Gamma spectrometry in the ITWG CMX-4 exercise.

Low enriched uranium samples of unknown origin were analyzed by 16 laboratories in the context of a Collaborative Materials Exercise (CMX), organized by the Nuclear Forensics International Technical Working Group (ITWG). The purpose was to compare and prioritize nuclear forensic methods and techniques, and to evaluate attribution capabilities among participants. This paper gives a snapshot of the gamma spectrometric capabilities of the participating laboratories and summarizes the results achieved by gamma spectrometry.

Introduction

This paper presents the state of practice in gamma spectrometry for nuclear forensics exercises. The Nuclear Forensics International Technical Working Group (ITWG) organized its fourth interlaboratory exercise in 2014, called Collaborative Materials Exercise (CMX-4) [1]. This paper documents the collective experience with gamma spectrometry during the CMX-4 exercise and it gives a snapshot of the applied approaches.

Nuclear forensics is the analysis of intercepted nuclear or other radioactive material to provide evidence for nuclear attribution in a legal context. The goal of the analysis is to identify the composition, origin, and intended use of interdicted nuclear or radiological samples, containers, and transport vehicles. Nuclear forensic analysis includes the characterization of the material and correlation with its production history [2]. The CMX-4 represents the second paired-comparison exercise organized to improve international cooperation and communication in case of a nuclear security event.

Three oxide samples of low enriched uranium (LEU) were selected as the materials to be characterized during the CMX-4 exercise. A scenario was included in which a seizure of nuclear material occurred and forensic analysis was requested. Laboratories were instructed to submit assessment reports in a 24 h, 1 week, and 2 month time frame. Participating laboratories categorized and characterized the exercise materials, and performed nuclear forensic evaluations. Each of the 16 participating laboratories was assigned a code name by the organizers to ensure anonymity and confidentiality of data.

Among the methods used in nuclear forensics, high-resolution gamma spectrometry (HRGS) is a relatively rapid nondestructive analytical technique. Advantages include preservation of evidence and no, or a minimal, need for sample preparation. A disadvantage is that it suffers from higher uncertainty compared to destructive techniques, such as mass spectrometry (MS) [3]. This paper presents the isotopic composition, age, and signatures of the neutron irradiation history of the three LEU samples determined by gamma spectrometry. These values are compared to the community average values determined by mass spectrometry.

Sample description

Participants were provided three samples of similar size and mass: ES-1, ES-2, and ES-3. The ES-1 sample consisted of a physically homogenous, fine black U-oxide powder. ES-2 and ES-3 samples were dark gray, homogeneous, UO2 pellets with smooth surfaces. Representative physical sample properties are shown in Table 1. Pellets are made of UO2, whereas ES-1 was a mixture of UO2 and higher U-oxides. Further details on the exercise samples are provided in the introductory article of this Special Section [1].

Table 1

Average physical properties of the samples used in the exercise

Sample ID	Physical form	Mass, g± 0.1	Size, mm ± 0.1		Approximate isotopic abundance, mass%
			Diam.	Height	234U	235U	238U
ES-1	Powder	2.9			0.025	2.9	97.1
ES-2	Pellet	2.4	9.2	3.7	0.018	2.19	97.81
ES-3	Pellet	2.5	9.1	3.7	0.025	2.9	97.1

Determining major U isotopic (234U, 235U, 238U) abundances

The samples were assayed first using HRGS for the 24 h and 1 week reports. Spectra were generally acquired by high purity germanium (HPGe) detectors for about 30–60 min in the case of the 24 h reports, whereas much longer measurement times were used for the 1 week reports.

Most participants of the exercise determined the major U isotopic (234U, 235U, and 238U) abundances using computer codes [4, 5] for automatic spectrum evaluation. These codes are based on so-called relative efficiency calibration (or intrinsic calibration) method [6]. The relative efficiency curve is obtained from the spectrum of the measured sample, thus the attenuation both in the sample (self-attenuation) and in absorbers (shielding) are taken into account. Therefore, the method does not depend on the sample size, geometry, physical, and chemical state. As all the information for determining the isotope ratios is present in the spectrum of the sample, no reference materials are required for calibration. However, for quality control purposes and for demonstrating laboratory performance, it is recommended to use a set reference materials.

