Synthesis of Bi₂Sr₂CaCu₂O_x superconductors via direct oxidation of metallic precursors

The starting metals are 0.074 mm bismuth powders, 0.5−1.5 μm copper powders, 19 mm strontium granules and 1 mm calcium granules, all with high purity (≥99.5%). For one powder batch, 0.5−1.0 μm silver powder (purity of 99.9%) is added as well. Before mixing, the Sr granules are charged within a Spex zirconia milling vial and ball-milled for 2 h using a Spex 8000 to reduce the particle size to 0.5 mm.

The handling of MPs requires particular caution because Ca and Sr are highly reactive with O₂ and water. Thus, all the powder handling requires a carbon-free, oxygen-free, dry environment. Here, the weighing, storing and milling of MPs are under an Ar atmosphere.

According to the phase diagram [35, 36], the central axes of the single-phase region point to a Bi stoichiometry of 2.18 in the temperature range of 820–870 °C in air. As the temperature decreases, the single-phase region becomes large. For the Bi2212 phase within the composition range Bi_2.18Sr_3−yCa_yCu₂O_8+d (y = 0.4−2.0), the single phase extends furthest when y = 1.10 and decreases slightly as y increases [26]. Here, the starting materials are mixed with a cation ratio of Bi:Sr:Ca:Cu of 2.18:1.90:1.10:2.00, a stoichiometry in the middle of the single-phase region. The endpoints of the Bi2212 solid solution region remain almost unchanged from 100% O₂ to 1.0% O₂. Therefore, this stoichiometry remains in the center of the single-phase region as the oxygen partial pressure varies from 100% to 1.0% [36].

Two batches of powders are milled with the same Bi:Sr:Ca:Cu ratio; in one batch 5 wt% Ag is added. After weighing and mixing all the powders in a high purity (<1 ppm oxygen) argon atmosphere in a glovebox, they are charged again within a Spex zirconia milling vial and mechanically alloyed for 20 h. After milling, the powders are stored in glass vials in an argon atmosphere. Table 1 lists the cation ratios of Bi, Sr, Ca, Cu and weight percentage of Ag as detected by chemical analysis after milling but before heat treatment. Batch I was milled continuously while batch II was milled with intermittent scraping. The cation ratios and Ag content deviate from the starting stoichiometry of Bi_2.18:Sr_1.90:Ca_1.10:Cu_2.00, which may be due to some elements sticking preferentially to the milling vial and balls; intermittent scraping of batch II was used to reduce this effect.

Pellets with a 6 mm diameter, 1.0–1.2 mm thickness and 0.18–0.20 g mass are made from the milled powders; three pellets are made from batch I (I-1–3) and six from batch II (II-1–6). A reference pellet is also made from oxide precursor (as received Nexans Bi2212 granules). All pellets are pressed using a uniaxial press at a pressure of up to 2 GPa and a density of up to 5.30 g cm⁻³. The milled metallic precursors are filled and sealed into a 6 mm die in a glovebox and then the die is taken out for pressing.

To avoid intense reactions between the metallic precursors and oxygen, and to ensure that the starting stoichiometry is in the single-phase region, a series of two-step heat treatments are carried out under 10% O₂ (Ar balance) atmosphere with different temperatures and times in a quartz tube furnace. Figure 1 shows a schematic drawing of a typical two-step heat treatment profile. For each heat treatment, the pellet sample is taken from the glovebox just before heat treatment and put on a silver foil in an alumina boat. The furnace is then sealed, vacuumed and refilled with pure Ar repeatedly to ensure that there is no oxygen in the quartz tube before heat treatment.

Zoom In Zoom Out Reset image size

Heat treatments include two dwelling steps, one at T_ox and the other at T_p. Step 1 is used to directly oxidize the metallic precursors at T_ox instead of T_p, avoiding melting as there are considerable amounts of Bi and Sr, with melting points of 271.4 and 768.8 °C, in the green pellets. Hence, early oxidation, before the temperature reaches T_p, is needed to prevent elemental diffusion within a liquid phase which might harm the final homogeneity of the pellet. Moreover, Bi2212 crystallization at 630 °C in air from glass oxide precursors has been reported [26]. Hence, two values of T_ox are studied, 600 and 650 °C. Compared to conventional PMP in 100% oxygen, the reduced oxygen partial pressure used here reduces the Bi2212 melting temperature to around 870 °C [37]. Thus, to avoid partial melting during heat treatment, T_p is studied in the range 800–855 °C.

