STRUCTURE AND THERMODYNAMICS OF LANTHANIDE AND ACTINIDE COMPLEXES IN SOLUTION GREGORY R. CHOPPIN Department of Chemistry Florida State University, Tallahassee, Florida 32306, USA ABSTRACT Several questions of importance in the study of lanthanide and actinide coordi- nation compounds are reviewed. There is considerable evidence that in aqueous solution the primary coordination number is nine for the ions La(ni) through Nd(m) and eight for the ions heavier than Gd(m). While it seems that some degree of covalency exists in the metal—ligand bonding a model of electrostatic bonding seems satisfactory for explaining the structure and formation of the complexes. The dehydration of the lanthanide ions upon complexation largely determines the enthalpy and entropy data. However, there seems to be a compensation effect in the hydration parts of those terms such that the free energy changes seem to reflect the metal—ligand reaction unobscured by hydra- tion factors. A number of individual inorganic and organic ligand systems are reviewed from these viewpoints. The lanthanide and actinide elements constitute two families of metals which often exhibit very similar chemical behaviour. This similarity is most easily observed with the cations in oxidation state iii, the most common one for all the elements of the lanthanide series and for the transpiutonium elements of the actinide series. This paper reviews the present state of under- standing of the complexes formed in aqueous solution by the trivalent ions of these two families ofelements. The coordination chemistry of these elements has been studied neither as intensively nor as extensively as that of the transition metal ions. One reason for this is that these elements have been less available generally than the transition metal elements. Although they are fairly abundant, a lack of industrial application until a relatively few years ago caused many of the lanthanide elements to be fairly expensive. Increased interest in the use of these elements in phosphors and lasers and as catalysts as well as improve- ments in the methods of separation has caused a substantial reduction in their cost. The availability of the transuranium elements of the actinide series was limited to a few nuclear energy laboratories where greater emphasis was placed on research of either a more nuclear or more applied nature. This latter situation has changed as the transpiutonium programme of the US Atomic Energy Commission has resulted in production of relatively large amounts of the elements as heavy as californium (Z = However, the 98). 23 GREGORY R. CHOPPIN intensely radioactive nature of these elements still imposes such constraints as the availability of a good radiochemical laboratory and the use of tracer techniques for many studies. A second major reason for the greater interest by coordination chemists in the transition metal systems is found in the rich variety of complexes formed by these metals with a multitude of ligands. By contrast, the lanthanide and actinide ions resemble the alkaline earth ions. They are type A or 'hard' acids which have a strong preference for oxygen donor ligands. In the solid it is possible to have coordination via nitrogen, sulphur etc. but in aqueous solution complexation almost always involves substitution of the solvate waters with their metal—oxygen bond by a ligand which forms another metal—oxygen bond. Finally, much of the great interest in transition metal complexes can be attributed to the rapid and exciting evolution of the application of ligand field theory to these systems. The f electrons of the lanthanide and actinide elements are rather well shielded compared to d electrons. As a result, the effect of external fields is sharply reduced in complexes of the lanthanides and actinides. Moreover, the limitation of complexation in aqueous solution to oxygen donor groups places further restriction on the extent and range of possible ligand field effects. Associated with this is the absence in the lanthanide and actinide complexes of marked variations in the nature of the bonding. Nevertheless, study of the lanthanide and actinide elements does have a number of attractive aspects. The absence of significant ligand field effects in the formation of the complexes in aqueous solutions means that other phenomena may be more easily studied. The effect of change in cation radius, through the series of 15 elements from La(ui) to Lu(In), the coordina- tion number in the two series of elements, the degree of covalency in the complexes and its variation between the two series are a few examples of the questions of interest in complex formation by these elements. Comparison of the differences between the two families may indicate the relative roles, if any, of 4f and 5f