facilities and methods

Ultracold —Physics and Production

Introduction Free neutrons are the most funda- mental ones of the easily accessible, electrically neutral -1/2 systems. They take part in all the four known interactions: they strongly interact with nucleons and nuclear matter, they undergo weak β-decay, they have magnetic moments and gravitational mass. Although the itself is a composite system, and thus, strictly speaking, not a fundamental particle, it is one of the finest probes for funda- mental physics. Neutrons can be and are used to study all interactions, to search for new interactions, to test fundamental quantum mechanics and symmetries of nature. Over the past five decades the field of slow neutron Figure 1. Dispersion relation of superfluid helium (c) and free neutron (a). precision physics developed and Neutrons with E » 1meV can excite a single phonon q » 0.7Å-1 with same energy could usually make major advance- and are thus down-scattered to the UCN energy range. The UCN production ments when new and more powerful rate (b) (circles) [16] shows the dominance of this single phonon process with neutron sources became available. respect to multiphonon processes. Very often experiments benefit from using slower neutrons and, thus, longer de Broglie wavelengths λ=2πh/(m · v), for example, thermal Besides numerous reviews, two new schemes for the production of neutrons at about 2,200 m/s average text books have been written on UCN UCN that have been studied in detail velocity have wavelengths of the typi- [2, 3]. For an update on the status of recently and are presently being real- cal atomic scale of about 1 Å. Fermi the field, readers are referred to two ized at various laboratories around the and Zinn [1] found that neutrons can recent workshops [4, 5]. In this article world. In particular, the UCN source be totally reflected from material we briefly review the peculiarities of project currently commissioning at surfaces under grazing angles of inci- ultracold neutrons, including their the Paul Scherrer Institut (PSI) in dence. To each material a critical generation and mentioning some of Villigen, Switzerland will allow for a (maximum) angle for total reflection their applications. next generation of more sensitive fun- of thermal neutrons exists that As several times already in the damental physics studies. becomes larger for smaller neutron past, we are today at a point at which velocities. Consequently, sufficiently fundamental physics applications slow neutrons will be reflected under require larger UCN intensities in Ultracold Neutrons all angles of incidence. These neu- order to further advance. Especially Ultracold neutrons are free trons are called “ultracold neutrons” the importance of the search for a (unbound) neutrons defined via their (UCN). Due to their specific proper- finite value of the electric dipole most important property: they can be ties (λ≥600 Å) they have opened a moment of the neutron pushes the stored! They have very low kinetic new door for the field of fundamental development of new and more power- energies below about 300 neV corre- physics. ful UCN sources. We describe the sponding to ~3.5 mK, hence their

Vol. 20, No. 1, 2010, Nuclear Physics News 17 facilities and methods

realistic storage containers (e.g., holes or slits) contribute considerably.

