- Email: [email protected]
Comparison of the preparation of cerium oxide nanocrystallites by forward (base to acid) and reverse (acid to base) precipitation
Chemical Engineering Science 91 (2013) 102–110
Contents lists available at SciVerse ScienceDirect
Chemical Engineering Science journal homepage: www.elsevier.com/locate/ces
Comparison of the preparation of cerium oxide nanocrystallites by forward (base to acid) and reverse (acid to base) precipitation V. Morris a, P.G. Fleming a, J.D. Holmes a,b,c, M.A. Morris a,b,c,n a
Department of Chemistry, University College Cork, Cork, Ireland Tyndall National Institute, Cork, Ireland c CRANN, Trinity College Dublin, Dublin, Ireland b
H I G H L I G H T S c
c
c
c
c
G R A P H I C A L
A B S T R A C T
Forward and reverse precipitation routes to CeO2 were examined. Reverse precipitation results in smaller particles. Reverse precipitation results in precipitation products containing cerium mostly in the tetravalent state. Particle shape varies between the two precipitation methods. Ce3 þ defect sites are at much lower defect concentrations in reverse precipitation.
a r t i c l e i n f o
a b s t r a c t
Article history: Received 21 May 2012 Received in revised form 6 December 2012 Accepted 9 January 2013 Available online 16 January 2013
The morphological and structural characterization of CeO2 nanocrystallites prepared by forward and reverse precipitation techniques were investigated and compared by powder x-ray diffraction (PXRD), nitrogen adsorption (BET) and high resolution transmission electron microscopy (HRTEM) analysis. The two routes gave quite different materials although in both cases the products were essentially highly crystalline, dense particulates. It was found that the reverse precipitation technique gave the smallest crystallites with the narrowest size dispersion. This route also gave as-synthesized materials with higher surface areas. HRTEM confirmed the observations made from PXRD data and showed the two methods resulted in quite different morphologies and surface chemistries. The forward route gives products with significantly greater densities of Ce3 þ species compared to the reverse route. Data are explained using known precipitation chemistry and kinetic effects. & 2013 Elsevier Ltd. All rights reserved.
Keywords: Cerium dioxide Forward precipitation Reverse precipitation Crystallite size Anion defects Nanoparticles
1. Introduction Cerium dioxide (ceria, CeO2) has been the subject of significant scientific research in recent years due to its distinctive chemical and physical nature and a number of realized and potential applications. Ceria exhibits unique UV absorptivity (Tsunekawa
n Corresponding author at: Department of Chemistry, University College Cork, Cork, Ireland. Tel.: þ353 21 902180; fax: þ 353 21 4274097. E-mail address: [email protected] (M.A. Morris).
0009-2509/$ - see front matter & 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.ces.2013.01.016
et al., 2000a), high thermal stability and mechanical hardness at high temperature coupled to high redox activity (Trovarelli et al., 1999). Unique optical characteristics such as Raman-allowed modes shifting and broadening (Spanier et al., 2001), lattice expansion and a strong crystallite size related blue shift in ultraviolet absorption spectra (Tsunekawa et al., 2000b) have been reported. Applications for ceria include: as a support and promoter for exhaust-gas conversion (three-way exhaust catalyst) (Bekyarova et al., 1998), oxygen ion conductor in fuel cells (Yahiro et al., 1988) and gas sensors (Izu et al., 2002) amongst others.
V. Morris et al. / Chemical Engineering Science 91 (2013) 102–110
With such a variety of applications, different forms (e.g. particulate, mesoporous membrane/film) are needed and many methods of preparing these materials have been reported from these laboratories (Lyons et al., 2002; Chen et al., 2010). Ceria nanoparticles are of particular research interest because of the high chemical activity and the high surface area of small particulates. Note here that ceria is almost exclusively prepared as nanocrystallites and agglomerations of these form larger particles (care is needed in the use of terminology). The sensitivity of the cerium electronic band structure to the crystallite size suggests that changes in the chemical properties of ceria crystallites at nanoscale dimensions might be expected and the effects of crystallite size changes on e.g. the lattice parameter have been reported by Morris and co-workers amongst others (Chen et al., 2010). Another example of crystallite size effects, it has been noted that in oxygen sensor applications, by changing the particle size from micrometers to the nanometer region, the response time shortened from minutes to milliseconds (Ivanov et al., 2010). Further, the use of nanocrystalline CeO2 particles have allowed 200–400 1C decreases in sintering temperature when compared with micron sized CeO2 crystallites for the production of dense ceramics (Chen and Chen, 1993). It is clear from the work reported that the continued use of ceria in materials science requires both careful study and realization of rapid and scalable methods to synthesize different sizes/morphologies of the nanocrystallites whilst maintaining narrow size distributions. There are numerous chemical methods for the production of nano-dimensioned crystallites of ceria. These include sol–gel techniques (Chu et al., 1993; Makishima et al., 1986), forced hydrolysis (Dong et al., 1997), microemulsion (Masui et al., 1997) and precipitation (Hsu et al., 1988; Djuricic and Pickering, 1999; Zhou et al., 2002) to name but a few. In sol–gel processing, alkoxide or organometallic compounds are typically used as precursors which are expensive. Hydrothermal synthesis and forced hydrolysis often require severe conditions, such as higher temperatures and pressures and can involve longer reaction times. Microemulsion techniques are an efficient method for preparing highly size monodispersed CeO2 nanocrystallites. However, all of these methods are challenging and expensive to scale to commercially viable quantities. Precipitation is an attractive route due to the availability of cheap salt precursors, simple operation and easily scaleable for mass production (Zhou et al., 2002). Various simple precipitation agents have been used to generate small crystallite sizes (Li et al., 2001) and by the correct design, optimization and control of conditions, spherically shaped nanoparticles of around 13.5 nm can be prepared (e.g. Mazaheri et al., 2009). In a very recent study, Ji et al. (2012) carried out extensive studies of homogeneous precipitation methodology to generate good size monodispersed ceria crystallites of around 10 nm. In general, the use of simple precipitation methods to yield sub 10 nm particles is ineffective and, hence, the development of the methodologies described earlier using more complex micelle, forced precipitation and homogeneous precipitation methodologies. However, whilst academic synthesis methodology has become ever more complex, there has been relatively little systematic study of the simple precipitation process reported. In particular, we note that ‘‘reverse’’ precipitation of ceria has been scarcely reported. Hassanzadeh-Tabrizi et al. (2010) and Jalilpour and Fathalilou (2012) have used reverse precipitation to generate small crystallites but these have provided little insight to the mechanism and comparison of products to forward precipitation (Fleming et al., 2011; Taniguchi et al., 2011). In this paper, we analyze the products from precipitation of CeO2 nanocrystallites using the reverse precipitation technique and compare the results with those produced using forward precipitation methods and provide a rationale for the materials produced.
103
2. Experimental procedure 2.1. Synthesis of ceria crystallites Forward precipitation is the standard methodology used in large quantity production. Briefly, the base is added to the acid under stirring. However, despite the widespread use, care must be taken when precipitation reactions are carried out. As the base is added rapid pH variation occurs locally (at point of addition) due to mixing and more slowly through the course of the synthesis. The reaction might also be exothermic and heat variation may also occur. Crystallites may appear only later to be re-dissolved as stirring occurs. This precipitation process is complex and these transient changes in the solution can cause composition inhomogeneity (Hsu et al., 1988), wide size ranges (Chen and Chen, 1993) as well as other product variation. The sensitivity of product to varying pH can cause effects such as phase or composition coring when product stability is sensitive to pH (Fleming et al., 2011). Forward precipitation was investigated using the forced hydrolysis of cerium nitrate hexahydrate with ammonia hydroxide solutions (Dong et al., 1997). The resultant precipitates were then peptised to a sol–gel to provide more uniform product. Three different peptizing acids for comparison; citric acid, oxalic acid and nitric acid were examined, but citrate proved the most effective giving the smallest crystallites and a narrow size distribution and only these results are presented. A 1 M solution of cerium (III) nitrate hexahydrate, Ce(NO3)3 6 H2O was used. 100 ml of this solution was poured into a 250 ml beaker and stirred continuously at room temperature. The pH of this solution was measured at about 2.2. A 2 M solution of ammonium hydroxide solution was added at the rate of about 0.1 ml s 1 to a final pH of 12. This resulted in an immediate white or yellow emulsion which on stirring became brownishpurple. The precipitate was recovered by filtration and was vacuum dried on a Buchner funnel. Once dry, the precipitate was scraped into a crucible and dried overnight in the oven at 80 1C. On drying, the precipitate was lemon yellow in color and was ground down to a fine powder using a pestle and mortar. This sample was then suspended in water and peptized using drop-wise addition of citric acid until a clear sol–gel was formed. The resultant gel was dried as above and samples were calcined at 350 1C for 8hours and then portions of this were calcined at 50 1C intervals to vary the crystallite size. The ‘‘reverse’’ precipitation technique was identical to the forward precipitation technique, except that in this case the Ce (III) solution was added (at same rate above) to 100 ml of the ammonium hydroxide. All subsequent processing steps were as described above for the forward precipitation route. The initial and final pH values during precipitation were measured at around 13.5. The initial precipitate formed was yellow/white but then became brown/purple before finally turning green/yellow. The precipitate formed was aged, recovered and subsequently processed as detailed above. Unlike the forward precipitation where heat was needed to affect color change to a white/lemon color and remained purple until drying, the recovered solid precipitate turned bright yellow color during washing, filtering and vacuum drying. The color changes observed can be tentatively assigned to changes in oxidation state as it has been suggested that a cerium III precipitate product is a light purple color and a cerium IV precipitate product is light yellow (Taniguchi et al., 2011). Thus, it seems likely that the reverse precipitation material contains significantly less Ce3 þ states than the forward precipitation product. 2.2. Analysis To characterize the morphology of the nanocrystallites, powder x-ray diffraction (PXRD) patterns were recorded on a Phillips PW 3710 Xpert MPD diffractometer (y 2y mode), equipped with
104
V. Morris et al. / Chemical Engineering Science 91 (2013) 102–110
a Cu Ka radiation source (operating at an accelerating of 40 kV ˚ A nickel and an anode current of 30 mA) of wavelength 1.54 A. mask was used to eliminate Kb radiation and a standard scintillation detector was used. The instrument was calibrated against a ceria sample aged at 12001 for 24 h to ensure accuracy. Data were collected over the range of 10–7012y, using a step size of 0.031. Note that in order to measure small lattice parameter shifts ( o0.5%) and ensure maximum accuracy it is critical to ensure the sample surface is exactly (to within 0.05 mm) at the center of the axis of rotation of the goniometer and sample holder. This achieved by lowering the sample below the straight through beam and raising it to the point where 50% of the beam intensity is lost. The sample must also be level and this is ensured by rotation in both directions having equal intensity modulation effects. Unfortunately, depending on sample density, sample settling and sample holders this can lead to anomalous scattering from the sample holder which can result in ghost features. Where they are observed, they are marked in the figure. The measurement of lattice parameters were not made by simple analysis of peak positions, but rather by using detailed Rietveld simulation studies where the position of the features were quantitatively matched in simulation by changing lattice spacings within the crystal framework. The Phillips Analytical Rietveld 1.0b software package used also permitted quantitative analysis of each phase present. The crystallite size was calculated using the Scherrer equation, t ¼Kl/B cos y. All sizes are quoted for samples are an average result of 5 different reflections, between 25–601 2y. XPS data were collected on a vacuum generators ESCALAB (Mk II) using AlKa radiation. The pass energy was 50 eV and the binding energies were referenced to an adventitious C1s signal of 285.0 eV. Note that for XPS, as precipitated samples were recovered by filtration and immediate vacuum drying. The samples were then stored in vacuum to prevent air oxidation. The nitrogen adsorption and desorption isotherms at 77 K were measured using a Micromeritics Gemini 2375 volumetric analyzer (Micromeritics Instrument Corporation, UK). All samples were degassed under the flow of ultra high grade 5.0 nitrogen at 200 1C for 4 h prior to each measurement. Scanning electron micrographs (SEM) were collected on a JSM-5510 apparatus (JEOL) using a beam voltage of 5 kV. Transmission electron microscopy (TEM) was used for structural characterization. Each powder was dispersed onto holey carbon support grids and examined at 200 kV in a JEOL 2000FX microscope.