orbitals in the complexation. Since these trivalent cations increase the net structuring of solvent water, these elements also allow an excellent approach for studying many solution effects such as the role of solvation and solvent structure in complex formation. First we shall discuss, in general, the questions of coordination number, ionic size and degree of covalency. Then we shall review a few specific ligand systems with emphasis on the thermodynamics of complex formation. In no way is this intended as a complete review since references have been chosen primarily for illustrative purposes. COVALENCY There is general but not complete agreement that the metal—ligand bonds in lanthanide complexes are predominantly electrostatic. For example, Hoard and co-workers' have studied the structure of HLnEDTA 4H20 in which Ln(iii) has a coordination number of ten and KLnEDTA 8H20 in which the coordination number of Ln(iii) is 8(Tb) or 9(La). They conclude that in these complexes the f-electrons of Ln3 + have no role in determining 24 THERMODYNAMiCS OF LANTHANIDE COMPLEXES gross configuration or the pattern of primary bonding even though crystal field stabilization byf-electrons is substantial. Techniques such as optical and n.m.r. spectroscopy have provided convinc- ing evidence that some small degree ofcovalency must be present in the bonding in lanthanide complexes. Oxygen- 17 chemical shifts in solution of the lan- thanide perchiorates have been attributed to formation in the hydrated cations of covalent bonds involving the lanthanide 6s orbital2. This inter- pretation has been supported by Rueben and Fiat3 in studies of proton and 170 shifts in solutions of various lanthanide compounds. These results indicate that although the bonding must be primarily electrostatic, a small degree of covalent bonding between water molecules and the lanthanides is present and, moreover, remains relatively constant along the series with little effect on the lability of the water molecules of the aquo complexes. Some additional insight into the extent of covalency has been obtained from optical spectroscopy. The small effects of ligands on the spectra of lanthanide ions is observed as peak shifts, band splitting and intensity changes. The peak shifts have been used to conclude that the metal—oxygen bond in complexes in aqueous solution is weakly covalent4 and involves hybrid orbitals with 4f contribution. In the solid state, for example, the change in covalency from Pr—F bonds to Pr—0 bonds has been estimated to be approximately two per cent by the same criteria of peak shifts5. A recent study of Am(iii) complexes withf aminocarboxylate ligands has shown larger spectral shifts in americium f —* absorption bands than for analogous lanthanide bands6. This would be consistent with less shielding of 5forbitals than 4forbitals and, possibly, a greater degree of covalency in actinide bond- ing than in lanthanide bonding. Certain transitions, termed hypersensitive, exhibit relatively large sensitivities to the ligands. The theory of the origin of hypersensitivity is unsettled although it is likely that both symmetry and covalency play roles7' 8• Once the factors involved in hypersensitivity are understood to a satisfactory degree, the study of the effect of ligands on hypersensitive spectral transitions may be a valuable tool in lanthanide research, particularly in regard to questions of symmetry and covalency. In summary, then, it seems reasonable to agree that the degree of covalency is small in metal—ligand bonds of the lanthanides and the structure and thermodynamics of complex formation in aqueous solution may be inter- preted on the basis of electrostatic attraction. Since the 5f orbitals are less shielded in the actinide elements, it is probable that the covalency is larger. However, the similarity in complexation of the two families supports the view that in the actinides, as in the lanthanides, a model of electrostatic bonding is satisfactory for interpreting complex formation. COORDINATION NUMBER At this time we know little about the coordination number of trivalent lanthanide and actinide ions in complexes in aqueous solution. While the coordination numbers observed in solid complexes may reflect the values for the complexes in aqueous solution, such an assumption must be approached with caution. Nevertheless, let us cite a few cases of the coordination in solids. 