Superconductivity at 38 K at an electrochemical interface between an ionic liquid and FeSe0.8Te0.2 on various substrates

Superconducting FeSe0.8Te0.2 thin films on SrTiO3, LaAlO3 and CaF2 substrates were electrochemically etched in an ionic liquid, DEME-TFSI, electrolyte with a gate bias of 5 V. Superconductivity at 38 K was observed on all substrates after the etching of films with a thickness greater than 30 nm, despite the different Tc values of 8 K, 12 K and 19 K observed before etching on SrTiO3, LaAlO3 and CaF2 substrates, respectively. Tc returned to its original value with the removal of the gate bias. The observation of Tc enhancement for these thick films indicates that the Tc enhancement is unrelated to any interfacial effects between the film and the substrate. The sheet resistance and Hall coefficient of the surface conducting layer were estimated from the gate bias dependence of the transport properties. The sheet resistances of the surface conducting layers of the films on LaAlO3 and CaF2 showed identical temperature dependence, and the Hall coefficient was found to be almost independent of temperature and to take values of −0.05 to −0.2 m2/C, corresponding to 4–17 electrons per FeSe0.8Te0.2 unit cell area in two dimensions. These common transport properties on various substrates suggest that the superconductivity at 38 K appears in the surface conducting layer as a result of an electrochemical reaction between the surface of the FeSe0.8Te0.2 thin film and the ionic liquid electrolyte.

We have previously reported T c enhancements in FeSe 1−x Te x thin films on various substrates [23][24][25][26] . The observed T c values depend on both the Te content and the substrate material. For example, the T c values for FeSe 0.8 Te 0.2 films on CaF 2 and LaAlO 3 substrates are enhanced to 20 K and 12 K, respectively, in contrast to the T c of 8 K observed for FeSe 0.8 Te 0.2 on SrTiO 3 . This can be explained by differences in the a-axis lattice constants of FeSe 0.8 Te 0.2 films on different substrates 24 . In this paper, we report the enhancement of T c up to 38 K for thick FeSe 0.8 Te 0.2 films on various substrates prepared via EDLT fabrication with the ionic liquid DEME-TFSI. By means of electrochemical etching with the ionic liquid 16 , the film thickness was varied. The application of a gate bias resulted in the formation of a surface conducting layer with a T c of 38 K; with the removal of the gate bias, the surface conducting layer disappeared, causing T c to return to its original value. We also estimated the transport properties of the surface conducting layer. The surface conducting layer exhibited electron conduction with a common dependence of the mobility on temperature on various substrates.

Results
Characterization of FeSe 0.8 Te 0.2 thin films. The film thickness and crystal quality were examined via XRD measurements. Figure 1(a) shows the XRD patterns of FeSe 0.8 Te 0.2 thin films fabricated on LaAlO 3 (LAO), CaF 2 and SrTiO 3 (STO) substrates. All samples exhibited (001), (002) and (004) peaks, while the (003) peak for the film on the LAO substrate was obscured by a (002) peak of the substrate. The c-axis lattice constants for the films on the LAO, CaF 2 and STO substrates were 5.69, 5.72 and 5.70 A, respectively, consistent with previous reports 23,24 . The full widths at half maximum (FWHMs) of the rocking curves for the (001) peak were 0.4, 0.7 and 1 deg. for the films on the LAO, CaF 2 and STO substrates, respectively; these values are also almost the same as those reported previously (0.2-0.6 deg.) 23 , demonstrating that all of these films consisted of high-quality single-phase samples. X-ray reflectivity (XRR) measurements revealed clear thickness fringes for all films, as shown in Fig. 1(b), indicating a smooth surface and a sharp interface between the film and the substrate. In addition, all XRR curves were well fitted by the model structure, and we estimated the thicknesses as shown in Fig. 1(b). The film thicknesses were also confirmed by atomic force microscopy (AFM) images of the films.
