Mg-substituted PrFeO₃ perovskite oxides, PrMg _x Fe_1-x O₃ (x = 0.1, 0.2, and 0.3), were synthesized via a sol-gel auto-combustion method using waste-derived precursors for iron and magnesium. High-purity praseodymium oxide (Pr₆O₁₁, Sigma Aldrich, ≥99.9%) was dissolved in concentrated nitric acid (HNO₃, 68%) under mild heating (70 °C) and continuous stirring for 30 min to obtain a clear praseodymium nitrate solution. Industrial red mud (rich in Fe₂O₃) was used as the iron source. 20 g of dried red mud powder was treated with 200 mL of 2.0 M hydrochloric acid (HCl) and heated at 80 °C for 4 h under constant magnetic stirring (600 rpm) in a reflux setup to enhance leaching efficiency. The leachate was filtered using Whatman No. 42 filter paper to remove silicates and residual solids. The resulting Fe³⁺-enriched solution was reacted with 30 mL of concentrated nitric acid (HNO₃) and evaporated at 60 °C on a rotary evaporator under reduced pressure (150 mbar) to obtain crystalline Fe(NO₃)₃·9H₂O (Bartel et al., 2019; Agrawal et al., 2004; Rai et al., 2012). Scrapped magnesium alloy (AZ31 type) was washed with acetone, dried, and ground into small flakes. 10 g of alloy was dissolved in 100 mL of 1.5 M nitric acid (HNO₃) at 60 °C for 3 h under stirring (500 rpm). After complete dissolution, the solution was filtered to remove insoluble residues like aluminum and zinc. The clear filtrate containing Mg²⁺ ions was subjected to slow evaporation at 60 °C to yield purified Mg(NO₃)₂·6H₂O crystals (Cui et al., 2019; Royani et al., 2018; Jiao et al., 2019). The schematic view of the sol-gel auto-combustion route shown in Figure 1. The stoichiometrically required amounts of Fe(NO₃)₃·9H₂O, Mg(NO₃)₂·6H₂O, and praseodymium nitrate were then dissolved in 100 mL of triply distilled deionized water and thoroughly mixed for 1 h at 60 °C. Citric acid (molar ratio 1:1 with total metal cations) was added as a chelating agent, followed by ethylene glycol (molar ratio 2:1 with citric acid) to promote the formation of a polymeric gel. The resulting solution was stirred continuously at 80 °C until it formed a homogeneous gel (∼3 h). The gel was then dried in a hot-air oven at 250 °C, where auto-combustion spontaneously occurred within 5–10 min, yielding a loose, voluminous ash-like powder. This as-prepared precursor was calcined at 700 °C for 6 h in a muffle furnace to remove organic residues and initiate perovskite phase formation. Finally, the powders were finely ground and sintered at 900 °C for 2 h to achieve better crystallinity and phase purity.

The structural, morphological, chemical, and gas sensing characteristics of PrMg _x Fe_1-x O₃ (x = 0.1, 0.2 & 0.3) were meticulously analyzed employing diverse methodologies. X-ray diffraction (XRD) was performed using a Bruker D8 X-ray diffractometer (Cu Kα radiation, λ = 1.54 Å, 40 kV, 30 mA) to verify the single-phase orthorhombic structure. Morphological examination was conducted utilizing Field Emission Scanning Electron Microscopy (FESEM) with a Carl Zeiss Supra 55, operating at 20 kV, while compositional analysis was executed using Energy Dispersive X-ray Spectroscopy (EDX) with an Oxford Instruments X-MaxN 51-XMX0004 detector. Surface area and porosity assessments were performed utilizing the Brunauer–Emmett–Teller (BET) technique with a Micromeritics ASAP 2020 analyzer, while pore size distribution was ascertained by the Barrett–Joyner–Halenda (BJH) method. An investigation of the oxidation state of elements of the prepared composition was conducted using X-ray Photoelectron Spectroscopy (XPS) using a Thermo Scientific K-Alpha + spectrometer, employing a monochromatic Al Kα source at 1486.6 eV. The gas detecting performance was assessed utilizing a Keithley 2,450 SourceMeter for real-time resistance measurements.

A pellet is self-standing and mechanically stable. Therefore, sintered sample pellets were taken, and two electrodes of silver paste were drawn on one of their surfaces with a fine brush. The paste was then allowed to dry at room temperature (Singh et al., 2024). Finally, the sensors were connected to a measurement system to monitor their electrical responses.

