A comparison of oxide decay heat simulations and nuclear data libraries with fusion irradiation experiments

We compare existing experimental decay heat data sets measured at the JAEA fusion neutron source (FNS), which employed a fast extraction rabbit system that in certain cases allowed the measurements to capture, at short cooling times, the decay profile of N in a range of oxides. Focussing on those data points and timescales that can be attributed to N, we compare measurements to simulations performed using the FISPACT-II inventory code together with evaluated nuclear data libraries. Making small corrections for other contributions at these short times, we derive integral cross section data estimates for O(n,p)N from 12 oxide sample measurements and compare with previously obtained measurements in the IAEA EXFOR database and evaluations in the nuclear data libraries themselves.


INTRODUCTION
Cooling water in the primary circuit in both fission and fusion nuclear reactors is unavoidably exposed to neutrons leading to the generation of problematic short-lived isotopes, most importantly, 16 N (T 1/2 =7.3s) via the 16 O(n,p) 16 N reaction. This presents a significant itinerant radiation source to consider for operational dose and impact on radiation sensitive equipment, such as electronics and cryogenic components such as superconducting magnets, particularly due to the intense 6.13 MeV gamma ray that is emitted during 16 N decay. This issue is well-known from water-cooled fission reactors [1], though in fusion devices such as ITER, where water coolant will be used extensively in first wall and other components throughout the device, the intensity of the phenomena is expected to be more severe. The reasons for this is partly because the neutron energy threshold of the 16 O(n,p) 16 N reaction (the primary pathway to the production of 16 N) is above 10 MeV, and D-T fusion neutrons are born around 14 MeV energy. ITER is designed to operate at 500 MW fusion power, where the 14 MeV neutron emission rates from the deuterium-tritium plasma will be 1.77 × 10 20 n s −1 and the neutron fluence experienced by some first wall water-cooled components will exceed 10 14 n cm −2 s −1 . Presently due to uncertainties in modelling and nuclear cross section data large safety factors, between 8.2 and 4.7, are applied within the ITER project. Activities are ongoing to seek experimental justification to reduce these factors via coolant loop experiments at the Frascati Neutron Generator [2][3][4], which has incorporated an ITER first wall mock up component incorporating water coolant channels, irradiated with 14 MeV neutrons and the 16 N emissions measured via a high efficiency CsI detector at different water flow rates.
To complement to the body of work on this important phenomena, in this paper we provide new insight from experimental decay heat data sets measured at the JAEA 14 MeV fusion neutron source (FNS) [5][6][7] with irradiated samples where the 16 O(n,p) 16 N reaction contributes to decay heat. The FNS experimental campaigns employed a fast extraction rabbit system that in certain cases allowed the measurements to capture, at short cooling times, the decay profile of 16 N in a range of oxides: Focussing on those data points and timescales that can be attributed to 16 N, we have compared measurements to simulations performed using the FISPACT-II [8,9] inventory code together with the TENDL-2017 nuclear data library [10]. Extensive decay heat analysis and comparison with the FISPACT-II inventory code using various nuclear data libraries has been performed in [11]. However, here we provide analyses to derive integral cross section data estimates for 16 O(n,p) 16 N from 12 oxide sample measurements and compare with previously obtained measurements in the IAEA EXFOR database and evaluations in the nuclear data libraries themselves.

Description of FNS experiments
14 MeV neutrons were generated at the JAEA FNS facility via a 2 mA deuteron beam incident on tritium loaded titanium target. A series of experiments took place between 1996-2000 where thin samples, nominally 25 × 25 mm 2 in area, and approximately 10 µm thick were irradiated for 5 minutes in a neutron flux typically in range 10 9 -10 10 n cm −2 s −1 . Figure 1 (right-hand y-scale) shows a calculated neutron spectrum at the FNS at the irradiation position. Full details of the earlier of these experimental campaigns are provided in [5][6][7] with details from the 2000 experiments analysed in [11]. A sample irradiation 'stack' comprising the irradiation sample together with an Al monitor foil were placed into a plastic 'rabbit' capsule for each irradiation. The decay heat from the irradiated sample post-irradiation was measured using a Whole Energy Absorption Spectrometer (WEAS)-a system based on two large bismuth-germanate BGO (Bi 4 Ge 3 O 12 ) scintillators, associated photo-multiplier tubes, high voltage and pulse counting electronics, and a multi-channel analyser. Decay heat from both β − and γ emissions were measured from the irradiated samples,  typically 30-50 s following the end of the irradiation. Important to the design, the BGO scintillator does not suffer with hygroscopic degradation (as is the case with NaI scintillators) and hence encapsulation or 'window' materials are not needed. The lack of 'window' materials means that energy loss, particularly from β − emissions, are minimal. Thin samples were centrally located between the detectors in a tight 'sandwich' arrangement (gap width was approximately 1 mm), with close to 4π sr counting geometry, giving almost 100 % efficiency. The samples were measured for several minutes, dumping data into periodic time bins (typically 16 seconds) which allowed the decay heat response to be monitored. Some samples were in powder form (CaO, for example) and in these cases the powder was spread uniformly and encapsulated in plastic adhesive tape. Blank tape samples were irradiated to subtract background decay heat. The irradiated Al monitor foil 24 Na activity was separately measured following each irradiation using a high-purity germanium detector; the 27 Al(n,α) 24 Na reaction was used to determine the total D-T neutron fluence.

