Bi-2212/1T-TaS2 Van der Waals junctions: Interplay of proximity induced high-Tc superconductivity and CDW order

Understanding the coexistence, competition and/or cooperation between superconductivity and charge density waves (CDWs) in the transition metal dichalcogenides (TMDs) is an elusive goal which, when realized, promises to reveal fundamental information on this important class of materials. Here, we use four-terminal current-voltage measurements to study the Van der Waals interface between freshly exfoliated flakes of the high-T c superconductor, Bi-2212, and the CDW-dominated TMD layered material, 1T-TaS2. For highly transparent barriers, there is a pronounced Andreev reflection feature providing evidence for proximity-induced high-T c superconductivity in 1T-TaS2 with a surprisingly large energy gap (~20 meV) equal to half that of intrinsic Bi-2212 (~40 meV). Our systematic study using conductance spectroscopy of junctions with different transparencies also reveals the presence of two separate boson modes, each associated with a “dip-hump” structure. We infer that the proximity-induced high-T c superconductivity in the 1T-TaS2 is driven by coupling to the metastable metallic phase coexisting within the Mott commensurate CDW (CCDW) phase and associated with a concomitant change of the CCDW order parameter in the interfacial region.

CDW (ICCDW) phase. Interestingly, in the pressure experiments on pristine 1T-TaS 2 , the low-T c superconductivity (T c ≈ 5 K) persists until the pressure is high enough to convert the CDW phase to a metal 9 .
The TMD, 1T-TaS 2 , is a particularly interesting CDW material for such studies because it exhibits a pronounced first order CDW transition with hysteretic resistance transitions in the 180-230 K temperature range. It is generally recognized that electron-electron as well as electron-phonon interactions in 1T-TaS 2 13-16 are responsible for the evolution of a nearly commensurate CDW (NCCDW) to a Mott commensurate CDW (CCDW) phase dominating at low temperatures. Two opposing arguments focus on where and how low T c superconductivity forms in 1T-TaS 2 . The first argues that the superconductivity is formed within the metallic interdomain spaces separating the CCDW domains where tens of "star of David" clusters clump into rough hexagonal domains reproducing the Mott-CCDW phase locally 9,10 . The second argues that the superconductivity, characterized by a shallow electron pocket at the Brillouin-zone center, is formed exactly within the clusters of stars 11 in the NCCDW phase in real space. We note that low-T c superconductivity at ambient pressure is only found in non-pristine (i.e., doped) 1T-TaS 2 where the NCCDW phase dominates.
In this study, we find evidence for proximity-induced high-T c superconductivity in the topmost layers of pristine 1T-TaS 2 at the interface of Van der Waals bonded Bi-2212/1T-TaS 2 junctions, where pristine Bi-2212 is the high-T c superconductor Bi 2 Sr 2 CaCu 2 O 8+δ with a transition temperature T c = 85 K and intrinsic energy gap Δ 0 = 40 meV. Andreev reflection, marked by excess current and a wide zero-bias conductance peak, is observed at temperatures up to 80 K where the 1T-TaS 2 is in the Mott-CCDW state. The proximity induced gap Δ a in the 1T-TaS 2 is found to be a surprisingly large 20 meV, thereby implying a strong coupling limit (2Δ sc /k B T c ~ 5.8) instead of the BCS weak coupling limit of 3.5. In addition to the induced gap in the 1T-TaS 2 , we also observe a depressed gap compared to the intrinsic gap Δ 0 in the vicinity of the interface on the Bi-2212 side.
Our observation of a transparency-dependent superconducting proximity effect of Bi-2212/1T-TaS 2 junctions strongly indicates that superconductivity is induced in a metallic phase of 1T-TaS 2 . The transparency is defined as a dimensionless normal conductance, σ = + N Z 1 1 2 , where the parameter Z is extracted from theoretical fitting of conductance spectroscopy data using the extended BTK model described below. Such a finding is somewhat surprising since, in the absence of intimately contacting Bi-2212, pristine 1T-TaS 2 is in an insulating CCDW state, and a proximity effect is only expected to occur in metallic systems. The puzzle here is that at the temperatures where Bi-2212 is superconducting, the stoichiometric pure 1T-TaS 2 is in the nonmetallic Mott-CCDW state and a proximity effect is in fact observed thereby implying that the proximity of the Bi-2212 induces changes in the CDW order parameter to achieve a more metallic phase. Said in another way, the mutual interaction of the CDW and superconducting order parameters is such that the proximity-induced gap in the 1T-TaS 2 can only appear if the CDW order parameter is changed in the interfacial region as is the superconducting gap associated with the Bi-2212. This interpretation is consistent with evidence for a metastable metallic phase, induced by voltage pulses, laser pulses or current excitations, residing within the Mott-CCDW phase [17][18][19][20][21] . This metastable state is ascribed to the reduction of onsite Coulomb interaction U and increase of Hubbard band width W via phase shifts of the CDW order parameter or interplay between the electron-electron and electron-phonon interactions in the topmost layers [17][18][19][20][21] .
