Intensification of triphasic photocatalytic microreactor performance via high-frequency ultrasound for enhanced mass transfer and dye degradation Authors: Mahdieh Daneshvar, Arsalan Parvareh, Neda Azimi Published: 2026 ABSTRACT AbstractThis study demonstrates that high-frequency ultrasound can significantly enhance mass transfer efficiency in a three-phase (gas–liquid–solid) photocatalytic microreactor, improving dye degradation performance. A custom-designed microreactor, consisting of four diamond-shaped mixing chambers, was equipped with piezoelectric plate (PZT) transducers operating at 1.7 MHz, with a nominal input power of 5.1–5.2 W per transducer. The effective acoustic power delivered to the reaction medium was determined using a calorimetric method based on the temperature rise of the liquid, and was found to correspond to 51.3–61.6% of the nominal electrical input power depending on the number of activated PZT transducers. To investigate the synergistic effect between ultrasound and photocatalysis, methylene blue (MB) degradation was evaluated under three operational modes: photocatalysis alone (UV), sonolysis (US), and combined sonophotocatalysis (UV + US). A magnetically recoverable Fe2O3/TiO2 photocatalyst was synthesized and used together with oxygen as the gas phase. Under optimal conditions, the sonophotocatalytic process removed 44.3% of MB, representing a 13% improvement over UV photocatalysis alone and a 55% increase compared with sonolysis. This enhancement was associated with substantial improvements in key performance indicators, including an apparent overall removal coefficient (kORC) of 2470 min−1 and a significantly elevated reaction rate constant under mixed conditions. Statistical analyses using multiple linear regression (MLR) and Pearson correlation confirmed that catalyst dosage had the strongest positive correlation with removal efficiency (r = 0.47, p < 0.01), while flow rate and initial dye concentration negatively influenced performance. The PZT transducer arrangement was also critical: simultaneous activation of three transducers yielded 37.83% removal efficiency, more than doubling the performance of single-transducer configurations. These results demonstrate that high-frequency, plate-type ultrasound integrated into microreactors enhances cavitation, microstreaming, and mass transfer, offering a powerful approach to intensifying triphasic photocatalytic processes and advancing wastewater treatment technologies.Keywords: Sonophotocatalysis, Photocatalyst, Microreactor, Mass transfer, Ultrasound Introduction Recent advancements in novel technologies have significantly improved the efficiency and productivity of chemical engineering processes, particularly in optimizing energy consumption and reducing operational costs [1]. Within this context, the study of mass and energy transfer mechanisms in multiphase systems, mixing enhancement, improvement of reaction yield and selectivity, and the advancement of separation units represent core research domains in established scientific literature [2]. Since multiphase flows play such a central role in chemical processing, being able to accurately evaluate how these systems behave is crucial. Gas–liquid flows, in particular, have long been recognized as some of the most demanding to work with because the phases do not distribute evenly and mass transfer tends to be inherently limited. These ongoing difficulties have made it increasingly clear that new strategies are needed to intensify operations and improve the overall performance of multiphase processes [3]. Recent research shows that the use of microsystems, especially microreactors, can overcome a significant part of these limitations. Microreactors can improve the overall process efficiency by increasing the mass transfer coefficient [4]. Due to the very small dimensions of the channels, these systems create stable and uniform two-phase flows and provide a high specific surface area. Despite these advantages, the dominance of the laminar flow regime in microchannels reduces the effective turbulence and the internal circulation in the liquid slugs is usually not sufficient to create rapid mixing. As a result, the mass transfer coefficient in the liquid phase (kL) is limited [4], [5]. Increasing molecular diffusion and creating turbulent convection can significantly improve the mixing quality by increasing the contact area between fluids and shortening the mixing path [6]. Two types of methods are commonly used to enhance this process: passive and active. Passive methods employ elements such as baffles, curved channels, convergent–divergent structures, and asymmetric geometries; approaches that have been extensively investigated in both numerical simulations and practical experiments [7]. In contrast, active methods rely on the application of external fields including acoustic, magnetic, electric, or thermal fields to increase the intensity and quality of mixing [8], [9]. One of the most effective active methods is the use of ultrasound in microchannels; a method that is implemented with the help of ultrasonic transducers and thin piezoelectric layers and is capable of creating localized turbulence and very effective mixing [10]. However, in catalyst-containing reactors, channel clogging and blockage remain major challenges; these problems cannot be solved by passive methods, but sometimes the use of designs such as baffles or spiral channels can exacerbate the situation [11], [12]. Microreactors equipped with piezoelectric elements and capable of generating ultrasonic waves have attracted much attention due to their significant improvement in mixing and reduced channel clogging [13], [14]. These waves significantly increase the efficiency of chemical reactions by promoting phase mixing and increasing mass and energy transfer [15], [16]. Although the effect of ultrasound on mixing and system cleaning has been well studied, specific research focusing on gas–liquid flows at the microscale is limited, and most studies have been conducted on the scale of conventional reactors [17]. In microreactors, the application of ultrasonic vibrations can improve the gas–liquid mass transfer coefficient by increasing turbulence in the liquid phase [18]. These waves cause acoustic cavitation, which prevents the aggregation and sedimentation of solid particles by generating swirling currents [19]. Although the sonolysis process is important due to its ability to prevent secondary pollution and irreversible degradation of pollutants, its efficiency in removing hydrophilic organic compounds is still limited [20]. For this reason, more research has focused on the combined sonophotocatalytic (SPC) process, a method that simultaneously utilizes ultrasonic waves and photocatalysts and shows great potential for increasing the efficiency of pollutant removal [18], [21]. By simultaneously using UV light and ultrasonic waves, this technique reduces the limitations of UV light absorption and overcomes the challenges of mixing, which in turn increases the overall reaction rate [21]. Several studies have shown that the intense oscillations of cavitation bubbles in microchannels act as micromixers and effectively remove deposits from the wall surfaces [22]. Furthermore, increasing ultrasonic power is directly related to the improvement of the volumetric mass transfer coefficient (kLa) [22], [23]. The efficiency of sonophotocatalysis coupled with a catalyst has also been confirmed in the degradation of organic dyes such as methyl orange, methylene blue, RhB, and Orange II [19]. Despite extensive research in the field of sonophotocatalysis, most previous studies have been limited to batch systems and low ultrasonic frequencies. For this reason, the effects of high-frequency ultrasound (around 1.7 MHz) on mass transfer enhancement in continuous and three-phase (gas–liquid–solid) systems, along with the importance of optimal design of piezoelectric (PZT) plate transducer configurations, have not yet been fully investigated and have received little attention. This lack of simultaneous research on mass transfer and reaction kinetics in continuous systems with advanced geometries has created a significant knowledge gap and requires further studies. Ultrasound waves are generally classified into low-frequency and high-frequency regimes due to the different physical mechanisms they induce. The primary mechanism responsible for ultrasound-related chemical and physical effects is acoustic cavitation, which involves the cyclic nucleation, growth, and implosive collapse of microbubbles in a liquid medium [24]. At low frequencies (20–100 kHz), the relatively long acoustic cycle allows significant bubble growth, leading to violent transient cavitation and near-adiabatic collapse that generates extreme localized conditions (temperatures of ∼5000 K and pressures of ∼1000 atm). This regime is highly effective for radical generation and pyrolytic degradation via homolytic bond cleavage; however, it may be less suitable for microscale flow systems due to intense localized erosion and inefficient spatial distribution of energy [13]. In contrast, high-frequency ultrasound (≥1 MHz) does not eliminate cavitation but modifies its dynamics. The shorter acoustic period limits bubble growth, resulting in smaller, more stable cavitation bubbles and a reduced likelihood of violent collapse. Under these conditions, the dominant physical effect shifts toward acoustic microstreaming and stable cavitation, which generate localized shear forces and enhance fluid motion [25]. These oscillating microbubbles act as distributed micro-stirrers, effectively thinning the concentration boundary layer around photocatalyst particles, enhancing convective mass transfer, and reducing particle agglomeration without causing structural damage to the catalyst surface [11], [26]. This effect is particularly advantageous in continuous-flow microreactors, where laminar flow conditions dominate and mixing is inherently limited. Therefore, in this study, high-frequency