Powerful tool for probing molecules

X-ray absorption fine structure spectroscopy determines local structure and bonding in actinide-oxide structural variants

Actinide oxides (AnO₂) have crystalline structures that are relatively open. They are potentially "reactive" because the openness of their structures makes them easily invaded by or reactive toward other atoms. Actinide oxides are of great scientific interest because understanding how oxygen atoms are added and removed in actinide-oxide solids is essential for predicting their behavior. For example, some scientists believe that the presence of small amounts of plutonium(VI) in nonstoichiometric plutonium dioxide (PuO_2+x) may increase the solubility of this environmentally hazardous compound. Likewise, corrosion and other chemical reactions are affected by changes in molecular structure of these solids.

Actinide oxides have relatively open crystalline structures that make them easily invaded by other atoms. The actinide-oxide (AnO₂) lattice structure on the left exhibits a fluorite crystalline structure with alternating empty and occupied oxygen cubes. The lattice structure on the right shows how Pu₂O₃ is produced from plutonium dioxide (PuO₂) by removing every fourth oxygen atom from the lattice.

Data obtained through x-ray diffraction show that actinide oxides of general composition AnO_2-x form a cubic fluorite-type phase over the composition range of AnO_1.6 to AnO_2.0 for thorium through californium, as illustrated in the figures below. The stoichiometric AnO_2.0 lattice consists of a face-centered cubic (fcc) arrangement of metal atoms in six-fold "octahedral" lattice sites. Each metal atom is surrounded by eight oxygen atoms at the corners of a cube, and each adjacent cube has a vacant metal site. The intermediate phase compositions and structures displaying oxygen:metal ratios between 1.6 and 2.0 are realized by omission of oxygen atoms in a regular way. Except for small shifts away from the vacant oxygen sites, the metal atoms always maintain their fcc positions.

Some actinide-oxide solids can be further oxidized, as in the case of uranium or plutonium dioxide, with the simultaneous addition of oxygen atoms to the lattice. For the composition range AnO_2+x, where x falls between 0 and 0.25, x-ray diffraction analyses reveal a single fcc phase with a fluorite-related structure and a slightly expanded lattice, consistent with the formation of a solid-solution for AnO_2+x. Many actinide-oxide-type materials also react with water that can lead to substitution of the O^2- ion with a hydroxide ion (OH^-) or water molecule (H₂O), leading to a more complex overall description of coordination where y and z are stoichiometry descriptors that account for this net coordination of OH^- and H₂O, respectively (AnO_2+x-y(OH)_2y.zH₂O for example).

One of the more remarkable aspects of the class of solids AnO_2�x is the ability to add or remove oxygen atoms with simultaneous oxidation or reduction of the actinide ions while conserving the basic crystallographic structure. In many cases the addition or removal of oxygen (O) atoms results in diffraction patterns from the cubic actinide sublattice that translate into changes of only a few hundredths of an angstrom (�) in lattice constant. This conservation of the structure indicates the formation of structurally similar phases in which the oxygen atoms occupy interstitial lattice sites that retain most of the functional features of the original unit cell.

For example, cubic Pu₂O₃ is produced from PuO₂ by the sequential removal of every fourth oxygen atom from the lattice, as shown in the right-hand figure on the previous page. Similarly, sixteen distinct phases have been identified as the oxygen content in uranium oxide compounds increases between UO₂ and UO₃. These structures transform into each other by small expansions and contractions of the uranium sublattice, apparently induced by different, ordered arrangements of the oxygen atoms as more are added to the original uranium dioxide structure. The extra oxygen atoms or vacancies within the crystal form clusters within the host crystal because of the stability of these putative phases relative to a random distribution of oxygen atoms.

Based on interpretations of both x-ray and neutron diffraction data following addition or loss of oxygen atoms, highly detailed mechanisms of phase and structure formation have been described for the uranium dioxide UO_2+x system. The salient attribute of these mechanisms is that extra lattice oxygen atoms are added through displacements and rearrangements within the oxygen sublattice, using the empty space of the unoccupied cubic sites. The additional charge on the uranium ion tends to be dispersed rather than localized and is accommodated by small (less than 0.15 �) reductions in some uranium-oxygen bond lengths (balanced by expansions of others) and increases the number of nearest neighbors bonded to the uranium atoms in the lattice (i.e., the coordination number). There are, however, problems with these models.

