Review pubs.acs.org/CR
Analytical Methods for Characterizing High-Mass Complex Polydisperse Hydrocarbon Mixtures: An Overview A. A. Herod,*,† K. D. Bartle,‡ T. J. Morgan,§,# and R. Kandiyoti† †
Department of Chemical Engineering, Imperial College London, London SW7 2AZ, U.K. Energy Research Institute, University of Leeds, Leeds LS2 9JT, U.K. § European Commission Joint Research Centre, Institute for Energy, Westerduinweg 3, 1755 ZG Petten, The Netherlands ‡
5.6.1. Electrospray Ionization Mass Spectrometry (ESIMS) 5.6.2. Field Desorption and Atmospheric Pressure Photoionization Methods 5.7. Analysis of Complex, Polydisperse Samples by Laser Desorption/Ionization Mass Spectrometry (LDTOFMS) 5.8. Upper Mass Detection Limits of LD-MS Systems 6. Examining Higher Mass Fractions by Solution State 13C NMR 7. Comparing Calculated Parameters from Three Distinct Samples 7.1. Coal Tar Pitch Fractions 7.2. Fractions of Maya (Mexican) Heavy Crude 7.3. Examining Fractions of Synthetic Crude Prepared from the Athabasca Tar Sands 7.4. Common Features of Results from the Three Sets of Samples 8. Summary and Conclusions 8.1. Aims of the Review 8.2. The Need for Fractionation 8.3. Limitations of Individual Analytical Techniques 8.4. Several Novel Approaches of the Work 8.5. Closing Emphatic Remarks Author Information Corresponding Author Present Address Notes Biographies Acronyms References
CONTENTS 1. Introduction 2. Fractionation Methods Used for Coal and Biomass Liquids and Petroleum Asphaltenes 2.1. Solvent Separation 2.2. Planar Chromatography (PC)/Thin Layer Chromatography 2.3. Column Chromatography 2.4. Recovering Successive Fractions from SEC 2.5. Ultrafiltration (UF) 3. Examining High Molecular Mass Fractions by UVFluorescence Spectroscopy (UV−F) 3.1. UV-Fluorescence Spectra in Static Mode 3.2. Trends Observable by UV-Fluorescence Spectroscopy 3.3. Examining UV-Fluorescence Spectra 3.4. High Mass Limits to UV-Fluorescence 4. Examining High Mass Fractions by Size Exclusion Chromatography (SEC) 4.1. Limitations of Using Tetrahydrofuran As Eluent in SEC 4.2. Limitations of Using NMP as Eluent in SEC 4.3. How To Explain the “Excluded Peak”? 4.4. The Use of NMP−Chloroform Mixtures As Eluent 5. Examining High Mass Fractions by Mass Spectrometry 5.1. Gas-Chromatography − Mass Spectrometry 5.2. Pyrolysis-GC-Mass Spectrometry (Py-GC-MS) 5.3. Heated Probe-Mass Spectrometry 5.4. Field Ionization Mass Spectrometry 5.5. Laser Induced Acoustic Desorption (LIAD) 5.6. Complex, Polydisperse Samples by FT-ICRMS and Different Ionization Methods © 2012 American Chemical Society
3892 3894 3894 3894 3894 3895 3895 3895 3895 3895 3897 3899 3900 3901 3902 3902
3907 3908
3908 3909 3910 3912 3912 3913 3914 3917 3918 3918 3919 3919 3919 3919 3920 3920 3920 3920 3920 3921 3921
1. INTRODUCTION This review addresses issues involved in attempting to define the complete molecular mass ranges of polydispersed hydrocarbon mixtures, such as those found in products from the pyrolysis and gasification of coal, biomass, and waste materials as well as heavy fractions and residues encountered in petroleum processing. Literature coverage ranges from 1978 to 2011.
3903 3904 3904 3904 3906 3906 3907
Received: November 14, 2011 Published: April 24, 2012
3907 3892
dx.doi.org/10.1021/cr200429v | Chem. Rev. 2012, 112, 3892−3923
Chemical Reviews
Review
GC-mass spectrometry (GC-MS) are less precise than GC and GC-MS. Above this range, no single method appears unambiguously capable of determining molecular mass distributions or indicating the predominant chemical structural features present within the samples. Advances in this field have required using and comparing evidence from several independent analytical methods [e.g., cf. refs 7 and 8]. In this type of work, it is essential to distinguish between inherent limitations of particular analytical techniques and the inferred properties of complex sample fractions. In this context, UV−F intensities fall off rapidly above 1000−1500 u, decaying to near-zero9−14 above 2800−3000 u. It is not surprising therefore that UV−F based methods show far narrower molecular mass distributions than several other techniques. In other words, UV−F based estimates of mass ranges of heavy petroleum fractions are found to reflect the fundamental limitations of fluorescence based methods themselves. Similarly, methods seeking to characterize the whole of a widely polydisperse sample in a single analytical step tend to reflect the properties of more the abundant components within the sample mixture, or the subfraction that is most amenable to analysis by that particular technique. For example, nearly half the asphaltene sample extracted from a Maya (Mexican) crude oil was found not to contribute a discernible signal to the UV− F spectrum of the bulk asphaltene.2 Fractionation prior to analysis of samples of such wide polydispersity has been shown to sidestep some of these difficulties. Useful fractionation methods include solvent solubility, planar and column chromatography, preparative size exclusion chromatography (SEC) and, less commonly, ultrafiltration.7,8,15 The characterization of such fractions in isolation tends to allow arriving at more detailed information about structural features and molecular mass distributions of complex samples. Mass spectrometric methods play an important part in characterizing such complex samples. A critical evaluation and an overview of mass-spectroscopic work relevant to the characterization of high-mass, polydisperse samples, highlighting several recent useful applications will be presented later. More recent developments in size exclusion chromatography (SEC) have also been outlined, with a view to superposing findings from SEC on observations from various mass spectrometric techniques. These developments may be combined with several new applications in nuclear magnetic resonance spectroscopy (13C and 1H), for elucidating the structural characteristics of complex samples. In the inexact world of fossil fuel derived “macromolecules”, it is difficult to attain certainty, still less aspire to levels of precision expected from the analysis of more conventional chemical species by GC-MS. One alternative view from the upstream petroleum industry suggests that asphaltene molecules are typically quite small, made up of 5−11 aromatic rings with heteroatom and alkyl substituents. It has been proposed that such molecules form “nanoaggregates” which can themselves cluster together in stable arrangements. According to this view, the formation of still larger clusters from nanoaggregates may then be initiated by a change in conditions such as exploratory drilling.16,17 Wiehe18 notes, on the other hand, that in process chemistry, resid asphaltenes do not volatilize out of cokers, operated at about 500 °C or distill over during distillation, as would be expected if these fractions consisted of aggregates of smaller molecular species. There is, in fact, nothing unexpected about the proposition that petroleum derived complex molecules can form aggregates.
This review will pursue several parallel lines of enquiry. Number and weight (mass) average molecular masses of heavy hydrocarbons may be estimated by numerous analytical methods. On their own, the utility and the information carried by these parameters are limited. They do not address questions regarding the upper mass limits of sample fractions and those focusing on the nature of structural features making up material present in low and high molecular mass fractions. Furthermore, most of the analytical methods used for estimating these parameters have their particular limitations; often there are questions regarding whether the particular technique allows accessing the entire range of molecular masses. Vapor pressure osmometry is known to suffer from the formation of aggregates, while UV-fluorescence (UV−F) spectroscopy tends to show a low mass bias that tends to skew molecular mass distributions of complex mixtures. Mass spectrometric methods suffer from limitations of mass discrimination when applied to complex mixtures, but such limitations are not always evident without detailed investigation. In petroleum refining, the phrase “bottom of the barrel” refers to the heavier parts of a crude oil that cannot be processed easily. Refineries routinely treat these heavier fractions as little better than waste products. Coker units draw small amounts of hydrogen and light hydrocarbons out of vacuum resids during refinery coking processes. Much of the residual solids (“petroleum coke”) are used as low grade, highly polluting fuel, or simply thrown away to landfill. However, high oil prices are now probably here to stayat least for a while. With a barrel of benchmark crude costing upward of US$ 100 (at the time of writing), refineries worldwide are now looking for better ways to utilize the “ultracomplex petroleum macromolecules” that constitute the heavier fractions of crude oils.1 There is an overarching need for designing new, efficient process routes and more selective catalyst systems for improving the product slate from the upgrading of heavy petroleum macromolecules. However, such a transformation requires a better understanding of the structures and compositions of these heavier fractions. Asphaltenes and resids generally tend to be structurally complex and widely polydisperse. What we know about their structures suggests they are generally made up of molecules containing one or more “large” polynuclear aromatic ring systems. “Large” is not specific but suggests such molecules would be beyond the range of gas chromatography, that is, more than five or six rings. The latter may be in configurations that range from “continental” to “archipelago” type formations, embedded in alicyclic or heterocyclic matrices. Some of these clusters are thought to be linked by aliphatic or heteroetheric bridging structures.2,3 There is an active debate on the proportion of continental versus archipelago structures.4−6 The proportion of structural types appears to depend on the nature of the particular crude oil. Work on asphaltenes from Maya crude oil (Mexico) described later2 has shown that the 1-methyl-2-pyrrolidinone (NMP) soluble fraction (up to 50% by weight of the asphaltene) consisted mainly of archipelago structures, while the fraction insoluble in NMP consisted mainly of continental structures. The peculiarities of these heavy fractions partly explain why it has been difficult to achieve greater certainty regarding their molecular mass distributions and their structural make-ups. Analytical tools available for characterizing materials boiling above the range detectable by gas-chromatography (GC) or 3893
dx.doi.org/10.1021/cr200429v | Chem. Rev. 2012, 112, 3892−3923
Chemical Reviews
Review
findings from specialized applications of solution state 13C NMR methods will be described. The methods under review naturally adapt to characterizing tars and heavy residues from the thermochemical processing of other solid fuels, such as biomass materials, plastics, waste and tar-sands derived fractions.
It is clear that when heavy fractions are concentrated and cooled, aggregation due to some form of colloidal association precedes the processes that eventually lead to the formation of a solid phase. The question that concerns analytical work relating to heavy hydrocarbon fractions (asphaltenes, resids and their fractions), however, is whether the large molecules observed during analysis in dilute phase are composed of smaller molecules forming high mass aggregates or made up of larger complex molecules of corresponding molecular mass. The view that high mass materials in crude oils are made up of “aggregates” of smaller molecules has been adhered to for much of the long history of the oil industry. It has shed little light on the structural problems at hand and has not enabled making better use of the vast tonnages of the “bottom of the barrel” that, historically, have been systematically discarded. In the meantime, markets have entered a period of relentlessly rising crude oil prices. Historically, this may be dismissed as a cyclical phenomenon. However, present high prices are now coupled to increasing tonnages of heavier crude oils coming on the market. This combination of factors has highlighted the multiplicity of new needs and some of the persistent shortcomings in the analysis of these materials. It is no coincidence that in countries where extraction of heavy crudes has become part of daily reality, viz. Canada and Mexico, research aiming to improve the processing of heavier fractions has largely ignored the proposition that large molecules may be composed of aggregates of smaller ones. Meanwhile, there are no clear processing alternatives available to oil refiners. Many of the more recent upgrading processes borrow heavily from such interrelated technologies as hydro-demetalation (HDM), hydro-desulphurization (HDS), and hydro-denitrogenation (HDN). These approaches are largely based on the corpus of knowledge and know-how developed for upgrading lighter, more tractable fractions such as light and heavy naphthas.19 As their names suggest, they require much additional and expensive fresh hydrogen. Many of the solutions aimed at wasting less of the oil coming out of the ground, thus, retain some of the fundamental difficulties they were intended to resolve. The foregoing strongly argues in favor of attempting to bring greater clarity to our understanding of the molecular mass distributions and structural features of heavy “liquid” hydrocarbons derived from processing fossil fuels, waste material and biomass. In the short term, the effort may not change existing processing strategies designed to discard these materials as discretely as possible. Nevertheless, smarter processing solutions will need to be based on useful new analytical approaches and the resulting improved appreciation of potential opportunities. The present review will focus on evaluating analytical tools that have proved more useful in examining the molecular mass distributions and structural features of complex, widely polydispersed heavy hydrocarbons. The discussion will focus on samples with molecular masses greater than those normally identifiable by GC or GC-MS; the latter are usually limited to m/z 350−500. While mostly working with petroleum derived asphaltenes and resids, where relevant, reference will be made to work on the characterization of coal derived tars, pitches, and extracts and on biomass-derived tars. In what follows, the focus will be on results from UV−F spectroscopy, size exclusion chromatography (SEC), and several mass spectrometric methods. Recent work on a powerful approach linking this battery of techniques with
2. FRACTIONATION METHODS USED FOR COAL AND BIOMASS LIQUIDS AND PETROLEUM ASPHALTENES 2.1. Solvent Separation
Widely used, this method requires relatively large volumes of solvent, when the fractionation of gram size samples is required. A wide range of solvents can be used and it is usually necessary to dry the insoluble residue before the next solvent is used. The method can be quantitative provided all solvent residues can be removed, without loss of volatile components from the sample fraction. Problems arise when using stabilized solvents such as THF. The stabilizer can appear in one or more of the recovered sample fractions. Polymerization related effects have also been observed during the evaporation of NMP from sample fractions, when measures are not taken to exclude air/oxygen.20 2.2. Planar Chromatography (PC)/Thin Layer Chromatography
This is a rapid, low cost fractionation method, which uses small quantities of solvent and readily available chromatography plates. Photography of the developed chromatograms allows a permanent record to be kept and standard compounds can be included in the same experiment. When analyzing samples of low solubility, none of the sample material is lost and all of it remains within the surface area of the chromatographic coating, with the least mobile (usually most insoluble) fraction left at or near the origin, that is, the point of application of the sample. PC requires solvents to be significantly more volatile than the sample. A sequence of solvents can be used in any order. The solvent from each development stage is dried before the next solvent can be applied, which enables the sequential use of solvents that are mutually immiscible. Sample fractions may be recovered after the final development by scraping coating material into a suitable solvent. Larger sample size “preparative” developments are possible using thick coating layers with multiple developments. The technique allows the repeated use of the same solvent to achieve more complete separations. However, the method is not quantitative and recovered sample fractions are ordinarily no greater than a few milligrams. Examples of the application of PC to the fractionation of a pine wood tar pitch known as Stockholm tar21 as well as to coal tar pitch and petroleum derived samples have been described elsewhere.22−24 2.3. Column Chromatography
This method uses a column of silica in a glass tube, with the silica heated before use to activate it and to remove adsorbed organics. The sample is initially adsorbed onto an aliquot of silica, placed on top of the column already containing the cleaned silica and eluted with a sequence of solvents. Solvents flow through the column in sequence, washing out soluble material. The “next” solvent to be used needs to be miscible with the previous solvent. Typical solvents include acetonitrile, pyridine, and NMP.25,26 Fractions collected in solution may be recovered quantitatively by careful evaporation of the solvent. 3894
dx.doi.org/10.1021/cr200429v | Chem. Rev. 2012, 112, 3892−3923
Chemical Reviews
Review
3. EXAMINING HIGH MOLECULAR MASS FRACTIONS BY UV-FLUORESCENCE SPECTROSCOPY (UV−F) For a general introduction, the reader may wish to refer to one of numerous texts available (e.g., cf. ref 39).
