Abstract
Nonequilibrium phase transitions play a pivotal role in broad physical contexts, from condensed matter to cosmology. Tracking the formation of nonequilibrium phases in condensed matter requires a resolution of the longrange cooperativity on ultrashort timescales. Here, we study the spontaneous transformation of a chargedensity wave in CeTe_{3} from a stripe order into a bidirectional state inaccessible thermodynamically but is induced by intense laser pulses. With ≈100 fs resolution coherent electron diffraction, we capture the entire course of this transformation and show selforganization that defines a nonthermal critical point, unveiling the nonequilibrium energy landscape. We discuss the generation of instabilities by a swift interaction quench that changes the system symmetry preference, and the phase ordering dynamics orchestrated over a nonadiabatic timescale to allow new order parameter fluctuations to gain longrange correlations. Remarkably, the subsequent thermalization locks the remnants of the transient order into longerlived topological defects for more than 2 ns.
Introduction
The remarkable feature associated with spontaneous symmetry breaking (SSB) is emergent scaleinvariant dynamics in approaching a thermal critical point^{1}. There have been strong incentives to understand how this selforganization may proceed out of thermal equilibrium^{1,2}. Studying the nonequilibrium phase transition introduced via a swift change of the system parameters, also called a quench, is one of the most active areas in nonequilibrium physics, impacting diverse fields from condensed matter^{3,4}, quantum gases^{2,5,6}, to cosmology^{4,7}. It is widely believed that, after an interaction quench, isolated systems generically approach a thermal state^{2}; however, before thermalization can occur a transient nonthermal stationary state may emerge with properties unlike their equilibrium counterparts^{8}. Such investigations have been carried out using ultracold atoms^{2,5}. Meanwhile, recently ultrafast pumpprobe studies made surprising discoveries of lightinduced superconductivity^{9} and insulatormetal transitions in hidden chargedensity wave (CDW) states^{10,11,12}, hinting undisclosed routes towards new brokensymmetry ground states through light excitation far from equilibrium.
In this article, we demonstrate a prototypical case of nonequilibrium SSB into a hidden ground state through a laserassisted nonthermal quench of the system parameters that define competing orders. The system is CeTe_{3}^{13} in which the naturally occurring SSB ground state is the stripephase cCDW order^{14,15,16}; see Fig. 1a. Upon femtosecond (fs) nearinfrared pulse excitation the preexisting order is transiently suppressed; however, the system develops a new preference of SSB from the stripe order to a bidirectional state beyond a welldefined critical excitation. Furthermore, this ultrafast phase transition displays key characteristics of nonequilibrium SSB processes^{4,5,17}: the spontaneous emergence of softmode instabilities of the new CDW order, followed by a slow onset of the phase ordering stage to develop the longrange correlations. Finally, after the system relaxation back to the thermal ground state the remnants of transient orders survive as longlived topological defects^{4,6,7} for more than 2 ns. The dynamics observed here open an intriguing perspective of controlling phase transitions in quantum materials far from equilibrium.
The CeTe_{3} studied here belongs to the rareearth tritelluride (RTe_{3}) family^{13,14,16} where the incommensurate CDW develops inside the double Te square lattice sheets, which are isolated by the buckled insulating CeTe layer (Fig. 1a). The 2Dlayered RTe_{3} compounds are ideal systems for studying SSB because in the square net of Te layers two competing stripe density waves (along the a or caxis) can appear^{14,15,16,18}. The formation of different types of CDWs is subject to the nesting in the electronic structure^{16,18} but the strong momentumdependent electronphonon coupling (EPC) is believed to be essential to form singlewavevector CDW at dimensionality higher than one^{19,20,21,22}. The shape of the 2D metallic Fermi surface (FS) depends on the relative coupling strengths between neighboring 5p_{x} and 5p_{z} orbitals (t_{⊥} and t_{//} in Fig. 1a), which play a key role in these interactions^{16,18,23,24}. Nonetheless, the tendency for the RTe_{3} system to form cCDW over aCDW as the preferred brokensymmetry ground state is facilitated by a subtle bilayer coupling that weakly breaks the C4 symmetry in the Te lattice^{18}. Especially in CeTe_{3}, a large cCDW gap removes a significant amount of the potential aCDW spectral weight^{16,25}, which completely excludes subsequent formation of aCDW in its equilibrium phase diagram (Fig. 1a)^{15}. Only in the heavier members of RTe_{3} (see Fig. 1a) where the smaller lattice constant weakens the cCDW can the aCDW develop at a second lower critical temperature^{14,15,16}.
Our CeTe_{3} sample is a single crystal prepared by tape exfoliation to a thickness of ≈25 nm^{26} with a transverse size ≈30 μm; see “Methods” section. The sample film thickness is matched to the penetration depth of the pump laser pulse^{27,28}, establishing a nearly homogeneous excitation profile facilitated by the thinfilm interference effect (Supplementary Fig. 1)^{29}. The uniformity and thickness of the exfoliated samples are checked using a TEM, where we also conduct the temperaturedependent studies; see Supplementary Fig. 2 for the results. The sample is gently placed on a thin TEM grid in the vacuum specimen chamber where a relatively large pump laser beam (500 μm) illuminates the sample uniformly. Due to the good thermal isolation in this sample setting, the experiments show no visible dissipation of the pump energy absorbed into the materials over the observation window (2 ns). To optimize the probe beam brightness, we deliver intense electron pulses (~10^{6} e/pulse) at 100 keV generated from a silver photocathode just below the virtual cathode limit^{30,31}. The phase space of the spacechargedominated pulse is actively manipulated through placing a radiofrequency cavity in the optical column acting as the longitudinal lens^{30,32} (Fig. 1b) to reach ≈100 fs pulse duration. This is accomplished under a condition that its transverse phase space, controlled through a series of magnetic lenses, produces a highly collimated beam^{33} with a beam coherence length up to 40 nm^{30}. A highquality pattern produced by this fs coherent scattering setup is shown in Fig. 1c, where the incommensurate CDW state evidenced by the satellite peaks at Q_{c} ≈ 2/7c*^{13,14} stands sharply apart from the lattice Bragg peaks (G) in Miller indices. Due to the large intensity difference between the two (~100:1), typically the lattice peaks are intentionally saturated to provide sufficient dynamical range for investigating the nonequilibrium CDW dynamics.
