Detailed Characterization of Heavy Crude Oils and Asphaltenes by Ultrahigh Resolution Fourier Transform Ion Cyclotron Resonance Mass Spectrometry Amy Marilyn Mckenna

Home , Asphaltene, Heavy crude oil, Light crude oil

Florida State University Libraries

Electronic Theses, Treatises and Dissertations The Graduate School

2009 Detailed Characterization of Heavy Crude Oils and Asphaltenes by Ultrahigh Resolution Fourier Transform Ion Cyclotron Resonance Mass Spectrometry Amy Marilyn McKenna

Follow this and additional works at the FSU Digital Library. For more information, please contact [email protected] THE FLORIDA STATE UNIVERSITY

COLLEGE OF ARTS AND SCIENCES

DETAILED CHARACTERIZATION OF HEAVY CRUDE OILS AND

ASPHALTENES BY ULTRAHIGH RESOLUTION FOURIER TRANSFORM

ION CYCLOTRON RESONANCE MASS SPECTROMETRY

AMY MARILYN MCKENNA

A Dissertation submitted to the Department of Chemistry and Biochemistry in partial fulfillment of the requirements for the degree of Doctor of Philosophy

Degree Awarded: Fall Semester, 2009

Alan Marshall Professor Directing Dissertation

Phillip Froelich University Representative

Ken Goldsby Committee Member

Michael Roper Committee Member

Ryan Rodgers Committee Member

Approved:

William Cooper, Assistant Chair, Department of Chemistry and Biochemistry

Joseph Travis, Dean, College of Arts and Sciences

The Graduate School has verified and approved the above-named committee members.

To my family, past and present.

and

To my husband,

David Matthew McKenna whose selfless sacrifice and support allowed me to continue my education and raise our family.

iii ACKNOWLEDGEMENTS

First and foremost, I would like to thank Alan Marshall. The experience of working with such a truly dedicated, diverse scientific mind has contributed to any future success I will have. It is not often that you find someone who leads so thoroughly by example, and Alan has provided an environment where logic, commitment and analytical thought is of utmost importance, thereby facilitating the growth of the entire research group in a collaborative effort. Thank you, Alan. Never has anyone forced me to think, act and to be an analytical chemist more than Chris Hendrickson. “Methodical logic” is the best way to explain the way that he approaches a problem and being privy to it has made me a better scientist. For that, I am forever grateful. Thank you, Chris, for teaching me to be diligently analytical in my research. I am most grateful to Ryan Rodgers and his support and expertise in nearly every facet of my research. His sense of humor, patience and support molded me into a semi-confident analytical chemist with a solid comprehension of petroleum. Although I swamped your inbox with data, I am appreciative for you always taking time out of your day to help me work through a problem. More importantly, I value and appreciate your friendship and your mad skills as a traveler. Boddington’s. Say no more. John Quinn, for all of your patience and assistance over the years and for teaching me the workings of a true instrumentation lab, I thank you sincerely. I also need to thank Jerry Purcell, for teaching me everything I know about APPI and for teaching me “how to take it apart and see how it works”. A skilled analytical chemist with the power and heart of a mechanic, Jerry, you are a force to be reckoned with. Thank you for teaching me how to be a “button turner”. Don Smith, who has yet to stop answering all of my questions, I can not thank you enough. Not only did you teach me that packing material was not oil sands, but you had the heart to not laugh until you

iv left the lab (and saved the vial for the Hall of Shame). From Styrofoam to asphaltenes, you have helped me every step of the way. From one side of the pond to the other, from the bottom of my heart, I thank you. Nate Kaiser, even though you are my mortal sworn enemy as a Wolverine fan, thank you for all that you have taught me in the past year. Your knowledge of FT-ICR instrumentation and theory is impressive, even more that you are able to explain it in such a way that I understand it better. Thank you for being a friend and for always helping me, no matter what was wrong. (“Not getting any ions? Want me to help you out? Turn on the excitation amplifier.”) I would also like to thank the entire Marshall Research group, past and present. I am truly standing on the shoulders of everyone who has come before me. Greg Blakney, for countless advice and technical wisdom, thank you. Jeremiah Tipton, high bay resident extraordinaire, for always lightening the mood with a joke or a laugh, thank you. I would also like to thank my family. Mom, thank you for being such a strong supporter of my education, from start to finish. For countless weekends and evenings spent helping me with the kids, for helping us from everything from diapers to donuts (literally), thank you. To my brother, Jason, who listened to me and supported me with a glazed over look in his eyes. I am very proud of you, thank you for all your help. Finally, I would like to thank my husband, David. You made this possible. Together we have accomplished so much amid more chaos than we ever thought we could handle. For uprooting your life to move our family to Tallahassee for me to come to graduate school, for putting your own education on hold, for working long nights and for sacrificing so much, thank you. It’s your turn now, babe. You are my best friend, my strongest supporter and especially for never letting me quit. To my children, Joey, Sammy and Charleigh, who all arrived at different points in my education, Joey in undergrad and Sammy and Charleigh during

v graduate school. I would be nothing without you three. You are the reason that I get up in the morning, maybe not always smiling. To Joey, for always being inquisitive, Sammy for being such a natural analytical thinker and Charleigh, for being a medical miracle, I thank you for your love and support. I love you guys from the bottom of my heart. I am so proud of all of you.

vi TABLE OF CONTENTS

LIST OF FIGURES ...... xiii

ABSTRACT...... xxiii

CHAPTER 1. INTRODUCTION TO HEAVY OIL & ASPHALTENES ...... 1

Light Crude, Heavy Crude & Bitumen ...... 1 Light vs. Heavy Crude ...... 1 Bitumen...... 3

Crude Oil Composition…………………………………………………………….4 Environmental Concerns……………………………………………………………5

Geologic Origin of Crude Oil and Bitumen...... 6 Origin of Crude Oil...... 4 Bitumen Formation...... 4

Recovery of Crude Oil and Bitumen ...... 7 Crude Oil Recovery...... 7 Bitumen Recovery...... 7 Bitumen Extraction ...... 8

Crude Oil Refining...... 8 Distillation ...... 9 Conversion Processes ...... 11 Characterizing Distillation Products ...... 11 Bulk Property Measurement ...... 11 Light Fractions ...... 12 Middle Distillates ...... 12 VGOs ...... 13 Residue ...... 13 Nondistillable Residues ...... 13

Introduction to Asphaltenes : The Bottom of the Barrel...... 14 What are Asphaltenes? ...... 14 Problems Associated with Asphaltenes ...... 14 Asphaltene Aggregation ...... 16 Asphaltene Molecular Weight ...... 16 Asphaltene Structure ...... 17 Separation of Asphaltenes ...... 18

vii

CHAPTER 2. CHARACTERIZATION OF HYDROCARBON RESOURCES USING HIGH RESOLUTION FT-ICR MASS SPECTROMETRY : A PRIMER...... 20

SUMMARY ...... 20

Ionization Techniqes ...... 21 Electrospray Ionization...... 22 Ionization Mechanism ...... 22 Atmospheric Pressure Photoionization ...... 22 Ionization Formation in APPI ...... 25 Dopant-Assisted APPI…………………………………………………………….. 25

9.4 Tesla FT-ICR Mass Spectrometer ...... 26

Brief Overview of the Theory of FT-ICR Mass Spectrometry ...... 27

Petroleum Analysis by FT-ICR Mass Spectromery ...... 29 Kendrick Mass Sorting……………………………………………………………..30 Mass Resolution...... 32 Spectral Complexity ...... 37 Isotopic Signatures...... 37 Mass Accuracy ...... 38 Dynamic Range ...... 39

Conclusions...... 41

CHAPTER 3. OPTIMIZATION OF ATMOSPHERIC PRESSURE PHOTOIONIZATION NEBULIZATION TEMPERATURE FOR ATHABASCA BITUMEN DISTILLATION CUT POINT DETECTED BY FT- ICR MASS SPECTROMETRY...... 43

Summary...... 43

Introduction ...... 43

Experimental Methods...... 46 Sample Preparation ...... 46 Instrumentation ...... 47 APPI ...... 47

viii 14.5 Tesla LTQ/FT-ICR Mass Spectrometer...... 47

Results and Discussion ...... 47 LTQ-MS for Molecular Weight Distribution ...... 50 Determination of Optimal Sheath Gas Temperature...... 53 FT-ICR MS for Compositional Changes ...... 53 Determination of Optimal Sheath Gas Temperature...... 55 DBE vs. Carbon Number Plots...... 57 Heteroatom Class Distribution ...... 58

Conclusion ...... 62

CHAPTER 4. COMPARISON OF NONPOLAR AND POLAR SPECIES IN ATHABASCA BITUMEN HVGO DISTILLATES BY FT-ICR MASS SPECTROMETRY ...... 63

Summary...... 63

Introduction ...... 64

Experimental Methods...... 69 Sample Preparation ...... 69 Instrumentation ...... 69 APPI Source ...... 69 14.5 Tesla FT-ICR MS...... 70 Mass Calibration & Data Analysis ...... 70

Results and Discussion ...... 71 Elemental Formula Assignment...... 70 The Boduszynski Hypothesis ...... 72 Compositional Differences among HVGO Distillate Cuts: Test of the Boduszynski Hypothesis...... 75 The HVGO Compositional Continuum ...... 85 Cycloalkane Linkages ...... 85

Conclusion ...... 87

CHAPTER 5. THE COMPOSITION OF HEAVY PETROLEUM: EVOLUTION PF THE BODUSZYNSKI MODEL TO THE UPPER LIMIT OF DISTILLABLE PRODUCTS BY ULTRAHIGH RESOLUTION FT-ICR MASS SPECTROMETRY...... 91

Summary...... 91

Introduction ...... 92

ix Experimental Methods...... 98 Sample Preparation ...... 98 Instrumentation ...... 99 APPI Source ...... 99 9.4 Tesla FT-ICR MS...... 100 Broadband Phase Correction ...... 100 Mass Calibration and Data Analysis ...... 101

Results and Discussion ...... 101 Heteroatom Class Distribution ...... 105 DBE vs. Carbon Number Images ...... 106 The Boduszynski Model by FT-ICR MS ...... 108 The Continuum for Heavy Distillates ...... 110

Conclusion ...... 114

CHAPTER 6. MOLECULAR CHARACTERIZATION OF ASPHALTENES. PART I. MOLECULAR WEIGHT AND DISCOVERY OF DISTILLABLE ASPHALTENES...... 116

Introduction ...... 116

Experimental Methods...... 117 Sample Preparation ...... 117 9.4 Tesla FT-ICR MS...... 118 Mass Analysis ...... 119

Results and Discussion ...... 119 Asphaltene Molecular Weight...... 119 Distillable Asphaltenes ...... 122 Asphaltene Mass Defect...... 124 DBE vs. Carbon Number Images ...... 125 Hydrocarbon Class ...... 126 S1 Class ...... 128 Polar Classes...... 129 Composite Plots of DBE vs. Carbon Number Images ...... 129

CONCLUSION...... 133

CHAPTER 7. MOLECULAR CHARACTERIZATION OF ASPHALTENES. PART II. THE DEFINITION OF ASPHALTENE AND MALTENE COMPOSITION ...... 134

Introduction ...... 134

Experimental Methods...... 134

x Sample Preparation ...... 134 9.4 Tesla FT-ICR MS...... 135 Mass Analysis ...... 136

Results and Discussion ...... 136 Molecular Weight Distribution ...... 137 Mass Spectral Complexity...... 138 Asphaltene and Maltene Mass Defect...... 143 DBE vs. Carbon Number Images ...... 144

The Definition of Asphaltene and Maltene Composition ...... 144

CONCLUSION...... 146

CHAPTER 8. MOLECULAR CHARACTERIZATION OF ASPHALTENES. PART III. SOLUTION-PHASE AND GAS-PHASE AGGREGATION OF ASPHALTENES...... 147

Introduction ...... 147

Experimental Methods...... 149 Sample Preparation ...... 149 Silver Complexation ...... 150 LTQ-MS...... 150 9.4 Tesla FT-ICR MS...... 150 Mass Analysis ...... 151

Results and Discussion ...... 151 Molecular Weight Distribution ...... 153 Solution-Phase Aggregation ...... 153 50:50 Asphaltene/Maltene Mixture ...... 153 Heteroatom Class Distribution ...... 155 DBE vs. Carbon Number Images ...... 156 Gas-Phase Aggregation...... 159 Asphaltene Aggregates by LTQ-MS ...... 160 Maltene Molecular Weight by TOF-MS ...... 160 Asphaltene Molecular Weight by TOF-MS ...... 162 Effect of Increasing Focus Voltage ...... 163 Concentration Effects on Aggregation ...... 165

CONCLUSION...... 167

CHAPTER 9. IDENTIFICATION OF VANADYL PORPHYRINS IN A HEAVY CRUDE OIL AND RAW ASPHALTENE BY ATMOSPHERIC PRESSURE PHOTOIONIZATION FT-ICR MASS SPECTROMETRY ...169

xi Summary...... 169

Introduction ...... 169

Experimental Methods...... 173 Sample Preparation ...... 173 Atmospheric Pressure Photoionization (APPI) ...... 174 9.4 Tesla FT-ICR MS...... 174

Results and Discussion ...... 175 Elemental Composition Assignment...... 175 Double Bond Equivalents (DBE) Distribution...... 178 DBE vs. Carbon Number Images ...... 183 Heteroatom Class Distributions...... 186

Conclusion ...... 190

REFERENCES ...... 190

BIOGRAPHICAL SKETCH………………………………………………………221

xii LIST OF FIGURES

Figure 1.1. Images of light crude (left), heavy crude (center) and bitumen (right) are shown to illustrate the difference in viscosity between the three feedstocks which pose challenges for recovery and processing. Raw oil sands (top) is the mixture of sand, clays, water and a thick form of crude oil from which bitumen is extracted……………………………………………………………………3

Figure 1.2. Distribution of different compound types found in petroleum. The proportion of saturated hydrocarbons (paraffins) decreases as the molecular weight increases or at higher boiling point……………………………………………..5

Figure 1.3. Schematic of crude oil refinery. Crude oil enters the refinery unit and is first separated by volatility into various fractions which are then processed…………………………………………………………………………………………….9

Figure 1.4. Distillation unit of a typical oil refinery. Further separation, such as cokers, hydrocrackers of fluid catalytic cracking units (FCC) is often performed downstream in separate conversion units……………………………….10

Figure 1.5. An asphaltene deposit formed on the inside of a pipeline in a refinery. Asphaltenes deposit on the surface of the pipeline and reduce the flow of crude oil. Asphaltene deposition has been compared to coronary artery disease, in that both problems result in flow restriction through a pipe (or an artery) and cause major problems further downstream……………….………..…………………………………………………………….15

Figure 1.6. Structure of condensed aromatic ring systems. A pericondensed structure, coronene, is thought to dominate asphaltene structure…………….18

Figure 1.7 SARA fractionation procedure for the separation of heavy oil and residues into fractions of saturates, aromatics, resins and asphaltenes. The asphaltene fraction is first removed from the crude and the deasphalted oil, or maltene fraction, is further fractionated using adsorption chromatography…………………………………………………………………………………19

Figure 2.1. Schematic of electrospray ionization. 2kV voltage is applied to the tip of a capillary through which dilute sample flows. Ions are vaporized into an aerosol spray and desolvation occurs along with dry nitrogen gas…………...22

Figure 2.2. Two-dimensional schematic of the APPI ion source which is coupled to the 9.4 T FT-ICR mass spectrometer. The krypton vacuum ultraviolet gas discharge lamp is drawn on the z-axis along with the heated metal capillary. In practice, the three assemblies are mutually orthogonal…………………………24

Figure 2.3. Photoionization pathways in positive mode APPI. Direct photoionization is shown in (1) but is very limited, since the source is at atmospheric pressure and the photon undergoes approximately 2 x 1010 - collisions per second with atmospheric gases before reacting with the analyte………………………………………………………………………………………………25

Figure 2.4. Schematic of the 9.4 Tesla FT-ICR mass spectrometer located at the National High Magnetic Field Laboratory at Florida State University in

xiii Tallahassee, Florida. Differential pumping is used to reduce the base pressure in the ICR cell to 10-10 Torr to minimize collisions between ions during excitation/detection. Figure provided by the Marshall Research group courtesy of John Paul Quinn…………………………………………………………………27

Figure 2.5. Mass scale expanded zoom insets of positive-ion APPI FT-ICR MS of an Athabasca bitumen HVGO distillate. 14.01565 Da spacings (bottom) represent members of a homologous series which differ only in alkylation (CH2 units) and 2.0157 Da spacings represent compounds differing only by two hydrogen atoms, indicative of different aromaticity (DBE values)…………………………...31

Figure 2.6. Theoretical resolving power for FT-ICR mass spectrometry. Because of the complexity if crude oil, a minimum resolving power much be achieved to facilitate separation and correct identification of isobaric species. The 3.4 mDa split occurs between species with 36 Da nominal mass, but differing by SH4 and C3. The overlap between SH313C and C4 occurs between species weighing 48 Da…………………………………………………………………………………..33

Figure 2.7 Color-coded isoabundance contoured plots of DBE vs. carbon number for Middle Easter heavy crude protonated hydrocarbon species. The image exhibits a missing portion of the DBE and carbon number distribution for the sample which is due to the decrease in resolving power above a certain m/z value. At m/z 497, the 1.1 mDa mass doublet is resolved between a protonated hydrocarbon and isobaric 13C132SH3. However, the next member of the homologous series is not resolved from its [SH313C]+ counterpart and therefore elemental composition occurs erroneously for both hydrocarbon and sulfur species……………………………………………………………………………….34

Figure 2.8. Broadband positive-ion APPI 9.4 T FT-ICR mass spectrum of a Middle Eastern heavy crude. 31,232 mass spectral peaks are resolved at 6 times the signal-to-noise ratio baseline rms noise at an average resolving power, m/∆50% = 600,000…………………………………………………………………………………………..35

Figure 2.9. Broadband positive-ion APPI 9.4 T FT-ICR mass spectrum of a Middle Eastern heavy crude. 31,232 mass spectral peaks are resolved at 6 times the signal-to-noise ratio baseline rms noise at an average resolving power, m/∆50% = 600,000…………………………………………………………………………………………..36

Figure 2.10. Mass-scale expanded segment of positive-ion APPI FT-ICR mass spectrum of a processed vacuum bottom residue, 772 < m/z < 776, showing the monoisotopic peak for an S2 compound at m/z 772.50690 with corresponding 13C1, 13C2 and 34S113C1 isotopic contributions, with agreement between experimental relative abundances and those calculated from the assigned elemental composition (data not shown)……………………………………37

Figure 2.11. Internal calibration mass accuracy for more than 10,000 mass spectral peaks observed at 10 times the signal-to-noise ratio baseline rms noise collected by APPI FT-ICR MS at 9.4 T for European crude. Calculation of the rms mass error for all observed peaks across 350 < m/z <1025 was 260 ppb……………………………………………………………………………………………………39

Figure 2.12. Mass-scale expanded segment of positive-ion APPI FT-ICR mass spectrum of a processed vacuum bottom residue across a 420 mDa window at m/z = 616. The dynamic range of FT-ICR MS allows for observation of low

xiv signal-to-noise signals (zoom inset) simultaneously with high signal-to-noise peaks……………………………………………………………………………………………….41

Figure 3.1. Low resolution linear ion trap mass spectra (LTQ-MS) for an Athabasca bitumen HVGO distillation series. As the boiling point increases, the molecular weight distribution shifts to higher m/z and the molecular weight distribution covers a broader range indicating an increase in complexity associated with higher boiling fractions. At higher molecular weight, the increase in the number of carbon atoms per strucure results in an increase in the number of structural rearrangements (isomers) possible at a given moelcular weight, as indicated by the highest fraction covering the widest molecular weight range……………………………………………………………..49

Figure 3.2. Linear trap mass spectra for the IBP-343 °C (left) and 500-525 °C fraction (right) collected at increasing nebulization temperature. As the sheath gas temperature increases, there is no distinct change in the molecular weight distribution for either fraction. Furthermore, there is no change in the signal magnitude at higher sheath gas temperature. However, this is a low resolution analyzer and does not allow for any changes in speciation at higher temperature……………………………………………………………………………..51

Figure 3.3.a (Top) Linear trap mass spectra for the distillate residue (500-525 °C) collected at optimal sheath gas temperature (325 °C). Both low and high resolution mass spectra were collected for each boiling point range, since low resolution LTQ-MS analysis can not detect compostional changes as a function of nebulization temperature. Because there is inherent discrimination in the number of ions that can be trapped in the ICR cell prior to detection, the molecular weight distribution is truncated and represents a heart-cut of the most abundant species present, centered with the LTQ spectrum. Figure 3.3.b (Bottom) Ultrahigh resolution FT-ICR mass spectra for the distillate residue (500-525 °C) collected at optimal sheath gas temperature (325 °C). Over 20,000 peaks were detected above six times the baseline rms noise between 350 < m/z < 800 with approximately 77 unique mass spectral peaks per nominal mass. An average resolving power of m/∆m50% = 400,000 was achieved at m/z 600………………………………………….52

Figure 3.4. Isoabundance contour plots of double bond equivalents (DBE) versus carbon number for the hydrocarbon class (top) and S1 class (bottom) for the 475-500 ˚C fraction at increasing sheath gas temperature. At lower nebulization temperatures, the carbon number and DBE distribution does not change for either heteroatom class. A sheath gas temperature of 325 ˚C is too high and thermal breakdown of lighter compounds is evident, therefore, 300 ˚C is optimal for compounds boiling between 475-500 ˚C. See text for further discussion………………………………………………………………………………55

Figure 3.5. Isoabundance contour plots of double bond equivalents (DBE) versus carbon number for the hydrocarbon class (top) and S1 class (bottom) for the 500+ ˚C fraction at increasing sheath gas temperature. At lower nebulization temperatures, the carbon number and DBE distribution does not change for either heteroatom class. A sheath gas temperature of 325 ˚C is optimal for compounds boiling above 500 ˚C because it minimizes thermal degradation while efficiently ionizing the higher boiling (heavier) compounds present. See text for further discussion……………………………………………………………………58

xv Figure 3.6 APPI heteroatom class distribution for all classes above 1% relative abundance for all distillate cuts analyzed at their optimal nebulization temperatures. An increase in relative abundance of multiheteroatomic (i.e., S1 and S2) compounds is observed in higher boiling fractions along with a decrease in no or monoheteroatomic classes (i.e., hydrocarbon). Compounds with few or no heteroatoms, such as PAH’s and PAXH’s, for example) that have low molecular weights will have a high vapor pressure and therefore are concentrated in the lower boiling fractions. IBP-343 ˚C exhibits this trend and has the highest relative abundance of hydrocarbons and S1 classes across the entire series………………………………………………………………………..59

Figure 3.7 Color-coded isoabundance contour plots of carbon number vs DBE for four distillation fractions at optimal sheath gas temperature. Four heteroatom classes (hydrocarbon, S1, S2 and O1) are shown for each boiling range to shown how structures evolve within each class as a function of boiling point. Representative core structures are shown for thiophenic and furanic species to help highlight the growth of core structures within a distillation cut. At higher boiling points, the aromaticity also increases, shown here using DBE…………………………………………………………………………………………………..61

Figure 4.1 Broadband postivie-ion APPI FT-ICR mass spectrum of an Athabasca bitumen HVGO distillation cut (475-500 ˚C) at 14.5 tesla. 16,858 mass spectral peaks were observed at 6 times the baseline rms noise, at an average m/∆m50% = 400,000………………………………………………………………….72

Figure 4.2 The theoretical Boduszynski model illustrating the effect of molecular weight and structure on boiling point.1 Atmospheric equivalent boiling point (AEBP) is plotted versus molar mass for model compounds representative of compounds found in crude oil. Within a given boiling point, the paraffin class has the lowest boiling point, followed by naphthenic rings, aromatic rings, alkyl-substituted polyaromatic rings, heteroatom-containing ring aromatic rings and finally, by polar heteroatom-containing polyaromatic rings. To the right of the figure, we have included the decrease in the number of carbon atoms as heteroatom content increases within a given boiling point…………………………………………………………………………………………………74

Figure 4.3 Color-coded isoabundance contoured plots of DBE vs. carbon number for the hydrocarbon class for Athabasca bitumen HVGO distillate cuts. The carbon number abundance distribution maximum (red arrow) shifts from ~ C20 at IBP-343 ˚C to ~C40 at 500-538 ˚C. DBE values show a gradual increase in aromaticity from DBE = ~7 to DBE + ~10 with increasing boiling point………………………………………………………………………………………………...76

Figure 4.4 DBE vs. carbon number images for the S1 class for Athabasca bitumen HVGO distillate cuts. The carbon number abundance distribution maximum shifts from ~C18 at IBP-343 ˚C to ~C39 at 500-538 ˚C but for ~2 fewer carbons than for pure hydrocarbon analogues (Figure 3). DBE values increase from DBE =~5 to DBE =~10 with incresaing boiling point, as for the hydrocarbon class…78

Figure 4.5 DBE vs. carbon number images for the S2 class for Athabasca bitumen HVGO distillate cuts. The carbon number abundance distribution maximum shifts from ~C18 at IBP-343 ˚C to ~C39 at 500-538 ˚C but for ~2 fewer carbons than the S1 class (Figure 4) and ~4 fewer carbons than the hydrocarbons class (Figure 3). DBE values increase similarly from DBE =~6 to DBE= ~11 with increasing boiling point…………………………………………………….…………79

xvi

Figure 4.6 Composite DBE vs. carbon number images for the hydrocarbons, S1 and S2 classes for four of the eight HVGO distillate fraction shown in Figures 3-5. Within each boiling range, each increase in one sulfur shifts to lower carbon number which corresponds to results in ~2-3 fewer carbons per structure……………………………………………………………….…………….…………….81

Figure 4.7 DBE vs. carbon number images for four distillation cuts. Here, for a given carbon number (~24-25), each additional heteroatom is seen to increase the boiling point by ~ 25 ˚C…………………….……………………………………………82

Figure 4.8 DBE vs. carbon number images for the hydrocarbon (APPI), S1 (APPI), and acidic O2 (ESI) classes from the 425-450 ˚C and 475-500 ˚C HVGO distillation cuts of whole Athabasca bitumen. Proceeding from hydrocarbon to S1 for either cut, the carbon number decreases by 2. Polar O2 classes, most likely from carboxylic functionalities, contain 3 fewer carbons than hydrocarbons and 1 fewer than monoheteroatomic S1 classes…………………………..…………………….……………………………………………84

Figure 4.9 Combined DBE vs. carbon number images for all distillation cuts combined for the hydrocarbons class from Athabasca bitumen HVGO. Carbon number and DBE values increase monotonically with increasing boiling point across the entire series. The Boduszynski model is irrefutably supported by this Figure: crude oil composition is continuous in carbon number, DBE and boiling point………………….…………………….……………………………………….……88

Figure 4.10 DBE vs. carbon number images for four the S1 class of the Athabasca bitumen HVGO feedstock for all the distillation cuts combined. The number of aromatic rings corresponding to various DBE values are shown for representative structures. Because the abundance distribution is monomodal (i.e., no “magic numbers”), including significantly abundant species with DBE values intermediate between those of fused aromatic rings, cycloalkyl-ring addition must be invoked to account for the intermediate DBE values………………………………………………….……………………………………………90

Figure 5.1 Broadband positive-ion APPI FT-ICR mass spectrum of a Middle Eastern heavy crude oil 593+ °C distillate fraction. 26,896 mass spectral peaks were observed at 6 times the signal-to-noise ratio baseline rms noise, at an average mass resolving power, m/Δm50% = 580,000 at m/z 800……………………………………………………………………………………………..…..102

Figure 5.2 Broadband positive-ion APPI FT-ICR mass spectra of a full distillation series of Middle Eastern heavy crude oil. An increase in the center of the molecular weight distribution and a broadening of the molecular weight distribution accompanied an increase in boiling point…………………………..104

Figure 5.3 Heteroatom class distribution (heteroatom content) for Middle Eastern heavy crude oil distillation cuts and residue derived from positive-ion APPI FT- ICR MS. Relative abundances are normalized to the most abundant class within each distillate fraction………….………………………………………….…….106

Figure 5.4 The Boduszynski model of the effect of molecular weight and structure on boiling point for heavy crude oil composition. Reprinted with permission………………………………………………………………………………………107

xvii Figure 5.5 Color-coded isoabundance contoured plots of DBE vs. carbon number for the hydrocarbon class of a Middle Eastern heavy crude oil distillation series and residue………………………………………………….………………………….109

Figure 5.6 Color-coded isoabundance contoured plots of DBE vs. carbon number for the S1 class of a Middle Eastern heavy crude oil distillation series and the residue…………………………………………………………………………………………….110

Figure 5.7 Color-coded isoabundance contoured plots of DBE vs. carbon number for the S2 class of a Middle Eastern heavy crude oil distillation series and the residue…………………………………………………………………………………………….112

Figure 5.8 Composite color-coded isoabundance contoured plot of DBE vs. carbon number for the hydrocarbon class for Middle Eastern heavy crude oil distillation series and residue. Each boiling point is normalized to illustrate the global continuum in carbon number and DBE as a function of increasing boiling point…………………………………………………………………………………….113

Figure 5.9 Composite color-coded isoabundance contoured plot of DBE vs. carbon number for the S1 class for Middle Eastern heavy crude oil distillation series and residue. Each boiling point is normalized to illustrate the global continuum in carbon number and DBE as a function of increasing boiling point……………………………………………………………………………………………….114

Figure 6.1 Broadband positive-ion APPI LTQ mass spectra of an asphaltene fraction isolated from a Middle Eastern heavy crude vacuum bottom residue collected over 593+ ˚C. The asphaltene molecular weight distribution ranges from 250 < m/z < 2000……………………………………………………………………….120

Figure 6.2 Broadband positive-ion APPI LTQ mass spectra of asphaltenes isolated from the highest boiling distillate fraction (538-593 ˚C) from Middle Eastern heavy crude. At increasing concentration, the molecular weight distribution is constant over 250 < m/z < 900. The parent distillate covered a molecular weight distribution roughly nearly twice the distillable asphaltene fraction. Most asphaltene molecules self-associate in the crude oil matrix to form nanoaggregates (roughly 8 monomer units) and therefore share volatility properties associated with the aggregate. Therefore, only a small fraction of asphaltene molecules are distillable……………………………………………………122

Figure 6.3 Mass scale-expanded segment of a positive-ion electrospray FT-ICR mass spectrum of the 538-593 ˚C parent distillate and its asphaltene fraction. An increase in spectral complexity is observed for the distillable asphaltenes with a corresponding shift to lower mass defect indicating an increase in aromaticity. Since the mass defect of hydrogen is 0.007994, each addition of a hydrogen (increased saturation) shifts the total mass of a compound +0.007994………………………………………………………………………………………..124

Figure 6.4 Color-coded isoabundance contoured plots of DBE vs. carbon number for the hydrocarbon class derived from positive-ion APPI FT-ICR mass spectra for the distillate fraction collected between 538-593 ˚C (left) asphaltene (right) from Middle Eastern heavy crude. More aromatic compounds are observed in the asphaltene fraction relative to the parent distillate illustrate structural differences between the parent distillate and the distillable asphaltenes,

xviii noteably, an increase in aromaticity as indicated by higher DBE values obtained for the asphaltene fraction……………………………………………………126

Figure 6.5 Color-coded isoabundance contoured plots of DBE vs. carbon number for the S1 class derived from positive-ion APPI FT-ICR mass spectra for the distillate fraction collected between 538-593 ˚C (left) asphaltene (right) from Middle Eastern heavy crude. Overlap between asphaltenes and maltenes show a second, less abundant carbon number and DBE distribution indicating entrainment of non-asphaltene molecules during fractionation………………128

Figure 6.6 Color-coded isoabundance contoured plots of DBE vs. carbon number for the SO class derived from positive-ion APPI FT-ICR mass spectra for the distillate fraction collected between 538-593 ˚C (left) asphaltene (right) from Middle Eastern heavy crude. The asphaltene fraction has nearly twice the aromaticity as the parent distillate, and corresponds to the hydrocarbon and S1 classes…………………………………………………………………………………………129

Figure 6.7. Composite plot of DBE vs. carbon number for the distillable asphaltenes (red) and parent distillate (blue) for the hydrocarbon class. A pericondensed ring system, coronene, is representative of the structure of asphaltene compounds and is plotted as well. The boiling point of coronene is slightly lower than the distillation temperature and therefore coronene is not observed in the distillable asphaltene fraction……………………………………………………………………………………………130

Figure 6.8. Composite plot of DBE vs. carbon number for the distillable asphaltenes (red) and parent distillate (blue) for the S1 class. Dinphthothiophene is representative of the structure of S1 asphaltene compounds and is plotted as well. The boiling point of coronene is slightly lower than the distillation temperature and therefore coronene is not observed in the distillable asphaltene fraction…………………………………….132

Figure 7.1 Broadband positive-ion APPI LTQ mass spectra of a maltene (top) and asphaltene (bottom) fraction isolated from a Middle Eastern heavy crude vacuum bottom residue collected over 593+ ˚C. The asphaltene molecular weight distribution is centered at m/z 1100,, higher than the maltene fraction, centered at m/z 500. However, both fractions cover a similar molecular weight distribution between ~200

Figure 7.2 Mass scale-isolated 5 Da segment of a positive-ion APPI FT-ICR mass spectrum of an asphaltene isolated from a Middle Eastern heavy crude vacuum residue. A 5 Da window reveals the increased complexity observed for asphaltene fractions with over 140 peaks in a single nominal mass unit above six times the baseline rms noise level………………………………………….138

Figure 7.3 Mass scale-expanded segment of a positive-ion APPI FT-ICR mass spectrum of an asphaltene isolation from a Middle Eastern heavy crude vacuum residue. Isobaric species differing in mass by SH4 vs C3 and 13CH332S vs C4 are separated and identified due to the high resolving power afforded by FT-ICR MS. Other MS techniques are not able to routinely achieve resolving power to identify isobars in a broadband mass spectrum………………………139

xix Figure 7.4 Mass scale-expanded segment of a single nominal mass unit at m/z 553 for a maltene (top) and asphaltene (bottom) fractions from a Middle Eastern heavy crude vacuum residue. The mass defect, the difference between the exact mass and nominal mass, differs in spectral position in respect to the composition of the two fractions. Maltenes are more enriched in hydrogen and therefore have a higher mass defect than asphaltenes, which are composed mainly of condensed aromatic rings with little or no alkyl substitution………………………………………………………………………………………140

Figure 7.5 Color-coded isoabundance contoured plots of DBE vs. carbon number for maltene (left), asphaltene (center) and parent residue (right) from a Middle Eastern heavy crude. The DBE distribution and carbon number range of the maltene fraction is identical to the parent residue, indicating that ionization efficiencies differ between maltene and asphaltene molecules………………..141

Figure 7.6 Composite color-coded isoabundance contoured plots of DBE vs. carbon number for S1 and S2 classes from maltene and asphaltene fractions of Middle Eastern heavy crude. When viewed in compositional space defined by a plot of aromaticity (DBE) vs carbon number, asphaltenes and maltenes share similar carbon number space but asphaltenes are shifted to higher aromaticity relative to maltenes. This upward shift is defined by the planar limit for polyaromatic hydrocarbons………………………………………………………………..143

Figure 8.1. Broadband positive-ion APPI FT-ICR mass spectra for Middle Eastern heavy crude, distillate residue (593+ ˚C) and the asphaltene and maltene fractions derived from the residue. Each spectrum is normalized to the highest peak in all four spectra. The maltene fraction covers the exact same molecular weight distribution as the parent residue with comparable signal. However, the asphaltene fraction exhibits much lower signal with a narrow molecular weight distribution…………………………………………………………….152

Figure 8.2. Broadband positive-ion APPI FT-ICR mass spectra for Middle Eastern heavy crude maltenes (top) and a 50% (w/w) mixture of asphaltene and maltene fractions derived from the residue. Both spectra cover a similar molecular weight distribution between 250 < m/z < 950 centered at approximately m/z 550………………………………………………………………………155

Figure 8.3. Heteroatom class analysis for the maltene fraction and a mixture of 50% by weight asphaltenes and maltenes derived from Middle Eastern heavy crude. Both were collected using positive-ion APPI FT-ICR mass spectrometry…………………………………………………………………………………….156

Figure 8.4. Color-coded isoabundance contours for plots of DBE vs. carbon number for the hydrocarbon series of maltenes and a 50% by weight mixture of asphaltenes and maltenes derived from Middle Eastern heavy crude…………157

Figure 8.5. Color-coded isoabundance contours for plots of DBE vs. carbon number for the S1 series of maltenes and a 50% by weight mixture of asphaltenes and maltenes derived from Middle Eastern heavy crude………………………………..158

Figure 8.6. Low-resolution positive-ion ESI LTQ mass spectrum for asphaltene derived from Middle Eastern heavy crude. Because of time-of flight differences for ions of different masses in FT-ICR MS, the molecular weight distribution obtained from a linear trap is a more accurate depiction of the “true” molecular weight of a sample……………………………………………………………..159

Figure 8.7. TOF-MS mass spectra collected on maltene fraction isolated from Middle Eastern heavy crude. A molecular weight distribution between 250 < m/z < 1400 was observed with no significant signal detected from species above 2 kDa……………………………………………………………………………………..160

Figure 8.8. TOF-MS mass spectrum of asphaltene fraction derived from Middle Eastern heavy crude. At a concentration of 500 μg/mL, asphaltene aggregates are observed in the gas phase at approximately eight times the molecular weight of the monomer (~1200 Da). A focus voltage of 100V was used……..162

Figure 8.9. TOF-MS mass spectrum of asphaltene fraction derived from Middle Eastern heavy crude. At a concentration of 500 μg/mL, asphaltene aggregates are observed in the gas phase at approximately eight times the molecular weight of the monomer (~1200 Da). A focus voltage of 120V was used……..163

Figure 8.10. TOF-MS mass spectrum of asphaltene fraction derived from Middle Eastern heavy crude. At a concentration of 500 μg/mL, asphaltene aggregates are observed in the gas phase at approximately eight times the molecular weight of the monomer (~1200 Da). A focus voltage of 160V was used. Here, the low molecular weight distribution is bimodal, with a monomer and dimer distribution observed due to the increased thermal energy of the aggregated asphaltenes……………………………………………………………………………………..164

Figure 8.11. TOF-MS mass spectrum of asphaltene fraction derived from Middle Eastern heavy crude at 80V focus voltage. At a factor of 10 lower in concentration (50 μg/mL), asphaltene aggregates are observed at a lower m/z value. The monomer distribution is centered at m/z 850 and the aggregate distribution is eight times higher and corresponds to a stable asphaltene octamer. ………………………………………………………………………………………….165

