Forensic study of early stages of the Chernobyl accident: story of three hot
Andrey A. Shiryaev1,2,*, Irina E. Vlasova2, Vasily O. Yapaskurt3, Boris E. Burakov4, Alexey A.
Averin1, Ivan Elantyev5
1 A.N. Frumkin Institute of Physical Chemistry and Electrochemistry RAS, Leninsky pr. 31 korp.
4, Moscow, 119071, Russia
2 Department of Chemistry, Lomonosov Moscow State University, Leninskie gory, 1 bld.3,
Moscow, 119991, Russia
3 Department of Geology, Lomonosov Moscow State University, Moscow, 119991, Russia
4 V.G. Khlopin Radium Institute, 28, 2-nd Murinskiy Ave., St. Petersburg, 194021, Russia
5 Laboratory for microparticles analysis, Bolshaya Cheremushkinskaya str., 25, Moscow,
* - Corresponding author. email@example.com and firstname.lastname@example.org
Results of the investigation of three contrasting hot particles from the first fraction of
ejecta of the Chernobyl accident are presented.
Application of complementary methods provides new information about structure of the
Chernobyl hot particles.
Some of the particles consist of complex mixture of Zr-U-O phases often with minor Fe
admixture from spacer grids, reflecting processes preceding explosion of the reactor.
Three contrasting hot particles ejected from the core of the 4th Unit of Chernobyl nuclear
power plant at an early stage of Chernobyl accident were studied using complementary analytical
methods: including γ-spectrometry, SEM-E(W)DX, EBSD, Raman spectroscopy and Secondary
Ions Mass Spectrometry. The particles span range from dispersed UO2+x fuel to a fragment of
zirconia with traces of U and to chemically and structurally complex Zr-U-O particle. These
particles represent wide variety of processes in the reactor during the accident development and
likely originate in spatially distinct domains. Whereas the fuel particle is virtually unaltered, the
zirconia particle records interaction of zircaloy with fuel and structural steel, albeit rather short
one. Finally, the last particle comprises remarkable mix of various phases and was most likely
formed at an advanced stage of the accident when significant interaction of the fuel with
surrounding materials and eventual displacement of the reaction products took place. This
particle alone reproduces rather wide range of interactions encountered during various in-pile
experiments. Ubiquitous presence of Fe in the Zr-U-O phases reveals interaction of spacer grids
with Zr-U-O melt. Conditions leading to formation of these particles at early stages of the
accident are discussed.
Chernobyl accident on April 26, 1986, remains the most severe nuclear accident in
history. Despite numerous studies, knowledge of several important details of the accident is still
insufficiently clear. In part, these uncertainties are explained by scarcity of high-resolution
studies of hot particles and debris from the accident, which, in turn is due to both high
radioactivity of certain samples and related complications as well as due to limited access to
many analytical techniques in the first decade after the accident when most works were
performed. Nevertheless, we believe that investigation of available material is not only driven by
a scientific curiosity, but is important for validation of in- and out-of-pile experiments of nuclear
fuel behavior and for understanding of other accidents such as Fukushima Daichi and
development of prevention and remediation strategies.
Several detailed studies of Chernobyl hot particles (HPs) were published since 1986 (e.g.,
reviews [1,2]). Most of these contribution address particles of dispersed fuel [e.g. 3-8],
transuranium elements inventory in them [9-12], fission products needed for evaluation of the
fuel burn-up degree (e.g., [13, 14]), and temperatures during the explosion [15, 16]. However,
from the beginning it was clear that besides dispersed fuel, more complex particles, especially
with Zr admixture are present [1, 17-21]. Existence of different types of particles is partly
explained by the fact that extensive ejection from the destroyed reactor continued for 10 days
after the initiation of the accident and involved both explosion-related materials and those from
subsequent fire. Particles comprising simultaneously U and Zr and ejected during the actual
explosion are of significant interest, since details of their structure shed light onto processes at
the earliest stages of the accident.
Here we present results of detailed examination of three contrasting hot particles from the
Chernobyl accident by several complementary analytical methods. These data are used to
constrain processes during very first seconds of the accident. Note that in this paper radioactive
aerosols are not considered, for comprehensive review see .
1. Samples and methods
Three particles were collected in 1986 and 1990 at 0.5 and 5.5 km to the West from the
Chernobyl NPP. The Western Plume of the Chernobyl carried the very first fraction of the ejecta,
directly related to the explosion itself (for comprehensive review see ). The particles were
individually mounted into acrylic resin and manually polished. The sizes of the particles as seen
in the polished sections are: S2 – 140×110, S4 - 250×150, RL-1 - 80×50 μm (Fig. 1A, 2 A,B, 3
A,B). The particle S2 is a piece of dispersed UO2 fuel, the RL-1 is a piece of zirconia with small
U admixture and S4 is an amazingly complex particle containing U, Zr and other elements.
γ-spectra of individual particles were acquired using X-ray/γ spectrometer DSPec
(ORTEC, USA) with HPGe detector GEM-C.
