Skip to main content

Interrogating a fullerene-imprisoned water molecule

A H2O molecule (water) trapped inside a molecule of buckminsterfullerene, or a molecule of 60 carbon atoms. (Image: Malcolm Levitt)

Using an impressive bit of molecular surgery, an international group of scientists has entrapped individual H2O molecules within carbon cages and then battered them with neutrons, magnetic fields, and infrared light to make them give up their secrets. We interviewed team member Ronald Lawler of Brown University, also a AAAS member, in order to better understand what was learned from these experiments.

The carbon cages were composed of the molecule buckminsterfullerene, aka C60, and sometimes called a "buckyball." The molecule is composed of 60 symmetrical carbon atoms arranged in a pattern similar to the vertices of a geodesic dome, such as those designed by inventor and scientist Buckminster Fuller.   

By trapping a single water molecule in each cage, the scientists were able to study its properties free of intermolecular hydrogen bonds or solvent effects. Like molecular hydrogen, water can exist in two forms, called "ortho" and "para," due to the quantum property of spin. The hydrogen protons can spin in a parallel (ortho) or antiparallel fashion (para). So it was of interest to see whether conversion of the ortho state to the lower energy para state could be observed in isolated molecules.

AAAS Members Central: How much space is there inside a buckyball? Can you get anything much larger than a water molecule in there?"
Ronald Lawler, Brown University:
The carbon atoms in C60 form a sphere with a diameter of about 7.2 angstroms. That's known from X-ray diffraction. The actual space available for a caged molecule, however, is decreased by the van der Waals (vdW) radii both of the interior of the carbon-atom shell and the small molecule itself. The carbon shell contribution decreases the diameter of the shell by about 3 angstroms to an effective diameter of a little more than 4 angstroms. H2O and H2 molecules have surprisingly similar diameters, about 3 angstroms. So the answer to your question is that both water and a hydrogen molecule will fit with an angstrom or so of wiggle room. That's also borne out by some fairly high level quantum calculations.

Methane, ammonia, and carbon monoxide have all been encapsulated in somewhat larger fullerenes, with openings in the shell, but have not yet been successfully closed up in C60. The vdW radii of those molecules are borderline too large. Atoms as large as Xe, that have a vdW radius larger than H2O have, however, been incorporated in low yields using a brute-force method different from that used to prepare H2O@C60. So there still is hope for larger molecules.

AAASMC: Do the electric dipoles of the water molecules rotate freely? How can you tell that from the data?
The term "rotate freely" as used in the PNAS paper would describe a molecule whose rotational energy levels, as probed by far infrared (FIR) and inelastic neutron scattering (INS) spectroscopy, are essentially the same as the isolated molecule, i.e. in the gas phase. This is almost true for both H2O and H2 in C60. The small differences are detectable in the spectra at low temperature as shifts and splitting in the frequencies of the rotational transitions that correspond to weak interactions between rotational and translational motions of caged molecules. This has been calculated with some accuracy for H2@C60 by Zlatko Bacic and his group at NYU, but the computations are more difficult for H2O@C60 and haven't yet been completed. But the group is working on it.

We have also done some NMR relaxation time measurements on both H2@C60 and H2O@C60 in liquid solutions near room temperature that give an approximate value for an average rotation time of the small molecule as it changes randomly by collisions with the cage. The characteristic time for reorientation under those conditions is longer than that for the small molecule dissolved in a solvent but shorter than the rotation time for the larger cage. One could interpret that as indicating that the small molecule "feels" the cage wall at the higher temperature, but not as much as it would feel a solvent molecule if it were dissolved without the surrounding fullerene cage.

