Abstract
A state of matter with a multicomponent order parameter can give rise to vestigial order. In the vestigial phase, the primary order is only partially melted, leaving a remaining symmetry breaking behind, an effect driven by strong classical or quantum fluctuations. Vestigial states due to primary spin and chargedensitywave order have been discussed in ironbased and cuprate materials. Here we present the observation of a partially melted superconductivity in which pairing fluctuations condense at a separate phase transition and form a nematic state with broken Z_{3}, i.e., threestate Pottsmodel symmetry. Thermal expansion, specific heat and magnetization measurements of the doped topological insulators Nb_{x}Bi_{2}Se_{3} and Cu_{x}Bi_{2}Se_{3} reveal that this symmetry breaking occurs at \({{T}}_{\mathrm{nem}} \simeq 3.8\,K\) above \({T}_{{\mathrm{c}}} \simeq 3.25\,K\), along with an onset of superconducting fluctuations. Thus, before Cooper pairs establish longrange coherence at T_{c}, they fluctuate in a way that breaks the rotational invariance at T_{nem} and induces a crystalline distortion.
Introduction
Nematic electronic phases with vestigial order^{1,2} are known from ironbased superconductors and cuprates, where it has been suggested that the nematic phase and the nearby spin and chargedensity wave states are not independent competing but intertwined electronic phases^{1,2,3,4,5,6,7,8,9,10}. The densitywave states are the primary electronic phases and characterized by a multicomponent order parameter. The nematic phase is a fluctuationdriven phase and characterized by a composite order parameter. Then the spin or charge densitywave order melts partially, but leaves an Ising, i.e., Z_{2}nematic state as a vestige. Vestigial order whose primary order is superconductivity has not been observed. Such partially molten superconductivity requires a material with unconventional, multicomponent order parameter, and strong pairing fluctuations.
When the topological insulator Bi_{2}Se_{3} is doped with electrons, e.g., by intercalation of Cu, Sr, Nb, or other metal ions in its layered structure, a superconducting state is formed^{11,12}. The presence of a strong spinorbit coupling, which also manifests itself in a topological surface state of the parent insulator^{13}, led to the proposal of unconventional pairing with an oddparity symmetry and topological superconductivity^{14}. The low carrier concentration, the layered structure, and the low ratio ξ/λ_{F} of the superconducting coherence length and the Fermi wavelength^{11,12}, strongly enhance fluctuation effects. In addition, numerous experiments have shown that the superconducting state is accompanied by a spontaneous breaking of rotational symmetry with a pronounced twofold anisotropy within the Bi_{2}Se_{3} basal plane^{15,16,17,18,19,20,21,22}; see Ref. ^{23} for a recent review. The twofold symmetry can be observed in fieldangle resolved experiments where a magnetic field is rotated in the plane with respect to the crystalline axes and the corresponding physical quantity (e.g., spin susceptibility, specific heat, magnetoresistance, upper critical field, magnetization, magnetic torque) is represented as a function of angle^{15,16,17,18,19,20,21,22,23}. This behavior directly reflects the anisotropy of the superconducting state. Thus, doped Bi_{2}Se_{3} is an unconventional nematic superconductor with a pairing wave function in either the twocomponent E_{u} or E_{g} point group representation, the only pairing states that spontaneously break the trifold crystal symmetry within the basal plane. The temperature dependence of the penetration depth of Ref. ^{24} supports point nodes, consistent with E_{u} oddparity pairing.
In this article we report on highresolution thermal expansion experiments on a superconducting monocrystalline Nbdoped Bi_{2}Se_{3} sample in combination with electrical transport, DC magnetization, and specific heat data demonstrating a Z_{3}vestigial nematic phase with enhanced superconducting fluctuations. We have measured the linear thermal expansion in three different crystalline directions in the Bi_{2}Se_{3} basal plane and observed a strong anisotropic expansion occurring at a temperature of ~0.5 K above the superconducting transition. Our highresolution magnetization, electrical resistivity, and specific heat data—after zooming near T_{c}—show that an anomaly with increasing superconducting fluctuations occurs at the nematic transition. As we will explain below, these observations are perfectly consistent with a vestigial nematic phase of symmetrybreaking pairing fluctuations, recently predicted in Ref. ^{25}. This observation of a genuine symmetry breaking of pairing fluctuations above T_{c} is qualitatively distinct from the gradual onset of orderparameter fluctuations in the disordered phase^{26} or the crossover to Bose–Einstein condensation of pairs^{27}. It corresponds to a sharply defined state of matter that might, e.g., undergo a separate quantum phase transition when a magnetic field is applied in the plane at low temperatures. Qualitatively similar results are obtained both on another Nbdoped Bi_{2}Se_{3} single crystal and on a Cudoped Bi_{2}Se_{3} single crystal, thus, demonstrating the reproducibility and universality of the observed features.
