Fate Determination and Characterization of Food Additive Silicon Dioxide and Titanium Dioxide in Commercial Foods

Background : Silicon dioxide (SiO 2 ) and titanium dioxide (TiO 2 ) are ones of the most widely used food additives as an anti-caking and a coloring agent, respectively, in the food industry. Understanding particle, aggregate, or ionic fates of two additives in commercial products is of importance to predict their potential toxicity. Methods : Triton X-114 (TX-114)-based cloud point extraction (CPE) methods for two additives were optimized in food matrices. Their particle or ionic fates in various commercial foods were determined by the CPE, and the physico-chemical properties of separated particles were further characterized. Results : SiO 2 and TiO 2 were primarily present as particle forms without changes in constituent particle size, size distribution, and crystalline phase. The maximum solubilities of SiO 2 and TiO 2 were 5.5% and 0.9%, respectively, depending on food matrix type, supporting their major particle fates in complex food matrices. Conclusions : These findings will provide basic information about the fates and safety aspects of SiO 2 and TiO 2 additives in commercial processed foods.


Introduction
Silicon dioxide (SiO 2 ) and titanium dioxide (TiO 2 ) are the most widely used food additive particles and have been widely applied to the food industry as an anti-caking and a coloring agent, respectively [1][2][3].SiO 2 is used in confectionary, powdered mixtures, and seasonings to prevent ingredients from clumping together.TiO 2 is mainly added in confectionary including chocolates, candies, bakeries, snacks, and chewing gums due to its whitening pigment property.In the Unites States, the Food and Drug Administration recommends that the amounts for SiO 2 and TiO 2 are below 2% and 1% by weight of the food, respectively [1,2].In the European Union (EU), SiO 2 and TiO 2 were registered as E551 with maximum levels at 1% in dried powdered foods and as E171 with no maximum level specified, respectively [4,5].Current regulations do not specify the particle size range or size distribution of food additive SiO 2 and TiO 2 .Rapid development of nanotechnology may lead to manufacture nano-sized SiO 2 and TiO 2 particles, which have large specific surface area to volume ratio, high reactivity, and different biological responses compared with bulk-sized particles [6,7].Indeed, recent studies demonstrate that nanoparticles (NPs) ranged from 1 to 100 nm are present in commercially available food additive SiO 2 and TiO 2 , raising concerns about their potential toxicity [8][9][10][11].
Food additives are utilized in processed foods where thermal, mixing, and agitating treatments are often applied, which can cause the formation of aggregates or dissolution of particles [4,12].Moreover, the interactions between food additives and food components can occur, which can also affect the fates of food additive particles [13][14][15][16].Determining whether SiO 2 and TiO 2 are present as intact particles, aggregates, or dissolved forms is important to understand and predict their potential toxicity.Orally taken SiO 2 is generally considered to be not toxic at actual usage levels [17][18][19].However, some contradictory results were also reported, showing its potential toxicity in terms of oxidative stress, inflammation response, and intestinal barrier dysfunction, although most studies were performed using in vitro systems [20][21][22].In the EU, food additive TiO 2 (E171) is considered no longer safe due to uncertainty of its genotoxicity [23].TiO 2 is known to induce reactive oxygen species (ROS), which is critical to cause oxidative stress-related diseases [15,24].Recently studies demonstrated that food-grade TiO 2 can be involved in the intestinal barrier dysfunction, colorectal cancer, and development in offspring rodents [22,25,26].Hence, the information about the fates of SiO 2 and TiO 2 will be necessary and useful to answer the question as whether their toxicity is related to particle or ionic fates.
Most studies on the fate determination of NPs have applied harsh conditions such as acid and heat treatments to digest organic matrices, which can surely alter their dissolution and particle fates [15,27].A detergent-based cloud point extraction (CPE) was used for the detection of ions in matrices and environments after filtering, acid treatment, and dry-ashing [28,29].In the previous study, we developed a Triton X-114 (TX-114)-based CPE method to separate zinc oxide (ZnO) particles as intact forms from com-mercial foods and biomatrices without pre-treatments [14].The CPE approach was also developed for SiO 2 in biomatrices and its fates were determined in cell lines and tissues [17].In this study, the CPE method was further optimized for the most widely applied food additive particles, SiO 2 and TiO 2 , in food matrices to determine their dissolved, aggregated, or particle fates in commercial food products.Furthermore, the physicochemical properties of separated SiO 2 and TiO 2 particles from commercial foods by the CPE were characterized in terms of constituent particle size, crystalline phase, and solubility.