Conceptually, this delocalized charge distribution is inconsistent with the known molecular coordination chemistry of the lighter actinides. In molecular complexes of higher-valent light actinides, discrete oxo groups are found at very short (less than 1.85 �) actinide-oxygen distances as well as at lower total coordination numbers. The higher actinide charge is thus stabilized by fewer oxygen near-neighbors but some shorter, much more covalent bonds.

It is difficult to determine the correct structures of the disordered, mixed-valent AnO_2+x solids by conventional crystallographic analysis from diffraction data alone, typically because a range of structural solutions is possible. In highly ordered materials, atoms in a crystalline lattice exhibit very regular repeat units; this is called long-range order and is typically studied using x-ray or neutron diffraction techniques. To directly probe chemical speciation and local structure that will include aperiodic components in AnO_2+x-y(OH)_2y.zH₂O solids independent of their long-range order, we have made extensive use of extended x-ray absorption fine structure (EXAFS) spectroscopy. This experiment is performed at Stanford Synchrotron Radiation Laboratory and can provide information about the metal oxidation state, bond distances, and types and numbers of atoms coordinated to the metal center. The application of this technique provides unique characterization and insight into local bonding in a number of PuO_2+x and UO_2+x solids where x varies from 0 to 0.25. (See ARQ 1st quarter 2004 for applications in actinide solution chemistry.)

For plutonium, we have investigated more than two dozen PuO_2+x compounds prepared by a variety of methods, including hydrolysis and precipitation of the aqueous plutonium(IV) ion, and by the heterogeneous oxidation of plutonium metal and oxide with water vapor, oxygen, or both. The first finding, based on a simple comparison of the spectra with the one calculated from the crystallographic structure, is that simple, highly ordered plutonium dioxide is rather rare and is not typically obtained simply by firing or calcining plutonium compounds in air at high temperatures.

Extended x-ray absorption fine structure (EXAFS) data shows the splitting of the first oxygen shell (indicated by the double peaks) in the PuO_2.26 spectrum (purple) . The spectra also show that the relative amplitude of the peaks decreases as x increases from PuO_2.00 to PuO_2.26. The periodic oscillations also become more washed out as x increases, meaning the atoms become more disordered.

Fourier transforms of EXAFS data for selected plutonium oxide compounds are shown in the figure below. (See Page 24 for an explanation of XAFS data analysis.) For stoichiometric, ordered PuO_2.0, the first peak in the Fourier transform is the contribution of the eight nearest-neighbor oxygen atoms at 2.33 �, well separated from the more distant second nearest-neighbor peak of twelve plutonium atoms at 3.80 � (the peaks in the figure at left are all phase-shifted to lower R).

Regular features from the well-ordered extended structure subsequently continue out through very high distance from the central absorbing atom. This spectrum is consistent with the crystal structure shown on Page 16 on the left. An alternative drawing of the basic structure from the perspective of a central plutonium atom is shown at different internuclear distances on the next page to emphasize the different "shells" of atoms observed in an EXAFS experiment.

However, as x increases from PuO_2.00 to PuO_2.25, the amplitudes of all of the peaks in the Fourier transform decrease monotonically, indicative of diminished order via displacements of the plutonium and oxygen atoms from their lattice sites coupled to the incorporation of the nonstoichiometric oxygen atoms into interstitial, essentially defect, sites. What contradicts the current models in this process is the splitting of the first oxygen shell and the appearance of a short plutonium-oxygen bond distance of 1.84 �.

Diffraction techniques that probe long-range order have never observed this phenomenon, but the bond distance is similar to those found in discrete plutonium-oxygen bonds in molecular compounds that range from 1.73 � for plutonium(VI) compounds to 1.85 � for plutonium(V) compounds. These results show that as the plutonium center becomes partially oxidized in PuO2+x, there is a strong driving force to form short, strong, covalent plutonium-oxygen bonds.