Quantitative analyses of manually polished and carbon-coated (thickness ~30 nm)
samples were performed using Energy- (EDX) and Wavelength-dispersive (WDS) methods. For
the EDX study JEOL JSM-6480LV Scanning Electron Microscope equipped with W thermal
emission cathode and Oxford X-Maxn 50 detector (active area 50 mm2) was employed. XPP
correction was applied using Oxford Instruments INCA software. Accelerating voltage was
20 kV; electron beam current – 10 nA, working distance – 10 mm. Count rate was 14-17×103 cps
with 20-30% deadtime. At acquisition lifetime of 100 sec detection limits for most elements of
interest were 0.04-0.07 mass% (the given values are for Ca and Fe as measured using K-lines).
For optimisation of emission lines’ profiles and standardization reference metals and simple
oxides were used: for Ca (Kα) – wollastonite, for Cr(Ka), Fe(Ka), Zr(La) и Nb(La) –
corresponding metals, for U(Ma) – UO2 (dense ceramics). The references and samples were
measured at identical operation conditions. Oxygen was calculated from stochiometry.
To address ZrO2 structure Electron Backscattering Diffraction attachment was used with
the EDX detector. Diffraction patterns were analysed using AZtec («Oxford Instruments»)
software with American Mineralogist database .
Energy-dispersive method allowed investigation of elements with concentrations
exceeding 0.05 – 0.1 mass%. However, low spectral resolution hampers quantitative
determination of elements with overlapping lines, for example, L-lines of Zr and Nb. For Zr-rich
phases with minor Nb admixture the concentration of the latter is 2-3 times overestimated even
when advanced deconvolution of emission lines profiles is used. Therefore, some regions were
investigated also by Wavelength-dispersive spectroscopy using JXA-8230 (Jeol Ltd., Japan)
microprobe. To optimize excitation of Zr and Nb L-lines the acceleration voltage was set to
15 kV. For all measurements focused beam with 40 nA current was employed. Ca (Ka1), Zr
(La1), Nb (Lb1), U (Mb1) were measured using PET crystals and same references as for EDX; for
Fe (La1) TAP crystal and crystalline Fe2O3 were used. Dwell time at the line peak was 60 sec,
the background acquisition on every side of the peak was 30 sec. Oxygen concentration was
calculated from stochiometry assuming tetravalent U and Zr, Nb5+, Fe2+ and Ca2+. ZAF-factors
were accounted using XPP-correction.
Raman spectra in quasi-backscattering geometry were acquired with Senterra (Bruker)
spectrometer with an Olympus BX-51 microscope with long working distance objectives.
Excitation wavelengths of 532 and 785 nm were employed. The laser power was kept
sufficiently low to prevent sample modification; the spot size was 2-5 µm depending on
Secondary ion mass spectrometer Cameca IMS-1280 was used for isotopic analysis of
uranium in RL-1 microparticle. The instrument was operated in multicollector mode providing
simultaneous registration of ion currents of 234U, 235U, 236U, 238U and hydride ions 238U1H. The
mass resolution was 2500, primary beam current was ~5 nA. An electron gun was used to
compensate sample charging during analysis.
3.1. Fuel particle S2
The particle S2 represents rather typical piece of a dispersed fuel (Fig. 1). Intergrain
boundaries and cavities from gas bubbles are observed, the grains are approx. 30 μm, although
this value could be a slight overestimation, since the particle was not etched prior to the analysis
and some boundaries could remain hidden. In fresh RBMK fuel grains are usually less than
20 μm (e.g., ), therefore, the S2 particle probably corresponds to an intermediate part of a
fuel pellet (midway from the center to periphery) subjected to moderate annealing during normal
operation of the reactor. This conclusion is supported by relatively large size of intra-granular
and inter-grain pores, being 1-1.5 and up to >10 μm respectively.
An interesting feature of some of the intergranular pores is channel-like morphology.
Elongated pores are common for fuel with the burn-up of 50-60 GWd/tU as the high burn-up rim
structure is commencing and precipitation of 5-metal particles (e.g., ). In the same time,
average ChNPP fuel burn-up was much lower with an average value of 13 GWd/tU only (see
 for review of fuel in Chernobyl NPP). Existence of the 5-metal particles in ChNPP fuel was
inferred from ejected fuel particles , but this issue remains debatable . However, the
ramping of medium burn-up fuel in the Riso project has shown that ramping of the fuels can
considerably accelerate this process as is seen clearly in Figs. 9 & 10 in . Also, the rapid
ramping to fuel melting in the Phébus-FP project has shown that temperature ramps in severe
accidents can also generate the same phenomena in the non-melted parts of the fuel . Thus, it
is considered that the advanced development and coalescence of the grain boundary gas bubbles
for such a low burn-up as in the studied hot particle is the result of the temperature ramp of the
