like phase in these catalyst precursors, taking the anion exchangeability dissolution) can also be observed. In particular, the bands at 1356–1365 cm-1 (n3), 835–828 cm-1 (n2) and 669–644 cm-1 into consideration. The formation of a hydrotalcite-like phase is confirmed by (n4) are assigned to the carbonate anion included in the material matrices while the sharp peaks at 1383–1385 cm-1 the XRD investigation.As shown in Fig. 3 for the anion-rich sample MC1, the broad diffraction peaks are indeed character- are assigned to the NO3- anion (n3 mode).20,21 The absorption at 2361 cm-1 in all the spectra is due to the detection of CO2 istics of hydrotalcite-like compounds.21,23,24 In the spectrum of MC2, however, this hydrotalcite feature is not observable; gas.Finally, peaks/bands atwavenumbers lower than 800 cm-1 can be attributed to metal–oxygen vibrations.21,22 only pronounced brucite-like diffraction peaks can be seen. This observation is consistent with the low anion content For the monohydroxide Mg(OH)2 (brucite), no NO3- inclusion is observable. In contrast to this, both NO3- and found in the elemental analysis.The presence of the hydrotalcite- like phase in the MC1 samples indicates the formation of CO32- absorptions are pronounced for Co(OH)2 prepared at room-temperature. Nevertheless, CO32- absorptions are the trivalent cation (M3+).21,23,24 Owing to the presence of air atmosphere and common oxidation states of +2 and +3 for reduced significantly in Co(OH)2 which underwent hydrothermal treatment.Note that for the sample of Co(OH)2 dried at Co, a certain portion of the original Co2+ cations lose one of their 3d7 electrons and become Co3+ during the precipitate 60°C [Fig. 1(d)], two sharp peaks attributable to OH can be observed at 3698 and 3631 cm-1. The former peak, with a formation.25 The anion inclusion in MC1 indeed reflects the above Co2+ oxidation process. When a Co2+ ion is oxidized value in close agreement with that for Mg(OH)2 (3700 cm-1), may suggest that a small portion of Co(OH)2 has been to give Co3+, the brucite-like sheet gains a positive charge.This positive gain can be balanced by an anion which possesses transformed to a more brucite-like phase upon the heat treatment. The sharp peaks of OH vibrations at 3644 and negative charge(s). Based on the elemental analysis data (Table 1), a deduced value of Co3+ 3d6:Co2+ 3d7 is ca. 23577, 3654 cm-1 for MC1 and MC2 have intermediate values between those of Mg(OH)2 and Co(OH)2, indicating the or, atomic ratio Co3+:(Co2++Mg2+)#19:81 in sample MC1 (and MC1C).formation of the double hydroxides of Mg–Co, i.e. mixed cations of both Mg and Co in the brucite-like structure. Based Nevertheless, it should be pointed out that both the hydrotalcite- like and brucite-like phases coexist in MC1, while the on the relative intensities of these peaks/bands, it is recognized that the water and carbonate anion contents in MC1 are much brucite-like phase is predominant in the MC2 samples (Fig. 3). In Fig. 4, the CO32-/OH- peak ratio of the MC1 series varies higher than in MC2. As revealed by the elemental analyses (Table 1), MC1 has a greatly with the NO3-–CO32- exchange time whereas that of MC2 remains almost constant over the same time range. This Co/Mg atomic ratio of 4.37 and MC2 has a Co/Mg=2.35.Clearly, the Co/Mg atomic ratio in the precipitates depends indicates either a gradual loss of the hydrotalcite structure (i.e. a decrease in the inclusion capacity of CO32-) or an increase mainly on the ageing temperature. The above elemental analysis data are also in good agreement with the FTIR results. For in the brucite-like phase (i.e.an increase in the number of OH- groups) for the prolonged anion-exchange experiments example, the OH absorption at 3644 cm-1 (MC1; high Co/Mg) is closer to 3631–3632 cm-1 of Co(OH)2 while the 3654 cm-1 (>60 min). In other words, MC1 is not as stable as MC2 with additional ageing in the basic aqueous solution. (MC2; lower Co/Mg ratio) is nearer to 3700–3698 cm-1 of Mg(OH)2. With the higher Mg content in the double hydroxide (MC2), the brucite-like sheet would generally be anticipated.21 Chemical/thermal stabilities and hydrothermal synthesis Furthermore, since the MC2 is hydrothermally treated at 60 °C In Fig. 5, the observed single endothermic peak of the for 18 h, higher crystallinity will also be expected. These two Mg(OH)2 sample corresponds to brucite decomposition, points are confirmed by the SEM morphological study although this peak is shifted to higher temperature (420.0 °C) reported in Fig. 2. Better crystallinity, greater grain sizes, more in the sample subjected to hydrothermal treatment at 60°C. regular aggregates and flatter surfaces for MC2 are indeed Compared to this, the decomposition occurs at a much lower observed. However, the calcination treatment at higher tem- temperature of 188.9 or 188.1 °C (Fig. 5) for the Co(OH)2 peratures obviously led to considerable modifications in surface samples without hydrothermal treatment. Interestingly, this morphology, which will be addressed later. peak shifts to 206.4 and 267.0 °C for the sample which underwent hydrothermal ageing. The latter (267.0 °C) can be attri- NO3- and CO32- anion exchange buted to a more brucite-like phase since the anion population (Fig. 1) is significantly lower for this sample. The small When the MC1 and MC2 samples are immersed into the Na2CO3 solution, the NO3- group can be replaced by CO32-, peaks located at low temperatures (166.7 and 170.2 °C) can be assigned to the removal of the water molecules trapped which is evidenced by the emergence of the CO32- absorption at 1365 cm-1 (n3)20,21 that is originally masked by the NO3- in the interlayer spacing between the brucite-like sheets, i.e.the gallery water molecules,21,23–27 because these cobalt absorption (1383–1385 cm-1) in most NO3--containing samples (Fig. 1). Based on the elemental analyses (Table 1), it hydroxides also show hydrotalcite-like features in their XRD spectra.is found that total negative charges borne by the anions of Table 1 Preparation conditions and elemental analysis results for MC1, MC2 and their CO32--exchanged samples elemental analysis initial precipi- precipi- ageing/drying Co5Mg tation tating temperature CO32- NO3- NO3-/CO32- Mg Co Co/Mg Co3+/Co2+ sample ratio pH atmosphere /°C (mass%) (mass%) (mol ratio) (mass%) (mass%) (mol ratio) (mol ratio) MC1 1 9.5 air 25;60 1.25 7.69 5.96 4.06 42.96 4.37 23577 MC2 1 9.5 air 65;60 1.30 0.40 0.30 7.60 43.44 2.35 — MC1Ca 1 9.5 air 25;60 5.35 0 0 3.90 45.21 4.79 23577 MC2Ca 1 9.5 air 65;60 2.45 0.35 0.14 7.39 44.10 2.46 — aMC1 or MC2 impregnated in a 0.5 mol dm-3 Na2CO3 aqueous solution for 1 h under stirring at room temperature.J. Mater. Chem., 1997, 7(3), 493–499 495Fig. 2 SEM images of: (a) as-prepared MC1; (b) as-prepared MC2; (c) MC1 after calcination at 350 °C for 2 h; (d) MC2 after calcination at 350 °C for 2 h In line with the XRD results (Fig. 3), the DSC plot of MC1 calcined MC1 and MC2 samples. For the as-prepared MC1 reported in Fig. 6 also suggests the formation of a hydrotalcite- and MC2, metal–oxygen vibration absorptions are located at like phase.The endothermic peak at 113.1 °C can be assigned 511 and 482 cm-1 respectively. With increasing calcination to gallery water molecule depletion,21,23–27 since water adsorp- temperatures, the spectrum evolution reveals the formation of tion/desorption is reversible at this temperature. The main mixed metal oxides, i.e.a spinel-type phase.22 The doublet peak at 229.6 °C, however, is attributable to the collapse of the peaks for the high-temperature calcined samples are similar to hydrotalcite-like phase.21,23–27 Based on the DSC results for those reported in the literature for pure Co3O4 (661 and hydrothermally treated Co(OH)2, the high-temperature peak 568 cm-1).28,29 In the current study, doublet peaks of monohyof 282.7 °C can be ascribed to the decomposition of the droxide Co(OH)2 calcined at 500 °C for 2 h are located at 664 hydroxide phase remaining in MC1, which was detected in the and 567 cm-1, indicating the formation of a Co3O4 phase.XRD study. Although no direct evidence is given, the tiny Note that the wavenumber of the first peak of MC1 (652 cm-1) peak at 192.8 °C may be due to the desorption of surface is higher than that of MC2 (646 cm-1), i.e.with increasing Mg adsorbed water prior to the collapse of the hydrotalcite frame- content, the peak shifts towards the lower wavenumber region. work, using a similar explanation in other hydrotalcite The departure from 664 cm-1 (of Co3O4) can be attributed to compounds.26 the inclusion of Mg (more M2+ species) into the Co3O4 spinel For hydrothermally heated samples MC2, the thermal phase.From the decoupling temperature of the doublet peaks, behaviour is substantially different. For example, the gallery it can be concluded that MC2 is thermally more stable. The water desorption at low temperatures is barely observed. The decomposition of the catalyst precursors is also reflected in decomposition temperatures, on the other hand, have been the XRD spectra of Fig. 3(c) and (d ) (at 350 °C), indicating that shifted to higher values, owing to the low Co/Mg ratio in the resultant oxides are largely amorphous, although a weak MC2. Therefore, the thermal behaviour of MC2 can be spinel-type feature can be observed in Fig. 3(c).30 described as being between the hydrothermally treated Fig. 8 shows the results of BET measurements for the above Mg(OH)2 and Co(OH)2 (Fig. 5), noting that with the pro- two calcined sample series. When the hydroxide/hydrotalcite longed thermal drying, endothermic peaks for MC2 move phases are converted to metal oxides, the surface area varies further to the high-temperature side. substantially. For example, the surface area of MC1 reaches a maximum at 250 °C and later declines at higher temperatures Calcination and surface areas since it is Co-rich and decomposes readily.For the MC2 series, however, the surface area does not reach a maximum until As was addressed in the previous section, MC2 with the high 300°C, since it is more stable to sustained thermal treatment. Mg content is more thermally stable than MC1.This observation is further reflected in the FTIR spectra of Fig. 7 for In both cases, the maximum surface area is obtained when the 496 J. Mater. Chem., 1997, 7(3), 493–499Fig. 5 DSC plots of the monohydroxide samples: (a) Mg(OH)2 (25+60); (b) Mg(OH)2 (65+60); (c) Co(OH)2 (25+25); (d) Co(OH)2 (25+60); (e) Co(OH)2 (65+60) Fig. 3 XRD spectra of: (a) as-prepared MC1 (HT indicates the hydrotalcite phase); (b) as-prepared MC2; (c) MC1 calcined at 350°C for 2 h; (d) MC2 calcined at 350 °C for 2 h Fig. 4 FTIR peak ratio of CO32-/OH- vs. NO3-–CO32- exchange Fig. 6 DSC plots of the as-prepared double-hydroxide samples MC1 time; the FTIR peak height of CO32- is based on the absorption at (25+60), MC2 (65+60), and prolonged heat-dried MC2 (65+60): 1365 cm-1 (n3 ), and the height of OH- is measured from the sharpest/ (a) MC2 with a total drying time of 42 h at 60 °C (65+60/42h); strongest peak assigned for the double hydroxide framework (b) MC2 (65+60/66h); (c) MC2 (65+60/114h) Catalytic performances hydroxide/hydrotalcite framework collapses, i.e.it occurs at metal oxide formation temperatures. The decrease in surface In the current investigation, all the Mg–Co oxides studied exhibit reasonable catalytic activities for N2O decomposition.area at higher calcination temperatures can be explained by oxides grain growth. As shown in Fig. 9, MC1 and MC2 give a noticeable activity J. Mater. Chem., 1997, 7(3), 493–499 497Fig. 8 BET measurements for the calcined MC1 and MC2 sample series; two sets of data are averaged for each sample series Fig. 9 Catalytic activity [X, eqn. (1)] evaluation for the MC1 and MC2 catalysts; prior to the catalytic test, as-prepared MC1 or MC2 was decomposed in situ under the flow of the reaction mixture at 350°C optimization of the Mg/Co ratio and other processing parameters are needed in further study. Conclusions In summary, binary metal oxide catalysts of Mg–Co have been synthesized from Mg–Co hydroxide/hydrotalcite precursors for low-temperature N2O decomposition.Higher Mg contents and higher crystallinity are observed in the hydrothermally treated precursor. With the anion-exchange method and XRD, ahydrotalcite-like phase is also found due to a partial oxidation of Co2+ to Co3+. The catalyst prepared from the hydrothermally treated precursor is more active and stable compared Fig. 7 FTIR spectra of calcined MC1 (a) and MC2 (b) sample series with the one aged at room temperature. All precursor materials at different elevated temperatures give amorphous Mg–Co oxides after thermal decomposition. The resultant oxides have shown high catalytic activity for N2O decomposition. At 350 °C, approximately 6 moles of N2O at temperatures as low as 250–275 °C and 275–300 °C respect- per kg of precursor hydroxide/hydrotalcite can be decomposed ively.Under these experimental conditions, approximately 6 within 1 h. moles of N2O per kg of the hydroxide/hydrotalcite can be decomposed at 350 °C within 1 h, which is comparable to some The authors gratefully acknowledge the research funding of the most active catalysts reported so far.1–18 It is recognized (RP3950619) supported by the National University of that the activity difference between MC1 and MC2 is not due Singapore for the experimental study of catalytic materials.solely to surface area variations. For example, the surface area H. C. Z. would also like to thank Dr. T. A. Koch and Dr. of MC2 is ca. 40% higher than that of MC1 after calcination J. C. Wu of E. I. DuPont de Nemours & Co., Inc., USA, for (Fig. 8), whereas MC2 is ca. 200% more active than MC1 at the recommendation of this research topic. 350 °C (Fig. 9). Since the atomic mass of Mg is much lower than that of Co, the total amount of metal cations in the MC2 References catalyst is higher than that in MC1. It is therefore suggested that the higher catalytic activity observed for MC2 is attribu- 1 M.H. Thiemans and W. C. Trogler, Science, 1991, 251, 932. table to the larger number active sites formed by Mg–O–Co 2 Y. Li and J. N. Armor, Appl. Catal. B, 1992, 1, L21; 1993, 3, 55; US Pat., 1992, 5 149 512; 1992, 5 171 553. with an appropriate atomic ratio of Mg/Co. Apparently, the 498 J. Mater.Chem., 1997, 7(3), 493–4993 A. P. Kinzig and R. H. Socolow, Phys. T oday, 1994, 47, 24. 18 X. Y. Pang, H. C. Zeng, J. C. Wu and K. Li, Appl. Catal. B, 1996, 9, 149. 4 H. C. Zeng, J. Lin, W. K. Teo, J. C. Wu and K. L. Tan, J. Mater. Res., 1995, 10, 545. 19 F. Trifiro, A. Vaccari and G. D. Piero, in Characterization of Porous Solids, ed. K. K. Unger, Elsevier Science, Amsterdam, 5 H.C. Zeng and M. Qian, J. Mater. Chem., 1996, 6, 435. 6 Y-F. Chang, J. G. McCarty, E. D. Wachsman and V. L. Wong, 1988, p. 571. 20 E. C. Karuissink, L. L. Van Reijen and J. R. H. Ross, J. Chem. Soc., Appl. Catal. B, 1994, 4, 283. 7 G. D. Lei, B. J. Adelman, J. Sarkany and W. M. H. Sachtler, Appl. Faraday T rans. 1, 1981, 77, 649. 21 F. Cavani, F. Trifiro and A. Vaccari, Catal. T oday, 1991, 11, 173. Catal. B, 1995, 5, 245. 22 G. Busca, F. Trifiro and A. Vaccari, L angmuir, 1990, 6, 1440. 8 L.M. Aparicio and J. A. Dumesic, J.Mol. Catal., 1989, 49, 205. 23 W. T. Reichle, Solid State Ionics, 1986, 22, 135. 9 J. Valyon, W. S. Millman and W. K. Hall, Catal. L ett., 1994, 24, 24 K. A. Carrado, A. Kostapapas and S. L. Suib, Solid State Ionics, 215. 1988, 26, 77. 10 G. I. Panov, V. I. Sobolev and A. S. Kharitonov, J. Mol. Catal., 25 H. C. Zeng, M. Qian and Z. P. Xu, unpublished work. 1990, 61, 85. 26 S. K. Yun and T. J. Pinnavaia, Chem. Mater., 1995, 7, 348. 11 B. W. Riley and J. R. Richmond, Catal. T oday, 1993, 17, 277. 27 F. Rey, V. Fornes and J. M. Rojo, J. Chem. Soc., Faraday T rans., 12 P. Pomonis, D. Vatts, A. Lycourghiotis and C. Kordulis, J. Chem. 1992, 88, 2233. Soc., Faraday T rans. 1, 1985, 81, 2043. 28 H. D. Lutz and M. Feher, Spectrochim. Acta, Part A, 1971, 27, 357. 13 A. K. Ladavos and P. J. Pomonis, Appl. Catal. B, 1993, 2, 27. 29 J. Preudhomme and P. Tarte, Spectrochim. Acta, Part A, 1971, 14 R. Sundararajan and V. Srinivasan, Appl. Catal., 1991, 73, 165. 27, 1817. 15 S. L. Raj, B. Viswanathan and V. Srinivasan, J. Catal., 1982, 75, 30 G. Fornasari, S. Gusi, F. Trifiro and A. Vaccari, Ind. Eng. Chem. 185. Res., 1987, 26, 1500. 16 D. D. Eley, A. H. Klepping and P. B. Moore, J. Chem. Soc., Faraday T rans. 1, 1985, 81, 2981. 17 S. Kannan and C. S. Swamy, Appl. Catal. B, 1994, 3, 109. Paper 6/07627K; Received 8th November, 1996 J. Mater. Chem., 1997, 7(3), 493–499 499

ISSN:0959-9428    

DOI:10.1039/a607627k

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

Synthesis of yttrium iron garnet nanoparticles via coprecipitation in microemulsion Paz Vaqueiro,a M. Arturo Lo�pez-Quintelaa and Jose� Rivasb aDepartamento de Quý�mica-Fý�sica, Facultad de Quý�mica, Universidad de Santiago de Compostela, E-15706, Spain bDepartamento de Fý�sica Aplicada, Facultad de Fý�sica, Universidad de Santiago de Compostela, E-15706, Spain We describe a new technique using microemulsions to produce ultrafine precursors of yttrium iron garnet. A coprecipitation of hydroxide or carbonate precursors was made in a W/O microemulsion medium.These precursors (ca. 3 nm in size), when heated above 700 °C, transformed to YIG phase. Once prepared, YIG nanoparticles were characterized by X-ray diffraction, transmission electron microscopy and dc magnetic measurements. Yttrium iron garnet,1 Y3Fe5O12 (YIG), is a material used heptane as the continuous oil phase and an aqueous solution widely in electronic devices for the microwave region as well as the dispersed phase.Two microemulsions, A and B, with as in magnetic bubble domain-type digital memories. Some identical compositions but with different aqueous phases were magnetic properties, such as saturation magnetization, reman- prepared:in microemulsion A the aqueous phase was a solution ence and coercitivity, depend critically on the structure and of iron and yttrium nitrates in stoichiometric ratio, whereas microstructure of the materials.Therefore, it is important to in microemulsion B the aqueous phase was a solution of develop techniques to produce garnets with a strict control of ammonium hydroxide or ammonium carbonate.Table 1 shows the composition, homogeneity, size and particle shape. Also the composition of the prepared microemulsions, and in Fig. 1 the garnets have a uniquely defined cation distribution and do the preparation scheme is presented. Table 2 shows the samples not present any site inversion problems which can arise in prepared and the concentrations of the precipitating agents other ferrites;because of this, this kind of ferrimagnetic material [aq.NH3 and (NH4)2CO3] used.is very suitable for magnetism studies. Chemical precipitation can be performed by the normal In recent years, great interest has been focused on the study strike method, i.e. adding the basic solution to the acidic of the dependence of the physical and chemical properties on solution, or by the reverse strike technique, adding the acidic the particle size.For this reason, there is renewed interest in solution to the basic solution. Iron and yttrium ions precipitate the development of new techniques to produce particles of at quite different pH values, so better results can be achieved different sizes, sufficiently monodispersed and with a good by using the reverse strike technique.9 The addition of the salt grade of homogeneity.Water/oil (W/O) microemulsions are solution to the base results in an almost instantaneous increase ideal media for the preparation of ultrafine particles owing to of the pH, well above the precipitation pH values of both salts. the small water microdroplets (ca. 10 nm) contained inside This will produce more homogeneous precipitates. We used which, under appropriate conditions, will restrain the growth this technique in all cases, adding the microemulsion A into of the particles. Metal particles and metal oxide particles the microemulsion B. have been synthesized already by reaction in W/O micro- After mixing microemulsions A and B, precursor particles emulsions.2–4 Recently, some mixed-metal oxides have been appeared within the aqueous domains of the microemulsion.prepared from oxalate or carbonate precursor particles which These precipitate particles were separated in an ultracentrifuge were precipitated in microemulsion media: high-temperature at 5000 rpm for 10 min. The precipitate was washed twice with superconductors, YBa2Cu3O7-x, have been synthesized by ethanol and then dried at 80°C.To obtain the YIG particles, coprecipitation of oxalate particles;5 barium ferrite, BaFe12O19, the dried precipitates were calcined at 300 °C for 1 h, then at was obtained by coprecipitation of carbonate particles,6 and 400°C for 1 h and finally annealed at different temperatures.some perovskites, like LaNiO3, La2CuO4 and BaPbO3 , were Before the use of microemulsions with Igepal CA-520 as also obtained by coprecipitation of oxalate particles.7 In this surfactant, we studied the coprecipitation of YIG precursors study, we describe the use of W/O microemulsions to obtain in the CTAB/butanol/octane/water system and we observed Y3Fe5O12 particles. that it was quite difficult to remove all the surfactant from the precipitate. After 6 washings with ethanol and ethanol– chloroform, we detected the presence of Br- (from the CTAB) Experimental by adding a drop of a solution of silver nitrate to the washing liquid.Because of this problem we changed the composition All chemicals used in this work were reagent grade. of the microemulsion and used Igepal CA-520, which can be Fe(NO3)3 9H2O, Y(NO3)3 5H2O, Igepal CA-520 (penta- removed more easily with the thermal treatment of the ethyleneglycol monoisononyl phenyl ether), (NH4)2CO3 and precipitate.heptane were provided by Aldrich and aqueous NH3 by The hydrodynamic size of the microemulsion droplets was Panreac. Igepal CA-520 is a commercially available non-ionic determined from photon correlation spectroscopy (PCS) surfactant which forms W/O microemulsions in the system measurements made with an Ar laser Liconix series ALV-SP80 Igepal CA-520/heptane/water with the concentrations used controlled automatically by means of an ALV-LSE (Light here (see Table 1).A detailed study of the system Igepal Scattering Electronics) goniometer controller unit.The size CA-520/decane/water is reported elsewhere.8 and the morphology of the precursors and the annealed The precursor to YIG was precipitated from a solution samples were studied by transmission electron microscopy containing iron(III ) nitrate and yttrium(III ) nitrate, by treatment (TEM), using a Philips CM-12 instrument. The annealed with ammonium hydroxide or ammonium carbonate.To samples were characterized by X-ray diffraction (XRD), with restrain the growth of precursor particles, a water-in-oil microemulsion was used with Igepal CA-520 as the surfactant, a Philips PW-1710 diffractometer using Cu-Ka radiation J. Mater. Chem., 1997, 7(3), 501–504 501Table 1 Compositions of the microemulsion systems surfactant oil phase aqueous phase Igepal CA-520 heptane Fe(NO3)3 (0.2 mol dm-3)+Y(NO3)3 (0.12 mol dm-3) microemulsion A microemulsion B Igepal CA-520 heptane aq.NH3 or (NH4)2CO3 mass fractions 38.13% 53.87% 8% observed in the morphology of precipitates with different precipitating agents; for both precipitates the TEM images show aggregates constituted of very fine particles with sizes of ca. 3 nm (see Fig. 2). YIG particles XRD patterns of the samples annealed at 600 °C exhibit a broad band that can be attributed to an amorphous material, but also some weak peaks appear, attributed to maghaemite (c-Fe2O3 ), haematite (a-Fe2O3) and some small quantity of YIG (see Fig. 3). After annealing at 700 °C for 2 h, XRD patterns show the characteristic peaks of YIG, although the samples are not completely crystallized because the amorphous broad band remains.XRD patterns recorded on samples annealed above 700 °C do not exhibit any broad band. However, depending on the precipitating agent, phases other Fig. 1 Schematic representation of the synthesis process Table 2 Annealed samples precipitating agent/ annealing samplea mol dm-3 temp./°C annealing time/h N1 1.7 600 2 N2 1.7 700 2 N3 1.7 800 2 N4 1.7 800 4 N5 1.7 800 12 N6 1.7 900 2 N7 1.7 1000 2 N8 1.5 800 2 N9 2 800 2 C1 0.6 600 2 C2 0.6 700 2 C3 0.6 800 2 C4 0.6 900 2 C5 0.4 800 2 aN, coprecipitation with aq.NH3; C, coprecipitation with (NH4)2CO3. (l=1.54186 A° ). The different phases were identified using the JCPDS Powder Diffraction Files.10 Magnetic measurements Fig. 2 TEM image showing the precipitate obtained using were performed using a vibrating meter (VSM) microemulsion B with 0.6 mol dm-3 ammonium carbonate as the model DMS 1660, and hysteresis loops were recorded at room aqueous phase temperature up to 2 kOe.Results and Discussion Microemulsion system and precursor particles The hydrodynamic radius of the microemulsion droplets (without any salt added to the aqueous phase) as determined from PCS measurements was 7 nm.The addition of a salt to the aqueous phase can change the radius values, but we did not make measurements of the size of microemulsions with salts in the aqueous phase because microemulsions containing metal salts are coloured and absorb the light. After mixing microemulsions A and B, it was observed that the aggregation rate of the particles within the microemulsion depends on the concentration of the precipitating agent, and this rate increases with the increase of the concentration of precipitating agent in microemulsion B.The washed precipitates were dispersed in ethanol and a Fig. 3 XRD pattern of the C1 sample (annealed at 600°C). &, Maghaemite; #, YIG; 1, haematite. drop was deposited on a TEM grid.No differences were 502 J. Mater. Chem., 1997, 7(3), 501–504than YIG appear. The diffractogram of the C5 sample [pre- is an intrinsic property of a magnetic material; it is reported that small magnetic particles exhibit lower magnetization pared with 0.4 mol dm-3 (NH4)2CO3] exhibits the lines of YIG, haematite and perovskite (YFeO3; see Fig. 4), and these values than that of the bulk.13 The coercive field decreases with the annealing temperature due to the fact that coercitivity three phases also appear in the XRD pattern of the N8 sample, prepared with 15 mol dm-3 aq.NH3.The diffractograms of depends strongly on the size of the particles, and as observed by TEM measurements, the particle size increases with the samples synthesized with larger amounts of precipitating agent exhibit only the characteristic peaks of YIG (see Fig. 5). annealing temperature. By coprecipitation of hydroxides or carbonates in an aqueous solution, pure crystalline YIG was produced after heating at 760 °C,11 so the crystallization temperature does not decrease with the use of microemulsion media. However, for the La–Ni, La–Cu and Ba–Pb oxalate precursors synthesized by Gan et al.7 using a microemulsion technique, a decrease of the crystallization temperature of the samples prepared by the aqueous coprecipitation was reported.Fig. 6 shows a micrograph of C3 (annealed at 800°C). Very large aggregates (ca. 1mm) can be observed, which are composed of irregular fine particles. With the increase of the annealing temperature, the particles grow and a sintering process occurs.Crystallized particles are very much larger than precursor particles, as observed also for other mixed oxides synthesized by coprecipitation in microemulsion media. Fig. 7 shows hysteresis loops recorded at room temperature for the C1 sample, annealed below the crystallization temperature, and for the C3 sample, annealed above this temperature. A marked effect on the magnetization value with the crystallization process is observed.Table 3 shows the values of coercive field, Hc, and magnetization, Ms, measured at 2 kOe for the crystallized samples. The magnetization increases with the annealing temperature, but always remains lower than the bulk value of 26 emu g-1.12 This discrepancy might be due to the existence of amorphous impurities undetected by XRD or Fig. 6 TEM image showing the morphology of the C3 sample perhaps to the small particle size. However, the magnetization (annealed at 800 °C) Fig. 4 Diffractogram of the C5 sample, showing the obtained phases. &, Haematite; $, YIG; %, perovskite. Fig. 7 Hysteresis loops of samples C1 (upper) and C3 (lower) recorded at room temperature Table 3 Magnetic measurements sample Ms/emu g-1 Hc/Oe N3 21.8 25 N4 21.7 27 N6 22.4 17 N7 23.9 17 C3 23.0 36 C4 24.1 20 Fig. 5 Diffractogram of the C3 sample (annealed at 800°C) J. Mater. Chem., 1997, 7(3), 501–504 5032 P. Barnickel, A. Wokaun, W. Sager and H-F. Eicke, J. Colloid Conclusions Interface Sci., 1992, 148, 80. 3 J. B. Nagy and A. Claerbout, in Surfactants in Solution, ed.A new method for the synthesis of yttrium iron garnet has K. L. Mittal and B. Lindmann, Plenum Press, New York, 1990, been developed, using W/O microemulsions as the reaction vol. 11, p. 366. medium to produce ultrafine precursor particles. These precur- 4 M.A.Lo� pez-Quintela and J. Rivas, J. Colloid Interface Sci., 1993, sors are calcined to produce yttrium iron garnet, as confirmed 158, 446. 5 P. Ayyub, A. N. Maitra and D. O. Shah, Physica C, 1990, 168, 571. by XRD and magnetization measurements. The synthesis 6 V. Pillai, P. Kumar and D. O. Shah, J.Magn. Magn.Mater., 1992, temperature is 500 °C lower than that for yttrium iron garnet 116, L299. prepared by the conventional ceramic method. This technique 7 L.M. Gan, L. H. Zhang, H. S. O. Chan, C. H. Chew and B. H. Loo, can also be applied to other mixed oxide systems. J. Mater. Sci., 1996, 31, 1071. 8 A. Ferna�ndez No�voa, PhD Thesis, University of Santiago de Compostela, Spain, 1991. P. V. wishes to acknowledge Xunta de Galicia for its 9 P. Apte, H. Burke and H. Pickup, J.Mater. Res., 1992, 7, 706. 10 JCPDS files, Swarthmore, Pennsylvania, PA. financial support (XUGA209031395). 11 K. R. Nair, Ceram. Bull., 1981, 60, 626. 12 Landolt-Bo�rnstein, Numerical Data and Functional Relationships in Science and T echnology, new series, ed.K. H. Hellwege, Springer- Verlag, Berlin, 1978, Series III, vol. 12-a. References 13 K. Haneda and A. H. Morrish, Nucl. Instrum. Methods B, 1993, 76, 132. 1 M. A. Gilleo, in Ferromagnetic Materials: a handbook of the properties of magnetically ordered substances, ed. E. P. Wohlfarth, North-Holland, Amsterdam, 1980, vol. 2. Paper 6/05403J; Received 2nd August, 1996 504 J. Mater. Chem., 1997, 7(3), 501&n

