Long-length YBCO coated conductors for ultra-high field applications: gaining engineering current density via pulsed laser deposition/alternating beam-assisted deposition route

In-field performance of DD-YBCO-based tapes with different substrate thicknesses is shown in figure 2, which plots the critical current versus flux density for a set of samples (end pieces of long tapes). The areas viewed in figure 2 as high-field (HF) and ultra-high-field (UHF) tapes are defined via a simple condition, I_c,UHF(B) > 2020·B^−0.73 , that allows to distinguish the DD-YBCO tapes of HF or UHF class. Such definition is possible because the I_c(B)-dependence is well defined by α-law [1] in the wide field range. The tapes with lower I_c correspond to HF tapes because even with moderate current level they are able to carry sufficiently high currents (>300 A) in 4–10 T -range. UHF tapes must transport high currents in 10–31 T range. Thus, their I_c level in wide field range must be higher.

Zoom In Zoom Out Reset image size

There is a scattering of I_c-s, which, nevertheless is not degrading in tapes with thinner substrates. In general, the overall scattering of in-field I_c per cm-width, as shown in figure 3, for routinely produced tapes indicates technological progress regarding increase of reproducibility and yield.

Zoom In Zoom Out Reset image size

In the 4 mm wide tape, Q021-18N1, based on a 50 μm thick substrate, the highest I_c value observed was 454 A at 18 T, B//c, and 4.2 K (the red star in figure 2). Considering the Cu plating of 2 × 20 μm, the total tape thickness is 90 μm, which is 50 μm thinner than that of our thick substrate tapes. This results in a champion level J_e = 1261 A mm⁻² (that nevertheless should not be interpreted as a standard tape quality, more a level that is potentially achievable). However, we can expect an increase in J_e because of the reduction in the tape thickness. The thickness reduction factor of 1.56 (140 μm/90 μm) leads to an increase of J_e from 800 to 1244 A mm⁻², where 800 A mm⁻² represents a high level J_e for a 100 μm substrate of HTS-coated tapes [3]. This is in good agreement with the observed value of 1261 A mm⁻² and indicates the potential to gain even higher J_e values because the J_e of the tapes with 100 μm thick substrates may exceed the 800 A mm⁻² [3] mentioned earlier.

Figures 4(a) and (b) show the reduction of I_c in the course of tape bending. A new device with a variable bending radius enabled measurements without dismounting the tape sample. The critical bending radius for the 50 μm case corresponded to R_cr = 4.5 mm. This value is half that of R_cr = 8.9 mm observed at 100 μm (see figure 4(b)).

Zoom In Zoom Out Reset image size

A characteristic feature of the ABAD–YSZ buffered tapes is a bowing that originates from compressive stress occurring mainly during the growth of the buffer layer. This is a well-known effect caused by atomic peening [9] occurring in the course of the sputter-based ABAD employed in our processing route. Furthermore, a significant degree of bowing is retained by the tapes after deposition of all layers and intermediate annealing. This bowing effect can be seen in figure 4(c), where a cross-section of a 12 mm wide tape is depicted (left tape). The bowing height may exceed several millimeters depending on the thicknesses of the buffer layer and substrate. Such tapes show inhomogeneity of the in-plane texture, which results in an inhomogeneous I_c over the tape length. This can be seen in figure 2 for samples Q020-N1 and Q020-N2 which represent different parts of the same tape. The difference in I_c at 20 T corresponds to ∼6%. In some cases, this difference may exceed 10%.

We found that that this effect can be suppressed by the adjustment of the deposition parameters and introduction of additional layers. As a result, the tape has a significantly lower bowing height as is shown in the right-side tape in figure 4(c). Nevertheless, some residual bowing should be considered during tape processing, especially for thin substrates with a width of 12 mm.

Any technology based on reel-to-reel tape translation [10, 11] may expect intermittent longitudinal defects, e.g. scratches, especially in bowed tapes. In this work, we simulated such defects by scratching the finished tape (without the Cu layer) by using a hard material tool.

We measured the impact of an artificially introduced, 2 cm long surface scratch on the critical current at 77 K. The total dimension of the sample was 12 cm × 0.4 cm; the voltage contacts were separated by a distance of 3 cm and the initial critical current was 38.5 A (curve 1 in figure 5).

Zoom In Zoom Out Reset image size

We used the criterion of 1 μV cm⁻¹ voltage drop between contacts (i.e. 3 μV at 3 cm distance) to define I_c. With a scratch, the critical current (curve 2) degraded to 33.6 A, i.e. by ∼13%. This degree of deterioration also seems to be typical of the slitting of the manufactured tapes.