Analyses methodology
We have focussed on a subset of the data obtained in the year 2000 experimental campaign-a set of 12 oxide materials; Figure 2 (see the bottom plot) shows an example of decay heat measurements taken over the first three minutes of cooling for the Tm 2 O 3 sample measurement. We have com-pared the data with a calculation of the equivalent experimental scenario using the FISPACT-II code and the TENDL-2017 nuclear data library. In general we have concentrated on using experimental data points and timescales that can largely be attributed to 16 N. In this case for Tm 2 O 3 , and similarly in other cases analysed in this paper, the first data point that is highlighted at 0.6 minutes cooling (the shortest measurement time following irradiation). Since there are other contributions to decay heat from other nuclides (usually small in the cases studied here), other than the contribution from 16 N emissions, we have estimated these contributions at the relevant cooling time using the FISPACT-II code together with the TENDL-2017 nuclear data library so that corrections can be made in our later analyses to estimate 16 O(n,p) 16 N cross sections. In the example case for Tm 2 O 3 the contribution to decay heat from 16 N at 0.6 minutes is predicted to be 99.4 %, with the remaining, rather small contribution, mostly coming from 168 Tm. The various contributions at different cooling times post-irradiation are illustrated in this case by the top plot shown in figure 2, which indicates that at longer cooling times, >1.5 minutes, 168 Tm dominates, with minor contributions from 166 Ho and 170 Tm. In another example, at the other end of the scale in these studied cases, for Eu 2 O 3 (see figure 3), the contribution to decay heat from 16 N at 0.85 minutes cooling is much larger, 51.2 % with the remainder from 152n Eu, 152m Eu and 150m Eu contributions. These other contributions can be partly explained by the later first measurement point at 0.85 minutes, compared to 0.6 minutes in all other cases studies in this paper. The time difference is about 15 seconds, which exceeds two half lives decay of 16 N. In general, throughout our analyses we have accounted for these corrections in an expanded uncertainty estimate, which accounts for both the measurement uncertainty and the decay heat correction from other isotopic contributions-here added in quadrature. The uncertainty estimate is likely to be an overestimate, since it implies no knowledge of cross sections from other reactions; we have effectively treated these correction with a 100 % uncertainty estimate.
To enable the measured (and corrected) decay heat to be converted into activity we have used the Q value for 16 N decay, which is 10438.869 ±58.844 per disintegration [12], with 7263.658 ±52.302 of this attributed to the absorbed quantity (the total calculated energy minus the neutrino energy). The absorbed value has been used to convert the measured total decay heat, corrected for 16 N contribution, to activity in Bq. A further correction factor of 0.9588 [7] has been applied to correct for the BGO decay heat measurement efficiency for 16 N emissions. Activity estimates for 16 N have then been decay corrected to the end of the irradiation period so that cross section estimates can be made via the following where A ∞ is the saturated activity of 16 N, φ is the neutron flux and N16 O is the number of 16 N atoms cm −3 within the sample. The irradiation time, 5 minutes, is sufficiently long compared to the half-life of 16 N (7.13 s) that the saturated activity assumption can be made.

Discussion and Conclusions
In this paper we have used experimental decay heat data sets measured at the JAEA fusion neutron source (FNS) to derive 16 O(n,p) 16 N cross section estimates from the decay profile of 16 N in a range of oxides. Focussing on those data points and timescales that can largely be attributed to 16 N, we have compared measurements to simulations performed using the FISPACT-II inventory code together with evaluated nuclear data libraries. We have made small corrections for contributions to decay heat arising from contributions other than 16 N.
The spread in cross section results can be seen in Figure 4. In particular the estimates from the Cs 2 O 3 , Lu 2 O 3 , Tb 4 O 7 , and Yb 2 O 3 measurements are higher than the other results presented. However, the measurement uncertainty associated with these samples were comparatively high, between 15 and 24 %. The overall weighted average integral cross section data estimate across the full 12 sample set was 33.1 ±17.2 mb. This compares with a spectrum averaged value of 33.6 mb using the TENDL-2017 library and our result is therefore in good agreement. Given the narrow peak neutron energy of the FNS irradiation around 14.7 MeV one may also compare the weighted average result with differential experimental data shown in Figure 1 at 14