Local Coordination about La3+ in Molten LaCl3 and Its Mixtures with

Local Coordination about La3+ in Molten LaCl3 and Its Mixtures with Alkali Chlorides ... of different metals from a complex mixture(4) or allow contro...
1 downloads 0 Views 794KB Size
Download PDF
4664

J. Phys. Chem. A 2010, 114, 4664–4671

Local Coordination about La3+ in Molten LaCl3 and Its Mixtures with Alkali Chlorides Yoshihiro Okamoto* and Shinichi Suzuki Quantum Beam Science Directorate, Japan Atomic Energy Agency, Shirakata-Shirane 2-4, Tokai-mura, Ibaraki 319-1195, Japan

Hideaki Shiwaku, Atsushi Ikeda-Ohno, and Tsuyoshi Yaita Quantum Beam Science Directorate, Japan Atomic Energy Agency, Kouto 1-1-1, Sayo-cho, Hyogo 679-5148, Japan

Paul A. Madden The Queen’s College, UniVersity of Oxford, High Street, Oxford OX1 4AW, United Kingdom ReceiVed: NoVember 8, 2009; ReVised Manuscript ReceiVed: February 22, 2010

The local structure around the La3+ ions in molten LaCl3 and its mixtures with alkali and alkaline earth chlorides has been investigated by using extended X-ray absorption fine structure (XAFS) and molecular dynamics (MD) techniques. Such mixtures, which are of current technological interest, are known to be thermodynamically nonideal, and there has been a good deal of work to understand the structural effects factors that contribute to the nonideality. New experimental methods allow observations at the La K-absorption edge at the high temperatures of interest, and the ability of the technique to obtain reliable information even at very low La3+ concentrations in multicomponent mixtures is demonstrated. Both the mean La3+-Clinterionic separation and the mean La3+ coordination number are found to decrease as the concentration of La3+ in the mixture decreases. The rate of decrease depends on the identity of the alkali and alkaline earth cations present in the mixtures in a way that parallels the degree of nonideality of the different systems; it is greatest for those alkali cations that coordinate Cl- weakly. In dilute mixtures with such cations La3+ is able to adopt a very stable octahedral coordination geometry but this is inhibited by the presence of more strongly coordinating cations like Li+ and Mg2+. 1. Introduction It has been known for many years that mixtures of trivalent metal chlorides with alkali chlorides show significant departures from ideality that depend on the nature of the alkali cation.1,2 The observation is relevant to numerous technologies, such as the electrochemical reprocessing of nuclear fuels,3 since the thermodynamic activity of the metal ion may, in principle, be tuned by using appropriate mixtures of alkalis and this could allow selective extraction of different metals from a complex mixture4 or allow control of corrosion.5 As such, the excess thermodynamic properties of the mixtures have been studied extensively.6 In the case of the La3+ ion in LaCl3-ACl mixtures with different alkalis (A) it is found that the departures from ideality are most pronounced for the larger alkalis, like Cs, and smallest when A ) Li, and this trend is found for many trivalent ions. The melts of the halides of the larger alkalis are more “basic”5 than those of the smaller cations in the sense that they more readily release the halide anion to bind with other species. At a more fundamental level, these systems are ideal to examine the basic chemistry of ionic solvation and coordination since a large number of metals form a trivalent metal cation so that a very wide range of ionic radii may be spanned in experimental studies. Similar principles should govern the coordination behavior of other types of mixture, such as molten fluorides7 but also solutions in molecular solvents. * Corresponding author. E-mail: [email protected]. Phone: +81-29-284-3928. Fax: +81-29-284-3747.

Relating the observed thermodynamic behavior to the local coordination structures that form around the metal cation has motivated many experimental and theoretical studies. In particular, it has inspired a comprehensive examination of these structures by Raman spectroscopy.8,9 This has shown a remarkable isomorphous behavior10 of the Raman spectra of a wide range of systems in mixtures of varying concentration with different alkali chlorides. This finding has been interpreted as consequence of a conservation of an octahedral coordination for trivalent ions with a broad range of radii from the pure MCl3 melt to dilute solution.10,11 Neutron12-15 and X-ray diffraction studies16-18 may be applied to the pure melts and have indicated that the larger cations (including La3+) have more than six Clneighbors, approaching the nine neighbors of the LaCl3 crystal structure, whereas intermediate-sized cations, like Y3+ or Dy3+, which form six-coordinate chloride crystals, have coordination numbers close to six in the melt. Unfortunately, the diffraction methods cannot readily be applied to the MCl3-ACl mixtures, so as to examine the behavior on dilution from this perspective, as too many partial structure factors contribute to the total scattering to allow the signal from the M-Cl coordination structures to be isolated. The MCl3-ACl systems have been extensively studied by computer simulation using a common family of polarizable interaction potentials that were tuned to give good agreement with the neutron data for the pure melts.19-22 They have subsequently been shown to predict other measurable properties accurately. In the simulations, the La3+ cation is almost 8-fold

