Mechanochemical Synthesis of Water-Based Magnetite Magnetic Fluids Mechanochemical Synthesis of Water-Based Magnetite Magnetic Fluids

Magnetite, Fe 3 O 4 (Fe II O-Fe III2 O 3 ), is a member of the spinel group as well as a common ferrite with a cubic inverse spinel structure. Aqueous colloidal solutions of Fe 3 O 4 , i.e. water-based Fe 3 O 4 magnetic fluids, have attracted substantial attention in biomedical applications, such as drug delivery, magnetic resonance imaging, and magnetic hyperthermia. In this study, to readily prepare water-based magnetic fluids with biocompatible dispersant-coated Fe 3 O 4 nanoparticles that are stably dispersed in water medium, a mechanochemical synthesis method was developed. In this method, an iron-free citric acid solution is milled in a tumbling ball mill with steel balls at room temperature, reducing the production costs and environmental impacts. The initial gas phase in the milling vessel is air, and pressure is varied to control the formation of Fe 3 O 4 nanoparticles. Although no iron species are contained in the starting solution, Fe 3 O 4 nanoparticles form in the solution according to the reaction mechanism based on the oxidation-reduction processes of the corrosion of steel. At the same time, the Fe 3 O 4 nanoparticle surface is modified with citrate ions, resulting in a stable dispersion. The magnetic fluids prepared using this mechanochemical method possess good induction heating properties in an alternating current magnetic field.


Introduction
Magnetic fluids are colloidal solutions containing magnetic nanoparticles stably dispersed in a liquid medium so that the entire fluid behaves like a ferromagnet. Moreover, no solid-liquid separation occurs, even under centrifugal force fields [1]. Therefore, magnetic fluids are widely used in industrial products, such as rotating-axis seal material, lubricants, and liquid damper. In magnetic fluids, iron-based oxide nanoparticles are often employed as the magnetic material. In particular, magnetic fluids that consist of superparamagnetic magnetite (Fe 3 O 4 ) nanoparticles are used in biomedical and environmental fields, such as magnetic resonance imaging (MRI) contrast agents [2], magnetic hyperthermia in cancer therapy [3], drug delivery systems (DDS) [4], and so on, due to their high performance, low toxicity, and low environmental impact. The appropriate size of the Fe 3 O 4 nanoparticles depends on their application. For example, when used for MRI, hyperthermia, and DDS, the particle size is required to be below 30 nm. Appropriate surface modification of the Fe 3 O 4 nanoparticles can bind them to biomolecules as a result of the large surface area, leading to good bio-circulation [5]. In addition, the enhanced permeability and retention (EPR) effect can selectively accumulate Fe 3 O 4 nanoparticles into cancer tissue [6]. Accordingly, the sizes of primary particles and aggregates must be controlled, and good stable dispersion is needed. Therefore, many studies on controlling the particle size as well as dispersion and aggregation using additives and organic solvents have been conducted. Wen et al. [7] prepared Fe 3 O 4 nanoparticles with a size of 4-5 nm in an organic solvent by adding sodium oleate as a surfactant. Zhang et al. [8] synthesized Fe 3 O 4 nanoparticles with good dispersibility in an organic solvent using surface modification with polystyrene. Frimpong et al. [9] controlled the sizes of the primary particle and aggregate using citric acid. Elisa de Sousa et al. [10] obtained citrate-adsorbed Fe 3 O 4 nanoparticles that were dispersed under neutral conditions. In Fe 3 O 4 nanoparticles that are used in vivo, biocompatible molecules, such as dextran, polyethylene glycol, polyvinyl alcohol, citric acid, polyacrylic acid, and phospholipid, are frequently employed as additives. In particular, citric acid is non-toxic and can form ultrafine primary particles. Furthermore, citrate ions can adsorb onto Fe 3 O 4 nanoparticles, leading to aggregation inhibition due tothe steric hindrance and electrostatic repulsion. Therefore, citric acid is widely used as an anti-aggregation agent.
