Florida State University Libraries
Electronic Theses, Treatises and Dissertations The Graduate School
2018 Clustering in Light Nuclei with Configuration Interaction Approaches Konstantinos Kravvaris
Follow this and additional works at the DigiNole: FSU's Digital Repository. For more information, please contact [email protected] FLORIDA STATE UNIVERSITY
COLLEGE OF ARTS AND SCIENCES
CLUSTERING IN LIGHT NUCLEI WITH CONFIGURATION INTERACTION
APPROACHES
By
KONSTANTINOS KRAVVARIS
A Dissertation submitted to the Department of Physics in partial fulfillment of the requirements for the degree of Doctor of Philosophy
2018 Konstantinos Kravvaris defended this dissertation on May 11, 2018. The members of the supervisory committee were:
Alexander Volya Professor Directing Dissertation
David Kopriva University Representative
Ingo Weidenhoever Committee Member
Simon Capstick Committee Member
Laura Reina Committee Member
The Graduate School has verified and approved the above-named committee members, and certifies that the dissertation has been approved in accordance with university requirements.
ii ACKNOWLEDGMENTS
This thesis would not have been possible without the constant encouragement of Professor Alexan- der Volya. His patience was a critical ingredient throughout the research and writing process. It was always a pleasure discussing research, administrative and general topics. I am also grateful for the committee members’ guidance when necessary. I would also like to thank Professor Jorge Piekarewicz for being a great teacher and exposing me to more problems in nuclear physics. Finally, special thanks goes to Professor Efstratios Manousakis for all his help before and during my PhD studies. I would also like to thank everyone who made the non-research part of my studies enjoyable and interesting, Maria Anastasiou, Konstantinos Kountouriotis, Katerina Chatziioannou, David Fernandez, Hailey Nicol (soon to be Fernandez), and Daniel Lanford. Finally, for navigating the bureaucratic maze, I would like to thank Jonathan Henry, Brian Wilcoxon and Felicia Youngblood.
iii TABLE OF CONTENTS
List of Tables ...... vi List of Figures ...... viii List of Symbols ...... ix List of Abbreviations ...... x Abstract ...... xi
1 Introduction 1
2 Methods 5 2.1 Harmonic Oscillator ...... 5 2.1.1 Coordinate Transformations ...... 6 2.2 Configuration Interaction Approach in a HO Basis ...... 9 2.2.1 HO Basis Expansion ...... 9 2.3 Traditional Shell Model ...... 11 2.4 No Core Shell Model ...... 12 2.5 Resonating Group Method ...... 12 2.6 Spectroscopic Factors ...... 15 2.6.1 Traditional Spectroscopic Factors ...... 16 2.6.2 Normalized Spectroscopic Factors ...... 16 2.6.3 Dynamic Spectroscopic Factors ...... 17 2.7 Coupling to Continuum ...... 18 2.8 Rotational Bands ...... 20 2.8.1 Elliott’s SU(3) Model for Rotations ...... 22
3 Constructing Cluster Channels 23 3.1 Antisymmetry ...... 23 3.2 The Gloeckner-Lawson Procedure ...... 24 3.3 SU(3) Method ...... 25 3.4 Center of Mass Boosting Method ...... 27 3.5 Constructing the Channels ...... 29
4 Traditional Shell Model Studies 33 4.1 α Particle Wave Function ...... 33 4.2 Clustering in the sd Shell ...... 34 4.2.1 The Case of 20Ne...... 37 4.2.2 Evolution of Clustering with Increasing Nucleons ...... 39 4.3 Superradiance in 13C and α Clustering ...... 41
iv 5 No Core Shell Model Studies 43 5.1 The Nucleus 8Be ...... 43 5.2 The Nucleus 9Be ...... 46 5.3 The Nucleus 10Be...... 48 5.4 The 3α System ...... 50
6 Conclusions 52
Appendix A Second Quantization 54
B CM Boosting Operator Normalization Coefficients 56
C Calculation of Radial Integrals with HO Wave Functions 59
D CM Boost and Recoupling Examples 61
E Hamiltonian Kernels for the 2α System 65
F Tables of Spectroscopic Amplitudes 67
Bibliography ...... 82 Biographical Sketch ...... 94
v LIST OF TABLES
3.1 Select configuration content of NCSM wave functions for 4He withhω ¯ = 20 MeV boosted to a CM boosted state with n = 4 and ` =0...... 29
3.2 Moshinsky coefficients for different mass ratios d = AD/Aα coupling to a final HO wave function with CM quantum numbers N = L = 0 and relative quantum numbers n = 2, ` =0...... 31
4.1 Transition rates and α spectroscopic factors for the 20Ne rotational band members with the USDB interaction...... 37
4.2 Comparison between new and experimental SF for low-lying states in 21Ne...... 40
4.3 Energies, widths in MeV and α spectroscopic factors (structural) for the two lowest 3/2+ states in 13C...... 42
5.1 Alpha decay widths and spectroscopic factors for the first 3 states in 8Be using different oscillator parameters ¯hω. The S(dyn) are calculated with the lowest energy solution of 8 the RGM in each case projecting on the Nmax = 4 corresponding Be wave function. Widths are calculated with a fixed channel radius of R = 3.6 fm ...... 45
5.2 Contributions to the Hamiltonian from different interaction components in MeV. Then numbers in parentheses indicate number of nodes n and angular momentum ` in the relative coordinate of the 2α channel. V0,1 correspond to the different |Tz| values for the interaction. The second column refers to contribution to a single α particle. We use the JISP16 interaction withhω ¯ = 25 MeV...... 46
5.3 Negative parity band energies and SF for 9Be. The numbers in parentheses denote the number of available quanta in the relative motion between the 2 clusters. The SF are calculated withhω ¯ = 20 MeV. The sixth column marked with a single star refers to dynamic SF calculated with additional channels built on the 5He(1/2−) state. The final column marked with two stars refers to a calculation with channels restricted to having an 8Be +n form...... 48
5.4 Dynamic SF for rotational band members in 10Be. The last columns labeled with a (*) refers to calculations with channels constructed with the excited 2+ state of 6He. The states belonging to the two different rotational bands are separated by a line. The SF is evaluated with the dominant channel(s)...... 48
5.5 Leading SU(3) irreducible representation components and particle-hole content of states recognized as belonging to rotational bands in 10Be. The ordering is the same asin5.4...... 50
vi 5.6 Ground state binding energy and excitation energies for rotational band members of 12C using the NCSM and RGM with ¯hω = 20 MeV...... 51
D.1 Index ordering (k) of single particle orbitals. The ± superscirpt refers to the m = ±1/2 projection of each orbital...... 62
D.2 Index ordering (k) of neutron single particle orbitals for the sp space. The proton indexing follows the same order but offset by 8...... 63
vii LIST OF FIGURES
1.1 Ikeda diagram showing the location of different modes of clusterization and the re- spective decay thresholds...... 3
2.1 Broad alpha clustering states in light nuclei. The data for 10Be are taken from [88] and for 18O from [89] ...... 19
3.1 A schematic depiction of the process of transfering the quanta of motion from the CM of an α cluster to the relative motion between the α and a daughter nucleus. The primes indicate the (invariant) internal wave functions of both nuclei...... 30
4.1 Ab initio calculations for the ground state energy (left) and the s4 component of 4 the He wave function (right) with the JISP16 interaction spanning multiple Nmax truncations and ¯hω values...... 34
4.2 Old (dashed) and New (solid) spectroscopic factors comparison with experimental data from [115, 25] using the USDB [61] interaction. Red (blue) poinst correspond to α knockout (pickup). The dotted line corresponds to a calculation with a realistic α wave function, obtained using the NCSM withhω ¯ = 14 MeV and Nmax = 8...... 35
4.3 Comparison between (new) α SF calculated with different wave functions for the α particle. For the Nmax = 0, 4, 8 there are 1, 3, 5 channels open respectively...... 36
4.4 (a) Experimental population of states in 20Ne from (6Li,d) and (b) theoretical pre- dictions for SF in a (0+1)ph calculation of 20Ne. To match the experimental figure, energy is increasing to the left. Details of the calculation are given in main text. Figure (a) is reproduced with permission from [123]...... 38
4.5 Low-lying RGM, full USDB and experimental spectra of 21Ne...... 40
5.1 Low-lying spectra of 8Be calculated with the JISP16 interaction at varioushω ¯ values in units of MeV...... 44
5.2 RGM calculations of the rotational band of 8Be for ¯hω = 20 MeV with the JISP16 interaction and with the α particle wave function taken from a NCSM calculation with the corresponding value of Nmax. The value N in the plot reflects the total number of HO quanta available to the relative motion between the two clusters...... 47
5.3 Cluster rotational bands in 9Be. N refers to the total number of quanta available in the relative motion; even (odd) N corresponds to negative (positive) parity bands. Top (bottom) plot corresponds to positive (negative) parity states...... 49
viii LIST OF SYMBOLS
π 3.1415926 ... ¯hc 197.327 ... MeV fm n The number of nodes of in the radial coordinate, n = 0, 1, 2 ... l The orbital angular momentum of a single particle orbital. m The z - axis projection of the angular momentum. ` The relative orbital angular momentum of two clusters. N The total number of oscillator quanta, N = 2n + l.
