Fabrication and Characterization of Hybrid Films Based on NiFe2O4 Nanoparticles in a Polymeric Matrix for Applications in Organic Electronics

Hybrid films for applications in organic electronics from NiFe2O4 nanoparticles (NPs) in poly(3,4 ethylene dioxythiophene), poly(4-styrenesulfonate) (PEDOT:PSS), and poly(methyl methacrylate) (PMMA) were fabricated by the spin-coating technique. The films were characterized by infrared spectroscopy, atomic force microscopy, scanning electron microscopy, and energy-dispersive spectroscopy to subsequently determine their optical parameters. The electronic transport of the hybrid films was determined in bulk heterojunction devices. The presence of NiFe2O4 NPs reinforces mechanical properties and increases transmittance in the hybrid films; the PEDOT:PSS-NiFe2O4 NPs film is the one that has a maximum stress of 28 MPa and a Knoop hardness of 0.103, while the PMMA-NiFe2O4 NPs film has the highest transmittance of (87%). The Tauc band gap is in the range of 3.78–3.9 eV, and the Urbach energy is in the range of 0.24–0.33 eV. Regarding electrical behavior, the main effect is exerted by the matrix, although the current carried is of the same order of magnitude for the two devices: glass/ITO/polymer-NiFe2O4 NPs/Ag. NiFe2O4 NPs enhance the mechanical, optical, and electrical behavior of the hybrid films and can be used as semi-transparent anodes and as active layers.


Introduction
Spinel ferrites with general formula AB 2 O 4 (where A is a divalent cation and B is Fe 3+ ), are amongst the most studied iron-based systems. As nanomaterials, spinel ferrites are interesting due to their magnetic properties, which suggests their potential applications, as follows: in ferrofluidics [1] and high-density memory systems [2]; as adsorbents and catalysts for water decontamination and treatment [3,4]; as contrast agents in magnetic resonance imaging [5]; and in controlled drug delivery [6,7], hyperthermia [8][9][10][11], and other biomedical applications. Moreover, ferrites have applications in high-frequency drive technologies due to their low conductivity as well as their low current loss. Ferrites are high-resistivity dielectrics; therefore, they are used in the fabrication of transformer cores operating at high frequencies, in audiovisual memory storage, and in recording heads [12]. Depending on the cations making up a ferrite, it will have different thermal, electrical, magnetic, optical, and mechanical (etc.) properties [13]. In particular, nickel ferrite (NiFe 2 O 4 ) is an inverse spinel-type oxide with a bandgap of 1.5-5.0 eV for the bulk material. This ferrite crystallizes in a cubic system with space group Fd-3m and has low coercivity and saturation magnetization values as well as a high electrical resistivity; these properties intensity was measured between 10 • and 70 • with a 2θ step of 0.01 • and 0.25 s per point. The average crystal size (D) of the NPs was estimated from their diffractograms using the Debye-Scherer formula, D = κλ/βcos θ, where κ is the shape factor equal to 0.9, λ is the CuK α radiation, β is the full width at half maximum intensity (FWHM) of the selected peaks, and θ is the Bragg angle. Transmission electron micrographs (TEM) were obtained with an FEI Tecnai F20 microscope (Thermo Fisher, Waltham, MA, USA) operating at 200 kV S/TEM field emission with an X-TWIN lens and a high-brightness field emission electron gun (FEG).

Hybrid Film Deposition
Hybrid films composed of NiFe 2 O 4 NPs embedded in poly (3,4- Initially, the Corning glass and the ITO glass substrates were sequentially washed in an ultrasonic bath with dichloromethane, methanol, and acetone. The silicon substrates were washed with a "p" solution (10 mL HF, 15 mL HNO 3 , and 300 mL H 2 O) to remove surface oxide. After deposition, the films were dried at 85 • C for 2 min on a hot plate. The PEDOT:PSS-NiFe 2 O 4 NPs films underwent a post-treatment by vapor exposure of isopropanol (IPA) and were heated at 40 • C for 10 min.

Hybrid Film Characterization
FTIR spectroscopy analysis was performed using a Nicolet iS5-FT spectrometer (Thermo Fisher Scientific Inc., Waltham, MA, USA) at a wavelength range of 4000 to 400 cm −1 . Raman spectra of 100 to 900 cm −1 were acquired in a Nicolet Almega XR (Thermo Scientific Nicolet, Waltham, MA, USA) dispersive Raman spectrometer and detected by a CCD camera at 25 s and a resolution of~4 cm −1 . The excitation beam was a Nd:YVO 4532 nm laser, and the incident power on the sample was~3 mW. The roughness, topography, thickness, and mechanical properties of the hybrid films deposited on the silicon substrate were investigated with an atomic force microscope (AFM) using an Ntegra platform (Nanosurf, Liestal, Switzerland). In order to measure film thickness, an edge of the film was initially removed from the substrate in order to have a clean area so that only the substrate would be present and thus generate a step between the film and the substrate. The AFM tip was then placed on the film surface and made to pass through it in contact mode until it touched the clean substrate [34,35]. Subsequently, the images were analyzed with the Gwyddion software platform to determine the thickness through the height difference between the highest point and the lowest point. Due to the structural difference of the films and the type of technique used for the deposit, their average thicknesses changed slightly: 7.6 µm (PEDOT:PSS), 11.1 µm (PEDOT:PSS-NiFe 2 O 4 NPs), 6.5 µm (PMMA) and 11.7 µm (PMMA-NiFe 2 O 4 NPs).
