Quantification ofMineralOil AromaticHydrocarbons ( MOAH ) in Anhydrous Cosmetics Using 1 H NMR

In cosmetic products, hydrocarbons frommineral oil origin are used as ingredients in a wide variety of consistency, from liquid oil to solid wax. Refined mineral oil hydrocarbons consist of MOSH (mineral oil saturated hydrocarbons) and a low proportion of MOAH (mineral oil aromatic hydrocarbons). MOSH and MOAH comprise a variety of chemically similar single substances with straight or branched chains. In the context of precautionary consumer protection, it is crucial to determine hydrocarbons from mineral oil origin of inferior quality quickly and efficiently. /is publication presents a rapid method for quantifying MOAH by proton nuclear magnetic resonance spectroscopy (H qNMR) in anhydrous cosmetics such as lipstick, lip gloss, and lip balm. A sample clean-up using solid-phase extraction (SPE) was developed for the complete removal of interfering aromatic substances to improve the robustness of the method for analysing compounded cosmetics. In preliminary trials using silica gel thin-layer chromatography, the retention behaviour of 21 common aromatic compounds was tested in eluents with different solvent strength including EtOAc, MeOH, cyclohexane, and dichloromethane. Based on these results, the SPE sample cleanup with silica gel and cyclohexane as an eluent was suggested as best suitable for the purpose. /e SPE cleanup was successfully achieved for all tested potentially interfering aromatic cosmetic ingredients except for butylated hydroxytoluene./e recovery for lipophilic cosmetics is more than 80% based on naphthalene as calculation equivalent. Furthermore, a specific sample preparation for the examination of lipsticks was implemented. /e SPE cleanup was validated, and the robustness of the method was tested on 57 samples from the retail trade. /e H qNMR method is a good complement to the LC-GC-FID method, which is predominantly used for the determination of MOSH and MOAH. Chromatographic problems such as migration of MOSH into the MOAH fraction during LC-GC-FID can be avoided.


