Ambipolar Superconductivity with Strong Pairing Interaction in Monolayer 1T′-MoTe2

Gate tunable two-dimensional (2D) superconductors offer significant advantages in studying superconducting phase transitions. Here, we address superconductivity in exfoliated 1T′-MoTe2 monolayers with an intrinsic band gap of ∼7.3 meV using field effect doping. Despite large differences in the dispersion of the conduction and valence bands, superconductivity can be achieved easily for both electrons and holes. The onset of superconductivity occurs near 7–8 K for both charge carrier types. This temperature is much higher than that in bulk samples. Also the in-plane upper critical field is strongly enhanced and exceeds the BCS Pauli limit in both cases. Gap information is extracted using point-contact spectroscopy. The gap ratio exceeds multiple times the value expected for BCS weak-coupling. All of these observations suggest a strong enhancement of the pairing interaction.


Comparison with other 2D superconductors
15. References

Methods
Sample preparation. Bulk single crystals of 1T'-MoTe2 were synthesized using the flux method with NaCl 1,2 . Monolayer and few layer films of 1T'-MoTe2 were mechanically exfoliated from these bulk crystals using sticky tape inside a glovebox with a residual amount of water and oxygen of less than 0.1 ppm. Instead of transferring the flakes on the sticky tape directly onto the Si substrate, covered with the usual 300 nm thick dry thermal SiO2 to analyze their thickness, a softer polydimethylsiloxane (PDMS) stamp was used as an intermediate to place the flakes on to such a Si substrate in order to obtain larger sized flakes. The flakes were typically not homogeneous in thickness, but were composed of the desirable monolayer or few layer region as well as an undesirable area with larger thickness. In order to remove the latter, a tear-and-release procedure using hBN on the PDMS stamp was used 3  the help of a load lock arrangement, the sample was mounted into the sample rod and subsequently transferred to the cryostat without exposure to ambient air. The entire procedure avoided not only air exposure of the 1T'-MoTe2 sample, but also any exposure to solvents or elevated temperatures, all of which would cause a degradation of the sample quality. An accurate determination of the MoTe2 layer thickness was performed a posteriori by combining three pieces of information: the optical contrast produced by the flake during exfoliation and stacking, the onset of the superconducting transition and confocal Raman spectroscopy after completion of the magneto-transport measurements. The thickness determination works particularly well for monolayer and bilayer devices, as shown in Section 2, but becomes less reliable for thicker layers. Measurements. The low-temperature magneto-transport measurements were performed in a dry cryostat system equipped with a superconducting magnet offering an axial field of up to 14 T and a variable temperature insert that allows temperature tuning between 1.6 K and 400 K.
Low-frequency lock-in detection at 17.77 Hz was deployed for noise suppression to record longitudinal and transverse resistances. Measurement currents ranged between 1 nA and 10 µA depending on the two-point resistance of the sample. For both transport and point-contact measurements, the back-gate voltage, top-gate voltage and the dc bias current were applied with the help of a Keithley 2400 Source Measure Unit, a YOKOGAWA 7651 DC Source and an Agilent B2911A precision Source Measure Unit, respectively. The dc-voltage appearing as a result of the bias current was amplified by a voltage pre-amplifier and measured with an Agilent 34401A digital multimeter.

