An investigation of oxygen-containing compounds - phenols and indanols - in coal liquid distillates

A renewed emphasis on the development of liquid synthetic fuels from coal has prompted an increase in the variety of analytical methods employed for the chemical characterization of these products. These characterizations have been necessitated as product composition directly affects the choice of process variables, and inherently, the overall success of the process. Complete characterization, however, is often inhibited due to the complex nature of these liquids. The successful operation of any coal liquefaction process is ultimately dependent upon the efficiency of hydrogen consumption. Ideal levels of hydrogen consumption involve not only increased hydrogen to carbon (H/C) ratios of initially formed hydrocarbon fragments, but also removal of heteroatoms in reduced form, principally as NHg, h^S and 1^0. Of the major heteroatomic species found in coal, less attention has been devoted to the analysis of oxygen-containing compounds, although recent evidence indicates that oxygen functionalities may play important roles in the effectiveness of hydrogen utilization as well as in the yield of quality end-products. Several distillates from the Exxon Donor Solvent (EDS) and H-Coal processes were investigated for oxygen-containing compounds by a combination of analytical techniques including open-column liquid chromatography (LC), capillary gas chromatography (GC), gas chromatography coupled with mass spectrometry (GC/MS) and pyrolysis mass spectrometry (Py/MS). These analyses resulted in the tentative identification of several series of hydroxyaromatic and hydroxyhydroaromatic compounds. The two major classes of oxygen-containing compounds found in the distillates examined were a series of alkylphenols and a second series thought to consist of 4-indanol, 5-indanol and their respective homologs. This latter series was deemed noteworthy because of unexpectedly high abundances and the apparent reactivity of 5-indanol (and corresponding homologs) during hydrotreatment. Whether this reactivity occurred with a correspondingly beneficial or detrimental effect on the overall process has not been determined. Exact identities of the indanols could not be confirmed directly due to a lack of reference spectra. Since unambiguous identification of mass spectra usually requires matching unknown spectra with spectra obtained from standard compounds and/or reference libraries, and only a limited number of suitable standards were available, computerassisted pattern recognition was used in an attempt to assign mass spectra of these "unknown" species to correct compound classes. Selected spectra from GC/MS analyses were extracted to form two sets of data. The first data set contained spectra of a variety of hydrocarbons as well as spectra of suspected alkylphenols and alkylindanols and was used to examine the degree of separation possible between hydroxy- and nonhydroxy-containing compounds. The second data set consisted of spectra of suspected alkylphenols, alkylindanols and several reference compounds and was used to examine the degree of separation possible between these classes and the indanol relationships to isomers of differing molecular structure. Examinations were achieved through the formation of data files of two types. From each original data file (containing fragment ion spectra) a corresponding "reverse" file (containing neutral-loss spectra) was formed. Original files were used to obtain molecular core information while the "reverse" files were employed to investigate fragmentation tendencies of the various classes. Each data file was subjected to specific data processing routines, as available in the ARTHUR computer program package, including factor analysis, discriminant analysis and non-linear mapping. The combined results obtained from these analyses indicated separation not only between the various classes, but in certain instances, between class homologs and isomers of the same molecular weight. In spite of the absence of sufficient reference spectra, the classification of spectra into correct categories was successful and further validated previously assigned class identities.