Introduction
e term "mineral oil hydrocarbons" summarizes a complex combination of numerous saturated and aromatic chemically similar hydrocarbons. Mineral oil raw materials are used in cosmetics in different consistencies, for example, as liquid oil or solid wax. ey are classified according to their consistency into the groups listed in Table 1.
Hydrocarbons from mineral oil origin are often used in cosmetics due to their different positive properties, such as good skin compatibility, good cleaning performance, and high stability. ese substances are subject to mandatory declaration requirements and are indicated in the ingredients list as mineral oil, paraffin, paraffinum liquidum, petrolatum, cera microcristallina (microcrystalline wax), ceresin, or ozokerite (Table 1). In addition, hydrocarbons from mineral oil origin are inexpensive and can be produced in consistent quality. ese properties make them interesting for a wide range of cosmetic products such as skin care products, face and body cleanser, sun screens, and hair and lip care products [1][2][3][4][5].
Analytically, hydrocarbons are usually divided into two groups: MOSH (mineral oil saturated hydrocarbons) and MOAH (mineral oil aromatic hydrocarbons). MOSH consists of saturated aliphatic and cyclic hydrocarbons and MOAH of aromatic partially hydrogenated and highly alkylated compounds. In general, MOSH and MOAH are determined as sum parameters using convention methods (e.g., by LC-GC-FID). Toxic polycyclic aromatic compounds may not be contained, specifically potentially carcinogenic 3-7 ring polycyclic aromatic hydrocarbons (PAH). ey need to be removed by a comprehensive refining process to fulfil the prerequisites of the European cosmetics regulation EC/1223/09. Mineral oils used in cosmetics often also meet the purity requirements for medicinal products, i.e., they are in compliance with Pharmacopeia standards [6].
In terms of precautionary consumer health protection and in the light of a large variety of mineral oil containing cosmetics, there is a need for efficient methods for characterizing the profile of mineral oil in cosmetic products. e aim is to distinguish high-quality mineral oil hydrocarbons (pharmaceutical, cosmetic, or food-grade mineral oil) from less refined hydrocarbons (technical-grade mineral oil material) and to check the compliance with the European regulations. According to the current state of the art, different analytical methods are used, for example, online coupled high-performance liquid chromatography-gas chromatography-flame ionization detection (LC-GC-FID), comprehensive two-dimensional gas chromatographymass spectroscopy (GCxGC-MS), and nuclear magnetic resonance spectroscopy (NMR). A recently published review provides an overview of the current state of literature about the methods for detecting MOSH and MOAH in food, food contact materials, tissues, and cosmetics [1].
From the available methods, LC-GC-FID is currently the most widely used procedure for the analysis of mineral oil hydrocarbons.
is method uses the excellent separation performance of liquid chromatography (HPLC) to separate the hydrocarbons into saturated and aromatic hydrocarbons, predominantly MOSH and MOAH. Each single fraction is then separated by online coupled gas chromatography and detected by flame ionization detection (GC-FID) according to the volatility and carbon amount of the substances. As a result of the complex composition of mineral oils, no separated peaks but so-called humps are obtained, which still provide a relatively characteristic profile for the different mineral oil compositions [2,7]. For individual samples, GCxGC-MS after offline HPLC preseparation may be used to further characterize MOAH by ring number and degree of alkylation [3,8].
e chromatographic methods described above separate mineral oil hydrocarbons by the different elution force of the compounds. e proton NMR spectroscopy separates a sample into different signal areas based on the individual atom's electronic environment. For the MOAH fraction, the spectral aromatic region between δ 9.2 and 6.5 ppm can be integrated (excluding the solvent signals) and for nonmineral oil compounds the spectral region between δ 6.5 and − 3.0 ppm. A detailed description of the high-resolution 1 H quantitative NMR method ( 1 H qNMR) to evaluate MOSH and MOAH in pure mineral hydrocarbon-based cosmetics and cosmetic raw materials is given in Reference [3]. Furthermore, low-field NMR may offer a more economic approach for MOSH/MOAH screening [1].
In our experience, the quantitative 1 H NMR spectroscopy is a good complement to the LC-GC-FID method. It is useful as a simple screening tool to get information about the MOSH/MOAH distribution. By this evaluation, the MOSH/ MOAH ratio can be estimated and migration of MOSH into the MOAH fraction as evident in the LC-GC method can be circumvented. e LC-GC-FID method elutes firstly MOSH and then MOAH. e typically large excess of MOSH in cosmetic products (>99.5% [3]) can lead to carryover of the MOSH fraction into the MOAH fraction. By prior estimation of the MOSH/MOAH ratio using NMR, the sample weight for LC-GC-FID can be adjusted accordingly or a depletion of MOSH can be carried out as sample preparation [1]. Additionally, the determination of mineral oil via NMR spectroscopy provides a retention behaviour-independent validation of the LC-GC-FID results and additional information on further minor components that are visible in the 1 H NMR chemical shift range (δ 6.5-3.0 ppm) is possible. Nevertheless, the previously described NMR procedure [1,3] was only suitable for pure mineral oil-based cosmetic products. However, compounded cosmetic products containing hydrocarbons from mineral oil origin are often composed of many ingredients including aromatic substances, e.g., preservatives such as parabens, UV-filters, perfumes, antioxidants such as BHT, or active compounds such as tocopherol. ese aromatics must be  Ozokerite A complex combination of hydrocarbons. It mainly consists of saturated straight chain hydrocarbons (carbon numbers predominantly in the range of C20-C50).