Dynamics of CDW phase transition in RTe_{3} have been investigated using the ultrafast pumpprobe techniques^{24,25,28,34,35,36,37,38,39,40}. The pump fs laser pulses couple to the CDW materials with a broad excitation spectrum across the gap, leading to rapid carrier heating. Hence, the natural setting of these experiments has been to investigate the melting of the preexisting cCDW in this system. The time and angleresolved photoemission spectroscopy (trARPES) found that within ≈250 fs upon applying laser pulses, the spectroscopic gap at the Brillouin zone momentum c* − Q_{c} is smeared^{24,25,37,38,39}. The gap dynamics is coupled to the amplitude suppression of the density wave at the momentum wavevector Q_{c} as resolved with the fs electron^{36,41} and xray scattering^{40,42}. Timedomain spectroscopies gain accesses to the lower frequency collective modes, which are coherently excited upon applying a weaker laser pulse^{20,28,40,43,44}. The BCStype behavior^{14,23,45} has been identified by monitoring the collective mode softening in temperature (T) and the laserfluence (F) dependent investigations near T_{c,1}^{20,43,44}. Focused on the dynamics of the recovery, a study joining ultrafast electron diffraction (UED), optical reflectivity, and trARPES to investigate LaTe_{3} reported a much faster recovery of gap size than in recuperating the diffraction correlation length. The authors attributed the phenomenon to the presence of topological defects induced in the processes^{38}. Interestingly, a more recent MeVUED investigation following the work found the emergence of a lightinduced aCDW state and interpreted the phenomenon as a transient order developed by nucleating around the topological defects^{41}. Whereas the various pumpprobe investigations have illustrated the ultrafast dynamics of melting and the possibility of defectinduced state in RTe_{3}, the symmetrybreaking aspect that underpins the formation of the hidden order and the associated nonequilibrium phase ordering dynamics have not been clearly elaborated.
Here, with a high momentum resolution fs scattering probe, our work has addressed the nonequilibrium structural formation of a hidden aCDW in CeTe_{3}. The slow onset of the new phase and the coupled soft modes directly involved in the ordering processes lead us to conclude that a nonequilibrium SSB is responsible. The new understanding also bears on a recent clear demonstration of hidden coinstabilities occurring along both symmetrybreaking axes approaching T_{c,1} in a similar DyTe_{3} system, probed by the inelastic xray scattering (IXS)^{23}. This result suggests that the underlying energy landscape supports two stripephase CDW ground states – even though aCDW remains unsubstantiated throughout the phase transition. The reversal of the symmetrybreaking preference by light excitations evidenced here hence opens a window of opportunities for controlled investigation of the competitively ordered systems.
A phenomenological Landau–Ginzburg (L–G) theory is adopted to capture the essences of nonequilibrium phase transitions. We describe the two CDW states in terms of the order parameter η_{l} = η_{l}e^{iϕl}, where l = a or c, and (η_{l}) and ϕ_{l} represent the amplitude and phase fields. The L–G free energy modeled on the amplitude field is written as^{18,46}
with α_{l} and A_{l} > 0 for supporting the two symmetrybreaking states at T^{(l)} < T_{c}. The inclusion of the repulsive biquadratic coupling term where Ã > 0 makes the eventual symmetry breaking a competition between the two. In this regard, an anisotropy in α_{l}/A_{l} makes cCDW instead of aCDW the global ground state. The competing strengths of the order parameters can be evaluated at the stationary conditions, \(\frac{{\partial f}}{{\partial \left {\eta _a} \right}} \,=\, \frac{{\partial f}}{{\partial \left {\eta _c} \right}} \,=\, 0\), which define the coordinates of the free energy minimum where \(\left {\eta _a} \right \,=\, \sqrt {\frac{{\alpha_a\left( {T_c \,\, T} \right) \,\, \widetilde A\left {\eta _c} \right^2}}{{A_a}}}\) . From this, it is established that a stationary state with a nonzero η_{a} occurs only if η_{c} is suppressed beyond a critical threshold \(\left {\eta _{c,th}} \right \,=\, \sqrt {\frac{{\alpha_a\left( {T_c \,\, T} \right)}}{{\tilde A}}}\), which can occur under the nonequilibrium condition.
In Fig. 2a this nonequilibrium route of phase transition is illustrated. The key step, modeled after the scenario of F = 1.85 mJ/cm^{2}, is the sudden unfolding of the L–G landscape from the uniaxial to the bidirectional one, driven by an interaction quench that shifts the preexisting order parameter η_{a} from 1 to ≈\(\sqrt {0.17}\) (see stage II). Here, governing the aCDW formation following this change is the projected free energy, f(η_{a}), as depicted in the inset panel. The f(η_{a}) unfolds from an uphill into a double well under the presupposition that the subsystem effective temperature, T^{(a)}, remains near that of the ambient, in contrast to the excited cCDW state. This nonergodicity presupposition is justified by the high quench speed and the involvement of slow modes mediating the phase transitions. The nonequilibrium scenario makes the bidirectional ordering possible as opposed to the equilibrium route (also shown on the right of Fig. 2a). For details of parameterizing the L–G equation to describe the thermal and nonthermal phase transitions, see “Methods” section.
One chief goal in the studies of nonequilibrium SSB dynamics is to see how the rapidly quenched system adopts the longrange correlations from the deeply unstable order parameter fields^{3,4,5,7,17}, often referred to as coarsening^{1,3}. While at the late stage of coarsening the system is expected to follow universal laws^{1,3}, at the shortest time the evolution is dominated by the spinodal (unfolding) instabilities^{17,47}. In CDW phase transitions, the instabilities are mediated by soft modes, which occur in both the amplitude and phase fields^{23,48,49,50}, and the order parameter dynamics are governed by a 2D Mexican hat free energy (see Fig. 2b) in which the phase fluctuation acts as a Goldstone mode with essentially a zero energy cost^{48,49,50,51}. It has been theorized that to excite a collective mode in the CDW fields at the momentum wavevector k ≡ q − Q_{a}, two lattice soft phonons at q = k ± Q_{a} must be joined coherently^{49}.
The static singlewavevector lattice distortion wave (LDW), u_{η}(Q_{l} r), is a distinct physical manifestation of the CDW state which can be probed directly with the scattering techniques. The conversion between the order parameters (with η_{c} = 1, t < 0) modeled by the L–G theory and the LDW can be established with coefficient \(A_{\eta u} \,=\, \frac{{\left {u_\eta } \right}}{{\left \eta \right}}\). The nonequilibrium LDW with field variations is described by \({\mathbf{u}}_\eta \left( {{\mathbf{r}},t} \right) = u_{0,\eta }{\hat{\mathbf{e}}}_\eta \left( {1 \,+\, \delta u_\eta \left( {{\mathbf{r}},t} \right)} \right) \sin\left[ {{\mathbf{Q}} \cdot {\mathbf{r}} \,+\, \delta \phi \left( {{\mathbf{r}},t} \right)} \right]\), where u_{0,η} and \({\hat{\mathbf{e}}}_\eta\) are the amplitude and polarization vector. These field variations, which can be expanded in series of fluctuation waves of wavevector k, manifest in the diffuse scattering centered at Q_{1}; see Supplementary Note 1. The sum rules allow a simple expression \(S_{Q_l}\left( {{\mathbf{k}},t} \right) \approx  {u_{0,\eta _l}(t)} ^2f\left( {{\mathbf{k}},t} \right)\) in which f(k; t) is a normalized function containing the spatial and ensemble averaging of a system much greater than the CDW correlation length ξ; see Supplementary Note 2. In this expression, the structure factor S_{Ql} (k;t) simply is the Fourier counterpart of the autocorrelation function, \(S_{\eta _l}\left( {{\mathbf{r}},t} \right) \approx {\mathbf{u}}_{\eta _l}\left( {{\mathbf{r}}^\prime,t} \right){\mathbf{u}}_{\eta _l}\left( {{\mathbf{r}}^{\prime\prime},t} \right)\), with r = r′ − r″ and the bracket denoting the averaging^{1,3,17,48}. The coarsening process, during which fluctuation waves condensed into the LDW, is evidenced in the dynamical structure factor that sharpens over time, as illustrated in Fig. 2b. It is easy to see the amplitude of LDW is given by integrating structure factor over the momenta, \(m_{Q_l} = {\int} {S_{Q_l}\left( {{\mathbf{k}},t} \right)d{\mathbf{k}} \approx  {u_{0,\eta _l}(t)} ^2}\). Meanwhile, to probe the fluctuation effects, it is key to resolving \(S_{Q_l}\left( {{\mathbf{k}},t} \right)\), typically a Lorentzian with \(f\left( {\mathbf{k}} \right)\sim \frac{1}{{1 + \xi ^2k^2}}\) (assuming the correlation decays exponentially over the length ξ), in the momentum space.
