- Email: [email protected]
3−xTiO3 nanoparticles with different morphologies and self-organization, obtained from simple solution precipitation methods
Materials Letters 137 (2014) 182–187
Contents lists available at ScienceDirect
Materials Letters journal homepage: www.elsevier.com/locate/matlet
Li3xLa2/3 xTiO3 nanoparticles with different morphologies and self-organization, obtained from simple solution precipitation methods K.V. Kravchyk a,b,n,1, G. Brotons a,1, A.G. Belous b, O. Bohnke a a b
Université du Maine, Institut des Matériaux et des Molécules du Mans (UMR 6283 CNRS), Av O. Messiaen, 72085 Le Mans Cedex 9, France V.I. Vernadskii Institute of General and Inorganic Chemistry, 32/34 Palladina Avenue, 03680, Kyiv-142, Ukraine
art ic l e i nf o
a b s t r a c t
Article history: Received 27 May 2014 Accepted 30 August 2014 Available online 6 September 2014
We present a novel synthesis strategy of Li3xLa2/3 xTiO3 (LLTO) softly agglomerated nanoparticles by simple precipitation from solution and based on controlling the fractal self-organization. In this way we avoid the use of bulky organic ligands frequently used to avoid the irreversible aggregation of nanoparticles but that also often degrade the LLTO functional properties. We show that the pH of precipitation of titanium and lanthanum hydroxides (LLTO precursors) strongly affects the so-called “filtration coefficient” and the hierarchic structure of LLTO precursors (mass or surface fractal self-organization). For mass-fractal aggregation of the precursors, a high filtration coefficient was obtained with soft and readily friable aggregates of the hydroxide precipitates. Moreover, the fractal self-organization of the synthesized precursors strongly modifies the final LLTO nanoparticles morphology and the material connectivity. The LLTO nanoparticles synthesized from mass-fractal agglomerated precursors obtained from the so-called sequential precipitation of hydroxides (SPH, prepared at pHTiðOHÞ4 ¼3.5 and pHLaðOHÞ3 ¼8.5) had a much lower agglomeration ability than the surface-fractal agglomerated precursors obtained from the so-called co-precipitation of hydroxides (CPH, at pHTiðOHÞ4 ¼ pHLaðOHÞ3 ¼ 8.5). & 2014 Elsevier B.V. All rights reserved.
Keywords: Self-organisation Fractal structure Li3xLa2/3 xTiO3 (LLTO) Precipitation from solution
1. Introduction The search of new materials to improve electrochemical systems, particularly for the transformation and the conservation of electric energy and for chemical sensors, is a strong catalyst for research. The electrolyte is one of the important parts of these electrochemical systems and the improvement of the conductivity of solid electrolytes remains, at the time, a great challenge in research. Inorganic Li ionconducting solid conductors always attract great interest since they may present several advantages, i.e. high electrochemical stability window, high thermal stability, high mechanical resistance and the possibility of miniaturization by preparing thin or thick films. In such a context, Li-conducting ceramics based on Li3xLa2/3 xTiO3 (named hereafter LLTO) hold an important place among them because of their high ionic conductivity at room temperature, i.e. σ¼10 3 S cm 1 for x¼0.10 [1,2]. However, besides this high bulk conductivity, the grainboundary resistance of these ceramics remains high and limits considerably the applications of these polycrystalline ceramics. The n Corresponding author. Current address: ETH Zurich, Laboratory of Inorganic Chemistry, Wolfgang-Pauli-Str. 10, 8093 Zurich, Switzerland. Tel.: þ41 58 765 5944. E-mail address: [email protected] (K.V. Kravchyk). 1 These authors contributed equally to this work.
http://dx.doi.org/10.1016/j.matlet.2014.08.153 0167-577X/& 2014 Elsevier B.V. All rights reserved.
electrical properties of the grain boundaries differ essentially from the bulk because of the presence of strained or missing bonds in the interface region that may lead to a change of the mobile species mobility. Structural or space charge effects in this interfacial region have been suggested to explain these variations [3–5]. Further, based on the grain-boundary core-space-charge layer model, it has been shown that grain boundary resistance can change depending on the grain size. These suggestions were experimentally confirmed on stabilized zirconium oxide nanomaterials, where it has been shown that specific grain boundary conductivity can be increased by obtaining nanodimensional grains [6,7]. In such a context, the attempt to synthesize nanosized LLTO ceramics appeared to be of great interest. For obtaining nanosized ceramics of oxides, it is necessary not only to prepare nanoparticles of precursors but also to control the process of their aggregation during the synthesis. In order to avoid the irreversible agglomeration of inorganic nanoparticles, bulky organic surfactants can be used. By this way, it is also often possible to control their size and shape. However, drawbacks resulting from surface adsorbed surfactants are the unpredictable influence on the toxicity of nanoparticles, and the diminished accessibility to the particle surface [8,9]. The latter can be serious issue regarding grain boundary ionic transport even after high-temperature heat-treatment.
K.V. Kravchyk et al. / Materials Letters 137 (2014) 182–187
Another way to control the formation of nanoparticles and their agglomeration is to grow self-organized fractal structures forming during the synthesis. Thus, it has been shown that nanoparticles have the ability to self-organize in networks, with formation of fractal structures. By changing the synthesis conditions, control of the type of fractal self-organization is possible, resulting in the formation of either soft or hard agglomerated nanoparticles. Thus, in recent papers, Belous et al. [10,11] have shown that, by using precipitation from solution as synthesis method, by controlling the pH of the solution and by changing the sequence of hydroxides precipitation, it was possible to obtain different types of fractal self-organization of nanoparticles of zirconium and yttrium hydroxides (bulk and surface). This fractal self-organization of the precursor powder affects the morphology of the final oxide powder and the properties of the ceramics (sintering temperature, size of grains). In this work, we investigated and compared self-organization of the precursor and final LLTO nanoparticles synthesized by two routes: the sequential precipitation (SPH) and the coprecipitation (CPH) synthesis. The LLTO precursors used are a mixture of titanium and lanthanum hydroxides to which a lithium hydroxide solution is added after hydroxide precipitation in an adequate quantity. The aim of this paper is to investigate the self-organization of the LLTO precursor nanoparticles obtained and its influence on final morphology. We first investigated separately, the precipitation conditions of titanium and lanthanum hydroxides (pH and concentration of reduction agent). As is known, these factors can efficiently affect the dispersion (or self-organization) of the precipitates and their agglomeration. Afterwards, we investigated the influence of the sequence of precipitation of the two hydroxides (sequential or co-precipitation) on the fractal structure and on the microstructure of the precursor and final LLTO nanoparticles.
