Traditionally, reported methods for the determination of dyes focused on food matrices [
20,
21,
22,
23]. The analytical methodology for cosmetics ingredients regulated in the positive lists according to the European Regulation 1223/2009 [
17] was reviewed in 2016 [
16]. Thus, this paper only considered permitted dyes, and so, methodologies for the analysis of banned compounds were not included. In the current review, the field has been extended to include analytical methods for both permitted and prohibited dyes in cosmetics, which are summarised in
Table 2. This table compiles analytical methodologies developed for the determination of synthetic dyes in any kind of cosmetics and reported in international journals between 2008 and 2018. These methodologies are scarce with respect to all those reported for other cosmetic ingredients such as preservatives or UV filters. Moreover, there are no official methods to date in UE, and the existing methods just consider a part of the total regulated dyes. Many methods have also been applied in other matrices such as foodstuff, beverages or wastewater.
Out of 54 dyes considered in the papers reviewed, 16 are banned according to European Regulation (colour index underlined in
Table 2) and 38 are permitted, which represents a limited share of the colorants regulated by Annex IV [
17].
Figure 1 shows the 10 permitted dyes most determined in cosmetics. They mainly correspond to red shades and seven of them belong to the azoic group. Regarding banned dyes, the majority of studies are focused on the determination of Rhodamine B. Given its genotoxic effects, many methods have been exclusively developed for the analysis of this compound [
9,
25,
27,
28,
36,
38,
39], and it will be commented on separately. With the exception of these cases, the current trend is the development of methods for the simultaneous analysis of mixtures of colorants: up to 32 for qualitative analysis [
32] or 19 for quantitative analysis [
26]. Although out of the period considered in this review, it is worthy to note that in 1997 Rastogi et al. [
41] built a useful spectral library consisting of retention times and UV-Vis spectra of 130 organic cosmetic colorants for the purpose of identifying the colouring matter in cosmetic products, and a method based on extraction by SPE and analysis by HPLC-DAD with ion-pair mobile phase was optimized for routine control of colorants in lipsticks and varnishes.
3.1. Sample Preparation
Cosmetics are very different in their form (liquid, semisolid, powder, wax, etc.) and composition. The content of colorants in cosmetics may also vary widely depending on the type of product. Decorative cosmetics such as lipsticks, blushes, face powders, mascaras, eye products or nail polishes, contain the highest percentage of dyes. Therefore, these matrices have been the subject of most of the analytical studies [
15].
Sample preparation is a necessary step in the analytical process and it often requires time and energy-consuming operations that imply large amounts of organic solvents, acids or chemicals. Currently, the main priority is the development of sustainable methodologies in accordance with the principles of green chemistry. Simple and rapid methodologies with minimal consumption of reagents and solvents are increasingly demanded by control laboratories. In recent years, there has been a tendency towards the use of microextraction techniques or the miniaturization of conventional procedures of sample preparation.
Figure 2 briefs the sample extraction strategies proposed in the methods summarized in
Table 2. The amount and type of solvent employed is an important factor to develop more environmental friendly alternatives. In general, methanol or mixtures of methanol and other polar solvents were preferred as eluting or extracting solvent and in particular when the extracts are subsequently analysed by LC, given its compatibility with the mobile phase. However, extraction of dyes from wax-based cosmetics such as lip products often involved the use of toxic organic solvents such as chloroform or dichloromethane.