To study the morphology and ensure the presence of the NiFe 2 O 4 NPs in the films deposited on the glass substrate, energy-dispersive spectroscopy (EDS) and scanning electron microscopy (SEM) were performed on a ZEISS EVO LS 10 scanning electron microscope (Carl Zeiss AG, Oberkochen, Germany) with a coupled Bruker microanalysis system (Bruker Nano GmbH, Berlin, Germany). Optical properties, absorbance, and transmittance of the hybrid films on the glass substrate were obtained using a UV-Vis 300 Unicam spectrophotometer (Thermo Fisher Scientific Inc., Waltham, MA, USA), in a wavelength range from 190 to 1100 nm. An auto-ranging picoammeter Keithley 4200-SCS-PK1 (Tektronix Inc., Beaverton, OR, USA) with a four-point probe and a lighting controller circuit from Next Robotix (Comercializadora KMox, S.A. de C.V., Mexico City, Mexico), was used to obtain the electrical properties through current-voltage (I-V) measurements of the fabricated devices, using an ITO glass substrate as the anode and silver as the cathode. Finally, to modify the behavior of the device, a 2,3,7,8,12,13,17,18-octaethyl-21H,23H-porphine iron(III) chloride (PFeCl: C 36 H 44 ClFeN 4 ) film was deposited over PEDOT:PSS-NiFe 2 O 4 NPs and PMMA-NiFe 2 O 4 NPs films using a high-vacuum thermal evaporation system (Intercovamex, Morelos, Mexico). The PFeCl film was heated to 250 • C and sublimated at a vacuum pressure of 2.6 × 10 −6 Torr; the deposition rate was 35.1 Å/s.

Results and Discussion
The NiFe 2 O 4 NPs were characterized by powder XRD and TEM microscopy. Figure 1 shows the XRD pattern of the NiFe 2 O 4 NPs (Figure 1a), which corresponds to an inverse spinel-type structure and a cubic phase with space group Fd-3 m and crystal lattice parameter a = 8.33 Å according to the ICDD 00-086-2267 card. No additional peaks from other faces were observed ( Figure 1b). Scherrer's equation estimated the average crystallite size at 6 ± 0.5 nm. TEM micrographs (Figure 1c,d) also corroborated the formation of nanocrystals with dimensions close to those determined from the corresponding X-ray diffraction pattern.
To study the morphology and ensure the presence of the NiFe2O4 NPs in the films deposited on the glass substrate, energy-dispersive spectroscopy (EDS) and scanning electron microscopy (SEM) were performed on a ZEISS EVO LS 10 scanning electron microscope (Carl Zeiss AG, Oberkochen, Germany) with a coupled Bruker microanalysis system (Bruker Nano GmbH, Berlin, Germany). Optical properties, absorbance, and transmittance of the hybrid films on the glass substrate were obtained using a UV-Vis 300 Unicam spectrophotometer (Thermo Fisher Scientific Inc., Waltham, MA, USA), in a wavelength range from 190 to 1100 nm. An auto-ranging picoammeter Keithley 4200-SCS-PK1 (Tektronix Inc., Beaverton, OR, USA) with a four-point probe and a lighting controller circuit from Next Robotix (Comercializadora KMox, S.A. de C.V., Mexico City, Mexico), was used to obtain the electrical properties through current-voltage (I-V) measurements of the fabricated devices, using an ITO glass substrate as the anode and silver as the cathode. Finally, to modify the behavior of the device, a 2,3,7,8,12,13,17,18-octaethyl-21H,23H-porphine iron(III) chloride (PFeCl: C36H44ClFeN4) film was deposited over PEDOT:PSS-NiFe2O4 NPs and PMMA-NiFe2O4 NPs films using a high-vacuum thermal evaporation system (Intercovamex, Morelos, Mexico). The PFeCl film was heated to 250 °C and sublimated at a vacuum pressure of 2.6 × 10 −6 Torr; the deposition rate was 35.1 Å/s.

Results and Discussion
The NiFe2O4 NPs were characterized by powder XRD and TEM microscopy. Figure  1 shows the XRD pattern of the NiFe2O4 NPs (Figure 1a), which corresponds to an inverse spinel-type structure and a cubic phase with space group Fd-3 m and crystal lattice parameter a = 8.33 Å according to the ICDD 00-086-2267 card. No additional peaks from other faces were observed ( Figure 1b). Scherrer's equation estimated the average crystallite size at 6 ± 0.5 nm. TEM micrographs (Figure 1c,d) also corroborated the formation of nanocrystals with dimensions close to those determined from the corresponding X-ray diffraction pattern. Hybrid films were fabricated through the spin-coating technique. After deposition, IR spectroscopy measurements of the PEDOT:PSS-NiFe 2 O 4 NPs and PMMA-NiFe 2 O 4 NPs films were performed in order to verify that no degradation of the polymer matrix had taken place during deposition. In Figure 2a, the spectrum of the PEDOT:PSS-NiFe 2 O 4 NPs film shows peaks corresponding to the polymer at (i) 1523 cm −1 for the C-C bond, (ii) 1305 cm −1 for the C-O-C bond, (iii) 1077 cm −1 for the S-O bond, and (iv) 1077 cm −1 for