2. Experimental procedure Synthesis and characterization of La(OH)3 and Ti(OH)4 hydroxides: La(OH)3 was precipitated from an aqueous solution of La(NO3)3 by aqueous solution of NH4OH to adjust the pH from 8 to 11. Ti(OH)4 was precipitated from an isopropanol solution of TiCl4 by aqueous solution of NH4OH to adjust the pH from 2 to 8.5. The concentration of NH4OH used for adjusting the pH was varied (1 mol%, 5% or 25% NH3) since it can influence the morphology of the precipitates. The so-obtained precipitates of La and Ti hydroxides were filtered and washed from the mother solution with distilled water until Cl and NO3 ions disappear from the washing solutions. Afterwards, hydroxide precipitates were washed additionally with 500 ml of distilled water in order to determine filtration coefficient Kf. Filtration coefficient, Kf, of each fresh precipitate was calculated using Darsi formula [12]: Q ¼ Kf S
H t L
ð1Þ
where Q—filtrate volume, cm3; Kf—filtration coefficient, cm s 1; S—precipitate surface on the filter, cm2; H—pressure under filter, cm of Hg; L—thickness of precipitate layer, cm; t—time of filtration, s. Precipitates were then dried at 80 1C. The compactness of the hydroxide pellets (g cm 3) was determined from the mass, diameter and height of pellets after sintering. Pellets were obtained by uniaxial compression at 16 MPa (kg cm 2) of hydroxide powders, followed by a heat treatment at 600 1C. Infra Red (IR) spectroscopy and thermal analyses (DTA, TGA) were used to characterize the La and Ti hydroxides. IR spectra were recorded on a Specord-M80 spectrometer in the 250–4000 cm 1 frequency range. Samples consisted of pellets prepared by pressing
183
the hydroxide with dehydrated KBr (1 wt%). DTA and TGA experiments were carried out with a Setaram TGDTA92 equipment, at a heating rate of 5 1C min 1, in air using Pt crucible. Preparation of the precursors of LLTO: The synthesis of the precursors of LLTO has been carried out either by sequential precipitation of the hydroxides (SPH) or by their coprecipitation (CPH). The mixture of titanium and lanthanum hydroxides obtained by SPH was carried out in the following way: the titanium hydroxide was first precipitated at pH ¼3.5 and afterwards lanthanum hydroxide was precipitated at pH ¼ 8.5. In the case of CPH method, lanthanum and titanium hydroxides were precipitated simultaneously at pH ¼8.5. After obtaining the mixture of these two hydroxides, the precipitates were kept for 2 h in solution to control the loss of metal cations during washing procedure. Then, hydroxide mixtures washed with distilled water and an appropriated amount of an aqueous solution of LiOH was added in order to obtain the chemical composition Li0.3La0.57TiO3. The obtained suspension was dried afterwards in an oven at 80 1C. The microstructure of these precursors was studied by small angle X-ray scattering (SAXS). Measurements were performed in transmission mode trough vacuum with sample pellets of known density and thicknesses obtained by uniaxial pressure. The instrument source is a high flux Rigaku™ copper rotating anode and the beam is set by three pinholes after an Osmic™ confocal mirror that delivered a monochromatic (λ 1.54 Å) and cylindrical X-ray beam of 2 107 photons s 1 over 350 μm in diameter with an angular divergence below 0.5 mrad. To obtain a wide range of wave vectors (0.015 Å 1 to 2 Å 1) we combined a 2D gas detector and a “flat image plate” for larger angles, with a broad overlap from 0.1 to 0.3 Å 1. The absolute q-range was calibrated from the known diffraction lines of Ag-Behenate powder and intensities were given in absolute scale units calculated from the water scattering level (intensities in cm 1, i.e. cm² per cm3 of sample material). Morphological observations were performed using Transmission Electronic Micrsocopy (TEM). Thin specimens were obtained by ultrasonically dispersing particles in ethanol and depositing one droplet of the resulting suspension on a Cu grid covered with a holey carbon film. After drying, the grid was fixed in a side-entry 301 double-tilt specimen holder and was introduced in a JEOL2010 electron microscope operating at 200 kV. Synthesis of LLTO powder: LLTO has been synthesized from these precursors by heat treatment at 900 1C in air for 2 h. Powder X-ray diffraction (XRD) patterns have been recorded at room temperature with a Philips X’Pert PRO diffractometer (Cu Kα radiation), equipped with a linear X’Cellerator detector, in the 2Θ range from 5 to 701 with an interpolated step of 0.081. TEM has also been carried out to determine the microstructure of the final oxide. The lithium content in the LLTO powders was determined using chemical analysis. K2S2O7 was added in excess to LLTO powders and alloyed during 24 h on a sand bath. The alloy was then dissolved in acidified water. The analysis of the solutions was carried out with an atomic absorption spectrometer (SP-9 PueUnicom) showing a lithium content below 2%. To overcome this substantial loss of lithium after the heat treatment of LLTO precursor, we thus used an excess of LiOH for the LLTO synthesis.
3. Results and discussion Characterization of individual precipitates of La(OH)3 and Ti(OH)4: Previous investigations performed during the precipitation of zirconium and yttrium oxides have shown that the pH of the solution affects efficiently the filtration coefficient, Kf, of the precipitates. This coefficient is correlated to the fractal structure of the precipitates and to the strength of the bounds formed between particles during precipitation. The higher the filtration
184
K.V. Kravchyk et al. / Materials Letters 137 (2014) 182–187
coefficient is, the softer the agglomerated powder and the weaker the bounds between particles are [10]. According to these results, we determined the value of Kf for the individual lanthanum and titanium hydroxide precipitates. Fig. 1a shows the values of the filtration coefficient obtained for lanthanum and titanium hydroxides as a function of the pH of the solution (the concentration of NH4OH was fixed at 5 mol% NH3). Maxima are obtained at pH ¼3.5 and 8.5 for Ti(OH)4 and La(OH)3 precipitates respectively. These pH values indicate the experimental conditions of precipitation to obtain a soft precipitate with weak bounds between particles. Out of these values the obtained precipitates have a low filtration coefficient and therefore have strong bounds. Fig. 1a also displays the values of the compactness of the pellets obtained after drying, pressing and annealing in air at 600 1C. It can be observed that the compactness follows the Kf coefficient. The highest compactness is obtained for powders precipitated
Fig. 1. (a) Filtration coefficients of titanium and lanthanum hydroxides (left axis) and compactnesses of annealed titanium and lanthanum oxides at T¼ 600 1C (right axis), as a function of precipitation pH; (b) filtration coefficients of titanium and lanthanum hydroxides as a function of concentration of added ammonia solution; (c) IR spectra of titanium and lanthanum hydroxides obtained at different pH of precipitation.
