Bulk-like ferroelectricity and magnetoelectric response of low-temperature solution-processed BiFeO₃–PbTiO₃ films on Ni for metallic MEMS Open Access

Metallic microelectromechanical systems (MEMS) integrating ferroelectric films are attracting increasing attention, especially for applications that involve large dynamic strain/stress levels, such as vibration energy harvesting, because they provide a better mechanical performance than stiff and brittle silicon-based ones.^1,2 In addition, added functionalities can be obtained if an active metal substrate is selected.³ A ferroelectric film deposited on a magnetic substrate is inherently a multiferroic composite, capable of providing magnetoelectricity as a product property of the piezoelectricity and magnetostriction of the ferroic components through their elastic coupling.⁴ Indeed, this specific configuration has been proposed to avoid substrate clamping effects that appear when bulk technologies are evolved into thin films,⁵ known to limit the magnetoelectric response of multilayer structures deposited on inactive silicon substrates.^6,7 Magnetoelectric composites are being extensively investigated⁸ and are under consideration for a range of applications, such as magnetic field sensing⁹ or energy harvesting.¹⁰ 

Among the magnetic base metals, nickel stands out for its magnetostriction and commercial availability as a range of polycrystalline plates and foils with variable geometry. Magnetoelectric cantilevers consisting of Ni plates bonded to high-sensitivity piezoelectric ceramic or fiber composite elements have been shown to allow simultaneous scavenging of vibrations and magnetic fields¹¹ and are suitable for powering electronics from stray magnetic fields.¹² The fabrication of a magnetoelectric metallic MEMS to realize an analogous but miniaturized harvesting device requires solving the deposition of the ferroelectric film on Ni. The problem is the thermodynamic incompatibility between the metal and state-of-the-art high-sensitivity-piezoelectric perovskite oxides, such as morphotropic phase boundary (MPB) Pb(Zr,Ti)O₃ (PZT) or Pb(Mg_1/3Nb_2/3)O₃–PbTiO₃ (PMN-PT). Ellingham diagrams show that there is not a processing window available regarding temperature and partial oxygen pressure where Ni metal would not be oxidized and PbO would not be reduced. Therefore, Ni⁰ and Pb-containing ferroelectric oxides cannot coexist in equilibrium at any temperature or oxygen partial pressure, so the simultaneous formation of NiO_x and reduction of the perovskite oxide is unavoidable with time.¹³ This leads to the unpreventable formation of in-series NiO_x interfaces between Ni substrates and lead-containing ferroelectric perovskite films. Such interfaces would play the role of a low-dielectric-permittivity layer in series with the ferroelectric film, hindering its poling. It might also potentially be a source of strain relaxation and a barrier to the transmission of the magnetostrictive deformation to the piezoelectric layer, ruining magnetoelectric performance. Actually, the presence of uncontrolled inactive layers between the piezoelectric and magnetostrictive components is known to strongly degrade the magnetoelectric response of a composite.¹⁴ This is why previous studies explored the introduction of barrier coatings on Ni substrates to protect them during the subsequent deposition of these ferroelectric oxides, which may also play additional roles as buffers to reduce strain levels and/or promote specific, favorable crystallographic orientations of the ferroelectric perovskite.^15,16

We have alternatively investigated the possibility of kinetically limiting the formation of nickel oxide by taking advantage of recent advances in low temperature solution processing of ferroelectric thin films.¹⁷ BiFeO₃–PbTiO₃ was selected instead of PZT for the study. To the knowledge of the authors, these piezoelectric BiFeO₃–PbTiO₃ perovskite films have not been prepared before on Ni substrates, despite their strong potential for their use in MEMS. As in the case of PZT, this system also presents an enhancement of the dielectric and piezoelectric crystal coefficients in a MPB between ferroelectric rhombohedral and tetragonal polymorphs, but conversely to PZT, BiFeO₃–PbTiO₃ also shows strongly reduced ferroelectric/ferroelastic domain wall activity and a high Curie temperature above 600 °C.^18,19 As a result, comparatively much more stable functional responses are anticipated in the BiFeO₃–PbTiO₃ system than in the PZT one, even when the microstructure is refined across the submicron range down to the verge of the nanoscale, where ferroelectricity is expected to be canceled for very small nanosize-grained materials. This is a general characteristic of polycrystalline thin films, where grain growth would be additionally hindered if low temperature annealings were used for their processing.^17,20 Therefore, this feature of BiFeO₃–PbTiO₃ is very appealing for the research carried out in this work, which involves a challenge for attaining ferro-piezoelectric BiFeO₃–PbTiO₃ films directly deposited on Ni by solution deposition methods with appropriate magnetoelectric responses.