Characterization
Particle size and shape of SiO 2 or TiO 2 were analyzed by field emission transmission electron microscope (FE-TEM; JEM-2100F, JEOL, Tokyo, Japan) equipped with energy-dispersive X-ray spectroscopy (EDS).The particles were dispersed in distilled and deionized water (DDW) and the suspensions (0.1 mg/mL) were prepared in ethyl alcohol solution and sonicated for 15 min (160 W, Bransonic 5800, Branson Ultrasonics, Danbury, CT, USA).The suspensions (5 µL) were dropped on a carbon-coated copper grid (200 mesh, PELCO® TEM Grids, Ted Pella Inc., Redding, CA, USA).After drying at room temperature, TEM images were acquired at an accelerating voltage of 200 kV.The average particle sizes and size distributions of SiO 2 or TiO 2 particles were determined using ImageJ software (version 1.53k, National Institutes of Health, Bethesda, MD, USA).
Crystal structure of SiO 2 or TiO 2 was determined by powder X-ray diffraction (XRD) patterns using X-ray diffractometer (SmartLab, Rigaku Co., Tokyo, Japan) with Ni-filtered CuKα radiation (λ = 1.5418Å, a voltage of 40 kV, a current of 40 mA, a scan range of 5-80°with a step size of 0.02°, and a scanning rate of 3°/min).
Hydrodynamic diameters and zeta potentials of SiO 2 or TiO 2 were measured by dynamic light scattering (DLS) and electrophoretic light scattering (ELS), respectively, using a Zetasizer Nano System (Malvern Instruments, Worcestershire, UK).SiO 2 or TiO 2 suspension (0.1 mg/mL) was prepared in DDW or humic acid (HA) solution (10 µg/mL) and stirred for 30 min.After further sonication for 15 min, the suspensions (1 mL) were immediately put in plastic cuvettes and the measurements were carried out at room temperature.CPE method for SiO 2 and TiO 2 was optimized with SiO 2 or TiO 2 dispersions (0.1 mg/mL) in DDW or HA solution (10 µg/mL), respectively.After stirring for 30 min and sonication for 15 min at room temperature, the suspensions (7 mL) were transferred to bottom glass centrifuge tubes (15 mL) and the pH was adjusted to 3.0 and 4.0 for SiO 2 and TiO 2 , respectively, with NaOH or HNO 3 solution.Next, TX-114 (0.5 mL of 5% (w/v)) and NaCl (0.75 mL of 0.2 M) solutions were added in the suspensions and diluted to 10 mL with DDW.The mixed solutions were incubated for 30 min at 45 °C to promote phase separation and centrifuged at 2500×g for 5 min at 25 °C.The precipitates and the supernatants where particles and ions were present, respectively, were digested for inductively coupled plasmaatomic emission spectroscopy (ICP-AES) analysis (JY2000 Ultrace, HORIBA Jobin Yvon, Longjumeau, France) as described in "2.7.Digestion of Organic Materials, Microwave digestion, and ICP-AES Analysis".

CPE Application for SiO 2 and TiO 2 in Food Matrices
The CPE developed for SiO 2 or TiO 2 was applied by spiking SiO 2 or TiO 2 into representative food matrices, powdered sugar or casein.The contents of SiO 2 and TiO 2 were adjusted to be 2% (w/w) and 1% (w/w) based on FDA recommendations, respectively.SiO 2 suspension (0.1 mg/mL) was spiked into 0.1 g of powdered sugar or casein at the concentration of 2% (w/w) and dispersed in DDW (7 mL).TiO 2 (0.1 mg/mL) was spiked into 0.1 g of food matrices at the concentration of 1% (w/w) and dispersed in HA solution (7 mL).After stirring for 30 min and sonication for 15 min, the same procedure was applied as described in "2.3.Optimizations of CPE for SiO 2 and TiO 2 ".

