A Novel Method for Titanium Dioxide Quantification in Cosmetic Products via Borate Fusion by Flame Atomic Absorption Spectroscopy

Abstract

A new method for extraction of titanium dioxide (TiO2) from cosmetic matrices using borate salts for its quantification by flame atomic absorption spectroscopy (FAAS) was developed and validated. Following International Commission for Harmonization (ICH) and the United States Pharmacopeial Convention (USP) guidelines, the parameters of the method considered in this study were specificity, linearity, sensitivity, precision, and accuracy. In addition, critical factors of the method were assessed using a Youden–Steiner model. The method was able to differentiate the titanium signal from matrix and background signals, for which it is considered specific. The method is also linear for all cosmetic matrices and the raw material in the range 20–80 ppm with LOD and LOQ around 2 ppm and 7 ppm, respectively. Repeatability and intermediate precision were below 5.0%RSD, and Global Reproducibility was below 8.0% RSD. A digestion step free of HF or strong oxidizers makes this method a safer and easily transferable alternative to classical methods for quality control. It is thus a convenient tool for routine analysis of cosmetic products that need to comply with current regulations to ensure the safety of consumers and to guarantee batch-to-batch quality.

1. Introduction

Skincare and makeup products are highly innovative and profitable segments of the cosmetic industry as they represent, together, near 60% of the total market size, which was estimated at around € 200B for 2019 [1]. Current consumer trends show an increased demand for products with a wide variety of functions but, among them, protection against radiation from natural as well as human-made sources stands out as one of the most desirable features. Thus, sun protection claims are not exclusive to sunscreen lotions but are also demanded in makeup compositions such as lipsticks and foundations. Furthermore, sun-blocking properties are generally achieved by the inclusion of ultraviolet (UV) filters in cosmetic formulas [2, 3].

UV filters are incorporated into cosmetic formulas to build up a protective screen that blocks most photons from penetrating the skin. UV radiation negatively affects skin cells by promoting DNA damage mediated by reactive oxygen species (ROS), by chemical modification that includes pyrimidine dimer formation, or by inducing replication errors, which can lead to erythema or ultimately increase the risk of melanoma [4, 5]. As UV radiation comprises three main subclasses of radiation based on their wavelength (UVA for the lowest energy photons ranging from 320 to 420 nm; UVB, in the middle range from 280 to 320 nm; and UVC for high energy photos with wavelengths from 100 to 280 nm), different chemical compounds are used to cover those ranges as few materials are completely effective to cover the entire UV region. Organic molecules such as benzophenone and avobenzone confer protection from UVA while others such as PABA derivatives and octocrylene protect from UVB radiation. UVC is generally not considered when designing cosmetic formulas because most of this type of radiation is lost at the upper layers of the atmosphere, where it is involved in ozone synthesis. Inorganic compounds are also used as UV filters as some of them possess broad spectrum blocking properties, which include kaolin, talc, calcium carbonate (CaCO₃), zinc oxide (ZnO), and titanium dioxide (TiO₂) [6, 7].

Sustainability initiatives focused on preserving the environment have spotlighted the raw materials used to produce many of the goods we consume regularly. The cosmetic industry is no stranger to such scrutiny, and specifically, recent studies have dealt with the potential threats organic UV filters pose to water sources, coral reefs, and the food chain [8]. Future directions point to an increase in the relevance of inorganic filters as an alternative to organic compounds to act as physical blockers of UV radiation in cosmetic formulas, as they are currently considered less harmful to the environment, and they pose, in general, a low risk to human health [9]. The US Food and Drug Administration (FDA) approves the use of TiO₂ and ZnO as photoprotective agents up to a limit of 25% w/w. They are commonly combined in formulas to give wide-spectrum protection, but a major drawback of this is the undesired white residue they leave when applied on the skin. This can be solved by working with engineered nano-sized ingredients for which this effect is absent. Nevertheless, some environmental and human health concerns are associated with nano-sized TiO₂ and ZnO, but more data is needed to fully assess their effects [10].

