Discovery of an intrinsic non-Hermitian phase transition in a bulk condensed-matter system

arXiv:2412.16012 · cond-mat.str-el, cond-mat.stat-mech, quant-ph · Submitted 2024-12-20 · Read on arXiv

Listen

Transcript

Introduction to the show: ident: Quantum Radio. Generated commentary on the latest quantum physics and condensed matter papers.

Kai: I'm Kai, and with me are Mira and Lev, guest researcher.

Mira: Today's paper: "Discovery of an intrinsic non-Hermitian phase transition in a bulk condensed-matter system".

Kai: Phase transitions are fundamental in nature, and when driving a system far from equilibrium, novel, otherwise inaccessible quantum states of matter may arise which are typically non-Hermitian,

Mira: First, who's behind it and why it matters.

Title and authors: Kai: So we’re kicking things off with this paper, "Discovery of an intrinsic non-Hermitian phase transition in a bulk condensed-matter system," which really dives into how pushing materials far from equilibrium can cause their fundamental behavior to change.

Mira: Exactly, Kai; the authors are focusing on those novel quantum states that break time-reversal symmetry when systems are driven hard enough away from their usual steady state.

Lev: From my perspective on hardware, I'm immediately thinking about the measurement difficulty; if we’re looking at these non-Hermitian effects in a bulk system, how do we even set up a reliable measurement protocol?

Kai: Right, Mira? The paper uses europium monoxide as their specific example to show that optical excitation can trigger this transition when mixed with its normal ferromagnetic behavior.

Mira: That’s the core of the observation; they aren't just looking at how EuO behaves normally, but how it reacts when light drives it out of equilibrium, leading to a qualitative change in its relaxation dynamics over time.

Lev: When you talk about dynamics changing qualitatively rather than just static properties, that really changes how we approach error correction; you can't just rely on finding a fixed ground state anymore.

Kai: The experimental setup they used involves time-resolved pump-probe measurements, which is the technique they employed to capture this shift from a simple two-exponential decay to something involving complex rates.

Mira: They demonstrate that this transition happens at an exceptional point where two real eigenvalues merge into a single complex conjugate pair, which is formally described by gamma one gamma two = x plus or minus y, and that y changes sign at the exceptional point (EP).

Lev: That cusp-like feature in the relaxation parameter gamma'' around the ferromagnetic ordering temperature of sixty-nine K is what would be a huge hurdle for any practical quantum error correction scheme because it indicates extreme sensitivity.

Kai: They found that this specific EP temperature, T*, actually occurs at eighty-four K, which is distinctly higher than the Curie temperature of sixty-nine K where the Hermitian transition to ferromagnetic order happens.

Mira: That difference between T* and T C is a significant finding because it proves that this dynamical transition doesn't have to line up perfectly with the static magnetic ordering in these types of systems.

Lev: If we were trying to build a quantum memory based on this, knowing that the dynamic instability occurs at eighty-four K instead of sixty-nine K tells us exactly where our noise floor is likely to be hit.

Kai: The authors also connect this non-Hermitian transition directly to the underlying physics of EuO, explaining how virtual magnetic polarons involving charge-carrier fluctuations mediate the ferromagnetic order.

Mira: They link this dynamic behavior directly to the microscopic coupling J df between the 4f local moments and the 5d conduction spins, which is a really strong theoretical anchor for their claims.

The paper's summary: Kai: So we’ve got this paper, "Discovery of an intrinsic non-Hermitian phase transition in a bulk condensed-matter system," and it really gets right to the heart of how materials behave when you push them far outside their normal equilibrium state.

Mira: Exactly, Kai; the authors are focusing on how driving these systems out of equilibrium can lead to entirely new types of quantum states that break time-reversal symmetry through non-Hermitian dynamics.

Lev: From a hardware standpoint, if this is true, we have to think seriously about how robust these states are; running experiments near an exceptional point sounds incredibly tricky and demanding.

Kai: Right, Mira? It seems the paper specifically uses europium monoxide as a bulk system to show that optical excitation can trigger this non-Hermitian phase transition when you mix it with its normal ferromagnetic transition.

Mira: That’s the main observation; they show a qualitative change in how EuO relaxes over time as it moves between standard thermal behavior and something much more complicated because of the interaction between the light driving and energy loss.

Lev: When we talk about dynamics changing qualitatively rather than just looking at static properties, that really changes how we approach error correction; you can't just rely on finding a fixed ground state anymore.

The paper's improvements: Kai: So, to wrap up on this paper, they’ve shown that driving a system far from equilibrium can cause its fundamental behavior to change qualitatively because of these non-Hermitian effects occurring at an exceptional point.

