Improvement of the critical current density on in situ PIT processed Fe/MgB₂ wires by oleic acid addition

Monofilament conductors have been fabricated by the PIT method using in situ reactions of the components and mechanical conformation by drawing through round dies. Fe tubes (99.5%, Goodfellow, with outer diameter 5 mm and 0.25 mm wall thickness) were filled by hand with different precursor powder mixtures (details are given below). During drawing, an intermediate annealing was necessary to reduce the Fe sheath's work hardening. The final wires have 1.1 mm outer diameter.

The starting materials were commercial Mg powders (99.8%, maximum particle size 250 μm, Goodfellow), amorphous B powders (99% and mean particle size lower than 1 μm, New Metal & Chemicals Ltd) with stoichiometric proportions, and liquid oleic acid (99%, Alfa Aesar). All the precursors were mixed for 30 min using a Retsch MM 2000 mixer mill. Results on four different wires are reported.

• W0 (Undoped reference wire). This was made from a dry mixture of Mg and B powders filled into Fe tubes. An intermediate heat treatment at 550 ° C × 0.5 h in Ar atmosphere was done during drawing.
• W5-n (Mg + B powders mixed with oleic acid, non-preheated). Oleic acid (5 wt% of total MgB₂) was mixed with Mg and B powders, and the obtained mixture directly filled into the Fe tubes without preheat treatments, thus keeping inside the full oleic amount. The intermediate annealing during drawing was done at 320 ° C × 24 h (below the boiling point of oleic acid, 360 ° C) in Ar atmosphere.
• W10 (Mg + B powders mixed with oleic acid and preheated). Oleic acid (10 wt% of total MgB₂) was mixed with Mg and B powders, and before filling the Fe tube the mixture was preheated at 400 ° C × 1 h in Ar to evaporate the less-bounded and unreacted molecules, i.e. to eliminate the oleic acid excess. Two wires with different intermediate softening annealing treatments during drawing were prepared: 550 ° C × 0.5 h and 400 ° C × 0.5 h, in Ar atmosphere.
• W10-B (B powders treated with oleic acid and preheated before Mg mixtures). First, the B powders were mixed with oleic acid (10 wt% of total MgB₂) and heat treated at 500 ° C × 1 h in vacuum (0.1 mbar) or Ar, to evaporate the oleic acid excess. The resulting powders were mixed with Mg and used to fill the Fe tubes. An intermediate annealing for Fe softening at 550 ° C × 0.5 h in Ar atmosphere was done during drawing.

Two different final annealing conditions to form the superconducting phase have been studied: 700 ° C × 2 h and 670 ° C × 5 h, in vacuum (0.1 mbar).

The phase composition and microstructure of the wires was analysed by x-ray diffraction (XRD, RIGAKU D/max 2500 Cu Kα), field-emission scanning electron microscopy (FESEM) using angle-selective backscattered electrons (AsB), which gives information about the phase composition and crystal orientation, and energy-dispersive x-ray spectroscopy (EDX).

Thermogravimetric analysis (TGA) was performed using a TA Instruments SDT Q600 system to register the weight loss along programmed temperature scans of oleic acid and mixtures of oleic acid with Mg, B, and Mg + B powder samples. Pt crucibles and a heating ramp of 20 °C min⁻¹ from room temperature up to 550 ° C in Ar flow were used.

The magnetic measurements were carried out on 5 mm long cylindrical samples using a superconducting quantum interference device (SQUID, Quantum Design MPMS-5T) and a physical properties measurement system (Quantum Design PPMS-9T and PPMS-14T) with the field perpendicular to the sample axis and zero-field cooling. The ac magnetic susceptibility χ_ac (in-phase, χ', and out-of-phase, χ'', components) was measured as a function of temperature at a frequency of 120 Hz and magnetic field amplitude of 0.1 mT. Magnetization measurements were done at 5 and 20 K. To avoid the contribution of the Fe sheath to the magnetic measurements, it was removed by careful mechanical polishing.

Electric transport measurements were performed inside the bore of an 8 T coil magnet immersed in liquid helium using wire sample lengths of 3.7 cm, maximum currents about 220 A, and magnetic fields perpendicular to the wire. In order to avoid the heating of the sample, current pulses of 2 s were applied to measure the voltage versus current curves, V(I). A four-point technique was used, and the critical current was estimated by the usual 1 μV cm⁻¹ criterion.