ISSN:0959-9428    

DOI:10.1039/a605403j

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

Phases occurring in the Si3N4–YN system Thommy C. Ekstro�m, Kenneth J. D. MacKenzie, Martin J. Ryan, Ian W. M. Brown and G. Vaughan White New Zealand Institute for Industrial Research and Development, P.O. Box 31-310, L ower Hutt, New Zealand A method is described for synthesising complex silicon lanthanide nitrides by carbothermal reduction and nitridation (CRN) of both SiO2–Ln2O3 mixtures and elemental Si–Ln2O3 mixtures at 1400–1450°C.Syntheses in the Si3N4–YN system confirm the existence of the compounds YSi3N5 and Y2Si3N6, which can be prepared virtually free of oxygenated impurity phases from mixtures containing SiO2 rather than from mixtures containing elemental Si. A third compound was also prepared in this system, with an XRD pattern corresponding to that reported for Y6Si3N10; mass change data and EDX analyses suggest, however, that this phase is much less Y-rich than Y6Si3N10, with an apparent composition nearer Y3Si6N11.X-Ray powder diffraction data and 29Si and 89Y solid-state MAS NMR spectra are reported for all three phases. The NMR results suggest the presence of at least three Si sites in each phase, and at least two types of Y site, one of which, characterised by a resonance at d 490–510 (ref.YCl3), is common to all three compounds. The phase relationships in the Y–Si–Al–O–N system have and in the interior of the Si3N4–YN–AlN system the a-sialon phase has an extended linear solid solution Yy/3Si12-yAlyN16. been researched extensively.1–6 The Si3N4-rich corner of the The extent of this solid solution range has been variously phase diagram is of special interest as this contains phases reported as y=1.8–3.4 at 1750 °C1 or y=1.3–2.4 at 1800 °C.5 such as a- and b-sialon, which are important for tailoring Previous studies of the pseudo-binary Si3N4–YN system engineering ceramics.7 The b-sialon phase, Si6-zAlzOzN8-z have mainly concentrated on very high temperature reactions coexists with a-sialon, YxSi12-(m+n)Alm+nOmN16-n, which between the nitrides.Three phases, YSi3N5, Y2Si3N6 and forms in a restricted two-dimensional compositional area in Y6Si3N10, were reported by Thompson,1 but in a later work the plane Si3N4–Al2O35AlN–YN53AlN (Fig. 1). The Si3N4 only Y2Si3N6 was found, the XRD patterns reported for the corner of the illustrated Ja�necke prism contains, in addition other two nitrides by Thompson being assumed by the sub- to the two sialon phases, the nitrogen-rich phases N-melilite sequent authors5 to be a mixture of N-melilite and other and JEM.Melilite solid solutions, Ln2Si3-xAlxO3+xN4-x, have oxygen-containing phases. It should be stressed, however, that been found to form with Y and in Ln-sialon systems with the later paper5 reports only five preparations in the Si3N4–YN lanthanides of atomic number Z=58–66.8,9 Recent studies system, all of which include at least one oxygen-containing have revealed a new N-rich phase called JEM with a composiphase in addition to Y2Si3N6; the additional oxygenated phases tion LnAl(Si6-zAlz)(N10-zOz).This has so far been found to were typically N-melilite (Y2Si3O3N4 ), J-phase (Y4Si2O7N2) or form only with Ln ions of atomic number Z=57–62; a even Y2O3.5 corresponding Y-JEM phase has not been found.10,11 Discrepancies in the literature seem to be due, at least in The possible formation of other N-rich compounds is diffi- part, to difficulties in achieving an absolutely oxygen-free cult to assess because of the difficulty of synthesising pure environment when working with nitride starting materials.nitrides in the Si3N4–YN–AlN system. Silicon and aluminium Alternative lower-temperature synthesis techniques include nitride powders inherently contain some oxygen, and yttrium carbothermal reduction of silica and yttria at about 1400 °C nitride powder is a very sensitive starting material which accompanied by simultaneous nitridation, or a similar reaction hydrolyses readily unless manipulated in a dry-box, and even in which silica is replaced by elemental silicon.The use of very at moderate temperatures is oxidized rapidly. Further, the fine grained carbon will initiate reduction of the oxides, and reaction rate between YN and covalently bonded nitrides is in the presence of an excess of nitrogen gas, the reaction will very low even at high temperature. produce Si–Y nitrides with compositions depending on the No phase has been reported along the Si3N4–AlN tie line, ratios of the starting materials.A small excess of carbon assures an oxygen-free reaction environment. Although carbothermal reduction and nitridation (CRN) has been used for the preparation of b-sialon powders from natural clays and minerals,12,13 the technique has not been reported previously for syntheses in the Si–Y–N system.In the synthesis of sialons, the CRN process involves a complex reaction sequence which depends on the character of the precursors and the process parameters. It is generally accepted that gas-phase reactions involving SiO and CO are involved in sialon formation. A number of possible reactions have been suggested, which may proceed at different temperatures, sometimes even parallel to each other.13–15 This paper describes the phase relationships in the system Si3N4–YN, deduced from samples prepared by the CRN technique, using both SiO2–Y2O3 mixtures and elemental Si–Y2O3 mixtures, both in combination with carbon.We report Fig. 1 Schematic illustration of the Ja�necke prism for Ln–Si–Al–O–N X-ray powder diffraction data for the product phases YSi3N5, systems where Ln=Y or the rare-earth elements. All the N-rich phases Y2Si3N6 and Y3Si6N11, and also their 89Y and 29Si solid-state in the Si3N4 corner are marked, but note that melilite, a-sialon and JEM form only with a limited number of Ln elements, see text. MAS NMR spectra. J.Mater. Chem., 1997, 7(3), 505–509 505acted Y2O3 (where the yttria content was very high). Raising Experimental the temperature to 1450–1475 °C increased the degree of The compositions of the present CRN samples were calculated nitridation, making it possible to prepare mixtures of YSi3N5 so as to lie along the Si3N4–YN tie line. Mixtures were and a phase with the XRD pattern reported1 for Y6Si3N10.prepared corresponding to Si3N45YN ratios of 251, 151, 152, However, single-phase preparations were not achieved using 153 and 156. The starting materials were carbon (Degussa elemental Si and, surprisingly, Y2Si3N6 was never formed in Lampblack), SiO2 (Silica Superfine, d50=1.45 mm, Commercial this system. Minerals Ltd), elemental Si (Sicomill, grade 4D, d50=4.6 mm, By contrast, CRN reactions in mixtures of SiO2 and Y2O3 KemaNord Industrikemi, Sweden) and Y2O3 (Fine, grade C, produced phases with the XRD patterns1 of YSi3N5, Y2Si3N6 H.C. Starck, Berlin). The amount of carbon added corre- and Y6Si3N10, and, in some samples, even YN. It was possible sponded to the calculated amounts for complete reaction plus to prepare almost pure Y2Si3N6 from a starting mix of the an excess of 10%.The samples were heated in alumina crucibles appropriate composition; EDXA indicated the bulk Y5Si ratio placed in an alumina boat in a horizontal tube furnace (40 mm of this phase to be 0.65, in good agreement with the theoretical i.d.). The dynamic reaction atmosphere was nitrogen (high- ratio of 0.66. purity, oxygen-free), flow rate 100 ml min-1. The holding Mixtures with the composition Y6Si3N10 were found to be temperature (1400–1450 °C) was approached at a heating rate not very reactive below ca. 1450 °C, while above this tempera- of 10°C min-1 with the holding time at the final temperature ture, the loss of volatile Si species was too great for the varying from 8 to 48 h.A small amount (ca. 70mmol g-1) of stoichiometry to be maintained. However, mixtures of YSi3N5 Eu2O3 was added before heating to the samples prepared for composition were found to give an essentially pure phase with examination by MAS NMR spectroscopy. The purpose of thiseported Y6Si3N10 XRD powder pattern after reaction at addition, which did not change the phase composition of the 1475°C for 8 h.This reaction was accompanied by a mass loss products, was to increase the relaxation rate of both 89Y and 8–12% greater than the theoretical loss for formation of 29Si and enable these spectra to be acquired.16 YSi3N5 from this mixture, suggesting the removal of additional All samples were carefully weighed before and after the Si and the formation of a more Y-rich phase.However, the reaction, and the reaction products were stored in a desiccator mean bulk Y5Si ratio determined by EDXA (0.39) indicates before examination by X-ray powder diffraction (XRD) in a that only slight Y-enrichment has occurred, and certainly not dry flowing nitrogen atmosphere using a Philips PW1700 to the extent required for the formation of Y6Si3N10 (Y5Si computer-controlled diffractometer with graphite mono- ratio 2).That this phase is not Y6Si3N10 is also suggested by chromator, Co-Ka radiation and APD1700 software. Selected the observed mass loss, which is much less than the theoretical specimens were examined using a Huber 620 Guinier X-ray value for the conversion of 6YSi3N5�Y6Si3N10 (48%). On the camera and Y5Si ratios were determined on the bulk material basis of the mass change and the EDXA data, a possible by EDXA (the materials were too fine-grained to permit composition for this phase may be Y3Si6N11; this would have analysis of the individual particles). The room-temperature 89Y a Y5Si ratio of 0.5 and its formation would be accompanied and 29Si MAS NMR spectra were acquired at 11.7 T using a by a mass loss of 19.2%.It should be noted that the original Varian Unity 500 spectrometer with a 5 mm Doty MAS probe spinning at 10–12 kHz, under the following conditions: 89Y: 20 ms p/2 pulse, recycle delay 3 s, 170 ms ringdown delay between excitation and sampling, shifts referenced to aqueous 1 mol dm-3 YCl3 solution; 29Si: 6 ms p/2 pulse, recycle delay 30 s, shifts referenced to tetramethylsilane. Results and Discussion Regardless of the starting mixture, all the CRN samples experienced a mass loss 3–5% higher than the theoretical loss, due to the removal from the sample of gaseous SiO, which condensed as a fine wool-like material on the cooler downstream furnace tube walls.XRD showed this condensed material to be amorphous; solid-state 29Si NMR studies indicated a preponderance of SiMO bonds, with evidence of some SiMC bonds also, probably formed by the presence of CO gas during SiO condensation.This migration of material made it difficult to predict the final composition of the product, which, however, proved in all cases to be a Y–Si nitride with a Y content slightly elevated over the batched composition. Once formed, the Y–Si nitride appeared to be stable under prolonged heating in nitrogen and did not decompose with time. In the preliminary experiments, the odour of ammonia was sometimes detected during grinding of the products in air, denoting the presence of hygroscopic nitrides. All subsequent preparations were therefore immediately desiccated and the XRD traces obtained under dry nitrogen.Further investigations showed that the samples were only hygroscopic when YN or a small amount of amorphous nitride (probably YN) was indicated by XRD.The different crystalline Y–Si nitrides, once formed, proved to be fully stable to ambient air. It proved difficult to obtain oxygen-free products by CRN reaction of Si–Y2O3 mixtures; the products at 1400 °C always Fig. 2 Portions of typical XRD powder diffractograms of Y–Si–N contained, in addition to the Y–Si–N phases, significant phases: A, YSi3N5; B, Y2Si3N6; C, Y3Si6N11.Impurity phases: *, Y3Si6N11; †, YSi3N5; o, J-phase. amounts of J-phase (Y4Si2O7N2 ) and sometimes even unre- 506 J. Mater. Chem., 1997, 7(3), 505–509Table 1 XRD powder diffraction data for YSi3N5,a Y2Si3N6b and Y3Si6N11c YSi3N5 Y2Si3N6 Y3Si6N11 h k l dobs/A° (I/I0 )obs h k l dobs/A° (I/I0)obs h k l dobs/A° (I/I0)obs 1 0 0 8.48 6 2 1 0 6.44* 1 1 1 0 5.1583L 11 1 0 1 6.63 12 0 0 1 4.975 9 0 2 0 5.1333K 0 0 2 5.32 9 1 0 1 4.754 25 0 0 2 4.9455 4 1 1 0 4.905 8 2 2 0 4.441 11 1 1 1 4.5737L 20 1 0 2 4.509 13 4 0 0 4.034 8 0 2 1 4.5563K 2 0 0 4.249 12 2 1 1 3.935 10 1 1 2 3.5699L 5 2 0 1 3.946 8 3 0 1 3.653 20 0 2 2 3.5616K 1 1 2 3.607 5 1 2 1 3.546 5 0 1 3 3.1391d 4 2 0 2 3.320 6 3 1 1 3.456 11 2 0 0 2.9830 15 1 0 3 3.272 18 2 3 0 3.249 17 1 3 0 2.9685 49 2 1 0 3.212 9 4 2 0 3.218 9 1 1 3 2.7780L 100 2 1 1 3.074 33 4 0 1 3.136 3 0 2 3 2.7741K 3 0 0 2.833 23 3 2 1 3.012 64 2 2 0 2.5792 4 2 1 2 2.749 88 1 3 1 2.844 100 2 0 2 2.5543L 52 3 0 1 2.737 100 5 2 0 2.762 3 1 3 2 2.5452K 2 0 3 2.722 99 2 3 1 2.721 33 2 2 1 2.4957 31 0 0 4 2.659 13 4 2 1 2.702 9 0 4 1 2.4844 31 1 0 4 2.638 32 6 0 0 2.691 12 1 4 1 2.2935d 3 3 0 2 2.5001d 2 0 4 0 2.661 3 1 0 4 2.2843dL 3 2 2 0 2.4534 45 1 4 0 2.626 68 0 4 2 2.2781dK 2 1 3 2.3805 27 6 1 0 2.609 15 1 1 4 2.2298dL 8 3 1 0 2.3576 2 3 3 1 2.546 15 0 2 4 2.2278 K 1 1 4 2.3379 10 2 4 0 2.527 6 2 2 3 2.0314 7 3 1 1 2.3012 10 0 0 2 2.4876 3 0 4 3 2.0253 10 2 0 4 2.2547 2 1 0 2 2.4594 13 2 4 0 1.9456dQ 3 2 2 2 2.2279 3 5 2 1 2.4142 5 0 1 5 1.9425dR 3 0 3 2.2136 1 6 2 0 2.4020 15 1 5 0 1.9416dS 3 1 2 2.1553 1 2 0 2 2.3766 4 3 1 1 1.9154dL 3 4 0 0 2.1249 1 6 0 1 2.3674 4 2 4 1 1.9090dK 1 0 5 2.0638 3 1 4 1 2.3221 4 2 0 4 1.9037dQ 5 2 1 4 2.0485 9 6 1 1 2.3098 2 1 3 4 1.9000dR 4 0 2 1.9728 1 3 0 2 2.2585 11 0 5 2 1.8964dS 3 1 3 1.9633 2 1 2 2 2.2321 2 1 1 5 1.8471L 7 3 2 0 1.9497 1 4 4 0 2.2222 5 3 0 2 1.8451K 3 0 4 1.9394 2 2 2 2 2.1705 5 2 4 2 1.8105dL 3 3 2 1 1.9173 10 6 2 1 2.1635 3 1 5 2 1.8073dK 2 0 5 1.9016 2 5 3 1 2.1536 6 2 2 4 1.7849dL 3 3 2 2 1.8303 24 7 0 1 2.0939 1 0 4 4 1.7808dK 4 1 1 1.8268 11 3 2 2 2.0785L 16 3 3 0 1.7194d 8 4 0 3 1.8222 3 4 1 2 2.0768K 0 6 0 1.7111d 6 2 2 4 1.8032 5 2 5 0 2.0590 4 3 1 3 1.6799Q 23 2 1 5 1.7734L 11 1 3 2 2.0213 14 2 4 3 1.6756R 0 0 6 1.7732K 2 3 2 1.9755 13 1 5 3 1.6730S 3 1 4 1.7639 2 7 2 1 1.9478 5 2 0 5 1.6486d 3 1 0 6 1.7361 1 5 1 2 1.9386 9 2 1 5 1.6278dL 3 3 2 3 1.7082 13 2 5 1 1.9019 5 0 1 6 1.6277dK 3 0 5 1.7008L 20 6 4 0 1.8927 7 3 3 2 1.6241dL 9 5 0 0 1.6996K 8 2 0 1.8871 5 1 6 1 1.6225dK 5 0 1 1.6781 6 4 5 0 1.8835 6 0 6 2 1.6171dL 5 1 1 6 1.6675 2 8 1 1 1.8427 3 3 2 3 1.6163dK 4 0 4 1.6598 1 0 4 2 1.8180 2 1 1 6 1.5703dQ 17 4 1 3 1.6432 3 7 3 1 1.8023 10 2 2 5 1.5697dR 2 0 6 1.6361L 18 0 6 0 1.7745 17 0 2 6 1.5696dS 3 3 0 1.6357K 4 5 1 1.7608 2 0 4 5 1.5668d 2 5 0 2 1.6189 14 8 3 0 1.7548 9 4 2 1 1.5881 1 6 2 2 1.7281 7 F24=20(0.013,94) 3 1 5 1.5793 2 3 4 2 1.7223 8 3 2 4 1.5722 17 7 0 2 1.6920 2 3 3 2 1.5632 6 9 0 1 1.6879 3 5 5 1 1.6745 15 F3=171(0.006,30) 7 1 2 1.6704 5 1 6 1 1.6626 8 0 0 3 1.6585 6 7 4 1 1.6457 3 2 0 3 1.6246 3 10 0 0 1.6146Q 6 7 2 2 1.6122R 1 5 2 1.6096S 6 5 1 1.5829 2 3 1 3 1.5679 1 F30=90(0.006,52) aYSi3N5 indexed on a hexagonal cell with a=9.8126(3) A° , c=10.6373(5) A° , V=887.0 A° 3.bY2Si3N6 indexed on an orthorhombic cell with a=16.1492(9) A° , b=10.6465(6) A° , c=4.9758(3) A° , V=855.5 A° 3 .cY3Si6N11 indexed on an orthorhombic cell with a=5.965(1) A° , b=10.267(2) A° , c=9.892(2) A° , V=605.8 A° 3. Intensities based on peak heights of the a1 reflections. dThese reflections were not resolved by Guinier–Hagg camera but are calculated from the cell parameters and demonstrated by pattern-fitting techniques. J. Mater.Chem., 1997, 7(3), 505–509 507identification of the XRD pattern of this phase1 with Y6Si3N10 was not accompanied by supporting analytical data, and that the pattern has been identified by subsequent authors5 with Y2Si3N6; the present results suggest it contains even less Y than this. The phase YSi3N5 could not be prepared totally free of the compound we now identify as Y3Si6N11, possibly owing to a slow reaction rate and the loss of SiO, which is particularly significant in high Si compositions.Although YSi3N5 forms more efficiently at higher temperatures, the loss of SiO is also greater under these conditions, leading to the additional appearan of Y3Si6N11 or even small amounts of YN. The EDX analyses of these samples are fairly variable, with a typical mean bulk Y5Si ratio of 0.45, approximating to a 151 mixture of the two phases.Portions of typical X-ray powder diffractograms of the three phases prepared from SiO2 starting materials are shown in Fig. 2. The XRD data obtained from the Guinier camera for the three Y–Si nitrides were used to determine the cell parameters, from which the X-ray powder diffractometer data presented in Table 1 were indexed.The data of Table 1 are presented in the accepted ICDD format, and include the associated Smith– Snyder figure of merit value F30 for the mono- and di-yttrium phases, expressed in the form F30=F(D2h,N), where F is the figure of merit, D2h is the mean deviation between the calculated and observed 2h values and N is the number of calculated lines up to the 30th observed line.The figure of merit value quoted for Y3Si6N11 is F24, calculated from the first 24 observed lines free of overlap. The quoted relative intensity values were Fig. 3 Representative 11.7 T 29Si (A–C) and 89Y (D–F) MAS NMR spectra of Y–Si–N phases. A,D, predominantly YSi3N5 with Y3Si6N11 derived from the diffractometer data for the a1 reflections; the impurity; B,E, predominantly Y2Si3N6 with YSi3N5 impurity; C,F, values for overlapping peaks were derived by pattern-fitting predominantly Y3Si6N11 with a trace of J-phase.procedures. The X-ray data are in general agreement with Thompson1 (re-defining his Y6Si3N10 as Y3Si6N11), but are reported here Table 2 29Si and 89Y NMR assignments for the present Y–Si–N phases for a wider range of d-values, which has permitted the identifi- and related compounds (resonances listed in order of decreasing intensity) cation of the major peak for each phase.The compound Y2Si3N6 has been indexed on the basis of a primitive ortho- phase d(29Si) d(89Y) rhombic unit cell. Typical 29Si and 89Y MAS NMR spectra are shown in Fig. 3. YSi3N5 -44.7, -41.5, -47.9 391.1, 131, 505.8(?) Although the samples of Fig. 3 are predominantly of one phase, Y2Si3N6 -59.7, -57, -64.4 394, 510(?) they are known from XRD measurements to contain various Y3Si6N11 -37.3, -35.5, -48.2 497.3, 97 amounts of a second phase which can, however, be allowed Y4Si2O7N2 -71 to -73a,b 214.5a for, as follows. The sample of Y3Si6N11 (Fig. 3C,F) contains a Y2Si3O4N3 -56.7 to -57.5a,d 160.5a Y5Si3O12N -73.7 to -74.9, -67.4d — small amount of J-phase (Y4Si2O7N2), of which the 29Si YSiO2N -64.7 to -65.3d — spectrum contains a resonance at d -71 to -73,16,17 and the Y10Al2Si3O18N4 -76.1c — 89Y spectrum contains a peak at d 214.5.16 The absence of Y2SiO5 -79.8 to -80.0d 237, 148b these resonances from the present 29Si and 89Y spectra suggests c-Y2Si2O7 -92.8d 198b that the J-phase is not detected in the spectra shown in a-Y2Si2O7 — 114b Fig. 3C,F, which are therefore assumed to arise solely from b-Y2Si2O7 — 208b d-Y2Si2O7 — 122b Y3Si6N11. On this basis, the peaks at d -37.3 and -35 in the Y2O3 — 314, 272.5b 29Si spectrum of YSi3N5 (Fig. 3A) are assigned to the Y3Si6N11 impurity known to be present in this sample; by a process of aRef. 16. bRef. 17. cRef. 18. dRef. 19. elimination, the peaks at d -44.7 and -41.5 are due to YSi3N5.On the basis of relative intensities, the resonance at d -47.9 (Fig. 3A) must contain contributions from both YSi3N5 arise solely from the YSi3N5 impurity, but the possibility that and Y3Si6N11. By similar reasoning, the 89Y resonances at d it contains a contribution from Y2Si3N6 cannot be ruled out. 391 and 131 in this sample are assigned to YSi3N5, although These spectral assignments are summarised in Table 2, which the possibility cannot be ruled out that the resonance intensity includes the NMR data of related compounds for comparison.at d 506 contains a contribution from this phase as well as The NMR results suggest the presence of at least three Si from the impurity phase Y3Si6N11. sites in each of these phases, those in YSi3N5 having shifts in On this basis, the 29Si resonances of YSi3N5 can be eliminated the general region of those of Si3N4.The 89Y spectra suggest from the spectrum of Y2Si3N6 which contains the former as at least two types of Y site in each phase, one of which (at d an impurity phase (Fig. 3B); the group of peaks at d -64.4, 490–510) appears in all the spectra. In addition, YSi3N5 and -59.7 and -57 are therefore associated with Y2Si3N6. The Y2Si3N6 have a similar Y site, resonating at d ca. 390. All three 89Y spectrum of this phase (Fig. 3E) consists predominantly of Y spectra contain intensity at d ca. 100–200, but peaks can be the resonance at d 394 (which will, however, also contain a resolved in this region only in YSi3N5 and Y3Si6N11. The shifts component from the YSi3N5 impurity).The origin of the of the peaks at d ca. 390 and 500 are distinctly different from those reported for oxygenated Y compounds including Y2O3, smaller peak at d 510 in this spectrum is less clear; it could 508 J. Mater. Chem., 1997, 7(3), 505–5094 W. Y. Sun, T. Y. Tien and T. S. Yen, J. Am. Ceram. Soc., 1992, the various yttrium silicates18 and J-phase and N-melilite,16 74, 2547.which appear at d 100–300. The present shifts may reflect the 5 W. Y. Sun, T. Y. Tien and T. S. Yen, J. Am. Ceram. Soc., 1992, predominantly nitrogen environment of the yttrium, with the 74, 2753. minor spectral features at d 100–200 arising from small 6 G. Z. Cao and R. Metselaar, Chem.Mater., 1991, 3, 242. amounts of oxygenated phases. 7 T. Ekstro�m and M. Nygren, J. Am. Ceram. Soc., 1992, 75, 259. 8 Y. B. Cheng and D. P. Thompson, J. Am. Ceram. Soc., 1994, 77, In summary, this work shows CRN to be an excellent 143. method for preparing complex silicon yttrium nitrides at 9 P. L. Wang, H. Y. Tu, W. Y. Sun, D. S. Yan, M. Nygren and considerably lower temperatures (1400–1450 °C) than reported T. Ekstro�m, J.Eur. Ceram. Soc., 1995, 15, 689. previously. The usefulness of this technique for determining 10 Z. J. Shen, T. Ekstro�m and M. Nygren, J. Am. Ceram. Soc., 1996, phase relationships in other Si3N4–LnN systems will be 79, 721. explored in future studies. The present work confirms the 11 J. Grins, Z. J. Shen, M. Nygren and T. Ekstro�m, J. Mater. Chem., 1995, 5, 2001. existence in the Si3N4–YN system of the phases YSi3N5 and 12 I. Higgins and A. Hendrey, Br. Ceram. T rans. J., 1986, 85, 161. Y2Si3N6, but suggests that the phase identified previously as 13 M. E. Bowden, K. J. D. MacKenzie and J. H. Johnston,Mater. Sci. Y6Si3N10 may be Y3Si6N11. The X-ray powder diffraction data Forum, 1988, 34–36, 599. and 29Si and 89Y solid-state MAS NMR spectra of these three 14 I. Higgins and A. Hendry, Proc. Br. Ceram. Soc., 1986, 39, 163. phases are reported. 15 K. J. D. MacKenzie, R. H. Meinhold, G. V. White, C. M. Sheppard and B. L. Sherriff, J.Mater. Sci., 1994, 29, 2611. 16 R. H. Meinhold and K. J. D. MacKenzie, Solid State Nucl. Magn. T. E. is grateful for a senior research fellowship from the New Zealand Institute for Industrial Research and Development. Reson., 1995, 5, 151. 17 R. Dupree and M. E. Smith, Chem. Phys. L ett., 1988, 148, 41. 18 K. J. D. MacKenzie and R. H. Meinhold, J. Mater. Chem., 1994, References 4, 1595. 19 R. Dupree, M. H. Lewis and M. E. Smith, J. Am. Chem. Soc., 1988, 1 D. P. Thompson,Mater. Sci. Res., 1986, 20, 79. 110, 1083. 2 I. K. Naik and T. Y. Tien, J. Am. Ceram. Soc., 1979, 62, 642. 3 Z. K. Huang, P. Greil and G. Petzow, J. Am. Ceram. Soc., 1983, 66, C96. Paper 6/05992I; Received 30th August, 1996 J. Mater. Chem., 1997, 7(3), 505&ndash

