EXECUTIVE SUMMARY SPONSORED CONTENT

Measuring TiO2 in Foods to Meet New

Regulatory Requirements
Advanced testing for titanium dioxide in various
food matrices.

INTRODUCTION Daniel Clayton-Cuch

Application Scientist
Titanium dioxide (TiO2) is an inorganic compound typically used in food products Agilent Technologies
as a pigment labelled E171. TiO2 provides a white color and reflects light across
most of the visible spectrum. When used as a food additive, TiO2 particles have
been shown to exhibit a broad size distribution within the final food products. A
Sponsored by
proportion of these particles are nanoparticles (NPs), which are of the greatest
concern to human health (1). TiO2 may derive from different sources within the
food supply chain, however, the direct addition of TiO2 as a food additive is the
focus of this work.

TiO2 UNDER SCRUTINY

Food grade TiO2 was first approved by the FDA in 1966, and then by the EU in 1969.
However, in 2006, the International Agency for Research on Cancer classified
TiO2 NPs as potentially carcinogenic to humans (1). A series of studies followed,
first showing that TiO2 contains NPs up to 36%, further showing it may cause
precancerous lesions and immune problems in animal studies (1). By 2021, the
European Food Safety Administration concluded that TiO2 could no longer be
considered safe when used as a food additive which led to its eventual ban within
the EU starting in 2022.

Following the complete ban of TiO2 within the EU, the industry has been left
with many unaddressed concerns, one of which includes the analysis of food
products for TiO2. Since TiO2 is present in a range of food matrices, from
beverages and chewing gum to dairy, grains, candy, sauces, and more, any
standardized sample digestion methods must be able to digest a range of
different food materials. Currently, no standardized analytical methods exist for
detecting TiO2, nor does the industry have a clear maximum detectable limit of
TiO2 for food samples.
EXECUTIVE SUMMARY SPONSORED CONTENT

THE AGILENT SOLUTION emptied into 50 mL centrifuge tubes and the vessels were
Agilent developed a preliminary method for the quick and rinsed multiple times with deionized water to ensure complete
simple chemical analysis of TiO2 in a broad range of food sample recovery. This left the samples at a final sulfuric acid
products using the Agilent 5800 inductively coupled plasma concentration of 18%.
optical emission spectrometer (ICP-OES). The instrumentation
setup included: The Agilent ICP Expert software features a function called
IntelliQuant. IntelliQuant provides rapid semi-quantitative
• Agilent 5800 ICP-OES instrument analysis, as well as primary data regarding the elements
• ICP Expert instrument control software, version 7.6 present in the sample. This sample screening produces a
• Agilent SPS 4 Autosampler color-coded periodic table heat map, showing the relative
• 1.8 mm internal diameter one-piece torch (supplied with concentrations of each element. Higher concentrations are
instrument) demonstrated by orange and red, while lower concentrations
• SeaSpray nebulizer (also supplied with instrument) are represented as yellow, as shown in FIGURE 1. Knowing
• Double-pass cyclonic spray chamber (also supplied with this information early in the process helps analysts design a
instrument. calibration curve which ultimately saves time and resources.

In the past, many methods used hydrofluoric acid (HF) to In addition to IntelliQuant Screening, the ICP Expert software
digest samples when analyzing for TiO2. However, HF is offers Outlier Conditional Formatting (OCF), which alerts
hazardous to work with, and requires extra steps as it is analysts to problematic results in real time so corrective action
aggressive to glass components. HF should generally be can be taken. Results are based on custom limits, and will be
avoided if alternatives exist. In response, Agilent developed a flagged if:
method using only concentrated sulfuric acid and hydrogen
peroxide. This is safer and equally efficacious for TiO2 recovery, • A sample is above the defined concentration range
allowing more labs to adopt the methodology, as many are • Samples have high relative standard deviation
not equipped to handle HF. • Samples are below the method detection limit
• An internal standard exceeds the user-defined
All the samples in this work were digested using the CEM acceptable ratio
Mars6 digestion system with iPrep vessels, which can handle • The elemental concentration %RSD is greater than the
strong acid and high temperatures over a long duration. The user-specified pass value
digestion parameters were as follows:
This feature is useful for high throughput labs that want to
• Maximum Power: 1800 watts automate their analysis and avoid sifting through datapoints
• Temperature: 250 °C from thousands of samples.
• Ramp time: 25 minutes
• Hold time: 60 minutes The ICP Expert software also includes an automated feature
called Fitted Background Correction (FBC), which is an
The sample digestion mixture consisted of 0.5 g sample (or algorithm correcting for the variable background structures
water), 9 mL sulfuric acid, and 1 mL 30% hydrogen peroxide. arising from different sample matrices. FBC automatically
Following the digestion, the contents of the vessels were models the background signal under the analyte peak,
EXECUTIVE SUMMARY SPONSORED CONTENT

