A major source of renewable energy is solar energy (Ellabban et al., 2014). Nevertheless, the production of fuels and electricity from solar power is still costly, mainly because of the materials used in building the cells (Ellabban et al., 2014; Hussein, 2015). Except for Si and copper indium gallium selenide solar cells (CIGSSe), which are principal targets for photovoltaic (PV) applications, CdTe and GaAs are also significant photoactive materials. However, the mass production of such photovoltaics is limited due to the high production costs, the indirect bandgap energy (for Si) and the dependence on elements which are expensive (In, Ga, Te) or even hazardous (As, Cd) (Rohatgi, 1996; Mitzi et al., 2011; Kosten et al., 2013; Wang, 2014). Contrarily, hybrid lead halide perovskite solar cells (PVSCs) have been in the epicenter of the current solar cell research because of their facile fabrication process and due to the fact that they appear a high PCE of over 23% (Jung et al., 2019). However, PVSCs have disadvantages as well, that is, ultraviolet light absorption, and humidity and atmospheric oxygen affecting decomposition (Li and Liu, 2017; Fu et al., 2018387). Along with the presence of toxic lead, these factors restrain the advancement of the PVSCs. Therefore, it is essential to find alternatives for inorganic PVs that use inexpensive, eco-friendly and Earth abundant photoactive materials with appropriate semiconductor structure, while at the same time seek improvements in solar cell efficiency.

Metal oxide (MeOx) based solar cells have the potential to resolve some of the issues which arise in conventional solar cells. The all-oxide perspective is advantageous due to its excellent chemical stability, minor toxicity and ample quantities of metal oxides that effectively permit the manufacturing of solar cells under ambient conditions (Pérez-Tomás et al., 2018). MeOx are typically used as functional layers in solar cells such as transparent conducting front electrodes (ITO, FTO), (Kim et al., 1999; Han et al., 2007) electron (TiO₂, SnO₂, ZnO, Fe₂O₃ etc.) (Seo et al., 2018; Xiong et al., 2018; Shin et al., 2019; Papadas et al., 2019) or hole (Cu:NiOx, CuGaO₂, NiCo₂O₄, CuLi:NiCo₂O₄ etc.) transporting layers, (Galatopoulos et al., 2017; Papadas et al., 2018a; Papadas et al., 2018b; Ioakeimidis et al., 2019) whereas a very small number of MeOx have been used as photoactive layers, primarily Cu₂O, CuO and Co₂O₃ (Pérez-Tomás et al., 2018).

Ferroic semiconductors are now being continually included in the list of materials which are employed to investigate and push the efficiency limits in all-oxide photovoltaics. Ferroic semiconductor materials are considered to be light absorbers, such as Pb(Zr,Ti)O₃ (energy band gap (E_g) = 3.63 eV) (Pérez-Tomas et al., 2019), KNbO₃ (E_g = 3.8 eV) (Grinberg et al., 2013) and BiFeO₃ (E_g = 2.67 eV) (Papadas et al., 2015a). However, what seems to confine their implication to solar cells is their wide energy bandgap that results to low absorption of visible light and, thus, low conductivity. Other ferroelectric oxide semiconductors, like Bi₂FeCrO₆ (Nechache et al., 2014) and BiMnO₃ (Sun et al., 2017), have a suitable bandgap between 1 and 2 eV and are considered as more efficient in absorbing solar light. It is worth mentioning that a power conversion efficiency (PCE) up to ∼8.1% has been recently achieved using a single ferroelectric Bi₂FeCrO₆ layer fabricated by pulse laser deposited technique with the following structure: SrTiO₃/SrRuO₃/Bi₂CrFeO₆/ITO (Nechache et al., 2014). However, without the mentioned method of the thin film deposition which is considered to be a very complicated and energy demanding procedure, it is very unlikely to gain high PCE values (Calnan, 2014). Furthermore, the usage of the low-bandgap KBiFe₂O₅ material (E_g ∼1.6 eV) in the photovoltaic cells appears to be restricted due to low PCE (∼3 10^–3%) (Zhang et al., 2013). Therefore, the above mentioned examples demonstrate that further research is needed in finding more efficient narrow-bandgap, non-toxic and low-cost materials for solar cell devices.

