Positive and negative-mode laser desorption/ionization-mass spectrometry (LDI-MS) for the detection of indigoids in archaeological purple

Authors


Correspondence to: Josefina Pérez-Arantegui, Environmental Sciences Institute (IUCA), Faculty of Sciences, University of Zaragoza, 50009 Zaragoza, Spain. E–mail: jparante@unizar.es

Abstract

Laser-based ionization techniques have demonstrated to be a valuable analytical tool to study organic pigments by mass spectrometric analyses. Though laser-based ionization techniques have identified several natural and synthetic organic dyes and pigments, they have never been used in the characterization of purple.

In this work, positive and negative-mode laser desorption/ionization mass spectrometry (LDI-MS) was used for the first time to detect indigoids in shellfish purple. The method was used to study organic residues collected from archaeological ceramic fragments that were known to contain purple, as determined by a classical high-performance liquid chromatography-based procedure.

LDI-MS provides a mass spectral fingerprint of shellfish purple, and it was found to be a rapid and successful tool for the identification of purple. In addition, a comparison between positive and negative mode ionization highlighted the complementarity of the two ionization modes. On the one hand, the negative-ion mode LDI-MS showed a better selectivity and sensitivity to brominated molecules, such as 6,6'-dibromoindigo, 6-monobromoindigo, 6,6'-dibromoindirubin, 6- and 6’-monobromoindirubin, thanks to their electronegativity, and produced simpler mass spectra. On the other hand, negative-ion mode LDI-MS was found to have a lower sensitivity to non-brominated compounds, such as indigo and indirubin, whose presence can be established in any case by collecting the complementary positive-ion LDI mass spectrum. Copyright © 2013 John Wiley & Sons, Ltd.

Introduction

In the last few years, laser-based ionization techniques, e.g. laser desorption/ionization (LDI), graphite-assisted LDI and matrix-assisted LDI (MALDI), have become popular and useful for mass spectrometric analyses of organic materials in the Cultural Heritage [1-3]. These techniques require negligible sample manipulation and no sample pre-treatment. This reduces problems such as contamination and sample destruction, which in some cases occur with wet-chemical procedures.

Laser-based ionization techniques only need a few nano/micrograms of sample, and just a few minutes for the analysis and data acquisition. In addition, the ‘soft’ mechanism that involves desorption/ionization of the analyte, without or with negligible fragmentation, enables thermally labile and no-volatile compounds, and polymeric substances to be analyzed and structurally characterized.

In the case of plant organic dyes and pigments, these techniques have been extensively employed by Wyplosz [4, 5]. For a wide range of traditional natural organic dyes in the form of plant extracts, lakes and dyestuff at the surface of dyed fibers, he obtained promising results and demonstrated that LDI provides abundant signals and detailed structural information without the use of a matrix (MALDI). More recently, LDI-mass spectrometry (LDI-MS) has been used to analyze Roman and Greco-Roman pink–reddish cosmetics [6, 7]. In all the cosmetics except one, the technique revealed the presence of alizarin, purpurin, munjistin and pseudopurpurin, madder characteristic compounds, in which the presence of a lead-containing matrix probably hid the peaks in the mass spectrum due to the madder. MALDI-MS has been also used to detect carminic acid, which is a characteristic compound of carmine lake, in a paint on paper [8] and in a linseed oil painting [9]. Kirby et al. [10] demonstrated the value of LDI-MS as an analytical tool to identify synthetic organic pigments in the investigation of painting techniques and materials used by Jackson Pollock and James Castle.

Though laser-based ionization techniques have identified several natural and synthetic organic dyes and pigments, they have never been used in the characterization of purple. In this study, for what we believe is the first time, LDI-MS was applied to study shellfish purple and to identify it in archaeological samples.