Response and efficiency measurement of a gas sensor involves quantifying its response to the presence and concentration of specific gases in the environment, primarily through changes in its electrical properties. Initially, the sensor’s baseline electrical characteristic, particularly resistance (R₀), was measured in a clean-air environment to serve as a reference. After the sensors were exposed to a known concentration of the target gas in a controlled gas chamber. The concentration of the target gas was measured by the formula (Jia et al., 2015): C = 22.4 × Φ × ρ × V 1 M ×   V 2

Where the Concentration of the target gas is denoted by C, measured in ppm, the volume of the gas fraction is mentioned as Φ, and the liquid density is given by the symbol ρ (g/mL). V₁ is the volume of the liquid measured in μL. The chamber volume is expressed as V₂ (L), and M is the molecular weight of the liquid in g/mol.

The interaction between the gas molecules and the sensing material caused changes in its resistance, denoted by Rg. This change was due to the adsorption and reaction of gas molecules on the sensor’s surface, altering its conductivity.

The sensor’s responses ‘S’ can be denoted by the formula: S = R 0 R g

Where R₀ is the resistance in clean air, and R_g is the resistance when exposed to the target gas (Bertocci et al., 2017; Moseley, 1992; Qin et al., 2021; Ma et al., 2018a; Shella et al., 2020).

The Goldschmidt tolerance factor (t) is a key parameter used to predict the stability and structural distortion of perovskite oxides. It is defined as: t = r A + r O 2 r B + r O where r A is the ionic radius of the A-site cation, r B is the ionic radius of the B-site cation, and r O is the ionic radius of oxygen. (Bui and Shin, 2023) Upon Mg²⁺ substitution in PrMg_xFe_1-xO₃, where Mg²⁺ (0.72 Å) replaces Fe³⁺ (0.645 Å), the average B-site ionic radius increases (Shannon and Prewitt, 1970). The calculated tolerance factors for x = 0.1, 0.2, and 0.3 are 0.951, 0.947, and 0.944, respectively. This progressive decrease suggests modifications in the lattice symmetry and octahedral connectivity. The crystal structure of pristine PrFeO₃, visualized using VESTA, illustrated in Figure 2, highlights the perovskite framework and FeO₆ octahedral arrangement. The decreasing tolerance factor with Mg substitution suggests an increasing degree of structural distortion due to enhanced FeO₆ octahedral tilting, further influencing the overall symmetry and stability of the material.

The X-ray diffraction (XRD) patterns confirm the formation of a single-phase orthorhombic perovskite structure for PrMg_xFe_1-xO₃ (x = 0.1, 0.2 & 0.3), with space group Pbnm (No. 62), consistent with PDF No. 01-074–1472 for PrFeO₃, shown in Figure 3, indicating the successful incorporation of Mg²⁺ into the perovskite lattice. The primary diffraction peaks at 2θ ≈ 22.7 °, 32.4 °, 46.9 °, and 57.9 ° correspond to the (101), (121), (202), and (123) planes, respectively. The crystallite size (D) was estimated using the Scherrer equation: D = K λ β Cos θ

Where K is the shape factor (0.9), λ is the X-ray wavelength (Cu Kα = 1.5406 Å), β is the full-width at half maximum (FWHM), and θ is the Bragg angle. The calculated crystallite sizes showed a decreasing trend with increasing Mg content: 13.21 nm (x = 0.1), 12.20 nm (x = 0.2), and 10.75 nm (x = 0.3), indicating that Mg substitution introduces structural distortions, thereby limiting crystal growth. To further analyze lattice imperfections, the dislocation density was determined using the Williamson–Smallman relation: δ = 1 D 2

The calculated dislocation densities were found to be 5.72 × 10¹⁵ m^-2, 6.72 × 10¹⁵ m^-2, and 8.65 × 10¹⁵ m^-2 for x = 0.1, 0.2, and 0.3, respectively, demonstrating a significant increase in lattice defects with higher Mg substitution. Additionally, the microstrain (ε) was determined using Williamson–Hall analysis: ε = β 4 ⁡ tan ⁡ θ

Where β is the FWHM in radians. The microstrain values also exhibited an increasing trend, from 7.39 × 10⁻³(x = 0.1) to 7.58 × 10⁻³(x = 0.2) and 8.03 × 10⁻³(x = 0.3), suggesting enhanced lattice distortions.