ISSN:0959-9428    

DOI:10.1039/a605992i

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

Nature of the stacking faults in orthorhombic LiMnO2 Laurence Croguennec,a Philippe Deniard,a Raymond Breca and Andre� Lecerfb aL aboratoire de Chimie des Solides, I.M.N. UM 110 CNRS, 2 rue de la Houssinie`re, 44072 Nantes Cedex 03, France bL aboratoire de Chimie des Solides, INSA, 20 avevue des Buttes de Couesnes, 35043 Rennes Cedex, France The synthesis of orthorhombic LiMnO2 (O-LiMnO2 ) with very small crystals (diameter#0.3 mm) leads to peculiar X-ray diffraction patterns.Some reflections (with k even) remain thin allowing for cell parameter refinements, showing that, compared to phases with bigger crystals, b and c remain unchanged, whereas an important increase of the a parameter is observed. Other reflections (with k=2n+1 and h0) are widened substantially, while the remnant peaks (k=2n+1 and h=0) undergo a strong asymmetrization.These features have been related successfully to faults corresponding to a b/2 translation of a basic unit constituting O-LiMnO2. A simulation made with the Diffax program allowed good reproduction of the experimental X-ray diffraction data, showing a statistic distribution of the faults, at least for the low fault concentrations corresponding to the samples under study.The insertion of the fault corresponds to the insertion of a monoclinic cell between two blocks of orthorhombic symmetry. This cell (a#5.53 A° , b#2.80 A° , c#5.30 A° ) corresponds to a newly obtained monoclinic LiMnO2 phase obtained through a topotactic deintercalation of a-NaMnO2. The fault percentage of the compounds studied goes from 1 to 6%and is well correlated to the substitution ratio between lithium and manganese when the fault occurrence is treated as a cationic disorder (only in the case of small disorder for which the lines remain treatable with the Rietveld refinement program).The fault percentage can also be determined easily from the cell parameter relation a=x0a0+xmamsinc, where a0 and am are the parameters of the orthorhombic and monoclinic cell of the pure phases and x0 and xm their relative fractions, a being the parameter of the faulted phase as refined from the fault-unaffected thin reflection peaks.Since the higher electrochemical capacities of orthorhombic to 0.99. The ground powders of the starting phases were placed in an aluminium boat in a nitrogen-filled silica tube. Some LiMnO2 (O-LiMnO2) used in lithium batteries have been oxygen was added to obtain the targeted mean manganese related to small crystal size,1 attempts have been made to oxidation states.The samples were heated at 500°C for 5 h, decrease the O-LiMnO2 crystal size as much as possible in ground after cooling and heated again at 700 °C for 10 h, order to obtain cathodic materials with still higher capacities.before final slow cooling (10 h) to room temperature. Under Under certain experimental conditions, the synthesis of Othese conditions, samples with different LiMn2O4 contents LiMnO2 may lead to preparations containing some LiMn2O4 were obtained (Table 1). spinel impurity.2 Based on the usual X-ray diffraction analyses, Mn and Li elemental analyses were carried out by EDTA these spinel-containing O-LiMnO2 compounds show XRD complexation and emission plasma spectrometry, respectively, patterns similar to those obtained for the pure samples, except whereas the mean oxidation states of manganese were deter- that their X-ray diffraction lines appear generally very ill mined by oxidoreduction analysis using Fe2+ as the reducing defined.This was first ascribed to a poor crystallization state agent. It was found that increasing lithium content corre- due to very small crystals with short coherence lengths, which sponded to the decreasing manganese oxidation state, in was related to the synthesisprocedures. These materialsshowed agreement with a decreasing spinel content (Table 1).particularly good electrochemical behaviour, and questions The X-ray diffraction patterns were recorded on a Siemens arose about the respective roles of the crystal size vs. the D5000 diffractometer, without monochromator (Cu-Ka1= spinel content. 1.540598 and Cu-Ka2=1.544390 A° ) in Bragg–Brentano Going beyond the simple observation of the widening and geometry. distortion of the diffraction peaks, this study led us to determine The X-ray patterns showed that the sample mixtures were the true origin of the structural phenomenon responsible for mostly made up of O-LiMnO2 with some LiMn2O4 spinel. A the X-ray pattern characteristics.This article thus describes small amount of tetragonal Li2Mn2O4 and Mn3O4 could also the nature and the extent of the faults found in O-LiMnO2 be detected.A Rietveld quantitative analysis3,4 was carried out samples (with and without spinel impurity) and demonstrates to determine the percentage of each phase present (see later). how these faults can satisfactorily model the XRD pattern The Diffax program5 allows the calculation of the diffracted features of the phases for various fault concentrations. It is intensities of crystals containing imperfections such as twinning also demonstrated that the spinel presence seems only to be and/or two-dimensional stacking faults.This software was an indicator (but an important one) of the particle size. chosen to model the poorly defined X-ray diffraction pattern of O-LiMnO2. Diffax is a modelling, not arefinement, program; it is necessary to make hypotheses and to compare visually Experimental the calculated and observed X-ray powder diffraction diagrams.To calculate the structure factors, it is thus necessary to know A series of six samples (labelled as shown in Table 1) containing mostly the spinel LiMn2O4 as impurity have been synthesized not only the imagined layers sequences, but also the transition probability aij between two layers i and j within the stacks, under conditions very similar to those used to prepare X-ray diffraction pure O-LiMnO2.1 The usual synthesis time was and the interlayer translational vectors Rij.The structural data of O-LiMnO2 (cell parameters, atomic positions and pseudo- shortened in order to try to decrease the crystallinity of the phases. The samples were prepared from mixtures of Mn2O3 Voigt u,v,w and g profile function parameters) were determined from the Rietveld refinement performed with the Fullprof [obtained by pyrolysing MnO2 (Sedema T.R.)] and LiOH H2O (Chemetals) with Li/Mn ratios varying from 0.83 program.J. Mater. Chem., 1997, 7(3), 511–516 511Table 1 Manganese and lithium elemental analyses of several samples (O-LiMnO2–LiMn2O4 mixtures) with nominal and experimental Li/Mn ratios MN438 MN436 MN440 MN435 MN439 MN441 (Li/Mn)nom. 0.83 0.91 0.93 0.94 0.98 0.99 Li (%) 6.01 6.66 6.85 6.97 7.24 7.14 Mn (%) 58.70 58.43 58.24 58.74 58.30 57.55 (Li/Mn)exp. 0.81 0.90 0.93 0.94 0.98 0.98 Mn mean oxidation statea 3.18 3.08 3.07 3.05 3.03 3.01 D200b/A° 185 205 205 210 185 235 D002b/A° 270 235 210 230 235 240 aThe mean manganese oxidation state are results from redox analyses.bCoherence lengths (D) calculated using theWarren–Averbach method (ref. 6). Characterization of O-LiMnO2 samples The sample colour changed from light to dark green with increasing spinel content. In our studies, in contrast to what had been observed before,1 the colour change is not due to the crystallite size but rather to the LiMnO2/LiMn2O4 ratio.In effect, the crystallite and crystal size of the six samples as calculated by the Warren–Averbach method6 on those peaks not affected by the stacking faults in the compounds (see below) and observed by scanning electron microscopy (diameter#0.3 mm) (Fig. 1) were found to be about the same. The crystallite size was ca. 200 A° in the a and c directions (Table 1).The XRD patterns of these newly prepared spinel-containing orthorhombic O-LiMnO2 phases (Fig. 2) showed important changes compared to those of the well crystallized spinel-free samples studied previously.1 In effect, one observes a marked widening of certain peaks, leading to the disappearance of the weaker lines (see, for example, lines around 2h#23° and 41°), Fig. 2 Comparison between the X-ray diffraction pattern of (a) a ‘pe’ and a strong asymmetry of others towards high h values. Using O-LiMnO2 (ref. MN412)1 and (b) the spinel-containing MN438 sample. Widening and asymmetry of certain peaks are clearly observed. the orthorhombic cell parameters of a well crystallized and The extra reflections on the diagram of the latter sample are due to characterized O-LiMnO2 phase, an attempt to index and the occurrence of LiMn2O4, Li2Mn2O4 and Mn3O4.Peaks marked model the profiles of all the diffraction lines failed: only the with asterisks are enlarged strongly. narrow and symmetric peaks could be considered, the widening and asymmetry of the other peaks preventing any fit. Narrow Table 2 Profiles fitting obtained from the MN438 sample X-ray peaks correspond to the planes with k even, and their profile diffraction pattern with the PROLIX program (only the reflection modelling with a pseudo-Voigt-type function with the help of lines of O-LiMnO2 are given)a the PROLIX7 program was carried out successfully (see Table 2 hkl 2h/degrees d/A° I FWHM/degrees g given as an example for the sample MN438).Enough of these peaks were left out to allow the cell parameter refinement of 010 15.37(3) 5.76(1) 100.00 0.2721 0.857 the six samples under study (program U-FIT8) (Table 3). 110 25.0(3) 3.57(5) 38.30 1.5720 0.045 Whereas the b and c parameters are identical to those of pure 011 35.57(7) 2.522(5) 20.23 0.2687 0.000 O-LiMnO2, an important increase of ca. 0.3% of the a 120 36.84(6) 2.438(4) 29.85 0.2860 0.982 parameter is recorded for these spinel impurity containing 101 37.58(8) 2.391(5) 22.60 0.3672 0.000 200 39.13(3) 2.300(2) 69.15 0.4432 0.071 samples.The widened reflections not taken into account in the 111 — — — — — refinement procedure correspond to k=2n+1 and h0 and 210 — — — — — the asymmetric ones to k=2n+1 and h=0 (Fig. 2). In the 021 45.10(2) 2.0089(8) 97.75 0.3026 0.665 211 — — — — — 221 61.26(3) 1.5119(8) 55.54 0.3915 0.319 131 — — — — — 040 64.8(2) 1.439(3) 12.32 0.4644 0.000 002 66.6(2) 1.404(3) 7.57 0.2806 0.360 112 — — — — — 240 78.47(6) 1.2178(8) 9.43 0.3727 0.000 202 80.11(5) 1.1970(6) 8.46 0.2900 0.000 aEntries in italics correspond to those peaks which were too wide and too asymmetric to be considered in the parameter refinements.Non- fitted lines are those which were so broad that they were lost in the background. case of the asymmetric lines, even a Thomson–Cox–Hasting4 type of modelling did not work because of the very strong right-side asymmetry of the diffraction lines. These different features of the ‘non-classical’ reflection peaks (shift, widening, asymmetry) are well documented in the case Fig. 1 Electronic microscope photographs of one spinel containing O- of the c-MnO2 series9,10 and have been explained through LiMnO2 sample (ref. MN441) showing the regular size distribution (diameter#0.3 mm) of the crystals stacking faults resulting from the alternating of pyrolusite and 512 J. Mater. Chem., 1997, 7(3), 511–516Table 3 Least-squares refinement of the O-LiMnO2 cell parameters for the six samples under study; comparison with the parameters refined for a well crystallized pure O-LiMnO2 sample (from ref. 1) MN438 MN436 MN440 MN435 MN439 MN441 Hoppe ref. 1 a/A° 4.593(2) 4.593(1) 4.593(1) 4.5904(7) 4.604(3) 4.5903(9) 4.572 4.5792 b/A° 5.754(2) 5.747(1) 5.750(1) 5.7462(9) 5.759(5) 5.754(2) 5.757 5.7509 c/A° 2.8061(9) 2.8051(5) 2.8045(5) 2.8044(3) 2.8088(9) 2.8067(6) 2.805 2.8060 V /A°3 74.15(8) 74.05(5) 74.07(5) 73.97(3) 74.5(1) 74.13(5) 73.8 73.89 ramsdellite blocks.Accordingly, De Wolff9 described the c- MnO2 phases as a statistic distribution of simple and double chains. Following the same idea, we studied the possible occurrence of stacking faults in the O-LiMnO2 materials. Quantitative analysis of sample mixtures The phases present in the samples under study contained essentially the O-LiMnO2 majority phase and the spinel LiMn2O4 minority one.Some small amounts of tetragonal Li2Mn2O4 and Mn3O4 could also be detected. The percentage of each phase was calculated, whenever possible, by using the Fullprof program. Rietveld analyses were carried out with some specific constraints because of the particularities of the diagram line shapes as explained above.The refinement conditions were as follows: angular regions were excluded from Fig. 3 Projection along the c axis of the structure of O-LiMnO2. The phase can be described as a stacking, along b, of corrugated layers of the refinement for those peaks widened, markedly; the atomic oxygen, lithium, oxygen and manganese.positions, cell parameters and profile function parameters were refined only in the case of O-LiMnO2 and LiMn2O4; owing to the very small amounts of Mn3O4 and Li2Mn2O4, only the the structure as based on the stacking of very similar layers cell parameters were refined in these cases. For these two made of O/Li/O/Mn/O rows perpendicular to the Fig. 3 plane compounds, the profile function parameters were not allowed projection, the layers being stacked along the a axis.In order to vary with h. The atomic parameters were in agreement with to describe the faults in O-LiMnO2, four different basic units those of the literature.11,12 with the parameters a/2, b and c with the atomic arrangements Obviously, and owing to the complexity of the sample shown in Fig. 4 have been chosen. Table 5 shows the atomic compositions, the quality of the refinement proved not perfect, positions for the four units. Units 1, 3 and 2, 4 are structurally but it seemed to us to be of sufficient quality to quantify the identical but are defined by their distinct transition probability ratio of the four phases. A validation of the multiple excluded regions procedure was performed by refining the structure of pure O-LiMnO2 .It was found that the exclusion of domains of the same magnitude as those considered for the mixedphase samples resulted essentially in the lowering of the data accuracy, without sizeable drift of the atomic positions. The quantitative analyses were easy to carry out owing to the similar compositions of the phases and their similar linear absorption coefficients (in which case the Brindley coefficient13–15 can be considered equal to unity).Table 4 gives the quantitative analyses of the six samples. It can be seen that the mean oxidation states of manganese Fig. 4 Perspective views of the layers chosen to describe the faulted calculated from the phase percentages agree well with those O-LiMnO2 determined by the redox analyses.Note also that only one phase out of the six exhibited a measurable amount of Mn3O4 Table 5 Atomic coordinates of the four layers of the stacking faults [5.3(6)%]. layer 1 (�3) layer 2 (�4) Determination of stacking faults in O-LiMnO2 Mn 1/2,ca.0.63,1/4 1/2,ca.0.37,3/4 O-LiMnO2 structure and fault choice Li 1/2,ca.0.10,1/4 1/2,ca.0.90,3/4 O(1) 1/2,ca.0.87,3/4 1/2,ca.0.13,1/4 Fig. 3 shows the O-LiMnO2 structure projected along the b O(2) 1/2,ca.0.39,3/4 1/2,ca.0.61,1/4 axis (with a#4.58 A° , b#5.75 A° , c#2.80 A° ). One can consider Table 4 Ratios of O-LiMnO2 , LiMn2O4, Li2Mn2O4 and Mn3O4 in the different samples under study, with the calculated and measured manganese mean oxidation states MN438 MN436 MN440 MN435 MN439 MN441 LiMnO2 (%) 67(3) 86(3) 90(2) 92(3) 96(3) 99(2) LiMn2O4 (%) 22(1) 10.3(7) 8.1(5) 5.6(5) 2.3(3) 0.9(2) Li2Mn2O4 (%) 5.7(7) 3.4(7) 2.2(5) 2.1(5) 2.0(4) — Mn3O4 (%) 5.3(6) — — — — — Mn calc.mean oxidation state 3.12 3.09 3.07 3.05 3.02 3.01 Mn exptl. mean oxidation state 3.18 3.08 3.05 3.05 3.03 3.01 J. Mater. Chem., 1997, 7(3), 511–516 513aij and their translation vector Rij. The occupation ratios of all positions were equal to unity and the atomic displacement parameters were 0.5 A° 2 for Mn and 1.0 A° 2 for Li and O.Because of the O-LiMnO2 atomic arrangement, we first tested the possibility of a fault resulting from a translation of b/2 of a basic unit. The transition probability aij between layers i and j and the translation vectors Rij are given in Table 6.Results of simulation Fig. 5(a) shows the simulated X-ray diffraction pattern obtained from the proposed fault (translation of a slab of b/2), the fault concentration going from 0 to 1. Fig. 5(b) shows the strong influence of even small amounts of faults. Also, the main characteristics of the ill defined diagrams obtained for the six samples under study are very well reproduced, with in particular a marked asymmetry on the right of certain peaks [e.g.(010)] and very strong widening of others [e.g. (110), (111) and (210)]. Clearly, one of the first main lines of the diagrams, the (110) peak, is highly sensitive to even small stacking faults. This rules out the first explanation given for this peak broadening through the occurrence of some Mn2O3 impurity.1 This model is thus to be considered to account for the structural arrangement of faulted O-LiMnO2 .Indeed, other types of faults were considered but proved to be unable to show the expected influence on the X-ray diffraction pattern. Consequences of the occurrence of faults in OLiMnO2 The insertion of the fault corresponds to the occurrence of a monoclinic cell between two blocks of orthorhombic symmetry.The so-called ‘fault cell’ has, then, an am parameter defined as: am=a0-b0/2 (Fig. 6), while b0 and c0 still correspond to the translation of the well ordered phase network. Indeed, this interpretation holds only for small amount of faults, i.e. in the case of isolated faults, which is the case here [Fig. 5(a)]. This model implies that the diffraction patterns observed yield the Fig. 5 (a) Simulated X-ray diffraction patterns for O-LiMnO2 with mean structure, the a parameter observed corresponding to faults corresponding to a b/2 translation of an isolated slab (lowermost the weighted average of the parameters a0 and am sin c of the pattern, orthorhombic LiMnO2; uppermost pattern, monoclinic orthorhombic and monoclinic cells. Fig. 7 shows the ortho- LiMnO2). The simulation was made for fault contents from 0 to 1.(b) For a 0.1 fault concentration, one can see the marked asymmetry rhombic and monoclinic LiMnO2 reciprocal networks pro- of the (010) and (030) peaks and the important widening of the (110), jected along the c axis, an identical vector for both symmetries. (111) and (210) lines. For even k, the diffraction lines will be neither shifted nor widened since both lattice nodes fall at the same positions [peaks (120), (101) and (200), for instance].Clearly, and as can be shown through simulation with a random distribution of faults, the concentration of which can be simulated up to Table 6 Values attributed to the transition probability aij and translation vector Rij between two layers i and j allowing the description of the chosen fault [the x values (between 0 and 1) allow for the probability variation of the faults] transition aij Rij Fig. 6 Projection along the c axis of the orthorhombic cell of O- 1–1 0.000 — LiMnO2 with the inserted monoclinic fault 1–2 1-x a/2 1–3 0.000 — 100% (i.e. up to pure monoclinic LiMnO2) [Fig. 5(a)], those 1–4 x a/2+b/2 2–1 0.000 — lines not affected by the faults are reflections common to the 2–2 0.000 — two phases.For odd k and h0, the nodes of the monoclinic 2–3 1.000 a/2 cell are in between two nodes of the orthorhombic cell. For 2–4 0.000 — the reflections (h, k=2n+1, l ) there results a widening corre- 3–1 0.000 — sponding to the shorter and longer diffusion vectors on each 3–2 1-x a/2 side of the orthorhombic cell vector.This is what is observed 3–3 0.000 — 3–4 x a/2+b/2 for the lines (110), (111) and (210), for instance. For odd k 4–1 1.000 a/2 and h=0, the nodes of the monoclinic cell are also in between 4–2 0.000 — two nodes of the orthorhombic cell, but the diffusion vector is 4–3 0.000 — longer on each side of the orthorhombic cell vector. This 4–4 0.000 — results in an increase of the diffusion vector which induces a 514 J.Mater. Chem., 1997, 7(3), 511–516Fig. 7 Projection along the c axis of the reciprocal networks of OLiMnO 2 and of the fault cell M-LiMnO2 . The circles correspond to the peaks presenting a strong right-side asymmetry, the squares correspond to the peaks keeping their normal profile (nodes common to both networks), and the triangles represent the peaks enlarged strongly.marked asymmetry towards the higher diffraction angles, i.e. on the right side of the (0, k=2n+1, l) peaks (see Fig. 7). Fig. 8 shows the calculated pattern of a faulted O-LiMnO2 phase (2.5% faults) compared with the experimental one. This Fig. 9 Projection along the c axis of monoclinic LiMnO2 correspond- simulation demonstrates how well reproduced the experimental ing to the fault found in O-LiMnO2 .The parameters of the broken data are (see how the fine but also abnormal lines are well lines cell allow a description in the monoclinic M-NaMnO2-type modelled), allowing for a good evaluation of the faults concen- reference axes. tration in a given sample. This evaluation is carried out by simple visual comparisons between several simulated diagrams Table 7 Comparisons between the monoclinic cells of a-NaMnO2, the and the diagram under study [see also below the (110) LiMnO2 fault in O-LiMnO2, and M-LiMnO2 prepared via a soft linewidth vs.the fault content]. chemistry route Fig. 9 represents the projection along c of the LiMnO2 a-NaMnO2 LiMnO2(fault)a M-LiMnO2b monoclinic fault structure.This new structure represents an anionic ABC close stacking with a succession of lithium and a/A° 5.63(1) ca.5.53 5.439(3) manganese layers, these two cations remaining in octahedral b/A° 2.86(4) ca.2.80 2.809(2) sites. This structural arrangement is that of a-NaMnO216 and c/A° 5.77(1) ca.5.30 5.395(4) this is confirmed by recent works17,18 reporting the successful b/degrees 112.9(5) 116 115.9(4) synthesis of isotypic monoclinic LiMnO2 (M-LiMnO2) by a aOnly approximate values can be given for the fault cell, since it is topotactic substitution of sodium by lithium.In M-LiMnO2, only a model used in Diffax. bRef. 17. Li and Mn are in the (2d) positions of the C2/m space group (0,1/2,1/2 and 0,0,0, respectively, for the atomic coordinates), whereas the oxygen atoms are in (4i) with x#0.25 and y#0.75 orthorhombic and monoclinic cells and x0 and xm their respect- [these values are precisely calculated at 0.270(1) and 0.772(1) ive fractions.The knowledge of a and of a0 (taken as 4.572 A° from the structure calculation of ref. 17]. Table 7 compares the from a single-crystal study)19 allows the determination of all estimated M-LiMnO2 parameters of this study with those of the fault concentrations of the samples under study along with the Rietveld refinement.the spinel-free samples studied in a previous work1 (Table 8). It appears that the phases containing LiMn2O4 in detectable amounts (0.9–22%) exhibit a large number of faults from 3.4 Scaling of stacking faults in O-LiMnO2 to 6.1%, whereas the spinel-free materials1 (at least at the The fault percentage can easily be determined from the relation X-ray diffraction detection threshold level, but simulations a=x0a0+xmamsinc, where a0 and am are the parameters of the show that this threshold can be as low as 0.4%) contain percentages lower than 2.9%, with a sample at 0.9%.If the presence of LiMn2O4 is indicative of a more disordered O-LiMnO2, (probably due to a particular synthesis process), it is also of interest to observe that the spinel concentration is not well correlated to that of the faults, although it can be seen that the less faulted compounds contain no spinel.Table 8 Estimated fault concentrations of the monoclinic LiMnO2 type in orthorhombic LiMnO2 for the six samples under study spinel-free samplesa this study sample faults(%) sample LiMn2O4(%) faults(%) MN389 1.7 MN441 0.9 3.4 MN384 2.5 MN439 2.3 6.0 MN386 2.9 MN435 5.6 5.4 MN399 1.1 MN440 8 4.8 MN401 1.4 MN436 10 6.2 Fig. 8 An example of the very good agreement between the simulated MN412 0.9 MN438 22 6.1 and experimental X-ray diffraction patterns for 2.5% faults in OLiMnO 2 (sample MN384 from ref. 1) aRef. 1. J. Mater.Chem., 1997, 7(3), 511–516 515The recently synthesized M-LiMnO2 phase17 also shows a Li/Mn substitution ratio of ca. 3%: it is thus possible that this disorder originates in the occurrence of orthorhombic faults. This situation would then be similar to that of O-LiMnO2, and one may wonder whether a continuous increase of faults would allow an intergrowth phase for a high fault ratio (25, 50, 75%, for instance).Beyond these structural considerations, the influence of these faults on the electrochemical behaviour of O-LiMnO2 as a cathodic material in lithium batteries is under study and will be the subject of a forthcoming article. Conclusions Two previous groups2,20 had proposed the hypothesis of stacking faults occurring in O-LiMnO2 . We have obtained evidence for their existence and have identified their nature.In agreement with their random distribution, these faults model Fig. 10 Evolution of the (110) peak full width at half maximum as a function of the fault percentage. All the samples of this study have very well the experimental X-ray diffraction patterns and been considered along with spinel-free samples from a previous study.1 explain the widening and asymmetrization of certain reflection lines.This study emphasizes the rather systematic behaviour of the manganese oxides to exhibit structural disorders in their structures. These disorders are very sensitive to the synthetic procedures which must be tightly controlled to obtain the targeted materials. In particular, smaller crystals correspond to higher fault concentrations, with the occurrence of spinel impurity.Since the stacking faults and the crystal size are expected to influence greatly the electrochemical behaviour of the manganese oxides, it is now important to relate the sample structural features to the electrochemical characteristics of faulted O-LiMnO2. The authors thank F. Boucher for helpful discussion and J.Gareh for his help in handling the Diffax program. References 1 L. Croguennec, P. Deniard, R. Brec and A. Lecerf, J.Mater. Chem., Fig. 11 Variation of the Li/Mn substitution ratio vs. the fault 1995, 5, 1919. percentage in pure O-LiMnO2 obtained previously.1 Within the 2 J. N. Reimers, E. W. Fuller, E. Rosen and J. R. Dahn, estimated standard deviation, the zero fault compound corresponds J.Electrochem. Soc., 1993, 140, 3396. to zero substitution ratio. 3 H. M. Rietveld, J. Appl. Crystallogr., 1969, 2, 65. 4 J. Rodriguez-Carjaval, Fullprof Manual, ILL Report, 1992. Since some X-ray reflections of O-LiMnO2 are very sensitive 5 M. M. Treacy, M. W. Deem and J. M. Newsam, Diffax, V1.76, to monoclinic fault occurrence, it was of interest to see whether 1990. 6 B. E. Warren, in Progress in Metal Physics, Pergamon Press, one could quantify easily the fault concentration without London, 1959, vol. 8, p. 521. modelling the diagrams. Fig. 10 shows the variation of the 7 M. Evain, J. M. Barbet, P. Deniard and R. Brec, presented at the more disorder-sensitive (110) linewidth vs. the fault content. A Powder Diffraction Meeting, Toulouse, France, 1990. clear correlation exists and one can see that the (110) linewidth 8 M.Evain, U-FIT Manual, I.M.N. Internal Report, Nantes, at half maximum is a good measure of the fault occurrence. France, 1992. The presence of the stacking faults found in O-LiMnO2 is 9 P.M. DeWolff, Acta Crystallogr., 1959, 12, 341. 10 M. Ripert, Thesis, I.N.P.G., Grenoble, 1990. in perfect agreement with the previous cationic disordered 11 D.Jarosch,Mineral. Petrol., 1987, 37, 15. model chosen to better refine the diffraction data:1 the slab 12 W. I. F. David, M. M. Thackeray, L. A. De Picciotto and shift is approximately equivalent to a cationic substitution (see J. B. Goodenough, J. Solid State Chem., 1987, 67, 316. Fig. 3). To prove the point further, Fig. 11 shows the good 13 R. J. Hill and C. J. Howard, J. Appl. Crystallogr., 1987, 20, 467. correlation between the Li/Mn ‘substitution’ ratio and the 14 D. L. Bish and S. A. Howard, J. Appl. Crystallogr., 1988, 21, 86. fault concentration. It should be recalled (see ref. 1) that in a 15 G. W. Brindley, Philos.Mag., 1945, 36, 347. 16 J. P. Parant, R. Olazwaga, M. Devalette, C. Fouassier and previous study and for slightly faulted materials, the diagrams P. Hagenmuller, J. Solid State Chem., 1971, 3, 1. had been refined by the Rietveld program, the faults showing 17 C. Delmas, personal communication. up as Li/Mn disorder. For a hypothetical zero-fault material, 18 A. R. Amstrong and P. G. Bruce, Nature (L ondon), 1996, 381, 499. the substitution ratio is, within error, equal to zero, as expected. 19 R. Hoppe, G. Brachtel and M. Jansen, Z. Anorg. Allg. Chem., 1975, The Li/Mn substitution had been considered as a completely 417, 1. random distribution. The present structural study demon- 20 T. Ohzuku, A. Ueda and T. Hirai, Chem. Express, 1992, 7, 193. strates that the disorder is in fact based on a well identified Paper 6/04947H; Received 15th July, 1996 shear mechanism, involving entire slabs shifts. 516 J. Mater. Chem., 1997, 7(3), 511–516