Figure 8.12. TOF-MS mass spectrum of asphaltene fraction derived from Middle Eastern heavy crude at 100V focus voltage. The monomer distribution is centered at m/z 850 but the aggregate distribution shifts to lower m/z and shows a slightly bimodal distribution, indicating the presence of two stable core aggregates containing five and seven asphaltene monomers……………166

Figure 8.13. TOF-MS mass spectrum of asphaltene fraction derived from Middle Eastern heavy crude at 150V focus voltage. The monomer distribution is centered at m/z 850 and the aggregate distribution once again becomes monomodal, centered at m/z 4200 consistent with a stable aggregate containing five asphaltene monomers………………………………………………….167

Figure 9.1. Possible core structures of vanadyl porphyrins found in petroleum. The two major structural forms, DPEP (CnH2n-28N4VO) and Etio (CnH2n-30N4VO), are shown at the top with elemental compositions assigned from experimental mass measurements (see text). DBE (double bond equivalents) is the number of rings plus double bonds to carbon (DBE = c - h/2 + n/2 +1 for elemental composition, CcHhNnOoSs)……………………………………………………..171

Figure 9.2. Broadband positive-ion APPI FT-ICR mass spectrum of an Athabasca bitumen raw asphaltene fraction without preconcentration or isolation. 14,475 mass spectral peaks were observed at 6 times the signal-to-noise ratio baseline rms noise, at an average m/Δm50% = 400,000. An unknown

xxi contaminant peak at m/z 637, presumably resulting from the asphaltene fractionation process…………………………………………………………………………175

Figure 9.3. Mass scale-expanded segment of a positive-ion APPI FT-ICR mass spectrum of an Athabasca bitumen raw asphaltene, 527 < m/z < 529, showing the monoisotopic peak for a DPEP vanadyl porphyrin at m/z 527.20104 with corresponding 13C1 and 13C2 isotopic contributions. Note the close agreement between experimental relative abundances and those calculated from the assigned elemental composition………………………………………………………….177

Figure 9.4. Mass scale-expanded segment of a positive-ion APPI FT-ICR mass spectrum of a South American heavy crude, 541 < m/z < 542, showing the monoisotopic peak for a DPEP vanadyl porphyrin at m/z 541.21665 with corresponding 13C1 isotopic contributions. Note the close agreement between experimental relative abundances and those calculated from the assigned elemental composition……………………………………………………………………….178

Figure 9.5. DBE distribution for vanadyl porphyrins in a raw Athabasca bitumen asphaltenes fraction. The DPEP class corresponds to DBE = 18 and is the most abundant structure. Etio structures are also observed and form radical cations and protonated species of comparable abundance. Di-DPEP, Rhodo- Etio and Rhodo-DPEP structures are also seen………………………………………179

Figure 9.6. DBE distribution for vanadyl porphyrins in a whole South American heavy crude oil. In contrast to the asphaltene fraction, the DPEP (DBE=18) and Etio (DBE=17) types are present in almost equal abundance. The etio porphyrins also protonate. Di-DPEP, rhodo-etio and rhodo-DPEP structures are also observed as radical molecular cations…………………………………….180

Figure 9.7. Color-coded isoabundance contoured plots of DBE vs. carbon number for Athabasca bitumen asphaltenes. The image exhibits multiple domains and higher rms error (0.88 ppm) if vanadyl porphyrins are misassigned as O2 species (left) rather than separate images for the correctly assigned elemental compositions with rms errors of 0.21 ppm for the vanadyl porphyrins and 0.31 ppm for the O2 species (right)………………………………..184

Figure 9.8. Color-coded isoabundance contoured plots of DBE vs. carbon number for a South American heavy crude oil. Interpretation is as for Figure 6.7…………………………………………………………………………………………………..185

Figure 9.9. Heteroatom class distribution for Athabasca bitumen asphaltenes. Vanadyl porphyrins are observed at ~3% relative abundance without preconcentration or isolation……………………………………………………………..186

Figure 9.10. Heteratom class distribution for a South American heavy crude oil for all species of >1% relative abundance, including vanadyl porphyrins……….187

xxii ABSTRACT

Eventually, the world will deplete the global oil supply and a new form of energy will supplant fossil fuels as the main global energy source. However, technological advances are still decades away from finding a unified solution to meet the energy needs of the global community. Wind and solar energy can be harvested, but only provide a fraction of the required energy. In the meantime, heavy conventional and unconventional crude oil production has increased due to the depletion of low viscosity, sweet crude. Oil companies sell molecules. Comprehensive compositional characterization of refinery feeds allows for production and refinery strategy development. Oil companies convert high-boiling fractions to increase product yields for valuable, low-boiling fractions. Development of chemical processes to maximize profits requires an exhaustive characterization of each molecule in heavy crude fractions. Bulk property measurements combined with analytical techniques are limited in their ability to separate and characterize each of the tens of thousands of compounds in a single crude oil. The use of mass spectrometry has facilitated the characterization of low-boiling, light crude oils; however, more complex, heavy feeds require extensive separation techniques to produce meaningful compositional information. Ultrahigh resolution FT- ICR mass spectrometry provides the most detailed, comprehensive examination of high-boiling fractions of crude oil. Chapter 1 provides an introduction to heavy crude oil and bitumen processing and refining. In Chapter 2, a brief introduction to FT-ICR principles and the figures of merit which make it indispensable for complex crude oil mixtures. Chapter 3 establishes the correlation of atmospheric pressure photoionization (APPI) nebulization temperature with boiling point for an Athabasca bitumen HVGO distillation series. Here, we establish optimal

xxiii temperatures which are used for the following chapters to ensure complete desorption/ionization while assuring thermal reactions do not occur in the ionization process which would affect subsequent data interpretation. Chapter 4 begins a four part examination of heavy oil composition and investigates the relationship between molecular weight, structure and boiling point of nonpolar and polar species boiling between 427-538 ˚C. Extensive characterization of a heavy vacuum gas oil (HVGO) distillation series provides the first comprehensive test of the Boduszynski model of heavy oil composition. Both electrospray ionization (ESI) and APPI provide insight into the evolution of polar and nonpolar species and show the continuity of crude oil composition as boiling point increases. In Chapter 5, part II extends the crude oil continuum to the limits of distillation through detailed examination of a Middle Eastern heavy crude distillation series. Extension of the Boduszynski model transitions into a detailed examination of the composition of nondistillable residues. Chapter 6 characterizes asphaltenes, a solubility fraction of crude oil known to be high-boiling, highly polar and problematic for production, transport and refining of crude oil through deposition, catalyst fouling, and viscosity increases, to name a few. Part III introduces a new fraction of asphaltenes we deem “distillable asphaltenes”. Here, we further prove the Boduszynski model through the discovery of asphaltene compounds which boil much lower than previously thought. Chapter 7 defines asphaltene compositional space in conjunction with their counterpart fraction, maltenes. Asphaltenes and maltenes are separated by their solution-phase behavior through solubility differences in paraffinic solvents (ie, heptane or pentane). However, for twenty years, asphaltene molecular weight has been the subject of a heated debate between those who think they are high molecular weight (<10 kDa) and those who think they are relatively low in molecular weight (> 2 kDa).

xxiv Mass spectrometry is used to show that asphaltenes are not abnormally high in molecular weight and in fact share carbon number space with maltenes. Compositional differences between asphaltene and maltene species show that asphaltenes are shifted to higher DBE values at the same carbon number than maltenes. Results are combined with reported results in the literature to provide a unified theory of asphaltene composition. Chapter 8 investigates the phenomena of asphaltene self- association (aggregation) to form stable aggregate structures through noncovalent interactions. Here, we employ time-of-flight mass spectrometry, which allows for much wider molecular weight distributions in a single mass spectrum. We are able to identify and show that asphaltenes are aggregated at concentration levels above those routinely used for mass spectral analysis through a bimodal distribution for the monomer and aggregate at roughly eight times the monomer molecular weight. Solution-phase aggregation of asphaltenes is also examined through high resolution analysis of a mixture of 50/50 asphaltene/maltene by weight to explore the composition of asphaltenes in a mixed matrix. Chapter 9 identifies and characterizes metal-containing petroporphyrins in raw asphaltene and whole crude oil for the first time by FT-ICR MS. Petroporphyrins are important for removal prior to refining because they are highly corrosive and are known to deactivate catalysts used in conversion processes. Porphyrins have been extensively characterized with other techniques, but tedious separation and isolation is required. Here, different structural classes of vanadyl porphyrins are characterized by APPI FT-ICR MS without prior separation or isolation.

xxv CHAPTER 1. INTRODUCTION TO HEAVY OIL AND ASPHALTENES

As the global supply of light, sweet crude is exhausted, refinery feedstocks are exceedingly shifting towards heavy conventional and unconventional crude. Highly viscous crudes, such as bitumen, are rapidly displacing light ends on the global market. Lighter crude oil has a low viscosity, low heteroatom content, contains a high percentage of desirable low molecular weight hydrocarbons and therefore, is more expensive. High gravity, high boiling, low solubility and heteroatom-rich, heavy feeds introduce enormous technical processing challenges. First, compounds that compose heavy crude are higher in molecular weight and heteroatom content than light crudes. Heteroatoms are organic compounds of nitrogen, oxygen and sulfur and trace metals such as nickel, vanadium, iron and copper that are responsible for a multitude of problems encountered throughout oil production and refining. Second, heavy crude oil is more viscous than light crude oil and therefore has a greater resistance to flow, requiring additional measures for transporting oil to a refinery. Because of the wide range of chemical moieties present in crude oil, a vast number of techniques determine processing techniques for different feeds. Understanding the composition, chemical and physical properties of petroleum, heavy oil and bitumen is paramount to meet future energy needs.

Light Crude, Heavy Crude and Bitumen

Light Crude. Conventional (or light crude) has a low viscosity and therefore flows easily in pipelines and contains a low amount of heteroatoms – nitrogen, oxygen, sulfur and metals (vanadium, nickel and iron). Light crude produces high yields of paraffinic low-boiling distillates, such as naptha and gasoline, with even the heaviest fractions of light crude containing mainly large paraffin molecules or alkyl chains. 2 Heteroatom-containing compounds are less volatile are concentrate in

1 higher boiling fractions. 3 Crude oils can have very different properties, (e.g., viscosity, specific gravity) but the carbon and hydrogen content remains relatively constant for all crude, usually between 83-87% carbon and 11-14% hydrogen by weight. 3 The atomic ratio of hydrogen to carbon (H/C) increases with molecular weight indicative of an increased amount of condensed ring systems composed of aromatic and naphthenic rings. Heavy crude. Heavy crude oil has a higher viscosity than light crude oil, and therefore requires enhanced recovery techniques for removal from reservoirs. 3 Generally less saturated, heavy crude molecules contain more ring systems and heteroatoms than light crudes. 2 Most heavy oil requires heat or dilution to flow to a well or through a pipeline. 4, 5 Figure 1.1 shows images of light crude, heavy crude, oil sands and bitumen derived from oil sand to highlight the viscosity differences between the three crude oil types. Table 1.2 compares the properties of conventional crude oil and bitumen. Upgrading converts bitumen and heavy oil into a product with a density and viscosity similar to light crude oil.5, 6

Bitumen. Bitumen is a thick, highly viscous form of crude oil that is found mixed with sand, clay and water in what are called “oil sands”. 7 Usually, bitumen refers to petroleum with a density greater than ~960 kg/m3. 7 Oil sands are saturated mixtures of bitumen, water, sand and clay. At room temperature, bitumen and fine-grained quartz sand mix together with clay particulates and form a dark, sticky mass called oil sands. 3, 7 Bitumen is a sticky, highly viscous form of crude oil extracted from oil sands that requires dilution with lighter hydrocarbons or heating in order to flow through pipelines. Bitumen lacks low-boiling fractions but may encompass a wide range of paraffinic, naphthenic and aromatic carbon distributions and contains a high concentration of heteroatoms. 2 3 Compared to conventional crude oil, bitumen requires additional upgrading prior to refining along with dilution with lighter hydrocarbons and/or heat for transport through pipelines. 8

Table 1.2 Properties of Light Crude Oil, Heavy Crude Oil and Bitumen3, 7, 9

Property Light Crude Heavy Crude Bitumen

Elemental Analysis (wt%) Carbon 86.0 83.0 Hydrogen 13.5 10.3-10.6 Nitrogen 0.2 0.3-0.5 0.4-0.5 Oxygen <0.5 <0.1 0.9-1.1 Sulfur <2.0 3.0 4.9 Nickel (ppm) <10.0 16 250 Vanadium (ppm) <10.0 50 100

Molecular Weight (Da) 540-800

Fractional Composition (wt%) Asphaltenes 17-25 <10.0 Resins 29-35 <20.0 Aromatics 32-35 >30.0 Saturates 12-16 >30.0

Viscosity, cP 38˚C/100˚F <200 750,00 100˚C/212˚F 11,300 Specific Gravity 0.85-0.90 0.89 1.03

Crude Oil Composition

Petroleum is composed of alkanes, naphthenes and aromatic compounds. Figure 1.2 presents a model of the distribution of various 3 compounds in petroleum. Alkanes, or paraffins, (CnH2n+2) are saturated hydrocarbons composed entirely of straight or branched alkyl chains. Naphthenes are saturated hydrocarbons ring structures that may have varying degrees of alkyl substitution. Aromatic hydrocarbons contain one or more conjugated five- to six-carbon member rings, such as benzene or naphthalene and may be bonded to naphthenic rings and alkyl side chains. 3 Heteroatoms are atoms other than carbon and hydrogen found 4 in crude oil, such as nitrogen, oxygen, sulfur and metals, such as nickel and vanadium.

Figure 1.2. Distribution of different compound types found in petroleum. The proportion of saturated hydrocarbons (paraffins) decreases as the molecular weight increases or at higher boiling point.3

Environmental Concerns. Heteroatoms are highly problematic in crude oil but only account for 5-15% by weight of petrocompounds. For example, upon combustion in automobile engines, sulfur-containing compounds form sulfuric acid and cause acid rain. Nitrogen and metals poison catalysts used in refinery processes. Heteroatom characterization is essential to the development of effective removal processes. The increased emphasis on minimizing air pollution has led to substantial modifications to the formulation of motor fuel. 9 As previously mentioned,

5 sulfur combusts to produce sulfuric acid which then causes acid rain. For this reason, the characterization of the structure, composition and functionality of sulfur in petrocompounds is of utmost importance for the development of removal strategies. Aromatic compounds, such as benzene, are especially important because they are responsible for particulate emission and exhaust smoke; however, aromatics increase the octane rating of fuel, which is a measure of a fuels tendency to burn in a controlled manner.9 Consequently, lowering the concentration of aromatic compounds in gasoline requires new ways of achieving high octane ratings and reduces the need for upgrading.

Geologic Origin of Crude Oil and Bitumen

All forms of crude oil, conventional and bitumen alike, are formed from the remains of lacustrine plants and bioorganisms that deposited and buried in the ocean floor. Over millions of years, layers of earth formed on top of the biomass, increasing temperature and pressure. Simpler chemicals such as hydrocarbons, water, carbon dioxide and hydrogen sulfide were created from the conversion of the integrated organic material. 5 The biological origin of crude oil results in a complex mixture of tens of thousands of compounds. 2 Two kinds of liquid hydrocarbons are found in crude oil - light hydrocarbons and heavy hydrocarbons. Light hydrocarbons contain a few carbons surrounded by hydrogen atoms, and heavy hydrocarbons, contain more carbon atoms and fewer hydrogen atoms and often form condensed ring structures. 8 On a geologic time-scale, hydrocarbons migrated to the earth’s surface but were trapped in porous rocks, mainly sandstone, beneath impermeable rock layers. 5 An oil reservoir is really a porous, permeable sedimentary rock in which gas and oil are trapped and is capped by impermeable rock or salt dome, preventing further migration. Petroleum geologists explore for possible rock structures where oil could be trapped.

6 Oil sands, on the other hand, are more debated. There are two main branches of thought on the origin of bitumen. One thought believes that bitumen is a precursor of crude oil since it contains roughly the same concentration of metals as other crudes from the same region and has a low coking temperature, indicating that bitumen is “autochthonous” (indigenous) to its region. The opposing view believes that bitumen migrated to its present locations more than 50 million years ago. Huge volumes of oil migrated east and upwards towards the earths surface through hundreds of kilometers of rock and settled in large sandstone deposits. Bacterial degradation and pressure changes removed lighter constituents, leaving behind heavier compounds. 5, 7, 8 Since bacteria preferentially degrade simple hydrocarbons, an increased abundance of heavy hydrocarbons and organic sulfur occurs in bitumen. The two largest oil sands resources are located in Canada and Venezuela. 7 Canada’s Alberta province contains 2.5 trillion barrels in the ground, with the world’s largest known hydrocarbon resource, the Athabasca deposit containing more than 1.3 trillion barrels alone. 5, 7

Recovery of Crude Oil and Bitumen

Crude Oil Recovery. Oil companies want to maximize the yield of light hydrocarbons since these fractions are the most valuable and require the least amount of processing. As heavier crude supplants light crude on the global market, more processing and conversion techniques are necessary to convert heavy feeds into more valuable lighter products. 9 Some heavy oil can be produced by drilling wells and using pumps to lift the oil while reservoir pressure is maintained through water injection, recovering roughly 20% of heavy oil in the reservoir. 5 Bitumen Recovery. Bitumen cannot be produced from a well unless it is heated or diluted. In situ recovery techniques are used for recovery of deep deposits and shallow deposits are recoverable by surface mining techniques. Approximately two thirds of the total deposit is too 7 deep for surface mining and too shallow for thermal techniques, which accounts for the disparity between oil that is the ground and oil that is recoverable. 10, 11 Bitumen Extraction. There are several technologies used to extract oil sands and bitumen and the depth of the deposit determines which method is used. Surface deposits, those that are less than 50 m from the surface, are mined in vast open pits. Deeper deposits require in situ recovery techniques which reduce oil viscosity and facilitate oil flow. 5, 12 Bitumen is then separated from mined oil sands. The Clark hot water method separates bitumen from the other components in oil sands. Oil sands are mixed with hot water and form a slurry of oil sand and hot water which is then hydrotransported to the extraction plant. Tiny air bubbles trapped in the bitumen to form a thick froth at the top, water falls below the froth with heavy sand falling to the bottom in large separation vessels. The majority of the bitumen is entrained in the froth layer, which is then mixed with solvent and centrifuged to remove water and dissolved salts. 5 Sand and water are byproducts of bitumen production referred to as tailings, and pose a serious environmental problem. Sand is used to help fill in mined-out areas in the mine site, but water now contains sand, clay and traces of bitumen and therefore cannot be released into the water table. Instead, massive settling ponds, visible from outer space, collect the waste water isolated from the natural water bodies. Bitumen that floats to the surface is skimmed off, sand settles to the bottom but clay particles remain suspended in the water before eventually settling to the bottom. Remaining water can then be recycled back to the extraction plant, if possible. 5, 12

Crude Oil Refining

Refinery processes can be classified into three general types : separation, where the feedstock is divided into various fractions; conversion, where economically viable products are produced through 8 alteration of the feedstock constituent molecules; and finishing, where product stream are purified. 3 Figure 1.3 shows a general overview of high-conversion refinery similar to many in the United States. 8 Crude oil is first sent to a desalter where clean water removes any dissolved salts. 3, 8 Distillation separates crude oil is then separated by distillation into straight-run fractions which are then characterized based on boiling point. 9

Figure 1.3. Simplified layout for a high conversion refinery in the United States adapted from8. Here, we show a refinery configured for maximum fuel production.

Distillation. Heteroatom composition varies greatly between fractions and determines the amount and severity of conversion that is required for each distillate cut. Distillation limits the molecular-weight

9 range of each compound type and therefore can contain only a limited range of chemical species. 2, 13 First, crude oil is heated in a distillation column at atmospheric pressure and different products boil off at different temperatures. Light fractions such as naphtha and liquid petroleum gases are removed at the lowest temperatures and collect at the top of the column and higher boiling species, such as heavy fuel oil and residues, are recovered at high temperatures ( < 1000 ˚F) and settle to the bottom. 8 The residue from the atmospheric distillation is then sent to a vacuum distillation unit for increased recovery of lighter fractions. Vacuum is pulled from the top of the tower through a steam ejector or by vacuum steam. The overhead stream, called light vacuum gas oil, consists of lube base, heavy fuel or can be fed into a conversion unit; heavy vacuum gas oil is pulled from a side unit on the tower and the vacuum residue can be sent to a coker or visbreaking unit for further processing or left simply as asphalt. 8 Figure 1.4 shows a general schematic of a typical crude oil distillation.

Conversion. Oil companies increase the amount of high-demand transportation fuel from a barrel of crude by conversion of heavier fractions into light hydrocarbons, which have lower molecular weights, lower boiling points, lower density and higher H/C ratios. Decomposition, rearrangement or recombination of the molecular structure of heavy molecules increases the amount of light fractions. 3 The majority of conversion processes produce light hydrocarbons through carbon rejection; however; hydrotreating and hydrocracking increase the H/C ratio through hydrogen addition. 8 Carbon rejection techniques breaks carbon – carbon bonds to break molecules into two smaller molecules, one with a higher, more desirable H/C ratio and one

11 with a lower H/C ratio. The more condensed, polyaromatic hydrocarbon that is formed with the lower H/C ratio can condense to form coke. 8 Hydrotreating. Hydrotreating removes heteroatoms at low temperatures whereas hydrocracking is a catalytic, high-pressure, high- temperature process for conversion of petroleum feeds. 3, 9 In hydrotreating reactions, there is minimal carbon-carbon bond breaking. Reactions include hydrodesulfurization and hydrodenitrogenation, which 8, 9 remove sulfur and nitrogen and produce H2S and NH3. Other hydrotreating reaction include hydrodemetallation, saturation of aromatic compounds, saturation of olefins and isomerization. Hydrocracking, on the other hand, breaks carbon-carbon bonds to increase naphtha and middle distillate (jet and diesel) yields and produces olefin plant feeds and ultra-clean lube stocks. 5, 8

Analytical Characterization of Distillate Fractions

Bulk property measurement. The complexity of heavy crudes poses significant challenges for correlating physical/chemical properties to compositional trends. Therefore, bulk property measurements on refinery feeds such as specific gravity, refractive index or viscosity (pour point or oxidation stability) are used to anticipate behavior during processing. 3 The boiling-point distribution, density or API gravity and viscosity are the most important properties of a crude oil. 3 A boiling profile allows determination of the yield of various distillation cuts from a particular feed and therefore can determine how much transportation fuel can be made before any conversion processes are required. Density is used to estimate the paraffinic nature of the crude and viscosity measurements help determine if the crude has a tendency to form residues that will restrict flow through pipelines. The higher the boiling point of a fraction, the more difficult it is to analyze its composition due to the increased complexity. Higher boiling points contain higher molecular weight species, which have more carbon 12 atoms per structure and therefore the number of structural rearrangements increases rapidly. 14 The reduced complexity of low boiling fractions facilitates use of a single analytical technique to determine molecular composition. However, up until recently, the complexity of boiling fractions required more than one analytical technique to elucidate the same compositional information. The overwhelming complexity of the hydrocarbon constituents in higher boiling fractions as well as the inclusion of sulfur, oxygen and nitrogen containing compounds makes isolation of individual compounds difficult.3 Light Fractions. Naphtha has a normal boiling range between 40- 220 ˚C and composition can be determined readily using gas chromatography (GC).2, 3, 9 Naphtha is also referred to as raw gasoline. Naphtha makes up approximately 20% (w/w) of crude oil and approximately ~50% alkane, ~40% napthene and ~10% aromatic structures.9 The H/C ratio ranges between 2.0-2.2. Gas chromatography (GC) and gas chromatography/mass spectrometry (GC/MS) can be used for both the light and heavy naphtha’s Middle Distillates. Middle distillates boil between 220-345 ˚C and are also known as kerosene and jet fuels, and make up ~10% of crude oil. Slightly more than 60% of the compounds in middle distillates are naphthenic ring systems. Analysis using GC/MS becomes challenging. GC can no longer separate all of the components complicating mass analysis. Chromatographic techniques such as liquid chromatography (LC) and supercritical fluid chromatography (SFC) can help determine the composition of middle distillates based on saturated, aromatic and polar compound groups.2, 3, 13 VGOs. Vacuum gas oils (VGOs) boil between 345-540 ˚C and constitute ~25-30% of a crude oil. Since VGOs have a lower paraffinic content, the H/C ratio is lower than in the middle distillates, typically between 1.8-1.9. Direct analysis of vacuum gas oils becomes complicated with a high concentration of sulfur and nitrogen-containing compounds. 13 Preparative separation techniques such as high-performance liquid chromatography (HPLC) fractionates compound classes into distinct fractions for subsequent analysis by mass spectrometry or spectrographic techniques, such as nuclear magnetic resonance (NMR).2, 13 Residue. Vacuum residues boil at temperatures greater than 500 ˚C and therefore contain the highest molecular weight species found in a crude oil. An associated decrease in the amount of free alkanes in the residue results in a decreased H/C ratio of approximately 1.4.2, 13 Extensive separation techniques have been required in the past to completely characterize residues by spectroscopic methods. Nondistillable residues. Nondistillable residues, the material that is completely nonvolatile and remains after all distillable material has been removed, remains in the condensed phase above 700 ˚C and represent the most complex fraction of crude oil. They have very limited solubility and volatility and therefore are difficult to analyze. Because they can make up 20-45% of crude, knowledge of their composition is critical. However, up until recently, compositional analysis of the highest boiling fraction of crude oil was very limited due to their complexity. The asphaltene fraction, long believed to be completely nonvolatile, concentrates in the nondistillable residues and is discussed in-depth in the following section. However, as we will discuss in chapter 8, there are

some asphaltene compounds that are volatile.

The Bottom of the Barrel : Asphaltenes

Nondistillable residues of crude oil are the most complex of all crude oil fractions. 2, 3, 9 Between 40-70% by weight of the nondistillable residue is attributed to the asphaltene fraction of crude oil. 3, 15 Compounds found in residues contain several heteroatoms and account for 30-60% of the total sulfur and 70-90% of the total nitrogen and nearly all of the total metal in a crude oil. 3, 15 For decades, the chemical 14 composition and structure of the molecules that make up the nondistillable fraction of crude oil has eluded analytical chemists. Without a doubt, the complexity of residues has complicated analytical techniques routinely used to characterize lighter fractions of crude oil.

What are asphaltenes?

Petroleum asphaltenes are dark brown/black, crumbly solids produced by the treatment of crude oil with low-boiling, paraffinic hydrocarbon solids. 2, 3, 15, 16 Asphaltenes are insoluble in nonpolar hydrocarbon solvents (e.g., n-heptane or n-pentane) but soluble in aromatic solvents such as toluene and benzene. The portion of crude oil that is soluble in paraffinic solvents is called the maltene fraction, or simply de-asphaltened oil. 2 Asphaltenes have a high concentration of heterocompounds and trace metals and are the most refractory portion of crude oil and are often resistant to chemical treatment. 17 Therefore, heavy oil and bitumen have high asphaltene content making asphaltene composition critical for removal. Problems associated with asphaltenes. Asphaltenes are problematic in the oil field and the refinery, causing deposition in nearly every aspect of petroleum production. Figure 1.5 shows a picture of an asphaltene deposit in a pipeline. Clogged wells, flowlines, surface facilities as well as the formation of deposits below the surface of the well are just a few production-level problems attributed to asphaltenes. 18 Asphaltenes cause problems on the molecular level as well as at the production level. A fundamental knowledge of asphaltene properties and composition is crucial to prevent and predict problems caused by asphaltenes in the oilfield and refinery. On the molecular level, the chemical and physical properties of asphaltenes can determine the types of processing treatments used on a crude. Molecular characterization techniques emphasize molecular weight and compositional determination and are more relevant for downstream processes. For example, catalysts 15 are widely used in refineries, but asphaltene molecules can block active sites on catalysts and deactivate or poison them, especially at high temperatures and pressures where asphaltenes exist. On the larger scale, asphaltene behavior during transport and production needs to be understood as well.

Figure 1.5. An asphaltene deposit formed on the inside of a pipeline in a refinery. Asphaltenes deposit on the surface of the pipeline and reduce the flow of crude oil. Asphaltene deposition has been compared to coronary artery disease, in that both problems result in flow restriction through a pipe (or an artery) and cause major problems further downstream. However, arterial asphaltene deposition is questioned by many in the oil industry and is currently being investigated.

Asphaltene aggregation. Asphaltenes self-associate and form aggregates in dilute toluene solutions and are thought to be aggregated in whole crude oil.19 Techniques are used to understand asphaltenes on the large scale involve careful examination of the colloidal properties of asphaltenes and their tendency to aggregate at low concentrations. 17, 18 Asphaltenes can flocculate of precipitate under a variety of conditions. For example, a decrease in pressure has been shown to cause asphaltene deposition in undersaturated crude oils.19, 20 The addition of solvents

16 during any stage of production can also cause asphaltenes to precipitate out of solution, much as they are generated in the laboratory through excess addition of paraffinic solvents. Erroneous measurements of asphaltene molecular weight are attributed to aggregation since the onset of asphaltene self-association occurs at concentrations much lower than many techniques used for molecular weight measurements. 21

Asphaltene Molecular Weight. One of the earliest parameters examined to correlate was the molecular weight of asphaltenes to help correlate their behavior.17 This topic has been heatedly debated over more than 20 years but recently has been essentially resolved as multiple techniques have consistently agreed that asphaltene molecular weight is less than 1 kDa.2, 18, 19, 22-28 The controversy over asphaltene molecular weight arises due to the tendency of asphaltene molecules to self- associate and form aggregates at very low concentrations.29 The controversy over molecular weight stems from several analytical techniques that are poorly-suited for asphaltenes. First, vapor pressure osmometry (VPO) is operated at concentrations that are nearly two orders of magnitude higher in concentration than the apparent onset concentration for asphaltene aggregation. 19 VPO measurements are the results of measurements made on the weight of the aggregate, not the monomer. Second, the variability in laser power and high surface concentration used in laser desorption/ionization mass spectrometry has reported erroneously high molecular weights of asphaltenes. 30, 31 Gas phase aggregation can produce molecular weights by LDI MS and lead to incorrect data interpretation.32 Lastly, size exclusion chromatography produces high molecular weights for asphaltenes because several factors inherent to the technique impart error into the measurement. First, chromatography columns are incompatible with toluene which, by definition, is the best solvent for asphaltenes. It is not known how asphaltene aggregation changes or occurs in other solvent systems, since asphaltenes are not fully soluble in other solvents. Also, polystyrene, a 17 standard, behaves differently and exhibits much different behavior than asphaltenes and therefore is not suited to optimize the separation. Solvents such as N-methyl pyrrolidinone have been used and are known to cause flocculation of more than half of the asphaltene sample. 19, 32

Asphaltene Structure. Asphaltene carbon is slightly more than half aromatic and slightly less than half saturated as determined by 13C NMR techniques. 19, 33 The aromatic carbon in asphaltenes has been determined to be pericondensed and not catacondensed as once believed.33 Figure 1.6 shows two different structures, on pericondensed and one catacondensed. Pericondensed rings or pericyclic structures are the main forms of asphaltene structure because they are enriched in aromatic sextet carbon, which is more stable that isolated double bond carbons. 18, 19, 34 The fused aromatic core of asphaltenes has been determined to have a diameter of approximately 10 Ǻ, which corresponds to six or seven fused aromatic rings. 35

Figure 1.6. Structure of condensed aromatic ring systems. A pericondensed structure, coronene, is thought to dominate asphaltene structure.

Asphaltene Separation. Because asphaltene content is an important factor in determining the processing and refining paths of crude oil, a convenient laboratory method has been developed to quantify the asphaltene fraction.18 The saturate-aromatic-resin-asphaltene method (SARA) was developed to reduce the immense complexity of heavy crude and residua.3, 15, 36, 37 The SARA fractionation method is a simplified version of the United States Bureau of Mines-American Petroleum Institute (USBM-API) method and uses coordination chemistry and adsorption chromatography to fractionate crude oil into chemically significant fractions for compound type analysis. 3, 37 A schematic of the SARA fractionation technique is shown in Figure 1.7.

Originally, the SARA method was developed to further fractionate nondistillable fractions but has been applied to heavy oil and bitumen. 3 Asphaltenes are first removed by the addition of a paraffinic solvent and the deasphaltened oil, or maltene fraction, is further separated using adsorption chromatography.2 Each component is removed from a silica column by flushing with various solvents. SARA fractionation is a widely used separation scheme for heavy crude oil compositional analysis due to its reproducibility as well as its applicability to the heaviest feedstocks, such as bitumen and nondistillable residues. However, a limitation of SARA fractionation is the careful and complete separation of the asphaltene and maltene fractions. Asphaltenes that are not removed from the maltene fraction can adsorb onto chromatography columns and can alter any subsequent analysis on the resin or aromatic fractions.

20 CHAPTER 2. CHARACTERIZATION OF HYDROCARBON RESOURCES USING HIGH RESOLUTION FT-ICR MASS SPECTROMETRY: A PRIMER

Crude oil composition can affect product yields and quality, and market prices can influence process operating strategies. Therefore, it is important to understand the complex composition of crude oil as completely as possible. However, many routinely used analytical techniques are limited by the complexity of heavy ends and residues. For material boiling below 177 ˚C (350 ˚F) gas chromatography (GC) techniques identifies and quantitates volatile species. Compounds boiling between 177-527 ˚C (350-980 ˚F) can be classified by molecular type, but exact molecular structure is complicated by the increased complexity. The increased complexity of material volatile only <527 ˚C makes even group-type identification challenging. 38 Models to elucidate the structure of petrocompounds have been used in the past. For example, Quann and Jaffee introduced a structure-oriented lumping method to describe the composition and properties of hydrocarbon samples. 39 However, these techniques lack the ability to identify and speciate each of the thousands of samples in crude oil. Such a complex mixture requires a highly sophisticated analytical technique. Fourier transform ion cyclotron resonance mass spectrometry is the only mass spectrometric technique with enough resolving power to identify and characterize individual compounds from a complex crude oil matrix. With the invention of electrospray ionization by John B. Fenn, the use of mass spectrometry as an analytical technique exploded.40 Electrospray ionization provided the means by which to ionize petroleum compounds with little to no fragmentation, an important challenge, since the complexity of crude oil makes discerning fragments from complete molecules impossible. Chapter 2 discusses the merits of FT-ICR mass spectrometry that make it uniquely well-suited for petroleum characterization. “Petroleomics” couples the highest resolution mass

21 spectrometer with arguably the most complex natural mixture and correlates the exact chemical composition of petrocompounds to the properties of crude oil. 41, 42

Ionization Techniques

Electrospray Ionization

Electrospray ionization earned John Fenn the Nobel prize in 2001. Its ability to transform analyte species in solution to free ions in the gas phase continuously and to do so on large, complex, fragile compounds not ionized by other ionization techniques quickly expanded its applicability. Electrospray quickly became the primary ionization technique operated at atmospheric pressure and the technique of choice for coupling a liquid chromatographic technique to a mass spectrometer. 40 Figure 2.1 shows a schematic of an electrospray source. Solution- phase anions or cations, depending on the polarity of the dispersing field, create tiny, gas-phase, charged droplets by application of an intense electric field. 40, 43 Dilute sample solution is pushed through a syringe pump through a needle where 2-4 kV electric potential is applied. Excess charge on the surface of the droplet creates a charged ion. As the drops slowly evaporate, they reach their Rayleigh limit and eject short bursts of charge through Taylor cone structures. 40 A small portion of the sprayed material enters the mass spectrometer at atmospheric pressure through a capillary that is coupled to the first pumping stage of the instrument (mTorr).

Fenn and Zhan first applied electrospray ionization to fossil fuels. 44 One of the main limitations of electrospray is its inefficient ionization of nonpolar species, highly abundant in crude oil. However, polar compounds comprise only ~15% of crude oil, but are the most problematic for both upstream and downstream processes and therefore their molecular characterization is crucial.

Atmospheric Pressure PhotoIonization

23 One of the main limitations of electrospray is its limited ability to ionize nonpolar species. Other ionization techniques, such as field desorption (FD) and field ionization (FI) have been used in the past for analyzing nonpolar species. 2, 45, 46 However, this technique is tedious and time-consuming, since heavy crude oil tends to deposit on fragile FD emitters. Since ionization occurs under vacuum conditions, changing emitters requires a break in vacuum and subsequent pump down of the system making a single analysis time consuming and tedious. For this reason, Atmospheric Pressure PhotoIonization (APPI) has become the technique of choice for the characterization of the nonpolar fraction of crude oil. The principal advantage of APPI over ESI, as mentioned above, is its ability to efficiently ionize compounds of low-polarity; however, APPI also ionizes polar compounds simultaneously making it an excellent ionization technique for coupling liquid chromatography to a mass spectrometer just as electrospray. Figure 2.2 shows a schematic of the APPI source used at NHMFL. A custom-built adapter was used to interface the APPI source to the first stage of pumping in the mass spectrometer. The sample solution is dissolved in toluene to a concentration between 10-75 g/mL and is supplied to a fused silica capillary by a syringe pump at a rate of 25-75

L/min. The sample mixes with a nebulization gas, typically N2 or CO2, at approximately 50 kPa inside a heated chamber. The nebulization temperature is controlled by an external heating supply which can be operated between 200-500 ˚C, depending on the sample (See Chapter 3). Once nebulized, the sample exits the chamber as a confined jet and passes orthogonal to a vacuum gas discharge lamp, often krypton, where photoionization occurs at atmospheric pressure. The ions are then swept into the mass spectrometer through a resistively heated capillary into the mass spectrometer.

The mechanisms of ion formation in APPI are shown in Figure 2.3. The fundamental principle in positive mode APPI is the absorption of a photon by a molecule causing the ejection of an electron and the formation of a molecular radical cation (1). 47, 48 Direct photoionization occurs if the photon energy is greater than the ionization potential (IP) of the molecule. The probability of this occurring is very low, since photons collide with gases and other molecules in the source before they reach the analyte. Ionization of the dopant (2) followed by subsequent charge exchanges with the analyte (3) increases ionization efficiency of the analyte. If the proton affinity of the deprotonated dopant molecule is less

25 than the proton affinity of the analyte, solvent molecules can act as an intermediate between dopant ions and the analyte through proton transfer (4) and charge exchange (5) reactions. However, toluene acts as the dopant and solvent for petroleum analysis and serves and increases ionization efficiency alone.

