We performed resistivity and magnetic susceptibility measurements on CsV₃Sb₅ under pressure to track the evolution of the CDW state and superconductivity in the compressed material. Temperature dependence of resistivity for CsV₃Sb₅ under various pressures is shown in Fig. 1. As shown in Fig. 1a, an anomaly due to the CDW transition in the resistivity¹⁵ is clearly visible for sample 1 with PCC. The CDW transition temperature T* gradually decreases with increasing the pressure, and the anomaly becomes much weaker at high pressures. We can determine the CDW transition temperature T* precisely in the derivative resistivity curves as shown in Fig. 1b. The anomaly clearly shifts to lower temperatures with increasing the pressure, and disappears at the pressure of P2 ≈ 2 GPa. The anomaly of dρ_xx/dT at T* gradually evolves from a broad peak to a peak-dip feature around P1. Further increasing the pressure above P1, the peak gradually weakens and the dip becomes more pronounced. The dρ_xx/dT anomaly is associated to the change of the band structure and electron scattering rate in the CDW state. Therefore, such change of the shape of dρ_xx/dT anomaly indicates the possible change of the CDW transition around P1. To track the evolution of T_c with pressure, the low-temperature resistivity of CsV₃Sb₅ under various pressures is shown in Fig. 1c, d for sample1 with PCC and sample 2 with DAC, respectively. The T_c first increases with increasing pressure as shown in Fig. 1c, however, the superconducting transition becomes much broader with pressure larger than P1 of ~0.7 GPa, reminiscent of an inhomogeneous superconductivity in which multi superconducting phases coexists. In the high-pressure magnetic susceptibility measurements for sample 3 with PCC as shown in Supplementary Fig. 2, the superconducting transition temperature \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${T}_{c}^{M2}$$\end{document}TcM2 determined by the magnetic susceptibility gradually increases with increasing the pressure below P1. With the pressure above P1, the superconducting volume fraction suddenly decreases. Further increasing the pressure, the bulk superconductivity \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${T}_{c}^{M2}$$\end{document}TcM2 emerges below 4 K around 1.1 GPa, but the magnetic susceptibility shows a weak reduction at \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${T}_{c}^{M1}$$\end{document}TcM1 with higher temperature, indicating an inhomogeneous superconductivity, consistent with our resistivity measurements. It is striking that the superconducting transition becomes sharp again around P2. Meanwhile the highest T_c of 8 K is obtained. It should be addressed that the maximum T_c achieved at high pressure is 3 times higher than that (2.7 K) at ambient pressure. With further increasing the pressure, the T_c is monotonically suppressed and completely disappears above 12 GPa as shown in Fig. 1d for sample 2 with DAC. The pressure inhomogeneity or pressure induced disorder may cause the broadening of the superconducting transition. Such extrinsic effects become more pronounced with increasing the pressure. However, in our case, the superconducting transition at P2 is very sharp, which indicates that such extrinsic effect is negligible at least below P2 and the broadening of the superconducting transition between P1 and P2 is intrinsic.Fig. 1

a Temperature dependences of resistivity under various pressures for sample 1 measured with PCC. b The derivative dρ_xx/dT curves under various pressures. The blue arrows indicate the CDW transition temperature T*. c, d The evolution of superconducting transition temperatures under pressure for sample 1 with PCC and sample 2 with DAC. The red and green arrows represent the \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${T}_{c}^{{onset}}$$\end{document}Tconset and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${T}_{c}^{{zero}}$$\end{document}Tczero, respectively. All the curves were shifted vertically for clarity.

Combing the high-pressure electrical and magnetic susceptibility measurements on sample 1 and 3 with PCC and sample 2 with DAC, we can map out the phase diagram of CsV₃Sb₅ with pressure, as shown in Fig. 3a, b. The CDW is monotonically suppressed with increasing pressure, and the maximum T_c locates at the endpoint of CDW. A subsequent monotonic reduction of T_c is followed at higher pressure. The relatively high T_c (8 K) achieved at high pressure makes it the record in the kagome lattice materials. Such competition between CDW state and superconductivity is usual since the gap opening at CDW state would dramatically reduce the density of states at Fermi surfaces, leading to the suppression of superconductivity within Bardeen-Cooper-Schrieffer (BCS) scenario. It is striking that the superconducting transition becomes much broad with pressure between P1 and P2, and T_c together with H_c2 are strongly suppressed. The T* shows a weak anomaly around P1, which suggests the transformation of CDW state. Above P1, the superconductivity shows much stronger competition with CDW order, which will be discussed later. We can estimate the H_c2 at zero temperature (H_c2(0)) by using the two-band model²² of H_c2 vs T curves as shown in Fig. 2c, d. The pressure dependence of the extracted H_c2(0) is shown in Fig. 3c, and two peaks are clearly located at P1 and P2. The magnitude of H_c2(0) at P1 and P2 is one order larger than that at ambient pressure.Fig. 3

