High Jc and low anisotropy of hydrogen doped NdFeAsO superconducting thin film

The recent realisations of hydrogen doped LnFeAsO (Ln = Nd and Sm) superconducting epitaxial thin films call for further investigation of their structural and electrical transport properties. Here, we report on the microstructure of a NdFeAs(O,H) epitaxial thin film and its temperature, field, and orientation dependencies of the resistivity and the critical current density Jc. The superconducting transition temperature Tc is comparable to NdFeAs(O,F). Transmission electron microscopy investigation supported that hydrogen is homogenously substituted for oxygen. A high self-field Jc of over 10 MA/cm2 was recorded at 5 K, which is likely to be caused by a short London penetration depth. The anisotropic Ginzburg–Landau scaling for the angle dependence of Jc yielded temperature-dependent scaling parameters γJ that decreased from 1.6 at 30 K to 1.3 at 5 K. This is opposite to the behaviour of NdFeAs(O,F). Additionally, γJ of NdFeAs(O,H) is smaller than that of NdFeAs(O,F). Our results indicate that heavily electron doping by means of hydrogen substitution for oxygen in LnFeAsO is highly beneficial for achieving high Jc with low anisotropy without compromising Tc, which is favourable for high-field magnet applications.


Microstructure. Microstructural analysis by transmission electron microscopy (TEM) confirmed that our
NdFeAs(O,H) film is almost free of defects in the matrix as well as at the interface (Fig. 1a). The atomic-resolution annular dark-field (ADF) image agrees well with the crystal structure of NdFeAs(O,H) projected along the b-axis, as shown in the inset of Fig. 1a (top left). This ADF image also revealed the atomic arrangement at the NdFeAs(O,H)/MgO interface. The first atomic layer in the NdFeAs(O,H) film exhibits brighter contrast than surroundings, indicating that a Nd layer is firstly formed on the MgO substrate at the beginning of film growth. In this interfacial Nd layer, a large density of dislocations is introduced, as shown in the inset of Fig. 1a (bottom left). Those misfit dislocations compensate the large lattice parameter difference, i.e., a (NdFeAsO) = 3.99 Å while a (MgO) = 4.23 Å, resulting in the defect-free matrix inside the NdFeAs(O,H) film. Figure 1b,c shows magnified ADF images of NdFeAsO and NdFeAs(O,H), respectively, clearly indicating a shrinkage of the lattice in the c-axis direction by H substitution for oxygen. The c-axis lattice parameter decreased from 8.64 to 8.50 Å, as shown in the extracted intensity profiles (Fig. 1d). It is reported that the c-axis lattice parameter decreases with increasing hydrogen content x in LnFeAsO 1−x H x 7,10,11 with a rate of Δc/ Δx ~ −2-3 × 10 -3 Å/at%. The lattice parameter c of our NdFeAs(O,H) film determined by X-ray diffraction (XRD) was 8.437 ± 0.003 Å, which also supports the lattice shrinkage due to hydrogen doping although the value was slightly shorter than the average value evaluated from TEM. In order to check the homogeneity of hydrogen doping, the c-axis lattice parameters in the vicinity of the MgO substrate and near the film surface were compared, resulting in the same value (Fig. 1d). This result implies a homogeneous H substitution for oxygen, which guarantees that the transport properties shown below are not affected by local inhomogeneity.
Resistivity measurements for determining the magnetic phase diagram. Figure 2a,b summarises the field dependence of resistivity for both major field directions, H parallel to the ab-plane and to the c-axis. T c is recorded at 44 K, which is 2 K lower than the as-processed NdFeAs(O,H) film (Supplementary information fig. S1). The reason for the reduced T c may be that the sample was slightly damaged during bridge fabrication.
A clear shift of T c to lower temperatures with magnetic fields is observed for both directions. This shift together with a broadening of the transition is more obvious for H || c than || ab. The temperature dependencies of H c2 , Fig. 2c, show slopes of − 11.8 T/K for H || ab and − 2.7 T/K for H || c in the range 0 ≤ µ 0 H ≤ 4 T. Hence, the anisotropy of H c2 near T c is around γ Hc2 = 4.4, which is lower than for NdFeAs(O,F) film (γ Hc2 = 5.1) of similar thickness (22 nm) 14 . For cuprate superconductors, it has been shown that the anisotropy decreased with doping because of the increase in the interlayer coupling 15 . The decreased γ Hc2 for NdFeAs(O,H) may be explained similarly.