The main chromophores of shellfish purple, known also as Tyrian or royal purple, are brominated indigoids (6,6'-dibromoindigo and 6-monobromoindigo) and indirubins (6,6'-dibromoindirubin, 6- and 6’-monobromoindirubin), along with indigo and indirubin [11]. These compounds are not present in the mollusks, but are generated by enzymatic and photochemical reactions undergone by molecular precursors found in the hypobranchial glands of the mollusks [11]. A variety of analytical techniques have been employed in the identification of purple dyes in art works and archaeological objects [12-18]. However, the identification and quantification of indigoids and indirubins from purple are usually carried out by analytical methods based on high-performance liquid chromatography (HPLC) [19-22]. In any case, as has also recently been reported and discussed [22], HPLC analysis of purple compounds, especially if they contain bromine atoms, is not always straightforward, and wide, tailing and asymmetric chromatographic peaks may be obtained due to the low solubility of these compounds in commonly used mobile phases. To overcome these possible problems and to improve the solubility of purple, harsher conditions of the analysis have been proposed [22].

The aim of this paper is to evaluate the use of LDI-MS for rapidly identifying the compounds present in shellfish purple. We undertook this research to characterize the pigments on archaeological objects where minimal sampling and preparation are critical issues. The results were compared with those obtained by a well-established HPLC method.

Experimental

Instruments and methods

A Bruker MicroFlex matrix-assisted LDI time-of-flight mass spectrometer, equipped with a 337-nm pulsed nitrogen ultraviolet laser (20 Hz), was used for all the experiments. Samples were adsorbed on the surface of the microScout MALDI target plate by depositing a droplet of ethanol (0.5 µl) with discrete particles in suspension, with subsequent evaporation of the ethanol. Desorption and ionization was performed directly (LDI) without the assistance of a matrix. Mass spectra were obtained in the reflectron configuration with an accelerating voltage of −19 kV (negative-ion mode) or +19 kV (positive-ion mode). The laser intensity was set just above the ion-generation threshold to obtain peaks with the highest possible signal-to-noise ratio without significant peak broadening. The mass spectrometer was calibrated by the so-called ‘close external’ method, using dithranol in the negative mode (DIT, 1,8-dihydroxy-10H-anthracen-9-one, molecular weight: 226.2274 g/mol), and polyethylenglycol (PEG300) in DIT matrix in the positive mode. The spectra were obtained by pointing the laser at at least three different areas of the sample target to ensure that the results were consistent and representative.

The HPLC analytical procedure (modified slightly from [13, 20]) can be summarized as follows. The samples (roughly 1 mg) were dissolved in 200 µl of dimethylsulfoxide and heated at 60°C for 10 min in an ultrasonic bath. After dissolution, the samples were immediately filtered on a PTFE 0.2 µm syringe filter (Alltech) and analyzed using HPLC-diode array detection (DAD). The HPLC analyses were carried out using a gradient elution program (see Table 1). The eluents were acetonitrile with 0.1% trifluoroacetic acid (A) and water with 0.1% trifluoroacetic acid (B). The chromatographic column was thermostated at 40°C. The wavelength chosen for the compound detection was 275 nm, and the flow rate was 0.4 ml/min. The DAD spectra were recorded between 200 and 650 nm.

Table 1. Gradient of elution for HPLC analyses (A: acetonitrile with 0.1% trifluoroacetic acid; B: water with 0.1% trifluoroacetic acid)
Time (min)AB
04060
14060
58020
109010
159010
181000
251000
304060
404060

An HPLC was used consisting of a PU-2089 Quaternary Gradient Pump with degasser (Jasco International Co., Japan), equipped with a 20 µl Rheodyne Model 7125 injection valve and coupled to a spectrophotometric diode array detector MD-2010 (Jasco International Co., Japan). The data were processed using ChromNav® software. The chromatographic separation was performed using an analytical reverse phase C-18 column (Agilent TC-C18 II 4.6 × 250 mm, 5 mm, Agilent Technologies, US) connected to a C-18 pre-column (Agilent TC-C18, analytical guard column 4.6 × 12.5 mm, Agilent Technologies, US).

Samples

The two following reference materials were purchased from Kremer Pigmente GmbH & Co. KG (Aichstetten, Germany): 36000A Natural Indigo from India (powder, natural organic product from Indigofera tinctoria; analysis: indigotin (photometric) 46.2%, ash (DGF-M-V-4) 12.1%); and 36010 Tyrian Purple (natural 6,6'-dibromoindigo with impurities).

