Nucleon scattering on actinides using a dispersive optical model with extended couplings
Abstract
Tamura coupling model (Rev.Mod.Phys. 17 (1965) 679) has been extended to consider the coupling of additional lowlying rotational bands to the ground state band. Rotational bands are built on vibrational bandheads (eveneven targets) or single particle bandheads (odd targets) including both axial and nonaxial deformations. These additional excitations are introduced as a perturbation to the underlying axiallysymmetric rigid rotor structure of the ground state rotational band. Coupling matrix elements of the generalized optical model are derived for extended multiband transitions in eveneven and odd nuclei. Isospin symmetric formulation of the optical model is employed.
A coupledchannels optical model potential (OMP) containing a dispersive contribution is used to fit simultaneously all available optical experimental databases including neutron strength functions for nucleon scattering on Th, U and Pu nuclei. Quasielastic (,) scattering data on Th and U to the isobaric analogue states of the target nucleus are also used to constrain the isovector part of the optical potential. Lane consistent OMP is derived for all actinides if corresponding multiband coupling schemes are defined. For eveneven (odd) actinides almost all lowlying collective levels below 1 MeV (0.5 MeV) of excitation energy are coupled. OMP parameters show a smooth energy dependence and energy independent geometry. A phenomenological optical model potential that couples multiple bands in odd actinides is published for a first time.
Calculations using the derived OMP potential reproduce measured total crosssection differences between several actinide pairs within experimental uncertainty for incident neutron energies from 50 keV up to 150 MeV. The importance of extended coupling is studied. Multiband coupling is stronger in eveneven targets due to the collective nature of the coupling; the impact of extended coupling on predicted compoundnucleus formation cross section reaches 5% below 3 MeV of incident neutron energy. Excitation of multiple bands in odd targets is weaker due to the singleparticle nature of the coupling. Coupling of groundstate rotational band levels in odd nuclei is sufficient for a good description of the compoundnucleus formation cross sections as long as the coupling is saturated (a minimum of 7 coupled levels are typically needed).
pacs:
11.55.Fv, 24.10.HtI Introduction
The cross sections for the interaction of neutrons with actinide nuclei are crucially important for design of various fission reactor systems. Tight target uncertainties on the capture and inelastic scattering data for major actinides were derived from advanced reactor sensitivity studies by Nuclear Energy Agency WPEC Subgroup26 WPECSG26:08 (). However, the present status of evaluated nuclear data files for inelastic scattering on major actinides is not satisfactory as discussed at the IAEA Technical Meeting on “Inelastic Scattering and Capture Crosssection Data of Major Actinides in the Fast Neutron Region” INDC0597 () held in 2011. Significant differences in evaluated inelastic cross sections were observed in evaluated nuclear data files from 200 keV up to a few MeV of incident neutron energy for major actinides INDC0597 (). Such differences lead to large uncertainties in the inelastic scattering cross section (), as well as in calculated multiple neutron emission cross sections (,2) and (,3). The improvement of evaluated scattering cross sections and neutron emission spectra and reduction of their uncertainties for neutron induced reactions on actinides is an important issue that should initiate new theoretical studies.
Many theoretical models are being developed to describe nucleon scattering on actinide nuclei. The optical model is one of the fundamental theoretical tools which provides the basis of nucleonscattering data evaluations hodg63 (). An accurate calculation of the nucleon scattering from a welldeformed actinide nucleus using the optical model must include the coupling to the lowlying collective states. A very successful computational method to account for the importance of the multistep processes is the coupledchannels (CC) method bu63 () using Tamura’s formalism Tamura:65 (), which permits an exact solution of the CC equations. Many of deformed optical potentials suggested for actinides are based on Lagrange work at CEA, France la75 (); but coupledchannels equations were a formidable challenge for the computing capabilities of 70s and 80s. Many approximations were suggested and used; it was customary to couple only a few levels (usually 3) of the ground state rotational band in those calculations. In 2004 two of the authors studied the convergence of neutron cross sections on U as a function of the number of coupled target states in the ground state rotational band. It was shown that the common practice of calculating neutron cross sections with 3 coupled levels in groundstate bands is inadequate Soukhovitskii:04 (). This result was confirmed and generalized in a new comprehensive study by Dietrich et al. Dietrich:12 (), but none of these papers included additional vibrational bands in the coupling scheme of eveneven nuclei, nor additional rotational bands in odd nuclei.
Needed higher accuracy of data for fast reactors requires improving the description of scattering data at incident neutron energies from a few keV up to 5–6 MeV to cover the region with the maximum yield of fission neutrons. While the energies of excited states of the groundstate band of eveneven actinides below 500 keV are well described by a rigid rotor model, above 500 keV several vibrational bands are observed that need to be considered. The situation for odd actinides is even more complex, as no pairing gap exists, therefore lowlying excited states are dominated by rotational bands built on singleparticle (1QP) bandheads (e.g., particlehole configurations with , and , and corresponding rotational bands, dominate the lowlying excitation spectra for U and Pu nuclei). Therefore, lowlying rotational bands, built on vibrational bandheads for eveneven targets, and on singleparticle bandheads for odd targets, need to be taken into account to describe neutron inelastic scattering on actinides. This fact has long been recognized for eveneven nuclei; a vibrationalrotational description within the coupledchannels approach has been used to describe scattering data on eveneven actinides by University of Lowell group Chan:1982a (); Chan:1982b (); Sheldon:1986 (), and later used by Kawano et al. Kawano:94 (), Minsk group porsuk96 (); Maslov:06 () and BruyèresleChâtel group (e.g., Refs. Lopez:05 (); Romain:16 ()). Authors presented preliminary findings of this work for axialsymmetric eveneven nuclei in Ref. Quesada:2013 (). However, authors are not aware of similar published work undertaken for odd actinides^{1}^{1}1P. Romain has used extended coupling schemes for calculations of the +U reaction, but no results have been published., for which typically only the groundstate rotational band had been considered in optical model studies^{2}^{2}2Preliminary findings for oddnuclei were presented at ND2013 Quesada:2015 (), however current results supersede older publications as further understanding both in derivation and results was achieved.. Additionally, those works used nondispersive potentials to describe scattering data, except the dispersive potential used to describe neutron scattering on U target Lopez:05 () and other actinides Romain:16 ().