ultrasound (1.7 MHz) generated by piezoelectric plate transducers was selected to intensify overall removal and phase interaction in a continuous three-phase microreactor system. The aim of this study is to fill the existing research gap by developing an effective method that integrates high-frequency ultrasound with photocatalysts to significantly enhance the mass transfer efficiency in three-phase continuous microreactors (gas–liquid–solid). The novelty of this work lies in the systematic investigation of a three-phase continuous system under the influence of 1.7 MHz ultrasound and, in particular, the analysis of the role of the spatial configuration of piezoelectric elements in improving the overall process efficiency. The removal of methylene blue (MB) from aqueous solution was chosen as a model process. MB, a cationic aromatic heterocyclic dye of the thiazine class, is widely used in the textile industry and recognized as a representative organic pollutant in advanced oxidation process (AOP) studies [27], [28]. Owing to its stable aromatic structure, MB exhibits high resistance to conventional biological treatment, rendering it a relevant and challenging target for intensified AOPs. Moreover, MB poses significant environmental and health hazards, being toxic to aquatic organisms even at low concentrations. Its concentration can be measured simply and accurately via spectrophotometry throughout experiments, attributable to its distinct optical properties, specifically the absorption band at 663 nm. Finally, as one of the most extensively studied model dyes in photocatalytic and sonophotocatalytic research, MB facilitates meaningful comparison of the present results with the existing literature. The evaluation involved the comparison of three separate processes: photocatalysis (UV), sonocatalysis (US), and sonophotocatalysis (UV + US) in a custom-designed microreactor. The liquid-phase apparent overall removal coefficient (kORC) and apparent reaction rate constants (kapp) were calculated and analyzed from the experimental data. The microreactor used was made of Plexiglas with convergent-divergent geometry and contained methylene blue aqueous phase, Fe2O3/TiO2 magnetic nanocomposite as solid phase, and injected oxygen as gas phase. The Fe2O3/TiO2 nanocomposite, containing 6% iron oxide, was synthesized via ultrasonically assisted impregnation hybrid method and used as photocatalyst. Key operating parameters including liquid phase flow rate, initial pollutant concentration, photocatalyst dosage, ultrasonic wave intensity, and PZTs position were experimentally investigated. In addition, multivariate linear regression modeling and Pearson correlation analysis, performed in Python environment provided quantitative evaluation of operating parameters and statistically confirmed the superior performance of the hybrid process and identified the priority of catalyst dosage optimization. The superparamagnetic property of iron oxide nanoparticles allows for easy recovery and redispersion of the catalyst after the reaction, which is of great importance for continuous processes [29]. Materials and experimental methods Materials In this study, a Fe2O3/TiO2 magnetic nanocomposite containing 6 wt% iron oxide was synthesized via a combined ultrasonic-assisted impregnation method and used as a photocatalyst to investigate the role of ultrasonic irradiation in improving mass transfer and phase mixing in photocatalytic microreactors. The final material was thoroughly washed, dried, and calcined before use. The reagents used in this study were Fe3O4 nano powder (≥98% purity), tetra-n-butyl orthotitanate (≥98% purity), absolute ethanol, a 25% ammonia solution, and methylene blue (≥98% purity). Methods Photocatalytic degradation experiments were conducted in a custom-designed three-phase microreactor equipped with four piezoelectric plate transducers (PZTs), each operating at 1.7 MHz and connected to an ultrasonic driving module. In ultrasonic systems, calorimetry is used to determine the amount of ultrasonic energy absorbed by the liquid phase, which is primarily converted into heat. Therefore, the resulting temperature rise can be used to estimate the effective acoustic power delivered to the medium. Assuming adiabatic conditions, the acoustic power was calculated from the initial slope of the temperature–time profile using the following equation [30], [31]: Where Pacoustic is the effective acoustic power (W), m represents the mass of liquid (kg), Cp is the specific heat capacity of liquid (J/kg·K), and dT/dt is the initial slope of temperature rise (K/s). The effective acoustic power of the system was evaluated using a calorimetric method for different numbers of activated piezoelectric transducers. The results show that the electro–acoustic conversion efficiency is not strictly constant but varies with the number of operating PZTs. Specifically, the effective acoustic power corresponded to 51.3% of the nominal input power for one PZT, increasing to 54.6% for two PZTs, 59.7% for three PZTs, and 61.6% for four PZTs. This increasing trend suggests that the overall acoustic coupling efficiency improves with the number of simultaneously activated transducers, likely due to enhanced acoustic field overlap and more uniform energy distribution within the microreactor, which facilitates more efficient energy transfer into the liquid phase. Importantly, these experimentally determined efficiencies are in good agreement with previously reported values in the literature (50–63%) [30], [31], confirming the reliability of the calorimetric measurements and the consistency of the system performance with established ultrasonic reactor behavior. However, the observed variation with the number of PZTs indicates that the acoustic energy distribution and coupling conditions evolve with reactor configuration, leading to a non-linear dependence of effective acoustic power on the number of active transducers. UV radiation was provided by a 125 W high-pressure mercury lamp and oxygen was continuously injected from a gas cylinder with a gas-to-liquid volumetric ratio of 0.2:1 to provide the necessary conditions for photocatalytic reactions. The liquid flow rate was carefully controlled using a syringe pump (SP2000HSM) and the experiments were performed at 20 and 40 mL/min. To investigate the effect of the operating parameters, catalyst doses of 0.1, 0.2, and 0.4 mg/L and initial methylene blue concentrations of 30, 50 and 60 ppm were evaluated. The methylene blue concentration at the reactor outlet was measured using a UV–Vis spectrophotometer (Cecil-7250, UK) at a maximum absorption wavelength of 663 nm. Experimental setup A custom-designed three-phase microreactor consisting of four diamond-shaped mixing chambers was used to improve mass transfer and phase mixing. The reactor consisted of a Plexiglas microchannel with a T-shaped inlet, such that the liquid phase containing methylene blue and magnetic photocatalyst, and the gaseous oxygen phase were introduced separately before merging into the main channel. The combined flow then passed through successive mixing chambers, where the contact surface and mass transfer were significantly increased. Key geometric specifications are provided in Table S1. Each chamber was connected at its center to a 1.7 MHz PZT, which was connected to an ultrasonic generator and allowed for independent acoustic excitation. Continuous ultrasonic irradiation was applied simultaneously with the gas–liquid injection, creating localized microcurrents, cavitation, and effective mixing near the transducers. UV radiation was provided by a 125 W high-pressure mercury lamp located 10 cm above the reactor surface. Effluent was collected downstream and the photocatalyst nanoparticles were separated from the solution using magnetic properties. The residual methylene blue concentration was determined by UV–Vis spectrophotometry at 663 nm, and the removal efficiency (RE) was calculated using Eq. (2). All degradation experiments were repeated three times, and the reported removal efficiencies are the average values of these replicates, ensuring reproducibility. Table S2 provides the experimental uncertainty data, namely standard deviations and standard errors of the mean, for all replicated conditions. Where C0 and C denote the initial and final methylene blue concentrations, respectively. A schematic of the experimental setup is presented in Fig. 1. Integrated Continuous-Flow System for Combined US/UV Degradation in Triphasic Microreactor. Synthesis of magnetic Fe2O3/TiO2 photocatalyst One of the main challenges in using heterogeneous catalysts for liquid-phase processes is the separation of solid catalysts from the reaction mixture. To address this issue, a magnetically recoverable Fe2O3/TiO2 photocatalyst containing 6 wt% iron oxide was synthesized using a combined saturation method assisted by ultrasound [32]. Ultrasonic energy was employed in this method to enhance the dispersion of the magnetic phase within the titanium dioxide matrix, making catalyst recovery both simpler and more cost-effective. The synthesis proceeded as follows: 0.96 g of Fe3O4 nano powder was combined with 6 mL of ammonia and 160 mL of ethanol. This mixture was homogenized for 15 min using a probe-type ultrasonic homogenizer (20 kHz, 200 W nominal power), providing a power density of approximately 1.2 W/mL relative to the initial volume of 166 mL and effectively preventing particle aggregation. Next, 12 mL of tetra-n-butyl orthotitanate was added, and the mixture was further subjected to ultrasonic treatment in a bath (40 kHz, 200 W nominal power, 9 L capacity) for 1 h. The bath’s power density of roughly 22.2 W/mL ensured uniform hydrolysis and condensation of the titanium precursor. The combination of probe and bath ultrasonication at different frequencies optimized particle dispersion and reaction conditions. After the initial precipitate formed, 10 mL of deionized water, 30 mL of ethanol, and 1.5 mL of ammonia were introduced, followed by continuous