Interpretation of Mass Spectra for Elemental Speciation Studies
A dissertation submitted to the
Division of Research and Advanced Studies
of the University of Cincinnati
in partial fulfillment of the requirements for the degree of
Doctor of Philosophy
in the Department of Chemistry
of the College of Arts and Sciences
2005
by
Juris Meija
B.Sc., Chemistry, University of Latvia, 2001
Committee Chair: Professor Joseph A. Caruso ABSTRACT OF DISSERTATION
During the last decade, mass spectrometry has become a powerful tool in understanding the various aspects of molecular processes occurring in biological systems. This paves the way to the understanding of complicated life processes that are among the greatest challenges in the contemporary bioscience. Sample preparation is a critical area in elemental speciation analysis and it is important that techniques used to extract elemental species are efficient yet correctly reflecting the chemical species present in native sample. However, with all respect to sample preparation, data analysis and critical evaluation of experimental observations based on the molecular-level explanation is becoming a major obstacle in transferring the experimental knowledge into valid conclusions. Many problems in the context of mass spectrometry can be solved using techniques of computer sciences, graph theory, and discrete mathematics. The aim of this dissertation is to recollect several essays that demonstrate the power and the need for developing skills in mass spectrometry data interpretation. These include the study of chemical behavior of selenium bio-volatiles using mass spectrometry, element-specific fingerprinting of mass spectra using isotope patterns, data analysis for accurate isotope ratio measurements and isotope pattern analysis as an extension of isotope dilution analysis.
PREFACE
When Alice stepped through the looking glass, she did and so I sketched the corresponding equations on not know what she will do there nor did she had even a the back of my airplane itinerary. Later we slightest clue what ever will happen to her. My refined the mathematics with Nacho and what we dissertation is just such a journey. When I joined Doc’s called the “general approach to the method of group, he listed about four or so potential project ideas isotope dilution” was presented later that year in and I seem to like the one that he thought was perhaps Plymouth, UK (Chapter 6). This is how the main research projects initiated. the least interesting. When I saw a slide of Maria’s [Montes-Bayon] preliminary work that plants emit I would most like to thank my advisor (whom we
CH3SeCH3 it sounded interesting, but then she showed call simply Doc) Professor Joe Caruso not only that plants also can emit CH3SeSeCH3. For a moment I for his guidance throughout my graduate career, thought this was the coolest thing in the world and I I thank him for allowing me to be myself, for immediately became fascinated with the chemistry of allowing me to pursue projects even though selenium volatiles, even though all I knew about volatile sometimes they had not much to with the Se was that these things “exist”. This lead to about four research interests of the group. He supported me publications (Chapters 2-4), including one pirated (!) in all those weird “virtual” studies that I publication (Sepu 2004 (22), 16–19). In autumn of 2003, performed sitting at my laptop computer. I thank shortly before going to Oviedo, I submitted a tutorial him for being such an interesting advisor. This manuscript on isotope pattern reconstruction (J. Am. Soc. had a large impact on my attitude towards the big Mass Spectrom. 2004 (15), 654–658). About two weeks things in the life. It is a great art to present a later Nacho [Garcia Alonso] came to me and told that he research seriously, yet attractive and in a very liked the manuscript (he was one of the reviewers). In a charming manner. Joe is a master of this art. two minute hallway talk later that day I told that I could Anne Vonderheide had tremendous impact on apply the algorithm to explain some odd things my English writing abilities and she encouraged regarding the accuracy of isotope ratio measurements in me so much. Not to mention that without our electron impact mass spectra of tin compounds. About extraordinary extensive library at the University week later we had the general fragmentation mechanism of Cincinnati I could not even pursue a single of these compounds (J. Mass. Spectrom. 2005 (40), in manuscript. And last, I thank Maria so much press). This scheme also explained the origin of about 2- again and again about those two wonderful 3% systematic error observed by few other research autumns that I could spend in Oviedo research groups. Few months later I was listening to some talk in group. There were so many other nice people the winter conference on plasma spectrochemistry (Jan involved in my research projects that I simply 2004, Ft. Lauderdale, FL) and it just occurred to me that cannot list them here. I just want to say that I the isotope pattern reconstruction algorithm is essentially was very delighted to collaborate with all of a reverse method of isotope dilution. I was quite excited them. juris
TABLE OF CONTENTS
CHAPTER 1 | ELEMENTAL SPECIATION………………………………………………… 3
1.1. Elemental Speciation: The concept and the need 1.2. Conceptual approaches to elemental speciation 1.3. Instrumental approaches
CHAPTER 2 | MASS SPECTRA OF SELENIUM BIOVOLATILES…………………………… 13
2.1. Abstract 2.2. Introduction to selenium biovolatiles 2.3. Preparation of selenium standards 2.4. Mass spectra of diselenides and selenosulfenates i. EI+ ionization ii. EI– ionization iii. CI+ ionization iv. CI– ionization 2.5. Conclusions 2.6. Experimental
CHAPTER 3 | POLYCHALCOGENIDES AND THEIR MASS SPECTRA……………………… 38
3.1. Abstract 3.2. Chalcogenide exchange reaction i. Formation of polychalcogenides 3.3. Mass spectra of trichalcogenides i. Triselenides ii. Branched trichalcogenides 3.4. Experimental
1 CHAPTER 4 | THEORY OF ISOTOPE PATTERN RECONSTRUCTION……………………… 58
4.1. Abstract 4.2. Introduction 4.3. Isotope pattern reconstruction 4.4. Signal deconvolution in the mass domain i. Signal peak shape analysis ii. Peak centroid mass analysis 4.5. Application to dimethyl diselenide mass spectra i. Deconvolution in the intensity domain ii. Deconvolution in the mass domain ii. Peak centroid mass shift analysis 4.6. Aspects of isobaric interferences in mass spectra
CHAPTER 5 | INTERPRETATION OF BUTYLTIN MASS SPECTRA FOR ISOTOPE RATIO MEASUREMENTS…………………………………. 73
5.1. Abstract 5.2. Introduction 5.3. Mass spectra of butyltin compounds i. Interpretation of mass spectra using isotope pattern reconstruction ii. Fragmentation behavior of butyltin compounds 5.4. Isotope ratio measurements in fragment ions 5.5. Conclusions 5.6. Experimental
CHAPTER 6 | ALTERNATIVE LOOK AT THE METHOD OF ISOTOPE DILUTION…………… 91
6.1. Abstract 6.2. Introduction 6.3. Theory 6.4. Results and Discussion i. Effects of carbon isotope variations ii. Application to butyltin quantitation iii. Uncertainty of isotope pattern reconstruction iv. Applications to multiple spike isotope dilution 6.5. Conclusions 6.6. Experimental
CHAPTER 7 | PROSPECTIVES……………………………………………………………. 110
Mathematical speciation
CHAPTER 7 | REFERENCES……………………………………………………………… 114
2
“Perhaps Looking-glass milk isn't good to drink…”
/Lewis Carroll, Through the Looking Glass/
CHAPTER 1 | ELEMENTAL SPECIATION
3 1.1. Elemental speciation: the concept and the need
For more than 450 years, toxicologists have relied on an idea expressed by Paracelsus in the fifteenth century: "Alle Dinge sind Gift und nichts ohn Gift; alein die Dosis macht das ein Ding kein Gift ist” [all things are poison and not without poison; only the dose makes a thing not a poison”]. With the exception of E = mc2, perhaps no other single statement has wielded such force in establishing the popular notoriety and the professional stature of an individual in the history of science as the words just quoted. The statement of Paracelsus is clearly one of the main driving forces in modern elemental speciation studies. Nowadays it is known that molecular structure governs properties of compounds and not the individual constituents of that molecule. For example, from two different arsenic-containing compounds one might be toxic
th (such as Me2AsOH) and the other - harmless (such as arsenobetaine). In the middle of the 18 century Louis Pasteur established the fact that the biological properties of organic compounds are structure specific and soon after the very same idea was advertised by English mathematician/writer Lutwidge Dodgson (known as Lewis Carroll). In his work “Through the looking glass” he incorporates the monologue “perhaps Looking-glass milk isn't good to drink…”, most likely referring to the properties of lactic acid mirror isomers.
Elemental speciation refers to the characterization of the species of one or more particular elements. The ultimate goal for this analytical activity is to identify and possibly quantify one or more species associated with a certain sample. Speciation (speciation analysis) is driven by the need for better risk/benefit assessments than total elemental analyses can provide. The toxicity of an element depends on its physical-chemical form present in the sample and on its capacity to move through the intestinal barrier. However, elemental species identification is not always an easy task, especially for high molecular weight species. Speciation includes the elucidation
4 of the oxidation state, total charge or molecular weight of the species, binding strength of the elements, etc. In all the cases, the information obtained leads to the ultimate structural identification of the particular species of interest and, as necessary, its quantification. Sample preparation, analyte extraction, pre-concentration or enrichment play important roles in trace element speciation and sometimes they are the key factors, since often speciation analyses are in sub-ppb range.1,2 Elemental speciation in foods is of particular interest because of their high consumption. For example, characterization of Al species in tea or heavy metals in wine is of great interest because of the toxicity and concentrations of these elements. It is essential to understand the nature of these species within the beverages to predict their potential interaction with the human body.
The field of elemental speciation has rapidly expanded in the past decade.3,4 Starting with metal speciation for environmental interest (Hg and Pb) and followed by other metals and non- metals, speciation now is evolving toward biological systems, including chiral speciation and isotopic profile characterization of the compounds. This shift is driven by the availability of coupled separation and detection techniques. One of the most important detectors to date is inductively coupled plasma mass spectrometry (ICP-MS). In addition, ICP-MS is not just a sensitive and selective metal detector but it can access other elements in the periodic table.
Elements currently unavailable for detection with ICP-MS are only those restricted by Ar ionization potential (He, Ne and F) or atmospheric gases (O, C and N). ICP-MS is emerging as a versatile ultra-trace level detection technique capable for selective and sensitive detection of metals and non-metals.
5 1.2. Conceptual approaches to elemental speciation
Dissection of the molecular mechanisms involved in (semi)metal hyperaccumulation in land plants is one of the biggest challenges in molecular biology. A biochemical approach to solve such problems is the identification of genes involved in the accumulation process that can provide also valuable genetic resources for the future development of plants, which would ideally be suited for phytoremediation. In this regard, molecular approaches are increasingly used to provide new insights into the role of metal transporters and assimilation enzymes in uptake and metabolism of (semi)metals. An alternative of approaching an element’s metabolism in plant tissues is to study the metabolic pathway of (semi)metals through the elucidation of their complexes with bioligands and their compartmentalization within the plant. For this purpose, the joint research effort of analytical chemistry and plant physiology is providing fruitful results in molecular characterization of metallo-complexes (metallomics).4
The use of molecular biology methods and those of analytical chemistry have not overlapped to a large extent in research studies. Developments in (semi)metal metabolism are gathered largely on a basis of separate collection of knowledge. However, it is a becoming more of a fashion to combine the best of both worlds. For example, in a recent study by Vacchina et al., hyphenated analytical chemistry techniques (HPLC and CE coupled to ICP-MS and ESI-MS) are assisted by molecular biology techniques (functional yeast complementation) offering a new and attractive way of investigating the processes of metal accumulation and metal tolerance in hyper-accumulating plants (Figure 1.1).5
6
Figure 1.1 | Elucidation of main Ni-containing species in Ni hyperaccumulator plant T. caerulescens by complementary use of integrated mass spectrometry and molecular biology techniques. Adapted from Vacchina et al.5
After the metal-complex species have been detected and identified using chromatography and mass spectrometry techniques, it is common among analytical chemists to simulate the chelation chemistry in isolated ligand-(semi)metal systems. This approach allows careful exploration of the complexation phenomena in easily controlled conditions. Almost as a rule, all the (semi)metal and biomolecule interactions have been studied using this approach. Among most popular examples are the studies of complexes of glutathione and arsenic (using ICP and
ESI)6 or selenium (ICP and ESI)7, phytochelatin complexes with arsenic (ICP and ESI)8, mercury9, cadmium (ESI)10, mercury complexes with biological thiols (MS)11 and S/Se chalcogenide exchange reaction simulations (ICP and EI)12. The majority of these studies were conducted using both atomic and molecular mass spectrometry techniques. It is of importance to mention that the balance between the ionization efficiency and species stability is a ubiquitous problem in metallobiomolecule analysis.
7 Some other trends deal with the use of mass spectrometry to provide information about chemical behavior. Direct correlations between mass spectral ion intensities and calculated solution equilibrium distributions have been found, thus opening a novel avenue for further exploration of complexation phenomena. ESI-MS can be used to determine metal-ligand binding energies.13 Also, the metal binding site to peptides and other biomolecules can be determined from ESI-MS analysis. For example, in low energy collision activated dissociation
ESI-MS calcium/peptide complexes undergo fragmentations that are controlled by Ca2+ binding, thus providing the binding site information.14 The question of whether ESI-MS may be used to probe solution-phase properties now is clearly demonstrated. Recent studies of mercury(II)-bis-thiolate ESI mass spectra showed a peculiar fragmentation processes observed, loss of neutral ammonia from protonated Hg(II)-bis-thiolates with free, protonated amino groups leads to the formation of thiirane-carboxylic bound species:
COOH
+ H3N COOH Loss of NH3 S Hg S + H O S Hg SH O McLafferty NH rearrangement 2
This may hint at unforeseen mechanisms for the interaction of mercury(II) ions with biological thiols, ultimately leading to cellular and organ toxicity.11
Finally, computational approaches for studying various chemical problems have lately gained increased popularity, especially in the biochemical field. This is a routine practice among organic chemists, who often perform electronic structure calculations to explore or verify their experimental results. Besides the benefit of investigating separate isolated processes
(such as bonding, or pKa), modern computational approaches achieve a fairly high level of
8 accuracy within reasonable computational time. Electronic structure modeling can address the stability of the species, bond energies, and also predictions of molecular interactions. Recent computational and synthetic studies also showed that bis(thio)selenide-type compounds are likely formed during the sequestration of selenite by phytochelatins.15 Besides the electronic structure modeling, computations are also used to interpret the mass spectra of metallobiomolecules. For example, isotope pattern deconvolution of the observed molecular ion for PC-Cd complex has been used to suggest the presence of internal S–S bridges.10 A similar approach is used to account for the Fe3+ and Cu2+ induced oxidized plant flavonoids in ESI mass spectra.16 Also, the location of Se-peptides in MALDI-TOF-MS spectra of plant Se- protein proteolytic digest is assisted by the changes in the peptide isotope patterns.10
1.3. Instrumental approaches
Although an increasing number of publications are addressing the capabilities of analytical tools to study the effect of certain genetic modifications in the metabolism of metals and metalloids, they are still scarce. This is probably due to the limitations of conventional biochemical approaches in terms of specificity and sensitivity. However, the instrumental developments in organic and inorganic mass spectrometric techniques over the last decade have enhanced our ability to do such complex studies. Thus, elemental speciation studies, which combine a powerful separation with atomic (elemental) detectors, provide selective and sensitive detection of the (semi)metals present in the complexes formed in vivo.
Hyphenated techniques (coupled techniques) play an important role in elemental speciation at trace levels. Methods are hyphenated because of the synergy resulting by coupling powerful separation and detection techniques. One of the most recent tributes to hyphenated techniques
9 came with a number of articles in the special issue of Analytical and Bioanalytical Chemistry in
July of 2002. By using an element specific detector, ICP-MS or ICP-AES, constraints are minimized with respect to carefully separating species containing different elements. The separation only requires separating different compounds of the same element and simultaneous multi-elemental monitoring (time resolved analysis) allows one to obtain several element specific chromatograms within a single injection. Multi-element specificity and accurate quantification capabilities are the key features of ICP-MS whereas molecular weight determination and structural information are those contributed by electron impact, ESI and
MALDI techniques. Both elemental and molecular mass spectrometry, supported by on-line or off-line coupling to chromatographic or electrophoretic separation methods, have created the basis for new structural and/or quantitative insights, demonstrating the analytical excellence of this approach.17
For detecting molecules, electrospray-mass spectrometry (ESI-MS), introduced in 1985, combines a soft ionization source with mass spectrometric detection. The capability of ESI-MS for studying weak, non-covalent interactions between metal cations and organic ligands was recognized almost immediately after the introduction of this technique and a significant number of papers on this subject have already been published. This advantage, however, was not commonly acknowledged among analytical chemists until five years ago, as discussed on the analysis of intact molecules produced during metal accumulation in plant tissues.18 MALDI-
TOF-MS is the most used alternative to ESI. While the advantage of this technique is higher sensitivity and lesser solvent adduct formation, it cannot be easily coupled to a chromatographic separation system. Also, the ability to detect non-covalent complex ions in MALDI/MS is highly dependent on the choice of the matrix. MALDI/MS has had little use in studying
10 coordination complexes of (semi)metals in biological tissues, although some examples can be found in the literature. The structural features of organic compounds are most often elucidated from tandem mass spectra (ESI-MS2, ESI-Q-TOF, etc.), however, MALDI/post-source decay is a less expensive alternative to this and its use in peptide structure analysis has been demonstrated.19
Choice of detection method. In ICP-MS a high temperature plasma discharge generates positively (and mostly singly) charged atomic ions. Ions formed in ICP can be detected according to their mass (with quadrupole mass filters, magnetic sectors or time-of-flight mass analyzers) or by measuring their characteristic emission (ICP-AES). Among the most widely used plasma-based detection techniques are ICP-AES and ICP-MS. Probably the most exclusive feature of ICP-MS in comparison to photon-based plasma spectroscopy (ICP-AES) is the ability to discriminate between the isotopes. Along with this feature, ICP-MS offers the advantage of doing isotope dilution analysis and stable isotope tracer analysis, thus eliminating the need for radioactive tracers. Compared with AES, plasma MS detection limits are generally 2-3 orders of magnitude lower (Figure 1.2). On the other hand, AES offers higher sample throughput, higher tolerance to the dissolved particles and a lower price. One of the biggest problems in AES is spectral interferences which cause considerable problems in quantification.
SEC/ICP-MS AAS
CE/ICP-MS ICP-AFS
SFC/ICP-MS ICP-AES
HPLC/ICP-MS GF-AAS
GC/ICP-MS ICP-MS
10-4 10-3 10-2 10-1 100 101 102 103 104 10-4 10-3 10-2 10-1 100 101 102 103 104 ppb ppb
Figure 1.2 | Relative detection limits (in parts-per-billion) of major hyphenated ICP-MS time-resolved techniques (left) and plasma and flame based detection techniques (right).3
11 TOF mass analyzers allow for effective fast transient signal acquisition with multi-element capabilities, not as easily done with the quadrupole analyzer, so fast transient signals as with
GC are not limited to only one element. However, the ability of performing simultaneous multi- element analysis with TOF-based instrument sacrifices the detection levels required for biological sample analysis. Sector field instruments have the advantage of high mass resolution and thus isobaric interferences and polyatomic interferences from sample matrices is less of a problem. High resolution, however, is achieved by narrowing the entrance slit width, thereby compromising the sensitivity. Thus, there is always a trade between the selectivity and sensitivity. Similar to quadrupole instruments, the mass scanning rate is relatively low because of the hysteresis of the magnetic field.
To conclude, hyphenated techniques are approaching common use in analytical laboratories in order to selectively separate, detect or determine metal or metalloid containing species. The interpretation of mass spectra together with the help of computational devices permits even a closer approach to the identification and characterization of the naturally occurring semi(metallo) complexes.
12
CHAPTER 2 | MASS SPECTRA OF SELENIUM BIOVOLATILES
13 2.1. ABSTRACT
The mass spectral fragmentation of aliphatic diselenides and selenosulfenates is analyzed to gain a better understanding of the behavior of these species. The main fragmentation pathways of these species include the fragmentation along the Se–C bond, fragmentation along the Se–Se or Se–S bonds and intra-molecular rearrangements. In general, negative ionization favors the fragmentation along the Se–Se or Se–S bonds while positive ionization leads to stable molecular ions. Density functional theory calculations of bond dissociation energies and molecular orbital analysis was undertaken to explain the observed trends in molecular fragmentation. Besides the analysis of molecular fragmentation, a phenomenon of molecular association in negative electron impact and positive chemical ionization conditions was observed and investigated using a high resolution time-of-flight mass spectrometer. Molecular association that occurs during the ionization of species includes the formation of symmetrical diselenides from asymmetrical selenosulfenates and formation of alkylseleno adducts from the corresponding diselenides. For species which are hard to resolve by mass analysis, such as isobars of CHSe, CH2Se and CH3Se, the isotope pattern superimposition procedure was applied to define the overlapping clusters.
14 2.2. INTRODUCTION TO SELENIUM BIOVOLATIES
Interest in selenium species has recently increased due to their anti-oxidative and anti- carcinogenic properties. Biological activity of selenium compounds is of high importance since this element is now recognized as an essential element. Interestingly enough, selenium is the only trace-element that is found in the human genetic code. Most of the selenium compounds are red.-ox. active and as such they account for the anti-oxidative action of Se (Scheme 2.1).
ONOO- ONO- ROOH ROH
Se– SeX–
GSSG GSH
GSH HX (X = OH, I)
SeSG
Scheme 2.1 | Formation of selenium conjugates in regeneration of several antioxidant systems in human body.
