---
title: 'CuGeO3: Frustrated Spin Chains & Charge Transfer'
url: https://www.emergentmind.com/topics/cugeo3
type: topic
---

# CuGeO3: Frustrated Spin Chains & Charge Transfer

CuGeO3 is a quasi-one-dimensional insulating cuprate built from edge-sharing CuO4 plaquettes, or equivalently strongly distorted CuO6 octahedra, arranged in chains along the crystallographic $c$ axis. It is a canonical spin-Peierls material, undergoing a transition near $T_{\mathrm{SP}} = 14$–$14.2\ \mathrm{K}$ to a dimerized singlet ground state, but it is also a charge-transfer insulator, a benchmark system for O $K$-edge resonant inelastic x-ray scattering (RIXS), a platform for ultrafast nonthermal control of correlated states, and a model material for vibrational strong coupling and magnetoelastic spectroscopy. Across these contexts, CuGeO3 is distinguished by the coexistence of frustrated $S=1/2$ chain magnetism, pronounced electron–phonon coupling, and a lattice geometry in which small atomic displacements strongly reshape superexchange pathways [2104.03557][2507.12409][2507.19412].

## 1. Crystal chemistry and local electronic structure

At room temperature CuGeO3 is orthorhombic, described as Pbmm (No. 51), with lattice constants $a = 4.796\ \text{\AA}$, $b = 8.466\ \text{\AA}$, and $c = 2.940\ \text{\AA}$. The structure contains chains of edge-sharing Cu-centered oxygen plaquettes running along $c$, linked by oxygen into layers parallel to the $b$–$c$ plane and weakly coupled along $a$. In low-temperature descriptions of the spin-Peierls phase, the orthorhombic cell doubles along both $a$ and $c$, corresponding to a dimerization wave vector $Q_{\mathrm{SP}} = (1/2,0,1/2)$ in Pbmm notation, while the true primitive low-temperature cell is monoclinic $P2/m$ with $a = 5.622\ \text{\AA}$, $b = 8.402\ \text{\AA}$, $c = 5.622\ \text{\AA}$, and $\beta = 116.84^\circ$ [2507.12409].

The local coordination is strongly anisotropic. CuGeO3 contains two strongly deformed edge-sharing CuO6 octahedra per unit cell, with Cu–O\(_{\mathrm{apex}}\) $= 2.77\ \text{\AA}$ and Cu–O\(_{\mathrm{inplane}}\) $= 1.94\ \text{\AA}$. The Cu–O–Cu bond angle is reported as $\simeq 98^\circ$ or $\approx 99^\circ$ in the cited studies, i.e. close to the edge-sharing limit near $90^\circ$, and this geometry is central to both its magnetic frustration and its optical selection rules. Oxygen ligands and Ge side groups break the $p$-orbital degeneracy and help render the nearest-neighbor superexchange antiferromagnetic even in this edge-sharing environment [0908.0406][2104.03557].

Electronically, CuGeO3 is a charge-transfer insulator. A cluster-model description writes
$$
H = \sum_l E_d(l)\, d_l^\dagger d_l + \sum_l E_p(l)\, p_l^\dagger p_l + \sum_l T_{pd}(l)\,(d_l^\dagger p_l + p_l^\dagger d_l) + H_U,
$$
with $H_U$ collecting on-site Cu interactions. In optical language, the low-energy local manifold is the Cu $d$–$d$ sector between about $1.5$ and $2\ \mathrm{eV}$, while the charge-transfer edge lies at higher energy, with Urbach-fit edge positions $E_{0,c} \simeq 3.46\ \mathrm{eV}$ for $E \parallel c$ and $E_{0,b} \simeq 3.67\ \mathrm{eV}$ for $E \parallel b$ [0908.0406].

Wavefunction-based embedded-cluster calculations further resolve the local crystal-field spectrum. In the rotated coordinate convention used for edge-sharing cuprates, the ground-state hole is labeled $d_{xy}$; the authors explicitly note that this corresponds to the conventional $b_{1g}$, $d_{x^2-y^2}$-like hole after a $45^\circ$ in-plane rotation. The MRCI energies for CuGeO3 are $1.41\ \mathrm{eV}$ for $d_{x^2-y^2}$, $1.61\ \mathrm{eV}$ for $d_{xz}$, $1.64\ \mathrm{eV}$ for $d_{yz}$, and $1.71\ \mathrm{eV}$ for $d_{z^2}$ above the ground state. The relatively low $d_{z^2}$ excitation reflects the presence of two apical oxygens at $2.76\ \text{\AA}$ [1111.2231].

