Influence of Metal Diboride and Dy2O3 Additions on Microstructure and Properties of MgB2 Fabricated at High Temperatures and under Pressure

High temperatures and under pressure (HTP) processing has been used to study the effects of chemical doping in MgB2. ZrB2, TiB2 and NbB2 were selected as additives since, like MgB2, they have an AlB2-type structure and similar lattice parameters. Dy2O3 was selected as it has been reported to generate nanoscale, secondary intragrain phases in MgB2. While C is known to enter the B-sublattice readily, attempts to dope Zr and other elements onto the Mg site have been less successful due to slow bulk diffusion, low solubility in MgB2, or both. We have used high-temperature, solid-state sintering (1500 °C), as well as excursions through the peritectic temperature (up to 1700 °C), to investigate both of these limitations. Bulk MgB2 samples doped with MB2 (M = Zr, Ti and Nb) and Dy2O3 additions were synthesized and then characterized. Lattice distortion and high densities of crystal defects were observed in the MgB2 grains around nano-sized MB2 inclusions, this highly defected band contributed to a large increase in Bc2 but was not large enough to increase the irreversibility field. In contrast, distributed intragrain precipitates were formed by Dy2O3 additions which did not change the lattice parameters, Tc, Tc distribution or Bc2 of MgB2, but modified the flux pinning.

. Further studies focusing on the AlB 2 -like metal diborides ZrB 2 , TiB 2 and NbB 2 , (e.g. refs 11-16) yielded contradictory results. Feng et al. 11,12 reported an enhancement in B c2 in response to 10 mol% Zr doping; Bhatia et al. 13 observed a significant increase in B c2 (from 20.5 T to 28.6 T at 4.2 K) after adding 7.5 mol% ZrB 2 to MgB 2 bulks. On the other hand, Zhang et al. 14 reported no B c2 enhancement in ZrB 2 doped MgB 2 tapes. In any case, while B c2 enhancements have been noted by various researchers working with MgB 2 PIT or powder type processes, no one has reported enhanced transport current, suggesting that the effect may be in a surface layer. The one effort to date which has clearly injected Zr deeply into the grain, resulting in a pulsed laser deposition (PLD) synthesized ZrB 2 -doped MgB 2 thin film 15,16 , showed a much stronger response to the presence of Zr, and in this case a decrease of T c and B c2 with increasing Zr content. These various observations give rise to the question: what is the actual influence of Zr doping in MgB 2 ? The possible roles of Zr in MgB 2 can be summarized in terms of: 1) extrinsic effects, such as modified intergranular connectivity and reduced grain size 11,12 ; 2) intrinsic effects, such as an influence on B c2 of Zr substitution for Mg [13][14][15][16] , or increased flux pinning by a distribution of nano-sized ZrB 2 /Zr Scientific RepoRts | 6:29306 | DOI: 10.1038/srep29306 precipitates 14 . On the other hand, incomplete microscopic evidence of Zr substitution for Mg has been provided, at least for materials made by equilibrium processes (contrasting to the non-equilibrium processing of the films of 15,16 ). Therefore further study on the limits of Zr doping was deemed necessary.
There are many roadblocks to clarifying the true role of chemical doping in MgB 2 . Chief among them is that homogeneous doping is very hard to achieve. Traditional powder synthesis is generally performed at 600-1000 °C -too low to form homogeneously doped samples. To overcome this problem a high temperature under pressure (HTP) route (see below) was developed to explore solubility limits of dopant species in MgB 2 and maximize diffusion during reaction. MgB 2 bulks synthesized by HTP should have a greater depth of dopant penetration into the MgB 2 for any species introduced (if it is soluble) given the increase in diffusion rate at higher temperatures. HTP samples also have large grain size (over 5 μ m) making chemical analysis easier. The purpose of the HTP method is to minimize any diffusion limitations so that we can explore the solubility limits of doping, rather than to fabricate MgB 2 bulk samples with high J c . Also, the use of existing metal diborides (MB 2 ) with a structure isomorphous to MgB 2 (P6/mmm) as a vector for effective metal element doping is a promising way to investigate possible changes of superconducting properties like B c2 and T c . Thus, three sets of MgB 2 bulks doped with ZrB 2 , TiB 2 and NbB 2 powders were prepared.
