Harvesting Vibration Energy for Efficient Cocatalyst-Free Sonocatalytic H2 Production over Magnetically Separable Ultra-Low-Cost Fe3O4

The cavitation effect is an important geochemical phenomenon, which generally exists under strong hydrodynamic conditions. Therefore, developing an economical and effective sonocatalyst becomes a vital method in capitalizing on the cavitation effect for energy generation. In this study, we first report a novel Fe3O4 sonocatalyst that can be easily separated using a magnetic field and does not require any additional cocatalysts for H2 production from H2O. When subjected to ultrasonic vibration, this catalyst achieves an impressive H2 production rate of up to 175 μmol/h/USD (where USD stands for dollars), surpassing most previously reported mechanical catalytic materials. Furthermore, the ease and efficiency of separating this catalyst using an external magnetic field, coupled with its effortless recovery, highlight its significant potential for practical applications. By addressing the key limitations of conventional sonocatalysts, our study not only demonstrates the feasibility of using Fe3O4 as a highly efficient sonocatalyst but also showcases the exciting possibility of using a new class of magnetically separable sonocatalysts to productively transform mechanical energy into chemical energy.


Introduction
Addressing the pressing demand for energy and confronting environmental concerns necessitate the transition from conventional fossil fuels to renewable and eco-friendly energy alternatives.Among these alternatives, hydrogen (H 2 ) energy emerges as particularly promising, boasting advantages such as efficient combustion, abundant availability irrespective of fossil fuels, high energy content, lack of pollution, and superior efficiency.However, achieving environmentally friendly, user-friendly, and economically feasible hydrogen production remains a formidable challenge [1,2].Photocatalytic materials have emerged as promising contenders for hydrogen generation through light-driven processes [3][4][5][6][7][8].Despite their potential, challenges persist, notably in harnessing visible light effectively and optimizing electron-hole transfer rates, impeding their widespread adoption for large-scale hydrogen production.Similarly, electrocatalytic methods for hydrogen production, while efficient, face drawbacks such as energy consumption and the production of economically less viable by-products, hindering their extensive development [9][10][11][12][13].However, Materials 2024, 17, 1463 2 of 14 a promising alternative in the realm of catalysis is mechanical catalysis.This innovative approach operates without the need for light or electrical stimulation, instead relying on mechanical stimuli to induce physical or chemical electron transfer, thereby facilitating catalytic processes.Since its inception, mechanical catalysis has garnered significant attention and found applications across various catalytic fields due to its distinct advantages and versatility.By leveraging mechanical forces, such as shear, compression, or vibration, mechanical catalysis offers a unique pathway to enhance reaction kinetics and selectivity, circumventing the limitations associated with traditional catalytic methods.
Furthermore, the versatility of mechanical catalysis extends beyond conventional catalytic reactions.It has been explored in diverse areas, including sustainable energy production, environmental remediation, and chemical synthesis, showcasing its potential to address complex challenges across multiple domains.Moreover, the simplicity and scalability of mechanical catalysis make it an attractive option for industrial applications, promising efficient and cost-effective solutions for hydrogen production and beyond.Piezocatalytic H 2 production, as a subtype of mechanical catalysis, offers an intriguing avenue to substantially reduce catalysis costs by using mechanical energy, making it particularly attractive as it operates independently of light [14][15][16][17].Piezoelectric effects are indeed restricted to crystals lacking a center of symmetry, limiting the range of semiconductor catalytic materials that can utilize mechanical energy for catalysis.However, an innovative approach in the form of sonoluminescence offers a promising alternative.Unlike piezoelectric and ferroelectric effects, sonoluminescence does not rely on material symmetry, allowing for broader application possibilities.This unique phenomenon involves the generation of light through the collapse of bubbles in a liquid medium, where appropriately positioned electrons and holes on the energy bands can initiate a cascade of reactions.
Recent research has witnessed a growing interest in leveraging non-piezoelectric semiconductor materials for effective sonocatalysis [18][19][20].This trend indicates a shift towards exploring materials with diverse properties to overcome the limitations posed by crystal symmetry.Nevertheless, challenges persist in the preparation processes of these nanomaterials, which are often complex and yield unstable products.Additionally, the high rates of charge recombination associated with sonogenerated species present obstacles to the widespread application of these materials.Moreover, traditional heterogeneous sonocatalysts face issues regarding their separation from the reaction system post reaction, leading to potential losses and increased separation costs.This concern poses a significant barrier to the scalability and practicality of sonocatalyst development, hindering its transition from laboratory-scale research to real-world applications.Addressing these challenges requires a concerted effort to develop robust preparation methods for nanomaterials, improve their stability, and enhance their separation efficiency.Moreover, exploring novel catalytic architectures and materials with enhanced sonocatalytic properties could offer promising solutions to overcome existing limitations.By addressing these critical issues, the field of sonocatalysis can progress towards realizing its full potential as a viable and sustainable approach for various catalytic applications.
In light of the challenges associated with traditional sonocatalytic materials and their separation processes, the development of sonocatalytic materials tailored specifically for H 2 production without the need for co-catalysts becomes imperative.These materials should possess key attributes such as cost-effectiveness, simplicity in synthesis, non-toxicity, minimal separation expenses, and exceptional performance.One promising approach involves the utilization of magnetic sonocatalytic materials, which offer several advantages over conventional counterparts.The application of an external magnetic field for assistance in material separation presents a non-destructive, straightforward, and user-friendly technique compared to other methods of liquid/solid separation.This approach not only simplifies the separation process but also reduces costs and minimizes potential losses associated with traditional separation methods.Ferro-ferric oxide (Fe 3 O 4 ), commonly known as magnetite, exhibits a distinctive inverse spinel structure and occurs naturally, making it an attractive candidate for magnetic sonocatalytic materials.Notably, the synthesis cost of Fe 3 O 4 is lower compared to many previously reported sonocatalytic materials, enhancing its appeal for large-scale applications [21].Furthermore, Fe 3 O 4 has demonstrated wideranging applications in sonocatalytic degradation reactions, showcasing its potential as a versatile catalyst material.Despite its promising attributes and extensive applications in degradation reactions, surprisingly, there have been no reports on H 2 production through sonocatalysis using magnetic Fe 3 O 4 material.This presents an exciting opportunity for further exploration and research in the field of sonocatalysis, particularly in the context of hydrogen generation.By leveraging the unique properties of Fe 3 O 4 and exploring its catalytic potential in H 2 production, researchers can potentially unlock a novel and efficient pathway for sustainable hydrogen generation.
In this study, we first present the groundbreaking use of magnetic Fe 3 O 4 material in sonocatalytic H 2 production reaction.When exposed to ultrasonic vibration, this catalyst achieves a remarkable H 2 production rate of up to 175 µmol/h/USD.Additionally, this catalyst can be quickly separated using an external magnetic field and easily retrieved, showcasing its considerable promise for real-world applications.This pioneering work opens up new horizons for the application of magnetic Fe 3 O 4 in the field of sonocatalytic H 2 production, offering potential benefits in terms of cost-effectiveness and environmental sustainability.One of the key advantages of this catalyst is its ease of separation using an external magnetic field.This feature allows for quick and efficient retrieval of the catalyst from the reaction mixture, highlighting its practical utility for real-world applications.Moreover, the simplicity of the separation process contributes to the overall cost-effectiveness and feasibility of utilizing magnetic Fe 3 O 4 in sonocatalytic H 2 production.Furthermore, our pioneering work not only showcases the impressive performance of magnetic Fe 3 O 4 as a sonocatalyst but also opens up new avenues for its application in the field of hydrogen production.By leveraging the unique properties of Fe 3 O 4 and exploring its catalytic potential, we have laid the foundation for further advancements in sustainable H 2 production technologies.

