Volatility in the Trinity Fireball as Recorded in Glassy Fallout Debris

Chemical and isotopic analysis of glassy fallout debris from a nuclear explosion is a powerful forensic tool for identifying nuclear devices and preventing future nuclear attacks. It has, however, long been recognized that the composition of glassy fallout debris does not directly reflect the nature of the originating device. Physical, chemical, and nuclear processes within the fireball lead to the preferential segregation of elements and isotopes—i.e., fractionation—into different regions of the fallout, which hinders forensic reconstruction of the device. In order to retrieve as much information as possible about the device composition, it is necessary to identify the processes occurring in the fireball and account for the systematic fractionating effects of these processes.

Using minimally-destructive analysis to observe microscale variations

In order to study processes occurring within the fireball, we selected aerodynamic beads of glassy fallout (Fig. 1) from the first nuclear test, Trinity (July 16, 1945, New Mexico). Vesicular (cavity-rich), amorphous crusts of glassy “trinitite” are widely available for research by both federal and nonfederal entities; however, these ground-crust samples were avoided in the present study because it is still unknown to what extent they formed in place by melting of the ground surface versus in the fireball. The beads clearly formed within the fireball and solidified before falling out, and therefore unambiguously retain information about fireball processes.

Recent studies of trinitite have shown that the debris is extremely heterogeneous at the microscopic scale (Fig. 1). Therefore, analytical methods for studying trinitite composition that involve bulk dissolution necessarily result in averaging of the data derived from the debris. Because this microscale variation is important for understanding the formation of the debris, we have characterized the trinitite beads using several non-destructive or minimally-destructive techniques that allow for in situ determination of the chemical and isotopic composition of regions of the glass ~10–40 μm in diameter. Data presented here were acquired using a combination of scanning electron microscopy (SEM), energy-dispersive X-ray spectroscopy (EDS), digital autoradiography, electron microprobe analysis (EMPA), and secondary ion mass spectrometry (SIMS).

Our study focused on identifying fine-scale variations in both the major- and trace-element chemistry of the trinitite beads in order to better understand the trans-formation of material into glassy fallout. The major-element composition of debris (elements making up ≥ 1 wt.%) was determined by the ground material incorporated into the fireball. The trace-element composition (elements making up ≤ 0.1 wt.%, or ~1000 ppm) meanwhile was determined by a combination of trace elements within the ground material and dispersed device and detonation structure (tower) material.

Composition of fallout debris

Phase maps (Fig. 1) and EMPA data show that the trinitite beads studied are composed of three distinctly different types of glass: an essentially pure silica glass, a glass rich in alkali elements K and Na, and a glass characteristically containing Ca, Mg, and Fe. The silica and alkali glasses have compositions that are directly equivalent to mineral constituents in the test-site sand, as would be expected for glasses formed by melting individual sand grains. However, the compositional variations within the CaMgFe glass are large and complex.

In order to understand the chemical relationship of the CaMgFe glass to the test-site sand, we normalized the measured CaMgFe glass compositions to the estimated average bulk composition of the test-site sand (Fig. 2). Thermodynamic calculations, experiments, and studies of natural magmatic rocks show that melting of a quartzofeldspathic starting-material (like the Trinity test-site sand) will produce glasses enriched in elements such as Na, K, and Si, but depleted in elements like Ca, Mg, and Fe. The patterns of elemental enrichment and depletion in the CaMgFe glass are therefore not consistent with melting. Instead, the element variation patterns, which can range from highly enriched to quite depleted (Fig. 3), are more consistent with those expected for a melt formed by condensation from a vapor. We conclude that the elemental variation in CaMgFe glass arises from characteristic differences in condensation temperature (volatility) of different elements. In other words, the CaMgFe glass is chemically distinct because it formed by condensation from a vapor, not by melting of a solid.

Digital autoradiography is a technique that images radioactivity similar to the way photographic film images light; in this context, it can identify radioactive constituents that derive from the originating device. We found that these autora-diography maps of trinitite beads (Fig. 1) show no activity in unmelted mineral fragments, silica glass, or alkali glass. Debris components formed through simple melting of ground material are therefore unlikely to retain detectable chemical or isotopic information about a device. Detectable radioactivity occurs only in the condensation-formed CaMgFe glass, and thus we conclude that only liquids (and, subsequently, glasses) formed in this manner contain significant quantities of radioactive, device-derived constituents. Consequently, CaMgFe glass is the primary repository for compositional signals from the nuclear device.

Because ground-crust trinitite has been shown to have significant compositional signal from the device, it is probable that much of the ground crust is made up of CaMgFe glass that, like glasses of the aerodynamic beads, formed initially by condensation within the cloud. However, whereas the melt composing the beads formed high enough in the cloud to solidify before reaching the ground, the melt that constituted the ground-crust material must have formed lower in the cloud, splashing to the ground while still molten.

