The high-temperature superconductor YBa₂Cu₃O_7-δ (YBCO) is still attracting intense attention.^[1,2] Particularly, ultrathin films of YBCO have attracted significant research interest since the discovery of YBCO.^[3–7] In the early days, a fundamental question as to what is the minimum thickness needed for the occurrence of superconductivity was raised because the layered structures of YBCO suggest a high degree of anisotropy.^[3] Interestingly, it has been demonstrated that while single YBCO thin films lose superconductivity quickly when their thickness decreases to a few unit cells (uc), superconductivity has been achieved in as thin as 1-uc-thick YBCO layer embedded in superlattices that comprise YBCO and nonsuperconducting PrBa₂Cu₃O₇ (PBCO)^[5–7] or sandwiches of PBCO/YBCO/PBCO.^[3,8,9] In addition to the above fundamental question, the major interest of ultrathin YBCO films lies in its promising application of electric-field-effect devices such as three-terminal high-temperature superconductor transistors.^[10] Compared with conventional Bardeen–Cooper–Schrieffer (BCS) superconductors, YBCO has a much smaller carrier density, 3×10²¹/cm³–5×10²¹/cm³, which makes it intriguing to control superconductivity by electric field.^[11] For this purpose, ultrathin YBCO films are needed because the tuning ability of the field effect is generally small. In addition, ultrathin YBCO films can also be used to construct an ultra-sensitive YBCO-based single-photon detector.^[12]

As far as field effect application is concerned, unfortunately, superlattices^[5–7] and sandwiches^[3,8,9] are not very useful because an electric field can hardly penetrate these structures. For the optimization and future applications of YBCO-based field-effect devices, a high-quality ultrathin YBCO film that is directly accessible to gating materials is crucial.^[10] Over the past two decades, electric-field effects on YBCO have been realized by either back gating through SrTiO₃^[13–15] or top gating through SrTiO₃,^[16,17] ferroelectrics,^[18–20] or ionic liquids.^[21–23] The ultrathin YBCO films used in these studies are either single^[13,14,23] or buffered with PBCO.^[17–20,22] Compared with the single YBCO, the PBCO-buffered YBCO has much better superconducting properties, and thus has been widely used in top-gating devices.^[17–20,22] However, this strategy cannot be applied to the SrTiO₃-based back gating devices because PBCO, rather than YBCO, is directly contacting with SrTiO₃, the gating material. Instead, one would expect that a high-quality ultrathin YBCO film that is made directly on SrTiO₃ and covered with a suitable capping layer should work well for the back gating. Despite the intense previous studies of YBCO,^[24] little attention has been paid to optimizing ultrathin YBCO films in this way.

In this work, we explore the possibility of optimizing ultrathin YBCO films directly grown on SrTiO₃ by controlling the termination of SrTiO₃,^[25] and by capping nonsuperconducting oxides whose structures are compatible with YBCO. Our result shows that the superconductivity of YBCO is insensitive to the terminations of SrTiO₃, but can be generally enhanced by capping different oxides, including dielectrics such as SrTiO₃ and LaAlO₃.

All the films were grown on TiO₂-terminated (001)-oriented SrTiO₃ substrates by pulsed laser deposition using sintered polycrystalline targets. A KrF excimer laser (248 nm) was used with a laser fluence of 1.0 J/cm² and a repetition rate of 1 Hz. SrO-terminated SrTiO₃ was obtained by inserting 1-uc SrO on the TiO₂-terminated SrTiO₃ before growing YBCO (as shown in Fig. 1). The growth of YBCO was performed at 750 °C, under a mixed gas atmosphere of 0.03-mbar N₂O and 0.10-mbar O₂ (1 bar = 10⁵ Pa). To avoid disturbing the properties of YBCO, all the capped oxides were deposited using the same growth conditions as YBCO. After growth, the samples were postannealed in situ at 500 °C, under 500-mbar O₂, for 30 minutes or 60 minutes, and then cooled to room temperature under the same atmosphere. The transport properties were measured using a standard four-probe method with evaporated silver electrodes.

Fig. 1.

Figure Option ViewDownloadNew Window

Fig. 1. (a) RHEED intensity oscillations and (b) atomic force microscope (AFM) image of a 5-uc YBCO film grown on TiO2-terminated SrTiO3 substrate. (c) RHEED intensity oscillations and (d) AFM images of a 5-uc YBCO film grown on SrO-terminated SrTiO3 substrate that is made by inserting 1-uc SrO layer on the TiO2-terminated SrTiO3. Insets in panels (a) and (c) show the sample configuration. The root mean square roughness of panels (b) and (d) is 0.41 nm and 0.52 nm, respectively. Lower panels in panels (b) and (d) show the typical height profiles along the white cross lines.