Materials
All chemicals were obtained from commercial sources and used without further purification.Commercial magnetic Fe 3 O 4 powder was purchased from Tianjin Damao Chemical Reagent Co., Ltd.(Tianjin, China).Basic information of commercial magnetic Fe 3 O 4 powder is displayed in Table 1.The deionized water was used throughout the experiment.

Sample Characterization
The crystalline structure of the sample was analyzed using X-ray powder diffraction (XRD) with Cu Kα radiation (λ = 0.15418 nm) via a Bruker D8 instrument (Karlsruhe, Germany).Field emission scanning electron microscopy (SEM) from ZEISS (Oberkochen, Germany), and transmission electron microscopy (TEM) from JEM-2100F (Tokyo, Japan), were employed for visualizing the morphology and structure of the sample.The surface state and element composite of the samples were examined by X-ray photoelectron spectroscopy (XPS) using an XPS spectrometer (ESCALAB 250Xi, Thermo Fisher Scientific, Waltham, MA, USA) with an Al Kα X-ray source.Fourier transform infrared spectroscopy (FTIR) data were collected using a NICOLET IS10 spectrophotometer from Thermo Fisher Scientific (Waltham, MA, USA).Magnetic properties were characterized using a LakeShore7404 Vibrational Sample Magnetometer (VSM) instrument (Westerville, OH, USA).Thermogravimetric analysis (TGA) was carried out in a nitrogen environment with a METTLER TOLEDO TGA/SDTA851 instrument (Columbus, OH, USA).

Electrochemical Tests
Electrochemical impedance spectroscopy (EIS), Nyquist plots, and sonocurrent measurements were conducted on an electrochemical workstation (CHI 760E, Shanghai, China).The electrolyte used was a 0.2 M Na 2 SO 4 solution, with Pt serving as the counter electrode and Ag/AgCl as the reference electrode.To prepare the working electrode, 10 mg of the synthesized samples was uniformly dispersed in 1 mL of ethanol via ultrasound treatment.Subsequently, this suspension was evenly coated onto indium tin oxide (ITO) glass.The ultrasound frequency and ultrasound power used in the electrochemical tests were fixed at 50 kHz and 240 W.

