Microstructure, pinning properties, and aging of CSD-grown SmBa₂Cu₃O_7−δ films with and without BaHfO₃ nanoparticles

The TFA-MOD precursor solutions were prepared following the recipe of [12, 36]: the acetates of Sm, Ba, and Cu (>99.99%, Alfa Aesar) of ratio 1:2:3 were dissolved in water, mixed with surplus of trifluoroacetic acid (99.5+%, Alfa Aesar), and stirred until complete dissolution of the metalorganic salts at room temperature. These solutions were concentrated with a rotary evaporator, re-diluted in ultra-dry methanol (>99.9%, H₂O < 50 ppm, Carl Roth) and filled up to the final concentration of the RE of 0.25 mol l⁻¹. For the nanocomposite solutions, hafnium(IV)-2,4-pentanedionate (Hf(acac)₄, 97+%, Alfa Aesar) and an according molar amount of barium acetate, dissolved in water, were added to the solutions for nominally 12 mol% of BHO in the films. For BZO as nanoparticles, it had been reported that the amount should be limited to 10 mol% in order to avoid nanoparticle coarsening [37, 38]. For BHO, our earlier (unpublished) experiments with several REBCO phases suggest a quite similar amount of 12 mol% as optimum because the coarsening is still not excessive but the pinning improvement considerable. Afterwards, minute amounts of acetylacetone (0.6 acac/RE ion or 1.5 vol% in final solution) were added to protect them from impurities such as water. The solutions were then filtered through polytetrafluoroethylene with 0.2 μm pore size and spin-coated with 6000 rpm for 30 s on cleaned 10 mm ×10 mm (001)-oriented single crystals of LaAlO₃ (LAO) resulting in a film thickness of around 220 nm. The films were heat-treated (for details see [36]) with a crystallization oxygen partial pressure p_O2 of 50 or 150 ppm in nitrogen at 1 atm total pressure (dew point T_dew = 19 °C) and crystallization temperatures between 780 °C and 860 °C. Afterwards, the sample cooled down in dry nitrogen with the same p_O2 and gas flux as used during the film growth to 450 °C for the final oxygenation in 1 bar oxygen.

Structural features of the films were measured by x-ray diffraction (XRD; D8 Discover, Bruker, Cu-Kα radiation) and scanning electron microscopy (SH-5000P, Hirox, tungsten cathode, SE-detector, 10 kV acceleration voltage). Layer thickness and surface topography were determined by atomic force microscopy (Dimension Edge, Bruker) in tapping mode on bridges used for electrical transport measurements, and occasionally cross-checked by transmission electron microscope (TEM).

Cross-section samples for scanning TEM (STEM) were prepared by an in-situ lift-out technique [39] in a FEI Strata 400S focused-ion beam (FIB)/scanning electron microscope system. High- and low-angle annular dark-field (HAADF/LAADF)-STEM images were taken with a FEI Titan³ 80–300 TEM operated at 300 kV. The HAADF-STEM image intensity is sensitive to the average atomic number Z, where a larger Z results in higher HAADF-STEM intensity for a sample with reasonably constant TEM-sample thickness. In contrast, LAADF-STEM images also show diffraction contrast and reveal more clearly crystalline defects and lattice strain. Chemical analysis with energy-dispersive x-ray spectroscopy (EDXS) was carried out with a FEI Tecnai Osiris microscope operated at 200 kV with ChemiSTEM technology [40]. The EDXS mapping data set was denoised with principal component analysis [41] before extracting the elemental signals with a peak-fitting routine using the HyperSpy Python-package [42, 43].

The films were scanned for their superconducting properties by inductive techniques: the transition temperature T_c (defined as 10%, 50% and 90% of the normal-state value) by a self-made mutual-inductance device, and the self-field critical current density at 77 K in a Cryoscan (THEVA, 50 μV criterion). The ∼50 μm wide and 1 mm long bridge structures for 4-point transport measurements on selected samples were photolithographically prepared with an image-reversal photoresist (AZ5214E, Microchemicals) plus wet-chemical etching in an aqueous 0.6 wt% HNO₃ solution. For lowering the contact resistances, Au pads were prepared by PLD. The electrical contacts were provided by Cu wires attached with silver paint.

Field, temperature, and orientation dependence of J_c in magnetic fields up to 14 T were determined via the measurement of voltage–current (V–I) characteristics in a Physical Properties Measurement System (Quantum Design) with rotator stage and a self-made LabVIEW measurement software controlling a Nanovoltmeter 2182 A and a SourceMeter 2460 by Keithley. J_c was determined with an electrical field criterion E_c of 1 µV cm⁻¹. The N value (V ∼ I^N ) was determined by fits of log(V) vs log(I) in 1–2 decades above E_c.