Results
CDW evolution after pump/quench
In Fig. 3, we show the dynamics of nonequilibrium SSB, with the scattering to probe the main event sequences phenomenologically captured by the L–G model. The fluctuation effects associated with the symmetry breaking, which are beyond the meanfield description, are addressed by simultaneously following the scattering structure factors of the main and CDW superlattice(s). We first focus on the longrange ordering manifested in the LDW. The corresponding structure factors of the bidirectional state, S_{Qc}(k, t) and S_{Qa}(k, t), are extracted from the intensity plots over the two satellites across the ordering axes c* and a*, shown in Fig. 3a. Here, the structure factors are obtained in the region near the G_{401} (inset pattern); however, we also confirm the results obtained in other regions are similar. Concerning the immediate impact of the interaction quench, we examine S_{Qc}(k, t) at Q_{c} = 0.28c*, where the scattering in the longwavelength limit, i.e., at k ≈ 0, best captures the impacts on the preexisting longrange state by laser excitations (red curve in Fig. 3a). At F = 1.85 mJ/cm^{2}, the ordering is suppressed nearly instantaneously, reflected in the halving of the amplitude within the first 250 fs. To study the emergence of aCDW, we turn to the orthogonal sector where new satellites are spotted at ±0.30a* (Q_{a}). The scattering intensity at k ≈ 0 shows a visible delay, by ≈300 fs relative to the suppression of cCDW. Interestingly, the sampling at a finite spatial frequency, at k = 0.07c*, under the extended envelope of S_{Qc}(k, t) shows a different response, apparently mimicking the dynamics at Q_{a}.
Stabilization of subdominant order
The results here can be mapped into the L–G order parameter fields. Following the preceding introduction, we calculate the normalized integrated intensity, \(\hat m_{Q_l}\left( t \right) \equiv \frac{{m_{Q_l}\left( t \right)}}{{m_{Q_c}\left( {t < 0} \right)}} \,=\, \left {\eta _l} \right^2\), which gives straightforward mapping to the order parameter fields, and with the coefficient A_{ηu} = u_{0,ηc} = 0.15 Å, set by the initial LDW amplitude^{52}, the LDW amplitudes are also obtained. In Fig. 3b, we plot two trajectories, \(\hat m_{Q_c}\left( t \right)\) and \(\hat m_{Q_a}\left( t \right)\), in response to the transformed free energy. The delayed onset observed in the diffuse scattering retrieved at Q_{a} can be explained with the L–G picture of a serial process for forming the bidirectional order. Following the initial quench, the slow increase of scattering at Q_{a} and Q_{c} + 0.07c* (Fig. 3a) is a signature of coarsening of the system in a newly created but softened 2D landscape (Stage II in Fig. 2a). The system assumes the minimum of the new energy basin, identified by \(\hat m_{Q_c}\left( t \right)\) and \(\hat m_{Q_a}\left( t \right)\) in Fig. 3b, and becomes metastable at ≈1 ps. The field fluctuations that limit the correlation lengths ξ_{c} (t) and ξ_{a} (t) are extracted from fitting structure factors \(S_{Q_l}\); see Supplementary Note 2. The correlation lengths, ξ_{c} (t) and ξ_{a} (t), are presented by the colored regions along with \(\hat m_{Q_c}\left( t \right)\) and \(\hat m_{Q_a}\left( t \right)\) in Fig. 3b. The two order parameters straddle for ≈1.5 ps until the relaxation sets in and that drives the system back to the thermal state of cCDW (stage III). The timescale for this thermal recovery (τ_{e}; see Fig. 3b) is ≈4.5 ps, after which the order parameter does not evolve for a very long time.
Nonequilibrium critical thresholds
We find the three main stages of evolution, namely the metastable state, the recovery, and the long plateau, that are also identified at other fluences. We may assume the system reaches thermal equilibrium at the late plateau stage. Figure 3c illustrates the common dynamics from two selected fluences; for the dynamics over a broad range of fluences, see Supplementary Fig. 3. We focus on the intensities \(\hat m_l^{meta}\) and \(\hat m_{c,\infty }\), obtained at the metastable and thermal stages. The Fdependent intensities are depicted in Fig. 3d. The results show a threshold behavior for introducing aCDW at F_{c,a} = 0.59 mJ/cm^{2}. Furthermore, two different critical thresholds for melting, F_{c,c} ≈ 2.0 mJ/cm^{2} and F_{c,T} ≈ 6.3 mJ/cm^{2}, are obtained by extrapolating \(\hat m_{Q_c}\) to zero at the metastable and thermal stages. Here, given no dissipation is observed at the late stationary state, the energy density deposited by the laser pulse is well calibrated by equating the absorbed energy at the thermal threshold (F_{c,T}) to the enthalpy increase^{53} to reach T_{c,1} = 540 K (Supplementary Fig. 1), calculated to be E_{c,T} = 2.0 eV nm^{−3} (“Methods” section). Interestingly, the nonthermal critical energy E_{c,c} is only ≈1/3 of E_{c,T}, signaling the enthalpy increase is localized to only phonons connected to cCDW; namely, the shorttime active modes accounts ~1/3 of the heat capacity according to the difference between E_{c,c} and E_{c,T}. This nonergodicity is enabled by high efficiency of the coupling to the soft phonons/amplitude modes (AM) upon laser excitations, well documented in a number of pumpprobe studies^{20,28,40,43,44}. This leaves the rest lattice modes only weakly excited at the metastable stage.