Speciation of selenium metabolites in the environment has been undertaken by many research groups to gain more understanding about bio-transformations and the occurrence of the various selenium compounds. Inductively coupled plasma mass spectrometry (ICP-MS) is usually utilized for efficient detection and screening of the various selenium species at ultra- trace levels; however, the ultimate identification necessitates characterization by molecular mass spectrometry 20.
Many plants (and living organisms) volatilize or emit sulfur species mainly as a means of self-defense. For example, isothiocyanates in Brassicaceae plants are produced to protect them against insect attack and fungal infection. Similarly, Allium sulfur volatiles (disulfides, trisulfides and thiols) are released to reveal the presence of herbivores to its natural
15 parasitoids.21 As early as 1894, Hofmeister proposed that selenium in animals is detoxified by releasing volatile dimethyl selenide from the lungs. He based this proposal on the fact that the odor of dimethyl telluride was detected in the breath of dogs injected with sodium tellurite.
Using the same logic, it was suggested that the garlicky odor of plants that accumulate selenium may indicate the release of volatile selenium compounds. Lewis was the first to show that both selenium non-accumulator and accumulator species volatilize selenium.22 This was later confirmed by other authors.23-25 The volatile selenium compound released from the selenium accumulator Astragalus racemosus was identified as dimethyl diselenide.24
Volatile selenium species are of interest as the end terminal in Se metabolism, and are emitted by plants as a means of self-detoxification.26 Selenium methylation and eventual volatilization is one of the most important processes of the biogeochemical cycling of selenium in aqueous and terrestrial environments. Selenium metabolism in plants starts with the uptake of the corresponding species from the rhizosphere. In many plant species selenate is generally taken up actively by the same mechanism as sulfate is absorbed. Selenite uptake, on the other hand, seems to be passive. Rhizosphere bacteria increase the plant uptake of Se through the formation of proteinaceous intermediates. After selenate is taken up by the sulfate membrane
2- transporter, SeO4 is activated to adenosine phosphoselenate (although the identity of this compound has not been established to date), which can then react non-enzymatically with the tripeptide glutathione (GSH) or other abundant thiols. The formed bis(alkylthio)selenides eventually are reduced to the corresponding alkylthioselenides which are the precursors for selenocysteine biosynthesis. Similarly, selenite is scavenged by GSH followed by the formation of corresponding thioselenide. Biological in vivo synthesis of selenomethionine from selenocysteine leads to the formation of the dimethyl selenide, mainly from the hydrolysis of
16 Se-methyl-selenomethionine and dimethyl selenopropionate. This is a typical case for Se non- accumulating plants. Selenium hyperaccumulating plants are apparently able to tolerate high Se concentrations by methylating selenocysteine, which is stored or volatilized as a dimethyl diselenide. During the period of these studies, ICP-MS was an important tool which, combined with reverse phase HPLC, allowed selective screening of Se-containing amino acids, thus effectively informing the outcome of genetic modification.27 Also, the coupling of gas chromatography to ICP-MS allowed very fast and effective monitoring and quantitation of Se- and S-containing volatile species emitted from the plants.28 It is interesting to note that Se non- accumulating plants differ from the Se accumulators with respect to the Se volatiles emitted. In general, non-accumulators produce Se-methionine which results in volatile dimethyl selenide; however, accumulator plants store Se in form of Se-methyl-SeCys, which leads to dimethyl
26 diselenide. Thus, the ratio of volatilized CH3SeCH3 and CH3SeSeCH3 serves as a proxy to the pathway of Se-accumulation (Figure 2.1).
Figure 2.1 | Monitoring the Se metabolites in wild type and genetically modified Brassica juncea using Se-specific HPLC/ICP-MS detection. Adapted from Montes-Bayón et al.29
17 2.3. PREPARATION OF SELENIUM STANDARDS
The availability of selenium-containing volatile standards is limited to a small number due to their instability and infrequent use in environmental analysis, despite the growing interest in Se biospecies relative to both their toxic and health benefit potential. Selenium-containing volatiles of environmental interest can be classified into four main groups: selenides (–Se–), diselenides
(–SeSe–), selenosulfenates (–SeS–) and bis(alkylthio)selenides (–SSeS–). Symmetrical diselenides and selenides are usually available commercially and may be used as standards.
They can also serve as the starting point for the preparation of other species of interest, such as selenols and selenosulfenates.
Selenols. Thiols are strong reducing agents, stronger than selenols, and thus, diselenides may be reduced to selenols by thiols. This reduction process is reversible and thus the thiols that provide the highest equilibrium constant for this reaction are chosen. 1,4-dithio-2,3-butanediol
(DTT) is a good candidate because of the high equilibrium constant and resistance towards the oxidation from dissolved oxygen:
HO OH HO OH HO OH CH3
H3C Se DTT Se HS HS HS Se S SS
(DTT) H3C CH SeH CH SeH 3 3
Similarly, reduction with NaBH4 can be used to obtain selenols from the corresponding diselenides.
Selenosulfenates. Aliphatic selenosulfenates are easily prepared in a mixture utilizing the sulfur/selenium exchange reaction, which occurs at room temperature in aqueous or organic solvent medium:
Me Et Me Et Se Se Se Se Se Se Se Se Me Et Me Et
18 The equilibrium is reached within few hours at room temperature. This type of exchange is an entropy driven process with the equilibrium constant K ≈ 4.30 Asymmetrical diselenides and selenosulfenates are prepared by mixing the corresponding symmetrical precursors. For example, Boss et al. reported the formation of all 28 asymmetrical diselenides (after 2 h) when eight symmetrical diselenides were mixed together in n-hexane.31 The yield of dimethylselenosulfenate, MeSeSMe, by mixing MeSSMe and MeSeSeMe is very small and the process is slow. It was found that reaction of dimethyltrisulfide, MeSSSMe, and MeSeSeMe is statistically more favorable for MeSeSMe production and this species is formed to a much greater extent when MeSSSMe and MeSeSeMe are mixed together.
In the case of the acyclic selenium volatiles, the identification based upon retention time matching has strong reliability because of the interaction chemistry. Diselenide exchange and interaction of selenols offers an additional identification confirmation. In other words, it is unlikely there will be high levels of dimethyl selenosulfenate without the presence of dimethyl diselenide and dimethyl disulfide or methyl thiol. Moreover, element isotope specific detection allows one to confirm the correct Se isotope pattern within each of the chromatographic peaks, thus confirming the absence of isobaric interferences.
In addition to preparing selenium and sulfur-containing volatiles, retention times of homologues may be predicted using retention indices and boiling point correlations.32 Using such an approach, for example, the confusion between the MeSeSMe and MeSe(O)2Me was easily resolved.33
19 2.4. MASS SPECTRA OF DISELENIDES AND SELENOSULFENATES
Despite the interest in the selenium volatiles, very few reports are devoted to the investigation of mass spectral fragmentation of heavier dichalcogenides – diselenides (R-SeSe-R) and selenosulfenates (R-SeS-R). Early mass spectrometric studies of organic selenium compounds were undertaken by Rebane 34-36 and recently by Prabhakar et al.37. However, diselenides and selenosulfenates were not a subject of interest in either of the studies. In this study we analyzed the mass spectral fragmentation of methyl-, ethyl- and ethylmethyl- diselenides and selenosulfenates to gain a better understanding of the behavior of these species. In order to explain the fragmentation behavior of diselenides and selenosulfenates, density functional theory calculations were performed to obtain the bond dissociation energies and molecular orbital information.
2.4.i EI+ ionization
In general, alkyl disulfides, selenosulfenates and diselenides give intense molecular ions in EI+ spectra. This can be easily explained by the properties of electronic structure of these species.
The highest occupied molecular orbital (HOMO) in these compounds is a set of two perpendicular (non-bonding) lone pairs on chalcogen atoms (Figure 2.2).
20
Figure 2.2 | Removal of one electron from diselenides results in planar geometry of the obtained radical cation. Note that the electron is removed from the anti-bonding π-π molecular orbital. Calculated using ub3lyp/6-311+G(2d).
The removal of an electron from diselenides or selenosulfenates (and disulfides) results in planar radical cation where the electron is removed from the anti-bonding π-π orbital, thus increasing the Se−Se, Se−S and Se−C, S−C bond energies (Figure 2.2 and Table 2.1). This explains the inherent stability of the positively charged molecular ions of diselenides and selenosulfenates observed in positive ionization.
Table 2.1
Calculated bond dissociation energies of dimethyldiselenide E(C−Se), E(Se−Se), Species kJ mol−1 kJ mol−1 MeSeSeMe 198 202a MeSeSeMe.+ 274 433 MeSeSeMe._ 98 100 Method of calculation: ub3lyp/6-311+G(2d) a) Experimental value of 192 ± 12 kJ mol−1 is given by Tel’noi et al.38
21 EI+ fragmentation of dimethyldiselenide starts with the removal of electron from the molecule. From here, further transformations occur according to the three main pathways: 1)
Fragmentation along the Se−C bond, 2) Fragmentation along the Se−Se bond and 3)
.+ Intramolecular rearrangement yielding selenoseleninyl radical cation, R2Se−Se (Scheme 2.2).
Scheme 2.2 | Mass spectral fragmentation of the dimethyldiselenide in EI+ ionization mode (mass based on 80Se).
Fragmentation along the Se−C bond is the main fragmentation pathway in
. dimethyldiselenide and similar compounds. This seems to occur by (sequential) loss of CH3 radicals leading to the MeSeSe+ and SeSe.+ (see Figure 2.3). In positive ionization mode, fragmentation along the Se−Se bond is not highly favorable as the least energy demanding route is the cleavage of Se−C bond (see Table 2.1). Besides that, MeSe−Se+ cations formed during the Se−C bond cleavage can be stabilized by atomic charge dislocation via the resonance equilibria [Me−Se−Se+ ↔ Me−Se+=Se], which is reflected in the atomic charges on the selenium atom in this species (both Se atoms bear equal positive charge of 0.5). Evidently, cleavage along the Se−Se bond is not greatly favored, as indicated by the low abundance of
+ CH3Se at m/z = 95 (see Figure 2.4).
22
Figure 2.3 | High resolution mass spectra of dimethyldiselenide in EI+, EI–, CI+, and CI– ionization modes.
Intra-molecular rearrangement of dimethyldiselenide or dimethylselenosulfenate is evidenced by the formation of a C2H5Se cluster as it can logically occur only via the loss of
. .+ HSe radical from the dimethylselenoseleninyl radical cation, Me2Se−Se , yielding
+ .+ .+ CH2=Se−CH3 (m/z = 108.9556). The presence of both - C2H5S and C2H5Se ions in the EI+ mass spectra of dimethylselenosulfenate suggests that intramolecular methylgroup
.+ .+ rearrangement of MeSeSMe yields both – thioseleninyl (Me2Se−S ) and selenosulfaninyl
.+ (Me2S−Se ) ions (Table 2.2).
23 Table 2.2
Alkylgroup rearrangements in EI+ ionization of dichalcogenides (see Scheme 2.2) Compound Rearrangement Fragmentation Fragment ion (intensity)a Measured mass (error)b
+ Me–SeSe–Me Me-migration [M – HSe] CH3–Se=CH2 (15%) 108.9566 u (+1.0 mu) 1 2 2 + Me –SeS–Me Me -migration [M – HS] CH3–Se=CH2 (2%) 108.9559 u (+0.3 mu) 1 + Me -migration [M – HSe] CH3–S=CH2 (8%) 61.0115 u (+0.3 mu) + Me–SeSe–Et Me-migration [M – HSe] CH3CH2–Se=CH2 (0.4%) 122.9703 u (-1.0 mu) + Et-migration CH3–Se–CH2CH2 + + Me-migration [M – CH3Se] CH3–Se=CH2 (-) overlaps with EtSe Et-migration + Me–SSe–Et Et-migration [M – HSe] CH3CH2–S=CH2 (0.4%) 75.0271 u (+0.3 mu) + Et-migration [M – CH3Se] CH3–S=CH2 (1%) 61.0117 u (+0.5 mu) + Me-migration [M – HS] CH3CH2–S=CH2 (0.4%) 75.0271 u (+0.3 mu) + CH2–S–CH2CH2 + + Me-migration [M – CH3S] CH3–Se=CH2 (-) overlaps with EtSe + Et–SeSe–Et Et-migration [M – HSe] CH3CH2–Se–CH2CH2 (0.6%) 136.9892 u (+2.2 mu) + Et-migration [M – CH3Se] CH3CH2–Se=CH2 (0.4%) 122.9730 u (+1.7 mu) 1 2 1 + Et –SeS–Et Et -migration [M – HSe] CH3CH2–S–CH2CH2 (0.2%) 89.0421 u (-0.4 mu) 1 + Et -migration [M – CH3Se] CH3CH2–S=CH2 (0.3%) 75.0272 u (+0.4 mu) 2 + Et -migration [M – HS] CH3CH2–Se–CH2CH2 (-) not detected 2 + Et -migration [M – CH3S] CH3CH2–Se=CH2 (-) not detected a) Relative to the molecular ion. b) Based on the 80Se ion. Error is given in milli-mass units
This is also in agreement with the total atomic charge distribution in dimethylselenosulfenate radical cation (molecular ion). Atomic charges on sulfur and selenium are close (+0.3 and +0.6 respectively), and so are the atomic charges on both carbon atoms (−1.0). In such a case atomic charge-driven methylgroup migration can occur for either of them in MeSeSMe.+. Table 2.2 summarizes the fragments characteristic of the alkylgroup migration that cannot be attributed to the direct fragmentation. Note that the abundance of some these ions approaches up to 10−15% of the molecular ion.
Rearrangements seen from dimethyldiselenide or selenosulfenate suggest 4-member ring hydrogen transfer. However, species like diethyldiselenide or diethylselenosulfenate, in addition
. . to the loss of HSe radical show also the loss of CH3Se radical (Table 2.2), which is consistent with a 5-member ring hydrogen transfer rearrangement (Scheme 2.3).
24
Scheme 2.3 | Alkylgroup rearrangements in EI+ ionization of dichalcogenides.
The observed selenoseleninyl rearrangement is disadvantageous because it may be confused
+ .+ with the presence of ethylseleno ion (CH3CH2Se vs. CH2SeCH3 ), which is the only mass spectrometric (in EI+) evidence that the ethyl group is directly attached to the selenium atom.
Internal rearrangement is especially disturbing in the case of asymmetrical species, such as
MeSeSEt and EtSeSMe, where differences between the structural isomers cannot be conclusively elucidated by inspecting EI+ fragmentation patterns. Additionally, the use of GC retention time information is limited because of the very close structural properties and the lack of pure standards for asymmetrical selenosulfenates.
80 + The relatively low abundance of methylseleno cation, CH3 Se (m/z = 95) with respect to
78 + 39 m/z = 93 (usually ascribed to CH3 Se ) has been a subject of confusion . The fact that m/z =
93 abundance in EI+ spectra of dimethyldiselenide is of higher abundance than of m/z = 95
+ 78 cannot be explained only by the signal assignment to CH3Se because the abundance of Se is about two times lower than that of 80Se (see Figure 2.4). The distorted natural selenium isotope
80 pattern suggests the presence of several isobaric species. Isobars of CHnSe (CH Se and
78 CH3 Se for example) can be resolved only on instruments with high mass resolving capability
25 of m/∆m > 10000. To elucidate the identity of species that form the isotopic pattern around m/z
= 89 − 98 in the mass spectra of dimethyldiselenide, the cluster observed in EI+ spectra was
40 mathematically reconstructed allowing for contributions of CHnSe species where n = 0 − 4 .
The results show that the CHnSe cluster in CH3Se-containing species (such as MeSeSMe,
+ .+ + MeSeSeMe, MeSeSEt etc.) consists of CHSe , CH2Se and CH3Se in the amount ratio of
+ + about 2: 1: 1. Note that CH3Se contributes only about 25% to the CHnSe cluster. CHnSe cluster is observed also in CI+ ionization while negative ionization renders pure CH3Se pattern
+ .+ (see Figure 2.4). This suggests that the formation of CHSe and CH2Se is associated with the
Se−C bond rupture, which is favored in positive ionization (see Scheme 2.2). Similar to methyl-
+ .+ diselenides and selenosulfenates, CHS and CH2S also contributes the most in the CHnS pattern of dimethyl disulfide EI+ mass spectra.
Figure 2.4 | CHnSe cluster in mass spectra of dimethyldiselenide obtained under EI+, EI–, CI+, and CI– ionization modes.
Reconstruction of EI+ spectra of ethylselenides has shown that −CH2CH3 eliminates mainly via the neutral loss of CH2=CH2 (80%) accompanied by minor (20%) loss of ethyl radical
26 (Scheme 2.4). This is also reflected in the composition of Se2 and SeS ion clusters. For
.+ example, the Se2 cluster in MeSeSeMe consists exclusively of Se2 . Replacement of one methyl
.+ group by an ethyl group (in MeSeSeEt) results in a cluster of 75% Se2 and 25% contribution of
HSeSe+. Further replacement of the methyl group (EtSeSeEt) results in a cluster of about 60%
.+ + .+ HSeSeH and 20% HSeSe (besides Se2 itself).
Scheme 2.4 | Loss of ethyl group from diselenides in EI+ ionization mode.
Detailed inspection of the C2HnSe cluster composition in dimethyldiselenide, ethylmethyldiselenide and diethyldiselenide shows that the C2HnSe pattern, similar to CHnSe, consists of several species. In the presence of an ethyl group attached to a selenium atom
(MeSeSeEt and EtSeSeEt), the C2HnSe cluster consists of about 35% C2H3Se, 20% C2H4Se and
40% C2H5Se. On the other hand, isotope pattern deconvolution of EI+ C2HnSe pattern in
DMeDSe (allowing for n = 0 − 6) shows the prevailing presence (> 95%) of only C2H5Se
+ species, which is consistent with the selenoseleninyl rearrangement yielding CH2=Se−CH3 .
MeSeSeEt and EtSeSeEt fragmentation, however, yields the species pattern consisting of
C2H3Se, C2H4Se and C2H5Se in the ratio of about 2: 1: 2.
Although EI+ is established as a standard ionization technique for structural confirmation, extensive fragmentation of aliphatic diselenides may lead to the loss of information regarding
27 the internal bonding. For example, the methylseleno cluster (around m/z = 93 − 95) is formed also from ethylseleno species and thus MeSeSEt and EtSeSMe are hard to distinguish from their
EI+ spectra, as the positive charge is exclusively retained by Se-species due to the low ionization energy of Se (9.75 eV) compared to S (10.36 eV) and C (11.26 eV).
2.4.ii EI− ionization
Negative electron impact ionization renders M.− molecular ions. However, the inherent instability of the negatively charged diselenides and selenosulfenates is apparent in the EI− mass spectrum of dimethyldiselenide, where the molecular ion is no longer the most abundant
(as in EI+). The reason behind this is that in negative ionization mode a valence electron is being added to the lowest unoccupied molecular orbital (LUMO) of the molecule. In aliphatic
41 disulfides, diselenides and selenosulfenates the LUMO is a σπ(X−X)* anti-bonding orbital and thus the addition of an electron results in σπ(X−X)* anti-bonding orbital now being the
HOMO (see Figure 2.5). This leads to a substantial increase of the X−X bond length and
* eventual fragmentation along this bond favored by the repulsion of the σπ(Se-Se) anti-bonding
HOMO. In agreement with this, it is experimentally estimated that low energy electron capture in disulfides lowers the S−S bond dissociation energy by about 40% 42.
28
Figure 2.5 | Highest occupied and lowest unoccupied molecular orbitals (HOMO and LUMO) in dimethyldisulfide, selenosulfenate and diselenide. Calculated using ub3lyp/6-311+G(2d).
The Se−C and Se−Se bond energies in negatively charged dimethyldiselenide are very close
(Table 2.1) and therefore the rupture of Se−C and Se−Se bonds is assumed to be comparably favorable, which is observed experimentally (see EI− and CI− in Figure 2.3).
Note that the secondary fragmentation in EI− occurs to a much lesser extent: the CHnSe
− cluster is entirely CH3Se and no selenoseleninyl rearrangement is observed (which would lead to the formation of CH2SeCH3). Also, elimination of the ethyl group occurs only via the loss of
. ethyl radical, CH2CH3. The loss of ethyl radical in negative ionization (and not neutral ethylene loss as seen in EI+) is favored as it leaves a negatively charged even-electron selenium- containing anion.
It is interesting to note that the formation of various adducts has been observed for asymmetrical diselenides and selenosulfenates in EI− conditions (see Figure 2.6). Formation of symmetrical diethyldiselenide from the asymmetrical CH3SSeCH2CH3 in the EI− conditions is
29 evident. Similarly, formation of diethyldiselenide has been observed in the EI− mass spectra of ethylmethyldiselenide. Formation of these species might seem like a reverse chalcogen exchange reaction (toward symmetrical species) which occurs between the neutral species and the most abundant negative fragments, RXSe−. However, in this case a distorted Se isotope pattern might be expected in the EI− mass spectra of MeSeSeEt due to the overlap of the
EtSeSe− and MeSeSeMe.−, which is not observed. Similarly, this reaction cannot be explained by the interaction of the negatively charged molecular ion and its main fragmentation product,
− CH3SSe . In such a case, formation of negatively charged dimethyldisulfide should be observed, which is not apparent.
