Florida State University Libraries

Transcription

1 Florida State University Libraries Electronic Theses, Treatises and Dissertations The Graduate School 2007 Petroleum Analysis by Atmospheric Pressure Photoionization Fourier Transform Ion Cyclotron Resonance Mass Spectrometry Jeremiah Michael Purcell Follow this and additional works at the FSU Digital Library. For more information, please contact

2 THE FLORIDA STATE UNIVERSITY COLLEGE OF ARTS AND SCIENCES PETROLEUM ANALYSIS BY ATMOSPHERIC PRESSURE PHOTOIONIZATION FOURIER TRANSFORM ION CYCLOTRON RESONANCE MASS SPECTROMETRY By JEREMIAH MICHAEL PURCELL 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: Spring Semester, 2007 Copyright 2007 Jeremiah Michael Purcell All Rights Reserved

3 The members of the Committee approve the Dissertation of Jeremiah M. Purcell defended on March 19, Alan Marshall Professor Directing Dissertation Vincent Salters Outside Committee Member William Cooper Committee Member Timothy Logan Committee Member Ryan Rodgers Committee Member Approved: Christopher Hendrickson Committee Member Joseph Schlenoff, Interim Chair, Department of Chemistry and Biochemistry Joseph Travis, Dean, College of Arts and Sciences The Office of Graduate Studies has verified and approved the above named committee members. ii

4 To Shannon Elodie Willkens Hand-in-Hand Together and Mom and Dad John Edward Purcell Mary Ann Hobbs Purcell iii

5 ACKNOWLEDGEMENTS Foremost, I owe a debt of gratitude to Alan Marshall, my advising professor. I will always be humbled by Professor Marshall s breadth of knowledge and his unique ability to communicate exquisitely either verbally or written. Alan, my hat is off, thanks. I thank Chris Hendrickson and Ryan Rodgers. I was fortunate to be exposed to leading scientist in the field of mass spectrometry. Chris and Ryan individually are accomplished analytical chemist but the combination of their abilities is unparalleled. Thanks Chris and Ryan. I want to thank Mark Emmett. In my excursion to find buckybowls, I had a steep learning curve in the field of liquid chromatography. Mark s vast knowledge was irreplaceable. A big Texas thank you Mark. I thank John Quinn. There is always someone in a group (and in the US Air Force) who is the go-to person to find the answer. I can t count how many times John pointed this grad student in the right direction. Countless thanks John. I want to also thank all the Marshall group members, past and present. It has been a privilege to work with you and I look forward to future endeavors. I want to thank all my family. A special thanks to Shannon, Emalee and Sarah. To Emalee and Sarah, two wonderful daughters who I know will achieve great things, I love you. To Shannon, the love of my life and a life companion, I love you completely. iv

6 TABLE OF CONTENTS LIST OF TABLES...IX LIST OF FIGURES... X ABSTRACT...XVII CHAPTER 1. INTRODUCTION... 1 Fourier Transform Ion Cyclotron Resonance Mass Spectrometry... 1 Key Scientific Events... 1 Ion Cyclotron Motion Theory... 2 Perturbation of Cyclotron Motion Tesla FT-ICR Mass Spectrometer at the National High Magnetic Field Laboratory (NHMFL)... 4 Atmospheric Pressure Photoionization... 6 Photon Ionization... 6 Photoionization Pathways... 7 Thermo Fisher Scientific APPI Source... 9 Speciation of Non-polar Petroleum Compounds Kendrick Data Analysis and Double Bond Equivalents Calculations12 CHAPTER 2. ATMOSPHERIC PRESSURE PHOTOIONIZATION FOURIER TRANSFORM ION CYCLOTRON RESONANCE MASS SPECTROMETRY FOR COMPLEX MIXTURE ANALYSIS Summary Introduction Experimental Methods Solvents and Compounds Crude Oil Results And Discussion Model Compounds The Nitrogen Rule Complex Mixture Analysis Negative Ions Mass accuracy v

7 Conclusions CHAPTER 3. COMPARISON OF ATMOSPHERIC PRESSURE PHOTOIONIZATION AND ELECTROSPRAY IONIZATION OF CRUDE OIL NITROGEN CONTAINING AROMATICS BY FOURIER TRANSFORM ION CYCLOTRON RESONANCE MASS SPECTROMETRY Summary Introduction Experimental Methods South American Crude Oil Nitrogen Class Compounds ESI Experimental Conditions Results and Discussion Nitrogen Compounds Nitrogen Class Speciation Ion Fragmentation Conclusions CHAPTER 4. ATMOSPHERIC PRESSURE PHOTOIONIZATION PROTON TRANSFER FOR COMPLEX ORGANIC MIXTURES INVESTIGATED BY FOURIER TRANSFORM ION CYCLOTRON RESONANCE MASS SPECTROMETRY Summary Introduction Experimental Methods Solvents and Compounds Crude Oil Results And Discussion Nitrogen Class Compounds Bitumen Distillation Cuts Deuteration versus Protonation Negative and Positive Ion Class Distribution Comparison Conclusions CHAPTER 5. SULFUR SPECIATION OF PETROLEUM BY ATMOSPHERIC PRESSURE PHOTOIONIZATION FOURIER vi

8 TRANSFORM ION CYCLOTRON RESONANCE MASS SPECTROMETRY Summary Introduction Experimental Methods Middle East Crude Oil Results And Discussion Middle East Crude Analysis Conclusions CHAPTER 6. LIMITATIONS OF AROMATIC SULFUR CHEMICAL DERIVATIZATION ANALYSIS OF PETROLEUM BY ESI AND APPI FOURIER TRANSFORM ION CYCLOTRON RESONANCE MASS SPECTROMETRY Summary Introduction Experimental Methods Vacuum Bottom Residue SARA Fractionation CHNOS Analysis Results And Discussion APPI FT-ICR MS Raw Vacuum Bottom Residue Raw Methylated Vacuum Bottom Residue Saturate and Aromatic Fraction of the Vacuum Bottom Residue Conclusions CHAPTER 7. CONCLUSIONS AND APPI FT-ICR MS APPLICATION AND COLLABORATION WITH THE INSTITUTE OF PETROLEUM AT FRANCE; A REAL WORLD APPLICATION Assessment of APPI Technology APPI FT-ICR MS Applied to Current Petrochemical Challenges Introduction Residue Sample Overview Asphaltene Analysis vii

9 Overall Conclusion APPENDIX A. CARBON CLUSTER STRUCTURAL CHARACTERIZATION BY GAS PHASE ION-MOLECULE REACTION IN AN FT-ICR MASS SPECTROMETER Fullerene Introduction Instrumentation Cluster Source Cluster Source Coupled to Existing 9.4 T FT-ICR Mass Spectrometer Retarding Potential Study Cluster Spectra Mass Range In-Cell Gas-Phase Ion-Molecule Reactions Conclusions APPENDIX B. REACTION OF HYDROGEN GAS WITH C60 AT ELEVATED PRESSURE AND TEMPERATURE: HYDROGENATION AND CAGE FRAGMENTATION Summary Introduction Experimental Section Results and Discussion APPI FT-ICR MS of Hydrogenated Samples APPI versus FD FT-ICR MS Low Mass Ions Elemental Composition of Hydrofullerene Mixtures Conclusion REFERENCES BIOGRAPHICAL SKETCH viii

10 LIST OF TABLES Table 1.1. Example of a Crude Oil Homologous Series Categorized by Class, DBE and Carbon Number Distribution. DBE and Molecular Formulas Correspond to the Ion as Opposed to the Neutral Compound Table 2.1 Elemental Compositions Assigned to Peaks in the Negative-ion APPI FT- ICR Mass Spectral Segment Shown in Figure 2.5. All elemental compositions are for the deprotonated molecule, (M-H) -. Note that measured and calculated masses are uniformly identical to six places, and differ only at the sub-ppm level (shown in red) Table 3.1. List of DBE 9 positive-ion N1-class APPI species (M + ) and the DBE 9.5 negative ion N1-class ESI species (M-H) -. Each molecular formula (and the DBE value computed from it (Eq. 1.19)) is for the stated ion, not its neutral precursor Table 4.1. Positive-ion APPI FT-ICR MS ion relative abundances for the five aromatic nitrogen compounds of Figure 4.1. Parenthetical values show the percentages of M +, [M + H] +, and [M + D] + for each compound Table 7.1. Total Elemental Peak Assignment and Root-Mean-Square Mass Error (mass error, difference between experimentally measured mass and the exact mass corresponding to the elemental composition assigned to that mass spectral peak). The asphaltene alpha-numeric designators correspond to Figure Table A.1. Spectra Instrument Parameters Table A.2. Percent Relative Abundance of Cluster Reaction Products ix

11 LIST OF FIGURES Figure Tesla FT-ICR mass spectrometer. Graphical presentation of the differentially pumped vacuum chambers and ion optics. Differential pumping achieves ultra low pressure (10-10 Torr) at the ICR cell. At atmosphere pressure, ions are introduced through a heated metal capillary to the first radio frequency (rf) octopole ion guide for external accumulation, transferred to the middle octopole (collisionally cooled with helium) and then pulsed to the ICR cell Figure 1.2. Two-dimensional layout of the APPI ion source. For simplicity, the vacuum UV lamp is drawn along the z axis with the heated metal capillary. In practice, the lamp is along the x axis so that the three assemblies are mutually orthogonal Figure 2.1. Class distribution from an APPI positive-ion FT-ICR mass spectrum of Middle East crude oil Figure 2.2. APPI positive-ion FT-ICR mass spectra of 30 µm naphtho[2,3-a]pyrene in toluene (top) and hexanes (bottom). Top: The insets show the two kinds of ions formed in the APPI source region; protonated molecules and radical molecular cations. Nine acquisitions were summed with an external ion accumulation of 5 seconds each, resulting in a SNR of Bottom: The insets show the reduction in formation of the protonated molecule in the absence of a dopant. Nine acquisitions were summed with an external ion accumulation of 10 seconds each Figure 2.3. APPI FT-ICR mass scale-expanded segment for a South American crude oil. The mass doublets document the requirement for ultrahigh mass resolving power with an APPI source for complex mixture analysis. The 3.4 mda mass doublet corresponds to species differing by C3 vs. SH4 and the 4.5 mda mass doublet to 12 CH vs. 13 C. Two hundred acquisitions were summed with an ion accumulation of 3 seconds each. Starred peaks were assigned to elemental compositions not shown in the Figure Figure 2.4. APPI FT-ICR mass scale-expanded segment of a high-sulfur Middle East crude oil, showing a very close 1.1 mda mass doublet, 12 C4 vs. SH3 13 C. Two hundred acquisitions were summed with an external ion accumulation of 5 seconds each. Starred peaks were assigned to elemental compositions not shown in the Figure Figure 2.5. Negative ion APPI FT-ICR broadband mass spectrum of a South American crude oil. Bottom: Across a 400 Dalton mass window, 12,449 unique elemental compositions (a new record for a single mass spectrum) were assigned (> 99% deprotonated molecules), based on an average mass resolving power of ~400,000 and an rms mass accuracy of 260 parts per billon. Top: At a S/N ratio > 8 σ of baseline noise, there are 63 spectral peaks of nominal mass 377 Da of which unique elemental compositions could be assigned to 62 (see Table 2.1) x

12 Figure 2.6. APPI FT-ICR mass spectral peak magnitude vs. mass error (measured mass minus the exact mass for the assigned chemical formula) for the elemental compositions assigned to 12,449 spectral peaks from Figure 2.5. Ninety percent of the peaks exhibit less than 500 ppb mass error. As predicted, 53 mass accuracy increases with increasing mass spectral S/N ratio. 29 Figure 2.7. Mass error distribution for the 12,449 spectral peaks from Figure 2.5. Each bar represents the number of assigned masses within a 50 ppb "bin" mass error range. At half-maximum height, the errors span a range of ±200 ppb Figure 3.1. ESI and APPI ionization pathways for acridine and carbazole. For ESI, negative-ion and positive-ion spectra would be necessary to detect both species. However, both compounds yield APPI positive ions. Double bond equivalents (DBE) are calculated from Eq for each ion Figure 3.2. Nitrogen class compounds. Carbazole and 7H-dibenzo[c,g]carbazole are pyrrolic (acidic), whereas acridine and 7,9-dimethylbenz[c]acridine are pyridinic (basic) species. Ellipticine, with two nitrogen heteroatoms, has both pyrrolic and pyridinic moieties Figure 3.3. Negative-ion and positive-ion ESI FT-ICR mass spectra of representative nitrogen-class compounds. For both spectra, an equimolar solution was electrosprayed. The pyrrolic species were ionized by negativeion ESI and the pyridinic species by positive-ion ESI. Ellipticine was detected in both negative-ion and positive-ion spectra Figure 3.4. Negative-ion and positive-ion APPI spectra of representative nitrogenclass compounds. For both spectra, an equimolar solution was infused into the APPI source. The pyrrolic species were detected in the negative-ion APPI spectrum, and all five compounds were detected in the positive-ion APPI spectrum Figure 3.5. Positive-ion APPI FT-ICR broadband mass spectrum of a South American crude oil. Both radical molecular ions and protonated compounds are formed in the APPI source. The mass scale-expanded inset shows two common spectral peak doublets for APPI. Naphtho[2,3-a]pyrene was added to the petroleum sample to test for possible fragmentation. No fragment ions were observed Figure 3.6. ESI (positive-ion and negative-ion) and APPI (positive-ion) DBE distributions for the N1 class from the petroleum sample. The ion DBE is calculated from Equation The total relative ion abundance for each DBE is plotted on the y-axis. For ESI, protonation or deprotonation yield ions of half-integer DBE values. For APPI, radical molecular ions yield integer DBE values. Note that positive-ion APPI can distinguish pyridinic (M+H) + from pyrrolic (M + ) nitrogen ions based on their respective integer and half-integer DBE values Figure 3.7. Schematic representation (not to scale) of the Heated Metal Capillary (HMC), tube lens, and skimmer housed in the first differentially pumped stage of the mass spectrometer. Ions are transferred through the HMC and are focused by the tube lens before reaching the skimmer conductance limit. The gas dynamics in the tube lens/skimmer region can cause fragmentation xi

13 but fragmentation can be negated by appropriate pressure and voltage adjustments (see text) Figure 3.8. Positive-ion APPI FT-ICR mass spectra of naphtho[2,3-a]pyrene (C24H14, neutral monoisotopic mass, Da) for various choices of tube lens voltage and skimmer region pressure. Fragmentation is evident at higher tube lens potential and/or lower pressure. At a tube lens potential of 200 V DC and a skimmer region pressure of 2.1 Torr, no fragmentation was observed Figure 4.1. Five aromatic nitrogen compounds chosen to model petroleum acidic and/or basic compounds. Five-membered ring nitrogen structures are acidic and six membered ring nitrogen species are basic Figure 4.2. Negative ion APPI FT-ICR mass spectrum of an equimolar solution of the model compounds of Figure 4.1 in deuterated toluene. Only the acidic compounds containing a pyrrole ring are deprotonated to yield [M - H] - ions, none of which contained deuterium Figure 4.3. Positive ion APP FT-ICR mass spectrum of an equimolar solution of the model compounds of Figure 4.1 in deuterated toluene. All five compounds yielded positive molecular (M + ) or quasimolecular ([M - H] - ) ions. The compounds containing a six-membered pyridinic ring are sufficiently basic to readily protonate (or deuterate) (along with ~1% of radical molecular radical cations), whereas the more acidic compounds containing a fivemembered pyrrolic ring form molecular radical cations, and <1% protonation (or deuteration). For the even-electron species, the extent of deuteration was ~15% for acridine (see the mass scale-expanded inset spectrum), ~10% for ellipticine, and ~14% for 7,9-dimethylbenz[c]acridine. Also, at nominal mass 268 (right mass scale-expanded inset), 7H-dibenzo[c,g]carbazole exhibits slight hydrogen-deuterium exchange Figure 4.4. Heteroatom class distribution for a bitumen mid-range distillate positive ions. Each class represents the relative ion abundance of species which contain the stated heteroatom(s) in the assigned molecular formula. The error bars are standard deviation computed from 3 separate sample preparations and analysis Figure 4.5. Broadband APPI FT-ICR mass spectra of a bitumen mid-range distillate. The positive- and negative-ion spectra were collected without source interruption and with appropriate instrument polarity changes. Although APPI produces both molecular radical cations (M + ) as well as [M - H] - and [M + H] + ions, the N1 class positive-ion mass spectrum is dominated (~97%) by protonated compounds Figure 4.6. Positive-ion APPI FT-ICR mass scale-expanded segment of a bitumen mid-range distillate in deuterated toluene. This figure emphasizes the ultrahigh resolving power required to resolve the deuterated species in complex petroleum mixtures Figure 4.7. Heteroatom class distribution for the positive and negative ions from a bitumen mid-range distillate. Generic structures are shown for the most abundant positive and negative species. DBE is the number of rings plus double bonds, and is calculated from Eq Because only ~5% of the evenelectron N1 class ions contain deuterium, the acidic neutrals in the original xii

14 sample are likely proton donors to form the even-electron species from basic neutrals Figure 5.1. Broadband positive APPI FT-ICR mass spectra of the whole crude and its SARA fractions. The samples were analyzed at the same concentration and experimental conditions Figure 5.2. Summed relative ion abundance for heteroatom classes in the whole crude. Middle East crude oils have a high sulfur content. The graph includes those heteroatom classes above 1% relative abundance. Eight of the eleven classes contain one or more sulfur atoms Figure 5.3. Summed relative ion abundance class graphs for the SARA fractions. The saturate, aromatic, and asphaltene fractions show sulfur species most abundant. For the resins, more polar heteroatom classes are dominant Figure 5.4. Three-dimensional relative abundance contoured DBE versus carbon number plot for selected heteroatom classes of the whole crude and its SARA fractions. Carbon number is represented on the x-axis and double bond equivalents (equation 1.19) on the y-axis. The z-axis is color scaled to relative ion abundance. All plots are scaled equally Figure 6.1. Heteroatom class distribution for the raw vacuum bottom residue (not methylated). All classes ionized by ESI and APPI above 1 % relative abundance are represented. The non-polar classes, e.g., S1, S2 and HC (hydrocarbon), were not detected by ESI. APPI analysis detected both the polar and non-polar species Figure 6.2. Iso-abundant contoured DBE versus carbon number images for heteroatom species of the raw vacuum bottom residue. Relative ion abundance within the class is color scaled in the z-axis. The ESI and APPI N1S1 classes have similar carbon number distributions. However, the DBE distribution for the APPI N1S1 image extends to higher DBE. The non-polar S1 species was not detected by ESI Figure 6.3. Heteroatom class distribution for the raw methylated vacuum bottom residue. All classes ionized by ESI and APPI above 1 % relative abundance are represented. The ESI distribution exhibits a remarkable change in highest relative abundance to the S1 class. The S2, HC and O1 classes are also detected by ESI in the methylated sample. The APPI heteroatom class distribution is similar to Figure Figure 6.4. Iso-abundant contoured DBE versus carbon number images for heteroatom species of the raw methylated vacuum bottom residue. Relative ion abundance within the class is color scaled in the z-axis. The N1S1 images are similar to the images produced from the unmethylated sample (Figure 6.2). However, the S1 class images differ dramatically between ESI and APPI. The low DBE species in the ESI S1 image are absent in the APPI image Figure 6.5. APPI analyzed heteroatom class distribution for the saturates, aromatics and a solution of saturates and aromatics fractionated from the vacuum bottom residue. The saturates and aromatic solution was an equal molar concentration prepared by mixing equal volumes of equal mass/volume solutions. All classes above 1 % relative abundance are represented xiii

15 Figure 6.6. S1 DBE distribution of the saturates, aromatics, and the saturate/aromatics solutions. The calculated DBE values (equation 1.19) are for the cation (not the neutral species). Therefore, non-integer values are possible for protonated compounds [M + H] +. The analysis of the combined saturates and aromatics (equal weight/concentration) show a broad distribution which encompasses both the individual saturate and aromatic DBE distributions Figure 6.7. Iso-abundant contoured image for the S1 class of the saturates, aromatics and saturates/aromatics solutions. Relative ion abundance within the class is color scaled in the z-axis. The same trend seen in the DBE distribution graph (Figure 6.7) is represented in the images. The lower DBE species are found in the saturate fraction, higher DBE species in the aromatic fraction, and a combination of the individual DBE distribution values in the combined image Figure 7.1. Residue hydroconversion scheme. Sample designations A1, A2, A11 and A22 reflect hydroconversion in fixed bed conditions. Sample A1 and A2 were reacted at different temperature (hydrodemetalization) and further reacted to produce A11 and A22 (hydrodesulfurization). Samples B1, B2, and B3 were obtained in ebullated bed conditions at different increasing residence times Figure 7.2. Broadband APPI FT-ICR mass spectrum of the IFP aromatic sample (bottom). Zoom insets (top) identify [C30H18S3 + H] + (64 % relative abundance), [C29H18S3 13 C1 + H] + (22 % relative abundance, and [C29H18S2 13 C1 34 S1 +H] + (3 % relative abundance) Figure 7.3. Heteroatom class distribution for the IFP asphaltene sample. Forty heteroatom classes were assigned. For classes above 1 % relative abundance (top), 15 of the 17 classes contain one or more sulfur atoms. Also note the hydrocarbon class (HC) is present in low abundance Figure 7.4. Class distribution for samples A1 and A2 (Figure 7.1). Each sample is normalized to the most abundant class within its class distribution, i.e., they are mutually exclusive. Sample A1 was reacted at 380 C and sample A2 at 400 C Figure 7.5. Iso-abundant contoured DBE versus carbon number plot of the feed asphaltene and A1 sample; sulfur and hydrocarbon classes. Relative ion abundance is color scaled in the z-axis. A dashed reference line is added to enhance graph-to-graph comparison Figure 7.6. Iso-abundant contoured DBE versus carbon number plot of the feed asphaltene and A2 sample; sulfur and hydrocarbon classes. Relative ion abundance is color scaled in the z-axis. A dashed reference line is added to enhance graph-to-graph comparison Figure 7.7. Heteroatom class distribution of the final reaction products (A11 and A22) Figure 7.8. Iso-abundant contoured DBE versus Carbon number plot for the hydrocarbon classes of A11 and A22. A22 shows an increase in aromaticity (increase in carbon number-to-dbe ratio) xiv

