Identification of organic pigments in tattoo inks and permanent make-up using laser desorption ionisation mass spectrometry

Chemicals

As reference materials, more than 150 individual pigments from different producers (mainly BASF (Basel, Switzerland), Clariant (Muttenz, Switzerland) supplied by Omya (Oftringen, Switzerland), Kremer Pigmente (Aichstetten, Germany) and Sigma–Aldrich (Buchs, Switzerland)) were collected over the past years. Mass calibration was performed with the following dyes all from Sigma-Aldrich: Crystal Violet (C.I. 42555), Basic Blue 7 (C.I. 42595), Ethyl violet (C.I. 42600), Basic Blue 11 (C.I. 44040), Basic red 2 (C.I. 50240), Basic Blue 17 (C.I. 52040). MALDI matrix 2,5-Dihydroxybenzoic acid (DHB, 99%) was purchased from Sigma–Aldrich. Ethanol (EtOH) and methanol (MeOH), both analytical grades, were from J.T. Baker (Center Valley, USA).

Suspensions and solutions

Stock suspensions and/or solutions of reference pigments in MeOH (1 mg mL^-1) were individually prepared. For mass calibration, a mixture containing six pigments at levels of 50 μg mL^-1 for C.I. 42555, C.I. 42595 and C.I. 42600, of 250 μg mL^-1 for C.I. 50240 and C.I. 52040 and of 100 μg mL^-1 for C.I. 44040 was prepared by mixing homogenous aliquots of each stock suspension and diluting with MeOH.

Samples and sample preparation

Tattoo inks and PMU were taken from the Swiss market over the last eight years. An overview of the analysed samples is given in Table 1.

Ink samples were shaken intensely by hand for about one minute. Then, about 25 µL (20–30 mg) of the suspension or 20 mg of PMU were weighed into a tared 2 mL-tube, mixed with 1 mL of EtOH, sonicated (5 min, 25°C) and centrifuged (5 min, 15000 rpm). The supernatant was removed and the residue containing the colour pigments was diluted with 1 mL of MeOH. After vortexing for some seconds, 1 µL duplicates of the homogenous suspension were immediately spotted onto a steel target plate, air dried at room temperature and used for LDI-TOF-MS.

Pigment verification with standard addition

20 µL of the methanolic ink suspension were transferred to a 96-Well reaction plate (MicroAmp Optical, Applied Biosystems) and mixed with 10 – 50 µL of the appropriate reference pigment suspension (10 mg mL^-1 MeOH). One microliter of the homogenous suspension was used for LDI-TOF-MS.

Improvement of detection using DHB as a matrix

20 µL of the methanolic ink suspension were transferred to the reaction plate, mixed with the same volume of DHB (10 mg mL^-1 MeOH) and 1 µL of the homogenous suspension was used for MALDI-TOF-MS.

MALDI-TOF-MS

Pigment mass spectra were obtained using a MALDI-TOF Mass Spectrometry Axima™ Confidence machine (Shimadzu-Biotech Corp., Kyoto, Japan) with detection in the reflectron mode with pulsed extraction (optimised at 450 Da) at a N₂-laser frequency of 50 Hz and within a mass range from 50 to 2000 Da. The transmission of the laser power was in the range of 60 – 100 units (33 – 55%). The target plate was scanned by the laser (diameter of 30 μm) in the regular rectangular mode and serpentine style (1000 × 1000 μm, spacing 166.666 μm and 49 points). A minimum of 20 laser shots was accumulated per profile and for each sample a total of 50 mass profiles was averaged and processed using Launchpad™ version 2.9.3 software (Shimadzu-Biotech Corp., Kyoto, Japan). This software was also used for peak processing of all raw spectra with the following settings: the advanced scenario was chosen from the parent peak clean up menu, peak width was set to 3 channels, smoothing filter width to 2 channels, baseline filter width to 10 channels, and the threshold apex was chosen as the peak detection method. The threshold apex peak detection was set as a dynamic type and the offset was set to 0.300 mV with a response factor of 1.0. The processed spectra were exported as peak lists with m/z values for each peak and signal intensity in the ASCII format. Calibration was conducted for each target plate using spectra of the reference standard mixture consisting of the following exact masses: 271.1137, 315.1609, 372.2439, 422.2596, 456.3378 and 478.3222 amu.

Identification of pigments with NIST library software

Exported ASCII files from MALDI-TOF-MS were converted to NIST software compatible peak lists with a mass resolution of 1 D (*.msp), and the identification of pigments was carried out with our home-made mass spectra library using the software of NIST 2.0 (Standard Reference Data Program of the National Institute of Standards and Technology, USA).