- - by Gary Steven Metcalf A thesis submitted to the faculty of The University of Utah in partial fulfillment of the requirements for the degree of Master of Science Department of Chemical Engineering The University of Utah March 1983 AN INVESTIGATION OF OXYGEN-CONTAINING COMPOUNDS AND INDANOLS -- IN COAL LIQUID DISTILLATES Hetcalf PHENOLS © 1983 Gary Steven Metcalf AIT Rights Reserved 1983 All SCHOOL SUPERVISORY COMMITTEE APPROVAL Gary Steven Metcalf satisfactory. December 22. 1982 December 22. 1982 December 22. !982 j&M^UiA^LLr^ Henk L. C. Meuzelaar "i (; . ~ , >, ! ~ I,' THE UNIVERSITY OF UTAH GRADUATE SCHOOL of a thesis submitted by This thesis has been read by each member of the following supervisory committee and by majority vote has been found to be satisfactory . 1982 22, 1982 ~ J. D. Seader December 22, 1982 Meuze1aar THE UNIVERSITY OF UTAH GRADUATE SCHOOL F I N A L R E A D I N G A P P R O V AL To the Graduate Council of The University of Utah: 1 Gary $t?yen MetCfllf in its final form and have found that (1) its format, citations, and bibliographic style are consistent and acceptable; (2) its illustrative materials including figures, tables, and charts are in place; and (3) the final manuscript is satisfactory to the Supervisory Committee and is ready for submission to the Graduate School. December 22, 1982 D a , e Henk L. C. Meuzelaar Member. Supervisory Committee Approved for the Major Department A. Lamon^Tylejj^ Chairman Dean Approved for the Graduate Council James L. Clayl Dean of The Graduate School . ~, FINAL READING APPROVAL I have read the thesis of Ga ry Steven Metcalf In Its I) format. citations. tables. December 22. 1982 Dale S<;hool ABSTRACT A renewed emphasis on the development of liquid synthetic fuels from coal has prompted an increase in the variety of analytical methods employed for the chemical characterization of these products. These characterizations have been necessitated as product composition directly affects the choice of process variables, and inherently, the overall success of the process. Complete characterization, however, is often inhibited due to the complex nature of these liquids. The successful operation of any coal liquefaction process is ultimately dependent upon the efficiency of hydrogen consumption. Ideal levels of hydrogen consumption involve not only increased hydrogen to carbon (H/C) ratios of initially formed hydrocarbon fragments, but also removal of heteroatoms in reduced form, principally as NHg, h^S and 1^0. Of the major heteroatomic species found in coal, less attention has been devoted to the analysis of oxygen-containing compounds, although recent evidence indicates that oxygen functionalities may play important roles in the effectiveness of hydrogen utilization as well as in the yield of quality end-products. Several distillates from the Exxon Donor Solvent (EDS) and H-Coal processes were investigated for oxygen-containing compounds by a com­bination of analytical techniques including open-column liquid chromatography (LC), capillary gas chromatography (GC), gas chromatog­raphy coupled with mass spectrometry (GC/MS) and pyrolysis mass . '" -'- li,- ~ NH3, H2S H20. containing containing GC/MS) spectrometry (Py/MS). These analyses resulted in the tentative identi­fication of several series of hydroxyaromatic and hydroxyhydroaromatic compounds. The two major classes of oxygen-containing compounds found in the distillates examined were a series of alkylphenols and a second corresponding homologs) during hydrotreatment. Whether this reactivity occurred with a correspondingly beneficial or detrimental effect on the overall process has not been determined. Exact identities of the indanols could not be confirmed directly due to a lack of reference spectra. Since unambiguous identification of mass spectra usually requires matching unknown spectra with spectra obtained from standard compounds and/or reference libraries, and only a limited number of suitable standards were available, computer-assisted pattern recognition was used in an attempt to assign mass spectra of these "unknown" species to correct compound classes. Selected spectra from GC/MS analyses were extracted to form two sets of data. The first data set contained spectra of a variety of hydrocarbons as well as spectra of suspected alkylphenols and alkyl-indanols and was used to examine the degree of separation possible between