25 GREGORY R. CHOPPIN The evidence for eight, nine and ten-coordination in the hydrates of the lanthanide EDTA complexes has been mentioned in the previous section. Lanthanum carbonate octahydrate, La2(C03)3 8H20, has also been shown to be ten-coordinate9 while the monohydrate and trihydrates of the lanthanide acetylacetonates are seven and eight coordinate respectively1 O• The solid neodymium bromate nonahydrate, Nd(BrO3)3 9H20, has been found to have the Nd(iii) ion surrounded by nine water molecules1 .Coordinationnumber six has also been found in a number of systems. Thus the coordination number varies from six to ten depending upon the particular system. Brady studied concentrated solutions of ErCl3 and Er!3 by x-ray diffrac- tion and concluded that six water molecules are arranged octahedrally around the cation and two halide ions are located in the centre of two opposite edges of the octahedron'2. Some interesting work has been reported by Karraker in which he compared the spectra of solids and solutions13. He found that the spectra of Nd(Iii) in dilute solutions and in crystalline Nd(Br03)3 .9HO were very similar. From this he concluded that the coordination number of aquo-Nd(In) is nine. As the concentration of the solutions increased in HCI, LiC1 and HC1O4, the spectra also changed to resemble that of the 8-coordinate Nd(iii) in NdCl3 8H20. The heavier lanthanides such as Ho(Iii) and Er(iIi) do not show such spectral changes so it is assumed that the coordination is unchanged for these ions as the con- centration of electrolyte increases. Studies by Spedding's group on such thermodynamic properties as molal volumes, heat capacities and relative viscosities'4 indicated a change in coordination number along the lanthanide series. The best interpretation of the spectral and thermodynamic data is that the hydrated ions in dilute solution have acoordination number of nine from La(in) to Nd(In) and one of eight from Tb(Iii) to Lu(m). Pm(iu), Sm(m), Bu(iii) and Gd(iii) presumably have a mixture of these coordination numbers. As is obvious from the variety of coordination numbers found in solid complexes, we may not assume that the same coordination numbers exist in complexes in solution as in the hydrated ions. However, studies with aminopolycarboxylate ligands have led to the suggestion that the coordina- tion number of Ln(ni) in complexes is greater than six'5. Thompson provided evidence from mixed ligand complexes that coordination numbers greater than six are, in fact, more likely the norm16. He showed that six-coordinate transition metal ions would not form mixed chelates with N-hydroxyethyl- ethylenediaminetriacetic acid, HEDTA, and iminodiacetic acid, IMDA. However, the lanthanides formed Ln(HEDTA) (IMDA)2 -and Ln(EDTA) (IMDA)3 - from which Thompson concluded the coordination number to be eight for the lanthanide ions. More recently the study of other mixed complexes of the lanthanides supports a coordination number of eight' '. It seems likely, then, that in complexes the coordination number is variable and dependent on steric effects of the ligands rather than on the geometry of directed bonds. This is consistent with the electrostatic nature of the bonding. Very little has been done on coordination numbers of trivalent actinide ions in solution or in coipplexes and the few studies available indi- cate a situation very similar to that of the lanthanide ions. It has been sug- gested that a coordination number of ten exists for Am(in) in complexes6. 26 THERMODYNAMICS OF LANTHANIDE COMPLEXES IONIC SIZE A favourite relationship of researchers in lanthanide complexation is that between the stability constants or some other thermodynamic parameter and the reciprocal of the cationic radius. A linear relationship would be expected from a model of purely electrostatic interaction. Invariably the ionic radii used for the lanthanides are those reported by Templeton and Dauben18 which relate to a coordination number of six, whereas, as we have seen, the actual coordination number is likely larger and may even change between Nd(iii) and Tb(iii); for example from nine to eight. In Figure 1 the experimental radii for CN =6 and calculated radii for CN =8 and 1.10 1.00 UN) U) U C0 0.90 CN=9 CN=8 CN = 6 _________ I II I I I La Ce Pr Nd PmSrnEu 3d Tb Dy Ho Er Tm Yb Lu Figure 1. Ionic radii for coordination numbers six, eight and nine CN = 9 are presented. In the calculations a factor of 1.032 was used for CN =8 and 1.046 for CN =9 as recommended by Pauling'9. The heavy line shows the trend in radii if a change from CN =9 to CN =8 occurs between Nd(m) and Tb(iii). We can see that a deviation from linearity in plots involving the reciprocal of the radius may reflect use of the wrong set of values as the trend of the heavy line differs from that of the line for CN = 6. Hopefully, sets of lanthanide ion crystal radii for CN =8, 9 and 10 may be determined by crystallographers for use in the correlations with thermodynamic properties of these elements. However, the existence of different Ln—O bond lengths within the same compound poses a serious question to the definition of the radius. 