Dopant-Assisted APPI. One of the main limitations of APPI is that ionization occurs at atmospheric pressure, where collisions between photons and atmospheric gases can occur and limit analyte ionization efficiency. Robb and Bruins developed a technique called dopant-assisted APPI to help increase analyte ionization through the addition of an easily 26 ionizable substance at high relative ratio to the analyte. Benzene and toluene are two commonly used dopants. Photons first react with the dopant molecule which then undergoes charge exchange or proton transfer reactions with the analyte. However, if toluene is used as the solvent, it is already at high concentration relative to the analyte molecule and proton transfer or charge exchange reactions should prevail minimizing neutralization reactions of the analyte.

The Flagship 9.4 Tesla FT-ICR Mass Spectrometer

Figure 2.4 shows a schematic of the custom-built FT-ICR mass spectrometer equipped with a passively-shielded 22 cm room temperature bore 9.4 Tesla superconducting magnet (Oxford Corp., Oxford, U.K.) controlled by a modular ICR data station. 49-52 Ions generated at atmospheric pressure in the external ionization region (ESI or APPI) enter the skimmer region operated at ~2 Torr through a heated metal capillary into the first rf-only octopole. Ions then pass through a quadrupole to a second octopole where they are accumulated 250-5000 ms. Collisional cooling with helium gas occurs priors to transfer through an rf-only octopole to an open cylindrical Penning ion trap (10 cm i.d. x 30 cm long). Octopole ion guides are operated between 1.5-2.0 MHz and

170-240 Vp-p rf amplitude. Broadband frequency chirp excitation accelerates the ions to a cyclotron orbital radius detected by the differential current induced between two opposed electrodes within the ICR cell. Multiple (100-500) time-domain acquisitions are summed for each sample, Hanning-apodized, and zero-filled once before fast Fourier transform and magnitude calculation.

Figure 2.4. Schematic of the 9.4 Tesla FT-ICR mass spectrometer located at the National High Magnetic Field Laboratory at Florida State University in Tallahassee, Florida. Differential pumping is used to reduce the base pressure in the ICR cell to 10-10 Torr to minimize collisions between ions during excitation/detection. Figure provided by the Marshall Research group courtesy of John Paul Quinn.

A Brief Overview of the Theory of FT-ICR Mass Spectrometry

In 1973, Alan Marshall and Melvin Comisarow combined Fourier transforms, ion cyclotron resonance and mass spectrometry to create FT- ICR mass spectrometry. A fixed magnetic field and an rf pulse applied excited trapped ions to cyclotron motion through electrodes parallel to the magnetic field. 53 Coherent ion packets were excited close enough to another pair of detection electrodes to induce an “image” current that

28 was measured as a time-varying differential voltage. Sinusoidal signals were subjected to Fourier transformation after conversion from analog to digital. The first FT-ICR mass spectrum was collected on methane ions in 1973 at the University of British Columbia.53 Ion cyclotron motion occurs from the interaction between an ion and a spatially homogenous magnetic field. As an ion enters a magnetic field, it encounters a force which bends the ion’s path into a circle. This is the Lorentz force (FL), and the applied force on the ion is always perpendicular to the ion motion and is expressed mathematically by Eq

(3.1), in which q is ion charge, v is ion velocity and Bo is magnetic field strength.

FL = mass x acceleration = q v x Bo (3.1)

The cross product indicates that the force is perpendicular to the velocity and the magnetic field. The angular acceleration of uniform circular motion is shown in Eq. (3.2) where v and r are velocity and radius.

a = v 2 / r (3.2)

Substituting Eq. (3.2) into Eq. (3.1)

m v 2/ r = q v Bo (3.3)

Angular velocity (ω) is equal to

ω r = v (3.4)

29 Substitution of Eq. (3.4) into Eq. (3.3) and simplification produces the conventional form of the cyclotron equation Eq. (3.5) where ω is the cyclotron frequency.

ω = q Bo / m (3.5)

A more useful form of the cyclotron equation is given in Eq. (3.6)

where vc is the cyclotron frequency in Hertz, Bo is the magnetic field strength in Tesla, m is the ion mass in Da and z is multiples of elementary charge.

× 7 ωc 1.535611 10 B0 c = = (3.6) 2π m z Ion cyclotron motion is independent of ion velocity and is what makes ion cyclotron resonance a valuable attribute for mass spectrometry. 54

FT-ICR Mass Spectrometry for Petroleum Analysis : A Primer

To separate and identify the tens of thousands of compounds in a single crude oil, a powerful technique must be used. Mass spectrometry techniques have evolved along with the oil industry over the past 50 years; however, only in the past 15 or so years has a technique been applied to and successfully characterized heavy crude oil. FT-ICR MS was first applied to the analysis of hydrocarbons in 1994. 55 Since then, FT-ICR mass spectrometry has exploded as a tool for use by the oil industry, with more than 10 commercial instruments being used in- house by oil companies worldwide. Many studies have provided detailed characterization of the polar fraction of many fractions of crude oil and bitumen. 56-67 “Petroleomics” aims to establish the connection between the behavior of a crude oil and its chemical composition. For example, to understand why two crude oils from the same reservoir can behave very

30 differently during production, their chemical composition needs to be interrogated thoroughly. Ultrahigh resolution FT-ICR mass spectrometry is unmatched in its ability to analyze complex mixtures quickly and concisely. Other techniques require tedious and time-consuming wet chemical separations, such as extraction, precipitation, distillation, etc. Sample preparation is minimal, with crude oil simply being diluted and analyzed with no prior sample treatment.

Kendrick Mass Sorting The m/z spacing between 12C and 13C12C versions of identical species in crude oil differ in mass by 1.0033 Da, which identifies only singly charged species in crude oil. Figure 2.5 shows two mass-scale expanded insets for an Athabasca bitumen HVGO distillation cut, identical to one previously presented. 42. A 100 Da window shows the 14.01565 Da spacing representative of members of a homologous series, differing in CH2 units with the same heteroatom content and DBE (bottom). A 30 Da window shows the spacing of 2.0157 Da which is compounds differing in elemental composition by two hydrogens, equivalent to a double bond or a ring and differ in DBE value only. Even at sub-ppm mass accuracy, assignment of elemental formulas above 400 Da becomes challenging as the number of structural rearrangements exponentially increases. Kendrick mass sorting can be used to assign formulas to ions of higher m/z by extending the mass range of a homologous series from low m/z to span the entire molecular weight distribution. 61, 68 The Kendrick mass scale is normalized to the mass of a CH2 unit equal to 14.00000 Da

(versus IUPAC where CH2 = 14.01565 Da) Eq (3.8).

Kendrick Mass = IUPAC mass X (14.0000/14.01565) Eq. (3.8)

Complex natural mixtures, such as organic matter and crude oil, benefit by using the Kendrick scale because compounds with the same 31 heteroatom content and same aromaticity differ only in the degree of alkylation make up homologous series and can be sorted by their Kendrick mass defect ( Eq. (3.9).

Kendrick Mass Defect = (exact Kendrick mass – nominal Kendrick Mass) Eq. (3.9)

Kendrick normalization or Kendrick mass sorting then identifies homologous series that span the entire molecular weight distribution of a

32 sample. Accurate mass alone can assign elemental formulas up to 400 Da and extension of the series allows for identification of all the other members of that series. Kendrick mass sorting extends elemental formula assignment to formulas up to nearly 1400 Da.

Mass Resolution

Ultrahigh resolution (m/∆m50% > 350,000, where ∆m50% is the magnitude mode mass spectral peak width and half-maximum peak height) is essential for separation of isobaric species highly abundant in crude oil. A minimum resolving power must be achieved in order to separate signals from ions of very similar masses, i.e. compounds having the same nominal mass but differing in Kendrick mass. 68 For example, the 3.4 mDa split between isobars which differ in elemental composition

by SH4 vs. C3, both having a nominal mass of 36 Da. To accurately assign compositions in crude oil, these species must be separated from one another, and separation requires a minimum resolving power. Crudes that are high in sulfur, like heavy crudes and residua, cannot be correctly assigned if the 3.4 mDa split is not resolved. In APPI, the

13 overlap between SH3 C and C4, occurs between a protonated and radical cation, both with 48 Da nominal mass. Correct elemental assignment requires sufficient resolving power to separate and identify these isobaric species. Figure 2.5 shows the theoretical resolving power in FT-ICR MS and the minimum resolving power required to separate the 3.4 mDa split and the 1.1 mDa split. Separation of the 1.1 mDa and 3.4 mDa isobaric overlap is the proverbial “line in the sand” required to correctly assign elemental formulas to mass spectral peaks.

Figure 2.7 illustrates the problem associated with assignment of elemental formulas below the resolving power threshold. Here, a DBE vs carbon number image is shown for only the protonated hydrocarbon

13 class, which overlaps with C1SH3 radical cation. A compound at m/z 497 is resolved from isobaric species; however, the next member of the homologous series at 14.01565 Da higher (mass of a CH2 unit) is not resolved, thus affecting the mass accuracy and subsequent elemental

13 + composition assignment for the more abundant [ C1SH3] .

Figure 2.7. Color-coded isoabundance contoured plots of DBE vs. carbon number for Middle Easter heavy crude protonated hydrocarbon species. The image exhibits a missing portion of the DBE and carbon number distribution for the sample which is due to the decrease in resolving power above a certain m/z value. At m/z 497, the 1.1 mDa mass doublet is resolved between a protonated hydrocarbon and isobaric 13C132SH3. However, the next member of the homologous series is not resolved from its [SH313C]+ counterpart and therefore elemental composition occurs erroneously for both hydrocarbon and sulfur species.

Figure 2.8 shows broadband APPI FT-ICR MS at 9.4 T for a processed vacuum residue. 26,359 mass spectral peaks from 350 < m/z < 1000 were observed at 6 times the signal-to-noise ratio baseline rms noise, at an average m/∆m50% = 900,000 at m/z = 687. To the best of our knowledge, the mass spectrum represents the highest resolving power at 9.4 Tesla for a petroleum broadband mass spectrum by FT-ICR MS.

Figure 2.8 Broadband positive-ion APPI FT-ICR MS at 9.4 Tesla. 26,359 mass spectral peaks above 6 times the signal-to-noise ratio baseline rms noise were observed from 400 < m/z < 1100 with m/∆m50% = 900,000 at m/z 687, currently the world record for resolving power at 9.4 Tesla of a petroleum sample.

Spectral Complexity. Spectral complexity can hinder correct identification of elemental compositions if sufficient resolution is not achieved. Routinely, FT-ICR MS of petroleum results in more than 30,000 spectral signals in a single mass spectrum. As boiling point increases, so too does complexity and therefore, heavy crudes, residua and bitumen produce extremely complex mass spectra. Between 50-80 peaks per single nominal mass unit is common for APPI. Figure 2.9 shows a broadband positive-ion APPI FT-ICR MS collected at 9.4 Tesla for Middle Eastern heavy crude, containing more than 31,000 peaks (each

36 with magnitude higher than at least 6σ of baseline noise) between 300 and 1250 Da, at a mass resolving power m/Δm50% (in which Δm50% denotes the full mass spectral peak width at half-maximum peak height) of 600,000 at m/z = 675. The 3.4 mDa isobaric overlap (each with 36 Da Kendrick exact mass but differing 3.4 mDa in nominal mass) is displayed in a mass-scale expanded inset at m/z = 605 combined with the 4.5 mDa mass doublet between [ 13C ]+ vs [ 12CH + H ]+ containing species.

Isotopic signatures. To ensure that elemental compositions are assigned correctly, isotopic signatures are used in conjunction with mass accuracy. One commonly used isotopic signature is 13C. Since petroleum is composed mainly of compounds containing carbon and hydrogen, the 13C peak can be detected and identified for nearly every compound. The exact mass difference between 12C and 13C is 1.0033 Da at an abundance of 1%; therefore, once a molecular formula is assigned it can be further validated from its 13C isotope. Heavy crudes contain a high concentration of sulfur compounds, and the 34S isotope is used to verify the elemental

38 composition assignment for sulfur containing species. Isotopomers, compounds with the same elemental composition differing by an isotope, of 32S and 34S differ in mass by 1.9958 Da at 4.2% abundance and are routinely used in FT-ICR MS. Figure 2.10 shows the isotopic signatures

13 13 34 13 for an compound containing two sulfurs, its C1, C2 and S C1 isotopomers.

Mass Accuracy Inside of the ICR cell, the act of trapping ions inside an electrostatic cell shifts their natural cyclotron frequency slightly. 69 A frequency-to-m/z calibration can be applied to correct the m/z measurement across the molecular weight distribution. The most widely used calibration equation is shown in Eq. (3.7). 70

2 m/z = A/v + B/v (3.7)

A and B are constants that are obtained by fitting at least two ICR frequencies of ions of known m/z to the equation. Internal calibration produces mass accuracies of less than 1 ppm because calibrant and analyte ions experience the same electric field inside the ICR cell during detection. Internal calibration in petroleum samples is based on calibration on a homologous, highly abundant alkylation series of ions

differing in mass by 14.01565 Da, the mass of a CH2 unit, across the entire molecular weight distribution of the sample.69 Internal calibration yields mass accuracies between 100-400 ppb for petroleum and allows for unambiguous elemental formula assignment. Figure 2.11 plots mass error (in ppm) vs m/z for a European crude oil analyzed by APPI 9.4 T FT-ICR MS. Elemental composition assignment for 10,656 peaks mass spectral peaks observed at 10 times the signal-to-noise ratio baseline rms noise resulted in rms mass error of 260 ppb from 350 < m/z < 1025.

Dynamic Range Petroleum crude oil is arguably the world’s most compositionally complex organic mixture with the number of chemically distinct constituents within a dynamic abundance range between 10,000 – 100,000. 71 Dynamic range is generally thought of as the ability of an analyzer, such as a mass spectrometer, to measure harmonically related signals. In mass spectrometry, it is the ratio between the largest and smallest signals simultaneously present in a mass spectrum and allows measurement of the smaller signal to a given degree of uncertainty. FT-

40 ICR mass spectrometry has a high dynamic range therefore making it uniquely sorted for complex mixture analysis, since less abundant ions are able to be resolved along with highly abundant ions in the same spectrum. Other techniques with a lower dynamic range have difficulty identifying the less abundant species in a sample. The complexity of crude oil, because there are thousands of different species present in a sample, requires a high dynamic range. Often, the most problematic chemical constituents in a crude are in low relative concentration to the overall composition. For example, naphthenic acids and porphyrins are very problematic but are not often present in crude about a few ppm by weight.) Figure 2.12 visually represents the advantage of dynamic range across a 421 mDa window of a mass spectrum collected at 9.4 Tesla. The largest signal occurs at m/z 616.50364 with a signal-to-noise ratio of 170 which results in 50 ppb mass error in elemental composition assignment. However, at lower signal, a mass scale expanded inset across a 20 mDa window shows four peaks above six times the baseline rms signal-to-noise level at much lower signal-to-noise ratio. A 1.1 mDa split is separated and identified at two times the signal-to-noise with a slightly higher mass error of -100 ppb.

Conclusions

Here, we have presented a review of the analytical merits which make FT-ICR MS s powerful technique for crude oil compositional characterization. The complexity of high boiling, heavy crudes is well- suited for the high resolving power afforded by FT-ICR MS. High mass accuracy alone can assign elemental compositions below ~400 Da and Kendrick mass sorting exploits patterns in crude oil and extend the upper mass limit based on homologous series. However, there is a 42 threshold for resolving power that must be attained in order to separate compounds which differ by less than 40 mDa in molecular weight. Isobaric overlaps between sulfur-containing compounds exist in both ESI and APPI and resolution must exceed the minimum resolving power at a given m/z to correctly identify compounds. Currently, the Marshall research group holds the world record for the highest number of peaks assigned in a single spectrum and the highest achieved resolution for a complex mixture. Mass accuracy across a ~1000 Da window routinely produces rms mass error values less than 400 ppb for crude oil samples. Assignment of elemental formulas to monoisotopic peaks is verified by identification of isotopic signatures, namely isotopomers 13C and 34S. High dynamic range inherent to FT-ICR MS facilitates simultaneous detection of high and low mass spectral signals in a single broadband mass spectrum.

43 CHAPTER 3. OPTIMIZATION OF ATMOSPHERIC PRESSURE PHOTOIONIZATION NEBULIZATION TEMPERATURE FOR ATHABASCA BITUMEN DISTILLATION CUT POINT DETECTED BY FT- ICR MASS SPECTROMETRY

Summary

The ultrahigh mass resolving power (450,000 – 650,000 at m/z 500) and high mass accuracy (<300 ppb) of Fourier transform ion cyclotron resonance mass spectrometry (FT-ICR MS) allows for the assignment of molecular formulas to more than 20,000 mass spectral peaks in a single mass spectrum. Consequently, it is especially useful for complex petroleum mixture analysis. APPI is uniquely suited for petroleum analysis due to its ability to ionize both nonpolar species (e.g., polycyclic aromatic, thiophenic and furanic) as well as polar compounds (e.g., pyridinic and pyrrolic nitrogen) in a single analysis. A heavy vacuum gas oil (HVGO) and its distillation series of Athabasca bitumen are analyzed at varying nebulization temperatures by atmospheric pressure photoionization (APPI) FT-ICR MS. A series of nebulization temperatures (from 250 - 500ºC) was selected to determine the optimal sheath gas temperature for each boiling point range. FT-ICR mass spectra for all eight samples plus the feedstock were compared and analyzed at 7 different nebulization temperatures to observe changes in heteroatom class, double bond equivalents (DBE = number of rings plus double bonds) and carbon number distribution as a function of source/ionization temperature.

Introduction

The exhaustion of easily accessible light, sweet crude oil reserves has increased the need for detailed characterization of heavy oil reservoirs, such as the Athabasca oil sands of Alberta, Canada. 7 Heavy 44 feedstocks such as bitumen create technical challenges during recovery, transport, storage and upgrading due to its high gravity (7-15 °API) and high viscosity (> 100,000 cP) and high TAN values between 3-4 mg KOH/g oil (TAN = total acid number). 7, 12, 72 Compositional knowledge of refinery feedstocks is essential for predicting how oil will behave in reservoirs, pipelines and during upgrading. 73 Bulk property and chromatographic measurements are helpful for behavior prediction, but cannot provide detailed compositional analysis. Petroleomics is the understanding of the structure-function relationship between components in crude oil and with sufficient detailed molecular characterization, property and behavior prediction is possible.27, 42, 71, 74, 75

The ultrahigh resolution (mass resolving power = m/Δm50% > 400000, in which Δm50% is magnitude-mode FT-ICR mass spectral peak full width at half-maximum peak height) of FT-ICR MS and high mass accuracy (<100-300 ppb) allows for characterization of complex crude oil mixtures at the level of elemental formula assignment. 55, 76, 77 Due to the complexity of crude oil, distillation is used to separate crude oil into boiling point fractions prior in the refinery. A feed can be characterized based on the yield of low-boiling, high value distillates are present in each barrel without costly upgrading processes (cracking, conversion, etc). Conversely, feeds that contain a high amount of nondistillable fractions (e.g. asphaltenes) require more refining to produce valuable light hydrocarbons. Bitumen contains approximately 15-17% asphaltenes by weight and therefore requires considerable upgrading to be economically feasible to produce. Distillation reduces the number of different chemical groups and molecular weight ranges into fractions based on volatility. A variety of analytical techniques has been applied to the characterization of distillate fractions such as thermogravimetry 78, HPLC 22, 79, NMR, 80, 81 XPS, 82 and thin-layer chromatography with flame ionization detection.83 However, these techniques can only provide general compositional information.

45 Ionization methods successfully coupled to ultra-high resolution FT-ICR MS include electrospray,64, 66, 84 laser-induced acoustic desorption85, field ionization,86 field desorption45, and recently, APPI. 87-91 Electrospray ionization (ESI) has been used to characterize polar components in crude oil, 24, 49, 60, 64, 92-95 but inefficiently ionizes highly abundant nonpolar hydrocarbons, thiophenes and furans. Field desorption (FD) ionization forms ions from nonpolar species, and several studies show its application to petroleum analysis. 86, 96, 97 Schaub et al. introduced continous-flow FD ionization which increases the signal-to- noise ratio and dynamic range for characterization of petroleum fractions by FT-ICR mass spectrometry. 24, 46, 96 Smith et al.98 used both positive- and negative-ion electrospray (ESI) and automated liquid injection field desorption ionization (FDI) coupled to FT-ICR MS to characterize the acidic, basic, and non-polar species in a bitumen HVGO distillation series. However, FD is a pulsed-ionization source, which increases data acquisition time and replacement of fragile FD emitters requires a tedious break in source chamber vacuum. FD analysis routinely produced narrower carbon number and DBE distributions than ESI and APPI analysis of the same samples. 67 APPI can mitigate the inconvenience of FD analysis at reduced pressure and the need for mutiple ESI experiments since ionization occurs at ambient pressure and ionization is initiated by 120 nm photons which ionize polar and nonpolar species in a single experiment. There are two main pathways for ionization by APPI. First, direct ionization occurs if the energy of the photon exceeds the first ionization potential of the analyte. However, at atmospheric pressure, the mean-free path of the photon is less than ~1 picometer, resulting in inefficient analyte ionization. Robb et al.99 introduced dopant-assisted APPI to increase ionization efficiency in reversed-phase LC-MS. A dopant, often toluene, is first photoionized and the high collision rate in the source ensures that dopant photoions react to completion with analyte molecules through charge exchange and proton transfer reactions.99-101 46 Purcell et al. first coupled dopant-assisted APPI to FT-ICR MS analysis of crude oil and identified >12,000 unique elemental compositions across a 400 Da mass window. 91 In APPI, heated sheath gas facilitates sample nebulization, for example, crude oil dissolved in toluene. However, because of the heated nebulizer region, the source has an inherent thermal desorption limit. Too high of a sheath gas temperature for a given crude oil (light or heavy) can cause thermal reactions to occur between analyte molecules, such as dealkylation, condensation and result in an inaccurate description of sample composition. Too low of a sheath gas temperature for a given boiling point results in incomplete desorption of high boiling compounds and result in aberrantly low molecular weight distributions and heteroatom composition (since these increase in higher boiling crudes). The main objective of this work is to determine the minimal sheath gas temperature that results in successful desorption (and ionization) of crude oil compounds within a distinct boiling point range. Athabasca bitumen HVGO was fractionated into eight boiling point ranges and analyzed by APPI FT-ICR MS at incremental sheath gas temperatures. Heteroatom class, type (double bond equivalents or DBE = the number of rings plus double bonds) and carbon number distributions are used to examine trends within each boiling point range as a function of source temperature and distillation cut temperature to determine optimal nebulization temperature.

Experimental Methods

Sample Preparation. A bitumen heavy vacuum gas oil (HVGO) was fractionated by ASTM D-1160 into eight distillate fractions in 25 °C sections from the initial boiling point (IBP) to the final fraction (500 – 525+ °C) which is the residue left in the distillation pot after collection of the 475 – 500 °C fraction. Each distillation cut was prepared in toluene

47 at 500/mL and analyzed by APPI FT-ICR MS without additional modification.

Instrumentation: APPI FT-ICR MS. Samples were ionized with an IonMaxx™ API source (ThermoFisher) in photoionization configuration coupled to a hybrid FTMS. Ions are generated at atmospheric pressure as the gas stream containing the vaporized analyte flows toward the heated metal capillary inlet and orthogonal to the krypton vacuum UV lamp. The solvent flow rate was 50 L/min and the nebulizer heater was operated at a series of temperatures (200-500 °C) to determine the optimal temperature for each boiling point range (i.e. distillation cut). Nitrogen served as a sheath and auxiliary gas and was regulated by Xcaliber™ software (set to 50 and 5 respectively, arbitrary units). Low resolution and high resolution mass spectra were collected with a modified hybrid LTQ-FT-ICR (ThermoFisher Corp., Bremen, Germany) mass spectrometer equipped with a 14.5 Tesla superconducting magnet (Magnex Scientific, Oxford, UK).102 One modification consists of an additional wired storage octopole 103 directly behind the LTQ that allows for a second method of ion accumulation. Ions can first be mass- selected in the linear trap and then passed to the storage octopole for numerous cycles prior to transfer to the ICR cell.102, 104

Results and Discussion

Due to the complexity of conventional and unconventional crude oil, ultrahigh resolution is required to resolve and assign molecular formulas to every peak in the mass spectrum. Each distillation cut consists of a limited selection of chemical species and it is easier to distinguish between compound classes in a narrow distillation cut than in a whole crude. Distillation limits the molecular weight range of each compound type within a given fraction but the molecular weight range

48 betweeen compound types in the mixture can be quite different, e.g., the n-paraffins and the n-alkylnaphthalenes. 14 A given boiling point range contains a wide range of compound types and molecular weights. Bodusysnki and Altgelt extensively characterized heavy crude oil composition as a function of atmospheric equivalent boiling point (AEBP). 13, 14, 22, 105, 106 As the molecular weight, aromaticity and polarity of a compound increases, vapor pressure decreases and results in a higher boiling point. 14 Distillation is widely used in refineries to reduce the complexity of crude oil into different volatile fractions. Distillation separates based on vapor pressure differences between molecules and is arguably the most important separation method used in petroleum refining since it reduces the number of species present based on volatility and reduces complexity. Distillation reduces sample complexity, and is useful for ultrahigh resolution compositional analysis such as APPI FT- ICR MS at 14.5 Tesla which routinely results in over 20,000 unique elemental assignments from a single whole crude oil. 107

The hybrid FTMS system is equipped with a linear trap quadrupole (LTQ) mass spectrometer for low resolution sample interagation and mass selection. Each HVGO fraction was analyzed with the LTQ to provide an independent verification of the molecular weight distribution,

50 shown in Figure 3.1. The IBP-343 °C cut has a narrow distribution of 150 < m/z < 375, whereas the higher boiling 500+ °C cut has a molecular weight range of 300 < m/z < 750, a factor of 2 greater. An increase in complexity is observed with increasing boiling point as evident by the broader molecular weight range as the distillation cut temperature increases. Furthermore, as the number of carbon atoms per molecule increases (higher molecular weight), the number of possible combinations of C,H,N,O and S increases and results in an increase in complexity. Hydrogen-bonding polar heteroatoms are found in greater abundance in the higher boiling fractions since stronger intermolecular forces exist within these structures 2, 22. Optimal APPI source temperature is critical to determine whether or not the species desorbed into the gas phase are representative of the species in solution.

LTQ-MS Analysis

As a first approach, the optimal sheath gas temperature was explored with low resolution LTQ spectra. Increasing the temperature, it was thought, would shift the molucular weight distribution to higher mass. However, within a given boiling point range, increased sheath gas temperature did not significantly alter the molecular weight distribution. Figure 3.2 shows LTQ mass spectra for two distillates, the lowest-boiling (IBP-343 °C) and the residue (500+ °C). At incrementally higher sheath gas temperatures, the mass center increases slightly for the residue and not at all for the IBP-343 °C. Futhermore, sheath gas temperature does not affect the molecular weight distribution within a given boiling point. A molecular weight distribution of 150< m/z < 400 for the light distillate and 250 < m/z < 750 for the residue was observed at sheath gas temperatures from 250-450 °C.

Therefore, ultrahigh resolution FT-ICR MS was employed to assign elemental compositions and determine if the higher boiling compounds are being desorbed without degradation of the lower boiling compounds within a distillation cut.

52 Figure 3.3.a (Top) Linear trap mass spectra for the distillate residue (500-525 °C) collected at optimal sheath gas temperature (325 °C). Both low and high resolution mass spectra were collected for each boiling point range, since low resolution LTQ-MS analysis can not detect compostional changes as a function of nebulization temperature. Because there is inherent discrimination in the number of ions that can be trapped in the ICR cell prior to detection, the molecular weight distribution is truncated and represents a heart-cut of the most abundant species present, centered with the LTQ spectrum. Figure 3.3.b (Bottom) Ultrahigh resolution FT-ICR mass spectra for the distillate residue (500- 525 °C) collected at optimal sheath gas temperature (325 °C). Over 20,000 peaks were detected above six times the baseline rms noise between 350 < m/z < 800 with approximately 77 unique mass spectral peaks per nominal mass. An average resolving power of m/∆m50% = 400,000 was achieved at m/z 600.

Figure 3.3a shows LTQ and FT-ICR spectra for the 500+ °C distillate at a sheath gas temperature of 325 °C. The LTQ mass spectrum shows a mass distribution from 300 < m/z < 750 with mass centroid at ~490 m/z. The lower FT-ICR mass spectrum shows a similar mass centroid but is slightly truncated. The truncation is primarily the result of time-of-flight dispersion as the ions travel from the external ion trap to the ICR cell. Figure 3.3b shows the need for ultrahigh resolution for the analysis of crude oil. Two common mass doublets observed in APPI are shown in the zoom inset. The 3.4 mDa split (the mass difference between compounds differing only by C3 vs. SH4) and 1.1 mDa split

13 (SH3 C vs. C4) are common in heavy crude oil. The 3.4 mDa split is also observed in electrospray ionization, however, since positive ion APPI can form two ion types – protonated compounds and radical cations – ultrahigh resolution mass spectrometry is necessary to resolve the 1.1

13 mDa split (SH3 C1 vs C4 – one protonated compound and one radical cation).

Determination of Optimal Sheath Gas Temperature. While the mass distribution remains unchanged at higher nebulization temperatures, the ability to ionize higher boiling compounds does vary with sheath gas temperature. Compounds within the same molecular weight range can differ in structure and heteroatom content (greater aromaticity and more

53 heteroatoms per compound) that would shift their boiling point higher. Therefore, even though the molecular weight distribution does not vary with nebulizer temperature (Figure 3.2), carbon number, type and heteroatom class desorbed may change with ionization temperature. To determine optimal sheath gas temperature for each distillation cut, compositional trends in and between distillation cuts were determined with ultrahigh resolution.

Heteroatom class analysis (i.e., molecules with the same number of N, O, and S atoms and differ only by carbon and hydrogen) combined with color-coded isoabundance contour plots of double bond equivalents56, 108 versus carbon number display a large amount of compositional data in a compact form. 61 Compositional analysis at different sheath gas temperature shows variations within each heteroatom class. Figure 3.4 shows the three dimensional isoabundance color contour plots for the two most abundant classes, hydrocarbon and

+ S1, for the 475-500 °C and 500 °C distillation cuts. Carbon number is plotted on the x-axis and double bond equivalents on the y-axis with relative abundance color weighted in the z-axis. Since different compounds are ionized more efficiently at varying sheath gas temperatures, the optimal sheath gas temperature will ionize higher- boiling compounds without causing thermal degradation of light, low- boiling compounds. Necessarily, the higher boiling fractions contain higher boiling compounds and require increased sheath gas temperatures for efficient sample vaporization.

55 For each distillation cut, the optimal sheath gas temperature was determined by analysis of the four most abundant heteroatom classes at incrementally higher sheath gas temperatures. For example, the high boiling 500+ °C fraction showed that 275 °C resulted in incomplete desorption/ionization of higher molecular weight species as evident by “holes” in the carbon number versus DBE plot. As the sheath gas temperature is increased to 300 °C, higher carbon number species are observed. A sheath gas temperature of 325 °C produces an image with a DBE range of 5 - 16 and 30-50 carbon atoms per molecule for the hydrocarbon class. Because too high of a sheath gas temperature can result in possible thermal cracking of low molecular weight, hydrogen- rich compounds, the lowest sheath gas temperature that results in the widest carbon number and DBE range is deemed optimal. For the 500- 525+ °C sample, a sheath gas temperature of 325 °C desorbs the high molecular weight (carbon number) compounds and higher sheath gas temperature does not ionize material not observed at lower temperatures.

The S1 heteroatom class exhibits the same trend but with a lower boiling point distillation cut, and therefore, a lower optimal sheath gas temperature, 300 °C corresponds to the highest abundance of the hydrocarbon and S1 compounds. Above the optimal temperature, the carbon number distribution narrows and the appearance of low carbon number, high DBE species can be observed. At sheath gas temperatures above optimal, thermal degradation of low carbon number compounds can occur and one possible explanation could be dehydrogenation reactions of lighter compounds that result in higher DBE, low carbon number species. This method was used for each distillate sample to determine the sheath gas temperature that results in efficient desorption without degradation for the five most abundant heteroatom classes with

Table 3.1. Optimal sheath gas temperatures for each boiling point range for an Athabasca bitumen HVGO distillation series.

Distillation Optimal Cut Temperature (°C) Nebulization Temperature (°C)

IBP – 343 200 343 – 375 250 375 – 400 250 400 – 425 250 425 – 450 250 450 – 475 250 475 – 500 300 500 – 525 325

results listed in Table 3.1. What is important to note is that for each distillate fraction, the optimal sheath gas temperature is significantly lower than the boiling point range. The highest nebulization temperature, 325 °C, was only required for compounds with a boiling point above 500 °C. Distillation cuts from 343 – 475 °C were vaporized and ionized at 250 °C sheath gas temperatures at a level able to be detected by FT-ICR MS. The ability to characterize the higher boiling components of crude oil for thermal comparison of nonpolar speciation is a unique aspect of APPI.

58 Class distributions for four distillation cuts are shown in Figure

3.6. The S1 and hydrocarbon classes are the most abundant classes across all temperatures, however, for the IBP-343 °C cut, their relative abundance is greater.

Figure 3.6 APPI heteroatom class distribution for all classes above 1% relative abundance for all distillate cuts analyzed at their optimal nebulization temperatures. An increase in relative abundance of multiheteroatomic (i.e., S1 and S2) compounds is observed in higher boiling fractions along with a decrease in no or monoheteroatomic classes (i.e., hydrocarbon). Compounds with few or no heteroatoms, such as PAH’s and PAXH’s, for example) that have low molecular weights will have a high vapor pressure and therefore are concentrated in the lower boiling fractions. IBP-343 ˚C exhibits this trend and has the highest relative abundance of hydrocarbons and S1 classes across the entire series.

59 Boiling point is an indicator of intermolecular force strength between molecules and therefore dependent upon structure and molecular composition. Compounds with low molecular weight and low heteroatom content are dominated by weak van Der Waals dispersion forces which increase with surface area (carbon number). Within a given boiling point range (such as that of an HVGO), compounds with lower carbon number and fewer heteroatoms per molecule will have the lowest boiling point. This trend correlates with the high abundance of hydrocarbon and S1 classes in the lowest boiling cut. As the distillation cut temperature increases (Figure 3.6), the abundance of multiple- heteroatoms species mono-heteroatom species shift toward a more equitable distribution. As the number of heteroatoms per molecule increases, the strength of the intermolecular forces within a molecule increase and results in a higher boiling point as evident in the S1 and S2 classes which gradually increase in abundance for the higher distillation cuts. Isoabundance color contour plots of carbon number versus DBE are shown for four distillation cuts in Figure 3.7. Four heteroatom classes are shown for each boiling point range to illustrate the structural evolution within each class as a function of boiling point. The number of carbon atoms per molecule is less than 35 regardless of the functional groups present in the IBP – 343 °C cut. The upper boiling limit of the distillation cut limits the aromaticity to ~10 DBE per molecule. The higher distillation cuts (400-425 °C, 450-475 °C and 500+ °C) have a similar narrow carbon number range and different heteroatom classes within a given boiling point differ by 2-3 carbon number between classes. Higher carbon number compounds have a higher boiling point and therefore are found in the higher distillation cuts. Representative possible core structures are shown for the thiophenic and furanic species and help highlight probable core structure growth within each distillation cut. That is to say, dibenzothiophene is a known

61 to help highlight the growth of core structures within a distillation cut. At higher boiling points, the aromaticity also increases, shown here using DBE.

stable core structure at DBE 9 and since olefins are not stable in crude oil due to their reactivity, the S1 DBE = 10 species is likely formed through cycloalkane ring addition. As the boiling point increases, so too does the DBE range which indicates multiple stable core structures with alkyl substitution increasing the molecular weight range.

Conclusions

We have analyzed Athabasca bitumen HVGO distillate fractions for detailed compositional analysis by positive-ion APPI FT-ICR MS. The (sheath gas) temperature was optimized for each boiling point range (distillation cut). Optimal temperatures produced a wide molecular weight distribution without thermal degradation. Futhermore, optimization allowed for direct comparison of heteroatom class, type (DBE) and carbon number (molecular weight) trends between and within the distillation cuts. Surprisingly, the optimal sheath gas temperature is much lower than the boiling point ranges (Table 3.1). Thus, it appears the method can be used to successfully desorb and subsequently ionize higher boiling compounds than expected below the maximum allowable source temperature (500°C).

