1 Introduction

The fossil energy reserves will soon start declining and the photo-electrochemical (PEC) conversion of visible light to electrical and/or chemical energy continues to attract a much attention in the scientific community [1,2,3]. Algeria has a high solar potential with an annual insulation of ~ 3200 h and an average solar constant exceeding 1200 W m−2 (South). However, the intermittence of this energy requires its storage to be released on demand. In this respect, hydrogen with environmentally friendly characteristic is suitable for long-term storage with a capacity of 28,540 cal g−1. It can be produced photo-electrochemically and considerable works have been focused on ferromagnetic materials active under UV–Visible light, and the 3d metal-based oxides remain privileged because of their chemical stability, high saturation magnetization, thermal properties, and facile synthesis [4,5,6,7,8,9] compared to sulfide. However, many of them suffer from limitations including a wide gap (Eg), due to the crystalline structure with a valence band (VB) made up of O2−: 2p orbital, while the conduction band (CB) derives mostly from metal character [10]. This drawback can be overcome by raising the energy of VB and this occurs by introduction of cationic orbital, which not only decreases the gap but maintains a high reducing ability of CB. In this category, the spinel magnetic nanomaterials MFe2O4 (M=Sr, Te, Ni, Co, etc.) continue to attract a growing interest in electronic, optic, magnetic resonance, biosensors, and photocatalysis [11,12,13,14,15,16,17,18,19,20] due to their magnetism and absorption over the visible region. In addition, they have pH-independent double layer through which the potential of H2O/H2 couple, which varies by ~ 60 mV pH−1, can be judiciously located. This constitutes a further advantage to have an optimal band bending (~ 0.4 V) at the semiconductor/electrolyte junction [21].

On the other hand, the preparation of nanocrystallites is attractive for small polaron oxides with a low carriers' mobility [22], and the spinels were the object of nano-sized effect inquiry [23, 24]. Hence, the conversion of the incident light to electrical current or chemical energy increases with decreasing the crystallite size till it becomes comparable with both the diffusion length of the minority carriers and the light preparation (α−1λ), where αλ is the absorption coefficient which depends on the wavelength (λ) [25]. In such a case, the electrons have a high probability to reach the liquid phase to reduce water into hydrogen [26, 27].

The main goal in photocatalysis is to produce large surface areas with crystallites size as small as possible and the sol–gel route seems appropriate in such a case. Our approach is then to prepare SrFe2O4 with nano-morphology of the kind obtained with the spinel NiFe2O4 [28]. The choice of SrFe2O4 for the hydrogen production is motivated by the friendly environmental characteristic, chemical inertness in aqueous medium, and absorption over the visible region. To our knowledge, the photo-electrochemical characterization of SrFe2O4 is missing, and the photocatalytic properties have been little investigated. The main purpose of this work was to describe the synthesis of SrFe2O4, to study its optical, magnetic, morphological, and photo-electrochemical properties in order to be applied for the hydrogen production upon visible light irradiation. Hydrogen is reported because of its chemical stability, low cost, and optimal gap. The loss by recombination is minimized by using a hole scavenger, and SO32− is appropriate in our case.

2 Experimental

SrFe2O4 was synthesized by sol–gel method from Sr(NO3)2 (Merck, 99%) and Fe(NO3)3, 9H2O (Flucka, > 98%), dehydrated in a dessicator under static vacuum for several days. The stoichiometric amounts {molar ratio Sr/Fe:1/2) were dissolved in distilled water containing agar-agar (1 g L−1); the solution was heated at 70 °C (3 h) and 100 °C (4 h) until total evaporation. Then, the powder was crushed in an agate mortar and fired at 800 °C with intermediate regrinding. The phase has been confirmed by X-ray diffraction (XRD) using a Phillips diffractometer with Cu Kα radiation (λ = 0.15418 nm) in the 2θ range (20°–80°).