AAASMC:  Do ortho and para forms interconvert? I gather from the FIR data that ortho converts to para where it is trapped at low temperature. Can para convert to ortho? What physically causes the spin to flip in orientation?"
The FIR data show clearly that the intensities of the transitions assigned to the ortho and para forms change with time [which] is just the way one would expect if the more stable, para, form were being enriched slowly at a lower temperature. This is analogous to a simple chemical rate process, so it proceeds in both directions, eventually reaching an equilibrium composition of the two isomers appropriate to the temperature. Equilibration of the spin isomers of H2O and of H2 turns out to be of great interest to planetary astronomers because the ratio of the two forms, if at equilibrium, can be used as a spectroscopic thermometer for comets (H2O) or the gas giants like Jupiter (H2).

The simple answer to what causes the spin flip in H2O@C60 is "Nobody knows! (yet)." The most common mechanism for ortho-para conversion, that was worked out by Eugene Wigner in 1933 (!), involves having the two nuclear spins interact differently with an internal magnetic field, usually generated by a paramagnetic material like oxygen. We don't know for sure whether that's happening in the solid H2O@C60 samples, although some care is taken to exclude O2. There are other mechanisms for conversion known for polyatomic molecules like H2O and they may be operating in this case. There has also been a fair amount of work done on conversion of H2O spin isomers in rare gas matrices at temperature similar to those in the PNAS report.

AAASMC:  Ortho water is metastable by definition, isn't it, as para is lower energy? The experiment shows that ortho converts to para—so you can trap it originally in ortho, but it doesn't have to stay that way, correct?
 What you say is mostly correct, and is a corollary of the question above. Most often, however, the "metastability" corresponds to an excess of the para form if the temperature of the sample is raised before equilibration occurs. In fact, we are hoping to be able to find a way to raise the temperature of a sample of H2O@C60 equilibrated at cryogenic temperatures rapidly enough, and keep in the metastable state long enough, to make use of the enrichment of the single nuclear spin state of the para form. We could, in principle, then transfer that spin state to other molecules and greatly increase the sensitivity for detection by NMR or in MRI measurements.

Actually obtaining a sample enriched in ortho-water is, of course, much more difficult than enriching in more stable para-water. It involves finding a property, like the boiling point or the frequency of a spectral transition, which is different for the two spin isomers. This works well for ortho- and para-H2 that can even be separated by gas chromatography. There is some controversy about whether this has yet been accomplished for the water spin isomers and the possibility has not yet been explored for H2O@C60.

There is an important practical effect of metastable H2 as the temperature is lowered. It turns out that as hydrogen gas at ordinary temperature is condensed to form liquid hydrogen at low temperature (think "Space Shuttle fuel"), the sample may become unstable if the high concentration of the ortho isomer present at high T persists at low T. Consequently large scale hydrogen liquifiers employ a magnetic catalyst to ensure that the sample maintains the composition appropriate for low T.

AAASMC: What is the potential describing the interaction of water with the curved carbon surface?  Is there a simple English way of explaining this that would make sense to me and my nonphysicist readers?
This is just a fancy way of wondering what force a water molecule would feel as it approaches a carbon surface. The "curved" part is a somewhat new requirement since most previous work was done on graphite that forms a planar surface.

We presently have a pretty good qualitative picture of the distance at which the water molecule starts to feel the carbon (see "van der Waals radius" above). It also seems likely that the force does not simply increase suddenly like a rock hitting a brick wall. Instead a water molecule hitting a fullerene cage wall probably is a lot like throwing a pillow onto a sofa. An accurate description of the magnitude of the force, and how it varies with distance, is needed if one is to calculate the rotation and translation energies of the caged small molecule. This has been done with some success for H2@C60 (Zlatko Bacic) but has not yet been looked at in detail for H2O@C60.