Experimental Procedure
The sonocatalytic H 2 production performance of sonocatalysts was evaluated in an ultrasonic cleaner (Kunshan) with an adjustable power range of 120, 180, and 240 W and an adjustable frequency range from 50 to 75 kHz.The temperature of the reaction system was controlled at 25 • C (with a variation of ±2 • C) through the circulation of water and an ice bag to minimize the influence of temperature fluctuations.Typically, in a sealed Pyrex reactor with a total volume of 500 mL, 0.05 g of Fe 3 O 4 powder was added to a mixture of 180 mL deionized (DI) water and 20 mL methanol (MeOH) or Triethanolamine (TEOA) or glycol (EG).Subsequently, the reaction system was completely evacuated to eliminate any residual air, and internal and external pressures were equalized using high-purity (99.99%)N 2 for approximately 60 min.Following this, the suspension underwent ultrasonic treatment for sonocatalytic H 2 production reaction.To quantify the produced amount of H 2 , 1 mL of gas was collected from the reaction system at 30 min intervals and analyzed using a gas chromatograph (GC-9790, Taizhou, China).

Material Systhesis and Characterization
The investigation into the morphology and microstructure of the sample was conducted using a combination of advanced microscopy techniques, including scanning electron microscopy (SEM) and transmission electron microscopy (TEM).The results, showcased in Figure 1a,b, provide detailed insights into the particle-like morphology of magnetic ferro-ferric oxide (Fe 3 O 4 ).In addition, we also utilized particle size analysis software Im-ageJ V1.8.0 for a comprehensive evaluation of the sample's particle size distribution, which showed that the particle size distribution range was 50~400 µm, and the average particle size distribution range was about 206 ± 9 nm (Figure S1).The particle-like morphology of Fe 3 O 4 plays a crucial role in facilitating good diffusivity within the sonocatalytic system.This morphology promotes the efficient diffusion of reactant molecules to the active sites on the catalyst surface.By enhancing the accessibility of reactants to these active sites, the Fe 3 O 4 particles effectively contribute to accelerating the catalytic reaction rates.Further analysis through energy-dispersive spectrometer (EDS) element mapping images, as depicted in Figure 1c-f, not only confirmed the presence of Fe and O within the magnetic Fe 3 O 4 structure but also unveiled the uniform distribution of these elements throughout the material.The overall EDS distribution chart, representing the elemental composition of magnetic Fe 3 O 4 , can be observed in Figure S2, further clearly indicating the presence of two elements: Fe and O. Notably, a large peak located at about 1.7 keV was observed, which is an intrinsic attribute of the silicon substrate in the measurement of SEM-EDS mappings.The total spectral distribution diagram, as illustrated in Table 2, presents the total spectrum of the synthesized material.An evident observation from the data is the striking similarity between the mass fraction ratios of Fe and O obtained through the EDS test and the theoretical mass fraction values of Fe and O.This close alignment between the experimental and theoretical values provides compelling evidence affirming the successful synthesis of the material.The X-ray diffraction (XRD) analysis conducted at room temperature provided crucial insights into the crystal structure and phase composition of the catalytic material.Figure 2a displays the XRD patterns of magnetic Fe3O4, where the diffraction peaks at 2θ = 30.46°,35.35°, 43.68°, 54.31°, 57.13°, and 63.75° corresponded to the (220), (311), (400), (422), (511), and (440) crystal planes of Fe3O4, consistent with JCPDS 65-3107.Furthermore, no additional peaks were detected, confirming the absence of crystalline phases of FeO and Fe2O3 and indicating the high purity of the sample.To investigate the chemical composition and atomic chemical environment of the catalyst, X-ray photoelectron spectroscopy (XPS) analysis was performed.The binding energy of all elements was corrected with reference to C 1s.The XPS survey spectra revealed the presence of O, Fe, and C elements in magnetic Fe3O4, with element C attributed to instrument contamination (Figure 2b).Furthermore, high-resolution XPS spectra of Fe 2p are presented in Figure 2c, where the peak at 710.8 eV corresponded to Fe 2p3/2, while the peak at 724.2 eV corresponded to Fe 2p1/2.This indicated the existence of Fe elements in Fe3O4, consistent with previous findings in the literature [22].The high-resolution XPS spectrum of O 1s also verified the existence of O (Figure 2d).The X-ray diffraction (XRD) analysis conducted at room temperature provided crucial insights into the crystal structure and phase composition of the catalytic