This item has the following downloads:
1 LASER DESORPTION/IONIZATION TANDEM MASS SPECTROME TR Y OF ANTHRAQUINONE DYES AND LEAD WHITE PIGMENT FOR PAINTED WORKS OF ART By MICHAEL PATRICK NAPOLITANO A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF F LORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 201 3
2 201 3 Michael Patrick Napolitano
3 To all the educators in my life who have inspired me and nurtured my love of empiricism, sc ience, and knowledge; and to the preservation and continuance of the culture and heritage of the people of the Occident
4 ACKNOWLEDGMENTS Certainly, I must first humbly extend my most sincere gratitude to my research advisor, Dr. Richard Yost. The comp letion of my degree would not have been possible without his unwavering support. As my life as a chemist develops, I shall hold inviolate his advice and hope to demonstrate that his efforts have not been in vain. I am also very grateful for the advice fr om and friendship of Dr. James Horvath. During our many discussions in his office, he instilled in me the sense of forthrightness and pride an educator must have. I thank my committee members Dr. John Eyler, Dr. Leonid Moroz, Dr. Nicolo Omenetto, and Dr. David Powell for their patience and guidance. Thanks are also given to Julie Arlsanoglu, Yelena Bobkova Dr. Phil Brucat, Dr. Mari Prieto Conaway, Dr. Ron Heeren, Dr. Jodie Johnson, Andras Kiss, Dr. Lennaert Klerk Dr. Katrien Kuene, Dr. Ping Chung Kuo, Dr. Ben Smith, and Dr. Don Smith for all of their assistance. I thank all the fellow students and friends that have made my time at UF memorable and have also provided stimulating, scientific discussions including, in alphabetical order, Dr. Dodge Baluya, Dr. Stacey Benjamin, Dr. John Bowden, Dr. Tim Garrett, Dr. Fabrizio Guzzetta, Chris Hilton, Dr. Lloyd Horne, Dr. Kaan Kececi Antoinette Knight, Dr. Rachelle Landgraf, Hillary Lathrop, Jessica Leigh, D r. Dan Magparangalan Dr. Antonio Masello, Dr. Ro b Meng er, Funda Mira, Dr. Giovennella Moscovici, Dr. David Pirman, Karla Radke, Dr. Rich Reich Dr. Dav e Richardson, Anna Sberegaeva, Dr. Dosung Sohn, Dr. Jen nifer Garrett Williams and Dr. Alex Wu ; with particularly special attention to Dominic Colosi, Dr. Fran k Kero, Whitney Stutts and Dr. Mari lyn Prieto Tourn. Thanks to the two NSF REU students, Vivian Estavam Cornlio
5 and Jennifer Webber, who provided both assistance to my research and a platform to hone my mentoring skills. Finally, I want to thank my pa rents, family, and friends back home in New Jersey. Their love and constant, unfaltering enthusiasm and support have sustained me during my seemingly unending pursuit of higher education.
6 TABLE OF CONTENTS page ACKNOWLED GMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 8 LIST OF FIGURES ................................ ................................ ................................ .......... 9 ABSTRACT ................................ ................................ ................................ ................... 12 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 15 Cultural Heritage ................................ ................................ ................................ ..... 15 Conservation Science ................................ ................................ ............................. 16 Archaeometry ................................ ................................ ................................ ... 16 History of Conservation Science ................................ ................................ ...... 17 Methods of Ana lysis ................................ ................................ ......................... 18 Instrumentation ................................ ................................ ................................ ....... 20 Laser Desorption/Ionization ................................ ................................ .............. 20 Matri x Assisted Laser Desorption/Ionization ................................ .................... 22 Electrospray Ionization ................................ ................................ ..................... 23 Linear Quadrupole Ion Trap ................................ ................................ ............. 25 Orbitrap ................................ ................................ ................................ ............ 26 Figures of Merit ................................ ................................ ................................ 28 Overview of Dissertation ................................ ................................ ......................... 30 2 TANDEM MASS SPECTROMETRY OF ANTHRAQUINONES ................................ 34 Background ................................ ................................ ................................ ............. 34 Experimental Methods ................................ ................................ ............................ 42 Chemicals and Materials ................................ ................................ .................. 42 Recrystallization of Standards ................................ ................................ .......... 42 Preparation of Chemicals ................................ ................................ ................. 43 Electrospray Ionization Parameters ................................ ................................ .. 44 Laser Desorption/Ionization and Matrix Assisted Laser Desorption/Ionization Parameters ................................ ................................ 45 Ultraviolet visible (UV) Light Exposure ................................ ............................ 47 Results and Discussion ................................ ................................ ........................... 47 Electrospray Ionization of Anthraquinones ................................ ....................... 48 Laser Desorption/Ionization of Anthraquinones ................................ ................ 54 Matrix Assisted Laser Desorption/Ionization of Anthraquinones ...................... 61 Conclusion ................................ ................................ ................................ .............. 64
7 3 TANDEM MASS SPECTROMETRY OF CLUSTERS FROM LEAD WHITE .......... 89 Background ................................ ................................ ................................ ............. 89 Experimental Methods ................................ ................................ .......................... 105 Chemicals and Materials ................................ ................................ ................ 105 Preparation of Chemicals ................................ ................................ ............... 105 Ionization and Instrumental Parameters ................................ ......................... 105 Results and Discussion ................................ ................................ ......................... 107 Full Scan Spectra Analysis ................................ ................................ ............. 107 Tandem Mass Spectrometric Analysis ................................ ........................... 112 Final Mass Assignments ................................ ................................ ................. 115 Conclusion ................................ ................................ ................................ ............ 117 4 LASER DESORPTION/IONIZATION TANDEM MASS SPECT ROMETRY OF MADDER AND LEAD WHITE DIRECTLY FROM ARTISTIC SAMPLES .............. 136 Background ................................ ................................ ................................ ........... 136 Experimental Methods ................................ ................................ .......................... 139 Samples: Painting Fragments and Dyed Silk Swatches ................................ 139 Laser Desorption/Ionization Parameters ................................ ........................ 140 Instrumental Parameters ................................ ................................ ................ 141 Results and Discussion ................................ ................................ ......................... 141 In Situ Detection of Alizarin from Painting Samples ................................ ....... 141 In Situ Detection of Madder from Swatches of Dyed Silk ............................... 142 In Situ Detection of Lead White from Painting Samples ................................ 145 Conclusion ................................ ................................ ................................ ............ 146 5 CONCLUSION AND FUTURE DIRECTIONS ................................ ....................... 161 Conclusions ................................ ................................ ................................ .......... 161 Future Directions ................................ ................................ ................................ .. 163 REFERENCES ................................ ................................ ................................ ............ 167 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 177
8 LIST OF TABLES Table page 2 1 Summary of results for the spectral abundances of all anthraquinones analyzed. ................................ ................................ ................................ ............ 83 2 2 Abundance of P as a percentage of the total abundance of the P, P+1, P+2 envelope for both LDI and MALDI. ................................ ................................ ..... 86 3 1 Most abundant and monoisotopic masses of cluster ions from th e LTQ analyzed spectrum with the normal scan rate. ................................ ......... 123 3 2 Most abundant and monoisotopic masses of cluster ions from the LTQ analyzed spectrum with the normal enhanced rate. .......................... ...... 124 3 3 Most abundant and monoisotopic masses of cluster ions from the Orbitrap analyzed spectrum. ................................ ................................ ............ 125 3 4 LTQ analyzed daughter ions following MS/MS of pa rent ion clusters .............. 132 3 5 Orbitrap analyzed daughter ions following MS/MS of parent ion clusters ........ 133 3 6 Final elucid ation of observed cluster ions of lead white. ................................ ... 134
9 LIST OF FIGURES Figure page 1 1 Schematic of the linear quadrupole ion trap ................................ ...................... 32 1 2 Diagram of the Thermo LTQ Orbitrap. ................................ ................................ 33 2 1 Structures of all anthraquinones investigated and the carbon numbering scheme for anthraquinone. ................................ ................................ ................. 66 2 2 Optimization of isolation width 1,2 dihydroxyanthraquinone during ESI MS/MS on an LCQ. ................................ ................................ ..................... 67 2 3 Optimization of CID energy for the m/z 241 ion of 1,2 dihydroxyanthraquinone during ESI MS/MS on an LCQ. ............................. 68 2 4 The mass spectra of 1,2 dihydroxyanthraquinone (alizarin). .............................. 69 2 5 The tandem mass spectra of 1,2 dihydroxyanthraquinone (alizarin). ................. 70 2 6 The mass spectra of 1,2,4 triydroxyanthraquinone (purpurin). ........................... 71 2 7 The tandem mass spectra of 1,2,4 triydroxyanthraquinone (purpurin). .............. 7 2 2 8 The mass spectra of 1,5 dihydroxyanthraquinone. ................................ ............. 73 2 9 The tandem mass spectra of 1,5 dihydroxyanthraquinone. ................................ 74 2 10 The mass spectra of 2,6 dihydroxyanthraquinone. ................................ ............. 75 2 11 The tandem mass spectra of 2,6 dihydroxyanthraquinone. ................................ 76 2 12 The mass spectra of 1,5 diaminoanthraquinone. ................................ ................ 77 2 13 The tandem mass spectra of 1,5 diaminoanthraquinone. ................................ ... 78 2 14 The mass spectra of 2,6 diaminoanthraquinone. ................................ ................ 79 2 1 5 The tandem mass spectra of 2,6 diaminoanthraquinone. ................................ ... 80 2 16 The mass spectra of anthraquinone. ................................ ................................ .. 81 2 17 The tandem mass spectr a of anthraquinone. ................................ ..................... 82 2 18 ESI+ mass spectra of anthraquinone after exposure to n ambient light. ...... ........ 84 2 19 UV vis spectrographs of anthraquinone with and without exposure to UV light. ................................ ................................ ................................ .................... 85
10 2 20 Extents of reduction under both MALDI and LDI conditions as a function of analyte concentration for alizarin and anthraquinone ................................ ....... 87 2 21 Plot of reduction extent as a function of matrix to analyte ratio for alizarin and anthraquinone. ................................ ................................ ................. ........... 88 3 1 Theoretical is otopic distribution pattern for lead white ((PbCO 3 ) 2 2 ) ..... 119 3 2 L DI generated, full scan, positive mode spectrum of lead white analyzed by LTQ with normal and e nhanced scan rates and Orb itrap ...................... ......... 120 3 3 Theoretical and experimental capital clusters.. ................................ ................. 121 3 4 Isotopic pattern matching for the cluster with the monoi sotopic mass of 706. .. 126 3 5 LTQ analyzed daughter ion clusters from MS 2 of the cluster with the monoisotopic mass of 432 Da. ................................ ................................ ......... 127 3 6 Daughter ion clusters from MS 2 of the cluster with the monoisotopic mass of 691 Da. ................................ ................................ ................................ ............. 128 3 7 Daughter ion clusters from MS 2 of the cluster with the monoisotopic mass of 897 Da. ................................ ................................ ................................ ............. 129 3 8 Daughter ion clusters from MS 2 of the cluster with the monoisotopic mass of 1139 Da. ................................ ................................ ................................ ........... 130 3 9 Daughter ion clusters from MS 2 of the cluster with the monoisotopic mass of 1363 Da. ................................ ................................ ................................ ........... 131 3 10 Isotopic pattern matching for the cluster with the monoisotopic mass of 1363. 135 4 1 Paint ing samples containing alizarin Silk swatches dyed w ith alizarin Painting samples containing lead white. ................................ ................ .......... 147 4 2 Painting sample I: LDI mass spectrum and zoomed regio n for alizari n. .......... 148 4 3 Painting sample II: LDI mass spectrum and zoomed region for alizarin ......... 149 4 4 Tandem mass spectra from sample I of supposed alizarin at the location for the [M+H] + and [M+2H ] + ................................ ................................ ........ ........ 150 4 5 Tandem mass spectra from sample II of supposed alizarin at the location for the [M+H] + and [M+2H ] + ................................ ................................ ......... ...... 151 4 6 Silk sample III: LDI mass s pectrum and zoomed region for alizarin ........ ....... 152 4 7 Silk sample IV: LDI mass spectrum and zoomed reg ion for alizarin ....... ........ 153
11 4 8 Tandem mass spectra from sample III (silk) of supposed alizarin ............ ...... 154 4 9 Tandem mass spectra from sa mple IV (silk) of supposed alizarin ............ ...... 155 4 10 Tandem mass spectra from sample III of supposed purpurin .................. ....... 156 4 11 Tandem mass s pectra from sample IV of supposed purpurin. .................. ...... 157 4 12 Painting sample V: LDI mass spectrum which shows lead white in the sample. ................................ ................................ ................................ ...... ...... 158 4 13 Painting sample VI: LDI full scan, positive mode mass spectrum which shows lead white in the sample. ................................ ................................ ..... 159 4 14 Painting sample VII: LDI full scan, positive mode mass spectrum, whi ch shows lead white in the sample. ................................ .................. .................... 160
12 A bstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy LASER DESORP TION/IONIZATION TANDEM MASS SPECTROMET R Y OF ANTHRAQUINONE DYES AND LEAD WHITE PIGMENT FOR PAINTED WORKS OF ART By Michael Patrick Napolitano August 2013 Chair: Richard A. Yost Major: Chemistry The goal of sustaining cultural heritage is manifested by diagnosing and preserving objects of historical and corporeal significance and artistic beauty. To achieve this goal conservation scientists develop and use methods commensurate with new technologies of traditional analytical chemists. For instance, gas chromatography/mass spectrometry (GC/MS) has been extensively used for many years for the characterization of oil painting components ; h owever, th is method often requires derivatization, cannot be used for direct surface analyses and, most importantly, i s destructive Consequently laser desorption/ionization (LDI) and matrix assisted laser desorption/ionization (MALDI) MS have been gaining popularity and interest among the conservation community for their non destructiveness and ability to interrogate i ntact surfaces. Furthermore, the single quadrupole mass analyzer of most GC/MS instruments and the time of flight (ToF) mass analyzer of most LDI and MALDI instruments lack the capability of tandem MS (MS/MS and in general MS n ),
13 which is a powerful attrib ute available on a linear quadrupole ion trap (LIT) to provide unambiguous structural identification of known and unknown dye and pigment components of a painted work of art. This work will present three projects for the analysis of painted works of art th at use d both LDI and MALDI MS on an LIT and Orbitrap which have been conducted in collaboration with The Metropolitan Museum of Art, New York. The first project examine d the peculiar ionization of anthraquinone dyes such as alizarin For instance when ionized by either LDI or MALDI, alizarin (MW = 240) exhibit ed a dominant ion of [M+2H] + at m/z 242 with a far greater abundance than the expected ion of [M+H] + at m/z 241. For the first time, MS 2 analysis of these anomalous [M+2H] + ions suggest ed a laser induced reduction of one of the anthraquinone s carbonyl groups as well as different relative abundances of neutral losses from either water or ammonia, depending The second project involved the elucidation of formulae from a suite of (PbCO 3 ) 2 Pb(OH) 2 Using LDI on both an LIT and Orbitrap, the low and high resolution full and tandem mass spectra were used in concert with deductive and inductive reasoning to decipher the peculiar spectra. The results are reported for the first time ubiquitous since antiquity. The third project empl oyed the results from the above two projects to directly detect the red anthraquinone dyes alizarin and purpurin and the white pigment lead white in various artistic samples. For the first time, the LDI MS 2 detection of alizarin and
14 madder was achieved in both paintings and textiles and the LDI MS detection of lead white was achieved in paintings, all without any sample pretreatment whatsoever. The results described herein provide new insight into a unique ionization phenomenon found in a specific class o f dye molecules, have elucidated the isotopically complex cluster ions formed from a common pigment, and have shown the detection of those dyes and pigment in artistic samples. Moreover, these results offer new avenues for the conservation science communi ty to diagnose and ultimately preserve painted works of art for the defense of cultural heritage.
15 C HAPTER 1 INTRODUCTION Cultural Heritage A proper grammar on the notion of cultural heritage and the science applied to that notion must first be establish ed before any meaningful logic or rhetoric may ensue. After all, the term heritage may have difficulty being either accurately or precisely defined in the lexicon of either scientists or the general population The United Nations Educational, Scientific and Cultural Organization (UNESCO) established a convention in destruction, need for economic and scientific investment, and requisite of diffused and increased knowledge thereof. 1 heritage to include monuments, such as works of architecture, monumental sculpture and painting, inscriptions, and cave dwellings; groups of buildings; and sites; which are of outs tanding universal value from the point of view of history, art, or science. However structures, with possible political or economic overtures, and never mentions painted wor ks of art. Cultural heritage was succinctly defined by Ciliberto in the introductory chapter of his excellent text as the whole of human cultural patrimony, which is everything that refers to the history of civilization and may include all works and docum ents that are of value from an archaeological, historical, or artistic perspective. 2 For the purpose of this dissertation, cultural heritage will be exemplified by painted works of art whose appreciation is embraced by the public and whose need for analysis is growing among archaeometrists, conservation scientists, and analytical chemists.
16 Conservation Science The orthodoxy of contemporary science regrettably perpetuates the atomization of fields to the unfortunate exclusion of compre hensive disciplinary approaches and the hierarchical stratification of fields and subfields. Such hypercategorization of fields may be exemplified by the analysis of painted works of art presented in this dissertation, which may simultaneously be describe d broadly as materials science and specifically as diagnostic conservation science To understand this seemingly disparate nomenclature, its grammatical and historical context must first be addressed. Archaeometry Conservation science is defined in t text and it is traced back to the broader fields whence it came. 3 considered to be the highest strata, a field so vast that a concise definition may not be readily applied. Thereafter, archaeomet ry follows, which may be defined as the 4 Archaeometry may be a sufficient term to describe the majority of related work conducted under its umbrella; nevertheless, it is then divided into 1) archaeology and 2) conservation science understand past societies from their surviving cultural 3 which is accomplished by dating via physical and chemical methods 4 and studying how artifacts were obtained 4 used 4 traded 3 and distributed 3 One common misconception is that archaeology is restricted to the realm of qualitative methodologies, but dating methods, such as the exploitation of carbon 14 dating achieved by accelerator mass spectrometry (AMS), certainly negate that notion.
17 Conservation science is used to diagnose, authenticate, and preserve artifacts, objects, and works of art for the purposes of curatorship, art history, and museum conservation and restoration, thereby making it distinct from archaeology. 3 Artioli cleverly describes conservation science as acting with respect to a function of time, tha t 3 Consideration must be given to via degradation processes. 3 Conservation science is practically employed by dating, characterizing (analyzing), and establishing provenance and original usage of materials and objects, and consequently identifying and preventing degradation of materials and objects. 2,3 Overall, archaeology may be considered as the attempt to understand the ext, whereas conservation science may be used to operate on an object for analyses and preservation. History of Conservation Science Though conservation science has recently been steadily gaining recognition as an independently viable and legitimate field when considered separately from the distinction of its field label, it may be seen as simply the application of chemistry and other natural sciences that have been conducted since the early days of modern science, as detailed in a interesting survey by C aley 5 For instance, Klaproth determined the metallic composition of Greek and Roman coins using gravimetry ca 1795, Chaptal published the first ever qualitative analysis of the chemical compositio n of ancient pigments in 1809, Humphry Davy examined ancient pigments of Rome and Pompeii in 1815, Faraday studied for the first time the use of lead in glazes of Roman pottery ca 1867, and Kekul analyzed ancient samples of wood tar that may have contain ed benzene derived compounds. 4,5 Moreover, not until the e arly 1950s was the
18 term archaeometry was first coined, with an eponymous journal launched in 1958. 3,4 The Journal of Archaeologi cal Science was started in 1974, but it should be noted that articles on conservation science are often found in traditional journals of analytical chemistry. Though conservation science might be perceived as an ancillary research field, it has long been above. The field has taken root in several museums and some universities; independent departments for conservation science are commonly housed in the former, but only rarely in the lat ter. Many of the most notable conservation science laboratories the United States with dedicated conservation science departments or laboratories are (with the year a depar tment or laboratory was established, if known): The Museum of Fine Arts Boston (1930s), The National Gallery of Art (1950), The Smithsonian Institution (1963), The J. Paul Getty Museum (1985), The Art Institute of Chicago (2003), The Indianapolis Museum of Art (2008), Winterthur, and The Metropolitan Museum of Art. By far, the most prolific and well known department is the Getty Conservation Institute at the Getty Museum. Some examples of U.S. colleges or universities with academic departments in conserva tion are The University of Delaware, Buffalo State College, Scripps College, and New York University, though they primarily teach preservation and not scientific analysis. Methods of Analysis As noted above, diagnostic conservation science may be consider ed subordinate to materials science and, more relevant to this dissertation, analytical chemistry. Therefore, virtually all analytical instrumentation and methodologies can be and
19 indeed have already been employed for the chemical diagnosis of works of ar t and other artifacts and objects. To provide a historical or literature review of the various types of instrumentation and their myriad range of applications would be unnecessarily cumbersome. Instead, the reader is directed to the introductory sections in Chapter s 2, 3, and 4 of this dissertation for thorough reviews of the analytical methods relevant to the topics in those Chapter s. Critical to the analysis of objects, artifacts, and works of art is the extent of destructiveness. Three categories of destructiveness are considered, which are, in order of increa sing degree of destructiveness: non invasive, non destructive, and micro destructive. 2 Non invasive te chniques require no sampling of the object and leave the object unalter ed; examples include micro X ray fluorescence on a medieval painted wooden reliquary 6 and micro Raman spectroscopy for an overpainted reproduction of a Byzantine icon 7 Non destructive techniques require sampling of the object, yet the sample can be analyzed by succeedin g methods; examples include laser ablation inductively coupled plasma MS of ancient African glass beads 8 and time of flight (ToF) secondary ion MS of ancient Tuscan ceramics 9 Micro destructive techniqu es require m inimal sampling of the object and completely destroy or consume the sample; examples include pyrolysis gas chromatography/MS of the coating on an Oriental lacquered wooden dish 10 and direct temperature resolved MS of organic residues in Roman er a pottery 11 Laser desorption/ionizati on MS is an example of a non destructive technique, and will be thoroughly reviewed in the following Chapter s, particularly Chapter 4. It is the primary means of analysis for the work presented in this dissertation and its theory and instrumentation shall henceforth be explained.
20 Instrumentation A mass spectrometer may be conceived to be primarily constructed of three main, sequential components: 1) ionization source, 2) mass analyzer, and 3) detector. Early mass spectrometers displayed and recorded spec tra on a phosphorescent screen or a photographic plate. Since the adoption of the microprocessor and the personal computer, it is quite appropriate to consider a computational/processing unit as a fourth component. The research presented in this disserta tion may be classified as applied mass spectrometry that ventures into the realm of fundamentals, particularly those related to ionization. Therefore, the following section shall be presented in a manner proportional to that of the research. For instance topics on ionization will include laser desorption/ionization (LDI), matrix assisted laser desorption/ionization (MALDI), and electrospray ionization (ESI); topics on mass analyzers will include the linear quadrupole ion trap (LIT), and orbital trap (Orb itrap). For thorough reviews of the literature regarding the precedent of research and historical perspective of the instrumentation used that is specific to the three projects in this dissertation, the reader is directed to the introductory sections of t he Chapter s for those respective projects. Laser Desorption/Ionization Soon after the laser was developed in the late 1950s, it was first employed as an ionization source for a (double focusing) mass spectrometer in 1963 by Honig and Woolston 12 Although the instrument suffered from low spectral resolution from space charge effects and low (ppm) sensitivity (that is, to current standard), it w as able to detect singly charged ions from metals, semiconductors, and insulators. Using an arc discharge just after the sample stage, the instrument was, in effect, also the first to
21 produce singly and doubly charged ions with both laser ablation and ele ctron ionization (EI). In 1966, a laser was first coupled to a time of flight (ToF) mass analyzer, which was used for the analysis of metal foils and organic compounds, though only atoms, not molecular ions, were observed from the latter. 13 The mechanism of ionization was considered to be a result of thermionic electrons being released following the laser heating the sample, a theory repeated 14 without proof for many years afterwards. Interestingly, the authors a feature still well known to present day users of an LDI source. The first successful generation of intact i ons using LDI (with a ToF) from organic compounds was reported in 1970. 15 LDI generated molecular ions are typically comprised of either odd electron radica l ions ([M ] + ) or even electron ions ([M+H] + or [M+Z] + where Z = metal cation). The prior developments reached a practical conclusion with the 1975 introduction of the laser microprobe mass analyzer (LAMMA) by Hillenkamp and Kaufmann, which was able to a chieve both high spatial resolution (0.5 m) through the simple use of a microscope objective to focus the laser beam and slightly increased sensitivity (0.4 ppm). 16 commercialized laser based mass spectrometer, the LAMMA 500, an d was primarily used for probing thin sliced tissue sections that were embedded in a matrix of polymer resin. 14,17 Indeed, it that contributed to the discovery of the MALDI principle. 17
22 M atrix A ssisted L aser D esorption/ I onization form gas phase molecular ions with neither excessive fragmentation nor p rior evaporation. 17,18 1969 19 ; plasma desorption (PD), developed in 1974 20 ; static secondary ion MS (SIMS), developed in 1978 21 ; and fast atom bombardment (FAB), developed in 1981 22 Much like the polymer resin used with LAMMA, the glycerol matrix used with FAB also contributed to the MALDI principle. 17 The significant difference with MALDI compared to the just mentioned ionization methods, is relatively soft ionization event that minimizes source fragmentation, creat ing primarily singly charged ions and allow ing ionization of (biological) molecules in excess of 100,000 Da. 17 photons from a laser that are absorbed by an infrared or ultraviolet (UV) absorbing matrix, typically a n organic acid of low molecular weight; two of the most commonly used are 2,5 dihydroxybenzoic acid ( DHB ) cyano 4 hydroxycinnamic acid (CHCA). Yet, t he success of a molecule as a matrix is not solely dependant on its ability to absorb UV light, as c rystallization properties and gas phase acidity also play significant roles. 23 Crucial to the desorption process is the effective co crystallization of the analyte with excess matrix. When photons bombard the sample surface, through the intermediary ionization of the matrix an ablation plume forms that concurrently desorbs and ionizes the analyte. Interestingly, the exact ionization pathways are still not completely known, though there are two currently recognized theories : energy pooling 24 and excited state proton transfer 25 Energy pooling is theorized to occur when multiple matrix molecules, in the excited state after irradiation, combine their internal energy to
23 form one matrix radical ion (M + ) which subsequently ionizes an analyte molecule in the gas phase by charge transfer 24 Excited state proton transfer is theor ized to occur when a single photon is absorbed to excite one matrix molecule which then reacts with a ground state matrix molecule to transfer a proton forming an [ M + H ] + ion of the matrix 25 This protonated matrix molecule subsequently protonates an analyte molecule in the gas phase. 26 As in LDI, MA LDI generated molecular ions are typically comprised of either odd electron radical ions ([M ] + ) or even electron ions ([M+H] + or [M+Z] + where Z = metal cation). Electrospray I onization Alongside MALDI, electrospray ionization, whose d evelopment in 1989 is credited to Fenn 27 is a soft ionization method and h as emerged as the premier ionization method for the analysis of large molecules, particularly peptides and proteins. 28 ESI has also earned popularity for features such as relative ease of use, high sensitivity, compatibility with liquid chromatographic methods, and allowance for extension of the ionizable mass range, the last feature being a consequence of multiply charged an alytes, which permits high mass ions to be detected at lower mass to charge ( m/z ) values. 29 An electrospray interface may appear rudimentary in design, yet the physics of droplet formation and charge transfer are quite involved. Briefly, a solv ated analyte is pumped through capillary tubing to the ESI needle, which is kept at a high potential kV) relative to a counter electrode. It is this applied potential on the needle, and ultimately on the analyte by proxy of the solvent, that is at the heart of electrospray. As
24 charge applied to the needle is transferred to those droplets as they are propelled toward the entrance capillary via carrier gas. 30 Then, as droplets travel toward the inlet of the mass analyzer, the solvent evaporates, rapidly decreasing the volume of the droplets. Despite th e decrease in volume, and concurrent decrease in surface area, the amount of charge on the droplet remains constant. Eventually, a point is reached where the force of charge repulsion overcomes the force of surface tension of the solvent; smaller droplets. 30,31 As this process repeats, the droplets become small enough that only one or a few ions remain, which are then transported through the ion optics toward the mass analyzer. Two competing theories offering explanations of the final ion formation have been proposed. 32 The first proposed theory is the charged residue model (CRM), which supports the mechanism of continual droplet size reduction until the droplets are ultimately small enough to contain only a single ion, yet retain mul tiple charges. 33 The second theory, ion evaporation mechanism (IEM), also assumes a decrease in droplet size, yet it proposes that the electric field from the charge potential is great enough as the droplet decreases in diameter to eject an ion out of the droplet. 34 This ejected ion, free from solvent is then carried through to the mass analyzer. For many years, these two theories have had their fair share of detractors and supporters but m ost ESI experiments seem to support the CRM model over the IEM model. 32 ESI generated molecular ions are typically comprised of even electron ions ([M+ n H] n + or [M+ m Y + n Z] n + where Y = mobile phase adduct and Z = cation).
