Continuous Operation on Synthesis and Surface Modification of Rutile Nanoparticles in Designed Microfluidic Reactors.

Synthesis and surface modification of rutile nanoparticles (NPs) are two distinct processes. Conventionally, they should be conducted separately. In this work, synthesis and surface modification of rutile NPs are consecutively performed in a designed microfluidic system, thereby avoiding the pilot processes, giving a high controllability and low-energy consumption of the process, and the preparation process of the coated TiO2 is simplified effectively. Samples synthesized using different strategies are compared, and the results demonstrate that the sample prepared using the microfluidic method shows a smaller particle size (60 nm) and a narrower particle size distribution range than those synthesized using the other two methods. Rutile NPs are most commonly used in terms of suspensions, the stability of the suspensions consisting of the naked and coated samples are assessed in terms of turbidity, agglomeration size, and settlement rate. Response surface methodology is employed to quantify the effects of the factors on the stability of suspensions.

Introduction

Since its commercial production in the early 20th century, TiO2-based materials have been used for various applications such as pigments,[1] UV sunscreens,[2] cosmetics,[3] and sensors.[4] There are four typical crystalline polymorphs of TiO2, namely anatase, rutile, brookite, and T(B). Among them, the rutile phase TiO2 has attracted extensive attention in both engineering and academic territories. Additionally, it is widely acknowledged that new physical and chemical properties emerge when the size of the material made down to the nanometer scale. Nanoparticles (NP) are a typical powdery material, and by far, rutile powder is the most widely used among all the rutile-based materials. Over the past few decades, rutile NPs have been synthesized using various methods, such as sol–gel, hydrothermal, and hydrolysis methods.[5−9] Rutile NPs are most frequently used in terms of suspensions. However, naked rutile NPs exhibit poor dispersibility in an aqueous solution because of their high surface energy, which limits their applications to a large extent. Significant efforts have been devoted to synthesize well-dispersed rutile NPs. To date, surface modification has been considered as an ideal way to enhance the dispersibility of rutile powder in an aqueous suspension. Godnjavec and coworkers presented an investigation on surface modification of rutile NPs with SiO2/Al2O3, and their results showed that surface treatment of rutile NPs with SiO2/Al2O3 could improve dispersion and UV protection property of rutile NPs.[10] Zhang et al. published an article concerning the surface modification of rutile by using the liquid-phase deposition method. They prepared binary amorphous Al2O3/SiO2-coated layers on TiO2 surfaces, and their surface-modified samples showed a high dispersibility in water.[11] Liu et al. prepared SiO2-coated TiO2 powders using a chemical deposition method starting from rutile TiO2 and Na2SiO3. They found that the TiO2 powders with continuous and uniform SiO2-coated layers exhibited higher dispersibility than that of naked TiO2.[12]

Conventionally, the coated rutile NPs are prepared via two steps: (1) synthesis of naked TiO2 NPs; (2) surface modification of the naked TiO2 NPs. Calcination of the intermediate is an essential procedure, which could lead to particle agglomeration easily. It presents great challenges for surface modification of the rutile NPs. Consequently, the conventional methods are energy-wasting and rather complicated, thereby losing precise control over the process. Therefore, it is still challenging to control the synthesis and surface-modification processes precisely. Thus, it is necessary to develop a reliable route for synthesis and surface modification of rutile NPs.

Recently, the microfluidic method finds more and more applications in the field of sustainable chemistry and engineering. Compared to the experiments in classical batch processes, interest in materials preparation is mainly concentrated on the abilities to precisely control the synthesis process because of the high specific interfacial area as a microreactor offers a higher boundary area and mass transfer efficiency, better working conditions, and lower energy consumption.[13−15] Inspired by the advantages of the microfluidic method, some works are published on the synthesis of TiO2 using a microreactor. Indeed, nano-sized TiO2 has been successfully prepared by using the microfluidic method. Cottam et al. produced TiO2 nanorods with a yield of 85% after 6 h and 88% after 22 h using the microfluidic method. They obtained branched nanorods with Y- and H-shaped structures after a reaction of 22 h.[16] Gong et al. synthesized monodisperse hollow titania microspheres. They also controlled the morphology by adding butanol into the system, and it was found that butanol had a critical effect on the morphology of TiO2.[17] Zhang and co-workers synthesized highly dispersed colloidal anatase phase TiO2 nanocrystals in a microfluidic reactor, and they also reported that the microfluidic reactor approach could provide the TiO2 sol with enhanced uniformity of the physical and chemical properties.[18]