27 GREGORY R. CHOPPIN INORGANIC LIGANDS Let us now turn our attention to a consideration of some specific ligand systems. Among the inorganic ligands whose complexation has been studied with lanthanide and actinide ions are the halides, nitrates, thiocyanates and suiphates. With the lanthanide and actinide ions, chloride, bromide and iodide form very weak complexes whose order is Cl> Br> I. On the basis of the small, negative values of the enthalpy and entropy terms for the formation of the monochioro species, the complexes have been characterized as outer sphere20. This model of solvent separated ion pairing was supported by studies of the spectra of Nd(iii) which showed• no change up to concentra- tions of 5M chloride ion2 1 A spectral change was observed at higher concen- trations which may be related to a change in coordination number from nine to eight as proposed by Karraker. There is no evidence for formation of inner sphere complexes of the halides with lanthanide ions. The trivalent actinides, however, are reported to form such species to the extent of about one per cent of that of the outer sphere complexation22. This corresponds to an increase of about 2.7kcalin the free energy as a ligand moves from an outer to an inner sphere position and a water molecule is eliminated. Even in concentrated LiC1 solutions, only two chlorides can enter the inner sphere to form [AmCl2(H2O)]1 + with a likely coordination number of eight. The observation of the inner sphere complexation in actinides may reflect the greater probability of participation of the 5forbitals in bonding. The nitrate complexes are slightly stronger than the halide complexes. Again, the thermodynamic data on the formation of the mononitrato species is indicative of outer sphere complexation23. The net enthalpy and entropy changes in complex formation are the result of a positive contribution to both AH and AS as a result of the disruption of the hydration layer about the cation (anion hydration is usually small) and a negative contribution to both terms from the combination of the ions in the complex. In outer sphere complexation, the hydration zone should only be slightly affected with the result being rather small values of AH and AS, often negative. Ahrlapd has recently reviewed these arguments for a number of systems and found them generally valid24. He has pointed out that values of AH1 close to zero is a more reliable criterion of formation of outer sphere complexes since the value of AS1 depends rather strongly on the ionic strength. Although the dominant mode of complexation is outer sphere in the nitrate complexes, both spectral21'25 and n.m.r.3 data indicate the presence of a small degree of inner sphere formation, possibly two per cent8. Moreover, it seems that the nitrate ions function as bidentate ligands in these complexes. The thiocyanate complexes of both the actinides and lanthanides have been studied and the thermodynamic and spectral data for the monothiocyanato complexes agree with an interpretation of outer sphere formation21' 26• However, in concentrated solutions of ammonium thiocyanate, elution from anion exchange resins provides evidence for .the formation of such species as M(SCN) which are most likely inner sphere27. Again, the actinides seem to form these anionic species to a greater extent than do the lanthanide ions. 28 THERMODYNAMICS OF LANTHANIDE COMPLEXES As an example of complex formation with divalent inorganic ligands, let us consider the sulphate system. The stability constants for LnSO are an order of magnitude greater than those for LnNO + but still rather small (ca. 20 at an ionic strength of 2.OM). The enthalpy and entropy values for the lanthanide and actinide monosuiphate formation are very similar and posi- tive. From this it was concluded that an appreciable amount of inner sphere complexation was present28. Earlier, a study by infra-red spectroscopy had led to an estimate of only 12 per cent inner sphere formation29. However. spectral data8'3° as well as the rates of formation of the monosulphate complexes studied by ultrasonic absorption3' are in agreement with the interpretation from the thermodynamic data that the complexes are pre- dominantly inner sphere in nature. The fluoride complexes are interesting as the magnitude and the trend of the thermodynamic parameters for the formation of the monofluoride species are unusual-for monovalent inorganic ligands32. Figure 2 shows the a, 0 E a, 0 a0, •1 63 z Figure 2. Entropy of formation of some monoligand complexes 29 GREGORY R. CHOPPIN entropy data compared with similar data for acetate, glycolate and the multi- dentate EDTA systems. The values of fluoride are surprisingly large and account for the formation of the complex since the enthalpy values are quite endothermic. These data are