However, the weights of individual fractions recovered indicate some loss of volatile components and of material irreversibly bound to the silica. NMP can be removed without forming polymeric products by using established procedures, which require excluding contact with air/oxygen.20 Column chromatography gives similar separations as planar chromatography but allows the recovery of larger (typically gram) quantities of sample.
3.1. UV-Fluorescence Spectra in Static Mode
The method requires the sample to be dissolved completely. NMP and chloroform are typical of solvents with sufficient strength to dissolve many of the samples being considered in this paper. Acquisition of spectra may be performed in the form of excitation, emission, or synchronous spectra. The method is sensitive to small concentrations of sample (less than 0.1 mg per liter of solution); indeed, dilute sample solutions are necessary in order to minimize self-absorption effects. Self-absorption occurs when the fluorescence emitted within the body of the solution in the cell is absorbed by another molecule in the same solution before escaping from the cell; this reduces the intensity of fluorescence and can result in emission from the second molecule at a longer wavelength if the photon from the initial fluorescence is of sufficient energy. When a solution of a strongly fluorescent sample is sequentially diluted, observed fluorescence intensities are initially observed to increase. This is due to the dilution process reducing the loss of UV-light by self-absorption effects. Eventually, on further dilution, the intensity of the spectrum begins to diminish and the relative intensities of peaks no longer change. At this level of dilution, the concentration is considered to be optimal for generating UV−F spectra. The method is capable of high levels of reproducibility. The relevance of UV−F spectroscopy arises from the fluorescence emitted due to transitions between the π-bonding and π*-antibonding electronic energy levels of aromatic and polycyclic aromatic ring systems. Diverse types of structural features may contribute to the positions (and shapes) of particular bands in the spectra. There are overlaps between characteristic emission bands of structurally distinct aromatic ring systems. UV−F spectroscopy does not, therefore, allow precise interpretations in terms of the occurrence of individual structural features. In particular, there is little scope in attempting to infer precise numbers of fused rings and/or structural configurations of aromatic ring systems, by using this technique alone. Overall, it is necessary to be aware that changes in signal due to large numbers of distinct structural features may contribute to particular shifts in the spectra. The difficulty is common to the evaluation of data from most analytical techniques applied to ultracomplex materials such as coal or petroleum derived heavy hydrocarbons. Nevertheless, the method is useful in fingerprinting individual compounds as well as comparing the spectra of complex samples containing polycyclic aromatic ring systems.
2.4. Recovering Successive Fractions from SEC27
Analytical SEC columns may be used but allow only very small amounts of sample to be recovered. Injecting a relatively large amount of sample requires the use of a wide-bore preparative SEC column. The SEC effluent may be collected in successive vials for further analysis, sometimes without further concentration. The SEC of many relevant samples leads to a chromatogram with an apparently bimodal distribution of molecular masses, with part of the sample eluting in the excluded region of the column. It is possible to overload the column if a large amount of sample is injected in relation to column capacity. This results in the broadening of the excluded peak and the separation is compromised.28 One example of the overloading of the excluded region involved the collection of fractions from the analytical separation where the third fraction on reinjection eluted at the collection time of the second fraction.29 In general refractive index detectors are not sufficiently sensitive to operate at the low concentrations required to avoid overloading the front end of the chromatogram. When overloading occurs, the excluded peak broadens toward later elution times and eventually fills the valley between excluded and retained peaks.28 2.5. Ultrafiltration (UF)
This is an established technique commonly used for purifying and concentrating protein solutions.30−33 UF membranes can effectively act at the molecular level, to remove larger molecules from solution on the basis of size. Suspended solids and solutes of high molecular mass (size) are retained by the membrane, while solvent and lower molecular mass sample molecules pass through the membrane. Details of the mechanism of separation can be found elsewhere.32−34 Permeability through the membrane depends on a number of factors such as the composition of the solute, its molecular size distribution, solubility properties, and the chemical and physical properties of individual molecules, in addition to solvent properties and the properties of the solution as a whole. These factors cannot be readily isolated and quantified. When the difference in mass between solvent and solute is greater than 10-fold, a separation of >90% can be achieved.32 The use of UF has generally been restricted to the separation of simple solutions, for example, a light solvent and a large protein. With regard to the use of UF in the study of complex mixtures such as coal and petroleum derived heavy hydrocarbons, it is known that the technique cannot provide discrete separations. To date, it has been rarely used for studying complex samples. Only a few reports may be found in the literature for its application to coal or petroleum derived complex mixtures. Nonetheless, some useful insights on molecular sizes have been put forward.15,35−38 UF is particular useful when used in parallel with planar or column chromatography.15 The different fractionation mechanisms of each of the methods make it possible build a more complete picture of complex samples.
3.2. Trends Observable by UV-Fluorescence Spectroscopy
UV−F spectra are observed to shift to longer wavelengths and emit lower fluorescence intensities with increasing sizes of aromatic ring systems. Zander and Haenel40 have found clear correlations between the positions of spectral peaks and the average molecular masses of coal-tar pitch fractions. For the samples concerned, this implies a direct relationship between aromatic ring system sizes and average molecular masses. More recent work supports these observed trends. For a number of coal tar pitch, bitumen, and crude oil derived solubility fractions, broad correlations have been noted between 3895
dx.doi.org/10.1021/cr200429v | Chem. Rev. 2012, 112, 3892−3923
Chemical Reviews
Review
wavelengths of the synchronous UV−F spectra presented in Figure 1b will be discussed in section 3.3. Spectral shifts toward longer wavelengths were also observed when the spectra of coal and petroleum derived fractions soluble in lighter solvents (e.g., acetone, acetonitrile) were compared with material insoluble in such solvents, but soluble in stronger solvents such as pyridine or NMP. For such samples, increasing molecular mass distributions (determined by SEC and LD-MS) and aromaticity (from NMR) tend to correlate with shifts to longer wavelengths observed by UV− F.2,12,15,41−48 Such shifts may also be termed as “bathochromic shifts” or “red-shifts”. As already noted, overlaps of UV−F bands arising from distinct structural features do not allow reading back distinct structural features from already acquired spectra. Keeping that caveat in mind, there is in fact considerable agreement in the literature that increasing sizes of aromatic chromophores lead to shifts of the maxima of UV-absorbance and UV−F spectra to longer wavelengths.13,52−54 Complex molecules such as porphyrins behave in much the same manner. The maximum intensity of absorbance55 for an aromatic system comprised of a Ni-porphyrin system fused to four anthracene groups with eight symmetrically substituted aryl O- pendant groups was observed at 1417 nm, well inside the infrared range. This compound did not fluoresce at all in the 250−800 nm range.55 In contrast, petroporphyrins56,57 are not so heavily substituted by aromatic groups and vanadyl and nickel porphyrins show a characteristic absorbance maximum at 408 nm. In another study, the UV−F spectra of pyrolysis tars from a set of rank ordered coals have shown good correlations with broadening size exclusion chromatograms.9 The coal rank parameter is generally related to the C-content of the original coal. With increasing coal rank (up to semianthracites), tar yields increased and the spectra of the tars shifted to longer wavelengths by UV−F and to larger molecular masses by SEC. Beyond semianthracites, however, only small amounts of lighter material in high rank coals are released from the solid matrix to form tar. Thus, an abrupt reversal of the trend was observed for semianthracites and anthracites, where the coals gave progressively lower tar yields, which appeared to have shifted toward smaller molecular masses by SEC and toward shorter wavelengths by UV−F spectroscopy. In the UV−F spectroscopy of heavy hydrocarbon liquids derived from coals, biomass materials and petroleum derived high-mass, polydisperse materials (oils, tars and pitches) shifts to longer wavelengths are nearly always accompanied by a drop in fluorescence intensity. For a Stockholm tar21 fractionated by planar chromatography, the material immobile in pyridine or THF (on the PC plate) emitted no fluorescence, although their structural makeups are clearly aromatic and the material clearly absorbs UV light. Similarly, an asphaltene sample,58 washed with repeated aliquots of NMP left an insoluble residue, but produced a series of NMP-soluble subfractions showing increasingly red-shifted spectra, with increasing molecular size. Similarly, comparing the fluorescence spectra of the maltene, asphaltene, and NMP-insoluble fractions with that of the overall crude oil showed steady progressive red-shifts from maltenes to NMP insolubles. A parallel progressive decrease in fluorescence intensity was also noted. As outlined previously, intermolecular (molecule-to-molecule) energy transfer in solution can be minimized by sequentially reducing bulk sample concentrations. Unlike individual molecules in solution, however, aromatic ring
average numbers of rings in polynuclear aromatic ring systems, determined by average structural parameter (ASP) calculations from NMR and LD-MS data, and the wavelengths of maximum fluorescence intensity in the UV−F spectra.2,41−49 Figure 1a shows synchronous UV−F spectra for toluene and the standard polycyclic aromatic hydrocarbon (PAH) perylene,
Figure 1. (a) Synchronous spectra of standard compounds and essential oils; toluene: 1 ring; perylene: 5 rings; components of essential oilstea tree and lavenderare mainly terpenes; coniferyl alcohol: 1 ring (data for essential oils from ref 50). (b) Synchronous UV-fluorescent spectra of PAH standards toluene (1 ring), naphthalene (2 rings), anthracene oil (2−5 rings), and Maya asphaltene fractions soluble (MNS) and insoluble (MNI) in NMP; CHCl3 was the solvent for Maya samples and NMP for anthracene oil.
the essential oils from tea tree and lavender50 as well as coniferyl alcohol containing a single ring. The chromophores in the essential oils are mainly terpenes. Only the spectrum for perylene showed a signal between 400 and 450 nm. The synchronous UV−F spectra of heavier fractions of coal tars and petroleum asphaltenes shown in Figure 1b and in section 7 normally present a signal well beyond these wavelengths, strongly suggesting that they contain larger (more highly conjugated aromatic) chromophores. Figure 1b compares UV−F spectra of an anthracene oil51 with Maya asphaltene fractions soluble and insoluble in NMP, as recorded, showing the reduced fluorescence intensity of the insoluble fraction. The relationship between average numbers of rings in these molecular systems and the ranges of 3896
dx.doi.org/10.1021/cr200429v | Chem. Rev. 2012, 112, 3892−3923
Chemical Reviews
Review
Figure 2. Peak normalized synchronous UV−F spectra of the Maya solubility-fractions dissolved in CHCl3. MM - Maya maltene fraction (heptanes soluble); MA - Maya asphaltene fraction (heptane insoluble, toluene soluble); MNS - NMP soluble fraction of the Maya asphaltene; MNI - NMP insoluble fraction of the Maya asphaltene. Reproduced with permission from ref 71. Copyright 2010 Elsevier B.V.
extracts of coal.63−66 Fused five-membered rings are usually more reactive and the expected shift of UV-absorbance and UV−F spectra to shorter wavelengths, occasioned by their destruction, is somewhat greater than the loss of six-membered rings.67 Hydrocracking the pyridine-insoluble fraction of a coal tar pitch68 was observed by SEC to produce smaller molecular mass material. The product mixture presented a fluorescence spectrum similar to that of the whole original pitch, with a maximum intensity at about 400 nm. By contrast the pyridineinsoluble fraction (prior to hydrocracking) showed weak fluorescence intensity with a broad maximum between 500 and 600 nm. These data are consistent with the cracking of large molecules to produce smaller ones. Recalling the high levels of dilution required for UV−F measurements and the earlier discussion on “aggregates”, these findings are consistent with a model involving the breakup of large molecular mass material.
systems embedded in large molecules are not distributed randomly, nor can distancesand interactionsbetween them be interfered with through dilution. Consequently, their local concentrations within large molecules can be high. The resulting intramolecular energy transfer is accompanied by a drop in frequency (i.e., drop in energy level) resulting in redshifting and dissipation of some of the energy as heat, which tends to reduce fluorescence intensity. An example of the influence of molecular size can be observed in comparison of the synchronous spectra of toluene and polystyrene; both molecules contain benzene groups attached to one alkyl substituent. The toluene spectrum is intense while the polystyrene spectrum shows the same peak at about 260 nm but at much lower intensity. UV−F spectroscopy is not without its complications. The NMP soluble fraction of a Maya (heavy Mexican crude) asphaltene sample has recently been discussed in terms of consisting predominantly of “archipelago” type structures.2 Solution state 13C NMR work on this sample suggests the presence, on average, of two separate polynuclear aromatic ring systems per molecule, with 5−6 conjugated rings in each. In UV−F spectroscopy, such molecules would be emitting in the region representing 5−6 rings, rather than 10−12 rings. On the other hand, “continental” type structures with 10 or more conjugated rings would be expected to emit at far longer wavelengths but would not be expected to give much intensity, as was found for the NMP insoluble fraction of a Mayan asphaltene,2 cf. Figures 1b and 2. During hydrocracking processes, where large molecules are broken up, UV−F spectra are typically shifted toward shorter wavelengths.59−61 Structural changes likely to cause such shifts include the loss of alkyl and alkoxy substituents and the loss of heteroatom containing aromatic substituent groups, with the phenolic−OH group possibly being the most abundant. Shifts of UV−F spectra to shorter wavelengths may also result from the degradation of linearly conformed polynuclear aromatic ring systems (e.g., pentacene to tetracene) and loss of formally fixed double bondsas in zethrenes.62 Similar shifts may be observed upon the loss of structures with fused five-membered rings (with no ring carbons available for substitution) such as fluoranthene, found in components of coal tar pitches and
3.3. Examining UV-Fluorescence Spectra
Figure 1a,b present synchronous UV−F spectra for a series of standard compounds, coal and petroleum derived samples and their subfractions. On the basis of ASP analyses (discussed in section 6), the data suggest that the maximum in fluorescence intensity shifts steadily to longer wavelengths, by about 30 nm per additional aromatic ring in a polycyclic aromatic system. The correlation shown in Figure 1b is derived from data on coal tar pitch, petroleum- and bitumen-derived solubility fractions. The UV−F spectra are presented in Figures 1b, 2, 3, 4, 5 and the number of conjugated aromatic rings are in Tables 1, 5, and 9. Additional data was drawn from refs 2, 41−48. Fluorescence intensities in the UV−F spectrum of the NMP insoluble fraction of the asphaltene were very low and showed a broad wavelength distribution, ranging from 350 to more than 650 nm. These findings suggest the presence of a wide range of chromophores in this fraction, some corresponding to polycyclic aromatic ring systems of more than 10 conjugated rings, on average. As outlined later, solution state 13C NMR and LD-MS data were used to refine the estimates of average sizes of the conjugated ring systems. 3897
dx.doi.org/10.1021/cr200429v | Chem. Rev. 2012, 112, 3892−3923
Chemical Reviews
Review
Figure 3a compares the height normalized synchronous spectrum of a (whole) coal tar pitch with those of the three fractions derived from it: (1) the acetone-soluble, (2) acetoneinsoluble but pyridine soluble, and (3) the pyridine insoluble. Figure 3b shows the same spectra as recorded, before the normalization process; this diagram shows the less intense fluorescence observed from progressively heavier fractions, as follows. The pyridine-soluble fraction showed significant fluorescence intensity compared to the pyridine-insoluble fraction. However, both fractions required significantly more sample in solution than proved necessary for the acetonesoluble fraction, or the whole pitch, to give the observed signal strengths. These spectra from ref 69 were previously shown in this form in ref 70. The differences observed in the sample set, between the acetone-soluble and pyridine-insoluble fractions, illustrate the level of structural differentiation that can be achieved through simple fractionation. The molecular masses of these fractions,11 examined by SEC with reference to standard polystyrene mass equivalents, show an analogous progression. Comparing the maximum intensities of the retained peaks in SEC chromatograms suggests that the whole pitch and the acetone-soluble fraction consist of relatively small molecules, mainly of a few hundred mass units. The acetone-insoluble/pyridine soluble fraction showed larger molecular masses and a wider mass range. Meanwhile, the pyridine-insoluble fraction corresponded to an average mass equivalent of over 1000 mass units. These differences in mass distribution were confirmed by LD-MS.12,42 Figure 1b shows a similar trend for the NMP-soluble and -insoluble fractions of the Maya asphaltene.2,71 It may be noted that the examination of NMP-insoluble fractions was carried out in chloroform solution; sample concentrations were increased by adding sample, in small steps, until the fluorescence signal appeared sufficiently above background to give a weak spectrum although still relatively noisy.58,71−73 In Figure 3a, the intensity maximum in the spectrum of the pyridine insoluble fraction (curve 4) may be observed to have shifted toward longer wavelengths, compared to the other spectra. This appears as a common feature for fractions containing larger molecules (as determined by SEC and/or LDMS) and also larger conjugated aromatic ring systems as determined by NMR [e.g., cf. 2, 40, 41, 42]. Finally, it may be observed that the spectrum of the “whole” pitch sample (Curve 1 in Figure 3a) did not show a signal indicating the presence of material detected in the pyridine-insoluble fraction. The characteristics of the bulk sample (Curve 1) were in fact closer to those of the spectrum of the acetone-soluble fraction. As has been systematically observed in examining sets of petroleum and bitumen asphaltenes,2,12,42−44,71 the spectrum of the whole pitch (Figure 3a) appears to reflect properties of the apparently more abundant lighter fraction and to mask the features of the less abundant heavier fractions. The three fractions were approximately of similar proportions (∼33% of the pitch), but the components of the lightest fraction had the strongest fluorescence and thus apparently contained the highest proportion of the more intensely fluorescing material. UV−F may thus be considered a useful technique in comparing sample solutions, for broad indications of greater/ lesser presence (concentration) of larger/smaller polycyclic aromatic groups. However, as outlined in the next subsection, comparison with other techniques has shown strong evidence that UV−F spectroscopy is of limited utility in studying
Figure 3. (a) Normalized synchronous UV−F spectra of (1) the coal tar pitch, (2) fraction soluble in acetone, (3) insolubles in acetone soluble in pyridine, (4) fraction insoluble in acetone and insoluble in pyridine. These spectra are from the data of ref 69. Reproduced with permission from ref 69. Copyright 2005 John Wiley & Sons Ltd. (b) Pitch and fractions synchronous UV−F spectra as determined, not normalized; copyright permission and numbering as in Figure 3a.
