powder mixtures: Influence of transition-metal oxide on reaction rate

Solid State Ionics 190 (2011) 1–7

Contents lists available at ScienceDirect

Solid State Ionics j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / s s i

Mechanochemical reactions in Na2CO3–M2O5 (M = V, Nb, Ta) powder mixtures: Influence of transition-metal oxide on reaction rate Tadej Rojac ⁎, Živa Trtnik, Marija Kosec Jožef Stefan Institute, Jamova cesta 39, 1000 Ljubljana, Slovenia

a r t i c l e

i n f o

Article history: Received 27 December 2010 Received in revised form 17 March 2011 Accepted 21 March 2011 Available online 15 April 2011 Keywords: Mechanochemical synthesis Reaction mechanism Reaction rate

a b s t r a c t We have studied the mechanochemical reactions between Na2CO3 and the transition-metal oxides M2O5 (M= V, Nb, Ta), giving special attention to the course of the individual reaction and to the influence of the transitionmetal oxide on the reaction rate. A common mechanism links the studied reactions and is characterized by the formation of an intermediate amorphous carbonato complex, in which the CO2– 3 ions that are present initially in the Na2CO3, coordinate to the transition-metal cation. In a subsequent milling stage the carbonato complex decomposes, leading to the formation of the final binary compounds, i.e., NaVO3, NaNbO3 and NaTaO3. Quantitative experimental results, based on the Rietveld refinement method and thermogravimetric analyses, showed that the reaction rate follows the order Na2CO3 + V2O5 N Na2CO3 + Ta2O5 N Na2CO3 + Nb2O5. By analyzing the acid–base properties of the participating reagents, we found that the experimental observations agree with the acid–base interaction mechanism, which means that the larger the acidity of the transition-metal cation involved, the faster and more complete the mechanochemical reaction. Finally, it was found that only partial decomposition of the carbonato complex, even after a prolonged mechanochemical treatment, can be expected in reactions characterized by lower acid–base potential. © 2011 Elsevier B.V. All rights reserved.

1. Introduction In recent years, we have been involved in a continuous emerging of mechanochemical synthesis in the area of complex ceramics oxides, which has resulted in new possibilities for the preparation of highly homogeneous oxides with improved piezoelectric, ferroelectric or other functional properties [1]. Although the technique has already provided useful results, the understanding of the basic reaction mechanisms and kinetics has still not reached a satisfactory level, with the main reason probably being the complexity of the mechanochemical reactions and the shortage of systematic studies with a quantitative basis. With their pioneering work on the mechanisms of the mechanochemical reactions in hydroxide–oxide mixtures, such as Me(OH)x–SiO2 (Me = Ca, Mg, Sr, Al), Senna and co-workers [2–5] contributed essentially to the understanding of the mechanochemical interactions between solids. They showed that the reactions involving –OH groups proceed via an acid–base mechanism on the surface of dissimilar particles, which results in mechanochemical dehydration and the formation of typically amorphous products containing heterobridging M’\O\M bonds (M and M’ are dissimilar metals). Further studies on the mechanisms of mechanochemical reactions in which oxide starting compounds are involved were made by Wang et al. [6,7] and Kuscer et al. [8,9]. The mechanochemical synthesis of various

⁎ Corresponding author. Tel.: + 386 1 477 38 34; fax: + 386 1 477 38 87. E-mail address: [email protected] (T. Rojac). 0167-2738/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.ssi.2011.03.011

Pb-based perovskites, such as, for example, Pb(Mg1/3Nb2/3)O3 (PMN), was interpreted as an “nucleation-and-growth” process from the amorphous phase [6]. However, in addition to this and on the basis of a quantitative phase analysis, Kuscer et al. [9] proposed that after prolonged milling the formation mechanism of the perovskite switches to a pyrochlore-to-perovskite type of transformation, which proceeds in a similar way to that characteristic for a conventional solid-state synthesis. Important knowledge has also been acquired in the understanding of mechanochemical reactions from starting mixtures of carbonates and oxides. Our recent studies showed that the reaction between Na2CO3 and Nb2O5 proceeds through the formation of an intermediate carbonato complex, which subsequently decomposes upon further mechanochemical treatment, leading to the crystallization of NaNbO3 [10,11]. Recently, these findings turned out to have an important practical impact, i.e., in order to obtain highly homogeneous (K,Na,Li) (Nb,Ta)O3 (KNLNT) ceramics, a prerequisite for an enhanced piezoelectric response, the formation of the carbonato complex in the stage of mechanochemical activation, was found to be essential. In fact, applying milling conditions that did not result in the formation of the carbonato complex (for example, by milling in the friction mode) resulted in inhomogeneities in the KNLNT ceramics [12,13]. In spite of the progress made recently in the understanding of the basic mechanisms of mechanochemical reactions from different starting compounds, there are still several questions that need to be clarified, particularly in relation to the kinetics of mechanochemical reactions and the feasibility of certain oxides to be synthesized

