From Selective Adsorption to Photocatalytic Removal: Complementary Cobalt Oxides for Pollutant-Oriented Treatment of Dye Mixtures

Abstract

The development of photocatalysts for water treatment generally focuses on maximizing removal efficiency, whereas the role of material–pollutant interactions in determining pollutant-dependent removal behaviour remains comparatively underexplored. Herein, pristine Co₃O₄ and a CuCo₂O₄-based mixed oxide were comparatively investigated with two chemically distinct anionic dyes, methyl orange (MO) and Congo red (CR). Physicochemical characterization revealed differences between the materials, which were associated with distinct removal behaviours. Under solar irradiation, Co₃O₄ achieved 94% MO removal within 180 min at acidic pH, whereas CR removal showed a stronger adsorption contribution. Conversely, the CuCo₂O₄-based oxide exhibited near-complete CR removal through adsorption but limited photocatalytic contribution toward MO. After regeneration, substantial recovery of both photocatalytic and adsorption performance was observed. These complementary behaviours were exploited in sequential treatment of a binary MO/CR mixture, combining CR removal by adsorption over the CuCo₂O₄-based material with photocatalytic removal of the residual MO by Co₃O₄. A simultaneous one-pot configuration achieved comparably high final removal of both dyes, showing that temporal separation was not required for high overall removal under the investigated conditions. Overall, matching material functionalities to pollutant characteristics enables pollutant-oriented treatment strategies integrating complementary adsorption and photocatalytic functions.

1. Introduction

The continuous discharge of dye-containing wastewater from the textile, leather, paper, and printing industries represents a persistent environmental concern due to the high chemical stability, poor biodegradability, and potential toxicity of many synthetic dyes [1,2]. Besides their direct ecological impact, these compounds reduce light penetration in aquatic environments and may generate hazardous transformation products, posing additional risks to ecosystems and human health [3,4,5]. Industrial effluents may also contain dyes with different molecular structures, charge distributions, and functional groups, making their simultaneous removal more challenging than the treatment of individual model pollutants [6,7,8,9]. Among the wide variety of azo dyes released into industrial effluents, methyl orange (MO) and Congo red (CR) are frequently adopted as model contaminants because they exhibit different molecular structures and surface-interaction behaviours despite both being classified as anionic dyes [10,11]. MO is a monoazo dye (MW ≈ 327 g·mol⁻¹) containing a single sulfonate group and exhibiting a pKa of ca. 3.4, resulting in pronounced pH-dependent changes in molecular structure and electrostatic interactions with catalyst surfaces [12,13]. In contrast, CR is a larger diazo molecule (MW ≈ 697 g·mol⁻¹) containing two azo linkages, two sulfonate groups, and an extended aromatic backbone, which generally favours stronger adsorption through multiple electrostatic and π-related interactions [14,15]. These differences can affect adsorption affinity, photocatalytic reactivity, and competition for surface active sites [16].

Nevertheless, most photocatalytic studies still investigate single model pollutants, while treatment strategies specifically addressing mixed dye systems remain comparatively scarce [8,17].

Among the various technologies proposed for wastewater remediation, heterogeneous photocatalysis has emerged as a promising advanced oxidation process for the removal of organic contaminants under relatively mild operating conditions using solar or artificial light [17,18]. Considerable efforts have therefore been devoted to developing visible-light-responsive photocatalysts through crystal engineering, heterojunction construction, morphology control, elemental doping, and defect engineering [19,20]. Representative semiconductor families include TiO₂-based photocatalysts [21,22,23], graphitic carbon nitride [24,25], ZnO [26,27], WO₃ [28,29,30], Bi-based oxyhalides [31,32,33], spinel ferrites [34,35] and numerous heterostructured systems [36,37]. These strategies can improve light absorption, charge separation and overall photocatalytic performance [38], although the resulting activity remains strongly dependent on the target pollutant, solution chemistry, pH, and operating conditions [8,17].

Alongside photocatalysis, adsorption represents one of the most widely applied technologies for dye removal owing to its operational simplicity, relatively low energy demand and high removal efficiency [39,40,41]. Activated carbon, biochar, zeolites, clays and porous carbonaceous materials have been extensively investigated because of their large surface area and high adsorption capacity [42,43,44,45]. However, adsorption transfers contaminants from the aqueous to the solid phase rather than destroying them, making regeneration, disposal, and long-term stability important considerations [17,46,47]. Importantly, adsorption and photocatalysis need not be regarded as competing processes, since both are influenced by the interfacial interactions established between the material and the target pollutant [48].

In heterogeneous photocatalysis, pollutant–surface interactions can influence the overall removal process by controlling pollutant proximity to photogenerated charge carriers and reactive species [17,49]. Surface hydroxylation, surface charge, oxygen vacancies, exposed crystal facets, and the chemical nature of accessible sites can affect these interactions [50,51], while pollutant charge, molecular size, aromatic structure, and functional groups determine surface affinity and may influence subsequent photocatalytic oxidation [8]. Thus, similar concentration decreases may arise from different contributions of adsorption and photocatalytic conversion, which cannot be resolved when material performance is evaluated solely in terms of overall removal efficiency.

This distinction becomes particularly relevant in contaminant mixtures, where chemically different molecules coexist and compete for accessible surface sites [52,53]. Yet, catalyst modification is commonly pursued primarily to enhance overall activity, whereas comparatively less attention has been devoted to whether related materials can develop distinct pollutant-dependent functions. For Co-based materials in particular, systematic relationships between physicochemical properties, pollutant chemistry, and the relative contributions of adsorption and photocatalytic removal remain insufficiently explored [54,55,56,57]. Such understanding could support an alternative design strategy in which materials with complementary pollutant-dependent functions are deliberately combined in sequential treatment schemes.

Among visible-light-active transition metal oxides, spinel Co₃O₄ has attracted considerable attention owing to its narrow band gap, favourable redox properties arising from the coexistence of Co²⁺ and Co³⁺ species, chemical stability, low cost, and earth abundance [54,55,56,57]. These characteristics have promoted its application in environmental photocatalysis, electrocatalysis, energy conversion, and heterogeneous oxidation processes [58,59,60]. Its mixed-valence chemistry also provides a versatile platform for compositional modification and for investigating how changes in material properties affect pollutant-dependent removal behaviour [61]. Partial substitution with other transition metals has been explored to tailor the physicochemical properties of cobalt-based spinels [62,63]. In particular, CuCo₂O₄-based mixed oxides are promising because compositional modification can alter phase composition, electronic and surface properties, and charge-transfer characteristics, potentially affecting both photocatalytic activity and pollutant–surface interactions [64].

On this basis, the present work comparatively investigates pristine Co₃O₄ and a CuCo₂O₄-based mixed oxide to examine how their different compositions and physicochemical properties are associated with pollutant-dependent removal behaviour. Structural, morphological, textural, optical, electronic, and surface properties are correlated with the removal of two structurally distinct anionic dyes, MO and CR. Independent adsorption and photocatalytic experiments are used to distinguish the relative contributions of the two removal pathways, and the complementary behaviours identified for the two materials are subsequently exploited in the sequential treatment of a binary dye mixture. A simultaneous one-pot configuration is also investigated to directly compare the combined use of the two materials with their sequential application. This work examines how structurally related Co-based oxides can provide complementary functions that may be translated into pollutant-oriented treatment strategies based on different operational configurations.

2. Results and Discussion

In this work, pristine Co₃O₄ and a CuCo₂O₄-based material were investigated to elucidate how differences in their physicochemical properties correlate with the removal pathways of anionic dyes. The materials were first characterized in terms of crystal structure, morphology, texture, surface chemistry, and optical and electronic properties, and their removal behaviour toward MO and CR was subsequently evaluated under different operating conditions. Finally, their complementary functions were evaluated in a binary dye mixture using sequential and simultaneous one-pot treatment configurations.

2.1. Physicochemical Properties of the Cobalt-Based Materials

The main physicochemical properties of the synthesized materials are summarized in Table 1, whereas the corresponding structural and morphological characterization is reported in Figure 1.

Figure 1. (a) XRD patterns of the Co₃O₄ and CuCo₂O₄-based samples compared to the related reference diffraction patterns (● and ◊ indicate reflections assigned to CuCo₂O₄ and CuO, respectively). (b,c) FESEM images of the Co₃O₄ and CuCo₂O₄-based samples, respectively. (d) N₂ adsorption/desorption isotherms at −196 °C of the Co₃O₄ and CuCo₂O₄-based samples.

The phase composition of the synthesized materials was first investigated by X-ray diffraction (XRD), and the corresponding diffraction patterns are shown in Figure 1a. The diffraction pattern of pristine Co₃O₄ can be indexed to the cubic spinel structure (PDF 00-042-1467), with the characteristic reflections at 2θ ≈ 19.0°, 31.3°, 36.9°, 44.8°, 59.4°, and 65.2°, corresponding to the (111), (220), (311), (400), (511), and (440) crystallographic planes, respectively. No additional diffraction peaks ascribable to secondary cobalt oxide phases were detected, indicating the successful formation of phase-pure spinel Co₃O₄, in agreement with the literature [65,66]. For the CuCo₂O₄-based material, the diffraction profile shows the characteristic reflections of the Cu-Co spinel (PDF 00-025-0270), together with additional peaks centered at approximately 2θ ≈ 32.5°, 35.5°, 38.7°, 48.8°, 58.3°, 61.5°, and 66.2°, which can be assigned to monoclinic CuO (PDF 00-048-1548) [67]. Rietveld refinement of the XRD patterns indicates an approximately 70:30 wt.% ratio between the Cu-Co spinel and CuO phases, respectively, with comparable phase ratios obtained across the independently synthetized batches. To further assess whether the optical response of the CuCo₂O₄-based material could be reproduced by a simple physical combination of cobalt and copper oxides, its diffuse reflectance spectrum was compared with those of pristine Co₃O₄ and CuO, as well as with that of a physical Co₃O₄/CuO mixture prepared at a 70:30 wt.% ratio (Figure S1a). The spectrum of the CuCo₂O₄-based material differed from that of the corresponding physical mixture, indicating that its optical response cannot be reproduced by a simple physical combination of Co₃O₄ and CuO at the investigated ratio. Accordingly, the material is hereafter referred to as CuCo₂O₄-based rather than as phase-pure CuCo₂O₄. The coexistence of CuO and CuCo₂O₄ has previously been reported for Cu–Co mixed oxides prepared by co-precipitation followed by calcination. The phase composition of the Cu–Co–O system depends on both Cu content and thermal treatment, with increasing Cu content reported to favour CuO coexistence with the spinel phase and to affect its thermal stability [68,69,70].