3. Results The precipitation reactions can be described by reference to known chemistry and the observed color changes noted above can be explained by this reaction. For both forward and reverse precipitation, reaction begins with solution of cerium nitrate to yield Ce3 þ as expressed in Eq. (1). The solvated cations react with OH- ions formed from water molecules via protonation of NH3 molecules (Eq. (2)). Eq. (3) shows reaction of the aqueous cerium cations with hydroxyl ions. The product Ce(OH)3 cerium (III) hydroxide is quickly precipitated out due to an extremely low solubility constant (Ksp ¼6.3 10 24 at 25 1C) (Zhou et al., 2003). It is clear that this species is formed in both reactions and results in the strong purple color. Chen et al. (2010) suggested that a strongly basic solution favors Ce4 þ compared to Ce3 þ via a base oxidation process and the precipitate can be oxidized as shown in Eq. (4). As shown in Eq. (5) the product Ce4 þ ions undergo strong hydration and form aqueous coordination shells. This reaction scheme is important because it has been shown that the rate of reaction of Eq. (4) is critical in determining the final crystallite size and Zhou et al. (2003) showed that the size of CeO2
crystallites can be significantly decreased by introducing oxygen bubbles into solution to react with the as-formed Ce(OH)3. This is presumably because it limits growth of the Ce(OH)3 particles due to formation of highly insoluble CeO2 type materials. 2Ce(NO3)3-2Ce3 þ þ6NO3
(1)
NH3 þH2O2NH4þ þOH
(2)
3þ
Ce
þ3OH 2Ce(OH)3(aq)
(3)
In large excess of base oxidation can occur: 2Ce(OH)3(aq) þ1/2O2-2CeO2(aq)þ3H2O y) þ -CeO2 2(H2O)þ (4 y)H3O þ Ce(H2O)x(OH)(4 y
(4) þx 2(H2O) (5)
This reaction scheme suggests that there may be significant differences in the product of forward and reverse precipitation. In forward precipitation, nucleation and growth occurs at quite low solution pH since base is being added to acid. Thus it might be expected that the reaction will produce much higher concentrations of Ce3 þ particularly at the early stage of the process. The observation of an initial white/yellow color is probably due to high local concentrations of base favoring Ce4 þ which then become Ce3 þ as the solution is stirred. As the reaction proceeds and the solution becomes more basic, it might be expected that the reaction begins to yield Ce4 þ , however, the formation of the tetravalent species might be limited because precipitation is complete. In reverse precipitation, the acid is added to base and material is precipitated in strongly alkaline conditions favoring initial formation of Ce4 þ . As the reaction proceeds, there is an increased tendency to precipitate Ce3 þ as the solution becomes less alkali. This is consistent with the observations made above where the precipitate was in initially white/yellow but became brown/purple. The reaction also explains why the precipitate changed to yellow during vacuum drying as a rapid surface reaction is enough to convert the surface trivalent cations to Ce4 þ whilst for the forward precipitation reaction extended times and temperatures are required for oxidation as the bulk is Ce3 þ . With such changes in the precipitation mechanism, the physical and chemical properties of the products should be markedly different. Data showing this is the case is described below. Fig. 1A shows the PXRD pattern for materials prepared via forward precipitation following calcination at various temperatures. The reflections observed are characteristic of the cubic fluorite Fm3m structure (JCPDS file 34-394, Wycoff, 1963) and the main reflections observed are indicated in the figure. The calculated lattice parameter after heating was measured at 0.541070.0005 nm in excellent agreement with the expected value of 0.5411 nm (JCPDS file 34-394, Wycoff, 1963). The asprecipitated material was dried at 50 1C and the XRD signals are weak. High background (as in the rising signal seen at low angle) and the rather weak intensity of the reflections suggests considerable XRD-amorphous content to the samples. Between 350 and 550 1C the peaks are broad and typical of very small crystallite sizes. It is only after heating to 850 1C, the peaks become sharp and symmetrical consistent with crystallite growth. There is no tendency to form any other phases, regardless of age, time or temperature. Fig. 1B shows typical Ce 3d XPS data from these materials. The as precipitated sample shows two main features typical of a sample containing significant amounts of Ce3 þ (Chen et al., 2010). These peaks are the Ce 3d 5/2 and Ce 3d 3/2 spinorbit doublet. Final state effects broaden these peaks into a poorly resolved quadruplet (see Chen et al., 2010 for more detail). A third high binding energy feature marked with the arrow in the figure is typical of Ce4 þ states and its present indicates a combination of Ce3 þ (majority) and Ce4 þ (minority) states in the materials. Following air calcination, the high binding energy feature
V. Morris et al. / Chemical Engineering Science 91 (2013) 102–110
105
Fig. 1. (A) shows XRD data of CeO2 prepared by forward precipitation technique calcined at various temperatures between 350 1C to 950 1C. (B) shows typical Ce 3d XPS data from these materials.
Fig. 2. (A) shows typical PXRD data of CeO2 prepared by the citric acid reverse precipitation technique calcined at temperatures between 350 1C to 1000 1C as shown. (B) shows typical Ce 3d XPS data from these materials.
increases in relative intensity and, after a 700 1C treatment, the spectrum contains a sextuplet structure typical of a material dominated by the Ce4 þ state (Chen et al., 2010). The Ce 3d spectra of Ce4 þ containing materials is highly complex because of final state effects caused by the transfer of electrons from O 2p to Ce 4 f states during the photoionization process (see Chen et al., 2010 for more detail). It would, thus, appear that the forward precipitated material contains cerium ions in largely the trivalent oxidation state but these are oxidized on treatment in air. Fig. 2A shows similar PXRD patterns for CeO2 prepared by reverse precipitation following calcination at various temperatures. As with forward precipitation, the characteristic cubic Fm3m fluorite structure is observed. As for the forward reverse as-precipitated materials the dried only sample is consistent with a material having a large XRD-amorphous content. Note by XRD amorphous we are not suggesting the particles have a completely disordered structure, simply that they do not coherently diffract
the x-rays. Between 350 and 500 1C the peaks are broad and typical of small crystallites and, significantly, the broadening of the diffraction features at low calcination temperature is much more apparent for these materials compared to those prepared via forward precipitation. Following calcination at higher temperatures, the peaks become significantly sharper and more symmetrical typical of rapid crystallite growth. As in forward precipitation, the samples show no tendency to form any other phases, regardless of age, time or temperature and there is no evidence of new diffraction peaks. In Fig. 2B, Ce 3d XPS data are described. As can be seen immediately, the as precipitated material has significantly more Ce4 þ content than that of the equivalent forward precipitated material. The sextuplet structure is not as well developed as at higher temperature suggesting it contains a small amount of Ce3 þ (Chen et al., 2010). Following air calcination, the spectra strongly resemble data expected from material which contains predominantly cerium in the 4 þ valence. The data strongly suggest that reverse precipitation produces
106
V. Morris et al. / Chemical Engineering Science 91 (2013) 102–110
Fig. 3. Comparison of XRD derived data sets. (A) and (B) represent the crystallite size variation as function of calcination temperature and the lattice parameter against crystallite size respectively for the forward precipitation route. (C) and (D) are similar data for the reverse precipitation route.