The sol-gel synthesis has been the most frequently used method in the manufacture of MeOx. Nevertheless, to achieve crystallinity and to guarantee an efficient charge–carrier mobility, for metal oxide based active layers, high temperatures are necessary, which increases the cost of manufacturing and also restricts printable applications. These restrictions call for alternative techniques which could operate at lower temperatures. Compared to sol-gel synthesis of MeOx nanoparticles (NPs), the solution combustion synthesis (SCS) of NPs displays considerable advantages, such as use of a simple experimental setup, production of NPs with high crystallinity and pure phase, and exact control of the size and crystal structure of the particles by simple adjusting the fabrication conditions. (Suresh et al., 1991; Mimani and Patil, 2001; Patil et al., 2002a; Bansal, 2005; Deganello et al., 2009; Jadhav et al., 2011; Jiang et al., 2014) The SCS seems to be adaptable and effective for the growth of high crystalline MeOx layers at relative lower temperatures. As an exothermic procedure with a high rate of heat release, the necessity for high temperatures is circumvented and high purity MeOx NPs can be produced at moderate reaction conditions. In SCS process, the metal salts (e.g., nitrates) dissolved in saturated aqueous or alcoholic solutions act as oxidizing agents and react with organic fuels (such as urea, glycine, acetylacetonate, citric acid etc.) under relatively lower temperatures compared to other commonly used solution process methods to give rise to a combustion reaction and to produce the corresponding metal oxide NPs (Kim et al., 2011; Hsieh, 2014; Yu, 2015).

Iron manganite (FeMnO₃, FMO) is a mixed perovskite material with the chemical formula ABO₃, where the Fe atom is placed at the center of a cube formed by eight corner-sharing MnO₆ octahedra (Habibi and Mosavi, 2017). FeMnO₃ has been examined for applications such as lithium-ion batteries, catalysis, humidity sensors, energy storage and antibacterial devices (Doroftei et al., 2014; Cao et al., 2016; Cetin et al., 2019; Vasiljevic et al., 2020a; Nikolic et al., 2020). A large number of synthesis methods, such as co-precipitation, hydrothermal, ball milling, solid state reaction and sol-gel chemistry, have all been employed for the fabrication of FeMnO₃ materials (Sundari et al., 2013; Doroftei et al., 2014; Cao et al., 2016; Soni and Pal, 2016; Bin et al., 2017; Mungse et al., 2017; Gowreesan and Ruban Kumar, 2017; Saravanakumar et al., 2018; Cetin et al., 2019; Fix, 2019; Lobo and Rubankumar, 2019; Vasiljevic et al., 2020a; Nikolic et al., 2020; Vasiljevic et al., 2020b). Despite this, not all these techniques are viable to synthesize FeMnO₃ nanomaterials, as there are some drawbacks such as the expense of the source materials, chemical non-uniformity, high impurity, aggregated nanoparticles, and non-stoichiometry of some ferrite systems (Buonsanti et al., 2012; Alves et al., 2013; Bennet et al., 2016; Skliri, 2018). FeMnO₃ is a semiconductor which consists of plentiful and environmentally friendly elements with an ideal direct optical bandgap (∼1.5 eV) to absorb solar photons, while it has a deep lying valence band (VB ∼ 5.3 eV) (Skliri, 2018) that corresponds well to the VB edges of several p-type materials (e.g., CuO, NiCo₂O₄ etc.) (Savva et al., 2017; Papadas et al., 2018b). Additionally, it has high photochemical stability which is necessary for long-term optoelectronic devices while furthermore its intrinsic electric polarization field can enable charge-carrier separation within the semiconducting structure. Such characteristics make FeMnO₃ a promising light absorber for optoelectronic uses. On the other hand, between the numerous metal oxides that have been used as p-type active layers, nickel oxide (NiO) is a promising candidate for PVs due to its excellent electrochemical behavior (Kerli and Alver, 2016). NiO demonstrates a rock salt structure and exhibits adequate p-type conductivity with a wide bandgap in the range of 3.5 eV (Chrissanthopoulos et al., 2011; Mahmood et al., 2011; Zhang, 2015). NiO has been reported as p-type material in all-oxide solar cells in combination with the n-type TiO₂ and ZnO materials, while also solution combustion synthesized NiOx is commonly used as HTL in perovskite solar cells (Warasawa et al., 2013; Kawade et al., 2015; Kerli and Alver, 2016; Karsthof et al., 2016; Galatopoulos et al., 2017; Patel, 2017; Ukoba et al., 2018).