A sol-gel procedure previously demonstrated for obtaining BiFeO₃–PbTiO₃ on Si based substrates has been used here.²¹ Precursor solutions with nominal compositions of 0.65BiFeO₃–0.35PbTiO₃, close to the MPB, were synthesized. For that, bismuth nitrate [Bi(NO₃)₃ · 5H₂O, 99.99%; Sigma-Aldrich], iron nitrate [Fe(NO₃)₃ · 9H₂O, 99.99%; Sigma-Aldrich], lead acetate trihydrate [Pb(CH₃COO)₂ · 3H₂O, 99.99%) (Sigma-Aldrich)], and titanium di-isopropoxide bisacetylacetonate [CH₃)₂CHO]₂Ti(C₅H₇O₂)₂ (Aldrich)] were used as chemical reagents. 1,3-Propanediol [OH(CH₂)₃OH (diol)] was used as a solvent and N-methyldiethanolamine [CH₃N[CH₂CH₂OH]₂ (Aldrich, 99%) (mdea)] as a metal complexing agent for Bi(III) and Fe(III). Two different solutions of the former cations were prepared by dissolving bismuth nitrate and iron nitrate in a mixture of diol and mdea, using molar ratios of metal:mdea:diol = 1:5:10, further heating at 150 °C for 1 h, and refluxing at 190 °C for 5 min. The obtained solutions were mixed in molar ratios of Bi(III):Fe(III) = 1.05:1.00 and stirred for 30 min to get the BiFeO₃ precursor solution, containing a 5 mol. % excess of Bi. This solution was diluted in dried ethanol [C₂H₅OH (Merck)] to an equivalent BiFeO₃ perovskite concentration of 0.25 mol/l. For the preparation of the PbTiO₃ precursor solution, lead acetate trihydrate was refluxed in diol for 1 h, and then titanium di-isopropoxide bisacetylacetonate was added with molar ratios of 1:1:5 of Pb:Ti:diol. The mixture was refluxed in the air for 8 h. The resulting solution was diluted in dried ethanol to obtain an equivalent PbTiO₃ perovskite concentration of 0.25 mol/l. The BiFeO₃ and PbTiO₃ solutions were mixed and diluted in dried ethanol to obtain a precursor solution of 0.65BiFeO₃–0.35PbTiO₃ with an equivalent concentration of 0.1 mol/l. The solution was spin-coated at 2000 rpm for 45 s on Ni metal plates (Goodfellow 1.6 mm thick Ni metal sheets cut to 15 × 15 mm², mirror polished on both sides). The wet layers were dried at 100 °C for 5 min and subsequently pyrolyzed at 250 °C for 5 min. Finally, crystallization was accomplished at 500 °C for 10 min by Rapid Thermal Processing (RTP, JIPELEC, JetStar 100T Processor) in oxygen with a heating rate of 30 °C s⁻¹. This procedure was repeated ten times to reach a film thickness of ∼500 nm.

The in-plane stresses developed during the film processing were estimated from the changes in substrate curvature after deposition and crystallization. For that, the film surfaces were measured using a Taylor Hobson Form Talysurf 50 mm Intra 2 profilometer, in which a total of 400 scans were performed across a 10 × 10 mm area.

The crystalline structure of the films was monitored by x-ray diffraction (XRD) with a Bruker D8 Advance apparatus and Cu anode (λ_Cu = 1.5418 Å), while surface and cross section microstructures were studied with a field emission gun scanning electron microscope (FEG-SEM). The chemical profiles of the films were studied by x-ray photoelectron spectroscopy (XPS), as described elsewhere.^21,22 Special emphasis was put on identifying the formation of nickel oxide at the interface or the occurrence of interdiffusion between film and substrate.

