Toward the Insulator-Metal Transition

At higher levels of alkali metal uptake, it was originally thought that intracrystalline, ultrafine particles (of colloidal dimension) were formed in the large supercages of zeolites X and Y.^1,2 Interestingly, Kruger⁵⁶ in 1938 postulated that concentrated alkali metal/ammonia solutions were colloidal metal systems rather than genuine solutions.⁵⁷ In both cases this view is now superseded! The revised picture for the zeolite systems arises as a result of a detailed magnetic resonance study⁵⁸ of the progressive dissolution of sodium into Na⁺-Y. The emergence of the singlet ESR line in Na_u/Na⁺-Y, originally interpreted as being due to ultrafine colloidal metal particles, was monitored carefully as the sodium concentration u was increased. These observations revealed that, although the faujasite unit cell contains eight potential sites (i.e., the sodalite cages) to solvate excess electrons as Na₄³⁺ units, the ESR singlet is already manifestly present at nominal concentrations as low as u = 3 atoms per unit cell and quite dominant at u = 8 (Figure 3). Its first appearance in the spectrum comes at between u = 1 and u = 3, when only about a quarter of the sodalite cages should contain Na₄³⁺ species.

Figure 3 The ESR spectra of Na+-Y containing (a) 3, (b) 8, (c) 13, and (d) 32 extra sodium atoms per unit cell (ref 58).

Significantly, the ESR singlet appeared⁵⁸ just at the stage where the probability that two Na₄³⁺ centers occupy adjacent sodalite cages becomes appreciable. The unmistakable implication here is that the change in ESR spectrum, represented by the emergence of the singlet resonance, is caused by the interaction of neighboring Na₄³⁺ centers. Excess electrons in Na₄³⁺ centers occupying adjacent sodalite cages are sufficiently close to one another that the wave functions overlap and they are coupled through quantum mechanical exchange forces. If two neighboring Na₄³⁺ centers are coupled in this way, each spin in effect experiences the hyperfine field of eight ²³Na nuclei instead of four, and the magnitude of the interaction with each nucleus is reduced by the same factor, 2. The ESR spectrum expected under such circumstances was simulated by Anderson and Edwards,⁵⁸ who found that, even for just two interacting centers, individual hyperfine lines can hardly be resolved. With quite small "superclusters" (i.e., four to eight interacting Na₄³⁺ clusters) the hyperfine structure of the individual tetrameric ion vanishes and the ESR envelope becomes a smooth symmetric line.

Ursenbach et al.⁵⁹ have estimated the exchange interaction of nearest neighbor paramagnetic Na₄³⁺ clusters at ca. 8 × 10^-7 eV. This value is comparable to the inverse of the observed ²³Na hyperfine coupling constant for Na₄³⁺. Put another way, the ESR time scale is dictated by the hyperfine coupling constant (ca. 32 G) and corresponds to a time scale of ca. 10 ns while the nearest neighboring interaction corresponds to electron exchange on a time scale of ca. 1 ns. We would therefore expect to see a collapse of the hyperfine structure at doping levels where each cluster had at least one cluster neighbor. Strong support for this picture comes from the structural work of Armstrong et al.,⁴⁷ which identifies an array of interacting Na₄³⁺ centers (Figure 4), located in the sodalite cages, as the source of the ESR singlet.

Figure 4 Representation of an array of interacting Na43+ clusters, located in the sodalite cages of zeolite Na+-Y.

Despite this false start the question of metallic behavior in alkali metal loaded zeolites is still firmly on the agenda. It is now known, for example, that dehydrated zeolite Na⁺-X will take up, reversibly, up to 100 extra sodium atoms per unit cell without any significant loss in crystallinity. The ability to achieve such high concentrations of excess electrons within zeolites has led us to speculate on the possibility of a solvent matrix-bound insulator-metal transition,⁵¹ in short, the possible synthesis of a conducting metallic zeolite. Just what degree of metal loading is required for this transformation-and indeed what type of metal would ensue-remains unknown.