## 2. Frustrated chain magnetism and the spin-Peierls state

Magnetically, CuGeO3 is a frustrated $S=1/2$ chain system with antiferromagnetic nearest-neighbor $J_1$ and next-nearest-neighbor $J_2$ exchanges. It does not develop long-range magnetic order; instead, its physics is governed by short-range spin correlations in a low-dimensional magnetoelastic environment. Below $T_{\mathrm{SP}} \approx 14\ \mathrm{K}$ it undergoes a spin-Peierls transition in which the chains dimerize and a singlet ground state with a spin gap emerges [2104.03557].

A minimal time-dependent spin-chain form used in ultrafast analysis is
$$
H = \sum_i \Big[ J_1(1 + (-1)^i \delta(t))\, \mathbf{S}_i \cdot \mathbf{S}_{i+1} + J_2\, \mathbf{S}_i \cdot \mathbf{S}_{i+2} \Big] + H_{\mathrm{phonon}} + g \sum_i u_i(t)\, \mathbf{S}_i \cdot \mathbf{S}_{i+1},
$$
which makes explicit that both the dimerization $\delta$ and the local displacement field $u_i$ feed back onto spin correlations [2104.03557].

Recent high-resolution neutron spectroscopy and tensor-network analysis sharpen this picture quantitatively. A global fit of the dynamical structure factor yields
$$
J_1 = 13.79(2)\ \mathrm{meV}, \qquad J_2/J_1 = 0.35, \qquad \delta = 0.04,
$$
with a minimum spin gap $\Delta_{\min} = 2.1\ \mathrm{meV}$ at $q = (0,2n+1,0.5)$ and an empirical triplon-to-spinon crossover near $16\ \mathrm{meV}$. Since $J_2/J_1 = 0.35 > \alpha_c = 0.2411$, CuGeO3 lies in the spontaneously dimerized regime of the $J_1$–$J_2$ chain and is close to the Majumdar–Ghosh point. The resulting spectrum is energy dependent: low energies are dominated by tightly bound, long-lived triplons, while higher energies retain a deconfined two-spinon continuum with a coherent upper-edge enhancement attributed to frustration-suppressed spinon interactions [2507.19412].

This recent picture modifies a common oversimplification of spin-Peierls physics. Robust triplons in CuGeO3 do not require strong external dimerization: the extracted $\delta = 0.04$ is small, yet low-energy triplons are well defined because the underlying frustration already places the chain deep within a spontaneously dimerized regime [2507.19412].

At the same time, exchange extraction remains model sensitive. A DFT+total-energy-mapping analysis in the dimerized phase reproduces the hierarchy of couplings and identifies $J_3$ as the only significant interchain exchange, with $J_2 \approx (0.15$–$0.25)J_1$, but it does not simultaneously capture the observed small gap-to-bandwidth ratio without renormalization. This establishes that CuGeO3 is unambiguously frustrated and dimerized, while the precise effective parametrization depends on whether the target is the full spin spectrum, low-energy triplon dispersion, or static total energies [2507.12409].

## 3. Equilibrium spectroscopic fingerprints

A defining equilibrium signature of CuGeO3 is the O $K$-edge Zhang–Rice singlet (ZRS) exciton in RIXS. High-resolution measurements at the ADRESS beamline with $60\ \mathrm{meV}$ energy resolution identify three principal features in the energy-loss spectrum: a sharp $d$–$d$ excitation near $1.9\ \mathrm{eV}$, a strong excitation centered at $\sim 3.8\ \mathrm{eV}$ assigned to a ZRS exciton on neighboring CuO4 plaquettes, and fluorescence-like intensity above $\sim 4.5\ \mathrm{eV}$. The spectra were recorded at the O $K$-edge pre-peak with $\pi$ polarization, at specular geometry and without momentum transfer along the chain direction [1211.1925].

