14 MeV calibration of JET neutron detectors-phase 1: calibration and characterization of the neutron source

PAPER 14 MeV calibration of JET neutron detectors—phase 1: calibration and characterization of the neutron source To cite this article: P. Batistoni et al 2018 Nucl. Fusion 58 026012 View the article online for updates and enhancements. Related content 14 MeV calibration of JET neutron detectors—phase 2: in-vessel calibration P. Batistoni, S. Popovichev, Z. Ghani et al. - Neutron measurement techniques for tokamak plasmas O N Jarvis - Monoenergetic fast neutron reference fields: II. Field characterization Ralf Nolte and David J Thomas - Recent citations Pre-conceptual study of the European DEMO neutron diagnostics M. Cecconello et al - Multiphysics approach to plasma neutron source modelling at the JET tokamak Žiga Štancar et al - Overview of the JET preparation for deuterium–tritium operation with the ITER like-wall E. Joffrin et al - This content was downloaded from IP address 87.218.223.151 on 17/07/2020 at 16:55 1 © 2017 EURATOM Printed in the UK Nuclear Fusion 14 MeV calibration of JET neutron detectors—phase 1: calibration and characterization of the neutron source P.Batistoni1, S.Popovichev2, A.Cufar3, Z.Ghani2, L.Giacomelli4, S.Jednorog5, A.Klix6, S.Lilley2, E.Laszynska5, S.Loreti1, L.Packer2, A.Peacock7, M.Pillon1, R.Price2, M.Rebai8, D.Rigamonti8, N.Roberts9, M.Tardocchi4, D.Thomas9 and JET Contributorsa EUROfusion Consortium, Culham Science Centre, Abingdon, Oxon, OX14 3DB, United Kingdom 1 ENEA, Department of Fusion and Nuclear Safety Technology, I-00044 Frascati (Rome), Italy 2 CCFE, Culham Science Centre, Abingdon, Oxon, OX14 3DB, United Kingdom 3 Reactor Physics Division, Jožef Stefan Institute, Jamova cesta 39, SI-1000 Ljubljana, Slovenia 4 Istituto di Fisica del Plasma ‘P. Caldirola’, CNR, Milano, Italy 5 Institute of Plasma Physics and Laser Microfusion, Hery 23, 01-497 Warsaw, Poland 6 Karlsruhe Institute of Technology, D-76344 Eggenstein-Leopoldshafen, Karlsruhe, Germany 7 JET Exploitation Unit, Abingdon, Oxon, OX14 3DB, United Kingdom 8 Dipartimento di Fisica ‘G. Occhialini’, Università degli Studi di Milano-Bicocca, Milano, Italy 9 National Physical Laboratory, Hampton Road, Teddington, TW11 0LW, Middlesex, United Kingdom E-mail: [email protected] Received 1 September 2017, revised 27 October 2017 Accepted for publication 8 November 2017 Published 21 December 2017 Abstract In view of the planned DT operations at JET, a calibration of the JET neutron monitors at 14 MeV neutron energy is needed using a 14 MeV neutron generator deployed inside the vacuum vessel by the JET remote handling system. The target accuracy of this calibration is ±10% as also required by ITER, where a precise neutron yield measurement is important, e.g. for tritium accountancy. To achieve this accuracy, the 14 MeV neutron generator selected as the calibration source has been fully characterised and calibrated prior to the in-vessel calibration of the JET monitors. This paper describes the measurements performed using different types of neutron detectors, spectrometers, calibrated long counters and activation foils which allowed us to obtain the neutron emission rate and the anisotropy of the neutron generator, i.e. the neutron flux and energy spectrum dependence on emission angle, and to derive the absolute emission rate in 4π sr. The use of high resolution diamond spectrometers made it possible to resolve the complex features of the neutron energy spectra resulting from the mixed D/T beam ions reacting with the D/T nuclei present in the neutron generator target. As the neutron generator is not a stable neutron source, several monitoring detectors were attached to it by means of an ad hoc mechanical structure to continuously monitor the neutron emission rate during the in-vessel calibration. These monitoring detectors, two diamond diodes and activation foils, have been calibrated in terms of neutrons/counts within ±5% total uncertainty. A neutron source routine has been developed, able to produce the neutron spectra resulting from all possible reactions occurring with the D/T ions in the beam impinging on the Ti D/T target. The neutron energy spectra calculated by combining the source routine with a MCNP model of the neutron generator have been validated by the measurements. These numerical tools will be key in analysing the results from the in-vessel calibration and to derive the response of the JET neutron detectors to DT plasma neutrons starting from the response to the generator neutrons, and taking into account all the calibration circumstances. P. Batistoni etal Printed in the UK 026012 NUFUAU © 2017 EURATOM 58 Nucl. Fusion NF 10.1088/1741-4326/aa98f6 Paper 2 Nuclear Fusion IOP International Atomic Energy Agency a See the author list of [31]. 2018 1741-4326 1741-4326/18/026012+24$33.00 https://doi.org/10.1088/1741-4326/aa98f6 Nucl. Fusion 58 (2018) 026012 (24pp) P. Batistoni etal 2 Keywords: neutron calibration, neutron generator, fusion reactor (Some figuresmay appear in colour only in the online journal) 1. Introduction Currently, JET is the only experimental fusion device that can operate with tritium. The JET neutron source intensity, which relates directly to the fusion yield, ranges from ≈108 n s−1 in hydrogen and deuterium ohmic operations to nearly 1019 n s−1 in deuterium–tritium (DT) operations. In order to measure the absolute neutron emission rate over the whole variation range, JET is equipped with several types of neutron detectors (figure 1) [1]: (a) 235U/238U fission chambers (KN1) mounted in moderator packages at mid-plane locations close to the transformer magnet limbs in Octants 2, 6 and 8, and (b) the invessel activation system (KN2) which pneumatically delivers and retrieves capsules containing activation foils to/from ‘Irradiation Ends’ located inside the torus structure, e.g. to the edge of the vacuum vessel. Capsules are delivered before and retrieved after the pulse for counting of the induced gamma radioactivity in the foils. Both KN1 and KN2 need to be accurately calibrated to provide an absolute measurement of the neutron yield and fusion power. A neutron calibration consists of the deployment of a neutron source of known intensity and energy spectrum at different toroidal/poloidal locations inside the JET vacuum vessel to simulate the volume plasma source, and in recording the resulting signals in the JET neutron detectors located inside (KN2) and outside (KN1) the machine. An accurate calibration, at 2.5 MeV neutron energy, of the JET neutron detectors was performed in 2013 where the four systems, KN2 and the three KN1 fission chambers were independently calibrated for JET D plasma operations [1, 2]. A 252Cf source, which emits neutrons with a mean energy of 2.1 MeV, was placed at about 234 different positions, covering the whole JET invessel space. After the calibration, the neutron yields from D plasmas measured by the four systems agree within ±3%, i.e. within the combined uncertainty of the different systems of ~10% [2]. In view of the new DT campaign planned in 2019 [3], a new calibration at 14 MeV neutron energy is needed to allow accurate measurements of the fusion power and of plasma ion parameters. The target accuracy is ±10% to maximise the exploitation of DT operations within the total allowed JET neutron budget. Moreover, the JET 14 MeV neutron calibration has also the objective to benchmark the ITER neutron calibration, where 10% accuracy is required for tritium accountancy [4]. Neutron calibrations at 14 MeV energy are much more challenging than at 2.5 MeV as there are no naturally occurring neutron sources at the relevant energy that could be conveniently employed, and neutron generators have to be used. 14 MeV neutron calibrations have been previously attempted in JET [5] and performed in TFTR [6], overall accuracies of ±10.4% and ±13% were estimated, respectively. At JET a 14 MeV neutron generator is used and deployed inside the vacuum vessel by means of the remote handling system (RH), mounted on one of the two existing RH booms and its ‘MASCOT’(MAnipolatore Servo COntrollato Transistorizzato) robotic arms (figure 2) which can bear a maximum weight of 10 kg. In order to achieve the target accuracy a calibration strategy with different phases has been developed including: (1) technical preparations and selection of a suitable 14 MeV neutron generator (definition of functional requirements, design constraints, personnel safety requirements, RH compatibility, resistance to failure); (2) calibration and characterization of the 14 MeV neutron generator at a neutron metrology laboratory, and analyses and calibration of ‘monitoring detectors’; (3) in-vessel calibration with the neutron generator deployed by the RH system and analyses. The final validation of the calibration and of the related accuracy will be obtained with measurements of neutron yields during DT operations, when the results obtained by the different