ISSN:0959-9428    

DOI:10.1039/a604947h

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

Electron paramagnetic resonance in pyrolusite and cryptomelane manganese dioxides M. V. Ananth* and K. Dakshinamurthi Central Electrochemical Research Institute—Madras Unit, CSIR Complex, Madras-600113, India Electron paramagnetic resonance spectroscopy has been applied to identify the singular structural features of pyrolusite (b) and cryptomelane (a) manganese dioxides. At 77 K, the derivative linewidth of pyrolusite is reduced to that of cryptomelane and the MS=-3/2 spin states in cryptomelane are affected by the changes in crystal field. Isothermal heating of cryptomelane in air at 100 °C for two days affects the MS=-5/2 andMS=-3/2 levels.The results seem to confirm the partial collapse of the lattice due to release of molecular water on heating. The use of manganese dioxide as a cathode material for Results and Discussions different types of electrochemical power sources is well Electron microscopy known.1–4 Amongst the different polymorphs of this unique material, the a and b forms (Fig. 1 and 2) are notable in the As the morphologies of a- and b-MnO2 are different, SEM sense that these two fall on the extreme edges of a wide was used to confirm the polymorphs.Pure and fully crystalline spectrum of different non-stoichiometric structures. Also, the b-MnO2 occurs in globular particles of 50–100 mm diameter a form hollandite-type MnO2 has recently been found to be (Fig. 3). Platelets of width 3 mm and length 20 mm are found useful in lithium rechargeable batteries5 whereas b polymorphs in these globules (Fig. 4). The crystallites are smaller in readily find applications in non-aqueous batteries.6 So far KMn8O16 (Fig. 5), and cryptomelane is not as crystalline as characterisation of these groups has been carried out using pyrolusite (Fig. 6). conventional techniques, like X-ray diffraction, magnetic susceptibility, 7 electrical conductivity and thermoanalytical EPR experiments at 300 K measurements,8 which cannot be used for elucidating the subtle The EPR (first derivative) characteristics are shown in Fig. 7. structural features of the dioxides. But electron paramagnetic In pyrolusite [Fig. 7(a)] only a very broad singlet resonance (EPR) spectroscopy enables precise investigation of (DHp–p>2000 G) is obtained. Since this phase is highly stoi- the chemical environments around the manganese spins and chiometric with the presence of Mn4+, the occurrence of only lends itself as an elegant technique to the evaluation of one EPR signal is justified.11 The large linewidth is indicative manganese dioxides.9 To our knowledge no exhaustive report of the powdered polycrystalline nature of the sample, as b- is available on electron paramagnetic resonance in different MnO2 is known to possess a preferred crystal orientation and MnO2 samples.Hence, a variety of manganese dioxides are is found to occur in globular particles8 as seen in the electron being studied by EPR spectroscopy in our laboratory to obtain micrographs (Fig. 3 and 4). This is also in concurrence with valuable information about the structural parameters useful the higher electrical conductivity of the sample.12 The good for battery assembly.9 electrochemical performance in aprotic electrolytes6 is prob- In this paper pyrolusite (b-MnO2 with near rutile structure) and cryptomelane (K+-containing a-MnO2) have been characterised with EPR measurements both at room- and liquidnitrogen temperatures.A comparative study of the various EPR parameters for these samples is made using these experimental results.Experimental Fine powders of pyrolusite and cryptomelane chosen for the study were supplied by Union Carbide India Ltd., Calcutta, India. The samples were verified for their polymorph group by X-ray diffraction and scanning electron microscopy (SEM) Fig. 1 Structure of a-MnO2 examinations. For electron microscopy, the samples were moistened with doubly distilled water and compacted into pellets of diameter 8 mm.Samples were glued on a specimen holder and sputtered with gold. A JEOL JSM-CF 35 scanning electron microscope operating at 15 kV was used. EPR spectra of the samples were recorded with a Varian E- 112 spectrometer at 77 and 300 K. DPPH was used as the ‘g’ marker since it has a sharp, single EPR spectrum with an accurately known g factor equal to 2.0037±0.002.10 EPR measurements were also made on the international common sample IBA-18 (International Battery Associations common Fig. 2 Structure of b-MnO2 sample 18) as a reference. J. Mater. Chem., 1997, 7(3), 517–520 517Fig. 6 Magnified view of particles in a-MnO2 Fig. 3 Crystalline nature of b-MnO2 Fig. 4 Globular agglomerates of plates in b-MnO2 Fig. 7 EPR spectra of pyrolusite (a), cryptomelane (b) and heat-treated cryptomelane (c) at 300K (*, DPPH) However, the observed admixture of the absorption spectra due to MS=-1/2 and MS=+1/2 is indicative of spin–orbit coupling. This is in line with expectation since a-MnO2 is nonstoichiometric and contains interactions with H+ species (i.e. a-MnO2 is actually MnO2-d, the d part is composed of H+ ions in constituent water which interact with the manganese oxide lattice) and contains K+ ions in its 2×2 tunnels.13 Interestingly the first three lines occur at the same fields as in Fig. 5 Small crystallites in a-MnO2 c- and e-MnO2 samples (IBA 18) [Fig. 8(a)], which have very high electrochemical activity in alkaline media.9 The remaining EPR signals occur at relatively high field as compared to IBA ably supported by the large DHp–p values.This may be attributed to several factors, e.g. the degree of stoichiometry, 18. This observed shift in g values for Mn positive spins may again be attributed to excess spin–orbit coupling in the a form the constituent water content and the particle size. By contrast, five EPR lines are obtained for cryptomelane as compared to the c and e samples.We find that the resonance occurs only over a narrow range as compared to the broad KMn8O16 [Fig. 7(b)]. This is reminiscent of the six well resolved hyperfine structure components characteristic of EPR spectra that usually occur in the c and e varieties. This could be one reason for the observed higher aqueous electro- Mn2+ which usually occur in its hydrated salts.These lines arise from the interaction of the 55Mn nuclear spin with the chemical activity of the e and c polymorphs over the pure a samples. The spectrum is extended in lower fields in IBA-18 electronic spins of Mn2+: -5/2, -3/2, -1/2, +1/2, +3/2, +5/2 (nuclear magnetic spin quantum number, MI). Only one as compared to pure a-MnO2. Also there is a possibility that Mn2+ ions could be located on K+ sites with a K+ vacancy set of hyperfine components appears in the observed spectrum. 518 J.Mater. Chem., 1997, 7(3), 517–520Fig. 8 EPR spectra of IBA-18 MnO2 at (a) 300 K and (b) 77K (*, DPPH) nearby, and the hopping process could be connected with equivalent Mn2+ off-centre positions.14 Additional complexities may arise due to the fact that in C3 symmetry, the total of 252 degenerate levels of the d5 configuration for the Mn2+ ion can be condensed into 126 doubly degenerate levels i.e.Kramer’s doublets.15 Fig. 9 EPR spectra of pyrolusite (a), cryptomelane (b) and heat-treated In general, the spectra are mostly not well resolved, and cryptomelane (c) at 77 K (*, DPPH) although it appears that Mn2+ is present in some of them, this may only be a minority species.Thus some of the conclusions drawn may be rather speculative. up to 150–250°C and (iii) condensing of OH groups or EPR experiments at 77 K constituent water released in the range 105–500 °C. b-MnO2 has less than 0.1% adsorbed water8 and is not Interesting changes are revealed by low-temperature measure- expected to be very sensitive to heat treatments in the tempera- ments in b-MnO2 [Fig. 9(a)].The linewidth is reduced to that ture range 25–105 °C. It will be interesting to study the of a-MnO2. Thus, as the temperature increases, broadening of structural changes occurring in a-MnO2 on isothermal pro- the paramagnetic resonance lines occurs due to spin–lattice longed heating in air at around 100 °C as oxidation of Mn3+ interactions.ions to Mn4+ and changes in lattice structure due to the As MnO2 is an n-type semiconductor, as revealed by our release of constituent water may occur, as has been observed thermopower measurements,16 it is worth studying EPR at by Ruetschi et al.19 low temperature as in the case of silicon.17 The EPR lines from The effect of heat treatment at 100 °C for 2 days on the a-MnO2 at 77 K [Fig. 9(b)] are almost the same as those at structure was investigated for cryptomelane to study the 300 K except that, as expected, they occur at lower fields. The changes occurring in the EPR lines. The 300 K spectra derivative linewidth is also practically unchanged. This is [Fig. 7(c)] indicate that although the a structure is maintained unlike the c and e modifications which exhibit singular random at this temperature, a tendency for EPR line broadening is variations between the 300 and 77 K spectra [Fig. 8(b)]. This observed.4 The observed inhomogeneous broadening of Mn2+ is as expected as Mn2+ is a paramagnetic ion in an S state. may be due to random internal stresses depending on the Due to the fact that the functional orbital level is in an S state, thermal history of the sample or due to distribution of point the Mn2+ ion is only very slightly sensitive to changes in defects.In the new spectrum, the MS=+3/2 signal is well crystal field and broadening is not expected. Thus the spin– developed in contrast to the broadened nature before the heat lattice interactions through electric field modulations (Kronig treatment.Also, the upper symmetry of central peak is more Van Vleck’s mechanism) should not be effective in such ions. resolved in the new spectrum whereas it was as high as the However, S-state ions can act only through admixtures of MS=-3/2 peak before the heat treatment. Thus it is seen upper orbital states with non-zero orbital momenta. This that, owing to isothermal heating in air at 100°C, the manga- explains the interactions made by the -3/2 levels at 77 K as nese negative spin states are affected because of changes in revealed from the state of ions which appear to be located at local environment and oxidation state.non-cubic sites. The new 77 Kspectrum [Fig. 9(c)] reveal interesting features. The broadened nature of the spectrum (characteristic of b- Influence of heat treatment MnO2) is clearly visible.But in the structure, the a lines are maintained. However, there is an admixture of MS=-5/2 and Heat treatment is carried out on some manganese dioxides to improve their electrochemical activity.18 As such, studies in MS=-3/2 levels. Thus it is inferred that the constituent water, which is essential for the stability of the lattice according to this direction are important.Three kinds of water have been observed in manganese dioxides:8 (i) adsorbed (loosely bound) the earlier reports,20 has been partially released, leading to partial collapse of the lattice as revealed by the XRD results molecular water which is desorbed in the range 25–105 °C; (ii) interlayer molecular water bound more tightly and released (Fig. 10 vs. 11), which indicate partial amorphisation. Such J. Mater. Chem., 1997, 7(3), 517–520 519MnO2 and the MS=-3/2 levels in the a form seem to be affected by the changes in crystal field. Isothermal heating of KMn8O16 in air at 100 °C for 2 days affects the manganese negative spin states owing to changes in local environment and oxidation state.The EPR curves have shown evidence that the manganese spin states in cryptomelane are greatly affected by the release of molecular water on heating. We thank RSIC, IIT, Madras, for EPR measurements and Union Carbide India Ltd., Calcutta, for donating the samples. We thank Shri Y. Mahadeva Iyer, CECRI, Karaikudi, for technical arrangements of the sample analysis by SEM and XRD.References 1 Manganese dioxide batteries, ed. K. V. Kordesch, Marcel Dekker, New York, 1974, vol. 1. Fig. 10 XRD patterns of cryptomelane (a) and pyrolusite (b) 2 K. V. Kordesch and M. Weissenbacher, J. Power Sources, 1994, 51, 61. 3 P. R. Roberge, M. Farahani, K. Tomanstcher and E. Oran, J. Power Sources, 1994, 47, L13. 4 H. Huang and P. G. Bruce, J. Electrochem. Soc., 1994, 141, 76. 5 Q. Feng, H. Kanoh, K. Ooi, M. Tani and Y. Nakacho, J. Electrochem. Soc., 1994, 141, L135. 6 O. N. Khodarev, Elektrochimiya, 1991, 27, 1046. 7 M. V. Ananth, V. Venkatesan and K. Dakshinamurthi, Mater. Sci. Eng. B, in the press. 8 R. Giovanoli, T hermochim. Acta, 1994, 234, 303. 9 M. V. Ananth and K. Dakshinamurthi, J. Power Sources, 1992, 40, 355. 10 ESR Elementary theory and practical applications, ed.J. E. Wertz and J. R. Bolton, Chapman and Hall, New York, 1986. Fig. 11 XRD pattern of cryptomelane after isothermal heating at 11 M. Voinov, Electrochim. Acta, 1982, 27, 833. 100 °C for 2 days 12 V. I. Gryaznoc, Dokl. Akad. Nauk SSSR, 1958, 121, 159. 13 T. Ohzuku, M. Kitagawa, K. Sawai and T. Hirai, J. Electrochem. Soc., 1991, 138, 360. materials may be of interest to electrochemists and battery 14 A.P. Pechenyi, M. D. Glinchuk, C. B. Azzoni, F. Scardina and scientists. A. Paleari, Phys. Rev. B., 1995, 51, 12165. 15 T. H. Yeam, S. H. Choh, Y. M. Chang and C. Rudowics, Phys. Status Solidi B, 1994, 185, 417. Conclusions 16 M. V. Ananth and K. Dakshinamurthi, unpublished results. We have detected the manganese spin states in pyrolusite and 17 G. F. Lancaster, ESR in semiconductors, Hilger &Watts, London, 1966. cryptomelane manganese dioxides at both room- and liquid- 18 H. Tamura, in Electrochemistry of manganese dioxide and manga- nitrogen temperature. From the observation of the roomnese dioxide batteries in Japan, ed. S. Yoshikawa, U.S. Branch temperature EPR signals, it can be presumed that Mn4+ is office, Electrochem. Soc. Japan, Cleveland, OH, 1973, p. 189. present in b-MnO2 and Mn2+ in a-MnO2. In a-MnO2 the 19 P. Ruetschi and R. Giovanoli, J. Electrochem. Soc., 1988, 135, 2663. observed admixture of spectra due to MS=-1/2 and MS= 20 G. Buttler and H. R. Thirsk, Acta Crystallogr., 1952, 5, 288. +1/2 is indicative of spin–orbit coupling. Interestingly, at low temperatures, narrowing of the EPR lines is observed for b- Paper 6/07792G; Received 18th November, 1996 520 J. Mater. Chem., 1997, 7(3), 517–520