Electrical characterization was accomplished on an array of parallel capacitors, fabricated by depositing ∼100 and ∼200 µm diameter Pt top electrodes on the BiFeO₃–PbTiO₃/Ni films by magnetron sputtering through a shadow mask (BAL-TEC SCD 050). The capacitors were RTP annealed at 350 °C for 10 min to improve electric contact. Current density vs electric field (J-E) curves were measured at room temperature under sine waves of 1 kHz frequency, applied with a HP 8116A function generator. Generated currents were amplified with a Keithley 428 current amplifier and collected with a Tektronix (TDS 520) oscilloscope. Ferroelectric hysteresis loops were then obtained by integration of the J-E curves. Nonswitching contributions (leakage currents and conductivity) were calculated and subtracted from the experimental hysteresis loop by a fitting procedure that simulates the linear contributions with the capacitance and resistance of the films and the nonlinear leakage contribution.²³ 

Films were poled at room temperature by the application of amplitude-biased sine waves of 1 kHz, resulting in an oscillating field between 0 and 1.5 MV cm⁻¹ that was maintained for 100 signal cycles. Positive and negative poling were carried out by changing the signal polarity in the sample holder. Correct poling and retention were checked by dynamic pyroelectric measurements as described elsewhere,²⁴ after which the magnetoelectric response was characterized. This was carried out with a system consisting of two Helmholtz coils (Serviciencia S.L.), one high power and one high-frequency, designed to independently provide a static magnetic field up to 1 kOe (to magnetically bias the material) and an alternate magnetic field up to 10 Oe at 1 kHz (the stimulus) on the sample, while magnetoelectric output voltages were monitored with a lock-in amplifier (Signal Recovery model 7265). The typical configuration for layered composites was chosen so that magnetic fields were applied in the film plane and voltages were measured across the film thickness between the top electrodes and the Ni substrate so that the 31-piezoelectric mode was excited. A scheme showing how the magnetoelectric measurements have been carried out in the films is shown in Fig. 1. The final data were the transverse magnetoelectric coefficient α₃₁^E under varying bias magnetic fields along the ferromagnetic loop. Piezoelectric coefficients were obtained from the magnetoelectric response obtained by simulation with the COMSOL Multiphysics 6.0 software, considering the geometry of the Ni substrate in the demagnetization of the sample and adapted with special procedures for layered composite materials.

The XRD pattern of the 0.65BiFeO₃–0.35PbTiO₃ film deposited on Ni is compared to that of a film deposited on Si in Fig. 2(a). In the latter case, a single phase perovskite layer with a pseudocubic symmetry and a significant ⟨100⟩ preferred orientation was detected in the XRD pattern, which has been associated with a major rhombohedral component in coexistence with a minor tetragonal one with a ⟨100⟩ texture because of the tensile stress imposed by silicon.²¹ The perovskite phase was also formed in the film on Ni. Contrary to the film on Si, the polymorphic phase coexistence of the rhombohedral and tetragonal phases is clearly observed in the corresponding XRD pattern. Roughly, similar percentages of both phases are present in the film. The tetragonal component seems to show a ⟨001⟩ texture that could be associated with compressive stresses. This would be in agreement with previous reports for solution processed PZT films on Ni foils.¹⁶ It is worth comparing these patterns for films with those of bulk ceramics with the same composition, also included in the figure.¹⁸ The same polymorphic phase coexistence is observed. However, the relative intensities of the peaks of the tetragonal and rhombohedral phases in the ceramic and the film on Ni are not the same, probably indicating different percentages of the tetragonal and rhombohedral phases when the material is prepared in bulk or thin film forms, being the tetragonal phase higher in the former than in the latter.

The phase coexistence for bulk materials has been reported across a compositional range that can be as wide as 0.2 ≤ x ≤ 0.4,^25,26 but that is narrowed down to 0.27 ≤ x ≤ 0.31 when chemical homogeneity is enhanced.²⁷ A shift of the MPB toward higher x-values seems to take place for films on both Ni and Si, being more significant for the latter case since only the rhombohedral phase is detected in the XRD pattern in Fig. 2(a). This observation is independent of the distinctive preferred ⟨100⟩ orientation because no tetragonal peaks are found, thus totally proving the movement of the MPB toward higher x-values for films compared with bulk materials. Phase diagrams, and specifically MPB regions, are sensitive to different parameters, such as grain size. For instance, MPB polymorphs of PMN-PT are known to evolve with the reduction of grain size in the submicron range, down to the nanoscale.²⁸ The SEM images for the films on Si and Ni are given in Figs. 2(b) and 2(c), respectively. The microstructures differ, so that a polycrystalline film with a grain size of around 100 nm and non-negligible porosity was obtained on Ni [Fig. 2(c)], while on Si [Fig. 2(b)], highly dense columnar microstructures with a column width of around 150 nm resulted. Nonetheless, the actual differences in characteristic length scale (grain size or column width and film thickness) are not large and all lie within the submicron range, as in the bulk ceramic chosen for comparison here (∼500 nm).¹⁸ It should be noted that a relatively large grain size is obtained for these films on Ni processed by RTP at only 500 °C. Grain growth could have been promoted by introducing a 5% excess of Bi, which is expected to enhance mass transport during crystallization through the formation of a liquid phase (melting point of Bi is ∼271.5 °C), which would contribute to the reduction of the processing temperature of the film.^17,29 Regarding the small peak recorded at 2θ ∼ 27.8° in the XRD pattern of the film on Ni, it should not be related to the traditional formation at low temperatures of a bismuth-rich ternary bismuth iron oxide (JCPDS 46-0416), since it is not detected by XRD in the counterpart film on the Pt-coated Si substrate [Fig. 2(a)].

