Research Article

Characterization of berkelium(III) dipicolinate and borate compounds in solution and the solid state

See allHide authors and affiliations

Science  26 Aug 2016:
Vol. 353, Issue 6302, aaf3762
DOI: 10.1126/science.aaf3762

Bonding to berkelium

A geographical theme prevailed in the recent naming of the heaviest chemical elements. The choices brought to mind berkelium (Bk) and californium (Cf), the names chosen for elements 97 and 98 over half a century ago. Silver et al. now revisit the chemistry of Bk, which has proven fiercely challenging to study over the years on account of its vigorous radioactive decay. Synthetic crystallized Bk borate and dipicolinate compounds structurally resembled Cf analogs in the solid state but manifested distinct electronic and magnetic characteristics stemming from spin-orbit coupling effects.

Science, this issue p. 888

Structured Abstract


Developing the chemistry of late actinides is hindered by the lack of availability of isotopes, the need for specialized research facilities, and the nuclear instability of the elements. Berkelium represents one of the last elements that can be prepared on a milligram scale in nuclear reactors. However, its only available isotope, 249Bk, has a half-life of only 320 days, which has greatly curtailed the expansion of its chemistry and fundamental exploration of how large relativistic and spin-orbit coupling effects alter its electronic structure. Furthermore, data gathered from Bk(III) in aqueous media suggest that its coordination may be different from that of earlier actinides. However, a single-crystal structure of a berkelium compound has remained elusive, leaving unanswered whether these structural changes occur in the solid state.


This work focuses on characterizing two distinct berkelium compounds on the milligram scale. In particular, the goal was to obtain crystals of these compounds that could be used in structure determinations and physical property measurements. Two compounds were selected: a coordination complex of dipicolinate and a borate. Dipicolinate complexation occurs with most other lanthanides and actinides in the +3 oxidation state, facilitating comparisons across the series to discern periodic trends. In the borate family, the structural frameworks are hypersensitive to the nature of the bonding at the metal center and are rearranged accordingly. Modeling the experimental data using a variety of computational techniques allows us to deconvolute the role of covalent bonding and spin-orbit coupling in determining the electronic properties of berkelium.


Experiments with milligram quantities of 249Bk were choreographed for 6 months before the arrival of the isotope because the total quantity used in the studies was 13 mg, which corresponds to a specific activity of 21 Ci. Although this isotope is a low-energy β emitter, it decays to 249Cf at a rate of about 1.2% per week, and the latter produces hard γ radiation that represents a serious external hazard. In addition, the samples described in this work undergo about 1012 decays per second. This rapid decomposition necessitated the development of techniques for swiftly preparing and encapsulating samples and for collecting all structural and spectroscopic data within 24 hours of crystal formation. After this preparation, the single-crystal structures of Bk(III)tris(dipicolinate) and Bk(III) borate were determined. The latter compound has the same topology as that of californium(III) (Cf) and contains an eight-coordinate BkO8 unit. This reduction in coordination number is consistent with previous solution-phase x-ray absorption measurements and indicates that a drop in coordination number in the actinide series from nine to eight begins at berkelium. The magnetic and optical properties of these samples were also measured. The red luminescence from Bk(III) was similar in nature to that of curium(III) and is primarily based on an f-f transition. The ingrowth of the broad green luminescence from Cf(III), which is caused by a ligand-to-metal charge transfer, was shown to be distinct in nature from that originating from Bk(III). Ligand-field, density functional theory, and wave-function calculations were used to understand the spectroscopic features and revealed that the single largest contributor to the unexpected electronic properties of Bk(III) is spin-orbit coupling. This effect mixes the first excited state with the ground state and causes a large deviation from a pure Russell-Saunders state. The reduction in the measured magnetic moment for these samples from that calculated for an f8 electron configuration is primarily attributable to this multiconfigurational ground state.


The crystallographic data indicate that Bk(III) shares more structural similarities with Cf(III) than with Cm(III). However, ligand-field effects are more similar between Bk(III) and Cm(III). Terbium (Tb), in the lanthanide series, represents the closest analog of Bk because the trivalent cations possess 4f8 and 5f8 configurations, respectively. Spin-orbit coupling in Bk(III) creates mixing of the first excited state (5G6) with the ground state. In contrast, the ground state of the Tb(III)tris(dipicolinate) contains negligible contributions of this type. An overall conclusion from this study is that spin-orbit coupling plays a large role in determining the ground state of late actinide compounds.

Crystal structure of a berkelium coordination compound.

The central Bk(III) ion is coordinated by three monoprotonated dipicolinate ligands in tridentate O,N,O fashion. Bk, yellow; C, gray; N, blue; O, red; H, white.


Berkelium is positioned at a crucial location in the actinide series between the inherently stable half-filled 5f7 configuration of curium and the abrupt transition in chemical behavior created by the onset of a metastable divalent state that starts at californium. However, the mere 320-day half-life of berkelium’s only available isotope, 249Bk, has hindered in-depth studies of the element’s coordination chemistry. Herein, we report the synthesis and detailed solid-state and solution-phase characterization of a berkelium coordination complex, Bk(III)tris(dipicolinate), as well as a chemically distinct Bk(III) borate material for comparison. We demonstrate that berkelium’s complexation is analogous to that of californium. However, from a range of spectroscopic techniques and quantum mechanical calculations, it is clear that spin-orbit coupling contributes significantly to berkelium’s multiconfigurational ground state.