Figure 4. Syncrude synchronous UV−F spectra of maltene (1), syncrude (2), and asphaltene (3). Previously shown in ref 44. From Preprints ACS Fuel Div., Natl. Meeting Salt Lake City 2009. Copyright 2009 American Chemical Society.
Many pure PAH compounds deviate from this correlation such as perylene (5 rings), which is shown in Figure 1a and has a peak maximum at 440 nm that equates to 6−7 conjugated rings according to the correlation in Figure 1b. However, for complex polydisperse PAH mixtures the correlation provides a useful estimate of the extent of conjugation, according to the available data. 3898
dx.doi.org/10.1021/cr200429v | Chem. Rev. 2012, 112, 3892−3923
Chemical Reviews
Review
Figure 5. Coal tar pitch fractions soluble in heptane, soluble in toluene but insoluble in heptane, and insoluble in toluene. (A) SEC chromatograms in (a) NMP and (b) NMP:CHCl3 eluents on mixed A column and (B) UV−F in NMP solution. From ref 143. UV−F data not previously shown. Panel (a) reproduced with permission. Copyright 2009 John Wiley & Sons Ltd.
Table 1. Mass Values for Pitch Fractions Derived from LD-MS Spectraa fraction
low mass limit/m/z
most intense mass/m/z
upper mass limit/m/z
Mn/ m/z
peak max/nm
number fused Ar-rings
heptane solubles toluene solubles toluene insolubles acetone soluble acetone insoluble − pyridine soluble pyridine insoluble
200 200 400 200 300 500
300 400−500 2000 350 1200 1800
1000−2000 2000−3000 10000 2500 >4000 >10000
n/a n/a n/a 300 1600 >2500
390 390 480 390 440 480
n/a n/a n/a 3−5 7−9 >30
a
Data for heptane and toluene fractions reproduced with permission from ref 143. Copyright 2009 John Wiley & Sons Ltd. NMR data from refs 41, 42; LD-MS and Mn values from ref 42; method for mass estimates is outlined in ref 71; n/a = data not available.
In UV−F, solution concentrations are required to be low [order of 0.00001 g L−1] for strongly fluorescent standards such as toluene, in order to avoid self-absorbance related effects. This solution strength is normally barely sufficient to give a good response on the UV-absorbance detector, but adequate to show the absence of a significant change in the SEC chromatogram with increased dilution. The resulting chromatograms were analyzed after correcting for the time lag for solvent flow between the two detectors connected in series. At short elution times, the UV-absorbance
materials above a threshold of about 1500 u and is entirely blind to material above 2800−3000 u. 3.4. High Mass Limits to UV-Fluorescence
In earlier work, a UV-absorbance and UV−F spectrometer were placed in series and used as detectors for recording the retention times of samples eluting from an SEC column.74 Initially, the study was undertaken to explore the possibilities of detection by UV−F in the size exclusion chromatography of widely polydisperse samples with suspected high mass content. 3899
dx.doi.org/10.1021/cr200429v | Chem. Rev. 2012, 112, 3892−3923
Chemical Reviews
Review
for the low molecular mass materials and least for the high mass material. These workers concluded that fluorescence-based studies were unsuitable for molecular mass measurements in such samples. In subsequent work,13,14 Strausz and co-workers have considered the mechanism of excitation of aromatic materials by photons in LD-MS work, to yield either fluorescence or ionization. The conclusion was that the fluorescence intensity would decrease as molecular mass increased. Concurrently the ionization of large molecules would become less efficient with increasing molecular mass. Further references to this study will be made in the following sections. Ascanius and co-workers78 have also found that NMP insoluble fractions of petroleum asphaltenes (approximately 9− 53% of the total) “...hardly exhibits any ultraviolet−visible light absorption or fluorescence.” In one part of their work,78 NMPinsoluble fractions were examined by UV−F in toluene solution at about 10 mg/L concentration, similar to concentrations of NMP-soluble material recovered from the SEC outlet. The NMP-insoluble fraction of these asphaltenes gave no significant UV−F signal in toluene solution. They concluded that a substantial proportion of the asphaltenes was not represented in the fluorescence spectrum, indicating significant limitations in the analysis of asphaltenes by UV−F spectroscopy. Mullins and co-workers79,80 have disagreed with the conclusion from evidence that shows fluorescence methods are not able to detect the entire mass range of asphaltene molecules. Considerations of optical spectroscopy of polycyclic aromatic hydrocarbons using molecular orbital theory suggest there is difficulty in proposing molecular structures able to reproduce the UV, visible, and near-infrared spectra of asphaltenes.81−83 However, this theoretical work overlooks the experimental evidence that some asphaltene molecules do not fluoresce or have low absorbance of UV light.13,14,77,78 There remains the fact that several independent laboratories have reported that larger molecular mass materials have been observed not to emit fluorescence. Equating the inability of large mass species to emit UV−F with their absence seems to require some revision. Although the conviction seems firmly held, it does not seem possible to avoid the evidence and accept the interpretation.84 Meanwhile, it still seems difficult to speculate about what the structures of large molecules present in asphaltenes might be. The simple addition of arene groups to known smaller molecular structures does not appear to give internally consistent results. Once again, the preferred route in examining complex sample mixtures appears to pass through sample fractionation. Examining f ractions of crude oil derived samples would have pinpointed whether and how some of the techniques used are capable, or indeed incapable, of showing a signal for palpably existing sample fractions.
detector gave signals corresponding to material coming through at the exclusion limit of the column. By contrast, little or no signal could be observed from the UV−F cell at short elution times, corresponding to larger molecular mass material. However, at longer elution times corresponding to smaller molecular masses, broad correspondence could be observed between the two detectors. Similar experiments were performed during subsequent studies, to explore this apparent limitation of detection by UV−F.10 A more recent set of such experiments11 was undertaken after fluorescence depolarization experiments were claimed to have shown far lower molecular mass distributions in petroleum asphaltenes,75 than have been reported using SEC and MALDI-MS (see later). Taken together, comparison of signal from the UVabsorption and UV−F detectors showed that fluorescence intensities fall off rapidly above 1000−1500 u, decaying to nearzero above 2800−3000 u. The observation is thought to be secure, since up to about 3000 u, there is numerical agreement between MALDI-mass spectrometry and SEC, calibrated by using polystyrene and polymethylmethacrylate (PMMA) molecular mass standards.12,76 Figure 6 shows chromatograms acquired using the two detectors in series, during an SEC run with NMP as eluent.73
Figure 6. Comparison of detectors (UV−F and UV−A) in series during SEC in NMP eluent of an asphaltene. Vacuum bottoms C by SEC with UV−A and UV−F detectors. Reproduced with permission from ref 73. Copyright 2005 Elsevier B.V.
The two spectrometric methods showed a broadly similar signal for the smaller mass material eluting at longer times (after 18 min). The two peaks were separated by about 1.5 min, suggesting that the UV-absorption detector was somewhat more sensitive to larger mass material within this mass range. However, there were radical differences between the two curves between 12−16 min. UV-absorbance showed a large peak for material excluded from the SEC column porosity, while little signal was observed from the UV−F detector. The latter clearly underestimated the presence of larger mass material eluting in the 12−16 min range. Comparing with colloidal silica standards, this elution time corresponds to a 10 nm diameter particle. During LD-MS examination of subfractions of these samples (collected at the outlet of the SEC apparatus), materials eluting over the 12−16 min time range (in the “excluded zone”see later) showed average masses (m/z) of ∼1500−3000 by LD-MS. There was little evidence12,42 for ions with mass (m/z) greater than 10 000. During an independently conceived study, Strausz et al.77 reported that fluorescence of asphaltene fractions was strongest
4. EXAMINING HIGH MASS FRACTIONS BY SIZE EXCLUSION CHROMATOGRAPHY (SEC) Gas chromatography (GC) and GC-MS are preferred techniques for identifying and quantifying aromatic compounds up to molecular masses of about 300 u and predominantly aliphatic material85 up to about 500 u. In characterizing samples with higher masses and more limited volatilities, however, liquid-phase separations such as high performance liquid chromatography (HPLC) tend to be more useful. Both normal and reverse phase HPLC of polycyclic aromatic hydrocarbon compounds (PAH) are well understood,86−88 particularly in 3900
dx.doi.org/10.1021/cr200429v | Chem. Rev. 2012, 112, 3892−3923
Chemical Reviews
Review
relation to the influence of solute shape and configuration on retention times. However, HPLC analyses are limited by the lack of solvents able to elute large PAH molecules with molecular masses much above 500 u from a bonded silicaparticle column.89,90 The limited availability of well-defined standard compounds over a wide range of masses has also tended to restrict the amount of information that can be obtained from HPLC studies above the 500 u level. In size exclusion chromatography (SEC), analyte molecules are not separated by partition according to chemical properties. Instead, samples are injected into columns packed with crosslinked polymer gels. The degree of penetration of molecules of different sizes within the porous structure of the packing then determines the retention time distribution of sample molecules. SEC has found wide application in the examination of diverse types of samples.12,15,23,28,44,45,58,71,72,91−95 These include heavy coal liquids, soot, bitumen and petroleum asphaltenes and their related residual fractions,12,15,23,28,44,58,71,72,92 tars from the thermochemical treatment of other solid fuels such as biomass or waste45,93,94 as well as geochemically interesting samples such as contaminated soil95 and amber extracts.91 Catalytic hydrotreatment of coal tars and asphaltenes have also been examined and, as would be expected, showed shifts to smaller molecules.61,96−99 Several reviews are available for applications of SEC to the characterization of high-mass, polydisperse hydrocarbon mixtures.7,8,100,101
polysaccharides). The elution times of the PAHs showed the opposite of the trend expected from SEC: the smaller PAHs eluted earlier and the largest later. This is strong evidence of the larger PAHs interacting with the column packing material more strongly and eluting at longer times than expected from their mass. Similar trends were observed when chloroform was used as eluent in analyzing the same set of standard PAH compounds. However, when NMP was used as eluent the correct trend was observed for the PAHs and polymer standards.42,72,104,109 No such problems were reported when a coal tar pitch and a petroleum vacuum residue were fractionated by planar chromatography, sequentially using pyridine, acetonitrile, toluene and pentane to develop the chromatograms. The bands of separated material were recovered in NMP and examined by size-exclusion chromatography (SEC), with NMP as eluent. These chromatograms were observed to shift to earlier elution times as fractions with progressively less mobility on the plate were examined.22 The same correlations were also observed for coal tar derived creosote and anthracene oils when fractionated by planar chromatography and examined by SEC with NMP as eluent; the SEC mass estimates were confirmed by LD-MS analyses.45 In general, decreasing mobility in planar chromatography is associated with both increasing molecular mass and with increasing polarity. The two effects cannot be differentiated unless observations are supplemented by other analytical tests. The difficulties encountered in using THF as eluent in SEC are compounded by the reactivity of pure THF, which tends to polymerize on contact with air during storage. Not only does this contaminate the solvent, but the mixture becomes potentially explosive. This is normally countered by mixing proprietary polymeric additives in with the THF, which serve to block polymerization reactions during storage. Either way, the use of THF in analytical work leads to complications, involving the presence of high mass material of indeterminate composition. The case for abandoning the use of THF as a solvent and as an eluent in the SEC and general characterization of high-mass, polydisperse, and often highly polar and aromatic materials is thus quite compelling. Pyridine appears as a likely contender and has been occasionally used in the analysis of complex, polydisperse samples, as both solvent and eluent. However, pyridine is observed not to dissolve coal derived tar and pitch samples completely. Furthermore, its vapor pressure at room temperature is quite high and its strong smell easily recognizable. In other work, Behrouzi and Luckham110 have proposed “a proof” of the unsuitability of SEC in examining the molecularmass distributions of petroleum asphaltenes. The claim was based on two propositions: first, that samples aggregate during SEC in three solvents tested [THF, toluene, and NMP] and, second, that adsorption phenomena take place between sample and column packing, in the presence of NMP. At first glance, a generalized criticism of the SEC method on account of the imputed failure of one or more solvents appeared as a rather sweeping statement. Had the claim been correct, the conclusion to draw would have had to be about the unsuitability of the particular solvents used, rather than that of the method as a whole. In any case, bringing toluene and THF into the discussion was spurious, then as now. In the event, no evidence was presented to show sample aggregation in dilute NMP solution, while the SEC work reported included several elementary mistakes. Despite the
4.1. Limitations of Using Tetrahydrofuran As Eluent in SEC
In early characterization work on coal and petroleum derived heavy fractions by SEC, tetrahydrofuran (THF) has been widely used as the preferred eluent [e.g., cf. ref 102−106]. The limitations of THF as solvent and eluent have since become apparent. Several cases outlined later show some of its drawbacks. In one case, fractions of a coal tar pitch separated on and recovered from a planar chromatographic plate were examined by SEC using THF as the eluent.107 Different fractions which had showed increasing mobility on the plate showed no indication of shifting to longer or shorter SEC-elution times. The range of elution times for the fractions broadened out, but all fractions showed similar elution times. Furthermore, using the same SEC setup (with THF as the eluent), much of the pitch sample was observed to elute later than the permeation limit of the column, as defined by polystyrene standards. These observations were interpreted as indicating the presence of adsorptive interactions between analyte molecules and column packing, due to the insufficient solvent power of THF. Similar problems have been reported from work on an Athabasca bitumen sample,108 which was fractionated into five fractions, by preparative SEC using THF as the eluent. The fractions were examined by analytical SEC, using NMP as the eluent, a far stronger solvent for predominantly aromatic materials. The chromatograms acquired in NMP eluent of the first two fractions (prepared by preparative SEC in THF) showed only small differences; the later eluting three fractions were found to give nearly identical chromatograms. It was concluded that the use of THF as eluent in the preparative SEC step had failed to separate the sample mixture by size exclusion. Once again, these results suggested that THF was unable to break-up analyte-column packing interactions. In other work, the elution times of a number of small standard PAH molecules were compared with those from polymer standards (polystyrenes, polymethylmethacrylates, and 3901
dx.doi.org/10.1021/cr200429v | Chem. Rev. 2012, 112, 3892−3923
Chemical Reviews
Review
performance117 before construing the effect in terms of sample dissociation. The view in much of the oil industry120−122 seems to be that apparently large molecules (say, >1500 or 2000 u) observed under the great dilution levels prevalent in analytical work are essentially “aggregates” of far smaller molecules. Apart from the unsuccessful attempt outlined previously, however, no systematic work has been reported in the literature and no clear evidence has been presented to support what in effect may be described as a legitimate working hypothesis.