Ceresin
"A complex combination of hydrocarbons produced by the purification of ozokerite." removed completely by a reliable cleanup to avoid falsepositive MOAH results by 1 H NMR spectroscopy because all aromatic compounds are detected in the spectral region between δ 9.3 and 6.5 ppm. is publication is a follow-up of the publication by Lachenmeier et al. [3] which focused on the determination of MOSH/MOAH in mineral oil raw materials and mineral oil-based cosmetic products without other aromatic ingredients. Using the sample preparation presented in this article, the method can be expanded to anhydrous cosmetic products (e.g., lip products) containing other aromatic ingredients. Furthermore, a specific sample preparation for the examination of lipsticks was implemented. A homogeneity experiment of 12 lipsticks and 12 lip care sticks showed that choosing a specific sample preparation for solid lip products is imperative. e procedure was validated for lip cosmetics (liquid, creamy, and solid), and the robustness of the method was tested on 57 samples from the retail trade. e entire process from sample preparation to the evaluation of the NMR data was validated.

in-Layer Chromatography (TLC)
. TLC was used to determine the appropriate eluent for the solid-phase extraction (SPE) cleanup on the silica 60 stationary phase. As a test system, 21 common aromatic compounds were used (Table 2), and their elution behaviour compared to naphthalene (as calculation equivalent for MOAH in NMR) was investigated in eluents with different solvent strength. As a matrix, a retail-available petroleum jelly was used to simulate a real product cleanup. e TLC plates were analysed under an UV lamp (Camag, Muttenz, Switzerland) at 254 nm.

NMR Method.
All 1 H NMR measurements were performed using a Bruker Ascend 400 spectrometer (Bruker Biospin, Rheinstetten, Germany) equipped with a 5 mm SEI probe PA BBI 400S1 with Z-gradient coils and a Bruker automatic sample changer (Sample Xpress, Bruker Biospin). All spectra were acquired at 300.0 K. e spectra (for samples dissolved in 120 µL acetone-d 6 /TMS + 480 µL cyclohexane sample solution) were acquired using the Bruker noesygppr1d pulse program with 32 scans, 2 prior dummy scans (DS), and receiver gain of 45 (RG). is pulse program is used to suppress the H 2 O signal and has been further optimized for the cyclohexane parameters. e 1 H NMR method largely corresponded to the one described in Reference [3]. For investigations in cyclohexane, cyclohexane suppression was applied. 1 H quantitative NMR in this publication has been established using the ERETIC methodology (electronic reference to access in vivo concentrations) based on the PULCON principle (pulse length-based concentration determination) [9]. e ERETIC factor was established using a quantification reference (quantref ) sample containing ethyl benzene (10.24 mg mL −1 in cyclohexane/acetone-d 6 ) and diethyl phthalate (8.86 mg mL −1 in cyclohexane/acetone-d 6 ). e quantref sample was prepared by dissolving 54 mg diethyl phthalate in 5 mL cyclohexane/ acetone-d 6 . e recovery had to be 100 ± 5%.

Samples and Sample Preparation.
For the robustness of the 1 H qNMR method, it is crucial that all aromatic compounds except mineral oil aromatic hydrocarbons are removed during sample preparation. e sample cleanup depends on the matrix, as shown in Table 3.
Sample preparation for pure hydrocarbons from mineral origin and hydrocarbon-containing cosmetics without further aromatic ingredients can be conducted as previously published [3]. In short, 50 mg of cosmetic product was dissolved in 1.5 mL CDCl 3 in an ultrasonic bath. To enhance the solubility of the sample, the ultrasonic bath was heated up to an elevated temperature (Table 3). When turbidity occurred, syringe filters with PET membrane or GF/PET membrane were used for membrane filtration depending on the degree of it. A given amount of sample solution (Table 3) was then taken for the NMR measurement.
Most cosmetic products contain, in addition to mineral oil aromatic hydrocarbons, other aromatic compounds, such as BHT, which would interfere with the 1 H qNMR measurement (Table 2). A matrix-dependent cleanup was developed to remove a large number of these components. e cleanup can be universally applied to anhydrous cosmetics (e.g., lip care products and lipsticks). For this, 50 mg of cosmetic product was dissolved in 1.5 mL cyclohexane in an  (Table 3).
As additional preparation step for lipstick, the following two steps have to be carried out to ensure homogeneity of the sample. A cross section of the lipstick is cut from the core of the product (2-3 mm from the edge), and a peanut-sized potion is scraped out (Figure 1(a). e obtained mass was filled with a spatula to homogeneity on a watch glass (Figure 1(b)).