The FTIR spectrum was performed and plotted with a Bruker Alpha (P1000) spectrometer in the range (400–4000 cm−1). The X-ray photoelectron spectroscopy (XPS) was performed with Source Gun AA < Ê = X-Ray 000 350 µm—FG ON (350 µm). The calibration was made by considering the C1s peak (289.4 eV). The SEM images were visualized with a Jeol JSM-7610 F equipped with a Noran EDS detector. The N2 adsorption–desorption isotherms were obtained by BET surface area analyzer (ASAP 2010 V5.02). The magnetic measurement was carried out at room temperature by using a Vibrating Sample Magnetometer (Microsense easy VSM software) in a magnetic field range ± 24,000 Oe which gives the variation among the applied field H and magnetization M.

The band gap (Eg) of SrFe2O4 was determined from the diffuse reflectance using a UV–Visible spectrophotometer (Model: Specord 200 Plus) and polytetrafluoroethylene as reference. The pellets, sintered at 850 °C, were electrically contacted to a copper wire with silver paint; the electrical conductivity was measured by the two probe method using GW Instek equipment (GDM 8255A). The data were accurate to ± 5 Ω cm, and the temperature was controlled with a Cr–Al thermocouple.

The photo-electrochemical study was done in a standard Pyrex cell; the pellets were encapsulated in glass tubes with insulating araldite, leaving a geometric surface area of 1.12 cm2. A PGZ301 Potentiostat was employed for the acquisition of electrochemical data; a saturated calomel electrode (SCE) and Pt foil were used as reference and auxiliary electrodes, respectively. The electrochemical impedance spectroscopy (EIS) was performed over the frequency range (1 mHz–100 kHz), while the Mott–Schottky characteristic was plotted at a frequency of 10 kHz. All experiments were initiated after stabilization of the open circuit potential (OCP).

The photocatalytic tests were carried in a water-cooled Pyrex reactor housing three tungsten lamps (3 × 200 W); the flux intensity (29 mW cm−2) was measured with a digital light meter (Testo 545). The temperature was maintained at 50 (± 1 °C) thanks to a thermo-stated bath. SrFe2O4 powder (200 mg) was immersed in 200 mL of SO32− (0.025 M) solution, and the pH was adjusted by the addition of KOH. The powder was uniformly dispersed under moderate agitation at 210 rpm to occupy the whole reaction space and to preclude the vortex phenomenon. The solution was bubbled by N2 gas at a purging rate of 10 mL mn−1; after 35 min, the light was switched on. Hydrogen was qualitatively identified by gas chromatography (Agilent Technology 7890A); because of its low solubility, the amount was quantified volumetrically in a graduated burette by water displacement due to the pressure produced inside the reactor.

3 Results and discussion

The synthesis of oxides by solid state reaction requires high temperatures, leading to irregular and large crystallites with low active surface areas. By contrast, the sol–gel route produces more or less homogenous powder with a narrow size distribution [29]. The XRD pattern of SrFe2O4 (Fig. 1) is characteristic of a single phase crystallizing in a spinel structure; the peaks are indexed with a cubic symmetry in agreement with the JCPDS Card N° 001-1027. The lattice constant (a) and spacing between crystal planes (dhkl) are linked by the relation [16, 30]:

$$a = d_{hkl} \sqrt[2]{{h^{2 } }} + k^{2} + l^{2}$$
(1)
Fig. 1
figure 1

X-ray diffraction pattern of SrFe2O4 prepared by sol–gel route and calcined at 800 °C

We do not detect any peak due to secondary phase like the deficient perovskite Sr2Fe2O5 [31] or SrO which converts into Sr(OH)2 when exposed to free atmosphere. The sharp peaks indicate small crystallites size (D = 31 nm) (β in rd., = /βcosθ), where K (= 0.94) is the shape factor and β (rd.) is the broadening at half maximum. The dislocation density (δ) was calculated with the help of the value of crystallite size using the following relations: [32]

$$\delta = \frac{1}{{D}^{2}}$$
(2)
$$V= {a}^{3 }{(\AA }^{3})$$
(3)

An active surface area of SrFe2O4 was calculated from the relation (= 6/ρXRDD), where ρXRD is the theoretical density (5.36 g cm3). The lattice constant, unit cell volume, crystallite size, and dislocation density of the spinel oxide are given in Table 1.