at the pH that shows the highest filtration coefficient suggesting that the nature of the bounds between particles influences the compactness of the annealed pellet. The pellets made from weak bounded particles, are more compact. It is also important to note that the filtration coefficient data of Ti(OH)4 and La(OH)3 precipitates show a good correlation with the mechanical properties of the annealed pellets. The pellets obtained at pH ¼3.5 and pH ¼8.5 for Ti(OH)4 and La(OH)3 respectively were soft, suggesting that weak bounds exist between particles. On the other hand, the annealed pellets obtained from the hydroxides precipitated at higher or lower pH were very difficult to grind, suggesting that the particles are more strongly bounded. We also investigated the influence of the ammonia concentration on Kf. Fig. 1b presents the values of the filtration coefficient as a function of ammonia solution concentration for precipitation performed at pH ¼3.5 and 8.5 for Ti(OH)4 and La(OH)3 respectively. The concentration of 1 mol% NH3 is the best one to obtain La (OH)3 precipitate with the highest filtration coefficient and a concentration of 5 mol% NH3 for Ti(OH)4 precipitates. These results clearly show that the pH and the ammonia concentration of the solution greatly influence the morphology of the precipitates. IR-spectra of individual titanium and lanthanum hydroxides, precipitated at different pH, are presented in Fig. 1c. The absorption bands observed on the spectra are related to the presence of OH-groups, adsorbed H2O molecules and to the presence of carbonates and adsorbed CO2 molecules. The absorption bands in the interval 3700–2700 cm 1 correspond to the stretching vibrations of OH-groups. The broadening and the shift of the absorption bands, typical for stretching vibrations of isolated OH-groups (3750– 3500 cm 1), indicate the presence of hydrogen bonds in individual lanthanum and titanium hydroxides. Bands in the frequency range from 1600 to 1685 cm 1 correspond to deformation vibrations of H2O molecules and may be attributed to adsorbed water. Absorption bands in the frequency range from 1330 to 1430 cm 1 can be attributed to deformation vibrations of OH-groups. All these absorption bands are observed on both lanthanum and titanium hydroxides. Bands around 2275 to 2400 cm 1 correspond to deformation vibrations of molecules of carbon dioxide. They are slightly observed on Ti(OH)4 precipitated at pH¼3.5, they are more intense on La(OH)3 precipitated at pH¼8.5. They can be attributed to CO2 adsorption. The presence of absorption bands in IR-spectra of lanthanum hydroxide at 1050 and 850 cm 1 indicates the presence of carbonate groups, namely the peak at 1050 cm 1 can be assigned to stretching vibration of CO and the peak at 850 cm 1 to deformation vibration of CO. To resume the above obtained results, on titanium hydroxide IR spectra revealed the presence of OH groups, adsorbed H2O molecules and adsorbed CO2 molecules. These last molecules are in a very small amount and mostly on the hydroxide prepared at pH¼3.5. On the lanthanum hydroxide the presence of OH-groups and adsorbed H2O molecules is observed but also the presence of carbonate species and of CO2 molecules in a greater amount than in titanium hydroxide (bands at 2275–2400 cm 1) and especially for the hydroxide prepared at pH¼ 8.5. The intensity of absorption bands correlates clearly with the dependence of both, the filtration coefficient and the density of precipitates, on pH. Samples corresponding to the highest Kf and to the highest compactness are characterized by deeper absorption bands of carbon dioxide suggesting that these samples have the highest surface of adsorption for CO2. Further, the presence of carbonates on lanthanum hydroxide is clearly related to the more basic character of lanthanum compared to titanium. Fig. 2 shows the curves of differential thermal analysis (DTA) and thermal gravimetry (TG) of dried samples of titanium and lanthanum hydroxides precipitated at different pH. The weight loss of titanium hydroxides is characterized by a constant rate up to 250 1C, that is
K.V. Kravchyk et al. / Materials Letters 137 (2014) 182–187
185
decomposition temperature of lanthanum oxycarbonate La2O2(CO3). Indeed, the higher is pH for the precipitation of lanthanum hydroxide, the higher is the decomposition temperature of lanthanum oxycarbonate La2O2(CO3). From the comparison of TG curves of lanthanum hydroxides it follows that lanthanum hydroxides precipitated at lower pH are characterized by a higher weight loss compared to precipitates obtained at higher pH. Characterization of LLTO precursor precipitates obtained either by SPH or CPH methods: The precipitate of LLTO precursors has been obtained either by SPH or CPH, as described in the experimental part of the paper. Fig. 3a and e show SAXS curves of SPH and CPH precipitates respectively. The two SAXS curves are different but where analyzed with the same model. We used the Beaucage’s unified exponential/power-law scattering function [13–15]. The equation describing an arbitrary number of interrelated structural levels has the form n
I ðqÞ ¼ ∑
i¼1
Fig. 2. Thermal analyses plots of titanium and lanthanum hydroxides obtained at different pH of precipitation.
related to the loss of adsorbed water molecules at lowest temperatures and of structural water molecules around 200 1C. The loss of adsorbed water is confirmed by the presence of an endothermic peak around 100 1C and the loss of structural water is confirmed by the presence of an exothermic peak around 220 1C on DTA curves that reveals more clearly at higher pH of precipitation (pH46). The exothermic peak at 430 1C corresponds to the crystallization of the anatase phase of titanium oxide. Above 500 1C, rutile phase begins to appear. It allows supposing that the transition from anatase phase to rutile phase takes place between 500 and 1000 1C. The comparison of DTA curves of titanium hydroxides obtained at different pH allows asserting that the pH of precipitation affects both the temperature interval of transition from titanium hydroxide to oxide and the crystallization temperature of the anatase phase of titanium oxide. Besides, it is important to note, that the exothermic peaks, which correspond to the dehydration of titanium hydroxide and the crystallization of anatase phase become more diffused at low pH. This might be connected with simultaneous processes of loss of adsorbed water, dehydration and crystallization. The comparison of weight loss of titanium hydroxides precipitated at different pH indicates that titanium hydroxide precipitate obtained at pH¼ 3.5 has the lowest amount of adsorbed water and hydroxyl groups. The analysis of TG and DTA curves of lanthanum hydroxide allows one to suggest that when increasing the temperature, the following processes take place: weight loss of adsorbed water around 100 1C, partial dehydration of lanthanum hydroxide to La2O2(OH)2 phase and interaction of the latter with carbon dioxide to form lanthanum oxycarbonate La2O2(CO3) corresponding to the endothermic peak at 325 1C. The formation of this oxycarbonate phase La2O2(CO3) was displayed on IR-spectra of lanthanum hydroxides at 1500, 1050 and 850 cm 1, which indicate the presence of carbonate groups. In the temperature interval 450–550 1C, partial decomposition of lanthanum oxycarbonate takes place, what is indicated by the endothermic peak and sharp weight loss on TG curve. Endothermic peak at 650–750 1C is connected obviously with full decomposition of lanthanum oxycarbonate to form lanthanum oxide. Comparison of the DTA and TG curves of lanthanum hydroxides obtained at different pH shows that at increasing pH of precipitation the behavior of curves changes partially. Namely, at pH¼10 the processes of decomposition of lanthanum oxycarbonate do not take place until 650–750 1C. It reflects in absence of endothermic DTA peak at 500 1C and almost lacking decreasing of weight loss on TG curves at 450–650 1C comparing to lower pH values. It means that a change of the precipitation pH of lanthanum hydroxide affects the
pffiffiffi3 P i Gi exp q2 R2gi =3 þ Bi exp q2 R2gði þ 1Þ =3 erf qRgi = 6 =q
!
ð2Þ where Gi is the Guinier prefactor for level i, Bi is the Porod prefactor (in the power-law dependence of the logarithm of the scattered intensity on the logarithm of the scattering vector). Accordingly, this relationship continuously describes the small angle Guinier regime due to scattering from large scale domains (fractal aggregates at level i) with radius of gyration Rgi and the Porod regime due to scattering from these domain interfaces. At first level 1, we expect smaller nanoparticles (building blocks of the fractals). The exponent P i , defining the fractal dimension of the domains at level i, is responsible for the q Pi asymptotical regime. Depending on the values of Pi, we distinguished two different fractal types: (1) a mass fractal system with 1 o P o 3 with a massfractal dimension Dm ¼P. The domains are tenuous for smaller values of Dm and denser for larger values. (2) A surface fractal system with 3 o P o 4 and a surface-fractal dimension Ds ¼ 6 P (while Dm ¼ 3). A perfectly flat interface corresponds to Ds ¼2 and the Ds value increases with roughness. Both, mass and surface fractals are common types of domains found in dispersive systems made of small clusters and nanoparticles [16]. The SAXS slope is then a strong indication of the type of aggregation and dimension of fractals. The SPH curve is characterized by a linear region with a slope of 2.99, when plotted in log–log scale up to q¼ 0.27 Å 1. At larger angles, intensity reaches a constant level due to the background and the broad tail of a diffraction peak (Fig. 3a). We first fitted the data successfully using only one level of structuration in (equation 2). For this, we obtained large fractal domains (Rg1 75– 85 nm) with a Guinier plateau that appeared at angles that are smaller than our measurement window (i.e. the G1 term could be neglected). Thus, the second term B1 fitted the data up to 0.27 Å 1 and corresponds to the surface scattering from a dense mass fractals (P1 2.99) made of very small building blocks (Rg2 o1.5 nm). By introducing a second level of structuration in (equation 2), we could fit the data with parameters corresponding to large mass-fractal domains (Rg1 75–85 nm), dense (P1 2.75) and made of small nanoparticles with a diameter of d1 9 nm (Rg2 3.5 nm and for a plain particle d1 2.58 Rg2). The sum of three terms reproduced the constant slope in log–log scale: the B1 term fitted the first part up to q¼ 0.05 Å 1; for q40.1 Å 1 the strongest contribution came from B2 and the G2 terms dominated the intermediate range scattering. The CPH curve (Fig. 3e) shows three regimes with less intensity at smaller angles than CPH. For 0.01 oq o0.07 Å 1 a slope of 0.77 was measured, 3.13 for 0.07 oq o0.3 and a slope of 0.26 at larger angles before reaching background. Using two structural levels, the data are well fitted from the Guinier and Porod scattering terms (G2 dominates for q o0.06 Å 1 and then B2 dominates up to 0.3 Å 1) from
186
K.V. Kravchyk et al. / Materials Letters 137 (2014) 182–187
Fig. 3. Small-angle X-ray scattering curves of SPH (a) and CPH (e) samples dried at 80 1C; I(q), with absolute intensity (cm 1) and q wave vector (Å 1) in log–log scale (a scheme of the fractal structures is given on top of the figures); microphotographs of SPH and CPH samples dried at T ¼ 80 1C (b) and (f) and annealed at T ¼900 1C for 2 h (c), (d) and (g), (h).