The ZRS feature is spin selective because O $K$-edge RIXS conserves total spin. In a transparent nearest-neighbor form, its intensity follows the singlet projector
$$
P_{\mathrm{singlet}} = \frac{1}{4} - \langle \mathbf{S}_i \cdot \mathbf{S}_j \rangle,
$$
so that stronger antiferromagnetic nearest-neighbor correlations increase the ZRS spectral weight. Experimentally, the $\sim 3.8\ \mathrm{eV}$ peak indeed gains intensity upon cooling in CuGeO3, opposite to the trend in ferromagnetically correlated Li\(_2\)CuO\(_2\), demonstrating that a high-energy exciton can track short-range magnetic correlations on the meV scale [1211.1925].

The underlying cross section is the Kramers–Heisenberg form
$$
I(\omega) \propto \sum_f \left| \sum_n
\frac{\langle f|T^\dagger|n\rangle \langle n|T|g\rangle}
{E_g + \hbar \omega_{\mathrm{in}} - E_n + i\Gamma}
\right|^2
\delta(E_f - E_g - \hbar \omega),
$$
with the relevant intermediate states built from O $1s \rightarrow 2p$ excitation into the upper Hubbard band. Many-body calculations on Cu\(_3\)O\(_8\), Cu\(_4\)O\(_{10}\), and Cu\(_5\)O\(_{12}\) clusters reproduce both the $\sim 3.8\ \mathrm{eV}$ exciton energy and its temperature dependence, validating the interpretation of the ZRS peak as a local reporter of antiferromagnetic bond correlations [1211.1925].

Optical spectroscopy resolves the same electronic hierarchy from another angle. For light polarized along $c$, the equilibrium absorption shows a charge-transfer edge with an excitonic shoulder at $3.1$–$3.2\ \mathrm{eV}$; for polarization along $b$, only the charge-transfer edge remains. The lower-energy $d$–$d$ manifold appears between $1.5$ and $2.1\ \mathrm{eV}$, with oscillators near $1.6\ \mathrm{eV}$, $1.75\ \mathrm{eV}$, and $1.9\ \mathrm{eV}$ in one parameterization, and with resolved bands at $\approx 1.50$–$1.55\ \mathrm{eV}$, $\approx 1.70\ \mathrm{eV}$, and $\approx 1.88$–$1.90\ \mathrm{eV}$ in another. The absorption edge follows an Urbach form shaped by exciton–phonon scattering, with characteristic optical phonon energies $E_P \simeq 53\ \mathrm{meV}$ for $c$ polarization and $E_P \simeq 56\ \mathrm{meV}$ for $b$ polarization, consistent with a $B_{2g}$ oxygen bond-bending mode at $427\ \mathrm{cm}^{-1}$ [0908.0406][2003.13447].

## 4. Ultrafast nonthermal dynamics of charge, spin, and lattice sectors

Ultrafast optical pump–probe spectroscopy established early that photoexcited CuGeO3 can enter nonthermal states. With $60\ \mathrm{fs}$ pumps at $3.14\ \mathrm{eV}$ or $1.55\ \mathrm{eV}$, absorbed fluence $\Phi_{\mathrm{pump}} \simeq 20\ \mathrm{mJ/cm^2}$, and white-light probing over $\simeq 400$–$800\ \mathrm{nm}$, the charge-transfer edge shows an electronic response with $\tau_1 \simeq 10\ \mathrm{fs}$, a cooling component $\tau_2 \simeq 170\ \mathrm{fs}$, and a slow metastable recovery extending to several hundred ps. For $c$-polarized $3.14\ \mathrm{eV}$ pumping, the Lorentz-oscillator resonance energy shifts from $3.46\ \mathrm{eV}$ to $2.9\ \mathrm{eV}$ within $\sim 100\ \mathrm{fs}$ and the linewidth broadens impulsively to $\simeq 2.0\ \mathrm{eV}$, later settling near $\simeq 1.2\ \mathrm{eV}$. Differential Urbach analysis then requires an unphysical effective local temperature $T \simeq 1630\ \mathrm{K}$ with $\delta T \simeq 150\ \mathrm{K}$, ruling out a purely thermal explanation. The interpretation advanced is an impulsive modification of electronic interactions, notably $T_{pd}$ and $\Delta_{pd}$, producing a nonthermal metastable state in the charge-transfer sector [0908.0406].