Results
Magnetoresistance and nematicity
The advantage of the Nbdoped Bi_{2}Se_{3} system is that single crystals with a high superconducting volume fraction and a complete zero resistance can be found, as the results presented here show. All our bulk thermodynamic data (thermal expansion, magnetization, and specific heat) show relatively large anomalies at the superconducting transition.
Figure 1a shows magnetoresistance data recorded at 0.35 K with the magnetic field applied strictly parallel to the Bi_{2}Se_{3} basal plane for different directions in the plane with respect to the trigonal crystalline axes. We determine the approximate H_{c2} values from the fields in which 75% of the normal state resistance is reached and plot these values in Fig. 1b in a polar diagram as a function of angle ϕ. A significant angular variation of H_{c2} can be observed, ranging from 0.43 T at 90° where H_{c2} is minimal to 1.42 T at 0° with a clear maximum of H_{c2}. The data show a pronounced twofold symmetry, at odds with the trifold crystalline symmetry. This is the characteristic property of nematic superconductivity in Nb_{x}Bi_{2}Se_{3}^{19,20}, also known from Cu_{x}Bi_{2}Se_{3}^{15,16,22} and Sr_{x}Bi_{2}Se_{3}^{17,18,21}. A fit with a theoretical model for nematic SC of the form^{20}
yields \(\Gamma \approx 3.32{\mathrm{\Omega }}\) and H_{c2}(0) = 1.42 T as a measure of the anisotropy in the basal plane. It should be noted that the normal state resistance well above the upper critical field has no variation for the different orientations of the magnetic field in the plane, indicating an isotropic normal state within the trigonal basal plane. We have previously found that the orientation of this nematic superconducting order parameter for this sample always appears to be pinned along the same of the three equivalent crystal directions in the Bi_{2}Se_{3} basal plane, even if the sample is warmed to room temperature between different experiments^{19}. The origin of this preference for a particular direction is unknown, but likely associated with microscopic details of the sample morphology, such as internal strain or microcracks (see Supplementary Discussion and Supplementary Fig. 6 for more details).
Thermal expansion and vestigial order
Through thermal expansion experiments, we have a highly sensitive bulk thermodynamic probe that is not only sensitive to the anharmonicity of phononic contributions but also to electronic degrees of freedom including nematic and superconducting order. We focus here on \({\mathrm{\Delta }}L(T,H)_\mu /L_0\) measured along different directions μ in the Bi_{2}Se_{3} basal plane, which directly represents the change in length ΔL of the sample as a function of temperature or magnetic field, normalized to the length L_{0} at ambient temperature. This quantity is directly related to the linear thermal expansion coefficient α_{μ} (T,H) = 1/L_{0} dL_{μ}(T)/dT. Figure 2a shows the linear thermal expansion ΔL_{μ}/L_{0} measured along the three directions within the Bi_{2}Se_{3} basal plane of 90°, 155°, and 215°. All data fall perfectly on each other in the normal state, but begin to deviate gradually from each other below 3.8 K. In the following, we will refer to this characteristic temperature as T_{nem}, because here the onset of a twofold crystalline distortion and thus nematicity occurs. The twofold crystalline distortion with a relative length change ΔL/L_{0} = 2 × 10^{−7} amounts to a distortion of less than 0.1 femtometers within the unit cell. Still, these minute changes smaller than the size of the proton, are clearly resolvable in our measurements. The distortion is correlated with the upper critical field, with a small negative length change along 90° where the H_{c2} minimum occurs (Fig. 1b), and large positive anomalies at 155° and 215°, both near the mean H_{c2} value. At 3.25 K, much smaller anomalies are visible that can be identified as the superconducting transition, as the comparison with the specific heat (shown in the same graph) reveals. In the specific heat, the anomaly at T_{nem} is obscured by the phonon background, but becomes visible in the temperature derivative of C/T (Fig. 2b), where a small stepshaped anomaly occurs. The anomalies in ∆L/L_{0} at T_{nem} show up as a somewhat broadened step. As a firstorder derivative of the free energy, a steplike transition in ∆L/L_{0} is the characteristic signature of a firstorder transition, while a secondorder transition would appear as a kink. In distinction, the superconducting transition at T_{c} remains the standard secondorder transition, as evidenced by the jump in the specific heat. Figure 2c shows the Meissner signal in the zerofield cooled and fieldcooled DC magnetization, which agrees with T_{c} ≈ 3.25 K obtained from the specific heat. We also show the same data, but with a magnification of 10^{5}, to illustrate that an enhanced diamagnetic response, signaling superconducting fluctuations, already sets in at T_{nem}. The electrical resistivity in Fig. 2d shows a similar trend with a drop in resistivity consistent with paraconductivity, i.e., superconducting fluctuations, well above the main transition.
