Tetrabutyl titanate (TBT, >98%) was bought from Aladdin Biochemical Technology Co., Ltd. (Shanghai, China). Absolute ethanol (C₂H₅OH, AR), acetic acid (CH₃COOH, AR), oxalic acid dihydrate (H₂C₂O₄·2H₂O, AR), and calcium chloride (CaCl₂, AR) were obtained from Sinopharm Chemical Reagent Company (Shanghai, China). All reagents were of analytical-grade and had no purification. Deionized Millipore-Q water (18.2 MΩ cm) was used during the whole experimentation.

The TiO₂ nanoparticles were synthesized through the sol–gel method. TBT was used as the inorganic precursor, and ethanol and water were used as solvents. First, TBT and ethanol were mixed at a TBT/ethanol volume ratio of 1/2. Meanwhile, a small amount of acetic acid was added under the condition of intense agitation to prevent TBT from reacting with water in the air during the stirring process. Then the deionized water was dropped into the TBT solution with vigorous stirring. In the preparation, the volume ratio of H₂O, TBT, and ethanol is kept at 7:17:34. The ER fluid was stirred for 12 h and then aged for 10 h to deposit the particles. Then the white precipitate was filtrated and, using water and ethanol, it was washed three times. Finally, the precipitation was transferred to a vacuum drying oven, vacuum dried at 60°C for 12 h, and then vacuum dehydrated at 120°C for 2 h. The bulk density of the as-prepared TiO₂ particles was 2.0 g/cm³. The fabrication processing schematic diagram is presented in Figure 1.

The calcium–titanium–oxygen precipitate (Ca-Ti-O) nanoparticles were prepared by the co-precipitation method and no special additives were added to the particles (Lu et al., 2007). In brief, calcium chloride, TBT, and absolute ethyl alcohol were the initial solution, and oxalic acid solution was the precipitant. White precipitate came into being as soon as oxalic acid solution was added. After 8 h of aging at room temperature, the precipitation was filtrated and washed. The drying steps were the same as those of TiO₂ nanoparticles. The bulk density of the as-prepared Ca-Ti-O particles was 1.8 g/cm³.

The ER fluids were obtained by blending the prepared TiO₂ and Ca-Ti-O nanoparticles with silicone oils (50 mPa s at 25°C) via ball milling. The blended ER fluid was dried at 60°C for 1 h before use. The solid content of the ER fluids is denoted by the mass of the ER nanoparticles, e.g., 0.5 g of the TiO₂ nanoparticles blended with 1 ml of silicone oil is denoted by 0.5 g/ml.

The microstructures of the samples were acquired by using Hitachi S4800 field scanning electron microscopy (FE-SEM; Hitachi, Japan) and Tecni F20 transmission electron microscopy (TEM; FEI, United States). By virtue of the Nicolet 6700 transform infrared spectrometer record, Fourier transform infrared (FT-IR; Thermo-Fisher, United States) spectra were from 400 to 4,000 cm⁻¹ using KBr pellets. All infrared spectra were collected 32 times the scan data accumulation at a resolution of 4 cm⁻¹. X-ray diffraction (XRD; Bruker, Germany) spectra were obtained by a D8 Advance/Discover diffractometer using Cu kα radiation. Dielectric properties of the powders and ER fluids were performed via a broadband dielectric/impedance spectrometer with an Alpha-A analyzer (Novocontrol, Germany). The performed voltage is 1.0 V during the dielectric measurements.

The rheological behaviors of the ER fluid were obtained by the Haake RS6000 rotational rheometer with a 15-mm diameter circular plate, and the gap of the height of the parallel plate is 1 mm (Thermo-Fisher, United States). To get the static yield stress, we adopted a stress–strain measurement at a very low shear rate (0.2 rad/s), and the yield stress is the stress value when the viscosity decreases abruptly. The shear stress vs. shear rate curve of the ER was measured using the CR mode with shear rates from 1 to 100 s⁻¹. The external electric field was generated from the SL 300 DC high-voltage generator (Spellman, United States). The controlled humidity environments were achieved by placing in a temperature humidity chamber for calibration at a set humidity for 5 h. Experimental data were carried out with the aid of the software package Rheowin. All data were accomplished at room temperature.