mechanical stirring for 2 h. The final product was washed with water and ethanol, filtered, and dried at 60°C for 24 h. The resulting powder was then calcined at 450°C for 3 h in a muffle furnace (see Fig. 2 for synthesis flowchart). Ultrasound-Assisted Impregnation Synthesis of Magnetically Recoverable 6 wt% Fe2O3/TiO2 Photocatalyst. Photocatalyst characterization VSM analysis The magnetic properties of the synthesized Fe2O3/TiO2 photocatalyst were characterized using a vibrating sample magnetometer (VSM) at 298 K. The applied magnetic field was swept to evaluate the saturation magnetization, remanent magnetization and coercivity of the sample, which are key indicators of its suitability for magnetic separation and redispersion in continuous flow applications. XRF analysis The composition of the synthesized photocatalyst was investigated in detail using X-ray fluorescence (XRF) analysis. As can be seen in Table 1, this material is mainly composed of two main components: titanium dioxide (TiO2) and iron oxide (Fe2O3). Quantitative analyses showed that TiO2 and Fe2O3 accounted for about 91.88 wt% and 6.02 wt% of the total composite, while other elements were present only in minor amounts. The dominance of TiO2 in the composite structure confirms its role as the main photoactive semiconductor. Titanium dioxide is of great interest in the field of photocatalysis due to its high chemical stability, non-toxicity, and ability to absorb light and produce reactive oxygen species. These properties make TiO2 a key and effective component in various photocatalytic studies and applications [33]. The controlled incorporation of Fe2O3 (6.02 wt%) provides the intended magnetic functionality, which is a recognized strategy for enabling effective magnetic recoverability of nano catalysts [32]. Furthermore, the introduction of iron species can enhance the catalytic performance by potentially reducing the recombination rate of photogenerated electron-hole pairs [34]. The balanced composition ensures good magnetic recovery while preserving optical and catalytic properties for photocatalysis. XRF and VSM results align, confirming successful synthesis of the magnetic photocatalyst, validating the strategy for environmental uses. Table 1Chemical composition of the synthesized Fe2O3/TiO2 catalyst obtained by XRF analysis.ComponentContent (wt.%)TiO291.88Fe2O36.02Other Components2.10●SiO20.84●CaO0.45●Al2O30.19●Others0.62**Sum of MgO, Na2O, K2O, MnO, and trace elements (Total purity ≈ 97.9 wt%). Chemical composition of the synthesized Fe2O3/TiO2 catalyst obtained by XRF analysis. *Sum of MgO, Na2O, K2O, MnO, and trace elements (Total purity ≈ 97.9 wt%). Kinetic analysis and performance evaluation Pseudo-first-order kinetic model The degradation of methylene blue (MB) with photocatalysts like TiO2 and Fe2O3/TiO2 follows pseudo-first-order kinetics, based on the Langmuir-Hinshelwood mechanism [35], [36]. This model assumes the reaction occurs at the photocatalyst surface, where pollutants interact with the solid catalyst [18]. The pseudo-first-order approximation is valid because the concentration of reactive oxidative species (e.g., •OH radicals) remains steady, while MB concentration stays relatively constant [35]. The process is heterogeneous-homogeneous, involving liquid-phase pollutants and the solid catalyst surface. Under stable hydrodynamic conditions, the residence time was constant and used as the effective reaction time [37]. While absolute rate constants in conventional units aren't directly measurable, the apparent rate constant (kapp), derived from residence time, provides a reliable metric for comparing reaction efficiency across different conditions [38]. The kinetic equation for the pseudo-first-order model is expressed as [19]: where C0 is the initial MB concentration in the liquid phase (mg·L−1), C is the MB concentration after the sonophotocatalytic reaction (mg·L−1), t is the residence time (s), and kapp is the apparent rate constant (s−1). In this study, the measured residence times were 60 s and 50 s, respectively, while the flow rates were 20 and 40 mL·min−1. This nonlinear decrease in residence time with doubling the flow rate is mainly due to the effect of the continuous gas phase (oxygen) and the converging–diverging channel geometry on the effective reaction volume (Vr) and the complex hydrodynamic patterns in the three-phase system. These effects must be carefully modeled through the effective volumetric flow rate (Q). Using equation (3) and substituting the experimental values of C0 and C, the apparent rate constants were calculated and applied to analyze the influence of different operational parameters on the performance of the triphasic sonophotocatalytic microreactor. Apparent overall removal coefficient Microreactors (MRs) play a central role in modern process intensification strategies, owing to their unique geometrical features, particularly their extremely small characteristic dimensions and exceptionally high surface-to-volume ratios (SVR) [39]. This structural advantage facilitates remarkably enhanced transport phenomena, both mass and photon transfer, crucial for efficient photocatalysis [10]. Despite this inherent advantage, microreactors often face mass transfer limitations, mainly because the low flow rates typical of microchannels generate a laminar flow regime, where mass transport relies primarily on molecular diffusion [22]. Therefore, a rigorous approach is mandatory to accurately quantify the transport capacity of the system, achieved here by employing the apparent overall removal coefficient (kORC) as the primary performance indicator [2]. To decouple the effects of physical transport from intrinsic chemical kinetics, the overall process is analyzed based on three key assumptions: (1) steady-state operation of the continuous-flow reactor, (2) the degradation of the model cationic pollutant, methylene blue (MB), is typically approximated using pseudo-first-order kinetics derived, especially at low substrate concentrations [36], [40], and (3) to isolate the mass transfer rate contribution, the concentration profile within the reactor volume must be mathematically linearized using the logarithmic mean driving force [38]. Under these premises, the apparent overall removal coefficient, kORC (min−1) can be calculated from the concentrations of MB at the reactor inlet (Ci) and outlet (C) using the following expressions [6]. It should be emphasized that the volumetric mass transfer coefficient (kLa) is strictly defined for the interphase transfer of a non-reactive species (e.g., oxygen transfer from the gas phase to the liquid phase). In the present three-phase reactive system, methylene blue removal does not represent a pure mass-transfer process; rather, it results from the combined effects of adsorption on the catalyst surface, photocatalytic degradation, sonochemical radical reactions, and enhanced catalyst dispersion. Therefore, the parameter used in this study is defined as an apparent overall removal coefficient (kORC), which serves as a lumped performance metric representing the overall intensification of transport and reaction processes occurring within the system: Here, kORC is apparent overall removal coefficient, ΔC represents the concentration drop of MB across the reaction volume, and Vr is the total reaction chamber volume (m3). The logarithmic mean driving force,C0-CM, is given by [38]: In equations (4), (5), C and C0 represent the inlet and outlet concentrations of methylene blue in the microreactor, respectively, measured directly from the feed and effluent streams. The parameter C* shows the saturation concentration of methylene blue under operating conditions and can be determined experimentally or for each set of conditions through a pre-equilibrium protocol: before the sonophotocatalytic process was carried out, an aqueous solution of methylene blue was mixed with the photocatalyst and oxygen at a standard volume ratio of 0.2:1. The mixture was stirred in the dark without ultrasound, and effluent samples were collected at 15 and 30 min. The measured concentrations at these time points were designated C (15 min) and C* (30 min). The slight decrease in concentration at 30 min compared to 15 min confirmed that the system was approaching equilibrium. Since the 15 min dark contact period is much longer than 60 s reaction residence time, and the majority of adsorption occurred within the first 15 min, the concentration measured at this point served as the baseline for all calculations. The reported results represent net photocatalytic and sonochemical degradation. This method ensures that the driving force used in the kORC calculation accurately reflects the true equilibrium of the system. In addition to its role in determining C*, the controlled introduction of a gas (such as O2) transforms the reaction medium into a dynamic three-phase (gas–liquid–solid) system. This is of great importance because the continuous supply of oxygen acts as a strong electron absorber, reducing electron-hole recombination and significantly enhancing photocatalytic kinetics, a feature not observed in traditional two-phase liquid–solid systems [35].The presence of microbubbles, however, alters the hydrodynamic regime, introducing mechanical perturbation (Microturbulence and Surface Self-Refreshment) while simultaneously complicating the volumetric flow rate calculation [41]. Consequently, in equation (4) the effective volumetric flow rate (Q) is given by [35]: In this equation, η represents the nominal flow rate delivered by the syringe pump (m3·s−1), while x/y denotes the volumetric gas-to-liquid flow ratio. This ensures that the calculated apparent overall removal coefficient accurately reflects the true hydrodynamic conditions of this highly intensified system, which is essential for separating mass transfer effects from intrinsic reaction kinetics and for obtaining a quantitative understanding of the reactor’s performance. Results and discussion Magnetic properties of the Fe2O3/TiO2 photocatalyst