The addition of extra lattice oxygen into the solids produces a number of other systematic changes in the spectra from the various plutonium oxide compounds. Disorder in the nearest-neighbor oxygen shell resulting from plutonium-oxygen bonding that can be assigned to hydroxyl and possibly water ligands is evident in some cases.

Furthermore, the directly coordinated oxygen near-neighbor shells are supplemented by longer plutonium-oxygen bond distances that result from the mirroring of these displacements on the neighboring plutonium atoms and show up as peaks between the crystallographic oxygen and plutonium shells at 1.8 and 3.8 �, respectively (the peaks are phase shifted from their actual distances).

What is unexpected, however, is that these additional shells of oxygen atoms are often found in materials following thermal treatments (to 1,000 degrees Celsius) in an oxygen-containing environment. Following this calcination step, nominal plutonium dioxide solids are cooled in an ambient processing environment. Unless preventative steps are taken, such as cooling in an extremely dry ambient atmosphere with immediate packaging, the rates of water uptake, hydrolysis, and coordination to plutonium ions within the lattice are sufficient to generate measurable quantities of extra lattice oxygen as protonated ligands (ligands with added protons) that are not associated with oxidation.

In addition, the plutonium-hydroxide bonds remain stable to moderately high temperatures (typically 600 degrees Celsius; see the following article). Thus, plutonium dioxide displays a tenacious affinity for water-derived extra lattice oxygen constituents.

X-ray absorption fine structure (XAFS) spectroscopy provides information about the number of atoms and their interatomic distance from a central target atom. The crystalline structure of plutonium dioxide (PuO2) shows cubic symmetry. There are eight near-neighbor oxygen atoms (shown in red) that all sit at 2.33 angstroms (�) from the central plutonium atom (black). There are also 12 neighboring plutonium atoms (gray) at an interatomic distance of 3.81 � from the central plutonium atom. It is this combination of the number of near-neighbor atoms, their elemental identities, and their interatomic distances that uniquely define the chemical structure using XAFS spectroscopy.

For PuO_2+x compounds processed at elevated temperature, the reduced XAFS data exhibit narrower features with much greater relative amplitudes for the nearest-neighbor oxygen peak, thus indicating a corresponding higher degree of local order in the plutonium-oxygen distributions in the compounds.

For the PuO_2+x spectra shown on the previous page, the near-neighbor, singly bound oxygen region with plutonium-oxygen bond distances between 2.15 and 2.45 � always requires at least two scattering shell contributors for an adequate fit, with one above and one below 2.3 �. The 0.10 to 0.18 � separation between these two shells is right at the 0.11 � data resolution limit. In addition, fits to some spectra are significantly improved by including a third oxygen shell to this region. In these cases, the short plutonium-oxygen bond distance decreases to less than 2.15 �, one remains between 2.25 and 2.30 �, and the longer one shifts above 2.30 �.

These limitations demonstrate that EXAFS also suffers from the problem that more than one solution can be obtained for the overall distribution function described by multiple curve fits. The solution(s) does not necessarily fit to the precise values of the explicit parameters such as numbers of coordinating atoms, distances, and Debye-Waller factors (the Debye-Waller factor of a near-neighbor scatter accounts for the mean-square fluctuations and thermal disorder). Nevertheless, general aspects of the XAFS curve-fit results indicate that the type of ligand largely determines bond lengths, as is typical of coordination chemistry of molecular complexes.

In fact, the plutonium-oxygen bond lengths determined by curve fits of the XAFS data are so similar to those found in molecular coordination compounds that they can be assigned to specific ligands based on these internuclear distances: coordinated hydroxyls for the plutonium-oxygen bond distance between 2.25 and 2.35 �, coordinated oxide ions for the plutonium-oxygen bond distance between 2.35 and 2.40 �, and coordinated water for the plutonium-oxygen bond distance beyond 2.45 �. This scheme is not necessarily rigorous; the longer distance(s) could also reflect particular bridging ligand geometries where a coordinated oxygen entity in effect bridges two plutonium atoms with an elongation of bond length.