25 Linear Q uadrupole I on T rap The primary mass analyzer used in this dissertation is the two dimensional linear quadrupole ion trap. The LIT is a direct descendant of the three dimensional version developed in 1953 by Wolfgang Paul, for which he was awarded the 1989 Nobel Prize in Physics. 35 The current 2 D version offers impressive enhancements including extension of the analytical mass to charge ( m/z ) range and increases in trapping efficiency, space charge limit, ejection rate and sensitivity. 36,37 The 2 D trap is constructed of four radially symmetric hyperbolic rods split into th ree sections as shown in Figure 1 1A 36 The end sections are 12 mm in length and the center section is 37 mm in length with 30 mm 0.25 mm ejection slit s in two of the rods The two y axis rods are spaced 8 mm apart, while the two x ax is rods (with the slits) have an additional 0.75 mm stretch to correct for the field imperfections arising from the slit, much as the spacing of the endcaps of the 3 D trap was stretched to corre ct for the holes in the endcaps 38 Different voltages are applied to the various rod sections of the trap as shown in F igure 1 1, parts B, C, and D 36 The pr imary alternating current (AC) voltage responsible for trapping is applied with matching polarity to opposing rods and opposite polarity to adjacent rods, typically 5 kV at 1 MHz This voltage, when applied to the hyperbolic rods of the specific geometry creates a quadrupolar electric field capable of trapping ions which may be located in a z and q z space according to the solutions given for the general form of the Mathieu equation 38 which are provided in Equation s 1 1 & 1 2 (1 1)
26 (1 2) where a z is the dimensionless parameter of stability, e is the charge of an electron, U is the DC amplitude, m is the mass of ion, x 0 = y 0 is the radial dimension, is the drive frequency, q z is the dimensionless parameter of stability, and V is the RF amplitude. The AC voltage s for isolation, excitation and ejection are applied with opposite polarity to opposing rod s in the x axis, typically 80 V at 5 500 kHz I f the frequency of this AC voltage matches (is in resonance with) the oscillatory frequency of a given ion the amplitude of its oscillations will increase and the ion is ejected through the radial exit slits. A broadband AC waveform such as SWIFT 39,40 or sum of sines 41 can be applied to perform mass isolation of a given packet of ions. After isolation, a relatively small AC voltage, referred to as resonant excitation or tickle voltage, adds sufficient kineti c energy to the ions for fragmentation for collision induced dissociation (CID) experiments. Increasing this RF voltage q z value) linearly allows for the mass selective ejection of ions from low to high mass to charge. Ejected ions exit through both radial slits, and impact onto dual 15 kV conversion dynodes producing secondary charged species (electrons or ions) that are detected by the dual electron multipliers. The current received on the multipliers is matched by a data acqui sition system to the instance when the aforementioned ion packet was ejected during the RF ramp resulting in a mass spectrum. Orbitrap The Orbitrap is a relatively new mass analyzer capable of high resolving power and mass accuracy. Both the geometric a nd mathematical descriptors of the Orbitrap can be quite complex; hence, only a rudimentary level of detail will be provided,
27 sufficient for the purview of this dissertation. The Orbitrap is based upon an ion trapping device first invented by Kingdon in 1 922, whose shape and applied electric fields were modified by Knight in 1981. 42 It was Makarov and colleagues who adapted the Knight modified Kingdon trap for use as an mass analyzer, first as a stand alone instrument with an ESI source 43 in 2003, then as the mass analyzer in an LIT Orbitrap hybrid instrument 44 in 2006 (Figure 1 2) 45 simplicity, robustness, performance characteristics, and ability to function without a cryogen or magnet, it has become a principal instrument for proteomic s 46 and other applications where high resolution and accurate mass measurements are of paramount importance. 29 The Orbitrap is constructed of a spindle shaped inner electrode surrounded by a coaxial barrel shaped outer electrode, which is split in two halves, and whose inner surface is nonparallel to that of the spindle, as seen in Figure 1 2D. 29 A constant electrostatic field potential is applied between the two electrodes, which is not constant across the central axis of the electrodes due to their non complementary shapes, thus causing a radial electric field. Following tangential injection, a packet of ions are out the spindle electrode with a radius proportional to the initial kinetic energy of the ion packet from the injection process and inversely proportional to the electric field from the electrodes, as provided by Equation 1 3 (1 3)
28 where r is the radius of the spindle electrode, eV is the energy, and eE is the force of the applied electric field. 47 The ion packet is injected via a specially designed trapping device (C trap, Figure 1 2C) new to the Orbitrap mass spectrometer that allows a very narrow pulse of ions from the LIT to be collected and 48 Following trapping, the ion packet will naturall y oscillate along the axis of the device due to the inhomogeneous electric field caused by the nonparallel electrodes as noted above. The frequency of oscillation may be modeled by the equation of a simple harmonic oscillator 47 as in Equation 1 4 (1 4) where is the frequency of axial oscillation, z is the charge of an ion, k is the oscillator constant, m is the mass of an ion. The ion oscillation is detected as an image current across the two halves of the outer barrel electrode, whose frequency ( ) can be then be converted via Fourier transform to mass ( m ) to generate a mass spectrum. 29 Because oscillation frequencies can be detected with very high accuracy and precision even at low signal, the Orbitrap can obtain a resolving power as high as 1,0 00,000 and a mass accuracy as high as 0.1 ppm. 29,45,46 Figures of Merit Three figures of merit critical to this dissertation, specifically Chapter 3, are resolution, resolving power, and ma ss accuracy. These figures are not often considered when operating an instrument with a mass analyzer capable of only low resolution, resolving power, and mass accuracy, such as an ion trap, but are of significant concern with mass analyzers such as an Or bitrap
29 Resolution ( ) is defined as the difference in mass (or m/z ) between two ions given by Equation 1 5 (1 5) where m 2 is the mass of ion 2, m 1 is the mass of ion 1, and m 2 > m 1 For a single ion, resolution is defined at the intersection at the full width at half maximum (FWHM) of that 49 Resolution is constant throughout the m ass to charge scale with mass analyzers that use quadrupolar fields, such as the LIT, 29,49 but resolution is not constant for mass analyzers such as a ToF and those that use double focus ing or Fourier transform methods 29 Resolution is a value to quantitate the separation between two peaks; higher resolutions are better than lower resolutions. Resolving power ( R ) is defined as the difference in masses (or m/z ) between two ions divided into a specific mass (or m/z ) value such that those two ions can be sufficiently separated, as given by Equation 1 6 (1 6) where M is the mass (or m/z ) of an ion and is the resolution as defined above. Therefore, R must always be provided in reference to a particular location in the mass to charge scale. Resolving power is a function of the mass spectrometer, are often incorrectly applied. 29 Resolving power is a value to quantitate the quality of separation for a mass spectrometer; higher resolving powers are better than lower resolving powers.
30 Mass accuracy is, in eff m/z ) of an ion compared to that of either a known calibrant or theoretical mass. Mass accuracy (or mass error) is defined as the difference between an experimentally determined mass and a theoretical mass divided by the theoretical mass as in Equation 1 7 (1 7) where is an oft used symbol for mass error, not to be confused with resolution, m e is the experimental mass, m t is the theoretical mass. For the convenience of whole numbers the multiplicative term 10 6 is used to provide parts per million (ppm). High mass accuracy is often incorrectly used interchangeably with high resolution and can be realized as a direct result from an increase in the significant figures of an experimental mass. 29 Mass accuracy is a value to quantitate the error of a mass; low er errors are better than higher errors. Overview of Dissertation This dissertation is intended to bolster the burgeoning field of conservation strength of tandem MS, the high resolution and accurate mass of the Orbitrap and the non destructive properties of LDI were all exploited to provide methodologies commensurate with the requirements of conservation scientists. Also, the introductory sections of the first t wo data Chapter s ( Chapter s 2 and 3) present a deliberate, exhaustively thorough review of the literature related to the Chapter projects. Many of the results observed and achievements in analysis are presented for the first time without prece dence in the literature.
31 Anthraquinones have been used extensively throughout history in both painted works of art and fabric as a source of red dyes. Dye analysis in the literature was performed mostly by chromatographic and ESI methods, so knowledge of their response to laser based ionization, such as LDI and MALDI, is lacking. Prior analysis of anthraquinones includes EI, where reduction was observed. Chapter 2 attempts to reconcile the reduction observed by EI with LDI and MALDI. Indeed, laser indu ced reduction of anthraquinones was observed and the site of reduction was determined by MS 2 The pigment lead white has never been the subject of any published work for cluster ion analysis, or any other types of analyses, despite being continually used since antiquity. Chapter 3 uses LDI MS 2 on both an LIT and Orbitrap to elucidate the of many isotopically complex cluster ions. Using the high resolution and accurat e mass capabilities of the Orbitrap in conjunction with the LIT, the cluster ions of lead white were successfully identified. Lastly, in Chapter 4, the results obtained in Chapter s 2 and 3 for the analysis of anthraquinone dyes and lead white, respectivel y, were applied to real samples. Using MS 2 alizarin and purpurin were determined to be present in both sections of painted works of art and swatches of dyed silk. Lead white was determined to be present in full scan spectra from sections of painted work s of art. The detection of the colorants was achieved without any sample pretreatment.
32 Figure 1 1. Schematic of the linear quadrupole ion trap (LIT). 36 The dimensions of the trap are shown in part A and the applied voltages are shown in parts B, C, and D. 12 mm 37 mm 12 mm 0 r (X,Y) = 4 x 4.375 mm 100 V 5 kV 1 MHz front section center section back section X Z Y Y X Y X Y Z A B C D 80 V 5 500 kHz
33 Figure 1 2. Diagram of the Thermo LTQ Orbitrap. 45 The MALDI source (A) is followed by ion transfer guides toward the linear ion trap (B), C trap (C), and the Orbitrap (D). A B C D
34 C HAPTER 2 TANDEM MASS SPECTROM ET R Y OF ANTHRAQUINONES Background for beauty. That basal desire is manifested in the addition, alteration, and manipulation of color to common and distinct objects such as garments, dcor, and works dedicated as art. Before the advent of modern chemistry, molecules for coloring were deve loped from natural sources. Inorganic, water insoluble pigments were usually obtained from minerals, ores, and rudimentary chemical preparation 50 while organic, water soluble dyes were harvested from animal and plant sources. Organic dyes are classified i nto three color groups 51,52 which arise (with some exceptions) from different chemical cl asses: blue (indigotins), yellow (carotenoids and flavonoids), and red (anthraquinones), the last of which is the focus of this Chapter Anthraquinones can be further sub classified as either animal, such as lac from the insect Kerria lacca and cochineal from the insect Dactylopius coccus or plant, such as madder from the root of Rubia tinctorum 52,53 Moreover, there are many different isomers and analogues of dyes from a particular source, and dyes are also found in many ge nera and species of a family, as in the case of madder. For instance, madder is the general term for at least thirty six different anthraquinone based red dyes that may be found in at least four different species of the genus Rubia which is within the fa mily Rubiaceae. 54,55 Because of such factors as color quality and light fastness, by far the most significant, notable, and well studied anthraquinone in madder is alizarin (1,2 dihydroxyant hraquinone), and to a lesser extent its analogue purpurin (1,2,4 trihydroxyanthraquinone). 55,56
35 The use of madder dyes have been known since antiqu ity, originating first from the East then west to ancient Persia and Egypt before arriving in ancient Greece and Rome. 57 In fact, madder is the oldest known textile dye, having been found as the dye of a belt in Tutankhamun tomb from 1350 BC. 55 Rubia the feedstock for red dye led it to be widely cultivated in Europe until alizarin was first synthesized in 1868, the first chemical synthesis of a natural dye. 57,58 synthesis, up to one half of a million acres, roughly the area of Luxembourg, was used in Europe to cultivate the Rubia crop. 59 Yet, not all species of the Rubiaceae family contain alizarin, as is the case of some South American species of the genus Relbunium and certainly not all species of the family contain the same relative amount of anthraquinones. 60 62 Indeed, determining the particular species of Rubia from where madder root extracts were obtain has been a factor in the analyses of madder, accomplished by HPLC diode array detection (DAD) HPLC/ESI MS, and GC/EI MS. 54,62 Because of the many similarly structured anthraquinones in madd er from extracts of Rubia spp., analyses of madder dyes have understandably used separation s such as HPLC, CE, and GC prior to DAD or MS detection. 60,62 65 as a textile dye, many analyses have focused on the detection of anthraquinones extracted from dyed textiles using GC/EI MS, 66 HPLC/TSP MS 67 HPLC DAD 61 and HPLC DAD/ESI MS 54,68 70 When alizarin or purpurin were able to be detected with MS, in cases not using EI they were observed as the [M+H] + or [M H] ion for positive or negative mode, respectively. Recently, direct analysis in real time (DART) MS was demonstrated to provide rapid analysis of madder on a textile and even directly from
36 madder root. 71 Alizarin and purpurin were both detected as [M+H] + ions. Although DART may be amenable to the conservation science field because of its non destructiveness, DART still suffers from being a relatively esoteric, underutilized ionization technique that cannot provide the separation of similar mole cules with in line chromatographic separation. Moreover, the cited DART analysis was conducted on a improvement compared to the destructiveness of extraction required for H PLC based methods, direct detection is still dominated by laser based ionization sources such as LDI and MALDI. The first laser based ionization m ass spectrometric method for the analysis of dyes. 72 Then in 1993, a second article was published on the two step infrared laser de sorption ultraviolet laser ionization time of flight mass spectrometry (L 2 ToFMS) of natural dyes. 73 Although twelve dyes were analyzed in that work, the one most relevant to this Chapter was disperse blue 1 (1,4,5,8 tetraaminoanthraquinone) since it is an anthraquinone d ye and was noted to appear as [M] + [M+Na] + and [M+2Na] + ions. In 2003, much highly cited laboratory on the LDI ToFMS, working directly from paper and illuminated manuscripts. 74 76 However, these works primarily deal with inorganic pigments, hence they will considered in the Chapter s 3 and 4 of this dissertation. The first and still one of the most comprehensive laser based analys e s of both anthraquinones particularly alizarin, among other dyes and direct analysis of painted works of art was the 2003 dissertation of Nicolas Wyplosz as a doctoral student of Jaap Boon. 77 Alizarin samples,
37 both neat and from a dyed textile, were analyzed with both LDI and MALDI on a 2 D QIT, yet no tandem MS was conducted. The displayed LDI spectrum shows alizarin in the positive ion mod e as the [M+H] + ion at m/z 241. Critically important to the work present in this Chapter an ion with very high intensity was shown at m/z 242, which Wyplosz claimed was the 13 C isotope of m/z 241. However, the m/z 242 ion had an intensity that was 95% of the m/z 241 ion even though the theoretical distribution for the 13 C isotope should be 15.4% of an [M+H] + ion of alizarin, given just 14 carbon atoms in the molecule. This discrepancy will be thoroughly covered in this Chapter of this dissertation. In 2007, the MALDI ToF analysis of dyes and pigments was published in which the major ions of 58 neat samples were catalogued. 78 In this work, alizarin was detected only in t he negative ion mode as the [M H] ion at m/z 239. In 2008, the LDI ToF analysis of madder standards was published in which alizarin and purpurin were observed only in the negative ion mode. 79 Both molecules were shown in spectra to appear at mass to charge ratios for their respective [M H] and [M ] ions. Lastly, in 2009, work was published on the LDI ToF analysis of modern pigments and dyes from standards and samples of painted works of modern art, yet no anthraquinones were tested. 80 Most significant for this Chapter was the work published in 2007 on the MALDI ToF analysis of synthesized anthraquinone derivatives. 81 The authors, whose expertise is in synthetic organic chemistry, 82 made the remarkable claim that their anthraquinones underwent never before protons or sodium ions were adducted, yet the ion remained singly charged. The pecul iar ions observed were [M+2H] + [M+2Na] + and [M H] + This phenomenon is
38 purportedly caused by the electron deficient anthraquinones from the dipoles of their carbonyl groups, although this claim was poorly substantiated by the lack of both direct evidence and proper citation. Hence, in the hypothesiz ed gas phase ionization mechanism, addition of an extra proton occurred concomitantly with the addition of an electron. The authors justify their claim with the following experimental findings: 1) The electron capture ability (double cationization) of th e anthraquinones was significantly increased with the electron withdrawing capability of the substituents. 2) Negative ion mode shows only negative radical ions, [M ] 3 ) Whether using the thin layer deposition method or a basic matrix, the phenomenon still was observed. 4) Matrix acidity had no considerable influence. 5) Phenomenon was not observed with ESI+. Lastly, without support by direct evidence or citation the authors rule out the possibility Yet, MALDI reduction of analytes has been extensively researched and published by some notable mass spectrometry groups. The evidence supporting the presence of introduced in 2000 by Karas et al. 83 Experimental confirmation of free electrons generated by the MALDI laser occurred in 2002 and 2003 by Zenobi et al. when they documented the photoelectric effect, from the UV photons interac ting with the metal MALDI sample plate, as the fundamental source of electrons produced. 84,85 It was shown that a decreased number of electrons correlated to the impediment of photons impinging the sample plate: samples with a thickness >1 mm had negligible production
39 of photoelectrons 84 and samples completely separated from the plate with a piece of adhesive tape experienced a >100 fold increase in positive ions 85 With an elegant set of experiments, Zenobi et al. also definitively proved that laser ionization caused copper salts to be reduced from Cu(II) to Cu(I) with higher yields a t higher laser fluence and without regard to the presence of matrix, steel substrate, or combination thereof, as observed in both positive and negative ion modes. 86 In 2004, a Japanese group published observations on the MALDI reduction of the chloride salts of four organic dye ca tions, which contained fused rings similar to anthraquinone, although without a quinone moiety. 87 The extent of reduction, via electron transfer and protonation, was shown to increase as a function of both increased matrix to analyte ra upon addition of Cu(II) as an electron scavenger. Yet, the work tested the dyes only with MALDI matrix, on a steel sample plate, in positive ion mode, without discussion of the possible site of proto nation, and wit hout CID since a ToF was used. Lastly, in 2008 a Ukrainian group published comprehensive work on the reduction of four imidazophenazine dye derivatives, which have four fused nitrogen containing rings. 88 In the positive ion mode singly and doubly reduced dyes, [M+2H ] + and [ M +3H] + respectively, were observed by FAB in glycerol, MALDI, LDI on steel, LDI on graphite, and low te mperature SIMS. In the negative ion mode singly reduced dyes, [M ] were observed by MALDI, LDI on steel, LDI on graphite, ESI, and low temperature SIMS; and doubly reduced dyes, [M+H ] were observed for only LDI on graphite. Interestingly, the extent of reduction correlated with the electron affinity of the dyes in t he positive ion mode for FAB in glycerol, LDI on steel, and LDI on graphite; and
40 in the negative ion mode for LDI on graphite and ESI. It should be noted that reduction is not the only effect laser based ionization methods may have on an analyte. Recent observations were published on the laser induced oxidation of cholesterol, which was thought to have been caused by hydroxyl radicals from irradiated MALDI matrix, although no definitive proof was provided. 89 Reduction ionization anomalies of anthraquinones do not occur exclusively from laser based methods or as protonated cations. Moreover, it is the central ring of anthraquinones indeed the benzoquinone moiety that appears to be the operative portion to generate ions of the form M+2. Intriguingly, M+2 ions of p benzoquinone were first reported in 1966 in a succinct article by Pike et al. in which radical cations were produced using an E I source at an elevated temperature (250 C). 90 Pike cleverly conditioned the source region with D 2 O and observed M+4 peaks in the resultant mass fo 2 O was repeated in 1967 by Ukai et al. who also observed increases of M+2 of benzoquinones as functions of source temperature and time. 91 In 1971, Park et al. reported on the isolation and characterization of a benzoquinone derivative from fungal cultures. 92 As a brief side note in the article, they commented that [M+2] + ions, as EI generated radical cations, had formed if the ion and Ukai regarding the role of water. In a pair of papers, Oliver et al. studied the M+2 93,94
41 of M+2 and their aqueous redox potential. Furthermore, Taylor et al. published in 1974 comprehensive experiments to expl ore the M+2 phenomenon in particular, the effect of water using derivatives of polyporic acid, which are centered on benzoquinone. 95 They concluded that the abundance of M+2 EI generated radical cations was dependent on the 1) temperature of th e sample, 2) pressure in the source, and 3) partial pressure of water in the instrument. In a r 2 O was added in the source, M+4 ions were also observed. Taylor concluded that the M+2 M+2 EI generated radical cations were reported fo llowing the characterization of benzoquinone and its derivatives from various sources 96 99 Following the exhaustive literature review and background above, it is clear that ionization of quinone containing molecules particularly anthraquinone based dyes can form anomalous M+2 ions, yet the exact proc ess is still not definite. It is of surprising and significant relevance to note that there exist many articles that use either an EI or a laser source to ionize quinone and anthraquinone containing dyes and other molecules in which M+2 ions are not obse rved. Unfortunately, scant experimental details particularly from works of the 1960s and 1970s that use EI on sector instruments preclude critical analyses that might have revealed a trend to explain this discrepancy. Although work over twenty years ago has shown interesting results regarding EI, it is laser particularly dyes, that is most applicable to the contemporary, growing field of conservation science.