Most additions which have been attempted for MgB 2 tend to accumulate at the grain boundaries, with the exception of the above-mentioned C-bearing additions. On the other hand, several studies have shown that a very small amount of Dy 2 O 3 17 can form nanosize precipitates within the MgB 2 grains and thereby enhance flux pinning without changing T c . Thus a Dy 2 O 3 doped MgB 2 bulk was also included in this study for comparison. The changes in microstructure and lattice parameter, as well as the superconducting properties B c2 , T c and flux pinning were studied for all samples and are discussed below.

Results and Discussion
Influence of MB 2 and Dy 2 O 3 doping on XRD and lattice constants. The X-ray diffraction data for all HTP bulks are presented in Fig. 1 where the Bragg reflections are indexed for only the MgB 2 phase for simplicity. MgO and Mg were present at some level in all samples (MgO was less than about 2 wt% for samples reacted below the peritectic). MgB 4 peaks are also present for samples HT at 1700 °C. Only a very small peak shift (less than 0.2 degree) at both (110) and (002) was observed in MB 2 doped samples while no peak shift was observed for the Dy 2 O 3 doped sample. Peaks corresponding to MB x impurity phases were observed in all MB 2 doped samples. The lattice parameters extracted from pseudo-Voigt fitting the MgB 2 peak reflections and the calculated lattice parameters by Vegard's law are given in Table 1. For the Dy 2 O 3 added sample HTP-DY, similar to Chen's report 17 , the lattice parameters a and c did not change with Dy 2 O 3 addition. Similarly, for the MB 2 added samples, even though both lattice parameters a and c were slightly changed, these changes were very small and none of the doped samples obeyed Vegard's law, unlike C-doped MgB 2 HTP bulks 6 . Therefore it seems that even under HTP processing the metal borides (ZrB 2 , TiB 2 and NbB 2 ) mainly acted as impurity phases and did not form homogeneous solid solutions with MgB 2 , at least not to an extent detectable by XRD. However, the small changes in lattice parameters suggest that a distortion of the MgB 2 lattice was present. Such distortion, caused by strain generated around these dopant impurities (see below) rather than elemental substitution, appears to be the driver for the modified upper critical field B c2 of the doped samples.   (110) and (002), as these two peaks are directly related to lattice parameter a and c, respectively. Note for the MB 2 doped samples, small peak shifting and MB 2 peaks are observed, while no peak shifting is observed in the Dy 2 O 3 doped sample.

