Porous silicon (PSi), both as single layers and multilayers, was obtained from high conductivity (0.01-0.02 Ωcm), p-type, silicon wafers. The anodization in 1:2 (volume) HF:ethanol solutions (from commercial HF 48% (w/v) in water, Sigma-Aldrich) was carried out in a homemade Teflon® electrochemical cell, with a Pt reference electrode. The Si wafers were galvanostatically etched under illumination from a 100 W halogen lamp. Single layer PSi were obtained at 100 mA/cm² for 100 s, resulting in layers of about 10 microns. Different current densities were set for obtain different porosities from 40 to 120 mA/cm² in case of multilayer configurations (Figure  1). Two different porosity gradients were chosen to electrochemically grow Co NPs into PSi: A negative gradient (more porous towards the surface) and a positive gradient (more porous towards the substrate). In both cases the current densities applied were in four steps: 20, 40, 60 and 80 mA/cm². After etching, the Si/PSi substrates were rinsed in ethanol and dried with N₂.

Electroinfiltration of Co nanoparticles into PSi single- and multi-layers was carried out in a electrochemical bath with a Watt’s solution at RT (CoSO₄·7H₂O 0.2 M, CoCl₂ 0.05, H₃BO₃ 0.4 M, Na-Saccharine 25 g/l H₂SO₄ 1 mM). Boric acid was used as a buffer, to avoid pH fluctuations, sodic saccharine as crystallization catalyzer and sulfuric acid to decrease the pH of the solutions below 3. A well parameterized pulsed current process was used to allow the nucleation of metal nanoparticles into the pores of pulses of 40 mA/cm² and 10 s. An important parameter in pulsed mode is the equilibrium time or rest time, between pulses, set at 60 s; estimated by observing the dynamics of the voltage versus time curves (not shown). A overall view of the whole process of PSi single/multi layer fabrication and Co electroinfiltration is summarized in Figure  2.

Field emission scanning electron microscopy (FESEM) images were obtained in a XL 30S-FEG (PHILIPS). No metallization was required to observe the samples. Samples for cross section observation were prepared according to previously optimized protocols for mechanical and ion bean milling 9 . TEM/STEM characterization was carried out using a Jeol JEM 3000F with HAADF (High Angle Annular Dark Field) system included (300kV).

Preliminary structural studies of Co-PSi hybrid structures are carried out by Transmission Electron Microscopy (TEM) in a milled sample of single layer Co-PSi. As Figure  3 shows, Co ions infiltrate inside the PSi matrix and form spherical nanoparticles (NP) into the pores. In this case, nanoparticles of circa 5 nm are formed. Scanning TEM (STEM) images plus EDX analysis of these Systems (Figure  3) allow the identification of both elements in the NP.

Magnetic properties of single layer PSi-Co are evaluated by Alternating Gradient Field Magnetometry (AGFM). Samples electroinfiltrated at an increasing number of cycles with Co, from 5 to 20 cycles, are evaluated. Saturation magnetization values normalized to mass increase with increasing number of cycles in the studied range of cycles. Results point to the possibility of tailoring the magnetization of Co-PSi by controlling the amount of Co inside the PSi matrix (Figure  4). This is achieved by setting the number of electroinfiltration pulses during the synthesis process. Coercivity magnitude of the material depends on the amount of Co in the hybrid material: less concentrated systems present lower coercivity values 4 6 ; systems with higher Co concentration present higher values. This feature can be due to the appearance of inter-particle correlations inside the matrix or by the increase of the size of particles 12 favoured by an increasing density of NPs into the PSi matrix.

