Computational Screening of 670,000 Materials For Optimal Separation of Krypton From Xenon.

[NNadir]

As I poke around this long weekend through scientific papers that I was inspired to collect but never actually read and filed properly I came across a really cool one, this one: What Are the Best Materials To Separate a Xenon/Krypton Mixture?

(Cory M. Simon†, Rocio Mercado‡, Sondre K. Schnell†§, Berend Smit†⊥, and Maciej Haranczyk*¶ Chem. Mater., 2015, 27 (12), pp 4459–4475)

Xenon and Krypton are very rare gases in the Earth's atmosphere, from which they are obtained industrially, because they are very useful, at considerable expense. (The most common use for xenon is in automotive headlights because its excited states decay to produce light that is very similar to daylight.)

The authors of the cited paper do quite a nice job in describing some of the background on these two gases, and I'll quote what they say in their introduction:

The noble gases xenon (Xe) and krypton (Kr) have several important applications.28 Xenon is used as an anesthetic 29−31 and for imaging 32 in the health industry and as a satellite propellant in the space industry.33 Both xenon and krypton are used in lighting,34 in lasers,35,36 in double glazing for insulation,37,38 and as carrier gases in analytical chemistry.39 Since krypton and xenon are present in Earth’s atmosphere at concentrations of 1.14 and 0.087 ppm, respectively,40 the conventional method to obtain xenon and krypton is as a byproduct of the separation of air into oxygen and nitrogen by cryogenic distillation.41 This byproduct stream from air separation consists of 80% krypton and 20% xenon.42 At Air Liquide, this mixture is compressed to 200 bar and stored in cylinders, then sent to a separate Xe−Kr separation plant to undergo another cryogenic distillation to obtain pure xenon and pure krypton.43 Cryogenic distillation for the separation of krypton and xenon has a very high energy and capital requirement, reflected by the cost of high-purity xenon, about $5000/kg.44

Both gases are considered in general general to be inert, and until 1962 they were thought not to exhibit any kind of chemistry at all. Since that time, it has been discovered that both gases can, in fact, react to form compounds, krypton only at very low temperatures. The chemistry of xenon, by contrast, is quite extensive. (I touched briefly on the discovery of xenon chemistry here: Neil Bartlett's superpowerful oxidants NiF6- and AgF4- and the preparation of RhF6.)

As noted by the authors, the separation of these two gases is energetically and economically expensive, and the purpose of their paper is to examine, by computer modeling, approaches to designing materials that can reduce the costs of their separation by a technique known as "pressure swing absorption," PSA which relies on a process in which a gas consisting of multiple components is pressured in a chamber in the presence of a solid material that has a capability to absorb one component in preference to another. (Home oxygen generators for medical use, and nitrogen generators in some scientific laboratories utilize a PSA approach to separate these bulk gases.)

It is possible to exploit, I suppose, the chemical differences between the gases to effect their separation, however in most cases the chemistry involves the use of highly reactive, corrosive and toxic fluorine gas. In the presence of water, xenon fluorides can hydrolyze to form xenon oxides which can be highly explosive.

By contrast, pressure swing absorption is orders of magnitude safer and most probably considerably cheaper.

Modern computers are extremely powerful compared with computers from even a short time ago, but computational power is not necessarily cheap or free when one considers very, very, very complex calculations.

Nevertheless in silico calculations can save far greater expense in screening for molecular structures - be they involved in medicinal chemistry or in materials chemistry, such as being explored here - that accomplish these kinds of tasks.

Candidate materials for the separation of these gases exist in quite an array of differing types, again, I'll let the authors describe them:

By combining different molecular building blocks in their synthesis, advanced classes of nanoporous materials are highly tunable. For example, in metal organic frameworks (MOFs),6metal nodes or clusters form a coordination network with organic linkers. Other highly adjustable materials include covalent organic frameworks (COFs),7 zeolitic imidizolate frameworks (ZIFs),8 and porous polymer networks (PPNs).9High chemical tunability not only enables one to tailor-make a material for each application under a variety of conditions, but also inundates researchers with practically endless possibilities. Because of limited resources and time in practice, only a small subset of the possible materials can be synthesized and tested for each application.

The separations, as the author's note, in a pressure swing situation rely mostly on the size difference between the two types of atoms: The mean diameter of a xenon atom is 198.5 picometers, of krypton 183 picometers, a small, but significant difference. The idea is to structure the pores in materials so that xenon cannot fit into the pores while krypton can, or conversely that krypton can easily diffuse out of the pores while xenon can do so only slowly. Besides size, differences in their electronic structure - which accounts for the differences in their chemistry - can also be exploited without actual chemical reactions taking place.