Influence of MB 2 and Dy
in 6,18,19 ). Two phases are visible: MgB 2 (majority phase, dark grey) and ZrB 2 (minority phase, white) in HTP-Zr-01; while in HTP-Zr-02, Mg and MgB 4 phases were present since it experienced (upon cooling) the reaction Figure 2(c) represents the bright-field (BF) TEM image obtained from a thin foil extracted from HTP-Zr-02 (1700 °C). This foil contains a cross section of several grains. The results from the energy dispersive spectroscopy analysis (EDS) and selected area diffraction (SAD) confirm that these large grains are MgB 2 . An intragranular crack and dislocation loops are present in one of the grains. Since HTP-Zr-02 was processed above the peritectic, the crack presumably resulted from volume expansion taking place during cooling as the MgB 4 converted into MgB 2 . TEM examination revealed a number of impurity phases in the form of 30-80 nm inclusions around the MgB 2 grains, Fig. 2(d). The results of EDS analysis performed on these inclusions are presented in Fig. 2(e). It is clear that these inclusions contain Zr, or possibly ZrB 2 , which is likely dispersed around the MgB 2 grains in the bulk. Compared to other areas inside the MgB 2 grains, the regions around these ZrB 2 inclusions have much higher contrast under BF condition, which suggests that strain fields were generated around these inclusions and that the MgB 2 lattice was distorted locally. This local lattice distortion may be the origin of the slight lattice parameter changes observed by XRD analysis in Section 3.1. Figure 2(f) shows HAADF imaging for one of these nano-size inclusions. Since HAADF imaging is sensitive to variations in the atomic number (Z-contrast), these white inclusions should have a higher average atomic number than MgB 2 . A STEM-EDS line scan applied using 21 distinct points over ~100 nm across this inclusion confirmed it was ZrB 2 . The Zr signal dropped to zero quickly outside of the inclusion, beyond the ZrB 2 /MgB 2 interface. The spatial resolution of the STEM-EDS line scans is about 5 nm, thus these observation indicate that Zr did not notably penetrate into the MgB 2 lattice. Figure 3 shows the BSE images of the TiB 2 doped samples, HTP-Ti-01 (1500 °C, below the peritectic) and HTP-Ti-02 (1700 °C, above the peritectic). Similar to the behavior of ZrB 2 , TiB 2 mainly acts as an impurity phase (light grey in Fig. 4(a,b)) and is widely distributed. A TEM thin foil containing a cross section of both TiB 2 and MgB 2 grains was carefully extracted from HTP-Ti-01. A BF image including MgB 2 , TiB 2 , and their interface is represented in Fig. 3(c). A large number of defects can be observed inside the MgB 2 grains close to the MgB 2 / TiB 2 interface. Inclusions 100-200 nm in size are found at the interface as well as at MgB 2 grain boundaries. EDS analysis confirms that these inclusions are MgO, Fig. 3(f). Dark-field (DF) imaging was also used to examine the dislocations and the interface since crystal defects have stronger contrast under DF conditions. In Fig. 3(d), the DF image clearly confirms that the MgB 2 grain contains a high density of defects. Detailed in-grain analysis was performed using HAADF imaging. Figure 3(e) shows a HAADF image of MgB 2 grains with high defect density. Nano-size inclusions (~10-30 nm, white) dispersed both in and around MgB 2 grains were observed. EDS analysis performed on randomly selected white inclusions confirmed that they were TiB 2 (EDS spectrum of spot E (inclusion) in Fig. 3(f)). No Ti was detected by EDS in the other regions of MgB 2 grains (EDS spectrum of spot D (matrix) in Fig. 3(f)). STEM-EDS analysis was applied across inclusion E, and beyond the TiB 2 /MgB 2 interface the intensity of the Ti signal quickly dropped to zero, indicating that Ti did not dissolve into the MgB 2 lattice. These nano-size TiB 2 inclusions can also contribute to the high defect density observed in MgB 2 grains in Fig. 3(c,d).
In the NbB 2 -added sample HTP-Nb-01 (HT below the peritectic), three phases are visible in the BSE images of Figs 4(a) and 5(b): MgB 2 (majority phase, dark grey), MgO (minority phase, light grey) and NbB 2 (minority phase, white). In Fig. 4(b), based on fractured secondary electron (SE) imaging by 'through the lens' (TTL) detection, NbB 2 particles are observed outside the MgB 2 grains. These particles are small (~300-500 nm), well connected with the MgB 2 grains, and dispersed throughout the bulks.