That conclusion depends strongly on polarization and excitation channel. For $b$-polarized $3.14\ \mathrm{eV}$ pumping, which suppresses the delocalized exciton channel, or for $c$-polarized $1.55\ \mathrm{eV}$ pumping, the transient absorption can be fit with effective temperatures near $300\ \mathrm{K}$, though still with apparent heating larger than simple energy-balance estimates, attributed to selective heating of strongly coupled optical phonons [0908.0406].

Time-resolved O $K$-edge RIXS later extended this nonequilibrium picture directly into the magnetic sector. At LCLS, CuGeO3 at $20\ \mathrm{K}$ was pumped with $4.7\ \mathrm{eV}$ ($266\ \mathrm{nm}$), $\approx 50\ \mathrm{fs}$ ultraviolet pulses, typically at $F = 37.4\ \mathrm{mJ/cm^2}$, and probed with $\approx 531\ \mathrm{eV}$, $70\ \mathrm{fs}$ x-ray pulses. The combined trRIXS resolution was $\approx 365\ \mathrm{meV}$ and the effective time resolution $\approx 400\ \mathrm{fs}$. In this configuration the ZRS exciton at $\approx 3.8\ \mathrm{eV}$ loss acts as a proxy for the nearest-neighbor antiferromagnetic correlator [2104.03557].

The transient ZRS response is strongly selective. Its intensity drops within $\approx 0.5\ \mathrm{ps}$, exhibits a plateau or peak around $\approx 1$–$1.5\ \mathrm{ps}$, continues to decrease out to $\approx 10\ \mathrm{ps}$, and remains depleted beyond $100\ \mathrm{ps}$ and at least to $\approx 500\ \mathrm{ps}$. At $\Delta t = 100\ \mathrm{ps}$ the suppression saturates for fluence $F \gtrsim 5\ \mathrm{mJ/cm^2}$ and maps onto an effective magnetic quasi-temperature of only $\approx 230\ \mathrm{K}$, substantially below the room-temperature saturation of the equilibrium ZRS intensity. A standard electronic–lattice two-temperature model reproduces the slow evolution only if the absorbed energy density is reduced well below the experimental estimate, indicating out-of-equilibrium decoupling between magnetic and lattice baths on $10$–$100\ \mathrm{ps}$ timescales [2104.03557].

The short-time feature near $\approx 1\ \mathrm{ps}$ implies a characteristic scale of $\approx 1\ \mathrm{THz}$ ($\approx 4\ \mathrm{meV}$). The study identifies a known $0.9\ \mathrm{THz}$ excitation in the spin-Peierls phase, interpreted as a bound pair of magnons with a strong phononic component, as a plausible channel for the coherent modulation. Higher-frequency optic phonons associated with Cu–O–Cu bending at $\approx 3.3$ and $6.8\ \mathrm{THz}$, or Ge motion at $\approx 5$–$18\ \mathrm{THz}$, are too fast to explain the $\approx 1\ \mathrm{ps}$ modulation. This does not identify the atomic motion uniquely, but it localizes the relevant nonequilibrium coupling to a low-energy magnetoelastic channel [2104.03557].

## 5. Coherent vibrational control and cavity polaritonics

CuGeO3 is also a model system for coherent phonon control of localized electronic transitions. Mid-infrared pump–visible probe experiments on a $100\ \mu\mathrm{m}$ crystal at base temperature $8\ \mathrm{K}$ used tunable $4$–$18\ \mu\mathrm{m}$ pump pulses of $260 \pm 40\ \mathrm{fs}$ duration and $\sim 1\ \mathrm{mJ\,cm^{-2}}$ fluence, together with $30 \pm 5\ \mathrm{fs}$ probes spanning $650$–$950\ \mathrm{nm}$. When both pump and probe were polarized along $c$, resonant driving near $\lambda_{\mathrm{pump}} \approx 9\ \mu\mathrm{m}$ produced distinct energy-dependent responses across the $d$–$d$ manifold: below $\approx 1.45\ \mathrm{eV}$ the transient $\Delta T/T$ changes sign around $\tau \approx 300$–$400\ \mathrm{fs}$, yielding a transient transparency, while above $\approx 1.75\ \mathrm{eV}$ the response remains negative [2003.13447].