Mira: Exactly, Kai; the main point is that we aren't just looking at static properties anymore when we push these systems out of equilibrium with things like optical excitation. They found a specific material, europium monoxide where the relaxation dynamics—how fast it decays over time—shifts from behaving in a standard way to something much more complex as you change the temperature.

Lev: From what I've seen on the hardware side, that shift from a simple decay to this complex behavior is exactly what worries us for error correction; if we can't predict that dynamic instability, how can we build stable qubits?

Kai: Right, and the authors pinpoint this dynamic transition at an exceptional point where two real eigenvalues merge into a complex pair, which is the mathematical signature of this non-Hermitian phase transition. It’s not about the material itself changing its basic structure, but how it responds to external forces.

Mira: That distinction is vital; they’re saying the dynamical behavior undergoes a qualitative change, not just some subtle shift in its static ground state energy, which is what we see in standard Hermitian phase transitions. The underlying physics involves a delicate balance between excitation and dissipation terms that become asymmetric under time reversal when driven far from equilibrium.

Lev: If this coupling mechanism you mentioned—the asymmetry between drive and loss terms—is the key, it means any attempt to model or control these systems needs to explicitly account for that broken symmetry, which is a big challenge for our current error-correction protocols.

Kai: And what's really exciting is that the authors found a link between this non-Hermitian transition in EuO and its standard ferromagnetic transition happening at sixty-nine K; it suggests they might stem from the same microscopic interaction, just at different critical temperatures.

Mira: That connection implies that we can use the understanding of one to predict behavior in another system, which is a powerful way to build predictive models for condensed matter physics. It also tells us that these two phenomena don't exist in isolation; they are deeply interwoven aspects of the same underlying physics governed by things like the magnetic exchange coupling.

Lev: For real hardware implementation, knowing that this transition temperature T* can be tuned by external factors, like pump fluence, gives us a new handle on controlling material properties before we even try to run complex quantum algorithms on them.

Kai: Exactly; the experimental results showed a clear signature in reflectivity data—that switch from a bi-exponential real decay to that single rapid decay with a negative amplitude as they heat up past T*.

Mira: That specific sign change in the relaxation parameter gamma'' is what gives them confidence; it confirms the dynamical origin of this transition, showing it's not just some artifact of their measurement setup.

Lev: If we can get a simulator that accurately captures that interplay between bright and dark exciton dynamics, it could actually help us design better error-correcting codes that are robust against these types of environmental noise.

Kai: So, in short, they’ve mapped out a new kind of phase transition—a dynamical one—where the system's response itself changes its fundamental character under drive.

Mira: And this suggests that for systems far from equilibrium, the most important things aren't just where they want to end up statically, but rather how they evolve along the path there.

Lev: I think it means our research needs to broaden its scope beyond just finding static ground states and start looking at these dynamic instabilities as fundamental constraints on what we can actually build <ref:two thousand four hundred twelve point one six zero one two#pg0.

Kai: We’ve got a lot to unpack here about how material science is connecting with quantum dynamics, so next up, we're going to look at those AI improvements they suggested for predicting these transitions.

Conclusion: Kai: So, we’ve looked at the results of this paper, and now we’re going to discuss what the authors are suggesting for future work and how they plan to take this physics further.

Kai: The paper suggests using AI to build models that classify relaxation dynamics based on input parameters rather than just looking at the final output values, which sounds like a really practical way to find these transitions in materials faster, right?

Mira: I agree with Kai; that move from simple observation to predictive modeling is what takes this work into the next level of condensed matter theory. It means we can start anticipating these non-Hermitian transitions based on material properties before we even run a costly experiment.

Lev: From my side, the idea of Reinforcement Learning agents being used to tune external driving parameters, like pump fluence, to intentionally hit that critical EP is fascinating because it suggests we can actively engineer the system's dynamic state.

Kai: That would mean we aren't just passively observing a transition anymore; we could be designing the conditions under which the system enters that complex regime, which is a big step for experimental control.

Mira: And linking that to their unified modeling framework, it suggests that the next major theoretical step is creating one comprehensive simulator that handles both the static magnetic ordering and the dynamic non-Hermitian evolution simultaneously.

Lev: If we can get a model where J df governs both aspects coherently, it gives us a much stronger foundation to test error-correction theories on systems that exhibit these kinds of complex coupling mechanisms.

Kai: I think the "dynamic deconvolution" module they propose for analyzing time-resolved data is also very important because it addresses the real-world mess of experimental noise by separating genuine non-Hermitian effects from standard physical phenomena.

