Characterisation of aerosols produced in a simulated severe nuclear accident using electron microscopy

Characterisation of aerosols produced in a simulated severe nuclear accident using electron microscopy Contents lists available at ScienceDirect Journal of Aerosol Science journal homepage www elsevie[.]

Contents lists available atScienceDirect Journal of Aerosol Science journal homepage:www.elsevier.com/locate/jaerosci

Characterisation of aerosols produced in a simulated severe nuclear

accident using electron microscopy

K Knebela,b,⁎,1, P.D.W Bottomleya, V.V Rondinellaa, A Lähdeb, J Jokiniemib

a European Commission, Joint Research Centre, Institute for Transuranium Elements, P.O Box 2340, D-76125 Karlsruhe, Germany

b University of Eastern Finland, Department of Environmental Science, Fine Particle and Aerosol Technology Laboratory, P.O Box 1627, FIN-70211

Kuopio, Finland

A B S T R A C T

The chemical composition of aerosols formed by revaporisation and subsequent condensation of fission products and simulant material was investigated using electron microscopy A stage dedicated only to the collection of aerosol particles on carbon coated copper grids was designed and tested As raw material for these experiments core material from the TMI-2 accident,fission product deposits from the Phébus FPT3 tests and Cs2MoO4powder were used The examination

of aerosols was made by transmission and scanning electron microscopy and energy-dispersive X-ray spectroscopy The results indicate the range of species encountered at different temperatures showed the necessity for a time dependent sampling method The investigated aerosols were in the nanoparticle range A clear pattern was found for the use offission product deposits from the Phébus FPT3 experiments In oxidising atmosphere in the temperature range of 450 K to 910 K a species of 2 nm sized spherical caesium-rhenium particles were observed that coagulated to spherical about 20 nm sized particles Other elements such as Mo and Te occurred at various temperature ranges and atmospheres Also the vaporisation of caesium molybdate, as predicted

in thermodynamic equilibrium calculations, was examined

1 Introduction

In the case of a severe nuclear accident the main goal is to prevent the release of radioactive fission products into the environment After the Great Eastern Tsunami on March 11, 2011 in Japan,fission products (FP´s), especially the volatile species, leaked into the environment It is estimated that 0.6−1.5×1019

Bq of133Xe, 0.7−5.0×1017

Bq of131I and 1.0−5.0×1016

Bq137Cs have been released into the atmosphere (Koo, Song & Yang, 2014) Traces of the released isotopes were measured in locations in Asia, for example China (Tuo et al., 2013) and Vietnam (Long et al., 2012) and also in Europe, for example Finland (Kettunen, Kontro, Leppanen & Mattila, 2013) and France (Evrard et al., 2012) This shows that such an accident is not a regional but a global threat To assess these threats, nuclear safety codes, such as ASTEC (Chatelard et al., 2016) are used to simulate possible scenarios and their impact Although the state of the art nuclear safety codes use thermodynamic equilibrium calculations to model the chemical reactions between the different elements interacting during an accident, they rely on experimental data for the chemical species formed (Allelein et al., 2009) The objective of this study is to enhance the knowledge of Cs behaviour in severe accident conditions Due to release mechanism offission products vaporising from the degraded fuel and subsequent condensation an aerosol formation when released into the atmosphere, a special attention is given to vaporisation and condensation in this study In 1988 the Phébus FP

http://dx.doi.org/10.1016/j.jaerosci.2017.01.008

Received 8 May 2016; Received in revised form 27 January 2017; Accepted 31 January 2017

⁎ Corresponding author at: European Commission, Joint Research Centre, Institute for Transuranium Elements, P.O Box 2340, D-76125 Karlsruhe, Germany.

1 present address: Hauptstraße 270, 50169 Kerpen, Germany

E-mail address: KevinKnebel@gmx.de (K Knebel).

Journal of Aerosol Science 106 (2017) 68–82

Available online 01 February 2017

0021-8502/ © 2017 Elsevier Ltd All rights reserved.