Fate Determination of SiO 2 and TiO 2 in Commercial Foods
Commercial foods (10 g), such as candy, powder, coffee mix, milk tea, chocolate, sauce, and snacks, containing SiO 2 or TiO 2 as a food additive on product labeling were homogenized in an agate mortar.Homogenized commercial products were dispersed in 7 mL of DDW and 7 mL of HA solution for the determination of SiO 2 and TiO 2 , respectively.After stirring for 30 min and sonication for 15 min, the same procedure was applied as described in "2.3.
Optimizations of CPE for SiO 2 and TiO 2 ".

Dissolution Properties of SiO 2 and TiO 2 in Food Matrices and Commercial Foods
Dissolution properties of SiO 2 and TiO 2 in food matrices were evaluated in powdered sugar or casein, spiked with SiO 2 (2% (w/w)) or TiO 2 (1% (w/w)).Commercial foods (10 g) containing SiO 2 or TiO 2 as a food additive indicated on product labeling were homogenized in an agate mortar.Food matrices (0.1 g) or homogenized commercial products (0.1 g) spiked with SiO 2 and TiO 2 were dispersed in 10 mL of DDW and HA, respectively, and stirred for 30 min, followed by sonication for 15 min at room temperature.Further incubation was carried out for 30 min at 45 °C, as described in "2.3.Optimizations of CPE for SiO 2 and TiO 2 ".The suspensions were then centrifuged at 16,000×g for 15 min to collect the supernatants containing dissolved Si or Ti.The Si and Ti concentrations in the supernatants were determined by ICP-AES analysis after pre-digestion with HNO 3 and H 2 O 2 as described in "2.7.Digestion of Organic Materials, Microwave digestion, and ICP-AES Analysis".

Digestion of Organic Materials, Microwave Digestion, and ICP-AES Analysis
Si or Ti concentrations were quantified by measuring total Si or Ti contents using ICP-AES analysis, with Si or Ti standard solutions of different concentrations based on our previous reports [15,30].It was reported that the amount of SiO 2 and TiO 2 in commercial foods are ranged from 1.5 to 7 mg/g and 0.2 to 10 mg/g, which are enough to be detected by ICP-AES [15,31,32].Organic materials in the supernatants obtained after dissolution or CPE experiments were digested with 10 mL of ultrapure HNO 3 and 1 mL of H 2 O 2 at 180 °C until the solution was colorless and entirely evaporated.The precipitates containing SiO 2 or TiO 2 particles after CPE application were digested in perfluoroalkoxy microwave digestion vessels using a microwave system (ETHOS EASY, Milestone Srl, Sorisole, Italy).Briefly, SiO 2 particles were digested with 6 mL of 70% HNO 3 and 1 mL of 40% HF.TiO 2 particles were digested with 6 mL of 70% HNO 3 and 2 mL of 40% HF [33,34].The samples were digested for 55 min at 1600 W by irradiation at 120, 160, and 210 °C for 15, 10, and 30 min, respectively, followed by holding for 1 min.After digestion, all samples were diluted to suitable volumes with DDW, shaken up and down, and analyzed by ICP-AES.Method blanks were determined by performing the same procedure in the absence of Si or Ti samples.Instrument operating conditions are described in Supplementary Table 1.

Statistical Analysis
Results were presented as means ± standard deviations.One-way analysis of variance with Tukey's test was performed using the SAS Ver.9.4 (SAS Institute Inc., Cary, NC, USA) to determine the significances of intergroup differences.Statistical significance was accepted for p values of <0.05.