Back to the shift in MPB, this should then be associated with a different parameter, the stress state being the most likely candidate. Epitaxial strain is known to strongly modify the phase diagram of ferroelectric perovskite oxides.³⁰ For BiFeO₃–PbTiO₃ films, in-plane strains would drive the MPB toward high x-values, either tensile or compressive. The smaller shift for the film on Ni would indicate partial stress relaxation, consistent with the polycrystalline porous microstructure. Indeed, the substrate curvature was followed during film deposition and crystallization [Figs. 2(d)–2(g)], and very small compressive stresses were measured, as shown in the 3D surface profile shown in Fig. 2(f) and the same figure at an amplified scale [Fig. 2(g)].

In addition, the small peak at 2θ ∼ 27.8°, no additional peaks were found in the XRD pattern that could indicate the formation of NiO_x. Therefore, the possible occurrence of interface reactions between film and substrate was further investigated by XPS. The spectra at increasing sputtering times and the obtained chemical profiles are given in Fig. 3. During the first seconds of the Ar⁺ bombardment, Bi³⁺ and Pb²⁺ are reduced to their metallic states at the film surface and, subsequently, volatilized. A steady situation is reached after several minutes and is maintained until the Ni signal starts appearing 250 min later [Fig. 3(f)]. This indicates a homogeneous compositional profile, although a quantitative analysis is not possible because of the initial Bi and Pb preferential losses. Some indication of interdiffusion between Fe and Ni is found for longer sputtering times. Note the Bi, Pb, and Ti profiles shown in Figs. 3(a)–3(c), which start decaying after sputtering for 250 min and vanish 50 min later. However, this is not the case for Fe [Fig. 3(d)], which is still present after 300 min, while the Ni signal is already observed after 250 min [Fig. 3(e)]. Indeed, these two species show opposite evolutions between 250 and 300 min. The presence of oxidized nickel is confirmed by the two contributions in the Ni 2p_3/2 profile with maxima at ∼853.4 and ∼851.3 eV, assigned to Ni²⁺ and Ni⁰, respectively. In addition, the formation of mixed Fe and Ni oxides at the substrate-film interface is also inferred from these data [see Figs. 3(d) and 3(e)], which may be responsible for the small peak at 2θ ∼ 27.8° detected in the XRD pattern of the film on Ni [Fig. 2(a)].

The electrical characterization confirmed the ferroelectric nature of the BiFeO₃–PbTiO₃ films on Ni. Ferroelectric hysteresis loops at increasing electric fields are shown in Fig. 4(a), while the loop at the maximum applied field after subtracting the non-ferroelectric contributions²³ is shown in Fig. 4(b). High electric fields could be applied for the polarization of these films, which indicates a low number of defects and second phases in the films. As a result, the ferroelectric hysteresis loops show low leakage and conduction contributions. A loop for the submicrostructured bulk material is also included in Ref. 18. Note the very large value of ∼60 µC cm⁻² for the remnant polarization P_r, virtually equal to that of the bulk material. In addition, these films on nickel exhibit a high retention of polarization (P_r), as calculated from the dynamic pyroelectric measurements shown in Fig. 4(c). A pyroelectric coefficient of γ = 8 × 10⁻⁹ C K⁻¹ cm⁻² is obtained from these results, which remains stable after measuring one day later, indicating a large polarization retention of these films on nickel.²⁴ These retention values are more than enough for using these materials in applications related to piezoelectricity where polarization retention is demanded for their reliable use in sensors. Large ferroelectric polarization is characteristic of BiFeO₃ and its solid solutions, but it is remarkable that the bulk-like value can be attained in thin film form after processing at only 500 °C. Actually, a significantly lower P_r was obtained for the same films on Si (∼30 µC cm⁻²).²¹ Therefore, this cannot be associated with a size effect since the grain size of the film on Ni is actually slightly smaller than the column width in the textured film on Si. Indeed, this result clearly demonstrates the very good downscaling behavior of the switchable polarization of the ferroelectric BiFeO₃–PbTiO₃ system until grain sizes down to the verge of the nanoscale.^18,19 The better properties on Ni than on Si must then be related to the different stress states and their effect on the polymorphic phase coexistence and domain texturing. Thus, the observed enhancement of the switchable polarization in the film on Ni compared with that on Si is a consequence of a larger fraction of ⟨001⟩ tetragonal phase in the former case, as indicated by the XRD results in Fig. 2(a). The coercive field for the film on Ni is distinctively high, ∼1.4 MV cm⁻¹, which is about seven times more than the value of the same films on Pt-coated Si substrates, ∼200 kV cm⁻¹. This strongly suggests the presence of a dead layer between the Ni and the ferroelectric film that has to be related to the interdiffusion phenomena detected by XPS. In addition, a defect in the electrode charge screening can also produce an increase in the coercive field at different frequencies. Indeed, both phenomena are supported by the low effective permittivity ε_r ∼ 50 obtained from capacitance values when compared to the values of ε_r ∼ 140 measured for the film on Si.