the bond between S and the phenyl group in PEDOT:PSS; (v) the last four peaks at 972, 917, 824, and 685 cm −1 reflect C-S group stretches in PEDOT:PSS [36][37][38]. Regarding PMMA, Figure 2b shows the IR spectrum for the PMMA-NiFe 2 O 4 NPs film. The following signals corresponding to the polymer can be observed: (i) 3000 and 2949 cm −1 for the methyl group, (ii) 1729 cm −1 for the C=O bond, (iii) 1640 cm −1 for the C=C bond, and (iv) 1136 cm −1 for the C-O bond [39]. IR spectroscopy results indicate that the polymers used as a matrix in the fabrication of the hybrid films did not undergo decomposition during the deposition process. Regarding the NPs, due to their chemical nature and resistance to organic solvents and their resistance to high temperatures, no decomposition was expected during the preparation of the hybrid films. However, their distribution in the polymeric matrix was verified later in the morphological studies of the hybrid films. Hybrid films were fabricated through the spin-coating technique. After deposition, IR spectroscopy measurements of the PEDOT:PSS-NiFe2O4 NPs and PMMA-NiFe2O4 NPs films were performed in order to verify that no degradation of the polymer matrix had taken place during deposition. In Figure 2a, the spectrum of the PEDOT:PSS-NiFe2O4 NPs film shows peaks corresponding to the polymer at (i) 1523 cm −1 for the C-C bond, (ii) 1305 cm −1 for the C-O-C bond, (iii) 1077 cm −1 for the S-O bond, and (iv) 1077 cm −1 for the bond between S and the phenyl group in PEDOT:PSS; (v) the last four peaks at 972, 917, 824, and 685 cm −1 reflect C-S group stretches in PEDOT:PSS [36][37][38]. Regarding PMMA, Figure  2b shows the IR spectrum for the PMMA-NiFe2O4 NPs film. The following signals corresponding to the polymer can be observed: (i) 3000 and 2949 cm −1 for the methyl group, (ii) 1729 cm −1 for the C=O bond, (iii) 1640 cm −1 for the C=C bond, and (iv) 1136 cm −1 for the C-O bond [39]. IR spectroscopy results indicate that the polymers used as a matrix in the fabrication of the hybrid films did not undergo decomposition during the deposition process. Regarding the NPs, due to their chemical nature and resistance to organic solvents and their resistance to high temperatures, no decomposition was expected during the preparation of the hybrid films. However, their distribution in the polymeric matrix was verified later in the morphological studies of the hybrid films. Raman spectroscopy was used to monitor the distribution of NPs in the polymeric matrix and the possible transformation of the benzoid structure to the quinoid structure of PEDOT before and after its treatment with IPA. Figure 3 shows the Raman spectra of PEDOT:PSS pristine and PEDOT:PSS-NiFe2O4 NPs films. The post-treated PEDOT:PSS-NiFe2O4 NPs film had a change in relative intensity in comparison to the PEDOT:PSS pristine film. Changes in the Raman spectral intensity confirm the interaction between the NiFe2O4 NPs and PEDOT chains; this may be caused by the redistribution of PEDOT:PSS and the NPs as a result of the IPA treatment. The increase in relative intensity for the Raman spectroscopy was used to monitor the distribution of NPs in the polymeric matrix and the possible transformation of the benzoid structure to the quinoid structure of PEDOT before and after its treatment with IPA. Figure 3 shows the Raman spectra of PEDOT:PSS pristine and PEDOT:PSS-NiFe 2 O 4 NPs films. The post-treated PEDOT:PSS-NiFe 2 O 4 NPs film had a change in relative intensity in comparison to the PEDOT:PSS pristine film. Changes in the Raman spectral intensity confirm the interaction between the NiFe 2 O 4 NPs and PEDOT chains; this may be caused by the redistribution of PEDOT:PSS and the NPs as a result of the IPA treatment. The increase in relative intensity for the treated film means that the PEDOT:PSS distribution becomes more uniform through the film, enhancing the detection of their main planes in the Raman spectrum. Additionally, the strongest band between 1380 and 1620 cm −1 corresponds to the C α = C β symmetric stretching of the five-membered thiophene ring on the PEDOT chains [40][41][42]. The symmetric stretching vibration C α = C β has a red-shifted peak when the chain structure changes from the benzoid to the quinoid form [36][37][38]. The displacement of the 1430 cm −1 signal in the spectra in Figure 3 is practically negligible, which is an indication that in this film the transformation of the PEDOT structure was not carried out. This result is relevant because the presence of NPs inhibits the transformation of PEDOT to the quinoid form; however, the post-treatment with IPA helps to improve the distribution of NPs in the polymeric matrix.