Our Xray diffraction results (see section “Methods”) show that the doped Bi_{2}Se_{3} phase of the R3m space group is the majority phase responsible for T_{c} at 3.25 K. As minority phases Bi_{2}Se_{3} of space group P3m1 and NbBiSe_{3} were found. The latter occupy far too little volume to explain such large anomalies in thermal expansion. From this, we conclude that the observed crystalline distortion below T_{nem} = 3.8 K is caused by a transition separate from the main superconducting transition, but linked to the occurrence of superconducting fluctuations, which cause a weak Meissner effect and decrease in resistivity below this temperature.
Two stage transition
Our data thus show that nematic superconductivity occurs in the form of a twostage transition, see also Fig. 3 for an illustration. The distortion forms near the higher onset temperature T_{nem}, where superconducting fluctuations in the magnetization and the firstorder derivative of the specific heat are visible. The superconducting transition occurs at a lower temperature T_{c}, which corresponds to the formation of a global phasecoherent superconducting state. The signs and magnitudes of the length changes for the three measuring directions are consistent with the indicated distortions, see “Methods” section.
It could be argued that thermal expansion reveals a separate structural transition at T_{nem} that has nothing to do with superconductivity^{28}. Such a sequence of independent or competing transitions would be allowed within the Landau theory of phase transitions. Obviously, the anomaly in the diamagnetic response at T_{nem} is already strong evidence that this is not the case. Furthermore, in Fig. 4 we show magnetostriction data for the 155° direction, along which we observed the greatest change in length in Fig. 2a. Here, ∆L/L_{0} was measured at a fixed temperature of 350 mK as a function of the magnetic field. A broad steplike transition occurs with a total length change of ∆L/L_{0} ≈ 0.22 × 10^{−7} with onset at ~1.0 T. Figure 1 shows that the resistively determined H_{c2} for this direction occurs at 0.7 T. Magnetostriction shows that the crystalline distortion is constant up to this field where a kink occurs, then it is gradually removed up to 1.0 T. The overall anomaly represents a broad step in ∆L/L_{0}, which is the expected characteristics of a firstorder transition. A small hysteresis can be seen in the data measured upon sweeping the field up and down. Experimental artifacts as source for the hysteresis have been carefully excluded, especially since the field scanning speed was kept very slow at 0.02 T/min, which typically does not cause any hysteresislike effects in reversible samples. Given the layered structure of the sample with the field aligned in parallel, this hysteresis is most likely a consequence of flux pinning effects, which are typically strong at such a low temperature. Therefore, the hysteresis should not necessarily be regarded as evidence of a firstorder H_{c2} nature. The fieldinduced length change at low temperature shown in Fig. 4 corresponds to the temperatureinduced length change at zero field shown in Fig. 2a. This observation provides further evidence that the nematic distortion is closely linked to the superconducting state, with a separate nematic transition H_{nem} occurring above the main superconducting H_{c2} transition, as shown in Fig. 4.
Further data on thermal expansion measured on a second sample of the same batch are shown in Supplementary Figs. 1 and 2 and are discussed in Supplementary Note 1. They show a similar behavior, although the crystalline distortion occurring below T_{nem} is weaker due to a multidomain structure. For this sample we have also measured thermal expansion in fixed magnetic fields, and it can be seen that T_{nem} is suppressed by the magnetic field together with T_{c}, further confirming that the two transitions are closely related. We also show similar data of a Cu_{0.2}Bi_{2}Se_{3} single crystalline sample (Sample 3) in Supplementary Fig. 3–5 and the data are discussed in Supplementary Note 2.
Discussion
The two thermodynamic properties, the specific heat and the linear thermal expansion coefficient α_{μ}(T) = 1/L_{0} dL_{μ}(T)/dT are closely related in the vicinity of a phase transition through the Clausius Clapeyron (2) and Ehrenfest (3) relation for first and second order phase transitions, respectively. The proportionality is the uniaxial pressure dependence dT_{c}/dp_{μ} of the transition (T_{c} is the critical temperature at which the phase transition occurs, p_{μ} is the uniaxial pressure applied along a certain crystalline direction μ, V_{mol} is the molar volume, ΔS is the jump in entropy at a first order transition and ΔC_{p} is the jump in the specific heat at constant pressure at a second order phase transition).
A large anomaly in thermal expansion and a small anomaly in specific heat means in both cases that T_{nem} is strongly dependent on uniaxial pressure and the electronic nematic order is strongly coupled to the crystalline lattice. The strong crystalline distortion observed here using linear thermal expansion therefore means that the nematic transition is strongly dependent on pressure or strain. Such a behavior can also be observed, for example, in iron based superconductors, where a nematic transition occurs in the vicinity to a spin density wave transition and causes large anomalies in thermal expansion^{29}.
