Development of UCN sources at PNPI

2.1.Cold beryllium source (CBS)

The history of UCN source developments at PNPI started in the early 1970s with a beryllium converter at ambient temperature, located in the center of the WWR-M reactor. While that first device was still rather ineffective for down-conversion of thermal neutrons to UCNs, the UCN yields were significantly improved after implementation of cryogenic infrastructure to cool the block of beryllium down to 30 K (see Fig. 2). A helium refrigerator of the type HGU-500/15 (HGU abbreviating «Holodilnaya Gelivaya Ustanovka» that translates as «Cold Helium Plant») was used for this purpose. To increase the thermal neutron flux in the converter, it was installed in a water cavity. The assembly was surrounded by a lead shield, which reduced the heat generated in the converter from 3 W/g to 1 W/g, so that less than 450–500 W had to be removed from the converter by the cold helium. A modernization of the refrigerator made it later possible to remove about 1300 W from the beryllium so that it could be operated without a lead shield, which made it possible to increase the thermal neutron flux in the beryllium converter by about a factor 2 and achieve a more stable mode of operation. Cooling the converter from 350 K down to its base temperature of 30 K increased the UCN flux by a factor up to 12 (see Fig. 2, right). This is what we call «UCN gain factor». The UCN flux at the entrance of the EDM spectrometer installed at this second version of source was 1.2 × 10⁴ s⁻¹ [3].

Fig. 2.

Cold beryllium UCN source at the WWR-M reactor. Left: artistic view of the central channel with the cooled UCN source in the reactor core (1 – fuel elements; 2 – lead shield; 3 – neutron guide mirrors; 4 – cryogenic pipes). Right: dependence of the UCN flux on the temperature of the beryllium converter (open and closed symbols refer to experimental results measured at two different channels. The solid curves represent simulations indicating the range of values for different samples of melted beryllium tested by transmission of UCN. Values of the heat load Q at 16 MW reactor power are given for source configurations with and without the lead shield.

2.2.Small liquid hydrogen source (SLHS)

The next step in the UCN source development was the use of liquid hydrogen as a more effective converter [4]. The Small Liquid Hydrogen Source (SLHS) was installed in the beryllium reflector outside the core of the WWR-M reactor (Fig. 3, left and middle), in a thermal neutron flux of 6 × 10¹³ cm⁻²s⁻¹. Key component of the SLHS was a heat exchanger in which the flow of cold helium from a HGU-500/15 refrigerator made it possible to condense 150 cm³ of normal hydrogen (75% ortho, 25% para). The total heat load on the low-temperature parts of the source was comprised of 0.3 W/g for the source material and 4 W/g for liquid hydrogen at 18 MW reactor power (which became available after an upgrade of the reactor). At a reactor power of 16 MW, the measured gain factor in the yield of UCNs was about 25, taking the UCN count rate at a hydrogen temperature of 280 K and a pressure of 3.2 bar for normalization. With a measured UCN flux of 4 × 10⁴ s⁻¹, the SLHS operated at PNPI from 1980 to 1985 was at that time the most powerful UCN source in the world. Note that the SLHS also became a prototype for a CN source at the Budapest Research Reactor [15].

The SLHS was also used to perform first studies of UCN production in solid hydrogen and deuterium, replacing the LH₂ used during normal source operation (Fig. 3, right). These works had to be carried out at a reduced reactor power of only 1.2 MW, because otherwise the refrigeration plant did not allow solidification of the converter material. For LH₂ at the boiling point, UCN gain factors of 20–30 were measured. Lowering its temperature further down increased the gain factor to 40–43, which finally reached a value of 47 after solidification. For deuterium, the maximal UCN gain factor determined after solidification was 57. These experiments demonstrated for the first time the potential of sD₂ for UCN production.

Fig. 3.

Small liquid hydrogen UCN source at the WWR-M reactor. Left: artistic view of the SLHS (1 – UCNS vessel; 2 – beryllium reflector; 3 – fuel elements; 4 – lead shield to decrease the heat load on the UCN source; 5 – UCN guide; 6 – helium cryo-pipes, 7 – hydrogen supply pipe). Middle: UCN source details (1 – cold helium buffer vessel; 2 – cold helium pipes; 3 – LH₂ pipes; 4 – LH₂ buffer vessel; 5 – hydrogen supply pipe; 6, 7 – cold helium inlet/outlet pipes, 8 – UCN extraction guide). Right: dependence of the UCN yield on the temperature of the hydrogen (∘ – while cooling down; ● – while warming up) and of the deuterium (□) in the converter (I – hydrogen condensation phase; II – increase of LH₂ density with decreasing temperature; III – hydrogen crystallization phase).