62 CHAPTER 4. HEAVY PETROLEUM COMPOSITION 1. EXHAUSTIVE COMPOSITIONAL ANALYSIS OF ATHABASCA BITUMEN HVGO DISTILLATES BY FOURIER TRANDFORM ION CYCLOTRON RESONANCE MASS SPECTROMETRY: A DEFINITIVE TEST OF THE BODUSZYNSKI MODEL

Summary

Fourier transform ion cyclotron resonance mass spectrometry (FT- ICR MS) allows detailed characterization of complex petroleum samples at the level of elemental composition assignment. Ultrahigh resolution (450,000 – 650,000 at m/z 500) enables identification of isobaric species that differ in mass by 3 milliDalton or less, and high mass accuracy (better than 300 ppb mass error) combined with Kendrick mass sorting allows for unambiguous molecular formula assignment to each of more than ten to twenty thousand peaks in each mass spectrum. Thus it is possible to identify, sort and monitor simultaneously thousands of elemental compositions as a function of boiling point. Here, the detailed FT-ICR MS characterization of an Athabasca bitumen heavy vacuum gas oil (HVGO) distillation series exposes the progression of heteroatom class, type (double bond equivalents (DBE), number of rings plus double bonds to carbon) and carbon number for tens of thousands of crude oil species as a function of boiling point. Specifically, we analyze a distillation series of Athabasca bitumen HVGO with cut temperatures from initial boiling point (IBP) to 538 °C (in eight cuts) by atmospheric pressure photoionization (APPI) as well as positive and negative electrospray ionization (ESI) FT-ICR MS to determine the distributions of nonpolar and polar species as a function of HVGO boiling point. Compositional distributions reveal definitive heteroatom class, type, and carbon number trends among distillation cuts, and provide the first detailed compositional evidence in support of the Boduszynski model that describes the progression of petroleum composition and structure as a function of boiling point. Quantitation of aromaticity and carbon number

63 profiles of both polar and nonpolar species in all distillate cuts further affirms the validity of the Boduszynski model for the HVGO distillate range, and provides evidence for cycloalkane linkages in addition to polyaromatic cores. Introduction The acceptance of heavy refinery feeds requires extension and improvement of analytical techniques routinely used for light feeds to the characterization of heavy feeds.3 However, the task is daunting, given the increased compositional complexity encountered in heavy crude oils and higher boiling distillate cuts. The increased complexity arises from the evolution of narrow molecular weight ranges (in light distillate cuts) to broader molecular weight distributions with higher heteroatom contents for heavy crudes and distillates. Generally, molecules progress from single heteroatom-containing compounds in the middle distillate fractions to multi-heteroatom containing compounds in the vacuum residue. Distillation is an important refinery process to reduce crude oil compositional complexity and produce meaningful product yields for species in each boiling range, and to predict problems encountered during processing.13, 22, 105, 106 Longstanding analytical techniques for characterizing compositional diversity in individual boiling cuts include gel permeation chromatography109 and high-performance liquid chromatography110 to discern compositional changes in distillate fractions in refinery processes. Recently, two-dimensional gas chromatography (GC x GC) has characterized and quantitated lighter distillate fractions in crude oil.111-113 Thermogravimetry114 and nuclear magnetic resonance82, 115, 116 have been applied to evaluate medium and heavy fractions of crude oil. However, high-boiling fractions and vacuum residues are often separated by supercritical fluid extraction and fractionation coupled with Fourier transform infrared, nuclear magnetic resonance and x-ray photoelectron spectroscopy for compositional characterization.117-119 Nevertheless, most analytical techniques access

64 only lower boiling fractions due to the increase in compositional complexity that accompanies and increase in boiling point, and/or measure only bulk properties in heavy crudes or heavy distillate fractions. Mass spectrometry provides detailed characterization of crude oil composition and has been coupled to ionization/separation83, 120-122 techniques such as gas chromatography-MS123-125, supercritical fluid chromatography 126, 127, field ionization 13, 22, 106, 128, 129, field desorption, 24, 67, 130 electron ionization131, 132 and electrospray ionization. 67, 130 Zhan and Fenn first applied electrospray mass spectrometry (ESI-MS) to the analysis of polar molecules in petroleum distillates, but lacked sufficient resolving power for complete compositional assignment (a problem common to all but the highest resolution mass spectrometers). 44 Since then, electrospray ionization coupled to ultrahigh resolution Fourier transform ion cyclotron resolution mass spectrometry mass spectrometry (FT-ICR MS) has extensively characterized the polar species in petroleum. 64, 67, 93, 133-137 Although polar compounds represent a small fraction (0- 15%) of the total species present in petroleum, they are implicated in production issues such as corrosion, and catalyst deactivation/fouling in upgrading processes such as hydrotreatment. 138 Thus, electrospray ionization selectively ionizes acidic and basic polar heterocompounds from the hydrocarbon matrix of petroleum samples with little or no matrix effects65 and as a result, is widely accepted as the preferred ionization technique for polar compound characterization by mass spectrometry. For nonpolar species, field ionization/desorption has historically been used to characterize HVGO and other heavy distillate fractions. 67 However, field desorption is a pulsed ionization source and ionization occurs in vacuo, necessitating tedious replacement of emitters and time- consuming data collection. Atmospheric pressure photoionization (APPI)

65 was first used as an ionization method for mass spectrometry by Robb et al.,99 with toluene as a dopant to increase analyte ionization efficiency. 101 Subsequently, APPI has analyzed a wide range of compounds including pharmaceutical and medicinal drugs,139-142 lipids, 143, 144 polyaromatic hydrocarbons,145-147 explosives148, and steroids. 149-151 APPI was first coupled to FT-ICR MS for analysis of corticosteroids. 152 Purcell et al. first coupled APPI to FT-ICR MS for characterization of nonpolar species in crude oil and resolved and identified more than 12,000 unique elemental compositions across a 400 Da mass window. 87, 89-91 APPI FT-ICR MS has recently been used to identify and characterize vanadyl porphyrins in unfractionated asphaltene and whole crude oil, and nebulization temperature has been correlated to boiling point range in heavy crude oil. 153, 154 Although limited to species that contain π-electron moieties, APPI is ideally suited for coupling to high resolution FT-ICR MS because APPI is a continuous atmospheric pressure ion source with little or no down time between sample analyses. Boduszynski et al. conducted a prescient, noteworthy comprehensive analysis of heavy crude oil composition in the late 1980s and early 1990s. 13, 22, 105, 106 Their series of papers was unique because unlike crude oil models/reaction networks based on compositional analysis of a target crude, 38, 39, 155 Boduszynski and Altgelt proposed a model based on the progression of crude oil composition and molecular weight as a function of boiling point and proposed an extrapolation to an upper limit of molecular weight for all crudes. Remarkably, if expanded to encompass current collective compositional predictions, the Boduszynski/Altgelt model (now 20 years old) predicts what is now known as the Petroleome. Combined with Quann and Jaffe, early contributions in the development of composition-dependent models for specific crudes have pioneered the current field of “Petroleomics”. 41, 42, 71, 75

66 The aim of the current work is fundamental. Oil companies sell molecules; therefore an oil’s composition determines its economic value. The ultimate goal of the detailed speciation of all the components of crude oil is simply a model to predict (for example) phase, deposition, distillation and upgrading behavior as well as bulk properties such as viscosity. However, even methods and results used to obtain bulk properties for heavy ends and asphaltenes are hotly debated. Therefore, rather than attempting to generate a predictive model directly from the exhaustive detailed composition of a specific heavy crude, we instead begin from a 20 year-old petroleum composition model that presumes to describe all crudes and evaluate whether or not the exhaustive compositional analysis of whole fractions and narrow distillation cuts (which span the high vacuum gas oil (HVGO) range centered in Boduszynski’s model (see Figure 1) supports the Boduszynski model. This simple but important issue can appropriately focus future research efforts and put an end to controversy about the molecular weight of petroleum. 28, 125, 156-165 Boduszynski related atmospheric equivalent boiling point to molecular weight/structure for heavy crude oil distillate fractions and concluded that (all) crude oil composition is continuous in molecular weight, structure, and heteroatoms (N, O, and S) in the distillables and made the inductive leap that the same trend extends to asphaltenes and nondistillable residues. 1, 13, 22, 106, 129 Based on his model, developed from boiling point trends of standards and backed by mass spectrometric results on a collection of heavy crude oils, Boduszynski concluded that "most of petroleum components do not exceed a molecular weight of about 2000." He acknowledged that the results are controversial, "These findings are significant because of the existing controversy over whether there is an appreciable concentration of molecules in petroleum having molecular weights greater than 2000 Da. Data show there is not."

67 Subsequently an increasing number of bulk measurements support his original claim 25, 26, 28, 166-172. Nevertheless, 20 years later, petroleum science remains bogged down in the same arguments about petroleum molecular weight. A definitive proof of Boduszynski's model requires direct, complete compositional characterization of complex distillate cuts unavailable at that time. If substantiated, the Boduszynski model would impose strict limits on molecular weight distributions for both distillable and nondistillable petroleum fractions that contradict many previously published assertions about petroleum molecular weight and composition. 31, 32, 173-175 Although distillation separates components based on volatility and thereby limits the observable carbon number and aromaticity, each distillate fraction nevertheless remains compositionally complex and contains a wide variety of different heteroatom functionalities. 106 Therefore, detailed compositional characterization of each fraction is paramount to understand the structural progression of crude oil compounds as a function of boiling point: from the economically beneficial, low-boiling fractions (e. g., gasoline) to the problematic, high- boiling and nondistillable fractions (e.g., resids and asphaltenes). Such detailed characterization is now possible with a single analytical technique: FT-ICR mass spectrometry. The inherent high resolution and mass accuracy of FT-ICR MS make it an effective tool for compositional analysis of complex mixtures such as heavy crude oil and distillate fractions. Here, we analyze an Athabasca bitumen HVGO distillation series (fractions critical to Boduszynski's model) to characterize nonpolar and polar species enabling characterization of molecular composition as a function of boiling point. We compare detailed compositional results obtained for polar species by ESI and nonpolar species by APPI coupled to FT-ICR mass spectrometry to access the validity of the model proposed by Boduszynski et al. in the HVGO boiling range. 13, 22, 106, 128 The current

68 results are the first of a four-part series of publications on the composition of heavy petroleum and asphaltenes.

Experimental Methods

Sample preparation. A bitumen heavy vacuum gas oil (HVGO) was fractionated by ASTM D-1160 into eight distillate fractions: (IBP-343, 343-375, 375-400, 400-425, 425-450, 450-475, 475-500, 500-538 °C). Additional distillation information can be found elsewhere. 67 Each distillate fraction (~10 mg) of each distillate was diluted with 5 mL of toluene (HPLC Grade, Sigma-Aldrich Chemical Co., St. Louis, MO) to make a stock solution (2 mg/mL) that was either further diluted to yield a final concentration of 500 g/mL for APPI or diluted with an equal part (vol:vol) methanol spiked with 2% by volume ammonium hydroxide (negative ESI) prior to FT-ICR MS analysis.

Instrumentation: APPI Source. Samples were ionized with an Ion Maxx™ API source (ThermoFisher Corp., Bremen, Germany) in photoionization configuration. The sample flows through a fused silica capillary at a rate of 50 L/min and is mixed with nebulization gas (N2 introduced at ~100 kPa) inside a heated vaporizer operated between 250- 350 °C. Nitrogen served as an auxiliary and sheath gas and was regulated by Xcaliber™ software (set to 50 and 5 arbitrary units, respectively). Source parameters were set in Xcaliber™ as follows : sweep gas rate 5 arbitrary units; capillary voltage 11 V; tube lens 50 V. Once nebulized, the sample exits the vaporizer in a confined jet and flows orthogonal to the krypton VUV lamp that produces 10 eV photons (120 nm) where photoionization occurs at atmospheric pressure. Ions are then swept into the first pumping stage of the mass spectrometer by differential pressure through a heated metal capillary. Toluene is used as a solvent/dopant to increase analyte ionization through proton-

69 transfer and charge exchange reactions.99 Electrospray ions were generated externally by a micro-electrospray source50 and were delivered by a syringe pump at a rate of 500 nL/min. 2.5 kV was applied between the capillary needle and ion entrance to the mass spectrometer.

Instrumentation: 14.5 Tesla FT-ICR MS. Low resolution and high resolution mass spectra were collected with a customized hybrid linear quadrupole ion trap/FT-ICR MS (LTQ-FT, ThermoFisher Corp., Bremen, Germany) adapted to operate in an actively-shielded 14.5 Tesla superconducting magnet (Magnex, Oxford, UK), as described in detail elsewhere.107 The octopole directly behind the LTQ was modified to include tilted wire extraction electrodes and serves as a secondary ion accumulation method and improves ion injection efficiency.103 Ions may be mass selected in the LTQ and accumulated in the wired octopole for numerous cycles before the entire octopole ion population is transferred to the ICR cell for excitation and detection.

Mass Calibration and Data Analysis. Positive-ion APPI FT-ICR mass spectra were internally calibrated70, 176 with respect to a highly

32 abundant homologous alkylation series containing one S atom and verified by identification of the corresponding 34S signal at the correct relative abundance. Singly charged ions with relative abundance greater than six standard deviations of baseline rms noise (6σ) were exported to a spreadsheet after conversion to the Kendrick mass scale68 for easier identification of homologous series. For each elemental composition,

CcHhNnOoSs, the heteroatom class (NnOoSs), type (double bond equivalents, DBE = number of rings plus double bonds to carbon)177 and carbon number, c, were tabulated for generation of heteroatom class relative abundance distributions and isoabundance-contoured DBE vs. carbon number images constructed for each heteroatom class. The molecular weight distribution for each sample was first verified by LTQ

70 analysis to ensure the validity of the molecular weight distribution based on FT-ICR MS.

Results and Discussion

Positive-ion APPI increases mass spectral complexity by the potential formation of two kinds of ions from a single neutral analyte: radical molecular cations, M+, resulting from removal of an electron and [M+H]+ species due to protonation. For accurate elemental formula assignment, two key isobaric overlaps must be resolved. Species differing

in elemental composition by C3 vs. SH4 both have a nominal mass of 36 Da (but differ by 3.4 mDa in exact mass), are generated by both ESI and APPI and define the minimum required mass resolving power.42 However, for APPI of heavy, high-sulfur crude oil, an additional 1.1 mDa doublet

12 13 arises from a protonated C4 vs. a radical molecular cation SH3 C (both with nominal mass of 48 Da) that must be resolved for correct elemental assignment. 91 Figure 4.1 shows a broadband positive-ion APPI FT-ICR mass spectrum for the 475-500 °C distillation cut for an Athabasca bitumen HVGO, containing more than 16,000 peaks (each with magnitude higher than at least 6σ of baseline noise) between 300 and 700 Da, at a mass resolving power, m/Δm50% (in which Δm50% denotes the full mass spectral peak width at half-maximum peak height) of 800,000 at m/z = 400. High mass accuracy alone can provide elemental assignments of peaks below ~400 Da. However, Kendrick mass sorting highlights alkylation and hydrogenation patterns found in crude oil and allows for unambiguous (mass error <100-200 ppb) assignment of elemental compositions for ions of much higher mass. 42, 61, 68 Heteroatom class analysis combined with color-coded isoabundance contour plots of DBE vs. carbon number creates a visual image that is especially helpful for

71 sorting the thousands of elemental compositions into chemically and structurally informative patterns.61

Figure 4.1. Broadband positive-ion APPI 14.5 T FT-ICR mass spectrum of an Athabasca bitumen HVGO distillation cut (475-500+ °C). 16,858 mass spectral peaks are resolved at 6 times the signal-to-noise ratio baseline rms noise at an average resolving power, m/Δm50% = 400,000.

The Boduszynski Hypothesis. Boduszynski et al. proposed rules to account for the dependence of boiling point on molecular weight and elemental composition for organic components of heavy crude oil. 13, 22 A fundamental principle is that “diverse compounds with similar molecular weights cover a broad boiling range; and conversely, a narrow boiling point cut can contain a wide molar mass range”. 13, 22 Figure 4.2 shows plots of molecular mass vs. atmospheric equivalent boiling point (AEBP),

72 yielding positive-sloped, wedge-shaped envelopes for each compound type. For a given boiling point, the the highest molecular weight components are paraffins, followed by naphthenes, aromatic hydrocarbons, heteroatom-containing compounds, polar heteroatom- containing compounds, and finally polar, unsubstituted aromatic heteroatom-containing compounds. 13, 22 Vertical progression from one compound class to the next reveals a difference by 2-3 carbon atoms per molecule at a given boiling point. For example, an unsubstituted polycyclic aromatic hydrocarbon (PAH) from a low-boiling HVGO cut (427 °C) contains ~17-18 carbon atoms per molecule. Monoheteroatomic compounds with the same boiling point, such as S1 or N1 classes, contain ~2-3 fewer carbon atoms per molecule than their PAH analogs. Compounds with two heteroatoms at that same boiling point contain ~5- 6 fewer carbon atoms than their PAH analogs. A polar, polyfunctional heteroatom-containing compound molecule would contain 6-7 fewer carbon atoms than its PAH analog of the same boiling point. Boduszynski proposed that the most polar compounds with the highest heteroatom content at a specific mass would have the highest boiling point; conversely, hydrocarbons (devoid of heteroatoms) would have the lowest boiling point for a given mass. 13, 22 Table 1 shows boiling point data for several core structures known to exist in crude oil.

Figure 4.2. Left: Boduszynski and Algelt model illustrating representation of the effect of molecular weight and structure on boiling point. 1 Atmospheric equivalent boiling point (AEBP) is plotted versus molar mass for compounds known to exist in crude oil. Right: At a given boling point, carbon number decreases as heteroatom content increases for compounds in the HVGO boiling range: pure hydrocarbons have ~2-3 more carbon atoms than monoheteroatomic analogues; addition of a second heteroatom reduces the carbon number by another 2-3. Polar functional groups exhibit the lowest carbon number within each boiling range.

Table 4.1 Boiling points of several core structures known to exist in crude oil. In crude oil, alkyl substitution off of core structures produces compounds with a higher molecular weight than their nonalkylated counterparts. The degree of alkylation differs with crude oil type, viscosity, and boiling point.

Compositional Differences among HVGO Distillate Cuts: Tests of the Boduszynski Hypothesis. We compared Boduszynski's predictions with experimental carbon number distributions obtained by

75 ESI and APPI FT-ICR MS for a series of distillation cuts varying by ~25 °C increments. Figure 4.3 shows isoabundance-contoured plots of DBE vs. carbon number for members of just the hydrocarbon class for all eight

distillate fractions from initial boiling point (IBP) to 538 °C. An increase in abundance-weighted average carbon number results in an increase in boiling point, from ~20 carbons for the lightest fraction (IBP-343 ˚C) to ~40 carbon atoms for the highest boiling fraction (500-538 ˚C). Each 25 ˚C increase in boiling point results in the addition of ~1-4 carbon atoms to the average carbon number. The large 6 carbon number increase from IBP-343 ˚C to 343-375 ˚C most likely due to wide boiling point range associated with the vaguely defined "initial boiling point". A similar jump

(from ~C30 to ~C38) occurs in proceeding to the final distillation cut, because the final cut has an undefined upper limit in boiling point.

Figure 4.3. Color-coded isoabundance contoured plots of DBE vs. carbon number for the hydrocarbon class for Athabasca bitumen HVGO distillate cuts. The carbon number abundance distribution maximum (red arrow) shifts from ~C20 at IBP-343 ˚C to ~C40 at 500-538 ˚C. DBE values show a gradual increase in aromaticity from DBE = ~7 to DBE = ~10 with increasing boiling point.

Progression from the first to the last well-defined boiling cuts (343- 375 ˚C to 475-500 ˚C) results in a gradual increase from 26 to 35 carbons in 1 or 2 carbon increments. Careful scrutiny of the fractions above the 343-375 ˚C cut reveals a slight widening of the carbon number distribution toward lower carbon number for the most abundant, higher DBE species. The shift tilts the otherwise vertical distributions slightly to the left. That behavior highlights a fundamental principle of the Boduszynski model: as the aromaticity (and thus DBE) increases, the number of carbons for the more aromatic species must decrease to remain in the same distillation cut. However, the tilt diminishes in the highest boiling fractions, presumably due the increased structural degeneracy in and between aromatic and cycloalkane species at higher DBE, as well as variation in ionization efficiency and dynamic range limitations of FT-ICR mass spectrometry. The number-average DBE (aromaticity) also increases, from compounds with DBE = 7 (IBP-343 ˚C fraction) to DBE = 10 (500-538 ˚C fraction). An increase in aromaticity accompanied by an increase in boiling point matches the increase in carbon number predicted by Boduszynski et al. 13, 22 Based on that model, the hydrocarbon class (compounds with no heteroatoms) should exhibit the highest molecular weight for a given boiling point.

Figure 4.4 shows DBE vs. carbon number images for the S1 class spanning the HVGO distillation series. As for the hydrocarbon class, the carbon number increases from ~18 f (IBP-343 ˚C cut) to ~39 (500-538 ˚C

77 cut). Again the most pronounced difference in carbon number occurs between the lightest (heaviest) cut and the next nearest cut. The average molecular weight again increases steadily with increasing boiling point. and the high DBE tilt to the left indicates higher DBE species of the same boiling point must have slightly lower carbon number.

Figure 4.4. DBE vs. carbon number images for the S1 class for Athabasca bitumen HVGO distillate cuts. The carbon number abundance distribution maximum shifts from ~C18 at IBP-343 ˚C to ~C39 at 500-538 ˚C but for ~2 fewer carbons than for pure hydrocarbon analogues (Figure 3). DBE values increase from DBE = ~5 to DBE = ~10 with increasing boiling point, as for the hydrocarbon class.

The compositional dependence on boiling point is further elucidated by the DBE vs. carbon number images for the S2 class (Figure

4.5). The average carbon number jumps from C17 to C22 from the lowest

78 to next-highest boiling point cut, but steadily increases thereafter to a

maximum of C38 for the highest-boiling cut. Similar plots for all 10+ classes in the bitumen HVGO distillation series reveal identical trends. All classes exhibit an abnormally high jump in carbon number from the IBP fraction to the 343-375 ˚C cut with steadily increasing carbon numbers through cuts 2-7 and another jump from the 475-500 ˚C to 500-538 ˚C cut.

Figure 4.5 DBE vs. carbon number images for the S2 class for Athabasca bitumen HVGO distillate cuts. The carbon number abundance distribution maximum shifts from ~C18 at IBP-343 ˚C to ~C39 at 500-538 ˚C but for ~2 fewer carbons than the S1 class (Figure 4) and ~4 fewer carbons than the hydrocarbon class (Figure 3). DBE values increase similarly from DBE = ~6 to DBE = ~11 with increasing boiling point.

79 Comparison of the average carbon number in each distillate cut between classes (hydrocarbon, S1, and S2) provides clinching support for the Boduszynski model. Specifically, each heteroatom addition (from hydrocarbon to S1 and then to S2) results in a decrease in approximately 2-3 carbon atoms within a given boiling point range from the hydrocarbon (Figure 4.3) to S1 (Figure 4.4) and finally, the S2 class (Figure 4.5). For example, the hydrocarbon class for the 450-475 ˚C cut exhibits an average carbon number of 33. For the same boiling cut, the addition of a sulfur atom shifts the average carbon number to 31. Similar trends are readily apparent for all distillate fractions, directly validating the Boduszynski model. The distillate results for the 343-450 ˚C distillate ranges (4 cuts) are collected for the hydrocarbon, S1, and S2 classes in Figure 4.6. Progression from the hydrocarbon class to the S1 and S2 classes yields the predicted decrease of 2-3 carbons for all four distillate cuts. For a given distillate cut, progression from a hydrocarbon to monoheteroatom-containing species drops the number of carbons by

2-3. Going from mono- (S1) to di- (S2) heteroatom-containing class, the carbon number drops by another 2-3. Importantly, trends in carbon number and aromaticity, accompanied by increased heteroatom content as a function of boiling point are gradual and continuous within and between all classes and match those proposed by Boduszynski.

Figure 4.6 Composite DBE vs. carbon number images for the hydrocarbon, S1, and S2 classes for four of the eight HVGO distillate fractions shown in Figures. 3-5. Within each boiling range, each increase in one sulfur shifts to lower carbon number which corresponds to results in ~2-3 fewer carbons per structure.

Summarized another way, Figure 4.7 illustrates the relation between heteroatom class (hydrocarbon, S1, S2, and S1O1) and boiling point for a fixed carbon number (~25 It is clear that, for a given molecular weight (carbon number) each additional sulfur atom corresponds to an increase of ~25 °C in boiling point. Finally, although the SO and S2 classes both contain two heteroatoms per molecule, the increased polarity of the S1O1 class displays a higher boiling point than the (likely) dithiophenic S2 class. In addition, a decrease in aromaticity is

81 noted for the S1O1 class. That behavior is predicted by Boduszynski, because polar atoms such as oxygen are capable of hydrogen bonding and can participate in other polar interactions that result in a higher boiling point. 2

Figure 4.7 DBE vs. carbon number images for four distillation cuts. Here, for a given carbon number (~24-25), each additional heteroatom is seen to increase the boiling point by ~25 ˚C.

Differences in DBE and carbon number among nonpolar

hydrocarbon and S1 classes and polar, acidic O2 species are shown in Figure 4.8. (Because carboxylic compounds are not efficiently ionized by

positive-ion APPI, data for the O2 class was acquired by negative-ion ESI 9.4 T FT-ICR MS.) For the 425-450 ˚C boiling cut the average carbon number decreases by 2 carbons (from ~C30 for hydrocarbons to ~C28 for

82 the S1 class). Proceeding from a monoheteroatom-containing species (S1) to a highly polar diheteroatom-containing species (O2) results in a slightly lower carbon number (~C27) but a large drop in the aromaticity from DBE = 7 to DBE = 4), because polar compounds exhibit stronger intermolecular forces that result in higher boiling points at a given molar mass. Thus, acidic O2 species display slightly lower molecular weights and strikingly lower aromaticity than their nonpolar counterparts for each boiling cut. Similar trends were noted for both acidic and basic species across distillation range.

84 Figure 4.8. DBE vs. carbon number images for the hydrocarbon (APPI), S1 (APPI), and acidic O2 (ESI) classes from the 425-450 ˚C and 475-500 ˚C HVGO distillation cuts of whole Athabasca bitumen. Proceeding from hydrocarbon to S1 for either cut, the carbon number decreases by 2. Polar O2 classes, most likely from carboxylic functionalities, contain 3 fewer carbons than hydrocarbons and 1 fewer than monoheteroatomic S1 classes.

The HVGO Compositional Continuum. To further illustrate the continuity of HVGO compositional variation with boiling point, we combined data for the hydrocarbon class for each boiling point fraction into a single DBE vs. carbon number image (Figure 4.9). Data for any one spectrum for a given boiling point are presumably scaled to 100 as the highest-magnitude peak and (individually scaled) mass spectra were added together. As the boiling point increases, the carbon number shifts continuously from ~C15 (IBP-343 ˚C) to ~C55 (500-538 ˚C). The high

abundance of relatively low molecular weight species (>C20) has been discussed previously and is attributed to the lack of a well-defined starting temperature for the lowest-boiling fraction. Carbon number

increases gradually from ~C22 for boiling cuts defined in the HVGO temperature range. Aromaticity also gradually increases from DBE = ~2 to DBE = ~20, (average of ~3-4 aromatic rings). All other heteroatom classes identified by positive-ion APPI and positive/negative-ion electrospray resulted in similar plots. These results irrefutably demonstrate that crude oil is continuous in composition, structure, and boiling point across the HVGO temperature range.

Cycloalkane Linkages Characterizing the structure of high- boiling, highly polar compounds such as asphaltenes and resins requires a thorough understanding of the structural progression of lower boiling crude oil compounds. Bitumen is also referred to as "extra-heavy oil" and contains ~15-17% asphaltenes by weight with cP viscosity > 100,000 and API gravity of 7-15°.7, 12 To characterize species found in the heaviest crude oil ends, we first examine the combined bitumen HVGO fractions

85 and note the structural progression of aromatic ring systems by APPI FT- ICR MS.

Figure 4.10 shows DBE vs. carbon number images for the

hydrocarbon S1 class for HVGO derived from Athabasca bitumen. Thiophene corresponds to a DBE of 3. Addition of one or two phenyl rings yields benzothiophene (DBE = 6) and dibenzothiophene (DBE = 9) core structures (DBE = 9). However, the do not exhibit high relative abundance for DBE "magic number" values corresponding to those aromatic cores, and there are many species with intermediate DBE values. The intermediate DBE values cannot be due to alkene linkages, because alkenes are not found in crude oil.178 Also, APPI can result in proton transfer, which changes the DBE by 0.5, but doesn't affect the present argument. 179 Thus, cycloalkane ring addition must account for the species found with DBE values of 4-5 and 7-8 and HVGO must contain species that progress in DBE through the addition of cycloalkane rings. The effect is most easily evident at low DBE values where structural degeneracy in aromatics is minimized. At higher DBE values, the possibility of multiple structural isomers clouds assignment of integer DBE value increments. Whether or not the cycloalkane substituted aromatic structural motif extends into the heavy ends and asphaltenes has yet to be demonstrated by high resolution mass spectrometry. Nevertheless, the Boduszynski model predicts that the seeds of structural diversity in the lower boiling fraction augur that the continuity model demands their presence in the heavier ends. If large condensed aromatic ring systems indeed occur in heavy crude oil and are bridged by cycloalkane or heteroatom-containing 5 membered rings, it should be possible to crack across those linkages and produce the distillable hydrocarbons produced in heavy end and asphaltene conversion. We are currently exploring that possibility.

Figure 4.9 Combined DBE vs. carbon number images of for all distillation cuts combined for the hydrocarbons class from class from Athabasca bitumen HVGO. Carbon number and DBE values increase monotonically with increasing boiling point across the entire series. The Boduszynski model is irrefutably supported by this Figure: crude oil composition is continuous in carbon number, DBE, and boiling point.

Figure 4.10 shows DBE vs. carbon number images for the hydrocarbon S1 class for HVGO derived from Athabasca bitumen. Thiophene corresponds to a DBE of 3. Addition of one or two phenyl rings yields benzothiophene (DBE = 6) and dibenzothiophene (DBE = 9) core structures (DBE = 9). However, the do not exhibit high relative abundance for DBE "magic number" values corresponding to those aromatic cores, and there are many species with intermediate DBE

87 values. The intermediate DBE values cannot be due to alkene linkages, because alkenes are not found in crude oil.178 Also, APPI can result in proton transfer, which changes the DBE by 0.5, but doesn't affect the present argument. 179 Thus, cycloalkane ring addition must account for the species found with DBE values of 4-5 and 7-8 and HVGO must contain species that progress in DBE through the addition of cycloalkane rings. The effect is most easily evident at low DBE values where structural degeneracy in aromatics is minimized. At higher DBE values, the possibility of multiple structural isomers clouds assignment of integer DBE value increments. Whether or not the cycloalkane substituted aromatic structural motif extends into the heavy ends and asphaltenes has yet to be demonstrated by high resolution mass spectrometry. Nevertheless, the Boduszynski model predicts that the seeds of structural diversity in the lower boiling fraction augur that the continuity model demands their presence in the heavier ends. If large condensed aromatic ring systems indeed occur in heavy crude oil and are bridged by cycloalkane or heteroatom-containing 5 membered rings, it should be possible to crack across those linkages and produce the distillable hydrocarbons produced in heavy end and asphaltene conversion. We are currently exploring that possibility.

Figure 4.10 DBE vs. carbon number images for the S1 class of the Athabasca bitumen HVGO feedstock for all distillation cuts combined. The number of aromatic rings corresponding to various DBE values are shown for representative structures. Because the abundance distribution is monomodal (i.e., no "magic numbers"), including significantly abundant species with DBE values intermediate between those of fused aromatic rings, cycloalkyl-ring addition must be invoked to account for the intermediate DBE values.

Conclusions

89 Detailed compositional characterization of an HVGO distillation series exposes the relationship between molecular weight, aromaticity and boiling point. More than 70,000 mass spectral peaks provided elemental compositions with mass errors below 400 ppb and allowed for calculation of DBE values, and Kendrick sorting identified homologous series within each distillation cut. Within each heteroatom class, an increase in carbon number increases the boiling temperature. Among classes for a given boiling point, the compositional changes (carbon number and DBE) conform closely to the Boduszynski model, even for polar species. DBE vs. carbon number image "tilt" suggests that the highest DBE species must have fewer carbons to reside in the same distillate cut with heavier, less aromatic species. However, that trend becomes less pronounced as the complexity (boiling point) of the cut increases, most likely due to a pronounced increase in the number of possible structures that contain combinations of aromatic and cycloalkane moieties. The lack of "magic" DBE values corresponding to polyaromatic cores (e.g., thiophene, DBE = 3; benzothiophene, DBE = 6; and dibenzothiophene, DBE = 9), even at low DBE, clearly underscores the importance of cycloalkane structures in the overall contribution to DBE. In summary, we have provided detailed, comprehensive analysis of compositional trends within the HVGO boiling range to provide the first definitive evidence for the validity and accuracy of the Boduszynski model. In the next paper in this series, we shall extend our analysis through the distillation upper limit and into nondistillable residues.

90 CHAPTER 5. THE COMPOSITION OF HEAVY PETROLEUM: EVOLUTION OF THE BODUSZYNSKI MODEL TO THE UPPER LIMIT OF DISTILLABLE PRODUCTS BY ULTRAHIGH RESOLUTION FT-ICR MASS SPECTROMETRY

Summary

Heavy petroleum fractions present a structural and compositional complexity that complicates characterization by routine analytical techniques. Here, we present the detailed characterization of a Middle Eastern heavy crude oil distillation series to provide further evidence in support of the Boduszynski continuity model, which states that molecular weight, aromaticity and heteroatom content of heavy crude oil fractions increases with boiling point. Ultrahigh resolution Fourier transform ion cyclotron resonance mass spectrometry (FT-ICR MS) provides ultrahigh resolving power and mass accuracy and thereby allows for elemental assignment for each peak in a crude oil sample. Previous work has provided supportive evidence for the Boduszynski continuity model for heavy vacuum gas oil distillation series. Here, we extend the continuity model from low boiling distillate fractions to the upper limit of distillable products for a conventional crude oil and results shown within will provide further evidence to support the Boduszynski continuity model.

Introduction

One of the fundamental analytical techniques used in the oil industry is distillation, which limits the complexity of crude oil by reducing the number of compounds based on volatility. Complete compositional characterization of the each boiling point range allows companies to develop refinery strategies based on product yields for different crude oil feeds. Distillation is the main separation process used

91 in refineries and one of the most important properties of a whole crude oil is its boiling point distribution.2, 3 Distillation separates compounds in a whole crude based on volatility and reduces the number of molecules present by limiting molecular weight and structure of compounds to those boiling within a given temperature range. Oil companies us distillation assays (also known as distillation or boiling profiles) for feedstock evaluation and to determine the amount of light, low-boiling gasoline and transportation fuel within a crude without upgrading and to help predict and minimize possible adverse reactions during processing due to incompatibility, storage or refining processes associated with a particular feed.3,180 Identification and separation of compounds in distillate fractions is simplified compared to the whole crude, since each compound type covers only a small molecular weight range and the molecular weight range of each type of compound type is unique.2 Boiling point ranges of refinery feeds and products assist oil companies in strategy development and help predict the economic impact prior to production of a crude. One of the most comprehensive studies on heavy oil composition was conducted by Boduszynski et al. who used distillate fractions to characterize heavy crude oil as a function of increasing boiling point.13, 22, 106 Heavy crude is defined as crude with a high density (API gravity between 10-20°) and density increases with decreasing H/C ratio due to the increasing hydrogen deficiency of the molecule and increasing aromaticity (DBE).13 For a crude oil, the pure hydrocarbon content varies between more than 90% for a light, paraffinic petroleum and 50% (by weight) for heavy crude oil.3 Compounds containing heteroatoms (such as nitrogen, oxygen, sulfur and metals such as vanadium, nickel and iron) are distributed over the entire boiling range of straight-run distillate fractions of crude oil, but increase in concentration in higher boiling fractions and nondistillable residue.9 These species are responsible for problems in the refining, transportation, storage and deposit formation of

92 crude oil and are in higher concentration in heavy crude oils than light crude. However, the ratio of carbon to hydrogen remains constant for all crude oil densities, approximately 83-87% carbon and 11-14% hydrogen (by weight) for light and heavy crudes.3 The atomic hydrogen-to- carbon ratio (H/C) decreases in the higher-boiling fractions and resids, which are dominated by polynuclear aromatics, such as multiring cycloalkane, aromatic- and poly-aromatic structures with minimal alkyl branching and are less reactive than lighter distillate fractions with higher H/C ratios and higher paraffinic content.3, 9 As refinery feedstocks shift to heavier ends, comprehensive compositional characterization is key for strategy development for increasing the H/C ratio and facilitating heavy feed conversion into high-value products.3 Analytical techniques routinely used for lighter feedstocks require additional modification to successfully characterize heavy crude oil. For lighter boiling fractions, such as light and heavy naphthas (boiling range of IBP – 220 °C), a single analytical technique such as gas chromatography (GC) or GC/MS can provide detailed compositional analysis, but produces little to no information for more complex, higher boiling feeds. The advent of 2-dimensional GC (GC x GC)113 and high- temperature GC,123 has made detailed characterization of branched and normal alkanes, alkylcyclopentanes, alkylcyclohexanes and alkyl aromatics possible but still is unable to resolve heavy, high boiling species. 19, 166 The increased complexity of higher boiling middle distillate, kerosene and diesel fractions (220 – 345 °C boiling range) further complicates GC/MS analysis and resolution of individual compounds complicates data interpretation. 181 Comprehensive two-dimensional GC (GC x GC) increases the resolving power of traditional GC by subjecting each petroleum compound to two different stationary phase selectivities.111, 182 GC x GC coupled to various detectors has developed widespread use for hydrocarbon characterization of middle distillates, whose final boiling points are compatible with the maximum allowable

93 column temperatures. 112, 113, 183, 184 However, for characterization of higher boiling residues (> 540 °C), extensive separation and is necessary prior to most analytical techniques, including GC, GC x GC and HPLC. Boduszynski et al. provided the first detailed examination of the compositional analysis of heavy petroleum fractions summarized in a comprehensive series of papers an a book.2, 13, 22, 105, 106 Boduszynski et al. combined complementary separation techniques with mass spectrometry to characterize the molecular nature of heavy crude oil to show that “…compositional trends in fractions of increasing boiling point are continuous and that this continuity extends even to nondistillable residues”.2 Through use of the atmospheric equivalent boiling point (AEBP), calculated from molecular weight and densities or molecular weight and H/C ratio, Boduszynski characterizes the boiling point range of distillable species but importantly postulates the hypothetical extension to nondistillable residues. 2, 105 The “continuity concept” states that crude oil composition and AEBP is continuous between the light, low boiling species to complex, highest-boiling fractions of crude oil.2 With an increase in boiling point, an increase in complexity, molecular weight range, molecule type and structure existed. Therefore, property prediction for more problematic, heavier fractions could be extrapolated from results obtained for lower-boiling fractions, more suited to analytical techniques of the day.2 Thus, the “continuity concept” of the Boduszynski model predicted that crude oil is a continuum in molecular weight, structure and functionality from low boiling fractions to the nondistillable residues by extrapolation of results obtained from lower boiling fractions. A plot of molecular mass versus AEBP can be used to compare different crude oils on a common basis and emphasizes the importance of knowing the boiling range of the fraction before deciding the best technique for characterization.2 Also, the correlation between molar mass, heteroatom type and boiling point can easily be observed across the entire range of petroleum products.

