Suppression of superconductivity at the nanoscale in chemical solution derived YBa2Cu3O7−δ thin films with defective Y2Ba4Cu8O16 intergrowths

The analysis of the microstructure and superconducting behavior of chemical solution deposited epitaxial YBa2Cu3O7−δ films, with thickness going down to 5 nm has been carried out with the purpose to disclose the behavior of the most common intergrowth in these films, the Y2Ba4Cu8O16. The analysis of ultrathin films is a unique opportunity to investigate the superconducting behavior of these nanoscale defects because of the high concentration created as a consequence of the elastic energy associated to the misfit strain. Magnetic susceptibility and X-ray diffraction measurements evidence a strong decrease of the superconducting volume correlated with an increase of the intergrowth volume fraction. We demonstrate that these intergrowths are non-superconducting nanoscale regions where Cooper pair formation is disrupted, in agreement with their key role as artificial pinning centers for vortices in YBa2Cu3O7−δ films and coated conductors.


Introduction
Since the discovery of high temperature superconductors (HTS) there has been an intensive analysis of the complex relationship between lattice structure, defects and superconducting properties. 1 This correlation, however, becomes even more relevant when one intends to develop high current superconducting tapes and wires for power applications where the defect structure strongly inuences the most relevant superconducting parameter, i.e. critical current density J c (H,T). 2 Aer having achieved epitaxial lms and coated conductors (CCs), i.e. HTS lms grown on biaxially buffered metallic substrates not inuenced by grain boundary disorder, the main issue has become to understand which defects behave as effective articial pinning centers (APC) of vortices and hence increase J c (H,T) at high temperatures and magnetic elds. 3,4 Vortex pinning in HTS is actually a complex research topic because one needs to correlate the intricate physical behavior of vortices with the defect landscape of these materials in order to disclose how the different defects contribute to pin vortices at different temperatures, magnetic elds and eld orientations. [5][6][7][8][9] In the case of YBa 2 Cu 3 O 7Àd (Y123) lms and CCs extensive analyses have been carried out to sort out how to enhance vortex pinning. Particularly, huge progress has been achieved with the development of several processing approaches to produce nanocomposite lms where secondary nanometric nonsuperconducting phases coexist with the Y123 matrix. 7-13 A very relevant issue in this eld has been to disclose how the Y123 matrix is modied by the secondary phases, such as the perovskites BaZrO 3 , BaHfO 3 , Ba 2 YTaO 6 , or other nonsuperconducting oxides, and how they inuence vortex pinning. 7,8,10,12,14,15 These secondary phases leads to the formation of Y 2 Ba 4 Cu 8 O 16 (Y248) intergrowths having an extra Cu-O layer 16 so extensive efforts to disclose its inuence on vortex pinning has been undertaken. 3,9,10 Chemical solution deposition (CSD) has been demonstrated to be a very attractive technique for large-scale production of Y123 lms and CCs owing to its cost effectiveness advantage. 17,18 CSD is a prototypical example of Y123 epitaxial growth where secondary phases in the form of nanoparticles can be easily included. 12,15,[19][20][21][22][23] They usually remain randomly oriented, generating a high interfacial energy which partially relaxes through the formation of induced defects such as the Y248 intergrowths. 10,24 Particularly, in CSD nanocomposites it has been shown that vortex pinning is strongly enhanced by the concentration of these Y248 intergrowths. 9,10,25 The structure of the planar Y248 intergrowth defects has been recently analyzed in detail by Scanning Transmission Electron Microscopy (STEM). The expected composition of the Y248 intergrowth is Y 2 Ba 4 Cu 8 O 16 because they consist on an extra Cu-O chain layer inserted between two Ba-O layers, leading to a lattice expansion in c-axis direction (from 11.7 to 13.3Å). 16,26 Considering the stoichiometric ratio for a pristine Y123 lm (i.e. Y : Ba : Cu ¼ 1 : 2 : 3), the formation of these intergrowths has been recently described as double chains including defect clusters formed by two Cu vacancies decorated by three O vacancies. 27 The stoichiometric Y248 phase (without vacancies) is superconducting with a lower T c 28 than the Y123 phase. However, the superconducting performance of the Cu off-stoichiometric Y248 phase is still unknown because these planar defects are dispersed within CSD Y123 lms, as it was shown in previous reports, 10,19 and hence it becomes very difficult to sort out their superconducting properties. Actually, recent studies have shown, on one hand, that ferromagnetic clusters are formed around the Cu vacancies defects in the double chains and so one should wonder if this causes a pair breaking effect. 27,29 On the other hand, it has also been noticed that the defective double chains induce distortions at the nanoscale, in the neighboring CuO 2 planes, and oxygen vacancies and also generate highly strained localized areas at the partial dislocations surrounding the Y248 intergrowth, 23 as detected by atomic scale STEM studies. 30,31 Both of them may induce Cooper pair breaking effects. 30 In order to have a better understanding of the superconducting performance of this type of defective Y248 intergrowths it is crucial to grow Y123 epitaxial lms with a high concentration of them.