ISSN:0959-9428    

DOI:10.1039/a607792g

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

A method for the preparation of high purity lead titanate zirconate solid solutions by carbonate–gel composite powder precipitation T. R. Narayanan Kutty and P. Padmini Materials Research Centre, Indian Institute of Science, Bangalore 560 012, India A novel wet chemical route for the preparation of perovskite titanates (ABO3 type, where A is a divalent cation and B is a tetravalent cation) such as PbTiO3 and its solid solutions, is described.The method involves the coprecipitation of the divalent cations (A) as fine particles of carbonates with hydrated gels of titania or zirconia (BO2 xH2O, 12<x<130; B=Ti4+, Zr4+) by the addition of ammonium carbonate. Such coprecipitation is possible because of the instability of the carbonates and oxycarbonates of Ti and Zr in aqueous media in comparison to the polymerised hydroxides, whereas lead carbonate is precipitated readily.The method gives rise to composite powders in which the submicron crystalline particles of the carbonates, crystalline to X-ray diffraction, are embedded within amorphous gels of BO2 xH2O. The composite nature of the precipitate is confirmed by transmission electron microscopy.The precipitate is dried and calcined at elevated temperatures. Upon heating to 350–400 °C, the reaction between the carbonate and the amorphous dry gel proceeds via the formation of the intermediate PbO zTiO2 (z<0.09) solid solution which then converts to a defect pyrochlore phase (A2B2O7-d, d=1). Above 450 °C, the latter converts to an isocompositional perovskite phase.The process is superior to ceramic methods because of the high purity, uniform chemical homogeneity and lower particle size of the final product. The simple ABO3 perovskite compounds such as BaTiO3, electronic ceramics require raw material powders with much higher purity, better homogeneity and high reactivity.3 It is SrTiO3, PbTiO3, PbZrO3 and their solid solutions are archetypal ceramics with a long history of technological applications well known that the characteristics of the fine powders, such as particle size and shape, point defect contents, microstrain in the electronics industry.1 Increasing demands on the quality of electronic ceramics have led to greater sophistication in the and dislocation distribution, influence the final electrical properties of the ceramics.In this respect, the present route, processing of these materials at both the powder synthesis stage and subsequent densification to solid components or referred to as carbonate–gel composite powder precipitation, yields high-purity ultrafine powders with predictable dopant thin dielectric layers. The growing awareness of the need for scientific elegance in the processing of ceramic powders before levels and a high degree of compositional homogeneity.The peculiarity is that the product on annealing crystallises via the firing is clear in much of current emphasis on novel lowtemperature synthesis techniques. formation of an intermediate defect pyrochlore phase which has the same composition as that of the perovskite. Lead titanate has been a very attractive material for many years from both scientific and practical aspects.This material has the highest Curie temperature (490 °C) and the largest anisotropy (c/a=1.064 in the tetragonal phase) in the simple Experimental perovskite family. PZT has been attracting the attention of the Gels of the hydrated metal hydroxides Bn+(OH)n or microelectronics community in recent years owing to its poten- Bn+On/2 xH2O (B=Ti4+, Zr4+) were coprecipitated with the tial applications in non-volatile, radiation-hard, random access carbonates of Pb2+ by the addition of ammonium carbonate memories.It offers high remanent polarisation and sufficiently at 30–40°C to the corresponding mixed nitrate solutions until low coercive field strength. It also finds importance in room pH 6–8.The precipitate was then washed free of the anions temperature operated IR sensors from the viewpoint of their and the ammonium ions. The precipitate was finally dried at applications such as remote sensing, biomedical thermography, 120°C for 12 h and then calcined at elevated temperatures. gas detection and alarms. Calcination at the appropriate temperature is necessary to Various wet chemical routes have been described for the (i) obtain a stoichiometric crystalline phase, (ii) remove any preparation of perovskites.2 The different processes are listed residual solvent and (iii) obtain powders with the desired in Table 1.The selection of the processing method depends to agglomerate size, surface area and crystal structure according a large extent on the final applications.Understandably, to the end application. This heat treatment has an important effect on the phase content, microstructure and electrical Table 1 Different chemical processes for the preparation of perovskite titanates properties of the ceramics prepared from these powders. Phase identification of the powders was carried out by X- sol–gel processinga – mixed alkoxide route ray powder diffraction using a Scintag/USA diffractometer – carboxy–alkoxide route using Cu–Ka radiation.For precise lattice parameter measure- – hydroxide–alkoxide route ments, higher angle reflections were chosen. The reflection hydrothermal synthesisb peaks were recorded at a scanning speed of 0.125° min-1. synthesis from complex precursorsc – oxalate route – acetate route Electronic-grade silicon was used as an internal standard. – citrate route Particle size and shape were evaluated by the intercept method evaporative decompositiond – liquid mix process on the micrographs from the transmission electron microscope – spray pyrolysis (TEM; JEOL 200 CX, 200 kV) having 2 A° resolution.The ceramic routese – mixed oxides/solid–solid chemical composition of the products and the contaminants, reaction if any, were determined by wet chemical analysis using atomic absorption spectroscopy (AAS; Perkin-Elmer 2980).Thermal aRef. 3, 4. bRef. 5–7. cRef. 8, 9. dRef. 10. eRef. 11. J. Mater. Chem., 1997, 7(3), 521–526 521analyses were performed on a simultaneous thermogravimetry– resorbed by the oven-dried samples when handled in the air, (ii) around 260 °C (mass change ca. 4%) corresponding to the differential thermal analysis (TG–DTA) instrument from Polymer Laboratory, STA 1500, at a heating rate of decomposition of PbCO3 and (iii) around 300–320 °C (mass change ca. 2.5%) due to the reaction of amorphous 5–8°C min-1. hydroxylated titania with PbO accompanied by the release of water due to dehydroxylation. There is only minor mass loss Results and Discussion above 350 °C (mass change ca. 0.5%). The corresponding endothermic maxima are observed around 50, 255 and 300 °C The method reported gives rise to composite powders wherein [Fig. 2(c)] in the TG–DTA curves. The change in the course submicron crystalline particles of lead carbonate are embedded of the PbCO3 decomposition is as a result of the fine-grained within the amorphous gels of titania or zirconia.The carbon- nature of the carbonate particles embedded in the amorphous ates of titanium or zirconium are not formed as they are gel. The gel-derived material did not exhibit the well known unstable. Moreover, titanium and zirconium tend to form endotherm due to the ferroelectric transition around 490 °C in polymeric chains such as (Ti–O–Ti)n and (Zr–O–Zr)n in prefer- the DTA curve.This result seems to indicate that although ence to isolated Ti4+ or Zr4+ ions. It is well known that crystalline PbTiO3 is observed by X-ray diffraction at 450 °C, independent, hydrated Ti4+ in the form of Ti(OH)4 does not the particle size is low so that the phase transformation is exist because of the high charge to ionic radius ratio (5.9 for broadened off, whereas the PbTiO3 powder after annealing at the six-coordinate ion).In addition, Ti4+, which prefers octa- 950°C for 48 h showed a reversible DTA peak around 490°C, hedral coordination, cannot be stabilised by carbonate ligands indicative of the coarsening of the particles and also the for steric reasons.Therefore, neither titanium carbonate nor elimination of the non-thermodynamic defects present during zirconium carbonate is formed from aqueous solutions. their formation. The above interpretation of the TG–DTA Hydrolysed products are produced readily from TiIV solutions results is supported by the X-ray powder diffraction data even at intermediate pH ranges. The often proposed (TiO)2+ given below.and (ZrO)2+ ions are in fact made up of TiMOMTi and ZrMOMZr bridged polymers. Thus the precipitate obtained X-Ray diffraction. The sequence of reactions was studied by by increasing the pH of the TiIV solutions in the presence of identifying the intermediate phases through X-ray diffraction, ammonium carbonate gives rise to only hydrated titania in as shown in Fig. 3, after annealing the composite powder for the form of a gel. At best, the hydrated oxides thus formed various times at selected temperatures. The studies revealed may have partial substitution of the terminal hydroxy groups the presence of intermediate phases appearing during the by carbonate groups. conversion of the carbonate–gel composite powder to tetra- The metal hydroxide gels are in general polymeric chains gonal PbTiO3 .forming an entangled network in which the solvent is The X-ray diffractogram of the as-prepared powder dried at entrapped. The effective molar volume of this network is very 120°C for 12 h [Fig. 3(a)] showed the presence of lead carbon- large in comparison to that of lead carbonate, formed simul- ate (PbCO3) as the only crystalline phase.No titanium dioxide taneously as crystalline fine particles. As a result of the large (TiO2) is detected since it is in the amorphous form, as difference in the molar volume, lead carbonate becomes embed- hydrated titania. Heat treatment of the powder at ca. 275 °C ded within the polymeric network of the gel. This accounts for for 4 h results in the formation of PbO with tetragonal the generation of the composite precipitate.symmetry, whereas no reflections of TiO2 (anatase or rutile) are observed [Fig. 3(b)]. Lead titanate (PbTiO3) As the calcination temperature is increased to 350 °C for 4 h TEM studies. The TEM studies clearly illustrate the com- [Fig. 3(c)], reflections of PbO (tetragonal) shifted to larger 2h posite nature of the powder. Fig. 1(a)–(c) shows micrographs values, indicative of the decrease in d-spacings and cell param- of the as-prepared powder of PbTiO3, at different magnifi- eters. This arises from the partial dissolution of TiO2 in PbO cations, showing the titania gel particle with PbCO3 embedded (tetragonal) to form a solid solution, (PbO zTiO2)ss, where in it. The corresponding electron diffraction pattern [Fig. 1(d)], z<0.12. This solid solution is known from the PbO–TiO2 which shows the spotty pattern superimposed with the haloes, phase diagram.13 The solid solubility arises from the partial further confirms the presence of two phases in the system, i.e replacement of Pb2+ by Ti4+ ions. The presence of the TiO2 as the amorphous phase and PbCO3 as the crystalline PbO zTiO2 solid solution [designated herein as (PbO)ss] phase.The broad haloes seen in Fig. 1(e), for the PbCO3-free could also be detected by chemical methods. PbO, present in gel of hydrated titania precipitated from the ammonium car- the products of incomplete reaction, on dissolution in acetic bonate medium, further supports the presence of an amorphous acid showed the presence of titanium ions.The other phases phase in the system. Furthermore, the powder X-ray diffraction are not soluble in weak acids such as acetic acid. Quantitative studies show no reflections of TiO2 (anatase or rutile), indicat- analysis showed that the maximum solubility of TiO2 is ca. 9 ing that the hydrated titania is X-ray amorphous. mol%. The experimentally determined ao and co values are plotted on the curve given by Matsuo et al.14 and the estimated solubility of TiO2 in PbO is ca. 8.2 mol%, which is in Thermoanalytical studies. The TG–DTA curves of the hydrated titania [Fig. 2(a)] show an endotherm below 100 °C agreement with the chemical analysis results. The residue after acetic acid washing of the product contained a significant (mass change 16%) and another around 300–320 °C (mass change 17%).The TG curves indicate that essentially all the amount of Pb, as shown by chemical analysis. Since uncombined PbO is soluble in acetic acid, this should indicate a mass loss in the sample occurred below 400 °C. This mass loss is due to the removal of water from the xerogel. The recrystallis- chemically combined form of PbO, such as Pb2Ti2O6, PbTiO3, Pb2Ti2O7, etc.Takai et al.15 assigned the reflections of (PbO)ss ation step has been reported to be above 500°C for titania gel.12 In the TG–DTA curves of PbCO3 [Fig. 2(b)] two endo- to arise from another form of lead titanate (designated as PTI) which they considered as the first intermediate in the therms are observed, at 275 °C (mass change 10%) and 330 °C (mass change 6%), corresponding to the decomposition of hydrothermal precipitation.Simultaneous to the shifts in the X-ray reflections of PbO is the appearance of additional PbCO3 to PbO in two steps, due to non-equivalent carbonate groups in the crystalline structure. In contrast, the TG–DTA reflections which could be matched with those of Pb2Ti2O6.16 When the calcination temperature is increased to 375 °C, results of the as-prepared carbonate–gel composite powder [(Fig. 2(c)] show the mass loss to occur in three steps: (i) below the intensities of the reflections from Pb2Ti2O6 increase [Fig. 3(d)]. The formation of the defect pyrochlore Pb2Ti2O7-d 100 °C (mass change ca. 6%), the mass loss arises from water 522 J. Mater. Chem., 1997, 7(3), 521–526Fig. 1 (a)–(c) TEM images of a composite precipitate of PbCO3 and hydrated titania under varying magnification; (d) the corresponding electron diffraction pattern and (e) the diffraction pattern of the amorphous titania hydrate gel (d=1) from the borosilicate glass matrix containing lead oxide (0<z<0.12) solid solution and Pb2Ti2O6 (defect pyrochlore) are observed. and titanium oxide has been reported.17 This defect pyrochlore phase has the same composition as that of the perovskite, Further calcination at 400 °C for 4 h [Fig. 3(e)] results in the evolution of the perovskite phase, in addition to although it has a different crystal structure. Pb2Ti2O6 is a metastable cubic phase with a lattice parameter of 10.44 A° . (PbO zTiO2)ss and the defect pyrochlore phase, Pb2Ti2O6. Calcination at 450 °C for 4 h leads to the complete disappear- The X-ray diffraction lines observed in this study correspond to the characteristic pattern of the pyrochlore.The formation ance of (PbO zTiO2)ss reflections accompanied by a drastic decrease in the intensities of the defect pyrochlore reflections of the pyrochlore phase is attributed to its lower density in comparison to that of the perovskite phase.Thus, after calci- and enhanced intensities for the perovskite reflections. Further annealing at 450 °C for extended periods (ca. 24 h) leads to the nation at375 °C for 4 h, the X-ray reflections of the PbO zTiO2 J. Mater. Chem., 1997, 7(3), 521–526 523formation of crystalline PbTiO3 [Fig. 3(f )] with complete phase purity. At no stages of calcination are the X-ray reflections correspondingto TiO2 (anatase, rutile or brookite) observed, indicating that the amorphous phase of titania hydrate gives rise to PbTiO3 through the formation of the intermediates (PbO zTiO2)ss and Pb2Ti2O6.The conversion of the defect pyrochlore to the perovskite is a continuous process with no mixed phase formation, as is evident from the X-ray diffractograms.The powders prepared are highly reactive because of the low temperature of formation of the product. The product yield is nearly quantitative (99%). Highly crystalline and phase-pure powders are obtained. The sequence of reactions in the preparation of PbTiO3 can be formulated as follows: TiO2 xH2O�TiO2 yH2O+(x-y)H2O (1) PbCO3�PbO+CO2 (2) PbO+zTiO2 yH2O�(PbO zTiO2)ss+yzH2O (3) 2PbO zTiO2+2(1-z)TiO2 yH2O �Pb2Ti2O6+2(1-z)yH2O (4) Fig. 2 TG–DTA data in air for (a) hydrated titania, (b) PbCO3 and Pb2Ti2O6�2PbTiO3 (5) (c) a composite precipitate of PbCO3 and hydrated titania These steps differ completely from the solid–solid reaction starting from the constituent oxides PbO and TiO2. The preparation of PbTiO3 via the ceramic route is well known,9–11,13 wherein the reaction between the crystalline phases takes place between 600 and 700 °C.Under the dynamic conditions of thermal analysis, reactions (4) and (5) are not detectable in the DTA curves, as there are no large energy changes. Moreover, they are kinetically slower processes. This accounts for the necessity of annealing these composite powders for longer durations at temperatures between 350 and 450 °C.Lead zirconium titanate (PZT) Fig. 4 shows the TG–DTA curves of the as-prepared PZT powder after drying at 120°C for 12 h. The sequence of reactions followed is the same as that observed for PbTiO3. Three endotherms are observed, one at 65°C (mass change 17%), the second around 249 °C and the third at 299 °C (combined mass change ca. 9%). In comparison, the DTA curve of the solid–solid reaction between PbO, TiO2 and ZrO2 shows an endotherm followed by an exotherm (Fig. 5). This pattern is essentially a composite of the two binary mixtures, as the formation of PbTiO3 is exothermic in nature and that of PbZrO3 is endothermic.11 It can be seen that these reactions occur at much higher temperature in comparison to that observed for the present route.Thus, it is very clear that the nature of the reaction is completely different in the case of the product from carbonate–gel precipitation. Fig. 6 shows the X-ray diffraction patterns of the powders calcined at different temperatures. The as-prepared powder Fig. 3 X-Ray diffractograms [at (a) 120, (b) 275, (c) 350, (d) 375, (e) 400 and (f ) 450 °C] showing the evolution of crystalline PbTiO3 from the composite powder.D, PbCO3 ; #, PbO; $, (PbO zTiO2 )ss; Fig. 4 TG–DTA data in air of the Pb(Zr0.5,Ti0.5)O3 composite powder *, Pb2Ti2O6; +, PbTiO3 524 J. Mater. Chem., 1997, 7(3), 521–526Fig. 7 TEM image of the Pb(Zr0.5,Ti0.5)O3 powder indicating particle sizes of ca. 0.3–1.2 mm Fig. 5 DTA plots in air for (a) PbO+TiO2 , (b) PbO+ZrO2 and (c) PbO+0.5 TiO2+0.5 ZrO2 mixtures (for more details see ref. 11) Table 2 Compositions of PZT prepared PZT compositions phase symmetry PbZrO3 pseudo-tetragonal (orthorhombic) Pb(Zr0.8,Ti0.2)O3 rhombohedral Pb(Zr0.6,Ti0.4)O3 rhombohedral Pb(Zr0.5,Ti0.5)O3 rhombohedral Pb(Zr0.48,Ti0.52)O3 rhombohedral+tetragonal Pb(Zr0.45,Ti0.55)O3 tetragonal Pb(Zr0.4,Ti0.6)O3 tetragonal Pb(Zr0.2,Ti0.8)O3 tetragonal PbTiO3 tetragonal Fig. 6 X-Ray diffractograms of the Pb(Zr0.5,Ti0.5)O3 composite precipitate calcined at (a) 400 and (b) 450 °C (dried at 120°C) showed the presence of the crystalline lead carbonate phase. Heat treatment at 400 °C for 4 h results in the formation of the mixed phase perovskite [Pb(Zr0.5,Ti0.5)O3] (PZT), (PbO)ss and the defect pyrochlore phase [Pb2(Zr0.5,Ti0.5)2O6]. Further annealing at 450 °C for 24 h results in the evolution of a crystalline perovskite with Fig. 8 Variation of lattice parameters with composition in the single-phase characteristics. Here again, no X-ray reflections PbTiO3–PbZrO3 system corresponding to any of the crystalline phases of ZrO2 and TiO2 are observed at any stage of the thermal annealing.Well defined acicular crystallites of PZT are obtained, as pathway followed by the composite powders is completely shown by the TEM image (Fig. 7). The particle size, calculated different from both the products of other wet chemcial pro- by the linear intercept method from the micrograph, is of the cesses and the solid–solid reaction between the binary oxides. order of 0.3–1.2 mm. The absence of a ferroelectric anomaly may be the result of The various compositions of PZT prepared are listed in the low particle size. Homogeneous mixed oxides of the Table 2.Depending upon the Zr/Ti ratios in the starting raw perovskite series are obtained at relatively low temperatures material, the final product formed will be tetragonal, rhombo- in comparison to the ceramic processes, in which reactions go hedral or orthorhombic (pseudo-tetragonal). The cell dimen- to completion only above 700°C.The new method has the sions of the product show systematic variations with changes following advantages: (1) it is simple to adopt; (2) the raw in the Zr/Ti ratios (Fig. 8), with the end member, PbZrO3, materials involved are commonly available reagents; (3) highly exhibiting a pseudo-tetragonal symmetry with cell parameters reactive powders are obtained; (4) the product yield is >98%, of a¾=4.158 A° and c¾=4.108 A° .and (5) highly phase-pure and crystalline products are obtained. Conclusions References The carbonate–gel method results in composite powders with submicron particles of lead carbonate embedded within the 1 A. J. Moulson and J. M. Herbert, Electroceramics, Chapman and Hall, London, 1990.amorphous hydrated titania or zirconia gel. The reaction J. Mater. Chem., 1997, 7(3), 521–526 5252 P. P. Phule and S. H. Risbud, J. Mater. Sci., 1990, 25, 1169. 11 S. S. Chandratreya, R. M. Fulrath and J. A. Pask, J. Am. Ceram. Soc., 1981, 64, 422. 3 Better Ceramics T hrough Chemistry, ed. C. J. Brinker, D. E. Clark and D.A. Ulrich, Mater. Res. Soc. Symp. Proc., North-Holland, 12 T. R. N. Kutty, R. Vivekanandan and P. Murugaraj,Mater. Chem. Phys., 1988, 19, 533. Amsterdam, 1984, vol. 32. 4 L. M. Brown and K. S. Mazdiyasni, J. Am. Ceram. Soc., 1972, 13 B. Jaffe, W. R. Cook and J. Jaffe, Piezoelectric Ceramics, Academic Press, London, 1971. 55, 541. 5 S. Kaneko and F. Imoto, Bull. Chem. Soc. Jpn., 1978, 51, 1739. 14 Y. Matsuo and H. Sasaki, J. Am. Ceram. Soc., 1963, 46, 409. 15 K. Takai, S. Shoji, H. Naito and A. Sawaoka, Proc. 1st Int. Symp. 6 A. N. Christensen and S. E. Ramussen, Acta Chem. Scand., 1963, 17, 845. Hydrothermal Reactions, ed. Somiya, Association for Scientific Document Information Publications, Tokyo, 1983, p. 877. 7 R. Balachandran and T. R. N. Kutty, Mater. Res. Bull., 1984, 19, 1479. 16 JCPDS Powder Diffraction File, Inorganic Volume, card no. 26–142. 8 H. S. Gopalakrishnamurthy, M. Subbarao and T. R. N. Kutty, J. Inorg. Nucl. Chem., 1976, 38, 417. 17 F. W. Martin, Phys. Chem. Glasses, 1965, 6, 143. 9 K. Aykan, J. Am. Ceram. Soc., 1968, 51, 577. 10 J. Thomson, Jr., Bull. Am. Ceram. Soc., 1974, 53, 421. Paper 6/06449C; Received 18th September, 1996 526 J. Mater. Chem., 1997, 7(3), 521

ISSN:0959-9428    

DOI:10.1039/a606449c

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

Thermal oxidation of carbothermal b¾-sialon powder: reaction sequence and kinetics Kenneth J. D. MacKenzie,*a Shiro Shimadab and Takenori Aokib aNew Zealand Institute for Industrial Research and Development, P.O. Box 31–310 L ower Hutt, New Zealand bFaculty of Engineering, Hokkaido University, Sapporo 060, Japan The oxidation of carbothermally synthesised b¾-sialon powder (z=2.45) was found by X-ray powder diffraction and solid-state 29Si and 27Al MAS NMR spectroscopies to result in the initial formation of amorphous SiO2 and a non-crystalline mullite-like aluminosilicate which becomes crystalline at higher temperatures. The evolution of crystalline mullite is accompanied by a decrease in both the sialon z value and in the proportion of tetrahedral Al.The ratio of 29Si in the SiO2 to 29Si in the mullite oxidation products is about 4, consistent with the formation of Al2O3 as an additional oxidation product.The oxidation kinetics at 1100–1300 °C are described by a parabolic rate law with an activation enthalpy of 161 kJ mol-1, suggesting the rate-determining step to be the permeation of oxygen through the oxidised layer coating the sialon grains.Many applications for advanced ceramics call for their use at trometry indicated the following typical impurity levels (in elemental mass%), carried over from the clay starting material: elevated temperatures in air. The thermal oxidation of nonoxide ceramics such as silicon nitride and sialons is therefore 0.13% Fe, 0.03% Ti, 0.06% Ca, <0.01% K, <0.005% Mg, <0.001% Cr.The only impurity phase detectable by XRD a limiting factor in such applications, although in practice the progress of oxidative degradation can be slowed significantly was a trace of X-phase sialon, with no evidence of glass. Samples (1.0 g) were oxidised in air at 950–1400 °C for 1 h in by the limited access of air in highly dense bodies, and, in silicon nitride, by the formation of a structurally dense subsur- platinum crucibles, using an electric muffle furnace.The powders were then examined by XRD (Philips PW 1700 computer- face Si2N2O layer interposed between the bulk Si3N4 and the SiO2 oxidation product.1 Of the many published kinetic studies controlled diffractometer with graphite monochromator and Co-Ka radiation). The 29Si and 27Al MAS NMR spectra were of Si3N4 oxidation,2–7 most were performed on dense sintered bodies which contained sintering additives such as MgO2–4 or obtained at 11.7 T using a Varian Unity 500 spectrometer and a 5 mm Doty MAS probe spun at 10–12 kHz under the rare-earth-metal oxides5 which influence the rate of the typically parabolic oxidation kinetics by becoming incorporated following conditions: 29Si: 6 ms p/2 pulse, recycle delay 300 s, shifts referenced to tetramethylsilane; 27Al: 1 ms p/10 pulse for in the subsurface interface layer.5 The one published study of the oxidation kinetics of undoped Si3N4 powder6 should solution, recycle delay 5 s, shifts referenced to 1 mol dm-3 aqueous Al(NO3)3 solution.represent a limiting case of the fastest reaction parameters uncomplicated by side reactions, and provide a benchmark The oxidation kinetics were measured by isothermal thermogravimetry of the sialon powder at 1105–1305 °C in a Rigaku against which more slowly oxidising systems may be compared.More recently, the oxidation of Si3N4 powder containing 4 thermoanalyser with an infrared image furnace. Samples (50 mg) were brought to the reaction temperature at 20°C mass% Y2O37 has indicated the role played by the lanthanide in forming an interfacial phase below the surface SiO2 layer.min-1 under a flowing Ar atmosphere, then, after equilibration, the atmosphere was changed to flowing air (200 ml min-1), By comparison with silicon nitride, the oxidation of sialon has been much less thoroughly studied. Singhal and Lange8 this point being taken as t=0 in the kinetic experiment.The oxidation mass change was then monitored continuously for reported that the oxidation of hot-pressed sialon compacts followed a parabolic rate law, with increased oxidation resist- the duration of the experiment (2 h). ance as the Al content increased. Mullite (Al6Si2O13) was the only identifiable crystalline oxidation product.Work by Results and Discussion Chartier et al.9 on b¾-sialon (z=0.4) sintered with 14 mass% Oxidation reaction Y2O3 suggests that in such samples the oxidation kinetics are controlled by yttrium migration in the intergranular glassy Fig. 1 shows the relative intensities of the major XRD diffrac- phase. tion lines for the various crystalline phases identified in samples The aim of the present work is to establish the oxidation oxidised between 950 and 1400 °C.The diffraction peaks used kinetics of undoped b¾-sialon powder as a baseline against in this diagram were: b¾-sialon: JCPDF no. 36–1333, 3.32 A° which more oxidation-resistant sialon systems may sub- (020); mullite: JCPDF no. 15–776, 3.38 A° (210); cristobalite: sequently be compared, and to investigate the formation of JCPDF no. 11-695, 4.04 A° (101); X-phase sialon: JCPDF no. both the crystalline and non-crystalline product phases using 36-832, 3.62 A° (320). X-ray powder diffraction (XRD) and 29Si and 27Al solid-state Between room temperature and 950°C, the intensity of the MAS NMR respectively. b¾-sialon diffraction pattern shows only a very slight decrease, but above 950 °C, the intensity decreases regularly, with a plateau at ca. 1150–1200 °C (Fig. 1). The small amount of X- Experimental phase present appears to be more resistant to low-temperature oxidation, its intensity remaining virtually unchanged until its The b¾-sialon was synthesised from high purity New Zealand halloysite clay by carbothermal reduction, as described else- abrupt disappearance <1150 °C (Fig. 1). Fig. 2 shows the changes during oxidation of the mean z where.10 XRD showed it to be predominantly monophasic with z=2.45 and a particle size distribution, determined value of the sialon phase, deduced by the method of Ekstro�m et al.11 from careful measurements of the a and c cell parameters by laser interferometry, of 100%<6 mm, 50%<2 mm, 10%<1 mm. Chemical analysis by X-ray fluorescence spec- using elemental Si as the external calibration standard.J. Mater. Chem., 1997, 7(3), 527–530 527Fig. 1 Semi-quantitative representation of the phases in carbothermal b¾-sialon powder as a function of the oxidation temperature, 1 h holding time at each temperature Fig. 3 Typical 11.7 T MAS NMR spectra of carbothermal b¾-sialon powder oxidised for 1 h at the indicated temperatures.A–E, 29Si spectra (A, unheated; B, 950°C; C, 1000–1100°C; D, 1100–1200°C; E, 1300–1400°C); F–I, 27Al spectra (F, unheated; G, 950–1000°C; H, 1050–1200°C; I, 1250–1400 °C). Asterisks indicate spinning side bands. Fig. 2 Mean z value of carbothermal b¾-sialon powder as a function of oxidation temperature, 1 h holding time at each temperature small amount of uncombined amorphous silica; this resonance The experimental error in these measurements will be great- increases markedly from the first stages of oxidation, becoming est at the higher oxidation temperatures, owing to the weaker narrower and shifting at higher temperatures towards the diffraction patterns as the sialon phase disappears; nevertheless, reported value for cristobalite (d -108.5).14 Another small the results indicate a decrease in the amount of aluminium in resonance at d -88 to -90, which can be discerned even at the sialon as oxidation proceeds, with an apparent disconti- the lowest oxidation temperature used here, evolves at higher nuity at ca. 1250 °C. temperatures into a sharp peak corresponding to the major The early stages of oxidation are not accompanied by the resonance of mullite (d -86.415).Thus, from the onset of appearance of any crystalline oxidation product; mullite makes oxidation, both amorphous silica and a trace of an aluminosil- its appearance 1100 °C (Fig. 1), and grows steadily in icate similar to mullite are detected by 29Si NMR spectroscopy intensity and crystallinity above that temperature.Careful but not by XRD. By 1100°C, the mullite structure has become measurements of the cell parameters of the evolving mullite, sufficiently organised for XRD detection, but the silica oxi- made using elemental Si as the external calibration standard, dation product remains X-ray amorphous until 1300 °C, when were used to deduce the composition of the mullite, from the some is converted to cristobalite. These changes were charted relationship of Cameron.12 The least reliable results for this by integration of the 29Si spectra over the diagnostic spectral phase were those of the lower-temperature samples, in which regions of the three Si-containing phases, to give a represen- the mullite was only poorly crystalline.Using the measured tation of the changes in partitioning of Si within the system mullite cell volumes, the mullite composition was found in during oxidation (Fig. 4). most of the samples to be 60.2–62.0 mol% Al2O3, i.e. corre- Fig. 4 indicates that the ratio of Si in SiO2 to Si in mullite sponding to 352 mullite. The alumina contents deduced from is about 4 at 1400°C. On this basis, the oxidation reaction the mullite a parameters alone were slightly poorer in alumina.approximates to: No trend was found between the alumina content of the mullite and the oxidation temperature. 6Si7Al5O5N11CA O2 4Al6Si2O13+34SiO2+3Al2O3+33N2 Above 1300 °C, a small amount of crystalline cristobalite (SiO2) becomes evident (Fig. 1). A selection of typical 29Si and (1) 27Al MAS NMR spectra of the oxidised sialons are shown in Fig. 3. giving a predicted ratio of Si in SiO2 to Si in mullite of 4.25, and indicating the formation of some Al2O3 . Although XRD During the course of oxidation, the 29Si spectra show a progressive intensity decrease of the diagnostic peak for b¾- evidence for the latter is lacking, the 27Al spectra of samples heated to higher temperatures confirm the presence of a small sialon, which occurs at d -47.6 to -48.4.13 The progressive downfield shift of this peak during oxidation is apparently amount of alumina (see below).No NMR evidence was found for the significant formation of oxynitride intermediates in the real, but the reason is not presently understood. The broad feature at d -116 in the unheated sialon (Fig. 3A) is due to a oxidation process. 528 J. Mater. Chem., 1997, 7(3), 527–530parabolic rate law which has been found most frequently to describe the oxidation behaviour of sialons and Si3N4: Dm2=kt+c (2) where Dm is the oxidation mass change per unit area, k is the reaction rate constant and c is a numerical constant. Scanning electron micrographs of the unoxidised and variously oxidised samples indicated clearly that the particle size and morphology are unchanged over the temperature range of the kinetic experiments, consistent with the well known difficulty of sintering undoped b¾-sialon.For the purpose of this kinetic analysis, the particle area was therefore treated as a constant. Eqn. (2) was found to fit the present data reasonably well, although, as was found for the oxidation of Si3N4 powder,6 the results suggest an initial induction period which may be Fig. 4 Partitioning of 29Si over the various phases in oxidised car- due to the presence of an oxidised layer on the original bothermal b¾-sialon powder as a function of oxidation temperature particles, or to genuine initial linear oxidation behaviour, or both. The parabolic rate law was also found to describe the The 27Al spectra (Fig. 3F–I) show a change in the ratio of initial portions of the kinetic curves reasonably well, and gave tetrahedral Al sites (at d ca. 55 to 68) to octahedral (d -5 to a similar activation enthalpy; the full data set was therefore 1.6) as oxidation proceeds. The change in the position and used to determine the parabolic rate constants, shown in Fig. 6. shape of the tetrahedral peak corresponds to the conversion The slope of the resulting Arrhenius plot (Fig. 7) gave a of Al in b¾-sialon (d 66–6913) to mullite, in which two tetra- value for the activation enthalpy (temperature dependence of hedral peaks can sometimes be resolved, at d 42.2–48.3 and the oxidation rate) of 161 kJ mol-1. This enthalpy is low by 56.6–66.1.15 The position and squarish shape of the tetrahedral Al peaks in the present samples oxidised at 1250–1400 °C (Fig. 3I) reflects these two unresolved tetrahedral sites. The proportion of tetrahedral sites, determined by curve-fitting the 27Al spectra, decreases progressively as oxidation progresses, achieving a high-temperature value (52%) consistent with that deduced from the published 11.7 T spectrum of sintered synthetic mullite.15 At the higher oxidation temperatures, an additional small resonance appears at d ca. 14 (Fig. 3I); this arises from a small amount of a-Al2O3, present in too small a concentration to be detected by XRD. In summary, the combined XRD and NMR evidence indicates that undoped b¾-sialon powder oxidises at >900 °C with the formation of X-ray amorphous SiO2 , the associated Al residing in aluminosilicate regions which subsequently form crystalline mullite >1050 °C.During oxidation, the Al content of the sialon, as reflected in its z value, decreases slightly, although the Al content of the mullite product remainsconstant at about the 352 mullite composition. The concentration of Fig. 6 Kinetic data for oxidation of carbothermal b¾-sialon powder excess Al is insufficient to be detected by XRD, but appears plotted according to the parabolic rate law: (a) 1105 °C; (b) 1150°C; at higher temperatures in the 27Al NMR spectrum as a-Al2O3.(c) 1175°C; (d) 1260 °C; (e) 1305 °C Crystalline SiO2 (cristobalite) is also detected at higher temperatures. No evidence was found for the formation of signifi- cant oxynitride intermediates in the oxidation process.Oxidation kinetics The isothermal mass-gain curves for five oxidation temperatures are shown in Fig. 5. These were analysed in terms of the Fig. 5 Isothermal oxidation curves for carbothermal b¾-sialon powder Fig. 7 Arrhenius plot for oxidation of carbothermal b¾-sialon powder in air: (a) 1105°C; (b) 1150 °C; (c) 1175 °C; (d) 1260°C; (e) 1305°C J. Mater. Chem., 1997, 7(3), 527–530 529comparison with the only other literature value for b¾-sialon amorphous SiO2 and an aluminosilicate with NMR character- (440 kJ mol-1).9 The samples used in that study were, however, istics similar to mullite, but only acquiring sufficient ordering very different, being sintered bodies of z=0.4, containing Y2O3 to be detected by XRD above 1100 °C.At higher temperatures as a sintering additive, all of which factors are expected to (>1300 °C) some of the amorphous silica is converted to affect the oxidation kinetics; sintered bodies oxidise more crystalline cristobalite.slowly than powders, whereas the oxidation rate decreases As the reaction proceeds, the sialon z value decreases, with with decreasing z value.8 Additives such as Y2O3 are reported a discontinuity at 1250 °C.The proportion of tetrahedral Al to give rise to linear rather than parabolic kinetics in Si3N4 (deduced by 27Al MAS NMRspectroscopy) decreases progress- powders,7 and might exert a similar influence on sialon pow- ively with temperature, to a value consistent with crystalline ders. For these reasons, a more meaningful comparison might 352 mullite. be made with undoped Si3N4 powder, for which activation At 1400 °C the distribution of 29Si in the two principal enthalpies ranging from 14716 to 285 kJ mol-1 6 have been oxidation products (SiO2 and mullite) suggests Al2O3 should reported. As expected, these activation enthalpies are generally be a further reaction product; this phase was detected by 27Al lower than the range reported for sintered Si3N4 compacts MAS NMR spectroscopy but not by XRD.(2933 to 440 kJ mol-1 4), but since these samples normally The oxidation kinetics of carbothermally synthesised b¾- contain oxide sintering additives which are known to enter sialon powder at 1100–1300 °C can be described by a parabolic into the oxidation process, the relative effects of densification rate law with an activation enthalpy Ea of 161 kJ mol-1 and and sintering additive on the oxidation kinetics are difficult to a negative activation entropy.The Ea value is similar to that distinguish. An even greater spread of activation enthalpies for oxygen diffusion in both SiO2 and mullite, suggesting that has been reported for the oxidation of Si3N4 CVD films (88 the rate-determining step is the permeation of oxygen through to 628 kJ mol-1 17).the oxidised product layer coating the sialon grains. The rather low activation enthalpy found for the present samples is of similar magnitude to that reported for the diffusion of oxygen in fused SiO2 (122 kJ mol-1 18), suggesting that in our b¾-sialon powder containing no additives, the rate- References determining process might be the diffusion of oxygen through 1 H.Du, R. E. Tressler and K. E. Spear, J. Electrochem. Soc., 1989, the SiO2 surface layers. Alternatively, oxygen diffusion through 136, 3210. mullite may be a rate-determining process in sialon oxidation, 2 S. C. Singhal, J.Mater. Sci., 1976, 11, 500. since mullite is another principal product. Data for oxygen 3 W.C. Tripp and H. C. Graham, J. Am. Ceram. Soc., 1976, 59, 399. self-diffusion in mullite are not available, but oxygen ion 4 D. Cubicciotti and K. H. Lau, J. Am. Ceram. Soc., 1978, 61, 512. conductivity measurements in mullite19,20 indicate an acti- 5 D. M. Mieskowski and W. A. Sanders, J. Am. Ceram. Soc., 1985, 68, C160. vation enthalpy which is dependent on the thermal history, 6 R.M. Horton, J. Am. Ceram. Soc., 1969, 52, 121. and ranges from 63 to 112 kJ mol-1 for undoped 352 mullite. 7 P. S.Wang, S.M. Hsu, S.G. Malghan and T. N.Wittberg, J.Mater. These enthalpies can be equated to the activation enthalpy for Sci., 1991, 26, 3249. random diffusion of oxygen in mullite, and their similarity to 8 S. C. Singhal and F. F. Lange, J. Am. Ceram. Soc., 1977, 60, 190.the present oxidation enthalpy, while tending to confirm 9 T. Chartier, J. L. Besson and P. Goursat, Int. J. High T ech. Ceram., oxygen diffusion as rate determining, militate against an unam- 1986, 2, 33. biguous identification of the phase presenting the diffusion 10 K. J. D. MacKenzie, R. H. Meinhold, G. V. White, C. M. Sheppard and B. L. Sherriff, J.Mater. Sci., 1994, 29, 2611.barrier. 11 T. Ekstro�m, P. O. Kall, M. Nygren and P. O. Olsson, J. Mater. From absolute rate theory,21 the Gibbs activation energy Sci., 1989, 24, 1853. DG* and activation entropy DS* are given by: 12 W. E. Cameron, Am. Ceram. Soc. Bull., 1977, 56, 1003. DG*=RT [ln(RT/Nh)-ln k] (3) 13 J. Sjo�berg, R. K. Harris and D. C. Apperley, J. Mater. Chem., 1992, 2, 433. DS*=(Ea-DG*)/T (4) 14 J.V. Smith and C. S. Blackwell, Nature (L ondon), 1983, 303, 223. 15 L. H. Merwin, A. Sebald, H. Rager and H. Schneider, Phys. Chem. where R is the gas constant, N is Avogadro’s number and h is Miner., 1991, 18, 47. Planck’s constant. 16 P. Goursat, P. Lortholary, D. Tetard and M. Billy, Proc. Int. Symp. The values of DG* thus calculated fall in the range React. Solids, Bristol, 1972, ed. J. S. Anderson, Chapman and Hall, 504–587 kJ mol-1, and the DS* values fall in the range -249 London, 1972, pp. 315–326. to -271 J K-1 mol-1. The negative value of the activation 17 J. D. Choi, D. B. Fischbach and W. D. Scott, J. Am. Ceram. Soc., 1989, 72, 1118. entropy is consistent with the transition state being substan- 18 E. L. Williams, J. Am. Ceram. Soc., 1965, 46, 190. tially more ordered than the reactants, as would be expected 19 G. Meng and R. A. Huggins, Solid State Ionics, 1984, 11, 271. for sorption of oxygen on to the surface of the sialon grains. 20 G. Meng, W. Cao and D. Peng, Solid State Ionics, 1986, 18&19, 732. 21 H. Eyring, J. Chem. Phys., 1935, 3, 107. Conclusions b¾-Sialon powder (z=2.45) produced by carbothermal synthesis oxidised in air above 900 °C with the initial formation of Paper 6/04354B; Received 24th June, 1996 530 J. Mater. Chem., 1997, 7(3), 527&ndash