Nmax Number of allowed excitation quanta in No Core Shell Model. s, p, d, f . . . Spectroscopic notation for orbitals with l = 0, 1, 2, 3 ... α Ln(x) Generalized Laguerre polynomial.
Ylm(θ, φ) Usual Spherical Harmonic. JM Cj1m1j2m2 Clebsch Gordan Coefficient coupling j1m1 and j2m2 to JM [41]. Mn`NL Moshinsky Coefficient for a transformation with angle ϕ [45] ϕ n1l1n2l2 Mn`NL n1l1n2l2 Coefficient for a transformation to the CM of equal mass particles (ϕ = π/4)
ix LIST OF ABBREVIATIONS
HO Harmonic Oscillator CI Configuration Interaction SM Traditional Interacting Shell Model NCSM No Core Shell Model RGM Resonating Group Method CSM Continuum Shell Model Harmonic Oscillator Representation HORSE of Scattering Equations SF Spectroscopic Factor NN Nucleon-Nucleon JISP J-Inverse Scattering Potential
x ABSTRACT
The formation of sub-structures within an atomic nucleus, appropriately termed nuclear clustering, is one of the core problems of nuclear many-body physics. In this thesis, we put forward a new method for the study of nuclear clustering relying on the completely microscopic Configuration Interaction approach. We construct reaction cluster channels in a Harmonic Oscillator many- body basis that respect the symmetries of the Hamiltonian, are fully antisymmetrized, and carry a separable and controlled Center of Mass component. Such channels are then used to explore cluster signatures in Configuration Interaction many-body wavefunctions. The Resonating Group Method is then applied, utilizing the reaction channels as a basis to capture the essential cluster characteristics of the system. We investigate the emergence of nuclear clustering in 2α, 2α + n, 2α+2n and 3α systems using a No Core Shell Model approach from first principles, and traditional Shell Model studies of clustering in heavier nuclei.
xi CHAPTER 1
INTRODUCTION
One of the central questions that nuclear many-body physics seeks to answer revolves around the emergence of regularities such as nuclear shell structure, rotational bands and vibrational states to name a few, all observed in nuclear spectra [1, 2]. It has proven to be difficult to devise a model that describes the properties of atomic nuclei across the nuclear chart, and at various excitation energies, let alone from a completely ab initio view; namely starting from QCD interactions or at least interactions that describe the nucleon - nucleon phase shifts and properties of mass A = 3, 4 nuclei. While certain patterns, seen across the nulear chart, such as rotations and vibrations of the nucleus, are somewhat understood, nuclear clustering is yet another exotic phenomenon that is more difficult to study not only from a theoretical prespective, but also due to lack of experimental data. Nuclear clustering, the subject of this thesis, refers to the formation of substructures within the atomic nucleus. Light nuclei, with states that can be almost perfectly described as two or more substructures in relative motion, exhibit perhaps the most extreme degree of clusterization. Clusterization greatly affects not only the spectra of nuclei but also, and perhaps more importantly, the reaction rates for astrophysical processes. In this work we take a completely microscopic point of view in our approach. The main question we seek to answer concerns the emergence of clustering from nucleon-nucleon (NN) interactions, however, we also seek a general method that can probe the cluster features of nuclear systems and to study the role of coupling to the continuum of reaction states in altering spectroscopic charac- teristics of highly excited states.
The concept of a nucleus being formed as an assortment of α particles has existed since the beginning of nuclear science [3, 4, 5]. This idea was predominantly supported by both the fact that the α particle is very well bound and requires a rather large amount of energy to excite, and by experimental evidence showing α particle emission as the main decay channel for heavy elements.
1 This shows that the quartet arrangement involving the 2 protons and 2 neutrons that makes up the α particle is somehow energetically favorable. In support of clustering, early theories [3, 4, 6] emphasized that the binding energies of N = Z nuclei seem to follow a linear relation with the number of bonds that can be made between the α constituents of the nucleus. With the discovery of the neutron [7] these α models became disfavored, and attempts to describe the nucleus from its single particle constituents gained momentum. While geometric arrangements of α particles in multi-α nuclei were the first ones to be con- sidered, different types of fragmentation were subsequently probed, identifying a vast number of excited nuclear states possessing cluster characteristics that contained non−α fragments [8, 9]. These excited clustered states were found to lie at energies near the respective cluster decay thresh- old, prompting Ikeda [10] to exemplify this property in the, now famous, Ikeda diagram (Figure 1.1). This idea ties in nicely with the Gammow picture of α decay, in which an α particle is formed in the nucleus and at some threshold energy can penetrate the Coulomb barrier. From an astrophysical prespective, the most important clustered state is the excited 0+ state in 12C. In 1954 Sir Fred Hoyle predicted [11, 12] the existence of this state as a neccessary doorway for the creation of 12C in the universe. Given that the 2α system 8Be is unbound, and there exists no bound A = 5 system, the formation of elements heavier than A = 4 would otherwise be hindered. From this argument the state is expected to have a strong 3α character, a result verified experimentally as it is seen to decay into a 3α final state, predominantly through the short-lived 8Be 0+ ground state resonance [13, 14, 15, 16, 17]. In the 1970s a large body of experimental data was collected aimed at elucidating the cluster aspects of light nuclei, see [18, 19, 20] for some specific examples, or [21] for a more complete survey of experimental data. The main tools were the α transfer reactions, such as (6Li, d), (7Li, t), and α knockout reactions such as (p, pα). These reactions were found to be selective to populating or depopulating states with some degree of clusterization. The nuclear shell model was seen as a conflicting view of nuclear structure at the time due to its single particle nature. Attempts to describe α clusters from within the shell model resulted in the powerful SU(3) symmetry based techniques [22, 23]. Despite some degree of success in select cases [24, 25], there were still significant deviations from experiment, making further theoretical work regarding quantitative agreement neccessary. This led to the rise of the Resonating Group Method (RGM) [26, 27], which aimed
2 α. α. α. α. α. α. α. α. α. α. α. α. α. α. α. α. α. α. α. α. 7.27 14.44 19.17 28.48 12.C 12.C α. 12.C α. α. 12.C α. α. α. 7.16 11.89 21.21
16.O 16.O α. 16.O α. α.
4.73 14.05 12.C 12.C Ex 13.93
Ikeda Diagram 20Ne. 20Ne. α.
9.32
24Mg.
Figure 1.1: Ikeda diagram showing the location of different modes of clusterization and the respective decay thresholds. at an explicit description of the nucleus as a multi-fragment clustered state, while maintaining a microscopic description for the cluster fragments and employing a fully antisymmetric wave function on the nucleonic level. Later, clustering was shown to emerge from a nucleon-nucleon (NN) interaction describing the NN phase shifts [28]; The Green’s Function Monte Carlo calculations showed the formation of two distinct centers in the density distribution of 8Be. At about the same time the method of Antisymmetrized Molecular Dynamics began to be extensively applied to the problem of molecular- type states in clustered nuclei [29]. Such states are known to exist in N > Z nuclei [30, 31], with the extra neutrons forming valence bonds between the α (or heavier) cores, similar to the role of electrons in molecules. Relying heavily on Group Theory, symmetry based approaches were also applied, treating the relative motion of the α particles as bosons, and, borrowing from molecular physics, exploiting triangular and tetrahedral spatial symmetries to describe the spectra, finding great success in both 12C and 16O [32, 33]. To this day, nuclear clustering is still under intense theoretical and experimental investigation. On the theory side, mean field approaches [34], lattice models [35, 36], large-scale ab initio shell
3 model calculations [37] and Bose-Einstein condensate wave functions [38, 39, 40] have all been em- ployed to better understand how the clusters are formed, and which aspects of nuclear structure and reactions are most affected by the formation of substructures within the nucleus. Experimentally, the identification of clustered states becomes harder as the states are often broad (Γ ∼ 1 MeV) and quite high in excitation energy. The work presented here is organised as follows. In Chapter 2, we review the properties of the quantum Harmonic Oscillator (HO) and the related Configuration Interaction (CI) methods that are used throughout this thesis. Both the phenomenological traditional interacting shell model (SM) and the ab initio No Core Shell Model (NCSM) are introduced. The Resonating Group Method (RGM) is discussed as a way to merge the structure and reaction aspects of the nuclear system. We define the different types of Spectroscopic Factors (SF) and compare the methods of evaluating the degree of clustering in a nuclear state. Finally, a method for coupling the physics of the continuum to a finite quantum system is described. Chapter 3 serves to explore the various methods of creating explicitly clustered states within the HO many particle basis. We demonstrate our novel method that controls the Center-of-Mass (CM) coordinate of a many-body wave function in order to construct translationally invariant binary and ternary cluster channels. This approach overcomes various problems concerning the construction of α cluster channels and is general enough to be applicable to various clustering scenaria. Chapter 4 contains several explorations of clustering in various light nuclei using the Traditional Shell Model. Chapter 5 uses the RGM as a way to explore clustering from completely ab initio principles. Finally, Chapter 6 concludes the thesis by providing a brief review and summary of all the topics covered. Select theoretical topics, tables, and derivations can be found in the Appendices.
4 CHAPTER 2
METHODS
In this chapter we describe the methods used in this work. First we discuss the HO Hamiltonian and the properties of the wave functions. We briefly go over the basics of rotational bands in low- lying nuclear spectra and Elliott’s SU(3) model. Variational solutions to the many-body problem in terms of HO expansion of the wave function are reviewed, followed by discussion of the RGM. We discuss the concept of the spectroscopic factor and present different approaches to defining it. Finally, we show how the physics of the continuum can be incorporated in the shell model through the Feschbach projection formalism.
2.1 Harmonic Oscillator
The Harmonic Oscillator (HO) is perhaps the most widely encountered system in physics. It is defined by a quadratic confining potential and the Hamiltonian is given by p2 1 H = + mω2r2. (2.1) 2m 2 As the HO is analytically solvable, it becomes useful in approximating the solution of more complex systems near the stable point. In quantum mechanics, the wave function of a single particle trapped in a spherically symmetric HO potential is given by
s 2 3 n+l+2 l − −1/2 2 n! r −r2/2b2 l+1/2 r h~r|nlmi = ψnl (r) = ψnl (r, θ, φ) = b 2 π e L Yl (θ, φ), m m (2n + 2l + 1)!! b n b2 m (2.2) p where b = ¯h/mω is the so-called characteristic length of the oscillator. For alternative forms of the HO wave function see [41]. Here, the radial part of the wave function is positive near the origin. A phase convention used by some authors has an extra phase factor of (−1)n that assures that the radial wave function is positive as r → ∞. The energy corresponding to the wave function with quantum numbers |nlmi is E = ¯hω(N + 3/2), where N = 2n + l (2.3)
5 is the total number of HO quanta. Since the potential is a second power of r and the |nlmi are eigenstates we can invoke the Virial theorem and obtain for the expectation values of the kinetic T = p2/2m and potential V = mω2r2/2 energy ¯hω 3 hnlm|T |nlmi = hnlm|V |nlmi = 2n + l + . (2.4) 2 2 The off-diagonal matrix element for the kinetic energy reads
¯hω p hnlm|T |n + 1lmi = (n + 1)(n + l + 3/2), (2.5) 2 and since the Hamiltonian is diagonal in its eigenbasis |nlmi, the off diagonal matrix elements for the r2 operator are nearly identical
¯h p hnlm|r2|n + 1lmi = − (n + 1)(n + l + 3/2). (2.6) mω The range of the wave function can be evaluated by the rms value for r s p ¯h 3 hr2i = 2n + l + . (2.7) mω 2
The wave functions ψnlm (r) form a complete basis and can be used to expand another single-particle function. If the expansion is truncated by N, eq. (2.7) gives a measure of the range of validity of the expansion. While the HO possesses attractive properties, some of which are discussed later, it suffers from a gaussian fall-off of the wave function, a direct consequence of the potential being infinite as r → ∞. In real systems, the potential at large distances becomes 0, thus leading to a wave function decaying as exp (−r), not as exp −r2 which are the asymptotics of the HO wave functions. As a result systems that are loosely bound require a large number of excitations to be properly described in the HO basis [42].
2.1.1 Coordinate Transformations
Apart from the HO Hamiltonian being exactly solvable for a single particle, another advantage comes from its symmetry property under orthogonal transformations of coordinates in the many- body case. For example, when considering two non interacting particles in a HO potential, the Hamiltonian takes the form p2 1 p2 1 H = 1 + mω2r2 + 2 + mω2r2 , (2.8) 2m 2 1 2m 2 2
6 which allows us to write down the wave function of the system as a product of HO wave functions
Ψn1l1m1n2l2m2 (r1, r2) = ψn1l1m1 (r1)ψn2l2m2 (r2) (2.9)
Since the Hamiltonian is a rotational scalar, we can couple the two single particle wave functions into a total wave function with good angular momentum λ using the usual Clebsch Gordan coefficients
X Ψλµ (r , r ) = [ψ (r ) × ψ (r )]λ = Cλµ ψ (r )ψ (r ). (2.10) n1l1n2l2 1 2 n1l1 1 n2l2 2 µ l1m1l2m2 n1l1m1 1 n2l2m2 2 m1m2
The Moshinsky or Talmi-Moshinsky recoupling scheme [43, 44] reflects the symmetry of the Hamiltonian to remain invariant under orthogonal coordinate transformations by an angle ϕ
r cos ϕ sin ϕ r + = 1 , (2.11) r− sin ϕ − cos ϕ r2
2 2 2 2 due to the orthogonality of the transformation, r1 + r2 = r+ + r−. Therefore the potential part of the Hamiltonian remains invariant under a coordinate canonical transformation. This is a subset of a more general set of canonical transformations that leave the Hamiltonian unchanged. The eigenstates of the multi-particle system are highly degenerate as the energy is given only by the total number of quanta; in the two particle case E = 2n1 + l1 + 2n2 + l2. The coordinate rotation leaves us within the same representation of the symmetry group, which in turn implies that the total wave function in the new coordinates (2.10) can be written as a linear sum of the wave functions in the old coordinates that are members of this irreducible representation
X n l n l ;λ Ψλµ (r , r ) = M + + − − Ψλµ (r , r ). (2.12) n+l+n−l− + − ϕ n1l1n2l2 n1l1n2l2 1 2 n1l1 n2l2
This requirement enforces HO quanta and parity conservation
2n+ + l+ + 2n− + l− = 2n1 + l1 + 2n2 + l2 (2.13)
(−1)l++l− = (−1)l1+l2 . (2.14)
The transformation coefficients are known as the Moshinsky brackets and are matrix elements of the Wigner D-matrix. They depend on the angle of rotation between the two sets of coordinates
7 ϕ hn+l+n−l−|n1l1n2l2iλ ; see [45] for derivations of symmetries amongst different angles and permu- tations of indices. An alternative shortened notation used here for the brackets is defined as
Mn+l+n−l−;λ ≡ hn l n l |n l n l iϕ. (2.15) ϕ n1l1n2l2 + + − − 1 1 2 2 λ
The angular momenta coupled here, ~λ = l~1 +l~2, always obey the triangle conditions. The symmetry is limited to particles in oscillator potentials with the same frequency ω. Up until now, we have implicitly assumed that the particles are of the same mass. To treat 0 √ different mass particles (m1, m2) we can first use transformed coordinates (ri = miri), to eliminate an explicit dependence on mass from eq. (2.8) as
p02 p02 1 H = 1 + 2 + ω2 r02 + r02 (2.16) 2 2 2 1 2
0 p A particular transformation of the new coordinates (ri) is the one by an angle tan ϕ = m2/m1, which results in the new set of coordnates
√ 0 √ 0 m1r1 + m2r2 r+ = √ (2.17) m1 + m2 √ 0 √ 0 m1r1 − m2r2 r− = √ , (2.18) m1 + m2 √ with the Hamiltonian remaining invariant. A final transformation of R = r+/ m1 + m2 and p r = r− m1m2/(m1 + m2) yields the following form for the Hamiltonian
P2 p2 1 1 H = + + Mω2R2 + µω2r2, (2.19) 2M 2µ 2 2 where we have defined the total M = m1 + m2 and reduced µ = m1m2/M masses of the system and the new coordinates are given by
m1r1 + m2r2 R = , (2.20) M
r = r1 − r2. (2.21)
We recognize these as the CM and relative coordinates of the system. This exercise shows that the coefficients for the transformation to the CM of the system are the ones corresponding to the angle p tan ϕ = m2/m1. The frequency of the HO for this transformation to be applicable needs to be the same for all constituents. There is a lot to gain from the availability of such a transformation; the HO basis is recognized as the only basis of expansion where the CM motion of a system of particles
8 can be separated exactly from the internal (Jacobi) coordinates [44, 46]. Since the recoupling can work between any HO states, we will use this to our advantage to construct states that have a well defined wave function in the relative coordinate between two clusters and an HO wave function for the total CM motion with N = 0 quanta. Varying definitions, derivations of the Moshinsky coefficients, sample values and symmetry properties can be found in references [43, 47, 48, 49, 50].
2.2 Configuration Interaction Approach in a HO Basis
We begin this section by describing how to use the variational method to solve for single particle motion in a potential. We then discuss how the method can be generalized in the full Configuration Interaction approach for the many-body problem, and conclude with discussion of the Traditional Shell Model (SM) and the newer No Core Shell Model (NCSM) approaches to nuclear structure.
2.2.1 HO Basis Expansion
As an introductory example and explanation of our RGM strategy in Section 2.5, we solve the problem of a single particle with its motion governed by a Hamiltonian H with rotational symmetry.
One way to go about this is to expand the unknown wave function φ`m (r) in terms of some known complete basis of functions and then obtain the expansion coefficients variationally. A special case of this is the HO expansion assuming a fixed ` and m ∞ ∞ X X φ`m (r) = Cnψn`m (r) = CnRn`(r)Y`m (θ, φ) (2.22) n=0 n=0 where we seek to determine the coefficients Cn. The infinite sum is truncated at some value nmax which converts the minimization problem to an eigenvalue equation
n Xmax Hnn0 Cn0 = ECn, (2.23) n0=0
0 where Hnn0 = hn`m|H|n `mi. The diagonalization yields not only an estimate for the ground state of the particle, but also the excited states. The Ritz-Rayleigh variational principle ensures that the ground state eigenvalue obtained this way is an upper bound to the actual solution to the problem. The upper bound for the ground state energy is dependent on the oscillator parameter ¯hω and has minimum at some optimal ¯hω value. As the size of the basis grows, the minimum becomes broader and thus thehω ¯ dependence of the ground state energy estimate diminishes. It is not a priori
9 guaranteed that the same oscillator parameter as the one for the ground state will yield the best estimate for all excited states. The method can be extended to the many-body problem, with the expansion now carried over many-body Slater determinants Φi, or other antisymmetrized configurations, thus guaranteeing full antisymmetrization of the solution
X hr1, r2 ... |Ψi = Ψ(r1, r2 ... ) = ciΦi(r1, r2 ... ). (2.24) i The single particle states used in the Slater determinant can again be HO wave functions and therefore all properties obtained will be ¯hω dependent. With a large enough basis all observables will converge to their exact values.
Depending on the orbitals included in the Slater determinant, we can define a number Ntot which is given by the sum of all HO quanta in the single particle wave functions of the determinant. At the same time, for an A fermion system, there exists a number Nmin which corresponds to the minimum number of HO quanta the system must have due to the Pauli exclusion principle. The truncation of the basis is performed based on the number of allowed excitations from this minimum number Nmax. For a space with a Nmax truncation of the basis, we use only Slater determinants that have Nmin ≤ Ntot ≤ Nmax. While the single particle case has matrix dimensions that scale linearly with Nmax, the many body basis size“explodes” combinatorially with the number of allowed quanta, reaching well in the millions even for a modest number of excitations Nmax = 6 in light nuclei. Although we discussed here the expansion in terms of HO wave functions, the method is quite general, allowing for any type of single particle functions to be used. If we choose to give up the symmetry properties associated with the HO, we can use wave functions defined as solutions to the mean-field Hartree-Fock equations as they are the ones that will bring us closer to the true ground state of the system expressed with a single Slater determinant, potentially reducing the size of space required to achieve convergence of results. In both cases the inclusion of the excited configurations and subsequent diagonalization of the Hamiltonian in the configuration space yields what is known as the Configuration Interaction (CI) variational method for solving the quantum many-body problem. In the next sections we will examine two variations of CI approaches, the Shell Model (SM) and No-Core Shell Model (NCSM).
10 2.3 Traditional Shell Model
The SM was one of the first CI methods used to study nuclear structure, and still it remains one of the most popular methods to approach the nuclear many-body problem. The large jumps in particle separation energies as one moves across an isotopic (or isotonic) line are indicators of the so-called magic numbers 2, 8, 20, 28, 50, 82, 126 which in turn can be explained by the picture of independent particles filling up orbital shells [51, 52]. In contrast to the electronic structure of atoms (another system exhibiting such magic number structure) there is no externally created
field like the Coulomb potential. Nevertheless, nucleons themselves create a mean field (VMF) and one can use it for nuclear structure studies along with the residual interactions of the Hamiltonian
(Vres)
H = Trel + V = Trel + VMF + Vres. (2.25)
Gradual filling of the spherical mean field orbitals leads to the shell structure observed in nuclei. The model, just like atomic models, assumes that some of the nucleons in the core will be inert, at least for the low-lying states, and treats the remaining (valence) nucleons as active. This assumption works well, however, multiple effects, such as core polarization are treated effectively. The nucleons used in the shell model are now not bare nucleons, but dressed quasi particles whose interaction is expected to reproduce the spectra of nuclei. The SM is in some way a configuration space model, containing effective particles with effective properties and effective operators. All have appropriate structures and symmetries, but semi- phenomenological parameters are used (effective proton/neutron charges, etc). For initial estimates, Hartree Fock mean field or simple HO are commonly used. In particular, the residual part of the interaction can be obtained by either performing a series of unitary transformations on a realistic Hamiltonian [53, 54, 55, 56], selecting phenomenologically the coordinate form of the residual NN interaction and fitting the parameters to experiment[57, 58], or directly fitting the two body matrix elements of the interaction to reproduce a set of experimental data [59, 60, 61]. The last of these approaches has proven to be an accurate predicting tool for nuclear spectra, reduced transition rates and SF values, establishing the SM as one of the core techniques in nuclear physics.
11 2.4 No Core Shell Model
Building on the successes of the SM, the NCSM treats all A nucleons in a nucleus as active, point-like non relativistic fermions. In this non-perturbative approach, the NN interactions are used without any additional renormalization. The method is the many-body extension of the variational HO expansion. Due to the basis size explosion discussed earlier, various truncation and expansion schemes are employed in an attempt to circumvent the problem [62, 63, 64], with varying degrees of success. The many-body Schr¨odinger Hamiltonian in the NCSM has the form
2 1 X (pi − pj) X X H = + V (ri, rj) + V (ri, rj, r ) + ... (2.26) A 2m 2b 3b k i 2.5 Resonating Group Method The Resonating Group Method (RGM) was introduced in 1937 by Wheeler [26] to deal with a problem of a nucleus being partitioned into multiple different sub-structures. In the RGM, the wave function of a nucleus is projected onto the relative motion of sub-structures, which are the clusters of interest. For multiple reasons, it is an attractive model to study clustering as discussed in references [73, 74]. Here we present some key elements of the RGM which carry over into our approach 1. The wave function is totally antisymmetrized. 2. The non orthogonality of different factorizations is treated exactly. 12 3. The interaction used can be a nucleon - nucleon interaction designed to fit particle-particle scattering. 4. The structure and reaction parts of the problem are treated on equal footing. 5. The CM motion of the problem is also treated exactly. While many RGM calculations have been performed since the 1960s [74, 27], the RGM combined with the NCSM in the NCSM/RGM method [75, 76] has recently re-emerged in ab initio approaches. We outline here the basic principles of the RGM and defer the details of how it will be combined with the shell model to later chapters. In the RGM the binary cluster wave function can be written as a totally antisymmetrized product 0 0 0 0 hξ1ξ2ρ|Ψ i = Ψ (ξ1, ξ2, ρ) = A Ψ1(ξ1)Ψ2(ξ2)f(ρ) , (2.27) where we use primes to denote wave functions depending only on internal coordinates, we define 0 Ψ1,2 the wave functions describing the internal dynamics of the two clusters (ξ1, ξ2 are internal cluster coordinates) and f(ρ) is the wave function for the motion between the CM coordinates of the two clusters (ρ = RCM,1 − RCM,2). The total CM motion wave function is ommitted in all expressions in this section. We are interested in variationally calculating the relative wave function f(ρ). This can be done by expanding the function using a set of known functions. The δ functions were used originally, representing a channel where the two cluster fragments are a definite distance (here ρ0) apart 0 0 0 0 0 hξ1ξ2ρ|Φch(ρ )i = A Ψ1Ψ2δ(ρ − ρ ) (2.28) Z 0 0 0 0 hξ1ξ2ρ|Ψ i = f(ρ )|Φch(ρ )idρ . (2.29) The wave function f(ρ) is to be determined variationally using the Hamiltonian of eq. (2.26). Using eq. (2.29), the variational condition transforms into the Hill-Wheeler equation [77, 78, 79] Z H(ρ, ρ0) − EN (ρ, ρ0) f(ρ0)dρ0 = 0. (2.30) The functions H(ρ, ρ0) and N (ρ, ρ0) are known as the Hamiltonian and Norm kernels, respec- tively. They are found as matrix elements of the Hamiltonian and Norm operators in the channel 13 space 0 0 H(ρ, ρ ) = hΦch(ρ)|H|Φch(ρ )i (2.31) 0 0 N (ρ, ρ ) = hΦch(ρ)|1|Φch(ρ )i. (2.32) The existence of the norm kernel enforces the Pauli principle at the channel level and controls normalization and orthogonality. Since the Norm kernel is positive semidefinite, all eigenvalues are non-negative, with the ones that are equal to zero representing the Pauli forbidden states. The positive semidefiniteness can be demonstrated by expanding in some orthonormal basis function k set |Φchi X k |Φch(ρ)i = bk(ρ)|Φchi (2.33) k 0 X ∗ 0 X ∗ 0 hΦch(ρ )|Φch(ρ)i = bk(ρ )bm(ρ)δkm = bk(ρ )bk(ρ), (2.34) km k which, if we pick an arbitrary function g(ρ) and evaluate the expectation value yields Z Z 2 X ∗ 0 ∗ 0 0 X g(ρ)bk(ρ)g (ρ )b (ρ )dρdρ = g(ρ)bk(ρ)dρ ≥ 0. (2.35) k k k When solving the RGM equations, our first step is the diagonalization of the Norm kernel and the subsequent discarding of any eigenvector with a corresponding eigenvalue below some pre-set threshold governed by numerics. The δ-function choice for the expansion is not unique; in our work we choose a basis consisting of HO wave functions to expand the relative motion, allowing for both the Hamiltonian and Norm kernels to be written down as matrices. For convenience we choose the same HO basis as the one used in the NCSM for the cluster and parent nucleus. In the HO expansion, the full channel wave function |χ(`)i for a given set of asymptotic quantum numbers (which in the following we abbreviate with a single label `) is given as an expansion of basis channel states (Φn`) 0 0 hξ1ξ2ρR|Φn`i = ψ0(R)A ψn`(ρ)Ψ1(ξ1)Ψ2(ξ2) (2.36) (`) X (`) |χ i = χn |Φn`i. (2.37) n In the Slater determinant HO basis, the CM motion is unavoidable and thus we explicitly include it in the channel with a N = 0 HO quanta wave function. The expansion is similar to the one-body 14 variational CI treatment discussed in Section 2.2.1, however the bases used here are not orthogonal. The quantum number n here is used only for the number of nodes in the relative motion of the (`) clusters. The full function f(ρ) is now determined via the coefficients χn . We re-write the RGM equation (2.30) in the HO channel basis X (`) (`) X (`) (`) Hnn0 χn0 = E Nnn0 χn0 , (2.38) n n where the matrix elements are given by: (`) Hnn0 = hΦn`|H|Φn0`i (2.39) (`) Nnn0 = hΦn`|Φn0`i. (2.40) P (`)∗ (`) (`) The channel normalization requires n χn Nnn0 χn0 = 1. For large n, where the two fragments have a significant probability of being found seperated by a distance that makes the effects of the Pauli principle negligible, the basis channels Φn` become approximately orthogonal. The removal of the Pauli blocked states is done by working in the image space of the Norm kernel. The channel (`) P (`) vectors |χ i, and the associated relative wave functions F`(ρ) = n χn Φn`, have proper orthog- (`) 0(`) onality and normalization properties hχ |χ i = δEE0 . With this reduction of the RGM equation to a matrix diagonalization problem, the only remaining task is the construction of the HO channels which will be the topic of the next chapter. 2.6 Spectroscopic Factors In the study of clustering and reactions we are concerned with the role nuclear structure plays in rearrangement reactions or fragmentation into clusters; the Spectroscopic Factor (SF) is a relevant quantity [80, 81]. We limit our discussion to weak decays where the Fermi Golden rule and lowest order perturbation theory are valid, allowing for a formal definition of the SF. In this section we discuss various ways the SF have been historically defined and their relations, as well as a sum rule for the strength distribution. 15 2.6.1 Traditional Spectroscopic Factors The traditional SF is a measure of the overlap of a state with a particular cluster-core configu- ration. For single nucleon pickup reactions for example, the traditional SF is given by: 2 † 2 (A + 1Jf ||a`j||AJi) (old) X † S`j = hA + 1Jf Mf |a`jm|AJiMii = , (2.41) 2Ji + 1 Mf m where the initial |AJiMii and final |A + 1Jf Mf i states belong to nuclei with A and A + 1 particles, respectively. The double bar here refers to the reduced matrix element as defined in the Wigner- Eckart theorem [41]. The detailed balance relation for pickup, knockout, or decay processes is observed (old) S`j (i → f) 2Jf + 1 (old) = . (2.42) 2Ji + 1 S`j (f → i) In this definition, we ignored the recoil of the heavy nucleus in the decay process. This results in some of the strength being absorbed into CM-excited states in the calculation. To correct for this, one uses the so-called recoil coefficient which we will discuss in the next chapter. The SF can be generalized to cluster (many-particle) transfer/decay processes defined by the relative partial wave and total angular momentum of the process, and subsequently constructing a multi-particle operator in second quantized form that can take us from the initial (target) nuclear many-body state to states belonging to the final system. Many-body correlations play an important role in the calculation of SF, with the well-known example of significant enhancement in pair transfer by superconducting pairing correlations [82]. 2.6.2 Normalized Spectroscopic Factors Unlike single-particle operators that also provide basis states for the many-body problems, cluster operators and resulting channels are not following the same normalization and orthogonality conditions. The whole fractional parentage picture is obscured by multiple ways of fragmentation and by Pauli blocked components. Mathematically this is controlled by the Norm Kernel. Let us start with the traditional SF and discuss how they are calculated from the reaction channels with the HO expansion for the relative motion. The parent state wave function here will be denoted by ΨP and the channel follows the definitons of Section 2.5. The coefficients Fn` = hΨP|Φn`i, (2.43) 16 are fractional parentage amplitudes, leading to the following expression for the traditional spectro- scopic factor (old) X 2 S` = |Fn`| (2.44) n Here, ` is an asymptotic observable but n is an internal channel label. Under the assumption that channels are orthonormal, eq. (2.44) is just a sum of probabilities of transition to each n. As discussed in [83, 84], the implicit assumption is that the relative motion channels obey proper normalization conditions both for bound state and scattering boundary conditions. We see that 0 the channels defined in the form of |Φch(ρ)i or |Φn`i do not obey such condition. However, channels obtained as eigenvectors of the norm kernel (in the image space) do obey the orthogonality condition. Therefore, without even considering the Hamiltonian kernel, we obtain such channels by diagonalizing the Norm kernel. The new channels |G˜k`i are orthonormal, and therefore the binary cluster relative motion wave function and new SF are written as −1/2 ˜ 1 ˜ |ψk`i = Nk` |Gk`i = √ |Gk`i (2.45) Nk` 2 2 (new) X 2 X 1 ˜ X 1 X ˜ S` = |hΨP|ψk`i| = hGk`|ΨPi = hGk`|Φn`ihΦn`|ΨPi . (2.46) Nk` Nk` k k k n The index k is used to enumerate the different eigenvectors of N˜ and Nk` denotes the k-th eigenvalue of the Norm kernel. The new SF still has the same interpretation as a total sum over the probabilities of structurally separating the system into given clusters, but now with proper orthonormalization. This renormalized definition of the SF has a fixed sum rule for the total strength of a state [85]. If we now sum over all parent states with a given set of quantum numbers P , and use the fact that they form a complete set, we obtain the sum rule X (new) X X X X S` = hψk`|ΨPihΨP|ψk`i = hψk`|ψk`i = 1, (2.47) P P k k k which is the number of channels included in the calculation. 2.6.3 Dynamic Spectroscopic Factors In processes such as cluster decay, it is the dynamics, i.e. the Hamiltonian, that determine the weights of different cluster basis configurations and thus summing all probabilities can be 17 inappropriate. The dynamic SFs provide an appropriate definition. We define them via the overlap of the parent nucleus many-body state with a RGM generated channel function. The advantage of this approach is that it allows the dynamics of the Hamiltonian to define the relative motion between the clusters. The channels obtained as RGM solutions are linear combinations of the channels used in the previous section and thus have the same normalization properties. The new channels now have some hierarchy defined from the RGM energy eigenvalues. The dynamic SF is defined ”per channel”, as the decay proceeds through a single channel at a given energy 2 (dyn) (`) X 2 2 S` = hΨP|χ i = χn|Fn`| . (2.48) n The SFs can vary significantly when considering decays proceeding through different channels; when considering a sum over all open channels the sum rule given in eq. (2.47) is still observed. 2.7 Coupling to Continuum We now turn to the physics of the continuum and how it can be tied to the nuclear shell model. The discussion here follows closely the one found in [86]. The presence of open decay channels in a nucleus leads to a restructuring of the many-body states, separating them into two classes: broad short-lived states, known as superradiant, and long-lived, trapped states, that become nearly decoupled from the continuum. This phenomenon of superradiance has been discussed by many authors, whose efforts are summarized in the recent review [87]. Our interest lies in how the physics of clustering can affect, and in turn be affected, by the openness of the finite quantum system. While the phenomenon is well known, in the physics of nuclei it remains elusive; our inability to tune the continuum coupling, the short lifetimes of states of interest, the elaborate experimental tools required for this research, and the general complexity of the many-body physics make it dif- ficult to unambiguously pinpoint the superradiance. Nevertheless, experimental advances towards exotic nuclei far from the valley of stability motivate the quest for superradiance in nuclei while hints of superradiant states are seen in states of light nuclei, shown in Figure 2.1 . The Fesh- bach projection formalism [90, 91] appears to provide a straightforward theoretical strategy for exploration of this phenomenon [87]. Under this formalism a full Hilbert space is divided into two subspaces, an intrinsic or internal subspace Q that describes the isolated system, while the remain- ing continuum of external reaction states belongs to the second subspace P . The full problem can 18 + ���� � ���� �+ + � + �+ � + ���� � �+ + � �+ ��� � �(���) � ��+α ��+α ���+α �� ��� �+ �+α �+ �� + �+α � + ��� � �+ + + + + + + ��� � � � � � � ��� ���� ��� ��� ��� ���� Figure 2.1: Broad alpha clustering states in light nuclei. The data for 10Be are taken from [88] and for 18O from [89] be exactly projected, leading to the energy-dependent non-Hermitian effective Hamiltonian H(E), 1 H(E) = HQQ + HQP HPQ. (2.49) E − HPP Here, HQQ acts only in the intrinsic space Q, the operator HPP connects only states in the external P space, and HQP is the coupling between the intrinsic and external spaces. The effective Hamil- tonian operator (2.49) belongs to the intrinsic space and describes the evolution of states in the projected space. Therefore, the matrix elements of the transition matrix between reaction channels |Ac(E)i involved in the scattering process are c 1 c0 T 0 (E) = hA (E)| |A (E)i. (2.50) cc E − H(E) It is important that the same reaction channels define the effective Hamiltonian itself. The second term in (2.49) involves virtual processes going through the continuum channels; using the Cauchy 19 principal value (PV), we can separate real and imaginary parts 1 i HQP HPQ = ∆(E) − W (E) (2.51) E − HPP 2 given as X Z |Ac(E0)ihAc(E0)| X ∆(E) = PV dE0 and W (E) = 2π |Ac(E)ihAc(E)|. (2.52) E − E0 c c(open) The imaginary part here is of factorized form containing a summation over channels that are open at a given energy. This form assures unitarity of the scattering matrix [92, 93]. Indeed, taking the effective Hamiltonian in the form where the imaginary part is explicitly factorized H = H˜ − iπ AA† (2.53) allows us to write the S-matrix in the channel space as 1 1 − iπ K S = 1 − 2πT = 1 − 2πA† A = , (2.54) E − H 1 + iπ K which is explicitly unitary, being expressed using the Hermitian K-matrix that comes from the Hermitian part of the effective Hamiltonian 1 K = A† A. (2.55) E − H˜ In the above formalism bold symbols correspond to matrices in the channel space spanned by vectors |Ac(E)i for channels that are opened at a given energy. For most decays, the number of channels that are open is relatively small; cases with just one dominating decay channel are very common. Thus, mathematically, the factorized term iπ AA† is of low rank; if this term is dominating the effective Hamiltonian (2.53), then this would force segregation of eigenvectors into those orthogonal to channels |Ac(E)i and to eigenvectors that are nearly parallel to |Ac(E)i in the Hilbert space. The number of nearly parallel ones, those referred to as superradiant, equals to the number of open channels and these states accumulate almost all width (imaginary part) thus absorbing a large part of the decay strength [94, 95]. 2.8 Rotational Bands The existence of localized clusters in light nuclei results in rigid structures whose rotation gen- erates rotational bands observed in experiment. An outstanding example is 8Be, seen in Figure 2.1 20 with the 0+, 2+, 4+ states forming a rotational band. Rotational spectra are recognised as a part of nuclear structure, with clear experimental signatures, seen throughout the nuclear chart. In this section we briefly present some of the basic components in the physics of rotating many-body systems neccessary for the description of cluster rotations [77, 96]. For rotations to be possible, there must be some relatively rigid deformed system. The rigid nature of some clustered systems allows for the rotational motion to be nearly decoupled from other types of excitations [97]. Therefore, the low-lying spectra can be described in terms of a single rotating intrinsic shape. This leads to the excitation energies of nuclear states described by the rotation of a rigid (non-vibrating) body to be given by 2 2 ¯h HR = AJ = J(J + 1). (2.56) 2I While higher orders of J2 are possible, this is sufficient [77]. Here, I is the only parameter and is commonly known as the moment of inertia of the band in question in a semiclassical analogy. The other constant of motion is the projection K of the angular momentum on the deformation axis, which is used to label families of states belonging to the same rotational band. The case of K = 1/2 bands is special as it sometimes shows effects of Coriolis decoupling. This can be understood in a particle-plus-rotor model where the odd particle system is considered as a single particle on top of a rigid rotor [96, 98]. The excitation energies for the band now further depend on the so-called Coriolis decoupling parameter a 2 ¯h h J+1/2 i Ex = J(J + 1) + a(−1) (J + 1/2) . (2.57) 2I In cases where the Coriolis decoupling parameter is large, this results in a zig-zag shape for the spectra of the states as one goes up in J. Let us make a note on how rotational bands are identified, apart from the spacings of the states. Deformations are probed experimentally by measuring reduced electric quadrupole transition rates B(E2), as it is directly related to the quadrupole moment of the nucleus [77] 2 5 JK 2 B(E2,J + 2 → J) = Q C (2.58) 0 16π J+2K2K Here, Q0 denotes the intrinsic quadrupole moment of the rotational band, labeled by K, considered in this case. Within a rotational band these transition rates are tens or even hundreds of times larger 21 than what one might expect from a single particle picture [77]. From an independent particle model prespective, the phenomenon of rotation and transitions within a rotational band are collective in nature where nucleons transition coherently from one state to the other. The large rates coupled with the spacing of different J values given by eqs. (2.56,2.57) serve as signatures of a deformed rotating system. 2.8.1 Elliott’s SU(3) Model for Rotations In Elliott’s seminal work [99, 100], dating back to 1958, the foundation was laid out on how rotational bands can be described within the nuclear shell model with a HO single particle basis. Traditionally there had been a separation between microscopic (shell model) and macroscopic (liquid drop) descriptions of nuclei. Elliott bridged the two concepts by presenting a microscopic formulation that displayed rotational band characteristics. By exploiting the U(3) symmetry of the harmonic oscillator potential to label the states according to their irreducible representation, the representation of collective rotational states within a microscopic model was uncovered. While Elliott worked within a single major HO shell, the results hold for spaces spanning multi- ple shells. The U(3) irreducible representations are labeled by three integer numbers, traditionally denoted by f1, f2, f3, each of them being the number of boxes in each of the rows of the Young pattern associated with the irreducible representation [101]. Since the U(3) representation of the HO states concerns only the spatial components, it follows that any symmetry specified here refers to the spatial permutational symmetry; the spin-isospin part is given by the conjugate Young pat- tern, so as to maintain total antisymmetry of the wave function [102]. The 9 generators are taken to be the HO single particle Hamiltonian, the five (algebraic) Quadrupole operators and the three angular momentum operators. The difference between the algebraic Quadrupole operators and the usual mass/electric quadrupole operators is the absence of matrix elements between shells that differ by two quanta. By further removing transformations that only yield an overall sign change, we obtain the SU(3) group for the description of nuclear states. The two integers labeling the SU(3) irreducible representation (λ, µ) are given with respect to their U(3) counterparts by λ = f1−f2 and µ = f2−f3. The stretched SU(3) irreducible representations (λ, 0) imply full permutational symmetry in the spatial part of the wave function and will be used later on for the construction of α particle wave functions. 22 CHAPTER 3 CONSTRUCTING CLUSTER CHANNELS In this chapter, we present the main contribution of this work. We construct many-particle fully antisymmetric channel states for a system consisting of two fragments in relative motion described by a HO wave function. By construction, the channels have the same oscillator parameter ω, carry no HO quanta in their CM coordinate and respect rotational symmetry. 3.1 Antisymmetry The Pauli exclusion principle is an important consideration when constructing many-body states. Let’s consider here, for example, the 8Be system clustering into two α fragments. Due to the compact nature of 4He, the ground state of the α particle can be approximated by a s4 configuration. Antisymmetrizing the total wave function for a channel with both α particles in the ground state configuration of the same HO well would result in a trivial 0 as the state is Pauli blocked. The lowest, in terms of HO quanta, allowed configuration for the 2α system would 4 4 then be the s p (Ntot = 4). Such issues were recognized in the past, requiring the relative wave function between the cluster and the nucleus to have a fixed number of quanta, so as to not be Pauli blocked [103, 104]. The four quanta in the above example are in the relative motion between the clusters, leaving their respective internal s4 structures unchanged. When utilizing CI methods with a HO single particle basis we assume non-spurious configurations with no quanta in the CM motion. We briefly comment on the state-operator polymorphism through which antisymmetrization is enforced in every step of the cluster channel construction. For more in depth discussion on the topic, Appendix A describes second quantized states and operators, and two examples are found in Appendix D. Since configuration interaction approaches utilize a second quantization occupation basis to expand the many-body states, the states we start with are by construction antisymmetrized. In order to enforce complete antisymmetry we write the many-body wave functions as second quantization many-body creation operators acting on the vaccum, with each operator weighed by 23 an appropriate coefficient. For example, the state describing a trivial s4 α particle is a single Slater Determinant which can be written as 4 creation operators acting on the vaccum † † † † † |Ψα(s4)i = aν+aν−aπ+aπ−|0i = Ψα|0i. (3.1) Here, the subscript π, ν denotes protons and neutrons respectively, while the ± denotes the m = ±1/2 angular momentum projection of the total angular momentum. The rest of the quantum numbers are those of the s shell {n`j} = {0 0 1/2}. The (s4) notation is omitted in the rest of the text. We are working here in the isospin formalism so the proton and neutron labels refer to the two projections of isospin. Similarly, for the daughter state there exists an AD-body creation † operator ΨD that creates the state when acting on the vaccum. The channel state can then be constructed as a sequential action of many-body creation operators † † |Ψi = ΨDΨα|0i (3.2) hr|Ψi = A [Ψα(rα)ΨD(rD)] = Ψ(r), (3.3) where bold symbols denote all single particle coordinates for each wave function. We are not dealing here with Jacobi coordinates, so the wave function in eq. (3.1) carries a CM contribution. † † The comment we started from implies that, with the definition (3.1), ΨαΨα|0i = 0 because of Pauli blocking. Thus to obtain relative channels, we have to manipulate the CM part of the wave function. 3.2 The Gloeckner-Lawson Procedure Before constructing the cluster channels we mention the method used to construct so-called non-spurious states, meaning here states that have zero HO quanta in the CM motion, i.e. the many-body wave function can be written as 0 hr|Ψi = ψ000(R)Ψ (ξ), (3.4) with ξ the internal nucleus coordinates and R the CM coordinate. We mentioned this subject for two particles in Section 2.1.1, but the Moshinsky bracket recoupling scheme is not efficient for many particles. 24 The technique presented here is applicable in CI approaches where the HO is used as the single particle basis. Alternative approaches either eliminate the CM contaminants from the states after calculation [23], or construct the many-body states excluding the CM excitations [105]. A computationally efficient procedure was proposed by Palumbo and Prosperi and further developed by Gloeckner and Lawson [106, 107]; we discuss it next. The method augments the many-body Hamiltonian in eq. (2.26) by explicitly adding a HO Hamiltonian for the CM coordinate, multiplied by an tunable parameter β 1 2 2 3 Hˆ = H + βHCM = H + β(TCM + AMω R − ¯hω), (3.5) 2 CM 2 and then proceed to solve the new Hamiltonian Hˆ . The existence of the explicit CM HO contri- bution to the Hamiltonian implies that now any CM excited states will have an additional energy related to their CM HO quanta ECM = β¯hωNCM. The parameter β is tuned so the CM-excited states are shifted higher in the spectrum and are absent in the low-lying spectrum of interest (a typical number is β = 10). While in NCSM calculations with the Nmax truncation prescription the separation is exact, this does not hold for an arbitrary space truncation, including some traditional shell model valence spaces. Remarkably, in the non-exact case the separation of the CM motion of the system is good, with expectation values of the CM hamiltonian hβHCMi in the resulting states being in the order of ∼ 10 keV [85]. 