nanoparticles (Rg1 5.5 nm) that aggregated in the form of a corrugated surface fractal with Rg1 30 nm and P1 3.1. These large aggregates agglomerated at larger scales without visible fractal structure. This latter behavior is shown by the saturation of intensity for q o0.07 in comparison to the SPH sample. It is worth noting that all SAXS data are in good agreement with the TEM results. Fig. 3b, f present the micrographs of LLTO precursors obtained by SPH and CPH methods. As it appears in Fig. 3b and f both precipitates have different mesoscopic organization. Fig. 3b clearly evidences the formation of mass fractal with a branched structure. Fig. 3f shows the formation of dense precipitates of 300–500 nm size. This correlates with the agglomeration of surface fractal structure. By considering the difference in the self-organization of the particles in SPH and CPH precipitates and their conditions of preparation (pHTiðOHÞ4 ¼3.5 and pHLaðOHÞ3 ¼8.5 for SPH; pHTiðOHÞ4 ¼ pHLaðOHÞ3 ¼8.5 for CPH), it is easy to understand why the filtration coefficients of titanium and lanthanum hydroxides, obtained at pH ¼ 3.5 and 8.5 respectively, are higher than the ones obtained at other pH (Fig. 1a). During CPH precipitation both hydroxides are precipitated at pH ¼8.5. At this value of pH, compact surface aggregates of Ti(OH)4 are formed with strong bounds between particles. Those compact surface aggregates form compact particles of big size with non-fractal behavior. Simultaneously, La(OH)3 precipitates but the morphology of the final product is dictated by the Ti hydroxide one which is in a greater amount. On the other hand, during SPH precipitation Ti hydroxide is formed first at pH ¼ 3.5. In such a condition the primary particles aggregate and form a mass fractal with a branched structure. Afterwards, La hydroxide is precipitated at pH ¼ 8.5 adopting a mass fractal morphology and an open structure too. Consequently, water will flow easily through such precipitates in contrast to the compact precipitates with surface fractal structure formed by CPH method. The precipitates obtained by SPH method are characterized by weak bounds between particles in comparison
to precipitates obtained by CPH method. This reflects the friability of the annealed pellets obtained with the former precipitate (SPH aggregates can be easily broken up by a low pressure, whereas a certain force is required to crush CPH aggregates). Characterization of LLTO powders after synthesis: Heat-treated LLTO powders (900 1C/2 h) were characterized by a single perovskite phase. However, as it was confirmed by TEM experiments, LLTO powders obtained by SPH and CPH methods showed different mesoscopic structures. As shown in Fig. 3c, d and g, h LLTO powders prepared from precursors with a mass-fractal agglomeration structure (SPH route at pHTiðOHÞ4 ¼3.5 and pHLaðOHÞ3 ¼8.5; Fig. 3c and d) revealed lowest agglomeration and density than LLTO powders prepared from precursors with a surface-fractal agglomeration structure (CPH route at pHTiðOHÞ4 ¼pHLaðOHÞ3 ¼8.5; Fig. 3g and h). Thus, the self-organization of precursor particles and their preparation route can drastically change the morphology and technological properties of annealed LLTO nanoparticles. All these results demonstrate the topochemical memory of such materials; the influence of the synthesis conditions on the self-assembly of synthesized precursors as well as the final LLTO nanoparticles aggregation.
4. Conclusions The effect of precipitation conditions (pH and concentration of ammonia solution) of titanium and lanthanum was investigated on the filtration coefficient of precipitates and their properties. It has been shown that precipitates of titanium and lanthanum hydroxides synthesized at pH¼3.5 and 8.5 were characterized by the highest filtration coefficients and the higher “carbon dioxide” amount comparing to the other pH. High values of filtration coefficient and amount of adsorbed carbon dioxide at such pH may be related to the formation of more branched structure of hydroxide nuclei. This allows carbon dioxide to penetrate more into pores between particles.
K.V. Kravchyk et al. / Materials Letters 137 (2014) 182–187
It has been shown that the difference in the pH of precipitation of titanium and lanthanum hydroxides and sequence of hydroxide precipitation (SPH sequential, CPH coprecipitation) strongly affects the hierarchic structure of LLTO precursors. Namely, the LLTO precursors synthesized by SPH (at pH ¼ 3.5 and 8.5 for titanium and lanthanum hydroxides) and CPH methods (at pH ¼8.5 for both titanium and lanthanum hydroxides) were characterized by different types of fractal self-organization: either mass (SPH method) or surface (CPH method) type. It has been shown that mass type of fractal self-organization contributes to a high filtration coefficient of hydroxide precipitates and allows obtaining LLTO precursors with soft and readily friable aggregates. We report that the self-organization of precursor particles and their preparation route can drastically change the morphology of LLTO nanoparticle assemblies and technological properties of annealed LLTO. The more we form branched structures during the synthesis of the LLTO precursors, the less we obtain agglomerated LLTO powders. These results open new possibilities for synthesis of free-ligand soft agglomerated LLTO nanoparticles or similar functional compounds by controlling from simple methods the self-organization of precursors in the synthesis procedure. Acknowledgments K.V. Kravchyk thanks the University of Maine for the financial support and for the opportunity to perform present investigations in the Laboratory of Oxides and Fluorides of the University. The authors are grateful to Stéphanie Kodjikian for the TEM measurements.
187
References [1] Belous AG, Novitskaya GN, Polyanetskaya SV, Gornikov YI. Russ J Inorg Chem 1987;32:156. [2] Inaguma Y, Liquan C, Itoh M, Nakamura T, Uchida T, Ikuta H, et al. Solid State Commun 1993;86:689. [3] Maier J. Solid State Ionics 2002;154-155:291. [4] Maier J. Solid State Ionics 2003;157:327. [5] Maier J. Solid State Ionics 2002;148:367. [6] Hui Sh, Roller J, Yick S, Zhang X, Dec’es-Petit C, Xie Y, et al. J Power Sources 2007;172:493–502. [7] Kosacki I, Rouleau CM, Becher PF, Bentley J, Lowndes DH. Solid State Ionics 2005;176:1319. [8] Pinna N, Niederberger M. Angew Chem Int Ed 2008;47:5292. [9] Nag A, Kovalenko MV, Lee J-S, Liu W, Spokoyny B, Talapin DV. J Am Chem Soc 2011;133:10612. [10] Belous AG, Kravchyk KV, Pashkova OV, Gomza YuP, Nesin SD. J Eur Ceram Soc 2010;30:141. [11] Kravchyk KV, Gomza YuP, Pashkova OV, V’yunov OI, Nesin SD, Belous AG. J Non-Cryst Solids 2009;355:2557. [12] Vaserman IM. Khimicheskoe osazhdenie iz rastvorov (chemical precipitation from solutions). Leningrad: Khimiya; 1980; 16. [13] Beaucage G. J Appl Crystallogr 1995;28:717. [14] Beaucage G. J Appl Crystallogr 1996;29:134. [15] Hyeon-Lee J, Beaucage G, Prausnits SE, Vemury S. Langmuir 1998;14:5751. [16] Pattier B, Henderson M, Brotons G, Gibaud AJ. Phys Chem B 2010;114:5227.