MARK

programme was launched by the French‘‘Institut de Radioprotection et de Sûreté Nucléaire’’ (IRSN) in cooperation with the European Commission (EC) and other international partners to investigate light water reactor accidents and the subsequent release of fission products (March & Simondi-Teisseire, 2013) As a follow-up to these experiments, parts of the experimental line withfission product deposits, were used for separate effect tests to examine the behaviour of137

Cs more closely.137Cs is of special interest as it is produced in high amounts during thefission process; it has a high volatility and with 30.2 years’ half-life that makes it a long term problem The initial facility at JRC-Karlsruhe was used to study the revaporisation of137Cs of relatively active samples by monitoring the gamma-radiation emitted fromfission products remaining on the sample inside the furnace (Auvinen, Bottomley, Jokiniemi, Knebel & Rondinella, 2014a) and the total deposits collected onfilters to give the total revaporisation of all other (gamma –emitting) FP´s Thus a decision was made to upgrade the device with a means to collect the aerosols and gather information about the chemical composition and size of the aerosols that form during the rapid cooling of the revaporisedfission products To get the particles in an untainted state the collection was made directly at their place of origin by impaction on to 3 mm in diameter copper grids with a carbon substrate layer used for Transmission Electron Microscopy (TEM)

In this paper two sorts of samples were examined: TMI-2 molten core samples and Phébus PF FPT3 vertical line deposits dominated by137Cs In addition, two simplified experiments with Cs2MoO4were also performed The TMI-2 and Phébus samples were chosen due to the fact that they represent the very rare case of deposits of irradiated material that was produced under accident conditions Cs2MoO4was chosen because it represents the chemical form, which was predicted from thermodynamic calculations, but could not be observed in our previous experiments (Auvinen, Bottomley, Jokiniemi, Knebel & Rondinella, 2014b) The grids were exchanged during the sample heating, particularly at high revaporisation rates, and were examined by either TEM or SEM microscopy

to determine size and composition where possible Cs as the major gamma emitting isotope was observed along with Re originating from thermocouples in the original Phébus tests Small amounts of Mo and Te also observed in the Phébus FPT3 samples

2 Materials and methods

2.1 Experimental set-up

2.1.1 Fundamental set-up

The basic design of the revaporisation set-up is shown inFig 1and consists of a furnace, a diluter and afilter stage that are connected to a process gas and a diluter gasflow A stainless steel (1.4301) tube with 375 mm length and 21.5 mm diameter runs through the furnace The tube can be opened to insert the sample into the furnace and has the process gas connected at the furnace inlet The furnace can reach up to 1273 K while being continuously provided with a steady gasflow through the process gas inlet Hydrogen, air, nitrogen and argon are available as process gases with 0.1 to 8 l/minflow rates Additionally, steam is available at 5 to

120 g/h massflow rates (~0.1 to 0.5 l/min nitrogen is needed as carrier gas for the steam) Any vapour produced during the experiment is transported down the steel tube in the furnace and passes to the diluter The diluter consists of metallic cylinder housing a porous sinter metal tube The outside of the porous tube is provided with aflow of 150 l/min nitrogen gas at room temperature whichflows through the porous tube and mixes with the hot process gas flow resulting in a rapid cooling of the process gas along its length and any vapour transported by it As the temperature of the process gas is lowered from up to 1173 K (the process gas never reaches the maximum furnace temperature of 1273 K as there is always an offset between furnace and sample temperature) down below 300 K most condensable species solidify and form aerosols while remaining suspended in the turbulent gasflow These aerosols are collected inside the aerosol sample stage, which is connected to the diluter outlet The gasflow finally enters the filter stage where itflows along a high flow rate quartz tube filter to collect any aerosols on its inner cylindrical surface before the gas flow

Fig 1 Schematics of the revaporisation set-up used to examine radioactive sample with dose rates up to 25 mSv/h The surrounding 5 mm thick lead shielding is not shown.

leaves the glove box drawn by a vacuum pump into the glove box exhaust system The vacuum pump also ensures a steadyflow in the system as well as preventing anyfission product from contaminating the glove box It also prevents any hydrogen from leaking during the experiments into the glove box and thus prohibits the potential build-up of an explosive atmosphere Additionally, the revaporisation set-up is equipped with a scintillation gamma-ray detector above the glove box used to measure the revaporisation rate of detectablefission products A more detailed description of the set-up can be found in (Knebel et al., 2014a)