Validation Concept.
For the validation of lip cosmetics, 3 lip cosmetics with different consistency were chosen: an intensely colored lipstick (solid product), a lip balm (creamy product), and a colored lip gloss with glitter particles (liquid product). First, all cosmetic agents were analysed for their MOAH content by 1 H qNMR spectroscopy. For all 3 products, this was below the detection limit of 0.03 g MOAH/100 g sample. erefore, all 3 products were suitable for spiking experiments according to the preparation specified in Table 4. e INCI list for each matrix is listed in Table 5.
A vaseline of known MOAH content was added to each matrix (additive weighing). e vaseline was stirred into the cosmetic product by melting the matrix (Table 4). Vaseline was added in nonequivalent proportions (9 points in the range of 0.00-1.19 g MOAH/100 g sample) [10] (the results are summarized in Tables 6-8). e validation series was prepared as described in the sample preparation (Section 2.4).

Initial TLC Experiments for Solvent Selection.
e retention behaviour of 21 frequently used aromatic ingredients ( Table 2) was tested by a TLC experiment. As a TLC system, silica gel and eluents with different solvent strength (ethyl acetate, methanol, tert-butyl methyl ether, cyclohexane, dichloromethane, water/methanol (25/75), tert-butyl ethyl ether/cyclohexane (20/80), and tetrahydrofurane/cyclohexane (20/80)) were investigated experimentally. All 24 compounds retard slightly with polar eluents. ere is no sufficient elution difference between polar compounds and MOAH. A desired low retention for MOAH and sufficiently good retention for    the remaining compounds is achieved for the eluents cyclohexane, dichloromethane, and n-hexane/cyclohexane. e result reflects the elution behaviour corresponding to the eluotropic series of silica gel. For the SPE cleanup, cyclohexane was selected as the preferred eluent, because it is suitable for 1 H qNMR spectroscopy. In addition to the correct elution behaviour, the signal ranges of the eluent in the 1 H NMR spectroscopy were a decisive decision criterion, with cyclohexane (1.44 ppm), dichloromethane (5.30 ppm), and n-hexane/cyclohexane (1.28 and 0.88 ppm).

SPE Cleanup.
Based on the results of thin-layer chromatography, the SPE cleanup for silica gel/cyclohexane was developed. To check the robustness of the SPE cleanup, the breakthrough rate was screened for some frequently used cosmetic ingredients (Table 9). e breakthrough rates of these critical ingredients provide information about the robustness of the implemented SPE cleanup step. As shown in Table 9, most of the investigated polar aromatic compounds are well retained due to their interactions with the hydrophilic groups of the deployed silica gel and showed no interfering signals in the 1 H NMR spectra between δ 9.3-6.5 ppm after SPE cleanup. Only a few substances, such as benzyl benzoate and tocopherol, elute partially with breakthrough rates of less than 1%. Furthermore, some perfuming agents such as eugenol, coumarin, benzyl alcohol, and several cinnamic acid derivates were tested as mixtures in different commercially available perfume oils. e breakthrough rates determined were generally below 8%. Usually, preservatives, antioxidants, and perfume oils are used in percentages below 2% in cosmetics while these experiments were performed with weights similar to cosmetics sample preparation (50 mg ± 5 mg) to achieve significant results. e actual signals in real cosmetic products would therefore be most probably below the detection limit for NMR measurements. Figures 2 and 3 show two application examples of a mineral oil containing an UV filter and a lip gloss sample.
is situation is different with butylated hydroxytoluene (BHT), a frequently used antioxidant with additional masking functions. It can reduce oxidant reactions and the intrinsic smell and taste (lipstick) of a product to improve its storage life and scent. Owing to the two tert-butyl groups strongly shielding the hydroxyl group, BHT elutes almost entirely resulting in more than 90% breaking through. e steric hindrance leads to visible signals for BHT even at the commonly low concentrations in cosmetic products. A comparison to the structural and functional similar butylated  hydroxyanisole (BHA) shows that the steric hindrance is obviously only given with more than one tert-butyl group as the aromatic BHA signals were missing after the cleanup (Figure 4). BHTcan be detected by the characteristic singlets at δ 6.88 ppm and δ 5.32-5.29 ppm (shifting signal) and therefore needs to be taken into account for MOAH detections in cyclohexane/acetone-d 6 to avoid false positive results.