10.1021/jp910637p  2010 American Chemical Society Published on Web 03/15/2010

Coordination about La3+ in Molten LaCl3 coordinated in the pure melt but the coordination number falls to a limit of six as the degree of dilution of LaCl3 into ACl increases in a way that depends on the identity of the alkali cation (A).22 In a 25% solution in CsCl the average coordination of the La3+ ion is already close to 6 but in an NaCl solution at the same concentration it remains high. For the intermediatesized cations (like Y3+ and Dy3+) the coordination number is close to six at all concentrations, in agreement with the Raman findings. In an attempt to resolve the disagreement between the simulation (and diffraction) and the Raman studies for the larger cations, the simulations were used to calculate Raman spectra for various models of the fluctuating polarizability of the mixtures.22 The simulated spectra resembled the experimental ones quite closely and evolved in the same way on dilution. The simulation study indicated, somewhat surprisingly, that the change in average coordination number from eight (pure melt) to six (dilute solution in CsCl) did not strongly affect the overall structure of the Raman spectrum. However, the simulated Raman spectra certainly did not reproduce the experimental spectra in fine detail23 and cannot be taken in isolation to overturn the long-standing interpretation which has been given to them:10 a study using high temperature (HT) NMR methods would be useful.24,25 Rollet et al.7 successfully obtained information on the local structure of molten LaF3-alkali fluoride mixtures from the chemical shifts of 139La and 19F in the HT NMR studies. In this work we describe the application of recently developed high-temperature, high-energy XAFS techniques26,27 to mixtures of LaCl3 with an LiCl-KCl eutectic mixture and the effect of adding MgCl2. XAFS has the advantage over the diffraction probes of structure that it is sensitive to the local structure around the atoms of a particular element and therefore may be applied to multicomponent mixtures as well as pure melts,28,29 even if the element of interest is in low concentrations. XAFS can yield very accurate values for nearest-neighbor separations but it is less reliable for assigning absolute Values of coordination numbers.30 Since the shape of partial radial distribution function beyond the first peak is not unambiguously determined from the experiment alone, some form must be assumed and input into the analysis. On the other hand, as we will show below, the XAFS signal is very sensitive to a change of coordination number and in the particular case of LaCl3 does show a change as the degree of dilution is increased. The choice of the LiCl-KCl eutectic mixture and of MgCl2 as an additive is designed to illustrate the potential relevance of these methods to technologies of the type mentioned above. Eutectic mixtures are often used in molten salt technologies as they enable working at lower temperature, which has an important impact on corrosion inter alia. A material like MgCl2, which is more “acidic” than an alkali chloride, is often added to alter the activity of the species of interest (La3+ here) and it is of considerable interest to understand the origin of this effect on the atomic scale. In addition to presenting the XAFS results per se, we also illustrate the direct calculation of the XAFS spectrum from a computer simulation31 in a way that involves no additional assumptions about structure. The method may be used to validate the simulation model but also, since the structural information presented by XAFS is incomplete, gives a way of providing a more complete picture of the structure if the agreement between simulation and experiment is satisfactory. In this context we will use the same simulation models as used in the study of diffraction and Raman spectroscopy of the LaCl3, which were referred to above.22