Further analysis was performed on a TEM thin foil sectioned from HTP-Nb-01. Both BF (Fig. 4(c)) and DF images (Inset of Fig. 4(c)) show nano-size inclusions (~300 nm) embedded in the MgB 2 grain boundaries. Moreover, a large number of defects can be observed inside the MgB 2 grains around these inclusions, while the other MgB 2 grains have fewer intragranular defects. The EDS results in Fig. 4(e) confirm that these inclusions are NbB 2 . HAADF imaging performed on MgB 2 grains with high density of defects is presented in Figs 4(d) and 5(e). Nano-size inclusions (~10-50 nm, white) were found inside these grains and high strain fields were observed around them. EDS analysis was applied on these distinct inclusions and several randomly selected spots in the matrix; those for spot C (matrix) and spot D (inclusion) are presented in Fig. 4(f). These EDS results confirm that these white inclusions were NbB 2 . A STEM-EDS line scan was applied across inclusion D. The intensity of Nb signals abruptly decreased from ~10 4 to 0 across the NbB 2 /MgB 2 interface and no Nb was detected in other regions of the MgB 2 grains. The microstructure of HTP-DY (HT above the peritectic) was investigated by BSE in Fig. 5(a). Five phases are visible: MgB 2 (majority phase, dark grey), Mg (main phase, grey), MgB 4 (minority phase, black), MgO (minority phase, light grey) and Dy-containing inclusions (minority phase, white). These Dy-containing inclusions (DyB 4 according to XRD results) with a size of ~100 nm were dispersed throughout the bulk. Bright-field TEM examination revealed a number of impurity phases in the form of ~10-50 nm inclusions inside the MgB 2 grains in Fig. 5(b). A low density of large inclusions (over 100 nm) was also observed (Fig. 5(c)). HAADF imaging (Fig. 5(c)) showed that these nano-size inclusions had a higher average atomic weight. EDS indicates that these inclusions contained Dy and B suggesting they are the previously XRD-identified  respectively. The unchanged T c s suggest that a portion of these ZrB 2 doped samples was unaffected with a T c equal to that of the undoped sample. Figures 6(a,b), show very broad transitions and bi-modal peak in the T c distribution. This effect became more severe in HTP-Zr-02 indicating the presence of regions with various T c s. The results for each of the MB 2 samples was similar-after MB 2 doping, the onset T c s were relatively unaffected, however their transition widths were significantly enhanced. We interpret this effect in terms of the presence of nanoscale MB 2 (where M = Zr, Nb, or Ti) second phases which produce locally distorted regions separated by large regions of unaffected MgB 2 , leading to broadened T c distributions with a wide T c variation ranging from 39 K to low values. Since these MB 2 additives are isomorphous to MgB 2 and their lattice parameters are close to those of MgB 2 , the localized distortion is probably due to the coherent strain generated around the MB 2 inclusions. However, in HTP-DY both the onset T c and FWHM did not change by adding Dy 2 O 3 , which is consistent with Chen's observation 17 . The lattice parameters and crystal structures of Dy 2 O 3 (cubic with space group Ia-3) 20 and DyB 4 (tetragonal with space group P4/mbm) 21 are very different from MgB 2 (hexagonal with space group P6/ mmm) 22 , therefore it is unlikely that the Dy-contained inclusions in HTP-DY can generate coherent strain in the MgB 2 grains.
As indicated above the MB 2 dopants were mostly found as distinct impurity inclusions that only influenced the surrounding MgB 2 grains through the MgB 2 /MB 2 interfaces. Increasing the concentration of MB 2 inclusions produced more "affected zones" leading to a wider T c distribution, Fig. 6(b). The behaviors of MB 2 doped samples are quite different from those of C-doped MgB 2 bulks 6 . After doping with 6.2 at.% C Susner et al. 6 observed a significant decrease in the onset T c , from 39.5 K to ~24 K, while the FWHM changed from 0.65 K to 1.4 K 6 . Since C is known to be a substitutional defect, if homogeneous C doping is achieved, the onset T c and the lattice parameter a will decrease simultaneously with increasing C doping levels 6 . Under MB 2 doping, it seems that Zr, Ti and Nb did not substitute for Mg or form homogeneous solid solutions with MgB 2 , even under 1700 °C and 10 MPa. However, the properties of the host lattices in the vicinities of these dopants were indeed affected and their T c s were clearly altered, possibly due to local compositional changes caused either by Mg diffusion into MB 2 particles or by local strain. Both of these possibilities could cause T c reduction, comparable to the effect of Al doping in MgB 2 [23][24][25] . Based on the results in the previous section, the affected vicinities probably had thicknesses similar to or smaller than 5 nm-the resolution of STEM-EDS line scans used in this study.