Special Preparation Step for Lipstick and Solid Lip
Products. Lipsticks were found to be challenging to analyse due to homogeneity issues. ese may arise during the production process of lipsticks. e shape of lipsticks is   achieved by pouring the hot mass into special silicone or metal casting molds. When the lipstick cools down in the casting mold, different temperature zones are created, which can lead to inhomogeneity with respect to the composition of the pencil as a result of the flow behaviour. e process of shaping itself can also contribute to the inhomogeneity of the surface. e potential inhomogeneity within a lipstick is irrelevant to the user of the product, but may be relevant to the comparability of analytical results. In an experiment, 12 lipsticks and 12 lip care sticks were examined with regard to the homogeneity of MOSH and MOAH. For this purpose, for each lip product, at 3 different points of the lipstick, a sample was taken (from the surface of the lipstick to the core). Figure 5 shows the results of the 36 samples from 12 lipsticks. It becomes apparent that the results regarding MOSH led to a highly significant inhomogeneity of the results, whereas the MOAH values did not lead to inhomogeneity. In the process of making a lipstick, the ingredients are melted and poured into metal or silicone molds. e sticks are then cooled to freezing. e solidification occurring during this process does not lead to a perfect homogeneous distribution of oils and waxes within the lipstick. During the cooling process, there is a temperature gradient from the core of the lipstick to the edge. e different temperature zones lead to surface phenomena at the edge zone of the lipstick. A uniform distribution of oils and waxes in the pencil is not to be expected. e MOSH value recorded by 1 H NMR spectroscopy encompasses all saturated hydrocarbon building blocks, including side chains. An inhomogeneity of the stick is therefore mainly visible in the MOSH value. e homogeneity experiment was repeated using the additional sample preparation step for lipstick (Section 2.4).
e results are shown for 3 lipsticks that in the first experiment exhibited results strongly deviating from each other. It turns out that the same lipstick leads to homogeneous results with suitable sample preparation ( Figure 6). To achieve comparable results, the sample preparation of lipstick or solid lip care products therefore requires the additional step as described to ensure comparable results.

Validation Results and Applicability in Routine Analysis.
ree experimental runs were carried out, each at nine concentration levels with varying different parameters (first, matrix; second, NMR spectrometer; and third, sample preparation). For each particular run, an individual blank matrix was used for spiking: a liquid product (lip gloss), a creamy lip balm, and a solid product (lipstick). e validation series were measured on two different NMR spectrometers (400 MHz) to include the variation of the devices during validation (second robustness criteria). e 3 rd robustness criterion takes into account the sample cleanup (as described in Section 2.4) by two different technicians. In order to calculate the performance characteristics (recovery, 0.95 prediction band, and critical limits), all obtained results were regressed on the spiked amounts by weighted least squares regression [10] (Figures 7-9).
A 95% prediction band of 0.02 to 0.05 g MOAH/100 g for a spiked MOAH content of 0.05 g/100 g and a 95% prediction band of 0.91 to 1.27 g MOAH/100 g for a spiked MOAH content of 1.19 g/100 g are shown in Figure 7. Recovery was sufficient above 0.2 g MOAH/100 g (Figure 8). e measurement uncertainty is sufficient below 20% also at concentration above 0.2 g MOAH/100 g (Figure 9).