A major difficulty in studying EI− fragmentation pathways lies in the fact that this technique has about three orders of magnitude higher detection levels in comparison to EI+ and lower abundance ions thus are hard to measure with high confidence. However, regardless of the mechanism, the formation of symmetrical chalcogen exchange reaction in EI− conditions has negative implications for molecular ion assignment. Due to this reaction, odd-electron symmetrical diselenides appear in the mass spectra at the masses larger than the molecular ion
(as shown in Figure 2.6). As a result, this might lead to erroneous molecular ion assignment taking into account the low intensity of the molecular ion, itself.
30
Figure 2.6 | High resolution EI– mass spectra of ethylmethylselenosulfenate (EtSeSMe) and diselenide (EtSeSeMe) showing the formation of symmetric diselenides in negative ionization.
2.4.iii CI+ ionization
In positive chemical ionization a formation of [M + H]+ ions takes place and the ratio of M.+ to
[M + H]+ under normal operating conditions (using isobutane) is close to 1: 1. Very small amounts of reagent gas based adduct formation are observed when using isobutane, the main
+ + adducts being [M + C3H5] and [M + C4H9] . The relative abundance of these adducts with respect to the molecular ion cluster usually did not exceed 1−4% for MeSeSeMe, MeSeSeEt and EtSeSeEt. As seen in Figure 2.7, the loss of the methyl group from the M.+ and [M+H]+ ion cluster renders clear the isotopic pattern of selenium, suggesting the presence of a single species
(supported also by isotope pattern superimposition calculations). This, however, is possible only
. .+ + by CH3 radical loss from M and neutral CH4 loss from the protonated [M+H] . Similar behavior, however, is not observed in the case of the ethyl group. Isotope cluster calculations of diethyldiselenide mass spectra suggest that protonated diethyldiselenide loses ethyl radical
. .+ + (C2H5 ), ethane (CH3−CH3) and ethene (CH2=CH2) yielding clusters of EtSeSeH , EtSeSe and
+ EtSeSeH2 .
31
Figure 2.7 | High resolution CI+ mass spectra of EtSeSeEt (a), MeSeSeEt (b), and MeSeSeMe (c) showing the intermolecular association of the species.
32 Interestingly, formation of [M+RSe]+ adducts from diselenides and selenosulfenates is observed in the CI+ ionization process. This occurs as a minor process and the [M + RSe]+ ion abundances are usually about 1−5% relative to the M.+. Similar adduct formation is observed also in the EI+ mass spectra, however, due to the extensive fragmentation processes, the abundance of [M + RSe]+ adducts in EI+ spectra is only about 0.1% relative to M.+. The formation of [M + RSe]+ adducts can be used for the structure confirmation studies of asymmetrical diselenides or selenosulfenates. For example, CI+ spectra of MeSeSEt exclusively
+ show the formation of [M + CH3Se] adduct, and thus MeSeSEt can be easily distinguished from EtSeSMe from their chemical ionization spectra. No formation of [M + RS]+ adducts has been observed in selenosulfenates. If the molecular association occurs by interaction of the molecular species, such a process is suspected then to occur in the field ionization (FI+) experiments. In field ionization the valence electron tunnels through its confining boundaries in the presence of the strong applied electric field. This is a very soft ionization technique, which results in the formation of the odd-electron molecular ion. No adduct formation (or fragmentation) from diselenides or selenosulfenates was observed in FI+ suggesting that the formation of the [M + RSe]+ adducts occurs by interaction of fragment ions (not present in FI+) with neutral molecular species as shown in Scheme 2.5.
+ Scheme 2.5 | Explanation for the formation of [M + CH3Se] adducts from dimethyldiselenide in EI+ and CI+ conditions.
33 This process is similar to the selenium exchange reaction; however, the increased bond energies in the presence of the positive charge greatly favor molecular association rather than exchange.
2.4.iv CI− ionization
Negative chemical ionization renders an almost exclusively intact pattern of the molecular ion,
M.−, which is beneficial for fast isotope pattern recognition. This is an advantage over CI+, where the molecular ion cluster is superimposed with [M + H]+ isotopic pattern. This leads to the normal (Gaussian) isotope distribution pattern thereby losing the characteristic pattern of
Sen clusters and thus only high resolution MS instrumentation can confirm the number of selenium atoms in the molecular ion cluster when doing CI+. Similar to EI−, CI− equally favors the cleavage of Se−Se (and Se−S) in addition to the Se−C rupture (see Figure 2.3). However, fragmentation occurs to a much lesser extent. Although CI− shows intense molecular ions for dichalcogenides, it should be used with care because of the existing possibility of fragmentation along the chalcogen-chalcogen bond molecular ion might be absent in lesser stable species, such as trichalcogenides (see Figure 3.7).
2.5. CONCLUSIONS
Mass spectral study of aliphatic diselenides shows that the principles of physical and organic chemistry are applicable to the fragmentation and rearrangement of ions in the gas phase. In general, positive ionization techniques favor the rupture of the C−Se bond, while negative ionization favors the rupture along the Se−Se bond. This principle is observed with EI+, CI+,
EI− and CI− ionization modes. In addition, mass spectral characterization of diselenides and selenosulfenates can be achieved not only by detailed inspection of the molecular fragmentation
34 (for example in EI+ or CI−) but also by inspecting the pathways of molecular association, as observed in CI+ and EI− ionization conditions. Due to the inherent stability of the positively charged radicals of diselenides and selenosulfenates, EI+ along with CI+ are safe choices for obtaining molecular weight information. Although electron impact ionization gives intense molecular ion signals, extensive ion fragmentation and ion rearrangements can inhibit the conclusive identification in more complicated species. As a result of the different fragmentation pathways of positively and negatively charged diselenides or selenosulfenates, the complementary use of positive and negative ionization techniques is recommended for characterization of diselenides and related species.
2.6. EXPERIMENTAL
Reagents and Standards. All reagents were of analytical grade and were used without any further purification. Dimethyldisulfide was purchased from Fluka (Milwaukee, WI, USA). Dimethyltrisulfide, dimethyldiselenide, diethyldisulfide and heptacosafluorotributylamine were purchased from Sigma-
Aldrich (Milwaukee, WI, USA). Diethyldiselenide was purchased from Strem Chemicals (Newburyport,
MA, USA). Chloro- and bromo-pentafluorobenzene, and 2,4,6-tris(trifluoromethyl)-1,3,5-triazine were purchased from Lancaster (Pelham, NH, USA).
Instrumentation. A high resolution Micromass GCTTM orthogonal time-of-flight mass spectrometer
(Micromass, Manchester, UK) coupled to GC was used for mass spectral characterization of the synthesized selenium species. High mass accuracy and precision is achieved by continuous leaking of a single reference compound directly into the ionization source from the reference inlet. A single fragment of the reference compound is used as an internal reference signal (lock mass). Accurate mass measurements were performed after calibrating the instrument using heptacosafluorotributylamine in
EI+, EI−, CI+, CI− and FI+ ionization modes (with no reagent gas in CI+ mode during the calibration).
35 Heptacosafluorotributylamine was used also as a lock mass compound in EI+ ionization mode
(218.9856 u), 2,4,6-tris-(trifluoromethyl)-1,3,5-triazine in CI+ mode (286.0027 u) and chloro- and bromo-pentafluorobenzene in CI− mode (201.9609 u and 245.9104 u, respectively). Instrumental grade isobutane was used as a reagent gas. If the average mass accuracy exceeded 0.001-0.002 u, the instrument was re-calibrated. The estimated resolution of the instrument m/∆m = 3700. Field ionization filaments were obtained from Carbotech (Carbotech, Germany). An Agilent 6890N gas chromatograph was utilized in this work (Agilent Technologies; Palo Alto, CA, USA).
Synthesis of the standards. Dimethylselenosulfenate, MeSeSMe, was prepared in solution by mixing equal volumes of 1000 ppm methanol or pentane solutions of dimethyl trisulfide and dimethyl diselenide in a closed vial. The resulting solution was allowed to equilibrate at room temperature for a few hours and, after dilution with pentane, the obtained mixture was subjected to chromatographic separation.
Diethylselenosulfenate, EtSeSEt, was obtained by the mixing of equal volumes of diethyldisulfide and diselenide. Similarly, ethylmethyldiselenide, MeSeSeEt, was obtained from dimethyldiselenide and diethyldiselenide. Methylselenol, MeSeH, was prepared by adding a small amount of crystalline DTT to the methanol solution of dimethyldiselenide in a closed vial. After a few minutes, the resulting mixture was diluted with pentane and subjected to chromatographic characterization. Similarly, ethylselenol,
EtSeH, was prepared from diethyl diselenide solution.43
The obtained reference compounds were separated on a HP-5 column and characterized by their EI+ mass spectra using a GC/TOF-MS. Boiling points, Tb.p., of the species were estimated from their retention times, tR, using the linear relationship between logtR and Tb.p. (under constant temperature ramp conditions).44 Symmetrical species were used as calibrants for the boiling point estimate of the asymmetrical analogues. Results are summarized in Table 2.3. Boiling point estimate of MeSeSMe and
EtSeSEt gives the values of 135 ± 5 oC and 182 ± 5 oC. This is in close agreement to the experimental values (128-131 oC and 170-175 oC) obtained by Potapov et al.45
36 Table 2.3
Characterization of the synthesized reference standards o (a) (b) Compound Tb.p., C Mass spectra (70 eV, EI+) .+ MeSeH - 96 (M , 100), 93 (M − 3H, 90), 80 (M − CH4, 80) .+ .+ EtSeH - 110(M , 100), 108(M − 2H, 50), 93(CHSe, 20), 82(H2Se , 70) .+ . .+ MeSeSMe 135 142(M , 100), 127(M − CH3, 60), 112(SeS , 15), 93(CHSe, 20) .+ .+ .+ EtSeSMe 162 156(M , 100), 128(M − C2H4, 80), 112(SeS , 30), 80(Se , 10) .+ .+ EtSSeMe 155 156(M , 100), 128(M − C2H4, 80), 112(SeS , 30), 93(CHSe, 15) .+ .+ EtSeSEt 182 170(M , 100), 142(M − C2H4, 55), 114(HSeSH , 90) .+ . .+ EtSeSeMe 178 204(M , 100), 189(M − CH3, 8), 176(M − C2H4, 70), 160(SeSe , 40), 93(CHSe, 30) (a) Calculated from the retention times on the HB-5 capillary column using symmetrical selenides and diselenides as a calibration species. Estimated error: ± 5 oC. (b) Based on 80Se.
Computational details. All calculations were performed with the Gaussian 98 implementation of
Becke’s three-parameter hybrid density functional (B3LYP) theory.46 All geometries for the studied compounds were fully optimized without using symmetry or structural constraints using the Berny optimization and the tight Gaussian 98 convergence criteria.46 The locally dense triple split valence basis set, 6-311+G(2d), was found to give good results (in terms of result consistency and computational cost) and was used for geometry optimization and normal vibrational mode analysis. Vibrational frequency calculations were performed on each of the optimized structures and no imaginary values were found.
The zero-point energies and the corresponding thermal corrections to the enthalpy and Gibbs free energy at 298.15 K were obtained by using the harmonic oscillator and rigid rotor approximations and were subsequently added to the total electronic energies to estimate the enthalpy and Gibbs free energy of the species. Population analysis was calculated using Mulliken and natural bond orbital method 47, both implemented in Gaussian 98. As the population analysis is basis-set sensitive, the results of both methods were compared for various basis sets until consistent results were obtained (with 3-21G and 6-
31G basis sets).
80 78 Isobaric interferences, such as Se and H2 Se were evaluated using isotope pattern superimposition principle in conjunction with the least square optimization principle, where the isotope pattern is reconstructed from its possible contributions using the pattern matching optimization (see Chapter 4).40
37 Alice laughed. “There's no use trying,” she said “one can't believe
impossible things.”
“I daresay you haven't had much practice,” said the Queen.
“When I was your age, I always did it for half- an-hour a day.
Why, sometimes I've believed as many as six impossible things
before breakfast.”
/Lewis Carroll, Through the Looking Glass/
CHAPTER 3 | POLYCHALCOGENIDES AND THEIR MASS SPECTRA
38 3.1. ABSTRACT
Various selenium and sulfur-containing volatiles have been detected and characterized as products of chalcogen exchange reactions. Formation of all possible selenium and sulfur containing trichalcogenide isomers (–SeSS–, –SSeS–, –SeSeS–, –SeSSe– and –SeSeSe–) was observed at room temperature in the solutions containing diselenide and trisulfide. Due to the low activation barrier of the selenium exchange reaction, these species are expected also to form in biological systems in the presence of diselenides and higher sulfides. Methyl and ethyl derivatives of these species were characterized using gas chromatography coupled to time-of- flight mass spectrometry with electron impact, chemical and field ionization. The high reactivity of Se–Se and Se–S bonds is also reflected in the mass spectral behavior of these species. Several mass spectrometric observations at first sight seem to be artifacts and only upon closer inspection reveal the nature of their cross-interactions. Rearrangements of ethyl triselenides lead also to the formation of isomeric branched-structure selanadiselenides, which were distinguished from the linear chain triselenides due to the presence of the rare fragmentation pathway: [M – CH2] loss from the ethyl group.
39 3.2. CHALCOGENIDE EXCHANGE REACTION
Diselenide/trisulfide interactions are of importance in biological systems and in this regard, it is useful to examine the possible interaction pathways of these species. In this study we report the formation of all selenium- and sulfur-containing trichalcogenides and several higher polychalcogenides from the Se/S exchange reaction at room temperature. Rearrangements of these volatiles may lead to the formation of isomeric branched structures. Because of very similar structural properties, the behavior of these novel species was studied under various ionization modes (EI+, CI+, EI– and CI–).
The dichalcogenide exchange reaction45 (Scheme 3.1), whose mechanism has not yet been fully elucidated, is probably one of the most intriguing riddles in Se chemistry. This reaction occurs through the rupture of the X−X bond as the RX moiety is conserved. For linear alkyl chains containing dichalcogenides the enthalpy change is zero within experimental error and, therefore, this is an entropy-driven reaction and the equilibrium constant does not vary with temperature.45,48 Such exchange reactions are also reported to occur also with biologically active bis(alkylthio)selenides.48,49
Z Z X Y X Y XY XY
X X X Y X Y X Y X Y XY XY Z Y Z Y
X, Y = S, Se, Te X, Y, Z = S, Se, Te (?)
Scheme 3.1 | General view of the chalcogenide exchange reaction. White and black dots represent non-hydrogen substitution.
In the case of linear chain alkyl groups, the equilibrium constant of dichalcogenide exchange is governed solely by the statistical factors and is close to the K = 4.45 Low activation energy
40 barriers and an equilibrium constant close to unity results in accumulation of the exchange reaction products. Thus, pure asymmetric species, such as MeSeSeEt or MeSeSMe, will be in equilibrium with its exchange products (MeSeSeMe + EtSeSeEt and MeSSMe + MeSeSeMe).
This implies that asymmetric dichalcogenides cannot be isolated in a pure form unless the chalcogenide exchange is suppressed. On the other hand, products of the exchange reaction can serve as important biomarkers of diselenides and selenosulfenates.
3.2.i Formation of polychalcogenides
Analogous to the dichalcogenide exchange (Scheme 3.1), trisulfides interact with diselenides through steps of several consecutive exchange reactions as shown in Scheme 3.2.
Scheme 3.2 | Consecutive Se/S exchange reactions between dimethyldiselenide and dimethyltrisulfide.
As the change in enthalpy is zero, this reaction can be considered from a statistical point of view. With respect to this, trichalcogenides can interact with dichalcogenides in two different routes as shown in Scheme 3.1. Products of such interaction processes can be easily detected with GC/ICP-MS (Figure 3.1). The abundance of the species diminishes in each of the consecutive steps and the decreasing volatility also inhibits the detection of reaction products with more than four chalcogen atoms per molecule. Under constant temperature ramp conditions (∆T = const) linear chain dimethyl polychalcogenides possess systematic incremental
41 retention time behavior on a nonpolar polysiloxane DB-1 capillary column. Elution of these species can be summarized in a two-dimensional retention time grid as shown in Figure 3.2.
Existence of multidimensional retention time grids for selenium-substituted dimethyl chalcogenides were first recognized by Hillen and Werner in 1973. However, the detection and characterization of the various isomeric species was out of the question due to the technical limitations at the time.50
1 2
5
6 4 8 7 10 78Se
3 9
1
5 4 34S
2345678 t, min
Figure 3.1 | 34S and 78Se GC/ICP-MS chromatograms of S/Se exchange reaction products identified from dimethyldiselenide
(MeSeSeMe) and dimethyltrisulfide (MeSSSMe).
(1) MeSeSMe, (2) MeSeSeMe, (3) MeSSSMe, (4) MeSSeSMe, (5) MeSSSeMe, (6) MeSeSeSMe, (7) MeSeSSeMe, (8)
MeSeSeSeMe, (9) MeSSSSMe impurity from MeSSSMe, (10) MeSeSSSMe.
Increments in retention times due to sulfur replacement with selenium also suggest that the –
SeS– moiety can be considered as an average between the corresponding diselenides and disulfides. In other words, the behavior of MeSeSMe can be considered as an average between
MeSSMe and MeSeSeMe. This is also true not only for the boiling points or molecular weights
42 of these species, but also for ionization energies, bond lengths and bond dissociation energies.
An initial estimate of various properties of Se–S species can be generated by averaging those of the corresponding Se–Se and S–S species.
Figure 3.2 | 2D GC retention time grid of the linear chain dimethyl polychalcogenides, CH3SnSemCH3 (0 ≤ n ≤ 5 and 0 ≤ m ≤ 3).
Due to the very low abundance, MeSe2S2Me shown in Figure 3.2 was only identified from the molecular ion and retention time. The actual structure, however, was not elucidated. For higher sulfur- and selenium-containing chalcogenides several structural isomers are possible as shown in Scheme 3.3. Both MeSeS2Me isomers, MeSeSSMe and MeSSeSMe, were detected from the mixture of MeSeSeMe and MeSSSMe. MeSeSSMe is always present at the highest levels as it is formed directly from the interaction of the main diselenide and trisulfide, while the formation of MeSSeSMe requires triselenide, which is a minor component (Scheme 3.3).
–SeSe– –SeSe– –SSSe– –SeSSe– –SSS– –SeSeSe– ? –SeSe– –SSeS– –SeSeS– –SeSe– –SSS– –SSS–
Scheme 3.3 | Formation of all the Se/S trichalcogenides in MeSeSeMe/MeSSSMe system.
43 Thus, chalcogen exchange reaction between dimethyl diselenide and dimethyl trisulfide leads to the formation of all the six Se/S trichalcogenides plus several higher polychalcogenides providing an attractive way to obtain an all-in-one standard mixture of these otherwise commercially unavailable species. Se/S exchange between trisulfides and diselenides can also be used as an alternative approach to bis(thio)selenides over the Painter reaction48,51 or selenium transfer reagents52. To date the Se/S exchange reaction has been successfully used for preparation of otherwise commercially unavailable dichalcogenides43,53, however, it can be easily extended also to higher chalcogenides.
Biogenic bis(thio)selenides are of special importance as they are formed in vivo from selenite
2− 48 (SeO3 ) with thiols (Painter reaction). Volatile bis(methylthio)selenide (4) has been also found in natural elephant garlic and onion oil and its EI+ mass spectra is in agreement with our results.53
Isomeric polychalcogenides 4, 5 and 6, 7 have very similar retention behavior on a non- polar GC column as expected due to their structural similarities. Formation of selenosulfenates
(–SeS–) and more complex species, such as selanyldisulfides (–SeSS–) or selanediylsulfides (–
SeSSe–) can be easily detected by simultaneous monitoring of the S and Se isotopes in GC/ICP-
MS. Also, the elemental ratios in the eluting species are accessible by recording the isotope ratio chromatogram with respect to the reference compound for which the S: Se ratio is known
(such as MeSeSMe in Figure 3.3).
44
Figure 3.3 | 78Se and 34S isotope ratio GC/ICP-MS chromatogram of the MeSSSMe and EtSeSeEt interaction products.
(1) MeSeSMe (Se/S = 1: 1); (2) EtSeSMe (1: 1); (3) EtSeSEt (1: 1); (4) EtSeSSMe (1: 2); (5) EtSeSSeEt (2: 1)
Reductive cleavage of the Se−Se bond in diselenides or selenosulfenates occurs easily in the presence of 1,4-dithio-2,3-butanediol (DTT) and addition of DTT is widely used to elucidate whether species of interest contain Se−Se or Se−S bond (in such case the chromatographic peak will disappear after the addition of DTT).27 Triselenides, on the other hand, are not highly susceptible to Se−Se bond cleavage in the presence of DTT. Symmetrical trichalcogenides such as MeSSeSMe and MeSeSSeMe are reduced in the presence of DTT in contrast to their
45 asymmetric isomers MeSeSSMe and MeSSeSeMe (see Figure 3.4). This finding, however, requires more research regarding the structural effects on reduction efficiency of higher selenium-containing chalcogenides.
1 2
MeSeH
5 7
6 8 4
+ DTT 1234567 t, min
Figure 3.4 | Addition of DTT to the MeSSSMe/MeSeSeMe system: MeSeSeSeMe (8) and the two asymmetrical trichalcogenides
MeSSSeMe (5) and MeSeSeSMe (6) are not reduced (80Se GC/ICP-MS chromatogram).