2

T. Rojac et al. / Solid State Ionics 190 (2011) 1–7

mechanochemically. Our recent study based on a milling map showed, for example, that the critical cumulative kinetic energy (a measure of the overall kinetic energy, transferred from the milling balls to the powder that is necessary to obtain a certain oxide product during the mechanochemical treatment) for a variety of reactions found in the literature has a broad span, i.e., over an order of magnitude (10–200 kJ/g), for ball-impact energies ranging between 120 and 180 mJ [14]. Another good example are the mechanochemical reactions in the systems Na2CO3–Nb2O5 and K2CO3–Nb2O5; while the formation of NaNbO3 requires a low cumulative kinetic energy (around 10 kJ/g), a more than 500 times larger cumulative energy (5200 kJ/g) is required for the reaction involving K2CO3[15]. The present work aims to clarify some important aspects related to the mechanism and kinetics of mechanochemical reactions involving oxide-carbonate starting mixtures. A systematic selection of the reaction systems was made, i.e., Na2CO3 was chosen as the carbonate compound and was reacted with 5th group transition-metal oxides, i.e., V2O5, Nb2O5 and Ta2O5. The sequence of the rates of these three reactions, consisting of the formation of the intermediate carbonato complex, its decomposition and the crystallization of the final oxides (NaVO3, NaNbO3 and NaTaO3), agrees with the concept of the acid– base interaction mechanism. Therefore, the reaction rate depends strongly upon the acid–base potential between the participating reagents and should be taken into account when estimating the total reaction time of such mechanochemical reactions. 2. Experimental The mechanochemical syntheses of NaVO3, NaNbO3 and NaTaO3 were carried out starting from stoichiometric mixtures of Na2CO3 (99.96%, Sigma-Aldrich) and V2O5 (N99.9%, Alfa Aesar), Nb2O5 (N99.9%, Sigma-Aldrich) or Ta2O5 (N99.9%, Alfa Aesar), respectively. The transition-metal oxides were used as received. The median particle size d50, as determined by the area distribution measured by a laser granulometer (Microtrac S3500), of all the three oxides was around 0.3 μm. Na2CO3, being composed of larger particles ranging from 1 to 30 μm (d50 = 3 μm), was milled prior use in a planetary mill at 175 min− 1 of disc rotational frequency for 4 h using acetone and YSZ (yttria-stabilized zirconia) milling balls with diameters of 3 and 10 mm. Due to its hygroscopic nature, the carbonate was dried after milling at 200 °C overnight, cooled down in a desiccator and weighed immediately. The homogenization of the starting compounds was performed in acetone at 200 min− 1 for a duration of 2 h. Twenty grams of Na2CO3–M2O5 (M=V, Nb, Ta) mixtures were highenergy milled using a Fritsch Pulverisette 4 planetary mill (Fritsch GmbH, Idar-Oberstein, Germany). A 125-ml cylindrical tungsten carbide milling vial (height 4.4 cm, diameter 6 cm) filled with 23 tungsten carbide milling balls of 10 mm diameter was used. The rotational frequency of the supporting disc was set to 300 min− 1, while the transmission ratio (vialto-disc rotational frequency ratio) was fixed at −2. According to the model from ref. [14], these milling conditions correspond to 93 mJ of ballimpact energy and 520 s− 1 of ball-impact frequency. X-ray powder diffraction (XRD) analyses were conducted using a Panalitical X'Pert PRO diffractometer with Cu Kα1 radiation (PANalytical B.V., Almelo, Netherlands). The data were collected in the 2θrange from 10° to 90° with a step of 0.017° per 100 s and a fully opened X'Celerator detector. In some cases a polymeric foil was used to cover the powders during the measurement in order to prevent reactions with environmental H2O and CO2. Quantitative XRD phase analyses were made with the Rietveld refinement method using the Topas R software package (Bruker, AXS, Karlstuhe, Germany). The crystallographic data for V2O5[16], Nb2O5 [17], Ta2O5[18], NaVO3[19], NaNbO3[20] and NaTaO3[21] were taken from the ICSD database. To refine the high XRD-background a polynomial of the order of 20 was used. A modified pseudo-Voigt function was chosen as a profile-shape function. The scale factor, the

