Introduction

Iron-based superconductors has triggered great interest1 because of the high transition temperature2,3, ultra-high critical magnetic field4,5,6 and potential applications as a group of high TC superconductors7,8. FeSe, with the simplest structure and less toxicity in iron-based superconductors (among all five families), has become one of attractive materials but the TC is relatively low in bulk state (~8 K)9. In the meanwhile, heterostructure based interface engineering has been proved an effective method for raising TC due to the enhancement of electron-phonon coupling10 or epitaxial strain11. In previous work, in situ scanning tunneling microscopy/spectroscopy (STM/STS)12 and angle resolved photoemission spectroscopy (ARPES)13,14,15 detections on 1-UC thick FeSe films on Nb-doped STO (conductive STO) substrates revealed a superconducting energy gap as large as 20 meV and above 15 meV (closing at a temperature of 65 ± 5 K) respectively, indicating a possible TC higher than 60 K. Following, the TC above 40 K in 1-UC FeSe films on insulating STO has been demonstrated by direct transport measurements and Meissner effect7. However, direct evidence of TC above 60 K for FeSe films on conductive substrates and systematic transport studies of FeSe films on insulating substrates are still absent.

Experimental

Previous STM/STS study shows square-like 1 × 1 structure with the in-plane lattice constants a = b = 3.82 Å and 20 meV superconducting-like gap for the 1-UC FeSe films on STO substrates7. However, by STS detection, semiconducting behavior is observed for the second UC of the FeSe film on STO. Thus, carefully comparison of the transport properties of high quality 1-UC and 2-UC FeSe films on STO substrates becomes necessary. In this paper, we mainly report the electronic transport and diamagnetic results from five typical FeSe films grown by molecular beam epitaxy (MBE) system. Sample No.1 is 1-UC FeSe on conductive STO substrate, samples No.2 and No.3 are 1-UC FeSe films on insulating STO substrates, while samples No.4 and No.5 are 2-UC FeSe films on insulating STO substrates. The 10-UC FeTe protection layers are deposited before the ex situ measurements of the films12. All FeSe films are 1.5 mm wide and 8 mm long. The distance between the voltage electrodes for the measurement is 1.8 mm for sample No.2, 2.25 mm for sample No.3, 2.0 mm for sample No.4 and 2.75 mm for sample No.5. As for sample No.1, the diamagnetic measurement exhibits that the onset of magnetization drop starts around 85 K, indicating a possible superconductivity up to this high temperature. For samples No. 2–5, thermally activated flux flow (TAFF) behavior and Hall effect are carefully studied. It is found that single-vortex pinning dominates vortex dynamics in low magnetic field region, whereas collective creep becomes important at higher magnetic fields. More interestingly, sign reversal behavior of Hall coefficient (RH) with temperature is observed.