3.3. MASS SPECTRA OF TRICHALCOGENIDES
Mass spectral behavior of heteroatom-containing trichalcogenides in fact is a challenge due to the variety of rearrangements under EI+ ionization conditions. Selenium-containing trichalcogenides very easily undergo the Se/S exchange reaction, thus asymmetric species always will be contaminated with the exchange reaction products, which can also appear at masses higher than that of the molecular ion. As a consequence to this, in the CI+ ionization spectra of all the CH3Se-containing trichalcogenides (MeSeSeSeMe, MeSeSSeMe,
MeSeSeSMe and MeSeSSMe) one can easily see abundant signals at m/z = 190, corresponding to the dimethyl diselenide formed during the self-exchange reaction of these species (Figure
3.5). CH3S- and CH3Se- containing trichalcogenides (MeSeSSMe and MeSeSeSMe) show the
46 presence of dimethyl selenosulfenate (at m/z = 142) in their CI+ mass spectra. CH3S-containing species show a dimethyldisulfide signal.
200 7 150 5 8
100 MeSeSeMe.+ m/z = 189.88 ± 0.04 u 50 6
0 2000 5 1500
.+ 1000 MeSeSMe m/z = 141.94 ± 0.04 u 6 500
0 2000 9 1500 3
1000 MeSSMe.+ 500 4 5 m/z = 93.99 ± 0.04 u
0 200000 3 5 9 150000
100000 4 6 50000 7 8
8 9 10 11 12 13 14 t, min
Figure 3.5 | Selected ion monitoring in EI+ ionization GC/TOF-MS detection showing the structure specific signals in chromatogram of the S/Se exchange reaction products identified from MeSeSeMe and MeSSSMe.
The phenomena of Se-containing chalcogenide self-exchange should also be considered
13 when analyzing asymmetric isotopically labeled species. In other words, pure CH3-SeSe-CH3,
13 for example, is expected to equilibrate in a 1:2:1 mixture of CH3-SeSe-CH3, CH3-SeSe-CH3
13 13 and CH3-SeSe- CH3.
47 Besides the self-exchange reaction, evidenced in CI+ spectra, trichalcogenides also show structure specific adduct formation as summarized in Scheme 3.4.
[M+XR]+ R-XY-R.+ [M].+ + [M+YR]+
[M+XR]+ [M+YXR]+ R-XYZ-R.+ [M].+ + + [M+ZR]+ [M+YZR]+
Scheme 3.4 | Adduct formation in CI+ mass spectra of Se/S chalcogenides (X, Y, Z = S, Se, Te(?)).
Similar adduct formation is also recently reported for diselenides.54 This is valuable in structural elucidation. For example, both isomers of MeSe2SMe could be easily identified solely from the adduct information in addition to the molecular mass information provided from CI+
+ + mass spectra. While MeSeSSeMe shows only [M+SeCH3] and [M+SSeCH3] , its isomer
+ + + MeSeSeSMe shows abundant adducts [M+SeCH3] , [M+SeSeCH3] and [M+SeSCH3] (Figure
3.6). Adduct formation in the CI+ mass spectra of di-, tri- and higher selenium-containing chalcogenides is an attractive approach of species identification at trace levels. With regard to this, CI+ offers lower detection levels compared to EI+ (due to the ionization chamber geometry and no fragmentation-based loss of the signal), also providing structural information.
48
Figure 3.6 | CI+ ionization mass spectra of S/Se dimethyl trichalcogenides showing the adduct formation.
Symmetrical MeSSeSMe can be distinguished from its asymmetric isomer MeSSSeMe also from their negative chemical ionization mass spectra. Similar to dichalcogenides54, the presence of the negative charge (in CI– or EI– ionization) greatly reduces the energy of dichalcogen bond thus making the molecular ion extremely fragile. In CI– ionization, symmetrical MeSSeSMe
49 leads only to the MeSSe– fragment whereas MeSSSeMe gives MeSe–, MeSeS– and MeSS– fragments and almost no molecular ion, in contrast to MeSSeSMe (Figure 3.7).
Figure 3.7 | CI– ionization mass spectra of isomers MeSSeSMe and MeSSSeMe.
3.3.i Triselenides
Several studies have shown that diselenides take part in disproportionation equilibria
2RSeSeR = RSeR + RSeSeSeR (3.1)
Such a reaction is reported for R = −Cl55, −Br55 and −C≡N56 with an equilibrium constant of K ≈
1×10−4. Similar to these studies, dimethyl- (8), diethyl- (12) and ethylmethyl triselenides (11) were detected in the mixtures of diselenides after prolonged equilibrating time (several months).
Relative amounts of these three triselenides were in agreement with the K ≈ 1×10−4. These species can be formed similar to dichloro- and dibromo- diselenides55 as shown in Scheme 3.5.
50 CH CH3 3 Se Se H3C H C Se 3 Se CH3 CH3 Se Se Se Se CH CH 3 3
Scheme 3.5 | Suggested mechanism of triselenide formation from diselenides (adapted from Milne et al.55).
Asymmetrical ethylmethyl triselenide is presumably formed from dimethyl- and diethyl diselenides according to this scheme or in an exchange reaction between dimethyl- and diethyl triselenides. All three triselenides 8, 11, 12 were characterized by electron impact and chemical ionization in both positive and negative modes.
Statistically, the most abundant ion in the triselenide ion cluster is 78Se80Se80Se and not
80Se80Se80Se as might be expected, in analogy to diselenides or selenides (see Figure 3.8).
78Se78Se82Se 76 80 82 74Se80Se82Se Se Se Se 76Se78Se82Se 78Se80Se80Se Se3 76Se80Se80Se 78Se82Se 78Se78Se80Se 76Se82Se82Se 76Se82Se 80Se80Se 78Se80Se82Se Se2 78Se80Se 74Se80Se80Se 77Se77Se80Se 78Se78Se78Se 74Se82Se 80Se80Se80Se 76Se78Se80Se 78Se78Se 80 82 Se Se 78Se82Se82Se 76Se80Se 80Se80Se82Se
80 82 82 82Se82Se Se Se Se
152 154 156 158 160 162 164 230 232 234 236 238 240 242 244 246 m/z
Figure 3.8 | Formation of polyselenide isotope patterns: 80Se80Se80Se is not the most abundant ion in triselenide clusters.
Ignoring this issue may erroneously lead to incorrect molecular formula assignment.
Dimethyl triselenide (C2H6Se3) mass spectra, for example, have been interpreted erroneously as
39 cyclic 1,2,4-triselenolane (C2H4Se3) as illustrated in Scheme 3.6.
51 Se Se ×6 ×12 ×8 C H 80Se80Se80Se 2 4 Se 78 78 80 78 80 80 Se Se 78 80 80 C2H6 Se Se Se Se 80 80 80 Se3
Scheme 3.6 | Mass assignment of molecular ions in triselenides.
It is interesting to note, that Figure 3.8 also explains the appearance of 78Se-80Se-82Se isotope pattern if the most abundant 160 u or 238 u ions (or their equivalents) are subjected to tandem mass spectrometry.
3.3.ii Branched trichalcogenides
When analyzing the chromatograms of triselenides using post-acquisition selected accurate mass monitoring, minor triselenide satellite peaks were found corresponding to the molecular formula of C3H8Se3 and C4H10Se3 (see Figure 3.9). The validity of this molecular formula was also confirmed by the correct isotope pattern distribution in the eluting satellite species (as calculated by polynomial expansion method57).
52 8 C H Se 2 6 3 14000 11
C H Se 3 8 3 12 12000 C H Se 4 10 3
10000 13 14 Total ionTotal counts C H Se 3 8 3 C H Se 4 10 3
8000
12 13 14 15 16 17 18 t, min
8 R1 = CH3 13 R1 = C2H5 EtR2 EtR2 R2 = CH3 R2 = CH3 Se Se Se Se 11 R1 = CH3 14 R1 = C2H5 EtR1 Se EtR Se R2 = C2H5 1 R2 = C2H5
12 R1 = C2H5 R = C H 2 2 5
Figure 3.9 | GC/TOF-MS CI+ total ion mass chromatogram showing the presence of isomeric triselenides 13 and 14.
Mass spectra of these isomeric species (13, 14) show several differences between their corresponding linear chain triselenides (11, 12). Presence of Se=Se.+ ion (center at m/z =
159.8332 u) in EI+ conditions is not characteristic for the satellite species, while Se=Se.+ cluster is one of the most abundant in the EI+ spectra of triselenides (Figure 3.10). The extrusion of selenium atom is apparent in the mass spectra of triselenides via the formation of the corresponding diselenide, while this feature is not observed for the isomeric species. While the major clusters in EI+ mass spectra of linear triselenides are characterized by the consecutive loses of H2C=CH2 and extrusion of the central selenium, ethylated selanadiselenides also show
. . . major rearrangement leading to the loss of CH2 and CH2Se (in EI+) and CH2SeH (in CI+) as
. shown in Figure 3.10. Although the loss of CH2 from ethyl group as observed in EI+ ionization is rare in mass spectrometry58, such a behavior is consistent with the 4-member ring rearrangement in selanadiselenides as shown in Scheme 3.6. In EI+ conditions, this terminal
53 CH3- group rearrangement leads to the loss of CH2 and Se which is reflected in two clusters [M
+ + + – CH2] and [M – CH2Se] . CH2=SeR species are also formed (m/z = 122.97 u for R = Et and m/z = 108.96 u for R = Me) and they are not observed for linear chain species. In CI+ conditions, however, loss of protonated [CH2Se] moiety is evident leading to the abundant
+ cluster [M – CH2SeH] (Scheme 3.7 and Figure 3.10).
Scheme 3.7 | Proposed 4-member ring rearrangement of ethylmethyl- (13, R = CH3) and diethyl- selanadiselenide (14, R = C2H5) in CI+ and EI+ ionization.
The apparent instability of the positively charged molecular ion of selenadiselenides is also supported by EI+ mass spectra where its abundance is less than 20% (for linear chain triselenides M.+ is one of the most abundant ions).
54
Figure 3.10 | CI+ and EI+ mass spectra of diethyl-selanadiselenide (14) and diethyl-triselenide (12).
Based on all these properties, it was concluded that the isomeric species represent a new class of branched-structure triselenides – selenadiselenides (Figure 3.9, compounds 13 and 14).
The formation mechanism of these species was not investigated, however, due to their low abundance it may be that linear chain triselenides are their precursors. The search for neutral and ionic branched chalcogenides has been widely discussed in the past59,60. Branched structure
61 isomers (thiosulfoxides, R2S=S) are known in sulfur chemistry , however, analogous selenides are rare in the literature and require further research. To our knowledge only dichlorodiselenide62 and di(β-naphtyl) diselenide63 are reported as possibly having a branched structure.
55 In conclusion, it has been shown that diselenides and triselenides easily participate in intermolecular chalcogen exchange reactions leading to the formation of all the Se/S trichalcogenides. This feature can be used in identification of polychalcogenides mass spectra and also to obtain the standards of all the trichalcogenides simultaneously. The high reactivity of Se–Se and Se–S bonds in trichalcogenides is also reflected in the mass spectral behavior of these species.
3.4. EXPERIMENTAL
Reagents and Standards. All reagents were of analytical grade and were used without any further purification. Dimethylselenide, dimethylsulfide and dimethyldisulfide were purchased from Fluka
(Milwaukee, WI, USA). Dimethyldiselenide, dimethyltrisulfide, diethyldisulfide and dithiothreitol were purchased from Sigma-Aldrich (Milwaukee, WI, USA). Diethyldiselenide was purchased from Strem
Chemicals (Newburyport, MA, USA). Stock solutions of 1000 ppm were prepared by dilution of 2.5 µL of compound with 2500 µL HPLC grade methanol (Fisher Scientific; Fair Lawn, NJ, USA).
Instrumentation. An Agilent 6890 (Agilent Technologies; Palo Alto, CA, USA) gas chromatograph was utilized in this work and coupled to a Micromass GCTTM orthogonal time-of-flight mass spectrometer (Micromass, Manchester, UK) and Agilent Technologies 7500c inductively coupled plasma mass spectrometer (Agilent Technologies, Tokyo, Japan).
In TOF-MS a single fragment of the reference compound is used as an internal reference signal (lock mass). Accurate mass measurements were performed using heptacosafluorotributylamine as a mass calibration compound in EI+, EI−, CI+ and CI− ionization modes (with no reagent gas during the calibration in CI+ mode). Heptacosafluorotributylamine was used as a lock mass compound in EI+ ionization mode (218.9856 u), 2,4,6-tris-(trifluoromethyl)-1,3,5-triazine in CI+ mode (286.0027 u) and chloro- and bromopentafluorobenzene in CI− mode (201.9609 u and 245.9104 u, respectively).
Instrumental grade isobutane was used as a reagent gas. If the average mass accuracy exceeded 0.001-
56 0.002 u the instrument was re-calibrated. The estimated resolution of the instrument m/∆m = 3700 in the mass range of 100-200 u. ICP-MS was used with an hydrogen octapole reaction system and the instrument was operated at standard operating conditions described elsewhere.43
GC conditions. A splitless injection mode was used and the temperature of the injector was programmed at 220 oC. The column oven was initiated at a temperature of 75 oC and immediately ramped at 10 oC min−1 to a temperature of 220 oC. Helium was used as the carrier gas and the column flow was set at a constant flow of 1.5 mL min−1. An HP-5 (5% phenyl, 95% methyl−polysiloxane) capillary column (30 m, 0.25 mm i.d., 0.25 µm film thickness) was used for separation.
Procedures. All the oligochalcogenides were obtained in solution by mixing equal volumes of methanol solutions of dimethyltrisulfide (1000 ppm) and dimethyl- or diethyldiselenide (1000 ppm) in a closed vial. The resulting solution was allowed to equilibrate at room temperature for a few hours and, after dilution with pentane, the obtained mixture was subjected to chromatographic separation.
Formation of triselenides and selana-diselenides appears to be much slower process and these species were detected only few months after the mixing of the dimethyl trisulfide and methyl- or ethyl diselenide. No cross-interaction products were obtained if the mixture is subjected to GC analysis immediately after the mixing of the reagents. Effect of temperature and acid- or base- catalysis was not studied.
57
Fragment of the Cosmati floor mosaic at the Anagni Cathedral (Italy, 1226 AD)
/Anal. Bioanal. Chem. 2005, 381(5) cover image/
CHAPTER 4 | THEORY OF ISOTOPE PATTERN RECONSTRUCTION
58 4.1. ABSTRACT
The concept of isobar deconvolution using the mass domain and the signal intensity based domains is described. The intensity domain-based approach employs the reconstruction of the observed isotope pattern from the isolated patterns of the isobaric species. The quantitative information is adjusted with the use of least squares algorithm. The mass domain-based approach, however, employs signal deconvolution by forming Gaussian components for which the peak width and position can be predicted a priori. The latter method is applicable to medium resolution instruments, such as TOF-MS while the pattern reconstruction approach is applicable also to low resolution instrumentation, such as quadrupole-based ICP-MS or
GC/MS. An example of CHnSe cluster analysis in dimethyl diselenide mass spectra is given to illustrate the concepts underlying both approaches.
59 4.2. INTRODUCTION
Very recently Roussis and Proulx clearly demonstrated the need of mathematical data treatment when interpreting mass spectra.64 The presence of isobaric interferences is a common issue in mass spectrometry. In plasma mass spectrometry (ICP-MS), this potentially leads to false identification of the elements and consequently, affects quantitation of a particular element.
Instrumental approaches through either high resolution mass spectrometers or collision/reaction cell quadrupole instruments are often used to minimize or eliminate polyatomic interferences.65
Analyte m/z signals in ICP-MS can be interfered with by isobaric and polyatomic species.
These problems may be overcome in several ways.3 1) Time-resolved separation techniques can be used. For example, chromatographic chloride separation from As-containing species eliminates the formation of 40Ar35Cl+ interfering with 75As+ detection. 2) Hydride generation can be used to eliminate matrix related interferences for elements such as As, Se, Pb and Sn. 3)
Use of mixed-gas plasma is effective in removal of matrix-based interferences, such as carbon loading. Addition of O2 to the Ar carrier gas eliminates the carbon deposit on the extraction cones (sampler and skimmer cones) and leads to better analytical performances. Additionally, mixed-gas plasmas (He-Ar, N2-Ar etc.) usually allow lower detection levels of problematic elements by at least to one order of magnitude. 4) Increase of resolution with sector field ICP-
MS is a powerful tool for elemental analysis in a number of complicated matrices. 5)
Implementation of a collision or reaction cell system. This technology offers an efficient way of alleviating isobaric interferences in ICP-MS through selective ion-molecule reactions. This allows one to attenuate spectral interferences even at low mass resolving powers of MS.
Interference removal can be accomplished by collision induced dissociation, which occurs on
+ some matrix polyatomic species (such as ArO ), or through the charge transfer reaction of H2 or
60 other reacting gas. Besides these two modes of interference removal, kinetic energy discrimination is possible with the collision/reaction cell.
The general mathematical strategy in practice to correct for isobaric interferences in atomic mass spectrometry is the subtraction of the contribution of interfering isotopes from the measurements of non-interfering isotopes. For example, the 40Ar35Cl+ contribution on the m/z =
75 signal (determination of 75As+) can be estimated by monitoring 40Ar37Cl+ at m/z = 77. From the signal at m/z = 77, one can then estimate the abundance of 40Ar35Cl+ by knowledge of the natural chlorine isotope ratio. The idea behind this approach is very simple; however, it is applicable only for a single polyatomic ion interference.
In molecular mass spectrometry, isobaric interferences lead to a shift in the observed peak mass centroid. In isotope dilution GC/MS, for example, the presence of isobaric interferences affects the apparent isotope ratios and thus they have to be accounted for to achieve accurate results.66 It is therefore useful to understand the main aspects of isobar deconvolution as a tool to enhance the validity of the data obtained.
In this study we utilize several approaches to isobaric interference estimation, as recently outlined by Roussis and Proulx64. In particular, these approaches were demonstrated to be useful in interpretation of diselenide and selenosulfenate mass spectra54 and herein is given a more technical report of the various isobar deconvolution approaches. These are isotope cluster reconstruction using isotope pattern superimposition combined with the least square optimization algorithm, signal deconvolution to its components in the mass domain, and mass shift analysis of the overlapping clusters.
61 4.3. ISOTOPE PATTERN RECONSTRUCTION
The presence of isobaric interferences can be easily detected by the distortion of the isotope pattern. One can use the advantage of an isotopic pattern to estimate the amount of overlapping species. Reconstruction of the observed isotope pattern is achieved by the superimposition of the separate isotope profiles of the overlapping species in conjunction with least square optimization of the pattern intensities.
Assuming the mass spectral pattern of interest (Y) can have a contribution (η) from n
64 components, then according to the superposition principle , signal abundance (Yi) at a certain mass (j) is a linear combination of the isotope patterns of the isobaric components (I):
m YIiiji∝⋅∑η (4.1). j=1
Isotope pattern deconvolution involves selection of contributing species and sequentially finding the relative contributions (η) for each of the components. Isotopic profiles of the individual component can be easily calculated by the classical polynomial expansion approach.57 According to this, compound isotope patterns (as observed in mass spectra) form as a combinatorial convolution product of element isotope patterns. For example, isotope pattern of methane is obtained from the following polynomial (Eq. 2):
12 12 13 13 1 1 1 2 2 4 P(CH4) = [x( C)·{ C} + x( C)·{ C}] ·[x( H)·{ H} + x( H)·{ H}] (4.2).
The only input in this equation is the structure and the isotopic composition of elements. {12C},
{13C}, {1H} and {2H} are simply symbolic identifiers for isotopic species and can be considered a book-keeping device. Expansion of this polynomial with substitutions a = x(12C), b = x(13C), c
= x(1H), d = x(2H) yields the following expression:
4 3 2 2 3 4 4 3 2 2 3 4 P(CH4) = ac + 4ac d + 6ac d + 4acd + ad + bc + 4bc d + 6bc d + 4bcd + bd .
62 4 12 1 3 Here ac corresponds to the methane ( C)( H)4, which has a mass of 16 u. The term ac d,
12 1 2 however, belongs to the ( C)( H)3( H) with the mass 17 u, etc. The coefficients before these terms describe the statistical abundance of a particular isotopic combination for a molecule. The polynomial expansion can be also performed using fast Fourier transformation methods. 67
Although the classical polynomial expansion method is slow, it can be used in a form of cellular automata, which easily generates complex isotopic patterns in an educationally friendly manner (Figure 4.1). Such construction of complex isotope patterns captures the very essence of a problem – complex isotope patterns can be generated using very simple rules.
Figure 4.1 | Cellular automata evolution of complex isotope patterns from elemental isotope patterns.
Isotope pattern abundances for each of the n individual components can be represented as an m×n matrix F (usually non-symmetric) where m is the size of the pattern. Iji represents the abundance of mass j of the i-th component (each of the i-th components in the F matrix has the same mass):
63 ⎛⎞I11II 21... n 1 ⎜⎟ II... I F = ⎜⎟12 22n 2 (4.3) ⎜⎟...... ⎜⎟ ⎝⎠II12mm... I nm
The matrix of the experimental abundances for the cluster of interest is an m×1 vector Y and solution of the least squares is an n×1 vector a, from which the relative contributions of the individual components (ηi) then can be calculated:
−1 aFFFY= ()TT (4.4)
ai ηi = (4.5) ∑ ai i
Graphically, this problem can be depicted as the following:
Error estimate of ηi can be obtained from the diagonal elements of the inverse information
T −1 matrix (F F) via the error estimate of ai:
−1 22T ssFF=⋅r (4.6) ai ()ii
Here sr is the residual standard deviation, obtained from the residual matrix (Y − Fa).