samples with very significantly less Ce3 þ and more Ce4 þ than those prepared via forward precipitation. The PXRD were quantified for accurate comparison of lattice parameter and crystallite size. The crystallite size and lattice parameter data are shown in Fig. 3 for both forward and reverse precipitation methods. As was seen in Fig. 1, the crystallites grow continually through the calcination regime (Fig. 3A) with more rapid growth beginning at temperatures greater than 650 1C. The minimum crystallite size observed is after the lowest temperature calcination and is 16.5 nm and this has increased to 38.9 nm after 650 1C calcination. The lattice parameter varies with crystallite size as revealed in Fig. 3B and shows a decreasing value with increasing crystallite size eventually reaching the value of around 5.411 A˚ expected for bulk ceria at the largest crystallite sizes. Lattice parameter expansion at low crystallite sizes has been ascribed to increased amounts of anion vacancies and accompanying Ce3 þ defects (since the trivalent cation has a larger ionic radius than that of the tetravalent cation) at the surface of nanocrystallites and the change with temperature due to the decreasing contribution of the surface as crystallite size increases (due to sintering) and/or oxidation at higher temperatures. As discussed by Chen et al. (2010) lattice parameter reduction may also be explained by the presence of tri-valent cerium hydroxyl states rather than anion vacancy production. In reverse precipitation the crystallite growth with calcination temperature follows a similar trend (Fig. 3C). However, the crystallites grow at an increased rate. After 350 1C calcination the crystallite size compared to the forward precipitation route is lower at a value of 10.5 nm. However, following 950 1C calcination, the crystallite size is higher at 57.2 nm. This is consistent with the expected increase in sintering and densification known for smaller crystallites and is due to increased crystallite– crystallite contact and lower mass transport limitations (Chen
and Chen, 1993). The change in lattice parameter with crystallite size shows the opposite trend to that of forward precipitation. At smaller crystallite sizes the lattice parameter is reduced compared to the bulk value and increases with crystallite size until the bulk value is attained. Chen et al. have argued such lattice contraction arises from surface tension effects in small particles (crystallites) (Chen et al., 2010). Chen et al. argued the precipitation route is important in understanding lattice parameter changes for CeO2 since the amount of surface hydroxyl states causing lattice expansion at small crystallite sizes can mask the inherent lattice contraction due to surface tension (Chen et al., 2010). It would, thus, appear that the reverse precipitation route is linked to lower concentrations of Ce(OH)3 type states and suggests that the majority of the product has been converted to CeO2 (2H2O) consistent with the higher pH during the precipitation process. This is supported by the XPS data provided in Figs. 1 and 2 which showed significant concentration of Ce3 þ states for only the forward precipitation materials. As reverse precipitation is the addition of the metal nitrate (acid) to a large excess of base, the lattice parameter measurements observed are consistent with the methodology. The surface area of these materials was also measured by the BET (nitrogen) technique and the results are consistent with the PXRD measurements showing higher surface areas for those samples with smaller crystallites. Typical isotherms for forward and reverse precipitation are shown in Fig. 4(A and B). All isotherms, bar some of the samples at highest calcination temperatures, can be classified as type IV (Rouquerol et al., 1994) typical of mesoporous materials. The shape of the isotherms and the measured crystallite sizes suggests inter-crystallite (i.e. voids between particles) pores are present. As can be seen in Fig. 4, the mesopore structure is essentially lost for the reverse precipitation process (but not the forward precipitation) as little hysteresis is observed. This is consistent with the suggested higher sintering
V. Morris et al. / Chemical Engineering Science 91 (2013) 102–110
107
Fig. 4. Illustrative pore volume vs. relative pressure for forward (A) and reverse (B) precipitation. Samples at various calcination temperatures shown as detailed in the figure.
rate suggested from analysis of the PXRD data. This is confirmed in the data shown in Fig. 5 which describes the quantification of the BET derived surface area data for the samples prepared via reverse precipitation compared to values calculated via the Scherrer crystallite sizes. The latter was estimated assuming an octahedrally shaped crystallite which is a reasonable assumption considering previous work (Wang, 2000; Wang and Feng, 2003) and the TEM data shown below. The surface area was calculated from the ratio of the surface area ( ¼2O3a2) and volume (1/3O2a3) using the Scherrer calculated values as a. It can be seen that the data are in reasonable agreement at low calcination temperatures but the measured surface area decreases more markedly than the calculated data at temperatures 4650 1C. The low temperature agreement (any variance can be assumed to be due to the simple geometric calculations and disregarding of the contact between crystallites) again suggests the crystallites are dense and have negligible intra-crystallite porosity (i.e. pores within the particles). It is suggested (but other explanations may be possible) that the data diverge at higher temperatures because significant crystallite–crystallite condensation is occurring as well as the inter-crystallite pores becoming closed and thus non-available to nitrogen adsorption.
Fig. 5. Surface area vs. calcination temperature for reverse precipitation. ’ are BET measured data and m are calculated from XRD derived crystallite size data assuming octahedrally shaped crystallites.
108
V. Morris et al. / Chemical Engineering Science 91 (2013) 102–110
Fig. 6 compares the variation in surface area of materials prepared via forward and reverse precipitation following different calcination treatments. The surface area of the materials prepared via forward precipitation is more complex than that of similar forward precipitation prepared samples and shows a quite clear maximum surface area at 550 1C as can be seen in Fig. 6. The data suggest some type of thermal decomposition is taking place from around 400 1C. The surface area increases from 26.8 to
Fig. 6. Surface area (via N2 adsorption isotherms) vs. calcination temperature for forward (m) and reverse (’) precipitation.
54.9 m2 g 1 following calcination at 350 and 550 1C and then decreases until it follows the same curve as that shown by the reverse precipitated materials (comparison data shown in Fig. 6). It is suggested that the different behavior derives from strong a mixture of strong hydrogen bonding and condensation of –Ce–OH groups (loss of water and formation of –Ce–O–Ce– linkages) at the points of contact between crystallites and so decreasing the surface area below that expected from the crystallite size. As the temperature increases, the Ce(OH)3 species are oxidized to CeO2 and the surface area apparently increases (as clearly shown in the XPS data in Fig. 1). Again all the data is consistent with higher concentrations of Ce(OH)3 for the forward precipitation technique. It should be noted that the decomposition of cerium hydroxyl species formed in nanocrystallites is seen around 400 1C to support the suggestion made here (Chen and Chen, 1993; Djuricic and Pickering, 1999). TEM confirm the suggestions made above and typical micrographs of samples prepared via forward and reverse precipitation following 350 1C are shown in Fig. 7. In all cases only crystalline particles were observed and there was no evidence of amorphous material or pores within the crystallites as was suggested by the BET and XRD studies. Average crystallite sizes are in reasonable agreement with XRD measurements at about 15–20 nm for forward precipitation and 8–12 nm for reverse precipitation and confirm the smaller size of crystallites formed via the reverse precipitation method. The value of 15–20 nm for the crystallite size via forward precipitation is very much consistent with the limits seen in other work and the lower values of 10 nm and below seen by reverse precipitation represent significant improvement. The crystallite shape of each preparation route is
Fig. 7. TEM images (high and low magnification) of samples following 350 1C calcination. (A) and (B) represent a sample prepared via forward precipitation and (C) and (D) via reverse precipitation.
V. Morris et al. / Chemical Engineering Science 91 (2013) 102–110
also markedly different. From forward precipitation the products are not only larger but also have more well-defined shape with two different morphologies present (Figs. 7A and B). A number of crystallites have cubic and pseudo-cubic geometry (as marked with dashed lines in Fig. 7A) whilst others have a platelet structure (see the darker image shown in Fig. 7B). Cubic CeO2 crystallites are well known and are thought to arise via Oswald Ripening of more spherical shapes precipitated in the early stages of growth (Mai et al., 2005). In the case of reverse precipitation, the crystallites are noticeably smaller and the size dispersion much less (Fig. 7C). The crystallites now appear to adopt largely one shape have a more dimensionally uniform shape appearing almost spherical in nature. Closer examination reveals them to be polyhedral with relatively well-defined crystal facets as indicated in Fig. 7D and are similar to shapes seen in previous work (Wang, 2000; Wang and Feng, 2003) and can be best described as truncated octahedra (Zhang et al., 2007). The smaller size of the materials prepared via reverse precipitation can be explained by increased oxidation rate (Wang, 2000) and/or kinetic and thermodynamic limitations where good evidence has been provided that the truncated octahedra morphology is the precursor to platelet and cubic morphologies due to orientated attachment along the well-defined crystallite faces (Niesz et al., 2010).