Optimization of CPE for SiO 2 and TiO 2
To optimize TX-114-based CPE methods for SiO 2 and TiO 2 , commercially available food-grade SiO 2 and TiO 2 were purchased and the characterization of each particle was carried out.Supplementary Fig. 1A,B show that the constituent particle sizes of SiO 2 and TiO 2 by TEM analysis were 14 ± 4 nm and 109 ± 35 nm, respectively.Zeta potential values of SiO 2 and TiO 2 under different pH conditions are presented in Supplementary Fig. 1C, showing isoelectric points (IEP) of the former and the latter were pH 1.9 and 3.1, respectively.It was reported that the electrostatic repulsion of particles is minimized at pH close to the IEP due to reduction of Brownian motion, and thus, the pH where zeta potentials are close to zero is optimal for the formation of NPs captured in TX-114-based micelles [35,36].But, the addition of NaCl during the CPE process elevated zeta potential values of NPs.Hence, the pH values for the CPEs were set at higher pHs than IEP values, 3.0 and 4.0 for SiO 2 and TiO 2 , respectively.As a consequence, particles can be captured in TX-114-based micells as precipitates, whereas ionized forms are present in supernatants after phase separation using CPE followed by centrifugation.
Table 1 demonstrates the changes in hydrodynamic diameters and zeta potentials before and after CPE application.The hydrodynamic diameters of SiO 2 suspension in DDW were statistically same to those obtained by CPE, whereas SiO 2 in HA solution had increased hydrodynamic diameters after CPE application (Table 1).In case of TiO 2 , no increase in hydrodynamic diameters after CPE was found when it was dispersed in HA solution.It is worth noting that particles captured in TX-114 micelle by CPE must have the same particle size distribution compared with pristine particles without aggregation or dissolution, which is critical for fate determination as intact forms.Hence, SiO 2 and TiO 2 were dispersed in DDW and HA solution, respectively, for CPE application.This result also suggests that the CPE process can capture SiO 2 and TiO 2 as intact forms without aggregate formation or degradation.On the other hand, Table 1 shows that the zeta potential values of both SiO 2 and TiO 2 changed to less negative charges after CPE, probably resulted from NaCl addition during the CPE process.
Total recoveries of SiO 2 and TiO 2 were about 92.8% and 97.5%, respectively, and only 0.6% and 0.1% of SiO 2 and TiO 2 were detected as Si and Ti ionic forms, respectively, after CPE application (Fig. 1A,B).Meanwhile, the solubility of pristine SiO 2 and TiO 2 was assessed under the CPE conditions, showing 0.5% and 0.2% solubilities for the former and the latter, respectively, without significant differences between solubility and ionic fates (Fig. 1B).These

CPE Application for SiO 2 and TiO 2 in Food Matrices
Representative food matrices such as powdered sugar and casein were spiked with SiO 2 or TiO 2 , and the CPE methods optimized were applied.Powedered sugar and casein were chosen as food matrices due to their frequent usage in confectionary.Fig. 2A shows that there were no statistical changes in hydrodynamic diameters between pristine SiO 2 or TiO 2 and particles recovered from particlesspiked powdered sugar or casein by CPE application, suggesting that the CPE methods can recover SiO 2 and TiO 2 particles from food matrices as intact forms without aggregation or dissolution.The recoveries of SiO 2 and TiO 2 in powdered sugar or casein as particle forms were ranged from 93.2% and 102.6% (Fig. 2B,C).Only ~0.7% and 0.1% of SiO 2 and TiO 2 were detected as ionic forms, respectively, regardless of matrix type.Total recoveries of both particles and ions were 93.3%-103.3% for all cases, implying reliability of the results.All the results suggest that SiO 2 and TiO 2 added in food matrices can be separated as intact particle forms by applying the CPE.
On the other hand, the solubility of SiO 2 and TiO 2 was also evaluated because particles can dissolve to some extent in food matrices.As shown in Fig. 2D, the solubilities of SiO 2 and TiO 2 were ~0.9% and 0.1%, respectively, without significant differences compared with ionic forms (%) obtained by CPE (Fig. 2B,C).This result clearly supports that the CPE methods can separate particles from ionic forms in food matrices.