42 Considering that in all of th e aforementioned laser based publications known molecules were analyzed, it begets the question on how the ionization phenomena would complicate the elucidation of unknown analytes or molecules forming ions with isobaric mass to charge ratios. Moreover, n one of the publications with laser based methods were able to perform tandem mass spectrometry which might have revealed sites of reduction, glimpses on mechanisms, or isobaric interferences. With the exception of the seminal work by Pike, all of the pub lications with EI based methods did provide EI gener ated fragmentation. Yet, the fragmentations, which will be discussed later, were not able to provide conclusive evidence of the cause or site of the M+2 phenomenon. Certainly, the need for MS/MS is ap parent and fulfilled herein. Experimental Methods Chemicals and Materials Formic acid (FA) and MALDI matrix 2,5 dihydroxybenzoic acid (DHB), were purchased from Acros Organics (Morris Plains, NJ) Whatman filter paper, acetonitrile (ACN), HPLC grade metha nol (MeOH), tetrahydrofuran (THF), and water were purchased from Fisher Scientific (Fairlawn, NJ). Anthraquinone standards of varying purity (a nthraquinone 99.5% ; 1,2 dihydroxyanthraquinone (alizarin) 85% ; 1,2,4 trihydroxy anthraquinone (purpurin) 90% ; 1,5 dihyrdroxyanthraquinone 85% ; 2,6 dihydroxyanthraquinone 90% ; 1,5 diaminoanthraquinone 85% ; and 2,6 diaminoanthraquinone 97% ) were purchased from Sigma Aldrich (St. Louis, MO) (Figure 2 1) Recrystallization of Standards The standards of both anthr aquinone and 1,2 dihydroxyanthraquinone were first recrystallized using established techniques. 100 Although there were slight variations in
43 recrystallization parameters such as total solvent volume, required time for dissolution, and required time for recrystallization, a generalized workflow for the r ecrystallization procedure follows: 1) heat 50 mL of MeOH to boiling 2) transfer 0.5 g of unpurified anthraquinone standard to a clean 50 mL Erlenmeyer flask 3) dispense hot MeOH in a slow, dropwise manner 4) stop addition of hot MeOH immediately following full disso lution of standard 5) cover the saturated solution and store in a cool, dark place 6) filter crystals following overnight recrystallization using filter paper and gravity filtration 7) rise filtered crystals with only a few drops of cold MeOH 8) transfer crystals to a watch glass to allow drying The recrystallized anthraquinones were stored in darkened vials until use to prevent exposure to ambient light. Preparation of Chemicals All recrystallized anthraquinone standards were first dissolved at a concentration of 100 0 ppm in THF. Initial dissolution in THF was a necessary step due to the limited solubility of most of the anthraquinones in a mobile phase suitable for electrospray. Furthermore, THF was selected as the initial solvent since it possesses the necessary l ower hydrophobicity to dissolve the anthraquinones, yet is still itself fully soluble in water. Thereafter, stock solutions were serially diluted in 50/50 ACN/H 2 O to concentrations of 100 ppm, 10 ppm, and 1 ppm and stored in a dark drawer at room temperat ure until use. For preparation of electrospray mobile phase, 0.1% FA was
44 added. The MALDI matrix DHB was prepared at a concentration of 40 mg/mL in 70/30 MeOH/H 2 O and stored in a freezer until use. Electrospray Ionization Parameters As a set of control e xperiments that ionize anthraquinones without the use of light (i.e., laser generated UV photons), electrospray ionization mass spectrometry was used. A Finnigan LCQ (San Jose, CA) was used with its standard electrospray source. All solutions (10 ppm) we re directly infused at a flow rate of 10 L/min. The applied potential on the ESI needle was +4.00 kV. The heated capillary was kept at 200.0 C and at +40.0 V. All other ESI and ion optics parameters such as auxiliary and sheath gas flow rates, tube le ns voltage, and lens and multipole offsets, were separately tuned for each standard analyzed. All spectra were recorded with automatic gain control (AGC) enabled for the target value of a normalized signal of 7.0 10 7 and with three microscans averaged pe r recorded analytical scan. All of the aforementioned parameters for recording a full scan were kept constant for tandem mass spectrometry. Additionally, the isolation width for all MS/MS spectra was adjusted to isolate only one profile peak in an isotopi c envelope of a given analyte, which is critical for the experiments in this Chapter As shown in F igure 2 2, the optimized value for isolation width was determined to be a 0.8 Da window, which was based upon the balance between a) obtaining maximal daugh ter ion intensities and b) preventing truncation of the peak of the desired parent ion while omitting undesired, interferent neighboring parent ions. Collision induced dissociation (CID) parameters were optimized using manual tuning. The parent mass was first isolated and the CID energy was increased in increments of 0.5 (arbitrary units) from 20 to 50. Optimized
45 CID values were declared for the observed CID interval at which the signal intensity of the parent ion dropped to 10% of the sum of intensitie s for the parent ion and the two most abundant daughter ions. A representative plot of this CID determination is shown in Figure 2 3, which would yield an optimized CID energy of 38. The average value for CID energies was 41 among all tested anthraquinon es. Laser Desorption/Ionization and Matrix Assisted Laser Desorption/Ionization Parameters All LDI and MADLI experiments were performed on a Thermo Finnigan LTQ XL (San Jose, CA) equipped with an intermediate pressure ( 170 mTorr) MALDI source and a N 2 las er ( = 337 nm) Standard solutions of anthraquinones were spotted with a volume of 1 L on a polished sta inless steel MALDI sample plate or glass microscope slides. For LDI experiments, the deposited solutions were allowed to dry unaided at ambient cond itions before being inserted and analyzed in the instrument. For MALDI experiments, a modified dried droplet method was employed. 101 The drie d droplet method as indicated in the literature has a separate vial that is used to pre mix the analyte and matrix solutions before that newly mixed solution is spotted on a sample plate. The modified dried droplet method that was employed for this work a voids the use of a separate vial since the analyte solution (1 (1 the deposited solution was allowed to dry unaided at ambient conditions before be ing inserted and analyzed in the instrument. Instrumental parameters such as the front lens voltage were automatically tuned to maximize the abundance of the base peak, which, in the case of MALDI spectra was unavoidably the DHB ion at m/z 154. Laser para meters were manually tuned to obtain
46 maximal signal for the ion of interest, yet minimizing the consequent increase in both baseline and space charge effects. Typical laser energies were approximately 10 J/pulse for LDI and 5 .0 J/pulse for MALDI. The desired ionization metric of laser fluence can only be estimated at 1.3 10 3 J/m 2 for LDI and 6.4 10 2 J/m 2 for MALDI due to the difficulty in an accurate and precise measurement of the laser spot size, whi ch has an approximate 100 m spot. The number of laser pulses for each analytical scan was controlled by the automatic gain control (AGC) to maximize the total ion signal, yet not exceed the target value of 3.0 10 4 Most often, the AGC determined number of laser pulses was nine. When samples were spotted on a stainless steel MALDI sample plate with dedicated sample wells, the plate was moved with respect to the stationary laser in either a spiral outward motion or automatically controlled via the soft crystal positioning system Considering that samples spotted on glass slides did not have dedicated sample wells that were recognized by the software, the sample holder was moved in a raster pattern within a user imaging mode. All of the aforementioned parameters for recording a full scan were kept constant for tandem mass spectrometry. The isolation width for all MS/MS spectra was adjusted to isolate only one profile peak in an isotopic envelope of a given analy te. The optimized value for isolation width was determined to be a 0.8 Da window. CID was isolated and the CID energy was increased in increments of 0.5 (normalized instrum ental values) from 20 to 50. Optimum CID values were defined as the point at which the parent ion intensity was reduced to 10% of the most intense peak in the
47 daughter ion spectrum. Typical values for CID energies were approximately 30 for all tested ant hraquinones. Ultraviolet visible (UV) Light Exposure To explore the model of photoreduction, anthraquinone was exposed to UV light for an hour and its UV vis spectrum was then recorded. A freshly prepared 10 ppm solution of anthraquinone was prepared via dilution according to the method above. The control aliquot was dispensed in to a quartz cuvette and placed in a dark drawer until analyzed. The test aliquot was dispensed in to a quartz cuvette and placed in a custom made chamber ( 30 cm l w h) constru cted from black poster board. The cuvette was placed 15 cm from a UVP, Inc. UVG 11 (San Gabriel, CA) 4 W ultraviolet TLC lamp ( = 254 nm) for 1 h. Thereafter, both cuvettes were analyzed in a Hewlett Packard 8450A UV vis spectrophotometer (Paolo Alto, CA). Results and Discussion U pon preliminary analyses for the mass spectrometric imaging of the dyes known to be present in the obtained painting cross sections, data revealed the peculiar and initially quite perplexing ionization of alizarin. Both laser desorption/ionization and matrix assisted laser desorption/ionization analyses of alizarin standards had shown a base peak whose identity was not readily understood, even after preliminary tandem mass spectrometric analysis. An initial review of the liter ature generated the recently published and seemingly promising article by Meijer et al. 81 which suggested a manner counter to all previously published mass spectral data. This phenomenon and how it may affect conservation scientist
48 brought the further and thorough tandem mass spectrometric investigations described herein. As will be hence forth detailed, alizarin and similar anthraquinones undergo peculiar ionization via reduction (but not double cationization), which is supported by both the thorough literature review in the introduction of this Chapter and the following experimental evide nce, as discussed later in this Chapter : 1) LDI and MALDI cause reduction of anthraquinones, which is observed as an ion in the form [M+2H ] + 2) ESI does not cause appreciable reduction of anthraquinones despite the aqueous, acidic mobile phase. 3) elucidated by tandem mass spectrometry. 4) Anthraquinone undergoes photoreduction at ambient conditions, which was confirmed by ESI MS/MS. 5) The extent of reduction for ali zarin and anthraquinone increases as a function of analyte concentration with both LDI and MALDI. 6) The extent of reduction for alizarin and anthraquinone decreases as a function of matrix to analyte ratio with MALDI. Electrospray Ionization of Anth raquinones Electrospray ionization was used, in effect, as the control ionization method. With the lack of UV photons, ESI may provide confirmation of the null hypothesis, which states that the ionization phenomenon, specifically the formation of [M+2H ] + ions, of anthraquinones would not be possible without the presence of UV photons. Naturally, the fundamental difference between ESI and both LDI and MALDI regarding factors such as causal mechanisms of ionization (i.e., without or with UV photons), phase (i.e., aqueous or solid), and source pressure (i.e., ambient or low vacuum), precluded any variable control among the two classes of ionization methods. Also, all ESI experiments
49 were performed on an LCQ, which is a three dimensional quadrupole ion trap, whereas all LDI and MALDI experiments were performed on an LTQ, which is a two dimensional quadrupole ion trap. Despite the geometric differences between the 3 D and 2 D traps and all the associated ion optics and requisite voltages applied therein, the two versions of the QIT played no discernable part in experimental variability. That is, experiments were designed as such to only observe changes in ionization after altering the ionization variable. When performing tandem mass spectrometry, there is a slight difference in some instrumental parameters for MS/MS. The internal kinetic energy of the analytes as imparted by the CID energy, which is a normalized value set by the manufacturer, may vary between the LCQ and the LTQ. Therefore, all daughter ion spectra will slightly vary depending on the instrument used. Nevertheless, the respective daughter ion spectra are still amenable for comparison since differences in the CID among the two instruments would have only differences in imparted not delivery m odes of kinetic energy and would simply be observed as providing slightly varying intensities of daughter ions and not differences in actual fragmentation pathways and consequent daughter ions. Lastly, the amount of internal energy imparted into the ions during ionization will vary among the three methods (i.e., LDI, MALDI, and ESI). As observed i n Figure 2 4A, the p ositive mode ESI spectrum of 1,2 dihydroxyanthraquinone (alizarin, MW = 240) produced a singly protonated, non reduced ion at m/z 241, [M+H] + Of course, this is the ion one would expect from alizarin and shall be henceforth referred to as P (parent), which is a nomenclature historically used, albeit increasingly infrequent, by mass spectrometrists The ion at m/z 242 shall be henceforth r eferred to as P+1 (parent+1 Da ) and incorporates the 13 C
50 isotope contribution of P. The experimental P+1 peak has an intensity of 20.6, which is normalized to the experimental P peak, and is marginally higher than the normalized P+1 peak (15.4) for the theoretical spectrum of a 14 carbon compound. The reason for the higher experimental P+1 peak for anthraquinone, and some of the other compounds tested, may be a result of spectral interference that may arise from both impurity of the standard and limited ESI induced electrochemical reduction 102 The ion at m/z 243 shall be henceforth referred to as P+2 (parent+2 Da ) and incorporates both the 18 O isotope contribution of P and the 13 C contribution of P+1. The experimental P+2 peak has a n intensity of 2.72, which is normalized to the experimental P peak, and is quite close to the normalized P+2 peak (1.93) for the theoretical spectrum given two oxygens. Considering that the ESI mobile phase was 50% water, the role of water in the product ion of M+2 ions as was observed with prior experiments cited in the introduction, is questionable and inconclusive. 90,92,95 These three ion peaks are all predicted by the spectrum of theoretical isotopic distributi on for alizarin, which is shown in Figure 2 4B as an overlay on the experimental spectrum. The two spectral plots closel y overlap since they share peaks with similar abundances at the same m/z yet there is a slight shift of the experimental spectrum to a higher m/z This shift might be explained by the possibilities of both small space charging and unoptimized calibration For instance, the lowest mass calibrant used for the LTQ is the peptide MRFA, which has an [M+H] + ion at the monoisotopic m/z of 524 .27 close to double the m/z of the tested anthraquinones. Alizarin was discussed as the exemplar case. Indeed, for all of the compounds tested, the ESI generated experimental and theoretical spectra overlap, as shown in
51 parts A and B in Figure s 2 4, 2 6, 2 8, 2 10, 2 12, 2 14, 2 16. The tabulated summary of theoretical and experimental spectra for all anthraquinones tested appear s in Table 2 1 The reason for the higher experimental P+1 peak for anthraquinone; 1,2 dihydroxyanthraquinone; 1,5 dihydroxyant hraquinone; and 2,6 dihydroxyanthraquinone may be a result of spectral interference that may arise from impurity of the standard and limited ESI induced electrochemical reduction 102 Although, there was no apparent correlation observed between a higher P+1 peak and purity of the standards. Tandem mass spectrometry was performed on both the P and P+1 peaks for all an thraquinones tested. Again, alizarin will be treated as the exemplar case. As shown in Figure 2 5 parts A and B, the daug hter ion spectrum of m/z 241, P, closely corresponds to the daughter ion spectrum of m/z 242, P+1. Although the m/z values of the daughter ions differ by 1 Da the neutral loss (NL) values, which indicate the fragmentation of the parent ion, are the same. The NL of 28 is derived from the loss of CO from one carbonyl functional group about the central quinone backbone that forms the fluorenone ion, which is supported by numerous articles, albeit by the fragmentation of EI generated, odd electron, radical c ations. 66,95,97,103 108 Furthermore, as Beynon et al. proved with 18 O labeled 1 hydroxyanthraquinone, the loss of 28 is highly favorable as the loss of carbon monoxide from the carbonyl rather than the hydroxyl portion. 107 The loss of 28 is seen with high abundance in all anthraquinones except the two diaminoanthraquinone compounds (Figures 2 13 and 2 15, parts A and B). Those two amino compounds preferentially fragment with a cross ring cleavage, observed as a NL of 93, which incorporates carbon positions 1, 2, 3, 9, and 13 (or 5, 6, 7, 10, an d 11) and
52 respective substituents. Interestingly, when the same two amino compounds were analyzed in 1960 by Beynon et al. as EI generated odd electron, radical cations, a similar cross ring cleavage was not observed, but the loss of carbon monoxide was 103 ; however, Bowie et al. did observe cross ring cleavage for derivatives of benzoquinones. 104 The NL of 46 is observed in ESI spectra only for alizarin This loss is attributed to the loss of one carbo nyl functional group and one water. Although this loss, which may be realized by the possible interaction of the carbonyl and the hydroxy at the 9 and 1 positions, respectively, has a significant role in the fragmentation pathway of anthraquinones reduced via LDI or MALDI, its appearance is not completely understood at this time. Also, the other two compounds that might experience a similar interaction with a hydroxy at the 1 position (1,2,4 trihydroxyanthraquinone, Figure 2 7 A and B, and 1,5 dihyroxyant hraquinone, Figure 2 9 A and B) do not exhibit a loss of 46. The NL of 56 is derived from the loss of two CO moieties from two carbonyl functional groups to form an ion of biphenylene. This consecutive loss has been well documented, albeit for EI generate d odd electron, radical cations. 66,97,103 108 The three compounds that exhibit cross ring cleavage thereby losing one c arbonyl [1,5 and 2,6 diaminoanthraquinone (Figures 2 9 and 2 11, parts A and B) and 1,2,4 trihydroxyanthraquinone (Figures 2 7 A and B), the latter with cleavage that incorporates carbon positions 1, 9, and 13 with respective substituents] have the daught er ion for NL 56 at a relative abundance of less than one percent. Other losses with relative abundances of note are some cross ring cleavages, which ultimately are inconsequential for the aim of this Chapter For alizarin, the NL of
53 84 suggests a cross r ing cleavage, although the positions about which the cleavage takes place is unknown (Figure 2 5 A and B). For purpurin, the NL of 42 must be the loss of carbons 2 and 3 with the attached hydroxyl group (Figure 2 7 A and B). Interesting bifurcations are observed with the four symmetrical molecules 1,5 and 2,6 dihydroxy (Figures 2 9 and 2 11, parts A and B); and 1,5 and 2,6 diaminoanthraquinone (Figures 2 13 and 2 15, parts A and B) at neutral losses of 120 and 119, respectively. Obviously, the split ma y occur indistinguishably across either positions 9 13 and 10 11 or 9 12 and 10 14. A serendipitous accident allowed for the reduction of anthraquinone to occur under ambient conditions and be detected with ESI. A volumetric flask containing a 10 ppm sol ution of anthraquinone was unintentionally left on a lab bench for the weekend. Thereafter, an aliquot was used for analysis by ESI, following the addition of 0.1% FA similar to the normal samples. The resultant spectrum is in Figure 2 18A and shows a si gnificant P+1 to In fact, the ambient reduction of anthraquinone has a reduction extent greater than the 1.95 ratio for the LDI generated spectrum. To confirm that the ion at m/z 21 0 was indeed the [M+2H ] + ion of anthraquinone, tandem mass spectrometry was performed. As seen in Figure 2 18B, the daughter ion spectrum from m/z 210 clearly shows a relatively abundant NL of 29 that is assigned as one reduced carbonyl. The reduction o f anthraquinone at ambient and other conditions especially in an aqueous medium is a well established feature and has been studied spectroscopically for many years. 51,59,109 114 T o further explore light induced reduction, an aliquot from a freshly prepared 10 ppm solution of anthraquinone was irradiated with a UV lamp for 1 h and analyzed
54 by a UV vis spectrophotometer. When compared to an aliquot of the same solution that was not exposed to the UV lamp, a hyperchromic (higher) shift of the relatively low shoulder at 400 nm is apparent, as seen in Figure 2 19A and B. This spectrographic change in the absorbance of anthraquinone can be reasonably attributed to the UV induced reduction phenomenon observed with LDI and MALDI. However it must be mentioned that any direct correlation of the two observations warrant caution since the correlation of molecular structure to UV vis spectra is ultimately impossible. 51 The UV irradiated sample was not analyzed with ESI mass spectrometry. Laser Desorption/Ionization of Anthraquinones Again using alizarin as the exemplar case, its positive mode LDI spectrum is shown i n Figure 2 4C What is immediately apparent in t his portion of the full spectrum is that the base peak is not the [M+H] + ion at m/z 241 (P), but rather, the [M+2H ] + ion at m/z 242 (P+1). In fact, as shown in Table 2 1, P+ 1 is 77% greater than P. Indeed, P+1 is the reduced form of the alizarin ion. A lthough P is not the dominant ion, it is still relatively abundant P is 33.6% of the total ion abundance of the P, P+1, P+2 envelope (Table 2 2) and its presence may be explained by the lack of total conversion because of a limited number of available prot ons, which will be supported in the following section on MALDI. The ion at m/z 240 is also present at a very low abundance, which is undoubtedly the unprotonated radical cation [M ] + Interestingly, relatively dominant unprotonated radical cations were o bserved for 1,2,4 trihydroxy (Figure 2 6C); 1,5 dihydoxy (Figure 2 8C); 2,6 diamino (Figure 2 14C); and 1,5 diaminoanthraquinone (Figure 2 12C), the last of which has that ion at significantly higher abundance 28.1% of the P 1 (i.e., radical), P, P+1, P+2 envelope. No discernable trend that correlates the
55 abundance of the radical cations to either structure or functional groups is apparent. It is important to note that abundance of the P+1 ion peak is produced not solely from the reduced ion; rather it in cludes the 1.1% 13 C isotope contribution from P. For the diamino isomers, P+1 also includes the 0.37% 15 N isotope from P. Moreover, the abundance of the P+2 ion peak is a combination of both the 13 C and the 0.20% 18 O isotopes contribution of P+1 and P, r espectively. The LDI generated experimental and theoretical overlay spectra for all the anthraquinones tested are shown in parts C and D in F igur es 2 4, 2 6, 2 8, 2 10, 2 12, 2 14, 2 16 The tabulated summary of theoretical and experimenta l spectra for a ll anthraquinones is in Table 2 1. Tandem mass spectrometry was performed on both the P and P+1 peaks for all anthraquinones tested. The daughter ion spectrum for alizarin, again as the exemplar case, is shown i n Figure 2 5 parts C and D In Figure 2 5C, the MS/MS spectrum of the unreduced [M+H] + ion at m/z 241 P, precisely corresponds to both the P and P+1 daughter ion spectra of the unreduced alizarin ions produced by ESI, as shown i n Figures 2 5A and B. As expected, the neutral losses of the daughter ion mass spectra of P for all LDI generated ions of the anthraquinones tested correspond to the neutral losses of the daughter ion mass spectra of both P and P+1 for all the ESI generated ions (Parts A, B, C for Figures 2 5 2 7 2 9 2 1 1 2 1 3 2 1 7). The only exception is the daughter ion at m/z 195 for 2,6 diaminoanthraquin one (Figure 2 15C) th at stems from a neutral loss of 44 Da which suggests fragmentation to lose CO 2 ; however, this loss is not seen in any other anthraquinone except as a very low abundant daughter ion for 1,5 diaminoanthraquinon e (Figure 2 13C). It follows that the most likely assignment for the loss that includes nitrogen is one carbonyl and one of the amine groups (i.e.,
56 CONH 2 ), although the exact structure is neither known nor crucial. This loss seems to be a violation of the nitrogen rule, which maintains proper parity, but is actually a commonly observed neutral loss, albeit as a radical. Nevertheless, the appearance of a relatively abundant NL of 44 from the P ion for 2,6 diaminoanthraquinone in the LDI but not the ESI generated tandem mass spectrum is perplexing. The heart of these experiments is to elucidate the anomalous P+1 ions, which, with the following comparison of daughter ions from P+1 to P, will be explained by m odel of reduction of one of the carbonyls. Briefly, two critical observations are made to support the model: 1) reduction of carbonyl produces a decrease in relative abundance of NLs of 28 and 56, which are loss es of CO from one and two carbonyls, respect ively, with the concomitant increase in relative abundance of NL of 29 (HCO ), which is from one reduced carbonyl (C O H); 2) reduction of the carbonyl adjacent to a substituent at the 1 position causes intramolecular hydrogen bonding and proton transfer that produces a significant increase in the relative abundance of NLs of 18 (water) or 17 (ammonia) for hydroxyl or amino groups, respectively. The simplest case to support the first critical observation on the reduction model is the LDI generated daughter ion spectrum of reduced anthraquinone, P+1 (Figure 2 17D). It is quite clear that the relative abundance of the daughter ion from the lo ss of CO from one carbonyl (NL 28) has drastically decreased while the dau ghter ion from the loss of HCO from one red uced carbonyl (NL 29) is now present. The reason why the abundance of NL 28 is 40% of the abundance of NL 29 must be due to differences in the energetics and, perhaps, the kinetics of the two fragmentations For anthraquinone, the two carbonyls are indis tinguishable, so it would be safe to assume an equimolar
57 reduction of either carbonyl (in a 1:1 ratio). Therefore, the observed 5:2 ratio of the two daughter ions must be explained by the fragmentation, which stems from the loss of an odd electron neutral fragment ( HCO ), from the odd electron parent ion being more likely to occur than the loss of a n even electron neutral molecule ( CO ) Dramatic decreases in the NL of 28 from P to P+1 can be seen for all other anthraquinones tested (parts C and D, Figure s 2 5 2 7 2 9 2 1 1 2 1 3 2 1 5 ). For the two compounds that do not exhibit intramolecular hydrogen bonding with a reduced carbonyl (2,6 hydroxy and 2,6 diaminoanthraquinone) similar appearances of NL 29 are seen, also with ratios that are explained by differences in the energetics of fragmentations (Part D in Figures 2 11 and 2 15). Also, it should be noted that 2,6 diaminoanthraquinone shows a NL of 29 and not the NL of 28 as a daughter ion from the unreduced form, P (Figure 2 15C). When judged by it self, the NL of 29 may have been attributed to HCO from a reduced carbonyl. Yet, considering the peculiar NL of 44 for this compound noted above it is apparent that 2,6 diaminoanthraquinone exhibits fragmentation not observed with the 1,5 isomer; therefo re, the NL of 29 for P of 2,6 diaminoanthraquinone should be attributed as loss of CNH 3 This NL will nonetheless not violate the reduction model for this compound, which will be expounded below. To support the second critical observation on the reducti on model, the LDI generated daughter ion spectra of the reduced (P+1) isomers of dihydroxy anthraquinone must first be compared (Part D in Figu res 2 9 and 2 1 1 ). The 2,6 isomer has the hydroxyl groups distal to the carbonyl groups at positions 9 and 10; t herefore, no intramolecular hydrogen bonding can occur between a reduced carbonyl and a hydroxyl group. Observing the daughter ion spectrum from P to P+1, the relative
58 abundance of NL 18 from the loss of water is dramatically decreased and the NL of 17 fr om the loss of OH is now pr esent (Figure 2 11C and D). Cons idering the NL of 17 was not observed from P, this loss might have been erroneously assigned to the OH of the reduced carbonyl. However, because of the lack of intramolecular hydrogen bonding, w hich will soon be described as allowing the hydroxyl group to be lost as stabilized water, the OH is allowed to be lost on its own. Furthermore, the NL of 45 is the losses of CO and OH from both the unreduced carbonyl and a hydroxyl group, while the NL o f 46 is the simultaneous losses from both the reduced carbonyl and a hydroxyl group. The NL of 46 can also stem from losses of both water and HCO ; yet, this additional contributing loss would only cause an increase in its abundance relative to the other important daughter ions (i.e., NLs of 17, 18, 28, 29, 45) and not the observed decrease. Why the relative abundance of NL 46 is lower than the other important daughter ions is immediately not clear, except for the possibility of it simply being a reflecti on of the difference in energetics arising from the different losses. The 1,5 isomer of dihydroxyanthraquinone has hydroxyl groups proximal to the carbonyl groups at positions 9 and 10; therefore, both intramolecular hydrogen bonding and proton transfer ca hydrogen transferring to a hydroxyl group that leads to the highly favorable neutral loss of water. Intramolecular hydrogen bonding in substituted anthraquinones is a well established featu re, particularly in acidic, aqueous media, which has been extensively studied by spectroscopic methods for many years. 51,59,108,109,115,116 As seen in Fig ure 2 9D, prot on transfer leads to a significant daughter ion ( m/z 224) with NL of 18 that dwarfs all other important ions in the spectrum. Since 18 O isotopic labeling was not
59 employed for these experiments, it is difficult to differentiate between the lo ss of water from the carbon at the 1 (hydroxyl) or 9 (reduced carbonyl) position. MS 3 was performed on the m/z 242 daughter ion (data not shown), generating granddaughter ions at neutral losses 28 (base peak, bp), 17 (7.0% of bp), and 56 (11% of bp). Sin ce the NL of 56, which is attributed to the loss of two CO, was seen from the daughter ion, it is reasonable to assume that the neutral loss of water occurred with the oxygen at the 1 position. Also, there is a still a very low abundant daughter ion at m/ z 214 that is a NL of 28 (CO), which, at 2.02% of m/z 224, may erroneously be the remnant of the 13 C contribution from P. Yet, a contribution from P is not supported since NL of 46 is seen as an abundant daughter ion in the reduced form, P+1, of purpurin (Figure 2 7D) a nd aliza rin (Figure 2 5D), whic h both have a daughter ion from the NL of 28 from one unreduced carbonyl. Therefore, by proxy of the presence of a NL of 28 in purpurin and alizarin, it is reasonable to infer that the NL of 28 in 1,5 dihydrox yanthraquinone is also from one unreduced carbonyl and, subsequently, the NL of 46 is the combination of losses of both water and one unreduced carbonyl. Also, along with the very abundant loss of water, the absence of the loss of a reduced carbonyl (NL 2 9) for 1,5 dihyrdoxyanthraquinone reinforces the prominence of proton transfer from a hydroxyl group proximal to the reduced carbonyl and it also suggests proton transfer of all the reduced carbonyls. Lastly, the bifurcation seen both from P of the 1,5 an d 2,6 isomers of dihydroxyanthraquinone and P+1 of the 2,6 isomer is not seen from P+1 of the 1,5 isomer, which may be attributed to the increased stability given by intramolecular hydrogen bonds.