2.3.Universal cold neutron source (UCNS)

The aim to achieve still higher UCN fluxes led to the development of a new method for heat removal, which made it possible to place the next UCN source directly to the center of the WWR-M reactor core (Fig. 4, left) [5]. Key component was a thermosiphon described further below. The project was successfully implemented and provided high fluxes not only of UCNs, but also of cold, and polarized cold neutrons – that is why it was called a «universal» cold neutron source (UCNS).

The operation of the UCNS at the WWR-M reactor was launched at the end of 1985. A lead shield surrounded its one-liter LH₂ moderator, to reduce the heat load on the low-temperature vessels. The thermal neutron flux incident on the source was 2 × 10¹⁴ cm⁻²s⁻¹, and the total heat release in the source was 2 kW. For normal operation, a refrigerator plant with a cooling power of 3 kW at 20 K was used. The linear circulation velocity of LH₂ in the source reached 1 m/s at full reactor power of 18 MW.

The temperature dependence of the UCN gain factor for gaseous and liquid H₂ is shown in Fig. 4 (right), again referring to the UCN yield at a H₂ temperature of 300 K at 3 bar. At 23.5 K and 2.3 bar, the UCN gain factor sharply increases due to the density increase of the moderator material on liquefaction, supporting thermalization of the incident neutron flux. A further decrease in the temperature of liquid hydrogen leads to an additional increase of the UCN output, with the gain factor reaching a value of 55. This source was also used for experiments with LD₂ and H₂-D₂ mixtures.

Fig. 4.

Universal Cold Neutron Source at the WWR-M reactor. Left: artistic view of the installation. Right: temperature dependence of the UCN gain factor (referring to the UCN intensity at hydrogen at 300 K at 3 bar).

The operational principle of the thermosiphon for subcooled LH₂ is shown in Fig. 5. Motion of the hydrogen is caused by hydrostatic forces, which are due to the density difference in the right and left parts of the loop. An increase of the heat supplied to the source also increases the speed of circulations, so that the system is self-regulating to an external heat load. The hydrogen in the left part of the loop is cooled down in the heat exchanger by cold helium from the cryogenic plant. While the reactor was operated at its full power of 18 MW, the speed of movement of LH₂ along the loop tubes was 1 m/s. So, it turned out to be a very efficient circulation pump. Based on this principle of heat removal, PNPI designed and manufactured CN sources for the OPAL reactor (Lucas Heights, Australia) [8] and the CMRR (China Mianyang Research Reactor) [16], which are successfully operating right now.

Fig. 5.

Thermosiphon used to circulate the subcooled moderator material (LH₂, LD2, or H₂-D₂ mixtures) of the UCNS.

2.4.Solid Deuterium Source (SDS)

After the aforementioned successful studies with sD₂ in the SLHS vessel, more detailed studies focused on developing for the WWR-M reactor a UCN source dedicated to this converter material [32]. As shown in Fig. 6 (left), this sD₂ source called SDS was placed in a vacuum tube located inside a channel with an inner diameter of 210 mm which passed through the reactor’s thermal column. Its vessel made of 0.5 mm thick zirconium alloy had a cylindrical shape with elliptical ends. The length of the SLD cell was 350 mm, and its internal diameter was 160 mm. The source was cooled by a flow of liquid helium from a 150 W refrigerator. The weights of the zirconium cell and the sD₂ during operation were 1.7 kg and 1 kg, respectively. The SDS was irradiated through a conical channel in the reactor’s thermal column, separated from the reactor core by a beryllium reflector. The distance from the core center to the SDS center was 150 cm.