The main purpose of this comprehensive contribution is to improve the description of neutron scattering on actinides at incident neutron energies below 6 MeV by extending a previously derived dispersive optical model potential for nucleon induced reactions on actinides based on a rigidrotor description socaqu05 (); casoqu05 (); Capote:08 (). That previously published by authors rigidrotor potential of Refs. socaqu05 (); casoqu05 (); Capote:08 () will be referred in the rest of this work as the “RIPL 2408” potential by using the keyword from the RIPL opticalmodel database ripl3 ().
Special focus will be on predicted compoundnucleus formation cross section , which is a critical input quantity for statistical reaction modelling. It has been shown that strongly depends on employed coupledchannels couplings Dietrich:12 (). The proposed extension should account for multipleband couplings, while keeping the achieved quality of description of the whole set of scattering data (e.g., total cross section data above 5 MeV were described by the RIPL 2408 potential within the quoted experimental uncertainty of about 1.5%). The coupling of vibrational bands is expected to improve the description of neutron scattering on eveneven actinides in the energy region from 500 keV up to about 6 MeV, which is critical for fast neutron fission reactors. The coupling of rotational bands built on singleparticle bandheads for odd nuclei is expected to reduce observed discrepancies from 100 KeV up to 1 MeV incident neutron energy in inelastic scattering on major fissile actinides U, U and Pu INDC0597 (). The authors are unaware of previously published optical model studies for odd actinide targets involving coupling of multiple bands.
The paper is structured as follows. Sections II and III provide descriptions of the nuclear shape parametrization, nuclear hamiltonian and nuclear wave functions needed in the solution of the Schrödinger equation describing the nucleon scattering problem. Section IV contains the new coupling matrix elements derived within Tamura’s formalism allowing for multiple band coupling; this is one of key results of this paper. For interested readers a detailed derivation is given in Appendices 1 and 2. Section V describes the formulation of dispersive Laneconsistent potential for the simultaneous description of neutron and proton induced reactions on actinides. In section VI, we discuss derived optical potential parameters to describe nucleon scattering on actinides using a proper multiband level scheme; we also compare the derived potential with selected experimental data, including highly accurate averaged total crosssection differences measured for incident neutron energies from 50 keV up to 150 MeV. Comprehensive coupling schemes and multiband coupling parameters derived from a leastsquare fit for Th, U, and Pu targets are tabulated in Appendix 3. Finally, section VII contains our conclusions.
Ii Nuclear shape parameterization for extended coupling scheme
Actinides are well deformed nuclei, where lowlying collective levels are strongly excited in nucleon inelastic scattering. The employed nuclear shape parameterization combines a rigid rotor description of the ground state rotational band, with small quadrupolar and octupolar vibrations around the deformed equilibrium shape. Vibrations are described in the spirit of the softrotator model SRM1 (); SRM2 (), which considers nuclei to be deformable with both axial and nonaxial vibrations; deformations with are considered axial, assuming that such deformations are usually small. Rotational bands are built on top of vibrational (singleparticle) bandheads for eveneven (odd) nuclei, respectively. The deformed nuclear optical potential arises from deformed instant nuclear shapes of actinide nucleus. The instant nuclear shape in a bodyfixed system can be described as follows
where means spherical harmonics; and are the angular coordinates in the bodyfixed (intrinsic) system; are the static axial deformations, and are the quadrupolar and octupolar deformations, and and are the quadrupolar () and octupolar () nonaxial deformation parameters, respectively. We introduced the notation . We followed Bohr Bohr52 () and assumed that the bodyfixed frame is chosen to be the principal nuclear symmetry axes, such that and , therefore quadrupolar deformations are described by two parameters and . We also followed Lipas and Davidson Lipas:1961 () and assumed that octupolar vibrations with even projection () best describe the octupole contribution to the nuclear vibrations for the lowlying negative parity states. Therefore, the nonaxial deformations , and octupolar deformations are described by two parameters and (such assumption could be easily removed if needed by introducing one additional parameter). In what follows we will use the fact that for nonaxial contributions () in Eq. (II) only even values of the projection are allowed both for quadrupolar and octupolar vibrations.
Nuclear radii () are defined for five potential geometries described in the next section, being the target mass number. The soft rotator model of nuclear structure has been successfully applied in coupledchannels optical model analyses for many nuclei porsuk96 (); porsuk98 (); Lee:2011 (). However, until recently we were unable to derive a dispersive coupledchannels optical model potential for actinides based on softrotator couplings. Such work is still in progress. Dispersive optical model features a reduced number of parameters compared to traditional opticalmodel analyses, therefore parameter compensation of model defects become more difficult for dispersive potentials. Mathematically, the failure of the traditional softrotator model applied to actinides is due to the very slow convergence of the multipolar expansion of the softrotor potential around the spherical shape for large equilibrium deformations typically encountered in actinides. To solve this problem it was assumed that the quadrupole variable of the soft rotator model can be considered as a large equilibrium axial component plus a small contribution (i.e. ). This assumption allows a much better convergence of the potential expansion, as we use a truly small parameter , instead of the relatively large parameter originally used by Tamura Tamura:65 (). The departures from the axiallysymmetric rigidrotor shape are considered to all orders in quadrupole nonaxiality parameter and octupole nonaxiality parameter . Under these assumptions we can rewrite the Eq. (II) as follows
The first term of the sum in curly brackets in Eq. (II) corresponds to the axiallysymmetric equilibrium shape (rigid rotor). There are two additional quadrupole and octupole axiallysymmetric terms (proportional to and , respectively), and nonaxial quadrupole and octupole terms. If we further assume that the octupole deformation parameter is small, we can expand the nuclear potential around the equilibrium rigid rotor shape (instead of the equilibrium spherical shape used by Tamura – see Eq. (5) of Ref. Tamura:65 ()). If we insert the nuclear shape given by Eq. (II) into the optical potential, and make a first order Taylor expansion around the equilibrium (axial) shape, we can obtain the following expression for the deformed optical potential expansion
where

,

,
are the axiallysymmetric components of the deformed optical model potential, and the nonaxial components are included in Eq. (II) within curly brackets. A general expression for the multipolar expansion of the deformed optical model potential is derived in Appendix 1, see Eq. (Appendix 1. Multipolar expansion of the deformed optical model potential in the laboratory system).
