**2.4. Other physicochemical properties of food-grade TiO2**

Titanium dioxide is insoluble in water, hydrochloric acid, dilute sulphuric acid and organic solvents. It dissolves slowly in hydrofluoric acid and hot concentrated sulphuric acid. It is almost insoluble in aqueous alkaline media (COMMISSION DIRECTIVE 2008/128/EC).

The physicochemical characteristics of particles, including morphology (spherical and cylindrical), size (smaller or larger than <100 nm), surface charge (negative, neutral or positive), structure (crystallinity), agglomeration (aggregates, agglomerates and primary particles) and surface composition, are assumed or demonstrated to play a role in nanoparticle uptake through the gut [24]. Therefore, the five criteria detailed before have to be completed by a deeper characterization of food-grade TiO2 , which unfortunately received much less attention.

#### *2.4.1. Content in nanoparticles and size distribution*

matter must represent less than 0.5%. For impurities soluble in 0.5 N hydrochloric acid, their amount must be lower than 1 mg/kg for arsenic, cadmium and mercury, lower than 2 mg/kg for antimony and lower than 10 mg/kg for lead. These specifications are very similar to those

In Europe, titanium dioxide is authorized *at quantum satis*, whereas it is used in the USA in the limit of 1% by weight of food. Although no maximum use level is specified for this additive in Europe, it shall be used in accordance with the good manufacturing practices (GMPs), that is, at a level not higher than is necessary to achieve the intended technical effect. This decision

and that neither significant absorption nor tissue storage following the ingestion of TiO<sup>2</sup>

possible. In its last report, the Panel of EFSA concluded that definitive and reliable data on the reproductive toxicity of E 171 are not yet available to enable the Panel to establish an accept-

products richest in titanium dioxide [2, 4]. They contain between 0.7 and 5.4 mg Ti/g of food. The next category is sweets with 0–2.5 mg Ti/g food, followed by pastry with 0–0.5 mg Ti/g food [2]. In the report of EFSA, including more numerous food categories and data provided

level, it is a little bit higher in processed nuts (3.8 mg Ti/g food) than in chewing gums (3.4 and 2.8 mg Ti/g food, depending on manufacturers), food supplements (2.8 mg Ti/g food) and

called this fact into question [23]. In the recent re-evaluation of titanium dioxide (E171) as food additive [7], the EFSA Panel estimated that the absorption of orally administered TiO<sup>2</sup> particles, including micro- and nano-sized (less than 3.2% by mass) fractions, was negligible, reaching at most 0.02–0.1% of the administered dose. They also indicated that no adverse effect resulting from the eventual accumulation of the absorbed particles was expected, based on the results of long-term studies which did not highlight any toxicity up to the highest administered dose. The lowest value found in the literature for the no-observed adverse effect

Titanium dioxide is insoluble in water, hydrochloric acid, dilute sulphuric acid and organic solvents. It dissolves slowly in hydrofluoric acid and hot concentrated sulphuric acid. It is almost insoluble in aqueous alkaline media (COMMISSION DIRECTIVE 2008/128/EC).

/kg bw/d.

**2.4. Other physicochemical properties of food-grade TiO2**

/g food which corresponds to 12 mg Ti/g food, a little bit above the maximum level

was considered as an inactive ingredient in human food,

is in decorations, coatings and fillings [7] with

/g, i.e., 9.6 mg Ti/g food). Considering the mean use

in commercial products indicates that chewing gums are the food

is considered as safe for use in food. Since this time, some authors

was

given by JECFA [14].

8 Application of Titanium Dioxide

was motivated by the fact that TiO2

by industry, the highest maximum level in TiO2

salads and savoury-based sandwich spreads (2.5 mg Ti/g food).

reported for chewing gums (16 mg TiO<sup>2</sup>

able daily intake (ADI) [7].

The quantification of TiO<sup>2</sup>

*2.3.5. Innocuousness of TiO2*

Since the early 1960s, TiO<sup>2</sup>

levels (NOAEL) was 2250 mg TiO<sup>2</sup>

20 mg TiO2

*2.3.4. Amounts*

Considering the food use of TiO2 as whitening agent, the size distribution is expected to be centred on a mean pigment size of 250 nm to obtain an optimal effect [25]. However, the mean size of food-grade TiO2 is actually rather comprised between 106 and 145 nm and the size distribution spans between 30 and 300 nm [4, 18, 19, 26, 27] or 60 and 300 nm [2]. For example, several size distribution spans and mean sizes are reported in **Figure 2**. Overall, they span between 30 and 300 nm. In these batches, the fraction of nanoparticles (<100 nm) ranged from 17 to 36%. In the whole set of samples investigated in the literature, the nanoparticle size distribution expressed in number was always smaller than 50%. In chewing gums, this fraction mounts to 43.7% [1].