hydroxy- and nonhydroxy-containing compounds. The second data set consisted of spectra of suspected alkylphenols, alkylindanols and several reference compounds and was used to examine the degree of separation possible between these classes and the indanol relationships to isomers of differing molecular structure. Examinations were achieved through the formation of data files of two types. From each MS). hydroxyaromat;c hydroxyhydroarol11atic containing alkyl phenols series thought to consist of 4-indanol, 5-indanol and their respective homologs. This latter series was deemed noteworthy because of unex­pectedly high abundances and the apparent reactivity of 5-indanol (and Sincs frool computer­assisted cl asses. ' alkyl­indanols containing v original data file (containing fragment ion spectra) a corresponding "reverse" file (containing neutral-loss spectra) was formed. Original files were used to obtain molecular core information while the "reverse" files were employed to investigate fragmentation tendencies of the various classes. Each data file was subjected to specific data processing routines, as available in the ARTHUR computer program package, including factor analysis, discriminant analysis and non-linear mapping. The combined results obtained from these analyses indicated separation not only between the various classes, but in certain instances, between class homologs and isomers of the same molecular weight. In spite of the absence of sufficient reference spectra, the classification of spectra into correct categories was successful and further validated previously assigned class identities. vi linear { CONTENTS ABSTRACT iv LIST OF TABLES ix LIST OF SYMBOLS x ACKNOWLEDGEMENTS xf 1i 1 Background Perspectives 1 Development of Coal-Liquefaction Processes . . 4 Analyses 8 Motivation for Investigations of Oxygen-Containing Compounds . 13 Sample Collection and Process Description . . . . . 17 H-Coal (hydro!iquefaction) process [11, 14, 29] 18 Exxon Donor Solvent (solvent-extraction) process [11, 14, 15] 19 20 ANALYTICAL METHODS AND RESULTS 22 22 Injection . 28 Basis of Curie-point technique 29 31 Operating Conditions for Curie-point GC Analyses 38 Preliminary GC Results 39 Pre-separation Methods 41 Solvent-extraction technique 43 45 Description of GC/MS System 50 GC/MS Analysis of Whole Liquid Distillates 52 GC/MS and Direct-Probe, Low-Voltage MS Analyses of Whole Liquid Distillate Sub-fractions 63 DATA PROCESSING ...... 81 Parametric Methods (PBM-STIRS) 81 parametric ARTHUR) . . 84 ... INTRODUCTION • . • . . xii; . . . . • . . . . . . . . ... . General Trends in Coal Liquid,Analyses ......... . Containing ..... . hydro1iquefaction) . ll,14,15J • . . . . . . . ... University of Utah process .. Capillary Gas Chromatography ............... . Curie-point Flash Vaporization Device for Splitless Injection. ....... . Steps in the Curie-point method ....... . ........... . ............• ....... . Open-column liquid chromatographic technique Descri pti on . . . . . . . . . . . ..... . Voltage fractions . . . . . . . . . . r~ethods PBM-Non-parametric Methods (ARTHUR) Unsupervised learning 84 learning 120 CONCLUSIONS . 130 Development of Analytical Techniques 130 of Analytical Results 131 Computerized Pattern Recognition Results 133 APPENDIX 135 Introduction to Computerized Pattern Recognition Techniques . . 135 Parametric Methods (PBM-STIRS) . . . . 138 parametric ARTHUR) 138 Pre-processing . . . . 141 Scaling 141 Feature weighting . 143 Feature reduction via the Karhunen-Loeve transform ... 144 Linear and Non-linear Display Methods 147 149 The k-nearest neighbor method 150 Multicategory [112] . 151 LITERATURE CITED 152 vi i i ,.' . Supervised learning. . • . • . . . Interpretation APPENDI X 0 0 0 0 0 • 0 0 0 • • 0 0 0 0 0 0 0 0 • PBM-Non-parametric Methods (ARTHUR) . 0 0 0 0 0 0 o. 0 0 0 0 0 • . 0 0 0 0 0 0 0 ~ 0 •• 0 •• linear Methods. 0 0 • Supervised Learning or Discriminant Analysis 0 0 0 •• 0 0 0 0 0 • 0 • • • • • • • • • SIMCA (Statistical Isolinear r~u1ticategory Analysis) 112J . . . . . . . . . . 0 • • • • • • • 0 • • • • • • vi;; LIST OF TABLES Table I Process Sample Descriptions 22 II 28 49 Fraction of EDS Atmospheric Bottoms and Hydrotreated Atmospheric Bottoms 75 V (Alkyl)Phenol/(Alkyl)Indanol and Multi-class Data Members . . 