IRS was collected from Hutumuri, Maluku, Indonesia. All chemicals were of analytical grade, including hydrochloric acid (HCl, PT. Smarlab Indonesia, 37%), zinc nitrate hexahydrate (Zn(NO₃ )₂ ·6H₂ O, Merck, 98%), urea (GR for analysis ACS, Reag. Ph Eur), sodium hydroxide (NaOH, Sigma-Aldrich, reagent grade, ≥98%, pellets anhydrous), potassium bromide (KBr, Sigma-Aldrich, ACS reagent, ≥99.0%), MB (Merck, 80%), and demineralized water.

The IRS was washed with demineralized water. The solid was separated and dried overnight. The dried solid was mixed with 0.1 M HCl (diluted from HCl 37%) and stirred for one hour. The solid was separated and washed using demineralized water. The final product was dried at 105°C for 24 h and labelled as IRSH.

The ZnO/IRSH composite was synthesized via co-precipitation. An amount of Zn(NO₃ )₂ ·6H₂ O was dissolved in 5 mL of demineralized water. To this solution, 0.5 g of IRSH was added while stirring continuously. Next, urea was introduced to the mixture, maintaining a 1:1 molar ratio with the zinc nitrate. The mixture was then heated to 70°C, at which point a NaOH solution was added, creating a 2:1 molar ratio of NaOH to zinc nitrate. The final mixture was stirred at 70°C for one hour. The solid product was collected, thoroughly rinsed with 100 mL of demineralized water, and dried overnight at 60°C. To ensure crystallinity and stability, the dried solid was annealed at 400°C for two hours. The final annealed product was labeled xZnO/IRSH, where ‘x’ represents the weight percentage of ZnO, varied at 5, 10, and 20% to investigate its effect on the composite’s properties.

Phase and crystallinity confirmation was performed by X-ray diffraction (XRD) using a Rigaku Miniflex600. The samples were measured at 2θ = 5–90° with a scan rate of 10°/minute and a scan width of 0.02°. The functional groups in the samples were observed using a Fourier transform infrared spectrophotometer (FTIR, Shimadzu 8400S). The sample was blended with KBr in an agate mortar, placed on a pellet holder, and pressed to form a pellet. The resulting pellet was placed on a sample holder and measured over the wavenumber range 4000–400 cm⁻¹. The morphology of samples, including ZnO, IRSH, and ZnO/IRSH (10% as the optimal sample), was investigated by field-emission scanning electron microscopy (FESEM, Thermo Fisher Apreo 2), equipped with an energy-dispersive X-ray spectrometer to analyse the elemental distribution. Micromeritics TriStar II Plus 3.03 was employed to perform the physisorption of nitrogen (N₂). Prior to analysis, the samples were degassed at 423K to remove any remaining surface water. The sample was subsequently analysed at 77K. The data obtained was used to determine the textural features of samples, including surface area, pore volume, and pore distribution. The surface composition of ZnO/IRSH was also analyzed using X-ray photoelectron spectroscopy (XPS, XPS Kratos AXIS SUPRA PLUS/ESCA). The absorbance profiles of the samples were analysed using a UV-Vis diffuse reflectance spectrophotometer (UV-Vis DRS, Agilent Cary 60). The sample was analysed over the wavelength range 200–800 nm at a scan rate of 10 nm/s. The Photoluminescence (PL) spectra of samples were recorded using Horiba MicOS Photoluminescence microspectrophotometer (iHR320) at emission wavelength of 350–1100 nm and excitation laser of 325 nm.

40 mg of ZnO/IRSH was mixed with 50 mL of MB solution (37.5 mg·L⁻¹). The mixture was stirred in the dark for 30 minutes to reach the adsorption-desorption equilibrium of MB on ZnO/IRSH. The mixture was subsequently irradiated for 4 hours using a UV-LED lamp with a 400 nm monochromatic wavelength and a power of 12 W. The use of a 400 nm light source is significant because it marks the boundary between the UV and visible spectra. This approach allows for precise assessment of the catalyst’s ability to harvest low-energy photons in the near-UV and violet regions, providing a more accurate simulation of the near-UV and violet components of natural solar radiation than traditional short-wave UV lamps. 3 mL of solution was collected at 1-hour intervals. The solution was filtered using a PVDF membrane syringe filter to separate the solution from the solid. The solution was subsequently analysed using a UV-Vis spectrophotometer (Drawell’s UV-Vis N2S) to obtain the absorbance. The optimization of the ZnO amount deposited on IRSH was investigated to determine the optimal photocatalytic activity for MB degradation.