FTIR-ATR spectroscopy (Figure S1b,c) indicates that both materials display the characteristic vibrational features of the spinel structure [71]. The spectrum of pristine Co₃O₄ shows bands at approximately 655, 550 and 390 cm⁻¹, assigned to ν₁, ν₂, and ν₃ modes (T_1u symmetry), respectively, corresponding to Co–O stretching vibrations [72]. Additionally, weak shoulders are observed at around 675 and 585 cm⁻¹, consistent with the TO–LO splitting reported for nanostructured spinel oxides [73]. In the spectrum of the CuCo₂O₄-based material, similar spinel-related vibrational features are observed, although slightly shifted toward lower wavenumbers, indicating a different local chemical environment of the Cu-containing spinel [72]. Notably, the ν₂ band exhibits a pronounced asymmetry, with an intense shoulder at approximately 480 cm⁻¹, attributable to the coexistence of the CuO secondary phase [74], in agreement with XRD data.

Analysis of the entire spectral range reveals further differences between the samples. The spectrum of Co₃O₄ exhibits broad bands centered at around 3350 and 1630 cm⁻¹ attributed to the stretching and bending vibrations of adsorbed water molecules interacting via hydrogen bonding. Additionally, the multiple bands between 1600 and 1300 cm⁻¹ can be associated with surface oxygen-containing species [75]. These bands are absent in the spectrum of the CuCo₂O₄-based material, which instead displays signals attributable to hydrocarbon species (C–H stretching modes around 2900 cm⁻¹ and their corresponding bending modes). Weak signals at 3630 and 3560 cm⁻¹ may be assigned to terminal hydroxyl species.

Overall, XRD and ATR confirm the spinel nature of pristine Co₃O₄ and the coexistence of Cu–Co spinel and crystalline CuO in the CuCo₂O₄-based material, together with differences in their local and surface chemical environments.

The morphology of the synthesized materials was subsequently investigated by FESEM (Figure 1b,c, Figures S2 and S3). Pristine Co₃O₄ consists of loosely aggregated nanoparticles forming a relatively homogeneous porous network (Figure 1b). At higher magnification, these aggregates are composed of nearly spherical or slightly elongated (i.e., rod-like) primary particles a few tens of nanometers in size, consistent with the nanoscale morphology typically reported for precipitation-derived Co₃O₄ spinels [76]. The CuCo₂O₄-based sample displays a different particle morphology, characterized by larger, more compact agglomerates of interconnected flake-like crystallites and a rougher, more heterogeneous surface texture (Figure 1c). The larger flakes are elongated in one preferential direction and are very thin (average thickness ca. 10 nm). Numerous small, approximately spherical nanoparticles (average diameter of ca. 20–25 nm) are also observed (Figure S3a). These observations show that the two materials differ substantially in particle morphology and aggregation, consistent with morphological variations previously reported for Cu–Co mixed oxides prepared by co-precipitation [70].

The corresponding EDS elemental maps (Figures S2 and S3) show a homogeneous distribution of the constituent elements at the investigated scale. Co and O are uniformly distributed in pristine Co₃O₄, whereas Co, Cu and O exhibit no evident elemental segregation in the CuCo₂O₄-based sample. No distinct Cu-rich domains are resolved, indicating an overall homogeneous Cu distribution at the FESEM-EDS scale, despite the coexistence of Cu–Co spinel and crystalline CuO established by XRD.

Following the structural and morphological characterization, the textural properties of the two materials were investigated by N₂ adsorption–desorption measurements (Figure 1d). Both samples exhibit type IV isotherms [77] with a hysteresis loop at high relative pressures, indicative of mesoporous solids formed by the aggregation of nanosized particles. Such behaviour is consistent with the morphologies observed by FESEM and is commonly reported for precipitation-derived cobalt oxide materials [70].

The surface area value increases from 9 m²·g⁻¹ for pristine Co₃O₄ to 17 m²·g⁻¹ for the CuCo₂O₄-based sample (Table 1), which also exhibits a larger adsorbed nitrogen volume over the entire relative pressure range (Figure 1d).

The pore size distribution, obtained by the BJH method (Figure S4), further confirms that both materials are predominantly mesoporous, with most of the pore volume associated with pores between 2 and 5 nm. The CuCo₂O₄-based sample exhibits a broader pore size distribution and a greater contribution from larger mesopores, in agreement with its different particle morphology and aggregation.

Overall, despite their relatively low surface areas (Table 1), the two materials exhibit distinct textural properties, with the CuCo₂O₄-based sample showing an approximately twofold higher surface area and a broader pore-size distribution than pristine Co₃O₄.

These differences may affect the accessibility of surface sites and should therefore be considered when interpreting their adsorption behaviour.

Table 1. Main properties of Co₃O₄ and CuCo₂O₄-based samples.

Code	Surface Area a (m2·g−1)	Surface Composition (Atomic %)				pHpzc b	Optical Transitions c (eV)
		C	O	Co	Cu
Co3O4	9	51.7	33.8	14.5	-	5.8	1.18, 1.90
CuCo2O4-based	17	72.2	25.5	1.70	0.6	6.4	~1.70 (weak), 1.90

^a by BET equation (two parameters); ^b from procedure described by Ref. [78]; ^c values estimated from Tauc plots. The CuCo₂O₄-based sample exhibits an additional absorption feature at lower energy (~1.70 eV).

Table 1. Main properties of Co₃O₄ and CuCo₂O₄-based samples.

Code	Surface Area a (m2·g−1)	Surface Composition (Atomic %)				pHpzc b	Optical Transitions c (eV)
		C	O	Co	Cu
Co3O4	9	51.7	33.8	14.5	-	5.8	1.18, 1.90
CuCo2O4-based	17	72.2	25.5	1.70	0.6	6.4	~1.70 (weak), 1.90

^a by BET equation (two parameters); ^b from procedure described by Ref. [78]; ^c values estimated from Tauc plots. The CuCo₂O₄-based sample exhibits an additional absorption feature at lower energy (~1.70 eV).

The surface charge properties of the two materials were then evaluated by determining their point of zero charge (pH_pzc), and the corresponding results are reported in Figure 2a,b. Both materials exhibit pH_pzc values within the mildly acidic region, equal to 5.8 for pristine Co₃O₄ and 6.4 for the CuCo₂O₄-based sample (Table 1). The slightly higher pH_pzc of the CuCo₂O₄-based material indicates a shift in its acid–base surface properties compared with pristine Co₃O₄, consistent with differences in the nature and distribution of surface functional groups. Similar shifts toward higher pH_pzc values have been reported for Cu-containing cobalt spinels and are generally associated with changes in surface composition and chemical environment [79].

Figure 2. (a,b) Point of zero charge of the Co₃O₄ and CuCo₂O₄-based samples, respectively. (c) Nyquist plots of Co₃O₄ electrode recorded in Na₂SO₄ 0.1 M. Experimental conditions: Co₃O₄-ITO (0.25 cm²) as working electrode, Pt (1 cm²) as counter electrode, SCE as reference electrode. Marks indicate experimental data, while lines represent the corresponding fitted curves.

Under the investigated conditions, these pH_pzc values determine the prevailing surface-charge regime: below the pH_pzc, the material surface is predominantly positively charged, whereas above it the surface progressively acquires a negative charge [45,80]. Since both MO and CR are anionic azo dyes [8,81], electrostatic attraction toward positively charged surface sites may therefore contribute to their interaction with both materials under acidic conditions. However, the surface-charge regime alone cannot predict either the extent of dye uptake or the selectivity between structurally different anionic molecules.

The optical properties of the synthesized materials were investigated by UV–vis diffuse reflectance spectroscopy (UV–DRS), and the corresponding Tauc plots are reported in Figure S5. The estimated optical transition energies are summarized in Table 1. Pristine Co₃O₄ exhibits two characteristic optical transitions at ca. 1.18 and 1.90 eV, according to the literature [82,83]. These transitions are generally associated with charge-transfer processes involving Co²⁺ and Co³⁺ cations in the spinel lattice [82,83]. For the CuCo₂O₄-based material, an optical transition at approximately 1.90 eV is retained, together with an additional contribution at approximately 1.70 eV (Figure S5b, Table 1). The latter may be related to the CuO phase identified by XRD, which typically exhibits a narrower optical band gap and may contribute to visible-light absorption [82,83,84,85].

Therefore, the UV–DRS results are consistent with the multiphase composition identified by XRD, with an additional lower-energy contribution observed for the CuCo₂O₄-based material.

The recombination behaviour of photogenerated charge carriers was investigated by photoluminescence (PL) spectroscopy, and the corresponding spectra are reported in Figure S6. Independently of the excitation wavelength (290 and 340 nm), both materials exhibit similar emission profiles, with no additional bands or significant spectral shifts. A slightly higher emission intensity is observed for the CuCo₂O₄-based sample over the investigated spectral range. Since PL intensity is commonly used as a qualitative indicator of radiative electron–hole recombination [86], these results suggest broadly comparable radiative recombination behaviour in the two materials, with only a limited difference under the investigated conditions. Accordingly, the PL results do not indicate a sufficiently pronounced difference in radiative charge-carrier recombination to account, by themselves, for the different photocatalytic behaviours subsequently observed for the two materials. The PL data are therefore considered as complementary information on radiative recombination rather than as a direct predictor of photocatalytic performance.

The surface chemical composition and oxidation states were investigated by X-ray photoelectron spectroscopy (XPS). The corresponding survey and high-resolution spectra are reported in Figure 3 for pristine Co₃O₄ and Figure S7 for the CuCo₂O₄-based sample. The survey spectra (Figure 3a and Figure S7a) confirm the presence of the expected elements, C, O, Co, and Cu in the CuCo₂O₄-based sample, with no additional elements detected within the sensitivity of the analysis. The quantitative surface atomic composition (Table 1) confirms the presence of Cu at the outermost surface of the CuCo₂O₄-based material, with Co remaining the more abundant transition metal.

Figure 3. Survey (a) and HR XP spectra of C 1s (b), O 1s (c), and Co 2p (d) for pristine Co₃O₄. Deconvolutions in sub-bands and the resulting fitted spectra (black lines) are also reported.

