Nuclear Atomic Data Needs for NDA Measurements Metrology

Nuclear & Atomic Data Needs for NDA Measurements & Metrology in Nuclear Safeguards Stephen Croft (ORNL) Ramkumar Venkataraman (ORNL) Louise G. Worrall (ORNL) Robert D. Mc. Elroy Jr. (ORNL) Andrea Favalli (LANL) ORNL is managed by UT-Battelle for the US Department of Energy

Acknowledgement This work was sponsored in part by the U. S. Department of Energy (DOE), National Nuclear Security Administration (NNSA), Office of Defense Nuclear Nonproliferation Research and Development (NA-22) and in part by ORNL and LANL. 2 Presentation_name

Introduction - 1 • International Nuclear Safeguards and Nuclear Material Accounting and Control (NMAC) rest on accurate physical inventory measurements. • A wide suite of non-destructive assay (NDA) techniques are employed to perform nuclear material inventory measurements, including: – Passive gamma spectrometry – Passive and active neutron correlation counting – Nuclear calorimetry – X-ray fluorescence techniques • Other methods such as photo-fission, NRTA, NRF are of course of interest too but not in routine use for SG today. • From implementation to analysis and interpretation, NDA techniques depend on nuclear & atomic data. – Design and optimization of measurement systems using forward predictive models – Characterization and calibration of instruments – Correction factors (to reference conditions), interference corrections – Inverse analyses to determine source term using measurement data 3 Presentation_name

Introduction - 2 • Nuclear and atomic data uncertainties are often the limiting factors in the overall uncertainties achievable using an NDA technique in the long run. • For accurate uncertainty quantification (UQ), it is important to consider covariances in nuclear data but it is typically not! • We often don’t use nuclear data as it is traditionally evaluated. • Minimizing systematic uncertainties due to nuclear and atomic data would improve the accuracies that can be achieved by the NDA instruments – Will in turn drive the revision of International Target Values (ITV) [see e. g. STR-368 (2010), ESARDA Bulletin 48 (2012)], resulting in better measurements. – The ITVs reflect the current state of practice, given the knowledge of the uncertainties (they are not a goal per se or what is possible). – We are obligated by IAEA agreements to use best metrological 4 Presentation_name practices

Status of nuclear data and their uncertainties – Fission Yields • Fission Yields: Accurate estimation of neutron absorbing fission products is vital. – Build-up of neutron absorbing fission products, reduces the net neutron population inside and escaping from the source, and therefore, the count rate measured by a NDA instrument. – ORIGEN estimations of fission products such as 133 Cs, 143 Nd, 149 Sm, 154 Eu are within a few % of experimental values. – Absorption cross sections of some of the fission products (155 Gd) have relatively large uncertainties (~5. 3%). – Calculated/Experimental ratios for 109 Ag, 106 Rh, and 125 Sb: 170%, 67%, and 100%, respectively. – Inconsistencies have been observed with respect to quoted uncertainties on legacy nuclear fission yield data on noble gas fission products; (e. g. ) 85 Kr • 244 Cm is the dominant source of spontaneous fission neutrons as well as delayed neutrons from spent fuel: the nuclear data uncertainties are relatively high (8%). 5 Presentation_name

Using FPY data in practice 6 Presentation_name

Status of nuclear data and their uncertainties – Actinide reaction cross sections • High-fidelity covariance matrices for evaluated ENDF/B-VII files are available for 3 major actinides, 235, 238 U and 239 Pu [P. Talou et al. , Nuclear Data Sheets 112 (2011) 3054– 3074] • Covariance matrix evaluations for all major reaction cross sections are available- total, capture, fission, elastic, total inelastic, and (n, xn). [P. Talou et al. , Nuclear Data Sheets 112 (2011) 3054– 3074. ] – Need: Angular distribution, uncertainties for discrete inelastic reaction cross sections • Quotable quotes [P. Talou et al. , Nuclear Data Sheets 112 (2011) 3054– 3074] – “In many cases, the evaluation of the nuclear data was performed prior to the quantification of uncertainties, thereby creating a somewhat inconsistent approach. ” – “While such detailed approach (for uncertainty quantification and covariance calculation) has been used for the standards evaluation, much less has been done for other reactions and isotopes. ” – “The unresolved resonance region represents an interesting challenge…Such work would lead to correlations between the resolved resonance range and the fast energy range, which are totally absent from the current covariance matrices. ” 7 Presentation_name