HPLC

More or less soluble colourants detected by LDI-TOF-MS were occasionally confirmed by high performance liquid chromatography (HPLC). Approximately 5 mg of tattoo or PMU samples were extracted with different solvents, starting with N,N-Dimethylformamide, 1-Chloronaphthalene and N-Methylpyrrolidone, 1 mL each. Coloured extracts were centrifuged at 12’000 g, filtered with 0.45 µm PTFE syringe filters and analysed with Ion Pair Reversed Phase HPLC with Ultraviolet Diode Array (UV/DAD) detection under the following conditions: Kromasil-column C18, 5 µm, 150 × 2 mm (30°C); eluent A: aqueous solution of dodecyltrimethylammonium bromide (3 g/L) and ammonium bromide (1 g/L), eluent B: ethanolic solution of dodecyltrimethylammonium bromide (3 g/L) and ammonium bromide (1 g/L); run time = 30 minutes; flow rate = 0.35 mL/min; gradient conditions: 0 min 45% eluent B, 2 min 55% eluent B, 10 min 65% eluent B, 20 min 100% eluent B, 24 min 100% eluent B, 24.1 min 45% eluent B.

Method overview

After clean-up, samples were screened with LDI-TOF-MS using the NIST program with our pigment library for identification. Tentatively identified pigments were then confirmed by comparing the high-resolution mass spectra of the native and the reference spiked sample. As a further check, the resulting colour of found pigments had to give the sample colour. Whenever the results of the analysis did not match with the colour of the samples, the ionisation was optimised by changing the laser power and/or adding DHB as a matrix to the sample.

Effect of ethanolic sample clean-up

Modern tattoo inks not only contain pigments but also binders, solvents, surfactants, preservatives and thickening agents¹⁵. Surfactants adsorb to the surface of pigment agglomerates and decrease the surface tension of the solvent. This eliminates residing air bubbles and improves particle coating with other additives. Adsorbed surfactants, however, can interfere with analysis; spiking C.I. 74265 (pigment green 36) with the non-ionic tenside Triton X-100 leads to a nearly complete suppression of mass signals (Figure 1). This problem was met by washing samples with EtOH before LDI-TOF-MS, which proved to be effective in most cases¹⁶.

Figure 1.

High resolution mass spectra of Pigment Green 36 (C.I. 74265) without any discrimination effect (A) and affected by the surfactant Triton X-100 (B).

Screening of samples with NIST software

With no commercial library of pigments available, we made our own library by exporting ASCII files of the mass spectra of analysed reference pigments to NIST compatible unit resolution mass peak lists. We preferred the NIST software for screening samples because its reversed match algorithm (R. Match) allows the simultaneous identification of several pigments in mixtures. This advantage is demonstrated for a red tattoo ink in Figure 2. Although the three existing pigments have overlapping ion clusters, all could be identified by their characteristic ions: C.I. 73915 (P.R. 122) with the ions [M+H]⁺ (m/z 341) and the sodium adduct [M+Na]⁺ (m/z 363), C.I. 21110 (P.O. 13) with the molecular ion [M+H]⁺ (m/z 623) and the two fragment ion clusters at m/z 437 and m/z 251 and C.I. 12475 (P.R. 170) with the ions [M+H]⁺ (m/z 455), [M+Na]⁺ (m/z 477) and the fragment ion m/z 318. As a further advantage of the NIST software compound specific information such as structural formula, molecular mass, synonyms, colour and CAS-no. can be added to the mass spectra library, and therefore also be used as search criterions. In the case of unknown mass spectra, the software can also extend the search to commercial GC-MS mass libraries (e.g. NIST main library).

Figure 2. Head to tail plot of three pigments identified in the mass spectra of a red tattoo ink sample by NIST software: C.I.73915 (Pigment Red 122, R. Match = 878), C.I.21110 (Pigment Orange 13, R. Match = 833) and C.I.12475 (Pigment Red 170, R. Match = 780).

Verification of pigments

Standard addition and high-resolution mass spectrometry were used for the verification of particularly interesting pigments tentatively identified by NIST. As a quality control standard addition also allows checking for possible discriminations caused by the sample matrix. Verification is shown for a violet coloured tattoo ink containing the prohibited C.I. 51319 (P.V. 23), which was not declared (Figure 3). After being tentatively identified in a first run, spiking the sample with C.I. 51319 (Figure 3B) showed the same typical molecular ion cluster [M]⁺ (m/z 588) and two fragment ion clusters with m/z 554 and m/z 520 as in the original sample (Figure 3A). Instead of C.I. 51319, the product label listed two legal pigments red C.I.12477 (P.R. 210) and blue C.I.74160 (P.B. 15) both non-present in the ink. LDI spectra did not reveal their characteristic ions [M+Na]⁺ (m/z 463) and [M+Na]⁺ (m/z 477) for the red pigment (a mixture of C.I. 12474 and 12475) and the ion cluster [M+H]⁺ (m/z 575) of the blue one (Figure 3C). This demonstrates a typical case of label forging for tattoo inks.

Figure 3.

Verification of the prohibited violet pigment C.I. 51319 in a tattoo ink sample by high resolution mass spectra (A) and standard addition (B), as well as the exclusion of a legal red (C.I. 12477) and blue (C.I. 74160) reference pigment (C) in the sample.