The size and shape of the TiO₂ particles were examined by SEM and TEM. From Figure 2A, it can be observed that the particles have a spherical shape and a rather uniform size distribution, or some irregular agglomerates. The primary particle size is less than 500 nm. The TEM image (inset in Figure 2A) reveals that the surface of the TiO₂ particle is rough and has tiny little balls sticking around it. The powder XRD patterns of the TiO₂ particles is shown in Figure 2B. The pattern almost is a smooth line without the characteristic peak of a TiO₂ crystal, which indicates the amorphous structure of the TiO₂ particles. Figure 2C is the infrared spectra of the TiO₂ particles. The wide peak from 2,900 to 3,500 cm⁻¹ is derived from the asymmetric and symmetric stretching vibration of the −OH group. The peak around 1,647 cm⁻¹ is assignable to the H-O-H bending vibration. The Ti-O band sorption is found in the broad peaks from 900 to 500 cm⁻¹. The peak from 1,455 to 1,380 cm⁻¹ is assignable to CH₂ and CH₃ in-plane deformation distortion, which indicates a small amount of CH₃ and CH₂ groups present on the TiO₂ surface. The peak around 1,038 cm⁻¹ is attributed to the C-O stretching vibration of butanol. All these data show that the titanium oxide has some residual butanol. These are typical precipitated TiO₂ particles. The dipoles of C=O and O-H are 2.3∼2.7 and 1.51 Debye, respectively. Based on the mechanism of the giant ER effect proposed, sufficient active groups on the surfaces of the particles would promote ER response (Shen et al., 2009).

The particle distribution of the TiO₂ ER fluids (1.0 g/ml) was observed by an optical microscope (Figure 3). Without the electric field, the TiO₂ particles are randomly dispersed in the silicone oil. When the electric field strength reaches 0.5 kV/mm, these TiO₂ particles are polarized, and then swiftly form a thick chain-like structure along the direction of the electric field between the positive and negative electrodes. The fibril phenomenon of the TiO₂ particles at the low electric field indicates that the TiO₂ fluids might have high yield stress at the low electric field.

The rheological properties of the TiO₂ ER fluid were evaluated under different electric field strengths. With increasing intensity of electric field strength, the yield stress of the TiO₂ ER fluid significantly rises by several orders of magnitude. When the electric field strength is 2 kV/mm, the yield stress reaches 120 kPa. As the electric field strength increases further, the yield stress deviated from the linearity. Ca-Ti-O is a kind of high-performance electrorheological material (Wang et al., 2005; Cheng et al., 2010; Wu et al., 2016). To compare with our TiO₂ material, we refer to the literature to synthesize Ca-Ti-O electrorheological fluid (The XRD pattern and IR spectrum of Ca-Ti-O particles are shown in Supplementary Figures S1 and S2). The yield stress of the Ca-Ti-O ER fluid is only 14.4 kPa at 2.0 kV/mm, which is 15% that of the TiO₂ ER fluid under the same conditions (Figure 4A and Supplementary Figure S3). Moreover, we noted that the TiO₂ ER fluid demonstrates a much higher ER efficiency [defined as (τ_E–τ₀)/τ₀, where τ_E is the shear stress with an electric field, and τ₀ is the shear stress at zero field] than that of Ca-Ti-O ER fluid at the lower applied electric field. The yield stress of TiO₂ ER fluid reached 122.4 kPa at an electric field of 2.0 kV/mm, and the ER efficiency is about 204. The current density is only 8 μA/cm² leaking through the fluids (the relative curves are shown in Supplementary Figures S4, S6). In contrast, the yield stress of Ca-Ti-O ER fluid is only 14.4 kPa with an ER efficiency of 6 at 2.0 kV/mm (Supplementary Figure S5), which is consistent with the previous report of Shen et al. (2009).