Deep into the periodic table, the rising magnitudes of relativistic effects and spin-orbit coupling and the associated subtle rearrangement of orbital energies challenge our ability to predict and understand the chemical and physical properties of heavy elements. Recent experiments and computations have revealed that models based on monotonic changes in electronic structure across heavy element series are overly simplistic and that perhaps our current arrangement of the periodic table is subject to debate. A case in point is the recent determination that the outermost electron of lawrencium (Z = 103) lies in a p orbital rather than d as would have been predicted by extrapolating from earlier elements (1). Beyond the actinide series, the situation is no less complex with superheavy elements like seaborgium (Z = 106) exhibiting chemical behavior similar to the triad above (i.e., Cr, Mo, and W) (2), whereas reactivity of rutherfordium (Z = 104) and dubnium (Z = 105) can deviate significantly from expectations (3, 4).

Although the exotic electronic structure of superheavy elements is now accepted, the validity of simple chemical principles to predict electron configuration, and thus chemical behavior, is questionable throughout the entire actinide series. Plutonium demonstrates these qualities best, and after seven decades of interrogation, only now do we understand that its electronic structure must be framed within the context of electronic states that are fluctuating on a subpicosecond time scale (5). Similarly, at californium (Z = 98) a metastable electronic state is observed near ambient conditions that also manifests in the spectroscopic features of complex molecules containing Cf(III) ions (610).

However, as is often true with heavy elements, our ability to probe chemical and physical properties is hampered by nuclear instability. This creates gaps that impede the feedback between experiment and theory. One such gap occurs just before californium at the element berkelium. Berkelium has no long-lived isotopes that can be isolated. The only accessible isotope is 249Bk, with a half-life of only 320 days. Compounds that contain isotopes with half-lives of less than hundreds of thousands of years undergo rapid degradation because of the high energetics and ionization nature of nuclear decay processes that can be up to a million times larger than the strength of any chemical bond.

Although measurements and crystal-field modeling of the optical spectra of aqueous solutions of Bk(III), as well as solid-state samples doped with low levels of Bk(III) and Bk(IV), have been reported (1113), single-crystal x-ray analysis of a berkelium compound proved elusive (1416). Spectroelectrochemical measurements of berkelium’s speciation in aqueous media have been accomplished using x-ray absorption spectroscopy and provided evidence that a break in the trend of complexation with actinides might begin at berkelium (17). This latter study points to the need for obtaining high-resolution structural data for berkelium compounds to investigate whether a transition also occurs in the solid state (1820). Furthermore, the optical measurements obtained from hydrated Bk(III), LnCl3:Bk (where Ln is generic for lanthanide), and CeF4:Bk do not reveal the full range of the effects of ligation on the electronic properties of Bk(III) because water and chloride coordination induce relatively small electronic perturbations (1113, 20). Data fitting using early crystal-field models, based on electrostatic interactions between point charges, did not account for the possible involvement of the frontier orbitals in bonding (21).

For meaningful comparisons to be made between the chemistry of berkelium and other actinides, two divergent systems were selected for investigation. The first of these is an archetypal coordination complex that forms via the chelation of An3+ cations by 2,6-pyridinedicarboxylate [dipicolinate (DPA)] yielding An(HDPA)3·nH2O in the solid state (An is Pu, Am, Cm, Bk, or Cf) or [An(DPA)]+, [An(DPA)2], and [An(DPA)3]3− in solution (6, 19, 2224). These complexes have a predictable coordination environment (a feature often absent in the f block), and serve as a benchmark for examining trends in bond distances, spectroscopy, and thermodynamics. The second system is one that has proven to be exquisitely sensitive to the nature of the actinide ions employed in the reaction. In this case, we have shown that polyborates form unique structural topologies with each of the different actinide ions from Pu(III) to Cf(III) (79, 25).

Experiments with milligram quantities of 249Bk were choreographed for 6 months before the arrival of the isotope because the total quantity used in the studies was 13 mg, which corresponds to a staggering activity of 21 Ci. Although this isotope is a low-energy β emitter, it decays to 249Cf at a rate of ~1.2% per week, and the latter produces hard γ radiation that represents a serious external hazard. In addition, the samples described in this work undergo ~1012 decays per second. This rapid decomposition necessitated the development of techniques for swiftly preparing and encapsulating samples and collecting all structural and spectroscopic data within 24 hours of crystal formation. Crystals of these compounds underwent Coulombic explosions within 4 days, although they did not exhibit the color changes that often occur with high specific activity α emitters (6). With the exception of the magnetic susceptibility measurements, all of the data described below were acquired from isolated single crystals.