In this paper, we present an analysis of the micro/ nanostructure and superconducting properties of Y123 epitaxial lms of different thicknesses grown by CSD where a high concentration of Y248 intergrowths is developed when ultrathin lms are formed. Y123 lms with thickness down to 5 nm have been successfully fabricated using an optimized CSD growth process allowing to reach such small thickness keeping a high lm homogeneity. Actually, ultrathin Y123 lms have been recently used as seed layer to improve the epitaxial quality of nanocomposite lms and this provided some hints that a high concentration of Y248 could be achieved. 10,20,21,23 Our purpose here is to maximize this opportunity to depict the inuence of this microstructural defect on the superconducting properties. We show rst by X-ray diffraction and Scanning Transmission Electron Microscopy (STEM) that, indeed, a high concentration of Y248 intergrowths is achieved in Y123 ultrathin lms. We show then that these defects have a remarkable inuence on the superconducting properties of the lms. Particularly, we discern a linear relationship between the Y248 intergrowth concentration and the superconducting volume, determined from magnetic shielding measurements. Our results allow to infer the non-superconducting character of the Y248 intergrowths observed in Y123 thin lms. 32-35

Experimental section
The Y123 precursor solution was prepared by the reaction from solid Y123 ceramic powders (yttrium-barium-copper oxide, Solvay) with triuoroacetic anhydride as described in detail elsewhere. 36 The original obtained anhydrous precursor solutions were diluted from 1.5 M to 0.3-0.03 M using anhydrous methanol for the purpose of achieving an adjustment of lm thicknesses, ranging from 250 nm to 5 nm. Aer depositing the solutions on 5 Â 5 mm 2 commercially available (100) LAO or STO single-crystal substrates by spin-coating at a typical rotation speed of 6000 rpm for 2 min, the low-temperature (310 C) pyrolysis process 37 in a humid oxygen atmosphere was conducted to prepare solid precursor lms. We have found that an improved lm quality is achieved on LAO substrates where lattice mismatch induces an in-plane compressive strain to Y123 (f LAO ¼ (a LAO À a Y123 )/a Y123 ¼ À1.6%), as compared to STO substrates which induce an in-plane tensile strain to Y123 (f STO ¼ +1.3%) (see ESI †). Therefore, our main analysis was performed on lms grown on LAO substrates.