The thermal neutron flux density in the center of the SDS was (9.2 ± 0.8) × 10¹⁰ cm⁻²s⁻¹MW⁻¹. The heat generated in the zirconium due to γ-rays amounted to (1.3 ± 0.2) × 10⁻³ Wg⁻¹MW⁻¹, while that due to scattering of fast neutrons in the deuterium was (9 ± 1) × 10⁻⁴ Wg⁻¹MW⁻¹. At 12 MW reactor power, the total heat load amounted to 37 W (20 W in the zirconium, 11 W in the deuterium and additional 6 W due to thermal heat flows from outside). The source was connected to a 6 m³ reservoir via a tube of 56 mm diameter. Cooling the SDS soaks deuterium from the reservoir, to condense and solidify in the source vessel, after which the remaining gas in the system settles at the saturated D₂ vapor pressure. In the solid phase, the bulk volume of D₂ was 85% of the source volume. Several measures were in place to keep the system always in a safe state. These included a fixed inventory of deuterium, a flow of warm helium through a tube located in the deuterium outlet channel to avoid blockages during warmup, and an always open vent for dumping the deuterium into the reservoir.

The neutron guide system of the SDS is shown in Fig. 6 (right). A straight guide with cross Section 70 × 120 mm² connected to a guide system located outside the reactor shields, which consisted of a straight CN guide, a curved VCN guide and a curved UCN guide. The cross section of the straight and curved guides was 70 × 60 mm². They were all made of highly polished stainless steel and coated with the alloy ⁵⁸Ni-Mo providing a critical transverse velocity of 7.8 m/s for neutron reflection. Guide elements were connected by means of electron beam welding. The UCN trap (item 7 in the right part of Fig. 6) allowed us to select UCNs up to a critical velocity of 6.2 m/s defined by the trap material. The curved guide (4) selected the neutrons with wavelengths larger than 30 Å. Measurements of the VCN spectrum were carried out using a chopper (9) and a detector (10). The full neutron spectrum from the source was measured on the straight guide, also with the help of a chopper (11) and a detector (12).

Fig. 6.

Solid Deuterium Source at the WWR-M reactor. Left: location in a through-going channel of the reactor (1 – reactor core; 2 – beryllium reflector; 3 – SDS cell; 4 – source containment; 5 – UCN guide). Right: neutron guide system (1 – biological shield of the reactor; 2 – biological shield of the guides; 3 – straight CN guide; 4 – curved VCN guide; 5 – main guide from the SDS; 6 – curved UCN guide; 7 – UCN trap; 8 – UCN detector; 9 – VCN chopper; 10 – VCN detector; 11 – CN chopper; 12 – CN detector).

Experimental results on the neutron yields from the SDS at different ranges of wavelength are shown in Fig. 7. The left part of this figure presents the yields versus D₂ pressure in the system, normalized to the yield from LD₂ at a temperature of 20 K, calling this quantity the «relative gain factor». The pressure of 128.5 torr corresponds to the triple point at 18.7 K, where the transition to the solid phase occurs. Figure 7 (middle) demonstrates the same relative gain factor, but in dependence on the temperature as derived from the saturated vapor pressure. For UCNs, a maximal value close to 10 occurs at 10 K. The UCN yield increases on solidification of the D₂, because the transitive atom motions in the liquid freeze out and the cross-section for inelastic UCN scattering decreases, so that the source becomes more transparent. However, further reduction of the temperature below 10 K would not be effective, since the total cross section contains a then dominating constant contribution, σincel=2.2b, due to incoherent elastic scattering caused by the spin dependence of the neutron scattering length of the deuteron. However, for the deuterium molecule this value can increase close to ∼5 barn.

The strong dependence of the UCN gain factor on σincel is demonstrated in Fig. 7 (right), which compares the data points of measured absolute gain factors for the SDS versus temperature, calculated with and without account for incoherent scattering (curves 1 and 2, respectively). There, a different normalization is used («absolute gain factor»), which relates the UCN yields from the source at low temperatures to the UCN yield at room temperature and an equilibrium pressure of the D₂ gas of 5 atm. These calculations were performed with the effective neutron temperature fixed to 40 K, as it would be for a source of size sufficient to thermalize the neutron flux. As can be seen, the difference between these curves reaches one order of magnitude at low temperatures, which demonstrates the negative influence of incoherent scattering on the UCN yield. Density inhomogeneities in sD₂ lead to additional incoherent scattering, which thus causes a further suppression of the UCN yield. The maximum experimental gain factor, 1230, was obtained for a source with a volume of 6 liters, irradiated with a thermal neutron flux [35].