unit-cell parameters and the parameters of the profile-shape function were refined for each phase. In addition to this, a displacementcorrection factor was also refined. In order to keep the number of refinement parameters reasonably low and to avoid correlations with the scale factor, the atomic positions and occupancies were kept constant during the refinements. The Rwp factors of the refined patterns were in the range of 5–10%. The infrared spectra were recorded using a Perkin-Elmer Spectrum 100 FT-IR spectrometer, equipped with a Specac Golden Gate Diamond ATR as a sample support. The data were collected in the wavenumber range of 600–4000 cm− 1 with a spectral resolution of 2 cm− 1 and 4 scans. A NETZSCH STA 409 analyzer was used for thermogravimetric (TG) analyses. A total of 50 mg of powder was placed in a Pt/Rh crucible and heated up at a rate of 10 °C/min until it reached a constant mass. The measurements were performed in an atmosphere of flowing air. A simultaneous evolved-gas analysis (EGA) on H2O and CO2 using a Balzers Thermostar GSD 300T mass spectrometer was also performed. The amount of the carbonate, equivalent to Na2CO3, after various milling times (subsequent referred to as the carbonate fraction), was calculated taking into account the total measured CO2 loss from the samples and the theoretical CO2 loss for the complete decomposition of Na2CO3 in a 1:1 stoichiometric mixture with V2O5, Nb2O5 or Ta2O5. In order to estimate the acidic properties of the transition-metal cations involved in this study, data on the coordination number (CN) in the individual oxide structures were required (see Table 1 and Section 3.3 for details). Being made up of VO5 tetragonal pyramids [22,23], CN of 5 was used for V5+ in orthorhombic V2O5. In contrast, orthorhombic Nb2O5 and Ta2O5 are composed of 6- and 7-coordinated Nb and Ta, forming distorted octahedra and pentagonal bipyramids as coordination polyhedral [24,25]; in these two cases, an average of CN = 6.5 was used for calculations. 3. Results and discussion 3.1. Mechanochemical synthesis of NaMO3 (M = V, Nb, Ta) The formation of NaVO3, NaNbO3 and NaTaO3 as the final products of the mechanochemical reactions between Na2CO3 and M2O5 (M = V, Nb, Ta) was followed by XRD analysis (Fig. 1). In the case of the Na2CO3–V2O5 mixture (Fig. 1a), 1 h of milling resulted in the appearance of NaVO3. Unreacted V2O5 was still present at this milling stage and also low-intensity Na2CO3 peaks, close to the background level, could be detected. With further milling for 2 h the amount of NaVO3 increased at the expense of the V2O5, while Na2CO3 was no longer detected. According to the XRD, 4 h of milling was sufficient to bring the reaction to completion since only NaVO3 was present as a crystalline phase. Further milling up to 48 h did not result in any significant changes to the XRD patterns, confirming the stability of NaVO3 under high-energy ball-impacts. The major phase present in the Na2CO3–Ta2O5 mixture (Fig. 1b) after 8 h of milling was Ta2O5. A small amount of the newly formed NaTaO3 could also be identified, while no Na2CO3 could be detected. Further milling brought about a gradual formation of the tantalate and, simultaneously, a decrease of the Ta2O5 content in the mixture. In a sharp contrast to the reaction involving V2O5, much longer milling, i.e., 72 h, was necessary to obtain pure NaTaO3, according to the XRD. The reaction between Na2CO3 and Nb2O5 (Fig. 1c) proceeded in a similar way to the reaction with Ta2O5 (Fig. 1b); however, some important differences can be found. After 8 h of milling, besides Nb2O5 and NaNbO3, also Na2CO3 could be detected, which was not the case in the Na2CO3–Ta2O5 mixture after the same milling time. This suggests that Na2CO3 is more likely to become amorphous and, thus, undetectable with XRD, by milling with Ta2O5 than with Nb2O5. Furthermore, a much longer milling time, i.e., 150 h, was necessary to finally obtain NaNbO3 that is free of Nb2O5. This is in contrast to the 72 h needed to obtain NaTaO3 (Fig. 1b).


3

Fig. 1. XRD patterns of Na2CO3–M2O5 (M= V, Nb, Ta) mixtures after different milling times (Notations: ○…Na2CO3 (PDF 19-1130), V…V2O5 (PDF 77-2418), T…Ta2O5 (PDF 79-1375), N…Nb2O5 (PDF 71-0336), ●…NaVO3 (PDF 70-1015), ▲…NaTaO3 (PDF 73-0878), ■…NaNbO3 (PDF 33-1270), F…polymeric foil).

In order to analyze the XRD patterns quantitatively, we determined the amount of crystalline phases using the Rietveld refinement method. The fractions of NaMO3 (M = V, Nb, Ta), which are given according to all crystalline phases in respective mixtures, including Na2CO3 and M2O5 (M = V, Nb, Ta), as a function of milling time are shown in Fig. 2. According to the quantitative XRD data, the rate of formation of the final oxides is in the order NaVO3 N NaTaO3 N NaNbO3, where the vanadate is formed much faster in comparison with the tantalate and niobate. The results lead to some key questions: i) what governs the kinetics of these complex reactions and ii) how do the transition-metal oxides affect the decomposition of the carbonate? In order to provide answers, we made a careful analysis of the reaction mechanism, which is described in the following section.

3.2. Mechanochemical formation and decomposition of the carbonato complex In order to examine the changes in the symmetry of the carbonate ions due to their mechanochemical interaction with transition-metal oxides, we used infrared spectroscopy analysis. Fig. 3 shows the IR spectra of the mixtures after various milling times. The main absorption band in all the initial Na2CO3–M2O5 (M = V, Nb, Ta) mixtures (denoted as 0 h) belongs to the asymmetrical C–O stretching vibration of the CO2– 3 ions present in the Na2CO3 (ν3) [26]. In the case of the reaction between Na2CO3 and Nb2O5 (Fig. 3c), considerable changes can be observed in the IR spectra as the milling time increases: i) the ν3 vibration of the CO2– 3 ions shifts

Fig. 2. Fraction of NaMO3 (M = V, Nb, Ta) as a function of milling time. The lines are drawn as a guide for the eye.