The superconducting proximity effect, attributed to the leakage of Cooper pairs into a conducting metallic phase of the 1T-TaS 2 , is also confirmed by the good agreement of our c-axis conductance spectroscopy (dI/dV vs. V) with an extended BTK model for d-wave superconductors 22 . Additionally, our conductance measurements reveal the presence of two dip-hump structures which can be interpreted to reflect an inherent electron-phonon interaction in 1T-TaS 2 that assists the formation of high-T c superconductivity in the metastable metallic domains residing within the Mott-CCDW phase. This evidence of electron-phonon interaction assisted high-T c superconductivity within the Mott CCDW phase of 1T-TaS 2 , presents a new paradigm for understanding the correlation of CDW order and high-T c superconductivity in 1T-TaS 2 .

Results
Characteristics of Samples. Nearly optimally doped crystals of high-T c cuprate Bi 2 Sr 2 Ca 1 Cu 2 O 8+δ (Bi-2212) and stoichiometically pure layered transition metal dichalcoginide (TMD) 1T-TaS 2 crystals were used. The critical superconducting temperature T c of Bi-2212 (Supplementary Fig. S.1(a)) and the CCDW-NCCDW phase transition temperatures on cooling/warming processes (Supplementary Fig. S.1(b)) were verified with AC transport measurements to be at 85 K and 180 K/230 K respectively.
The Bi-2212 and 1T-TaS 2 crystals were mechanically exfoliated in a dry atmosphere as thin flakes with approximately rectangular shapes and nominal thicknesses of 0.5-2.0 μm and 2.0-5.0 μm for the Bi-2212 and 1T-TaS 2 flakes respectively. The cleaved surfaces were clean and flat, with a mean roughness of 1.55 Å for Bi-2212 and 1.52 Å for 1T-TaS 2 from AFM images as shown in Fig. 1(c). Two cleaved thin flakes were placed against each other and naturally bonded via Van der Waals forces 23 ; high quality normal metal-superconductor (NS) or normal metal-insulator-superconductor (NIS) junctions were then formed. Four terminal tunnel junction configurations with perpendicularly oriented top and bottom electrodes shown in the Fig. 1(d) schematic have the advantage that contact resistances are eliminated and the active area common to both electrodes (0.2 mm 2 ) can be accurately calculated 24,25 . Advantageously the c-axis of both the bottom (Bi-2212) and top (1T-TaS 2 ) electrodes are perpendicular to the substrate and thus colinear, thereby minimizing currents flowing along the ab-plane. Our c-axis conductance spectroscopy measurements were performed in a Quantum Design Physical Properties Measurement System (PPMS) at temperatures ranging from 2.5 K to 120 K. Using this set up we find the intrinsic Bi-2212 superconducting gap to be in the range 38  Experimental measurements of Bi-2212/1T-TaS 2 junctions. Highly transparent junctions. C-axis AC differential conductance spectroscopies and DC current-voltage (I-V) characteristics were measured on Bi-2212/1T-TaS 2 junctions. To exemplify the superconducting proximity effect at the interface, junctions with high transparency 1, 3, 28, 29 or low Z parameter described in the BTK model 29 are needed. Under these conditions, Cooper pairs in the superconductor (S) can leak into the normal material (N) side resulting in a spatially dependent superconducting gap Δ(x) that extends across the interface along the c-axis and into the 1T-TaS 2 . Various gap features can be identified beginning with the intrinsic superconducting gap Δ 0 = 40 mV deep in the superconductor which decreases to a depressed superconducting gap Δ p = 28 mV at the S/N interface and then decreases to a proximity effect induced superconducting gap Δ a = 20 mV on the N side of the interface which decreases in magnitude with increasing distance from the interface 1, 3 . Moreover, the different proximity region widths within which these gaps exist depend sensitively on the transparency of junctions 1, 3, 30-32 . In addition, different quasi-particle lifetimes or scattering rates at the vicinity of interface might affect the spectroscopic line shapes 33 .