providing accurate correction of both simple and complex were then matrix-matched using sulfuric acid and further
background structures—no method development or manual diluted with Milli-Q water to form a final matrix of 18%
placement of background peak markers is required. A sulfuric acid. The calibration range used for each element
powerful background correction algorithm known as FACT is shown in FIGURE 2 and linear calibration curves were
provides accurate data when one element has strong achieved for all elements, as shown by the correlation
interference from another at a very similar wavelength. FACT coefficients being greater than 0.9996. The graph in
allows analysts to select a specific interferent and analyte, FIGURE 2 shows an example titanium calibration curve
removing the interferent intensity and ensuring accurate data. with a 1.0000 correlation coefficient, and a relative
standard error below 5 percent. This curve ranges from 5
Another popular software feature used in the TiO2 analysis is ppb to 50 ppm, covering a wide range of concentrations in
the Intelligent Rinse function. This function allows rinse times which samples can fall.
between samples to be optimized automatically to suit the
analyst’s priorities—for example, productivity or accuracy. Two certified reference materials (CRMs) were used to verify
This improves throughput and minimizes sample-to-sample the analytical method: oyster tissue (primarily composed
contamination. of proteins, up to 42%) and seaweed (primarily composed
of carbohydrates, up to 68%). Each CRM was prepared
RESULTS in triplicate and each digest was analyzed in triplicate.
Calibration standards were prepared using Agilent’s single Recoveries for all reported analytes fell within ± 10% of the
element standard calibration solutions. The standards certified value. The excellent recoveries demonstrated in
EXECUTIVE SUMMARY SPONSORED CONTENT

• Calibration standards matrix matched to samples with H2SO4

• All linear calibration curve correlation coefficients >0.9995

these tests confirm the suitability of this HF-free digestion Next, a variety of food samples containing TiO2 were
method for the preparation of different food matrices and purchased from a local Australian supermarket to evaluate
confirmed the accuracy of the 5800 ICP-OES method. the capacity of these digestion procedures and analytical
methods to measure TiO2 in food matrices. The samples
As there are currently no detection limit regulations varied in levels of protein, fat, carbohydrates, being: dairy
regarding the allowable concentration of TiO2 in food products, candy-coated chocolate, decorative gold dust
products in the EU, analysts used the GB5009.268-2016 spray, salad cream (similar to mayonnaise), and a protein
detection and quantification limits for TiO2, which were 0.2 bar. The quantitative data for these five samples were
and 0.5 mg/kg, respectively. The limit of detection (LOD) obtained using the Agilent 5800 ICP-OES and revealed TiO2
and the limit of quantification (LOQ) shown in FIGURE 3 was present in all samples.
were calculated based on the analysis of 10 method blank
solutions, measured over six separate occasions, on two To assess the stability of the 5800 ICP-OES, 300
separate instruments. The values were below the detection sample solutions were measured over 10 hours without
limit specified in the GB method. To test for non-spectral recalibration. As shown in FIGURE 4, the stability of the
interferences or matrix effects, TiO2 in rutile form at 99.9% results was within ±10% over the course of the run, and
purity was spiked at different concentrations into a method the %RSD was below 1.4% for all elements. These results
blank and into the seaweed CRM. Results from the spiked demonstrate the robustness, stability, and precision of the
recovery test show that both the digestion and analytical 5800 ICP-OES for the routine quantitative analysis of foods
method are suitable for the complete recovery of TiO2. and food additives such as TiO 2.
EXECUTIVE SUMMARY SPONSORED CONTENT

MEASURING NANOPARTICLES discrete pieces of material with one or more dimensions in

The largest concern surrounding food grade TiO2 is the the size of 1 to 100 nm. They have been an area of significant
presence of NPs. When attempting to determine the size interest in recent years, since several studies have shown that
distribution of NPs, a triple quadrupole ICP-MS, such as the the uptake of TiO2 in the bloodstream can lead to accumulation
Agilent 8900 ICP-QQQ is the most suitable choice. NPs are in the organs and various negative implications (2, 3, 4, 5).
EXECUTIVE SUMMARY SPONSORED CONTENT