The T dependencies of the upper critical fields B c2 of all the samples are presented in Fig. 6(c). The B c2 s at 20 K linearly extrapolated from Fig. 6(c) are listed in Table 2. It is clear that B c2 was increased by MB 2 doping, but not by Dy 2 O 3 doping. It is important to note, however, that since these MB 2 doped samples were not homogeneous (as evidenced by the microstructure and the T c distribution), the B c2 values represent the properties of only a small fraction of the bulk samples. In other words, some "affected zones" inside these doped bulks have higher B c2 s than those in the unaffected MgB 2 , therefore their measured B c2 was enhanced. The observed high defect densities in the MgB 2 grains, which increase electron scattering and reduce the electron mean free path, are likely responsible for the B c2 enhancement. These regions are at the edge of the grains, and can therefore act as connected percolative paths. In the Dy 2 O 3 doped sample, although nano-size inclusions were observed inside the MgB 2 grains, B c2 did not change. This observation together with the absence of changes in the lattice parameters, and lack of change in the T c and FWHM suggested that Dy 2 O 3 , unlike the metal diboride additions, did not cause a band of defect structure at the boundary of the MgB 2 grain.
The magnetic critical current density J cm s and flux pinning behaviors of selected samples were calculated based on Bean's critical state model: where Δ M is the width of the hysteresis loop at a given field B, a and b are the edge lengths of the sample orthogonal to B (a > b). The results at 15 K are shown in Fig. 7. The J cm s for most of the field range were either not changed, or even reduced after MB 2 doping; for HTP-DY, its J cm was slightly increased at all measured fields. A "tail" in J cm (B) can be observed in all MB 2 doped samples, Fig. 7(a). Based on the microstructural evidence and results of B c2 and T c , this "tail" in J cm of the MB 2 doped samples is probably caused by regions in the samples with different B c2 s. However these regions were too small to have significant influence on the overall J c (> 100 A/cm 2 ). The irreversibility field, B irr , defined as the point where flux pinning vanishes, is often taken as the field at which J c (B) = 100 A/cm 2 . The results for these samples are given in Table 2; no increase in B irr is seen, and in some cases there is a decrease. This definition of B irr does not capture the high field and super-low-J c "tail" observed in Fig. 7(a). MgB 2 is primarily a grain boundary pinner, and thus the starting place to describe its pinning is the Kramer function (although deviations are seen). In order to perform such analysis, a Kramer field is needed. A Kramer plot, J c 0.5 B 0.25 versus B, is shown in Fig. 7(b). The Kramer fields, B k , taken at the cross-intercepts of linear fittings (Black dash lines in Fig. 7(b)) are listed in Table 2. The values of B k , similarly to those of B irr , were stable or slightly reduced after MB 2 doping, unlike the values of B c2 which increased with doping. This effect is due to several factors: (1) the higher B c2 region was apparently small, presumably restricted to the defected zones near the grain boundaries, these regions will not substantially influence the measured B k ; (2) the B k is also affected by the sample connectivity, possibly reduced with second phases present. For samples with Dy 2 O 3 additions, B c2 , B irr , and B k were not affected. Figure 7 (c) is a plot of bulk pinning force density (F p = J cm × B) vs B; the maximum values, F pmax , are listed in Table 2. Compared to the literature values, the F p,max of these HTP processed bulks is quite small. Spark-plasma sintering 26 gives a J c at 2 T and 15 K of about 10 5 A/cm 2 , while our values here are closer to 2 × 10 3 A/cm 2 . The F p,max of the undoped sample HTP-01 is only ~0.095 GN/m 3 while according to Susner 6 at 15 K the F p,max of an undoped MgB 2 wire was about 2 GN/m 3 . The main reason for the difference is that the MgB 2 grains in these HTP bulks (> 5 μ m) are much larger than those in the traditional synthesized samples (typically 30-500 nm). Other factors could also contribute to this difference in F p,max , including some reduction of connectivity by small amounts