Although the microfluidic method has been proved to be a suitable alternative to classical batch methods for synthesizing TiO2 nanostructures, the processes of synthesis and surface modification of TiO2 are still too complicated, which makes the preparation processes physiochemically elusive. Here, we perform a continuous operation on synthesis and surface modification of rutile NPs in a designed microreactor to simplify the preparation of coated TiO2 NPs, thereby making the synthesis process more sustainable and efficient. Additionally, it is of interest to investigate the effects of operating parameters on the stability of the as-prepared rutile suspensions using response surface methodology (RSM).

Results and Discussion

Comparison of the Naked Samples Prepared using Different Approaches

To compare the samples prepared using different methods, the concentration of the solutions and temperatures are kept the same in these three strategies. The field emission scanning electron microscope (FESEM) images and particle size distribution of the naked rutile samples prepared via these three methods are illustrated in Figure . As measured using a laser particle size analyzer (Figure d), the average particle size of the three samples is approximately 60 nm(Figure a), 500 nm(Figure b), and 1000 nm(Figure c). In Figure d, the results are obtained from laser particle measurements, we plot the intensity versus the particle size. To further describe the uniformity of the acquired TiO2 powders shown in Figure a–c, we perform the analysis of variance of Figure d, the standard deviation values of the results in Figure d were calculated using the following equationwhere S is the standard deviation value, x is the particle size determined by the analyzer, and x̅ is the average particle size. The calculated value of S for the samples shown in Figure a–c are 22.29, 76.68 and 239.77, indicating that particle size varied within a wide range in each sample.

Figure 2

Comparison of the samples prepared using different strategies [(a): microfluidic strategy, HAc, 80 °C, channel width: 200 nm; (b): microfluidic and calcination strategies, 80 °C, channel width: 200 nm, 780 °C, 90 min; (c): classical batch strategy, 80 °C, 60 min, 780 °C, 90 min; (d) particle size distribution of (a–c)].

In Figure , the sample prepared using the microfluidic method possesses a much smaller average size than the other two samples, primarily because the particle agglomeration and undesired growth are avoided in the microfluidic synthesis of naked TiO2 NPs. The sample in Figure b is synthesized via the microfluidic strategy and the strategy of calcination of the precursor. The precursor (amorphous hydrated TiO2 particles) is prepared in the microfluidic cell and then followed by calcination of the precursor at 780 °C for 90 min. The particle size sharply increased to 500 nm, indicating that calcination can promote particle growth effectively. As for the results of the batch experiment (Figure c), the precursor is calcined at 780 °C for 90 min. The particle size increased to 1000 nm. Consequently, these results from the above methods present the following facts: (1) the growth of rutile NPs occurs in the precipitation and calcination processes of the precursor and the particle sizes of the rutile NPs are strongly dependent on the particle size of the intermediate. Thus, calcination of the precursor is unfavorable to synthesize TiO2 NPs; (2) microfluidic method exhibits a better controllability over the particle size than the other two methods, and it is an ideal way to synthesize ultrafine TiO2 NPs. In classical batch experiments, the TiO2 products are usually ground mechanically to obtain nanosized products, which significantly increases the surface energy of the acquired particles thereby producing a poor stability of TiO2 suspensions.