After pumping, the sealed furnace is heated to T_ox at a rate of 300 °C h⁻¹ in flowing Ar and held for 10 min to allow the temperature to stabilize, after which a flow of 10% O₂ mixed gas is introduced. After holding at T_ox for a specific amount of time, t₁, the system is heated to T_p at a rate of R₁ and held for t_p. Finally, the system is cooled to room temperature at a rate of 100 °C h⁻¹ or furnace cooled (FC) (∼400 °C h⁻¹). Detailed heat treatment parameters are listed in table 2.

Inductive-coupled-plasma emission spectroscopy (ICP-ES) (Perkin Elmer 2000 Dual View spectrometer) is used to determine the stoichiometry of milled batches (table 1). Samples are weighed on Teflon microwave digest sample holders to the nearest 10⁻⁵ g. 5 ml of 4:1 ([conc. HCl]: [conc. HNO₃]) is then added to each sample. The solution then sits for 48 h.

A JEOL 10LA scanning electron microscope (SEM) with energy dispersive spectrometry (EDS) is used to examine the homogeneity of the elemental distribution of milled batches. Hitachi S3200 and TM3000 SEMs are used to investigate the microstructures of the heat-treated samples.

An FEI Titan 80-300 probe aberration corrected scanning transmission electron microscope (STEM) with SUPER X EDS is used to characterize the Bi2201 intergrowths of a heat-treated sample.

A Rigaku Smartlab x-ray diffractometer (XRD) is used for phase analysis. The measurements are carried out with a Cu Kα source over a 2θ range of 5°–75°, using 40 kV and 44 mA. To avoid oxidation of the metallic precursors during handling, a thin mineral oil film is coated on the pellet surface in the glovebox before removal for analysis. For heat-treated samples, the pellets are pulverized into fine powders before scanning.

The superconducting critical temperature, T_c, is measured in a SQUID magnetometer (Quantum Design MPMS-5S) via magnetization versus temperature measurements. The samples are cooled to 4.2 K in zero field and measured in a 100 Oe applied field during warming.

Differential thermal analysis (DTA) (Perkin Elmer STA 6000) is used to study the melting behavior of the heat-treated samples to investigate the phase purity. All the tested samples are 40 ± 0.002 mg and are heated to 980 °C at 10 K min⁻¹ in flowing 10% O₂ (20 ml min⁻¹).

Figure 2 shows an SEM micrograph and corresponding EDS maps of the elemental distributions of Bi, Sr, Ca and Cu in batch I before heat treatment. A homogeneous distribution of elements is observed after milling. Similar homogeneity is observed in batch II. Figure 3 shows the XRD results for both batches after milling but before heat treatment. Note that in addition to elemental metals (Bi, Cu and Sr), some alloys formed during milling (Ca–Cu and/or Bi–Sr).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Figures 4 and 5 show the magnetic properties of heat-treated samples measured by SQUID. Figure 4 shows the absolute magnetic moment versus temperature for samples from batches I and II. All heat-treated samples show a superconducting transition in the range of 69–81 K, and only sample I-1 has a lower magnetic moment at 4.2 K than Nexans granules. Figure 5 shows the same data normalized by reference values at 4.2 K. Only sample I-1 exhibits a shallower transition than Nexans granules. The T_cs of samples I-1–3 (figure 5(a)) are 69.0 K, 77.5 K and 79.0 K, respectively, and the T_cs of samples II-1–4 are 79.0 K, 79.5 K, 80.5 K and 81.0 K, respectively (figure 5(b)). The Nexans granule sample has a T_c of 67.5 K.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Figure 6 shows the XRD patterns for three heat-treated samples, I-3, II-1 and II-4; the major peaks of each phase are indexed. XRD results (not shown) from batch I indicate that all samples contain Bi2212, Bi2201 and Cu-free phase. In contrast, for batch II, only sample II-1 contains Cu-free phase whereas the others contain only Bi2201 and Bi2212. Thus, the XRD patterns in figure 6 are indicative of all samples. Figure 7 displays enlarged sections for 2θ between 21° and 24°, showing the relative intensity of the 006 (Bi2201) and 008 (Bi2212) peaks for different samples from batch II.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

DTA curves of different heat-treated samples and Nexans granules are shown in figure 8. Additional peaks around the main melting peak are indicated by arrows. The melting onsets of samples II-1–6 are all 876 ± 1 °C. Figure 8(a) compares samples II-1–4 to show the effects of T_p and t_p while (b) compares samples II-3–6 to show the effect of R₁. The peak width for the Nexans granules is much wider than that for the converted metallic precursor samples. In (a), for samples II-1–3 and Nexans granules, there are additional peak(s) around the main melting peak, suggesting multiple melting events. There is no observable additional peak for sample II-4. In (b) there are also additional peaks around the main melting peak for samples with a low heating rate, i.e. samples II-5 and II-6.