Sample thickness determination
In order to determine the thickness of the 1T'-MoTe2 layer, we mainly use optical microscopy, Raman spectroscopy as well as the critical temperature of the superconducting transition.
Optical images of layers with different thicknesses are illustrated in Fig. S2A. Layers with a thickness of up to three monolayers can be identified using the optical contrast. Raman spectra on flakes of different thicknesses are plotted in panel B. For layer thicknesses of less than four layers the position of the vibrational modes can be used to confirm the layer thickness 4 . These Raman spectroscopy results also help to "calibrate" the optical contrast needed for a certain layer thickness. The observed modes are denoted as P1 through P9 in Fig. S2B. In previous studies, the P1, P4, P5, P6, P8 and P9 modes were identified as Ag modes in the 1T' phase, whereas the P2, P3, and P7 modes were assigned as Bg modes. The position of the P1 mode displays the most obvious layer thickness dependence with the peak occurring at ~86 cm -1 for 1L, ~81 cm -1 for 2L, ~79 cm -1 for 3L and ~78 cm -1 for 4L. Beyond four layers the thickness determination becomes less reliable. The data in Fig. S2B show good agreement with previously published results 4 .
The temperature-dependence of the resistance for samples of different layer thickness is summarized in Fig. S2C. The sheet resistivity increases significantly when the layer thickness decreases from four layers to one layer. At room temperature, the resistivity is about ~600 Ω for devices with a thickness of 3 or 4 layers, ~2 kΩ for devices with two layers and ~5 kΩ for monolayer devices. For a thickness above two layers, the samples display metallic behavior and a critical temperature ,onset below 3 K. The ,onset increases significantly with layer thickness reduction as seen in the inset of Fig. S2C. This behavior is consistent with previously reported data in the literature 1,5 . For monolayer devices, a metal-to-insulator transition is observed at lower doping density, followed by a sharp resistance drop near 7-8 K signaling the superconducting transition. Hence, the ,onset can also serve as a convenient criterion to distinguish samples of different layer thickness. The optical contrast helps to identify the layer thickness. (B) Raman spectra on samples with different layer thickness from the monolayer to the bulk. The layer thickness or layer sequence of the samples used in this Raman study are marked for each trace. (C) Temperature-dependent resistivity of samples with 1T'-MoTe2 layer thicknesses between one and four layers. The inset shows a zoom of each curve near the superconducting transition.

Carrier density estimates
The carrier density is estimated from a straightforward parallel plate capacitor model 6,7 Here e is the electron charge, 0 is the density at zero back and front gate voltage induced by disorder, 0 is the permittivity of vacuum, (SiO 2 ) ~ 3.9 and (hBN) ~ 4 are the relative The determination of the disorder induced density 0 can in principle proceed in two different ways. In the first method it is assumed that the sample resistance reaches its maximum at zero average density. The resistance maxima should then correspond to Because the resistance maxima occur in the insulating regime, this procedure suffers from strong fluctuations in the data as is apparent in panel A-C of Fig. S3. Alternatively, it is possible to use carrier density data points extracted from Hall measurements in the high density regime and extrapolate n0 from a fit of the data to the expression Hall = This method too is not without difficulties. Fig. S3D

Electronic phase diagram
All monolayer devices exhibit the same phase transitions from an insulating state at low density to a superconducting state for larger electron or hole doping. Fig. S4A    The large dV/dI peaks for zero bias current reflect the insulating state (Ins) at low densities.

Density and bias current dependence of the differential resistance
The peak value strongly depends on the device, which we attribute to the significant variation in the residual spatially inhomogeneous carrier density. As the electron (e) or hole (h) density is raised, the peak in the differential resistance at zero bias current vanishes and the differential resistance drops to zero at low bias current instead. This region of zero differential resistance expands with increasing electron or hole density and the critical current where the differential resistance becomes non-zero moves to higher bias current. The density dependence of the critical current is not symmetric when comparing electron and hole doping. This presumably reflects the lack of symmetry in the band structure. In some samples spikes appear in the dV/dI further away from the zero bias current point.
A very clean example is seen in Fig.2C of the main text. This data set is reproduced here in However, we consider the existence of a gap in our monolayer devices of 1T'-MoTe2 beyond doubt. Some of us previously reported a magnetotransport study on multilayer 1T'-MoTe2 devices 11 . At that time, devices were not fabricated such that they are not exposed to ambient conditions or elevated temperatures during processing and sample mounting. While a minor degree of oxidation might just cause some extra doping, more oxidation will induce significant disorder and the majority of the multilayer devices showed insulating behavior due to this disorder. As a matter of fact, it has been a common observation that samples turn insulating focusing on the high-temperature regime between 20 and 100 K. This yields a value for the bulk gap of ~7.3meV. The Arrhenius fit deviates significantly at temperatures below 20 K. This temperature regime is described well by the expression for variable-range hopping (VRH) in a 2D system with localization [ ( )~exp �−( 0 ) . Even if so, other potential mechanisms can't be totally ruled out 13 . In another device D5, the estimated bulk gap is ~7.0 meV for a net density of 3×10 11 cm -2 (Fig. S7B). This is close to the gap value obtained on device D4 in panel A.