Status of nuclear data and their uncertainties – Actinide reaction cross sections • Fission cross sections: Important for source term definition and interpretation of the response of active and passive neutron NDA system measurements for safeguards (e. g. Active Well Coincidence Counter, Neutron Coincidence Collar). • Neutron-induced fission cross section of 235 U was evaluated by the IAEA Standards Group [A. D. Carlson et al. , Nuclear Data Sheets, 110, 3215 (2009)], – ENDF/B-VII. 0 evaluation incorporates their findings without modification, including the associated covariance matrix for this reaction – UQ for the neutron-induced fission cross-section of 235 U is of major importance as most other actinide fission cross-section uncertainties are driven by it. • Evaluation by the IAEA Standards Group is the result of major efforts from experts in the domain. – Yet, unrecognized correlations between experiments can be expected to lead to an underestimation of the final uncertainties. 8 Presentation_name

Prompt neutron nu-bar value for 252 Cf spontaneous fission – 1: Review • 252 Cf neutron sources are used to characterize and calibrate instruments used to assay Pu. Ultimate accuracy of such techniques depends on how well the neutron multiplicity distribution of 252 Cf is known. • In the SORMA 2018 conference, Croft et al. presented a review (adj & eval) of 14 high quality absolute determinations of 252 Cf nu-bar [1 -14] and subjected them to a full covariant evaluation (in press in NIM A). • Croft et al. , observe that the youngest measurement of the above 14 is ~30 years old. “New measurements of the highest quality are overdue. ” • Croft et. al. (SORMA 2018) suggest an alternative approach to use a high density polyethylene moderated array of high pressure 3 He proportional counters (vs. liquid scintillators). See also Henzlova et al. , Metrologia (2019) ABCD. 9 Presentation_name

252 Cf source aging • Gamma spectra of 252 Cf can determine source age using the ratio of the intensity of gamma rays from 137 Cs to short-lived fission products • Short-lived fission product gamma rays were chosen for having no interfering gamma rays, branching ratios close to unity, and large spontaneous fission product yields (FPY) • However, some FPY have large uncertainties greater than 20% • Source age is meaningless if uncertainty, driven by available nuclear data, is large as demonstrated by 132 I and 136 I Isotope Energy[ke. V] 1 Half-life 1 Branching Ratio 1 Fission Product Yield (FPY)1 % Uncertainty FPY Source 3 Age [y] 137 Cs 661. 657(3) 30. 08(9) y 0. 851(2) 0. 0502(20) 4. 0% N/A 142 La 641. 285(9) 91. 1(5) m 0. 474(5) 0. 0602(24) 4. 0% 29. 9± 0. 3 132 I 667. 714(2) 2. 295(13) h 0. 987 0. 0215(138) 64% 28. 6± 3. 2 136 I 1313. 02(10) 46. 9(10) s 1. 00 0. 0228(52) 23% 29. 2± 1. 2 138 Cs 1435. 77(7) 32. 5(2) m 0. 763(5) 0. 0547(15) 2. 7% 30. 8± 0. 3 140 La 1596. 21(4) 1. 67855(12) d 0. 9540(8) 0. 0596(8) 1 1. 3% 30. 9± 0. 3 National Nuclear Data Center, Brookhaven National Laboratory.

Beta-Delayed neutron yields • The method of detecting delayed neutron signature from fissile materials is a cross cutting NDA technique with applications in nuclear forensics, non-proliferation, and safeguards (from components to waste). • Relative delayed neutron group abundances are among the poorest known nuclear data and have large associated uncertainties. – Kelley et al. [“Uncertainty analysis of delayed neutron fissile material assay using a genetic algorithm”, Annals of Nuclear Energy 80 (2015) 460– 466] carried out an analysis of the uncertainties associated with the delayed neutron NDA technique using a genetic algorithm. – The overall uncertainties of the mass estimates were 15%, and 30% for 235 U, 238 U, and 239 Pu, respectively. Reducing the first delayed neutron group abundances as proposed by a factor of three reduced the overall uncertainties to 10%, 3%, and 20%, respectively. • Delayed neutron yield of 252 Cf is typically based on 1950’s data and is long over due redetermination given that 252 Cf is the de facto fission standard source and uncertainty in the delayed neutron yield impact the nu-bar value. 11 Presentation_name