Our findings can be explained in terms of vestigial order due to superconducting fluctuations. In fact, recently it has been suggested that such vestigial order should emerge from the superconducting phase in doped Bi_{2}Se_{3}^{25}. On the one hand, the superconducting order parameter of either the E_{g} or the E_{u} representation has two components^{14,30,31,32}
that are characterized by the overall amplitude Δ_{0}, the global U(1) phase φ and three distinct values of the angle \(\theta = \left\{ {\frac{\pi }{6},\frac{\pi }{2},\frac{{5\pi }}{6}} \right\}\) that select a specific crystalline axis. Superconducting fluctuations will then induce a phase transition to a vestigial nematic state at a temperature T_{nem} above T_{c}. While superconductivity is signaled by a finite expectation value of Δ_{x} and/or Δ_{y}, the nematic phase is characterized by a finite expectation value of the composite order parameter
Upon increasing the temperature, superconducting fluctuations continue to break the rotational symmetry, even after restoration of global U(1) symmetry at T_{c}. The composite order parameter Q_{μν} is made up of combinations of the superconducting order parameter, similar to charge4e superconductivity proposed within the context of pairdensity wave order in cuprate superconductors^{33,34} or proton–electron superconducting condensate in liquid hydrogen^{35}. As a traceless secondrank tensor, \(\langle{\mathbf{Q}}_{\mu {\upnu}}\rangle = Q_0\left( {n_\mu n_{\upnu}  \frac{1}{2}\delta _{\mu {\upnu}}} \right)\) behaves, however, like a nematic order parameter with director \({\mathbf{n}} = \left( {\cos \theta ,\sin \theta } \right)\)^{36} and strongly couples to the strain tensor ε_{μν} via \(\kappa {\mathrm{tr}}({\mathbf{Q\varepsilon }})\) with nematoelastic coupling constant κ. A nonzero Q_{0} then induces a lattice distortion ε_{μν} \({\propto} \, {\kappa} {\mathbf{Q}}_{\mu {\upnu}}\), see Fig. 3. Thus, the lattice can be utilized to detect this unconventional electronic order. The point group analysis further yields a firstorder transition at T_{nem} into a state with Q_{0} ≠ 0, since it is in the threestate Potts model, i.e., the Z_{3} universality class. The superconducting transition continues to be of second order, all in agreement with our experimental findings. In Fig. 5a we show the nematic (Q_{0}) and superconducting (Δ_{0}) order parameters and in Fig. 5b the diamagnetic susceptibility obtained within the theory of Ref. ^{25}. The susceptibility is compared with the data of Fig. 2c, where the logarithmic axis is used to illustrate the rapid growth of diamagnetic fluctuations below T_{nem}. While the inplane anisotropy of the susceptibility is only finite below the nematic transition, the crystal symmetry allows χ_{zz} to be distinct already above T_{nem}. The magnitude of the outofplane anisotropy is determined by the ratio of the electron velocities in the corresponding directions. The anisotropy of \(H_{{\mathrm{c}}2}\left( \phi \right)\) shown in Fig. 1b (orange line) was also obtained within the same theory and is compared with the behavior without nematic order (Q_{0} = 0) where H_{c2} should have sixfold symmetry^{24,25}. Without nematic phase above T_{c}, the superconducting order parameter directly at \(H_{{\mathrm{c}}2}\left( \phi \right)\) is infinitesimal and no twofold rotational symmetry breaking should be visible, in clear contrast to experimental observations. In Ref. ^{31}, the twofold symmetric behavior of \(H_{{\mathrm{c}}2}\left( \phi \right)\) only occurred after an additional symmetrybreaking strain was added. Vestigial nematic order offers a natural explanation for this strain field.
After this work was completed, we learned about Ref. ^{37}, where a twofold symmetry breaking above T_{c} is reported. Our results agree with those of Ref. ^{37} and make evident that the high temperature phase is separated by an actual first order transition where superconducting fluctuations are enhanced. Furthermore, Refs. ^{38,39} reported on the control of nematic superconductivity by uniaxial strain, which is consistent with our observation of a coupling of the nematic order parameter to the crystal lattice.
To summarize, our data demonstrate that a separate nematic transition occurs in the doped topological insulator Nb_{0.25}Bi_{2}Se_{3} at T_{nem} = 3.8 K, i.e., about 0.5 K above T_{c}, with a distinct crystalline distortion occurring in the Bi_{2}Se_{3} basal plane. T_{nem} coincides with the onset temperature of superconducting fluctuations. The direction of the crystalline distortion is correlated with the direction of the twofold symmetry of the superconducting order parameter and is removed together with the superconductivity at or near the upper critical field H_{c2}. The two transitions are thus interconnected. Our observations are perfectly consistent with vestigial nematic order and a sequential restoration of U(1) and rotational symmetry. The new nematic phase is a state of matter in which Cooper pairs have lost their offdiagonal longrange order, yet fluctuate in a way that breaks the rotational symmetry of the crystalline lattice.