60 The daughter ion spectra for both isomers of diaminoanthra quinone (Parts C and D in Figur es 2 13 and 2 15) sho uld have the neutral losses similar to their respective dihydroxy analogues, but with a mass shift to one Da higher for those NLs that include a nitrogen. Indeed, this holds true for the P+1 spectra, but with one exceptio n : the 2,6 diaminoanthraquinone has a relatively abundant neutral loss of ammonia (Figure 2 15D), though the loss of water from its dihydroxy analogue was relatively less abundant (Figure 2 11D). Nevertheless, the model of P+1 stemming f rom a reduction process is still upheld. The only peak that lacks definite assignment is the daughter ion at m/z 210 for P of 2,6 diaminoanthraquinone (Figure 2 15C). If the NL 29 corresponds to HCO then it should be observed only from P+1 as a loss of a reduced carbonyl. If this loss includes nitrogen, its assignment could be methyl enei mine (CH 2 =NH). Also, the absence of this loss from the daughter ion spectra from P of all other anthraquinones tested including the ESI generated daughter ion spectra of this compound in question is further confounding. Considering purpurin has two hydroxyl groups proximal to both carbonyls, it is understandable that its daughter ion spectra from P+1 is quite similar to that of 1,5 dihydroxyanthraquinone (Part D i n Figu res 2 7 and 2 9). M oreover, considering alizarin has only one hydroxyl group proximal to one carbonyl, it is understandable that its daughter ion spectra from P+1 is a p seudo average between the 1,5 and 2,6 isomers of dihydroxyanthraquinone (Par t D in Fig ures 2 5, 2 9, and 2 11). T his assessment can be realized by comparing the absolute abundances from NLs 18 (H 2 O) and 29 ( H CO ) to their relative count of proximate carbonyl hydroxyl groups. A function of counts of proximate carbonyl hydroxyl groups (i.e. 0; 1; 2; for 2,6 dihydroxy; alizarin; and
61 1,5 dihy d roxy; respectively) versus the ratio of absolute abundances from NLs 18 and 29 (i.e., H 2 O/COH) quantifies the observed increase with 0.865, 4.14, and 316 being the H 2 O/COH ratio for their respective 0, 1 and 2 proximate groups. Matrix Assisted Laser Desorption/Ionization of Anthraquinones Because LDI and MALDI are similar ionization methods considering their use of laser generated UV photons, it is to be expected that their spectra have accordingly gene rated P+1. However, when MALDI employs the use of an acidic matrix, as is the case with the 2,5 dihydroxybenzoic acid used for this work, there is a preponderance of available protons. If the model of a reduction phenomenon that is, the addition of one e xtra proton along with an electron is to prevail, then it is reasonable to expect that MALDI generated P+1 would be correspondingly greater in abundance. Indeed, that is exactly the case, as observed in Part E and F in Figures 2 4, 2 6, 2 8, 2 10, 2 12, 2 14, 2 16, where the base peak of the shown m/z range is P+1 far exceeding the abundance of P for all compounds ionized by ESI, and for most compounds ionized by LDI. For alizarin, the extent of reduction, that is the ratio of P+1 to P, from LDI (1.77) to MALDI (7.48) is increased greater than four fold, as indicated in Table 2 2. Yet, P is still present at 10.0% of the total ion abundance of the P, P+1, P+2 envelope, although it was reduced by two thirds from total ion abundance of the envelope for LDI m entioned in the preceding section. According to Table 2 2, the compounds with the respective highest and lowest increase of the P+1 to P ratio are anthraquinone and 1,5 dihydroxyanthraquinone. No explanation readily exists for the results of these two co mpounds or otherwise apparent trend for all of the compounds as seen in Table 2 2. Tandem mass spectra for the MALDI experiments were obtained, but are not provided in this presentation. As expected, the spectra are quite similar to those
62 produced via LDI despite some added low abundant spectral complexity from MALDI matrix interferents that fall at the same nominal m/z as the selected parent ions. To evaluate the extent of reduction in MALDI as a function of both concentration and matrix to analyte ratio alizarin and anthraquinone were prepared at concentrations 1, 10, 100, and 1,000 ppm, spotted with a volume of 1 L, and analyzed with MALDI and LDI. As seen in parts A and B of Figure 2 20, the attenuated laser power for LDI was doubled from 30 to 60 so that the total ion count was sufficient to obtain usable spectra and to satisfy the AGC threshold. This differen ce in laser attenuation makes the plotted data (P+1/P ratio) more remarkable since considerably more reduction was observed with MALDI than LDI despite the reduced laser power. The extent of reduction for both alizarin and anthraquinone increase with both MALDI and LDI as concentration in increased. For a given spotted sample, if the concentration of analyte increases, the thickness of the sample should also increase. According to the Zenobi model of laser induced reduction, increasingly thi c k samples im pede UV photons from emitting photoelectrons from the steel sample pla te. 84 T he sample thickness, therefore, is inversely proportional to the number of free electrons produced and, consequently, the extent of reduction. 85,86 So, that model does not fit the results present here. To explore this model by using a glass surface to eliminate the production of photoelectrons, a single experiment was conducted with a 1000 ppm sample of alizarin on a glass surface with LDI at the same attenuated laser power of 60 (data not shown). The resultant extent of reduction, as reflected in the P+1/P ratio, was 0.851, which is slightly lower than the data point plotted (0.926) at the same c oncentration on stainless
63 on the relationship between extent of reduction and surface substrate. A different explanation for the increased extent of reduction at high sample concentrations in MALDI and LDI could be that as more analyte is desorbed (and ionized) in the desorption plume, more protons are also available. This idea is not necessarily new, considering the work by Taylor et al. mentioned in the introduction, which stated that the abundance of the M+2 ion was dependent on the pressure in the source, among other factors, albeit for EI generated ions. 95 There were multiple articles published, also with EI generated ions, that stated the presence of wate r in the source directly caused an increase in the M+2 ion. 90,92,95 More experiments are ultimately required to be confident as to the cause of the observed results. Data from the extent of reduction as a function o f sample concentration can also be extended to determine dependence on the MALDI matrix to analyte ratio. The MALDI matrix was prepared at a concentration of 40 mg/mL, which translates to 40,000 ppm. As seen in Figure 2 21, replotted from F igure 2 20, th e extent of reduction for both alizarin and anthraquinone decreases as the matrix to analyte ratio increases. Again, these results are not in accord with expectations considering two reasons. First, as the matrix to analyte ratio increases, it would be q uite safe to assume that the increase in protons from the acidic matrix would increase the extent of reduction. Second, the 2004 work by a Japanese group that was mentioned in the introduction 87 found that the extent of reduction increa sed as a function of increased matrix to analyte ratio. Though their result was for MALDI produced ions, the compounds tested were chloride salts of four organic dye cations, which contained fused rings similar to
64 anthraquinone, but without a quinone moie ty. 87 Again, more experiments are ultimately required to be confident as to the cause of the observed results. Conclusion For many years, the appearance of M+2 peaks in the mass spectra of anthraquinone and other quinone containing comp ounds has been a confounding problem. Although progress on the probable experimental origin of this phenomenon was gained many years ago with the analyses of EI generated radical cations, awareness of this progress seemed to have been overlooked when lase r based mass spectrometric analyses became in vogue. One of the intentions of this work was to marry the information gained from both types of analyses with the powerful utility of tandem mass spectrometry. Indeed, the conclusion drawn thus far from the data presented in this work validates a gas phase, UV photon induced reduction of a carbonyl functional group in anthraquinones. Reduction was not observed for ESI generated spectra of anthraquinones to an appreciable extent yet was clearly obvious with l aser based ionization, particularly with MALDI, which contributes an excess of protons. The LDI MS/MS spectra of the proof of reduction at the site of a carbonyl with t he daughter ion at NL 29 (HCO ) for P+1 and the concomitant decrease in both NL 28 (CO) and NL 56 (2 CO) for P. Furthermore, MS/MS of P+1 shows that anthraquinones with reduced carbonyls in close proximity to either hydroxyl or amino functional groups par ticipate in intermolecular hydrogen bonding to show very abundant NLs of water or ammonia.
65 The extent of reduction was seen generally to increase across four increasing orders of magnitude of analyte concentration for both LDI and MALDI. Surprisingly, the extent of reduction was seen to decrease as the matrix to analyte ratio increase d despite the supposition that an increase in availability of participating protons from the MALDI matrix would add to the reduction extent. These concentration studies shou ld be refined and expounded with further experimentation. This work presents the first ever tandem mass spectrometric analysis of any M+2 ions. Moreover, this work proves for the first time that the site of reduced anthraquinones occurs at one of the carb onyls.
66 Figure 2 1. Structures of all anthraquinones investigated and the carbon numbering scheme for anthraquinone.
67 F igure 2 2. Optimization of isolation width 1,2 dihydroxyanthraquinone during ESI MS/MS on an LCQ. 1,2 dihydroxyanthraquinone ESI+ MS 2 241 10 ppm 50/50 ACNH/H 2 O 0.1% FA width=0.8, CID=0 1,2 dihydroxyanthraquinon e ESI+ MS 2 242 10 ppm 50/50 ACNH/H 2 O 0.1% FA width=0.8, CID=0
68 Figure 2 3. Optimization of CI D energy for the m/z 241 ion of 1,2 dihydroxyanthraquinone during ESI MS/MS on an LCQ.
69 Figure 2 4. The mass spectra of 1,2 dihydroxyanthraquinone (alizarin) with its corresponding theoretical overlay for ESI (A) (B), LDI (C) (D), and MALDI (E) (F).
70 Figure 2 5. The tandem mass spectra of 1,2 dihydroxyanthraquinone (alizarin) for m/z 241 and m/z 242 for ESI (A) (B) and LDI (C) (D).
71 Figure 2 6. The mass spectra of 1,2,4 triydroxyanthraquinone (purpurin) with its corresponding theoretical overlay fo r ESI (A) (B), LDI (C) (D), and MALDI (E) (F).
72 Figure 2 7. The tandem mass spectra of 1,2,4 triydroxyanthraquinone (purpurin) for m/z 257 and m/z 258 for ESI (A) (B) and LDI (C) (D).
73 Figure 2 8. The mass spectra of 1,5 dihydroxyanthraquinone with its corresponding theoretical overlay for ESI (A) (B), LDI (C) (D), and MALDI (E) (F).
74 Figure 2 9. The tandem mass spectra of 1,5 dihydroxyanthraquinone for m/z 241 and m/z 242 for ESI (A) (B) and LDI (C) (D).
75 Figure 2 10. The mass spectra of 2,6 dihydroxyanthraquinone with its corresponding theoretical overlay for ESI (A) (B), LDI (C) (D), and MALDI (E) (F).
76 Figure 2 11. The tandem mass spectra of 2,6 dihydroxyanthraquinone for m/z 241 and m/z 242 for ESI (A) (B) and LDI (C) (D).
77 Figure 2 12. The mass spectra of 1,5 diaminoanthraquinone with its corresponding theoretical overlay for ESI (A) (B), LDI (C) (D), and MALDI (E) (F).
78 Figure 2 13. The tandem mass spectra of 1,5 diaminoanthraquinone for m/z 241 and m/z 242 for ESI (A) (B) and LDI (C) (D).
79 Figure 2 14. The mass spectra of 2,6 diaminoanthraquinone with its corresponding theoretical overlay for ESI (A) (B), LDI (C) (D), and MALDI (E) (F).
80 Figure 2 15. The tandem mass spectra of 2,6 diaminoanthraquinone for m/z 239 and m/ z 240 for ESI (A) (B) and LDI (C) (D).
81 Figure 2 16. The mass spectra of anthraquinone with its corresponding theoretical overlay for ESI (A) (B), LDI (C) (D), and MALDI (E) (F).
82 Figure 2 17. The tandem mass spectra of anthraquinone for m/z 209 and m/ z 210 for ESI (A) (B) and LDI (C) (D).
83 Table 2 1. Summary of results for the spectral abundances of all anthraquinones analyzed. Reduction Abundances Theoretical ESI+ LDI+ MALDI+ P P+1 P+2 P P+1 P+2 P P+1 P+2 P P+1 P+2 ant hraquinone 100 15.3 1.50 100 25.86 9.87 100 195 30.5 100 1190 208 1,2 dihydroxy anthraquinone 100 15.4 1.93 100 20.56 2.72 100 177 20.8 100 748 149 1,2,4 trihydroxy anthraquinone 100 15.4 2.13 100 15.65 2.15 100 76.2 12.3 100 265 46.1 1,5 dihyrdroxy anthraquinone 100 15.4 1.93 100 18.90 6.84 100 132 26.4 100 631 111 2,6 dihydroxy anthraquinone 100 15.4 1.93 100 28.96 6.83 100 208 33.6 100 193 37.8 1,5 diamino anthraquinone 100 16.0 1.61 100 15.97 1.50 100 65.7 13.2 100 124 40.3 2,6 diamino anthraquinone 100 16.0 1.61 100 17.07 2.21 100 150 26.8 100 140 26.7
84 Figure 2 18. ESI+ mass spectra of anthraquinone after a three day exposure to ambient light ; full scan (A) and daughter ion scan (B) of the m/z 210 (P+1) parent ion. [M H 2 O] + [M 18] + [M CO] + [M 28] + [M COH] + [M 29] + A B
85 Figure 2 19. UV vis spectrographs of anthraquinone with (A) and without (B) exposure to UV light. A B
86 Table 2 2. Abundance of P as a percentage of the total abundance of the P, P+1, P+2 env elope for both LDI and MALDI. Ratio of abundance of P+1 to P for both LDI and MALDI. Increase from LDI to MALDI of both P as a percentage of the total abundance of the P, P+1, P+2 envelope and the P+1 to P ratio. Numerical extrema of each column are col ored green (lowest) and red (highest). Compound %P of Envelope (P+1)/P %P of Envelope (P+1)/P %P of Envelope Increase (P+1)/P Increase LDI+ MALDI+ From LDI+ to MALDI+ anth raquinone 30.7 1.95 6.69 11.9 0.218 6.08 1,2 dihydroxy anthraquinone 33.6 1.77 10.0 7.48 0.298 4.23 1,2,4 trihydroxy anthraquinone 53.1 0.762 24.3 2.65 0.459 3.48 1,5 dihyrdroxy anthraquinone 38.7 1.32 11.9 6.31 0.307 4.78 2,6 dihydroxy ant hraquinone 29.3 2.08 30.2 1.93 1.03 0.932 1,5 diamino anthraquinone 55.9 0.657 37.9 1.24 0.678 1.88 2,6 diamino anthraquinone 36.2 1.50 37.5 1.40 1.04 0.933
87 Figure 2 20. Extents of reduction under both MALDI and LDI conditions as a function of analyte concentration for alizarin (A) and anthraquinone (B) For both figures, the laser power for MALDI and LDI was 30 and 6 0, respectively. The theoretical reduction extent s are 0.154 and 0.153 for alizarin and anthraquinone, respectively.
88 Figure 2 21. Plot of reduction extent as a function of matrix t o analyte ratio for alizarin and anthraquinone.
89 C HAPTER 3 TANDEM MASS SPECTROMET R Y OF CLUSTERS FROM L EAD WHITE Background The use and history of lead is inextricably tied to the developments of metallurgy, art, geology, archaeology, and mass spectrometry. The earliest uses of lead in its elemental, metallic form can be tra ced to circa 6,000 B.C. in the Balkans and Near East, circa 1,500 B.C. in Egypt where it was mixed with tin to form pewter, and the Roman Empire where it was extensively used as a conduit for water (indeed, the term plu mbum ). 3 Regar ding lead in artistic objects, it has seen use as galena (PbS) for the ancient Egyptian eye makeup known as black kohl 2 in Greek and Roman bronze coins and statues 117,118 as litharge and massicot (PbO) for ancient and Art Nouveau glass 119 and as cames for medieval stained glass 117 has been in the pigment known as either lead white or flake white (( PbCO 3 ) 2 Pb(OH) 2 ). Lead white may also b e called hydrocerussite, which is a rarer form of the mineral cerussite (lead carbonate, PbCO 3 ), but its use as a pigment stems from synthetic roots. Lead white has been in use as a pigment for at least two millennia: there are accounts of its use in Chin a, circa 300 B.C., and ancient Greece and Rome where Pliny documented its synthesis. 50,117 Indeed, lead white is one of the oldest known pigments synthetically produced and was the only white pigment used in painted works of art until t he twentieth century development of safer substitutes, such as ZnO and TiO 2 50,120 The process of preparing lead white consists of suspending strips of metallic lead over vinegar (acetic acid, CH 3 COOH) in clay pots stored in a shed with horse manure, to supply carbon dioxide (CO 2 ). After several weeks, the exterior flakes of lead white are
90 scraped off and pulverized. 50 Though less frequently used, Naples yellow is another artistic pigment based on lead, found in the natural mineral lead anti monate (Pb 2 Sb 2 O 7 ). Four other lead containing pigments are chrome orange ( PbCrO 4 Pb(OH) 2 ) chrome yellow (PbCrO 4 ), lead tin yellow (Pb 2 S n O 4 or Pb(Sn,Si)O 3 ) and red lead (Pb 3 O 4 ). In addition to galena, litharge/massicot, and cerussite (vide supra), six ot her minerals have historically been used for the smelting of lead: anglesite (PbSO 4 ), boulangerite (Pb 5 Sb 4 S 11 ), bournonite (CuPbSbS 3 ), laurionite (Pb(OH)Cl), phosgenite (Pb 2 (CO 3 )Cl 2 ), and semseyite (Pb 9 Sb 8 S 21 ). 3 It is from these nine minerals particularly the primary ore galena that quantitative geology and archaeology (archaeometry) were born, made possible by developments in mass spectrometry. Following J. J. Aston showed in 1927 that l ead has multiple isotopes. 2 Although there are many isotopes of lead, by far the most useful are 204 Pb, 206 Pb, 207 Pb, and 208 Pb. 204 Pb is 238 U (t = 4.468 10 9 yr) 235 U (t = 0.7038 10 9 yr), and 232 Th (t = 14.01 10 9 yr) as their respective parent nuclides and half lives of formation. 4 Amazingly, using just the decay scheme of uranium to lead in 1929, Ernest Rutherford, also a student of Thomson, was able to approximate the age of the Earth for the first time, alt hough he was off by a factor of 2. 2 It was not until Alfred Nier developed the first double focusing (i.e., electric and magnetic sectors) mass spectrometer in 1936 that sufficient precision and accuracy was available to allow for mod ern isotopic measurements in geology and archaeology. 2,121 Following digestion i n nitric acid and either thermal ionization (TI) or inductively coupled plasma (ICP) ionization, particular isotopic abundance ratios of lead
91 are used for analyses of lead containing minerals, metals, pigments, and other objects. To determine the age of a material, lead isotope abundance ratios of 207/204 and 206/204, and to a lesser extent 208/204, are used in the equations of the Holmes Houtermans model and subsequently plotted, extrapolated, and compared to reference values. 2,4 To determine the original location of the ore from which lead was extracted, the lead isotope abundance ratios of 208/ 206 and 207/206, and to a lesser extent 206/204, are plotted without formulaic manipulation and subsequently grouped via chemometric methods. 2,4 Mass spectrometric chronologic and geographic determination of lead containing materials both contribute to the provenance of artistic objects. In fact, analysis of the pigment lead white had been shown t o be successful for both dating paintings across five centuries 122 and grouping paintings by their Renaissance creators such as Rubens, Van Dyck, and Rembrandt 120 Yet, the unique relationship between mass spectrometry and lead goes well beyond establishing provenance from four isotopes of lead. For example, in 2008, when cerussite was ionized with laser desorption (LDI) to determine its effect on the ionization of Greco Roman cosmetics, lead was observed to form ions of integral clusters ( n Pb, n = 1 4) up to mass to charge ( m/z ) 1,000. 123 In fact, lead and many other elements a nd organometallic compounds can ionize to form cluster ions well above the expected m/z of their chemical formulae, which should not be confused with oligomeric ions often observed in mass spectra. Indeed, cluster ions and their experimental analyses and theoretical models have been an active field of research for both chemists and physicists since the early 1980s. 124 Lead has played a pivotal role in
92 this research, since clusters of lead were one of the first to ever have been documented in 1982, 125 and has remained important up to the present 126 132 Lead cluster ions, and cluster ions in general, allow for glimpses into the transition of matter from the atom to the bulk 130 because a countable number of atoms can be observed with mass spectrometr ic techniques 124 Cluster ion analysis was first achieved with the condensation of a desired cluster target (e.g., Xe, Pb, In, CO 2 ) i n helium followed by electron ionization (EI) and measurement by a time of flight (ToF) or sector mass analyzer. 125,126,133 The ionization technique has since evolved from thermal desorption with EI or chemical ionization (CI) 127 to secondary io n mass spectrometry (SIMS) 128 and finally to LDI 129 131,134 Many different types of cluster analysis have been pe rformed with mass spectrometry: observation of lead cluster ions up to Pb 110 125 fragmentation of lead cluster ions as a function of EI energy 126 observation of lead acetate clusters up to m/z 633 127 compariso n between lead cluster cations and anions 128 comparisons among different bime tallic clusters with lead (viz., M/Pb, M = Au, Ag, Cu) 129 fragmentation pathways of Sn N Pb and Pb N Sn clusters 130 and observation and fragmentation of lead acetate cluster ions up to m/z 8,000 131 The aforementioned techniques used for cluster ion analysis have allowed for greater understanding of the formation of bulk matter and the application of mass 135 and isotopic distributions. Cluster ion analyses has contributed to theoretical models such as geomet ric packing of atoms 125,128,129 stability of clusters based on valence electron cou nting 124,128,129,134 the ster 136 and magic numbers 125,128,129,133 Magic numbers, which were first observe d in 1981 and
93 are symbolized as n* are the cluster integers of relatively high abundance immediately preceding a drop of relative abundance (by about a factor of two) of the next highest integer ( n* +1 ). 133 Low mass magic numbers for cationic clusters of pure lead (Pb n + ) are 7, 10, 13, 17, and 19 125 Since lead has four valence electrons and the cationic clusters have a net charge of +1, the magic numbers of lead listed above correspond to the following valence electron counts: 27, 39, 51, 67, and 75. The valence ele ctron counts are all odd numbers and, critically, do not fit any of the electron counting models that explain the observation of magic numbers, which were succinctly presented in a detailed review by de Heer. 124 This discrepancy will not be further explored since it is beyond the scope of this work. Quite relevant to the work presented in this Chapter are organometallic clusters, partic ularly those with lead. In 1989, Bus c h et al. reported on the first observations of lead organic clusters from lead acetate (Pb(OAc) 2 OAc = C 2 H 3 O 2 ), along with clusters from acetates of magnesium and mercury, using thermal desorption with EI and CI on a sector analyzer with mass resolution of 1,000 or 7,500. 127 The EI generated spectrum did not contain clusters; in fact, the molecular ion was not observed, only a few fragment ions. However, the ammonia CI generated spectrum had cluster ions as large as Pb 2 (OAc) 3 C 2 O + with the appearance of the C 2 O moiety being both unexplained and unique to the mass spectra of lead acetate. In 2000, organometallic clusters of Rh 6 (CO) 16 were recorded with LDI on a ToF. Anionic clusters of (Rh 6 ) n (CO) y were observed from n = 1 10 with y varying from losses of CO a t the source. 134 The appearance of rhod ium atoms in core remained intact. Although clusters appeared up to m/z 8,000, mass assignments
94 of the ions were not difficult because rhodium has only one stable isotope. As mentioned above, in 200 8, clusters anions from lead carbonate were observed up to 4Pb with LDI ToF MS. 123 Only two ions were identified (PbO and PbO 2 ) and scant observations were noted since the LDI of lead carbonate was used solely to determine spectral suppression of organic dyes; this was a minor experiment subordinate to the major aim of the project. The most important work on lead orga nic clusters was a seemingly benign article published in 2010 in Rapid Communications in Mass Spectrometry 131 Lead acetate was analyzed in positive and negative modes and with tandem mass spectrometry (MS/MS) of selected ions using LDI on a quadrupole ToF (QToF) up to m/z 8,000. Interestingly, in the same mass spectrum, clusters were observed for both Pb n ( OAc) y up to Pb 5 O 5 ( OAc) + ( m/z 117 7); and Pb n up to Pb 38 + ( m/z 7,904). For the Pb n + clusters, magic numbers of 7, 10, 13, 17, and 19 were observed and, surprisingly, there was no signal for the Pb 14 ion. Tandem mass spectrometry of Pb 4 + Pb 6 + and Pb 13 + showed only the loss of one or t wo Pb. MS/MS of m/z 1038, along with isotopic pattern matching, was able to differentiate between two isobaric ions: Pb 5 + and Pb 4 O 2 (OAc) 3 + Lastly, in 2012, a unique application of lead containing organometallic ions was published using nano electrospray ionization with a hybrid linear quadrupole ion trap orbital trap mass spectrometer (nano ESI LIT Orbitrap MS). 137 The metal binding oligopeptide phytochelatin (PC n ), which participates in metal detoxification in plants and algae, was observed to form small clusters with lead up to Pb 2 (PC 2 ) 2 + at m/z 1491. Clusters were positively identified not only by the high resolving power and accurate mass provided by the Orbitrap, but also by isotopic pattern matching, as will be discussed further below. In
95 contrast to all of the above cited works, despit e lead white being continually used since antiquity 117 and the most important of all white pigments 50 other than its analysis by ICP 120 or TI 122 for provenance, or acting in a minor test on its spectral effect on organic dyes by LDI 123 it has astoundingly not been the sole su bject of any published experiment for cluster analysis or any other analysis whatsoever. The mass spectrometric analyses of clusters from pure lead and lead organic compounds coincided with developments of new ionization methods that heralded a new era in mass spectrometry that was made possible an expanded m/z range and insight to isotopic distributions. Prior to the early 1980s, ions of m/z greater than 1,000 were seldom observed in spectra due to limitations of available ionization techniques, primaril y EI, CI, and TI. With the proliferation of new methods such as field desorption, m/z 135,138 New and improved mass analyzers, detectors, and electronics, such as the Fourier transform ion cyclotron resonance mass spectrometer (FTICR MS), also enhanced the analyses of middle molecules by offering resolving p owers sufficient to separate individual peaks among a distribution of ions created from 135,138 142 The increased awareness of isotopic envelopes lead Fenselau et al. to strictly define four its respective e lements 139 : 1) nominal mass, which is calculated from the whole number, integer value of the most abundant isotope
96 3) most abundant mass, which is the ion o f greatest abundance within an isotopic envelope weighted average mass and may not represent a real ion. For example and as shown in Figure 3 1, the four types of mass for the theoretical spectrum of lead white (( PbCO 3 ) 2 Pb(OH) 2 ) are: 778 for the nominal mass, which does not represent a real ion; 777.894 for the monoisotopic mass; 775.9 for the most abundant mass; and 775.6 for the average mass, which does not represent a real i on. Mass spectrometry of clusters and middle molecules brought three questions to the forefront of the science: 1) Why do clusters and middle molecules have an isotopic envelope? 2) Which of the four masses of an isotopic envelope should be used to report the mass of a cluster or middle molecule? 3) Which ion or ions of an isotopic envelope should be selected as the precursor for tandem mass spectrometric analysis? generally have more than one stable isotope with different natural abundances. Therefore, an ion may appear at different m/z abundances due to the combinations of those different isotopes. The mathematical and computer a lgorithms used to predict isotopic envelopes can be cumbersome, so the reader is directed to the excellent review article by Valkenborg et al. 142 What follows a re brief explanations of the mathematics used for isotopic envelopes and how they apply to clusters of lead white. In 1960, John Beynon calculated the probability of occurrence of any one monoisotopic variant of a peptide. 142 His calculation, as applied to one monoisotopic variant of a lead white cluster ( Pb w C x O y H z ) is shown in Equation 3 1,
97 P = Pr ( 208 Pb) w Pr ( 12 C) x Pr ( 16 O) y Pr ( 1 H) z (3 1) where P is t he probability of the monoisotopic variant and Pr is the probability, given by possible variants for lead white (Pb 3 C 2 O 8 H 2 ) based on the total number of permutations is actually 6,718,464 (calculated from 4 3 2 2 3 8 2 2 ) 142,143 Though quite large, this number incorporates all the possibilities of one isotope simply moving to a new position within the same molecule (i.e., structural isotopomer), which would appear at the same exact mass and thus would not be differentiated by a mass spec trometer. So that only observable variants are considered, in 1962 Klaus Biemann sums them stepwise until the molecule is accumulated. 142 It is this stepwise procedure far reaching text. 144 Yet, the stepwise procedur e is laborious, practical only for very small molecules, and relied on whole number rounding of the ratios of isotopic abundances. To use exact abundances and to include middle and large molecules, in 1977 Yamamoto and McCloskey developed the polynomial ex pansion method. 142,145 Applying this method, the i sotopic envelope for a lead white cluster (Pb w C x O y H z ) can be created using Equation 3 2. ( 204 Pb + 206 Pb + 207 Pb + 208 Pb) w ( 12 C + 13 C) x ( 16 O + 17 O + 18 O) y ( 1 H + 2 H) z (3 2) Each isotopic term for an element is the probability of that iso tope, which is just the natural abundance. The superscripts w x y z
98 the cluster. Though a significant improvement over the stepwise method, polynomial expansion requires significant computation because of the ma ny possibilities (6,718,464 for lead white) and those possibilities with the same exact mass must then be combined; once combined, the total number of possibilities will be significantly reduced and can be calculated using combinatorial algebra to find the number of simplified terms. For a polynomial expansion of two, three, and four terms, the respective formulae are shown in Equation 3 3, and are already multiplied to find the total number of simplified terms (3 3) where r is the power of the binomial term, s is the power of the trinomial term, and t is the power of the quadrinomial term. The number of simplified terms for lead white using Equation 3 3 yields 8,100 different permutations. Note that since lead white has two elements with o nly two stable isotopes (C and H), the binomial term must be included twice. To ease the burden of computation and of combining the isobaric masses, in 1983 Yergey developed the method of multinomial expansion. 142,143 Rather than collecting identical masses as with polynomial expansion, multinomial expansion calcula tes multinomial coefficients, which are equal to the frequency a variant appears in the expansion. The equation for multinomial expansion, which is a direct corollary to 146 Equation 3 4 shows the general form of a multinomial expansion for an element that has two isotopic variants (e.g., C).