calibrated detectors will be compared. The constraints and early decisions which defined the main calibration approach, e.g. the RH compatibility and the limited space available, the technical requirements of the 14 MeV neutron generator and the deployment method, the safety and engineering aspects have been discussed in a separate paper [7]. The present paper describes the outcome of the accurate calibration and characterization of the 14 MeV neutron generator selected as the calibration source. In particular, it describes the measurements performed at the Neutron Metrology Laboratory of the National Physical Laboratory (NPL, Teddington, London) using different types of neutron detectors, spectrometers, calibrated long counters and activation foils which allowed us to measure the neutron emission rate and the anisotropy of the neutron generator, i.e. the neutron flux and energy spectrum dependence on emission angle, and to derive the absolute emission rate in 4π sr. The use of high resolution diamond detectors made it possible to resolve for the first time the complex features of the neutron energy spectra resulting from the mixed D/T beam ions reacting with the D/T nuclei present in the target, and to derive information of beam ion composition. As the neutron generator is not a stable neutron source, several monitoring detectors are attached to it by means of an ad hoc mechanical structure to continuously monitor the neutron emission rate and the total neutrons emitted during the invessel calibration. These monitoring detectors, single crystal diamonds and activation foils, located in the same positions at NPL as when in the JET vessel, have been calibrated in terms of (neutrons s−1)/(counts s−1). Since the neutron emission by the neutron generator is anisotropic, and the MASCOT cannot rotate the neutron generator in different directions when inside the JET vessel to Nucl. Fusion 58 (2018) 026012 P. Batistoni etal 3 account for its anisotropy, the neutron emission anisotropy has to be measured. Finally, a neutron source routine has been developed, able to produce the neutron spectra resulting from all possible reactions occurring with the D/T (mono-atomic, bi-atomic etc) ions in the beam impinging on the TiD/T target. The neutron source routine is used in combination with a very detailed MCNP model of the neutron generator which has been valid ated by the measurements. These numerical tools will be key in analysing the results from the in-vessel calibration to derive the response of the JET neutron detectors to plasma neutrons starting from the response to the generator neutrons, and taking into account all the calibration circumstances. 2. The selected 14 MeV neutron generator The 14 MeV neutron generator type ING-17 (figure 3) provided by VNIIA [8] was identified as a suitable source complying with the JET physical and technical requirements. The system consists of a power supply and control unit (PSCU) and a neutron generator (NG), connected by a power supply cable. The VHV (very high voltage) unit is enclosed within the NG, together with the sealed tube containing tritium and deuterium. One PSCU and two NGs were purchased because of the need to avoid delays in the JET programme in the case of a NG failure. The main parameters of the ING-17 are given in table1. In the NGs, a mixed D+ x / T+ y / D x T+ y beam (x, y = 1,2,3 …), with nominally 50% D and 50% T, is accelerated to a nominal energy of 100 keV onto a titanium target containing T/D (nominally 50%/50%) inside a sealed tube thus producing beam-target fusion reactions. 3. The ‘monitoring’ detectors The NG intensity can be subject to fluctuations due to variations in the voltage/current, target heating or general aging of the target and other components. Achieving stability of emission down to a few percent level is still very challenging in present neutron generator technology. Therefore, during the in-vessel calibration the neutron emission intensity, or the NG total neutron yield, needs to be monitored by compact detectors mounted in suitable positions close to the NG. When Figure 1. Left: top view of JET machine showing the position of fission chambers (KN1) close to the magnetic limbs, and of the 3-upper irradiation end (KN2). Right: cross sectionof JET showing the 3-upper irradiation end (KN2). Figure 2. Left: remote Handling MASCOT inside the JET vacuum vessel. Right: the neutron generator deployed by the MASCOT and the power supply control unit (PSCU) attached to the MASCOT body. Nucl. Fusion 58 (2018) 026012 P. Batistoni etal 4 inside the JET vessel, the absolute neutron yield produced by the NG can only be obtained by using the calibrated ‘monitoring detectors’. Multiple monitoring detectors were chosen, a single crystal diamond detector (SDD) and a Si diode, and a set of activation foils, which are attached to the NG at fixed and well-defined positions by means of an ad hoc mechanical support (figure 4). The Si-diode did not prove to be as stable as required and was later replaced by a second SDD. The mechanical support also provides attachment points to allow for the MASCOT arms to grip the NG, the housing of a preamplifier for the active detectors, and two lasers to be used whenever precise positioning of the NG at well-defined locations inside the vessel is required. Two SDDs have been chosen by ENEA to provide the time resolved neutron yield. They are located at nominally symmetrical positions with respect to the target (figure 4). Neutron detection in a SDD is based on the collection of electrons and holes generated in the detector active volume (5.5 eV are needed to produce an e–h pair) by charged particles produced via neutron-induced nuclear reactions on 12C. The main reactions are: (i) the n-α reaction: 12C(n,α)9Be (Q = −5.70 MeV, threshold energy Ethr = 6.18 MeV); (ii) the elastic neutron scattering channel: 12C(n,n′)12C′; (iii) the n-3α reaction (Carbon breakup): 12C(n,n′)3α (Q = −7.27 MeV, Ethr = 7.89 MeV). The first reaction represents the detection channel for 14MeV neutrons. It represents also the spectroscopic channel used in the SDD spectrometer as described in section6.1. In this case, the deposited energy, ideally, equals the incoming neutron energy minus the reaction Q value, i.e. ∼14 MeV– 5.70 MeV = ∼ 8.3 MeV. In the elastic scattering, the energy deposited into the detector, Ed, depends on the recoil angle, i.e. Ed = En · 0.284 · cos2θ and its maximum value can therefore be about 30% of the incoming neutron energy. The third reaction, consisting in the Carbon breakup into 3 alpha particles, gives rise to a continuum up to a shoulder at Ed,max = En − Q. The SDDs used as monitoring detectors and the data acquisition electronics have been supplied by CAEN Company. Both detectors have been realized for CAEN by the Istituto di Struttura della Materia (ISM), Consiglio Nazionale delle Ricerche (CNR), Rome, Italy. ISM-CNR has produced the detectors using ‘electronic grade’ (with Nitrogen concentration [N] < 5 ppb and Boron concentration [B] < 1 ppb) CVD single crystal diamond plates (4.5 × 4.5 mm2, with thickness d = 500 µm), provided by Element Six Ltd [9]. Square Figure 3. Left: the power supply and control unit. Right: the two neutron generators. Table 1. Main characteristics of ING-17 neutron generator. Characteristic Specification Max nominal neutron emission rate 2 × 108 n s−1 Dimensions/weight: Neutron emitting unit ∅70 mm × 459 mm/2.8 kg Power supply and control unit 356 mm × 315 mm × 110 mm/4.6 kg Operation mode Continuous Continuous operation time >20 min Target temperature limit 60 °C Recovery time after switch off <0.5 h Power supply 220 VAC, 50 Hz Power consumption <150 W Tritium content <370 GBq Lifetime: Neutron generator 300 h Power supply and control unit >5000 h Figure 4. Neutron Generator with the mechanical structure needed for MASCOT gripping and to support the ‘monitoring detectors’ and pre-amplifier (red box at the back). The two SDDs and the activation foils are symmetrically hosted in the support around the neutron generator, at the same distance from the target centre. Nucl. Fusion 58 (2018) 026012 P. Batistoni etal 5 200 nm thick multilayer gold finished contacts, 4.2 × 4.2 mm2, are deposited on both faces of a diamond plate which is then mounted in an anodized aluminium casing using an alumina plate holder. A standard SMA connector is used to bias the SDD and to pick-up the signal output, see figure5(a). The first SDD (Dia#0), used since the beginning, was acquired in 2014 and fully characterized at the Frascati 14 MeV neutron generator (FNG) [10]. The second SDD (Dia#1), identical to Dia#0 apart from improved metal casing, was acquired in 2015 and also tested at FNG before being used at NPL. The electronics are standard CAEN catalogue units and include a one channel charge sensitive preamplifier, and a DT5780 dual digital multi channel analyzer (MCA). DT5780 accepts directly pulses from the charge sensitive preamplifier performing a digital trapezoidal