film, enhancing the detection of their main planes in the Raman spectrum. Additionally, the strongest band between 1380 and 1620 cm −1 corresponds to the Cα = Cβ symmetric stretching of the five-membered thiophene ring on the PEDOT chains [40][41][42]. The symmetric stretching vibration Cα = Cβ has a red-shifted peak when the chain structure changes from the benzoid to the quinoid form [36][37][38]. The displacement of the 1430 cm −1 signal in the spectra in Figure 3 is practically negligible, which is an indication that in this film the transformation of the PEDOT structure was not carried out. This result is relevant because the presence of NPs inhibits the transformation of PEDOT to the quinoid form; however, the post-treatment with IPA helps to improve the distribution of NPs in the polymeric matrix. In the EDS analysis of the films after the IPA treatment, as shown in Figure 4a,b, the chemical elements corresponding to the polymer and the NPs can be observed. In both spectra, the presence of Ni and Fe is due to the NPs, and the presence of S and C in Figure  4a is due to the presence of PEDOT:PSS. The presence of O is due to the polymer as well as the NiFe2O4 NPs. As for the PMMA films, the presence of C in Figure 4b is due to the polymer and that of O is related to the NPs. SEM analysis (Figure 4c,d) was used to study the morphology of the films; a homogeneous distribution of NPs throughout the entire matrix was observed in PEDOT:PSS-NiFe2O4 NPs and PMMA-NiFe2O4 NPs. However, in the PEDOT:PSS film some NP agglomerations were observed which produced larger structures than those observed in the PMMA film. According to the EDS and SEM results, it is evident that the NiFe2O4 NPs were adequately incorporated into the polymer matrix, thus forming dispersed heterojunction films; however, a further topographic study of the films was necessary due to the important requirements in organic electronics related to the homogeneity and quality of the films needed in these devices. In the EDS analysis of the films after the IPA treatment, as shown in Figure 4a,b, the chemical elements corresponding to the polymer and the NPs can be observed. In both spectra, the presence of Ni and Fe is due to the NPs, and the presence of S and C in Figure 4a is due to the presence of PEDOT:PSS. The presence of O is due to the polymer as well as the NiFe 2 O 4 NPs. As for the PMMA films, the presence of C in Figure 4b is due to the polymer and that of O is related to the NPs. SEM analysis (Figure 4c,d) was used to study the morphology of the films; a homogeneous distribution of NPs throughout the entire matrix was observed in PEDOT:PSS-NiFe 2 O 4 NPs and PMMA-NiFe 2 O 4 NPs. However, in the PEDOT:PSS film some NP agglomerations were observed which produced larger structures than those observed in the PMMA film. According to the EDS and SEM results, it is evident that the NiFe 2 O 4 NPs were adequately incorporated into the polymer matrix, thus forming dispersed heterojunction films; however, a further topographic study of the films was necessary due to the important requirements in organic electronics related to the homogeneity and quality of the films needed in these devices.
The topographic characterization of the hybrid films was supplemented by means of AFM studies. Figure 5 shows the AFM images for the films of the pristine PEDOT:PSS and PMMA polymers (Figure 5a,c, respectively) and for the hybrid PEDOT:PSS-NiFe 2 O 4 NPs and PMMA-NiFe 2 O 4 NPs films (Figure 5b,d, respectively). In the case of the films with PEDOT:PSS, a different topography is observed in terms of the growth direction of the polymer; in the pristine film, the direction is perpendicular to the surface (~90 • ), and in the PEDOT:PSS-NiFe 2 O 4 NPs film with IPA post-treatment, a preferential growth direction is observed with an angle smaller than 90 • . The interaction between NiFe 2 O 4 and PEDOT:PSS can be a decisive factor in the topography of the film because NPs can behave as nuclei around which the polymer is deposited in a preferential direction. In Table 1, the root mean square roughness (RMS) and the average roughness (Ra) are presented; RMS represents the average of the squared deviations concerning the average height of the films and is more significant than Ra, which represents the arithmetic average of the absolute values of the heights of the films. The PEDOT:PSS pristine film had a lower roughness than the PEDOT:PSS-NiFe 2 O 4 NPs film. This result is to be expected considering that the NPs have an average crystallite size of 5 ± 0.4 nm [33] and that when they are embedded in the PEDOT:PSS matrix they increase the roughness of the film. In the case of the PMMA polymer films, the topography, RMS, and Ra changed drastically upon incorporating NiFe 2 O 4 NPs. The presence of NPs not only increases roughness but also allows the topography to become heterogeneous, with a larger grain size. When comparing the PEDOT:PSS films with the PMMA films, it is observed that although RMS and Ra do not differ significantly, their topographies are very different from those of the PEDOT:PSS-NiFe 2 O 4 NPs and PMMA-NiFe 2 O 4 NPs films. PEDOT:PSS seems to have better interaction and behavior as a matrix and generates more homogeneous films than PMMA. The topographic characterization of the hybrid films was supplemented by means of AFM studies. Figure 5 shows the AFM images for the films of the pristine PEDOT:PSS and PMMA polymers (Figure 5a,c, respectively) and for the hybrid PEDOT:PSS-NiFe2O4 NPs and PMMA-NiFe2O4 NPs films (Figure 5b,d, respectively). In the case of the films with PEDOT:PSS, a different topography is observed in terms of the growth direction of the polymer; in the pristine film, the direction is perpendicular to the surface (~90°), and in the PEDOT:PSS-NiFe2O4 NPs film with IPA post-treatment, a preferential growth direction is observed with an angle smaller than 90°. The interaction between NiFe2O4 and PE-DOT:PSS can be a decisive factor in the topography of the film because NPs can behave as nuclei around which the polymer is deposited in a preferential direction. In Table 1, the root mean square roughness (RMS) and the average roughness (Ra) are presented; RMS represents the average of the squared deviations concerning the average height of the films and is more significant than Ra, which represents the arithmetic average of the absolute values of the heights of the films. The PEDOT:PSS pristine film had a lower roughness than the PEDOT:PSS-NiFe2O4 NPs film. This result is to be expected considering that the NPs have an average crystallite size of 5 ± 0.4 nm [33] and that when they are embedded in the PEDOT:PSS matrix they increase the roughness of the film. In the case of the PMMA polymer films, the topography, RMS, and Ra changed drastically upon incorporating NiFe2O4 NPs. The presence of NPs not only increases roughness but also allows the topography to become heterogeneous, with a larger grain size. When comparing the PE-DOT:PSS films with the PMMA films, it is observed that although RMS and Ra do not differ significantly, their topographies are very different from those of the PEDOT:PSS-NiFe2O4 NPs and PMMA-NiFe2O4 NPs films. PEDOT:PSS seems to have better interaction and behavior as a matrix and generates more homogeneous films than PMMA.    Regarding mechanical behavior when considering a maximum applied force of 900 N, the maximum stress (σ max ), the unitary deformation (ε unit ), and the Knoop micro hardness (HK) were obtained by AFM and are presented in Table 1. In the case of films with PEDOT:PSS, it is observed that the presence of NiFe 2 O 4 NPs increases the stress that the film is capable of withstanding by a factor of about 2.6 times. In the case of PMMA films, there is a very small increase in the stress that the film can withstand when introducing NiFe 2 O 4 NPs. The above may be a result of the increase in grain size observed in the PMMA-NiFe 2 O 4 NPs film ( Figure 5d) and a deficiency of this polymer in its function as a matrix. On the other hand, HK was calculated from the length of the long penetration diagonal in the films, and according to Table 1 the values obtained are very small, although they increase with the presence of NPs. This increase is more significant in the films with PEDOT:PSS, which once again demonstrates better behavior as a matrix as post-treatment with IPA favors the interaction and distribution of NPs in the PEDOT:PSS. Finally, and as expected, it is observed that the deformation in the hybrid films is smaller than in the films with the pristine polymers; the NPs exert the function of anchoring sites that prevent the flow and deformation of the matrix. For applications in organic electronics, it is required that the films have adequate mechanical resistance and dimensional stability, which according to the results obtained is favored by incorporating the NiFe 2 O 4 NPs in the PEDOT:PSS matrix.
