Beyond Californium – A Neutron Generator Alternative for Dosimetry and Instrument Calibration in the U.S.

Abstract

Evaluations of neutron survey instruments, area monitors and personal dosimeters rely on reference neutron radiations, which have evolved from the heavy reliance on (α,n) sources to a shared reliance on (α,n) and the spontaneous fission neutrons of ²⁵²Cf. Capable of producing high dose equivalent rates from an almost point source geometry, the characteristics of ²⁵²Cf are generally more favorable when compared to the use of (α,n) and (γ,n) sources or reactor-produced reference neutron radiations. Californium-252 is typically used in two standardized configurations: unmoderated, to yield a fission energy spectrum, or with the capsule placed within a heavy-water moderating sphere to produce a softened spectrum that is generally considered more appropriate for evaluating devices used in nuclear power plant work environments. The U.S. Department of Energy (DOE) ²⁵²Cf Loan/Lease Program, a longtime origin of affordable ²⁵²Cf sources for research, testing and calibration, was terminated in 2009. Since then, high-activity sources have become increasingly cost-prohibitive for laboratories that formerly benefited from that program. Neutron generators, based on the D-T and D-D fusion reactions, have become economically competitive with ²⁵²Cf and are recognized internationally as important calibration and test standards. Researchers from the National Institute of Standards and Technology (NIST) and the Pacific Northwest National Laboratory (PNNL) are jointly considering the practicality and technical challenges of implementing neutron generators as calibration standards in the U.S. This article reviews the characteristics of isotope-based neutron sources, possible isotope alternatives to ²⁵²Cf and the rationale behind increasing favor of electronically-generated neutron options. The evaluation of a D-T system at PNNL has revealed characteristics that must be considered in adapting generators to the task of calibration and testing where accurate determination of a dosimetric quantity is necessary. Finally, concepts are presented for modifying the generated neutron spectra to achieve particular targeted spectra, simulating ²⁵²Cf or workplace environments.

Background

Neutron detection devices, including survey instruments, area monitors and personal dosimeters, used in radiation protection, security and other applications must be tested and periodically calibrated. For rate detectors in particular, this requires the use of one or more sources of neutrons to cover the response range over which the detectors are to be used. Calibrations have historically relied upon a variety of isotopes, but evaluations performed at high ambient or personal dose equivalent rates (e.g., 20 mSv h^-1 and above) are most readily achievable using high-activity ²⁵²Cf sources.

Prior to the availability of ²⁵²Cf in the 1960s, most neutron calibration sources consisted of alpha-emitting isotopes intimately mixed with light elements. Early sources were primarily based on the use of ²²⁶Ra and ²¹⁰Po, but were eventually supplanted by ²³⁹Pu, ²³⁸Pu and ²⁴¹Am. The light elements paired with these alpha-emitting isotopes are typically beryllium, boron, fluorine and lithium. The average neutron energies produced using alpha-neutron (α,n) reactions vary depending on the light element used. Beryllium-based sources provide the highest neutron intensity with typical average energy from 4 to 5 MeV, while those using boron, fluorine and lithium range from hundreds of keV to a few MeV [USNRC 1991]. The specific neutron intensity for all common (α,n) sources are significantly less than for ²⁵²Cf, ranging from about 200 times less for ²¹⁰Po-Be to over ten million times less for ²³⁹Pu-Be as shown in Table 1.

Table 1. Overview of Neutron Calibration Sources and Properties.

Neutron source (type)	Radionuclide half-life	Number of sources needed for 15 yearsa	Relative neutron intensity of individual sourcea	Radionuclide nominal specific neutron intensity	Nominal mass of radionuclide needed for 109 s-1 source
	(years)			(s-1 g-1)	(g)
DD (NG)	n/a	3-5	1.000	n/a	n/a
DT (NG) b	12.3	3-5	1.147-1.087	2.28 × 1013	0.00004
252Cf (SF)	2.645	2	2.285	2.31 × 1012	0.0004
210PoBe (α,n)	0.378	13	2.409	1.08 × 1010	0.1
244CmBe (α,n)	18.1	1	1.315	2.06 × 108	5
238PuBe (α,n)	87.74	1	1.060	4.12 × 107	24
248Cm (SF)	348000	1	1.000	3.94 × 107	25
226RaBe (α,n)	1620	1	1.003	1.27 × 107	79
244Cm (SF)	18.1	1	1.315	1.11 × 107	90
241AmBe (α,n)	433.6	1	1.012	8.23 × 106	122
239PuBe (α,n)	24100	1	1.000	1.49 × 105	6700

^aNumbers are estimated based on assumption that source is used for the length of its three half-lives of corresponding radionuclide or the manufacturer's specified neutron tube or target lifetime.

^bNeutron tube lifespans may differ significantly. Typical working life of 1000-2000 hours can last for three and even five years, depending on how often the generator is used and power level. Specific neutron intensity is calculated as the maximum neutron yield divided by tritium quantity in the neutron tube of a particular DT generator.

Other sources were developed based on gamma-neutron (γ,n) reactions, emitting relatively discrete low energies (keV range). These photoneutron sources are employed in industries (applications) other than radiation protection (dosimetry). While some of these sources provide high neutron intensities, they are unfavorable for dosimetry and calibration applications because of short half-life and intensive associated gamma contribution.