claim that UV-absorbance was used for detection, in all cases, negative peaks in the chromatograms clearly indicated the use of a refractive index detector rather than the elution of “compounds devoid of UV absorbance” as stated in the text. These authors presented SEC chromatograms in solvents that were known to have shortcomings, applied erroneous calibrations to produce Mw values from them, and, because the Mw values calculated were not similar, concluded that SEC is not a reliable technique for the analysis of petroleum asphaltenes. Finally the atomic force microscope experiments reported in the paper110 did not show adsorption of asphaltenes on the SEC column packing material under the conditions used for analyzing these samples during SEC. The observed interaction forces were between particles of solid asphaltenes and dry column packing. The experiments did not detect interactions between the column packing and a solution of asphaltenes in NMP. This suggests that adsorption phenomena in SEC are avoided if the right eluent is chosen, that is, the opposite of what the authors claimed to have demonstrated. Criticism of this work in the open literature111 has failed to elicit any response by the authors.
4.3. How To Explain the “Excluded Peak”?
Concerted efforts have been made to isolate and examine the molecular mass distributions of the material that elutes in the excluded SEC region.12,15,42,76 As with other aspects of characterizing complex polydisperse hydrocarbons, it appears difficult to arrive at definitive conclusions from this work. The working principle of SEC relies on molecular separation by size. So long as there is some sort of correlation between molecular mass and molecular size, the technique may be used, with the aid of appropriate calibrations for determining molecular masses. However, the chain of information for estimating molecular masses is known to fail where samples such as soots, silica particles, and fullerenes are concerned. In the case of such samples, the column works entirely on the basis of molecular or particle size, in a manner entirely disconnected from molecular masses.7,123 According to existing polystyrene-PMMA calibrations,72,123 probably unrealistically large masses (105−106 u) would need to be assigned to material eluting in the excluded zone. On the other hand, bimodal distributions of SEC chromatograms may be interpreted in a way that is consistent with the short elution times observed for three-dimensional samples such as fullerenes and silica particles.7,123 One tentative explanation would have sample molecules above a certain molecular mass threshold adopt threedimensional conformations (analogous to the behavior of fullerenes), where hydro-dynamic volume no longer relates to molecular mass. It appears tempting to consider that threshold as being close to the front, that is, higher mass, edge of the resolved (longer elution time) peak, that is, the upper mass end of the second peak. What is proposed, as a working hypothesis, is a change of conformation of molecules above a certain molecular mass/size threshold from planar or worm-like conformations (able to penetrate the porosity of the SEC column) to three-dimensional structures that would result in a far larger hydrodynamic volume (i.e., size in solution) of the molecule. Estimates of molecular masses for the leading edge of the resolved peak have differed between samples and between techniques, from an upper limit of 5000−8000 u (by SEC, of the pyridine insoluble fraction of a coal tar pitch) down to about 1500−2000 (by LD−MS, of a petroleum asphaltene). The change of conformation for molecules larger than those identified at the leading edge (highest mass point) of the resolved peak may be visualized as allowing the already quite large molecules to adopt shapes (and sizes) that would exclude them from column porosity. The common example for this type of behavior is the elution of fullerenes in SEC columns.72 These materials appear at far shorter elution times than their known molecular masses would warrant, giving far larger apparent molecular masses when evaluated by the PS-PMMA calibration.
4.2. Limitations of Using NMP as Eluent in SEC
NMP dissolves nearly all coal derived heavy hydrocarbon mixtures completely.41,112 Its wider use as a suitable solvent and eluent has followed on from the work of Lafleur and Nakagawa.113 As already signaled, atomic force microscopy experiments have shown interactions between column packing and solute in NMP solution to be negligible.110,111 NMP is also a more powerful solvent than THF. When an SEC guard column, used for characterizing pitch fractions with THF as eluent was later washed through with NMP, a dark, concentrated solution of material soluble in NMP was reportedly recovered.101 When this THF-insoluble material was examined by SEC with NMP as eluent, signal was observed at markedly shorter elution times compared to THF-soluble material from the same pitch sample. Similar findings have been reported in work relating to the characterization of a coal liquefaction extract.10 When relatively high mass fractions are examined by SEC with NMP used as eluent, two separate peaks are observed. The early eluting peak is thought to show signal for material excluded from column porosity, eluting through interparticle voids within the column packing. Meanwhile, the later eluting peak corresponds to material resolved by permeation of column porosity.78,113 The reason for the appearance of two peaks is not yet clear. Lafleur and Nakagawa113 attempted to examine the material showing signal under the excluded peak by heatedprobe MS; they detected only small ions ( m/z > 2000), limits the detailed compositional analysis to the center of the molecular-weight distribution”. This statement may indicate that more than one experiment is required to detect all of the sample (suggested by a reviewer). As a result, compositional trends highlighted were limited. The changes in compound classes with reaction were discussed in terms of aggregated molecules not being observed by the mass spectrometer before reaction but being observable in the products after disaggregation; this statement is a recognition that the FT-ICR-MS system does not detect the full mass range of asphaltene samples, and a suitable test would be the examination of sequential solvent soluble fractions as in the SEC and LD-MS sections of this review. A one-step analysis is therefore insufficient for such complex samples as shown in this review. 5.7. Analysis of Complex, Polydisperse Samples by Laser Desorption/Ionization Mass Spectrometry (LDTOFMS)
Several advances have taken place since the publication of a wide-ranging review7 summarizing applications of laserdesorption based mass spectrometric methods to complex, polydisperse samples. A more recent review of mass spectrometric methods is in press.149 Preliminary work indicated that mass spectra were able to reveal differences between coal derived liquids, prepared through different processes. In this context, liquefaction extracts of coals were observed to present much broader mass ranges than pyrolysis tars from the same coals.149 Much of the earlier work has focused on attempts to define upper mass limits for both coal-derived tars and extracts and petroleum derived fractions, such as maltenes and asphaltenes. Refocusing efforts on determining average molecular masses of distinct fractions of samples has tended to be more informative. As already explained, samples of wide polydispersity (Mw/Mn > 1.1) tend to suffer from mass discrimination effects, whereby the more abundant fractions (usually made up of smaller molecules) can be ionized and detected more easily than less abundant components. In complex, polydisperse samples, the 3908
dx.doi.org/10.1021/cr200429v | Chem. Rev. 2012, 112, 3892−3923
Chemical Reviews
Review
possible to observe molecular ions. Solution state 13C NMR suggested (see later) that these samples contained material with multiple aromatic chromophores, with 2−4 fused rings in each, connected by long aliphatic bridges. SEC derived estimates suggest that these materials have masses between approximately 500 and 1500 u, and could be clearly identified as bands of yellow to brown material on the PC plate. It is also not possible to observe the long chain alkanes present in most maltenes and possibly in some asphaltene fractions by LD-MS using a UVlaser, since alkanes have no UV absorbance. Some recent work done by varying laser power and ion intensity to define conditions when fragmentation and cluster ion formation begin suggests that LD-MS might in fact be used to generate mass spectra for maltenes and asphaltenes that reflect relative differences in their mass ranges, without further fractionation by PC. The evidence for this is shown in the discussion on Maya and Syncrude samples, sections 7.2 and 7.3, respectively.
less abundant material tends to be of larger molecular mass, which fail to ionize, or give ion intensities well below levels expected from their molar concentrations within the samples. Fragmentation commonly occurs when laser power is ramped to improve levels of ionization of larger molecules,45,71 while formation of gas phase aggregates from small molecules can erroneously indicate the apparent presence of very high mass materials.45,120,150 Once again, such mass discrimination effects can be at least partially side-stepped by fractionating complex samples, into subfractions with narrower ranges of molecular masses, that is, with reduced sample polydispersity.45,71 There is also evidence that the method of applying sample to the mass spectrometer target plate prior to shining the laser can influence the degree of mass discrimination. Hortal et al.150 suggested that a very dilute solution of the sample should be applied and the solvent evaporated, as for single compounds; this procedure reduces the probability of forming ions consisting of aggregates of molecules. However, our work45,70,71 suggests this procedure increases discrimination against the high molecular mass components. A better procedure is to apply a thicker layer of sample and closely control the laser power, including comparison of the spectra with those of fractions derived from the whole sample. Knockenmuss151 noted that laser desorption from a structured surface such as silica may reduce the formation of secondary reactions leading to aggregate ions in the reactive plume compared with desorption from a planar (smooth) metal surface. Later, we will describe experiments where after developing the chromatogram, laser desorption was carried out directly from the silica layers of planar chromatography plates. Planar chromatography plates with flexible backing were reported to be useful in transferring involatile fractions of sample into the mass spectrometer, with desorption taking place directly from the plate surface. Using these methods, the low-mass fractions of petroleum derived asphaltenes and “whole”, that is, unfractionated, asphaltenes were reported to give mass spectra with maximum ion intensities below m/z 1000; signal for high mass material extended to several thousand mass units (m/z). It was also reported that the relatively high-mass fractions of asphaltenes can show maximum ion intensities of several thousand mass units (m/ z) with upper mass limits extending to m/z 10 000 and possibly beyond.2,12,15,45,71 The estimation of molecular masses by LD-TOF-MS with samples introduced using fragments of planar chromatographic plates presents several shortcomings. When fractionation is carried out by solvent solubility or by column chromatography, the relative proportions of fractions separated from a “whole” sample can be estimated with reasonable accuracy. This cannot be done easily when the separation is done by planar chromatography. Furthermore, it is not possible to determine whether the material ionized by the laser was representative of the sample fraction used. Because of the complexity of sample types studied, it is assumed that the mass spectra generated nearly represent the complete range of chemical species present in the fraction. There is also evidence from recent studies that certain types of alkyl-aromatics cannot be observed by LD-MS.42,44,71 During work on the planar chromatographic fractions of asphaltenes from a Maya crude and a heavy fraction derived from the Athabasca tar sands, only fragment ions with masses (m/z) < 200 could be observed from certain PC fractions. It was not
5.8. Upper Mass Detection Limits of LD-MS Systems
There is evidence suggesting that upper mass (m/z) limits of the LD-MS technique are being reached, during the examination of complex samples of broad molecular mass distributions. In earlier work, traces of signal suggesting molecular masses extending up to 100 000 u had been observed.152 A number of statistical tests were applied to distinguish between sample derived signal and instrument noise. However, a broadly common upper mass limit of about m/z 10 000 was observed, when the work was repeated in more modern LD-MS spectrometers. A very wide range of samples displayed this apparent upper limit, from crude oil, coal and bitumen derived heavy fractions, subfractions of pitches and asphaltenes, through to pyrolysis tars and mesophases.2,8,12,15,42−44,46−48,71,143 Considering the widely varying origins and thermal histories of this range of samples, it is difficult to arrive at an explanation of an apparently common upper mass ceiling. Provisionally, it is thought this may be due to limitations of the method rather than a common of feature of samples from such diverse origins. Such an upper limit appears consistent with an independent theoretical study by Strausz et al.13 on the UV−F and ionization of polynuclear aromatic hydrocarbons. These researchers have proposed a theoretical upper size limit for aromatic chromophores that can be ionized under UV-light. Further work clearly seems necessary to examine whether and how this effect corresponds to the upper mass limits observed in recent LD-MS work. A two-step laser desorption technique has also been investigated with a view to avoid significant mass discrimination effects.120,153 The technique has been applied to asphaltenes and hydrocarbon mixtures. However, results to date do not provide evidence that larger mass species may be desorbed and ionized by this method. For example,120 four asphaltenes from geographically different crudes gave very similar mass spectra by this method although there seems no obvious reason why this should be so. Possible lack of mass discrimination could be determined by fractionation of involatile complex mixtures to demonstrate that the spectrum of the whole sample reflects the mass ranges of all the fractions. The limitations of absorbance of UV wavelengths with increasing molecular mass would apply to the ionization laser, but the near IR laser pulse used for desorption has not been shown to desorb all, or even a 3909
dx.doi.org/10.1021/cr200429v | Chem. Rev. 2012, 112, 3892−3923
Chemical Reviews
Review
Figure 9. 125 MHz 13carbon (inverse gated) NMR spectrum of Maya fractions; integration areas and ppm labels for TKS and TCE-d2 are shown. Reproduced with permission from Supporting Information from ref 2. Copyright 2010 American Chemical Society. (a) The maltene fraction (MM) in TCE-d2 solvent; acquired with a delay time (D1) of 2.5 s, 10 240 scans were summed. (b) the Maya asphaltene fraction (MA) in TCE-d2 solvent; acquired with a delay time (D1) of 2.5 s, 5 120 scans were summed. (c) the Maya asphaltene NMP soluble fraction (MNS) in TCE-d2 solvent; acquired with a delay time (D1) of 2.5 s, 10 240 scans were summed. (d) The Maya asphaltene NMP insoluble fraction (MNI) in TCE-d2 solvent; acquired with a delay time (D1) of 2.5 s, 5 120 scans were summed.
groups, introducing errors that are significant when it is intended to use the NMR data quantitatively.41,42 The present review will focus on solution state NMR methods suitable for obtaining quantitative structural information on aromatic carbon and hydrogen content of complex hydrocarbon mixtures as well as on the proportions of the different types of aromatic and aliphatic groups. The NMR pulse sequences of choice for the analysis of complex hydrocarbon mixtures are standard proton-NMR conditions (1H), 13C inverse gated (IG),156 distortionless enhancement by polarization transfer (DEPT), and quaternary only spectroscopy (QUAT)157,158 Although the latter two methods are not quantitative, they can provide useful addition structural information and confirmation of chemical shift classifications.41,156−158 Figure 9 shows the 125 MHz 13carbon (inverse gated) NMR spectrum of the Maya fractions; integration areas and ppm labels for TKS and TCE-d2 are shown. By combining information from these pulse sequences, the following empirical parameters may be determined:157−161 (1) Proportion of aromatic hydrogen (Har) (2) Proportion of aliphatic hydrogen (Hali), further subdivided into:
representative fraction, of an asphaltene. The claim that the method works is based on the coincidence of molecular mass ranges from this method with those from UV−F and ESI-MS, where doubts have been expressed (as discussed previously) regarding the ability of these techniques to detect all of a polydisperse sample in a single analytical step.
