Supporting a photocatalyst, such as titania nanoparticles (TiO2 NPs), is a good strategy to improve its performance since it can facilitate the photocatalyst recovery from the aqueous media and provides a high surface area for pollutant adsorption. Among the several advanced functional materials used as TiO2 NP support, the hierarchical meso/macroporous SiO2 spheres not only show the advantages associated to its chemical nature but also the dendritic fibrous structure provides a porous network that offers many benefits to be exploited in optical and catalytic devices. In this chapter, different synthetic approaches to design hierarchical meso/macroporous silica and the strategies to support TiO2 NPs regarding the photocatalytic performance of these materials are shown.
- hierarchical SiO2 spheres
- meso/macroporous network
- dendritic silica
- supported TiO2 nanoparticles
- TiO2 photocatalysis
The practical applications of semiconductor photocatalysis using TiO2 nanoparticles (TiO2 NPs) are still limited by the catalyst recovery from the aqueous media. Supporting TiO2 NPs on suitable materials is an advantageous strategy to overcome this issue. The effectiveness of this process requires that the supporting material has mechanical and thermal strength, high surface area for the loading of NPs, and surface sites for covalent binding to avoid NPs detachment. Nevertheless, the support must not modify the electronic structure of TiO2 nor reduce the density of the surface defects that promotes the adsorption of reactive molecules [1, 2, 3]. Silica-based materials are attractive supports for TiO2 due to several reasons: (i) it enhances the thermal stability of the photocatalytically active anatase phase, (ii) it avoids the particles agglomeration , (iii) the establishment of an Si-O-Ti interaction prevents the NPs detachment, and (iv) SiO2 is transparent in UV-vis region and it does not interfere with the generation of electron-hole pairs in TiO2 . Additionally, SiO2@TiO2 core@shell structures enhance the absorption of photons due to the contrast between the refractive index producing a light-harvesting effect, which favors the photoactivity .
The optimal TiO2 photocatalysts are 5–50 nm NPs in the anatase phase . High loadings of NPs within this size range could be challenging for supports with low surface areas. Mesoporous silica with high surface area (500–1000 m2/g), such as MCM-41, MCM-48, SBA-15, and KIT-6 has been widely used to deposit TiO2 NPs [8, 9, 10, 11]. However, the pore diameters in these materials usually range from 2 to 15 nm ; hence, the crystallization of NPs with sizes comparable to the pores entrances will cause, inevitably, the clogging of the pores and a significant decrease in the surface area .
Hierarchically porous materials have the advantage that micro- and mesopores provide the size and shape selectivity for guest molecules, enhancing the host-guest interactions, whereas macropores favor the diffusion and accessibility of guest molecules . The need for tailored porous materials for different applications has motivated the design of hierarchical meso/macroporous silica by different synthetic approaches ; most of them are based on the sol-gel chemistry at liquid interfaces using arrangements of amphiphilic molecules as templates [16, 17, 18]. Interesting examples of these materials are the so-called dendritic fibrous nanosilicas (DFNS) [19, 20, 21], such as KCC-1 spheres (KAUST Catalysis Center-1) that are potentially useful for solving problems in catalysis, photocatalysis, DNA adsorption, CO2 capture, optical devices, and drug delivery .
For photocatalytic applications, the location of the TiO2 NPs plays an important role in the photoactivity, since NPs should be accessible for both, photons and reactive molecules. Figure 1 illustrates the different locations of the TiO2 NPs deposited on the fibrous silica KCC-1 in comparison with conventional mesoporous silica such as MCM-41 or SBA-15. In case of NPs deposited on KCC-1, the broad size of pore entrances facilitates the accessibility to the particles located closer to the center of the sphere.
In this chapter, we describe the synthesis pathways and the strategies for turning fibrous silica as support for photocatalytic TiO2 NPs for applications in dye-sensitized solar cells, water splitting, and photodegradation of dyes. The advantages of using these materials as support for TiO2 in relation to photocatalytic processes are addressed by comparing the photocatalytic activity of TiO2 NPs deposited by different routes, as well as with TiO2 NPs on mesoporous materials. Since the structure of these materials in the nanometer scale presents many analogies with forms of the micro- and macroscopic world: dendrites, carnations, wrinkles, fibers, in what follows, we will use indistinguishably the terms fibrous, dendritic, or wrinkled to refer to structures like those in Figure 2.