UCN in Fundamental Physics The possibility to store UCN for relatively long observation times makes them unique and highly sensitive probes, testing our understanding of fundamental physics [2, 3, 6]. Most of these experiments today are statistically limited and further advancements strongly depend on new high-intensity sources for UCN. A key experiment is the search for a permanent electric dipole moment of the neutron (nEDM). A finite nEDM violates time-reversal invariance and, therefore, might help to understand the matter–antimatter asymmetry in our universe. It is tightly linked to some of the open problems in Figure 2. Experimentally determined temperature dependence of UCN modern physics, the so-called “strong production in deuterium [22]. The sharp increase with solidification is obvious. CP-problem” and the “SUSY CP- problem.” Its observation would be a clear indication for physics beyond the electro-weak Standard Model of name: ultracold. They can be confined In inhomogeneous magnetic particle physics [7–9]. Other impor- by the strong interaction (total reflec- fields, the kinetic energy change of tant studies with UCN include deter- tion at any angle of incidence from the neutrons can be expressed as mination of the neutron lifetime and Δ =± Δ| | surfaces of certain materials like Ni, Be, Ekin 60 · neV/T B , taking the decay parameters, strongly influenc- stainless steel), the positive sign if the spin component is ing our understanding of weak inter- interaction (repulsion of one spin com- antiparallel to the field B. actions and big bang nucleosynthesis ponent from field gradients due to the Also the gravitational interaction [6]. UCN are also being used to study neutron magnetic moment) and due to is on the same scale for UCN, fundamental quantum mechanics, search Δ =Δ Δ gravitation (limited vertical reach). Ekin h · 103 neV/m, where h is for exotic interactions, test baryon num- UCN reflection on walls via the the height difference. ber conservation, and measure proper- strong interaction can be well described The effects of all three interactions ties of the neutron itself, such as the by a potential step model solving the can be combined and used for UCN search for a tiny but finite charge of Schrödinger equation with a step traps. The energy spectrum of stored the neutron. height equal to the so-called Fermi UCN depends on peculiarities of the = π 2 (optical) potential: VF (2 h / m)·Na; trap and is a function of height with a where N is the number density in the material or magnetic field dependent The Production of UCN material assumed to be homogeneous, cut-off, typically below about Δh=3m. The typical kinetic energy of UCN a is the coherent scattering length, and m The storage time of UCN is fun- (~100 neV) is orders of magnitude is the mass of the neutron. For exam- damentally limited by the time con- below the typical neutron energy in a = β ple VF(Ni) 252 neV, VF () stant of neutron -decay of almost thermal moderator (~25 meV) of a = 304 neV. For kinetic energies below 15 minutes. Practically, also loss factors , such as a nuclear this threshold energy total reflection like nuclear absorption, inelastic reactor. The energy distribution of the occurs at any angle. up-scattering and imperfections of thermalized neutron gas is often

18 Nuclear Physics News, Vol. 20, No. 1, 2010 facilities and methods

slowed down when climbing the grav- itational potential. Various developments have allowed one to increase the intensity of UCN considerably over the years. Clearly, the extracted intensity is proportional to the initial one, thus high initial neu- tron flux helps and could be increased by more than 4 orders of magnitude as compared to the first experiments. Also, the intensity in the tail of the Maxwell- Boltzmann distribution increases sig- nificantly when the temperature of the neutron gas is lowered, thus extract- ing UCN from a cold moderator of Figure 3. UCN production as function of incoming neutron energy [20]. The liquid hydrogen or deuterium gained dashed, colored curves display individual cold neutron energy distributions almost another 2 orders of magnitude. (intensity: vertical right scale) prepared from the “white” cold neutron beam Improvements were made on UCN using a neutron velocity selector and measured by time-of-flight. The measured guide quality by developing high UCN production cross-section (left vertical scale) for the individual energies is Fermi potential, very low roughness displayed as open circles. The red squares are the cross-sections calculated for surfaces, allowing a low-loss trans- the energy distributions, the solid black line is calculated as continuous port of UCN over many meters. function of energy. The cut-off at around 9 meV comes from cut-off of the Today’s highest UCN intensity is phonon density of states in the simple Debye model at the Debye temperature. obtained at the instrument PF2, which Above this energy, multiple phonon excitations still permit UCN production. has been in operation for more than 20 years at the high flux reactor of the Institute Laue-Langevin (ILL) in approximated by a Maxwell-Boltzmann control the background count rate of Grenoble. Vertical extraction over distribution and the amount of neutrons UCN detectors on a level of a few 17 m height through a curved replica in the very low energy tail is therefore counts per 1000 s allowed at that time guide system is used for initial deceler- rather small. Lower energy neutrons to detect similar UCN rates. Curved ation and background suppression. Then become further suppressed because of neutron guides were used to filter out a mechanical decelerator, the “neutron longer travel times in the moderator faster neutrons that could not fulfill turbine,” transforms very cold neutrons medium before extraction and the the necessary conditions for total of approximately 40 m/s into the UCN temperature of the extracted neutron reflection. Shapiro’s group in Dubna energy range. The neutron turbine gas is accordingly shifted to higher succeeded in detecting UCN this way (developed by A. Steyerl since 1975) values. Historically it was therefore in 1968 [10]. Steyerl (Munich) in consists of a set of curved blades mov- not even clear that a sizeable amount of 1969 [11] reported, independently, ing with a peripheral velocity 20 m/s ultracold neutrons could be extracted measured neutron cross-sections for in the same direction as the neutrons. from a moderator, having in mind also neutron velocities as low as 5 m/s The neutrons are thus decelerated by additional material that the neutrons (E = 130 neV). In this case, vertical several total reflections from the mov- must transmit, such as enclosures of extraction of the neutrons out of the ing curved blades. Up to 50 UCN/cm3 the moderator medium and safety reactor was used, which helps to miti- have been observed in a storage set- windows. While such considerations gate some of the extraction loss prob- up directly at the turbine exit [12]. In might have been somewhat discour- lems: the neutrons are still faster when specific experiments, the achievable aging, UCN have the distinct advan- penetrating window materials (loss density is usually somewhat lower, for tage that they can be detected with cross-sections are usually inversely example, about 2 polarized UCN/cm3 very high efficiency. The ability to proportional to velocity) and then get in the nEDM experiment [13].