The high-resolution (HR) C 1s spectra (Figure 3b and Figure S7b) are mainly dominated by the adventitious carbon component centered at 284.6 eV, which was used as the reference for charge correction and is commonly assigned to C–C/C–H species [87,88]. For pristine Co₃O₄, the second contribution centered at ca. 287.3 eV can be attributed to oxygenated carbon species, such as carbonyl or O–C–O groups, originating from surface-adsorbed contaminants [87]. In the CuCo₂O₄-based sample, the C 1s envelope requires an additional component at about 286.0 eV, commonly assigned to C–O species, together with the higher-binding-energy contribution located at 287.3 eV [89,90,91]. These contributions are attributed to adventitious surface carbon species adsorbed upon air exposure.

The HR O 1s spectra (Figure 3c and Figure S7c) can be deconvoluted into three contributions. The main component located at 529.5–530.0 eV is assigned to lattice oxygen (O²⁻) belonging to the oxide framework, whereas the peaks centered around 531.0–531.7 eV and 532.4–532.6 eV are commonly associated with surface hydroxyl groups, oxygen-deficient environments and chemisorbed oxygen species [87]. These contributions indicate differences in the surface oxygen environment of the two cobalt-based materials and are consistent with their different surface chemical characteristics.

The HR Co 2p spectra (Figure 3d and Figure S7d) exhibit the characteristic spin–orbit doublet together with well-defined shake-up satellite features, confirming the mixed-valence nature of cobalt in both materials [92]. The Co 2p_3/2 envelope can be deconvoluted into components centered at ca. 779.4–779.9 eV and 780.9–781.8 eV, assigned to Co³⁺ and Co²⁺ species [92], respectively, whereas the corresponding Co 2p_1/2 contributions are detected at ca. 794.1–794.3 eV and 796.1–797.1 eV. The satellite peaks observed at higher binding energies (around 789–790 eV and 805–806 eV) are typical of Co₃O₄-type spinel electronic structures containing both Co²⁺ and Co³⁺ species [87,89,92]. No pronounced shifts in the binding energies of the Co components are observed between the two materials, indicating broadly comparable Co²⁺/Co³⁺ chemical environments within the resolution of the analysis.

For the CuCo₂O₄-based material, the HR Cu 2p spectrum (Figure S7e) displays the characteristic Cu 2p_3/2 and Cu 2p_1/2 peaks centered at ca. 933.1 eV and 952.9 eV, respectively, together with shake-up satellite features located at ca. 942.7 eV and 961.8 eV. These satellite features are characteristic of Cu²⁺ species [93,94,95]. This is consistent with the XRD evidence of crystalline CuO coexisting with the Cu–Co spinel phase, although the Cu 2p spectrum alone does not allow the different Cu-containing phases to be quantitatively distinguished.

Overall, XPS analysis reveals broadly comparable Co²⁺/Co³⁺ chemical environments in the two materials, while confirming surface Cu²⁺ species and a different surface oxygen environment in the CuCo₂O₄-based sample. Together with the pH_pzc and ATR results, these observations establish distinct surface chemical characteristics for the two materials.

Since pristine Co₃O₄ showed the clearest photocatalytic contribution to MO removal, electrochemical impedance spectroscopy (EIS) was performed exclusively on this material under dark and illuminated conditions to further investigate its photoinduced charge-transfer behaviour. The Nyquist plots (Figure 2c) exhibit a depressed semicircle followed by a low-frequency contribution, indicating that the impedance response deviates from ideal RC behaviour and originates from distributed electrochemical processes occurring at the oxide/electrolyte interface [96]. This interpretation is further supported by the corresponding Bode plots (Figure S8), which display broad phase maxima rather than sharp peaks, consistent with a distribution of relaxation times commonly associated with heterogeneous and non-ideal electrode interfaces [97].

To quantitatively analyze the impedance response, the experimental spectra were fitted using an equivalent circuit consisting of two R–CPE branches connected in series (inset of Figure 2c), consistent with two distributed relaxation processes. The corresponding fitting parameters are summarized in Table S1. The higher-frequency contribution is assigned to an interfacial charge-transfer process, whereas the lower-frequency contribution is associated with slower interfacial polarization and transport phenomena. Constant phase elements (CPEs) account for the non-ideal capacitive response associated with surface heterogeneity and distributed electrochemical sites [98]. Under illumination, the resistance associated with the high-frequency process decreases from 10.54 to 8.77 kΩ, consistent with facilitated interfacial charge transfer upon photoexcitation [99]. In contrast, the CPE exponent of the high-frequency process remains nearly unchanged (n₁ = 0.820 in the dark and 0.828 under illumination), suggesting that illumination does not appreciably alter the degree of non-ideal capacitive behaviour associated with this process. The low-frequency process exhibits a CPE exponent close to 0.5 under both conditions, indicating a strongly distributed response with a significant transport-related contribution [100]. The resistance associated with this low-frequency process could not be reliably resolved under illumination and was therefore not considered for quantitative comparison.

Overall, the characterization establishes clear differences between the two materials in phase composition, morphology, texture, surface properties, and optical response. For pristine Co₃O₄, EIS additionally shows a decrease in interfacial charge-transfer resistance under illumination, supporting a photoresponsive interfacial behaviour.

2.2. Influence of pH on Methyl Orange Removal over Co₃O₄

The photocatalytic performance of pristine Co₃O₄ toward MO was first investigated under different pH conditions to assess the influence of interfacial interactions on the removal process. The pH-dependent surface charge is expected to play an important role because MO is predominantly present in an anionic form over much of the investigated pH range. Since the pH_pzc of Co₃O₄ was determined to be 5.8 (Table 1, Figure 2a), changes in solution pH modify the electrostatic interactions between the catalyst surface and the dye [7]. In addition, MO undergoes acid–base speciation around its pKa (~3.4), which may also affect its interaction with the catalyst under the most acidic conditions investigated.

Before discussing the removal results, it should be clarified that throughout this study the term “removal” refers to the decrease in the analytical signal of the parent dye measured by UV–vis spectroscopy and, where applicable, supported by HPLC analysis. Since TOC measurements could not be reliably quantified at the investigated dye concentration, the present data do not allow chromophore bleaching, transformation of the parent dye, and mineralization to be quantitatively distinguished. Accordingly, removal observed under irradiation should not be interpreted as direct evidence of complete degradation or mineralization. The terms ‘photocatalytic removal’ and ‘photocatalytic contribution’ are used here in an operational sense to describe the additional dye removal observed in the presence of the catalyst under irradiation relative to the corresponding dark and photolysis controls, rather than to imply complete degradation or mineralization.

Figure 4a reports the temporal evolution of MO concentration during photolysis, adsorption and photocatalysis at pH 3. Under dark conditions (red line), Co₃O₄ removes ca. 33% MO during the 30 min dark equilibration period. Under the adopted experimental conditions (10 mg·L⁻¹ MO and 0.5 g·L⁻¹ catalyst), complete dye removal would correspond to an apparent uptake of 20 mg·g⁻¹; thus, the 33% dark removal corresponds to ca. 6.6 mg·g⁻¹. These values refer specifically to the investigated concentration and solid loading and should not be interpreted as equilibrium or maximum adsorption capacities. Photolysis results in only ca. 16% removal after 180 min of irradiation (black line). When Co₃O₄ is irradiated under simulated solar light (green line), MO concentration progressively decreases after the dark equilibration period, reaching 94% removal after 180 min of irradiation. Thus, the removal observed under irradiation substantially exceeds that attributable to dark adsorption or direct photolysis under the investigated conditions.

Figure 4. (a) MO abatement in UW at initial dye solution pH = 3 on Co₃O₄ by adsorption, and photocatalysis. (b) Effect of initial solution pH on the MO abatement by photolysis, adsorption and photocatalysis on Co₃O₄ at the end of the tests (210 min). Experimental conditions: MO concentration = 10 mg·L⁻¹; catalyst dosage (when used) = 0.5 g·L⁻¹; irradiance (when used) = 35 W·m⁻².

The corresponding pseudo-first-order kinetic fitting (Figure S9) shows good linearity (R² = 0.997), with a reaction rate constant (k) of 0.011 min⁻¹. This kinetic parameter describes the disappearance of MO under the adopted analytical conditions and should not be interpreted as a mineralization rate constant.

Qualitative HPLC analysis did not reveal additional chromatographic peaks of appreciable intensity; however, this observation does not exclude the formation of transformation products (TPs) outside the detection conditions of the adopted method. Moreover, total organic carbon (TOC) measurements could not be reliably performed because the initial MO concentration (10 mg·L⁻¹) was below the quantification limit of the available instrument. Accordingly, the present results demonstrate MO removal but do not allow its mineralization or complete degradation to be established.

The effect of solution pH on the MO removal is summarized in Figure 4b. Photocatalytic removal is highest at pH 3 (94% after 180 min of irradiation), while substantially lower values are obtained at the natural solution pH (pH ≈ 5.5, ca 23%). At pH 11, no appreciable MO removal was observed under the investigated conditions. Dark removal follows the same general pH dependence decreasing from 33% at pH 3 to ca. 21% at pH ≈ 5.5 and becoming negligible at pH 11. This behaviour is consistent with the experimentally determined pH_pzc of Co₃O₄. Below the pH_pzc, the positively charged oxide surface favours interactions with negatively charged MO species, whereas these attractive interactions progressively weaken as the pH approaches the pH_pzc and electrostatic repulsion is expected above it. Similar pH-dependent behaviour has been reported for anionic dyes interacting with metal oxide photocatalysts [101,102,103]. At pH 3, however, the proximity to the pKa of MO should also be considered, and the observed behaviour therefore cannot be attributed exclusively to catalyst surface charge.

Importantly, the pH dependence of dark removal and photocatalytic removal is not identical. Although favourable dye–surface interactions under acidic conditions can increase the amount of MO associated with the catalyst surface, the substantially larger decrease observed under irradiation indicates an additional photocatalytic contribution. Enhanced interfacial availability of MO may facilitate its interaction with photogenerated charge carriers and/or short-lived reactive species. The relative involvement of these species was therefore further investigated by scavenging experiments under the acidic conditions showing the highest MO removal. The proposed interpretation of the pH-dependent interfacial behaviour is summarized in Figure 5. Under acidic conditions (pH < pH_pzc), favourable dye–surface interactions promote MO–Co₃O₄ contact, whereas these interactions weaken close to the pH_pzc and become unfavourable under alkaline conditions. MO speciation may additionally contribute at pH 3. Overall, the results identify acidic conditions as the most favourable for MO removal over Co₃O₄ and support a contribution of pH-dependent interfacial interactions to the observed photocatalytic response.