Iii Nuclear Hamiltonian and targetnucleus wave functions
Excited states in actinides are usually classified into singleparticle and collective excitations, being the latter divided into vibrational and rotational states. Such division works well for eveneven nuclei due to the pairing gap; only collective states are located inside the gap. However, for odd actinides the energy of singleparticle excitations could be of the same order of the rotational energies, and so the independence of the singleparticle excitations from the collective movement is hardly justifiable. These cases have been extensively studied by Davidov and coworkers davidov60 (); davidov61 (); davidov62 (); davidovbook (), and Davidson davidson65 () (for additional information see references therein). Davidov showed that in eveneven nuclei whose equilibrium shape is close to the axiallysymmetric shape (e.g., actinides), rotation cannot be regarded independently of vibrations davidov61 (). For odd nuclei, the interaction of the nuclear rotation with the unpaired nucleon may change both the structure of the rotational spectra and the energies of the singleparticle excited states davidovbook (). Therefore, it is expected that lowlying excited states of odd nuclei have a complex structure which does not permit the separation of the singleparticle, rotational and vibrational degrees of freedom. However, all excited states in actinides are expected to have definite values of parity , being the orbital angular momentum, and of the total angular momentum . The analysis above clearly showed that eveneven and odd actinides have a very different nuclear structure.
Let’s define the nuclear structure model to be used in this work. We assume that a nucleus is comprised of the eveneven core where only paired nucleons are present. Additionally, for odd nuclei, we consider a single unpaired nucleon that moves in the nuclear mean field created by the eveneven core. We further assume that the nuclear ground state may be statically deformed. Dynamical deformations are assumed to be small, and , i.e., the adiabatic assumption holds for the separation of the rotational and vibrational motion for eveneven and odd targets. Under those assumptions, the nuclear Hamiltonian can be written as,
(4) 
where is the rotational energy operator; is the vibrational energy operator; is the energy operator of the unpaired nucleon (singleparticle operator); and is the interaction energy operator of the unpaired nucleon with the nuclear eveneven core field. The corresponding Schrödinger equation can be written as , where the nuclear eigenfunction describes singleparticle, vibrational and rotational motions.
iii.1 Eveneven nuclei
Following our assumptions for actinides, we do not consider unpaired nucleons for eveneven nuclei (nor singleparticle excitations), therefore the nuclear Hamiltonian describes the collective motion only, neglecting the interaction of vibrational and rotational states. Adiabatic approximation for the collective motion means that the eveneven nuclear wave function can be factorized into a rotational and a vibrational part , where the rotational wave function depends on the Euler angles , and the vibrational wave function depends on the number of excited vibrational phonons of multipolarity and, implicitly, on the corresponding vibrational deformation variables , , , and (e.g., represents quadrupolar axial and nonaxial phonons and depends on and quadrupolar deformations, respectively). The phonon parity is . In this work we consider only onephonon vibrational states. The phonon description of vibrational states also implies that interband transitions when the number of phonons changes by more than one () are forbidden (this is a well known selection rule of the vibrational phonon model). Therefore, strongest interband transitions occur between the bands built on onephonon states and the ground state band.
Assuming that the nucleus behaves as a triaxial rotor and denoting as the total angular momentum projection on the symmetry axis and as the total angular momentum, then the nuclear wave function is mixed in , and can be written in the form Bohr52 (); Preston75 ()
(5) 
The coefficients are the mixing coefficients that depend on all deformation parameters and satisfy orthonormality conditions
(6) 
where denotes the additional quantum numbers. The expression (5) can be considered a generalization of the wave function given in Eq. (91) of Ref. Bohr52 () and Eqs. (959) and (960) of Ref. Preston75 () (all restricted to quadrupolar deformations) for any phonon multipolarity. The phonon wave function has been ignored in those references, since for eveneven nuclei the lowestlying states are given by rotational and vibrational modes with intrinsic spin 0 Preston75 (). Note that if , then the condition holds and the phase factor becomes unity. The phonon parity operator in Eq. (5) is missing in Eq. (3) of Ref. Hecht62 () as only quadrupolar vibrations featuring positive parity were considered. In this work, the phonon parity has been considered as needed for the proper inclusion of both quadrupolar and octupolar vibrations.
The coefficients , which determine mixing in a given rotational state due to the nonaxiality, are to be determined from the solution of the rotational problem, and can be obtained as described in Ref. SHEMMAN () by solving the Schrödinger equation corresponding to the nonaxial soft nuclear Hamiltonian. The code SHEMMAN adjusts the softrotator Hamiltonian parameters by comparing the energies of the experimental and calculated collective levels. Therefore, the coefficients are determined independently from coupledchannels equations.
For axiallysymmetric nuclei the projection becomes a good quantum number and only one coefficient is different from zero. Therefore, the sum over and the mixing disappears in Eq. (5), and we obtain,
(7) 
The optical model calculations and fitting undertaken in this work assumed that eveneven nuclei were axially symmetric rotors in their ground state. However, nonaxial formulation of the wave function given by Eq. (5) was retained for the calculation of most general coupling matrix elements.
iii.2 Odd nuclei
Following our assumptions for odd actinides, we neglect vibrational excitations of the core. The total angular momentum of the nucleus , a constant of the motion, is now the sum of two parts, , the eveneven core’s angular momentum, and , the nucleon angular momentum, where . It is customary to take the projections of and along the intrinsic axis to be and , respectively, while the component of is . Since the potential in which the unpaired nucleon moves is, in general, neither spherically nor axially symmetric, , , and , are not constants of the motion. Following Refs. Hecht62 (); Preston75 (), we assume the extreme extracore singleparticle model for the Hamiltonian describing the unpaired nucleon. The nucleon singleparticle wavefunction is the eigenfunction of the singleparticle Hamiltonian that describes a nucleon moving in the deformed nuclear mean field (e.g., Nilsson or WoodsSaxon potential), i.e., , being the singleparticle energy corresponding to the singleparticle state .