Despite this dead layer, and thanks to the high dielectric strength, poling was possible, and a distinctive magnetoelectric response was measured using the procedure in Fig. 1. The room temperature magnetic responses of the films on Ni and the bare substrate have not shown an appreciable difference, which supports the kinetical limitation of nickel oxidation by the BiFeO₃–PbTiO₃ film preparation process here used. The magnetoelectric response of this film is shown in Fig. 4(d). A maximum transverse magnetoelectric coefficient α₃₁ of ∼25 mV cm⁻¹ Oe⁻¹ was obtained at a bias magnetic field of ∼360 Oe. This value is below that reported for optimized PZT films on Ni foils,^15,16,31 although any comparison must take into account the substrate geometry and the associated demagnetization effects, which are known to decrease as the aspect ratio increases.³¹ 

The expected response for the (1 − x)BiFeO₃–xPbTiO₃ system on a Ni substrate with the geometry used here (15 × 15 × 1.6 mm) and with the bulk piezoelectric coefficients of ceramics with a composition of x = 0.325 (d₃₃ = 87 pC N⁻¹ from¹⁹ d₃₁ = d₃₃/2) is also given in Fig. 4(e). The COMSOL Multiphysics 6.0 software, using special procedures specially developed in this work to simulate the magnetoelectric response of layered composites, was used, considering appropriate modifications to take into account demagnetization effects.³² Clamping on the opposite substrate surface was also considered, while Ni parameters were taken from the literature.^15,16

The results of this simulation show that a α₃₁ of ∼160 mV cm⁻¹ Oe⁻¹ should be expected, so the obtained value would be lower than that anticipated. This is often the case for magnetoelectric composites, where this behavior is associated with non-ideal elastic coupling between components. XPS data have indicated some interdiffusion and formation of Fe/Ni mixed oxides at the interface so that an inactive layer is present between the two ferroic materials with a detrimental effect on the response. This possibility and its effect on the magnetoelectric response can be evaluated by simulation with the COMSOL tool, considering the response of three-component structures, which would be formed by an additional non-active layer between the film and the substrate. This intermediate layer is most likely a mixed Fe/Ni oxide, so the elastic coefficients of the oxide film have been assumed in the calculations. This assumption is reasonable considering that the elastic mechanical impedances of Ni and typical piezoceramics are similar.¹⁶ Simulations then revealed a minor effect of this inactive layer, so that the magnetoelectric coefficient only decreased by ∼2% after inserting inactive layers of increasing thickness up to 25 µm.

However, neither XPS nor FEG-SEM showed any indication of such a thick layer, so the comparatively low response must be associated with the actual piezoelectric response of the two components, most probably that of the film. Piezoelectric coefficients for ferroelectric thin films are always lower than those for bulk materials of the same composition, for which different mechanisms have been raised. The grain size and stress effects on polymorphic phase coexistence, domain configurations and dynamics, or polarization retention after poling have been demonstrated. In addition, the COMSOL tool has provided us with a powerful means of evaluating the actual feasibility of this effect by obtaining the d₃₁ coefficient that would result from the experimental α₃₁. The simulated responses for decreasing piezoelectric coefficients are given in Fig. 4(e). The experimental α₃₁ of 25 mV cm⁻¹ Oe⁻¹ is obtained for a d₃₁ of 8 pC N⁻¹, which would then be the piezoelectric coefficient value for the film. This implies a reduction of ∼80% compared with that of the bulk material, which is within the expected decrease produced for a nanostructured material.²⁰ 