Methods
Sample characterization
The monocrystalline Nb_{0.25}Bi_{2}Se_{3} sample used in this study was selected because of its particularly large T_{c} anomalies in the specific heat, which indicates a high superconducting volume fraction, and because of its particularly large nematic inplane H_{c2} anisotropy^{19}. Our previous work also demonstrated that it forms one large nematic domain comprising ~90% of the superconducting volume fraction in which the nematic order parameter is pinned in one crystalline direction, while the remaining 10% is due to a minority domain in which the orientation is rotated by 60°. Data from a second monocrystalline Nb_{0.25}Bi_{2}Se_{3} sample with broader transition anomalies and somewhat less nematic anisotropy are shown in Supplementary Figs. 1 and 2. We also show data of a Cu_{0.2}Bi_{2}Se_{3} single crystalline sample (Sample 3) in Supplementary Fig. 3–5. These samples show a qualitatively similar behavior as Sample1.
A laboratory Xray Laue equipped with CCD camera (Photonic Science) was used to characterize the crystal quality and to determine the crystalline directions in the basal Bi_{2}Se_{3} plane. The Laue images (Fig. 6) were taken on the shiny surface of the sample after cleaving off a thin layer with shooting Xray beam along the caxis, revealing the crystal orientation and proving the hexagonal structure. We took data on different spots on the sample surface and found that the change in the crystalline direction was less than 0.2° over a distance of 0.8 mm, proving the sufficiently good single crystalline quality of the sample.
Powder XRay Diffraction pattern was collected at room temperature in the BraggBrentano geometry using a Bruker AXS D8 Advance diffractometer equipped with a Nifiltered Cu Kα radiation and a 1D LynxEye PSD detector (Fig. 7). The reflections were indexed to Bi_{2}Se_{3} (SG R3m; No 166) as a main phase. In addition, some minority phases were found as following: Bi_{2}Se_{2} and NbBiSe_{3} with corresponding space groups: P3m1 (No 164) and P2_{1}2_{1}2_{1} (No 19), respectively. This demonstrates that, while most of the Nb must be intercalated between the Bi_{2}Se_{3} layers, some Nb is incorporated into the layers on the Bi sites. This agrees with literature data^{40}. The Rietveld refinement^{41} of the diffraction patterns was done by the package FULLPROF SUITE^{42} (version July2019) using a previously determined instrument resolution function (based on the small line width polycrystalline sample Na_{2}Ca_{3}Al_{2}F_{14} measurements^{43}). Refined parameters were: scale factor, zero displacement, lattice parameters, atomic positions, isotropic Debye–Waller factors, and peak shape parameters as a ThompsonCoxHastings pseudoVoigt function. Determined lattice parameters of the rhombohedral Bi_{2}Se_{3} are equal a = b = 4.1854(4) Å, and c = 28.4633(7) Å. Because of the habit of the powdered crystal, a preferred orientation as a MarchDollase multiaxial phenomenological model was implemented in the analysis.
Experimental techniques
The highresolution linear thermal expansion was measured with a capacitive technique using a dilatometer, in which the sample is pressed by a fine screw mechanism against a cantilever forming one of the two plates of a capacitor. A change in the sample length leads to a change in the separation of the capacitor plates, which can be determined with a General Radio 1615A capacitance bridge in combination with a Stanford Research SR830 digital lockin amplifier. Before the experiments we have carefully checked that the empty dilatometer does not show any anomalies in the temperature range of interest (Supplementary Fig. 7). We have measured the thermal expansion as a function of temperature in three different directions within the basal Bi_{2}Se_{3} plane (90°, 155°, and 215°). 0° is the a direction normal to the mirror plane and corresponds to the magneticfield direction in the plane providing the maximum upper critical field, while 90° corresponds to the a* direction parallel to the mirror plane. The crystalline directions have been obtained by Laue Xray diffraction as explained above. The other directions were chosen as representative of other characteristic directions in the plane, but largely dictated by the crystal shape, which allowed a stable mounting of the sample only for certain directions. All directions were within 5° from the three different a* directions. For technical reasons, all data was taken at a slow rate of 0.02 K/min upon increasing temperature. Figure 8 shows a photograph of the mounted crystal for the three different orientations in the dilatometer. The dilatometer^{44} was well characterized using separate calibration measurements: e.g., a measurement with a Cu sample of 1 mm length of the same material as the dilatometer body shows only a very small temperature dependence without significant anomalies in the temperature range of interest. The absolute value of the change in length was calibrated using a 2 mm long undoped silicon sample, which gave a linear thermal expansion coefficient that agrees well with literature^{45} (Supplementary Fig. 8).
The specific heat was measured with a homemade calorimeter, which can be used either in AC modulated temperature mode or in long relaxation mode. The long relaxation mode provides high accuracy in the absolute value of 1% precision, while the AC mode provides high relative resolution with a high density of data points of 1000 points per K. The data presented in this letter has been acquired using the AC technique, but the absolute value has been calibrated using the relaxation technique.
DC magnetization was measured using a commercial Quantum Design Vibrating Sample SQUID magnetometer and the electrical resistance was measured using a standard 4probe technique with a Keithley 6221 AC current source combined with a SR830 digital lockin amplifier. For the latter, a low temperature piezo rotator was used to precisely align the magnetic field along the different crystalline directions and to study the H_{c2} anisotropy that reflects the nematic superconductivity. The rotator allowed millidegree precision for relative changes in orientation. However, a systematic error of less than 5° can occur with respect to the measured crystalline axes.