fuel at the time of the accident. It appears to have suffered an intermediate ramp rate compared
to ramps close to the initiation point in the reactor core. It therefore suffered a shortened or
slower transient before breaking up and perhaps cooling in contact with the steam or
γ-spectra of the particle S2 were obtained twice, in 1999 and 2016, allowing analysis of
different nuclides with low background. Short-living nuclides (144Ce, 134Cs, 106Ru, 125Sb,
154,155Eu) were measured in 1999; by 2016 all of them have decayed whereas 241Am accumulates
from the parent 241Pu (table S-1 in Supplementary Materials). Thirty years after the accident only
137Cs, 241Am, 60Co and Eu isotopes could be detected. In 2016 the 137Cs activity was 1025
Bq/particle, 241Am – 0.11 Bq/particle. High 137Cs concentration could be caused by relatively
low local temperature of the fuel fragment and/or short duration of overheating during the
Raman spectra of the particle show clear features of damaged stochiometric urania with
peaks at 446 cm-1 (T2g mode of UO2), 562 cm-1 (presumably damage-related peak), and 1145-
1150 cm-1 due to crystal electric field transition . The shoulder at 630 cm-1 increases with the
growth of O concentration in the lattice  and may be a sign of U4O9 . At present no
conclusions on timing of the oxidation – during normal operation/explosion/ejection – could be
3.2. Zirconia particle RL-1
This particle represents a rare case among the Chernobyl HPs where the matrix is almost
pure zirconia. In 2016 the activity of 137Cs was 1.1 Bq/particle, 241Am – 0.28 Bq/particle. Its
chemical composition is close to ZrO2 in the main body of the particle with traces of U at the
level few hundredth of a wt%. Raman scattering confirms that the particle consists of monoclinic
ZrO2 without admixture of tetragonal or cubic modifications (fig. 2C; for assignment of the
Raman lines see [32,33]). Uranium is observed in some spots, mostly at the periphery of the
particle (see whitish regions on the BSE-SEM images in fig. 2 E-G) where urania content reach
13 wt% and probably even higher in small domains. In the central part of the particle U
concentration is 0-0.2 wt%. According to SIMS results, the 235U enrichment is 0.968±0.005 at%,
236U - 0.18±0.002 at%, 234U - 0.0165±0.008 at%. Such ratio of uranium isotopes is within the
range of the ChNPP fuel with different degrees of burn-up at the moment of accident, as was
shown for HPs from the nearest fallout zone, for example, in [8; 34]. Note that the 238U/235U ratio
of 102.1 falls between the values reported for zircons from Chernobyl lava-like materials, where
mixing of uranium from different fuel rods took place .
Niobium concentration (WDX data recalculated for Nb2O5) is rather constant in the range
0.42-0.49 wt%; only at few points its concentration is higher and reach 1.3 wt% (Supplementary
materials Table S-2). RBMK fuel cladding (the E110 alloy, below called “zircaloy” for
simplicity) consists of metallic Zr with admixture of 1 wt.% Nb. All these enriched spots also
show high U levels. At some places iron (up to 1.5%) and occasionally Ca are observed. Fe
concentration appears to positively correlate with urania. Detailed compositional data are given
in Supplementary materials.
In addition to the large voids there are numerous submicron voids (Fig. 2 apart from Fig.
2C). Most of them appear to mark grain boundaries. Some large (0.5-0.7 µm) voids are clearly
facetted (Fig. 2D). We do not have a unique explanation, but they could either be voids formed
during slow straining of hot zircaloy  or inherited irradiation-induced cavities (e.g., ).
Shapes of the U-enriched regions are variable. Sometimes it is irregular with occasional
submicron spots, presumably representing U-rich (U1-xZrx)O2 particles. In other cases the U-rich
domain is clearly limited by ZrO2 grain boundaries. Such distribution of U likely indicates that
these domains were formed by intragrain uranium diffusion from “point sources” at high
temperatures with the grain and interphase boundaries serving as an effective obstacle. One
presumes small UO2 pieces became stuck to the metal and started to form eutectic melt. This
process was rapidly quenched due to the reactor explosion.
A convincing evidence of rapid quench of a (partly) molten system comes from
observation of a symplectite structure typical for decay of solid solutions region (Fig. 2H). This
region is relatively enriched in U (6.6 wt% UO2).
One of the unusual features of the particle is that despite being ceramic-like, the grain
sizes inferred from voids tracing the boundaries are close to those in the initial E110 alloy. No
columnar grains typical for thick ZrO2 layers formed during prolonged oxidation of zircaloy
tubes are observed. The rugged particle periphery suggests that it was torn away from the matrix
along the grain boundaries without notable plastic deformation, i.e. in a brittle regime. Taking
into account small size of the particle (<150 microns), it is possible that the oxidation process
largely took place after the ejection.