Following Eq. (3) of reference Hecht62 () the nuclear wave function in the extreme singleparticle model is given by
(8) 
where is the parity of the intrinsic wave function, and are the mixing coefficients that depend on deformation parameters, and are determined from the solution of the rotational problem (e.g., see Refs. davidov60 (); davidov62 ()). In the asymmetric case ellipsoidal symmetry imposes additional restrictions on quantum numbers and (that characterizes the singleparticle wave function – see Appendix 2), namely that () must be an even integer Hecht62 (). This restriction on values is indicated by the apostrophe on the summation symbol in Eq. (8).
It is remarkable that the odd nucleus wave function given by Eq. (8) has exactly the same collective angular operator structure (i.e., Wigner functions) that the evennucleus wave function given by Eq. (5). Such analogy allows using the same matrix element derived for the eveneven nuclear wave function to obtain the odd matrix element as will be shown in Appendix 2. There are only two differences regarding the odd and eveneven nuclear wave functions: 1) the phase in front of the second Wigner function changes to (odd wave function) from the eveneven case phase ; and 2) there is a single particle wave function () that multiplies the first (second) Wigner functions in the odd case in Eq. (8).
For axiallysymmetric nuclei the sum over disappears, becomes a good quantum number that labels the singleparticle state (i.e., ), which automatically leads to Eq. (9139) of Ref. Preston75 ():
(9) 
Note that axial symmetry is an acceptable approximation for odd nuclei, as demonstrated by an extensive use of the rigidrotor assumption in oddactinide optical model potentials.
Iv Coupledchannels matrix elements for extended coupling scheme
Starting from the derived potential multipolar expansion of the optical model potential (see eq. (Appendix 1. Multipolar expansion of the deformed optical model potential in the laboratory system) in Appendix 1), and using nuclear wave functions given above, we can derive the general expressions (VII.1) and (VII.2) for eveneven and odd targets, correspondingly (see Appendix 2 for a detailed derivation) for coupling matrix elements between nuclear states and . In those expressions, the reduced matrix elements have to be defined, and the radial functions are given by Eq. (28), which for are being equal to the radial functions used in matrix elements of the conventional rigidrotor potential.
Reduced matrix elements entering Eqs. (VII.1) and (VII.2) are different for couplings in eveneven and odd nuclei as the corresponding nuclear wave functions given by Eqs. (5)–(9) are different. Ready to use expressions for calculations of the reduced matrix elements are given by Eq. (VII.1) for eveneven targets, and by Eq. (VII.2) for odd targets. The reduced matrix elements also define the selection rules determining couplings, which are discussed in Appendix 2.
The general expressions mentioned above constitute an important result of this work allowing the extension of Tamura’s coupledchannels formalism to describe the coupling of additional excited bands both for eveneven and odd targets.
An example of the coupling scheme employed in the current work is shown in Fig. 1 for a neutron induced reaction on U target. The corresponding effective deformations (e.g., ) that define the interband coupling strength are shown near the transitions. The highestenergy state considered in the ground state rotational band of U was the located at 775.9 keV; the rigid rotor assumption breaks down above that excitation energy (level energies do not follow the rigid rotor prediction ).
V Dispersive optical model formalism
Pioneering dispersive optical model analyses for nucleon scattering have been carried out by Lipperheide li66 (); li67 (), Passatore pa67 () and Lipperheide and Schmidt lisc68 (). Both bound and scattering states were calculated using the same nuclear mean field constrained by dispersive relations mang84 (); mangsa86 (); masa86 (); masa87np (); johoma87 (); masa88 (); masa89 (); fiwida89 (); dewara89 (); tochde90 (); masa91 (); masa91rev (); moro04 (). Recent works to improve the description of the bound states using the dispersive optical model, and describe reactions on unstable targets off the stability valley have been published by a group at Washington University, St. Louis, Missouri (see Ref. di10 () and references therein). Additional constraint imposed by dispersion relations helps to reduce the ambiguities in deriving phenomenological OMP parameters from the experimental data, and effectively also reduces the number of phenomenological parameters.
However, a few studies have been devoted to derive dispersive optical model parameterizations for strongly deformed nuclei, where coupledchannels formalism should include potential corrections arising from dispersion effects Merchant:92 (); rode97 (); sm01 (); sm02 (); sm04 (). Recently we derived an isospin dependent optical model potentials for actinides based on rigidrotor couplings casoqu05 (); Capote:08 (). We will summarize below the formalism employed in this work.
Assuming that the geometry of the imaginary terms of the OMP is energyindependent, then the deformed optical model potential for incident nucleons may be written as
(10)  
where the first term is the real smooth volume potential . Successive complexvalued terms are the volume, surface, and spinorbit potentials, all containing the corresponding dispersive contributions and discussed in section V.1. The Coulomb potential term is needed for incident protons. We followed our previous studies socaqu05 (); casoqu05 (); Capote:08 () and connected the imaginary spinorbit potential to the real spin orbit potential by a dispersion relation as discussed by Walter wa96 ().
The geometrical WoodsSaxon form factors are given as
(11) 
where deformed radii are described by Eq. (II). An spherical spinorbit potential was used with constant radius equal to . The deformed Coulomb potential was calculated using a multipole expansion of charged ellipsoid with a uniform charge density within the Coulomb radius and zero outside as suggested by Bassel et al. badrsa62 (). The spherical term of the Coulomb potential was calculated by taking account of the diffuseness of the charge density distribution of the form chiwya97 ().