2.1.2 Aerosol sampling

Thefirst step in examining the aerosols produced in the diluter was to take small pieces of the inner surface of the cylindrical fibrous quartz filter where they were collected and examine them with scanning electron microscopy (SEM) This proved to be unsuccessful, as single particles could not be identified due to their small size and the uneven structure of the quartz fibres Nevertheless, the energy-dispersive X-ray spectroscopy (EDS) showed the presence of traces of uranium and caesium Due to the fast cooling process in the diluter and the turbulentflow conditions, it was anticipated that the condensed aerosols that form are in the submicron region and therefore TEM combined with EDS would be a more successful analysis method This also has the benefit that the samples produced are very small and thus also emit lower radiation than accumulated samples produced with a cascade impactor

Atfirst a simple collection device using TEM-grids was used (Fig 2) and the grids did have deposits offission product aerosols; however, a shorter stage was designed that was able to exchange the samples during the experiment As the furnace is heated slowly over several hours with 2 K/min this enables the aerosols collected to be assigned to a corresponding temperature Thisfilter stage is shown in the middle ofFig 2 The TEM-grid is placed on a 4 mm wide stainless steel sample holder and isfixed with a small hinge and screw The aerosol sampling stage itself consists of a short stainless steel pipe withflanges on each side, a hole on its upper side to insert the TEM-grid and a simple locking mechanism to fix the holder of the TEM-grid while sealing the system Finally polyoxymethylene was used instead of steel for the sampling stage This has the benefit that one flange connection is required The assembled aerosol sampling stage /filter cartridge combination is shown at the bottom ofFig 2 Each combined sampling stage

is now used only for one experiment and discarded afterwards The experiments showed that even within the limited space of the glove box, the exchange of a TEM-grid is possible in about 1 min The combination of this short amount of time and the under pressure produced by the vacuum pump excluded any significant contamination of the glove box

2.2 Fission product samples

The initial material for the aerosols examined in this study originated from two different sources The first were degraded fuel samples from the TMI-2 severe nuclear accident that were examined during an OECD/NEA project (Akers, 1992)

The TMI-2 samples were taken from the completely molten central mass of the degraded core and were in the form of afine dark powder The initial examination (Bottomley & Coquerelle, 1992) showed three different oxide phases; these were a uranium-rich [(U,Zr)O2], a zirconium-rich [(Zr,U)O2] and a ferrous phase containing structural material from the reactor core, (iron, chromium, nickel and aluminium) The examination showed also the overall atomic composition of the metallic phases of this sample to be 18%

U, 23% Zr and 59% (Fe, Cr, Ni, Mn and Al) The core samples were in powder form since they are remnants from the cutting performed in the initial examination These samples were used for the experiments as they still contained a significant amount of radioactive137Cs, which is the main target of the revaporisation studies

Fig 2 The upper photo shows the initial sampling method, in the middle a CAD drawing of the improved section is shown and at the bottom is a photo of the resulting combined filter and sampling section.

K Knebel et al. Journal of Aerosol Science 106 (2017) 68–82

The second type of sample werefission product deposits formed on the upper vertical line of the Phébus FPT3 experimental circuit (Haste, Payot, & Bottomley, 2013;Haste et al., 2013;Hache, Schwarz, von der Hardt, 1999), which was kept at a temperature of

973 K during the Phébus experiments The samples from the Phébus FPT3 experiment were of a different nature as they contained fuel only in trace amounts They consist of layers offission products that first vaporised from the test bundle during the FPT3 experiment and then condensed on the inside of the stainless steel/Inconel pipes in the upper vertical line just above the degrading bundle Ring-shaped pieces were cut out of the piping and used as the source material for these experiments The rings were cut as needed to smaller pieces and thenfixed with alumina glue on a stainless steel sample holder with the fission product deposit facing upward Thefission product deposit is very brittle and substantial amounts are lost during the cutting process Nevertheless, the activity emitted by the samples is still quite high at about 1–20 mSv/h dose rate for a 10×10 mm² sample The gamma radioactivity was over 97% due to the134Cs and predominantly137Cs isotopes Previous analysis (Bottomley et al., 2014) showed that an roughly