Manual evaluation also occurred, after primary processing (measurement control, peak shape fitting, determining peak area, and deconvolution of overlapping peaks) by codes (FitzPeaks, PeakEasy, and Gammavision), followed by application of intrinsic self-calibration. For example, peaks of 214Bi (for age dating) and descendants of 232U were evaluated manually.

Table 2 summarizes the gamma spectrometers and software used for determining the isotopic composition of the samples in the CMX-4 exercise. Only participants sharing their results for this paper are listed. High-efficiency coaxial HPGe detectors used for age dating and identifying reprocessed uranium are not included here.

Table 2

HPGe detectors and software used by participants to determine the abundance of the major U isotopes

Lab code	Detector	Software
Michelangelo	ORTEC Micro-Detective-HX, coaxial, diam. 50 mm, height 30 mm, electrically cooled,	IdentifyFRAM 5.1, (In-house param. set: V_CX120-1010keV_microdetective)
Van Gogh	Canberra Falcon5000, broad-energy, diam. 61.8 mm, height 31.70 mm, electrically cooled	Manual, intrinsic efficiency calibration
Monet	Ortec GMX40P, coaxial, diam 63.0 mm, height 63.8 mm, electrically cooled	FRAM 5.2, (Param. set: ULEU-coax120-1010)
Rembrandt	Canberra GL0510, planar, diam. 24.8 mm, height 10 mm, active area 500 mm2	MGAU V.3.2
Renoir	Canberra GL0515R, planar diam. 25.2 mm, height 15 mm, active area 500 mm2	MGAU V.4.2
Picasso	For 24 h report: Canberra GL2020R, planar, diam. 50.5 mm, height 20 mmFor 1 week report: ORTEC GLP10180/07P4, planar, diam. 10 mm, height 7 mm	MGAU V.4.2
Buonarroti	Canberra GL0210R, planar,diam. 16 mm, height 10 mm, active area 200 mm2	MGAU V.4.2
Pollock	Canberra BE3820, broad-energy Ge, diam. 70 mm, height 20 mm	U235 v1.51 (MGA ++)
Gauguin	Ortec, coaxial (averages from different detectors were used)GEM30P4-70: diam. 54.8 mm, height 51.0 mmGEM-20180-S: diam. 51.0 mm, height 50.7 mmGEM-10195: diam. 42.7 mm, height 49.0 mmGEM-13180: diam. 4.0 mm, height 50.1 mm	GAMANAL

For 234U most participants obtained results that correlate well with the average of mass spectrometric results for all samples (Table 3). Biases fall within expanded uncertainty (k = 2) limits, amounting to 10–20% relative uncertainty. Older versions of MGAU tend to underestimate 234U due to inaccuracies in the extrapolation of the intrinsic efficiency curve to 120.9 keV [7]. Evaluation by the GAMANAL code also resulted in unrealistically low 234U abundance.