progressively to higher wavenumbers and decreases in intensity until it completely disappears after 48 h of milling, ii) the ν3 vibration is gradually replaced by new absorption bands appearing at 1610, 1530 and 1340 cm− 1, and iii) a new band arises during milling, located at 1060 cm− 1, which belongs to the symmetrical C–O stretching vibration of the CO2– 3 ions (ν1) [27,28]. The ν1 vibration is infrared inactive for the free CO2– 3 ion and also for the Na2CO3 (see initial mixtures denoted as 0 h) as the influence of Na+ cations on the CO2– 3 symmetry is small [29,30]. The splitting of the ν3 vibration, being doubly degenerate, coupled with the activation of ν1 is characteristic for the lowering of the symmetry of the CO2– 3 ions. As was also reported in our previous studies [10,15], the lowered symmetry is a consequence of the mechanochemically driven coordination of CO2– 3 ions with niobium, i.e., formation of new bonds, which results into a carbonato complex as an intermediate phase. These characteristic changes in the IR spectra, related to coordination, are well documented in the literature for a variety of carbonato complexes with different central cations [27,28,31]. Thus, the formation of the carbonato complex in the Na2CO3–Nb2O5 system was also confirmed for the milling conditions applied in this study. The mechanism of incipient mechanochemical interaction between the Na2CO3 and Nb2O5 is general since it is unaffected by the ball-impact energy as the formation of the carbonato complex was observed over a wide range of ball-impact energies, i.e., at 15 mJ [10], 93 mJ (this work) and 370 mJ [11]. Complexation was also observed during milling the K2CO3–Nb2O5 mixture; in addition to the lowered CO2– 3 symmetry, the complex was also identified by new characteristic Nb\O bonds using Raman spectroscopy [15]. Like in the case of the Na2CO3–Nb2O5 mixture (Fig. 3c), the characteristic splitting of ν3 and the activation of the ν1 vibration of the CO2– 3 ions, related to complexation, also occur during the milling of Na2CO3 and Ta2O5 (Fig. 3b). The ν3 vibration splits into three bands at 1645, 1550 and 1340 cm− 1, while the ν1 vibration appears at 1060 cm− 1. The main differences in comparison with the Na2CO3–Nb2O5 system are the intensities and positions of the bands related to the splitting of the doubly degenerate ν3 vibration. We associate this difference with the presence of Ta5+ as the central ion in the carbonato complex, which exhibits different characteristic properties (such as polarizing power) than Nb5+ (see next section for detailed discussion). The V2O5 is also prone to react with Na2CO3, resulting in the formation of the carbonato complex, as evidenced by the splitting of −1 the ν3(CO2– , 3 ) vibration into bands at 1615, 1350 and 1290 cm which can be seen after 4 h of milling (Fig. 3a). Note that the same

4


Fig. 3. IR spectra of Na2CO3–M2O5 (M = V, Nb, Ta) mixtures after different milling times.

shift and intensity decrease of the ν3 vibration (see spectra from 0 to 4 h in Fig. 3a), indicative of the gradual change in the symmetry of the CO2– 3 ions, is observed as in the Na2CO3–Nb2O5 (Fig. 3c) and Na2CO3– Ta2O5 (Fig. 3b) mixtures. This characteristic splitting appears to occur faster in the Na2CO3–V2O5 mixture as the ν3 vibration almost disappears after 4 h of milling, while at least 16 h are needed in the case of Nb5+ and Ta5+ as the central cations. Since the band at 1290 cm− 1 (Figs. 3a, see 4 h) appears narrower than those observed in other systems, we verified its origin by performing quenching experiments. On the basis of the TG analysis, which gave us the temperature range of the carbonate decomposition in the 4-hours-milled sample, we were able to release CO2 selectively from this sample by quenching it in air at selected temperatures. The results of the IR spectroscopy performed on quenched samples (not shown) confirmed that the split ν3(CO2– 3 ) bands at 1615, 1350 and 1290 cm− 1 (Figs. 3a, see 4 h) correspond to the residual CO2– 3 ions present in the sample. The absorption band at 1025 cm− 1 from Fig. 3a (0 h) belongs to the stretching vibration of the double vanadyl V_O bonds present in V2O5[32]. The decrease in the intensity of this band with increasing milling time until its disappearance after 16 h of milling confirms the

reaction of V2O5 with Na2CO3. The characteristic activation of the ν1 (CO2– 3 ) vibration, as it was observed in the cases of Nb and Ta, could not be detected, most probably due to overlapping with the V_O absorption band at 1025 cm− 1. Note that in all three systems no XRD peaks were detected, at any stage of the reactions, that could be assigned to a carbonate compound formed between the Na2CO3 and the transition-metal oxides (Fig. 1). In addition, transmission electron microscopy analysis on the Na2CO3–Nb2O5 high-energy milled powder mixture (not shown) revealed the existence of an amorphous phase, similarly like it was observed in the K2CO3–Nb2O5[15] and Na2CO3–K2CO3–Li2C2O4– Nb2O5–Ta2O5[13] mixtures. The results, therefore, confirm the absence of a long-range periodic structure of the newly formed carbonato complex, in line with observations in other systems [10,13,15]. The decomposition of Na2CO3 (for correctness, we subsequently refer to this as the carbonate phase since the CO2– 3 ions coordinate during milling) upon annealing mechanochemically treated mixtures was followed by a thermogravimetric analysis. The TG and DTG (derivative thermogravimetric) curves of the Na2CO3–M2O5 (M = V, Nb, Ta) mixtures after various milling periods

Fig. 4. TG and DTG curves of Na2CO3–V2O5 (a, d), Na2CO3–Ta2O5 (b, e) and Na2CO3–Nb2O5 (c, f) mixtures after different milling times. The temperature ranges of the H2O and CO2 release, identified using EGA, are marked on the DTG curves.