4. Discussion Forward and reverse precipitation gives markedly different products. Pourbaix diagrams show that Ce3 þ is stable in solution at normal pH values and that a minimum pH value of 4–6 must be reached before formation of Ce(OH)3 occurs (O’Keefe et al., 2004). However, the Ce(OH)3 formed has some solubility to a pH of about 10 when precipitation of the hydroxide is rapid. As the pH is raised further, the formation of Ce4 þ (Ce(OH)4 or CeO2 2H2O) products becomes more likely, consistent with the reaction model given above in Eqs. (1)–(5). It thus seems that the reverse precipitation, where the pH through the process is uniformly high, is likely to give a CeO2 type product whilst the forward precipitation process is more likely to give Ce3 þ precipitates particularly during the early stages of the process where the pH changes from acidic through to alkaline. This suggestion is apparently borne out by both the color changes observed during synthesis and the lattice dimensions measured by XRD as well as previous XPS measurements (Chen et al., 2010). What is also important is that the products of the precipitation are smaller and have lower size dispersions. This is also partly explained by the Pourbaix diagrams as the entire precipitation process occurs at a pH value where solubility is limited and thus nucleation, re-solution and the size related thermodynamic stability (Dirksen and Ring, 1991) of the crystallites has a much lesser effect and the statistical nature of the crystallite growth process is limited. Zhou et al. offer a detailed rationale for the size homogeneity of the product (Zhou et al., 2002). These authors point out that the solutions with high pH are highly supersaturated and this provides conditions for homogeneous nucleation. It should also be noted that growth of large crystallites is reliant on two processes; the normal Ostwald Ripening and also orientated attachment (Zhou et al., 2003). Orientated attachment is enhanced by surface hydroxyl groups (Ce(OH)3) and so will tend to form larger crystallites. It is also interesting to note that the orientated attachment mechanism promotes formation of cubic (Zhou et al., 2003) and linear (Ivanov et al., 2010) crystallite morphologies. It can, therefore, be reasonably asserted that this is the cause of the cubic and platelet structures seen for the forward precipitation route (Fig. 7) and supports the higher Ce(OH)3 content suggested by experimental evidence presented here for
109
this preparation route. It might be thought that the oriented attachment is limited for CeO2 type products because of the lower surface hydroxyl densities and thus the truncated octahedral shape is maintained. From this work, it is suggested that reverse precipitation, which is often ignored as a synthesis method, can provide a useful methodology to produce small crystallites producing a more size homogenous product and smaller crystallite sizes/ higher surface areas. However, the products can have markedly different crystallite shapes and surface chemistries. They are likely to be more ‘passive’ to crystallite–crystallite attachment and also contain lower defect (including structural defects such as grain boundaries, pores and point defects such as anion vacancies/Ce3 þ centers) densities as they consist of single crystallites following homogeneous precipitation. It should be recognized that this methodology of reverse precipitation should not be considered as a technique to rival current state-of-the-art laboratory work but rather to enhance large-scale precipitation methods. Often, in small scale synthesis, researchers have used surfactant, ligand attachment, forced synthesis etc., to produce smaller, size monodisperse and non-aggregated nanoparticles. Whilst these techniques can produce almost ideal materials, the cost of the chemicals and processing might prevent scale-up to large quantities. The reverse precipitation technique described here could be used in simple batch or continuous flow apparatus at relatively low costs and provide a robust and reproducible method for producing small particles.
Acknowledgments We wish to thank Dr. Simon Newcomb of Glebe Scientific for TEM data. We would also like to thank SFI through the FORME SRC Grant for research support. References Bekyarova, E., Fornasiero, P., Kaspar, J., et al., 1998. CO oxidation on Pd/CeO2–ZrO2. Catal. Today 45, 179–183. Chu, X., Chung, W.I., Schmidt, L.D., 1993. Sintering of sol–gel prepared submicrometer particles by transmission electron microscopy. J. Am. Ceram. Soc. 76, 2115–2118. Chen, L., Fleming, P., Morris, V., et al., 2010. Size-related lattice parameter changes and surface defects in ceria nanocrystals. J. Phys. Chem. C 114, 12909–12919. Chen, Pei-Lin, Chen, I-Wei, 1993. Reactive cerium(IV) oxide powders by the homogeneous precipitation method. J. Am. Ceram. Soc. 76, 1577–1583. Dirksen, J.A., Ring, T.A., 1991. Fundamental of crystallization: kinetic effects on particle size distributions and morphology. Chem. Eng. Sci. 46, 2389. Djuricic, B., Pickering, S., 1999. Nanostructured cerium oxide: preparation and properties of weakly agglomerated powders. J. Eur. Ceram. Soc. 19, 1925–1934. Dong, X., Hong, G., Yu, D., et al., 1997. Synthesis and properties of cerium oxide nanometer powders by pyrolysis of amorphous citrate. J. Mater. Sci. Technol. 13, 113–116. Fleming, P.G., Holmes, J.D., Otway, D.J., et al., 2011. The complex synthesis and solid state chemistry of ceria-lanthana solid solutions prepared via a hexamethylenetetramine precipitation. J. Solid State Chem. 184, 2595–2600. Hassanzadeh-Tabrizi, S.A., Mazahere, M., Aminzare, M., et al., 2010. Reverse precipitation and characterization of CeO2 nanopowder. J. Alloys Compd. 491, 499–502. Hsu, W.P., Ronnquist, L., Matijevic, E., 1988. Preparation and properties of monodispersed colloidal particles of lanthanide compounds. 2. Cerium (IV). Langmuir 4, 31–37. Ivanov, V.K., Polezhaeva, O.S., Tret’yakov, D., 2010. Nanocrystalline ceria: synthesis, structure-sensitive properties, and promising applications. Russ. J. Gen. Chem. 80, 604–617. Izu, N., Shin, W., Murayama, N., et al., 2002. Resistive oxygen gas sensors based on CeO2 fine powder prepared using mist pyrolysis. Sens. Actuators B-Chem. 87, 95–98. Jalilpour, M., Fathalilou, M., 2012. Int. J. Phys. Sci. 7, 944–958. Ji, H., Chen, Y., Chen, S., 2012. Synthesis and characterization of nanocrystalline ceria powder through homogeneous precipitation method. Adv. Mater. Res., Adv. Mater. Process. II 557–550, 567–571. Li, J.G., Ikegamai, T., Wang, Y., Mori, T., 2001. Reactive ceria nanopowders via carbonate precipitation. J. Am. Ceram. Soc. 85, 2376–2378.
110
V. Morris et al. / Chemical Engineering Science 91 (2013) 102–110
Lyons, D.M., Ryan, K.M., Morris, M.A., 2002. Preparation of an ordered mesoporous phase of ceria with enhanced thermal stability. J. Mater. Chem. 12, 1207–1212. Mai, H.-X., Sun, L.-D., Zhang, Y.-W., et al., 2005. Shape selective synthesis and oxygen storage behavior of ceria nanopolyhedra, nanorods and nanocubes. J. Phys. Chem. B 109, 24380–24385. Makishima, A., Kubo, H., Wada, Y., Shimohira, T., 1986. Yellow coatings produced on glasses and aluminium by the sol-gel process. J. Am. Ceram. Soc. 69, c127–129. Masui, T., Fujiwara, K., Machida, K.I., et al., 1997. Characterization of Ce (IV) oxide ultrafine particles prepared using reversed micelles. Chem. Mater. 9, 2197–2204. Mazaheri, S.A., Hassanzadeh-Tabrizi M., et al., 2009. Synthesis of CeO2 nanocrystalline powder by precipitation method. In: Proceedings of the 11th ECERS Conference, Krakow. Niesz, K., Reji, C., Nielson, J.R., et al., 2010. Unusual evolution of ceria nanocrystal morphologies promoted by a low-temperature vapor diffusion based process. Cryst. Growth Des. 10, 4485–4490. O’Keefe, T., Yu, P., Hayes, S., et al., 2004. Fundamental evaluation of the deposition of cerium oxide for conversion coating applications. Office of the Department of Defense, DOD Corrosion Policy and Oversight, Washington, DC. Rouquerol, J., Avnir, D., Fairbridge, C.W., et al., 1994. Recommendations for the characterization of porous solids. Pure Appl. Chem. 66, 1739–1758. Spanier, J., Robinson, R., Zhang, F., et al., 2001. Size dependent properties of CeO2-y as studied by Raman Spectroscopy. Phys. Rev. B 64 (245407), 1–8.
Taniguchi, T., Katsumata, K.-I., Omata, S., et al., 2011. Tuning growth modes of ceria-based nanocubes by a hydrothermal method. Cryst. Growth Des. 11, 3754–3760. Trovarelli, A., de Leitenburg, C., Boaro, M., et al., 1999. The utilization of ceria in industrial catalysis. Catal. Today 50, 353–367. Tsunekawa, S., Fukuda, T., Kasuya, A., 2000a. Blue shift in ultraviolet absorption spectra of monodisperse CeO2 x nanoparticles. J. Appl. Phys. 87, 1318–1321. Tsunekawa, S., Sahara, R., Kawazoe, Y., et al., 2000b. Origin of the blue shift in ultraviolet absorption spectra of nanocrystalline CeO2 x particles. Mater. Trans. 41, 1104–1107. Wang, Z.L., 2000. Transmission electron microscopy of shape-controlled nanocrystals and their assemblies. J. Phys. Chem. B 104, 1153. Wang, Z.L., Feng, X., 2003. Polyhedral shapes of CeO2 nanoparticles. J. Phys. Chem. B 107, 13563–13566. Wycoff, R.W.G., 1963. 2nd Ed.Crystal Structures, vol. 1. Interscience Publishers, New York, UK. Yahiro, H., Baba, Y., Eguchi, K., et al., 1988. High temperature fuel cell with ceria– yttria solid electrolyte. J. Electrochem. Soc. 135, 2077–2088. Zhang, J., Ohara, S., Umetsu, M., et al., 2007. Colloidal ceria nanocrystals: a tailormade crystal morphology in supercritical water. Adv. Mater. 19, 203–206. Zhou, X.D., Huebner, W., Anderson, H.U., 2002. Nucleation synthesis and thermal stability of nanometer CeO2 single crystals. Appl. Phys Lett. 80, 3814. Zhou, X.D., Huebner, W., Anderson, H.U., 2003. Processing of nanometer-scale CeO2 particles. Chem. Mater. 15, 378–382.