The magnetic hysteresis loop of the synthesized Fe2O3-TiO2 nanocomposite, measured by VSM at 298 K, is presented in Fig. 3a. The saturation magnetization (Ms) was determined to be 2.97 emu/g, which is lower than that of pure Fe2O3, as expected, due to dilution by the non-magnetic TiO2 matrix. However, this level of magnetization was sufficient for efficient separation of the catalyst from the aqueous suspension using a conventional magnet, as shown in Fig. 3b. The remanent magnetization (Mr) and coercivity (Hc) values were recorded as 0.51 emu/g and 140 Oe, respectively (Table 2). The low Mr/Ms ratio (0.17) indicates soft magnetic behavior with almost superparamagnetic properties. This soft magnetic property is of great importance for use in continuous flow microreactors, as it prevents permanent particle aggregation after the magnetic field is removed. Therefore, the catalyst can be easily redispersed in fresh solution, preventing activity loss in subsequent reaction cycles [42]. Collectively, these magnetic properties make the Fe2O3-TiO2 nanocomposite a promising candidate for environmental relevant processes, combining facile magnetic recovery with good colloidal stability. This aligns with reports on similar magnetic composites demonstrating both notable photocatalytic performance and recyclability [32]. Magnetic behavior and Separation Efficiency of 6 wt% Fe2O3/TiO2 Photocatalyst (a) VSM Hysteresis Loop at 298 K (Ms = 2.97 emu/g, Hc ≈ 140 Oe), (b) Visual demonstration of magnetic separation capability using external magnet. Table 2Magnetic parameters of Fe2O3-TiO2 photocatalyst.ParameterValueUnitSaturation magnetization (Ms)2.97emu/gRemanent magnetization (Mr)0.51emu/gCoercivity (Hc)140OeSquareness ratio (Mr/ Ms)0.17− Magnetic parameters of Fe2O3-TiO2 photocatalyst. Mechanistic basis of sonophotocatalysis in triphasic microreactor systems The efficiency of heterogeneous photocatalytic oxidation in gas–liquid–solid systems is often limited by interfacial mass transfer, particularly the transport of organic reactants, such as methylene blue (MB), and dissolved oxygen to the catalyst surface [43]. The simultaneous application of power ultrasound and ultraviolet irradiation, here defined as the sonophotocatalysis (SPC) mode, has been widely employed to mitigate these transport limitations through the combined effects of acoustic cavitation and photonic activation. The selection of a high operating frequency (1.7 MHz) in this study is pivotal, as it fundamentally shifts the cavitation regime compared to conventional low frequency ultrasound (20–100 kHz) [13], [25]. At low frequencies, longer acoustic cycles allow cavitation bubbles to grow larger through rectified diffusion, leading to violent inertial collapse that generates intense local hotspots with extreme temperatures and pressures [13]. While this regime is highly effective for molecular degradation, it is less efficient for bulk mixing and may cause catalyst erosion in micro scale systems [13], [25]. In contrast, the short acoustic period at 1.7 MHz limits bubble growth, resulting in a dense cloud of small bubbles undergoing stable, non-inertial oscillations [25]. This transitions the dominant physical effect from violent collapse to intense acoustic microstreaming. This dense cloud of oscillating microbubbles acts as a distributed array of microscopic stirrers [13]. The microstreaming fields generated induce strong shear forces that effectively disrupt the concentration boundary layer around the photocatalyst particles, enhancing the convective transport of reactants to the active sites and substantially increasing the overall removal [11], [26]. The significant increase in kORC, reaching up to 2470 min−1, highlights that microstreaming-induced convective transport offers a much more effective overall removal in continuous-flow microreactors than the turbulence generated by cavitation at lower ultrasonic frequencies [25]. Furthermore, the milder conditions at 1.7 MHz better preserve the structural integrity of the magnetic Fe2O3/TiO2 photocatalyst, ensuring long-term stability in continuous operation [14]. The degradation of methylene blue in this sonophotocatalytic microreactor is governed by four distinct mechanisms operating in concert, each supported by specific experimental evidence. First, adsorption/desorption effects were isolated by a 30-minute dark pre-equilibration step performed before each experimental run, regardless of the irradiation mode. Second, UV-only experiments (Fig. 4, Fig. 5) isolated the photocatalytic pathway, driven by electron-hole pair generation and sustained by continuous oxygen supply, which prevents charge recombination [38], [44]. Third US-only experiments (Fig. 4, Fig. 5) demonstrated that sonochemical radical formation is catalyst-independent, with acoustic cavitation producing additional •OH radicals through water pyrolysis [24]. Finally, mass transfer, measured by kORC, couples transport to reactions. Acoustic microstreaming at 1.7 MHz disrupts the boundary layer, accelerating reactant delivery to the catalyst surface and intensifying both radical generation routes. The observed synergy under UV + US conditions thus arises from dual-source ROS production, driven by enhanced mass transfer and measured free of adsorption contributions. Comparative evaluation of different irradiation type (UV, US, and combined UV + US) on the process performance in a triphasic microreactor under fixed conditions: liquid flow rate = 20 mL·min−1, catalyst dosage = 0.1 mg·L−1, and initial MB concentration = 50 ppm. (a) MB removal efficiency, (b) apparent rate constant (kapp), and (c) apparent overall removal coefficient (kORC). Dye removal performance as a function of photocatalyst dosage, irradiation mode (UV, US, and UV + US), flow rates, and initial concentrations of methylene blue. Symbols represent: ♦ UV + US at 20 ml·min−1, ■ UV at 20 ml·min−1, ▲US at 20 ml·min−1, × UV + US at 40 ml·min−1, ж UV at 40 ml·min−1, ● US at 40 ml·min−1. Subfigures (a), (b), and (c) correspond to initial MB concentrations of 30, 50, and 60 ppm, respectively. In parallel to the acoustic effects, the UV-induced photonic pathway activates the semiconductor catalyst by generating electron-hole pairs (e−/h+), which subsequently form reactive oxygen species (ROS), most notably hydroxyl radicals (•OH), responsible for oxidative degradation of MB [44]. Building on the distinct pathways identified above, applying both energy sources simultaneously led to marked improvements in reactant transport and reaction performance. As shown in Fig. 4c, the combined UV + US mode outperformed both standalone processes across all three performance indicators. Notably, the apparent overall removal coefficient (kORC) under sonophotocatalytic (SPC) conditions reached 1718.41 min−1, exceeding the values measured for standalone photocatalysis (PC, 1614.09 min−1) and sonolysis alone (S, 1360.33 min−1). While kORC lumps together transport and reaction contributions, the enhancement arises from the synergistic action of all mechanisms. This increase highlights the role of acoustic microstreaming, which accelerates convective transport and intensifies both the photocatalytic and sonochemical radical generation routes. In addition, ultrasound helps keep the catalyst well-dispersed, prevents particle aggregation, and continuously removes adsorbed intermediates from the catalyst surface, collectively ensuring sustained catalytic activity over time [19]. The photochemical pathway also benefits indirectly from acoustic effects: improved O2 transport to the catalyst surface promotes electron scavenging, thus diminishing e−/h+ recombination and supporting higher photonic utilization efficiency [41]. The enhanced performance is evident in both the apparent reaction rate constant (kapp) and the apparent overall removal coefficient (kORC). Under SPC conditions, kapp reached 0.33 min−1, exceeding the values measured for standalone photocatalysis (PC, 0.29 min−1) and sonolysis (S, 0.21 min−1) under similar operating conditions (Fig. 4b). These measurements were taken at a fixed flow rate of 20 mL·min−1, a catalyst dosage of 0.1 mg·L−1, and an initial methylene blue concentration of 50 ppm. Further optimization, discussed in Section 4.6, revealed that significantly higher performance up to 44.3% removal efficiency with kORC = 2470 min−1 can be achieved by fine-tuning both operational and acoustic parameters. In the unique hydrodynamic and optical conditions of the SPC system, the observed reaction rate represents the combined contributions of cavitation-driven mass transport and photocatalytic activation. Moreover, the way cavitation bubbles are distributed and their density during ultrasonic irradiation can scatter and partially absorb UV light, affecting the actual photon flux that reaches the catalyst surface [41]. In parallel, the coexistence of multiple ROS generation pathways may alter radical lifetimes and distribution within the microreactor environment, influencing the overall oxidative activity, as observed in complex photocatalytic systems like Fe3O4/TiO2 nanocomposites [32]. These interacting physical and chemical phenomena collectively determine the observed kinetic behavior of the SPC system. Moreover, the high operating frequency of 1.7 MHz used in this study fundamentally alters the cavitation behavior compared to conventional ultrasonic systems operating at 20–100 kHz. At these megahertz frequencies, cavitation tends to be dominated by non-inertial effects, resulting in smaller bubble sizes, higher bubble densities, and more stable oscillatory motion [25]. This frequency-dependent behavior significantly reduces violent inertial collapse while promoting uniform microstreaming patterns around the oscillating bubbles [13]. The dense cloud of stably oscillating bubbles generates intensive microstreaming throughout the reactor volume, effectively reducing the concentration boundary layer around photocatalyst particles and enhancing mass transfer rates in the sonophotocatalytic system [22]. The significant increase in the apparent overall removal coefficient (kORC, reaching up to 2470 min−1) observed in this system is one manifestation of the cavitation behavior optimized at high frequency, where microstreaming-driven transport predominates over the collapse-induced effects typical of lower ultrasonic frequencies. Overall, the combined UV and ultrasound (UV + US) process functions as an intensified oxidation strategy in triphasic microreactor systems: the acoustic field alleviates external mass transfer limitations, while UV irradiation sustains the photochemical pathways necessary for reactive oxygen species (ROS) generation. The synergistic effect of these two mechanisms, as illustrated in Fig. 4, leads to markedly higher degradation performance than either photocatalysis or sonolysis alone, demonstrating the effectiveness of sonophotocatalysis under the tested operating conditions [18]. Influence of operational parameters on degradation efficiency The efficiency of hybrid degradation processes is highly sensitive to the careful adjustment of operational parameters, including photocatalyst dosage, initial pollutant concentration, and liquid flow rate. In this work, a series of experiments were conducted to systematically investigate these effects. Liquid flow rates of 20 and 40 mL·min−1 were tested to assess the influence of residence time and shear conditions, while photocatalyst loadings of 0.1, 0.2, and 0.4 mg·L−1 allowed evaluation of how the available active surface area impacts reaction kinetics. Additionally, initial methylene blue (MB) concentrations of 30, 50, and 60 ppm were employed to explore the effect of substrate availability on degradation efficiency. The results, presented in Fig. 5, highlight the combined influence of these parameters on both reaction rates and mass transfer characteristics. Fig. 5 presents the results of the three different processes: US-only data (▲, ●) isolate the sonochemical radical generation routes, UV-only data (■, ж) show the net photocatalytic contribution, and the UV + US data (♦, ×) capture the synergistic coupling of both. During all experiments, ultrasonic waves were applied continuously and in synchrony with the gas and liquid phases throughout the microreactor. The PZTs generated localized high-frequency vibrations that induced microstreaming, promoted acoustic cavitation, and enhanced mixing in the adjacent microchannels. These effects not only enhanced the convective transport of reactants to the catalyst surface but also prevented particle agglomeration, ensured consistent catalyst dispersion, and enabled the continuous removal of adsorbed intermediates. Together, these mechanisms led to a more efficient and stable photocatalytic degradation process under the applied operating conditions [26]. The analysis of the experimental data consistently indicates that increasing the liquid flow rate from 20 to 40 mL·min−1 results in a measurable reduction in removal efficiency (Fig. 5a-c). This phenomenon is principally linked to the inverse relationship between the flow velocity and the residence time of the reactants within the microreactor [45]. As the flow velocity increases, the reduced residence time curtails the necessary duration for effective contact between the MB molecules and the photocatalyst surface, thereby limiting the overall extent of surface adsorption and subsequent catalytic oxidation [46]. While increasing the flow rate can theoretically enhance the convective mass transfer rate, the detrimental effect of diminished contact time acts as the dominant kinetic bottleneck, overriding the advantages of improved mixing in the tested range [47]. Similarly, the initial concentration of methylene blue (MB) plays a key role in balancing photon utilization and radical consumption. The results show that reducing the initial MB concentration significantly enhances the removal efficiency (Fig. 5a and c). This improvement is attributed to the increased ratio of active radical species (•OH and O2∙) to pollutant molecules, which is essential for sustaining high degradation rates [48]. Conversely, at elevated initial concentrations, the increased number of dye molecules occupying the catalyst surface leads to surface site saturation, effectively suppressing the primary photocatalytic reactions [49]. Furthermore, during photo-assisted degradation, higher MB concentrations amplify the light-screening effect, in which dye molecules absorb incoming UV and visible light before it can reach and activate the catalyst surface [46]. This reduction in effective photon flux directly hinders the photogeneration of electron-hole pairs and subsequent production of •OH radicals [44]. Concomitantly, high initial concentrations generate a higher load of intermediate by-products which then compete with the parent MB molecule for the limited supply of active radicals, contributing further to the overall decrease in efficiency [50]. The photocatalyst dosage demonstrates a clear positive correlation with degradation efficiency within the operational limits of this study, as evident across all panels of Fig. 5. The increased dosage (from 0.1 to 0.4 mg·L−1) provides a greater availability of active surface sites for MB adsorption and enhances the overall efficiency of photon capture [51]. In the hybrid sonophotocatalytic system, this increase in particulate load is doubly advantageous, as the solid particles act as additional nucleation sites for acoustic cavitation, thereby augmenting the frequency of bubble collapse and enhancing the physical generation of •OH radicals [8]. Although the catalyst dosages examined in this study fall within a relatively low range, it is important to note that while an increasing trend in performance was observed across this range, excessively high dosages generally lead to a decrease in efficiency. This decline is attributed to two concurrent effects: (i) particle agglomeration, which reduces the available active surface area, and (ii) increased turbidity and light scattering, which limit the penetration of light into the bulk solution. Together, these effects create an optical barrier similar to the light-screening phenomenon seen at high dye concentrations [51]. Notably, the effects of all tested operational parameters were more pronounced under sonophotocatalytic conditions, highlighting the synergistic benefits of the combined approach [45]. The continuous acoustic field provides essential mechanical energy that enhances overall removal by generating acoustic microstreaming and turbulence at the fluid–solid interface [11], [26]. Furthermore, the acoustic field prevents catalyst deactivation by mitigating particle agglomeration and continually cleaning the active surface sites [19]. Overall, these results confirm that the complementary action of ultrasoundenhancing mass transfer through acoustic microstreaming, preventing catalyst deactivation, and providing an independent source of radicals establishes an efficient, multi-mechanism platform for methylene blue degradation [40]. Statistical data analysis, regression modeling, and quantitative assessment of process enhancement Regression model and statistical analysis All statistical analyses and numerical computations in this study were carried out using Python (version 3.9.12) within the Google Colab environment. Advanced numerical operations and array management were handled with the NumPy library (version 1.22.3), while scientific data visualization and calculation of descriptive statistics such as mean and standard deviation were performed using Matplotlib (version 3.5.1). The Standard Error of the Mean (SEM) was also computed to evaluate data reliability and quantify experimental uncertainty. The experimental design followed a full factorial approach, examining four independent variables. To ensure rigorous statistical evaluation and accurately estimate pure experimental error, replications were assigned to specific design points- including a center point- according to the experimental design generated by Design Expert. A total of 81 experimental runs were performed, comprising six conditions with n = 4 and three conditions with n = 3. After averaging the replicates, the resulting 54 mean observations were used for ANOVA and MLR modeling. A MLR model was subsequently developed to quantify the individual contribution and directional effects of the operational variables on the removal efficiency (RE). The model’s statistical significance and predictive power were rigorously evaluated using Analysis of Variance (ANOVA). The derived MLR model, which established the combined sonophotocatalytic process (UV + US) as the baseline scenario, is presented in equation (7): Where: RE = Removal efficiency (%); X1 = Flow rate (mL·min−1); X2 = Methylene blue concentration (ppm); X3 = Photocatalyst dose (mg·L−1); X4 = Radiation type (UV); X5 = Radiation type (US). The “Radiation type” was treated as a categorical variable using indicator (dummy) coding, where a value of 1 was assigned for the specified type and 0 otherwise. The variable names reported in the ANOVA results (Table 3) correspond directly to the predictors defined in the regression equation. The coefficients represent the model intercept and the primary effects derived from the Ordinary Least Squares (OLS) regression. Table 3ANOVA results for the developed OLS regression model.SourceDFSum SqMean SqF-valueP-valuemodel52384.14476.83148.36< 0.001C (Radiation Type)2.01185.07592.53184.38< 0.001Catalyst Dose1.0566.09566.09176.15< 0.001MB Concentration1.0435.18435.18135.41< 0.001Flow Rate1.0197.80197.8061.55< 0.001Residual48.0154.243.21Lack of Fit46.0153.433.348.24< 0.05Pure Error20.810.405R2 = 0.939Adjusted R2 = 0.933RMSE = 1.79Residual SD = 1.79Pure Error SD = 0.64 ANOVA results for the developed OLS regression model. Analysis of variance (ANOVA) The ANOVA results (Table 3) indicate that all examined operational variables had a statistically significant effect on methylene blue removal efficiency, as reflected by their very low