249BkCl3·nH2O, freshly prepared at the High Flux Isotope Reactor at Oak Ridge National Laboratory, was treated with a five-fold excess of dipicolinic acid in a 1:1 water:ethanol mixture. The initial solution was lime green and emitted faint green light as the result of electronic excitation from the high specific activity of 249Bk (i.e., self-luminescence). The addition of DPA resulted in an immediate color change and precipitation of a golden-yellow crystalline product. Mild heating of the reaction mixture led to ripening of the microcrystals and formation of golden crystals of Bk(HDPA)3·nH2O with an approximate hexagonal prismatic shape (see fig. S1). Much like the reaction with Cf(III) (6), the precipitation of Bk(HDPA)3·nH2O was nearly quantitative, and little Bk(III) remained in solution at the end of the reaction. This sharply contrasts with curium(III), where a substantial portion of the 248Cm remained in solution in the form of [Cm(HDPA)(H2DPA)(H2O)2Cl]+ (20).


Single-crystal x-ray diffraction studies revealed that Bk(HDPA)3·nH2O is isomorphous with the other members of the trivalent lanthanide and actinide series (table S1) (6). In short, the Bk(III) ions are complexed by three tridentate, monoprotonated dipicolinate ligands, constituting a nine-coordinate, tricapped, trigonal prismatic coordination environment with approximate D3 symmetry. The nitrogen atoms from the pyridine rings are located in the capping positions, and the oxygen atoms from the carboxylate anions fill the six prismatic sites. Tris-chelate complexes, where the metal center is octahedral or nine-coordinate, are chiral. The two enantiomers are designated Δ and Λ, and unless stereoselective processes are involved, a racemic mixture of both enantiomers should be present in every system. Bk(HDPA)3·nH2O crystallizes in a centrosymmetric space group, indicating that a racemic mixture is present. In addition, a hydrogen-bonding network is present between the cocrystallized water molecules and the berkelium complexes. These interactions cause minor distortions of the local coordination around the Bk(III) centers. Each enantiomer is present in the structure in two distinct positions (see Δ and Δ′ in Table 1), one more distorted than the other because of these intermolecular interactions. A view of the Δ and Λ enantiomers is shown in Fig. 1A. Table 2 highlights bond distance changes in the tris(dipicolinate) series spanning Am(III) through Cf(III).

Table 1 Selected bond lengths (Å) for An(HDPA)3 (An is Am, Cm, Bk, or Cf) complexes.

View this table:
Fig. 1 Crystallography.

(A) Views of the Λ (lower left) and Δ (upper right) enantiomers of Bk(HDPA)3, showing the tricapped trigonal prismatic N3O6 coordination around Bk(III). (B) Depiction of the distorted square antiprismatic geometry of Bk(III) in Bk[B6O8(OH)5].

Table 2 Comparison of bond lengths (Å) between the Δ and Δ′ molecules of An(HDPA)3 (An is Am, Cm, Bk, or Cf) complexes.

View this table:

The preparation of Bk(III) borate followed the same procedure used for preparing all other An(III) borates (An is Pu, Am, Cm, or Cf) (9, 2527) and yielded golden tablets of Bk[B6O8(OH)5] (see fig. S2). The synthesis and crystal growth of Bk[B6O8(OH)5] required 10 days during which time nearly 2% of the 249Bk converted to 249Cf. We initially posited that this compound would remain stable under steady β emission, but the buildup of positive charge in the crystals led to violent fracture within 4 days of isolation. Bk[B6O8(OH)5] has the same formula but is not isomorphous with Cf[B6O8(OH)5] (table S1), but distinct from all other actinide borates previously synthesized. The Bk(III) ions are present as eight-coordinate, distorted square antiprisms (Fig. 1B), in contrast to the nine- and ten-coordinate motifs in the Pu(III), Am(III), and Cm(III) borates (9, 2527).

Thermodynamics of complexation

The thermodynamics of Bk(III) complexation with dipicolinic acid are presented in Table 3, with the association constants presented in table S2. The Bk(III) β101 [Bk(DPA)+], β102 [Bk(DPA)2], and β103 [Bk(DPA)33−] constants at 25°C and 1.0 M ionic strength are larger than values reported for Sm(III), Eu(III), or Gd(III). These lanthanides have ionic radii comparable to those proposed for Bk(III) (28). The enthalpic binding contribution for the berkelium 1:3 DPA complex is more exothermic than observed with any of the lanthanide dipicolinic acid complexes (23). The 1:1 and 1:2 complexes are comparable or slightly more exothermic than other lanthanide DPA complexes.

Table 3 Thermodynamic parameters for Bk(III)-dipicolinic acid complexation at 1.0 M ionic strength and various temperatures (in molality).

View this table:

The correlation between coordination exothermicity and degree of bonding covalency is a subject of current debate. Comparisons of the interactions of aromatic nitrogen donors with Nd(III) and Am(III), using 2-amino-4,6-di-(pyridin-2-yl)-1,3,5-triazine (ADPTZ) have been assessed (2831). The ADPTZ studies show the formation of the Am(III)-ADPTZ complex to be 9 kJ·mol‒1 more exothermic than the Nd(III)-ADPTZ complex. When the thermodynamics of the 1:3 berkelium and samarium DPA complexes are considered, a Δ(ΔH)Bk/Sm for the 1:3 metal:dipicolinic acid complex of 9 kJ·mol‒1 is observed.