1 cm² sample of the upper vertical line above the FPT3 degrading bundle would have after cutting (thicker outer layers may be broken off during cutting) an adherent deposit with the following approximate composition:

Fission Products (natural isotopes excluded - except Sb):

i) Major: Cs ~50 μg, Te ~75 μg, Mo ~20 μg (+50% nat), Rb ~5 μg,

ii) Minor: Tc-99: ~1.5 μg, Sb ~ 0.4 μg, Ag ~0.3 μg; Zr ~0.3 μg,

Fissile material: U ~4 μg, Control material B (from B4C rod): 200 μg

Substrate materials: eg.: Fe, Cr, Ni, Zn, Cu

Nevertheless, ICP-MS analysis of the deposits themselves and the aerosols collected on the main outletfilters demonstrated that many otherfission products were present and could also revaporise under these conditions Substrate material (Fe, Cr, Ni) and thermocouples materials (W, Re) as well as fuel traces (U, Pu) and B absorber were also present in the deposit and on thefilters Pb traces detected were impurities from the revaporisation shielding The results are reported in (Bottomley et al., 2014)

2.3 Experimental procedure

The most important parameters of the experiments are the samples used, the furnace/sample temperatures; the process gas that is fed into the furnace and the time-frame in which a TEM-grid was inserted for aerosol sampling The information regarding process gas, temperature of furnace and sample and the period that a TEM-grid remained inside the furnace is shown for each experiment in Fig 3andFig 4 The TEM-grids were also labelled according to the experiment and the position in that experiment, so 5-2 is the second TEM-grid in experiment 5 The choice of the heating profile and the carrier gas was based on previous experiments (Knebel

et al., 2014b) and the results of the137Cs kinetics measured by gamma spectroscopy As these experiments were thefirst tests after an upgrade of the facility, the same mixtures of the process gases air, steam and hydrogen were used However, in the present study the focus is based on the sampled aerosols The data collected with gamma spectroscopy regarding the Cs kinetics will be analysed and discussed in a future study As mentioned inSection 2.1.1., the process gas and thus also the sample never reach the temperature to which the furnace is programmed There is an offset that is at its maximum of up to 180 K (experiment 6), depending on the furnaces temperature, the gasflow and mixture used If not described separately, all temperatures in the following are the sample temperature 2.3.1 Experiment 1: TMI-2 powder in air atmosphere

After a number of preparatory tests, experiment 1 used a sample from the TMI-2 core consisting of 1.2 gfine powder produced during sample cutting in the initial TMI-2 fuel exploration programme The sample wasfirst heated under nitrogen as inert gas (1 l/ min) to 515 K by 10 K/min, then the process gas was switched to air (1 l/min) and the sample remained at 573 K for 10 min until the next heating phase with 2 K/min began This continued until the maximum temperature of 1171 K was reached with a dwelling time

of 60 min Afterwards the cooling phase started with 10 K/min (seeFig 3) Three TEM-grids were present using the initial sampling method and retrieved the next morning

2.3.2 Experiment 2: TMI-2 powder in hydrogen atmosphere

The same sample from experiment 1 was reused as the online gamma measurements showed no loss of137Cs from the sample under air Also, as shown inFig 3, the same heating profile was used but the atmosphere after the initial heat up (10 K/min) to 530 K was hydrogen (1 l/min) Three TEM-grids were used to collect the aerosols (exposed for the whole experiment with the initial sampling method), but only two of them were retrieved successfully

2.3.3 Experiment 3 Phébus sample in alternating air and hydrogen atmospheres

In experiment 3 a Phébus sample was used and was thefirst using the improved sampling method which allows the TEM-grids to

be exchanged while the experiment is running This process took ~5 minutes, later it was possible to perform this in about 1 min