Contents lists available at ScienceDirect
Materials Letters journal homepage: www.elsevier.com/locate/matlet
Li3xLa2/3 xTiO3 nanoparticles with different morphologies and self-organization, obtained from simple solution precipitation methods K.V. Kravchyk a,b,n,1, G. Brotons a,1, A.G. Belous b, O. Bohnke a a b
Université du Maine, Institut des Matériaux et des Molécules du Mans (UMR 6283 CNRS), Av O. Messiaen, 72085 Le Mans Cedex 9, France V.I. Vernadskii Institute of General and Inorganic Chemistry, 32/34 Palladina Avenue, 03680, Kyiv-142, Ukraine
art ic l e i nf o
a b s t r a c t
Article history: Received 27 May 2014 Accepted 30 August 2014 Available online 6 September 2014
We present a novel synthesis strategy of Li3xLa2/3 xTiO3 (LLTO) softly agglomerated nanoparticles by simple precipitation from solution and based on controlling the fractal self-organization. In this way we avoid the use of bulky organic ligands frequently used to avoid the irreversible aggregation of nanoparticles but that also often degrade the LLTO functional properties. We show that the pH of precipitation of titanium and lanthanum hydroxides (LLTO precursors) strongly affects the so-called “filtration coefficient” and the hierarchic structure of LLTO precursors (mass or surface fractal self-organization). For mass-fractal aggregation of the precursors, a high filtration coefficient was obtained with soft and readily friable aggregates of the hydroxide precipitates. Moreover, the fractal self-organization of the synthesized precursors strongly modifies the final LLTO nanoparticles morphology and the material connectivity. The LLTO nanoparticles synthesized from mass-fractal agglomerated precursors obtained from the so-called sequential precipitation of hydroxides (SPH, prepared at pHTiðOHÞ4 ¼3.5 and pHLaðOHÞ3 ¼8.5) had a much lower agglomeration ability than the surface-fractal agglomerated precursors obtained from the so-called co-precipitation of hydroxides (CPH, at pHTiðOHÞ4 ¼ pHLaðOHÞ3 ¼ 8.5). & 2014 Elsevier B.V. All rights reserved.
Keywords: Self-organisation Fractal structure Li3xLa2/3 xTiO3 (LLTO) Precipitation from solution
1. Introduction The search of new materials to improve electrochemical systems, particularly for the transformation and the conservation of electric energy and for chemical sensors, is a strong catalyst for research. The electrolyte is one of the important parts of these electrochemical systems and the improvement of the conductivity of solid electrolytes remains, at the time, a great challenge in research. Inorganic Li ionconducting solid conductors always attract great interest since they may present several advantages, i.e. high electrochemical stability window, high thermal stability, high mechanical resistance and the possibility of miniaturization by preparing thin or thick films. In such a context, Li-conducting ceramics based on Li3xLa2/3 xTiO3 (named hereafter LLTO) hold an important place among them because of their high ionic conductivity at room temperature, i.e. σ¼10 3 S cm 1 for x¼0.10 [1,2]. However, besides this high bulk conductivity, the grainboundary resistance of these ceramics remains high and limits considerably the applications of these polycrystalline ceramics. The n Corresponding author. Current address: ETH Zurich, Laboratory of Inorganic Chemistry, Wolfgang-Pauli-Str. 10, 8093 Zurich, Switzerland. Tel.: þ41 58 765 5944. E-mail address: [email protected] (K.V. Kravchyk). 1 These authors contributed equally to this work.
http://dx.doi.org/10.1016/j.matlet.2014.08.153 0167-577X/& 2014 Elsevier B.V. All rights reserved.
electrical properties of the grain boundaries differ essentially from the bulk because of the presence of strained or missing bonds in the interface region that may lead to a change of the mobile species mobility. Structural or space charge effects in this interfacial region have been suggested to explain these variations [3–5]. Further, based on the grain-boundary core-space-charge layer model, it has been shown that grain boundary resistance can change depending on the grain size. These suggestions were experimentally confirmed on stabilized zirconium oxide nanomaterials, where it has been shown that specific grain boundary conductivity can be increased by obtaining nanodimensional grains [6,7]. In such a context, the attempt to synthesize nanosized LLTO ceramics appeared to be of great interest. For obtaining nanosized ceramics of oxides, it is necessary not only to prepare nanoparticles of precursors but also to control the process of their aggregation during the synthesis. In order to avoid the irreversible agglomeration of inorganic nanoparticles, bulky organic surfactants can be used. By this way, it is also often possible to control their size and shape. However, drawbacks resulting from surface adsorbed surfactants are the unpredictable influence on the toxicity of nanoparticles, and the diminished accessibility to the particle surface [8,9]. The latter can be serious issue regarding grain boundary ionic transport even after high-temperature heat-treatment.
K.V. Kravchyk et al. / Materials Letters 137 (2014) 182–187
Another way to control the formation of nanoparticles and their agglomeration is to grow self-organized fractal structures forming during the synthesis. Thus, it has been shown that nanoparticles have the ability to self-organize in networks, with formation of fractal structures. By changing the synthesis conditions, control of the type of fractal self-organization is possible, resulting in the formation of either soft or hard agglomerated nanoparticles. Thus, in recent papers, Belous et al. [10,11] have shown that, by using precipitation from solution as synthesis method, by controlling the pH of the solution and by changing the sequence of hydroxides precipitation, it was possible to obtain different types of fractal self-organization of nanoparticles of zirconium and yttrium hydroxides (bulk and surface). This fractal self-organization of the precursor powder affects the morphology of the final oxide powder and the properties of the ceramics (sintering temperature, size of grains). In this work, we investigated and compared self-organization of the precursor and final LLTO nanoparticles synthesized by two routes: the sequential precipitation (SPH) and the coprecipitation (CPH) synthesis. The LLTO precursors used are a mixture of titanium and lanthanum hydroxides to which a lithium hydroxide solution is added after hydroxide precipitation in an adequate quantity. The aim of this paper is to investigate the self-organization of the LLTO precursor nanoparticles obtained and its influence on final morphology. We first investigated separately, the precipitation conditions of titanium and lanthanum hydroxides (pH and concentration of reduction agent). As is known, these factors can efficiently affect the dispersion (or self-organization) of the precipitates and their agglomeration. Afterwards, we investigated the influence of the sequence of precipitation of the two hydroxides (sequential or co-precipitation) on the fractal structure and on the microstructure of the precursor and final LLTO nanoparticles.