6. EXAMINING HIGHER MASS FRACTIONS BY SOLUTION STATE 13C NMR The versatility of NMR spectroscopy allows a wide range of different pulse programs to be applied to examine specific aspects of coal and petroleum derived materials. Numerous reports in the literature have described NMR methods capable of investigating internal and peripheral structural features in complex molecules and their mixtures.154−170 Solid-state NMR often seems a useful analytical route for tackling heavy coal and petroleum derived fractions, as it avoids the need for a solvent. However, it is often found that the resolution of signal is not sufficiently high for identifying detailed structural differences between samples.8,154,155 Moreover, instrument artifacts known as “spinning side bands” often overlap with signal from the reference material or aliphatic 3910
dx.doi.org/10.1021/cr200429v | Chem. Rev. 2012, 112, 3892−3923
Chemical Reviews
Review
contents of up to about 7% (w/w) are possible. For samples with high heteroatom contents, such as bitumen tar asphaltenes (9 wt %), the second approach tends to be more realistic. ASP calculations allow the following parameters to be calculated:2,41,156,167 • Fraction of quaternary aromatic carbon (Car-Q) • Fraction of protonated aromatic carbon (Car-US) • Fraction of peri-condensed aromatic carbon (Car-PC) • Fraction of cata-condensed aromatic carbon (CCata) • Fraction of quaternary aromatic carbon substituted by alkyl groups (Car‑S) • Fraction of aromatic carbons substituted by aromatic groups (Car‑AS) • Average number of aromatic rings per average molecule (RA) • Average number of naphthenic rings per average molecule (RN) Clearly, information derived from ASP methods only provide averaged bulk parameters. Nevertheless, useful information can be determined such as the average size of fused aromatic ring systems and extents of substitution, among others.2,41,156,159,160,167 Fractionation of the “whole” sample helps in inferring more detailed structural information. This type of quantitative structural information has proved virtually impossible to gain by any technique other than by the use of NMR.2,7,8,41 In general, most attempts to study the heavier fractions of coal and petroleum derived materials prove to be problematic, in particular when quantitative data are required.126,160−165,172 This appears due to the use of solvents of insufficient strength to dissolve the whole sample, the inadequate choice of experimental conditions, or a combination of the two. Many of the earlier ASP investigations have only been able to gain useful information regarding lighter fractions, equivalent to acetone and chloroform soluble fractions of heavy coal derived samples. In studies examining petroleum asphaltenes, the sample is completely soluble in chloroform and is usually examined as a single bulk sample, that is, without fractionation into narrower fractions. As already signaled, this approach severely limits the amount of information that can be obtained from these analyses.2 Heavy coal tar pitch fractions that are insoluble in common NMR solvents may be dissolved by the use of (undeuterated) NMP as the NMR solvent. NMP is able to dissolve41 these samples in the high concentrations needed for NMR (>10% weight per volume). This approach tends to obscure important bands of the aliphatic part of the spectrum, through signal from the NMP itself. Thus, analysis in NMP solution provides information on only the aromatic part of the spectrum. A second analysis of the same sample is then required, by dissolving the sample in (undeuterated) quinoline. This allows quantitative data to be obtained from the aliphatic part of the spectrum. Using undeuterated solvents is useful for keeping the cost of the analyses within bounds. Coaxial NMR tubes may be used, so a second, more common, deuterated solvent could be used for locking the spectrometer, as well as holding the quantitative reference material. Details of the method may be found elsewhere.2,41 For petroleum-derived materials, solubility is not an issue, since they mostly tend to be completely soluble in common NMR solvents such as chloroform or tetrachloroethane. The latter is the preferred solvent.2 However, a key factor in
(i) methine hydrogen (CH) (ii) methene hydrogen (CH2) (iii) hydrogen in methyl groups (CH3) (3) Proportion of aromatic carbon (Car) (4) Proportion of aliphatic carbon (Cali), further subdivided into: (i) quaternary aliphatic carbon (Cali-Q) (ii) tertiary aliphatic carbon (CH) (iii) secondary aliphatic carbon (CH2) (iv) primary aliphatic carbon (CH3). In this review, particular attention has been paid to the solution state NMR analysis of coal tar pitches and heavy petroleum (crude) fractions.2,41,126,159−170 In practice, these methods are not straightforward to use for the types of materials being addressed and require considerable experience, time, and expense. Many attempts have been made to gain more detailed NMR structural information on coal and petroleum samples, such as aromaticity, aromatic fused rings configuration, and degree of alkyl substitution.156,160,166,167 The mathematical relationships enabling calculation of average structural parameters (ASPs) have gone through phases of evolution: first proposed by Bartle,169 the approach has evolved through the work of Dickinson156 and Rongbao.167 To calculate ASPs requires data from ultimate analysis, the appropriate NMR spectra, and estimates of average molecular masses (Mn). Within this framework, two distinct approaches will be summarized: (i) where only material containing carbon and hydrogen are considered,156,167 and (ii) where the role of heteroatoms is also incorporated.169,170 The choice of which ASP approach to use depends on the samples (among others, their heteroatom content) and the information ultimately expected from the calculations. A comparison of the two approaches for the maltene and asphaltene fractions of a heavy Mexican Maya crude oil, which has a relatively high heteroatom content, indicated that the information given by the two approaches were broadly similar.2 Information derived regarding alkyl chain lengths, protonated aromatic carbons, peri-condensed carbon, and peripheral carbons of aromatic clusters, were found to be similar.2 The main benefit of including heteroatoms is that a parameter termed “number of aromatic rings and rings containing ringjoining methylene groups (Ra + RJM)” can be determined.169 This includes certain heteroatoms and ring joining CH2 groups as well as aromatic carbon atoms. The resulting total ring size can be significantly greater than the total number of aromatic and naphthenic rings found when considering only carbon and hydrogen.2,47 Larger differences were found between the two approaches when the comparison was extended to a set of solubility fractions from a bitumen tar (tar-sand).44,47,171 This appears to be due to the higher heteroatom contents of these samples. As expected, more information could be inferred when heteroatom contents were included in the calculations. However, uncertainties introduced through the requirement of the additional experimental data tend to partly diminish the advantages of the approach. For coal derived materials, heteroatom contents are generally found to be relatively low ( 10 000 with a maximum intensity71 at m/z 2000. The information from LD-MS has been combined with NMR and ultimate analysis data.2 It was found that the NMP insoluble fraction (MNI) contained, on average, one large PAH system (>10 rings) surrounded by alkyl chains (n = 4.5, aromaticity = 54%). It may be recalled that NMP is a good solvent for aromatic materials but a poor solvent for predominantly aliphatic material. The position of the alkyl chains around the aromatic core appears to have rendered these materials insoluble in NMP. By contrast, the NMP soluble fraction (MNS) was found to be less aromatic (aromaticity: 50%) but on average contained two PAH groups (5−6 rings in each) connected via aliphatic bridges. This material is thus characterized by small aromatic cores connected by long alkyl-chains; the solvent appears to access the aromatic cores relatively easily to solvate these molecules. These findings on the sizes of PAH groups correlated well with the positions of the peak maxima in their UV−F spectra (Figure 1b). This combination of analytical approaches made it possible to determine two distinct structural types present in the Maya asphaltene, the NMP soluble portion having mainly archipelago structures and the NMP insoluble fraction containing mainly continental ones; the maltene had mainly “island” structures.2 Tables 5−8 compare mass values derived from the four analytical methods. As with the pitch fractions discussed previously, the values show a trend of increasing mass as solubility and/or mobility during PC separation decrease.
Figure 12. LD-mass spectra of pitch fractions showing the influence of increasing laser power (1 = lowest laser power and 6 the highest): (a) the heptane-soluble fraction, (b) the heptane-insoluble but toluenesoluble fraction, and (c) the toluene-insoluble fractions. Reproduced with permission from ref 143. Copyright 2009 John Wiley & Sons Ltd.
spectra are shown in Figure 2, the SEC chromatograms in Figure 7, and LD-MS spectra in Figure 13. Figure 14 shows Table 2. Mass Estimates from SEC and LD-MS for PC Mobility-Fractions (F1−F5) from the Pitch Acetone Soluble Fractiona SEC retained peak
LD-MS m/z
mass/u sample fraction fraction fraction fraction fraction a
1 2 3 4 5
peak max
lower limit
upper limit
peak max
lower limit
upper limit
tail
740 425 300 300 300
100 100 100 100 100
6400 3000 2700 2400 2400
800 600 n/a 270 250
250 200 n/a 170 180
4500 1600 n/a 1400 700
>20k ∼2500 n/a ∼2500 ∼1500
7.3. Examining Fractions of Synthetic Crude Prepared from the Athabasca Tar Sands
Analyses of a saturate fraction and the volatile components of an aromatic fraction of a similar sample174,175 derived from an Athabasca bitumen, using GC and field ionization mass spectrometry, have shown the presence of both multicyclic saturates up to m/z 750 in the saturate fraction and in the aromatic fraction, the molecular ions from 6,000 individual aromatics, in the mass range up to m/z 800. Clearly the tar
From ref 42. n/a − data not available. 3914
dx.doi.org/10.1021/cr200429v | Chem. Rev. 2012, 112, 3892−3923
Chemical Reviews
Review
Table 3. Mass Estimates from SEC and LD-MS for PC Mobility-Fractions (F1−F6) from the Pitch Pyridine Soluble Fractiona
sample fraction fraction fraction fraction fraction fraction a
1 2 3 4 5 6
SEC retained peak
LD-MS
mass/u
m/z
peak max
lower limit
upper limit
peak max
lower limit
upper limit
tail
1384 1417 1310 1211 409 409
100 100 287 287 100 227
22000 22000 22000 22000 4400 2700
1850 1000 1100 n/a 500 400
350 350 350 n/a 190 200
6000 2500 4000 n/a 3000 1000
∼20k ∼6000 ∼15000 n/a ∼7500 ∼2000
From ref 42. n/a - data not available.
Table 4. Mass Estimates from SEC and LD-MS for PC Mobility-Fractions (F1−F5) from the Pitch Pyridine Insoluble Fractiona SEC retained peak
LD-MS m/z
mass/u sample fraction fraction fraction fraction fraction a
1 2 3 4 5
peak max
lower limit
upper limit
peak max
lower limit
upper limit
tail
1870 1870 1380 1200 400
100 100 100 100 100
12000 12000 12000 12000 1400
1650 1500 1400 1400 1100
600 450 400 400 400
5000 4000 4750 4750 2000
>20000 ∼20000 ∼20000 ∼20000 ∼4000
From ref 42.
sample. The asphaltene component was barely apparent in the chromatogram of the “whole” sample. As with the Maya samples, the asphaltene showed the highest proportion of material excluded from the SEC column. Similar features were observed in Figure 16, showing the SEC chromatograms of the whole synthetic crude and four fractions separated by PC on silica plates. Fraction 1, the least mobile fraction, consisted almost entirely of material excluded from column porosity in SEC, while (at the other extreme) fraction 4 showing the narrowest molecular mass range did not extend to the same small molecule limit as for the original synthetic crude sample. The discrepancy is probably linked to the most volatile components of the crude sample evaporating with the solvent during drying of the PC plate after development. Figure 17 shows the LD-MS spectra of the “whole” synthetic crude and the four fractions separated by PC. The unfractionated synthetic crude did not give a mass spectrum that accurately reflected the mass spectra of the separate fractions. The mass spectrum of the whole crude ranged from m/z 200 to 5000 >7000 >7000 >10000
500−900 1100−1500 900−1100 1500−1700
390 405 405 490
2−5 8−10 5−6 >10
a
Table 6. Mass Estimates from SEC and LD-MS for PC Mobility-Fractions (F1−F5) from the Maya Maltene Sample (MM)a SEC - 300 nmb sample MM
LD-MS (m/z)c
PCfraction
peak max (u)
upper mass (u)
peak max
upper limit
tail
1 2 3
3750 1300 350
10,400 6,100 2,700
1250 920 615
3000 2600 1250
6000 6000 3000
a
Data from refs 42, 71. Reproduced with permission from ref 71. Copyright 2010 Elsevier. bThe SEC mass estimates are based on the method outlined in Supporting Information of ref 71 and only account for the retained SEC region; therefore these mass estimates do not accurately reflect their entire mass ranges. cThe LD-MS mass estimates are based on the method described in supplementary data of ref 71.
Data from refs 2, 42, 71.
These data44,47,176 have been presented at several conferences and more detailed papers are being prepared for publication. The NMR-based analytical approach outlined previously, which was applied to the analysis of the Maya crude samples (∼11.5% asphaltene content) to determine ASPs, was similarly used to examine analogous solubility fractions of the synthetic crude (∼13% asphaltene content). The maltene fraction of the synthetic crude was somewhat more aromatic (33%) compared to 31% for the Maya maltene. The difference of 2% is experimentally significant for this parameter as the experimental error is no more than ±0.5% of the absolute value. The NMP soluble fraction of the synthetic crude asphaltene was less aromatic than the equivalent Maya
fraction (42% vs 50%) as was the NMP insoluble fraction (45% vs 54%). It may be noted that the difference in aromaticity between the NMP soluble and insoluble fractions is 3−4% for both the petroleum and synthetic crude derived samples. The synthetic crude was found to have shorter aliphatic side chains than the Maya samples, about 2.2−2.5 carbons long, versus 3.5−4.6 for the Maya samples (quoting for NMP soluble and insoluble asphaltene fractions, respectively). The NMR, ASP, and UV−F data all showed evidence for the NMP-insoluble 3916
dx.doi.org/10.1021/cr200429v | Chem. Rev. 2012, 112, 3892−3923
Chemical Reviews
Review
Table 7. Mass Estimates from SEC and LD-MS for PC Mobility-Fractions (F1−F5) from the Maya Asphaltene Fraction Soluble in NMPc SEC - 300 nma sample MNS
LD-MS (m/z)b
PCfraction
peak max (u)
upper mass (u)
peak max
upper limit
tail
1 2 3 4 5
270 940 1,200 730 330
4500 5700 6600 4500 2500
1200 720 1000 850 760
3000 2500 3200 2400 1800
8000 7000 7000 5000 4000
a
The SEC mass estimates are based on the method outlined in Supporting Information of ref 71 and only account for the retained SEC region; therefore, these mass estimates do not accurately reflect their entire mass ranges. bThe LD-MS mass estimates are based on the method described in supplementary data of ref 71. cData from refs 42, 71. Reproduced with permission from ref 71. Copyright 2010 Elsevier.
