Operation principle of a so-called Be reflector-filter, used to enhance the flux of cold neutrons (Eâ¤5 meV), has been known for decades [1]. A recent study by ...
ICANS-XVIII 18th Meeting of the International Collaboration on Advanced Neutron Sources April 25–29, 2007 Dongguan, Guangdong, P R China
Implementation of a Be reflector-filter in the design of the Manuel Lujan Jr. Center Target Moderator Reflector System M. Mocko, G. Muhrer Los Alamos National Laboratory, LANSCE-LC, Mail Stop H805, Los Alamos, NM 87545, USA
Abstract Operation principle of a so-called Be reflector-filter, used to enhance the flux of cold neutrons (E≤5 meV), has been known for decades [1]. A recent study by Muhrer et al. [2] explored the possibilities of incorporating a cold Be reflector-filter system in the Manuel Lujan Jr. Center Target Moderator Reflector System (TMRS). Based on this simplified model the engineering group at LANSCE has developed a conceptual design for the new TMRS. A series of Monte Carlo neutron transport calculations (using MCNPX [3]) has been carried out to study the engineering details of the next generation TMRS. Optimization constraints and the investigated geometries will be discussed. We will show a factor of 2 gain in integral cold neutron flux when using our optimized geometry for the Be reflector-filter hydrogen moderator as compared to a supercritical liquid hydrogen moderator of the current-generation Lujan TMRS.
1.
Introduction
The potential usefulness of a cold beryllium reflector-filter to enhance cold neutron source brightness has been recognized for decades. The concept takes advantage of the Bragg edge in the beryllium scattering cross section at 5 meV (see Figure 1). Above 5 meV the total scattering cross section is approximately 6 b. Below 5 meV the cross section drops at room temperature (≈ 300 K) to approximately 0.5 b. By cooling the Be to 20 K, the cross section drop is much more apparent—more than 3 orders of magnitude—making it highly transparent to low energy (≤ 5 meV) neutrons. At the cryogenic temperature of 20 K, the mean free path for low energy neutrons is approximately 400 m, whereas a 7 meV neutron has only 1-cm mean free path. So a thick slab of beryllium at 20 K placed at the emission surface of a cold moderator will reflect (scatter) neutrons with energy higher than 5 meV back into the moderator, while transmitting all neutrons below this energy. The neutrons reflected back into the moderator have another opportunity to downscatter to lower energy and then the Be material is completely transparent to them. In this way the cold neutron source brightness can be significantly enhanced. Several experiments to evaluate the effectiveness of a Be reflector-filter have been carried out in early 1980s by Carpenter et al. [1]. They measured the neutron energy spectrum for a coupled liquid hydrogen moderator with and without a warm (300 K), 3.8-cm Be reflector-filter. Unfortunately, their measurements were not absolute, so no conclusion about the long-wavelength gain could be made. However, the published normalized spectra show an enhancement for neutron energies less than 10 meV. Another simulation study by Pitcher et al. [4] looked at potential (calculated) enhancements of brightness at long wavelengths when using a cold Be reflector-filter. The simulations were done for three types of neutron sources: short-pulse, and long-pulse spallation sources, and a steady-state source. Authors concluded that the brightness of cold neutron part of the spectrum can be boosted as much as 57% for both short and long pulse spallation sources. They reported only marginal gains (≈ 4%) for the steady state neutron source. The previous investigations prompted an experimental study, carried out at the Weapons Neutron Research Facility of the Los Alamos Neutron Science Center (LANSCE), to measure the gain in cold source brightness resulting from the use of a cold Be reflector-filter [5]. The experiment was done using a 5-cm-thick cold polyethylene moderator followed by a 10-cm Be reflector-filter, both cooled to 75 K. Neutron energy spectra for the coupled polyethylene moderator with and without the Be reflector-filter were measured. The analysis resulted in a 64% integral brightness ˚ gain for neutrons with wavelengths greater than 4 A. The experimental confirmation of efficiency of a cold Be reflector-filter at LANSCE sparked interest of the scientists to actually implement it in the next generation (Mark III) Lujan Target Moderator Reflector System (TMRS). 1 459
10
300 K
σtotal (b)
1 10-1 10-2 10-3
20 K -4
10
10-2
10-1
1
10
102
103
Energy (meV) Figure 1: Total cross section of beryllium at 20 K (dashed line) and 300 K (solid line).