Table 1 Lattice constant (a), unit cell volume (V), crystallite size (D), and the dislocation density of SrFe2O4 nanoparticle

The FTIR spectroscopy is a useful technique for identifying the chemical absorption. The two broad bands in the spectrum are assigned to metal–oxygen of the ferrite spinel (Fig. 2). The highest one, ν1 in the range (800–600 cm−1), corresponds to stretching vibrations of the metal in tetrahedral site (Td), Mtetra MO, while the lowest ν2 band in the region (550–400 cm−1) is attributed to octahedral—metal stretching (Oh), Mocta MO [33,34,35,36]. The SEM micrographs of SrFe2O4 (Fig. 3) show give grains morphology. Some grains are necked in shape, due to the diffusion among the particles during sintering. This means that the smaller particles join together to form the larger particles[37]. Images show an intense and irregular dispersion of conglomerates without specific morphology [38]. Agglomeration occurs between the particles, due to the dipole–dipole interaction among the particles [39], and the elemental composition (Fig. 3) gives 38.56% of Sr, 32.18% of O, and 24.95% of Fe.

Fig. 2
figure 2

FTIR spectrum of SrFe2O4 prepared by sol–gel route

Fig. 3
figure 3

SEM images showing the morphology of SrFe2O4 at different magnifications. a ×5000, b ×10,000, c ×25,000, d ×50,000 ,and f EDX composition of SrFe2O4

The XPS analysis was used to determine the composition and surface valence states (Fig. 4), The characteristic lines of O, Sr, and Fe of the spinel are observed, and the carbon peak C 1 s (289.23 eV) is due to the grid used for analysis and/or CO2 adsorption. The analysis of the curves fitting revealed that Sr2+ and Fe3+ species are located in different sites (tetrahedral-A and octahedral-B). The deconvoluted spectrum of O 1 s orbital (Fig. 4a) shows one typical peak due to chemical state of oxygen (531.5 eV) in Sr–O–Fe. Figure 4b displays the spectrum of Sr 3d orbital with two peaks at 135.28 and 133.4 eV assigned to Sr 3d3/2 and Sr 3d5/2 orbitals, respectively. Figure 4c illustrates the spectrum of Fe 2p orbital, where the peaks 724.28, 710.78, and 718.68 eV correspond to Fe 2p1/2 and Fe 2p3/2 orbitals respectively [47]. The peaks at 710 eV and 718.68 eV are assigned to the presence of Fe3+ located respectively in the A- and B-sites, indicating that Fe3+ favors rather octahedral coordination in the spinel structure.

Fig. 4
figure 4

Survey XPS spectra for SrFe2O4, resolved core level spectra. O (1 s) (a), Sr (3d) (b) and Fe (2p) (c)

The N2 adsorption–desorption isotherm of SrFe2O4 (Fig. 5) gives a specific surface area of 26 m2 g−1.

Fig. 5
figure 5

N2 adsorption–desorption isotherms of the spinel SrFe2O4 at − 196 K

The magnetic properties of SrFe2O4, such as saturation magnetization (Ms), residual magnetization (Mr), and coercive field (Hc), as well as the squareness ratio Mr/Ms, were measured at room temperature in a magnetic field ranging from − 24,000 to 24,000 Oe (Fig. 6 and Table 2). An hysteresis loop was observed from the magnetization curve M(H), characteristic of a hard ferrite with ferromagnetic behavior and a Curie temperature above 300 K, due to the magnetization Mr (6.045 emu/ g) when the field (H) is zero. The significant field Hc (3892 Oe) makes this spinel ferrite resistant to demagnetization [40, 41]. Our sample with a squareness ratio of 0.47 (< 0.5) has a single domain structure, whereas materials with a ratio greater than 0.5 have a multi-domain structure [42]. So, we conclude that the prepared ferrite exhibits better magnetic properties, a desired property in photocatalysis.