94 With the advent of modern ionization techniques, such as electrospray ionization, mass spectrometry has increased in popularity as a tool for petroleum characterization at the molecular level. The development of electrospray (ESI) as an ionization technique for mass spectrometry by John Fenn expanded the applications of mass spectrometry to polar species in complex biological and environmental samples.40 Zhan and Fenn later applied electrospray ionization to analyze petroleum distillates to examine the highly problematic polar species present in crude oil samples thought responsible for corrosion, deposition and conversion problems during refining.44 Their work helped pave the way for the development of petroleomics, the prediction of chemical and physical properties and behavior from chemical composition to aid in processing problems.42, 71, 75 The work of Quann and Jaffee in the early 1990s further facilitated the development of petroleomics and concluded that detailed measurement of compound classes, types and carbon number distributions of feedstocks are critical to managing refinery processes.38, 39 Before the complete molecular characterization of crude oil could begin, a single analytical technique was needed that could meet the challenges associated with complex mixtures such as crude oil. With the development of Fourier transform ion cyclotron resonance mass spectrometry (FT-ICR MS),53, 185 complex mixture analysis without prior separation became possible. The development of high-field, high- homogeneity, temporally stable superconducting magnets allowed for FT- ICR mass spectrometers to evolve into important tools for complex mixture analysis.41, 64, 186, 187 Compared to other mass spectrometry techniques, FT-ICR MS is unparalleled in the ability to characterize heavy crude oil on the molecular level. The ultrahigh resolving power

(m/Δm50% > 400,000, in which Δm50% is the magnitude-mode mass spectral peak full width at half-maximum peak height) and sub-ppm mass accuracy (<400 ppb rms error) of FT-ICR MS allow for baseline

95 resolution of closely spaced isobaric species and unambiguous molecular formula assignment of the tens of thousands of species present in a single mass spectrum of crude oil. Electrospray ionization produces protonated and deprotonated ions from neutral analytes was coupled to FT-ICR for characterization of polar species in petroleum products.108 Mass spectral analysis via positive-ion electrospray (basic species) and negative-ion electrospray (acidic species) resulted in two mass spectra that when combined, result in the resolution and identification of over 17,000 polar compounds from a single crude oil.64 However, ESI is not able to ionize nonpolar species such as polyaromatic hydrocarbons, thiophenes and furans, all highly abundant in crude oil. Continuous-flow field desorption/ionization produces ions from nonpolar species not attainable by ESI but is time-consuming since the current to the FD emitter must be ramped over several minutes for complete volatilization/ionization of species of increasing boiling point.45, 67 With the implementation and development of atmospheric pressure photoionization (APPI) FT-ICR for petroleum analysis,87, 89, 91 nonpolar species are easily ionized at atmospheric pressure. However, the need for ultrahigh resolving power for APPI FT-ICR MS of petroleum is critical to assign elemental formulas to each peak in a single spectrum since APPI produces ions from polar and nonpolar aromatic species, approximately five times as many species (and five times as many peaks per spectrum) as ESI.91 APPI ionizes through irradiation with a Krypton lamp (~10 eV, 120 nm photons) after sample desorption into the gas phase via a pneumatically-assisted nebulizer chamber. A single neutral analyte molecule may therefore produce both radical molecular cations (M+) and protonated ions (M + H)+ in positive mode APPI.89 Ultrahigh resolving power is critical for APPI analysis of petroleum since a single neutral

13 analyte can produce isobaric species differing on composition by SH3 C

12 89, 91 vs. C4 (corresponding to 0.0011-Da mass difference).

96 Previous work characterized a bitumen heavy vacuum gas oil (HVGO) distillation series by positive- and negative-ion ESI to characterize the acidic and basic species as a function of boiling point.67 Subsequent work on the same HVGO distillation series characterized nonpolar (and polar) species by APPI FT-ICR MS and provided the first and most comprehensive data to support the Boduszynski continuity concept.179 The continuity concept postulates that compositional trends observed in low and middle petroleum distillates continue into higher boiling species not able to be analyzed by available technology of the time, relying on extrapolation to follow the continuum to the limits of distillation. Previous work by Qian et al. used field desorption ionization mass spectrometry (FDMS) to characterize the molecular weight distributions of heavy petroleum fractions and to outline the continuity of molecular weight as a function of boiling point.24 However, FDMS can create high molecular weight species through gas-phase reactions and produce erroneously high molecular weight measurements. Since then, advancements in analytical techniques, such as FT-ICR MS, have made possible compositional characterization the more complex, higher boiling fractions of crude oil. Here, we combine APPI FT-ICR MS analysis of a Middle Eastern heavy crude oil distillation series to extend the Boduszynski continuity model for heavy crude oil composition to the upper limits of distillation. We selected a heavy crude oil distillation series to compare observed changes in molecular weight, heteroatom content (type) and aromaticity (DBE, double bond equivalents, the number of rings plus double bonds) as a function of boiling point to the hypothetical model proposed by Boduszynski.

Experimental Methods

Sample preparation.

97 Distillation of Middle Eastern crude oil was performed in two stages. Four distillate cuts were generated from the initial distillation in a traditional pot still : IBP – 191 (data not shown), 191- 315, 315 and 371+ °C. The residue, 371 °C, was further fractionated in a vacuum flash unit described as follows. A pre-warmed feed is trickled into a heated vessel held at ~ 1520 Torr to facilitate the flash volatilization of material over an atmospheric equivalent temperature related to the vacuum and flash vessel temperature. Since this method does not employ a column, there is only one theoretical plate in this separation which results in poor separation. However, since contact times at each temperature are short, yield is maximized with minimal coking and decomposition. This method produced additional distillate cuts from 371-482 (data not shown), 371- 510, 510-538, 538-593 and the residue left in the distillation chamber

+ (593 °C)after collection of the 538 – 593 °C fraction. Each fraction (~10 mg) was diluted with 5 mL of toluene to make a stock solution. The stock solution was further diluted in toluene to yield a final concentration of 500 g/mL for analysis by APPI FT-ICR MS without additional modification.

Instrumentation.

Atmospheric Pressure Photoionization (APPI). The APPI source (ThermoFisher Scientific, San Jose, CA) was coupled to the first pumping stage of a custom-built FT-ICR mass spectrometer (see below) through a custom-built interface.89, 91 A Hamilton gastight syringe (5 mL) and syringe pump were used to deliver the sample at a rate of 50 L/min to

the heated vaporizer region of the APPI source where N2 sheath gas was introduced at 50 p.s.i. to facilitate nebulization while the auxiliary port remained plugged. After nebulization, the sample flows from the heated vaporizer as a confined jet and passes under a krypton vacuum ultraviolet lamp (10 eV photons, 120 nm) where photoionization occurs.

98 Dopant-assisted APPI often uses toluene to enhance ionization efficiency for nonpolar aromatic compounds.99, 147 In the APPI source, the nebulizer heater is operated between 250-375 °C according to previous nebulization temperature optimization for crude oil boiling point ranges.154 Charge exchange and proton transfer reactions occur between ionized toluene molecules and neutral analytes through collisions in the ionization region at atmospheric pressure.91, 99 Protonated ions form half- integer DBE values (DBE = c – h/2 + n/2 + 1, calculated from the ion

elemental composition, CcHhNnOoSs) and may be used distinguished from radical cations with integer DBE values.

9.4 Tesla FT-ICR MS. Middle Eastern crude oil distillate fractions were analyzed with a custom-built FT-ICR mass spectrometer equipped with a 9.4 Tesla horizontal 220 mm bore diameter superconducting solenoid magnet operated at room temperature (Oxford Corp., Oxney Mead, U.K.) and a modular ICR data station (PREDATOR) facilitated instrument control, data acquisition and data analysis.50-52 Positive ions generated at atmospheric pressure were accumulated in an external linear octopole ion trap49 for 250-1000 ms and transferred by rf-only octopoles to a 10 cm diameter, 30 cm long open cylindrical Penning ion

trap. Octopoles were operated at 2.0 MHz and 240 Vp-p amplitude. Broadband frequency sweep (chirp) dipolar excitation (70 – 700 kHz at 50

Hz/s sweep rate and 350 Vp-p amplitude) was followed by direct-mode image current detection to yield 8 Mword time-domain data sets. Time- domain data sets were co-added (200-300 acquisitions), Hanning apodized, and zero-filled once before fast Fourier transform and magnitude calculation.188

Broadband Phase Correction. Due to an increase in complexity at higher m/z values, , a broadband phase correction was applied to the entire mass spectrum for the 593+°C fraction to increase resolution for

99 isobaric species.189, 190 Briefly, this technique provides as much as a factor

191,69 of 2 increase in mass resolving power (m/Δm50%). In standard FT-ICR mass spectra, the frequency domain is a linear combination of absorption- and dispersion-mode spectral components.190 With an applied broadband phase corrections, absorption- and dispersion-mode line shapes are fitted to a subset of resolved peaks to accurately calculate the true phase vs. frequency spectrum for the entire mass spectrum.189 The detailed theory of this method can be found elsewhere.189, 190, 192-194 Calibration and data analysis was subsequently performed in the same manner as non-phased spectra and is described below.

Mass Calibration and Data Analysis. ICR frequencies were converted to ion masses based on the quadrupolar trapping potential approximation176, 195, 196 and internally calibrated with respect to a highly abundant homologous alkylation series differing in mass by integer 61 multiples of 14.01565 Da (mass of a CH2 unit). Experimentally measured masses were converted from the International Union of Pure and Applied Chemistry (IUPAC) mass scale to the Kendrick mass scale68 to identify homologous series for each heteroatom class (i.e., species with the same CcHhNnOoSs content, differing only by their degree of alkylation).41, 71 Peak assignments were performed by Kendrick mass defect analysis as previously described.61 For each elemental composition, CcHhNnOoSs, the heteroatom class, type (double bond equivalents, DBE = number of rings plus double bonds involving carbon)177 and carbon number, c, were tabulated for subsequent generation of heteroatom class relative abundance distributions and graphical DBE vs. carbon number images.197

Results and Discussion

100 Figure 5.1 shows a broadband positive-ion APPI FT-ICR mass spectrum of the 593+ °C residual fraction for a Middle Eastern heavy crude oil. The achieved resolving power of 580,000 at m/z results in 26,896 mass spectral peaks, each with magnitude greater than 6σ of baseline rms noise. The time-domain signal duration was 5.6 s (figure 5.1, zoom inset) and the signal did not damp completely to zero during the acquisition period, indicating that the resolving power achieved could be slightly higher than what we report. Due to the increased complexity (~ 91 mass spectral peaks per nominal mass > 6σ baseline rms noise) associated with residual samples, such the 593+ °C fraction, we have applied a broadband phase correction to improve resolution as discussed previously.

101 Figure 5.1 Broadband positive-ion APPI FT-ICR mass spectrum of a Middle Eastern heavy crude oil 593+ °C distillate fraction. 26,896 mass spectral peaks were observed at 6 times the signal-to-noise ratio baseline rms noise, at an average mass resolving power, m/Δm50% = 580,000 at m/z 800.

Figure 5.2 shows the broadband mass spectrum for each distillate fraction for Middle Eastern heavy crude oil. As the distillation temperature increases, so too does the upper molecular weight limit and the molecular weight distribution for each fraction. For example, the 191 – 315 °C fraction ranges from 175 < m/z < 400 and is centered at m/z 275 whereas the 593+ °C fraction ranges from 500 < m/z < 1200 and is centered at m/z 800. Each distillation cut contains a variety of compound types each comprised of a homologous alkylation series and higher-boiling fractions contain higher molecular weight species than lower boiling fractions. 13 An increase in the width of the molecular weight distribution is also observed with increasing boiling point. The broadening of the molecular weight distribution reflects the increased complexity attributed to an increase in the number of carbon atoms per molecule due to the increasing number of isomers so that the most complex fraction, 593+ °C resid, contains the most mass spectral peaks and produces the most complex mass spectra. Boduszynski outlined these two principles using field ionization and field desorption mass spectrometry for five distillation cuts and three sequential elution fractionation (SEF) fractions from Boscan atmospheric residue. 13 Here, we show APPI FT-ICR MS results that support the Boduszynski model by with a Middle Eastern heavy crude oil distillation series and provide evidence of the global continuum in carbon number and DBE of crude oil composition.

102

103 Heteroatom Class Distribution. A condensed class distribution graph is shown in Figure 5.3 for five distillate fractions and the resid from Middle Eastern heavy crude oil. Compounds containing a single sulfur atom were the most abundant across the entire boiling range,

followed by the hydrocarbon class, S2 and the SO classes. At lower boiling point, the lighter fractions (191 – 315 °C and 315 – 371 °C) have a higher relative abundance of pure hydrocarbon species than other heteroatom class. Low molecular weight compounds will have the lowest boiling point and therefore will be most abundant in lower-boiling fractions. As the boiling point increases, a decrease in relative abundance of the hydrocarbon class is observed along with an increase

in the multiple-heteroatom classes (S1 and S2). The same results were observed for an Athabasca bitumen HVGO distillation series previously reported.154, 179 Again, the results agree with the Boduszynski model which states that “diverse compounds with similar molecular weights cover a broad boiling range; conversely, a narrow distillation cut can 13 contain a wide range of molecular weights”. Increasing intermolecular forces affect the heat of vaporization (and therefore the boiling point) of a compound, with weak dispersion forces dominating in alkanes and increasing with chain length, followed by stronger intermolecular forces between aromatic rings systems and polar compounds.2

104

DBE vs. Carbon Number Images. A plot of carbon number versus boiling point introduced by Boduszynski illustrates the effect of boiling point on molecular weight, structure and heteroatom functionality in crude oil. The normal paraffins, or saturated hydrocarbons, have the highest molecular weight (carbon number) for a given boiling point, followed by the naphthenic rings, aromatic hydrocarbons, non- alkylsubstituted aromatic hydrocarbons, alkylsubstituted aromatic hydrocarbons, mono- and multiple-heteroatomic compounds, aromatic polar species with the condensed polyaromatic molecules with several polar groups and little alkyl substitution falling on a hypothetical line with the lowest molecular weight for a given boiling point. 2, 13 The

105 original plot created by Boduszynski et al. is reproduced in figure 5.4 with permission. Ionization of saturated hydrocarbons is problematic since nearly all techniques result in extensive fragmentation, complicating the identification of neutral precursors. Therefore, we begin our comparison of APPI FT-ICR results with the hypothetical Boduszynski model using compounds containing one or more cycloalkyl- or aromatic ring structures. 75 Careful examination of the theoretical plot of molar mass vs. AEBP shows that the incorporation of a single heteroatom to a hydrocarbon core structure results in a loss of approximately 2-3 carbon atoms per molecule to remain within the same boiling range.

106

Figure 5.4 The Boduszynski model of the effect of molecular weight and structure on boiling point for heavy crude oil composition. Reprinted with permission.

To correlate these results, we use color-coded isoabundance- contoured plots of DBE vs. carbon number for the hydrocarbon class for the five distillate fractions plus the residue, shown in Figure 5.5. An increase in distillation temperature is accompanied by a shift to higher carbon number from 15 – 28 centered at ~18 for the 191 – 315 °C cut to 40 – 90 centered at ~61 for the 593+ °C residue. The average DBE value increases from 8 DBE for the 191 – 315 °C cut to 16 DBE for the 593+ °C resid. Compounds with multiple fused aromatic rings (higher DBE values) and polar functional groups (e.g. asphaltenes) have higher boiling points due to the increase in intermolecular forces within the compounds which results in a lower molar mass with increased aromaticity and heteroatom content. 13 According to the Boduszynski model, within a given boiling point, pure hydrocarbons have the highest molar mass with an increase in aromaticity (DBE) and heteroatom content. An increase in the number of heteroatoms and aromatic structures per molecule results in a shift to lower carbon number within each boiling point. Heavy crude oils, such as the Middle Eastern heavy crude we have selected, are known to be highly abundant in sulfur-containing species. Therefore, a comparison of the carbon number and DBE distributions for the hydrocarbon and S1 classes can be used to correlate the hypothetical model to a true data set.

107

108 Figure 5.5 Color-coded isoabundance contoured plots of DBE vs. carbon number for the hydrocarbon class of a Middle Eastern heavy crude oil distillation series and residue.

109

110 Figure 5.6 Color-coded isoabundance contoured plots of DBE vs. carbon number for the S1 class of a Middle Eastern heavy crude oil distillation series and the residue.

111 Figure 5.6 shows the color-coded isoabundance-contoured plot of

DBE vs. carbon number for the S1 class for each of the six distillate fractions. Again, a shift to higher carbon number is observed with increasing distillation temperature from 15-18 for the 191 – 315 °C cut to 30-90 for the 593+ °C resid. DBE values shift from 6 DBE for the 191 - 315 °C cut to 12 DBE for the 593+ °C resid. If we compare the carbon number of highest relative abundance for each distillation cut for the hydrocarbon class and the S1 class, we see that the limits proposed by Boduszynski hold true. For example, the hydrocarbon class for the 371 – 510 °C cut has a carbon number distribution of 22-52 with the highest abundance at 38 carbon atoms per molecule. Within the same boiling point range, the addition of a sulfur atom shifts the carbon number distribution to 20-50 with the highest abundance at 35 carbon atoms per molecule. The trend towards lower carbon number (in multiples of approximately 3 carbon atoms per molecule) with the addition of each heteroatom class correlates to the hypothetical model proposed by Boduszynski. Similar trends are observed for the other distillate fractions, with the shift being less noticeable in the resid fraction, since it is not a “true” distillation cut with a defined upper boiling point.

To correlate the trend with increasing heteroatom content, Figure

5.7 shows color-coded isoabundance-contoured plots for the S2 class from the five distillate fractions and the resid. Using the same example as above, the carbon number distribution for the 371 – 510 °C shifts even lower to 18-50 with the highest abundance at 32 carbon atoms per molecule. Again, the loss of approximately 3 carbon atoms per molecule is associated with the addition of each heteroatom class. The same observation exists for the rest of the distillate fractions with the exception of the 191 -315 °C fraction where no S2 species were observed above 1 % relative abundance.

112

113 Figure 5.7 Color-coded isoabundance contoured plots of DBE vs. carbon number for the S2 class of a Middle Eastern heavy crude oil distillation series and the residue.

114 To further exemplify the idea of continuity in composition and boiling point that extends across the entire boiling range of a crude oil,2 a single color-coded isoabundance-contoured plot for the hydrocarbon class is shown for the entire distillation series for Middle Eastern heavy crude in Figure 5.8. We have normalized all relative abundances such that the entire series can be observed simultaneously. Here, the gradual progression to higher carbon number (molecular weight) and DBE (aromaticity) extends from the low boiling species to the high boiling resid fraction. The crude oil continuum in carbon number and DBE extends to the limits of modern distillation and can therefore be used to characterize the most complex, challenging nondistillable fractions, i.e. asphaltenes.

115

Figure 5.9 shows the same continuity trend in carbon number and DBE for the S1 class in a single color-coded isoabundance-contoured

plot for the entire distillation series. A similar trend is observed for the S1 class which further validates the proposed continuity model of crude oil.

Conclusions Here, we present evidence that agrees with the theoretical model proposed by Boduszynski and others which argues that crude oil is a continuum in molecular weight, structure and boiling point. With the

116 advances made in ultrahigh resolution FT-ICR MS of heavy crude oil in the past two decades, we are able to provide undisputable data that supports the continuity model through comparison of molecular weight distribution, carbon number and DBE for a Middle Eastern heavy crude oil. Previous work has supported this model for middle distillate fractions (HVGO) from an unconventional crude.154, 179 We correlate the full distillation series for a single whole conventional crude oil to the Boduszynski model and further solidify proof of the global continuum of petroleum composition. The most problematic fraction of crude oil is the nondistillables (i.e. residuals and asphaltene fractions) and future work will focus on detailed compositional characterization of these fractions.

117 CHAPTER 6. THE TRUE MOLECULAR CHARACTERIZATION OF ASPHALTENES PART I. MOLECULAR WEIGHT AND DISCOVERY OF DISTILLABLE ASPHALTENES

Introduction

The controversy surrounding the molecular weight of the asphaltene fraction of crude oil is central to understanding and predicting asphaltene behavior. Ultimately, to eliminate problems that are attributed to asphaltenes in crude oil production requires a comprehensive understanding of the composition and interactions between asphaltene molecules. The most important characteristic of any chemical is its constituent elements : its chemical composition. Second is its molecular weight. 19 Elemental analysis and bulk property measurements have completely characterized asphaltene composition with little controversy; 28 however, asphaltene molecular weight is still controversial although a general consensus agrees that asphaltene molecular weight is less than 2 kDa. 2, 74 Because there is a general agreement that asphaltenes contain ~6- 7 fused rings (by direct imaging techniques)with alkane substituents, the discussion then is focused on whether asphaltenes are monomers or polymeric. 28, 35, 198 The chemical and physical properties of monomers and polymers are different significantly and it is therefore fundamental to the progression of asphaltene chemistry to resolve the debate on asphaltene molecular.28 The classification of asphaltenes is rooted in solubility producing a broad, general definition. Asphaltenes concentrate in residues and are therefore have been erroneously referred to as “nondistillables”. However, if asphaltene molecules have molecular weight distributions ranging from

118 300-2000 Da composed of condensed aromatic rings with minimal alkylation, a small fraction must be volatile. Very few compounds, if any, composed of C, H, O, N, S and trace metals with such properties boil above 500 ˚C. For decades, researchers have worked under the incorrect assumption that asphaltene fraction is nondistillable. 3 pg. 53. “The asphaltene constituents, being insoluble in low-boiling hydrocarbon liquids such as n-heptane, are also nondistillable, no matter from what source they are isolated.” 3 Here, we present the molecular weight of asphaltenes as determined by FT-ICR mass spectrometry by electrospray ionization (ESI) and atmospheric pressure photoionization (APPI) and in agreement with a host of analytical techniques. The discovery of a sub-fraction of asphaltene compounds which we refer to as “distillable asphaltenes” provide further evidence that the majority of the compounds that make up the asphaltene fraction are less than 2 kDa.

Experimental Methods

119 drops/minute for 60 min until all asphaltenes were completely desorbed from the filter paper. 199 Both asphaltene samples were rotary vacuum- evaporated to dryness, weighed and redissolved in toluene to produce a stock solution of 10 mg/mL. Two stock solutions were prepared by dissolving ~20 mg of asphaltene sample in 20 mL of toluene. A one mL aliquot was diluted with 1 mL of methanol that contains 2% by volume

formic acid or NH4OH for positive- or negative-ion mode electrospray analysis. Samples were further diluted in toluene to yield final concentrations for APPI analysis without additional modification. 9.4-T FT-ICR MS. A custom-built FT-ICR mass spectrometer is equipped with a 22 cm horizontal room temperature bore 9.4 Tesla magnet (Oxford Corp., Oxney Mead, UK) and a modular ICR data station (PREDATOR).50, 52 Positive ions generated at atmospheric pressure in the external APPI source enter the skimmer region at ~2 Torr through a heated metal capillary into the first rf-only octopole. Ions pass through a quadrupole to a second octopole where they accumulate for 250-1000 ms. Helium gas was introduced during accumulation to collisionally cool the ions before transfer through a 200 cm rf-only octopole into an open cylindrical Penning ion trap (10 cm i.d. x 30 cm long). Octopole ion

guides were operated at 2.0 MHz and 240 Vp-p rf amplitude. Broadband frequency chirp excitation (70 – 700 kHz at a sweep rate of 50 Hz/s and

amplitude, Vp-p = 350 V) accelerated the ions to a cyclotron orbital radius that was subsequently detected by the differential current induced between two opposed electrodes of the ICR cell. The experimental event sequence was controlled by a MIDAS (modular ICR data acquisition and analysis software) data station.51, 52 Multiple (100-300) time-domain acquisitions were summed for each sample, Hanning-apodized, and zero- filled once prior to fast Fourier transform and magnitude calculation.200

Mass Analysis. Asphaltene samples were analyzed at the National High Magnetic Field Laboratory (NHMFL) with a custom 9.4 Tesla Fourier

120 transform ion cyclotron resonance mass spectrometer. 50 Ions were generated externally by an ESI or APPI (ThermoFisher Scientific Corp., Bremen, Germany) ion source and accumulated for a period of 0.5 – 5s prior to introduction into the ICR cell. Multiple (100-300) time-domain acquisitions were summed for each sample, Hanning-apodized and zero- filled once prior to fast Fourier transform and magnitude calculation. A custom modular ICR data system (MIDAS) data station provides instrument control, data acquisition and data analysis. 52 Mass spectra were internally calibrated with respect to a known homologous series of heteroatom class specific to the ionization method. Homologous series were separated and grouped by nominal Kendrick mass and Kendrick mass defect to facilitate rapid identification. 61

Results and Discussion

Asphaltene molecular weight. Figure 6.1 shows a broadband positive-ion APPI FT-ICR mass spectrum of an asphaltene from a Middle Eastern heavy crude vacuum bottom residue. The molecular weight distribution is centered at m/z 800 between 200 < m/z < 2000 in agreement with previous results which find that asphaltene compounds are below 2 kDa. 2, 28, 167, 168, 171, 201 Asphaltenes concentrate in the high boiling, polar fractions and therefore typically contain the highest molecular weight compounds. 2-4 However, the low end of the molecular weight distribution contains compounds between 2—500 Da. Asphaltenes have generally been regarded as nonvolatile and having the highest boiling point of all petrocompounds. The consensus is that asphaltenes are small molecules composed of one fused condensed ring system with approximately 60% of the total carbon being in alkyl chains. By definition, asphaltene compounds of low molecular weight should

121 have boiling points attainable by distillation. Since the molecular weight distribution of asphaltenes shown in figure 6.1 contains compounds with m/z values between 200-500, at the very least, these compounds will have boiling points below 400 ˚C. Asphaltenes have long been characterized as being nonvolatile and the term “nondistillables” has been used to discuss asphaltenes in the same category as coke and residua. However, by definition, asphaltenes are a solubility fraction of crude oil. Therefore, compounds that are found in the asphaltene fraction at low molecular weight should be accessible through distillation. Phenanthrene, C14H10 (DBE = 10), for example, has a boiling point of 340 ˚C and is a common PAH core found in crudes.

122 Figure 6.1 Broadband positive-ion APPI LTQ mass spectra of an asphaltene fraction isolated from a Middle Eastern heavy crude vacuum bottom residue collected over 593+ ˚C. The asphaltene molecular weight distribution ranges from 250

If asphaltenes are composed mainly of compounds of low molecular weight compounds consisting of a single fused aromatic core, with slightly more than 50% saturated carbon, then to some degree, asphaltenes should be distillable. However, central to the asphaltene debate is their tendency to self-associate and form aggregates of approximately 6-12 monomer units through noncovalent interactions between aromatic cores (i.e., π-π stacking, hydrogen bonding, etc.). 29 Stable aggregates would then have a much higher boiling point than the individual monomer. Since asphaltene fused aromatic core structures can have varying degrees of alkyl substitution, the tendency to form aggregates varies between compounds. As the number of carbon atoms increases in alkyl chains, steric hindrance prevents noncovalent associations between fused cores, preventing a small amount of asphaltenes from being incorporated into aggregates. Low molecular weight asphaltenes with slightly longer alkyl chain lengths therefore remain as monomers in the crude oil matrix and can then distill based on volatility. More abundant asphaltene compounds with little or no alkyl substitution form nanoaggregates at very low concentrations in crude oil and consequently have volatilities representative of the associated aggregate molecular weight (~3-8 kDa). Therefore, asphaltene molecules have been referred to as “nondistillables” because the tendency to self- associate increases the boiling point of the asphaltene aggregate structure to temperatures far above the limit of distillation, which is approximately 600 ˚C. Since asphaltenes behave differently than their maltene counterparts, they must differ in composition. However, asphaltenes and maltenes are similar in carbon number distribution (figure 6.1), with both fractions mainly of relatively small molecules

123 weighing less than 2 kDa. 2 By definition, asphaltenes are n-heptane insoluble and toluene soluble whereas maltenes are soluble in paraffinic solvent systems. The ability to self-associate into aggregate structures inherent for asphaltene molecules increases the molecular weight of the monomer to the molecular weight of the aggregate, preventing small asphaltene molecules from being distilled at high abundance. Since most asphaltenes are aggregated, only a small fraction remains monomeric and is accessible by distillation.

Distillable asphaltenes. Small fused aromatic rings such as phenanthrene and naphthalene have boiling points of less than 350 ˚C and are known to exist in crude oil. 2, 3 Figure 6.2 shows positive-ion APPI LTQ mass spectra for asphaltenes fractionated from the 538-593 ˚C distillation cut of Middle Eastern heavy crude oil at increasing concentration. At 100 g/mL, the molecular weight distribution can be observed at very low signal to noise ratio. At an increased concentration of 250 g/mL, the molecular weight distribution ranges from 250 < m/z < 900 centered at m/z 450. A twofold increase in concentration does not shift the molecular weight distribution of the distillable asphaltene fraction. The molecular weight of the parent distillate, the 538-593 ˚C, was nearly twice the molecular weight as the distillable asphaltene fraction.

124

The mass scale-expanded segment of 538-593 ˚C distillate fraction (figure 6.3, top) and its asphaltene fraction (figure 6.3, bottom) at m/z 600 show an increase in spectral complexity associated with asphaltenes. A shift to lower mass defect for the asphaltene is indicative of an increase in aromaticity through dehydrogenation and dealkylation reactions compared to the parent. For example, the two most abundant ions in the parent correspond to a 9 DBE N1 species and 11 DBE N1S1

125 species with H/C ratios equal to 1.6 and 1.5, respectively. In the asphaltene fraction, the two most abundant ions have DBE’s of 17 and 19 with much lower H/C ratios between 1.1-1.2. The shift to lower mass defect is attributed to the aromaticity of the core structures. Since the exact mass of hydrogen is 1.007994 Da, the mass defect of the molecule shifts +0.007994 Da to the right in nominal space with each hydrogen atom. Saturated molecules with higher H/C ratios would therefore have higher exact masses and therefore larger mass defects. An increase in complexity is observed in the asphaltene fraction relative to the parent fraction, as evident by the increase in mass spectral peaks in a 1 Da window. Positive-ion APPI FT-ICR mass spectra also showed an increase in complexity (Figure 6.4).

126 Figure 6.3 Mass scale-expanded segment of a positive-ion electrospray FT-ICR mass spectrum of the 538-593 ˚C parent distillate and its asphaltene fraction. An increase in spectral complexity is observed for the distillable asphaltenes with a corresponding shift to lower mass defect indicating an increase in aromaticity. Since the mass defect of hydrogen is 0.007994, each addition of a hydrogen (increased saturation) shifts the total mass of a compound +0.007994.

DBE vs. carbon number images. Figure 7.4 shows color-coded isoabundance-contoured plot of DBE vs. carbon number for the hydrocarbon class obtained with positive-ion APPI for the parent distillate (left) and distilled asphaltenes (right). The parent distillate image is centered at DBE = 12 with a carbon number ranging from 35 to 70 which corresponds to a pyrene ring with alkyl chains to account for additional carbon and hydrogen. On the right, the distillable asphaltene image is bimodal, with two apparent hydrocarbon distributions. Between carbon number 27 to 42, compounds with DBE values between 17 and 28 are present in the distilled asphaltenes. For example, in the parent distillate, a compound with 38 carbons has a DBE values between 5-20 whereas the same carbon number in the asphaltene fraction corresponds to a DBE = 25, indicating an increased aromaticity and smaller H/C ratio. The second observed distribution at lower DBE indicates that nonasphaltene molecules were also removed simulataneously from the crude oil matrix, despite the additional Soxhlet extraction with additional n-heptane to remove entrained resin molecules.

127

The S1 class yielded similar isoabundance color-coded contoured plots of DBE vs. carbon number, shown in figure 6.5. The distillable asphaltene fraction covers a lower carbon number distribution and therefore S1 asphaltene compounds presume to have a lower molecular weight than their distillate homologues. However, the increased complexity (figure 6.3) found in the asphaltene fraction limits the number of ions that can be trapped and subsequently detected in the

128 ICR cell at a given time. Since there is nearly a three-fold increase in the number of compounds present at a given m/z, only a limited molecular weight distribution can be covered in a single spectrum. Stated another way, if the molecular weight distribution of the asphaltene sample ranges from 200 < m/z < 700, only the most abundant ions are detected at one time. There are more ions at a given m/z for the asphaltene and therefore only a limited m/z range can be collected at one time, since there is a limit to the number of ions the ICR cell can trap, excite and detect in a single acquisition without interferences from ion-ion interactions. For samples with exceedingly high spectral complexity, inherent to asphaltenes, a “heart-cut” is collected in which the most abundant ions are detected in a single spectrum. For the S1 class, the most abundant species have a carbon number ranging from 24 to 42, with 16 < DBE < 28 which is consistent with a thiophenic ring attached to 10-12 benzene rings. The most abundant species in the parent distillate have much lower DBE values between 9 and 12, consistent with a thiophene ring attached to 2 to 3 aromatic rings.

129

FT-ICR mass spectral analysis of the polar species derived from positive-ion APPI yielded similar results as nonpolar hydrocarbon and S1 classes (figure 6.3 and 6.4). Isoabundance color-coded contoured plats are shown for the sulfoxide class, S1O1, in figure 6.6. The DBE distribution for the parent distillate is centered at DBE = 10 whereas the distillable asphaltene fraction is considerably more aromatic, with the most abundant species 20 < DBE < 26.

130

Composite plots of DBE vs. carbon number. To determine where distillable asphaltenes lie in compositional space, a composite plot was made to show DBE vs. carbon number for the hydrocarbon series. Figure 6.7 shows the overlap between the hydrocarbon compounds found in the parent distillate (blue) and those collected in the asphaltene fraction (red). Here, compounds that are identified are marked in two-

131 dimensional space to determine their relation to one another. Compounds in the parent distillate and the distillable asphaltene are striking similar and therefore would be expected to have similar solubility. However, even a slight increase in alkylation prevents asphaltene molecules from self-associating through steric hindrance, therefore allowing asphaltene species to collect in fractions according to volatility along with maltene counterparts. A broad, defined distribution in carbon number (25 to 65) and DBE (3 to 28) is shown in blue for hydrocarbon species for the parent distillate. The distillable asphaltenes, or asphaltenes with a boiling point within the limit of distillation, cover a narrow carbon number range (15 to 40) due to the increase in complexity discussed earlier and are more aromatic with DBE values between 17 and 27.

132

Importantly, the distillable asphaltenes lie adjacent to maltene species, but are distinctly different in aromaticity, and therefore have different solubility. For example, a n-heptane insoluble (asphaltene) compound with 30 carbon atoms has a DBE value of 20. Compounds soluble in paraffinic solvents (i.e., heptane and pentane) with 30 carbons per molecule have lower boiling points than 538-593 ˚C and therefore distill at lower temperatures. Second, asphaltene compounds increase incrementally in carbon number and aromaticity with the addition of aromatic rings. Coronene, a pericondensed PAH commonly found in heavy crude has an elemental formula of C24H12 with 19 DBE, has a boiling point of 525 ˚C, slightly lower than the boiling range of the parent distillate. The addition of a benzene ring, C27H14, increases the DBE value to 21, a fused polycyclic hydrocarbon which is consistent with the distillable asphaltenes. A second benzene ring addition increases the elemental formula to C30H16, DBE = 23, also found in the range of detected asphaltene hydrocarbons. Integer DBE values can be accounted for through cyclohexane ring incorporation on the exterior of condensed aromatic ring structures.

133

Figure 6.8 shows the composite plot of DBE vs. carbon number for the S1 class. Similar trends are observed as figure 6.7 with the distillable asphaltene fraction shifted to slightly higher aromaticity for a given carbon number and heteroatom class than the parent distillate.

Dinaphthothiophene, C20H12S1, has a DBE value of 15 and is slightly below the collection temperature of the parent distillate. However, the addition of one benzene ring results in a DBE of 17 (C23H14S1)l, which is consistent with structures found in the distillable asphaltenes. A second benzene ring results in a DBE of 19 (C26H16S1) and is also consistent

134 with structures observed in the distillable asphaltenes. Similar to the hydrocarbon series discussed in figure 6.7, incremental DBE values can be accounted for through cycloalkane ring addition.

Conclusions

Historically, asphaltenes have been categorized as “nondistillables” and relegated to the heaviest, highest boiling fractions of crude oil. The general structure of asphaltenes as condensed polycyclic ring systems has been corroborated by a host of analytical techniques, including diffusion techniques, NMR and direct imaging as well as mass spectrometry. Asphaltenes are thought to contain 6-7 fused aromatic rings, slightly more than 50% saturated hydrocarbon with molecular weight distributions from 300 < m/z < 2000. At the low end of the molecular weight distribution are light polycyclic ring systems, with boiling points attainable by distillation. The majority of asphaltene molecules engage in noncovalent interactions between the aromatic cores (i.e., π-π stacking, hydrogen bonding) and form aggregates which have much higher molecular weights and therefore higher boiling points. However, a small abundance of asphaltenes remain as monomers and therefore distill with maltene counterparts. Distillable asphaltenes are shifted only slightly higher in aromaticity than the parent distillate for a given carbon number and exist in compositional space directly below asphaltenes. Since these compounds have smaller aromatic cores and slightly increased alkyl chain length relative to the majority of asphaltenes, interactions between aromatic cores is limited preventing aggregation. Here, for the first time,we introduce a new fraction of crude oil compounds called distillable asphaltenes, which are asphaltenes by definition (n-heptane insoluble/toluene-soluble) but have boiling points comparable to small, polycyclic ring systems. Distillable asphaltenes, long thought to be an oxymoron, are one of the fundamental keys to

135 unlocking the puzzle of the true structure of asphaltenes. Future research will explore the detailed definition of asphaltene and maltene composition.

136 CHAPTER 7. THE TRUE MOLECULAR CHARACTERIZATION OF ASPHALTENES PART II. THE DEFINITION OF ASPHALTENE AND MALTENE COMPOSITION

Introduction

Asphaltenes are the heptane-insoluble, toluene-soluble subfraction of crude oil and are responsible for a collection of problems associated with processing of heavy ends or crude oil recovery. Asphaltenes slow the overall rate of catalytic hydroprocessing, act as coke precursors which in turn causes catalyst deactivation and form sludge thereby limiting the maximum level of conversion possible in hydroconversion.3, 9, 33, 201 Increased viscosity, coke formation and product stability also are associated with the asphaltene content of a feedstock. Asphaltene deposition during production and transport reduces flow and can even cause well-blockage. 29, 33 Understanding the molecular nature of asphaltenes can be a valuable aid to understanding their behavior.

Experimental Methods

137 residue to complete the transfer of solids. The filter paper with the asphaltenes was then refluxed with heptane at a rate of 3-5 solvent drops/minute for 60 min until all asphaltenes were completely desorbed from the filter paper. 199 The asphaltene sample was then rotary vacuum- evaporated to dryness, weighed and redissolved in toluene to produce a stock solution of 10 mg/mL. A stock solution was prepared by dissolving ~20 mg of asphaltene sample in 20 mL of toluene. A one mL aliquot was diluted with 1 mL of methanol that contains 2% by volume formic acid or

NH4OH for positive- or negative-ion mode electrospray analysis. Samples were further diluted in toluene to yield final concentrations for APPI analysis without additional modification.