Extensive conceptual study of a pre-moderated Be reflector-filter hydrogen moderator system was done by Muhrer et al. [2]. The investigation focused on three aspects: size and shape of the Be reflector-filter, size of pre-moderator, and changes due to different ortho/para hydrogen ratio. Current generation (Mark II) as-built Lujan TMRS model has been used with a makeshift water pre-moderated cold Be reflector-filter hydrogen moderator. The conceptual modeling study not only investigated various size/shape influence on the overall enhancement of the cold neutron brightness but also studied the change of the neutron spectra in different configurations. Extensive MCNPX modeling by Muhrer et al. [2] concluded an enhancement of approximately a factor of two for cold neutron flux (En ≤ 5 meV) as compared to a coupled hydrogen moderator of Lujan TMRS Mark II. The present paper reports on a series of simulation studies to understand the influence of various engineering details of a premoderated cold Be reflector-filter hydrogen moderator implementation in Lujan TMRS Mark III on the cold neutron brightness. This paper focuses on determination of the optimal size of the water premoderator, supercritical hydrogen moderator, and Be reflector-filter, taking into account all structural details of the engineering model which were neglected in earlier studies [2].
2.
Lujan TMRS Mark III
The Manuel Lujan Jr. Neutron Scattering Center (Lujan Center) is a Short-Pulse Spallation Source, providing intense neutron beams for scientific research. The Los Alamos Neutron Science (LANSCE) linear accelerator in conjunction with the Proton Storage Ring (PSR) produce intense pulses of 800-MeV protons impinging on a water cooled tungsten target. Figure 2 displays the tungsten target as part of the Lujan TMRS which employs four flux trap moderators (lower tier), two back-scattering moderators (upper tier) and a composite reflector/shield. The fourth generation of the Lujan TMRS is based on the currently-used Mark II model differing only in the lower tier flux-trap moderator design. Configuration of the upper tier partially coupled back-scattering moderators (light water and supercritical hydrogen moderator; see Figure 2) is identical to the Lujan TMRS Mark II. The new configuration for the lower tier flux trap moderators is shown in Figure 2 in the form of an engineering drawing. It consists of three light water decoupled moderators (5 in Figure 2) and one premoderated Be reflector-filter supercritical hydrogen moderator (4 in Figure 2). Our study is focused only on the Be reflector-filter hydrogen moderator which serves three neutron scattering instruments SPEAR, LQD, and Asterix via neutron Flight Paths (FP) 9, 10, and 11, respectively. Before commencing the modeling work we defined these four aspects as the guidelines (constraints) for our simulations: 1. maximize the cold neutron flux (En ≤ 5 meV), 2. heat generated by all particles must not exceed 2.20 W/µA of proton beam, 3. minimize fluctuations of the cold neutron flux with varying ortho/para hydrogen content, 4. geometry must respect structural stability and heat transfer limits.
460
protons
Upper tier 2
3
Upper tier
1
4
5 Lower tier Lower tier Figure 2: Engineering model of the Lujan TMRS Mark III. Elevation view of the target system (left) and cross sectional views of upper and lower tiers (right) are shown. Legend: 1–tungsten target, 2–liquid hydrogen moderator, 3–water moderator, 4–Be reflector-filter hydrogen moderator, 5–three flux trap water moderators.
In the following sections we will report on a series of Monte Carlo neutron transport calculations performed with the above mentioned constraints to find the optimal configuration of the Be reflector-filter hydrogen moderator.
3.