Contents lists available at SciVerse ScienceDirect
Chemical Engineering Science journal homepage: www.elsevier.com/locate/ces
Comparison of the preparation of cerium oxide nanocrystallites by forward (base to acid) and reverse (acid to base) precipitation V. Morris a, P.G. Fleming a, J.D. Holmes a,b,c, M.A. Morris a,b,c,n a
Department of Chemistry, University College Cork, Cork, Ireland Tyndall National Institute, Cork, Ireland c CRANN, Trinity College Dublin, Dublin, Ireland b
H I G H L I G H T S c
c
c
c
c
G R A P H I C A L
A B S T R A C T
Forward and reverse precipitation routes to CeO2 were examined. Reverse precipitation results in smaller particles. Reverse precipitation results in precipitation products containing cerium mostly in the tetravalent state. Particle shape varies between the two precipitation methods. Ce3 þ defect sites are at much lower defect concentrations in reverse precipitation.
a r t i c l e i n f o
a b s t r a c t
Article history: Received 21 May 2012 Received in revised form 6 December 2012 Accepted 9 January 2013 Available online 16 January 2013
The morphological and structural characterization of CeO2 nanocrystallites prepared by forward and reverse precipitation techniques were investigated and compared by powder x-ray diffraction (PXRD), nitrogen adsorption (BET) and high resolution transmission electron microscopy (HRTEM) analysis. The two routes gave quite different materials although in both cases the products were essentially highly crystalline, dense particulates. It was found that the reverse precipitation technique gave the smallest crystallites with the narrowest size dispersion. This route also gave as-synthesized materials with higher surface areas. HRTEM confirmed the observations made from PXRD data and showed the two methods resulted in quite different morphologies and surface chemistries. The forward route gives products with significantly greater densities of Ce3 þ species compared to the reverse route. Data are explained using known precipitation chemistry and kinetic effects. & 2013 Elsevier Ltd. All rights reserved.
Keywords: Cerium dioxide Forward precipitation Reverse precipitation Crystallite size Anion defects Nanoparticles
1. Introduction Cerium dioxide (ceria, CeO2) has been the subject of significant scientific research in recent years due to its distinctive chemical and physical nature and a number of realized and potential applications. Ceria exhibits unique UV absorptivity (Tsunekawa
n Corresponding author at: Department of Chemistry, University College Cork, Cork, Ireland. Tel.: þ353 21 902180; fax: þ 353 21 4274097. E-mail address: [email protected] (M.A. Morris).
0009-2509/$ - see front matter & 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.ces.2013.01.016
et al., 2000a), high thermal stability and mechanical hardness at high temperature coupled to high redox activity (Trovarelli et al., 1999). Unique optical characteristics such as Raman-allowed modes shifting and broadening (Spanier et al., 2001), lattice expansion and a strong crystallite size related blue shift in ultraviolet absorption spectra (Tsunekawa et al., 2000b) have been reported. Applications for ceria include: as a support and promoter for exhaust-gas conversion (three-way exhaust catalyst) (Bekyarova et al., 1998), oxygen ion conductor in fuel cells (Yahiro et al., 1988) and gas sensors (Izu et al., 2002) amongst others.
V. Morris et al. / Chemical Engineering Science 91 (2013) 102–110
With such a variety of applications, different forms (e.g. particulate, mesoporous membrane/film) are needed and many methods of preparing these materials have been reported from these laboratories (Lyons et al., 2002; Chen et al., 2010). Ceria nanoparticles are of particular research interest because of the high chemical activity and the high surface area of small particulates. Note here that ceria is almost exclusively prepared as nanocrystallites and agglomerations of these form larger particles (care is needed in the use of terminology). The sensitivity of the cerium electronic band structure to the crystallite size suggests that changes in the chemical properties of ceria crystallites at nanoscale dimensions might be expected and the effects of crystallite size changes on e.g. the lattice parameter have been reported by Morris and co-workers amongst others (Chen et al., 2010). Another example of crystallite size effects, it has been noted that in oxygen sensor applications, by changing the particle size from micrometers to the nanometer region, the response time shortened from minutes to milliseconds (Ivanov et al., 2010). Further, the use of nanocrystalline CeO2 particles have allowed 200–400 1C decreases in sintering temperature when compared with micron sized CeO2 crystallites for the production of dense ceramics (Chen and Chen, 1993). It is clear from the work reported that the continued use of ceria in materials science requires both careful study and realization of rapid and scalable methods to synthesize different sizes/morphologies of the nanocrystallites whilst maintaining narrow size distributions. There are numerous chemical methods for the production of nano-dimensioned crystallites of ceria. These include sol–gel techniques (Chu et al., 1993; Makishima et al., 1986), forced hydrolysis (Dong et al., 1997), microemulsion (Masui et al., 1997) and precipitation (Hsu et al., 1988; Djuricic and Pickering, 1999; Zhou et al., 2002) to name but a few. In sol–gel processing, alkoxide or organometallic compounds are typically used as precursors which are expensive. Hydrothermal synthesis and forced hydrolysis often require severe conditions, such as higher temperatures and pressures and can involve longer reaction times. Microemulsion techniques are an efficient method for preparing highly size monodispersed CeO2 nanocrystallites. However, all of these methods are challenging and expensive to scale to commercially viable quantities. Precipitation is an attractive route due to the availability of cheap salt precursors, simple operation and easily scaleable for mass production (Zhou et al., 2002). Various simple precipitation agents have been used to generate small crystallite sizes (Li et al., 2001) and by the correct design, optimization and control of conditions, spherically shaped nanoparticles of around 13.5 nm can be prepared (e.g. Mazaheri et al., 2009). In a very recent study, Ji et al. (2012) carried out extensive studies of homogeneous precipitation methodology to generate good size monodispersed ceria crystallites of around 10 nm. In general, the use of simple precipitation methods to yield sub 10 nm particles is ineffective and, hence, the development of the methodologies described earlier using more complex micelle, forced precipitation and homogeneous precipitation methodologies. However, whilst academic synthesis methodology has become ever more complex, there has been relatively little systematic study of the simple precipitation process reported. In particular, we note that ‘‘reverse’’ precipitation of ceria has been scarcely reported. Hassanzadeh-Tabrizi et al. (2010) and Jalilpour and Fathalilou (2012) have used reverse precipitation to generate small crystallites but these have provided little insight to the mechanism and comparison of products to forward precipitation (Fleming et al., 2011; Taniguchi et al., 2011). In this paper, we analyze the products from precipitation of CeO2 nanocrystallites using the reverse precipitation technique and compare the results with those produced using forward precipitation methods and provide a rationale for the materials produced.
103
2. Experimental procedure 2.1. Synthesis of ceria crystallites Forward precipitation is the standard methodology used in large quantity production. Briefly, the base is added to the acid under stirring. However, despite the widespread use, care must be taken when precipitation reactions are carried out. As the base is added rapid pH variation occurs locally (at point of addition) due to mixing and more slowly through the course of the synthesis. The reaction might also be exothermic and heat variation may also occur. Crystallites may appear only later to be re-dissolved as stirring occurs. This precipitation process is complex and these transient changes in the solution can cause composition inhomogeneity (Hsu et al., 1988), wide size ranges (Chen and Chen, 1993) as well as other product variation. The sensitivity of product to varying pH can cause effects such as phase or composition coring when product stability is sensitive to pH (Fleming et al., 2011). Forward precipitation was investigated using the forced hydrolysis of cerium nitrate hexahydrate with ammonia hydroxide solutions (Dong et al., 1997). The resultant precipitates were then peptised to a sol–gel to provide more uniform product. Three different peptizing acids for comparison; citric acid, oxalic acid and nitric acid were examined, but citrate proved the most effective giving the smallest crystallites and a narrow size distribution and only these results are presented. A 1 M solution of cerium (III) nitrate hexahydrate, Ce(NO3)3 6 H2O was used. 100 ml of this solution was poured into a 250 ml beaker and stirred continuously at room temperature. The pH of this solution was measured at about 2.2. A 2 M solution of ammonium hydroxide solution was added at the rate of about 0.1 ml s 1 to a final pH of 12. This resulted in an immediate white or yellow emulsion which on stirring became brownishpurple. The precipitate was recovered by filtration and was vacuum dried on a Buchner funnel. Once dry, the precipitate was scraped into a crucible and dried overnight in the oven at 80 1C. On drying, the precipitate was lemon yellow in color and was ground down to a fine powder using a pestle and mortar. This sample was then suspended in water and peptized using drop-wise addition of citric acid until a clear sol–gel was formed. The resultant gel was dried as above and samples were calcined at 350 1C for 8hours and then portions of this were calcined at 50 1C intervals to vary the crystallite size. The ‘‘reverse’’ precipitation technique was identical to the forward precipitation technique, except that in this case the Ce (III) solution was added (at same rate above) to 100 ml of the ammonium hydroxide. All subsequent processing steps were as described above for the forward precipitation route. The initial and final pH values during precipitation were measured at around 13.5. The initial precipitate formed was yellow/white but then became brown/purple before finally turning green/yellow. The precipitate formed was aged, recovered and subsequently processed as detailed above. Unlike the forward precipitation where heat was needed to affect color change to a white/lemon color and remained purple until drying, the recovered solid precipitate turned bright yellow color during washing, filtering and vacuum drying. The color changes observed can be tentatively assigned to changes in oxidation state as it has been suggested that a cerium III precipitate product is a light purple color and a cerium IV precipitate product is light yellow (Taniguchi et al., 2011). Thus, it seems likely that the reverse precipitation material contains significantly less Ce3 þ states than the forward precipitation product. 2.2. Analysis To characterize the morphology of the nanocrystallites, powder x-ray diffraction (PXRD) patterns were recorded on a Phillips PW 3710 Xpert MPD diffractometer (y 2y mode), equipped with
104
V. Morris et al. / Chemical Engineering Science 91 (2013) 102–110
a Cu Ka radiation source (operating at an accelerating of 40 kV ˚ A nickel and an anode current of 30 mA) of wavelength 1.54 A. mask was used to eliminate Kb radiation and a standard scintillation detector was used. The instrument was calibrated against a ceria sample aged at 12001 for 24 h to ensure accuracy. Data were collected over the range of 10–7012y, using a step size of 0.031. Note that in order to measure small lattice parameter shifts ( o0.5%) and ensure maximum accuracy it is critical to ensure the sample surface is exactly (to within 0.05 mm) at the center of the axis of rotation of the goniometer and sample holder. This achieved by lowering the sample below the straight through beam and raising it to the point where 50% of the beam intensity is lost. The sample must also be level and this is ensured by rotation in both directions having equal intensity modulation effects. Unfortunately, depending on sample density, sample settling and sample holders this can lead to anomalous scattering from the sample holder which can result in ghost features. Where they are observed, they are marked in the figure. The measurement of lattice parameters were not made by simple analysis of peak positions, but rather by using detailed Rietveld simulation studies where the position of the features were quantitatively matched in simulation by changing lattice spacings within the crystal framework. The Phillips Analytical Rietveld 1.0b software package used also permitted quantitative analysis of each phase present. The crystallite size was calculated using the Scherrer equation, t ¼Kl/B cos y. All sizes are quoted for samples are an average result of 5 different reflections, between 25–601 2y. XPS data were collected on a vacuum generators ESCALAB (Mk II) using AlKa radiation. The pass energy was 50 eV and the binding energies were referenced to an adventitious C1s signal of 285.0 eV. Note that for XPS, as precipitated samples were recovered by filtration and immediate vacuum drying. The samples were then stored in vacuum to prevent air oxidation. The nitrogen adsorption and desorption isotherms at 77 K were measured using a Micromeritics Gemini 2375 volumetric analyzer (Micromeritics Instrument Corporation, UK). All samples were degassed under the flow of ultra high grade 5.0 nitrogen at 200 1C for 4 h prior to each measurement. Scanning electron micrographs (SEM) were collected on a JSM-5510 apparatus (JEOL) using a beam voltage of 5 kV. Transmission electron microscopy (TEM) was used for structural characterization. Each powder was dispersed onto holey carbon support grids and examined at 200 kV in a JEOL 2000FX microscope.