Vol. 20, No. 1, 2010, Nuclear Physics News 19 facilities and methods

While many ideas for more intense UCN sources have been discussed over the past 4 decades, only over the past 10 years has there been a major effort to realize those. The obvious reason is that most UCN experiments nowadays are again statistically lim- ited. In order to obtain higher UCN intensities and densities the trend is to build so-called “superthermal sources.”

Superthermal UCN Sources In “conventional” UCN sources neutrons are extracted from a distribu- tion almost in thermal equilibrium with a moderation system. The con- cept of “superthermal” sources is dif- ferent: Thermal or cold neutrons are inelastically scattered and transfer Figure 4. a) Schematic view of the main components of the ultracold neutron their kinetic energy to an excitation of source at PSI contained in a large moderator and vacuum tank system. b) the scattering medium (e.g., to a Picture of the UCN vacuum tank taken during construction in spring 2009, phonon). The loss in kinetic energy is showing the main neutron guides’ vacuum tubes being prepared for welding. called down-scattering. By detailed balance, the reversed process of up- scattering is suppressed by a Boltz- In fact, the achievable UCN density The two factors, production rate P − can be much larger than the one that and average storage time τ, can be mann factor, exp( DE/kBT). If the medium is sufficiently cold would correspond to the thermal equi- very different for different media. and the excitation energy introduced librium density at the given moderator The research and development by the neutrons can be cooled away, temperature. Therefore, and because effort of the past few years concen- up-scattering becomes negligible. the process is a one-step conversion trated mostly on two possible Then, the UCN production becomes of cold neutrons to UCN rather than a choices, superfluid helium and solid insensitive to a further temperature moderation, the medium is often deuterium. Qualitatively, helium decrease of the medium. called a “superthermal converter.” can provide a very long storage The UCN production rate Q can The maximum achievable UCN time but has a comparatively low be expressed as: density is given as the product of pro- production rate, while deuterium duction rate per volume P times the has a high production rate but a UCN average storage time τ in the rather limited τ. Below, we give = QE()()UCN VPE UCN system, which depends on the rates some more detail on the status of (1) τ =→VEd ΦΣ() E E for neutron decay (1/ β), up-scattering research with both these materials. ∫ 00 0 UCN τ in the medium (1/ up), absorption on As a matter of fact, UCN sources the nuclei of the medium or contami- have been constructed already for τ where V = UCN production volume, nants (1/ abs), and, for a containment, both materials, and the next genera- E = neutron energy in the UCN also according losses during wall col- tion of powerful sources will use UCN τ = lisions (1/ wall): these media. Research on other range, E0 incoming neutron Φ = promising materials (e.g., solid energy, 0 incoming , Σ= −1 oxygen, solid heavy methane, and macroscopic cross-section for ⎛ 11 1 1⎞ down-scattering, P = UCN production t =+++⎜ ⎟ nanoparticles), is ongoing but much ⎝ tt t t⎠ rate per unit volume. β up abs wall (2) less advanced [4, 5].