Table 3

Available data of isotope abundances in mass %, measured by gamma spectrometry

Lab code/software	ES-1 powder			ES-2 pellet			ES-3 pellet
	234U	235U	238U	234U	235U	238U	234U	235U	238U
Michelang./IdentifyFRAM	_	1.93 ± 0.14 (Identify) 3.03 ± 0.18 (FRAM)	_	_	1.50 ± 0.06 (Identify)2.14 ± 0.1 (FRAM)	_	_	1.93 ± 0.32 (Identify) 3.41 ± 0.41 (FRAM)	_
Van Gogh/manual	_	3.3 ± 0.5	_	_	2.6 ± 0.2	_	_	2.6 ± 0.5	_
Monet/(FRAM)	0.023 ± 0.004	2.614 ± 0.140	97.34 ± 0.140	0.019 ± 0.006	2.290 ± 0.138	97.677 ± 0.140	0.016 ± 0.012	2.659 ± 0.32	97.308 ± 0.322
Rembrandt/MGAU V3.2	0.028 ± 0.016	2.85 ± 0.16	97.12 ± 0.18	0.014 ± 0.012	2.11 ± 0.15	97.86 ± 0.15	0.016 ± 0.012	2.72 ± 0.15	97.25 ± 0.16
Renoir/MGAU V4.2	0.023 ± 0.0024	2.868 ± 0.047	97.11 ± 0.0475	0.0154 ± 0.0032	2.202 ± 0.0424	97.78 ± 0.0430	0.0244 ± 0.0041	2.833 ± 0.059	97.143 ± 0.06
Picasso/MGAU V4.2	0.026 ± 0.004	2.823 ± 0.039	97.151 ± 0.040	0.018 ± 0.004	2.138 ± 0.034	97.844 ± 0.035	0.025 ± 0.004	2.787 ± 0.041	97.188 ± 0.041
Buonarroti/MGAU V4.2	0.025 ± 0.0014	2.851 ± 0.038	97.124 ± 0.040	0.0192 ± 0.0012	2.154 ± 0.030	97.827 ± 0.030	0.0250v± 0.0010	2.861 ± 0.036	97.114 ± 0.036
Pollock/U235 V1.51	0.025 ± 0.010	2.91 ± 0.020	97.07 ± 0.020	0.019 ± 0.010	2.16 ± 0.020	97.83 ± 0.020	0.019 ± 0.010	2.86 ± 0.020	97.12 ± 0.020
Gauguin/GAMANAL	0.01565 ± 0.00072	2.752 ± 0.080	97.233 ± 0.080	0.00747 ± 0.00075	2.21 ± 0.12	97.78 ± 0.12	0.00916 ± 0.00092	2.99 ± 0.15	97.00 ± 0.15
Community average MS without outliers	0.02433 ± 0.00093	2.875 ± 0.040	97.076 ± 0.0082	0.01881 ± 0.00108	2.190 ± 0.040	97.789 ± 0.046	0.02441 ± 0.0083	2.875 ± 0.043	97.084 ± 0.084

Expanded uncertainties’ coverage factor k = 2. The community average of mass spectrometric results is calculated without the outliers, and its expanded standard deviation (k = 2) is given

Concerning 235U, all participants recognized the LEU character of the samples, regardless of the detector-type and software used. Participants found similar enrichment values of around 2.6–2.9% for samples ES-1 and ES-3, and 2.1–2.3% for the sample ES-2 (Table 3). The majority of values reported by participants provided a means to differentiate ES-1 and ES-3 from ES-2, regardless of the method used.

MGAU v4.2 results from spectra taken by a planar detector compared very well to the average 235U enrichment measured by mass spectrometry for all three samples. The slight underestimation of the 235U content by MGAU may come from the coincidence summing losses in the peaks of 235U [8]. Some evaluations of spectra taken by coaxial and broad-energy detectors resulted in significantly higher 235U content. This could be due to coincidence summing losses in the high-energy peaks from 234mPa, which leads to an underestimation of the activity ratio 238U(234mPa)/235U and therefore an overestimation of 235U [9]. The 235U enrichment estimates by the Identify software were significantly lower than values reported by mass spectrometry. One reason for the discrepancy could be wrong assumptions on the sample matrix, similar to what previously was observed for low resolution measurements [10].

Regarding 238U, mass abundances of 97.2 ± 0.1% were reported for ES-1 and ES-3, whereas 97.8 ± 0.1% for ES-2, in agreement with mass spectrometric results.

It can be concluded that the inventory of the three major uranium isotopes established by HRGS generally agreed with mass spectrometric results within expanded uncertainties and confirmed the LEU character of the samples. The accuracy was generally enough for distinguishing ES-1 and ES-3 from ES-2. Exceptions for 235U results came from some participants using coaxial germanium detector. Discrepancies for 234U results took place mainly for participants using outdated computer routines [7], or efficiency transfer algorithms based on point-source efficiency calibrations instead of relative, intrinsic efficiency calibrations.

Results of uranium isotopic abundance measurements by gamma spectrometry and community average values by mass spectrometry for comparison are summarized in Table 3. The mass spectrometry average is calculated from the data given in graphical form in the CMX-4 After-Action report [1]. For this calculation the outliers were removed. A three-isotope plot of the relative biases with respect to mass spectrometry for the 234U/238U and 235U/238 isotopic ratios is shown Fig. 1.