2. Synthesis of the hierarchical meso/macroporous SiO2 spheres
Most of the novel methods for building nanostructures with tunable morphology and pore size are based on the application of an oil—water interface using amphiphilic molecules as template, which allows the preparation of materials with well-defined structures in a confined reaction media [24, 25]. By changing the oil phase, surfactant, or reaction parameters, several morphologies can be obtained, such as rod-like , vesicle-like , cage-like , or fibrous structures .
The morphological and textural characteristics of a fibrous structure are shown in Figure 2. Transmission electron microscopy (TEM) image reveals a darker core surrounded by a meso/macroporous shell (Figure 2a). Unlike the narrow pore-size distribution of common mesoporous silica materials (e.g., MCM-41 and SBA-15), these particles have radially oriented pores whose diameter increases from the center of the sphere. The scanning electron microscopy (SEM) images show a narrow distribution of particle size, and the surface of each sphere presents cavities with a broad distribution of sizes limited by folded walls (Figure 2b). The associated nitrogen adsorption—desorption isotherm shown in Figure 2c could be classified as type IV isotherm with H3 hysteresis loop. Materials with H3 hysteresis do not show any limiting adsorption at high P/Po associated with nonrigid aggregates of plate-like particles and have slit-shaped pores . These particles have a complex porous structure (Figure 2d) with pores of 3.5 nm, resulting from the surfactant assemblies that define the pore structure of the walls , as derived from BJH analysis of adsorption—desorption isotherms. Additionally, there is a broad distribution of larger pores related to the lamellar structure unfolding toward the particle surface.
This type of morphology with radially oriented channels can be obtained in a ternary confined reaction media composed by a surfactant and a co-surfactant, a hydrocarbon, and water. In case of particles presented in Figure 2, the reaction media is composed by cetyltrimethylammonium bromide (CTAB) and 1-butanol, cyclohexane, and water. Usually, the silicon sources are alkoxides, such as tetraethyl or tetramethyl orthosilicates (TEOS, TMOS) that give rise to the SiO2 structure through sol-gel reactions.
The water-surfactant (+co-surfactant)-oil system may present one or more phases depending on the chemical composition. For instance, a thermodynamically stable microemulsion phase or multiphase systems containing a microemulsion layer can be obtained . To follow the description of the formation mechanism, it is necessary to introduce some concepts related to the system.
Micelle: structures formed by self-assembly of amphiphiles in aqueous solutions giving separate regions with opposite affinities toward a given solvent. Normally, micelles have characteristic size within the 2–50 nm range and form spontaneously under certain conditions of concentration and temperature .
Reverse micelle: spheroidal aggregates formed by amphiphiles in organic solvents. They can be formed both in the presence and in the absence of water. In a medium free of water, the aggregates are small and polydisperse .
Microemulsion: system composed of water, oil, and an amphiphile, which is a single optically, isotropic, and thermodynamically stable liquid solution . The microemulsions can exist as oil-swollen direct micelles dispersed in water (O/W microemulsions), water-swollen reverse micelles dispersed in oil (W/O microemulsions), or with both aqueous and oily continuous domains as interconnected sponge-like channels (bicontinuous microemulsions) .
Winsor emulsion systems: the Winsor classification of ternary systems, i.e., types I–IV, can be made according to the numbers and type of phases in the equilibrium state. Except for the single-phase microemulsion (type IV), the structure of the microemulsion is determined through the separation of the solvent from the dispersed phase of the microemulsion layer. Type I: at equilibrium, superfluous oil separates from the oil-in-water microemulsion. Type II: superfluous water separated from the microemulsion layer. Type III: both superfluous oil and water separated from the water-in-oil microemulsion .
Mesophase: a phase occurring over a defined range of temperature, pressure, or composition within the mesomorphic state. This is a state of matter in which the degree of molecular order is intermediate between the three-dimensional, long-range order found in solid crystals and the order found in isotropic liquids, gases, and amorphous solids .
Sol-gel reactions: these are kinetically controlled hydrolysis and polycondensation reactions of a metallic precursor (e.g., inorganic or alkoxide), which give rise to a metal oxide structure. The precursor, building blocks, evolves in several steps from dispersions of colloidal particles in a liquid (Sol) to interconnected polymeric chains forming a porous rigid network (gel) . The most generally accepted model to describe silica polymerization in aqueous media is nucleation and growth .