In the following crystallization step, an optimized thermal process was necessary to be developed to achieve ultrathin lms with enough good quality (see ESI †). Flash Heating (FH) is a novel recently developed process based on high heating ramps ($30 times faster than in conventional thermal annealing (CTA), i.e. $750 C min À1 ), thus leading to a reduced total heating time #1 min. 23,25 A series analysis of several processing parameters, e.g. heating process, crystallization temperature and thermal annealing time, were carried out to avoid dewetting effects in ultrathin lms. It is known that dewetting effects are promoted at long annealing times and high temperature annealing. 38 To minimize the detrimental dewetting effects we used FH. The optimal annealing temperature was found at 810 C. At lower temperatures the lms have a tendency to include large pores and misoriented grains. At higher temperatures, some secondary phases are identied by SEM-EDX and X-ray diffraction. The crystallization stage was performed during 20 min in a wet N 2 -0.02% O 2 mixed gas atmosphere with a water partial pressure (P(H 2 O)) of 23 mbar which was introduced at 110 C. Aer that, an extra 10 min dwell was proceeded at the crystallization temperature in dry N 2 -0.02% O 2 gas to minimize lm imperfections which generated from the grain boundary zipping, mist strain, porosity, etc. 18,39 Reasonably smooth ultrathin lms with uniform lm thickness (5-15% variation depending on thickness) were only achieved using FH and short enough annealing times, otherwise lm dewetting was originated, as evidenced by SEM, TEM and AFM images (see ESI †). Finally, the oxygenation process of the well crystallized and grown lms were performed at 550 C for 3 h in dry oxygen atmosphere.
Film surface morphology was studied by scanning electron microscopy (SEM) in a planar view and Atomic force microscopy (AFM) analysis on tapping mode with a molecular imaging system. AFM images were processed and analyzed with the commercial soware package MOUNTAINS (Digital Surf); see Fig. S7 and S8. † The phase analysis and texture characterization of the fully converted Y123 ultrathin lms were carried out by twodimensional X-ray diffraction (XRD) patterns using a Bruker AXS GADDS diffractometer. As a supplement for the limited resolution of the GADDS system, a high-resolution XRD (HRXRD) q-2q scan using a Bruker-AXS (model A25 D8 Discover) X-ray diffractometer was also applied for the phase identication. Non-uniform r.m.s. strain (nanostrain) (3) was determined using the Williamson-Hall (WH) method 40,41 by analyzing the symmetric (00l) 2q Bragg diffraction integral breadth b acquired in a Siemens D5000 diffractometer. The tting was made following the following equation: where q is the Bragg angle, l a 1 is the wavelength of the Cu K a radiation and L t is the size of the coherent volume perpendicular to the scattering vector (c-axis in our case). Nanostrain 3 corresponds, therefore, to the disorder in (00l) plane separation along the c-axis.
In-plane and out-of-plane texture analysis were analyzed from the (103) Y123 phi-scan (f-scan) and (005) Y123 rocking curve (u-scan), respectively. The microstructural characteristics of Y123 ultrathin lms were described by scanning transmission electron microscopy (STEM) using a FEI Titan 60-300 microscope equipped with an X-FEG gun, a CETCOR probe corrector and a Gatan TRIDIEM 866 ERS energy lter operated in STEM mode at 300 kV. Superconducting properties were investigated from magnetization measurements performed with a superconducting quantum interference device (SQUID) magnetometer (Quantum Design, San Diego, CA) equipped with a 7 T magnet. Low eld ($0.2 mT) Zero Field Cooled (ZFC) temperature dependent magnetic susceptibility measurements with Hkc were used to determine the superconducting volume, T c and DT c . Critical current densities J c (H,T) with Hkc were determined from isothermal hysteretic magnetization measurements M(H,T) (see ESI †) or from temperature dependent remnant magnetization M(T) measurements, performed aer applying and suppressing a magnetic eld of 7 T to assure full eld penetration, to calculate J sf c (T). The Bean model approximation to thin discs, J c (H,T) ¼ 3M(H,T)/R, where R is the effective radius of the sample and M(H,T) is the hysteretic magnetization, was used to calculate the critical current densities. 42,43 Optimization of the annealing process of ultrathin lms was also based on the study of the isothermal magnetic eld dependence of the critical current densities J c (H) determined from isothermal magnetization measurements ( Fig. 1(c) and (d)). The isothermal critical current densities of the different thin lms were estimated from the recorded isothermal magnetization hysteresis loops, using the Bean model approximation to thin lms, J c (H,T) ¼ 3M(H,T)/R, where R is the effective radius of the sample. The hysteretic magnetization is M ¼ (m p À m n )/V, calculated from the positive and negative values of magnetic moment and V is the volume of the lm.