The observed reduction in crystallite size, increase in dislocation density, and rise in microstrain strongly indicate that Mg incorporation leads to higher structural disorder and defect density. These effects can be attributed to oxygen vacancies and strain-induced lattice deformations, which are further supported by the broadening of XRD peaks.

Field Emission Scanning Electron Microscopy (FESEM) was performed to analyse the grain morphology and porosity of PrMg _x Fe_1-x O₃ (x = 0.1, 0.2 & 0.3), as shown in Figure 4. The micrographs at 100kX magnification exhibit a polycrystalline structure characterized by distinct round grains, whilst the 50kX pictures elucidate differences in porosity. The grain size, analysed with ImageJ software, decreases from 143 nm (x = 0.1) to 134 nm (x = 0.2) and thereafter declines to 117 nm (x = 0.3), following the XRD crystallite size trend. A significant rise in porosity is noted with Mg doping; x = 0.1 displays a dense structure, while x = 0.2 and x = 0.3 reveal more pronounced voids and grain boundary separation. Energy Dispersive X-ray Spectroscopy (EDX) of PrMg _x Fe_1-x O₃ (x = 0.3), shown in Figure 5, verifies the existence of Pr, Fe, O, and Mg, with atomic ratios consistent with the anticipated stoichiometry. Elemental mapping reveals a consistent distribution of Mg and Fe, signifying homogenous doping within the perovskite structure.

The textural characteristics of PrMg _x Fe_1-x O₃ (x = 0.3) validate its mesoporous nature, as demonstrated by a Type IV nitrogen adsorption-desorption isotherm exhibiting a hysteresis loop illustrated in Figure 6. The material exhibits a Brunauer–Emmett–Teller (BET) surface area of around 6.24 m² g^-1, calculated using the multi-point BET method. The total pore volume is approximately 0.030 cm³ g^-1, Additionally, the Barrett–Joyner–Halenda (BJH) analysis reveals a pore surface area of ∼ 5.61 m²/g and a narrow pore size distribution centered ∼ 3.83 nm, indicating homogeneous and well-developed mesoporosity and promoting effective gas diffusion and adsorption-desorption processes. The interlinked pore network and augmented surface area improve the accessibility of adsorption sites, increasing the material’s structural efficacy.

X-ray Photoelectron Spectroscopy (XPS) was utilized to examine the oxidation states and defect structure of PrMg _x Fe_1-x O₃ (x = 0.1, 0.2 & 0.3), concentrating on the development of oxygen vacancies and Fe valence states. The survey spectrum depicted in Figure 7 validated the existence of Pr, Mg, Fe, and O, with no extraneous elements identified, hence confirming the phase purity of the synthesized materials. The O 1 s spectra were deconvoluted into three separate components: lattice oxygen (O_L, ∼529 eV), oxygen vacancies (O_V, ∼531–532 eV), and surface-adsorbed oxygen species (O_A, ∼532–533 eV), as seen in Figure 8. The quantitative analysis of these peaks offer information into the defect chemistry of the system summarised in Table 1. As the Mg concentration increased, the O_L peak area diminished from 31.70% (x = 0.1) to 25.86% (x = 0.2) and subsequently to 22.71% (x = 0.3), signifying a gradual reduction of lattice oxygen. Concurrently, the O_V peak area increased from 37.39% (x = 0.1) to 53.43% (x = 0.2) and 60.73% (x = 0.3), thus corroborating the augmented production of oxygen vacancies. The O_A peak area associated with chemisorbed oxygen species, demonstrated reduction with Mg doping from 30.91% (x = 0.1), 20.71% (x = 0.2) to 16.56% (x = 0.3), indicating that vacancy generation prevails over surface adsorption. The Fe 2p spectra corroborate these findings illustrated in Figure 9, indicating shifts in the peak locations of Fe³⁺ and Fe²⁺, as well as deviations in the Fe³⁺/Fe²⁺ ratio are detailed in Table 2. The Fe³⁺ 2p_3/2 peak changed from 711.90 eV (x = 0.1) to 711.73 eV (x = 0.2) and 711.26 eV (x = 0.3), whereas the Fe²⁺ 2p_3/2 peak shifted from 710.50 eV (x = 0.1) to 710.21 eV (x = 0.2) and 709.72 eV (x = 0.3), signifying an increase in Fe²⁺ content owing to charge compensation. The Fe³⁺/Fe²⁺ area ratio for Fe 2p_3/2 diminished from 2.80 (x = 0.1) to 1.57 (x = 0.2) and 1.05 (x = 0.3), corroborating the reduction of Fe³⁺ to Fe²⁺ linked to the creation of oxygen vacancies. The Fe³⁺-S satellite peaks exhibited a comparable pattern, with Fe³⁺-S (2p_3/2) decreasing from 719.92 eV (x = 0.1) to 719.31 eV (x = 0.2) and 717.39 eV (x = 0.3), indicating alterations in the electronic structure resulting from defect generation. The Fe³⁺-S (2p_1/2) peak was seen at x = 0.1 and x = 0.2, but it was absent in the x = 0.3 spectrum, indicating a substantial change in the electronic environment at elevated Mg concentrations. The observed shifts in Fe 2p binding energies and Fe³⁺/Fe²⁺ ratios are consistent with the increasing concentration of oxygen vacancies (as derived from the O 1 s spectra), confirming the charge compensation mechanism. The findings are further substantiated by XRD and FESEM investigations, indicating that increased Mg doping resulted in lattice distortions and alterations in grain size. The gradual rise in oxygen vacancies and Fe²⁺ concentration due to Mg substitution signifies substantial changes in the electronic structure, affecting charge transport characteristics across the compositions. The conversion of Fe³⁺ to Fe²⁺ and the increased concentration of oxygen vacancies strongly indicate n-type conduction characteristics in the produced materials.