Eventually, spontaneous fission sources (e.g., ²³⁸Pu, ²⁴²Cm, ²⁴⁴Cm and ²⁵²Cf) emerged as important sources for calibration and testing. In terms of the ambient or personal dose equivalent, ²⁵²Cf offers extremely high neutron output coupled with low photon contribution. Partly due to the availability through the ²⁵²Cf Loan/Lease Program [Sherman and Patton 2013] at the Oak Ridge National Laboratory (ORNL), this isotope became readily available in large quantities, enabling some laboratories to attain multiple sources for establishing a broad range of fluence or dose equivalent rates. The spontaneous fission neutron spectrum of ²⁵²Cf has been well-studied and the isotope was identified as a potential calibration source as early as 1962 [IAEA 1971]. In 1978, ²⁵²Cf was one of several options identified for testing and calibration of radiation protection instrumentation in the U.S. [ANSI 1978].

A characteristic of the more commonly applied neutron detection methods used for personal dosimeters and dose rate measuring instruments is significant neutron energy dependence. Unfortunately, the characteristic neutron energies emitted by most neutron sources used for calibrating radiation protection-related devices are not representative of most workplace environments in which neutrons pose an occupational hazard. It is often an erroneous assumption that neutron detectors calibrated using (α,n) or spontaneous fission sources will assure indicated dose equivalent response accuracy under workplace conditions unless potential response differences are understood and appropriate corrections are applied. The U.S. Nuclear Regulatory Commission especially took note of this, leading researchers at the National Institute of Standards and Technology (NIST), in 1976 (then the National Bureau of Standards), to develop a “workplace field” spectrum created by “moderating” ²⁵²Cf with a cadmium-covered, 15 cm radius sphere containing deuterium oxide (D₂O) [USNRC 1980]. The D₂O-moderated ²⁵²Cf field was developed in order to produce approximately the same dose equivalent response in albedo thermoluminescence dosimeters and hand-held neutron survey instruments as was observed when used in nuclear power plant work environments.

In 1989, D₂O-moderated ²⁵²Cf, along with the unmoderated ²⁵²Cf fission spectrum and (α,n) sources ²⁴¹Am-Be and ²⁴¹Am-B were internationally standardized for calibration and testing [ISO 1989]. Over the past 30 years, ²⁴¹Am-Be, ²⁵²Cf and D₂O-moderated ²⁵²Cf have been prominently used in U.S. calibration and testing laboratories and have also been incorporated into many published American National Standards Institute (ANSI) standards [ANSI/IEEE 2004a, ANSI/IEEE 2004b, ANSI/IEEE 2004c, ANSI/HPS 2009, ANSI/IEEE 2013]. In addition to routine detector calibration, these standards advise ²⁵²Cf and D₂O-moderated ²⁵²Cf for the proficiency testing of personal dosimeters and for detector type tests to determine such characteristics as integrated dose equivalent and dose equivalent rate linearity, energy and angular dependence, alarming capabilities, interfering radiation fields and overrange behavior. Fluence spectra for these isotope-based reference neutron radiations are compared in Fig. 1.

Fig. 1.

Comparison of ²⁵²Cf, D₂O-moderated ²⁵²Cf and ²⁴¹Am-Be isotope-produced neutron spectra specified in ISO 8529-1. Not shown is the fourth isotope-based reference field cited in the standard, ²⁴¹Am-B, which has not been prominently applied within the U.S.

The ORNL ²⁵²Cf Loan/Lease Program made sources available, for relatively low cost, primarily to universities and government agencies and contractors. A five year loan would typically include a modest loan fee, along with costs for encapsulation, restocking (at the end of the loan period) and shipping (i.e., no cost for the isotope itself). Even before the broad dissemination of ²⁵²Cf sources through the loan arrangement, source or bulk isotope could be purchased through the ²⁵²Cf Sales Program at variable isotope costs that escalated from about $10 μg^-1 in the 1970s to $68 μg^-1 in 2002. In 2009, funding for production of ²⁵²Cf was discontinued by the U.S. Department of Energy (DOE), which concerned those industries that heavily relied on the isotope. In response to continued demand for ²⁵²Cf, several private sealed source manufacturers consolidated resources to continue funding U.S. production and purification of ²⁵²Cf, thus preserving the availability of this important isotope. Unfortunately, the ²⁵²Cf Loan/Lease Program did not survive this transition [Sherman and Patton 2013] and, in the first years of privately-funded production and purification, the cost of procurement for an encapsulated ²⁵²Cf source increased approximately tenfold over the purchase cost just 10 years earlier. By 2016, cost was nearly 15 times more than in 2002. The relative increase was even larger compared to the cost of obtaining encapsulated sources through the former ²⁵²Cf Loan/Lease Program. For users of relatively small activity sources, the cost increase appears tolerable, based on continued demand. However, for some laboratories that depend upon high-intensity sources (e.g., neutron emission rate of 10⁹ s^-1 and above), the marked increase in cost is prohibitive and an alternative will eventually be needed.