2.1 Proposed mechanism for the formation of the hierarchical porous SiO2 using oil-in-water emulsions
The general mechanism for the formation of hierarchical porous silica using oil-in-water emulsion made from a type III Winsor system can be described in the following steps, also schematized in Figure 3 :
The silicon source dissolved in the oil layer meets the water layer—an aqueous solution of a base—at the emulsion interface; the sol-gel reactions take place at this stage. In the basic solution, the reaction mixture contains ionized silicate monomers and oligomers negatively charged. As the silica condensation proceeds, the amount of partially condensed silicates decreases and that of fully condensed silicates increases.
Silicate oligomers interact with the surfactant heads giving rise to a change in the curvature of the interface.
As the curvature increases, a closed structure such as spherical or cylindrical shapes can be formed. These surfactant-silicate systems with closed structure lead to the formation of micelles or micellar emulsions inside the channels of the bicontinuous microemulsion.
The aggregation of these micelles leads to the formation of a repetitive mesophase.
However, in this mechanism, the growth of the particle is still unclear. It can be thought that the formed mesophase may act as a seeding site were the wrinkles grow through the water channels. As particles start growing, the structure of the microemulsion is rearranged, and according to Moon and Lee , the walls of wrinkles are formed in the water layers and the valleys are consequences of the oil phase in the microemulsion.
An alternative mechanism was proposed by Febriyanti et al. , considering the optimal conditions for the synthesis of KCC-1 . They suggest the formation of reverse micelles in which the sol-gel reactions take place at higher rate in the center of the droplet. This mechanism is schematized in Figure 4. Although this mechanism explains the TEM images in which the density of the particles seems to decrease from the center to the surface of the sphere, experiments using real-time small angle X-ray scattering and 3D tomography demonstrate that the fibers grow uniformly from the center of the spheres along the free radial directions to form fibrous spherical silica particles. Based on these results, Maity et al.  described the microemulsion-droplet-coalescence mechanism.
2.2 Tuning the morphology and pore size distribution of the hierarchical meso/macroporous silica spheres
The possibility to obtain tailor-made supports for specific guest molecules and tunable pore volume for loading control is the most attractive feature of the three-phase synthesis approach. In case of fibrous nanosilica KKC-1, the surface area may vary from 450 to 1244 m2/g, and the tunable pore sizes (3.7–25 nm) allow to increase the pore volume up to 2.18 cm3/g. Likewise, the particle size can be controlled from 40 to 1120 nm. . Parameters controlling particle size, pore volume, and surface area are discussed below.
2.2.1 Oil phase
Shen et al. developed a stratification approach to obtain a three-generation pore size by changing the oil phase in a biphasic system . They used 1-octadecene, decahydronaphthalene (decalin), and cyclohexane to pores of sizes 2.8, 5.5, and 7.0 nm in the same particle. Each mesoporous shell was obtained in a synthesis step using the corresponding oil phase. This confirms that the organic solvent not only provides a storage medium for the silicon source (TEOS), but also interacts and assembles with the surfactant molecules in the interface to form oil-in-water microemulsion droplets. Figure 5 shows the dendritic particles obtained by the biphasic stratification approach.
Depending on the molar ratio of surfactant to silicon, the surfactant participates in the synthesis with different functional roles: controlling the particle diameter, the particle dispersion in the medium, and the degree of mesophase formation. As the surfactants favor the dispersion of the alkoxysilanes in water, the hydrolysis reactions are favored in comparison with the condensation reactions leading to an increased nucleation rate. The net effect is that increasing the surf/Si ratio, the particle diameter decreases. Moreover, at a high surf/Si ratio, free micelles may adsorb in the primary particles, allowing the dispersion of the particles due to electrostatic and/or steric repulsion. In contrast, at low surf/Si ratios, the surfactants mainly act as porogen . For the KCC-1 synthesis using benzyldimethylhexadecyl ammonium chloride (16- BAC) instead of CTAB, the fibrous morphology is lost due to the differences in the packing parameters, which in turn depend on the length of the nonpolar chain and the effective area of the polar head .
By increasing the alkyl chain length of the co-surfactant (linear C3-C20 alcohols), the distance between wrinkle walls increases, as shown in Figure 6. As the affinity between oil and surfactant defines the valleys of the wrinkles, an increase in the hydrophobicity of the aqueous phase by enlarging the co-surfactant alkyl-chain broadens the distance between the wrinkle walls. It can be considered that this distance is directly related to the density of fibers by changing this parameter, the pore size distribution in the region related to the lamellar structure and the pore volume also changes, but the surface area remains unaltered .