Figure 15. SEC of Syncrude (1) and maltene (2) and asphaltene (3) fractions in NMP:CHCl3 eluent. Previously shown in ref 44.
Table 8. Mass Estimates from SEC and LD-MS for PC Mobility-Fractions (F1−F4) from the Maya Asphaltene Fraction Insoluble in NMPa SEC - 300 nmb sample MNI
observable when the unfractionated asphaltene samples were examined using the same analytical techniques. 7.4. Common Features of Results from the Three Sets of Samples
LD-MS (m/z)c
PCfraction
peak max (u)
upper mass (u)
peak max
upper limit
tail
1 2 3 4
590 590 300 400
4500 7700 1400 4500
1700 1600 1000 1400
7000 4600 2500 3400
>15000 >15000 >10000 10000
Fractionation of the samples produced subsamples with narrower bands of different (albeit overlapping) molecular mass and size ranges. Diminishing UV−F intensities were observed with increasing molecular masses of the fractions. A clearer correlation was found between UV−F spectra and sample structures, when the average size of the fused aromatic ring systems (PAH) was taken into account. The synchronous UV−F peak maximum shifted by about 30 nm with each additional aromatic ring in the fused (conjugated) system, as determined by ASP. This relationship appears to hold until the number of rings reaches about eight; for chromophores containing more than about eight fused rings, the fluorescence intensity rapidly decreased and no further red-shifting could be observed. No evidence could be found for the larger molecules being composed of aggregates of small molecules, at the levels of dilution used in SEC and UV−F spectroscopy. In fact size exclusion chromatograms were acquired at sample dilutions sufficient to allow the use of a fluorescence detector. Instead, at short elution times (large molecular masses) the presence of sample molecules could be observed by UV-absorption, while no signal was observed from the UV−F detector. This is consistent with diminishing fluorescence intensities observed for high mass material during static-cell UV−F measurements.
a
Data from refs 42, 71. Reproduced with permission from ref 71. Copyright 2010 Elsevier. bThe SEC mass estimates are based on the method outlined in Supporting Information of ref 71 and only account for the retained SEC region; therefore these mass estimates do not accurately reflect their entire mass ranges. cThe LD-MS mass estimates are based on the method described in supplementary data of ref 71.
fractions containing molecules with significantly larger PAH systems than the corresponding NMP-soluble fractions. In both the Maya and synthetic crude asphaltenes, it was the NMP insoluble fraction that was the most aromatic, contained the largest conjugated aromatic chromophores, and had the highest average mass (>1500 u, with masses up to ∼10 000 u). LD-MS analysis of their PC fractions revealed that within the synthetic crude asphaltene, the NMP-insoluble fraction contained materials with a broader mass (m/z) range than the equivalent sample from Maya crude oil. Once again, neither the presence nor the characteristics of these materials were
Table 9. Mass Value Data for the Synthetic Crude Fractions Derived from Three Analytical Methodsa LD-MS fraction maltene acetone soluble (SyMAcS) maltene acetone insoluble (SyMAcI) asphaltene NMP soluble (SyANMPS) asphaltene NMP insoluble (SyANMPI)
low mass limit/m/z
most intense mass/m/z
200
280
200
1000
200
350 (1500)
300
3000
upper mass limit/ m/z 3500−5000 >6000
b
>5000 >10000
fraction maltene acetone soluble (SyMAcS) maltene acetone insoluble (SyMAcI) asphaltene NMP soluble (SyANMPS) asphaltene NMP insoluble (SyANMPI)
UV−F
NMR
low mass limit/m/z
most intense mass/m/z
200
280
200
1000
200
350 (1500)b
300
3000
a
Methods outlined in refs 2, 71. bSpectra show strong evidence of fragmenting even at very low laser power; main peak possibly from fragments, second peak max at m/z 1500 is probably the real peak max. 3917
dx.doi.org/10.1021/cr200429v | Chem. Rev. 2012, 112, 3892−3923
Chemical Reviews
Review
Figure 16. SEC of Syncrude (S) and fractions from TLC fractionationFraction 1 immobile and Fraction 4 the most mobile in PC from ref 44 shown in ref 70. Reproduced with permission. Copyright 2010 John Wiley & Sons Ltd.
consistent with problems raised by Strausz et al.13 regarding energy transfer to large aromatic chromophores. The combined evidence from experimental and theoretical studies thus strongly suggests that molecules above about m/z 10000 that may be present in the samples are observed by LDMS with difficulty, if at all. Indeed observations of higher mass fringes in the past have been at relatively low levels.152 The features of the least soluble fractions are not generally observable by examination of unfractionated (i.e., “whole”) samples, even when they account for as much as 50% of the material. In the work outlined, information on the less volatile (distillate residues or asphaltenes) fractions only became available (observable) following fractionation and separation from the more abundant lighter fractions. Thus “single-step” analyses of complex samples are observed to be limited by the masking effect due to the preponderance of signal from more abundant fractions. These results outlined show that it is perilous to accept structural and molecular mass related information as complete, unless they are shown to display all the features observed in examining separated fractions.
Figure 17. LDMS of Syncrude and the TLC fractions: fraction 1 was immobile and fraction 4 was the most mobile from ref 44. Shown in ref 70. Reproduced with permission. Copyright 2010 John Wiley & Sons Ltd.
8. SUMMARY AND CONCLUSIONS LD-MS analysis of the material eluting in the excluded region of SEC chromatograms showed significantly higher average mass compared to material eluting in the retained region. The excluded material, which showed little or no UV−F, was found to have average mass (m/z) between ∼2000 and 3500, with higher mass limits ranging to about m/z 10 000. Materials at such high masses appear difficult to ionize by LD-MS, despite containing large aromatic chromophores. This finding is
8.1. Aims of the Review
By attempting to arrive at improved definitions of the heavy ends of coal and petroleum derived liquids, the paper aims to facilitate process development for improved utilization of heavy fractions (asphaltenes, resids). Historically much heavy petroleum derived material has been used as cheap fuel or simply thrown away. 3918
dx.doi.org/10.1021/cr200429v | Chem. Rev. 2012, 112, 3892−3923
Chemical Reviews
Review
no sample solubility in the solvents used during the determinations, electrospray ionization becomes incapable of providing an analysis of the heavier fractions that might reflect their presence. The combined use of SEC and laser desorption ionization time-of-flight mass spectrometry (LD-TOF-MS) was found useful in indicating mass ranges of complex polydisperse hydrocarbons. In the case of a heavy crude oil, the asphaltenes have been found to extend well beyond those corresponding to peak-mass intensity [normally less than (m/z) 1000]. Signal was reliably identified at up to m/z 10 000. Similarly the combined evidence from experimental and theoretical studies shows that, above about m/z 10 000, it is unlikely for larger molecular mass materials that may be present in the sample to be easily observed by LD-MS. Observations of higher mass fringes in the past have been at very low levels.
The proposition that the large molecular mass components consist of molecular aggregates lacks evidence to sustain it. While it is understood that aggregates may form in a dense phase, what is at issue in the present discussion is the elucidation of fundamental structures and molecular masses of high mass material, as identified in dilute phase. Although the misapprehension is widespread, no experimental evidence has been presented, to date, showing that large molecules encountered in dilute phase are aggregates, or indeed that they may be disaggregated. Furthermore, the “aggregates” model of large molecules, widely upheld in the petroleum industry over the decades, has led to no improvements in processing methods. With the rising cash value of the “bottom of the barrel,” greater emphasis on the detailed chemistry of these materials has become an imperative. This paper has reviewed analytical techniques used for estimating molecular mass distributions and structural features of complex polydispersed hydrocarbons, with masses above those identifiable by gas-chromatography (or GC-MS). The features of the least solubleand usually large molecular massmaterials in complex mixtures are not generally observable by examination prior to the fractionation of (i.e., “whole”) complex samples. This has been observed even in cases where the “heavy” fraction accounts for about 50% of the total sample. The major approach reviewed covers the combined use of UV−F spectroscopy, size exclusion chromatography, mass spectrometry and NMR-spectroscopy. This is preceded by a description of useful sample fractionation methods found in the literature.
8.4. Several Novel Approaches of the Work
In performing laser desorption/ionization time-of-flight mass spectrometry (LD-TOF-MS), the introduction of fractionated samples on cut-outs of planar chromatographic plates is capable of suppressing sample recombination reactions and giving more reliable sample masses than was hitherto found possible. LDMS (coupled to planar chromatography), combined with SEC is found to be useful for indicating mass ranges in petroleum asphaltenes, extending well beyond values corresponding to peak-mass intensity [normally less than (m/z) 1000]. Using these techniques, higher mass components in asphaltenes may be reliably identified up to about m/z 10 000. The parallel application of solution state NMR methods has shown that average sizes of fused ring systems may be correlated with average molecular masses of sample fractions. The NMR pulse sequences of choice for the analysis of complex hydrocarbon mixtures are standard proton-NMR conditions (1H), 13C inverse gated (IG),152 distortionless enhancement by polarization transfer (DEPT), and quaternary only spectroscopy (QUAT). Although the latter two methods are not quantitative, they can provide useful addition structural information and confirmation of chemical shift classifications. By combining information from these pulse sequences, several structural parameters may be determined. Work on asphaltenes from Maya crude oil (Mexico) described previously2 has shown that the 1-methyl-2pyrrolidinone (NMP) soluble fraction (up to 50% by weight of the asphaltene) consisted mainly of “archipelago” type structures, while the fraction insoluble in NMP consisted mainly of “continental” structures.
8.2. The Need for Fractionation
Information on the less volatile fractions, such as distillate residues or asphaltenes, may only be observed after fractionation and separation from more abundant lighter fractions. Results of “single-step” analyses are therefore limited by the masking effect due to the preponderance of signal from more abundant fractions and that it may be perilous to accept such information as complete, unless they are shown to contain all the features of the individual fractions. The evidence presented shows that the larger molecular size aromatics of coal and petroleum asphaltene fractions have no significant fluorescence. This feature only becomes apparent when the samples are fractionated to separate large and small molecules. 8.3. Limitations of Individual Analytical Techniques
Where reported results tend to depend on the nature of the analytical methods employed, distinguishing between the limitations inherent in the characteristics of equipment used and results of measurements on samples requires special care. It was found that the intensity of signal in fluorescence based techniques tends to fall off rapidly above 1000−1500 u cutting out signal altogether above 2800−3000 u. Recent work has also shown that in order to perform analyses by ESI-MS, it is necessary to completely dissolve the samples in the solvent flowing to the electrospray source. Unless the solvent mixture carrying the sample in ESI applications (e.g., methanol, toluene) is able to dissolve the sample completely, only a partial analysis, presumably of more soluble components, will be made. Meanwhile, the discussion presented in this paper regarding solvent strengths necessary to dissolve some samples completely indicates that the use of methanol or toluene falls well short of dissolving whole samples of coal tars and petroleum asphaltenes. When there is little or
8.5. Closing Emphatic Remarks
(1) Above the GC range, no single analytical method appears unambiguously capable of determining molecular mass distributions or indicating the predominant chemical structural features present within samples of heavy hydrocarbon liquids. Advances in this field have required (and will require) using and comparing evidence from several independent analytical methods. (2) The analysis of complex polydispersed hydrocarbon mixtures requires fractionation of the “whole” sample in order to arrive at a measure of structural detail. (3) It is critically important to distinguish between the limitations of individual analytical techniques and the characteristics of the samples themselves. (4) The combination of structural information from NMR and UV−F spectroscopy, with information from 3919
dx.doi.org/10.1021/cr200429v | Chem. Rev. 2012, 112, 3892−3923
Chemical Reviews
Review
molecular mass determinations (using size exclusion chromatography and laser-desorption mass spectrometry), is useful in attaining an improved level of understanding of the nature of heavy hydrocarbon liquids.
(Chemical Engineering Department) with Prof. R. Kandiyoti in 2008. His PhD focused on developing and applying analytical chemistry methods to complex (heavy) coal and petroleum derived materials. Prior to this he worked with the same group for eight years providing research-technical support. After the Ph.D., he worked as a consultant to a major oil company, developing analytical chemistry methods. This was followed by post-doctoral positions at Imperial College with Prof. R. Kandiyoti, also with Prof. D. Dugwell and Prof, J. Hewitt. In 2009 he joined the European Commission − JRC − Institute for Energy, to work on the gasification of biomass. Now he is researching the pyrolysis of biomass at the Hawaii Natural Energy Institute, University of Manoa (2011). Throughout his career, he has continued his research into analytical chemistry related to thermochemical processes.
AUTHOR INFORMATION Corresponding Author
*E-mail:
[email protected]. Present Address #
Hawaii Natural Energy Institute, University of Hawaii at Manoa, 1680 East West Road, POST 109, Honolulu, HI 96822. Notes
The authors declare no competing financial interest. Biographies
Keith Bartle has worked in the area of the analytical chemistry of fuels since his early days at the Coal Tar Research Association (1956−1962) while taking his first degree by part-time study at the Bradford Institute of Technology. His Ph.D., in the NMR of coal derived hydrocarbons at the University of Leeds, was awarded in 1966. Periods as a postdoctoral fellow at the Universities of Bradford (as ICI Fellow) and Stockholm followed, before his appointment to the staff of the School of Chemistry of the University of Leeds in 1969, where has remained, successively as Instructor, Senior Lecturer, and Professor; he is currently Visiting Professor in the Energy and Resources Research Institute, University of Leeds. Other visiting appointments have been at Indiana and Brigham Young Universities, and Victoria (Australia) Colleges. His research interests have centered mainly on the development of gas and supercritical fluid chromatography, and their applications in fossil fuel and atmospheric chemistry and, currently, biomass combustion.
Alan Herod received a Ph.D. from Leeds University in 1967, followed by postdoctoral work at Toronto University (with A. G. Harrison) and at Heriot-Watt University, Edinburgh. His first employment was with British Coal Utilization Research Association (BCURA), Leatherhead UK and subsequently until 1992, at British Coal, Coal Research Establishment in Gloucestershire, UK. In 1993, he joined Imperial College London, Chemical Engineering Department with Prof. R. Kandiyoti until his retirement in 2009. Now an Honorary Research Fellow, he is able to write reviews on coal liquids, petroleum mass spectrometry, and the analytical methods that have interested him in application to these materials. He has developed and modified these methods together with the group at Imperial to overcome problems experienced with the analysis of these very complex materials.