252 Cf Shuffler for 235 U and Waste Applications • • 239 Pu assay – Nuclear Safeguards and Passive/Active Neutron Interrogation System – Passive measurement of the even isotopes of plutonium – Active Neutron Interrogation for 235 U and 239 Pu Active Technique – Large (~109 n/s) 252 Cf neutron source is used to induce fission within the sample – 235 U and 239 Pu mass is determined by counting delayed neutrons • The shuffler is the only accurate method for high density, high mass 235 U assay. • The same system can be used for safeguards and waste applications – Mass Range: to 3000 g U-235 – MDA: 0. 06 to 0. 40 g U-235 Now let’s peak into a possible future 12 Presentation_name

Laser-driven Neutron Source for Research and Global Security • High intensity short-pulse laser produces highintensity short-duration neutron-beam bursts • Laser irradiates deuterated sub-µm plastic foils to drive intense deuterium beam • Neutrons produced in beryllium converter by D-Be interactions: • Forward-peaked neutron beam demonstrated at LANL short-pulse laser facility: ~ 2 x 1010 neutron/sr in ~few ns • Applications • Active Neutron Interrogation of Special Nuclear Material • Treaty verification (warhead & nuclear material signatures) and stockpile stewardship • Cross sections • Production of 9 Li (beta-delayed neutron emitter , 257 ms decay constant) from the nuclear reactions: 9 Be(d, 2 p)9 Li, threshold 18. 42 -Me. V (*) and 9 Be(n, p)9 Li , threshold 14. 26 -Me. V, in the Be converter. • These delayed neutrons form the basis of non-intrusive diagnostics for determining the features of deuteron acceleration as well as monitoring neutron production for laser-driven neutron sources. -> need the experimental reaction cross section as a function of energy from threshold to around 100 Me. V • *W. L. Gardner, N. Knable, B. J. Moyer, Li 9 -New delayed neutron emitter, Phys. Rev. 83, 1054 (1951). • Spent fuel assay (e. g. Fukushima debris) • Nuclear physics experiments (e. g. neutron resonance spectroscopy) 13 Presentation_name

Prompt neutron nu-bar value for spontaneous fission – 2: DN 252 Cf • If we are to rely on the 252 Cf SF system as a standard we additionally need a new delayed neutron assessment • The best experimental delayed neutron yield (0. 0086± 0. 0010) remains that of Cox et al PR 112(3)(1958)960 and Smith et al 2 nd UN Intern Conf on the Peaceful Uses of At En 15/P/960(1958). We should be able to do better including E-dependence of efficiency (BF 3 -array in this case calibrated to fission energy n’s), and access to the short lived 0. 25 sec group. • The best dn spectra by Chulick et al NP A 168(1971)250 -258 is also rather limited (and the unfolding should be checked). Needed in our ABCD (Absolute Cf-252 Determination) method (rel eff to PFNS) • New comparison to calculations based on dn precursor yields? • Comparison to empirical correlations (e. g. Tuttle)? Because we are not going to have a first principles account of fission in my lifetime 14 Presentation_name

Direct relative worth of SF isotopes • We typically work in terms of Pu-240 -eff or U-238 -eff masses • Absolute error in e. g. (λ/A). <ν(ν-1)> are huge compared to the SG need • We can side step by using ‘representative standards’ • BUT we want to create a metrology train to Cf-252 (recall ABCD – Absolute Cf-252 Determination) – We have evaluated Cf-252 nu-bar, Diven parameters in progress (to replace current unweighted mean of ‘good’ measurements) • Fix U-238 (including CR B/G), need known masses Cf-250, Pu-238, 240, 242, Cm-242, 244 and we eliminate the half-life unc. 15 Presentation_name