DC current-voltage I-V curves of Junction-1, with BTK parameter Z = 0.25, show clear zero-bias excess current relative to the normal current (indicated by the red vertical arrows labeling the induced gap, ±Δ a ) as shown in Fig. 2(a) for temperatures from 5 K to 100 K. The slope within the excess current range from −20 meV to 20 meV at 5 K, as indicated by the solid gray triangle, is nearly twice the slope within the normal current range indicated by the solid hollow triangle. At temperatures well below the T c (~85 K) of Bi-2212, the excess current in the vicinity of zero bias is approximately twice the normal current, and therefore attributable to Andreev reflection at the interface 29 , where the electrons injected from N side are reflected as holes with time reversal symmetry tracing the injected electrons' track back into N side. To conserve current across the interface, Cooper pairs flow at the Fermi energy within S. These Cooper pairs have electron-like (ELQ) and hole-like (HLQ) quasiparticle character above/below the edges of the energy gap and the conversion process only occurs when the incident electrons on the N side lie within the conductance plateau defined over the energy range ±Δ a shown in the normalized differential conductance plot of Fig. 2(b).
The two additional nonlinear features marked as blue and magenta vertical arrows in Fig. 2 are attributed respectively to the two superconducting gaps Δ p and Δ 0 mentioned at the beginning of this section. The AC differential conductance spectroscopies (dI/dV) S of Junction-1, normalized by the normal state conductance (dI/dV) N at 100 K, at various temperatures from 5 K to 100 K shown in Fig. 2(b) confirm the features observed from I-V curves. The proximity effect induced superconducting gap Δ a = ±20 meV at 5 K on the 1T-TaS 2 side of  25, at various temperatures from 5 K to 100 K. Δ a (Red), Δ p (Blue) and Δ 0 (Magenta) respectively represent the induced superconducting gap on 1T-TaS 2 side at interface, the depressed superconducting gap on Bi-2212 side at interface, and the intrinsic superconducting gap of Bi-2212. The hollow solid triangle indicates the normal current of the junction, where the normal conductance at low temperatures in the superconducting state is approximately equal to that of the normal state. The solid gray triangle indicates the excess current due to Andreev reflection at N-S interface, where the conductance within the Andreev reflection region amounts to nearly twice the normal state conductance well below T c . (b) Normalized AC differential conductance dI dV dI dV ( / ) /( / ) S N for Junction-1, where the differential conductance at 100 K is considered as the normal state conductance. The curves are shifted for clarity. (c) DC I-V curves of Junction-1 (Z = 0.25, black) and Junction-2 (Z = 0.27, blue) at 5 K. The induced superconducting gap Δ a and excess current within ±Δ a of Junction-2 are slightly lower relative to Junction-1. The normal conductance of Junction-2, around 0.62 μA/mV, is lower than the normal conductance around 1.85 μA/mV for Junction-1. (d) AC differential conductance dI dV dI dV ( / ) /( / ) S N for Junction-2, normalized by the differential conductance at 100 K. Curves are shifted for clarity. Note: for panels (b,d), the orange/purple markers and arrows are discussed in the text.
Scientific RepoRts | 7: 4639 | DOI:10.1038/s41598-017-04645-1 the interface delineates the voltage region within which Andreev reflection occurs. Additional gap features, relevant to the nonlinear features observed from I-V curves, are believed to be the density of states (DOS) features corresponding to the depressed superconducting gap Δ p on the Bi-2212 side at the interface and the intrinsic superconducting gap Δ 0 of Bi-2212. For Junction-1, the sizes of the two gaps are Δ p = 28 meV and Δ 0 = 40 meV respectively at 5 K. As temperature increases, the zero-bias conductance peak evolves from a flat mesa to a rounded hump while the peak's width and height are depressed up to temperatures near 80 K. The gaps, Δ p and Δ 0 merge with increasing temperature until all gaps disappear near 80 K. Similar features are also observed on Junction-2 with BTK parameter Z = 0.27, as shown in Fig. 2(c) inset for DC I-V curves and (d) for AC differential conductance spectroscopies.
Both Junction-1 and Junction-2 show characteristics of highly transparent junctions, exhibiting high normal state conductance of 1.85 μA/mV and 0.62 μA/mV at 5 K respectively as shown in Fig. 2. As revealed by the conductance spectroscopies of Junction-2 (seen in Fig. 2(d)), the sizes of the three gaps Δ 0 = 39 mV, Δ p = 23 mV and Δ a = 16 mV at 5 K are close to the values of Junction-1. The consistency of the Bi-2212 intrinsic superconducting gap size measured for Junction-1 and Junction-2 with previous c-axis point-contact tunneling studies on single crystal Bi-2212 26, 27 is satisfying. Any variation of the differential conductance spectroscopies' line-shape for Junction-1 and Junction-2 might be ascribed to the discrepancy of proximity regions' width and scattering rate of quasiparticles at the vicinity of the interface 1, 3, 30-33 .