The upper critical field B_c2 and the irreversibility field B_irr were determined via temperature dependent resistivity measurements, ρ(T), at constant applied fields B||c and a bias current of 100 µA. The criteria were 75% of normal state resistivity above T_c and ρ_irr = E_c/100 A cm⁻², respectively. T_c and T_irr, i.e. zero-resistivity T_c, were determined accordingly. For evaluating B_c2 additionally, extrapolation to zero of the logarithmic derivative (dlnR/dT)⁻¹ was used in order to be more criterion-independent, figure S1. The activation energy U₀ of thermally activated flux motion was determined by linear fits of the Arrhenius plots ln(R) vs. 1/T in the region of highest linearity.

The two top-performing samples of the pristine and the nanocomposite series were analysed further in d.c. fields up to 24 T in the 25 T cryogen-free superconducting magnet at IMR Tohoku University. For measuring the angular dependence of J_c, the magnetic field B was applied in maximum Lorentz force configuration at different angles θ measured from the c-axis.

From experience, we regarded the crystallization temperature T_crys and the oxygen partial pressure p_O2 during crystallization as the most significant parameters. Consequently, while keeping the water vapour, i.e. the dew point T_dew constant at 19 °C, we varied T_crys at two oxygen partial pressures, which were promising from former studies [12] taking into account that the optimum partial pressure decreases with increasing RE ion size [17, 62]. The lower oxygen partial pressure of 50 ppm yielded higher critical temperatures and self-field critical current densities compared to 150 ppm, figure 6. The optimum crystallization temperature, especially with respect to J_c, lies at 820 ± 10 °C for nanocomposites at both partial pressures as well as the pristine films at 150 ppm, whereas 50 ppm O₂ allows even lower T_crys < 800 °C for the pristine films. Crystallization temperatures below 780 °C were not reasonable anymore because of increasing porosity and formation of misorientations. Also for lower oxygen partial pressures, the system seems to be not stable anymore. For extreme-low-fluorine CSD-grown SmBCO films for example, 20 ppm yielded lower J_c values than 50 ppm [13].

Zoom In Zoom Out Reset image size

In this optimized parameter range, the resistive transition width, the zero-resistivity T_c value, and the temperature dependence of the irreversibility field B_irr are very similar, except for the highest T_crys values at 50 ppm and the nanocomposites at 150 ppm, which show lower irreversibility lines B_irr(T), figure 7.

Zoom In Zoom Out Reset image size

The pinning force densities F_p(B), measured 1.5 years after deposition, show some variability for the optimized parameter range, see figure 8 for the 65 K data as an example. Where the pristine samples have a maximum pinning force density F_pmax of 10 ± 2 GN m⁻³, the nanocomposites show slightly better values of 14 ± 3 GN m⁻³. This trend is also visible for other measurement temperatures. This increase due to the BHO nanoparticles of ∼30% (best samples) compares well with earlier results on GdBCO + BHO [30], however are well below values for recent GdBCO, as well as GdYBCO, and YBCO systems, see table 1. So, the achievable increase in F_pmax due to randomly oriented nanoparticles seems to depend in part also on RE ion size (most likely via the respective microstructures) besides nanoparticle size [34] and content. However, one has to take into account the much larger advance and data availability for the other systems.

Zoom In Zoom Out Reset image size

Table 1. Comparison of achievable increase in maximum pinning force density ${F_{{\text{pmax}}}}$ by randomly oriented BaMO₃ (M: Hf or Zr) nanoparticles in MOD-grown REBCO films at 65 or 77 K.

RE	Additive	Content (mol%)	$F_{{\text{pmax}}}^{{\text{BMO}}}/F_{{\text{pmax}}}^{{\text{prist}}}$	T (K)	References
Sm	BHO	12	1.5	77	This work
Sm	BHO	12	1.3	65	This work
Gd	BHO	12	1.4	77	[30]
Gd	BHO	12	3.8	77	[33]
Gd	BHO	12	4.5	65	[33]
Gd0.5Y0.5	BZO	25a	2.5	65	[34]
Gd0.5Y0.5	BHO	25a	6.5	65	[34]
Y	BHO	16b	1.5	77	[63]
Y	BHO	12	3.3	77	[30]
Y	BZO	10	3.0	77	[64]
Y	BZO	10	7.0	77	[37]
Y	BZO	10	9.8	77	[65]
Y	BZO	10	10.4	65	[38]

^aCalculated from 12 vol% in [34]. ^bCalculated from 1.5 at% in [63].