( , n) reaction data for safeguards science • Light (low atomic number, Z) elements have relatively low Coulomb barriers and so, when energetically allowed, can often be a significant source of neutrons throughout the nuclear fuel cycle, via (α, n) reactions when they are in intimate contact with α-emitters including special nuclear materials of the actinide family. • The nuclear data need is summarized as follows: – thick target (α-particles stop in the material) integrated-over-angle (TT IOA) (total yield into 4π, not the angular distribution) – yield (n. α-1) functions (vs. α-energy) for the elements/compounds of Li to K – over the energy range (threshold or from about) 3 Me. V to 10 Me. V with reasonable energy resolution (50 ke. V spacing or better over the interval 4 -6 Me. V) – with a total measurement uncertainty of about 2 -3% (which should be a realistically, almost readily, achievable goal, for carefully performed modern methods), – together with a measured data base of associated neutron emission spectra (40 energy groups would be sufficient in most applications) emerging into 4π. 16 Presentation_name

Why do we need ( , n) source-term data? • Bulk measurements on bulk Pu and U oxides where (α, n) production is often calculated from the isotopic composition when coincidence counting is applied • Bulk measurements of uranium fluorides, including UF 6 in storage cylinders and UO 2 F 2: n. H 2 O holdup in process equipment, via gross neutron counting, and which is usually dominated by the 234 U contribution • Am. O 2 -Li. OH source term representation in active neutron interrogation assay systems • For safety, security and safeguards priority elements of immediate concern are F (yield and energy), Li(energy) and O(energy). • For advanced fuel cycles (e. g. MSR) Na, P, S, Cl and K data is also needed and particularly scant. • Our review paper [Venkataraman et al. INMM 2018] provides a comprehensive list of applications that require high quality ( , n) nuclear data. Croft et al [NIM A In Press] report on a new determination of the n/s/g 234 U in UF 6 at sub 2 % total rsd. At INMM 2018 we also reported how both thick and thin target data need large adjustments. 17 Presentation_name

Comparison with thin target data using SRIM-2013 stopping powers (ref: INMM-2018) ‘Blended’ microscopic cross-section: 592 points threshold to 9. 92 Me. V Peters et al x 1. 00* 3. 92 – 6. 67 Me. V Balakrishnan et al x 1. 34 3. 1043 – 3. 9122 Me. V Wrean and Kavanagh x 2. 68 2. 3635 – 3. 10054 Me. V Norman et al x 1. 14 6. 67 – 9. 92 Me. V 18 *Note the non-unity scaling factors! Peter’s et al’s data is not fully mined. • To match our UF 6 data (508 n/s/g 234 U) need to scale by x 0. 84 • BUT caution U alpha spectrum taxes only a very limited range. Presentation_name

Summary of (α, n) thick target yield data available • Best individual thick target measurements are those of West and Sherwood. Quality of new measurements must aspire to do even better than this. – D. West and A. C. Sherwood, Harwell, report AERE-R-10502(1982). See also Annals of Nuclear Energy 9(1982)551 -557. Target Li Be B C Yield 5 -20% Good, 2% Spectrum Good for Am. O 2/Li. OH Good Modest N O F Ne Na Mg Al Si P S Cl Ar K 25% Good, 2% Poor, 30% N/A ~10% 2% Good, 2% Poor N/A Poor ? Modest 19 Presentation_name ? ? ? ? ? Comment Important for active interrogation ISO Am. Be needs updating Am. O 2/B used for characterization 238 Pu. O 13 C hard spectrum; 6. 13 Me. V gamma 2 Forbidden below about 6 Me. V Extend UO 2 data; Am. O 2 UF 6 and Am compound benchmarks Croft & Mc. Elroy, rough estimate

Types of ( , n) measurements needed • Sealed source measurements – Convenient, transferable for inter-comparison exercises, calibrate different kinds of instruments, QC, validate spectroscopy and yield measurements methods which might be developed. – Can be certified for absolute neutron emission using the Mn. SO 4 -bath technique and thereby provided an important link to national standards. – In formulating an (α, n) yield and spectrum measurement program, sealed sources can play an essential role as well as providing important benchmark/integral/normalization data. – In the realm of nuclear safeguards Am 2 O 3, Am. O 2 and Am. F 3 (or Pu. F 3) are of especial immediate interest to provide fiducial reference points (or benchmarks) for both yield and energy spectra for a realistic α-spectrum. • Accelerator measurements – TT IOA yield measurements in steady state using flat a well characterized energy 4π are needed for the elements mentioned. – A variety of target materials should be measured to check consistency and scaling rules. Using the same instrumentation is recommended to avoid bias. Target degradation under bombardment should be included as part of the experimental evaluation. Materials used previously, e. g. Ca. F 2, La. F 3, Pb. F 2 need to be re-measured. – Energy range of interest: in the vicinity of 4. 75 Me. V for U materials, 5. 5 Me. V for Pu. A broader energy range, say 3 – 10 Me. V is important to normalize thin target data. 20 Presentation_name