Sample Characterization

X-ray Diffraction Analysis

To investigate the phases in the prepared samples, we characterize the naked and coated samples using X-ray diffraction (XRD) technology (Figure ). In the XRD patterns of the naked rutile NPs, no peaks associated with anatase and brookite are detected, indicating that the pure phase of rutile is successfully synthesized in the present work. Interestingly, no peaks associated with the aluminum compound are detected in the coated samples, indicating that the aluminum compounds coated on the surface of TiO2 NPs are amorphous. The surface-modified samples are comparable with the naked samples in crystallinity. Zhang et al. presented a study on surface modification of rutile NPs using a liquid-phase deposition method starting from Na2SiO3·9H2O and NaAlO2. They found that the intensity and peak width of rutile did not alter with the deposition of the Al and Si components.[11] The results presented in this work coincide with those reported in the reference mentioned above, and Dong et al. draw similar conclusions.[19] The results in the present work and the literature demonstrate that the crystal structure of rutile is not affected by the surface-modification process. The reactions between NaAlO2 and water just proceed on the surface of the naked rutile NPs.

Figure 3

XRD patterns of the naked and coated rutile NPs.

Fourier-Transform Infrared Analysis

Fourier-transform infrared (FTIR) analysis is carried out to investigate the chemical bonds of the naked and modified powders, and the results are shown in Figure . The wide absorption region below 1000 cm–1 is because of the vibrations of the Ti–O–Ti bond.[20] Noticeably, the redshift and broadening of the absorption band before 1000 cm–1 occur in the coated samples, which could be ascribed to the combination of Ti–O–Ti and Ti–O–Al vibrations.[19] The FTIR absorption peaks around 1100 cm–1 (1091.26, 1102.04, and 1098.87 cm–1) are related to the Al–O asymmetric stretch.[10] The absorption peaks around 1630 and 3450 cm–1 are assigned to the bending vibrations of the physically surface-adsorbed water (H2O) molecules and stretching vibrations of the surface hydroxyl groups (−OH) on the surface of TiO2.[20] Compared to the Al–O peaks in normal Al-bearing compounds, the Al–O stretching vibration bands are found to be blue shifted, which implies that the alumina-coated layers anchor at the naked TiO2 surfaces in terms of the Ti–O–Al bonds.[21]

Figure 4

FTIR spectra of the naked rutile NPS prepared using HAc acid with different temperatures, (1) 75, (2) 80, (3) 85 °C, and the coated TiO2 samples prepared at 70 °C with different mole ratios of NaAlO2 to TiO2 of (1#) 1:25, (2#) 1:50, and (3#) 1:75.

Morphologies of the Naked and Surface-Modified Rutile NPs

Figure shows the morphologies and particle size distribution of the prepared naked and coated samples using the microfluidic method proposed in this work. As measured using the laser particle size analyzer, the average particle size of the as-synthesized TiO2 particles shown in Figure a–c is approximately 60, 30, and 20 nm. In the present work, the rutile NPs are produced via the following reaction

Figure 5

HRTEM images of the naked rutile NPS prepared using different acids at 80 °C and the coated TiO2 samples prepared at 70 °C; (a) naked NPs prepared by nitric acid; (b) coated sample of a; (c) naked NPs prepared by acetic acid; (d) coated sample of c; (e) naked NPs prepared by benzoic acid; and (f) coated sample of e.

Thus, hydrogen ions are a by-product in the preparation of rutile NPs. Compared with nitric acid, acetic acid and benzoic acid possess weak acidity, which led to lower H+ in the microchannels in the preparation process of rutile NPs. At lower pH, more water molecules are bound to the Ti4+ center, which favors corner-sharing and leads to a faster formation rate of rutile TiO2. Therefore, the rutile NPs obtained by using nitric acid have a longer time to grow up thereby yielding a larger particle size of rutile NPs. A clearer image of the rutile NPs prepared by nitric acid on the nanoscale is illustrated in Figure b. The parallel lattice fringe distances of the samples are measured, the obtained values in the sample are 0.25 and 0.325 nm, which are close to the (1 0 1) and (1 1 0) planes of the rutile crystals.[22−24] As for the surface-modification samples, the coatings are successfully prepared on the surface of the naked rutile NPs. Morphologies of the coatings are uniform, indicating stable mass transfer processes in the surface-modification processes of the rutile NPs in the microreactor. This implies the microfluidic method proposed in this work is feasible to perform surface modification of the naked rutile NPs.