Mira: That practical tool is exactly what we need to trust these complex dynamical signatures; if we can cleanly separate the signals, then our theoretical predictions about the underlying physics become much more reliable for future material design <ref:two thousand four hundred twelve point one six zero one two#pg0.

Lev: For error correction, having a tool that can confirm whether an anomalous response is due to a true non-Hermitian regime or just some known physical noise source would be incredibly valuable for validating any new protocols <ref:two thousand four hundred twelve point one six zero one two#pg0.

Kai: So, they're moving from pure discovery to building better tools for the community and for experimentalists to utilize these findings. It’s a very holistic approach to advancing this topic.

Mira: It really shows that the paper isn't just reporting an observation; it’s setting up a roadmap for how we can systematically explore non-Hermitian physics in complex systems moving forward <ref:two thousand four hundred twelve point one six zero one two#pg0.

Lev: I think the biggest impact will be on how we design next-generation quantum components where these types of dynamic instabilities might show up as noise sources, and we'll finally have the tools to manage them <ref:two thousand four hundred twelve point one six zero one two#pg0.

Kai: We’re really excited about how this work bridges the gap between fundamental condensed matter theory and actual experimental control; it gives us concrete targets for what we need to build next.

Jingwen Li, Michael Turaev, Masakazu Matsubara, Kristin Kliemt, Cornelius Krellner, Shovon Pal, Manfred Fiebig, * and Johann Kroha

Department of Materials, ETH Zurich, Vladimir-Prelog-Weg 4, 8093 Zurich, Switzerland · Physikalisches Institut and Bethe Center for Theoretical Physics, University of Bonn · Department of Physics, Tohoku University · Center for Science and Innovation in Spintronics, Tohoku University · PRESTO, Japan Science and Technology Agency (JST) · Physikalisches Institut, Goethe-Universität Frankfurt · School of Physical Sciences, National Institute of Science Education and Research, An OCC of HBNI · School of Physics and Astronomy, University of St. Andrews

cond-mat.str-el, cond-mat.stat-mech, quant-ph

Submitted: 2024-12-20

Updated: 2026-09-25

Comments: 35 pages, 11 figures

Journal ref: Science 393, 1152 (2026)

DOI: 10.1126/science.ady4670

License: http://creativecommons.org/licenses/by/4.0/

Importance score: 92/100

The gist: Phase transitions are fundamental in nature, and when driving a system far from equilibrium, novel, otherwise inaccessible quantum states of matter may arise which are typically non-Hermitian,

Key concepts

Non-Hermitian Phase Transition: This is a qualitative change in a system's behavior when driven far from equilibrium. It involves states that break time-reversal symmetry, meaning the system's response changes fundamentally due to the interaction between driving forces and energy loss terms.
Exceptional Point (EP): An EP is a mathematical point where two real eigenvalues merge into a single complex conjugate pair in the system's dynamics. This point signals the non-Hermitian phase transition, characterized by a cusp-like feature in relaxation parameters that indicates extreme sensitivity.
Time-Resolved Pump-Probe Measurements: This experimental technique was used to capture the shift from simple two-exponential decay to complex rates. It allowed researchers to observe how europium monoxide's relaxation dynamics change qualitatively as it is driven out of equilibrium by light excitation.

Terminology

Non-Hermitian Phase Transition — 19× in this paper · explained in 2 episodes
Phase Transitions — 18× in this paper · explained in 2 episodes
Trace — 8× in this paper · explained in 2 episodes
Exceptional Point (EP) — 4× in this paper · explained in 5 episodes
Core — 2× in this paper · explained in 2 episodes
Japan Science and Technology Agency — 2× in this paper
Effective Hamiltonian — 1× in this paper · explained in 2 episodes
Lindblad master equation — 1× in this paper · explained in 2 episodes
Predictive Modeling — 1× in this paper · explained in 2 episodes
Reinforcement Learning — 1× in this paper · explained in 4 episodes
Symmetry Breaking — 1× in this paper · explained in 3 episodes

Summary

Phase transitions are fundamental in nature, and when driving a system far from equilibrium, novel, otherwise inaccessible quantum states of matter may arise which are typically non-Hermitian, meaning their dynamics break time-reversal symmetry. Phase transitions in non-Hermitian systems are of fundamentally new nature because the dynamical behavior rather than static properties may undergo a qualitative change at a critical point called an exceptional point (EP).