Andreev reflection such as observed here for our low-Z junctions (Junction-1 and Junction-2) exists at the normal metal-superconductor (N-S) interface, thereby indicating that the observed Andreev reflection feature on our Bi-2212/1T-TaS 2 junctions implies at minimum a metallic component on the N side. We conclude that the metastable metallic phase of 1T-TaS 2 residing in the Mott-CCDW state at low temperatures and revealed by STM studies 17-21 is the requisite metallic state for proximity coupling to pristine 1T-TaS 2 , where the metallic phase has a smaller parameter U/W than the Mott-CCDW phase. Importantly, the Andreev enhanced zero-bias peak disappearing at high temperature around 80 K in Junction-1 and Junction-2 reflects proximity induced high-T c superconductivity in the metastable metallic phase residing in the layers of 1T-TaS 2 which are adjacent to the Bi-2212.
Besides the evidence for a robust superconducting proximity effect existing in our two low-Z Bi-2212/1T-TaS 2 junctions, we find two dip-hump structures positioned at U e-ph and − ⁎ U e ph for both junctions, which are indicated at their respective peaks (humps) by orange diamond and purple inverted triangles in Fig. 2(b,d). As suggested by neutron resonance and ARPES results 34,35 , a single dip-hump structure in Bi-2212 probably stems from the electron-boson coupling, or combined electron-boson coupling and pseudogap 36 corresponding to a boson mode energy Ω. As the fingerprint of a boson mode, the energy at the maximum slope of the dip-hump structure in conductance spectroscopies or the peak/dip in the dI 2 /dV 2 spectra at positive/negative bias regions for NIS junctions is positioned at E p = Δ 0 + Ω [37][38][39][40] . Indicated by orange and purple arrows for the lowest temperatures in Fig. 2(b,c) respectively in the conductance spectroscopies, the determination of experimental values of E p and ⁎ E p from dI 2 /dV 2 spectra are described in Fig. S.4 of Supplementary information. Consequently, with the information summarized in Table 1  For intrinsic Bi-2212, the boson mode energy found in STM spectra is around 52 ± 8 meV 41 , leading us to conclude that the hump peaked at U e-ph is from intrinsic Bi-2212 accompanied by a boson mode energy Ω, while the hump peaked at − ⁎ U e ph is a heretofore unseen hump due to another boson mode with lower energy Ω*. As temperature increases, the two humps merge to a single hump near 50-60 K for both junctions, a temperature somewhat more than the temperature near 40 K where Δ 0 and Δ p merge as shown in the differential conductance spectroscopies in Fig. 2(b,d).
Two boson modes. Our Bi-2212/1T-TaS 2 Junction-3, with slightly lower transparency (BTK parameter Z = 0.5), offers more information about the dip-hump structures observed in Junction-1 and Junction-2. As shown in  Table 1. Parameters used in calculation (marked as "theory") or from measurements (marked as "exp") of the conductance spectroscopies at 5 K for various junctions. Z: BTK parameter; σ  N : normal conductance from measurement; Δ 0 : Bi-2212 intrinsic superconducting gap; Δ p : suppressed superconducting gap on S side; Δ a : proximity induced superconducting gap on N side; Γ 0 : quasiparticle lifetime parameter on S side; Γ p : quasiparticle lifetime parameter at interface on S side; Γ a : quasiparticle lifetime parameter at interface on N side;  Besides the nonlinear feature of the Bi-2212 intrinsic superconducting gap Δ 0 indicated by the magenta vertical arrow, the other two nonlinear features taking place above Δ 0 , as indicated by the purple and orange vertical arrows, are also clearly observed. Junction-3 with lower normal conductance around 0.21 μA/mV at 5 K than Junction-1 and Junction-2, however, demonstrates the normal conductance varies with temperatures indicating the 1T-TaS 2 for Junction-3 might behave more like a Mott-CCDW material rather than the metastable metallic material in the more transparent junctions. The temperature dependence of normal conductance in Junction-3, referring to a less metallic NS junction, probably originates from the carrier delocalization in the Mott-insulating state in 1T-TaS 2 with increasing temperature [13][14][15][16] .