J. Phys. Chem. A, Vol. 114, No. 13, 2010 4665 We previously reported32 the Y K-edge (E0 ) 17.08 keV) XAFS spectra of a 15% YCl3 mixture with LiCl-KCl and confirmed that the octahedral coordination structure (YCl6)3in pure YCl3 was stabilized by mixing with the LiCl-KCl eutectic. More recently, we have developed high-energy (over 35 keV) XAFS for application to molten lanthanide chlorides at SPring-8; the K-absorption edge energy of lanthanide elements ranges from 39 to 63 keV, much higher than that of Y. We have already reported a study of pure molten LaCl3 using the La K-edge (38.93 keV).26 The use of a high-energy beam has another advantage. In XAFS studies of high-temperature molten chlorides, a vessel such as glass or metal is usually used to separate the sample from air and moisture, since the chloride sample is generally hygroscopic, and this creates a problem for X-ray beam absorption by the cell. The absorption effect decreases as X-ray energy increases. We have used a quartz cell (typically 0.5 mm thickness)32 and for the Y K-edge, for example, only 15% of the incident X-ray beam passes through the blank cell. At the La K-edge, the transmission of the X-ray beam reaches 85% for the La K-edge; this is a big advantage for obtaining high quality XAFS data and working with dilute mixtures. The high-energy XAFS measurement is especially suitable for the XAFS measurement of a heavy element absorber (lanthanide elements) mixing with a light element solvent (light alkali chlorides such as LiCl), which applies to our present study. A simple calculation suggests that obtaining good quality La K-edge XAFS data even for a dilute mixture of 5% LaCl3(LiCl-KCl eutectic) is quite possible using the quartz cell. 2. Experimental Section (1) XAFS Experiment. The high-temperature XAFS measurements were performed in transmission geometry at the BL11XU beamline at SPring-8, Harima, Japan. The incident radiation was monochromatized with liquid nitrogen-cooled double Si(311) crystals. Details of the optics and of the XAFS measurement system in the BL11XU beamline were described in ref 27. The constituent samples of LaCl3 (99.9% purity, Aldrich), the LiCl-KCl eutectic (99.9% purity, Aldrich), and MgCl2 (99.9% purity, Aldrich) were dried at 523 K under reduced pressure for 2 days to remove moisture. These were then combined in the appropriate proportions and melted in a quartz crucible to obtain the mixtures studied. In addition to pure LaCl3, mixtures containing 50, 30, 11, 5, and 1% LaCl3 in the LiCl-KCl eutectic mixture (molar compositions) were prepared as well as a 2% LaCl3-49% (LiCl-KCl eutectic)-49% MgCl2 sample. The samples were sealed off after drying at 573 K under reduced pressure for 1 day in the quartz cell. The XAFS measurement was performed on the La K-edge in an energy range 38.5-40.5 keV using a stepped scan method for 1 s at each energy. We used cells with a melt path of 0.2 mm thickness for the pure LaCl3 melt and the 50% LaCl3 mixture, 1.0 mm thickness for the 30%, 11%, and 5% LaCl3 mixtures, and 5.0 mm thickness for the 1% LaCl3 mixture. The temperatures of the XAFS measurements were 1173 K for the pure LaCl3 melt, 1073 K for the 50% LaCl3 mixture, 1023 K for the 30% LaCl3 mixture, and 773 K for other compositions. The measurement temperature for the mixture melt was reduced below 1073 K, to avoid the LiCl-KCl eutectic’s instability at higher temperatures and to avoid a reaction between LiCl and the quartz cell. The XAFS data were analyzed by using the computer program code WinXAS ver.2.3. developed by Ressler33 and the XAFS simulation code FEFF Ver.8.34 Coordination number Nj,

4666

J. Phys. Chem. A, Vol. 114, No. 13, 2010

Yoshihiro et al.

interionic distance rj and Debye-Waller factor σj2 are obtained by curve fitting in r-space. In the present work, the following equation containing the third and fourth cumulants32 was used in the fitting procedure.

χ(k) )

∑

NjSj(k) Fj(k) exp(-2σj2k2) exp(-rj /λ) ×

j

2 4 exp C4jk4 sin 2krj + φj(k) - C3jk3 /(krj2) 3 3

(

) (

)

(1)

where Nj ) coordination number of ion j around central ion i Sj(k) ) amplitude reduction factor mainly due to many-body effect Fj(k) ) backscattering amplitude for each neighboring atom σj2 ) Debye-Waller factor corresponding to thermal vibration λ ) electron mean free path φij(k) ) total phase shift of experienced by photoelectron rj ) average distance of ion j from the central ion i C3, C4 ) third and fourth cumulants The anharmonic vibration effect expected at high temperature is treated by the cumulant expansion to fourth degree.32,33 This allows for the asymmetry in the shape of the first peak of the radial distribution function. (2) Molecular Dynamics Simulation. The XAFS function k3χ(k) was also predicted by using molecular dynamics (MD) simulation and the XAFS simulation code FEFF Ver.8 to compare with the experimental signals. This gives an alternative way of examining the experimental data to the “conventional” XAFS analysis using WinXAS. Atomic configurations obtained from the MD simulation were directly used as input data in the FEFF8, which allows for a calculation of the energy dependence of the scattering of the emerging photoelectrons by the surrounding Cl- ions around each La3+. The overall XAFS signal is computed as a superposition of the signals calculated from a very large (∼104) set of local configurations extracted from the MD run. The XAFS simulation method is described in ref 31. The MD simulation was performed by using the polarizable ionic model (PIM). The ions carry formal valence charges, and the effects of the induced dipoles20 caused by polarization of the ions are included in the PIM simulation. We had already established18 that simple pair potentials (rigid ionic model, or RIM) did not reproduce the structure and other physical properties of molten LaCl3. On the other hand, the X-ray and neutron diffraction data of molten LaCl3 are in good agreement with the PIM simulation results.18-20 The simulation model of the LaCl3-LiCl-KCl mixture proposed by Brookes et al.35 was used in the present work. They successfully reproduced the diffusion coefficient and shear viscosity of the LaCl3(LiCl-KCl eutectic) mixture melt. Very similar simulations were used in the examination of the Raman spectra of LaCl3-CsCl mixtures described in the Introduction.22 A full description of the potential and a table of parameters is given in the appendix to ref 35. Considerations about the nature of the potential and the parameter determination are given in the references cited in refs 35 and 22. The MD simulations of the pure LaCl3 were carried out by using 2688 ions (672 La3+ and 2016 Cl- ions). Parameters of the basic cell in the MD simulation are listed in Table 1. The system was initially annealed for 105 MD steps at 3000 K and cooled to the prescribed temperature using more than 2 × 105 MD steps. We then collected positional information for 5 × 106 MD steps for subsequent analysis. Mixtures were prepared by randomly placing La3+, K+, Li+, and Cl- ions in the