material.Figure 2a displays the XRD patterns of magnetic Fe 3 O 4 , where the diffraction peaks at 2θ = 30.46, and 63.75 • corresponded to the (220), (311), (400), (422), (511), and (440) crystal planes of Fe 3 O 4 , consistent with JCPDS 65-3107.Furthermore, no additional peaks were detected, confirming the absence of crystalline phases of FeO and Fe 2 O 3 and indicating the high purity of the sample.To investigate the chemical composition and atomic chemical environment of the catalyst, X-ray photoelectron spectroscopy (XPS) analysis was performed.The binding energy of all elements was corrected with reference to C 1s.The XPS survey spectra revealed the presence of O, Fe, and C elements in magnetic Fe 3 O 4 , with element C attributed to instrument contamination (Figure 2b).Furthermore, high-resolution XPS spectra of Fe 2p are presented in Figure 2c, where the peak at 710.8 eV corresponded to Fe 2p 3/2 , while the peak at 724.2 eV corresponded to Fe 2p 1/2 .This indicated the existence of Fe elements in Fe 3 O 4 , consistent with previous findings in the literature [22].The high-resolution XPS spectrum of O 1s also verified the existence of O (Figure 2d).Moreover, Figure 3a presents the Fourier Transform Infrared Spectroscopy (FT-IR) spectra of magnetic Fe3O4 nanoparticles.The FT-IR spectra revealed a prominent stretching absorption peak at 567 cm −1 , which corresponded to the Fe-O-Fe group of magnetic Fe3O4.Additionally, peaks around 1635 and 3420 cm −1 correspond to hydroxyl groups associated with water, aligning with previously reported FT-IR data for magnetic Fe3O4 [23].The Raman spectrum of magnetic Fe3O4 is shown in Figure 3b, and its typical peak position is consistent with the previously reported literature [24].In addition, thermogravimetric (TG) and differential thermal gravimetry (DTG) analysis of these nanoparticles (Figure 3c) showed negligible mass loss throughout the heating process.Additionally, Vibrating Sample Magnetometer (VSM) measurements were used for characterizing the magnetic properties of materials.VSM measurements under a ±15 kOe magnetic field (Figure 3d) revealed a saturation magnetization (Ms) of approximately 78.80 emu/g.The hysteresis loop of Fe3O4, passing through the origin without hysteresis, residual magnetization, or coercivity, confirmed the superparamagnetic nature of the material, facilitating easy separation from solutions via an external magnetic field.N2 adsorption/desorption measurements for Brunauer-Emmett-Teller (BET) surface area analysis (Figure 3e) indicated a specific surface area (SBET) of 9.60 m 2 /g for magnetic Fe3O4.This relatively low SBET suggested that the sonocatalytic hydrogen production primarily occurs on the surface of the magnetic Fe3O4, rather than within the material's internal structure.Through further analysis of the data, it can be seen that the pore size distribution is about 2.4 nm, which belongs to the mesoporous range, and the pore volume is about 0.03 cm 3 /g (Figure 3f).Moreover, Figure 3a presents the Fourier Transform Infrared Spectroscopy (FT-IR) spectra of magnetic Fe 3 O 4 nanoparticles.The FT-IR spectra revealed a prominent stretching absorption peak at 567 cm −1 , which corresponded to the Fe-O-Fe group of magnetic Fe 3 O 4 .Additionally, peaks around 1635 and 3420 cm −1 correspond to hydroxyl groups associated with water, aligning with previously reported FT-IR data for magnetic Fe 3 O 4 [23].The Raman spectrum of magnetic Fe 3 O 4 is shown in Figure 3b, and its typical peak position is consistent with the previously reported literature [24].In addition, thermogravimetric (TG) and differential thermal gravimetry (DTG) analysis of these nanoparticles (Figure 3c) showed negligible mass loss throughout the heating process.Additionally, Vibrating Sample Magnetometer (VSM) measurements were used for characterizing the magnetic properties of materials.VSM measurements under a ±15 kOe magnetic field (Figure 3d) revealed a saturation magnetization (Ms) of approximately 78.80 emu/g.The hysteresis loop of Fe 3 O 4 , passing through the origin without hysteresis, residual magnetization, or coercivity, confirmed the superparamagnetic nature of the material, facilitating easy separation from solutions via an external magnetic field.N 2 adsorption/desorption measurements for Brunauer-Emmett-Teller (BET) surface area analysis (Figure 3e) indicated a specific surface area (S BET ) of 9.60 m 2 /g for magnetic Fe 3 O 4 .This relatively low S BET suggested that the sonocatalytic hydrogen production primarily occurs on the surface of the magnetic Fe 3 O 4 , rather than within the material's internal structure.Through further analysis of the data, it can be seen that the pore size distribution is about 2.4 nm, which belongs to the mesoporous range, and the pore volume is about 0.03 cm 3 /g (Figure 3f).