The low density regime
In the low density regime, the resistance behaves as an insulator for temperatures typically down to about 7-8 K marked by the gray area in the exemplary density-dependent RT traces recorded on device D1, D2 and D4 in Fig. S9A-C. The kink in the traces in this temperature regime signals the onset of the superconducting transition and does not change significantly when varying the doping level. This is reminiscent of a superconductor-to-insulator transition in granular films of various metals 18,19 . The behavior is attributed to the existence of isolated superconducting puddles in an insulating background 20 , as shown in a cartoon like fashion in the inset of Fig. S9C. This landscape originates from the spatial density inhomogeneity across the sample where areas of high-density remain conducting. Upon reducing the temperature below the transition temperature ,onset Cooper pairs form within these conducting puddles and cause at least a kink or a pronounced downturn of the resistance. The carrier density averaged across the entire sample area is still very small, but inside these puddles the carrier density should be on the order of ~10 12 cm -2 and can be tuned by changing the applied gate voltages. The transition temperature ,onset does not or depends only weakly on the applied gate voltage even down to lowest average sample densities as can be seen in Fig. S9D. The traces plotted in panel A and B for higher average carrier densities indicate that the resistance continues to decrease as temperature is lowered, but the resistance values remain non-zero and large. This suggests that there is only weak link coherence among the puddles and overall superconductivity cannot be achieved.
The existence of superconducting puddles is also corroborated in temperature dependent resistance data in the presence of a perpendicular magnetic field. When the magnetic field is increased, the superconducting puddles should gradually convert into the normal state at relatively high fields (> 2T) and the insulating behavior should then persist also for temperatures below 7 K. This is indeed observed in perpendicular field data plotted in Fig. S10.
The low-temperature resistance reaches values as high as ~10 5 ohms when the superconducting puddles are entirely quenched in high magnetic fields.

Density dependence of the superconducting parameters
In Fig. S11, we summarize the density dependence of some key sample and superconductivity parameters, such as the superconducting transition temperature ,0 , the out-of-plane critical field 2,⊥ , the BCS coherence length 0 and the mean free path . Within the density range we can cover, the superconductivity strengthens with density, since both ,0 in Fig. S11A and 2,⊥ in Fig. S11B continue to increase with increasing density without observing a maximum that would signal optimal doping. Note that a distinction must be made between the superconducting transition temperature Tc,0 determined from the criterion that the resistance drops to 50% of the normal state resistance and the temperature Tc,onset determined from a kink or downturn of the resistance marking the onset of the superconducting transition. A resistance value of 50% of the normal state resistance reflects a macroscopic average across the entire sample and hence Tc,0 is a "global" property. Tc,0 therefore diminishes upon reducing the carrier density as seen in Fig. S11A. This behavior is expected. In contrast, Tc,onset does not reflect global behavior, but local behavior and at zero magnetic field varies little even in the lowdensity regime. We attribute this to the spatial density inhomogeneity across the sample. Even in the insulating regime, some areas exist where the local carrier density is sufficiently high to enter the superconducting regime. Hence, superconducting puddles are present when the temperature approaches Tc,onset even in the insulating regime. When the temperature drops below Tc,onset, Cooper pairs form in these areas and cause at least a kink or downturn in the resistance. Tc,onset therefore does not vary with the average carrier density as ,0 does.
In order to judge whether the superconductor is in the dirty limit ( 0 ≫ ) or the clean limit ( ≫ 0 ), the BCS coherence length is compared with the mean free path for different densities 21,22 . The mean free path follows from the Drude model: with the sheet resistivity, the net charge carrier density. For electron densities, = = 2 to account for both the spin and valley degeneracy. For hole doping, = 2 and = 1 5,6 . We can obtain the BCS coherence length from the in-plane Ginzburg-Landau (GL) coherence length using the relation 0 = 1.35 (0K) . The in-plane coherence length (0K) is calculated using the expression 2,⊥ (0K) = 0 2 2 , where 0 stands for the superconducting flux quantum ℎ/2 and 2,⊥ (0K) is the out-plane critical field at 0 K. The latter is obtained from a linear extrapolation of temperature dependent measurements of the out-of-plane critical field. A comparison of and 0 is displayed in Fig. S11C for device D4 and in Fig. S11D for device D2. Since 0 is always larger than , both devices are in the dirty limit. When decreasing the carrier density down to the insulating regime, 0 increases, whereas decreases. Hence, the devices are located even deeper into the dirty regime with decreasing density.