As in previous experiments (Knebel et al., 2014b) the revaporisation of137Cs did not start before reaching a furnace temperature

of 773 K, the heating profile was changed, as shownFig 3 The sample was heated under air (2 l/min) by 10 K/min up to 680 K and left for 30 min at this temperature, during which thefirst TEM-grid (3-1) was replaced The following heating phase was at 2 K/min

up to 827 K followed by a dwell phase of 60 min during which the atmosphere was changed to hydrogen (2 l/min) for 20 min, simulating a short hydrogen surge in a severe accident scenario After this hydrogen phase the process gas was switched back to air (2 l/min) and the TEM-grid (3-2) was replaced and the third heating phase up to 1165 K at 2 K/min started At temperature of

~1050 K the gamma detectors showed137Cs revaporising at a high rate therefore the TEM-grid (3-3) was replaced again and afterwards the atmosphere was changed to hydrogen (2 l/min) After reaching maximum temperature of 1165 K, the furnace rested there for 60 min and then the cooling phase at 10 K/min was initiated After changing the process gas from hydrogen to nitrogen (8 l/ min) the TEM-grid (3–4) was changed and the system was left overnight to cool to room temperature under nitrogen The final TEM-grid (3–5) was retrieved the next morning

2.3.4 Experiment 4: Phébus sample in alternating steam and hydrogen atmosphere

For experiment 4 another sample from Phébus was treated with the same temperature profile as the previous experiment, as shown inFig 4 The atmosphere was steam (60 g/h) and nitrogen traces (0.5 l/min) or hydrogen (2 l/min) The steam atmosphere required small amounts of nitrogen in the mixing system to remain stable and prevent condensation before the furnace TEM-grid 4-1 was taken out at the beginning of thefirst dwell phase at 523 K and 4-2 was introduced at 608 K that was left inside during the following steam phase The temperature difference between taking 4-1 out and introducing 4-2 was due to the amount of time that was needed to change the samples As the137Cs showed a less volatile behaviour compared to the previous experiment the hydrogen phase was brought forward to 772 K with the TEM-grid 4-3 being introduced beforehand Subsequently the process gas was changed

to steam and later back to hydrogen Most of the137Cs had revaporised at this time and the hydrogen phase was interrupted by a short nitrogen phase to change the grid to 4-4 which remained in place until the beginning of thefinal cooling using nitrogen The cooling grid 4–5 was then retrieved the next morning

2.3.5 Experiment 5: Phébus sample in alternating air and hydrogen atmosphere

In experiment 5 the sample originated from a slightly higher position in the Phébus upper vertical line where the deposit showed a higher (predominantly > 97%) Cs radioactivity This sample was treated with either air or hydrogen (2 l/min each) in an alternating pattern and with the same heating profile as the previous two experiments (seeFig 4) Grid 5-1 was introduced shortly before

Fig 3 Temperature profile of furnace and sample and process gas as used in experiments 1, 2 and 3 The sampling period of each TEM grid is represented by the dashed black horizontal lines The process gas is shown by the background colour as described at the top of this figure.

K Knebel et al. Journal of Aerosol Science 106 (2017) 68–82

reaching thefirst dwelling phase As the sample showed no137Cs release at ~800 K furnace temperature, as with every previous experiment (Knebel et al., 2014b) with oxidising atmosphere, the planned hydrogen phase was delayed and the grid left inside until thefirst significant activity loss was observed around 994 K, at which point grid 5-1 was taken out The following grid 5-2 was exposed until the beginning of the dwell phase at 1133 K and replaced by 5-3 Due to the low137Cs revaporisation rate, compared to previous experiments, the phase at maximum temperature was prolonged to 120 min before cooling The sample reached a maximum temperature of 1145 K during that dwell phase Grid 5-3 was retrieved the next morning

2.3.6 Experiment 6: Phébus sample in steam atmosphere

For experiment 6 a second upper vertical line Phébus sample was used as in experiment 5, but this time an experiment with pure steam was performed, as can be seen inFig 4 The sample was heated in nitrogen at 10 K/min to 562 K followed by a 20 min dwell phase during which thefirst grid 6-1 was replaced by 6-2 and the atmosphere was switched to steam (60 g/h steam with nitrogen traces (0.5 l/min) as carrier gas) Thereafter, the temperature was raised by 2 K/min for 50 min followed by a 10 min dwell phase during which the TEM-grid was replaced This was pursued until the maximum temperature of 1090 K was reached and thefinal dwell phase of 60 min started The last TEM-grid 6–7 was retrieved shortly after the cooling phase started and afterwards the gas was switched to nitrogen