Quantification of TiO₂ is important not only to ensure regulatory compliance but also to control the batch-to-batch quality of cosmetic products containing it to claim UV-blocking properties or even use it as a pigment. Quantification is particularly challenging because of the complexity of cosmetic formulas [11], and, in contrast with studies concerned with the analysis of organic filters, not many literature works are devoted to the analysis of TiO₂ in such matrices [12]. TiO₂ quantification can be achieved via redox titration [13], portable X-ray fluorescence (pXRF) [14, 15], Raman spectroscopy [16], laser-induced breakdown spectroscopy [17], ICP/AAS approaches [18, 19, 20, 21, 22, 23] and nanomaterial-based approaches [24]. Remarkably, most of these methods were developed for emulsion matrices only, which leaves a gap for the analysis of makeup matrices, which might involve additional considerations such as the presence of inorganic compounds commonly used as pigments.

When analyzed by methods such as AAS or ICP, samples that contain TiO₂ are processed with aggressive treatments that normally involve a digestion step with hydrofluoric acid (HF) or mixtures of HF with other reactive species such as concentrated nitric acid (HNO₃) and hydrogen peroxide H₂O₂ [25]. Other protocols use sulfuric acid (H₂SO₄) [26], which provides high recovery yields of TiO₂, but may lead to the formation of sulfur oxide species (S∙O) with a similar m/z ratio as the main Ti isotope, which is inconvenient in some applications [27]. Different approaches include the formation of emulsion slurries before injection in the ICP [19, 20, 22]. For this study, we wanted to implement a borate fusion method, which has been used in treating geological and environmental samples, where it has shown better recovery compared to HF digestion methods [28, 29, 30] but, to the best of our knowledge, has not been implemented and validated for the analysis of cosmetic samples. In this approach, the mixture is pre-oxidized to remove organic interferences and then dissolved at the melting temperature of the borate salt in a platinum crucible (flux). Different borate salts are available to dissolve inorganic oxides depending on their acidity, and mixtures are normally used to work with samples that contain more than one type of oxide [31]. This is especially important for samples such as makeup, where inorganic oxides such as Fe₂O₃ and ZnO are used as pigments or UV filters.

For the quality assessment of cosmetic products, specifically in compliance with current regulations that set a maximum limit for TiO₂ of 25% w/w, we are mainly concerned with the development of methods that facilitate the routine analysis of products that have been designed to contain important amounts of inorganic filters. This is in clear contrast with methods aimed at the analysis of traces, where LOD and LOQ parameters are expected to be as low as the methods permit it. In this study, we sought to develop and validate a cost-effective method for TiO₂ extraction from emulsion, foundation, and lipstick matrices, and its quantification using FAAS, which is an affordable and widely implemented analytical technique for most quality control laboratories in the cosmetic industry. We assessed the specificity, linearity, precision, accuracy, and robustness of the method following ICH and USP guidelines [32, 33].

2. Materials and methods

2.1 Reagents and materials

Titanium 1000 mg/mL standard solution (Certipur^®), reagent grade potassium chloride, reagent grade anhydrous aluminum chloride, 66% di-lithium tetraborate/34% lithium metaborate (SPECTROMELT ® A 12), and analytical grade 65% HNO₃ (EMSURE ®) were purchased from Merck (Darmstadt, Germany). TiO₂ raw material (UV-Balance Powder 100™) was obtained from KOBO Industries (New Jersey, USA). Lipsticks, liquid foundations, and cream emulsions were obtained from Belcorp (Lima, Perú). Deionized water (resistivity ≥18 MΩ*cm) was prepared with a Barnstead™ Easy Pure™ II water purification system.

2.2 FAAS conditions

Measurements were performed using a Thermo Scientific™ iCE™ 3000 atomic absorption spectrometer with an air-acetylene flame. Acetylene flux was 4.6 L min⁻¹. Ti absorbance was measured using a Hollow Cathode Lamp at a wavelength of 365.4 nm, with a bandpass of 0.5 nm. A deuterium lamp was used for background correction.