TABLE 1: Parameters of the Basic Cell in the MD Simulation for Molten LaCl3-LiCl-KCl Mixture Systems mol % LaCl3

T (K)

N(Cl-)

N(La3+)

N(Li+)

N(K+)

100 75 50 30 11 5 1

1200 1200 1073 1023 773 773 773

2016 1920 1792 1656 1478 1408 1360

672 576 448 312 134 64 16

112 264 424 633 712 768

80 184 296 443 504 544

appropriate proportions in a cubic simulation cell and then annealing at 3000 K for 105 MD steps. Typical mixture cells contained more than 3000 ions. The simulations were then cooled to the target temperature over 2 × 105 steps. The simulations were carried out at NPT at zero external pressure. The densities of the ternary mixtures do not appear to have been measured, but we have previously compared the densities of the binary mixtures (LaCl3-KCl, etc.) predicted by the simulations with these potentials with the formulas given in the Janz database36 and found very satisfactory agreement. Plots showing these comparisons are presented in the Supporting Information. 3. Results and Discussion (1) Molten LaCl3-(LiCl-KCl Eutectic) Mixtures. XAFS measurements of the La K-absorption edge were performed on solid and molten LaCl3 and on five mixtures (50%, 30%, 11%, 5%, and 1% LaCl3) with the LiCl-KCl eutectic. Figure 1 shows the absorbance curves of solid and molten 30% and 5% LaCl3 in the LiCl-KCl eutectic mixture. Clear oscillations after the La K-absorption edge jump are apparent despite the small amount of the absorber. The experimental XAFS functions

Figure 1. Raw EXAFS spectra of molten (a) 30% and (b) 5% LaCl3-(LiCl-KCl eutectic) mixtures.

Coordination about La3+ in Molten LaCl3

Figure 2. EXAFS functions k3χ(k) of molten LaCl3-(LiCl-KCl eutectic) mixtures.

k3χ(k) and Fourier transform magnitude functions |FT(k3χ(k))| for molten LaCl3 and the five mixture melts are shown as solid curves in Figures 2 and 3, respectively. In the XAFS functions the period of the XAFS oscillation is seen to shift to higher k as the degree of dilution of LaCl3 increases. In the Fourier transform functions, a sharp isolated peak is observed in each melt, which corresponds to the La3+-Cl- nearest-neighbor correlation. As indicated by the “arrow” in the figure, this firstneighbor separation shortens with increasing dilution in accord with the shifting period of the oscillation in k3χ(k). This shows that the average first La3+-Cl- distance decreases on mixing with the LiCl-KCl eutectic. The results of a direct curve fitting WinXAS analysis in r-space for the first La3+-Cl- correlation are shown as a dashed curve in Figure 3, and the structural parameters obtained in this way are listed in Table 2. The four most significant structural parameter values (the coordination number, the nearest-neighbor distance, the Debye-Waller factor, and the third cumulant) in the fitting analysis are plotted in Figure 4. Two important tendencies can be observed from the plot. First, the highest values of the coordination number and the interionic distance in the pure melt are greater than those of any of the mixture melts. The coordination number and distance of the first La3+-Cl- correlation were ∼8 and 2.87 Å for the pure melt. Second, all the fitting parameters decrease as a consequence of the mixing. Below 10% LaCl3 concentration, the coordination

J. Phys. Chem. A, Vol. 114, No. 13, 2010 4667

Figure 3. EXAFS Fourier transform magnitudes |FT(k3χ(k))| of molten LaCl3-(LiCl-KCl eutectic) mixtures.

