In two 1970 papers in the Journal of Low Temperature Physics, A. A. Abrikosov argued that a semimetal in an extremely high magnetic field — the extreme quantum limit, where only the lowest Landau level is occupied — becomes unstable toward formation of bound electron-hole pairs (excitons), that is, toward an insulating excitonic state. The analysis was carried out at leading-logarithmic accuracy with a parquet summation of the electron-hole vertex (a resummation of the leading logarithmic diagrams in the two scattering channels simultaneously), introducing an auxiliary short-range coupling and yielding a binding-energy formula that contains undetermined order-unity constants which enter the result exponentially. The problem left open by that work and never completed since is a controlled quantitative theory of this instability: computing the vertex pole structure beyond leading logarithms (including non-logarithmic diagram classes and the Landau-level wave-function structure), fixing the numerical constants so that the binding energy becomes predictive, deriving the effective short-range coupling microscopically from the band structure, and delimiting the regime of validity. The field-induced insulating phases now observed in graphite (30-80 T) and the still-unexplained valley-symmetry-broken state of bismuth (bismuth carries three equivalent conduction-band pockets, or valleys, and the observed state spontaneously breaks their threefold symmetry) give this completion its experimental motivation; they postdate the source and therefore do not enlarge the open problem it left. Later literature reformulated parts of the problem — mean-field phase diagrams for bismuth and graphite, renormalization-group theories of the graphite phases, numerical parquet studies of simplified semimetal models — but the specific quantitative completion is untouched, and current reviews explicitly identify the missing microscopic theory of the observed quantum-limit phases.

Background

A semimetal is a metal in which the conduction band slightly overlaps in energy with the valence band, so that small pockets of electrons and holes coexist even at zero temperature; bismuth and graphite are the classic examples. In a magnetic field HH (quoted in oersteds in this literature; 10410^{4} Oe =1=1 tesla), the transverse motion of charge carriers is quantized into Landau levels, and when the field is so high that only the lowest Landau level remains occupied — the extreme quantum limit — the carrier spectrum becomes effectively one-dimensional along the field direction. In a compensated semimetal, with equal electron and hole densities, the Fermi momentum then grows with field as p0≈eH/cp_{0}\approx\sqrt{eH/c} (Gaussian units), so that at sufficiently large HH the Coulomb interaction can exceed the Fermi energy εF\varepsilon_{F} in relative strength, and a paired electron-hole state becomes energetically possible.

An exciton is a bound electron-hole pair; a semimetal or narrow-gap semiconductor in which such pairs condense spontaneously is called an excitonic insulator, because the binding gaps the Fermi surface and destroys metallic conduction. Part I of Abrikosov's series analyzed this question under a weak point-like interaction, appropriate when the Coulomb logarithm is of order unity, and found that electron-hole excitons form while Cooper pairing of like charges is absent; the source paper (part II) treats the opposite case of extremely high fields, in which the Coulomb logarithm is large and the long-range part of the Coulomb interaction must be kept explicitly. The technical difficulty is that the effectively one-dimensional kinematics makes both the particle-hole and the particle-particle channels of perturbation theory logarithmically divergent, so that the electron-hole vertex Γ\Gamma — the two-particle scattering amplitude — acquires double logarithms of the momentum transfer; the parquet method, the diagram technique developed for the X-ray absorption edge problem and systematized by Nozières, Gavoret and Roulet, which sums the leading logarithms in all channels simultaneously, is the standard tool in this situation. Abrikosov approximates screening by that of the lowest Landau level, ε0ϰ2=2(me+mh)e2eH/(πp0c)\varepsilon_{0}\varkappa^{2}=2(m_{e}+m_{h})e^{2}eH/(\pi p_{0}c), where ϰ\varkappa is the inverse screening radius, ε0\varepsilon_{0} is the dielectric constant contributed by the deep bands — treated as field-independent for H<107H<10^{7} Oe, below the threshold at which the field would reshape the deep levels themselves — and me,mhm_{e},m_{h} are the electron and hole effective masses. He shows that the vertex with small momentum transfer factorizes as Γ∼γ({ln⁡})/ky2+kz2+ϰ2\Gamma\sim\gamma(\{\ln\})/\sqrt{k_{y}^{2}+k_{z}^{2}+\varkappa^{2}}, with γ\gamma a logarithmic amplitude; that in the symmetric kinematic regime, in which all external longitudinal momenta are of the same order, the leading parquet contributions from the parallel and antiparallel channels cancel pairwise, so that this vertex remains at its bare value at leading-logarithmic accuracy; and that the vertex Γ2\Gamma_{2} with longitudinal momentum transfer of order 2p02p_{0} satisfies a closed parquet equation whose solution, in the large-Coulomb-logarithm regime considered, has a single pole (at smaller interaction constants the same equation yields an infinite ladder of poles). A pole in the two-particle vertex signals the excitonic instability, and its position translates into a binding energy ω0\omega_{0} given by the exponential formula of Eq. (24) of the source, transcribed below; where that transcription differs from [1], [1] governs:

ω0=α εF exp⁡ ⁣[πp0ε04m∗e2 ln⁡ ⁣(ρln⁡(p0/ϰ))ln⁡(p0/ϰ)]4,\omega_{0}=\alpha\,\varepsilon_{F}\,\exp\!\left[\frac{\pi p_{0}\varepsilon_{0}}{4m^{*}e^{2}}\,\frac{\ln\!\left(\rho\ln(p_{0}/\varkappa)\right)}{\ln(p_{0}/\varkappa)}\right]^{4},

where m∗m^{*} is the reduced effective mass and α\alpha and ρ\rho are dimensionless constants of order unity left undetermined by the calculation. Because the constants enter the result exponentially, the formula is exponentially sensitive to them and to non-logarithmic corrections, and is therefore not yet a prediction.

The calculation is exact only at leading logarithmic accuracy under several idealizations: the Landau-level magnetic wave functions were factored out of the internal vertex integrations after checking only low-order diagrams; only the lowest Landau level was retained; an auxiliary short-range coupling λ\lambda, with e2/ε0≪λ≪(e2/ε0)ln⁡(p0/ϰ)e^{2}/\varepsilon_{0}\ll\lambda\ll(e^{2}/\varepsilon_{0})\ln(p_{0}/\varkappa), was introduced by hand to represent non-logarithmic effects, only the limiting identification λ0→e2m∗/ε0p0\lambda_{0}\to e^{2}m^{*}/\varepsilon_{0}p_{0} being considered; and the derivation requires ln⁡(ε0 p0/m1e2)≫1\ln(\sqrt{\varepsilon_{0}}\,p_{0}/m_{1}e^{2})\gg 1, with m1m_{1} the smaller of the effective masses. The open problem is the quantitative completion of this theory. Its later history branched rather than completed it: 1970s work on bismuth (Fukuyama and Nagai; Nagai and Tsuzuki; Yoshioka) reformulated the excitonic instability with mean-field methods and realistic bands; Yoshioka and Fukuyama predicted the analogous field-induced transition of graphite, and the field-induced phases of graphite above roughly 30 tesla are now experimentally established; and modern studies have both observed still-unexplained high-field states in bismuth, proposed spin-nematic excitonic-insulator orderings for the graphite phases, and re-derived the parquet program numerically for simplified models. No later work has completed the original calculation itself.

Problem Statement

Construct a controlled quantitative theory of exciton formation in a compensated semimetal in the extreme quantum limit, completing Abrikosov's leading-logarithmic parquet analysis for the semimetal model of that paper. The problem comprises four interlocking scientific subgoals that support this single objective.

(1) Vertex structure beyond leading logarithms. Determine the electron-hole vertex of the lowest-Landau-level semimetal beyond leading logarithmic accuracy: which non-logarithmic diagram classes contribute, whether the factorization of the vertex into a logarithmic amplitude and a Coulomb factor and the pairwise cancellation of the parallel- and antiparallel-channel contributions in the symmetric kinematic regime (all external longitudinal momenta of the same order) survive, and whether the vertex Γ2\Gamma_{2} with longitudinal momentum transfer of order 2p02p_{0} still has a single pole in the large-Coulomb-logarithm regime, where the leading-logarithmic theory finds one. The Landau-level wave functions must be tracked through the internal vertex integrations rather than factored out after checking only low orders.