16 Figure 7.9. Class distribution for samples B1, B2 and B3. With increasing hydroconversion time, there is a corresponding reduction in sulfur species with an increase in hydrocarbon species Figure Iso-abundant carbon number versus DBE contoured plots for the hydrocarbon classes of samples B1, B2 and B3. The plots display an overall migration of the most abundant species toward greater aromaticity. Structures a, b, and c (bottom) represent the predicted stable structures which correspond to elemental species hot spots for sample B Figure A.1. Diagram of the 5.3 Split-pair FT-ICR mass spectrometer (unshielded magnet) in its original configuration at Lucent Technologies (not to scale). The eight inch bore of the magnet is vacuum sealed and was designed with four access ports which allows trapped ion interrogation within the ICR cell Figure A.2. Ion optics lens stack. This lens stack is mounted to the source block. The first electrode (labeled source potential) is physically connected to the source block and, therefore at the same potential as the source. Note the mechanical 10 offset of the final two electrodes Figure A.3. Ion optics and differential pumping diagram. The zoom inset details the source block and static potential optics. Conductance limits 2 and 3 are held at trapping potentials during external ion accumulation Figure A.4. Interface octopole ion guide. The octopole operated at MHz and ~300 volts peak-peak amplitude with a -30 V DC offset. The overall length was ~27 inches Figure A.5. Retarding potential profile. Total ion current (y-axis) is measured on accumulator octopole rods and CL2 potential (x-axis) is varied. CL2 is the accumulator entrance lens. Source, extraction, and CL1 potentials are plotted to investigate ion kinetic energy Figure A.6. Carbon cluster broadband FT-ICR mass spectrum. Spectra was collected on the initial day of instrument operation. Additional instrument parameter are reported in Table A Figure A.7. Broadband carbon cluster mass spectrum (2600 m/z 4600). Instrument parameters favored accumulation and transfer of high m/z ions. The zoom inset shows the C278 monoisotopic peak and its isotopic variants Figure A.8. Low mass carbon cluster mass spectrum. Although this is a single acquisition, 400 laser pulses were required to accumulate this ion population Figure A.9. SWIFT isolated C28 with a 30 msec NO pulse gas event. The [C28NO] + spectral peak magnitude is 13% relative abundance. The loss of C2 ([C26] + ) spectral peak is 1 % relative abundance Figure A.10. Theoretical stable structure of fullerene C28. The four red carbon atoms are at the vertices of triplet pentagons. In this isomer form, the red atoms have sp 3 orbitals with a lone electron xv

17 Figure A.11. Double resonance SWIFT isolation of C28 with NO pulse gas. Experimental sequence spectrum A (top): SWIFT isolation C28, 30 msec NO pulse gas event, SWIFT isolation C28, 30 msec NO pulse gas event. Spectrum B (bottom) was recorded immediately (2 minutes) after spectrum A without NO pulse gas Figure A.12. Reaction of carbon cluster C50 with NO gas. The cluster (spectral peak at 600 m/z) was SWIFT isolated and reacted with a 30 msec NO gas pulse. The reaction product [C50NO] + was detected at 3 % RA. The reaction also showed a strong loss of C Figure B.1. APPI FT-ICR mass spectra of C60 samples with different degrees of hydrogenation Figure B.2. APPI FT-ICR MS from the 3.8 wt % sample. The most abundant ions are assigned Figure B.3. APPI FT-ICR mass spectra from the samples with maximum hydrogenation (5.0 wt % (top) and 5.3 wt % (bottom)) Figure B.4. Scale-expanded m/z segment, Da, for samples with different hydrogen contents Figure B.5. Carbon and hydrogen compositions obtained from APPI FT-ICR mass spectra (for mass spectral peaks with S/N > 7) Figure B.6. FT-ICR mass spectra 5.0 wt % samples: (top) FD; (bottom) APPI Figure B.7. APPI FT-ICR MS of the 5.3 wt % sample under conditions that favor higher abundance of low-mass ions. Filled triangles denote higher-abundance species xvi

18 ABSTRACT Petroleum and petroleum products are an integral part of today s society. Although petroleum is projected to be the dominant energy source for the next fifty years, the depletion of light sweet crude oil reserves has led to the refinement of heavier feedstocks. Heavier petroleum feedstocks contain higher weight percent sulfur-, nitrogenand oxygen-containing species. Not only is the combustion of these species harmful to the environment, they can also poison catalytic and hydrotreatment refining equipment. The United States Environmental Protection agency has limited allowable heteroatom weight percents in petroleum products. Moreover, sulfur is the third most abundant element in petroleum and has been regulated to parts-per-million levels and further reduction slated for the year To meet the more stringent environmental regulations, refineries are facing major challenges. Mass spectrometry has proven to be a valuable tool for the molecular speciation of petroleum. Notably, electrospray ionization Fourier transform ion cyclotron resonance (FT- ICR) mass spectrometry has proven invaluable for the speciation of the polar compounds in crude oil. This analysis has added to the understanding of specific refinery problems, e.g., solid deposition and flocculation. However, hydrocarbons and non-polar sulfur species are not accessible by ESI mass spectrometry. Atmospheric Pressure PhotoIonization (APPI) can produce ions from non-polar (and polar) species. Chapter 1 is a brief discussion of basic ICR principles, APPI pathways, instrumentation and data analysis. In Chapter 2, I describe an APPI source coupled to the in-house-built 9.4 Tesla Fourier transform ion cyclotron resonance (FT- ICR) mass spectrometer at the National High Magnetic Field Laboratory (NHMFL) in Tallahassee, Florida. This chapter highlights the complexity xvii

19 of crude oil analysis with an APPI source. The possibility of forming two ion types (protonated compounds and radical molecular ions) from one compound complicates an already complex spectrum. Model compound spectra demonstrate the necessity of ultra-high resolution mass spectrometry to resolve common mass doublets (3.4 mda, the mass difference between C 3 vs. SH 4 ; 4.5 mda, the mass difference between 12 CH and 13 C) found in petroleum spectra. Also, this report establishes the highest number of resolved (and assigned elemental formulas) spectral peaks (>12,000 peaks in a single mass spectrum and up to 63 peaks of the same nominal mass) in one mass spectrum. Although APPI is considered to be a soft ionization technique, the analyte is nebulized and heated before ion formation. On the other hand, ESI is a well established soft ionization process. Therefore, in Chapter 3, I compare ESI and APPI data from the same crude oil and also pyridinic and pyrrolic nitrogen model compounds. The chapter defines instrument parameters which can cause fragmentation (loss of H 2 ) and parameters which do not. ESI and APPI crude oil spectra yield the same elemental species, providing evidence that APPI can produce an ion population without fragmentation. A dopant (proton donor) is advantageous for APPI mass spectrometry because proton transfer reactions are enhanced. For simple mixture analysis, the proton donor is predominantly the dopant. However, for complex mixture analysis (crude oil), the solution matrix can contain species which could also participate in proton transfer reactions. In Chapter 4, I investigate the proton transfer reaction for a Canadian bitumen petroleum in deuterated toluene (C 7 D 8 ). Nitrogen class compounds are also analyzed in deuterated toluene. The dopant percent contribution to the even-electron ions (protonated and deuterated compounds) of the petroleum is ~5 %. The nitrogen model compounds exhibited a similar trend. xviii

20 Petrochemical analysis commonly employs the saturates-aromaticresins-asphaltenes (SARA) separation method. In Chapter 5, the sulfur containing compounds of a Middle East crude oil are speciated. The crude oil is additionally fractionated by the SARA method and its fractions are analyzed by APPI FT-ICR mass spectrometry. Molecular species from the whole crude oil and its fractions are compared to ascertain differences and similarities between sulfur species in the fractions. Non-polar sulfur species are not efficiently ionized by ESI. However, derivatization chemistry can methylate polycyclic aromatic sulfur species and form cations in solution with subsequent analysis by ESI mass spectrometry. In Chapter 6, the derivatized and nonderivatized samples of a petroleum vacuum bottom residue (the highest boiling point fraction of petroleum and hence, the most complex heteroatom content) are analyzed by ESI and APPI. Significant differences in the double bond equivalent values (DBE, value equal to the number of rings plus double bonds in the molecular structure calculated from the elemental formula) between the ESI and APPI analyzed sulfur species are identified. Furthermore, this report provides data that probes APPI ionization efficiency. Chapter 7 is a synopsis of the APPI technology applied to petroleum analysis. The chapter also includes a real world application of APPI FT-ICR mass spectrometry. The Institute of Petroleum at France (IFP) is interested in the development of new hydroconversion processes to upgrade vacuum bottom residue to more useful petroleum products. A substantial fraction of vacuum bottom residue is the asphaltenes; the most heteroatom-rich fraction in petroleum. The chapter presents molecular speciation from intermediate stages of a hydroconversion process; a first step in hydroconversion catalytic technology improvement. xix

21 A Ph.D. thesis may also include research outside the scope of the primary dissertation research to achieve a broader understanding of the sciences. Appendix A describes the ongoing construction and adaptation of an ion cluster source to an existing FT-ICR mass spectrometer. The primary investigator is Professor Harry Kroto, Nobel prize laureate for the discovery of fullerenes. Fullerenes are closed cage molecules consisting of 12 pentagonal and several hexagonal rings. Fullerenes with 60 carbon atoms or larger follow the isolated pentagon rule (IPR). Smaller fullerenes (< 60 carbon atoms) consist of isomers with adjoined pentagon rings. Perhaps one of the more interesting small fullerenes is C 28. The structure in part consist of four reactive carbons bonded in sp 3 orbitals located at the apex of triplet pentagons which form 4 tetrahedral vertices. The research focuses on the formation of C 28 by laser vaporization and gas phase reaction products in the ICR cell. In appendix B, the reaction products of C 60 and hydrogen at high temperature and pressure are resolved and identified. The product species formed at elevated temperature and hydrogen pressure are characterized by APPI FT-ICR mass spectrometry. Only the APPI analysis (and Field Desorption, FD) were accomplished at Florida State University and the first report (of three published reports) is presented. xx

22 CHAPTER 1. INTRODUCTION Fourier Transform Ion Cyclotron Resonance Mass Spectrometry Key Scientific Events Fourier Transform Ion Cyclotron Resonance mass spectrometry is an example of different aspects of science combining to form one analytical technique, i.e., Fourier transforms, ion cyclotron resonance, and mass spectrometry. Quoting Professor Alan Marshall from a 2001 Florida State University Alumni newsletter, In research, I encourage my students to maintain broad interests, because most new ideas don t come completely out of the blue; rather, they consist of connecting two existing ideas and/or methods from initially different fields. Fourier, a brilliant mathematician, developed a theorem subsequently named the Fourier transform. 1 The theorem proved that a waveform could be divided into the summation of sine and cosine functions and then transformed into its frequency components. In essence, changing the waveform from a time domain to a frequency domain. Ion cyclotron resonance was introduced in 1932 as a method to produce high energy ions to study collision processes. 2 Mass spectrometry was born from a need to characterize atomic isotope abundance 3 and was further advanced with the development of high resolution instrumentation. 4 Alan Marshall and Melvin Comisarow developed the mass spectrometric method that combined Fourier transforms, ion cyclotron resonance and mass spectrometry. 5 Today, FT- ICR mass spectrometry has evolved into the most powerful mass determination technique primarily because of the inherent ultra-high mass resolving power and ultra-high mass accuracy. 6 1

23 Ion Cyclotron Motion Theory The interaction of a charged particle (ion) with a spatially homogenous magnetic field is the basis of ion cyclotron motion. An ion s path moving through a magnetic field is bent in a circular motion from an applied force. The equation for the force is F = mass acceleration = qv 1.1 B 0 where q, v, and B 0 are ionic charge, velocity and magnet field strength respectively. The cross product indicates the force is perpendicular to the velocity and magnetic flux planes. The acceleration component of uniform circular motion is a 2 v r = 1.2 where v and r are velocity and radius. Substituting equation 1.2 into equation v m = qvb r Also, angular velocity (ω) is equal to v = or ωr = v r ω 1.4 Therefore, substituting equation 1.4 into 1.3 and simplifying 2

24 qb ω = 0 (S.I. units) 1.5 m and a more useful form 7 ZqB B 0 ν = = C 2 π m u Z where v c is cyclotron frequency (Hz), u is in Dalton and Z is multiples of elemental charge. From equation 1.5 and 1.6, it is clear that the cyclotron frequency is independent of the ion velocity. This precludes the need to focus the translational energy of the ions for determining their mass/charge. It is this which makes ion cyclotron resonance a valuable phenomenon for mass spectrometry. 7 Perturbation of Cyclotron Motion The interaction of the ion with the magnetic field confines the ion in an xy plane perpendicular to the magnetic flux plane (defined as the z-axis). The trapping effect of the magnetic field, however, does not confine ion motion along the z-axis. A static electric field potential (trapping potential) is applied at the ends of the ICR trap to prevent ion loss along the z-axis. The electric field perturbs the natural cyclotron motion. Furthermore, ions of the same mass to-charge (m/z) ratio enter the ICR trap with incoherent cyclotron motion and inconvenient ion cyclotron radii. A broadband radio frequency (rf) excitation is applied to two opposing plates in the ICR cell. The rf is applied as a frequency 3

25 sweep excitation (chirp) with a bandwidth equal to the ions of interest. Ions with the same m/z ratio resonate with the applied chirp at their cyclotron frequency and merge into coherent cyclotron orbits. Also, the power absorbed from the chirp excites the ion packets into detectable ion cyclotron radii. Coulombic interactions between ions and the radial component of the trapping potential perturb the ion cyclotron frequency. However, the perturbation can be corrected with a mass calibration equation 8 m z A B + 2 v v = 1.7 where terms A and B are constants derived from two (or more) known m/z values of calibrant ions. A calibration derived from calibrant ions coexisting in the ICR cell works well because the calibrant ions and analyte ions are equally perturbed. 9.4 Tesla FT-ICR Mass Spectrometer at the National High Magnetic Field Laboratory (NHMFL) Figure 1.1 is a schematic drawing of the home built 9.4 Tesla ICR mass spectrometer located at the NHMFL. This instrument was used to acquire the FT-ICR mass spectral data for chapter 2-7 and appendix B. The instrument is equipped with a passively shielded Oxford 9.4 Tesla superconducting magnet. 9, 10 The mass spectrometer is controlled by a modular ICR data system. 11, 12 Ions are produced at atmosphere pressure (e.g., ESI or APPI) and traverse the heated metal capillary to the first stage of vacuum pumping into a skimmer region. The skimmer provides a conductance limit to the second stage of differential pressure where the ions enter the first radio frequency (rf)-only octopole. In the 4

26 Front Octopole Heated Metal Capillary Quadrupole Gas Inlet Middle Octopole Transfer Octopole ICR Cell ~10-3 Torr Torr Figure Tesla FT-ICR mass spectrometer. Graphical presentation of the differentially pumped vacuum chambers and ion optics. Differential pumping achieves ultra low pressure (10-10 Torr) at the ICR cell. At atmosphere pressure, ions are introduced through a heated metal capillary to the first radio frequency (rf) octopole ion guide for external accumulation, transferred to the middle octopole (collisionally cooled with helium) and then pulsed to the ICR cell. 5

27 first octopole, ions are accumulated (1-20 sec) 13 before transfer through a quadrupole (not operated in mass-resolving mode) into a second rf-only octopole where they are collisionally cooled (10-20 ms) with helium before transfer through an rf-only octopole to a 10-cm diameter, 30-cmlong open cylindrical Penning ion trap. The octopole ion guides (1.6 mm diameter titanium rods with a 4.8 mm i.d.) are typically operated between 1.5 and 2.0 MHz and 190 < V p-p < 240 V rf amplitude. Broadband frequency-sweep excitation (~ khz at a sweep rate of 150 Hz/µsec and a 190 V peak-to-peak amplitude) applied to two opposed electrodes accelerates the ions to a detectable cyclotron orbital radius. Ion cyclotron resonant frequencies are detected from induced current on two opposed detection electrodes of the ICR trap. Multiple time-domain acquisitions are summed for each sample, Hanningapodized, and zero-filled once before fast Fourier transform and magnitude calculation. 6 Negative ion data is collected with similar parameters and appropriate instrument polarity changes. Atmospheric Pressure Photoionization Photon Ionization Photoionization can take place in the liquid or gas phase but at different energy levels, e.g., water has an ionization potential of ~12.6 ev in the gas phase but ~10.5 ev in the condensed phase. 14, 15 An early application of photoionization was detection of gas phase compounds. In the early 1960's, photon sources were used as a detection method for gas chromatography. 16, 17 Analytes eluted from the GC column where ionized, and an electrode collected the electrons to produce a corresponding signal response. These early photoionization detectors (PID) were not vacuum sealed and, thus, were problematic. 6

28 The detectors were maintained at low pressure with a vacuum pump, were prone to coating problems from column bleed and were very complex to operate. 18 In 1976, Driscoll 18 introduced a photoionization detector with a sealed UV lamp which enabled the PID to be operated at atmospheric pressure. The PID then became more functional for chromatographic detection and began to replace the less sensitive Flame Ionization Detector (FID). More recently, photoionization (with a vacuum UV lamp) has changed from a detector to an ionization source for mass spectrometers. The first reports demonstrated photoionization coupled to a mass spectrometer for the detection of hydrocarbons, ketones, alcohols and amines. 19, 20 Syage et al., 21, 22 demonstrated the application of photoionization for pharmaceutical mass spectrometry methods. Syage s method primarily produced radical molecular ions [M] + (odd electron ions) through direct photoionization of the analyte. However, Bruins et al, were the first to demonstrate dopant-assisted atmospheric pressure photoionization (APPI). 23 Dopant-assisted APPI increased ionization efficiency through proton transfer reactions which produces evenelectron ions [M + H] +. Since the introduction of dopant-assisted APPI, there has been a plethora of APPI mass spectrometry application research in the biological 24 and pharmaceutical 25 sciences. Photoionization Pathways Direct photoionization ionization can be a one step process where an electron is ejected when the absorbed photon energy is equal to or higher than the first ionization energy (IE) of the molecule. Represented as, ionization: + A + hv A + e 1.8 7

29 where A is the molecular species and hν represents photon energy greater than the IE of A. However, there is an intermediate step where the ionizable species is in an excited state and other pathways are possible that result in neutral analyte 26 ionization: photodissociation: radiative decay collisional quenching collisional quenching electron capture + A + hv A A + e 1.9 AB A + B 1.10 A A + hv 1.11 A + S A + S 1.12 A + gas A + gas 1.13 A + gas A + gas where S is the solvent and gas is any gas in the source region. There are several pathways which do not result in charged species. At atmosphere pressure and room temperature, the mean free path for a 10 Ǻ diameter molecule is ~ 9 nano-meters. Therefore, the molecule would undergo ~ 2 X collisions/second. Statistically, the relative abundance of charged species under these conditions is low. Bruins et al. demonstrated dopant assisted APPI. A dopant can be infused into the ionization source region in several ways. It can be mixed with the solvent system directly, injected into the sample flow tube or infused into the APPI source through an auxiliary port. A dopant adds to the possible reaction pathways + protonated analyte [ ] [ ] + radical ion D + A D H + A + H D + A D + A

30 where D is the dopant and A is the analyte. If dopant ion molar concentration far exceeds the analyte concentration, reactions 1.15 and 1.16 should prevail over charge quenching reactions. 27 It is evident from equations 1.15 and 1.16 that two ion types can form in the APPI source, i.e., protonated compounds and radical molecular ions. Furthermore, most APPI applications use a dopant. A common dopant is toluene, and this is fortunate for APPI analysis of petroleum because toluene is also a good solvent for petroleum products. Thermo Fisher Scientific APPI Source The APPI source was supplied by Thermo Fisher Scientific. The vaporized analyte gas stream flows orthogonally to the mass spectrometer inlet (heated metal capillary) and the Krypton vacuum UV lamp (Figure 1.2) that produces 10 ev photons. The source is mounted to a home-built adapter which interfaces the first differentially-pumped stage of the 9.4 Tesla FT-ICR mass spectrometer through a heated metal capillary. The heated metal capillary is.030 inches inside diameter and resistively heated with direct current (3-4 amperes). The source-adapter apparatus construction provides a closed area such that the nebulizer gas (CO 2 ) provides a slight positive pressure. A Harvard stainless steel syringe (8 ml) and syringe pump are utilized to deliver solution to the heated nebulizer of the APPI source. In the APPI source, solvent flow rate is µl/min, the nebulizer heater is operated at C with carbon dioxide sheath gas at 50 p.s.i., and the auxiliary gas port is plugged. 9