of MgO. However, as J c ∝ 1/grain size, and the grain size in our samples is roughly 50 times larger than the highest performing MgB 2 , the grain size effect is expected to be dominant. Among all samples, the highest value of F p,max (~0.135 GN/m 3 ) was observed in HTP-DY. The normalized bulk pinning force density = f F F / p p p, max is plotted against normalized magnetic field b = B/B k in Fig. 7(d). The functions of grain boundary (GB) pinning from the Dew-Hughes model 27 2 , are also plotted for comparison. Although the undoped sample HTP-01 followed GB pinning function quite well, all doped samples show a deviation from the standard function. The peaks of f p in the doped samples were shifted from the value of b = 0.2 (the peak position of the GB pinning) to lower values. For example, the peaks in HTP-Zr-01, HTP-Ti-01, HTP-Nb-01 and HTP-DY were 0.12, 0.13, 0.11 and 0.10, respectively. This observation which was also reported by Matsushita et al. 28 in C doped MgB 2 bulk samples can be explained by two possibilities: (1) These doped samples might contain a set of local B k s instead of one distinct value (just like the B c2 s in the MB 2 doped samples), which can lead to an artificial error in the estimation of the peak positions; while the variation in B k of the undoped sample was small, thus the undoped sample followed the GB pinning function. (2) The deviation from b peak = 0.2 might be caused by the operation of other pinning mechanisms (e.g., normal volume pinning in which f p maximizes at b → 0.0 27 ) in association with the GB pinning. As noted above, we clearly see a distribution of Dy-based second phases, consistent with 17 . Also present were modest levels of MgO, known both to act as a pinner and in some cases reduce connectivity 29   on this expression, the values of a vary from ~50 nm at 1 T to ~20 nm at 6 T. By definition, the size of the volume pins needs to be larger than a. Considering the fact that these doped samples contained intragrain inclusions some of which were bigger than the FLL parameter a at every measured field, it appears that volume pinning contributed to these shifts in F p,max .
In summary, after adding MB 2 , the B c2 of MgB 2 HTP bulks increased, the T c distributions were broadened, but T c , B k and J c remained unchanged (or slightly reduced). Considering the microstructural evidence, this observation can be explained as follows: only very small regions (possible ≤ 5 nm in thickness) around dopant particles of the MgB 2 grains are influenced by doping, leaving the majority of MgB 2 unaffected. To the contrary, the Dy 2 O 3 doping did not change the T c , T c distribution and B c2 , instead it increased the J c and flux pinning apparently associated with the nano-size precipitates in MgB 2 grains.

Conclusion
In this work we have used our HTP method for synthesizing doped MgB 2 bulks at high temperatures (up to 1700 °C) and at pressure (10 MPa) to explore solubility limits of dopant species in MgB 2 , maximize diffusion, and (alternatively) attempt to form dense, nanoscale secondary phases during the sample synthesis. We explored both metal diborides (MB 2 , where M = Zr, Ti and Nb) for attempted Mg site substitution and Dy 2 O 3 for nanoscale intragrain precipitate formation. Using the HTP process we conclusively show that the large increases in B c2 with metal diboride additions are due to a highly defected band within the grain, rather than substitution or inclusion within the grain, or grain boundary effects. High defect densities observed in MgB 2 grains around/with these MB 2 inclusions, cause electron scattering and therefore contribute to the B c2 enhancement and T c distribution broadening. On the other hand, these regions (≤ 5 nm in thickness) were not large enough to significantly influence the high field J c , B irr , or B k . This observation explains the frequently observed increases seen for B c2 in materials with no accompanying increase in transport current. We also confirm the previously observed but sparsely distributed intragrain precipitates formed with Dy 2 O 3 additions. Dy 2 O 3 additions not change the lattice parameters, T c , T c distribution and B c2 of MgB 2 , but increased the J c and flux pinning by forming an array of nano-size precipitates in MgB 2 grains.

