Discrete spectrum of curved quantum layers: the Lu–Rowlett open problems on uniform thin-layer existence, the positive-total-Gauss-curvature domination inequality, and the thick-layer dichotomy

A quantum layer is the region trapped between two parallel copies, a fixed distance $2a$ apart, of a smooth complete non-compact surface $\Sigma\subset\mathbb{R}^3$, equipped with the Dirichlet Laplacian as Hamiltonian; bound states are discrete eigenvalues below the essential-spectrum threshold $\pi^2/(4a^2)$. Since Duclos, Exner and Krejčiřík (2001) showed that bending alone can create such bound states, the guiding question has been how generally curvature guarantees a non-empty discrete spectrum. Lu and Rowlett (2012) settled the non-uniform version — every asymptotically flat surface with integrable Gauss curvature, and indeed every parabolic surface, binds for all sufficiently thin layers — and posed the questions that remain: Conjecture 2, whether the thin-layer width threshold can be chosen depending only on the supremum of the norm of the second fundamental form; Conjecture 3, whether positive total Gauss curvature forces a sign-sensitive mean-curvature domination inequality that would imply Conjecture 2; and the thick-layer dichotomy, whether every admissible thickness over a parabolic surface still binds or a smooth asymptotically flat example with empty discrete spectrum exists.

Background

The model. A quantum layer describes a quantum particle hard-wall confined between two copies of a surface $\Sigma\subset\mathbb{R}^3$ a fixed distance $2a$ apart: the configuration space is $\Omega_a\cong\Sigma\times(-a,a)$, the points at distance less than $a$ from $\Sigma$, with the Riemannian metric induced from $\mathbb{R}^3$, and the Hamiltonian is the Dirichlet Laplacian $-\Delta_{\Omega_a}$ (Dirichlet boundary conditions enforce the hard walls). For the flat layer $\mathbb{R}^2\times(-a,a)$ the spectrum is the single band $[\pi^2/(4a^2),\infty)$, whose bottom is the lowest Dirichlet eigenvalue of the transverse interval $(-a,a)$ of width $2a$. Eigenvalues below this threshold, if any, are isolated eigenvalues of finite multiplicity — the discrete spectrum — and correspond to bound states localized near the curved part of the layer rather than propagating to infinity. That purely geometric curvature can produce bound states was discovered for two-dimensional strips (quantum waveguides) by Exner and Šeba in 1989, and for layers by Duclos, Exner and Krejčiřík in 2001.

Geometry. Throughout, $\Sigma$ satisfies the standing hypotheses of Lu–Rowlett: it is a $C^2$, orientable, complete, non-compact, embedded surface that is not totally geodesic (not a plane) and is asymptotically flat, meaning its second fundamental form $B$ — the tensor measuring extrinsic bending, whose eigenvalues are the principal curvatures — tends to zero at infinity. The Gauss curvature $\kappa$ is the product of the principal curvatures (the intrinsic curvature), and the mean curvature $H$ is their average. Writing $B_\infty:=\sup_{p\in\Sigma}\|B(p)\|$, the map $(p,t)\mapsto p+t\vec N(p)$ is an immersion for all $a\in(0,B_\infty^{-1})$, so the layer does not fold over itself; these are the admissible thicknesses, thin ($a$ small) or thick ($a$ anywhere in the range). A complete surface is called parabolic if it admits no positive Green's function; equivalently, every metric ball has zero capacity, i.e. there exist cutoff functions equal to one on the ball with arbitrarily small Dirichlet integral $\int|\nabla\varphi|^2$ — a quantitative recurrence property. By the structure theory of surfaces with integrable Gauss curvature ($\kappa\in L^1(\Sigma)$, finite total curvature), such surfaces have at most quadratic area growth and are parabolic; and if the total Gauss curvature $\int_\Sigma\kappa\,d\Sigma$ is positive, then $\Sigma$ is diffeomorphic to $\mathbb{R}^2$ and $0<\int_\Sigma\kappa\,d\Sigma\le 2\pi$.