Figure 5 shows the yield stress of TiO₂ ER fluids with different solid particle fractions. It was discovered that the yield stress is positively correlated with the solid concentrations. The yield stress is enhanced with the concentration increasing. When the concentration is less than 1.0 g/ml, the yield stress is small, and the change is not obvious. The yield stress increased significantly from 1.0 to 1.5 g/ml, and the yield stress increased slowly from 1.5 to 2.0 g/ml. When the solid content increases from 1.0 to 1.5 g/ml, the yield stress appears remarkably improved. The yield stress is related to the concentrations of the TiO₂ closely. As for the GER fluids, the high yield stress major comes from the interfacial polarization. Polarized particles will generate interaction force, when the concentration is higher, the distance will be closer. Hence, the interaction force becomes stronger with the concentration. As the solid content further increases, the yield stress is saturated gradually. It is attributed to the polarization (P) of the particles slowed down at high field strengths for high nanoparticle concentrations.

The electric field dependencies of the measured shear stress and solid particle concentrations are shown in Figure 6. The shear stress was measured as a function of the shear rate under various electric field strengths (0–3 kV/mm). The ER fluid with low solid content (0.5 and 1.0 g/ml) is a Newtonian fluid when the electric field is zero, and the diagram of shear stress and the shear rate was a constant slash (Supplementary Figure S7). After applying the electric field (the ER fluid behaved as a Bingham fluid under an electric field), the dynamic shear stress can reach 3.3 kPa (0.5 g/ml) and 23.5 kPa (1.0 g/ml) at 3 kV/mm, respectively. When the shear rate is added to nearly 100 s⁻¹, the shear stress of the two ER fluids is maintained almost constant (plateau region), which suggested that the process of forming the chain structure is still dominated by the electrostatic force at low ER fluids. However, when the solid–liquid ratio increased to 2.0 g/ml, the expelling of the ER fluid materials from the center region of the testing apparatus was observed in the concentration and electric field strength applied. Indeed, the expelling behavior is a generic problem encountered in all the parallel plate types of ER measurement apparatus when a higher concentration of ER fluid was used (Gamota and Filisko, 1991). Interestingly, it has been found that the yield stress results from the potential energy of the repulsion between particle chains in an ER fluid (Song et al., 2012). As for the ER fluids shown in Figure 6C, when the shear rate was increased to 8∼12 s⁻¹, the shear stress has a net reduction due to expelling behavior. Another reason for such behavior may be that the nearest particles separate from the electrode induced by the weaker interaction at the fluid–electrode interface since the commercial electrodes were not made rough (Wang et al., 2007b).

The TiO₂ ER fluids maintain high yield stress during the wide temperature scope from 20°C to 100°C. For representation, the yield stress as a function of temperature for the ER fluid with 2.0 g/ml under various electric field strengths is shown in Figure 7. The best yield stress is around 60°C, which is consistent with the permittivity of the TiO₂ particles prepared in a similar way. Taking in the negligible variation of ε in the tested temperature range, the mismatch in ε between TiO₂ and silicone oil would reach a maximum value and leads to a maximum yield stress at around 60°C. Thus, the TiO₂ fluids presented here are preceded by other known ER fluids and have potential applications in fabricating novel ER devices, for instance, clutches and shock absorbers.