Ursenbach et al.⁵⁹ have investigated the species which arise when Na atoms are dissolved in zeolite Na⁺-Y using Car-Parrinello molecular dynamics simulations. These studies indicate that, at high levels of loading/solubility, the excess electrons reside in ionized, high-nuclearity clusters, rather than the neutral metallic particles we had postulated^1,2 in 1984. This work gives a very clear message of an unquestionable tendency toward the formation of extended clusters, which appear to spread continuously throughout the zeolite cages (Figure 5), cavities, and channels, in contrast to the localized excess electron states (e.g., Na₄³⁺) described at low excess electron concentrations. Even at this early stage of investigation, the situation is reminiscent of the behavior both of expanded alkali metal fluids⁶⁰ as the elemental density is continuously increased and of metal-ammonia solutions^27,34 as the solute concentration is increased toward the electrolyte-metal transition. Clearly, the formation of a continuous ribbon of excess electron density throughout the zeolite void structure could be seen as a necessary, if not sufficient,⁶¹ criterion for the onset of metallic character in these systems.

Figure 5 The initial and final configurations for a molecular dynamics simulation of sodium-loaded Na+-Y (taken from ref 59), demonstrating the tendency of cations and excess electrons to spread out and form extended structures rather than the discrete clusters of the starting configuration.

The possibility of observing in alkali metal loaded zeolites a very large periodic array of closely-spaced alkali metal cations, all accessible to delocalized excess electrons, has been noted by Anderson and Edwards.^58,62 This situation may already have been realized in the "cationic continuum" found by Sun and Seff⁶³ in cesium-loaded Cs⁺-X. Similar structures are also seen in potassium-loaded zeolites,^42,53,54 though in all cases the extent of electron delocalization remains unknown. Recently, through microwave cavity loss measurements,⁶⁴ Anderson et al. have demonstrated for the first time that certain zeolites do indeed exhibit an increasing electronic conductivity as more and more alkali atoms are continuously incorporated into the host structure. Already, Terasaki and co-workers^65,66 have interpreted both the optical properties and observed ferromagnetism of potassium-loaded K⁺-A using a model involving itinerant electron states. To our knowledge this is the first recorded example of the occurrence of ferromagnetism centered on alkali metal ions.

A Glance to the Future

There are still a number of important issues to be resolved concerning the behavior of cations and excess electrons within zeolites. One such issue, crucial for an understanding of the dissolution of alkali metal and the concomitant formation of ionic clusters at low concentration, relates to the overall thermodynamics of the process. The spontaneous ionization of, for example, sodium atoms in zeolite Na⁺-Y or Na⁺-X requires a total stabilization/solvation of both excess electron and extra cation to the extent of at least 5-6 eV, the first ionization energy of atomic sodium. We note also that Park et al.⁶⁷ have reported a reaction between dehydrated zeolite and solid alkali metal, and this, inevitably, would require a much higher energy cost, involving the enthalpy of atomization of the metal. The urgent need for thermodynamic data for the solvation process is clear.

For the case of sodium atoms entering Na⁺-Y, the molecular dynamics simulations of Ursenbach et al.⁵⁹ indicate that the excess electron is indeed spontaneously transferred from the incoming atom to the interior of the sodalite cage to form Na₄³⁺, leaving the extra cation coordinated to a suitable coordination site in the supercage. A quantitative estimate of the energies involved in this process was obtained for the situation in which an Na₄³⁺ cluster is in one sodalite cage and the accompanying, excess cation is complexed to this same sodalite cage at an adjacent site in the supercage. Yet again we note important similarities to the metal-ammonia situation;²⁷ this could be viewed as an "ion-pairing" or association process in the solid polar solvent, viz.,

Turning now to the more concentrated samples, we again find a fascinating situation. Here the properties of high-nuclearity cationic clusters are determined both by electronic structure effects, reminiscent of the situation in gas phase clusters, and by the inexorable necessity of the included cluster to seek out suitable coordination sites on the interior surface of the supercages. Part of our own interest in these systems is derived from such considerations where one sees the physics of free clusters naturally merging with the chemistry of extended clusters. A representation of a possible scenario for the heavily loaded samples is given in Figure 5, taken from the molecular dynamics studies of Ursenbach et al.⁵⁹ To initiate these simulations, the authors take as a starting point four sodium trimers placed in separate cage windows, two as Na₃²⁺ clusters and two as neutral Na₃ molecules. Snapshots of the unit cell contents close to the initial and final configurations are shown in Figure 5. Here, all supercage Na⁺ ions are shown, with the extra cations arising from ionized incoming atoms shaded more darkly. Note that the final cluster is appreciably more spatially extended than the starting configuration. Interestingly, all the excess electrons appear spin paired, and it would seem that at this loading level the onset of metallic behavior may be imminent.