Experimental electronic structure analysis

The absence of the greenish self-luminescence of Bk(III) upon complexation by DPA is an indication of increased ion-ligand interactions that alter Bk(III) electronic structure and transition dynamics. Absorption, excitation, and photoluminescence spectra are shown in Fig. 2 and in figs. S3 and S4. The low-lying energy levels (<20,000 cm‒1 or 2.5 eV) of the 5f8 electronic states do not differ grossly from the corresponding states in the aquo complex (11), indicating relatively small changes in ligand-field splitting. Similar to the 5f photoluminescence in Cm(HDPA)3 (6, 20), the narrowband emission near 14,700 cm‒1 (680 nm) most likely arises from an intra-5f8 transition. As shown in Fig. 2C, the 5000 cm‒1 energy gap between the emitting state and the next 5f8 state is critical for the red luminescence to be competitive with nonradiative phonon relaxation, whereas the smaller gap of ~35,000 cm‒1 above the emitting state effectively eliminates green luminescence from the excited state near 20,000 cm‒1 (11, 12).

Fig. 2 Optical spectra of Bk(HDPA)3·nH2O obtained from a single crystal.

(A) Room-temperature absorption. (B) Photoluminescence spectrum upon excitation at 420 nm (at 110 K) and excitation spectra monitored at 670 nm. (C) Predicted energy levels of 5f8 and 5f76d states of Bk(III) in DPA and assignment of the 680-nm photoluminescence.

However, starting from 21,000 cm‒1 in the blue and ultraviolet (UV) region, a broad band lacking 5f characteristics appears in the absorption and excitation spectra. The 6d levels of Bk(III) are much lower in energy than those of Am(III) or Cf(III) with the same ligands (8, 11, 12, 32), and we attribute this transition in the Bk(III) spectra to 5f8 to 5f7 6d transitions. The narrow and intense peak at 21,350 cm‒1 in the excitation spectrum has an energy and bandwidth consistent with the scheme of the 5f8 states but an intensity on the same scale as the parity allowed 5f-6d transitions. These observations are consistent with crystal-field induced coupling of the 5f and 6d orbitals, which is expected in a molecule with D3 symmetry (11, 12). The overlap of the 5f8 and 5f76d states enhances this orbital hybridization. The low-lying 6d states and the enhanced 5f-6d hybridization contribute to Bk(III)-ligand bonding and coordination and therefore, put Bk(III) in a unique position in the heavy actinide group. These studies also provide a direct comparison between Bk(III) photoluminescence and the in-growth of Cf(III)-based luminescence. The latter is ligand-to-metal charge transfer–based and is therefore an allowed transition. In contrast, although the 5f/6d mixing does provide some relaxation of the selection rules and hence greater intensity than a pure intra-f transition, the intensity from the Cf(III) is larger than that of the Bk(III) despite a sample composition of ~97% 249Bk and only ~3% 249Cf.

Magnetic properties

To further elucidate the ligand-field effects on Bk(III), as well as enhance our understanding of electronic structure, magnetic susceptibility data were measured in the 4 to 300 K temperature range for Bk(HDPA)3·nH2O and 7 to 300 K range for Bk[B6O8(OH)5] under an applied field of 1 kOe (Fig. 3). The experimental μeff values of 9.24 μB {Bk[B6O8(OH)5]} and 9.27 μB [Bk(HDPA)3·nH2O] at 300 K are close to the simulated value of 9.32 μB derived from spectroscopic data by Carnall (12). The susceptibility data were fitted to a full-model Hamiltonian (33), where the relevant spin-orbit coupling parameter (ζ = 3210 cm−1) and Slater-Condon parameters (F2 = 57,697 cm−1, F4 = 45,969 cm−1, and F6 = 32,876 cm−1) were chosen on the basis of the optical spectra of BkCl3 (11, 12) and used as constants. To restrict the number of independent ligand-field parameters, a least-squares fit of the magnetic data for Bk(HDPA)3·nH2O requires assumption of an idealized C3v symmetry for the BkN3O6 coordination sphere. Initial sets of most ligand-field (Bkq) (Wybourne notation) parameters were again adopted from spectroscopically determined energy levels for BkCl3; the Bk3+ site symmetry reduction from D3h (BkCl3) to C3v mandates two additional ligand-field parameters (B43 and B63). The signs of the Bkq values were assigned from point-charge electrostatic model calculations for a regular tricapped trigonal prism and fixed throughout the fitting procedure. This approach resulted in a near-perfect fit (quality of fit, SQ = 0.5%), yielding a ligand field characterized by B20 = 59 cm−1, B40 = –870 cm−1, B43 = 150 cm−1, B60 = –2310 cm−1, B63 = –420 cm−1, and B66 = 800 cm–1. The composition of the ground term [7F (70%); 5G (23%)] is in good agreement with CASSCF-SO results (vide infra). The total splitting of the J = 6 ground state into mJ substates amounts to 147.9 cm–1, with a small energy gap of 4.9 cm–1 between the ground state and the first excited state. The observed temperature dependence of μeff (~T1/2) below 15 K arises from the composition of the ground state (mixture of 50% mJ = 0 and 50% mJ = 6 states), not from exchange interactions.