Nonequilibrium critical dynamics and structure refinement
In Fig. 4a, we outline how the nonequilibrium phase transformation driven beyond the nonthermal critical threshold is expected to proceed. The scenario is given for a deep quench, F >> F_{c,a}, where f(η_{a}) unfolds on a timescale over which the critical state dynamics cannot follow adiabatically. Instead, field instabilities would be introduced and the shortrange orders appear randomly across the entire sample and coarsen indpendently^{3,4,17}. Statistically, the phase dynamics, described by the autocorrelation function, are resolved through Fourier inversion from the structure factors (Supplementary Note 2). At this stage of evolution, the diffuse scattering of the collective modes and the structure factor due to the shortrange ordering are indistinguishable^{3}.
An iterative fitting procedure is established, aiming for structure refinement based on a realspace model of LDWs to capture the nonequilibrium evolutions. Based on LDWs, the system autocorrelation function, S_{tot}(r, t), is calculated to 75 nm, which in turn reproduces the structure factors via a Fast Fourier Transform. The initial input parameters for LDWs, including \(Q_l(t)\), \(u_{0,\eta _l}(t)\), and \(\xi _l\left( t \right)\), are deduced from fitting the experimental S_{Qc} and S_{Qa}. To deal with the dual order parameter system, with the fluctuations likely subject to the nonlinear effects from the softened modes, a second Voigt function is included to account for the different dynamics shown in the diffuse scattering of S_{Qc} (Fig. 3a). Comparisons between the experimental S_{Qc} (in red dots) and S_{Qa} (in green dots) and the theoretical reconstructions (in solid lines) are shown in Fig. 4b. The theoretical S_{Qc} can reproduce the fine features, such as the highorder peaks at c* − 2Q_{c}, and the diffuse scattering connected with the ordering; for details of the refinement protocol, see Supplementary Note 2. The structure factors derived here are plotted in the semilog scale in order to adequately depict the evolution of the diffuse scattering. The fitted scattering weight of each CDW state is shown in the colorcoded pie chart.
The refinement successfully identifies the second cCDW component, referred hereafter as the c^{+}component, by its broader scattering feature. The new component (plotted in purple in Fig. 4b) is shown to have a delayed onset and evidently is responsible for the dynamics extracted at k = 0.07c* in Fig. 3a. Another notable feature is the higher ordering wavevector of the c^{+}component, which is distinguishably greater than the Q_{c} of the initial state. In Fig. 4d, we plot the change in the mean position \(\bar Q_c\left( t \right)\) for the entire course of the evolution. The \(\bar Q_c\left( t \right)\) shows similar dynamical features found in the intensity evolution of aCDW. These results, taken together, support that the c^{+}component is part of the 2Dordered state, predicted by the L–G theory. We note that the sizeable shift in ordering vectors, from the ≈0.28c* to ≈0.3–0.32c*, has in fact been identified by ARPES in the heavier RTe_{3} members upon their transitions into the bidirectional states^{16,54}. The empirical results show the aCDW would not appear unless the cCDW is weakened. This phenomenon has been attributed to a competition for the spectral weight in the nested FS; the shift in nesting vector is the result of a reduction of FS curvatures^{16,54}. Here, the reordering of the cCDW state is triggered by the introduction of the aCDW ordering. Dynamically, the reorganization unfolds in a highly softened landscape in which the modes are anharmonic. The field fluctuations along η_{a} can thus effectively couple to the instabilities in the cCDW field, resulting in a cocoarsening. Indeed, during the phase ordering of the aCDW state, the c^{+} structure factor also shows sharpening, which means the established bidirectional region is expanding over time. This transformation, presented in the autocorrelation function of the system on the length scale of 27 nm, is depicted in Fig. 4c.
Identification of softened phonons
Precursors to the symmetry breaking are nonequilibrium lattice excitations in response to the laser quench. The strongest responses occur in modes wellconnected to symmetry breaking^{22,23}; in particular, in the continuous phase transition^{55} as is the case here, the soft phonons are directly responsible for the critical instabilities observed in the dynamical phase transitions^{20,43,44}. Unlike the longrange parameters that have been retrieved from analyzing the satellite peaks, the dynamics of incoherent phonons (soft modes included) are encoded in the intensity decay of lattice Bragg peaks, namely the Debye–Waller factor (DWF)^{56}, \(e^{  2M_{hkl}\left( t \right)}\). Here, the \(2M_{hkl}\), which is the summed lattice meansquared (ms) vibrations projected along G_{hkl}^{56,57}, could inform the lattice kinetic energies—equipartitioned among all the modes at equilibrium. However, immediately upon laser excitations, the partitioning of the kinetic energies is necessarily nonergodic. The lattice responses here are analyzed via the change of DWF (Supplementary Note 1), which is expected to show highly momentum (G_{hkl})dependent features due to the symmetry breaking (and recovery) and the nonergodicity in the phonon system.
In RTe_{3}, the instabilities from soften phonons approaching the thermal phase transition have been probed by the IXS technique^{23}. The relevant phonon dispersion curves^{23} are schematically reproduced in Fig. 5a. In the nonequilibrium phase transition, the transformations of the dispersion curves are sudden with the most changes occurring in the softmode regions (shaded area)^{58,59}. Such dynamics can be analyzed through calculating the ms vibrational profile change \(\Delta \bar u_{hkl}^2\left( t \right) \,=\, \Delta 2M_{hkl}\left( t \right)/\left {{\mathbf{G}}_{hkl}} \right^2\). We note extracting Δ2M_{hkl} is an integral part of the systematic scattering profile analysis, which is detailed in Supplementary Note 1. The scheme builds on the scattering weight transfer between the lattice (S_{G}) and CDW (S_{Q}) structure factors and the sum rules to extract the LDW and the DWF dynamics, as outlined in Fig. 5b. The orthogonality of the two CDW systems makes the analyses relatively straightforward. It is worth noting that to form a singlewavevector CDW, the pairing of soft modes into the collective modes and their coalescence into the static LDW shall occur near the minimum of the dispersion curves^{45,58}. The dynamical DWF analysis here hence provides us with the rare opportunity to probe the nonequilibrium and momentumselective events at the earliest stage of symmetrybreaking process, before the order parameter is even substantiated.
Taking advantage of the highenergy electron probe, which enables a large Ewald sphere to cover a broad range of momentum range, the entire momentum (G_{hkl})dependence can be mapped without rotating the crystals. The clean way to retrieve the vibrational angular profile rests with analyzing the Bragg peaks over a circular reciprocal lattice zone at the same radius, G_{hkl} (see Fig. 5b). When the symmetry breaking is instigated by the soft modes of a specific polarization and over a limited wavevector range, the angular profile shall exhibit a cos^{2}θ distribution (also cos^{2n}θ where n > 1 with the involvement of nonlinear interactions) with θ the angle between G_{hkl} and \(\widehat {\mathbf{e}}_q\).