TABLE 2: Structural Parameter Obtained from the XAFS Curve Fitting Analysis for Molten LaCl3 and the Mixture Melts with LiCl-KCl Eutectic % LaCl3

S02

N

r (Å)

C3 C4 σ2 (Å2) (104 Å3) (104 Å4)

100 50 30 11 5 1 2a

0.80 0.80 0.80 0.80 0.80 0.80 0.80

7.80 7.19 6.80 6.58 6.27 6.20 6.40

2.869 2.862 2.830 2.831 2.797 2.804 2.810

0.0265 0.0216 0.0192 0.0161 0.0146 0.0135 0.0201

a

3.339 2.154 1.775 1.264 0.725 0.658 2.575

1.641 0.347 0.220

0.334

∆E0

residual

2.23 2.38 2.00 1.91 -0.39 0.30 1.63

26.68 21.70 19.80 5.96 7.41 13.57 13.25

The solvent is 50%MgCl2-(LiCl-KCl eutectic) mixture.

number approaches 6 and the nearest-neighbor distance 2.80 Å. The behavior of the coordination number agrees with the Raman study9 of molten LaCl3 mixture systems in this dilute regime. The nearest neighbor distance decreases by about 0.07 Å between the pure melt and dilute mixture, which we associate with the decrease in coordination number. The Debye-Waller factor (the second cumulant), which corresponds to the width of the first peak of the La3+-Cl- radial distribution function, and the third cumulant, which characterizes the asymmetry in this function, also decrease by the mixing. They suggest that a well-ordered 6-fold coordination structure (LaCl6)3- is formed and is stabilized by mixing with the LiCl-KCl eutectic at sufficiently low LaCl3 concentrations. (2) Comparison with MD Simulations. The XAFS functions k3χ(k) calculated from the MD simulation and the FEFF

4668

J. Phys. Chem. A, Vol. 114, No. 13, 2010

Yoshihiro et al.

Figure 4. Structural parameter (open circles); coordination number (CN), the nearest distance, the Debye-Waller factor, and the third cumulant for the La3+-Cl- correlation obtained from the EXAFS curve fitting analyses of molten LaCl3-(LiCl-KCl eutectic) mixtures. The parameter obtained from the MD simulation is plotted as open squares. The EXAFS curve fitting results of molten LaCl3-MgCl2-(LiCl-KCl eutectic) mixture is also shown as closed squares.

Figure 5. Coordination number distribution of Cl- ion around La3+ obtained from the MD simulation of three kinds of LaCl3 mixture systems.

computation are plotted as a dashed curve in Figure 2. The experimental XAFS functions are well reproduced by the simulation, in particular the lengthening of the period of the XAFS oscillations with increasing dilution. The reproduction of the experimental data at low LaCl3 concentrations could be improved by a small shift of the phase. An analysis of the Cl- ion coordination around La3+ ion yields simulation results for the mean coordination number and the mean nearest-neighbor separation, which are compared with the results of the XAFS analysis in Figure 4. The coordination number in the simulation was calculated from the first La3+-Cl- peak of the pair distribution function g(r) with the cutoff distance, which is the first minimum after the first peak. The simulation results show the same decrease of coordination number and separation as the direct analysis of the XAFS data, though the decrease in the coordination number is not as large. In the previous simulation studies22 of the LaCl3-ACl mixtures the coordination number had already dropped to 6 for a 20% mixture with the very basic CsCl but decreased more slowly, and in a similar way to the LiCl-KCl eutectic data shown here, for the more acidic NaCl. In addition to the mean coordination number the simulation may also be analyzed for the distribution of coordination numbers, which is shown in Figure 5 by the heavy dashed curve. In the pure LaCl3 melt, the coordination number distribution shows a peak at value 8. It is compatible with the XAFS fitting results in the Table 2, as well as the analyses of neutron and X-ray diffraction data.12,18 Even in the pure melt, only about

half the ions are actually 8-coordinate at any instant, the remainder are 7- or 9-coordinate in roughly equal proportions. As the degree of dilution increases, the peak of the distribution shifts to lower values so that at the most dilute concentrations the mixture contains roughly equal amounts of 6- and 7-coordinate species. The facts that the mixtures contain a distribution of ions with different coordination numbers and that we calculate an XAFS signal by superimposing the signals from individual La3+ ions suggests a way of illustrating how the observed XAFS signal is truly sensitive to the shift in coordination number types. In Figure 6, we show the comparison of the overall XAFS signal with those calculated from only those ions having a particular coordination number. We have used the 50% LaCl3 mixture for illustration. The top panel shows the comparison of the total calculated signal with the experimental one and below the comparison the experimental signal with that calculated for ions with 6-, 7-, 8-, and 9- nearest neighbors separately (the total signal being made up of these separate subspectra weighted by the probability of finding that coordination number). It can be seen that the signals calculated for the different coordination numbers show a systematic shift in the period of the XAFS oscillations that is consistent with the trend of the larger coordination numbers having the largest mean La3+-Clinterionic separation. The spectra of the 7-, 8-, and 9- coordinate species, which are the most probable ones in this mixture, bracket the one observed experimentally. In addition to the mixture with the LiCl-KCl eutectic, simulation results for the coordination number (CN) distributions