99 (3 4) For Equation 3 4, a and b are the abundance of the isotopic variants and n is the number of atoms of that element in the molecule. Certainly, the left side of the Equation is just a binomial; but, the right side of the Equation is the sum of multinomial prob ability like terms. Equation 3 5 shows the general form of a multinomial expansion for an element that has three isotopic variants (e.g., O) and is derived as a series of binomial expansions. 146 (3 5) For Equation 3 5, the variables from Equation 3 4 are conserved, and c is the third isotopic variant. Similarly, Equation 3 6 shows the general form of a multinomial expansion for an elemen t that has four isotopic variants (e.g. Pb). (3 6) For Equation 3 6, the variables from Equations 3 3 and 3 4 are conserved, and d is the fourth isotopic variant. The multinomial expansions indicated in Equation s 3 4, 3 5, and 3 6 are o nly used to find the probability from one isotopic variant from only one element. To obtain the probability for an entire molecule, following reassignment of the proper variables, the probabilities for the element specific isotopic variants must be multip lied. For larger middle molecules, and certainly for biomolecules such as
100 proteins, polynomial and multinomial expansions can result in a combinatorial explosion and, hence, require significant amounts of computation time and memory. Therefore, methods f pruning of isotopic variants of low abundance and binning of isotopic variants with differences of mass far below the resolving power of a mass spectrometer. 142 Although the methods of polynomial and multinomial expansion most accurately describe the fundamental mechanisms that create an isotopic envelope and are sufficient for most appli cations, more advanced techniques and algorithms have been introduced to mitigate the expense of computation. Most noteworthy are the methods by Alan Rockwood, who devised aggregation of isotopic variants with the same nucleon number and convolution by FT techniques, among others. 142 Unfortunately, expansion. Finally, it should be m entioned that all of the above methods for calculating the matrix of masses and their abundances. The graphical representation of a spectrum, similar to that generate d by a mass spectrometer, is achieved by applying a Gaussian or Laplacian approximation over the pattern spectrum with a user defined resolution. 142 The second q uestion brought about during the era of middle molecules asks which of the four defined masses outlined above or, more broadly applied, which of the many masses within an isotopic envelope should be used to report the mass of a molecular ion. Certainly, f or the majority of small molecule mass spectrometry, only the most abundant mass is provided, which is indeed appropriate since most elements have
101 either one isotope with a relative abundance far greater than the other isotopes (e.g., C, N, O, H, S) or onl y one isotope (e.g., F, Na, P). For most small molecules the most abundant mass is the first (lowest mass) peak in an isotopic envelope and also the monoisotopic mass. For mass analyzers of low resolving power this is often indistinguishable from the no minal and average masses. Once elements containing isotopes with relative abundances that are not dominated by the low mass isotope (e.g., Cl, Br) are introduced into a small molecule, however, the most abundant mass may no longer be the monoisotopic mass The effect of reducing the prominence of the monoisotopic mass is much more pronounced when elements such as lead are introduced, whose most abundant isotope is actually the isotope of highest mass, as shown with the following relative abundances 147 : 204 Pb, 1.4%; 20 6 Pb, 24.1%; 20 7 Pb, 22.1%; 20 8 Pb, 52.4%. Moreover, when molecules contain an increasingly larg e number of a particular atom, the isotopes of low relative abundance have an increasingly large effect on the relative abundance of the monoisotopic and most abundant mass. 135,140 For example, when a hypothetical molecule composed solely of carbon reaches 92 atoms, the abundance of t he 12 C 9 1 1 3 C peak is only 0.538 % lower than the 12 C 9 2 peak; when 93 carbons are reached the 12 C 9 2 1 3 C peak overtakes the 12 C 9 3 peak as it then becomes 0.543% higher The aforementioned effects are clearly apparent with lead sly shown in Figure 3 1. For middle to large mass molecules, the average mass was typically reported prior to the advent of high resolving power instruments. 139 But since the implementation of FTICR for middle to large mass molecules particularly biomolecules such as proteins and large peptides report ing the appropriate mass became a priority. 140,141 In
102 the early days of high resolution MS, the most abundant mass was reported; although accurate, th e actual mass may have had up to a one Dalton error due to incorrectly assigning contributions from heavy isotopes. 140 What remains is still the monoisotopic mass as the most reliable mass to report for middle to large mass molecules. But, as noted above, as the number of atoms increase, the monoisotopic mass sharply decreases in relative abundance. However, the monoisotopic mass as it is known suffers from the bias in assuming that only the lowest mass isotopes are relevant. being composed of only t he highest mass isotopes can also be just as useful, assuming of course, that it, too, is of sufficient abundance. 148 This monoisotopic mass composed of highest mass isotopes should be particularly crucial to exploit for molecules containing lead since its isotopes have unusual relative abundances. Lastly, monoisotopic masses are useful for mi ddle to large mass molecules because they do not suffer from natural isotopic variance. 140,141 For instance, although the relative abundances of 12 C and 13 C are commonly cited as which are composite values and may not represent the true abundance from any single terrestrial source, according to the International Union of Pure and Applied Chemistry (IUPAC). 147 The isotopic abundances of 12 C and 13 C can actually have a range of 98.85 99.02% and 1.15 0.98%, respectively. 147 So, an ion composed of only one isotope from each of its c onstituent elements should not vary from among different terrestrial samples or portions of that same sample. The third question brought about during the era of middle molecules asks which of the ions within an isotopic envelope should be selected as the p recursor/parent ion for
103 fragmentation in MS/MS experiments. Overall, the selection of a monoisotopic peak is preferred so that daughter ion spectra will be minimally complex and is routinely performed on small molecules. 148,149 Yet, fragmentation of non monoisotopic peaks may be preferable either if the monoisotopic peaks lack sufficient abundance or if increased information about the chemical composition and structure o f the precursor is desired. 149 The latter reason has been well documented in the literature with experimental evidence 141,150 156 and mat hematical formulae 148,149,152,153,157,158 The information gathered from the fragmentation of non monoisotopic peaks may be well considered with the following two exemplar, seminal works. In 1982, McLafferty et al. published the claim that product ions, following MS/MS of a single non monoisotopic ion from a precursor with a multi isotope element, will have a different isotopic envelope pattern than either full scan fragments (from EI) or even the natural ulti isotope element. 151 Yet, th e work was shown only for small molecules with simple, two isotope elements such Br and Cl. In 1983, Cooks the different isotopic patterns acquired from varying scan mo des (i.e., neutral loss, parent, and daughter) on a triple quadrupole MS. 152 The cl aim was made that in daughter ion spectra, fragment ions have isotopic ratios that reveal both the variant of the parent and the number of atoms of the multi isotopic element in the neutral fragment. Accordingly and critically important if a monoisotopic variant was selected as the parent, the daughter ion would show only as a single ion. Subsequently from the two works mentioned just above, several publications were spawned with both expanded mathematical theories and applications. 153 155,158
104 Lastly, fragmentation of not one ion from an isotopic envelope, but indeed, the entire isotopic envelope itself, was surmised as a valid technique by Rockwood. 148 Although he discussed that fragmentation of an envelope would lead to a complicated spectrum, he also mentioned reasons when such a process is desired, including to increase sensitivity for triple quadrupole MS/MS, and when the monoisotopic peak is of low abundance with FTICR MS/MS. It should also be added that non selective precursor fragmentation also occurs in some experiments with either skimmer con e fragmentation 29 during ESI, or with post or in source decay 17 duri ng matrix assisted laser desorption/ionization (MALDI) ToF MS. Considering the richness of information from the peculiar relative abundances of the isotopes of lead, fragmentation of the entire isotopic envelope of lead organic clusters should prove fruit ful. The thorough historical and literature review above, regarding lead and its intimate connection to some underlying aspects of mass spectrometry, provides the foundation onto which the experimental results herein will be placed as capstone. Preceding advances in analyses of lead and lead organic clusters, theories and formulae of isotopic distributions, and considerations of reporting and fragmentation of isotopic envelopes, have allowed for the following results to be appreciated with greater signific pigment lead white by MALDI LIT MS/MS and MALDI LIT Orbitrap MS/MS. Unique clustering was observed and, following fragmentation of complete isotopic envelopes, daughter io n spectra revealed chemical formulae for clusters that would otherwise be difficult to elucidate. Furthermore, high resolution and accurate mass analysis on the O rbitrap, in conjunction with MS/MS, allowed for an alternative, compl e mentary means
105 of cluste r identification. The results have contributed to the breadth of knowledge of organometallic clusters, isotopically complex compounds, and the tandem mass spectrometry thereof. Experimental Methods Chemicals and Materials HPLC grade methanol (MeOH) and wa ter were purchased from Fisher Scientific (Fairlawn, NJ). Lead(II) carbonate basic ((PbCO 3 ) 2 Pb(OH) 2 ), commonly referred to as Preparation of Chemicals A slurry of lead(II) carbonate basic (( PbCO 3 ) 2 Pb(OH) 2 ) was prepared at a concentration of 100 g/mL 70:30 MeOH:H 2 O and stored at room temperature. The slurry was sonicated for approximately 20 minutes just prior to use. Ionization and Instrumental Parameters All samples were analyzed by laser desorption/ionization with an intermediate pressure (70 mTorr) source and a N 2 laser ( = 337 nm) A volume of 1 L of sonicated slurry was spotted on a polished stainless steel MALDI sample plate and allowed to dry unaided under ambient conditions. Las er parameters (i.e., laser energy and number of laser shots) were manually tuned to obtain maximal signal, yet minimize the consequent increase in both baseline and space charge effects. Typical laser energies were approximately 10 J/pulse. The desired ionization metric of laser fluence can only be estimated at 1.3 10 3 J/m 2 d ue to the difficulty in an accurate and precise measurement of the laser spot size, which has an approximate 100 m spot The number of laser pulses for each analytical scan was co ntrolled by the automatic gain control to maximize the total ion signal, with fewer than 10 laser shots being typical. The sample plate was
106 one scan per spot. Experim ents were conducted using a Thermo Finnigan LTQ XL (hereafter LTQ) (San Jose, CA) and a Thermo Finnigan LTQ Orbitrap (hereafter Orbitrap) (San Jose, CA) hybrid instrument. The LTQ was used for initial experiments at relatively resolving power and mass acc uracy using both the normal ( 60 s/Da ) a nd enhanced scan rates. The Orbitrap was used for later experiments at relatively high resolving power and mass accuracy. All of the aforementioned parameters for recording a full scan were kept constant fo r tandem mass spectrometry. All collision induced dissociation (CID) was conducted in the linear ion trap of both the LTQ and Orbitrap instruments. The isolation width (W) for all MS/MS spectra was adjusted to isolate the entire isotopic distribution of a cluster and centered on the most abundant ion. CID parameters were manually optimized to ensure maximal abundance from daughter cluster ions while maintaining a relatively low abundance of the parent cluster ions, and are presented using the instruments unit. The W and CID values for the fragmented clusters were maintained for both LTQ and Orbitrap experiments. The parameters for MS/MS follow, in the form (center mass, W, CID): (430.2, 8, 30), (689.0, 8, 55), (895.0, 8, 45), (1135.0, 14, 75 ), (1358.0, 14, 50). Theoretical spectra were generated with Qual Browser version 2.0.7 (Thermo width half maximum (FWHM) re solution adjusted to match that of an experimental spectrum. Automatically generated formulae were limited to the elements Pb, C, O, and H. After contacting the manufacturer to no
107 avail, it is presumed that the model used to create theoretical spectra is proprietary information. Results and Discussion The identification of ions in a complex spectrum using deductive reasoning alone can present undue difficulties; this is certainly the case in the spectra of lead white. The issue at the core of these diffi culties is the determination of the ion, out of a broad distribution of ions with an isotopic envelope, to be selected for the elucidation of an ion cluster. The compl e mentary approach of inductive reasoning alleviates these difficulties by 1) finding the number of lead atoms to generate a matching isotopic envelope, 2) matching daughter ion clusters from a tandem mass spectrometric spectrum to ion clusters in full scan spectra, 3) exploiting patterns in the formulae of cluster ions to fill in the gaps in the list of elucidated formulae, and 4) matching isotopic distributions between experimentally and theoretically derived spectra. Finally, confirmation of assignments of ion clusters using the superior mass accuracy of the Orbitrap was a fail safe arbite r among competing formulae, with few exceptions. Such an approach was employed to identify a large number of LDI pigment lead white and is reported herein without any precedence in published literature. Full Scan S pectra Analysis LDI generated spectra were recorded on an LTQ in both normal and enhanced scan modes. The slower scan rate of the enhanced mode was used to obtain spectra of increased mass resolution in an attempt to better identify ions. The even slowe r scan rates of zoom and ultrazoom modes caused peak broadening and mass shifts from space charge effects, and thus were not used here. An Orbitrap was ultimately used to
108 obtain superior resolution and mass accuracy. Among the different spectra, mass ass with some exceptions. The spectra obtained from the LTQ and Orbitrap are in Figure 3 2. What is quite apparent is the grouping of ions into what will be termed supercl usters which is analogous to the grouping observed with oligomers. Supercluster delineation was obtained following determination of the isotopic pattern of the most abundant ion cluster within a supercluster, which will be termed the capital cluster All clusters other than the capital cluster within a supercluster will be termed minuscule clusters For example, as seen in Figure 3 2 there are six superclusters of n Pb; six capital clusters with a most abundant ion at (rounded) m/z 208, 430, 689, 89 5, 1135, and 1359; and many miniscule clusters. Interestingly, the maximum abundance of the two lead supercluster and the decreasing relative abundances of the larger superclusters matches the trend that was observed in the published spectra from the LDI generated spectra from lead carbonate 159 and lead acetate 131 Besides the resolution and mass accuracy, there are no other major differences between the two LTQ and Orbitrap generated spectra such as relative intensities, save for the appearance of some low abundance mini scule clusters. Delineation of superclusters was obtained by isotopic pattern matching of the capital cluster to the theoretical isotopic pattern of pure Pb clusters, as shown in Figure 3 3. It is quite apparent that the unique isotopic distribution of l ead ( Figure 3 3, top row) allows for rapid matching and consequent identification of the number of lead atoms in a capital cluster ( Figure 3 3, bottom row), and hence allows for the delineation of superclusters in Figure 3 2. Also, since the heaviest isot ope of lead ( 208 Pb) has a very
109 high relative abundance (52.4%), it is readily identified and, more importantly, allows for identification of the monoisotopic ion in clusters. Moreover, since the isotopes of the other elements in lead white have low relati ve abundances ( 13 C, 1.07%; 18 O, 0.205% ; 2 H, 0.0115%), they have only small contribution to the monoisotopic ion of a cluster when they are incorporated at experimental resolutions. Therefore, the monoisotopic ion of an experimentally generated cluster can be identified by matching its relative location within the isotopic envelope to the highest monoisotopic ion of a theoretically derived n Pb ion and will be primarily composed of 208 Pb, 12 C, 16 O, 1 H. As observed in the bottom row of Figure 3 3, the monois otopic ion for the superclusters are (rounded) m/z 208, 432, 691, 898, 1139, and 1363 which contain 1, 2, 3, 4, 5, and 6 lead atoms, respectively. Determining the number of lead atoms in the ions of a supercluster and subsequently determining the monoisot opic ion permits the elucidation of chemical formulae of clusters. By selecting the monoisotopic ion, automatic formula generation via the Qual Browser software can commence. Only formulae generated with the lowest mass error for the normal and enhanced scan rates of the LTQ and for the Orbitrap spectra are shown in Table s 3 1, 3 2, and 3 3. What is quite apparent is that limiting the number of lead atoms for a cluster, which was based on supercluster delineation, is indeed required. The most obvious ex ample of the need to limit the number of lead atoms is with the monoisotopic mass of 208.14 in the LTQ spectra, as shown in Table s 3 1 and 3 2. Presumably, m/z 208.14 in the LDI spectra of lead white results only from the Pb + ion and not from C 10 O 5 H 8 + or C 13 O 2 H 20 + as generated from normal or enhanced scan rates, respectively. Furthermore, the need to limit Pb atoms
110 is also shown in the Orbitrap spectrum for the miniscule cluster at the monoisotopic ion of m/z 705.92415 ( Table 3 3). Again, because of supe rcluster delineation, the cluster must be considered as Pb 3 O 5 H 3 + and not Pb 2 C 12 O 9 H 2 + despite the former assignment having a mass error three times higher than the latter. Therefore, by limiting the software to a designated number of lead atoms for a clust er, more reliable formulae can be generated. The assignment of the cluster with the monoisotopic ion at m/z 705.9 is further complicated due to lower mass errors from the LTQ with both normal (Pb 3 C 4 O 2 H 2 + 2.40 ppm) and enhanced (Pb 3 C 4 O 2 H 2 + 4.60 ppm) scan rates, as compared to the higher mass error from the Orbitrap (Pb 3 O 5 H 2 + 6.463 ppm). All of the assignments have the required number of lead atoms, but vary with the number of carbon and oxygen atoms. Two methods were devised to reconcile this conflict: isotopic pattern matching and inductive formula generation. The first method was that of isotopic pattern fitting whereby experimentally and theoretically generated spectra were compared, as shown in Fi gure 3 4, which has precedence in the literature 131,137 Unfortun ately, spectral differences regarding either peak presence or relative abundances are not great enough for unambiguous visual matching and, hence, the assignment is inconclusive. The second method, inductive formula generation, which is essentially fitting to the trend, may be an adequate arbitrator. For instance, the formula trend within a supercluster in Table 3 3 appears to be Pb x O y H z rather than Pb x C w O y H z Therefore, it is proper to assign the cluster at m/z 706 as Pb 3 O 5 H 2 Lastly, MS/MS provides an additional level of confirmation, as will be later shown in the discussion.
111 The very large error ( for the miniscule cluster with the monoisotopic ion at m/z 246.96251 in the Orbitrap spectrum indicates that the generated formula is erroneous ( Table 3 3). Considering that that ion is 22 Da higher than the monoisotopic ion of the minuscule cluster immediately preceding it (i.e., PbOH, m/z 224.98037) the presence of sodium (minus a hydrogen atom) was supposed. So, one sodium atom was added to override the originally designated pool of potential isotopes used for automatic formula generation which resulted in the formula of PbONa with the low error of 7.038 ppm; therefore the formula for that mass will be identified as such. Building upon the formula of PbONa, the minuscule cluster with the monoisotopic ion at m/z 264.7330 in the Orbitrap s pectrum was also assigned to include one sodium atom, which then makes apparent the addition of water ( 18 Da). Therefore, the ion at m/z 264.7330 will be assigned the formula PbO 2 H 2 Na, which reduces the error from 3.720 to 1.963. The large errors for th e clusters with the monoisotopic masses at (rounded) m/z 208, 225, 247, 265, 432, and 449 in the LTQ spectrum ( T able s 3 1 and 3 2 ) were initially surprising; yet, a plausible reason for such errors of these low m/z ions may be that they lie outside of the calibrated range. The LTQ uses a mixture of six peptides for normal mass range calibration: MRFA, bradykinin 1 7, bradykinin, angiotensin I, neurotensin, and rennin substrate that have [M+H] + ion s at the monoisotopic m/z of 524.27, 757.40, 1060.7, 1296.69 1672.92, and 1758.93, respectively; thus, a reasonable assumption for those large errors is that the lack of a calibrant below the m/z of MRFA. Indeed, as the experimental masses deviate from MRFA (i.e., from m/z 449 to 208), their error concomitantly i ncreases in an almost linear correlation.