Monte Carlo model of the Lujan TMRS
Since the fourth generation of the Lujan TMRS is significantly similar to the current (Mark II) system our MCNPX geometry model was based on an as-built model of the Lujan spallation target system [6]. The updated geometry model faithfully reproduces the detail and complexity of the engineering model (Figure 2) that includes moderator canisters, influence of the piping systems, vacuum and liquid hydrogen canisters, target cooling details, etc. Figure 3 shows the elevation (left) and plane (right) views of the MCNPX geometry used in our calculations. The split tungsten target is marked as well as the composite beryllium (Be) iron (Fe) reflector/shield. The Be reflector-filter hydrogen moderator is emphasized with a black solid frame in both views (see Figure 3). The above-described MCNPX geometry model of the Lujan TMRS Mark III was used in our Monte Carlo transport calculations together with cold neutron scattering kernels for beryllium [7], liquid hydrogen [7], iron, and aluminum. The total scattering cross sections for cold neutrons and these materials are shown as a function of neutron energy in Figures 1, 4, respectively. The cold neutron flux was tallied at 10 m distance using a variance reduction technique called “next event estimator” realized as a point detector in the MCNPX code. Reference calculation was carried out for a partially coupled supercritical hydrogen moderator using an as-built Lujan TMRS model (Mark II) and the integral cold neutron fluxes for FP-9, FP-10, and FP-11 are listed in Table 2 for ortho/para liquid hydrogen content of 25%/75% [6]. The results of the optimization calculations are given in terms of relative enhancement of the integral cold neutron flux for neutrons with energy smaller or equal than 5 meV, which is depicted as “Mark3/Mark2” in the following plots.
461
Fe
1
Fe
Fe
Fe
Be Be
Be
Be
Be
Be
Be
Be
Be 2
Be
Fe
Be
Be
Be Fe
Fe
Fe
3
4
Figure 3: Monte Carlo geometry model used in our transport calculations. Elevation view of the Lujan TMRS Mark III is shown in the left panel along with a plane view in the right. Legend: 1– Be reflector-filter hydrogen moderator, 2– upper (segmented) tungsten target, 3– lower tungsten target, 4– flux trap water moderators.
10
σtotal (b)
σtotal (b)
102
10
1 Iron
Ortho
1
Aluminum
Para
10-1 10-2
10-1
1
10
102
10-2
Energy (meV)
10-1
1
10
102
103
Energy (meV)
Figure 4: Total scattering cross sections for cold neutrons. Scattering cross sections for ortho and para configurations of liquid hydrogen are shown in the left panel. Right panel depicts the total cross sections for iron and aluminum used in our transport calculations.
462
Be
H2O LH 2
1
Be 1
3
LH 2
2
H2O
2 3
Figure 5: Elevation (left) and plane (right) views of the Be reflector-filter hydrogen moderator. Beryllium reflector-filter (1), liquid hydrogen moderator (2), and water pre-moderator (3) thicknesses are shown in both views.
4.
Results
Zoomed view of the proposed pre-moderated Be reflector-filter hydrogen moderator is shown in Figure 5. The Be reflector-filter in consideration is composed of three main components: beryllium, liquid hydrogen, and water pre-moderator. Since the outer dimensions of the Lujan target system are fixed by the outer composite reflector/shield our task is reduced to two-dimensional maximization problem. In the first step we varied the water pre-moderator and liquid hydrogen moderator thicknesses. Once the optimal configuration (thickness) was found we investigated the influence of different ortho/para content in liquid hydrogen moderator on the neutronic performance of the Be reflector-filter. We also tried to determine the most influential structural components of the system on the neutronics. The above mentioned topics of our study will be described in more details in the following sections. Table 1: Heat generated by neutrons, protons, photons, π + , π − , and π 0 in the cryogenic volume (liquid hydrogen, beryllium, and cryo-tank walls) for different thicknesses of the water premoderator (H2 O). The range in the energy deposition corresponds to different thickness of liquid hydrogen (3.0–4.5 cm). H2 O (cm) Heat (W/µA)
4.1
1.0 2.41–2.52
1.5 2.25–2.34
2.0 2.11–2.20
2.5 1.80–2.03
3.0 1.76–1.90
3.5 1.61–1.78
Premoderator/moderator size