Alternatives to High Intensity ²⁵²Cf

To avoid the heavy cost of procuring high-intensity ²⁵²Cf sources, two general alternatives were considered at PNNL for addressing future needs. The first included procuring lower-intensity sources, leading to a lower ceiling of available neutron dose equivalent rates for tests and calibrations. The second alternative was to identify other means of delivery of controlled neutron fluence, including alternative isotopic sources or other means of neutron production (e.g., accelerator- or reactor-produced fields).

Utilizing lower activity ²⁵²Cf sources

Determining the minimum quantity of ²⁵²Cf source needed must account for the decay half-life of about 2.65 years and physical and practical limitations of detector placement relative to the source. At PNNL, newly acquired ²⁵²Cf sources have historically ranged from 1.5 (3.5 × 10⁹ s^-1) to 2.5 mg (5.9 × 10⁹ s^-1) with an average replacement time of about 7 ± 1 years (or about three half-lives). At acquisition, these sources provided nominal ambient dose equivalent rate ceilings of 1 Sv h^-1 and 250 mSv h^-1 at 25 cm in unmoderated and D₂O-moderated configurations, respectively. These rates decrease by an order of magnitude after approximately three half-lives.

The procurement of sources significantly less intense results in ambient dose equivalent rate ceilings proportionally reduced. Some detectors may be placed closer to the source to compensate for reduced source activity; however, for some large, moderator-equipped, neutron survey instruments, proximity to the source is a limited option. A non-uniform neutron field produced across the volume of the moderator at short distances induces a bias in the response, relative to the normal inverse-squared relationship. Although geometry correction methods for this phenomena have been developed for some standard types of detectors [Kluge et.al. 1990], all corrections impose additional uncertainty upon the measurements. Many neutron detector meters may be calibrated at upper ranges via the use of electronic means (e.g., pulse generators), but this alternative also induces additional uncertainty. In the case of instruments conventionally calibrated or tested using D₂O-moderated ²⁵²Cf, response linearity (across ranges) could be evaluated using the higher dose equivalent rates of unmoderated ²⁵²Cf, with the final instrument response efficiency determined using lower ambient dose equivalent rates of the D₂O-moderated ²⁵²Cf spectrum. Where application of these alternatives are not possible or practical, detector calibrations would be range-limited. This option may be technically acceptable, depending on the required range and uncertainty tolerance of instruments to be calibrated, however, source costs would still be substantial.

Consideration may also be given to procuring lower activity sources with increased frequency to help limit the upper range ambient dose equivalent rates lost through longer (e.g., three half-lives) decay intervals. With each source procured, there are peripheral expenses incurred, including multiple shipments, NIST-calibration, various characterization efforts and more frequent need for reevaluation of uncertainties. These supplemental costs would diminish potential savings.

Utilizing other isotopic sources

Besides ²⁵²Cf, only a few available isotopes with a significant half-life produce neutrons via spontaneous fission. Curium-244 and ²⁴⁸Cm, the daughter product of ²⁵²Cf, both have attractive longer half-lives. For these isotopes; however, relatively massive quantities would be required to yield 10⁹ s^-1 (see Table 1) and the associated spontaneous fission energy spectra are not widely reported. Availability of these isotopes is uncertain but, regardless, their cost may be similar to ²⁵²Cf, since the cost of ²⁵²Cf is largely dictated by the high cost of curium target material. Consequently, these isotopes are considered impractical replacements for ²⁵²Cf.

Of the several types of historically available (α,n) sources, ²⁴¹Am-Be currently is the most available and widely-applied alternative for calibrations, despite the fluence spectrum mean energy of about twice that of a spontaneous fission spectrum. To obtain an ²⁴¹Am-Be source with intensity near 10⁹ s^-1 would require 120 g or more of ²⁴¹Am (approximately 18.5 TBq). Based on scaling the dimensions of available smaller activity sources, a capsule containing 120 g of isotope, suitably mixed with beryllium, may occupy a volume ranging from 120 to 230 cm³; massive compared to the almost point-source geometry of ²⁵²Cf sources. The large dimension of ²⁴¹Am-Be sources could be problematic for calibrations from a geometry consideration, but a similar issue is represented by the 14,000 cm³ configuration of D₂O-moderated ²⁵²Cf. In that case, the geometry correction methodology of Kluge [1990] is applied. Should close detector proximity to a massive ²⁴¹Am-Be source be necessary, an adaptation of the Kluge methodology or a separate approach may be derived for use. Nevertheless, the influence of the physical geometry could also affect the energy spectrum. While significantly different masses of ²⁵²Cf can be encapsulated without concern for appreciable self-moderation, differently configured ²⁴¹Am-Be sources may present unique low-energy spectral characteristics [Bedogni et.al. 2014]. Such influences potentially create source configuration-dependent dose conversion coefficients that may be difficult for source owners/users to characterize or simulate (i.e., for determining source-dependent fluence-to-dose equivalent conversion coefficients).

Several additional factors make such a large magnitude ²⁴¹Am-Be source less attractive. A concern with all (α,n) sources as aging occurs is the possibility of dissociation of the ²⁴¹Am and beryllium constituents, which requires extra vigilance of the emission rate through frequent NIST manganese bath calibrations or intermediary, high-precision constancy verifications. The ability to contain such a source in a single, currently-approved (e.g., International Atomic Energy Agency Special Form certificate) encapsulation has not been thoroughly explored, but is unlikely. Finally, based on extrapolation from previously procured PNNL sources, it is estimated the cost for a source of this magnitude would be of the same order as an equally capable ²⁵²Cf source.