Fig. 7.

Relative and absolute (see text) gain factors of UCN, VCN and CN production in a range of temperatures/pressures near solidification. (Graphs taken from references [35]) left: as a function of D₂ pressure in the buffer tank. Middle: as a function of temperature. Right: temperature dependence of the absolute gain factor of the SDS.

The development of the SDS, despite all its successes, also revealed some issues of sD₂ as a UCN source material:

After successful demonstration of the SDS at the WWR-M reactor in 1995, it was proposed to use sD₂ at spallation neutron sources based on pulsed accelerators [34]. Solid D₂ as a UCN converter material had received great interest after the «The First UCN Workshop» held in 1999 in Pushkin, Russia. Experimental facilities based on sD₂ were subsequently constructed at the Los Alamos National Laboratory [30] and at the PSI accelerator [6].

3.1.Brief historical background and operational principle

The history of UCN source developments at PNPI continues even right now, with a new development of a high-intensity UCN source using superfluid ⁴He (He-II) as a converter. Neutron scattering in He-II was analyzed theoretically by Ahiezer and Pomeranchuk as early as 1945 [2], which may have inspired Zeldovich in 1959 to consider this completely neutron absorption free medium for UCN production [54]. Likely, due to the challenge to maintain this cryogenic liquid near the core of a nuclear reactor, this idea was not immediately pursued further in Russia. In 1977 then, Golub and Pendlebury published the idea to exploit the dispersion of coherent excitations in He-II for production of UCNs [14], where they also noted the possibility to produce large UCN densities in a He-II converter installed at an extracted cold neutron beam.

The basic mechanism of UCN production in He-II exploits the Landau dispersion curve relating the energy and momentum of excitations (phonons, rotons) of He-II. At the point where it intersects with the parabolic dispersion curve of the free neutron, at a neutron wavelength of 8.9 Å, the neutron may convert into a UCN in a coherent inelastic process by emitting a single phonon, thereby imparting most of its energy of 1 meV to the quantum liquid. The dominant processes leading to up-scattering back to CN energies involve two quasiparticles in He-II and become negligible with respect to neutron beta decay (τβ≈880 s) only well below 1 K (for instance, the time constant for the two-phonon processes at 0.6 K is 4000 s). Still, at temperatures above 1.1 K, UCN storage lifetimes of several tens of seconds are possible.

To exploit such long time constants requires other UCN losses from the He-II converter to be suppressed as well. On the one hand, there are losses at the walls of the converter vessel, which can be kept small by a proper choice of the material (at an extracted beam one may also use a magnetic reflector [57]). On the other hand, one needs to remove the strongly neutron absorbing isotope ³He from the converter below a relative atomic abundance of 10−12. Such purity of helium is achievable using the heat flush technique in He-II [26], or a superleak as a filter for ³He [52,55]. At PNPI we will use a superleak to produce the required amount of isopure ⁴He. A first prototype of such a filter was already tested at IFP RAN in 2012, showing a helium flow rate of 2.5 g/s at 1.2 K. At temperatures in the range 2.3–2.4 K, the average flow rate was 4.3 × 10⁻⁴ g/s. With a fraction of ∼10−6 of ³He in natural helium, the penetration of ³He through the filter thus does not exceed 4×10−10 g/s. In March 2021, a cryostat was designed and launched at PNPI, which will allow us to fill 4 tanks (10 m³ each) with isopure ⁴He. The system throughput was 0.5 g/s, using a superleak with a diameter of 8 mm. Filling of these receivers is planned for the beginning of 2023.

Developments of UCN sources based on He-II are currently advancing in various countries, including Canada, France, USA and Japan (see Table 2). While the principle of UCN production by cold-neutron conversion in He-II has experimentally been well established in several laboratories on the scale of prototypes, the next major step will be its implementation in the harsh environment near the core of a high-flux neutron source. The current developments pursued at PNPI aim towards this goal.