R. Kandiyoti received his B.S. degree in Chemical Engineering from Columbia University in New York (1965) and his Ph.D. degree from the University of London (1969). He served as academic staff in Chemical Engineering Departments of the Middle East Technical
Trevor J. Morgan received a B.Sc. degree in Chemistry from University of London (2005) and a Ph.D. from Imperial College London 3920
dx.doi.org/10.1021/cr200429v | Chem. Rev. 2012, 112, 3892−3923
Chemical Reviews
Review
(13) Strausz, O. P.; Safarik, I.; Lown, E. M.; Morales-Izquierdo, A. Energy Fuels 2008, 22, 1156. (14) Strausz, O. P.; Safarik, I.; Lown, E. M. Energy Fuels 2009, 23, 1555. (15) George, A.; Morgan, T. J.; Alvarez, P.; Millan, M.; Herod, A. A.; Kandiyoti, R. Fuel 2010, 89, 2953. (16) Ball, P. Column: The Crucible; Chemistry World, 2010; 7(10), 32. (17) Mullins, O. C. Energy Fuels 2010, 24, 2179. (18) Wiehe, I. A. Process Chemistry of Petroleum Macromolecules; CRC Press, Taylor and Francis: Boca Raton FL, USA, 2008; Chapter 2.6.3. (19) Ancheyta, J.; Speight, J. G. Hydroprocessing of Heavy Oils and Residua; CRC Press: Boca Raton, London, NY; 2007. (20) Berrueco, C.; Á lvarez, P.; Venditti, S.; Morgan, T. J.; Herod, A. A.; Millan, M.; Kandiyoti, R. Energy Fuels 2009, 23, 3008. (21) Lazaro, M.-J.; Domin, M.; Herod, A. A.; Kandiyoti, R. J Chromatogr A 1999, 840, 107. (22) Li, W.; Morgan, T. J.; Herod, A. A.; Kandiyoti, R. J. Chromatogr. A 2004, 1024, 227. (23) Deelchand, J.-P.; Naqvi, Z.; Dubau, C.; Shearman, J.; Lazaro, M.-J.; Herod, A. A.; Read, H.; Kandiyoti, R. J. Chromatogr A. 1999, 830, 397. (24) Suelves, I.; Islas, C. A.; Herod, A. A.; Kandiyoti, R. Energy Fuels 2001, 15, 429. (25) Islas, C. A.; Suelves, I.; Carter, J. F.; Li, W.; Morgan, T. J.; Herod, A. A.; Kandiyoti, R. Rapid Commun. Mass Spectrom. 2002, 16, 774. (26) Suelves, I.; Islas, C. A.; Millan, M.; Galmes, C.; Carter, J. F.; Herod, A. A.; Kandiyoti, R. Fuel 2003, 82, 1. (27) Islas, C. A.; Suelves, I.; Carter, J. F.; Herod, A. A.; Kandiyoti, R. Rapid Commun. Mass Spectrom. 2002, 16, 481. (28) Pipatmanomai, S.; Islas, C. A.; Suelves, I.; Herod, A. A.; Kandiyoti, R. J. Anal Appl. Pyrol. 2001, 58, 299. (29) Lazaro, M.-J.; Islas, C. A.; Herod, A. A.; Kandiyoti, R. Energy Fuels 1999, 13, 1212. (30) Rodríguez, J.; Requena, T.; Fontecha, J.; Goudédranche, H.; Juárez, M. J. Agric. Food Chem. 1999, 47, 558. (31) Huisman, I. H.; Prádanos, P.; Hernández, A. J. Membr. Sci. 2000, 179, 79. (32) Millipore web-site, Ultra filtration membrane information: http://www.millipore.com/publications.nsf/docs/ tn1000enus?open&lang=en (33) Marshall, A. D.; Munro, P. A.; Tragardh, G. Desalination 1993, 91, 65. (34) Rippe, B.; Haraldsson, B. Physiol. Rev. 1999, 74, 163. (35) Duong, A.; Chattopadhyaya, G.; Kwok, W. Y.; Smith, K. J. Fuel 1997, 76, 821. (36) Lai, W.-C.; Smith, K. J. Fuel 2001, 80, 1121. (37) Marques, J.; Merdrignac, I.; Baudot, A.; Barré, L.; Guillaume, D.; Espinat, D.; Brunet, S. Oil Gas Sci. Technol.−Rev. IFP 2008, 63, 139. (38) Zhao, B.; Shaw, J. M. Energy Fuels 2007, 21, 2795. (39) Lakowicz, J. R. Principles of Fluorescence Spectroscopy; Springer: New York, 2010. (40) Zander, M.; Haenel, M. W. Fuel 1990, 69, 1206. (41) Morgan, T. J.; George, A.; Davis, D. B.; Herod, A. A.; Kandiyoti, R. Energy Fuels 2008, 22, 1824. (42) Morgan, T. J. Ph.D. Thesis, Imperial College London, London, UK, 2008. (43) Herod, A .A.; Morgan, T. J.; Alvarez, P.; George, A.; Millan, M.; Kandiyoti, R. Analysis of Maya crude oil. Preprints ACS Fuel Division, ACS National Meeting, Salt Lake City, Utah; March, 2009, 54(1) 28. (44) Á lvarez, P.; Berrueco, C.; Venditti, S.; George, A.; Morgan, T. J.; Millan, M.; Herod, A. A.; Kandiyoti, R. Syncrude Characterization. Preprints ACS Fuel Division, ACS National meeting, Salt Lake City, Utah, March, 2009; American Chemical Society: Washington, DC, 2009; Vol. 54(1), p 33. (45) Morgan, T. J.; George, A.; Alvarez, P.; Millan, M.; Herod, A. A.; Kandiyoti, R. Energy & Fuels. 2008, 22, 3275.
University (Ankara, Turkey, 1969−1972) and Boğaziçi University (Istanbul, Turkey, 1974−1980), before joining Imperial College London (1980), where he served as Professor of Chemical Engineering and co-ordinator of the Energy Engineering Group. On retirement in September 2008, he was appointed as “Distinguished Research Fellow” in the Department of Chemical Engineering at Imperial College London. R. Kandiyoti has worked on topics relating to experimental reactor design for pyrolysis, gasification, and liquefaction, the thermochemical characterization of fossil fuels, biomass and waste, the characterization of heavy hydrocarbon liquids, and environmental aspects of power generation.
ACRONYMS APPI DIE ESI FD FI FT-ICR-MS
atmospheric pressure photoionization delayed ion extraction electrospray ionization field desorption (ionization) field ionization Fourier transform ion cyclotron resonance mass spectrometry GC-MS gas chromatography−mass spectrometry LD-MS laser desorption mass spectrometry LD-TOF-MS laser desorption time-of-flight mass spectrometry MALDI-MS matrix assisted laser desorption ionization mass spectrometry NMP N-methyl pyrrolidinone; 1-methyl 2-pyrrolidinone; 1-methyl 2-pyrrolidone NMR nuclear magnetic resonance spectroscopy PAH polycyclic aromatic hydrocarbon PC planar chromatography, or thin layer chromatography SEC size exclusion chromatography THF tetrahydrofuran (solvent) UV−A ultraviolet absorbance spectroscopy UV−F ultraviolet fluorescence spectroscopy
REFERENCES (1) Wiehe, I. A. Process Chemistry of Petroleum Macromolecules; CRC Press: Boca Raton, Fl, USA; 2008, pp 1−3. (2) Morgan, T. J; Alvarez-Rodriguez, P.; George, A.; Herod, A. A.; Kandiyoti, R. Energy Fuels 2010, 24, 3977. (3) Murgich, J. Mol. Simul. 2003, 29, 451. (4) Sabbah, H.; Morrow, A. L.; Pomerantz, A. E.; Zare, R. N. Energy Fuels 2011, 25, 1597. (5) Alshareef, A. H.; Scherer, A.; Tan, X.; Azyat, K.; Stryker, J. M.; Tykwinski, R. R.; Gray, M. R. Energy Fuels 2011, 25, 2130. (6) Klee, T.; Masterson, T.; Miller, B.; Barrasso, E; Bell, J.; Lepkowicz, R.; West, J.; Haley, J. E.; Schmitt, D. L.; Flikkema, J. L.; Cooper, T. M.; Ruiz-Morales, Y.; Mullins, O. C. Energy Fuels 2011, 25, 2065. (7) Herod, A. A.; Bartle, K. D.; Kandiyoti, R. Energy Fuels 2007, 21, 2176. (8) Kandiyoti, R.; Herod, A. A.; Bartle, K. D. Solid Fuels and Heavy Hydrocarbon Liquids: Thermal Characterisation and Analysis; Elsevier: Oxford, 2006; Chapter 8. (9) Li, C.-Z.; Wu, F.; Cai, H.-Y.; Kandiyoti, R. Energy Fuels 1994, 8, 1039. (10) Herod, A. A.; Zhang, S.-F.; Johnson, B. R.; Bartle, K. D.; Kandiyoti, R. Energy Fuels 1996, 10, 743. (11) Morgan, T. J.; Millan, M.; Behrouzi, M.; Herod, A. A.; Kandiyoti, R. Energy Fuels 2005, 19, 164. (12) Morgan, T. J.; George, A.; Alvarez, P.; Herod, A. A.; Millan, M.; Kandiyoti, R. Energy Fuels 2009, 23, 6003. 3921
dx.doi.org/10.1021/cr200429v | Chem. Rev. 2012, 112, 3892−3923
Chemical Reviews
Review
(46) Á lvarez, P.; Morgan, T. J.; Granda, M.; Sutil, J.; Menendez, R.; Fernandez, J.; Viña, J.; Millan, M.; Herod, A. A.; Kandiyoti, R. Energy Fuels 2008, 22, 4077. (47) Alvarez, P.; Morgan, T. J.; George, A.; Vendetti, S.; Berrueco, C.; Millan, M.; Herod, A. A.; Kandiyoti, R. NMR characterization of an Athabasca bitumen. Chemistry of Petroleum and Emerging Technologies, Division of Petroleum Chemistry, ACS National Meeting, Washington, DC, August, 2009; American Chemical Society: Washington, DC, 2009 (48) Á lvarez, P.; Morgan, T. J.; George, A.; Granda, M.; Sutil, J.; Herod, A. A.; Millan-Agorio, M.; Menéndez, R.; Kandiyoti, R. Predicting properties of carbon materials from the advanced analysis of the parent anthracene oil based pitch. Keynote Paper at Carbon 2009; Biarritz, France. (49) Clark, E. R.; Darwent, J. R.; Demirci, B.; Flunder, K.; Gaines, A. F.; Jones, A. C. Energy Fuels 1987, 1, 392. (50) Morgan, T. J.; Morden, W. E.; Al-Muhareb, E.; Herod, A. A.; Kandiyoti, R. Energy Fuels 2006, 20, 734. (51) Bermejo, J.; Menendez, R.; Fernandez, A. L.; Granda, M.; Suelves, I.; Herod, A. A.; Kandiyoti, R. Fuel 2001, 80, 2155. (52) da Silva-Souza, R.; Nicodem, D E.; Garden, S. J.; Correa, R. J. Energy Fuels 2010, 24, 1135. (53) Yamaguchi, Y.; Matsubara, Y.; Ochi, T.; Wakamiya, T.; Yoshida, Z. J. Am. Chem. Soc. 2008, 130, 13867 ja8040493. (54) Hanson, K.; Roskop, L.; Djurovich, P. I.; Zahariev, F.; Gordon, M. S.; Thompson, M. E. J. Am. Chem. Soc. 2010, 132, 16247. (55) Davis, N. K. S.; Thompson, A. L.; Anderson, H. L. J. Am. Chem. Soc. 2011, 133, 30. (56) McKenna, A. M.; Purcell, J. M.; Rodgers, R. P.; Marshall, A. G. Energy Fuels 2009, 23, 2122. (57) Baker, E. W.; Lauda, J. W.; Orr, W. L. Org. Geochem. 1987, 11, 303. (58) Al-Muhareb, E.; Morgan, T. J.; Herod, A. A.; Kandiyoti, R. Petroleum Sci. Technol. 2007, 25, 81. (59) Zhang, S.-F.; Xu, B.; Moore, S. A.; Herod, A. A.; Kandiyoti, R. Fuel 1996, 75, 597. (60) Zhang, S.-F.; Xu, B.; Herod, A. A.; Kandiyoti, R. Energy Fuels 1996, 10, 733. (61) Begon, V.; Suelves, I.; Li, W.; Lazaro, M.-J.; Herod, A .A.; Kandiyoti, R. Fuel 2002, 81, 185. (62) Clar, E. Polycyclic Hydrocarbons, vols 1 and 2; Academic Press: London, 1964. (63) Burchill, P.; Herod, A. A.; James, R. G. In Carcinogenesis; Jones, P. W., Freudenthal, R. I., Eds.; Raven Press, New York, 1978, 35; Vol. 3. (64) Herod, A. A.; Ladner, W. R.; Snape, C. E. Phil. Trans. R. Soc. London A 1981, 300. (65) Burchill, P.; Herod, A. A.; Pritchard, E. J. Chromatogr. 1982, 242, 51. (66) Parker, J. E.; Johnson, C. A. F.; John, P.; Smith, G. P.; Herod, A. A.; Stokes, B. J.; Kandiyoti, R. Fuel 1993, 72, 1161. (67) Begon, V.; Megaritis, A.; Lazaro, M.-J.; Herod, A. A.; Dugwell, D. R.; Kandiyoti, R. Fuel 1998, 77, 1261. (68) Begon, V.; Lazaro, M.-J.; Suelves, I.; Herod, A. A.; Dugwell, D. R.; Kandiyoti, R. Eur. J. Mass Spectrom. 2000, 6, 39. (69) Millan, M.; Morgan, T. J.; Behrouzi, M.; Karaca, F.; Galmes, C.; Herod, A. A.; Kandiyoti, R. Rapid Commun. Mass Spectrom. 2005, 19, 1867. (70) Herod, A. A. Rapid Commun. Mass Spectrom. 2010, 24, 2507. (71) Morgan, T. J.; George, A.; Alvarez-Rodriguez, P.; Millan, M.; Herod, A. A.; Kandiyoti, R. J. Chromatogr. A 2010, 1217, 3804. (72) Berrueco, C.; Venditti, S.; Morgan, T. J.; Á lvarez, P.; Millan, M.; Herod, A. A.; Kandiyoti, R. Energy Fuels 2008, 22, 3265. (73) Millan, M.; Behrouzi, M.; Karaca, F.; Morgan, T. J.; Herod, A. A.; Kandiyoti, R Catal. Today 2005, 109, 154. (74) Li, C.-Z.; Wu, F.; Xu, B.; Kandiyoti, R. Fuel 1995, 74, 37. (75) Groenzin, H.; Mullins, O. C. Energy Fuels 2000, 14, 677. (76) Islas, C. A.; Suelves, I.; Millan, M.; Apicella, B.; Herod, A. A.; Kandiyoti, R. J. Sep. Sci. 2003, 26, 1422.