Figure 5. Schematic representation of the proposed pH-dependent interfacial behaviour of MO over Co₃O₄ under simulated solar irradiation.

To further investigate the reactive species involved in the photoinduced removal of MO over Co₃O₄, scavenging experiments were performed at pH 3 using IPA, BQ, and EDTA as probes for ·OH, O₂·⁻, and photogenerated holes (h⁺), respectively (Figure S10). In the absence of scavengers, MO removal reached ca. 94%. The addition of EDTA decreased MO removal to ca. 33%, whereas substantially smaller effects were observed in the presence of IPA and BQ, with final MO removal remaining at ca. 90 and 85%, respectively. These results indicate that photogenerated holes play a major role in the photoinduced removal of MO over Co₃O₄ under the investigated acidic conditions, whereas ·OH and O₂·⁻ appear to provide comparatively smaller contributions. Scavenging experiments have also been employed to investigate the reactive species involved in MO photooxidation over Co₃O₄-based photocatalytic systems using EDTA, IPA, and BQ as probes [104], while other Co₃O₄-containing systems have shown the involvement of multiple reactive species [105]. These findings highlight that the relative involvement of reactive species may depend on the specific catalyst composition and reaction conditions. Together with the photoinduced electrical response independently evidenced by EIS, these results support the involvement of photogenerated charge carriers in MO removal over Co₃O₄ under irradiation. Nevertheless, the scavenging experiments provide indirect mechanistic evidence and should not be interpreted as establishing a complete photocatalytic mechanism or as completely excluding possible contributions from dye photosensitization or light-induced changes in adsorption behaviour.

2.3. Influence of Pollutant Chemistry on the Removal Pathways over Co₃O₄

The influence of pollutant chemistry on the removal pathway was investigated by comparing MO and CR over pristine Co₃O₄ under the same operating conditions. Based on the results described in the previous section, the comparison was performed at pH 3, where favourable dye-surface interactions and the highest MO removal were observed.

The results are summarized in Figure 6a. Despite belonging to the same class of anionic azo dyes, MO and CR exhibit distinct removal behaviours under the investigated conditions. MO reaches ca. 94% removal after 180 min of irradiation, compared with ca. 33% during the 30 min dark equilibration period. In contrast, CR shows ca. 68% removal already under dark conditions, with irradiation producing only a limited additional decrease in concentration. Under the adopted experimental conditions (10 mg·L⁻¹ dye and 0.5 g·L⁻¹ material), the dark removal of MO and CR corresponds to apparent uptakes of approximately 6.6 and 13.6 mg·g⁻¹, respectively. As discussed in Section 2.2, these values refer specifically to the investigated concentration and solid loading and should not be interpreted as equilibrium or maximum adsorption capacities. Thus, while MO exhibits a substantial additional photocatalytic contribution under irradiation, CR removal is predominantly associated with adsorption under the investigated conditions.

Figure 6. (a) Dyes removal in UW by photolysis and adsorption and photocatalysis on Co₃O₄ at the end of the test (210 min). Experimental conditions: dye concentration = 10 mg·L⁻¹; catalyst dosage (when used) = 0.5 g·L⁻¹; irradiance (when used) = 35 W·m⁻². (b) Schematic representation of the different potential interfacial interactions of MO and CR with positively charged Co₃O₄ at pH (pH < pH_pzc = 5.8).

The different behaviour of the two dyes can be related, at least in part, to their different molecular structures (Figure 6b). Although both molecules belong to the class of anionic azo dyes, CR possesses a larger molecular structure containing two azo bridges, two sulfonate groups and a more extended aromatic conjugated system than MO. These features provide a greater number of potential interaction sites and may favour more extensive interactions with the Co₃O₄ surface. Thus, although the positively charged surface expected at pH 3 provides favourable electrostatic conditions for both anionic dyes, it does not imply comparable surface uptake. MO speciation may also contribute under these conditions, since pH 3 lies close to its reported pKa (~3.4). In addition to electrostatic attractions, hydrogen bonding and interactions involving the aromatic framework may contribute to surface affinity [106,107,108,109], consistent with the higher dark removal observed for CR.

The greater dark removal of CR, however, is not accompanied by a comparable increase under irradiation. The limited additional CR removal suggests that strong surface affinity alone does not necessarily favour subsequent photocatalytic conversion. Molecular size, adsorption configuration, and accessibility of surface-bound molecules to photogenerated charge carriers and/or short-lived reactive species may contribute to this behaviour, although their individual roles cannot be resolved from the present experiments. Similar effects have been reported for bulky aromatic dyes interacting with metal oxide photocatalysts [110,111]. By contrast, the substantially larger additional removal of MO upon irradiation indicates a more pronounced photocatalytic contribution under otherwise identical conditions [7]. Thus, the comparison shows that adsorption affinity and photocatalytic response do not necessarily follow the same trend for different pollutants over Co₃O₄.

2.4. Comparative Assessment of Dye-Removal Behaviour over Co-Based Materials

It is important to stress that the two materials are not simply two independent oxides with unrelated properties selected for their individual performance. CuCo₂O₄ represents a Cu-containing Co-based spinel system synthesized through a precipitation route related to that used for pristine Co₃O₄. As shown by XRD, however, the Cu-containing material also contains crystalline CuO and should therefore not be regarded as a single-phase CuCo₂O₄ spinel. The comparison provides a chemically meaningful basis for assessing how the compositional and physicochemical differences introduced in the Cu-containing system correlate with the different removal behaviours observed for the two dyes.

Having investigated the removal pathways of MO and CR over pristine Co₃O₄, the behaviour of the Cu-containing system was then examined to determine how the physicochemical differences identified between the two materials are reflected in the balance between adsorption and photocatalytic removal.

Figure 7 summarizes the removal behaviour of both dyes over the CuCo₂O₄-based sample under photolysis, adsorption and photocatalytic conditions, together with a direct comparison with the pristine Co₃O₄.

Figure 7. (a) MO and CR abatement in UW by photolysis, adsorption and photocatalysis on the CuCo₂O₄-based sample. (b) Comparison of the adsorption and photocatalytic performances of the Co₃O₄ and CuCo₂O₄-based samples toward MO and CR. Experimental conditions: MO concentration = 10 mg·L⁻¹; initial dye solution pH = 3; catalyst dosage (when used) = 0.5 g·L⁻¹; irradiance (when used) = 35 W·m⁻².

A marked change in removal behaviour is observed for the CuCo₂O₄-based material (Figure 7a). Unlike pristine Co₃O₄, which exhibited high photocatalytic MO removal under acidic conditions (Figure 4), the CuCo₂O₄-based sample displays only limited MO removal. After the initial 30 min dark equilibration, adsorption remains below 5%, while 180 min of irradiation provides only a marginal additional contribution, resulting in a final removal below 15%. These results indicate that the pronounced photocatalytic contribution observed for pristine Co₃O₄ is not retained in the CuCo₂O₄-based material.

A different scenario emerges for CR, where the adsorption and photocatalytic profiles nearly overlap throughout the experiment, indicating that irradiation provides only a negligible contribution to the overall CR removal. Therefore, as already observed over pristine Co₃O₄ but to a substantially greater extent, CR removal over the CuCo₂O₄-based sample is governed predominantly by adsorption.

The comparison between the two materials (Figure 7b) clearly highlights their contrasting functional behaviours. Whereas pristine Co₃O₄ shows a pronounced photocatalytic contribution toward MO and moderate adsorption of CR, the CuCo₂O₄-based material exhibits substantially greater CR removal by adsorption under the investigated conditions. Thus, the compositional modification of the Co-based oxide system does not result in a uniform improvement in removal performance but shifts the relative contributions of adsorption and photocatalysis in a pollutant-dependent manner.

The origin of this behaviour must be interpreted considering the simultaneous physicochemical differences between the materials. XRD and ATR confirmed the presence of characteristic spinel features in both materials, while XRD additionally demonstrated the coexistence of crystalline CuO with the Cu–Co spinel phase in the CuCo₂O₄-based sample. UV–DRS likewise revealed a different optical response, including a lower-energy contribution potentially associated with CuO. PL spectroscopy, conversely, showed broadly similar emission profiles and only a limited difference in intensity, providing no evidence for a major difference in radiative recombination behaviour under the investigated conditions (Figure S6). Importantly, this modest PL difference contrasts with the pronounced difference in MO removal, which reaches approximately 94% over Co₃O₄ but remains below 15% over the CuCo₂O₄-based material. Therefore, the different photocatalytic responses cannot be rationalized solely in terms of the radiative recombination behaviour inferred from steady-state PL. Instead, they likely reflect the combined influence of the different phase composition, optical response, surface properties and interfacial processes of the two materials.

Differences relevant to adsorption are particularly evident in the textural and surface properties. The surface area increases from 9 m²·g⁻¹ for Co₃O₄ to 17 m²·g⁻¹ for the CuCo₂O₄-based material, providing almost twice the accessible surface area and therefore representing an important contribution that must be considered when comparing their adsorption performances. In parallel, the pH_pzc shifts from 5.8 to 6.4 (Table 1), while XPS reveals surface Cu²⁺ species and differences in the surface oxygen environment (Figure S7). The substantially greater CR removal observed for the CuCo₂O₄-based material under the investigated conditions is consistent with its almost twofold larger specific surface area, which likely represents an important contribution to the higher apparent uptake. Differences in pH_pzc and surface chemical environment may also influence dye–surface interactions; however, their individual contribution cannot be separated from the effect of surface area on the basis of the present dataset. Moreover, because adsorption was evaluated at a single initial dye concentration and material loading, these experiments do not allow the equilibrium adsorption capacity of the material to be determined or the near-complete CR removal to be attributed specifically to a high intrinsic adsorption capacity. The time-resolved dark profiles indicate rapid CR uptake under the investigated loading conditions, but they do not by themselves distinguish adsorption kinetics from adsorption capacity. The available characterization therefore does not permit the respective contributions of accessible surface area, surface chemistry, and dye-specific interactions to be quantitatively separated, nor does it allow the higher CR removal to be assigned to a specific adsorption site or mechanism. Accordingly, the limited MO photocatalytic response of the CuCo₂O₄-based material cannot be unambiguously attributed to a single electronic or interfacial factor. EIS analysis of pristine Co₃O₄ show a modification of its interfacial electrical response under illumination (Figure 2). These results are therefore discussed specifically in relation to pristine Co₃O₄ and are not used to infer differences in interfacial charge-transfer properties between the two materials. The different photocatalytic behaviours of the two materials are therefore discussed in the context of their overall differences in phase composition, optical response, surface chemistry, and textural properties rather than attributed to a single charge-transfer parameter.