Other (α,n) sources, although currently uncommon, may be worth considering. The specific neutron intensity of ²¹⁰Po-Be is about 1000 times higher than for ²⁴¹Am-Be and less than 0.1 g would produce an emission rate of 10⁹ s^-1, but its half-life of 138 days would necessitate frequent replacement. A 5 g mass of ²⁴⁴Cm, in the form of an (α,n) source, could yield 10⁹ s^-1 with a half-life of 18.1 years; however, no previous references have been identified related to such application of this source configuration.

From this review, ²⁴¹Am-Be is a logical (α,n) candidate for replacement of ²⁵²Cf; however, its high cost and the difficulty of attaining the substantial activity needed to meet high-intensity neutron output needs, especially contained within a single capsule, weaken this option.

Accelerator- or reactor-produced fields

A variety of neutron-emitting nuclear reactions have been identified [ISO 2001] for use in testing or calibrating radiation detection devices. Reactor-produced reference neutron radiations (e.g., attainable at research reactors) are filtered using various materials (e.g., scandium, iron/aluminum or silicon) in order to produce approximately monoenergetic reference neutron radiations. Accelerated proton interactions with tritium [³H(p,n)³He] and ⁷Li [⁷Li(p,n)⁷Be] and the deuterium-deuterium (D-D) and deuterium-tritium (D-T) fusion reactions, ²H(d,n)³He and ³H(d,n)⁴He, respectively, are also listed. Reactors and particle accelerators require large capital and maintenance investments and may be attainable at some research institutions, including NIST, but are unlikely to be available at second-tier metrology or dosimetry calibration laboratories (e.g., accredited dosimetry or ionizing-radiation calibration laboratories) responsible for calibration and testing of field-level detection devices.

Relatively compact, fusion-based accelerators – commonly referred to as neutron generators – have been developed to produce the D-D and D-T reactions noted above and are being increasingly applied in research and industry. These devices also are becoming more capable in terms of operational efficiency and increased neutron emission rate while gradually decreasing in cost. Such D-D and D-T neutron generators produce relatively monoenergetic neutrons of approximately 2.5 MeV and 14 MeV, respectively.

With respect to calibration applications, reference neutron radiations produced using the D-D and D-T fusion reactions offer both advantages and challenges relative to those based on radioactive decay. Included among the general advantages are well-studied and documented fusion reaction kinematics, which enables the calculation of energy and angular distribution of the emitted neutrons. Such calculations are necessary for the development of the source term in conducting the detailed modeling of the neutron energy and directional distribution at a reference point that complements measurements of the neutron energy spectrum. The neutron yield of the D-T reaction is two orders of magnitude higher than the D-D reaction at commonly used accelerator voltages, but D-T generators must employ a sealed tube design for retaining tritium. A D-D generator may be implemented as an “open vacuum” system that allows more flexibility in design and a target assembly that can be more easily serviced and maintained. Current and commercially available D-D and D-T generators offer similar neutron emission rates exceeding 10¹⁰ s^-1. The procurement and configuration of a neutron generator system is significantly more costly when compared to obtaining a similar yield ²⁵²Cf source through the former ²⁵²Cf Loan/Lease Program, however, under the current ²⁵²Cf cost structure, a neutron generator may represent less capital expense.

At first, the choice of a neutron generator for the specific purpose of replacing a ²⁵²Cf spectrum may seem obviously in the favor of D-D generators, given the production of a narrow spectrum with energy relatively similar to the average energy of the fission spectrum. However, the neutron energy spectrum of the D-D reaction does not extend to the high end of the fission spectrum, above 10 MeV, which may be significant for the evaluation/testing of certain detectors with high sensitivity in this region. Deuterium-tritium generators produce neutrons well above the median fission energy.

Neutrons from both D-D and D-T reactions may be shaped, via scattering media, to attain neutron energy spectra concentrated at particular regions below the peak energy emitted by the respective generators. A fission reaction may also be induced from the emitted neutrons of either D-T or D-D reactions impinging on a suitable converter, such as ²³⁸U.

In 2015, researchers at NIST and PNNL began investigating alternatives to the use of ²⁵²Cf for neutron detector calibrations and testing. Options regarding the neutron spectra to be produced were considered as well as the production means. The initial focus was on 14 MeV neutrons produced using a D-T neutron generator. A few laboratories – mostly national measurement institutes – have employed such capabilities, among others, for evaluating neutron detector energy response, dose equivalent rate dependence and directional dependence, but not as replacements to isotope-based reference neutron radiations. At the Cadarache Laboratory in France, the Insitut de radioprotection et de Sûreté Nucléaire^‡ (IRSN) has produced and reported realistic neutron calibration fields based on D-D and D-T generators, facilitated by supplemental nuclear reactions and scattering methods [Chartier et.al. 1997]. These concepts are now at the forefront of the collaborative evaluation of alternatives to the current use of ²⁵²Cf. A detailed investigation was initiated at PNNL for the D-T reaction, in particular focusing on producing spectra similar to D₂O-moderated ²⁵²Cf and a hypothetical average workplace spectrum derived from historical measurement data [Mozhayev et.al. 2016].