The mechanism is explicitly coherent rather than thermal. In a minimal phonon-assisted model,
$$
H = \omega b^\dagger b + \epsilon d^\dagger d + M d^\dagger d (b + b^\dagger),
$$
with light–matter interaction
$$
H_{\mathrm{int}} = \mu_0 (b + b^\dagger)(d + d^\dagger)\sum_k (a_k + a_k^\dagger).
$$
Thermal lattice fluctuations increase the total absorption uniformly through
$$
\Gamma_{\mathrm{eq}} = \mu_0^2 \coth(\beta \omega/2),
$$
whereas a displaced thermal state adds an explicitly coherent term,
$$
\Gamma_{\mathrm{tot}}^{\mathrm{displaced}} = \mu_0^2\left[4B_t^2 + \coth(\beta \omega/2)\right].
$$
Experimentally, the driven response is tied to an IR-active $B_{2u}$ mode near $702\ \mathrm{cm}^{-1}$ and to an anharmonically generated Raman $A_g$ oscillation at $\approx 182\ \mathrm{cm}^{-1}$. A single effective phonon energy $\omega \approx 131\ \mathrm{cm}^{-1}$ describes the equilibrium phonon-assisted absorption of the full $d$–$d$ band, and model explorations use $M \approx \omega \approx 16\ \mathrm{meV}$. The extracted coherent displacement scale is tiny, $\Delta B_t \approx 10^{-3}$ or $\Delta x \approx 10^{-4}\ \text{\AA}$, yet sufficient to induce measurable spectral modulations $\Delta\langle E\rangle \approx 60\ \mu\mathrm{eV}$ and $\Delta \sigma \approx 18\ \mu\mathrm{eV}$ [2003.13447].

Under these conditions the spin sector is not the dominant actor. The optical coherent-control study states that no strong signature of spin physics is observed; the phenomena are governed primarily by electron–phonon coupling and phonon-assisted $d$–$d$ absorption [2003.13447].

A different low-energy route uses cavity electrodynamics. CuGeO3 served as the demonstration crystal for a tunable cryogenic Fabry–Perot THz cavity. The relevant material excitation is an IR-active phonon at $\nu_0 = 1.45\ \mathrm{THz}$ for $E \parallel b$, with free-space linewidth $\gamma_f = 76\ \mathrm{GHz}$ at $80\ \mathrm{K}$; no absorption at this photon energy is observed for $E \parallel c$. A $20\ \mu\mathrm{m}$ crystal placed at the antinode of the fundamental cavity mode and tuned to $L = 103\ \mu\mathrm{m}$ yields $\nu_c \approx 1.45\ \mathrm{THz}$ and cavity quality factor $Q = 6.8$ [2112.01560].

At zero detuning the coupled system shows a lower–upper polariton splitting $\Omega_R = 0.32\ \mathrm{THz}$, corresponding to $g \approx 0.16\ \mathrm{THz}$ through $\Omega_R = 2g$, and a time-domain beating period $T_R \approx 3.1\ \mathrm{ps}$. Using the measured $Q$ and phonon linewidth gives a cooperativity
$$
C = \frac{4g^2}{\kappa \gamma} \approx 6.3,
$$
consistent with strong vibrational coupling despite the low cavity $Q$. The polariton anticrossing red-shifts by $\approx 0.04\ \mathrm{THz}$ between $295\ \mathrm{K}$ and $80\ \mathrm{K}$, following the known red-shift of the bare CuGeO3 phonon in the normal phase. These measurements were restricted to $80\ \mathrm{K}$ and $295\ \mathrm{K}$, i.e. above the spin-Peierls transition [2112.01560].

## 6. Phonon spectrum, magnetoelastic channels, and current perspective

The most complete mode-resolved lattice picture in the spin-Peierls phase comes from recent DFT calculations and high-resolution neutron spectroscopy. In the dimerized phase the primitive cell contains four formula units and therefore $60$ zone-center phonons, split equally into Raman-active ($g$) and IR-active ($u$) irreducible representations. The measured and calculated dispersions agree closely across multiple Brillouin zones and at temperatures $5$, $20$, and $110\ \mathrm{K}$ [2507.12409].