The asymmetry between MO and CR removal over the CuCo₂O₄-based material is particularly informative in this respect. At pH 3, the solution pH lies below the measured pH_pzc of the material (6.4), and its surface is therefore expected to be predominantly positively charged. Nevertheless, MO adsorption remains below 5%, whereas CR is almost completely removed under dark conditions. This observation demonstrates that favourable electrostatic conditions alone are insufficient to account for the experimentally observed dye-dependent behaviour. MO and CR differ substantially in molecular size, structure and number of functional groups: MO is a relatively small azo dye containing one sulfonate group [8,107], whereas CR possesses two sulfonate groups and a much larger conjugated aromatic framework, providing multiple potential interaction sites with the oxide surface [112]. Such structural differences may favour multipoint interactions of CR with the heterogeneous CuCo₂O₄-based surface, involving electrostatic and additional non-electrostatic contributions. MO speciation may also contribute under the investigated conditions, since pH 3 lies close to its reported pKa (~3.4). The different uptake of the two dyes should therefore be interpreted as the result of their different molecular and interfacial behaviours rather than as a direct consequence of surface charge alone. The present dataset does not allow the individual contributions of molecular structure, speciation, surface chemistry and texture to be quantitatively separated or a specific adsorption mechanism to be assigned. Because adsorption was evaluated at a single initial dye concentration (10 mg·L⁻¹) and material loading (0.5 g·L⁻¹), corresponding to a maximum theoretical uptake of approximately 20 mg·g⁻¹, these results should be interpreted as a comparison of removal behaviour under the investigated conditions rather than as a quantitative assessment of adsorption capacity.

Overall, the comparison shows that pristine Co₃O₄ and the CuCo₂O₄-based material favour different removal pathways under the investigated conditions. Co₃O₄ provides a substantial photocatalytic contribution toward MO removal, whereas the Cu-containing material shows limited MO removal but substantially greater CR removal by adsorption under the investigated conditions. The latter behaviour occurs together with an almost twofold increase in specific surface area and changes in surface chemistry; the present dataset does not allow their individual contributions to be separated or an intrinsic adsorption capacity to be established.

2.5. Reusability and Regeneration of Co-Based Materials

Since the two cobalt-based materials exhibited different removal pathways, their reusability and regeneration behaviour were investigated to assess whether these distinct functional roles could be preserved during repeated operation.

The obtained results are summarized in Figure 8, where the left panels report the concentration profiles recorded during consecutive reuse cycles, while the right panels compare the performance of the fresh and regenerated materials. The regeneration efficiency (RE %) was calculated as reported in the Experimental Section, using the removal efficiencies associated with photocatalysis or adsorption, respectively, to quantify the recovery of the initial removal performance after the UV regeneration treatment.

Figure 8. Reuse and regeneration results: profiles of C/C₀ versus time of consecutive reuse cycles (left) and comparison of the removal efficiencies achieved during the material’s first use and after post-use UV regeneration of the recovered material for 4 h. (a,b) Co₃O₄ during MO photocatalytic removal; (c,d) Co₃O₄ during CR adsorption; (e,f) CuCo₂O₄-based sample during CR adsorption. Data are reported as mean values from at least three independent experiments; error bars represent the standard deviation. Experimental conditions: dye concentration = 10 mg·L⁻¹; initial dye solution pH = 3; catalyst dosage = 0.5 g·L⁻¹; irradiance (when used) = 35 W·m⁻².

The reuse behaviour of pristine Co₃O₄ during MO photocatalytic removal is reported in Figure 8a. The material maintains photocatalytic activity over three consecutive cycles, although a progressive decrease in performance is observed. While 94% MO removal is achieved after 180 min of irradiation in the first cycle, increasingly longer irradiation times are required in the following runs to reach comparable removal levels. In the absence of post-cycle structural characterization, the origin of this gradual activity loss cannot be unambiguously established. The partial recovery of photocatalytic performance after the regeneration treatment, discussed below, nevertheless indicates that at least part of the observed activity loss is reversible. No specific deactivation mechanism can be assigned on the basis of the available data.

To assess the recoverability of the photocatalytic function, the spent material was regenerated under UV irradiation for 4 h and reused under the same operating conditions (Figure 8b). The regenerated catalyst recovers most of its original activity, yielding a regeneration efficiency of approximately 77%. Although the regeneration is not complete, this result confirms a substantial functional recovery after UV treatment. Although recovery of activity alone does not demonstrate structural stability of the materials, ICP-OES analysis of the liquid phase collected after the catalytic tests showed Co concentrations below the corresponding detection limit (LOD = 0.010 mg·L⁻¹) for pristine Co₃O₄. For the CuCo₂O₄-based material, both Co and Cu concentrations were below their respective detection limits (0.010 and 0.005 mg·L⁻¹, respectively). These results indicate that metal release, if present, remained below the detection capability of the adopted analytical method under the investigated conditions, although post-use structural characterization would be required to assess possible changes in the solid-state properties of the materials.

A different behaviour is observed when pristine Co₃O₄ is employed for CR removal through adsorption (Figure 8c). In this case, the adsorption performance decreases much more rapidly during reuse, with the second cycle exhibiting a drastically lower removal efficiency than the fresh material. Unlike photocatalysis, adsorption directly depends on the availability of accessible surface sites [48]. Therefore, retention of CR on the material after the first cycle is expected to reduce the number of sites available for subsequent adsorption, providing a plausible explanation for the pronounced decrease observed upon direct reuse. Under the conditions adopted here (10 mg·L⁻¹ CR and 0.5 g·L⁻¹ material), the maximum apparent uptake corresponding to complete removal would be 20 mg·g⁻¹; accordingly, these experiments should be interpreted as reuse tests under a fixed pollutant loading rather than as measurements of the adsorption capacity of the material.

Following the same UV regeneration treatment, the adsorption performance is largely restored (Figure 8d). The regenerated material exhibits a regeneration efficiency of about 92%, calculated as detailed in the Experimental Section. This result indicates that the CR-removal function of Co₃O₄ can be effectively recovered under the adopted regeneration conditions. Since UV regeneration was performed on the recovered dry material in the absence of an aqueous phase, the treatment should be regarded as a post-use photoinduced surface treatment rather than as a desorption step in solution. However, the present data do not establish whether the recovery of performance results from photochemical transformation of residual adsorbed species or from other UV-induced surface-cleaning processes.

The regeneration behaviour of the CuCo₂O₄-based sample was finally evaluated for CR adsorption (Figure 8e). Here, the adsorption performance decreases upon direct reuse, again indicating a reduced removal capability after the first adsorption cycle. However, after UV regeneration the original performance is recovered within the experimental resolution (Figure 8f), corresponding to a full regeneration efficiency. As for Co₃O₄, this recovery should be interpreted as functional regeneration of the dry recovered material, without assigning a specific regeneration mechanism.

Moreover, Co and Cu concentrations remained below their respective ICP-OES detection limits after treatment, indicating that the recovery of adsorption performance was not accompanied by detectable metal release under the investigated conditions.

Overall, the reuse experiments show that direct operation progressively decreases the removal performance of the investigated materials, whereas UV regeneration allows substantial functional recovery. RE values of approximately 77%, 92%, and 100% were obtained for MO photocatalytic removal over Co₃O₄, CR adsorption over Co₃O₄, and CR adsorption over the CuCo₂O₄-based material, respectively. These results support the reversibility of a substantial fraction of the performance loss under the investigated conditions, while structural stability and the mechanisms responsible for regeneration remain to be independently established.

2.6. Comparison with Literature on Co-Based Materials for Dye Removal

To position the obtained results within the context of Co-based materials for dye removal, representative literature data are compared in Table 2. Direct ranking of the reported performances should be approached with caution, since the experimental conditions vary considerably among studies in terms of dye concentration, material dosage, solution pH, irradiation source and treatment time. Nevertheless, the comparison provides a useful overview of the different functional roles reported for Co₃O₄ and CuCo₂O₄-based materials and highlights how the present results relate to the existing literature.

Table 2. Literature comparison of representative Co-based materials reported for dye removal.

Material	Dye	Process	[Dye] (mg·L−1)	Material Dosage (g·L−1)	pH	Irradiation Conditions	Treatment Time (min)	Performance	Reuse	Ref.
Co3O4	MO	Adsorption	50	10	6	Dark	60	46.08 mg·g−1	n.r.	[113]
Co3O4	Safranine	Photocatalysis	10−5 M	0.10	n.r.	UV irradiation	80	~81% removal	4 cycles; 81% → 79%	[114]
Co3O4	Methylene blue	Photocatalysis	40	0.8	11	Solar light	90	88%	5 cycles	[115]
10 wt.% Sr-Co3O4	Methylene blue	Photocatalysis	40	0.8	11	Solar light	90	96%	5 cycles	[115]
Co3O4	Malachite green	Photocatalysis	40	0.8	9	Solar light	90	85%	5 cycles	[115]
10 wt.% Sr-Co3O4	Malachite green	Photocatalysis	40	0.8	9	Solar light	90	92%	5 cycles	[115]
Sn-doped Co3O4	Methylene blue	Photocatalysis	n.r.	n.r.	n.r.	Visible light	180	~75%	3 cycles	[116]
Bentonite/Co3O4	Malachite green	Adsorption/photocatalysis	100	0.5	6	400 W metal halide lamp (~490 nm)	180	qmax = 223 mg·g−1; complete removal	n.r.	[117]
Biogenic Co3O4	CR	Photocatalysis	120	n.r.	n.r.	UV irradiation	60	~100%	5 cycles	[118]
CuCo2O4	Basic fuchsine	Adsorption	50	0.20	8	Dark	5	~97%	n.r.	[119]
CuCo2O4/MgO	Celestine blue B	Adsorption	50	0.005 g a	5	Dark	30	~97%	n.r.	[120]
CuCo2O4	MO	Photocatalysis	300	0.075	n.r.	sunlight	120	~99%	6 cycles; >88% retained	[121]
CuCo2O4	CR	Photocatalysis	300	0.075	n.r.	sunlight	120	~99%	6 cycles; >88% retained	[121]
Co3O4	MO	Photocatalysis	10	0.5	3	Simulated solar light, 35 W·m−2	180	94% removal	3 cycles; RE = 77% after UV regeneration	This work
Co3O4	CR	Adsorption	10	0.5	3	-	180	~68% removal	RE = 92% after UV regeneration	This work
CuCo2O4-based	CR	Adsorption	10	0.5	3	-	180	Near complete removal	RE = 100% after UV regeneration	This work

n.r. = not reported: ^a volume not reported.