shaping on exponential decaying signals. Two HV channels able to supply a bias voltage up to ±0.5 kV, 300 µA and two connectors to power the preamplifier are part of the DT5780. The ‘horseshoe’ activation foil holder surrounding the target can hold 4 Al foils, 4 Fe foils and 8 Nb foils, each 1 mm thick and 18 mm in diameter (figure 5(b)), to measure the four reactions described in table2. All foils are at the same distance from the target centre. The horseshoe assembly is designed in such a way that during the in-vessel calibration it can be conveniently removed at the end of each NG operational period. In particular, the NG can be removed from the vacuum vessel within the Octant 1 boom tent where the foils used can be easily exchanged by remote handling methods at the end of each day. The activation of used foils can then be measured at a remote γ-ray spectrometer and the horseshoe with fresh foils returned to the vacuum vessel ready for the next NG operation period. The activation reactions chosen for the monitoring foils were selected based on numerous requirements: the reaction thresholds should be sufficiently high to discriminate lower energy neutron scatter, and the reaction must be a standard dosimetry reactions with well known cross sections. Finally, the nuclear reactions chosen for NG characterization should also be used during the DTE2 campaign in the JET KN2 diagnostic. Given the above requirements, the intensity of the NG and the available irradiation time, the following nuclear reactions were selected as activation monitors for 14.1 MeV neutrons: 27Al(n,p)27Mg, 56Fe(n,p)56Mn, 27Al(n,α)24Na and 93Nb(n,2n)92mNb. Nuclear data for analysis was taken from the International Reactor Dosimetry and Fusion File (IRDFF, volume 1.05) [11]. Fundamental nuclear data parameters relating to the products of nuclear reactions, the gamma lines, intensities and half-lives were taken from the National Nuclear Data Center, Brookhaven National Laboratory, US. The main parameters for the selected reactions are given in table2. All the activation foils used for the NPL measurement campaigns were of 18 mm diameter and 1 mm thickness, except for niobium foils which were 2 mm thick. The monitoring detectors have to be attached to the NG in the same fixed positions both during the calibration\characterization campaign and during the in-vessel calibration. In this way, the active monitoring detectors can be ‘absolutely calibrated’ in their operating conditions, i.e. their absolute response in the operating position can be derived. However, the monitoring detectors will experience a different neutron spectrum during the in-vessel calibration with respect to that occurring in the calibration campaign due to the presence of neutrons scattered by the vacuum vessel itself. This effect is minimized by the fact that the detectors are very close to the NG, and by the choice of high energy threshold reactions (the 93Nb(n,2n)92mNb activation reaction and, for the SDD, the 12C(n,α)9Be reaction with a threshold at 6.18 MeV). (a) (b) Figure 5. (a) The two ‘monitoring’ diamond detector: Dia#0 (left) and Dia#1 (right). (b) The activation foils in the horseshoe holder. Table 2. Dosimetry reactions selected for the ‘monitoring’ activation measurements. Reaction Energy threshold (MeV) Cross sectionat E = 14 MeV (b) Half-life Isotopic abundance Gamma energy (keV) Gamma decay probability 27Al(n,p)27Mg 3 0.10 9.458 min 1 843.7 0.718 56Fe(n,p)56Mn 4 0.09 2.579 h 0.918 846.8 0.989 27Al(n,α)24Na 5 0.12 14.997 h 1 1368.6 0.999 93Nb(n,2n)92mNb 9 0.46 10.15 d 1 934.5 0.991 Nucl. Fusion 58 (2018) 026012 P. Batistoni etal 6 Nonetheless, the effect will have to be taken into account by neutronics simulations to correct for the different operating conditions. 4. Modelling of neutron generator and source Extensive neutronics analyses are required to derive the JET detectors’ calibration factors related to the plasma neutron source from those measured by deploying the neutron generator inside the vessel, and to take into account the many particular circumstances such as the presence of the RH system. For these reasons, a detailed and validated neutronics model of the NG and a source routine are needed. This is required also to analyse the calibration/characterization measurements at NPL, as discussed in the following sections. 4.1. Neutronics model A detailed MCNP model of the NGs has been first developed using the generic technical documentation provided by VNIIA. In addition, during the NG characterisation campaign, x-ray computed tomography (CT) scans of the generator were carried out at NPL, using a Nikon XT H 225M micro CT system, with a maximum x-ray beam energy of 225 kV. These scans revealed, in 3D, details that were not readily available in the manufacturer’s documentation. The materials information provided by the manufacturer was used in conjunction with the CT scans to create a detailed model of the NG target, backing and housing (figure 6). Furthermore, a very precise MCNP model of the mechanical support, activation monitoring foils/holder, CAEN preamplifier, positioning lasers and monitoring detectors has also been developed, based on CAD models (figure 7). Models of the monitoring diamond detectors were incorporated by extracting their geometry from inverted MCNP models and integrated into the reference NG CAD model. The CAD models of the RH apparatus and monitoring detectors were simplified using SpaceClaim [12]. Overly detailed minute features, such as screw threads and hexagonal bolt heads were rounded, spline surfaces were removed and approximated with conical or spherical surfaces. Any overlaps in geometry (interferences) were fixed and the final model was converted to MCNP Constructive Solid Geometry (CSG) form, using MCAM [13]. Where available, the FENDL 3.0 [14] nuclear data library was used for neutron transport simulations. The full model (NG with mechanical support and monitoring detectors) is used to accurately simulate the neutron field around the neutron source, the neutron fluence and spectrum at the monitoring detectors, the neutron spectra at the DePangher Long counter positions and finally, to predict the activation of all components after operation as a function of irradiation time and cooling time [15]. 4.2. Neutron source routine The simulations of the neutron source were performed using the ENEA-JSI source subroutine [16]. The subroutine extends the standard MCNP6 by adding the ability to simulate the neutron production due to the more general case of the interaction of beam D/T ions slowing down in the target of the NG and the D/T nuclei present there. The subroutine allows also accounting for atomic and molecular species in the beam which result in different effective energies on the target. The stopping power data from SRIM2013 [17] and DT fusion reaction cross-section from ENDF/B-VI.1 [18] were used. Other codes for the source description were tested in the process of the preparation for the use of the NG as a calibration source [19]. The ENEA-JSI source subroutine was selected as it has been extensively validated by the neutronics experiments performed at FNG [20]. The spectrum of neutrons produced by the commercially available DT neutron generators is difficult to predict due to the lack of reliable information about the composition of both the ion beam and the target. The presence of a mixed D/T beam, and of different molecular species, gives rise to Figure 6. Left: CAD model of NG target end. Right: single slice from CT scan showing accelerator body and target/backing. Nucl. Fusion 58 (2018) 026012 P. Batistoni etal 7 different neutron emission components whose relative intensity in the present work has been experimentally determined, as further discussed later in section6. In the source subroutine these source components were simulated separately, using a large number of source events (2 × 108 events) which were written to a file and then sorted according to their direction and energy. In this way SDEF cards describing each of the neutron source components in 400 directional (equidistant in cosine) and 170 energy bins (10 keV wide bins) were generated. Once the relative intensities of the source components were determined, the resulting neutron source was produced as a weighted sum of the components, based on the standard source description card (SDEF card in MCNP) to speed-up the simulations. The DT beam-target reactions are characterised by an energy-angle dependence and anisotropy of the neutron emission intensity due to the reaction kinematics. The latter, being of the order of a few percent in the forward/backward directions for ≈100 keV beam energy, cannot be neglected. The tube components are responsible for a much larger anisotropy profile, with sharp features, as the source neutrons interact with materials surrounding the target. They are also responsible for the presence of neutrons at lower energy resulting from the elastic and inelastic scattering. Figure8 shows the neutron energy spectra of neutrons exiting from the neutron generator at different angles with respect to the beam direction. The shown spectra are calculated for the ING-17 NG assuming, for simplicity, only a 100 keV D beam impinging on the target containing only T. According to calculations, the fraction of neutrons in 4π sr in the energy ranges En > 13 MeV, 1.5 MeV < En < 13 MeV, En < 1.5 MeV amount to 75.6%, 14.5% and 9.9% of the total, respectively (implying that the neutrons with En > 13 MeV are 83.2% of all neutrons in the range E > 1.5 MeV). Also, the neutrons escaping from the neutron generator amount to 1.012 times the neutrons generated in the target due to neutron multiplication reactions. These numerical predictions are validated by neutron spectrum measurements (see section6). 