2. Experimental procedure Synthesis and characterization of La(OH)3 and Ti(OH)4 hydroxides: La(OH)3 was precipitated from an aqueous solution of La(NO3)3 by aqueous solution of NH4OH to adjust the pH from 8 to 11. Ti(OH)4 was precipitated from an isopropanol solution of TiCl4 by aqueous solution of NH4OH to adjust the pH from 2 to 8.5. The concentration of NH4OH used for adjusting the pH was varied (1 mol%, 5% or 25% NH3) since it can influence the morphology of the precipitates. The so-obtained precipitates of La and Ti hydroxides were filtered and washed from the mother solution with distilled water until Cl and NO3 ions disappear from the washing solutions. Afterwards, hydroxide precipitates were washed additionally with 500 ml of distilled water in order to determine filtration coefficient Kf. Filtration coefficient, Kf, of each fresh precipitate was calculated using Darsi formula [12]: Q ¼ Kf S
H t L
ð1Þ
where Q—filtrate volume, cm3; Kf—filtration coefficient, cm s 1; S—precipitate surface on the filter, cm2; H—pressure under filter, cm of Hg; L—thickness of precipitate layer, cm; t—time of filtration, s. Precipitates were then dried at 80 1C. The compactness of the hydroxide pellets (g cm 3) was determined from the mass, diameter and height of pellets after sintering. Pellets were obtained by uniaxial compression at 16 MPa (kg cm 2) of hydroxide powders, followed by a heat treatment at 600 1C. Infra Red (IR) spectroscopy and thermal analyses (DTA, TGA) were used to characterize the La and Ti hydroxides. IR spectra were recorded on a Specord-M80 spectrometer in the 250–4000 cm 1 frequency range. Samples consisted of pellets prepared by pressing
183
the hydroxide with dehydrated KBr (1 wt%). DTA and TGA experiments were carried out with a Setaram TGDTA92 equipment, at a heating rate of 5 1C min 1, in air using Pt crucible. Preparation of the precursors of LLTO: The synthesis of the precursors of LLTO has been carried out either by sequential precipitation of the hydroxides (SPH) or by their coprecipitation (CPH). The mixture of titanium and lanthanum hydroxides obtained by SPH was carried out in the following way: the titanium hydroxide was first precipitated at pH ¼3.5 and afterwards lanthanum hydroxide was precipitated at pH ¼ 8.5. In the case of CPH method, lanthanum and titanium hydroxides were precipitated simultaneously at pH ¼8.5. After obtaining the mixture of these two hydroxides, the precipitates were kept for 2 h in solution to control the loss of metal cations during washing procedure. Then, hydroxide mixtures washed with distilled water and an appropriated amount of an aqueous solution of LiOH was added in order to obtain the chemical composition Li0.3La0.57TiO3. The obtained suspension was dried afterwards in an oven at 80 1C. The microstructure of these precursors was studied by small angle X-ray scattering (SAXS). Measurements were performed in transmission mode trough vacuum with sample pellets of known density and thicknesses obtained by uniaxial pressure. The instrument source is a high flux Rigaku™ copper rotating anode and the beam is set by three pinholes after an Osmic™ confocal mirror that delivered a monochromatic (λ 1.54 Å) and cylindrical X-ray beam of 2 107 photons s 1 over 350 μm in diameter with an angular divergence below 0.5 mrad. To obtain a wide range of wave vectors (0.015 Å 1 to 2 Å 1) we combined a 2D gas detector and a “flat image plate” for larger angles, with a broad overlap from 0.1 to 0.3 Å 1. The absolute q-range was calibrated from the known diffraction lines of Ag-Behenate powder and intensities were given in absolute scale units calculated from the water scattering level (intensities in cm 1, i.e. cm² per cm3 of sample material). Morphological observations were performed using Transmission Electronic Micrsocopy (TEM). Thin specimens were obtained by ultrasonically dispersing particles in ethanol and depositing one droplet of the resulting suspension on a Cu grid covered with a holey carbon film. After drying, the grid was fixed in a side-entry 301 double-tilt specimen holder and was introduced in a JEOL2010 electron microscope operating at 200 kV. Synthesis of LLTO powder: LLTO has been synthesized from these precursors by heat treatment at 900 1C in air for 2 h. Powder X-ray diffraction (XRD) patterns have been recorded at room temperature with a Philips X’Pert PRO diffractometer (Cu Kα radiation), equipped with a linear X’Cellerator detector, in the 2Θ range from 5 to 701 with an interpolated step of 0.081. TEM has also been carried out to determine the microstructure of the final oxide. The lithium content in the LLTO powders was determined using chemical analysis. K2S2O7 was added in excess to LLTO powders and alloyed during 24 h on a sand bath. The alloy was then dissolved in acidified water. The analysis of the solutions was carried out with an atomic absorption spectrometer (SP-9 PueUnicom) showing a lithium content below 2%. To overcome this substantial loss of lithium after the heat treatment of LLTO precursor, we thus used an excess of LiOH for the LLTO synthesis.
3. Results and discussion Characterization of individual precipitates of La(OH)3 and Ti(OH)4: Previous investigations performed during the precipitation of zirconium and yttrium oxides have shown that the pH of the solution affects efficiently the filtration coefficient, Kf, of the precipitates. This coefficient is correlated to the fractal structure of the precipitates and to the strength of the bounds formed between particles during precipitation. The higher the filtration
184
K.V. Kravchyk et al. / Materials Letters 137 (2014) 182–187
coefficient is, the softer the agglomerated powder and the weaker the bounds between particles are [10]. According to these results, we determined the value of Kf for the individual lanthanum and titanium hydroxide precipitates. Fig. 1a shows the values of the filtration coefficient obtained for lanthanum and titanium hydroxides as a function of the pH of the solution (the concentration of NH4OH was fixed at 5 mol% NH3). Maxima are obtained at pH ¼3.5 and 8.5 for Ti(OH)4 and La(OH)3 precipitates respectively. These pH values indicate the experimental conditions of precipitation to obtain a soft precipitate with weak bounds between particles. Out of these values the obtained precipitates have a low filtration coefficient and therefore have strong bounds. Fig. 1a also displays the values of the compactness of the pellets obtained after drying, pressing and annealing in air at 600 1C. It can be observed that the compactness follows the Kf coefficient. The highest compactness is obtained for powders precipitated
Fig. 1. (a) Filtration coefficients of titanium and lanthanum hydroxides (left axis) and compactnesses of annealed titanium and lanthanum oxides at T¼ 600 1C (right axis), as a function of precipitation pH; (b) filtration coefficients of titanium and lanthanum hydroxides as a function of concentration of added ammonia solution; (c) IR spectra of titanium and lanthanum hydroxides obtained at different pH of precipitation.
at the pH that shows the highest filtration coefficient suggesting that the nature of the bounds between particles influences the compactness of the annealed pellet. The pellets made from weak bounded particles, are more compact. It is also important to note that the filtration coefficient data of Ti(OH)4 and La(OH)3 precipitates show a good correlation with the mechanical properties of the annealed pellets. The pellets obtained at pH ¼3.5 and pH ¼8.5 for Ti(OH)4 and La(OH)3 respectively were soft, suggesting that weak bounds exist between particles. On the other hand, the annealed pellets obtained from the hydroxides precipitated at higher or lower pH were very difficult to grind, suggesting that the particles are more strongly bounded. We also investigated the influence of the ammonia concentration on Kf. Fig. 1b presents the values of the filtration coefficient as a function of ammonia solution concentration for precipitation performed at pH ¼3.5 and 8.5 for Ti(OH)4 and La(OH)3 respectively. The concentration of 1 mol% NH3 is the best one to obtain La (OH)3 precipitate with the highest filtration coefficient and a concentration of 5 mol% NH3 for Ti(OH)4 precipitates. These results clearly show that the pH and the ammonia concentration of the solution greatly influence the morphology of the precipitates. IR-spectra of individual titanium and lanthanum hydroxides, precipitated at different pH, are presented in Fig. 1c. The absorption bands observed on the spectra are related to the presence of OH-groups, adsorbed H2O molecules and to the presence of carbonates and adsorbed CO2 molecules. The absorption bands in the interval 3700–2700 cm 1 correspond to the stretching vibrations of OH-groups. The broadening and the shift of the absorption bands, typical for stretching vibrations of isolated OH-groups (3750– 3500 cm 1), indicate the presence of hydrogen bonds in individual lanthanum and titanium hydroxides. Bands in the frequency range from 1600 to 1685 cm 1 correspond to deformation vibrations of H2O molecules and may be attributed to adsorbed water. Absorption bands in the frequency range from 1330 to 1430 cm 1 can be attributed to deformation vibrations of OH-groups. All these absorption bands are observed on