31 Nebulizer Heater APPI Source Vacuum UV Lamp Y Heated Metal Capillary X Atmospheric Pressure millitorr Pressure Z Figure 1.2. Two-dimensional layout of the APPI ion source. For simplicity, the vacuum UV lamp is drawn along the z axis with the heated metal capillary. In practice, the lamp is along the x axis so that the three assemblies are mutually orthogonal. 10

32 Speciation of Non-polar Petroleum Compounds Current legislative trends for ultralow-sulfur fuels necessitate a better understanding of the structure of sulfur species in petroleum to facilitate the development of better hydrodesulfurization catalysts and optimize processes conditions Furthermore, with world petroleum production shifting toward heavier heteroatom-rich crude oils, the upgrading capacity of world refineries must increase to deal with the large volume of heavy crudes. 31 Hence, detailed sulfur speciation is of paramount importance, from a refinery as well as environmental point of view. Fourier transform ion cyclotron resonance (FT-ICR) mass spectrometry significantly contributes to sulfur speciation through its ability to correctly assign class, type and carbon number, providing unambiguous molecular formulas for heteroatom-containing species from mixtures as complex as unfractionated heavy petroleum. Electrospray ionization mass spectrometry (ESI MS) has identified polar compound classes from crude oil and its associated fractions. 32, 33 However, sulfur compound classes (that are not sufficiently acidic or basic, i.e., non-polar) and hydrocarbons are not efficiently ionized by ESI. Structural elucidation of sulfur species by ESI may often require chemical derivatization for enhanced detection. In contrast to ESI, Atmospheric Pressure Photoionization (APPI) can efficiently ionize gasphase non-polar species 21, 23 (and polar species) through direct photon ionization or proton transfer (with a toluene dopant) and charge exchange reactions. Hence, APPI coupled to FT-ICR mass spectrometry can provide elemental composition of non-polar petroleum species to help mitigate environmental and refinery problems. 11

33 Kendrick Data Analysis and Double Bond Equivalents Calculations Because crude oil primarily consist of homologous series differing only by nch 2 (n is a positive integer), it is convenient to convert the experimental m/z (mass/charge) values to Kendrick mass. 34, 35 ( 14 ) Kendrick mass = m 1.17 z The Kendrick masses of a homologous series differ by exactly 14 Da and will have the same Kendrick Mass Defect (KMD). ( nominal mass - Kendrick mass) 1000 KMD = 1.18 The data set can then be sort by KMD in an Excel spreadsheet to enhance assignment of elemental composition of a homologous series. Furthermore, each homologous series can be categorized by class, double bond equivalents (DBE) and carbon number. Double Bond Equivalents = H N C Where C, H and N corresponds to the number of carbon, hydrogen and nitrogen atoms in the elemental formula. For example, a compound with an assigned elemental formula of C 39 H 57 N 1, belongs to the N 1 class and has a DBE value of 12. DBE is related to hydrogen deficiency. A fully hydrogen saturated compound has a DBE of zero. Each loss of two hydrogen atoms corresponds to a structural addition of one double bond or ring. Also, APPI forms radical molecular ions, M +, and protonated compounds [M + H] +. Calculation of the DBE for a protonated compound 12

34 (one additional hydrogen atom) can, thus, result in a non-integer value, e.g., the protonated compound C 39 H 58 N 1 calculates to a 11.5 DBE value (Table 1.1). Hence, by simply calculating the DBE value from the molecular formula of the detected ion (not the neutral species), type ion formed in the APPI source can be determined. Table 1.1. Example of a Crude Oil Homologous Series Categorized by Class, DBE and Carbon Number Distribution. DBE and Molecular Formulas Correspond to the Ion as Opposed to the Neutral Compound. Carbon IUPAC Kendrick Nominal KMD DBE Molecular Class Number Mass Mass Mass Formula C 28 H 24 N 1 S 1 N 1 S C 29 H 26 N 1 S 1 N 1 S C 30 H 28 N 1 S 1 N 1 S C 31 H 30 N 1 S 1 N 1 S C 32 H 32 N 1 S 1 N 1 S C 33 H 34 N 1 S 1 N 1 S C 34 H 36 N 1 S 1 N 1 S C 35 H 38 N 1 S 1 N 1 S C 36 H 40 N 1 S 1 N 1 S C 37 H 42 N 1 S 1 N 1 S C 38 H 44 N 1 S 1 N 1 S C 39 H 46 N 1 S 1 N 1 S 1 Table 1.1 is an example of a petroleum homologous series from a positive ion APPI mass spectrum. The half integer DBE value (17.5) indicates the type ion formed (protonated compound) in the APPI source. 13

35 CHAPTER 2. ATMOSPHERIC PRESSURE PHOTOIONIZATION FOURIER TRANSFORM ION CYCLOTRON RESONANCE MASS SPECTROMETRY FOR COMPLEX MIXTURE ANALYSIS Summary We have coupled atmospheric pressure photoionization (APPI) to a home-built 9.4 Tesla Fourier transform ion cyclotron resonance (FT-ICR) mass spectrometer. Analysis of naphtho[2,3-a]pyrene and crude oil mass spectra reveals that protonated molecules, deprotonated molecules and radical molecular ions are formed simultaneously in the ion source, thereby complicating the spectra (>12,000 peaks per mass spectrum and up to 63 peaks of the same nominal mass), and eliminating the "nitrogen rule" for nominal mass determination of number of nitrogens. Nevertheless, the ultrahigh mass resolving power and mass accuracy of FT-ICR MS enabled definitive elemental composition assignments, even for doublets as closely spaced as 1.1 mda (SH 13 3 C vs. 12 C 4 ). APPI efficiently ionizes nonpolar compounds that are unobservable by electrospray and allows nonpolar sulfur speciation of petrochemical mixtures. Introduction Advancements in atmospheric pressure ionization (API) techniques have broadened the analytical possibilities for mass spectrometry. Notably, electrospray ionization (ESI) 36 and atmospheric chemical ionization (APCI) 37 have expanded the application of mass spectrometry to the biological and pharmaceutical sciences. 38, 39 Both ESI and APCI mechanisms attach a charge to the analyte and ionization efficiency correlates with analyte polarity. Electrospray ionization of a neutral analyte typically occurs by addition or loss of a proton. The APCI charge 14

36 carrier is the product of a corona discharge, typically CH 5 + from methane, but can vary with different gas systems. These API techniques have the advantages of ready coupling with Liquid Chromatography (LC), can efficiently ionize polar species and to some extent less polar species, and are robust. However, non-polar compounds are inaccessible by ESI and can be problematic for APCI. Atmospheric Pressure PhotoIonization (APPI) was initially introduced as a soft ionization method through direct photoionization and later with dopant-assisted ionization coupled to LC, 23 and can produce ions of low-polarity and even non-polar species not efficiently ionized by ESI and APCI. Field Desorption (FD) ionization 40, 41 can also produce ions from non-polar species, but (less conveniently) at less than atmospheric pressure. An APPI ion source typically uses a vacuum ultraviolet (VUV) gas discharge lamp (e.g., krypton at ~120 nm) and can produce radical molecular ions from species with first ionization energies (IE) below the photon energy. However, some typical LC solvents (acetonitrile, methanol and/or water) deplete much of the photon flux resulting in poor analyte ionization efficiency. 26, Poor ionization efficiency by direct photoionization is problematic. Robb et al. have shown that the addition of a dopant, toluene, increases sensitivity by promoting proton transfer reactions and charge exchange reactions. 23 Robb eluted four model compounds with and without a dopant and noted a 100-fold increase in signal with a dopant for some compounds. Consequently, most APPI configurations have coupled LC to APPI with a toluene dopant. 45 The dopant is introduced directly into the solvent flow post-column or infused into a stream of hot gas through the auxiliary gas port of the APPI source heated nebulizer. Toluene has a first IE of 8.3 ev 14 (lower than that of the photons from the VUV lamp) and is typically infused at a flow rate that results in a relative molar concentration much higher than that of the analyte. The combination of 15

37 low first ionization energy and high molar concentration increases the statistical probability that an analyte ion will form because the abundant dopant radical molecular ions collide reactively with the analyte. 23, 27 There are two primary ionization products for a neutral analyte in the APPI source. With toluene dopant, if the proton affinity of the analyte is higher than the proton affinity of the benzyl radical, a protonated molecule can form. 23, 27 If the electron affinity of the toluene radical cation is higher than the electron affinity of the analyte (lower or equal ionization energy than toluene), a radical molecular ion can form. The possibility of forming two ion types from a single analyte can further complicate an already complex spectrum. High mass resolving power and mass accuracy are particularly essential for APPI MS. Greig et al. first coupled APPI to Fourier Transform Ion Cyclotron Resonance (FT-ICR) mass spectrometry, and applied it to the analysis of corticosteroids. 46 In this work, we couple APPI with a 9.4 Tesla FT-ICR mass spectrometer 10 to evaluate model compounds as well as complex mixtures. We chose petroleum crude oil for demonstration, because it contains both polar and non-polar constituents, and represents the most complex natural mixture over a relative abundance dynamic range of ~10 4. We further demonstrate that the ultrahigh mass resolving power and ultrahigh mass accuracy of FT-ICR MS 7 are essential for analysis of complex mixtures by APPI mass spectrometry. Experimental Methods Solvents and Compounds All solvents were HPLC grade and purchased from Fisher. Naphtho[2,3-a]pyrene was purchased from Sigma-Aldrich and dissolved in toluene or an isomeric mixture of hexanes to produce a 3 mm stock 16

38 solution. Two serial dilutions resulted in a 30 µm final concentration in toluene or hexane. Crude Oil Each crude oil was supplied by ExxonMobil and a sample (2.5 g) was fractionated according to the Saturates-Aromatics-Resins- Asphaltenes (SARA) method. 47 The crude oil was dissolved in 20 ml of toluene and rotary vacuum evaporated to approximately 5 ml volume to remove the volatiles. The sample was then completely dried under a stream of nitrogen gas. The dried sample was dissolved in n-hexane (25 ml) and gravity filtered through Whatman 2V grade paper to remove the asphaltenes. The filtrate (maltenes) was rotary vacuum-evaporated to 5 ml volume, absorbed onto aluminum oxide (Al 2 O 3, 5.2 g), and dried under a stream of nitrogen gas with gentle stirring. The alumina was then packed on top of neutral alumina (15.0 g) in an 11 x 300 mm open column. The aliphatics were eluted with hexane (80 ml) and the aromatics subsequently eluted with toluene (80 ml), and the resins with 80:20 toluene/methanol (80 ml). The aromatic sample was rotary vacuum-evaporated to dryness and weighed (1.13 g) and re-dissolved in toluene (11 ml) to produce a stock solution of 100 mg/ml. The stock solution was further diluted to 1 mg/ml in toluene for analysis. Results And Discussion FT-ICR MS can simultaneously analyze ions spanning several decades in mass-to-charge ratio (m/z), over a vertical dynamic range of up to 10 4, with ultrahigh resolving power and mass accuracy. 7 Time-offlight and quadrupole mass spectrometers have lower mass resolving power and usually require a liquid chromatography (LC) preseparation step prior to mass analysis of complex mixtures. Therefore, FT-ICR MS 17

39 can negate the need for a preseparation step by combining both resolution and mass measurement in one step. Although crude oil is the most compositionally complex organic mixture (over a dynamic range of 10 4 ); ESI FT-ICR MS has enabled the detailed speciation of its polar components. 33, 48 However, because the electrospray ionization mechanism involves proton transfer reactions, it selectively ionizes acids (to produce negative ions) or bases (to produce positive ions). Thus, a limitation of ESI is that it will most efficiently ionize the most acidic/basic species. Atmospheric pressure photoionization, on the other hand, can ionize both polar and non-polar compounds. Because petroleum crude oil is composed ~90% of hydrocarbons, APPI generates mass spectral signals for species not accessible by ESI or APCI. For example, Figure 2.1 shows heteroatom class relative abundances for ions in a positive-ion APPI FT-ICR mass spectrum of a Middle East crude oil. The starred classes are non-polar classes observed by APPI and are not detected by ESI, the most notable of which are those that contain sulfur. Sulfur speciation is particularly important to the petroleum refining industry due to continued regulatory decreases in the allowable sulfur levels for petroleum products. A detailed discussion of the species observed in ESI and APPI, their respective class based ionization trends, and APPI ionization mechanisms in complex petroleum matrices will be reported elsewhere. Model Compounds Figure 2.2 shows APPI-FT-ICR positive-ion mass spectra of a polycyclic aromatic hydrocarbon, naphtho[2,3-a]pyrene, dissolved in toluene (top) and a mixture of isomeric hexanes (bottom) and injected directly into the APPI source. Both spectra were collected under the same instrumental conditions except for a doubled ion accumulation 18

40 * (+) APPI FT-ICR MS * Middle East Crude Oil * (Non-Polar Classes) * * S1 S2 N1 N1 S1 HC O1 S1 S3 N1 O1 N1 S2 Figure 2.1. Class distribution from an APPI positive-ion FT-ICR mass spectrum of Middle East crude oil. 19

41 period for Figure 2.2 (bottom) to enhance the signal-to-noise ratio. Even so, the signal to noise ratio in substantially lower for the hexanes sample. The higher signal-to-noise ratio in Figure 2.2 (top) is attributed to charge exchange between the toluene dopant and the analyte. In this example, ionization efficiency is enhanced 20-fold by addition of the dopant. The mass scale-expanded insets in Figure 2.2 reveal another difference between the spectra. At nominal mass 303 Da, the protonated molecule ([C 24 H 14 + H] + ) relative abundance in hexanes (Figure 2.2, bottom) is significantly lower than in toluene (Figure 2.2, top). In Figure 2.2 (top), the signal from the 12 C 24 protonated molecule at nominal mass 303 is comparable in magnitude to that for the 13 C 12 1 C 23 radical cation. Lower resolving power mass analyzers would not resolve that doublet and the resulting broadened and asymmetrical peak would yield poor mass accuracy leading to ambiguity in peak assignment. Also in Figure 2.2 (top), three species are detected at nominal mass 304 Da. The use of a dopant (toluene) clearly increases signal-to-noise ratio, but at the cost of increased spectral complexity due to formation of two ion types (radical molecular ions and protonated molecules), thereby increasing the need for ultrahigh resolving power for reliable assignment of elemental compositions. The Nitrogen Rule At nominal mass accuracy, the "nitrogen rule" 49 states that an odd-electron ion (e.g., M + ) has an even (odd) nominal mass if it contains an even (odd) number of nitrogen atoms." Conversely, an even-electron ion (e.g., (M+H) + or (M-H) - ) has an odd (even) nominal mass if it contains an even (odd) number of nitrogen atoms. Thus, it is possible to 20

42 [C 24 H 14 ] + (+) APPI FT-ICR MS [C 23 H C] + [C24 H14 + H]+ [C 22 H C 2 ] + [C 23 H C+ H] + Naphtho[2,3-a]pyrene m/z [C 24 H H] + in Toluene m/z m/z [C 23 H C] + [C 24 H 14 + H]+ [C 24 H 14 ] + [C 23 H C+ H] + [C 22 H C 2 ] + Naphtho[2,3-a]pyrene m/z in Hexanes m/z m/z Figure 2.2. APPI positive-ion FT-ICR mass spectra of 30 µm naphtho[2,3-a]pyrene in toluene (top) and hexanes (bottom). Top: The insets show the two kinds of ions formed in the APPI source region; protonated molecules and radical molecular cations. Nine acquisitions were summed with an external ion accumulation of 5 seconds each, resulting in a SNR of Bottom: The insets show the reduction in formation of the protonated molecule in the absence of a dopant. Nine acquisitions were summed with an external ion accumulation of 10 seconds each. 21

43 determine whether the number of nitrogens is even or odd based on appearance of a mass spectral signal at even or odd nominal mass, provided that all ions are either even-electron (as in electrospray ionization or matrix-assisted laser desorption ionization) or odd-electron (as in electron ionization). However, because APPI can produce both even- and odd-electron ions in the same spectrum, the "nitrogen rule" can no longer be used to determine the number of nitrogens based on nominal mass alone. Again, ultrahigh-resolution and mass accuracy are needed to derive the correct elemental composition. Complex Mixture Analysis Formation of both protonated molecules and radical ions obviously increases the number of peaks per nominal mass. In Figure 2.2 (top), at nominal mass 303, the difference between [C 23 H C] + and [C 24 H 14 + H] + is 4.5 mda ( 13 C vs. CH), requiring mass resolving power of at least 130,000 (and proportionately higher at higher mass and/or for unequal relative abundance) for correct elemental composition assignment. In a complex mixture, FT-ICR MS routinely achieves ultrahigh mass resolving power (e.g., 400,000 resolving power at 400 Da), m/ m 50%, in which m 50% is the mass spectral peak full width at half-maximum peak height. 7 For example, Figure 2.3 shows a mass scale-expanded segment of a positive-ion APPI FT-ICR mass spectrum of a South American crude oil. A single crude oil mass spectrum can contain thousands of peaks. 50, 51 The 3.4 mda separation seen in Figure 2.3 is the mass difference between C 3 and SH 4, a common mass doublet in crude oil. That mass difference is also seen by electrospray ionization because both species are amenable to proton transfer reactions and can produce protonated molecules. Furthermore, the formation of both radical 22

44 APPI FT-ICR MS 4.5 mda South American Crude Oil [C 34 H 45 N 1 13 C 1 + H] + [C 35 H 47 N 1 ]+ [C 32 H 48 O 1 S 1 + H] mda 3.4 mda C 3 versus SH mda CH versus 13 C [C 35 H 44 O 1 + H] + * * * * * * * * * * m/z Figure 2.3. APPI FT-ICR mass scale-expanded segment for a South American crude oil. The mass doublets document the requirement for ultrahigh mass resolving power with an APPI source for complex mixture analysis. The 3.4 mda mass doublet corresponds to species differing by C3 vs. SH4 and the 4.5 mda mass doublet to 12 CH vs. 13 C. Two hundred acquisitions were summed with an ion accumulation of 3 seconds each. Starred peaks were assigned to elemental compositions not shown in the Figure. 23

45 cations and protonated molecules in the APPI source can produce an additional 4.5 mda split ( 13 C vs. CH). Figure 2.4 shows a mass scale-expanded segment of a positive-ion APPI FT-ICR mass spectrum of a Middle East crude oil known to be high in sulfur content. 52 The combination of high sulfur content and two ionization pathways produces yet another mass doublet, separated by only 1.1 mda, corresponding to the mass difference between C 4 and SH 13 3 C from the protonated molecule [C 24 H 29 N 1 S 13 1 C 1 + H] + and the radical molecular ion [C 28 H 27 N 1 ] +. That doublet is not seen in ESI spectra because radical cations are not observed by ESI for crude oil samples. Negative Ions Negative ions are formed along with positive ions in the APPI source and may be detected with appropriate instrument polarity changes. Figure 2.5 (bottom) is a negative-ion APPI FT-ICR broadband mass spectrum of a South American crude oil and further demonstrates the remarkable and necessary analytical power of FT-ICR MS. Across the spectrum, unique elemental compositions could be assigned to 12,449 spectral peaks: the most compositionally complex resolved mass spectrum to date. Within the 400 Dalton mass window, an average mass resolving power of ~400,000 and an rms mass accuracy of 260 parts per billion was achieved. An example of the complexity is shown in the mass scale expansion (Figure 2.5 top). There are 63 spectral peaks with magnitude exceeding 8 σ of rms baseline noise, and unique elemental compositions could be assigned to 62 of them (see Table 2.1) based solely on mass accuracy and Kendrick analysis. 24

46 Middle East Crude Oil 1.1 mda APPI FT-ICR MS [C 24 H 29 N 1 S 1 13 C 1 + H] + * [C 28 H 27 N] + * * 1.1 mda SH 3 13 C 1 versus C 4 * * * * * m/z Figure 2.4. APPI FT-ICR mass scale-expanded segment of a high-sulfur Middle East crude oil, showing a very close 1.1 mda mass doublet, 12 C4 vs. SH3 13 C. Two hundred acquisitions were summed with an external ion accumulation of 5 seconds each. Starred peaks were assigned to elemental compositions not shown in the Figure. 25

47 Table 2.1 Elemental Compositions Assigned to Peaks in the Negative-ion APPI FT- ICR Mass Spectral Segment Shown in Figure 2.5. All elemental compositions are for the deprotonated molecule, (M-H) -. Note that measured and calculated masses are uniformly identical to six places, and differ only at the sub-ppm level (shown in red). Peak Elemental Measured Calculated ppm No. Composition Mass Mass error 1 C 24 H 15 O 4 S C 20 H 15 O 9 S C 24 H 15 O 6 S C 21 H 19 O 6 S C 28 H 15 O 3 S C 24 H 15 O C 21 H 19 O 8 S C 28 H 15 O C 25 H 19 O 5 S C 22 H 23 O 5 S C 29 H 19 O 2 S C 25 H 19 O C 22 H 23 O 7 S C 19 H 27 O 7 S C 29 H 19 O C 26 H 23 O 4 S C 23 H 27 O 4 S C 26 H 23 O C 25 H 24 N 1 O 3 S 13 1 C C 23 H 27 O 6 S C 20 H 31 O 6 S C 30 H 23 O C 27 H 27 O 3 S C 24 H 31 O 3 S C 20 H 31 O 8 S C 29 H 24 N 1 O 13 2 C C 27 H 27 O C 26 H 28 N 1 O 2 S 13 1 C C 24 H 31 O 5 S C 21 H 35 O 5 S C 31 H 27 O C 28 H 31 O 2 S C 24 H 31 O C 23 H 32 N 1 O 4 S 13 1 C C 21 H 35 O 7 S C 30 H 28 N 1 O 13 1 C C 28 H 31 O