Sonocatalytic H2 Production
The sonocatalytic H2 production performance of magnetic Fe3O4 was assessed under ultrasonic vibration, without any co-catalyst deposition, using various sacrificial agents.Notably, compared with other sacrificial agents, when using methanol (MeOH) as a sacrificial agent, the magnetic Fe3O4 showcased superior H2 production performance at 45.1 µmol/g/h (Figure 4a,b), which may be because MeOH can capture the holes most effectively.Prior research indicated that power is a critical factor in sonocatalysis.Consequently, the impact of varying ultrasonic powers on H2 production performance by magnetic Fe3O4 was investigated (Figure 4c,d).The results showed an increased trend in sonocatalytic performance with increasing power.This can be attributed to the fact that higher vibrational energy leads to increased stress, which in turn generates more sonoelectric charges on the catalyst's surface, consequently enhancing the hydrogen production rate [25].In addition, the impact of varying ultrasonic frequences on H2 production performance by magnetic Fe3O4 was investigated (Figure S3).The results showed an increased trend in sonocatalytic performance with increasing frequences.This improvement could be ascribed to the enhanced mass transfer phenomenon achieved by increasing the ultrasonic frequency [26].To eliminate the influence of inherent hydrogen generation capabilities of methanol/H2O, control experiments without any variables were carried out (Figure S4).These experiments revealed that, in the absence of a sonocatalyst, a minimal hydrogen production rate of 0.44 µmol/h was observed in a methanol/water mixture subjected to ultrasonic vibration.Nonetheless, the introduction of Fe3O4 into this methanol/water system resulted in an increase in hydrogen production, elevating the rate from 0.44 µmol/h to 2.25 µmol/h.

Sonocatalytic H 2 Production
The sonocatalytic H 2 production performance of magnetic Fe 3 O 4 was assessed under ultrasonic vibration, without any co-catalyst deposition, using various sacrificial agents.Notably, compared with other sacrificial agents, when using methanol (MeOH) as a sacrificial agent, the magnetic Fe 3 O 4 showcased superior H 2 production performance at 45.1 µmol/g/h (Figure 4a,b), which may be because MeOH can capture the holes most effectively.Prior research indicated that power is a critical factor in sonocatalysis.Consequently, the impact of varying ultrasonic powers on H 2 production performance by magnetic Fe 3 O 4 was investigated (Figure 4c,d).The results showed an increased trend in sonocatalytic performance with increasing power.This can be attributed to the fact that higher vibrational energy leads to increased stress, which in turn generates more sonoelectric charges on the catalyst's surface, consequently enhancing the hydrogen production rate [25].In addition, the impact of varying ultrasonic frequences on H 2 production performance by magnetic Fe 3 O 4 was investigated (Figure S3).The results showed an increased trend in sonocatalytic performance with increasing frequences.This improvement could be ascribed to the enhanced mass transfer phenomenon achieved by increasing the ultrasonic frequency [26].To eliminate the influence of inherent hydrogen generation capabilities of methanol/H 2 O, control experiments without any variables were carried out (Figure S4).These experiments revealed that, in the absence of a sonocatalyst, a minimal hydrogen production rate of 0.44 µmol/h was observed in a methanol/water mixture subjected to ultrasonic vibration.Nonetheless, the introduction of Fe 3 O 4 into this methanol/water system resulted in an increase in hydrogen production, elevating the rate from 0.44 µmol/h to 2.25 µmol/h.In addition, the cyclic stability of magnetic Fe3O4 was also examined (Figure 5a,b), revealing a minor decrease in H2 production performance over three successive cycles, likely due to the gradual loss of the Fe3O4 catalyst.The reusability and stability of magnetic Fe3O4 in sonocatalytic hydrogen generation were further confirmed by a series of characterization analysis before and after the sonocatalytic reaction.Figure 5c shows the PXRD pattern of the sample before and after undergoing a sonocatalytic reaction.While the range of diffraction angles remained largely unchanged, there was a noticeable reduction in peak intensity.This showed that the crystal form of magnetic Fe3O4 had not changed after the sonocatalytic reaction, but the crystal structure had become imperfect.In addition, Figure 4d shows the FT-IR spectrum of the sample before and after the sonocatalytic reaction.The peak position has basically not changed, but the intensity has decreased, which also verified the high stability of the catalytic material.All these characteristics proved that the sonocatalytic materials have excellent cyclic stability.In fact, Fe3O4 demonstrates versatility as it can serve not only as a sonocatalyst but also as a thermal and triboelectric catalyst.To ascertain that the hydrogen production attributed to ultrasound is a result of the material's sonoelectric effect rather than frictional or thermal effects, a comprehensive analysis was conducted along with a controlled trial.While Fe3O4 could theoretically serve as a thermal catalyst, our experiments were carefully designed with stringent temperature control.A cooling water system was utilized to maintain constant temperatures, minimizing any thermal effects.Based on this, the hydrogen production In addition, the cyclic stability of magnetic Fe 3 O 4 was also examined (Figure 5a,b), revealing a minor decrease in H 2 production performance over three successive cycles, likely due to the gradual loss of the Fe 3 O 4 catalyst.The reusability and stability