To find the optimal premoderator/moderator size we created 24 different geometries with different water premoderator and liquid hydrogen moderator thicknesses, while the overall size of the Be reflector-filter system was constant. The water premoderator thickness varied 1.0–3.5 cm in six steps and the liquid hydrogen thickness was changed in four steps in 3.0–4.5 cm range. All simulations were carried out using ortho/para hydrogen fraction of 75%/25%. The results of these 24 simulations are presented in Figure 6 in terms of relative boost of the integral cold neutron flux with respect to the Lujan TMRS Mark II. We see minimum variation of the histograms for different neutron flight paths (marked FP-9, FP-10, and FP-11). The maximum enhancement of the cold neutron flux is achieved for 1.5-cm water premoderator and 3.5-cm liquid hydrogen moderator. Before proceeding any further we must take into account the heat generated by all particles in the cryogenic volume of the Be reflector-filter since the cooling system limit of the hydrogen refridgerator is 2.2 W/µA of proton beam. In the above described calculations we transported 6 particle species (neutrons, protons, photons, π + , π − , and π 0 ) and tallied their energy depositions in the cryogenic volume of the Be reflector-filter system (beryllium, liquid
463
Mark3/Mark2
Mark3/Mark2
FP-9 2
1.5
1 5
1.5
1 5
Mo
de 4 rat or
Mark3/Mark2
FP-10
2
(cm 3 )
1
3
Mo
4
2
m) rator (c
de
Premo
de
rat
4
or
(cm 3 )
1
2
3
4
(cm) derator
Premo
FP-11
2
1.5
1 5
Mo
de 4 rat or
(cm
3
)
1
2
3
4
m) rator (c
de
Premo
Figure 6: Relative boost in integral cold neutron flux with respect to Lujan TMRS Mark II plotted as a function of water premoderator thickness and liquid hydrogen moderator thickness. The histograms for three neutron flight paths (FP-9, FP-10, FP-11) show 24 calculations with different premoderator (1.0–3.5 cm) and moderator (3.0–4.5 cm) thicknesses.
hydrogen, and cryo-tank walls—see Figure 5). The results of heat generated by these 6 transported particle species are summarized in Table 1 for different water premoderator thicknesses. The particle energy deposition varies greatly with changed water premoderator thickness, as expected. Since these changes move the cryogenic part of the Be reflectorfilter system in the radial direction from the tungsten target generating vast majority of neutrons and other secondary particles. Unfortunately, the energy deposition calculations exclude the smallest water premoderator thicknesses (1.0 and 1.5 cm), where the heat generated is greater than the cooling capacity of the hydrogen refridgerator (2.2 W/µA). Now when re-examining Figure 6 with the energy deposition results in mind we pronounce the 2.0-cm water premoderator optimal. This result fixes the outside dimensions of the cryostat holding the liquid hydrogen moderator and Be reflector-filter piece (see Figure 5). 4.2
Ortho/para liquid hydrogen concentration
Molecular hydrogen (H2 ) exists in two states, ortho and para-H2 . In the ortho form, the proton spins are aligned, and in the para form the spins are anti-parallel resulting in molecules with spins of 1 and 0, respectively. The energy difference between these two states is only 14.7 meV; para form is the lower energy state and at temperature of the liquid predominates (≈ 99.8%) at equilibrium. On the other hand, hydrogen in gas form at room temperature has para form concentration at only about 25%. Once the hydrogen is liquefied, ortho form naturally converts to para-hydrogen resulting in above mentioned equilibrium. This process is relatively well known in non-radiation environment but it is not fully understood in high-radiation environment. Since the total scattering cross section for these two molecular hydrogen states differs significantly below approximately 20–30 meV (see Figure 4), neutronic performance of a liquid hydrogen moderator depends strongly on
464
2
2
FP-9
FP-10 1.8
Mark3/Mark2
Mark3/Mark2
1.8
1.6
3.0 cm 1.4
3.5 cm 4.0 cm
1.2
1
0.2
0.4
3.0 cm 1.4
3.5 cm 4.0 cm
1.2
4.5 cm 0
1.6
1
0.6
Ortho concentration
4.5 cm 0
0.2
0.4
0.6
Ortho concentration
2
FP-11
Mark3/Mark2
1.8
1.6
3.0 cm 1.4
3.5 cm 4.0 cm
1.2
1
4.5 cm 0
0.2
0.4
0.6
Ortho concentration Figure 7: Relative boost in the integral cold neutron flux with respect to Lujan TMRS Mark II plotted as a function of ortho hydrogen concentration in the hydrogen moderator. The calculations were done for 2.0-cm water premoderator and four different liquid hydrogen thicknesses (3.0–4.5 cm) shown with different markers.