The XRD patterns obtained on wire cores, after removing the Fe sheath, show in all wires the characteristic peaks of the MgB₂ main phase together with small amounts of MgO and without signatures of other crystalline compounds. In figure 1, the characteristic Bragg reflexion peaks (002) and (110) of the MgB₂ structure are displayed. It is observed that the position of the (001) peaks remains invariant for all wires while the (110) peak moves towards higher 2θ angles for the doped wires. This indicates that the doping has not affected the c-axis cell parameter whereas the a-axis length decreases (see table 1), which confirms the random substitution of B by C in the MgB₂ lattice [13]. The averaged doping amount, x, of the Mg(B_1−xC_x)₂ grains in the wire core have been obtained from the estimated a unit-cell parameter and by using the nearly lineal dependence a(x) observed by Kazakov et al [27] for doped single crystals. The error in the estimation of carbon doping is about 15–20%; therefore, the x-value of wire W10 would be higher but close to the W10-B value. In the figure it is also observed that the height of the MgO peak is similar for all samples, although it is slightly higher for W5-n and lower for W0.

Zoom In Zoom Out Reset image size

Considering the full width at half-maximum (FWHM) values of the reflexion peaks, it is observed that the FWHM of the (001) peaks remain constant for all wires, while the (110) peaks become broader for doped wires, with higher width for W5-n (see table 1). The similarities in FWHM of the (002) peak and the differences for the (110) one in the wire with larger C doping is a strong indication of an increase of crystallographic disorder or a reduction of crystallinity, but it also may be due to segregations of phases with different C content or to lack of homogeneity in the doped grains.

χ_ac(T) measurements have been used to determine the critical temperatures, T_c, and to analyse the sharpness of the superconductor to normal state transitions in all wires. The measured χ_ac(T) curves are shown in figure 2. The T_c-values of the wires, determined by the onset of diamagnetism by χ'(T), have been collected in table 1. They decrease from 37.5 K for the undoped wire (W0) down to 34.5 K for W5-n, and are in good agreement with the expected variations from the estimated C substitution [27].

Zoom In Zoom Out Reset image size

The transition widths, ΔT_w, defined as the temperature interval from 90% to 10% of χ'(5 K) (i.e., of complete diamagnetism), are also collected in table 1. Both W0 and W10 wires have the same value ΔT_w = 0.7 K for all annealing conditions, while, for the W10-B wire, ΔT_w = 2 K for final annealing at 700 ° C × 2 h, and the value decreases to ΔT_w = 1 K for 670 ° C × 5 h. Wire W5-n shows a very wide transition, ΔT_w = 13.6 K, which likely indicates the presence of important amounts of non-superconducting phases in this wire.

The FESEM micrographs on transverse cross-sections of these wires with different magnifications are displayed in figure 3. The overall transverse cross-section of W10, displayed in figure 3(a), shows the porous microstructure of the core, typical of in situ PIT wires. The surrounding Fe sheath shows reactivity with B in a layer of thickness 5–8 μm, which is similar for all fabricated wires and cannot be seen in this low-magnification micrograph. There are voids of several micrometres, left by Mg when it diffuses into the B particles, and MgB₂ regions quite uniformly distributed. W10 wires present very homogeneous MgB₂ phase compositions for all annealing conditions, as may be seen in the enlarged micrograph of figure 3(b), which is similar to the microstructure observed on undoped wire W0.

Zoom In Zoom Out Reset image size

A similar FESEM analysis was performed for W10-B. We observed that in this wire there are large areas of size 250–400 μm without voids, unlike W0 and W10 wires, which show a homogeneous distribution of voids. For annealing condition 700 ° C × 2 h, areas with two-phase compositions are abundant, as seen in figure 3(c). EDX analysis showed that darker grey areas correspond to boron-rich zones, while light grey ones are MgB₂. In the overall view of the cross-section of this wire, it is seen that these darker grey zones are more abundant in the centre of this dense large area, while the lighter grey ones are closer to voids. When longer annealing times and slightly lower temperatures are used, 670 ° C × 5 h, more areas with homogeneous phase composition are observed (see figure 3(d)) in agreement with [11], although zones with boron- rich phases still persist. These results indicate that there is a lack of precursor distribution homogeneity in wire W10-B, probably due to the agglomeration of boron powders.