The incident nucleon energy is equal to the incident energy for neutrons; for incident protons we have to take into account the modification of the nuclear potential by Coulomb repulsion . We assume that the “effective” interacting energy of the proton is , being , the target mass and atomic numbers. The term is an estimate of the kinetic energy loss of the incident proton in the interaction region due to Coulomb repulsion. The use of the “effective” interacting energy of the proton is a generalization of the previously used Coulomb corrections, which considers such corrections in all orders. The constant is an adjustable constant meant to account for the ”effective” radius of interaction of proton in nucleus. Avoiding the use of Coulomb corrections is a precondition to the Lane consistency, otherwise the resulting potential is not symmetric with respect to the nucleon charge.
The present optical potential includes relativistic corrections as discussed by Elton el66 (). Firstly, the nucleon wave number was calculated in the relativistic form where denotes the total energy of projectile, the projectile rest mass, and the light velocity. Secondly, projectile and target masses were replaced by corresponding relativistic energies in reduced mass formulae. However, the change of the potential depth related to the transformation from the Dirac equation was not considered in this work.
v.1 Dispersive relations
In a dispersion relation treatment, the real potential strength consists of a term which varies slowly with energy, the so called HartreeFock (HF) term, , plus a dynamic (polarization) term, , which is calculated using a dispersion relation. Under favorable conditions of analyticity in the complex plane the real part can be constructed from the knowledge of the imaginary part on the real axis through the dispersion relation
(12) 
where means that the principal value of the integral should be taken. Assuming that , where is the Fermi energy, Eq. (12) can also be be written in the subtracted form
(13) 
Here denotes the Fermi energy, determined as for neutrons and for protons, where denotes the separation energy of nucleon from a nucleus labeled by and The symmetry condition (for symmetric imaginary potentials) is used to extend the imaginary part of the OMP for energies below the Fermi energy, allowing for the calculation of the dispersive integral.
It is known that the energy dependence of the depth is due to the replacement of a microscopic nonlocal HF potential by a local equivalent pebu62 (). For a Gaussian nonlocality is a linear function of for large negative and is an exponential for large positive . Following Mahaux and Sartor masa91 (), the energy dependence of the smooth “HartreeFock” part of the nuclear mean field is taken as that found by Lipperheide li67 ():
(14) 
where the parameters and are undetermined constants. Eq. (14) can be used to describe the HF potential in the scattering regime masa91 () for .
It is useful to represent the variation of surface and volume absorption potential depth with energy in functional forms suitable for the dispersive optical model analysis, which are integrable analytically PRCdisp (). An energy dependence for the imaginary volume term has been suggested in studies of nuclear matter theory by Brown and Rho brrh81 ():
(15) 
where and are undetermined constants. An energy dependence for the imaginarysurface term has been suggested by Delaroche et al. dewara89 () to be:
(16) 
where and are undetermined constants.
The isospin dependence of the potential (the Lane term la62a (); la62b ()) was considered in real and imaginary surface potentials as follow,
(17)  
(18) 
where and are undetermined constants. Many authors found that the imaginary volume potential does not depend on the isospin.
For the energy dependence of the spherical spinorbit potential we used the functional form suggested by Koning and Delaroche kode03 (), which is convenient for the calculation of the dispersive contribution PRCdisp (), namely:
(19)  
(20) 
where and are undetermined constants.
v.2 High energy behavior of the volume absorption
The DOM analysis of neutron scattering on Al mocaqu02 () showed the importance of the dispersive contribution to describe data for energies above 100 MeV using a nonsymmetric version of the volume absorptive potential for large positive and large negative energies as proposed by Mahaux and Sartor masa91 (). Similar behavior was confirmed in Th dispersive coupledchannels analysis of nucleon induced reactions socaqu05 () and in combined dispersive coupledchannels analysis of nucleon induced reactions on Th and U casoqu05 (); Capote:08 (). We use the same formalism in this work, which is briefly described below.
Following Mahaux and Sartor masa91 (), the assumption that the imaginary potential is symmetric about (according to equation ) is modified above some fixed energy , which is expected to be close to 60 MeV. However this value is fairly arbitrary masa91 () and we will use it as a fitting parameter. Let us assume the imaginary potential to be used in the dispersive integral is denoted by , then we can write masa91rev ()
(21) 
and
(22) 
These functional forms are chosen in such a way that the function and its first derivative are continuous at . At large positive energies nucleons sense the ”hard core” repulsive region of the nucleonnucleon interaction and diverges like . Using a model of a dilute Fermi gas hardsphere the coefficient can be estimated to be equal to 1.65 MeV masa86 (), but the actual value is a model parameter. On the contrary, at large negative energies the volume absorption decreases and goes asymptotically to zero. The asymmetric form of the volume imaginary potential of Eqs. (21) and (22) results in a dispersion relation that must be calculated directly from Eqs. (12) and (13), and separates into three additive terms vaweto00 (); CPC (). Therefore, we write the dispersive volume correction in the form
(23) 
where is the dispersive correction due to the symmetric imaginary potential of Eq. (15) which is calculated following Ref. PRCdisp (), and the terms and are the dispersive corrections due to the asymmetric terms of Eqs. (21) and (22), respectively calculated as described in Ref. CPC (). Unfortunately, the equation giving the dispersive correction in Ref. CPC () contains several typos, so we give the correct expression below.
Using the notation , the contribution for volume nonlocality correction is exactly:
The resulting dispersive correction for the asymmetric case starts to increase already for energies above 50 MeV, making a significant contribution to the real part of the OMP at high energies. It should be noted that nonlocality corrections (Eqs. (21) and (22)) can be used either for the volume or surface imaginary potential; however, Mahaux and Sartor masa91 () have shown that nonlocality consideration for the surface imaginary potential has a very small effect on the calculated cross sections. Therefore in this work we followed Ref. masa91rev () and only considered the effects of nonlocality in the volume absorption.
v.3 Dispersive coupledchannels optical model analysis
A survey of the experimental data for nucleon interaction on actinide nuclei spanning from 0.001 to 150 MeV used in the current work coincide with the data used by Soukhovitskii and coworkers about 10 years ago Soukhovitskii:04 (). The total cross section data considered cover all the critical energy points which are necessary to reveal the structure due to the Ramsauer effect and extends up to 150 MeV. The experimental database available for actinide nuclei other than Th and U is very scarce, especially in the high energy range; but some angular distributions of scattered neutron, and total cross section measurements are available for fissile actinides and were considered in the fitting. Quasielastic scattering (,) data on Th and U were also considered in the fit to fix the isovector terms of the potential as described in Ref. Quesada:2007 ().