Magnetic fluids were originally developed by Papell of the United States National Aeronautics and Space Administration (NASA) in 1965 for position control of liquid fuel under zero gravity [11]. The initially developed magnetic fluid was prepared by ball-mill grinding of Fe 3 O 4 grains in kerosene containing oleic acid for several hundred hours, followed by the removal of coarse particles by centrifugal separation. At present, various methods for synthesizing Fe 3 O 4 nanoparticles using chemical reactions in gas, liquid, and solid phases have been developed. Among them, liquid phase synthesis has been actively studied because the component and concentration of the reactants can be controlled fairly easily and the formation reaction can progress, even under moderate conditions. In particular, coprecipitation methods that produce Fe 3 O 4 by adding a base as a precipitant to a solution containing Fe 2+ ions and Fe 3+ ions are industrially employed because they can easily provide homogeneous Fe 3 O 4 nanoparticles with smaller primary sizes [12]. The thermal decomposition method is also frequently used. Jeyadevan et al. [13] synthesized Fe 3 O 4 nanoparticles that are suitable for magnetic hyperthermia via thermal decomposition of iron pentacarbonyl in oleic acid-containing dioctyl ether. Sun and Zeng [14] prepared mono-dispersed superparamagneticFe 3 O 4 nanoparticles using iron acetylacetonate, oleic acid, and amine oleate. As a result, conventional methods can provide magnetic fluids that are desirable for various applications. For the previously mentioned biomedical applications, water-based Fe 3 O 4 magnetic fluids are suitable. However, in many cases, the conventional methods require iron salt, base, and organic solvents, requiring the removal of unnecessary components from the product to clean water-based magnetic fluids. This may increase the environmental impact and production cost. As a result, an innovative process is required to more readily prepare water-based Fe 3 O 4 magnetic fluids without any additional operations.
To meet the demand, we developed a new mechanochemical process with an iron-free aqueous solution containing citric acid (CA) as a reaction accelerator and anti-aggregation agent milled at room temperature with a ball mill using steel balls, resulting in the production of waterbased Fe 3 O 4 magnetic fluid [15]. The formation of crystalline Fe 3 O 4 nanoparticles in this process may consist of several steps, as follows: (1) corrosion of steel balls, (2) oxidation of released ferrous ions, (3) reduction of ferric ions, and (4) formation and crystal growth of Fe 3 O 4 . Therefore, the mechanochemical effect can enhance the formation and crystallization of Fe 3 O 4 . Furthermore, this method does not use iron salts as iron sources or a base as the precipitant because iron ions and hydroxide ions form during the corrosion, which can provide a waterbased Fe 3 O 4 magnetic fluid without any post-operations, such as the removal of unnecessary ions and solvent displacement.
This paper presents the properties of Fe 3 O 4 magnetic fluids prepared by this mechanochemical process and a detailed analysis of the reaction mechanism based on changes in the composition of gas, liquid, and solid phases in the formation of Fe 3 O 4 as well as ferrous, ferric, and hydroxide ions. Furthermore, the formation reaction of Fe 3 O 4 is kinetically analysed under different gas phase conditions and mechanical energy fields.

Experimental
All chemicals used in this work were purchased from Wako Pure Chemical Industries and were used without further purification. Some preliminary experiments determined that the CA concentration in the starting solution was desirable to be 5 mmol/L, and a single Fe 3 O 4 phase was finally obtained. Ninety millilitres of the CA solution (pH = 2.7) was placed in a tumbling ball mill consisting of a Teflon-lined gas-tight vessel (capacity 500 mL, diameter 90 mm) and carbon steel balls (Fe > 99 mass%, diameter 3 mm). The charged volume of the balls (includes the voids among balls) was 40% of the vessel capacity. The initial gas phase in the vessel was air at atmospheric pressure. The solution was milled at room temperature for anappropriate time period. The rotational speed of the vessel was 140 rpm, corresponding to the theoretically determined critical rotational speed. After milling, the fluid was removed from the vessel and characterized. The weight loss of balls during milling was also measured.
The particle size distribution and zeta potential of samples were determined using a particle size analyser (Malvern Zetasizer Nano ZS), and the phase evolution was evaluated using a powder X-ray diffractometer (Rigaku RINT-1500) after drying the sample. The average crystallite size was determined using Scherrer's equation for the diffraction peak from the Fe 3 O 4 (311) plane at 2θ = 35.4°. The FT-IR spectrum was measured using a Fourier transform infrared spectrophotometer (Shimadzu IRAffinity-1). The magnetic properties (i.e. magnetization-magnetic field hysteretic cycle) were analysed with a superconducting quantum interference device (SQUID) magnetometer (Quantum Design MPMS XL) at room temperature over a magnetic field range of −10 kOe to +10 kOe. The magnetic hyperthermia-related induction heating properties of the fluid were evaluated by an alternating magnetic field generator consisting of a radio frequency power source (Thamway T162-5723A), an impedance matching box (Thamway T020-5723F), and a solenoid coil (inner diameter 70 mm) with 21 copper tube (outer diameter of 4 mm and inner diameter of 3 mm) turns [13]. Cooling water flowed inside the copper tube. A proper amount (approximately 1.0 g) of fluid was charged in a glass tube with a diameter of 16 mm, and the test tube was placed in the coil centre. The temperature increase in the fluid from 37°C (corresponding to normal body temperature) in an alternating magnetic field was measured with an optical fibre thermometer (FISO Technologies FTI-10 equipped with FOT-L-NS-967). The frequency and amplitude of the magnetic field were 600 kHz and 3 kA/m, respectively.
