Upward shift of the vortex solid phase in high-temperature-superconducting wires through high density nanoparticle addition

We show a simple and effective way to improve the vortex irreversibility line up to very high magnetic fields (60T) by increasing the density of second phase BaZrO3 nanoparticles. (Y0.77,Gd0.23)Ba2Cu3Oy films were grown on metal substrates with different concentration of BaZrO3 nanoparticles by the metal organic deposition method. We find that upon increase of the BaZrO3 concentration, the nanoparticle size remains constant but the twin-boundary density increases. Up to the highest nanoparticle concentration (n ~ 1.3 × 1022/m3), the irreversibility field (Hirr) continues to increase with no sign of saturation up to 60 T, although the vortices vastly outnumber pinning centers. We find extremely high Hirr, namely Hirr = 30 T (H||45°) and 24 T (H||c) at 65 K and 58 T (H||45°) and 45 T (H||c) at 50K. The difference in pinning landscape shifts the vortex solid-liquid transition upwards, increasing the vortex region useful for power applications, while keeping the upper critical field, critical temperature and electronic mass anisotropy unchanged.

simple argument that once the pinning centers have been outnumbered by vortices, their effect on increasing H irr decreases or may even be washed out. However, this idea overlooks the fact that vortices are interacting elastic objects, which form a lattice with different degrees of order depending on the underlying pinning landscape. The understanding of the physical system adds a layer of complexity when pinning centers of different dimensions (point defects, lines, planes or nanoparticles) are taken into account.
In this paper, we show that the combination of planar twin boundaries (TBs) and a higher density of three dimensional (3D) nanoparticles (NPs) increases H irr at up to 60 T. In the case of TBs there is a clear signature of the matching field at very low fields (0.5 T) but nevertheless, their presence continues to increase H irr up to 60 T. The effect of the NPs is also seen up to 60 T with the additional benefit of their being effective in the entire angular range. We find no signs of saturation in the improvement of H irr up to 3 wt% BaZrO 3 (BZO) added to REBCO wire up to the highest field measured (60 T), and obtain record high values.

Results
The results shown are distributed as follows. We start by explaining the growth methods and show superconducting and microstructural studies. We find that the use of (Y,Gd)BCO is beneficial (as compared to YBCO) and that the addition of BZO NPs does not degrade the general properties, and increases twin boundary density. Then we focus on transport measurements in DC and pulsed fields as a function of temperature and field orientation. For the upper critical field, (H c2 ), the addition of NPs produces no significant changes. For H irr , the scenario is complex and depends on the field magnitude and orientation. For H||c, at low fields, twin boundaries dominate the irreversibility line with a matching field observed and a peak as function of field orientation. As the field increases the effects of TBs are less prominent as nanoparticles compete in the vortex localization. At high fields, NPs and TBs cooperate to enhance the H irr . At intermediate angles (e.g. H||45°) nanoparticles produce the greatest increase in H irr , showing no saturation up to the highest fields measured. Furthermore, at high fields we obtain a constant improvement of 25% with respect to YBCO films. For H|| ab, the changes due to NPs addition are small as compared to other orientations.
Uniformly dispersed BaZrO 3 nanoparticles in (Y 0.77 Gd 0.23 )Ba 2 Cu 3 O y wires. The samples used in this study were 0.5 μ thick films of standard YBCO, (Y 0.77 Gd 0.23 )Ba 2 Cu 3 O y ((Y,Gd)BCO) and BZO-doped (Y,Gd) BCO ((Y,Gd)BCO+ BZO) derived from the metal organic deposition (MOD) process, grown on metal substrates. The content of BZO in the films was 1-3 wt%. The film microstructure was studied by several techniques (see Methods). The planar view transmission electron microscopy (TEM) images of standard YBCO, (Y,Gd)BCO with 1wt% BZO (+ 1BZO) and 3wt% BZO (+ 3BZO) wires are shown in Fig. 1a-c, respectively. For clarity, the corresponding bottom panels in those figures show schematics of the main defects. All films have a low density (∼ 0.1 × 10 21 /m 3 ) of large RE 2 Cu 2 O 5 (225) precipitates (shown as open circles in the bottom panels) formed during REBCO crystallization from the precursors containing BaF 2 , RE 2 Cu 2 O 5 and CuO. In both + 1BZO and + 3BZO wires, BZO NPs (modal size ∼ 23 nm) are randomly distributed and uniformly dispersed. The