of magnetic Fe 3 O 4 in sonocatalytic hydrogen generation were further confirmed by a series of characterization analysis before and after the sonocatalytic reaction.Figure 5c shows the PXRD pattern of the sample before and after undergoing a sonocatalytic reaction.While the range of diffraction angles remained largely unchanged, there was a noticeable reduction in peak intensity.This showed that the crystal form of magnetic Fe 3 O 4 had not changed after the sonocatalytic reaction, but the crystal structure had become imperfect.In addition, Figure 4d shows the FT-IR spectrum of the sample before and after the sonocatalytic reaction.The peak position has basically not changed, but the intensity has decreased, which also verified the high stability of the catalytic material.All these characteristics proved that the sonocatalytic materials have excellent cyclic stability.In fact, Fe 3 O 4 demonstrates versatility as it can serve not only as a sonocatalyst but also as a thermal and triboelectric catalyst.To ascertain that the hydrogen production attributed to ultrasound is a result of the material's sonoelectric effect rather than frictional or thermal effects, a comprehensive analysis was conducted along with a controlled trial.While Fe 3 O 4 could theoretically serve as a thermal catalyst, our experiments were carefully designed with stringent temperature control.A cooling water system was utilized to maintain constant temperatures, minimizing any thermal effects.Based on this, the hydrogen production observed in our study is not a result of thermal influences.Moreover, to address the possibility of friction-induced hydrogen production, a specific experiment was conducted by focusing on triboelectric catalysis.This involved examining hydrogen production of Fe 3 O 4 in a methanol aqueous solution under room temperature stirring conditions, with the findings presented in Figure S5.The observed hydrogen production was negligible, reinforcing our stance that the hydrogen generation in our experiments is not attributable to triboelectric effects.In addition, when evaluating semiconductor catalytic technologies, considering economic factors is crucial.Currently, existing semiconductor catalytic technologies, while claiming to achieve efficient water splitting for hydrogen production, often entail high catalyst synthesis costs.Although these reported hydrogen-producing catalysts appear to perform exceptionally well based on simple hydrogen production performance data, the actual production costs in practical applications often far exceed the economic benefits they bring.This is an issue that deserves attention.Therefore, assessing the hydrogen production performance of semiconductor catalysts in terms of hydrogen yield per unit cost (µmol/h/USD) is more reasonable compared to the commonly used hydrogen yield per gram per hour (µmol/g/h).This approach better reflects the economic feasibility of semiconductor catalysts as it directly relates their performance to production costs, aiding in determining the competitiveness of specific semiconductor catalysts in real-world applications.As a result, conducting a comprehensive cost-benefit analysis regarding the application potential of magnetic Fe 3 O 4 becomes imperative.This study showcases a remarkable H 2 production rate of 175 µmol/h/USD, surpassing the performance of numerous previously reported mechanical catalytic materials, as delineated in Table 3 [27][28][29][30][31][32].This finding underscores the promising prospects of magnetic Fe 3 O 4 as a highly efficient and cost-effective sonocatalyst for hydrogen production.
Materials 2024, 17, x FOR PEER REVIEW 9 of 14 observed in our study is not a result of thermal influences.Moreover, to address the possibility of friction-induced hydrogen production, a specific experiment was conducted by focusing on triboelectric catalysis.This involved examining hydrogen production of Fe3O4 in a methanol aqueous solution under room temperature stirring conditions, with the findings presented in Figure S5.The observed hydrogen production was negligible, reinforcing our stance that the hydrogen generation in our experiments is not attributable to triboelectric effects.In addition, when evaluating semiconductor catalytic technologies, considering economic factors is crucial.Currently, existing semiconductor catalytic technologies, while claiming to achieve efficient water splitting for hydrogen production, often entail high catalyst synthesis costs.Although these reported hydrogen-producing catalysts appear to perform exceptionally well based on simple hydrogen production performance data, the actual production costs in practical applications often far exceed the economic benefits they bring.This is an issue that deserves attention.Therefore, assessing the hydrogen production performance of semiconductor catalysts in terms of hydrogen yield per unit cost (µmol/h/USD) is more reasonable compared to the commonly used hydrogen yield per gram per hour (µmol/g/h).This approach better reflects the economic feasibility of semiconductor catalysts as it directly relates their performance to production costs, aiding in determining the competitiveness of specific semiconductor catalysts in real-world applications.As a result, conducting a comprehensive cost-benefit analysis regarding the application potential of magnetic Fe3O4 becomes imperative.This study showcases a remarkable H2 production rate of 175 µmol/h/USD, surpassing the performance of numerous previously reported mechanical catalytic materials, as delineated in Table 3 [27][28][29][30][31][32].This finding underscores the promising prospects of magnetic Fe3O4 as a highly efficient and cost-effective sonocatalyst for hydrogen production.