Archaeological samples consisted of residues of a pink–violet substance sampled from ceramic fragments (I-II c. AD) (Fig. 1). The fragments were discovered at the archaeological site of Sumhuram, the most important pre-Islamic harbor in the Khor Rori area (Oman), where the Italian Mission to Oman, directed by Prof. A. Avanzini (University of Pisa), has been working since 1997. Microsamples of these pink–violet residues were collected from the pottery with a scalpel and stored in glass vessels until the analysis.

Figure 1.

One of the ceramic fragments (I-II cent AD, Sumhuram, Oman) showing residues of a pink–violet substance.

Results and discussion

The samples were found to be positive to shellfish purple dyes using the HPLC method, described in the Experimental section and commonly used in our laboratory for the determination of indigoid compounds. HPLC-DAD enabled us to separate, identify and semi-quantify the following molecules present in all the archaeological samples: indigo, indirubin, 6-monobromoindigo, 6,6’-dibromoindigo and dibromoindirubin. Figure 2 shows the distribution of the indigoid molecules in the various samples (chromatographic peak areas were measured, and the data expressed as percentages of the total).

Figure 2.

The relative abundance distribution of the indigoid molecules in the various archaeological samples obtained by HPLC analysis.

In order to compare the LDI mass spectra of archaeological purple at a later stage, two reference materials (36010 Tyrian Purple and 36000A Natural Indigo) were also analyzed. The first was later used to reveal the LDI mass spectra of the indigo molecule and the second one as a reference of the Royal purple dye.

In the case of the indigo, the most relevant peak was at m/z 261 in the negative-ion LDI mass spectrum (Fig. 3a). The peak corresponded to the deprotonated indigo molecule, [In-H]-. In the positive-ion LDI mass spectrum (Fig. 3b), the most characteristic peaks were due to the radical ion [In] at m/z 262 and the group of sodium and potassium adducts, [In+Na]+ and [In+K]+, at 285 m/z and 301 m/z. The dominant peak at m/z 232 remained unidentified. According to Wyplozs’ studies [5], this ion at m/z 232 could be a marker of indigo of natural origin and could be ascribed to the presence of an additional compound or a degradation product.

Figure 3.

LDI mass spectra of natural indigo (Kremer 36000A), obtained in (a) negative mode and (b) positive mode.

The negative-ion LDI mass spectrum of the reference purple material (Fig. 4a) showed a very intense isotopic cluster peaking at m/z 419 and corresponded to the deprotonated 6,6'-dibromoindigo, [diBrIn-H]-. The isotopic cluster is due to the presence of two bromine atoms present on the molecule. The same isotopic cluster could also be ascribed to 6,6'-dibromoindirubin, an isomer of 6,6'-dibromoindigo. In fact, it should be noted that LDI-MS does not differentiate between the isomers present in shellfish purple, such as indigo and indirubin, monobromoindigo and monobromoindirubin, and 6,6'-dibromoindigo and 6,6’-dibromoindirubin. In the mass spectrum, another isotopic cluster peaking at m/z 339 and a peak m/z 261 corresponded to the deprotonated 6-monobromoindigo, [BrIn-H]-, and indigo (and/or its isomers), [In-H]-, respectively, are clearly visible. The positive-ion LDI mass spectrum also confirmed the presence of these components (Fig. 4b). Protonated [diBrIn+H]+ (m/z 421) and sodiated dibromoindigo [diBrIn+Na]+ (m/z 443) ions were detected, as the most intense peaks. The peak at m/z 341, [BrIn+H]+, showed a very low intensity, and no group of ions at m/z 262–263 due to indigo were detected. In both LDI spectra, other unknown peaks were also present, but with a very low intensity and probably arising from impurities.

Figure 4.

LDI mass spectra of 6,6'-dibromoindigo, with impurities (Kremer 36010), obtained in (a) negative mode and (b) positive mode. Insets show the isotopic distribution of peaks at m/z 339 and m/z 419.