average density of these nanoparticles was determined to be n = 2.9 × 10 21 /m 3 for + 1BZO, n = 7.1 × 10 21 /m 3 for + 2BZO (not shown), and n = 13 × 10 21 /m 3 for + 3BZO wires, respectively. Note that the greater than linear dependence of n on BZO% derives from the slight decrease in size of the NPs with increasing BZO density. In addition to the uniformly randomly distributed 3D defects (BZO NPs), all samples, including the standard YBCO, contain high densities of two types of planar (2D) defects, namely c-axis correlated TBs (Fig. 1a-d) and stacking faults (SFs) parallel to the ab plane (Fig. 1e). In some previous studies of MOD REBCO with nanoparticles, it was observed that the SFs cut into the TBs, resulting in discontinuous short length TBs 14 . In contrast, although our + 3BZO wire has a high density of SFs as shown in Fig. 1e, the TBs remain connected from the bottom to the surface (Fig. 1d), indicating that the TBs in the BZO NP-doped wire are c-axis correlated pinning centers. Table 1 shows a summary of the crystallographic and superconducting properties for the YBCO, (Y,Gd)BCO, + 1BZO, + 2BZO and + 3BZO films. No decrease in T c , or increase of δω or δφ is observed for the films with BZO inclusions. The only difference found is positive, being that films with Y,Gd have higher T c and self-field J c (J c s.f. ) than the YBCO film, consistent with good crystallinity (unchanged δω, δφ). The J c s.f. values of the + 3BZO film at 65 and 77 K are 12.5 and 4.0 MA cm −2 , respectively. These J c s.f. values are almost as high as those of nanocomposite MOD-YBCO thin films on a single crystal substrate 6 . The fact that we maintained (or even improved) J c s.f. and T c after adding defects is significant given that in some cases, for films grown by pulsed laser deposition (PLD), T c and J c s.f decreased with increasing BZO content due to poor crystallinity, local strain and oxygen deficiency near second phases 16 . The difference in T c and J c s.f behavior with BZO% between MOD and PLD processes comes from differences in the formation mechanisms of the BZO/REBCO matrix 17 .

Irreversibility line at low and intermediate DC magnetic fields.
To investigate the influence of the density of BZO NPs on the superconducting properties of the glassy phase, we measured the irreversibility temperature T irr (H,θ). For the YBCO and (Y,Gd)BCO wires, at low magnetic fields we observe two maxima in T irr , centered at H||c (θ = 0°) and H||ab (θ = 90°). The peak at H||c comes from c-axis correlated defects. The maximum at H||ab originates from the electronic-mass anisotropy and the SFs. At intermediate fields (see Fig. 2b), the c-axis peak of T irr in both YBCO and (Y,Gd)BCO becomes very weak, but is evident again at higher fields (see Fig. 2c). We note that the T irr (θ )/T c for the YBCO and (Y,Gd)BCO wires are almost identical as shown in inset of Fig. 2, which validates the use of YBCO for the comparison that we are making.
All wires with BZO NPs show an enhanced T irr (θ) as compared to that of the YBCO and (Y,Gd)BCO wires, and the magnitude of the T irr (θ) enhancement increases with increasing BZO NP content at all magnetic fields (0.5, 4 and 15 T) and at all angles. The T irr anisotropy, T irr (||ab)/T irr (||c), is also smaller for the + BZO samples and Scientific RepoRts | 6:20436 | DOI: 10.1038/srep20436 decreases monotonically with BZO NP content. At low fields, the + 3BZO wire exhibits nearly isotropic properties with a small (in height) c-axis peak. At 15 T, T irr (θ) shows a small c-axis peak indicating a contribution from correlated pinning. This is consistent with J c (θ) data 10 at high fields where a c-axis peak is also found. In both cases (J c and T irr ) the c-axis peak height is less pronounced in samples with BZO additions, partially because of the general increase in J c and T irr produced by the nanoparticles.
The results in Fig. 2 can be interpreted as follows. At low fields, below the matching field of the TBs (μ 0 H φ,TB ) given by μ 0 H φ,TB = (φ 0 /d TB 2 ), where φ 0 and d TB are the flux quantum and TB spacing respectively, most vortices get localized by the TBs (Fig. 3a). At intermediate H, above μ 0 H φ,TB , the density of vortices is high enough that the TBs become "saturated", the extra vortices sit outside them and become localized by the randomly distributed NPs which are strong pinning centers (Figs 2b and 3b). At even higher fields, once all the NPs are saturated, no elastic energy is lost by achieving pinning at the TBs, allowing for the localization effect of the TBs to be observed through the caging effect (Fig. 3c), similar to what happens for the standard YBCO sample (Fig. 2c).