2.3.7 Experiment 7: Phébus sample in alternating steam and hydrogen atmosphere

For experiment 7 the Phébus upper vertical line sample showed a behaviour similar to that of the sample in experiment 6 and with previous experiments compared to experiment 5 Again alternating atmospheres were used during the experiment The previous heating sequence was used, as shown inFig 4 The short plateaus every 50 min were again used to exchange the TEM-grids, culminating in the use of 6 grids (7-1 to 7-6) The process gas was changed at the start of each dwell phase from 60 g/h steam (+0.5 l/min nitrogen) for 10 min to 2 l/min hydrogen

Fig 4 Temperature profile of furnace and sample and process gas as used in experiments 4 to 7 The sampling period of each TEM grid is represented by the dashed black horizontal lines The process gas is shown by the background colour as described at the top of the previous figure.

2.3.8 Experiment 8 & 9: Cs2MoO4 powder in air and hydrogen atmosphere

Calculations performed by (Kissane & Drosik, 2006) indicate Cs2MoO4as the most likely source for the vapour phases carrying caesium This led to the decision to test Cs2MoO4powder (Goodfellow, 99.9% purity, particle size: 75 μm) and collect the aerosols The experiments were performed with the same heating profile as in experiment 6 (Phébus - steam) & 7 (Phébus - alternating steam

& H2) Starting with inert nitrogen atmosphere the process gas was switched during thefirst dwell phase to air for experiment 8 and

to hydrogen for experiment 9 respectively To lower the number of TEM-grids that had to be examined only 3 grids were used in each experiment for 673–873 K, 873–1073 K and 1073–1273 K temperature regimes

2.4 Post-test analysis

Most of these examinations were performed using a Field Emission (Schottky type) Transmission Electron Microscope (200 kV JEOL JEM2100F, JEOL Ltd., Tokyo, Japan) equipped with liquid nitrogen cooled (Si(Li)) 30 mm2 energy dispersive X-ray spectrometer (Thermo Noran System 7, Madison, WI, USA) Afirst short examination was performed to get an overview of the grids and already revealed a significant number of aerosols deposited on their surface For experiments 7–9, grids were used that did not have any carbon coating and so did not permit the examination with a TEM Hence these samples were examined with a Field Emission Scanning Electron Microscope (Zeiss Sigma HD VP, Carl Zeiss NTS, Cambridge, UK) equipped with two 60 mm2(SDD) energy dispersive x-ray spectrometers (Thermo Noran) The digital micrographs were then analysed using the ImageJ software For micrographs showing a large number of particles the automatic particle analysis tool was used to calculate the Feret´s diameter (FD) which is the diameter of the smallest possible circle covering the whole projected surface area of the particle

3 Results

Representative micrographs will be shown of the main particle species on the grid along with size distribution plots EDS measurements were performed with TEM or SEM examination The proximity of the characteristic X-ray peaks of molybdenum, sulphur and lead can cause difficulties Lead can often be recognized by the higher energy L lines Previous analyses had shown fission product molybdenum is present in the deposits whereas sulphur was absent (Bottomley et al., 2014) Lead is an inevitable impurity from the large amount of lead shielding in the glove box Due to the relatively large number of samples that were examined with TEM & SEM only the micrographs of the Cs-containing aerosols are shown inFig 5and the according FD distribution inFig 6 Additionally, the elemental compositions and size characteristics of the aerosols that were found during TEM and SEM examination are shown inTable 1