20 Nuclear Physics News, Vol. 20, No. 1, 2010 facilities and methods

Superfluid Helium as Ultracold the source for a comparable time in order to avoid up-scattering on para- Neutron Converter order to approach equilibrium density. deuterium. The loss cross-section for Already in 1977, Golub and This more than compensates for a low UCN is not negligible as in 4He. The Pendlebury [14] developed ideas to production rate and allows for UCN UCN lifetime in solid deuterium it is use superfluid helium as a converter densities of 103–104 cm−3 at existing ultimately limited to τ ~ 150 ms by for UCN. The characteristic phonon cold neutron beam lines. Two types of neutron absorption on deuterons. In dispersion curve of He-II is an ideal UCN sources based on superfluid practice τ ~ 30 ms has been demon- two-level system for superthermal cool- helium are being operated and further strated and up to 70 ms may be ing (see Figure 1). As explained before, developed: achievable. In contrast to superfluid an incident neutron can excite a quasi- helium, integral sources make no particle by transfer of energy and • Integral sources: These are sources sense; the issue is rather to extract the momentum while the inverse process is where experiment and source are UCN from the converter system as suppressed by a Boltzmann factor. In combined in one apparatus and the fast as possible [19]. Advantages of superfluid helium, up-scattering becomes measurement is performed in-situ solid deuterium over helium are the negligible below T ≈ 0.7 K. The domi- inside the superfluid helium. fact that it can be operated at tempera- nant process, shown in Figure 1, is the Experiments to measure the elec- tures of 4–8 K and it provides a larger excitation of a single phonon at the tric dipole moment of the neutron UCN production cross-section. The crossing of the free neutron and (ILL, SNS) or the neutron lifetime higher operating temperature allows phonon dispersion curves, with a (NIST) are being pursued. one to operate at higher heat loads − momentum transfer q ≈ 0.7 Å 1 and • Multipurpose sources: The source (i.e., in a higher flux closer to the E ≈ 1 meV, corresponding to a neutron is an apparatus on its own and actual neutron source). Concepts have wavelength of 8.9 Å. The availability delivers neutrons to any experi- been developed that allow one to of 8.9 Å cold neutrons is crucial and ment connected to it by UCN obtain densities of 103–104 cm−3. The their flux must be maximized. An guides. Such sources had prob- physics of solid deuterium UCN advantage of this narrow useful initial lems in the past to extract UCN sources has been experimentally stud- neutron energy distribution is that the through cryogenic windows being ied in quite some detail over the past input beam into the UCN source can prone to growing deposits of 10 years. The relevant total slow neu- be tailored to provide exactly the absorbing contaminants. However, tron cross-sections of gaseous, liquid, required neutrons. Reflecting these such problems have been over- and solid deuterium have been mea- neutrons out of a white beam permits come as UCN accumulation and sured, as well as the integral UCN one to use the rest of the spectrum for window-less extraction from the production cross-section from a cold other applications and at the same superfluid has been realized [15]. neutron beam on a deuterium target. time minimizes the activation of The most recent experiments deliv- source equipment with useless neu- ered the UCN production as a func- trons that would only contribute to Solid Deuterium as Ultracold tion of incoming neutron velocity [20] backgrounds in most experiments. Neutron Converter and the generalized density of states A key issue for a long UCN life- The potential of solid deuterium as with inelastic [4, time in superfluid helium, besides the a good cold neutron (CN) moderator 21]. All observations, as the tempera- low temperature, is a low contamina- and UCN converter was also recog- ture dependence of UCN lifetimes, tion with 3He (3He/4He ≤ 10−12), which nized almost 30 years ago, experi- the total scattering and UCN produc- requires 4He purification. In principle, mentally in Gatchina [17] and tion cross-sections as well as the the lifetime (see Eq. 2) is only limited theoretically by Golub and Böning velocity dependent production cross- by neutron decay; in practice, how- [18]. It has been repeatedly demon- sections could so far be well described ever, also non-negligible wall losses strated in test experiments over the by a simple theoretical model [18] in contribute and may lead to lifetimes past years. A suitable deuterium UCN which the down-scattering via phonon on the order of ~200 s. This still rather converter has low temperature to creation is calculated using a simple long lifetime can be used most effi- avoid thermal up-scattering and high Debye model for solid deuterium. Multi- ciently by accumulating UCN inside ortho-deuterium concentration in phonon excitations in the calculation