3.3. Multiphase particle S4
This particle shows remarkable complexity and contains phases ranging from UO2 to
ZrO2. In 2016 activity of 137Cs was 19 Bq/particle, that of 241Am – 5.1 Bq/particle. According to
microprobe analysis (both Wavelength- and Energy dispersive spectroscopy) and Electron back-
scattering diffraction (EBSD) the particle largely consists of solid solution (Zr,U)O2 interspersed
with abundant microscopic UO2. EBSD patterns suggested tetragonal structure of the (Zr,U)O2,
although presence of monoclinic material could not be excluded. However, Raman
microspectroscopy (fig 3C) proved the tetragonal phase [e.g., 32; 38] or a mixture of tetragonal
and cubic zirconia phases  in most part of the particle. Possibly very small grains of
incompletely oxidized Zr (monoxide and/or α-Zr(O)?) are also present. The particle consists of
several clearly distinguishable zones, which could be roughly described as: 1) Zr-rich (the major
part), 2) U-rich (thick whitish veins in the middle of Fig. 3A), and 3) a complex structures
showing a decomposed solid solution. Numerous cavities inherited from gas bubbles and/or
contraction voids are also present. The main body consists of (Zr,U)O2 matrix with Zr-rich
grains with sizes 2-5 µm (depending on particular location) surrounded by numerous submicron
precipitates of (U,Zr)O2 phase (see also Fig. 46 in ). A single 3-4 µm UO2 rounded inclusion
is also present (fig. S-5, point 84; EDX data in tables S-6, S-7, row S4_84 in Supplementary
materials). Despite clear variations in visual appearance, the chemical composition of the Zr-rich
part of the S4 particle is relatively uniform (see WDX data in Supplementary materials, Tables
S-4, S-5, rows S4_07-12). Small sizes of the grains and their close spatial arrangement precludes
analysis of individual grains, but compositions averaged for regions of 100-150 µm2 are rather
similar: 72-76 wt% (81-82 mol%) ZrO2, 29-31 wt% (14-16 mol%) UO2 with minor admixture of
niobium 0.8-1.3 wt% (0.6-0.9 mol%; calculated as Nb2O5) and 0.5-1.6 wt% (0.9-2.9 ml%) of
iron (calculated as FeO). Small amounts of Cr2O3 (≤0.22 wt%) and of CaO (≤0.025 wt%) were
also detected. The four latter elements appear to be distributed non-systematically. Whereas the
chromium admixture most likely stems from structural steel, explaining of the CaO presence
(although at a trace level) is less obvious. Production of the E110 alloy does not involve Ca (e.g.,
). The most plausible source of Ca is the involvement of deposits from the cooling water,
which form on VVER and RBMK fuel rods during reactor operation (e.g., ). Other common
components of the deposits (e.g., Na, Cu, Mg) were not detected; possibly due to their
volatilization during the accident. Such deposits were obviously absent in most in-pile
experiments. As in case of the RL-1 particle in most cases the totals of WDS analyses deviate
from 100% and often lie between 103-105%. Direct measurements of O content may indicate
that the discrepancy could be due to understochiometric ZrO2-x. For detailed data on chemical
composition see Supplementary material.
The U-rich veins contain 57.5-64 wt% ZrO2, 39.3-43.3 wt% UO2 and 0.3-0.6 wt% FeO.
The thermal expansion coefficients of the veins should differ from that of the Zr-rich matrix,
which is reflected in cracks continuing the veins, which were likely formed during cooling.
The third part of the particle is partly surrounded by the U-rich veins and in its center
domains with crystallographically-oriented lamellae due to decomposition of U-Zr-O solid
solution are observed. Measurements of chemical composition of the lamellae are influenced by
the surroundings due to their small size, but the variations are rather limited and are within 76.7-
78.4 wt% ZrO2, 24.3-25.4 wt% UO2 and 0.1-0.2 wt% FeO. Adjacent regions (see points like 29
at Fig. S-2 of Supplementary materials) are somewhat enriched in urania and Fe (72, 29-30, 0.3-
0.4 wt % for the three oxides ZrO2/UO2/FeO). Angle between the lamellae is 120 degrees.
Therefore, the lamellae represent crystalline hexagonal Zr-rich phase, which has precipitated
during cooling of the particle. Similar phases were occasionally observed in FZK core melting
experiments (e.g., fig. 53 in  and fig. 72 in ).
The amount of the (U,Zr)O2 precipitates and size of the Zr-rich grains vary across the
particle and probably correlates with relative distance from the U-rich veins. Namely, the urania
precipitates become smaller and grainy structure of the Zr-rich grains is less obvious with
distance from the veins. In the lower part of the grain (see Fig. 3A) the precipitates are virtually
absent. At the same time, average chemical composition varies only slightly. This observation
can be explained if the hot U-rich vein was injected into the Zr-rich matrix. High temperature of
this vein should have caused high diffusion rate of O and U into the Zr-rich material, forming
supersaturated solid solution, which subsequently decomposed to sub-μm precipitates (e.g.,
). The injection scenario also explains enlargement and columnar shape of the (Zr,U)O2
grains on contact with the vein (see upper right part of Fig. 3A). Higher residual temperatures of
this material appear to be important for formations of the peculiar domains with
crystallographically-oriented lamellae. Note that in the upper part of the particle the (Zr,U)O2
grains seem to be surrounded not only by individual (U,Zr)O2 sub-μm precipitates, but also by
compositionally different veinlets. Such texture resembles observations of zirconia intergrain
attack by U-Zr-Fe-Cr melt found in , see their Fig. 10, but in our case both the grains and
injected material decomposed on cooling, presumably indicating more advanced stage of the
Several papers discuss evolution of various parameters of the reactor at the 4th Unit of the
Chernobyl NPP in course of the accident (e.g., [46-49]). During last hours of operation the
reactor showed rather complex behavior with progressive xenon poisoning, retraction of control
rods and complex hydraulics. Activation of the emergency (AZ-5) button by the personnel is
often taken as a starting moment of the active phase of the accident. Both direct examination of
the logged data  and various types of numerical modeling (e.g., [47-49]) show that increase
of fuel temperature took approx. 3-5 seconds, following dramatic increase of neutron power and
reactivity with a 1-2 sec. lag, which culminated with a (prompt) supercriticality event, leading to
the reactor’s destruction. Due to large size of the RBMK core and highly heterogeneous neutron
flux the maximum fuel temperature prior to the final explosion was highly variable and whereas
core-averaged values reach ~2000-2300 K [46, 47], calculated peak temperatures in energy-
stressed regions are much higher and considerably exceed the fuel melting temperature. For
example, according to  at 6.3 sec after the accident initiation up to 40 fuel assemblies reach
3300 K, by the 8th second more than 200 rods attain such temperatures (note that the exact values
are not very precise due to uncertainties of the model employed after the fuel melting). These
theoretical values are supported by examination of structural peculiarities [17, 18] and of Cs
content of hot particles [15, 16]. Relatively small size of the core fraction (0.01-0.1) where the
neutron flash (supercriticality event) took place was also inferred from measurements of
133mXe/133Xe ratio .