(2) Numerical constants. Evaluate the order-unity constants α\alpha and ρ\rho defined by the exponential binding-energy formula of the source (its Eq. (24)), with an error analysis reflecting the exponential sensitivity of ω0\omega_{0} to these constants, so that the binding energy ω0\omega_{0} becomes a quantitative prediction; equivalently, compute the binding energy and threshold field directly at the higher accuracy.

(3) Microscopic effective coupling. Derive the short-range coupling λ\lambda from the underlying Coulomb interaction and the band structure, replacing the hand-introduced auxiliary constant and its limiting identification λ0→e2m∗/ε0p0\lambda_{0}\to e^{2}m^{*}/\varepsilon_{0}p_{0} by a controlled approximation with a stated error.

(4) Regime of validity. Establish the range of magnetic fields, mass ratios, and dielectric parameters over which the completed theory applies, including the fate of the condition ln⁡(ε0 p0/m1e2)≫1\ln(\sqrt{\varepsilon_{0}}\,p_{0}/m_{1}e^{2})\gg 1 and of the assumed field independence of the deep-band dielectric constant below 10710^{7} Oe, and whether the asymptotic large-logarithm regime overlaps the fields accessible in experiments (tens of tesla).

These subgoals are linked because each feeds the next: the beyond-logarithmic vertex structure determines the constants, the constants and the microscopic coupling determine the prediction, and the validity regime determines where the prediction applies. In the limit in which non-logarithmic corrections are formally dropped, the completed theory must reduce to the leading-logarithmic parquet results of the source — the single pole of Γ2\Gamma_{2} and the exponential binding-energy form of its Eq. (24). Within the same controlled framework, a leading-logarithmic result — the single pole of Γ2\Gamma_{2}, the factorization and pairwise cancellation of the small-momentum vertex, or the exponential form of Eq. (24) — may turn out to be an artifact rather than an established result. Abrikosov's own 1973 continuation evaluated the theory for real materials (bismuth-like band structures) but remained within logarithmic accuracy, and comparison with the observed field-induced phases of graphite (roughly 30-80 T) and with the valley-symmetry-broken state of bismuth motivates the problem's significance.

