Oxygen Adsorption Induced Superconductivity in Ultrathin FeTe Film on SrTiO3(001)

The phenomenon of oxygen incorporation-induced superconductivity in iron telluride (Fe1+yTe, with antiferromagnetic (AFM) orders) is intriguing and quite different from the case of FeSe. Until now, the microscopic origin of the induced superconductivity and the role of oxygen are far from clear. Here, by combining in situ scanning tunneling microscopy/spectroscopy (STM/STS) and X-ray photoemission spectroscopy (XPS) on oxygenated FeTe, we found physically adsorbed O2 molecules crystallized into c (2/3 × 2) structure as an oxygen overlayer at low temperature, which was vital for superconductivity. The O2 overlayer were not epitaxial on the FeTe lattice, which implied weak O2 –FeTe interaction but strong molecular interactions. The energy shift observed in the STS and XPS measurements indicated a hole doping effect from the O2 overlayer to the FeTe layer, leading to a superconducting gap of 4.5 meV opened across the Fermi level. Our direct microscopic probe clarified the role of oxygen on FeTe and emphasized the importance of charge transfer effect to induce superconductivity in iron-chalcogenide thin films.

Zheng et al. [12] found that interstitial oxygen, rather than oxygen substitution [23], was responsible for the emergence of superconductivity. The interstitial oxygen [13,16] would dope holes into FeTe [17], suppress the AFM ordering and then induce superconductivity. Yamazaki et al. [19] and Sun et al. [18] have studied the dynamics of oxygen annealing and concluded that the superconductivity first emerged on the surface and the superconducting regions moved inside with non-superconducting materials left on the surface, such as Fe 2 O 3 , TeO x , FeTe 2 . The controversial results from different experimental probes call for a unified study of the crystal, chemical and electronic structures on thin films with controlled oxygen incorporation. However, this is generally challenging to achieve for ex situ measurements because samples exposed to the atmosphere inevitably change surface morphology and chemistry. In this study, we overcame this challenge by integrating one ultra-high vacuum environment for the sample growth and oxygenation process along with spectroscopic tools with elemental and spatial resolution.
First, we grew nearly stoichiometric 10 Unit-Cell (UC) FeTe epitaxy ultrathin films by molecular beam epitaxy (MBE) and then studied the adsorption of O 2 incorporation by combined in situ STM and X-ray photoelectron spectroscopy (XPS). We found that O 2 molecules physically adsorbed on FeTe surface under low O 2 partial pressure at room temperature and crystallized into c (2/3 × 2) structure, which gradually disappeared after heating up to 100 • C in vacuum. No surface oxidation can be found. The band shift observed in STS and XPS suggests hole doping from the O 2 overlayer to FeTe, which induced a superconducting gap in FeTe. It agrees with the ex situ transport measurement that long-time exposure to air can induce superconducting transition in 10UC FeTe on STO.

Materials and Methods
High-quality 10UC FeTe thin films were grown on SrTiO 3 (001) by using a Unisoku ultrahigh vacuum (UHV) low-temperature STM system (Unisoku 1300 9T-2T-2T, 2016, Hirakata, Osaka, Japan), equipped with an MBE chamber for film preparation. The base pressure of our MBE chamber is 1 × 10 −10 mbar and does not exceed 5 × 10 −10 mbar during FeTe epitaxial growth. The Nb doped-STO substrate (Nb: 0.7% wt, KMT comp.) was firstly annealed at 1000 • C for 1 h and then kept at 280 • C during FeTe growth. High purity Fe (99.9999%) and Te (99.9999%) sources were co-evaporated onto SrTiO 3 (001) substrates from two standard Knudsen cells. To improve the quality of crystallization, FeTe films were annealed at 280 • C for 1 h after growth. The O 2 gas (99.999%) was introduced to a small preparation chamber with a base pressure of 2 × 10 −9 mbar. Then the oxygen absorption was carried out at room temperature under O 2 partial pressure of 1.6 × 10 −4 mbar (if not specified otherwise). After O 2 absorption, the samples were in situ transferred to the STM chamber or XPS equipment for further characterizations. For electronic transport measurement, 10UC FeTe films were grown on high-resistance STO and covered with the Te protection layer.
All STM/STS measurements were performed at a temperature of 4.7 K with a bias voltage applied to the sample. Polycrystalline Pt/Ir tips were cleaned by electron bombardment and then carefully calibrated on Ag films grown on Si (111). All STM topographic images were obtained under a constant current mode. The differential conductance dI/dV spectra were acquired by using a standard lock-in technique at a frequency of 983 Hz. The STM topographic images were processed with WSXM software (WSxM 5.0 Develop 10.0, 2020, Julio Gómez Herrero & José María Gómez Rodríguez, Madrid, Spain) [24].
After STM/STS measurements, the samples were transferred via a UHV interconnection pipeline in Nano-X for XPS/UPS characterization without exposure to air. XPS/UPS experiments were carried out by using a commercial analyzer (PHI 5000, VersaProbe, ULVAC-PHI,~5 × 10 −10 mbar, 2016, Chigasaki, Kanagawa, Japan), equipped with a monochromatic Al K α X-ray source of 1486.7 eV. The binding energy (BE) of core-level peaks were calibrated with respect to the C-C 1s bond (BE = 284.8 eV). A monochromatized He I with an energy of 21.22 eV were used for UPS measurement. During the UPS measurement, a voltage of −5 V was applied between the sample and the spectrometer.