To better determine the effects of the c-axis correlated defects, we measured μ 0 H irr (T) at low fields for different θ (see Fig. 4). The c-axis irreversibility lines of the wires with and without BZO NPs exhibit a pronounced kink indicating a crossover field (see H cr arrows). For fields smaller than μ 0 H cr , H irr shows a rapid increase with  Fig. 1 (a,b,c,d) and Fig. 1 Table 1. Structural and superconducting properties. Sample data for the reference YBCO wire and for the (Y,Gd)BCO + BZO wires. δω and δφ denote the full-width at half-maximum (FWHM) values of the out-ofplane rocking curves (ω scans) of the 005 diffraction peaks and in-plane rocking curves (φ scans) of the 103 diffraction peaks, respectively. The particle-size distribution was extracted from high-magnification planar-and cross-section views of several TEM images. The size of the BZO NPs ranged from 17 to 28 nm with a modal size of ∼ 23 nm. The average TB spacing was estimated from several planar view TEM images. T c was determined using a criterion of 0.01ρ n . * 225 precipitates. decreasing T; above μ 0 H cr the irreversibility line is closer to a linear temperature dependence. For the + 3BZO wire, μ 0 H cr is 0.55 T, which is 1.6 times higher than that of YBCO (μ 0 H cr = 0.34 T). From TEM images, we observe that this correlates with the higher TB density found in + 3BZO as compared to YBCO wires. Indeed, we obtain μ 0 H φ,TB = 2.3T and 1.0T, for + 3BZO and YBCO, respectively. This cross-over is remarkably similar to that observed in YBCO bulk with nanoscale TBs 18,19 and columnar defects 20 . For TBs and columnar defects, the cross over field is usually not observed precisely at the matching field but around 1/3-1/2 of H φ . This also correlates very well with our observation of the crossover fields with H cr /H φ,TB ∼ 1/3 for samples with and without BZO NPs,  namely 0.34 and 0.24 for YBCO and + 3BZO, respectively. To further corroborate that the observed crossover field arises from correlated defects and is not an artifact, we confirmed that no kink is found in H irr -T at H||45° (see Fig. 4). Therefore, we can conclude that the array of TBs is the main source of correlated pinning at H||c in our MOD REBCO wires. Note that the TBs continue to contribute at μ 0 H ≫ μ 0 H cr as seen in the peak of T irr (θ) around H||c in Fig. 2c and the inset of Fig. 5d.

Irreversibility lines and upper critical fields in pulsed fields.
To obtain the temperature and angular dependence of μ 0 H c2 and μ 0 H irr for the YBCO and (Y,Gd)BCO + BZO wires at even higher fields, we measured ρ(T) and ρ(H) at various θ values by using pulsed fields up to 60 T (see insert Fig. 5b). See Methods for further details. Figure 5a shows μ 0 H c2 (T) at H||c, H||45°, and H||ab. The results demonstrate that μ 0 H c2 (T) is remarkably similar for all samples in spite of their very different pinning landscapes. The inset of Fig. 5a shows the electronic-mass anisotropy (γ) vs. T obtained from the H c2 measurements for the YBCO and (Y,Gd)BCO+ BZO wires (see ref. 9 for details). The average values for γ are 5.1, 4.6 and 4.7 for the YBCO, + 1BZO and + 3BZO wires, respectively. Although γ is slightly smaller for the wires with BZO, all values are very close to 5. Further confirmation of the obtained value of γ, comes from the angular-dependent H c2 shown in Fig. 5b. The angular dependence of H c2 can be fit very well using Hε(θ) = H(cos 2 (θ)+ γ -2 sin 2 (θ)) (see ref. 21) with γ = 4.7, in agreement with the experimental result shown in the inset of Fig. 5a. This is also in accordance with our previous work, where we reported that the μ 0 H c2 (θ) dependence follows a single curve consistent with γ ~ 5 even with different additions and growth methods 22 . From the μ 0 H c2 data, we conclude that μ 0 H c2 is not greatly affected by the pinning landscape. It is worth noting that the value of γ ~ 5 is indeed smaller than the values found for YBCO single crystals 23,24 , as previously noted 25 . The lack of sensitivity in H c2 upon increase of disorder can be found in the small coherence length of REBCO (also the reason for high H c2 ). The small coherence length places REBCO in the clean limit, thus making its superconducting properties less sensitive to disorder.
On the other hand, the changes in pinning landscape do influence H irr , as a clear enhancement is found for all + BZO wires compared with YBCO wire (see Fig. 5c,d). Figure 5c displays the μ 0 H irr -T at various angles for the YBCO and (Y,Gd)BCO+ BZO wires. For H||c and H||45°, wires with BZO NPs show a higher μ 0 H irr (T) than that of standard YBCO. Furthermore, we observe that μ 0 H irr (T) continues to increases with higher densities of BZO NPs. For H||ab, + BZO samples show a much smaller increase in μ 0 H irr (T).