2.3 Sample preparation

Cosmetic matrix (lipstick, foundation, and emulsion) effects were studied through the analysis of placebos (no added TiO₂) specially prepared for this study. Calibration curves were completed for each matrix by combining placebo stock solutions and an appropriate amount of titanium standard solution. A defined amount of placebo (2000 mg of lipstick, 1400 mg of emulsion, 1200 mg of foundation) was weighted in a platinum crucible and mixed with 0.6 g of Spectromelt ® A 12. The crucibles were heated at 800°C for 1 h in a muffle furnace. Later, they were further heated at 1100°C for 15 min. After slow cooling to room temperature, the crucibles were transferred to 100-mL beakers, and a volume of 70 mL of 1 M HNO₃ was added. The solution was stirred with a magnetic bar at 80°C until complete dissolution of the solid residue. The cool solution was transferred to a 100-mL volumetric flask and completed to volume with 1 M HNO₃ to obtain the placebo stock solutions. Calibration standards for each matrix were then prepared by mixing 5.0 ml of placebo stock, 2.5 mL of 10% KCl, and 1.0 mL of 5% AlCl₃ with the appropriate quantity of titanium standard solution. The mixture was diluted to volume with 0.6% Spectromelt ® A 12 in 1 M HNO₃. For system and raw material calibration curves, deionized water was used instead of placebo stocks. Cosmetic and raw material samples were prepared by placing an accurately weighed amount, between 10 and 500 mg, in a platinum crucible. The contents were extracted following the protocol already described. The resultant mixture was filtered with a 0.45 μm nylon syringe filter. Next, 1 ml of the filtered solution was transferred to a 100-mL volumetric flask, mixed with 5 mL 10% KCl, 2 mL 5% AlCl₃, and diluted to volume with 1 M HNO₃ for FAAS analysis.

2.4 Method validation

The method was validated following ICH and USP guidelines [32, 33]. Specificity, linearity, precision, accuracy, and robustness assays were conducted as described below.

2.4.1 Specificity

Placebo blanks were prepared by following the entire extraction method from placebos to yield placebo stocks without further addition of Ti standard. Each blank was prepared in duplicate. To identify matrix interferences and assess the ability of the method to quantify Ti in their presence, the method specificity was determined by comparing the signal of placebo blanks with the background and the Ti signal.

2.4.2 Linearity and sensitivity

System linearity was evaluated by preparing three calibration plots of absorbance versus standard concentration (20, 35, 50, 65, and 80 ppm). The same procedure was repeated for each cosmetic matrix using placebo standards. Data regression and correlation significance were estimated using analysis of variance (ANOVA) and a Student’s t-test, respectively. To evaluate sensitivity, the limit of detection (LOD) and limit of quantification (LOQ) were calculated using data from the regression analysis. LOD was calculated using the formula 3 * (σ/S), where σ is the intercept of the regression equation and S is its slope. Similarly, LOQ was computed using 10 * (σ/S). The LOQ was validated by measuring six replicates of stock dilutions or placebo stock dilutions to the calculated concentration and assessing their accuracy and precision. %RSD limit was set at 5.0%.

2.4.3 Precision

The method precision was determined considering repeatability, intermediate precision, and reproducibility. Repeatability was assessed by measuring six replicates in low, middle, and high concentration levels (each level in duplicate) for the system and each cosmetic matrix. System levels were prepared from the stock solution by dilution to a final concentration of 20, 50, and 80 ppm. Cosmetic matrix levels were prepared from their corresponding placebo stock by adding an appropriate amount of titanium standard and diluting to a final concentration of 3.2, 4.0, and 4.8% w/w for lipstick, 4.8, 6.0 and 7.2% w/w for emulsion and 5.6, 7.0 and 8.4% w/w for foundation. For raw material repeatability, only 1 level, the nominal concentration declared in its certificate of analysis, was considered. Six replicates of this level were prepared, as previously described, and measured. %RSD maximum limit was established at 5.0% for each concentration level. Intermediate precision was estimated at the nominal concentration from three cosmetic matrix samples and raw material samples. The sample duplicates were measured in two different days and by two different analysts. %RSD maximum intra-day and inter-day limits were set at 5.0%, whereas the global %RSD maximum limit was set at 8.0%, following USP validation criteria [33]. Reproducibility was evaluated by measuring 3 sample preparations of each cosmetic matrix and raw material in two different laboratories.