9.4-T FT-ICR MS. A custom-built FT-ICR mass spectrometer is equipped with a 22 cm horizontal room temperature bore 9.4 Tesla magnet (Oxford Corp., Oxney Mead, UK) and a modular ICR data station (PREDATOR).50, 52 Positive ions generated at atmospheric pressure in the external APPI source enter the skimmer region at ~2 Torr through a heated metal capillary into the first rf-only octopole. Ions pass through a quadrupole to a second octopole where they accumulate for 250-1000 ms. Helium gas was introduced during accumulation to collisionally cool the ions before transfer through a 200 cm rf-only octopole into an open cylindrical Penning ion trap (10 cm i.d. x 30 cm long). Octopole ion guides were operated at 2.0 MHz and 240 Vp-p rf amplitude. Broadband frequency chirp excitation (70 – 700 kHz at a sweep rate of 50 Hz/s and amplitude, Vp-p = 350 V) accelerated the ions to a cyclotron orbital radius that was subsequently detected by the differential current induced between two opposed electrodes of the ICR cell. The experimental event sequence was controlled by a MIDAS (modular ICR data acquisition and analysis software) data station.51, 52 Multiple (100-300) time-domain

138 acquisitions were summed for each sample, Hanning-apodized, and zero- filled once prior to fast Fourier transform and magnitude calculation.200

Mass Analysis. Asphaltene samples were analyzed at the National High Magnetic Field Laboratory (NHMFL) with a custom 9.4 Tesla Fourier transform ion cyclotron resonance mass spectrometer. 50 Ions were generated externally by an ESI or APPI (ThermoFisher Scientific Corp., Bremen, Germany) ion source and accumulated for a period of 0.5 – 5s prior to introduction into the ICR cell. Multiple (100-300) time-domain acquisitions were summed for each sample, Hanning-apodized and zero- filled once prior to fast Fourier transform and magnitude calculation. A custom modular ICR data system (MIDAS) data station provides instrument control, data acquisition and data analysis. 52 Mass spectra were internally calibrated with respect to a known homologous series of heteroatom class specific to the ionization method. Homologous series were separated and grouped by nominal Kendrick mass and Kendrick mass defect to facilitate rapid identification. 61

Results and Discussion

Figure 7.1 shows broadband positive-ion APPI LTQ mass spectra of the maltene (top) and asphaltene (bottom) fractions of the vacuum bottom residue (593 + ˚C) from a Middle Eastern heavy crude. The maltene fraction has a molecular weight distribution between approximately 175

139 spectral results agree with the literature and maltenes and asphaltenes share similar carbon number space.

Figure 7.2 shows a mass-isolated segment from an asphaltene fraction. A 1 Da window (top) shows the immense complexity of an asphaltene fraction, with more than 140 unique mass spectral peaks per nominal mass above six times the baseline rms noise level. A typical

140 whole crude oil contains approximately 50-70 peaks per nominal mass. The increased complexity complicates analysis by FT-ICR MS for asphaltene fractions.

A mass spectral zoom inset of a positive-ion APPI FT-ICR mass spectrum of an asphaltene from Middle Eastern heavy crude is shown in figure 7.3. Mass doublets are shown to illustrate the need for sufficient

141 resolving power is critical for correctly assigning elemental formulas.

Isobaric species that differ in elemental composition by SH4 vs C3 both have a nominal mass of 36 Da are observed in ESI and APPI; isobars

13 32 differing by CH3 S vs C4 both with a nominal mass of 48 Da are observed in APPI. In order to speciate chemical classes in a crude oil, sufficient resolving power must be achieved to separate the signal produced from ions of very similar masses. FT-ICR MS is able to routinely produce a resolving power m/m∆50% = 400,000 capable of identifying species not observed with other mass spectrometry techniques. 38, 42

142 vs C4 are separated and identified due to the high resolving power afforded by FT-ICR MS. Other MS techniques are not able to routinely achieve resolving power to identify isobars in a broadband mass spectrum.

Since the repeating mass unit for a hydrocarbon family is CH2, it becomes possible to recognize a series of compounds differing only in the 61, 62 degree of alkylation, or 14.01565 Da, the mass of a CH2 unit. Each nuclide (atom or nucleus with a specific number of protons and neutrons) has a different mass defect, the exact mass minus the nearest- whole integer mass, therefore each ion of different elemental composition has a different mass. 61, 202 For example, the mass of hydrogen is 1.007825 Da. Therefore, the addition of each hydrogen atom to a molecule shifts the exact mass 0.007825 Da higher. In mass spectral compositional space, the spectral peak shifts to the right, towards the next highest integer mass. Compounds with a higher H/C ratio are more saturated and have higher mass defects, with exact masses which are closer to the next highest integer mass. Conversely, aromatic compounds have lower H/C ratios and therefore lower mass defects and are closer to the lower integer mass. Hughey et al first identified the difference between “coarse” spacings in petroleum mixtures, namely the 1 Da

12 12 13 spacings between Cc vs Cc-1 C1 elemental compositions of the same molecule and “fine structure” which is different mass defects for different elemental compositions. 61 Figure 7.4 shows a 1 Da window at m/z 553 for positive-ion APPI

FT-ICR mass spectra for a whole crude (top) and its asphaltene (bottom).

The asphaltene sample contains compounds with a smaller mass defect indicating more aromatic species which is in agreement with proposed asphaltene structure being condensed aromatic rings with a small degree of alkyl substitution with an H/C ratio ~1.0-1.1. 19

143 Maltene fractions have a greater degree of alkylation, contain more aliphatic hydrogen and more saturated carbon and have higher H/C ratios (1.25-2.0) resulting in a shift to higher mass defect. The two most abundant peaks in the whole crude spectrum (top) are DBE 11 and 4. The asphaltene fraction is shifted to lower mass defect and contains a high abundance of compounds with DBE of 19 and 25, a difference of several aromatic rings.

Figure 7.4 Mass scale-expanded segment of a single nominal mass unit at m/z 553 for a maltene (top) and asphaltene (bottom) fractions from a Middle Eastern heavy crude vacuum residue. The mass defect, the difference between the exact mass and nominal mass, differs in spectral position in respect to the composition of the two fractions. Maltenes are more enriched in hydrogen and therefore have a higher mass defect than asphaltenes, which are composed mainly of condensed aromatic rings with little or no alkyl substitution.

144 DBE vs. Carbon Number Images. Since asphaltenes exhibit different solution-phase behavior, they must differ in composition. Asphaltenes are a solubility-defined fraction of crude oil that is insoluble in the normal parafinnic solvents, such as n-heptane and n-pentane, but soluble in more aromatic solvents, such as toluene. A host of analytical techniques agrees upon the molecular weight of asphaltenes being approximately between 500-1500 Da; therefore maltenes and asphaltenes share similar carbon number space. More specifically, asphaltenes are not abnormally high molecular weight species (<2000 Da) but are composed of relatively small molecules. However, since asphaltenes and maltenes share the same carbon number space, they must differ in aromaticity since they differ in solubility. Figure 7.5 shows a color-coded isoabundance-contoured plot of DBE vs. carbon number for the parent vacuum bottom residue (left), its asphaltene fraction (center) and maltene fraction (right) for the S2 and S3 classes. The parent residue and the maltene fraction share the exact same carbon number range (between 32-80) for the S2 class with a DBE distribution of ~3-34. The maltene fraction and parent residue are identical in molecular weight and aromaticity for a given heteroatom class indicating that the majority of the species observed from a parent crude are maltenes. However, upon fractionation, asphaltenes are removed from the crude oil matrix and can be characterized compositionally without interference from the highly abundant, more efficiently ionized maltene compounds. The asphaltene fraction for the S2 covers a carbon number range between ~28-60 with DBE values between 18-35. Both fractions have virtually the same carbon number range and therefore the same molecular weight. However, the asphaltene fraction is shifted to higher DBE indicating a predominance of more aromatic species relative to maltenes within the same carbon number and heteroatom class. A similar trend is observed for the S3 class. For example, maltene compounds for the S2 class with a DBE of 11 corresponds to 1-

145 benzothieno[3,2-b]1-benzothiophene molecule, a compound known to 8 exist in crude oil. The molecular formula of C14H8S2 contains nearly 35 less carbons per structure than the most abundant species found in the parent and maltene fractions with a DBE value equal to 8 with a carbon number of 50. Alkyl substitution off of the core structure accounts for the increased carbon number in the maltene fraction. However, asphaltene compounds that contain two sulfur atoms and contain 50 carbon atoms have a much higher DBE values between 25-30.

146 Degradation of alkyl chains or dehydrogenation reactions result in condensed ring structures that contain little to no alkyl-substitution in the asphaltene relative to their maltene counterparts. However, the shift upward in aromaticity for a given class is not without limits; it is defined by the highest possible aromaticity for a planar aromatic structure. Beyond this aromaticity, the compound is no longer a planar structure.

Figure 7.6 shows normalized composite color-coded isoabundance plots of DBE vs carbon number for the S1 and S2 classes for maltene and asphaltene fractions of Middle Eastern heavy crude. In order to show two different spectral results on the same relative abundance scale, each fraction was normalized to one another. Here, the purpose is to show the different regions of compositional space between maltenes and asphaltenes not relative abundances.

147

When plotted in a single composite figure, it is clear that asphaltenes are more aromatic than their maltene homologues within the same carbon number. FT-ICR MS results agree with a variety of analytical techniques and observations. First, asphaltenes are more aromatic than maltenes. Their solution-phase behavior supports this observation since asphaltenes are defined by their insolubility in the normal alkanes such as n-heptane. Like dissolves like; more aromatic

148 solvents such as toluene and benzene are excellent solvents for asphaltenes. Maltenes are soluble in paraffinic solvents since they are less aromatic than asphaltenes. Second, maltenes have a higher degree of alkyl-substitution off of core structures sterically hindering interactions between PAH cores. Asphaltenes, on the other hand, have little to no alkyl-substitution allowing the PAH cores to interact through non-covalent interactions such as hydrogen bonding and π- π stacking. One of the fundamental problems with asphaltenes is their ability to self- associate and form highly-stable aggregates that flocculate and precipitate forming deposits in the reservoir, pipelines and refineries. The mechanism by which asphaltene monomers aggregate is highly studied but not fully understood. Therefore, asphaltenes have to be compositional separated from maltenes since maltenes do not typically self-associate. Finally, bulk property measurements of a typical asphaltene produce elemental composition by mass percent of C 81.07, H 7.11, N 1.02, O 1.6, S 8.94 and yield an H/C ratio of 1.045. 19 Maltenes have higher H/C ratios, usually between 1.25-2. FT-ICR MS agree with elemental analysis for both asphaltenes and maltenes, since there are two distinct regions in DBE space for each fraction.

Conclusions

The molecular weight of asphaltenes has widely been determined in recent years through a consensus of analytical techniques. Fundamentally, asphaltenes are defined by their solubility, or rather, their lack of solubility in paraffinic solvents. Because asphaltenes and maltenes have similar molecular weights, and therefore cover similar carbon number ranges, they must be somehow separated in structure and composition than their maltene homologues. Although asphaltenes are known to have an increase in heteroatoms, the amount of heteroatoms per structure is not distinct enough to cause solution-phase

149 behavioral differences alone. Structurally, asphaltenes must be distinctly different than maltenes. Compositional analysis by ultrahigh resolution FT-ICR MS defines maltene and asphaltene compositional space. Several conclusions are reached in this study. First, asphaltenes are not abnormally high molecular weight compounds and are in agreement with the Boduszysnki continuity model (Chapters 4 and 5) which stated that 95% of the compounds found in crude oil are below 2 kDa. 13, 22 Second, asphaltenes are maltenes are defined by similar carbon number range. Third, asphaltenes exist in DBE space above maltenes but are below the planar limit for polyaromatic ring systems in the condensed phase. Finally and importantly, FT-ICR MS results agree with and support bulk property measurements, 13C/1H NMR, dispersion techniques such as time-resolved fluorescence depolarization, Taylor dispersion and fluorescence correlation spectroscopy and other mass spectral techniques such as APCI, ESI, FI, FD and LD-MS. Here, we present a unified theory of the definition of asphaltene and maltene composition that is supported by numerous research groups and techniques. FT-ICR MS results agree with the Boduszynski continuity concept and serve as the basis for a comprehensive definition of asphaltene structure and composition.

150 CHAPTER 8. THE TRUE MOLECULAR CHARACTERIZATION OF ASPHALTENES PART III. SOLUTION-PHASE AND GAS-PHASE AGGREGATION OF ASPHALTENES

Summary

Most asphaltene molecules self-associate to form aggregates at very low concentrations. 29, 30, 168, 171 Mass spectral analysis is performed at concentrations where aggregation occurs; therefore, the majority of the asphaltene molecules are in aggregate structures of approximately 8 monomers and are therefore at much high m/z values. The observed signal is produced by the low abundance of asphaltene compounds that are nonaggregated. One of the fundamental problems with asphaltene mass spectral characterization in recent years is their inability to ionize efficiently and consistently. One of the proposed reasons is that highly conjugated, polyaromatic ring systems such as asphaltene monomers, have low ionization efficiencies. It has also been postulated that a fraction of asphaltene molecules are simply not able to be ionized, thereby limiting the ability to produce enough signal for mass spectral analysis. Here, we provide direct evidence that shows that asphaltenes are aggregated at concentrations routinely used for mass spectral analysis. Low signal that is observed for asphaltene monomers is affected by the degree of aggregation, which pushes the m/z value of the monomer approximately eight times higher to the m/z value of the aggregate. Introduction

A previous study explored the use of silver cationization (Ag+) mass spectrometry to determine the molecular weight distribution of nonboiling petroleum fractions. 203 Rousiss and Proulx ionized asphaltene

151 molecules and examined them with ToF-MS and observed a bimodal distribution for asphaltenes, one below 1000 Da and a broad, wide distribution from 5000 – 20,000 Da. The authors attributed the high molecular weight component to prove the existence of high molecular weight asphaltene compounds, and the low m/z portion to fragmentation. However, the high m/z molecular weight distribution is 7- 10 times the molecular weight distribution at low m/z in direct correltation with fluorescence depolarization techniques which indicate that asphaltene aggregates composed of eight monomers are the most stable. 29, 30, 168, 171 The results, however misinterpreted, offered great insight into ionization mechanisms possible for asphaltenes. Previous work from our group has optimized solvent systems to explore Ag+ cationization for preferential ionization of sulfur compounds in crude oil. 204 Controversy exists over whether or not asphaltenes are able to be ionized, due to the constant poor ionization efficiency they exhibit over a wide range of ionization techniques. However, the molecular weight of asphaltene monomers (> 2 kDa.) is well below ionization thresholds by molecular weight. However, the molecular weight of the aggregate structure, between 10-30 kDa further complicates ionization of a singly- charged species. Under current review is why asphaltenes, the most polar fraction of crude oil, do not show multiple ionization sights and form multiply charged species. A likely reason is that the polar functional groups are turned towards one another in the center of the aggregate eliminating functional sights for protonation/deprotonation by routine electrospray. APPI has always been challenging for asphaltenes as well, since the inherent thermal parameter results in asphaltene deposition similar to problems that occur in pipelines. Here, we apply Ag+ cationization to validate the molecular weight distribution of asphaltene monomers and aggregates and examine changes that occur in the monomer/aggregate distribution as a function

152 of concentration. Since asphaltene aggregation is thought to occur above ppb level (~5 g/mL), concentrations of mass spectral samples are typically a factor of five greater than aggregation onset. If the concentration can be decreased to a level below aggregation threshold, aggregates will presumably dissociate leaving only asphaltene monomers for characterization, the ultimate goal of this entire dissertation.

Experimental Methods

Sample preparation. Middle Eastern heavy crude oil was supplied by General Electric Global Research (Niskayuna, NY). Distillation was performed in a still pot and produced seven fraction : IBP-191, 191-315, 315-371, 371-510, 510-538, 538-593 and 593+ ˚C. The residue (593 + ˚C) was fractionated according to the saturates- aromatics-resins-asphaltenes (SARA) method. 36, 37 Briefly, 500 mL of n- heptane was added to ~10 g of the residue sample, refluxed for 1 hour in a 1 L round-bottom flask and stored in the dark (12 h). The solids (asphaltenes) were isolated by gravity filtration through Whatman (Kent, UK) 2V grade filter paper. Hot heptane was added to the asphaltene residue to complete the transfer of solids. The filter paper with the asphaltenes was then refluxed with heptane at a rate of 3-5 solvent drops/minute for 60 min until all asphaltenes were completely desorbed from the filter paper. 199 The asphaltene sample was then rotary vacuum- evaporated to dryness, weighed and redissolved in toluene to produce a stock solution of 10 mg/mL. A stock solution was prepared by dissolving ~20 mg of asphaltene sample in 20 mL of toluene. A one mL aliquot was diluted with 1 mL of methanol that contains 2% by volume formic acid or

NH4OH for positive- or negative-ion mode electrospray analysis. Samples were further diluted in toluene to yield final concentrations for APPI analysis without additional modification.

153 Silver complexation Silver complexation was achieved by mixing 3 parts of silvertrifluoromethyl sulfate (silver triflate) and one part crude oil. A solution of crude oil was prepared in 500 g/mL concentration in 1:1 (v/v) solution of toluene/methanol. Silvert triflate, roughly three times by weight was added to the crude and vortexed immediately prior to injection into the ESI source.

LTQ-MS. Positive-ion APPI mass spectra [broadband and collision- activated dissociation (CAD) MSn] were acquired with an LTQ mass spectrometer (Thermo Electron Corp., San Jose, CA). Optimum collision energy for CAD fragmentation varies linearly with m/z. Therefore, in the LTQ mass spectrometer, the collision energy is normalized for each m/z value selected for dissociation [normalized collision energy (NCE)]. APPI conditions were analogous to those described below. 205 9.4-T FT-ICR MS. A custom-built FT-ICR mass spectrometer is equipped with a 22 cm horizontal room temperature bore 9.4 Tesla magnet (Oxford Corp., Oxney Mead, UK) and a modular ICR data station (PREDATOR).50, 52 Positive ions generated at atmospheric pressure in the external APPI source enter the skimmer region at ~2 Torr through a heated metal capillary into the first rf-only octopole. Ions pass through a quadrupole to a second octopole where they accumulate for 250-1000 ms. Helium gas was introduced during accumulation to collisionally cool the ions before transfer through a 200 cm rf-only octopole into an open cylindrical Penning ion trap (10 cm i.d. x 30 cm long). Octopole ion guides were operated at 2.0 MHz and 240 Vp-p rf amplitude. Broadband frequency chirp excitation (70 – 700 kHz at a sweep rate of 50 Hz/s and amplitude, Vp-p = 350 V) accelerated the ions to a cyclotron orbital radius that was subsequently detected by the differential current induced between two opposed electrodes of the ICR cell. The experimental event sequence was controlled by a MIDAS (modular ICR data acquisition and analysis software) data station.51, 52 Multiple (100-300) time-domain

154 acquisitions were summed for each sample, Hanning-apodized, and zero- filled once prior to fast Fourier transform and magnitude calculation.200

Results and Discussion

Figure 8.1 shows broadband positive-ion APPI FT-ICR mass spectra of Middle Eastern heavy crude and its residue and the asphaltene and maltene fraction derived from the residue normalized to the highest peak in all four spectra. Whole heavy crude has a molecular weight distribution 300 < m/z < 1000 centered at m/z 550; the residue (593+ ˚C) contains the highest-boiling components of crude oil and therefore is shifted to higher molecular weight 450 < m/z < 1200 centered at m/z 825. As pointed out in chapter 3 and 4, crude oil composition is a continuum in molecular weight, composition and boiling point. The highest boiling compounds in a whole crude oil concentrate in

155 the residue fraction and gradually shift the molecular weight distribution of the residue higher relative to the whole crude.

156 Figure 8.1. Broadband positive-ion APPI FT-ICR mass spectra for Middle Eastern heavy crude, distillate residue (593+ ˚C) and the asphaltene and maltene fractions derived from the residue. Each spectrum is normalized to the highest peak in all four spectra. The maltene fraction covers the exact same molecular weight distribution as the parent residue with comparable signal. However, the asphaltene fraction exhibits much lower signal with a narrow molecular weight distribution.

Two major observations can be inferred from comparison of the maltene and asphaltene spectra, First, the maltene fraction derived from the residue covers a similar molecular weight distribution as the parent residue, but the asphaltene fraction has a lower and more truncated molecular weight distribution (450 < m/z < 800). Since asphaltenes are generally regarded as the highest boiling, heaviest compounds in crude oil, one would expect the molecular weight distribution to be higher than the maltene fraction. In chapter 8, we introduced the fraction of crude oil referred to as distillable asphaltenes, low molecular weight, highly aromatic compounds in crude oil that are asphaltenes by definition (n- heptane insoluble) but do not self-associate to form aggregates. Here, the asphaltene fraction is derived from the nondistillable material (593+ ˚C) and therefore distillable asphaltenes, by definition, are removed. Second, the signal magnitude for the asphaltene fraction is significantly less than the rest of the spectra, a factor of 3 less than the maltene fraction. As we already have pointed out, most asphaltene molecules self-associate to form aggregates at very low concentrations (~5 ug/mL). Therefore, most of the asphaltene monomers are aggregated and therefore are not efficiently trapped by FT-ICR mass spectrometry similar to the rest of the components in crude oil (<2 kDa). 50:50 asphaltene/maltene mixture. To determine whether or not asphaltenes were aggregated in the solution-phase, a mixture of 50/50 (by weight) asphaltene and maltene was analyzed. Each fraction was dissolved in toluene (250 g/mL) and mixed in a 1:1 by volume ratio to produce a final solution of 500 g/mL for analysis. Fundamentally, a solution that is 50% by weight asphaltene (in toluene) should produce

157 signal from asphaltene species that can be differentiated from maltenes by their mass defect (see figure 8.6). Since asphaltenes are highly condensed, polycyclic systems with little or no alkylation and maltenes have a high degree of alkylation, species from either fraction will occupy different regions of a nominal mass unit in m/z space. Figure 8.2 shows the broadband positive-ion APPI mass spectra for the maltene fraction (top) and a 50% (by weight) mixture of asphaltene and maltene (bottom) normalized to the highest peak in both spectra. The pure maltene sample had nearly double the signal intensity (y-axis) relative to the asphaltene:maltene mixture. Both spectra have a similar molecular weight distribution between 250 < m/z 950 centered at approximately m/z 550 and therefore share compounds of similar molecular weight. Interestingly, the asphaltene/maltene mixture has a factor of two less signal magnitude than the purely maltene species. Our theory is that asphaltenes are in solution at concentrations where they are already aggregated, are ionized in solution and therefore competing with maltene species and lowering the overall signal of the mixture. Since FT-ICR parameters are optimized to trap, excite and detect species below 2 kDa, these asphaltene aggregates are not detected, since they exist at the molecular weight of the aggregate (approximately 6-10 monomers per aggregate). Experiments have been conducted to optimize FT-ICR MS for m/z values of the aggregates; however, eventhough m/z values (2.6-4.0

kDa) have been reported in the past on simple samples such as C60 clusters, the increased complexity inherent to asphaltenes makes this virtually impossible by FT-ICR MS. 206

158

Heteroatom class distribution. Figure 8.3 shows the DBE distribution for the maltene fraction and the 50:50 asphaltene/maltene mixture. The S1 class is the most abundant in both the maltene and the mixture, followed by S2 and hydrocarbon. A sample that has 50% by weight asphaltene content, far exceeding typical asphaltene contents for example, bitumen (~15%), shows little change in the heteroatom composition relative to the maltene fraction.

159

DBE vs. Carbon Number. Figure 9.4 shows the color-coded isoabundance contoured plots for DBE vs carbon number for members of the hydrocarbon class. The purely maltene fraction contains hydrocarbon species with DBE = 2-28 with ~27-64 carbons. The asphaltene/maltene mixture covers a slightly broader carbon number distribution between 20-70 carbons with DBE values = 3-25. However, the most abundant hydrocarbons for both samples had DBE values = 10, consistent with phenanthrene, C14H10. Alkyl chains account for increased carbon number, therefore indicating a high degree of alkyl substitution for both samples, with the most abundant species having over 40

160 carbons. Asphaltenes are highly condensed polycyclic ring systems with low H/C ratios (1-1.1), and the compounds observed in the mixture are purely maltenic, with much higher H/C ratios, indicating that only maltenes are being detected from the mixture.

Figure 8.5 shows plots of the S1 class DBE vs. carbon number. Both samples cover a similar DBE range (0-28) with the most abundant species having DBE values of 6 and 9, consistent with benzothiophenic and dibenzothiophenic rings. As discussed in figure 8.4, the mixture appears to contain only maltene species with much lower DBE values for a given carbon number than expected for asphaltene species.

161

If asphaltenes are not detectable at high abundance in a mixture that is 50% asphaltenic, then asphaltenes are either not making it into the gas phase for ionization, or they are simply not being detected. Since asphaltenes are known to self-associate and form aggregates of approximately 6-10 monomer units, the m/z value they would be detected at would equal that of the aggregate, not the monomer. For this reason, the molecular weight of asphaltenes needs to be readdressed. Preliminary results obtained on a LTQ-MS (figure 8.6) indicate that the molecular weight of asphaltene monomers is less than 2 kDa, which is in

162 agreement with the Boduszynski model and previous results from this group.

The molecular weight distribution shown in figure 8.6 begins at m/z 200 and is centered at m/z 1200 and tails at approximately m/z 3500 which is attributed to asphaltene monomers. A sharp increase in ion abundance begins at m/z 3500 and continues on past the upper molecular weight limit of the linear trap (4000 Da). Because we are

163 unable to characterize species over 4000 Da by LTQ-MS, the nature of the observed high molecular weight species is uncertain. Asphaltene monomers have a molecular weight distribution centered at m/z 1200, Asphaltenes are thought to aggregate at concentrations of 60 g/mL in toluene, much lower than typical concentration (100-500 g/mL) used for mass spectral analysis. 168, 207 It is believed that asphaltene nanoaggregate structures consist of approximately 8 asphaltene molecules. 19 The molecular weight distribution ranges from 200 < m/z < 3000, therefore aggregate structures would range from 1600 < m/z < 24,000. Characterization of singly-charged compounds above m/z 4000 by LTQ and FT-ICR MS is difficult for even simple mixtures but the increased complexity of asphaltenes proves nearly impossible. To determine if the increase in signal magnitude between 3500 < m/z 4000 is indeed the onset of asphaltene aggregation, time-of-flight mass spectrometry (TOF-MS) was utilized. Ions are accelerated by an electrical field to the same kinetic energy independent of charge. The velocity of the ion depends on the mass-to-charge (m/z) ratio and the time it takes each ion to reach the detector can be measured.202 Therefore, a wide range of m/z values can be measured in a single mass spectrum.

164

Figure 8.7 shows a TOF-MS spectrum of the maltene fraction of Middle Eastern heavy crude at two different focusing voltages. Since maltenes have increased H/C ratios, steric hindrance from alkyl chains prevents noncovalent interactions between polycyclic cores preventing aggregation. Here, we show that the molecular weight distribution of the maltene fraction obtained by TOF-MS agrees with FT-ICR results which show molecular weights well below 2 kDa.2 Maltene molecular weight distribution ranges from 200 < m/z < 1000 centered at approximately m/z 700 collected by TOF-MS, in agreement with LTQ-MS and FT-ICR

165 MS spectra (data not shown). If maltenes also aggregated such as their asphaltene counterparts, their molecular weight distribution would be centered at m/z 5600. A mass-scale expanded zoom inset shows the lack of signal magnitude of any significance above 2 kDa indicating that the maltene fraction does not form aggregates. Two different focusing voltages, 100V and 120V, resulted in the same molecular weight distribution. Figure 8.8 shows the mass spectrum obtained of asphaltenes by silver cationization electrospray TOF-MS at a concentration of 500 g/mL. The molecular weight distribution is bimodal, with two apparent distributions distinctly separated from each other. First, the low molecular weight distribution ranges between 200 < m/z < 2000 and is centered at m/z 1200. As discussed previously, silver forms clusters with itself through noncovalent interactions which are responsible for the highly abundant peaks observed in the low m/z region. The low m/z distribution is consistent with other previously reported results for the molecular weight of asphaltene monomers. Self-associated into aggregates at this concentration, a distribution is observed separate from the monomer between 3000 < m/z < 24,000 attributed to aggregates. The distribution of the aggregate is broad (8 kDa) and of lower signal than the monomer. However, centered at m/z 10,000, the distribution is in agreement with proposed 6-10 asphaltene monomer per aggregate theory. 19, 25, 166, 171 Since both distributions have a well-defined start and end point, it is a clear indication of asphaltene aggregation.

166

A broad, nondescript molecular weight distribution is shown in figure 8.8 at a focusing voltage of 100 V. Figure 8.9 and figure 8.10 show the same distribution but at increased focusing voltage of 120 V and 160 V. At higher focusing voltages, more energy is being put into the ions. Figure 8.8 shows the distribution at 120 V focus with the aggregate distribution between 2500 < m/z < 25,000 centered at m/z 9500. An increase of 40 V (figure 8.9) produces the same distribution. However, the ratio of signal magnitude of monomer to aggregate changes as a function of focus voltage, with higher molecular weight ions being more

167 efficiently focused through the mass spectrometer inlet at higher voltages. Also, at higher focusing voltages, the increased internal energy of the ions has been known to disrupt noncovalent interactions, such as multimer formation. At 160 V focusing potential, the monomer distribution is bimodal, indicating a distinct distribution for asphaltene monomers and dimers.

168

Since asphaltene aggregation is a function of concentration in toluene or in crude oil, lowering the concentration should result in a shift to lower m/z for the aggregate. Figure 8.10 shows the TOF-MS mass spectrum collected at 50 g/mL at 80 V focus voltage. Here, the molecular weight distribution of the monomer remains unchanged between 200 < m/z < 2000 centered at m/z 850 comparable to the higher concentration of 500 g/mL shown in figures 8.7, 8.8 and 8.10. However, the aggregate distribution shifts lower between 3000 < m/z <

169 20,000 at lower concentration and is centered at m/z 6800, exactly eight times the apparent molecular weight of the monomer. Since the ionization technique used in ESI with silver cationization, an internal standard compound can be used to indicate if fragmentation occurs. Since silver noncovalently binds to itself, these complexes should fragment at much lower energies than covalent linkages in molecular bonds. The observation of silver cluster peaks at low m/z is indicative of the lack of fragmentation, since these interactions would be disrupted if ionization parameters were inducing fragmentation.

170

Figure 8.12 shows that at higher focusing voltage (100 V), the aggregate distribution shifts to lower m/z, as was shown in figures 8.8- 8.10. Similarly, the monomer distribution remains centered at m/z 850 Da, but the aggregate distribution center shifts to m/z 4200, which is consistent with seven asphaltene monomers per aggregate. An apparent bimodal distribution in the aggregate appears at m/z 4200 consistent with approximately five monomers per aggregate. At increasing focusing voltage, molecules have more thermal energy, thereby resulting in the disruption of noncovalent interactions, such as π- π stacking and hydrogen bonding.

Figure 8.12. TOF-MS mass spectrum of asphaltene fraction derived from Middle Eastern heavy crude at 100V focus voltage. The monomer distribution is

171 centered at m/z 850 but the aggregate distribution shifts to lower m/z and shows a slightly bimodal distribution, indicating the presence of two stable core aggregates containing five and seven asphaltene monomers.

An increase in focus voltage to 150 V (figure 8.13) shifts the aggregate molecular weight to a monomodal distribution centered at m/z 4200. As observed in figure 8.12, aggregates with a molecular weight of 4200 Da are consistent with pentamer structures consisting of five asphaltene monomers per aggregate.

172 Conclusions

Here, we present data that illustrates the fundamental problem associated with FT-ICR mass spectral analysis of asphaltenes : they are aggregated in solution and therefore not readily detectable. We present evidence of the solution-phase aggregation tendencies of asphaltenes by comparing a mixture that is 50% by weight asphaltene with a maltene fraction. All observed species were maltenic in structure, composition and aromaticity, indicating the aggregation of asphaltenes in solution. Next, low resolution LTQ-MS was used to define the molecular weight distribution of asphaltenes at low m/z and a possible second distribution exactly equal to 6-10 times the molecular weight of the monomer was observed. Since the linear trap is limited in the high m/z to less than 4 kDa, a time-of –flight mass spectrometer was employed to characterize the onset and conclusion of asphaltene aggregation by molecular weight. Two concentrations were analyzed, both above the published critical nanoaggregation concentration (CNAC). We are able to observe the changes in the molecular weight of the aggregate clusters as a function of concentration and of focusing voltage. At lower concentration (50 g/mL), asphaltene aggregation occurs at a much lower m/z value that corresponds to two stable core structures of five and seven monomers per aggregate. Increasing the focus voltage results in disruption of noncovalent interactions that bind asphaltenes to each other and consequently, higher focusing voltage results in fewer asphaltene monomers per aggregate. Here, we have outlined a major problem in the complete characterization of asphaltenes : they are aggregated. Although asphaltene self-associative tendencies has been well-researched in the past, the ability to characterize the monomer at required concentrations for mass spectrometry has been problematic. Now that we understand the inherent problem, future work will focus on methods to disrupt asphaltene aggregation to completely characterize asphaltene monomers.

173 CHAPTER 9. IDENTIFICATION OF VANADYL PORPHYRINS IN A HEAVY CRUDE OIL AND RAW ASPHALTENE BY ATMOSPHERIC PRESSURE PHOTOIONIZATION FT-ICR MASS SPECTROMETRY

Summary

Vanadyl porphyrins are detected and characterized by their double bond equivalents (DBE = number of rings plus double bonds) and carbon number in an unfractionated (raw) asphaltene and unaltered South American crude oil. Atmospheric Pressure Photoionization (APPI) Fourier Transform Ion Cyclotron Resonance Mass Spectrometry (FT-ICR MS) provides the high mass resolving power (450,000 – 650,000 at m/z 500) and accurate mass (<300 ppb) to unambiguously assign elemental compositions to each of more than ten thousand peaks in each mass spectrum. Kendrick mass sorting revealed unusually high mass errors

for peaks assigned to high DBE O2 species as well as a suspicious bimodal distribution in plots of DBE vs. carbon number for all O2 species. Inclusion of vanadium in chemical formula assignment resolved the bimodal distribution into lower DBE O2 species and vanadyl porphyrins with a subsequent decrease in mass assignment errors to the same level as those for the thousands of other identifieded species.

+ + Vanadyl porphyrins are detected both as M and [M+H] molecular and quasimolecular ions. Trends in the relative abundance of specific DBE values reveal the structural diversity of the vanadyl porphyrins in the asphaltene and heavy crude oil. To our knowledge, the current results are the first to directly identify and catalog the structural diversity of vanadyl porphyrins directly in raw (unfractionated) asphaltene and unaltered heavy crude oil.

Introduction

174 The depletion in the world’s supply of light, "sweet" crude has resulted in a demand for increased compositional information for medium and heavy crude oils. Heavy crude oils, as their name accurately reflects, contain a higher fraction of heavy, high-boiling species that are enriched in heteroatoms (N, O, and S) and metals. Metals such as Ni and V are known to exist in porphyrin structures, commonly called petroporphyrins, and accumulate in the higher-boiling fractions. Due to the global demand for refined transportation fuels, conversion of the heavy, high-boiling (problematic) fractions of medium and heavy crude oils to lower-boiling (useful) fractions is highly desirable. However, before crude oil can be converted to fuel, metal- containing components must be removed. Distillation concentrates metals in the residue and the subsequent asphaltene fraction is enriched in vanadium and nickel relative to the whole crude.3, 208 Characterization of vanadium (V) and nickel (Ni) complexes is important to the development of demetallation and catalytic strategies used to process heavy crudes.209 Porphyrins are problematic for refineries because they affect upgrading and conversion processes.3 Even at low concentration (<1%), vanadium alters catalyst selectivity and blocks active sites on catalysts used in cracking, increases coke formation, reduces gasoline yields3, 210 and forms sodium vanadates implicated in corrosion of metal surfaces.211 Deposits of vanadium and nickel formed on catalysts can cause bed plugging with heavier feedstocks and nickel porphyrins in the asphaltene fraction are believed to stabilize water-in-oil emulsions.212 Petroporphyrins were the first compounds to link crude oil to its biological origin by Triebs more than sixty years ago.213, 214 Metalloporphyrins can also serve as indicators of petroleum maturation because heavy, young oils contain a higher amount of vanadyl and nickel porphyrins than more mature, light crudes.3, 215 The concentration ratio of the two main petroporphyrins, deoxophylloerythroetioporphryin (DPEP)

175 and etioporphyrin (etio) that form vanadyl (VO) and Ni complexes is also an oil maturity indicator. A higher amount of the DPEP (or cycloalkanoporphyrins, CAP) indicates a more mature crude oil and the ratio of nickel to vanadyl porphyrins decreases as a crude matures.216, 217 Figure 9.1 shows the structures of proposed petroporphyrin classes found in crude oil: e.g., di-DPEP, rhodo-DPEP, and rhodo-etio formed by dehydrogenation of alkyl chains of DPEP and etio core structures, each consisting of a homologous alkyl series.210, 218-222 The structures of petroporphyrins have been characterized by X-ray diffraction223, NMR,224 and electron spin resonance.225, 226

176 the number of rings plus double bonds to carbon (DBE = c - h/2 + n/2 +1 for elemental composition, CcHhNnOoSs).

Characterization of petroporphyrins is usually preceded by extensive isolation and/or purification by methods such as oxidation,220 separation into acid, base and neutral fractions,210 solubility separations,227, 228 SARA fractionation, Soxhlet extraction,225 and vacuum sublimation,229 as well as a chromatographic technique such as HPLC,220, 222 reversed-phase LC,230 size exclusion chromatography,231, 232 or thin- layer chromatography.210, 233 The characteristic absorption at 408 nm in the UV-vis spectrum is often used to detect the presence of porphyrins in crude oil.215 Mass spectrometry of petroporphyrins provides molecular weights and empirical formulas.216 Electron impact ionization,234 chemical ionization,235 supercritical fluid chromatography/MS,127 GC/MS,236 and LC/MS, 237 as well as tandem MS (MS/MS)238 have been used to characterize petroporphyrins. Because petroporphyrins concentrate in more complex, high-boiling heavy crude oils, high resolution is necessary to distinguish chemically different components. Techniques such as continuum source graphite furnace atomic absorption spectrometry,211 high-resolution, low energy, electron ionization mass spectrometry, 239 magnetic sector mass spectrometry, 226 and time of flight mass spectrometry 240 have been used to characterize petroporphyrins at the molecular level. Electrospray ionization was first used to study geoporphyrins by Van Berkel et al., who first isolated vanadium and nickel porphyrins before analysis by mass spectrometry.241 Rodgers et al. first used ultrahigh resolution Fourier transform ion cyclotron resonance mass spectrometry (FT-ICR MS) to characterize nickel and vanadium porphyrins isolated from heavy crude. 61 Atmospheric pressure photoionization was first coupled to FT-ICR MS for characterization of nonpolar (and polar) species present in crude oil by Purcell et al. 91

177 Recently, Qian et al. coupled atmospheric pressure photoionization (APPI) to FT-ICR MS and observed vanadyl porphyrins in an asphaltene sample from a vacuum resid that had been subjected to solubility fractionation. 242 Here, for the first time, we detect and identify vanadyl porphyrins in a South American heavy whole crude oil and an Athabasca bitumen asphaltene without prior fractionation or sample treatment by APPI FT-ICR MS.