Vol. 20, No. 1, 2010, Nuclear Physics News 21 facilities and methods

are needed and have been included to delivering an equally good description few seconds of production, the den- obtain agreement [20, 22]. of the measured data points [4, 21]. sity of the UCN gas reaches equilib- Figures 2 and 3 show results from rium and bombarding the experiments [20, 22] at the cold neu- The UCN Source at PSI—The First target further does not help. At this tron beam line FUNSPIN at the Paul High Intensity Source Based on instant a large shutter at the bottom of Scherrer Institut. The cold neutrons Solid Deuterium the storage volume is closed and the interacted with deuterium in a cryogenic The UCN source at the Paul proton beam is switched back to the target cell. Ultracold neutrons leaving Scherrer Institut (see ucn.web.psi.ch) other users of the facility (pion and the target were separated from the is a pulsed spallation source and uses muon production targets, spallation cold neutron beam and detected with solid deuterium for the generation of neutron source SINQ). Two main low background. Because these mea- ultracold neutrons [23]. It takes full (south, west) and one R&D guide surements could be calibrated by UCN advantage of the large production made from NiMo (85%/15%) coated production from deuterium gas, which cross-section and circumvents prob- glass tubes are selectable with UCN can be calculated, absolute cross- lems with the finite lifetime of UCN shutters at the storage volume and can sections could be extracted. Figure 2 inside the converter by decoupling the be used to deliver the neutrons on shows the number of detected UCN as production region from the storage demand to experiments in two experi- a function of target temperature. The region [19]. Figure 4a sketches the mental areas. It is expected that a typ- empty target at 85 K indicates the important components of the UCN ical experiment of up to a few background level, temperatures above source. Neutrons are produced by hundred liters of volume at the height 24 K are for gas, below that for liquid spallation when the proton beam from of the beamline can be filled with a −3 and below 18.7 K for solid deuterium. the PSI ring cyclotron hits a lead target. UCN density of about 1000 cm , The yield of UCN from the solid The basic principle is to use the full which is almost two orders of magni- increases with lowering the tempera- beam (proton kinetic energy: 590MeV, tude more than possible today. A ture because of suppressing the up- dc-beam current >2mA, beam power superconducting 5 T magnet at the scattering of UCN before extraction. >1.2 MW) in macropulses of a few end of the south beamline is used to Figure 3 shows the results for velocity seconds at 1% duty cycle. Almost 10 provide polarized UCN. Figure 4b dependent UCN production on solid neutrons (average energy of a few MeV) shows the source hall during con- ortho-deuterium at 8 K. A velocity per incident proton are generated and struction. Commissioning has started selector was used providing the indi- thermalized in the 3.5 m3 heavy water in fall 2009. An international collab- cated energy distributions. The mea- moderator. The typical lifetime of the oration (see nedm.web. psi.ch), is sured data is displayed as open simultaneously setting up an experi- neutrons inside the D2O moderator is circles. The red squares indicate cal- ~5 ms. The neutrons are further ment to search for the electric dipole culated results using the measured cooled in 30 liters of solid deuterium moment of the neutron and plans on velocity distribution for the specific data taking, starting in 2010. (sD2) at 5–8 K, in which neutron life- point as input, and the black line is the times of 30–40 ms can be achieved. In result of the calculation as a continu- the crystal some of the cold neutrons UCN Research and Source Projects ous function of energy. The simple are converted to UCN and can leave Worldwide Debye model results in the steep cut- the deuterium converter vessel Besides the effort at PSI in off at the Debye temperature of solid upward into vacuum. On top of a short Switzerland, there are activities for deuterium of about 110 K correspond- (1.2m) vertical guide a storage volume improved UCN sources proposed, ing to an energy of about 9 meV. This of about 2 m3 gets filled with UCN. planned, or under way at the ILL explains the importance of including The walls of the storage volume are Grenoble, at Los Alamos National multi-phonon processes in the calcu- coated with diamond-like carbon, pro- Laboratory (LANSCE), at the Mainz lation in order to describe the data at viding a high Fermi potential and low TRIGA reactor, at the Technical Uni- larger neutron energies. UCN losses. The 1.2 m ascent bal- versity of Munich for the FRM-II, at Very recently, the simple Debye ances the boost effect UCN experience PNPI Gatchina for the WWR-M reac- model could be replaced by the due to the Fermi potential of solid deute- tor, at Indiana University for LENS, measured phonon density of states, rium when leaving the solid. After a at the North Carolina PULSTAR

