Dipole-Obstructed Cooper Pairing: Theory and Application to Superconductors
Penghao Zhu
zhu.3711@osu.eduDepartment of Physics, The Ohio State University, Columbus, OH 43210, USA
ââRui-Xing Zhang
ruixing@utk.eduDepartment of Physics and Astronomy, The University of Tennessee, Knoxville, TN 37996, USA
Department of Materials Science and Engineering, The University of Tennessee, Knoxville, TN 37996, USA
Abstract
Like electrons, Cooper pairs can carry a monopole charge if the pairing electrons come from two or more Fermi surfaces with different Chern numbers. In such an instance, a superconductor is necessarily nodal due to an inherent topological pairing obstruction. In this work, we show that a similar obstruction is also possible when there is only one Fermi surface involved in the pairing process. By developing a Chern-vorticity theorem, we have identified a class of Fermi surfaces with a quantized dipolar Berry flux pattern, where all intra-Fermi-surface Cooper pairings are âdipole-obstructedâ and nodal. As a real-world application, we find that the dipole obstruction plays a crucial role in stabilizing the superconducting nodal structure for half-Heusler compounds.
Introduction.â Superconductors (SCs) are generally classified by the symmetry pattern of their Cooper pairs. By definition, electron pairing in a conventional -wave SC is spatially isotropic, generating a uniform energy gap of the normal-state Fermi surface (FS) [1]. Meanwhile, an unconventional SC such as the cuprates can feature a gapless spectrum when its anisotropic pairing order has symmetry-enforced zeros in the momentum space [2, 3, 4]. In experiments, the SC gap structure can be feasibly probed by angle-resolved photoemission spectroscopy [5], scanning tunneling spectroscopy [6], penetration depth measurements [7], etc. Such gap information often offers valuable insights into unraveling the nature of Cooper pairs for a new SC candidate.
The pairing symmetry, however, does not fully account for the gap structure of SCs. For example, a SC is found to be necessarily nodal, once the electrons forming a Cooper pair come from two FSs with opposite Berry monopole charges [8, 9, 10, 11]. Such an inability to develop a full energy gap is intrinsic to the topological texture of the FS, which holds even when is constant. This striking phenomenon has sparked a growing research interest in uncovering similar mechanisms of topologically obstructed nodal pairing orders for FSs carrying an Euler index [12, 13, 14] or a index [15]. Notably, most existing theories have assumed the obstructed pairings to involve multiple FSs, while generalizations to intra-FS electron pairings are less explored.
In this work, we have identified a new class of FSs where all intra-FS Cooper pairings exhibit topology-enforced zeros. The FS of our interest encloses a quantized dipole of the Berry curvature [16], i.e., there exists a Berry flux of through either half of the FS, while the net Berry flux vanishes. As a proof of concept, we consider a minimal model with such a Berry-dipole FS and find all intra-FS Cooper pairings are dipole-obstructed from fully gapping out the FS. Specifically, the resulting SC state always exhibits zero-energy Weyl nodes and/or nodal loops in the Bogoliubov-de Gennes (BdG) spectrum. Finally, we revisit the half-Heusler SCs such as YPtBi [17, 4] and LuPdBi [18, 19, 20], which are believed to feature a mixed-parity singlet-septet pairing [21]. We find that the dipole obstruction naturally exists in the septet pairing channel and further clarify its contribution to the BdG line nodes observed in experiments.
Chern-Vorticity Theorem and Pairing Zeros.â We start by presenting a Chern-vorticity theorem that will guide us to the target FSs. As shown in Fig. 1(a), we consider two 2D closed or effectively closed manifolds and in -space, as well as two Bloch states and with . Here, we define an open manifold to be effectively closed if the Bloch states over each of its boundaries are identical. Generally, and are related by a -space transformation with Here denotes a point-group operation (e.g., rotation and mirror) that satisfies and is a translation in -space. We now consider the matrix element of a general two-particle operator bridging electrons on and :
(1)
When displays a vortex pattern around , the value of necessarily vanishes and is thus topologically obstructed. Such a vortex is captured by a vorticity index around , defined as . The Chern-vorticity theorem proven in the Supplemental Material (SM)Â [22] dictates that the net vorticity on , , is determined by
(2)
where is the single-particle Chern number of on . The loop integral is summed over all possible boundaries of , which, by default, vanishes for a closed .