Results and discussions

It is worthy to mention that the observed TC of 1-UC FeSe on insulating substrate by ex situ transport measurements is obvious lower than estimated TC of 1-UC FeSe on conductive substrate by STM and ARPES studies12,13,14,15. Indeed, high quality FeSe films are easier to be achieved by MBE growth on conductive STO substrates comparing to insulating STO substrates since the conductive STO substrate shows more flat and homogeneous surface for sample growth. In order to see if higher TC of 1-UC FeSe can be revealed by direct measurements other than energy gap detection, we did magnetization experiments for 1-UC FeSe on conductive STO substrate (sample No.1) in a magnetic property measurement system (MPMS-SQUID-VSM). DC magnetization as a function of temperature (M-T) during both zero field cooling (ZFC) and field cooling (FC) at 1000 Oe parallel field of sample No.1 is shown in Fig. 1(a). The M-T curves exhibit a drop crossover around ~85 K. Please notice that the raw data shown in Fig. 1(a) include the contributions from FeSe film, protection layer and STO substrate. Thus, the background signal from the protection layer and substrate is necessary to be subtracted for more precise result. However, it is hard to keep the quality of STO substrates exactly same for magnetic measurements. Therefore, we designed an experiment to exclude the background signal. After we found the 85 K drop crossover of 1-UC FeSe, we put the sample in low vacuum system and waited for a long time until the sample lost its superconductivity. Then, we measured this sample again as the background signal as shown in the inset of Fig. 1(a). We can see that when the sample becomes non-superconducting, the magnetization drop around 85 K disappears. Thus, the possibility of the relation between the drop and the superconductivity cannot be excluded. However, without resistivity evidence of TC at 85 K, the susceptibility drop alone cannot demonstrate the high TC up to 85 K in 1-UC FeSe films. Since the susceptibility of FeTe capping layer may also change in the degraded sample, the magnetization curves excluding the background signal shows negative value even at 300 K (Fig. 1b). Interestingly, after subtracting the influence of STO substrate and FeTe protection layer from the same sample, the M-T curves still exhibit the magnetization drop crossover around 85 K with decreasing temperature (Fig. 1(b)). The structural phase transition (from a tetragonal to an orthorhombic phase on cooling) of bulk FeSe was ever observed around 90 K16,17, at which there also is strong antiferromagnetic spin fluctuation18. In the meanwhile, phonon softening is a general feature in Fe-based superconductors and induced by either structural phase transition or the superconductive phase transition18. Thus, the consideration for the phonon softening may be valuable for understanding our observation of magnetization drop crossover around 85 K in the 1-UC FeSe films, which show much enhanced superconductivity comparing with bulk FeSe.

Figure 1
figure 1

Diamagnetic measurements of 1-UC FeSe films grown on conductive STO (LR STO) substrate (sample No.1).

The M-T curves without or with removing the influence of substrate and protection layer show a drop crossover around 85 K. (a) M-T curves of sample No.1 (LR STO/1-UC FeSe/10-FeTe) measured under a 1000 Oe parallel magnetic field. Inset: M-T curves of LR STO/non-superconducting 1-UC FeSe/10-UC FeTe. (b) M-T curves of 1-UC FeSe (removing the influence of STO substrate and FeTe protection layers).

The electronic transport measurements were carried out in a physical property measurement system with the magnetic field up to 16 T (PPMS-16T). The obtained superconducting parameters of the four measured samples are summarized in Table 1. Figure 2 shows the transport results of sample No.2, one typical 1-UC FeSe film grown on insulating STO substrate covered by non-superconducting FeTe protection layers with an excitation current of 500 nA. Fig. 2(a) shows the sheet resistance of the sample as a function of temperature Rsq(T) at zero magnetic field (μ0H). The resistance begins to drop at about 54.5 K. By extrapolating both the normal resistance and the superconducting transition curves, we obtain the onset and the resistance drops completely to zero at 23.5 K (). The zero resistance is defined when the measured voltage is within the instrumental resolution ± 20 nV. Figure 2(b) shows Rsq(T)curves at different perpendicular magnetic fields (μ0H) up to 16 T from 2 K to 60 K. The resistive transition becomes broader and shifts to lower temperatures with increasing magnetic field, characteristic of superconducting transition in thin films.

Table 1 Summary of the parameters of four FeSe samples on insulating STO. was obtained by extrapolating both the normal resistance and the superconducting transition curves. with star is the temperature at which the resistance starts to decrease. Since previous STM study11 indicates that the second UC of FeSe films grown on STO substrates shows semiconducting behavior and only the first UC FeSe is superconducting, JC of the four samples at 2 K is calculated from IC by using the thickness of 0.55 nm (1-UC FeSe)
Figure 2
figure 2

Transport measurement of the 1-UC FeSe film grown on insulating STO (HR STO) substrate (sample No.2).

(a) The temperature dependence of sheet resistance under zero field, showing and . (b) The temperature dependence of sheet resistance under various perpendicular magnetic fields up to 16 T, showing a typical broadened superconducting transition. (c) ln[Rsq(Ω)] vs. 1/T in various perpendicular magnetic fields. The corresponding solid lines are fitting results from the Arrhenius relation. (d) Field dependence of U0 (H). The solid lines are power-law fits using U0 (H) ~ H−α. For this 1-UC FeSe film, α = 0.14 for μ0H < 3.4 T and α = 0.60 for μ0H > 3.4 T.