We compared I_c degradation due to the scratch in a field produced in the gap between 2 disc-like permanent magnets creating a flux density B ∼ 0.4 T. In this particular field applied perpendicular to the tape, I_c is reduced to 12.4 A, i.e. by a factor of ∼3 compared to the initial tape. In scratched tape, I_c reduces to 11.8 A in the field, yielding a difference of 5%. In contrast to the 13% reduction, this minor influence of a scratch at 0.4 T seems surprising. Here, we must consider the 'zooming' effect, which is pronounced in the local magnetic field. Owing to the drastic suppression of the in-field critical current, voltage drop U_tot in the measured tape is mainly determined by the in-field tape area. This follows from the n-law, which in the considered case of the coordinate-dependent field, may be generalized as

$$\begin{eqnarray}&&{U}_{{\rm{t}}{\rm{o}}{\rm{t}}}=\displaystyle \frac{{U}_{cr}}{L}\displaystyle {\int }_{0}^{L}{\left[\displaystyle \frac{I}{{I}_{c}\left(B\left(x\right)\right)}\right]}^{n\left(B\left(x\right)\right)}dx,\end{eqnarray} \tag{ 1 }$$

where the in-field area with the lowest I_c-s determines the fraction of the integrand. Here, U_cr is a criterion used for defining I_c, L is the distance between voltage probe contacts, B(x) is a linear distribution of flux density in the longitudinal direction of the tape, and n(B(x)) is a coordinate-dependent power value used instead of n-values employed in the constant field case. The approximation via (1) (figure 5(a), curve 4a) sufficiently agrees with that obtained through the experiment, similar to the approximation suggested in [12]. However, the approximation predicts a 7% reduction of I_c instead of the observed 5%. This suggests that further features of tape behavior within field gradients should be considered. Nevertheless, the dominating influence of defects remaining in the field was confirmed.

In [3], we showed that the distribution of the deposited material that originates from the stationary (i.e. fixed) laser plume is extremely wide, considerably exceeding the diameter of the laser plume. Figure 6 shows the fragment of the interferogram (with fringes of equal thickness) of the pulsed-laser-deposited DD-YBCO on the Al-foil substrate attached to the PLD drum; the deposition was at 25 °C–35 °C. Despite the drum curvature (100 mm⁻¹), fringes of equal thickness on the flattened foil are circular (not elliptical as expected). The layer-thickness distribution derived from the interferogram is depicted in figures 7 and 8. The 2D distribution exhibits a Gaussian-like shape with an eroded peak caused by self-etching [3]. The radius of the distribution is rather large. For deposition from a moderately sized plume where the radius exceeds 50 mm and for a large sized plume, visible fringes indicating a layer thicknesses of 100 nm were observed at a radius of >100 mm. The shape of the distribution can be explained by the lateral flows (LFL) and is essentially axial-symmetric despite the drum curvature, confirming that LFL follows the drum surface, similar to elastic flow for Mach number M > 5 [13–15]. This similarity remains valid in the case that LFL is bypassing an obstacle on the surface: we observed thickness distributions that indicate characteristic 'jumps' of the ultrasonic flow. A rough estimation of the speed of the initial flow yielded ∼5 km s⁻¹ (for a target-to-substrate distance of ∼5 cm and plasma propagation time of <10 μs). Thus, M ∼ 15 implies that the process is within the high hypersonic range.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Deposition in plume-scan mode at standard deposition temperature confirmed this wide distribution remains under actual process conditions [3]. In figure 8, we plotted this distribution together with the critical currents measured at thickness slope area (i.e. over −20 mm < Z < 30 mm): they exhibit relatively good agreement.

Zoom In Zoom Out Reset image size

The observed wide distribution of layer thickness implies that nucleation of DD-YBCO starts far away from the scan area of the laser plume. Considering that film nucleation starts at the nanometer thickness range, the distance from the center of the laser plume may be considerably higher than the radii shown in figures 6 and 7. This offers a good opportunity to independently control layer nucleation via creating different conditions (e.g. temperature and oxygen pressure) in the nucleation zone. As such, the DD-nanostructure and as a result the J_e performance may be further improved.

The substrate temperature varies rapidly for any PLD process [16]. By using a drum-based tape support with fast drum rotation in the tubular heating zone, temperature oscillations are basically suppressed but not completely eliminated [10]. Especially, the amplitude of residual temperature oscillations is much more pronounced in the case of thin substrates (50 μm), where this amplitude may exceed 5 °C with a transient time of tens of milliseconds. This seems to be tolerable for PLD processing, as HTS layers with the highest J_e were prepared in this way.

Nevertheless, the process window for substrate temperature is not very wide, possibly less than 10 °C, especially in the case of DD-YBCO.

Recently, we found that there is another, even more powerful source for temperature perturbation. This source is the energy release of ions impinging upon the growth surface of the DD-YBCO layer. The key factor that motivated us to analyze this case was that the thermal conductivity of the YBCO layer is extremely poor in the c direction, i.e. perpendicular to the growth direction.