a Phase diagram of CsV₃Sb₅ with pressure. CDW transition temperature T* gradually suppressed with increasing the pressure. T* shows an anomaly around P1. The color inside the CDW phase represents the magnitude of magnetoresistance measured at 9 T and 10 K. b Pressure dependence of superconducting transition temperatures \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${T}_{c}^{{onset}}$$\end{document}Tconset, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${T}_{c}^{{zero}}$$\end{document}Tczero, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${T}_{c}^{M1}$$\end{document}TcM1 and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${T}_{c}^{M2}$$\end{document}TcM2 measured on various samples. T_c is strongly suppressed with the pressure between P1 and P2. c Pressure dependence of upper critical field at zero temperature, two peaks can be observed at P1 and P2.

The pressure dependence of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${T}_{c}^{{zero}}$$\end{document}Tczero clearly shows that two peaks are located at P1 and P2, and the superconducting transition width ΔT_c shows sudden enhancement with pressure between P1 and P2, as shown in Fig. 4a. In order to unveil the unusual suppression of superconductivity between P1 and P2, we plot the residual resistivity and residual-resistivity ratio (RRR) as a function of pressure, as shown in Fig. 4b. The residual resistivity suddenly increases around P1 and keeps at a relatively high value with the pressure between P1 and P2. Above P2, the residual resistivity suddenly drops. RRR also exhibits reduction with pressure between P1 and P2. The magnetoresistance (MR) measured under magnetic field of 9 Tesla and at the temperature of 10 K suddenly drops around P1. The MR measured at 3 T shows an anomaly at P1 and sudden drops at P2. In addition, the magnetoresistance of the low-field region at 10 K evolves from “V” shape to “U” shape at P2 as shown in Supplementary Fig. 4. Temperature dependence of MR at high pressure also evolves from “V” shape to “U” shape as shown in Supplementary Fig. 5. These results indicate that the low-field linear MR is an intrinsic property of the CDW phase, since the MR exhibits quadratic temperature-dependence when the CDW is suppressed by both of heating and pressure. However, the room-temperature resistivity gradually decreases with increasing the pressure, and does not show any anomaly at P1 and P2 as shown in Fig. 4c. These results indicate that the transitions at P1 and P2 are related to the CDW transitions rather than the normal state change at high temperature. The Shubnikov-de Haas (SdH) quantum oscillations (QOs) measurements at 2 K exhibit a rapid damping above P1 as shown in Supplementary Fig. 6, which is possibly associated to the enhancement of scattering rate for the resolved four bands, providing evidence for a transformation of the CDW state at P1.Fig. 4

a Pressure dependence of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${T}_{c}^{{zero}}$$\end{document}Tczero and ΔT_c for sample 1 with PCC. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${T}_{c}^{{zero}}$$\end{document}Tczero exhibits two peaks at P1 and P2. b Pressure dependence of residual resistivity and residual resistivity ratio (RRR) for sample 1 with PCC. c Pressure dependence of magnetoresistance (9 T and 3 T, 10 K) and room-temperature resistivity.

Single crystals of CsV₃Sb₅ were synthesized via a self-flux growth method similar to the previous reports¹⁵. In order to prevent the reaction of Cs with air and water, all the preparation processes were performed in an argon glovebox. After reaction in the furnace, the as-grown CsV₃Sb₅ single crystals are stable in the air. The excess flux is removed using water and millimeter-sized single crystal can be obtained.

Piston cylinder cell (PCC) was used to generate hydrostatic pressure up to 2.4 GPa for high-pressure electrical transport measurements. Daphne 7373 was used as the pressure transmitting medium in PCC. The pressure values in PCC were determined from the superconducting transition of Sn³⁶. Diamond anvil cell (DAC) was used to generate the pressure up to 12 GPa. Diamond anvils with 500 µm culet and c-BN gasket with sample chambers of diameter 200 µm were used. Four Pt wires were adhered to the sample and NaCl was used as the pressure transmitting medium. Pressure was calibrated by using the ruby fluorescence shift at room temperature³⁷. Electrical transport measurements were carried out in a Quantum Design physical property measurement system. High-pressure magnetic susceptibility measurements were performed in a miniature PCC using a SQUID magnetometer (MPMS-5 T, Quantum Design).

Note added: During the preparation of this manuscript, we noticed a similar high-pressure work³⁸.