22 −−11n sη 111T ⎛⎞⎛⎞TT i =−−⋅()()YFaYFa⎜⎟⎜⎟ FF + FF (4.7) ()ii∑ () kk ηiikmaa−1 ⎝⎠⎝⎠k =1
64 One of the characteristics of this approach is that the relative uncertainty of ηi is very similar across the values of i = 1…n. This approach gives better ηi estimate for elements with rich isotopic patterns (such as Te, Sn, Cd or Hg where each of the elements has more than 6 major isotopes). Eventually, higher numbers of included fitting components need to be justified with the amount of input information (via the size of the isotope pattern). In the case of selenium, which has 5 major isotopes, five species deconvolution of the isotopic pattern is readily accessible with a precision of less than 1%.
4.4. SIGNAL DECONVOLUTION IN THE MASS DOMAIN
4.4.i Signal peak shape analysis
Because of the different mass defects of the elements, isobaric interferences will differ slightly in their mass. This discrepancy will result in a broadening of the signal in the mass domain. For time-of-flight detection, the observed signal peak width is linearly proportional to the mass and the mass-peak width relationship can be easily obtained using a calibration standard, such as heptacosafluorotributylamine. Broadening is more apparent at the base of the peaks so peak width at 20–30% height (instead of 50%) can be efficiently used for broadening characterization, however higher precision is attained at the 50% level. Accordingly, the reference peak width is compared with the signals of interest and any significant signal width difference indicates the presence of isobaric interferences.
After the presence of an isobaric interference is established, its magnitude can be estimated by signal deconvolution using the Gaussian function (or any other more appropriate signal peak function, such as Lorentzian). Under normal operating conditions in time-of-flight mass spectrometry, ions will be normally distributed with respect to their arrival time (represented as
65 mass) and the peak width of two isobar signals can be safely assumed to be the same. Note that this procedure is not merely an empirical fitting. All the parameters of the Gaussian components have a relationship to the physical properties such as the resolution (peak width), exact mass of the component (peak position) and normalized area (relative amount of the interfering species).
4.4.ii Peak centroid mass analysis
Peak centroid mass, which is often used to represent the mass spectrometric data, is affected in the presence of isobaric interferences. To illustrate this, assume there are two interfering species
A and B with the relative abundance of x and (1 – x) respectively. The analytical signal as a function of mass can be represented with the following expression:
22 ⎛⎞mmii−−AB ⎛⎞ mm xx−−22⎜⎟1− ⎜⎟ II=+∝(A) I (B) e⎝⎠∆∆mm + e ⎝⎠ (4.8) ii∆∆mm
The maximum of the observed resulting signal profile (mmax) depends on the relative contributions of isobars A and B. Such relationships can be computed from equation 4.8. For example, when the peak width is larger than the mass difference of the isobars, ∆m ≥ |mA – mB|, it can be shown that the observed peak centroid mass is the weighted average of the isobar masses:
mxmxmmax≅⋅ A A +⋅ B B (4.9)
According to this, in the presence of isobaric interferences, the accurate mass measurements will be biased if uncorrected peak centroid mass is used. Moreover, care should be taken on the actual centroid algorithm itself. If the average mass is used (or 80% top average etc.), it no longer exactly represents the maximum mass, mmax, and additional mass bias thus can be introduced (see Figure 4.3b).
66 4.5. Application to dimethyl diselenide mass spectra
It is known that the electron impact fragmentation of dimethyl diselenide does not render a pure cluster of CH3Se as the region at m/z = 88…98 clearly shows a distorted isotope pattern of selenium54 (see also Figure 4.2e). The following example illustrates the deconvolution of this isobaric overlap.
4.5.i Deconvolution in the intensity domain
+ The isotope pattern reconstruction algorithm described above was applied to the CHnSe cluster of dimethyl diselenide EI+ mass spectrum. The contribution of the species with n = 0-4 was allowed in the pattern reconstruction and the F matrix therefore is constructed as follows:
CH0Se CH1Se CH2Se CH3Se CH4Se mz/88=−−−−⎛⎞ 8.9 ⎜⎟ 89⎜⎟ 7.6 8.9 − −− 90⎜⎟ 23.4 7.6 8.9 − − ⎜⎟ 91⎜⎟−− 23.4 7.6 8.9 92⎜⎟ 49.2− 23.4 7.6 8.9 ⎜⎟ 93F =⎜⎟−− 49.2 23.4 7.6 94⎜⎟ 9.1−− 49.2 23.4 ⎜⎟ 95⎜⎟−−− 9.1 49.2 ⎜⎟ 96⎜⎟−− 9.1 − 49.2 97⎜⎟−−− 9.1 − ⎜⎟ 98⎝⎠−−−− 9.1
From here the least squares optimized isobar abundance contribution matrix (ηi) was calculated according to the equations (4)-(7). The results were as follows (with the standard
+ + deviation given in the parenthesis): η0 = 2.0(0.8)% (CH0Se ); η1 = 49.8(0.6)% (CH1Se ); η2 =
+ + + 19.7(0.6)% (CH2Se ); η3 = 22.7(0.6)% (CH3Se ) and η4 = 5.9(0.6)% (CH4Se ), which is
+ + + consistent with the 2: 1: 1 contribution from CHSe , CH2Se and CH3Se , and the result obtained is in good agreement with the experimentally observed isotope pattern (Figure 4.2).
67
+ + + Figure 4.2 | Calculated isotope patterns of the CHSe (a), CH2Se (b), and CH3Se (c) clusters, and the result of the 2:1:1
+ superimposition of these species (d) along with the experimentally observed CHzSe cluster in EI+ mass spectra of dimethyl diselenide (e).
68 4.5.ii Deconvolution in the mass domain
A single signal from dimethyl diselenide mass spectrum at m/z = 92.924 u was selected for subsequent analysis. The results of the isotope pattern reconstruction procedure suggest that
+ + there are two contributions at this m/z value, namely CHSe and CH3Se (refer to the F matrix).
Accordingly, the m/z = 92.924 u peak is deconvoluted with two Gaussians. The predicted single species peak width was estimated from the linear regression of mass vs. peak width of the heptacosafluoro-tributylamine mass spectrum (Figure 4.3a).
Figure 4.3 | (a) Peak width (at half-height) and mass relationship in time-of-flight mass analyzer as obtained from the isobar-free signals in the mass spectra of heptacosafluoro-tributylamine (●). Squared region shows the isobar interference related peak broadening in mass spectra of dimethyl diselenide (○) around m/z = 93. (b) High resolution mass spectra deconvolution. m/z =
92.924 u fragment of the dimethyl diselenide EI+ mass spectra (●) and the reconstructed (fitted) signal peak profile (dash) showing
80 78 the contribution of CH Se and CH3 Se (in the ratio of 4:1). Note that 80% top-of-the-peak centroid mass does not correspond to
80 78 the mass of either of the components, CH Se (1) or CH3 Se (2), nor to the maximum signal.
Results of the Gaussian signal fit agree well with the predicted values given in the
80 + 78 + parenthesis: CH Se and CH3 Se peak width 18 mu (20 mu is the experimentally estimated peak width at the m/z = 93 u from the data shown in Figure 4.3a), peak area ratio 4.1 ± 0.5 (a
80 + 78 + value of 4.21 corresponds to the CHSe/CH3Se ratio of 2: 1), and CH Se and CH3 Se peak
69 separation 20 mu (theoretical value 17 mu). The obtained results clearly show the presence of
80 + 78 + CH Se and CH3 Se in the ratio of about 4: 1 which corresponds to the CHSe/CH3Se ratio of
2: 1 as estimated from the isotope pattern reconstruction approach.
4.5.iii Peak centroid mass shift analysis
In the above example of dimethyl diselenide, the difference between the mass of CH80Se
78 (92.9243 u) and CH3 Se (92.9408 u) is 17 mu and peak width (∆m) at this mass is estimated to be 20 mu. The observed centroid mass is 92.9273 ± 0.0005 u. Since equation (9) holds true in this case ( m ≥ |m 80 – m 78 |), the observed centroid mass is in a simple relationship with ∆ CH Se CH3 Se the isobar abundances:
92.9273 u≅⋅xx80 92.9243 u + 78 ⋅ 92.9408 u . CH Se CH3 Se
80 78 80 78 From here CH Se/CH3 Se = 4.5 ± 0.8. In conclusion, the CH Se/CH3 Se amount ratio in m/z
= 92.924 signal of dimethyl diselenide EI+ mass spectra obtained by centroid mass shift analysis (4.5 ± 0.8) is comparable to that obtained by isotope pattern reconstruction (4.6 ± 0.4) and peak shape deconvolution approaches (4.1 ± 0.5) thus verifying the validity of all methods.
4.6. ASPECTS OF ISOBARIC INTERFERENCES IN MASS SPECTRA
The presence of isobaric interferences is a very important issue in mass spectrometry that has a broad range of negative consequences. One of the most obvious problems is the distortion of the isotope pattern. Slightly over one half of about 90 naturally occurring elements have more than one stable isotope and isotope patterns offer attractive advantages of identifying various elements in mass spectrometry (Figure 4.4).
70
Figure 4.4 | Chemical fingerprints: isotope patterns of elements (modified from Agilent Technologies).
Loss of the isotope fingerprint therefore poses loss of analytical information and is undesirable. Figure 4.5 illustrates the loss of Se2 isotopic pattern in positive chemical ionization of dimethyl diselenide. In heteroatom-tagged speciation, the presence of a characteristic isotope pattern is very useful in locating the element-tagged species. For small molecular weight species isotope patterns are mostly governed by the heteroatom, however for large molecules isotope patterns eventually approach the shape of normal distribution regardless of the elemental composition as a direct consequence of the statistical “law of large numbers”.
71 CI+ EI+
175 180 185 190 195 200 175 180 185 190 195 200 m/z
Figure 4.5 | Distortion of isotopic pattern of dimethyl diselenide molecular ion due to the presence of isobaric interferences in CI+ ionization.
Besides the isotope pattern distortion, centroid mass shift is an important isobar interference consequence, as peak centering is common practice in mass spectrometry data analysis. The number of possible empirical formulas calculated from the data decreases rapidly with increasing mass accuracy. It has been shown that the presence of molecular interferences leads to erroneous mass assignments due to the shifts peak center.68 This leads to large errors in determining a unique correct empirical formula, and therefore the analysis can be highly ambiguous.
72
+
CHAPTER 5 | INTERPRETATION OF BUTYLTIN MASS SPECTRA
FOR ISOTOPE RATIO MEASUREMENTS
73 5.1. ABSTRACT
The fragmentation patterns of butyltin compounds (mono-, di- and tributyltin) in an electron impact ion source were studied using an isotope pattern reconstruction algorithm with emphasis on isotope ratio measurements from molecular clusters. For this purpose, standards of natural tin isotope abundance and a 119Sn-enriched mixture of the three compounds were both ethylated and propylated using sodium tetraalkylborates. The corresponding mass spectra of the various tetraalkyltin compounds prepared were obtained by GC/MS after their extraction with hexane.
The results showed that pure interference-free molecular clusters were obtained only for certain
+ + + R3Sn ions where no isobaric overlap with R2SnH ions occurred (e.g. BuEt2Sn overlaps with
+ Bu2SnH ). These ions are ideal candidates for accurate Sn isotope ratio measurements while isotope pattern perturbing interferences are observed for other molecular fragments down to
Sn.+. An isotope pattern reconstruction algorithm thus can be used as an analytical tool to ensure the absence of molecular interferences – a requirement for accurate isotope ratio measurements from molecular clusters. The relevance of these studies for the determination of butyltin compounds in environmental samples by isotope dilution GC/MS is also discussed.
74 5.2. INTRODUCTION
Since the 1960’s various organotin compounds have been extensively used as broad spectrum biocides and lately the need to measure butyltin species in the environment on a routine basis has increased due to the global ban on the application of tributyltin.69 To meet the legislative criteria, several analytical methods have been developed. Inductively coupled plasma mass spectrometry (ICP-MS) based methods with isotope dilution gas chromatography (ID-GC/ICP-
MS)70,71 or liquid chromatography (ID-HPLC/ICP-MS)71 are ideal for research purposes, however their cost may be high for routine testing laboratories. Isotope dilution GC/MS is regarded as an attractive alternative to this.66 However, electron impact (EI) ionization produces molecular ions in contrast to atomic spectrometric methods, and this complicates the use of ID methods, since elemental isotope ratios need to be measured from molecular clusters. With respect to this, EI ionization creates two challenges. First, any molecular ion signal at a particular m/z value represents not only a single Sn isotope abundance but also includes contributions from other lower mass Sn isotopes due to the presence of 13C and/or 2H in the molecular ions. This precludes any direct use of the measured signal to obtain Sn isotope ratios for isotope dilution analysis. Secondly, a particular molecular ion signal can also have contributions from isobaric molecular interferences as different species could overlap in the mass spectrum. Perhaps the biggest concern in accurate isotope ratio measurements is the loss or gain of protons. It is known that many mass spectral rearrangements involve loss or transfer of hydrogen atoms resulting in various fragment ions differing only in one or two hydrogen
. atoms. For example, ethyl groups are eliminated from diselenides both as C2H5 and neutral
54 CH2=CH2. Such phenomena lead to the distortion of the natural Se isotope pattern due to the isobaric superimposition of several molecular ions. This is easy to recognize for monoisotopic
75 elements (such as As or P) where distinct signals [M ± nH] appear without overlapping with the primary signal of interest [M]. In this case the identity and the amount of such molecular interferences can be readily estimated. For polyisotopic element species (Sn, Hg, Cd etc.), however, [M ± nH] seriously overlaps the signal of interest [M] and thus, visual inspection will not always lead to the identification and quantitation of such interferences.
The first problem, the contribution of 13C or 2H, can be easily resolved, since the natural isotope composition of carbon and hydrogen is known and correction equations can be easily developed.66 The second problem, however, requires the selection of a “pure” ion cluster in the mass spectrum for accurate isotope ratio measurements and to date this issue has not been addressed in a quantitative manner for butyltin compounds. This has only been assured by performing analysis with reference materials. Since the presence of isobaric interferences leads to inaccurate results, one can only indirectly validate the selection of a particular ion cluster for isotope dilution applications using this approach.
There is a clear need to study the fragmentation pathways of butyltin compounds to select adequate molecular ions for Sn isotope ratio measurements by GC/MS. First, the use of heteroatom-labeled compounds, such as 119Sn-enriched butyltin compounds70,66 for isotope dilution analysis is increasing due to its great advantages over the more traditional 13C or 2H labeling, since only one atom in the molecule needs isotopic alteration. Second, the isotope dilution equation used for elemental analysis can only be applied if a pure molecular cluster can be selected that avoids isobaric interferences and, third, one needs to assure that the fragmentation pathways of butyltin compounds are not isotope-dependent. It is the aim of this report to introduce a systematic approach for ensuring the absence of molecular interferences in isotope ratio measurements from molecular clusters. Here we apply an isotope pattern
76 reconstruction algorithm40 to study the butyltin mass spectral fragmentation pathways using both standards of natural Sn isotope abundance and mono-, di-, and tributyltin labeled with enriched 119Sn. To clarify differences in fragmentation, those standards were both ethylated and propylated and their electron impact mass spectra were studied regarding accurate tin isotope ratio measurements for isotope dilution applications.
5.3. MASS SPECTRA OF BUTYLTIN COMPOUNDS
The positive EI ionization mass spectra of ethylated and propylated monobutyltin (MBT), dibutyltin (DBT) and tributyltin (TBT) of both natural and 119Sn-enriched isotope abundances are shown in Figures 5.1, 5.2 and 5.3 respectively. All ions in the electron impact source contained only one tin atom, which is evident from the presence of its isotope pattern in the mass spectral clusters.
+ Sn + Sn
100 120 140 160 180 200 220 240 260 280 300 320 100 120 140 160 180 200 220 240 260 280 300 320 m/z m/z Sn + Sn + CH2CH2 CH2CH2CH2 + H CH2CH2 H + + + + + + .+ +
+ + + + + + + + + + .+ .+
+ 100 120 140 160 180 200 220 240 260 280 300 320 100 120 140 160 180 200 220 240 260 280 300 320
Figure 5.1 | EI mass spectra of natural (top) and 119Sn enriched (bottom) ethylated (left) and propylated (right) monobutyltin among with the modular representation of the fragment ion identity.
77 + + Sn Sn
100 120 140 160 180 200 220 240 260 280 300 320 100 120 140 160 180 200 220 240 260 280 300 320 m/z m/z
Sn + Sn CH2CH2 CH CH CH + 2 2 2 H CH CH + 2 2 + + + + H
+ + + + + +
+ + + + + + + + + + .+ .+ +
100 120 140 160 180 200 220 240 260 280 300 320 100 120 140 160 180 200 220 240 260 280 300 320
Figure 5.2 | EI mass spectra of natural (top) and 119Sn enriched (bottom) ethylated (left) and propylated (right) dibutyltin among with the modular representation of the fragment ion identity.
+ Sn
+ Sn
100 120 140 160 180 200 220 240 260 280 300 320 100 120 140 160 180 200 220 240 260 280 300 320 m/z m/z Sn Sn CH2CH2CH2 CH2CH2 + CH CH + H 2 2 + + H + + + + + + +
+ + + + + + + .+ .+ + + + +
100 120 140 160 180 200 220 240 260 280 300 320 100 120 140 160 180 200 220 240 260 280 300 320
Figure 5.3 | EI mass spectra of natural (top) and 119Sn enriched (bottom) ethylated (left) and propylated (right) tributyltin among with the modular representation of the fragment ion identity.
78 As can be observed, molecular ions of butyltin species are not present in the mass spectra
(their position is indicated by their molecular structure in the figures). In general, mass spectra of butyltin species can be sketched as the loss of one alkyl radical, .R, following with the consecutive neutral losses of alkene fragments, [R − H]. At first, mass spectra of these species resemble a simple set of consecutive alkyl and alkene group losses; however, such a fragmentation pathway would result in the correct Sn isotope pattern across the whole mass spectra. A closer look at Figures 5.1-5.3 shows that the Sn isotope patterns in lower m/z clusters are not identical to those at higher m/z values and this is more evident for the spectra obtained from the 119Sn-enriched compounds. From this it is clear that overlapping molecular clusters are present in the mass spectra of butyltin species. The tentative identifications of all possible molecular clusters, including isobaric overlaps, are also given in Figures 5.1-5.3 using a simple modular representation. This will be discussed in more detail below.
5.3.i Interpretation of mass spectra using isotope pattern reconstruction
The qualitative identification of the molecular clusters presented in the 12 mass spectra
(Figures 5.1-5.3) will be further elaborated by quantifying the relative abundances of the different fragment ions using isotope pattern reconstruction.
In mass spectrometry of polyisotopic elements, the most common form of molecular isobaric interferences are [X ± nH], where X is the fragment species containing the element of interest,
+ e.g. C2H5Sn . One can use the advantage of an isotope pattern reconstruction algorithm to estimate the amount of any isobars present. With this approach one can estimate the relative amount of isobars [X ± nH] as a function of n. The mathematical aspects of this algorithm are outlined elsewhere.40 In short, the observed m/z pattern of the species of interest (X) is reconstructed as a linear combination of molecular formulae differing only in hydrogen, [X ±
79 nH]. The observed isotope pattern then is reconstructed with respect to the relative amount of each of the possible isobars using a standard linear least squares fitting algorithm which yields the estimates of isobar abundances. For example, the tributyltin cluster at m/z = 179 is a combination of primarily two species, C4H9Sn and C4H11Sn. To elucidate their relative abundance it is necessary only to inspect the correlation between the experimentally obtained isotope pattern, Pexp, at m/z = 179 and artificially constructed pattern from C4H9Sn and
C4H11Sn, P = a·[C4H9Sn] + (1 – a)·[C4H11Sn], as a function of species ratio a. From here,
2 maximum correlation r (or minimum residual uncertainty s0) corresponds to the actual value of a. Since the superimposition algorithm is linear, there is only one minimum in the optimization space which leads to the unique deconvolution answer. Therefore, large residual uncertainty of the reconstructed isotope pattern or negative eigenvalues indicates incorrect or an incomplete set of chosen isobars (descriptors).