48 Table continued Peak Elemental Measured Calculated ppm No. Composition Mass Mass error 38 C 27 H 32 N 1 O 1 S 13 1 C C 25 H 35 O 4 S C 22 H 39 O 4 S C 29 H 35 O 1 S C 27 H 32 N 1 O 13 3 C C 25 H 35 O C 22 H 39 O 6 S C 31 H 31 N C 31 H 32 N 13 1 C C 29 H 35 O C 28 H 36 N 1 S 13 1 C C 26 H 39 O 3 S C 28 H 36 N 1 O 13 2 C C 26 H 39 O C 23 H 43 O 5 S C 30 H 39 O C 27 H 43 O 2 S C 29 H 40 N 1 O 13 1 C C 27 H 43 O C 24 H 47 O 4 S C 31 H 43 O C 28 H 47 O 1 S C 30 H 44 N 13 1 C C 28 H 47 O C 29 H 51 O Mass accuracy Figures 2.6 and 2.7 graphically demonstrate the unrivaled mass accuracy of FT-ICR MS, by showing the relation between peak magnitude and mass error (i.e., difference between experimentally measured mass and the exact mass corresponding to the elemental composition assigned to that mass spectral peak). The precision in measurement of peak position should be linearly proportional to the mass spectral peak signalto-noise ratio (SNR) and the square root of the number of data points per 27

49 63 Spectral Peaks above 8 σ 62 Spectral Peaks Assigned * Not assigned APPI FT-ICR MS 8 σ * m/z m/z Figure 2.5. Negative ion APPI FT-ICR broadband mass spectrum of a South American crude oil. Bottom: Across a 400 Dalton mass window, 12,449 unique elemental compositions (a new record for a single mass spectrum) were assigned (> 99% deprotonated molecules), based on an average mass resolving power of ~400,000 and an rms mass accuracy of 260 parts per billon. Top: At a S/N ratio > 8 σ of baseline noise, there are 63 spectral peaks of nominal mass 377 Da of which unique elemental compositions could be assigned to 62 (see Table 2.1)

50 Figure 2.6. APPI FT-ICR mass spectral peak magnitude vs. mass error (measured mass minus the exact mass for the assigned chemical formula) for the elemental compositions assigned to 12,449 spectral peaks from Figure 2.5. Ninety percent of the peaks exhibit less than 500 ppb mass error. As predicted, 53 mass accuracy increases with increasing mass spectral S/N ratio Number of Assigned Masses per Bin (-) APPI FT-ICR MS S. American Crude Oil 12,449 Assigned Masses 400 ppb 50 ppb Mass Bins Mass Error, ppm Figure 2.7. Mass error distribution for the 12,449 spectral peaks from Figure 2.5. Each bar represents the number of assigned masses within a 50 ppb "bin" mass error range. At half-maximum height, the errors span a range of ±200 ppb. 29

51 peak width. 53 Fig. 2.6 shows that the mass error does increase as peak signal-to-noise ratio decreases as expected; nevertheless, 90% of the peaks exhibit less than 500 parts-per-billion mass error. Figure 2.7 shows a more conventional mass error distribution, based on counting the number of peaks in each 50 ppb mass error "bin". The errors are Gaussian-distributed, with an rms deviation of ±200 ppb. The data in Figures 2.6 and 2.7 constitute the most definitive measures of broadband mass measurement accuracy, especially at low signal-tonoise ratio. Conclusions Atmospheric pressure photoionization is useful for analysis of lowpolarity and non-polar compounds. A dopant is typically necessary to increase sensitivity by promotion of proton transfer and charge exchange reactions. Toluene works well because it can act as a proton donor (or, the toluene radical cation, as an electron acceptor) and participate in reactions which produce both cations and anions. Protonated molecules (and deprotonated molecules) and radical molecular ions are formed simultaneously. As a result, APPI can add complexity to mass spectra. FT-ICR MS overcomes that complication. APPI FT-ICR MS is uniquely suited to analysis of complex petrochemical mixtures that naturally contain a large proportion of low-polarity or non-polar aromatic hydrocarbons, and for analysis of fullerene mixtures

52 CHAPTER 3. COMPARISON OF ATMOSPHERIC PRESSURE PHOTOIONIZATION AND ELECTROSPRAY IONIZATION OF CRUDE OIL NITROGEN CONTAINING AROMATICS BY FOURIER TRANSFORM ION CYCLOTRON RESONANCE MASS SPECTROMETRY Summary We determine the elemental compositions of aromatic nitrogen model compounds as well as a petroleum sample by Atmospheric Pressure Photoionization (APPI) and Electrospray Ionization (ESI) with a 9.4 Tesla Fourier transform ion cyclotron resonance (FT-ICR) mass spectrometer. From the double bond equivalents calculated for the nitrogen-containing ions from a petroleum sample, we can infer the aromatic core structure (pyridinic versus pyrrolic nitrogen heterocycle) based on the presence of M + (odd-electron) versus [M + H] + (evenelectron) ions. Specifically, nitrogen speciation can be determined from either a single positive-ion APPI spectrum or two ESI (positive- and negative-ion) spectra. APPI operates at comparatively higher temperature than ESI and also produces radical cations that may fragment before detection. However, APPI fragmentation can be eliminated by judicious choice of instrumental parameters. Introduction Heavy petroleum characterization by mass analysis requires the ultrahigh mass resolving power of Fourier Transform Ion Cyclotron Resonance Mass Spectrometry (FT-ICR MS) 7 to distinguish thousands of ionic species. 33, 48 Electrospray ionization (ESI) enables the analysis of the petroleum polar constituents. 32, 33 Electrospray ionization of petroleum 55 typically involves proton transfer reactions that selectively ionize acids (to produce negative [M-H] - ions) or bases (to produce positive [M+H] + ions). 56 The ionization efficiency correlates with acidity or 31

53 basicity. For example, carboxylic acids are efficiently ionized by negativeion ESI whereas basic species (e.g., pyridinic N 1 species) are efficiently ionized by positive-ion ESI. Acidic or basic species represent only a small fraction of the aromatics in a petroleum sample. 52 For the non-polar aromatic fraction of petroleum, Atmospheric Pressure Photoionization 21, 23 (APPI) can produce ions from species that are not efficiently ionized by ESI. The youngest of the soft ionization methods, dopant-assisted APPI was initially developed to interface liquid chromatography to mass spectrometry to analyze simple mixtures. The ionization technique is ideal for non-polar aromatic compounds because the photon energy (typically 10 ev) is great enough to ionize species with aromatic ring structures. However, direct photoionization is usually not efficient 45 and the addition of a dopant enhances ionization efficiency through proton transfer reactions and charge exchange reactions. Toluene is both a good dopant and an excellent solvent for crude oil. Although APPI is considered to be a soft ionization technique (i.e., produces minimal fragmentation of most analytes), it is considered less soft than ESI because the APPI heated nebulizer and the source region can reach >300 C, whereas ESI is conducted at room temperature. Furthermore, toluene dopant-assisted APPI produces radical cations 26 that can participate in further gas phase reactions. Here, we compare ESI and APPI for analysis of petroleum crude oils. The nitrogen atom in petroleum aromatic N 1 class species can reside in a five-membered ring (pyrrolic) or six-membered ring (pyridinic), and be readily protonated (pyrrolic) or deprotonated (pyridinic) by negative-ion or positive-ion ESI, respectively. 50 However, positive-ion APPI efficiently generates both pyrrolic and pyridinic nitrogen signals in a single mass spectrum. APPI can extend the characterization of the petroleome 33, 48 to include non-polar aromatic species and enhance the extent of molecular speciation for better understanding of petroleum processing and refining. 32

54 The nitrogen class species are efficiently ionized by both ESI and APPI, and therefore offer a good basis for comparison. In this work, we couple APPI and ESI to a 9.4 Tesla FT-ICR mass spectrometer 10 for detailed compositional comparisons of the ions produced from the same South American crude oil. Experimental Methods South American Crude Oil A South American Crude oil was supplied by ExxonMobil Research and Engineering Company (Annandale, NJ) and a sample (2.8 g) was fractionated according to the Saturates-Aromatics-Resins-Asphaltenes (SARA) method. 57 The SARA method produced an aromatic fraction solution of 100 mg/ml in toluene. Two serial dilutions yielded a 2 mg/ml solution that was further diluted in toluene to 1 mg/ml for APPI analysis and 1 mg/ml in toluene:methanol (1:1 v/v) for ESI analysis. Nitrogen Class Compounds Five aromatic nitrogen compounds (acridine, carbazole, 7,9- dimethylbenz[c]acridine, 7H-dibenzo[c,g]carbazole and ellipticine) were purchased from Sigma-Aldrich. Equimolar solutions (2 mm) of the five nitrogen compounds were prepared in toluene. For ESI, the 2 mm solutions were diluted by a factor of 5 in toluene to 400 µm, and subsequently equal aliquots (100 µl) of the five compound solutions were combined and diluted with methanol (1:1 v/v) to produce a final concentration of 40 µm for each compound. For APPI, equal aliquots (2 ml) of 2 mm solutions were combined to yield 400 µm (10 ml) for each compound, followed by a 1:10 dilution to 40 µm for each analyte. 33

55 ESI Experimental Conditions One end of a 50 µm i.d. fused silica tube was in-house ground to a point and used as the microelectrospray source 58 ] to produce electrosprayed positive and negative ions. General conditions were: needle voltage, 2 kv (-2 kv negative-ion electrospray); tube lens, 350 V (- 350 V negative-ion electrospray); and heated metal capillary current, 4.0 amperes. ESI negative-ion conditions were comparable to ESI positiveion conditions. Results and Discussion Nitrogen Compounds Figure 3.1 shows the possible ionization pathways for aromatic N 1 class compounds (acridine and carbazole) by APPI (positive-ion) and ESI (negative-ion and positive-ion). The two pathways involving proton transfer yield even-electron [M+H] + or [M-H] - ions with half-integer DBE values. Positive-ion APPI can also form odd-electron radical cations with integer DBE values. Five nitrogen compounds (Figure 3.2) were selected as petroleum model compounds: two pyrrolic (five-membered ring), two pyridinic (sixmembered ring), and one (ellipticine) with one of each ring type. Figure 3.3 shows the positive-ion (bottom) and negative-ion (top) ESI spectra of an equimolar solution of the nitrogen compounds. The pyrrolic nitrogen species deprotonate to form [M-H] - ions, whereas the pyridinic species protonate to form [M+H] + ions. 34

56 Acridine Positive APPI + N Positive ESI N + H 9.5 DBE 10 DBE - Carbazole Negative ESI N -H 9.5 DBE N H +. H 9 DBE Positive APPI N 9 DBE H Figure 3.1. ESI and APPI ionization pathways for acridine and carbazole. For ESI, negative-ion and positive-ion spectra would be necessary to detect both species. However, both compounds yield APPI positive ions. Double bond equivalents (DBE) are calculated from Eq for each ion. CH 3 H 3 C N Acridine C 13 H 9 N MW DBE N H Carbazole C 12 H 9 N MW DBE N C H 3 CH 3 N H Elipticine C 17 H 14 N 2 MW DBE N 7,9-dimethylbenz[c]acridine C 19 H 15 N MW DBE H N 7H-Dibenzo[c,g]carbazole C 20 H 13 N MW DBE Figure 3.2. Nitrogen class compounds. Carbazole and 7H-dibenzo[c,g]carbazole are pyrrolic (acidic), whereas acridine and 7,9-dimethylbenz[c]acridine are pyridinic (basic) species. Ellipticine, with two nitrogen heteroatoms, has both pyrrolic and pyridinic moieties. 35

57 Figure 3.4 shows positive-ion (bottom) and negative-ion (top) APPI spectra of an equimolar solution of the nitrogen compounds. Both positive and negative ions are formed simultaneously in the source, enabling detection of either positive or negative ions by appropriate instrument polarity changes and no source interruption. The APPI negative-ion spectrum shows the deprotonated pyrrolic species. However, in the APPI positive-ion spectrum, all five nitrogen compounds are detected simultaneously. The pyrrolic species form radical molecular ions and the pyridinic species form protonated compounds. The pyrrolic compounds (carbazole and 7H-dibenzo[c,g]carbazole) exhibit different ionization efficiency (and ion type) in negative-ion and positive-ion APPI mass spectra. Carbazole (7.6 ev ionization energy) 59 forms radical molecular cations less efficiently than the more extensively conjugated 7H-dibenzo[c,g]carbazole (7.1 ev ionization energy). 59 In contrast, negative-ion APPI generates much more similar [M-H] - ion abundances for both carbazoles (Figure 3.4 top). Nitrogen Class Speciation Figure 3.5 is the positive-ion APPI broadband mass spectrum of a South American crude oil, including various closely-spaced mass doublets. First, a 3.4 mda separation (zoom inset in Figure 3.5) results from two compounds that differ in elemental composition by C 3 vs. SH 4, a common mass doublet from crude oil. That mass difference is also seen by electrospray ionization because both species can produce protonated compounds. 36

58 A H N - Negative ESI -H N H -H - N CH 3 -H - C H 3 N H m/z B Positive ESI CH 3 H C 3 + H N + N CH 3 + H + C H 3 N H N + H m/z Figure 3.3. Negative-ion and positive-ion ESI FT-ICR mass spectra of representative nitrogen-class compounds. For both spectra, an equimolar solution was electrosprayed. The pyrrolic species were ionized by negative-ion ESI and the pyridinic species by positive-ion ESI. Ellipticine was detected in both negative-ion and positive-ion spectra. 37

59 A Negative APPI N C H 3 CH 3 N H -H - H N -H - N H -H m/z B Positive APPI CH 3 H C 3 + H + H N + N N + H + N CH 3 + H + + C H 3 N H N H m/z Figure 3.4. Negative-ion and positive-ion APPI spectra of representative nitrogenclass compounds. For both spectra, an equimolar solution was infused into the APPI source. The pyrrolic species were detected in the negative-ion APPI spectrum, and all five compounds were detected in the positive-ion APPI spectrum

60 Second, APPI can produce an additional 4.5 mda split ( 13 C vs. 12 CH) doublet not seen by ESI, due to formation of both a radical molecular ion (M + ) and protonated molecule ([M + H] + ) from the same neutral precursor. Naptho[2,3-a]pyrene was added as a standard to test for fragmentation (see below), i.e, no spectral peaks corresponding to loss of one or two hydrogen atoms from naptho[2,3-a]pyrene were observed. The spectral peaks from the broadband spectrum were assigned unique molecular formulas based on accurate mass measurement for homologous series. 34 Similar broadband spectra were acquired (data not shown) for positive- and negative-ion ESI. The N 1 class ions were sorted by DBE and carbon number. Figure 3.6 is the DBE distribution for the APPI positive-ion nitrogen class species and the ESI negative- and positive-ion nitrogen classes. The DBE values in Figure 3.6 were calculated from Eq for the ion molecular formulas. Accordingly, half-integer DBE values result for even-electron (protonated or deprotonated compound) ions, [M+H] + or [M- H] -, whereas integer DBE values result for odd-electron ions (e.g., radical cations, M + ). Hence, the DBE value calculated from Eq for any N 1 -class aromatic compound readily distinguishes even-electron (pyridinic, half-integer DBE) from odd-electron (pyrrolic, integer DBE) ions produced by APPI. From the DBE distribution (Figure 3.6 top), it is clear that APPI produces both protonated compounds (half integer DBE) and radical molecular ions (integer DBE). Interestingly, the radical molecular ions begin at DBE 9, the same DBE threshold observed for negative evenelectron [M-H] - ions. In fact, if the negative-ion ESI DBE distributions is shifted one-half DBE lower (to compensate for the one-proton difference 39

61 4.5 mda c 3.4 mda b a b c [C 31 H 29 N 1 13 C 1 ] + [C 32 H 29 N 1 + H] + [C 29 H 33 N 1 S 1 + H] + a m/z Naphtho[2,3-a]pyrene Internal Standard m/z Figure 3.5. Positive-ion APPI FT-ICR broadband mass spectrum of a South American crude oil. Both radical molecular ions and protonated compounds are formed in the APPI source. The mass scale-expanded inset shows two common spectral peak doublets for APPI. Naphtho[2,3-a]pyrene was added to the petroleum sample to test for possible fragmentation. No fragment ions were observed. 40

62 in mass between [M-H] - and M + ), then the negative- and positive-ion ESI DBE distributions are essentially the same as the combined [M+H] + and M + distribution from APPI. Table 3.1 lists the elemental compositions assigned to the APPI positive N 1 -class ions (DBE 9) and ESI negative N 1 -class ions (DBE 9.5). Note that the neutral molecular formulas are the same for both kinds of ions. The same behavior is seen throughout the nitrogen class DBE distributions: i.e, the neutral elemental compositions inferred from the deprotonated (negative-ion) ESI ions are also seen for the positive-ion APPI ions at one-half integer lower DBE. Furthermore, the relative DBE abundances (APPI DBE 9 vs. ESI DBE 9.5, Figure 3.6) agrees with the nitrogen class compound spectra (Figures 3.3 and 3.4): i.e., a carbazole type core structure compound will form a radical molecular ion (APPI positive-ion, 9 DBE) less efficiently (Figure 3.4 bottom) than in negativeion ESI (9.5 DBE) (Figure 3.3 top). Moreover, as the DBE increases for the pyrrolic species (which form radical molecular ions, integer DBE) in the positive-ion APPI DBE distribution, the relative DBE abundance also increases in agreement with the nitrogen compound spectra. Ion Fragmentation Ion dissociation is more prevalent in APPI than ESI due to hot gases and radical ion formation. Fragmentation can be minimized (or eliminated) by proper choice of source pressure and tube lens voltage. The APPI source interfaces with the first stage of the mass spectrometer through a heated metal capillary (HMC) mounted to a homebuilt 41

63 APPI Negative ESI Positive ESI DBE Figure 3.6. ESI (positive-ion and negative-ion) and APPI (positive-ion) DBE distributions for the N1 class from the petroleum sample. The ion DBE is calculated from Equation The total relative ion abundance for each DBE is plotted on the y-axis. For ESI, protonation or deprotonation yield ions of half-integer DBE values. For APPI, radical molecular ions yield integer DBE values. Note that positive-ion APPI can distinguish pyridinic (M+H) + from pyrrolic (M + ) nitrogen ions based on their respective integer and half-integer DBE values. 42

64 Table 3.1. List of DBE 9 positive-ion N1-class APPI species (M + ) and the DBE 9.5 negative ion N1-class ESI species (M-H) -. Each molecular formula (and the DBE value computed from it (Eq. 1.19)) is for the stated ion, not its neutral precursor. Experimental Ion DBE Calculated ppm m/z Formula Mass error APPI [M] C 33 H 51 N C 34 H 53 N C 35 H 55 N C 36 H 57 N C 37 H 59 N ESI negative [M - H] C 33 H 50 N C 34 H 52 N C 35 H 54 N C 36 H 56 N C 37 H 58 N adapter (Figure 3.7). At the exit of the HMC, an open-ended cylindrical tube (tube lens) focuses the ion flux in front of a conical shaped skimmer. The skimmer is a pressure conductance limit (1 mm orifice) between the first stage of the mass spectrometer and the second stage that houses the first octopole ion guide/trap. A convectron gauge senses the pressure in the first stage (~2.1 Torr for ESI conditions) and a direct current potential (350 V for ESI) is applied to the tube lens. Depending on pressure in the tube lens and the skimmer region, ion fragmentation can ensue. 60, 61 The APPI source produces hot gas at the inlet of the HMC, resulting in elevated gas temperature in the first stage (skimmer region) of the mass spectrometer and reduced pressure. Furthermore, tube lens voltage >200 V in combination with increased temperature for APPI can cause significant fragmentation. 43

65 Atmospheric pressure ~ 2 Torr 10-3 Torr Heated Metal Capillary Tube Lens Skimmer Not drawn to scale DC Voltage Figure 3.7. Schematic representation (not to scale) of the Heated Metal Capillary (HMC), tube lens, and skimmer housed in the first differentially pumped stage of the mass spectrometer. Ions are transferred through the HMC and are focused by the tube lens before reaching the skimmer conductance limit. The gas dynamics in the tube lens/skimmer region can cause fragmentation but fragmentation can be negated by appropriate pressure and voltage adjustments (see text). 44

66 The mass spectra of naphtho[2,3,a]pyrene (C 24 H 14 ) in Figure 3.8 at different skimmer region pressure and tube lens voltage illustrate fragmentation. In panels A and B, the tube lens voltage is held constant and the skimmer region pressure is lowered from 2.1 Torr (A) to 1.8 Torr (B) by partial valve closure at the vacuum pump. At 1.8 Torr, a fragmentation threshold is reached. The mass scale-expanded inset (B) shows formation of [C 24 H 12 ] +, a fragment formed by the loss of two hydrogen atoms. In panels C and D of Fig. 3.8, the skimmer region pressure is maintained at 2.1 Torr and the tube lens voltage is varied. At increased tube lens voltage (D), the loss of one hydrogen atom from the monoisotopic precursor, [C 24 H 13 ] + at mass 301 Da, is also observed. Furthermore, under conditions at which fragmentation occurs (B, C, and D in Fig. 3.8), the magnitude (zoom inset data not shown) of the [C 24 H 14 +H] + spectral peak relative to that of [C 24 H 14 ] + decreases as fragmentation increases. All other spectra in this work were acquired under Figure 3.8, panel A conditions. Conclusions Positive-ion APPI spectra yield the same information as combined positive- and negative-ion ESI spectra for N 1 class aromatics. Also, ESI and APPI mass analysis of a petroleum sample can yield aromatic core structural information. Although pyridinic and pyrrolic species ionization efficiency depends on ionization method, APPI can generate positive ions for both pyridinic and pyrrolic compounds, albeit at different efficiency. The DBE values for the APPI positive-ion species differ by 0.5 for odd- and even-electron ions, and can distinguish pyrrolic from pyridinic nitrogen class species in a petroleum sample. 45