The acoustic anisotropy is pronounced: along the chain direction $L$ the acoustic branch reaches $\sim 40\ \mathrm{meV}$, along $K$ it reaches $\sim 25\ \mathrm{meV}$, and along $H$ only $\sim 15\ \mathrm{meV}$. This establishes a stiffness hierarchy $c>b>a$. The optical spectrum falls into distinct energy groups: $1.6$–$3.5\ \mathrm{THz}$ ($6.6$–$14.5\ \mathrm{meV}$), $4.8$–$6.2\ \mathrm{THz}$ ($19.9$–$25.7\ \mathrm{meV}$), $7.6$–$9.0\ \mathrm{THz}$ ($31.5$–$37.3\ \mathrm{meV}$), $10.2$–$12.2\ \mathrm{THz}$ ($42.3$–$50.5\ \mathrm{meV}$), $15.1$–$17.3\ \mathrm{THz}$ ($62.6$–$71.5\ \mathrm{meV}$), and $19.7$–$23.8\ \mathrm{THz}$ ($81.6$–$98.6\ \mathrm{meV}$), with numerous nonmonotonic dispersions and anticrossings [2507.12409].

A central outcome is negative: no phonon softening or dispersion change is resolved across $T_{\mathrm{SP}}$, consistent with the long-standing conclusion that CuGeO3 is not driven into the spin-Peierls state by a soft phonon. The most strongly involved modes instead harden upon cooling, and earlier quasi-elastic scattering was associated with short-range fluctuations above the transition [2104.03557][2507.12409].

Mode-resolved eigenvector analysis identifies the phonons that most effectively modulate exchange geometry. The Raman-active modes $i=8$ ($A_g$, $2.94\ \mathrm{THz}$), $i=20$ ($A_g$, $6.05\ \mathrm{THz}$), $i=34$ ($A_g$, $10.56\ \mathrm{THz}$), and $i=55$ ($A_g$, $21.83\ \mathrm{THz}$) are singled out as “Peierls-active” in the sense that they strongly modulate intrachain Cu–O–Cu angles $\eta_1,\eta_2$ and interchain Cu–O–Ge–O–Cu angles $\delta_1,\delta_2$. Modes $i=23$ ($B_{1g}$, $7.65\ \mathrm{THz}$), $i=29$ ($A_g$, $8.60\ \mathrm{THz}$), and $i=26$ ($B_{2u}$, $8.16\ \mathrm{THz}$) strongly affect the interchain geometry relevant to $J_3$ [2507.12409].

The generic magnetoelastic structure is summarized by
$$
H_{\mathrm{sp}} = \sum_{i,\mu} g_\mu Q_\mu O_i,
$$
with $Q_\mu$ a phonon normal coordinate and $O_i$ a magnetic bond operator such as $\mathbf{S}_{1,i}\cdot \mathbf{S}_{2,i}$. In parallel, the exchange itself is treated geometrically as
$$
J \propto \frac{t(\theta,r)^2}{U}, \qquad
J(\theta,r) \approx J_0 + \frac{\partial J}{\partial \theta}\,\delta\theta + \frac{\partial J}{\partial r}\,\delta r.
$$
The recent phonon work does not compute explicit derivatives $dJ/dQ$, but it ranks the phonon modes by the magnitudes and symmetries of the angle and bond-length distortions they induce [2507.12409].

Despite repeated energy–momentum coincidences between phonons and magnetic excitations, equilibrium neutron spectroscopy does not reveal avoided crossings or linewidth anomalies attributable to spin–phonon hybridization. The interpretation proposed is that static mutual renormalization dominates: phonons are already dressed by magnetism, and triplons by lattice effects, without generating simple single-particle hybridization signatures in equilibrium spectra [2507.12409]. This aligns with the ultrafast trRIXS result that the most relevant nonequilibrium channel is likely a low-energy magnetoelastic excitation near $1\ \mathrm{THz}$ rather than the higher-frequency optic phonons that dominate the broader phonon catalog [2104.03557].

Taken together, these results place CuGeO3 at the intersection of several mature research programs. It is simultaneously a frustrated spin-Peierls chain near the Majumdar–Ghosh regime, a charge-transfer insulator with strongly polarization-selective optical excitations, a benchmark system in which the ZRS exciton provides a direct probe of nearest-neighbor antiferromagnetic correlations, and a material whose phonons can be both coherently driven and strongly coupled to THz cavity photons. The recurring theme is not a single dominant order parameter but a hierarchy of intertwined local processes—Cu–O–Cu bond-angle modulation, Ge-side-group motion, charge-transfer renormalization, and confinement of spinons into triplons—that collectively define the physics of CuGeO3 [1211.1925][2112.01560][2507.19412].

Source: https://www.emergentmind.com/topics/cugeo3