The large J c (H) values in Fig. 1(c) and (d) indicate that an improved homogeneity of the lms has been achieved, i.e. pores and dewetting have been minimized. Also the observation of a peak in J c (H) at nite magnetic elds ( Fig. 1(c) and (d)) has been previously attributed to granularity effects in porous thin lms, i.e. in lms exhibiting some dewetting for instance. [44][45][46] Fig. 1 Cross sectional STEM image of 50 nm Y123 ultrathin films grown during different annealing times or temperatures (a) FH 810 C with 20 min wet anneal and 10 min dry anneal process; (b) FH 810 C with 60 min wet anneal and 30 min dry anneal process; (c) magnetic field dependence of the critical current density measured at 5 K for the 50 nm Y123kLAO films grown by FH at 810 C following different wet + dry annealing times, as indicated in the caption, (20 + 10) (black square), (20 + 30) (red circle) and (60 + 30) (blue triangle); Films having the highest critical currents correspond to those not exhibiting dewetting, while the other ones show a progressive influence of dewetting; (d) magnetic field dependence of the critical current density measured at 5 K of the 50 nm Y123/LAO films grown by FH at temperatures of 750 C, 810 C and 830 C.
This journal is © The Royal Society of Chemistry 2020

Nanoscale Advances Paper
Optimally grown lms, instead, display a maximum in J c (H) at zero external magnetic eld.

Results and discussion
3.1. Structural characterization of the lms a X-ray diffraction study. Fig. 2 displays the X-ray diffraction (XRD) analysis of pristine Y123 lms with different thickness. A typical two dimensional q-2q XRD (GADDS) frame of pristine Y123 thin lms with thickness of 50 nm is shown in Fig. 2(a). Note that solely (00l) diffraction poles of Y123 are identied, indicating that the lms are epitaxial without any polycrystalline or randomly oriented Y123 grains. In Fig. 2(b) we present the high resolution XRD scans for lms with different thickness, ranging from 5 to 250 nm. It is observed that Y123 thin lms only show (00l) Bragg reections, demonstrating that c-axis oriented Y123 grains are obtained throughout the investigated lm thickness. This result suggests that c-oriented Y123 thin lm can be obtained in an extended thickness window down to 5 nm for CSD-based Y123 thin lms, in a similar thickness limitation compared with the vacuum-based deposition routes. [47][48][49] All the lms are pristine Y123 phase without any trace of residual secondary phases. Another detail we can note from the high resolution XRD plots ( Fig. 2(b)) is the shi of the (005) Y123 peaks to smaller angles when the lm thickness decreases below 25 nm, which indicates an increase in the c-axis lattice parameter, from 11.69Å to 11.87Å (Fig. 3(a)). 47 The observed maximum c-axis increase appears to be consistent  This journal is © The Royal Society of Chemistry 2020 Nanoscale Adv., 2020, 2, 3384-3393 | 3387

Paper
Nanoscale Advances with the Poisson's ratio n ¼ 0.314 of Y123 in the case of having a fully coherent epitaxy with the LAO substrate. 47,50 It's also worth to mention that the observed highest c-axis parameter (c ¼ 11.87Å) is larger than that of a fully oxygen decient Y123 phase (YBa 2 Cu 3 O 6 ) and so we cannot attribute this increase to an oxygen deciency of the ultrathin lm Y123 layers. 51 In Fig. 3(b) and (c), we present the Du and Df evolution with lm thickness, which gives us the estimation of texture quality in out-of-plane and in-plane direction, respectively. The Du increases from $0.5 to $1.2 when lm thickness decreases from 250 nm to 5 nm (Fig. 3(b)), revealing a decreased out-ofplane texture quality in our ultrathin lms. This tendency is similar to what other authors have been previously reported for Y123 lms deposited by sputtering method. 52 Moreover, the Df values were found to be constant (Df ¼ 1.0 AE 0.1 ) down to the minimum CSD-based Y123 lm thickness measured so far, i.e. 5 nm, indicating good in-plane texture quality. We also determined the evolution of nanostrain 3 in these lms and we realized that it increases in parallel with the c-axis expansion up to 3 ¼ 0.5% (see Fig. 3(d)) while at lm thicknesses above 50 nm a clear enhancement of nanostrain is detected for FH lms as compared to CTA lms. These results point us to the conclusion that the lattice cell is strongly distorted in CSD Y123 ultrathin lms on LAO. The increase of the c-axis parameter is due to the compressive mismatch with the substrate and, at the same time, we disclose that some disorder is generated in the periodicity along c-axis. Lattice increases have been extensively observed in vacuum grown Y123 ultrathin lms, attributing it either to the lattice mist induced lattice distortion 47,53,54 or to the oxygen content changes. 55,56 More detailed analysis of the structural disorder in FH CSD Y123 lms will be described hereinaer.