In this work, SCS of FeMnO₃ NPs is presented while indicating that tartaric acid can be used as a fuel and nitrate as an oxidizer agent. FeMnO₃ NPs characterized by an average size of ∼13 nm and a narrow particle-size distribution, were prepared using a low cost SCS process (6 h calcination at 450°C in air). The as-synthesized FeMnO₃ NPs were then functionalized with β-alanine and the ligand-capped NPs enabled the formation of compact and functional layers. These films were used, for the first time, as n-type photoactive materials and were incorporated in p-n heterojunction of all-oxide solar cells. For the purposes of this study, nanostructured NiO films were also synthesized by SCS method and applied as a p-type layer in the following structure ITO/NiO/FeMnO₃/Cu. The corresponding MeOx PVs show a high V_oc of 1.31 V with adequate FF of 54.3% and limited short current of 0.07 mA cm⁻² resulting to a PCE of 0.05%. Electrical characterizations by impedance spectroscopy reveled a high charge recombination resistance inducing high V_oc, whereas the limited current density is ascribed to the high charge transport resistance. Despite the low PCE values, these results provide a framework for further optoelectronic properties research on eco-friendly and cost-effective photoactive layers for fabrication of all solution processable inorganic photovoltaics.

Materials: Pre-patterned glass-ITO substrates (sheet resistance 4 Ω/sq) were purchased from Psiotec Ltd. All the other chemicals used in this study were purchased from Sigma Aldrich.

Solution combustion synthesis (SCS) of FeMn O ₃ NPs: For the synthesis of FeMnO₃ NPs, 0.5 mmol Mn(NO₃)₂.4H₂O, 0.5 mmol Fe(NO₃)₃.9H₂O and tartaric acid were blended in 5 ml of 2-methoxy ethanol solution. Subsequently, 150 μl HNO₃ (69% wt HNO₃) were added slowly into the mixture, and the solution stirred up to almost complete homogeneity. The whole solution was left under stirring for at least 3 h at room temperature (RT). The molar ratio of the total metal nitrates and tartaric acid was 1. Thereafter, the precursor solution was heated at 100°C under consecutive stirring until complete evaporation of the solvent. The dry black powder was then used for the combustion synthesis of the FeMnO₃ NPs in ambient atmosphere at 450°C in a preheated oven for 6 h, so that the combustion process be completed and then left to cool down at room temperature.

Perovskite FeMn O ₃ films preparation: The prepared FeMnO₃ (FMO) NP powder was used for the preparation of FMO dispersion for the deposition of corresponding film by spin coating technique. Firstly, the surface of NPs was modified with β-alanine. Briefly, as-made FMO NPs (50 mg) were added in 4 ml of deionized (DI) water containing β-alanine (10 mg), and the pH of the solution was adjusted to 4.2 with 1M HNO₃. (Papadas et al., 2015b; Skliri, 2018) To secure that NPs will transfer to the liquid phase and form a stable suspension, typically within 1 day, the resulting mixture was then intensively stirred at RT. The dispersion was aided with probe sonicator for about 30 min. To a stable colloidal dispersion of 30 mg ml⁻¹ be formed, the alanine-capped FMO NPs were isolated by centrifugation, rinsed several times with DI water, and finally dispersed in DMF. The obtained homogenous dispersion was then drop-casted and subsequently was spin coated on the top of NiO layer at 3,000 rpm for 40 s. The process of FeMnO₃ film formation was repeated about ten times to obtain a desired thickness of about 500 nm. To assemble a network of tightly connected metal oxide NPs, the deposition of the FeMnO₃ films was accomplished by spin coating technique of the colloidal NPs, followed by thermal annealing at 300°C for 30 min. In this way, β-alanine can enable direct NP–NP interactions upon ligand removal at growth temperature due to its small size, thus yielding high strength films consisted of firmly interconnected NP networks. This strategy is very important to obtain functional metal oxides photovoltaic devices with good charge transfer properties.

NiO NPs synthesis and films preparation by SCS: For the solution combustion synthesis of NiO, 1 mmol of Ni(NO₃)₂·6H₂O were dissolved in 2.5 ml of 2-methoxyethanol. After the solution was stirred at 50°C for 1 h, 0.1 mmol of acetylacetone was added to the solution, and the whole solution was allowed under further stirring for 1 h at RT. Spin coating technique was applied for the fabrication of the precursor films on the various substrates. The precursor’s solution was spin coated at 3,000 rpm for 40 s. The resulting light green colored films were dried at 100°C for 5 min and used as a precursor for the combustion synthesis of NiO NPs. Subsequently the obtained films were heated at 300°C in ambient atmosphere for 1 h in a preheated hot plate to complete the combustion process and then left to cool down at room temperature, forming a ∼50 nm thin layer.