Fig. 6
figure 6

Magnetic hysteresis loop for SrFe2O4 prepared by sol–gel method

Table 2 Magnetic parameters for SrFe2O4 spinel ferrite

The optical gap (Eg), vital in the photo-electrochemical conversion, quantifies the solar flux converted into useful energies. It is determined from the diffuse reflectance (R) by using the relation:

$$\left( {\alpha_{\lambda } h\nu } \right)^{m} = \, \left( {E_{g} - h\nu } \right)$$
(4)

The exponent m is equal to 0.5 or 2, respectively, for indirect and direct transitions. The Eg value (1.85 eV) is obtained by extrapolating the line (αλhν)2 to the energy-axis (Fig. 7), and the transition is directly allowed indicating that the top of the valence band (VB) coincides and the bottom of the conduction band (CB) is with no phonon assistance. The Eg value is close to that obtained elsewhere [43]; αλ (cm−1) is the absorption coefficient and its inverse represents the light penetration of the monochromatic light (λ). The gap (Eg) results from the octahedral crystal field splitting of Fe3+: 3d orbital into three lower energy levels t2g and two upper levels eg. It describes the lifting of orbital degeneracy in the Fe3+ complexes due to the presence of oxygen octahedra and the strength of the Fe–O bond.

Fig. 7
figure 7

Direct and indirect optical transitions of SrFe2O4 prepared by sol–gel route

The semi-conductivity is a required property in photocatalysis and depends on the extent of deviation from the stoichiometry. The Hall effect cannot be measured, indicating a resistive oxide (σ300K < 10–6 Ω−1 cm−1) with a low holes density (see below.) The electrical conductivity (σ) follows an exponential law {σ = σo exp {− Ea/kT)}, characteristic of a classical semiconductor; the activation energy (Ea) was deduced from the slope dlogσ/dT−1 (Fig. 8.). This seems to be not an intrinsic property but due to Fe3+ ions in interstitial site, dominated by small polarons lattice, thermally activated.

Fig. 8
figure 8

Logarithm of the electrical conductivity of SrFe2O4 vs. reciprocal temperature

The chemical inertness of the material is also crucial for the photocatalytic applications. By immersion for 1 week in the working solution (pH 12.5), SrFe2O4 shows no deterioration; the reaction of Fe3+, after acidification, by SCN was negative, and no red coloration due to the complex Fe(SCN)2+ was observed.

The intensity–potential J(E) profile of SrFe2O4 plotted in KOH electrolyte exhibits a plateau region over a large potential range, a behavior similar to a chemical diode, indicating a good electrochemical stability with a tunneling effect at large anodic polarization. The photocurrent (Jph) corresponding to the change in the current induced under visible irradiation, Jph, augments in the cathodic side, thus, implying p-type conductivity which becomes pronounced below − 0.4 V, and the onset photocurrent potential (Eon) is taken as the potential above which no photocurrent flows (Fig. 9).

Fig. 9
figure 9

The J(E) plots of SrFe2O4 electrode in KOH solution (0.5 M) under N2-atmosphere in the dark and under visible irradiation. Scan rate 10 mV/s

The ability of a semiconductor to drive the H2 evolution is governed by the position of its conduction band (CB), which must be more negative than the H2O/H2 level. In this regard, the capacitance−potential (C−2 − E) plot is used for a reliable determination of CB:

$${\text{C}}^{{ - {2}}} = \, \left\{ {2/\varepsilon \varepsilon_{0} N{}_{{\text{A}}}} \right\}\left\{ {E - E_{{{\text{fb}}}} } \right\}$$
(5)

All symbols have their usual significations. The flat band potential (Efb = 0.01 V) is simply deduced by extrapolating the linear plot to infinite capacitance (C−2 = 0) (Fig. 10), and is equivalent to the built-in potential in the solid state. The negative slope confirms the p-type conductivity where the majority carriers are holes. The semi-logarithmic (Fig. 11) confirms the electrochemical stability of the spinel. The energy of CB (= e × Efb + Ea − Eg) is positioned at 3.05 ± 0.10 V below vacuum [44] (Fig. 12). It corresponds to a potential of − 1.70 V, more cathodic than that of hydrogen and should trigger the H2 evolution under visible illumination. The valence band (VB: 4.85 eV/0.1 V) is simply deduced from the relation:

$${ }E_{{{\text{VB}}}} = { }E_{{{\text{CB}}}} - E_{{\text{g}}}$$
(6)
Fig. 10
figure 10