b Transmission electron microscopy analysis. Annular dark eld (ADF) STEM imaging investigations have been conducted in order to further disclose the particular nanostructural landscape of the Y123 ultrathin lms. Cross-sectional STEM images of Y123 thin lms are shown in Fig. 4. The STEM images of the 45-50 and 10 nm FH Y123 thin lms (Fig. 4(a), (c) and (e)) show that the Y123 lm has a high density of long intergrowths (horizontal dark stripes in the image), as compared to a CTA Nanoscale Advances Paper lm of 250 nm thickness ( Fig. 4(b)). Note that the Y248 intergrowths can have different homogeneity distribution throughout the whole cross-section. From a higher resolution Zcontrast STEM image, Fig. 4(d), it is clearly observed that these intergrowths consist on a structure having an extra Cu-O chain layer inserted within the normal Y123 matrix, and hence they are identied as the well-known Y248 phase (see structure identication in inset of Fig. 4(d)). 26,27 Occasionally, another type of intergrowth is also identied, indicated by arrows in Fig. 4(d), with two extra Cu-O chains being incorporated, resulting in a local composition of YBa 2 -Cu 5 O 8 (Y125) phase. 57 These Y248 and Y125 intergrowths are surrounded by partial dislocations which generate localized strained regions 57 reected in an enhanced inhomogeneous strain (nanostrain) of the Y123 structure along the c-axis, as measured by the inhomogeneous integral breadth of (00l) Bragg peaks in the X-ray diffraction patterns (Fig. 3(d)).
In coherent epitaxial lms, the lattice mismatch with the substrate leads to lattice deformation perpendicular to it, 47 and this can be modied depending on the amount of mist dislocations formed at the interface. In our ultrathin lms mist dislocations, previously observed in CSD Y123kLAO thick lms, 26,38 do not form in the Y248 layers identied by TEM at the interface (Fig. 4(c)). Therefore, the Y123 layers embedded in the lm can display a strong lattice expansion along c-axis, as it is experimentally demonstrated by X-ray diffraction (Fig. 3(a)).
It is not straightforward to understand why CSD Y123 lms have the high concentration of Y248 intergrowths identied by STEM. Actually, the generation of these intergrowths within the Y123 matrix has been described previously as a mechanism to accommodate lattice deformations at interfaces. 10,58 Therefore, our results suggest that the microstructural landscape of CSD Y123 ultrathin lms is also driven by a release of the elastic energy associated to the lattice mismatch at the LAO interface.
c Concentration of intergrowths versus lm thickness. A point of central interest is to investigate the evolution of the concentration of the intergrowths with lm thickness. The analysis of the XRD patterns of the Y123 thin lms did not show any Bragg peak associated to the Y248 phase, even if it was directly visible on the Z-contrast STEM images. As we have mentioned before, the Y248 intergrowths are characterized by a special structure having double Cu-O chains with a high concentration of defect clusters including two Cu vacancies decorated by three O vacancies which induce some additional disorder in the neighboring CuO 2 planes and preserve the overall Y123 (1 : 2 : 3) cation stoichiometry. 27,30 The disordered arrangement at the nanoscale of Y248 intergrowths perpendicularly to the c-axis will lead to a strong Xray diffuse scattering which will decrease the coherent contribution to the corresponding (00l) X-ray diffraction intensity peaks of Y248, at the limit of becoming invisible. 59, 60 We can then have a rough estimation of the concentration of the Y248 intergrowths via the calculation of the integrated intensity of a Y123 (00l) Bragg peak and comparing them with that expected for a lm of the same thickness having 100% of the volume with the Y123 phase. This estimation is based on two preconditions, one is that all the lms are free of secondary phases aer growth ( Fig. 2(a)-(e)) and the other one is that we can use the CTA 250 nm Y123 lms as a reference for a lm having 100% of its volume as the Y123 structure ( Fig. 4(b)).