We know in the thermally activated flux flow (TAFF) of vortex region, the lnρ− 1/T can be described by Arrhenius relation19,20

where ρ0 is a temperature independent constant and U0(H) is the activation energy of the flux flow. Thus, lnρ(T,H) versus 1/T should be linear in the TAFF region. As shown in Fig. 2(c), the experimental data in Fig. 2(b) can be well fitted by the Arrhenius relation (solid lines). The fitting lines obtained from lnρ(T,H) versus 1/T at different magnetic fields can be well extrapolated to the same temperature, Tm = 38 K, which is close to the value of . In addition, the activation energy U0 for different magnetic field can be obtained by the slope of the solid fitting lines. As shown in Fig. 2(d), the U0(H) shows a magnetic field dependent power law relation,

with α = 0.14 (μ0H < 3.4 T) and 0.60 (μ0H > 3.4 T), respectively.

For the 2-UC FeSe film grown on insulating STO substrate covered by non-superconducting FeTe protection layer (sample No.4), Fig. 3(a) shows the sheet resistance as a function of temperature Rsq(T) at zero magnetic field (μ0H). By extrapolating both the normal resistance and the superconducting transition curves, we obtain the onset and the resistance drops completely to zero at 22.5 K (), which are almost same with those of 1-UC FeSe (sample No.2). Figure 3(b) shows the Rsq of sample No.4 as a function of temperature, Rsq(T), at different perpendicular magnetic fields (μ0H). The Rsq(T) curves at different magnetic fields can also be fitted by the Arrhenius relation (solid lines) well as shown in Fig. 3(c). The fitting lines obtained from lnρ(T,H) versus 1/T at different magnetic fields cross to one point at about Tm = 36 K, which is consistent with the TC of the sample. The activation energy U0 varies with different magnetic field can be achieved by the slope of the solid fitting lines as shown in Fig. 3(d), α = 0.125 (μ0H < 3.7 T) and 0.79 (μ0H > 3.7 T) respectively by utilizing Eq. (2). These values obtained from the two samples are close. At the temperatures not far from TC, the weak power law decreases of U0(H) in low magnetic fields for both samples implies that single-vortex pinning dominates in this region, followed by a quicker decrease of U0(H) at high field where a crossover to collective flux creep regime occurs. The similar behaviors are observed in iron-based superconductors, such as FeSe21,22, FeTeSe23 and Nd(O,F)FeAs24 crystals. The exponent α = 0.5 and 1 corresponds to a planar-defect pinning and a point-defect pinning in high TC superconductors respectively25. For our FeSe ultrathin films, the fitted values obtained at high magnetic field vary between 0.5 and 1, suggesting that the pinning centers may be mixed with point and planar defects. The cross-over magnetic field is about 3 T for bulk FeSe, 2 T for FeTeSe and 3 T for Nd(O,F)FeAs, which are almost in the same level with our observation in ultrathin FeSe films. For cuprates, the range of the cross-over magnetic field is from 0.8 T - 5.5 T, where the exact value depends on the sample situations, such as defects and boundaries.

Figure 3
figure 3

Transport measurement of the 2-UC FeSe film grown on HR STO substrate (sample No.4).

(a) The temperature dependence of sheet resistance under zero field, showing and . (b) The temperature dependence of sheet resistance under various perpendicular magnetic fields. (c) ln[Rsq(Ω)] vs. 1/T in various perpendicular magnetic fields. The corresponding solid lines are fitting results from the Arrhenius relation. (d) Field dependence of U0 (H). The solid lines are power-law fits using U0 (H) ~ H−α. For this 2-UC FeSe film, α = 0.125 for μ0H < 3.7 T and α = 0.79 for μ0H > 3.7 T.