112 Therefore, using the known exact mass of the 208 Pb ion as a post acquisition lock mass may provide an alternative method to obtain lower errors for those low m/z clusters; but, this method was not tried here since the Orbitrap had already provided reliable identifications with low error. Tandem Mass Spectrometric Analysis Tandem mass spectrometry was conducted on all six capital clusters with both the LTQ at the normal scan rate and the Orbitrap, as shown in Figure s 3 5 to 3 9. As discussed in the articles by McLafferty 151 and Cooks 152 mentioned in the introduction, daughter ions have a different isotopic pattern from that of their parent ion that was being derived from a combination of the isotopes included in isolation window of the parent i on. Considering the cluster ions of lead are isotopically complex to begin with, the entire isotopic envelope of a parent ion cluster of lead white was selected for fragmentation via a large isolation window. As shown in the tandem mass spectra, using a large isolation window resulted in many daughter ion clusters retaining their isotopic distribution observed in the full scan spectra. However, some daughter ion clusters did not retain the same isotopic distributions observed in the full scan spectra, whi ch will be mentioned as they appear in the discussion below. The most readily plausible reason for that phenomenon is from signal attenuation of the parent cluster ions that resulted from either the isolation or CID steps. For instance, Reich et al. publ ished a convincing set of experiments whereby it was shown that the relative abundances of two daughter ions that were three Da apart, which came from their respective parent ions of the target analyte and the tri deuterated form of that analyte, varied as a function of the isolation width used prior to CID of the
113 parent ions. 160 Moreover, they ha ve shown that the isolation step, prior to CID, was the likely culprit since eliminating CID still caused the effect on the daughter ions. In short, selecting a sufficiently large isolation width is a critical procedure; any deviant daughter ion clusters shown in the tandem mass spectra herein was due, in part, to not thoroughly following that procedure. The two LTQ analyzed daughter ion cluste rs for the capital cluster at (rounded) m/z 432 ( Figure 3 5) appear at m/z 208.00 and 225.00. Their isotopic distribution patterns match those found for the same clusters in the full scan spectrum thereby confirming a proper isolation width was used and a iding in the identification of the monoisotopic mass. The Orbitrap was not used to an acquire MS/MS spectrum for this same capital cluster. The tandem mass spectra for the capital cluster at (rounded) m/z 691 are shown in Figure 3 6. The striking differe nce between the LTQ ( Figure 3 6A) and Orbitrap ( Figure 3 6B) spectra is the appearance of the daughter ion cluster with high monoisotopic mass at (rounded) m/z 449. Although this daughter ion cluster may be safely assigned as Pb 2 O 2 H + as shown in Table 3 3, its appearance in only the Orbitrap spectrum might be explained by an ion molecule reaction with water that may have occurred either within the ion trap or the C trap of the instrument, or anywhere in between. Although the reaction is very unlikely, th e reasoning is plausible since the difference in mass is 17 Da, when comparing the daughter ion cluster ion seen in the LTQ ( m/z 432 ) to the daughter ion cluster seen in the Orbitrap ( m/z 449). However, upon closer inspection of the isotopic distributions of the daughter ion clusters in the LTQ and Orbitrap spectra, only the latter has a distribution similar to a two Pb ion cluster, whereas the former does
114 not Therefore, the abundances observed in the LTQ daughter ion clusters were influenced by an inade quate isolation width. Lastly, the LTQ cluster at m/z 673 was also observed with the Orbitrap and has a distribution exactly matching that found in the full scan. The tandem mass spectra for the capital cluster at (rounded) m/z 897 are shown in Figure 3 7 Other than the daughter ion cluster with a most abundant ion at m/z 876, which has an indistinguishable monoisotopic ion, the LTQ spectrum was of very low signal and required significant amplification of the daughter ion clusters. Only the cluster at m /z 656 has an isotopic distribution exactly matching that found in the full scan. The Orbitrap spectrum shows only one daughter cluster ion with a discernable monoisotopic ion at m/z 449, presumably Pb 2 O 2 H + which was also observed with the LTQ. The tande m mass spectra for the capital cluster at (rounded) m/z 1139 are shown in Figure 3 8. For the LTQ spectrum, only the daughter ion cluster at the monoisotopic ion at m/z 897 both possessed an isotopic distribution matching that found in the full scan and w as observed with the Orbitrap. The other clusters did not have distinguishable monoisotopic ions among their isotopic envelope. For the Orbitrap spectrum, only the cluster at m/z 706 had an isotopic distribution matching that found in the full scan spect rum. The tandem mass spectra for the capital cluster at (rounded) m/z 1363 are shown in Figure 3 9. For the LTQ spectrum, only the daughter ion cluster at m/z 1120 has a distinguishable monoisotopic ion among their isotopic envelope. For the Orbitrap
115 spe ctrum, the daughter ion cluster at m/z 897 both possessed an isotopic distribution exactly matching that found in the full scan and was observed with the LTQ. The daughter ion clusters from tandem mass spectra are tabulated in Table 3 4 and 3 5. Blank ent ries are purposefully included to align the rows among both tables and to omit information for clusters with a monoisotopic ion that is indistinguishable among the isotopic envelope of a cluster. Entries in bold indicate clusters that are found in both LT Q and Orbitrap spectra, but not necessarily produced from the same parent ion cluster. Overall, the tandem mass spectrometric analysis provided further confirmation for mass assignments by identifying clusters outright or allowing for the inductive determ ination of formulae, as will be discussed next. Final Mass Assignments Table 3 6 provides the final chemical formulae for lead white as elucidated from the full scan and MS/MS spectra following both LTQ and Orbitrap analyses. The most reliable determinant was the low error, full scan spectra obtained from the Orbitrap. Yet, the two clusters at the monoisotopic mass of (rounded) m/z 706 and 1139 had lower mass errors from the LTQ, rather than the Orbitrap. Also, the formula for the cluster at m/z 247 was clearly not generated correctly and was altered to include one sodium atom. The formula for the cluster at m/z 265 was also altered to include one sodium atom. The cluster at m/z 897 was the only formula that was common among both analyzers, and with low mass errors as well. The cluster at m/z 482 was the only one observed only with the Orbitrap and not the LTQ. The clusters at m/z 656, 674, 1104, and 1122 were observed only with the LTQ and not the Orbitrap; therefore, their formulae required confirmat ion by other means. For instance, the clusters at m/z 656 and 1104 were observed and their formulae confirmed via MS/MS ( Table s 3 4 and 3 5).
1 16 The clusters at m/z 674 and 1122 were observed in the MS/MS spectra from the Orbitrap at one Da lower ( Table 3 5 ), which was attributed to the difference of one hydrogen atom. Therefore, the clusters at m/z 674 and 1122 were inductively assigned a formula based on daughter ion clusters with a very near m/z which also allowed their formulae to fit the trend apparen t in Table 3 6, namely Pb x O z H z As shown in Table 3 6, there are two clusters whose formula assignments were inconclusive: (rounded) m/z 1346 and 1363; their formulae were not able to be definitively determined using either the above deductive or inductive reasoning. Since the cluster at m/z 1346 was merely a miniscule cluster, it was the capital cluster at m/z 1363 whose correct formula was most desired and lack thereof troublesome. The LTQ spectrum supports a formula of Pb 6 O 7 H 3 with a much higher error ( 58.40 ppm) than the formula supported by the Orbitrap spectrum, Pb 6 C 4 O 4 H 3 ( 1.257 ppm). Though the former formula fits the trend observed in Table 3 6 (i.e., Pb x O y H z ), its high error precludes its credible use. Since the cluster at m/z 1363 is itself th e capital cluster of highest m/z with abundance easily detectable among the surrounding spectral noise, it of course, cannot be detected as a daughter ion cluster in an MS/MS spectrum. Consequently, the remaining reasonable method to determine its formula is with matching of its isotopic pattern, as shown in Figure 3 10. Unfortunately, any cluster within the six lead supercluster ( Figure s 3 3K and 3 3L) has an isotopic profile dominated by the distribution of the six lead isotopes, whereby the other atoms (i.e., C, O, and H) have only a marginal effect on the profile. Consequently, the attempt at matching isotopic patterns in Figure 3 10 was not conclusive, though Pb 6 O 7 H 3 would be a reasonable guess.
117 Final mass assignments were in the form Pb x O y H z with th e inclusion of sodium for two low mass clusters The reason for the exclusion of integer clusters of Pb n larger than Pb 1 was not immediately apparent from the data, but high laser energy might have been a cause 161 Furthermore, high laser energy may have also contributed to the exclusion of carbon. Experiments with the LDI generated clusters of Os 3 (CO) 12 resulted in an increasing loss of carbonyl groups following a rise in laser power 161 A s mentioned in the introduction c arbon was included in clusters of lead acetate generated by CI 127 and LDI 131 which was contained mostly within the acetate moiety, but was excluded from LDI generated clusters of lead carbonate 123 Lastly, loss of carbon, in the form of carbonyl, was observed and extensively discussed with the LDI generated clusters of Rh 6 (CO) 16 134 Thus, the omission of carbon from the final mass assignments are not necessarily cause for immediate concern. Conclusion For the first time, the LDI ge nerated, LTQ and Orbitrap analyzed full scan and experimental evidence combined with a variety of deductive and inductive reasoning allowed for the near complete elucidatio n of the peculiar mass spectrum of lead white, which was composed of numerous cluster ions with both relatively complex isotopic distributions and wide isotopic envelopes. Mass assignments appeared to fit a general pattern of Pb x O y H z with the exclusion of both carbon and integer clusters of Pb n larger than Pb 1 Key to the successful assignments of formulae was the multi step method used herein. The initial and most critical step was to determine the number of lead atoms of a cluster that allowed for bot h the identification of the monoisotopic ion and a limitation
118 to be placed on potential formulae assigned to a cluster. Second in importance was the high resolution, accurate mass analysis provided from the Orbitrap, which led to very low mass errors and, hence, heightened reliability of mass assignments. Thereafter, the use of daughter ion clusters, pattern recognition in the formulae of clusters, and matching of isotopic distributions also played integral roles, with varying degrees of success. Overall the Orbitrap was the most productive analyzer, though with some exception, and the LTQ in enhanced mode offered no benefit. With the ever increasing use of laser based ionization techniques in the field of conservation science igments will play a crucial role. The first ever analysis of lead white presented herein was an important step toward the goal of conservation scientists to better understand the very complex mass spectra among the milieu of components observed from the s urface of a painted work of art. The spectra present an oft misunderstood or underappreciated analyses of isotopically complex ions in the middle molecule range.
119 Fi gure 3 1. Theoretical isotopic distribution pattern for lead white ((PbCO 3 ) 2 Pb(OH) 2 ). The nominal mass is 778, which does not represent a real ion; the monoisotopic mass is 777.894; the most abundant mass is 775.90: and the average mass is 775.6, which does not represent a real ion. The masses average masses (Pb: 208, 207.976, 207.2; C: 12, 12.000, 12.011; O: 16, 15.994, 15.994; H: 1, 1.007, 1.008). most abundant mass nominal mass monoisotopic mass average mass
120 Figure 3 2. LDI generated, ful l scan, positive mode spectrum of lead white analyzed by LTQ with normal (A) and enhanced (B) scan rates and Orbitrap (C). Lead atom counts are bracketed into superclusters. 1Pb 2Pb 3Pb 4Pb 5Pb 6Pb B C A
121 Figure 3 3. Theoretical and experimental capital clusters. Top row: theoret ical spectra of Pb (A), Pb 2 (C), Pb 3 (E), Pb 4 (G), Pb 5 (I), and Pb 6 (K) with resolutions matching that of its respective portion below. Bottom row: portions of the LTQ analyzed spectrum for the capital cluster in a supercluster, which were matched to the top row to determine the number of PB atoms: 1Pb (B), 2Pb (D), 3Pb (F), 4Pb (H), 5Pb (J), 6Pb (L). Monoisotopic ions are marked with a blue asterisk. A B C D E F *
122 Figure 3 3. Continued G H I J K L 1138.83 * 1362.75
123 Table 3 1. Most abundant and monoisotopic masses of cluster ions from the LT Q analyzed spectrum with the normal scan rate. Formulae generated with both limitations on the number of Pb atoms and with no limitations, with their respective mass accuracies. Entries in bold indicate a formula with a mass error smaller than tha t obtained with the Orbitrap. Resolutions are FWHM. LTQ Normal Scan Rate M ost abundant ion: mass M ost abundant ion: resolution M onoisotopic ion: mass M onoisotopic ion: resolution F ormula with Pb limit F ormula with Pb limit, error (ppm) F ormula with n o Pb limit F ormula with no Pb limit, error (ppm) 208.14 669 208.14 669 Pb 762.44 C 10 O 5 H 8 471.58 225.16 699 225.16 699 Pb C H 5 625.70 C 10 O 6 H 9 518.44 247.14 720 247.14 720 Pb C 3 H 3 560.42 C 8 O 9 H 7 524.42 265.14 782 265.14 782 Pb C 4 H 9 367.94 Pb C 4 H 9 367.94 430.11 1125 432.10 1196 Pb 2 C H 4 267.78 Pb C 10 O 6 H 8 211.89 447.08 1216 449.10 1207 Pb 2 C 2 H 9 177.75 Pb 2 C 2 H 9 177.75 465.06 1217 467.05 1259 Pb 2 C 2 O H 11 36.69 Pb 2 C O 2 H 7 114.59 653.95 1588 656.00 1633 Pb 3 C 2 H 8 6.81 Pb 3 C 2 H 8 6.81 671.99 1702 673.98 1681 Pb 3 C O 2 H 6 21.03 Pb 2 C 10 O 8 H 10 14.80 688.91 1761 690.98 1750 Pb 3 C 5 H 7 5.79 Pb 3 C 5 H 7 5.79 703.96 1573 705.93 1719 Pb 3 C 4 O 2 H 2 2.40 Pb 3 C 4 O 2 H 2 2.40 894.89 2230 896.89 2270 Pb 4 O 4 H 6 .75 Pb 4 O 4 H 6.75 912.81 1937 914.85 1922 Pb 4 O 5 H 3 58.20 Pb 4 O 5 H 3 58.20 1099.86 2760 1103.89 2302 Pb 5 C 4 O 8.16 Pb 4 C 9 O 10 H 4 0.10 1117.79 2561 1121.83 2679 Pb 5 O 5 H 2 39.12 Pb 4 O 18 H 2 1.02 1134.81 2704 1138.80 2544 Pb 5 O 6 H 3 66.07 Pb 4 O 19 H 3 28.54 1149.76 2202 1153.77 2007 Pb 5 O 7 H 2 56.86 Pb 4 O 20 H 2 19.81 1341.76 2922 1345.81 2682 Pb 6 O 6 H 2 26.02 Pb 5 C 4 O 16 H 2 5.59 1358.77 3232 1362.77 3231 Pb 6 O 7 H 3 58.40 Pb 5 O 20 H 3 27.03 1372.79 3071 1376.83 2914 Pb 6 O 8 H 4. 72 Pb 5 C 5 O 17 H 5 1.74
124 Table 3 2. Most abundant and monoisotopic masses of cluster ions from the LT Q analyzed spectrum with the normal enhanced rate. Formulae generated with both limitations on the number of Pb atoms and with no limitations, with their respec tive mass accuracies. Entries in bold indicate a formula with a mass error smaller than that obtained with the Orbitrap. Resolutions are FWHM. LTQ Enhanced Scan Rate M ost abundant ion: mass M ost abundant ion: resolution M onoisotopic ion: mass M onoisoto pic ion: resolution F ormula with Pb limit F ormula with Pb limit, error (ppm) F ormula with no Pb limit F ormula with no Pb limit, error (ppm) 208.25 958 208.28 958 C 13 O 2 H 20 488.07 225.23 1031 225.23 1031 Pb C H 5 963.02 C 17 H 21 303.44 247.20 1048 2 47.20 1048 Pb C 3 H 3 802.28 C 16 O 2 H 23 115.80 265.20 1173 265.20 1173 Pb C 4 H 9 594.77 C 19 O H 21 171.79 430.10 1675 432.06 1735 Pb 2 C H 4 173.61 C 9 O 19 H 20 0.068 447.08 1848 449.05 1849 Pb 2 C 2 H 9 59.48 Pb C 11 O 6 H 13 5.69 465.05 1860 467.05 1924 Pb 2 C 2 O H 11 40.88 Pb C 18 O 2 H 11 1.75 653.94 2453 655.95 2457 Pb 3 C O H 4 11.97 Pb C 21 O 12 H 4 2.07 671.95 2608 673.94 2684 Pb 3 C 4 H 2 4.65 Pb C 17 O 16 H 6 0.03 688.94 2752 690.94 2701 Pb 3 O 4 H 3 10.64 Pb 2 C 5 O 13 H 7 2.24 703.88 2 411 705.93 2283 Pb 3 C 4 O 2 H 2 4.60 Pb C 17 O 18 H 6 0.13 894.86 3498 896.85 3371 Pb 4 O 4 H 47.16 Pb 3 O 17 H 0.50 912.84 2919 914.80 2563 Pb 4 O 5 H 3 114.76 Pb 3 O 18 H 3 68.04 1099.84 4245 1103.86 4740 Pb 5 C O 3 H 2 31.17 Pb 3 C 13 O 20 H 4 0.15 1117.81 4138 1121.79 4064 Pb 5 O 5 H 2 71.57 Pb 4 O 18 H 2 33.47 1134.86 3587 1138.84 3307 Pb 5 O 6 H 3 28.48 Pb 4 C 4 O 16 H 3 4.34 1358.78 3716 1362.80 4534 Pb 6 O 7 H 3 38.04 Pb 5 O 20 H 3 6.68
125 Table 3 3. Most abundant and monoisotopic masses of cluster ions from the Orbitrap analyzed spectrum. Formulae generated with both limitations on the number of Pb atoms and with no limitations, with their respective mass accuracies. Entries in bold indicate a f ormula with a mass error less than that obtained with the LTQ. Resolutions are FWHM. Orbitrap Scan M ost abundant ion: mass M ost abundant ion: resolution M onoisotopic ion: mass M onoisotopic ion: resolution F ormula with Pb limit F ormula with Pb limit, e rror (ppm) F ormula with no Pb limit F ormula with no Pb limit, error (ppm) 207.97738 175890 207.97738 175890 Pb 3.590 Pb 3.590 224.98034 168880 224.98034 168880 Pb O H 4.290 Pb O H 4.289 246.96251 160462 246.96251 160462 Pb C 3 H 3 152.240 C 3 O 13 H 3 20.8 50 264.97330 153630 264.9733 0 153630 Pb C 2 O 2 H 3.720 Pb C 2 O 2 H 3.720 429.94844 127030 431.95024 127778 Pb 2 O 6.019 Pb C 12 O 5 0.969 446.95108 126333 448.95311 121544 Pb 2 O 2 H 6.081 Pb C 12 O 6 H 0.649 464.96185 123026 466.96377 123188 Pb 2 O 3 H 3 6.046 Pb C 12 O 7 H 3 0.426 479.94898 122068 481.95103 119401 Pb 2 O 4 H 2 6.215 Pb C 12 O 8 H 2 0.056 688.93537 98763 690.93735 98948 Pb 3 O 4 H 3 7.028 Pb 2 C 12 O 8 H 3 2.653 703.92221 97571 705.92415 96549 Pb 3 O 5 H 2 6.463 Pb 2 C 12 O 9 H 2 2.182 894.89818 87266 896.90029 85496 Pb 4 O 4 H 7.589 Pb 3 C 5 O 13 H 5 2.329 912.90868 84620 914.9106 80490 Pb 4 O 5 H 3 7.168 Pb 3 C 5 O 14 H 7 2.555 1134.88069 71355 1138.88424 66273 Pb 5 C 4 O 3 H 3 5.801 Pb 4 C 5 O 15 H 7 0.216 1149.86766 66892 1153.87021 49567 Pb 5 O 7 H 2 6.525 Pb 4 C 5 O 16 H 6 1.184 1358.85293 52412 1362.86069 47783 Pb 6 C 4 O 4 H 3 1.257 Pb 4 C 17 O 20 H 7 1.192 1373.83830 56967 1377.81948 58534 Pb 6 O 8 H 2 10.705 Pb 5 C 8 O 15 H 2 1.826
126 Figure 3 4. Isotopic pattern matching for the clus ter with the monoisotopic mass of 706. A: theoretical spectrum of Pb 3 C 4 O 2 H 2 with matching resolution (1719) of the LTQ analyzed cluster (B). C: theoretical spectrum of Pb 3 O 5 H 2 matching resolution (1719) of the LTQ analyz ed cluster (B). D: theoretical spectrum of Pb 3 C 4 O 2 H 2 with matc hing resolution (96549) of the Orbitrap analyzed Pb 3 O 5 H 2 cluster (B). F: theoretical spectrum of Pb 3 O 5 H 2 with matching resolution (96549) of the Or bitrap analyzed cluster (E). A B C D E F Pb 3 C 4 O 2 H 2 LTQ Pb 3 O 5 H 2 Pb 3 C 4 O 2 H 2 Orbitrap Pb 3 O 5 H 2
127 Figure 3 5. LTQ analyzed daughter ion clusters from MS 2 of the cluster with the monoisotopic mass of 432 Da Inset: enlargement of a portion of the spectrum x20 x20
128 Figure 3 6. Daughter ion clusters from MS 2 of the cluster with t he monoisotopic mass of 691 Da. Top: LTQ analyzed spectrum. Bottom: Orbitrap analyzed spectrum. Insets are enlargements of portions from their respective spectrum.
129 Figure 3 7. Daughter ion clusters from MS 2 of the cluster with the monoisotopic mass of 897 Da. Top: LTQ analyzed spectrum. Bottom: Orbitrap analyzed spectrum. x25 x50 x50
130 Figure 3 8. Daughter ion clusters from MS 2 of the cluster with the monoisotopic mass of 1139 Da. Top: LTQ analyzed spectrum with enlargement of a portion (inset). Bottom: O rbitrap analyzed spectrum. x10 x15 x15
131 Figure 3 9. Daughter ion clusters from MS 2 of the cluster with the monoisotopic mass of 1363 Da. Top: LTQ analyzed spectrum. Bottom: Orbitrap analyzed spectrum. x10 x10 x10 x10 x10
132 Table 3 4. LTQ analyzed daughter ions following MS/MS of parent ion clusters (left column). Entries in bold are daughter ion clusters also found with the Orbitrap. Resolutions are FWHM. LTQ MS/MS M onoisotopic parent ion: mass M ost abundant daughter ion: mass M ost abundant daughter ion: resolution M onoisotopic ion: mass M onoisotopic ion: resolution F ormula with Pb limit F ormula with Pb limit, error (ppm) F ormula with no Pb limit F ormula with no Pb limit, error (ppm) E xact match in full scan? 432.10 208.02 701 208.02 701 Pb 43.39 C 6 O 8 H 8 1.89 Y 225.03 694 225. 03 694 Pb C H 5 16.41 C 6 O 9 H 9 7.52 Y 690.98 429.97 1257 431.96 1232 Pb 2 O 13.67 Pb C 5 O 10 H 4 4.78 461.21 1063 670.91 1780 672.90 17801 Pb 3 O 3 H 26.03 Pb 2 C 8 O 19 H 7.85 Y 896.89 446.97 13612 44 8.98 1367 Pb 2 C O H 5 12.11 Pb 2 C O H 5 12.11 653.91 1672 655.96 1554 Pb 3 O 2 17.38 Pb 3 O 2 17.38 Y 671.00 1311 673.02 1189 Pb 3 C 2 O H 9 37.11 Pb 3 C 2 O H 9 37.11 743.12 1591 875.85 2253 1138.80 894.85 2172 896.85 2256 Pb 4 O 4 H 50.36 Pb 4 O 4 H 50.36 Y 1097.81 2797 1116.79 2711 1362.77 892.82 2310 1115.77 2623 1119.77 2846 Pb 5 O 5 79.73 Pb 5 O 5 79.73 1271.92 2504 1295.96 2503 1319.94 3331 1335.75 3139
133 Table 3 5. Orbitrap analyzed daughter ions following MS/MS of parent ion clusters (left column). Entries in bold are daughter ion clusters also found with the LTQ. Resolutions are FWHM. Orbitrap MS/MS M ono isotopic parent ion: mass M ost abundant daughter ion: mass M ost abundant daughter ion: resolution M onoisotopic ion: mass M onoisotopic ion: resolution F ormula with Pb limit F ormula with Pb limit, error (ppm) F ormula with no Pb limit F ormula with no Pb lim it, error (ppm) E xact match in full scan? 690.93735 446.94702 33807 448.94907 33279 Pb 2 O 2 H 2.90380 Pb 2 O 2 H 2.90380 668.91625 28338 672.92029 29338 Pb 3 O 3 H 2.43392 Pb 3 O 3 H 2.43392 896.90029 448.95289 35935 448.95289 35935 Pb 2 O 2 H 5.59981 Pb 2 O 2 H 5.59981 875.32747 24298 1138.88424 703.91532 27367 705.91652 27508 Pb 3 O 5 H 2 4.33284 Pb 3 O 5 H 2 4.33284 Y 89 4.88976 26783 896.89626 28632 Pb 4 O 4 H 3.10507 Pb 4 O 4 H 3.10507 1014.90878 24607 1014.90878 24607 Pb 6 C 7 O 6 H 3 9.63710 1099.85697 23135 1103.85929 23020 Pb 5 O 4 2.72030 Pb 5 O 4 2.72030 1116.85854 22087 1120.86199 21960 Pb 5 O 5 H 2.71315 Pb 5 O 5 H 2.71315 1362.86069 894.88970 5954 896.89170 6204 Pb 4 O 4 H 1.98059 Pb 4 O 4 H 1.98059 Y 1338.82777 5263 1342.82692 5345 Pb 6 C O 5 H 3 22.63621 Pb 6 C O 5 H 3 22.6 3621
134 Table 3 6. Final elucidation of observed cluster ions of lead white. Entry in italics: cluster observed with the Orbitrap, but not the LTQ. Entries in bold: clusters observe d with the LTQ, but not the Orbitrap. Formulae in pink: determined solely by MS/MS. Formulae in blue: inductively determined after MS/MS. Key for column of determinants: O, Orbitrap; Oo, Orbitrap override; L, LTQ; L2, MS/MS with LTQ; O2, MS/MS with Orb itrap. Final Elucidation of Cluster Ions M ost abundant ion (rounded) M onoisotopic ion (rounded) F ormula D eterminant 208 208 Pb O 225 225 Pb O H O 247 247 Pb O Na Oo 265 265 Pb O 2 H 2 Na Oo 430 432 Pb 2 O O 447 449 Pb 2 O 2 H O 465 467 Pb 2 O 3 H 3 O 480 482 Pb 2 O 4 H 2 O 654 656 Pb 3 O 2 L2 672 674 Pb 3 O 3 H 2 O2 689 691 Pb 3 O 4 H 3 O 704 706 Pb 3 O 5 H 2 L 895 897 Pb 4 O 4 H L, O 913 915 Pb 4 O 5 H 3 O 1100 1104 Pb 5 O 4 O2 1118 1122 Pb 5 O 5 H 2 O2 1135 1139 Pb 5 O 6 H 3 L 1150 1154 Pb 5 O 7 H 2 O 1342 1346 inconclusive 1359 1363 inconclusive 1374 1378 Pb 6 O 8 H 2 O
135 Figure 3 10. Isotopic pattern matching for the cluster with the monoisotopic mass of 1363. A: theoretical spectrum of Pb 6 O 7 H 3 with matching resolution (3231) of the LTQ ana lyzed cluster (B). C: theoretical spectrum of Pb 6 C 4 O 4 H 3 with matching resolution (323 1) of the LTQ analyzed cluster (B). D: theoretical spectrum of Pb 6 O 7 H 3 with matching resolution (47783) of the Orbitrap analyzed cluster (B). F: theoretical spectrum of Pb 6 C 4 O 4 H 3 with matching resolution (47783) of the Orbitrap analyzed cluster (E). A B C D E F Pb 6 O 7 H 3 LTQ Pb 6 C 4 O 4 H 3 Pb 6 O 7 H 3 Orbitrap Pb 6 C 4 O 4 H 3
136 C HAPTER 4 LASER DESORPTION/ION IZATION TANDEM MASS SPECTROM ET R Y OF MADDER AND LEAD WHITE DIREC TLY FROM ARTISTIC SA MPLES Background The mass spectrometric analysis of artis tic samples often garners interest from the general public due to its appeal as a practical technology to preserve objects of cultural heritage, rather than as an unapproachably esoteric chemistry experiment. Such analyses also win acclaim from the scient ific community as way to make known the universal application of chemistry to preserve empathetic objects of beauty. Yet, interest in the analysis of artistic samples must be limited and focused: its novelty lies not necessarily with either the advancemen t or application of new technologies, but with the new sets of information obtained from such analyses for the conservation science community, which includes conservators curators, and art historians. To the analytical chemist, particularly the analytica l mass spectrometrist, artistic samples are simply another genre of analysis, whether it be derived from direct interrogation or following extraction, separation, or derivatization. It is not significantly different from the analysis of neat samples, geol ogical specimens, or biological tissues. Considering the above perspective on the mass spectrometric analysis of artistic samples, any attempt at an exhaustive review of the literature, such as was conducted for C hapters 2 and 3 would be unnecessarily cum bersome. Therefore, only a cursory review shall be provided, focusing on desorption/ionization methods applied directly to artistic samples of textile fibers and painting cross sections. For a primer on the application of analytical chemistry for artisti c samples and, more broadly, for the preservation of art and archaeological objects, the reader is referred to an existing selection of books 2 4,53,117,162,163 on this topic.