Volatility of glasses

Recent work has shown that as much as 50–60 vol.% of the glass within Trinity fallout formed as liquid condensed from the fireball plasma (i.e., the CaMgFe glass) as opposed to melting of local sediment. Thus, both the major-element and the radionuclide compositions of certain glasses are controlled by chemical volatility. Variations in the major-element compositions can be used to locate the different glass components in “volatility space”, as shown in Fig. 3. In this figure, we compare the relative proportions of low-, intermediate-, and high-volatility major elements from each location measured in the trinitite glass. Silica and alkali glass compositions show limited ranges of volatility; however, the condensation-formed CaMgFe glass shows a very large range in volatility, with compositions extending almost continuously from a very low volatility endmember, containing large quantities of Ca and Al, toward intermediate- and high-volatility compositions. Assuming that melts enriched in low-volatility elements condensed first, and that condensates became progressively enriched in intermediate- and high-volatility elements with time, the observed volatility trend in CaMgFe glass provides a timeline of condensation (Fig. 3).

The near-linearity of the CaMgFe glass major-element volatility trend allows for the formulation of a simple volatility index. By plotting trace element abundance as a function of the volatility index, we assessed the timing of the incorporation of trace elements into fallout glass. As might be expected, the abundance of high-volatility trace elements, such as Rb and Cs, increases with increasing volatility index. Conversely, low-volatility trace element abundance decreases with increasing volatility index. These patterns confirm the inference based on autoradiography (Fig. 1) that trace elements, including radioisotopes originating from the nuclear device, are incorporated into fallout glass by volatility-controlled co-condensation of vaporized ground and device material. 

Actinide fractionation

The fractionation of actinides and their isotopes during fallout formation can also be assessed using this volatility index, which shows the compositional changes in the fireball with time and decreasing temperature. The Trinity device contained both Pu and natural U, and minerals in the test-site sand contain abundant natural U. The ²³⁹Pu/²³⁸U ratios measured in trinitite beads decrease with increasing volatility index (Fig. 4a), indicating that Pu was less volatile than U within the fireball. While this relationship has been previously suggested, this is the first study to demonstrate it. This was achieved using an independent volatility scale based on stable element measurements. In contrast, the ²³⁵U/²³⁸U ratio remains the same over the entire volatility index range and, within analytical error, identical to the original, natural isotopic composition of both the device and sand uranium (Fig. 4b). Thus, while the elements Pu and U were fractionated from each other in the Trinity fireball, the isotopes of U were not. 

Uranium fractionation can be further understood by comparing data from the aerodynamic trinitite beads with ground-crust trinitite. The data shown in Fig. 2 suggest that, on average, U is more abundant in aerodynamic trinitite than in the original test-site sand or in the ground-crust trinitite. Our data demonstrate that aerodynamic trinitite was generally enriched in U at all times and over all volatility conditions, relative to ground-crust trinitite. We conclude that melts that condensed close enough to the ground surface to be rained out while still molten and form the ground crust contain less device-derived U than melts that condensed (and solidified) higher within the fireball to form the beads. Thus, U was likely less abundant in regions of the fireball closer to the ground.

Summary

Microanalysis of aerodynamic glassy fallout from the Trinity test suggests that some of the fallout glasses were formed by melting individual sand grains, but that much of the melt condensed from vaporized sediment (forming a glass characteristically containing Ca, Mg, and Fe). As would be expected, the order of condensation of both major and trace elements was determined by elemental volatility within the fireball. The major-element composition of a glass can be used to calculate a volatility index, which provides a volatility scale for trace elements, including those derived from the device. This volatility index can be used to show that Pu was less volatile than U in the fireball, that U isotopes were not fractionated during the explosion, and that U was more abundant higher in the fireball than near the ground surface.

Altogether, these findings demonstrate that microanalytical techniques can provide valuable information about how glassy fallout debris forms and how it records the chemical and isotopic composition of a nuclear device. This is a vital step in the forensic reconstruction of a nuclear explosion.

Acknowledgments

The author thanks Warren Oldham and Susan Hanson for trinitite bead samples; John Fournell and Richard Hervig for assistance with microanalysis data collection; Mindy Zimmer and Anthony Pollington for helpful discussions. This work was supported through the Glenn T. Seaborg Institute for Actinide Science, the Strategic Outcomes Office of LANL, and the US DOE National Nuclear Security Administration.

Further reading:

1. E. C. Freiling, “Radionuclide fractionation in bomb debris”, Science 1961, 133, 1991–1998.
2. G. N. Eby, N. Charnley, D. Pirrie, R. Hermes, J. Smoliga, and G. Rollinson, “Trinitite redux: Mineralogy and petrology”, Am. Mineral. 2015, 100, 427–441.
3. R. E. Hermes and W. B. Strickfaden, “A new look at trinitite”, Nuclear Weapons Journal 2005, 2, 2–7.
4. F. Belloni, J. Himbert, O. Marzocchi, and V. Romanello, “Investigating incorporation and distribution of radionuclides in trinitite” J. Environ. Radioact. 2011, 102, 852–862.
5. D. K. Smith and R. W. Williams, “The dynamic movement of plutonium in an underground nuclear test with implications for the contamination of groundwater”, J. Radioanal. Nucl. Chem. 2005, 263, 2, 281–285.
6. J. J. Bellucci, A. Simonetti, E. C. Koeman, C. Wallace, and P. C. Burns, “A detailed geochemical investigation of post-nuclear detonation trinitite glass at high spatial resolution: Delineating anthropogenic vs. natural components”, Chem. Geol. 2014, 365, 69–86.
7. C. E. Bonamici, R. L. Hervig, and W. S. Kinman, “Tracking radionuclide fractionation in the first atomic explosion using stable elements”, Anal. Chem. 2017, 89, 9877–9883