Expansion of the fuel due to increased power and lack of coolant flow closed the fuel-
zircaloy gap. Subsequent rupture of the tubes may lead to violent interaction of the molten fuel
(±zircaloy) with the coolant. According to observations immediately after the Chernobyl
accident ejected macroscopic fragments of the core (e.g., pieces of graphite masonry) were
heated to at least 1000 °C .
The fuel particle S2 bears no signs of interaction with zircaloy and/or melting. Large
grain size suggest that the particle came from relatively old fuel , which was probably far
from the explosion epicenter.
Complex microstructure and presence of Fe admixture in the particles S4 and RL-1
reflect processes, which occurred rather close to the epicenter of the explosion in the first
seconds of the accident. Both particles experienced fairly high temperatures. Presumably, the Zr-
based RL-1 particle was not fully molten which is suggested by preserved grain boundaries and
cavities. The initial metallic E110 interacted with fuel and steel components (mainly Fe) at
several spots, but uranium diffusion was not very extensive and was apparently limited by the
grain boundaries, suggesting moderate temperatures.
The S4 particle was likely formed by injection of U-rich melt and rapid subsequent
quenching. The proposed “injection” does not necessarily imply high speed process, it could
have been closer to molding. Some important characteristic features observed in this particle,
such as decay of Zr-U-O solid solution with characteristic formation of oriented lamellae and
abundant sub-μm UO2 precipitates have similarities with samples obtained in CORA-13 and
CORA-16 experiments at FZK [40, 43]. These features were encountered only at levels where
the highest temperatures leading to complete melting of fuel and zircaloy or very extensive
zircaloy oxidation has occurred. In the CORA-13 experiment temperatures at the corresponding
elevations were 1850 °C during the transient and up to 2200 °C during quench. Obviously the
timescale of the CORA experiments is much longer (the heating rate of 1 °/sec) than at
Chernobyl NPP and one may assume that in the latter case significantly higher temperatures
were required to form similar features.
A very important feature of the RL-1 and S4 particles is the presence of iron and
chromium, which are always associated with uranium. The only plausible source of these
elements is the melting of the spacer grids and their presence clearly indicate that relocation of
the steel-zirconium-fuel melt has occurred before the explosion. The RL-1 particle likely
represents a spot on freshly oxidized zircaloy on the very front of the relocating melt. The S4
particle shows more advanced stage of the fuel-Zr interaction and might have been formed on
contact of relocating molten fuel with severely oxidized cladding or oxidation of Zr-U-O eutectic
Absolute amounts of 137Cs and 241Am in the particles depend on several factors such as
fuel burn-up, temperature regime and on behavior of Cs and Am in the Zr-U-O melt. The highest
137Cs amount is found in the fuel particle S2, which corresponds to a relatively low-temperature
and/or brief heating of the particle. Though microstructural analysis of RL-21 and S4 suggests
notably higher temperatures for the latter particle, although the 137Cs/241Am ratio are similar –
3.92 and 3.72, respectively. This could be explained by an annealing sufficient for removal of a
significant Cs fraction, though absolute values of the temperature could have been different.
It is interesting to note that the Zr-U-Fe-O material is encountered not only in hot
particles, but also in lava-like ceramics formed after the core destruction. Figure 4 shows a
chemically and morphologically complex piece of Zr-U-O phase in the black Chernobyl “lava”
(for details see ). Mapping of Fe distribution reveals not only obvious round spots due to
spheres of molten steel, but also presence of iron in the Zr-U-O pieces. The most logical scenario
is that the Zr-U-Fe-O material was formed prior to the explosion and was subsequently
incorporated into the “lava”.
The three examined particles show high variability of material ejected from the core of the 4th
Unit ChNPP during the first moments of the accident. They differ not only in composition and
structure, but also in conditions experienced at their location prior to the ejection. One of them -
the S2 - a piece of dispersed fuel (UO2+x), was not melted and did not contact with the Zr
cladding. It retains grain boundaries with fission gas release bubbles and high content of volatile
cesium. Another one, #RL1, is a piece of the oxidized zircaloy cladding and mostly consists of
monoclinic zirconia with small domains manifesting uranium diffusion. The particle was likely
to be formed at an early stage of the interaction of zircaloy with fuel and structural steel (mainly
Fe). The temperature and/or reaction time was insufficient for extensive U diffusion into the
cladding and the diffusion front was largely limited by pre-existing grain boundaries. The third
particle, #S4, (mostly tetragonal Zr-U-O) has witnessed contrasting local conditions at the
boundary of the (oxidized?) zircaloy -–and UO2 fuel. As a result of the different melting points
of closely spaced phases, partial melting of the individual phases occurred along with mutual
interdiffusion of the melt and solid components.