In Fig. 5(a) we present the integrated area I 1 of the experimentally determined Y123 (005) Bragg peaks (see Fig. 2(b)), as a function of lm thickness. We include in the same gure the intensity expected for non-disordered Y123 lms I 2 , taking as a reference the intensity of a CTA 250 nm Y123 lm where Y248 intergrowths are practically absent (linear decrease of intensity with thickness). It is clearly seen that the experimental intensities of the FH lms (dotted line, I 1 ) are well below the expected values for a 100% volume of Y123 lms (solid line, I 2 ). Then, taking into account that I 1 is proportional to the volume of Y123 phase, we can estimate that the volume percentage of the Y248 phase r Y248 for the FH lms at different lm thicknesses corresponds roughly to: where for each thickness V T is the total volume of the lm, V 123 is the volume occupied by the Y123 phase and V 248 is the volume of the Y248 phase. The estimated evolution of r Y248 with lm thickness is displayed in Fig. 5(b). Note that the values of r Y248 increase continuously with the decrease of lm thickness. Especially, lms with thickness #25 nm display very high r Y248 Fig. 5 (a) Thickness dependence of the experimental integrated area I 1 of (005) Bragg peaks in the XRD patterns of FH Y123 films shown in Fig. 3(a). This intensity is compared to the expected intensity in films with 100% volume of non-disordered Y123 phase I 2 . The reference for I 2 corresponds to a Y123 CTA film with a thickness of 250 nm where no Y248 intergrowths are observed by STEM; (b) volume percentage of Y248 phase (r Y248 ) versus film thickness estimated from the observed differences I 2 -I 1 in (a). Dashed lines are guides to eyes.
In conclusion, the pristine Y123 CSD ultrathin lms grown by FH present a small percentage of the Y123 lattice structure, instead, a high density of the defective Y248 structure has been formed, making our lms very rich in Cu and O vacancies in the double chains of the Y248 structure which extends now over the whole lm thickness. This unique structural landscape enables us to investigate the superconducting behavior of the most common defect in CSD lms, i.e. Y248 intergrowths.

Superconducting properties
Superconducting properties of the Y123 lms have been investigated using isothermal and temperature dependent magnetization measurements. First of all, we should stress that, in agreement with the microstructural degradation mentioned before for Y123 ultrathin lms grown on substrates with tensile mist (i.e. STO), the low eld ZFC ($0.2 mT) magnetic susceptibility measurements showed smaller T c values (see ESI †). Therefore, we will concentrate our attention on the superconducting properties of lms grown on LAO substrates having a compressive mist.