The Mott−Schottky characteristic of SrFe2O4 plotted in KOH solution at a frequency of 10 kHz

Fig. 11
figure 11

Semi logarithmic plots of SrFe2O4 in KOH (10–2 M) solution under N2 bubbling; scan rate 10 mV/s

Fig. 12
figure 12

Energetic diagram of the junction SrFe2O4/KOH electrolyte, given in the electrochemical scale at pH 7

EIS analysis quantifies the different contributions of the solid/liquid junction and provides information on the electron transfer mechanism. It is a non-stationary technique and the excitation with small amplitude does not modify the system, thus permitting repeated tests at the free potential (Fig. 12). The occurrence of two loops in the EIS plot of SrFe2O4 is due to the bulk and grains boundaries. As expected, the diameter of the first semicircle decreases under illumination (Fig. 13), thus corroborating the semiconducting behavior. The EIS diagram does not show a straight line at low frequencies, indicating the absence of Warburg impedance. Such observation agrees with the existence of a plateau region in the J(E) characteristics. The center of the semicircle is located below the real axis, due to a deviation from a pure capacitive behavior with the existence of a constant phase element (CPE). An electrical equivalent circuit is suggested by fitting the data with the Z-View software.

Fig. 13
figure 13

The complex impedance spectroscopy of SrFe2O4 in KOH solution plotted in the dark and under illumination. Inset: the equivalent electrical circuit

4 Photocatalysis

The oxide is not deliberately doped and the low holes density extends the space charge region up to ~ 0.9 µm, allowing the participation of the most electron/hole (e/h+) pairs in the photoactivity. The distance traveled by the electrons to reach the interface decreases, and these electrons are then captured by the water adsorbed on the surface inside the pores.

As mentioned above, SrFe2O4 is a small polaron semiconductor with a low carrier mobility (µh ~ 10–6 cm2 V−1 s−1), the latter is calculated from the relation (σ = µeNA). Therefore, the adopted strategy was to reduce the crystallite size below diffusion length, and the advantage of sol–gel is to produce small crystallites with increased specific surface. Nanoparticles with sizes (D) in the range (30–50 nm) give enhanced active surface area with more photocatalytic sites. A specific surface area of ~ 26 m2 g−1 was evaluated from the BET analysis. The electrochemical characterization revealed chemical stability in KOH electrolyte under irradiation. The excited electrons have a cathodic potential, sufficient to convert water into hydrogen.

Hydrogen is a nonpolluting fuel with no greenhouse gases emission and could bring a partial solution for the clean energy supply and environmental protection [45, 46]. The maximum photovoltage produced by a semiconductor under irradiation cannot exceed Eg/e. Accordingly, to split water into oxygen and hydrogen, a gap greater than 2.2 eV is required because of the high O2-over-voltage [48]. So, O2 cannot be evolved through by a valence band process and a reducing agent must be used to capture the photoholes; this improves the charges separation and protects the oxide against the photo-oxidation [49]. Figure 12 depicts a possible photocatalytic reaction mechanism of SrFe2O4. The electrons are excited to the conduction band, leaving holes on the valence band (8). The photogenerated electrons focalized to H2O molecules to be reduced into H2, while the holes react on the surface with SO32−. On the other side, the pH has a great influence in photo-electrochemistry. The results (Fig. 14) show that SrFe2O4 is active at various pHs, leading to different performances. The best activity is obtained at pH ~ 11, followed by pH 7 and 3 because the potential E(H2) moves toward the cathodic direction, producing an optimal interfacial band bending at pH ~ 11 with a complete separation of (e/h+) pairs. The H2 formation needs two electrons, and the conversion efficiency is given by the relation:

$$\eta = { 2 } \times \, \left\{ {{\text{number of H}}_{{2}} \,{\text{mol s}}^{{ - {1}}} } \right\}/\left. {{\text{photonsflux}}\,{\text{s}}^{{1}} } \right\}$$
(7)
Fig. 14
figure 14