The Hall resistance (Rxy) of FeSe films is measured by sweeping the magnetic field at a fixed temperature. The temperature stabilization is better than 0.1%. The distance between the Hall voltage electrodes is about 1.5 mm for all measured samples. Figures 4(a)–4(b) show the data of Rxy vs magnetic field (Rxy(H)) at different temperatures from 40 K to 150 K of samples No.2 and No.4, which exhibit good linear relation. In order to subtract the influence of FeTe protection layer, transport properties of reference sample (10 UC FeTe grown on insulating STO substrate) were studied. The sheet resistance of the reference sample varies with temperature (Rsq(T)) under different magnetic fields (0 T, 9 T and −9 T) are exhibited in Fig. 4(c). Figure 4(d) shows the Rxy(H) curves of the reference sample at different temperatures. The Hall conductance of pure FeSe films (removed the influence of FeTe layers) can be calculated by . After removing the influence of FeTe protection layer, the Hall resistance (Rsq(xy)-FeSe) of the pure FeSe film sample is re-obtained by conversing from Hall conductance of FeSe films, (where ), which are plotted in Figs. 5(a) and 5(c) for sample No.2 and No.4, respectively.

Figure 4
figure 4

Hall results of FeSe films and FeTe protection layers.

(a) Hall resistance (Rxy) varies with magnetic field — Rxy(H) at different temperatures of sample No.2. (b) Rxy(H) curves at different temperatures of sample No.4. (c) Rsq(T) curves of the 10-UC FeTe protection layer under different perpendicular magnetic fields (0 T, 9 T and −9 T). (d) Rxy(H) curves at different temperatures of the 10-UC FeTe protection layer.

Figure 5
figure 5

Hall results of sample No.2 (1-UC FeSe films grown on HR STO) and sample No.4 (2-UC FeSe films grown on HR STO) after subtracting the influence of the protection layer.

(a) & (c), Rxy vs magnetic field curves at different temperatures of sample No.2 and No.4 subtracting the background of protection layers, respectively. (b) & (d), Hall sensitivity and the carrier density of sample No.2 and No.4 at different temperatures obtained from the data in (a) and (c).

In Fig. 5(b), the temperature dependence of the Hall sensitivity (VH is the Hall voltage, d is the thickness of the sample) and two-dimensional carrier density (, q is the charge of electron) are plotted. For a normal metal with Fermi liquid feature, the Hall coefficient is constant at different temperatures. However, it varies with temperature for multiband materials, such as MgB226, iron-based superconductors27, or a sample with non-Fermi liquid behavior such as cuprate superconductors28. Here, an interesting phenomenon—sign reversal behavior of is observed in the 1-UC FeSe film (sample No.2) with temperature increasing (shown in Fig. 5(a) and 5(b)). In Fig. 5(a), the slope of the Rxy(H) curves changes from negative to positive as the temperature is higher than 100 K. Correspondingly, the signs of and ns reversed at the same time as shown in Fig. 5(b). That is to say, as the temperature is lower than 100 K, the 1-UC FeSe film is electron-doped, but it is hole-doped while the temperature higher than 100 K. As for 2-UC FeSe film, Figure 5(c) shows the Rxy(H) curves of sample No.4 at different temperatures from 40 K to 150 K. The Hall sensitivity for sample No.4 decays continuously with increasing temperature as shown in Fig. 5(d). The value of at 150 K is almost 10 times smaller than that at 40 K, exhibiting strong temperature dependence of . This behavior is possibly induced by the multiband effect. For sample No.4, the Hall data manifest electron-doped property as increasing temperature till 150 K. Please notice that the carrier density (ns) in FeSe films is pretty huge (~1015 cm−2).