Results
High-quality FeTe films are epitaxially grown on STO substrates via step flow growth mode. Figure 1a shows the typical surface morphology of 10UC FeTe film, and Figure 1b shows the atomically resolved image with Te termination. Fast Fourier transform image yields enlarged lattice constant of~0.384 nm, similar to the case of FeSe on STO [1]. No Fe adatoms (or interstitial Fe atoms) and Te vacancies can be observed on the surface, which indicates nearly stoichiometric FeTe epitaxy films. To carry out electronic transport measurement, the protection Te layer was deposited on the 10UC FeTe surface, and the results were shown in Figure 1c. No superconducting transition but semiconducting behavior was observed in initially grown 10UC FeTe film down to 1.8 K, consistent with previous experiments [25,26]. Compared with bulk or thick FeTe films [11,13,23,[27][28][29], we have not found an anomaly in the R-T curve between 40 K and 80 K, which further demonstrates free of Fe interstitial atoms [26]. AFM state may be suppressed here. After exposing the sample to air for 8 days, the resistance turned downward at~3.2 K, as shown in the inset of Figure 1c. Considering the fact of long-time exposure-induced superconductivity in FeTe compounds, such downward should be attributed to superconducting transition. It is further verified by the suppression of transition under external magnetic fields in Figure 1d, characteristic of superconductivity. Non-zero resistance here suggests the developed superconductivity is not uniform. We found that exposure to air for a longer time would increase the superconducting transition temperature further.