References

  1. A. A. Abrikosov, "On the possibility of exciton formation in semimetals in extremely high magnetic fields. II", Journal of Low Temperature Physics 2, 175-186 (1970). DOI: 10.1007/BF00628174, https://doi.org/10.1007/BF00628174
  2. A. A. Abrikosov, "The metal-dielectric transition in semimetals in extremely high magnetic fields. I", Journal of Low Temperature Physics 2, 37-58 (1970). DOI: 10.1007/BF00628100, https://doi.org/10.1007/BF00628100
  3. A. A. Abrikosov, "Transition of a bismuth-type semimetal to an excitonic insulator in a strong magnetic field", Journal of Low Temperature Physics 10, 3-34 (1973). DOI: 10.1007/BF00655239, https://doi.org/10.1007/BF00655239
  4. A. A. Abrikosov, "Galvanomagnetic Phenomena in Metals in the Quantum Limit", Zhurnal Eksperimental'noi i Teoreticheskoi Fiziki 56, 1391 (1969); English translation in Journal of Experimental and Theoretical Physics 29, 746 (1969). No DOI; JETP archive record (English translation): http://jetp.ras.ru/cgi-bin/e/index/e/29/4/p746?a=list
  5. Hidetoshi Fukuyama and Tatsuzo Nagai, "Excitonic Instability and Ultrasonic Attenuations in Strong Magnetic Fields", Journal of the Physical Society of Japan 31, 812-822 (1971). DOI: 10.1143/JPSJ.31.812, https://doi.org/10.1143/JPSJ.31.812
  6. Tatsuzo Nagai and Toshio Tsuzuki, "Excitonic phase in a strong magnetic field. I. Phase diagram", Journal of Low Temperature Physics 25, 9-26 (1976). DOI: 10.1007/BF00654821, https://doi.org/10.1007/BF00654821
  7. Daijiro Yoshioka, "Excitonic Phase and Gas-Liquid Transition of Semimetals in a Strong Magnetic Field", Journal of the Physical Society of Japan 45, 1165-1173 (1978). DOI: 10.1143/JPSJ.45.1165, https://doi.org/10.1143/JPSJ.45.1165
  8. Daijiro Yoshioka and Hidetoshi Fukuyama, "Electronic Phase Transition of Graphite in a Strong Magnetic Field", Journal of the Physical Society of Japan 50, 725-726 (1981). DOI: 10.1143/JPSJ.50.725, https://doi.org/10.1143/JPSJ.50.725
  9. K. Takahashi and Y. Takada, "Charge- and spin-density-wave instabilities in high magnetic fields in graphite", Physica B: Condensed Matter 201, 384-386 (1994). DOI: 10.1016/0921-4526(94)91120-7, https://doi.org/10.1016/0921-4526(94)91120-7
  10. P. Nozières, J. Gavoret and B. Roulet, "Singularities in the X-Ray Absorption and Emission of Metals. II. Self-Consistent Treatment of Divergences", Physical Review 178, 1084-1096 (1969). DOI: 10.1103/PhysRev.178.1084, https://doi.org/10.1103/PhysRev.178.1084
  11. Anatoley T. Zheleznyak, Victor M. Yakovenko and Igor E. Dzyaloshinskii, "Parquet solution for a flat Fermi surface", Physical Review B 55, 3200-3215 (1997). DOI: 10.1103/PhysRevB.55.3200, https://doi.org/10.1103/PhysRevB.55.3200
  12. Benoît Fauqué, Baptiste Vignolle, Cyril Proust, Jean-Paul Issi and Kamran Behnia, "Electronic instability in bismuth far beyond the quantum limit", New Journal of Physics 11, 113012 (2009). DOI: 10.1088/1367-2630/11/11/113012, https://doi.org/10.1088/1367-2630/11/11/113012
  13. Jason Alicea and Leon Balents, "Bismuth in strong magnetic fields: Unconventional Zeeman coupling and correlation effects", Physical Review B 79, 241101 (2009). DOI: 10.1103/PhysRevB.79.241101, https://doi.org/10.1103/PhysRevB.79.241101
  14. Huan Yang, Benoît Fauqué, Liam Malone, Arlei B. Antunes, Zengwei Zhu, Ctirad Uher and Kamran Behnia, "Phase diagram of bismuth in the extreme quantum limit", Nature Communications 1, 47 (2010). DOI: 10.1038/ncomms1039, https://doi.org/10.1038/ncomms1039
  15. Benoît Fauqué, David LeBoeuf, Baptiste Vignolle, Marc Nardone, Cyril Proust and Kamran Behnia, "Two Phase Transitions Induced by a Magnetic Field in Graphite", Physical Review Letters 110, 266601 (2013). DOI: 10.1103/PhysRevLett.110.266601, https://doi.org/10.1103/PhysRevLett.110.266601
  16. Z. Zhu, R. D. McDonald, A. Shekhter, B. J. Ramshaw, K. A. Modic, F. F. Balakirev and N. Harrison, "Magnetic field tuning of an excitonic insulator between the weak and strong coupling regimes in quantum limit graphite", Scientific Reports 7, 1733 (2017). DOI: 10.1038/s41598-017-01693-5, https://doi.org/10.1038/s41598-017-01693-5
  17. Zengwei Zhu, Benoît Fauqué, Kamran Behnia and Yuki Fuseya, "Magnetoresistance and valley degree of freedom in bulk bismuth", Journal of Physics: Condensed Matter 30, 313001 (2018). DOI: 10.1088/1361-648X/aaced7, https://doi.org/10.1088/1361-648X/aaced7
  18. Zhiming Pan, Xiao-Tian Zhang and Ryuichi Shindou, "Theory of metal-insulator transitions in graphite under high magnetic field", Physical Review B 98, 205121 (2018). DOI: 10.1103/PhysRevB.98.205121, https://doi.org/10.1103/PhysRevB.98.205121
  19. Zhiming Pan and Ryuichi Shindou, "Ground-state atlas of a three-dimensional semimetal in the quantum limit", Physical Review B 100, 165124 (2019). DOI: 10.1103/PhysRevB.100.165124, https://doi.org/10.1103/PhysRevB.100.165124 (preprint: arXiv:1907.09153, https://arxiv.org/abs/1907.09153)
  20. René Côté, Gautier D. Duchesne and Santiago F. Lopez, "Internodal excitonic state in a Weyl semimetal in a strong magnetic field", Physical Review B 110, 035162 (2024). DOI: 10.1103/PhysRevB.110.035162, https://doi.org/10.1103/PhysRevB.110.035162