Fig. 1

Relative biases with respect to the community average mass spectrometric result. The bias of the mass ratio 234U/238U versus the bias of 235U/238 is shown for all three samples. Uncertainties are displayed with a coverage factor of k = 1

U age dating

The model age (time elapsed since the last chemical purification) of the material is important for determining of the origin of nuclear material outside of regulatory control. The daughter/parent ratio as a function of decay time is widely used for determining the age of radioactive samples [11, 12]. Gamma spectrometric uranium age dating is nondestructive and suitable for relatively rapid assay. The method does not require the use of reference materials of known ages, nor radionuclide standards for method calibration. Usually there is no need to take subsamples or dismantle the investigated items, so preservation of evidence can easily be ensured. The method works particularly well for high-enriched and aged material. Its limits appear for low-enriched material and in sensitivity to background.

Uranium age dating by HRGS is based on the 234U → 230Th → 226Ra chronometer [13-19]. 234U can be detected by its 120.9 keV gamma line, whereas 230Th has no measurable gamma rays. The next member of the 234U decay series is 226Ra, of which the only gamma-line at 186.2 keV overlaps with 185.7 keV line of 235U. However, its short lived descendants 214Pb and 214Bi have measureable gamma lines.

The time needed for secular equilibrium between 226Ra and 214Bi is about 3 weeks, so it can be assumed that the activities of 226Ra and 214Bi are equal at the time of the measurement. Although we are not aware of any experimental evidence that 222Rn would escape from the solid samples, it is useful practice to hermetically seal the samples in small containers.

For determining 214Bi activity, the intensity of the 609.3 keV gamma line (and some other 214Bi lines) can be recorded by a large coaxial HPGe spectrometer under low-background conditions. The same spectrum is used to determine 238U peaks for relative efficiency. As 214Bi is a cascade emitter, true coincidence summing losses can cause a bias for short sample-to-detector distances, and should preferably be corrected for. However, for larger sample-to-detector distances the bias due to true coincidence summing losses can often be neglected compared to other sources of uncertainty.

The line of 214Pb at 352 keV does not suffer from true coincidence summing losses. However, it is usually difficult to quantify due to its low intensity and high background continuum. Furthermore, it is far away from the peaks which are used to construct the relative efficiency curve, so the uncertainty of the relative efficiency at 352 keV is very high.

The 226Ra/238U activity ratio determined through measuring 226Ra descendants is divided by the 234U/238U ratio obtained during the measurement of the U isotopic composition. The age of the sample is then calculated from the 226Ra/234U ratio. 234U is preferentially enriched along with 235U in the enrichment process. Hence, for lower 235U abundances, the amount of 234U (and therefore of 214Bi) is lower as well, so the corresponding activity is more difficult to measure.

To extend the capabilities of the method, a high-efficiency, 293 cm3 well-type detector was acquired by a participant laboratory (Picasso). The first application of this kind of detector for uranium age dating was assaying the CMX-4 exercise samples [20]. The well-type HPGe detector (Canberra GCW6023) was in an iron chamber of 20 cm wall thickness. A spectrum of ES-3 acquired by the well-detector is shown in Fig. 2. Owing to their low enrichment and age, an upper limit of ≈ 11 years was estimated uniformly for the three samples. This result was consistent with the results from destructive measurements.

Fig. 2

Spectrum of ES-3 (live time = 55,008 s) and of the background (live time = 65,122 s) taken by a well-type HPGe detector in a low-background iron chamber. Both spectra are normalized to 60,000 s live time (Picasso). No surplus of the peak area of 214Bi at 609 keV was observed above background. Abundance of 232U was evaluated from the net peak areas at 583, 860, and 2614 keV of 208Tl, and those at 238 and 727 keV of 212Pb and 212Bi, respectively

Ten participant laboratories employed mass spectrometry and alpha spectrometry for determining 234U and 230Th in the samples [21]. One of those laboratories applied 231Pa/235U chronometer as well. Most of the measured ages (around 10 years for ES-1 and ES-3 and around 12 years for ES-2) were consistent with the known history of the material.

Identification of reprocessed material

Gamma spectroscopy is useful for screening uranium samples for fission products, and some of the actinide isotopes that would be produced by neutron activation. If these were detected, they would indicate reprocessed uranium, and also offer clues about the reprocessing technology by identifying deficiencies in the uranium purification chemistry. No fission or activation products were detected in the CMX-4 samples, and upper limits reported for representative nuclides were similar between the three samples.

The minor isotopes 236U and 232U are characteristic of reprocessed material (or blended/contaminated with reprocessed material) and their presence gives evidence of previous neutron irradiation (e.g., in a reactor) of the sample. The very low 232U concentration can be measured by alpha- [22, 23] or gamma spectrometry [18, 24–26].