3.1 Experiment 1 TMI-2 powder in air atmosphere

During experiment 1 all three grids were exposed in thefilter during the whole time and the population of nanoparticles was large enough to calculate the size distribution for each grid The micrographs of grid 1-1 revealed cubic crystals with about 100 nm edge length and a species of nanoparticles with a bimodal distribution centring at ~6 nm and ~21 nm Elemental analysis revealed the cubic crystals to be of sodium chloride but was not conclusive for the smaller particles, showing traces of silver and molybdenum The next grid, 1–2, showed no significant number of coarse particles but a high density of nanoparticles with a nuclei mode ~11 nm No conclusive EDS could be performed on those particles Grid 1–3 showed a particle density that was lower than 1–2 but considerably higher than 1-1 with a nuclei mode of ~6 nm Furthermore, SEM analysis showed uranium-zirconium particles with ~3 μm diameter and particles consisting of iron-based alloys in wide range of sizes and shapes This confirms the analyses in the initial investigation of the full molten core; it is surprising that they still show some volatility The variation in size (modes varying from 6 nm to 21 nm) shows that the aerosol capture varied with the exact positioning of the grid, namely the grids closer to the entrance of thefilter stage showed smaller particles than the ones behind (downstream) In this respect, the improved design withfixed position is expected to have reduced this variation As the caesium activity of the sample did not decrease during the experiment, it was not a surprise to observe any nanoparticles containing Cs

3.2 Experiment 2: TMI-2 powder in hydrogen atmosphere

Two out of three grids were retrieved successfully after the experiment Both show spherical nanoparticles with a size distribution mode at 24 nm for grid 2-1 and 10 nm for grid 2-2 Also bigger particles of varying size (100– 1000 nm) and consisting of steel or uranium-zirconium mixtures were found Thus, similar size ranges and compositions were found under the air and hydrogen atmospheres Even so, the activity measurements showed a loss of about 50% of the137Cs, although the elemental analysis was not able to identify nanoparticles containing any caesium

3.3 Experiment 3: Phébus sample in alternating air and hydrogen gas

Only the grids 3-2, 3-3 and 3–4 showed a sufficient number of nanoparticles to evaluate the size distribution Grid 3-1 (293–680 K under air) showed only a small number of particles with ~14 nm in diameter and some larger (200 nm) structures that were of low contrast and thus most likely to be of material with a low atomic number Grid 3-2 (exposed from 682 to 827 K under air then H2) showed a bimodal distribution of small spherical particles with a peak at 12 nm and a population of larger, elliptical particles with

K Knebel et al. Journal of Aerosol Science 106 (2017) 68–82

Fig 5 Micrographs with Cs-containing particles (left is 20 k and right is 200 k magnification) First row is grid 3 from experiment 4, second row grid 1 from experiment 5, third row grid 4 from experiment 6 and last row grid 5 from experiment 6 Note the many uniformly-sized particles at lower magnification that may be multiphase agglomerates at high magnification.

Feret´s diameter of 30–60 nm Grid 3-3 (828 to 1063 K under air) was densely populated with a species of very small particles with a peak in their size distribution at ~4 nm The next grid, 3–4 (used during final heat to 1165 K, dwelling and subsequent cooling to

923 K under H2), showed larger structures that appeared to be agglomerates of 3–5 nm sized spherical nuclei that are attached to each other by a second phase Due to their size, it was not possible to analyse the elemental composition of the small nuclei But due

to the fact that they were sampled during a period of high Cs release, it is very likely that they contain Cs The second phase attaching the small nuclei is represented by a brighter colour and thus consists of elements with a lower atomic number than the small nuclei The size distribution showed a broad spectrum of 6–16 nm Feret´s diameter peaking at ~10 nm Traces of molybdenum, silicon and lead were found in these agglomerates The last grid in this experiment was 3–5 (887 K to 293 K under N2) which showed relatively few particles on the micrograph, most notably were large structures in the size range of a few hundred nanometres EDS analysis indicated large amounts of silicon and some molybdenum in the particles of this grid

3.4 Experiment 4: Phébus sample in alternating steam and hydrogen atmosphere

During experiment 4,five TEM-grids were used, from which 4 size distributions could be derived Grid 4-1 (293–523 K under N2) had only a species of irregular shaped agglomerates (10–100 nm in diameter), consisting mainly of iron and some traces of phosphorus Grid 4-2 (608–750 K under steam) showed very similar large agglomerates consisting of iron, chromium, molybdenum and nickel These may be rust particles in the deposit or from corrosion of the sample holder Some of the agglomerates were silicon and silver-based The second type of spherical particles was considerably smaller The size distribution showed afirst mode at 5 nm, a second at around 30 nm and a third at 100–150 nm The next grid 4-3 (772–1029 K under H2then steam), showedfirstly a few large, cubic particles with an edge length of 100–150 nm that consisted of pure rhenium or rhenium with traces of manganese; secondly a dense layer of smaller spheroid particles with a broad size distribution centring at 32 nm Feret´s diameter (seeFig 5top row & Fig 6 top left) EDS showed traces of silicon and molybdenum in those particles There were also a few darker spherical particles of about