Fit of the KLB model to the ,∥ − data
Since the superconductivity in our samples occurs in the dirty regime, spin-orbit scattering can be a potential mechanism enhancing the in-plane critical field 2,∥ with respect to the BCS Pauli limit. This is captured by the KLB theory. It yields the following expression for the inplane field dependence of the critical temperature 6,23 ln � ,0 Here, is the spin-orbit scattering time, is the electronic g-factor and is the Bohr magneton. By fitting this expression to the experimental data of device D2, we obtain a of ~170 fs for an electron density of 2.38 × 10 13 cm −2 (Fig. S12A) and ~120 fs for a hole density of −1.10 × 10 13 cm −2 (Fig. S12B). To verify the applicability of the KLB model, it is important to confirm that the transport scattering time is smaller than . The transport scattering time can be estimated from the Drude model using the expression = / .
Here, the Fermi velocity = ℏ / * = ℏ * � 4 . The effective mass * is taken as 0.37 5 . We obtain a transport scattering time of ~20 fs for the electron case illustrated in Fig.   S12A and ~60 fs for the hole case in Fig. S12B. Since < , the KLB model is indeed applicable and spin-orbit scattering may also contribute to an enhancement of the in-plane critical field, in addition to the increase associated with a strengthening of the superconducting gap due to a pair interaction enhancement as discussed in the main text and section S12 (see below).

Magnetic field induced superconductor-to-metal transition
In Fig. S13A, a superconductor-to-metal transition is observed in device D2 at an electron density of 2.60 × 10 13 cm −2 when applying a perpendicular magnetic field larger than ~3 T. This transition is also confirmed in the magnetic field dependence of the differential resistance shown in Fig. S13B. A dip at low dc bias current changes into a peak near ~3 T. Fig. S13C illustrates the magnetoresistance for different temperatures from 1.65 K to 9 K. The isotherms show crossover points (Bc) within a small magnetic field region near ~3 T. The inset in Fig.   S13C shows a close-up view of this crossover area. The width of the crossover region changes with density as illustrated in Fig. S14. It shrinks with increasing density. This observation suggests a magnetic field induced quantum phase transition in this 2D superconductor.  However, in some cases a collection of rare ordered regions can completely destroy the sharpness of the phase transition and this may be the origin of the appearance of a crossover region rather than a single crossover point in Fig. S14. This is reminiscent of a Griffiths singularity. Such a singularity is frequently observed for weakly disordered quantum phase transitions 24,25 . It has been well studied for 3D systems and more recently multiple experimental observations in 2D crystalline superconductors with weak pinning potentials were reported [26][27][28] . In these studies, multiple critical points were observed and through a finite size scaling analysis, a divergent behavior of the dynamical critical exponent, as expected for a Griffiths singularity, was confirmed. The exponent follows from the power law ~( * − ) −0.6 , where C is a constant, * is the infinite randomness critical point and is the static critical exponent. The theoretically predicted value for for a superconductor to metal transition is 0.5 28 . The divergence implies activated scaling with a continuously varying dynamical critical exponent when approaching the infinite-randomness quantum critical point.
In our samples, when increasing the charge carrier density, the 2D superconductor becomes more homogeneous and more robust against fluctuations and disorder. This is likely responsible for the shrinking of the crossover region. Fig. S15 summarizes a typical scaling analysis. In order to extract the dynamical critical exponent , a set of R versus B curves were recorded at different temperatures from 1.7 K to 7.5 K on device D2 for an electron density of 2.60 × 10 13 cm −2 . There are multiple crossover points (Bc). The finite size scaling law for the isotherms for a proximate crossover point can be expressed as ( , ) = • (| − |( / 0 ) −1/ ), where F is an arbitrary function obeying the condition that F(0) = 1. Rc and Bc are the critical resistance and critical magnetic field, respectively. T0 is the lowest temperature for the R vs, B curves used in the scaling analysis. By adjusting the value of , curves for different temperatures can be made to collapse onto a single curve 7,28 . An example of this analysis for a narrow range of temperatures between 1.7 K and 1.9 K is shown in Fig.   S15B. Fig. S15C plots a quantity that quantifies how well the curves collapse onto each other for a given . The minimum corresponds to the value that fits best. Fig. S15D displays the optimized value of as a function of the magnetic field and indeed rises rapidly with increasing field consistent with divergent behavior. The corresponding change with temperature is included as an insert in Fig. S15C. The data points of Fig. S15D   Finally, in Fig. S13D we show a preliminary phase diagram for the transition of the superconducting ground state to the metallic state. Five different criteria are used to determine phase boundaries. The black squares correspond to the out-of-plane upper critical field, 2,⊥ (T). It is defined here as the field where the resistance has dropped to 50% of the normal state resistance using the data in Fig. 13A. The red circles mark the transition from the regime of thermally activated flux flow (TAFF) to the quantum metal (QM) regime. In experiment this transition is signaled by a deviation from thermally activated behavior as described by the Arrhenius expression. This transition has been marked with arrows in the inset of Fig. 13A.
The blue diamonds mark the onset of superconductivity ( ,onset ). The orange triangles mark the crossover point of the R versus B curves recorded at neighboring temperatures (see the inset of Fig. S13C). Green hexagons signal the transition to the true superconducting state when R < 0.1RN and T < TBKT at low magnetic field. With increasing magnetic field, the true superconducting state converts to a weakly localized metallic state via a quantum metal state and/or a quantum Griffith state.