In the analysis of the reaction mechanism, 0.5 mL of the gas in the vessel was sampled with a syringe after milling, and the gas component was analysed by a gas chromatograph (Shimadzu GC-8A). The pH of the resultant fluid was measured with a pH meter (Horiba D-21 equipped with 9625-10D electrode), and the iron(II) and iron(III) concentrations were determined by a colorimetric method with 1,10-phenanthroline and a spectrophotometer (JASCO Ubest V-530).
The influence of gas phase conditions in the vessel on the Fe 3 O 4 formation was studied. The 5 mmol/L CA solution was placed in the vessel; then, compressed air was charged into the vessel before milling. The total pressure in the gas phase was varied from 1 atm (atmospheric pressure) to 6 atm, corresponding to initial oxygen partial pressures of 0.21-1.26 atm. The milling time was 24 h at total pressures lower than 2.5 atm but 48 h at higher than 3 atm. The pressure in the vessel before and after milling was measured, and the oxygen consumption was determined from the pressure change. Furthermore, the rotational speed of the vessel was varied from 0 to 140 rpm at a total pressure of 1 atm, which can alter the intensity of the mechanical energy field. Based on the obtained results, the kinetics of the Fe 3 O 4 formation reaction were investigated.
Lastly, scaleup of the process was examined. In this investigation, a Teflon-lined milling vessel with a capacity of 2.6 L (diameter 150 mm) was used. The charged volume of the steel balls (diameter 3 mm) was 40% of the vessel capacity, and 476 mL of the 5 mmol/L CA solution was placed in the vessel and milled for 24 h. The gas phase was air at atmospheric pressure. The rotational speed of the large vessel was adjusted so that either the Froude number or the peripheral velocity in both vessels agreed with each other. Figures 1 and 2 show the X-ray diffraction (XRD) patterns and average crystallite sizes of solid products obtained after milling the CA solution, respectively. At milling times of less than 1.5 h, broad peaks that were attributed to ferrihydrite (Fe 5 O 8 H⋅4H 2 O), which is an amorphous or low crystalline oxyhydroxide [16][17][18], were observed in the XRD patterns. After 2 h, formation and crystal growth of Fe 3 O 4 occurred, and milling for more than 18 h provided relatively high crystalline Fe 3 O 4 . However, most of the Fe 3 O 4 particles settled in the fluid due to aggregation. To improve the dispersion, the total citrate concentration of fluids was increased by adding anhydrous CA and trisodium citrate dehydrate to the fluid obtained by milling for 18 h; and then the zeta potential was measured. The pH of the fluids was kept constant at approximately 8 in all fluids by adding proper amounts of CA and trisodium citrate. Figure 3 illustrates the change in the zeta potential with the changing citrate concentration. The absolute value of the zeta potential increased with the increasing citrate concentration. When the citrate concentration was 14 mmol/L, corresponding to an iron/citrate molar ratio of approximately 3 in the fluid, the zeta potential was less than −40 mV, leading to good dispersion. It was inferred that an increase in the citrate concentration can improve the dispersion. Using the fluid obtained at a milling time of 24 h, fluid with a total citrate concentration of 27 mmol/L was prepared and characterized. As shown in Figure 4, this fluid had good dispersion even after 2 weeks. Furthermore, when a permanent magnet was placed beside the glass bottle, the fluid was attracted to magnet, suggesting that the Fe 3 O 4 nanoparticles have superparamagnetic properties. Figure 5 shows the particle size distribution, magnetizationmagnetic field curve, and FT-IR spectrum of the solid product obtained by drying the fluid after magnetic separation. The median diameter was 7.3 nm, while the crystallite size was 8.8 nm. The primary particle size was almost the same as the crystallite size, suggesting that the obtained Fe 3 O 4 nanoparticles were monocrystalline. The saturation magnetization was 27 emu/g, which was much lower than that of bulk Fe 3 O 4 (92 emu/g) due to the smaller particle size. The residual magnetization was approximately zero, and the coercivity was very low, indicating superparamagnetism. Additionally, some strong absorption bands in the FT-IR spectrum were observed, which were attributed to citrate ions, indicating that the Fe 3 O 4 nanoparticle surface was modified by citrate ions. Figure 6 illustrates a temperature adjustment of the fluid within 43 ± 0.5°C, corresponding to a typical temperature range in hyperthermia treatments, in an on-off-controlled alternating magnetic field. This fluid was found to exhibit good magnetic hyperthermia properties. In addition, the temperature was successfully controlled within the temperature range, suggesting that the fluid can be used in hyperthermia therapies.