Another important feature to emerge, and a primary objective of our initial program, is that the geometric and electronic properties of structures within the zeolite pores may now be suitably "engineered" through the choice of the crystal architecture of the zeolite host. This crystal electronic engineering at the nanoscale is an exciting development within zeolite science and one which we have recently used to our advantage in the quest for ultrathin quantum wires. It was this basic proposition that led us recently to examine the fate of potassium atoms in low-dimensional zeolite structures, most notably zeolite L, where there are channels rather than interconnected cages.⁶⁸ It is found that incoming potassium atoms will spontaneously ionize and the resultant excess electron is then delocalized among the K⁺ cations that line the one-dimensional channels of the host. Interestingly, no K₄³⁺ or similar ions are formed in this instance.

With the ever-expanding corpus of information pertaining to these intriguing solids, the prospect of being able to harness the geometric and electronic properties through the creation of quasi-one-dimensional (or quasi-zero-dimensional) quantum structures, as recently discussed by Kelly,⁶⁹ can be more thoroughly assessed. That zeolites can indeed serve as hosts for atomically thin wires is hardly in dispute, for there is already promising information concerning the novel optical, electronic, and structural properties of individual chains of selenium⁷⁰ and a number of p-block metals^71,72 dispersed within the channels of the zeolite mordenite.

Figure 6 Representation of the structure of zeolite L, showing the location of potassium ions/atoms within the one-dimensional channels.

The potential that alkali-metal-loaded zeolites may have in the realm of heterogeneous catalysis has also yet to be fully explored. Recent work by Simon et al.⁷³ extending an earlier report by Martens et al.⁷⁴ has shown that Na₆⁵⁺-containing zeolite X catalyzes the isomerization of cyclopropane to propene.

Concluding Remarks

In the decade that has elapsed since we embarked in our study of alkali metal ionic clusters in zeolites, significant advances relating to the structural electronic, magnetic, and optical properties of excess electron states in zeolites have been made. Looking to the future, three major themes may be discerned. The first centers on molecular dynamics, where an exciting start has already been made by Ursenbach et al.,⁵⁹ but where there still exists enormous scope for further development. The second relates to new techniques of structure refinement, and in particular to the opportunites that have arisen from the availability of high-flux synchrotron radiation sources. These are especially relevant because one may employ a combined study (usually in situ) of long- and short-range structure by X-ray diffraction and X-ray spectroscopy.⁷⁵ Such approaches, when coupled with combined small-angle and wide-angle X-ray scattering (SAXS/WAXS),^76,77 afford much scope in the study of localized and extended electronic structures within zeolites.

The other major theme to evolve will surely be the use of metal-loaded zeolites as designed reduced-dimensionality electronic structures. Our vision is that metal-containing zeolites (possibly alkali metals, or indeed others from the periodic table) will be strong contenders in M. J. Kelly's quest⁶⁹ for "a dense bundle of quasi 1D conducting wires embedded in a 3D matrix" (see Figure 6). Given the advances in both the synthesis and characterization of metal-loaded zeolites highlighted here, we anticipate that the remarkably diverse range of zeolites (and metals) available will lend themselves naturally to this program. The added attraction of the alkali metal-zeolite system is the high degree of compositionally-induced tuning of structural, electronic and optical properties. What seems abundantly clear is that we can look forward to further major developments in the coming decades in the science and technology of these fascinating solids.

Acknowledgment

We express our thanks and gratitude of all our colleagues who have so effectively participated in the work reviewed here. In particular we thank Robert Armstrong and Lee Woodall for their unstinting help and major contributions to this Account. We thank the EPSRC for financial support.