Experimental Methods

Sample preparation. South American heavy crude oil (~20 mg) that had been previously analyzed63 was diluted with 5 mL of toluene (HPLC Grade, Sigma-Aldrich Chemical Co., St. Louis, MO) to make a stock solution. The stock solution was further diluted another 8-fold in toluene to yield a final concentration of 500 g/mL for mass analysis. Athabasca bitumen was supplied by the National Center for Upgrading Technology (Alberta, Canada) and fractionated according to the saturates-aromatics-resins-asphaltenes (SARA) method.243 Briefly, 500 mL of n-heptane was added to the bitumen sample (10 g), refluxed for 1 hour in a 1 L round-bottom flask and stored in the dark (12 h). The solids (asphaltenes) were isolated by gravity filtration through Whatman (Kent, UK) 2V grade filter paper. Hot heptane was added to the asphaltene residue to complete the transfer of solids. The filter paper with the asphaltenes was then refluxed with heptane at a rate of 3-5 solvent drops/minute for 60 min until all asphaltenes were completely desorbed from the filter paper.199 The asphaltene sample was then rotary vacuum-evaporated to dryness, weighed and redissolved in toluene to produce a stock solution of 10 mg/mL. The stock solution was further diluted to 500 g/mL in toluene prior to analysis.

178 Atmospheric Pressure Photoionization (APPI). A custom-built adapter interfaced the APPI source (ThermoFisher Scientific, San Jose, CA) to the front stage of a custom-built 9.4 T FT-ICR mass spectrometer (see below).91 The sample flows through a fused silica capillary at a rate of 50 L/min and is mixed with nebulization gas (N2 at 50 kPa) inside a heated vaporizer operated at 250 °C for whole crude and 275 °C for asphaltene according to previous nebulization temperature optimization based on crude oil boiling point ranges.154 The nebulized sample flows from the heated vaporizer as a confined jet and passes under a krypton vacuum ultraviolet lamp that produces 10 eV photons (120 nM). In dopant-assisted APPI, first introduced by Robb et al., toluene was selected as a solvent/dopant to increase analyte ionization.99 Charge exchange and proton transfer reactions occur between ionized toluene and neutral analytes via collisions in the ionization region at atmospheric pressure.91, 99 9.4-T FT-ICR MS. A custom-built FT-ICR mass spectrometer is equipped with a 22 cm horizontal room temperature bore 9.4 Tesla magnet (Oxford Corp., Oxney Mead, UK) and a modular ICR data station (PREDATOR).50, 52 Positive ions generated at atmospheric pressure in the external APPI source enter the skimmer region at ~2 Torr through a heated metal capillary into the first rf-only octopole. Ions pass through a quadrupole to a second octopole where they accumulate for 250-1000 ms. Helium gas was introduced during accumulation to collisionally cool the ions before transfer through a 200 cm rf-only octopole into an open cylindrical Penning ion trap (10 cm i.d. x 30 cm long). Octopole ion

guides were operated at 2.0 MHz and 240 Vp-p rf amplitude. Broadband frequency chirp excitation (70 – 700 kHz at a sweep rate of 50 Hz/s and

179 analysis software) data station.51, 52 Multiple (100-300) time-domain acquisitions were summed for each sample, Hanning-apodized, and zero- filled once prior to fast Fourier transform and magnitude calculation.200

Results and Discussion

Elemental Composition Assignment. Figure 9.2 shows a broadband positive-ion APPI FT-ICR mass spectrum of a raw asphaltene from Athabasca bitumen with 14,475 peaks of peak height >6σ of

180 baseline noise (400 < m/z < 900) and mass distribution centered at m/z 600. All ions are singly charged, as evident from the unit m/z spacing

12 13 12 between species differing by Cc vs. C1 Cc-1. Because petroporphyrins concentrate in the higher boiling fractions (asphaltene fraction), DPEP vanadyl porphyrins are observed at much higher relative abundance than other parent asphaltene components. Although positive-ion APPI can form both protonated and radical cations from other asphaltene aromatic components, the porphyrin core structure yields mainly radical molecular cations, plus protonated species at low abundance. Figure 9.3 shows a mass scale-expanded segment of a DPEP vanadyl porphyrin

+ (DBE = 18). The monoisotopic peak for [] at m/z 527.20104 appears at a

signal-to-noise ratio of 256:1 and mass resolving power, m/∆m50% =

440,464, in which /∆m50% is the mass spectral peak full width at half- maximum peak height. Mass spectral segments are also shown for the

13 13 C1 and C2 isotopomers peaks of the DBE 18 DPEP vanadyl porphyrin. The experimental relative abundances match well with those calculated for the assigned elemental composition. The mass error for each assigned elemental composition was <100 ppb.

181

The mass scale-expanded segment of a South American heavy crude at m/z 541 (Figure 9.4) shows a vanadyl porphyrin monoisotopic

13 peak and its C1 isotopomer 1.0033 Da higher in mass. The DPEP

+ homologue [C32H34N4O1V1] at m/z 541.21665 is identified to within ~50 ppb mass accuracy at 640,000 resolving power, and the corresponding

13 + [C31H34N4O1V1 C1] within <100 ppb mass accuracy at 650,000 resolving power. Vanadyl porphyrins are thus resolved and identified

182 unambiguously for the first time in an unprocessed, whole crude oil sample.

Double Bond Equivalents (DBE) Distribution. Figure 9.5 shows the DBE distribution for the vanadyl porphyrins observed in a raw asphaltene derived from Athabasca bitumen. Protonated molecules have half-integer DBE values (DBE = c – h/2 + n/2 + 1, calculated from the ion elemental composition, CcHhNnVvOoSs), and may thus be distinguished from radical cations with integer calculated DBE values. A

183 DBE value of 18 corresponds to a DPEP (CnH2n-30N4V1O1) structure that is the most abundant vanadyl porphyrin class detected in the raw asphaltene.218, 233 This result agrees with Rodgers et al., who reported a highly abundant DBE = 18 series of vanadyl porphyrins in fractions 63 isolated from heavy crude oil. The etio structure (CnH2n-28N4V1O1) corresponding to a DBE = 17 is also detected. A DBE value of 17.5 corresponds to the protonated form of the DBE = 17 etio cations, and both ion types are formed from the same parent species at roughly equal relative abundance. The etio-porphyrin structure corresponds to DBE = 17 and contains four pyrollic nitrogen rings and no exocyclic rings. The di-DPEP structure corresponds to a DBE = 19 with the addition of a pyrrole and benzene ring on the porphyrin core and forms both radical cations and protonated species. Table 9.1 lists the elemental formula, measured mass, theoretical mass, mass accuracy and resolving power for each of a series of etio (Table 9.1a) and DPEP (Table 9.1b) homologues in a bitumen raw asphaltene.

184

Figure 9.6 shows the DBE distribution for vanadyl porphyrins from whole heavy crude oil. The relative abundances of etio relative to DPEP structures in whole heavy crude is much higher than for raw asphaltenes (Figure 9.5). Etio structures form some protonated species, but at lower relative abundance in whole crude than in the asphaltene. Di-DPEP, rhodo-etio and rhodo-DPEP structures were also observed as radical molecular cations in the whole crude. We continue to analyze heavy crude oils and asphaltenes and have directly identified and speciated vanadyl porphryins in ten other petroleum samples.

185

186 Table 9.1a. An etio-vanadyl porphyrin homologous alkylation series in a raw asphaltene fraction of Athabasca bitumen

Elemental Composition Measured Mass (Da) Calculated Mass (Da) Mass Error (ppm) m/Δm50%

C26H24N4O1V1 459.13853 459.13843 0.22 409316 C27H26N4O1V1 473.15410 473.15408 0.05 486287 C28H28N4O1V1 487.16975 487.16973 0.01 478760 C29H30N4O1V1 501.18539 501.18538 0.03 460708 C30H32N4O1V1 515.20106 515.20103 0.06 451828 C31H34N4O1V1 529.21672 529.21668 0.08 426678 C32H36N4O1V1 543.23236 543.23233 0.07 407774 C33H38N4O1V1 557.24800 557.24798 0.04 384808 C34H40N4O1V1 571.26369 571.26363 0.11 343547 C35H42N4O1V1 585.27941 585.27928 0.23 635043 C36H44N4O1V1 599.29500 599.29493 0.12 715962

RMS Error = 0.116 ppm

187 Table 9.1b. A DPEP vanadyl porphyrin homologous alkylation series identified in a raw asphaltene fraction of Athabasca bitumen

Elemental Composition Measured Mass (Da) Calculated Mass (Da) Mass Accuracy (ppm) m/Δm50%

C27H24N4O1V1 471.13844 471.13843 0.03 481192 C28H26N4O1V1 485.15412 485.15408 0.09 468775 C29H28N4O1V1 499.16974 499.16973 0.03 463266 C30H30N4O1V1 513.18541 513.18538 0.06 450420 C31H32N4O1V1 527.20104 527.20103 0.03 440464 C32H34N4O1V1 541.21669 541.21668 0.02 420320 C33H36N4O1V1 555.23237 555.23233 0.08 405661 C34H38N4O1V1 569.24797 569.24798 -0.01 399390 C35H40N4O1V1 583.26363 583.26363 0.01 379118 C36H42N4O1V1 597.27924 597.27928 -0.06 355943 C37H44N4O1V1 611.29511 611.29493 0.30 339697 C38H46N4O1V1 625.31061 625.31058 0.05 324903 C39H48N4O1V1 639.32610 639.32623 -0.20 293459 C40H50N4O1V1 653.34200 653.34188 0.19 266821

RMS Error = 0.141 ppm

188 DBE vs. Carbon Number Images. The molecular weight distribution was independently verified with a low-resolution linear quadrupole ion trap mass spectrometer (LTQ-MS; ThermoFisher Scientific, Bremen, Germany) (data not shown). Contrary to a previous report,242 isobaric overlaps preclude direct assignment of vanadyl porphyrins to within the 99% confidence interval (~ ±1 ppm) required for accurate determination by FT-ICR MS. Without the necessary resolving power and mass accuracy, a vanadyl porphyrin could easily be

misassigned as an O2 species. Indeed, both elemental compositions fall within ±1 ppm of the measured mass. Arbitrarily constraining the allowed mass error (to say ±0.7 ppm) may exclude the O2 species from consideration, but at the expense of the assignment of low S/N ratio species, because mass precision varies directly with S/N ratio.244 For example, the previous assignment for an individual peak

+ [C31H32N4O1V1] at m/z 527.20104, there is also the possibility of an

+ isobaric overlap from a [C39H26O2 + H] species differing by 0.9 ppm from the experimentally measured mass. However, if all of the species are considered as individual members of a defined class and type, the mass errors within the class and type definitions may be used for confident reassignment of otherwise misassigned peaks. Specifically, Figure 9.7 (left) shows a color-coded isoabundance- contoured plot of DBE vs. carbon number for Athabasca bitumen asphaltene, in which vanadium is not included in the assignment of possible elemental compositions. The image is bimodal, with two apparent O2 class distributions centered at DBE = 18 with a carbon number ranging from 28 to 35 and (at much lower abundance) 1 < DBE < 24 and carbon number from 29 to 45. However, petroleum and bitumen universally exhibit monomodal (continuous) variation in DBE and carbon number. If vanadium is included in the possible elemental compositions, then the DBE vs. carbon number images separate into

189 monomodal distributions for O2 and N4O1V1 (vanadyl porphyrin) classes (Figure 9.7, right).

The vanadyl porphyrins, incorrectly assigned as O2 species have a root-mean-square (rms) error of +0.88 ppm, much higher than for other classes. After reassignment, the rms error drops to +0.21 ppm for N4O1V1 and +0.31 ppm for the correct O2 class. Thus, inclusion of vanadium in

13 13 the elemental composition, combined with detection of the C1 and C2

190 members of the isotopic distribution at the correct relative abundances (Figure 9.3) allow for unambiguous assignment of vanadyl porphyrins in the mass spectrum. The current example highlights the power of accurate mass, by enabling the Kendrick sorting procedure for elemental composition assignment for components of complex organic mixtures.

Figure 9.8 further illustrates the separation of O2 class from

N4O1V1 class components, this time for South American heavy crude oil

rather than asphaltenes. Again, the spurious bimodal O2 distribution is

resolved into separate O2 and N4O1V1 distributions, with two-fold reduction in mass error (to a level comparable to that for other classes).

Figure 9.8. Color-coded isoabundance contoured plots of DBE vs. carbon number for a South American heavy crude oil. Interpretation is as for Figure 6.7.

191

Heteroatom Class Distributions. Figure 9.9 shows the class distribution for all species of >1% relative abundance in the APPI FT-ICR mass spectrum of the raw asphaltene sample. The vanadyl porphyrins constitute the 9th most abundant class. The asphaltene fraction has high heteroatom content (N, O, S, Ni, V and other metals) and vanadyl porphyrins are therefore more abundant than in a whole crude oil.

Figure 9.9. Heteroatom class distribution for Athabasca bitumen asphaltenes. Vanadyl porphyrins are observed at ~3% relative abundance without preconcentration or isolation.

Nevertheless, the class distribution in Figure 9.10 shows that vanadyl porphyrins can be resolved and detected at <1% relative

192 abundance in a South American heavy crude oil, without prior extraction, even though the heavy crude is predictably dominated by sulfur and nitrogen polycyclic aromatics as well as furanic oxygen species.

Figure 9.10. Heteratom class distribution for a South American heavy crude oil for all species of >1% relative abundance, including vanadyl porphyrins.

Tables 9.2a and 9.2b list the elemental formula, measured mass, theoretical mass, mass accuracy and resolving power for a series of etio and DPEP homologues for the South American crude oil The rms error for the etio series was +0.26 ppm and +0.36 for the DPEP series.

193 Table 9.2a. An etio vanadyl porphyrin homologous alkylation series in a whole South American heavy crude oil

Elemental Composition Measured Mass (Da) Calculated Mass (Da) Mass Error (ppm) m/Δm50%

C27H26N4O1V1 473.15408 473.15408 +0.01 756354 C28H28N4O1V1 487.16972 487.16973 -0.01 662421 C29H30N4O1V1 501.18536 501.18538 -0.03 704176 C30H32N4O1V1 515.20102 515.20103 -0.01 656173 C31H34N4O1V1 529.21665 529.21668 -0.05 651711 C32H36N4O1V1 543.23226 543.23233 -0.12 631290 C33H38N4O1V1 557.24791 557.24798 -0.12 605232 C34H40N4O1V1 571.26357 571.26363 -0.10 591138 C35H42N4O1V1 585.27920 585.27928 -0.13 606111 C36H44N4O1V1 599.29480 599.29493 -0.21 582242 C37H46N4O1V1 613.31027 613.31058 -0.50 478361 C38H48N4O1V1 627.32618 627.32623 -0.08 540178 C39H50N4O1V1 641.34178 641.34188 -0.15 415665 C40H52N4O1V1 655.35720 655.35753 -0.50 423629 C41H54N4O1V1 669.37304 669.37318 -0.21 257934 C42H56N4O1V1 683.38851 683.38883 -0.46 238565 C43H58N4O1V1 697.40414 697.40448 -0.48 368923

RMS Error = 0.257 ppm

194 Table 9.2b. A DPEP vanadyl porphyrin homologous alkylation series in a whole South American heavy crude oil

Elemental Composition Measured Mass (Da) Calculated Mass (Da) Mass Error (ppm) m/Δm50%

C28H26N4O1V1 485.15407 485.15408 -0.01 606738 C29H28N4O1V1 499.16972 499.16973 -0.01 668546 C30H30N4O1V1 513.18534 513.18538 -0.07 677632 C31H32N4O1V1 527.20098 527.20103 -0.09 652291 C32H34N4O1V1 541.21665 541.21668 -0.05 642182 C33H36N4O1V1 555.23227 555.23233 -0.10 615755 C34H38N4O1V1 569.24789 569.24798 -0.15 597130 C35H40N4O1V1 583.26357 583.26363 -0.10 580700 C36H42N4O1V1 597.27923 597.27928 -0.08 573751 C37H44N4O1V1 611.29458 611.29493 -0.57 432094 C38H46N4O1V1 625.31019 625.31058 -0.62 367611 C39H48N4O1V1 639.32602 639.32623 -0.32 490400 C40H50N4O1V1 653.34162 653.34188 -0.38 489629 C41H52N4O1V1 667.35734 667.35753 -0.28 528049 C42H54N4O1V1 681.37314 681.37318 -0.06 452854 C43H56N4O1V1 695.38821 695.38883 -0.89 381154 C44H58N4O1V1 709.40416 709.40448 -0.45 229834 C45H60N4O1V1 723.41987 723.42013 -0.36 382258

RMS Error = 0.362 ppm

195 Conclusions

Here, we present the first identification and characterization of vanadyl porphyrins from a whole South American heavy crude and Athabasca bitumen asphaltene without prior sample treatment or fractionation. Future work will focus on characterization of vanadyl porphyrins in different crude oil samples and fractions.

196 REFERENCES

1. Boduszynski, M.M., Klaus H. Altgelt, Composition and Analysis of Heavy Petroleum Fractions. 1994, New York, NY: CRC Press.

2. Boduszynski, M.M. and K.H. Altgelt, Composition and Analysis of Heavy Petroleum Fractions. 1994, New York, NY: CRC Press.

3. Speight, J.G., Handbook of Petroleum Analysis. Chemical Analysis: A Series of Monographs on Analytical Chemistry and its Applications, ed. J.D. Wineforder. Vol. 158. 2001, New York: Wiley Interscience.

4. Speight, J.G., Environmental Analysis and Technology for the Refining Industry. Chemical Analysis : A Series of Monographs on Analytical Chemistry and its Applications, ed. J.D. Winefordner. Vol. 168. 2005, Hoboken, New Jersey: John Wiley and Sons, Inc.

5. Bott, R., Canada's Oils Sands and Heavy Oil, ed. B. Potyondi. 2000: Petroleum Communication Foundation.

6. Fan, Z., P. Rahimi, and T. Alem, Challenges in Processing Bitumen and Heavy Oils. Preprints-American Chemical Society, Division of Petroleum Chemistry, Salt Lake City, UT March 22-26,2009, 2009. 54(1).

7. Berkowitz, N., James G. Speight, The Oil Sands of Alberta. Fuel, 1975. 54: p. 138-149.

8. Hsu, C.S. and P.R. Robinson, Practical Advances in Petroleum Processing. Vol. 1. 2006, New York, NY: Springer 448.

9. Topsoe, H., C.S. Bjerne, and F.E. Massoth, Hydrotreating Catalysis: Science and Technology. Catalysis Science and Technology, ed. J.R. Anderson and M. Boudart. Vol. 11. 1996, New York: Springer- Verlag. 1-310.

10. Hein, F.J., Unconventional Energy Resources and Geospatial Information: 2006 Review. Natural Resources Research, 2007. 16(3): p. 243-261.

11. Czarnecki, J., et al., On the nature of Athabasca Oil Sands. Advances in Colloid and Interface Science, 2005. 114-115: p. 53-60.

12. Chow, D.L., et al., Recovery Techniques for Canada's Heavy oil and

197 Bitumen Resources. Journal of Canadian Petroleum Technology, 2008. 47(5): p. 12-17.

13. Boduszynski, M.M., Composition of Heavy Petroleums .1. Molecular-Weight, Hydrogen Deficiency, and Heteroatom Concentration as a Function of Atmospheric Equivalent Boiling-Point up to 1400- Degrees-F (760-Degrees-C). Energy & Fuels, 1987. 1(1): p. 2-11.

14. Boduszynski, M.M., Comprehensive Analysis of Heavy Petroleums. Abstracts of Papers of the American Chemical Society, 1989. 197: p. 56- Petr.

15. Speight, J.G. and S.E. Moschopedis, Chemistry of Asphaltenes, in Advances in Chemistry Series, J.W. Bunger and N.C. Li, Editors. 1981.

16. Ancheyta, J., et al., Asphaltene characterization as function of time on-stream during hydroprocessing of Maya crude. Catalysis Today, 2005. 109(1-4): p. 162-166.

17. Sheu, E.Y., Petroleum Asphaltene - Properties, Characterization, and Issues. Energy & Fuels, 2002. 16: p. 74-82.

18. Mullins, O.C., et al., Asphaltenes - Problematic but Rich in Potential. Oilfield Review, 2007. Summer 2007: p. 22-43.

19. Mullins, O.C., Review of the Molecular Structure and Aggregation of Asphaltenes and Petroleomics. Society of Petroleum Engineers Journal, 2008. March: p. 48-57.

20. de Boer, R.B., et al., Screening of Crude Oils for Asphalt Precipitation : Theory, Practice and the Selection of Inhibitors. SPEPF, 1995. 10(1): p. 55-61.

21. Sheu, E.Y., D.A. Storm, and M.M. De Tar, Fuel, 1992. 71: p. 299.

22. Boduszynski, M.M., Composition of Heavy Petroleums .2. Molecular Characterization. Energy & Fuels, 1988. 2(5): p. 597-613.

23. Merdrignac, I., et al., Secondary effects in steric exclusion chromatography separations and SEC-MS coupling, in AIChE Spring National Meeting. 2004: New Orleans, LA, United States.

24. Qian, K., et al., Desorption and Ionization of Heavy Petroleum Molecules and Measurement of Molecular Weight Distributions. Energy & Fuels, 2007. 21: p. 1042-1047.

198 25. Groenzin, H. and O.C. Mullins, Asphaltene molecular size and weight by time-resolved fluorescence depolarization, in Asphaltenes, Heavy Oils and Petroleomics, O.C. Mullins, et al., Editors. 2007, Springer: New York.

26. Guerra, R., A.B. Andrews, and O.C. Mullins, Comparison of asphaltene molecular diffusivity of various asphaltenes by fluorescence correlation spectroscopy. Fuel, 2007. 86: p. 2016-2020.

27. Mullins, O.C., Petroleomics and Structure-Function Relations of Crude Oils and Asphaltenes, in Asphaltenes, Heavy Oils and Petroleomics, O.C. Mullins, et al., Editors. 2007, Springer: New York.

28. Badre, S., et al., Molecular size and weight of asphaltene and asphaltene solubility fractions from coals, crude oils and bitumen. Fuel, 2006. 85: p. 1-11.

29. Mullins, O.C., et al., The Colloidal Structure of Crude Oil and the Structure of Oil Reservoirs. Energy & Fuels, 2007. 21: p. 2785-2794.

30. Hortal, A.R., et al., Molecular-Weight Distributions of Coal and Petroleum Asphaltenes from Laser Desorption/Ionization Experiments. Energy & Fuels, 2007. 21: p. 2863-2868.

31. Trejo, F., et al., Characterization of asphaltenes from hydrotreated products by SEC, LDMS, MALDI, NMR, and XRD. Energy & Fuels, 2007. 21(4): p. 2121-2128.

32. Herod, A.A., K.D. Bartle, and R. Kandiyoti, Characterization of Heavy Hydrocarbons by Chromatographic and Mass Spectrometric Methods: An Overview. Energy & Fuels, 2007. 21: p. 2176-2203.

33. Calemma, V., et al., Characterization of Asphaltenes Molecular Structure. Energy & Fuels, 1998. 12: p. 422-428.

34. Ruiz-Morales, Y., The Agreement Between Clar Structures and Nucleus-Independent Chemical Shift Values in Pericondensed Benzenoid Polycyclic Aromatic Hydrocarbons : An Application of the Y-Rule. J. Chem. Phys. A, 2002. 108(49): p. 10873-10896.

35. Zajac, G.W., N.K. Sethi, and J.T. Joseph, Molecular Imaging of Petroleum Asphaltenes by Scanning Tunneling Microscopy: Verification of Strucutre from 13C and Proton Nuclear Magnetic Resonance Data. Scan. Micros., 1994. 8(3): p. 463-470.

36. Jewell, D.M., et al., Ion-Exchange, Coordination, and Adsorption

199 Chromatographic Separation of Heavy-End Petroleum Distillates. Analytical Chemistry, 1972. 44(8): p. 1391-1395.

37. Jewell, D.M., R.G. Ruberto, and B.E. Davis, Systematic Approach to the Study of Aromatic Hydrocarbons in Heavy Distillates and Residues by Elution Adsorption Chromatography. Analytical Chemistry, 1972. 44(14): p. 2318-2321.

38. Quann, R.J. and S.B. Jaffe, Building Useful Models of Complex Reaction Systems in Petroleum Refining. Chemical Engineering Science, 1996. 51(10): p. 1615-1635.

39. Quann, R.J. and S.B. Jaffe, Structure-Oriented Lumping: Describing the Chemistry of Complex Hydrocarbon Mixtures. Industrial and Engineering Chemistry Research, 1992. 31(11): p. 2483-2497.

40. Fenn, J.B., et al., Electrospray Ionization for Mass-Spectrometry of Large Biomolecules. Science, 1989. 246(4926): p. 64-71.

41. Rodgers, R.P. and A.G. Marshall, Petroleomics: Advanced Characterization of Petroleum-Derived Materials Characterized by Fourier Transform Ion Cyclotron Resonance Mass Spectrometry (FT-ICR MS), in Asphaltenes, Heavy Oils and Petroleomics, O.C. Mullins, et al., Editors. 2007, Springer: New York.

42. Rodgers, R.P., T.M. Schaub, and A.G. Marshall, Petroleomics: MS returns to its roots. Analytical Chemistry, 2005. 77(1): p. 20A-27A.

43. Enke, C.G., The Science of Chemical Analysis and the Technique of Mass Spectrometry. Int. J. Mass Spectr., 2001. 212(1-3): p. 1-11.

44. Zhan, D.L. and J.B. Fenn, Electrospray mass spectrometry of fossil fuels. International Journal of Mass Spectrometry, 2000. 194(2-3): p. 197-208.

45. Schaub, T.M., et al., Instrumentation and Method for Ultrahigh Resolution Field Desorption Ionization Fourier Transform Ion Cyclotron Resonance Mass Spectrometry of Non-Polar Species. Analytical Chemistry, 2005. 77: p. 1317-1324.

46. Schaub, T.M., et al., Speciation of Aromatic Compounds in Petroleum Refinery Streams by Continuous Flow Field Desorption Ionization FT-ICR Mass Spectrometry. Energy & Fuels, 2005. 19(4): p. 1566-1573.

200 47. Bos, S.J., van Leeuwen, S.M., Karst, U., From fundamentals to applications : recent developments in atmospheric pressure photoionization mass spectrometry. Anal Bioanal Chem, 2006. 384: p. 85-99.

48. Raffaelli, A. and A. Saba, Atmospheric pressure photoionization mass spectrometry. Mass Spectrometry Reviews, 2003. 22(5): p. 318-331.

49. Senko, M.W., et al., External Accumulation of Ions for Enhanced Electrospray Ionization Fourier Transform Ion Cyclotron Resonance Mass Spectrometry. J. Am. Soc. Mass Spectrom., 1997. 8: p. 970-976.

50. Senko, M.W., et al., Electrospray Ionization FT-ICR Mass Spectrometry at 9.4 Tesla. Rapid Communications in Mass Spectrometry, 1996. 10: p. 1824-1828.

51. Blakney, G.T., et al. Further Improvements to the MIDAS Data Station for FT-ICR Mass Spectrometry. in Proc. 49th Amer. Soc. Mass Spectrom. Conf. on Mass Spectrom. & Allied Topics. 2001. Chicago, IL: Amer. Soc. Mass Spectrom.

52. Senko, M.W., et al., A High-Performance Modular Data System for FT-ICR Mass Spectrometry. Rapid Communications in Mass Spectrometry, 1996. 10: p. 1839-1844.

53. Comisarow, M.B. and A.G. Marshall, Fourier transform ion cyclotron resonance spectroscopy. Chem. Phys. Letters, 1974. 25(282-283).

54. Marshall, A.G., C.L. Hendrickson, and G.S. Jackson, Fourier Transform Ion Cyclotron Resonance Mass Spectrometry: A Primer, in Encyclopedia of Analytical Chemistry, R.A. Meyers, Editor. 2000, Wiley. p. 11694-11728.

55. Hsu, C.S., Z. Liang, and J.E. Campana, Hydrocarbon Characterization by Ultrahigh Resolution FTICRMS. Analytical Chemistry, 1994. 66: p. 850-855.

56. Rodgers, R.P., et al., FT-ICR mass spectral analysis of processed and unprocessed diesel fuels: An evaluation of the hydrotreating process. Abstracts of Papers of the American Chemical Society, 1998. 216: p. U597-U597.

57. Marshall, A.G., et al., Resolution and elemental composition identification of hydrocarbon and NSO compounds in petroleum distillates by fourier transform ion cyclotron resonance mass

201 spectrometry. Abstracts of Papers of the American Chemical Society, 2000. 219: p. U693-U693.

58. Qian, K., et al. Direct Speciation of Nitrogen-Containing Aromatics in Cerro Negro Extra Heavy Oil by Electrospray Ionization Fourier Transform Mass Spectrometry. in 48th Amer. Soc. for Mass Spectrom. Annual Conf. on Mass Spectrometry & Allied Topics. 2000. Long Beach, CA: Amer. Soc. Mass Spectrom.

59. Rodgers, R.P., et al., Efficacy of Bacterial Bioremediation: Demonstration of Complete Incorporation of Hydrocarbons into Membrane Phospholipids from Rhodococcus Hydrocarbon Degrading Bacteria by Electrospray Ionization Fourier Transform Ion Cyclotron Resonance Mass Spectrometry. Environmental Science and Technology, 2000. 34(3): p. 535-540.

60. Qian, K., et al., Reading chemical fine print: Resolution and identification of 3000 nitrogen-containing aromatic compounds from a single electrospray ionization Fourier transform ion cyclotron resonance mass spectrum of heavy petroleum crude oil. Energy & Fuels, 2001. 15(2): p. 492-498.

61. Hughey, C.A., et al., Kendrick Mass Defect Spectrum: A Compact Visual Analysis for Ultrahigh-Resolution Broadband Mass Spectra. Analytical Chemistry, 2001. 73(19): p. 4676-4681.

62. Qian, K., et al., Resolution and Identification of Elemental Compositions for More than 3000 Crude Acids in Heavy Petroleum by Negative-Ion Microelectrospray High Field Fourier Transform Ion Cyclotron Resonance Mass Spectrometry. Energy & Fuels, 2001. 15: p. 1505-1511.

63. Rodgers, R.P., et al., Molecular Characterization of Petroporphyrins in Crude Oil by Electrospray Ionization Fourier Transform Ion Cyclotron Resonance Mass Spectrometry. Canadian Journal of Chemistry, 2001. 79: p. 546-551.

64. Hughey, C.A., R.P. Rodgers, and A.G. Marshall, Resolution of 11,000 Compositionally Distinct Components in a Single Electrospray Ionization Fourier Transform Ion Cyclotron Resonance Mass Spectrum of Crude Oil. Analytical Chemistry, 2002. 74(16): p. 4145-4149.

65. Hughey, C.A., et al., Identification of acidic NSO compounds in crude oils of different geochemical origins by negative ion electrospray Fourier transform ion cyclotron resonance mass spectrometry. Organic Geochemistry, 2002. 33(7): p. 743-759.

202

66. Wu, Z., et al., Resolution of 10,000 Compositionally Distinct Components in Polar Coal Extracts by Negative-Ion Electrospray Ionization Fourier Transform Ion Cyclotron Resonance Mass Spectrometry. Energy & Fuels, 2003. 17: p. 946-953.

67. Smith, D.F., et al., Characterization of Athabasca Bitumen Heavy Vacuum Gas Oil Distillation Cuts by Negative/Positive Electrospray Ionization and Automated Liquid Injection Field Desorption Ionization Fourier Transform Ion Cyclotron Resonance Mass Spectrometry. Energy & Fuels, 2008. 22(5): p. 3118-3125.

68. Kendrick, E., A Mass Scale Based on CH2 = 14.0000 for High Resolution Mass Spectrometry of Organic Compounds. Analytical Chemistry, 1963. 35(13): p. 2146-2154.

69. Marshall, A.G., C.L. Hendrickson, and G.S. Jackson, Fourier Transform Ion Cyclotron Resonance Mass Spectrometry: A Primer. Mass Spectrometry Reviews, 1998. 17: p. 1-35.

70. Ledford, E.B., Jr., D.L. Rempel, and M.L. Gross, Space Charge Effects in Fourier Transform Mass Spectrometry. Mass Calibration. Analytical Chemistry, 1984. 56: p. 2744-2748.

71. Marshall, A.G. and R.P. Rodgers, Petroleomics: The Next Grand Challenge for Chemical Analysis. Accounts of Chemical Research, 2004. 37(1): p. 53-59.

72. Meyer, R.F., Attanasi, E.D, Heavy Oil and Natural Bitumen- Strategic Petroleum Resources. US Geological Survey: Reston, VA, 2003. Vol. FS-070-03.

73. Kharrat, A.M., New Approach for Characterizing Heavy Oils. Energy & Fuels, 2008. 22: p. 1402-1403.

74. Mullins, O.C., et al., Asphaltenes, Heavy Oil and Petroleomics. 2007, New York: Springer.

75. Rodgers, R.P. and A.G. Marshall, Petroleomics: Chemistry of the Underworld. Proceedings of the National Academy of Sciences of the United States of America, 2008. 105(47): p. 1-6.

76. Rodgers, R.P., et al., Compositional analysis for identification of arson accelerants by electron ionization Fourier transform ion cyclotron resonance high-resolution mass spectrometry. Journal of Forensic

203 Sciences, 2001. 46(2): p. 268-279.

77. He, F., C.L. Hendrickson, and A.G. Marshall, Baseline Mass Resolution of Peptide Isobars: A New Record for Molecular Mass Resolution. Analytical Chemistry, 2001. 73: p. 647-650.

78. Goncalves, M., Ribeiro, D., da Mota, D.,Teixeira, A., Teixeira, M., Investigation of petroleum medium fractions and distillation residues from Brazilian crude oils by thermogravimetry. Fuel, 2006. 85: p. 1151- 1155.

79. Woods, J.R., et al., Characterization of a Gas Oil Fraction and Its Hydrotreated Products. Petroleum Science and Technology, 2004. 22(3 & 4): p. 347-365.

80. Vishnoi, V., Agrawal, K.M., Singh, I.D., Raizada, B.B., Compositional Analysis of Waxy Distillate Fractions by 1H and 13C NMR Spectroscopy. Petroleum Science and Technology, 2005. 27: p. 931-937.

81. Clutter, D.R., Petrakis, L., Stenger, R.L. Jr, Jensen, R.K., Nuclear Magnetic Resonance Spectrometry of Petroleum Fractions. Analytical Chemistry, 1972. 44(8): p. 1395-1405.

82. Zhao, S., et al., Molecular Nature of Athabasca Bitumen. Petroleum Science and Technology, 2000. 18(5&6): p. 587-606.

83. Barman, B.N., Characterization of Feeds, Intermediates, and Products from Heavy Oil Processes by High-Temperature Simulated Distillation and Thin-Layer Chromatography with Flame Ionization Detection. Energy & Fuels, 2005. 19: p. 1995-2000.

84. Klein, G.C., et al., Petroleomics: Electrospray ionization FT-ICR mass analysis of NSO compounds for correlation between total acid number, corrosivity, and elemental composition. Abstracts of Papers of the American Chemical Society, 2003. 225: p. U844-U844.

85. Crawford, K.E., Campbell, J.L., Fiddler, M.N., Duan, P., Qian, K., Gorbaty, M.L., Kenttamaa, H.I., Laser-Induced Acoustic Desorption/Fourier Transform Ion Cyclotron Resonance Mass Spectrometry for Petroleum Distillate Analysis. Analytical Chemistry, 2005. 77: p. 7916-7923.

86. Schaub, T.M., et al., High-resolution field desorption/ionization Fourier transform ion cyclotron resonance mass analysis of nonpolar molecules. Analytical Chemistry, 2003. 75(9): p. 2172-2176.

204

87. Purcell, J.M., Juyal, P., Kim, D.G., Rodgers, R.P., Hendrickson, C.L., Marshall, A.G., Sulfur Speciation in Petroleum : Atmospheric Pressure Photoionization or Chemical Derivatization and Electrospray Ionization Fourier Transform Ion Cyclotron Resonance Mass Spectrometry. Energy & Fuels, 2007. 21(5): p. 2869-2874.

88. Purcell, J.M., Petroleum analysis by atmospheric pressure photionization Fourie transform ion cyclotron resonance mass spectrometry, in Department of Chemistry and Biochemistry. 2007, Florida State University: Tallahassee. p. 161.

89. Purcell, J.M., et al., Atmospheric pressure photoionization proton transfer for complex organic mixtures investigated by Fourier transform ion cyclotron resonance mass spectrometry. Journal of the American Society for Mass Spectrometry, 2007. 18(9): p. 1682-1689.

90. Purcell, J.M., et al., Speciation of nitrogen containing aromatics by atmospheric pressure photoionization or electrospray ionization Fourier transform ion cyclotron resonance mass spectrometry. Journal of the American Society for Mass Spectrometry, 2007. 18(7): p. 1265-1273.

91. Purcell, J.M., et al., Atmospheric Pressure Photoionization Fourier Transform Ion Cyclotron Resonance Mass Spectrometry for Complex Mixture Analysis. Analytical Chemistry, 2006. 78(16): p. 5906-5912.

92. Hughey, C.A., et al., Elemental composition analysis of processed and unprocessed diesel fuel by electrospray ionization Fourier transform ion cyclotron resonance mass spectrometry. Energy & Fuels, 2001. 15(5): p. 1186-1193.

93. Qian, K., et al., Resolution and identification of elemental compositions for more than 3000 crude acids in heavy petroleum by negative-ion microelectrospray high-field Fourier Transform ion cyclotron resonance mass spectrometry. Energy & Fuels, 2001. 15(6): p. 1505- 1511.

94. Hughey, C.A., et al., Identification of Acidic NSO Compounds in Crude Oils of Different Geochemical Origins by Negative Ion Electrospray Fourier Transform Ion Cyclotron Resonance Mass Spectrometry. Org. Geochem., 2002. 33: p. 743-759.

95. Qian, K.N., et al., Resolution and identification of elemental compositions for more than 3000 crude acids in heavy petroleum by negative-ion microelectrospray high-field Fourier transform ion cyclotron resonance mass spectrometry. Energy & Fuels, 2001. 15(6): p. 1505-

205 1511.

96. Schaub, T.M., et al., Continuous Flow Sample Introduction for Field Desorption/Ionization Mass Spectrometry. Rapid Communications in Mass Spectrometry, 2004. 18: p. 1641-1644.