Stability of Rutile Suspensions

Rutile NPs are most commonly used in terms of suspensions, stability of the rutile-containing suspensions is a key property for further application. To investigate the stability of suspensions consisting of the naked and surface modified rutile NPs and further understand the effects of surface modification on the stability of the suspensions, we took three factors, turbidity of the suspensions, agglomeration size of the naked and coated rutile NPs, and settlement ratio of NPs in suspensions after 25 days to evaluate the stability of the naked and coated rutile suspensions. The results are summarized in Table . All the naked samples possess larger turbidity and agglomeration size than the coated ones, while the trend for the settlement ratio of NPs in suspensions is opposite, and this is more evident for smaller particles, which is mainly because of the stronger particle agglomeration in the naked powder than that in coated ones. These results demonstrate that surface modification can significantly enhance the stability of the suspensions. It is reported that TiO2 molecules exhibit strong polarity in water, the lifetime of the hydrogen bond in interfacial water molecules is several times longer than that in bulk water because of the strong water–TiO2 interactions, and the surface polarity of TiO2 enhances the water–TiO2 interactions.[25] Thus, it is quite difficult to disperse the naked rutile NPs in water because of their small size and strong polarity.

Table 1

Stability of the Naked and Surface Modified Rutile NPs Suspensiona

sample	no.	turbidity (NTU)	agglomeration size (nm)	settlement rate of NPs in suspensions (mass fraction) (%)
Naked Samples
	1	32	190.53	97.50
	2	46	182.60	97.35
	3	52	163.58	95.15
	4	63	161.73	95.06
	5	72	156.84	95.02
	6	81	153.92	95.02
	7	83	149.87	94.59
	8	84	143.69	94.06
	9	90	142.73	93.58
	10	97	141.09	93.06
Coated Samples
	1#	276	85.63	6.75
	2#	287	73.50	6.73
	3#	301	62.50	5.75
	4#	312	59.74	5.68
	5#	318	55.62	5.56
	6#	321	52.75	5.53
	7#	327	48.59	5.21
	9#	347	40.27	4.75
	10#	359	35.70	4.23

Note: the number i# is the coated sample of the naked sample i.

Evaluation of Effects on the Stability of Suspensions by RSM

The main aim of surface modification of the rutile NPs is to obtain suspensions with high stability. However, it is challenging to quantify the relationships between the experimental conditions and stability of the rutile NP suspensions. In the present work, the surface-modification samples are placed in deionized water, and then we used the settlement ratio of the rutile NPs in the suspensions as the index to quantify the effects of the experimental conditions on the stability of the suspensions. RSM and Box–Behnken design (BBD) in RSM are employed to evaluate the stability of rutile NPs suspensions. According to this method, the experimental times can be represented as[26,27]where k is the number of influenced factors, and Cp is the central points, respectively. The common form of the second-order polynomial equation is presented in eq where Y is the predicted response value, β0 is the intercept, β is the linear coefficient, β is the quadratic coefficient, β is the interaction coefficient, x and x are the independent factors, and ε is the random error. Table shows the three factor design table for BBD experiments. The experimental data are analyzed using design-experiment version 8.0.6 software.

Table 2

Factor Design Table for BBD Experiments

	coding level
variables code	–1	0	1
mass fraction of rutile NPs/% (x1)	0.075	0.1	0.125
ultrasonic time/s (x2)	250	300	350
stirring time/min (x3)	308	318	328
pH of the suspension	6.5	7	7.5

According to the fitted results from RSM, the estimated equation for output response y in terms of coded factors can be represented aswhere y is the stable time, x1, x2,x3, and x4 represent the TiO2 content in suspension, ultrasonic time, stirring time, and pH of the suspension.