The dosage of ZnO/IRSH was varied from 10 to 50 mg, with a constant volume of MB solution, to determine the effective dosage of the samples for removing MB. The MB concentration was varied during the photocatalytic reaction to assess its effect on the photocatalytic performance of the samples. The MB concentration was varied at 22.5, 30, 37.5, 45, and 50 mg·L⁻¹. The effect of the MB solution’s pH was also investigated at 3, 5, 7, 9, and 11 to determine the optimal photocatalytic conditions. Prior to the reaction, the pH of the MB solution was adjusted by adding 0.1 M HCl or 0.1 M NaOH.

RSM optimization was employed to simultaneously analyse multiple variables, examine their relationships, and determine the optimal experimental conditions. RSM-Box Bechen (RSM–BB) was used to optimize the photodegradation of MB using ZnO/IRSH. The BB design, regression analysis, and surface plot were implemented using Python. ANOVA is a statistical method used to identify the various effects of control parameters on design outputs. Optimization of the MB adsorption process using BB was carried out using 15 experimental combinations involving four independent variables at low (−1), medium (0), and high (+1) levels. The three variables include: Dosage (mg) (X₀), MB concentration (mg·L⁻¹) (X₁), and pH of MB solution (X₂), as summarized in Table S1.

The intermediate analysis is important for detailing the mechanism of MB photodegradation using ZnO/IRSH. The solution after the photocatalytic reaction was separated by centrifugation, followed by filtration through a PTFE membrane syringe to remove any solids. The filtrate was subsequently analysed using Liquid chromatography tandem with mass spectrometry (LC–MS).

Fig. 1 shows the diffraction pattern of the samples. IRSH (Fig. 1a) shows several peaks correlated with the mineral composition. According to the analysis, IRSH is composed of hisingerite (Fe₂Si₂O₇·xH₂O, ICSD No. 05–0296), hematite (Fe₂O₃, ICSD No. 73–0603), chloritoid (Al₁₅Fe₇H_13.60Mg_1.20O_53.60Si₈, ICSD No. 72–1183), iron silicate (Fe₂O₄Si, ICSD No. 74–0715), and greenalite-1T (Fe₃H₄O₉Si₂, ICDS No. 11–0265). Fig. 1b shows the diffraction pattern of ZnO/IRSH with the control materials, including IRSH and ZnO. The ZnO impregnation on IRSH shows a combination of the typical ZnO and IRSH peaks. At 5% ZnO, the ZnO peak is not clearly visible, possibly due to the low ZnO loading on IRSH. Increasing the amount of ZnO to 10% and 15%, the typical peaks of ZnO are clearly observed. The typical peaks of ZnO are shown at 2θ = 31.74, 34.42, 36.27, 47.55, 56.58, 62.82, 67.91, and 69.05° according to the JCPDS No. 36–1451, which is ascribed to the respective (100), (002), (101), (102), (110), (103), (112), and (201). The intensity of ZnO also increases with increasing ZnO amount. Nonetheless, the IRSH peak is retained even after ZnO impregnation, indicating its stability.

Fig. S1 shows the FTIR spectra of the samples. ZnO shows broad peaks at 3600–3000 cm⁻¹, ascribed to the hydroxyl (OH) group of adsorbed water on the ZnO surface. A sharp absorption band at 650–400 cm⁻¹ is related to the vibration of the Zn–O bond in the structure of ZnO. IRSH shows the typical peaks of the OH group at 3600 cm⁻¹ and 3700 cm⁻¹ [26], which are attributed to the silanol group in the IRSH structure. The bending vibration of the OH group is observed at 1650 cm⁻¹. The absorption band of Si–O–Si is observed at 1063 cm⁻¹ [27, 28]. The stretching vibrations of tetrahedral units and double-ring 4-atom tetrahedral units are observed at wavenumbers of 900–700 cm⁻¹ and 568 cm⁻¹, respectively [29]. The absorption band below 500 cm⁻¹ is related to the Fe–O vibration [30], which is composed of the IRSH. The incorporation of ZnO into IRSH does not significantly alter its FTIR spectrum. All the IRSH absorption bands are maintained, indicating the stability of IRSH even after impregnation.

Fig. 2 shows the FESEM images of IRSH (top), ZnO/IRSH (centre), and ZnO (bottom). IRSH shows the appearance of large, irregular particles. At high magnification, the large particles are composed of flake-like particles that form a layered structure. The incorporation of ZnO into IRSH does not significantly alter the IRSH structure. The flake-layered structure is retained, showing the good stability of IRSH during the impregnation process. In contrast, ZnO nanoparticles exhibit a range of shapes and sizes. However, these nanoparticles are not clearly observed in ZnO/IRSH due to the low amount of ZnO loaded on IRSH.