The resistive transitions of pristine and BHO-added SmBCO films are remarkably similar, figure 9. Both kinds of samples, if optimized, show a very small zero-field transition width of ∼1 K and comparable normal state resistances of 120 Ω at 100 K (i.e. ∼130 µΩ cm). Via the Arrhenius plots of these data, the activation energy U₀ = −dlnR/d(kT)⁻¹ of thermally activated flux motion was determined, figure 10. The BHO–SmBCO nanocomposite shows slightly larger values especially for low fields (1 T values: 4.6 10⁴ K vs. 5.4 10⁴ K), and both samples show roughly a square root dependence in low fields (characteristic for plastic pinning) and a linear dependence at higher fields (characteristic for collective pinning). A linearity analysis between the residual resistance near T_c and U₀ yields an estimate for T_c, which is 93.9 K for both samples within error bars. This value is consistent with two other criteria: maximum slope (93.5 K) and extrapolation of the logarithmic derivative (dlnR/dT)⁻¹ to zero (94.0 K), which is related to the excess conductivity analysis (not followed here), see also supplement. This corresponds to the aforementioned 75% onset-T_c resistivity criterion for B_c2.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

At higher fields, both samples show a characteristic kink in their transitions (arrow in figure 9). Since this kink is not observed at low fields, inhomogeneities and grain boundary effects can be excluded as explanations. It rather characterizes the beginning of a transition region between the thermally activated flux motion and the fluctuation region near T_c with strong excess conductivity. The transition region itself shows a rather linear behaviour, characteristic for flux flow. This kink behaviour has been observed in different REBCO films, e.g. [66–69], for a theoretical description see e.g. [69, 70].

Figure 11 shows the B–T phase diagram of the BHO–SmBCO nanocomposite for B||c (the one for pristine SmBCO is very similar). B_c2(T) was linearly fitted with slope dB_c2/dT|_{T c} = 1.50 ± 0.02 and T_c = 93.9 ± 0.1 K. The werthamer-helfand-hohenberg (WHH) estimate [71] of the zero-temperature orbital upper critical field B_c2(0) = 0.69dB_c2/dT|_{T c} T_c yields 97 ± 1.5 T and consequently a coherence length ξ_ab(0) = 1.84 ± 0.01 nm. The irreversibility field could well be fitted with a power law B_irr(T) = B_irr(0) (1 T/T₀)^q , with B_irr(0) = 76 ± 1.6 T, T₀ = 93.1 ± 0.2 K, and an exponent q = 1.32 ± 0.02. The 95% confidence intervals for both fits are roughly the line thickness in figure 11. Within the error bars, q is 4/3, which is characteristic for a glass–liquid transition. This transition, however, could not be determined unambiguously from the present data set. B_irr(0) being much smaller than B_c2(0) can have two origins: either B_c2 is not orbitally limited at low T and/or B_irr(T) shows an upturn at low T due to a possible extra pinning component. High-field measurements may clarify this.

Zoom In Zoom Out Reset image size

The addition of BHO nanoparticles to the SmBCO films enhances the pinning force densities and hence the critical current densities mainly at low fields with a crossover at low temperatures around 10 T, figure 12. The self-field critical current densities of both films are very similar; only at 77 K does the BHO sample show slightly lower values. Whereas there is a clear power law region J_c(B) ∼ B^−α visible for the pristine sample, it is not as clear for the BHO nanocomposite, especially at elevated temperatures. Nevertheless, when analysed in the same field range of highest linearity for a certain temperature, it is obvious that α increases with increasing temperature for both films, and α_pristine < α_BHO for all temperatures, table 2. The latter is explained by an increase of the accommodation field (i.e. the length of the constant J_c plateau at lowest fields) and a decrease of the irreversibility field (especially at low T) by BHO addition.

Zoom In Zoom Out Reset image size

The medium-field range of increased J_c and F_p by BHO addition is also clearly distinguishable by an increase in N value, though in somewhat smaller ranges at high temperatures, figure 12. The N value, i.e. the exponent of power law E(J) curves E ∼ J^N , is in first approximation related to the flux creep rate $S\, \cong \,\frac{1}{{N - 1}}$. In lowest fields however, the BHO nanocomposites show somewhat lower N values for T ⩾ 50 K, which may be due to an increased inhomogeneity, i.e. in local J_c variance. For medium fields and temperatures (T ⩽ 65 K), the N value seems not to depend on temperature for both samples. For the lower temperatures, and especially for the nanocomposites, a region of nearly field-independent N value (i.e. creep rate) is observed before the N value tends towards 1 for fields going to B_irr.