The variances of the predicted results are summarized (Table ), a low p-value (≤0.05) implies a significant influence on the stability of the suspensions. It is highly significant when the p-value is smaller than 0.01.[28] In Table , F and p values of the model are 125.39 and less than 0.0001, indicating that the regression equation is highly significant. The single factors, x1, x2, and x4, are also highly significant. For interaction factors, x1x3, x12, x22, x32, and x42 are highly significant, and x2x4 is significant. The p-value of the lack of fit is 0.2742, much larger than 0.05, indicating a small error between the model and practical experiments. In summary, our model possesses a high fitness toward the experimental results. Additionally, the difference between Adj-R-squared and R-squared is 0.1155 < 0.2, which implies a good relationship between the predicted and experimental values. We can evaluate the effect of the factors by comparing the F-values of each factor. The effect of the single factor follows the trend: pH(x4, 515.47) > TiO2 content in suspension(x1, 37.10) > ultrasonic time (x2, 24.57) > stirring time(x3, 3.32). The order of the interaction factors is: x1x3(149.81) > x2x4(4.78) > x1x4(1.18) > x3x4 = x1x2(4.75 × 10–4) > x2x3(4.75 × 10–4). The effect of the square factors follows the trend: x42(950.55) > x12(173.93) > x32(101.34) > x22(16.03).

Table 3

Variance of the Results of the Phenol Degradation Rate as a Response Variablea

source	sum of squares	df	mean square	F value	p value	mark
Model	9.13	14	0.65	125.39	<0.0001	**
x1	0.19	1	0.19	37.10	<0.0001	**
x2	0.13	1	0.13	24.57	0.0003	**
x3	0.350.017	1	0.350.017	3.32	0.0935
x4	2.68	1	2.68	515.47	<0.0001	**
x1x2	2.47 × 10–6	1	2.47 × 10–6	4.75 × 10–4	0.9830
x1x3	0.78	1	0.78	149.81	<0.0001	**
x1x4	6.15 × 10–3	1	6.15 × 10–3	1.18	0.2981
x2x3	0.00	1	0.00	0.00	1.0000
x2x4	0.025	1	0.025	4.78	0.0493	*
x3x4	2.47 × 10–6	1	2.47 × 10–6	4.75 × 10–4	0.9830
x12	0.90	1	0.90	173.93	<0.0001	**
x22	0.083	1	0.083	16.03	0.0018	**
x32	0.53	1	0.53	101.34	<0.0001	**
x42	4.94	1	4.94	950.55	<0.0001	**
residual	0.062	12	5.20 × 10–3
lack of fit	0.059	10	5.85 × 10–3	3.02	0.2742
pure error	3.873 × 10–3	2	1.94 × 10–3
total	9.19	26

Note:* means significant, ** denotes highly significant.

Normal distribution of the residuals for the stable time is illustrated in Figure a, the linear correlation coefficient is 0.9856, indicating a good linear relationship between normal probability and internally studentized residuals. Reliability diagram of the quadratic regression equation with stable time is illustrated in Figure b, the linear correlation coefficient and correction factor in Figure are close to 1(0.9872 and 0.9863), which further confirms the good reliability of the model and feasibility of RSM in the present work.

Figure 6

Model reliability analysis diagrams [(a) normal plot of residuals; (b) predicted vs actual values].