The normalized AC differential conductance spectroscopies dI dV dI dV ( / ) /( / ) S N of Junction-3, obtained using the method described in the Methods section, at various temperatures from 3 K to 100 K are shown in Fig. 3(b). A strongly suppressed amplitude of the zero-bias peak for Junction-3 relative to Junction-1 and Junction-2 is observed. Such strong suppression suggests a stronger scattering rate or shorter quasiparticle lifetimes in the proximity-effect-induced superconducting region on the 1T-TaS 2 side at interface. The width of the zero-bias peak decreases with increasing temperature, and is totally suppressed near 40 K. Meanwhile, the feature representing the depressed superconducting gap Δ p is clearly not observed, as seen in Fig. 3(b), which we ascribe to the S N for Junction-3. The differential conductance at 100 K manually scaled into the range of the conductance at one specific temperature serves as the normal conductance. Relative to the NS Andreev reflection feature seen in Junction 1 and 2, the zero-bias excess differential conductance peak is significantly reduced and almost disappears at around 40 K. The Bi-2212 intrinsic superconducting gap Δ 0 and the humps referring to boson modes with energies Ω* and Ω are indicated respectively by solid magenta ellipses, inverted purple triangles and orange diamonds. Positions of E p and ⁎ E p are marked by orange and purple arrows. The curves are shifted for clarity. (c) DC I-V curves of Junction-3 at various temperatures from 40 K to 100 K. The conductance in the normal state at 100 K is around 0.4 μA/mV. The temperature dependence of the conductance for Junction-3 evolves consistently with intrinsic 1T-TaS 2 at temperatures from 3 K to 100 K shown in Fig. S.1(b). (d) Temperature dependence of the two humps' peak positions and the Bi-2212 intrinsic superconducting gap. closer proximity between Δ p and Δ a and temperature broadening. For Junction-3, the measured superconducting gap Δ 0 around 38 meV at 5 K is slightly smaller but still consistent with the highly transparent Junction-1 and Junction-2. Notably, the width of zero-bias peak is around 21 meV, which is not strongly depressed relative to Junction-1 and Junction-2.
Sharper and clearer signatures of two dip-hump structures are observed in Junction-3. The conductance spectroscopy indicates the hump (peak marked by the orange diamond) due to the boson mode energy Ω has a broader width than the hump (peak marked by the purple inverted triangle) due to the boson mode energy Ω*. The two humps of Junction-3 at 5 K which are peaked at the energy . More details about the temperature evolution (at temperatures below Bi-2212 T c ) of the Bi-2212 superconducting gap Δ 0 and two humps are presented in Fig. 3. Similar to the highly transparent Junction 1 and 2, the two humps merge toward each other with increasing temperature, and almost merge into a single hump at an energy position of = = − − ⁎ U U 39 meV e ph e ph at around 60 K. The zero-bias conductance peak is totally suppressed at a lower temperature around 40 K instead of 80 K as in Junction 1 and 2; however, still indicating proximity-induced high-T c superconductivity in 1T-TaS 2 .
Theoretical modeling and results. The superconducting proximity effect distinguishes two major regimes at the interface of NS or NIS junctions 1, 3 . The NS interface for electrons with energy E < Δ a injected from the N side is the boundary between the induced superconducting region and the normal material on the N side, whereas electrons within energy Δ a < E < Δ p and Δ p < E < Δ 0 will experience the NS interface from the S side. Two schemes of a proximity junction with thicknesses of the proximity regions on the S and N sides respectively marked as d S and d N are depicted in Fig. 4(a). Schemes Nos. 1 and 2 correspond respectively to the cases The c-axis conductance characteristics of various junctions, including Bi-2212/1T-TaS 2 and Bi-2212/graphite junctions, are calculated based on an extended BTK model of the tunneling spectrum for anisotropic superconductors 22 .  , the smearing effect on tunneling spectroscopy due to quasiparticle life time τ R or scattering rate 1/τ R is quantitatively described 33,45 .
The calculated conductance spectroscopies for various junctions show good agreement with measurement results, as shown in Fig. 4(b) for various junctions at 5 K, as well as the temperature-dependent conductance spectroscopies below Bi-2212 T c for Bi-2212/1T-TaS 2 and Bi-2212/graphite junctions seen in Fig. S.3(a-e). Figure 4(c) also compares the temperature dependent gaps (Δ 0 , Δ p and Δ a ) with theoretical calculations for various Bi-2212/1T-TaS 2 junctions. These calculations were only carried out up to 40 K because of high temperature smearing effects. The parameters used in calculation of the conductance characteristics at 5 K for various junctions are listed in Table 1. More details on theoretical modeling are described in Supplementary information (Section: Supplementary Discussion-Theoretical Modeling).

Discussion
The normalized differential conductance plots, as shown in Fig. 5(a), at temperatures from 5 K to 30 K for Bi-2212/1T-TaS 2 Junction-4 with BTK parameter Z = 0.8 exhibit the proximity effect described by Scheme No. 2 in Fig. 4(a). For this case there is no discontinuity in the gap at the interface (i.e., Δ p = Δ a ). For even lower transparency interfaces, such as Junction-5 with BTK parameter Z = 2.5 shown in Fig. 5(b), any remnants of a superconducting proximity effect with markers at Δ p and Δ a have disappeared. To briefly summarize, for relatively high transparent Bi-2212/1T-TaS 2 Junction 1-4, the superconducting proximity effect is clearly experimentally observed (seen in Figs 2, 3 and 5(a)) via c-axis conductance spectroscopies. The temperature dependence of the gaps Δ 0 , Δ p and Δ a for Junction 1-4, displayed in the four panels of Fig. 4(c), show a merging of Δ p and Δ a with decreasing transparency (or increasing Z) of the interface and increasing temperature.  5(a,b)), the normalized zero-bias conductance (NZBC) for Junction-1 with Z = 0.25 divided by the conductance at 100 K is calculated and shown in Fig. 6(a) as a function of temperature for the cooling cycle. At high temperatures above T c where the Bi-2212 is in the normal state, hysteresis dominates in the temperature range of 180 K to 230 K as it does in the four-terminal temperature-dependent resistance of a pristine 1T-TaS 2 flake shown in Supplementary Fig. S.1(b).