p-values (p < 0.001). The relative influence of each factor was assessed and ranked according to the magnitude of its F-statistic. Among all factors, the type of radiation was found to have the greatest impact, with the highest F-statistic of 184.38 (p < 0.001). This clearly reinforces the study’s main conclusion that the combined sonophotocatalytic process (UV + US) is vital for maximizing process intensification, as it simultaneously enhances mass transfer and radical generation. The catalyst dose was the next most influential factor (F = 176.15, p < 0.001), emphasizing the pivotal role of the solid phase in providing active sites and facilitating cavitation nucleation. Both initial pollutant concentration and flow rate also demonstrated highly significant effects (F = 135.41 and 61.55, respectively; both p < 0.001), confirming their importance as key operational parameters. The analysis of variance decomposition further reveals a statistically significant Lack of Fit (F = 8.24, p < 0.05). This indicates that, although the linear model is highly significant and successfully captures the main trends, it does not account for all the systematic variation present in the data. The standard deviation of the pure error (0.64), which represents the inherent experimental variability, is considerably lower than the standard deviation of the residuals (1.79). This suggests that the observed lack of fit may result from model misspecification, possibly due to unaccounted interaction effects or subtle non-linearities. Nevertheless, the primary aim of this study was to identify and rank the key operational parameters. The high R2 value (0.939) and the random, pattern-less distribution of residuals in the diagnostic plots (Fig. 6a) confirm that the model is more than adequate for this purpose, providing a robust and interpretable tool for factor screening and optimization. The investigation of interaction effects was separately addressed through Pearson correlation analysis (Section 4.4.2), aligning with the screening objective of this work. Diagnostic plots for the validated multiple linear regression model: (a) Residuals versus. predicted values plot, demonstrating random scatter and validating linearity and homoscedasticity; (b) Normal probability plot (Q-Q plot) of the residuals, confirming their normal distribution. Model validation: residuals analysis Prior to further interpretation of the regression model, the underlying statistical assumptions including linearity, constant variance (homoscedasticity) of the residuals, and normality of residuals were carefully examined. Validating these assumptions is essential to ensure the reliability of p-values, confidence intervals, and the overall inferential power of the model. Fig. 6a presents a plot of the residuals versus the predicted removal efficiency. The absence of any discernible patterns (such as curvature or funnel shapes) and the random scatter of residuals around zero provide strong confirmation that the assumptions of linearity and constant variance are satisfied [52]. This indicates that the model has successfully captured the underlying trend in the data without missing any systematic information. Furthermore, the normality of the residuals was assessed using a Normal Probability Plot (Q-Q plot), presented in Fig. 6b. The strong linear alignment of the data points with the theoretical diagonal line provides compelling evidence that the residuals follow a normal distribution. This satisfies a critical assumption underlying the validity of the F-tests and t-tests presented in the ANOVA. Taken together, these diagnostic checks confirm that the developed MLR model is statistically sound, robust, and reliable for subsequent interpretation and process optimization. Analysis of statistical coefficients and rate-limiting mechanisms While the ANOVA results confirm the statistical significance and relative impact of the operational parameters, examining the MLR coefficients (Equation 7) provides deeper mechanistic insights into the rate-limiting processes governing the degradation system. These coefficients indicate both the direction (positive or negative) and the strength of the relationship between each variable and the removal efficiency. Among all continuous variables, catalyst dosage (X3) exhibits the largest positive coefficient (+47.48), clearly demonstrating that increasing catalyst loading is the most effective strategy for enhancing removal efficiency in this microreactor. Mechanistically, this strong positive effect arises from the greater availability of active surface sites, which is essential for efficient photogeneration of electron-hole pairs and the subsequent formation of •OH radicals [53]. The analysis also implies an operational kinetic window, as excessively high doses are known to lead to limitations such as particle agglomeration and the light screening effect, which jointly reduce the effective surface area and incident photon flux [54]. Conversely, both flow rate and initial concentration exhibit negative correlations with removal efficiency: the initial pollutant concentration (X2) shows a significant negative coefficient (−0.36), confirming that high concentrations impose severe limitations, primarily driven by active site saturation and the profound inner-filtering effect (Light Screening Effect) [39]. At higher concentrations, methylene blue molecules absorb a portion of the incident radiation before it can penetrate and activate the deeper layers of the semiconductor, thereby impeding the primary activation step of the catalyst [55]. The liquid flow rate (X1) has the smallest, yet significant, negative coefficient (−0.51), reflecting that it drastically diminishes the contact time between the pollutant and the catalyst surface, thereby curtailing the overall reaction opportunity and restricting conversion [39]. Furthermore, the notable coefficients for the indicator variables (X4 = –21.71 and X5 = –13.27) provide statistical confirmation of the enhanced performance of the combined process, demonstrating that the superior efficiency of the sonophotocatalytic mode stems from the acoustic field alleviating mass transfer limitations and preventing catalyst deactivation. Validation of process enhancement and mechanistic basis for intensification A comparative analysis of the model’s indicator variables for the individual processes (UV-only: X4, US-only: X5) relative to the combined UV + US system highlights the clear advantage of the hybrid approach. The substantial negative coefficients, especially X5 (–21.71), reflect the inherent kinetic limitations of the standalone advanced oxidation processes, consistent with experimental observations showing very low degradation efficiencies (∼4%) for the individual photolysis and sonolysis treatments [40]. The results indicate that the sonophotocatalytic system achieves a degradation rate that surpasses what would be expected from the sequential or isolated application of the individual techniques [56]. This enhanced performance represents a true example of process intensification, where the acoustic field effectively alleviates mass transfer limitations and mitigates catalyst deactivation, challenges typically associated with conventional heterogeneous photocatalysis [57]. This intensification is driven by several interconnected mechanisms. The mechanical forces generated by acoustic cavitation, involving the formation and rapid collapse of microbubbles, play a central role. This process induces strong acoustic streaming and localized microstreaming near the solid–liquid interface [18], [26]. These hydrodynamic effects significantly reduce diffusion-layer resistance, enhancing mass transfer and accelerating the transport of reactant molecules (MB) to the catalyst’s active sites [11]. At the same time, the mechanical energy from the ultrasound helps prevent catalyst agglomeration and continuously cleans the catalyst surface, maintaining its activity throughout the reaction [18], [40]. This action helps maintain a high effective surface area for light absorption, maximizes photon utilization efficiency, and delays catalyst deactivation [58]. The enhancement is further reinforced by the complementary generation of reactive hydroxyl radicals (•OH) through both photocatalytic and sonochemical pathways [40]. While photocatalysis produces •OH radicals upon UV irradiation, sonolysis provides an auxiliary source via the thermal decomposition of water vapor within the high temperature cavitation hotspots [59]. Together, these mechanisms—supported by quantitative statistical evidence—explain the superior performance of the sonophotocatalytic system and highlight the role of ultrasound as a powerful tool for process intensification in continuous-flow microreactors [47]. Pearson correlation matrix analysis of operational parameters and MB removal efficiency The interrelationships among key operational parameters were examined by calculating the Pearson correlation matrix using Python’s pandas library, based on data from 54 experimental runs (Fig. 7). This quantitative analysis is essential for process optimization, particularly in advanced oxidation processes (AOPs), and aligns with established statistical approaches such as Response Surface Methodology (RSM) and Box-Behnken Design (BBD), which are widely used in photocatalytic research [60]. For each variable, the mean and standard deviation (SD) were calculated to assess data variability and ensure statistical robustness. The methylene blue (MB) concentration ranged from 30 to 60 ppm, with an average of 46.7 ± 15.3 ppm; the photocatalyst dosage varied between 0.1 and 0.4 mg·L−1, averaging 0.233 ± 0.153 mg·L−1; and the flow rate spanned 20 to 40 mL·min−1, with a mean of 30 ± 14.1 mL·min−1. Pearson correlation coefficients (r), which range from –1 to + 1, were calculated for each pair of variables to quantify the strength and direction of their linear relationships. To facilitate visualization and interpretation, the correlation matrix was presented as a color-coded