The temperature dependence of the irreversibility field H irr , Fig. 2c, for H || ab shows a kink around 4 T, which is due to a matching field effect. This effect has the same origin as reported for the 22-nm thick NdFeAs(O,F) film in ref. 14 . The matching field corresponds to the film thickness and is related to the Bean-Livingston barrier 16 . Hence, the origin of this matching field effect differs distinctly from the one commonly observed for H || c in    Thompson et al. 20 who argued that the exponents should be the same as the ones in the pinning force density analysis. These exponents (i.e., 0.5 and 2) suggest that Kramer's scaling for the pinning force density holds, which will be discussed later. For both directions, the exponent α is 0.07 at low fields, which can be explained by single vortex pinning 21 . The distinct feature for H || ab is that α changes from 0.07 to ~ 1 in the range 2-4 T, followed by 0.34 above 4 T, although the value of α ~ 1 may contain somewhat large uncertainty as we have only three data points in this field regime. Nevertheless, the exponent α = 1 indicates that collective pinning is dominating in this field regime 21 . The transition field at which the exponent α changes from 1 to 0.34 corresponds to the matching field shown in Fig. 2c. It is intriguing that the pinning mechanism for H || ab changes from single vortex pinning to collective pinning, followed by plastic pinning (i.e., α ~ 0.5 22 ).
Field dependence of J c and the pinning force density. Field dependence of J c for both H || ab and || c, and the corresponding pinning force density F p are summarised in Fig. 4a-d. Self-field J c of NdFeAs(O,H) at 5 K exceeds 10 MA/cm 2 . Another film with a T c of 45 K prepared by the same condition showed even a self-field J c of over 17 MA/cm 2 at 4 K 13 . These values are higher than our best-performing NdFeAs(O,F) film of similar   (Fig. 4a) and F p shows a linear increase above 4 T, indicative of strong single-vortex pinning. The reason for that is intrinsic pinning and will be discussed later. The elemental pinning force density per length for intrinsic pinning can be calculated by f The respective f p ' are 8.0 × 10 −5 N/m at 5 K, 4.2 × 10 −5 N/m at 10 K, 1.5 × 10 −5 N/m at 15 K, and 1.7 × 10 −6 N/m at 20 K. On the other hand, for H || c, J c monotonously decreases with increasing applied field, which reflects the absence of macroscopic defects in our film (i.e., a clean microstructure as can be seen in Fig. 1).
In order to understand the pinning mechanism for H || c, the normalised pinning force densities f p = F p /F p,max were plotted as a function of the reduced field h = H/H irr . H irr was evaluated from J c -H characteristics with a criterion of 1.4 kA/cm 2 in the temperature range 20 ≤ T ≤ 35 K. The fit of f p ~ h p (1 − h) q to each f p at given temperatures is shown in Supplementary information fig. S2, and the resulting fitting parameters p and q are plotted as a function of temperature (Fig. 4e). Although both p and q show a slight temperature dependence, the respective values of p and q are almost close to 0.5 and 2, suggesting that the Kramer model for shear breaking of the flux line lattice is mainly responsible for depinning 23 .
For T ≤ 15 K H irr cannot be evaluated from J c -H characteristics due to the experimental limitation. Hence, H irr was determined from fits to the pinning force density, on the assumption that the Kramer model prevails in the whole T range [i.e., (p, q) = (0.5, 2)].
The temperature dependence of H irr for H || c evaluated by three different methods (i.e., ρ(H, T), J c -H, and F p -H) is summarised in Fig. 2c. H irr in the temperature range 20 ≤ T ≤ 35 K from J c -H follow well the H irr -line expressed by Eq. (1) with an exponent k = 1.2, which is close to the theoretically predicted value of 4/3 for a glass-liquid transition 24,25 .