Neutron generator systems are not without challenges. High output systems may be of substantial size and mass, require special power service and likely require cooling systems. In addition, facility shielding needs – especially for the 14 MeV neutrons of D-T systems – may be substantial. Some generators may require periodic maintenance operations during extended times of non-use and the target assembly will erode with use – often being limited to 1000 to 2000 hours [IAEA 2012]. Thus, periodic replacement targets (possibly including the entire neutron tube) are necessary for long term applications, demanding associated cost and recharacterization efforts.

Even taking into account the above-identified challenges, both D-D and D-T neutron generators may offer suitable reference neutron radiation alternatives to ²⁵²Cf for detector testing and calibration. The use of neutron generators, in place of ²⁵²Cf or ²⁴¹Am-Be sources could also reduce the difficulties associated with the possession, protection, and use of accountable quantities of special nuclear materials.

Investigation of D-T Neutron Generator-Produced Fields

Employing a continuous operating mode D-T generator commissioned at PNNL in 2012 for the purpose of activation analysis research, investigations were initiated into its specific application for detector calibration and testing purposes. Initially, it was the goal of researchers to integrate the 14 MeV reference neutron radiation among the ISO 17025 accredited^§ neutron calibration capabilities at PNNL. To that end, the means to characterize and calibrate the field, and a thorough knowledge of the various uncertainties associated with its use are required. A secondary goal was the modification of the 14 MeV field using various energy-shaping moderators and scattering media to create one or more realistic workplace neutron field surrogate(s) [Mozhayev et.al. 2017], akin to D₂O-moderated ²⁵²Cf. Through these efforts, specific challenges were identified and addressed.

Operational Challenges

The particular neutron generator being used in this effort is housed within the Low Scatter Facility (LSF) at the PNNL Radiological Calibrations Laboratory (RCL). This facility is well-shielded with 1.3 m thick concrete external walls and a sliding, 0.9 m thick, concrete shield door to provide shielding within the building at full power operation of the generator. For routine operations of the pneumatically-controlled ²⁵²Cf sources, for which the room has been largely dedicated for over 30 years, a control room and concrete entry maze was constructed within the boundary of the sliding shield door to facilitate faster frequent entry than could be accommodated using the original door. That configuration provides adequate shielding to surrounding laboratories and offices during the use of at least 3 mg ²⁵²Cf sources within the facility. Providing sufficient attenuation for neutrons and capture gamma radiation associated with the 14 MeV neutron field produced at full power, estimated at 2 × 10¹⁰ s^-1, requires closure of the large shielding door. This results in delayed re-entry into the exposure room, which, in comparison to current ²⁵²Cf usage, translates to a general reduction in functional efficiency for potentially high throughput irradiations.

The initial ramp-up of the neutron generator to full power requires a relatively long interval of stepwise increases in voltage (from 120 to 160 kV) and current (from 0.5 to 3 mA). Minor instabilities of the system routinely occur during both the ramp-up phase and the steady state phase following ramp-up as shown in Fig. 2. A long ramp-up period is not conducive to establishing a stable neutron source calibration or evaluation of integrating dosimetry devices and detection systems. In consideration of the ramp-up and instabilities occurring at high-power, investigations of low-power use (i.e., 120 kV and 0.5 mA) were performed and revealed much improvement of the output stability while still producing a neutron yield of about 10⁹ s^-1. In addition to improved stability, working at low power significantly reduces the ramp-up time, increases the longevity of the generator target/tube and may eliminate the need for use of the large shielding door.

Fig. 2.

Accelerating voltage and beam current for a typical D-T generator run near full output capacity. The initial stepwise ramp-up (left) shows intervals in which adjustments induce short-term instabilities. Even at steady state operations (right), variations in current and voltage are observable but insignificant in comparison to typical neutron detector sensitivity.

Finally, operational protocols, developed originally for the case of full-power sample activation irradiations requires positive access control by radiation protection technologists (RPT), who perform entry surveys following each irradiation. Such long intensive irradiations commonly activate other materials near the generator. For detector test and calibration applications, typically for minutes-long duration at low power settings, researchers anticipate that activation-induced background will be minimal, thereby potentially reducing the need for continuous RPT coverage.

Calibration and Characterization Challenges

For testing and calibration of radiation survey and dose measurement devices, the neutron personal and ambient dose equivalent rate, Ḣ_p(10) and Ḣ*(10), respectively, must be known at the point at which the device is placed (i.e., the reference dose point). Determination of Ḣ*(10) requires knowledge of the fluence rate and distribution of incident neutron energy upon the reference dose point. Determination of Ḣ_p(10) requires knowledge of the fluence rate, distribution of incident neutron energy and direction of incident neutrons at the reference dose point. Acquisition of such information requires a combination of measurement and computations. Measurement of the neutron fluence rate, in particular, must be conducted in a manner that is traceable to the international system of units (SI). Furthermore, because the machine-generated fluence rate is subject to possible variation due to instabilities in beam current, accelerator voltage, wear of the target, etc., one or more flux monitoring devices must be configured and calibrated during the traceable measurement of fluence rate [ISO 2008].

Neutron fluence distribution

The angular-energy distribution of neutrons emitted by a neutron generator is dependent upon several fundamental parameters, including the ion source types, accelerating voltage applied as well as the structure and materials of the target assembly. Depending on these parameters, the D-D reaction will yield neutrons with a mean energy between 2 and 3 MeV and a D-T reaction will produce neutrons primarily in the 13 to 15 MeV range.