137 Desorption/ ionization methods such as laser desorption/ionization (LDI), secondary ion mass spectrometry (SIMS), and direct analysis in real time (DART) have been conducted on samples and surfaces such as fibers 77 textiles 71 newspaper 74 illuminated manuscripts 74,76 ancient cosmetics 123 ceramics 9 statuettes 164 and painting cross sections 74,77,165 170 Other direct MS methods used to characterize art objects include laser ablation inductively coupled plasma MS on glass 119 and matrix assisted laser desorption/ionization (MALDI) on painting fragments 171 Many of the other direct analytical methods for artistic samples are spectroscopic, including Raman, infrared (IR) and its Fourier transform variant (FTIR), X ray fluoresc ence, and particle induced X ray emission (PIXE). 2 4 Crucial for the work presented in this Chapter is the precedence of direct analyses of pai nting cross sections and textiles. Attention must be given to relevant direct analyses with respect to their sample preparation and introduction, since both seem to be the only real novelty gleaned from such experiments. Although seemingly important, ion ization and instrumental settings established in the literature are not necessarily worthy of note for two reasons. First, the majority of direct analyses have been conducted with SIMS for imaging experiments, which is not the method employed here. Secon d, the LDI 74,76,77 analyses followed the standard, long espoused practice of using minimal laser fluence, such th at it is just above the ionization threshold and below the fluence causing in source fragmentation and noise. As Wyplosz mentioned in his thorough dissertation 77 an important aspect to consider for painting cross sections is the surface morphology and localization of pigments or dyes, which may alter
138 spot to spot variation, and the risk of ablating away the sample, wh ich is critically important for precious samples. To account for the peculiar morphology and topography of artistic samples, particularly painting cross sections, and to allow these samples to be effectively introduced into the vacuum chamber of an instrum have generally been employed. The most rudimentary method is to simply affix the unaltered sample to the sample plate using double sided tape, which has been applied to painting cross sections. 74,77 Wyplosz created and used a specially modified sample holder to support fiber samples. 77 painting in linseed oil atop the sample plate. 74 A necessity for SIMS analysis is to create a smooth, flat surface by slicing or poli shing the sample. 9 To aid in the creation of a flat su rface for SIMS, the popular method of first embedding the sample in a polymer resin is often used 164 170 and with some variation such as washing the sample with hexane to remove contam inants 9,165 168 and coating the sample with sputtered gold 165,168 Considering that only LDI was use d in the work presented here, the simple method of affixing unaltered chips from paintings and fabrics with double sided tape was used with adequate results. The analyses were conducted to determine if the pigment lead white and the dye alizarin were dete ctable in chips of both authentic and experimental paintings, and if components of madder (i.e., alizarin or purpurin) were detectable in dyed fabrics of silk. For a comprehensive review of the LDI mass spectrometric analysis of alizarin and lead white, t he reader is referred to Chapter 2 and 3 in this dissertation, respectively. Briefly, the in situ detection by LDI of alizarin was attempted
139 by Wyplosz in his 2003 dissertation. 77 His analysis of alizarin in a painting produced no significant ions, but his analysis of an ancient dyed fiber did show an abundant ion indicative of alizarin; in both samples, tandem mass s pectrometry (MS/MS) was not attempted despite the use of an ion trap. The 2003 dissertation by Grim did show some LDI ToF generated spectra from a paint chip that may have been indicative of lead white, but she was unable to conclude that the signal did n ot originate from two competing lead containing pigments, lead chromate (PbCrO 4 ) or red lead (Pb 3 O 4 ). 74 There has been no published work on the in situ detection by LDI of lead white thus far; what follows is the first time LDI tandem mass spectrometric detection of (laser reduced) alizarin from samples of both a painting and dyed textile, and the first ever LDI detection of lead white from a painting sample, all without any sample preparation whatsoever. Experimental Methods Samples: Painting Fragments and Dyed Silk Swatches Several samples of paintings, from both historical and experimental paintings, and swatches of dyed silk were donated by Julie Arslanoglu conservation scientist at The Metropolitan Museum of Art, New York, NY, and are shown in Figure 4 1. The provided descriptions for samples I VII follows. I: Paint composition: linseed oil and alizarin in a single l ayer. Ground: calcium sulfate an d gelatine on panel. Created: 1997. II: Paint composition: linseed oil and alizarin, possibly more than one layer of paint. Ground: calcium sulfate and gelatine on panel. Created: 1997. III: Silk fabric dyes with madder, which contains alizarin and purpurin. IV: Silk fabric probably dyed with munjeet, which contains alizarin and purpurin.
140 V: Paint composition: unknown paint and oil in multiple layers with wax. Ground: c alcium carbonate and collagen glue. Created: 15 th century with extensive restoration. VI: Paint composition: lead white in oil w ith a dammar varnish. Ground: b oard. VII: Paint composition: yellow ochre (FeO(OH)), linseed oil and elemi (resin). Ground: paper. Created: 20 06. All samples were untreated and affixed with double sided adhesive tape to either a glass microscope slide or directly on the steel sample plate, which was machined to a lower height to fit a microscope slide. Laser Desorption/Ionization Parameters All laser desorption/ionization experiments were performed on a Thermo Finnigan LTQ XL (San Jose, CA) equipped with an intermediate pressure source and a N 2 laser ( = 337 nm) For all samples and for both full scan and tandem mass spectrometry the las er was attenuated to the lowest power just above the ionization threshold, which varied by sample, but was typically set between 5 .0 and 35 J per pulse. Thus, t he desired ionization metric of laser fluence can only be estimated between 6.4 10 2 J/m 2 and 4.5 10 3 J/m 2 due to the difficulty in an accurate and precise measurement of the laser spot size, which has an approximate 100 m spot. Au tomatic gain control was set to limit the number of laser shots per analytical scan, which was generated from just one microscan. After each analytical scan, the samples were moved in a raster pattern with respect to the laser, within a user defined area manually
141 Instrumental Parameters All spectra were recorded in positive ion mode at the normal scan rate. For tandem mass spectra, the isolation width was set at 1.0 Da Collision induced dissociation (CID) pa rameters were manually set based on maximal daughter ion intensity, while maintaining the parent ion at 10% of the base peak. CID values ranged from 30 to 60. Results and Discussion As an extension to Chapters 2 and 3 on the tandem mass spectrometry of laser reduced anthraquinones, particularly the madder components alizarin and purpurin, and lead white, this Chapter demonstrates that the detection of those compounds is feasible in real samples without any prior pretreatment. Using both fragments of pai nted works of art and swatches of dyed silk, the former set including an historically authentic painting from the 15 th century and several samples from experimental paintings recently made by a conservator, the in situ LDI MS and LDI MS/MS detection of ali zarin, purpurin, and lead white was achieved for the first time. In Situ Detection of Alizarin from Painting Samples Two painting cross sections ( Figure 4 1, I and II) that were described to contain alizarin were tested by LDI MS and MS/MS. The full scan positive mode spectrum of sample I ( Figure 4 2A ) showed a few relatively abundant ions (e.g., m/z 154, 155, 550) along with low abundance ions (often considered to be chemical noise) at every mass to charge. As a result, the very low abundance peaks at m/z 241 and 242, which represent alizarin in either its unreduced (P) or reduced (P+1) form, respectively, cannot be distinguished even in the zoomed region of the spectrum ( Figure 4 2B ). Yet, the ion
142 at m/z 223, observed distinctly above the surrounding chemical noise, may very well be an in source loss of wate r from m/z 241, which was also observed in the LDI MS/MS spectrum of the standard ( Figure 2 5C). Similar re sults were obtained for sample II ( Figure 4 3 A and B ), although an ion at m/z 242, which might be P+1 of alizarin, was seen with higher abundance. To confirm the presence of alizarin in both samples, despite the very weak ion signals observed in the full scan spectra, tandem mass spectrometry was performed for both P and P+1 at m/z 241 and 2 42, respectively. The tandem mass spectra of both P and P+1 for sample I in Figure 4 4 show neutral losses of 28, 29, 56 that match those observed from LDI MS/MS of alizarin standard ( Figure 2 5C and D). T he spectra include daughter ions arising from oth er compounds with isobaric parent ions that fall in the 1.0 Da isolation window. This would be expected considering the complex nature of tandem mass spectrum of P ( Figure 4 5A) shows daughter ions sufficient to positively identify alizarin. However, the expected daughter ions of P +1 ( Figure 4 5B) are present, but do not dominate the spectrum; rather, daughter ions of isobaric interferences at m/z 242. Overall, des pite low abundance, with the aid of tandem mass spectrometry, alizarin was confirmed to be present in both samples I and II. In Situ Detection of Madder from Swatches of Dyed Silk Two swatches of silk ( Figure 4 1, III and IV) that were described to be dye d with madder, whose components include alizarin and purpurin, were tested by LDI MS and MS/MS. The full scan, positive mode spectrum of sample III ( Figure 4 6A) showed a few relatively abundant ions ( m/z 223, 501, 550) along with low abundance ions at ev ery mass to charge. The low abundance peaks at m/z 241 and 242, which represent
143 alizarin in either its unreduced (P) or reduced (P+1) form, respectively, are not above background even in the zoomed region of the spectrum ( Figure 4 6 B ). Similar results we re obtained for sample IV ( Figure 4 7), though with different ions dominating the spectrum (e.g., m/z 305, 331). To confirm the presence of alizarin in both samples despite very weak ion signals observed in the full scan spectra, tandem mass spectrometry was performed for both P and P+1 at m/z 241 and 242, respectively. The tandem mass spectra of both P and P+1 for sample III in Figure 4 8 show a few daughter ions, yet only one with a NL indicative of unreduced alizarin (NL of 28) and only one NL indic ati ve of reduced alizarin (NL of 29) matching what was observed in the alizarin standard ( Figure 2 5 C and D). The MS/MS spectra of both P and P+1 for sample IV in Figure 4 9 show a NL of 29 for P, which is not expected: this NL should appear only in the red uced form ( Figure 2 5D). The spectra are certainly not free from aberrant daughter ions whose parent ions mus t have been included in the 1.0 Da isolation window; these are expected considering the appear many similar NLs ac ross both samples such as the NLs of 43 and 71 that were not seen with the alizarin standard ( Figure 2 5 C and D), hence their assignments remain indeterminate. Overall, despite the low abundance in the full scan and the addit ional NLs in the daughter ion scan, with the aid of tandem mass spectrometry, alizarin was confirmed to be present in both samples III and IV. Purpurin is also a component of madder, so its detection in the swatches of dyed silk was also sought via LDI MS /MS. The full scan, positive mode spectrum of samples III and IV were mentioned above ( Figure s 4 6 and 4 7). Similarly, the very low
144 abundance peaks at m/z 257 and 258, which represent purpurin in either its unreduced (P) or reduced (P+1) form, respectiv ely, cannot be easily distinguished above background even in the zoomed region of the spectra ( Figure 4 6 B and 4 7B). Though, for sample IV, an ion for P+1 at m/z 258 does appear with relatively low abundance ( Figure 4 7B). To confirm the presence of pur purin in both samples, despite the very weak ion signals observed in the full scan spectra, tandem mass spectrometry was performed for both P and P+1 at m/z 257 and 258, respectively. The tandem mass spectra of both P and P+1 for both samples III and IV i n Figure s 4 10 and 4 11 show remarkable similarity, with many of the same NLs (i.e., 28, 43, 74). It is the NL of 28 that is most indicative of an anthraquinone, such as purpurin, as seen in the MS/MS spectra of the standard ( Figure 2 7). The NL of 70 th at is seen from the MS/MS of P in sample IV ( Figure 4 11A) indicates the cross ring cleavage seen in the standard ( Figure 2 7C) Unfortunately, none of the NLs identified with either an unreduced or reduced anthraquinone such as NL of 56 or 29, respective ly, were observed with the same relative abundance as in the standard. The spectra are certainly not free from aberrant daughter ions whose parent ions must have been included in the 1.0 Da isolation window, which are not altogether unexpected considering the complex milieu of this h samples such as the NLs of 43 and 74, which were not seen with the purpurin standard ( Figure 2 7 C and D), hence their assignments remain indeterminate. Overall, d espite both the low abundance in the full scan and the additional NLs in the daughter ion scan, tandem mass spectrometry allows purpurin to be confirmed as present in both samples III and
145 om the tandem mass spectra. In Situ Detection of Lead White from Painting Samples Three samples from separate paintings ( Figure 4 1 V, VI, and VII) were tested by LDI MS to detect lead white, only one of which was described to definitely contain the pigme nt. As seen in Figure s 4 12, 4 13, and 4 14, the full scan, positive mode spectrum of samples V, VI, and VII all showed spectra quite similar to that of the standard ( Figure 3 2), having the same superclusters and many of the same capital clusters. Howev er, the relative abundances of many miniscule clusters and of the standard. The spectrum obtained from sample VII ( Figure 4 14) is interesting because it was produced f rom only one analytical scan, despite the laser interrogating the sample VII and sample V are interesting because the description of either sample did not include lead wh ite as being a component of the paint layer. The spectra from all the samples were significant considering that the samples were all described as having a coating on the paint layer, such as wax or varnish, and that no sample pretreatment was used to eith er lessen or eliminate the coatings. Overall, the three full scan spectra alone that is, without the need for MS/MS are certain confirmation that lead white Also provided with the spectra are nine zoomed re gions that are intended to show in greater detail the spectral resolution obtained for three capital clusters at the monoisotopic m/z of (rounded) 208, 432, and 897. In Chapter 3 the LTQ generated spectral resolutions at the normal sc an rate for the capi tal clusters at m/z 208, 432, and
146 897 were 669, 1196, and 2270, respectively ( Table 3 1). The resolutions obtained for sample V ( Figure 4 12 B, C, and D) for clusters at m/z 208, 432, and 897 were 614, 1089, and 2366, respectively, which were quite good. The resolutions obtained for sample VI ( Figure 4 13 B, C, and D) for clusters at 208, 432, and 897 were 436, 829, and 2085, respectively. The resolutions obtained for sample VII ( Figure 4 14 B, C, and D) for clusters at m/z 208, 432, and 897 were 375, 80 3, 1553, respectively, which were quite poor, but understandable considering the spectrum was obtained from only one analytical scan and that space charge effects were apparent Conclusion The LDI tandem mass spectrometric analyses of the dyes alizarin and purpurin and the pigment lead white, which were reported in Chapters 2 and 3 have now been fragments of paintings and swatches of dyed silk. Two components of madd er, alizarin and purpurin, were shown to have been present in their unreduced forms with tandem mass spectrometric analyses, despite having very low relative abundances in their full scan spectrum, in both fragments of paintings and dyed silk. Also, aliza rin and purpurin were shown with less confidence to be present in those samples in their reduced forms. The pigment lead white was shown to be present in three samples of paintings, two of which were not described to contain the pigment, without the use o f both the LDI MS/MS detection of madder was achieved in both paintings and textiles and the LDI MS detection of lead white was achieved in paintings without any sampl e pretreatment whatsoever.
147 Figure 4 1. Painting samples containing alizarin (I, II). Silk swatches dyed with alizarin (III, IV). Painting samples containing lead white (V, VI, VII). III m II m I IV m V m VII VI
148 Figure 4 2. Painting sample I: LDI full scan positive mode mass s pectrum (A); and zoomed region for alizarin (B ). A B
149 F igure 4 3. Painting sample II: LDI full scan, positive mode mass spectrum (A); and zoomed region for alizarin (B) A B
150 Figure 4 4. Tandem mass spectra from sample I of supposed alizarin at the locati on for the [M+H] + (A) and [M+2H ] + (B) parent ions. [M H 2 O] + [M 18] + [M CO] + [M 28] + A B MS 2 241 MS 2 242 [M CO] + [M 28] + [M 2CO] + [M 56] + [M H 2 O] + [M 18] + [M 2CO] + [M 56] + [M COH] + [M 29] +
151 Figure 4 5. Tandem mass spectra from sample II of supposed alizarin at the location for the [M+H] + (A) and [M+2H ] + (B) parent ions. A B [M 2CO] + [M 56] + [M H 2 O] + [M 18] + [M CO] + [M 28] + MS 2 241 MS 2 242 [M CO] + [M 28] + 223.83 [M H 2 O] + [M 18] +
152 Figure 4 6. Silk sample III: LDI full scan positive mode mas s spectrum (A); and zoomed region for alizarin (B). A B
153 Figure 4 7. Silk sample IV: LDI full scan positive mode mass spectrum (A); and zoomed region for alizarin (B). A B
154 Figure 4 8. Tandem mass spectra from sample III (silk) of supposed alizarin at the location for the [M+H] + (A) and [M+2H ] + (B) parent ions. A B MS 2 241 MS 2 242 [M 18] + [M 28] + [M 43] + [M 18] + [M 43] + [M 29] + [M 71] + [M 72] + [M 44] + [M 28] + 214.08
155 Figure 4 9. Tandem mass spectra from sample IV (silk) of supposed alizarin at the location for the [M+H] + (A) and [M+2H ] + (B) parent ions. A B MS 2 241 MS 2 242 [M 18] + [M 18] + [M 29] + [M 71] + [M 43] + [M 44] + [M 71] + [M 43] +
156 Figure 4 10. Tandem mass spectra from sam ple III of supposed purpurin at the location for the [M+H] + (A) and [M+2H ] + (B) parent ions. A B MS 2 257 MS 2 258 [M 28] + [M 43] + [M 18] + [M 74] + [M 74] + [M 18] + [M 43] + [M 28] + [M 56] +
157 Figure 4 11. Tandem mass spectra from sample IV of supposed purpurin at the location for the [M+H] + (A) and [M+2H ] + (B) parent ions. A B MS 2 257 MS 2 258 [M 43] + [M 28] + [M 70] + [M 74] + [M 18] + [M 28] + [M 43] + [M 18] + [M 56] + [M 56] + 202.00
158 Figure 4 12. Pa inting sample V: LDI full scan, positive mode mass spectrum (A), which shows lead white in the sample. Zoomed regions to show spectral resolution for clusters at monoisotopic m/z of 208 (A), 432 (B), and 897 (C). A B C D
159 Figure 4 13. Painting sample VI: LD I full scan, positive mode mass spectrum (A), which shows lead white in the sample. Zoomed regions to show spectral resolution for clusters at monoisotopic m/z of 208 (A), 432 (B), and 897 (C). A B C D
160 Figure 4 14. Painting sample VII: LDI full scan, posi tive mode mass spectrum (A), which shows lead white in the sample. Zoomed regions to show spectral resolution for clusters at monoisotopic m/z of 208 (A), 432 (B), and 897 (C). A B C D
161 C HAPTER 5 CONCLUSION AND FUTUR E DIRECTIONS Conclusions The preservation of cultural heritage is of critical importance to the continuance of a people, their connection to t heir past, their outlook toward their future, and their pride, celebration, and enjoyment thereof. The spectacularly rich and voluminous amount of art cr eated over millennia is resoundingly worthy of the efforts engaged toward its preservation. Over the past few decades, scientists particularly analytical chemists and conservation scientist s have applied and, to a lesser extent, developed state of the art analytical technologies for the analyses of artistic and archaeological objects for such purposes as component detection, dating, provenance, and taphonomy with the ultimate goals of restoration and preservation (viz., conservation). Thus, the underlying philosophy of this dissertation was to utilize scientific positivistic, empirical, and experimental approaches to further strengthen the union of analytical chemistry particularly mass spectrometry to the field of conservation science The practical go al of the research presented in this dissertation was to develop and employ laser desorption/ionization tandem mass spectrometric techniques (LDI MS/MS) in the analysis of the dye madder particularly the anthraquinones alizarin and purpurin and the pigment lead white in fragments of paintings and swatches of dyed silk. To achieve that goal, neat standards of selected anthraquinones and lead white were initially tested with LDI MS and LDI MS/MS to observe their ionization and fragmentation behavior and to e lucidate the structure of the ions formed, and then those data were applied to the detection of those colorants in real samples.
162 In the analyses of the anthraquinones, the peculiar ionization that formed anomalous [M+2H ] + ions was fully explored with LDI MS/MS for the first time. The results show that anthraquinones exhibit laser induced reduction about the carbonyl central to the anthracene skeleton, observed as the neutral loss of 29 (COH) with the concomitant decrease in neutral losses of 28 (CO) and 56 (2 CO). Furthermore, differences in fragmentation were exhibited for anthraquinones with functional groups (e.g., hydroxyl or amine) proximal to the reduced carbonyl versus anthraquinones with those groups distal to the reduced carbonyl. These results new to LDI mass spectrometry, stand to benefit the conservation science community by alerting them to the presence and correct interpretation of otherwise unexpected ionization of this important class of compounds in artistic dyes. The ubiquitous pigmen t lead white was analyzed by LDI MS/MS and LDI Orbitrap MS/MS for the first time. The distinct pattern of superclusters, capital clusters, and miniscule clusters makes the observation of the pigment relatively simple and quick. Through deductive and indu ctive reasoning, the formulae of the majority of clusters was determined via the superior spectral resolving power and mass accuracy of the Orbitrap for ions observed in both the full and daughter ion scans. Exploiting the distinctive isotopic pattern of lead to match isotopic patterns between experimental and theoretical spectra, the full scan spectrum was first divided into superclusters based upon the number of lead atoms. The isotopic pattern of lead was subsequently used to identify the monoisotopic mass of each cluster, from which a lead limited, accurate formula could be derived. Formulae that remained indeterminate were inductively determined using tandem mass spectra and pattern matching to the already determined
163 formulae. These methods, complet ely new to mass spectrometry, can benefit the communities of mass spectrometrists and conservation scientists by offering a new method of interpretation of isotopically complex cluster ions of a pigment ubiquitous since antiquity. Lastly, the preceding me thodologies were applied to the in situ detection of of dyed silk. Two components of the red dye madder, alizarin and purpurin, were detected in their reduced and unred uced forms using tandem mass spectrometry, despite their not being present at detectable levels in the full scan spectra. The pigment lead white was easily detected in painting fraqments using only full scan spectra despite the presence of wax or varnish atop the paint layer. The successful in situ LDI MS/MS analysis of alizarin and purpurin and LDI MS analysis of lead white were the first instances ever documented for fragments of paintings or swatches of dyed silk, and both were accomplished without any sample pretreatment whatsoever. Overall, the results obtained and conclusions drawn are not a dramatic leap forward for science as a whole, but they do represent a significant step forward for the relatively young and burgeoning field of conservation sci ence. Therefore, it is recommended that the material presented in this dissertation be disseminated throughout the community. Future Directions Though the work presented in this dissertation includes well researched literature reviews, complete experimen ts, and definitive conclusions, there exist several aspects open to further exploration and experimentation. These future directions can neither negate nor supplant the findings presented here, they may only bolster the findings.
164 Although the laser induc ed reduction of anthraquinones was definitively proven by the appearance of the [M+2H ] + ion, neutral loss of a reduced carbonyl, and decrease in the relative abundance of the neutral loss of one and two carbonyls, there are still three sets of experiments that should be conducted. First, anthraquinones that are deliberately exposed to UV light in a controlled environment should be analyzed with electrospray ionization tandem mass spectrometry. The anthraquinone that was analyzed was only unintentionally exposed to UV after not having been stored in a darkened container over several days. A controlled time course study of UV exposure may reveal a dependence on the abundance of P+1 relative to P, which may provide further insight to the underlying physioch emical process of photon induced reduction. Second, many more experiments could be devised and conducted to control the many variables influencing LDI in the source chamber. These variables include, but are not limited to, source pressure, partial pressu re of water, temperature, surface substrate, and the use of electron capturing additives. Third, to test the influence of UV photons on the laser induced reduction, a laser that delivers IR photons, such as a CO 2 laser ( 2 laser, which delivers UV photons ( = 337 nm). The analysis of lead white is novel not only because of the elucidation of the clusters themselves, but because of the limited precedent in the literature of decip hering isotopically complex middle molecules. To improve the elucidation of all isotopically complex middle molecules not just lead white experiments exercising thermodynamic control at the ionization source and kinetic control at the ion trap are propose d, though lead white will be discussed as the exemplar case.
165 Thermodynamically controlled experiments at the ionization source are intended to observe changes, if any, in the presence and relative intensity of clusters and should include variables such as source pressure and laser fluence. Indeed, the quality of the lead white spectra was negatively affected by high laser energies, since the pigment has a very low ionization threshold and produces significant signal at a very low laser energy. The analys is presented was focused on elucidating the clusters in clean, low noise spectra a challenge in itself and not did not venture in to creating even more complex spectra at higher laser power. Kinetic control at the ion trap should include varying the activ ation time with and without a change in the value of collision activated dissociation (CID). Since cluster formation may occur in the gas phase after ionization, varying activation time particularly with zero CID may produce different clusters in full sca n or daughter ion spectra. This de pendence on residence time with in the mass spectrometer may have already been observed in Figure 3 6, which shows different daughter ion clusters from the same precursor cluster using the ion trap and Orbitrap, the latter of which necessitates a much longer ion transport as an ion travels from the LIT to the Orbitrap Lastly, analysis of real art samples might be improved and expanded upon by ent other than the use of double sided tape to adhere them to the sample plate. However, the surface of the fragments of paintings could have been cleaned with solvent to remove the wax or varnish or other surface chemicals and contaminants to promote int
166 cleaning process. O ne objective of a conservation scientist is to have a totally non invasive analytical technique. Considering that LDI is a non destructive process, it will be useful to gather some information as to the extent of its destruction of the surface. This info rmation can be obtained via visible light microscopy or scanning electron microscopy, depending on the level of resolution desired, before and after LDI analysis. Other future directions for real samples might be dictated as needed simply by the types of sample and analysis at hand.