Utilizing Bragg peaks at G_{hkl} ~ 6 r.l.u. as shown in Fig. 5b, we have taken the snapshots of the vibrational excitation profile, \(\Delta \bar u^2\left( t \right)\), plotted in the polar coordinates to evaluate the angular dependence. The excitation profiles bear close resemblances to the cos^{2}θdistribution, which is indicative of the soft modes but centered around both axes of symmetrybreaking. We obtain the vibrational amplitude dynamics along \(\widehat {\mathbf{e}}_q \,=\, [100]\) and [001], i.e., \(\Delta \bar u_{100}^2(t)\) and \(\Delta \bar u_{001}^2(t)\), in Fig. 5d. The changes are clearly more pronounced along [100], yet both experience a rapid rise on the timescale close to the period of order parameter quench. We can connect the dynamics along [100] to the process of ultrafast melting of the preexisting cCDW state, where the unpairing the amplitude mode into soft phonons from the suppression of LDW drives the large ms vibrations along [100]. Naturally, the increase in the vibration amplitude along [001] is connected to the soft modes excited upon introducing the new brokensymmetry landscape, f(η_{a}), precursory to the aCDW formation in the system. This analysis also reaffirms the transverse nature of the critical softening, hence the CDW states, a predication made by the recent joint IXS and densityfunctionalperturbation theory study^{23}.
Nonthermal phonon manifolds and their equilibration
The investigation here illustrates the fundamental differences between exciting microscopic soft modes and generating long wave collective modes in densitywave formation. In particular, the time delay between the onsets of soft modes and the order parameter represented by u_{0,ηa}(t)^{2} (Fig. 5d), ~300 fs, defines a nonadiabatic time window, which is a distinct signature of the nonequilibrium dynamics near a critical point^{4,5}. The two softmode branches connected to c and aCDWs are isolated by the vibration polar plots in Fig. 5c. Clearly shown are the modes along the two symmetrybreaking axes having very different amplitude changes; see the bifurcated trajectories of \(\Delta u_{001}^2\left( t \right)\) and \(\Delta u_{100}^2\left( t \right)\) in Fig. 5d. Akin to the equilibrium state analyses but considering the nonergodicity of the transient state, we may use these amplitudes as a measure for the vibrational kinetic energy, or effective temperature, of the subsystems. As discussed, a key to sustaining the hidden state is the nonergodic environment. The results here thus can substantiate this claim and be used to follow the flow of the kinetic energies in the system. The larger \(\Delta u_{100}^2\) indicates the excitation is more readily coupled to the AM of the preexisting longrange state. The differences between the two are most pronounced in the first 2 ps. Over the longer time, the phonon kinetic energy spreads into the neighboring 2D lattice modes, shown in the reduced anisotropy of the polar plot. We attribute such spreading as an evidence of thermalization between the subsystems. Indeed, this spreading is linked to a reversal of the bifurcation in \(\Delta u_{100}^2(t)\) and \(\Delta u_{001}^2(t)\), commencing at ≈2 ps. The changes can be interpreted as a cooling of the cCDW state accompanied by a continuous, but more rapid, heating of the aCDW in the next ~4.5 ps. A direct consequence of this thermal reequilibration process, as predicted by the L–G model, is the unraveling of the hidden state which is manifested in dynamics of u_{0,ηa}(t)^{2}.
Discussions
We now elaborate on how the new bidirectionally ordered state emerges with a high momentum wavevector and relate it to the reduction in transverse interorbital coupling (t_{⊥} in Fig. 1c) upon laser quench. In RTe_{3}, the transverse coupling introduces the curvatures in the FS^{18}. The curved FS contour leads to different nesting scenarios and allows the anisotropy in the EPC matrix to select a larger nesting vector in the lowsymmetry state^{19,20,42}. The possibility of symmetry recovery from pulsed laser excitation is described by Rettig et al.^{24} in their trARPES experiment, where the FS curvature is reduced in just 250 fs at an absorbed fluence of 0.27 mJ/cm^{2} (equivalent to F = 0.9 mJ/cm^{2}). They map this onto a reduction of t_{⊥} as large as 25% to account the change. Our calculation of the static charge fluctuation susceptibility based on a model band structure (refs. ^{20,54}) shows that with t_{⊥} reduced by 25%, the nesting vector q_{χ,c} shifts towards the lower q value by ≈1.25%; see Supplementary Fig. 6. This could explain the increase of the ordering vector here—since Q_{c} = c* − q_{χ,c}^{54} and ≈2.5% is found at twice the fluence of the trARPES experiment. We point out that our susceptibility calculation does not explicitly consider the formation of aCDW. Nonetheless, empirical data do show the supporting trend that in the heavier RTe_{3} members (where aCDW is substantiated) the ordering vector of the weakened cCDW shifts to ≈0.30–0.32c*^{16,54}. The movements reflect the reduced FS curvature, hence a more symmetric FS as the system enters the bidirectional state. Given that our observed Q_{c+}≈0.30c* and Q_{a} ≈ 0. 30a* reflecting the transient FS topology already reach the same level of changes in the bidirectional regime^{16,54}, it may be concluded that the electronic reorganization induced by the laser excitation may be behind the ultrafast quench mechanism leading to the hidden state.
The longlived topological defects are widely anticipated in a nonequilibrium SSB process^{4,7,17}. Relevant to the ultrafast setting and in the studies of RTe_{3}, several types of defect generating mechanism have been discussed, including localized phase impurities (referred to as topological defects in ref. ^{38}) from direct highenergy pulse excitations^{38,41}, the transient domain wall formation driven by nonuniform excitation^{28,60}, the rapid (re)emergence of broken symmetry^{4,7}, and from competing brokensymmetry orders as discussed here. Generally, introducing point defects within the ordered lowsymmetry state has a high cost in local energy density, given creating such impurities must shift the phases away from those of the surrounding domains. Furthermore, the presence of defects also softens the CDW sublattice, causing amplitude modulation and their impact can be felt far beyond the point defects.
Exploiting the defectAM interactions, the longstanding annihilation dynamics of topological defects after the nonequilibrium phase transition^{4,7,17} were tackled in TbTe_{3} system with the timedomain reflectivity measurements^{28,60,61}. By applying a second weak pump laser pulse to coherently excite the AM in the quenched CDW system, hence allowing measurements of AM damping and linewidth broadening, the authors monitor the annihilation of the topological defects. Interestingly, at least two different timescales were identified. One that occurs in 5–8 ps, which they attribute to the annihilation of the coherent domain walls developed during the inhomogeneous laser quench; a slower process up to ≈30 ps timescale is associated with annihilation of incoherent defects created by Kibble–Zurek mechanism^{4,7}; and an even longer timescale of relaxation is inferred^{4,7}.