In Fig. 6(a), we present the ZFC magnetic susceptibility c(T) measurements of the Y123 lms, measured at 0.2 mT, with thicknesses ranging from 10 nm to 250 nm. In the thin lm approximation, the initial susceptibility of thin lms with a disk shape can be written as c 0 ¼ (8R/3pt), where R and t are the radius and thickness of the disk. 42,43 Therefore, the normalized susceptibility c(T)/c 0 is a measure of the shielding capacity of Fig. 6 (a) Temperature dependence of magnetic susceptibility c/c 0 measured at low magnetic field (0.2 mT) of Y123 films of different thicknesses and grown by flash heating or conventional thermal annealing (inset); (b) correlation of c(0)/c 0 with the volume percentage of Y248 (r Y248 ) quoted in Fig. 4(b), i.e. the fraction of superconducting volume versus the estimated fraction of Y248 phase; (c) dependence of T c and DT c with film thickness. Inset: oxygenation times of the thin films; (d) temperature dependence of self-field critical current density J sf c (T) for films with different thickness (inset); (e) thickness dependence of J sf c (5 K) and J sf c (77 K) values. Dots account for the mean value of each while error bars account for the statistical distribution. (f) J c (H) dependence with magnetic field measured at 5 K for pristine Y123 films (star) with thickness of 25 nm (blue), 50 nm (red) and 250 (black). All the measurements were performed with Hkc. the thin lms, i.e. for a full superconducting lms we have c(0)/ c 0 ¼ À1, while a decrease of this value is a measure of a reduced superconducting volume. Fig. 6(a) clearly indicates that while lms with large thicknesses ($250 nm) display a full superconducting shielding behavior, a progressive decrease of the superconducting volume occurs when the lm thickness decreases. Fig. 6(b) shows that, actually, a close correlation exists between the decrease of the ratio c(0)/c 0 , i.e. the superconducting volume, and the volume percentage of the Y248 intergrowths, as estimated from X-ray diffraction, see r Y248 in Fig. 5(b). Our results show that extrapolation to full suppression of superconductivity occurs very close to the limit of 100% Y248, thus suggesting that this defective double layer structure has a non-superconducting behavior and it is intermixed with the remaining Y123 phase which is responsible of the observed superconducting behavior. The shielding currents have, therefore, a percolative behavior along the Y123 layered structure while the Y248 intergrowths would allow full ux penetration.
The evolution of the corresponding critical temperature (T c ) and transition width (DT c ) of the Y123 lms with the total lm thickness is shown in Fig. 6(c). It is noteworthy that T c gradually decreases down to 50 nm lm thickness, followed by a sudden drop at lower thicknesses. Similar behaviors have also been observed in vacuum deposited Y123 lms 47,56 or strained superlattices, 54,61 even if the observed T c decrease is more severe in the present case. We should note, as well, the obvious increase of DT c which is very likely inuenced by a decrease of the shielding efficiency of the percolating currents at smaller lm thickness. This is consistent with the increase of the concentration of nonsuperconducting Y248 intergrowths, although an enhanced structural disorder, as revealed by the decrease of the out-of-plane texture quality and the increase of nanostrain ( Fig. 2(b) and (d)), could also have some inuence on DT c .
The dependence with the lm thickness of self-eld critical current density when Hkc, J sf c , calculated using the Bean model as indicated in section 2, is illustrated in Fig. 6(d) and (e). Using the thin lm approximation of the Bean model to estimate J sf c (T) assumes, in the present case, that the macroscopic ux prole across the lms is established in spite of the nanoscale inhomogeneous superconducting character of the lms having a high concentration of Y248 volume where superconductivity is supressed. This mixed superconducting and nonsuperconducting microscopic structure was already previously tested in several sorts of superconductors, such as for instance superconducting foams, 62 where the validity of establishing a critical state prole with an effective critical current density was assessed. A progressive degradation of J sf c (T) with the decrease of lm thickness is also clearly identied here. In particular, the lm with a thickness of 10 nm shows a practical absence of superconducting behavior at all temperatures. Fig. 6(e) displays the evolution, as a function of lms thickness, of J sf c values, both at 5 K (J sf c (5 K)) and 77 K (J sf c (77 K)). We observe that J sf c keep constant values, i.e. J sf c (5 K) ¼ 30.0 AE 2.0 MA cm À2 and J sf c (77 K) ¼ 3.2 AE 0.2 MA cm À2 , for lms with thicknesses ranging from 50 nm to 250 nm. On the other hand, a strong tendency towards J sf c degradation is clearly observed when the lm thicknesses