The EDX analysis was performed to determine the elemental composition of the samples, as shown in Fig. S2. In IRSH, silicon (Si), aluminium (Al), iron (Fe), and oxygen (O) are evenly distributed and powerfully present. Additionally, the carbon (C) signal indicates that the IRSH substance contains an organic or carbon-based component. The IRSH material’s complex inorganic nature, which is possibly a clay or mineral rich in silicates, aluminates, and iron oxides, is confirmed by this elemental fingerprint. The EDS spectra of Si, Al, Fe, O, and C in ZnO/IRSH are identical compared to IRSH, suggesting that the IRSH material serves as the base for the composite’s core structure. Additionally, ZnO/IRSH shows the appearance of the ZnO signal, which is distributed throughout the sample, suggesting that the zinc oxide particles are dispersed on the surface or within the agglomerates of the IRSH material. The co-localization of the zinc signal with the other elements further confirms the successful synthesis of the ZnO/IRSH composite, where the ZnO is intimately mixed with the IRSH component rather than existing as a separate phase.

Fig. S3a presents the profile of N₂ adsorption-desorption of ZnO, IRSH, and 10ZnO/IRSH. The parameter of textural properties of ZnO, IRSH, and 10ZnO/IRSH is summarized in Table S2. The profile of N₂ adsorption-desorption shares a similar pattern. Initially, adsorption of the samples is initiated at a slightly elevated N₂ pressure, followed by a gradual increase to P/P₀ of 0.22, demonstrating micropore filling. The gradual increase in N₂ adsorption is further observed at high N₂ uptake up to P/P₀ of 0.75, which correlates with the filling of mesopore features in the samples. Subsequently, N₂ uptake is significantly enhanced in the P/P₀ range of 0.75–0.99, which correlates with the filling of macropore features. The occurrence of a hysteresis loop in all samples is attributed to the condensation process. All samples exhibit a combination of type I, III, and IV isotherms, indicating the formation of micropore, mesopore, and macropore combinations. IRSH shows the highest N₂ adsorption, indicating the highest surface area. The incorporation of ZnO decreased N₂ uptake, indicating that the surface area of IRSH decreased after incorporation. This finding is due to pore blockage by ZnO nanoparticles and to the reduction in surface sites.

Further, the pore-size distribution of the samples in Fig. S3b shows a non-uniform distribution of pore diameters. All samples show the presence of micropores at 1.43 nm, 1.56 nm, and 1.56 nm for ZnO, IRSH, and 10ZnO/IRSH, respectively. The distribution of mesopore sizes is observed in all samples, indicating non-uniformity. ZnO exhibits a mesopore size of 4.89 nm, whereas IRSH shows two prominent diameter sizes at 5.29 nm and 10.49 nm. The incorporation of ZnO into IRSH yields mesoporous materials with similar properties, exhibiting two prominent pore sizes at 5.09 nm and 11.28 nm. The intensity of dV(d) is decreased after the incorporation of ZnO, which is in line with the decrease in surface area. Importantly, this incorporation strategy maintains all the pore characteristics of IRSH, which would be beneficial for harvesting MB molecules during photodegradation.

The composite surface nature of the samples was further analyzed by XPS. Fig. S4 shows the XPS spectra of 10ZnO/IRS in the core level of its constituent element. The Si 2p XPS spectrum (Fig. S4a) shows two deconvoluted peaks. The first peak at 102.36 eV is attributed to the Si–O–Zn linkage, whereas the peak at 103 eV is related to the Si–O–Si [31]. This deconvolution demonstrates that ZnO is chemically bound to the silicon-based material rather than merely physically blended. A single peak is observed at the Al 2p spectra (Fig. S4b), which is ascribed to the Al–O bonds in the IRSH structure [32]. The Fe 2p spectrum (Fig. S4c) is particularly informative; the presence of a strong satellite peak alongside the central Fe 2p_3/2 and Fe 2p_1/2 spin-orbit doublet is a hallmark of high-spin iron(III), likely as ferric oxide (Fe₂O₃), which is a common component in iron-rich materials [33].