The gas sensing properties of PrMg_xFe_1-xO₃ (x = 0.1, 0.2 & 0.3) perovskite materials were thoroughly examined to assess their potential for acetone detection. Response and recovery times, repeatability, long-term stability, operating temperature optimization, and selectivity against interfering gases were among the many factors that were examined. The sensing response of all three compositions was evaluated across a temperature range from 60 °C to 210 °C at an acetone concentration of 50 ppm, as seen in Figure 10a. The response was initially observed to increase with temperature, peaking at 150 °C, after which a decrease in sensitivity was recorded. The observed trend can be explained by the interplay between the kinetics of adsorption and desorption processes. At low temperatures, the restricted thermal energy impedes gas molecule interaction with the sensor surface, resulting in lowered responses. As the temperature rises to 150 °C, adequate energy is supplied to promote gas adsorption and interaction with surface oxygen species, thereby improving the sensing response. Increasing the temperature beyond 150 °C leads to significant thermal desorption of acetone molecules, which reduces the number of surface reactions and ultimately lowers the response. From the three compositions analysed, the sample with x = 0.3 demonstrated the most significant response at 150 °C. Following the determination of 150 °C as the ideal operating temperature, the gas sensing response of samples with x values of 0.1, 0.2, and 0.3 was methodically assessed across various acetone concentrations ranging from 5 to 500 ppm, as seen in Figure 10b. The observed response values exhibited a rising trend in relation to acetone concentration, thereby validating a conventional adsorption-controlled sensing mechanism. The sample x = 0.3 exhibited a response increase from 1.118 at 5 ppm to 1.394 at 500 ppm, indicating an improvement in sensitivity. In comparison, the compositions at x = 0.1 and x = 0.2 demonstrated reduced responses across all concentrations, highlighting that the x = 0.3 composition shows the highest efficiency in charge transfer and surface reactivity. The enhanced performance observed at x = 0.3 can be attributed to the increased density of oxygen vacancies, which act as active sites for acetone adsorption, thereby promoting swift electron transfer. The dynamics of response and recovery for the x = 0.3 composition were analysed at 150 °C to 50 ppm acetone as seen in Figure 10c, showcasing its enhanced sensing performance. The observed response time was found to be 33 s, while the recovery time was 20 s. The rapid response and recovery observed indicate the effective kinetics of adsorption and desorption of acetone molecules on the sensor surface. In comparison, the samples with x = 0.1 and x = 0.2 demonstrated slower response times of 47 s and 36 s, along with longer recovery times of 41 s and 30 s, respectively. The rapid response and recovery of x = 0.3 highlight its significant surface reactivity and refined defect structure, positioning it as a stable candidate for acetone detection.