3.2.UCN source project at the WWR-M reactor

In the mid-2000s, it was proposed to implement a UCN source based on He-II at the WWR-M reactor in its so-called thermal column [37]. This void with a diameter of one meter adjacent to the reactor core offers a high neutron flux and plenty of space to host a large cryogenic installation. The projected source assembly is shown on the left side of Fig. 8. A water-cooled lead shield on the front part of the source assembly reduces the heat load on the He-II due to nuclear radiations. The vacuum container is cooled by heat conduction through the metal and graphite from the lead shield, which is sufficient to keep the temperature below 80°C.

Fig. 8.

Proposed implementation of a He-II UCN source in the thermal column of PNPI’s WWR-M reactor. Left: horizontal cut view. Right: heat loads on various elements of the UCN source at 16 MW reactor power.

An analysis of source performance to be expected for this source model showed that LD₂ would be the best choice of a material for the pre-moderator [28]. The latter needs to be enclosed in a double-walled aluminum vessel and cooled down to 20 K by helium from a refrigerator. The central bore in the LD₂ pre-moderator contains the cylindrical UCN converter vessel with a diameter of 300 mm, a length of 500 mm and a wall thickness of 2 mm, which during source operation is filled with He-II at 1.2 K. The front wall of the chamber is located at a distance of 92 cm from the center of the reactor core.

The inner surface of the UCN vessel is foreseen to be coated with a 3–5 nm thick ⁵⁸Ni-Mo alloy, for which UCNs have a critical reflection velocity of 7.8 m/s. UCNs can be extracted via a ⁵⁸Ni-Mo-coated neutron guide, separated from the source by an 100 μm aluminum foil (foil of the minimum thickness, which will not break down under the pressure drop between the UCN chamber and UCN neutron guide). The thicknesses of the graphite moderator, the lead shield and the LD₂ pre-moderator were chosen to obtain the maximal possible cold neutron flux in the UCN converter vessel, while keeping the thermal load on the coldest part of the source below 40 W.

Monte-Carlo simulations, using the program MCNP, determined neutron fluxes and heat loads. The latter are listed in the table on the right side of Fig. 8, quoting 28 kW on the source components at ambient temperature, 287 W on the LD₂ pre-moderator, and 35.2 W on the filled UCN vessel; all at a reactor power of 16 MW. Calculations showed that UCN densities of about 10⁴ cm⁻³ might be achievable in the experimental trap of an EDM spectrometer [44].

For implementation of this project, modern cryogenic and vacuum equipment was purchased and put into operation at PNPI. It consists of a helium liquefier with a capacity of 100 liters per hour (Linde model L-280), a helium refrigerator providing 3000 W at 20 K, and a helium pumping station enabling circulation of 3 g helium per second at 50 Pa.

3.3.Full-scale model of the UCN source

Operation of the UCN source as described above requires a technological complex, capable to maintain the He-II filled UCN converter close to 1 K, under the heat load due to the radiations from the reactor. For validation of the feasibility of such a UCN source, we created a full-scale model, respecting the geometric conditions in the thermal column of the WWR-M reactor. This mockup included all the cryogenic and vacuum facilities as needed in its designated position. Thermal studies were performed using a heater, which imitated the heat flow from the reactor radiation. To make these studies as realistic as possible, a thermal screen at 20 K surrounded the entire low-temperature part of the model. The goals of experiments with this setup were:

Figure 9 shows a schematic of the technological complex implemented at PNPI. It includes the mockup of the UCN converter, a helium cryostat, a helium liquefier (L-280), a helium refrigerator (TCF-50), a helium gas pumping station, a helium storage system, and auxiliary technological and control systems. Figure 10 shows some photos of the whole technological complex, and Fig. 11 the UCN converter vessel.

Fig. 9.

Technological complex of the UCN source mockup (see text).

Fig. 10.

UCN source mockup at the WWR-M reactor (1 – helium liquefier L-280; 2 – helium refrigerator TCF-50; 3 – helium cryostat; 4 – UCN source model; 5 – pumping unit for isotopically pure ⁴He gas; 6 – helium vapor heaters; 7 – liquefier control cabinet; 8 – refrigerator control cabinet; 9 – helium roots pumps).

Fig. 11.

UCN converter vessel mockup (1 – helium vessel, 2 – helium vapor pipe, 3 – He-II supply pipe, 4 – heat screen, 5 – wire pipe, 6 – 250 Ohm nichrome electrical heater). The He-II vessel has a diameter of 300 mm and a length of 500 mm. The chamber is equipped with a helium level gauge and a temperature sensor.