Superconducting properties of pristine and etched FeSe 0.8 Te 0.2 thin films. The superconducting properties of the samples when subjected to gating and etching were examined for a gate bias (V G ) of 5 V. As shown in Fig. 2(a), the EDLT samples were patterned in Hall bars with six electrodes, and a Pt film was placed alongside each Hall-bar-instrumented sample to act as a gate electrode. Figure 2(c,d) show the temperature (T) dependences of the sheet resistance (R S ) for samples of FeSe 0.8 Te 0.2 films on LAO and CaF 2 substrates (LAO and CaF 2 samples), respectively. First, the R S -T curves of the pristine sample before etching were measured for V G values of of 0 V and 5 V. Then, the temperature was increased to 250 K with a V G of 5 V while monitoring the gate current (I G ), as shown in Fig. 2(b). The channel was electrochemically etched, and the drain current (I D ) was gradually decreased. The product of I G and time is a Faradaic charge (Q F ) that is proportional to the amount of charge of the reacted ions. After etching, T was decreased while maintaining V G = 5 V, and the R S -T curve was measured. After several cycles of etching, the V G dependence of the R S -T curve for the etched sample was measured via the following procedure: First, the R S -T curve was measured for a V G of 5 V. Then, the temperature was increased to 250 K without V G , and the R S -T curve for V G = 0 V was measured. Finally, a V G of 5 V was again applied at 250 K, and the R S -T curve for V G = 5 V was measured again. We show the R-T curves of the pristine and etched samples for V G = 5 V, V G = 0 V, and V G = 5 V. The R S -T curves of the pristine samples remained almost unchanged between the V G values of 0 V and 5 V. In contrast, T c was enhanced after several cycles of etching at V G = 5 V. T c increased from 12 K to 38 K for the LAO sample and from 19 K to 38 K for the CaF 2 sample. As shown in S. Fig. 3 in the Supplementary Information, critical magnetic field at 0 K was also enhanced from 47 T to 67 T on films on the LAO substrate. The coherence length at 0 K was 3.8 nm to 3.1 nm. With the removal of V G , R S increased, and T c returned to the value of the pristine sample. With the application of a V G of 5 V, R S slightly decreased, and T c was enhanced to 38 K for both samples. Notably, the R S value at V G = 5 V after the removal of the initial V G was larger than that before the removal of V G . Since the sheet resistance due to electrostatic carrier doping should always have the same value at V G = 5 V, this difference suggests that the change in R S can be attributed to some other origin than electrostatic carrier doping. In addition, the T c for V G = 0 V was different for the LAO and CaF 2 sample because of different lattice constant of the FeSe 0.8 Te 0.2 films. If the surface conducting layer is produced by an electrostatic carrier doping, it is natural that the surface conducting layer also show different T c values for V G = 5 V. Then, the T c enhancement to 38 K for both samples also suggests that the origin of the carrier doping is not the electrostatic doping. This will be discussed later. Figure 2(e,f) show the T dependences of the Hall coefficient (R H ) for the same samples shown in Fig. 2(c,d), respectively. R H was almost zero above 80 K and showed an increase with decreasing temperature below 40 K for the pristine samples. In contrast, R H was always negative at all temperatures for the etched samples at V G = 5 V. The R H -T behavior returned to almost the original one after the application of V G = 0 V. These results indicate that the etching at V G = 5 V resulted in the formation of a conducting layer on the surface, with a T c of 38 K, for both the LAO and CaF 2 samples. Since R H was negative, electron conduction dominated in the conductive layer. In addition, since the transport properties of the pristine Thickness dependence of the superconducting properties of FeSe 0.8 Te 0.2 thin films. The thickness dependence of the superconducting properties was also examined for the LAO, CaF 2 and STO samples. Figure 3(a-c) show the R S -T curves of the LAO, CaF 2 and STO samples, respectively, with various thicknesses (numbers of etching cycles). R S is normalized to the R S value at 100 K. For the LAO and CaF 2 samples, T c was enhanced after several cycles of etching. The CaF 2 sample showed a two-step superconducting transition during the initial stage of etching. A similar two-step transition has been reported for an EDLT configuration on an FeSe flake with a gate bias of approximately 4 V and has been ascribed to the inhomogeneity of the carrier distribution 15 . In addition, the normalized R S -T curves after the T c enhancement were nearly identical. In contrast, for the STO sample, T c was gradually enhanced from 8 K to 38 K over many cycles. We estimated the onset temperature (T c onset ) of superconductivity