The changes can be better understood by analyzing the H irr (θ) as shown in Fig. 5d. Consistent with Fig. 5c, the addition of BZO NPs produces a small increase near H||ab and a much bigger enhancement for H||c and intermediate orientations. The resulting angular dependence of H irr for the + 3BZO wire cannot be described using the electronic-mass anisotropy scaling model, unlike H c2 (θ) or YBCO's H irr (θ), which do follow ε(θ) for most of its angular dependence (see Fig. 5b,d). More information on the effect of the nanoparticles on H irr can be observed in the inset to Fig. 5d. In this inset we plot the difference of μ 0 H irr (θ)-ε(θ)μ 0 H irr for the YBCO and + 3BZO sample. For YBCO we observe that, except for the peaks around H||c and H||ab due to correlated disorder (TBs and SFs, respectively), over most of the angular range H irr (θ) can be very well described by the electronic-mass anisotropy, γ = 5 indicating that most of the angular dependence of H irr can be explained by pinning arising from weak random point defects such as oxygen vacancies (with diameter in the angstrom scale ≪ 2ξ (T)). For + 3BZO we find very similar behavior near the ab-plane, where it shows a peak almost identical to that of the YBCO wire indicating similar ab-plane correlated pinning like SFs (consistent with H irr (T) in Fig. 5c). However, a much broader maximum is found for |θ| < 80°. This is the consequence of the higher density of TBs coupled with the presence of strong random pinning from BZO NPs as shown in the TEM of Fig. 1b,c, and the schematics of Fig. 3. The NPs are responsible for the increases in the wider angular range and the TBs are more effective near the c axis. The broad dome-shape like increase in μ 0 H irr (θ)-ε(θ)μ 0 H irr due to the NPs resembles the increases found in J c in samples with nanosized inclusions, both in YBCO and Fe-based superconductors 10,[26][27][28][29] . The reason for this dome shape (or inverse anisotropy) is found in the fact that while NPs are randomly dispersed, they are not point defects, thus do not follow the electronic mass anisotropy scaling as point defects do 21 . Recently Mishev et al. proposed a model to explain this effect based on the relative size of the vortex core compared to the inclusion size 29 .
The leveling off observed in H irr (θ) due to nanoparticles translates into an smaller ratio of H irr (||ab)/H irr (||c), with values of 2.9 for + 3BZO as compared to 5.0 for YBCO. The decrease in the H irr ratio (also found in J c ) has been analyzed using an 'effective anisotropy" (γ eff ) both for of J c (θ) and H irr (θ) 6,7,30,31 . We attempted to scale the H irr (θ) data of our BZO NPs doped wires using γ eff . Using intermediate angles we obtain γ eff ~ 2, but the scaling is not successful, as the data do not collapse onto a single curve as shown in the literature 6,7,30,31 . It is worth mentioning that although in principle the samples in references 6,7,14 are also MOD with NPs, there are some differences in the microstructure with the wires in this study. Specifically, with the addition of BZO our samples have a higher density of twin boundaries, but the opposite is found in ref. 14.
As shown in Fig. 5c, overall the + 3BZO wire shows an extremely high μ 0 H irr at 65 K, at 30 T (H||45°) and 24 T (H||c) and at 50 K, 58 T (H||45°) and 45 T (H||c), values which to our knowledge are the highest reported for practical HTS wires 31 as well as for thin films on single-crystal substrates 32 . Recently, very high μ 0 H irr = 14.8 and 15.8 T at 77 K (H||c) were observed in a heavily BZO nanorod-doped (Y,Gd)BCO wire grown by metal organic chemical vapor deposition 33,34 and BaHfO 3 doped PLD GdBCO wire 35 ; these methods could also translate to great improvements of H irr at high fields. This indicates that a high density of NPs/columns is beneficial not only for MOD prepared materials but is a general trend for REBCO wires.

Discussion
In order to pinpoint H-T regions where the BZO additions are more effective, in Fig. 6 we plot the H irr enhancement Δ μ 0 H irr = μ 0 H irr,BZO − μ 0 H irr,Y for different concentrations of BZO NPs. This also set the bases to determine possible H-T saturation regions and if so, the dependence with magnetic field orientation. The improvement, Δ μ 0 H irr, is bigger for higher BZO NP densities. We also find that Δ H irr shows similar temperature dependences for samples with + 1BZO, + 2BZO (not shown) and + 3BZO for all orientations measured. At H||45°, Δ μ 0 H irr curves have a greater increase with respect to H||c for both concentrations, consistently with Fig. 5c, showing no signs of saturation up to the highest magnetic field measured.