The neutron energy distribution for the generator system currently utilized at PNNL was initially evaluated for the task of irradiating/activating samples placed very close to the generator target. That evaluation utilized Monte Carlo N-Particle (MCNP) simulations and spectrum analysis measurements with activation foils/wires placed at polar angles from 0° to 90° around the target [Hayes et.al. 2014]. For greater distances from the target, at which evaluations of detection devices would be conducted, using the same spectrum analysis method involving activation foils was anticipated to be challenging due to the relatively low fluence rate. Other conventional means of spectral analysis (e.g., multi-sphere measurements) were considered, but appeared to be incapable of resolving subtle details of the 13 to 15 MeV spectrum emitted via the D-T reaction. As such, initial evaluations of the energy distribution focused on detailed modeling of the generator. Investigation of the operation of the generator revealed several important details pertinent to the determination of dosimetric quantities. Although related to the generator system currently used at PNNL, these findings might be useful to consider for any D-T generator with dosimetric application.

First, the metal hydride target within the neutron generator is initially self-loaded via ion implantation at the factory during the final stages of preparation. The gas reservoir, from which ions are introduced into the accelerator, holds a nominal 50/50 mixture of deuterium and tritium [Thermo 2016]. Deuterons and tritons accelerated to the target that do not undergo fusion are deposited in the target matrix. The mechanism continually refreshes the target with tritium, enabling the tube to maintain very stable neutron yields over its operational lifetime. Thus, four nuclear reactions, namely D-T, D-D, tritium-tritium (T-T) and tritium-deuterium (T-D), occur at the target, although for accelerating voltages below about 1 MV, D-D and T-T reactions have significantly lower cross sections than the D-T and T-D reactions. Therefore, it is important to consider both D-T and T-D reactions in the simulation of output.

Second, accelerated ions are not limited to single atoms (e.g., H⁺) [Chichester 2009]. The generator at PNNL utilizes a low-pressure, cold-cathode technology, or Penning type, ion source from which 80% to 90% of accelerated ions are molecular and consist of a mix of diatomic (H₂⁺) and triatomic (H₃⁺) species [IAEA 2012]. Further, these species may include random combinations of D₂⁺, DT⁺ or T₂⁺. The molecular ions share the acceleration energy proportionally among the individual atoms. For instance, individual atoms of a D₂⁺ molecule accelerated through the potential difference of 150 keV will share the kinetic energy (about 75 keV each) and will have different fusion cross sections for each relevant reaction (i.e., D-D and D-T) from that of an atomic ion accelerated in the same electric field (see Fig. 3). Other ion source methods are capable of producing higher fractions of atomic ions. For instance, over 90% atomic ions have been achieved using a multicusp radiofrequency (RF) ion source with an internal antenna [Burns and Bischoff 1997].

Fig. 3.

Influence of the distribution of atomic (H⁺), diatomic (H₂⁺) and triatomic (H₃⁺) ion species accelerated through the potential difference of 150 kV in determining the complicated mixture of possible neutron energies. Kinetic energy of individual atoms of these species will be 150, 75 and 50 keV, respectively

Third, neutrons resulting from the fusion reactions within the target are emitted in all directions. Neutron energy distribution depends on the polar angle between their direction and that of an incident ion. This is important to consider with respect to the uniformity of neutron energy spectra across a large instrument or dosimetry phantom being irradiated.

Finally, the location of the target within the neutron tube, and to a lesser degree, the target shape (e.g., planar disc, conical, slanted) may present a variety of scattering possibilities after each neutron is produced. For the system used at PNNL, neutron scattering may occur within the electrode, heat sink, cooling cap and other structural materials. The construction and associated location of the target within different neutron generator models may vary significantly.

Given the number of potential influences on the angular-energy distribution of the field produced by a neutron generator, detailed system analysis and careful modeling of the emission spectrum and angular distribution is a necessity. Equally important is the measurement of the neutron spectrum using high-resolution neutron spectrometry, which can be used to corroborate or benchmark the computational determination of a reference field.

Calibration and flux monitoring

For encapsulated neutron sources, a well-established calibration practice includes submitting the source to NIST for calibration based on the use of the manganese sulfate bath [McGarry and Boswell 1988], followed by the calculation of fluence rate at the reference dose point. MCNP-based computations are then necessary to determine encapsulation-induced fluence anisotropy and possible resulting influences on the energy spectrum of neutrons incident upon the reference dose point. A neutron generator cannot be submitted to NIST for calibration in the same manner and establishing traceability is a non-trivial endeavor that must rely on one or more transfer standards.