Consider the tributyltin cluster at m/z = 290 u in the EI mass spectra of ethyltributyltin
(Figure 5.3, ethylation, natural Sn). When the experimentally observed cluster is reconstructed
(fitted) as a linear combination of [C12H25+nSn] species, where n = 0 to n = 4, the following abundances of [X ± nH] species are obtained: 1 ± 4% (C12H25Sn), 0 ± 4% (C12H26Sn), 99 ± 3%
(C12H27Sn), 0 ± 3% (C12H28Sn) and 0 ± 3% (C12H29Sn), thus confirming the presence of a pure
+ C12H27Sn cluster. The results of isotope pattern deconvolution using the isotope pattern reconstruction agree with direct experimental measurements at high mass resolution. As an
+ example, high resolution mass spectrometry experiments showed that SnHn cluster in the EI
72 mass spectra of Et4Sn consists of 26.7% (Sn), 66.2% (SnH) and 7.1% (SnH3). Isotope pattern reconstruction from low resolution EI mass spectra of Et4Sn agree well with these findings: 27
± 1% (Sn), 67 ± 2% (SnH) and 6 ± 1% (SnH3). Also, the SnEt/SnEtH2 ratio in the mass
80 spectrum of Et4Sn was estimated as 1.40 ± 0.05, which agrees well with experimental high-
72 resolution mass spectrometry data (1.44). Similar calculations on Et4Pb showed the agreement between direct experimental measurements and mathematical deconvolution, thus confirming the validity of the mathematical deconvolution approach in general. Each of the m/z cluster in mass spectra of ethylated and propylated MBT, DBT, and TBT was subjected to least squares isotope pattern reconstruction analysis allowing for [X ± nH] where n was varied. Tables 5.1 and 5.2 summarize all the quantitative results obtained showing the molecular formulas of species forming various mass spectra clusters.
Table 5.1
Molecular formula reconstruction from clusters in EI mass spectra of propylated butyltin species
BuSnPr3 (MBT) Bu2SnPr2 (DBT) Bu3SnPr (TBT) Fragment Abundance Abundance Abundance Abundance Abundance Abundance (natural Sn) (119Sn) (natural Sn) (119Sn) (natural Sn) (119Sn) H3Sn 271% 211% 181% 220% 291% 231% HSn 521% 601% 531% 601% 531% 601% Sn 201% 191% 281% 181% 171% 171%
C3H9Sn 781% 811% 801% 820% 822% 841% C3H8Sn 20% 11% 21% 10% 11% 11% C3H7Sn 191% 180% 181% 170% 171% 151% C4H11Sn 781% 762% 781% 781% 801% 801% C4H10Sn 21% 11% 21% 21% 11% 11% C4H9Sn 213% 180% 201% 200% 191% 191% C6H15Sn 971% 991% 971% 991% C6H14Sn 11% 11% 11% 11% C6H13Sn 21% 01% 11% 11% C7H17Sn 961% 981% 971% 991% 971% 991% C7H16Sn 21% 11% 21% 20% 11% 11% C7H15Sn 21% 11% 21% 01% 21% 01% C8H19Sn 962% 971% 971% 980% C8H18Sn 21% 21% 21% 10% C8H17Sn 21% 10% 21% 10% C9H21Sn 991% 1002% C9H20Sn 01% 12% C9H19Sn 11% 01% C10H23Sn 991% 1001% 991% 1000% C10H22Sn 01% 01% 01% 00% C10H21Sn 11% 00% 01% 00% C11H25Sn 991% 1000% 991% 1000% C11H24Sn 01% 00% 01% 00% C11H23Sn 11% 00% 01% 00% C12H27Sn 991% 1000% C12H26Sn 01% 00% C12H25Sn 11% 00% Values in subscript stand for uncertainty of the estimates (2s). 980% means uncertainty < 0.5%.
81 Table 5.2
Molecular formula reconstruction from clusters in EI mass spectra of ethylated butyltin species
BuSnEt3 (MBT) Bu2SnEt2 (DBT) Bu3SnEt (TBT)
Fragment Abundance Abundance Abundance Abundance Abundance Abundance (natural Sn) (119Sn) (natural Sn) (119Sn) (natural Sn) (119Sn) H3Sn 152% 81% 181% 112% 231% 162% HSn 612% 702% 622% 702% 591% 655% Sn 221% 221% 202% 192% 181% 183%
C2H7Sn 671% 641% 762% 742% 792% 794% C2H5Sn 282% 331% 192% 221% 162% 182% C2H3Sn 62% 31% 51% 22% 52% 31%
C4H11Sn 922% 931% 812% 812% 692% 682% C4H10Sn 21% 21% 11% 21% 11% 12% C4H9Sn 62% 61% 171% 171% 312% 311% C6H15Sn 984% 972% 982% 991% 982% 991% C6H14Sn 13% 22% 11% 12% 11% 11% C6H13Sn 14% 11% 22% 01% 22% 01% C8H19Sn 992% 1001% 983% 992% 962% 971% C8H18Sn 02% 01% 12% 01% 22% 11% C8H17Sn 01% 00% 13% 11% 22% 21% C10H23Sn 992% 1002% 992% 1001% C10H22Sn 01% 03% 02% 01% C10H21Sn 11% 01% 12% 01% C12H27Sn 1002% 1001% C12H26Sn 02% 00% C12H25Sn 02% 00% Values in subscript stand for uncertainty of the estimates (2s). 980% means uncertainty < 0.5%.
From the results shown in these tables we can draw several important conclusions:
First, and most important, fragment ions corresponding to trialkyltin ions appear to be free
+ from isobars. See, for example, fragments C12H27-nSn (Bu3Sn , 291 u) for Bu3SnEt and
Bu3SnPr.
+ Second, fragment ions corresponding to dialkyltin ions are mainly of the type of R2SnH
.+ + with a small contribution from R2Sn (1–2%) and the loss of H2 from R2SnH (1-2%). See, for
+ example, ions C8H19-nSn (around 235 u) for Bu2SnPr2, Bu3SnPr and Bu3SnEt in Tables 5.1 and
+ 5.2 where the R2SnH ion (C8H19Sn) appears to be on the order of 96–99%. Third, smaller
+ + fragments of the type RSnHx and HSnHx are always mixtures of species with x ∈ [0, 1, 2, 3], thus unsuitable for isotope ratio determinations.
82 Although isotope patterns of certain clusters show the presence of only one molecular formula cluster, it does not necessarily mean that the cluster represents a single species. This is clearly demonstrated in case of butyltin species where additional isobaric overlaps can be observed between the dialkyl and trialkyltin ions. For example, the ion cluster around m/z = 235 in the mass spectrum of Bu2SnEt2 appears to be pure C8H19Sn, however, it represents a mixture
+ + of two species Bu2SnH and BuEt2Sn . Note that these spectral overlaps are not so evident if only ethylation is performed. This was demonstrated experimentally using propyl- derivatives
+ + where isobars Bu2SnH (235 u) and BuEt2Sn (235 u) appear as two non-isobaric fragments
+ + Bu2SnH (235 u) and BuPr2Sn (263 u). Since we have demonstrated that dialkyltin ions are not completely free from isotope pattern-distorting interferences, trialkyltin ions that overlap with the dialkyltin ions should not be used for isotope ratio measurements. Similarly, m/z = 207 in
+ + + the BuSnEt3 mass spectrum is a mixture of BuEtSnH and Et3Sn and thus the Et3Sn fragment should not be selected for isotope ratio measurements.
The uncertainty in the molecular formula reconstruction is generally lower when using the
119Sn-enriched compounds and no discrepancies in fragmentation patterns was observed between natural and isotopically enriched Sn compounds (see Tables 5.1 and 5.2).
Table 5.3
Ideal butyltin fragment ion candidates for Sn isotope ratio measurementsa
x = 1 (MBT) x = 2 (DBT) x = 3 (TBT)
+ + + BuxSnEt4–x BuSnEt2 (C8H19Sn, 235 u) Bu2SnEt (C10H23Sn, 263 u) Bu3Sn (C12H27Sn, 291 u)
+ Bu2SnEt (C10H23Sn, 263 u)
+ + + BuxSnPr4–x BuSnPr2 (C10H23Sn, 263 u) Bu2SnPr (C11H25Sn, 277 u) Bu3Sn (C12H27Sn, 291 u)
+ + + Pr3Sn (C9H21Sn, 249 u) BuSnPr2 (C10H23Sn, 263 u) Bu2SnPr (C11H25Sn, 277 u)
a) Mass is given for 120Sn isotope.
83 In summary, for accurate tin isotope ratio measurements one should select trialkyltin ions where no overlap is possible with dialkyltin ions. Such fragment ions are ideal candidates for Sn isotope ratio measurements from molecular clusters and they are listed in Table 5.3. Since both ethylation and propylation provide trialkyltin fragments free from molecular interferences, either one of these derivatization approaches can be used in practice and the selection of the derivatization agent can be based on criteria such as stability, cost and chromatographic parameters, such as retention time and required temperature program.
5.3.ii Fragmentation behavior of butyltin compounds
Data shown in Tables 5.1 and 5.2 also provide insights into mass spectral fragmentation pathways. This already has been demonstrated for aliphatic diselenides.54 Of particular interest is that one can indirectly access isobaric overlap information without performing high- resolution mass spectrometry experiments.
It is evident that EI mass spectra of butyltins contain mainly [M – 2H] type of isobaric overlaps (molecular interferences) for C4H11Sn, C2H7Sn and H3Sn species. This suggests the
+ + + loss of hydrogen from triply substituted even-electron species BuSnH2 , EtSnH2 and HSnH2 .
+ + Decomposition routes leading to the triply substituted even-electron ions (R3E , R2EH and
+ 73 REH2 ) are also dominant in the mass spectra of alkyllead and alkylgermanium species. Main fragmentation pathways of butyltin species can be summarized as shown in Scheme 5.1.
Scheme 5.1 | EI mass spectral fragmentation pathways of tetraalkyltin compounds. Isobaric interferences are shaded gray.
84 + It is clear from here that severe isotope pattern distortion occurs for ions RSnH2 (including
+ HSnH2 ) where hydrogen elimination is possible. Fragment ion identity in the mass spectra of butyltin species can be attractively visualized using the modular representation (Scheme 5.2).
Modular analysis of mass spectra was recently introduced by Sweeney to facilitate mass spectral recogition.74 Such representation clearly shows the source of isotope pattern distortion
+ in the presence of RSnH2 ions. In fact, ∼99% of all the ions of the butyltin species mass spectra can be explained and represented in attractive modular representation with the exception being the rare methylene elimination leading to MeSn+ and EtSn+ from ethyl- and propyltin species respectively.72
.+ + + + + +.
+ + .+
+ +
+
+ + +
.+
+ + Sn
CH2CH2 H
+ +
Scheme 5.2 | Modular representation of main EI fragmentation pathways of diethyldibutyltin (Bu2SnEt2).
5.4. ISOTOPE RATIO MEASUREMENTS IN FRAGMENT IONS
Molecular ions in electron impact mass spectra are usually free from interferences or overlapping fragment ions, however, absence of molecular ions for butyltin species requires selection of the best candidate from the fragment ions. Also, the atomic tin clusters (Sn+)
85 + + contain overlapping molecular interferences of SnH and SnH3 , therefore adequate molecular clusters need to be selected for ID analysis.66 So far ID analysis of butyltin species with GC/MS has been performed using various fragment ions from their electron impact mass spectra. For example, ID analysis of ethylated tributyltin has been performed using clusters C8H19Sn (m/z =
75,76 75,77,78 66,75 235) , C10H23Sn (m/z = 263) and C12H27Sn (m/z = 291) and ethylated dibutyltin has
76 66,77,78 been determined using clusters C8H19Sn (m/z = 235) and C10H23Sn (m/z = 263) .
However, the only justification of such selections seems to be the higher abundance of the ion fragment. Indirect validity of such selection remains the accuracy of the obtained results. As we have seen (Tables 2 and 3) cluster C8H19Sn should not be used for TBT nor for DBT, since it is
+ a dialkyltin ion (Bu2SnH ) possessing minor interferences. The selection of the other ions listed above is valid according to our results.
In summary, after the fragment ion is known to be free from isotope pattern distorting interferences, the measured isotope ratios are valid for ID analysis. Also, there is no need to do
13C and 2H correction from the measured signal since 13C and 2H contributions could be incorporated into the certified isotopic abundances. Figure 5.4 shows good agreement between the natural isotope ratios calculated from the corresponding molecular clusters (taking into account the contribution of 13C and 2H) and those ratios directly measured from interference-
+ + free molecular clusters (Bu3Sn and Bu2SnEt ), demonstrating the ability to perform accurate isotope ratio measurements from molecular clusters. However, it is evident that isotope
+ measurements from the dialkyltin ion Bu2SnH (m/z = 235, Figure 5.4) have a large systematic error in order of ∼3%. Despite this fact, it should be noted that this fragment ion has been used in the literature for isotope dilution analysis even in international interlaboratory comparison
86 studies.75,76 This clearly demonstrates the need for a quantitative isobar detection algorithm for accurate isotope ratio measurements from molecular clusters.
measured certified 0.45 116Sn/120Sn
0.40 + + +
119Sn/120Sn 0.35
0.30 117Sn/120Sn
0.25 m/z = 235 m/z = 263 m/z = 291
Figure 5.4 | Natural isotope ratio measurements from three selected fragment ions in the EI mass spectrum of tributylethyltin.
5.5. CONCLUSIONS
We have designed a systematic approach to select interference-free fragment ions for isotope ratio measurements from molecular clusters. This procedure has shown to be particularly useful for the interpretation of mass spectra with focus on isotope ratio measurement. Although demonstrated only for butyltin species, this algorithm can be used for any other species to
+ ensure fragment ion validity for isotope dilution analysis. We have shown that only R3Sn clusters are free from isobaric interferences and we recommend the use of [M – 29]+ ions for isotope ratio measurements when butyltin compounds are ethylated or [M – 43]+ when propylation is performed.
87 Data analysis has been the Achilles heel in ID analysis in mass spectrometry for a long time76 and it is hoped that isotope pattern reconstruction will help in promoting a wider use of ID analysis for organic mass spectrometry in the future.
5.6. EXPERIMENTAL
Instrumentation. Chromatographic analysis was performed with an Agilent 6890N gas chromatograph (Agilent Technologies, Waldbronn; Germany) using a cross-linked 5%-phenyl- methylsiloxane capillary column (HP-5ms: 30 m × 0.25 mm i.d. × 0.25 µm coating). The GC was equipped with an Agilent 6973 Network MSD quadrupole mass detector (Agilent technologies,
Waldbronn; Germany).
The isotope composition for the 119Sn-enriched compounds was determined by GC/ICP-MS. A
Hewlett-Packard 6890 gas chromatograph (Palo Alto CA, USA) with a cross-linked 5%-phenyl- methylsiloxane capillary column (HP-5: 30 m × 0.25 mm i.d. × 0.32 µm coating) was coupled with the
HP-4500 inductively coupled plasma mass spectrometer (Yokogawa Analytical Systems, Tokyo; Japan).
Experimental parameters and references to more detailed instrumental descriptions are reported elsewhere.70,66
Reagents and materials. All solvents and reagents were of analytical grade. Tributyltin chloride
(Bu3SnCl, 96%), dibutyltin dichloride (Bu2SnCl2, 97%) and monobutyltin trichloride (BuSnCl3, 95%) of natural tin isotope composition were obtained from Aldrich (Steinheim, Germany). Sodium tetraethyl- and tetrapropylborates (NaBEt4 and NaBPr4) were obtained from Galab (Geesthacht, Germany). Metallic
119Sn-enriched tin was purchased from Cambridge Isotope Laboratories (Andover MA, USA). The mixture of 119Sn-enriched mono-, di-, and tributyltin compounds was obtained and characterized by Ruiz
Encinar et al.70
Procedures. Mixed standard solutions of butyltin compounds, either natural or 119Sn-enriched, were ethylated or propylated with alkaline NaBEt4 and NaBPr4 respectively in an acetate buffer solution (pH
88 = 5.4) followed by hexane extraction.66 GC separation of derivatized mono-, di- and tributyltin species was performed using the following temperature program: 50 oC (1 min) followed by a 30 oC min-1 ramp up to 250 oC (260 oC injection port temperature). Data acquisition was performed in full scan mode between m/z 100 and 320 for all compounds. The measurement of isotope abundances was performed using a 10 ms dwell time per mass. Detailed instrumental conditions are reported elsewhere.66
Calculations. Isotope pattern reconstruction was done using a non-weighted least-squares deconvolution algorithm as described in Chapter 4. Isobar deconvolution was performed for each mass spectra within the chromatographic peak resulting in approximately 20 calculations per chromatographic peak. Average values and standard deviations were calculated from these data. Calculations were performed using MathCad v.7 professional software (MathSoft Inc., USA). Tin isotope abundances used in this work are listed in Table 5.4.70,79
Table 5.4
Isotope composition of tin used in this study
Natural Sn* 119Sn-enriched Sn
116Sn: 0.1453 116Sn: 0.0003
117Sn: 0.0768 117Sn: 0.0011
118Sn: 0.2423 118Sn: 0.1433
119Sn: 0.0859 119Sn: 0.8240
120Sn: 0.3259 120Sn: 0.0313
122Sn: 0.0463 122Sn: < 0.0001
124Sn: 0.0579 124Sn: <0.0001
*The presence of low abundance 112Sn (0.0097),
114Sn (0.0065) and 115Sn (0.0034) was not relevant for this study.
* Isotope patterns for the 119-enriched Sn-containing molecular clusters Sn CnHm were calculated by
* combining the enriched tin (Sn ) and CnHm isotope patterns according to Eq. 5.1:
89 Sn*Cnm HSn* C nm H IIIzij=⋅∑ ()() (5.1) ij, ()ijz+=
Isotope patterns of CnHm were obtained by combining Cn and Hm patterns in a similar fashion and they were generated using a commercially available isotope pattern calculator implemented in MassLynxTM v4.0 (Micromass Ltd, England).
90
(c) Juris Meija, 2005
CHAPTER 6 | ALTERNATIVE LOOK AT THE METHOD OF ISOTOPE DILUTION
91 6.1. ABSTRACT
Since its introduction in 1939/40, isotope dilution (ID) analysis has been considered as a two- isotope ratio analysis. In practice, however, more than one isotope of the analyte carries the abundance information, thus there is inherent loss of information for elements having several major isotopes. About half of the naturally occurring elements fall into this category. Here we present a theory for ID analysis using complete isotope patterns of analyte, spike and their mixture. According to this approach, isotope dilution is mathematically treated as the superimposition of the natural isotope pattern of the analyte with the isotopically altered
(enriched) isotope pattern. From this point of view, the method of isotope dilution can be considered as an intermediate case between the method of standard additions and the method of internal standard, as the only difference between these being the isotope pattern similarity between the analyte and spike. The validity of the proposed approach has been demonstrated with analysis of butyltin standard reference materials and aspects of uncertainty propagation of the isotope pattern ID analysis are also discussed. The developed general approach is useful in performing isotope dilution from molecular clusters and, most importantly, it benefits the inherent mathematical simplicity which makes it attractive for wider audience.
92 6.2. INTRODUCTION
The isotope dilution (ID) method is widely recognized as one of the most accurate techniques in modern quantitative analysis. ID was first introduced by Rittenberg and Foster and independently by Ussing in 1939, based on the addition of a known amount of isotopically enriched species followed by the measurement of the altered isotope ratios of the analyte.80-82
Since then, this ratio conventionally is the ratio of two isotopes.83-86 In the presence of isobaric interferences ID method precision traditionally is increased by reporting the average result obtained from several two-isotope ratios. Although the inclusion of more isotope ratios at once is beneficial due to cancellation of random errors, it heavily complicates the mathematical expression and thus such a strategy has been abandoned in routine laboratory practice.
We have developed a general theory of ID analysis using the whole isotopic pattern image analysis. According to this approach, the ID process is mathematically treated as the mixing of the species natural isotope pattern with the isotopically enriched isotope pattern.40 The ratio of both isotope patterns that match experimentally observed values the best then is calculated and analyte concentration is estimated as the product of this ratio and the spike concentration.
Quantitation in analytical chemistry is usually achieved using the method of external calibration. In the presence of various interferences, however, the method of internal calibration has to be used to reduce or eliminate the various sources of errors. Among these, three fundamental approaches to quantitation are method of internal standard, method isotope dilution, and method of standard additions. ID is an intermediate case between the method of internal standard and method of standard additions. The only parameter differentiating among
93 these widely used quantitation approaches is the similarity (overlap) of isotope patterns between the analyte and spike (Figure 6.1).
Method of Internal Standard Method of Isotope Dilution Method of Standard Additions
258 260 262 264 266 268 270 258 260 262 264 266 268 270 258 260 262 264 266 268 270
0% Isotope pattern similarity 100%
Figure 6.1 | Illustrative comparison of three fundamental quantitation methods in analytical chemistry. Black columns represent the analyte and the off-set red columns – the spike.
Mathematically, the similarity between the two isotope patterns (S) can be defined as the normalized difference between the i-th isobaric ion intensities A1i and A2i:
∑()AA12ii− S =−1 i (6.1) ∑()AA12ii+ i
Ideally the information preserved in the two isotope ratio ID analysis is orthogonal, in other words, one isotope carries mostly the information regarding the natural species and the second measured isotope contains information about the enriched spike, thus leading to most precise quantitation. From this point of view, isotope ratios in the method of internal standard are totally orthogonal, thus giving the lowest uncertainty of isotope ratio measurements. Due to this reason, several ID analyses are performed using a naturally non-occurring isotope spike formally leading to the internal standard method. This is practiced, for example, in determination of uranium with the naturally non-existing 233U spike.87
94 Isotope pattern overlap
0% 20% 40% 60% 80% 100% 100.0% Uncertainty of
reconstructed 258 260 262 264 266 268 270 pattern ratio 10.0% case III: 258 260 262 264 266 268 270 identical patterns case II:
1.0% intermediate situation information limit noise limit 0.1% case I: orthogonal patterns 258 260 262 264 266 268 270
Figure 6.2 | Simulated uncertainty of the reconstructed 1:1 isotope pattern ratio as a function of isotope pattern similarity (overlap) assuming constant 0.1% relative signal uncertainty.