Acknowledgements. The work was partly supported by IAEA CRP Contract No. 20546 and by
RFBR grant 16-03-00944. Raman measurements were performed using equipment of CKP FMI
IPCE RAS. We appreciate useful discussions with Dr. M.Veshchunov and thorough reviews of
Prof. P.D.W.Bottomley and of an anonymous reviewer.
The raw and processed data required to reproduce results of the EDX and WDX
measurements are available as Supplementary materials to the paper.
 Burakov B.E., Shabalev S.I., Anderson E.B., Principal Features of Chernobyl Hot Particles:
Phase, Chemical and Radionuclide Compositions. In S. Barany, Ed. Role of Interfaces in
Environmental Protection, Kluwer Academic Publishers, NATO Science Series, Earth and
Environmental Sciences, 24, 2003, pp. 145-151.
 Sandalls F.J., Segal M.G., Victorova N. Hot particles from Chernobyl: a review, J. Environ.
Radioactivity 18 (1993) 5-22.
 Bogatov S.A., Borovoi A.A., Dubasov Yu.V., Lomonosov V.V., Form and parameters of the
particles of the fuel ejection in the Chernobyl reactor accident. Soviet Atomic Energy 69(1)
 Devell L., Tovedal H., Bergstrom U., Appelgren A., Chyssler J., Andersson L., Initial
observations of fallout from the reactor accident at Chernobyl. Nature 321 (1986) 192-193.
 Schubert P. and Behrend U. Investigations of Radioactive Particles from the Chernobyl Fall-
out. Radiochimica Acta, 41 (1987) 149-155
 Dubasov Yu.V., Krivokhatskii A.S., Savonenkov V.G., Smirnova E.A., State and behavior of
dispersed fuel particles ejected from 4th Unit of Chernobyl NPP, Soviet Radiochemistry, 33(1)
 Petelin G.I., Zimin Yu.I., Tepikin V.E., Rybalka V.B., and Pazukhin E.M., Fuel ”Hot”
Particles of the Chernobyl Accident in the Retrospective Evaluation of the Emergency Processes
at the 4th Block of the Chernobyl Nuclear Power Plant, Radiochemistry 45(3) (2003) 304-308.
 Ganeev A., Gubal A., Korotetski B., Bogdanova O., Burakov B., Titova A, Solovyev N.,
Ivanenko N., Drobyshev E., Iakovleva E., Sillanpää M. Direct isotope analysis of Chernobyl
microparticles using time-of-flight mass spectrometry with pulsed glow discharge,
Microchemical J., 132 (2017) 286-292.
 Lancsarics Gy., Feher I., Sagi L., Palfalvi J., Transuranium elements in the hot particles
emitted during the Chernobyl accident. Radiation protection Dosimetry, 22(2) (1988) 111-113.
 Loshchilov N.A., Kashparov V.A., Yudin E.B., Protsak V.P., Fractionation of radionuclides
in Chernobyl fuel hot particles. Soviet Radiochemistry, 34(5) (1992) 125-134.
 Tcherkezian (Cherkezyan) V.O., Kashkarov L.L. Shalaeva T.V., Korovaikov P.A.,
Investigation of composition and distribution of Chernobyl “hot particles”. Geochemistry
International, 32(2) (1994) 252-261.
 Boulyga S.F., Erdmann N., Funk H., Kievets M.K., Lomonosova E.M., Mansel A.,
Trautmann N., Yaroshevich O.I., Zhuk I.V., Determination of isotopic composition of plutonium
in hot particles of the Chernobyl area, Radiation measurements, 28(1-6) (1997) 349-352.
 Dubasov Yu.V., Krivokhatskii A.S., Savonenkov V.G., Smirnova E.A., Fractionation of
radionuclides in fall-out of near zone of the Chernobyl NPP, Soviet Radiochemistry, 33(4)
 Loshchilov N.A., Kashparov V.A., Polyakov V.D., Yudin E.B., Protsak V.P., Zhurba M.A.,
Parshakov A.E., Nuclear-physical properties of hot particles formed as a results of accident at
Chernobyl NPP. Sov. Radiochemistry, 34(4) (1992) 113-119.
 Kashparov V.A., Ivanov Yu.A., Khomutinin Yu.V., Pazukhin E.M., An estimate of
effective temperature and annealing duration of fuel particles ejected from Chernobyl reactor
during the accident, Radiochemistry, 38(1) (1996) 84-89.
 Kashparov V.A., Ivanov Yu.A., Protsak V.M., Khomutinin Yu.V., Yoshchenko V.I.,
Pazukhin E.M., An estimate of maximal effective temperature and of duration of non-isothermal
annealing of Chernobyl fuel particles during the accident. Radiochemistry 39(1) (1997) 63-67.
 Burakov B.E., Anderson E.B., Galkin B.Ya., Pazukhin E.M., Shabalev S.I., Study of
Chernobyl “hot” particles and fuel containing masses: implications for reconstruction the initial
phase of the accident, Radiochimica Acta, 65 (1994) 199-202.
 Burakov B.E., Anderson E.B., Shabalev S.I., Strykanova E.E., Ushakov S.V., Trotabas M.,
Blanc J.-Y., Winter P., Duco J., The behavior of nuclear fuel in first days of the Chernobyl
accident, in Scientific Basis for Nuclear Waste Management XX, edited by W.J. Gray, I.R. Triay
(Mat. Res. Soc. Symp. Proc. 465, Warrendale, PA), 1997, 465, pp. 1297-1308.