The optical and electrical behavior of the hybrid films must also be evaluated to determine if they have a potential for use in device manufacturing. In order to evaluate the ability of the films to behave as transparent anodes, the transmittances of the pristine polymers and the hybrid films PEDOT:PSS-NiFe 2 O 4 NPs and PMMA-NiFe 2 O 4 NPs were evaluated. The spectra of Figure 6 show that the presence of NiFe 2 O 4 NPs increases the transmittance in the hybrid films. This effect is more significant in the PMMA-NiFe 2 O 4 NPs film, where a higher transmittance of 87% is obtained compared to the 80% that is achieved at wavelengths between 340 and 470 nm in the hybrid film with PEDOT:PSS. This result suggests the possibility of using PMMA-NiFe 2 O 4 NPs film as a transparent anode, which could eventually replace costly or easily oxidized anodes such as ITO or FTO (fluorine-doped tin oxide), which are just some of the expensive anodes currently used. In solar cells, for example, radiation must enter the interior of the device and reach its active layer, so one of its electrodes, which is normally the anode, should be (semi)transparent to radiation. According to these results, the PMMA-NiFe 2 O 4 NPs film is a viable option and an even better one than PEDOT:PSS, which has already been studied for this particular application [43][44][45].
If the use of hybrid films as constituents of devices in organic electronics is required, it is important to determine their optical band gap (E g ), which indicates their charge transport capability. The model by Tauc [46,47] relates the E g of hybrid films to its photon energy (hν) and the absorption coefficient (α) according to: (αhν) = A(hν − E g ) r In the above expression, A depends on the type of transition, r takes the value 2 for indirect electronic transitions in amorphous heterostructures, h is Planck's constant, and ν is the frequency, which is given by: where c is the speed of light, λ is the wavelength, and α is experimentally obtained from [47]: Here, d is the film thickness as obtained from AFM and T is the transmission; the reflectivity (R) is neglected due to its low level for each film, which is caused by the presence of NPs [33,48]. The Tauc band gap is determined from the graphs of Figure 7 by plotting (αhν) 1/2 versus hν and finding the intercept on the hν axis through extrapolation of the plot to (αhν) 1/2 = 0. In these graphs, a similarity is observed when comparing the curves for the films of the pure PEDOT:PSS and PMMA polymers and those for their films with NiFe 2 O 4 NPs. The Tauc bandgap value reported in Table 2 is the same for PEDOT:PSS pristine and for the PEDOT:PSS-NiFe 2 O 4 NPs film. According to Jarząbek et al. [48], this constant value of the Tauc band gap means that the NPs do not influence conjugation in the main, rigid polymer chains. It seems that the structural changes of the composite film during the incorporation of the NPs and the post-treatment with IPA are due to the redistribution of NPs in the polymeric matrix, but the conjugation in the main PEDOT:PSS chains is preserved and the energy band gap remains constant. Regarding the band gap of the PMMA and PMMA-NiFe 2 O 4 films, the introduction of the NPs influences the conjugation in the main chains of PMMA. The films with PMMA have the lowest Tauc gap and would therefore have a behavior resembling that of a semiconductor. The Tauc gap is a consequence of several factors related to electronic-transition processes, including defects, structural disorder, and traps. The Urbach energy (E U ) can be used to determine the defects in the band gap and can be evaluated from the following equation [49,50]: In addition to the parameters defined above, A a is a constant of the material that conforms to α at the energy gap. Figure 7c,d displays the nearly linear relationship between ln(α) and hν for the hybrid films. The values of E U were determined from the reciprocal of the slope from this linear relation and have been recorded in Table 2. The Urbach energy corresponds to the width of the band tail, which is related to localized states within the energy gap, possibly caused by structural defects [48]. As a reference, the value of E U is zero in a perfect semiconductor [51]. The highest E U belongs to the PEDOT:PSS-NiFe 2 O 4 NPs film, and the lowest E U belongs to the PMMA-NiFe 2 O 4 NPs film. From the values of E g and E U , it is understood that the PMMA film shows the best semiconducting behavior, which is a required feature in organic electronics. In addition to performing as a transparent anode (see Figure 6b), it also has the potential for use as an active layer. The behavior of the NPs and their effect on the optical properties of the films depend on the matrix in which they are embedded. In the case of films with PEDOT:PSS, the presence of NPs slightly increases the value of E U ; however, the value of E U for the PMMA pristine matrix is higher than that obtained with the NiFe 2 O 4 NPs. Finally, it is important to consider that the values obtained for E U are in the same range as those for some inorganic semiconductor films, such as Bi 2 S 3 (0.26 eV) and Bi 2−x Cr x S 3 (0.35 eV) [51]. They are lower than for some hybrid films, such as PVOH (polyvinyl alcohol) film embedded with Bi-nanoparticles (0.604 eV) [52], while they are bigger than for hybrid films with poly(3-hexylthiophene) (0.11-0.17 eV) [48].