Focusing on SCs, we choose and in order to properly project the pairing operator , since physically describes hoppings between electrons and holes. Here, is the Bloch state of electrons on the Fermi surface . It is crucial to note that , the Chern number for , is exactly opposite to , the electronic Chern number for on  [22]. As a concrete example, we immediately arrive at when (i) are both closed FSs with and (ii) is the spatial inversion with , which exactly corresponds to the obstructed nodal pairing for monopole SCs in Ref. [9]. Therefore, the Chern-vorticity theorem manifests as a natural generalization of Ref. [9], but its application transcends the realm of SC systems with inter-FS pairings [23, 24, 25].
Berry-Dipole Fermi Surface.â The above Chern-vorticity theorem suggests that for a single Fermi surface , the intra-FS pairing between two patches can be topologically obstructed if each patch is effectively closed and carries a nonzero quantized Berry flux. Further requiring the net Berry flux to be zero, we are thus looking for a FS with a quantized dipolar texture of Berry curvature, i.e., a Berry-dipole FSÂ [26, 16, 27, 28, 29].
As discussed in Refs. [26, 16] and reviewed in the SM [22], quantization of a Berry-dipole can be achieved by enforcing a mirror symmetry along the dipole axis. As a concrete example, we consider a minimal model with Berry-dipole physics proposed in Ref. [16],
(3)
Besides mirror symmetry along -direction , respects a continuous rotation symmetry around with , where is a diagonal matrix. Regularizing will reduce the symmetry group to , which will be exploited for later pairing analysis. When , describes a 3D minimal Weyl semimetal (WSM) with two Weyl nodes at , respectively. At , the Weyl nodes merge at to form a quadratic band touching. The annihilation of Weyl nodes, however, does not lead to a mass generation for the low-energy electrons. Instead, we note that with features a 1D doubly degenerate nodal loop in the plane, i.e., a nodal-loop semimetal (NLSM). This unexpected robustness of Weyl nodes arises from the Berry-dipole delicate topological charges [26, 16].
Let us generally denote the FS patch with () as (). For , we always find a single spherical-like FS regardless of the value of , where and are hence the north and south hemispheres, respectively. Undergoing the Lifshitz transition with , the WSM with comprises a pair of closed spherical FSs (i.e., ), while a single torus-like FS is found for the NLSM () with now being both open and of an annulus shape. Evidently, the boundaries of and , if present, are always sitting in the plane regardless of . Consequently, the eigenstates of along each boundary can always be uniform, owing to the symmetry. Hence, we conclude that and must be either closed or effectively closed for all choices of and . This guarantees the FS(s) of to always feature a quantized Berry dipole, as explicitly confirmed in Fig. 1(b).
Dipole-Obstructed Pairing.â We are now ready to explore Cooper pairing physics on the above Berry-dipole FS by updating to a BdG Hamiltonian,
(4)
The pairing matrix generally takes the form and the Fermi statistics requires . For our purpose, we consider expanding up to and classify all pairing channels based on the irreducible representations (irreps) of group in Table. 1 [22]. For , describes the intra-FS pairings (as there is only one FS). Following Eq. 1, the projected Cooper pairing onto is given by , where are normal states over .
We first note that a general pairing with must be topologically obstructed since , which directly applies to all pairings. To see this, note that must be eigenstates of , since . As a result, any will vanish on the equator upon projection, further leading to . Meanwhile, the component of can contribute to by the winding phase of () around the equator for (), so long as . Here, we have defined and . Based on Table. 1, it is straightforward to see that and pairings are the only two pairing channels that can contribute to . Since must be an odd function of for both and irreps, the corresponding phase winding of around the equator must be with  [22]. This necessarily leads to a non-zero odd-integer-valued for all relevant and pairings. Since all other pairings feature and , we conclude that all intra-FS pairing channels for the Berry-dipole FS are obstructed and nodal.
Table 1: Pairing classification for Berry-dipole FS following the irreps. For -pairing, is a constant.
As a concrete example, let us focus on to explicitly illustrate the dipole-induced obstruction and relegate the discussions on other pairing channels in the SMÂ [22]. Without loss of generality, we set and the FS at is a sphere of radius , which can be parameterized by a polar angle and an azimuthal angle . For , we find the electron wavefunction to be , which is a constant spinor at the equator. Under this gauge choice, are both effectively closed, while becomes singular at the north pole (). This implies an obstruction to define a globally smooth gauge, thanks to the non-zero Chern number on each hemisphere.
To ensure the wavefunction is merely locally singular, we consider for , and choose a different gauge choice for with .
Straightforward calculations lead to:
(5)
Apparently, features two zeros on , one at the north pole and another at the equator. Based on the form of , it is easy to see that the vorticity of the north-pole zero is . Meanwhile, the vorticity for the equator zero is also , which can be achieved by performing a clockwise loop integral of the phase of at  [22]. Together, we find the net vorticity for , which is consistent with the prediction of the Chern-vorticity theorem.