heatmap using the seaborn library (version 0.11.2). The analysis revealed a significant positive correlation between photocatalyst dosage and MB removal efficiency (r = 0.47, p < 0.01), indicating that increasing the catalyst concentration enhances the number of active sites, promotes •OH radical formation, and accelerates the reaction rate. This positive correlation, consistent with previous studies, strongly supports the conclusion that catalyst loading is a key factor in improving removal efficiency [61]. This finding aligns with previous studies on integrated microreactors/continuous multiphase systems, which emphasize that increasing catalyst concentration in continuous multiphase systems is a more decisive factor than hydrodynamic parameters [62]. Conversely, MB concentration exhibited a strong negative correlation with removal efficiency (r = −0.41, p < 0.01), reflecting the detrimental effect of higher initial pollutant concentrations on process performance. This phenomenon is attributed to surface saturation of the catalyst at concentrations above 30 ppm, reducing effective access to active sites. The strong inverse relationship observed here (a relevancy factor < 0) is a consistent finding in the photocatalytic degradation of MB, where high initial feed concentrations lead to reduced efficiency [61]. Furthermore, the inherent toxicity and persistent nature of MB dye compounds contribute to this performance drop under high load conditions [60]. This observation is consistent with reports on reaction kinetics in nanostructured photocatalysts, indicating that adsorption-limiting mechanisms are the primary factor reducing removal efficiency under high pollutant load conditions [63]. The flow rate exhibited a modest negative correlation with MB removal efficiency (r = –0.28, p < 0.05), which is weaker than the correlations observed for MB concentration (r = –0.41) and catalyst dosage (r = +0.47). This indicates that, while flow rate does influence the process, its effect is secondary compared to the other two parameters. Nevertheless, the relatively small negative correlation suggests that the enhanced mass transfer arising from efficient hydrodynamic mixing can largely compensate for the reduced residence time at higher flow rates, particularly in microreactors with optimized channel geometries. These findings align well with previous studies on advanced microreactor designs, such as the work reported by Ghasemi et al [63]. The correlation matrix, presented in Fig. 7, identifies the main routes for process optimization, namely by increasing photocatalyst dosage, since this is the most effective method to enhance removal efficiency, and controlling pollutant concentration to prevent the saturation of the active surface and subsequent loss of efficiency. This analytical framework is very useful for the design of high-performance continuous treatment systems, in that it provides a quantitative understanding of the interactions among operational parameters, in addition to statistical support of experimental observations [63]. Pearson correlation matrix illustrating the linear relationships between key operational parameters and methylene blue (MB) removal efficiency (N = 54). Significance levels are denoted as p < 0.05 (*) and p < 0.001 (**). The correlation analysis and visualization (heatmap) were performed using Python (pandas and seaborn libraries, versions 1.4.2 and 0.11.2, respectively. Comparison of photocatalytic and mass transfer performance between TiO2 and Fe2O3/TiO2 catalysts The choice of photocatalyst is a critical factor in determining the overall effectiveness of a sonophotocatalytic microreactor [53]. In this context, we compared the performance of pure titanium dioxide (TiO2) with that of the iron oxide/titanium dioxide nanocomposite (Fe2O3/TiO2), as shown in Fig. 8. The comparison was based on two main metrics: (i) the removal efficiency of methylene blue (MB) across different catalyst loadings (Fig. 8a), and (ii) the apparent overall removal coefficient (kORC) (Fig. 8b). Results from the Pearson correlation analysis (Fig. 7) highlighted the catalyst dosage as the single most influential factor for enhancing MB removal in the continuous multiphase system (r = 0.47, p < 0.01), underscoring the importance of optimizing the solid phase for effective degradation performance. Building on this finding, which indicates that catalyst concentration is a more decisive factor than hydrodynamic parameters, the performance of both catalysts was evaluated across dosages ranging from 0.1 to 0.4 g∙L−1. Fig. 8a confirms that increasing the catalyst dosage significantly enhances removal efficiency for both photocatalysts [64]. From a heterogeneous kinetics perspective, this behavior is directly attributed to the greater availability of surface active sites, which in turn promotes the generation of hydroxyl radicals (•OH) and accelerates the photocatalytic degradation rate of the pollutant [40]. At all tested dosages, Fe2O3/TiO2 consistently exhibited superior performance compared to pure TiO2, showing an average increase of approximately 32% in removal efficiency. Although TiO2 remains the predominant photocatalyst for wastewater treatment due to its chemical stability and intrinsic photoactivity, the Fe2O3/TiO2 nanocomposite offers complementary functionalities [53]. The incorporation of Fe2O3 increases the specific surface area and the number of active sites, thereby improving the degradation rate of organic pollutants [40]. From an acoustic–chemical perspective, the heterogeneous catalyst also serves as an additional nucleation center, facilitating the formation of cavitation bubbles [58]. This enhanced nucleation intensifies cavitation activity and promotes the pyrolysis of H2O molecules, leading to the generation of highly reactive radicals, as the collapsing bubbles create localized “hotspots” [19]. Consequently, the degradation efficiency of sonophotocatalytic systems is substantially greater than that of single-mode photocatalysis [58]. This phenomenon highlights the critical role of the Fe2O3 phase in improving catalytic activity, particularly under ultrasonic excitation [49], [50]. This performance enhancement is also evident in Fig. 8b, which presents the apparent overall removal coefficient for both catalysts under identical operating conditions. The Fe2O3/TiO2 catalyst achieved a maximum kORC value of 3166.24 min−1, while TiO2 reached 2753.02 min−1 under the same conditions, representing a 15% increase. This improvement in kORC is a direct result of the interaction between the catalyst and the ultrasonic field [40]: oscillation and collapse of cavitation bubbles (micro-explosions) generate localized microstreaming, which greatly enhances hydrodynamic mixing and significantly reduces the mass transfer resistance within the concentration boundary layer surrounding the photocatalyst particles [17], [56]. The reduced resistance facilitates pollutant access to the active sites, thereby increasing the overall reaction rate. The combined improvement in mass transfer, along with the higher structural stability of the magnetic photocatalyst, ensures the robust performance of Fe2O3/TiO2 in continuous-flow systems [53]. Ultimately, the results clearly demonstrate that the use of Fe2O3/TiO2 nanocomposite represents a dual-optimization strategy in sonophotocatalytic microreactors: it simultaneously enhances the generation of reactive radicals through cavitation nucleation and mitigates the mass transfer limitations that typically govern continuous systems [17]. (a) Removal efficiency of methylene blue over TiO2 (■) and Fe2O3/TiO2 (♦) catalysts at a flow rate of 20 mL∙min−1 and an initial concentration of 30 ppm, evaluated at varying catalyst dosages. (b) apparent overall removal coefficient (kORC) for both catalysts under the same operating conditions. Optimization of piezoelectric configuration and ultrasound intensity Four piezoelectric plate transducers (PZT1–PZT4) were positioned beneath the four consecutive mixing chambers, numbered from upstream (PZT1) to downstream (PZT4). In this study, each “configuration” corresponds to the specific transducer or combination of transducers that were activated. For example, Configuration 1 indicates that only PZT1 was active, Configuration 2 activates only PZT2, and so on. Combined configurations (such as 1–2 or 1–2–3–4) denote simultaneous activation of the corresponding transducers. This approach allowed systematic evaluation of how the spatial position of activated piezos, individually or in combination, influences the acoustic field distribution and sono-photocatalytic performance. The performance enhancement in continuous-flow sonophotocatalytic microreactors is fundamentally dependent on the effective delivery of acoustic energy and the precise management of mass transport phenomena [65]. The experimental findings, summarized in Fig. 9a, clearly indicate that the spatial arrangement of the acoustic transducers had a greater impact on system performance than either the total power input or the number of active elements. When operated individually, the upstream PZT1 transducer delivered the highest single-element performance, with a removal efficiency of 17.7% and a kORC of 1620 min−1. This is attributed to the elevated pollutant concentration at the reactor inlet, which maximizes reaction opportunities. Combinatorial activation of transducers led to further improvements: the PZT2–PZT3 pair achieved 28.9% MB removal with a kORC of 2048 min−1, while the triple-element configuration PZT1–PZT2–PZT4 reached 37.83% degradation with kORC of 2309 min−1. The optimal performance was achieved by activating all four transducers (PZT1–PZT4) simultaneously, resulting in the highest MB removal efficiency of 44.3%, a peak apparent overall removal coefficient of 2470 min−1, and an apparent reaction rate constant (kapp) of 0.33 min−1. This exceptional activity arises from the finely tuned 1.7 MHz ultrasonic field, which creates overlapping acoustic zones and intense