Although these results show the affinity of the plutonium in PuO₂ for water, the primary issue in PuO_2+x remains the disposition of the added oxygen associated with oxidation that increases the valence of the plutonium. The critical comparative sample is PuO2.26. The first shell of oxygen near-neighbor atoms shows two clearly resolved peaks of relatively low but similar amplitude; the plutonium-oxygen nearest-neighbor distribution is obviously more complicated than the single oxygen shell of the highly ordered PuO₂. The peak at lower R, whose large relative amplitude clearly suggests a prominent structural feature, occurs almost 0.6 � below the oxygen peak obtained from the ordered PuO₂ samples.

Curve fits show that these two peaks require four shells to be completely fit. Despite the limitations of the curve-fitting process, three of these correspond to singly bound oxygen ligands, with overlapping plutonium-oxygen bond distances of 2.13, 2.28, and 2.41 �, giving a broad oxygen distribution. The fourth shell, completely resolved from the others and essential in fitting the lower peak in the Fourier transform, has a plutonium-oxygen bond distance of 1.84 �. Bond lengths this short in both molecular complexes and extended solids are found only for the multiply bound oxo groups associated with valences of (V) and (VI) in MO₂⁺ and MO₂²⁺ units.

Oxo groups can be defined as oxygen shells with actinide(V, VI, VII)-oxygen bond distances less than or equal to 1.9 � that involve actinide-oxygen bonds of multiple-bond order, and that usually impose additional constraints on actinide coordination geometry in unit cell models. The presence of oxo groups in PuO_2.26 demonstrates that an oxygen shell near this distance should be included in the basic structural model that is the starting point for the curve fits for the spectra of all samples. A more detailed inspection of this region shows that there are trends, especially in the real component of the transform, by which the presence of an oxo shell can be directly inferred from the spectra without resort to curve fitting, further corroborating its presence.

The observation of an apparently stable, disordered, mixed-valence form of PuO_2+x without any indication of phase separation poses a conundrum in our understanding of this material. Recent independent electronic-structure calculations place the extra lattice oxygen in the interstitial site formed by the empty cube of oxygen atoms and without consideration of oxo group formation. These calculations are inconsistent with XAFS measurements; a 1.9-� plutonium-oxygen bond distance places the oxygen within the oxygen cube already occupied by a plutonium ion.

In solid crystals comprised of many unit cells, cooperative and collective effects that include entropic statistical effects can be important, such as-type clusters. The microscopic structural attributes identified by EXAFS in the PuO_2+x series suggest that entropy in combination with different bonding types, the combination of which will only become significant in crystals containing many more atoms than can be included in the calculations, may be a significant contributing factor in their structural evolution. These structural attributes must, however, be inferred because XAFS data does not give direct information on nanoscale ordering.

Although there are exceptions, the most common structural motif for actinide ions in their higher oxidation states (V and VI) is the trans dioxo geometry, in which two oxygen atoms form strong covalent bonds with the actinide ion to form a linear actinyl unit, O=An=O. All other donor atoms form bonds to the molecule in an equatorial plane that is perpendicular to the linear O=An=O unit. The equatorial plane can accommodate four, five, or six atoms. In the figure at left the local bonding about a central actinide AnO₂²⁺ ion is shown for PuO₂Cl₄^2-, PuO₂(H₂O)₅²⁺, and PuO₂(CO₃)₃^4- as examples of four, five, and six coordination in the equatorial plane.

Actinide ions in the higher oxidation states (V and VI) generally exist as trans dioxo (or actinyl) ions. Two oxygen atoms form strong covalent bonds with the actinide metal center to form a linear actinyl unit, O=An=O. All other ligands form bonds within an equatorial plane perpendicular to the linear O=An=O unit. The equatorial plane typically contains four, five, or six other donor atoms as illustrated for (from top to bottom) PuO₂Cl₄^2- (four chloride atoms), PuO₂(H₂O)₅²⁺ (five oxygen atoms from water molecules), and PuO2(CO3)34- (six oxygen atoms from carbonate ligands). In these examples the metal atoms are black, chlorides are blue, oxygens are red, and hydrogens are gray.