For additional insight, we use a back-to-back structured Ag/1T-TaS 2 /Ag trilayer junction to detect the perpendicular transport characteristic of a metal/1T-TaS 2 junction. The linear I-V characteristics are shown in Fig. S.5(c,d). As shown in Fig. 6(b), the temperature-dependent NZBC of Ag/1T-TaS 2 /Ag reveals the signature of the CDW transition in 1T-TaS 2 by a strong suppression of the NZBC when 1T-TaS 2 transits from a metallic NCCDW phase to a Mott-CCDW phase. In this case the metal/1T-TaS 2 junction is expected to include a metal/ semiconductor barrier owing to the opened Mott gap of 1T-TaS 2 at low temperatures. Thus, the observed hysteresis in Junction-1 suggests the existence of CDW transitions when Bi-2212 is intimately contacted to 1T-TaS 2 either as a superconductor or a normal metal. In both cases the 1T-TaS 2 has clearly converted to a Mott-CCDW phase at low temperatures. Moreover, the NZBC of Junction-1 in panel (a) has increased with decreasing temperature up to almost 2 times that of the referenced 100 K conductance due to the Andreev reflection contributions at base temperature 5 K. Accordingly, the superconducting proximity effect and the CDW transition are simultaneously present in the same sample. We also note an obviously suppressed amplitude of the hysteresis window for Junction-1 compared to the Ag/1T-TaS 2 /Ag junction. Such a signature of suppression reveals the more metallic nature of 1T-TaS 2 when in contact with Bi-2212.
Consequently, the superconducting proximity effect observed in our highly transparent Bi-2212/1T-TaS 2 junctions strongly suggests that high-T c superconductivity forms within the metastable metallic phase with a smaller parameter U/W, residing in the Mott-CCDW phases located in the topmost layers of 1T-TaS 2 . The different heights and widths of the zero bias peak on the conductance characteristics for various Bi-2212/1T-TaS 2 junctions with different transparencies might be ascribed to the Andreev reflection corresponding to different configurations of Mott-CCDW phase and metallic phase in 1T-TaS 2 (See Supplementary information for more discussion). The curved and depressed zero-bias conductance peak (seen in Figs 2(b,d) and 3(b)) corresponds to the smearing effect with increasing quasiparticle scattering rate 33,45 , Accordingly, it is no surprise that the quasiparticle lifetime parameters Γ a at the interface on the N side (see Table 1) are lower by more than a factor of ten for the high transparency Junction 1 and 2 than they are for the lower transparency Junction 3 and 4.
We rule out alternative interpretations with the following arguments: Firstly, the Bi-2212 intrinsic superconducting gap Δ 0 measured from various Bi-2212/1T-TaS 2 and Bi-2212/graphite junctions shows consistency of the results at temperatures well below Bi-2212 T c , with the BCS gap ratio 2Δ sc /k B T c to be around 10.4-11.5 (seen in Fig. 5(c)) in good agreement with previous works on intrinsic Bi-2212 26,27 . Moreover, less dependence of the proximity induced gap Δ a on temperature (seen in Fig. 4(c)) rules out a contribution to the differential conductance from Andreev bound states caused by planar geometry or surface roughness [46][47][48] . (Here, we claim the larger deviation of measured superconducting gaps among various junctions at higher temperatures close to the T c of Bi-2212 is probably due to distortion of gap features on conductance spectroscopies by stronger scattering and temperature smearing effects, which could be self-consistently verified by the smaller deviations of gap parameters used in theoretical results shown in Fig. S3(f)). Secondly, though there is a strong suppression of amplitude for the zero bias conductance peak for various junctions (almost two times the normal conductance in Junction-1 and Junction-2, whereas 30% larger than normal conductance in Junction-3), less temperature dependence of the wide induced gap Δ a (41% to 55%Δ 0 on 1T-TaS 2 side as shown in Fig. 4(c)) rules out the feature of the complex Andreev reflection due to the phase conjugation of electrons and holes predicted for superconductor-semiconductor interfaces 49 . Thirdly, there is no dominant feature of periodic conductance peaks observed in any of our Bi-2212/1T-TaS 2 junctions, which rules out the possibility that our observed proximity feature is related to McMillan-Rowell oscillation observed in some normal metal/cuprate junctions 31,[50][51][52][53] . Lastly, with a very short c-axis coherence length (~1 Å 54 ) in intrinsic Bi-2212, the absence of evidence for Josephson junctions intrinsically formed by superconducting CuO 2 and non-superconducting Bi-O and Sr-O layers 55,56 observed at temperatures well below T c rules out the explanation of the superconducting proximity effect masquerading as interlayer tunneling within the Bi-2212.