The use of common neutron-sensitive survey instruments employing active detectors (e.g., rem-meters, tissue-equivalent proportional counters) were considered for use as transfer standards between a similar 14 MeV primary standard reference neutron radiation, possibly available at NIST or other national measurement institute having such a capability. Such instruments are routinely used within PNNL to extend calibration traceability of ²⁵²Cf within idealized geometry to other working-standard neutron radiation calibration ranges. Some of these detectors exhibit energy dependent responses and their application for this purpose would necessitate detailed modeling of both the primary standard and PNNL reference neutron radiation fields as well as the characteristic response of each detector within those fields in order to successfully establish a traceable fluence rate for the generated reference neutron radiation at PNNL. These devices may also be susceptible to uncertainties related to rate dependency, room-scattered neutron response and accumulated-dose dependence (i.e., dose soaking), which may discourage their use as transfer standards. Methods have been established for foil activation (threshold reactions) [ASTM 2014] to determine neutron yield as well as “average” energy. While these methods overcome some of the drawbacks associated with the use of active, but energy-sensitive detectors, they require relatively intense fields and, as identified earlier, would be a challenge to implement for characterizations at anticipated reference distances (e.g., 75 to 200 cm from the target) where fluence rates are relatively low.

Alternate means of indirect neutron generator calibration have been proposed and explored at NIST for use as transfer standards that could be evaluated in reference to a calibrated ²⁵²Cf source and shipped to offsite laboratory customers [Heimbach 2011]. These methods involve measurements using a ²³⁵U fission chamber and activation of aluminum and copper cylinders. Activation techniques could be more widely applicable for establishing fluence rate traceability because, unlike instruments subject to dead-time effects, they are not affected by count rate and could be applied to a generator utilized in either continuous or pulsed operation mode. These measurement capabilities have not yet been fully implemented at NIST as of the time of this publication. Due to the in-house availability of necessary resources, the ²³⁵U-lined fission chamber-based measurement method relating the D-T generator to a NIST-calibrated ²⁵²Cf (reference) source is under investigation at PNNL. This method requires the accurate knowledge of: (1) reference isotopic source energy, anisotropy and calibrated emission rate, (2) energy distribution of the unknown (i.e. D-T generated) field, and (3) characteristics of the fission chamber. Measurements carried out for both the reference ²⁵²Cf source and the D-T generator field at the reference dose point would provide two count rates that can be related based on the fissile material cross sections for the respective neutron spectra.

During calibration of the reference neutron radiation and during use of the generator for testing or calibration of neutron detectors, it is necessary to monitor the output due to possible instabilities, discussed earlier. In the longer term, such a monitoring system is needed to account for the gradual reduction in neutron output due to degradation of the target and decay of tritium, for possible alterations or inaccuracies of the applied voltage or current indicators or in the event that repairs or replacement components are incorporated.

The qualities of a suitable flux monitoring device include the ability to integrate response, long-term stability, sensitivity to subtle changes in the output of the generator, minimal sensitivity to facility-scattered neutrons and operational simplicity. A flux monitor's indication should be relatable to the calibration of the neutron generator and its position must be held stable relative to the generator target. Furthermore, it should be positioned in a manner unaffected by devices being tested or calibrated using the neutron generator. Some generators are equipped with associated particle imaging (API) systems, which relate emitted neutrons and their direction to the associated alpha particle emitted from the fusion reaction. Such systems have been applied as flux monitors, although such capability is not inherent in the PNNL generator. Monitor candidates initially considered for the PNNL system included a precision long counter, a tissue-equivalent ion chamber, a conversion reaction detector (e.g., BF₃, ³He) and/or a laboratory-grade tissue-equivalent proportional counter (TEPC). Three additional detectors currently being evaluated at PNNL include a cadmium-wrapped, ²³⁵U-lined fission chamber, a TEPC-based survey instrument, the REM 500^**, and a neutron-sensitive electronic dosimeter with dose profiling capability. These devices have undergone trials under varying current and voltage conditions of the generator and, similar to that shown for the REM 500 in Fig. 5, each detector appears to have sufficient sensitivity to changing conditions, especially when running count-rate averages are used.

Fig. 5.

Evaluation of a TEPC-based survey instrument in monitoring the output of a neutron field under varying current and voltage conditions. The top graph shows the flux monitor response to the accelerating voltage and beam current conditions shown in the bottom graph.

Standardization of Neutron Generator Fields

The international working group responsible for ISO 8529-1 [ISO 2001] has begun consideration of alternatives to ²⁵²Cf for the next anticipated update of that standard. The initial focus is limited to methods that can closely simulate the currently established neutron reference fields produced using unmoderated and D₂O-moderated ²⁵²Cf using energy-shaped generators. The retention of a capability for international harmonization in neutron measurements via a fission-like spectrum - currently fulfilled through the use of ²⁵²Cf - is currently viewed as important. Another consideration is whether to develop a capability based on using the D-D reaction. Use of a D-D based system is attractive due to the lower energy and similarity of the fusion reaction average energy to neutron fields found in nuclear power plant work environments. The degree to which it can be developed to replicate a fission spectrum and the reproducibility of systems among different laboratories - potentially using different generator models - are major concerns. An additional consideration is ingrowth of D-T reactions in D-D systems [Cecil and Nieschmidt 1986] via the 50% probability of a tritium atom (and proton) produced through the D-D fusion reaction. This influence is analogous to the relevance of the increasing ²⁵⁰Cf fluence rate and associated spectrum uncertainty of aging ²⁵²Cf sources. If not acknowledged and tracked, such build-up may lead to significant biases.

D-T systems generally offer higher output and perhaps more flexibility to alter the neutron spectrum via moderation and/or reflection; however, shielding the 14 MeV component can be difficult. Furthermore, although the tritium content of D-T generators have been compared to self-illuminated “EXIT” signs [Chichester and Simpson 2003], working with tritium may be a sizable burden for some laboratories.