Signal enhancement using DHB as matrix

In the case of some 3,3’-dichlorobenzidine based diazodiaryl-pigments, only weak molecular ion mass signals ([M]⁺, [M+H]⁺, [M+Na]⁺) but more pronounced fragmentation were observed. In order to enhance molecular mass signals, three of the most often used MALDI- matrices, dihydroxybenzoic acid (DHB), α-cyano-4-hydroxycinnamic acid (CCA), and sinapic acid (SA), were added to ink samples containing dichlorobenzidine based pigments. It turned out that all three had a positive effect, with DHB showing the best signal enhancement. The effect of DHB on the mass spectra of yellow pigment C.I. 21090 is shown in Figure 4. With DHB the signal intensity of the molecular ion [M]⁺ (m/z 628) was twice that of the sodium adduct [M+Na]⁺ (m/z 651), even twelve times as high as without matrix supplement.

Figure 4.

LDI mass spectra of pigment Yellow 12 (C.I. 21090) without matrix (A) and MALDI spectra of the same pigment with DHB as a matrix (B).

Method validation

Because the identification of pigments by LDI-TOF-MS is a qualitative method, validation was only performed for mass resolution, accuracy of pigment identification, a rough estimation of limits of detection and the precision of characteristic mass intensity.

At the mass of 450 amu, a mass resolution of R = M/ΔM = 6000 (full width at half maximum, FWHM) was achieved. The resolution quality depends on the laser power and is not the same for all pigments. Generally, mass resolution decreases with higher laser power and a shift to a higher mass of up to 0.1 u can be observed.

Accuracy of pigment identification in the screening mode (NIST, without sample washing and spiking) was tested with 48 different tattoo ink samples, their composition being unknown to the analyst. The results were checked for conformity using trustworthy declarations and/or additional in-house HPLC analysis of more or less soluble colourants. In 44 of the 48 samples (92%) all organic pigments were identified correctly (up to four in the same sample). Incomplete detection was found in two inks where the banned orange pigment C.I. 12075 remained undetected due to its weak signals in the mass spectra. The resulting poor library search matches did not give sufficient evidence for a positive identification. With the characteristic ion [M+H]⁺ (m/z 339), this pigment is usually well detectable. Therefore, a pigment concentration near the limit of detection is the most likely explanation for the weak signal. For the two other samples, results of LDI-TOF-MS were not consistent with those of HPLC-analysis: in the first case, C.I. 21110 was identified with LDI-TOF-MS, whereas HPLC-analysis gave C.I. 21115. In the second case C.I. 21100 was mistaken for C.I. 21095 because the characteristic mass signals of C.I. 21100 (m/z 522 and 550) were quenched.

The limits of detection (LOD) in a blue tattoo ink were estimated by standard addition with two red pigments, of which one was a well (C.I. 56110) and the other a poorly ionisable one (C.I. 14720). A correct identification of the pigment specific mass clusters by the NIST library was possible between sample concentrations of 1% (w/w) for the well ionisable pigment and of 10 – 20% (w/w) for the other one (C.I. 14720). As pigments are added to tattoo inks in the percent range, screening with LDI-TOF-MS proved to have a suitable LOD.

Nevertheless, we have to consider that sometimes the LOD can also be affected by discrimination effects of tensides or inorganic pigments, such as carbon black (C.I. 77266) or TiO₂ (C.I. 77891, white), present in the sample¹⁶. Whereas tensides can be easily removed by EtOH the separation of the two inorganic pigments remain an unresolved problem up to this date. However, with the application of a higher laser energy and colour plausibility checks, negative identification results due to matrix effects could be eliminated for all forbidden organic pigments.

Precision was determined as repeatability with ten determinations of the characteristic mass intensity of two pigments (C.I. 21108, C.I. 74260) and two different inks containing C.I.74160 and C.I.12475, respectively. The pigment specific relative standard deviations were in a range of 4% to 13%, which are by far sufficient for identification purposes. This good reproducibility is in contrast to common experience with MALDI-TOF-MS as inhomogeneous crystallisation of the matrix promotes the building of so called “hot-spots” with locally higher ionisation signals¹⁷. Obviously, the dispersion of the pigments spotted on the target plate using MeOH instead of a matrix seems to give a more homogenous distribution of the analytes. Good reproducibility was probably also due to laser scanning in the serpentine style over the whole sample with 20 accumulated laser shots per profile which evens out some inhomogeneity.

Market survey

Over the last eight years we used LDI- or MALDI-TOF-MS for market survey purposes in Switzerland. Samples were randomly but also risk based collected from tattooing and PMU studios and from different importers. Table 1 gives an overview of the organic pigments identified in tattoo ink and PMU samples. Data obtained show that Pigment Blue 15 (C.I. 74160) was the most often found colourant followed by the prohibited Pigment Green 7 (C.I. 74260). All prohibited pigments, especially Pigment Green 7, were more frequently found in inks (3 – 13%) than in PMU (2 – 4%). In contrast to detected legal pigments where only 7% were missing on product labels of the samples, the prohibited pigments were often not declared (68%), for details see 18. This clearly gives evidence that label forging is widespread. This fraud considers pigments that are banned due to toxicological concerns, showcasing the urgency for widespread market controls.