8 nm in diameter, embedded in the larger particles Under SEM two cubic caesium-containing particles were observed, very similar in

Fig 6 particle size (FD) Distribution of the micrographs for Cs-containing particles shown in Fig 5 Top left is from grid 3 of experiment 4 (mode: 32 nm), top right grid 1 from experiment 5 (modes 5 nm and 22 nm), bottom left grid 4 from experiment 6 (modes: 4 nm and 22 nm) and bottom right grid 5 from experiment 6 (modes:

6 nm and 52 nm).

K Knebel et al. Journal of Aerosol Science 106 (2017) 68–82

size and appearance to the rhenium particles observed also with TEM This coincides with the gamma measurements during the experiments that showed a loss of ~35% in137Cs activity during the exposure time of this grid The next grid, 4-4 (1052–1165 K under H2) showed a large number of agglomerates with an irregular shape and an even size distribution between ~4–20 nm and a semi distinctive peak at 15 nm can be observed Nearly all had lead as their main component sometimes withfluorine, sodium, tellurium and/or chlorine Several of these particle are attached to larger spherical objects (40–100 nm diameter) consisting mainly

of sodium andfluorine The last grid 4–5 (1044-293 K under N2) shows spherical nanoparticles of about 15 nm in diameter and a second species of ellipsoidal shape with ~35 nm diameter Only the later particles could be analysed with EDS These all had tellurium as their main component with some traces of molybdenum The traces of silicon andfluorine found in the analyses may result from the vacuum grease used to seal the sections of the revaporisation line Na and Cl are considered to be impurities; their origin is not certain

3.5 Experiment 5: Phébus sample in alternating air and hydrogen atmosphere

In experiment 5 only 3 TEM-grids were used to collect particles The reason for this was to provide a check on the data reliability Thefirst grid, 5-1 (450–994 K under air), was inside the experimental line for more than 5 h and thus the longest time of all For the micrographs seeFig 5second row from top and for the FD distribution seeFig 6top right It showed three particle types, one being spheroid with Feret´s diameter of 40–60 nm Higher magnification reveals those to be agglomerates consisting of small spherical nuclei of about 2 nm diameter EDS indicated rhenium and caesium as the main components with traces of potassium, chromium and chlorine A second type of spherical particles was observed A third type appeared to be condensed light elements on the spherical particles mentioned before In the size distribution the second type had a peak at 6 nm while the third species showed a peak at

20 nm Grid 5-2 (994–1133 K under air) showed a surface densely covered with spheroid- or ellipsoid-shaped particles with very low

Table 1

In this table the main results for the EDS analysis are shown for each grid in the experiments 1 to 7 The conditions during the sampling period are shown as well as the results of the elemental analysis and the modes of the ferrets diameter distribution The Temperature represents the actual sample temperature.

Exp No Sample origin Grid No Temperature Duration Process Elements observed FD modes

min [K] max [K] [min] gas a observed [nm]

1 TMI-powder 1 300 1171 500 air Na, Cl, Ag, 8/21

2 608 750 79 steam Fe, Cr, Mo, Ni, Si, Ag 5/30/125

3 772 1029 194 H 2 /steam Re, Cs, Mn, Mo, Si 32

4 1052 1165 138 H 2 Pb, Na, Cl, F, Te 15

5 Phébus FPT3 V15 1 450 994 308 air Cs, Re 6/20

2 562 639 60 steam Ag, Mo, Cl, F, Ca, Na, K –

3 639 723 60 steam Mo, Cl, K, Na –

4 842 932 60 H 2 /steam Cs, F, Cl, Mg, P –

a If a grid was exposed to two different process gases (N 2 , H 2 , air and steam) in an experiment, then it was consecutively.