ISSN:0959-9428    

DOI:10.1039/a604354b

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

Preparation of oriented cadmium sulfide nanocrystals Z. Y. Pan, G. J. Shen, L. G. Zhang, Z. H. Lu and J. Z. Liu National L aboratory ofMolecular and Biomolecular Electronics, Southeast University, Nanjing, 210096, China Cadmium sulfide (CdS) particulate films, composed of highly oriented, rod-like nanocrystals have been generated in situ by the exposure of stearic acid (SA) Langmuir monolayer-coated aqueous CdCl2 solutions to hydrogen sulfide (H2S).The SA-coated CdS particulate films were transferred to a solid substrate and examined by transmission electron microscopy (TEM) and Auger electron energy spectroscopy. It was found for the first time that the electron diffraction pattern was a composite one with six sets of diffraction patterns which were contributed by the different oriented CdS nanocrystals in these particulate films system.The epitaxial growth of rod-like CdS nanocrystals has been rationalized in terms of matching the d220 spacing of the cubic CdS crystals and the d101� 0 spacing of the hexagonal closed-packed SA monolayer. The presence of a negatively charged monolayer at the air/water interface was an essential requirement for the oriented growth of CdS nanocrystals.This leads to a novel means of fabrication of highly oriented semiconductor quantum wires. Quantum confinement effects in semiconductor systems with we reported that cadmium sulfide particulate films composed of highly oriented rod-like CdS nanocrystals can be prepared reduced space dimensions have attracted considerable attention. 1–9 In this field, it is very important to assemble semicon- at the monolayer/subphase interface by exposing the stearic acid Langmuir monolayer-coated salt solution to hydrogen ductor nanocrystals in an orderly form and, at the same time, maintain the properties of each individual nanoparticle.9–13 sulfide gas. The generation of CdS was proved by the Auger electron spectrum. Dark field images in TEM and transmission Fendler’s group first recognized that semiconductor nanoparticles can be synthesized by exposing fatty acid monolayer- electronic diffraction were used to investigate the structure of the CdS particulate film in detail.It was found for the first coated aqueous salt solutions to small molecule gases.14 The preparation of PbS and PbSe particulate films composed of time that the electron diffraction pattern was a composite one with six sets of individual electron diffraction patterns which high-oriented equilateral-triangular nanocrystals and some other particulate films have been reported in detail.14 –21 implied that the CdS particulate films are composed of six sets of CdS nanocrystals. From the consideration of the electron Because of the advantages of versatility and simplicity, Fendler’s method is a very interesting, significant and attractive diffraction pattern, a growth mechanism for the CdS particulate films was proposed and, at the same time, the epitaxial growth one for the formation of inorganic semiconductor nanocrystals in an orderly form.The Langmuir monolayer at the air/water of the rod-like CdS nanocrystals was reasonably attributed to the matching of the distance of the (220) plane of the CdS interface not only provides size, geometrical control and stabilization with a single dimension for nanocrystals, but also crystals and the {101�0} planes of the hexagonal closed-packed SA monolayer.influences the structure of the particulate films.22–24 After the successful preparation of CdS monolayers within LB films and copper layers at the monolayer/subphase interface,24–26 we Experimental have tried to prepare oriented CdS particulate films induced by an organic monolayer at the air/water interface.The materials and the assembly method are similar to those reported earlier.14–21 TEM analyses of these semiconductor particulate films have been published and growth models proposed, but not enough The solution (in chloroform, 1×10-3 mol dm-3) of SA was spread on a four-times distilled water subphase containing attention has been paid to some interesting and detailed phenomena in these semiconductor systems (for example, the CdCl2 at a concentration of ca. 2×10-4 mol dm-3 and NaHCO3 at a concentration of ca. 3×10-4 mol dm-3 at composition of the oriented CdS nanocrystals films). Recently, Fig. 2 Electron diffraction pattern of the CdS particulate film in Fig. 1 Fig. 1 Typical TEM image of the CdS particulate films J. Mater. Chem., 1997, 7(3), 531–535 531pH=6.42. The in situ generation of monolayer-supported CdS films were transferred to solid substrates by horizontal lifting through the surface layers.Amorphous carbon and formvar- semiconductor particulate films was achieved as follows. In a rectangular trough the surface of the subphase was cleaned by coated 300 mesh copper grids and fresh, well cleaned silica were used as the substrates for transmission electron sweeping it with an aspirator. The SA Langmuir monolayer were compressed to their solid states to give a coverage of microscopy and Auger electron spectroscopy, respectively.TEM observations were carried out by a JEOL-2000EX elec- 20 A° 2 molecule-1. Injection of H2S at 100 ml h-1 to the air/water interface led to the slow growth of CdS particulate tron microscope operating at 160 kV. Electron diffraction patterns of individual crystallites were also taken in the selected films at the monolayer/subphase interface.These CdS particulate films were prepared through the area. Auger electron spectroscopy was performed in a AES-350 Auger electron spectrometer. reaction of H2S and Cd2+ at the monolayer/subphase interface. The presence of the SA monolayer plays an important role in the formation of the CdS particulate films. In order to avoid Results and Discussion changing the deposition procedure,27 the horizontal transferring method was chosen to transfer the monolayer-supported The present system is different from bulk semiconductors and from dispersed semiconductor particles.It is a highly oriented CdS particulate films to the substrate after the reaction. After 2 h reaction, the monolayer-supported CdS particulate particulate film which consists of a large number of uniform Fig. 3 Top: The composite diffraction pattern composed of six sets of diffraction spots. Spots a–f were contributed by the crystals shown in a–f. 532 J. Mater. Chem., 1997, 7(3), 531–535rod-like CdS nanocrystals. TEM is a very powerful tool surface. The alkyl chains of SA, fully extended in the air in a planar zigzag conformation, are oriented approximately because it allows us to see the arrangements, sizes and other important physical characteristics of the crystallites.28 The normal to the surface in a hexagonal close-packed lattice.14–21 The hexagonal close-packed structure of the monolayer at the combination of the TEM dark field image with an electron diffraction pattern enables us to investigate the structure and air/water interface has been discussed previously.27,30,31 Fig. 4 shows the proposed hexagonal close-packed structure of the the growth mechanisms of these particulate films. Fig. 1 is a typical TEM image of a CdS particulate film. The monolayer. From the structure of the monolayer and the area per molecule (20 A° 2), it is shown easily that the lattice constant rod-like CdS particles of length ca. 100 nm aligned in three directions.These three directions were parallel to the edge a is 4.81 A° . The diffraction pattern of Fig. 2 shows that the d220 and d111 directions of an equilateral triangle. Some dot-like CdS particles with no regular shape were found also. Because the film spacingsof cubic CdS crystal are 1.94 A° and 3.16 A° respectively. A comparison of the double d220 spacing of the CdS crystallites is very thin, the bright light image is not very clear. Longer reaction times could result in thicker films which are more with d101� 0 of the SA monolayer (4.16 A° ) revealed a 6.8% mismatch between the template and the crystals.The mor- readily observed, leading to the formation of the layer-by-layer structure, such a multilayer structure wouvery difficult phology of these rod-like CdS nanocrystals [Fig. 1, 3(a)–(c)] is rationalized by the small mismatch (good fit) between d220 to analyse.14–21 Fortunately, the combination of the electron diffraction pattern and the corresponding TEM dark field of the CdS crystals and d101� 0 of the hexagonal close-packed SA monolayer. This is illustrated in detail in Fig. 5. Thus the image enables us to see the morphology and composition of the particulate films clearly.preferred orientation of the [110] axis of the CdS crystal is parallel to the monolayer and perpendicular to the electron Fig. 2 is the transmission electron diffraction pattern from the area shown in Fig. 1, displaying a symmetric pattern with beam. Therefore the rod-like CdS crystallites parallel to the edge directions of an equilateral triangle are reasonably attri- somewhat dispersed and elongated spots.This implies that the CdS nanocrystals are not oriented randomly. The orientation buted to the three equivalent (101�0) planes of the hexagonal close-packed SA monolayer. within the distribution is limited, i.e. there is pronounced texture. The TEM pattern shows that the diffraction arches There is a 30° angle between the line connecting the two diffraction arches produced by the corresponding rod-like CdS should be indexed as the (220) (outer circle) and (111) (inner circle) faces of the CdS cubic lattice of the zinc-blende struc- nanocrystals and that of the adjacent diffraction arches produced by the corresponding dot-like CdS nanocrystals.This ture.29 In order to analyse the composition and the growth mechanisms of CdS microcrystals under the SA Langmuir result implies that the dot-like CdS nanocrystals were induced by the {112�0} planes of the hexagonal close-packed SA mono- monolayer, the TEM dark field image technique was used.The results were totally different from those found for PbSe layer because the same angle is found between the [101�0] and [112�0] axes of the SA monolayer.The mismatch between the particulate films. It was found that the adjacent arches in one diffraction circle and the arches in the different diffraction d111 spacing (3.16 A° ) of the CdS crystal and the d112� 0 spacing (4.81 A° ) of the SA monolayer is 31.7%. On the other hand, circles were contributed by different CdS nanocrystals.Only the two arches symmetrical about the diffraction centre the mismatch between the double d111 spacing of the CdS crystal and the d112� 0 spacing of the SA monolayer is 34.2%. (symmetric inversion) were contributed by the same CdS nanocrystals, as shown in Fig. 3 (top). These two symmetrical arches compose a set of diffraction patterns. Six sets of individual electron diffraction patterns compose the diffraction pattern of CdS particulate films i.e.the diffraction pattern is actually a composite one. The different dark field images, corresponding to different sets of diffraction arches are shown in Fig. 3(a)–(f ) respectively. In Fig. 3(a)–(c) the CdS nanocrystals were aligned in only one direction, but these three directions were different.The nanocrystallites in Fig. 3(d)–(f ) were dot-like. In this experiment, the magnetic rotation angle of the TEM was 186°, the long axes of the rod-like CdS crystals in Fig. 3(a)–(c) were parallel to the line connecting the two corresponding diffraction arches and, at the same time, these three directions were parallel to the edges of an equilateral triangle as shown in Fig. 1 and Fig. 3. It was mentioned above that the CdS crystallites were formed by the exposure of the SA monolayer-coated CdCl2 solution to H2S gas. The surface charge density due to the head groups of the monolayer are important. Because of ionization of the carboxyl head groups of SA, negatively charged SA monolayers in their solid state consist of CH3(CH2)16COO- ions which are ordered two-dimensionally at the air/water interface.The electrostatic attraction between the positively charged Cd2+ and the negatively charged head group of the monolayer will lead to a very high Cd2+ concentration at the monolayer/subphase interface where the nucleation of CdS nanocrystals is initiated. The preferential two-dimensional growth of the CdS crystallites is also reasonably attributed to the high local reagent concentration at the monolayer/subphase interface.Therefore the structure of the head group (monolayer) plays an important role in the fabrication of the CdS particulate films. If the SA monolayer is compressed to its solid state, the Fig. 4 The hexagonal close-packed structure of the SA monolayer: (a) plan view; (b) three-dimensional representation carboxylate groups are aligned perpendicular to the water J.Mater. Chem., 1997, 7(3), 531–535 533Fig. 6 Auger electron spectrum of the surface of the CdS particulate films. Cd5S=(HCd/0.98):(HS/0.67)#1 (0.98 and 0.67 are the sensitivity factors of Cd and S respectively29). films formed at the monolayer/liquid interface in a future paper. The sizes (width 5–10 nm, length 100 nm) of each individual rod-like CdS crystallite are size-quantized; their sizes are comparable to the de Broglie electron wavelength, the mean free paths of excitons.21–26,33 This implies that the rod-like Fig. 5 (a) The crystal structure of the cubic CdS crystal and CdS is a kind of quantum wire, which may lead to novel (b) schematic diagram of the different CdS nanocrystal growth applications and devices.orientations. ———>, Growth direction of the rod-like crystals; The reaction of a monolayer-covered subphase with small ------>, growth direction of the dot-like crystals. gaseous molecules provides a potential method for nanofabrication of quantum dots and quantum wires. The most important factor in the preparation of this quantum confine- In other words, the mismatch between these two kinds of ment system is the match of the crystal and the template (the crystals faces is difficult to accommodate.The greater mismatch monolayer on the surface of the subphase), in other words the compared with that of the d220 spacing of the CdS crystal and match of the face distancesof the two kinds of crystal structures. the SA monolayer (6.8%) resulted in the morphology of the If the semiconductor and the monolayer are suitable, the dot-like nanocrystals [Fig. 3(d)–(f )]. Although the preferred fabrication of perfect and highly oriented quantum wires is growth orientation of dot-like CdS crystallites is [111] and possible. This result opens the door to the colloid chemical this direction is perpendicular to the electron beam, the generation of semiconductors with unusual crystal structures crystallites cannot grow long enough to produce rod-like and controllable dimensions with unique electric, optical and morphology.This result is shown in Fig. 5, which shows the electro-optical properties. hexagonal close-packed structure of the Langmuir monolayer. The other important phenomenon is that diffraction spots The generation of CdS was also studied by surface Auger were observed instead of the usual diffraction arches.This was electron spectroscopy (Fig. 6). Auger electron spectroscopy due to the structure of the compressed monolayer, which was shows that the Cd5S ratio was ca. 151.32 not a strict hexagonal close-packed structure. It was also the The PbSe studied in earlier work had a face-centred cubic reason why the rod-like CdS particles were not strictly aligned structure,29 similar to CdS.It was found that the morphologies in one direction; there were small direction differences for of the PbSe particulate films were different at different surface different CdS rod-like crystallites. This implies that the pressures, in other words the surface pressure of the Langmuir structure of the monolayer is the other important and neces- monolayer influences the morphology of the particulate film.sary factor for the fabrication of highly oriented quasi-two- We found that the pH of the subphase was another important dimensional quantum wire structures. factor which influences the morphology of the particulate film. In our experiment, NaHCO3 was used to adjuse The authors thank the reviewers for very informative subphase.We will report the influence of the subphase pH and the surface pressure on the morphology of the CdS particulate suggestions. 534 J. Mater. Chem., 1997, 7(3), 531–53519 X. K. Zhao and J. H. Fendler, J. Phys. Chem., 1991, 95, 3716. References 20 X. K. Zhao and J.H. Fendler, J. Phys. Chem., 1992, 96, 9933. 1 L. E. Brus, IEEE J. Quantum Electron., 1986, 22, 1909. 21 J. Yang, J. H. Fendler, T. C. Jao and T. Laurion, Microsc. Res. 2 R. R. Chandler and J. L. Coffer, J. Phys. Chem., 1993, 97, 9767. T ech., 1994, 27, 402. 3 H. Weller, Angew. Chem., Int. Ed. Engl., 1993, 32, 41. 22 B. R. Heywood, S. Rajam and S. Mann, J. Chem. Soc., Faraday 4 A.Mews, A. Eychmuller, M. Giersig, D. Schooss and H. Weller, T rans., 1991, 87, 727; 735. J. Phys. Chem., 1994, 98, 934. 23 B. R. Heywood and S. Mann, J. Am. Chem. Soc., 1992, 114, 4681. 5 Y. Wang and N. Herron, J. Phys. Chem., 1991, 95, 525. 24 G. P. Luo, Z. M. Ai, J. J. Hawkes, Z. H. Lu and Y. Wei, Phys. Rev. 6 H. Weller, Adv.Mater., 1993, 5, 88. B, 1993, 48, 15337 and refs. therein. 7 Y.Wang and N. Herron, Phys. Rev. B, 1992, 42, 42. 25 Z. Y. Pan, X. G. Peng, Z. H. Wu, T. J. Li, M. Zhu and J. Z. Liu, 8 L. Blus, Appl. Phys. A, 1991, 53, 465. L angmuir, 1996, 12, 851. 9 V. L. Clvin, A. N. Goldstein and A. P. Alivisatos, J. Am. Chem. 26 Z. Y. Pan, X. G. Peng, T. J. Li and J. Z. Liu, Chin. J. Sci. Instrum., Soc., 1992, 114, 5221. 1996, 17, 153. 10 J. R. Heath, Science, 1995, 270, 1315; C.B. Murray, C. R. Kagan 27 V. K. Gupta, J. A. Kornfield, A. Ferencz and G. Wegner, Science, and M. G. Bawendi, Science, 1995, 270, 1335. 1994, 265, 940. 11 C. R. Martin, Science, 1994, 266, 1961. 28 A. P. Alivisatos,MRS Bull., 1995, XX, 23. 12 S. Mann, Nature (L ondon), 1993, 365, 499. 29 O. Osugi et al., Rev. Phys. Chem. Jpn., 1966, 36, 59; Joint 13 I. Meriguchi, I. Tanaka, Y. Teraoka and S. Kagawa, J. Chem. Soc., Committee on Power Diffraction Stands, 1971. Chem. Commun., 1991, 1401. 30 J. B. Peng and G. T. Barnus, T hin Solid Films, 1994, 252, 44. 14 X. K. Zhao, Y. Yuan and J. H. Fendler, J. Chem. Soc., Chem. 31 R. Viswanathan, L. L. Madsen, J. A. Zasadzinski and D. K. Commun., 1990, 1248. Schwartz, Science, 1995, 269, 51. 15 X. K. Zhao, L. D. McCormick and J. H. Fendler, Chem. Mater., 32 Auger Electron Spectra Catalogue, A Data Collection of Elements, 1991, 3, 922. Anelva Corporation, 1979. 16 X. K. Zhao and J. H. Fendler, Chem.Mater., 1991, 3, 168. 33 Y. Wang, Acc. Chem. Res., 1991, 24, 133. 17 X. K. Zhao, S. Q. Xu and J. H. Fendler, L angmuir, 1991, 7, 520. 18 X. K. Zhao, L. D. McCormick and J. H. Fendler, L angmuir, 1991, 7, 1255. Paper 6/04867F; Received 10th July, 1996 J. Mater. Chem., 1997, 7(3), 531–535 535