As shown in Table 2, Co₃O₄ has been successfully employed both as an adsorbent and as a photocatalyst. For MO adsorption, a capacity of 46.08 mg·g⁻¹ was reported at a relatively high material dosage of 10 g·L⁻¹ [113], whereas photocatalytic studies have achieved removal efficiencies ranging from approximately 81% to 96% for different dyes and modified Co₃O₄ systems [114,115,116,117]. High photocatalytic removal has also been reported for CR using biogenic Co₃O₄ nanoparticles [118]. Similarly, CuCo₂O₄-containing materials have shown high adsorption or photocatalytic performances depending on their formulation and target dye [119,120,121]. For example, CuCo₂O₄-based adsorbents achieved approximately 97% removal of basic fuchsine and celestine blue [119,120], while photocatalytic removal approaching 99% has been reported for MO and CR over CuCo₂O₄ nanosheets under sunlight [121]. These results confirm the versatility of Co-based oxides for dye removal but also illustrate the strong dependence of the reported performance on material formulation and experimental conditions.

Under the conditions adopted in our work, pristine Co₃O₄ achieved 94% MO removal under simulated solar irradiation at a material dosage of 0.5 g·L⁻¹, while the CuCo₂O₄-based material provided near-complete CR removal through adsorption at the same dosage and pH. These values fall within the high-performance range reported for Co-based systems, despite the substantial differences in experimental conditions that preclude a strict numerical comparison. More importantly, however, the present results show that high removal efficiencies can originate from different functional pathways. Whereas pristine Co₃O₄ provides a substantial photocatalytic contribution toward MO removal, the CuCo₂O₄-based material behaves predominantly as an adsorbent toward CR under the investigated conditions.

A particularly relevant comparison is provided by Chen et al. [121], since CuCo₂O₄ nanosheets were reported to achieve ca. 99% removal of both MO and CR under sunlight. However, the two dyes were investigated separately rather than as components of a competitive binary mixture, and the material and operating conditions differed substantially from those adopted here. The CuCo₂O₄ nanosheets reported by Chen et al. [121] were prepared using a deep eutectic solvent, exhibited a BET surface area of 42 m²·g⁻¹, displayed an ultrathin leaf-like morphology, and were characterized by XPS contributions assigned by the authors to lattice oxygen, oxygen-vacancy-related species, and chemisorbed oxygen. These features were discussed by Chen et al. [121] in relation to the availability of surface active sites and dye-removal performance. Moreover, the material exhibited a substantially different surface-charge behaviour (point of zero charge of 11.8) and appreciable dark adsorption of both MO and CR. In contrast, the CuCo₂O₄-based material investigated in the present work has a lower BET surface area (17 m²·g⁻¹), contains crystalline CuO in addition to the Cu–Co spinel phase, and exhibits different morphological and surface characteristics, while showing limited MO removal under irradiation and near-complete CR removal predominantly through adsorption at pH 3. These differences are also consistent with the distinct synthesis routes adopted in the two studies and may affect the amount and nature of accessible surface sites, dye–surface interactions, and photoinduced interfacial processes. The different behaviour of these nominally related Cu–Co oxide systems therefore further illustrates that dye-removal performance and the relative contributions of adsorption and photocatalysis cannot be inferred from nominal composition alone, but depend strongly on phase composition, morphology, textural and surface properties, and operating conditions. At the same time, the present characterization does not allow the individual contributions of these factors to be isolated or differences in facet exposure or defect concentration between the two materials to be directly established.

The present study should also be distinguished from previous work in which pollutant-dependent behaviour was deliberately exploited. In a recent work of some of the authors [7], a two-step, pH-driven adsorption process combining PANI-modified loofah and pristine loofah was developed for the selective removal of anionic and cationic dyes from a multicomponent solution. Similarly, a successive study [8] demonstrated that a single floating BiOBr/biochar photocatalytic platform can exhibit different removal behaviour toward structurally distinct pollutants, including under competitive multi-pollutant conditions. Thus, neither pollutant-dependent removal nor sequential treatment is, by itself, unique to the present work. Here, these concepts are extended to structurally related Co-based oxides investigated under the same experimental conditions, for which different dominant functions are identified toward the same pair of anionic dyes. Independent dark and irradiation experiments allow the relative contributions of adsorption and photoinduced removal to be distinguished, while binary-mixture experiments establish whether the complementary behaviour identified in single-dye systems is retained under competitive conditions.

In this respect, the physicochemical characterization discussed above indicates that the different functional behaviours arise from the combined influence of phase composition, accessible surface area, surface charge and surface chemical environment, together with the molecular characteristics of the target dye, rather than from a single material descriptor. However, these factors should be regarded as concurrent descriptors of the observed behaviour rather than as independently established mechanistic determinants, since their individual contributions cannot be quantitatively separated from the present dataset. The comparison between pristine Co₃O₄ and the CuCo₂O₄-based material therefore provides information that would remain obscured if the two systems were evaluated exclusively on the basis of their final removal percentages.

Reuse and regeneration further complement this comparison. Although several literature studies report repeated photocatalytic operation, regeneration of the specific functional response is less consistently addressed (Table 2). Here, UV regeneration recovered approximately 77% of the initial MO photocatalytic performance of Co₃O₄ and 92% and 100% of the CR adsorption performance of Co₃O₄ and the CuCo₂O₄-based material, respectively. Thus, the experimentally identified photocatalytic and adsorption functions are not only pollutant-dependent but can also be substantially restored after use.

Overall, the literature comparison indicates that the significance of the present results does not lie in establishing a new maximum dye-removal efficiency for Co-based materials, nor in introducing sequential treatment as a concept in itself. Rather, the present study comparatively identifies, under a common experimental framework, distinct pollutant-dependent functions in structurally related Co-based oxides and evaluates whether these functions are retained when the materials are applied to a competitive binary dye mixture. The additional comparison between sequential and simultaneous one-pot configurations further shows that temporal separation of the two material functions is not required to achieve high final removal under the investigated conditions. The sequential configuration should therefore not be regarded as intrinsically superior in terms of removal efficiency, but as an alternative operational arrangement that physically separates the adsorption and photoinduced-removal stages, allowing independent material recovery and potentially facilitating separate regeneration. Taken together, these results shift the comparison from maximum removal efficiency toward the identification and operational use of complementary material functions according to pollutant-dependent behaviour.

2.7. Exploiting the Complementary Behaviour of Co-Based Materials in Binary Dye Treatment

The complementary behaviours observed for the two Co-based materials provided the rationale for their combination in a sequential treatment strategy for a binary MO/CR mixture. While pristine Co₃O₄ showed high photocatalytic removal of MO under acidic conditions, the CuCo₂O₄-based material exhibited near-complete CR removal by adsorption under the investigated conditions. Accordingly, the binary mixture was first treated with the CuCo₂O₄-based material under dark conditions to remove CR by adsorption. After solid separation, the resulting MO-enriched solution was treated with pristine Co₃O₄ under simulated solar irradiation (Figure S11).

As shown in Figure 9a, CR was rapidly and almost completely removed during the first stage, whereas the MO concentration remained essentially unchanged. The preferential adsorption of CR by the CuCo₂O₄-based material is therefore retained in the presence of a competing anionic dye, extending the behaviour observed in the single-dye experiments to a binary system. The larger aromatic framework and the presence of two sulfonate groups in CR may contribute to its stronger interaction with the material surface compared with MO, although the present results do not allow this selectivity to be assigned to a specific adsorption site or interaction mechanism. After removal of the CuCo₂O₄-based material, the residual MO was progressively removed upon irradiation in the presence of pristine Co₃O₄, resulting in near-complete removal of both dyes at the end of the sequential treatment (Figure 9a).

Figure 9. Comparison of sequential and simultaneous one-pot treatment of a binary MO/CR mixture in UW. (a) C/C₀ vs. time profiles during the sequential two-step treatment (i) step 1, CR adsorption by the CuCo₂O₄-based material under dark conditions, (ii) step 2, photocatalytic removal of the remaining MO by Co₃O₄ under solar irradiation. (b) C/C₀ vs. time profiles during simultaneous one-pot treatment in the presence of both Co₃O₄ and the CuCo₂O₄-based material. Experimental conditions: initial total dye concentration = 20 mg·L⁻¹ (10 mg·L⁻¹ each) at pH = 3; material dosage = 0.5 g·L⁻¹ for each material (total solid loading = 1.0 g·L⁻¹ in the one-pot configuration); irradiance (when used) = 35 W·m⁻².

To directly assess whether the sequential configuration provided an advantage over the simultaneous use of the two materials, an additional one-pot experiment was performed by introducing Co₃O₄ and the CuCo₂O₄-based material simultaneously into the binary MO/CR mixture (Figure 9b). During the dark period, CR was again almost completely removed, whereas MO remained essentially unaffected. Upon simulated solar irradiation, MO concentration progressively decreased, ultimately resulting in near-complete removal of both dyes. The simultaneous one-pot and sequential configurations therefore achieved comparable final removal efficiencies under the investigated conditions.

These results show that temporal separation of the two material functions is not required to achieve high overall removal efficiency under the investigated conditions. Rather, the sequential configuration provides a different operational arrangement in which CR adsorption and MO photoinduced removal are assigned to physically separated treatment stages. This allows the two materials to be independently recovered after their respective treatment stages and may facilitate their separate regeneration, while preserving the complementary pollutant-dependent functions identified in the individual dye experiments.