Coordination about La3+ in Molten LaCl3

J. Phys. Chem. A, Vol. 114, No. 13, 2010 4669

Figure 7. La3+-Cl- cage-correlation functions (CCF) obtained from the MD simulation of three kinds of LaCl3 mixture systems. Figure 6. Experimental and simulated EXAFS functions k3χ(k) of the molten 50% LaCl3-(LiCl-KCl) mixture. The top panel shows total EXAFS functions k3χ(k). In the other four plots, the simulated function is separated by using selected structures having a particular coordination number (from CN ) 6 to 9). The total simulated function (the top panel) is the weighted average of these separated functions.

of two binary systems LaCl3-LiCl and LaCl3-KCl are also plotted in Figure 5, which illustrates the effect of the basicity of the alkali halide on the La coordination number. In the mixture of 75% LaCl3, the distribution profile does not change from that of the pure melt for all the three systems. A small difference is observed at 50%. The profile of the 50% LaCl3-(LiCl-KCl eutectic) and 50% LaCl3-LiCl melt is almost the same as that of the pure melt, peaking at 8: on the other hand, the 50% LaCl3-KCl melt contains a higher number of 7-coordinate species. At first sight, this appears to contradict chemical intuition, as KCl is the more basic (Cl- releasing) solvent and might be expected to favor high coordination of Cl- about La3+; we will return to this paradox below. Drastic changes occur between 30% and 11% LaCl3 concentrations where the different solvation characteristics of the alkali halides become apparent. At 11% the most predominant CN is 6 for the LaCl3-KCl system, 7 for the LaCl3-(LiCl-KCl eutectic) system, and still 8 for the LaCl3-LiCl system, respectively. The CN of the pure LaCl3 decreases from 8 to 6 by the mixing with KCl, similar to the behavior previously observed for the CsCl mixtures, so that in the 5% and 1% LaCl3-KCl melts the proportion of CN 6 is over 80%. On the other hand, the solvent LiCl does not allow the formation of (LaCl6)3- and the proportion of CN 6 is less than 10% for all the compositions. It seems that the mixing with LiCl-KCl eutectic shows the average effect of the two alkali chlorides.

A measure of the stability of the coordination structure may be obtained by calculating the cage-correlation function (CCF); the decay of the CCF gives the time scale on which an ion leaves the coordination shell or a fresh one enters it. We have previously shown how the CCF provides a link between coordination shell structure and the transport properties of the melt.35,37 Figure 7 shows the CCFs for the different systems. The slowing of the decay of the CCF on increasing dilution in KCl shows that mixing with KCl gives a progressively stabilized coordination structure. On the other hand, in LiCl the decay curves are grouped around the pure LaCl3 one, suggesting that, in some sense, the tendency of a La3+ ion to exchange one of its coordinating Cl- ions with a Li+ ion in the “solvent” is comparable to the tendency to exchange with another La3+. This measure then, seems to parallel the behavior of the degree of thermodynamic ideality of the LaCl3-ACl mixtures discussed in the Introduction. The more basic alkali halides show a greater departure from ideality than the more acidic ones, and the CCF observations show that the latter (LiCl) leave the properties of the coordination shell more similar to that of pure LaCl3 in terms of both coordination number and coordination shell stability. A discussion that linked the La3+ coordination shell structure with the affinity of the alkali cation for Cl- was given in section V.A of ref 22. The ideal coordination number of La3+, as seen in high dilution in basic melts (CsCl and KCl), is 6, to form the ideal octahedral coordination complex. However, even in very basic solutions, a structure consisting of isolated (LaCl6)3units surrounded by alkali cations that are weakly coordinated to the outside of the shell of Cl- ions would only be possible until a molar concentration of 25% is reached (where the stoichiometry is A+3(LaCl6)3-). For higher concentrations there are not sufficient Cl- ions to go round and some degree of crosslinking in which La3+ coordination polyhedral share a common

4670

J. Phys. Chem. A, Vol. 114, No. 13, 2010

Figure 8. EXAFS Fourier transform magnitudes |FT(k3χ(k))| of molten LaCl3-MgCl2-(LiCl-KCl eutectic) mixtures.