Unfortunately, there is not much information available about the E U values for hybrid films with FeNi 2 O 4 NPs in a polymeric matrix.
Nanomaterials 2023, 13, x FOR PEER REVIEW 9 of 16 doped tin oxide), which are just some of the expensive anodes currently used. In solar cells, for example, radiation must enter the interior of the device and reach its active layer, so one of its electrodes, which is normally the anode, should be (semi)transparent to radiation. According to these results, the PMMA-NiFe2O4 NPs film is a viable option and an even better one than PEDOT:PSS, which has already been studied for this particular application [43][44][45]. If the use of hybrid films as constituents of devices in organic electronics is required, it is important to determine their optical band gap (Eg), which indicates their charge transport capability. The model by Tauc [46,47] relates the Eg of hybrid films to its photon energy (hν) and the absorption coefficient (α) according to: In the above expression, A depends on the type of transition, r takes the value 2 for indirect electronic transitions in amorphous heterostructures, h is Planck's constant, and ν is the frequency, which is given by: = where c is the speed of light, λ is the wavelength, and α is experimentally obtained from [47]: Here, d is the film thickness as obtained from AFM and T is the transmission; the reflectivity (R) is neglected due to its low level for each film, which is caused by the presence of NPs [33,48]. The Tauc band gap is determined from the graphs of Figure 7 by plotting (αhν) 1/2 versus hν and finding the intercept on the hν axis through extrapolation of the plot to (αhν) 1/2 = 0. In these graphs, a similarity is observed when comparing the curves for the films of the pure PEDOT:PSS and PMMA polymers and those for their films with NiFe2O4 NPs. The Tauc bandgap value reported in Table 2 is the same for PEDOT:PSS pristine and for the PEDOT:PSS-NiFe2O4 NPs film. According to Jarząbek et al. [48], this constant value of the Tauc band gap means that the NPs do not influence conjugation in the main, rigid polymer chains. It seems that the structural changes of the composite film during the incorporation of the NPs and the post-treatment with IPA are due to the redistribution of NPs in the polymeric matrix, but the conjugation in the main PEDOT:PSS chains is preserved and the energy band gap remains constant. Regarding the band gap of the PMMA and PMMA-NiFe2O4 films, the introduction of the NPs influences the conjugation in the main chains of PMMA. The films with PMMA have the lowest Tauc gap and would therefore have a behavior resembling that of a semiconductor. The Tauc gap is defects in the band gap and can be evaluated from the following equation [49,50]:

= ℎ
In addition to the parameters defined above, is a constant of the material that conforms to α at the energy gap. Figure 7c,d displays the nearly linear relationship between ln(α) and hν for the hybrid films. The values of EU were determined from the reciprocal of the slope from this linear relation and have been recorded in Table 2. The Urbach energy corresponds to the width of the band tail, which is related to localized states within the energy gap, possibly caused by structural defects [48]. As a reference, the value of EU is zero in a perfect semiconductor [51]. The highest EU belongs to the PEDOT:PSS-NiFe2O4 NPs film, and the lowest EU belongs to the PMMA-NiFe2O4 NPs film. From the values of Eg and EU, it is understood that the PMMA film shows the best semiconducting behavior, which is a required feature in organic electronics. In addition to performing as a transparent anode (see Figure 6b), it also has the potential for use as an active layer. The behavior of the NPs and their effect on the optical properties of the films depend on the matrix in which they are embedded. In the case of films with PEDOT:PSS, the presence of NPs slightly increases the value of EU; however, the value of EU for the PMMA pristine matrix is higher than that obtained with the NiFe2O4 NPs. Finally, it is important to consider that the values obtained for EU are in the same range as those for some inorganic semiconductor films, such as Bi2S3 (0.26 eV) and Bi2−xCrxS3 (0.35 eV) [51]. They are lower than for some hybrid films, such as PVOH (polyvinyl alcohol) film embedded with Bi-nanoparticles (0.604 eV) [52], while they are bigger than for hybrid films with poly(3-hexylthiophene) (0.11-0.17 eV) [48]. Unfortunately, there is not much information available about the EU values for hybrid films with FeNi2O4 NPs in a polymeric matrix.  