(77) Strausz, O. P.; Peng, P.; Murgich, J. Energy Fuels 2002, 16, 809. (78) Ascanius, B. E.; Merino-Garcia, D.; Andersen, S. I. Energy Fuels 2004, 18, 1827. (79) Mullins, O. C. Energy Fuels 2009, 23, 2845. (80) Mullins, O. C. Fuel 2007, 86, 309. (81) Ruiz-Morales, Y.; Wu, X.; Mullins, O. C. Energy Fuels 2007, 21, 944. (82) Ruiz-Morales, Y.; Mullins, O. C. Energy Fuels 2007, 21, 256. (83) Ruiz-Morales, Y.; Mullins, O. C. Energy Fuels 2009, 23, 1169. (84) Paraphrased from Bernal, M., Black Athena, Fifth paperback printing; Rutgers University Press: New Brunswick, NJ, 2002; Vol. 2, p 65. (85) Bartle, K. D. In Spectroscopic Analysis of Coal Liquids; Kershaw, J. R., Ed.; Elsevier: Amsterdam, 1989; pp 13−40. (86) Fetzer, J. C. Large Polycyclic Aromatic Molecules - Chemistry and Analysis; Wiley Interscience: New York, 2000. (87) Wise, S. A. Polycyclic Aromat. Compd. 2002, 22, 197. (88) Fetzer, J. C.; Briggs, W. R. Polycyclic Aromat. Compd. 1994, 4, 3. (89) Fetzer, J. C. Polycyclic Aromat. Compd. 2007, 27, 143. (90) Marsh, N. D.; Ledesma, E. B.; Wornat, M. J.; Tan, M. P.; Zhu, D.; Law, C. K. Polycyclic Aromat. Compd. 2005, 25, 227. (91) Islas, C. A.; Suelves, I.; Carter, J. F.; Herod, A. A.; Kandiyoti, R. Rapid Commun. Mass Spectrom. 2001, 15, 845. (92) Apicella, B.; Millan, M.; Alfè, M.; Herod, A. A.; Pucci, P.; Ciajolo, A. Rapid Commun. Mass Spectrom. 2006, 20, 1104. (93) Adegoroye, A.; Paterson, N.; Morgan, T. J.; Herod, A. A.; Dugwell, D. R.; Kandiyoti, R. Fuel 2004, 83, 1949. (94) Purevsuren, B.; Herod, A. A.; Kandiyoti, R.; Morgan, T. J.; Avid, B.; Gerelmaa, T.; Davaajav, Ya. Fuel 2004, 83, 799. (95) Morgan, T. J.; Herod, A. A.; Brain, S. A.; Chambers, F. M.; Kandiyoti, R. J. Chromatogr. A 2005, 1095, 81. (96) Trejo, F.; Ancheyta, J.; Morgan, T. J.; Herod, A. A.; Kandiyoti, R. Energy Fuels 2007, 21, 2121. (97) Millan, M.; Adell, C.; Hinojosa, C.; Herod, A. A.; Dugwell, D. R.; Kandiyoti, R. Energy Fuels 2007, 21, 1370. (98) Bodman, S. D.; McWhinnie, W. R.; Begon, V.; Millan, M.; Suelves, I.; Lazaro, M.-J.; Herod, A. A.; Kandiyoti, R. Fuel 2003, 82, 2309. (99) Suelves, I.; Lazaro, M.-J.; Begon, V.; Morgan, T. J.; Herod, A. A.; Dugwell, D. R.; Kandiyoti, R. Energy Fuels 2001, 15, 1153. (100) Herod, A. A.; Zhuo, Y.; Kandiyoti, R. J. Biochem. Biophys. Methods 2003, 56, 335. (101) Herod, A. A.; Johnson, B. R.; Bartle, K. D.; Carter, D. M.; Cocksedge, M. J.; Domin, M.; Kandiyoti, R. Rapid Commun. Mass Spectrom. 1995, 9, 1446. (102) Bartle, K. D.; Taylor, N.; Mulligan, M. J.; Mills, D. G.; Gibson, C. Fuel 1983, 62, 1181. (103) Li, C.-Z.; Bartle, K. D.; Kandiyoti, R. Fuel 1993, 72, 3. (104) Paul-Dauphin, S.; Karaca, F.; Morgan, T. J.; Millan-Agorio, M.; Herod, A. A.; Kandiyoti, R. Energy Fuels 2007, 21, 3484. (105) Nali, M.; Calemma, V.; Montanari, L. Org. Mass Spectrom. 1994, 29, 607. (106) Merdrignac, I.; Truchy, C.; Robert, E.; Guibard, I.; Kressmann, S. Pet. Sci. Technol. 2004, 22, 1003. (107) Herod, A. A.; Kandiyoti, R. J. Chromatogr. A 1995, 708, 143. (108) Domin, M.; Herod, A. A.; Kandiyoti, R.; Larsen, J. W.; Lazaro, M.-J.; Li, S.; Rahimi, P. Energy Fuels 1999, 13, 552. (109) Sato, S.; Takanohashi, T.; Tanaka, R. Energy Fuels 2005, 19, 1991. (110) Behrouzi, M.; Luckham, P. F. Energy Fuels 2008, 22, 1792. (111) Herod, A. A.; Kandiyoti, R. Energy Fuels 2008, 22, 4307. (112) Guillen, M.; Blanco, J.; Canga, J.; Blanco, C. Energy Fuels 1991, 5, 188. (113) Lafleur, A. L.; Nakagawa, Y. Fuel 1989, 68, 741. (114) Takanohashi, T.; Iino, M.; Nakamura, K. Energy Fuels 1994, 8, 395. (115) Masuda, K.; Okuma, O.; Knaji, M.; Matsumara, T. Fuel 1996, 75, 1065. (116) Chen, C.; Iino, M. Fuel 2001, 80, 929. 3922
dx.doi.org/10.1021/cr200429v | Chem. Rev. 2012, 112, 3892−3923
Chemical Reviews
Review
(117) Herod, A. A.; Shearman, J.; Lazaro, M.-J.; Johnson, B. R.; Bartle, K. D; Kandiyoti, R. Energy Fuels 1998, 12, 174. (118) Karaca, F.; Behrouzi, M.; Morgan, T. J.; Herod, A. A.; Kandiyoti, R. Energy Fuels 2005, 19, 187. (119) Karaca, F.; Behrouzi, M.; Morgan, T. J.; Herod, A. A.; Kandiyoti, R. Fuel 2005, 84, 1805. (120) Pomerantz, A. E.; Hammond, M. R.; Morrow, A. L.; Mullins, O. C.; Zare, R. N. Energy Fuels 2009, 23, 1162. (121) Arnaud, C. H. Digging into asphaltenes − mass spectrometry uncovers chemical details of petroleum’s most recalcitrant fraction. Chem. Eng. News 2009, September 21, p 12. (122) Mullins, O. C.; Martínez-Haya, B. A.; Marshall, A. G. Energy Fuels 2008, 22, 1765. (123) Karaca, F.; Islas, C. A.; Millan, M.; Behrouzi, M.; Morgan, T. J.; Herod, A. A.; Kandiyoti, R. Energy Fuels 2004, 18, 778. (124) Al-Muhareb, E. M.; Karaca, F.; Morgan, T. J.; Herod, A. A.; Bull, I. D.; Kandiyoti, R. Energy Fuels 2006, 20, 1165. (125) Herod, A. A. Mass Spectrometry of Coal Liquids. In Encyclopedia of Mass Spectrometry; Nico, M., Nibbering, M., Series Eds.; Gross, M. L., Caprioli, R. M., Eds.; Elsevier: Amsterdam, 2005; pp 790−803; Vol. 4, Fundamentals of and Applications to Organic (and Organometallic) Compounds. (126) Herod, A. A.; Islas, C.; Lazaro, M. J.; Dubau, C.; Carter, J. F.; Kandiyoti, R. Rapid Commun. Mass Spectrom. 1999, 13, 201. (127) Islas, C. A.; Suelves, I.; Carter, J. F.; Herod, A. A.; Kandiyoti, R. Rapid Commun. Mass Spectrom. 2000, 14, 1766. (128) Final Report of project B53, to BCURA, The Investigation of Large Coal-Derived Molecules: New Developments and Applications, October 2003. (129) Flego, C.; Zannoni, C. Energy Fuels 2010, 24, 6041. (130) Borton, D.; Pinkston, D. S.; Hurt, M. R.; Tan, X.; Azyat, K.; Scherer, A.; Tykwinski, R.; Gray, M.; Qian, K.; Kenttamaa, H. Energy Fuels 2010, 24, 5548. (131) Chapman, J. R. Practical Organic Mass Spectrometry, 2nd ed; John Wiley & Sons Ltd: Chichester, UK, 1993. (132) McLafferty, F. W.; Turecek, F. Interpretation of Mass Spectra, 4th ed; University Science Books: Mill Valley, CA, 1993. (133) Schaub, T. M.; Hendrickson, C. L.; Qian, K.; Quinn, J. P.; Marshall, A. G. Anal. Chem. 2003, 75, 2172. (134) Qian, K.; Edwards, K. E.; Siskin, M.; Olmstead, W. N.; Mennito, A. S.; Dechert, G. J.; Hoosain, N. E. Energy Fuels 2007, 21, 1042. (135) McEwan, C. N.; Pagnotti, V. S.; Inutan, E. D.; Trimpin, S. Anal. Chem. 2010, 82, 9164. (136) Purcell, J. M.; Hendrickson, C. L.; Rodgers, R. P.; Marshall, A. G. Anal. Chem. 2006, 78, 5906. (137) Purcell, J. M.; Merdrignac, I.; Rodgers, R. P.; Marshall, A. G.; Gauthier, T.; Guibard, I. Energy Fuels 2010, 24, 2257. (138) These, A.; Reemtsma, T. Anal. Chem. 2003, 75, 6275. (139) McKenna, A. M.; Blackney, G. T.; Xian, F.; Glaser, P. B.; Rodgers, R. P.; Marshall, A. G. Energy Fuels 2010, 24, 2939. (140) McKenna, A. M.; Purcell, J. M.; Rodgers, R. P.; Marshall, A. G. Energy Fuels 2010, 24, 2929. (141) Corilo, Y. E.; Vaz, B. G.; Simas, R. C.; Lopes Nascimento, H. D.; Klitzke, C. F.; Pereira, R. C. L.; Bastos, W. L.; Santos Neto, E. V.; Rodgers, R. P.; Eberlin, M. N. Anal. Chem. 2010, 82, 3990. (142) Herod, A. A.; Millan, M.; Morgan, T. J.; Li, W.; Feng, J.; Kandiyoti, R. Eur. J. Mass Spectrom. 2005, 11, 429−442. (143) Karaca, F.; Morgan, T. J.; George, A.; Bull, I. D.; Herod, A. A.; Millan, M.; Kandiyoti, R. Rapid Commun. Mass Spectrom. 2009, 23, 2087. (144) Stanford, L. A.; Kim, S.; Klein, G. C.; Smith, D. F.; Rodgers, R. P.; Marshall, A. G. Environ. Sci. Technol. 2007, 41, 2696. (145) Hsu, H.-J.; Oung, J.-N.; Kuo, T.-L.; Wu, S.-H.; Shiea, J. Rapid Commun. Mass Spectrom. 2007, 21, 375. (146) Purcell, J. M.; Hendrickson, C. L.; Rodgers, R. P.; Marshall, A. G. J. Am. Soc. Mass Spectrom. 2007, 18, 1682. (147) Purcell, J. M.; Juyal, P.; Kim, D. G.; Rodgers, R. P.; Hendrickson, C. L.; Marshall, A. G. Energy Fuels 2007, 21, 2869.
(148) Haapala, M.; Purcell, J. M.; Saarela; Franssila, V. S.; Rodgers, R. P.; Hendrickson, C. L.; Kotiaho, T.; Marshall, A. G.; Kostiainen, R. Anal. Chem. 2009, 81, 2799. (149) Herod, A. A. MALDI-MS of Polydisperse Hydrocarbon Samples. In Mass Spectrometry Handbook; Lee, M. S., Ed.; John Wiley & Sons: Hoboken, NJ, 2012; Chapter 33. (150) Hortal, A. R.; Hurtado, P.; Martínez-Haya, B.; Mullins, O. C. Energy Fuels 2007, 21, 2863. (151) Knockenmuss, R. Eur. J. Mass Spectrom. 2009, 15, 189. (152) Lazaro, M.-J.; Herod, A. A.; Domin, M.; Zhuo, Y.; Islas, C. A.; Kandiyoti, R. Rapid Commun. Mass. Spectrom. 1999, 13, 1401. (153) Elsila, J. E.; deLeon, N. P.; Zare, R. N. Anal. Chem. 2004, 76, 2430. (154) Botto, R. E.; Wilson, R.; Winans, R. E. Energy Fuels 1987, 1, 173. (155) Andersen, J. M.; Luengo, C. A.; Moinelo, S. R.; Garcia, R.; Snape, C. E. Energy Fuels 1998, 12, 524. (156) Dickinson, E. M. Fuel 1980, 59, 290. (157) Netzel, D. A. Anal. Chem. 1987, 59, 1775. (158) Bendall, M. R.; Pegg, D. T. J. Magn. Reson. 1983, 53, 272. (159) Guillen, M.; Diaz, C.; Blanco, J. Fuel Process. Technol. 1998, 58, 1. (160) Diaz, C.; Blanco, C. G. Energy Fuels 2003, 17, 907. (161) Strom, D. A.; Edwards, J. C.; Decanio, S. J.; Sheu, E. Y. Energy Fuels 1994, 8, 561. (162) Wilson, M.; Collin, P.; Pugmire, R.; Grant, D. Fuel 1982, 61, 959. (163) Snape, C.; Ladner, W. R. Fuel 1978, 57, 685. (164) Cookson, D.; Smith, B. J. Magn. Reson. 1984, 57, 355. (165) Gillet, S.; Rubini, P.; Delpuecg, J. J.; Escalier, J. C.; Valentin, P. Fuel 1981, 60, 221. (166) Qian, S. A.; Zhang, P. Z.; Li, B. L. Fuel 1985, 64, 1085. (167) Rongbao, L.; Zengmin, S.; Bailing, L. Fuel 1988, 67, 565. (168) Asphaltene Particles in Fossil Fuel Exploration, Recovery, Refining and Production Processes Sharma, M. K., Yen., T. F., Eds.; Plenum Press: New York and London, 1994; p 118. (169) Bartle, K. D.; Ladner, W. R.; Martin, T. G.; Snape, C .E.; Williams, D. F. Fuel 1979, 58, 413. (170) Kershaw, J. R.; Black, K. J. T. Energy Fuels 1993, 7, 420. (171) Alvarez, P.; Morgan, T. J.; George, A.; Vendetti, S.; Berrueco, C.; Millan, M.; Herod, A. A.; Kandiyoti, R. Papers in preparation on Athabasca bitumen. (172) Islas, C. A.; Suelves, I.; Li, W.; Morgan, T. J.; Herod, A. A.; Kandiyoti, R. Fuel 2003, 82, 1813. (173) Herod, A. A.; George, A.; Islas, C. A.; Suelves, I.; Kandiyoti, R. Energy Fuels 2003, 17, 862. (174) Strausz, O. P; Morales-Izquierdo, A; Kazmi, M.; Montgomery, D. S.; Payzant, J. D.; Safarik, I.; Murgich, J. Energy Fuels 2010, 24, 5053. (175) Strausz, O. P; Lown, E. M.; Morales-Izquierdo, A; Kazmi, M.; Montgomery, D. S.; Payzant, J. D.; Murgich, J. Energy Fuels 2011, 25, 4552. (176) Herod, A. A.; Bartle, K. D.; Morgan, T. J.; Kandiyoti, R. Presentation to PetroPhase 2011, Imperial College London, London, UK, 10−14 July 2011.
3923
dx.doi.org/10.1021/cr200429v | Chem. Rev. 2012, 112, 3892−3923