both lanthanum and titanium hydroxides. Bands around 2275 to 2400 cm 1 correspond to deformation vibrations of molecules of carbon dioxide. They are slightly observed on Ti(OH)4 precipitated at pH¼3.5, they are more intense on La(OH)3 precipitated at pH¼8.5. They can be attributed to CO2 adsorption. The presence of absorption bands in IR-spectra of lanthanum hydroxide at 1050 and 850 cm 1 indicates the presence of carbonate groups, namely the peak at 1050 cm 1 can be assigned to stretching vibration of CO and the peak at 850 cm 1 to deformation vibration of CO. To resume the above obtained results, on titanium hydroxide IR spectra revealed the presence of OH groups, adsorbed H2O molecules and adsorbed CO2 molecules. These last molecules are in a very small amount and mostly on the hydroxide prepared at pH¼3.5. On the lanthanum hydroxide the presence of OH-groups and adsorbed H2O molecules is observed but also the presence of carbonate species and of CO2 molecules in a greater amount than in titanium hydroxide (bands at 2275–2400 cm 1) and especially for the hydroxide prepared at pH¼ 8.5. The intensity of absorption bands correlates clearly with the dependence of both, the filtration coefficient and the density of precipitates, on pH. Samples corresponding to the highest Kf and to the highest compactness are characterized by deeper absorption bands of carbon dioxide suggesting that these samples have the highest surface of adsorption for CO2. Further, the presence of carbonates on lanthanum hydroxide is clearly related to the more basic character of lanthanum compared to titanium. Fig. 2 shows the curves of differential thermal analysis (DTA) and thermal gravimetry (TG) of dried samples of titanium and lanthanum hydroxides precipitated at different pH. The weight loss of titanium hydroxides is characterized by a constant rate up to 250 1C, that is
K.V. Kravchyk et al. / Materials Letters 137 (2014) 182–187
185
decomposition temperature of lanthanum oxycarbonate La2O2(CO3). Indeed, the higher is pH for the precipitation of lanthanum hydroxide, the higher is the decomposition temperature of lanthanum oxycarbonate La2O2(CO3). From the comparison of TG curves of lanthanum hydroxides it follows that lanthanum hydroxides precipitated at lower pH are characterized by a higher weight loss compared to precipitates obtained at higher pH. Characterization of LLTO precursor precipitates obtained either by SPH or CPH methods: The precipitate of LLTO precursors has been obtained either by SPH or CPH, as described in the experimental part of the paper. Fig. 3a and e show SAXS curves of SPH and CPH precipitates respectively. The two SAXS curves are different but where analyzed with the same model. We used the Beaucage’s unified exponential/power-law scattering function [13–15]. The equation describing an arbitrary number of interrelated structural levels has the form n
I ðqÞ ¼ ∑
i¼1
Fig. 2. Thermal analyses plots of titanium and lanthanum hydroxides obtained at different pH of precipitation.
related to the loss of adsorbed water molecules at lowest temperatures and of structural water molecules around 200 1C. The loss of adsorbed water is confirmed by the presence of an endothermic peak around 100 1C and the loss of structural water is confirmed by the presence of an exothermic peak around 220 1C on DTA curves that reveals more clearly at higher pH of precipitation (pH46). The exothermic peak at 430 1C corresponds to the crystallization of the anatase phase of titanium oxide. Above 500 1C, rutile phase begins to appear. It allows supposing that the transition from anatase phase to rutile phase takes place between 500 and 1000 1C. The comparison of DTA curves of titanium hydroxides obtained at different pH allows asserting that the pH of precipitation affects both the temperature interval of transition from titanium hydroxide to oxide and the crystallization temperature of the anatase phase of titanium oxide. Besides, it is important to note, that the exothermic peaks, which correspond to the dehydration of titanium hydroxide and the crystallization of anatase phase become more diffused at low pH. This might be connected with simultaneous processes of loss of adsorbed water, dehydration and crystallization. The comparison of weight loss of titanium hydroxides precipitated at different pH indicates that titanium hydroxide precipitate obtained at pH¼ 3.5 has the lowest amount of adsorbed water and hydroxyl groups. The analysis of TG and DTA curves of lanthanum hydroxide allows one to suggest that when increasing the temperature, the following processes take place: weight loss of adsorbed water around 100 1C, partial dehydration of lanthanum hydroxide to La2O2(OH)2 phase and interaction of the latter with carbon dioxide to form lanthanum oxycarbonate La2O2(CO3) corresponding to the endothermic peak at 325 1C. The formation of this oxycarbonate phase La2O2(CO3) was displayed on IR-spectra of lanthanum hydroxides at 1500, 1050 and 850 cm 1, which indicate the presence of carbonate groups. In the temperature interval 450–550 1C, partial decomposition of lanthanum oxycarbonate takes place, what is indicated by the endothermic peak and sharp weight loss on TG curve. Endothermic peak at 650–750 1C is connected obviously with full decomposition of lanthanum oxycarbonate to form lanthanum oxide. Comparison of the DTA and TG curves of lanthanum hydroxides obtained at different pH shows that at increasing pH of precipitation the behavior of curves changes partially. Namely, at pH¼10 the processes of decomposition of lanthanum oxycarbonate do not take place until 650–750 1C. It reflects in absence of endothermic DTA peak at 500 1C and almost lacking decreasing of weight loss on TG curves at 450–650 1C comparing to lower pH values. It means that a change of the precipitation pH of lanthanum hydroxide affects the
pffiffiffi3 P i Gi exp q2 R2gi =3 þ Bi exp q2 R2gði þ 1Þ =3 erf qRgi = 6 =q
!
ð2Þ where Gi is the Guinier prefactor for level i, Bi is the Porod prefactor (in the power-law dependence of the logarithm of the scattered intensity on the logarithm of the scattering vector). Accordingly, this relationship continuously describes the small angle Guinier regime due to scattering from large scale domains (fractal aggregates at level i) with radius of gyration Rgi and the Porod regime due to scattering from these domain interfaces. At first level 1, we expect smaller nanoparticles (building blocks of the fractals). The exponent P i , defining the fractal dimension of the domains at level i, is responsible for the q Pi asymptotical regime. Depending on the values of Pi, we distinguished two different fractal types: (1) a mass fractal system with 1 o P o 3 with a massfractal dimension Dm ¼P. The domains are tenuous for smaller values of Dm and denser for larger values. (2) A surface fractal system with 3 o P o 4 and a surface-fractal dimension Ds ¼ 6 P (while Dm ¼ 3). A perfectly flat interface corresponds to Ds ¼2 and the Ds value increases with roughness. Both, mass and surface fractals are common types of domains found in dispersive systems made of small clusters and nanoparticles [16]. The SAXS slope is then a strong indication of the type of aggregation and dimension of fractals. The SPH curve is characterized by a linear region with a slope of 2.99, when plotted in log–log scale up to q¼ 0.27 Å 1. At larger angles, intensity reaches a constant level due to the background and the broad tail of a diffraction peak (Fig. 3a). We first fitted the data successfully using only one level of structuration in (equation 2). For this, we obtained large fractal domains (Rg1 75– 85 nm) with a Guinier plateau that appeared at angles that are smaller than our measurement window (i.e. the G1 term could be neglected). Thus, the second term B1 fitted the data up to 0.27 Å 1 and corresponds to the surface scattering from a dense mass fractals (P1 2.99) made of very small building blocks (Rg2 o1.5 nm). By introducing a second level of structuration in (equation 2), we could fit the data with parameters corresponding to large mass-fractal domains (Rg1 75–85 nm), dense (P1 2.75) and made of small nanoparticles with a diameter of d1 9 nm (Rg2 3.5 nm and for a plain particle d1 2.58 Rg2). The sum of three terms reproduced the constant slope in log–log scale: the B1 term fitted the first part up to q¼ 0.05 Å 1; for q40.1 Å 1 the strongest contribution came from B2 and the G2 terms dominated the intermediate range scattering. The CPH curve (Fig. 3e) shows three regimes with less intensity at smaller angles than CPH. For 0.01 oq o0.07 Å 1 a slope of 0.77 was measured, 3.13 for 0.07 oq o0.3 and a slope of 0.26 at larger angles before reaching background. Using two structural levels, the data are well fitted from the Guinier and Porod scattering terms (G2 dominates for q o0.06 Å 1 and then B2 dominates up to 0.3 Å 1) from
186
K.V. Kravchyk et al. / Materials Letters 137 (2014) 182–187
Fig. 3. Small-angle X-ray scattering curves of SPH (a) and CPH (e) samples dried at 80 1C; I(q), with absolute intensity (cm 1) and q wave vector (Å 1) in log–log scale (a scheme of the fractal structures is given on top of the figures); microphotographs of SPH and CPH samples dried at T ¼ 80 1C (b) and (f) and annealed at T ¼900 1C for 2 h (c), (d) and (g), (h).