Validation of Analytical Methods for SiO 2 and TiO 2 in Food Matrices
Quantitative analytical methods using acid digestion in a microwave system were validated by calculating recovery (%), linearity (coefficient of determination values, R 2 ), accuracy (relative error, RE), precision (coefficient of variation, CV), limit of detection (LOD), and limit of quantification (LOQ).Table 2 shows that the recoveries of pristine SiO 2 and TiO 2 or SiO 2 -and TiO 2 -spiked powdered sugar or casein were range from 90.18% to 108.32%.The CV and RE values for all the cases were 0.40%-9.69%and -9.82% to 8.32%, respectively.Fig. 3 demonstrates a good linearity for pristine particles and particles-spiked in food matrices.All the results indicate reliable analytical parameters for SiO 2 and TiO 2 in food matrices, which is consistent with other previous reports [15,30,37,38].
On the other hand, the LOD and LOQ values were 0.03-0.10µg/mL and 0.10-0.30µg/mL for SiO 2 , and 0.01-0.13µg/mL and 0.03-0.38µg/mL for TiO 2 , respectively.These are similar or lower values compared with those obtained by other results, showing that the LOD and LOQ were 0.07 µg/g and 0.20 µg/g for SiO 2 , and 0.03-0.34µg/mL and 0.09-1.04µg/mL for TiO 2 , respectively [15,31].Sensitively low LOQ and LOD values were obtained, suggesting that the analytical methods by acid digestion in a microwave system are effective and reliable.

Separation of SiO 2 and TiO 2 from Commercial Foods and Their Fate Determination
The CPE methods were applied to eight commercial products available on the market indicating SiO 2 or TiO 2 as a food additive on product labelling.Fig. 4 shows that 90.9%-101.8% of SiO 2 were present as particle forms, only 0.02%-7.3%ionized forms were detected.When total recovery percentage (91.3%-108.3%) of both SiO 2 particles and Si ions from products was considered, about 93.5%-99.9% of added SiO 2 were found as intact particles.In case of TiO 2 , particle forms were detected in the range of 92.4% to 105.9%, whereas ionic forms were only found at 0.1%-0.9%(Fig. 5).Based on total recovery (93.3%-106.3%) of both TiO 2 particles and Ti ions, about 99.1%-99.9% of added TiO 2 were detected as intact particle forms.The results demonstrate that most of food additive SiO 2 and TiO 2 are not degraded and present as particles, and these two additives have low solubility in food matrices.When detected ionic forms were compared, higher Si ions than Ti ions were found, probably indicating high solubility of the former than the latter.Indeed, it was reported that the solubility of SiO 2 was higher than TiO 2 [4,39,40], which is in good agreement with our results.Dissolution property of SiO 2 and TiO 2 was further evaluated by spiking known amount of pristine SiO 2 and TiO 2 in commercial food matrices, the same products where the CPEs were applied (Figs.4,5).As shown in Fig. 6A,B, the solubilities of SiO 2 were ranged from 0.1% to 5.5%, whereas TiO 2 had 0.1%-0.9%solubilities depending on the matrix type.Overall results indicate that the solubility of two additives was low in commercial complex food matrices, but SiO 2 has high solubility than TiO 2 , which is in good agreement with previous report [4,5].It is worth noting that no significant differences between solubility (Fig. 6) and ionic fate (Figs.4,5) were found in the same products (p > 0.05), indicating that the solubility critically affects the fates of two additives in foods.The results support reliable fate determination of SiO 2 and TiO 2 in commercial foods by CPE, and also suggest that major fates of SiO 2 and TiO 2 are particle forms in commercial processed products.Relatively high solubility of SiO 2 in powder 1, powder 2, and milk tea seems to be related to its matrix interactions because their pH values were neutral (pH ~7.0).Indeed, the interactions between food additive SiO 2 and food matrices were demonstrated, although the interactions seem to be not strong [3,6,13].Quantitative analysis revealed that saccharides, proteins, fatty acids, and minerals can be adsorbed on the surface of SiO 2 [13].Hence, the adsorption of food matrices on SiO 2 may increase its solubility in commercial food products.Meanwhile, the interaction between TiO 2 and food matrices is likely to be negligible based on its extremely low solubility (Fig. 6B) and minor ionic fates (Fig. 5).