The starting values of static deformation parameters , , and were taken from FRDM deformations theoretically derived by Möller and Nix Moller:95 (). Evaluated neutron strength functions for actinide nuclei, and and potential scattering radius ripl3 (); CSWEG1991 () were used for finetuning deformation parameters together with measured (,’) angular distributions on excited states close to the Coulomb barrier ( MeV) for actinide nuclei of interest. The minimization procedure used the quantity as described in previous references socaqu05 (); casoqu05 (); Capote:08 ().
Both direct and statistical processes contribute to nucleonnucleus elastic scattering at these energies. However, according to our estimation, the statistical processes are not important above 3 MeV on actinides so we neglect them in the OMP derivation. The direct processes, increasingly dominant at higher energies, can be described by the optical model.
The customary coupledchannels calculations were performed by coupling the ground state rotational band with the selected members of vibrational (singleparticle) bands as shown in Appendix 3 for studied actinides. Optical model code OPTMAN sochiw05 (); somoch04 () was used for OMP parameter fitting. We were using symmetric surface and nonsymmetric volume imaginary absorptive potentials as described in previous section. The dispersion integrals were calculated using analytical solutions CPC (); PRCdisp () of dispersion relations with Eq. (V.2) for the contribution.
In our formulation of the OMP in Eq. (10) the geometrical parameters of the “HartreeFock” potential and are in general different from geometrical parameters of the volume and surface absorptive potentials; however the real and imaginary spinorbit terms share the same geometric and parameters. Also the surface and volume dispersive corrections share the same geometry of corresponding imaginary potentials and , respectively. The employed actinide optical potential energy dependence is very simple and we use energyindependent geometry and the same OMP parameters for both neutron and proton projectiles. A rather weak dependence of the dispersive potential geometry on mass number was previously observed casoqu05 (); Capote:08 (). The fitted dispersive potential parameters are listed in Table 1. These parameters have to be complemented by a corresponding coupling scheme, ground state deformations, and coupling strength parameters (listed in Appendix 3, tables 3, 4, 5, 6, 7, and 8) to become a complete OMP parameter set for studied actinides.
VOLUME  SURFACE  SPINORBIT  COULOMB  
Real  MeV  MeV  MeV  
potential  MeV  –  MeV  
parameters  MeV  
MeV  MeV  MeV  
Imaginary  MeV  MeV  MeV  
potential  MeV  MeV  
parameters  MeV  MeV  
fm  fm  fm  fm  
Potential  fm  fm  fm  fm 
geometry  fm  
fm 
Vi Results and discussion
Quantity  Reference  Th  U  U  U  Pu 

this work  0.86  0.97  0.95  1.02  1.15  
,(eV)  RIPL 2408 OMP casoqu05 (); Capote:08 ()  0.86  0.93  0.93  0.92  1.18 
RIPL3 evaluation ripl3 ()  0.84 (.07)  0.90 (.05)  0.98 (.07)  1.03 (.08)  1.20 (.10)  
Porodzinski et al. posuma98 ()  0.80 (.08)  1.07 (.14)  1.01 (.10)  1.17 (.10)  1.25 (.13)  
Mughabghab 2006 mu06 ()  0.71 (.04)  0.98 (.09)  0.98 (.07)  1.29 (.13)  1.3 (.10)  
this work  1.66  1.41  1.60  1.68  2.05  
,(eV)  RIPL 2408 OMP casoqu05 (); Capote:08 ()  1.72  2.20  1.93  1.72  2.19 
RIPL3 evaluation ripl3 ()  1.50 (.30)  –  –  1.60 (.20)  –  
Mughabghab 2006 mu06 ()  1.35 (.04)  –  1.8 (.30)  2.17 (.19)  2.30 (.40)  
CSEWG 1991 CSWEG1991 ()  1.60 (.60)  –  1.8 (.30)  1.70 (.30)  2.30 (.40)  
this work  9.68  9.60  9.58  9.52  9.50  
,fm  RIPL 2408 OMP casoqu05 (); Capote:08 ()  9.68  9.56  9.51  9.64  9.49 
Mughabghab 2006 mu06 ()  9.65 (.08)  9.75 (.15)  9.63 (.05)  9.60 (.10)  9.48 (.10)  
CSEWG 1991 CSWEG1991 ()  9.65 (.30)  9.75 (.20)  9.65 (.10)  –  9.60 (.10) 
The derived dispersive potential parameters shown in Table 1 for actinide nuclei combined with deformation parameters and coupling strengths tabulated in Appendix 3 were tested against selected experimental data. Average resonance parameters are consistently described by our calculations as can be seen in Table 2. A similar level of agreement in calculated strength functions is seen for rigidrotor OMP results (RIPL 2408) within quoted evaluated uncertainties. Current results for are closer to evaluated ones for U and U, but marginally worse for Pu. For , calculations using the derived potential are closer to evaluations than those using the RIPL 2408 potential, but evaluated uncertainties are large.
The aim of this work is to improve the description of neutron inelastic scattering on actinides at incident neutron energies from keVs to a few MeVs. Unfortunately, many neutronscattering physical observables (e.g., angular distributions of elastic and inelastic scattering, inelastic scattering cross sections, etc) measured on actinides for incident neutron energies below 3 MeV are influenced by a compound nuclear decay that can not be described by a coupledchannels optical model alone, which is a direct reaction model. Total cross section is one measured observable not affected by compound nuclear processes.
Total cross sections for U and Th nuclei were calculated with the derived potential parameters. An excellent agreement with data in the whole energy range is achieved as seen in Fig. 2. Similar agreement is also shown for the RIPL 2408 optical model potential based on a rigidrotor structure. Calculated total cross sections do not show an improvement for a new potential as both calculated results are within quoted experimental uncertainty. However, it is remarkable that the newly derived potential with extended couplings preserves the achieved agreement with total cross section and other scattering observables shown in Refs. casoqu05 (); Capote:08 (), while allows predicting directinteraction cross sections for additional coupledlevels in the whole energy range of interest.
