In the cosmetics industry, colorants such as dyes and pigments are incorporated into products like lipsticks, eye shadows, and hair tints to achieve the desired colors [1,2]. Most colorants are organic pigments, of which azo pigments account for more than 60 % [2,3]. Synthetic organic pigments comprise a large group of compounds with unique chemical and physical properties such as chromophore structure, solubility, and stability. The most commonly used pigments in many industries are azo, anthraquinone, indigo, xanthene, and triarylmethane [4]. The colorants used for tattoos and permanent make-up (PMU) often exhibit low purity, as they are not explicitly manufactured for these applications [5,6]. In humans, exposure to cosmetic dyes predominantly occurs in areas close to mucous membranes (such as the eyes and lips). Additionally, several color additives used in cosmetics can potentially cause adverse health effects [2,7,8]. Azo dyes have undergone extensive scrutiny to comprehend their applications and side effects. Previous research indicates that around 60–70 % of azo dyes are toxic, carcinogenic, and resistant to conventional physicochemical treatments [6,9]. Their characteristic chromophore azo groups can be reduced to aromatic amines, leading to mutagenic, genotoxic, and carcinogenic effects [6,10]. Anthraquinone dyes, despite being widely used, are considered highly toxic, although research on their biotransformation and toxicity of anthraquinones is limited [11]. Triarylmethane-based dyes, known for accumulating on and penetrating the skin, can severely impact metabolism and prove carcinogenic if ingested or inhaled [12,13]. Additionally, xanthene dyes make the skin rough by reacting with skin proteins [14]. In the case of Rhodamine B, its harmful effects have been proven [15], and its use in cosmetics has been banned in many countries. In addition, the use of various colorants in cosmetics is not permitted by regulatory authorities (Food and Drug Administration in the US, European Commission in the European Union, Ministry of Food and Drug Safety in South Korea, etc.)

The exploration of colorants in cosmetics has been ongoing for years. In 1997, Rastogi et al. [16] proposed the high–performance liquid chromatography with diode–array detection (HPLC–DAD) method for analyzing organic pigments in cosmetics. The following analytical methods have been used for analyzing colorants: Fourier–transform infrared spectroscopy [17], ultraviolet–visible analysis [18], surface-enhanced Raman spectroscopy [19], thin–layer chromatography [20], liquid chromatography (LC) [21], LC with tandem mass spectrometry (LC–MS/MS) [22], and matrix–assisted laser desorption ionization mass spectrometry [23]. The sensitivity, resolution, and measurement accuracy of LC–MS/MS analytical techniques are constantly improving, enabling the detection of various analytes. However, several limitations (such as a lack of standards) remain. Consequently, there is a need for analytical technologies capable of detecting and characterizing a diverse range of colorants in cosmetics.

Recently, molecular networking (MN) has emerged as a potent tool for large-scale annotation of LC–MS/MS data [24,25]. MN is performed by grouping MS/MS spectra based on their spectral similarities. Molecules that share similar fragmentation spectra are connected within the network [26], [27], [28]. This approach allows inferring structural similarities based on the spectral similarities between molecules, which in turn allows visualizing structural relationships between molecules [29,30]. It proves effective for analyzing vast amounts of data, exploring thousands or even billions of MS/MS mass spectra without requiring standard materials or knowledge of the sample's chemical composition [31].

This study aimed to establish a method for the simultaneous analysis of multiple compounds to detected banned colorants in cosmetics. To that end, an LC-MS/MS simultaneous analysis method was developed and verified to accurately quantify 26 banned colorants in commercially distributed cosmetics. In addition, the MN screening method was established to build a basis for tentatively estimating and identifying more colorants beyond the 26 already analyzed. Here, we present a new analytical technology to simultaneously screen for various colorants (such as azo and triarylmethane, anthraquinone, nitro, and other compounds) by applying a multi-information MN strategy based on non-targeted LC–quadrupole time–of–flight MS (LC–Q–TOF–MS). The MN analysis of the banned colorants was implemented in a program provided on the Global Natural Product Social Molecular Networking (GNPS) website and high–resolution TOF–MS data identified the structural similarities between the colorant compounds, thus confirming the detection of non-targeted pigments in cosmetics. The results of this study are expected to contribute to enhancing the safety management of commercial cosmetic products.