EDX analysis reveals the presence of similar oxygen content for W10 and W10-B, ∼5 at.%, which is slightly higher than for the undoped wire, ∼3.5 at.%. This may be due to the additional oxygen introduced by the carboxylic groups of the oleic acid. Then, the main difference between these wires is the homogeneity, which would be originated in the precursor powders. Oleic acid seems to agglomerate boron particles, adding difficulty to the subsequent homogeneous mixing with Mg powders. Nevertheless, this effect does not appear when oleic acid is added to the Mg + B mixture, as in W10.

The FESEM analysis of wire W5-n (not shown) displays phases different to the above-reported ones, and EDX analysis reveals the presence of regions with very high oxygen content (up to 15–20 at.%). Moreover, a considerable amount of carbon should be present as secondary phases, since just ∼5 at.% of carbon coming from oleic acid has substituted B in MgB₂. Nevertheless, since samples were coated with carbon for FESEM observations, it was not possible to quantify it with EDX. These carbonaceous phases will act as weak links between MgB₂ superconducting regions, which will be the reason for the large transition width observed in the magnetic susceptibility.

Thermogravimetric analysis has been used to analyse the effect of oleic acid addition on B, Mg, or B + Mg powders. The weight loss (%) and derivative weight—dW(T)—curves as a function of temperature are displayed in figure 4. A needed reference is the weight losses of liquid oleic acid, see figure 4(a), which has a single and smoothed step down, starting at 200 ° C, with maximum of the dW(T) curve at 330 ° C, due to the dragging off of its vapour by the Ar flow, and finishing around the atmospheric pressure boiling point (360 ° C).

Zoom In Zoom Out Reset image size

For B powders mixed with oleic acid in a proportion similar to the one used in W10-B wire precursors (see figure 4(b)), there is an initial weight loss with maximum of the dW(T) curve at 95 ° C. This is due to the elimination of the water produced by the chemical reaction of oleic acid carboxylic groups (–COOH) with boron oxides present in the B powders to form oleyl borates with water evolution. The dragging of loose bounded oleic acid molecules starts around the same temperatures as liquid oleic acid (∼200 ° C), and dW(T) has a large maximum at 260 ° C. This maximum occurs at lower temperatures than in figure 4(a), due to the small amount of oleic acid. The results of B + oleic are similar to those seen for oleic acid coatings in Co and Ni nanoparticles [26] analysed with TGA and NEXAFS techniques. Those authors also observed thermal desorption of oleic acid at around 240 ° C (i.e., weight loss in TGA and NEXAFS spectra corresponding to oleic acid), whereas at higher temperatures, between 360 and 430 ° C, they observed a weight loss, but also changes in the NEXAFS spectra, with disappearance of C to O double bonds, and only the peak of C=C double bonds in graphitic C remained. Consequently, the peak at 380 ° C in figure 4(b) should be associated with the thermal decomposition (pyrolysis) of oleic acid bounded to B grains producing a carbon-rich coating.

For Mg powders mixed with oleic acid samples (see figure 4(c)), two chemical reactions may take place associated with its carboxylic groups. At low temperatures, the acid may dissolve the MgO precipitates, usually placed at the Mg metallic surfaces, following the chemical reaction 2⋅C₁₇H₃₃–COOH + MgO → [C₁₇H₃₃–COO]₂Mg + H₂O, which is enhanced at higher temperatures. This effect is marked by a smooth weight loss due to dragging off of created water molecules, peaking at 144 ° C. At higher temperatures, less-bounded oleic acid molecules are dragged off by the Ar flow, but the weight loss rate is much slower than for B powders with oleic. That indicates that oleic acid also reacts with metallic Mg: 2⋅C₁₇H₃₃–COOH + Mg → [C₁₇H₃₃–COO]₂Mg + H₂. The Mg oleates produced in the above chemical reaction will start to decompose at slightly higher temperatures (pyrolysis), showing a marked maximum at 380 ° C. The pyrolysis is a complex reaction giving oleic acid fragmentation into short-chain hydrocarbons, CO, CO₂, and coke, causing fine MgO or MgCO₃ precipitates.

For B + Mg powders mixed with oleic acid, at temperatures below the MgB₂ formation, the weight losses, displayed in figure 4(d), almost correspond to the addition of the interaction effects observed on B + oleic and Mg + oleic mixtures. There is an enhancement of water losses by the combined effects of the oleic acid on the Mg and B oxides, with a peak at 120 ° C. The decomposition of Mg oleate and oleyl borates produces the peak at 390 ° C. At intermediate temperatures, less-bounded oleic acid molecules are dragged off by the Ar flow.