2.4.4 Accuracy

Samples of each cosmetic matrix were prepared by supplementing placebo stocks with titanium standard in three concentration levels (low, middle, high) and extracted following the procedure previously described. Raw material samples were analyzed assuming the concentration of TiO₂ reported in the certificate of analysis as the nominal concentration. Three samples for each concentration level were measured in duplicate. The accuracy was evaluated based on the recovery rate for each matrix.

2.4.5 Robustness

Robustness was assessed using a five-variable Youden-Steiner model [34]. Cosmetic matrix and raw material samples were studied under method standard conditions and alternative conditions in eight experiments, as described in Table 1. Samples were measured in duplicate for each set of conditions. Critical variables were defined as those with absolute differences greater than σ * $\sqrt{2}$, where σ represents the standard deviation calculated from the repeatability assays.

Parameter	Experiment numbera
	1	2	3	4	5	6	7	8
Spectromelt ® A 12 (g)	0.7	0.5	0.5	0.7	0.7	0.5	0.7	0.5
Muffle furnace initial temperature (°C)	820	820	780	780	820	780	780	820
Muffle furnace initial step (min)	40	40	40	80	80	80	40	80
Muffle furnace final step (min)	10	20	20	20	10	10	10	20
Final concentration (ppm)	30	30	70	70	70	30	70	30

Table 1.

Difference values for the variables studied under the Youden–Steiner model.

^a

Bolded values are the standard conditions of the method.

3. Results and discussion

3.1 Method development

Cosmetic products such as lipsticks, foundations, and emulsions contain different ingredients of both organic and inorganic nature [35]. For TiO₂ extraction from such matrices, the use of Spectromelt ® A 12, a 66% di-lithium tetraborate (LiT)/34% lithium metaborate (LiM) flux, common to geological and environmental sample treatment [36] but not so common outside this field [37], is proposed as an alternative to the use of hazardous acids such as HF, or mixtures of strong acids and oxidizing reagents under harsh conditions. The ratio LiM/LiT is chosen considering that basic oxides such as ZnO, present in cosmetic formulas to boost UV protection, are more soluble in LiT, whereas acidic oxides such as TiO₂ itself are soluble in LiM [31]. The first step of calcination at 800°C is performed in a muffle furnace to remove organic interferences. The next step is dissolving the sample in the melted flux, for which further heating is carried out at 1100°C. Combining these two steps yields a solid residue easily attacked by diluted acids such as 1 M HNO₃. Flux mass is an important factor for this method as it impacts the extent of solubilization of the oxides. In general, sample/flux ratios varying from around 2–20 are used, and for this method, this ratio can vary in the range of 1–10. Higher ratios must rely on the assumption of high purity of the flux, as any impurities would be magnified in the sample analysis. Melting temperature and time are also important as temperatures exceeding 1100°C might yield loss of sample and flux due to volatilization or spreading of the mixture outside of the crucible. Finally, oxidation time is considered relevant as well because only oxides are soluble in the mixture. Thus, complete oxidation of other components must be assured [31]. Consequently, these factors were studied under an experimental design to test the robustness of the method, as described below. The method scope can easily be extended to raw material samples, specifically the UV-Balance Powder 100™, and similar preparations due to their relatively simple formulas compared to the more complex cosmetic matrices.

3.2 Method validation

3.2.1 Specificity

Background and placebo blank signals, plotted in absorbance units as a function of time, are presented in Figure 1. Signals from placebo blanks show no difference compared to the background signal. This result confirmed that the extraction procedure removes matrix interferences if existing and that the final concentration of excipients is low enough to have negligible effects on the stability of the signal, as can be seen from their progression in time. Figure 1 also presents the Ti signal at the lowest concentration (20 ppm), showing that placebo signals do not interfere with Ti quantification.

Figure 1.

Method specificity. Absorbance as a function of time for background, placebo blanks, and Ti (20 ppm) signals.