5. NG calibration and characterization campaigns at NPL Two experimental campaigns have been undertaken at the Neutron Metrology Laboratory (NPL) in November 2015 and June 2016 for the calibration/characterization of the two NGs. Different ‘characterization detectors’ (to be distinguished from ‘monitoring detectors’) have been used according to the following strategy: • Characterize/calibrate both NGs: – Measure the neutron energy spectra at different emission angles using a single crystal diamond spectrometer (SDD) in the energy range E > ∼8 MeV, and a NE213 liquid scintillator in the energy range E > 1.5 MeV; – Measure the neutron emission rate as a function of angle (anisotropy profile) using absolutely calibrated long counters and activation foils, and the relative emission rate by monitoring diamond detectors; Figure 7. Left: original CAD model of NG and of mechanical support, including the monitoring detectors. Right: the resulting MCNP model. Figure 8. Calculated neutron energy spectra of neutrons emitted at different angles to the beam direction, due to the D+@100 keV → T reaction. Nucl. Fusion 58 (2018) 026012 P. Batistoni etal 8 – Derive from the above measurements the total neutron emission in 4π sr. • Calibrate the ‘monitoring detectors’ in terms of neutrons emitted by the NG per diamond count (neutrons/counts). • Use the calibration/characterization measurements to validate the neutron source routine and the MCNP model of the NGs. The experimental set up at NPL low-scatter area (24m × 18 m × 18 m) is shown in figure9. Measurements were performed for both NGs during both campaigns, with NGs located at the centre of a platform with rotating arms which allows simultaneous measurements to be performed. These consisted of ‘shots’ where each NG was run for 10, 15, or 20 min with a break in between shots to allow the generator to cool down and to move the instruments characterising the NG to different angles. Due to geometrical restrictions, both NGs had to be turned by 180 degrees about their vertical axis after approximately half of the measurements to realize a full 360 degree measurement. All of the electrical supplies for the NG and measured signals were transmitted using the same cables and scheme envisaged for the in-vessel calibration. The low-scatter area at NPL is associated with a Van de Graaff accelerator and beamlines from this facility and the presence of other detector transport arms limit the angular range of any one arm. Two LCs are available, but the one chosen for the fluence anisotropy measurement was a De Pangher precision long counter. It was chosen in preference to the other ‘home made’ LG because its characteristics are better known for 14MeV neutron detection. The De Pangher LC can move over the angular range from −30° to 135° relative to the axis of the NG when mounted with the Van de Graaff at its back side. For the NG the 0° direction is defined as being along the line of the charged particle beam impinging on the target. The angular range from 150° to −45° was covered by using the same range of low-scatter area angles, but rotating the NG by 180°. 5.1. Differences between campaign#1 and campaign#2 The first campaign at NPL in November 2015 served to test the selected detectors and methodology. The following differences between the first and the second campaigns at NPL are to be noted: – in the first campaign one SDD (Dia#0) and a Si diode were used as active monitoring detectors. As the Si diode was not stable, it was replaced by a second SDD (Dia#1) in the second campaign. – only one mechanical structure was available for the two generators in the first campaign. Moreover, many tests were performed, for example rotating the NGs around their axis to check the poloidal symmetry of the neutron emission. As a consequence, the mechanical structure was removed and re-installed several times. The error in the positioning of the detectors was estimated to be ±1 mm. – the mechanical structure, in particular the part carrying the monitoring detectors was improved in the second campaign to allow for a more reproducible positioning of detectors. Two mechanical structures were available during the second campaign, one for each generator. They were installed at the beginning of the campaign and never removed. – a splitter was used on the CAEN MCA for the NPL team to acquire the diamond counts in 2015. It was removed in the second campaign as it had caused interference with the SDD acquisition system. – NPL foil mounting involved a wheel with hollow aluminium spokes in 2015, whereas in 2016 a disc of art material (polyboard) was used to improve the accuracy of positioning the NPL activation foils. The final calibration data were obtained mainly in the 2016 campaign. Figure 9. Experimental set up at NPL in 2015. Nucl. Fusion 58 (2018) 026012 P. Batistoni etal 15 The direct counts were obtained by subtracting the scatter counts from the total and the uncertainties combined in quadrature. Dead-time corrections were made using previously determined values for the dead-time per BF3 counter pulse. Two uncertainty components were added to the counting statistics for the scatter counts. The measured scatter contribution at angles other than 0° are probably too large because the LC sees neutrons scattered in un-shadowed parts of the NG body which should be part of the direct component and hence should not be subtracted. For example, the measured scatter at 90° was 20% compared to 16% at 0°. To allow for the uncertainty caused by this effect the scatter subtraction was chosen to be mid way between the minimum possible percent age value, 16% measured at 0°, and the measured value at a particular angle, and an uncertainty was assigned that was large enough to just cover both extremes. When performing shadow cone measurements at NPL an uncertainty comp onent of 5% is assigned to the scatter counts in addition to the uncertainty derived from statistics. This is to allow for inexact shadowing as the alignment and positioning of the shadow cone can never be perfect. For most neutron fields produced at NPL this 5% contribution to the uncertainty in the scatter subtraction is unimportant because this correction is usually much smaller than in the present case, where the LC at 300 cm was further from the source of neutrons than usual, and the source was producing high-energy neutrons almost isotropically thus producing a significant scatter comp onent. For the present measurements this 5% uncertainty in the scatter was in most cases the largest contributor to the final uncertainties in the direct counts. These varied between 1.2% and 1.4% for angles between 0° and 150° although they increased on approaching 180°. Because the shadow cone has to be close to the source of neutrons, the mounting arrangements of the NG meant that shadow cone scatter measurements were not possible at 180°. Estimates for the scatter were therefore made by interpolating the data for scatter counts per diamond count as a function of angle and also by interpolating the scatter/total values. The predictions of the two approaches did not agree well so a scatter correction corresponding to the mean from these two approaches was adopted with an uncertainty large enough to cover both values. This resulted in an uncertainty in the direct counts per diamond count of about 40%. This is largely due to the direct count being so small at this angle (about 14% of that at 0°), and this is because the neutrons have to traverse the full length of the NG. The contribution of the neutrons at this angle (and also that at 0°) to the total emission into 4π sr is, however, very small because the solid angle about these two angles is small. The fluence at 300 cm from the NG target was calculated from the LC direct counts using the spectrum averaged LC efficiencies and effective centres. Corrections were made for small variations in the target to LC distance as the angle varied, and for air out-scatter. Uncertainties in these corrections were added in quadrature with the statistical uncertainties in the direct counts. One additional uncertainty component was added to allow for uncertainty in the LC