Here, T irr is the irreversibility temperature for self-field, which is 37.4 K. This result indicates that the criterion for determining H irr is quite reasonable and consistent. However, H irr starts to deviate from Eq. (1) at around 15 K. A steep increase of H irr at low temperatures was also observed in LaFeAs(O,F) 26 , where it was related to a similar increase of H c2 at the same temperature. This is due to the 2-dimensional multiband character of the  Angle dependence of J c . To further understand the pinning mechanism, the angular dependence of J c was measured at three different temperatures, T = 10, 20, and 30 K (Fig. 5). Simultaneously, the corresponding n values in E ~ J n is also plotted. As expected from the microstructural observation, the minimum J c is always observed at θ = 0° (i.e., H || c), whereas the maximum J c is located at θ = ± 90° (i.e., H || ab). Additionally, the J c peak at H || ab becomes sharper with increasing the applied field. Because the exponent n is proportional to the pinning potential U, J c (T, H, θ) should show a behaviour similar to n(T, H, θ) 28 . Indeed, this relation holds at 30 K. However, n(θ) at 20 K shows a dip at θ close to ± 90° for applied magnetic fields exceeding 3 T. At an even lower temperature of 10 K, a peak located at the local minimum around H || ab is observed (see, Fig. 5e: for clarity n(θ) at 14 T was plotted), which evolves with decreasing the field. Such behaviour can be explained by intrinsic pinning, as observed in REBCO [28][29][30] and FBS [31][32][33] , arising from the modulation of the superconducting order parameter along the crystallographic c-axis. Vortices depin from intrinsic pinning through the double-kink mechanism 34 , which easily creep along the ab-plane, resulting in small n. Here, the flux creep rate is proportional to the inverse of n − 1 35 . The cross-over temperature T cr from 3-dimensional Abrikosov to 2-dimensional Josephson vortices is, accordingly, located between 20 and 30 K. To determine T cr precisely, n(θ) around H || ab at 10 T with a step size of 1 K and n(T) for H || ab under magnetic fields 5 ≤ µ 0 H ≤ 14 T were measured (Supplementary information, Figs. S3 and S4). As a result, T cr is determined as 24.5 ± 0.5 K. Given that the FeAs layer spacing d is 0.8437 nm determined by XRD, the out-of-plane coherence length at zero kelvin, ξ c (0), can be estimated by The resultant ξ c (0) is 0.39 ± 0.01 nm, which is comparable to NdFeAs(O,F) 14,33 . To decouple the pinning contributions arising from uncorrelated and correlated defects, the anisotropic Ginzburg-Landau (AGL) scaling 36 for the angle dependence of J c can be applied. This approach has been widely used for REBCO 37 and FBS 26,32,33,38 . In the absence of correlated pinning centres (i.e., mainly randomly distributed and sufficiently small, isotropic pinning centres determine the pinning behaviour), all J c (θ) curves at a given temperature collapse onto a single curve if plotted as a function of effective field H eff : where γ J is the anisotropy parameter. The AGL scaling, Fig. 6, shows that some portion of J c (θ) curves at given temperatures indeed scale with H eff when γ J is appropriately chosen. γ J decreases from 1.6 to 1.25 with decreasing temperature in contrast to NdFeAs(O,F) 14,33 , where it increased. Clear deviations from the master curves due to the ab correlated pinning (here mostly intrinsic pinning because of the layered crystal structure) become obvious with decreasing temperature and also increasing field. www.nature.com/scientificreports/

Discussion
Our NdFeAs(O,H) film shows a high self-field J c exceeding 10 MA/cm 2 at 5 K, which is a record level value for pnictides without artificial pinning centres. According to Talantsev and Tallon 39 , self-field J c for type-II superconductors can be expressed by H c1 /λ, if the sample thickness is less than λ. Here, H c1 is the lower critical field and λ the relevant London penetration depth. Hence, the high self-field J c of NdFeAs(O,H) may be due to a short London penetration depth at heavily electron doping. Another effect of heavily electron doping is the reduction of anisotropy. The H c2 anisotropy near T c for NdFeAs(O,H) is γ Hc2 = 4.4, which is smaller than that of NdFeAs(O,F) (γ Hc2 = 5.1). Additionally, compared with NdFeAs(O,F), the temperature dependence of the anisotropy γ J evaluated from the AGL scaling for NdFeAs(O,H) 14,33 shows an opposite behaviour. It is also worth mentioning that γ J of NdFeAs(O,H) is comparable to that of Co-doped BaFe 2 As 2 38 . Heavily electron doping by means of hydrogen substitution for oxygen in LnFeAsO is a novel method to tune superconducting properties, whilst T c is maintained around 45 K, comparable to NdFeAs(O,F). For most FBS in contrast, a high carrier concentration reduces T c . Additionally, this method is rather simple, once the parent LnFeAsO films are fabricated. Now the parent compound can be fabricated by both pulsed laser deposition 40,41 and molecular beam epitaxy (MBE) 42,43 . Hence, our study motivates coated conductor preparation, for which films with thicknesses in the micrometer range are needed. However, a homogeneous H substitution for oxygen seems to be difficult in such thick films. Indeed, the H concentration showed to be inhomogeneous for 90-nm thick SmFeAs(O,H) films 11 . To realise LnFeAs(O,H) coated conductors and eventually applications of hydrogendoped LnFeAsO, new approaches to a homogeneous H substitution should be explored.