The pinning force density curves F_p(B) clearly show the above-mentioned crossover at around 10 T (at low T). This indicates a reduction in the irreversibility field by BHO addition and an already high pinning efficiency of the pristine films' microstructure. The pinning force curves of the pristine samples can be approximated by a Dew-Hughes function F_p(B) ∼ B^p (1 − B/B_irr)^q with (p,q) ∼ (0.5,2), which suggests surface pinning or random pinning at sufficiently small defects. The parameter q is actually often slightly larger than 2 (close to 3) due to the stronger flux creep effects near B_irr. The F_p curves of the BHO nanocomposites, more than for the pristine films, however cannot be described well by a single Dew-Hughes function with physically reasonable combinations of B_irr and q. This is due to an extra low-field component of the extended BHO nanoparticles at around 1 T. A similar behaviour has been observed recently on all-CSD-grown YBCO nanocomposite coated conductors with BZO and BHO nanoparticles [72].

The angular dependence of the critical current density and of the corresponding N value, figure 13, with respect to field orientation in fields up to 14 T of the pristine and the BHO nanocomposite film are very similar. A strong ab-peak in J_c(θ) due to electronic anisotropy (random pinning and intrinsic pinning at low temperatures) as well as correlated pinning at planar defects (SFs) is clearly visible. Close to the irreversibility line, i.e. at sufficiently high fields and temperatures, a small c-axis peak is appearing; most likely due to extended planar (anti-phase boundaries, twin boundaries, domain boundaries) or linear defects (dislocation networks) parallel to the c-axis. At medium fields and low temperatures—more for the pristine sample—J_c(θ) shows a dip near B||c. This also can be attributed to planar defects, namely vortex channelling effects at twin boundaries [73]. The BHO particles hence reduce this channelling effect either directly by increased pinning or indirectly by leading to a reduced density of extended twin boundaries. By BHO nanoparticles being effective in a wide angular range around B||c, cf figure 12, the J_c anisotropy is decreased by BHO incorporation at all temperatures. Figure 14 compares the J_c anisotropy at 24 T for several temperatures between the pristine and the BHO-nanocomposite film. Above 30 K, both films behave surprisingly similar.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Further information about the present pinning species and mechanisms are gained by the study of the N value anisotropy. First of all, it usually enhances features due to its averaging effect, see e.g. the extra pinning component of additional medium angle peaks in N(θ) (e.g. peak around 135° at 65 K, 3 T), which are not or barely visible in J_c(θ) and due to the sparsely distributed nanoparticles larger than ξ. In these regions, N scales well with J_c, since both depend on the pinning potential U₀, and the pinning behaviour is dominated by the elastic properties of the flux lines. In contrast, at sufficiently low temperatures, N(θ) shows a strong dip whereas J_c(θ) has a maximum. Here, N and J_c are anti-correlated and the creep rate is not determined by the pinning potential anymore but influenced by a second creep mechanism. In this region, the flux lines build staircase patterns of segments strongly pinned (trapped) within the ab-planes/ab-planar defects and weakly pinned segments with oblique angle to the ab-planes. The easy movement of the weakly pinned segments leads to a net hopping behaviour of the trapped segments from one planar pinning centre to the next, increasing the creep rate and hence decreasing the N value. At the lowest temperatures (see 30 K) and even smaller angles between B and the ab-planes, the whole flux lines will be trapped in the planar pinning centres, which hinders the flux lines from the net hopping movement, thus decreasing the creep rate, and hence increasing the N value. As is apparent in figure 13, especially at low T, the addition of BHO nanoparticles increases the widths of these regions. This is due to the shortening of the trapped segments due to the direct interaction with the nanoparticles and/or a shortening of the planar defects. Interestingly, the trapping angle seems rather field-independent for the pristine sample at medium temperatures, while a B^−3/4 dependence would be expected by theory. This had been observed also for NdFeAs(O,F) films and was explained by the vicinity of the 2D–3D transition [74]. Finally, we would like to point the attention to the slight c-axis peak in N(θ) at 14 and 24 T, 30 K for the pristine sample, indicating a decreased creep rate at minimum J_c. Most likely, the extra pinning contribution is hidden by the strong intermediate peaks in J_c(θ) and gets only apparent in the creep rate.