Some methodologies focused on the determination of a unique colorant in one or a few cosmetic matrices. In this line, Wang et al. [
37] proposed a simple sample treatment for the analysis of Lithol Rubine B (CI 15850) in nail preparation, lipstick and rouge, and Guo et al. [
35] developed a method based on ionic liquid dispersive liquid-liquid microextraction (IL-DLLME) for the determination of Brilliant Blue (CI 42090) in eau de toilette and shampoo. DLLME is a relatively recent micro-extraction technique based on a ternary component solvent system. An adequate extraction solvent and disperser solvent are added to the aqueous sample to form a cloudy solution. To collect the enriched phase with the analytes, a solvent with higher density than water must be used. The conventional solvents for DLLME are chlorobenzene, chloroform, carbon tetrachloride, and carbon disulfide. The use of ionic liquids (IL), considered more environmentally friendly, improve the sensitivity and selectivity, becoming an interesting approach; however, dispersing solvents, heating and cooling down, ultrasonication or additional chemical reagents are frequently required for the dispersion and sedimentation of ILs, resulting in long sampling time. To avoid these additional requirements, Guo et al. [
35] employed 1-decyl-3-methylimidazolium tetrafluoroborate ([C10MIM][BF
4]) as the ionic liquid for the sample preconcentration. A full optimization of parameters affecting the micro-extraction such as volume of IL, pH and time of incubation and centrifugation was performed. Since an aqueous media is needed for the dispersion of ILs, this is a selective and sustainable approach only for water soluble samples.
As mentioned before, Rhodamine B is a widely considered dye. In recent years, several methods have been developed for its determination in cosmetics, especially in lipsticks. In 2008, Wang et al. [
38] proposed a spectrofluorimetric method with minimal pre-treatment based on dissolution of lipstick in hot water by mechanical stirring. Adding an anionic surfactant during the flow-injection analysis increased the fluorescent sensitivity until 2.5 fold. With this procedure, the authors estimated a sampling rate higher than 100 samples h
−1. In the same year, Pourreza et al. [
39] developed a method based on cloud point extraction using Triton X-100 as non-ionic surfactant in acidic media. This preconcentration and extraction technique is considered in accordance with the “green chemistry” principles. It is based on the separation, at a certain temperature, of two phases in aqueous solution, one of small volume rich in surfactant and another with low concentration. After separation, the turbid solution is cooled down, so the surfactant rich phase becomes viscous and it is easily separated by decantation. The method was applied to different types of samples including a cosmetic one: hand washing liquid soap.
Solid phase extraction is a well-established preconcentration method generally applied to the analysis of trace amounts of analytes in environmental samples. It has also been proposed as an alternative extraction technique for lipstick [
9,
25,
36]. Soylak et al. [
9] developed a method using SPE with Sepabeads SP 70 resin as adsorbent. The purpose was to preconcentrate traces of Rhodamine B in waste water samples and soft drinks but it was also applied to the extraction of the dye in lipstick, previously dissolved in water by heating and stirring. Rhodamine retained in the adsorbent was eluted with 5 mL of acetonitrile and analysed by UV-Vis spectrophotometer. Validation was performed only for the liquid samples so extraction efficiency in lipsticks is not provided. Rhodamine B was identified and quantified in two lipsticks at concentrations of 39 and 127 µg·g
−1. Recently, Bişgin et al. [
25] developed two extraction methods based on SPE and CPE (cloud point extraction) for the determination of Rhodamine B in lipstick as well. In both cases, sample was previously dissolved in 25 mL of CCl
4, which is potentially a long-term carcinogenic solvent. For SPE, amberlite XAD-1180 was used as adsorbent and 5 mL of ethanol to elute Rhodamine. For CPE, Tergitol NP-7 was employed as surfactant. In view of the experimental procedure, SPE implies a shorter sampling time, higher preconcentration factor and better repeatability. On the other hand, sensibility is slightly better in the case of CPE. In 2013, Bagheri et al. [
36] synthetized a novel magnetic nanocomposite (Fe
3O
4–aniline-naphthylamine) as a high-efficient sorbent for micro-SPE for the determination of Rhodamine in several samples including shampoo, eye pencil and eye shadow. In this case, an amount of 7 mg of sorbent is added to the sample solution and stirred for 10 min. Then, the sorbent is collected with an external magnet and Rhodamine is desorbed with 2 mL of methanol. This alternative sorbent led to higher extraction efficiency due to its increased sorbent surface area and porosity.