are shown in Fig. 4. The Na2CO3–V2O5 mixture milled for 1 h loses its mass in several steps (Fig. 4a and d, 1 h). The loss of 1.6% between room temperature and 150 °C is due to H2O, which was adsorbed during the milling, whereas CO2 is released in the range 150–500 °C as evidenced by four broad DTG peaks at 180 °C, 270 °C, 310 °C and 425 °C (Fig. 4d, 1 h). The total CO2 loss in this temperature range, related to the decomposition of the carbonate, is 7.0%, which corresponds to 53% of the initial amount of Na2CO3. Thus, the carbonate partially decomposes during 1 h of milling; its presence in the mixture after 1 h is also confirmed by the ν3(CO2– 3 ) IR band (Fig. 3a, 1 h). Further milling to 4 h does not bring about any change in the H2O content (Fig. 4a, 4 h). In contrast, the CO2 loss decreases remarkably; in fact, only a small DTG peak, related to the decomposition of the rest of the carbonate in the mixture, could be noted at around 460 °C (Fig. 4d, 4 h). The presence of CO2– 3 after 4 h of milling was also identified by the split ν3(CO2– 3 ) IR vibration (see Fig. 3a, 4 h). Finally, the 48-h-milled sample loses 1.2% of adsorbed H2O, while nearly no loss of CO2 could be detected (Fig. 4a and d, 48 h). The absence of IR absorption bands corresponding to CO2– 3 vibrations after 48 h of milling (see Fig. 3a, 48 h) supports the results of the TG analysis and confirms the nearly complete decomposition of the carbonate after milling for 48 h. In the case of the Na2CO3–Ta2O5, the 16-hours-milled mixture decomposed by releasing 3.5% of H2O in the temperature range 25– 220 °C and 6.4% of CO2 in two steps between 220 °C and 600 °C (Fig. 4b and e, 16 h). With increasing the milling time up to 72 h some shifts in the H2O DTG peak and small variations in H2O content (within 1%) could be seen; however, more remarkable changes can be observed in the release of CO2. By increasing the milling time from 16 h to 48 h, the two DTG peaks at 330 °C and 470 °C related to CO2 loss (Fig. 4e, 16 h) merge into one asymmetric peak, appearing at 460 °C (Fig. 4e, 48 h). This coincides with the disappearance of the ν3 (CO2– 3 ) IR band (Fig. 3b, 48 h), which suggests the complete coordination of the initial CO2– 3 ions. Therefore, we infer that the observed change in the carbonate decomposition from 16 to 48 h of milling (Fig. 4e) could be related to the coordination of the CO2– 3 ions. The total loss of CO2 after 48 h decreased to 4.0%. Finally, after 72 h the CO2 DTG peak broadens, shifts to lower temperature (440 °C) and decreases in intensity (Fig. 4e, 72 h). In agreement with these observations, also some shifts in the ν3(CO2– 3 ) split bands and ν1 band can be noted from 48 to 72 h of milling (Fig. 3b). The total loss of CO2 after this milling period (72 h) amounts to 2.3%, which corresponds to 29% of residual carbonate. An additional TG analysis (not shown) showed that this amount does not change with further milling. Like in the cases involving V2O5 and Ta2O5, the Na2CO3–Nb2O5 mixture also adsorbs H2O during milling, which is released in the range between room temperature and 250 °C (Fig. 4c and f). The H2O content varies between 4% and 7.5%, without showing a systematic trend with increasing milling time. Note that for the Na2CO3–Nb2O5 mixture, qualitatively, similar changes occur in the carbonate decomposition upon annealing the milled samples as in the case of the Na2CO3–Ta2O5 mixture: the two DTG peaks at 350 °C and 400 °C (Fig. 4f, 16 h) eventually merge into one, which then broadens, shifts to lower temperature and decreases in intensity with further milling up to 150 h (Fig. 4f, from 16 h to 150 h). A total of 4.2% of CO2 is released after 150 h of milling, corresponding to 39% of residual carbonate. An important difference between the three examined reaction systems can be found in the fraction of residual carbonate after prolonged milling, where “steady-state” milling conditions are expected. This condition refers to a period of milling during which negligible or no detectable changes are observed with increasing milling time; Lin et al. [33] and Iguchi et al. [34], for example, adopted expressions such as “mechanochemical equilibrium” and “stationary state”. While, for example, the IR absorption bands of the carbonato

5

complex can be clearly observed even after 72 and 150 h of milling for the cases of Ta2O5 and Nb2O5, respectively (Fig. 3b and c), these bands completely disappear after just 16 h in the case of the reaction between Na2CO3 and V2O5 (Fig. 3a). These observations are in agreement with the results of the TG analyses (Fig. 4) and suggest that the carbonato complex decomposes nearly completely in the case of the reaction where V2O5 is involved, while this is not the case for Ta2O5 and Nb2O5. To have a quantitative picture, we plot the fraction of residual carbonate, determined by the loss of CO2 from the TG curves, after various milling times for the three mixtures (Fig. 5). As expected, in all the cases the carbonate fraction decreases with increasing milling time; this is related to the mechanochemically driven carbonate decomposition. However, a substantial difference exists in the decomposition rate between the three systems: the fastest is in the case of V2O5, followed by Ta2O5 and Nb2O5. Note also that the carbonate fraction reaches a plateau after prolonged milling (denoted here as the “steady-state” condition), which depends on the type of transition-metal oxides participating in the reaction. As anticipated, the carbonato complex decomposes nearly completely in the case of vanadium, while 29% and 39% remains in the mixture in the case of Ta and Nb, respectively (Fig. 5). At this point it should be also noted that the complex is amorphous and could not be analyzed using Rietveld analysis (Fig.2); instead, we were able to follow its formation and decomposition using IR and TG analyses (Figs. 3,4 and 5). From the presented results we can infer that a common mechanism, characterized by the formation of an amorphous carbonato complex, link the reactions between Na2CO3 and M2O5 (M = V, Nb, Ta); however, large differences can be found in the rate of the formation and decomposition of this carbonato complex and, consequently, in the crystallization of the final binary compounds. The sequence of the rates of these reactions can be interpreted by considering the acid–base properties of the reagents involved and will be discussed in the following section.