Device fabrication: The metal oxides solar cells under study were ITO/NiO-NPs/FeMnO₃-NPs/Cu. ITO substrates were sonicated in acetone and subsequently in isopropanol for 10 min and then heated at 100°C on a hot plate for 10 min before use. The substrates were further treated with ozone for 10 min to achieve a better contact with the active layer by reducing the contact resistance. To fabricate the devices, a layer of NiO as p-type and FeMnO₃ as n-type side of the p-n junction were formed in sequence. The deposition of corresponding metal oxides films was described in detail above. Finally, 200 nm Cu layers were thermally evaporated through a shadow mask to finalize the devices, giving an active area of 0.9 mm².

Characterization: Thermogravimetric Analysis (TGA) were performed on a Shimadzu Simultaneous DTA-TG system (DTG-60H). Thermal analysis was conducted from 40 to 600°C in air atmosphere using air gas with a flow rate of 200 ml min⁻¹ and a heating rate of 10°C min⁻¹. X-ray diffraction (XRD) patterns were collected on a PANanalytical X´pert Pro MPD powder diffractometer (40 kV, 45 mA) using Cu Kα radiation (λ = 1.5418 Å). Transmission electron microscope (TEM) images and electron diffraction patterns were recorded on a JEOL JEM-2100 microscope with an acceleration voltage of 200 kV. The samples were first gently ground, suspended in ethanol, and then picked up on a carbon-coated Cu grid. Quantitative microprobe analyses were performed on a JEOL JSM-6390LV scanning electron microscope (SEM) equipped with an Oxford INCA PentaFET-x3 energy dispersive X-ray spectroscopy (EDS) detector. Data acquisition was performed with an accelerating voltage of 20 kV and 60 s accumulation time. Absorption measurements were performed with a Schimadzu UV-2700 UV-Vis spectrophotometer. For UV-VIS and PL measurements, thick films of FeMnO₃ NPs have been fabricated on top of the quartz substrates employing the spin coating method. UV–vis/near-IR diffuse reflectance spectra were recorded with a Schimadzu UV-2700 UV-Vis spectrophotometer, using BaSO₄ powder as a 100% reflectance standard. The energy bandgap (E_g) of the samples were estimated from Tauc plots of (Fhv)² as a function of photon energy (hv), where F is the Kubelka–Munk function of the reflectance (R): F=(1−R)²/(2R) (Kubelka, 1948). The thickness of the films were measured with a Veeco Dektak 150 profilometer. The PL measurements were performed on FeMnO₃ film on quartz substrate at an excitation wavelength of 400 nm. Photoluminescence (PL) spectrum was obtained at room temperature on a Jobin-Yvon Horiba FluoroMax-P (SPEX) spectrofluorimeter (Singapore) equipped with a 150 W Xenon lamp and operated from 300 to 900 nm. The current density-voltage (J-V) characteristics were characterized with a Botest LIV Functionality Test System. Forward bias scans were measured with 10 mV voltage steps and 40 msec of delay time. For illumination, a calibrated Newport Solar simulator equipped with a Xe lamp was used, providing an AM1.5G spectrum at 100 mW/cm² as measured by a certified oriel 91150 V calibration cell. A shadow mask was attached to each device prior to measurements to accurately define the corresponding device area. EQE measurements were performed by Newport System, Model 70356_70316NS. Atomic force microscopy (AFM) images were obtained using a Nanosurf easy scan two controller under the tapping mode. Electrochemical Impedance Spectroscopy (EIS) and Mott-Schottky measurements were performed using a Metrohm Autolab PGSTAT 302N, where for the EIS a red light-emitting diode (LED) (at 625 nm) was used as the light source calibrated to 100 mW/cm². For EIS a small AC perturbation of 20 mV was applied to the devices, and the different current output was measured throughout a frequency range of 1 MHz-1 Hz. The steady state DC bias was kept at 0 V throughout the EIS experiments. Mott-Schottky measurements on FeMnO₃ films were performed in a 0.5 M Na₂SO₄ aqueous electrolyte (pH = 7) using a Metrohm Autolab PGSTAT 302N potentiostat. A three-electrode set-up, with a platinum plate (1.0 × 2.0 cm²) and a silver-silver chloride (Ag/AgCl, 3M KCl) as the counter and reference electrodes, respectively, was adopted to study the samples. The capacitance of the semiconductor/electrolyte interface was obtained at 1 kHz, with 10 mV AC voltage perturbation. All the experiments were conducted under dark conditions. The measured potential vs the Ag/AgCl reference electrode was converted to the normal hydrogen electrode (NHE) scale using the formula: E_NHE = E_Ag/AgCl + 0.210 V. The working electrode for impedance-potential measurement was fabricated as follows, 10 mg of FeMnO₃ NPs was dispersed in 1 ml DI water and the mixture was subjected to sonication in a water bath until a uniform suspension was formed. After that, 100 µl of the suspension was drop-casted onto the surface of fluorine-doped tin oxide (FTO, 9 Ω/sq) substrate, which was masked with an epoxy resin to expose an effective area of 1.0 × 1.0 cm². The sample was dried in a 60°C oven for 30 min.