Finally, two methods based on DLLME have been developed for the determination of Rhodamine B in decorative make-up (lipstick, rouge and nail polish) [
27] and Rhodamine B and 6G (CI 45160) in lipstick [
28]. The first method [
27] is based on the combination of two supramolecular solvents, polar and non-polar, such as THF or decanoic acid. Although the experimental procedure is considered fast, prior to micro-extraction a step consisting of solving 0.5 g of sample in 10 mL of ethyl alcohol and shaking for 2 h is needed. In the second case, Ranjbari and Hadjmohammadi [
28] optimized a method based on magnetic stirring assisted DLLME (MSA-DLLME), using 1-octanol and acetone as extraction and dispersing solvent, respectively. As mentioned before, DLLME needs extraction solvents with higher density than water, which frequently implies the use of toxic organic solvents. To overcome this limitation, magnetic stirring is introduced in the methodology in order to help to maintain the cloudy solution and to accelerate the mass transfer from aqueous solution to the extraction solvent, without the need for ultrasonication. A home-made glassware extraction cell has been designed to improve the magnet rotation and to simplify the collection of the supernatant organic solvent enriched with the analytes after centrifugation. To apply a technique based on liquid-liquid extraction to a solid sample such as lipstick, a previous treatment is obviously needed. In this case, 0.05 g of lipstick was solved in 50 mL water assisted by ultrasonic waves and mechanical stirring, and then it was filtered and transferred to the extraction cell.
In the context of a growing market of colour cosmetics, multianalyte analytical methodologies are indispensable to satisfy the increased demand of quality control. In the period under consideration, the first extraction method for the determination of a mixture of dyes was proposed by Xian et al. [
33] in 2013. Before this work, several multianalyte methods are worthy to consider. Noguerol et al. [
40] developed a method based on HPLC-UV and HPLC-ESI-MS/MS for the application to commercial products, but neither the type of sample nor sample preparation method were provided. Nizzia et al. [
34] proposed an analytical approach based on desorption electrospray ionization mass spectrometry (DESI-MS), where sample treatment is not required for the analysis. The work published by Xian et al. [
33] was focused on the development of a method for the analysis of 11 dyes in cosmetics. In this way, a simple methodology based on vortexing, ultrasounds and centrifugation was successfully applied achieving quantitative recoveries in eye shadow, lipstick and lip gloss. However, for the extraction from waxy matrices, longer sampling time and chlorinated solvents were required though in small volumes (2 mL of chloroform). Similar liquid-liquid extraction using combinations of dichloromethane, methanol, acetic acid and water, was proposed by Miranda-Bermudez et al. [
32]. The developed and validated methodology was employed to survey 29 colour additives, including water and methanol-soluble permitted dyes and the most prevalent non-permitted colour additives found by FDA’s district laboratories, in 38 samples of lip and eye products, nail polish, blush, body glitter, face paint, cream and toothpaste. Despite the broad range of analytes and samples studied, this methodology was developed only for the qualitative identification of dyes.
Barfi et al. [
31] and Franco et al. [
30] focused mainly on the determination of illegal dyes in cosmetics. In the first work, a novel, simple and eco-friendly method based on one step air-assisted liquid-liquid micro-extraction (OS-AALLME) was developed to extract Sudan dyes and Orange G in lipstick. In addition, because of the high risk of direct oral ingestion of sample, the validated extraction method was applied to estimate the concentration of a potential biomarker of these dyes, 1-amino-2-naphthol, in human bio-fluids. The advantage of this approach is the application of ILs as extraction solvent, avoiding the use of organic solvent. The immiscibility of ILs in water and their capability to solubilize organic species make them a valuable alternative to conventional solvents. To carry out the dispersion of the IL in aqueous solution, the mixture is repeatedly withdrawn into a glass syringe and pushed out (AALLME), while sonication is performed in order to increase the surface contact between the immiscible liquids and then, to enhance the extraction efficiency. In this way, micro-extraction takes place in one step. On the other hand, Franco et al. [
30] developed a method based on sample dilution in water and SPE for the determination of four basic dyes and one acid dye used as semi-permanent hair colorants. Oxidative hair dyes have been the subject of most of the analytical methodologies for the control of hair colorants due to their widespread use and there are very few methods for semi-permanent dyes. Hair dyeing formulations do not require a complex extraction method. After shampooing, the dyes are retained in the structure of hair through ionic interactions or van der Waal forces.