3.3. Influence of acid–base potential between reagents on the reaction rate In their extensive work on the mechanochemical reactions involving hydroxide–oxide mixtures, Senna and co-workers [2–5,35] showed that the mechanism in these mixtures is governed by an acid–base reaction between different hydroxyl groups on the solid surface. The driving force for these reactions is the acid–base potential (the difference in the acid–base properties) between an acidic and basic surface \OH group, which is determined by the type of metal on which it is bound, and therefore, on the strength of the M\OH bond (M denotes the metal). In the case of the reaction between Mg(OH)2 and SiO2, for example, they showed that the mechanochemical dehydration (milling-induced H2O release) occurs only during milling of the mixture and not when Mg(OH)2

Fig. 5. Fraction of carbonate, determined with TG analysis, as a function of milling time in Na2CO3–M2O5 (M = V, Nb, Ta) mixtures. The lines are drawn as a guide for the eye.

6

T. Rojac et al. / Solid State Ionics 190 (2011) 1–7 Table 1 Electronegativity values for V5+, Nb5+ and Ta5+. Cation

XZ⁎

XZ/CN⁎

V5+ Ta5+ Nb5+

2.02 1.88 1.77

0.40 0.29 0.27

⁎ XZ and CN denote electronegativity defined by Zhang [36] and coordination number, respectively.

is milled alone [5]. The reason lies in the acid–base potential between the two reagents, which favors a fast dehydration and amorphization of the mixture, resulting in the formation of heterobridging M\O\M’ bonds. Thus, as shown experimentally, a larger potential, like in the case of Ca (OH)2–SiO2 mixture, brings a faster interaction [2]. The acid–base reaction mechanism is not confined to the hydroxyl groups. Thermodynamic calculations show that a correlation exists between the Gibbs free energies of a variety of reactions between oxide compounds and the acid–base potential between the participating oxide reagents for two-component systems: the larger the potential, the more negative the value of the Gibbs energy and, thus, the faster and more complete the reaction [35]. As explained in details in the book of Avvakumov et al. [35], the acid– base properties of a given compound can be estimated quantitatively using the concept of average orbital electronegativity, where electronegativity scales for atoms can be used. However, to fully consider the acid–base properties of oxide compounds, one should take into account that the acidity of a cation, incorporated into a certain oxide compound, depends on the oxidation state and the coordination number. The influence of these parameters on the acid–base properties of cations can be considered by introducing the electronegativity of a cation, divided by the coordination number. This defines the cation–ligand force per one bond in a coordination polyhedron; the larger the force, the larger the acidity of the cation (i.e., the stronger is the ability to attract electron pairs forming covalent bonds). To address the acid–base properties of the transition-metal cations used in this work, we adopt the electronegativity scale for cations derived by Zhang [36]. Table 1 shows the electronegativities (Xz) for V5+, Ta5+ and Nb5+. The ratio Xz/CN (where CN refers to the coordination number of the cation), taken as being indicative of the acidity of the cations in their respective oxides, is the highest for V5+, followed by Ta 5+ and Nb5+. This order correlates with our experimental results; in fact, the reaction rate follows the same order, i.e., Na2CO3 + V2O5 N Na2CO3 + Ta2O5 N Na2CO3 + Nb2O5, which means that the greater is the acidity of the cation involved, the faster is the reaction, including the formation and decomposition of the carbonato complex (Figs. 3 and 5), and the crystallization of the final oxides (Fig. 2). The agreement between the reaction rate sequence and the cation acidity or acid–base potential (where Na2CO3 is taken as basic and transition-metal oxides as acidic compounds) suggests that the mechanochemical reactions studied here suit the concept of an acid–base interaction mechanism. This observation is also supported by the measurements of surface catalytic activity, where it was shown that the surface of the V2O5 is significantly more reactive, in terms of surface acidic sites, than that of the Nb2O5 and Ta2O5, the latter two being very similar in their surface acidic catalytic activity [37]. A similar correlation between the acid–base properties and the mechanochemical reaction rate was found for the reactions between CaO and Al2O3, SiO2, TiO2, V2O5 or WO3[38]. It is well known that V2O5 can be reduced more easily than Nb2O5 and Ta2O5[37], so one should not neglect possible effects of reduction on the course of the reaction. In principle, if present in a larger amount, one could anticipate that reduced V5+ ions might slow down the reaction due to their lower acidity [35,36]. However, the present study did not show any clear evidence of V2O5 reduction during

Fig. 6. Plot of reaction rate constant versus XZ/CN for the reactions between Na2CO3 and M2O5 (M = V, Nb, Ta).