Scenarios can be devised for incorporating this PuO₂⁺ local bonding geometry in the extended plutonium dioxide lattice. The easiest way is to place the extra lattice oxygen atoms midway between two plutonium ions, so that they reside on the midpoint of a [100] directional vector of a PuO₈ cube. This is indicated schematically in the figures on the next page. The top figure shows three adjacent cubes of oxygen atoms where the center cube is occupied by a central plutonium atom and the two adjacent cubes are vacant. In the second figure, extra oxygen atoms (blue) are added to these vacancies. The resulting Pu-O bond distance of 2.7 � is clearly not consistent with the EXAFS data.

The third structure shows the addition of two oxo groups with Pu-O = 1.9 � normal to a [110] plane of the PuO₈ cube. In this case the four planar equatorial ligands are already in place. The oxo groups are inserted between the O₂^� ions at the vertices of the PuO₈ cube that are not involved in the equatorial coordination, giving 1.95 � oxygen-oxygen distances. Since these oxo groups are midway between two plutonium ions, they would form the same type of bond to the plutonium in the diagonally adjacent cell and what began as one plutonium(V) oxo complex automatically becomes a chain of them.

The fourth structure shows an alternative model, the conversion of two of the O₂^- ions at opposite vertices of the PuO₈ cube into oxos so that the O=Pu=O vector is normal to a [111] plane through the cube. In this case they are wholly within the unit cell of the central plutonium(V), avoiding the problems of bridging to neighboring plutonium ions and overlapping with other oxygen ions. However, creation of the equatorial plane requires that some of the O₂^- ions from the other vertices must be displaced into the [111] plane to create a set of equatorial ligands and the others displaced towards the adjacent cation holes. The formation of the plane now creates additional plutonium-oxygen distances of less than 1.9 �, with asymmetric bridging. The collective effect here is that each new oxo group forces the formation of additional ones with neighboring plutonium atoms so that again a structural repeat chain develops.

This portion of the crystal structure of plutonium dioxide (PuO₂) illustrates the face-centered cubic arrangement of metal atoms in black, and three adjacent cubes of oxygen atoms (red). The figure on top shows the center cube occupied by one plutonium atom generating a central PuO₈ cube, while the two neighboring O8 cubes are vacant. In the second figure, two interstitial oxygen atoms (light blue) are added into the vacancy positions. This model would give a Pu-O distance of 2.7 angstroms (�), which is not consistent with the EXAFS data.

The third figure illustrates the addition of two interstitial oxygen atoms (light blue) normal to a [110] plane the central PuO₈ cube with plutonium-oxygen (Pu-O) distances of 1.9 �. In this case the four equatorial ligands (pink) are already in place. Since these new oxygen atoms bisect a cube edge, the new oxo groups are midway between two plutonium atoms, which results in the formation of a chain. The bottom figure shows the conversion of two oxygen atoms into short Pu=O groups along a body diagonal (normal to a [111] plane) of the central PuO8 cube. This scenario requires distortion of the O₈ cube to create equatorial sites (pink).

The essential aspect of this plausible structural model is that it demonstrates, within the crystal and under the assumption that the coordination geometry is analogous to that in molecular complexes, that the formation of an oxo group on a particular plutonium ion forces the formation of additional oxo groups on neighboring plutonium ions. This sequence propagates through the crystal to form chains, filaments, or other types of clusters containing the plutonium(V)-oxo component.

This structural model assumes conservation of the plutonium sublattice, which follows from the diffraction results that show disorder but not the formation of a new phase. This result holds true for the EXAFS as well. Whereas, relative to the crystal, the oxygen shell forms a multisite distribution with the oxygen atoms organized around a set of specific plutonium-oxygen distances, there is no evidence for any new plutonium-plutonium distances or significant shift from the crystallographic plutonium-plutonium distance.