The lack of saturation is also apparent in the inset of Fig. 6, which shows the ratio μ 0 H irr,BZO /μ 0 H irr,Y for the H||45° orientation for two different concentrations of BZO NPs, 1% and 3%. The enhancement factor is largest at low fields and initially decays with H, but then becomes constant up the highest accessible fields. In particular, H irr for + 3BZO is about 25% higher than for YBCO up to ~45 T. Also observed in Fig. 6 is a steeper growth in H irr at T > 80 K for both + 1BZO and + 3BZO wires. This temperature corresponds to μ 0 H ∼ 5 T, where the inter-vortex distance is a f ∼ 22 nm, which is close to the average BZO NP spacing (d BZO ∼ 30 nm) observed in TEM images. This suggests that NPs are most effective in increasing μ 0 H irr (at H||45°) when the intervortex spacing is similar the distance between NPs. This behavior is analogous to the J c enhancement we observed as function of field for different NP densities in ref. 36.
The fact that improvements are smaller for H||c compared to H||45°, is partly due to the fact that H irr is already higher due to the presence of correlated defects (insert Fig. 5d). After an initial increase of H irr , Δ μ 0 H irr flattens around 75 K and appears to increase again for T < 60 K. The initial increase can be related to the higher density of NPs and/or of TBs, with the latter being 2.3 times more numerous in samples with BZO than in standard YBCO. However, the increase at lower temperatures is most likely related solely to the presence of nanoparticles, given the similarity of the behavior found at H||45°.
The lack of saturation in the increase of H irr is remarkable. For + 3BZO at H||45° we have achieved a 25% enhancement up to the highest fields measured, with no sign of decrease. This corresponds to an absolute increase over 9 T for μ 0 H irr ~ 50 T, implying an increase of ~4-6 K in operational temperature (depending on the orientation). This enhancement is particularly significant knowing that at μ 0 H ∼ 60 T the inter-vortex spacing is at least 7 times smaller than the average distance between nanoparticles. These findings indicate that further improvement of H irr in practical REBCO wires is still possible, thus pushing the technically relevant region up to even higher magnetic fields.
In summary, we find that up to the highest BZO NPs density we tested, H irr can be increased further, with no indication of saturation up to 60 T. This allows us to report extremely high μ 0 H irr , the highest values reported so far for REBCO films and wires. Our results suggest that higher densities of NPs will increase H irr even more, and that the high-field limit is far from being reached in nanoengineered REBCO wires. These results are further proof that REBCO wire is an enabling technology for several high field applications.

Methods
HTS wires based on epitaxial REBCO nanocomposite films of standard YBCO and (Y,Gd)BCO + BZO were grown from metal organic solutions including Y-, Gd-, and Ba-trifluoroacetates and Cu-naphthenate with the cation ratio of 0.77 : 0.23 : 1.5 : 3 on ion-beam-assist deposited metal templates 37 . We added Zr-naphthenate into the (Y, Gd)BCO solutions; the content of BZO was 1-3wt%, and the concentration of starting solution was 1.2 mol/L. The total thickness of the superconductive layer for all samples was 0.5 μ m, which was confirmed by cross sectional transmission electron microscopy (TEM). The details of the calcinations and conversion steps have been published elsewhere 17 .
Films were patterned using photolithography and ion milling into bridges of ∼ 50 μ m width. The crystalline quality was examined by x-ray diffraction (XRD). The resistivity measurements, ρ vs. H, were realized in pulsed fields up to 60 T at the Pulsed Field Facility of the National High Magnetic Field Laboratory at Los Alamos National Laboratory. The four-probe technique was applied in AC mode with a low current corresponding to 400 A/cm 2 and a frequency of 100 kHz. Pulsed field measurements were performed at fixed orientations (θ), and the angular orientation was changed using a mechanical rotator. Additional ρ vs. T measurements were performed in a Quantum Design PPMS with a superconducting magnet (DC magnetic field) generating fields μ 0 H up to 15 T. For the transport measurements the current was always perpendicular to H. Criteria of 0.9ρ n and 0.01ρ n , where ρ n is the normal state resistivity, were used to define H c2 and H irr , respectively. Both planar and cross sectional TEM images were taken to evaluate the microstructure of the film.
The microstructures and elemental concentration mappings of the films were analyzed by transmission electron microscopy (TEM) and energy-dispersive x-ray spectroscopy (EDS), respectively.