For typical inorganic electrorheological materials, the yield stress, apparent viscosity, and leaking current density of the ER fluids passed through its maximum with the moisture content increasing (Espin and Płocharski, 2007), while the yield stress of organic and polymer ER fluids increases monotonously with water content (Wen et al., 1997; Zhang et al., 2001). For typical dielectric-type ER fluids, yield stress data are properly formulated by the power law: τ_y ∝ E ₀ ^m (m < 2), where E ₀ is the applied electric field (Choi et al., 2001). As indicated in Figure 8A and Supplementary Figure S8, the yield stress of TiO₂ and Ca–Ti–O ER fluids increases with the water content, and the leaking current increases with the water content as shown in Figure 8B, which conforms to the widely accepted dramatic effect of water on the ER effect. However, different trends of yield stress by increasing the electric field strength were observed for ER fluids with different water contents. For TiO₂ ER fluids with very low water content, the slope of E ₀ vs. τ_y is m ≈ 2.0, while the slope approaches 1.0 by increasing the water content. It also showed that there exists a critical value, which corresponds to the giant ER behavior where m = 1 at a high electric field. Unexpectedly, a saturation trend is observed over a range of unscaled water contents greater than the critical value, which also resembles the phenomenon in the log–log scale, the slope of 1 at low electric field and becomes 0.4 at high electric field, the reason is that the interficial polarization dominates at low electric field and approach saturation with electric field increasing resulting in the diminishing of the slope (Supplementary Figure S9).

So, the action mechanism of moisture in the giant electrorheological fluids is governed by its content. It is easy to form a hydrogen bond bridge between particles in fluids with higher moisture levels with the co-action of the electric field and, hence, the stronger the ER properties. When the moisture content is higher, the “current channel” along the fiber structure of the ER fluid will be formed through the free water in the ER fluids immigrating to the surface of the particles, which screened the dipole–dipole interaction and, hence, the saturation of induced the dipole. The increase in electric field will mainly assist the establishment of the current channel, which reduces the local electric field and increases the yield stress. As for those ER fluids with very low water content, no free water was present, and the absorbed water acts as surface polar molecules, and hence, dielectric-type ER behavior will be observed. Meanwhile, the effect of the temperature and humidity on the yield stress of the TiO₂ ER fluids is researched. The effect is similar for temperature and humidity; both have good yield stress during a wide temperature or humidity range, and the yield stress increases with the humidity at the same temperature and electric field. Compared with the curve of the temperature, they have the same tendency; the yield stress increases and then decreases when the temperature increases (Supplementary Figure S10).

It is known that the large polarizability and appropriate dielectric relaxation time of ER particles are critical in producing better and more rapid electrostatic interactions, wherein they can remain as stable structures and rheological properties under shear flow conditions. Polarizations can be divided into four contributions: electronic, atomic, Debye (related to the orientation of dipoles), and interfacial polarization (the Maxwell–Wagner polarization). Among them, electronic and atomic polarization are fast polarizations located in the high-frequency range, which could not supply optimal dielectric or polarization properties for high ER activity (Zhao and Yin, 2002a), while Debye and interfacial polarization are appear at the low-frequency range, and their response speeds are slower. It has long been found that in the heterogeneous systems with a clear interface, the strong polarization effect is originated from ion accumulation in the interfacial area, rather than the dipole polarization for the most part. The interfacial polarization, which was properly described by the Wagner model (Wagner, 1914), rather than other polarization including electronic polarization, atomic polarization, and dipole polarization, is generally admitted as that which reflects the underlying physics. When an electric field is applied, two dominant processes occur in the ER fluid simultaneously: one is the polarization of molecules, and the other is the interfacial polarization between clusters.

Figure 9 shows the relaxation spectrum of the as-prepared TiO₂ ER fluids. It can be observed that the permittivity (ε′) first decreases and approaches a constant value as the frequency increases. The value of ε′ of TiO₂ ER fluid is about six orders smaller than that of Ca-Ti-O ER fluid (Supplementary Figure S11), and an obvious dielectric relaxation peak is observed in the TiO₂ ER fluid, while this peak is not found in Ca-Ti-O ER fluid. This may be attributed to its high conductivity; conductivity plays an important role only in the low-frequency range. Through the measurement of the current density, shown in Figure 4B and Supplementary Figure S6, it is found that both Ca–Ti–O and TiO₂ fluids exhibit the Poole–Frenkel conduction, i.e., ln ⁡ j ∝ E 1 / 2 , one of the important features of giant ER fluids (i.e., polar molecule-dominated ER fluids). The current density is obtained by Ohm’s law: J = σ E (1) where σ is the electrical conductivity. The current density can be written as the product of the mean charge drift velocity v c and the volume charge density, i.e, J = ρ v c (2) and substituting this into Eq. 1, we have, σ = ρ | v c | / | E | = ρ μ = ( n q ) μ (3) where q is the charge carriers, μ is the mobility of the charge in the electric field, and n is the number density of the charge.