To improve the semiconducting behavior of the hybrid films, a PFeCl layer was deposited by evaporation under high vacuum, forming a planar heterojunction on the polymer-NiFe 2 O 4 NPs films. The porphine film was added as an electron transport layer. According to Figure 8, the presence of PFeCl film with a low band gap could favor a semiconductor behavior in the/polymer-NiFe 2 O 4 NPs/PFeCl/system. The narrow band gap can be attributed to (i) FePCl optical properties, (ii) their proper interaction with the components of the hetero-junction system [23], and (iii) heat-induced movements of polymer particles [48,53]. During the deposition of the PFeCl film by the high-vacuum evaporation technique, the porphine molecules in a gaseous state are deposited on the polymer-NiFe 2 O 4 NPs film, and at this site they turn to the solid state. While this deposition occurs, the polymer-NiFe 2 O 4 NPs film heats up, thus enlarging the free volume and decreasing the order between polymer chains, which causes a blue shift and a decrease in the band gap [48,53]. cording to Figure 8, the presence of PFeCl film with a low band gap could favor a semiconductor behavior in the/polymer-NiFe2O4 NPs/PFeCl/system. The narrow band gap can be attributed to (i) FePCl optical properties, (ii) their proper interaction with the components of the hetero-junction system [23], and (iii) heat-induced movements of polymer particles [48,53]. During the deposition of the PFeCl film by the high-vacuum evaporation technique, the porphine molecules in a gaseous state are deposited on the polymer-NiFe2O4 NPs film, and at this site they turn to the solid state. While this deposition occurs, the polymer-NiFe2O4 NPs film heats up, thus enlarging the free volume and decreasing the order between polymer chains, which causes a blue shift and a decrease in the band gap [48,53]. To evaluate the electrical behavior in the heterostructures, Figure 9 shows the schemes of the devices manufactured for the polymer-NiFe2O4 NPs (Figure 9a) and/polymer-NiFe2O4 NPs/PFeCl/ (Figure 9b) devices. Figure 10a,b shows the behavior of the devices with PEDOT:PSS before and after performing the IPA treatment. When comparing the graphs, a marked difference between the two is observed as a result of the rearrangement between the NPs and the polymer after post-treatment. Before the treatment, an ohmic behavior is observed in the device under the different lighting conditions; in addition, there is a tendency towards the ambipolarity of the device. However, under most lighting conditions and at V > 0.4 V, the device interferes with the current flow, probably because the NPs are not perfectly embedded in the polymeric matrix. After treatment, the behavior of this device changes completely; according to Raman spectroscopy, the posttreatment with IPA changes the distribution of NPs in the polymeric matrix, and this influences charge transport. When analyzing and comparing the results obtained for the glass/ITO/polymer-NiFe2O4 NPs/Ag devices with PEDOT:PSS (see Figure 10b) and PMMA (see Figure 10c), it is observed that the maximum current carried is of the same order of magnitude in the two devices; the device with PEDOT:PSS is affected by the type Figure 8. Variation of (αhν) 1/2 with hν for hybrid films with a PFeCl layer.
To evaluate the electrical behavior in the heterostructures, Figure 9 shows the schemes of the devices manufactured for the polymer-NiFe 2 O 4 NPs (Figure 9a) and/polymer-NiFe 2 O 4 NPs/PFeCl/ (Figure 9b) devices. Figure 10a,b shows the behavior of the devices with PEDOT:PSS before and after performing the IPA treatment. When comparing the graphs, a marked difference between the two is observed as a result of the rearrangement between the NPs and the polymer after post-treatment. Before the treatment, an ohmic behavior is observed in the device under the different lighting conditions; in addition, there is a tendency towards the ambipolarity of the device. However, under most lighting conditions and at V > 0.4 V, the device interferes with the current flow, probably because the NPs are not perfectly embedded in the polymeric matrix. After treatment, the behavior of this device changes completely; according to Raman spectroscopy, the post-treatment with IPA changes the distribution of NPs in the polymeric matrix, and this influences charge transport. When analyzing and comparing the results obtained for the glass/ITO/polymer-NiFe 2 O 4 NPs/Ag devices with PEDOT:PSS (see Figure 10b) and PMMA (see Figure 10c), it is observed that the maximum current carried is of the same order of magnitude in the two devices; the device with PEDOT:PSS is affected by the type of radiation that falls on it. The maximum current carried is generated with wavelength illumination corresponding to green light, while a current reduction occurs under blue and violet lighting or under natural lighting conditions. The behavior of this device in darkness also changes, and this result could be an indication of photovoltaic properties that may be studied in the future. The type and position of the anode and cathode, as well as the direction of current flow, determine device behavior; at negative voltages, the behavior is practically that of an insulator, while at positive voltages, and despite presenting interference in charge transport, the behavior is ohmic. On the other hand, in the device with PMMA, there is no significant influence on charge transport when lighting conditions change; the behavior is mostly ohmic, although it is somewhat similar to that of a Schottky diode. In this device, there is no ambipolar behavior; nevertheless, current flow is continuous and without barriers or interference generated by the NPs-polymer interface.