Their computational approach, which they describe as considerably streamlined in terms of the computational algorithms utilized previously for one class of possible absorbents, metal organic frameworks (MOF), the previous approach being described as "brute force" is described in the following text:

The workflow of our method to screen about 670,000 structures in the Nanoporous Materials Genome is illustrated in Figure 2 and consists of the following six steps. (1) Characterization: we compute the feature vector of each material, whose components are seven quickly computed structural descriptors. (2) Selection of the training set: we use a diversity selection algorithm76 to ensure that our training set of materials adequately covers our seven-dimensional feature space. (3)Label the training set: we perform binary grand-canonical Monte Carlo simulations to compute the Xe/Kr uptake in the training set. (4)Training of the forest: we use these training examples to grow a forest of decision tree regressors to predict Xe/Kr separation performance from the structural descriptors. (5) Prescreening: we run the structural descriptors of the remaining materials through the trained forest of decision tree regressors. (6) Materials discovery: if the forest predicts the material to be promising for Xe/Kr separations, we run grand canonical Monte Carlo simulations to refine the prediction.

The authors in this screening process describe two known materials which may be useful in these separations:

Many materials in our database are predicted to have better Xe/Kr separation performance than CC3,104 a leading material for Xe/Kr separations.44 Our models predict that the two most selective materials in the Nanoporous Materials Genome are JAVTAC, an aluminophosphate zeolite analogue,106 and KAXQIL, a calcium coordination network.105 Both materials have been synthesized but not yet tested for Xe/Kr separations. We hope that our open database of simulated Xe uptake and Xe/Kr selectivities (http://nanoporousmaterials.org/xekrseparations/) will inspire the synthesis and characterization of a new material for Xe/Kr separations.

This paper has been cited extensively since its publication two years ago, and it might be fun, if I find the time once my favorite academic libraries reopen after the holidays, to look into these citations to see if these predictions have been experimentally confirmed.

It is interesting to note that isotopic ratios of these gases tell us a lot about the history of this planet, owing to the fact that actinide elements, for example, long lived plutonium-244 - which is known to have been a constituent of the early Earth (because of xenon isotopes) - spontaneously fission in a characteristic way that causes a traceable signature of geological history.

(See for example: Xenon isotope constraints on the thermal evolution of the early Earth (Nicolas Coltice a,⁎, Bernard Marty b, Reika Yokochi c, Chemical Geology 266 (2009) 4–9))

As the authors of the original paper note, these separations would also be valuable in the processing of used nuclear fuels, because both elements are fission products. No radioactive isotopes of xenon are very long lived in nuclear fuels; they rapidly decay into other isotopes. Xenon-135 has the highest neutron capture cross section of any nuclide known, it is rapidly converted into non-radioactive xenon-136 in the neutron flux in the core of nuclear reactors. Because of this, it does not accumulate, and in any case its half life is on the order of hours, not days. (The presence of xenon-135 in reactor cores where it is the cause of an effect known as "xenon poisoning" played a role in the very stupid decisions made by the operators of the Chernobyl reactor that exploded: Their decisions, made late at night, to remove the control rods from the reactor was intended to overcome xenon poisoning effects.)

Fission gases contained in fuel rods are therefore highly enriched in these valuable gases, and in theory they could be collected from used nuclear fuel for use, especially xenon.

Krypton contains one relatively long lived radioactive isotope, krypton-85, and its detection has been utilized to identify nuclear explosions (both atmospheric and underground) as well as reprocessing of nuclear fuels around the world, since it is generally vented to the atmosphere rather than recovered for use. (Radiokrypton-85 I think is possibly a very useful material for providing a continuous portable light source for remote locations or as a continuous power source.) Venting it to the atmosphere is probably not a very dangerous practice however, certainly not at the level of venting dangerous fossil fuel waste to the atmosphere, since dangerous fossil fuel waste and dangerous biomass combustion waste combine to cause 7 million deaths every year.

Krypton 85 can also be stored to provide a source of non-radioactive rubidium-85 which would be less radioactive than natural rubidium, since the latter contains the naturally occurring long lived radioactive isotope rubidium-87. However, except for esoteric research purposes, I cannot imagine that there is much call for isotopically pure rubidium-85, but hey, you never know.

Some folks might find all of this stuff as interesting as I do.

Enjoy the rest of the labor day weekend.