Sonocatalytic Mechanism Investigation
The separation and transfer efficiency of sonocarriers was studied through a series of electrochemical measurements [33].As shown in Figure 6a, the impact of ultrasonic vibration on the system was vividly depicted, showcasing a remarkable increase in the average current density, reaching approximately 920 nA post introduction of ultrasonic vibration.This surge in current density signifies the effectiveness of ultrasound in facilitating the separation of sonocarriers within the Fe 3 O 4 medium, consequently driving the sonocatalytic H 2 production reaction.Moreover, upon cessation of the ultrasound, the sonocurrent swiftly plummeted to 0, further underscoring the direct correlation between ultrasoundinduced vibrations and the dynamic behavior of sonocarriers.Additionally, the arc radius of the electrochemical impedance spectrum (EIS) curve of magnetic Fe 3 O 4 under ultrasonic vibration was slightly smaller than that of magnetic Fe 3 O 4 without ultrasonic vibration (Figure 6b).These results both indicated that mechanical energy introduction into the catalyst system effectively reduced charge transfer impedance and enhanced the separation and transfer efficiency of charges.

Sonocatalytic Mechanism Investigation
The separation and transfer efficiency of sonocarriers was studied through a series of electrochemical measurements [33].As shown in Figure 6a, the impact of ultrasonic vibration on the system was vividly depicted, showcasing a remarkable increase in the average current density, reaching approximately 920 nA post introduction of ultrasonic vibration.This surge in current density signifies the effectiveness of ultrasound in facilitating the separation of sonocarriers within the Fe3O4 medium, consequently driving the sonocatalytic H2 production reaction.Moreover, upon cessation of the ultrasound, the sonocurrent swiftly plummeted to 0, further underscoring the direct correlation between ultrasoundinduced vibrations and the dynamic behavior of sonocarriers.Additionally, the arc radius of the electrochemical impedance spectrum (EIS) curve of magnetic Fe3O4 under ultrasonic vibration was slightly smaller than that of magnetic Fe3O4 without ultrasonic vibration (Figure 6b).These results both indicated that mechanical energy introduction into the catalyst system effectively reduced charge transfer impedance and enhanced the separation and transfer efficiency of charges.The mechanism of H2 production from magnetic Fe3O4 by sonocatalysis, as depicted in Figure 7, elucidates the intricate interplay of various physical and chemical processes underlying this phenomenon.This process begins with the sonocatalytic water splitting, facilitated by the Fe3O4 catalyst, where H2O acts as the reactant and MeOH is used as the sacrificial agent.Under the influence of ultrasonic vibrations, a pivotal driving force attributed to the phenomenon of ultrasonic cavitation, the Fe3O4 catalyst is activated.Ultrasonic cavitation involves the rapid oscillation, expansion, and subsequent collapse of small bubbles in a liquid medium when exposed to ultrasonic waves, leading to the generation of localized regions of high temperature and pressure [18, [34][35][36][37][38][39][40][41].This environment, accentuated by the phenomenon of sonoluminescence-the emission of light and energy from the collapsing bubbles-energizes electrons to the conduction band of Fe3O4, leaving behind holes in the valence band.This electron excitation primes the catalyst for The mechanism of H 2 production from magnetic Fe 3 O 4 by sonocatalysis, as depicted in Figure 7, elucidates the intricate interplay of various physical and chemical processes underlying this phenomenon.This process begins with the sonocatalytic water splitting, facilitated by the Fe 3 O 4 catalyst, where H 2 O acts as the reactant and MeOH is used as the sacrificial agent.Under the influence of ultrasonic vibrations, a pivotal driving force attributed to the phenomenon of ultrasonic cavitation, the Fe 3 O 4 catalyst is activated.Ultrasonic cavitation involves the rapid oscillation, expansion, and subsequent collapse of small bubbles in a liquid medium when exposed to ultrasonic waves, leading to the generation of localized regions of high temperature and pressure [18, [34][35][36][37][38][39][40][41].This environment, accentuated by the phenomenon of sonoluminescence-the emission of light and energy from the collapsing bubbles-energizes electrons to the conduction band of Fe 3 O 4 , leaving behind holes in the valence band.This electron excitation primes the catalyst for activity, enabling the reduction of hydrogen ions (H + ) to form H 2 , while oxygen atoms produced from water splitting typically react with sacrificial agents, producing oxidation products.The sacrificial agent MeOH is crucial in this process.It is oxidized during the reaction, providing more reactive electrons to the system by consuming the holes created during water splitting and preventing their recombination with electrons, thereby ensuring continuous H 2 production.This oxidation process typically results in the formation of formaldehyde, formic acid, and other oxidation products [42][43][44][45][46][47][48][49].Overall, the intricate interplay of sonoluminescence-induced activation, electron excitation, and sacrificial agent utilization collectively contribute to the efficient sonocatalytic production of hydrogen, highlighting the pivotal role of ultrasonic vibrations in driving catalytic reactions and unlocking the potential of Fe 3 O 4 -based materials for sustainable energy applications.
Materials 2024, 17, x FOR PEER REVIEW 11 of 14 activity, enabling the reduction of hydrogen ions (H + ) to form H2, while oxygen atoms produced from water splitting typically react with sacrificial agents, producing oxidation products.The sacrificial agent MeOH is crucial in this process.It is oxidized during the reaction, providing more reactive electrons to the system by consuming the holes created during water splitting and preventing their recombination with electrons, thereby ensuring continuous H2 production.This oxidation process typically results in the formation of formaldehyde, formic acid, and other oxidation products [42][43][44][45][46][47][48][49].Overall, the intricate interplay of sonoluminescence-induced activation, electron excitation, and sacrificial agent utilization collectively contribute to the efficient sonocatalytic production of hydrogen, highlighting the pivotal role of ultrasonic vibrations in driving catalytic reactions and unlocking the potential of Fe3O4-based materials for sustainable energy applications.