Fig. 3 Magnetic properties.

Magnetic moment μeff versus T and inverse susceptibility versus T (inset) of polycrystalline samples of (A) Bk(HDPA)3·nH2O and (B) Bk[B6O8(OH)5] at 1 kOe. Open circles, experimental data; solid black lines, least-squares fits to full model Hamiltonian. The energy-level diagrams display the splitting of the lowest multiplet states and their composition. The dashed black line represents the single-ion contribution of the Bk(III) centers in Bk[B6O8(OH)5] in the absence of (antiferromagnetic) coupling interactions.

For Bk[B6O8(OH)5], bridging borate anions provide exchange pathways between neighboring Bk3+ centers; therefore, 5f-5f coupling interactions were assumed to be significant and were accounted for in a molecular-field approach. Approximating the distorted square-antiprismatic geometry of the BkO8 environments as C4-symmetric, an analogous least-squares fit (SQ = 0.7%) then yielded B20 = 140 cm−1, B40 = –910 cm−1, B44 = –550 cm−1, B60 = 460 cm−1, B64 = –860 cm−1, and the molecular-field parameter λmf = −9400 mol m−3 (corresponding to a Weiss temperature of –1.3 K). In comparison to Bk(HDPA)3·nH2O, the composition of the ground term [7F (73%); 5G (24%)] is nearly identical; however, the ground state multiplet here is exclusively composed of the mJ = 6 states. The total splitting of the 7F6 Russell-Saunders ground state amounts to 173 cm–1, with a very low separation (0.4 cm–1) between ground and first excited state (Fig. 3).

Theoretical electronic structure analysis

To further understand the bonding in Bk(HDPA)3, the electronic structure was analyzed using a variety of computational approaches. At the density functional theory (DFT) level (B3PW91) (34, 35), geometry optimizations were carried out separately on both independent molecules in the asymmetric unit (tables S3 and S4). The optimized structures are in good agreement with those observed crystallographically, affirming the capability of DFT to reproduce geometry and bond distances even for elements as heavy as berkelium. These calculations were also repeated with the Cm(III) and Cf(III) analogs, with the specific goal of examining trends in the bond distances. These data are provided in table S5. The bonding situation, although similar for both Δ and Δ′, was analyzed by scrutinizing the molecular orbitals with a focus on the berkelium atomic orbitals involved in the bonding. Among others, the two highest bonding orbitals are depicted in Fig. 4. In particular, the 5fz3 orbital is delocalized with a 2p orbital from a carboxylate oxygen atom. These orbitals involve hybrid 7s/6d/5f orbitals on berkelium and are consistent with the energetic availability of the 6d orbitals. Therefore, to probe the contribution of the 6d orbitals to bonding, an f-in-core calculation, as applied to lanthanides, was carried out. In this calculation, the 5f8 configuration is kept frozen in the core of the relativistic effective core potential (RECP) so that the 5f orbital cannot participate in bonding. In the latter calculation, the bonding situation is very similar to that described in Fig. 4 (see fig. S5), indicating that the 6d orbitals are crucial to describing the bonding in Bk(HDPA)3, as is often true in actinide compounds.

Fig. 4 Illustrations of some of the molecular orbitals involved in bonding in Bk(HDPA)3.

These orbitals are singly occupied molecular orbital (SOMO)–11 (left) and SOMO-12 (right). The involvement of the 5fz3 orbital in SOMO-11 should be noted.

The electronic structure and magnetic properties of Bk(HDPA)3 were also probed using relativistic all-electron multireference wave-function theory, including spin-orbit (SO) coupling (see the supplementary materials). The lowest-energy level of the free Bk3+ ion is 7F6, with L = 3, S = 3, and J = L + S = 6. An idealized D3 symmetric ligand environment would split the 13-fold degeneracy (2J + 1) into four non-Kramers doublets (E) and five singlets (A). The SO interaction mixes ion levels with the same J = 6. The calculated electronic states deriving from the 7F6 level are characterized in Table 4 for the 100 K Δ′ structure. The ground state corresponds to a non-Kramers doublet that is slightly split due to the distorted D3 geometry in the crystal. An admixture (~17%) of states from the excited 5G6 level of the ion contributes to each entry in Table 4 (table S6 and fig. S7). The fact that these states derive mainly from the7F6 and 5G6 levels indicates that for the 5f shell the ligand-field interactions are secondary to the SO coupling, in line with the 5f-in-core DFT results. The split Δ′ ground-state doublet can be described by a pseudo-spin ½ Hamiltonian with a large electronic Embedded Image factor along the magnetic axis (Embedded Image). As shown in Fig. 5, the calculated orbital (Embedded Image) and spin (Embedded Image) angular momentum are parallel, as expected, leading to a ground state with a large magnetic anisotropy. The calculated magnetic susceptibility Embedded Image is in reasonable agreement with the experimental data (Fig. 5, inset). Due to the relatively small energetic spread of the states derived from the 7F6 level (417 cm−1) from the ligand field, χ reflects contributions from several electronic states even at relatively low Embedded Image. See figs. S8 and S9 for comparative results of Δ structure.