Theory: qstate Potts model
A qstate Potts model describes a spinlike variable s = 1, 2, …, q that can attain q different values. In our case q = 3 and the three values label the three axes of the crystal along which one assumes a local displacement. The energy of two neighboring variables s and s’ that are the same is then lower than for distinct variables, i.e., the bond energy goes like \( \frac{1}{2} J_0(q\delta_{s,{s^{\prime}}}  1)\). In general, the model has to be distinguished from a qstates clock model where an angle \(\varphi \in \left[ {0,2\pi } \right]\) can take q distinct values \({\upvarphi}_{\mathrm{s}} = \frac{{2\pi s}}{q}\) and two sites interact via an energy proportional to \( J_0{\mathrm{cos}}({\upvarphi}_{\mathrm{s}}  {\upvarphi}_{{\mathrm{s}}\prime })\). For q = 3 both models are equivalent though. Using the Potts language one can define the fraction n_{s} of the lattice in the sth state, which implies n_{1} +n_{2} +n_{3} = 1. At high temperatures one expects m_{s} = n_{s} \( \frac{1}{3}\) to have zero expectation value for all s. Below the transition temperature, one of the m_{s} becomes positive and the two others negative. Because of the condition on the sum of the n_{s} the three m_{s} are not independent: m_{1} +m_{2} +m_{3} = 0. Furthermore, the free energy of the system must be symmetric under m_{s} ↔ m_{s′}. These two conditions lead to the unique free energy expansion up to quartic order
Since there are really only two independent variables one can introduce the parametrization \({\mathrm{m}}_1 = \frac{2}{{\sqrt 3 }}{\mathrm{Q}}_1\) and \({\mathrm{m}}_{2,3} = \frac{1}{2}  \frac{1}{{\sqrt 3 }}{\mathrm{Q}}_1 \pm {\mathrm{Q}}_3\). The expansion in terms of the Q_{1,2} leads up to constants to
This is precisely the free energy expansion obtained in Ref. ^{25} in terms of the quadrupolar order parameter of Eq. (2) with \(Q_1 = \left {{\mathrm{\Delta }}_{\mathrm{x}}} \right^2\) − \(\left {{\mathrm{\Delta }}_{\mathrm{y}}} \right^2\) and \(Q_2 = {\mathrm{\Delta }}_{\mathrm{x}}^ \ast {\mathrm{\Delta }}_{\mathrm{y}} + {\mathrm{\Delta }}_{\mathrm{y}}^ \ast {\mathrm{\Delta }}_{\mathrm{x}}\). This demonstrates that the problem at hand is indeed in the universality class of the q = 3 Potts or clock models.
The sketch of the distorted unit cell in Fig. 3 is deduced from the linear coupling term κtr(Qε) between the strain tensor and the composite order parameter. Thus, the nematic order parameter acts in the same way as an applied external stress field in the E_{g} symmetry channel and hence distorts the unit cell. Moreover, the linear coupling term does not entail a change of the unit cell volume, which is assumed to be unaltered in the following. Figure 9 shows the relative length changes in the three directions 90°, 155°, and 215° caused by an E_{g} unit cell deformation as a function of the lattice parameter a/a_{0}. For the strongly exaggerated value a/a_{0} = 1.1 we also show the corresponding distorted hexagon in the inset, which is quantitatively similar to the distorted unit cell in Fig. 3. While the computed relative length changes qualitatively capture the measured thermal expansion behavior, the calculated magnitude of the 90^{o} direction is slightly larger than that observed in experiment when compared with the other two directions. This could be due to the higherorder coupling that gives rise to a change of the unit cell volume at T_{nem}.
Data availability
The experimental data supporting the findings of this work are available at https://doi.org/10.4121/uuid:8f2eed773db34d07965ad4900f5ff22d.
References
Fradkin, E., Kivelson, S. A. & Tranquada, J. M. Colloquium: theory of intertwined orders in high temperature superconductors. Rev. Mod. Phys. 87, 457–482 (2015).
Fernandes, R. M., Orth, P. P. & Schmalian, J. Intertwined vestigial order in quantum materials: nematicity and beyond. Annu. Rev. Condens. Matter Phys. 10, 133–135 (2019).
Fang, C., Yao, H., Tsai, W. F., Hu, J. & Kivelson, S. A. Theory of electron nematic order in LaFeAsO. Phys. Rev. B 77, 224509 (2008).
Xu, C., Müller, M. & Sachdev, S. Ising and spin orders in the ironbased superconductors. Phys. Rev. B 78, 020501 (2008).
Fernandes, R. M. et al. Effects of nematic fluctuations on the elastic properties of iron arsenide superconductors. Phys. Rev. Lett. 105, 157003 (2010).