67 Figure 3.8. Positive-ion APPI FT-ICR mass spectra of naphtho[2,3-a]pyrene (C24H14, neutral monoisotopic mass, Da) for various choices of tube lens voltage and skimmer region pressure. Fragmentation is evident at higher tube lens potential and/or lower pressure. At a tube lens potential of 200 V DC and a skimmer region pressure of 2.1 Torr, no fragmentation was observed. 46

68 APPI increases the thermal energy of analyte and atmosphere gases in the source region. The increase in thermal energy changes the gas dynamics of the mass spectrometer and can induce fragmentation. Nevertheless, the fragmentation can be monitored and negated by proper choice of instrumental parameters. In this work, the good agreement in speciation of the aromatic nitrogen class species between a proven soft ionization technique (ESI) and APPI suggest that APPI can be conducted so as to virtually eliminate fragmentation. 47

69 CHAPTER 4. ATMOSPHERIC PRESSURE PHOTOIONIZATION PROTON TRANSFER FOR COMPLEX ORGANIC MIXTURES INVESTIGATED BY FOURIER TRANSFORM ION CYCLOTRON RESONANCE MASS SPECTROMETRY Summary To further clarify the extent and mechanism for proton transfer in Atmospheric Pressure PhotoIonization (APPI), we employ ultrahighresolution FT-ICR mass analysis to identify M +, [M + H] +, [M - H] - and [M + D] + species in toluene or perdeuterotoluene for an equimolar mixture of five pyrollic and pyridinic nitrogen heterocylic model compounds, as well as for a complex organic mixture (Canadian Athabasca bitumen middle distillate). In the petroleum sample, the protons in the [M + H] + species originate primarily from other components of the mixture itself, rather than from the toluene dopant. In contrast to electrospray ionization, in which basic (e.g., pyridinic) species protonate to form [M + H] + positive ions and acidic (e.g., pyrrolic) species deprotonate to form [M - H] - negative ions, APPI generates ions from both basic and acidic species in a single positive-ion mass spectrum. Ultrahigh-resolution mass analysis (in this work, m/ m 50% = 500,000, in which m 50% is the mass spectral peak full width at halfmaximum peak height) is needed to distinguish various close mass doublets: 13 C vs. 12 CH (4.5 mda), 13 CH vs. 12 CD (2.9 mda), and H 2 vs. D (1.5 mda). Introduction Atmospheric Pressure Photoionization (APPI) forms positive ions through several mechanisms that include proton transfer reactions. 23, 45 As for Electrospray Ionization (ESI), a protic compound must be present in solution or the gas phase to facilitate efficient proton transfer. ESI 48

70 solutions typically include an acid (positive ion) or base (negative ion) at ~1 % by volume. Likewise, APPI can have an additional protic solvent added to solution. For example, acetonitrile or methanol has been used successfully to increase protonation of neutral analytes. 27, 62 Furthermore, toluene is often added as a dopant to increase ion yield by proton transfer and/or charge exchange reactions. 26, 63 Specifically for protonation, a compound with a higher proton affinity than the benzyl radical will form the desired [M + H] + ion. 23, 27 Conversely, charge exchange reactions can produce M + if the ionization potential of the toluene cation is higher than that of the analyte. Anions can also form in the APPI source. Acidic species can deprotonate to form (M H) - and positive electron affinity compounds can capture thermal electrons and form M -. Kostiainen et al. studied negative ion formation mechanisms by investigating ionization efficiency and ion type (even- or odd-electron ion) for analytes of different polarity in various solvents. 27, 64 Traldi et al. reported negative ion formation by resonant electron capture from thermal electrons originating from metal surfaces and dopant. 65 Appropriately, these positive and negative ion formation mechanistic studies have involved various compounds mass analyzed sequentially. The molecular structure (i.e., polarity) of the selected compounds and solvent system govern the efficiency of positive and negative ion formation and the proton donor species could be theoretically and experimentally determined. However, complex mixtures can contain many potential proton donors. Negative and positive ions can be formed simultaneously in the APPI source. Furthermore, negative-ion APPI Fourier Transform Ion Cyclotron Resonance (FT-ICR) mass spectrometry of a crude oil can produce > spectral peaks in which more than 99 % arise from deprotonated compounds (i.e., proton donors). 66 In this work, we couple APPI with a 9.4 Tesla FT-ICR mass spectrometer 9 for analysis of a 49

71 petroleum sample to investigate proton transfer reactions with deuterated toluene, and thereby determine the extent of toluene s contribution to protonation (deuteration) of analytes ions in a complex mixture. The complexity of the petroleum sample and the presence of close mass doublets (see below) requires the high resolution and mass accuracy afforded by FT-ICR mass spectrometry 7 to resolve and assign molecular formulas to the various deuterated vs. protonated compounds. Experimental Methods Solvents and Compounds Model compounds and deuterated toluene (C 7 D 8 ) were purchased from Sigma-Aldrich (St. Louis, MO). The aromatic nitrogen compounds were prepared in equimolar concentration (50 µm) in deuterated toluene. Crude Oil Three Athabasca Canadian bitumen distillates were provided by the National Centre for Upgrading Technologies (NCUT), Devon, Alberta, Canada. The bitumen distillation cuts were diluted in toluene (500 µg/ml) or deuterated toluene and analyzed without further preparation. A Thermo Fisher Scientific (Lakewood, NJ) CHNS-O Flash EA elemental analyzer provided element weight percent for the bitumen (2200 ppm nitrogen content). Results And Discussion Canadian bitumen tar sands have a relatively high nitrogen content. ESI FT-ICR mass spectrometry efficiently ionizes the acidic 50

72 (pyrrolic) and basic (pyridinic) nitrogen classes in petroleum through proton transfer reactions. 33, 51, 56 Prior APPI analysis of a petroleum sample and nitrogen model compounds showed similar ionization trends (compared to ESI) for aromatic nitrogen compounds with additional radical molecular cation formation (primarily for other heteroatom class compounds). Aromatic nitrogen species preferentially form (M + H) + or (M H) - in the APPI source and therefore, provide a good test bed for investigation of proton transfer reactions in complex mixtures. Nitrogen Class Compounds Five aromatic nitrogen compounds (Figure 4.1) were chosen to model the proton transfer reaction. The structure of the CH 3 H 3 C N N Acridine C 13 H 9 N MW DBE N CH 3 7,9-dimethylbenz[c]acridine C 19 H 15 N MW DBE C H 3 N H H N N H Carbazole C 12 H 9 N MW DBE Elipticine C 17 H 14 N 2 MW DBE 7H-dibenzo[c,g]carbazole C 20 H 13 N MW DBE Figure 4.1. Five aromatic nitrogen compounds chosen to model petroleum acidic and/or basic compounds. Five-membered ring nitrogen structures are acidic and six membered ring nitrogen species are basic. 51

73 nitrogen-containing ring determines the preferred ionization mechanism. For five membered nitrogen rings, the hydrogen bonded to the nitrogen atom is acidic and preferentially deprotonates. On the other hand, the six membered nitrogen ring species are basic (because of the electron lone pair on nitrogen) and efficiently protonate. Thus, two of the model compounds are basic; two are acidic; and the remaining compound (ellipticine) has both acidic and basic moieties. The negative-ion APPI FT-ICR mass spectrum (Figure 4.2) of an equimolar solution of all five nitrogen compounds in deuterated toluene shows ions only for the acidic species. No radical anions (M - ) appear. Under continuous APPI source operation, the mass spectrometer was reconfigured for positive ion detection (Figure 4.3). All five nitrogen model compounds yield positive ions by APPI. The pyrrolic compounds (carbazole and 7H-dibenzo[c,g]carbazole) primarily form radical cations, M + (Table 4.1), whereas the pyridinic compounds (acridine and 7,9- dimethylbenz[c]acridine) protonate to form (M + H) +. Ellipticine, which contains both a pyrrolic and a pyridinic moiety, preferentially forms (M + H) +. Table 4.1 list the ion relative abundances for each compound, and the parenthetical values are the percentages of M +, [M + H] + and [M + D] + for each compound. For the pyridinic species, the percent abundance of [M + D] + was (see upper left inset in Figure 4.3). In contrast, there was no detectable (M + D) + for the pyrrolic class compounds (which did form (M + H) + in low abundance). Interestingly, 7H-dibenzo[c,g]carbazole exhibited minor hydrogen-deuterium exchange [C 20 H 12 D 1 N 1 ] + (see Figure 4.3). Carbazole also participates in hydrogen-deuterium exchange, but in low relative abundance. 52

74 Negative Ion APPI 9.4 Tesla FT-ICR MS In C 7 D 8 H N - -H N H -H - N H 3 C CH 3 N H -H m/z Figure 4.2. Negative ion APPI FT-ICR mass spectrum of an equimolar solution of the model compounds of Figure 4.1 in deuterated toluene. Only the acidic compounds containing a pyrrole ring are deprotonated to yield [M - H] - ions, none of which contained deuterium. 53

75 Table 4.1. Positive-ion APPI FT-ICR MS ion relative abundances for the five aromatic nitrogen compounds of Figure 4.1. Parenthetical values show the percentages of M +, [M + H] +, and [M + D] + for each compound. Compound M + (M + H) + (M + D) + Carbazole 2.7 (87) 0.4 (13) -- Acridine 0.4 (1) 39.0 (84) 7.0 (15) Ellipticine 0.8 (5) 14.8 (85) 1.8 (10) 7,9-dimethylbenz- [c]acridine 1.2 (3) ) 4.2 (14) 7H-dibenzo- [c,g]carbazole 34.6 (99) 0.4 (1) -- Bitumen Distillation Cuts Three Canadian bitumen petroleum distillation cuts were analyzed by positive-ion APPI FT-ICR MS. Mass analysis (with ~100 ppb mass accuracy) for all three cuts showed that ~90 % of the compounds contain at least one nitrogen atom. Furthermore, the N 1 class ("class" denotes N n O o S s heteroatom composition) ions consist primarily of protonated compounds (> 97%). Figure 4.4 shows the heteroatom class distribution for the middle distillation cut ( C). Unlike electrospray ionization sources, 36, 58 atmospheric pressure chemical ionization, 37 or field desorption ionization, 41 the APPI source produces both positive and negative ions simultaneously. Figure 4.5 shows broadband APPI positive- and negative-ion mass spectra for the middle distillation cut. Both spectra were collected without ion source interruption by reversing dc voltage polarity for ion transfer and trapping. 54

76 Positive Ion APPI 9.4 Tesla FT-ICR MS In C 7 D 8 [C 12 H 9 N 1 13 C+ H] + [C 13 H 9 N 1 + D] + [C 19 H 13 N 1 13 C] + [C 20 H 12 D 1 N 1 ] + [C 20 H 13 N 1 + H] + N + + H m/z m/z H C 3 + H CH 3 N + H N + N CH H N H C H 3 N H m/z Figure 4.3. Positive ion APP FT-ICR mass spectrum of an equimolar solution of the model compounds of Figure 4.1 in deuterated toluene. All five compounds yielded positive molecular (M + ) or quasimolecular ([M - H] - ) ions. The compounds containing a six-membered pyridinic ring are sufficiently basic to readily protonate (or deuterate) (along with ~1% of radical molecular radical cations), whereas the more acidic compounds containing a five-membered pyrrolic ring form molecular radical cations, and <1% protonation (or deuteration). For the even-electron species, the extent of deuteration was ~15% for acridine (see the mass scale-expanded inset spectrum), ~10% for ellipticine, and ~14% for 7,9- dimethylbenz[c]acridine. Also, at nominal mass 268 (right mass scale-expanded inset), 7H-dibenzo[c,g]carbazole exhibits slight hydrogen-deuterium exchange. 55

77 Deuteration versus Protonation The Canadian bitumen middle distillation cut was dissolved in deuterated toluene and mass analyzed. The deuterated toluene sample produce a relative class distribution (not shown) identical to that for the toluene sample (Figure 4.4). Due to the sample complexity (~2000 molecular species in a 400 Da mass window), the ultrahigh resolving power of FT-ICR mass spectrometry was essential to resolve and identify the deuterated species. Figure 4.6 is the a mass scale-expanded segment from the bitumen broadband positive-ion mass spectrum. The 2.9 mda mass difference between the two compounds that differ elementally by 13 CH versus CD requires a minimum mass resolving power of 160,000 (m/ m 50%, in which m 50% is the mass spectral peak full width at half-maximum peak height). Furthermore, the 1.5 mda mass difference between H 2 and D requires a minimum mass resolving power of 300,000. Additional mass doublets unique to APPI mass spectra arise by virtue of the presence of radical molecular cations and (de)protonated compounds: namely, 4.5 mda separation 66 for 13 C of M + vs. 12 CH of [M + H] + and 1.5 mda separation for H 2 of M + vs. D of [M - H 2 + D] +. All of the above mass doublets could be resolved by FT-ICR mass spectrometry at an average mass resolving power of 500,000 for the seven assigned spectral peaks in Figure 4.6. It is thus possible to quantitate the relative abundances of deuterated ([C 33 H 49 N 1 + D] + ) vs. protonated ([C 33 H 49 N 1 + H] + ) ions identified in Figure 4.6. For the spectral segment in Figure 4.6, as well as the mass spectrum as a whole, the detected deuterated ions contributed ([M + D] + vs.[m + H] + ) 5% for the even electron N 1 class compounds. For other less abundant protonated species, the deuterated species if present were below the detection limit. 56

78 (+)APPI FT-ICR MS Athabasca Bitumen Distillation Temperature C 1801 Elemental Formulas RMS Mass Accuracy 149 ppb m/z N1 N1 O1 N1 S1 O1 S1 S1 S2 HC Figure 4.4. Heteroatom class distribution for a bitumen mid-range distillate positive ions. Each class represents the relative ion abundance of species which contain the stated heteroatom(s) in the assigned molecular formula. The error bars are standard deviation computed from 3 separate sample preparations and analysis. 57

79 Athabasca Bitumen APPI FT-ICR MS Distillation Temperature C Positive Ions Negative Ions m/z Figure 4.5. Broadband APPI FT-ICR mass spectra of a bitumen mid-range distillate. The positive- and negative-ion spectra were collected without source interruption and with appropriate instrument polarity changes. Although APPI produces both molecular radical cations (M + ) as well as [M - H] - and [M + H] + ions, the N1 class positive-ion mass spectrum is dominated (~97%) by protonated compounds. 58

80 Negative and Positive Ion Class Distribution Comparison Unique elemental compositions were assigned to both negative and positive ion mass spectral peaks based solely on accurate mass measurement 67 combined with sorting of homologous alkylation series 34, 35 to yield 1844 negative-ion elemental compositions with an rms mass error of 105 parts-per-billion, and 1801 positive-ion elemental compositions with an rms mass error of 149 parts-per-billion. Figure 4.7 displays the heteroatom classes for positive- and negative-ion APPI FT- ICR mass spectra. The elemental compositions for each class can be further sorted by DBE (double bond equivalents, in which DBE is the number of rings plus double bonds calculated from Eq. 4.1) Double Bond Equivalents (C c H h N n O o S s ) = c -h/2 + n/ Representative structures for the most abundant DBE components of the most abundant (N 1 ) class compounds are illustrated. Each molecular structure represents one of many possible isomers. Neverthelss, we can say that the nitrogen atom DBE 9 N 1 class positive ion resides in a six membered pyridinic ring. Conclusions APPI of the nitrogen class compounds in a petroleum sample preferentially forms ions by means of proton transfer reactions. The Athabasca bitumen N 1 class positive ions consist of >97 % protonated compounds and only ~3 % radical molecular cations. Based on their elemental compositions, all of the negative ions form by deprotonation of 59

81 Athabasca Bitumen Distillation Temperature C [C 32 H 49 N 1 13 C + H] + Positive Ion APPI 9.4 Tesla FT-ICR MS In C 7 D mda 13 CH vs. CD [C 32 H 48 N 2 + H] + [C 30 H 50 O 1 S 1 + H] + [C 33 H 48 O 1 + H] + [C 31 H 45 N 1 O 1 13 C + H] mda H 2 vs. D [C 33 H 49 N 1 + D] + [C 33 H 51 N 1 ] m/z Figure 4.6. Positive-ion APPI FT-ICR mass scale-expanded segment of a bitumen mid-range distillate in deuterated toluene. This figure emphasizes the ultrahigh resolving power required to resolve the deuterated species in complex petroleum mixtures. 60

82 Athabasca Bitumen Distillation Temperature C APPI 9.4 Tesla FT-ICR MS 9 DBE 3 DBE N O (CH 2 ) n (CH 2 ) n OH Basic Species Proton Acceptors Positive Ions Acid Species Proton Donors Negative Ions N1 N1 O1 N1 S1 S1 S1 O1 S2 O2 S1 O2 N1 O1 O3 Figure 4.7. Heteroatom class distribution for the positive and negative ions from a bitumen mid-range distillate. Generic structures are shown for the most abundant positive and negative species. DBE is the number of rings plus double bonds, and is calculated from Eq Because only ~5% of the evenelectron N1 class ions contain deuterium, the acidic neutrals in the original sample are likely proton donors to form the even-electron species from basic neutrals. 61

83 precursor neutrals. Aromatic nitrogen model compounds (Figures 4.2 and 4.3) exhibit similar ionization trends in which the majority of nitrogen species are ionized through proton transfer reactions (Table 4.1). Although pyrrolic species more efficiently form negative ions and pyridinic species positive ions, pyrrolic species can also form radical cations (M + ). Interestingly, the more aromatic pyrrolic compound (7H-dibenzo[c,g]carbazole) exhibited higher ion abundance (radical cation) than carbazole, suggesting that a more condensed aromatic core structure can add stabilty to a radical cation. Negative and positive ions form simultaneously in the APPI source, and therefore, there are many potential proton donors including toluene (or deuterated toluene) solvent. However, in the present bitumen sample, deuterated toluene donated a deuteron to only % of the even-electron ions formed from pyridinic nitrogen model compounds. Presumably, the proton donors for the pyridinic nitrogen compounds (for the model nitrogen compound spectra) are the pyrrolic (acidic) nitrogen compounds from the sample itself (even though the toluene dopant is present at orders of magnitude higher concentration). For the petroleum sample, only 5 % of the even-electron nitrogen class species were deuterated. The most abundant negative-ion species is the O 2 class, likely carboxylic acids. All species detected in the petroleum negative-ion spectrum were [M - H] -. Reasonably, the even electron nitrogen class that were protonated (not deuterated) are protonated through reactions with acidic species present in the sample. 62

84 CHAPTER 5. SULFUR SPECIATION OF PETROLEUM BY ATMOSPHERIC PRESSURE PHOTOIONIZATION FOURIER TRANSFORM ION CYCLOTRON RESONANCE MASS SPECTROMETRY Summary A Middle East petroleum and its Saturates-Aromatics-Resins- Asphaltenes (SARA) fractions are analyzed by Atmospheric Pressure Photoionization (APPI) with a 9.4 Tesla Fourier transform ion cyclotron resonance (FT-ICR) mass spectrometer. The Environmental Protection Agency has regulated the heteroatom content in petroleum products to low levels (<15 ppm) to reduce harmful combustion products and enable clean vehicle technologies to operate optimally. Sulfur is the third most abundant element in petroleum, and non-polar sulfur containing compounds are not efficiently ionized by electrospray ionization. APPI is a more general ionization technique which can ionize hydrocarbons and non-polar sulfur compounds through proton transfer reactions (dopantassisted APPI) and charge exchange. In this current work, we speciate the sulfur containing aromatic compounds in an unfractionated petroleum and also its SARA fractions and note differences and similarities between the sulfur species in the samples. Introduction Since the discovery of petroleum at Oil Creek in Titusville, Pennsylvania, 1859, 68 and the subsequent industrial revolution, societies dependence on petroleum products has increased. The consumption of "sweet" crude oil reserves has lead to an increase in the refinement of less desirable heavy crude oil reservoirs indicated by a steady trend in the feed stock crude oil in the United States toward lower 63

85 API gravity (heavier crude oils) and higher sulfur content. 31 The average sulfur content of all crude oils refined in the five regions of the U.S. increased from 0.89 wt.% in 1981 to 1.25 wt.% in 1997, while the corresponding API gravity decreased from in 1981 to in The United States Environmental Protection Agency (EPA) is charged with regulatory authority over the petroleum industry. More stringent EPA regulations have created a need to develop technologies designed to remove harmful heteroatoms. More specifically, sulfur (the third most common element in crude oil) content in petroleum products has been regulated to lower levels and ultra low level by year Moreover, large oil reserves in the Middle East and Venezuela have high sulfur content (>1%). 70 Initially, diesel fuel sulfur content was targeted primarily because of acid rain and SO 2 pollution. Oil fractions in the diesel boiling range typically contained 0.1 to 1.5 weight percent sulfur, and as late as the 1980s, sulfur specifications for diesel oils were decreased to 0.3 wt % or 3000 parts per million (ppm). 71 In the 1990s, EPA regulations (Clean Air Act, Code of Federal Regulations, Title 40) and European standards were enacted to reduce the sulfur content in diesel fuel to 15 ppm by 2007 and 10 ppm by 2009 respectively from the current standard of 500 ppm. 71 Diesel is not the only fuel affected by EPA regulations. The EPA's Tier 2 Vehicle and Gasoline Sulfur program treats vehicles and the fuels they use as a system. Thus, clean vehicle technologies will require lowsulfur gasoline for vehicles to run their cleanest This program requires the refinery industry to meet a 30 ppm sulfur content average with an 80 ppm sulfur cap. Prior high resolution mass spectrometric applications have focused on the characterization of polar compounds in crude oils, bitumens and their SARA isolated fractions by Electrospray Ionization (ESI) FT-ICR 64