Recently, Guerra et al. have been working on improving the simultaneous analysis of a great number of chemically different dyes in a broad range of cosmetics, including rinse-off and leave-on products [
24,
26,
29]. In 2015 [
29], they introduced the matrix solid-phase dispersion (MSPD) for the first time in the analysis of dyes in cosmetics. MSPD is a valuable approach for solid and semisolid samples for its simplicity and the possibility of performing a clean-up step simultaneously. In this case, miniaturization of the conventional procedure employing a Pasteur pipette as device in order to pack the sample dispersed (see
Figure 3), allowed extracting quantitatively nine dyes in many cosmetics including lipstick, nail polish or toothpaste. More recently, they published a new method [
26] based on miniaturized MSPD with significant improvements in relation to the previous one. In this work, 19 permitted and banned dyes were analysed, 10 more than in the former, in a broader range of matrices, which involved a re-optimization of the sample preparation procedure using experimental designs. The Pasteur pipette was replaced by a 2 mL syringe to enable an adequate solvent elution employing the dispersant chosen as the optimum, C18. The most recent multi-dye method was also reported by Guerra et al. [
24]. Single-step vortex extraction and simultaneous clean-up was applied to the analysis of dyes as well as preservatives in different cosmetics, many of whom have not been analysed yet. In comparison with other approaches based on simple solvent extraction assisted by ultrasounds or centrifugation, this method offers advantages in terms of simplicity, rapidity and minimal consumption of solvents. In addition, it may be considered more environmentally friendly since it avoids the use of solvents such as dichloromethane or chloroform.
3.2. Analytical Techniques
In order to identify and quantify properly the dyes in cosmetic samples, a chromatographic separation is frequently required. Liquid chromatography coupled to absorbance or mass spectrometry detectors were proposed in all analytical methodologies for the simultaneous determination of mixtures of dyes (
Table 3). Due to the capability of dyes to absorb in the UV-Vis spectrum, DAD or UV-Vis detectors were traditionally the preferred detectors [
28,
30,
31,
32]. In recent years, mass spectrometry has become a valuable choice [
24,
26,
29,
33,
34], given its enhanced selectivity and sensitivity; it is particularly desired in the analysis of banned compounds. Mass spectrometry overcomes limitations of DAD such as overlapping of UV-Vis spectra among matrix ingredients. In some studies, a comparison between LC-DAD and LC-MS/MS was established [
30,
40]. In this way, Noguerol et al. [
40] developed two LC methods with UV-Vis detection and with tandem mass spectrometry triple quadrupole in positive ion mode electrospray ionization (ESI) for the routine control of 10 dyes, mainly dyes banned in cosmetics. LC-ESI-MS/MS in full scan mass spectra mode allowed to obtain structural information about multiple peaks observed for some of the studied dyes through HPLC-UV/Vis analysis. The presence of more than one peak for a compound was mainly due to possible isomerization processes, impurities or degradation products. In addition, multiple reaction monitoring (MRM) mode was employed for quantification purposes. In terms of sensitivity, there was a remarkable difference between both methods. Limits of quantification were one or two orders of magnitude lower for the LC-MS/MS analysis, which is particularly valued given the high restriction of those target dyes in the commercial products. This work focused on the development of an analytical method for its implementation in the control of dyes in commercial products, but not specifically for cosmetics.