Figure 6.2 shows the simulated uncertainty of the reconstructed 1:1 isotope pattern ratio depending on the overlap between the two isotope patterns of interest. Uncertainty of the reconstructed isotope pattern ratio is obtained using the Monte-Carlo algorithm described later.
It is clear that the isotope pattern ratio cannot be extracted from two identical patterns and this situation is information limited. Consequently, one has to use the signal intensity domain for quantitation (as in standard additions). For two completely orthogonal (non-overlapping) isotope patterns, however, the result uncertainty is noise-limited and mass domain gives more precise quantitation possibilities.
6.3. THEORY
Considering the isotope pattern overlap, one can generally interpret the method of the isotope dilution as a quantitative isobar deconvolution problem if the enriched spike has some signal overlap with the isotopes present in the native sample. With respect to this one has to face the
95 problem of how to partition the observed (measured) isotope pattern into a linear combination of natural and spike isotope pattern images. For example, in the mixture consisting of the analyte (X) and spike (X*) both at concentrations cX and cX* respectively, the resulting
(observed) isotope pattern is simply the concentration-weighted sum of both isotope pattern
40,64 images – natural {PX} and enriched {PX*}:
IcIii∝⋅XX,X*X*,XX,X*X*, + cI ⋅ ii =⋅ aI + aI ⋅ i (6.2)
Consider the analyte and the isotopically enriched spike with the whole isotope patterns spanning n masses (n different isotopes). Isotope composition of the analyte are n×1 vectors
12 n 12 n ii ()AAaa, ,..., A a and ()AAs ,ss ,..., A where ∑ AAas= ∑ =1.
When Na molecules (or atoms) are mixed with Ns molecules (or atoms) of the spike, the isotope
12 n pattern of the obtained mixture of molecules (Nm = Na + Ns) is measured to be ()AAmm, ,..., A m.
The following also is true for the mixture:
⎛⎞⎛⎞⎛⎞NNN111 ⎜⎟⎜⎟⎜⎟as m NNN22 2 ⎜⎟⎜⎟⎜⎟as+= m (6.3) ⎜⎟⎜⎟⎜⎟...... ⎜⎟⎜⎟⎜⎟ ⎜⎟⎜⎟⎜⎟nn n ⎝⎠⎝⎠⎝⎠NNNas m
This can be rewritten using the isotope abundances:
⎛⎞AA11 ⎛⎞ ⎛⎞ A 1 ⎜⎟as ⎜⎟ ⎜⎟ m AA22 A 2 NNN⋅+⋅=⋅⎜⎟as ⎜⎟ ⎜⎟ m (6.4) as⎜⎟... ⎜⎟ ... m ⎜⎟ ... ⎜⎟ ⎜⎟ ⎜⎟ ⎜⎟nn ⎜⎟ ⎜⎟ n ⎝⎠AAas ⎝⎠ ⎝⎠ A m
⎛⎞AA11 ⎛⎞ ⎛⎞ A 1 ⎜⎟as ⎜⎟ ⎜⎟ m AA22 A 2 NN⋅+⋅=+⋅⎜⎟as ⎜⎟() NN ⎜⎟ m (6.5) as⎜⎟... ⎜⎟ ... as ⎜⎟ ... ⎜⎟ ⎜⎟ ⎜⎟ ⎜⎟nn ⎜⎟ ⎜⎟ n ⎝⎠AAas ⎝⎠ ⎝⎠ A m
96 NNasiii ⋅+AAAasm ⋅= (6.6) NNas++ NN as
We denote the relative number of the analyte atoms in the resulting mixture as α:
N N α = a and 1−=α s (6.7) NNas+ NNas+
Then, Eq. 6.6 can be simplified:
iii αα⋅+−⋅=Aasm()1 AA (6.8)
Equation 6.5 for the set of all isotopes then becomes:
⎛⎞AAA111 ⎛⎞⎛⎞ ⎜⎟asm ⎜⎟⎜⎟ AAA222 αα⋅+−⋅=⎜⎟asm()1 ⎜⎟⎜⎟ (6.8a) ⎜⎟... ⎜⎟⎜⎟ ...... ⎜⎟ ⎜⎟⎜⎟ ⎜⎟nnn ⎜⎟⎜⎟ ⎝⎠AAAasm ⎝⎠⎝⎠
The maximum likelihood value of α now can be obtained using the least squares optimization algorithm
−1 aFFFY= ()TT (6.9)
⎛⎞⎛⎞AA11 A 1 ⎜⎟⎜⎟as m α 22 2 ⎛⎞ ⎜⎟⎜⎟AAas A m aF==⎜⎟ Y = (6.9a) ⎝⎠1−α ⎜⎟⎜⎟...... ⎜⎟⎜⎟ ⎜⎟⎜⎟nn n ⎝⎠⎝⎠AAas A m
ii ii ii ii ∑∑∑∑AAs msaamss+− AA AA − AA α = ii ii ii (6.10) 2∑∑∑AAsa−− AA ss AA aa
ii Remembering that ∑∑AAas==1 for a two isotope ratio system (imax = 2) this equation becomes:
11 AAms− α = 11 (6.11) AAas−
97 In practice one measures only a subset of the whole isotope pattern and the actual relative
ii abundances in the given mass range do not satisfy the condition ∑ AAas= ∑ =1, hence this has to be taken into account. This can be done by replacing the isotope abundances in equation
6.8a, however, it is easier to apply the correction to the obtained concentration by introducing the isotope pattern coverage factor:
ms i ∑ As α M s ccas=⋅ ⋅ (6.12) 1−α ma i ∑ Aa M a
If the concentration of the spike is known (cs), the concentration of the analyte (ca) can be calculated from Eq. 6.12 using the least-squares isotope pattern reconstruction algorithm with variable α.40 According to this approach, the isotope pattern of the analyte is mixed with the isotopically enriched isotope pattern and the ratio of both patterns, α/(1 – α), that match the experimentally observed values the best is found (see Figure 6.3).
Figure 6.3 | Illustration of isotope pattern reconstruction using linear least squares algorithm.
98 These equations represent a simple least square optimization in a matrix notation. The input consists of X and X* species isotope pattern matrix (F) and the experimental (background- corrected) signal intensities or peak areas (Y).
From this point of view, isotope pattern dilution analysis is in principle the reverse problem of the isotope pattern reconstruction used to analyze the ion cluster purity in mass spectra of diselenides54 and butyltin species88, with emphasis on the isotope ratio measurements in the later report. A similar approach also has been used to eliminate the [M – nH] interferences from mass spectra.89
6.4. RESULTS AND DISCUSSION
In the two isotope dilution approach, the isotope ratio (I1/I2) is transduced into the analyte-to- spike amount ratio (nA/nB) by incorporating the isotopic abundances. For example, after measuring the signal intensity at masses 1 and 2 in a system of two species A and B, one solves the set of two linear equations to obtain the amount ratio of A and B:
I1B,1I II In1A⎛⎞IIA,1 B,1 ⎛⎞I1 2B,2 ⇒⇒=⎜⎟⎜⎟ (6.13) In2B⎝⎠IIA,2 B,2 ⎝⎠I2 IA,1I 1 IA,2I 2
Note that this approach is identical to classical estimation of two compound concentrations from absorbance measurements at two wavelengths according to Beer’s law. The set of n isotopes,
1 can be characterized with 2 nn()−1 unique two-isotope ratios. Therefore, in case of three major isotopes (elements like Sn, Pb, Se, Pt, Ge and Zn), one has three different two-isotope ratios and each of them will give slightly different ID analysis result. If the spike is enriched in the minor isotope (which usually is the case), the number of different major two-isotope ratios becomes twice as large (six), and one has formally has to decide the ultimate two-isotope ratio. This is
99 not a trivial task if the abundances of available major isotopes are similar in magnitude (for elements such as Ag, Br, or Pt). Isotope pattern dilution addresses this issue, providing average between the results of all possible two-isotope choices (Figure 6.4).
Figure 6.4 | Amount ratio of analyte to spike (as isotope pattern ratio) calculated using the isotope pattern reconstruction algorithm (P1/P2) compared to estimates obtained from various two-isotope ratios for determination of MBT, DBT and TBT in PACS-2 marine sediment reference material using 119Sn enriched spikes of MBT, DBT, and TBT.
6.4.i Effect of carbon isotope variations
The carbon isotopic pattern is subject to various natural fractionation phenomena and it is not always safe to assume the average IUPAC isotopic abundances for 12C and 13C. This can be overcome by introducing a 13C/12C ratio as a variable into the isotope pattern reconstruction algorithm. As an example, Figure 6.5 shows the quality of C8H19Sn isotope pattern reconstruction from m/z = 235 cluster in commercially available ethylated monobutyltin EI mass spectra as a function of 13C abundance.
100 2 0.99981 r
0.99980
0.99979
-80.0 -60.0 -40.0 -20.0 0.0 20.0 40.0 60.0 80.0
δ(13C), ppm
Figure 6.5 | Reconstruction of C8H19Sn cluster in EI mass spectra of commercially available ethylated monobutyltin allowing for variable 13C amount. Maximum regression coefficient corresponds to the lowest residual between the reconstructed and observed isotope patterns and yields δ(13C) ≈ –20 ppm. Background-corrected integrated ion peak areas were used for isotope pattern reconstruction.
Although the reconstruction quality improvements are in the order of 0.01‰, these improvements are, nevertheless, of a functional character and the obtained result of δ(13C) ≈ –
20 ppm is in agreement with 13C abundance in common industrial source organic compounds.90
6.4.ii Application to butyltin quantitation
In the last few years application of isotope dilution has gained popularity in elemental speciation where the aim is the determination of the individual chemical species in which an element is distributed in a given sample.91 Since tin has seven abundant isotopes (> 5%), it seems a good candidate for validation of quantitation using isotope pattern analysis. For this purpose, two certified marine sediment materials PACS-2 and BCR-646 were analyzed for their
MBT, DBT, and TBT content with GC/MS. 119Sn enriched MBT, DBT, and TBT were used as tracers. Results obtained from the developed whole isotope pattern ID were compared to those
101 calculated using conventional two-isotope ID analysis. As an example, determination of MBT,
DBT, and TBT in standard reference materials using ID/GC-MS was performed (Table 6.1).66
Table 6.1
Comparison of MBT, DBT and TBT concentration in standard reference materials
obtained using conventional two-isotope ratio and whole isotope pattern analysis.
MBT DBT TBT
pacs-2 bcr 646 pacs-2 bcr 646 pacs-2 bcr 646
Conventional two-isotope IDa 598 ± 16c 468 ± 8 984 ± 12 402 ± 32 824 ± 15 195 ± 22
Whole isotope pattern IDb 605 ± 18 468 ± 13 996 ± 7 404 ± 21 838 ± 14 195 ± 19
Certified values n/ad 410 ± 69 1090 ± 150 394 ± 36 980 ± 130 195 ± 18
a) 120Sn/119Sn.
b) Calculated using the same data set as for conventional ID with data background corrected peak areas (3 replicates).
c) Uncertainty is given as 2s.
d) Certified value 450 ± 50 ng Sn g-1 is found to be too low by various research groups.
It is easy to see that the whole isotope pattern isotope dilution approach gives results identical to those calculated from two-isotope ratios. Therefore, the whole isotope pattern approach can be regarded as an alternative look at the conventional two-isotope ratio isotope dilution analysis.
The uncertainty of the obtained results does not seem to improve when whole isotope pattern is used. Uncertainty budget calculations of the conventional two-isotope ratio ID for butyltin analysis shows that the main source of uncertainty was the concentration of spike solutions and that explains why the uncertainties in the concentrations are very similar for all experiments.92
6.4.iii Uncertainty of isotope pattern reconstruction
In the case of two-isotope ratio, the uncertainty of the ID analysis result can be expressed in a closed mathematical function. Although this uncertainty magnification function is quite
102 complicated87, it has been presented in numerous manuscripts and its U-shape profile with the minima at around 1:1 analyte-to-spike ratio is one of the most recognized images of conventional ID analysis (Figure 6.6).83,87,93
⎛⎞σ R 0.016 ⎜⎟ ⎝⎠R 0.014
0.012 I 0.010 11 I I11 + I21 22 0.008 R = I21 I12 I12 + I22 0.006 258 260 262 264 266 268 270
0.004
0.002
0.000 0.01 0.10 1.00 10.00 100.00 R
Figure 6.6 | Relative precision of the two-isotope ratio: classical U-shape profile with the minimum at around 1:1 analyte-to-spike ratio.
Under the assumption of constant relative standard deviation (s ∝ I), optimum uncertainty of two isotope ratio is achieved when the resulting ratio is the geometric average of two isotope ratios83,84,93:
I ⎛⎞⎛⎞II 1 =⋅⎜⎟⎜⎟X,1 X*,1 (6.14) ⎜⎟⎜⎟ III2⎝⎠⎝⎠ X,2 X*,2
For isotope pattern ID analysis, the result (analyte concentration) is obtained from the optimization procedure and thus there is no closed mathematical function for the analysis result and its uncertainty. In cases like this, where no expressions are available for error analysis,
Monte-Carlo simulations94 can be used to obtain the information about the uncertainty of the result.95,96 According to this method, each isotopic intensity is perturbed with normally distributed noise and large set of isotope pattern ratios are then calculated from the randomly perturbed isotopic intensities. Uncertainty then is calculated from the combined results.
103 Based on the analogy from two-isotope ID analysis, we suggested that in general the optimum uncertainty is achieved when the resulting isotope pattern approaches the geometric mean between the natural and spike isotope patterns (Scheme 6.1)
Simple case: Two-isotope ratio ID
I11 I22 σ R II11+ II21 11 21 = min R =≡⋅ I21 I12 R II1221+ II22 22
258 260 262 264 266 268 270
More complicated case: Isotope pattern ID P 1 P ? 2 σ R {P1} = min RP=≡{}1 ×{}P2 R {}P2 258 260 262 264 266 268 270
Scheme 6.1 | Hypothesis of the optimum precision conditions.
This hypothesis was verified using Monte-Carlo simulations which were in agreement with such postulate. Although such an approach does not prove the hypothesis, nevertheless it supports it. Rigorous mathematical proof of this hypothesis would be very helpful and instructive; however, there is none at the moment to our knowledge. In this study, several artificial analyte/spike isotope patterns were constructed. Then, a signal intensity proportional noise (s ∝ I, constant relative noise) was added to each pattern and the obtained isotope patterns then were combined in a various analyte/spike ratio followed by the isotope pattern deconvolution. Uncertainty of the isotope pattern ratio then was calculated from the set of the generated values. The results are summarized in Figure 6.7.
104 0.01 γ = 1.62
γ = 0.72
1E-3 Relative reconstruction uncertainty γ = 0.58
0.01 0.1 1 10 100 P /P 1 2
Figure 6.7 | Testing the hypothesis of the geometric mean using Monte-Carlo simulations: Relative isotope pattern reconstruction uncertainty as a function of pattern ratio. Isotope pattern ratio that results in a pattern, closest to that of geometric mean of both is shown below each example (see also vertical lines). Each data point is uncertainty from 1000 Monte-Carlo simulations with constant 0.10% relative noise. White noise (normal distribution) was generated by convolution of 20 uniformly-distributed random number generation functions. Black isotope patterns correspond to P2, and red to P1.
The obtained uncertainty profiles are in good agreement with the hypothesis of geometric mean. Monte-Carlo simulations were also performed on two-isotope patterns for which an analytical solution for minimum uncertainty exists. Both theoretical value and that from Monte-
Carlo simulations were in exact agreement. The uncertainty magnification function in isotope pattern reconstruction is noise-profile dependent. Figure 6.6 corresponds to the constant relative uncertainty condition which is common in mass spectrometry when ion peak areas are used. However, non-integrated signals usually follow Poisson statistics. These situations can be easily simulated using Monte-Carlo approach, however, rigorous analysis of Poisson statistics
105 effects were not undertaken in this study. It is clear at this point that more work needs to be done to investigate the rules of error magnification function beyond two-isotope ratios.
6.4.iv Applications to multiple spike isotope dilution
Isotope dilution analysis using the whole isotopic patterns has applications beyond the educational or theoretical point of view. The use of multiple spikes in isotope dilution is a natural two-isotope extension, where different species of the same element are enriched in alternate isotopes. As a consequence of this, isotope pattern analysis is well suited for multiple spike isotope dilution data analysis than the conventional multiple two-isotope ratio approach.
The main advantage of the multiple spike analyses is that exact analyte concentrations can be computed regardless of the extent of degradation. This is a very attractive benefit for metabolism studies of toxic species in complex organisms.
Let us consider Cr(III)/Cr(VI) system as an example of double spike isotope dilution analysis.97 In this situation we are interested in determination of Cr(III) and Cr(VI) concentration in a sample using a 50Cr-enriched Cr(III) spike and 53Cr-enriched Cr(VI) spike.
To account for Cr(III) oxidation and Cr(VI) reduction during the sample analysis, we introduce
α as relative amount of Cr(III) oxidized during the sample analysis, and β - relative amount of
Cr(VI) reduced. The observed isotope pattern of Cr(III), PCr(III), then is a weighted sum of natural Cr(III), enriched Cr(III) and contributions from natural and enriched Cr(VI) as a result of reduction. Cr(VI) pattern is formed similarly from natural and enriched Cr(VI) plus the contributions of natural and enriched Cr(III) as a result of oxidation during the sample analysis:
enr enr nat nat enr enr nat nat PcPcPCr(III)=⋅+⋅+−⋅+⋅βα() Cr(VI) Cr(VI) Cr(VI) Cr(VI)(1 )( cPcP Cr(III) Cr(III) Cr(III) Cr(III) ) (6.15) enr enr nat nat enr enr nat nat PcPcPcPcPCr(VI)=−()1 βα() Cr(VI) ⋅ Cr(VI) + Cr(VI) ⋅ Cr(VI) +() Cr(III) ⋅ Cr(III) + Cr(III) ⋅ Cr(III)
106 Experimental isotope patterns PCr(III) and PCr(VI) are then fitted using the above expressions with
α, β, cCr(III) and cCr(VI) as variables using the least square fitting algorithm as outlined in equations 6.8a and 6.9. Two fitting variable matrices are obtained which contain all the unknowns:
⎛⎞⎛⎞ββccenr()1− enr ⎜⎟⎜⎟Cr(VI) Cr(VI) ββccnat()1− nat ⎜⎟⎜⎟Cr(VI)and Cr(VI) (6.16) ⎜⎟⎜⎟()1−ααccenr enr ⎜⎟⎜⎟Cr(III) Cr(III) ⎜⎟⎜⎟nat nat ⎝⎠⎝⎠()1−ααccCr(III) Cr(III)
Inherent simplicity of the matrix algebra makes the calculations of the whole isotope pattern
ID analysis more intuitive and easier in comparison to the conventional two-isotope ratio ID approach. Complexity in two-isotope-ratio calculations arises from the fact that all the two- isotope ratio equations (analogous to Eq. 6.15) must be stitched together to obtain the unknown parameters (Eq. 6.16). Isotope pattern approach handles the same calculation in one step.
Uncertainty calculations are very complex when multiple spikes are used. This can be overcome only with numerical simulation algorithms, such as Kragten’s spreadsheet method98 or Monte-Carlo simulations.
6.5. CONCLUSIONS
Generally the more highly isotopically enriched materials are more expensive and less available. Accordingly, when the less enriched spike is used for isotope dilution analysis, spike abundance information is no longer confined to a single isotope and thus the inclusion of more isotopes in ID calculations can benefit the overall method precision. Loss of analytical information by selecting only two isotopes for quantitation is more relevant for multi-isotope elements, such as Sn, Mo and Cd. Although the isotope pattern ratio approach slightly
107 complicates the calculations as compared to two-isotope ratio approach, it may be advantageous for less enriched isotopic spikes where isobaric interferences have to be considered. Also, the isotope pattern ratio greatly simplifies the mathematical data treatment when multiple isotope spikes are used simultaneously including degradation factors.
6.6. EXPERIMENTAL
Butyltin speciation. All the analyses were performed on quadrupole GC/MS instrument using marine sediment certified reference material PACS-2 (National Research Council; Ottawa, Canada) and fresh-water sediment certified reference material BCR-646 (Institute for Reference Materials and
Measurements; Geel, Belgium). Mixed standard solutions of butyltin compounds, either natural or 119Sn- enriched, were ethylated or propylated with alkaline NaBEt4 and NaBPr4 respectively in an acetate buffer solution (pH = 5.4) followed by hexane extraction. GC separation of derivatized mono-, di- and tributyltin species was performed using the following temperature program: 50 oC (1 min) followed by a
30 oC min-1 ramp up to 250 oC (260 oC injection port temperature). Data acquisition was performed in full scan mode between m/z 100 and 320 for all compounds. The measurement of isotope abundances was performed using a 10 ms dwell time per mass. Detailed instrumental conditions are reported elsewhere by Centineo et al.66
Isotope pattern reconstruction. Isotope pattern reconstruction was done using a non-weighted least- squares deconvolution algorithm as described in Chapter 4.40 Background-corrected ion signal peak areas for each m/z values were used for isobar deconvolution within the chromatographic peak.