102 VI Twenty Features (Mass Peaks) with Highest Total Variance Contribution to Factors I and II Extracted from the Multi-class Fragment Ion Data Set 109 VII Twenty Features (Neutral-loss in Dal tons) with Highest Total Variance Contribution to Factors I, II and III Extracted from the Multi-class Neutral-loss Data Set . . . . 115 Mass Contribution to Factors I and III Extracted from the (Alkyl)Phenol/(Alkyl)Indanol Fragment Ion Data Set 119 Twenty Features (Neutral-loss in Dal tons) with Highest Total Variance Contribution to Factors I and II Extracted from the (Alkyl)Phenol/(AlkylJlndanol Neutral-loss Data Set 133 X k-Nearest Neighbor Classification* (k 1, 3, 5, 7, 9) . . . 139 SIMCA Misclassification Matrix 141 ... GC Liquid Substrate Description .. III LC Fractionation Yields .... IV Hydroxyaromatic Compounds Identified in Benzene/Ether . . . . . . . . . . . . . . . . . Members. .............. Daltons) .. VIII Twenty Features (Mass Peaks) with Highest Total Variance A1 kyl )Pheno1/(A1 kyl ).. .. IX XI Daltons) Alkyl)Alkyl)Indanol ........ . Classificat;on* = 1,3, 5,7,9) ... . . . LIST OF SYMBOLS AND ABBREVIATIONS • (Alkyl)phenols tentatively identified from GC/MS analysis of EDS atmospheric bottoms distillate. • (Alkyl)indanols tentatively identified from GC/MS analysis of the EDS atmospheric bottoms distillate. t Aliphatics tentatively identified from GC/MS analysis of the A Hydroaromatics tentatively identified from GC/MS analysis of O Naphthalenes tentatively identified from GC/MS analysis of EDS atmospheric bottoms distillate. % Aromatics tentatively identified from GC/MS analysis of the EDS atmospheric bottoms distillate. O (Alkyl)phenols tentatively identified from GC/MS analysis of the EDS atmospheric bottoms hydrotreated distillate. • (Alkyl)indanols tentatively identified from GC/MS analysis of the EDS atmospheric bottoms hydrotreated distillate. 4-indanol reference compound. V • * 1-indanol reference V * 1-hydroxytetralin Y Magnetic susceptibility. A Eddy diffusion term of HETP parameter. ACQUIRE/- HP5930A GC/MS system program for the recording/processing of SPEED MS ARFMT Computer program for data normalization and ARTHUR formatting. ARTHUR Computer program package used for univariate and multi-data analysis. ;'f i .~; :'\' , .. Alkyl)phenols MS atmospheri c di still ate. Alkyl)I,1S • t<lS EDS atmospheric bottoms distillate. ~ the EDS atmospheric bottoms distillate. ct tt MS atmospheri c di sti 11 ate. o CJ Alkyl)MS CJ+ \7 5-hydroxytetralin reference compound. []* l-indanol reference compound. \7* l-hydroxytetralin reference compound. l' suscepti bil i ty. GC/MS spectral data. a.u. Arbitrary units. B Longitudinal diffusion term of HETP parameter. ethyl ether B/M Benzene/methanol extracted distillate sub-fraction. C Resistance to mass transfer term of HETP parameter. CONAES Committee on Nuclear and Alternative Energy Systems. CORNY Computer program used for the extraction of selected GC/MS fragment ion spectra and the creation of corresponding neutral-loss spectra. EDS Exxon Donor Solvent coal-liquefaction process. FID Flame ionization detector used in GC analysis. B H-Coal Hydrocarbon Research coal-liquefaction process. HETP Height-equivalent of a theoretical plate. hf High frequency. I Intensity of object magnetization. K-L Karhunen-Loeve data transform. KARLOV ARTHUR sub-program for Karhunen-Loeve data transformation KNN ARTHUR sub-program for k-nearest neighbor data analysis. MASSA Computer program for the assignment of theoretical (integer) m/z to the recorded (real number) m/z of Py-MS spectral data MULTI Data set of two file types (fragment ion spectra and neutral loss spectra) used to contrast hydroxy!- with nonhydroxyl-containing compounds. N/L Number of theoretical plates per unit length of chromato­graphic column. N/L Ratio of the number of objects (spectra) to the number of measurements made on those objects. xi B/E Benzene/ethylether extracted distillate sub-fraction. M FlD H Hydrogen. H Magnetic field strength. H/8 Hexane/benzene extracted distillate sub-fraction. 1 (prinipal component analysis). mlz MS data. neutral­loss hydroxyl- nonhydroxyl­containing NIL (m/z) v ~<: .I.