We now make a few remarks. First of all, the vorticity-induced pairing zeros always exist, regardless of the detailed form of . The zeros of itself may lead to additional zeros of , beyond the topologically obstructed ones. Second, we highlight that the counting of for intra-FS pairing does depend on a special gauge choice with which are effectively closed. Crucially, symmetry guarantees the existence of such a gauge in our case [22], ensuring the counting is always possible. Apparently, the nodal structure of must be gauge-invariant. Therefore, applying the Chern-vorticity theorem under a proper gauge thoroughly informs the obstructed pairing zeros that are gauge-independent.
Finally, let us take the pairing as an example to numerically explore the nodal nature of the obstructed SC phases. In Fig. 2, we plot the pairing gap function on the corresponding FS and track its evolution as a function of and , where the BdG Weyl nodes and nodal loops are highlighted in black. When , we find a single sphere-like FS for both NLSM () and WSM (), which exhibits both a BdG nodal loop at the equator and a pair of BdG Weyl nodes at the poles. This directly follows our analysis of vorticity counting. Further reducing triggers a Lifshitz transition for NLSM, which makes the point nodes merge. Surprisingly, the merging of these oppositely charged Weyl nodes leads to a second nodal loop at . This is in contrast with the expected pair annihilation process in conventional Weyl SCs, as shown in the SM [22]. Meanwhile, a similar Lifshitz transition for WSM shrinks the BdG nodal loop to a point, which further splits into a pair of Weyl nodes. This robust conversion between Weyl nodes and nodal loops for the BdG bands directly arises from the effective closedness of  [22], which is reminiscent of similar behaviors in its normal state  [26].
Application to SCs. â As a real-world example, we now demonstrate that the BdG line nodes of half-Heusler-based SCs such as YPtBi and LuPdBi are deeply rooted in the mechanism of dipole obstruction.
We start with a generalized Luttinger-Kohn Hamiltonian as a normal state of interest [30, 31],
(6)
where , , ,
, ,
and . When , recovers the standard isotropic Luttinger semimetal, which nicely captures the low-energy normal state of general half-Heusler SCs. While several candidate pairing channels have been theoretically proposed [32, 33, 34, 35, 36], a recent experiment reports a doping-dependent nodeless-to-nodal transition in LuPdBi [20]. This observation strongly supports the mixed-parity singlet-septet pairing  [21], with and
(7)
which belongs to the irrep of group. Specifically, with just , the BdG spectrum of displays point nodes at axes [c.f. Fig. 3(b)]. Turning on further inflates the point nodes into nodal loops, eventually leading to a full gap when dominates. However, why the -wave pairing can lead to point nodes can be quite puzzling. For example, let us set and focus on the BdG physics along the axis. While for , the existence of point nodes implies that the projection of onto the FS must vanish.
To trace the origin of these point nodes, we note that when , exactly describes a Dirac-dipole semimetal [37] that comprises two decoupled copies of the Berry-dipole Hamiltonian at , up to some parameter rescaling. The block-diagonal nature of suggests an emergent spin conservation symmetry generated by , with which the two Berry-dipole blocks are carrying opposite spins. Note that an generally reduces to the time-reversal symmetry , with the complex conjugation. Besides, fully inherits the symmetry of with an updated rotation generator and . For a finite , the Dirac-dipole semimetal features a pair of decoupled FSs, one for each spin sector. Hence, all intra-FS pairings are intra-spin pairings, and vice versa.
Notably, the first three terms of describe the intra-spin pairing process. Within each spin sector, these pairings correspond to the pairings in Table 1, all of which are hence dipole-obstructed for . Applying the Chern-vorticity theorem in Eq. 2, we find this set of intra-spin pairing terms to carry a nontrivial pairing vorticity of  [22], which strictly enforces pairing zeros (or equivalently point nodes) on both the north and south poles of the FS. Meanwhile, the remaining pairing terms of are inter-spin and exhibit no obvious vortex pattern. Nonetheless, both inter-spin pairings vanish on the axis and their existence is thus invisible to the dipole-obstructed pairing zeros. In Fig. 3(a), we numerically confirm the predicted pairing zeros of by mapping out the pairing gap on the FS of the Dirac-dipole semimetal.