Zoom In Zoom Out Reset image size

Figure 9 shows top surface SEM images of samples II-1–4. By increasing T_p from 845 to 855 °C, the grain size increases from approximately 10−15 μm to more than 20−25 μm. By increasing t_p from 16 ((a) and (c)) to 48 h ((b) and (d)), the grain size increases, but the effect is less dramatic than temperature. Figure 10 shows cross-sectional SEM images of polished samples II-3 and II-4. Closely stacked Bi2212 grains with an average size of 25 μm are formed when T_p reaches 855 °C. Porosity exists between the aligned grains in sample II-3 and is reduced in sample II-4. Figure 11 shows the microstructure at the pellet/Ag interface of samples II-1 and II-2. Comparing the images from the tops of these samples with those at the Ag interface, the Bi2212 grains at the Ag interface are much larger than those at the top and are well connected.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Figure 12 shows STEM images of sample II-4. Bi2212 grains with Bi2201 half-cell intergrowths are seen. With increasing magnification, it is clear that each bright line in (a) and (b) is a half unit cell of Bi2201 with a 1.2 nm spacing. In (c), one Bi2212 unit cell with a 3.1 nm spacing is seen between the two Bi2201 half-cells. EDS mapping confirms that the two bright lines in the STEM images are Ca-deficient. No large Bi2201 grains are observed in this sample.

Zoom In Zoom Out Reset image size

Table 3 shows the XRD intensity ratio γ ($\gamma =\frac {I_{008}}{I_{00\underline {10}}}\times 100\%$) of the 008 and $00\underline {10}$ peaks of the Bi2212 phase for six samples. The γ values are used to calculate the fraction of Bi2201 intergrowths in the Bi2212 grains (f) based on the model developed by Rikel et al [38, 39]; these are also listed in table 3.

The XRD results for the milled precursors in figure 3 show a combination of elemental metals (Bi and Cu) and metal alloys. A Ca–Cu alloy is common to both batches, whereas batch II also shows a Sr–Bi alloy which might be a result of the intermittent scraping giving a more homogeneous mixing of powders. The chemical analysis in table 1 shows that the result for the cation ratios in batch II is closer to the nominal stoichiometry of the starting powders, indicating that Sr sticks preferentially to the mixing media and that the intermittent scraping is beneficial.

The metal alloys formed during milling may be a source of impurities upon oxidation, such as copper-free phases in samples I-1–4 and II-1, as shown in figure 6. As there is no detectable copper-free phase or other non-superconducting phases in samples II-2–6, any alloys formed during milling are reactive and not stable during heat treatment.

Magnetic measurements of the three batch I samples indicate that a higher peak temperature yields a larger magnetic response, indicative of enhanced Bi2212 formation. T_p is either 800 or 830 °C, which is 80 or 50 °C below the Bi2212 melting temperature in 10% O₂. The reactions of oxidized powders were not complete and a considerable amount of Cu-free phase and Bi2201 phase was found in heat-treated samples. This is probably due to the very low diffusion rate of cations and the high activation energy required to form Bi2212 via solid-state reactions [40, 41]. The activation energy of the transformation from Bi2201 to Bi2212 in the solid state, without the aid of liquid, is 3.0 to 6.7 times larger than with partial-melt processing. Kinetic studies of the Bi–Sr–Ca–Cu–O phases also imply a diffusion controlled transformation [41]. Thus, the non-superconducting phases that formed in the early stages of the heat treatments are not fully converted to Bi2212 by diffusion at low T_p. With T_p well below the melt temperature, the phase transformation is limited. Note that increased T_p also increases T_c, consistent with results reported by others [26, 42]. The transition width ΔT_c of sample I-1 is 5–6 K while those for samples I-2 and I-3 are within 2 K, indicating increased oxygen homogeneity in samples I-2 and I-3 [43]. This also confirms the more efficient diffusion among grains due to the increase in T_p.