Overall, the comparison between sequential and simultaneous treatment confirms that the material–pollutant complementarity identified from the individual dye tests is retained in a competitive binary system. The sequential configuration should therefore not be regarded as intrinsically superior to the one-pot treatment in terms of final removal efficiency, but as an approach that enables controlled separation of the adsorption and photoinduced-removal functions. These findings are currently limited to the controlled experimental conditions investigated here and should not be extrapolated to treatment performance in realistic wastewater matrices. Further experiments over broader concentration ranges and mixture compositions, and particularly in the presence of variable ionic strength, competing inorganic ions, natural organic matter, and real wastewater matrices, will therefore be required to establish the robustness and practical applicability of this behaviour.

3. Materials and Methods

3.1. Materials

Cobalt(II) nitrate hexahydrate (Co(NO₃)₂·6H₂O, 98%) and copper(II) nitrate trihydrate (Cu(NO₃)₂·3H₂O, 99–104%) were of analytical grade and purchased from Merck (Darmstadt, Germany).

For the physicochemical characterization of the materials, liquid nitrogen (N₂, >99.0%, Sapio, Monza, Italy), nitric acid (HNO₃, 65.0%, Carlo Erba, Milano, Italy), sodium hydroxide (NaOH, >99%, Carlo Erba, Milano, Italy), hydrochloric acid (HCl, 37%, Carlo Erba, Milano, Italy), sodium nitrate (NaNO₃, >99%, Merck, Darmstadt, Germany), barium sulfate (BaSO₄, >99.0%, Nacalai Tescue, Kyoto, Japan), and sodium sulfate (Na₂SO₄, >99%, Merck, Darmstadt, Germany) were used.

For the abatement tests, methyl orange (C₁₄H₁₄N₃NaO₃S, ACS reagent 85%) and Congo red (C₃₂H₂₂N₆Na₂O₆S₂, ACS reagent 85%) were supplied by Merck (Darmstadt, Germany) and used as model pollutants. Isopropyl alcohol (IPA, ACS reagent, ≥99.5%), p-benzoquinone (BQ, reagent grade, ≥98%), and ethylenediaminetetraacetic acid disodium salt dihydrate (EDTA, reagent grade, 98.5–101.5%) were supplied by Merck (Darmstadt, Germany) and used in the scavenging experiments. Acetonitrile (CH₃CN, HPLC grade, Carlo Erba, Milano, Italy), water (H₂O, HPLC grade, Carlo Erba, Milano, Italy), and formic acid (HCOOH, 98%, Sigma-Aldrich, Saint Louis, MO, USA) were used for analytical measurements.

Ultra-pure water was obtained using a Milli-Q^® system (Merck Millipore, Darmstadt, Germany).

3.2. Materials Preparation

Co₃O₄ and the CuCo₂O₄-based sample were prepared by precipitation of metal hydroxides followed by thermal treatment according to a procedure adapted from Ref. [115].

Co₃O₄ was prepared by dissolving ca. 2.90 g of Co(NO₃)₂·6H₂O in 100 mL of ultrapure water to obtain a 0.1 M solution into a 250 mL necked flask. The system was stirred at room temperature at 250 rpm for 30 min. Then, 50 mL of a NaOH solution (0.2 M) was added dropwise under vigorous stirring. The resulting suspension was heated to 100 °C in an oil bath and maintained under stirring at 250 rpm for 30 min, leading to the formation of a precipitate. The solid was recovered by centrifugation at 5000 rpm for 15 min, washed several times with ultrapure water, and dried at 100 °C for 3 h. The dried powder was finally calcined at 500 °C for 2 h in static air, using a heating rate of 8 °C·min⁻¹.

The CuCo₂O₄-based sample was prepared by co-precipitation of cobalt and copper precursors. Co(NO₃)₂·6H₂O (8.73 g) and Cu(NO₃)₂·3H₂O (3.62 g) were dissolved together in 100 mL of 0.4 M HCl to obtain initial Co²⁺ and Cu²⁺ concentrations of 0.3 and 0.15 M, respectively, corresponding to an initial Co/Cu atomic ratio of 2. The resulting solution was stirred at 80 °C for 20 min. Subsequently, 150 mL of an aqueous NaOH solution (1.5 M) was rapidly added, resulting in a final suspension pH of about 13, and the suspension was vigorously stirred at 250 rpm for 2 h. The resulting solid was recovered by centrifugation, thoroughly washed with ultrapure water until a neutral pH was reached, and dried overnight at 70 °C. The dried powder was then calcined at 500 °C for 2 h in static air, using a heating rate of 8° C·min⁻¹.

3.3. Physicochemical Characterization of the Materials

The crystal structure and phase composition of the samples were analyzed by X-ray powder diffraction (XRD) using a PANanalytical X’Pert PRO diffractometer (Malvern Panalytical, Malvern, UK) equipped with a Cu Kα radiation (λ = 1.54060 Å) and operating at 40 kV and 40 mA. Diffraction patterns were collected over a 2θ range of 10–70°, with a step size of 0.02° and ranging from 5 to 96 s per step. Phase identification was performed by comparison with reference patterns from the International Center for Diffraction Data (ICDD) database. Quantitative phase analysis was carried out by Rietveld refinement of the powder XRD pattern using the MAUD software (v. 2.98) [122], without the use of an internal standard. Phase weight fractions were refined directly as independent parameters via MAUD’s layer-based refinement scheme, which provides a quantification equivalent to the conventional scale-factor approach (Hill–Howard relation) while directly optimizing the phase weight fractions during the least-squares refinement. The two crystalline phases were modelled as CuO (tenorite, space group Cc) and a Co–Cu mixed spinel (Cu_0.72Co_2.28O₄, space group Fd-3m). Lattice parameters, crystallite size, and microstrain were refined isotropically for both phases, together with the instrumental and background parameters. The refinement converged to a spinel:CuO weight ratio of 70.8(7):29.2(7) wt%, with fit quality indices of Rwp = 1.54%, Rp = 1.20%, Rexp = 1.37%, and χ² (GOF) = 1.28, indicating a satisfactory agreement between the observed and calculated patterns (Table S2, Figure S12). As no internal standard was employed, the reported weight fractions refer to the crystalline phases identified by XRD and do not account for possible amorphous content.

FTIR-ATR (Attenuated Total Reflection) spectra were obtained with a Bruker Vertex70 spectrometer (Bruker, Billerica, MA, USA) equipped with Platinum ATR-QL diamond cell and DTGS detector (4 cm⁻¹ resolution, 32 scans).

Morphology and elemental distribution were investigated by field-emission scanning electron microscopy (FESEM) using a TESCAN S9000G microscope (TESCAN ORSAY HOLDING, Brno, Czech Republic) equipped with a Schottky field emission gun (nominal resolution: 0.7 nm at 15 keV in In-Beam SE mode) and an Oxford Ultim Max EDS detector (Oxford Instruments, High Wycombe, UK). Samples were mounted on aluminum stubs using conductive carbon tape and coated with a thin chromium layer to improve surface conductivity.

Textural properties were determined by N₂ adsorption–desorption isotherms at −196 °C using a Micromeritics Tristar 3Flex analyzer (Micromeritics, Norcross, GA, USA). Prior to analysis, ca. 1 g of each sample was dried at 120 °C overnight and subsequently degassed at 150 °C for 4 h under nitrogen flow (Air Liquide Alphagaz, Paris, France) to remove physisorbed species. The surface area was calculated using the Brunauer–Emmet–Teller (BET) equation (2-parameters) in the relative pressure range 0.05 < p/p⁰ < 0.30, assuming a nitrogen cross-sectional area of 16.2 Ǻ² [123]. The pore size distribution was determined using the Barrett–Joyner–Halenda (BJH) model from the desorption branch of the isotherms over the relative-pressure range 0.3 < p/p⁰ < 0.95.

The point of zero charge (pH_pzc) was determined using the pH drift method [78]. A series of 0.1 M NaNO₃ solutions with initial pH values adjusted between 4.0 and 10.0 was prepared. For each measurement, 50 mg of sample was dispersed in 20 mL NaNO₃ solution, and the resulting suspensions were stirred for 24 h. The suspensions were then centrifuged, and the final pH was measured. The pH_pzc was identified as the pH at which ΔpH (pH_final − pH_initial) equaled zero.

Optical properties were investigated by UV–vis diffuse reflectance spectroscopy (UV–vis DRS) in the 200–800 nm range using a Shimadzu UV-2600i UV–vis spectrophotometer (Shimadzu, Kyoto, Japan) equipped with an integrating sphere. Samples were finely ground and pressed into disks (ca. 4 cm in diameter), and BaSO₄ was used as the reflectance standard. The reflectance data were converted using the Kubelka-Munk function, and Tauc plots were constructed to estimate the optical transition energies.

The photoluminescence data were recorded using a Fluorolog F2 Horiba/Jobin-Yvon spectrofluorimeter (Horiba, Kyoto, Japan), on pure samples.

Surface composition and chemical states were analyzed by X-ray photoelectron spectroscopy (XPS) using an M-PROBE Surface Spectrometer (Surface Science Instruments SSI, Mountain View, CA, USA) equipped with an Al Kα source. Measurements were performed under ultra-high vacuum (10⁻⁷–10⁻⁸ Torr), using a spot size of 800 µm. Survey spectra were collected over a binding-energy range of 0–1100 eV using a dwell time of 100 ms. High-resolution (HR) spectra were acquired using a pass energy of 40 eV, a step size of 0.1 eV, and a dwell time of 200 ms. Binding energies (BE) were calibrated by setting the C 1s peak assigned to C-C/C-H at 284.6 eV [90,94]. Data processing was carried out using ESCA Hawk software (version 7.04.04).

The semiconductive properties of Co₃O₄ were investigated by electrochemical impedance spectroscopy (EIS) using a single-compartment three-electrode glass cell. Powdered Co₃O₄ was immobilized on an indium tin oxide (ITO)-coated glass sheet (exposed geometric area: 0.25 cm²) by depositing 20 μL of a well-dispersed catalyst suspension prepared by dispersing 10 mg Co₃O₄ powder in 500 μL of acetone followed by ultrasonication [8]. A saturated calomel electrode (SCE) and a platinum foil (1 cm²) were used as reference and counter electrodes, respectively. All electrochemical measurements were performed using a Metrohm Autolab PGSTAT204 potentiostat/galvanostat (Metrohm, Herisau, Switzerland).