From the physical point of view, a newly derived OMP utilizes a better description of the underlying nuclear structure compared to commonly used rigidrotor potentials. The inadequacy of the rigid rotor description is clearly seen in the deviation from the rule of the energies of excited states with large in the ground state rotational band of eveneven actinides. E.g., for U and Th targets increasingly larger differences from the rule can be seen for levels with in the ground state rotational band, corresponding to excitation energies above keV. Such differences are mainly due to the change in moment of inertia (stretching of the soft nucleus). Additionally, mixing increases for higherspin states as e.g., shown in a softrotator structure description SRM1 (); SRM2 (). Therefore, the accuracy of the rigidrotor description of the nuclear structure in eveneven actinides rapidly deteriorates above 500 keV.
A lowestenergy vibrational mode in actinides is the octupolar vibration (e.g., the octupolar bandhead in U is located at 680 keV as shown in Fig. 1). This vibrational mode results in an enhanced neutron inelastic scattering cross section to negative parity levels , , , … of the lowlying octupolar band due to the strong band coupling. As a result the neutron inelastic scattering cross sections increases from 680 keV up to 1 MeV in the region of utmost importance for fast reactors. An smaller increase in inelastic scattering cross sections of actinides is also observed due to the strong coupling of the ground state band to other vibrational bands (e.g.,  and vibrations). The OMP derived in this work offers a consistent description of the scattering of low energy neutrons on actinides that goes beyond the assumption of rigid rotor nuclear structure for those nuclei. However, the question is how to show a clear advantage of new potential using available experimental data?
It is generally accepted that differences of neutron total cross sections among neighboring nuclei provide an unusually stringent test of optical models shbera80 (); caphwh84 (); cadiph86 (); dianba03 (). At the same time it has been shown that the standard optical model treatment fails to reproduce the observed differences of total cross sections for tungsten isotopes dianba03 (); incident neutron energies below 5 MeV are especially challenging due to the increasing impact of the target nuclear structure on calculated cross sections. Indeed, when the incident neutron energy becomes comparable with the energy of excited nuclear levels, then the weakcoupling assumption is not valid as the coupling becomes strong. In such conditions a commonly used DWBA method typically overestimates the observed scattering cross sections on vibrational levels as shown in Ref. INDC0597 (). Only a coupledchannels calculation with a consistent optical model potential would allow describing the observed cross section due to the strong coupling of vibrational bands.
We have already shown a good description of total cross sections in Fig. 2. However, a description of total crosssection differences is a much greater challenge. A natural test of the success of our approach is to check calculated total cross section differences for several pairs of actinides vs measured data at neutron incident energies below 5 MeV, where nuclear structure plays an important role. Energyaveraged total cross section data measured by Poenitz in 1981 and 1983 Poenitz:81 (); Poenitz:83 () on Pu, U and Th targets have been selected to benchmark our potential. Poenitz high accuracy data goes from 50 keV neutron incident energy up to 20 MeV, and data were measured under the same conditions and at the same installation, thereby minimizing any uncertainty in the ratio. Energyaveraged total cross sections for Th and U nuclei measured by Abfalterer et al. abbadi01 () from 5 to 200 MeV were also used to calculate the corresponding experimental ratio and associated experimental uncertainty. An small shift () of Poenitz data was applied for the experimental Th,U ratio for a better matching of Abfalterer et al. abbadi01 () data, well within the estimated uncertainty of the vertical scale (0.02 dianba03 ()) arising from uncertainties in the areal densities of the employed actinide targets.
Calculated total neutron cross section data for Th and U and Pu nuclei were used to obtain the energydependent ratio of the total cross section difference of targets and to the corresponding averaged total cross section for the RIPL 2408 potential casoqu05 (); Capote:08 () and the potential derived in this work. Calculated ratios are compared with experimental ratios derived from Poenitz and Abfalterer data in Fig. 3 for the following target pairs: Pu,Th, Pu,U, Th,U, and U,U. The measurements are very well reproduced for all target pairs by the dispersive OMP parameters shown in Table 1 combined with the corresponding target deformation parameters and coupling strengths tabulated in Appendix 3. Therefore, the derived OMP for all five studied nuclei is validated. Rigidrotor (RIPL 2408) potential casoqu05 (); Capote:08 (); ripl3 () description of data is of inferior quality, especially for incident neutron energies below 3 MeV, where nuclear structure effects have the largest impact. Note that differences seen in the figure are typically around 2% being the largest for Pu,Th at 50 keV (reaching about 5%). Such differences are not noticeable in Fig. 2.
The depicted high quality description of data with derived potential at low incident neutron energies is due to the improved description of the nuclear structure of target nuclei considered in present calculations. The fact that such description was achieved with exactly the same regional potential for all targets with an energyindependent geometry, is an evidence of the predictive power of the dispersive optical model when combined with a proper description of the nuclear structure of target nuclei.
The elastic and inelastic angular distributions on selected groundstate band (GSB) coupled levels for the U reaction below 3.0 MeV were not included in the OMP fit as compoundnucleus contribution needs to be considered at those energies. A comprehensive calculation of U reactions was recently undertaken in Ref. Capote:2014 (); calculated angular distributions below 3 MeV were shown to be in very good agreement with existing experimental data. An equally good agreement was also shown for angular distributions of scattered neutrons (including contributions from the lowest levels of the ground state rotational band) measured at higher incident neutron energies from 4 up to 15 MeV Capote:2014 ().