( NLM ARTHUR sub-program for nonlinear mapping data analysis. PBM Probability Based Matching computer program for the assignment of mass spectral identities. PHI8 Identification label of contaminated alkylphenol spectrum. PH/IN Data set of two file types (fragment ion spectra and neutral-loss spectra) used to contrast alkylphenols with (alkyl)- indanols. PYRO Computer program for the recording/processing of Py-MS spectral data. SCOT Support-coated open-tubular capillary column. SCT Short-contact-time coal-liquefaction process and/or products. SCALE ARTHUR sub-program for measurement (feature) normalization. SELECT ARTHUR sub-program for extraction of statistically inde­pendent measurements (features). SIMCA Statistical Isolinear Multicategory Analysis computer program (computer sub-program as available in ARTHUR) for intra-category principal component analysis. STIRS Self-Training Interpretive and Retrieval System computer pro­gram for the detection of spectral sub-structures and assignment of mass spectral identities. Tc Curie-point temperature. TIC Total Ion Current chromatogram. v Carrier gas linear velocity. VARVAR ARTHUR sub-program for two-dimensional data plots. WEIGHT ARTHUR sub-program for the determination of measurement (feature) weights according to a criterion of reliable category separation. xii PH18 neutral­loss i ndanol s. MS intra­category tWO-dimensional ACKNOWLEDGEMENTS The author wishes to express his gratitude to Dr. George R. Hill who instilled the author with confidence, provided him this research opportunity and was responsible for the overall direction and super­vision of this study. Deep appreciation is extended to Dr. Henk L. C. Meuzelaar for his personal involvement throughout all phases of this thesis. The successful outcome of this project is due in no small part to the detailed instructions, criticisms and suggestions of Dr. Meuzelaar, for Thanks is also extended to Dr. Jean H. Futrell and Dr. Alice M. Harper for their suggestions and support in this work and to Dr. J. D. Seader and Dr. Reginald F. Cane for their critiques of this manuscript. Gratitude is also expressed to the staff of the Biomaterials Profiling Center for their expert assistance and cooperation. In this regard, the author is particularly indebted to William H. McClennen for his assistance and supervision in sample analyses and to David L. Pope for his assistance in compound identification and in the develop­ment of computer programs. The financial support provided by the Department of Energy is gratefully acknowledged as well as the contributions of Exxon Research and Engineering Co., Hydrocarbon Research Inc. and the University of Utah, Department of Mines and Fuels for providing coal liquefaction samples. wishei which the author is indebted. ~ INTRODUCTION The energy demands of the United States during this century have been and continue to be enormous. In 1980, petroleum, natural gas, and coal satisfied over 90% of the estimated 78 quads (quadrillion BTUs) used by this country, and of these three, petroleum was responsible for nearly half of the total U.S. consumption [1]. At that time, although the United States was the world's third largest producer of petroleum (behind the Soviet Union and Saudia Arabia), nearly twice as much petroleum was consumed domestically as was produced [2]. This partial dependency upon other countries (many of which were/are potentially unstable) for a significant portion of our energy supplies resulted in the gradual implementation, since 1973, of programs aimed towards providing energy independence. These programs have been centered primarily in two areas: 1) conservation and 2) providing incentives for increased development of both conventional and non-conventional fuel supplies. Despite the recent successes in energy conservation, the United States remains the world's largest consumer of energy [3]. Further­more, the economic costs of importing petroleum have been negligibly affected as reduced levels of imports have largely been offset by increases in price. However, the political gains scored through re­duced import levels may be substantial, and should not be ignored. Background Perspectives lJ. Stat~s 3J. 2 Efforts to stimulate the production of conventional fuel supplies of Federal price controls instigated in 1971 resulted in a record number of oil and gas wells drilled in 1980. Over 2.5 million wells have been drilled in the U.S., four times as many as in the rest of the world, yielding some 500,000 producing