angle. Values for the rate of change of the fluence with angle ( dΦ/dθ ) were taken from the MCNP calculations. The final values for the random uncertainties in the fluence (excluding the systematic uncertainty in the LC efficiency and effective centre) were between 1% and 2% in the majority of cases although they increased approaching 180° due to the uncertainty in the scatter correction at this angle. The calculations using the MCNP model of the NG provide values for the fluence at 300 cm per neutron emitted from the target. These values were normalised to the measured values using a normalisation factor N derived by taking a weighted mean of the measured/calculated ratios for all angles, n, where there were both measurements and calculations: N= n i=1( M i/ C i)/(σ 2 Ci+σ 2 Mi )  n i=1 1/(σ2 Ci +σ2 Mi ) (1) where: Mi is the measured fluence Ci is the calculated fluence σMi is the uncertainty in the measured fluence σCi is the uncertainty in the calculated fluence Figure 22. Measured fluence per diamond monitor count (Dia#0) at 300 cm from the neutron-producing target compared to normalised calculated values for NG#1 (left) and NG#2 (right). Nucl. Fusion 58 (2018) 026012 P. Batistoni etal 16 The ratio for 180° was excluded from this process because of large uncertainties in both the measurements and the calculations. Values for the measurement uncertainty, σMi, were derived as explained above. Those for the calculations are difficult to estimate. The statistical component is small, but a component must be present for inaccuracies in the model. Various options were tried for this component and to test them a χ2 value was determined from: χ2= n  i=1 (N·Ci−Mi)2/(σ2 Ci+σ2 Mi) . (2) Values for χ2 per degree of freedom (χ2/(n − 1)) close to unity implies that the combined uncertainties are reasonably correct. For angles near 180° the model is likely to be inaccurate because of uncertainties about the construction and the effect of the cable attached to the NG at this point and uncertainties of 5%, 3% and 2% were therefore assigned for the calculations for angles of 165°, 160° and 150° respectively. Values for χ2 close to 1 were obtained when all the other angles were assigned an uncertainty of 1% for the calculations implying uncertainties of this order in the calculated shape of the angular distribution at angles away from 180°. An examination of the available data revealed differences for the fluence per Dia#0 between 2015 and 2016, and between NG#1 and NG#2. Information on the ratios 2016/2015 and NG#1/NG#2 were available from, (a) data where LC results were available at a particular angle for both campaigns and/or for both NGs, (b) the factors derived to normalise the MCNP calculations to the measurements, (c) ratios of slab monitor counts to diamond counts. The slab monitor is a high-sensitivity neutron detector positioned permanently on the wall of the NPL low-scatter area. All three measurements gave very similar results and these are summarised in table3. The tabulated uncertainties are based on the uncertainties in the individual measurements and the spread of results between the three approaches. The differences between 2105 and 2016 and between NG#1 and NG#2 were assumed to be due in part to positioning of the diamond monitor, which is very close to the target and the readings are thus very sensitive to the exact positioning, and in part to other differences described in section5.1. Because of the difference between the 2016 and 2015 results, the data from 2015 were normalised by the average ratio to 2016 data for plotting in figure22. After the normalisation between the 2015 and 2016 data for the shape of the variation of fluence with angle and the absolute values are in very good agreement. Several conclusions can be drawn from figure 22. Data from both NGs show very similar features. The fluence also appears, as expected, to be symmetric in that results at negative angles (to the left of the 0° line) agree with those at positive angles (to the right of the 0° line). The overall shape is very well reproduced by MCNP calculations normalised to the experimental data and can be explained qualitatively by considering the construction of the NG. The target is mounted on a solid cooling block so neutrons emitted with angles less than ±90° are attenuated in the metal of this block. This attenuation is not present for angles greater than ±90° so the fluence increases only to decrease again as the angle approaches 180° because of attenuation along the length of the neutron generator. It must be noted that, according to MCNP calculations, the neutrons detected by LC are about 1.2% higher than neutrons produced in the target in DT reactions. Taking this correction into account, at the beginning of the first campaign the total neutron production rate into 4π sr was calculated to be about 2.6 × 108 s−1 for NG#1 in agreement with the manufacturers estimate. When the output fell by about 25% during a shot near the end of the 2015 campaign, the fluence per diamond monitor count remained constant. At the beginning of the 2015 campaign the total neutron emission of NG#2 was similar, i.e. about 2.5 × 108 s−1, and dropped during the two campaigns but only by about 12%. Values for the total NG emission into 4π sr were obtained by integrating the fluence values measured as a function of angle by the LC. For an anisotropically emitting neutron source, the neutron fluence φ(r,θ,α) , at a distance r, and at angle Ω(θ,α) , is related to the total source emission, B, by the double integral: B= 2π α=0π θ=0 φ(r,θ,α)r2sin(θ)dθd α (3) where the angle θ is in the horizontal plane and the azimuthal angle α is in the vertical plane. If the neutron emission is independent of the azimuthal angle, i.e. the source has rotational symmetry about its cylindrical axis of the NG, then the following simpler relationship holds: B=2πr2 π θ=0 φ(r,θ) sin(θ)dθ . (4) This simpler relationship was assumed to apply in the present situation. The integral was performed by using a numerical algorithm integration routine, or by using an Excel spreadsheet to calculate: B=2πr2 n  i=1 φi{[cos(θi)−cos(θl)] + [cos(θh)−cos(θi)] } (5) where θi is the angle at which the fluence was measured θl is an angle half way between θi−1 and θi θh is an angle half way between θi and θi+1. Both approaches gave very similar results. The uncertainty varied depending on the number of angles measured. The number of points for NG#1 in 2016 and NG#2 in 2015 were Table 3. Ratios of fluences per Dia#0 for 2016/2015 and NG#1/ NG#2. Quantity Ratio 2016/2015 from NG#1 data 1.074 ± 0.004 2016/2015 from NG#2 data 1.084 ± 0.005 NG#1/NG#2 from 2015 data 1.056 ± 0.007 NG#1/NG#2 from 2016 data 1.052 ± 0.004 Nucl. Fusion 58 (2018) 026012 P. Batistoni etal 17 limited and the uncertainties when integrating the measured data were thus higher. In view of the good agreement between measurements and calculations for the shape of the angular dependence of the fluence the total fluence into 4π sr was also derived by integrating the MCNP calculated data after normalising to the measurements. The different approaches gave very slightly different answers. This variation was taken as an indication of uncertainty introduced by the numerical integration process, and this was of the order of 1%. To this uncertainty must be added the systematic uncertainties, the predominant ones being those associated with the LC efficiency and effective centre. Other systematic components such as for air-out-scatter, distance, dead-time, or LC stability are almost negligible when added in quadrature. The final estimated uncertainty in the total emission per diamond detector count is 3%. The results are presented in table4. The diamond count rates in table4 are the average values over all the shots used in deriving the total neutron emission for a particular configuration. Shot to shot variations were sometimes quite large. For example, in 2015 the diamond count rates for NG#1 varied by more than 10% between the highest and lowest values. Day to day variations tended to be smaller. For NG#2 in 2015 the difference between the highest and lowest days was 3.4%. From data taken in 2016 the average ratio of Dia#0/ Dia#1 over all shots was 1.1700 ± 0.0049 for NG#1 and 1.1732 ± 0.0027 for NG#2 and these ratios can be used to derive neutron emission per Dia#1. 