nanoparticles (Rg1 5.5 nm) that aggregated in the form of a corrugated surface fractal with Rg1 30 nm and P1 3.1. These large aggregates agglomerated at larger scales without visible fractal structure. This latter behavior is shown by the saturation of intensity for q o0.07 in comparison to the SPH sample. It is worth noting that all SAXS data are in good agreement with the TEM results. Fig. 3b, f present the micrographs of LLTO precursors obtained by SPH and CPH methods. As it appears in Fig. 3b and f both precipitates have different mesoscopic organization. Fig. 3b clearly evidences the formation of mass fractal with a branched structure. Fig. 3f shows the formation of dense precipitates of 300–500 nm size. This correlates with the agglomeration of surface fractal structure. By considering the difference in the self-organization of the particles in SPH and CPH precipitates and their conditions of preparation (pHTiðOHÞ4 ¼3.5 and pHLaðOHÞ3 ¼8.5 for SPH; pHTiðOHÞ4 ¼ pHLaðOHÞ3 ¼8.5 for CPH), it is easy to understand why the filtration coefficients of titanium and lanthanum hydroxides, obtained at pH ¼ 3.5 and 8.5 respectively, are higher than the ones obtained at other pH (Fig. 1a). During CPH precipitation both hydroxides are precipitated at pH ¼8.5. At this value of pH, compact surface aggregates of Ti(OH)4 are formed with strong bounds between particles. Those compact surface aggregates form compact particles of big size with non-fractal behavior. Simultaneously, La(OH)3 precipitates but the morphology of the final product is dictated by the Ti hydroxide one which is in a greater amount. On the other hand, during SPH precipitation Ti hydroxide is formed first at pH ¼ 3.5. In such a condition the primary particles aggregate and form a mass fractal with a branched structure. Afterwards, La hydroxide is precipitated at pH ¼ 8.5 adopting a mass fractal morphology and an open structure too. Consequently, water will flow easily through such precipitates in contrast to the compact precipitates with surface fractal structure formed by CPH method. The precipitates obtained by SPH method are characterized by weak bounds between particles in comparison
to precipitates obtained by CPH method. This reflects the friability of the annealed pellets obtained with the former precipitate (SPH aggregates can be easily broken up by a low pressure, whereas a certain force is required to crush CPH aggregates). Characterization of LLTO powders after synthesis: Heat-treated LLTO powders (900 1C/2 h) were characterized by a single perovskite phase. However, as it was confirmed by TEM experiments, LLTO powders obtained by SPH and CPH methods showed different mesoscopic structures. As shown in Fig. 3c, d and g, h LLTO powders prepared from precursors with a mass-fractal agglomeration structure (SPH route at pHTiðOHÞ4 ¼3.5 and pHLaðOHÞ3 ¼8.5; Fig. 3c and d) revealed lowest agglomeration and density than LLTO powders prepared from precursors with a surface-fractal agglomeration structure (CPH route at pHTiðOHÞ4 ¼pHLaðOHÞ3 ¼8.5; Fig. 3g and h). Thus, the self-organization of precursor particles and their preparation route can drastically change the morphology and technological properties of annealed LLTO nanoparticles. All these results demonstrate the topochemical memory of such materials; the influence of the synthesis conditions on the self-assembly of synthesized precursors as well as the final LLTO nanoparticles aggregation.
4. Conclusions The effect of precipitation conditions (pH and concentration of ammonia solution) of titanium and lanthanum was investigated on the filtration coefficient of precipitates and their properties. It has been shown that precipitates of titanium and lanthanum hydroxides synthesized at pH¼3.5 and 8.5 were characterized by the highest filtration coefficients and the higher “carbon dioxide” amount comparing to the other pH. High values of filtration coefficient and amount of adsorbed carbon dioxide at such pH may be related to the formation of more branched structure of hydroxide nuclei. This allows carbon dioxide to penetrate more into pores between particles.
K.V. Kravchyk et al. / Materials Letters 137 (2014) 182–187
It has been shown that the difference in the pH of precipitation of titanium and lanthanum hydroxides and sequence of hydroxide precipitation (SPH sequential, CPH coprecipitation) strongly affects the hierarchic structure of LLTO precursors. Namely, the LLTO precursors synthesized by SPH (at pH ¼ 3.5 and 8.5 for titanium and lanthanum hydroxides) and CPH methods (at pH ¼8.5 for both titanium and lanthanum hydroxides) were characterized by different types of fractal self-organization: either mass (SPH method) or surface (CPH method) type. It has been shown that mass type of fractal self-organization contributes to a high filtration coefficient of hydroxide precipitates and allows obtaining LLTO precursors with soft and readily friable aggregates. We report that the self-organization of precursor particles and their preparation route can drastically change the morphology of LLTO nanoparticle assemblies and technological properties of annealed LLTO. The more we form branched structures during the synthesis of the LLTO precursors, the less we obtain agglomerated LLTO powders. These results open new possibilities for synthesis of free-ligand soft agglomerated LLTO nanoparticles or similar functional compounds by controlling from simple methods the self-organization of precursors in the synthesis procedure. Acknowledgments K.V. Kravchyk thanks the University of Maine for the financial support and for the opportunity to perform present investigations in the Laboratory of Oxides and Fluorides of the University. The authors are grateful to Stéphanie Kodjikian for the TEM measurements.
187
References [1] Belous AG, Novitskaya GN, Polyanetskaya SV, Gornikov YI. Russ J Inorg Chem 1987;32:156. [2] Inaguma Y, Liquan C, Itoh M, Nakamura T, Uchida T, Ikuta H, et al. Solid State Commun 1993;86:689. [3] Maier J. Solid State Ionics 2002;154-155:291. [4] Maier J. Solid State Ionics 2003;157:327. [5] Maier J. Solid State Ionics 2002;148:367. [6] Hui Sh, Roller J, Yick S, Zhang X, Dec’es-Petit C, Xie Y, et al. J Power Sources 2007;172:493–502. [7] Kosacki I, Rouleau CM, Becher PF, Bentley J, Lowndes DH. Solid State Ionics 2005;176:1319. [8] Pinna N, Niederberger M. Angew Chem Int Ed 2008;47:5292. [9] Nag A, Kovalenko MV, Lee J-S, Liu W, Spokoyny B, Talapin DV. J Am Chem Soc 2011;133:10612. [10] Belous AG, Kravchyk KV, Pashkova OV, Gomza YuP, Nesin SD. J Eur Ceram Soc 2010;30:141. [11] Kravchyk KV, Gomza YuP, Pashkova OV, V’yunov OI, Nesin SD, Belous AG. J Non-Cryst Solids 2009;355:2557. [12] Vaserman IM. Khimicheskoe osazhdenie iz rastvorov (chemical precipitation from solutions). Leningrad: Khimiya; 1980; 16. [13] Beaucage G. J Appl Crystallogr 1995;28:717. [14] Beaucage G. J Appl Crystallogr 1996;29:134. [15] Hyeon-Lee J, Beaucage G, Prausnits SE, Vemury S. Langmuir 1998;14:5751. [16] Pattier B, Henderson M, Brotons G, Gibaud AJ. Phys Chem B 2010;114:5227.