3.2.2 Linearity and sensitivity

The calibration plots obtained for raw material, lipstick, foundation, and emulsion matrices were adjusted using Least Squares Regression and judged regarding the value of the r² parameter, which was chosen as the acceptance criterion for linearity with a limit value of 0.99 and is presented in Figure 2. Table 2 presents the values for r² and the other parameters of the model. According to these values, the data is well-fitted by a linear correlation in the range of concentrations used in this study. Furthermore, the statistical significance of these results, tested via ANOVA and Student’s test, is demonstrated. Calculated LOD and LOQ values are presented in Table 2. LOQ precision is within limits defined following USP guidelines (%RSD ≤ 5.0%) [33], and, regarding the accuracy, the percentages of recovery at LOQ are not statistically different from 100%, as shown by using a Student’s t-test. In general, the quality of these parameters illustrates the suitability of FAAS for the quantification of TiO₂ in cosmetic products.

Figure 2.

Calibration plots for the raw material (a), lipstick (b), emulsion (c), and foundation (d) matrices.

		Raw material	Lipstick	Emulsion	Foundation
Linear range	Slope	0.003388	0.002680	0.003097	0.002759
	r^2	0.9934	0.9950	0.9931	0.9913
	t valuea	71.8	82.1	69.9	62.1
	F valueb/10^3	5.15	6.74	4.89	3.85
Sensitivity	LOD mg L^−1	2.2794	1.9935	2.3396	2.6353
	LOQ mg L^−1	6.9073	6.0409	7.0898	7.9859
	%RSD	2.11	0.91	2.96	2.17
	%Recovery	99.3	99.1	100.0	97.8
	t valuea	0.824	2.515	0.016	2.530

Table 2.

Linearity and sensitivity parameters for all cosmetic matrices.

^a

t critical value: 2.032 (linearity), 2.571 (LOQ).

^b

F critical value: 4.1366.

3.2.3 Precision

Table 3 summarizes the total variation in terms of %RSD observed in this study for repeatability, intermediate precision, and reproducibility. %RSD values below 4.0% are observed across all concentration levels for all cosmetic matrices and below 3.0% for the raw material, showing that the method exhibits good overall repeatability. Regarding intermediate precision, an %RSD below 3.0% and near 1.0% for the foundation matrix indicates that the method is not significantly affected by slight changes in environmental conditions or different analysts. Regarding reproducibility, %RSD values are well below the USP recommended limit of 8.0%, which shows that the protocol was smoothly transferred to a different laboratory without significant loss in precision. %RSD values are comparable and, in some cases, lower than the results of other techniques [21], which is noteworthy as those methods are mainly designed for sunscreen preparation, whereas this method covers different categories of cosmetic products.

		%RSD
	Level	Raw material	Lipstick	Emulsion	Foundation
Repeatability	1		3.68	2.89	3.57
	2	2.21a	1.12	3.48	2.22
	3		1.37	3.52	3.21
Global intermediate precision		1.78	2.73	1.98	0.98
Global reproducibility		2.90	2.50	2.30	1.60

Table 3.

Total variation (%RSD) for the precision levels considered in this study.

^a

Raw material was evaluated at its nominal concentration, as declared by the certificate of analysis.

3.2.4 Accuracy

As an acceptance criterium for the accuracy of the method, a recovery rate not statistically different from 100% and contained in the interval 95–105% was defined as desirable. Table 4 presents recovery rates for all cosmetic matrices and the raw material. Data were evaluated by comparing nominal and found concentrations of TiO₂. The results show that the extraction protocol is optimal to achieve a variability below 3.0% for all matrices and concentration levels, except for the lowest level from foundation matrices where variability is below 5.0%. Global recovery rates are not significantly different from 100%, as demonstrated by Student’s t-test t values.

	Raw material	Lipstick			Emulsion			Foundation
Added concentration (%p/p)	87.30c	3.200	4.000	4.800	4.800	6.000	7.210	5.600	7.010	8.380
Average found concentrationa (%p/p, %RSD)	88.30, 0.93	3.219, 2.93	4.058, 0.95	4.815, 1.30	4.855, 1.84	6.094, 1.04	7.231, 2.00	5.584, 4.10	6.832, 2.23	8.613, 1.06
Recovery rate (m ean ± SD)	101.15 ± 1.03	100.59 ± 3.28	101.45 ± 1.05	100.22 ± 1.42	101.12 ± 1.98	101.52 ± 1.07	100.35 ± 2.20	99.76 ± 4.51	97.48 ± 2.41	102.72 ± 1.20
Global recovery rateb (mean ± SD)	101.15 ± 1.03	100.76 ± 1.94			101.00 ± 1.65			99.99 ± 3.48

Table 4.