3. Results The precipitation reactions can be described by reference to known chemistry and the observed color changes noted above can be explained by this reaction. For both forward and reverse precipitation, reaction begins with solution of cerium nitrate to yield Ce3 þ as expressed in Eq. (1). The solvated cations react with OH- ions formed from water molecules via protonation of NH3 molecules (Eq. (2)). Eq. (3) shows reaction of the aqueous cerium cations with hydroxyl ions. The product Ce(OH)3 cerium (III) hydroxide is quickly precipitated out due to an extremely low solubility constant (Ksp ¼6.3 10 24 at 25 1C) (Zhou et al., 2003). It is clear that this species is formed in both reactions and results in the strong purple color. Chen et al. (2010) suggested that a strongly basic solution favors Ce4 þ compared to Ce3 þ via a base oxidation process and the precipitate can be oxidized as shown in Eq. (4). As shown in Eq. (5) the product Ce4 þ ions undergo strong hydration and form aqueous coordination shells. This reaction scheme is important because it has been shown that the rate of reaction of Eq. (4) is critical in determining the final crystallite size and Zhou et al. (2003) showed that the size of CeO2
crystallites can be significantly decreased by introducing oxygen bubbles into solution to react with the as-formed Ce(OH)3. This is presumably because it limits growth of the Ce(OH)3 particles due to formation of highly insoluble CeO2 type materials. 2Ce(NO3)3-2Ce3 þ þ6NO3
(1)
NH3 þH2O2NH4þ þOH
(2)
3þ
Ce
þ3OH 2Ce(OH)3(aq)
(3)
In large excess of base oxidation can occur: 2Ce(OH)3(aq) þ1/2O2-2CeO2(aq)þ3H2O y) þ -CeO2 2(H2O)þ (4 y)H3O þ Ce(H2O)x(OH)(4 y
(4) þx 2(H2O) (5)
This reaction scheme suggests that there may be significant differences in the product of forward and reverse precipitation. In forward precipitation, nucleation and growth occurs at quite low solution pH since base is being added to acid. Thus it might be expected that the reaction will produce much higher concentrations of Ce3 þ particularly at the early stage of the process. The observation of an initial white/yellow color is probably due to high local concentrations of base favoring Ce4 þ which then become Ce3 þ as the solution is stirred. As the reaction proceeds and the solution becomes more basic, it might be expected that the reaction begins to yield Ce4 þ , however, the formation of the tetravalent species might be limited because precipitation is complete. In reverse precipitation, the acid is added to base and material is precipitated in strongly alkaline conditions favoring initial formation of Ce4 þ . As the reaction proceeds, there is an increased tendency to precipitate Ce3 þ as the solution becomes less alkali. This is consistent with the observations made above where the precipitate was in initially white/yellow but became brown/purple. The reaction also explains why the precipitate changed to yellow during vacuum drying as a rapid surface reaction is enough to convert the surface trivalent cations to Ce4 þ whilst for the forward precipitation reaction extended times and temperatures are required for oxidation as the bulk is Ce3 þ . With such changes in the precipitation mechanism, the physical and chemical properties of the products should be markedly different. Data showing this is the case is described below. Fig. 1A shows the PXRD pattern for materials prepared via forward precipitation following calcination at various temperatures. The reflections observed are characteristic of the cubic fluorite Fm3m structure (JCPDS file 34-394, Wycoff, 1963) and the main reflections observed are indicated in the figure. The calculated lattice parameter after heating was measured at 0.541070.0005 nm in excellent agreement with the expected value of 0.5411 nm (JCPDS file 34-394, Wycoff, 1963). The asprecipitated material was dried at 50 1C and the XRD signals are weak. High background (as in the rising signal seen at low angle) and the rather weak intensity of the reflections suggests considerable XRD-amorphous content to the samples. Between 350 and 550 1C the peaks are broad and typical of very small crystallite sizes. It is only after heating to 850 1C, the peaks become sharp and symmetrical consistent with crystallite growth. There is no tendency to form any other phases, regardless of age, time or temperature. Fig. 1B shows typical Ce 3d XPS data from these materials. The as precipitated sample shows two main features typical of a sample containing significant amounts of Ce3 þ (Chen et al., 2010). These peaks are the Ce 3d 5/2 and Ce 3d 3/2 spinorbit doublet. Final state effects broaden these peaks into a poorly resolved quadruplet (see Chen et al., 2010 for more detail). A third high binding energy feature marked with the arrow in the figure is typical of Ce4 þ states and its present indicates a combination of Ce3 þ (majority) and Ce4 þ (minority) states in the materials. Following air calcination, the high binding energy feature
V. Morris et al. / Chemical Engineering Science 91 (2013) 102–110
105
Fig. 1. (A) shows XRD data of CeO2 prepared by forward precipitation technique calcined at various temperatures between 350 1C to 950 1C. (B) shows typical Ce 3d XPS data from these materials.
Fig. 2. (A) shows typical PXRD data of CeO2 prepared by the citric acid reverse precipitation technique calcined at temperatures between 350 1C to 1000 1C as shown. (B) shows typical Ce 3d XPS data from these materials.
increases in relative intensity and, after a 700 1C treatment, the spectrum contains a sextuplet structure typical of a material dominated by the Ce4 þ state (Chen et al., 2010). The Ce 3d spectra of Ce4 þ containing materials is highly complex because of final state effects caused by the transfer of electrons from O 2p to Ce 4 f states during the photoionization process (see Chen et al., 2010 for more detail). It would, thus, appear that the forward precipitated material contains cerium ions in largely the trivalent oxidation state but these are oxidized on treatment in air. Fig. 2A shows similar PXRD patterns for CeO2 prepared by reverse precipitation following calcination at various temperatures. As with forward precipitation, the characteristic cubic Fm3m fluorite structure is observed. As for the forward reverse as-precipitated materials the dried only sample is consistent with a material having a large XRD-amorphous content. Note by XRD amorphous we are not suggesting the particles have a completely disordered structure, simply that they do not coherently diffract
the x-rays. Between 350 and 500 1C the peaks are broad and typical of small crystallites and, significantly, the broadening of the diffraction features at low calcination temperature is much more apparent for these materials compared to those prepared via forward precipitation. Following calcination at higher temperatures, the peaks become significantly sharper and more symmetrical typical of rapid crystallite growth. As in forward precipitation, the samples show no tendency to form any other phases, regardless of age, time or temperature and there is no evidence of new diffraction peaks. In Fig. 2B, Ce 3d XPS data are described. As can be seen immediately, the as precipitated material has significantly more Ce4 þ content than that of the equivalent forward precipitated material. The sextuplet structure is not as well developed as at higher temperature suggesting it contains a small amount of Ce3 þ (Chen et al., 2010). Following air calcination, the spectra strongly resemble data expected from material which contains predominantly cerium in the 4 þ valence. The data strongly suggest that reverse precipitation produces
106
V. Morris et al. / Chemical Engineering Science 91 (2013) 102–110
Fig. 3. Comparison of XRD derived data sets. (A) and (B) represent the crystallite size variation as function of calcination temperature and the lattice parameter against crystallite size respectively for the forward precipitation route. (C) and (D) are similar data for the reverse precipitation route.