Calculations were performed using MathCad v.7 professional software (MathSoft Inc., USA). Tin isotope abundances used in this work are listed in Table 5.4. Isotope patterns for the 119-enriched Sn-
* * containing molecular clusters Sn CnHm were calculated by combining the enriched tin (Sn ) and CnHm isotope patterns according to Eq. 5.1. Isotope patterns of CnHm were obtained by combining Cn and Hm
108 patterns in a similar fashion and they were generated using a commercially available isotope pattern calculator implemented in MassLynxTM v4.0 (Micromass Ltd, England).
Monte-Carlo simulations. Normally distributed, N∈(µ, σ), random numbers were obtained by
94,96 summation of n = 20 independently generated uniformly distributed random numbers (ri):
n n ()r − ∑ i 2 x =+µσi=1 (6.17) n 12
Uniformly distributed random numbers were generated using “rand()” command in Microsoft® Excel v.10 and the quality of the obtained random numbers was inspected by plotting the ith random number vs. the (i + 1)th. No patterns were observed in such a graph thus confirming the absence of intrinsic correlations. Normally distributed random numbers with the mean µ = 0 were added to the simulated isotope pattern and the obtained pattern was subjected to weighted (w ∝ I–2) isobaric deconvolution algorithm to calculate the analyte/spike ratio. The procedure was repeated about N = 1000 times and the standard deviation of the obtained analyte/spike ratios was calculated as illustrated below:
{P1} {P1 +σεi ⋅ } σ R R = Ri = {}P2 {}P2 +σεi ⋅ R ε = N(0,1)
109
CHAPTER 7 | PROSPECTIVES
110
MATHEMATICAL SPECIATION
From the previous chapters it is clear that mathematical data analysis has become an important facet of bioanalytical chemistry. Although mathematics has long been intertwined with the biological sciences, an explosive synergy between biology, chemistry and mathematics seems poised to enrich and extend all three fields greatly in the coming decades.
Chemical noise is a ubiquitous problem in mass spectrometry (especially MALDI) arising from matrix impurities, fragmentation products or unidentified constituents can compromise spectral analysis. The use of isotope pattern and mass defect information can significantly reduce the complexity of mass spectra allowing efficient tracking of desired species. Taking advantage of mass defects of heteroelements, mass spectrometry data can be efficiently analyzed using “mass defect tags”.99 Using this technique, a mass defect mask is applied to observed signals and as a result, signals having certain mass defects are visualized while others are suppressed. This is a mathematical chemical noise suppression method. Such visualization approach is well suited for elemental speciation of trace elements in complex biological systems because of the large differences in mass defects between the trace elements of common interest
(such as Se, Hg, Sn and I) and the major elements C, N, O, and P.99 Mathematical mass defect analysis is already established as a compact visual tool for ultrahigh-resolution mass spectra in petroleum classification and characterization.100
Mathematics plays a vital role in understanding mass spectra of extremely complicated systems. Protein digestion and tandem MS peptide sequencing leads to vast data sets that can be only processed using automated mathematical algorithms. Similarly, study of lignins, humic substances or fuel components cannot be imagined with conventional approaches in
111 interpretation of mass spectra simply due to the enormous complexity of the systems. Consider, for example, an ESI/FT-ICR mass spectrum of South American crude oil taken on a 9.4 Tesla instrument with more than 11000 resolved components.101 How is one to comprehend such an amount of data without the use of data analysis? It has been recently shown that the van
Krevelen diagram can mathematically separate different compound classes in pyridine-extracted coal or petroleum samples and can also graphically distinguish fossil fuels according to their nature (coal vs. petroleum), maturation (coals of different rank), and processing (the same coal at two stages of liquefaction).102 In this diagram O/C atomic ratios of compounds are plotted on the x-axis and the H/C atomic ratios on the y-axis.102,103 Each molecular formula (compound) in this diagram is represented as a point whose coordinates are determined by elemental composition. Thus, changes in elemental composition result in reaction product displacement in this diagram. The van Krevelen and Kendrick diagrams are established as a simple and powerful graphical tools for amplifying and exposing compositional differences within and between complex organic mixtures (Figure 7.1).100,102,104
In summary, Kendrick mass defect plots allow for attractive visualization of gross compositional differences among sample sets, whereas van Krevelen plots show compound class-specific variations among samples.101 Both types of these images in principle can be thought as chemical fingerprints of extremely complicated systems.
112
Figure 7.1 | Drawing with mass spectrometry: Images of 3D van Krevelen plots (right). The 3D van Krevelen images of McDonalds branch and Rio Negro dissolved organic matter are constructed from ultra-high resolution ESI/FT-ICR-MS data (left). Adapted from
Kim et al.105
Currently elemental speciation considers not more than 10 species of a given element in a particular sample; therefore it does not require sophisticated data handling techniques.
However, recent advances in metallomics and a continually increasing interest in heteroelement speciation in large biomolecules will soon result in more complicated systems where data handling will become a major obstacle.
113
| REFERENCES
114 (1) Lobinski, R. Appl. Spectrosc. 1997 (51), 260A-279A. (2) Vela, N. P.; Olson, L. K.; Caruso, J. A. Anal. Chem. 1993 585A-597A. (3) Meija, J.; Mounicou, S.; Caruso, J. A. J. AOAC. Int. 2004 (87), 205-224. (4) Meija, J.; Montes-Bayón, M.; Caruso, J. A.; Sanz-Medel, A. Tr. Anal. Chem. 2005 in press. (5) Vacchina, V.; Mari, S.; Czernic, P.; Marques, L.; Pianelli, K.; Shaumlöffel, D.; Lebrun, M.; Lobinski, R. Anal. Chem. 2003 (75), 2740-2745. (6) Raab, A.; Meharg, A. A.; Jaspars, M.; Genney, D. R.; Feldmann, J. J. Anal. At. Spectrom. 2004 (19), 183-190. (7) Braga, P.; Montes-Bayón, M.; Alvarez, J.; López, J. M.; Sanz-Medel, A. J. Anal. At. Spectrom. 2004 (19), 1128-1133. (8) Montes-Bayón, M.; Meija, J.; LeDuc, D. L.; Terry, N.; Caruso, J. A.; Sanz-Medel, A. J. Anal. At. Spectrom. 2004 (19), 153-158. (9) Mehra, R. K.; Miclat, J.; Kodati, R. V.; Abdullah, R.; Hunter, T. C.; Mulchandani, P. Biochem. J. 1996 (314), 73-82. (10) Yen, T.-Y.; Villa, J. A.; DeWitt, J. G. J. Mass Spectrom. 1999 (34), 930-941. (11) Rubino, M. F.; Verduci, C.; Giampiccolo, R.; Pulvirenti, S.; Brambilla, G.; Colombi, A. J. Am. Soc. Mass Spectrom. 2004 (15), 288-300. (12) Meija, J.; Caruso, J. A. Inorg. Chem. 2004 (43), 7486-7492. (13) Satterfield, M.; Brodbelt, J. S. Inorg. Chem. 2001 (40), 5393-5400. (14) Nemirovskiy, O. V.; Gross, M. L. J. Am. Soc. Mass Spectrom. 1998 (9), 1020-1028. (15) Spain, S. M.; Rabenstein, D. L. Anal. Bioanal. Chem. 2004 (378), 1561-1567. (16) Tereza Fernandez, M.; Lurdes Mira, M.; Helena Florencio, M.; Jennings, K. R. J. Inorg. Biochem. 2002 (92), 105-111. (17) Wind, M.; Lehmann, W. D. J. Anal. At. Spectrom. 2004 (19), 20-25. (18) Chassaigne, H.; Vacchina, V.; Lobinski, R. Tr. Anal. Chem. 2000 (19), 300-313. (19) Ruiz Encinar, J.; Polatajko, A.; Szpunar, J.; Lobinski, R. Analyst 2004 (129), 846-849. (20) Montes-Bayón, M.; Grant, T. D.; Meija, J.; Caruso, J. A. J. Anal. At. Spectrom. 2002 (17), 1015- 1023. (21) Staedler, E. In Sulfur Nutrition and Sulfur Assimilation in Higher Plants: Molecular, Biochemical and Physiological Aspects; Brunhold, C., Rennenberg, H., De Kok, J., Stulen, I., Davidian, J.- C., Eds.; Paul Haupt: Bern, Switzerland, 2000, pp 187-202. (22) Lewis, B. G.; Johnson, C. M.; Delwiche, C. C. J. Agric. Food Chem. 1966 (14), 638-640. (23) Duckart, E. C.; Waldron, L. J.; Donner, H. E. Soil Sci. 1992 (53), 94-99. (24) Evans, C. S.; Asher, C. J.; Johnson, C. M. Austrian J. Biol. Sci. 1968 (21), 13-20. (25) Zayed, A. M.; Terry, N. J. Plant Physiol. 1992 (140), 646-652. (26) Terry, N.; Zayed, A. M.; de Souza, M. P.; Tarun, A. S. Annu. Rev. Plant Physiol. Plant Mol. Biol. 2000 (51), 401-432. (27) Montes-Bayón, M.; Yanes, E. G.; Ponce de Leon, C.; Jayasimhulu, K.; Stalcup, A.; Shann, J.; Caruso, J. A. Anal. Chem. 2002 (74), 107-113.
115 (28) Meija, J.; Montes-Bayón, M.; LeDuc, D. L.; Terry, N.; Caruso, J. A. Anal. Chem. 2002 (74), 5837- 5844. (29) Montes-Bayón, M.; LeDuc, D. L.; Terry, N.; Caruso, J. A. J. Anal. At. Spectrom. 2002 (17), 872- 879. (30) Potapov, V. A.; Amosova, S. V.; Petrov, P. A.; Romanenko, L. S.; Keiko, V. V. Sulfur Lett. 1992 (15), 121-126. (31) Boss, R.; Barnett, N. W. J. Anal. At. Spectrom. 1997 (12), 733-741. (32) Hillen, L. W.; Werner, R. L. J. Chromatogr. 1973 (79), 318-321. (33) Chasteen, T. G. Appl. Organometal. Chem. 1993 (7), 335-342. (34) Rebane, E. Chemica Scripta 1974 (7), 219-230. (35) Rebane, E. Chemica Scripta 1974 (5), 5-8. (36) Rebane, E. Chemica Scripta 1974 (6), 65-73. (37) Prabhakar, S.; Krishna, P.; Kundu, A.; Roy, S.; Vairamani, M. Rapid Comm. Mass. Spectrom. 1999 (13), 1564-1572. (38) Tel'noi, V. I.; Sheiman, M. S. Uspheki Khimii 1995 (64), 330-337. (39) Tsai, J. H.; Hiserodt, R. D.; Ho, C.-T.; Hartman, T. G.; Rosen, R. T. J. Agric. Food Chem. 1998 (46), 2541-2545. (40) Meija, J.; Caruso, J. A. J. Am. Soc. Mass Spectrom. 2004 (15), 654-658. (41) Laitinen, R.; Pakkanen, T. J. Molec. Struct. (Theochem) 1983 (91), 337-352. (42) Carles, S.; Lecomte, F.; Schermann, J. P.; Desfrançois, C.; Xu, S.; Nilles, J. M.; Bowen, K. H.; Berges, J.; Houee-Levin, C. J. Phys. Chem. A 2001 (105), 5622-5626. (43) Meija, J.; Bryson, J. M.; Vonderheide, A. P.; Montes-Bayón, M.; Caruso, J. A. J. Agric. Food Chem. 2003 (51), 5116-5122. (44) Amouroux, D.; Tessier, E.; Pécheyran, C.; Donard, O. F. X. Anal. Chim. Acta 1998 (377), 241-254. (45) Potapov, V. A.; Amosova, S. V.; Petrov, P. A.; Romanenko, L. S.; Keiko, V. V. Sulfur Lett. 1992 (15), 121-126. (46) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Zakrzewski, V. G.; Montgomery, J. A. J.; Stratmann, R. E.; Burant, J. C.; Dapprich, S.; Millam, J. M.; Daniels, A. D.; Kudin, K. N.; Strain, M. C.; Farkas, O.; Tomasi, J.; Barone, V.; Cossi, M.; Cammi, R.; Mennucci, B.; Pomelli, C.; Adamo, C.; Clifford, S.; Ochterski, J.; Petersson, G. A.; Ayala, P. Y.; Cui, Q.; Morokuma, K.; Malick, D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Cioslowski, J.; Ortiz, J. V.; Baboul, A. G.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.; Gomperts, R.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.; Gonzalez, C.; Challacombe, M.; Gill, P. M. W.; Johnson, B.; Chen, W.; Wong, M. W.; Andres, J. L.; Gonzalez, C.; Head- Gordon, M.; Replogle, E. S.; Pople, J. A.; Gaussian, Inc.: Pittsburgh, PA, 1998. (47) Glendening, E. D.; Reed, A. E.; Carpenter, J. E.; Weinhold, F. (48) Nakagawa, T.; Aoyama, E.; Kobayashi, N.; H., T.; Chikuma, M.; Sakurai, H.; Nakayama, M. Biochem. Biophys. Res. Commun. 1988 (150), 1149-1154. (49) Rabenstein, D. L.; Tan, K.-S. Magn. Reson. Chem. 1988 (26), 1079-1085. (50) Hillen, L. W.; Werner, R. L. J. Chromatogr. 1973 (79), 318-321. (51) Kice, J. K.; Lee, T. W. S.; Pan, S.-t. J. Am. Chem. Soc. 1980 (102), 4448-4455.
116 (52) Ryan, D. M.; Harpp, D. N. Tetrahedron Lett. 1997 (38), 8829-8832. (53) Cai, X.-J.; Uden, P. C.; Block, E.; Zhang, X.; Quimby, B. D.; J., S. J. J. Agric. Food Chem. 1994 (42), 2081-2084. (54) Meija, J.; Beck, T. L.; Caruso, J. A. J. Am. Soc. Mass Spectrom. 2004 (15), 1325-1332. (55) Milne, J.; J., W. A. Inorg. Chem. 1992 (31), 4534-4538. (56) Cataldo, F. Polyhedron 2000 (19), 681-688. (57) Yergey, J. A. Int. J. Mass Spectrom. Ion Phys. 1983 (52), 337-349. (58) McLafferty, F. W. personal communication 2004 (59) Foss, O. Acta Chem. Scand. 1950 (4), 404-415. (60) Foss, O. Organic Sulfur Compounds; Pergamon Press: Elmsford, N.Y., 1961; Vol. 1. (61) Höfle, G.; Baldwin, J. E. J. Am. Chem. Soc. 1971 (93), 6307-6308. (62) Lautenschlaeger, F. J. Org. Chem. 1969 (34), 4002-4006. (63) Loevenich, J.; Fremdling, H.; Four, M. Ber. 1929 (62B), 2856-2865. (64) Roussis, S. G.; Proulx, R. Anal. Chem. 2003 (75), 1470-1482. (65) Tanner, S. D.; Baranov, V. I.; Bandura, D. R. Spectrochim. Acta B. 2002 (57), 1361-1452. (66) Centineo, G.; Rodriguez-Gonzalez, P.; Blanco Gonzalez, E.; Garcia Alonso, J. I.; Sanz-Medel, A. J. Mass Spectrom. 2004 (39), 485-494. (67) Rockwood, A. L.; Van Orden, S. L.; Smith, R. D. Anal. Chem. 1995 (67), 2699-2704. (68) Sleno, L.; Volmer, D. A.; Marshall, A. G. J. Am. Soc. Mass Spectrom. 2005 (16), 183-198. (69) Donard, O. F. X.; Lespes, G.; Amouroux, D.; Morabito, R. In Trace Element Speciation for Environment, Food and Health; Ebdon, L., Pitts, L., Cornelis, R., Crews, H., Donard, O. F. X., Quevauviller, P., Eds.; Royal Society of Chemistry: Cambridge, 2001, p 391. (70) Ruiz Encinar, J.; Monterde Villar, M. I.; Gotor Santamaria, V.; Garcia Alonso, J. I.; Sanz-Medel, A. Anal. Chem. 2001 (73), 3174-3180. (71) Yang, L.; Mester, Z.; Sturgeon, R. E. Can. J. Anal. Sci. Spectrosc. 2003 (48), 211-218. (72) Chambers, D. B.; Glockling, F.; Weston, M. J. Chem. Soc. A 1967 (11), 1759-1769. (73) Glockling, F.; Light, J. R. C. J. Chem. Soc. A 1968 (3), 717-734. (74) Sweeney, D. L. Anal. Chem. 2003 (75), 5362-5373. (75) Sturgeon, R. E.; Wahlen, R.; Brandsch, T.; Fairman, B.; Wolf-Briche, C.; Garcia Alonso, J. I.; Rodriguez-Gonzalez, P.; Ruiz Encinar, J.; Sanz-Medel, A.; Inagaki, K.; Takatsu, A.; Lalere, B.; Monperrus, M.; Zuloaga, O.; Krupp, E.; Amouroux, D.; Donard, O. F. X.; Schimmel, H.; Sejeroe-Olsen, B.; Konieczka, P.; Schultze, P.; Taylor, P.; Hearn, R.; Mackay, L.; Myors, R.; Win, T.; Liebich, A.; Philipp, R.; Yang, L.; Willie, S. Anal. Bioanal. Chem. 2003 (376), 780- 787. (76) Bancon-Montigny, C.; Maxwell, P.; Yang, L.; Mester, Z.; Sturgeon, R. E. Anal. Chem. 2002 (74), 5606-5613. (77) Inagaki, K.; Takatsu, A.; Watanabe, T.; Kuroiwa, T.; Aoyagi, Y.; Okamoto, K. Anal. Bioanal. Chem. 2004 (378), 1265-1270. (78) Cardellicchio, N.; Giandomenico, S.; Decataldo, A.; Di Leo, A. Fresenius J. Anal. Chem. 2001 (369), 510-515.
117 (79) IUPAC Quantities, Units and Symbols in Physical Chemistry; 2 ed.; Blackwell Science, 1993. (80) Schoenheimer, R.; Ratner, S.; Rittenberg, D. J. Biol. Chem. 1939 (130), 703-732. (81) Rittenberg, D.; Foster, G. L. J. Biol. Chem. 1940 (133), 737-744. (82) Ussing, H. H. Nature 1939 (144), 977. (83) Heumann, K. G. Mass Spectrom. Rev. 1992 (11), 41-67. (84) Heumann, K. G. In Inorganic Mass Spectrometry; Adams, F., Gijbels, R., Van Grieken, R., Eds.; John Wiley & Sons, Inc.: New York, 1988, pp 301-376. (85) Fasset, J. D.; Paulsen, P. J. Anal. Chem. 1989 (61), 643A-649A. (86) Smith, D. H. In Inorganic Mass Spectrometry. Fundamentals and Applications; Barshick, C. M., Duckworth, D. C., Smith, D. H., Eds.; Marcel Dekker, Inc.: New York, 2000, pp 223-240. (87) Adriaens, A. G.; Kelly, R. W.; Adams, F. C. Analytical Chemistry 1993 (65), 660-663. (88) Meija, J.; Centineo, G.; Garcia Alonso, J. I.; Sanz-Medel, A.; Caruso, J. A. J. Mass Spectrom. 2005 (in press), (89) Moraes, M. C. B.; Brito Neto, J. G. A.; F., J. V.; do Lago, C. L. Int. J. Mass Spectrom. 1998 (178), 129-141. (90) Acsche, S.; Michaud, A. L.; Brenna, T. J. Current Org. Chem. 2003 (7), 1527-1543. (91) Rodriguez-Gonzalez, P.; Marchante-Gayón, J. M.; Garcia Alonso, J. I.; Sanz-Medel, A. Spectrochim. Acta B. 2005 (60), 151-207. (92) Rodriguez-Gonzalez, P.; Ruiz Encinar, J.; Garcia Alonso, J. I.; Sanz-Medel, A. J. Anal. At. Spectrom. 2004 (19), 685-691. (93) Riepe, W.; Kaiser, H. Fresenius J. Anal. Chem. 1966 (223), 321-335. (94) Güell, O. A.; Holcombe, J. A. Anal. Chem. 1990 (62), 529A-541A. (95) Schwartz, L. M. Anal. Chem. 1975 (47), 963-964. (96) Efstathiou, C. E.; Hadjiloannou, T. P. Anal. Chem. 1982 (54), 1525-1528. (97) Kingston, S. H. M.; Huo, D.; Lu, Y.; Chalk, S. Spectrochim. Acta B. 1998 (53), 299-309. (98) Kragten, J. Analyst 1994 (119), 2161-2165. (99) Hall, M. P.; Ashrafi, S.; Obegi, I.; Petresch, R.; Petersson, J. N.; Schneider, L. V. J. Mass Spectrom. 2003 (38), 809-816. (100) Hughey, C. A.; Hendrickson, C. L.; Rodgers, R. P.; Marshall, A. G.; Qian, K. Anal. Chem. 2001 (73), 4676-4681. (101) Rodgers, R. P.; Schaub, T. M.; Marshall, A. G. Anal. Chem. 2005 (77), 21A-27A. (102) Wu, Z.; Rodgers, R. P.; Marshall, A. G. Anal. Chem. 2004 (76), 2511-2516. (103) van Krevelen, D. W. Fuel 1950 (29), 269-283. (104) Kendrick, E. Anal. Chem. 1963 (35), 2146-2154. (105) Kim, S.; Kaplan, L. A.; Benner, R.; Hatcher, P. G. Mar. Chem. 2004 (92), 225-234.
118