from the R S -T curve as shown in Fig. 3(a). We also estimated the thickness after n cycles of etching, thickness(n), from the following equation: where Q F is the Faradaic charge for each cycle, thickness(XRR) is the film thickness before etching as estimated from the XRR measurement, and n total is the total number of cycles needed for the etching of the entire film. After the total etching of the film, the sample resistance is larger than MOhm. In addition, the film totally disappeared after the etching experiment. Therefore, we assumed the entire film was etched during the n total cycles of etching. Figure 3(d) shows the thickness dependence of T c onset for the LAO, CaF 2 and STO samples. T c started to increase only after two cycles of etching and saturated at 38 K after four cycles for the LAO and CaF 2 samples. As reported in the previous paragraph, T c decreased to its original value with the removal of V G at 250 K, but it returned to 38 K after the next application of V G = 5 V for the next cycle. The STO sample showed T c enhancement from 8 K to 16 K at a thickness of 30 nm and a further T c enhancement to 38 K at a thickness of 10 nm. Since T c enhancement was observed for thick samples on all substrates, and since T c changed with the application and removal of V G , we conclude that this T c enhancement was not affected by the interface between the substrate and the film but instead originated from the surface conducting layer produced by V G = 5 V.
The thickness of the samples which showed the T c enhancement was confirmed by means of transmission electron microscopy (TEM) and XRD measurements. We performed corresponding etching experiments using other samples on LAO, CaF 2 and STO substrates and terminated the etching process after several cycles. All of the etched samples showed superconductivity at T c values above 34 K. The film thickness after etching was directly obtained via TEM measurements. The thickness dependences of T c onset for these samples are shown in   Fig. 3(d). The FeSe 0.8 Te 0.2 films on LAO, CaF 2 and STO substrates all showed T c enhancement for films with thicknesses greater than 30 nm. As shown in Fig. 4(a,b) and in S. Fig. 1(a,e) in the Supplementary Information, the interface between the substrate and the film was smooth for all films, and a clear periodicity of the atomic arrays was observed. The bright region at the interface probably indicates Se diffusion from the film into the substrate 27,28 . For the film on the LAO substrate, the clear periodicity remained at the surface, and no additional layer was found on the film, as shown in Fig. 4(a). On the CaF 2 and STO substrates, a disordered FeO x layer was found on the ordered region with clear periodicity. The T c enhancement probably occurred in this ordered region. XRD patterns recorded before and after etching indicated that the peak position remained unchanged and that no new peak was present after etching, as shown in Fig. 4(c); the only observed difference was a decrease in the peak intensity. In addition, the FWHM of the rocking curve for the (001) peak remained unchanged, as shown in Fig. 4(d). These XRD data indicated that no new phase was created in the film by the etching process. Thus, no electrochemical reaction occurred in the bulk of the film; instead, such reactions took place only at the surface.
We also examined the thickness dependence of T c for FeSe films on LAO and STO substrates. As shown in S. Fig. 2(a,b), these films showed T c enhancements of up to 30 or 40 K upon etching. On STO substrates, only films with thicknesses below 12 nm showed T c enhancement. This finding coincides with those of previous reports 16 . In contrast, a film with a thickness of 30 nm on the LAO substrate showed T c enhancement, similar to the behavior of FeSe 0.8 Te 0.2 films. Notably, several FeSe 0.8 Te 0.2 samples on CaF 2 and STO substrates showed T c enhancement to above 37 K only for thicknesses below 10 nm. The different critical thicknesses for FeSe and FeSe 0.8 Te 0.2 films on different substrates were probably due to differences in the homogeneity of the films. As shown in S. Fig. 1(a,e) in the Supplementary Information, a disordered region was observed in the FeSe 0.8 Te 0.2 film on STO. In addition, a disordered Fe (or FeO x ) layer was observed on top of the FeSe 0.8 Te 0.2 films on CaF 2 and STO substrates. These TEM images indicate that the film quality depends on the substrate and that the best quality is achieved for FeSe 0.8 Te 0.2 films on LAO substrates. We consider that good film quality throughout the entire film is necessary for the occurrence of T c enhancement for a thick film. Thus, although we did not perform TEM measurements of all of these samples, the lack of T c enhancement to 38 K for the thick films on some samples might have been due to insufficient film homogeneity, especially near the surface.