ortho/para hydrogen fraction. The differences in performance of a supercritical partially coupled hydrogen moderator were studied previously at the Lujan Center [8, 9]. Since we know that the ortho/para fraction is going to change as a function of time in our Be reflector-filter system we tried to find an optimal configuration when the relative enhancement of the cold neutron flux varies minimally with changing ortho-hydrogen fraction. 20 different MCNPX transport calculations were carried out to address this issue. We ran calculations with 5 different ortho-hydrogen fractions (0, 0.15, 0.25, 0.50, and 0.75) for 4 different liquid hydrogen moderator thicknesses (3.0, 3.5, 4.0, and 4.5 cm). Results are plotted in Figure 7 for three neutron flight paths (FP-9, FP-10, and FP-11) as a function of ortho-hydrogen concentration. Four markers display calculations with different liquid hydrogen moderator thicknesses. After quantifying fluctuations in relative enhancements of the cold neutron flux for 0.25, 0.50, and 0.75 ortho-hydrogen fractions we pronounced the 4.0-cm moderator thickness optimal. For a 2-cm water premoderator and 4-cm liquid hydrogen moderator Be reflector-filter in Figure 7, we see approximately a factor of 1.8 improvement in the integral cold neutron flux for all three neutron flight paths (FP-9, FP-10, and FP-11). One observes a relatively sharp drop of the gain in integral cold neutron flux with decreasing orthohydrogen concentration in Figure 7. Hence, for a successful operation of the Be reflector-filter hydrogen moderator, it is necessary to maintain the ortho-hydrogen fraction above approximately 25%. 4.3
Other considered factors
Even though our optimization study resulted in a reasonably close enhancement ratio of 1.8 as compared to a factor of 2 as reported by Muhrer et al. [2], we decided to spend some time to identify the most influential construction details of the Be reflector-filter system, with a possibility to boost the integral cold neutron flux even higher. We
465
Neutrons (/s/p/energy bin)
10
10
-9
-10
FP-9 Mark3 Mark2
10-11
10-12 10-1
10
3
5
10
107
10
8
10
Energy (meV)
Ratio Mark3/Mark2
2.5
2
1.5
1
0.5
0 0
5
10
15
20
Wavelength (Å ) Figure 8: Top panel displays neutron energy spectra for a supercritical partially coupled liquid hydrogen moderator (dashed line) and the premoderated Be reflector-filter hydrogen moderator (solid line) tallied by a point detector. Both calculations were done with 25% orthohydrogen concentration. Bottom panel shows the relative enhancement of the integral cold neutron flux as a function of neutron wavelength.
466
Table 2: Integral flux of cold neutrons (En ≤ 5 meV) for the Lujan TMRS Mark II [6] and Mark III (this work). Both calculations were done with liquid hydrogen ortho/para content of 25%/75%.
Flight Path (FP) FP-9 FP-10 FP-11
Integral cold neutron flux Mark II Mark III (n/p/cm2 ) (n/p/cm2 ) 1.00×10−8 2.04×10−8 1.08×10−8 2.11×10−8 1.01×10−8 2.04×10−8
investigated these four engineering details of the Be reflector-filter: • The film of liquid hydrogen between the emission surface of the Be reflector filter piece and the inner surface of the cryogenic tank. The original engineering design called for a 1.6 mm gap which results in approximately 16% penalty in cold neutron flux as compared to no gap. Even thinner gap (0.8 mm) causes a 7% penalty. • Aluminum walls of the cryostat, the vacuum tank and the premoderator tank simply attenuate the neutron flux. Hence it is necessary to reduce the amount of aluminum material in the way of neutrons. Our calculations showed that by thinning the aluminum walls of the cryostat from 3 to 2 mm we can boost the cold neutron flux by approximately 10%. On the other hand, alteration of the premoderator tank walls does not result in any significant gains in cold neutron flux. • Original engineering design called for usage of MACOR insulating plate to support the cryogenic tank of the Be reflector-filter (see Figure 5). A small concentration of boron (5%) in MACOR costs approximately 8% in lost cold neutron flux. Hence we need to search for a substitute material to be used for the supporting plate. • We also investigated the influence of impurities in beryllium material based on the analysis of a real beryllium sample by Brush Wellman Inc. We ran the Monte Carlo transport calculations with these, more realistic, beryllium compositions for both the Be reflector-filter piece as well as the outer, room temperature, Be reflector/shield, but we concluded no influence of the impurities on the neutronics. Taking into account the above mentioned results of our optimization study we proposed the following geometry for the Be reflector-filter hydrogen moderator system. The new design uses a 2-cm water premoderator in conjunction with a 4-cm liquid hydrogen moderator followed by approximately 13-cm Be reflector-filter. In order to maximize the boost of the integral cold neutron flux we require thin 2-mm cryostat aluminum walls, no boron in insulating support plate, and no liquid hydrogen film on the outer surface of the Be reflector filter. A transport calculation carried out taking these considerations into account and using ortho/para fraction of 25%/75% resulted in a neutron energy spectrum shown in the upper panel of Figure 8 as a solid line. The dashed line shows the neutron spectrum of the current-generation Lujan TMRS utilizing a supercritical partially coupled liquid hydrogen moderator. The lower panel shows ratios of the two spectra as a function of neutron wavelength, λ. We clearly see a factor of two improvement of ˚ the neutron flux with λ ≥ 4 A.