 Shabalev S.I., Burakov B.E., Anderson E.B., General classification of “hot” particles from
the nearest Chernobyl contaminated areas. in Scientific Basis for Nuclear Waste Management
XX, edited by W.J. Gray, I.R. Triay (Mat. Res. Soc. Symp. Proc. 465, Warrendale, PA), 1997,
465, pp. 1343-1350.
 Poml P. and Burakov B., Study of a “hot” particle with a matrix of U-bearing metallic Zr:
Clue to supercriticality during the Chernobyl nuclear accident, Journal of Nuclear Materials 488
 Poml P. and Burakov B., Study of the redistribution of U, Zr, Nb, Tc, Mo, Ru, Fe, Cr, and
Ni between oxide and metallic phases in the matrix of a multiphase Chernobyl hot-particle
extracted from a soil sample of the Western Plume, Radiochimica Acta (2018), DOI:
 Ogorodnikov B.I.; Pazuhin E.M.; Klyuchnikov A.A., Radioactive aerosols of the "Shelter"
Object 1986-2006. Institute for safety problems of NPP NASU, 2008. 456 p.
 Izrael Yu.A., Radioactive Fallout After Nuclear Explosions and Accidents, Elsevier, 2002,
 Downs, R.T. and Hall-Wallace, M. The American Mineralogist Crystal Structure Database.
American Mineralogist, 88 (2003) 247-250. Arutyunyan R.V., Bolshov L.A., Borovoi A.A.,
Velikhov E.P., Klyuchnikov A.A., Nuclear Fuel in the «Shelter» Encasement of the Chernobyl
NPP. Nauka, Moscow, (2010) p. 240.
 Rondinella V.V. and Wiss T., The high burn-up structure in nuclear fuel, Materials Today,
13(12) (2010) 24-32.
 Walker C.T., Knappik P., and Mogensen M., Concerning the development of grain face
bubbles and fission gas release in UO2 fuel, Journal of Nuclear Materials, 160 (1988) 10-23.
 Bottomley P.D.W., Bremier S., Papioannou D., Walker C.T., EPMA and X-ray diffraction
of the degraded fuel bundle from the Phebus FPT1 test, Mikrochimica Acta, 139 (2002) 27-38.
 Graves P.R., Raman Microprobe Spectroscopy of Uranium Dioxide Single Crystals and Ion
Implanted Polycrystals, Applied Spectroscopy, 44(10) (1990) 1665-1667.
 He H. and Shoesmith D., Raman spectroscopic studies of defect structures and phase
transition in hyper-stoichiometric UO2+x. Phys. Chem. Chem. Phys. 12 (2010) 8108-8117.
 Desgranges L., Baldinozzi G., Simon P., Guimbretière G., Canizares A., Raman spectrum of
U4O9: a new interpretation of damage lines in UO2, J. Raman Spectroscopy 43(3) (2012) 455-
 Kim D-J., Jung H-J., Yang I-S.. Raman spectroscopy of tetragonal zirconia solid solution J.
Am. Chem. Soc. 76(8) (1993) 2106-2108.
 Kim B.-K. and Hamaguchi H., Mode Assignments of the Raman Spectrum of Monoclinic
Zirconia by Isotopic Exchange Technique, Physica status solidi (b), 203(2) (1997) 557-563.
 Vlasova I., Kalmykov S., Sapozhnikov Yu., Simakin S., Anokhin A., Aliev R., and Tsarev
D., Radiography and local microanalysis for detection and examination of actinide-containing
microparticles. Radiochemistry 48(6) (2006) 613 - 619.
 Pöml P., Burakov B., Geisler T., Walker C.T., Grange M.L., Nemchin A.A., Berndt J.,
Fonseca R.O.C., Bottomley P.D.W., Hasnaoui R., Micro-analytical uranium isotope and
chemical investigations of zircon crystals from the Chernobyl ‘‘lava’’ and their nuclear fuel
inclusions, Journal of Nuclear Materials, 439 (2013) 51-56.
 Crepin J. Bretheau T. (1996) Crystallographic growth of cavities in zirconium (1996) In:
IUTAM Symposium on Micromechanics of Plasticity and Damage of Multiphase Materials, Eds.
Pineau A. and Zaoui A., pp. 183-190.
 Griffiths M., A review of microstructure evolution in zirconium alloys during irradiation.
Journal of Nuclear Materials, 159 (1988) 190-218.
 Huang W., Yang J., Wang Ch., Zou B., Meng X., Wang Y., Cao X., Wang Zh., Effects of
Zr/Ce molar ratio and water content on thermal stability and structure of ZrO2-CeO2 mixed
oxides prepared via sol-gel process. Materials Research Bulletin, 47 (2012) 2349-2356.
 Kontoyannis C.G., Orkoula M., Quantitative determination of the cubic, tetragonal and
monoclinic phases in partially stabilized zirconias by Raman spectroscopy. Journal of Materials
Science, 29 (1994) 5316-5320.
 Burbach J. (1993) Ergebnisse von REM Mikrobereichsanalysen des DWR-
Bündelabschmelzexperiments CORA-13, KfK report 5162.