samples with very significantly less Ce3 þ and more Ce4 þ than those prepared via forward precipitation. The PXRD were quantified for accurate comparison of lattice parameter and crystallite size. The crystallite size and lattice parameter data are shown in Fig. 3 for both forward and reverse precipitation methods. As was seen in Fig. 1, the crystallites grow continually through the calcination regime (Fig. 3A) with more rapid growth beginning at temperatures greater than 650 1C. The minimum crystallite size observed is after the lowest temperature calcination and is 16.5 nm and this has increased to 38.9 nm after 650 1C calcination. The lattice parameter varies with crystallite size as revealed in Fig. 3B and shows a decreasing value with increasing crystallite size eventually reaching the value of around 5.411 A˚ expected for bulk ceria at the largest crystallite sizes. Lattice parameter expansion at low crystallite sizes has been ascribed to increased amounts of anion vacancies and accompanying Ce3 þ defects (since the trivalent cation has a larger ionic radius than that of the tetravalent cation) at the surface of nanocrystallites and the change with temperature due to the decreasing contribution of the surface as crystallite size increases (due to sintering) and/or oxidation at higher temperatures. As discussed by Chen et al. (2010) lattice parameter reduction may also be explained by the presence of tri-valent cerium hydroxyl states rather than anion vacancy production. In reverse precipitation the crystallite growth with calcination temperature follows a similar trend (Fig. 3C). However, the crystallites grow at an increased rate. After 350 1C calcination the crystallite size compared to the forward precipitation route is lower at a value of 10.5 nm. However, following 950 1C calcination, the crystallite size is higher at 57.2 nm. This is consistent with the expected increase in sintering and densification known for smaller crystallites and is due to increased crystallite– crystallite contact and lower mass transport limitations (Chen
and Chen, 1993). The change in lattice parameter with crystallite size shows the opposite trend to that of forward precipitation. At smaller crystallite sizes the lattice parameter is reduced compared to the bulk value and increases with crystallite size until the bulk value is attained. Chen et al. have argued such lattice contraction arises from surface tension effects in small particles (crystallites) (Chen et al., 2010). Chen et al. argued the precipitation route is important in understanding lattice parameter changes for CeO2 since the amount of surface hydroxyl states causing lattice expansion at small crystallite sizes can mask the inherent lattice contraction due to surface tension (Chen et al., 2010). It would, thus, appear that the reverse precipitation route is linked to lower concentrations of Ce(OH)3 type states and suggests that the majority of the product has been converted to CeO2 (2H2O) consistent with the higher pH during the precipitation process. This is supported by the XPS data provided in Figs. 1 and 2 which showed significant concentration of Ce3 þ states for only the forward precipitation materials. As reverse precipitation is the addition of the metal nitrate (acid) to a large excess of base, the lattice parameter measurements observed are consistent with the methodology. The surface area of these materials was also measured by the BET (nitrogen) technique and the results are consistent with the PXRD measurements showing higher surface areas for those samples with smaller crystallites. Typical isotherms for forward and reverse precipitation are shown in Fig. 4(A and B). All isotherms, bar some of the samples at highest calcination temperatures, can be classified as type IV (Rouquerol et al., 1994) typical of mesoporous materials. The shape of the isotherms and the measured crystallite sizes suggests inter-crystallite (i.e. voids between particles) pores are present. As can be seen in Fig. 4, the mesopore structure is essentially lost for the reverse precipitation process (but not the forward precipitation) as little hysteresis is observed. This is consistent with the suggested higher sintering
V. Morris et al. / Chemical Engineering Science 91 (2013) 102–110
107
Fig. 4. Illustrative pore volume vs. relative pressure for forward (A) and reverse (B) precipitation. Samples at various calcination temperatures shown as detailed in the figure.
rate suggested from analysis of the PXRD data. This is confirmed in the data shown in Fig. 5 which describes the quantification of the BET derived surface area data for the samples prepared via reverse precipitation compared to values calculated via the Scherrer crystallite sizes. The latter was estimated assuming an octahedrally shaped crystallite which is a reasonable assumption considering previous work (Wang, 2000; Wang and Feng, 2003) and the TEM data shown below. The surface area was calculated from the ratio of the surface area ( ¼2O3a2) and volume (1/3O2a3) using the Scherrer calculated values as a. It can be seen that the data are in reasonable agreement at low calcination temperatures but the measured surface area decreases more markedly than the calculated data at temperatures 4650 1C. The low temperature agreement (any variance can be assumed to be due to the simple geometric calculations and disregarding of the contact between crystallites) again suggests the crystallites are dense and have negligible intra-crystallite porosity (i.e. pores within the particles). It is suggested (but other explanations may be possible) that the data diverge at higher temperatures because significant crystallite–crystallite condensation is occurring as well as the inter-crystallite pores becoming closed and thus non-available to nitrogen adsorption.
Fig. 5. Surface area vs. calcination temperature for reverse precipitation. ’ are BET measured data and m are calculated from XRD derived crystallite size data assuming octahedrally shaped crystallites.
108
V. Morris et al. / Chemical Engineering Science 91 (2013) 102–110
Fig. 6 compares the variation in surface area of materials prepared via forward and reverse precipitation following different calcination treatments. The surface area of the materials prepared via forward precipitation is more complex than that of similar forward precipitation prepared samples and shows a quite clear maximum surface area at 550 1C as can be seen in Fig. 6. The data suggest some type of thermal decomposition is taking place from around 400 1C. The surface area increases from 26.8 to
Fig. 6. Surface area (via N2 adsorption isotherms) vs. calcination temperature for forward (m) and reverse (’) precipitation.
54.9 m2 g 1 following calcination at 350 and 550 1C and then decreases until it follows the same curve as that shown by the reverse precipitated materials (comparison data shown in Fig. 6). It is suggested that the different behavior derives from strong a mixture of strong hydrogen bonding and condensation of –Ce–OH groups (loss of water and formation of –Ce–O–Ce– linkages) at the points of contact between crystallites and so decreasing the surface area below that expected from the crystallite size. As the temperature increases, the Ce(OH)3 species are oxidized to CeO2 and the surface area apparently increases (as clearly shown in the XPS data in Fig. 1). Again all the data is consistent with higher concentrations of Ce(OH)3 for the forward precipitation technique. It should be noted that the decomposition of cerium hydroxyl species formed in nanocrystallites is seen around 400 1C to support the suggestion made here (Chen and Chen, 1993; Djuricic and Pickering, 1999). TEM confirm the suggestions made above and typical micrographs of samples prepared via forward and reverse precipitation following 350 1C are shown in Fig. 7. In all cases only crystalline particles were observed and there was no evidence of amorphous material or pores within the crystallites as was suggested by the BET and XRD studies. Average crystallite sizes are in reasonable agreement with XRD measurements at about 15–20 nm for forward precipitation and 8–12 nm for reverse precipitation and confirm the smaller size of crystallites formed via the reverse precipitation method. The value of 15–20 nm for the crystallite size via forward precipitation is very much consistent with the limits seen in other work and the lower values of 10 nm and below seen by reverse precipitation represent significant improvement. The crystallite shape of each preparation route is
Fig. 7. TEM images (high and low magnification) of samples following 350 1C calcination. (A) and (B) represent a sample prepared via forward precipitation and (C) and (D) via reverse precipitation.
V. Morris et al. / Chemical Engineering Science 91 (2013) 102–110
also markedly different. From forward precipitation the products are not only larger but also have more well-defined shape with two different morphologies present (Figs. 7A and B). A number of crystallites have cubic and pseudo-cubic geometry (as marked with dashed lines in Fig. 7A) whilst others have a platelet structure (see the darker image shown in Fig. 7B). Cubic CeO2 crystallites are well known and are thought to arise via Oswald Ripening of more spherical shapes precipitated in the early stages of growth (Mai et al., 2005). In the case of reverse precipitation, the crystallites are noticeably smaller and the size dispersion much less (Fig. 7C). The crystallites now appear to adopt largely one shape have a more dimensionally uniform shape appearing almost spherical in nature. Closer examination reveals them to be polyhedral with relatively well-defined crystal facets as indicated in Fig. 7D and are similar to shapes seen in previous work (Wang, 2000; Wang and Feng, 2003) and can be best described as truncated octahedra (Zhang et al., 2007). The smaller size of the materials prepared via reverse precipitation can be explained by increased oxidation rate (Wang, 2000) and/or kinetic and thermodynamic limitations where good evidence has been provided that the truncated octahedra morphology is the precursor to platelet and cubic morphologies due to orientated attachment along the well-defined crystallite faces (Niesz et al., 2010).