To provide a better visualization of the effects of the factors on the stability of the suspensions, 3-dimensional response surfaces are performed (Figure a–f). The corresponding counter maps are provided in Figure S1. Figure a shows the effects of the TiO2 content and ultrasonic time on the stability of the suspension, and the stirring time and pH are fixed at 45 min and 7.0. As shown in Figure a, with the increasing of the TiO2 content in the suspension, the stable time of the suspension changes significantly, while compared with the effect of the TiO2 content, the ultrasonic time shows less impact on stability of the suspension. This agrees with the results shown in Table . In Figure S1a, we can see a relatively round shape of the counter map, which indicates that the interactions between these two factors are weak.[27]Figure b shows the effects of the TiO2 content and stirring time with an ultrasonic time of 200 s and pH 7.0. Additionally, as shown in Figure S1b, the counter map is oval-shaped, which indicates a strong interaction between these two factors. The stability did not increase as the stirring time increased, and this is a counterintuitive phenomenon, simply because of the interactions between these two factors. Figure c shows the effect of the TiO2 content and pH on stability of the suspensions with an ultrasonic time of 200 s and a stirring time of 45 min. The stability is very sensitive to pH; while the alternation of stability is barely observed when the TiO2 content is varied from 0.25 to 0.75%. This is because of the different electrostatic forces among the NPs at different pH values. The round-shaped counter maps in Figure c indicate a rather weak interaction between the two factors. In Figure d, the TiO2 content and pH are 0.5% and 7.0, the stirring time and ultrasonic time did not show a critical effect on the stability of the suspensions. Figure S1d also demonstrates that there is little interaction between the two factors. Interestingly, the stability did not increase with the increasing of the stirring time. It is listed as a not significant factor in Table . Figure e,f shows a similar trend: the stability of the suspensions is very sensitive to pH. At pH 6.9, it yields the largest stable time, which is mainly because the NPs are electrically neutral around this pH. According to the RSM, the optimal conditions for stability of the TiO2 suspensions is: TiO2 content: 0.42 wt %, ultrasonic time: 271.37 s, stirring time: 48.73 min, and pH 7.06. A stability of 27.3295 days can be attained under these optimal conditions. We did the experiments under the optimal conditions; the results are 27.40 days. The relative error between the predicted and experimental results is 0.258%, indicting RSM is an ideal way to evaluate the stability of the suspensions. The microfluidic method is an exciting alternative to batch methods for the synthesis of nanostructures, a further study on how to scale it up and improve the production capability is still needed in order to use it on the industrial scale.

Figure 7

Response surfaces for the effects of different variables on stabilities of the suspensions [(a) effect of TiO2 content and ultrasonic time; (b) effects of TiO2 content and stirring time; (c) effect of TiO2 content and pH; (d) effect of stirring and ultrasonic time; (e) effect of sonic time of pH; and (f) effect of stirring and pH].

Conclusion

In this work, we proposed a continuous operation approach for synthesis and surface modification of rutile NPs by using a designed microfluidic system. The preparation process of coated rutile NPs is substantially simplified. The particle size of TiO2 is much smaller than that produced using batch methods. Predictably, it is a more energy-saving and simple method than classical batch methods to obtain TiO2 NPs. Additionally, well-defined coated rutile NPs with enhanced stability in water are obtained using the microfluidic method. Samples synthesized using different strategies are compared. The samples obtained using the microfluidic method are smaller (60 nm) and have more uniform particle sizes than those prepared using batch experiment and microfluidic-calcination methods. The aluminum compounds coated on the surface of TiO2 NPs are amorphous. The surface-modified samples are comparable with the naked samples in crystallinity. The surface-modification process did not influence the crystal structure of rutile. Stabilities of the suspensions are evaluated using RSM, the linear correlation coefficient and correction factor are 0.9872 and 0.9863, which demonstrated that the model presented in this study has good reliability and feasibility. The optimal conditions for stability of the TiO2 suspensions are obtained.

Experimentation

Instrumentation

In this work, we designed and commissioned Shenyang Zhongshan Precision Instrument Co., Ltd. to manufacture a microfluidic system for the synthesis and surface modification of TiO2 NPs (Figure a,b). This system mainly consists of three pumps (Fluid Equipment Co., Ltd, Lanzhou, China. Model: LSP01-1A), one microreactor with a channel width of 200 μm, and another micromixer with a channel width of 150 μm. The microchannels in the microreactors and micromixers are embedded in chips, which are fixed in stainless steel modules. The microchannels distribute uniformly in the chips with a spacing of 2 mm, the channel lengths of the microreactors and micromixer are 120 and 80 cm, respectively. The channel width can be varied by changing the modules if needed. The stainless-steel modules could be heated in an oil bath when necessary.

Figure 1

(a): Schematic illustration of the synthesis and surface-modification process of rutile NPs; (b): schematics of the microreactors in (a).

Materials

TiCl4 and Na2AlO2 are used as the titanium resource and surface-modification agents, respectively. Nitric acid (0.05 mol/L), acetic acid (0.05 mol/L), and benzoic acid (0.05 mol/L) are used to investigate the morphologies of the NPs obtained using different acids. All the mentioned chemical reagents are purchased from Sinopharm Chemical Reagent Co, Ltd., China, and they are used without any further purification.