Results
High-quality FeTe films are epitaxially grown on STO substrates via step flow growth mode. Figure 1a shows the typical surface morphology of 10UC FeTe film, and Figure 1b shows the atomically resolved image with Te termination. Fast Fourier transform image yields enlarged lattice constant of ~ 0.384 nm, similar to the case of FeSe on STO [1]. No Fe adatoms (or interstitial Fe atoms) and Te vacancies can be observed on the surface, which indicates nearly stoichiometric FeTe epitaxy films. To carry out electronic transport measurement, the protection Te layer was deposited on the 10UC FeTe surface, and the results were shown in Figure 1c. No superconducting transition but semiconducting behavior was observed in initially grown 10UC FeTe film down to 1.8 K, consistent with previous experiments [25,26]. Compared with bulk or thick FeTe films [11,13,23,[27][28][29], we have not found an anomaly in the R-T curve between 40 K and 80 K, which further demonstrates free of Fe interstitial atoms [26]. AFM state may be suppressed here. After exposing the sample to air for 8 days, the resistance turned downward at ~3.2 K, as shown in the inset of Figure 1c. Considering the fact of long-time exposure-induced superconductivity in FeTe compounds, such downward should be attributed to superconducting transition. It is further verified by the suppression of transition under external magnetic fields in Figure 1d, characteristic of superconductivity. Non-zero resistance here suggests the developed superconductivity is not uniform. We found that exposure to air for a longer time would increase the superconducting transition temperature further.
.  To figure out the superconducting mechanism in FeTe after exposure to O 2 , we carried out the oxygenated process under a well-control condition and probed the atomic and electronic structures microscopically. Figure 2a shows the surface morphology of the initial adsorption stage after exposure 10UC FeTe to O 2 under a partial pressure of 1.6 × 10 −4 mbar for 15 min. We can observe extra layers along the step edge, while other uncovered clean FeTe areas still remain unchanged, as illustrated by the atomically resolved STM image in the inset. No O Te substitution or interstitial O atoms can be observed. The enlarged image in Figure 2b shows there are small compacted islands with a height of 0.18 nm in the overlayer, as shown in Figure 2c. Some dispersed clusters can be also found at the domain boundaries, which should be the ones that have not coarsened into the islands. If further increasing the amount of inlet O 2 , nearly the whole surface can be covered by the adsorbed overlayers, as shown in Figure S1a. The uncovered areas are still very clean in Figure S1b. showing turn-down behavior after exposure to air. (d) Temperature-dependent resistance of 10UC FeTe after exposure to air for 8 days under an out-of-plane magnetic field up to 2 T. The turn-down behavior was suppressed under magnetic fields and recovered after magnetic field withdrawal.
To figure out the superconducting mechanism in FeTe after exposure to O2, we carried out the oxygenated process under a well-control condition and probed the atomic and electronic structures microscopically. Figure 2a shows the surface morphology of the initial adsorption stage after exposure 10UC FeTe to O2 under a partial pressure of 1.6 × 10 −4 mbar for 15 min. We can observe extra layers along the step edge, while other uncovered clean FeTe areas still remain unchanged, as illustrated by the atomically resolved STM image in the inset. No OTe substitution or interstitial O atoms can be observed. The enlarged image in Figure 2b shows there are small compacted islands with a height of ~0.18 nm in the overlayer, as shown in Figure 2c. Some dispersed clusters can be also found at the domain boundaries, which should be the ones that have not coarsened into the islands. If further increasing the amount of inlet O2, nearly the whole surface can be covered by the adsorbed overlayers, as shown in Figure S1a. The uncovered areas are still very clean in Figure S1b.  To unravel the nature of adsorbed overlayer (O 2 molecules, FeO x , TeO x , Fe-Te-O or others?), Figure 2d,e present the core-level XPS spectra of Te 3d and Fe 2p before and after inlet O 2 , respectively. After inlet O 2 under a partial pressure of 1.6 × 10 −4 mbar, we do not observe Te-O and Fe-O spectra signals. If exposure to O 2 partial pressure of 1.5 × 10 −1 mbar for 20 min, as shown in Figure S3c clear Te-O core level peak can be identified, which suggests surface oxidation under high O 2 partial pressure. The XPS spectra of Te 3d and Fe 2p show an obvious peak shift toward higher binding energy, which hints hole doping effect in FeTe. The overlayer would gradually desorb when heating up to 100 • C and disappeared after annealing at 200 • C for 2 h. Only a clean FeTe surface was left. The overlayer will re-adsorb after inlet O 2 again. Thus, we can rule out the possibility of Fe-or Te-related compounds because there was negligible residual Fe or Te atoms after growth. In addition, we notice there are some dispersed clusters higher than the compacted islands in Figures 2b and 3a, which helps us to conclude that the overlayer should be O 2 molecules rather than stacked O atoms. The shape of O 1s core level almost remains unchanged after O 2 adsorption under a partial pressure of 1.6 × 10 −4 mbar, as illustrated in Figure S3, but changes a lot after surface oxidation due to the signal from O-Te and O-Fe. It implies the signal of O 1s at 529.5 eV~531 eV are mainly from the STO substrate, which, as expected, exhibits a similar core level shift as Te 3d, as shown in Figure S3a. It also consists of the nature of physically adsorbed O 2 molecules, where characteristic core-level XPS peak always cannot be found. Thus, we name the overlayer heterostructure as FeTe : O 2 . UPS measurements in Figure 2f show that the work function just slightly changed from φ FeTe~4 .43 eV to φ FeTe:O 2~4 .38 eV after O 2 adsorption. The small difference in work function is also consistent with the nature of O 2 molecules in the overlayer. To explore how the electronic structure changes due to O2 absorption, we conducted detailed differential conductance dI/dV measurements, which are known to be proportional to the local density of state (LDOS). On as-prepared and uncovered clean 10UC FeTe surfaces, the STS remain nearly unchanged with a "V" shape across the Fermi level in Figure 4a, consistent with previous studies [2,30]. A hump at the negative bias voltage shift from −0.37 V to −0.325 V on FeTe : O2 after O2 adsorption, which indicates hole doping effect to FeTe, agreeing well with the observation of peak shift in XPS experiments in Figure 2. Such oxygen-induced hole doping effect [31] was recently observed on the CsV3Sb5 with an increased superconducting transition temperature. Additionally, hole doping effects were observed in Bi2Te3/FeTe [32,33] and Sb2Te3/FeTe [26] heterostructures and thought to be the origin of two-dimensional superconductivity in the heterostructures. Figure 4b shows the dI/dV spectra in a small energy range. A clear V-shape was observed on a clean FeTe surface, similar to the one on the monolayer FeTe [2] and thick FeTe film  is nearly equal to 2/3a FeTe . Thus, the lattice can be notated as c (2/3×2), as shown in Figure 3b,c. Such lattice arrangement demonstrates a close relationship between the O 2 adsorption overlayer (FeTe : O 2 ) and the FeTe surface. To locate the precise adsorption sites of O 2 on FeTe, we achieved an atomically resolved image on both the FeTe : O 2 island and FeTe surface in Figure 3d. The results are shown in the structural model in Figure 3e. As illustrated, along the n 1 direction, the adsorption sites of O 2 molecules are not well-defined relative to FeTe lattice and not at high-symmetry points. It not only signifies that the van der Waals interaction between O 2 molecules plays a vital role in the overlayer but also verifies the absence of chemical bonding between the O 2 interlayer and the FeTe surface.
To explore how the electronic structure changes due to O 2 absorption, we conducted detailed differential conductance dI/dV measurements, which are known to be proportional to the local density of state (LDOS). On as-prepared and uncovered clean 10UC FeTe surfaces, the STS remain nearly unchanged with a "V" shape across the Fermi level in Figure 4a, consistent with previous studies [2,30]. A hump at the negative bias voltage shift from −0.37 V to −0.325 V on FeTe : O 2 after O 2 adsorption, which indicates hole doping effect to FeTe, agreeing well with the observation of peak shift in XPS experiments in Figure 2. Such oxygen-induced hole doping effect [31] was recently observed on the CsV 3 Sb 5 with an increased superconducting transition temperature. Additionally, hole doping effects were observed in Bi 2 Te 3 /FeTe [32,33] and Sb 2 Te 3 /FeTe [26] heterostructures and thought to be the origin of two-dimensional superconductivity in the heterostructures. Figure 4b shows the dI/dV spectra in a small energy range. A clear V-shape was observed on a clean FeTe surface, similar to the one on the monolayer FeTe [2] and thick FeTe film [30]. Strikingly, on FeTe : O 2 islands two pronounced peaks with an energy gap of~9 meV were clearly identified across the Fermi level. Considering the facts of previous observation of hole-doping-induced superconductivity in the Bi 2 Te 3 /FeTe heterostructure and the superconductivity observed in 10UC FeTe/STO sample after long-time exposure to air, we conclude that the gap on FeTe : O 2 is just a superconducting gap with ∆~4.5 meV, larger than the one on Bi 2 Te 3 /FeTe heterostructure (~2.5 meV) [32,34]. It should be due to the interfacial effect on STO substrate because similar enhanced superconductivities have been observed on the Pb islands [35] and monolayer FeTe x Se 1-x [1,2,25]. We have studied the spatial distribution of the superconducting gap on the FeTe : O 2 islands in Figure 4c. Pronounced coherent peaks can be observed in all the dI/dV curves.  The observation of enhanced superconducting gap on FeTe : O2 islands is intriguing and distinct from all the previously proposed mechanisms. Physically adsorbed O2 molecules rather than interstitial oxygen or oxygen substitution are responsible for superconductivity in oxygenated FeTe. Previously, annealing under an elevated temperature or The observation of enhanced superconducting gap on FeTe : O 2 islands is intriguing and distinct from all the previously proposed mechanisms. Physically adsorbed O 2 molecules rather than interstitial oxygen or oxygen substitution are responsible for superconductivity in oxygenated FeTe. Previously, annealing under an elevated temperature or quite a long-time aging may be aimed at the formation of a uniform O 2 overlayer into the FeTe layer. Our study here also emphasizes the importance of the hole doping effect and interfacial effect to induce superconductivity on FeTe. We have not observed a superconducting gap on FeTe:O 2 overlayer on monolayer FeTe, which may be due to the extra electron doping from the STO substrate.