To determine the exposure scenario, the equivalent mass of NPs is more interesting. According to several studies, the mass (wt%) of nanoparticles present in E171 ranges between 0.31 and 12.5% [7, 10, 11, 18]. This explains some discrepancies in the different exposure to TiO2 nanoparticles in the literature and, for example, the factor of 10 in the estimate of NP


**Figure 2.** Size distribution (dashed rectangles), mean size (black dots) and percentage of nanoparticles in number (%) of food-grade TiO2 particles characterized by (E) Dudefoi et al. [18] and (S) Yang et al. [19]. The mean sizes of the distribution vary between 106 and 145 nm.

consumptions between the study by Rompelberg et al. [11] who considered 0.31% of NPs and the evaluation of EFSA [7] who used a weight ratio of 3.2%.

#### *2.4.2. Specific surface area*

The specific surface area (SSA) of a material is defined as the total surface area of the material per unit of mass. It is reversely proportional to the size of the particles: the smaller the size of a material, the higher its specific surface area and its reactivity with the environment. The SSA is usually determined from the volumetric adsorption isotherms at 77 K of nitrogen gas followed by the Brunauer-Emmett-Teller (BET) adsorption treatment (the so-called N<sup>2</sup> -BET isotherm) assuming a multilayer of adsorbates. The specific surface area of food-grade TiO<sup>2</sup> ranges between 8.6 and 10.7 m<sup>2</sup> /g [18, 20] with an average of 9.3 m<sup>2</sup> /g. These values are quite low in comparison to anti-caking agents, for example, which are around 200 m2 /g. This hints that TiO2 offers a low contact surface with its environment.

#### *2.4.3. Surface chemical composition*

The surfaces of food-grade TiO2 were found to be mainly covered by hydroxyl groups [18], phosphate groups [18, 19] and potassium ions [18]. Some phosphate groups may not be tightly bound to the surface and be released after washing [19]. In a few cases, TiO<sup>2</sup> was covered by silica [18] and alumina [19], thus modifying the surface chemistry.

#### *2.4.4. Surface potential*

As mentioned previously, surface hydroxyl groups, which behave as Brønsted acid or base sites, confer a charge to the particle surface. When TiO<sup>2</sup> particles are dispersed in an aqueous medium, this charge is mainly determined by two phenomena: protonation/deprotonation of surface hydroxyls controlled by pH and adsorption of electrolyte ions [28]. An electrostatic potential, exponentially decaying away from the surface, is associated to the overall charge distribution in the interfacial region. The experimental determination of this potential, called zeta potential, is generally performed by electrophoretic mobility measurements. All models converting electrophoretic mobility into zeta potential consider ideal spherical particles, which is a delicate assumption in the case of TiO2 due to the formation of agglomerates with non-spherical particles (subsequent section). An improved model exists to convert electrophoretic mobility measurements to zeta potential values taking into account the effect of the agglomerate size and surface conductance of TiO2 [29]. Zeta potential values depend not only on the parameters controlling the surface charges, namely, the nature of the medium where TiO2 particles are dispersed (pH, ionic strength and adsorbed species [20]) but also on the primary particle size [29, 30] and the crystallographic face [31]. The point where the zeta potential is zero defines the isoelectric point (IEP).

The isoelectric point of food-grade TiO2 samples measured by electrophoretic mobility measurements was found between 3 and 4 for most samples (**Table 1**), far below the classical value for anatase. Such a difference is interpreted by the presence of phosphate groups on the surface of TiO2 particles [18, 19] or by silica coating [18], which decrease the isoelectric point


**Table 1.** Isoelectric point (IEP) and zeta potential at pH 7 of various food-grade TiO<sup>2</sup> (E171) dispersed in water, without any protein.

towards lower pH values. It is interesting to note that the isoelectric point of a food-grade sample measured through electroacoustic measurements gave a value of 5.1 [20], close to the classical data for anatase. For all these samples, the zeta potential of their suspensions varies between −35 and −45 mV at a physiological pH value. Faust et al. compared the zeta potential of a food-grade TiO2 and an extract of chewing gum, and observed that the gum extract presented a largely more negative potential (−45 mV at pH 7) than food-grade TiO<sup>2</sup> (−20 mV at pH 7), which may be due to coating of TiO<sup>2</sup> in chewing-gum formulation [26].