Recovery rates for all the cosmetic matrices studied in this work.

^a

Sample size n = 6.

^b

t values: 1.961 (raw material), 1.175 (lipstick), 1.824 (emulsion), and 0.013 (foundation). t critical value: 4.303 (raw material), 2.306 (lipstick, emulsion, and foundation).

^c

Purity (nominal concentration) declared in the certificate of analysis.

3.2.5 Robustness

Flux mass, muffle furnace initial temperature, duration of muffle furnace initial and final steps, and the final concentration of the sample for FAAS analysis were considered relevant factors for the study of the method robustness. Method-predetermined values for these factors were defined as standard conditions and deviations from them, as shown in Table 1, were defined as alternative conditions under a Youden-Steiner model. Critical factors are given by values that are higher than the critical value of each cosmetic matrix, calculated as described elsewhere [34]. As can be observed from the results presented in Table 5, final concentration is a critical factor for all cosmetic matrices and the muffle furnace initial temperature. These findings are a consequence of the complexity of cosmetic matrices regarding the wide chemical variety of their formulas. The result is influenced by the temperature at which oxidation is carried out because incomplete oxidation of the organic matter present in the formulas might interfere with the solubilization of the oxides in the flux later on the sample preparation protocol. At the same time, the effect of final concentration on the result can be explained because increasing organic/inorganic load without increasing the time of the oxidation step might also yield incomplete oxidation, which later interferes with the solubilization in the flux. The effects seen on cosmetic matrices can be contrasted to those observed for the raw material, where increasing the final concentration of the solution for FAAS analysis does not increase the organic matter content to be oxidized, and for which a slight decrease in the oxidation step temperature does not have a significant impact because the amount of organic matter to oxidize, compared to cosmetic matrices, is not significant. In addition to these factors, the mass of flux is critical only for foundation matrices processing. This is reasonable considering that apart from inorganic UV filters such as TiO₂ and ZnO, these compositions usually include some inorganic pigment load and, specifically for the matrices considered in this study, pigments based on Fe₂O₃, an oxide which would also be solubilized in the flux and thus would affect the working sample/flux ratio of the method.

Parameter	Differencea
	Raw material	Lipstick	Foundation	Emulsion
Spectromelt ® A 12 (g)	1.761	0.015	0.382	0.180
Muffle furnace initial temperature (°C)	0.357	0.120	0.234	0.410
Muffle furnace initial step (min)	1.012	0.054	0.027	0.100
Muffle furnace final step (min)	1.116	0.063	0.068	0.090
Final concentration (ppm)	0.605	0.134	0.377	0.540
Critical value	2.725	0.064	0.214	0.300

Table 5.

Youden-Steiner design for relevant factors of the method.

^a

Bolded values correspond to critical factors.

4. Conclusions

We developed and validated a new method for the extraction and FAAS quantification of TiO₂ in cosmetic matrices such as emulsions, foundations, and lipsticks. The method was validated according to ICH and USP guidelines and showed to be specific, accurate, precise, and robust. Borate fusion was successfully implemented for the extraction of TiO₂ with no use of HF or harsh conditions for digestion and oxidation of the organic content in the sample, which translates into a safer method. We showed the method can also be extended to the analysis of raw materials and is easily transferred to other laboratories, which is ideal in the context of quality assurance. A method is thus a tool to guarantee the safety of the products according to FDA and European Commission regulations and to ensure batch-to-batch quality.

Acknowledgments

The authors would like to acknowledge the Research, Development & Innovation Department of Belcorp for the financial support and research opportunities to create methodologies that ensure the quality and safety of products to fulfill consumer expectations.

Conflict of interest

The authors declare that there is no conflict of interest regarding the publication of this paper.