The LDI mass spectra of the archaeological samples were more complex. The spectra are presented in Figs. 5 and 6, and the main peaks are summarized in Table 2. In the negative-ion spectra (Fig. 5), the most relevant peaks corresponded to the deprotonated dibromoindigo ion [diBrIn-H]-, isotopic cluster peaking at m/z 419, monobromoindigo [BrIn-H]-, isotopic cluster peaking at m/z 339, and indigo [In-H]-, peak at m/z 261, and/or their isomers. In some of the archaeological samples, the presence of deprotonated indigo ion [In-H]- (m/z 261) was not confirmed because the peak intensity was similar to the noise at this m/z value. The main peaks in the positive-ion LDI mass spectra (Fig. 6) corresponded to the protonated dibromoindigo [diBrIn+H]+, cluster peaking at m/z 421, monobromoindigo [BrIn+H]+, cluster peaking at m/z 341, and indigo [In+H]+, m/z 263 (and/or their isomers: monobromoindirubins and indirubin). The sodium and potassium adducts of these three molecules were also detected in the mass spectra, at m/z 443, m/z 459, m/z 363, m/z 379, m/z 285 and m/z 301, respectively.

Figure 5.

LDI mass spectra of archaeological samples obtained in negative mode.

Figure 6.

LDI mass spectra of archaeological samples obtained in positive mode.

Table 2. Peaks collected in the negative and positive-ion LDI mass spectra of the archaeological samples (in bold, more intense peaks)
SampleNegative-ion LDI spectrumPositive-ion LDI spectrum
 Peaks from identified ions (m/z)Peaks from unidentified ions (m/z)Peaks from identified ions (m/z)Peaks from unidentified ions (m/z)
19261223263219
 237285224
 247301235
339253341248
 260363264
  379279
419 421305
  -321
 461459348
390
429
492
538
572
27-223263219
 237285235
  301248
339 341 
  363279
  379305
419 421321
  443 
  459429
   539
30- 263 
 234285 
  301 
339 341 
  - 
 372- 
419 421 
  443373
  459393
37-223263219
 237-235
  301248
 339 -264
  -279
  -305
419 421321
  443348
  459375
 390
429
572
46261223263219
237285235
 247301248
339253341264
 259363279
  379305
419 421321
- 
459
43A- - 
  - 
  - 
339 341 
  - 
 372-305
419 421 
443 
459 
43B- 263 
 
  285 
  301 
339 341 
  363 
  379305
419 421321
  443
  459

The most characteristic peaks and their relative intensities grouped the samples into two different patterns. Samples 19, 27, 37 and 46 had more complex positive-ion LDI mass spectra, with a high number of peaks in common (see Table 2). The mass spectra recorded for samples 19 and 46 were especially similar (see Fig. 5); both negative-ion mass spectra showed peaks at m/z 223, m/z 237 and m/z 247, and a similar intensity proportion between peaks at m/z 339 and m/z 419. The second group included samples 30, 43A and 43B. In these samples, very clear LDI mass spectra were recorded (Figs. 5 and 6), highlighting the presence of the indigoid derivates responsible for the color of the Tyrian purple dye.

The differences between both subgroups of samples were mainly due to the unidentified peaks in the positive-ion mass spectra (m/z 219, m/z 235, m/z 248, m/z 264, m/z 279, m/z 305, m/z 321 …). These peaks were probably due to the presence of other organic components, either linked to the products used to prepare the dyes, or to origin of purple dyes or to ageing. For instance, a peak at m/z 248 has been previously found in mass spectra of aged indigo obtained by DTMS and ascribed to tryptanthrin, a degradation product of indigo [5, 23]. In fact, these peaks appeared especially higher in the samples where peaks at m/z 421 were smaller, as an indication of a possible debromination or degradation of the products.