167 REFERENCES (1) UN UNESCO World Heritage Centre Convention Concerning the Protection of the World Cultural and Natural Heritage. http://whc.unesco.org/?cid=17 5 (accessed April 2013) (2) Ciliberto, E.; Spoto, G. Modern analytical methods in art and archaeology ; Wiley: New York, 2000. (3) Artioli, G. Scientific Methods and Cultural Heritage: An introduction to the application of materials science to archaeometry and conservation science ; OUP Oxford: Oxford, 2010. (4) Pollard, A. M.; Heron, C. Archaeological Chemistry ; Royal Society of Chemistry: Cambridge, 2008. (5) Caley, E. R. J. Chem. Educ. 1951 28 64 66. (6) Fremout, W.; Saverwyns, S.; Peters, F.; Deneffe, D. J. Raman Spectrosc. 2006 37 1035 1045. (7) Andrikopoulos, K. S.; Daniilia, S.; Roussel, B.; Janssens, K. J. Raman Spectrosc. 2006 37 1026 1034. (8) Dussubieux, L.; Robertshaw, P.; Glascock, M. D. Int. J. Mass Spectrom. 2009 284 152 161. (9) Tognaz zi, A.; Dattilo, A. M.; Bracchini, L.; Aggravi, M.; Benetti, F.; Mugnaini, E.; Magnani, A.; Rossi, C. Surf. Interface Anal. 2011 43 1108 1119. (10) Niimura, N. Int. J. Mass Spectrom. 2009 284 93 97. (11) Oudemans, T. F. M.; Eijkel, G. B.; Boon, J. J. J Archaeol. Sci. 2007 34 173 193. (12) Honig, R. E.; Woolston, J. R. Appl. Phys. Lett. 1963 2 138 139. (13) Fenner, N. C.; Daly, N. R. Rev. Sci. Instrum. 1966 37 1068 1070. (14) Denoyer, E.; van Grieken, R.; Adams, F.; Natusch, D. F. S. Anal. Chem. 1 982 54 26A 41A. (15) Vastola, F. J.; Mumma, R. O.; Pirone, A. J. Org. Mass Spectrom. 1970 3 101 104. (16) Hillenkamp, F.; UnsoLd, E.; Kaufmann, R.; Nitsche, R. Nature 1975 256 119 120. (17) Hillenkamp, F.; Peter Katalinic, J. MALDI MS ; Wiley VCH: Wei nheim, 2007.
168 (18) Aberth, W.; Straub, K. M.; Burlingame, A. L. Anal. Chem. 1982 54 2029 2034. (19) Beckey, H. D. Int. J. Mass Spectrom. Ion Phys. 1969 2 500 502. (20) Torgerson, D. F.; Skowronski, R. P.; Macfarlane, R. D. Biochem. Biophys. Res. Commun. 1974 60 616 621. (21) Benninghoven, A.; Sichtermann, W. K. Anal. Chem. 1978 50 1180 1184. (22) Barber, M.; Bordoli, R. S.; Sedgwick, R. D.; Tyler, A. N. J. Chem. Soc., Chem. Commun. 1981 0 325 327. (23) Liao, P. C.; Allison, J. J. Mass Spectrom. 199 5 30 408 423. (24) Bourcier, S.; Bouchonnet, S.; Hoppilliard, Y. Int. J. Mass Spectrom. 2001 210/211 59 69. (25) Karbach, V.; Knochenmuss, R.; Zenobi, R. Rapid Commun. Mass Spectrom. 1998 12 (26) Karas, M.; Bachmann, D.; Hillenkamp, F. Anal. Chem. 19 85 57 (27) Fenn, J. B.; Mann, M.; Meng, C. K.; Wong, S. F.; Whitehouse, C. M. Science 1989 246 64 71. (28) Watson, J. T. Introduction to Mass Spectrometry Third ed.; Lippincott Raven: New York, NY, 1997. (29) Watson, J. T.; Sparkman, O. D. Introduc tion to Mass Spectrometry: Instrumentation, Applications, and Strategies for Data Interpretation ; Wiley: West Sussex, 2008. (30) Cech, N. B.; Enke, C. G. Mass Spectrom. Rev. 2001 20 362 387. (31) Gomez, A.; Tang, K. Phys. Fluids 1994 6 404 414. (32) Co le, R. B. J. Mass Spectrom. 2000 35 763 772. (33) Dole, M.; Mack, L. L.; Hines, R. L.; Mobley, R. C.; Ferguson, L. D.; Alice, M. B. J. Chem. Phys. 1968 49 2240 2249. (34) Iribarne, J. V.; Thomson, B. A. J. Chem. Phys. 1976 64 2287 2294. (35) Nobel Th e Nobel Prize in Physics 1989. http://nobelprize.org/nobel_prizes/physics/laureates/1989/index.html (accessed May 2013) (36) Schwartz, J. C.; Senko, M. W.; Syka, J. E. P. J. Am. Soc. Mass Spectrom. 2002 13 659 669.
169 (37) Blackler, A. R.; Klammer, A. A.; MacCoss, M. J.; Wu, C. C. Anal. Chem. 2006 78 1337 1344. (38) March, R. E. J. Mass Spectrom. 1997 32 351 369. (39) Julian, R. K.; Cooks, R. G. Anal. Chem. 1993 65 18 27 1833. (40) Guan, S.; Marshall, A. G. Anal. Chem. 1993 65 1288 1294. (41) Louris, J. N.; Taylor, D. M. Method and apparatus for ejecting unwanted ions in an ion trap mass spectrometer. U.S. Patent 5,324,939, 1994. (42) Hu, Q.; Noll, R. J.; Li, H.; Maka rov, A.; Hardman, M.; Graham Cooks, R. J. Mass Spectrom. 2005 40 430 443. (43) Hardman, M.; Makarov, A. A. Anal. Chem. 2003 75 1699 1705. (44) Makarov, A.; Denisov, E.; Kholomeev, A.; Balschun, W.; Lange, O.; Strupat, K.; Horning, S. Anal. Chem. 2006 78 2113 2120. (45) Strupat, K.; Kovtoun, V.; Bui, H.; Viner, R.; Stafford, G.; Horning, S. J. Am. Soc. Mass Spectrom. 2009 20 1451 1463. (46) Scigelova, M.; Makarov, A. Proteomics 2006 6 16 21. (47) Perry, R. H.; Cooks, R. G.; Noll, R. J. Mass Spectro m. Rev. 2008 27 661 699. (48) Olsen, J. V.; de Godoy, L. M. F.; Li, G.; Macek, B.; Mortensen, P.; Pesch, R.; Makarov, A.; Lange, O.; Horning, S.; Mann, M. Mol. Cell. Proteomics 2005 4 2010 2021. (49) Xian, F.; Hendrickson, C. L.; Marshall, A. G. Anal. Chem. 2012 84 708 719. (50) Roy, A. Artists' Pigments: A Handbook of Their History and Characteristics ; Cambridge University Press: Cambridge 1993. (51) Zollinger, H. Color Chemistry: Syntheses, Properties, and Applications of Organic Dyes and Pigments ; Wiley VCH: Weinheim, 2004. (52) de Graaff, J. H. H.; Roelofs, W. G. T.; van Bommel, M. R. The colourful past: origins, chemistry and identification of natural dyestuffs ; Archetype: London, 2004. (53) Mills, J. S.; White, R. The Organic Chemistry of Museum Objects 2 ed.; Butterworth Heinemann, 1994. (54) Mouri, C.; Laursen, R. Microchim. Acta 2012 179 105 113.
170 (55) Derksen, G. C. H.; Van Beek, T. A.; Atta ur, R. In Studies in Natural Products Chemistry ; Elsevier, 2002; Vol. 26, Part G, pp 629 684. (56) Sa unders, D.; Kirby, J. Nat. Gall. Tech. Bull. 1994 15 79 97. (57) Thomson, R. H. Naturally Occurring Quinones 2nd ed.; Academic Press: New York, 1971. (58) Graebe, C.; Liebermann, C. Chem. Ber. 1869 2 332. (59) Gordon, P. F.; Gregory, P. Organic Chemis try in Colour ; Springer: New York, 1983. (60) Puchalska, M.; Orlinska, M.; Ackacha, M. A.; Potec Pawlak, K.; Jarosz, M. J. Mass Spectrom. 2003 38 1252 1258. (61) Wouters, J.; Rosario Chirinos, N. J. Am. Instit. Conserv. 1992 31 237 255. (62) Boldizsar, I.; Szucs, Z.; Fuzfai, Z.; Molnar Perl, I. J. Chromatogr. A 2006 1133 259 274. (63) Ackacha, M. A.; Potec Pawlak, K.; Jarosz, M. J. Sep. Sci. 2003 26 1028 1034. (64) Degano, I.; Ribechini, E.; Modugno, F.; Colombini, M. P. Appl. Spectrosc. Rev. 2009 44 363 410. (65) Derksen, G. C. H.; Neiderlander, H. A. G.; van Beek, T. A. J. Chromatogr. A 2002 978 119 127. (66) Ahn, C.; Obendorf, S. K. Text. Res. J. 2004 74 949 954. (67) Shibayama, N.; Yamaoka, R.; Sato, M. Analysis of Colored Compounds found i n Natural Dyes by Thermospray Liquid Chromatography Mass Spectrometry (TSP LC MS) and a Preliminary Study of the Application to Analysis of Natural Dyes used on Historic Textiles. Instituut Collectie Nederland Dyes in History and Archaeology 20 Amsterdam The Netherlands 2001; Archetype Publications; 51 69. (68) Surowiec, I.; Szostek, B.; Trojanowicz, M. J. Sep. Sci. 2007 30 2070 2079. (69) Szostek, B.; Orska Gawrys, J.; Surowiec, I.; Trojanowicz, M. J. Chromatogr. A 2003 1012 179 192. (70) Epolito, W J.; Lee, Y. H.; Bottomley, L. A.; Pavlostathis, S. G. Dyes Pigm. 2005 67 35 46. (71) DeRoo, C. S.; Armitage, R. A. Anal. Chem. 2011 83 6924 6928.
171 (72) Bennett, J. A.; Schweikert, E. A.; van Vaeck, L.; Adams, F. C. J. Trace Microprobe Tech. 1990 7 2 79 292. (73) Dale, M. J.; Jones, A. C.; Langridge Smith, P. R. R.; Costello, K. F.; Cummins, P. G. Anal. Chem. 1993 65 793 801. (74) Grim, D. M. Dissertation, Michigan State University, 2003. (75) Grim, D. M.; Allison, J. Int. J. Mass Spectrom. 2003 222 85 99. (76) Grim, D. M.; Allison, J. Archaeometry 2004 46 283 299. (77) Wyplosz, N. Dissertation, FOM Institute for Atomic and Molecular Physics, Amsterdam, 2003. (78) Soltzberg, L. J.; Hagar, A.; Kridaratikorn, S.; Mattson, A.; Newman, R. J. Am. Soc. Mass Spectrom. 2007 18 2001 2006. (79) Van Elslande, E.; Gurineau, V.; Thirioux, V.; Richard, G.; Richardin, P.; Laprvote, O.; Hussler, G.; Walter, P. Anal. Bioanal. Chem. 2008 390 1873 1879. (80) Kirby, D. P.; Khandekar, N.; Sutherland, K.; Price, B A. Int. J. Mass Spectrom. 2009 284 115 122. (81) Lou, X.; Sinkeldam, R. W.; van Houts, W.; Nicolas, Y.; Janssen, P. G. A.; van Dongen, J. L. J.; Vekemans, J. A. J. M.; Meijer, E. W. J. Mass Spectrom. 2007 42 293 303. (82) Meijer, W. W. prof.dr. E.W. (Bert) Meijer Expertises. http://www.tue.nl/en/employee/ep/e/d/ep uid/19911632 (accessed November 1, 2012) (83) Karas, M.; Glckmann, M.; Schfer, J. J. Mass Spectrom. 2000 35 1 12. (84) Frankevich, V.; Knochenmuss, R.; Zenobi, R. Int. J. Mass Spectrom. 2002 220 11 19. (85) Frankevich, V. E.; Zhang, J.; Friess, S. D.; Dashtiev, M.; Zenobi, R. Anal. Chem. 2003 75 6063 6067. (86) Zhang, J.; Frankevich, V.; Knochenmuss, R.; Friess, S D.; Zenobi, R. J. Am. Soc. Mass Spectrom. 2003 14 42 50. (87) Okuno, S.; Nakano, M.; Matsubayashi, G. e.; Arakawa, R.; Wada, Y. Rapid Commun. Mass Spectrom. 2004 18 2811 2817.
172 (88) Kosevich, M. V.; Chagovets, V. V.; Shmigol, I. V.; Snegir, S. V.; Bor yak, O. A.; Orlov, V. V.; Shelkovsky, V. S.; Pokrovskiy, V. A.; Gomory, A. J. Mass Spectrom. 2008 43 1402 1412. (89) McAvey, K. M.; Guan, B.; Fortier, C. A.; Tarr, M. A.; Cole, R. B. J. Am. Soc. Mass Spectrom. 2011 22 659 669. (90) Aplin, R. T.; Pike, W. T. Chem. Ind. (London) 1966 2009. (91) Ukai, S.; Hirose, K.; Tatematsu, A.; Goto, T. Tetrahedron Lett. 1967 8 4999 5002. (92) Cornforth, J. W.; Ryback, G.; Robinson, P. M.; Park, D. J. Chem. Soc. C: Org. 1971 2786 2788. (93) Oliver, R. W. A.; Rashma n, R. M. J. Chem. Soc. B: Phys. Org. 1968 1141 1143. (94) Oliver, R. W. A.; Rashman, R. M. J. Chem. Soc. B: Phys. Org. 1971 341 344. (95) Grigsby, R. D.; Jamieson, W. D.; McInnes, A. G.; Maass, W. S. G.; Taylor, A. Can. J. Chem. 1974 52 4117 4122. (96) Beaumont, P. C.; Edwards, R. L. J. Chem. Soc. C: Org. 1969 2398 2403. (97) Gaylord, M. C.; Brady, L. R. J. Pharm. Sci. 1971 60 1503 1508. (98) Das, B. C.; Lounasmaa, M.; Tendille, C.; Lederer, E. Biochem. Biophys. Res. Commun. 1965 21 318 322. (99) P atai, S.; Rappoport, Z. The Chemistry of the Quinonoid Compounds ; John Wiley & Sons: New York, 1974. (100) Fessenden, R. J.; Fessenden, J. S.; Feist, P. Organic Laboratory Techniques ; Brooks/Cole: Pacific Grove, 2001. (101) Karas, M.; Hillenkamp, F. Anal. Chem. 1988 60 2299 2301. (102) de la Mora, J. F.; Van Berkel, G. J.; Enke, C. G.; Cole, R. B.; Martinez Sanchez, M.; Fenn, J. B. J. Mass Spectrom. 2000 35 939 952. (103) Beynon, J. H.; Williams, A. E. Appl. Spectrosc. 1960 14 156 160. (104) Bowie, J. H.; Cameron, D. W.; Giles, R. G. F.; Williams, D. H. J. Chem. Soc. B: Phys. Org. 1966 335 339. (105) Filyugina, A. D.; Kolokolov, B. N.; Bragina, S. I.; Rotermel, I. A.; Dubrovia, A. A. Russ. J. Org. Chem. 1972 8 2151 2155.
173 (106) Vasileva, A. D.; Roter mel, I. A.; Artamonova, N. N.; Gaeva, L. A.; Dubrovia, A. A.; Kolokolov, B. N. Russ. J. Org. Chem. 1972 8 2156 2159. (107) Proctor, C. J.; Kralj, B.; Larka, E. A.; Porter, C. J.; Maquestiau, A.; Beynon, J. H. Org. Mass Spectrom. 1981 16 312 322. (108) Pan, Y.; Zhang, L.; Zhang, T.; Guo, H.; Hong, X.; Qi, F. J. Mass Spectrom. 2008 43 1701 1710. (109) Patai, S.; Rappoport, Z. The Chemistry of the Quinonoid Compounds ; John Wiley & Sons: New York, 1988. (110) Carlson, S. A.; Hercules, D. M. Photochem. Pho tobiol. 1973 17 123 131. (111) Carlson, S. A.; Hercules, D. M. Anal. Chem. 1973 45 1794 1799. (112) Gorner, H. Photochem. Photobiol. 2003 77 171 179. (113) Tickle, K.; Wilkinson, F. Trans. Faraday Soc. 1965 61 1981 1990. (114) Wakisaka, A.; Ebbesen T. W.; Sakuragi, H.; Tokumaru, K. J. Phys. Chem. 1987 91 6547 6551. (115) Flom, S. R.; Barbara, P. F. J. Phys. Chem. 1985 89 4489 4494. (116) Gollnick, K.; Held, S.; Mrtire, D. O.; Braslavsky, S. E. J. Photochem. Photobiol., A 1992 69 155 165. (11 7) Young, W. J. Application of Science in Examination of Works of Art: Proceedings of the Seminar: June 15 19, 1970 ; Museum of Fine Arts: Boston, 1973. (118) De Wannemacker, G.; Vanhaecke, F.; Moens, L.; Van Mele, A.; Thoen, H. J. Anal. At. Spectrom. 2000 15 323 327. (119) Schultheis, G.; Prohaska, T.; Stingeder, G.; Dietrich, K.; Jembrih Simburger, D.; Schreiner, M. J. Anal. At. Spectrom. 2004 19 838 843. (120) Fortunato, G.; Ritter, A.; Fabian, D. Analyst 2005 130 898 906. (121) De Laeter, J.; Kurz, M. D. J. Mass Spectrom. 2006 41 847 854. (122) Keisch, B.; Callahan, R. C. Archaeometry 1976 18 181 193. (123) Van Elslande, E.; Guerineau, V.; Thirioux, V.; Richard, G.; Richardin, P.; Laprevote, O.; Hussler, G.; Walter, P. Anal. Bioanal. Chem. 2008 390 1873 1879. (124) de Heer, W. A. Rev. Mod. Phys. 1993 65 611 676.
174 (125) Mhlbach, J.; Sattler, K.; Pfau, P.; Recknagel, E. Phys. Lett. A 1982 87 415 417. (126) Hoareau, A.; Mlinon, P.; Cabaud, B.; Rayane, D.; Tribollet, B.; Broyer, M. Chem. Phys. Lett. 1988 143 602 608. (127) Dean, L. K. L.; Didonato, G. C.; Busch, K. L. Inorg. Chim. Acta 1989 157 175 185. (128) Katakuse, I.; Ichihara, T.; Ito, H.; Matsuo, T.; Sakurai, T.; Matsuda, H. Int. J. Mass Spectrom. Ion Processes 1989 91 93 97. (129) Xing, X.; Tian, Z.; Liu, H.; Tang, Z. Rapid Commun. Mass Spectrom. 2003 17 1411 1415. (130) Waldschmidt, B.; Barman, S.; Rajesh, C.; Majumder, C.; Das, G. P.; Schfer, R. Phys. Rev. B 2009 79 045422. (131) Prussakowska, M.; Sokalska, M.; Gierczyk, B.; Rapid Commun. Mass Spectrom. 2010 24 1925 1929. (132) Kelting, R.; Otterstatter, R.; Weis, P.; Drebov, N.; Ahlrichs, R.; Kappes, M. M. J. Chem. Phys. 2011 134 024311. (133) Echt, O.; Sattler, K.; Recknagel, E. Phys. Rev. Lett. 1 981 47 1121 1124. (134) Dyson, P.; McGrady, J.; Reinhold, M.; Johnson, B. G.; McIndoe, J. S.; Langridge Smith, P. R. J. Cluster Sci. 2000 11 391 401. (135) Fenselau, C. Anal. Chem. 1982 54 105A 116A. (136) Brack, M. Rev. Mod. Phys. 1993 65 677 732. (137) Scheidegger, C.; Suter, M. J. F.; Behra, R.; Sigg, L. Front. Microbiol. 2012 3 (138) Fenselau, C.; Cotter, R.; Hansen, G.; Chen, T.; Heller, D. J. Chrom. A 1981 218 21 30. (139) Yergey, J.; Heller, D.; Hansen, G.; Cotter, R. J.; Fenselau, C. Ana l. Chem. 1983 55 353 356. (140) Senko, M. W.; Beu, S. C.; McLafferty, F. W. J. Am. Soc. Mass Spectrom. 1995 6 229 233. (141) O'Connor, P. B.; Little, D. P.; McLafferty, F. W. Anal. Chem. 1996 68 542 545. (142) Valkenborg, D.; Mertens, I.; Lemire, F. ; Witters, E.; Burzykowski, T. Mass Spectrom. Rev. 2012 31 96 109.
175 (143) Yergey, J. A. Int. J. Mass Spectrom. Ion Phys. 1983 52 337 349. (144) Interpretation of Mass Spectra 4th ed.; Univ. Science Books: Mill Valley, 1993. (145) Yamamoto, H.; McCloskey, J. A. Anal. Chem. 1977 49 281 283. (146) Hibbert, D. B. Chemom. Intell. Lab. Syst. 1989 6 203 212. (147 ) Rosman, K. J. R.; Taylor, P. D. P. 4th International Symposium on Bioorganic Chemistry 1997 70 217 235. (148) Rockwood, A. L. Rapid Commun. Mass Spectrom. 1997 11 241 248. (149) Rockwood, A.; Kushnir, M.; Nelson, G. J. Am. Soc. Mass Spectrom. 2003 1 4 311 322. (150) Bozorgzadeh, M. H.; Morgan, R. P.; Beynon, J. H. Analyst 1978 103 613 622. (151) Todd, P. J.; Barbalas, M. P.; McLafferty, F. W. Org. Mass Spectrom. 1982 17 79 80. (152) Singleton, K. E.; Cooks, R. G.; Wood, K. V. Anal. Chem. 1983 55 762 764. (153) Bozorgzadeh, M. H. Rapid Commun. Mass Spectrom. 1988 2 61 64. (154) de Hoffman, E.; Auriel, M. Org. Mass Spectrom. 1989 24 748 758. (155) Glish, G.; McLuckey, S.; Asano, K. J. Am. Soc. Mass Spectrom. 1990 1 166 173. (156) Cataldi, T. R. I.; Lelario, F.; Orlando, D.; Bufo, S. A. Anal. Chem. 2010 82 5686 5696. (157) Tou, J. C. Anal. Chem. 1983 55 367 372. (158) Bozorgzadeh, M. H.; Lapp, R. L.; Gross, M. L. Org. Mass Spectrom. 1988 23 712 718. (159) Elslande, E.; Gurineau, V.; Thi rioux, V.; Richard, G.; Richardin, P.; Laprvote, O.; Hussler, G.; Walter, P. Anal. Bioanal. Chem. 2008 390 1873 1879. (160) Reich, R. F.; Cudzilo, K.; Levisky, J. A.; Yost, R. A. J. Am. Soc. Mass Spectrom. 2010 21 564 571. (161) Critchley, G.; Dyson, P. J.; Johnson, B. F. G.; McIndoe, J. S.; O'Reilly, R. K.; Langridge Smith, P. R. R. Organometallics 1999 18 4090 4097.
176 (162) Colombini, P. M. P.; Modugno, F. Organic Mass Spectrometry in Art and Archaeology ; John Wiley & Sons: West Sussex, 2009. (163) G lascock, M. D.; Speakman, R. J.; Popelka Filcoff, R. S. Archaelogical Chemistry #968: Analytical Techniques and Archaeological Interpretation ; OUP USA: Washington, DC, 2007. (164) Mazel, V.; Richardin, P.; Touboul, D.; Brunelle, A.; Walter, P.; Laprvote, O. Anal. Chim. Acta 2006 570 34 40. (165) Keune, K. Dissertation, FOM Institute for Atomic and Molecular Physics, Amsterdam, 2005. (166) Keune, K.; Boon, J. J. Anal. Chem. 2004 76 1374 1385. (167) Keune, K.; Boon, J. J. Anal. Chem. 2005 77 4742 4750. (168) Keune, K.; Hoogland, F.; Boon, J. J.; Peggie, D.; Higgitt, C. Int. J. Mass Spectrom. 2009 284 22 34. (169) Sanyova, J.; Cersoy, S.; Richardin, P.; Laprvote, O.; Walter, P.; Brunelle, A. Anal. Chem. 2011 83 753 760. (170) Richardin, P.; Mazel, V .; Walter, P.; Laprvote, O.; Brunelle, A. J. Am. Soc. Mass Spectrom. 2011 22 1729 1736. (171) Kuckova, S.; Nemec, I.; Hynek, R.; Hradilova, J.; Grygar, T. Anal. Bioanal. Chem. 2005 382 275 282.
177 BIOGRAPHICAL SKETCH Michael (Mike) Patrick Napolitan o was born and raised in Rahway, New Jersey. Both his father, whom emigrated alone from Italy and without a command of English, and mother, she a child of immigrants, instilled in him the value of higher education and the idea of the American Dream while holding true the virtues of both the Catholic Church and the Occident. To that effect, Michael attended twelve years of Catholic grammar and high school, and then earned his ACS certified Bachelor of Science in chemistry from the Pennsylvania State Univer sity in 2003. During his time at Penn State, he was a research assistant in the mass spectrometry research laboratory of Dr. A. Daniel Jones. Following graduation, Michael first traveled to Europe where, for the first time, he met many of his relatives i n Italy, and then interned at Honeywell in Chester, Virginia. In the fall of 2004, Michael started graduate school at the University of Florida and which was a collabora tion with Dr. Leonid Moroz from the Whitney Laboratory for Marine Bioscience at UF, explored methods for the MALDI MS analysis of both singly isolated and cultured neurons of the sea slug Aplysia californica To prepare for his literature seminar, Michael traveled in the fall of 2007 to New York City to take a tour of the conservation laboratories at the Metropolitan Museum of Art and interview Julie Arslanoglu, who is a conservation scientist there, which began a collaboration that led to this dissertatio n. With funding from the MSPIRE grant, Michael returned to Europe in the spring of 2010 to conduct experiments in the laboratory of Dr. Ron Heeren at FOM AMOLF, Amsterdam.
178 During most of his time at UF, Michael was the head teaching assistant for the gene ral chemistry laborator y course CHM 2045L, which garnered him departmental and university wide TA awards. He is also very proud of and thankful for his opportunity as instructor of the spring 2008 consumer chemistry co urse, CHM 1083. After departing Flo rida, Michael was an adjunct instructor at Middlesex County College for one year for general chemistry lecture and laboratory courses. Thereafter, he returned to State College, PA, and all of its bucolic splendor to complete his dissertation. ture plans will not be limited by the myopia of fortune or materialism, but to see beyond, toward the virtue and idealism of empiricism and truth, regardless of manifestation of employment.