Methods
Sample Synthesis. Three sets of MgB 2 bulks with various MB 2 (M = Zr, Ti and Nb) dopants were fabricated at high temperatures and under pressure. This HTP process 18,19 based on the reactive liquid Mg infiltration (Mg-RLI) method 30 . Three metal borides with a structure isomorphous to MgB 2 (P6/mmm) were selected as vectors for Mg-site substitution: ZrB 2 (99.5%, Alfa Aesar), TiB 2 (99.5%, Alfa Aesar) and NbB 2 (99.5%, Alfa Aesar). As the Dy 2 O 3 additive, Dy 2 O 3 (> 99.9%, < 100 nm particle size, ALDRICH) was used. Amorphous B powder (50-100 nm in size) manufactured by Specialty Metals Inc. 31,32 was hand mixed with the dopant powder and high energy ball milled for 15 min in an Ar atmosphere. These dopants and B powder mixtures were then pressed into   18,19 ). All samples were heat treated at 10 MPa in an Ar atmosphere. Two heat treatment routes were used: (1) heating up to 1500 °C and soaking for 30 min; (2) heating up to 1700 °C and soaking for 20 min. A slow cooling rate of 5 °C/min was used in both HT routes to maintain thermal equilibrium. The first route was designed to limit the temperature to just below the peritectic decomposition point of the reaction, thus preventing decomposition while maximizing the diffusion of the dopant species. The second route was designed to allow the reaction to occur on the temperature upswing, and hence to form MgB 2 directly from MgB 4 and Mg + dopant species on cooling: where T p is the peritectic temperature (~1500 °C in our experiments).

Measurements.
A Rigaku SmartLab X-ray diffractometer (using Cu Kα of 1.5406 Å) was used for structural characterization and the scanning angle 2θ ranged from 20° to 80°. A FEI/Philips Sirion scanning electron microscope (SEM) with a field-emission source and a through-the-lens (TTL) detector was used for microstructural imaging. An FEI Helios 600 dual beam focused ion beam instrument (FIB) with an Omniprobe micromanipulation tool was used to prepare TEM thin foils. The TEM imaging was performed on a FEI/Philips CM-200T transmission electron microscope (TEM) with a silicon drift detector (SDD) and energy-dispersive X-ray spectroscopy function (EDS). The high-angle annular dark-field imaging (HAADF) and EDS line scans with a resolution of 5 nm were performed on a Tecnai F20 system field emission 200 kV scanning transmission electron microscope (STEM) with an X-TWIN lens and high brightness field emission electron gun (FEG).
The magnetic properties were measured by using a Quantum Design Model 6000 PPMS with 4.2 K < T < 300 K and − 10 T < B < 10 T. The superconducting critical transition temperature T c and T c distribution were determined by DC magnetic susceptibility methods. The T c was defined as the onset of superconductivity from the normal state at 10 mT and the T c distribution was expressed in terms of dχ /dT, where χ is the DC susceptibility. M-T curves were taken at 1 T intervals from 0-14 T. The upper critical field, B c2 , was determined by the highest temperature point where the M-T curves deviated from M = 0 at each given field. The irreversibility field, B irr , was taken as the field at which J c (B) = 100 A/cm 2 . The Kramer field, B k , was taken as the point where J c 0.5 B 0.25 extrapolated to zero on a Kramer plot (Fig. 7(b)), and the bulk pinning force density, F p , was calculated from F p = J c B, where J c was extracted from the magnetization results at various temperatures.