Here, the bidirectionally ordered hidden state emerges as a new inhomogeneous component, which play a similar role as the domain wall formation in previous investigations and will impact the ensuing CDW state evolution. The transient inhomogeneities are the sources of Kibble–Zurektype topological defects created in the (re)emergence of the brokensymmetry state and the longterm residual effects expected following a rapid cooling^{4,7}. We focus on the \(\bar Q_c(t)\), which is directly sensitive to the reorganization; the dynamics at different stages are featured in Figs. 4d and 5d. The lack of persistent topological defects generations directly from the quench is reflected in the absence of detectable movement in \(\bar Q_c\) at the early stage (see Fig. 5d; also ref. ^{42}). The natural type of defects emerging following the formation of hidden states is the dislocation (see Fig. 4c), caused by the strain fields in the CDW superlattice induced by the local phase modulations—the bidirectional state has a larger wavevector than the uniaxial state. The strain fields and the majority of the topological defects created at the domain boundaries can coherently unwind as the domains of the hidden state recede over the ps thermal recovery process. This is followed by a much slower decay over tens of ps featuring an incoherent process for defect annihilation^{60} and some defects could even be frozen out after the system is cooled sufficently^{4,7}. Indeed, over the ns timescale in which the thermal state is well restored (\(\hat m_c\) = 0.7 – 0.8, consistent with a thermal state at ≈369 K estimated based on the fluence; see Supplementary Fig. 1), an anomalously large shift in \(\bar Q_c\) of ≈0.75% has persisted (Fig. 4d). This shift is more than three times of what expected of the thermal state (≈0.2%). The residual effect is a signature that the topological defects created in a rapid nonthermal symmetrybreaking dynamics can be preserved in the system and significantly modify the collective state properties well after the phase transition. Unfortunately, we are unable to monitor the changes beyond 2 ns here although the relaxation on a longer timescale eventually reset the system over the pumpprobe cycles (≤1 ms).
In summary, we have provided highresolution fs electron scattering data on the hidden state formation in CeTe_{3}, uncovering nonthermal critical onset and bidirectional nature of the lightinduced state. Evidence suggests the critical softening to be mediated by the transverse excitation of the phonons that can be traced to the change of the FS topology driven by the transverse interorbital coupling suppressed by the fs laser pulses. Perhaps the most important finding here is that, rather than engineering an entirely new metastable landscape, the quench pulses accentuate the intrinsic bidirectional competitive energy landscape that governs both the thermal and nonthermal phase transitions. The nonthermal suppression of the present dominating order tilts the symmetrybreaking preference, thereby introducing the hidden order obscured in the thermal competition. The nonequilibrium physics plays a central role; the system’s capability to maintain nonergodicity during the formation stage allows the new order to flourish and gain longrange correlation. RTe_{3}, while simplest in manifestation, is likely not unique to harbor such hidden states. In light of the phase competition to be a fundamental phenomenon in any highsymmetry, low dimensional systems, it is highly likely that the lightinduced hidden state formation would be a common phenomenon in any competitive system. In that case, the outofequilibrium approach to accentuate the hidden order and probe the coupling constants could be a powerful approach to explore the fundamental physics often obscured under the equilibrium condition.
Methods
Sample and experimental setup
The singlecrystal CeTe_{3} samples are grown using the chemical vapor transport technique^{13} and tapeexfoliated into 30–50 µm sized thin flakes^{26} before transferring to a TEM grid as freestanding. Before loaded into the ultrafast beamline, samples were prescreened for uniformity from the contrast in optical microscope and then the thickness (d = 25 ± 10 nm) was determined by performing zeroloss electron energy loss spectroscopy thickness map with TEM. In the pumpprobe experiments, the electron beam waist, >100 μm, covers the entire sample. The pump laser beam with SPolarization is directed at 45° incidence, with the pump fluence estimated by measuring waist radius, σ, and the pump power P: \(F \,=\, \frac{P}{{f_{rep}\pi \sigma ^2}}\), where f_{rep} = 1 kHz is the repetition rate of pumpprobe experiment. In Supplementary Fig. 1, the excitation field profile is obtained by solving the Maxwell equations with the boundary condition of the thin film^{29} based on the real and imaginary refraction index n = 1.04 and k = 3.45 determined for CeTe_{3}^{27}. It is worth noting that the profile of local absorbed energy density is not an exponential decay, but a modified flat distribution due to the interference effect between the forward and backward fields. Furthermore, the total absorbed energy density by the sample is calculated to be: \(E \,=\, \frac{{FA\cos (45^ \circ )}}{{2\ln \left( 2 \right)d}}\), where A ≈ 0.3 is obtained from solving the Maxwell equations and similarly the transfer matrix method^{29,62}. The absorbed fluence is independently verified by measuring the thermal melting threshold as discussed below.
Landau–Ginzburg freeenergy model based on inputs from the experiments
The normalized intensities \(\hat m_c^{meta}\) and \(\hat m_{c,\infty }\) (see Supplementary Fig. 3) across the fluence range are used to evaluate the energy density thresholds for the phase transitions at the thermal and nonthermal regimes; the results are summarized in Fig. 3c. Here, the fluence to the absorbed energy density conversion is well calibrated by the thermal melting threshold at F_{c,T} ≈ 6.3 mJ/cm^{2} (Fig. 3d), corresponding to an energy density gain of E_{c,T} 2.0 eV/nm^{3} needed for heating the sample to T_{c} = 540 K (Supplementary Fig. 2) based on the specific heat of ≈100 J/(mol·K)^{53}. At the nominal fluence of our experiment (1.85 mJ/cm^{2}), the effective temperature T_{eff} of the specimen is ≈ 369 K (a ≈70 K rise from the temperature of the initial state T_{i} = 298 K). The Landau coefficients can be determined based on the critical energy density scales determined by the experiments. In the L–G model, the critical energy density connected to forming the cCDW state is set by the free energy minimum. Experimentally, supplying an energy density E_{c,c} 0.64 eV/nm^{3} reduces η_{c} from 1 to 0, and from this determination we obtain \(\alpha \,=\, \frac{{4\left {E_{c,c}} \right}}{{T_c \,\, T_i}} \,=\, 1.05\times10^{  2}\) eV nm^{−3} K^{−1}, and A_{4} = α(T_{c} − T_{i}) = 2.54 eV nm^{−3}.