NIST and PNNL have jointly investigated possible concepts for creating a fission-like spectrum. These efforts have, so far, been limited to consideration of neutrons generated by the D-T reaction. Simulations have shown that two conversion interaction configurations may closely approximate a typical fission spectrum. One design concept relies on the use of a depleted uranium (DU) target to induce fission reactions along with a target-surrounding neutron “shaper” composed of a lead-bismuth eutectic (LBE). A second concept involves a plug of lead with a tungsten tip, placed close to the generator target, to reduce the 14 MeV neutron component, with a surrounding bismuth shaper. The energy-shaping concepts are illustrated in Fig. 6 with the corresponding spectra shown in Fig. 7.

Fig. 6.

Two energy-shaper concepts to convert D-T generated neutrons to fission-like spectra. On the left, Concept A - a depleted uranium (DU) convertor coupled with a shaping assembly made of a lead-bismuth eutectic (LBE). At right, Concept B - a lead (Pb) plug with tungsten (W) tip is combined with the surrounding bismuth (Bi) annular cones to shape the neutron spectrum.

Fig. 7.

Simulated spectrum utilizing concept energy shapers (see Fig.8) to induce a fission-like neutron spectrum originating from a D-T neutron generator. The resulting residual 14 MeV peak represent totals of 30% of total fluence for Concept A (dot-dashed line) and 15% of total fluence for Concept B (solid line).

Further simulations have been performed to create surrogate spectra that are intended to replicate the D₂O-moderated ²⁵²Cf “workplace” reference neutron spectrum described in ISO 8529-1 [ISO 2001]. These spectra, shown in Fig. 8, utilize both D-T and D-D supplied neutrons as the starting spectra. Neutron energy shapers consist of 10 cm of lead plus 20 cm of polytetrafluoroethylene (PTFE) and 30 cm of PTFE plus a thin layer of cadmium, respectively.

Fig. 8.

Concept spectra for replacement of D₂O-moderated ²⁵²Cf spectrum using neutrons generated from a moderated D-T system (solid line) and a moderated D-D system (dotted line).

Another alternative to replacing currently established ²⁵²Cf references is to generate realistic workplace field(s). Many other specific spectral characteristics are possible following similar techniques used among the above examples [Chartier, et.al. 1997, Mozhayev, et.al. 2017]. There are perhaps other paradigms that could be considered with the emergence of increasingly reliable, capable and affordable neutron generating systems. For instance, the ability to develop a series of nominally discrete spectra focused around (1) thermal to epithermal energies, (2) heavily-scattered fission energies and (3) fission to D-T energies that could be used to replicate modern workplace fields. Use of such fields, individually or in various combinations, may be more suitable for testing and/or calibration of neutron sensitive detectors utilized in the broad array of U.S. radiological protection programs and may eventually offset the heavy dependence on the limiting energies of ²⁵²Cf, D₂O-moderated ²⁵²Cf and ²⁴¹Am-Be sources.

Conclusions

For nearly 50 years, ²⁵²Cf has served an important role in the calibration and testing of neutron detection devices utilized for radiation protection purposes. While lower-intensity sources of ²⁵²Cf and ²⁴¹Am-Be are perhaps affordable for many such evaluations at lower levels of personal or ambient dose equivalent rates, ²⁵²Cf sources used for high-rate applications require a significant level of capital investment. At present, there are no alternative radionuclide sources that can offer the strong advantages and minimal disadvantages of the ²⁵²Cf isotope.

The development of compact neutron generators has matured to a level where neutron yields rival those of ²⁵²Cf sources and may now present viable potential alternatives for some laboratories. Currently available neutron generators make use of two primary reactions, D-D and D-T. Various generator designs are available, each with their own associated operational and characterization challenges. However, most of these challenges are not new and could be overcome, resulting in the potential for a neutron calibration system that is analogous to those X-ray generators used for evaluation of photon-detection devices.

Nearly monoenergetic reference neutron radiations produced by radionuclide sources and neutron generators have their place in the mandated testing of radiation detection devices to be qualified for use in radiation protection. In the U.S., however, there is a potential for increasing the accuracy of neutron dose estimates for occupational neutron environments. This objective could be attained by implementing shaped spectra to simulate a variety of potential workplace and test sources. Neutron generators appear to be well-suited for such a challenge, but this challenge will require time to evolve into capabilities that best serve the needs of the radiation protection community, including nuclear power plants, industrial users of neutron sources, medical applications, military and researchers in many of these fields. The possible transition to, or augmentation of radionuclide-based neutron sources with neutron generator-produced radiation fields would require establishing requirements for the best-suited neutron spectra to meet the above-noted communities, along with traceable radiological standardization methods and augmenting of consensus standards used to guide evaluations and calibrations of neutron-sensitive instruments and dosimetric devices.

Fig. 4.

Neutron energy distributions resulting from D-T reactions with the neutrons emitted at discrete polar angles relative to the incident ion direction. Solid lines represent the neutron energy distribution of interactions of individual atoms with kinetic energy of 75 keV. Dashed lines indicate corresponding extension of neutron energy distributions for atomic ions accelerated by the potential difference of 150kV.