97. Fu, J.-M., et al., Comprehensive Compositional Analysis of Hydrotreated and Untreated Nitrogen-Concentrated Fractions from Syncrude Oil by Electron Ionization, Field Desorption Ionization, and Electrospray Ionization Ultrahigh-Resolution FT-ICR Mass Spectrometry. Energy & Fuels, 2006. 20: p. 1235-1241.

98. Smith, D.F.S., T.M.; Rodgers, R.P.; Rahimi, P.; Marshall, A.G. In Characterization of Acidic Species from Athabasca Canadian Bitumen by Negative Ion ESI FT-ICR MS. in Oilsands 2006. Edmonton, Alberta, Canada.

99. Robb, D.B., T.R. Covey, and A.P. Bruins, Atmospheric pressure photoionisation: An ionization method for liquid chromatography-mass spectrometry. Analytical Chemistry, 2000. 72(15): p. 3653-3659.

100. Robb, D.B. and M.W. Blades, State-of-the-Art in Atmospheric Pressure Photoionization for LC/MS. Analytica Chimica Acta, 2008. 627(1): p. 34-49.

101. Robb, D.B. and M.W. Blades, Factors affecting primary ionization in dopant-assisted atmospheric pressure photoionization (DA-APPI) for LC/MS. Journal of the American Society for Mass Spectrometry, 2006. 17: p. 130-139.

102. Schaub, T.M., Blakney, G.T., Hendrickson, C.L., Quinn, J.P., Senko, M.W., Marshall, A.G. Performance Characteristics of a 14.5 Tesla LTQ FT-ICR Mass Spectrometer. in ASMS. 2007. Indianapolis, IN.

103. Wilcox, B.E., C.L. Hendrickson, and A.G. Marshall, Improved Ion Extraction from a Linear Octopole Ion Trap: SIMION Analysis and Experimental Demonstration. J. Am. Soc. Mass Spectrom., 2002. 13: p. 1304-1312.

104. Marshall, A.G., Hendrickson, Christopher L., Emmet, Mark R., Rodgers, Ryan P., Blakney, Greg A., Nilsson, Carol L., Schaub, Tanner M. Fourier transform ion cyclotron resonance mass spectrometry at 14.5 Tesla. in ACS National Meeting. 2007. Boston, MA, United States.

105. Boduszynski, M.M. and K.H. Altgelt, Composition of Heavy

206 Petroleums. 3. An Improved Boiling Point-Molecular Weight Relation. Energy and Fuels, 1992. 6(1): p. 68-72.

106. Boduszynski, M.M. and K.H. Altgelt, Composition of Heavy Petroleums .4. Significance of the Extended Atmospheric Equivalent Boiling-Point (Aebp) Scale. Energy & Fuels, 1992. 6(1): p. 72-76.

107. Schaub, T.M., Christopher L. Hendrickson, Stevan Horning, John P. Quinn, Michael W. Senko, and Alan G. Marshall, High Performance Mass Spectrometry: Fourier Transform Ion Cyclotron Resonance at 14.5 Tesla. Analytical Chemistry, 2008. 80(11): p. 3985-3990.

108. Rodgers, R.P., et al., Resolution, Elemental Composition, and Simultaneous Monitoring by Fourier Transform Ion Cyclotron Resonance Mass Spectrometry of Organosulfur Species Before and After Diesel Fuel Processing. Analytical Chemistry, 1998. 70: p. 4743-4750.

109. Coleman, H.J., D.E. Hirsch, and J.E. Dooley, Separation of crude oil fractions by Gel Permeation Chromatography. Analytical Chemistry, 1969. 41(6): p. 800-804.

110. Sullivan, R.F., M.M. Boduszynski, and J.C. Fetzer, Molecular transformations in hydrotreating and hydrocracking. Energy & Fuels, 1989. 3: p. 603-612.

111. Frysinger, G.S., R.B. Gaines, and C.M. Reddy, GC X GC - A New Analytical Tool for Environmental Forensics. Environ. Forensics, 2002. 3: p. 27-34.

112. Mullins, O.C., et al., Visible-Near-Infrared Spectroscopy by Downhole Fluid Analysis Coupled with Comprehensive Two-Dimensional Gas Chromatography to Address Oil Reservoir Complexity. Energy & Fuels, 2008. 22: p. 496-503.

113. Reddy, C.M., et al., Identification and Quantitation of Alkene-based Drilling Fluids in Crude Oils by Comprehensive Two-Dimensional Gas Chromatography with Flame Ionization. Journal of Chromatography A, 2007. 1148: p. 100-107.

114. Goncalves, M.L.A., et al., Investigation of petroleum medium fractions and distillation residues from Brazilian crude oils by thermogravimetry. Fuel, 2006. 85: p. 1151-1155.

115. Clutter, D.R., et al., Nuclear magnetic resonance spectrometry of petroleum fractions. Analytical Chemistry, 1972. 44(8): p. 1395-1405.

207 116. Vishnoi, V., et al., Compositional analysis of waxy distillate fractions by 1H and 13C NMR spectroscopy. Petroleum Science and Technology, 2005. 27: p. 931-937.

117. Zhao, S., et al., Feedstock characteristic index and critical properties of heavy crudes and petroleum residua. Journal of Petroleum Science and Engineering, 2004. 41: p. 233-242.

118. Zhao, S., et al., Correlation of processability and reactivity data for residua from bitumen, heavy oils and conventional crudes: Characterization of fractions from super-critical pentane separation as a guide to process selection. Catalysis Today, 2007. 125: p. 122-136.

119. Zhao, S., et al., Systematic characterization of petroleum residua based on SFEF. Fuel, 2004. 84: p. 635-645.

120. Robbins, W.K., Quantitative measurement of mass and aromaticity distributions for heavy distillates, 1. Capabilities of the HPLC-2 system. Journal of Chromatographic Science, 1998. 36(9): p. 457-466.

121. Coleman, H.J., Hirsch, D.E., Dooley, J.E., Separation of Crude Oil Fractions by Gel Permeation Chromatography. 1969. 6: p. 800-804.

122. Takeuchi, T. and D. Ishii, Gradient separations of complex mixtures by micro high-performance liquid chromatography. HRC & CC, 1983. 6: p. 310-315.

123. Roehner, R.M., et al., Comparative Compositional Study of Crude Oil Solids from the Trans Alaska Pipeline System Using High- Temperature Gas Chromatography. Energy & Fuels, 2002. 16(1): p. 211- 217.

124. Wang, Z., M. Fingas, and K. Li, Fractionation of light crude oil and identification and quantitation of aliphatic, aromatic and biomarke compounds by GC-FID and GC-MS, Part 1. Journal of Chromatographic Science, 1990. 32: p. 361-382.

125. Morgan, T.J., et al., Characterization of Molecular Mass Ranges of Two Coal Tar Distillate Fractions (Creosote and Anthracene Oils) and Aromatic Standards by LD-MS, GC-MS, Probe-MS and Size-exclusion Chromatography. Energy and Fuels, 2008. 22(5): p. 3275-3292.

126. Qian, K., Diehl, J.W., Dechert, G.J., DiSanzo, F.P., The coupling of supercritical fluid chromatography and field ionization time-of-flight high-resolution mass spectromerry for rapid and quantitative analysis of petreoleum middle distillates. European Journal of Mass Spectrometry,

208 2004. 10(2): p. 187-196.

127. Wright, B.W.S., R.D., Supercritical fluid chromatography-mass spectrometry: a potentially useful technique for porphyrin analysis. Organic Geochemistry, 1989. 14(2): p. 227-332.

128. Altgelt, K.H. and M.M. Boduszynski, Composition of Heavy Petroleums .3. An Improved Boiling-Point Molecular-Weight Relation. Energy & Fuels, 1992. 6(1): p. 68-72.

129. Boduszynski, M.M., Limitations of average structure determination for heavy ends in fossil fuels. Liquid Fuels Technology, 1984. 2(3): p. 211-232.

130. Stanford, L.A., et al., Characterization of Compositional Changes in Vacuum Gas Oil Distillation Cuts by Electrospray Ionization FT-ICR Mass Spectrometry. Energy & Fuels, 2006. 20: p. 1664-1673.

131. Guan, S., A.G. Marshall, and S.E. Scheppele, Resolution and chemical formula identification of aromatic hydrocarbons containing sulfur, nitrogen, and/or oxygen in crude oil distillates. Analytical Chemistry, 1996. 68: p. 46-71.

132. Fu, J.M., et al., Comprehensive compositional analysis of hydrotreated and untreated nitrogen-concentrated fractions from syncrude oil by electron ionization, field desorption ionization, and electrospray ionization ultrahigh-resolution FT-ICR mass spectrometry. Energy & Fuels, 2006. 20(3): p. 1235-1241.

133. Wu, Z., et al., Resolution of 10 000 Compositionally Distinct Components in Polar Coal Extracts by Negative-Ion Electrospray Ionization Fourier Transform Ion Cyclotron Resonance Mass Spectrometry. Energy & Fuels, 2003. 17(4): p. 946-953.

134. Stanford, L.A., et al., Compositional Characterization of Bitumen/Water Emulsion Films by Negative- and Positive-Ion Electrospray Ionization and Field Desorption/Ionization Fourier Transform Ion Cyclotron Resonance Mass Spectrometry. Energy & Fuels, 2007: p. ACS ASAP.

135. Kim, S., et al., Microbial Alteration of the Acidic and Neutral Polar NSO Compounds in Petroleum Revealed by Fourier Transform Ion Cyclotron Resonance Mass Spectrometry. Organic Geochemistry, 2005. 36: p. 1117-1134.

136. Smith, D.F., et al., Self-Association of Organic Acids in Petroleum

209 and Canadian Bitumen Characterized by Low- and High-Resolution Mass Spectrometry. Energy & Fuels, 2007: p. ACS ASAP.

137. Wu, Z., et al., Comparative Compositional Analysis of Untreated and Hydrotreated Oil by Electrospray Ionization Fourier Transform Ion Cyclotron Resonance Mass Spectrometry. Energy & Fuels, 2005. 19(3): p. 1072-1077.

138. Schaub, T.M., et al., Heat-Exchanger Deposits in an Inverted Steam-Assisted Gravity Drainage Operation. Part 2. Organic Acid Analysis by Electrospray Ionization Fourier Transform Ion Cyclotron Resonance Mass Spectrometry. Energy & Fuels, 2007. 21(1): p. 185-194.

139. Cai, Y., et al., Advantages of atmospheric pressure photoionization mass spectrometry in support of drug discovery. Rapid Communications in Mass Spectrometry, 2005. 19(12): p. 1717-1724.

140. Wang, G., Y. Hsieh, and W.A. Korfmacher, Comparison of atmospheric pressure chemical ionization, electrospray ionization, and atmospheric pressure photoionization for the determination of cyclosporin A in rat plasma. Analytical Chemistry, 2005. 77(2): p. 541- 548.

141. Cai, S.-S., K.A. Hanold, and J.A. Syage, Comparison of atmospheric pressure photoionization and atmopsheric pressure chemical ionization for normal-phase LC/MS chiral analysis of pharmaceuticals. Analytical Chemistry, 2007. 79(6): p. 2491-2498.

142. Silva, L.C., et al., Quantification of isosorbide 5-mononitrate in human plasma by liquid chromatography-tandem mass spectrometry using atmospheric pressure photoionization. Journal of Chromatography B, 2006. 832(2): p. 302-306.

143. Muller, A., et al., Identification of conjugated linoleic acid elongation and beta-oxidation products by coupled silver-ion HPLC APPI- MS. Journal of Chromatography B, 2006. 837(1-2): p. 147-152.

144. Cai, S.-S. and J.A. Syage, Comparison of atmospheric pressure photoionization, atmospheric pressure chemical ionization, and electrosprat ionization mass spectrometry for analysis of lipids. Analytical Chemistry, 2006. 78(4): p. 1191-1199.

145. Cai, S.-S., et al., Ultra Performance Liquid Chromatography- Atmospheric Pressure Photoionization-Tandem Mass Spectrometry for High-Sensitivity and High-Throughput Analysis of U.S. Environmental Protection Agency 16 Priority Pollutants Polynuclear Aromatic

210 Hydrocarbons. Analytical Chemistry, 2009. XX(xx): p. xxxx-xxxx.

146. Itoh, N., et al., Mechanism of Ionization of Polycyclic Aromatic Hydrocarbons by a Toluene/Anisole Mixture as a Dopant in Liquid Chromatography/Dopant-Assisted Atmospheric Pressure Photoionization/Mass Spectrometry. Polycyclic Aromatic Hydrocarbons, 2009. 29(1): p. 41-55.

147. Smith, D.F., D.B. Robb, and M.W. Blades, Comparison of Dopants for Charge Exchange Ionization of Nonpolar Polycyclic Aromatic Hydrocarbons with Reversed-Phase LC-APPI-MS. J. Am. Soc. Mass Spectrom., 2009. 20(1): p. 73-79.

148. Song, L. and J.E. Bartmess, Liquid Chromatography/Negative Ion Atmospheric Pressure Photoionization Mass Spectrometry : A Highly Sensitive Method for the Analysis of Organic Explosives. Rapid Communications in Mass Spectrometry, 2009. 23(1): p. 77-84.

149. Leinonen, A., T. Kuuranne, and R. Kostiainen, Liquid chromatography/mass spectrometry in anabolic steriod analysis- optimization and comparison of three ionization techniques: electrospray ionization, atmospheric pressure chemical ionization and atmospheric pressure photoionization. Journal of Mass Spectrometry, 2002. 3(7): p. 693-698.

150. Roy, S., et al., Liquid chromatography on porous graphitic carbon with atmospheric pressure photoionization mass spectrometry and tandem mass spectrometry for the analysis of glycosphingolipids. Journal of Chromatography A, 2006. 1117(2): p. 154-162.

151. Karuna, R., A. von Eckardstein, and K.M. Rentsch, Dopant Assisted Atmospheric Pressure Photoionization (DA-APPI) Liquid Chromatography-Mass Spectrometry for the Quantification of 27- Hydroxycholesterol in Plasma. Journa of Chromatography A, 2009. 877(3): p. 261-268.

152. Greig, M.J., et al., Fourier transform ion cyclotron resonance mass spectrometry using atmospheric pressure photoionization for high- resolution analyses of corticosteroids. Rapid Communications in Mass Spectrometry, 2003. 17: p. 2763-2768.

153. McKenna, A.M., et al., Identification of vanadyl porphyrins in a heavy crude oil and raw asphaltene by atmospheric pressure photoionization FT-ICR mass spectrometry. Energy & Fuels, 2009. 23: p. 2122-2128.

211 154. McKenna, A.M., Purcell, J.M. Rodgers, R.P., Rahimi, P. and Alan G. Marshall. Correlation of distillation and nebulization temperature of bitumen by use of atmospheric pressure photoionization Fourier transform ion cyclotron resonance mass spectrometry. in Abstracts of Papers, 235th ACS National Meeting, New Orleans, LA, United States, April 6-10, 2008. 2008. New Orleans, LA, United States.

155. Jaffe, S.B., H. Freund, and W.N. Olmstead, Extension of Structure- Oriented Lumping to Vacuum Residua. Industrial and Engineering Chemistry Research, 2005. 44(26): p. 9840-9852.

156. Herod, A.A., K.D. Bartle, and R. Kandiyoti, Comment on a Paper by Mullins, Martinez-Haya, and Marshall "Contrasting Perspective on Asphaltene Molecular Weight. This Comment vs The Overview of A.A. Herod, K.D.Bartle, and R. Kandiyoti". Energy and Fuels, 2008. 22(6): p. 4312-4317.

157. Herod, A.A. and R. Kandiyoti, Comment on "Limitations of Size- Exclusion Chromatography in Analyzing Petroleum Asphaltenes: A Proof by Atomic Force Microscopy" By M. Behrouzi and P.F. Luckman, Energy Fuels, 2008, 22 (3), 1792-1798, DOI: 10.1021/ef800064q. Energy and Fuels, 2008. 22(6): p. 4307-4309.

158. Mullins, O.C., Rebuttal to comment by professors Herod, Kandiyoti, and Bartle on "Molecular Size and Weight of Asphaltene and Asphaltene Solubilty Fractions from Coals, Crude Oils and Bitumens". Fuel, 2006. 86(1-2): p. 309-312.

159. Mullins, O.C., Rebuttal to comment by professors Herod, Kandiyoti, and Bartle on “Molecular size and weight of asphaltene and asphaltene solubility fractions from coals, crude oils and bitumen” Fuel, 2007. 86(1-2): p. 309-312.

160. Mullins, O.C., B. Matinez-Haya, and A.G. Marshall, Contrasting Perspective on Asphaltene Molecular Weight. This Comment vs The Overview of A.A. Herod, K.D. Bartle, and R. Kandiyoti. Energy and Fuels, 2008. 22(3): p. 1765-1773.

161. Behrouzi, M. and P.F. Luckman, Limitations of Size-Exclusion Chromatography in Analyzing Petroleum Asphaltenes: A Proof by Atomic Force Microscopy". Energy and Fuels, 2008. 22(3): p. 1792-1798.

162. Herod, A.A., R. Kandiyoti, and K.D. Bartle, Comment on 'Molecular size and weight of asphaltene and asphaltene solubilty fractions from coals, crude oils and bitumen' by S. Badre, C.C. Goncalves, K. Norinaga, G. Gustavson and O.C. Mullins; Fuel 85 (2006) 1-11. Fuel, 2006. 85: p.

212 1950-1951.

163. Morgan, T.J., et al., On the Limitations of UV-Fluorescence Spectroscopy in the Detection of High-Mass Hydrocarbon Molecules. Energy and Fuels, 2005. 19(1): p. 164-169.

164. Strausz, O.P., et al., A Critique of Asphaltene Fluorescence Decay and Depolarization-Based Claims about Molecular Weight and Molecular Architecture. Energy and Fuels, 2008. 22(2): p. 1156-1166.

165. Mullins, O.C., Rebuttal to Strausz et al. Regarding Time-Resolved Fluorescence Depolarization of Asphaltenes. Energy and Fuels, 2009. 23: p. 2845-2854.

166. Betancourt, S.S., et al., Nanoaggregates of Asphaltenes in a Reservoir Crude Oil and Reservoir Connectivity. Energy & Fuels, 2009.

167. Buch, L., et al., Molecular size of asphaltene fractions obtained from residuum hydrotreatment. Fuel, 2003. 82(9): p. 1075-1084.

168. Buenrostro-Gonzalez, E., et al., The overriding chemical principles that define asphaltenes. Energy and Fuels, 2001. 15(4): p. 972-978.

169. Cunico, R.L., E.Y. Sheu, and O.C. Mullins, Molecular Weight Measurement of UG8 Asphaltene Using APCI Mass Spectroscopy. Petroleum Science and Technology, 2004. 22(7 & 8): p. 787-798.

170. Groenzin, H. and O.C. Mullins, Asphaltene molecular size and structure. Journal of Physical Chemistry A, 1999. 103(50): p. 11237- 11245.

171. Groenzin, H. and O.C. Mullins, Molecular size and structure of asphaltenes from various sources. Energy and Fuels, 2000. 14(3): p. 677- 684.

172. Merdrignac, I. and D. Espinat, Physicochemical Characterization of Petroleum Fractions: the State of the Art. Oil and Gas Science and Technology, 2007. 62(1): p. 7-32.

173. Morgan, T.J., et al., Essential oils investigated by size exclusion chromatography and gas chromatography-mass spectrometry. Energy and Fuels, 2006. 20(734-737).

174. Johnson, B.R., et al., Improved size exclusion chromatography of coal derived materials usin N-methyl 2-pyrrolidinone as mobile phase.

213 Prepr. Pap.-Am. Chem. Soc., Div. Fuel Chem., 1995. 40(3): p. 457-460.

175. Herod, A.A., et al., Solubility limitations in the determination of molecular mass distributions of coal liquefaction and hydrocracking products: N-methyl 2-pyrrolidinone as mobile phase in size exclusion chromatography. Energy and Fuels, 1996. 10: p. 743-750.

176. Shi, S.D.-H., et al., Comparison and interconversion of the two most common frequency-to-mass calibration functions for Fourier transform ion cyclotron resonance mass spectrometry. International Journal of Mass Spectrometry, 2000. 195/196: p. 591-598.

177. McLafferty, F.W. and F. Turecek, Interpretation of Mass Spectra. 4th ed. ed. 1993, Mill Valley, CA: University Science Books. 371.

178. Lawrence, E.O. and M.S. Livingston, The production of high speed protons without the use of high voltages. Physical Review, 1931. 38(4): p. 834-834.

179. McKenna, A.M., et al. Speciation of Nitrogen, Oxygen and Sulfur Species in Unconventional Crudes by Electrospray Ionization and Atmospheric Pressure Photoionization Fourier Transform Ion Cyclotron Resonance Mass Spectrometry. in Division of Petroleum Chemistry, American Chemical Society. 2008. Philadelphia, PA August xx-xx, 2008.

180. Energy, U.S.D.o., Strategic Petroleum Reserve Crude Oil Assay Manual, A.S.f.F.E. U.S. Department of Energy, Office of Petroleum Reserves, Editor. 2008, Washington, D.C.

181. Bauserman, J.W., G.W. Mushrush, and D.R. Hardy, Organic Nitrogen Compounds and Fuel Instability in Middle Distillate Fuels. Industrial and Engineering Chemistry Research, 2008. 47: p. 2867-2875.

182. Dalluge, J., J. Beens, and U.A. Brinkman, Comprehensive Two- Dimensional Gas Chromatography: A Powerful and Versatile Analytical Tool Journal of Chromatography A, 2003. 1000: p. 69-108.

183. Vendeuvre, C., et al., Characterisation of Middle-Distillates by Comprehensive Two-Dimensional Gas Chromatography (GC x GC): A Powerful Alternative for Performing Various Standard Analysis of Middle Distillates. Journal of Chromatography A, 2005. 1086: p. 21-28.

184. Frysinger, G.S., et al., Resolving the Unresolved Complex Mixture in Petroleum-Contaminated Sediments. Environmental Science and Technology, 2003. 37: p. 1653-1662.

214

185. Comisarow, M.B. and A.G. Marshall, Frequency-Sweep Fourier Transform Ion Cyclotron Resonance Spectroscopy. Chemical Physics Letters, 1974. 26: p. 489-490.

186. Rodgers, R.P., et al., Stable isotope incorporation triples the upper mass limit for determination of elemental composition by accurate mass measurement. Journal of the American Society for Mass Spectrometry, 2000. 11(10): p. 835-840.

187. Marshall, A.G., et al., Resolution and identification of elemental compositions of hydrocarbon and NSO components of crude oil and petroleum distillates by Fourier transform ion cyclotron resonance mass spectrometry. Abstracts of Papers of the American Chemical Society, 2000. 220: p. U144-U145.

188. Marshall, A.G. and F.R. Verdun, Fourier transforms in NMR, optical, and mass spectrometry : a user's handbook. 1990, Amsterdam ; New York: Elsevier. xvi, 450.

189. Xian, F., et al. Broadband Phase Correction of Complex FT-ICR Mass Spectra. in Proceedings of the 56th American Society for Mass Spectrometry Conference on Mass Spectrometry and Allied Topics 2008. Denver, Colorado.

190. Beu, S.C., et al., Broadband Phase Correction of FT-ICR Mass Spectra via Simultaneous Excitation and Dectection. Analytical Chemistry, 2004: p. 5756-5761.

191. Sommer, H., H.A. Thomas, and J.A. Hipple, The Measurement of e/M by Cyclotron Resonance. Physical Review, 1951. 82(5): p. 697-702.

192. Marshall, A.G., Theory for Ion Cyclotron Resonance Absorption Line Shapes. Journal of Chemical Physics, 1971. 55: p. 1343-1354.

193. Marshall, A.G., Dispersion versus Absorption (DISPA): Hilbert Transforms in Spectral Line Shape Analysis, in Fourier, Hadamard, and Hilbert Transforms in Chemistry, A.G. Marshall, Editor. 1982, Plenum: New York. p. 99-123.

194. Marshall, A.G., Spectroscopic Dispersion versus Absorption: A New Method for Distinguishing a Distribution in Peak Position from a Distribution in Line Width. Journal of Physical Chemistry, 1979. 83: p. 521-524.

215 195. Grosshans, P.B., et al., Upper Mass and Energy Limits in FT/ICR Mass Spectrometry: Design and Testing of New Ion Traps. Prof. 36th Amer. Soc. Mass Spectrom. Conf. Mass Spectrom. & Allied Topics, 1988: p. 592-593.

196. Grosshans, P.B., P.J. Shields, and A.G. Marshall, Comprehensive Theory of the Fourier Transform Ion Cyclotron Resonance Signal for All Ion Trap Geometries. Journal of Chemical Physics, 1991. 94: p. 5341-5352.

197. Kim, S., R.P. Rodgers, and A.G. Marshall, Truly "exact" mass: Elemental composition can be determined uniquely from molecular mass measurement at similar to 0.1 mDa accuracy for molecules up to similar to 500 Da. International Journal of Mass Spectrometry, 2006. 251(2-3): p. 260-265.

198. Sharma, A., et al., High-resolution transmission electron microscopy of asphaltenes: alkane-induced disorder. Energy and Fuels, 2002. 16(2): p. 490-496.

199. Standard Test Method for Determination of Asphaltenes (Heptane Insolubles) in Crude Petroleum and Petroleum Products, in ASTM International, West Conshohocken, PA, www.astm.org, A.S.D.-. 00(2005), Editor. 2005.

200. Marshall, A.G., Fourier Transform Ion Cyclotron Resonance Mass Spectrometry. Spectroscopy, 1990. 5: p. 30.

201. Calemma, V., et al., Structural characterization of asphaltenes of different origins. Energy & Fuels, 1995. 9: p. 225-230.

202. Gross, J., Mass Spectrometry : A Textbook, ed. J. Gross. 2004, Berlin: Springer Verlag.

203. Roussis, S.G. and R. Proulx, Probing the molecular weight distributions of non-boiling petroleum fractions by Ag+ electrospray ionization mass spectrometry. Rapid Comm. Mass Spectrom., 2004. 18: p. 1761-1775.

204. Juyal, P., R.P. Rodgers, and A.G. Marshall, Rapid Speciation of Sulfur Species in Crude Oils by Electrospray FT-ICR Mass Spectrometry. Energy and Fuels, 2009. xx(xx): p. xxxx-xxxx.

205. Lopez, L.L., et al., Automated Strategies for Obtaining Standardized Collisionally Induced Dissociation Spectra on a Benchtop Ion Trap Mass Spectrometer. Rapid Communications in Mass

216 Spectrometry, 1999. 13: p. 663-668.

206. Marshall, A.G., et al., Characterization of petroleum by high resolution field desorptionaonization and atmospheric pressure photoionization FT-ICR mass spectrometry. Abstracts of Papers of the American Chemical Society, 2005. 229: p. U865-U865.

207. Sheu, E.Y. and D.A. Storm, eds. Asphaltenes: Fundamentals and Applications. ed. E.Y. Sheu and O.C. Mullins. 1995, Plenum Pub. Co.: New York.

208. Reynolds, J.G., Petroleum Chemistry and Refining, ed. J.G.Speight. 1998, Washington, D.C.: Taylor and Francis Publishers. Chapter 3.

209. Amorim, F.A.W., B.; Costa, A.S.; Lepri, F.G.; Vale, M.G.; and Ferreira, S.L., Determination of vanadium in petroleum and petroleum products using atomic spectrometri techniques. Talanta, 2007. 72: p. 349- 359.

210. Pearson, C.D., J.B. Green, Vanadium and Nickel Complexes in Petroleum Resid Acid, Base, and Neutral Fractions. Energy & Fuels, 1993. 7: p. 338-346.

211. Lepri, F.G.W., B.; Borges, D.L.; Silva, A.F.; Vale, M.G.; and Heitmann, U., Speciation analysis of volatile and non-volatile vanadium compounds in Brazilian crude oils using high resolution continuum source graphite furnace atomic absorption spectrometry. Analytica Chimica Acta, 2006. 558: p. 195-2000.

212. Lee, R.F., Agents which promote and stabilize water-in-oil emulsions. Spill Science and Technology Bulletin, 1999. 5(2): p. 117-126.

213. Treibs, A., Ann. Chem., 1935: p. 517.

214. Treibs, A., Ann. Chem., 1936. 172: p. 517.

215. Baker, E.W., J.W. Louda and W.L. Orr, Application of metalloporphyrin biomarkers as petroleum maturity indicators: The importance of quantitation. Organic Geochemistry, 1987. 11(4): p. 303- 309.

216. Gallegos, E.J.a.S., Padmanabhan, Mass spectrometry of geoporphyrins. Mass Spectrometry Reviews, 1985. 4(1): p. 55-85.

217. Barwise, A.J.G., Role of nickel and vanadium in petroleum

217 classification. Energy & Fuels, 1990. 4(6): p. 647-652.

218. Baker, E.W.Y., T.F.; Dickie, J.P.; Rhodes, R.E.; Clark, L.F., Mass Spectrometry of Porphyrins. II. Characterization of Petroporphyrins. Journal of the American Chemical Society, 1967. 89(14): p. 3631-3639.

219. Vaughan, G.B.T., Edmund C., and Yen, Teh Fu, Vanadium complexes and porphyrins in asphaltene, 2. The nature of highly aromatic substituted porphins and their vanadyl chelates. Chemical Geology, 1970. 6(5): p. 203-219.

220. Barwise, A.J.G.E.V.W., Separation and structure of petroporphyrins. Adv Org Geochem, 1979. 12: p. 181-192.

221. Popl, M.D., Vladimir; Sebor, Gustav; and Stejskal, Michal, Hydrocarbons and porphyrins in rock extracts. Fuel, 1978. 57(9): p. 565- 570.

222. Hajibrahim, S.K.Q., J.M.E.; G. Eglinton, Petroporphyrins V. Structurally-related porphyrin series in biutmens, shales and petroleums - Evidence from HPLC and Mass Spectrometry. Chemical Geology, 1981. 32(3-4): p. 173-188.

223. Fleischer, E.B., The structure of porphyrins and metalloporphyrins. Accounts of Chemical Research, 1970. 3(3): p. 105-112.

224. Wototwiec, S.L.-G., L.; Serebrennikova, O.V.; Czechowski, F., ID and 2D 1H NMR studies of iron(III) complexes on geoporphyrins of the deoxophylloerythroetio structural type derived from oil shale. Magnetic Resonance Chemistry, 2005. 33(1): p. 34-43.

225. Doukkali, A.S., A.; Zrineh, A.; Hamad, M.; Ferhat, M.; Barbe, J.M.; Guilard, R., Separation and identification of petroporphyrins extracted from the oil shales of Tarfaya: geochemical study. Fuel, 2002. 81(4): p. 467-472.

226. Premovic, P.I.D., D.M.; Pavlovic, M.S., Vanadium of petroleum asphaltenes and source kerogens (La Luna Formation, Venezuela): isotopic study and origin. Fuel, 2002. 81(15): p. 2009-2016.

227. Millson, M.F.M., D.S; Brown, S.R., Geochim Cosmochim Acta, 1965. 30: p. 207.

228. Spencer, W.A., Galobardes, J.A., Curtis, M.A., Rogers, L.B., Chromatographis studies of vanadium compounds from Boscan crude oil.

218 Separation Science and Technology, 1982. 17(6): p. 797-819.

229. Marquez, N., Ysambertt, F., and C. De La Cruz, Three analytical methods to isolate and characterize vanadium and nickel porphyrins from heavy crude oil. Analytica Chimica Acta, 1999. 395(3): p. 343-349.

230. Fish, R.H.K., John J.; and Wines, Brian K., Characterization and comparison of vanadyl and nickel compounds in heavy crude petroleums and asphaltenes by reversed-phase and size exclusion liquid chromatography/graphite furnace atomic absorption spectrometry. Analytical Chemistry, 1984. 56(13): p. 2452-2460.

231. Biggs, W.R.F., John C.; Brown, Rick J.; and Reynolds, John G., Characterization of vanadium compounds in selected crudes. I. Porphyrin and non-porphyrin separation. Liquid Fuels Technology, 1985. 3(4): p. 397-421.

232. Reynolds, J.G.a.B., W.R., Analysis of residuum desulfurization by size exclusion chromatography with element specific detection. Amer. Chem. Soc. Dive. Petr. Chem., 1987. 32(2): p. 398-405.

233. Frakman, Z.I., T.M.; Montgomery, D.S.; and O.P. Strausz, Nitrogen compounds in Athabasca Asphaltene : the vanadyl porphyrins. AOSTRA Journal of Research, 1988. 4(3): p. 171-179.

234. Chicarelli, M.I.M., J.R., Analysis of ancient porphyrins. Trends in Analytical Chemistry, 1987. 6(6): p. 158-164.

235. Shaw, G.J.Q., M.E.; Eglinton, G., Analysis of petroporphyrins by chemical ionization mass spectrometry. Journal of the Chemical Society, Perkin Transactions 1: Organic and Bio-Organic Chemistry 1972-1999, 1978. 12: p. 1655-1659.

236. Eckhardt, C.B.D., Lynne; Yendle, Peter W.; Eglington, Geoffrey, Multimolecular data processing and display in organic geochemistry: the evaluation of petroporphyrin GC-MS data. Organic Geochemistry, 1988. 13(4-6): p. 573-582.

237. McFadden, W.H.B., D.C; Eglington, G.; Hajlbrahim, S.K.; Nicolaides, N., Application of combined liquid chromatography/mass spectrometry (LC/MS): analysis of petroporphyrins and meibomian gland waxes, Journal of Chromatographic Science, 1979. 17(9): p. 518-522.

238. Johnson, J.V.B., Edward D.; and Yost, Richard A.; Quirke, J. Martin E.; Cuesta, Lilia L., Tandem mass spectrometry for

219 characterization of high-carbon number geoporphyrins. Analytical Chemistry, 1986. 58(7): p. 1325-1329.

239. Grigsby, R.D.a.G., J.B., High-resolution mass spectrometric analysis of a vanadyl porphyrin fraction isolated from the >700 degree C resid of Cerro Negro heavy petroleum. Energy & Fuels, 1997. 11(3): p. 602-609.

240. Xu, H.Y., Daoyong; and Que, Guohe, Characterization of petroporphyrins in Gudao reside by ultraviolet-visible spectrophotometry and laser desorption ionization-time of flight mass spectrometry. Fuel, 2005. 84(6): p. 647-652.

241. Van Berkel, G.J.Q., Miguel A.; Quirke, J. Martin E., Geoporphyrin analysis using electrospray ionization-mass spectrometry. Energy & Fuels, 1993. 7(3): p. 411-419.

242. Qian, K.M., A.S.; Edwards, K.E.; and Ferrughelli, D.T., Observation of vanadyl porphyrins and sulfur-containing vanadyl porphyrins in a petroleum asphaltene by atmospheric pressure photoionization Fourier transform mass spectrometry. Rapid Communications in Mass Spectrometry, 2008. 22: p. 2153-2160.

243. D. Vazquez, G.A.M., Identification and measurement of petroleum precipitates. Journal of Petroleum Science and Engineering, 2000. 26: p. 49-55.

244. Lee, H.-N. and A.G. Marshall, Theoretical Maximal Precision for Mass-to-Charge Ratio, Amplitude, and Width Measurement in Ion- Counting Mass Analyzers. Analytical Chemistry, 2000. 72: p. 2256-2260.

220 BIOGRAPHICAL SKETCH

November 7, 1975….………………………..……….………Born. Warren, Ohio

May 5, 2005………………………………….Bachelors of Science, Chemistry The University of Tampa, Tampa, Florida

August 21, 2009………………………………….Ph.D., Analytical Chemistry Florida State University

PUBLICATIONS

McKenna, A.M.; Purcell, J.M.; Rodgers, R.P.; Marshall, A.G., Identification of Vanadyl Porphyrins in a Heavy Crude Oil and Raw Asphaltene by Atmospheric Pressure Photoionization Fourier Transform Ion Cyclotron Resonance (FT-ICR) Mass Spectrometry, Energy & Fuels. 2009, 23 (4), 2122-2128

McKenna, A.M.; Purcell, J.M.; Rodgers, R.P.; Marshall, A.G., Part I. Exhaustive Compositional Analysis of Athabasca Bitumen HVGO Distillates by FT-ICR Mass Spectrometry: The First Detailed Test of the Boduszynski Model. Submitted to Energy & Fuels, August 2009.

McKenna, A.M.;Glaser, P.B.; Rodgers, R.P.; Marshall, A.G., Part II. The Composition of Heavy Petroleum: Evolution of the Boduszynski Model to the Upper Limit of Distillation by UItrahigh Resolution FT-ICR Mass Spectrometry. Submitted to Energy & Fuels, August 2009.

McKenna, A.M.; Rodgers, R.P.; Marshall, A.G., The True Molecular Characterization of Asphaltenes. Part III. Molecular Weight and Distillable Asphaltenes. Submitted to Energy & Fuels. August 2009.

McKenna, A.M.; Rodgers, R.P.; Marshall, A.G., The True Molecular Characterization of Asphaltenes. Part IV. The Definition of Asphaltene and Maltene Compositional Space. Submitted to Energy & Fuels. August 2009.

McKenna, A.M.; Glaser, P.B.; Rodgers, R.P.; Marshall, A.G., The Asphaltene Problem: Solution-Phase and Gas-Phase Aggregation as Detected by Mass Spectrometry. To be submitted to Energy & Fuels. September 2009

221

McKenna, A.M.; Purcell, J.M.; Rodgers, R.P.; Marshall, A.G., Optimization of Atmospheric Pressure Photoionization Nebulization Temperature for Athabasca Bitumen Distillation Cut Point Detected by FT-ICR Mass Spectrometry, Submitted to Energy & Fuels. September 2009.

Pomerantz, A.E.; Ventura, G.T.; McKenna, A.M.; Canas, J.A.; Auman, J.; Koerner, K.; Curry, D.; Nelson, R.K.; Reddy, C.M.; Rodgers, R.P.; Marshall, A.G.; Peters, K.E.; Mullins, O.C., The Geochemical Origin of a Viscosity Gradient in a Petroleum Reservoir, Submitted to Organic Geochemistry, July 2009.

Fernandez-Lima, F.A.; Becker, C.; McKenna, A.M.; Rodgers, R.P.; Marshall, A.G.; Russell, D.H., Petroleum Crude Oil Characterization Using IMS-MS and FT-ICR MS, Submitted to Analytical Chemistry, August, 2009.

Rummel, J.L.; McKenna, A.M.; Marshall, A.G.; Eyler, J.R.; Powell, D.H., The Coupling of Direct Analysis in Real Time Ionization to Fourier Transform Ion Cyclotron Resonance Mass Spectrometry for Ultrahigh Resolution Analyses, Submitted to Analytical Chemistry, August, 2009.

222