The remaining Landau coefficients are determined in accordance with \(\alpha^\prime \,=\, \frac{{\left {\eta _{c,th}} \right^2\widetilde A}}{{T_c \,\, T_i}}\), \(A_4^\prime \,=\, \frac{{\alpha\prime (T_c \,\, T_i) \,\, \widetilde A\left {\eta _c^{meta}} \right^2}}{{\left {\eta _a^{meta}} \right^2}}\), and \(\left {\eta _c^{meta}} \right^2 \,=\, \frac{{\alpha(T_c \,\, T^{(c)}) \,\, \widetilde A\left {\eta _a^{meta}} \right^2}}{{A_4}}\), where \(\widetilde A\) is floated as the fitting parameter to fit the experimental parameters, i.e., the η_{c,th} = 0.87, and the (\(\left {\eta _c^{meta}} \right^2,\,\left {\eta _a^{meta}} \right^2\)) reported as the coordinates of the freeenergy minima as a function of the excitation energy density. We find the data across the fluence range can be best fitted with Ã = 4.10 eV nm^{−3} and an estimate of T^{(c)} based on involving only 45% of the active specific heat at the metastable stage at ≈1 ps—the more restricted phase space that the early stage excitation couples to and leads to a lower nonthermal melting threshold. Accordingly, we determine A_{4} = 2.54 eV nm^{−3}, α′ = 1.29 × 10^{−2} eV nm^{−3} K^{−1}, and A′_{4} = 22.0 eV nm^{−3}. The meanfield energy gain from opening aCDW gap^{45} is given by \(E_{el} \,=\, n\left( {{\it{\epsilon }}_F} \right)\Delta _e^2\left[ {1/2 \,+\, \ln \left( {2\varepsilon _F/{{\Delta }}_e} \right)} \right]^2\), where Δ_{e} is the CDW gap and n(ε_{F}) = 1.48 state/eV/(u.c.v.) the ungapped density of states near the Fermi energy, ε_{F} = 3.25 eV^{16}. The implicit role of EPC lies in the large gap size. Here, the unit cell volume^{13} (u.c.v.) is 0.503 nm^{3}. We obtain E_{el} ~ 1.55 eV/nm^{3} based on Δ_{e} = 0.59 eV^{25} and the understanding that 33% of the original FS contour remains metallic in CeTe_{3}^{16}. The E_{el} deduced here is similar to the L–G free energy minimum at T = 0, i.e., the condensation energy E_{c,0} ≈ 1.63 eV/nm^{3}.
Data availability
The datasets generated and analyzed during the current study are available from the corresponding author upon reasonable request.
References
 1.
Tauber, U. C. Critical Dynamics: A Field Theory Approach to Equilibrium and Nonequilibrium Scaling Behavior (Cambridge University Press, 2014).
 2.
Eisert, J., Friesdorf, M. & Gogolin, C. Quantum manybody systems out of equilibrium. Nat. Phys. 11, 124 (2015).
 3.
Bray, A. J. Theory of phaseordering kinetics. Adv. Phys. 43, 357–459 (1994).
 4.
del Campo, A. & Zurek, W. H. Universality of phase transition dynamics: topological defects from symmetry breaking. Int. J. Mod. Phys. A 29, 1430018 (2014).
 5.
Clark, L. W., Feng, L. & Chin, C. Universal spacetime scaling symmetry in the dynamics of bosons across a quantum phase transition. Science 354, 606–610 (2016).
 6.
Zurek, W. H. Causality in condensates: gray solitons as relics of BEC formation. Phys. Rev. Lett. 102, 105702 (2009).
 7.
Kibble, T. W. B. Topology of cosmic domains and strings. J. Phys. A: Math. Gen. 9, 1387–1398 (1976).
 8.
Nasu, K. Photoinduced Phase Transitions. (World Scientific, 2004).
 9.
Fausti, D. et al. Lightinduced superconductivity in a stripeordered cuprate. Science 331, 189–191 (2011).
 10.
Stojchevska, L. et al. Ultrafast switching to a stable hidden quantum state in an electronic crystal. Science 344, 177–180 (2014).
 11.
Han, T.R. T. et al. Exploration of metastability and hidden phases in correlated electron crystals visualized by femtosecond optical doping and electron crystallography. Sci. Adv. 1, e1400173 (2015).
 12.
Basov, D. N., Averitt, R. D. & Hsieh, D. Towards properties on demand in quantum materials. Nat. Mater. 16, 1077 (2017).
 13.
Malliakas, C. D. & Kanatzidis, M. G. Divergence in the behavior of the charge density wave in RETe3 (RE = rareearth element) with temperature and RE element. J. Am. Chem. Soc. 128, 12612–12613 (2006).
 14.
Ru, N. et al. Effect of chemical pressure on the charge density wave transition in rareearth tritellurides RTe_{3}. Phys. Rev. B 77, 035114 (2008).
 15.
Banerjee, A. et al. Charge transfer and multiple density waves in the rare earth tellurides. Phys. Rev. B 87, 155131 (2013).
 16.
Brouet, V. et al. Angleresolved photoemission study of the evolution of band structure and charge density wave properties in RTe_{3} (R = Y, La, Ce, Sm, Gd, Tb, and Dy). Phys. Rev. B 77, 235104 (2008).
 17.
Rivers, R. J. ZurekKibble causality bounds in timedependent GinzburgLandau Theory and Quantum Field Theory. J. Low. Temp. Phys. 124, 41–83 (2001).
 18.
Yao, H., Robertson, J. A., Kim, E.A. & Kivelson, S. A. Theory of stripes in quasitwodimensional rareearth tellurides. Phys. Rev. B 74, 245126 (2006).
 19.
Johannes, M. D. & Mazin, I. I. Fermi surface nesting and the origin of charge density waves in metals. Phys. Rev. B 77, 165135 (2008).
 20.
Eiter, H.M. et al. Alternative route to charge density wave formation in multiband systems. Proc. Natl Acad. Sci. USA 110, 64–69 (2013).
 21.
Zhu, X., Cao, Y., Zhang, J., Plummer, E. W. & Guo, J. Classification of charge density waves based on their nature. Proc. Natl Acad. Sci. 112, 2367–2371 (2015).
 22.
Maschek, M. et al. Wavevectordependent electronphonon coupling and the chargedensitywave transition in TbTe_{3}. Phys. Rev. B 91, 235146 (2015).
 23.
Maschek, M. et al. Competing soft phonon modes at the chargedensitywave transitions in DyTe_{3}. Phys. Rev. B 98, 094304 (2018).
 24.
Rettig, L. et al. Persistent order due to transiently enhanced nesting in an electronically excited charge density wave. Nat. Commun. 7, 10459 (2016).
 25.
Leuenberger, D. et al. Classification of collective modes in a charge density wave by momentumdependent modulation of the electronic band structure. Phys. Rev. B 91, 201106 (2015).
 26.
Novoselov, K. S. et al. Twodimensional atomic crystals. Proc. Natl Acad. Sci. USA 102, 10451–10453 (2005).
 27.
Sacchetti, A., Degiorgi, L., Giamarchi, T., Ru, N. & Fisher, I. R. Chemical pressure and hidden onedimensional behavior in rareearth tritelluride chargedensity wave compounds. Phys. Rev. B 74, 125115 (2006).
 28.
Yusupov, R. et al. Coherent dynamics of macroscopic electronic order through a symmetry breaking transition. Nat. Phys. 6, 681–684 (2010).
 29.
King, S. W. & Milosevic, M. A method to extract absorption coefficient of thin films from transmission spectra of the films on thick substrates. J. Appl. Phys. 111, 073109 (2012).
 30.
Williams, J. et al. Active control of bright electron beams with RF optics for femtosecond microscopy. Struct. Dyn. 4, 044035 (2017).
 31.
Portman, J. et al. Computational and experimental characterization of highbrightness beams for femtosecond electron imaging and spectroscopy. Appl. Phys. Lett. 103, 253115 (2013).