Based on the batch I results, batch II studies focused on higher T_p (845 and 855 °C) and longer t_p (16 and 48 h). Samples II-1–4 were heated from 650 °C to T_p at 200 °C h⁻¹. Magnetization measurements show that the samples with higher T_p (II-3 and II-4) have approximately 35% larger magnetization than those with lower T_p (II-1 and II-2). Furthermore, the magnetic results show that longer t_p results in an 8%–12% decrease in magnetic moment at 4.2 K. The T_c results indicate that for T_ps of 845 and 855 °C, the influence of increasing T_p on T_c is small. Also, the T_cs of samples II-1–4 are all close to 80 K, with oxygen content × close to 8.0 [44]. The transition widths ΔT_c of these four samples are within 1 K, indicating a more homogeneous distribution of oxygen content with increased T_p [43].

Figure 7 shows the relative intensity of the 006 (Bi2201) and 008 (Bi2212) peaks as a function of T_p and t_p. With increasing T_p and t_p, the 006 (Bi2201) peak decreases and the 008 (Bi2212) peak increases. The XRD results show that for batch II, only sample II-1, which was heat-treated with lower T_p = 845 °C and shorter t_p = 16 h, has Cu-free phases, while the others have only Bi2201 and Bi2212 phases. There is no obvious peak for the ordered 4413 phase (an intermediate phase of mixed Bi2212 and Bi2201) as reported by others [39, 45, 46].

Rikel et al developed a model that correlates the Bi2201 intergrowth fraction in Bi2212 with the intensity ratio γ ($\gamma =\frac {I_{008}}{I_{00\underline {10}}} \times 100\%$) of the 008 and $00\underline {10}$ peaks of the Bi2212 phase [38, 39]. According to this model, for a Bi2212 single crystal that contains a fraction f of randomly distributed Bi2201 intergrowths, the average value 〈f(1 − f)〉 is inversely related to the intensity ratio γ. If f < 0.5, the average value 〈f(1 − f)〉 increases monotonically with increasing f. Thus, for Bi2201 intergrowth f < 50%, less Bi2201 intergrowth content should result in larger γ. Intensity ratio γ and f-values for samples II-1–6 are listed in table 3. For samples II-1–4, γ ranges from 53–73%, for which, according to the correlation map in [39], the Bi2201 intergrowth fraction f ranges from 5% for II-4 to 11% for II-2. These values are similar to what Heeb et al showed previously for a sample annealed at 850 °C for more than 80 h [47]; however, here the annealing was for only 48 h. Therefore, with further increase in t_p, the fraction of Bi2201 may be reduced further [48]. Furthermore, the f-values here are much less than the average values of a PMP Bi2212/Ag conductor, which range from 6.4% to 35%, explaining the absence of any observable 4413 phase peak. In addition, only a low content of Bi2201 half-cell intergrowths is observed in sample II-4 under STEM, indicating that large Bi2201 grains either do not form or are converted to Bi2212 with higher T_p and longer t_p.

The SEM images in figure 9 show that by increasing T_p from 845 to 855 °C, the grain size increases from 10−15 μm to more than 20−25 μm, a result of increased growth rate at elevated temperature. Closed-stacked Bi2212 grains are found when T_p reaches 855 °C (figures 9(c) and (d)). Cross-sectional images for sample II-3 and 4 give a clearer view of layered Bi2212 grains and flakes. With a short t_p, porosity is favored between the Bi2212 flakes, while with a longer t_p, a more dense structure is formed.

Samples II-5 and II-6 share the same T_ox = 650 °C, T_p = 855 °C, t_p and cooling rate (furnace cooled) as samples II-3 and II-4, but the heating rate R₁ is decreased to from 200 to 60 °C h⁻¹. Magnetization results show that by decreasing R₁, there is an ∼23% decrease in magnetic moment at 4.2 K. Again, there is no effect on T_c, indicating that the oxygen content is mostly influenced by T_p. The XRD results in figure 7(b) show that with the lower R₁, the 008 (Bi2212) peak decreases, indicating that more Bi2201 is formed. The γ values calculated for samples II-5 and II-6 are 49% and 50%, respectively, suggesting around 13–14% of Bi2201, significantly higher than samples II-3 and II-4. Furthermore, the DTA curves in figure 8(b) for samples II-5 and II-6 show additional peaks around the main melting peak, indicating reduced Bi2212 content when processed with lower R₁.