Fernandes, R. M., Chubukov, A. V. & Schmalian, J. What drives nematic order in ironbased superconductors? Nat. Phys. 10, 97 (2014).
Jang, H. et al. Ideal chargedensitywave order in the highfield state of superconducting YBCO. Proceed. Natl Acad. Sci. 113, 14645–14650 (2016).
Comin, R. et al. Symmetry of charge order in cuprates. Nat. Mat. 14, 796 (2015).
Kim, H.H. et al. Uniaxial pressure control of competing orders in a hightemperature superconductor. Science 362, 1040–1044 (2018).
Nie, L., Maharaj, A. V., Fradkin, E. & Kivelson, S. A. Vestigial nematicity from spin and/or charge order in the cuprates. Phys. Rev. B 96, 085142 (2017).
Hor, Y. S. et al. Superconductivity in Cu_{x}Bi_{2}Se_{3} and its implications for pairing in the undoped topological insulator. Phys. Rev. Lett. 104, 057001 (2010).
Kriener, M., Segawa, K., Ren, Z., Sasaki, S. & Ando, Y. Bulk superconducting phase with a full energy gap in the doped topological insulator Cu_{x}Bi_{2}Se_{3}. Phys. Rev. Lett. 106, 127004 (2011).
Fu, L. & Berg, E. Oddparity topological superconductors: theory and application to Cu_{x}Bi_{2}Se_{3}. Phys. Rev. Lett. 105, 097001 (2010).
Xia, Y. et al. Observation of a largegap topologicalinsulator class with a single Dirac cone on the surface. Nat. Phys. 5, 398 (2009).
Matano, K., Kriener, M., Segawa, K., Ando, Y. & Zheng, G.Q. Spinrotation symmetry breaking in the superconducting state of Cu_{x}Bi_{2}Se_{3}. Nat. Phys. 12, 852 (2016).
Yonezawa, S. et al. Thermodynamic evidence for nematic superconductivity in Cu_{x}Bi_{2}Se_{3}. Nat. Phys. 13, 123 (2017).
Pan, Y. et al. Rotational symmetry breaking in the topological superconductor Sr_{x}Bi_{2}Se_{3} probed by uppercritical field experiments. Sci. Rep. 6, 28632 (2016).
Du, G. et al. Superconductivity with twofold symmetry in topological superconductor Sr_{x}Bi_{2}Se_{3}. Sci. ChinaPhys. Mech. Astron. 60, 037411 (2017).
Shen, J. et al. Nematic topological superconducting phase in Nbdoped Bi_{2}Se_{3}. npj Quantum Mater. 2, 59 (2017).
Asaba, T. et al. Rotational Symmetry Breaking in a Trigonal Superconductor Nbdoped Bi_{2}Se_{3}. Phys. Rev. X 7, 011009 (2017).
Smylie, M. P. et al. Superconducting and normalstate anisotropy of the doped topological insulator Sr_{0.1}Bi_{2}Se_{3}. Sci. Rep. 8, 7666 (2018).
Tao, R. et al. Direct visualization of the nematic superconductivity in Cu_{x}Bi_{2}Se_{3}. Phys. Rev. X 8, 041024 (2018).
Yonezawa, S. Nematic superconductivity in doped Bi_{2}Se_{3} topological superconductors. Condens. Matter 4, 2 (2019).
Smylie, M. P. et al. Robust oddparity superconductivity in the doped topological insulator Nb_{x}Bi_{2}Se_{3}. Phys. Rev. B 96, 115145 (2017).
Hecker, M. & Schmalian, J. Vestigial nematic order and superconductivity in the doped topological insulator Cu_{x}Bi_{2}Se_{3}. npj Quantum Mater. 3, 26 (2018).
Larkin A. I. and Varlamov, A. A. in The Physics of Superconductors, Vol. 1, edited by K.H. Bennemann and J. B. Ketterson (Springer, Berlin, 2003)
Leggett, A. J. Quantum Liquids, Bose Condensation and Cooper Pairing in CondensedMatter Systems, Oxford University Press. (2006).
Kuntsevicha, A. Y. U. et al. Structural distortion behind the nematic superconductivity in Sr_{x}Bi_{2}Se_{3}. N. J. Phys. 20, 103022 (2018).
Böhmer, A. E. et al. Thermodynamic phase diagram, phase competition, and uniaxial pressure effects in BaFe2(As1−xPx)2 studied by thermal expansion. Phys. Rev. B 86, 094521 (2012).
Fu, L. Oddparity topological superconductor with nematic order: application to Cu_{x}Bi_{2}Se_{3}. Phys. Rev. B 90, 100509(R) (2014).
Venderbos, J. W. F., Kozii, V. & Fu, L. Identification of nematic superconductivity from the upper critical field. Phys. Rev. B 94, 094522 (2016).
Krotkov, P. L. & Mineev, V. P. Upper critical field in a trigonal unconventional superconductor: UPt_{3}. Phys. Rev. B 65, 224506 (2002).