In Section IIB-2 we discussed the presence of two dip-hump structures, U e-ph and − ⁎ U e ph which when associated with peak/dip features in the dI 2 /dV 2 spectra at positive/negative bias regions defined respectively the energies E p and ⁎ E p . We then referenced these features to the intrinsic Bi-2212 gap edge Δ 0 using the relations Ω = − ∆ E p 0 and Ω = − ∆ ⁎ ⁎ E p 0 and find that the temperature-dependent boson mode energy Ω is consistent with the STM spectrum on intrinsic Bi-2212 41 . This picture is incomplete however without comparing the energy of the second feature referenced to the intrinsic gap, Ω = − ∆ ⁎ ⁎ E p 0 and alternatively referenced to the reduced gap at the interface, Δ p , using the relation Ω = − ∆ ⁎ E p p p , revealing a slightly larger energy scale than Ω* both of which are derived from ⁎ E p . For Junction 1-3 these three boson modes, Ω, Ω* and Ω p are plotted in Fig. 5(d) as a function of temperature.
Here, we provisionally assume the two dip-hump structures in conductance spectroscopies, observed in Junction-1-3, are due to self energy effects related to electron-boson interaction where phonons serve as the relevant bosons. We then infer that the "glue" assisting the strong-coupled pairing of electrons responsible for the high-T c superconductivity in 1T-TaS 2 arises from phonons. The key point, here, is whether the additional hump peaked at − ⁎ U e ph is the hump corresponding to the suppressed superconducting gap Δ p because of a boson mode energy Ω p representing the difference between the junction interface and bulk Bi-2212, or a hump related to intrinsic superconducting gap Δ 0 due to a boson mode energy Ω* incorporating the effect of interplay between electron-electron and electron-phonon interactions in 1T-TaS 2 on the density of states. For all junctions, the boson mode energy Ω at low temperatures is within the range of 40 meV to 60 meV (as indicated by the gray background) as well as being less temperature dependent. The results are consistent with the STM spectrum on intrinsic Bi-2212 41 . The yellow background demarcates the possible reduction (~4 meV 41 ) due to the substitution of 16 O by 18 O in Bi-2212 crystals. Slightly lower energy Ω p relative to Ω is observed in all Junction 1-3. On the other hand, all junctions reveal a nearly temperature-independent boson mode energy Ω* around 25 meV indicated by the horizontal black dashed line (except Junction-2 over limited temperatures). The magnitude of Ω* is in good agreement with the energy scale of an observed 25 meV infrared optical phonon 57 corresponding to a CDW near the same energy 58 that provides insight into the disorder-induced quasimetallic phase of 1T-TaS 2 residing in the Mott-CCDW phase. Additionally, the observed evolution of broader (Junction-1-2 as seen in Fig. 2) to sharper (Junction-3 as seen in Fig. 3) line-shape of the hump peaked at − ⁎ U e ph is consistent with current evidence and explanations of the transition from a Mott-insulating to a metallic phase in 1T-TaS 2 due to reduced Coulomb interaction U and broadened band width W of lower Hubbard band [17][18][19][20][21]58 .
In conclusion, we have used differential conductance spectroscopy of Bi-2212/1T-TaS 2 junctions with varying transparencies to find high-T c superconductivity induced within pristine 1T-TaS 2 by the proximity effect. The CDW order in the 1T-TaS 2 appears to play an important role firstly by coexisting with an unexpected and surprisingly high T c of the proximity gap in the 1T-TaS 2 and secondly by revealing a heretofore unseen ancillary dip-hump feature that accompanies a primary dip-hump feature corresponding to a boson (phonon) mode Ω previously seen in STM work on intrinsic Bi-2212. This second dip-hump feature is clearly related to the proximity of the CDW dominated 1T-TaS 2 and implies one of two possible boson modes depending on whether the dip-hump feature is referenced to the reduced gap Δ p or the intrinsic gap Δ 0 at the interface. In the former case the temperature-dependent boson mode Ω = − ∆ ⁎ E p p p has values in the range 30-45 meV, somewhat less than the 40-50 meV range shown in Fig. 5(c) for Ω. The energy Ω p can probably be interpreted as the mode relating to the gap Δ p in the same manner as the mode Ω is related to the intrinsic gap Δ 0 . In the latter case however the temperature dependent boson mode Ω = − ∆ ⁎ ⁎ E p 0 has values in the range 20-25 meV which is close to the infrared active phonon mode associated with the CDW in 1T-TaS 2 as measured by infrared reflectance 57 . This more plausible latter interpretation provides independent evidence that the phonon associated with the CCDW phase coexists with and may even enhance the superconductivity in pristine 1T-TaS 2 . However, both of the above scenarios describing the second dip-hump structure provide reasonable rationales for the occurrence of a high-T c proximity effect in Bi-2212/1T-TaS 2 junctions. Our work posits a mutual interaction of the CCDW and superconducting order parameters in the interfacial region of Bi-2212/1T-TaS 2 contacts, thereby revealing rich phenomenology and confirming a strong interplay between high T c superconductivity and CDW order which is only beginning to be understood.