Impedance spectra were recorded in potentiostatic mode in 0.1 M Na₂SO₄ aqueous solution over the frequency range from 100 kHz to 0.1 Hz using an AC perturbation amplitude of 10 mV. Measurements were carried out under both dark and illuminated conditions (simulated solar lamp, OSRAM ULTRAVITALUX 300 W, OSRAM GmbH, Munich, Germany, irradiance of 35 W·m⁻²) to evaluate the influence of light irradiation on the electrical response of the investigated oxides. The experimental impedance spectra were analyzed as Nyquist and Bode plots and fitted using the NOVA software (Metrohm Autolab B.V., version 2.1.9, Utrecht, The Netherlands). The fitting quality was evaluated using the root mean square error (RMSE).

3.4. Adsorption, Photolysis and Photocatalytic Tests

Methyl orange (MO) and Congo red (CR) were selected as model dyes for the adsorption, photolysis, and photocatalytic removal experiments. Tests were carried out both in the dark and under simulated solar irradiation.

Photocatalytic experiments were performed under atmospheric conditions in a 250 mL cylindrical glass batch reactor placed inside a custom-made box with dark walls to minimize interference from external light. An OSRAM Ultra Vitalux 300 W lamp, providing an irradiance of 35 W·m⁻², was positioned 25 cm above the reactor and maintained at a fixed distance throughout all experiments.

In a typical experiment, the material was dispersed at a concentration of 0.5 g·L⁻¹ in 100 mL dye solution prepared in ultrapure water (UW). The initial pH was either left at its natural value (ca. 6, independently of the dye) or adjusted to acidic or basic conditions (3 and 11, respectively) by adding diluted HNO₃ or NaOH solutions (1 M), respectively. Before irradiation, the suspension was stirred in the dark for 30 min to assess the initial adsorption contribution. Photocatalytic tests were then carried out under continuous stirring at 250 rpm for 180 min.

Dye removal was monitored over a total experimental time of 210 min by withdrawing 2 mL aliquots at predetermined intervals, namely every 15 min during the first 30 min, and every 30 min thereafter. Before analysis, the aliquots were centrifuged at 5000 rpm for 10 min to remove suspended particles of the materials.

The collected aliquots were analyzed by UV–vis spectroscopy or HPLC-UV, depending on the dye and experimental configuration.

UV–vis measurements were performed using a Shimadzu UV-2600i UV–vis spectrophotometer (Shimadzu, Kyoto, Japan). MO concentration was monitored at 463 nm under neutral/basic conditions and at 508 nm under acidic conditions, whereas CR concentration was monitored at 566 nm (typically under acidic conditions). For single-dye experiments, dye concentrations were determined from the corresponding UV–vis calibration curves. In binary MO/CR mixtures, the two dyes were quantified independently using complementary analytical methods to avoid spectral overlap: MO was quantified by HPLC/UV, whereas CR was quantified by UV–vis spectrophotometry at 566 nm using its corresponding calibration curve. HPLC-UV analyses of MO were performed using an Agilent 1100 system equipped with a C18 Supelco column (25 cm × 4 mm, 5 μm) and a 20 μL injection loop. The mobile phase consisted of 49.9% water, 49.9% acetonitrile, and 0.2% formic acid, and was delivered at a flow rate of 1 mL·min⁻¹. The MO peak was monitored at 463 nm.

The dye abatement was computed by the equation:

$$\mathit{Pollutant\; abatement}\left(\%\right)={\displaystyle \frac{{\mathit{C}}_0-\mathit{C}}{{\mathit{C}}_0}}·100\%$$

(1)

where C₀ is the initial dye concentration and C is the dye concentration at time t. The results are reported as C/C₀ versus time profiles.

Metal leaching was evaluated at the end of the tests by analyzing the recovered liquid phase by Inductively Coupled Plasma Optical Emission Spectroscopy (ICP-OES). Co was quantified for pristine Co₃O₄, whereas both Co and Cu were quantified for the CuCo₂O₄-based material. An “Activa” Jobin Yvon ICP-OES instrument (HORIBA, Kyoto, Japan) was used. The limits of detection (LOD) were 0.010 mg·L⁻¹ for Co and 0.005 mg·L⁻¹ for Cu.

Single-dye experiments were carried out at an initial concentration of 10 mg·L⁻¹. For binary-mixture experiments, a total dye concentration of 20 mg·L⁻¹ was used, corresponding to 10 mg·L⁻¹ of each dye, mainly for analytical consistency.

Adsorption tests were performed under otherwise identical conditions, with the systems maintained in the dark for the entire experiment.

For the photocatalytic removal of MO, the rate constant was determined by fitting the irradiation-stage data to a pseudo-first-order model:

$$\mathit{ln}{\displaystyle \frac{{\mathit{C}}_0}{\mathit{C}}}=\mathit{kt}$$

(2)

where C₀ is the MO concentration at the beginning of the irradiation stage, C is the concentration at irradiation time t, and k is the pseudo-first-order rate constant. The obtained fitting showed correlation coefficients (R²) higher than 0.98.

Scavenging tests were performed to investigate the involvement of reactive species in the photoinduced removal of MO over Co₃O₄ under acidic conditions. Isopropyl alcohol (IPA), p-benzoquinone (BQ), and disodium ethylenediaminetetraacetate (EDTA) were used as scavengers for hydroxyl radicals (·OH), superoxide radicals (O₂·⁻), and photogenerated holes (h⁺), respectively [104]. The experiments were carried out using an initial MO concentration of 10 mg·L⁻¹, an initial solution pH of 3, and a Co₃O₄ dosage of 0.5 g·L⁻¹. After the 30 min dark-equilibration period, immediately before simulated-solar irradiation, each scavenger was added at 3 mmol per 100 mL of dye solution, corresponding to a final concentration of 30 mm. The same irradiation conditions described above were then applied.

Reusability tests were performed under the same operating conditions used for the single-dye experiments. After each treatment cycle, the material was recovered by centrifugation, washed with ultrapure water, and reused without additional treatment. For regeneration experiments, the material was recovered, dried, and successively exposed to UV-A irradiation (Jelosil, HG 500; 500 W nominal power; emission range: 315–400 nm, maximum emission at 365 nm, Vimodrone, Italy) at an irradiance of 20 W·m⁻² for 4 h. The regenerated material was then reused under the same experimental conditions adopted for the corresponding removal test.

For photocatalytic tests, regeneration efficiency was calculated as:

$${\mathit{RE}}_{\mathit{photo}}\left(\%\right)={\displaystyle \frac{{\mathit{k}}_{\mathit{regenerated}}}{{\mathit{k}}_{\mathit{fresh}}}}·100$$

(3)

For adsorption-driven tests, regeneration efficiency was calculated from the removal efficiency obtained at the end of the test:

$${\mathit{RE}}_{\mathit{ads}}\left(\%\right)={\displaystyle \frac{{\mathbf{\eta }}_{\mathit{regenerated}}}{{\mathbf{\eta }}_{\mathit{fresh}}}}·100$$

(4)

where η_fresh and η_regenerated are the final removal efficiencies obtained with the fresh and regenerated materials, respectively.

Based on the results obtained in the single-dye experiments, a sequential treatment of a binary MO/CR mixture was performed. The mixture contained 20 mg·L⁻¹ of total dyes, corresponding to 10 mg·L⁻¹ of each dye, in UW (100 mL final volume) at an initial pH of 3. In the first step, the CuCo₂O₄-based sample was added at a dosage of 0.5 g·L⁻¹, and the suspension was stirred in the dark for 210 min to remove CR by adsorption. The solid was then removed by filtration, and the resulting MO-enriched solution was treated in a second step with Co₃O₄ at a dosage of 0.5 g·L⁻¹. After a 30 min dark equilibration period, the suspension was irradiated under simulated solar irradiation (35 W·m⁻²) for 180 min. The sequential treatment procedure is schematically illustrated in Figure S11.

For comparison, a simultaneous one-pot experiment was performed using the same binary MO/CR mixture and operating conditions. Co₃O₄ and the CuCo₂O₄-based material were simultaneously added to the solution at a dosage of 0.5 g·L⁻¹ each, corresponding to a total solid loading of 1.0 g·L⁻¹. Both materials remained in the reactor throughout the experiment. The suspension was first stirred in the dark for 210 min and was subsequently irradiated under simulated solar light (35 W·m⁻²) for 180 min. Dye concentrations were monitored as described above for the binary-mixture experiments.

All dye-removal experiments were independently performed at least in triplicate (n ≥ 3). This replication scheme also applied to the reuse and regeneration experiments, for which each complete sequence of consecutive reuse cycles and subsequent regeneration was independently repeated at least three times. Data are reported as mean values, and error bars represent the standard deviation.

4. Conclusions

This work demonstrates that structurally related Co-based oxides can exhibit different functions in dye removal depending on both material properties and pollutant characteristics. Pristine Co₃O₄ showed effective MO photocatalytic removal under simulated solar irradiation at acidic pH, whereas CR removal involved a substantially stronger adsorption contribution. In contrast, the CuCo₂O₄-based material displayed a limited photocatalytic contribution toward MO but near-complete CR removal by adsorption under the investigated conditions. These differences are consistent with the combined influence of composition, surface and textural properties, and dye molecular characteristics.

Importantly, these pollutant-dependent behaviours were retained in a competitive binary MO/CR mixture. In the sequential configuration, near-complete CR removal by adsorption over the CuCo₂O₄-based material was followed by photocatalytic removal of the residual MO using pristine Co₃O₄, demonstrating that the complementary material functions can be assigned to successive treatment stages. A simultaneous one-pot configuration also achieved near-complete final removal of both dyes, with performance comparable to the sequential treatment. Thus, temporal separation of the two functions is not required to achieve high overall removal under the investigated conditions; rather, the sequential configuration provides an operational separation of the adsorption and photoinduced-removal functions, potentially facilitating independent material recovery and regeneration.

Overall, the results show that the potential of Co-based materials for water treatment lies not only in maximizing the performance of an individual material, but also in identifying the function for which each material is best suited. These findings provide a rational basis for integrating complementary adsorption and photocatalytic functions through different operational configurations according to pollutant-dependent behaviour. ICP-OES analysis further indicated that Co and Cu release remained below the corresponding detection limits under the investigated treatment conditions.

Further studies involving broader pollutant concentration ranges and mixture compositions, variable ionic strength, competing inorganic ions, natural organic matter, and realistic water matrices, together with post-use structural characterization, will be required to establish the robustness, stability and practical applicability of the proposed approach.