2.2.4 Composition of the water phase
The type and concentration of the base catalyst dissolved in the aqueous phase affects the diameter of the spheres, as it controls the hydrolysis rate of the silicon source. Due to their buffer capacity, weak bases are preferred. For instance, basic amino acids, such as arginine or lysine, keep the reaction under weakly basic conditions leading to spherical particles with dendritic pores [18, 45]. The same particles can be obtained with urea, whose thermal decomposition above 70°C maintains the solution pH between 7 and 9.5, depending on the urea concentration . Nevertheless, the increase in urea concentration rises the nucleation rate leading to smaller particles.
Jin et al.  found that keeping constant the emulsion composition to obtain rod-like structures at low stirring speeds (i.e., <300 rpm) produces twisted-ribbon or twisted-rod-like structures with changes in the length and the diameter of the rods. On the other side, high stirring speeds (i.e., 1200 rpm) yield spherical or irregular morphologies. The stirring speed has two main effects in the synthesis: (i) it alters the self-assembly of the template and (ii) it affects the diffusion of the reagents to the interface . In the synthesis of wrinkled SiO2, high stirring speeds (around 900 rpm) are preferred to obtain a spherical morphology.
2.2.6 Reaction time
Hydrolysis and condensation rates for silica at pH 7–9 are slow processes occurring simultaneously with changes in surfactant assemblies. The synthesis of dendritic particles in the conditions giving the product as depicted in Figure 2 requires around 20 h to be completed. The particle evolution during these conditions monitored by SEM is shown in Figure 7. After 4 h, slightly structured particles (≤100 nm) are formed. As the reaction proceeds, lamellae are defined and particles grow, reaching an average diameter of 300 nm with high size dispersion after 6 h. Once lamellae are formed, the effect of reaction time is a slow increase of particle size up to 500 nm in 24 h, less polydispersion, and well-defined planes of wrinkles with deep valleys.
Several processes converge to achieve these corrugated particles. On the one hand, the silicon precursor condenses on the already formed silica units. On the other hand, there is a simultaneous silicate dissolution and re-precipitation (Oswald ripening) that generates hydrophilic silica surfaces. This process induces the formation of hydrophobic regions where the organic phase is accumulated giving rise to the deep valleys observed for 24 h. Although data on the kinetics of dendritic particles formation are scarce, it is known that the reaction may be accelerated by the assistance of microwave radiation, as proposed for the synthesis of the KCC-1. Under these conditions, particles are formed in few minutes but for longer reaction times (>60 min); the lamellae collapse producing denser particles . This experiment reinforces the idea that the necessary condition for having a dendritic structure is the formation of hydrophobic “pockets,” a process that is inhibited when microwave radiation is applied.
In summary, the main advantage of the synthesis using oil-in-water Winsor III emulsion is the production of dendritic particles that can be tailored by the synthesis parameters for a specific need. The adjustment of pore sizes is important for trapping biomolecules or nanoparticles and for diverse catalytic applications that require high surface area and pore volume. Moreover, for environmental applications, the particle size is a key factor for catalyst recovery. The surfactant to Si ratio and the concentration of the base catalyst are important parameters for controlling the particle diameter, whereas the microemulsion phase behavior defines the distance between the wrinkle walls, i.e., the macroporosity. For a given surfactant, the microemulsion structure is affected by the organic solvent and the co-surfactant chain length, which will tailor the size of cavities and fiber density. The reaction time is an empiric variable for each reaction set-up, which needs to be optimized to get the required product.
3. Strategies for supporting TiO2 NPs on hierarchical meso/macroporous silica spheres and photocatalytic performance
Catalysts based on Ti-containing mesoporous silica, e.g., MCM-41. SBA-15 has been developed from different synthetic pathways for applications such as selective oxidation of bulky organic compounds and photoreduction of greenhouse gases [48, 49].