Point contact spectroscopy
In this section, we describe in some detail the point contact spectroscopy method deployed here and the results obtained on device D5. Point contact spectroscopy is a technique commonly used to extract information about the energy gap in a superconductor or also a pseudogap. It involves a measurement of either the contact resistance ( , , defined as = / when = 0 A) or the differential conductance ( = / , see Fig. 4 for measuring geometry) across the contact junction as a function of the dc voltage (Vdc) for different temperatures and magnetic fields. Here a three terminal measurement configuration is used. A bias current is passed through one of the voltage leads of the device (see Fig. 4 in the main text) and a current lead. We will refer to these leads as contact #1 and contact #2. To impose the bias current a voltage difference must be developed across these two contacts and it is composed of three distinct contributions: a voltage drop ∆V1 across the junction between voltage lead #1 and the MoTe2 layer, ∆VMoTe2 due to the resistance of the MoTe2 layer itself and ∆V2 due to the junction between the MoTe2 layer and the current lead #2. By measuring the voltage drop between voltage lead #1 and a second voltage lead (#3), that does not carry and current, we only measure ∆V1, the quantity of interest. The voltage drop across a small portion of the MoTe2 should be of no concern, since MoTe2 remains in the superconducting state during these point-contact measurements (see Fig. S16B). Hence, we can unambiguously obtain the voltage drop across the junction formed by voltage lead #1 and the MoTe2 layer and there is no risk of overestimating the gap size in the selected geometry. The same approach is followed to measure the differential resistance at fixed bias across the junction.
The quality of the contact area is of great importance for point contact spectroscopy. Here in this work, the point contact is formed by placing part of the thin film on top of a micron-sized Au electrode. The contact area is determined by the grain size of the sputtered Au electrodes, and it is inevitable that many parallel channels co-exist. Without exerting any additional pressure, the contact is of the van der Waals type [29][30][31] . Since multiple channels contribute in parallel, the recorded differential conductance represents a spatial average. The superconducting gap appears as a bump like structure in the differential conductance versus bias voltage when the sample is cooled below the critical temperature. The bump becomes stronger at the temperature is lowered and develops a center dip flanked by the two coherence peaks as seen in panel C and E of Fig. 4  The evolution of the differential conductance of contact #1 with an applied perpendicular magnetic field is shown in Fig. S16C. The magnetic field varies from 0 to 6 T.
The temperature dependence can be found in Fig. 4C  pairing. Previous results obtained on bulk materials provided strong evidence for unconventional ± -wave pairing 33,34 . We therefore resort to an extended single-band isotropic s-wave BTK model to fit the normalized data in Fig. S16D. The normalization of the data proceeds, as usual, by subtracting data obtained at some temperature above the superconducting transition temperature. Here, either data at 8K or 15K were used for this purpose. These data sets are vertically shifted until the data points at higher bias voltage match with those of the low-temperature data. Subsequently, the shifted data sets are used to normalize the low temperature data by division. Examples of normalized data traces are shown in Fig. S16D. The fit procedure yields the following parameters: the gap size , the parameter describing the potential barrier Z, and a phenomenological broadening factor Γ. Fits to the data traces using the BKT model yield the following approximate gap values: 3.0 meV (left panel) and 3.1 meV (right panel) using 15K and 8K data for normalization respectively). This result confirms the validity of the normalization procedure as there is no significant influence of the high temperature data set used. The temperature dependence of the differential conductance of contact #1 plotted in Fig. 4B suggests that the contact is in the insulating regime.
The feature associated with a gap persists up to temperatures much higher than ,onset and in Fig 4D there is a deviation from the BCS relation. Such behavior is unanticipated and consistent with pseudogap behavior. In contrast, the conductance data for contact #2 in Fig. 4E does not exhibit such pseudogap behavior. For this data set we are in the metallic regime and we observe Andreev reflection and enhanced conductance below the superconducting gap. These two different behaviors in the insulating and metallic regime are not well understood. It possibly reflects some extrinsic effect such as a gap distribution over a wide energy range due to spatial inhomogeneities 35 . The fitted gap amplitudes are summarized in Fig. 4D and F for both contacts. The phenomenological broadening parameter Γ is shown in Fig. S16E for contact #1 and #2. We note that extrinsic effects are "absorbed" in the phenomenological broadening parameter Γ in this BTK fitting, so that they have little influence on the gap size. For the sake of completeness, we note that the pseudogap behavior has been reported in a number of previous studies, such as low-density superconductors 36 and cuprates 37 .