4. Discussion Forward and reverse precipitation gives markedly different products. Pourbaix diagrams show that Ce3 þ is stable in solution at normal pH values and that a minimum pH value of 4–6 must be reached before formation of Ce(OH)3 occurs (O’Keefe et al., 2004). However, the Ce(OH)3 formed has some solubility to a pH of about 10 when precipitation of the hydroxide is rapid. As the pH is raised further, the formation of Ce4 þ (Ce(OH)4 or CeO2 2H2O) products becomes more likely, consistent with the reaction model given above in Eqs. (1)–(5). It thus seems that the reverse precipitation, where the pH through the process is uniformly high, is likely to give a CeO2 type product whilst the forward precipitation process is more likely to give Ce3 þ precipitates particularly during the early stages of the process where the pH changes from acidic through to alkaline. This suggestion is apparently borne out by both the color changes observed during synthesis and the lattice dimensions measured by XRD as well as previous XPS measurements (Chen et al., 2010). What is also important is that the products of the precipitation are smaller and have lower size dispersions. This is also partly explained by the Pourbaix diagrams as the entire precipitation process occurs at a pH value where solubility is limited and thus nucleation, re-solution and the size related thermodynamic stability (Dirksen and Ring, 1991) of the crystallites has a much lesser effect and the statistical nature of the crystallite growth process is limited. Zhou et al. offer a detailed rationale for the size homogeneity of the product (Zhou et al., 2002). These authors point out that the solutions with high pH are highly supersaturated and this provides conditions for homogeneous nucleation. It should also be noted that growth of large crystallites is reliant on two processes; the normal Ostwald Ripening and also orientated attachment (Zhou et al., 2003). Orientated attachment is enhanced by surface hydroxyl groups (Ce(OH)3) and so will tend to form larger crystallites. It is also interesting to note that the orientated attachment mechanism promotes formation of cubic (Zhou et al., 2003) and linear (Ivanov et al., 2010) crystallite morphologies. It can, therefore, be reasonably asserted that this is the cause of the cubic and platelet structures seen for the forward precipitation route (Fig. 7) and supports the higher Ce(OH)3 content suggested by experimental evidence presented here for
109
this preparation route. It might be thought that the oriented attachment is limited for CeO2 type products because of the lower surface hydroxyl densities and thus the truncated octahedral shape is maintained. From this work, it is suggested that reverse precipitation, which is often ignored as a synthesis method, can provide a useful methodology to produce small crystallites producing a more size homogenous product and smaller crystallite sizes/ higher surface areas. However, the products can have markedly different crystallite shapes and surface chemistries. They are likely to be more ‘passive’ to crystallite–crystallite attachment and also contain lower defect (including structural defects such as grain boundaries, pores and point defects such as anion vacancies/Ce3 þ centers) densities as they consist of single crystallites following homogeneous precipitation. It should be recognized that this methodology of reverse precipitation should not be considered as a technique to rival current state-of-the-art laboratory work but rather to enhance large-scale precipitation methods. Often, in small scale synthesis, researchers have used surfactant, ligand attachment, forced synthesis etc., to produce smaller, size monodisperse and non-aggregated nanoparticles. Whilst these techniques can produce almost ideal materials, the cost of the chemicals and processing might prevent scale-up to large quantities. The reverse precipitation technique described here could be used in simple batch or continuous flow apparatus at relatively low costs and provide a robust and reproducible method for producing small particles.
Acknowledgments We wish to thank Dr. Simon Newcomb of Glebe Scientific for TEM data. We would also like to thank SFI through the FORME SRC Grant for research support. References Bekyarova, E., Fornasiero, P., Kaspar, J., et al., 1998. CO oxidation on Pd/CeO2–ZrO2. Catal. Today 45, 179–183. Chu, X., Chung, W.I., Schmidt, L.D., 1993. Sintering of sol–gel prepared submicrometer particles by transmission electron microscopy. J. Am. Ceram. Soc. 76, 2115–2118. Chen, L., Fleming, P., Morris, V., et al., 2010. Size-related lattice parameter changes and surface defects in ceria nanocrystals. J. Phys. Chem. C 114, 12909–12919. Chen, Pei-Lin, Chen, I-Wei, 1993. Reactive cerium(IV) oxide powders by the homogeneous precipitation method. J. Am. Ceram. Soc. 76, 1577–1583. Dirksen, J.A., Ring, T.A., 1991. Fundamental of crystallization: kinetic effects on particle size distributions and morphology. Chem. Eng. Sci. 46, 2389. Djuricic, B., Pickering, S., 1999. Nanostructured cerium oxide: preparation and properties of weakly agglomerated powders. J. Eur. Ceram. Soc. 19, 1925–1934. Dong, X., Hong, G., Yu, D., et al., 1997. Synthesis and properties of cerium oxide nanometer powders by pyrolysis of amorphous citrate. J. Mater. Sci. Technol. 13, 113–116. Fleming, P.G., Holmes, J.D., Otway, D.J., et al., 2011. The complex synthesis and solid state chemistry of ceria-lanthana solid solutions prepared via a hexamethylenetetramine precipitation. J. Solid State Chem. 184, 2595–2600. Hassanzadeh-Tabrizi, S.A., Mazahere, M., Aminzare, M., et al., 2010. Reverse precipitation and characterization of CeO2 nanopowder. J. Alloys Compd. 491, 499–502. Hsu, W.P., Ronnquist, L., Matijevic, E., 1988. Preparation and properties of monodispersed colloidal particles of lanthanide compounds. 2. Cerium (IV). Langmuir 4, 31–37. Ivanov, V.K., Polezhaeva, O.S., Tret’yakov, D., 2010. Nanocrystalline ceria: synthesis, structure-sensitive properties, and promising applications. Russ. J. Gen. Chem. 80, 604–617. Izu, N., Shin, W., Murayama, N., et al., 2002. Resistive oxygen gas sensors based on CeO2 fine powder prepared using mist pyrolysis. Sens. Actuators B-Chem. 87, 95–98. Jalilpour, M., Fathalilou, M., 2012. Int. J. Phys. Sci. 7, 944–958. Ji, H., Chen, Y., Chen, S., 2012. Synthesis and characterization of nanocrystalline ceria powder through homogeneous precipitation method. Adv. Mater. Res., Adv. Mater. Process. II 557–550, 567–571. Li, J.G., Ikegamai, T., Wang, Y., Mori, T., 2001. Reactive ceria nanopowders via carbonate precipitation. J. Am. Ceram. Soc. 85, 2376–2378.
110
V. Morris et al. / Chemical Engineering Science 91 (2013) 102–110
Lyons, D.M., Ryan, K.M., Morris, M.A., 2002. Preparation of an ordered mesoporous phase of ceria with enhanced thermal stability. J. Mater. Chem. 12, 1207–1212. Mai, H.-X., Sun, L.-D., Zhang, Y.-W., et al., 2005. Shape selective synthesis and oxygen storage behavior of ceria nanopolyhedra, nanorods and nanocubes. J. Phys. Chem. B 109, 24380–24385. Makishima, A., Kubo, H., Wada, Y., Shimohira, T., 1986. Yellow coatings produced on glasses and aluminium by the sol-gel process. J. Am. Ceram. Soc. 69, c127–129. Masui, T., Fujiwara, K., Machida, K.I., et al., 1997. Characterization of Ce (IV) oxide ultrafine particles prepared using reversed micelles. Chem. Mater. 9, 2197–2204. Mazaheri, S.A., Hassanzadeh-Tabrizi M., et al., 2009. Synthesis of CeO2 nanocrystalline powder by precipitation method. In: Proceedings of the 11th ECERS Conference, Krakow. Niesz, K., Reji, C., Nielson, J.R., et al., 2010. Unusual evolution of ceria nanocrystal morphologies promoted by a low-temperature vapor diffusion based process. Cryst. Growth Des. 10, 4485–4490. O’Keefe, T., Yu, P., Hayes, S., et al., 2004. Fundamental evaluation of the deposition of cerium oxide for conversion coating applications. Office of the Department of Defense, DOD Corrosion Policy and Oversight, Washington, DC. Rouquerol, J., Avnir, D., Fairbridge, C.W., et al., 1994. Recommendations for the characterization of porous solids. Pure Appl. Chem. 66, 1739–1758. Spanier, J., Robinson, R., Zhang, F., et al., 2001. Size dependent properties of CeO2-y as studied by Raman Spectroscopy. Phys. Rev. B 64 (245407), 1–8.
Taniguchi, T., Katsumata, K.-I., Omata, S., et al., 2011. Tuning growth modes of ceria-based nanocubes by a hydrothermal method. Cryst. Growth Des. 11, 3754–3760. Trovarelli, A., de Leitenburg, C., Boaro, M., et al., 1999. The utilization of ceria in industrial catalysis. Catal. Today 50, 353–367. Tsunekawa, S., Fukuda, T., Kasuya, A., 2000a. Blue shift in ultraviolet absorption spectra of monodisperse CeO2 x nanoparticles. J. Appl. Phys. 87, 1318–1321. Tsunekawa, S., Sahara, R., Kawazoe, Y., et al., 2000b. Origin of the blue shift in ultraviolet absorption spectra of nanocrystalline CeO2 x particles. Mater. Trans. 41, 1104–1107. Wang, Z.L., 2000. Transmission electron microscopy of shape-controlled nanocrystals and their assemblies. J. Phys. Chem. B 104, 1153. Wang, Z.L., Feng, X., 2003. Polyhedral shapes of CeO2 nanoparticles. J. Phys. Chem. B 107, 13563–13566. Wycoff, R.W.G., 1963. 2nd Ed.Crystal Structures, vol. 1. Interscience Publishers, New York, UK. Yahiro, H., Baba, Y., Eguchi, K., et al., 1988. High temperature fuel cell with ceria– yttria solid electrolyte. J. Electrochem. Soc. 135, 2077–2088. Zhang, J., Ohara, S., Umetsu, M., et al., 2007. Colloidal ceria nanocrystals: a tailormade crystal morphology in supercritical water. Adv. Mater. 19, 203–206. Zhou, X.D., Huebner, W., Anderson, H.U., 2002. Nucleation synthesis and thermal stability of nanometer CeO2 single crystals. Appl. Phys Lett. 80, 3814. Zhou, X.D., Huebner, W., Anderson, H.U., 2003. Processing of nanometer-scale CeO2 particles. Chem. Mater. 15, 378–382.