5.
Summary and conclusions
We carried out an extensive study of the construction (structural) details of a Be reflector-filter hydrogen moderator implementation in the Lujan TMRS Mark III design. The investigation resulted in optimal size of the water premoderator, and the liquid hydrogen moderator to be used in conjunction with Be reflector-filter. Our study also pointed out most important details of the engineering model which must be eliminated in order to boost the integral cold neutron flux even higher (liquid hydrogen film at the outer surface of the Be reflector-filter piece, elimination of MACOR as an insulating plate, thin cryostat walls). With our optimized design for the Be reflector-filter hydrogen moderator we demonstrated a factor of 2 improvement in integral cold neutron flux. Our proposed model is now a subject of structural and heat transfer engineering analyses which are the next step towards successful implementation of a water pre-moderated Be reflector-filter hydrogen moderator in the next generation of the Lujan TMRS Mark III.
467
References [1] J. M. Carpenter, R. Kleb, T. A. Postol, and R. H. Stefiuk, “The liquid hydrogen moderator on the ZING-P’ pulsed spallation neutron source,” Nuclear Instruments and Methods, vol. 189, pp. 485–501, 1981. [2] G. Muhrer, E. J. Pitcher, and G. J. Russell, “The neutron performance of a pre-moderated beryllium reflectorfilter hydrogen moderator system for the Manuel Jr. Lujan Neutron Science Center,” Nuclear Instruments and Methods A, vol. 536, pp. 154–164, 2005. [3] LA-CP-05-0369, MCNPX manual v. 2.5.0. [4] E. J. Pitcher, G. J. Russel, and P. D. Ferguson, “Use of a cold Beryllium relfector-filter to enhance cold source brightness at long wavelengths,” in The 14th Meeting of the International Collaboration on Advanced Neutron Sources, Starved Rock Lodge, Utica, IL, USA, June 14–19 1998. [5] E. J. Pitcher, G. J. Russel, G. Muhrer, J. J. Jarmer, and R. K. Corzine, “Experimental investigation of a cold beryllium reflector-filer,” in The 16th Meeting of the International Collaboration on Advanced Neutron Sources, D¨usseldorf-Neuss, Germany, May 12–15 2003. [6] G. Muhrer, “private communication.”. [7] R. E. MacFarlane, “New Thermal Neutron Scattering Files for ENDF/B-VI Release 2.” LA-12639-MS, 1994. [8] M. Ooi, T. Ino, G. Muhrer, E. J. Pitcher, G. J. Russell, P. D. Ferguson, E. B. Iverson, D. Freeman, and Y. Kianagi, “Measurements of the change of neutronic performance of a hydrogen moderator at manuel lujan neutron scattering center due to conversion from ortho- to para-hydrogen state,” Nuclear Instruments and Methods A, vol. 566, pp. 699–705, 2006. [9] G. J. Russel, G. Muhrer, E. J. Pitcher, J. Majewski, R. A. MacFarlane, J. J. Jarmer, Y. Kiaynagi, M. Ooi, T. Ino, P. D. Ferguson, E. B. Iverson, and D. W. Freeman, “Changes in Neutron Spectra with Ortho-Hydrogen Fraction for the Lujan center Flux-Trap Hydrogen Moderator,” in The 16th Meeting of the International Collaboration on Advanced Neutron Sources, D¨usseldorf-Neuss, Germany, May 12–15 2003.
468