As a first step in preparation of a cooldown of the source mockup, the helium gas inventory (corresponding to 150 liters of liquid at standard conditions) is circulated through a cryogenic absorber in a cold trap filled with liquid nitrogen, which removes most of the nitrogen and water impurities from the helium. Then, the helium refrigerator TCF-50 is launched to cool the heat shields to 20 K. Next, the L-280 liquefier is launched. Temperatures below 4 K are attained using the helium gas pumping station, which includes four Edwards HV-30000 roots pumps, each with nominal pumping speed of 30000 m³/h. This reduces within about 40 minutes the helium temperature to below 1.1 K (see Fig. 12).

Fig. 12.

Cooldown of the liquid helium, using the system of roots pumps (see text).

The main goal of the first cooldown test experiment was to determine the real temperatures of the He-II in the converter at various heat loads, i.e., 0, 15, 30, and 60 W. The latter exceeds the calculated value for the heat load at the WWR-M reactor operated at a power of 16 MW, which is expected to be 52 W (broken down to 35.2 W due to reactor radiation, 0.8 W through thermal bridges, and 16 W due to heat carried by the constant inflow of liquid helium at 4.2 K to be cooled down to the converter temperature). Each level of the heater power was maintained for 10 minutes, which was sufficient to stabilize the temperature in the cryostat and the converter vessel. Figure 13 shows the experimental results, which demonstrate the feasibility of the studied UCN source at the WWR-M reactor. More details about this experiment can be found in Ref. [41].

Fig. 13.

Evolution of the liquid-helium temperature during the experiment with a resistive heater (see text). The values of pressure measured at the inlet of the pumping station were: 0.5 Pa at 0 W, 5.7 Pa at 15 W, 14.6 Pa at 30 W, and 24 Pa at 60 W power provided by the heater.

This experimental confirmation of maintaining superfluid helium under such high thermal loads opens up new possibilities in the technology of producing UCN sources. While sources of this kind were previously located under conditions of low heat inputs and, correspondingly, low neutron fluxes, it is now possible to significantly increase the UCN density by placing the converter in close proximity to a strong neutron source.

3.4.UCN sources on cold neutron beams at PIK reactor

In 2020, a Federal Program was implemented for creating experimental facilities for the PIK reactor. One of the main facilities to be constructed within this Program is a new He-II based UCN source for research in fundamental physics. Afterwards, the decision was taken not to relaunch the WWR-M reactor anymore. This motivated us to study possibilities to place a He-II converter in a cold neutron beam at the PIK reactor [38].

The main differences between the UCN sources studied for the PIK and WWR-M reactors are shown in Fig. 14. No in-pile position is available at the PIK reactor right now, so that a UCN source based on He-II has to be located on an extracted neutron beam, at a minimum distance of 3 meters from the reactor core, while in the thermal column at the WWR-M reactor this distance would have been only 40 cm. At a 3-meter distance from the neutron source, the solid angle is 10−4×4π, leading to a reduced initial neutron flux density. On the other hand, the thermal neutron flux at the WWR-M reactor was 3 × 10¹² cm⁻²s⁻¹, while at the PIK reactor it will be 7 × 10¹⁴ cm⁻²s⁻¹. Hence, moving from WWR-M to PIK, almost four orders of magnitude of the in-pile flux are lost, but about two orders are regained due to the higher flux in the PIK reactor.

Due to this lower flux incident on the UCN source, the total heat load on the He-II will be 10 times lower than it would have been at the WWR-M reactor. Moreover, it can be further reduced by placing a block of polycrystalline bismuth in front of the source, serving for protection against gamma radiation. A 100 mm thick block would reduce the heat release in the UCN source vessel by 30 times. This will greatly simplify the technique to maintain the converter at 1 K and correspondingly reduce the cost of the necessary cryogenic and vacuum equipment. Accordingly, the temperature of He-II can be lowered significantly, with a positive impact on the UCN density. Results of a work studying bismuth filters are given in Ref. [39].