In order to confirm that the sign reversal of is a universal behavior in ultrathin FeSe films, Hall resistance from another two samples (sample No.3 is 1-UC FeSe film and sample No.5 is 2-UC FeSe film) at various temperatures up to 300 K was measured. Figure 6(a) and 6(b) are the raw Rxy(H) data from sample No.3 and No.5. The signs of are obviously reversed in these two samples with increasing temperature. After subtracting the influence of FeTe protection layer utilizing the same method mentioned above, the Rxy curves of pure FeSe films at different temperatures are shown in Fig. 6(c)–6(f) with similar sign reversal behavior of . The parameters of Hall effect for all four samples can be found in Table 2 & 3. Thus, the crossover of conduction carrier type with temperature is demonstrated to be a universal phenomenon for ultrathin FeSe films. The reversal temperature is 80 K or 150 K for 1-UC FeSe films (sample No.3 or sample No.2) but 150 K or larger for 2-UC FeSe (sample No.5 or sample No.4).

Table 2 Hall sensitivity carrier density and mobility of four FeSe samples on HR STO at different temperatures (with FeTe protection layer)
Table 3 Hall sensitivity carrier density and mobility of four samples on HR STO subtracting the influence of FeTe protection layer at different temperatures (subtracting FeTe affection)
Figure 6
figure 6

Hall results of sample No.3 (1-UC FeSe films grown on HR STO) and sample No.5 (2-UC FeSe films grown on HR STO).

(a) Rxy(H) curves at different temperatures of sample No.3. (b) Rxy(H) curves at different temperatures of sample No.5. (c) & (d) Rxy(H) curves at different temperatures of sample No.3 and No.5 subtracting the background, respectively. (e) & (f) Hall coefficient and the carrier density of sample No.3 and No.5 at different temperatures after subtracting the background.

Interestingly, we found that the sign reversal behavior of the Hall results of our FeSe films always happens at the temperature before the hump of the Rsq(T) curves, thus, a possible structural transition of FeSe film may contribute to the observed sign reversal phenomenon16,17,29. Besides, we notice that similar sign reversal behavior was also observed in topological insulator Bi2Te330 and was attributed to the change of defects in the sample. Further experimental or theoretical investigations in this context would be valuable towards the comprehension of the observed sign reversal Hall effect in 1-UC FeSe films.

The origins of the huge interface-induced enhancement of TC in 1-UC FeSe films compared with bulk FeSe are still under debate and many possible factors may contribute. An early magnetic study on the single-crystalline Fe1.02Se reveals that TC can be increased from 7 K to above 30 K with the modest pressure range below 2 GPa31. Additionally, doubling the critical temperature of La1.9Sr0.1CuO4 using epitaxial strain was also reported11. However, for 1-UC FeSe films on STO, the in-plane lattice constant measured from the atomically resolved STM image is ~3.82 Å7, which is a little larger than the bulk value (3.77 Å) while still smaller than the in-plane lattice constant of STO (001) surface (3.91 Å), indicating that the tensile train in the 1-UC FeSe films is less than 1%. Furthermore, FeSe and Fe(Se,Te) films grown on different substrates with larger lattice constants, such as MgO, Si and Al2O3, were studied32. The observed TC of these films is not higher than 20 K although the tensile strain in these films is much larger than our situation. Therefore, we conclude that the small tensile strain alone in 1-UC FeSe films cannot explain the observed high TC superconductivity. As has been revealed by recent in situ ARPES study15 and calculation33,34, the charge transfer from the oxygen vacancy states of the STO substrate may play an important role in the high TC superconductivity33. Additionally, the soft phonons at the interface can enhance the energy scale of Cooper pairing and even change the pairing symmetry, then may increase the TC34,35.

In summary, superconducting properties of ultra-thin (1-UC or 2-UC) FeSe films grown on insulating STO and conductive STO substrates were studied by transport and magnetic measurements. M-T observation of FeSe films on conductive STO shows a susceptibility drop ~85 K. Further transport investigations are necessary to see if the drop is corresponding to the superconductivity. The results from films on insulating STO reveal Arrhenius TAFF behavior with a transition from single-vortex pinning region to collective creep region. More intriguingly, the observed sign reversal of Hall coefficient above TC demonstrates a crossover from hole transport to electron conduction in ultrathin FeSe films with decreasing temperature.