Table 4 Calculated energetic splitting (cm−1) of the low-energy electronic states of Bk(HDPA)3 and characterization of the wave functions in terms of Bk3+ ion levels.

View this table:
Fig. 5 Calculated magnetic susceptibility χ(T) for the Δ′ structure.

CASSCF-SO calculations. Orbital (Embedded Image) and spin (Embedded Image) magnetization isosurfaces and g factors for the ground-state doublet, calculated using the method described in (36) from the septet states only. Doublet components with Embedded Image along the easy magnetic axis.

Terbium represents the closest analog of berkelium because the trivalent cations possess 4f8 and 5f8 configurations, respectively. However, the distinct electronic properties of these elements is attributable to several factors. First, the 5fz3 orbital of Bk(III) overlaps with the 2p orbital on the oxygen atoms in the dipicolinate ligands. This covalency is absent with the 4f orbitals of Tb(III). These ligand-field effects, however, are secondary to spin-orbit coupling in Bk(III) that creates mixing of the first excited state (5G6) with the ground state. In contrast, the ground state of Tb(HDPA)3 contains negligible contributions of this type. Furthermore, although spin-orbit coupling also dominates the electronic structure of Cf(III), the magnitude of ligand-field splitting can be an order of magnitude larger in Cf(III) than in Bk(III). Ligand-field effects on berkelium are much more similar to that of curium. Hence, although the structural chemistry of berkelium is akin to that of californium, its electronic structure is more similar to that of earlier actinides.

Materials and methods


Caution! 249Bk (t1/2 = 320 d; specific activity = 1.6 · 103 Ci/g) β-decays to 249Cf (t1/2 = 351 y; activity = 4.1 Ci/g), which represents a serious external hazard because of its γ-emission (0.388 MeV). There is also a small α-decay branch for 249Bk that yields 245Am. While this does not contribute in a meaningful way to hazards, 245Am β-decays to 245Cm. Reports attributing luminescence near 600 nm to a second 249Bk emission feature in the red are in error. This peak is from 245Cm (38). All studies with transuranium elements were conducted in a laboratory dedicated to these studies. This laboratory is equipped with HEPA filtered hoods and negative pressure glove boxes that are ported directly into the hoods. A series of counters continually monitor radiation levels in the laboratory. The laboratory is licensed by the State of Florida (an NRC-compliant state). All experiments were carried out with approved safety operating procedures. All free-flowing solids were worked with in glove boxes, and products were only examined when coated with either water or Krytox oil. Thick lead sheets, respirators, and long lead vests were used as much as possible to shield researchers from radiation.

2,6-Pyridinedicarboxylic acid (99%, Sigma-Aldrich), ethanol (100%, Koptec), hydrobromic acid (ACS reagent 48%, Sigma-Aldrich), and berkelium (249Bk) obtained from ORNL in the form of BkCl3 were used without further purification. PTFE-lined Parr 4749 autoclaves with a 10 mL internal volume, and Millipore water were used in all of the following reactions. All solvents that were used in a glove box were sparged with argon.


Bk (4.33 mg, 0.0173 mmol) in the form of BkCl3 was combined with an excess of DPA (17.2 mg, 0.1029 mmol) in 200 μL of a 1:1 mixture of ethanol and water. The reaction mixture was heated in a PTFE-lined Parr 4749 autoclave with a 10 mL internal volume for 4 hours at 150°C, and then slowly cooled to 40°C over a 22-hour period. The reaction was performed inside a negative-pressure glovebox that was surrounded by thick lead sheets. The reaction yielded gold-yellow crystals that had both hexagonal prismatic and columnar habits (see fig. S1).


Bk (4.33 mg, 0.0173 mmol) in the form of BkCl3 was dissolved in deionized water (30 mL) and transferred to a PTFE-lined Parr 4749 autoclave with a 10 mL internal volume. Boric acid (69.6 mg, 1.125 mmol) was added to the autoclave, and the reaction mixture was heated to 240°C for seven days. The reaction was cooled slowly over the course of three days. The product was washed with deionized water to dissolve the excess flux and yield small, yellow-orange crystals with a tablet habit (see fig. S2).

Crystallographic studies

Single crystals of Bk2(HDPA)6·3H2O and Bk[B6O8(OH)5] were glued to Mitegen mounts with epoxy and optically aligned on a Bruker D8 Quest x-ray diffractometer using a digital camera. Initial intensity measurements were performed using a IμS x-ray source (MoKα, λ = 0.71073 Å) with high-brilliance and high-performance focusing multilayered optics. APEXII software was used for determination of the unit cells and data collection control. The intensities of reflections of a sphere were collected by a combination of multiple sets of exposures (frames). Each set had a different ϕ angle for the crystal and each exposure covered a range of 0.5° in ω. Both data collections presented in this paper were taken using a set of standard hemispheres. Single crystals of each compound were run both at room temperature and 100 K. The SAINT software was used for data integration including Lorentz and polarization corrections. The structure was solved by direct methods and refined on F2 by full-matrix least squares techniques using the program suite SHELX (tables S1 and S2). Parameters for Bk are not present in the SHELX software and have to be inputted manually. Solutions were checked for missed symmetry using PLATON (39).