milling at any stage of the reaction, at least not to such a level that could be detected by XRD analysis. Somewhat similar observation was made by other authors [39,40], who showed that the original V2O5 structure was preserved after high-energy milling with only a minor amount of V4+ detected. Even if the rate of the three examined reactions conforms the acid– base reaction concept, another important factor, that could influence the course of the reaction and should not be neglected, is adsorption of water during milling on the surface of the particles. To clarify this point, further milling experiments in controlled atmosphere are going to be necessary and will be reported separately. In order to directly verify the possibility of having other influences, we plot on Fig. 6 the reaction rate constant versus Xz/CN. The reaction rate constant was obtained by fitting the curves from Fig. 1 with a kinetic model proposed for mechanochemical transformations in binary mixtures [41]. A linear relationship would be expected if the reaction rate will be largely dominated by the cation acidity. While the sequence of the reaction rate constants, i.e., V N Ta N Nb, agrees with the acid–base reaction mechanism, the non-linear relationship between the rate constant and Xz/CN suggests that, in addition to the cation acidity, probably other factors influence the reaction rate. We showed that after reaching a specific milling time (“steadystate” milling condition), the fraction of the residual carbonate does not change any longer if further milling is applied (Fig. 5). These carbonate fractions correlate with the acidity of the cation: while 29% and 39% of the carbonate remains in the mixture in the case of Ta2O5 and Nb2O5, respectively, only 0.5% is present in the case of V2O5 (Table 2). Note that this difference is consistent with the much larger acidity of V5+ as compared to Ta5+ or Nb5+ (XZ/CN values in Table 1). The results seem reasonable if one considers the relation that was found between the acid–base potential and the reaction Gibbs free energy; however, insufficient thermodynamic data prevent us from making further steps in this direction. In general, a larger amount of residual (undecomposed) carbonate might be expected, even after a prolonged mechanochemical treatment of a carbonate–oxide mixture, if the reaction is characterized by a low acid–base potential. This observation, which is treated insufficiently in the literature, might be important in the design of the milling conditions for a given mechanochemical reaction. In all the studied cases, the intermediate stage of the mechanochemical reactions was identified as an amorphous carbonato

Table 2 Fraction of residual carbonate in Na2CO3–M2O5 (M = V, Nb, Ta) mixtures after prolonged milling (during “steady-state” milling conditions). Mixture

Carbonate fraction (%)

Na2CO3–V2O5 Na2CO3–Ta2O5 Na2CO3–Nb2O5

0.5 29 39

T. Rojac et al. / Solid State Ionics 190 (2011) 1–7 Table 3 Maximum splitting of ν3 vibration of carbonate ion (Δν3) in Na2CO3–M2O5 (M = V, Nb, Ta) mixtures. Mixture

Max Δν3 splitting (cm− 1)

Na2CO3–V2O5 Na2CO3–Ta2O5 Na2CO3–Nb2O5

325 305 270

complex, in which the CO2– 3 ions, as a result of the mechanochemical interaction between Na2CO3 and M2O5 (M = V, Nb, Ta), coordinate to the transition-metal cations. Coordination causes an electron rearrangement in the CO2– 3 ion as one or more of its oxygens, depending on whether the coordination is monodentate, didentate, bridged, etc., become bound to the metal cation. Calculations, which were later also confirmed experimentally, showed that, for a given type of coordination, this CO2– 3 polarization is more pronounced if the acidity of the central cation is high as it can attract electrons more strongly. The extent of the CO2– 3 polarization can be estimated from the splitting of the ν3 IR vibration of the carbonate ions (Δν3) [42–44]. In our case, the largest Δν3 splitting is observed for the case of vanadium, i.e., 325 cm− 1, compared to 305 cm− 1 and 270 cm− 1 for tantalum and niobium, respectively (Table 3). As expected, the extent of the CO2– polarization increases with increasing cation acidity 3 (compare Tables 1 and 3). This is in agreement with the experimental results of Jolivet et al. [44], who found a relation between the Δν3 splitting and the polarizing power of the central ion for various carbonato complexes that have a didentate coordination. However, even if a correlation was found, it should be interpreted carefully since in addition to the polarizing ability of the central ion, the Δν3 splitting depends strongly on other factors, such as the type of coordination (mono-, didentate, etc.) and the incorporation of water molecules, which might differ between the three examined systems. This question remains open for further studies. 4. Conclusions We studied the mechanochemical reactions between Na2CO3 and transition-metal oxides (V2O5, Nb2O5 and Ta2O5), in particular the reaction mechanism and the reaction rate. We found that all the studied reactions proceed following the same mechanism: i) in the first stage an amorphous carbonato complex is formed, in which the CO2– 3 ions, present initially in Na2CO3, coordinate to the central cations, i.e., V5+, Nb5+ and Ta5+, as a result of the mechanochemical interaction between the carbonate and the oxide, ii) upon further milling the complex decomposes and, finally, iii) the binary compounds, i.e., NaVO3, NaNbO3 and NaTaO3, crystallize. The overall reaction rate depends strongly on the type of transition-metal oxide involved and it appears in the order Na2CO3 + V2O5 N Na2CO3 + Ta2O5 N Na2CO3 + Nb2O5. This reaction-rate sequence can be interpreted by considering the acid–base properties of the participating reagents; the larger the acid–base potential between the interacting compounds, the faster the mechanochemical reaction.