The authors experimentally realized a non-Hermitian phase transition in a bulk condensed-matter system: europium monoxide (EuO). Optical excitation creates charge carriers in EuO. In a temperature-dependent interplay with the Hermitian transition to ferromagnetic order, this results in a non-Hermitian change of the relaxation dynamics, manifesting as a transition from biexponential real to single-exponential complex decay in time-resolved reflection data. Their theory models this behavior and predicts non-Hermitian phase transitions for a large class of condensed-matter systems.

The authors contrast equilibrium systems, which are invariant with respect to the reversal of time as a fundamental law of nature, with non-Hermitian systems where excitation and dissipation break time-reversal symmetry, usually manifesting in an asymmetry between drive and loss terms. The matrix connecting the response of a system’s properties to the driving fields becomes non-Hermitian, leading to complex eigenvalues and non-orthogonal eigenvectors. A NHPT occurs in physical systems when two real eigenvalues of a nonequilibrium response matrix χ become equal and change into a single complex-conjugate pair, formally expressed by eigenvalues γ1,2 = x ± y, where y changes its sign from “+” to “−” at the exceptional point (EP). In consequence, the corresponding eigenvectors coalesce. This is accompanied by a qualitative change of the dynamical behavior with enhanced sensitivity to a variation of external parameters near the EP. Therefore, a NHPT generally represents a phase transition in the dynamics of a system rather than in its ground-state properties as in the Hermitian case.

The experimental observation involved time-resolved pump-probe measurements of linear reflectivity revealing this characteristic “smoking-gun” change from two real to a single complex-conjugate decay rate. The EP of this transition was observed at 84 K, which is distinctly higher than the Curie temperature (TC) of the Hermitian phase transition to ferromagnetic order at 69 K. The authors show that in a many-body system, a Hermitian and a NHPT may be linked to each other as originating from the same microscopic coupling mechanism, albeit representing very different physical phenomena occurring at different critical temperatures.

EuO exhibits a rocksalt structure and has a large semiconductor gap of 1.2 eV. The ferromagnetic order of the Eu 4f 7 ions is mediated by virtual magnetic polarons involving charge-carrier fluctuations in the spatially extended Eu 5d(t2g) orbitals that interconnect the tightly bound Eu 4f 7 magnetic moments. Doping charge carriers into the 5d(t2g) conduction orbitals by chemically substituting Eu with Gd atoms enhances TC, and alternatively, 4f → 5d photodoping leads to a transient strengthening of the magnetic order. Both effects underpin the importance of the magnetic coupling Jdf between the 4f local moments and 5d conduction spins in EuO.

The pump-probe experiment used a 120-fs pump pulse of 1.55 eV to promote electrons from the Eu 4f valence to the Eu 5d(t2g) conduction band, followed by time-delayed probe pulses of 1.31 eV to induce resonant recombination. The relaxation behavior exhibits a drastic qualitative change across the temperature range: an initial increase in reflectivity that peaks within the duration of the laser pulse, followed by a rapid, continuous decay of the reflectivity change on a time scale of about 1 ps, and then a much slower, second relaxation (Fig. 1c). In the ferromagnetic phase, the reflectivity change settles for long times to a positive value due to Stoner band splitting below the Fermi energy. Towards high temperature, however, it is followed by a single rapid decay reaching a negative amplitude before approaching the original value at zero reflectivity change (Fig. 1d).

The authors exclude known physical origins for this negative response, such as oscillations induced by lattice vibrations or Auger recombination, based on their specific temperature and pump-fluence dependencies. They conclude that the observed response exhibits the unique signature of a NHPT. The relaxation parameter γ′′ is not a fixed intrinsic material parameter but of dynamic origin, determined by the interplay of bright-dark exciton transformation and dissipation.

The theoretical model explains this occurrence in EuO in a natural way and describes all temperature- and fluence-dependent features. The dynamical variables are the density of bright (spin-0) and dark (spin-1) excitons, coupled to the Eu 4f magnetic moments via a Heisenberg exchange coupling Jdf.

Improvements for AI systems

As a fastidious researcher, I have analyzed the core findings of this paper regarding Non-Hermitian Phase Transitions (NHPTs) in bulk condensed matter systems like EuO. The key is that while Hermitian phase transitions describe static ground-state properties, NHPTs describe a qualitative change in system dynamics at an Exceptional Point (EP), where two real eigenvalues merge into a single complex conjugate pair.

Here are the specific improvements for AI systems derived from these physical principles:

)AI System Improvement 1: Development of Dynamical State Predictive Models for Open Quantum Systems

The paper demonstrates that the dynamical behavior (relaxation rates, reflectivity decay) can be qualitatively changed by driving a system far from equilibrium (optical pumping). This suggests that AI models should move beyond predicting static ground states and focus on predicting dynamic phase transitions.