22 Nuclear Physics News, Vol. 20, No. 1, 2010 facilities and methods

development in the field of UCN 13. C. A. Baker et al., Phys. Rev. Lett. 97 physics over the next decade. (2006) 131801. 14. R. Golub and J. M. Pendlebury, Phys. Lett. 62A (1977) 337. References 15. O. Zimmer et al., Phys. Rev. Lett. 99 1. E. Fermi and W. H. Zinn, Phys. Rev. (2007) 104801. 70 (1946) 103. 16. C. A. Baker et al., Phys. Lett. A 308 2. V. K. Ignatovich, Ultracold Neutrons (2003) 67. (Clarendon Press, Oxford, 1990). 17. I. S. Altarev et al., Phys. Lett. 80A 3. R. Golub, D. J. Richardson, S. K. (1980) 413. Klaus Kirch, Philipp Schmidt- Lamoreaux, Ultra-cold Neutrons (Adam 18. R. Golub and K. Böning, Z. Phys. B51 Wellenburg, Geza Zsigmond, and Hilger, Bristol, 1991). (1983) 95. Bernhard Lauss. 4. Proceedings of International Work- 19. Y. Pokotilovski, Nucl. Instr. Meth. A shop on Particle Physics with Slow 356 (1995) 412. Neutrons, 29–31 May 2008, Grenoble, 20. F. Atchison et al., Phys. Rev. Lett. 99 Nucl. Instr. Meth. A 611(2–3) (2008). (2007) 262502. reactor, at RCNP Osaka, TRIUMF, 5. 7th International UCN workshop, 21. A. Frei et al., Phys. Rev. B 80 (2009) and J-PARC. 8–14 June 2009, St. Peterburg, http:// 064301. Over the past few years a consid- cns.pnpi.spb.ru/ucn/index.html 22. F. Atchison et al., Phys. Rev. C 71 erable number of institutions have 6. J. S. Nico and W. M. Snow, Annu. Rev. (2005) 054601. Nucl. Part. Sci. 55 (2005) 27; H. Abele, 23. A. P. Serebrov et al., JETP Lett. 66 entered the field of ultracold neutron Prog. Nucl. Part. Phys. 60 (2008) 1. (1997) 802. physics, for the development of new 7. N. F. Ramsey, Ann. Rev. Nucl. Part. sources as well as for their exploita- Sci. 32 (1982) 211. K. KIRCH tion in fundamental physics. We are 8. M. Pospelov and A. Ritz, Ann. Phys. Paul Scherrer Institut, now at the point that new sources will 318 (2005) 119. Villigen PSI, Switzerland 9. M. Raidal et al., Eur. Phys. J. C57 come online and new precision exper- ETH Zürich, Zürich, Switzerland iments will provide results with much (2008) 13. improved sensitivities within the 10. V. I. Lushchikov, Y. N. Pokotilovsky, A. V. Strelkov, and F. L. Shapiro, Sov. B. LAUSS, next few years. Major improvements Phys. JETP Lett. 9 (1969) 23. P. SCHMIDT-WELLENBURG, in technology have often produced 11. A. Steyerl, Phys. Lett. 29B (1969) 33. AND G. ZSIGMOND unexpected and spectacular results. 12. A. Steyerl and S. S. Malik, Nucl. Instr. Paul Scherrer Institut, CH-5232 We are eagerly waiting to see the Meth. 284 (1989) 200. Villigen PSI, Switzerland

Vol. 20, No. 1, 2010, Nuclear Physics News 23