Berg, E., Fradkin, E. & Kivelson, S. A. Charge4e superconductivity from pairdensitywave order in certain hightemperature superconductors. Nat. Phys. 5, 830 (2009).
Agterberg, D. F. & Tsunetsugu, H. Dislocations and vortices in pairdensitywave superconductors. Nat. Phys. 4, 639 (2008).
Babaev, E., Sudbø, A. & Ashcroft, N. W. A superconductor to superfluid phase transition in liquid metallic hydrogen. Nat. (Lond.) 431, 668 (2004).
DeGennes, P. G. The Physics of Liquid Crystals (Oxford University Press) (1974).
Sun, Y. et al. Quasiparticle evidence for the nematic state above T _{c} in Sr_{x}Bi_{2}Se_{3}. Phys. Rev. Lett. 123, 027002 (2019).
Kuntsevich, A. Y. U. et al. Straindriven nematicity of oddparity superconductivity in Sr_{x}Bi_{2}Se_{3}. Phys. Rev. B 100, 224509 (2019).
Kostylev, I., Yonezawa, S., Wang, Z., Ando, Y., Maeno, Y. Uniaxialstrain Control of Nematic Superconductivity in Sr_{x}Bi_{2}Se_{3}. Preprint at https://arxiv.org/abs/1910.03252 (2019).
Wang, J. et al. Investigate the Nb doping position and its relationship with bulk topological superconductivity in Nb_{x}Bi_{2}Se_{3} by Xray photoelectron spectra. J. Phys. Chem. Solids 137, 109208 (2020).
Rietveld, H. M. A Profile Refinement Method for Nuclear and Magnetic Structures. J. Appl. Cryst. 2, 65 (1969).
RodríguezCarvajal, J. Recent advances in magnetic structure determination by neutron powder diffraction. Phys. B 192, 55–69 (1993).
Gozzo, F. et al. The instrumental resolution function of synchrotron radiation powder diffractometers in the presence of focusing optics. J. Appl. Cryst. 39, 347–357 (2006).
Hou, J. et al. Possible coexistence of doubleQ magnetic order and chequerboard charge order in the reentrant tetragonal phase of Ba_{0.76}K_{0.24}Fe_{2}As_{2}. Phys. C. 539, 30 (2017).
Lyon, K. G., Salinger, G. L. & Swenson, C. A. Linear thermal expansion measurements on silicon from 6 to 340 K. J. Appl. Phys. 48, 865 (1977).
Acknowledgements
We thank U. Lampe for the technical support and acknowledge enlightening discussions with I. R. Fischer, C. Meingast, K. T. Law, and K. Willa. This work was supported by grants from the Research Grants Council of the Hong Kong Special Administrative Region, China (GRF16302018, SBI17SC14, IEG16SC03). J.S. was supported by the Gordon and Betty Moore Foundation’s EPiQS Initiative through Grant GBMF4302 and GBMF8686 while visiting the Geballe Laboratory for Advanced Materials at Stanford University. J.S. also acknowledges support by the German Research Foundation (DFG) through the Collaborative Research Center CRC TRR 288 “Elastic Tuning and Response of Electronic Quantum Phases of Matter”, project B01. Y.S.H. acknowledges the support from the NSFDMR 1255607.
Author information
Authors and Affiliations
Contributions
This work was initiated by R.L., J.Y.S. carried out the magnetotransport measurements with help of M.B., C.w.C., Q.C., and R.L. carried out the thermal expansion measurements with help of O.A., C.w.C., and R.L. carried out the specific heat and DC magnetization measurements with help of J.L.; the Nb_{x}Bi_{2}Se_{3} single crystal samples were provided by S.H.L. and Y.S.H., the Cu_{x}Bi_{2}Se_{3} single crystal sample was provided by E.P., M.H. and J.S. provided the theoretical simulations and further theoretical support. The Xray characterization of the sample was done by D.J.G. and J.Y.S., the manuscript was prepared by R.L. and J.S. with help of C.w.C. and M.H. All authors were involved in discussions and contributed to the manuscript.
Corresponding author
Ethics declarations
Competing interests
The authors declare no competing interests.
Additional information
Peer review information Nature Communications thanks GuoZhu Liu and the other, 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
Cho, Cw., Shen, J., Lyu, J. et al. Z_{3}vestigial nematic order due to superconducting fluctuations in the doped topological insulators Nb_{x}Bi_{2}Se_{3} and Cu_{x}Bi_{2}Se_{3}. Nat Commun 11, 3056 (2020). https://doi.org/10.1038/s41467020168719
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41467020168719
This article is cited by

Emerging symmetric strain response and weakening nematic fluctuations in strongly holedoped ironbased superconductors
Nature Communications (2021)

State with spontaneously broken timereversal symmetry above the superconducting phase transition
Nature Physics (2021)

Misfit phase (BiSe)1.10NbSe2 as the origin of superconductivity in niobiumdoped bismuth selenide
Communications Materials (2020)
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.