Some participants of CMX-4 detected 232U by HRGS, using heavy shielding for low-background counting: Presence of 232U in ES-1 and ES-3 indicates that the samples were manufactured from material mixed or contaminated with reprocessed uranium. According to mass spectrometric results [1], the concentration of 236U in both samples is about 0.0020 ± 0.0004% on average, but near the detection limit in ES-2 (≤ 10−4). This confirms the conclusion from HRGS that ES-1 and ES-2 contained some reprocessed material, while ES-2 did not.

Gauguin reported (5.6 ± 3.8) × 10−11% 232U concentration in ES-1, whereas the two pellets were given upper limits as 8.8 × 10−11 and 6.3 × 10−11%.

Vermeer identified 238, 583, and 727 keV gamma lines of 212Pb, 208Tl, and 212Bi, respectively, (descendants of 232Th and 232U alike) in ES-1 and ES-3 samples, using a low-background system (15 + 5 cm Pb; the inner 5 cm layer is of a low 210Pb source), but the 911 keV line of 228Ac (daughter of only 232Th) was missing. Hence, it was concluded that the former did not derive from 232Th, but rather from 232U (decay scheme of the two nuclides is common starting from 228Th). However, they could not find any of those lines in the ES-2 pellet.

Picasso evaluated 208Tl, 212Bi, and 212Pb peaks in spectra of ES-1 and ES-3 samples, in absence of 228Ac, but no such peaks for ES-2 above background. The 2614 keV 208Tl peak was observed in the spectrum of ES-1 (also ES-3, Fig. 2) sample, corresponding to (6 ± 0.5)10−11% 232U concentration, but difficult to exclude from background (≈ 10−11%) in the ES-2 sample. From the lack of the 911 keV line of 228Ac, Picasso estimated a detection limit of about 10−4 for the 232Th/238U mass ratio with a 12 h measurement time using the well detector.

Buonarroti performed analysis similar to that of Picasso, and obtained that ES-1 and ES-3 contain (1.40 ± 0.96)10−10% and (1.20 ± 0.70)10−10% of 232U, respectively, whereas ES-2 contains less than 1 × 10−10%.

236U cannot be analyzed in LEU by gamma spectrometry, only in extremely high-enriched (weapons-grade) material [13], because its peaks are masked by the much stronger peaks from the major U isotopes (e.g., the strongest, but still quite weak 49.369 keV line of 236U lies very near the 49.55 keV line of 238U.) Thus, 236U abundances were only quantified by mass spectrometry.

A correlation between the 236U and 232U contents exists. According to Picasso’s results, the 236U/232U abundance ratio was ≈ 3 × 107 in the ES-1 and ES-3 samples. This value is in agreement with earlier results [18, 24, 26] on U samples over the full range of enrichments, from LEU to the highest enriched (90%) uranium, and is in accordance with theoretical predictions [27].

It is not clear where the reprocessed U in samples ES-1 and ES-3 comes from. All the three samples were made in the same factory (fictitious “EA Fuel Products” or virtually HIFAR [1]) and from natural U. If they got contaminated with reprocessed U in the enrichment plant, all three should contain traces of 236U and 232U, unless the facility became contaminated between processing the two source materials A and B.

Conclusions

For identifying the provenance of unknown nuclear material, information on the isotopic composition, the age, and previous neutron irradiation of the material is relevant. In addition to previous exercises when weapon grade materials were examined, this exercise confirmed that gamma spectrometry also plays a significant role for the analysis of LEU in the comprehensive response to these issues.

Results of this exercise confirmed that LEU can be categorized as such via gamma spectrometry within 24 h, regardless of the detector, software, and calibration methodology. For accurate determination of isotope ratios, the best results were acquired with planar HPGe detectors and current versions of the MGAU and U235 software. The utility of high-efficiency HPGe detectors in low-background setup was demonstrated for detecting trace 232U, thus indicating contamination with reprocessed U. Challenges related to the age dating of low-enriched and young aged uranium (difficulty with determination of lower amounts of 214Bi) were identified.

The combination of different analytical techniques increases the confidence in the results and can help further narrow down the set of possible origins and intended uses of the examined materials.