Comparison with Reference UV/VIS Methods.
In order to determine the purity and quality of raw materials from mineral oil origin, photometric methods are still often applied, e.g., Ph. Eur. [6]. ese methods are used for refined products to ensure a minimized content of polyaromatic compounds. In the present work, the Ph. Eur. method [6] was used as a comparison method. e method is based on UV/ Vis measurements of DMSO extracts that need to fulfil certain requirements depending on the nature of a mineral oil raw material. us, e.g., vaseline can be checked for conformity to the established standards. e DMSO extracts are being compared to naphthalene solutions. In the case of white and yellow vaseline, the extracts' absorbance needs to be below a certain threshold predetermined by the absorbance maximum of different naphthalene concentrations in DMSO. Naphthalene therefore has also been chosen as calculation 5 Figure 4: Comparison of BHT (a) and BHA (b). e blue 1 H NMR spectra show the signals of a direct measurement in cyclohexane/ acetone-d 6 , while the red spectra represent the result of the SPE cleanup. While BHA is completely removed by SPE, a considerable amount of BHT remains. equivalent for MOAH determination of mineral oil raw materials by the previously described 1 H NMR method [3]. Figure 10 shows the comparison of the absorbance maxima in DMSO provided by the Ph. Eur. method and the MOAH content determined via 1 H qNMR in CDCl 3 for 13 di erent vaselines.
All samples appeared to be in conformity with the Ph. Eur. standard requirements although the MOAH contents varied between 0.06 and 1.10 g/100 g. is goes in line with the Ph. Eur. labelling of 9 of the examined vaseline (blue). e remaining four were of technical grade (red) usually containing more aromatic compounds than highly re ned mineral oil raw materials.
ere is no obvious correlation between the two di ering methods as higher MOAH contents do not always correspond with high absorbance maxima and vice versa. Furthermore, technical-grade products do not stand out with higher MOAH contents as this is, e.g., also

Results of Lip Cosmetics.
Di erent lip products from the retail trade were purchased as test products, and their MOAH content was analysed. As already presented in the validation concept (Section 2.4), the products were classi ed according to their consistency into liquid, creamy, and solid lip products. e results are shown in Figure 11: 19 liquid products, e.g., lip gloss (red), 10 products of creamy consistence, e.g., lip balm products (blue), and 27 solid products, e.g., lipstick or lip care stick (green). Figure 11 also illustrates the wide range in the magnitude of MOAH content among lip cosmetics (from 0.0 g/100 g to 1.6 g/100 g).
e majority of the products tested had an MOAH content of less than or equal to 0.1 g/100 g sample (Table 10). Higher MOAH values (MOAH in the range of 0.1 g/100 g-0.4 g/100 g) were determined for some liquid and creamy lip products. For some lip gloss products, very high MOAH levels were found (0.4 g/100 g-1.6 g/100 g). Such high MOAH contents were also obtained for mineral oil raw materials and vaseline of technical quality, as shown in the method comparison Ph. Eur./ 1 H qNMR (Figure 9). Such products should be further scrutinized if they are in compliance with the purity criteria of the European cosmetics regulation.

Conclusions
As demonstrated, most of the polar aromatic compounds such as aldehydes, carboxylic acids, and their esters, as well as alcohols di er signi cantly from MOAH in their chemical structure or polarity and hence are well retained during SPE due to their interactions with the hydrophilic groups of the deployed silica gel and can be removed by a matrix-based cleanup. Only for sterically very demanding compounds, a breakthrough rate is recorded. e comparative investigation of BHT and BHA shows that BHA retards well despite a sterically demanding tri-tert-butyl group. Only the strongly shielded BHT with two tri-tert-butyl groups shows poor elution behaviour and cannot be separated from MOAH by the SPE cleanup.
In conclusion, the proposed SPE cleanup enables the determination of MOAH for anhydrous cosmetic agents (e.g., lip care products) by 1 H qNMR spectroscopy as an important element amongst others such as LC-GC-FID to quantify MOAH-equivalents. In future, it will be essential to characterize MOAH fractions using those elements and hopefully gaining more information about the composition of those fractions by more speci c methodologies.
Data Availability e data used to support the ndings of this study are available from the corresponding author upon request.

Conflicts of Interest
e authors declare that there are no con icts of interest regarding the publication of this paper.