Typically, two synthesis strategies could be applied depending on the stage of incorporation of the TiO2 precursor: (i) one-pot: added directly to the SiO2 reaction mixture (co-condensation route) or (ii) postsynthesis: added after the consolidation of the SiO2 network by deposition, impregnation, atomic layer deposition, among others. In the one-pot approach, the co-condensation of titanium precursor with the silicon source faces with the problem that Ti alkoxides are much more reactive than Si(OR)4 ; thus, it is necessary to control the Ti precursor hydrolysis in order to have an appropriate distribution of Ti species that lead to the formation of anatase nanocrystals, a necessary component for efficient photocatalysis .
In this section, we focus on different postsynthesis pathways leading to photocatalytic systems for energy conversion, H2 production, or photodegradation of dyes. The consolidated SiO2 porous network must ensure high loadings of homogeneously distributed TiO2 NPs attached by covalent interactions, while the deposition method determines the location and structure of the nanoparticles, e.g., particle size and crystallinity. All these features affect the photocatalytic performance of the semiconductor.
TiO2 NPs deposited on KCC-1 by atomic layer deposition (ALD) alternating titanium tetra-butoxide and hydrogen peroxide produces anatase NPs of 3–12 nm highly dispersed on the fibers (Figure 8a) whose diameter increases with the number of ALD cycles . The anatase phase was obtained after the ALD cycles by calcination at 300–700°C. This material exhibits high photocatalytic activity and turnover frequency (TOF) towards the Rhodamine B (Rh-B) photodegradation under UV light, even better than the one obtained using the same ALD procedure with MCM-41 and SBA-15 substrates (Figure 8b and c). This higher photoactivity may be a consequence of the location and distribution of TiO2 NPs on the support allowing an optimal dye-TiO2 interaction for a hole transfer or improved OH• attack to the dye. The open structure of the fibrous silica admits higher TiO2 loading accessible to the dye, while the ordered array of mesoporous channels in MCM-41 and SBA-15 is easily clogged by TiO2 NPs as schematized in Figure 1. The TiO2 loading on SiO2 is determined by the surface silanol/siloxane ratio which is independent on synthesis method . Thus, even all substrates exhibit nearly the same TiO2 loading in weight (Figure 9), the more open structure of KCC-1 allow better accessibility to Ti precursors as ALD cycles increase.
Enhanced photocatalytic activity can also be explained by higher photon harvesting in the fibrous structure of the KCC-1 , as these structures enable incident light to pass through the cavities and reduce optical loss due to multiple light scattering and reflection . The efficient photon harvesting of dendritic particles was also exploited for the design of dye-sensitized solar cells (DSCs) . By modifying the interwrinkle distances in the silica spheres with the co-surfactant’s length of the carbon chain, the fiber density changes from narrow (NWSNs) to wide (WWSNs) wrinkled silica. In this case, TiO2 NPs were deposited by dispersing the SiO2 spheres in a mixture of acetonitrile- ethanol (1:1) using ammonia solution to conduct the hydrolysis/condensation reactions of titanium ter-butoxide. Using this approach, 4 nm anatase crystallites were obtained after calcination at 500°C under air atmosphere. The dependence of photovoltaic parameters with the fiber density is interpreted as a consequence of the effective refractive index resulting from the combination of two materials with different optical properties, supporting the idea of improved photon harvesting promoted by the fibrous nature of the SiO2 spheres.
In the latter cases, the crystallization of TiO2 was conducted by calcination at temperatures higher than 300°C. Despite the fact that porous silica spheres have good thermal stability, the calcination presents several drawbacks for the photocatalytic process: (i) it reduces the surface hydroxyl groups and surface area  and (ii) it induces the anatase to rutile transformation leading to a decrease in photoactivity . Although anatase-rutile phase transition can be managed with the calcination temperature and by the introduction of defects, hydrothermal synthesis of TiO2 NPs is a more convenient method for producing anatase nanocrystals on silica substrate .
On the hydrothermal route, the solvent has a strong impact on the TiO2 NP size and distribution . Under soft conditions (180°C, 20 h) in ethanol-water mixtures, the TiO2 NPs are homogeneously located in the outermost zone of KCC-1 particles when using pure water. This causes an increment in the specific surface area and pore volume; whereas, by using absolute ethanol, the particles are dispersed in the fibers (see TEM images and particle size distribution in Figure 10a). Regarding the TiO2 NPs sizes, it goes from 8 nm to 17.5 nm by increasing the ethanol/water ratio from 0/100 to 28.5/1.5. Then, it decreases to 5 nm in a 100/0 ratio; under this condition, a narrower size distribution is obtained. As far as photocatalytic activity is concerned, particles prepared with absolute ethanol showed the highest activity for water splitting under UV light (Figure 10b). This result points out that the size and location of the TiO2 NPs plays a significant role in the photocatalytic performance. The TiO2 NP distribution leads to different morphologies with a characteristic effective refractive index, which defines the interaction with the incident photons. Thus, by controlling the accessibility of the TiO2 NPs to the porous system, it is possible to design suitable morphologies for improving photocatalysis.