ISSN:0959-9428    

DOI:10.1039/a604867f

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

Crystal structures and electronic properties of UTixNb3-xO10 (x=0,1/3,1) and of the intercalation compound Li0.9UTiNb2O10 Peter G. Dickens,* Gavin J. Flynn, Saban Patat and Gary P. Stuttart Inorganic Chemistry L aboratory, South Parks Road, Oxford, UK OX1 3QR Complete crystal structures of the related phases UTixNb3-xO10 (x=0,1/3,1) and of the intercalation compound Li0.9UTiNb2O10 have been determined by Rietveld analysis of room-temperature powder neutron diffraction data.The new structural data combined with magnetic susceptibility measurements made in the range 5<T /K<300 support a common electronic formulation of the compounds as LiIyUV1+y-xUVIx-yTiIVx NbV3-xO10 (yx1) with UV(f1) being the only paramagnetic species present. UV3O10,1 UNb3O10 2 and UTixNb3-xO10 3 belong to a family These authors suggested, on the basis of the metal–oxygen vectors associated with each metal site, that NbIV(d1) and NbV of isostructural mixed oxides of uranium having pillared-layer structures based on edge- and vertex-sharing of UO8 and MO6 occupied the MI and MII sites, respectively, with uranium present exclusively as UVI.This valence assignment conflicts, polyhedra in which edge-shared UO6 hexagons and MO4 rectangles form extended planar arrays linked together by however, with the electronic structure of UNb3O10 determined by magnetic and PES measurements by Miyake and co- metal–oxygen chains running perpendicular to them.The common interlayer chain sequence is U–O–MII–O–MI–O– workers4 which identified UV (f1) and not NbIV(d1) as the paramagnetic centre in stoichiometric UNb3O10, suggesting a MII–O–U, where the transition-metal atoms M occupy two different crystallographic sites.The characteristic interlayer formulation of the compound as UVNbV3O10. This electronic arrangement makes better chemical sense since the equilibrium separation is ca. 4 A° and the repeat distance along the chain direction ca. 16 A° . Each uranium atom is thus surrounded by oxygen pressure at 300 K, calculated from tabulated thermodynamic data,5,6 for a NbO2–Nb2O5 mixture is very much a hexagonal-bipyramidal array of oxygen atoms, the two axial oxygens above and below the plane being much closer to the lower than that for a UO3–U3O8 (or even U3O8–U4O9) couple, and hence spontaneous conversion of NbIV to NbV and UVI uranium than are the six planar oxygens.The oxygen atom arrangement around the metal M is approximately octahedral to UV in a solid solution of the oxides is the likely outcome. A further single-crystal X-ray study on ‘UTiNb2O10’ (having for both MI and MII , but an off-centre distortion in the chain direction is associated with MII . A plan of the structure, which the actual composition UTi1/3Nb8/3O10) by Chevalier and Gasperin3 led to the conclusion that this solid-solution phase is orthorhombic, viewed along the [001] chain direction, and based on the coordinates for UV3O10,1 is shown in Fig. 1. also had the UNb3O10-type structure with (TiIV1/3, NbIV2/3) occupying the MI site, NbV the MII site, and uranium again Only for this phase have accurate oxygen positions been determined by neutron diffraction.An early single-crystal study present as UVI. In order to determine the light atom positions accurately, and thus establish the precise oxygen environments of UNb3O10 by Chevalier and Gasperin2 placed the heavymetal positions satisfactorily, but large temperature factors of the metal atom sites, we report herein structure determinations, by powder neutron diffraction, of the compounds together with a poor R value (0.1) suggested that significant uncertainty in the positions of the oxygen atoms remained.UNb3O10, UTi1/3Nb8/3O10 and UTiNb2O10; for the last compound no crystal structure has been reported previously. In addition, new magnetic susceptibility measurements have been made which enable changes in electronic behaviour to be associated with changes in metal environments and thus assist in the identification of the species responsible for the observed paramagnetism in this series of related compounds. The stoichiometric compound UTiNb2O10 contains all metal atoms in their highest oxidation states and, like other oxides and mixed oxides7 containing UVI in tunnelled or layered structures, undergoes intercalation reactions at ambient temperatures in which small electropositive elements, such as lithium, are incorporated interstitially with retention of the parent oxide structure.We report below a determination, by powder neutron diffraction, of the complete crystal structure of one such compound, Li0.9UTiNb2O10, for which the changes in electronic properties caused by intercalation have been monitored by magnetic susceptibility measurements.This work is part of an ongoing investigation of the structural and electronic properties of intercalation compounds formed by the oxides and mixed oxides of uranium. Experimental Preparations The starting materials used for making samples of the mixed Fig. 1 Structure of UV3O10 in the ab plane (z=0) oxides were AnalaR-grade Nb2O5 and TiO2 and synthetic J.Mater. Chem., 1997, 7(3), 537–543 537U3O8 and NbO2. U3O8 was prepared by decomposition of was enhanced by prolonged annealing at 473 K. The indexed powder diffraction pattern of Li0.9UTiNb2O10, which was used UO4 2H2O8 as described previously.9 NbO2 was made by reduction of Nb2O5 in a stream of dry hydrogen at 1273 K; for further structural investigations, is given in Table 2. its composition and structural identity were confirmed by thermogravimetry and its powder X-ray pattern.10 Three mem- Powder neutron diffraction data bers of the U–Nb–Ti–O system were prepared according to UNb3O10.Data were collected at room temperature on the reactions (1)–(3): instrument D2B at ILL, Grenoble, from a 10 g sample con- 1/3NbO2+1/3U3O8+4/3Nb2O5�UNb3O10 (1) tained in a thin-walled vanadium can using a neutron wavelength of 1.5946 A° .The standard Rietveld method for constant Nb2O5+TiO2+1/3U3O8+1/6O2�UTiNb2O10 (2) wavelength data refinement was used in the Brookhaven 4/3Nb2O5+1/3TiO2+1/3U3O8�UTi1/3Nb8/3O10 (3) National Laboratories version13 and its application followed in all important respects that described by us recently for the The starting mixtures for reactions (1) and (3) were ground, structure refinement of USbO5.14 The starting structural pelletised and heated in evacuated sealed silica tubes for 24 h parameters chosen for refinement of UNb3O10 were those at 1373 K.To ensure complete reaction the products were given by Chevalier and Gasperin2 from single-crystal reground, repelletised and reheated in the absence of air.data. Refinement proceeded smoothly for data in the range UNb3O10 was dark brown and its powder X-ray pattern was 15<h/degrees<145 and the final cycle included 25 variables consistent with that reported by Kovba et al.11 UTi1/3Nb8/3O10 and converged with the following R values:14 R1=6.8, Rwp= was also brown and its X-ray pattern was readily indexable 7.9, Rp=5.9, RE=1.7%.Cell and positional parameters are on an orthorhombic cell of similar lattice parameters to those given in Table 3 and selected interatomic distances and bond reported by Chevalier and Gasperin.3 The starting mixture for angles in Table 4. The profile fit is shown in Fig. 2. An attempt reaction (2) was finely ground, pelletised and heated in an to improve the refinement by removal of symmetry constraints open alumina boat at 1373 K; after regrinding, repelletising through transformation to P1� led to no significant improve- and reheating a yellow single-phase product resulted whose X- ment in the goodness of fit or to changes of atom positions ray diffraction pattern (Table 1) closely resembled those of from those found for the original orthorhombic (Fddd) space UTi1/3Nb8/3O10 and UNb3O10. group.The intercalation compounds Li0.90UTiNb2O10 and Li0.34UTiNb2O10 were prepared at ambient temperature in an inert atmosphere by adding BunLi to the parent oxide in dried Table 2 Powder X-ray diffraction data for Li0.9UTiNb2O10a hexane. The procedures used followey those employed previously for the preparation (and the subsequent chemical 2hobs/degrees dobs/A° intensity index dcalc/A° characterisation by redox titration) of other uranium oxide 14.979 5.909 m 111 5.916 intercalation compounds.12 The crystallinity of the products 17.754 4.991 m 022 5.000 21.598 4.111 m 113 4.115 21.950 4.046 s 004 4.050 Table 1 Powder X-ray diffraction data for UTiNb2O10 (l=1.54056 A° )a 24.889 3.574 m 131 3.579 26.634 3.344 m 202 3.342 2hobs/degrees dobs/A° intensity index dcalc/A° 28.077 3.175 vs 040 3.178 29.432 3.032 m 133 3.035 15.011 5.897 m 111 5.892 30.955 2.886 vw 115 2.886 17.684 5.011 m 022 5.011 35.911 2.499 s 224 2.500 21.417 4.146 vw 113 4.146 37.846 2.375 m 151 2.376 21.487 4.132 s 004 4.125 39.101 2.302 m 242 2.303 25.021 3.556 w 131 3.556 41.081 2.195 m 313 2.194 26.739 3.331 w 202 3.332 41.498 2.174 m 117 2.174 28.282 3.153 vs 040 3.154 44.618 2.029 m 008 2.025 29.397 3.036 m 133 3.036 45.969 1.972 w 333 1.972 30.552 2.924 vw 115 2.924 46.480 1.952 w 137 1.957 35.833 2.504 s 224 2.506 47.047 1.930 w 155 1.930 38.123 2.359 m 151 2.360 49.652 1.834 s 260 1.835 39.327 2.289 m 242 2.290 50.845 1.794 w 246 1.795 40.766 2.211 m 117 2.208 53.640 1.707 m 228 1.708 41.234 2.188 m 313 2.187 54.879 1.672 m 264 1.671 43.817 2.064 m 008 2.063 57.986 1.589 m 080 1.589 45.858 1.977 w 137 1.979 62.755 1.479 m 084 1.479 46.178 1.964 w 333 1.964 68.989 1.360 m 268 1.360 46.984 1.932 vw 155 1.933 50.050 1.821 m 260 1.821 aRefined orthorhombic cell parameters: a=7.338(5) A° , b=12.712(4) A° , 50.647 1.801 w 246 1.801 c=16.199(7) A° . 53.026 1.725 m 228 1.726 53.486 1.712 vw 422 1.711 54.717 1.676 vw 317 1.676 55.087 1.666 w 264 1.666 Table 3 Unit-cell and positional parameters for UNb3O10a 55.084 1.666 m 404 1.666 57.651 1.598 vw 02 10 1.596 site x y z B/A° 2 58.447 1.578 m 080 1.577 61.658 1.503 vw 20 10 1.503 U 8a 1/8 1/8 1/8 0.63(4) Nb1 8b 1/8 1/8 5/8 0.73(5) 63.047 1.473 m 084 1.473 63.739 1.459 vw 11 11 1.459 Nb2 16g 1/8 1/8 0.3836(1) 0.60(3) O1 32h 0.4279(2) 0.2261(2) 0.1231(2) 0.84(5) 65.406 1.426 vw 282 1.425 66.541 1.404 vw 177 1.405 O2 16g 1/8 1/8 0.5004(2) 1.20(4) O3 16g 1/8 1/8 0.2438(2) 1.15(4) 67.477 1.387 vw 13 11 1.387 68.721 1.365 m 268 1.365 O4 16f 1/8 0.3267(3) 1/8 0.90(9) aRefined orthorhombic cell parameters: a=7.283(2) A° , b=12.616(2) A° , aCell parameters: a=7.4173(3) A° , b=12.8418(5) A° , c=15.8269(3) A° ; Z=8; space group Fddd.c=16.501(2) A° . 538 J. Mater. Chem., 1997, 7(3), 537–543Table 4 Selected bond lengths (A° ) and angles (degrees) in UNb3O10 UMO1 (×4) 2.595(2) Nb1MO1 (×4) 1.955(1) Nb2MO1 (×2) 1.960(1) UMO3 (×2) 1.880(1) Nb1MO2 (×2) 1.972(2) Nb2MO2 1.848(2) UMO4 (×2) 2.590(1) Nb2MO4 (×2) 1.960(2) Nb2MO3 2.213(2) O1iMUMO1ii 60.0(3) O1viiMNb1MO2i(O1viiiMNb1MO2ii) 90.9(3) O1iiMUMO1iv(O1iMUMO1iii) 178.7(2) O1viiMNb1MO2ii(O1viiiMNb1MO2i) 89.1(3) O1iMUMO3i 90.7(2) O1vMNb2MO1vi 170.3(2) O1iMUMO3ii 89.3(2) O1vMNb2MO2i 94.9(2) O1iMUMO4i(O1iiiMUMO4ii) 60.0(5) O1vMNb2MO3ii 85.1(2) O3iMUMO3ii 180 O1vMNb2MO4iv (O1viMNb2MO4iii) 82.8(2) O4iMUMO3i 90 O1vMNb2MO4iii(O1viMNb2MO4iv) 96.5(3) O1viiMNb1MO1viii 83.2(4) O2iMNb2MO4iii 94.0(3) O1viiMNb1MO1x 96.8(4) O3iiMNb2MO4iii 86.0(3) O1viiMNb1MO1ix(O1viiiMNb1MO1x) 178.2(7) O4iiiMNb2MO4iv 172.0(3) were satisfactory for both compounds: UTi1/3Nb8/3O10: RI= 5.5, Rwp=3.7, Rp=4.1, RE=1.7%; UTiNb2O10: RI=8.0, Rwp= 4.7, Rp=5.9, RE=1.5%.Unit-cell parameters, atomic positions and principal bond lengths and angles are summarised in Tables 5–8.In UTi1/3Nb8/3O10 the refined site occupancies confirm that titanium is disordered randomly over both niobium sites. In UTiNb2O10 there is apparently a small preference for titanium to occupy the MI sites, which may be a real effect or merely an artefact of the level of refinement achieved. Li0.9UTiNb2O10. Powder neutron diffraction data for Li0.9UTiNb2O10 were collected at room temperature on the POLARIS diffractometer.Refinement proceeded initially by including only atoms of the parent oxide framework and taking Fig. 2 Observed (points), calculated (line) and difference (lower line) the atomic coordinates listed in Table 7 as a starting model. profiles for UNb3O10 This continued to a final cycle incorporating 39 variables which converged with the following R values: RI=9.4, Rwp= UTixNb3-xO10.Neutron time-of-flight data for the two 4.2, Rp=5.7. Difference Fourier maps based on observed and oxides UTi1/3Nb8/3O10 and UTiNb2O10 were collected at room calculated structure factors were synthesised for sections in the temperature from powder samples (ca. 10 g) contained in thin- interlayer region (z#0.25) and the strongest negative peak was walled vanadium cans using the medium resolution high- observed at ca.(0,0.2,0.23). Lithium was sited at this position intensity POLARIS diffractometer at ISIS. The details of data with an occupancy of 0.25. This led to improvement in the handling, manipulation and display, and the model refinement profile fit and allowed the positional, thermal and site occu- programs used were the same as those described by us in pancy parameters for lithium to be incorporated into the final recently published structure determinations of U2V2O11 and refinement culminating in a satisfactory profile fit and the UV2O8.15 For members of the UTixNb3-xO10 family of com- following R values: RI=7.1, Rwp=3.8, Rp=4.9, RE=0.6%.pounds the starting model adopted for refinement of both data Structural parameters are given in Table 9 and bond lengths sets was that proposed by Chevalier and Gasperin3 for and angles in Table 10.The site occupancy of the lithium UTi1/3Nb8/3O10. In space group Fddd, 8 (TixNb1-x) units were refined to a value (0.24) which was consistent with the extent placed at MI sites (8b) and 16 Nb at MII sites (16 g).For these of insertion (x=0.9) determined by redox titration and no partially ordered arrangements the refinements progressed additional lithium was located in the structure. smoothly but converged with poor R values,16 RI#11 and Rwp#8%, and unreasonable temperature factors for titanium Magnetic measurements and niobium. It became clear that Ti should not be restricted to an exclusive occupation of the MI sites.Hence, the trial Magnetic susceptibilities of pure phases of UTixNb3-xO10 (x= 0, 0.34, 0.90, 1.0) were measured over the temperature range model was modified by distributing Ti evenly between the (8b) and (16 g) sites in the statistical ratio of 152. The refinements 5–300 K and at field strengths 0.1–1 T using a model S600C SQUID susceptometer (Cryogenic Ltd.).Raw values were improved dramatically and the final cycle was extended to incorporate the site occupancies as additional variables subject corrected for atomic diamagnetic contributions and the data converted to molar susceptibilities, xm(T ). In the range only to the constraints of correct overall stoichiometry and complete occupancy of each Ti/Nb site. The final R values 150–300 K the susceptibilities of all the compounds followed Table 5 Unit-cell and positional parameters for UTi1/3Nb8/3O10a site x y z B/A°2 occupancy U 8a 1/8 1/8 1/8 0.73(4) Ti1/Nb1 8b 1/8 1/8 5/8 1.45(9) 0.111(9)/0.889(9) Ti2/Nb2 16g 1/8 1/8 0.3894(1) 0.72(4) 0.111(4)/0.889(4) O1 32h 0.4275(3) 0.2262(2) 0.1216(2) 0.75(4) O2 16g 1/8 1/8 0.5009(2) 1.07(3) O3 16g 1/8 1/8 0.2416(2) 1.03(6) O4 16f 1/8 0.3263(2) 1/8 0.27(5) aCell parameters: a=7.3554(5) A° , b=12.7218(9) A° , c=15.949(1) A° ; Z=8; space group Fddd. J.Mater. Chem., 1997, 7(3), 537–543 539Table 6 Selected bond lengths (A° ) and angles (degrees) in UTi1/3Nb8/3O10 UMO1 (×4) 2.572(1) Nb1(Ti1)MO1 (×4) 1.939(1) Nb2(Ti2)MO1 (×2) 1.954(1) UMO3 (×2) 1.861(1) Nb1(Ti1)MO2 (×2) 1.983(2) Nb2(Ti2)MO2 1.776(2) UMO4 (×2) 2.557(1) Nb2(Ti2)MO4 (×2) 1.955(1) Nb2(Ti2)MO3 2.355(1) O1iMUMO1ii 60.2(2) O1viiMNb1 (Ti1)MO2i 91.6(1) [O1viiiMNb1(Ti1 )MO2ii] O1iiMUMO1iv(O1iMUMO1iii) 177.6(1) O1viiMNb1 (Ti1)MO2ii 88.4(1) [O1viiiMNb1(Ti1 )MO2i] O1iMUMO3i 91.2(2) O1vMNb2(Ti2)MO1vi 163.3(2) O1iMUMO3ii 88.8(2) O1vMNb2(Ti2)MO2i 98.3(2) O1iMUMO4i(O1iiiMUMO4ii) 60.0(1) O1vMNb2(Ti2)MO3ii 81.7(2) O3iMUMO3ii 180 O1vMNb2(Ti2)MO4iv 82.0(2) [O1viMNb2 (Ti2)MO4iii] O4iMUMO3i 90 O1vMNb2(Ti2)MO4iii 96.0(2) [O1viMNb2 (Ti2)MO4iv] O1viiMNb1MO1viii 83.1(1) O2iMNb2(Ti2 )MO4iii 96.7(1) O1viiMNb1MO1x 97.0(1) O3iiMNb2(Ti2)MO4iii 83.3(1) O1viiMNb1 (Ti1)MO1ix 176.8(2) O4iiiMNb2(Ti2)MO4iv 166.6(2) [O1viiiMNb1 (Ti1)MO1x] Table 7 Unit-cell and positional parameters for UTiNb2O10a site x y z B/A°2 occupancy U 8a 1/8 1/8 1/8 0.98(7) Ti1/Nb1 8b 1/8 1/8 5/8 0.6(2) 0.40(1)/0.60(1) Ti2/Nb2 16g 1/8 1/8 0.3980(3) 0.61(8) 0.302(6)/0.698(6) O1 32h 0.4232(4) 0.2257(3) 0.1233(3) 0.85(5) O2 16g 1/8 1/8 0.5059(2) 1.04(5) O3 16g 1/8 1/8 0.2337(2) 1.16(6) O4 16f 1/8 0.3295(4) 1/8 0.56(8) aCell parameters: a=7.2511(5) A° , b=12.5628(9) A° , c=16.448(1) A° ; Z=8; space group Fddd.Table 8 Selected bond lengths (A° ) and angles (degrees) in UTiNb2O10 UMO1 (×4) 2.506(2) Nb1 (Ti1)MO1 (×4) 1.935(1) Nb2(Ti2)MO1 (×2) 1.950(2) UMO3 (×2) 1.788(1) Nb1 (Ti1)MO2 (×2) 1.959(2) Nb2(Ti2)MO2 1.775(2) UMO4 (×2) 2.570(2) Nb2 (Ti2)MO4 (×2) 1.938(1) Nb2(Ti2)MO3 2.702(2) O1iMUMO1ii 60.7(1) O1viiMNb1(Ti1)MO2i 90.8(1) [O1viiiMNb1 (Ti1)MO2ii] O1iiMUMO1iv(O1iMUMO1iii) 178.7(2) O1viiMNb1(Ti1)MO2ii 89.2(1) [O1viiiMNb1 (Ti1)MO2i] O1iMUMO3i 90.7(1) O1vMNb2(Ti2)MO1vi 156.0(2) O1iMUMO3ii 89.3(1) O1vMNb2(Ti2)MO2i 102.0(1) O1iMUMO4i(O1iiiMUMO4ii) 59.7(1) O1vMNb2(Ti2)MO3ii 78.0(2) O3iMUMO3ii 180 O1vMNb2(Ti2)MO4iv 81.0(1) [O1viMNb2(Ti2)MO4iii] O4iMUMO3i 90 O1vMNb2MO4iii 94.3(1) [O1viMNb2(Ti2)MO4iv] O1viiMNb1(Ti1 )MO1viii 81.7(2) O2iMNb2 (Ti2)MO4iii 101.2(1) O1viiMNb1(Ti1 )MO1x 98.3(1) O3iiMNb2(Ti2 )MO4iii 78.8(1) O1viiMNb1(Ti1 )MO1ix[O1viiiMNb1(Ti1 )MO1x] 178.3(2) O4iiiMNb2(Ti2 )MO4iv 157.5(1) Table 9 Unit-cell and positional parameters for Li0.9UTiNb2O10a site x y z B/A°2 occupancy U 8a 1/8 1/8 1/8 1.76(5) Ti1/Nb1 8b 1/8 1/8 5/8 1.0(1) 0.383(8)/0.617(8) Ti2/Nb2 16g 1/8 1/8 0.3924(2) 0.43(5) 0.308(4)/0.692(4) O1 32h 0.4275(3) 0.2268(2) 0.1211(1) 1.06(4) O2 16g 1/8 1/8 0.5012(2) 1.40(3) O3 16g 1/8 1/8 0.2450(2) 1.93(5) O4 16f 1/8 0.3263(2) 1/8 0.96(6) Li 32h 0.903(2) 0.191(2) 0.249(1) 0.77(3) 0.24(2) aCell parameters: a=7.3563(2) A° , b=12.7162(3) A° , c=16.1997(4) A° ; Z=8; space group Fddd.the Langevin–Debye relationship: xm=C/T+A, where C is the tities C and A derived from these data are shown in Table 11 together with the corresponding effective magnetic moments, Curie constant and A a temperature-independent paramagnetic term. Plots of xm vs.T and xm-1 vs. T for UNb3O10 are shown meff. This last quantity is defined through the expression meff/mB=(3kC/NAmB2m0n)1/2, where n is the number of paramag- in Fig. 3 and plots of xm vs. T for the intercalation compounds LiyUTiNb2O10 (y=0,0.34,0.90) are shown in Fig. 4. The quan- netic centres in a formula unit, identified as 1-x in 540 J. Mater. Chem., 1997, 7(3), 537–543Table 10 Selected bond lengths (A° ) and angles (degrees) in Li0.9UTiNb2O10 UMO1 (×4) 2.575(1) Nb2(Ti2)MO4 (×2) 1.961(1) LiMO1 2.088(4) UMO3 (×2) 1.945(2) Nb2(Ti2)MO1 (×2) 1.955(2) LiMO2 2.206(3) UMO4 (×2) 2.560(1) Nb2(Ti2)MO2 1.763(1) LiMO3 1.841(3) Nb1(Ti1)MO1 (×4) 1.946(2) Nb2(Ti2)MO3 2.387(2) LiMO3¾ 2.352(5) Nb1(Ti1)MO2 (×2) 2.006(2) LiMO4 2.058(4) O1iMUMO1ii 60.4(2) O1viiMNb1(Ti1)MO2i 91.8(1) [O1viiiMNb1 (Ti1)MO2ii] O1iiMUMO1iv(O1iMUMO1iii) 177.2(2) O1viiMNb1(Ti1)MO2ii 88.2(1) [O1viiiMNb1 (Ti1)MO2i] O1iMUMO3i 91.4(1) O1vMNb2(Ti2)MO1vi 159.7(2) O1iMUMO3ii 88.6(1) O1vMNb2(Ti2)MO2i 100.1(2) O1iMUMO4i(O1iiiMUMO4ii) 59.8(2) O1vMNb2(Ti2)MO3ii 79.9(2) O3iMUMO3ii 180 O1vMNb2(Ti2)MO4iv 81.7(1) [O1viMNb2(Ti2)MO4iii] O4iMUMO3i 90 O1vMNb2MO4iii 95.4(1) [O1viMNb2(Ti2)MO4iv] O1viiMNb1(Ti1 )MO1viii 83.4(2) O2iMNb2 (Ti2)MO4iii 98.3(2) O1viiMNb1(Ti1 )MO1x 96.7(1) O3iiMNb2(Ti2 )MO4iii 81.7(2) O1viiMNb1(Ti1 )MO1ix[O1viiiMNb1(Ti1 )MO1x] 176.3(3) O4iiiMNb2(Ti2 )MO4iv 163.5(3) O1MLiMO2 79.6(5) O2MLiMO3¾ 107.4(7) O1MLiMO3 108.4(7) O2MLiMO4 82.7(5) O1MLiMO3¾ 78.2(4) O3MLiMO3¾ 122.1(7) O1MLiMO4 156.9(7) O3MLiMO4 94.4(6) O2MLiMO3 130.5(6) O3¾MLiMO4 93.4(6) Fig. 3 (a) xm vs.T for UNb3O10. (b) xm-1 vs. T for UNb3O10: $, measured; —, calculated. UTixNb3-xO10 and y in LiyUTiNb2O10. C and the standard Fig. 4 xm vs. T for LiyUTiNb2O10: ', y=0; $, y=0.34; #, y=0.90 physical constants used in this equation are expressed in SI units.to the c axis, are shown in greater detail. The gross features of Discussion the structure agree with those deduced by Chevalier and Gasperin2 from X-ray data but significant differences in metal– UTixNb3-xO10 oxygen distances are found (Table 4) which have consequences for the interpretation of electronic structures for this and for The structure of UNb3O10 determined by the present powder neutron diffraction study is shown in Fig. 5(a); in Fig. 5(b) the the other UTixNb3-xO10 phases examined. The local environment about uranium in each phase is hexagonal bipyramidal atomic arrangements about the three metal sites U, MI (Nb1) and MII (Nb2 ), which are members of a chain running parallel with shorter axial and longer equatorial UMO bonds present.Table 11 Magnetic data for UNb3O10 and LiyUTiNb2O10 compound T /K C/10-9 m3 mol-1 K A/10-11 m3 mol-1 meff/mB UNb3O10 150–300 2500 698 1.26 Li0.34UTiNb2O10 200–300 672 780 1.12 Li0.90UTiNb2O10 200–300 2395 688 1.30 J. Mater. Chem., 1997, 7(3), 537–543 541cantly different axial NbMO distances occurring; for example, in UNb3O10 the values found are: Nb2MO2=1.848 (1.95 A° ) and Nb2MO3=2.213 (2.17) A° .This is a not uncommon arrangement for NbV which occurs, for example, in KNbO3 (and is also found for TiIV in BaTiO3). However, an important result, guiding the interpretation of electronic structure, is that the bond-valence sums about Nb1 and Nb2 in UNb3O10, calculated using the new bond-length data and atomic parameters tabulated by Altermatt and Brown,17 are virtually identical at 5.2 and 5.1 respectively, and correspond to a common oxidation state (V) for Nb; this implies the correct electronic formulation for the unsubstituted oxide is UVNbV3O10.This assignment contradicts the conclusion of Chevalier and Gasperin2,3 who suggested that NbIV was present (at the MI site), but supports the results of Miyake et al.4 who showed by XPS measurements that oxidation of UNb3O10 to UNb3O10+x was accompanied by the removal of electrons from uranium and not from niobium.In addition the magnetic measurements made in the present work (Fig. 3) show that UNb3O10 behaves as a dilute paramagnet having, in the range 150–300 K, a temperature-independent moment meff#1.3 mB. This agrees with measurements made by Miyake et al.4 who reported similar behaviour and a temperature-independent value of meff#1.1 mB .Such a value is compatible with an isolated UV(f1) species present in a predominantly axial environment.18 For a NbIV (d1) species, in a low symmetry environment, a temperature-independent ‘spin only’ value of meff#1.7 mB would be anticipated. Alternatively, if the d1 species were sited in an octahedral environment (in a 2T state) a temperature-dependent moment would be expected.Neither situation corresponds to the observed magnetic behaviour. The main conclusions concerning the UTixNb3-xO10 phases arising from the new structural data are thus: (a) the electronic structure of the unsubstituted oxide is UVNbV3O10; (b) in the isostructural series, UTixNb3-xO10, TiIV (r=0.61 A° ) replaces NbV (r=0.64 A° ) by random substitution at both Nb1 and Nb2 sites; (c) Associated with this substitution is the removal of electrons from uranium according to TiO2 (s)+1/2O2 (g) CCDA UNb3O10 TiNb¾+3OO+hUV Fig. 5 (a) Structure of UNb3O10; (b) chain parallel to c in UNb3O10 where hUV represents the formation of UVI at a UV site. (numbering of oxygen atoms refers to Table 3) Li0.9UTiNb2O10 The observed changes in UMO bond lengths along the The framework of the parent oxide is largely undisturbed by sequence x=0, 1/3, 1.0 are summarised in Table 12.The the intercalation of lithium (at this degree of insertion) though shortening observed in the axial and equatorial UMO bond the mean UMOax distance increases from its typical ‘uranyl’ lengths on passing from UNb3O10 to the fully oxidised (UVI) value of 1.788 A° in UTiNb2O10 to 1.945 A° in UTiNb2O10 suggests electron removal from uranium occurs Li0.9UTiNb2O10 in response to electron transfer from lithium when Ti is substituted for Nb.The calculated bond-valence to uranium. In the structure of Li0.9UTiNb2O10, shown in sums17 for uranium, vU, based on the new structural data, Fig. 6, lithium is distributed randomly over the 32h interlayer quantify this progressive trend and imply that oxidation of UV sites which provide a local five-fold, approximately trigonal to UVI occurs in this sequence. From the data provided in bipyramidal, coordination by oxygen about lithium. A similar Tables 4–8 Nb1 is found to be at the centre of a nearly regular coordination geometry was found by Dickens and Powell19 octahedron of oxygens for UNb3O10 and the same environment for a lithium intercalation compound of the pillared-layer type occurs for Nb1 /Ti1 in the titanium substituted oxides with oxide a-U3O8, Li0.88U3O8 , and the lithium environments in Nb1/Ti1MO bonds falling in a narrow range of 1.93–1.98 A° .the two compounds are compared in Fig. 7. Bond-valence In contrast the environment about the MII site, either Nb2 in sums calculated for lithium on the basis of the bond lengths UNb3O10 or Nb2/Ti2 in the titanium-substituted oxides, shown in Fig. 7 give a value of ca. 1.0, as expected for an approximates to a square-pyramidal arrangement with signifi- intercalated lithium ion. As the data given in Fig. 4 and Table 11 show, the magnetic Table 12 Observed changes in UMO bond lengths with varying x for susceptibility of the compounds LiyUTiNb2O10 increases with UTixNb3-xO10 (values in parentheses are those obtained by Chevalier y from the effectively non-magnetic state of UTiNb2O10 to and Gasperin2,3) paramagnetic states where the magnetic moment per mole of bond length/A° UNb3O10 UTi1/3Nb8/3O10 UTiNb2O10 inserted lithium is ca. 1.1–1.3 mB. This value approximately equals that found for UV in the same environment in UNb3O10. mean UMOax 1.88 (1.95) 1.86 (1.79) 1.79 This provides clear evidence that the intercalation process is mean UMOeq 2.59 (2.47) 2.57 (2.39) 2.53 accompanied by electron transfer, such as has been demon- vU 4.7 5.2 6.1 strated for other UVI compounds,20 Li=LiiV+eU¾, where LiiV 542 J.Mater. Chem., 1997, 7(3), 537–543UTiO5 18 and a-U3O8.19 The facility to alter the UV to UVI ratio easily and systematically by means of lithium insertion in compounds of this type could be found useful in modifying catalytic activities. Some decrease in the UV to UVI ratio in the closely related family of compounds USb3-xTixO9–10 by substitution of TiIV for SbV in USb3O10 was shown21 to lead to greatly increased catalytic activity for the ammoxidation of propylene to acrylonitrile.G.P.S. thanks AEA technology, Harwell, for a studentship and S.P. thanks the Turkish Government for financial assistance. References 1 A. M. Chippindale, S. J. Crennell and P. G. Dickens, J. Mater. Chem., 1993, 3, 33. 2 R. Chevalier and M. Gasperin, C.R. Acad. Sci.Paris, Ser. C, 1968, 267, 481. 3 R. Chevalier and M. Gasperin, C.R. Acad. Sci. Paris, Ser. C, 1969, 268, 1969. 4 C. Miyake, S. Ohana, S. Imoto and K. Taniguchi, Inorg. Chim. Fig. 6 Occupied lithium site in Li0.9UTiNb2O10, oxygen coordination Acta, 1987, 140, 133. sphere is highlighted 5 O. Kubaschewski and C. B. Alcock, Metallurgical T hermochemistry, Pergamon Press, Oxford, 1979, 5th edn. 6 E. H. P. Cordfunke, R. J. M. Konings, G. Prins, P. E. Potter and M. H. Rand, T hermochemical Data for Reactor Materials and Fission Products, AERE, Harwell, UK, 1988. 7 A. M. Chippindale, P. G. Dickens and A. V. Powell, Prog. Solid State Chem., 1991, 21, 133. 8 G. W. Watt, S. L. Achorn and J. L. Marley, J. Am. Chem. Soc., 1950, 72, 3341. 9 P. G. Dickens, S. D. Lawrence and M.T.Weller,Mater. Res. Bull., 1985, 20, 635. 10 ASTM powder diffraction file, 19–859. 11 L. M. Kovba, E. I. Sirotkina and V. K. Trunov, Zh. Neorg. Khim. (Engl. T ransl.), 1965, 10, 188. 12 P. G. Dickens, A. V. Powell and A. M. Chippindale, Solid State Ionics, 1988, 28–30, 1123. 13 D. E. Cox, B. H. Toby and P. Zolliker, Program PROFPV (Brookhaven National Labs./Union Carbide), Oxford VAX version, July 1988. 14 P. G. Dickens and G. P. Stuttard, J.Mater. Chem., 1992, 2, 691. 15 A. M. Chippindale, P. G. Dickens, G. J. Flynn and G. P. Stuttard, J. Mater. Chem., 1995, 5, 141. 16 W. I. F. David, R. M. Ibberson and J. C. Matthewman, Profile Analysis of Neutron Powder Diffraction Data at ISIS, Rutherford Appleton Laboratory Report RAL-92-032, 1992. Fig. 7 Comparison of lithium coordination in (a) Li0.9UTiNb2O10 and 17 D. Altermatt and I. D. Brown, Acta Crystallogr., Sect. B, 1985, (b) Li0.9U3O8 41, 240. 18 P. G. Dickens, G. P. Stuttard, R. E. Dueber, M. J. Woodall and S. Patat, Solid State Ionics, 1993, 63–65, 417. 19 P. G. Dickens and A. V. Powell, J. Solid State Chem., 1991, 92, 159. represents an intercalated cation and eU¾ an electron trapped 20 R. E. Dueber, S. Patat and P. G. Dickens, Solid State Ionics, 1995, at a UVI site. UTiNb2O10 in its intercalation chemistry behaves 80, 231. as a typical framework-type host where intercalation causes 21 R. A. Innes, A. J. Perrotta and H. E. Swift, ACS Symp. Ser., 1985, marked changes in electronic properties, due to facile reduction 279, 75. of UVI, but only minimal structural change occurs since no Paper 6/05883C; Received 27th August, 1996 interlayer bonds are broken. In this respect it resembles J. Mater. Chem., 1997, 7(3), 537–543 543