Prior results. Duclos–Exner–Krejčiřík (2001) and Carron–Exner–Krejčiřík (2004) proved existence of bound states, for layers thin enough or under extra hypotheses, whenever the total Gauss curvature is non-positive, when the total mean curvature is infinite with $\nabla H\in L^2$, or for cylindrically symmetric ends; Lin and Lu (2007) covered graphs of convex functions and surfaces ruled outside a compact set. The one regime not covered is sign-changing $\kappa$ with positive total Gauss curvature. Lu and Rowlett (2012) introduced weak $\kappa$-parabolicity — the requirement that the $\kappa$-capacity $\inf\int_\Sigma(2|\nabla\varphi|^2+\kappa\varphi^2)\,d\Sigma$ of a neighborhood of a point with $H\neq0$ be non-positive — and proved: (Theorem 1.3) weakly $\kappa$-parabolic surfaces, in particular all surfaces with $\int\kappa\le0$, bind for all thin layers, with a width threshold depending only on $B_\infty$; (Theorem 1.4) surfaces with $\kappa\ge0$ and $\kappa\in L^1$ bind for every admissible thickness $a\in(0,B_\infty^{-1})$; (Theorem 1.5) every parabolic surface binds for all sufficiently thin layers — which, since $\kappa\in L^1$ implies parabolicity, resolves the original surface-dependent existence conjecture in full (their Remark 5.1). Their method is variational: a bound state follows once a test function pushes the quadratic form below the threshold, and the binding energy is extracted from the sign-sensitive term $\int_\Sigma Hj\,d\Sigma$ for suitable cutoffs $j$ supported far out. What this method cannot handle is the subject of the open problems below.

Problem Statement

Settle the three questions Lu and Rowlett explicitly left open for the hard-wall quantum layer. Throughout, $\Sigma\subset\mathbb{R}^3$ is a $C^2$, orientable, complete, non-compact surface that is not totally geodesic and is asymptotically flat (its second fundamental form $B$ tends to zero at infinity); $\kappa$ and $H$ are its Gauss and mean curvatures; $B_\infty:=\sup_{p\in\Sigma}\|B(p)\|$; $\Omega_a\cong\Sigma\times(-a,a)$ is the quantum layer of width $2a$ with the induced metric and Dirichlet Laplacian $-\Delta_{\Omega_a}$; admissible thicknesses are $a\in(0,B_\infty^{-1})$; $B(R)$ denotes the closed geodesic ball of radius $R$ about a fixed base point of $\Sigma$.

(1) Conjecture 3 (curvature domination, a purely Riemannian statement). If $\kappa\in L^1(\Sigma)$ and $\int_\Sigma\kappa\,d\Sigma>0$, must there exist $\varepsilon_0>0$ such that for every $R>0$ one can find $j\in C_0^\infty(\Sigma\setminus B(R))$ with

$\Big(\int_\Sigma Hj\,d\Sigma\Big)^2>\varepsilon_0\int_\Sigma\big(|\nabla j|^2+j^2\big)\,d\Sigma\ ?$

Or does there exist a surface with $\kappa\in L^1(\Sigma)$ and $\int_\Sigma\kappa\,d\Sigma>0$ for which no such $\varepsilon_0>0$ exists?

(2) Conjecture 2 (uniform thin-layer existence). If $\kappa\in L^1(\Sigma)$, must there exist $\alpha=\alpha(B_\infty)>0$, depending only on $B_\infty$ and not otherwise on $\Sigma$, such that for every $a\in(0,\alpha)$ the discrete spectrum of $-\Delta_{\Omega_a}$ is non-empty? Or is there a uniformly bounded-curvature family of admissible surfaces with arbitrarily small admissible widths $a$ for which the discrete spectrum of $-\Delta_{\Omega_a}$ is empty? By Lu–Rowlett's Theorem 5.1 together with the earlier results for non-positive total Gauss curvature, an affirmative answer to (1) implies an affirmative answer to (2).