 Pylypenko M.M., Production of nuclear purity zirconium. Problems of atomic science and
technology, 92 (2008) 66-72
 Smirnova I.M., Kuchkina I.N., Kostyuchenko A.N., Kuprienko M.V., Determination of the
composition of the deposits and thickness of the oxide film on the surface of VVÉR and RBMK
fuel elements. Atomic Energy 98(1) (2005) 46-49.
 Burbach J. (1994) Ergebnisse von REM/EDX Mikrobereichsanalysen des
Siedewasserreaktor-Bündelabschmelzexperiments CORA-16, KfK report 5282.
 Veshchunov M.S., Boldyrev A.V., Shestak V.E., Mueller K., Analysis of molten pool
physico-chemical interactions and interpretation of the PHEBUS FP tests observations, Nuclear
Engineering and Design 238 (2008), 1728-1742.
 Bottomley P.D.W., Stalios A.D., Glatz J.-P., Satmark B., Walker C.T., Examination of
melted fuel rods and released core material from the first Phebus-FP reactor accident experiment.
Journal of Nuclear Materials, 278 (2000) 136-148.
 Information on the accident at the Chernobyl nuclear power plant and its consequences
prepared for IAEA. Soviet Atomic Energy, 61(5) (1986) 845-868.
 Fletcher C.D., Chambers R., Bolander M.A., and Dallman R.J., Simulation of the Chernobyl
accident, Nuclear Engineering and Design, 105 (1988) 157-172.
 Abagyan A.A., Arshavskii I.M., Dmitriev V.M., Kroshilin A.E., Krayushkin A.V.
Khalimonchuk V.A., Computational Analysis of the initial stage of the accident at the
Chernobyl' atomic power plant, Atomic Energy, 71(4) (1991) 785-795.
 Afanas'eva A.A., Fedosov A.M., Donderer R., Ehrrenstein D., Liermann R., Schumacher
O., and Ziggel H., Analysis of the Chernobyl accident taking core destruction into account,
Atomic Energy, 77(2) (1994) 573–579.
 Pakhomov S.A., Krivokhatsky K.S., and Sokolov I.A., Estimate of the instantaneous energy
release in the Chernobyl NPP reactor accident based on the ratio of of 133Xe and 133mXe activities
in Air, Radiochemistry, 33(6) (1991) 685-688.
 Platonov P. A. and Krayushkin A. V., Process resulting in the destruction of the No.4 unit
reactor at the Chernobyl nuclear power plant. Atomic Energy, 104(6) (2008) 417-427.
 Shiryaev A.A., Vlasova I.E., Burakov B.E., Ogorodnikov B.I., Yapaskurt V.O., Averin
A.A., Pakhnevich A.V., Zubavichus Y.V., Physico-chemical properties of Chernobyl lava and
their destruction products, Progress in Nuclear Energy 92 (2016) 104-118.
Figure 1. Fuel particle S2. A-C – SEM images in SE mode. C – unpolished barren surface.
D – Raman spectra from two spots. Arrows in B point to channel-like pores, see text for details.
Figure 2. Zirconia particle R-1. A – image of the particle prior to polishing. B – BSE general
view. C – Raman spectrum. D – Zoomed BSE image; faceted void is highlighted. E-G – BSE
images showing patchy distribution of U admixture. H – symplectite-like structure, SE-mode.
Figure 3. Complex particle S4. A, B – general view, BSE and SE modes, respectively. White
arrow in A shows the UO2 particle, the polygon shows the region with columnar (Zr,U)O2 grains
see text for detail. C – Raman spectra from several spots. Arrows indicate positions of peaks due
to tetragonal ZrO2 phase. D-F – zoomed BSE images of a region with complex mixture of
various phases. G – Chemical maps of U, Nb and Zr spatial distribution.
Figure 4. Piece of Zr-U-O particle in lava-like material from ChNPP, see text for details. Left –
SE image, right – enhanced map of Fe distribution.
Table S-1. Activity, Bq/particle, of γ-emitting nuclides in three Chernobyl “hot” particles
"Hot" particles from the Chernobyl soils
*measured in 1999, recalculated on the date of accident;
**ND - not detected
SEM-EDX of the particle RL1
Fig. S-1. Location of EDX analyses of the particle RL1.
Table S-2. Weight % of the oxides in different spots of S4, SEM-EDX
1 In spectra with strong Zr contribution a relatively weak Nb EDX-signal can not be correctly measured because of
the overlap of the L alpha lines of Nb and Zr. WDX data for Nb are certainly more correct (see Table S-2 for Nb
SEM-WDX of the particle RL1
Fig. S-2. Location of WDX analyses of the particle RL1.
Table S-3. Weight % of the oxides in different spots of S4, SEM-WDX
Point of analysis
Figure S-3. Atomic ratio per 24 atoms of O for Zr and U in the particle RL1 (WDX data).
SEM-WDX of the particle S4
04 05 06
Fig. S-4. Location of WDX analyses of the particle S4.
Table S-4. Weight % of the oxides in different spots of of S4, SEM-WDX
Table S-5. Molar % of the oxides in different spots of of S4. SEM-WDX
SEM-EDX of S4
Fig. S-5. Location of EDX analyses of the particle S4.
Table S-6. Weight % of the oxides in different spots of S4. SEM-EDX
Table S-7. Molar % of the oxides in the different areas of S4. SEM-EDX