7

Acknowledgements The work was carried out as part of the Research Program “Electronic Ceramics, Nano, 2D and 3D Structures” P2-0105 and Postdoctoral project “Mechanochemical Synthesis of Complex Ceramic Oxides” Z2-1195 (Slovenian Research Agency). Ms. Tanja Urh is acknowledged for the preparation of some of the samples. Mr. Edi Krajnc is thanked for the XRD diffraction analyses; Ms. Jana Cilenšek for thermal analyses; and Dr. Bojan Kozlevčar for the FT-IR analysis. We would like to give special thanks to Prof. Dr. Olivier Masson for valuable discussions on the quantitative XRD phase analysis.

References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40] [41] [42] [43] [44]

L.B. Kong, T.S. Zhang, J. Ma, F. Boey, Prog. Mater. Sci. 53 (2008) 84. J. Liao, M. Senna, Solid State Ionics 66 (1993) 313. T. Watanabe, J. Liao, M. Senna, J. Solid State Chem. 115 (1995) 390. T. Watanabe, T. Isobe, M. Senna, J. Solid State Chem. 122 (1996) 74. J. Liao, M. Senna, Thermochim. Acta 210 (1992) 89. J. Wang, J.M. Xue, D.M. Wan, B.K. Gan, J. Solid State Chem. 154 (2000) 321. J. Wang, J.M. Xue, D.M. Wan, Solid State Ionics 127 (2000) 169. D. Kuscer, J. Holc, M. Kosec, J. Am. Ceram. Soc. 89 (2006) 3081. D. Kuscer, J. Holc, M. Kosec, J. Am. Ceram. Soc. 90 (2007) 29. T. Rojac, M. Kosec, P. Šegedin, B. Malič, J. Holc, Solid State Ionics 177 (2006) 2987. T. Rojac, M. Kosec, B. Malič, J. Holc, J. Am. Ceram. Soc. 91 (2008) 1559. T. Rojac, A. Benčan, H. Uršič, B. Malič, M. Kosec, J. Am. Ceram. Soc. 91 (2008) 3789. T. Rojac, A. Benčan, M. Kosec, J. Am. Ceram. Soc. 93 (2010) 1619. T. Rojac, M. Kosec, B. Malič, J. Holc, J. Eur. Ceram. Soc. 26 (2006) 3711. T. Rojac, M. Kosec, M. Polomska, B. Hilczer, P. Šegedin, A. Benčan, J. Eur. Ceram. Soc. 29 (2009) 2999. V2O5, Inorganic crystal structure database (ICSD) N°24042. Nb2O5, Inorganic crystal structure database (ICSD) N°1840. Ta2O5, Inorganic crystal structure database (ICSD) N°9112. NaVO3, Inorganic crystal structure database (ICSD) N°2103. NaNbO3, Inorganic crystal structure database (ICSD) N°97669. NaTaO3, Inorganic crystal structure database (ICSD) N°23319. A. Bystrom, K.A. Wilhelmi, O. Brotzen, Acta Chem. Scand. 4 (1950) 1119. R. Enjalbert, J. Galy, Acta Crystallogr. C42 (1986) 1467. V.K. Kato, S. Tamura, Acta Crystallogr. B31 (1975) 673. N.C. Stephenson, R.S. Roth, Acta Crystallogr. B27 (1971) 1037. M.J. Harris, E.K.H. Salje, J. Phys.-Condens. Mat. 4 (1992) 4399. G. Busca, V. Lorenzelli, Mater. Chem. 7 (1982) 89. M. Gatehouse, S.E. Livingstone, R.S. Nyholm, J. Chem. Soc. (1958) 3137. K. Buijs, C.J. Schutte, Spectrochim. Acta 17 (1961) 927. M.H. Brooker, J.B. Hates, J. Chem. Phys. 54 (1971) 4788. J.P. Jolivet, Y. Thomas, B.J. Taravel, Molec. Struct. 60 (1980) 93. Y. Kera, K.J. Hirota, Phys. Chem. 78 (1969) 3973. I.J. Lin, S. Nadiv, Mater. Sci. Eng. 39 (1979) 193. Y. Iguchi, M. Senna, Powder Technol. 43 (1985) 155. E. Avvakumov, M. Senna, N. Kosova, Soft Mechanochemical Synthesis, Kluwer Academic Publishers, Boston, USA, 2001. Y. Zhang, Inorg. Chem. 21 (1982) 3886. I.E. Wachs, Y. Chen, J.M. Jehng, L.E. Briand, T. Tanaka, Catalysis Today 78 (2003) 13. E.G. Avvakumov, E.T. Devyatkina, N.V. Kosova, J. Solid State Chem. 113 (1994) 379. V.A. Zazhigalow, J. Haber, J. Stoch, A.I. Kharlamov, L.V. Bogutskaya, I.V. Bacherikova, A. Kowal, Solid State Ionics 101–130 (1997) 1257. S.V. Khalameida, V.A. Zazhigalow, K. Vechorek-Tsyurova, N.S. Litvin, Russ. J. Inorg. Chem. 54 (2009) 368. G. Cocco, F. Delogu, L. Schiffini, J. Mater. Synth. Process. 8 (2000) 167. H. Brintzinger, R.E. Hester, Inorg. Chem. 5 (1966) 980. R.E. Hester, W.E.L. Grossman, Inorg. Chem. 5 (1966) 1308. J.P. Jolivet, Y. Thomas, B.J. Taravel, J. Molec. Struct. 79 (1982) 403.