(3) Thick-layer dichotomy. Either prove that for every parabolic $\Sigma$ (parabolic meaning that the complete surface admits no positive Green's function, equivalently every geodesic ball has zero capacity) satisfying the standing hypotheses and every admissible thickness $a\in(0,B_\infty^{-1})$ the discrete spectrum of $\Omega_a$ is non-empty, or construct a parabolic $\Sigma$ satisfying the standing hypotheses and an admissible $a$ for which the discrete spectrum is empty.

These questions share one open core: the regime in which the mean curvature $H$ changes sign on every neighborhood of infinity, where the cutoff integrals $\int Hj$ can cancel and the variational method of the earlier works fails; the known partial results are the weaker domination inequality with $|H|$ (Proposition 5.1) and the eventual-fixed-sign case.

References

1. Zhiqin Lu and Julie Rowlett, 'On the discrete spectrum of quantum layers', Journal of Mathematical Physics 53 (2012), DOI 10.1063/1.4736412, https://doi.org/10.1063/1.4736412 (preprint arXiv:1110.6807, https://arxiv.org/abs/1110.6807)
2. P. Duclos, P. Exner, and D. Krejčiřík, 'Bound states in curved quantum layers', Communications in Mathematical Physics 223 (2001) 13–28, DOI 10.1007/pl00005582, https://doi.org/10.1007/pl00005582
3. G. Carron, P. Exner, and D. Krejčiřík, 'Topologically nontrivial quantum layers', Journal of Mathematical Physics 45 (2004) 774–784, DOI 10.1063/1.1635998, https://doi.org/10.1063/1.1635998
4. Christopher Lin and Zhiqin Lu, 'Existence of bound states for layers built over hypersurfaces in $\mathbb{R}^{n+1}$', Journal of Functional Analysis 244 (2007) 1–25, DOI 10.1016/j.jfa.2006.11.017, https://doi.org/10.1016/j.jfa.2006.11.017
5. P. Exner and P. Šeba, 'Bound states in curved quantum waveguides', Journal of Mathematical Physics 30 (1989) 2574–2580, DOI 10.1063/1.528538, https://doi.org/10.1063/1.528538
6. David Krejčiřík and Zhiqin Lu, 'Location of the essential spectrum in curved quantum layers', Journal of Mathematical Physics 55 (2014), DOI 10.1063/1.4893035, https://doi.org/10.1063/1.4893035 (preprint arXiv:1211.2541, https://arxiv.org/abs/1211.2541)
7. Pedro Freitas and David Krejčiřík, 'A lower bound to the spectral threshold in curved quantum layers', in Functional Analysis and Operator Theory for Quantum Physics, EMS Series of Congress Reports (2017) 261–269, DOI 10.4171/175-1/14, https://doi.org/10.4171/175-1/14
8. Pavel Exner and Vladimir Lotoreichik, 'Spectral Asymptotics of the Dirichlet Laplacian on a Generalized Parabolic Layer', Integral Equations and Operator Theory 92 (2020), DOI 10.1007/s00020-020-2571-x, https://doi.org/10.1007/s00020-020-2571-x (preprint arXiv:1805.12448, https://arxiv.org/abs/1805.12448)
9. David Krejčiřík and Jan Kříž, 'Bound States in Soft Quantum Layers', Publications of the Research Institute for Mathematical Sciences 60 (2024) 741–766, DOI 10.4171/prims/60-4-4, https://doi.org/10.4171/prims/60-4-4 (preprint arXiv:2205.04919, https://arxiv.org/abs/2205.04919)
10. F. Bakharev and S. Matveenko, 'Spectra of the Dirichlet Laplacian in 3-dimensional polyhedral layers', St. Petersburg Mathematical Journal 35 (2024) 597–610, DOI 10.1090/spmj/1818, https://doi.org/10.1090/spmj/1818 (preprint arXiv:2304.10575, https://arxiv.org/abs/2304.10575)
11. Pavel Exner, 'Geometrically induced spectral properties of soft quantum waveguides and layers', Reviews in Mathematical Physics (2023), DOI 10.1142/S0129055X23600036, https://doi.org/10.1142/S0129055X23600036