Once studied the magnetic potential of Co-PSi systems, the protective effect of the PSi matrix against the oxidation of Co is determined. Magnetic properties of Co strongly depend on the oxidation state and therefore determine the possible applications of these hybrid systems 13 . For this purpose a multilayer PSi is fabricated and infiltrated with Co. Two multilayer configurations are made, as Figure  1 shows, one with a negative porosity gradient (Co-PSi-n) and other with a positive one (Co-PSi-p). These systems are studied by XAS 14 15 16 (Figure  5). XANES spectra (Figure  5a) of both systems show clear differences in oxidation state. Co and Cobalt oxides references are obtained for comparing and adjusting the experimental spectra to known materials. From the fitting of the experimental data to references it can be observed that in Co-PSi-n systems, Co atoms are mainly in metallic state (close to 90% at.) and the remaining 10% at. is in the form Co₃O₄. Besides, studying Co-PSi-p systems shows Co to be composed by a mix of Co (40%at.), CoO (50%at.) and Co₃O₄ (10%at). The EXAFS spectra of the Co-PSi show the element environment and lattice structure in the Fourier space (Figure  5b). Comparing the EXAFS spectra of the Co-PSi systems with the reference it is straightforward to identify similarities of that of Co-PSi-n with the spectrum of metallic Co. The similarities are clearer if we obtain the transform the spectra to the real space by obtaining the Radial Distribution Function (RDF, Figure  5c). In this figure the spacial distribution of neighbours around the scattered element is represented and gives an overall situation of the short range environment of the Co. In Figure  5c the similarities of CoPSi-n with metallic Co are clearer. In case of CoPSi-p, EXAFS spectra has more common features with CoO than with Co₃O₄: position and shape of the RDF in the first neighbour matches with the O in the CoO, but the second one matches with the position of Co atoms in the metallic Co, confirming the observations in XANES spectrum.

The elemental concentration profile of these systems is studied by Rutherford Back Scattering (RBS). RBS spectra and elemental profiles of both systems are depicted in Figure  6. Figure  6a,b, show the raw RBS spectra of the CoPSi-n and CoPSi-p samples, respectively, and the simulated spectra with the SIMNRA software 11 . A reference of Ta-Ti alternated multilayers was used to perform the energy calibration of the experimental system. By using this reference, a fine adjustment of the set up parameters was achieved. In the analyzed PSi systems, the porosity degree does not contribute significantly to the measured oxygen amount. The void (or porous) part of the samples does not produce backscattering nor energy loss, and a porous material is indistinguishable of a bulk one by this technique. This is a reason to not be able to determine porosity profiles directly by RBS. In this respect, other authors have carried out these calculations by filling pores with hydrocarbon solvents like pentane to obtain porosity profiles in PSi multilayers 17 18 . In these cases the filling substance can be quantified and an indirect measurement of the porosity can be obtained.

From the raw spectra (Figure  6a,b) we observe a clear difference in the profiles. The feature in the higher energies region corresponding to the Co presents a localized broad peak in case of Co-PSi-p and an extended region for Co-PSi-n. In a first approximation there should exists differences in the Co distribution along the layer in both configurations. Both spectra present the Si (1600–1700 keV) and O (~1000 keV) edges with subtle differences that require finer calculations. By using the fitting software it has been determined that in Co-PSi-n structures, Co is uniformly distributed along the multilayer, and the oxygen concentrates mainly near the surface. This means that metallic Co is hosted in deeper zones of the multilayer and Co₃O₄ is concentrated in outer zones of the matrix. For Co-PSi-p systems, Co penetrates to less deep zones and the O concentration keeps fairly constant along the multilayer. This is due both to the presence of O bubbles inside the matrix and the coexistence of a mixture of Co oxides. Figure  6c,d summarize the elemental in-depth profiles of each systems, CoPSi-n and CoPSi-p, respectively, obtained from the simulations of the experimental spectra.

A typical cross section of a CoPSi-n system is observed by TEM to study the distribution of the Co deposits along the PSi structure. A representative image is presented in Figure  7. In this image, the distribution of the Co NPs inside the PSi matrix can be clearly identified and also the porous structure of the PSi layer. An image in the STEM mode (inset to Figure  7), which enhances the contrast with the differences in atomic weight, facilitates the identification of Co inside the Si matrix. Again, Co NPs of circa 5 nm and spherical shape are observed inside the PSi matrix, with a roughly uniform distribution of them inside the porous structure.