The density of carriers, n , is the fastest varying parameter in this case. Therefore, the electric field-dependent part of the carrier density can be expressed as: n = n 0 exp ( − Δ V / k B T ) (4) where n 0 is the carrier density in the absent electric field, Δ V is the electrical potential energy of the carrier, k B is the Boltzmann factor, and T denotes the temperature. Here, Δ V = q 2 / k x + q E x (5) where x is the relative distance between the carrier and the electrode. The minimum value of Δ V is at d Δ V / d x = 0 . The extremum results can be given by: d Δ V d x | x 0 = 0 (6)

Assuming Δ V = − 2 q 3 / 2 E / k , for this equation, then we can obtain hbb : Δ V = 2 q 3 / 2 E / K (7)

Therefore, ln ⁡ J ∝ ln ⁡ n ∝ Δ V ∝ E 1 / 2 (8)

This behavior is indeed confirmed experimentally. As shown in Figure 3A and Supplementary Figure S3, a fine linear relation was obtained for both samples, indicating the dominant effect of polar molecules on the ER effect.

For the TiO₂ samples, a distinct dielectric loss, originating from interfacial polarization, which may have a predominant effect on the ER fluid efficiency, was found at 10³ Hz. For the interfacial polarization, the relaxation time t of one system, when σ p (conductivity of the particle) >> σ m (conductivity of the medium), the relationship between σ p and σ m can be described as (Morgan, 1934; Hao et al., 1998): t = ε 0 ( 2 ε m + ε p ) / σ p (9) where ε 0 , ε m , and ε p are the dielectric constant of the vacuum, continuous (insulating) phase, and disperse (conducting) phase, respectively; σ_p is the conductivity of the dispersed particles (7.2 × 10⁻⁶ and 1.5 × 10⁻⁵ S/m for TiO₂ and Ca-Ti-O powders at 1 kHz, respectively). Therefore, by virtue of testing the ER fluid response time, we can assume that the relaxation time of the polarization the ER response time would control the relaxation time of the polarization and then govern the ER effect. If the Wagner polarization determines the ER effect in truth, according to Eq. 1, the ER fluid answering time should be equivalent to the Wagner polarization relaxation time (Block and Rattray, 1995), which is shown in Figure 10. There is an inverse proportional relation between the relaxation time and the particle conductivity. As shown in Figure 10, the TiO₂ suspension has a larger conductivity and, hence, a shorter response time. When the ER fluid has a particle conductivity of around 10⁻⁷ S/m, the response time should be around 1 ms. This improved polarization may be one factor to induce the enhancement of the ER effect of the titania suspension at a lower electric field. Thus, we assume that both Debye and interfacial polarization affect the ER performance. As for the Ca–Ti–O samples, its large dielectric constant is attributed to the polar molecule-coated nanoparticle, whose effective dielectric constant is dominated by its coating and the existence of the interfaces^. (Wen et al., 2003; Huang et al., 2006; Wang et al., 2007a; Cheng et al., 2008) This induces a large dipole and, thus, a high ER response at a high electric field. (Wen et al., 2003; Huang et al., 2006; Wang et al., 2007a; Cheng et al., 2008; Cheng et al., 2010), while for TiO₂ suspension, the strong ER effect was mainly induced by interfacial polarization.