In both types of LDI mass spectra, the proportion between the intensity of the main peaks highlighted a pattern in the relative proportions of the three possible kinds of molecules: dibrominated compounds (dibromoindigo and dibromoindirubin), monobrominated compounds (monobromoindigo and monobromoindirubins) and unbrominated compounds (indigo and indirubin), in each sample. This approximation provides a semiquantitative approach to determine the dye composition. For the calculations, each compound was represented by the sum intensity of all the peaks corresponding to the different m/z values of its isotopic cluster. Unbrominated compounds (called 0Br in Fig. 7) were described by the peaks at m/z 260, m/z 261 and m/z 262 in the negative-ion mass spectra, and by the peaks at m/z 262, m/z 263 and m/z 264 in the positive-ion mass spectra. Monobrominated molecules (called 1Br in Fig. 7) were represented by the peaks at m/z 337, m/z 338, m/z 339, m/z 340 and m/z 341 in the negative-ion mass spectra, and by the peaks at m/z 340, m/z 341, m/z 342, m/z 343 and m/z 344 in the positive-ion mass spectra. Dibrominated molecules (called 2Br in Fig. 7) were defined by the peaks at m/z 417, m/z 418, m/z 419, m/z 420, m/z 421 and m/z 422 in the negative-ion mass spectra, and by the peaks at m/z 418, m/z 419, m/z 420, m/z 421, m/z 422, m/z 423 and m/z 424 in the positive-ion mass spectra. The relative percentage of each compound was calculated as the ratio between the sum intensity of the compound and the total intensity of the three components.

Figure 7.

The relative abundance distributions of the unbrominated (0Br), monobrominated (1Br) and dibrominated molecules (2Br) in the various archaeological samples as obtained by LDI mass spectrometry in (a) negative mode and (b) positive mode.

If the negative-ion mass spectra were considered for the calculation (Fig. 7a), the relative proportion of dibromoindigo (or its isomer) seemed to be overestimated because the two other forms were sometimes found under the detection limits. When the proportions were calculated with peaks from the positive-ion LDI mass spectra (Fig. 7b), the semiquantitative distribution was similar to that obtained by HPLC-DAD (Fig. 2). In this second approach, the presence of the intense peak at m/z 264 in some samples (19, 37 and 46) caused a positive interference and could have led to the overestimation of the relative value of indigo (or its isomer). For this reason, the relative percentage of indigo and its isomer was finally calculated using only the intensity of the peak at m/z 263 in the positive-ion mass spectra (Fig. 7b). In both cases, using negative or positive-ion LDI mass spectra, the semiquantitative histogram highlighted the higher proportions of dibrominated compounds in samples 30, 37 and 43A.

Conclusions

We have reported the results of the exploitation of positive and negative-mode laser desorption/ionization MS in the analysis of indigoid compounds from shellfish purple. The technique provided a mass spectral fingerprint of purple and was found to be a rapid and successful tool. Thanks to LDI-MS sensitivity to purple characteristic molecules and, minimal sample requirement and destruction, it is very suitable for detecting purple in archaeological samples. In positive mode, ions are mainly in the form of protonated molecules and alkali adducts; in negative mode, deprotonated molecular ions predominate. In all cases, the negative-ion LDI mass spectra showed fewer peaks and highlighted a simpler pattern to confirm the presence of mono- and dibrominated molecules from the shellfish purple dye.

In fact, LDI-MS in negative-ion mode enabled us to obtain a better selectivity and sensitivity to the brominated compounds thanks to their electronegativity. On the one hand, the analysis of only negative ions acted as a filter; on the other, it enabled us to observe fewer peaks and to obtain less noise. In some samples, the method was not able to detect negative ions from indigo and indirubin. However, the presence of these unbrominated compounds could be confirmed by collecting the complementary positive-ion LDI mass spectrum on the same sample.

Acknowledgements

The authors would like to thank Dr. A. Pavan and Prof. A. Avanzini of the University of Pisa (Italy) for providing the archaeological samples. The study was financially supported by the research project CTQ2011-24882 (Spanish Ministry of Science and Innovation) and by the project Azioni Integrate Italia-Spagna 2010, IT109GB8AA (Italian Ministry of Education, University and Research).

Ancillary