Synthesis and Surface Modification of Rutile NPs

For synthesis of rutile NPs, TiCl4 and acid solution are injected into the microreactor via two pumps, which are operated at 0.3 MPa (for TiCl4) and 0.05 MPa (for acid solution). The flow rates of TiCl4 and acid solution are 50 μL/min and 0.5 mL/min. TiO2 suspensions are produced in microreactors in the oil bath, which is heated to experimental temperatures. The as-prepared TiO2 suspensions are mixed with the NaAlO2 solution by using a pump operated at 0.15 MPa in the micromixer in the oil bath at 70 °C. The mole ratios of NaAlO2 to TiO2 are varied from 1:25 to 1:75. Afterward, the mixture is centrifuged at 19,000 rad/min to separate the samples from the obtained suspensions. The obtained nano-powder is dried at 80 °C for 5 h in an oven. The experimental setup is schematically illustrated in Figure . To compare the samples prepared using different methods, the rutile samples are synthesized using the other two strategies: (1) the microfluid-calcination process; (2) the batch experimental process. For the microfluid-calcination process, the microfluidic part of the experiments is the same as the process as mentioned above but deionized water is used instead of acid. Consequently, the resulting product is amorphous TiO2. To obtain a pure phase of rutile TiO2, the samples are calcined in a furnace at 780 °C for 90 min. For the batch synthesis experiments of rutile powder, TiCl4 is added drop by drop into stirred water to obtain a TiO2 product. Afterward, the precipitate is separated and dried in an oven at 80 °C for 24 h. To obtain a pure phase of rutile TiO2, the samples are also calcined at 780 °C for 90 min.

A FESEM from Carl Zeiss company is employed at 20 kV to observe the morphologies of the naked and coated samples. In addition, a high-resolution transmission electron microscope (HRTEM; KEM-ARM200F, JEOL Ltd. Tokyo, Japan) is also used to further characterize the samples. 0.1 wt % suspension of TiO2 was used to prepare a specimen for HRTEM and FESEM, it is dispersed using ultrasound in absolute ethanol for 2 min, and then it is deposited on a metallic sheet and carbon-coated copper grids for FESEM and HRTEM experiments. Afterward, the specimen is dried in an oven at 80 °C, 30 min. The samples for FESEM observation are coated with gold using an ion sputter coater (108Auto, Cressington Scientific Instruments) to enhance the conductivity of the samples. To investigate the phase composition of the naked and coated samples, an XRD instrument (MPDDY2094, Netherlands) with Cu Kα irradiation is used and operated at 30 kV, and the scan range is 10–90°. A Thermo Nicolet-380 FTIR spectrometer (Thermo Fisher Scientific, US) is used via the KBr pellet pressing method to measure the structure of the samples. The following parameters are used in the measurement: scan range: 400–4000 cm–1, integration time: 100 ms; data pitch: 2 cm–1; and number of scans: 32. Prior to the measurement of TiO2 NPs, a KBr sample is measured to perform the baseline correction. The particle size of the samples is measured using a laser particle size analyzer (Nano 90, Malvern Instruments Co., Ltd. UK). Prior to particle size determination, 0.05 wt % suspensions of the powder are dispersed in absolute ethyl alcohol using ultrasound (FS-450, Shenxi Ultrasound Instruments, Shanghai, China) for 5 min to break particle agglomeration.

Stability Test of Suspensions

For a typical stability test, a 0.1 wt % dispersed suspension is placed into a small tube and kept stationary for 25 days, then the upper suspension is dumped, and the sediments at the bottom of the tube are dried in an oven at 80 °C. Finally, the weight of the precipitation is measured using an electronic balance to calculate the settling ratio of rutile NPs in the suspensions. To test the turbidity of the suspensions, the TiO2 suspensions are dispersed by ultrasound with the same solid content and then centrifuged in a centrifugal tube at 18,000 rpm for 25 min, and the upper dispersions with the same volume are obtained. The turbidities are determined using a turbid meter (2100P, HACH Company, US).