interpreted as evidence of inner sphere formation. Moreover, in this system we have an unusual case as the major contribution to both the entropy and the enthalpy changes comes not from the dehydration of the metal ion but rather from the disruption of the ordered hydrate - Calculated covalency enthalpy differences Cl-Br C >, 0 - Experimental enthalpy differences C - F-O > a I I I I I I I I I I I I I La Ce Pr Nd Pm SmEu GdTb Dy Ho Er Tm Yb Lu Figure3. Enthalpy of formation of LnF2 +comparedwith the calculated differences in covalent bonding energy of LnC12 + andLnBr2 structure associated with the fluoride ion. The difference, for example, in the aqueous entropies of fluoride and acetate ions is 23 cal/deg. mole which is roughly the difference between the entropy changes for formation of the mono fluoride and the monoacetate complexes. The strength of the fluoride complexes is a result, then, of the entropy contribution derived from the decrease in net solvent structure when the fluoride ion is complexed rather than to any unusual metal—fluoride attraction. The enthalpy curve for formation of the monofluoride species is also un- usual (Figure 3). Since this is a rare case where a metal—oxygen bond is replaced by a metal—fluoride bond, possible covalent effects may be present. 30 THERMODYNAMICS OF LANTHANIDE COMPLEXES We have seen that in the solids, an estimate of two per cent was made for the difference in covalency between Pr—F and Pr—O bonds. The shape of the enthalpy curve could reflect the variation in covalent effects across the lan- thanide series. To attempt to ascertain the validity of this proposal, the shape of the enthalpy curve is compared in Figure 3 with the calculated 'covalency enthalpy differences' between the lanthanide monochioride and bromide complexes based on the energy of their charge transfer bands in ethanol. The details of this calculation are to be found in ref. 32.Theagree- ment between the two curves is not perfect but does add some support to the proposal that the enthalpy trend in the fluoride complexation reflects cova- lent effects. The calculated covalent curve is very similar to the trend in the energy of the 4f6s2 -÷ 4fn_ 15d16s2change33. ORGANIC LIGANDS Organic ligands form stronger complexes with the lanthanide and actinide ions than do inorganic ligands. The complexes of the simplest carboxylic acids such as acetate, propionate and isobutyrate are inner sphere. As in the case of the inorganic ligands, the thermodynamic data for complexation with organic ligands in aqueous solution can be interpreted as primarily reflecting the effects on the hydrate structure about the cation34'35. The formation of an inner sphere complex sufficiently disrupts the hydration about the cation to result in net entropy and enthalpy values that are positive. As a result, as in the fluoride and sulphate systems, the spontaneous forma- tion of the complex is attributable to a favourable entropy change overcoming an unfavourable enthalpy change. The thermodynamics of the formation of the monoacetate complexes have been compared for the lanthanide and actinides36. As in many complex systems, the stability constants (and, therefore, the free energies) of the actinides are very similar to those of the lanthanides of comparable radii. As shown in Figure 4, however, the enthalpies and entropies are found to be quite different but the difference in these terms between the actinides and lanthanides exactly balance. This 'compensation effect' can be understood by separating the thermodynamic terms into two parts—a reaction part and a hydration part: AG =AG'+ AG" =AH— TAS ± AH" — TAS" If we assume that all the hydrated systems remain in equilibrium with the bulk solvent, AGh = and, consequently AH" = TAS" Therefore, the agreement in AG values for lanthanide and actinide complex- ing means that AG is similar for the two series of complexes. This is consistent with a model of ionic bonding since the cationic radii are similar. Moreover, we can conclude that the difference in AH and AS values indicates that the complexing of the actinide ions by acetate is attended by a greater dehydration than is the complexing of the lanthanide ions. The difference of 9—10 cal/deg. mole is consistent with the release of 0.5-4.0 additional 1-120. 31 GREGORY R. CHOPPIN 8.0 TLS (Act.) TiS(Ln) 6.0 4.0 - —o 2.0 - E. 0 H(Act.) -6.0 0.94 0.98 1.02 1.06 1.10 1.14 1/I., Figure 4. Thermodynamic parameters of formation of MAc2 + where 0 =lanthanides and x = actinides 0 E -o C U (1) La Ce Pr Nd Pm Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Figure 5. Entropy of formation of MX2 +complexeswith carboxylate and c&hydroxycarboxylate ligands; Ac =acetate, P =propionate; lB =isobutyrate, Gi glycolate, Lac = lactate, HIB =c&hydroxyisobutyrate 32