Gold clusters are gold compounds with a core of gold atoms and organic groups covalently bound to the surface gold atoms. An example is undecagold, Au11(P(C6H5)3)7, whose structure was solved by x-ray crystallography using 3-dimensional crystals. These differ from colloidal gold, which are suspensions of metal particles, usually formed by metal ion reduction; although the particles may be approximately the same size, they vary due to the statistical process of formation. Gold clusters are compounds with a definite formula, and should all be perfectly identical. However, it is known that there is a family of stable gold cluster compounds, such as Au6, Au11, Au13, Au55, Au67, etc. In a given preparation of gold clusters, there is usually some mixture of these, thus leading to some size variation. Methods such as gel filtration column chromatography and ultrafiltration can be used to separate most of these species, so that relatively pure preparations may be achieved. The UV-Vis spectra of the different clusters are also usually significantly different, aiding in identification of the various clusters. Since the larger nuclearity clusters (>/= Au55) have not been crystallized, their exact structure has not been determined.
We have found that clusters may be fairly successfully separated by gel filtration using a Pharmacia Superose 12 column. The first major peak observed has a green color, and is called Greengold. Mass spectrometry revealed that this cluster contains 75 gold atoms in its core.1 A second peak is brown in color, and the third major peak is yellow in color and is predominantly undecagold.
Upon microscopic investigation, the Greengold appears extremely regular in size.2 Upon standing it has some propensity to form small microcrystals. These are usually 2-dimensional planar sheets, as shown in Figs. 1 and 2. Occasionally, thin 3-dimensional crystals were also observed. The center-to-center spacing between gold clusters is 2.6 nm, consistent with a gold core ~1.4 nm in diameter and an organic ligand shell of 0.6 nm thickness. Sheets over holes do not charge up and have an integrated mass thickness equivalent to a 40 Angstrom carbon film.
Why does this cluster crystallize, and why in 2-D sheets? Generally crystallization occurs when units are highly purified, so that each may fit into the growing lattice properly. This particular gold cluster was synthesized with two organic ligands attached to its surface, one hydrophobic, and one hydrophilic. It would appear logical that the two ligands are phase separated, and the hydrophobic ligand may form a belt around the cluster. This would interact with other clusters to extend this domain to minimize the free energy, especially considering that the clusters are in aqueous solution. Two dimensional arrays would result from this type of structure.
The formation of gold cluster crystals may yield novel materials with interesting properties. For example, the spacing between the clusters is appropriate for electron tunneling conduction, and gold clusters exhibit non-linear optical qualities, which could be applied this way using molecular films. The 2-D films might also be used as exacting substrates over holes for quantitative microscopy, instead of amorphous carbon, where they could be completely subtracted from the image. Crystals are also a first step to solving the structures by x-ray diffraction.
Fig. 1. Darkfield STEM micrograph of single-layer gold cluster crystal self-supported over crack in 2 nm carbon film. Carbon film is seen in lower right and upper left of image, where it supports single gold clusters. Full image width, 128 nm.
Fig. 2. Darkfield STEM micrograph of single-layer gold cluster crystal. Full image width, 64 nm.
- E. Gutierrez et al., Eur. Phys. J., 9 (1999) 1.
- J. S. Wall, J. Struct. Biol., 127 (1999) 161.
- The authors wish to thank Dr. Edmund Gutierrez for assisting with gold preparation. We would also like to thank Dr. Martha Simon, Ms. Beth Lin, and Mr. Frank Kito for STEM operation. Research supported by the Office of Biological and Environmental Research of the U.S. Department of Energy under Prime Contract No. DE-AC02-98CH10886 with Brookhaven National Laboratory, and by National Institutes of Health Grant 2 P41 RR01777 and SBIR grants.
Thanks to Springer-Verlag for allowing us to reproduce this online.