UV−vis-NIR spectroscopy

Single crystals of each compound were placed on separate quartz slides under Krytox oil. The slide was kept inside a Linkam temperature control stage for an extra layer of containment throughout all the measurements. Using a Craic Technologies 20/20 microspectrophotometer the data were collected from 250 to 1700 nm (fig. S3). The exposure time was auto-optimized by the Craic software. Photoluminescence data were collected using the same microspectrophotometer with excitation wavelengths of 420 nm (fig. S4). The Linkam temperature control stage was used to control the temperature from room temperature down to 80 K.

Magnetic susceptibility measurements and magnetochemical modeling

Magnetism measurements were performed on polycrystalline samples using a Quantum Design VSM Magnetic Properties Measurement System under an applied field of 10 kOe for 4 K < 300 K, and for 0 < H < 70 kOe at T = 4 K and 50 K. The samples were sealed inside two different, custom-built Teflon capsules. The outermost capsule was also taped closed using Kapton. In order to extract the intrinsic magnetic behavior of the Bk-containing sample, we subtracted measurements of an empty capsule from those of the filled capsule. All magnetic parameters are given in SI units.

The dc-susceptibility data were analyzed with the program framework CONDON 2.0, employing a complete basis set (full f manifolds, i.e., 3,003 functions for Bk3+) (33). Generally, a full model is required to accurately reproduce all magnetic aspects of 5f compounds, and CONDON accounts for all relevant single-ion effects, in particular the ligand-field effect (Hlf), interelectronic repulsion (Hee), spin-orbit coupling (Hso), and the Zeeman effect of an applied field (Hmag). Whereas for Bk(HDPA)3·nH2O, the nearest-neighbor Bk-Bk distances exceed 1.1 nm, which rules out significant coupling interactions, for Bk[B6O8(OH)5] exchange interactions between neighboring Bk3+ centers in the solid state are considered in the molecular field approximation Embedded Imagewhere χm is the single-center susceptibility and λmf is the molecular field parameter. Positive and negative values of λmf indicate dominant ferromagnetic and antiferromagnetic interactions, respectively.

As full model magnetic simulations of 5f8 systems had not been performed previously, we benchmarked the spin Hamiltonian implemented in CONDON against the more comprehensively parameterized model Hamiltonian used by Carnall for the interpretation of optical spectra of TbCl3 (4f8) and BkCl3 (5f8), where the M3+ ions reside in D3h-symmetric ligand environments (12). Here, the ligand-field operator with reference to the threefold rotation axis for the angular part of the wave function readsEmbedded ImageBkq parameter values from spectroscopically determined energy levels for BkCl3 with D3h symmetry were used as starting values in the magnetochemical fitting procedure. In order to highlight the differences between the 4f and 5f analogs, the 4/5f ions are first assumed as free ions, while both spin-orbit coupling and interelectronic repulsion are considered. This results in temperature-independent Bohr magneton numbers (μeff), where the lanthanide and actinide values clearly differ (fig. S6), with a simultaneous reduction of the main component in the Russell-Saunders ground term. The latter arises from an increase in spin-orbit coupling from 4f to 5f and admixture of excited terms into the ground term. Addition of the ligand-field effect then results in the characteristic temperature dependence of μeff. A critical implication is that a simple approximation like the Curie-Weiss law generally has no physical meaning for such actinide species: For both the free Bk3+ ion and the BkCl3 simulation, the inverse molar susceptibility is linear in temperature (inset, fig. S6) over a broad temperature range, and the extrapolated plots do not pass through the origin. Note that a finite value of an apparent Weiss temperature θ in such cases cannot be interpreted as originating from exchange coupling. The Bk3+ multiplet energy levels resulting from the magnetic simulations match the energies derived from the crystal field analysis by Carnall with high accuracy (SQ = 2%), confirming that the full model spin Hamiltonian indeed reproduces the ligand-field splitting with the spectroscopically derived ligand-field parameterization.

Time-resolved emission, steady-state emission, and excitation spectra

Time-resolved emission, steady-state emission, and excitation spectra were collected at room temperature using an Edinburgh FLS980 spectrometer. The sample, between two microscope slide coverslips, was placed at a 45° angle relative to the excitation source and detector. For excitation and emission spectra, samples were excited using light output from a housed 450 W Xe lamp passed through a single grating (1800 l/mm, 250 nm blaze) Czerny-Turner monochromator. Emission from the sample was first passed through a 515 nm long-pass color filter, then a single grating (1800 l/mm, 500 nm blaze) Czerny-Turner monochromator and detected by a peltier-cooled Hamamatsu R928 photomultiplier tube. Time-resolved emission was monitored using the FLS980’s time-correlated single-photon counting capability (1,024 channels; 5 μs window) with data collection for 10,000 counts. Excitation was provided by an Edinburgh EPL-445 picosecond pulsed diode laser (444.2 nm, 80 ps FWHM) operated at 200 kHz.