Formation of Fe 3 O 4 magnetic fluids
Eqs. (7) and (8) include O 2 as a reactant. Therefore, by analysing the dissolved O 2 concentration, the validity of the proposed reaction mechanism was confirmed. The dissolved O 2 concentration was estimated from the O 2 partial pressure in the gas phase using Henry's law: where x is the molar fraction of O 2 in a liquid phase, p is the O 2 partial pressure in a gas phase in equilibrium with the liquid phase, and H is the Henry constant (= 4.38 × 10 4 atm at 25°C [19]).
In this process, the dissolved O 2 concentration can always vary during milling. However, the liquid phase is well mixed with the gas phase by milling, resulting in a relatively large gasliquid interfacial area. Therefore, when the dissolved O 2 is consumed, O 2 can immediately be supplied from the gas phase. Accordingly, it can be assumed that the liquid and gas phases are always in equilibrium with each other. Figure 9 shows the gas phase composition during milling. The open circles in this figure indicate the estimated values that were calculated using Eqs. (7) and (8) from the iron concentrations shown in Figure 6. It was confirmed that the volume of nitrogen (N 2 ) gas was almost constant and that hydrogen (H 2 ) gas evolution hardly occurred. In contrast, the O 2 gas was completely consumed within 2 h. Furthermore, the experimental data of the O 2 gas volume mostly agreed with the calculated values. The results demonstrate the validity of the reaction mechanism according to Eqs. (1)-(8). As seen in Figure 8, after 2 h, O 2 was absent from the vessel and the pH was almost constant at 7.8. Figure 10 shows the change in the Fe 2+  Consequently, the overall reaction under low O 2 conditions can be described by Eq. (12).  Figure 11 illustrates the evolution of H 2 during milling. The formation of H 2 was noticed for long milling times. As seen in Figure 10, the iron concentrations slightly increased even after 6 h. Thus, the following reduction reaction may occur, resulting in H 2 evolution.   Figures 12-14 illustrate the XRD pattern of the obtained solid products, pH after milling, and internal pressure change of the vessel, respectively, under various initial total pressures in the vessel. Regardless of the internal total pressure, crystalline Fe 3 O 4 was obtained, and the pH after milling was approximately 9. As seen in Figure 14, the internal pressure change was in agreement with the theoretical values (shown by a broken line) that had been calculated by assuming that O 2 in the vessel was completely consumed during milling, suggesting that under an initial oxygen partial pressure of less than 1.26 atm, O 2 was used in the Fe 3 O 4 formation process.

Effect of the oxygen partial pressure
From Eqs.    Figure 15 shows the iron concentration of fluids estimated based on the internal pressure change, as shown in Figure 14 with Eq. (16). The iron concentration increased with increasing initial O 2 partial pressure, indicating that the iron concentration can be controlled by the initial O 2 partial pressure.