Types of ( , n) measurements needed • Neutron measurements should always include measurements on another ‘trusted’ material which can be used as a transfer standard; e. g. ceramics BN and Be. O, and UO 2, which is both well-known and extremely important to the safeguards community. – It is therefore very important to build the metrological link between F- and O- alpha-n yields. • Pulsed accelerator facilities allow spectra to be determined at various directions with respect to the incident beam by neutron time-of-flight. – Accuracy of yield not as good as moderated 4 p neutron counters; – Better energy resolution; possibility of teasing out different neutron energy groups and their branching ratios. • Calibration of TOF equipment, and spectrometers (like organic liquid scintillators) for the sealed source measurements could be facilitated by having a modest 252 Cf source embedded in a fission chamber. – 21 252 Cf is the de facto prompt fission source spectrum and the FC provide a fission event trigger for n-TOF. Important to use complementary methods. Simple small scale things can be a great added value e. g. sealed sources are Presentation_name always available

Updates to SOURCES 4 C Code • The SOURCES 4 C code is a commonly used tool, both domestically and overseas, for calculating the (α, n) source-term, yields and spectra, in safeguards. A comprehensive update is long overdue. – Extend the alpha-energy range. – Refine the energy treatment. – Replace the mass stopping power data. – Ensure recommended data (rather than the first in the list) (α, n) production data gets used. – Provide experimental emission spectra data when available for comparison. – Replace the GNASH statistical model for partitioning the total cross section into differential partial cross sections (EMPIRE or TALYS suggest themselves), or where possible use experimental data. – Update angular distribution treatment to replace assumption of isotropy in CM – Give the user some much needed guidance on the reliability and accuracy of the reported results. – Create applications relevant benchmark example problems. – Modernize the user interface. – Provide a utility to import cross-sections from evaluated data files as they become available. – Make use of covariance information as it become incorporated onto libraries. 22 Presentation_name

XRF Spectrum for MOX Sample • MOX Solution • 5 elements • Typically each has 10 assigned peaks (some “peaks” represent multiple transitions. • Branching ratios and peak energies appear to be dependent on actinide concentration • Fitted Peak Positions sensitive to isotopic distribution (differences on the order of a few e. V). 23

Branching Ratios Histograms of ωK(Pu) and ωK(U) • From the collection of fits we can predict ωK(Pu) and ωK(U) and attach an error – ωK(U) = (9. 73 ± 0. 004)✕ 10 -1 – ωK(Pu) = (9. 74 ± 0. 004)✕ 10 -1 • But for HKED applications the important point is we can also take ratios for each sampled fit and take the average of the ratios Histogram of ωK(Pu)/ωK(U) – ωK(Pu)/ωK(U) = 1. 0009 ± 0. 0001 A Nicholson, S Croft, and RD Mc. Elroy K-shell fluorescent yields and their uncertainties for use in hybrid K-edge densitometry The Journal of Radioanalytical and Nuclear Chemistry 307(3)(2016)2069 -2074. 24 Presentation_name We call it bootstrap but same as total Monte Carlo

Mass Attenuation Coefficient from KED assay of DU • Voigt Broadened XCOM data (Scofield calculation!) vs. Observed Attenuation Coefficient • The plot shows a comparison of the inferred mass attenuation coefficients determined from the ratio of 2 KED transmission spectra. • What is the fine structure near the upper edge as the electrons just emerge and probe the electronic environment? Voigt smoothing does not fit well! Matches with simple Gaussian smoothing Voigt smoothing fits well 25 Presentation_name