The incorporation of the porphine film generates a significant change in the electrical behavior of the devices, which can be considered ambipolar (see Figure 10c show a behavior resembling that of a Schottky diode, and there are no significant changes when irradiating them with different wavelengths. Finally, it is important to mention that the device with the highest current is glass/ITO/PEDOT:PSS-NiFe 2 O 4 NPs/PFeCl/Ag/. From these results, it can be seen that the rectifying behavior of the devices without the porphine film acquires an ohmic character with the addition of PFeCl. In the PEDOT device, this also leads to an increased electric current, whereas the PMMA device undergoes a reduction in current, which can be attributed to changes in the local electronic structure at the interface between the polymer and the porphine films. of radiation that falls on it. The maximum current carried is generated with wavelength illumination corresponding to green light, while a current reduction occurs under blue and violet lighting or under natural lighting conditions. The behavior of this device in darkness also changes, and this result could be an indication of photovoltaic properties that may be studied in the future. The type and position of the anode and cathode, as well as the direction of current flow, determine device behavior; at negative voltages, the behavior is practically that of an insulator, while at positive voltages, and despite presenting interference in charge transport, the behavior is ohmic. On the other hand, in the device with PMMA, there is no significant influence on charge transport when lighting conditions change; the behavior is mostly ohmic, although it is somewhat similar to that of a Schottky diode. In this device, there is no ambipolar behavior; nevertheless, current flow is continuous and without barriers or interference generated by the NPs-polymer interface.  The incorporation of the porphine film generates a significant change in the electrical behavior of the devices, which can be considered ambipolar (see Figure 10c,d). Transported current increases by two orders of magnitude in glass/ITO/PEDOT:PSS-NiFe2O4 NPs/Ag devices relative to glass/ITO/PEDOT:PSS-NiFe2O4 NPs/PFeCl/Ag devices. On the other hand, through the incorporation of the porphine film, the electric current decreases by three orders of magnitude in the glass/ITO/PMMA-NiFe2O4 NPs/Ag/ device relative to the glass/ITO/PMMA-NiFe2O4 NPs/PFeCl/Ag/ device. The devices with the PFeCl film show a behavior resembling that of a Schottky diode, and there are no significant changes when irradiating them with different wavelengths. Finally, it is important to mention that of radiation that falls on it. The maximum current carried is generated with wavelength illumination corresponding to green light, while a current reduction occurs under blue and violet lighting or under natural lighting conditions. The behavior of this device in darkness also changes, and this result could be an indication of photovoltaic properties that may be studied in the future. The type and position of the anode and cathode, as well as the direction of current flow, determine device behavior; at negative voltages, the behavior is practically that of an insulator, while at positive voltages, and despite presenting interference in charge transport, the behavior is ohmic. On the other hand, in the device with PMMA, there is no significant influence on charge transport when lighting conditions change; the behavior is mostly ohmic, although it is somewhat similar to that of a Schottky diode. In this device, there is no ambipolar behavior; nevertheless, current flow is continuous and without barriers or interference generated by the NPs-polymer interface.  The incorporation of the porphine film generates a significant change in the electrical behavior of the devices, which can be considered ambipolar (see Figure 10c,d). Transported current increases by two orders of magnitude in glass/ITO/PEDOT:PSS-NiFe2O4 NPs/Ag devices relative to glass/ITO/PEDOT:PSS-NiFe2O4 NPs/PFeCl/Ag devices. On the other hand, through the incorporation of the porphine film, the electric current decreases by three orders of magnitude in the glass/ITO/PMMA-NiFe2O4 NPs/Ag/ device relative to the glass/ITO/PMMA-NiFe2O4 NPs/PFeCl/Ag/ device. The devices with the PFeCl film show a behavior resembling that of a Schottky diode, and there are no significant changes when irradiating them with different wavelengths. Finally, it is important to mention that  NPs/PFeCl/Ag devices, respectively, and 4.21 × 10 −7 and 2.98 × 10 −6 cm 3 for the glass/ITO/PMMA-NiFe 2 O 4 NPs/Ag and glass/ITO/PMMA-NiFe 2 O 4 NPs/PFeCl/Ag devices, respectively. As expected, there is a roughly exponential shape in the power density versus voltage graphs that can be associated to the rectifier-and ohmic-like behaviors of the devices. It is interesting to notice the discontinuous changes in slope in the graph of the PMMA-based device, which are possibly due to charge-storage effects at the interfaces between the nanoparticles and the matrix.
NPs/PFeCl/Ag devices, respectively, and 4.21 × 10 −7 and 2.98 × 10 −6 cm 3 for the glass/ITO/PMMA-NiFe2O4 NPs/Ag and glass/ITO/PMMA-NiFe2O4 NPs/PFeCl/Ag devices, respectively. As expected, there is a roughly exponential shape in the power density versus voltage graphs that can be associated to the rectifier-and ohmic-like behaviors of the devices. It is interesting to notice the discontinuous changes in slope in the graph of the PMMA-based device, which are possibly due to charge-storage effects at the interfaces between the nanoparticles and the matrix.

Conclusions
Hybrid films of NiFe2O4 NPs in PEDOT:PSS and PMMA were obtained by the spincoating technique. The NPs positively affect the films' mechanical properties by increasing the maximum stress they can withstand and their HK hardness. The behavior of the NPs and their effect on the optical properties of the films depend on the matrix in which they are embedded. Another important aspect of the use of NPs is their effect on the electrical behavior of glass/ITO/polymer-NiFe2O4 NPs/Ag devices, which changes depending on the matrix. In the case of PEDOT:PSS, IPA treatment influences charge transport, and the current carried in its device depends on the wavelength with which it is irradiated. In the case of the device with the PMMA matrix, an ambipolar behavior occurs which is independent of the light that falls on it. NiFe2O4 NPs may be considered for use in organic electronics applications; when used as film reinforcement with PEDOT:PSS and PMMA matrices, they enhance their mechanical, optical, and electrical properties. The addition of a porphine layer to devices based on these materials adds an ohmic component to their mostly rectifying behavior.

Conclusions
Hybrid films of NiFe 2 O 4 NPs in PEDOT:PSS and PMMA were obtained by the spincoating technique. The NPs positively affect the films' mechanical properties by increasing the maximum stress they can withstand and their HK hardness. The behavior of the NPs and their effect on the optical properties of the films depend on the matrix in which they are embedded. Another important aspect of the use of NPs is their effect on the electrical behavior of glass/ITO/polymer-NiFe 2 O 4 NPs/Ag devices, which changes depending on the matrix. In the case of PEDOT:PSS, IPA treatment influences charge transport, and the current carried in its device depends on the wavelength with which it is irradiated. In the case of the device with the PMMA matrix, an ambipolar behavior occurs which is independent of the light that falls on it. NiFe 2 O 4 NPs may be considered for use in organic electronics applications; when used as film reinforcement with PEDOT:PSS and PMMA matrices, they enhance their mechanical, optical, and electrical properties. The addition of a porphine layer to devices based on these materials adds an ohmic component to their mostly rectifying behavior.