First, we examine the effect of SrTiO₃ termination on the properties of ultrathin YBCO films. Previous studies showed that the termination of SrTiO₃ has a dramatic influence on the transport properties of LaAlO₃/SrTiO₃ heterostructures^[26] and manganite films,^[27] because it changes the sequence of atomic layers in LaAlO₃ or manganite, which in turn changes charge transfer within the heterostructures or films. By analogy, in YBCO, charge transfer between different atomic layers plays a key role in superconductivity. Therefore, changing the termination of SrTiO₃ may also have an important influence on the properties of YBCO.^[28]

Figure 1 shows the typical growth processes and surface topology of 5-uc YBCO films grown on TiO₂- and SrO-terminated SrTiO₃ substrates, respectively. The thickness of each layer is monitored by reflection high energy electron diffraction (RHEED) (Figs. 1(a) and 1(c)). Both types of surface are very smooth (Figs. 1(b) and 1(d)). The root mean square roughness of the TiO₂- and SrO-terminated samples are 0.41 nm and 0.52 nm, respectively. The cross-line profiles show that the roughness mainly comes from the height difference around 1 uc or 0.5 uc of YBCO. These characterizations confirm that our ultrathin YBCO films are of good quality.

In Fig. 2, we present the dependence of resistance on temperature for YBCO films of different thicknesses. As expected, as the YBCO thickness decreases, the critical temperature T_c decreases and the transition region broadens. This kind of thickness-dependent behavior can be attributed to several factors such as Kosterlitz–Thouless (KT) transition,^[3] strain,^[29] or defects. The superconductivity is fully suppressed when YBCO is thinner than 4 uc. One important observation is that the YBCO films grown on SrO-terminated SrTiO₃ (dotted lines) actually have quite similar transport properties as those grown on TiO₂-terminated SrTiO₃ (solid lines). Therefore, we conclude that simply changing the termination of SrTiO₃ cannot enhance superconductivity of the ultrathin YBCO films. In the following surface-capping studies, we will only focus on the YBCO films grown directly on the TiO₂-terminated SrTiO₃ substrates.

Fig. 2.

	Figure Option ViewDownloadNew Window
	Fig. 2. The dependence of resistance on temperature for uncapped YBCO thin films of different thicknesses. The solid and dotted lines represent samples grown on TiO2- and SrO-terminated SrTiO3 substrates, respectively. The postannealing time for all these samples is 60 minutes.

Next, we examine the effect of surface capping on the properties of ultrathin YBCO films. Unlike the buffer layer, the choice of the capping layer is almost unlimited because it is on the top and thus does not obviously influence the crystalline structure of the underneath YBCO. In the present study, in order to obtain a high-quality capping layer, we limit our choice to a few oxides whose crystalline structures are compatible with YBCO. In addition, we have reduced the postannealing time from 60 minutes to 30 minutes to optimize the surface quality of the capped samples. Most of these capped samples still have a quite flat surface with a root mean square roughness below 1 nm (not shown). Figure 3 shows the typical transport results of capping 5-uc YBCO with ∼7-nm different oxides. The most important observation is that almost all of these capped oxides enhance the superconductivity of the YBCO films, more or less.

Fig. 3.

Figure Option ViewDownloadNew Window

Fig. 3. (a) The dependence of resistance on temperature for 5-uc YBCO films capped with different oxides. The thickness of all the capping layers is around 7 nm. The result of the uncapped YBCO is also shown for comparison (the dashed line). The inset replots the result of the uncapped and the LaAlO3-capped samples in the linear coordinates. (b) The critical temperature, Tc, generated from the R–T curves shown in panel (a). Here, Tc is defined as the transition onset that is determined as the position of the maximum of the derivative curve, . The postannealing time for all these samples is 30 minutes.

The dashed line in Fig. 3(a) represents the R(T) curve of the uncapped 5-uc YBCO film. Note that it has a lower T_c and a broader transition region than the 5-uc YBCO shown in Fig. 2. This difference is due to the different postannealing time in oxygen because YBCO is very sensitive to oxygen content. As stated above, all the capped samples shown in this study are postannealed for 30 minutes. Therefore, we compare them with an uncapped sample that has been postannealed in the same condition. We emphasize that this uncertainty in the uncapped 5-uc YBCO film does not affect our main conclusion. To give a more quantitative comparison, we define T_c as the transition onset that is determined as the position of the maximum of the derivative curve, .^[30] The result is summarized in Fig. 3(b). As shown in Figs. 3(a) and 3(b), after capping PBCO or La₂CuO₄, the superconducting properties become much better as indicated by a much higher T_c and a narrower transition region. Similar but smaller enhancements are observed in samples capped with LaMnO₃ and La_0.67Ca_0.33MnO₃. The transition regions of the samples capped with La_1.56Sr_0.44CuO₄, and SrTiO₃ are relatively broad, but their T_c values are clearly enhanced. An enhancement of T_c is also observed for the LaAlO₃-capped sample, although the transition is very broad. This point can be better appreciated if the data are plotted in the linear coordinates (inset of Fig. 3(a)).