We now revisit the isotropic limit () that the SCs follow. Since all -relevant terms in vanish on the axis, turning on will not disrupt the obstructed pairing zeros in the Dirac-dipole limit. Moreover, respects a larger symmetry group (e.g., ) than that of , which features a three-fold rotation that permutes , and . Hence, BdG point nodes should also emerge at both and axes, which is explicitly confirmed in Fig. 3(b). In other words, enables us to designate any of the -axes as the -rotation axis for the previous Dirac-dipole discussion. In this case, all of the six -wave-induced point nodes for SCs have a dipole-obstructed origin.
Finally, let us turn on a subdominant to achieve the BdG line nodes, as shown in Fig. 3(c). Specifically, lifts the spin-degeneracy of FSs by breaking the inversion symmetry, which leads to accidental inter-FS band crossings that manifest as line nodes. While the dipole-obstructed pairing nodes [purple dots in Fig. 3(d)] now show up at a finite energy , their existence ensures the BdG line nodes to necessarily show up even for an arbitrarily small . Note that we have omitted a small asymmetric spin-orbit coupling (ASOC) term in Eq. 6, which is intrinsic to zinc-blende materials. While the ASOC is key to mixing and -wave pairings, its effect on the BdG band spectrum is similar to that of the -wave pairing. A detailed discussion on the ASOC term can be found in the SM [22].
Discussions.â To summarize, we have established a general theoretical framework for comprehending topologically obstructed nodal pairing over a single FS. This leads us to uncover a class of Berry-dipole FSs where the SC state induced by any intra-FS pairing will be dipole-obstructed and nodal. The hidden Berry-dipole physics in the Luttinger-Kohn model further motivates us to explore the pairings in half-Heusler SCs. Focusing on the singlet-septet pairing channel, we find that the dipole obstruction contributes significantly to the BdG line nodes observed in experiments. Notably, Berry and Dirac dipoles, as well as Luttinger semimetals, have been recently established as critical points for delicate topological phases [16, 37, 31]. Therefore, our dipole-obstructed pairing offers the first example of how delicate topological bands can decisively impact the correlated electronic orders in real-world quantum materials.
We emphasize that our Chern-vorticity theorem goes beyond SCs and applies to general electronic orders such as charge-density waves, excitons, magnetism, etc. It would be interesting to explore possible dipole-obstructed phenomena in other non-superconducting Luttinger semimetals such as HgTe [38] and Pr2Ir2O7 [39]. Besides, we find many inter-spin pairings for the system to feature emergent nodal structures in the BdG spectrum, despite carrying no vortex structure. As elaborated in the SM [22] for the Dirac-dipole model, these âunobstructedâ pairing zeros arise from the geometric textures of the Bloch states on the FSs, which are beyond the scope of the Chern-vorticity theorem. How to interpret these geometry-relevant nodal pairings is an absolutely intriguing question for future research.
Acknowledgements.
We thank J. Yu, Y. Wang, L.-H. Hu, A. Alexandradinata, and Y.-M. Lu for helpful discussions. P.Z was primarily supported by the Center for Emergent Materials, an NSF MRSEC, under award number DMR-2011876. R.-X.Z. is supported by a start-up fund of the University of Tennessee.
[22]See the Supplemental Material for (i) a detailed proof of the Chern-vorticity theorem; (ii) detailed analysis for general pairings in Berry-dipole semimetals and inter-spin pairings in Dirac-dipole semimetals; and (iii) case studies on WSM, Berry dipole SC, and perturbations from ASOC.
Bobrow et al. [2020]E. Bobrow, C. Sun, and Y. Li, Monopole charge density wave states in weyl semimetals, Phys. Rev. Res. 2, 012078 (2020).
Bultinck et al. [2020]N. Bultinck, S. Chatterjee, and M. P. Zaletel, Mechanism for anomalous hall ferromagnetism in twisted bilayer graphene, Phys. Rev. Lett. 124, 166601 (2020).
Zhu and Alexandradinata [2023]P. Zhu and A. Alexandradinata, Anomalous shift and optical vorticity in the steady photovoltaic current, arXiv preprint arXiv:2308.08596 (2023).
Sun et al. [2018]X.-Q. Sun, S.-C. Zhang, and T. c. v. Bzduƥek, Conversion rules for weyl points and nodal lines in topological media, Phys. Rev. Lett. 121, 106402 (2018).
Boettcher and Herbut [2018]I. Boettcher and I. F. Herbut, Unconventional superconductivity in luttinger semimetals: Theory of complex tensor order and the emergence of the uniaxial nematic state, Phys. Rev. Lett. 120, 057002 (2018).
Yu and Liu [2018]J. Yu and C.-X. Liu, Singlet-quintet mixing in spin-orbit coupled superconductors with fermions, Phys. Rev. B 98, 104514 (2018).