86 mass spectrometry. 32, 33 The selectivity of the ESI process efficiently ionizes acidic and basic species without interference from the bulk hydrocarbon matrix. However, less polar sulfur containing species are rendered unobservable due to their low ionization efficiency. Furthermore, polar sulfur species are often overwhelmed by the more abundant polar species of equal acidity/basicity or less abundant species of greater acidity/basicity and, therefore, are present at low signal to noise. Atmospheric Pressure Photoionization (APPI) selectively ionizes those species that can either undergo direct ionization from 10eV photons (aromatics) or gas phase proton transfer reactions and charge exchange reactions. 19, 21, 23 The benefit of APPI is that it efficiently ionizes many important classes (nonpolar sulfur species and polycyclic aromatic hydrocarbons (PAH s)) that are unobservable by ESI. In this work, we focus on the compositional characterization of both nonpolar and slightly polar sulfur species in SARA fractions isolated from a medium Middle Eastern crude oil and the whole crude by APPI FT-ICR MS to provide insight into nonpolar sulfur compositional variations in the whole crude. Experimental Methods Middle East Crude Oil A Middle East Crude oil was supplied by ExxonMobil, and a sample was fractionated according to the Saturates-Aromatics-Resins- Asphaltenes (SARA) method. 57 The crude oil (1 gram) was added to 100 ml of n-heptane. After stirring 4 hours, the solution was stored in the dark for ~12 hours. The solution was then filtered to collect the n- heptane-insoluble asphaltenes. Subsequent Soxhlet extraction (8 min/cycle with hot normal heptane) removed coprecipitant materials from the purified asphaltenes. The raw asphaltene was purified until the 65

87 hot normal heptane collected in the extraction region of the Soxhlet apparatus was uncolored. The maltene fraction was further separated into saturates, aromatics and resins. 57 An aliquot of the whole crude and SARA fractions were diluted (500 µg/ml) in toluene (Fisher HPLC grade) and analyzed without further preparation. Results And Discussion Middle East Crude Analysis Previous mass spectral analysis revealed the complexity of data that can result from petroleum analysis with an APPI source (Chapter 2). Crude oil fractionation methods have been developed to lessen the complexity of a sample. A common method used for petroleum analysis is the Saturates-Aromatic-Resins-Asphaltenes fractionation. The challenge for mass spectrometry of complex mixtures is resolving power. For FT-ICR mass spectrometry, the many components of petroleum, though challenging, can be identified in part because of the ultra-high resolving power afforded by ICR mass spectrometers without sample fractionation. Furthermore, high mass accuracy can yield unique elemental formulas for mass spectral peaks. Nevertheless, in this work, we employ a common SARA fractionation method before FT-ICR mass analysis for a Middle East crude oil to speciate the sulfur compounds and identify differences between the fractions and whole crude. Figure 5.1 is a comparison of the broadband mass spectra of the whole crude oil and its SARA fractions. The aromatic and resin fractions display similar mass spectral distribution while the saturate fraction distribution is shifted to lower mass and the asphaltene fraction to higher mass. This broad perspective agrees with the common mass characteristics for the SARA fractions, i.e., the saturates are considered 66

88 Figure 5.1. Broadband positive APPI FT-ICR mass spectra of the whole crude and its SARA fractions. The samples were analyzed at the same concentration and experimental conditions. 67

89 the lighter fraction, and the asphaltenes are more polycondensed aromatic structures and heavier, with the aromatics and resins inbetween the light and heavy species. Unique elemental formulas were assigned to spectral peaks based on accurate mass and homologues series. The summed relative abundance of all the spectral peaks assigned to a homologues series can be represented in a bar graph (Figure 5.2). In Figure 5.2, the class, e.g. S 1, relative abundance represents the summed spectral magnitude for all peaks assigned a elemental formula with only one sulfur atom and the remainder of the elemental formula composed of only carbon and hydrogen. Middle East crude oils have a high sulfur content and this is reflected in the whole crude analysis (Figure 5.2). Eight of the eleven classes above 1 percent summed relative abundance contain one or more sulfur atoms. Relative Abundance + APPI FT-ICR MS Class Distribution Unfractionated Middle East Crude Oil S1 S2 N1 N1 S1 HC O1 S1 Class S3 N1 O1 N1 S2 N1 O1 S1 O1 S2 Figure 5.2. Summed relative ion abundance for heteroatom classes in the whole crude. Middle East crude oils have a high sulfur content. The graph includes those heteroatom classes above 1% relative abundance. Eight of the eleven classes contain one or more sulfur atoms. 68

90 Figure 5.3 is the class analysis of the four SARA fractions obtained from the Middle Eastern crude oil by positive ion APPI FT-ICR MS. The saturate fraction (upper left) is dominated by the S 1 and S 2 classes and contains a high amount of the hydrocarbon (HC) class as well. The aromatic fraction (upper right) continues the high S 1 /S 2 trend, and also contains a high amount of S 3 and NS classes with a increased amount of O x species. The resins (bottom left) transition from high S x classes to high NO, NOS and OS classes but also contain the S 1, S 2 and S 3 classes observed in the aromatics. The asphaltene fraction reverts back to classes similar to those observed in the aromatic fraction with a large associated increase in the OS x and N x O x and N x S x classes. Summed relative abundance bar graph analysis is informative for heteroatom type distribution, but contains no information that pertains to aromaticity (DBE) and carbon number. Figure 5.4 is a threedimensional relative abundance contoured DBE versus carbon number plot for selected heteroatoms classes of the whole crude and its SARA fractions. The x-axis represents carbon number and the y-axis DBE. The z-axis is color scaled to relative ion abundance (all plots are scaled equally). This type of information format represents a complete picture of all elemental formula assignments for the class. The carbon number distributions (x-axis) for the whole crude and its SARA fractions are similar in Figure 5.4. In general, the DBE range of the whole crude is duplicated in the individual SARA fractions. However, the higher DBE species are not efficiently ionized for the whole crude but are apparent in the asphaltene fraction. For the S 1 class, the saturate plot species exhibit relatively low DBE and carbon number which progress to species higher in DBE and carbon number through the aromatics and resins with the highest DBE/carbon-number species seen in the asphaltenes. This trend is also repeated for the S 2, S 3 and N 1 S 1 species. 69

91 Absolute Abundance, Arbitrary Units S1 N1 O1 N1 O1 S1 S1 N1 N1 S1 S2 O1 S1 O2 O1 S2 HC S3 N1 O2 O1 N1 S2 N1 O1 S2 N1 O2 S1 O2 S1 N2 O1 S1 N2 S1 O2 S2 O1 S3 N3 S1 S2 S1 S3 O1 S2 O1 S1 N1 O1 S1 N1 S1 N1 O1 N1 S2 O1 S4 N1 O1 S3 HC O2 S1 N1 O1 S2 N1 O2 N1 S3 N1 O2 S1 N1 O2 S4 O2 S2 O2 N2 O1 S1 S2 HC Saturates O1 S1 N1 O1 N1 S1 Resins S3 O1 S2 O2 S1 O2 N1 O Aromatics S2 S1 N1 N1 S1 S3 HC N1 O1 O1 S1 O1 S2 N1 S2 N1 O1 S1 S4 Asphaltenes O1 O1 S3 O2 S1 Class Figure 5.3. Summed relative ion abundance class graphs for the SARA fractions. The saturate, aromatic, and asphaltene fractions show sulfur species most abundant. For the resins, more polar heteroatom classes are dominant. 70

92 Figure 5.4. Three-dimensional relative abundance contoured DBE versus carbon number plot for selected heteroatom classes of the whole crude and its SARA fractions. Carbon number is represented on the x-axis and double bond equivalents (equation 1.19) on the y-axis. The z- axis is color scaled to relative ion abundance. All plots are scaled equally. 71

93 Furthermore, the resin fraction has a lower abundance of sulfur species. This is also represented in the class distribution graph (Figure 5.3). The resins are dominated by more polar species with nitrogen and oxygen heteroatoms. Additionally, the saturates are deficient in the S 3 and N 1 S 1 species. The N 1 S 1 and S 3 species are found primarily in the aromatic and asphaltene fractions. Conclusions APPI ionizes the non-polar species in crude oil which are not efficiently ionized by ESI sources. There is a need to better understand the complex matrix of crude oil. APPI coupled to FT-ICR MS can provide elemental formula analysis for the whole crude oil without fractionation. The SARA fractionation method separates the components of crude oil into subgroups based on aromaticity and polarity. However, there is a significant bleed-over between the fractions. That is to say, the same elemental species are observed in more than one fraction (although structural differences may be present). Nevertheless, the most abundant species of each fraction are significantly different (DBE and carbon number) from other fractions. For any single class, an overlap of the same elemental assignments are found in the whole crude analysis, albeit at different relative abundances. After fractionation, the relative ion abundances increase for species that dominate that fraction. We postulate that the differences in peak spectral magnitude between the whole crude and its fractions can be partially attributed to the ratio of photon flux to species molar concentration and competitive ionization in the APPI source. I have presented a detailed analysis of the non-polar sulfur components for a Middle East crude oil. Further studies will investigate the elemental differences between crude oil samples at different catalytic hydrodesulfurization steps. 72

94 CHAPTER 6. LIMITATIONS OF AROMATIC SULFUR CHEMICAL DERIVATIZATION ANALYSIS OF PETROLEUM BY ESI AND APPI FOURIER TRANSFORM ION CYCLOTRON RESONANCE MASS SPECTROMETRY Summary Sulfur-containing compounds in petroleum are detrimental to the environment and refining processes. The molecular characterization of the sulfur-containing species is therefore an important subject. One characterization method is the derivatization of the sulfur species prior to Electrospray Ionization (ESI) mass analysis. However, Atmospheric Pressure PhotoIonization (APPI) mass analysis can provide molecular characterization without chemical derivatization. A recent report speciated the sulfur-containing compounds in a vacuum bottom residue and identified the most abundant sulfur compounds present with double-bond-equivalents (DBE) values between 4 and 12. Our extensive experience with heavy-end petroleum causes us to question the analytical technique of sulfur derivatization prior to mass analysis. Here, we investigate the sulfur speciation of a petroleum vacuum bottom residue by ESI and APPI coupled to a 9.4 Tesla Fourier transform ion cyclotron resonance (FT-ICR) mass spectrometer. A comparison of the sulfur-containing compounds between the methods reveals significant differences in sulfur species DBE values. We postulate differences in APPI ionization efficiency could account for the DBE discrepancy. However, an analysis of the saturate and aromatic fraction by APPI shows equal ionization efficiency across a broad DBE range. Furthermore, we conclude the methylation reaction is sterically hindered for large DBE species (> 20 DBE). 73

95 Introduction Despite the recent interest in renewable energy sources, fossil fuels are projected to be the major source of energy for the next fifty years. 72 The increase in global consumption of "sweet" crude oil reserves has led to an increase in the refinement of less desirable heavy crude oil reservoirs, as evident by a steady trend in feed stock crude oils in the United States toward lower API gravity (heavier crude oils) and higher sulfur content. 31 The heavier feed stocks, heavy oils and bitumens, contain a large weight percent of sulfur, nitrogen and oxygen heteroatoms. The heteroatoms in the heavy petroleum are harmful to the environment, are detrimental to hydrogen addition and carbon rejection processes in petroleum refineries and therefore must be removed. 72 Even before the current need for low sulfur petroleum products, the refining industry had employed hydrodesulfurization for other reasons, e.g., to decrease corrosion, increase gasoline stability, and decrease smoke formation in kerosene. 72 Therefore, the petroleum industry has significant experience in desulfurization processes but primarily for lighter feed stocks. With current feedstock trends and a desire to process atmospheric and vacuum bottom residue into marketable lighter petroleum products, there is a need to develop new refining technologies and processing methods better suited for the heavier feedstocks and residues. Crude oil is a complex mixture of hydrocarbons with varying amounts of nitrogen, sulfur, oxygen and trace amounts of metal (iron, nickel and vanadium). An industry standard for the bulk characterization of crude oil is through the study of its fractional components. However, the fractional compositions vary significantly with laboratory isolation procedures; especially for heavier feed stocks. 73 In order to develop more efficient refining processes for heavier petroleum 74

96 feedstocks and residues, mass spectrometry can provide detailed compositional information on whole crude oils. Specifically, Fourier transform ion cyclotron resonance (FT-ICR) mass spectrometry is capable of resolving >12,000 spectral peaks in a single spectrum (which is paramount for petroleum) and provide unambiguous molecular elemental formulas based on mass accuracy and homologue series. 66 Electrospray ionization coupled to FT-ICR has provided elemental formula composition for the polar constituents in petroleum. 32, 33 Acidic molecular species, e.g., compounds with carboxylic acid or sulfonic acid groups, are deprotonated and basic species, e.g., pyridinic nitrogen, are protonated in the ESI process to form charged species. However, sulfur compounds (which are not sufficiently acidic or basic) and hydrocarbons are not efficiently ionized by ESI. One possible analytical/characterization method for non-polar sulfur species is ESI mass spectrometry of derivatized sulfur compounds. The derivatization chemistry involves electrophilic attack on the heterocyclic sulfur by a strong alkylating (methylating) reagent forming S-alkyl (methyl) sulfonium salts in solution prior to the ESI process. However, an alternate analytical method is Atmospheric Pressure PhotoIonization (APPI). APPI can efficiently ionize gas phase non-polar species (and polar species) through direct photon ionization 19, 21 or proton transfer (with a toluene dopant 23 or proton transfer reactions with acid species in a complex mixtures 74 ) and charge exchange reactions. Thus, APPI precludes the need for derivatization. Recently, sulfur derivatization of a vacuum bottom residue and ESI FT-ICR MS analysis was reported. 75 The analysis provided elemental molecular characterization of a feed stock (~900 species) before hydroprocessing and an effluent (~1000 species) after hydroprocessing. The derivatization was preceded by a saturates-aromatics-resinsasphaltene (SARA) fractionation and a ligand exchange chromatographic procedure to enrich sulfur species. The results indicated that the bulk of 75

97 the sulfur compounds for the feedstock and effluent of the vacuum bottom residue exhibited Double Bond Equivalents (DBE, a value equal to the number of ring plus double bonds in the molecular structure calculated from the elemental formula) between 4 and 12. However, from our experience with vacuum residues and heavy crude oils, the reported DBE ranges for the sulfur-containing species were much lower than previously observed by APPI and lower than expected for species concentrated in a vacuum residue. In this report, we utilize ESI and APPI coupled to a 9.4 Tesla FT-ICR mass spectrometer to speciate the sulfur-containing compounds in a vacuum bottom residue and identify similarities and/or differences in the species identified between the ionization techniques. We then compare the results (ESI and APPI) of the chemically derivatized vacuum bottom residue to highlight limitations of the derivatization process. Experimental Methods Vacuum Bottom Residue Methyliodide, silver tetrafluoroborate, 1,2-dichloroethane, methylene chloride and acetonitrile were purchased from Sigma-Aldrich (high purity, St Louis MO.). Established methodology was adopted for the derivatization chemistry A sample of the Canadian bitumen residue (13.5 mg) was dissolved in a (conical) vial containing 1,2- dichloroethane (3 ml), methyl iodide (1 mmol) and a stir bar. While mixing, a solution of silver tetrafluoroborate (1 mmol) in 3 ml of 1,2- dichloroethane was added. A yellow-brown precipitate immediately formed upon addition. The solution was stirred vigorously on a magnetic stir plate at ambient temperature for 48 hours. The precipitate was filtered and further washed with 1,2-dichloroethane. The combined washings and filtrate were evaporated to dryness under reduced pressure 76

98 to remove solvent and excess methyl iodide. The methylated samples were re-dissolved in a 1:1 (v/v) solution of methylene chloride/acetonitrile (10 mg/ml stock solution) for ESI analysis (1mg/mL analysis concentration). The stock methylated solution was diluted (1:10) into toluene for APPI analysis. Untreated residue was prepared (1mg/mL in 60:40 toluene : methanol and 1% formic acid) for ESI analysis and APPI analysis (1 mg/ml in toluene). SARA Fractionation Solvents were purchased from Fisher Chemical (HPLC grade). A SARA fractionation 57 of the vacuum bottom residue (VBR) was accomplished (517.1 mg). The sample was mixed with n-heptane (50 ml), stirred with a magnetic stir bar for 90 minutes and stored in the dark overnight. A Whatman No. 1 filter paper was used to separate the n-heptane insolubles (asphaltenes) from the maltenes and the filter paper asphaltenes were dried at room temperature. The maltenes n-heptane solution was rotary evaporated to dryness under reduced pressure. The dry maltenes were re-dissolved in n-heptane (6 ml). The maltene solution was then adsorbed onto the surface of activated alumina (3 g) and the maltene alumina slurry was dried while stirring under a stream of nitrogen. A glass column (11 mm i.d. x 300 mm length) was packed with activated alumina adsorbent (6 g) and the adsorbed maltenes were packed on the top. In sequence, 40 ml of n- heptane, 80 ml of toluene and 50 ml of a toluene:meoh (8:2, v/v) mixture were used to elute the saturates, aromatics and resins, respectively. The eluants were rotary evaporated under reduced pressure until dry and weighed. The recovered mass for each fraction and percent recovered follows: saturates, mg (23.7 %); aromatics, mg (29.0 %); resins, mg (29.0 %); asphaltenes, mg (33.4 %). The 77

99 saturates and aromatics were diluted in toluene (1 mg/ml) for APPI analysis. CHNOS Analysis A Flash Elemental Analyzer (C.E. Elantech, Inc.) model 1112 was used for CHNS/O weight percent determination of the vacuum bottom residue. Quadruplicate samples (~2 mg) were weighed for the CHNS analysis (combustion) and O analysis (pyrolysis). Calibration values were developed using sulfanilamide and 2,5-Bis-(5-tert.-butylbenzoxazol-2- yl)-thiophen (BBOT) standards. Percent composition follows: carbon 81.3 ± 0.19 % RSD, hydrogen 9.5 ± 0.28 % RSD, nitrogen 0.75 ± 2.3 % RSD, sulfur 5.7 ± 3.8 % RSD, oxygen 1.5 ± 3.1 % RSD. Results And Discussion APPI FT-ICR MS Although crude oil is the most compositionally complex organic mixture, ESI FT-ICR MS has enabled the detailed speciation of its polar components. 33, 48 Atmospheric Pressure PhotoIonization (APPI) can produce ions from non-polar sulfur-containing petroleum compounds and when combined with the ultra-high mass resolving power and unmatched mass accuracy of FT-ICR MS, sulfur speciation of petroleum is achieved. Petroleum fractionation methods can be employed to lessen the sample complexity and, even though the previous report 75 fractionated a vacuum bottom residue (VBR) before mass analysis, our attention is initially focused on the analysis of the raw VBR (a less complicated sample preparation) and the raw methylated VBR by ESI and APPI coupled to a 9.4 T FT-ICR MS. 78

100 Raw Vacuum Bottom Residue The residue analysis (raw unmethylated) yielded approximately 5800 and 2000 unique elemental formulas for positive ion APPI and ESI respectively. The elemental formulas were sorted by class, DBE and carbon number. Figure 6.1 represents the summed relative abundances of each class above 1 % relative abundance. The nitrogen class is the most abundant for both ionization techniques. ESI efficiently ionizes the basic pyridinic N 1 class 78 at twice the relative abundance as the APPI N 1 class. For other polar classes, e.g., N 1 S 1 and N 1 O 1, the relative ionization efficiency for the two ionization methods is comparable. For the nonpolar species, e.g., S 1, S 2, HC (hydrocarbon), only APPI produced spectral signal. Relative class abundance graphs are sufficient representations of class distribution but contain no information about carbon number and DBE distribution. One format that represents carbon number and DBE distribution within a class is an iso-abundant contoured DBE versus carbon number plot. Figure 6.2 compares the ESI and APPI carbon number and DBE distribution for the S 1 and N 1 S 1 classes of the raw VBR. Carbon number is plotted along the x-axis and DBE on the y-axis. The relative ion abundance within each class is color scaled in the z-axis. The APPI S 1 class has a carbon distribution of 22 C# 45 and more interestingly, a DBE distribution of 6 DBE 35. The greatest magnitude spectral peaks for the APPI S 1 have DBE values between 20 and 30. In comparison, the ESI and APPI N 1 S 1 classes have similar magnitude in the lower DBE range, however, the APPI N 1 S 1 class DBE range extends to higher DBE (similar to the APPI S 1 class). Similarly, other mutually common classes (ESI and APPI Figure 6.1) display comparable DBE and carbon number range for the greatest spectral magnitude ions with APPI generated species extending to higher DBE. Hence, for the raw unmethylated VBR, mutually ionized species have similar DBE range 79

101 Sum Relative Abundance ESI / APPI Class Distribution Comparison Raw Vacuum Bottom Residuum APPI ESI N1 N1 S1 S2 N1 O1 S1 N1 S2 S3 HC N1 O1 S1 N2 O1 S2 O1 S1 N2 S1 O1 S4 N2 O1 Class (>1% R.A.) Figure 6.1. Heteroatom class distribution for the raw vacuum bottom residue (not methylated). All classes ionized by ESI and APPI above 1 % relative abundance are represented. The non-polar classes, e.g., S1, S2 and HC (hydrocarbon), were not detected by ESI. APPI analysis detected both the polar and non-polar species. 80