Cl- ion must occur. For the larger trivalent cations, the crosslinking is associated with an increase in coordination number beyond six, reaching a value close to eight in pure LaCl3. The rate at which the coordination number decreases as dilution increases is then governed by the ability of the alkali halide to break up these cross-linked complexes, this seems to be greatest for the most basic alkalis. One might surmise that because they coordinate more weakly to Cl- they are more ready to accommodate an arrangement of Cl- ions that has been dictated by the preferred coordination structure of La3+. In ref 22 the relative tendency of Na+ and Cs+ cations to break up these crosslinked structures was followed by a study of the local structure. (3) Molten LaCl3 in a MgCl2-LiCl-KCl Mixture. Further experiments were performed with MgCl2 as a third component added to the LaCl3-(LiCl-KCl) eutectic mixture. Mg2+ is an even more acidic cation than Li+. It has a similar crystal radius (Mg2+ (0.72 Å) vs Li+ (r(VI) ) 0.76 Å), much smaller than that of the K+ ion (1.38 Å), but a higher charge and therefore coordinates to Cl- more strongly. Consequently, on the basis of the above discussion, we would expect the willingness of Mg2+ to accommodate the ideal Cl-coordination geometry around La3+ would be even lower than for the eutectic mixture. As such, Mg2+ will be even poorer at disrupting the cross-linking of the La3+ coordination complexes and therefore be likely to induce high coordination numbers to persist down to low concentrations in the mixtures. The Fourier transform magnitude |FT(k3χ(k))| of molten 2% LaCl3 in MgCl2-(LiCl-KCl eutectic) equimolar mixture is shown in Figure 8, together with those of molten 1% and 5% LaCl3 in the LiCl-KCl eutectic mixture for comparison. It can be seen that the first peak is broader and more asymmetric than either of these systems; i.e., it more closely resembles the pure melt. The results of the WinXAS curve fitting analysis are listed in the Table 2 and plotted in Figure 4. The broadening and deformation of the first peak in the |FT| function leads to an increase in the Debye-Waller factor and the third cumulant C3 relative to the eutectic. The increases in the coordination number and the distance between La3+ and Cl- ions are small but consistent with the expectation described above. In other words, MgCl2 does not destabilize the cross-linking of the La3+ coordination polyhedra and the tendency to lower the coordination number on dilution is even lower than in LiCl. 4. Conclusions In this work we have demonstrated the applicability of highenergy XAFS measurements to the study of the local coordination about the lanthanum ion in multicomponent chloride melts,

Yoshihiro et al. even when the ion is at low concentration. A direct analysis of the data yields information about the La3+-Cl- separation and about the La coordination number, provided that the asymmetry of the first peak of the La3+-Cl- coordination number is allowed for in the analysis. XAFS spectra were also computed from a large sample of local configurations generated in MD simulations of the molten mixtures. These spectra agreed well with the experimental ones and analysis of the simulations gave information on the coordination structures, which agreed well with that which had been obtained by the direct analysis. The MD simulations had previously been shown to predict a number of other properties of LaCl3-alkali halide mixtures reliably, including diffraction patterns, thermodynamic data, transport coefficients, and Raman spectra, and the reproduction of the XAFS data provides further confirmation of the reliability of the simulation structures. The mean coordination number of Cl- around La3+ was found to decrease from a value close to eight in the pure melt to a limiting value of six (octahedral coordination) in dilute solutions in alkali halides. However, the rate of decrease depends on how strongly the alkali cation binds to Cl-, so that the limiting value is reached at moderate concentrations for the larger alkalis, like Cs+ 22 or K+, but not for Li+ and Na+. Addition of Mg2+ appears to strongly inhibit the formation of the isolated octahedral coordination complex. The simulations indicate that these differences are associated with the ease of breakdown of the network structure of pure LaCl3. They parallel the degree of ideality of mixtures of LaCl3 with different alkali chlorides,1,2,6 with the large alkali systems (which bring about the greatest change in La coordination from the pure melt) associated with the largest departures from ideality. It is interesting to note, in closing, that other trivalent metal ions show different behavior when mixed with alkali halides. Ions smaller than La3+, like Y3+ or Dy3+, which form sixcoordinate chloride crystals, appear not to change their coordination number on dilution in alkali chlorides.10,32 The much smaller Al3+ ion increases its coordination number on increasing dilution in alkali fluoride melts.38 Acknowledgment. We thank Dr. J. Mizuki and Dr. K. Aoki for supporting the XAFS measurements at the SPring-8. The molecular dynamics simulation was performed by using the supercomputer Altix3700Bx2 at the Japan Atomic Energy Agency. We acknowledge to Dr. M. Salanne and Dr. H. Matsuura for their help in the XAFS simulation work. Supporting Information Available: Image of quartz cell used in the high-temperature XAFS measurements. Plots of molar volumes from the MD simulation. Figure showing the XAFS data analysis simulation. This material is available free of charge via the Internet at http://pubs.acs.org. References and Notes (1) Papatheodorou, G. N.; Ostvold, T. J. Phys. Chem. 1974, 78, 181. (2) Papatheodorou, G. N.; Kleppa, O. J. J. Phys. Chem. 1974, 78, 178. (3) Ogawa, T.; Minato, K.; Okamoto, Y.; Nishihara, K. J. Nucl. Mater. 2007, 360, 12. (4) Salanne, M.; Simon, C.; Turq, P.; Madden, P. A. J. Phys. Chem. B 2008, 112, 1177. (5) Mamentov, G., Marassi, R., Eds. Molten Salt Chemistry; Reidel: Dordrecht, The Netherlands, 1987. (6) Rycerz, L.; Gadzunc, S.; Gong, W.; Ingier-Stocka, E.; GauneEscard, M. J. Mol. Liq. 2007, 131, 246. (7) Rollet, A.-L.; Godier, S.; Bessada, C. Phys. Chem. Chem. Phys. 2008, 10, 3222. (8) Papatheodorou, G. N. J. Chem. Phys. 1977, 66, 2893. (9) Papatheodorou, G. N. Inorg. Nucl. Chem. 1975, 11, 483.