There are many ways to measure particle size or particle size most sensitivity. However, an isobaric interference from
distribution. However, single particle ICP-MS analysis has calcium can be observed in many food samples at high
significant advantages over other options, including: concentrations; there are also polyatomic interferences
on titanium 48 from other elements, including sulfur and
• Easy implementation phosphorus. For this reason, many analysts avoid using
• Speed this isotope on a classic ICP-MS instrument and choose
• High sensitivity (capable of measuring down to ppt) other isotopes for testing.
• Element-specificity
• Information on size distribution, particle number, and Triple-quadrupole technology (ICP-QQQ or ICP-MS/MS) can
mass concentration be very useful in this scenario. The first quadrupole is set to a
mass-to-charge ratio of 48, and the second is set to a mass-
A typical signal is shown in FIGURE 5, where the spikes are to-charge ratio of 64. Placed in the collision reaction cell is
related to the particle concentration, and the height of the a reaction gas consisting of hydrogen and oxygen. As the
spikes is related to particle mass. If the particle shape and titanium 48 isotope passes through the first quadrupole into
density are known, analysts can also use this data to convert the collision cell, it forms TiO2. This has a mass-to-charge
particle mass into particle size and, ultimately, into particle (m/z) ratio of 64. Since calcium 48, a common interferent,
mass concentration. reacts with hydrogen to form calcium hydroxide—which
doesn’t have a m/z ratio of 64—it will not pass into the
Titanium has five isotopes with different abundances—the second quadrupole. This is one technique in which the
most abundant is titanium 48 with 73.72% abundance. method can eliminate unwanted interferences to get a
This isotope is ideal for testing since it provides the signal from only the high abundance titanium 48 isotope.
EXECUTIVE SUMMARY SPONSORED CONTENT

CONCLUSION Chaverri J, Hernández-Pando R, Rodríguez-Sosa M, León-Cabrera

Agilent offers a range of instrumentation that can be used S, Terrazas LI, van Loveren H, Chirino YI. Food-grade titanium
to quantify and characterize TiO2 in food products. The dioxide exposure exacerbates tumor formation in colitis associated
Agilent ICP-OES 5800 is an ideal instrument for the food cancer model. Food Chem Toxicol. 2016 Jul; 93:20-31. doi: 10.1016/j.
manufacturing industry, specifically to identify and quantify fct.2016.04.014.
the presence of TiO2 in a range of different food products, 3. Bettini S, Boutet-Robinet E, Cartier C, Coméra C, Gaultier E, Dupuy
without the use of HF during sample preparation. J, Naud N, Taché S, Grysan P, Reguer S, Thieriet N, Réfrégiers M,
Thiaudière D, Cravedi JP, Carrière M, Audinot JN, Pierre FH, Guzylack-
Further, particle sizing of nanoparticles of TiO2 in different Piriou L, Houdeau E. Food-grade TiO2 impairs intestinal and systemic
food samples is achievable on the Agilent ICP-QQQ 8900, immune homeostasis, initiates preneoplastic lesions and promotes
which provides superior sensitivity, particularly for NPs. These aberrant crypt development in the rat colon. Sci Rep. 2017 Jan
technologies will become critical for food companies as public 20;7:40373. doi: 10.1038/srep40373.
scrutiny and regulations alike impact the use of TiO2 in foods. 4. Proquin H, Rodríguez-Ibarra C, Moonen C, Urrutia Ortega IM, Briedé JJ,
de Kok TM, van Loveren H, Irasema Chirino Y. Titanium dioxide food
REFERENCES additive (E171) induces ROS formation and genotoxicity: contribution
1. Younes, M., Aquilina, G., Castle, L., Engel, K.H., Fowler, P. Frutos of micro and nano-sized fractions. Mutagenesis. 2018 Sep 17;33(3):267-
Fernandez, M.J, et al. (2021). Safety Assessment of Titanium 268. doi: 10.1093/mutage/gey011.
Dioxide (E171) as a food additive. EFSA Journal, 19(5). Doi: 10.2903/j. 5. Pele, LC, Thoree V, Bruggraber SF, Koller D, Thompson RP, Lomer
efsa.2021.6585 MC, Powell JJ. Pharmaceutical/food grade titanium dioxide particles
2. Urrutia-Ortega IM, Garduño-Balderas LG, Delgado-Buenrostro NL, are absorbed into the bloodstream of human volunteers. Part Fibre
Freyre-Fonseca V, Flores-Flores JO, González-Robles A, Pedraza- Toxicol. 2015 Sep 2.12.26. doi: 10.1186/s12989-015-0101-9.