9. Activation measurements 9.1. Monitoring activation measurements The monitoring activation foils located on the mechanical support provide a complementary and independent measurement of the absolute neutron emission rate (n · s−1) for a given exposure period [27]. The foils were normally installed every morning and removed after the ninth irradiation cycle every day to allow the gamma spectrometry measurements to be started at a practical time. The main parameters for the selected reactions are given in table2. The neutron-induced radioactivity in a foil material can be expressed as: A=YnRNT(1−exp (−λtA)) (6) R = ∞ 0 φ(E)σ(E)dE=φ(E)σ(E)  (7) where: A is the activity of a particular radionuclide induced by neutron activation (Bq), Yn is the neutron yield (n · s−1), R is the reaction rate (reaction·s−1), NT is the number of target nuclei in the sample, λ is the decay constant of the activation product, tA is the activation time, ϕ(E) is the neutron energy spectrum at the foils (cm−2 s−1 eV−1), and σ(E) is the reaction cross section(cm2). The radioactivity after the series of subsequent irradiations each with irradiation time tAi (tAi ≠ tAi+1) and the subsequent cooling time tCi (tCi ≠ tCi+1) is expressed by following equation: A n= Y n/ t · N T·φ( E )·σ( E )·  i B i( 1 −exp (−λ· t Ai)) ·exp (−λ· t Ci ) (8) where t is the total irradiation time, Bi is the relative amplitude of YN, recorded by the Dia#0 during subsequent NG pulses of duration tAi. Each aluminium sample was measured twice. The activity of 27Mg and 24Na was in the range of 269–450 Bq·g−1 and 25–99 Bq·g−1, respectively. Between the two aluminium measurements, the 56Mn from the iron samples was detected. 56Mn activity was in the range of 51–141 Bq·g−1. The 92mNb activity, usually measured during the night, varied over the range of 1.64–7.11 Bq·g−1 The MCNP-calculated neutron reaction rates at the foil locations and the measured radioactivity were used to derive the neutron emission rate. The calculated reaction rates per source neutron are as follows: 8.34 · 10−5 s−1 for 27Al(n,p)27Mg, 1.24 · 10−4 s−1 for 27Al(n,α)24Na, 1.15 · 10−4 s−1 for 56Fe(n,p)56Mn, and 4.45 · 10−4 s−1 for 93Nb(n,2n)92mNb (note that these are normalised to one neutron produced in the target). The neutron emission rates derived for the four reaction products for each day of the experimental campaigns are presented in tables5 and 6 and in figure23. The total uncertainty in the neutron emission rate was calculated based on the quadratic sum of different contributions. They include: Table 4. Neutron emission rate for NGs into 4π sr per diamond count rate (peak ROI Dia#0). N.B. the emission rates are for all neutrons emitted from the NG which according to the MCNP calculations corresponds to 1.012 times the number from the target. Configuration Total neutron emission per Dia#0 count Average Dia#0 count rate (s−1) Average total emission rate (s−1) NG#1 in 2015 (1.011 ± 0.030) × 10726.41 ± 0.05a(2.672 ± 0.080) × 108 NG#1 in 2016 (1.090 ± 0.032) × 10719.32 ± 0.16 (2.106 ± 0.065) × 108 NG#1 all data normalised to 2016 (1.094 ± 0.033) × 107 NG#2 in 2015 (9.67 ± 0.029) × 10626.25 ± 0.17 (2.538 ± 0.077) × 108 NG#2 in 2016 (1.035 ± 0.031) × 10723.90 ± 0.13 (2.474 ± 0.075) × 108 NG#2 all data normalised to 2016 (1.038 ± 0.031) × 107 a Data are for measurements performed before the decrease in the neutron emission rate at the end of the 2015 campaign. Nucl. Fusion 58 (2018) 026012 P. Batistoni etal 18 – the uncertainty on the activity measurement: 8.8% for 27Mg, 8.7% for 24Na, 8.7% for 56Mn, 8.7% for 92mNb. – the uncertainty on the number of target nuclei, equal 0.39% for Al samples, 0.12% for Fe samples and 0.06% for Nb samples. – the uncertainty of the sums in equation(4) amounting to Mg: 1.51%, Na: 1.16%, Mn: 1.16%, Nb: 1.16%. The relatively high uncertainty of magnesium is due to its short half-life. – the uncertainty on the reaction rates which is less than 0.1%. The uncertainties of cross sectionsin the activation calculation were not considered. Discrepancies between reactions have been observed. These discrepancies are within the combined uncertainties of the measurements and can occur for measurements taken on the same day. The standard deviation for the daily average neutron emission rates for NG#1 was within the range of 2.3–6.1%. For NG#2 the standard deviation was in the range of 4.6–5.9%, where more than 5% was observed only on the second day of the 2016 NPL campaign. Finally, the absolute measurements provided by the monitoring activation foils are compared to the neutron yield measured by the NPL Long Counters. The comparison is shown in tables5 and 6 and in figure24. Once the 1.012 correction for neutrons from the NG (rather than the target) is included, the two methods are in agreement within 5.8% (if the first two days in NPL1 are not considered) for NG#1 and within 4.0% for NG#2, within the combined uncertainty of the two methods. Table 5. Mean values for neutron emission rate from target of NG#1. All rates are in 108 n s−1. 27Mg Unc (%) 24Na Unc (%) 56Mn Unc (%) 92mNb Unc (%) Daily average SD (%) Mean rate from LC Activation/ LC ratio 2015 Day1 2.61 8.2 2.96 8.1 2.78 8.0 2.98 8.8 2.83 6.1 2.57 1.101 Day2 2.68 8.2 2.90 8.1 2.60 8.1 2.91 8.7 2.77 5.7 2.56 1.081 Day3 2.67 8.2 2.73 8.1 2.54 8.0 2.71 8.7 2.66 3.2 2.60 1.025 Day4 2.74 8.2 2.55 8.2 2.56 8.0 2.71 8.7 2.64 3.8 2.63 1.005 Day5 2.67 8.2 2.73 8.1 2.59 8.0 2.71 8.7 2.68 2.3 2.65 1.009 Day8 2.69 8.2 2.90 8.1 2.70 8.0 2.86 8.7 2.79 3.8 2.65 1.050 Day9 2.56 8.0 2.79 8.8 2.67 6.0 2.53 1.058 2016 Day1 1.93 8.2 2.13 8.1 1.99 8.2 2.12 8.7 2.04 4.9 2.02 1.010 Table 6. Mean values for neutron emission rate from target of NG#2. All rates are in 108 n s−1. 27Mg Unc (%) 24Na Unc (%) 56Mn Unc (%) 92mNb Unc (%) Daily average SD (%) Mean rate from LC Activation/ LC ratio 2015 Day6 2.40 8.2 2.52 8.1 2.27 8. 2.32 8.7 2.38 4.6 2.44 0.98 Day7 2.58 8.2 2.59 8.1 2.36 8.0 2.41 8.7 2.49 4.6 2.51 0.99 2016 Day2 2.31 8.2 2.62 8.1 2.35 8.0 2.49 8.7 2.44 5.9 2.54 0.96 Day3 2.21 8.2 2.45 8.1 2.27 7.9 2.38 8.7 2.33 4.6 2.38 0.98 Figure 23. Neutron emission rates derived from the four activation measurements and daily average (horizontal bars) for each day of the experimental campaigns. Nucl. Fusion 58 (2018) 026012 P. Batistoni etal 19 9.2. NPL activation measurements Measurement of the neutron output from the NG was performed by irradiating thin iron and aluminium foils of diameter 25.4 mm. They are routinely used at NPL for measuring neutrons from the d + T reaction in the energy range 14–20 MeV, and have been used in international neutron fluence comparison exercises [26, 28, 29]. The foils were held in paper attached to lightweight aluminium holders mounted in an aluminium ring for the 2015 campaign and in a foam board disc for the 2016 campaign (as shown in figure25) at approximately 15 cm from the target of the NG. Some foils were also irradiated in contact with the surface of the NG, wrapped in cigarette paper (see figure26). Foils were positioned at 15° steps from −120° to +120° with 0° corresponding to the direction of the deuterium–tritium beam in the NG. The distance of each foil from the surface of the NG was measured using a measuring rod with the foil at either the 0° or ±90° position, before rotating the aluminium ring or foam board disc to set the foil at the desired angle. The height of each foil was set using the laser installed on the laboratory wall. Foil irradiations ranged from 1 to 6 shots in length (not the same as for the monitoring activation foils), depending on which reaction was being used and the position of the foil. In total 29 aluminium and 52 iron foils were irradiated over the Figure 24. Comparison of neutron emission rates as derived from the Dia#1 calibrated by the NPL Long Counters (squares and triangles) and those derived by the monitoring activation measurements (daily average values, full circles). Figure 25. Mounting arrangement of the foils in the 2015 (left) and the 2016 (right) campaign. Figure 26. Mounting of foils on surface of NG. Figure 27. MCNP model of NG and NPL activation foils around the head of the neutron generator. Nucl. Fusion 58 (2018) 026012 P. Batistoni etal 20 two campaigns. The start time and stop times of each irradiation were taken from the NPL scaler data file recording the LC counts, the PC clock having been checked against the speaking clock periodically during the two campaigns. The activities of the foils were measured as soon as possible after the end of the irradiation. Four separate highly efficient gas flow proportional counters were used to determine the activities of the foils from the beta particles emitted in the decay. The counters use a 90:10 Ar/CH4 gas mixture and two out of the four used have anti-coincidence rings to suppress the background. Foils were placed on gold-coated VYNS plastic mounts and counted for cycles of a fixed time period until sufficient β counts had been recorded to give an acceptably low statistical uncertainty. Where possible the foils were also counted in more than one counter to provide an additional measurement of the activity. The background rates of the counters were measured before and after each measurement campaign. Corrections are made for the counter’s additional sensitivity to gamma rays from the foil (K-correction), for variations in the beam flux during the irradiation, for the non-saturation of the activity and for the decay since the end of the irradiation. A least squares fitting program (MHLIFE) was used to analyse the decay data. The program calculates the saturated β count rate, Nβ(sat) (i.e. that which