Conclusions
In summary, we have systematically investigated the changes of morphology and electronic structure of 10UC FeTe films after O 2 adsorption. Ex-situ transport measurements demonstrate long-term exposure to air could induce superconducting transition in 10UC FeTe ultrathin film, which verifies the effect of oxygen incorporation. Initially, O 2 molecules physically adsorbed on the FeTe surface and formed compacted FeTe : O 2 islands with c(2/3×2) structure at low temperatures. No oxygen-related defects can be detected on the uncovered clean FeTe surface. Energy shifts in STS and XPS measurements point out obvious hole doping into FeTe from the oxygen overlayer. Such hole doping results in a superconducting gap of~4.5 meV opened with apparent coherent peaks. Our study here directly provides microscopical insights about the role of oxygen on FeTe thin film. Author Contributions: Conceptualization, F.L. and Y.W.; methodology, F.L. and W.R.; software, W.R.; validation, W.R., H.R. and K.P., H.L., S.L., A.C., P.W., X.F., Z.L., R.H. and L.W.; formal analysis, F.L., W.R., H.R. and Y.W.; investigation, F.L. and W.R.; resources, F.L. and Y.W.; data curation, F.L. and W.R.; writing-original draft preparation, F.L. and W.R.; writing-review and editing, F.L. and W.R.; visualization, F.L. and W.R.; supervision, F.L. and Y.W.; project administration, F.L. and Y.W.; funding acquisition, F.L. and Y.W. All authors have read and agreed to the published version of the manuscript.

Data Availability Statement:
The data presented in this study are available on request from the corresponding authors.