#### *2.4.5. Agglomeration*


/g. This hints

was covered by

particles are dispersed in an aqueous

due to the formation of agglomerates with

[29]. Zeta potential values depend not only

samples measured by electrophoretic mobility mea-

/g. These values are quite

consumptions between the study by Rompelberg et al. [11] who considered 0.31% of NPs and

The specific surface area (SSA) of a material is defined as the total surface area of the material per unit of mass. It is reversely proportional to the size of the particles: the smaller the size of a material, the higher its specific surface area and its reactivity with the environment. The SSA is usually determined from the volumetric adsorption isotherms at 77 K of nitrogen gas followed by the Brunauer-Emmett-Teller (BET) adsorption treatment (the so-called N<sup>2</sup>

isotherm) assuming a multilayer of adsorbates. The specific surface area of food-grade TiO<sup>2</sup>

phosphate groups [18, 19] and potassium ions [18]. Some phosphate groups may not be tightly

As mentioned previously, surface hydroxyl groups, which behave as Brønsted acid or base

medium, this charge is mainly determined by two phenomena: protonation/deprotonation of surface hydroxyls controlled by pH and adsorption of electrolyte ions [28]. An electrostatic potential, exponentially decaying away from the surface, is associated to the overall charge distribution in the interfacial region. The experimental determination of this potential, called zeta potential, is generally performed by electrophoretic mobility measurements. All models converting electrophoretic mobility into zeta potential consider ideal spherical particles,

non-spherical particles (subsequent section). An improved model exists to convert electrophoretic mobility measurements to zeta potential values taking into account the effect of the

on the parameters controlling the surface charges, namely, the nature of the medium where

surements was found between 3 and 4 for most samples (**Table 1**), far below the classical value for anatase. Such a difference is interpreted by the presence of phosphate groups on the

particles [18, 19] or by silica coating [18], which decrease the isoelectric point

 particles are dispersed (pH, ionic strength and adsorbed species [20]) but also on the primary particle size [29, 30] and the crystallographic face [31]. The point where the zeta

low in comparison to anti-caking agents, for example, which are around 200 m2

bound to the surface and be released after washing [19]. In a few cases, TiO<sup>2</sup>

offers a low contact surface with its environment.

silica [18] and alumina [19], thus modifying the surface chemistry.

sites, confer a charge to the particle surface. When TiO<sup>2</sup>

which is a delicate assumption in the case of TiO2

agglomerate size and surface conductance of TiO2

potential is zero defines the isoelectric point (IEP).

The isoelectric point of food-grade TiO2

/g [18, 20] with an average of 9.3 m<sup>2</sup>

were found to be mainly covered by hydroxyl groups [18],

the evaluation of EFSA [7] who used a weight ratio of 3.2%.

*2.4.2. Specific surface area*

10 Application of Titanium Dioxide

ranges between 8.6 and 10.7 m<sup>2</sup>

*2.4.3. Surface chemical composition*

The surfaces of food-grade TiO2

*2.4.4. Surface potential*

that TiO2

TiO2

surface of TiO2

The dispersion state of particles in aqueous solution is governed by the surface chemistry of the oxide and depends on the composition of the dispersion medium (pH, ionic strength, nature of electrolyte and presence of proteins). Traditionally, zeta potential measurements are used to assess the stability of colloidal dispersions: the higher the zeta potential absolute value, the more stable the dispersion. Around the IEP or when ionic strength is high in solution, the system is unstable and agglomeration of particles occurs, leading to settling of the suspension. It is thus important to consider agglomeration in the experimental medium, as this may alter the size of the particles which will be 'seen' later by the organism after ingestion.

In usual conditions of pH and ionic strength, TiO<sup>2</sup> particles tend to form large-sized agglomerates (particles relatively loosely bound) which settle after a few hours, partially due to the large density of TiO2 (3.9 g/cm3 for anatase as powder). For neutral pH values (around 6–7) and in the absence of any salt, E171 particles present agglomerates with a diameter of 200–400 nm, in agreement with the largely negative-measured zeta potential. When pH becomes closer to IEP, the measured diameter is larger than 1 µm, which is the sign of agglomeration due to low electrostatic repulsions [18].

Once particles are agglomerated or aggregated, they do not fragment easily and are difficult to disperse as primary particles. Ultrasound sonication can be used to break the agglomerates prior to zeta potential and size measurements, providing ultrasounds do not alter the surface chemistry of the material [32]. The hydrodynamic diameter of E171 particles dispersed in ultrapure water (pH not mentioned) and bath sonicated (for 5–30 min) comprises between 120 and 400 nm [19, 26]. Another possibility to stabilize the suspension and avoid agglomeration of particles consists in adding a dispersant which is able to cover the particles and create steric hindrance between them [33]. Bovine serum albumin (BSA) was typically used to stabilize E171 TiO2 particles, in combination with ultrasound sonication (30 min), leading to a mean hydrodynamic diameter of 150 nm [4]. In solutions added with salts (NaCl and NaHCO3 ), E171 particles dispersed by sonication presented a moderate stability, with a particle size of agglomerates remaining between 360 and 390 nm for at least 2 h. The same experiment conducted with P25 sample showed rapid and extensive aggregation of the particles [4].