further decreases. We have conrmed the observed decrease of the critical currents in ultrathin lms by investigating the isothermal magnetic eld dependence J c (H) at 5 K (Fig. 6(f)). The sudden drop with thickness reduction of J sf c (T) and J c (H) values follows closely the observed decrease of T c while the superconducting volume determined through magnetic susceptibility measurements has a steady decrease ( Fig. 6(a) and (b)). Very likely the decrease of the inductively estimated J sf c values of the remaining superconducting Y123 layers arise from a combined effect of a reduced superconducting order parameter (reduced T c ) and from the geometrical effect of the non-superconducting volume in the lms (Y248 intergrowths) which reduces the cross section of the percolating currents and so leads to reduced effective J sf c values. Finally, we should stress that several authors have previously reported that superconductivity can be either enhanced or degraded at interfaces in strained high temperature superconducting lms due either to the in-plane strain with the substrate, 47,63 to strain induced oxygen deciency or to atomic disorder. 64,65 For the purpose of assessing the role of oxygenation time on the superconducting performance of the ultrathin lms, we have also analysed the inuence of an extension of the oxygenation time from 100 min to 360 min in the superconducting properties of these lms. Note that in our case further increase of the oxygenation time up to 360 min presents very little changes of T c or DT c , see Fig. 6(c). This suggests that superconductivity quenching in the ultrathin CSD Y123 lms is very likely not the result of oxygen deciency, in agreement with our analysis of the increase of c-axis parameter (Fig. 3(a)). 49,66 It's, however, well known that oxygen kinetics in oxides may be strongly inuenced by local strain and surface barriers and so we cannot fully disregard that some oxygen deciency remains. [67][68][69] In summary, we have provided evidence for the suppression of superconductivity at the nanoscale in the Y248 intergrowths, although the microscopic origin of this behavior remains an open issue. As it was recently reported, the Cu-O double chains of Y248 intergrowths include a high concentration of defective clusters consisting of two Cu vacancies decorated by three O vacancies. 27,30 These defects also were shown to lead to the formation of a nanoscale ferromagnetic (superparamagnetic) behavior, a highly distorted Y123 matrix around them, including apical oxygen vacancies, and a modied electronic structure in the neighboring CuO 2 planes, as detected by EELS and XMCD analysis. 27,29,30 Very likely quenching of the Cooper pair formation occurs at the nanoscale in the full volume of the Y248 intergrowths. 10, 70 We suggest now that the double chains of the defective Y248 structure, and the CuO 2 planes next to them, have an absence or a very perturbed superconducting behavior. The observed progressive reduction of the superconducting volume with lm thickness decrease is then associated to an enhanced volume percentage of the Y248 intergrowths. The decrease of T c , on the other hand, should reect the lattice deformation of the Y123 layers remaining in the lms.

Conclusions
We report an investigation of the inuence of the micro/ nanostructure changes of chemical solution deposited Y123 This journal is © The Royal Society of Chemistry 2020 Nanoscale Adv., 2020, 2, 3384-3393 | 3391

Paper
Nanoscale Advances lms varying thicknesses, down to 5 nm, and their consequences on the superconducting properties. Ultrathin Y123 epitaxial lms have been successfully grown on LaAlO 3 substrates based on optimized crystallization conditions. Detailed microstructural investigations of the Y123 ultrathin lms by means of XRD and STEM have demonstrated that these thin lms are epitaxial with an increased concentration of Y248 intergrowths modifying the lms nanostructure when the lm thickness decreases, even if the overall Y:2Ba:3Cu stoichiometry is preserved. The progressive increase of the volume percentage of Y248 intergrowths when the lm thickness decreases has been closely correlated with a corresponding shrinking of the superconducting volume as measured by low eld magnetic shielding, thus suggesting a suppression of the superconducting behavior at the nanoscale. Defective Y248 intergrowths include a high concentration of Cu and O vacancies and the present work has shown that this defective structure is strong enough to suppress its superconducting behavior thus making to behave fundamentally different from the stoichiometric Y248 phase displaying T c values in the range of $80 K. This conclusion gives support to the idea that defective Y248 intergrowths play a key role as articial pinning centers of vortices in Y123 nanocomposite lms and coated conductors.

Conflicts of interest
The authors declare no competing interest