Methods
Sample fabrication. High quality, optimally doped single crystals of Bi 2 Sr 2 CaCu 2 O 8+δ (Bi-2212) were synthesized using the method of Mitzi et al. 59 , with the modification that a Pt crucible was used in place of an alumina one. This avoided possible contamination of the melt via reaction with the crucible walls. Single crystals of 1T-TaS 2 flakes were prepared using iodine vapor transport 60 . The transport measurements shown in Fig. S1 verify the critical superconducting temperature of Bi-2212 to be 85 K, and the CCDW-NCCDW transition temperatures for cooling and warming processes of 1T-TaS 2 to be 180 K and 230 K respectively. Prior to junction fabrication, thick flake Bi-2212 (thickness around 10 μm was mechanically exfoliated via Nitto-REVALPHA thermal release tape. Then, thick flake Bi-2212 was transferred to a clean glass substrate by thin double sided tape. Thin flake Bi-2212 (0.5 to 2 μm) was then cleaved via Scotch tape. A cleaved thin flake of 1T-TaS 2 (2 to 5 μm) was immediately placed against the cleaved surface of Bi-2212 after similar mechanical exfoliation. The two cleaved flakes strongly adhere to each other via Van der Waals forces, naturally forming high quality NIS or NS heterostructures. Junctions with different transparencies, characterized with theoretical modeling by the extended BTK model described in the manuscript, were randomly achieved. No significant dependence of the transparency upon area of junction or thickness of flake was recognized. All steps of fabrication were performed in a dry atmosphere, and all cleaved surfaces of thin flake Bi-2212 and 1T-TaS 2 were clean and flat. Based on multiple measured local-areas on the exfoliated pieces, the mean roughness for both materials is within 2 Å as indicated by the AFM images shown in Fig. 1(c)). Also, to implement four-terminal c-axis differential conductance measurements, the thin flake Bi-2212 and 1T-TaS 2 were fashioned into narrow rectangular shapes oriented perpendicular to each other with an overlap area around 0.2 mm 2 , as shown in Fig. 1(d).
Experimental measurements set-up. All measurements including the four-terminal AC differential conductance measurements and the DC current-voltage (I-V) measurements were performed over a wide range of temperature (2.5 K to 120 K) using a Quantum Design Physical Properties Measurement System (PPMS). Samples were mounted on a commercial PPMS puck and all measurements were performed in a low-noise screen room. For AC measurements, we used 23.3 Hz as the AC output frequency. The DC source voltage was supplied by a Keithley 2400 source meter and the AC source was supplied by a Agilent 33120 A AC generator with ΔV ~ 0.2 mV. The DC and AC source signals were added using a homemade DC + AC adder and then applied to the junction. The DC bias across the junctions was measured with HP 3456A multimeter, the AC voltage signal with a SR830 DSP lock-in amplifier, and the AC current signal with a second SR830 lock-in amplifier after converting the current to a voltage using a SR570 current preamplifier.
Data normalization. For non-ideal NS junctions which have different density of states on the N and S sides, the conductance at one specific temperature cannot be simply normalized because of temperature dependence of the normal conductance. Thus, we normalized the AC differential conductance of non-ideal NS junctions, such as Junction-3, Junction-4 and Junction-5 discussed in the manuscript, by using the conductance at 100 K manually scaled into the range of the normal conductance at one specific temperature as the effective normal conductance. For instance, in Junction-3 the normal conductance at 5 K is around 0.2 μA/mV and the conductance of the normal state at 100 K is around 0.4 μA/mV. Hence, the ratio factor r eff is the ratio of the normal conductance at 5 K to conductance at 100 K, which is around 0.5. Then, the normalized differential conductance spectroscopy at 5 K can be determined by multiplying a ratio factor r eff when calculating the division of (dI/dV) S by (dI/dV) N .
Data availability. The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.