Volume of evolved H2 on SrFe2O4 in alkaline medium (KOH, 10–2 M + Na2SO3, 0.025 M)

Water is reduced into hydrogen, while SO32− is concomitantly oxidized according to the following sequences:

$${\text{SrFe}}_{{2}} {\text{O}}_{{4}} + \, h\nu \, \to {\text{ CB }}\left( {{\text{e}}^{ - } } \right) \, + {\text{ VB }}\left( {{\text{h}}^{ + } } \right),$$
(8)

Cathodic pole

$${\text{CB }}\left( {{\text{2 e}}^{ - } } \right) \, + {\text{ 2H}}_{{2}} {\text{O }} \to {\text{ H}}_{{2}} + {\text{ 2OH}}\quad E^{^\circ } \sim - 0.{\text{2V}},$$
(9)

Anodic pole

$${\text{2 SO}}_{{3}}^{{{2} - }} + {\text{ 2 h}}^{ + } \to {\text{S}}_{{2}} {\text{O}}_{{6}}^{{{2} - }},$$
(10)
$${\text{SO}}_{{3}}^{{{2} - }} + {\text{ H}}_{{2}} {\text{O }} + {\text{ 2 h}}^{ + } \to {\text{SO}}_{{4}}^{{{2} - }} + {\text{ 2 H}}^{ + },$$
(11)
$${\text{2 SO}}_{{3}}^{{{2} - }} + {\text{ 3 H}}_{{2}} {\text{O }} + {\text{ 4 e}}^{ - } \to {\text{S}}_{{2}} {\text{O}}_{{3}}^{{{2} - }} + {\text{ 6 OH}}^{ - }.$$
(12)

The photocatalytic hydrogen formation occurs in ~ 20 min above which a net tendency to saturation was observed. This indicates that the reaction (11) is disfavored, otherwise the saturation should not be observed since the sulfate with a redox potential of ~ 2 V is stable and does not compete with photo-electrons. Therefore, one can conclude that the deactivation of the catalyst by H2-adsorption is due to competitive reaction (12) where the thiosulfate S2O32− consumes the photo-electrons. The saturation of catalytic sites also accounts for the regression of the photoactivity. More interestingly, the initial activity of the catalyst powder is completely restored during the second cycle.

Our research suggests that pursuing the synthesis of spinel AFe2O4 by Sol–Gel method, A denotes a alkaline earth metals. This is the goal to study the impact on the physico-chemical properties on the H2 photo-production.

5 Conclusion

The spinel SrFe2O4 was synthetized by sol–gel route with an average crystallite size of 40 nm. It is an attractive photocatalyst, due to its chemical stability, low cost, and optical gap of ~ 1.8 eV close to the optimal conversion value. The electronic bands are made up of 3d: orbital with a high reducing ability and appropriately positioned with respect to H2 level. The formation of cubic phase was confirmed by XRD analysis. The irregular and intense morphology of the spinel was observed by the SEM images. The VSM reveals a hard ferrite with a ferromagnetic behavior due to the presence of the magnetization Mr (6.045 emu g−1) for a zero applied magnetic field. The XPS analysis revealed that Sr2+ and Fe3+ are located respectively in four- and six-fold coordination. The d-d optical transition comes from the crystal field splitting of Fe3+: 3d in octahedral site. The thermal dependence of the electrical conductivity showed a semiconductor behavior with an activation energy of 0.09 eV. The capacitance measurement gave a flat band potential of 0.01 VSCE and a hole density of 1.4 × 1015 cm−3. The electrochemical characterization revealed chemical stability and the potentiality of SrFe2O4 in photocatalysis. Indeed, the spinel was tested for the hydrogen formation under illumination and the best activity occurs at pH ~ 11 with H2 evolution rate of 35 μmol mn−1 g−1 and a quantum yield of 1.8%. Therefore, the oxidation of sulfite permits to protect the spinel against photo-corrosion and to improve the charges separation. The saturation over time course evolution is due to the competition reduction of thiosulfate S2O32− and occupation of photo-electrochemical sites by adsorbed hydrogen.