102 Figure 6.2. Iso-abundant contoured DBE versus carbon number images for heteroatom species of the raw vacuum bottom residue. Relative ion abundance within the class is color scaled in the z-axis. The ESI and APPI N1S1 classes have similar carbon number distributions. However, the DBE distribution for the APPI N1S1 image extends to higher DBE. The non-polar S1 species was not detected by ESI. 81

103 which suggest the polar species generated by either ESI or APPI are equally represented. Raw Methylated Vacuum Bottom Residue An identical positive-esi and APPI FT-ICR MS analysis was performed on the methylated raw vacuum bottom residue. Figure 6.3 is the class distribution for the methylated sample. For the ESI classes, there is a dramatic shift in highest relative abundance from the N 1 class (Figure 6.1) to the S 1 class. Furthermore, the methylation reaction formed hydrocarbon (HC) and O 1 class ions not previously detected in the raw VBR ESI data. The ESI O 1 class most likely contains furanic structures that are known to also react with the derivatization reagent forming oxonium ions. The APPI class distribution (Figure 6.3) is nearly identical to the raw (underivatized) VBR in Figure 6.1. Figure 6.4 is the iso-abundant contoured plots for the methylated APPI and methylated ESI data for the S 1 and N 1 S 1 classes. The APPI S 1 and N 1 S 1 images are similar to those in Figure 6.2 (unmethylated VBR), with the exception of slightly lower DBE values present in the APPI raw VBR S 1 image are absent in the methylated VBR. For example, the APPI raw VBR S 1 class begins at DBE 8 (Figure 6.2) and for the APPI methylated VBR, the S 1 DBE range begins at 13. Also, the ESI N 1 S 1 image (Figure 6.4) is comparable (in the most abundant DBE and C#) to the ESI N 1 S 1 image in Figure 6.2. However, the differences between the ESI S 1 class and the APPI S 1 class in the derivatized VBR sample are remarkable. Although the carbon number distributions are similar, the DBE distributions are well separated with little overlap. We postulate a number of possible explanations for the DBE distribution discrepancy between the S 1 species. 82

104 Sum Relative Abundance ESI / APPI Class Distribution Comparison Methylated Raw Vacuum Bottom Residual APPI ESI N1 N1 S1 N1 O1 S1 HC S2 N1 O1 S1 N1 S2 O1 N2 O1 S1 N1 O2 S3 O2 O1 S2 Class (>1% R.A.) Figure 6.3. Heteroatom class distribution for the raw methylated vacuum bottom residue. All classes ionized by ESI and APPI above 1 % relative abundance are represented. The ESI distribution exhibits a remarkable change in highest relative abundance to the S1 class. The S2, HC and O1 classes are also detected by ESI in the methylated sample. The APPI heteroatom class distribution is similar to Figure

105 Figure 6.4. Iso-abundant contoured DBE versus carbon number images for heteroatom species of the raw methylated vacuum bottom residue. Relative ion abundance within the class is color scaled in the z-axis. The N1S1 images are similar to the images produced from the unmethylated sample (Figure 6.2). However, the S1 class images differ dramatically between ESI and APPI. The low DBE species in the ESI S1 image are absent in the APPI image. 84

106 One possible cause is that APPI more efficiently ionizes larger DBE species and thereby biases the DBE images to high values. In fact, it is clear from the ESI methylated VBR data that lower DBE species are present in the raw vacuum bottom residue. However, the APPI S 1 DBE distribution for the VBR does not reflect the lower DBE S 1 species. This issue will be addressed in the next section. Another possibility is that the methylation reaction is sterically hindered for larger DBE species. Thus, the low DBE S 1 species could represent only a small fraction (low mass fraction) of the total raw VBR sulfur species but be highly abundant in the ESI mass spectral results due to their inequitable methylation efficiency (compared to high DBE S 1 species) in the derivatization step. Saturate and Aromatic Fraction of the Vacuum Bottom Residue To test whether APPI more efficiently ionizes larger DBE species and thereby biases the DBE images to high values (ionization efficiency differences between low and high DBE S 1 species), we probe the APPI ionization efficiency for species of widely different aromaticity (DBE). The experiment is to fractionate the VBR into its saturate and aromatic fractions and subsequently analyze each by APPI. An equal weight combination of the saturates and aromatics would then be analyzed to identify gross differences in ionization efficiencies between low DBE S 1 species (present in the saturate fraction) and high DBE S 1 species (present in the aromatic fraction). Figure 6.5 is the heteroatom class distribution of the saturate, aromatic and a combined mixture (1:1 by weight) of the saturate and aromatic fractions analyzed by APPI. The combined saturate and aromatic solution was 85

107 APPI Class Distribution Saturates and Aromatic Fractions Sum Relative Abundance Saturates Aromatics Aromatics and Saturates 0 N1 S1 S2 N1 S1 HC S3 N1 O1 Class (>1% R.A.) N1 S2 O1 S1 O1 N1 O1 S1 O2 S1 Figure 6.5. APPI analyzed heteroatom class distribution for the saturates, aromatics and a solution of saturates and aromatics fractionated from the vacuum bottom residue. The saturates and aromatic solution was an equal molar concentration prepared by mixing equal volumes of equal mass/volume solutions. All classes above 1 % relative abundance are represented. 86

108 prepared by mixing equal volumes of equal weight per volume solutions. Interestingly, the saturate fraction has a high relative abundance for the S 1 class and significantly less for the N 1 class. This suggests that many sulfur species elute into the saturate fraction in a SARA isolation procedure. As expected, the combined saturate and aromatic fractions yield a heteroatom class distribution which is similar to the raw VBR distribution (Figure 6.2). The S 1 DBE distribution of the three solutions is depicted in Figure 6.6. The calculated DBE values (equation 1.19) are for the cation (not the neutral species), and therefore, a non-integer value for protonated compounds [M + H] + is possible because of the additional hydrogen atom. The integer values are the result of DBE calculations for radical molecular ions (M + ). The saturate distribution includes DBE values normally associated with a saturate fraction, i.e., low DBE values which can include zero for a hydrogen saturated compound. However, the aromatic fraction begins at DBE 6 and extends past the saturate fraction to DBE 28. Importantly, the analysis of the combined saturate and aromatic (equal weight/concentration) fractions yields a broad distribution that encompasses both the individual saturate and aromatic DBE distributions. Figure 6.7 is the iso-abundant contoured image for the saturate, aromatic and combined saturate and aromatic fractions S 1 class. The same trend established in the DBE distribution graph (Figure 6.6) is now represented in DBE image format. The lower DBE species are found in the saturate fraction, higher DBE species in the aromatic fraction, and a combination of the individual fraction DBE values in the combined fraction image. Thus, it appears from Figures 6.6 and 6.7 that APPI ionizes a wide range of DBE values with comparable efficiency. This suggests that the abnormally high abundance of the low DBE S 1 species observed in the methylated VBR ESI analysis and whose presence in the VBR is confirmed in the APPI mass spectral analysis of the SARA fractions, is a result of the enhanced methylation reactivity toward low 87

109 Saturates Vacuum Bottom Residuum APPI DBE Distribution S 1 Class Aromatics Saturates + Aromatics Double Bond Equivalents Figure 6.6. S1 DBE distribution of the saturates, aromatics, and the saturate/aromatics solutions. The calculated DBE values (equation 1.19) are for the cation (not the neutral species). Therefore, non-integer values are possible for protonated compounds [M + H] +. The analysis of the combined saturates and aromatics (equal weight/concentration) show a broad distribution which encompasses both the individual saturate and aromatic DBE distributions. 88

110 Figure 6.7. Iso-abundant contoured image for the S1 class of the saturates, aromatics and saturates/aromatics solutions. Relative ion abundance within the class is color scaled in the z-axis. The same trend seen in the DBE distribution graph (Figure 6.7) is represented in the images. The lower DBE species are found in the saturate fraction, higher DBE species in the aromatic fraction, and a combination of the individual DBE distribution values in the combined image. 89

111 DBE S 1 species and not due to DBE specific APPI ionization efficiencies. Therefore, the low DBE S 1 species in the saturate fraction (Figure 6.6 and 6.7) must have a low abundance (a reasonable assumption because this is a vacuum bottom residue) and these low abundance, low DBE S 1 species are not observed because they are simply below the detection limit. These results suggest that although chemical derivatization methods allow for nonpolar sulfur speciation by ESI, it does not efficiently methylate high DBE sulfur species. Thus, the chemical derivatization method presented here does not provide accurate sulfur speciation for petroleum materials that contain high DBE (>10) sulfur species and is therefore inherently limited to light distillate fractions. Conclusions In this report, we speciate the S 1 class of a vacuum bottom residue with two ionization methods coupled to a 9.4 Tesla FT-ICR mass spectrometer. The ionization methods demonstrated remarkably different S 1 DBE distributions. The ESI analysis for the methylated raw VBR showed an S 1 class with a distribution maximum centered at 7-10 DBE. The APPI analysis for the raw (not methylated) VBR showed an S 1 class with a distribution maximum centered at DBE. The APPI analysis of the saturate, aromatic and combined saturate and aromatic fractions showed equal ionization efficiency over a broad DBE distribution. This leads us to conclude that APPI does not discriminate against lower DBE species at similar molar concentration. Also, the APPI S 1 DBE distribution for the VBR (methylated and not methylated, Figures 6.2 and 6.4) differ in the low DBE range, i.e., the methylated APPI S 1 image (Figure 6.4) is lacking the low DBE species seen in the unmethylated sample (Figure 6.2). We postulate that the methylation of the lower DBE S 1 species render them unobservable by 90

112 APPI. Furthermore, we conclude the methylation reaction is sterically hindered for large DBE species (> 10 DBE). Not only because larger DBE species (> 10 DBE) were detected with an APPI source but also because this sample is a vacuum bottom residue which form our experience should contain high DBE species. Future research will further investigate the steric hindrance for the methylation reaction and comparison analysis of other high sulfur petroleums. 91

113 CHAPTER 7. CONCLUSIONS AND APPI FT-ICR MS APPLICATION AND COLLABORATION WITH THE INSTITUTE OF PETROLEUM AT FRANCE; A REAL WORLD APPLICATION Assessment of APPI Technology The application of APPI coupled to FT-ICR mass spectrometry can provide elemental composition of complex mixtures. Chapter 2 demonstrated the unmatched resolving power and mass accuracy of FT-ICR mass spectrometry. The dual ionization pathways of APPI can complicate mass spectra. Formation of even and odd electron cations translates to closely spaced mass spectral doublets that must be resolved for meaningful elemental assignments. For example, the difference between [C 23 H C] + and [C 24 H 14 + H] + is 4.5 mda ( 13 C vs. CH), and a 1.1 mda mass doublet corresponds to the mass difference between C 4 and SH 13 3 C, e.g., from the protonated molecule [C 24 H 29 N 1 S 13 1 C 1 + H] + and the radical molecular ion [C 28 H 27 N 1 ] + (Figures 2.3 and 2.4). Deuterated solvent experiments present their own unique mass doublets. Nevertheless, FT-ICR MS can also resolve these mass doublets, e.g., the 1.5 mda mass split which results from compounds that differ in elemental composition by H 2 and D (Figure 4.7). Conveniently, toluene is a good solvent for petroleum and an excellent dopant for APPI. However, toluene radical cation formation and solution vaporization temperature causes concern. One approach to validate new experimental techniques is through comparative studies. In Chapter 3, analyte fragmentation was addressed with model compounds and ESI comparison studies. Fortunately, petroleum analysis by ESI FT-ICR mass spectrometry is a well established soft ionization method in this laboratory. Therefore, comparison of ESI spectra with APPI spectra provided a foundation to build upon. The results from Chapter 3 92

114 identified fragmentation sources and established instrument parameters to avoid fragmentation. Confidence was gained through model compound spectra and APPI versus ESI petroleum analysis comparison. Spectra presented in Chapter 3 shows that proton transfer reactions are the dominant ionization pathway for pyridinic and pyrrolic nitrogen compounds. Likewise, the nitrogen class compounds in a petroleum sample preferentially form ions through proton transfer reactions. Chapter 4 investigated the source of the proton for APPI generated ions. For high nitrogen content petroleum, e.g., Canadian Athabasca bitumen, a majority of the nitrogen classes formed positive ions through proton transfer. To model the proton transfer, nitrogen compounds with acidic and basic moieties were analyzed in toluene and deuterated toluene. It was determined that a majority of the proton (deuteron) charge arises from reactions with other species present in the solution matrix even though toluene (or deuterated toluene) is present in much greater molar concentration. This was true for the petroleum sample also. Interestingly, also presented in this chapter, APPI sources produce a cloud of positive and negative ions simultaneously. Chapter 5 presented the thrust for APPI application, i.e., speciation of sulfur and other non-polar species in petroleum. As a first step, a crude oil was analyzed without any sample preparation to demonstrate the power of the technique. Furthermore, the crude oil was separated by a common petrochemical fractionation method and each fraction was individually analyzed. This chapter established the first detailed sulfur speciation a crude oil and its elemental composition yielded the most aromatic sulfur species ever detected by mass spectrometry. Chapter 6 investigated sulfur speciation by APPI FT-ICR MS and comparatively, ESI FT-ICR MS of a chemically derivatized petroleum sample. The ionization methods demonstrated remarkable differences for the sulfur containing compounds, i.e., different S 1 DBE distributions. The derivatized ESI sample spectrum was limited to lower DBE sulfur 93

115 species relative to the APPI sulfur species. It seems that the chemical derivatization is hindered for larger DBE species. Overall, evidence is provided that suggest APPI is a more generalized (equal ionization efficiency) ionization technique and may provide equal ionization efficiency across a broad DBE and polarity of compounds. APPI FT-ICR MS Applied to Current Petrochemical Challenges Chapters 2 through 6 developed and established a new analytical ionization technique for the elemental speciation of petroleum. Throughout the chapters, the need to characterize the sulfur containing species in petroleum to enhance refinery technology was emphasized. The ultimate utility of APPI FT-ICR mass spectrometry applied to petrochemical analysis will be found in monitoring molecular changes before, during and after refinery processes. In the following sections, I present data collected in collaboration with the Institute of Petroleum at France (IFP). A real world application for APPI FT-ICR mass spectrometry; Stepwise Molecular Characterization of an Asphaltene Hydroconversion Process. Introduction The following introduction was provided by Isabella I. Merdrignac, an IFP research scientist. While residue fuel oil demand has drastically decreased over the last decade, the demand for motor fuel derived from light and middle distillates is still increasing. However, light conventional crude oil production is declining and is being replaced by heavier, non-conventional crude oils. The main characteristics of heavy crudes (high viscosity and high heteroatom content) are directly related to the significant abundance of compounds such as resins and asphaltenes. 94

116 Concentrated in residue, they constitute the most polar fractions of these products. 79 To enhance the valorization of such oils, various upgrading processes have been developed. 80, 81 Hydroconversion processes (Hyvahl 82 and H-Oil) require catalysts to remove and accumulate metals (nickel and vanadium) and to desulfurize the feed. Although our knowledge has considerably improved in the last two decades, some industrial processes are not fully understood. In the refining process, we know that feedstocks with very similar composition convert in different ways or can induce variable aging of catalysts. The compositional analysis of heavy oil products has become a key step in various developments. However, the characterisation of such species still remains time consuming despite substantial efforts in this field. The complexity of these oil matrices tends to increase with their boiling point. A large variety of compounds which vary in structure and molecular weight are present. Asphaltenes are defined by their insolubility in a normal paraffinic solvent. They consist of a heterogeneous mixture of highly polydispersed molecules either in terms of size or chemical composition, with a high content of heteroatoms and metals. In solution, they exhibit self-assembly and colloidal behaviour, depending on the operating conditions. 83 Dissolved in residue, they may precipitate easily during hydroconversion due to chemical composition changes of the residue. Asphaltenes are known to be coke precursors and catalyst inhibitors. The deactivation of catalysts is thus strongly dependant on the heavy fractions concentration. Numerous analytical techniques regarding asphaltenes characterization have been performed. 84 Among them, APPI FT-ICR mass spectrometry can provide elemental molecular speciation and is the focus of our collaboration. 95

117 Residue Sample Overview The IFP selected a crude oil residue for analysis and performed a standard SARA method fractionation. As previously stated, asphaltenes are the most complex fraction in a crude oil and, even more so, asphaltenes precipitated from a vacuum bottom residue. Crude oil residue is the portion of petroleum which remains after successive distillation steps to remove lighter fractions under vacuum. Therefore, the asphaltene fraction of a vacuum bottom residue are heteroatom rich highly condensed polycyclic aromatic compounds. The asphaltene fraction was subjected to hydroconversion in a bench-top unit representative of industrial conditions. 85 A general scheme of the asphaltene conversion process and samples analyzed is depicted in Figure 7.1 Figure 7.1. Residue hydroconversion scheme. Sample designations A1, A2, A11 and A22 reflect hydroconversion in fixed bed conditions. Sample A1 and A2 were reacted at different temperature (hydrodemetalization) and further reacted to produce A11 and A22 (hydrodesulfurization). Samples B1, B2, and B3 were obtained in ebullated bed conditions at different increasing residence times. 96

118 All samples (to include the saturates, aromatics and resins) were diluted to the same concentration (1 mg/ml) and analyzed by APPI FT-ICR mass spectrometry in negative and positive ion mode. Table 7.1 is a compilation of all samples analyzed (negative and positive ion). The elemental peak assignment values represent all spectral peaks that could be assigned a unique molecular formula based on mass accuracy and homologues series. Although the negative ion spectra were acquired, initial data reduction presented here will focus on positive ion data. Table 7.1. Total Elemental Peak Assignment and Root-Mean-Square Mass Error (mass error, difference between experimentally measured mass and the exact mass corresponding to the elemental composition assigned to that mass spectral peak). The asphaltene alpha-numeric designators correspond to Figure 7.1. Elemental Assignments Elemental Assignments Sample Positive ion (RMS error)* Negative ion (RMS error)* Saturates 3526 (228) 3021 (103) Aromatics 5675 (176) (257) Resins 2783 (145) 8305 (323) Asphaltenes Feed 5624 (423) 3268 (282) A (432) 4939 (319) A (232) 5812 (450) A (303) 6792 (410) A (296) 4589 (441) B (452) 5466 (470) B (278) 4128 (384) B (262) 2997 (434) * error reported in parts per billion (ppb) The total unique elemental formulas assigned for the complete sample set (all samples in Table 7.1) was 104,630. This includes isotopic variants and elemental species overlap between samples. Noteworthy, the rms mass error for all samples is < 500 ppb (parts per billion). 97

119 [C 30 H 18 S 3 + H] + [C 29 H 18 S 13 2 C 34 1 S 1 + H] + [C 29 H 18 S 13 3 C 1 + H] m/z m/z IFP Aromatic Fraction APPI FT-ICR MS m/z Figure 7.2. Broadband APPI FT-ICR mass spectrum of the IFP aromatic sample (bottom). Zoom insets (top) identify [C30H18S3 + H] + (64 % relative abundance), [C29H18S3 13 C1 + H] + (22 % relative abundance, and [C29H18S2 13 C1 34 S1 +H] + (3 % relative abundance)

120 Figure 7.2 is the broadband positive ion APPI FT-ICR mass spectrum (bottom) of the aromatics. The zoom insets (top) identify an S 3 class compound and one of its isotopic variants. High sulfur content raw crude oils (not vacuum bottom residue) typically have S 1 and S 2 class species at relatively high abundance and S 3 species at lower abundance. For this sample, the multiple heteroatom species are concentrated, and the S 3 species is present at high spectral relative abundance. Hence, the [C 29 H 18 S 13 2 C 34 1 S 1 + H] + isotope of the [C 29 H 18 S 13 3 C 1 + H] + isotope is detected (which has never before been detected for a petroleum sample). Asphaltene Analysis Positive ion APPI FT-ICR mass spectra for each sample was acquired and spectral peak assignments were sorted by heteroatom class. The feed asphaltene analysis revealed a heteroatom-rich class distribution. Figure 7.3 is the relative abundance of the classes detected for the positive ion spectrum. For the greater abundant classes (> 1 % relative abundance, Figure 7.3 top), fifteen of the seventeen classes contained one or more sulfur atoms. Not surprising, the hydrocarbon class was detected in low abundance. Samples A1 and A2 represent the catalyzed reaction products at two different temperatures. The class analysis (Figure 7.4) reveal a reaction threshold was crossed between the temperatures (380 C and 400 C). For sample A-1, a high percent of sulfur containing compounds are still present; although, the hydrocarbon (HC) relative abundance has increased in comparison to the feed asphaltene (Figure 7.3). Sample A-2 shows a large relative reduction in the sulfur classes which is accompanied by an increase in the HC class. Relative abundance of S x species in the partial-conversion A1 sample are similar to those of S x+1 in 99

121 Sum Relative Abundance S2 S3 S1 N1 S2 N1 O1 N2 N1 O1 O1 S1 S3 S4 N1 S1 N2 N2 N2 N3 S2 O1 S1 N1 S3 O1 S2 N1 O1 S1 N1 O1 N4 O1 N2 S4 S4 O1 S2 APPI FT-ICR MS Asphaltene Feed Heteroatom Class Distribution HC N1 O1 S2 N4 S1 N3 S2 O2 S2 N1 O1 S3 O2 S1 N1 O2 S1 O1 S1 N1 O2 Class Distribution Continued Class S5 N2 S1 O3 N4 O4 N1 O4 O4 O1 O2 S1 S2 S2 Figure 7.3. Heteroatom class distribution for the IFP asphaltene sample. Forty heteroatom classes were assigned. For classes above 1 % relative abundance (top), 15 of the 17 classes contain one or more sulfur atoms. Also note the hydrocarbon class (HC) is present in low abundance. 100