Therefore, a distinctive magnetoelectric response consistent with the good polarization retention observed in the nanostructured thin films was obtained by low temperature solution processing of BiFeO₃–PbTiO₃ films on Ni. The demonstrated fabrication method compares favorably with previous solution processing of the ferro-piezoelectric model based PZT perovskite thin films on Ni, which required both much higher processing temperatures (over 650 °C) and the previous deposition on the Ni surface of inactive barriers, such as Pt or HfO₂, to avoid extensive nickel oxidation. Hence, LaNiO₃ conductive layers have been used to simultaneously induce film texture and serve as a bottom electrode.^3,16 Their deposition, however, involved complex and costly deposition techniques and additional thermal treatments at high temperatures, after which some degree of Ni oxidation was still observed.³³ The low temperature solution deposition method proposed here also compares favorably with attempts to reduce the NiO_x layer formed in situ during the fabrication of films on Ni, for example, BaTiO₃ on Ni.¹⁴ In this case, chemical solution treatments in H₂O₂ were applied to the pristine nickel substrate to control the thickness of this NiO_x layer. This surface layer was subsequently reduced to nickel metal after the solution deposition of the film by carrying out the film crystallization at high temperatures to form gas (N₂/H₂). Films with good properties were attained, but one more time, the fabrication procedures were tedious and expensive.

The low-temperature solution process developed in our group for the preparation, for the first time, of BiFeO₃–PbTiO₃ films on Ni substrates is a low-cost deposition method capable of providing films with bulk-like ferroelectric properties and distinctive magnetoelectric response at only 500 °C.

This remarkable functional response is thought to be the consequence of the kinetic limitation for the nickel oxidation, the coexistence of a ⟨001⟩-oriented tetragonal phase with the rhombohedral one, and the very good downscaling behavior of properties in the BiFeO₃–PbTiO₃ system. As an additional effect of the low temperature processing, the formation of charge carrier species, such as Fe²⁺ or second minor phases, commonly present in BiFeO₃ based materials, is strongly hindered. This is proven by the large electric fields that can be applied for the poling of these films on nickel without producing an appreciable increase in film conductivity or short-circuiting of the capacitors. The presence of charged defects and secondary phases in the films would produce non-negligible conduction and leakage currents, which are mainly responsible for the degradation of the ferroelectric behavior of the films processed by typical chemical solution deposition at high temperatures, well above 650 °C.

This work was funded through the Spanish Projects TED2021-130871B-C21/AEI/10.13039/501100011033/Unión Europea NextGenerationEU/PRTR, and PID2022-136790OB-I00 and PID2019-104732RB-I00, funded by MCIN/AEI/10.13039/501100011033. E.R.C. thanks the Spanish Ministry of Science and Innovation (PID2021-126235OB-C32, funded by MCIN/AEI/10.13039/501100011033/and FEDER funds, and TED2021-130756B-C31, MCIN/AEI/10.13039/501100011033, and “ERDF: A way of making Europe” by the European Union NextGenerationEU/PRTR). L.Z. acknowledges the Higher Education Commission of Pakistan (HEC) for the financial support for his stay at ICMM-CSIC under the Indigenous 5000-PhD Fellowship Program (PIN: 112-27186-2PS1-143) and the International Research Support Initiative Program (IRSIP) (PIN: IRSIP 43 PSc 48).

The authors have no conflicts to disclose.

Miguel Algueró: Conceptualization (equal); Investigation (equal); Methodology (equal); Resources (equal); Supervision (equal); Writing – original draft (equal). Layiq Zia: Investigation (equal); Methodology (equal). Ricardo Jiménez: Conceptualization (equal); Investigation (equal); Methodology (equal); Supervision (equal). Harvey Amorín: Investigation (equal); Methodology (equal); Supervision (equal). Iñigo Bretos: Investigation (equal); Methodology (equal); Supervision (equal). Adriana Barreto: Investigation (equal); Methodology (equal). G. Hassnain Jaffari: Visualization (equal). Enrique Rodríguez-Castellón: Investigation (equal); Methodology (equal); Visualization (equal). Pablo Ramos: Data curation (equal); Software (equal). M. Lourdes Calzada: Conceptualization (equal); Investigation (equal); Methodology (equal); Resources (equal); Supervision (equal); Writing – original draft (equal).

Data sharing is not applicable to this article as no new data were created or analyzed in this study.