Discussion
Finally, we discuss the origin of the T c enhancement. The T c enhancement has been reported to be due to charge accumulation on the surface of the FeSe 15,16,18 . However, an electrochemically reacted layer on the surface may also show a high T c , since FeSe samples intercalated with alkali ions and/or organic molecules present T c values above 40 K 8,9 . To distinguish electrostatic charge accumulation from electrochemical reaction, we estimated the sheet resistance and Hall coefficient of the surface layer. The resistance tensor, ρ, and the conductance tensor, σ, are represented by the following equations: where B is the magnetic field applied during the Hall measurement. The σ of a sample at V G = 5 V is equal to the sum of the σ values of the sample at V G = 0 V and of the surface conducting layer produced by a V G of 5 V. Therefore, the sheet resistance and Hall coefficient of the surface conducting layer, R S surface and R H surface , obey the following equations: As shown in Fig. 5(b-d), we examined the changes in the sheet resistance and Hall coefficient for one V G cycle of a LAO sample (LAO-1, as shown in Fig. 2(c)) and two V G cycles of a CaF 2 sample (CaF 2 -1 and CaF 2 -2, where the data for CaF 2 -1 are shown in Fig. 2(d)). Figure 5(a) shows the temperature dependence of R S surface normalized to the value at 90 K. The R S -T curves for the LAO and CaF 2 samples just before the last etching cycle, R S (last), are also plotted. All curves follow almost the same profile. This indicates that the transport properties, such as the electron mobility and scattering time, of all samples exhibited identical temperature dependences. As shown in the inset of Fig. 5(a), R H was almost independent of temperature and negative for all samples. Electron-type conduction is a common feature in previous reports on the T c enhancement of FeSe 7,15-18 , and the vanishing of the hole pocket at the Fermi level has been considered to be the origin of the T c enhancement 7 in the accumulated carrier density 29 . However, no such change was observed in the R-T curves. In addition, as shown previously in Fig. 2(c,d), the removal of V G irreversibly changed R S , suggesting that the origin of the change in R S is some other phenomenon than electrostatic carrier doping. Therefore, a different cause of carrier doping other than the electric field effect is likely responsible for the T c enhancement.
Both the irreversible change in R S with V G and the lack of variation in the electron mobility with the carrier doping can be explained by assuming that the surface conducting layer is formed not by the accumulation of electrostatic charge but by an electrochemical reaction between the FeSe 0.8 Te 0.2 and the ionic liquid. We hypothesize that the etching of the film and the formation of the surface conducting layer occurred simultaneously with the application of the V G of 5 V. One potential candidate of the forming reaction of the surface conducting layer is an electrochemical intercalation of DEME + ion, . .