Another alternative to control the TiO2 NP location is by taking advantage of the reagents used during the synthesis of the support. As shown in the previous section, the synthesis of wrinkled SiO2 spheres (see Figure 2) requires high amounts of CTAB acting as a pore template. Thus, it is possible to imagine that before removing the template, the pores are “filled” with CTAB molecules blocking the transport of Ti precursors. On this basis, our group developed a practical approach to handle the TiO2 NP distribution on wrinkled SiO2 spheres (460 nm in diameter, B.E.T. area = 550 m2/g) produced by hydrothermal synthesis using Ti isopropoxide in isopropanol. With this hydrothermal synthesis, TiO2 NPs of 10 nm with 6-nm anatase crystallites are obtained. The loading and location of the NPs significantly differ depending on the presence of CTAB during the hydrothermal treatment. Figure 11a summarizes the experimental approach for controlling the TiO2 NP location and their corresponding TEM and SEM images (Figure 11b). In Figure 11, w-SiO2@TiO2 refers to NPs deposited after the removal of CATB and w-SiO2/CTAB@TiO2 to NPs deposited in the spheres containing CTAB. By eliminating CTAB prior to TiO2 deposition, NPs are highly dispersed in a hybrid SiO2-TiO2 shell, resulting in a greater decrease in surface area than when TiO2 NP growth is performed in the presence of CTAB. This morphology resembles that obtained by atomic layer deposition (ALD) , as shown in Figure 8a. On the other side, when the hydrothermal process is carried out in the presence of CTAB, the NPs are arranged as a mesoporous TiO2 layer in the same way as the particles obtained by hydrothermal crystallization described in Figure 10a . Therefore, this synthetic approach provides different morphologies simply by taking advantage of CTAB molecules from the synthesis of SiO2 particles. The photoactivity of these particles was tested in the degradation of crystal violet (CV). As seen in Figure 11c, CV photodegradation occurs at higher rate when using SiO2-TiO2 hybrid layer particles (w-SiO2@TiO2) whose morphology allows better photon harvesting. In case of w-SiO2@CTAB TiO2 particles, the presence of CTAB increases the adsorption of the dye, thus providing a basis to functionalize the SiO2 particles for a more specific interaction with the targets. Finally, the dimensions of the support are adequate for a good recovery of the reaction medium providing an efficient system for environmental applications.
4. Summary and outlook
Meso/macroporous silica spheres are remarkable supports for TiO2 NPs as compared to conventional silica mesoporous materials, such as MCM-41 and SBA-15. The main advantage of this morphology is the accessibility of the molecules to the silica surface provided by the broad pore size distribution. For photocatalytic applications, the location of TiO2 nanoparticles is a critical parameter to optimize the photoactivity. Supported TiO2 NPs in these materials exhibit:
better accessibility to the active sites ensuring high TiO2 loadings;
high surface area, which favors higher adsorption of the pollutants;
enhanced photocatalytic performance for several target molecules;
improved light harvesting; and
facile recovery from the reaction media compared to unsupported TiO2 NPs.
The easy tailoring of the surface properties provides photocatalysts for specific applications that can be rationalized by considering the location of TiO2 NPs on the porous support, which in turn can be managed by the synthesis variables. Understanding the internal nature of the hierarchical supports of the photoactive TiO2 allows the development of value in photocatalysis for energy and environmental applications—a contribution of material’s scientists to a better world.
The authors are grateful with CONICET for the postdoctoral scholarship of K.M.F., and University of Buenos Aires for funding, UBACyT 20020170200298BA. We are in indebted to Nanoandes Network and ERANET-LAC FP7 project Recola for the continuous support. Also, we acknowledge Stefanie Kodjikian (Institut NÉEl, Grenoble) and Maria Claudia Marchi (CMA-CONICET) for TEM and SEM images, respectively.
Conflict of interest
There are no conflict of interest to declare.