Conclusions
In summary, our study has demonstrated the exceptional potential of magnetically responsive Fe3O4 nanoparticles in facilitating H2 production via sonocatalysis, achieving an impressive rate of 175 µmol/h/USD.This accomplishment highlights the efficacy of harnessing vibrational energy, a key aspect of sonocatalysis, to drive catalytic reactions efficiently.Moreover, the unique property of Fe3O4 nanoparticles to be rapidly and efficiently separated and recovered using an external magnetic field adds another layer of practicality to their utilization in real-world applications.This capability not only simplifies the catalyst recovery process but also minimizes material loss and enhances the overall sustainability of the process.Furthermore, the robust durability, economic viability, widespread accessibility, environmental compatibility, and remarkable stability exhibited by Fe3O4-based materials underscore their significance as highly promising candidates for future applications in energy conversion processes.These materials hold immense potential in converting mechanical energy into chemical energy, particularly in the context of sustainable energy technologies.

Supplementary Materials:
The following supporting information can be downloaded at www.mdpi.com/xxx/s1.

Conclusions
In summary, our study has demonstrated the exceptional potential of magnetically responsive Fe 3 O 4 nanoparticles in facilitating H 2 production via sonocatalysis, achieving an impressive rate of 175 µmol/h/USD.This accomplishment highlights the efficacy of harnessing vibrational energy, a key aspect of sonocatalysis, to drive catalytic reactions efficiently.Moreover, the unique property of Fe 3 O 4 nanoparticles to be rapidly and efficiently separated and recovered using an external magnetic field adds another layer of practicality to their utilization in real-world applications.This capability not only simplifies the catalyst recovery process but also minimizes material loss and enhances the overall sustainability of the process.Furthermore, the robust durability, economic viability, widespread accessibility, environmental compatibility, and remarkable stability exhibited by Fe 3 O 4 -based materials underscore their significance as highly promising candidates for future applications in energy conversion processes.These materials hold immense potential in converting mechanical energy into chemical energy, particularly in the context of sustainable energy technologies.
through the EDS test and the theoretical mass fraction values of Fe and O.This close alignment between the experimental and theoretical values provides compelling evidence affirming the successful synthesis of the material.

Figure 5 .
Figure 5. (a) Time profiles of sonocatalytic H2 production by Fe3O4 under 3 cycles (reaction conditions are as follows: ultrasonic frequences = 50 kHz, ultrasonic powers = 240 W, and sacrificial agent = MeOH); (b) H2 production rates of Fe3O4 after 3 cycles (reaction conditions are as follows: ultrasonic frequences = 50 kHz, ultrasonic powers = 240 W, and sacrificial agent = MeOH); (c) the XRD spectra over magnetic Fe3O4 before and after sonocatalytic water splitting reactions; (d) the FT-IR spectra over magnetic Fe3O4 before and after sonocatalytic water splitting reactions.

Figure 5 .
Figure 5. (a) Time profiles of sonocatalytic H 2 production by Fe 3 O 4 under 3 cycles (reaction conditions are as follows: ultrasonic frequences = 50 kHz, ultrasonic powers = 240 W, and sacrificial agent = MeOH); (b) H 2 production rates of Fe 3 O 4 after 3 cycles (reaction conditions are as follows: ultrasonic frequences = 50 kHz, ultrasonic powers = 240 W, and sacrificial agent = MeOH); (c) the XRD spectra over magnetic Fe 3 O 4 before and after sonocatalytic water splitting reactions; (d) the FT-IR spectra over magnetic Fe 3 O 4 before and after sonocatalytic water splitting reactions.

Figure 6 .
Figure 6.(a) Current obtained from magnetic Fe3O4 under ultrasonic or without ultrasonic vibration; (b) EIS plots obtained from magnetic Fe3O4 under ultrasonic or without ultrasonic vibration.

Figure 6 .
Figure 6.(a) Current obtained from magnetic Fe 3 O 4 under ultrasonic or without ultrasonic vibration; (b) EIS plots obtained from magnetic Fe 3 O 4 under ultrasonic or without ultrasonic vibration.
Figure S1: The size distribution bar chart of Fe3O4; Figure S2: The EDS distribution chart of the total number spectrum of Fe3O4; Figure S3: The sonocatalytic H2 production rates of catalyst under different frequencies; Figure S4: The sonocatalytic H2 production blank

Figure 7 .
Figure 7.The illustration of sonocatalytic H 2 production process over Fe 3 O 4 .
Figure S1: The size distribution bar chart of Fe 3 O 4 ; Figure S2: The EDS distribution chart of the total number spectrum of Fe 3 O 4 ; Figure S3: The sonocatalytic H 2 production rates of catalyst under different frequencies; Figure S4: The sonocat-

Table 2 .
Total spectral distribution diagram.

Table 2 .
Total spectral distribution diagram.

Table 3 .
The H 2 production rate for the preparation cost of different mechanical catalytic materials.

Table 3 .
The H2 production rate for the preparation cost of different mechanical catalytic materials.