Monolayer device showing no evidence of a bulk gap
The monolayer devices that were fabricated in an argon glove box environment and encapsulated immediately after exfoliation and placement on the prepatterned substrate showed insulating behavior. We attribute this to the opening of a gap. Here we wish to document the data obtained on a monolayer device with different transport behavior (device D6) from all other measured monolayer samples in this work. It exhibits superconductivity for the full range of gate voltages without any sign for the emergence of a bulk gap. This sample does not exhibit the onset of superconductivity at a higher temperature than the other monolayer devices and, hence, we conclude that the superconductivity is not enhanced compared with the insulating monolayer devices. The disorder induced density in this sample is on the order of 10 12 cm -2 .
The behavior of this sample resembles what has been reported previously in the literature 5 .
While this sample was fabricated using the same processing steps as all other samples, it was left on the Si substrate surface for approximately 5 days prior to protective encapsulation with a thin hBN flake and subsequent transfer onto the substrate with pre-patterned electrodes. Even though the sample was stored inside the argon atmosphere of the glove box, sample degradation may have occurred during this long delay between exfoliation and encapsulation. Partial oxidation may increase the doping and be responsible for the distinct behavior of this sample as compared to the other monolayer samples that were protected by encapsulated immediately.
Fig. S18A displays an optical image of the device. In panel B the band structure schematic for a semi-metal is shown that appears to be applicable for this device. Transport results are shown in this figure as well: R versus T (C), R versus B (D) and the differential resistance dV/dI as a function of the dc bias current Idc (E). In all cases, the charge carrier density serves as an additional parameter. The thick black trace in Fig. S18C corresponds to average zero carrier density using the criterion that at the applied gate voltages, the resistance exhibits a peak in the normal state at 10 K. At this gate voltage condition the critical current also reaches its minimum (Fig. S18E). Fig. S18F plots the temperature at the transition of superconductivity ,0 as well as the critical current Ic as a function of the charge carrier density. (C) Temperature dependence of the sheet resistance for different carrier densities. Data are recorded in the absence of a magnetic field. The thick black line is recorded at gate voltages corresponding to zero average density. (D) Sheet resistance as a function of the perpendicular magnetic field. The charge carrier density serves as an additional parameter. All data taken at 1.65 K. (E) Differential resistance as a function of the dc-bias current. The charge carrier density serves as an additional parameter. These data are used to extract the critical current Ic in panel F. Data are recorded at T = 1.65 K and B = 0 T. (F) Temperature at the transition of superconductivity, ,0 , and critical current, Ic, as a function of the average carrier density. The density range in panel C-F is -1.51 ×10 13 < n < 1.08×10 13 cm -2 .

Comparison with other 2D superconductors
Key parameters of our superconducting samples are compared with other 2D superconductors in the table below. ,⊥

(T)
,∥  Table S1. Comparison of key parameters of the superconducting 1T'-MoTe2 monolayers with some other 2D superconductors, that have been reported in the literature. They include type-I and type-II Ising superconductors, twisted graphene layers (TBLG = twisted bilayer graphene, TTLG = twisted trilayer graphene) and monolayer 1T'-WTe2. In some cases the density is not known.