A three-cycle test was conducted on the x = 0.3 composition at 50 ppm acetone to assess the consistency of the sensing performance shown in Figure 10d. The fluctuations in resistance across several cycles demonstrated a reliable response and recovery, highlighting the reproducibility and stability of the sensing properties. Additionally, the extended stability of the x = 0.3 sensor, shown in Figure 10e, was evaluated over a period of 30 days, with data collected at five-day intervals through continuous observation of its response to 50 ppm acetone. The sensor exhibited a consistent response throughout the duration of the study, demonstrating its reliability and endurance for extended use. The x = 0.3 sensor’s response to 50 ppm acetone at 150 °C was evaluated in relation to typical interfering gases such as ethanol, propanol, and dimethylformamide (DMF) as seen in Figure 10f. The findings demonstrated that the reaction to acetone (1.215) was markedly greater than that of ethanol (1.023), propanol (1.014), and DMF (1.084). The improved selectivity for acetone arises from the more robust molecular interactions and greater binding affinity of acetone molecules with the sensor surface, in contrast to other volatile organic compounds.

The gas sensing mechanism of n-type semiconductors, such as Mg-doped PrFeO₃, involves an acetone-sensing mechanism that involves a series of adsorption, charge transfer, and chemical reactions. The parent perovskite PrFeO₃ is intrinsically a p-type semiconductor due to cation vacancies that serve as electron donors and foster a hole-rich conduction mechanism. However, when Mg²⁺ is substituted into the parent perovskite to form PrMg _x Fe_1-x O₃, it starts behaving differently than it normally would. Doping Mg²⁺(divalent ion) at the Fe³⁺(trivalent) site in PrFeO₃ creates a charge imbalance. To maintain charge neutrality, the lattice compensates by generating oxygen vacancies (V_O^••) as per the defect reaction: MgO  → M g Fe ′   +   V _ O · · + 12 O 2

The presence of oxygen vacancies facilitates the release of electrons into the conduction band, resulting in an elevated concentration of free electrons, leading to a transition from p-type to n-type conductivity. The sensing mechanism illustrated in Figure 11 begins with the adsorption of atmospheric oxygen molecules (O₂) onto the surface of PrMg _x Fe_1-x O₃. The oxygen molecules ionize by capturing electrons from the conduction band, resulting in the formation of chemisorbed oxygen species (O₂ ⁻, O⁻, O^2-). The operational temperature determines which oxygen species are going to get involved. The following are the conditions involved: O 2 g a s → O 2 a d s O 2 a d s + e − → O 2 −   a d s   low temperatures , <   150   °   C O 2 − a d s + e − → 2 O − a d s   moderate temperatures ,   150 − 300   °   C O 2 − a d s + e − → O 2 − a d s   high temperatures , > 300   ° C

These chemisorbed species induce a surface depletion layer through electron removal, leading to an increase in resistance in the n-type semiconductor (Bartel et al., 2019).

Upon interaction with the sensor surface, acetone (CH₃COCH₃) molecules engage in adsorption and subsequently participate in oxidation reactions with the chemisorbed oxygen species. The process of oxidation facilitates the liberation of trapped electrons into the conduction band. The responses may be articulated in the following manner: C H 3 H 3 C O C H 3 g a s → C H 3 H 3 C O C H 3 a d s C H 3 H 3 ⁢ C O C H 3 ⁢ a d s + 8 O − ⁢ a d s → 3 C O 2 ⁢ g a s + 3 H 2 ⁢ O ⁡ g a s + 8 e − ⁢ low moderate temperatures C H 3 H 3 ⁢ C O C H 3 ⁢ a d s + 4 O 2 − ⁢ a d s → 3 C O 2 ⁢ g a s + 3 H 2 ⁢ O ⁡ g a s + 8 e − ⁢ h i g h e r temperatures

The aforementioned reactions lead to a reduction in the depletion layer, consequently resulting in a decreased resistance of the sensor. The alteration in the sensor’s resistance is directly related to the concentration of the target gas. Afterwards, the elimination of acetone, atmospheric oxygen re-adsorbs onto the surface, facilitating the extraction of electrons and reinstating the depletion layer, thereby reverting the resistance to its original condition, a comparison mentioned in Table 3. The reversible modulation of resistance facilitates the effective detection of reducing gases, a behavior that is emblematic of n-type semiconducting materials.