Characterization of Separated SiO 2 and TiO 2 from Commercial Foods
Physico-chemical characterization of separated and recovered SiO 2 and TiO 2 from commercial food products by CPE was further carried out.The constituent particle sizes and size distributions were analyzed by TEM.Fig. 7A,B and Fig. 8A,B show that the constituent particle sizes of separated SiO 2 and TiO 2 were ranged from 12 ± 3 nm to 21 ± 4 nm and from 112 ± 30 nm to 132 ± 37 nm, respectively, without significant differences from pristine    The results indicate that pristine SiO 2 and TiO 2 have typical amorphous and anatase phases, respectively, and the same crystalline phases were observed in separated SiO 2 and TiO 2 particles from commercial foods (Figs.9,10).It should be noted that amorphous SiO 2 and anatase TiO 2 are authorized as food additives [3,23].All the results imply that food additive SiO 2 and TiO 2 are present as intact particle forms without changes in crystalline phase and constituent particle size.

Conclusions
Particle, aggregate, or ionic fates of food additive SiO 2 and TiO 2 in commercial processed foods were determined by optimizing CPE approaches in food matrices.The results reveal that most SiO 2 and TiO 2 in commercial foods were present as particle forms without significant differences in constituent particle size and size distribution compared with pristine food-grade each particle.Moreover, the crystalline phases of separated SiO 2 and TiO 2 from processed foods were not affected, showing amorphous and anatase crystalline phases for the former and the latter, respectively.The ionic fates of two additives by CPE were highly consistent with the solubility results, demonstrating maximum ~5.5% and ~0.9% ionized forms for SiO 2 and TiO 2 , respectively.Higher ionic fate and dissolution of SiO 2 than TiO 2 in foods were found, indicating that the former can more easily dissolve in food matrices.The interactions between NPs and matrices may affect their dissolution fates in food products.All the results suggest that most food additive SiO 2 and TiO 2 were present as intact particle forms, implying that particle fates of food additives should be considered for understanding and predicting their potential toxicity.

Fig. 1 .
Fig. 1.Particle or ionic fates and total recoveries of pristine food additive (A) SiO2 and (B) TiO2 by CPE.

Fig. 2 .
Fig. 2. Hydrodynamic diameters, fates, and solubility of SiO2 and TiO2 particles in food matrices.(A) Hydrodynamic diameters of SiO2 and TiO2 in food matrices before (pristine) and after CPE.No significant differences between before and after CPE were found (p > 0.05).Particle or ionic fates and total recoveries of pristine food additive SiO2 and TiO2 in (B) powdered sugar and (C) casein by CPE.(D) Dissolution properties of food additive SiO2 and TiO2 in food matrices.

Fig. 3 .
Fig. 3. Quantitative analysis of SiO2 and TiO2.Standard curves for (A) pristine food additive SiO2 and TiO2, (B) powdered sugar spiked with SiO2 or TiO2, and (C) casein spiked with SiO2 or TiO2 obtained by acid digestion in a microwave system, followed by inductively coupled plasma-atomic emission spectroscopy (ICP-AES) analysis.

Fig. 7 .
Fig. 7. Characterization of separated SiO2 from commercial foods.(A) Transmission electron microscopy (TEM) images, (B) size distribution obtained by the TEM images, and (C) energy dispersive X-ray spectroscopy (EDS) of separated SiO2 from commercial foods.

Fig. 8 .
Fig. 8. Characterization of separated TiO2 from commercial foods.(A) Transmission electron microscopy (TEM) images, (B) size distribution obtained by the TEM images, and (C) energy dispersive X-ray spectroscopy (EDS) of separated TiO2 from commercial foods.

Table 1 . Hydrodynamic diameters and zeta potentials of pristine food additive SiO2 and TiO2 under different CPE conditions.
± 1 B,b Different upper-case letters (A,B) indicate significant differences between different CPE dispersion conditions (p < 0.05).Different lower-case letters (a,b) indicate significant differences between before and after CPE steps (p < 0.05).Abbreviation: CPE, cloud poidnt extract; DDW, distilled and deionized water; HA, humic acid.resultsclearly indicate that SiO 2 and TiO 2 particles can be obtained as intact particle forms by CPE approaches.