K-absorption Edges Shape & Fine Structure • Fine structure about the U and Pu L-Edges is well known and much data exists. • Fine structure data is available for low to intermediate elements but little data exists for fine structure about the High-Z KEdges • Expect similar structures at the U K-edge which may explain the apparent lack of Lorentzian broadening. 26 Presentation_name Allen et al EXAFS: A new approach to the structure of uranium oxides J Phys Chem 89(1989)13341336

Pu and U/Pu K-Edge Fits • The Pu K-edge also fits well to the half and half Voigt and Gaussian shape. • 27 Presentation_name NOTE: The reference atomic mass in XCOM is based on longest half-life for Pu its m = 244. 06 • However, there is indication that the Lorentzian width narrows for UPu mixtures.

Dream • Progress from Th, to DU to Pu-239 • Solutions, ρ. t <± 0. 05%, homogeneous! (vs foils) • E. g. for U (115. 6 ke. V) cover 114. 8 to 116. 4 ke. V with 150 points with a 10 e. V beam (CW beam/Ge-det to remove other orders) • TMU < 0. 1% • Explore temperature and concentration and mixed solutions 28 Presentation_name

Why investigate 233 U nuclear data? • Quantitative assay of 233 U – Safeguards technology tools underpinned by nuclear data – Thorium fuel cycles • Large uncertainties in 233 U nuclear data Objectives • Explore whether the uncertainties in 233 U gamma-ray absolute emission probabilities can be reduced – Understand the limitations – Understand signitures 29 Presentation_name

Nuclear Data Uncertainties for 233 U • Evaluated Nuclear Structure Data File (ENSDF) Can we put 5 9’s pure U-233 is a gamma-sphere Can we mix with U-235 as an internal standard 30 Presentation_name

Role of microcalorimetry • Photon or spin information carriers offer far superior resolution than HPGe – Gamma (eg for Am-241 or Np-237 in Pu; or to resolve U-238 (Th 234) and U-235 (Th x-ray) in enrichment determination – Total Energy (eg resolved overlapping alpha-lines such as Cm-244 in Pu) … needs incorporated samples • I would be really interested to see if they could improve U 233 decay scheme knowledge • If they can improve U and Pu x-ray branching ratio data – which is a far larger topic of course! • If we can use them with attenuated Bremsstrahlung to explore the K-absorption edge 31 Presentation_name

Conclusions • Knowledge of nuclear and atomic data can become the limiting factors in design and calibration of NDA systems and physics-based modeling of responses from NDA systems used in safeguards applications. • Among the nuclear data that are known very poorly are the neutron yields from ( , n) reaction on low Z nuclides. • Uncertainty quantification, taking into account covariances, is needed for cross-section (fission and other reactions) data in the evaluated nuclear data libraries. • Relative abundances of delayed neutron groups available in the literature have large uncertainties. • Branching ratios of gamma-rays emitted by uranium, plutonium, and other actinide isotopes are needed with greater accuracies so that the uncertainties in the isotopic analyses can be driven down. • Atomic data such as interaction cross sections and X-ray yield data have large uncertainties. These limit the accuracy of U and Pu elemental concentration results that are of importance to nuclear safeguards. 32 Presentation_name

Thank you for your interest in atomic & nuclear data What questions or comments do you have? Please make the breakout lively Nuclear data is a cross cutting challenge problem (Re)Develop and maintain measurement capability Stretch technology Supports basic and applied science and multiple users Safeguards needs specific data in specific forms for direct use All data sets a nuances and UQ requires expert judgement 33 Presentation_name

Status of nuclear data and their uncertainties – Actinide reaction cross sections • Using a time-projection chamber (TPC) for sub-percent fission cross -section measurements – represents a very different approach than what has been done in the past. Essential to identify and understand systematic error • • TPC measurements reduce uncertainty components – Particle ID, Beam & Target uniformity, uncertainty in reference cross section (235 U) • • TPC measurements would be mostly uncorrelated to past data sets and would represent a strong test for the IAEA evaluation. 34 Presentation_name Actinide target is placed on the central cathode and irradiated with a neutron beam that passes axially through the target, inducing fissions. 4 p acceptance for fission fragments (~100% efficiency) • Complete charged particle track reconstruction