122 Heteroatom Class Distribution Comparison A1 (380 C) versus A2 (400 C) Relative Abundance A1 A2 S1 S2 HC N1 S1 N1 S3 N1 S2 N2 Class O1 S1 O1 N1 O1 N2 S1 O1 S2 N1 O1 S1 S4 Figure 7.4. Class distribution for samples A1 and A2 (Figure 7.1). Each sample is normalized to the most abundant class within its class distribution, i.e., they are mutually exclusive. Sample A1 was reacted at 380 C and sample A2 at 400 C. 101

123 the feed and therefore suggest a reaction pathway like S x A1 = S x feed + S x+1 feed. Figure 7.5 is a three dimensional iso-abundance DBE versus carbon number plot of the sulfur classes of the feed compared to the A1 sulfur classes (and HC). This type of plot represents all the detected species for a given class displayed in one visual image. However, each plot is normalized to the highest spectral magnitude ion within that plot. Hence, relative abundances between classes are represented in bar graphs (Figure 7.4), and relative ion abundance within a class is depicted in iso-abundance contoured plots (Figure 7.5 and 7.6). The relative abundance of the sulfur species remained high for the A1 sample. The S 3 species in the feed would, presumably, be detected as S 2 species in the A1 sample and so forth. However, the A1 S 2 species exhibit a higher DBE (3-4 DBE) than the feed S 3 species and also an increase in carbon number (~6 carbons). Likely, there are condensation reactions occurring (forming rings and double bonds) and, the A1 S 2 species also include unreacted feed S 2 species because there is only a small percent of the sulfur species converted at this reaction temperature. In any case, there should have been a rearrangement of the sulfur families since the sulfur content decreased from 8 wt% in the feed to 7.2 wt% which is not much compared to the asphaltene conversion (47 %). Figure 7.6 is the comparison of the feed asphaltene to the A2 sample (reacted at higher temperature). At higher temperature, the A-2 sample showed a larger reduction of the sulfur species (Figure 7.4) which corresponds to elemental analysis (8, 7.2 and 3.5 wt% sulfur in the feed, A1 and A2, respectively). However, comparison of Figures 7.5 and 7.6 show little, if any, difference between A1 and A2 sulfur and hydrocarbon species. Hence, higher reaction temperature increased the rate at which sulfur is removed but produced similar hydrocarbon species. 102

124 Figure 7.5. Iso-abundant contoured DBE versus carbon number plot of the feed asphaltene and A1 sample; sulfur and hydrocarbon classes. Relative ion abundance is color scaled in the z-axis. A dashed reference line is added to enhance graph-to-graph comparison. Figure 7.6. Iso-abundant contoured DBE versus carbon number plot of the feed asphaltene and A2 sample; sulfur and hydrocarbon classes. Relative ion abundance is color scaled in the z-axis. A dashed reference line is added to enhance graph-to-graph comparison. 103

125 The heteroatom abundance of the final reaction products (A11 and A22) for this reaction branch (A1 A11 and A2 A22) is represented in Figure 7.7. Sample A1 and A2 reaction vessels utilized a macroporous (micrometer pore size) catalyst at different temperatures. These samples (A1 and A2) were the feedstock for further sulfur reduction which produced samples A11 and A22 (reacted with a mesoporous catalyst (angstrom pore size)). The final class composition of A11 and A22 are nominally identical in species and relative abundance. However, elemental analysis exhibited differences. Sulfur content for A11 decreased from 7.2 wt% (A1) to 4.5 wt% (A11), and for A22 from 3.5 wt% (A2) to 1.6 wt% (A22). Relative Abundance Heteroatom Class Distribution Comparison A11 A22 HC N1 O1 N1 O1 S1 O2 Class Figure 7.7. Heteroatom class distribution of the final reaction products (A11 and A22). Figure 7.8 is a comparison of the hydrocarbon species from A11 and A22. The graph highlights differences in aromaticity (defined as the carbon number to DBE ratio). For A11, the highest abundant species are centered at ~45 carbon and 27 DBE. Sample A22 shows a distinct shift toward the line of demarcation which corresponds to an increase in aromaticity. The line of demarcation is a carbon number to DBE ratio 104

126 Figure 7.8. Iso-abundant contoured DBE versus Carbon number plot for the hydrocarbon classes of A11 and A22. A22 shows an increase in aromaticity (increase in carbon number-to-dbe ratio). 105

127 boundary between planar carbon structure and bowl shaped structure, e.g., coronene versus corranulene. The second hydroconversion reaction branch (feed B1, B2, B3) was carried out under similar catalytic conditions as A2 with increased temperature (427 C) and increasing conversion time. Elemental sulfur analysis showed a decreasing weight percent, 5.5, 2.45 and 1.38 wt% sulfur for B1, B2 and B3, respectively. Figure 7.9 is the class distribution for samples B1, B2, and B3. The increased temperature (B1 427 C versus A2 400 C) shows no significant increase in the reduction of sulfur species, i.e., the HC, N 1, S 1, and S 2 relative abundances are comparable between A2 Figure 7.4 and B1 Figure 7.9. As expected, increased catalytic reaction time results in further reduction of sulfur accompanied by an increase in hydrocarbon. Class Distribution Comparison B1, B2 and B3 Relative Abundance B1 B2 B3 Increasing Time HC N1 S1 O1 S2 N2 N1 O1 Class N1 S1 O1 S1 S3 Figure 7.9. Class distribution for samples B1, B2 and B3. With increasing hydroconversion time, there is a corresponding reduction in sulfur species with an increase in hydrocarbon species. Finally, Figure 7.10 is a comparison of the hydrocarbon classes for samples B1, B2, and B3. The plots in Figure 7.10 display a progression 106

128 a b c C 36 H 16 C 38 H 16 C 40 H 16 Figure Iso-abundant carbon number versus DBE contoured plots for the hydrocarbon classes of samples B1, B2 and B3. The plots display an overall migration of the most abundant species toward greater aromaticity. Structures a, b, and c (bottom) represent the predicted stable structures which correspond to elemental species hot spots for sample B3. 107

129 to more aromatic structures with increased catalytic residence time. The proposed compounds (Figure 7.10 bottom) represent stable aromatic structures that correspond to high ion magnitude in the spectrum. The structures are only one possible isomer. However, a polycyclic aromatic hydrocarbon with a given number of aromatic sextets (highlighted red) is more stable than an isomer with fewer. 86 Figure 7.10 structures are the isomers which allow the highest number of sextets. Overall Conclusion I have presented a stepwise analysis of a hydrotreated asphaltene by APPI FT-ICR mass spectrometry. The hydroprocessing analysis revealed no hydrocracking (carbon rejection) and an increase in DBE associated with sulfur reduction. Temperature changes demonstrated significant differences in sulfur reduction, and longer catalytic reaction time exhibited structural migration to stable condensed polycyclic aromatic compounds. One unanswered question remains surrounding sample A11 s elemental analysis which showed a 4.5 wt% sulfur content and yet the mass spectrum analysis was similar (low sulfur species) to A22 (1.6 wt% sulfur): Are the sulfur species in sample A22 unobservable by APPI? The answer to this question will require further experiments. APPI FT-ICR mass spectrometry will be a useful tool for petrochemical analysis! 108

130 APPENDIX A. CARBON CLUSTER STRUCTURAL CHARACTERIZATION BY GAS PHASE ION-MOLECULE REACTION IN AN FT-ICR MASS SPECTROMETER Fullerene Introduction Fullerenes are closed cage molecules consisting of 12 pentagonal and two or more hexagonal rings. Fullerenes with 60 carbon atoms or larger follow the isolated pentagon rule (IPR). 87 Smaller fullerenes consist of isomers with adjoined pentagon rings, e.g., fullerene C 50. Perhaps one of the more interesting small fullerenes is C 28. The most stable theoretical structure in part consists of four reactive carbons bonded in sp 3 orbitals located at the apex of triplet pentagons. 88 It has been suggested that fullerene C 28 could form a crystal lattice (hyperdiamond) because the four dangling bonds are centered at tetrahedral vertices. 89 Breda et al., suggest C 28 fullerene could possibly be a room temperature organic superconductor. 90 Guo et al. has provided experimental evidence that the tetravalent C 28 fullerene is stabilized with endohedral metals (U@C28). 91 Laser vaporization of graphite followed by supersonic expansion is known to produce stable (magic number) fullerenes, e.g., truncated icosahedral C Lower mass C n clusters (10 n 18) are reported to consist of linear chain and monocyclic ring structures, and higher mass clusters (32 n 60) are the beginning of fullerene structures. 93 Cluster mass distribution is source condition dependent. A bimodal distribution (C n (10 n 18) and (32 n 60)) or single distribution inclusive of C n (19 n 31) is possible. Small carbon clusters between the bimodal distribution (C 28 ) are the focus of this report. 109

131 Instrumentation Cluster Source An FT-ICR mass spectrometer with a split-pair 5.3 Tesla superconducting magnet 94 and associated ion optics and electronic equipment was purchased from Lucent Technologies, New Jersey. Figure A.1 is a sketch of the complete instrument configuration at Lucent Technologies. 10 flange Conflated Bore Cryo pump Cryo pump 73 Source Tesla X2 X Diff Pumps Diff Pumps Figure A.1. Diagram of the 5.3 Split-pair FT-ICR mass spectrometer (unshielded magnet) in its original configuration at Lucent Technologies (not to scale). The eight inch bore of the magnet is vacuum sealed and was designed with four access ports which allows trapped ion interrogation within the ICR cell. The instrument included an external cluster ion source. The cluster ion source produces laser ablated clusters in the presence of a continuous helium gas stream. The carbon clusters experience supersonic 110

132 expansion into an optics lens stack constructed with a mechanical 10 offset from the z-axis (bore axis of the magnet). Figure A.2 is a diagram of the optics. X Y Deflection Potentials Common Center Einzel Lens Extraction Potential (-100 to -900 V) Source Potential (5-50 V) Figure A.2. Ion optics lens stack. This lens stack is mounted to the source block. The first electrode (labeled source potential) is physically connected to the source block and, therefore at the same potential as the source. Note the mechanical 10 offset of the final two electrodes. The first electrode is electrically connected to the source block and the conical shape protrusion comprises the exit wall (when mounted to the source block) of the cluster formation chamber. The source electrode orifice is 1 mm in diameter, and other electrodes with larger orifices were manufactured (1.5 mm and 2 mm diameter) but not installed. All electrodes are mutually insulated. Nominal electrode potentials are as follows: source, +20 V; extraction lens, -110 V; center Einzel lens, -150 V; all other electrodes are held at ground potential. The mechanical 10 offset prevents a pathway for the cluster neutral beam to the ICR cell. The instrument was not configured with neutral beam ionization capability. 111

133 Gas Valves 1 Einzel Lens 2 Carbon Rod Carbon Laser Vaporization FT-ICR MS Laser Port Conductance Limits 1, 2, and 3 Accumulator Octopole Transfer Octopole Cell Pulse Gas ICR Cell 10-6 Torr 10-7 Torr 10-7 Torr 10-9 Torr Torr Cluster Source Existing 9.4 Tesla FT-ICR MS Figure A.3. Ion optics and differential pumping diagram. The zoom inset details the source block and static potential optics. Conductance limits 2 and 3 are held at trapping potentials during external ion accumulation. 112

134 Cluster Source Coupled to Existing 9.4 T FT-ICR Mass Spectrometer Figure A.3 is a diagram of the cluster source chamber coupled to the existing 9.4 Tesla FT-ICR mass spectrometer. 40 The original cluster source chamber was configured with three Edwards 250M diffusion pumps (ion source chamber) and a Varian (model unknown) diffusion pump for the second source chamber. Two source chamber ports were sealed with iso-250 blank flanges. The third source port and second chamber port were adapted for Pfeiffer Balzer turbo molecular vacuum pumps. When not in operation (no helium gas flow) and under vacuum, the mass spectrometer s pressure is indicated in Figure A.3. Figure A.4. Interface octopole ion guide. The octopole operated at MHz and ~300 volts peak-peak amplitude with a -30 V DC offset. The overall length was ~27 inches. 113

135 Figure A.4 is a diagram of the interface ion optics that coupled the cluster chamber to the mass spectrometer. The cluster source lens stack was positioned.030 inches from conductance limit 1 (CL1) orifice. The octopole shell is supported in two locations. The first support allows z-axis translation and x-y plane alignment and a set screw for a ridged mount near CL1. The second support, is a slip fit mount that allows z- axis translation and adjustment in the x-y plane without a set screw. The end of the octopole (opposite CL1) was positioned.050 inches from CL2. Interestingly, after installation of the interface octopole (A.4), the system was evacuated for the first time. The outer wall of the source chamber (surface facing the mass spectrometer) is constructed of 3/4 inch stainless steel. With the system under vacuum, the interface octopole rods contacted CL2. The 3/4 inch thick steel wall had deformed at least.050 inch and, therefore, translated the octopole toward CL2. The octopole was positioned an additional.030 inches (total.080 inch) from CL2 to compensate. The source target (carbon) rod (1/4 inch diameter, 5 inch length) is simultaneously rotated and translated. The original drive motor failed. A replacement motor was purchased from MicroMo Inc. (part number, 1516E012ST+15/8 1670:1+X0583). The motor rotates the rod at ~300 mhz and a translation distance of one inch. Micro limit switches positioned on the source reverse the motor direction, and thus, the motor drive system provides continuous rotation and translation of the target rod. An Nd:YAG laser (2nd harmonic) produces 25 mj pulses (1-15 Hz) to ablate the rod. The laser head is rigidly mounted, and an array of mirrors and a 500 mm focal length lens focused the laser on the target rod. Without rod rotation, the laser was fired multiple times to ablate a hole into the rod. The diameter of the hole was.035 inches. 114

136 Retarding Potential 600 Extraction -110 V, CL1-161 V Picoampere Source Potential +40 V Source Potential +30 V Source Potential +20 V Source Potential +10 V CL2 potential, Volt 600 Extraction -110 V, Source +30 V Picoampere CL1-160 V CL1-110 V CL1-60 V CL2 potential, Volt 600 CL1-160 V, Source +30 V 500 Picoampere Extraction Potential -50 V Extraction Potential -80 V Extraction Potential -110 V CL2 potential, Volt Figure A.5. Retarding potential profile. Total ion current (y-axis) is measured on accumulator octopole rods and CL2 potential (x-axis) is varied. CL2 is the accumulator entrance lens. Source, extraction, and CL1 potentials are plotted to investigate ion kinetic energy. 115

137 Retarding Potential Study The laser optics were adjusted to maximum ion current measured on the common lens (see Figure A.2, common). With continuous laser shots (10 Hz) and a continuous stream of helium into the source block, the maximum obtainable current (on the common electrodes) was 1.5 nano-amperes (measure with a Keithley pico-ampere meter). For this measurement, the extraction lens was held at -110 volts, and the helium gas leak rate was increased until the pressure in the source chamber was 1 x 10-4 Torr. For the retarding potential study, the pico-ampere meter was connected to the accumulator octopole rods. Source, extraction and CL1 potentials were varied sequentially (two potentials held constant and one varied) and pico-ampere (collected on the accumulator rods) versus CL2 potential were recorded. The source potential has the greatest affect for ion current collected on the accumulator octopole rods (Figure A.4 (top)). The source potential sets the initial ion kinetic energy. The extraction and CL1 potential have minimal (if any, Figure A.4 middle and bottom)) affect on accumulator ion current. However, the accumulator ion current is the total ion current composed of a broad cluster mass range. Experimentally, to optimize instrument parameters for a desired mass range, a source potential of volts was not always optimal. The average kinetic energy of the ion clusters is ~60 ev. Cluster Spectra Mass Range 116

138 The detected mass distribution can be altered by rf-octopole parameters and ICR cell trap plate timing. Preferred m/z ion abundance C 50 Carbon Target Rod 10 Acquisitions summed 200 Laser Pulses/Acquisition Continuous Helium Stream C 60 Carbon Laser Ablation FT-ICR MS Jeremiah Purcell Don Smith John Quinn m/z ,040 1,120 1,200 Figure A.6. Carbon cluster broadband FT-ICR mass spectrum. Spectra was collected on the initial day of instrument operation. Additional instrument parameter are reported in Table A.1. Table A.1. Spectra Instrument Parameters Fig. A.5 Fig. A.6 Fig. A.7 Source, Volt Extraction, Volt Einzel Lens a, Volt CL1, Volt RF Oct 1, MHz 2.0 (1 V) 1.4 (1 V) 2.2 (1 V) RF Oct 2, MHz 2.0 (1 V) 1.4 (1 V) 3.2 (1.6 V) RF Oct 3, MHz 2.0 (1 V) 1.4 (1 V) 2.2 (1 V) a Center Einzel lens 117

139 Single Acquisition 5 Laser Pulses Continuous Helium Stream 12 C C C 3 + 3,334 3,336 3,338 3,340 m/z 3,342 3,344 3,346 2,600 2,800 3,000 3,200 3,400 3,600 m/z 3,800 4,000 4,200 4,400 4,600 Figure A.7. Broadband carbon cluster mass spectrum (2600 m/z 4600). Instrument parameters favored accumulation and transfer of high m/z ions. The zoom inset shows the C278 monoisotopic peak and its isotopic variants. can be increased with longer accumulation times and instrument parameter adjustments. Figure A.6 is a representative spectrum acquired on the first day of operating the instrument. Spectral peaks at m/z 600 (C 50 ) and m/z 720 (C 60 ) are enhanced. Other instrument parameters for Figure A.6 (and other spectra figures) are reported in Table A.1. Figure A.7 is a high m/z broadband mass spectrum and was acquired by adjusting instrument parameters to optimize accumulation and transfer of high m/z ions. Note that Figure A.7 is a single acquisition mass spectrum, and Figure A.6 is 10 acquisitions summed. Both display similar signal-to-noise ratios. Furthermore, the spectrum 118

140 in Figure A.6 was acquired with an ion population from 200 laser pulses and Figure A.7 from 5 laser pulses. With the instrument configured for continuous helium sweep gas, the bulk of the ion population formed is high m/z ions. Nevertheless, with the instrument in a continuous gas stream configuration, a significant low m/z ion population could be preferentially accumulated. Figure A.8 is an example of the low m/z ion population centered at m/z 336 (C 28 ). Although the spectrum is a single acquisition, it required 400 laser pulses. C 28 + Single Acquisition 400 Laser Pulses Continuous Helium Stream m/z Figure A.8. Low mass carbon cluster mass spectrum. Although this is a single acquisition, 400 laser pulses were required to accumulate this ion population. In-Cell Gas-Phase Ion-Molecule Reactions In-cell ion-molecule reactions were accomplished by SWIFT (stored waveform inverse Fourier transform) isolation followed by cell pulse gas events. Lecture bottles of nitric oxide and chlorine were purchased from Sigma Aldrich and fitted with a regulator. The output of the regulator 119

141 was connected to a gas bomb with Sulfurinert 1/16 inch o.d. tubing. The gas bomb (~5 Torr gas pressure) was connected to the ICR cell region through a pulse valve (General Valve Co.). Figure A.9 represents the in-cell gas phase reaction of C 28 with nitric oxide (NO). The addition of one NO adduct to C 28 is seen at m/z 366. Other experiments with longer (and multiple) pulse gas duration [C 28 ] [C + 26 ] + C 28 SWIFT Isolated 30 msec Pulse NO gas m/z [C 28 NO] m/z Figure A.9. SWIFT isolated C28 with a 30 msec NO pulse gas event. The [C28NO] + spectral peak magnitude is 13% relative abundance. The loss of C2 ([C26] + ) spectral peak is 1 % relative abundance. (20 msec 50) showed no increase in the relative abundance (RA) of the [C 28 NO] + species. Additionally, the mass spectral peak corresponding to [C 28 H 1 (NO) 2 ] + at m/z 397 was detected at 1 % RA. The source of the hydrogen impurities are unknown. As stated earlier, the theoretical stable structure of fullerene C 28 contains four carbons atoms at the vertices of triple pentagon structures (Figure A.10). The reaction of fullerene C 28 at the four reactive carbon 120

142 sites (Figure A.10) should be exothermic. 88 Furthermore, the experimental conditions for Figure A.9 were such that the pulse gas event resulted in a momentary spike in the pressure to ~1 x 10-6 Torr (from 2 x Torr). At this pressure, the C 28 clusters undergo hundreds of collisions with NO molecules. However, in Figure A.9 the predominate reaction product is the addition of only one Figure A.10. Theoretical stable structure of fullerene C28. The four red carbon atoms are at the vertices of triplet pentagons. In this isomer form, the red atoms have sp 3 orbitals with a lone electron. adduct (NO) and not multiple additions of NO. Furthermore, a large percent of the parent ion ([C28] + ) remains unreacted. Figure A.11 A (top) is the spectrum of a double resonance SWIFT isolation experiment. C 28 was SWIFT isolated in the ICR cell, reacted with NO gas (30 msec pulse at 5 Torr), followed by another SWIFT isolation of C 28 and another pulse gas event. The reaction products of the first pulse gas event are ejected from the ICR cell by the second SWIFT isolation event. The second pulse gas event should produce more product [C 28 NO] + if more reactive species are present. However, only a small 121