Taking all effects into account, the UCN density attainable in an external experimental trap drops by one order of magnitude when moving from the «in-pile position» at the WWR-M reactor to an «in-beam position» at PIK. It is noteworthy that simulations of two rather different schemes of a UCN source, with a cold source implemented inside or outside the PIK reactor (see Fig. 14) predict the same UCN density. However, operating a CN source in-pile would be very expensive due to the necessary powerful cryogenic equipment able to remove about 7 kW of heat loads. The next subsection presents our currently preferred solution for a UCN source at PIK.

Fig. 14.

Schematic comparison of UCN source implementations (in «zero approximation») at the PIK reactor with a CN source in-beam (upper figure) and in-pile (middle), and at the WWR-M reactor (lower). Φ denotes the thermal neutron flux.

3.5.UCN source on the thermal neutron beam GEK-4 at PIK reactor

This UCN source will be installed on the thermal neutron beam channel GEK-4. With its inner diameter of 220 mm, it is the largest available channel of the PIK reactor (see Fig. 15, left). Our plan is to create a 1-meter wide aperture in the outer PIK biological shielding. Similar to the source design for the WWR-M thermal column, this hole is large enough to host the entire UCN source, with a graphite reflector, a low-temperature pre-moderator and the He-II converter, everything included in an outer vacuum vessel. The front part of the UCN source will start directly after the flange of the GEK-4 channel. The thermal neutron flux density in the He-II in this geometry is expected to be 6.6 × 10¹⁰ cm⁻²s⁻¹.

The works on optimization of the UCN source included analyses of the source configuration as shown in Fig. 15 (left), but also of a so-called «inverse pre-moderator design». In this latter scheme, the low-temperature pre-moderator is placed behind the He-II, therefore mainly working as a reflector, which returns slowed-down CNs to the He-II source chamber. Liquid D₂ and solid methane were considered as pre-moderator materials. Preliminary calculations showed that LH₂ would give a significantly lower 9 Å neutron flux at the same heat load, and was therefore discarded. The use of solid methane would entail strict measures to comply with government safety rules, as clogging of cavities with solid methane can lead to disruption of the pre-moderator vessel. We have therefore decided to use LD₂ as a pre-moderator.

Fig. 15.

UCN source implemented in the horizontal experimental channel GEK-4 at the PIK reactor. Left: source model studied by MCNP calculations (see text); 1 – UCN converter vessel containing isopure superfluid ⁴He; 2 – heat exchanger containing helium at 1 K for cooling the converter vessel; 3 – LD₂ pre-moderator; 4 – graphite; 5 – lead shield; 6 – aluminum biological source shield; 7 – steel reactor biological shield; 8 – borated polyethylene reactor biological shield). Right: optimized parameters of the UCN source (for 100 MW reactor power).

Neutron flux densities and heat loads on the main parts of the optimized UCN source are summarized in Fig. 15 (right). Compared to the WWR-M project described in Section 3.2, heat loads on the low-temperature parts of the source are sufficiently low to use gaseous helium as a coolant for the lead screen. Helium, compared to water, has a lower neutron cross section, which gives 1.57 times more 9 Å neutrons in He-II. The results of the thermal calculation showed that to ensure the proper thermal condition of the lead shield, a minimum helium mass flow rate of 1.1 g/s is required, at which the temperature of the shield will be 59.5°C, and the maximum temperature of the aluminum vacuum vessel will be 58.8°C.

The optimization of the UCN source for GEK-4 resulted in the same design of the pre-moderator part (with the same material composition) as for the WWR-M reactor. However, at this new location, the heat release in the pre-moderator vessel has noticeably decreased from 287 W to now 10 W. In addition to changing the material of the UCN source chamber (from AlMg5 to stainless steel), the design of the helium part itself was also changed: a heat exchanger was added, the evacuation pipeline was removed, the diameter of the neutron guide was increased, etc. The total value of heat loads not related to radiations from the reactor (i.e., those left due to thermal bridges and thermal radiation) was calculated to be 3 W. Thus, the total heat load on the helium part, including the heat from the reactor at 100 MW, will be 6.85 W. This heat will be removed through a heat exchanger, using ⁴He as a coolant.