Owing to the short lifetime of the laser plume, the integrated incident ion energy of ∼0.05 J cm⁻² was delivered to the interface within 5–10 μs. Conditions for thermal accumulation and release for the given case are schematically shown in figure 9. Owing to the low thermal conductivity of ∼1 W Km⁻¹ at T > 300 K [17, 18], YBCO represents a 'bottleneck' for heat release in the normal direction to the substrate that, in this case, together with the buffer layer plays the role of a heat reservoir. The alternative route for heat release via IR emission, which increases with increasing temperature, was shown to be negligible in comparison with heat conduction.

Zoom In Zoom Out Reset image size

The heating–cooling kinetics evaluated for a heating pulse with a 10 μs duration are shown in figure 10 for an arbitrary offset temperature of 781 °C, used here as an example. We assumed that energy flow originating from the laser plume is constant during the 10 μs pulse (see curve 1 in figure 10). Power delivered to the growth interface during the pulse was estimated as 4.5 kW cm⁻². The specific heat capacity of YBa₂Cu₃O_7−δ was assumed at 0.8424 J g⁻¹ K⁻¹ and the layer density was 5.985 g cm⁻³ [19].

Zoom In Zoom Out Reset image size

At the beginning of YBCO growth (see curve 2 in figure 10), when the layer thickness is 0.5 μm, the instantaneous temperature at the YBCO surface increased quickly by ∼22 °C and decreased rapidly after the heat pulse ended. This could have been a result of two factors: increasing heat capacity of the layer and increasing heat resistance with growing thickness.

At an intermediate thickness of 0.75 μm, the maximal increase of the instantaneous temperature exceeded 33 °C. The kinetics of temperature saturation and temperature release become considerably slower (curve 3 in figure 10).

At the end of growth when the layer thickness is about 1.5 μm, the increase of temperature becomes more linear and does not reach saturation (curve 4 in figure 10). After the pulse a temperature increase above 42 °C occurs, and the YBCO temperature decays much more slowly, taking tens of μs, because (i) much more energy is stored in a thicker layer, and (ii) of increased heat resistance introduced by the HTS layer.

Another possible reason for the instability of the instantaneous temperature of the YBCO interface is the impact of energy released during material condensation caused by the deposition pulse. Because of the short (<10⁻⁵ s) migration time of adatoms on the surface of the growing layer [20, 21] a rapid condensation should take place. Calculation of the energy released via condensation enthalpy is shown in figure 11 that similarly to figure 10 evaluates the kinetics of instantaneous temperature assuming a 10 μs long PLD 'condensation' pulse.

Zoom In Zoom Out Reset image size

The vaporization enthalpy of the DD-YBCO composition was evaluated and employed in modeling of the condensation process in which the energy should be released. The energy was evaluated as about 5% of the energy transferred by ions. However, the pulse temperature was significantly lower, determined by only 2% of the energy triggered pulse, due to LFL which distributes the energy over a large substrate area. Because of the pulsed nature of PLD, energy and temperature effects occur as a sequence of short pulses [3, 22]. Temperature pulses originating from condensation enthalpy should be added to the temperature pulses caused by 'condensation' of ion energy.

Thus, the evaluated pulses of the instantaneous temperature are shown to be rather high, exceeding a tolerance for deposition temperature by a factor of 4 at least. Because the temperature pulses are dependent on layer thickness, one may employ a compensation of the perturbation effect via decrease of the background substrate temperature during deposition process. This may be employed as a tool for improvement of DD-YBCO growth kinetics and, finally, increase of critical current density.

The DD-layer growth employing the PLD with LFLs has already been used in the processing of ultra-high field tapes, which yield a very high engineering current density at a high magnetic field (see figure 2). Nevertheless, this process, in combination with compensation of the instantaneous temperature of YBCO surface, has not yet been sufficiently developed.

The first steps in this direction resulted in the improvement of the critical current density J_c by at least 10% (at 4.2 K, 5 T, B//c). As shown in table 1 representing two tapes, tape Q047C is deposited without variation in deposition temperature T_dep, and tape Q068V is deposited under 17 K higher nucleation temperature T_n. In the case of tape Q068V, the nucleation occurred inside a preheating area of the tubular heater used in PLD [3]. The material for this nucleation (which corresponds to ∼10% of the layer thickness) was delivered by the lateral flows (LFLs). The improvement of J_c and I_c in this case is reproducible enough to confirm the presence of the gaining effect. This can be observed despite insufficient amplitude of temperature variation (17 K instead of 35 K). Moreover, the temperature during the major part of layer deposition remained almost constant, while a further variation was needed according to new results (figure 10). Following the 'sensitivity' of I_c to the deposition temperature, a 20%–30% I_c improvement should be reachable when the foreseen temperature variation is introduced during DD-YBCO growth. In our drum technology, with a one-directional path of the tape through the deposition zone, the implementation of such a variation could be possible in future studies.