To conclude, we have grown hydrogen-doped NdFeAsO epitaxial thin films. TEM investigations supported that hydrogen is homogenously distributed. Detailed electric transport measurements revealed the benefits of heavily electron doping to LnFeAsO in terms of high self-field J c and low anisotropy without compromising T c .

Methods
Thin film fabrication. Parent NdFeAsO was grown on MgO(001) at 800 °C by MBE 40 . The structural characterisation by X-ray diffraction (XRD) confirmed that the 24-nm thick film was phase pure and epitaxially grown with (001)[100]NdFeAsO || (001)[100]MgO. After structural characterisation by XRD, the NdFeAsO films were cut into pieces of approximately 5 × 5 mm 2 and subsequently sealed in an evacuated silica-glass tube filled with ~ 0.5 g of CaH 2 powder that serves as a hydrogen source. Here, it is important that the film surface is in direct contact with the CaH 2 powders to promote a topotactic chemical reaction. The sealed silica-glass tube was heated to 490 °C at a rate of 100 °C/h, held at this temperature for 36 h, and then cooled to room temperature at a rate of 100 °C/h. The NdFeAs(O,H) film was also phase pure after processing, indicating that the crystalline quality is not compromised.
Microstructural analysis by TEM. The cross-sectional samples for TEM observation were fabricated by focused ion beam. Atomic-resolution observations were performed using a transmission electron microscope (Titan Cubed 60-300 G2, Thermo Fisher Scientific) which is equipped with a spherical aberration corrector (DCOR, CEOS GmbH) for the probe-forming lens system. The microscope was operated in the scanning TEM (STEM) mode at an accelerating voltage of 300 kV. The convergence semi-angle of the electron probe was set to 18 mrad. The typical probe diameter was less than 0.1 nm. An annular dark field (ADF) detector was positioned to detect scattered electrons of an angular range from 38 to 184 mrad. In order to measure the lattice parameters as accurately as possible, we employed a drift corrected frame integration available in Velox software (Thermo Fisher Scientific) to avoid image distortion due to sample drifting. The magnification of each image was calibrated by the lattice parameters of the MgO substrates.
Electrical transport properties. For temperature (T)-, field (H)-, and direction (θ)-dependence measurements of resistivity, ρ(T, H, θ), and critical current density, J c (T, H, θ), the NdFeAs(O,H) film was photolithographically patterned and Ar-ion beam etched to fabricate a small bridge of 30 µm width and 1 mm length. The sample was mounted on a rotator with maximum Lorentz force configuration, where the direction of the bias current is always perpendicular to that of the applied field. The angle θ is measured from the crystallographic c-axis. The critical temperature T c was determined as the intersection between the fit to the normal state resistivity and the steepest slope of resistivity. By measuring T c at various fields, the upper critical field H c2 versus T diagram was obtained. The bias current for resistivity measurements was 10 µA, corresponding to a current density of J b ~ 1.4 kA/cm 2 . The irreversibility field H irr was evaluated from ρ(T, H) and J c (T, H) data. For the former H irr is determined by the intersection between the ρ(T, H) curves and the resistivity criterion ρ c = E c /J b ~ 7.2 × 10 −7 mΩ cm, where E c (1 µV/cm) is the electric field criterion for determining J c (Supplementary information fig. S5). For the latter H irr was determined by the intersection between J c (T, H) curves and J b . At H irr , the electric field-current density J characteristics showed a relation that can be expressed as E ~ J n , where n was close to 1.