Dietrich et al. Dietrich:12 () showed an strong impact on calculated compoundnucleus formation cross section , of the number of coupled levels in the rigidrotor model. Calculated for U reactions using rigidrotor RIPL 2408 potential (dashed line) is compared in Fig. 4 with calculations using the new potential with full coupling (green solid line), or with coupling reduced to GSB levels (black solid line). Left panel shows results for U target; calculated from the potential derived in this work is much lower than the RIPL 2408 results (using 5 coupled levels) below 3 MeV. The observed difference is about 100 mb above 1 MeV (%) and reaches 300 mb (%) at 100 keV. Such differences are mainly due to the lack of saturation of the GSB coupled levels as discussed by Dietrich et al. Dietrich:12 (). If we restrict the number of coupled levels for the current potential to 7 GSB levels ( in Table VII) the results remain practically the same below 3 MeV. An small difference is observed at higher incident neutron energies due to the extended coupling for U target, being the full coupling result slightly lower than the one with GSB coupling. This result was expected as the singleparticle coupling of excited bands in odd nuclei is much weaker than the collective coupling for vibrational bands in eveneven nuclei. For many applications, it will be a very good approximation to couple 7 levels or more of the GSB in odd nuclei, and neglect the multipleband coupling.
Right panel in Fig. 4 shows calculations for U target. Again significant differences are seen below 3 MeV with the RIPL 2408 results (using 5 coupled levels) compared to present work mainly due to the lack of coupling saturation. For this eveneven target, stronger coupling due to vibrational collective levels induces larger differences between the full coupling results (green solid line) and those calculated with the GSB couplings (black solid line). Such differences reach 3%, while no differences were practically observed for odd targets. The impact of collective levels on calculated remains important from 50 keV up to 3 MeV. Therefore, we may conclude that multipleband coupling including vibrational bands is very important for eveneven targets to accurately calculate in the whole energy range of interest for applications.
Vii Conclusions
Tamura’s coupling formalism Tamura:65 (); Tamura:65a () has been extended to consider lowlying bands of vibrational (singleparticle) nature observed in eveneven (odd) actinides. These additional excitations are introduced as a perturbation to the underlying rigid rotor structure that is known to describe well the ground state rotational band of major actinides.
A dispersive isospin dependent coupledchannels optical model analysis of nucleon scattering on actinides nuclei from 1 keV to 150 MeV has been undertaken. A new optical model potential is derived which couples multiple vibrational (singleparticle) bands and is valid for eveneven (odd) targets. The isovector terms and the very weak dependence of the geometrical parameters on mass number allowed applying the derived potential parameters to neighboring actinide nuclei with a great confidence., provided that a suitable multiband coupling scheme could be defined. A single set of potential parameters given in Table 1 is able to describe all available scattering data on major actinides provided that the given coupling schemes and effective deformation parameters are used. While the potential is common for all actinides, the coupling scheme depends on the lowlying excitations of a given target. Fitted deformations of the ground state rotational band are in reasonable agreement with FRDM deformations theoretically derived by Möller and Nix Moller:95 () as seen in Table 3.
Excellent agreement in the whole energy range between calculations using derived dispersive coupledchannel optical model potential and the experimental total cross section for Pu, U and Th nuclei is obtained. A better nuclear structure description of the target nucleus also allowed an excellent description of total crosssection differences between pairs of actinide nuclei for neutron incident energies from 50 keV up to 3 MeV, fulfilling the main goal of this study: to improve the description of the neutron scattering cross sections on actinides at low neutron incident energies.
In summary, we have shown that a proposed dispersive coupledchannel phenomenological optical model with multipleband coupling is capable of predicting “optical” nucleon induced cross section on actinides at a few percent level and gives an excellent description of the total crosssection differences among neighboring actinide nuclei (both odd and eveneven) from 50 keV up to 150 MeV. Multipleband coupling is needed for a proper calculation of compoundnucleus formation cross sections on eveneven targets in the whole energy range of interest. However, it will be a very good approximation to couple only groundstate band levels in odd nuclei as long as the coupling is saturated (i.e., the cross section does not change with increasing number of coupled levels).
Current results inspire confidence in the use of the proposed OMP with the corresponding coupling schemes to improve evaluations of neutron scattering data at low neutron incident energies as needed for many applications. Open issues like the use of the softrotator model to describe structure of eveneven targets within the generalized optical model potential will be addressed in future works.
Acknowledgements
This work was partially supported by International Atomic Energy Agency, through the IAEA Research Contract 13067, by the European Union and Japan under the ISTC Project B1319, by GrantsinAid for Scientific Research (18560805) of JSPS and by the Spanish Ministry of Economy and Competitivity under Contracts FPA201128770C0302, FPA201453290C22P and FIS201128738C0201. We would like to thank anonymous referee for the extensive and detailed revision of the paper and valuable comments and suggestions which significantly improved the original manuscript. We also would like to thank our colleagues T. Kawano and Ian Thompson for their very useful advises and encouragement.
Appendix 1. Multipolar expansion of the deformed optical model potential in the laboratory system
Let’s derive the expressions for nuclear potential expansion in spherical harmonics. If we insert nuclear shape Eq. (II) into the optical potential to get deformed nuclear potential, and make a Taylor expansion up to the first order, we can obtain the following expression for the potential expansion (similar to Eq. (II) in section II)
The expansion was made around the equilibrium axiallysymmetric spheroidal rotor shape (instead of the spherical shape used by Tamura Tamura:65 ()), where
(26) 
The radial shape corresponds to the shape used in the traditional rigid rotor model, and is also mirror symmetric (as only even are allowed in the sum). We know that the rigid rotor model is an excellent approximation for the ground state rotational band of actinide nuclei. On the other side, it is clear that and differ. The coupledchannel potential derived by using a Taylor expansion around the spherical shape does not describe the nucleon scattering on deformed nuclei with large deformation (e.g., rare earth, actinides, …). Using the rigid rotor axial potential as a zero approximation solves this problem, therefore the aforementioned expansion is a key point in our derivation. It should be noted that =, i.e., the axial rigid rotor corresponds to the stable static deformation if , , and deformation parameters are equal zero.
Introducing the notation:


,
the multipole expansion of both quantities (with ) in terms of the spherical harmonics with axial symmetry is performed:
(27) 
where
(28) 
Note that the radial function corresponds to the Legendre expansion coefficient of the rigid rotor potential, which is usually already available in existing coupled channel codes.
Replacing the introduced notations into Eq. (Appendix 1. Multipolar expansion of the deformed optical model potential in the laboratory system) leads to: