Continuum Level Density in Nuclear Systems

• Continuum Level Density is defined as the interaction-induced change in state density beyond the free-particle contribution, derived from phase-shift derivatives.
• It plays a crucial role in nuclear pairing by unifying bound-state and scattering contributions in pairing Hamiltonians.
• The concept extends to complex and quasi-continuum regimes, impacting statistical spectroscopy and time delay interpretations in open quantum systems.

Searching arXiv for recent and foundational papers on continuum level density and related nuclear applications. Continuum level density denotes the density of states associated with spectra that are not purely discrete. In nuclear structure, scattering theory, and continuum approximations of lattice models, it is not simply the raw counting density of positive-energy states in an infinite volume. Rather, the physically relevant quantity is often the interaction-induced modification of that density relative to an appropriate free reference. In the nuclear literature this appears most explicitly as the continuum single-particle level density (CSPLD), obtained from scattering phase shifts and used to place bound and positive-energy contributions on a common footing in pairing problems (Betan et al., 2017). In quasi-continuum phenomenology, the same term refers to the dense excitation-energy region where individual levels are unresolved and statistical level-density models become operative (Firestone, 2021). In open quantum systems it can be extended to a complex quantity whose real and imaginary parts encode complementary aspects of transmission through a potential (Stránský et al., 2021).

1. Foundational definitions and phase-shift representation

For finite nuclei described by a mean field, the single-particle spectrum separates into bound states at negative energy and continuum scattering states at positive energy. A direct sum over positive-energy states is ill-defined in infinite volume, so a standard construction is to place the system in a large spherical box, discretize the continuum, and then take the limit $R\to\infty$. In this limit, the divergent free-particle contribution must be removed. The resulting interaction-induced density is the continuum single-particle level density, defined channel by channel through the derivative of the scattering phase shift (Betan, 2017).

For each partial wave $(l,j)$,

$$g_{lj}(\varepsilon) = g^{(\mathrm{total})}_{lj}(\varepsilon)-g^{(\mathrm{free})}_{lj}(\varepsilon) = \frac{1}{\pi}\frac{d\delta_{lj}}{d\varepsilon}.$$

Summing over partial waves with degeneracy gives

$$g(\varepsilon)=\sum_{lj}(2j+1)\,g_{lj}(\varepsilon).$$

This subtraction embodies the Beth–Uhlenbeck viewpoint: the relevant density of states is the change induced by the interaction, not the absolute free-gas density. A recurrent misconception is that the raw box density is itself physical. The phase-shift construction shows otherwise: the divergent free continuum dominates the total density in infinite volume, whereas localized nuclear correlations depend on where the mean field enhances or suppresses the continuum through resonances, antibound states, and non-resonant background (Betan, 2011).

Bound and continuum parts may be combined into a single mixed representation,

$$\tilde g(\varepsilon) = \sum_{n_b}(2j_{n_b}+1)\,\delta(\varepsilon-\varepsilon_{n_b})\,\theta(-\varepsilon) + g(\varepsilon)\,\theta(\varepsilon),$$

so that discrete sums become a bound-state sum plus a continuum integral. This mixed representation is the formal basis of most continuum pairing calculations (Betan et al., 2017).

2. Continuum level density in pairing Hamiltonians

In constant-pairing Hamiltonians, the decisive technical step is the replacement

$$\sum_n f_n \;\rightarrow\; \sum_{n_b}(2j_{n_b}+1)f_{n_b} + \int_0^\infty d\varepsilon\, g(\varepsilon)\,f(\varepsilon),$$

which promotes the usual discrete-shell pairing problem to a bound-plus-continuum one (Betan, 2011).

In BCS and Lipkin–Nogami formulations, the formal structure of the equations is unchanged; only the summation measure changes. The gap and number equations become

$$\frac{4}{G} = \sum_{n_b}(2j_{n_b}+1)\frac{1}{E_{n_b}} + \int_0^\infty d\varepsilon\, g(\varepsilon)\,\frac{1}{E(\varepsilon)},$$

$$N = \sum_{n_b}(2j_{n_b}+1)\,v_{n_b}^2 + \int_0^\infty d\varepsilon\, g(\varepsilon)\,v^2(\varepsilon).$$

The same substitution applies in the Lipkin–Nogami correction term $\lambda_2$, so the continuum enters self-consistently into the gap, Fermi level, and particle-number balance (Betan et al., 2017).

The exact Richardson solution admits an analogous generalization. For pair energies $E_{p_i}$, the continuum Richardson equations become

$(l,j)$0

so the continuum is included on the same footing as bound orbits through the CSPLD (Betan, 2012).

These formulations establish two points that recur across the literature. First, the continuum can make an important contribution to the pairing parameter even when the continuum is weakly populated. Second, the approximate BCS solution depends on the model space, whereas the exact Richardson solution does not after the pairing strength is refitted to the same observable (Betan, 2011). This separates genuine continuum effects from artifacts of approximate particle-number nonconservation.

3. Borromean nuclei and drip-line systems

The role of continuum level density is especially transparent in Borromean nuclei. In $(l,j)$1He and $(l,j)$2Li, the core–neutron subsystems are unbound, so all neutron single-particle states in the core field are in the continuum. Pairing must therefore be formulated directly in continuum space rather than through bound orbitals or box-discretized pseudostates (Betan, 2017).

In this setting, the two-neutron dispersion relation is weighted by the CSPLD: $(l,j)$3 and the partial-wave occupation probabilities are

$(l,j)$4

Because $(l,j)$5 reflects resonances and near-threshold antibound structure, the nominally constant pairing interaction becomes effectively channel dependent (Betan, 2017).

For $(l,j)$6He, a narrow $(l,j)$7 resonance at $(l,j)$8 MeV generates a pronounced bump in $(l,j)$9, while the very broad $$g_{lj}(\varepsilon) = g^{(\mathrm{total})}_{lj}(\varepsilon)-g^{(\mathrm{free})}_{lj}(\varepsilon) = \frac{1}{\pi}\frac{d\delta_{lj}}{d\varepsilon}.$$0 resonance produces almost no visible structure. For $$g_{lj}(\varepsilon) = g^{(\mathrm{total})}_{lj}(\varepsilon)-g^{(\mathrm{free})}_{lj}(\varepsilon) = \frac{1}{\pi}\frac{d\delta_{lj}}{d\varepsilon}.$$1Li, an antibound $$g_{lj}(\varepsilon) = g^{(\mathrm{total})}_{lj}(\varepsilon)-g^{(\mathrm{free})}_{lj}(\varepsilon) = \frac{1}{\pi}\frac{d\delta_{lj}}{d\varepsilon}.$$2 state very close to threshold, with $$g_{lj}(\varepsilon) = g^{(\mathrm{total})}_{lj}(\varepsilon)-g^{(\mathrm{free})}_{lj}(\varepsilon) = \frac{1}{\pi}\frac{d\delta_{lj}}{d\varepsilon}.$$3 keV, produces a very sharp peak near zero energy, and a $$g_{lj}(\varepsilon) = g^{(\mathrm{total})}_{lj}(\varepsilon)-g^{(\mathrm{free})}_{lj}(\varepsilon) = \frac{1}{\pi}\frac{d\delta_{lj}}{d\varepsilon}.$$4 resonance around $$g_{lj}(\varepsilon) = g^{(\mathrm{total})}_{lj}(\varepsilon)-g^{(\mathrm{free})}_{lj}(\varepsilon) = \frac{1}{\pi}\frac{d\delta_{lj}}{d\varepsilon}.$$5 MeV produces another peak. These structures govern the balance of $$g_{lj}(\varepsilon) = g^{(\mathrm{total})}_{lj}(\varepsilon)-g^{(\mathrm{free})}_{lj}(\varepsilon) = \frac{1}{\pi}\frac{d\delta_{lj}}{d\varepsilon}.$$6, $$g_{lj}(\varepsilon) = g^{(\mathrm{total})}_{lj}(\varepsilon)-g^{(\mathrm{free})}_{lj}(\varepsilon) = \frac{1}{\pi}\frac{d\delta_{lj}}{d\varepsilon}.$$7, and $$g_{lj}(\varepsilon) = g^{(\mathrm{total})}_{lj}(\varepsilon)-g^{(\mathrm{free})}_{lj}(\varepsilon) = \frac{1}{\pi}\frac{d\delta_{lj}}{d\varepsilon}.$$8 components in the Borromean wave functions (Betan, 2017).

A broader drip-line application is the Sn isotopic chain treated in BCS and Lipkin–Nogami approximations. There the CSPLD is constructed from Woods–Saxon plus spin–orbit phase shifts and used up to $$g_{lj}(\varepsilon) = g^{(\mathrm{total})}_{lj}(\varepsilon)-g^{(\mathrm{free})}_{lj}(\varepsilon) = \frac{1}{\pi}\frac{d\delta_{lj}}{d\varepsilon}.$$9 and $$g(\varepsilon)=\sum_{lj}(2j+1)\,g_{lj}(\varepsilon).$$0 MeV. The Fermi level crosses the continuum threshold at $$g(\varepsilon)=\sum_{lj}(2j+1)\,g_{lj}(\varepsilon).$$1, the continuum contribution to the particle number remains small up to $$g(\varepsilon)=\sum_{lj}(2j+1)\,g_{lj}(\varepsilon).$$2Sn, and the two-neutron drip line is predicted around $$g(\varepsilon)=\sum_{lj}(2j+1)\,g_{lj}(\varepsilon).$$3Sn (Betan et al., 2017). The same work stresses that the CSPLD is an economical way to include explicitly the correlations with the continuum spectrum of energy in large scale mass calculation (Betan et al., 2017).

4. Quasi-continuum level density and statistical nuclear spectroscopy

A distinct but related usage of continuum level density appears in reaction spectroscopy below particle thresholds. Here the relevant regime is the quasi-continuum: excitation energies where individual levels are too dense to be resolved, but the nucleus remains bound. In Oslo-method analyses this region is accessed through the factorization

$$g(\varepsilon)=\sum_{lj}(2j+1)\,g_{lj}(\varepsilon).$$4

which separates the first-generation $$g(\varepsilon)=\sum_{lj}(2j+1)\,g_{lj}(\varepsilon).$$5-ray matrix into a $$g(\varepsilon)=\sum_{lj}(2j+1)\,g_{lj}(\varepsilon).$$6-transmission coefficient and a level density (Tornyi et al., 2014).

For heavy actinides the extracted level density follows a constant-temperature law,

$$g(\varepsilon)=\sum_{lj}(2j+1)\,g_{lj}(\varepsilon).$$7

In $$g(\varepsilon)=\sum_{lj}(2j+1)\,g_{lj}(\varepsilon).$$8Th and $$g(\varepsilon)=\sum_{lj}(2j+1)\,g_{lj}(\varepsilon).$$9U, the quasi-continuum level densities display constant temperature behavior with $$\tilde g(\varepsilon) = \sum_{n_b}(2j_{n_b}+1)\,\delta(\varepsilon-\varepsilon_{n_b})\,\theta(-\varepsilon) + g(\varepsilon)\,\theta(\varepsilon),$$0 MeV, except $$\tilde g(\varepsilon) = \sum_{n_b}(2j_{n_b}+1)\,\delta(\varepsilon-\varepsilon_{n_b})\,\theta(-\varepsilon) + g(\varepsilon)\,\theta(\varepsilon),$$1U where $$\tilde g(\varepsilon) = \sum_{n_b}(2j_{n_b}+1)\,\delta(\varepsilon-\varepsilon_{n_b})\,\theta(-\varepsilon) + g(\varepsilon)\,\theta(\varepsilon),$$2 MeV (Guttormsen et al., 2013). In odd-odd $$\tilde g(\varepsilon) = \sum_{n_b}(2j_{n_b}+1)\,\delta(\varepsilon-\varepsilon_{n_b})\,\theta(-\varepsilon) + g(\varepsilon)\,\theta(\varepsilon),$$3Np, the level density follows closely the constant-temperature level density formula with $$\tilde g(\varepsilon) = \sum_{n_b}(2j_{n_b}+1)\,\delta(\varepsilon-\varepsilon_{n_b})\,\theta(-\varepsilon) + g(\varepsilon)\,\theta(\varepsilon),$$4 MeV and reaches $$\tilde g(\varepsilon) = \sum_{n_b}(2j_{n_b}+1)\,\delta(\varepsilon-\varepsilon_{n_b})\,\theta(-\varepsilon) + g(\varepsilon)\,\theta(\varepsilon),$$5 at $$\tilde g(\varepsilon) = \sum_{n_b}(2j_{n_b}+1)\,\delta(\varepsilon-\varepsilon_{n_b})\,\theta(-\varepsilon) + g(\varepsilon)\,\theta(\varepsilon),$$6 MeV (Tornyi et al., 2014).

This quasi-continuum usage should not be conflated with the open scattering continuum. In the actinide papers, “continuum” means the dense unresolved region up to the neutron separation energy, where statistical observables such as $$\tilde g(\varepsilon) = \sum_{n_b}(2j_{n_b}+1)\,\delta(\varepsilon-\varepsilon_{n_b})\,\theta(-\varepsilon) + g(\varepsilon)\,\theta(\varepsilon),$$7, $$\tilde g(\varepsilon) = \sum_{n_b}(2j_{n_b}+1)\,\delta(\varepsilon-\varepsilon_{n_b})\,\theta(-\varepsilon) + g(\varepsilon)\,\theta(\varepsilon),$$8, entropy, microcanonical temperature, and heat capacity become meaningful (Guttormsen et al., 2013).

A major criticism of the standard constant-temperature model is developed in the spin/parity-dependent CT-JPI framework. The paper argues that the standard CT model describing the total level density is fatally flawed due to discontinuities at the Yrast energies, the onset of new $$\tilde g(\varepsilon) = \sum_{n_b}(2j_{n_b}+1)\,\delta(\varepsilon-\varepsilon_{n_b})\,\theta(-\varepsilon) + g(\varepsilon)\,\theta(\varepsilon),$$9 sequences, that disrupt the exponential formula and cause the back shift parameter to become nonphysically negative. The proposed alternative assigns a common $$\sum_n f_n \;\rightarrow\; \sum_{n_b}(2j_{n_b}+1)f_{n_b} + \int_0^\infty d\varepsilon\, g(\varepsilon)\,f(\varepsilon),$$0 and separate back shifts $$\sum_n f_n \;\rightarrow\; \sum_{n_b}(2j_{n_b}+1)f_{n_b} + \int_0^\infty d\varepsilon\, g(\varepsilon)\,f(\varepsilon),$$1 to each spin-parity sequence, with

$$\sum_n f_n \;\rightarrow\; \sum_{n_b}(2j_{n_b}+1)f_{n_b} + \int_0^\infty d\varepsilon\, g(\varepsilon)\,f(\varepsilon),$$2

and constrains the spin distribution at $$\sum_n f_n \;\rightarrow\; \sum_{n_b}(2j_{n_b}+1)f_{n_b} + \int_0^\infty d\varepsilon\, g(\varepsilon)\,f(\varepsilon),$$3 through Ericson’s formula (Firestone, 2021). Applied to 46 nuclei with $$\sum_n f_n \;\rightarrow\; \sum_{n_b}(2j_{n_b}+1)f_{n_b} + \int_0^\infty d\varepsilon\, g(\varepsilon)\,f(\varepsilon),$$4–92, this analysis finds that the fitted spin cutoff parameters show no mass dependence and instead substantial variation at all mass regions (Firestone, 2021).

5. Complex continuum level density in open quantum systems

For scattering systems, the conventional continuum level density is defined by the Green-function difference

$$\sum_n f_n \;\rightarrow\; \sum_{n_b}(2j_{n_b}+1)f_{n_b} + \int_0^\infty d\varepsilon\, g(\varepsilon)\,f(\varepsilon),$$5

In one-channel problems this reduces to the phase-shift derivative, and in one-dimensional tunneling it can be expressed through the transmission phase (2002.03874).

A complex extension replaces the real-energy limit by a genuinely complex energy $$\sum_n f_n \;\rightarrow\; \sum_{n_b}(2j_{n_b}+1)f_{n_b} + \int_0^\infty d\varepsilon\, g(\varepsilon)\,f(\varepsilon),$$6 and defines

$$\sum_n f_n \;\rightarrow\; \sum_{n_b}(2j_{n_b}+1)f_{n_b} + \int_0^\infty d\varepsilon\, g(\varepsilon)\,f(\varepsilon),$$7

On the real axis this yields

$$\sum_n f_n \;\rightarrow\; \sum_{n_b}(2j_{n_b}+1)f_{n_b} + \int_0^\infty d\varepsilon\, g(\varepsilon)\,f(\varepsilon),$$8

so that

$$\sum_n f_n \;\rightarrow\; \sum_{n_b}(2j_{n_b}+1)f_{n_b} + \int_0^\infty d\varepsilon\, g(\varepsilon)\,f(\varepsilon),$$9

The real part is therefore the usual continuum level density, while the imaginary part tracks the energy dependence of the transmission probability (Stránský et al., 2021).

The same object is proportional to a complex time shift,

$$\frac{4}{G} = \sum_{n_b}(2j_{n_b}+1)\frac{1}{E_{n_b}} + \int_0^\infty d\varepsilon\, g(\varepsilon)\,\frac{1}{E(\varepsilon)},$$0

Its real part is the Eisenbud–Wigner time shift, and its imaginary part is interpreted semiclassically as an instanton-like tunneling time accumulated in classically forbidden regions (Stránský et al., 2021). Complex scaling provides the numerical framework: the continuum is rotated into a discrete set of complex eigenvalues, and the resulting $$\frac{4}{G} = \sum_{n_b}(2j_{n_b}+1)\frac{1}{E_{n_b}} + \int_0^\infty d\varepsilon\, g(\varepsilon)\,\frac{1}{E(\varepsilon)},$$1 reproduces both the real “time shift” and the imaginary time associated with tunneling (2002.03874).

Stationary points of the potential generate singularities in both components of the complex density. Quadratic maxima lead to logarithmic divergences, quadratic minima to step-like singularities, and higher-order stationary points to power-law behavior. These singularities are presented as close analogues, and in the later formulation a dual extension, of excited-state quantum phase transitions from bound to continuum systems (Stránský et al., 2021).

6. Implicit continuum densities and discrete-to-continuum limits

Not all continuum formulations introduce a phase-shift density explicitly. In density-dependent deformed relativistic Hartree–Bogoliubov theory in continuum, the notion of continuum level density is built into the quasiparticle spectrum obtained in a large Woods–Saxon basis in a box. The continuum level density is then the distribution of quasiparticle eigenvalues above the Fermi surface, including resonant and non-resonant states, that contribute to densities and pairing (Chen et al., 2012). The continuum is represented as a quasi-discrete spectrum with realistic asymptotics, and deformation reshapes its effective level density by splitting spherical levels into $$\frac{4}{G} = \sum_{n_b}(2j_{n_b}+1)\frac{1}{E_{n_b}} + \int_0^\infty d\varepsilon\, g(\varepsilon)\,\frac{1}{E(\varepsilon)},$$2 components and modifying resonant structure (Chen et al., 2012).

A mathematically distinct but conceptually related development appears in the analysis of a tight-binding model with a slowly-varying periodic potential. There the density of states of the discrete model, after rescaling, converges to the density of states of a continuum Mathieu-type operator,

$$\frac{4}{G} = \sum_{n_b}(2j_{n_b}+1)\frac{1}{E_{n_b}} + \int_0^\infty d\varepsilon\, g(\varepsilon)\,\frac{1}{E(\varepsilon)},$$3

and the convergence is proved pointwise in the Lorentzian-smoothed density of states with rate $$\frac{4}{G} = \sum_{n_b}(2j_{n_b}+1)\frac{1}{E_{n_b}} + \int_0^\infty d\varepsilon\, g(\varepsilon)\,\frac{1}{E(\varepsilon)},$$4 (Hofhansel et al., 12 Dec 2025). In that setting, continuum level density is the density of states of a continuum differential operator obtained as the scaling limit of a discrete spectrum.

This suggests a unifying theme: continuum level density is not a single formula but a family of constructions adapted to the spectral problem at hand. In nuclear pairing it is the interaction-induced change in single-particle state density relative to free motion; in quasi-continuum spectroscopy it is the statistical density of unresolved many-body levels below separation energy; in open scattering systems it becomes a complex spectral quantity tied to phase, attenuation, and complex time; and in discrete-to-continuum limits it is the spectral density of the effective continuum operator (Betan et al., 2017).