 32.
van Oudheusden, T. et al. Compression of subrelativistic spacechargedominated electron bunches for singleshot femtosecond electron diffraction. Phys. Rev. Lett. 105, 264801 (2010).
 33.
Zhou, F., Williams, J. & Ruan, C.Y. Femtosecond electron spectroscopy in an electron microscope with high brightness beams. Chem. Phys. Lett. 683, 488–494 (2017).
 34.
Eichberger, M. et al. Snapshots of cooperative atomic motions in the optical suppression of charge density waves. Nature 468, 799–802 (2010).
 35.
Schmitt, F. et al. Transient electronic structure and melting of a charge density wave in TbTe_{3}. Science 321, 1649–1652 (2008).
 36.
Han, T. R. T. et al. Structural dynamics of twodimensional chargedensity waves in CeTe_{3} investigated by ultrafast electron crystallography. Phys. Rev. B 86, 075145 (2012).
 37.
Schmitt, F. et al. Ultrafast electron dynamics in the charge density wave material TbTe_{3}. N. J. Phys. 13, 063022 (2011).
 38.
Zong, A. et al. Evidence for topological defects in a photoinduced phase transition. Nat. Phys. 15, 27–31 (2019).
 39.
Rohwer, T. et al. Collapse of longrange charge order tracked by timeresolved photoemission at high momenta. Nature 471, 490–493 (2011).
 40.
Trigo, M. et al. Coherent order parameter dynamics in SmTe_{3}. Phys. Rev. B 99, 104111 (2019).
 41.
Kogar, A. et al. Lightinduced charge density wave in LaTe3. Nat. Phys. 16, 159–163 (2020).
 42.
Moore, R. G. et al. Ultrafast resonant soft xray diffraction dynamics of the charge density wave in TbTe_{3}. Phys. Rev. B 93, 024304 (2016).
 43.
Yusupov, R. V., Mertelj, T., Chu, J. H., Fisher, I. R. & Mihailovic, D. Singleparticle and collective mode couplings associated with 1 and 2directional electronic ordering in metallic RTe_{3} (R = Ho, Dy,Tb). Phys. Rev. Lett. 101, 246402 (2008).
 44.
Lavagnini, M. et al. Raman scattering evidence for a cascade evolution of the chargedensitywave collective amplitude mode. Phys. Rev. B 81, 081101 (2010).
 45.
Gruner, G. The dynamics of chargedensity waves. Rev. Mod. Phys. 60, 1129–1182 (1988).
 46.
Hinton, J. P. et al. New collective mode in YBa2Cu3O6 observed by timedomain reflectometry. Phys. Rev. B 88, 060508 (2013).
 47.
Boyanovsky, D., Lee, D.S. & Singh, A. Phase transitions out of equilibrium: domain formation and growth. Phys. Rev. D. 48, 800–815 (1993).
 48.
Forster, D. Hydrodynamic Fluctuations, Broken Symmetry, and Correlation Functions (CRC Press, 1975).
 49.
Overhauser, A. W. Observability of chargedensity waves by neutron diffraction. Phys. Rev. B 3, 3173–3182 (1971).
 50.
Bruce, A. D. & Cowley, R. A. The theory of structurally incommensurate systems. III. The fluctuation spectrum of incommensurate phases. J. Phys. C: Solid State Phys. 11, 3609–3630 (1978).
 51.
Overhauser, A. W. Spin density waves in an electron gas. Phys. Rev. 128, 1437 (1962).
 52.
Malliakas, C., Billinge, S. J. L., Kim, H. J. & Kanatzidis, M. G. Square nets of tellurium: rareearth dependent variation in the chargedensity wave of RETe_{3} (RE = rareearth element). J. Am. Chem. Soc. 127, 6510–6511 (2005).
 53.
Ru, N. & Fisher, I. R. Thermodynamic and transport properties of YTe_{3}, LaTe_{3}, and CeTe_{3}. Phys. Rev. B 73, 033101 (2006).
 54.
Moore, R. G. et al. Fermi surface evolution across multiple charge density wave transitions in ErTe_{3}. Phys. Rev. B 81, 073102 (2010).
 55.
Cochran, W. Crystal stability and the theoy of ferroelectricity. Adv. Phys. 9, 387–423 (1960).
 56.
Guinier, A. Xray Diffraction in Crystals, Imperfect Crystals, and Amorphous Bodies (Dover, 1994).
 57.
Krivoglaz, M. A. Diffuse Scattering of Xray and Neutrons by Fluctuations (SpringerVerlag, 1996).
 58.
Lamago, D. et al. Measurement of strong phonon softening in Cr with and without Fermisurface nesting by inelastic xray scattering. Phys. Rev. B 82, 195121 (2010).
 59.
Kohn, W. Image of the Fermi surface in the vibration spectrum of a metal. Phys. Rev. Lett. 2, 393–394 (1959).
 60.
Mertelj, T. et al. Incoherent topological defect recombination dynamics in TbTe_{3}. Phys. Rev. Lett. 110, 156401 (2013).
 61.
Yusupov, R. et al. Femtosecond coherent nonequilibrium electronic ordering and dynamics of topological defect in charge density waves. J. Superconductivity Nov. Magn. 24, 1191–1193 (2011).
 62.
Katsidis, C. C. & Siapkas, D. I. General transfermatrix method for optical multilayer systems with coherent, partially coherent, and incoherent interference. Appl. Opt. 41, 3978–3987 (2002).
Acknowledgements
The authors acknowledge help from K. Sun with the TEM measurements, and the helpful contributions from E.A. Nowadnick in the simulation. We also thank M. Maghrebi for insightful discussions. The work at MSU was funded by the US Department of Energy, grant DEFG0206ER46309. The experimental facility was supported by US National Science Foundation, grant DMR 1625181. A.F.K. acknowledges support from the National Science Foundation under Grant No. DMR1752713. The work conducted at Argonne National Laboratory was supported by the US Department of Energy, Office of Science, Basic Energy Sciences, Materials Sciences, and Engineering Division.
Author information
Affiliations
Contributions
C.Y.R., F.Z., and J.W. designed and executed the experiments. C.D.M. and M.G.K. prepared the samples. C.Y.R. developed the software for analyzing the data. F.Z. and C.Y.R. did the data analysis. S.S. did the calculations on the laser excitation profile. A.F.K. did the calculations on the susceptibility and ordering vector.
Corresponding author
Ethics declarations
Competing interests
The authors declare no competing interests.
Additional information
Peer review information Nature Communications thanks the anonymous reviewer(s) for their contribution to the peer review of this work. Peer reviewer reports are available.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary information
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Zhou, F., Williams, J., Sun, S. et al. Nonequilibrium dynamics of spontaneous symmetry breaking into a hidden state of chargedensity wave. Nat Commun 12, 566 (2021). https://doi.org/10.1038/s41467020208345
Received:
Accepted:
Published:
Further reading

Ultrafast formation of domain walls of a charge density wave in SmTe3
Physical Review B (2021)
Comments
By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.