The XPS spectra of Zn 2p (Fig. S4d) show the presence of a doublet peak at 1023 eV and 1045 eV, which is correlated with the respective Zn 2p_3/2 and Zn 2p_1/2 [34]. The absence of other peaks indicates that the oxidation state of Zn²⁺ is contributed by ZnO. The O 1s XPS spectra (Fig. S4e) show five deconvoluted peaks. The first peak at 529.79 eV is attributed to the Zn–O bond in the ZnO lattice structure [16]. The presence of the Si–O–Zn feature at 530.45 eV indicates an interaction between Zn and SiO–during impregnation, which aligns with the finding in Si–O–Ti [35]. The formation of Si-O-Zn linkage at the interface of ZnO/IRSH would be beneficial to stabilize holes or transfer electrons from the ZnO to the silicon component [36]. As a result, the recombination rate decreases, providing more charge carriers to participate in surface redox processes and thereby enhancing photocatalytic activity. The Si–O–Si is further observed at 531.41 eV, which is related to the IRSH composition. The Si–O–H bond is also observed, which is related to the hydroxyl surface group in the samples [37]. Further, several carbon species are observed in the C 1s spectrum (Fig. S4f), with peaks denoting C–H, C–O, and C=O bonds, which support the findings from the FTIR results.

The absorption properties of samples are important for assessing their light adsorption ability, which can affect photocatalytic activity. Fig. 3a shows the adsorption profile of the samples. IRSH exhibits strong absorption across the UV and visible regions, demonstrating its ability to adsorb a broad range of light. This ability is due to the presence of iron oxide in the IRSH structure, which activates light absorption in the visible light range. ZnO exhibits a high reflectance of over 90% in the visible light range, whereas it is relatively low in the UV region. This characteristic is typical of a wide-bandgap semiconductor, indicating that ZnO can only be activated by UV light irradiation. The adsorption profile of ZnO/IRSH is similar to that of IRSH. The presence of ZnO increases reflectance intensity, indicating a decrease in the samples’ ability to absorb UV or visible light.

Figs. 3b-f presents the Kubelka-Munk plot to determine the band gap of the samples [38]. The results indicate that the optical band gap of ZnO/IRSH composites increased (blue-shifted) from 2.29 eV to 2.51 eV with higher ZnO loading. This increase indicates that incorporating ZnO alters the IRSH framework, leading to a slight increase in the energy required for electronic transitions. Although these values exceed the band gap of pristine IRSH, they remain significantly lower than that of bulk ZnO (3.37 eV). As the band gaps fall within the 2.29–2.51 eV range, these materials exhibit high visible-light absorption efficiency. These composites have the potential to be useful for technologies that require specific responses across different parts of the solar spectrum, such as photocatalysis or solar energy conversion. This tunability is essential for a variety of applications because it enables precise control over a material’s light-absorption characteristics.

Fig. S5 presents the PL spectra of ZnO, IRSH, and 10ZnO/IRSH. The PL spectrum of the pristine ZnO shows a relatively low peak below 400 nm, which is related to the near-band-edge free exciton transition because the localized level below the conduction band gives way to the valence band. The red emission at 650 nm is also observed and attributed to defect emission, such as an oxygen vacancy [39]. The highest red emission intensity indicates that the pristine ZnO has a high defect density. However, the high PL intensity indicates a rapid recombination rate, thereby limiting ZnO’s photocatalytic activity. IRSH shows lower PL intensity than pristine ZnO, indicating a lower charge-carrier recombination rate. However, the PL profile of IRSH differs from that of ZnO, which exhibits distinct charge-carrier dynamics and defect states, likely due to IRSH’s diverse mineral composition. The intensity of PL is decreased after ZnO is introduced to IRSH, suggesting that the recombination rate in ZnO/IRSH is lower than in IRSH and ZnO [40]. The intimate interaction between ZnO and IRSH enables effective charge-carrier separation and transfer. This may be related to the heterostructure that forms between ZnO and the Fephotocatalyst in IRSH, as ZnO/IRSH exhibits a lower recombination rate. As a result, more charge carriers are available for surface redox reactions, which are essential for the photocatalytic degradation of pollutants.

BBD-RSM was used to identify the ideal condition for the photodegradation of MB using 10ZnO/IRSH. The photodegradation process of 15 was performed as presented in Table S3. BBD is composed of three independent variables (X): dosage (X₀), initial MB concentration (X₁), and pH of the MB solution (X₂). As a result, a quadratic model of the photodegradation of MB using 10ZnO/IRSH is shown in Eq. (1).