would be obtained during an infinitely long irradiation at the mean fluence rate). Each β-count cycle is first dead-time corrected, then decay corrected to the end of the irradiation. The decay correction is performed by considering the following integral expression for the counts in each cycle, C: C = t 2 t1 Nβe−λtd t (9) where: t1 is the time from the end of the irradiation to the start of the cycle, t2 is the time from the end of the irradiation to the end of the cycle, Nβ is the β count rate at the end of the irradiation λ is the decay constant of the foil activity. Solving equation(9) yields the following expression: Nβ= Cλ ( e− λt1 −e− λt2) . (10) Values for Nβ for each β counting cycle are obtained from equation (10). The β count rate at saturation, Nβ (sat), is related to the β count rate, at the end of the irradiation, Nβ, by: N β(sat)= N β (1 − e − λT) (11) where T is the length of the irradiation, including breaks between shots. Another program (NVVARY) analysed the monitor data to calculate the small correction factor for the variation of the fluence rate during the irradiation, FT. The saturated activity, N0, is related to the β count rate at saturation, Nβ(sat) by: N 0= N β(sat) ε β FTFa(d) T TNG (12) where: ε β is the effective β-efficiency ε β=ε β (1+K) (allowing for the γ-ray sensitivity of the β−detector by means of the K-correction), Table 7. Parameters used in the analysis of the foils. Al(n,p)27Mg Al(n,α)24Na Fe(n,p)56Mn Nominal thickness (mm) 0.3 0.6 0.3 0.6 0.1 Mass range (mg) 395–411 811–818 395–413 811–818 369–424 εβ range 0.7837–0.7738 0.6202–0.6160 0.7337–0.7232 0.5382–0.5359 0.7270–0.7125 (1 + K)1.002 1.005 1.007 1.017 1.016 λ (s−1) [30]1.2209 × 10−31.2866 × 10−57.4664 × 10−5 Table 8. Mean values for neutron emission rate from target of NG#1. 27Al(n,p) × 108 n s−1 27Al(n,α) × 108 n s−1 56Fe(n,p) × 108 n s−1Mean × 108 n s−1 Long counter × 108 n s−1 2015, pre decrease 2.54(11) (7 foils) 2.68(10) (13 foils) 2.53(9) (13 foils) 2.59(9) 2.64(8) 2015, post decrease — — 2.11(12) (2 foils) 2.11(12) 2016 1.95(14) (3 foils) 2.13(8) (2 foils) 2.01(10) (4 foils) 2.01(9) 2.08(6) Table 9. Mean values for neutron emission rate from target of NG#2. 27Al(n,p) × 108 n s−1 27Al(n,α) × 108 n s−1 56Fe(n,p) × 108 n s−1Mean × 108 n s−1 Long counter × 108 n s−1 2015 —2.58(12) (4 foils) 2.31(8) (12 foils) 2.38(9) 2.51(8) 2016 —2.43(10) (4 foils) 2.27(9) (7 foils) 2.33(9) 2.44(7) Nucl. Fusion 58 (2018) 026012 P. Batistoni etal 21 FT is the fluence rate variation correction factor, Fa(d) is the self-shielding factor for a foil, thickness d, TNG is the time during the irradiation for which the NG was switched on. For the foil thicknesses used the self-shielding is negligible, hence Fa(d) = 1, and the effect of the front foils shielding the back ones is also negligible where two foils were placed together in a mount. In the standard approach in use at NPL the total neutron fluence, Φ, is derived from the saturated foil activity, N0 . However, because of the extended geometry of the NG, the variation in the energy spectrum with angle, and that the main purpose of the exercise was to validate the MCNP model of the NG, it was considered more appropriate to compare the measured saturated activities of each foil with the calculated reaction rate for each foil obtained from the MCNP simulation, R, normalised to the mass of the modelled foil (in mg) and to the total number of neutrons emitted by the NG into 4π sr, Q, derived from the model. All the foils were added to a single MCNP input file, the geometry of which is shown in figure27. The measured activity of each foil was divided by the mass of the foil (in mg), m, and the mean count rate in Dia#0, for the time during the foil irradiation for which the NG was switched on. When divided by the MCNP calculated reaction rate, each activity yielded a value for the total number of neutrons emitted by the NG into 4π sr per Dia#0, as follows: Q D = N 0 m T NG D 1 R (13) where D is the total count in Dia#0 during the foil irradiation. No correction was made for room and air scattered neutrons and there are two reasons why this correction was deemed unnecessary. Firstly, the target to detector distance was small so that the scattered fluence component was only a very tiny fraction of the direct, and secondly the fact that the cross sectiondecreases rapidly below 13 MeV, going to zero at about 6 MeV, means that the detection efficiency for scattered neutrons is low. The parameters used in the analysis of the foils are given in table7. Total emission rates were derived from each foil using the MCNP calculations of reaction rates. The spread of the points is quite large (about ±10%) and this is believed to be primarily due to the positioning of the foils. As they were close to the NGs small changes in the distance or angle of the foil could make a significant difference to the activity, par ticularly for foils which were mounted for several shots where they may have been disturbed while other equipment was moved and checked between shots. Foils which were attached to the surface of each NG give an even greater spread of emission rate values and this is probably due to their closeness to the source of neutrons making them even more sensitive to position. The mean values for the neutron emission rates from the targets of the NGs are given in tables8 and 9, with the 2015 values for NG#1 split into pre- and postthe large decrease in output that occurred near the end of the first campaign. Values derived from the long counter measurements are included for comparison. Note that the LC data from table 4 have been reduced by a factor of 1.012 for comparison with the foil Figure 28. Neutron emission per Dia#0 count for the 27Al(n,p)27Mg reaction (left) for the 27Al(n,α)24Na reaction (right). Figure 29. Neutron emission per Dia#0 count for the 56Fe(n,p)56Mn reaction. Open points correspond to foils on the surface of the NG, positions indicated by the corresponding labels. Nucl. Fusion 58 (2018) 026012 P. Batistoni etal 22 data which is per neutron from the target rather than from the whole NG body. The total uncertainties in the reported foil emission rates include components for the following: – distance from NG to foil (±2 mm) which equates to ±2.67% – foil parameters m (±0.06%), εβ (±0.2%), 1 + K (±0.1–0.2%), λ (±0.1%), cross section(±0.5%) – R as calculated from the model of the NG (±2%) - background in the foil beta counters (±0.33%) – timing of length of irradiation and time to start of foil counting (±0.11–0.35%) – correction for flux variations FT (±0.0-0.5%) – statistical uncertainty in N0/R from standard error on the mean. The emission rate per Dia#0 count rate was determined for each angle and reaction by averaging results where more than one foil of the same type had been irradiated at the same angle (taking positive and negative angles to be equivalent) for the same NG and during the same campaign. These are plotted in figures28 and 29, with the error bars corresponding to the statistical error in the foil counting or the standard deviation where more than one foil of the same type was irradiated at the same angle. There is no apparent trend in the emission rate as a function of angle indicating that the MCNP model has the correct angular distribution of neutrons from the NG. Mean values of the neutrons emitted from the target of each NG per Dia#0 for each reaction are given in tables10 and 11, excluding those foils placed on the surface of the NG. Values from 2015 have been normalised to the 2016 campaign using the overall mean of Slab 1/Dia#0, Slab 2/Dia#0 and LC/ Dia#0 (1.080). The overall mean value for NG#1 is 1.7% lower than the LC value once the 1.012 correction for neutrons from the NG (rather than the target) is included. For NG#2 the foil value is 4.6% lower than the LC value. 10. Conclusions Two 14 MeV neutron generators have been calibrated and characterised during two experimental campaigns at the NPL Neutron Metrology Laboratory in view of using them for calibrating the JET neutron detectors. In parallel, a detailed MCNP model of the neutron generators has been developed together with a neutron source routine capable of describing the neutron production from fusion reactions by D, T beams impinging on solid targets containing T, D. The results are summarised in the following: • The neutron energy spectra of both generators have been measured at different emission angles using a single crystal diamond spectrometer (in the energy range E >8.7 MeV) and a NE213 liquid scintillator (in the energy range E > 1.5 MeV). Several components were observed in the measured spectra, due to D+, T+, D+ 2 , T+ 2 , DT+ beam monoatomic and molecular species reacting with D, T in the target. The measured spectra could be very well reproduced by the numerical simulations at all emission angles with a beam acceleration voltage equal to 73 keV, with beam composed of ~80.44% DT, ~8.95% D2, 7.32% T2, 2.66% T and ~0.63% D, and assuming that in the target there is the same quantity of deuterium and tritium. • The neutron emission rate has been measured for both generators as a function of angle (anisotropy profile) using absolutely calibrated long counters available at NPL, and the relative emission rate by monitoring diamond detectors. The anisotropy profiles are identical for both generators and could be very well produced by numerical simulations within about ±1%. This statement is made on the basis of a χ2 test on the difference between the normalised calculated fluence and the measured values which gave a χ2 per degree of freedom close to unity when using the measured uncertainties and an uncertainty of 1% on the calculations—see section8. Table 10. Mean values for neutrons emitted from target of NG#1 per Dia#0, normalised to 2016 campaign. 