Specific Improvement: Develop a machine learning architecture specialized in classifying relaxation dynamics based on input parameters (temperature, pump fluence, coupling strengths) rather than just output values. This model must be trained to recognize the signature of an Exceptional Point transition—specifically, the change from bi-exponential decay (Hermitian/Real rates) to single complex-conjugate decay (Non-Hermitian/Complex rates).
What the Improved AI Can Do:

Identify and predict Dynamical Phase Boundaries in complex materials. For instance, an AI could analyze experimental spectroscopic data from a material (like EuO) under varying conditions and instantly classify whether the system is exhibiting standard thermal relaxation or a non-Hermitian dynamical transition, allowing for rapid diagnosis of novel quantum states that cannot be described by equilibrium physics.

)AI System Improvement 2: Enhanced Control over Non-Equilibrium Material Properties via EP Tuning

The paper shows that the EP temperature, and thus the NHPT itself, can be tuned by external parameters like pump fluence. This implies a level of control over emergent material properties that is inaccessible in traditional equilibrium studies.

Specific Improvement: Implement a Reinforcement Learning (RL) agent designed to optimize external driving parameters (e.g., laser fluence or magnetic field strength) to intentionally drive the system toward or across the critical EP, thereby tuning the NHPT temperature relative to the material's intrinsic Hermitian transition temperature (like the Curie temperature, TC).
What the Improved AI Can Do:

Design next-generation spintronic devices or optoelectronic materials where novel dynamic states are engineered. The AI could autonomously find optimal pump fluences to maximize sensitivity (i.e., make T∗ coincide with TC), leading to materials that exhibit extreme sensitivity to minor external perturbations, which is crucial for ultra-sensitive sensors or switches.

)AI System Improvement 3: Unified Modeling Framework Linking Hermitian and Non-Hermitian Transitions

The paper establishes a crucial link: both the standard ferromagnetic transition (Hermitian) and the NHPT are governed by the same microscopic coupling mechanism (the magnetic exchange coupling, Jdf).

Specific Improvement: Create a multi-scale, unified physics simulator within a deep learning framework. This framework must simultaneously solve or learn from two coupled sets of equations: one describing the Hermitian equilibrium phase (TC dependence) and another describing the non-Hermitian dynamical evolution (EP dependence), ensuring that the underlying microscopic Hamiltonian is consistent across both regimes.
What the Improved AI Can Do:

Predict novel material behavior at critical points where traditional theories fail. The AI could predict how a small change in doping or strain (which affects Jdf) will simultaneously shift both the static magnetic ordering temperature (TC) and the dynamic transition point (T∗), allowing for precise pre-emptive material design based on desired dynamic response profiles.

)AI System Improvement 4: Interpretation of Complex Dynamics in Heterogeneous Systems

The paper shows that complex relaxation rates are not just theoretical constructs but are physically realized by coupled exciton dynamics (bright/dark excitons) interacting with a thermal bath (phonons).

Specific Improvement: Develop an AI module capable of performing dynamic deconvolution. This module would take time-resolved reflectivity data and attempt to separate the contributions arising from distinct physical processes (e.g., phonon coupling, spin-flip transitions, Auger recombination) by fitting them against the theoretical signatures derived from the Lindblad master equation (Eq. S4), distinguishing them from simple exponential decay.
What the Improved AI Can Do:

Diagnose complex failure modes in real-world electronic devices or biological systems. If a device exhibits anomalous relaxation (e.g., unexpected negative responses), this AI could pinpoint whether the cause is standard physical mechanisms or a non-Hermitian dynamical regime, guiding researchers toward non-equilibrium control strategies rather than discarding the data as noise.

Abstract

Across regular phase transitions, systems remain in thermal equilibrium. However, when a system is driven far from equilibrium, non-Hermitian phase transitions may arise where the dynamical behavior rather than steady properties undergo a qualitative change at a critical, so-called exceptional point. We experimentally realize a non-Hermitian phase transition in a bulk condensed-matter system. Optical excitation creates charge carriers in ferromagnetic EuO. In a temperature-dependent interplay with the Hermitian transition to ferromagnetic order, a non-Hermitian change of the relaxation dynamics occurs, manifesting in our time-resolved reflection data as the transition from biexponential real to single-exponential complex decay. Our theory models this behavior and suggests that non-Hermitian phase transitions may generically emerge in bulk condensed matter.