FeSe Te DEME e F eSe Te (DEME) Since observed Q F during the reaction was much larger than Q F needed for this reaction, other electrochemical reaction, such as electrochemical decomposition of DEME-TFSI and dissolution of FeSe 0.8 Te 0.2 , also occurs. We discussed on the possible electrochemical reactions in the Supplementary Information. In addition, we hypothesize that when this V G was removed, the surface conducting layer disappeared, probably due to decomposition or peeling off from the surface. Then, the abrupt decrease in the sheet conductance occurred with the removal of V G . In addition, both the electron mobility and the volume charge carrier density should be identical among LAO-1, CaF 2 -1 and CaF 2 -2. This hypothesis was also supported by the change in the sheet conductance with the repeated etching of the LAO and CaF 2 samples, as shown in Fig. 5(d). The sheet conductance at 50 K decreased with the removal of V G and, with repeated etching, gradually increased after this reduction. This behavior can be explained by an increase in the thickness of the surface conducting layer with repeated etching. If the conductance of the surface conducting layer is higher than that of the bulk FeSe 0.8 Te 0.2 film at 50 K, then repeated etching will increase the sheet conductance at low temperatures. As the number of etching cycles increases, the ratio of the thickness of the surface conducting layer to the total film thickness will increase. Then, just before the film is totally removed, the surface conducting layer will cover almost the entire film. Consistent with this picture, the temperature dependences of the sheet resistance just before the last etching cycle for both the LAO and CaF 2 samples were also identical to that for the surface conducting layer, as shown in Fig. 5(a). We also examined two dimensionality of the superconductivity on the surface conducting layer. When the superconducting layer is thinner than the superconducting coherence length, it behaves as a two dimensional superconductor. However, as shown in S. Fig. 4 in the Supplementary Information, the surface conducting layer did not behave as a two dimensional superconductor. This suggests that superconducting layer is electrochemically formed and relatively thick.

Conclusion
In conclusion, a surface conducting layer with a T c of 38 K was formed with the electrochemical etching of FeSe 0.8 Te 0.2 thin films on LAO, CaF 2 and STO substrates. Since the thicknesses of all etched samples with T c values of 38 K were greater than 30 nm, the enhancement of T c cannot be related to any interaction between the film and the substrate. In addition, T c enhancement was also observed for an FeSe thin film on a LAO substrate with a thickness of approximately 30 nm. The surface conducting layer again showed almost identical temperature dependences between the sheet resistance and the Hall coefficient. This finding suggests that the surface conducting layer is formed not by the accumulation of electrostatic charge on the FeSe 0.8 Te 0.2 surface but by an electrochemical reaction between the FeSe 0.8 Te 0.2 and the ionic liquid electrolyte. Hall coefficient measurements showed that the surface conducting layer contained 4-17 electrons per unit cell in two dimensions, with an overall negative charge. From TEM measurements, we could observe a smooth interface between the substrate and the film and a clear periodicity of the atomic arrays in the etched FeSe 0.8 Te 0.2 film on the LAO substrate. These observations indicate that the formation of the surface conducting layer did not affect the bulk region of the film, and the surface conducting layer completely disappeared with the removal of V G . We consider that previous studies on carrier doping in ultrathin FeSe film can be classified into two groups: those that show a T c of approximately 40 K for a several-layer FeSe film and those that show a T c above 65 K for a monolayer FeSe on STO. Our results indicate that the electrochemical doping of FeSe and FeSe 0.8 Te 0.2 can result in the formation of a superconducting layer with a T c of approximately 40 K and that no interfacial interaction is necessary for the enhancement of T c to 40 K. However, we think that the interface between FeSe and STO is probably essential for the enhancement of T c above 65 K. We believe that it will be possible to prepare monolayer FeSe with a T c above 65 K with further study of the electrochemical etching of FeSe.

Methods
FeSe and FeSe 0.8 Te 0.2 thin films were deposited by means of pulsed laser deposition (PLD) with a KrF eximer laser and an FeSe or FeSe 0.8 Te 0.2 polycrystalline target. The fabrication conditions are described in detail elsewhere 30,31 . We used a commercially available STO (001) substrate with a step-and-terrace surface, a LAO (001) substrate and a CaF 2 (001) substrate. AFM and XRD measurements were carried out prior to device fabrication. Au(100 nm)/ Ti (20 nm) films were formed via electron-beam evaporation at a base pressure of 10 −5 Torr. Since FeSe 0.8 Te 0.2 thin films can be damaged by exposure to high temperatures (above 100 deg. Celsius) and water during the standard processes of photolithography and dry etching, we employed a sandblasting technique at room temperature for the fabrication of Hall bar electrodes. The films were coated with a dry film resist patterned via photolithography and were etched by sandblasting with alumina emery (#220). The Hall bar electrodes and wires were coated with a silicone sealant to prevent electrochemical reactions between the electrolyte and the Au/Ti electrode.