The solution processing technique provides an extensible low cost deposition procedure to fabricate high quality metal-oxide films and to replace costly and time consuming vacuum-deposition methods (Cochran et al., 2019). SCS has recently been employed in the low temperature manufacture of spinel nickel cobaltite (NiCo₂O₄) thin films as hole transport layers (HTLs) in inverted p-i-n perovskite solar cells (Papadas et al., 2018b). SCS possesses the benefit of speedily creating homogenous metal oxide materials of fine grain size. Most notably however, at a lower temperature than standard solid–state reaction, sol-gel and co-precipitation techniques (Chick et al., 1990; Rajeshwar and de Tacconi, 1998). Metal nitrates are distinguished for their capacity to synthesize metal oxides films of superior quality. Furthermore, it is important to choose the appropriate fuel agent for combustion, so as to circumvent the creation of sizeable clusters and particle agglomeration (Bansal, 2005; Verma et al., 2008). Tartaric acid was therefore used as the fuel agent in this work as it results in the formation of a single-crystalline phase of FeMnO₃. In general, tartaric acid leads to the formation of stable heterometallic polynuclear complexes (Selbach et al., 2007) because of its carboxylate and hydroxyl groups which can bind different metal ions from the solution, such as Mn²⁺ and Fe³⁺ (Wang et al., 2011). Basically, the growth of FeMnO₃ NPs is the result of combustion reaction of these polynuclear complexes while being heated in the presence of concentrated HNO₃ (Patil et al., 2002b; Cochran et al., 2019).

This study successfully proves that the synthesis of FeMnO₃ NPs can be achieved by a solution combustion technique, using tartaric acid as a fuel. Furthermore, the ultimate control of the nanoparticle’s size can be readily attained due to the multiple binding ability of the tartaric acid that resulted in the formation of single-phase FeMnO₃ with an average particle size of 13 ± 2 nm. X-ray diffraction and electron microscopy measurements verified the high phase purity and crystallinity of FeMnO₃. Additionally, we used a method to construct a network of tightly connected FeMnO₃ nanoparticles by spin coating of the colloidal solution. In this study, β-alanine was used as surface capping agent to produce a stable colloidal dispersion of FeMnO₃ NPs (β-alanine-capped FeMnO₃ NPs) in DMF. The short chain length of β-alanine allows direct interactions between the nanoparticles through ligand removal by thermal annealing (at 450°C in air), thus yielding a thick absorbing film (∼500 nm) consisting of continuous layers of interconnected nanoparticles and exhibiting a relative low roughness of ∼27 nm. The proposed strategy is crucial to obtain functional all-oxide photovoltaic devices that are developed using a process technique based on a simple solution. Furthermore, the inorganic perovskite FeMnO₃ was tested as a light absorber for photovoltaic applications for the first time. The band gap (∼1.54 eV) of the synthesized FeMnO₃ nanostructure was found to be very close to the hybrid lead perovskite CH₃NH₃PbI₃ material (∼1.55 eV). To this end, all-inorganic NiO/FeMnO₃ heterojunction photovoltaics were fabricated by solution combustion synthesis, using spin coating techniques. The corresponding all-inorganic solar cells reveal a high open circuit voltage (V_oc) of 1.31 V with a fill factor (FF) of 54.3% but exhibit a PCE of 0.05% under 100 mW cm⁻² illumination due to the limited short circuit current 0.07 mA cm⁻². Electrical characterization by impedance spectroscopy showed that the ITO/NiO/FeMnO₃/Cu device exhibits a high recombination resistance justifying the high V_oc. The high charge transport resistance indicates charge transport limitations within the relative thick (∼500 nm) n-type FeMnO₃ active layer. Optimizing the active layer thickness and improving the charge carrier transport properties are the main limiting processes of the n-type FeMnO₃ which results in low J_sc values for the presented solution processed based heterojunction inorganic photovoltaics. Moreover, further PCE improvement could be achieved by the incorporation of suitable charge selective contacts within the solar cell device architecture and the replacement of NiO with a more appropriate p-type material. The obtained results encourage more intense research on solution processed and environmentally friendly inorganic solar cells with suitable opto-electronic properties and high photon to electron conversion efficiency.