Fig. 5a shows the parity plot of the predicted photodegradation percentage versus the experimental photodegradation percentage, with a determination coefficient (R₂) of 0.979 (see Table 1). This result implies that the model accounts for 97.93% of the data’s variability. The empirical model is adequate for defining the variability in essay reading, which should be greater than or equal to 0.75 [44]. Besides, the adjusted R₂ (0.942) is s in close agreement with the R₂, confirming the adequacy of the model. Table 1 presents the regression analysis results. The F-statistic is 26.28, and the corresponding p-value is 0.00109. Since the p-value is much less than 0.05, the overall model is statistically significant, meaning the independent variables collectively have a significant effect on the dependent variable. The JB test statistic is 0.795 with a p-value (Prob(JB)) of 0.672. Since the p-value is high, we cannot reject the null of normality for the residuals, suggesting they are likely normally distributed.

Furthermore, the probability plot (Fig. 5b) shows that the residuals follow a normal distribution, suggesting that the errors are random and unbiased. These results suggest that the model is reliable and provides a valid representation of the system [45]. Fig. 5c shows the relative effect of variables on the photodegradation performance of 10ZnO/IRSH. It can be noted that X₁ and X₀ are the dominant variables driving the photodegradation efficiency of 10ZnO/IRSH. X₁ shows the highest contribution, which is nearly half (49.32%), followed by X₀ (31.435).

In detail, the contour plot of the interaction between two variables is shown in Fig. 6. The photodegradation of MB increases with both the initial MB concentration (20–30 mg L⁻¹) and the catalyst dosage (20–35 mg). Similarly, the photodegradation of MB increases with pH over the range 2–9. However, the photodegradation of MB decreases with increasing catalyst dosage above 35 mg and pH above 9. The high dosage reduces light penetration into the surface-active layer, reducing the formation of active species and decreasing photocatalytic activity. Similarly, the decline in MB removal at high pH (over 9) is attributed to competition between ·OH formation and the interaction of MB⁺ and OH⁻. The interaction of MB⁺ and OH⁻ reduces the adsorption of MB on 10ZnO/IRSH and reduces the photocatalytic activity of MB. A similar finding is also reported by Santoso et al. [46] and Khalaji [47], who observed a decrease in the MB photocatalytic efficiency at solution pH values above 10.

The intermediate product of MB photodegradation is essential for elucidating the mechanism of MB degradation. Fig. 7 shows the schematic of the MB degradation pathway based on LC-MS analysis. The MS spectra of each retention time are presented in Figs. S8–S9. Initially, two pathway reactions occur during the light illumination. Demethylation of MB molecules (m/z = 284, Fig. S9b) occurs, producing molecules with m/z = 270, which is commonly found in the first stage of the MB photodegradation pathway [48, 49]. Subsequently, demethylation and deamination occur, removing the methyl and amino groups from the Azure B molecules, generating molecules with m/z = 240 (Fig. S9a) [50].

Besides, the ring-opening of the thiazine group at the N–C bond is followed by hydroxylation, yielding molecules with m/z = 303 (Fig. S9b). This compound is subsequently cracked into two compounds due to ring-opening of the S–C bond. The continuous ring opening occurs in the right compound, followed by hydroxylation, which produces the compounds with m/z = 130 (Fig. S9d). This formed compound (m/z = 130) is similar to that reported in previous works [51, 52]. Alternatively, the A compound is partially hydroxylated to generate a compound with m/z = 153 (Fig. S9e). This compound (m/z = 153) is also created by the ring opening and hydroxylation of the compound with m/z = 240. The compound with m/z 153 is also reported in previous work by Wolkis and co-workers [53]. Both compounds with m/z = 153 and compounds with m/z = 130 are subsequently degraded by reactive oxygen species to produce CO₂ and H₂O.

Radical scavenging experiments were conducted to elucidate the photocatalytic mechanism and identify the primary reactive species involved in the degradation process. DMSO, isopropanol (IPA), methanol, and ascorbic acid (AA) were used as scavengers of electron [12], hydroxyl radicals (·OH) [54–56], hole and superoxide radicals (·O₂⁻) [57], respectively. As illustrated in Fig. S10, the photocatalytic activity of the ZnO/IRSH composite decreased markedly after the addition of IPA and AA, which serve as scavengers for ·OH and ·O₂⁻, respectively. The substantial reduction in degradation efficiency observed with IPA suggests that ·OH is the principal reactive species responsible for the reaction. These findings provide empirical support for the degradation pathway proposed by LC–MS analysis, confirming that the identified intermediate fragments result from the systematic oxidative action of these reactive oxygen species (ROS) on the parent molecule.