Why can the superconducting properties of ultrathin YBCO films be enhanced by these capped oxides? One general possibility is that the capped oxides can work as a protection layer. It is well known that the surface of YBCO is chemically instable when it is exposed to air, mainly due to the reaction with water humidity and carbon dioxide. These deteriorations can be avoided by capping a protection layer. However, if this mechanism dominates, one would expect that all the capped oxides will enhance the superconductivity to a similar extent, which is clearly inconsistent with the result shown in Fig. 3. Capping-induced strain or optimization of the surface structure of YBCO may be excluded since it has been shown that the effect of capping is similar for crystalline or amorphous PBCO.^[8] Therefore, the most likely scenario is charge transfer. That is, surface capping increases the hole concentration of YBCO. This scenario is supported by the observation that even the normal-state resistance of the capped samples drops a lot (Fig. 3), which cannot be simply attributed to the parallel conduction from the capped oxides.

We consider two charge-transfer processes. For the single YBCO film, the defects on its surface can easily react with water humidity and carbon dioxide in air, which can generate high-energy-level electrons that will be transferred into the YBCO.^[31] This process will reduce the hole concentration of YBCO and thus deteriorate the superconductivity. This process can be largely suppressed by surface capping. This is the reason why almost all the capped oxides can enhance superconductivity. The second charge-transfer process takes place between the capped layer and the YBCO. This process determines the apparent difference of capping different oxides. Hole injection by capping PBCO has been discussed in previous studies.^[3] Besides PBCO, La₂CuO₄, La_1.56Sr_0.44CuO₄, LaMnO₃, and La_0.67Ca_0.33MnO₃ are either a parent insulator of hole-type conductor or a hole-type conductor. They can provide holes to, or at least will not annihilate too much holes in, the YBCO. Therefore, the net effect of capping these materials is a significant enhancement of superconductivity in both T_c and transition region. The slightly lower T_c of LaMnO₃- and La_0.67Ca_0.33MnO₃-capped samples may be partially attributed to the interaction between superconductivity and magnetism.^[32] SrTiO₃ and LaAlO₃ have different physical properties from others and cannot supply holes to YBCO. Particularly, the surface of LaAlO₃ tends to react with water humidity. The high-energy-level electrons generated at this surface can still transfer across LaAlO₃ into YBCO.^[31] Therefore, it is reasonable that the net enhancement of superconductivity by capping SrTiO₃ and LaAlO₃ is relatively weak. The broad transition region observed in La_1.56Sr_0.44CuO₄-, SrTiO₃-, and LaAlO₃-capped YBCO samples is a signature of inhomogeneity of superconducting region. It might be related to the capping-induced local inhomogeneity in structure, charge transfer, and oxidization.

Finally, let us determine what is the thinnest YBCO film that is directly grown on SrTiO₃ and still shows superconductivity. Here, PBCO is used as the capping material because it enhances the superconductivity most effectively (Fig. 3). Figure 4 shows the R(T) curves of PBCO-capped YBCO thin films of different thicknesses. As expected, the superconducting properties in all these films are significantly enhanced. The capped 3-uc YBCO still shows a complete superconducting transition. But the capped 2-uc YBCO only shows an onset of superconductivity transition. Therefore, a YBCO film of at least 2 uc–3 uc is necessary for back gating. Although it is still quite challenging to gate this YBCO with the SrTiO₃ substrate, Mannhart et al.^[13,14] have demonstrated a revised back-gating structure in which much larger tuning range can be obtained through the pre-deposited SrTiO₃ film. In addition, our present study suggests that depositing gating materials on the top of YBCO will enhance, rather than reduce, its superconducting properties. Therefore, dual gating, from both back and top, could be a useful strategy to control ultrathin YBCO by an electrical field.

Fig. 4.

	Figure Option ViewDownloadNew Window
	Fig. 4. The dependence of resistance on temperature of PBCO (∼7 nm)-capped YBCO thin films of different thicknesses. The inset shows the onset of Tc, determined as the position of the maximum of the derivative curve, , for the PBCO-capped YBCO thin films. The postannealing time for all samples is 30 minutes. The “10-uc uncapped” sample (dashed line) is regenerated from Fig. 2 for comparison.

In summary, we have studied the effect of both the SrTiO₃ termination and the nonsuperconducting capping layer on the properties of ultrathin YBCO films grown directly on SrTiO₃. Our result demonstrates that the superconducting properties of the ultrathin YBCO films are insensitive to the termination of SrTiO₃, but can be significantly enhanced by surface capping. This enhancement effect is very general and even applicable to dielectric materials. The minimum thickness of the superconducting YBCO ultrathin film obtained by capping is ∼2 uc–3 uc. This result may be of interest for designing more effective YBCO-based field effect devices.