Coordination about La3+ in Molten LaCl3 (10) Papatheodorou, G. N.; Yannopoulos, S. N. In Molten Salts: from Fundamentals to Applications; Gaune-Escard, M., Ed.; NATO ASI Symposium Series 2; Kluwer: Boston, 2002; Vol. 52, pp 47-106 (see also references therein). (11) Iwadate, Y.; Matsuura, H.; Kajinami, A.; Takase, K.; Ohtori, N.; Umesaki, N.; Fujita, R.; Mizucuchi, K.; Kofuji, H.; Myochin, M. Electrochemistry 2009, 77, 736. (12) Wasse, J. C.; Salmon, P. S. J. Phys.: Condens. Matter 1999, 11, 1381. (13) Wasse, J. C.; Salmon, P. S.; Delaplane, R. G. Physica B 2000, 276/278, 433. (14) Wasse, J. C.; Salmon, P. S. J. Phys.: Condens. Matter 1999, 11, 9293. (15) Saboungi, M.-L.; Price, D. L.; Scamehorn, C.; Tosi, M. P. Europhys. Lett. 1991, 15, 283. (16) X-ray Diffraction Analysis of Ionic Liquids; Ohno, H., Igarashi, K., Umesaki, N., Furukawa, K., Eds.; Molten Salt Forum Vol. 3; Trans Tech Publications: Stafa-Zurich, Switzerland, 1994. (17) Okamoto, Y.; Hayashi, H.; Ogawa, T. Jpn. J. Appl. Phys. 1999, 38, 1569. (18) Okamoto, Y.; Madden, P. A. J. Phys. Chem. Sol. 2005, 66, 448. (19) Hutchinson, F.; Rowley, A. J.; Walters, M. K.; Wilson, M.; Madden, P. A.; Wasse, J. C.; Salmon, P. S. J. Chem. Phys. 1999, 111, 2028. (20) Hutchinson, F.; Wilson, M.; Madden, P. A. Mol. Phys. 2001, 99, 811. (21) Okamoto, Y.; Madden, P. A. J. Nucl. Mater. 2005, 344, 109. (22) Glover, W. J.; Madden, P. A. J. Chem. Phys. 2004, 121, 7293.

J. Phys. Chem. A, Vol. 114, No. 13, 2010 4671 (23) Zissi, G. D.; Chrissanthopoulos, A.; Papatheodorou, G. N. Vibr. Spectrosc. 2006, 40, 110. (24) Bessada, C.; Rakhmatullin, A.; Rollet, A. L.; Zhanghi, D. J. Nucl. Mater. 2007, 360, 43. (25) Bessada, C.; Rakhmatullin, A.; Rollet, A. L.; Zhanghi, D. J. Fluorine Chem. 2009, 130, 45. (26) Okamoto, Y.; Shiwaku, H.; Yaita, T.; Narita, H.; Tanida, H. J. Mol. Struct. 2002, 641, 71. (27) Shiwaku, H.; Mitsui, T.; Tozawa, K.; Kiriyama, K.; Harami, T.; Mochizuki, T. AIP Conf. Proc. 2004, 705, 659. (28) Okamoto, Y.; Yaita, T.; Minato, K. J. Non-Cryst. Solid 2004, 333, 182. (29) Okamoto, Y.; Yaita, T.; Minato, K. J. Mol. Struct. 2005, 749, 70. (30) Filipponi, A. J. Phys.: Condens. Matter 2001, 13, R23. (31) Okamoto, Y. Nucl. Instrum. Methods Phys. Res. A 2004, 526, 572. (32) Okamoto, Y.; Akabori, M.; Motohashi, H.; Shiwaku, H.; Ogawa, T. J. Synchrotron Radiat. 2001, 8, 1191. (33) Ressler, T. J. Synchrotron Radiat. 1998, 5, 118. (34) Aukudinov, A. L.; Rehr, J. J. Phys. ReV. B 2000, 62, 2437. (35) Brookes, R.; Davis, A.; Ketwaroo, G.; Madden, G. J. Phys. Chem. B 2005, 109, 6485. (36) Janz, G. J. J. Phys. Chem. Ref. Data 1998, 17, 1. (37) Salanne, M.; Simon, C.; Turq, P.; Madden, P. A. J. Phys. Chem. B 2007, 111, 4678. (38) Robert, E.; Materne, T.; Tixhon, E.; Gilbert, B. Vibr. Spectrosc. 1993, 6, 71.

JP910637P