The inductive critical current densities of the analysed wires have been estimated magnetically from the hysteresis magnetization loop width, ΔM(H) as

$$\begin{eqnarray} &&{J}_{\mathrm{c}}=\frac{3 \pi }{8}\frac{\Delta M}{R}, \end{eqnarray} \tag{ 1 }$$

where R is the radius of the cylindrical superconducting core and the magnetic field is applied perpendicular to the wire axis. The local distortions of the core cylindrical shape produced during polishing are estimated to be less than 10%. The measurements were done with this orientation in order to have most current paths along the sample's axis, as in transport measurements. Equation (1) is obtained under the assumption of homogeneous induced currents on the length scale of the sample, which will be discussed later. It should be noted that generally it is not valid for fields lower than the penetration field, H_p [28]. However, we have plotted the J_c-values at low fields in order to show the presence of flux jumps.

Figure 5(a) shows a logarithmic representation of J_c as a function of the magnetic field at 5 K for all doped samples for two different annealing conditions: 700 ° C × 2 h and 670 ° C × 5 h. Wire W5-n has the lowest J_c-values, while W10 presents the best results, with slightly higher J_c for lower annealing temperatures and longer times. Wire W10-B has higher critical current densities for treatments at 670 ° C × 5 h than at 700 ° C × 2 h, in agreement with the differences observed on their microstructure.

Zoom In Zoom Out Reset image size

The comparison between the inductive J_c-values of the optimized doped wire W10 and the undoped one W0 at 5 and 20 K are shown in figure 5(b), together with the transport J_c measured at 4.2 K. There is a large improvement of J_c at medium and high fields at both temperatures for the doped wire W10. The measured transport values are higher than the magnetic ones, especially for the undoped wire W0. A remarkable parallelism between magnetic and transport values is found for W10, although the transport experimental facility limits only just enable one to establish the tendency. It should be noted that the J_c-values of wire W10 are similar to those obtained on Fe tapes fabricated with B powders treated with stearic acid [24].

The analysis of the differences between transport and magnetic J_c-values has been studied by several groups (see, for example, [29–31]). In general, the use of equation (1) is justified by the non-granular behaviour of MgB₂ [32]. However, wire inhomogeneities may lead to the appearance of granularity effects, which at high fields are usually indicated by strong downturns of magnetic J_c(H)-curves [29]. When granularity exists, the inductive J_c-values are lower than the transport ones in high fields [29], as in our case. Even so, it must be noted that the high transport critical current values indicate the existence of good connectivity between grains also in these high magnetic fields. In addition to the apparition of granularity, flux creep phenomena would intensify the differences between magnetic and transport values, due to the more restrictive criterion used to define J_c in magnetic measurements compared with the transport ones [31].

The measured transport I–V-curves (V ∝ Iⁿ) of W0 and W10 wires are steep. For example, at μ₀H = 8 T, the estimated n-values are n ∼ 20–24 and n ∼ 30–33 for W0 and W10, respectively. For this field, the transport J_c-values of the doped W10 wires are 4 times higher than for W0. The transport critical currents of wire W10-B are much lower, by approximately one order of magnitude, than their corresponding magnetic values (for example transport J_c = 1.3 × 10⁷ and 6 × 10⁷ A m⁻² at 8 T and 5 T, respectively, were measured at 4.2 K). Moreover, its I–V-curve is smooth and shows a resistive tail, in agreement with previous indications of non-homogeneous phase formation on this wire. Similar results are obtained for pre-annealing in vacuum or in Ar.

For the wire with best superconducting properties (W10), a preliminary analysis of the influence on J_c of the annealing conditions at the different steps of wire manufacturing has also been done. First, no relevant difference was observed for wires W10 manufactured with annealing at 400 ° C or at 550 ° C during drawing, although the former seems to get slightly higher J_c-values. Second, in order to obtain the optimum pre-annealing conditions of the precursor, we prepared a wire (with precursors: Mg + 2B + 10 wt% oleic acid), which was drawn up to 2.5 mm diameter without any heat treatment of the precursors or during drawing. Three pieces of 1 cm length were cut and heat treated at 400, 450, or 550 ° C, respectively, for 1 h in Ar atmosphere, and cooled down to room temperature. These pieces were chosen to be short enough to allow the vapours generated during the pre-annealing to get out easily. Afterwards, these three pieces were sealed and annealed at 700 ° C × 2 h in vacuum. The highest magnetic J_c was obtained for the first one. For this reason, the precursors of wire W10 were pre-annealed at 400 ° C × 1 h before wire manufacturing.