Acid association constants

The acid association constants of dipicolinic acid were assessed at appropriate temperatures to allow the calculation of free DPA2- available for complexation. Titrations were performed manually using a Thermo/Ross semi-micro combination pH electrode. To allow work in NaClO4 media, the filling solution of the electrode, originally potassium chloride (KCl) was replaced with 3.0 M sodium chloride (NaCl) to prevent potassium perchlorate (KClO4) precipitation at the glass frit. Titrations were also maintained at 1.0 M ionic strength and were completed under a nitrogen atmosphere to prevent carbon dioxide contamination. The measured mV data were converted to hydrogen ion concentrations using a mV versus pcH calibration curve generated in a strong acid-strong base titration of HClO4 with NaOH at 1.0 M total ionic strength. All titrations were repeated in at least triplicate and data fitting was performed using the fitting program Hyperquad.

Extraction measurements

All complexation thermodynamics were assessed by using competitive solvent extraction investigations with bis-2-ethyl-hexyl-phosphoric acid (HDEHP) dissolved in o-xylene. The extraction constant (KEx) was assessed for berkelium at various temperatures. To assess β101, β102, and β103 metal-DPA stability constants, partitioning of berkelium and europium between the HDEHP/o-xylene organic phase and aqueous phase with increasing DPA concentration was measured. The ionic strength of the aqueous phase was maintained at 1.0 M using NaClO4. All phases were pre-equilibrated with an appropriate aqueous or organic phase prior to use in the distribution study at the temperature of a given study. Pre-equilibration contact times were five minutes and contact times for thermodynamic measurements were fifteen minutes. The pcH of the aqueous phase was measured after contact by using a series of standardized acid solutions at 1 M NaClO4. Conversions from molality were afforded by density determinations at 22°C. The partitioning of 249Bk was monitored using a Packard 2500 Liquid Scintillation counter. All thermodynamic constants were fit in QtiPlot using nonlinear regression model weighting the distribution data (D values) using w = 1/σ2 weighting. Metal KEx values were fitted assuming equilibria and mass balance relationships previously established in the literature (24).

DFT computational details

All the structures reported in this study were fully optimized with the Becke’s three-parameter hybrid functional (34) combined with the nonlocal correlation functional provided by Perdew/Wang (denoted as B3PW91) (35). To represent the berkelium atom, a relativistic energy-consistent small-core pseudopotential obtained from the Stuttgart-Köln ECP library was used in combination with its adapted segmented basis set (4042). For the f-in-core calculations in which the berkelium’s oxidation state is fixed to +3, the corresponding 5f-in-large-core ECP (augmented by a f polarization function, α = 1.0) was used (43). For the remaining atoms the 6-31G(d,p) basis set was used (4448). In all computations, no constraints were imposed on the geometry. All stationary points have been identified as minima (number of imaginary frequencies Nimag = 0). The vibrational modes and the corresponding frequencies are based on a harmonic force field. Enthalpy energies were obtained at T = 298.15 K within the harmonic approximation. GAUSSIAN09 program suite was used in all calculations (47). Finally, for the 3D representation of the structures, the Chemcraft (49) program was used, as well as for the visualization of the molecular orbitals. For details regarding the CASSCF calculations, see the supplementary materials.

Supplementary Materials

Figs. S1 to S9

Tables S1 to S6

References (5055)

References and Notes

Acknowledgments: This material is based on work supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences, Heavy Elements Chemistry Program under Award Number DE-FG02-13ER16414 (Florida State University) and DE-SC0012039 (Colorado School of Mines), and DE-SC0001136 (formerly DE-FG02-09ER16066) (F.G. and J.A.). M.S. and P.K. were supported by European Research Council StG 308051 MOLSPINTRON. The isotopes used in this research were supplied by the U.S. Department of Energy, Office of Science, by the Isotope Program in the Office of Nuclear Physics. The 249Bk was provided to Florida State University and the Colorado School of Mines via the Isotope Development and Production for Research and Applications Program through the Radiochemical Engineering and Development Center at Oak Ridge National Laboratory. We are especially grateful for assistance and supervision by the Office of Environmental Health and Safety at Florida State University/Colorado School of Mines and the Office of Radiation Safety for their facilitation of these studies. Magnetization measurements using the vibrating sample magnetometer SQUID magnetic properties measurement system were performed at the National High Magnetic Field Laboratory, which is supported by National Science Foundation Cooperative Agreement no. DMR-1157490, the State of Florida, and the U.S. Department of Energy. We are indebted to the Office of Safety at the National High Magnetic Field Laboratory for helping to facilitate these studies as well. J.C.W. is supported by the National Science Foundation Graduate Research Fellowship under grant no. DGE-1449440. This research was support in part by an appointment to the CBFO Fellowship Program, sponsored by the U.S. Department of Energy and administered by the Oak Ridge Institute for Science and Education. Metrical parameters for the structures of Bk(HDPA)3·nH2O and Bk[B6O8(OH)5] are available free of charge from the Cambridge Crystallographic Data Centre under accession numbers CCDC-1451021 and 1490887, respectively.
View Abstract

Stay Connected to Science

Navigate This Article