Kinetics of the reaction for the formation of Fe 3 O 4
Based on the concentrations of Fe 2+ and Fe 3+ ions during milling, the Fe 3 O 4 formation reaction was kinetically analysed. Figure 16 shows the change in the O 2 partial pressure and iron concentrations with the milling time in the initial stages at an initial total pressure of 1 atm. A monotonous decrease in the O 2 partial pressure with increased milling time was observed. Because the dissolved O 2 concentration is proportional to the O 2 partial pressure, the rate of reactions related to the dissolved O 2 concentration, as shown by Eqs. (7) and (8), can also vary depending on the milling time. However, at less than 1.5 h, O 2 was consumed at a constant rate, and the rates of Fe 2+ and Fe 3+ ion formation were almost constant regardless of the milling time, as seen in Figure 7b. The results imply that the reactions are independent of the dissolved O 2 concentration. Therefore, the reaction rates of Eqs. (7) and (8) can be described by a zero-order model. This model is effective because O 2 quickly dissolves into the solution due to vigorous gas-liquid mixing by milling and milling accelerates the corrosion of steel due to the improvement in the diffusion rate of O 2 to the steel surface. Accordingly, the Fe 3 O 4 formation process may be the oxidation-reduction reaction control, and both the dissolution of O 2 from the gas phase to the liquid phase and diffusion rate of O 2 in the liquid phase can be much faster than the rate of the oxidation-reduction reaction. Using the data shown in Figure 16, the rates of Fe 2+ and Fe 3+ ion formation were calculated. Based on Eqs. (7) and (8), the O 2 consumption rate was determined to be approximately 0.40 µmol/s, which nearly agreed with the experimental data (0.41 µmol/s) shown in Figure 16. This result suggests that the consumed O 2 was spent on the release and oxidation of Fe 2+ ions.
Eqs. (7) and (8) Assuming that the reaction rates of Eqs. (17) and (18), r 1 and r 2 , can be described by a zeroorder model and expressed as follows: Here k 1 and k 2 are the rate constants of Eqs. (17) and (18), respectively. As a result, the rate of Fe(OH) 2 formation reaction, r, can be expressed by Eq. (21).   Figure 18 illustrates the rate constant K calculated using Eqs. (19)-(21) as a function of the initial O 2 partial pressure. K was found to decrease with the increasing initial O 2 partial pressure. As shown in Figure 19, when the initial O 2 partial pressure was relatively high, goethite (α-FeOOH) and Fe 5 O 8 H⋅4H 2 O formed as intermediates. In general, the rate of steel corrosion is affected by the mass fraction of iron oxyhydroxides and iron oxides in a corrosion product on the steel. In particular, when a high level of α-FeOOH is contained in the corrosion product, the corrosion rate can decrease because α-FeOOH prevents O 2 from penetrating into the steel surface [20]. At high initial O 2 partial pressures, Fe 5 O 8 H⋅4H 2 O can make a phase transition to α-FeOOH, and a dense corrosion product layer may form on the steel surface. This inhibits the mass transfer of O 2 , resulting in a decrease in the rate of Fe(OH) 2 formation reaction. Next, the effect of the vessel rotational speed on the O 2 consumption was studied in atmospheric pressure. Figure 20 depicts the change in the O 2 partial pressure with the milling time.
The O 2 consumption rate was almost the same, regardless of the rotational speed. Figure 21 illustrates the relationship between the rotational speed and K calculated assuming that the O 2 consumption can be expressed by a zero-order equation. At rotational speeds higher than 10 rpm, K was almost constant, suggesting that the milling generated a sufficient level of energy required for the mass transfer and consumption of O 2 , even at low rotational speeds.

Scaleup of the process
Using a milling vessel with a capacity of 2.6 L, which was approximately 5 times as large as the small vessel used in the above investigations, the scaleup of the process was studied based on two cases, constant Froude number and constant peripheral velocity of the vessels. The rotational speed of 140 rpm for the small vessel corresponds to 108.4 rpm for the large vessel at a constant Froude number (= 0.0500) and 87.7 rpm at a constant peripheral velocity (= 0.735 m/s). Figures 22-24 show the XRD pattern of solid products, average crystallite size, and pH after milling, respectively. The results demonstrate that the large scale process can fabricate a similar fluid, suggesting that scale-up of the process can be successful based on either the Froude number or peripheral velocity of the vessel.

Conclusions
This study analysed in detail a new mechanochemical process for readily synthesizing waterbased magnetic Fe 3 O 4 fluids. Major conclusions are summarized as follows: a. The Fe 3 O 4 formation mechanism in this process has been clarified, which can be described by several oxidation-reduction reactions, such as the corrosion of steel, oxidation of Fe(OH) 2  c. Kinetic analysis of the process clarified the effects of the dissolved O 2 and the vessel rotational speed on the rate of the Fe(OH) 2 formation reaction. In the process, the reaction rate of O 2 consumption can be expressed by a zero-order equation.
d. Scaleup of the process can be successful by considering either the Froude number or peripheral velocity of the vessel.