A place for peculiar ideas.

Research article / 6 minute read

Magnetic Proximity and CISS Magnetoresistance Beyond the Jullière Limit (Zhang, Sun et al., 2025)

Electrical tests of Chirality-Induced Spin Selectivity (CISS) (CISS) keep producing magnetoresistance values that violate the Jullière model, a discrepancy the paper says has puzzled researchers for a long time. A theory paper by Tian-Yi Zhang, Yue Mao, Peng-Yi Liu, Ai-Min Guo and Qing-Feng Sun proposes an explanation. The authors are based at Peking University's International Center for Quantum Materials, Central South University in Changsha and the Hefei National Laboratory, and the paper is posted as arXiv:2509.05588 (version 2, 3 December 2025). Their proposal is that the ferromagnetic electrode magnetises the end of the chiral molecule it touches. That effectively moves the magnetic boundary into the molecule. Within this dossier, that sits as an alternative to other explanations of the same puzzle, such as the electrostatic account of Tirion and van Wees and the spinterface model.

The puzzle: magnetoresistance larger than the magnet allows

In a typical transport experiment, a chiral molecule sits between a nonmagnetic electrode and a ferromagnetic one. The experimenters measure how the current changes when the ferromagnet's magnetisation is flipped. The paper analyses this with the Jullière model, the textbook account of spin-dependent tunnelling. In that model the magnetoresistance (MR) is the product of two numbers: the spin polarisation the molecule gives to the electrons, and the spin polarisation of the ferromagnet. The molecule's polarisation cannot exceed 100 per cent, so the MR can never be larger than the electrode's own polarisation. The paper gives typical values for the common electrode metals: about 33% for nickel, 44% for iron and 45% for cobalt. Yet it notes that CISS experiments have widely reported MR as high as 80%. According to the paper, this violation has been noticed for a long time and still lacks a satisfactory answer.

How the model works

The group uses a tight-binding model of a single-helical protein, carried over from earlier Guo–Sun work. In this model an electron can hop between neighbouring sites and also between more distant ones, so it can travel along several paths at once. The molecule is 30 sites long. Each site is twisted 5π/9 from the last for a right-handed helix and −5π/9 for a left-handed one. The paper explains the spin selectivity as interference between these paths. Spin-orbit coupling (SOC) adds a phase that depends on spin, so spin-up and spin-down electrons interfere differently.

The main parameter change concerns SOC. The paper notes that earlier models of this kind set it at about one-tenth of the hopping integral; here it is 0.01 of the intrachain hopping, two orders of magnitude below the hopping and roughly ten times weaker than in that earlier work. With hopping taken as 100 meV, that means an SOC of only 1 meV, which the authors say matches realistic values. Even so, the model with two nonmagnetic electrodes gives a spin polarisation of up to 74.4%. That polarisation drops to exactly zero if either SOC or chirality is removed, and it reverses sign for the opposite handedness.

What the magnetic proximity effect adds

Next the authors replace one electrode with a ferromagnet of 30% polarisation. Without any extra ingredient the computed MR is small and always stays below that 30%. The paper stresses that this holds for every parameter choice tried, so the plain model obeys Jullière.

The new ingredient is a magnetic proximity effect. Electron states on the molecule's last site mix with the many states of the ferromagnet next to it. In the model, this gives that site a magnetisation M and an energy shift ε_FM. The paper cites literature putting both at roughly hundreds of meV, but its headline cases simply set M = −ε_FM = t1, which is 100 meV with the stated hopping. With this term the MR reaches 67.3%. The paper also states that longer molecules exceed 80%, citing its supporting figures for that result. The effect is largest when M and ε_FM have opposite signs and equal size. When they have the same sign, the MR flips sign but its size can still exceed the electrode's polarisation.

The authors also give an analytical picture. Once the last site is magnetised, the boundary between nonmagnetic and ferromagnetic material sits one site further into the molecule. The Jullière product then uses that site's modified polarisation instead of the electrode's. For suitable parameters, one spin's density of states on that site becomes very small, and the modified polarisation can approach 100%. The MR then approaches the molecule's own spin polarisation. The authors report that this analytical formula closely matches their numerical results.

Results at finite bias, and robustness checks

The authors also calculate current-voltage curves. With the proximity term included, the MR is the same for positive and negative voltage, with a peak above 60%. That is twice the electrode polarisation. The MR falls as the coupling to the nonmagnetic electrode increases. The paper reports that the MR stays above the electrode polarisation when dephasing or disorder in the twist angle is added, and in a double-stranded DNA version of the model. With SOC switched off, the MR is exactly zero whatever the other settings. On that basis the authors conclude that weak SOC is essential to CISS and that the proximity effect is what pushes MR past the limit.

What the paper does and does not establish

This is a model calculation and reports no new measurements. The proximity parameters are chosen within a range taken from the literature, not calculated for any specific molecule-electrode contact; the headline results use M = −ε_FM = t1 (100 meV), and the electrode polarisation of 30% is likewise fixed as a typical value rather than taken from a particular device. The authors themselves hope first-principles calculations will check the result. The paper asserts "excellent agreement" with experiment and cites several transport studies, but the main text shows no quantitative comparison with, or fit to, any measured data. Despite this, the conclusion says the results are "expected to end the dispute on the issue of the origin of the CISS".

The approach directly answers a criticism reported in the Sarkar–Dubi perspective. That paper argues that non-interacting models such as Guo–Sun's cannot push polarisation beyond the electrode's. Zhang and colleagues keep the single-particle picture and get around the limit by changing where the effective ferromagnetic boundary lies. They also argue against substrate-based explanations such as the spinterface model, noting experiments that show CISS without any metal substrate. The paper computes MR from conductances that differ under magnetisation reversal, but it never mentions Onsager reciprocity. Read alongside other pages in this dossier, that is a notable omission: the Evers–Kronik review sets out reciprocity as a constraint on two-terminal linear-response MR, and it is central to Fransson's and Tirion–van Wees's proposals. Those connections are this wiki's cross-references, not points the paper itself makes. Likewise, in contrast to the contact-potential shifts on which Tirion and van Wees build their account, the proximity model places the key effect in a magnetised end site instead; the paper does not discuss that alternative.

References

  • Zhang, T.-Y.; Mao, Y.; Liu, P.-Y.; Guo, A.-M.; Sun, Q.-F. "Anomalous Magnetoresistance beyond the Jullière Model for Spin Selectivity in Chiral Molecules." arXiv:2509.05588v2, 3 December 2025.
  • Guo, A.-M.; Sun, Q.-F. "Spin-dependent electron transport in protein-like single-helical molecules." Proc. Natl. Acad. Sci. U.S.A. 111, 11658 (2014).
  • Julliere, M. "Tunneling between ferromagnetic films." Phys. Lett. A 54, 225 (1975).
  • Liu, T.; Weiss, P. S. "Spin polarization in transport studies of chirality-induced spin selectivity." ACS Nano 17, 19502 (2023).
Source notes & attribution
  1. https://arxiv.org/pdf/2509.05588