A complete description of all the detail embedded in the Fourier transform spectra is beyond the scope of this article; however, the prevailing thought is that the diminution in total amplitude in the Fourier transform spectra represents a material with coexisting crystalline and glassy characteristics. The poor coherence in the latter arrangement of atoms would result in a material with diminished diffraction characteristics so that the diffraction pattern displays only a single phase. The wide range of plutonium-oxygen near-neighbor distances is also consistent with a periodicity and the diffuse scattering from glasslike materials.

These off-stoichiometric plutonium dioxide subunit structures could exhibit some local order, but their small size and a periodic arrangement precludes efficient diffraction from them. The preponderance of oxo groups, as well as coordinated water, is suggested to therefore reside in domains of disordered material coexisting with more stoichiometric plutonium dioxide, thus resulting in a material that is heterogeneous on the nanometer scale.

Comparative behavior: PuO_2+x versus UO_2+x

It is interesting to compare the local speciation and structural characteristics of PuO_2+x with the extensively studied UO_2+x system. We have also performed XAFS measurements on a series of UO_2+x samples, where x varies from 0 to 0.20. The presence of multiple, coexisting, ordered phases correlated with the oxygen stoichiometry is well understood in this system, and specific structural models for these phases and the mechanism of oxygen addition have been developed.

Controlling the oxygen activity during calcination can produce exact oxygen stoichiometries in UO_2+x. Domains of the UO_2.25/U₄O₉ phase have been confirmed by x-ray diffraction with a slightly contracted lattice constant relative to that of the original UO₂. This phase separation was easy to see and corroborated in XAFS data, where the high R features in the Fourier transform from the extended, ordered uranium dioxide structure all decrease linearly as the oxygen content increases.

A primary characteristic of extra lattice oxygen in both PuO_2+x and UO_2+x is the clear signature of oxo groups as x approaches 0.25. For both actinide oxides, the extra lattice oxygen is incorporated with the formation of oxo complexes analogous to the higher-valent molecular complexes. This incorporation is demonstrated by the Fourier transform spectra for the UO_2+x (analogous to those of PuO_2+x), which unequivocally show a resolved absorption feature at lower R comparable in amplitude to that of the O₂^- contribution at high x (oxygen content) where the disorder has substantially lowered the amplitude in the Fourier transform spectra from the original oxygen shell.

For UO_2+x, the appearance of this new peak at low R is accompanied by the growth of other features at higher R that signify the simultaneous formation of oxygen shells with uranium-oxygen distances greater than 2.4� and a multisite uranium-oxygen distribution. Curve fits demonstrate that this low R feature is well fit by an oxygen shell with a uranium-oxygen distance of about 1.75 �, significantly shorter than the oxo distance in PuO_2+x and more consistent with the oxo bond distances for uranium(VI).

This result is significant in comparison to plutonium, where plutonium(V) is the preferred high-valence species resulting from oxygen or water oxidation of plutonium dioxide (see the previous and following articles). Another characteristic similar to PuO_2+x is that the total uranium amplitude also decreases with increasing x, implying the separation into a phase where the uranium sublattice retains the UO₂ structure and a separate phase that is more glasslike. The UO_2+x compounds, however, do not show the hydroxyl-water ligation common in the plutonium system and either have a much lower affinity for water or were well isolated from it during the entire course of their preparation.

Conclusion

From a critical interpretation of XAFS data, we have been able to identify specific, local coordination components for uranium and plutonium in the AnO_2+x solids. These components have included coordinated oxide ions, coordinated hydroxyls, and coordinated water. Furthermore, our data show that participation of repeat chain subunits of actinide ion coordinated to oxo subunits contribute greatly to the overall structural development of these materials.

Our research has shown that these high-valence actinide oxide complexes can be stabilized through the development of a complex heterogeneous environment on the nanometer-length scale. Much of this additional local structure possesses glasslike characteristics. More detailed information on the structures, additional characteristics of the reactivity, the behavior of other actinides, and the origins of these effects await further study.

This article was contributed by Steven Conradson of the Materials Science and Technology Division.

NMT | LANL | DOE
Phone Book | Search | Help/Info

L O S  A L A M O S  N A T I O N A L   L A B O R A T O R Y
Operated by the University of California for the US Department of Energy