ISSN:0959-9428    

DOI:10.1039/a605883c

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

Synthesis of carbon nanotubes containing metal oxides and metals of the d-block and f-block transition metals and related studies Y. K. Chen, A. Chu, J. Cook, M. L. H. Green, P. J. F. Harris,† R. Heesom, M. Humphries, J. Sloan, S. C. Tsang† and J. F. C. Turner‡ Inorganic Chemistry L aboratory, University of Oxford, Oxford, UK OX1 3QR The filling of carbon nanotubes with metals and metal oxides via one- and two-step processes is presented.Both molten media and wet chemistry solution methods have been used to introduce foreign materials into the hollow nanotube cavities. Chemical reactions inside the tubes have been carried out, including the reduction of encapsulated materials to the metals. The nature of the crystalline filling has been found to be highly dependent on the techniques used.Wet chemical methods tend to result in filling which consists of discrete crystallites, whereas molten media methods tend to give long, continuous single crystals. The macroscopic synthesis of carbon nanotubes by the arc Experimental vaporization of graphite was first reported by Ebbesen and Production of carbon nanotubes Ajayan in 1992.1 The tubes which are formed consist of 2–20 multilayers of graphene sheets that are arranged concentri- Carbon nanotubes were prepared from ca. 1 cm diameter cally in a ‘Russian doll’ manner.2 Typically, they have graphite rods by the standard modified arc-discharge method hollow internal cavities with diameters of 3–10 nm and are in 0.13 atm of helium using a dc voltage of 30 V and a current 100–500 nm in length. of ca. 180 A.8 Nanotubes prepared by the arc-vaporization method are closed at both ends. It has been shown that these caps may be removed, leading to the filling of the inner cavity with metals,3 Typical preparation for the one-step filling method metal oxides3–5 and even biomolecules such as cytochrome c.6,7 A round-bottomed flask containing a sample of closed nano- The first example of opened and filled nanotubes was reported tubes (0.5 g), azeotropic nitric acid (68%; ca. 100 ml) and the by Ajayan and Iijima, who showed that the treatment of closed soluble metal nitrate (ca. 0.5–1 g) was heated to reflux for tubes with lead in the presence of air, leads to filling with 4.5–12 h. The nitric acid solution was decanted off, and the continuous regions of metal oxide material.4,5 The tube ends black sludge was pipetted onto glass filter paper.The sample can also be removed in a selective oxidation process using was dried overnight in an oven at 60°C, and then calcined by carbon dioxide8 or oxygen,9 although both methods give poor heating in a stream of argon at 450°C for 5 h for conversion yields of opened nanotubes and cause damage to the graphitic of the nitrate to the corresponding metal oxide.Reduction to walls. Recently, we have shown that the treatment of closed the metal, where possible, was carried out in a similar manner, tubes with refluxing nitric acid gives very high yields of opened usually by heating to 500 °C under a continuous stream of H2. tubes and does not etch the tube walls.3 When this reaction is carried out in the presence of a metal nitrate, after annealing at ca. 400 °C, many of the opened tubes are found to contain Typical preparation for the two-step filling method discrete crystals of the metal oxides. The nanotubes were opened by oxidation with azeotropic Carbon nanoparticles filled with metal materials have also HNO3 at 11°C for 8–24 h.3 The tubes were then filtered onto been formed via an in situ synthesis, involving the arc evapor- glass filter paper, washed copiously with deionized water and ation of composite carbon electrodes made by blending graph- then dried in the air in an oven overnight at 160 °C.This ite powder with metals or metal oxides.10–14 The resulting method of opening nanotubes results in the formation of acid materials have closed carbon shells with the encapsulated functionalities (CO2H, OH) on the surfaces of nanotubes.18 materials being either the pure metals or metal carbides.This Heating nanotubes slowly to 900°C results in the removal of method can also give nanotubes filled with crystalline materials these acidic groups, as shown by the loss of CO2 and CO from which, depending on the conditions and the metal in question, the samples.This pre-treatment is also used when the material vary from continuous crystals that fill the entire hollow cavity which is desired to fill the nanotubes is sensitive to oxygen to discrete crystals which are positioned in different locations groups (i.e. UCl4, as described below). along the length of the tube.15 In this work, we describe the preparation and characterization of carbon nanotubes containing a wide range of metal Two-step method with metal complexes oxides, elemental metals and related materials.Preliminary reports of parts of this work have been published A sample of opened carbon nanotubes (250 mg) was added to elsewhere.3,16–18 a solution of a metal complex (ca. 1 g) dissolved in a minimal volume of solvent (ca. 5 ml). This procedure may be carried out for air-sensitive materials, using organic or aqueous solvents. The mixture was then stirred overnight and the excess † Present address: The Catalysis Research Centre, Department of solution decanted off. The resulting black sludge was pipetted Chemistry, University of Reading, Whiteknights, Reading, UK onto glass filter paper and dried in an oven overnight at 60°C.RG6 6AD. Calcination of the materials was carried out as described for ‡ Present address: Department of Chemistry and Biochemistry, University of Delaware, Newark, Delaware, DE 19711, USA. the one-step procedure. J. Mater. Chem., 1997, 7(3), 545–549 545Typical preparation for the molten media method material both inside and outside the nanotubes.If the calcination process is not carried out slowly (<5 °C min-1), few A sample of opened carbon nanotubes (ca. 0.5 g) was ground filled nanotubes are observed. Presumably this results from in a mortar and pestle with the inorganic material (ca. 0.5 g). the rapid expulsion of the solution molecules present in the The nanotubes used with air-sensitive halides, such as ZrCl4, nanotube cavities, causing the metal complex to be forced out were decarboxylated prior to use.18 The mixture was placed in of the tubes.a silica ampoule and evacuated under high vacuum for 2 h Examination of the specimens by HRTEM revealed that, in before sealing. The ampoule was placed in a temperature- many cases, it was possible to obtain firm evidence for the programmed furnace, and then heated from room temperature identity of the encapsulated materials from the direct obser- to the desired temperature at a heating rate of 10°C min-1.vation of lattice fringes of single crystals. Indeed, this technique The ampoule was kept at that temperature for 1–8 h and proved to be more reliable in terms of characterizing the allowed to cool slowly to room temperature at a rate of 1°C composition of the encapsulates than X-ray diffraction analysis, min-1.Air-sensitive samples were oxidized after heating by as the latter was capable of characterizing the bulk specimen opening the ampoule to air and allowing contact with oxygen only. The measured magnitudes of the layer separations were and water. then compared with literature values. These results are summarized in Table 1.Methods of characterization Fig. 1(a)–(c) show HRTEM images of encapsulated oxide The samples were dispersed in CH2Cl2 and sonicated for 5 min crystallites of NiO, Sm2O3 and Nd2O3 . Carbon nanotubes before being deposited onto a lacey carbon film. High- containing oxides of the metals Co, Pd, Cd, Fe, Y, La, Ce, Pr, resolution transmission electron microscopy (HRTEM) was Nd, Eu and U18 have also been prepared using this one-step performed using JEOL 4000EX, JEM 2010F and JEOL method.It is interesting to note that in the case of the Nd2O3 2000EX microscopes operated at optimum defocus with acce- filled tubes, insertion of amorphous and crystalline material lerating voltages of 400, 200 and 200 kV, respectively.Lattice between the carbon layers has occurred. This insertion causes spacings on encapsulated crystals were calculated using the the carbon layers to buckle, giving an interlayer separation of 3.4 A° separation in the graphite layers of the nanotubes as an up to 20 A° , in contrast to the normally observed 3.4 A° separa- internal standard. The overall yields of filled nanotubes were tion.Similar insertion of amorphous materials has been estimated by the visual inspection of portions of the sample observed previously when V2O5 was melted into carbon nano- using HRTEM. EDS was performed on the JEOL 2010F tubes and AuCl3 encapsulated using the two-step method. microscope operating at 200 kV using the smallest available Extensive and detailed HRTEM studies showed that this probe (0.7 nm). A Philips PW1710 diffractometer with Cu-Ka1 method of filling carbon tubes gave samples in which there radiation operating in the h–2h mode was used for the XRD were also metal oxide crystals exterior to the tube cavities.measurements which were obtained on all samples. The data These are formed from the traces of mother liquor left on the was then compared to known literature spectra obtained from filter paper after collection.Attempts to remove this extraneous the Powder Diffraction Data File (JCPDF). residue by washing, for example with water or dilute nitric acid, invariably led to the elution of the nitrates from the interiors of the tubes. Results and Discussion The TEM studies also showed that, with exception of PdO One-step method and UO2-x, samples of the oxide crystallites appeared to fill the entire internal diameter of the nanotubes (ca. 3–6 nm). This treatment results in the encapsulation of the metal nitrate Most of these crystallites are in the form of elongated particles, inside nanotubes and its conversion to the crystalline oxide. typically ranging from 10 to 40 nm in length. Occasionally, Attempts to observe metal nitrates by HRTEM were inconclusive (i.e.without heat treatment) and revealed only amorphous ellipsoidal or spherical crystallites which are considerably Table 1 HRTEM observed lattice spacings of encapsulated metals and metal oxides encapsulated metals and metal starting observed lattice correlating planes literature oxides materials spacinga/A° {hkl} spacing/A° La2O3 La(NO3)3 ·6H2O 2.8, 3.5 {-303}, {202} 2.799, 3.59 Pr2O3 Pr(NO3)3·6H2O 2.8, 3.1, 3.5, 4.1 {012}, {302}, {400}, 2.875, 3.15, 3.55, 4.17 {102} CeO2 Ce(NO3)3·6H2O 3.08 {111} 3.123 Y2O3 Y(NO3)3 ·5H2O 2.65 {400} 2.652 Nd2O3 Nd(NO3)3·xH2O 3.01 {002} 2.998 Sm2O3 Sm(NO3)3·6H2O 3.21 {222} 3.155 FeBiO3 Fe(NO3)3 ·9H2O+ 2.8, 3.9 {110} or {-110}, {100} 2.811 or 2.783, 3.95 Bi(NO3)3·5H2O UO2-x UO2(NO3 )2·6HO 3.15 {111} stoichiometry varies NiO Ni(NO3)2·6H2O 2.40 {111} 2.41 MoO3 MoO3 3.85 {110} 3.81 MoO2 MoO3 3.44, 2.47 {-111}, {111} 3.42, 2.43 ZrO2 ZrCl4 3.16, 2.89 {-111}, {1-11} 3.165, 2.84 ZrO2 ZrO(NO3)2·xH2O 2.98 {111} 2.96 Re metal KReO4 2.30 {002} 2.226 Pd metal Pd(NO3)2 2.26 {111} 2.245 Ag metal Ag(NO3) 2.40 {111} 2.359 AuCl AuCl3 5.32 {101} 5.33 Au metal AuCl3 2.325 {111} 2.355 CdO Cd(NO3)2·4H2O 1.9, 2.6 {220}, {200} 1.877, 2.655 CdS CdO+H2S 2.5, 1.8, 1.7 {102}, {200}, {004} 2.450, 1.791, 1.679 a±0.1 A° . 546 J. Mater. Chem., 1997, 7(3), 545–549case of Nd1.85Ce0.15CuO4 , a definite assignment of the Nd and Ce stoichiometries of encapsulated crystallites could not be made, with the observed lattice fringes corresponding to [Ln]CuO4 , where [Ln]=Nd2 or Nd2-xCex.It was also possible to identify large amounts of Nd2O3 located external to the nanotube cavities. The two-step method This method has been useful for filling tubes with materials which are neither soluble nor stable in refluxing nitric acid. Carbon nanotubes filled with H4SiW12O40 have been obtained by stirring the opened tubes with a concentrated solution of H4SiW12O40 in deionized water for 16 h.Nanotubes filled with RhCl3, RuCl3, PdCl2, [NH4 ] IrCl6, [Ni(g-C5H5)2] and Co2(CO)8 have also been prepared by this method. This twostep method normally gave lower percentage of filled tubes (ca. 20–30%) than the one-step in situ method. As described previously, when the two-step method was used to encapsulate AuCl3 or AgNO3, an unusually high percentage of opened nanotubes were filled with the gold chloride or silver nitrate, respectively (ca. 70%).16 In the case of gold, most of the encapsulated material consisted of spherical crystallites of the metal, ranging from 10 to 50 A° in diameter. Also present were crystallites of AuCl, which resulted from the incomplete decomposition of the starting material AuCl3. At 150°C, AuCl3 decomposes to AuCl which at higher temperatures dissociates into the elements.These AuCl crystallites could also be reduced completely to Au metal by treatment of this sample with H2 at 300 °C. Molten media methods Fig. 1 (a) Nanotube filled with nickel material; the observed fringes of 2.4±0.05 A° correspond to the distance between the (111) planes in In this variation on the two-step method, the opened nanotubes NiO.(b) Encapsulated single crystal of Sm2O3 with two sets of lattice are filled using a pure liquid media, typically a molten metal fringes seen inside a carbon nanotube. The Sm2O3 lattice fringes 90° oxide or halide. The molten material must have a surface to the nanotube wall correspond to the (400) lattice planes of Sm2O3. tension less than ca. 100–200 mN m-1 in order to be able to (c) Nanotube containing cavity intercalated crystalline Nd2O3 (in the bore of the nanotube) and intralayer intercalated Nd2O3 (arrowed). wet and fill the nanotubes.19 This filling method usually results in the formation of long, continuous crystals which occupy the entire internal diameter of the nanotube, as has been observed smaller than the cross-section of the inner diameter of the previously for an unidentified oxide phase of Pb5,20 and also nanotube are observed, as was the case for PdO andUO2-x.3,18 V2O5.21 This filling is unlike the discrete crystals obtained from Stoichiometric mixed metal oxides were also be prepared solution methods. This difference may result from the removal using solutions containing the ions of two different metals.For of solvation spheres which surround each ion and occupy part example, refluxing closed nanotubes in a nitric acid solu- of the nanotube cavity. tion containing equimolar amounts of Fe(NO3)3·9H2O and Using the molten salt technique, carbon nanotubes have Bi(NO3)3·5H2O gave, after annealing, nanotubes filled with been filled with MoO3. In a typical procedure, MoO3 was crystals of the mixed oxide, FeBiO3 (Fig. 2). Attempts were crushed with a sample of pre-opened tubes and heated in a also made to synthesize NiCo2O4, LaCrO3, MgCeO3, LaFeO3, sealed silica ampoule to 800 °C for 3 h.22 Approximately 50% and Nd1.85Ce0.15CuO4. It was found, however, that the forma- of the nanotubes were observed to be filled with long single tion of mixed oxides inside nanotubes via their mixed nitrate crystals of MoO3 usually several thousand A° in length (not solutions was far from straightforward.For example, in the shown). Molten ZrCl4, KCl/CuCl2, and UCl4 and KCl/UCl417 have also been used to fill nanotubes. The air-sensitive halides UCl4 and ZrCl4 were observed by TEM as UxOyClz and ZrO2 , after oxidation. In the case of ZrO2, HRTEM examination of the sample indicated that a large number of nanotubes had been destroyed during the reaction.The bulk of the sample was composed of nanoparticles and long, finger-shaped graphitic shards. A HRTEM image showing a surviving nanotube in which ZrO2 is encapsulated is presented in Fig. 3. Similar destruction of carbon nanotubes has been observed to occur when the molten media method is used with other metal halides which are strong Lewis acids (e.g.TiCl4, GaCl3). Reactions inside nanotubes Fig. 2 A high-resolution electron micrograph of an encapsulated FeBiO3 crystallite possessing a lattice with a 2.8 A° spacing. This Attempts to fill empty tubes with molten metals were unsuc- distance corresponds to the inter-layer spacing between the (110) cessful, presumably because the surface tension of the metals planes of FeBiO3.The crystal appears to have grown where an was too high to wet and fill the tubes.19 Therefore, an indirect internal cap is present, as the internal cavity changes diameter at this point. strategy was adopted whereby samples of a metal complex J. Mater. Chem., 1997, 7(3), 545–549 547oxide samples with hydrogen.As can be seen in Fig. 5(a), the encapsulated Ru metal occurs as small, spherical crystals and as somewhat larger elongated crystals which grow along the axis of the nanotube cavities. Another example of an in situ reaction is the formation of cadmium sulfide crystals from cadmium oxide by treatment with hydrogen sulfide at 400°C. Electron diffraction obtained from encapsulated single crystals gave patterns which match well with CdS [Fig. 5(b)]. It is interesting to note that the obtained electron diffraction pattern corresponds to thermodynamically stable hexagonal CdS rather than the kinetic product, the cubic phase. Fig. 3 A high-resolution TEM image of encapsulated polycrystalline ZrO2. The observed lattices fringes (1,2) of 2.98 A° correspond to the A related study was aimed at the preparation of high yields distance between the (111) planes of the orthorhombic form of ZrO2.of nanotubes containing gold sulfide. First, pre-opened tubes were mixed with a concentrated solution of AuCl3 in dry diethyl ether. After stirring for 24 h, the sample was treated were first introduced into the nanotubes by one of the methods with hydrogen sulfide in order to precipitate gold sulfide.described above. The metal complex was then reduced to the Examination by TEM showed a high percentage of filled tubes elemental state by treatment with hydrogen gas at elevated (ca. 70%); however, there was also a large amount of material temperatures. Fig. 4 shows a typical TEM image of a single located outside the nanotubes.Fig. 6(a) shows a typical sample crystal of nickel metal formed by reduction of an NiO crystal. of nanotubes containing polycrystalline gold sulfide. The EDS The characterization of the reduced Ni metal sample was analysis [Fig. 6(b)] shows a strong Au peak, with a clearly carried out by XRD and the direct determination of the defined shoulder corresponding to sulfur.However, based on distance between the lattice fringes of 2.05±0.05 A° , which the information obtained from the EDS microanalysis and the corresponds to the (111) spacing of Ni metal. This reaction fringe separations, a distinction between Au2S and Au2S3 could may also be carried out starting directly from nickel nitrate not be made. filled nanotubes and treating them with H2 as described above.Using the same approach, encapsulated crystals of pure palladium and cobalt have been prepared.3,18 Small crystallites Conclusions of ruthenium metal located inside nanotubes are shown in We have described several methods for the filling of nanotubes Fig. 5. These crystallites are formed by treating ruthenium with metal oxides, pure metals and other materials.The identification of the encapsulated crystals was carried out by HRTEM and the direct imaging of lattice spacings, XRD and EDS. The nature of the filling is dependent on the method used to introduce the materials to the nanotube cavities, with both the one- and two-step methods giving discrete crystalline filling and molten media giving long, continuous crystals. We have also demonstrated that once encapsulated, materials may undergo reactions inside nanotubes.This finding suggests that filled nanotubes may find potential use in catalysis. Fig. 4 A high-resolution electron micrograph of a nanotube filled with nickel metal, the faceting of the particles is typical of crystalline nickel metal. Sets of lattice fringes of 2.05±0.05 A° corresponding to the (111) spacing in Ni metal can be clearly seen.Fig. 5 (a) A high-resolution electron micrograph of nanotubes filled Fig. 6 (a) Nanotube filled with polycrystalline AuxSy particles (ca. with both small spherical and elongated ruthenium metal particles. (b) HRTEM image and electron diffraction pattern (inset) obtained 50 A° in length). (b) EDS microanalysis on the encapsulated material showing an Au peak and S shoulder indicating the formation of AuxSy.from encapsulated CdS. The indexing of the pattern is shown. 548 J. Mater. Chem., 1997, 7(3), 545–54910 M. Liu and J. M. Cowley, Carbon, 1995, 33, 225. We wish to thank Dr. J. L. Hutchinson and the Department 11 M. Liu and J. M. Cowley, Carbon, 1993, 33, 749. of Materials for the use of JEOL 4000 EX and JEOL 2010 F 12 Y.Murakami, T. Shibata, K. Okuyama, T. Arai, H. Suematsu and microscopes and the Royal Society for a University Y. Yoshida, J. Phys. Chem., Solids, 1993, 54, 1861. Fellowship (S.C.T.). 13 S. Subramoney, R. S. Ruoff, D. C. Lorents, B. Chan, R. Malhotra, M. J. Dyer and K. Parvin, Carbon, 1994, 32, 507. 14 Y. Yosida, Appl. Phys. L ett., 1994, 22, 64. 15 C. Guerret-Pie�court, Y.Le Bouar, A. Loiseau and H. Pascard, References Nature (L ondon), 1994, 372, 761. 1 T.W. Ebbesen and P. M. Ajayan, Nature (L ondon), 1992, 358, 220. 16 A. Chu, J. Cook, R. Heesom, J. L. Hutchison, M. L. H. Green and 2 S. Iijima, Nature (L ondon), 1991, 354, 56. J. Sloan, Chem.Mater., 1996, submitted. 3 S. C. Tsang, Y. K. Chen, P. J. F. Harris and M. L. H. Green, Nature 17 J. Cook, J. Sloan, A. Chu and M. L. H. Green, in preparation. (L ondon), 1994, 372, 159. 18 R. M. Lago, S. C. Tsang, K. L. Lu, Y. K. Chen and M. L. H. Green, 4 R. Seshadri, A. Govindaraj, H. N. Aiyer, R. Sen, G. N. Subbanna, J. Chem. Soc., Chem. Commun., 1995, 1355. A. R. Raju, and C. N. R. Rao, Curr. Sci., 1994, 66, 839. 19 E. Dujardin, T. W. Ebbesen, H. Hiura and K. Tanigaki, Science, 5 P.M. Ajayan and S. Iijima, Nature (L ondon), 1993, 361, 333. 1994, 265, 1850. 6 S. C. Tsang, J. J. Davis, M. L. H. Green, H. A. O. Hill, Y. C. Leung 20 P. M. Ajayan, T. W. Ebbesen, T. Ichihashi, S. Iijima, K. Tanigaki and P. J. Sadler, J. Chem. Soc., Chem. Commun., 1995, 2579. and H. Hiura, Nature (L ondon), 1993, 362, 522. 21 P. M. Ajayan, O. Stephan, P. Redlich and C. Colliex, Nature 7 J. J. Davis, M. L. H. Green, A. O. Hill, Y. C. Leung, P. J. Sadler, (L ondon), 1995, 375, 564. J. Sloan and S. C. Tsang, Eur. J. Biochem., 1996, in press. 22 Y. K. Chen, M. L. H. Green and S. C. Tsang, Chem. Commun., 8 S. C. Tsang, P. J. F. Harris and M. L. H. Green, Nature (L ondon), 1996, 2489. 1993, 362, 520. 9 P. M. Ajayan, T. W. Ebbesen, T. Ichihashi, S. Iijima, K. Tanigaki and H. Hiura, Nature (L ondon), 1993, 362, 522. Paper 6/05652K; Received 13th August, 1996 J. Mater. Chem., 1997, 7(3), 545–549 5

ISSN:0959-9428    

DOI:10.1039/a605652k

出版商:RSC    

年代:1997

数据来源: RSC