The UCN density calculated for this source would be 2.2 × 10³ cm⁻³ in case of the closed UCN converter. However, we plan to work in the mode of filling experimental traps with UCN together with the source, so the UCN density in traps is of practical importance. For predictions of the UCN fluxes and densities available in an experimental setup placed in the main hall of the reactor to search for the neutron EDM, a Monte-Carlo model has been developed. The model is based on realistic assumptions on UCN transport in the guide system connecting the source with the experiment and also takes into account the presence of 2 aluminum foil (100 μm each) that separate the volumes of the UCN source, neutron guide and nEDM facility. The simulations predict a UCN density of 200 cm⁻³ in the setup. This density will allow us to achieve a sensitivity of 1×10−27e· cm/year in the measurement of the neutron EDM [12].

In general, a broad research program is planned. The UCN guide system has been designed to feed up to five experimental facilities [43]. At the start of its operation, it is planned to equip the UCN source with experimental setups already available at PNPI: an nEDM spectrometer [40] and two neutron lifetime experiments, one with a gravitational [42] and one with a magnetic trap [10]. In Fig. 16 the corresponding beam lines are designated as UCN1, UCN2 and UCN3, respectively.

Fig. 16.

Floor plan showing the UCN source at GEK-4 at the PIK reactor with its UCN guide system (in navy blue) and beam ports to five locations for experiments (in magenta); UCN1 – nEDM experiment, UCN2 – gravitational trap, UCN3 – magnetic trap, UCN4 – free, UCN5 – free.

4.1.Option 1 – inside twister

From our simulations of the various source options for the WWR-M and PIK reactors as described in Section 3, we expect the largest UCN fluxes achievable for a UCN source with a pre-moderator placed closest to the primary neutron source. At the ESS, this would mean to implement the source inside the twister, replacing the foreseen LD₂ cold source (see Fig. 17). This location will obviously be accompanied by huge heat loads on the He-II converter and its vessel. If it does not exceed 150 W, a bismuth screen surrounding this vessel (see Fig. 18) might reduce it to a level of 5–8 W, at which modern cryogenic and vacuum equipment as described above would be able to keep the He-II converter close to 1 K. It is however most likely that the heat load will be too high, so that the He-II converter will have to be located outside the LD₂ cold source, but then in closest possible vicinity. This situation is discussed in the next subsection.

Fig. 17.

Option 1 – inside twister (1 – He-II converter; 2 – bismuth shielding with water cooling, which will also thermalize fast neutrons; 3–LD₂ pre-moderator; 4 – vacuum vessel; 5 – UCN guide; 6 – cryogenic supply lines).

4.2.Option 2 – in a standard beam port viewing the LD₂ cold source

In this case, the maximal lateral thickness of the He-II converter cannot exceed 100 mm. Depending on the total heat loads, its length can be 2–3 meters. The heat load can be further reduced by a block of polycrystalline bismuth in front of the converter, serving for protection against gamma radiation. The inner part of the beam port can be covered with a CN reflector. It would be useful to foresee some flexibility for placement of the UCN source along the axis of the beam port, for optimization of the CN yields and the heat loads. This scheme for a UCN source is similar to what we proposed for the PIK reactor with a cold neutron source inside reactor (see Section 3.4).

Fig. 18.

Option 2 – in a standard beam port.

4.3.Option 3 – in the large beam port

With respect to heat loads and neutron fluxes, this solution looks most realistic and promising of all options discussed here (see Fig. 19). Moreover, the data presented in Ref. [53] indicates a heat load of only 8.4 W on a 57.6 liter He-II converter and its vessel, which would be manageable for the cryogenic complex described in Section 3.3. We would recommend further optimization of this geometry, for which we expect the inverse scheme of the pre-moderator to be advantageous.

Fig. 19.

Option 3 – in the large beam port.

4.4.Option 4 – outside the shielding monolith

This case corresponds to the scheme of a UCN source coupled to a cold source at the PIK reactor (see Section 3.4). We propose this option only as a solution if for any reason none of the others can be done. An advantage of this option would be that it can be realized at any time, even after the actual launch of the ESS. Note the «inverse scheme» arrangement of LD₂, shown in Fig. 20. Working as a reflector instead of a CN generator, one may expect a significant increase of 9 Å neutrons impinging on the He-II converter. However, this option will unfortunately most likely provide the lowest UCN densities, due to the large distance of more than 5 meters from the CN source.

Fig. 20.

Option 4 – outside the shielding monolith.