Although both Bi2201 and Bi2212 form in the temperature range 650–790 °C [37], Bi2201 is more stable than Bi2212 and is thus preferred. Additionally, the stability range of Bi2201 is only slightly reduced by decreasing the oxygen partial pressure, but is further reduced by increasing the temperature [45]. Hence, with a lower heating rate, the time for heating is increased from 1.0 to 3.4 h, allowing more time for Bi2201 formation and growth.

Figure 11 shows that the Bi2212 grains at the Ag interface are much larger than those at the sample's free surface. These grains are well connected, which is commonly observed in Ag-clad Bi2212 conductors after PMP [18, 49, 50]. In PMP Bi2212/Ag tapes, nucleation and grain alignment begin at the oxide/Ag interface [51]. Also, based on a grain alignment model developed by Buhl et al, Bi2212 grains bend during growth when they reach a Ag surface and turn parallel to the Ag surface [52].

Here, rapid diffusion at the pellet/Ag interface is the essential factor influencing Bi2212 formation and growth. The peak temperature is 845 °C for samples II-1 and II-2, which is nearly 30 °C below the melting temperature (876 ± 1 °C as determined by DTA), so the bulk does not melt during the heat treatment. A 5.2 wt% silver addition to Bi2212 powders reduces the melt temperature by more than 20 °C [53]. Thus, in the region influenced by the Ag interface, the melting temperature of Bi2212 is at least 20 °C lower than in the bulk, ensuring rapid diffusion. Because the growth in the a−b plane direction is much faster than in the c direction, very large platelet Bi2212 grains form at the interface. When dwelling at T_p, Bi2212 grains grow rapidly, reach the Ag interface, and then turn parallel to the Ag surface [52]. The smoothness of the Ag surface also influences Bi2212 grain orientation [54]. For the remainder of the pellet (which is not proximate to the Ag), the Bi2212 grains form and grow in colonies but without any alignment relative to the Ag sheath. With the layer thickness greater than the grain size (about 150 μm), the colonies of Bi2212 grains are no longer parallel to the Ag substrate [55]. Also, the rest of the pellet is still nearly 30 °C lower than its melting temperature, so grain growth via diffusion is limited in the solid state.

The results show several potential advantages of MP powders with a direct oxidation heat treatment over conventional OPIT with PMP. Several potential challenges must also be considered for wire processing.

Through mechanical alloying, MP powder can have a controllable stoichiometry and homogeneous elemental distribution. Composition is a potentially significant factor in future Bi2212 wire optimization. With sufficient intermittent scraping and optimized milling time, an elementally homogeneous MP powder with the desired stoichiometry can be obtained, avoiding phase segregation and inhomogeneous oxidation after heat treatment.

The results here show that by using a direct oxidation approach to forming Bi2212 grains without a partial melt, Bi2201 is the only phase impurity formed, and the Bi2201 content is much lower than in tapes and wires. Recent results indicate that although large Bi2201 grains reduce J_c, Bi2201 intergrowths within Bi2212 grains do not, and in fact may provide flux pinning [19, 56]. Through direct oxidation of MPs, any CF or AEC grains that may form in the early stage of heat treatment are fully reacted, whereas this is not the case with PMP of oxide precursors. The solid-state reaction converting MPs →Bi2212 also limits void agglomeration, a significant limiting factor for transport J_c in OPIT wire [57, 58].

The results here show that high density Bi2212 colonies are formed during the MP →Bi2212 transformation and that large, textured grains are aligned at the Ag interface. For multifilamentary wires with filament sizes between 10 and 30 μm, highly aligned Bi2212 grains are expected after heat treatment.

One important issue for MP wires that has not been addressed here is the need for sufficient and rapid oxygen diffusion through the Ag sheath to ensure homogeneous oxidation and to avoid phase segregation. In the work reported here, the MP pellets have a free surface exposed to the flowing oxygen, so even with a 10% oxygen atmosphere, access to oxygen is sufficient. In wires, the time constant for oxygen diffusion through the Ag sheath must be compared to those of solid-state metallic reactions and/or phase segregation within the precursor. This will require significant study with Ag-sheathed MP wires.