27Al(n,p) × 107 neutrons/count (Dia#0) 27Al(n,α) × 107 neutrons/count (Dia#0) 56Fe(n,p) × 107 neutrons/count (Dia#0) Mean × 107 neutrons/ count (Dia#0) Long counter × 107 neutrons/count (Dia#0) 2015 1.051(45) (7 foils) 1.102(40) (13 foils) 1.037(38) (15 foils) 1.064(39) 1.076(34) 2016 0.979(46) (3 foils) 1.109(58) (2 foils) 1.038(46) (4 foils) 1.034(43) 1.077(32) Mean 1.030(43) 1.103(40) 1.037(37) 1.058(37) 1.076(33) Table 11. Mean values for neutrons emitted from target of NG#2 per Dia#0, normalised to 2016 campaign. 27Al(n,p) × 107 neutrons/count (Dia#0) 27Al(n,α) × 107 neutrons/ count (Dia#0) 56Fe(n,p) × 107 neutrons/count (Dia#0) Mean × 107 neutrons/count (Dia#0) Long counter × 107 neutrons/count (Dia#0) 2015 —1.053(51) (4 foils) 0.964(34) (12 foils) 0.986(37) 1.026(31) 2016 —1.014(38) (4 foils) 0.942(36) (7 foils) 0.968(37) 1.023(31) Mean —1.033(40) 0.956(34) 0.979(36) 1.024(31) Nucl. Fusion 58 (2018) 026012 P. Batistoni etal 23 • From the above combined measurements of the long counter s and monitoring diamond detectors, the total neutron emission in 4π sr has been derived. It was about 2.6 × 108 s−1 for NG#1 at the beginning, in agreement with the manufacturer’s estimate, but it fell by about 25% during a shot near the end of the 2015 campaign. At the beginning of the 2015 campaign the total neutron emission of NG#2 was similar, i.e. about 2.5 × 108 s−1, and dropped during the two campaigns but only about 12%. According to numerical calculations, these correspond to 1.012 times the neutrons produced in the target by fusion reactions. • The ‘monitoring diamond detectors’ were calibrated against the NPL long counters in terms of neutrons emitted by the NG per count. The resulting diamond calibration factors were derived within ±3%, and are summarised in table4. • The absolute measurements of the neutron emission rates by the monitoring activation foils have been found to be in agreement with those derived by the calibrated diamond detectors within 4%–5.8% and 4.0% for NG#1 and NG#2, respectively. • The absolute measurements of the neutron emission rates by the NPL activation foils have been found to be in agreement with those derived by the calibrated diamond detectors within and 2% and 4.6% for NG#1 and NG#2, respectively. • The neutron source routine and the MCNP model of the NGs have been fully validated by the calibration/characterization measurements. We conclude that, the absolute neutron emission rate during the JET in-vessel calibration could be derived from the diamond detectors with a total uncertainty equal to ±3.1% (uncertainty on calibrated LC + uncertainty on diamond (n,α) peak count rate), and from monitoring activation measurements with a total uncertainty equal to ±5.9%, provided that no other sources of uncertainties arise in the much more challenging environment due to the deployment of the NGs and their equipment by the RH system inside the JET torus. The neutron source routine and the MCNP model of the NGs can be reliably used in the analysis of the in-vessel calibration at JET. At the end of 2016 campaign, the two neutron generators with the mechanical structures and all equipment were put back without any modification in their ad-hoc containers and stored ready for the in-vessel calibration at JET. Acknowledgments This work has been carried out within the framework of the EUROfusion Consortium and has received funding from the Euratom research and training programme 2014-2018 under grant agreement No 633053. The views and opinions expressed herein do not necessarily reflect those of the European Commission. References [1] SymeD.B. etal 2014 Fusion yield measurements on JET and their calibration Fusion Eng. Des. 892766–75 [2] BatistoniP., PopovichevS., ConroyS., LengarI., CufarA., AbhangiM., SnojL., HortonL. and JET Contributors 2017 Calibration of neutron detectors on the Joint European Torus Rev. Sci. Instrum. 88 103505 [3] BatistoniP. etal 2016 Technological exploitation of deuterium–tritium operations at JET in support of ITER design, operation and safety Fusion Eng. Des. 109–11278–85 [4] SasaoM., IshikawaM., YuanG., PatelK., JakharS., KashchukY. and BertalotL. 2013 Issues on the absolute neutron emission measurement at ITER Plasma Fusion Res. 82402127 [5] JarvisO.N. etal 1985 Further calibrations of time-resolved neutron yield monitor (KN1) JET Joint Undertaking Internal Report JET-IR(85)06, JET Joint Undertaking [6] JassbyD.L., BarnesC.W., JohnsonL.C., RoquemoreA.L., StrachanJ.D., JohnsonD.W., MedleyS.S. and YoungK.M. 1994 Absolute calibration of TFTF neutron detectors for dt plasma operation Princeton Plasma Physics Laboratory Report, Princeton Plasma Physics Laboratory [7] BatistoniP. etal 2017 Technical preparations for the in-vessel 14 MeV neutron calibration at JET Fusion Eng. Des. 117107–14 [8] www.vniia.ru [9] www.e6.com/en/Home [10] PillonM., AngeloneM., BatistoniP., LoretiS. and MiloccoA. 2016 Characterization of a diamond detector to be used as neutron yield monitor during the in-vessel calibration of JET neutron detectors in preparation of the DT experiment Fusion Eng. Des. 10693–8 [11] www-nds.iaea.org/IRDFF/ [12] SpaceClaim 16.2 (http://spaceclaim.com) [13] WuY. and FDS Team 2009 CAD-based interface programs for fusion neutron transport simulation Fusion Eng. Des. 841987–92 [14] www-nds.iaea.org/fendl3/SLib.html [15] GhaniZ. etal and JET Contributors Characterisation of a neutron generator and monitoring detectors, to be used for the in vessel calibration of JET Fusion Eng. Des. submitted [16] MiloccoA., TrkovA. and PillonM. 2012 A Monte Carlo model for low energy D–D neutron generators Nucl. Instrum. Methods Phys. Res. B 2716–12 [17] www.srim.org/ [18] BrownD. 2012 Release of the ENDF/B-VII.1 evaluated nuclear data file Brookhaven National Laboratory Report BNL-98708-2012-IR, Brookhaven National Laboratory [19] CufarA., BatistoniP., ConroyS., GhaniZ., LengarI., MiloccoA., PackerL., PillonM., PopovichevS. and SnojL. 2017 Calculations to support JET neutron yield calibration: modelling of neutron emission from a compact DT neutron generator Nucl. Instrum. Methods Phys. Res. A 847 199–204 [20] AngeloneM., BatistoniP., MoroF., PillonM. and LoughlinM. 2012 Neutronics analysis and nuclear heating measurement up to the TFC in a mock-up of the ITER inboard shield Fusion Eng. Des. 87910–5 [21] RigamontiD. etal Unprecedented very high resolution neutron spectroscopy measurements of 14 MeV neutrons Nucl. Fusion 58 (2018) 026012 P. Batistoni etal 24 and (new) prospects for diagnosing DT fusion plasmas Meas. Sci. Technol. submitted [22] ReginattoM., GoldhagenP. and NeumannS. 2002 Spectrum unfolding, sensitivity analysis and propagation of uncertainties with the maximum entropy deconvolution code MAXED Nucl. Instrum. Methods Phys. Res. A 476242 [23] GuldbakkeM., KleinH., MeisterA., SchelerU., UnholzerS., PulpanJ. and TichyM. 1993 Response matrices of NE213 scintillation detectors for neutrons Proc. of the 8th ASTMEURATOM Symp. on Reactor Dosimetry (29 August–3 September 1993) ed H Farrar [24] De PangherJ. and NicholsL.L. 1966 A precision long counter for measuring fast neutron flux density Pacific Northwest Laboratory Report BNWL-260, Pacific Northwest Laboratory [25] TagziriaH. and ThomasD.J. 2000 Calibration and Monte Carlo modelling of neutron long counters Nucl. Instrum. Methods Phys. Res. A 452470–83 [26] GressierV. etal 2014 International key comparison of neutron fluence measurements in monoenergetic neutron fields: CCRI(III)-K11 Metrologia 5106009 [27] JednorogS. etal 2017 Activation measurements in support of the 14 MeV neutron calibration of 1 JET neutron monitors Fusion Eng. Des. 125 50–6 [28] ChenJ. etal 2007 International key comparison of neutron fluence measurements in mono-energetic neutron fields: CCRI(III)-K10 Metrologia 4406005 [29] LövestamG., NolteR. and RobertsN.J. 2014 EUROMET Project 822 ‘Comparison of neutron fluence measurements for neutron energies of 15.5 MeV, 16 MeV, 17 MeV and 19 MeV’ (EUROMET.RI(III)-S2) Metrologia 5106023 [30] IAEA TECDOC-619 1991 X-Ray & Gamma-Ray Standards for Detector Calibration (Vienna: IAEA) [31] LitaudonX. etal 2017 Nucl. Fusion 57102001 Nucl. Fusion 58 (2018) 026012