microstreaming, effectively overcoming mass transfer limitations and boosting degradation kinetics in the continuous-flow microreactor [15], [65]. This performance enhancement stems from synergistic physical and chemical mechanisms. The optimized configuration generates overlapping acoustic fields that intensify cavitation bubble collapse through constructive wave interference [22], significantly promoting the production of reactive radicals essential for contaminant degradation [54]. Concurrently, ultrasonic vibrations induce strong microstreaming that drastically enhances hydrodynamic mixing, reducing the concentration boundary layer thickness around photocatalyst particles and diminishing mass transfer resistance [22]. Systematic evaluation of sonochemical microreactor performance under optimized conditions (20 mL∙min−1, 0.4 mg∙L−1, 30 ppm). (a) Dual-axis plot demonstrating piezoelectric arrangement optimization, showing simultaneous enhancement of methylene blue removal efficiency (right y-axis) and apparent overall removal coefficient (left y-axis). (b) Removal efficiency versus number of active piezoelectrics. The importance of an optimized transducer arrangement is further supported by numerical simulations, which show that simply increasing the number of piezoelectric plate transducers in a microfluidic system does not necessarily produce a proportional increase in acoustic pressure or intensity. In fact, activating multiple transducers simultaneously can often result in complex, non-uniform pressure fields or even destructive interference between vibrations, highlighting the need for careful spatial configuration to maximize performance [66]. Our empirical optimization successfully overcomes acoustic field non-uniformity through precise geometric design. Furthermore, the strong positive correlation between number of active piezoelectrics and removal efficiency (16.4% to 44.3% with increasing transducer activation) is evident (Fig. 9b). These results establish that optimized spatial configuration and the collective contribution of piezoelectric elements are essential design parameters for maximizing mass transfer efficiency and degradation kinetics in continuous-flow sonophotocatalytic microreactors [18]. The geometric-energetic coupling demonstrates that performance enhancement arises primarily from improved ultrasonic-induced mixing and optimal acoustic field distribution rather than simply increasing actuator quantity [55], [56]. The induced microstreaming is paramount for diminishing mass transfer resistance through concentration boundary layer reduction, providing a fundamental basis for designing advanced wastewater treatment systems [11]. Comparative performance analysis and process intensification context Placing the sonophotocatalytic microreactor in the broader context of advanced oxidation processes (Table S3) highlights its unique contribution to process intensification. Notably, when compared with the Microfluidic Photo electrocatalytic (PEC) Reactor (Table S3, Row 5), our system demonstrates a competitive edge. The PEC reactor achieves 25–40% pollutant removal within extremely short residence times of 3–8 s, providing a benchmark for evaluating the efficiency and throughput of continuous-flow microreactor systems [39]. This performance exemplifies the continuous single-pass operation paradigm, where the performance metric shifts from the batch-centric goal of ultimate conversion to the intensification-centric goal of maximizing the intrinsic kinetic rate (kapp), enabled by the reactor's exceptionally high surface-to-volume ratio [67]. Our system achieves a carefully optimized balance between high reaction rates and practical operational feasibility. With a measured apparent rate constant (kapp) of 0.33 min−1, it ranks among the top-performing continuous-flow remediation systems in terms of kinetics. Notably, the 44.3% single-pass degradation is accomplished without applying any external electrical bias. This stands in contrast to PEC systems, which rely on an applied voltage to create an electric field that suppresses the rapid recombination of photogenerated electron-hole pairs, a fundamental limitation in conventional heterogeneous photocatalysis [39]. In our system, the role of the electric field is effectively supplanted by the optimized 1.7 MHz ultrasonic field, which provides a twofold synergistic enhancement. First, the acoustic waves intensify mass transfer by generating microstreaming and associated shear forces that drastically reduce the concentration boundary layer thickness around the photocatalyst particles, ensuring a continuous supply of pollutant to active sites and compensating for mass transport limitations [68]. Second, the ultrasonic energy itself, through phenomena such as piezoelectric effects or acoustic perturbation of the semiconductor space-charge layer, facilitates the separation of photogenerated charges, thereby suppressing electron-hole recombination and increasing the overall quantum yield [69]. This comparison highlights that our sonophotocatalytic microreactor successfully addresses a key challenge in process intensification. It outperforms conventional batch systems, which are often limited by low productivity, while avoiding the material and operational complexities associated with advanced PEC setups. By harnessing the synergistic effects of high-frequency sonolysis and photocatalytic oxidation within a continuous-flow framework, the reactor achieves a high degradation rate (kapp = 0.33 min−1) while maintaining operational simplicity. This combination establishes a practical and efficient approach for intensified water treatment. Conclusion This study demonstrates the successful integration of high-frequency ultrasound into a gas–liquid–solid photocatalytic microreactor, resulting in significant enhancement of mass transfer and overall reaction performance. The combined sonophotocatalytic process (UV + US) achieved a methylene blue degradation efficiency of 44.3% under optimal conditions, corresponding to a 1.13-fold improvement over photocatalysis alone and a 1.55-fold increase over sonolysis, confirming a clear synergistic effect. This improvement is supported by a substantial increase in the apparent overall removal coefficient (kORC = 2470 min−1) and the apparent reaction rate constant. The enhanced performance is attributed to ultrasound-induced microstreaming and acoustic turbulence, which reduce diffusion boundary layers, improve transport of reactants (MB and O2) to the catalyst surface, and suppress nanoparticle agglomeration, thereby maintaining active catalytic sites and improving photon utilization. In addition, the coupling of sonochemical and photocatalytic pathways enhances the generation of reactive oxygen species, particularly •OH radicals, contributing to accelerated degradation. A key outcome of this work is the identification of piezoelectric transducer (PZT) spatial configuration as a critical design parameter. Optimized placement of PZT elements enables a more uniform acoustic field distribution, which enhances cavitation activity and improves mass transfer efficiency in the microreactor. The use of a magnetically recoverable Fe2O3/TiO2 photocatalyst further improves operational feasibility, with a saturation magnetization of 2.97 emu·g−1 and low remanence (Mr/Ms = 0.17), enabling efficient magnetic separation, reuse, and reduced secondary processing costs. From a process evaluation perspective, statistical analysis using multiple linear regression (MLR) revealed that catalyst dosage positively influences removal efficiency, while higher flow rate and initial pollutant concentration negatively affect performance due to reduced residence time and light screening effects. These findings confirm that combined UV and US operation is essential for achieving maximum process intensification compared to individual processes. From a practical standpoint, the system is inherently scalable through a numbering-up strategy, where multiple microchannels are operated in parallel to increase throughput while preserving the intrinsic advantages of microfluidic reactors. This makes the platform promising for textile wastewater treatment, particularly for persistent dyes such as methylene blue that are poorly removed by conventional biological processes. However, real industrial wastewater contains complex mixtures of dyes and additives, and therefore validation using authentic effluents is necessary. In addition, several engineering challenges remain for industrial implementation, including optimization of energy consumption for combined UV/US operation, reduction of fabrication complexity of microreactors, ensuring uniform acoustic field distribution in parallelized systems, and long-term stability of the photocatalyst under real wastewater conditions. Overall, this work demonstrates that coupling high-frequency ultrasound with optimized piezoelectric transducer configuration is an effective strategy for process intensification in continuous-flow photocatalytic microreactors. The system achieves performance comparable to advanced electro/photoelectrocatalytic systems while avoiding the need for external electrical bias, offering a simpler and more flexible approach for wastewater treatment applications. CRediT authorship contribution statement Mahdieh Daneshvar: Writing – original draft, Software, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization. Arsalan Parvareh: Writing – review & editing, Validation, Supervision, Software, Methodology, Investigation, Formal analysis. Neda Azimi: Writing – review & editing, Supervision, Methodology, Investigation. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Footnotes Supplementary data to this article can be found online at https://doi.org/10.1016/j.ultsonch.2026.107925. Supplementary data The following are the Supplementary data to this article: Supplementary Data 1 Comparison with similar works. References Associated Data Supplementary Materials Supplementary Data 1 Comparison with similar works.