3.3 SU(3) Method Earlier attempts dealing with α clustering in nuclei [22] utilized the SU(3) symmetry properties of the HO and dealt exclusively with cases where the valence-particle nuclear wave function lied completely in a single major oscillator shell. Furthermore they required the cluster structure to be described by a configuration having no intrinsic HO quanta, so as the total ground state wave function of the cluster would be a (0,0) SU(3) irreducible representation. This approach is especially well suited to the α cluster, where one does not expect significant additional contributions to the α ground state. The s4 structure for the α wave function in (3.1) has a full permutational symmetry on the spatial part of the wave function and all angular momenta and spin quantum numbers (L, S, J, T ) are coupled to 0. The spatial part is a product of gaussians so any exchange of particles 25 will leave the total wave function unchanged. This structure has no quanta, including the CM degree of freedom. Let us now consider an α particle moving with 4 quanta of motion relative to some p shell nucleus. We need to project the motion of the α particle with 4 quanta in the relative motion onto the valence space of interest. The answer to this problem is found to be the stretched SU(3) irreducible representation. As discussed in Section 2.8.1, stretched representations are of the form (N, 0) where N is the total number of HO quanta available in the relative cluster motion. In this case we need to find the (4,0) irreducible representation in the p4 configuration subspace. The reason behind the need for a stretched representation becomes apparent if one considers the Young diagram of the representation; a single line for all boxes, indicating full permutational symmetry in the spatial part. The conjugate pattern for the spin-isospin part will be fully antisymmetric [102] restoring the proper symmetry properties of the full wave function. Exactly how these stretched irreducible representations are constructed depends on the many- body scheme one chooses to represent the basis. A basis that uses SU(3) symmetry might seem like an ideal choice, however this would result in complications when trying to deal with interactions not written in the SU(3) scheme and require heavy use of SU(3) Clebsch-Gordan coefficients [108, 109]. We take a different approach and work in an M-scheme [41, 97] basis where we truncate the space of many-body states to our chosen configuration (particle distribution in the shells) and diagonalize an operator whose eigenstates are the configurations of interest and which lifts all possible degeneracies within the configuration [85, 110, 111]. The correct irreducible representation can then be chosen using the eigenvalues of some operator. Such an operator can be built as a sum of the following operators 2 2 2 2 Oi = {C2(SU3),J ,T ,L ,S ,MJ }, (3.6) 1 3 where C2(SU3) = 4 Q · Q + 4 L · L denotes the second order casimir operator of SU(3) and MJ is the Majorana exchange operator. The operator Q is the so-called algebraic quadrupole operator, with no matrix elements between major oscillator shells differing by 2 quanta. Each of the Oi operators is multiplied by an irrational number to avoid any accidental degeneracies in the eigenvalues. It should be noted that C3(SU3) is not necessary as it only helps to tell the difference between the (N, 0) and (0,N) irreducible representations, though the latter do not possess the permutational 26 symmetry we seek. At the end of this process we have a four-body state for the configuration labeled by η, written as an operator in the second quantization approach, that has the following structure SU(3) |Ψαη:L i = |[4](N, 0) : L = J, S = T = 0i , (3.7) and as such it represents a component of an α-like state in the desired configuration. While this method works flawlessly, it relies on diagonalization of matrices that quickly expand in dimension. Furthermore, it is only applicable to clusters that can be modelled in a Ntot = 0 space (all particles in the s shell). The method was heavily used in [85]. These limits drive us to develop new methods for the construction of cluster channels that both scale better computationally and are applicable to heavier and more complex clusters. 3.4 Center of Mass Boosting Method In the next two sections we present our newly developed method that can be used to construct cluster channels with the fragments in some definite relative HO motion. In Section 2.1.1 we saw how one can transform from single particle coordinates to CM-relative coordinates in the HO basis using a linear transformation. We will use such a transformation in Section 3.5, for the CM coordinates of the two fragments, transforming to the system of total CM and relative cluster coordinates. To proceed with the transformation, we devise a method that can construct the fragment CM excited (boosted) states to be recoupled. We do this via HO quantum raising operators acting directly on the CM coordinate of each cluster. The CM creation operators (and annihilation correspondingly) are defined in the usual way as † 1 B = √ (AMωRm − iPm ) (3.8) m 2AMω¯h where m denotes a specific magnetic projection of vectors . The nucleon mass is denoted by M to avoid confusion with the projection index. The operator is constructed using an isoscalar mass- density dipole E1 operator r r 4π ¯h D = (B† + B ), (3.9) m 3 2AMω m m and by taking the part that increases the number of quanta. It should be stressed that these operators are one-body operators that act on the many-body wave functions in second quantization of Section 3.1, see also Appendix A. 27 To raise the number of nodes in the CM coordinate, we need to construct a rotational scalar that contains two raising operators since each node contributes two quanta of energy. This can be achieved by taking a dot product of two creation operators † † † † † † † † B · B ≡ B+1B−1 + B−1B+1 − B0B0 . (3.10) The application of this operator on a harmonic oscillator wave function gives, 1p B† · B†ψ (R) = (2n + 2)(2n + 2` + 3) ψ (R). (3.11) n`m 4 n+1`m The normalization of the operator is derived in Appendix B. In a similar fashion, to control the † angular momentum part, we act with the B+1 operator to force the wave function up along the chain of m = ` states. s (` + 1)(2n + 2` + 3) B† ψ (R) = ψ (R). (3.12) +1 n`` 4(2` + 3) n`+1`+1 Finaly the m projection can be brought to the required value by repeated action of an angular momentum lowering operator. The angular momentum operator needs to be an axial vector and leave the total number of oscillator quanta unchanged. Thus the operator B† ×B is the only choice. The action of this operator is defined as √ † † p L+ψn`m (R) ≡ 4 2 B0B+1 − B+1B0 ψn`m (R) = (l − m)(l + m + 1)ψn`m+1(R) (3.13) √ † † p L−ψn`m (R) ≡ 4 2 B−1B0 − B0B−1 ψn`m (R) = (l + m)(l − m + 1)ψn`m−1(R). (3.14) An alternative way of dealing with the angular momentum part was outlined in [23, 44]. The result of a sequential application of node and angular momentum operators is a many-body wave function with a definite structure in the CM coordinate. The relation between the SU(3) irreducible representations and the CM boosted states can now be elucidated. The CM boosted state extends over all configurations and, as the original state, has a norm of 1 by construction. On the other hand, the SU(3) irreducible representations also have a norm of 1 but there is one for every configuration that exists. The boosted wave function of an s4 α particle is a sum of all possible (N, 0) SU(3) irreducible representations going over all allowed distributions of particles with coefficients known analytically s 1 N! 4! Xη = . (3.15) N N Q αi Q 4 i(ni!) i αi! 28 They are commonly known as Cluster Coefficients [85, 23, 112], depending only on the distribution of the particles (αi) in the HO shells (ni). The label η is used as in the previous section to denote the configuration and N is the total number of quanta. A resummation of the wave functions of all irreducible representations weighted by their respective Cluster Coefficients would result in the CM boosted wave function. Table 3.1 lists the cluster coefficients for different configurations in CM boosted by 8 HO quanta α wave functions. The right-most column refers to an α particle wave function that is not purely s4 but contains contributions from excited configurations. Table 3.1: Select configuration content of NCSM wave functions for 4He withhω ¯ = 20 MeV boosted to a CM boosted state with n = 4 and ` = 0. Configuration Nmax = 0 Nmax = 4 (sd)4 0.038 0.035 (p)(sd)2(pf) 0.308 0.282 (p)2(pf)2 0.103 0.094 (p)2(sd)(sdg) 0.154 0.141 (p)(sd)(sdg)(pfh) 0.000 0.005 (p)(sd)(pf)(sdg) 0.000 0.009 This new method of controlling the number of CM quanta directly bypasses all problems dis- cussed, and can raise any many-body wave function to higher quanta content. The boosted states with HO quantum numbers for the CM coordinate NLM are written in second quantization with their coordinate representation † 0 X η SU(3) hr|ΨNLM i = hr|ΨNLM |0i = ψNLM (R)Ψ = XN Ψαη:L , (3.16) η completely separating the intrinsic and CM parts of motion. It can therefore act on heavier clusters that extend beyond the s shell and deal with realistic cluster wave functions that might extend into multiple shells. 3.5 Constructing the Channels These CM boosted states for each cluster now need to be recoupled into a complete channel with some non-spurious wave function with definite relative motion. In Figure 3.1 we illustrate the construction of a cluster basis state which has the two clusters in relative motion described by the 29 D0 D0 nl(⇢)