The combination of liquid chromatography and mass spectrometry in the analysis of dyes in cosmetics was reported for the first time by Xian et al. [
33] in 2013. Afterwards, Guerra et al. developed several improved methods based on this analytical technique [
24,
26,
29]. Most of the dyes used in cosmetics and so considered in this review are sodium or calcium salts which contain in their structures one or more ionized groups such as sulphonic groups. This fact implies the possible formation of multicharged ions in the ionization source. In addition, the separation of ionic compounds by reverse phase liquid chromatography is a challenging task and more efforts must be done in relation to the separation of neutral compounds. In this respect, the mobile phase (ionic strength, pH, and composition) plays an important role. In some cases, a mobile phase without additives, composed by water and an organic modifier (acetonitrile or methanol), has been employed achieving good sensitivity and a fast analysis [
29,
33]. The use of UPLC allowed performing the analysis of 11 dyes in 4 min [
33]. Similar results were achieved for a mixture of nine dyes with a conventional porous C18 column [
29]. Although the use of mass spectrometry allows a selective identification of co-eluted compounds, a chromatographic separation is generally recommended. For this purpose, the addition of volatile neutral salts to mobile phase is necessary in order to avoid interactions between ionized negatively charged compounds and partially ionized residual silanols in the stationary phase [
46]. However, the presence of salts in the ion source may cause a suppression of the ionization. Thus, the composition of mobile phase must be investigated to achieve a compromise between good separation and performance. In this way, the use of only 3 mM ammonium acetate in the aqueous mobile phase was proposed [
24,
26]. This salt concentration was enough to avoid peak tailing while improving the chromatographic separation for quite a number of analytes with satisfactory limits of quantification. In the most recent work [
24], other chromatographic parameters were optimized to separate dyes and preservatives.
The matrix effect is the ionization suppression or enhancement of the target compound by others present in the sample and it is very frequent in LC-MS/MS analysis, in particular when electrospray sources are used. In every method for dyes analysis using MS/MS detector, a study of matrix effect was performed. The most comprehensive study was carried out for 19 dyes in seven cosmetic matrices (lip balm, nail polish, hairspray, eye shadow, toothpaste, face painting and gel) [
26]. In all cases, the optimized sample extraction procedure allowed obtaining an extract clean enough to perform the analysis with negligible matrix effects, with particular exceptions for some compounds in very few matrices.
In contrast to conventional mass spectrometry techniques, Nizzia et al. [
34] investigated the use of DESI-MS for the analysis of common semi-permanent hair colorants in two semisolid cosmetic formulations: a blemish cream and a hair-dye gel. As a novelty, the use of an ambient MS technique allowed a direct analysis without prior sample preparation or chromatographic separation. A thin layer of sample is deposited onto porous Teflon and a pneumatically assisted electrospray is employed to release neutral analytes present on this surface as secondary ions.
On the other hand, UV-Vis and DAD detectors are still often used in the literature for the analysis of dyes in commercial products. Concerning cosmetics, there have been several methods reported in the last 10 years that combine LC and DAD for the quantitative [
28,
30,
31,
32] and qualitative [
32] analysis of multiple dyes or that apply direct spectrophotometric measurement when only one dye is considered [
9,
25,
27,
35,
39] (see
Table 4). In these latter methods, good limits of detection were obtained because of the concentration achieved using an extraction technique such as SPE, CPE or DLLME previous to the analysis. Most of these methods were developed to identify and quantify Rhodamine B, so they established a fixed wavelength, 556 nm, which corresponds with the maximum of absorbance of this compound. When a mixture of dyes is analysed, DAD is a useful tool that allows the simultaneous recording of absorbance data from 190 to 800 nm and to match the UV-Vis spectra obtained with spectral libraries.
For the determination of basic dyes through LC-DAD, Franco et al. [
30] proposed the use of ILs in the mobile phase to enhance peak shapes and to reduce the chromatographic retention times. ILs usually compete with basic groups for the residual silanols on the stationary phase or they can form an ion pair with cationic solutes to improve the shielding efficiency of these silanols. In this way, the analysis was completed in 40 min using an isocratic mode with a mobile phase composed by acetonitrile and water and 2 mL of 0.040 mol·L
−1 BMIm[NTf
2] solution. The results obtained were compared with those obtained though LC-MS/MS by means of a t-student test. Concentration values obtained in both cases were not significantly different (at 95% confidence level). However, it must be pointed out that the mass spectrum provided chemical structural data that were employed to confirm the presence of the target compounds in the hair-dyeing formulations found by LC-DAD analysis.
Other analytical approaches were reported based on fluorescence measurements [
36,
38] or voltammetry [
37] for the determination of Rhodamine B or Lithol Rubine, respectively. The main advantage of the fluorometric measurement is its high selectivity.