 AAASMC: There is the suggestion here of a ferroelectric effect, or a cooperation between molecules in adjoining cages. How far away can a water molecule be from another water molecule and still feel its influence? What exactly is it feeling?"
The possibility of an electrical interaction between endofullerene molecules is what especially sets H2O@C60 apart from H2@C60. The electric dipole interaction would fall off as 1/r3. About the closest H2O molecules in adjacent C60 cages could get would be about the sum of the effective radii of the cages. For C60 this turns out to be about 10 angstroms. This is quite a bit larger than water molecules in ice, for example, and, of course in the fullerene case there is no possibility of hydrogen bonding. And then there is the necessity that the electric field of the H2O dipole reach outside the walls of the carbon cage. There is the possibility that the fullerene will act as a Faraday cage and block most of the electrical interaction. So far, we know qualitatively that solvent polarity doesn't have a detectable effect on the proton relaxation time of H2O@C60 as might be expected if a polar solvent were able to exert some drag on the rotational motion of the cage H2O and slow down its rotational motion.

 I'm more of a magnetic field guy and haven't really thought much about this particular part of the H2O@C60 story. Malcolm Levitt, and at least one of his colleagues, and also Yas Murata in Japan are very much interested and working on relevant theory and experiments.

AAASMC:  Is there any contribution of the C60 cage to any of the parameters measured?  I understand that there is no hydrogen, so no NMR signature, but shouldn't there be a neutron scattering peak?
  Subtle question. For FIR, we are lucky because the lowest frequency vibration of the carbon cage is way outside the range needed to study H2O rotations. In the case of INS there is weak, broad background scattering from the carbon atoms but this is easily filtered out of the data by comparing with the empty C60 signal.

For NMR there is actually a 13C (an isotope of carbon) signal that occurs at a frequency about one-fourth that of the proton and therefore doesn't interfere. It's actually of some interest because the 13C signal should be able to report some information about both the position and time dependence of the proton magnetic dipoles of the H2O molecule inside. This hasn't been looked at carefully yet. 13C NMR is actually easier than usual for C60 because statistically there is a 60-fold increase in the probability per molecule that it will contain 13C, relative to the more common situation where only one or a few carbons are symmetrically equivalent  (normally 12C outnumbers 13C by a 99:1 ratio).

AAASMC: Several years ago, Harry Dorn at Virginia Tech was experimenting with "trimetaspheres," which are rare earth metals trapped in a C80 molecule.  The thinking at the time was that these molecules might have some uses for medical imaging, or possibly in electronics.  Have you heard anything further about this?
I was not aware specifically of Dorn's work, but I do know that there is a lot of interest in the larger fullerenes that can incorporate metal atom clusters. Nitrogen atoms trapped in C60 are also extremely stable and are being considered as possible systems for quantum computing.

Applications of the rare earths would probably rely on magnetic interactions between the spins in a lattice of the fullerenes, a lot like the proposed ferroelectric interactions between H2O@C60 molecules. The imaging aspects probably would come through their use as contrast agents. These are mostly complexes of gadolinium. The fullerenes might be more sensitive and more versatile, especially if the cages can be functionalized.

AAASMC: What other molecule would you like to put in a fullerene cage and why?
We all have our own dreams. At the moment I am intrigued by fullerenes containing more than one small molecule. It has been possible, for example, to put two H2 molecules into a C70 cage. (Dorn's trimetal compounds are, I think, more like putting in a single molecule since the atoms interact strongly). Two H2O molecules would be even better and I'm sure it's being attempted.

In a sense what putting two molecules in a fullerene does is get you a "bimolecular collision in a bottle." The molecular beam people have been studying things like this indirectly for quite awhile, and clusters can be studied optically in the gas phase using lasers. The fullerenes, however, make it possible to, in effect, study gas phase properties using liquid or solid phase techniques like those used in the PNAS paper.

It would also be great to put formaldehyde, CH2O, into a fullerene because one then has a chromophore that reaches into the near UV. H2O and H2 are transparent there. This also opens up the possibility of photochemistry occurring from a reactive intermediate within the cage. Nobody knows how this would work. And there is also the possibility that photoexcited formaldehyde could add to the inside of the cage to form a four-membered ring that extends into the interior. Lots of possibilities....

Related Links:


Steven A. Edwards, Ph.D.

Related Scientific Disciplines