wells [1]. However, the large number of U.S. wells is largely offset by lower rates of "maximum sustainable production" as increases in production beyond these rates can damage oil fields by reducing pressure too quickly and thus render future supplies inaccessible [2]. In this regard, the average U.S. well is outproduced by its Saudi a Arabian counterpart by a.factor of nearly 800 [1]. Therefore, providing incentives for increased drilling addresses only part of the problem. Many feel that the key to the energy problem facing the U.S. is the increased production and use of coal. Coal is the world's most abundant fossil fuel and three countries, the U.S., the U.S.S.R., and China, own nearly 90% of total estimated resources [4]. Of these three, the U.S. has the largest share of coal reserves with more than one-fourth of the 786 billion tons of known world reserves [1]. It is somewhat of a paradox that coal, the most abundant fossil fuel resource of this country, contributed less in 1980 to total domestic energy usage than either petroleum or natural gas [3]. This was not always the case as coal was the primary source of energy for this country from the mid 1880s until 1950, when it was surpassed by petroleum [2]. Interestingly enough, two years before petroleum reached this level of dominance as the main U.S. energy source, the stage for the recent "energy crisis" was set. In 1948, the U.S., almost unnoticed, became a have enjoyed varying degrees of success. For example, the phasing out lJ. production ll i~accessible 2J. Saudia 4J. lJ. IllOSt 3J. . 2J. lIenergy net importer of oil [1]. Not only was the oil relatively cheap, but it became even cheaper during the 1950s and 1960s. In constant 1948 dollars, the price per barrel of crude oil at the wellhead actually fell from $2.50 in 1948 to $1.85 in 1972 [3]. In addition to the economic factors, oil was also convenient, particulary with regard to the growing demands of transportation. The above factors helped pro­mote an ever increasing demand for petroleum and a corresponding decrease in the reliance on coal. The energy crisis facing the U.S. has been, in large part, a crisis in transportation. Almost half of all the oil used in the U.S. is used in transportation [5]; one out of every nine barrels of oil used in the world is burned in American auto engines [1]. It is in this light that the production of liquid synthetic fuels from coal, oil shale and tar sands is receiving renewed emphasis. These interests are not new, but their development has been invariably limited by access to ample supplies of petroleum. The situation con­fronting the United States may perhaps best be described by the introductory remarks of the CONAES report [3]: "The energy problem now faced by the U.S. began to be recognized 10 years or more ago. Still, occasional symptoms (the oil embargo of 1973, the natural gas shortage of 1976-1977, and the gasoline lines of the summer of 1979) are fre­quently mistaken for the problem itself. As each symptom is relieved, the public sense of crisis fades. The seeds of future crisis, however, remain." The idea that synthetic fuels production will play a role in the U.S. energy picture is becoming more and more feasible. However, to 3 lJ. 19505 3J. particu1ary " pro-mote 1 1 ,; 5J; ,i.;. r lJ. ~;' J, i!,~· ~' r ·}~' ) ~j I,' con- ,'" i;;': fronting l, 3J: liThe t rH ;l~j:::, Sti 11 , ':d;'+;' fre-quent1y remain. II playa 4 what extent and over what time frame this role will develop is, at present, uncertain. The successful commercial production of synthetic fuels will be based largely upon interacting political, sociological, environmental and economical constraints. These constraints have already, to some extent, been reduced by advances in technology, and the potential exists for even further reductions as coal-liquefaction technology progresses. Development of Coal-Liquefaction Processes Of all the fossil fuels, coal generally has the greatest number of carbon atoms per unit weight, which in turn, is responsible for the higher energy content exhibited by most coals at the expense of a less fluid physical state. The structural and chemical differences between oil and coal are due principally to their different origins. The dominant sources of oil are believed to come from the "fossilization" of simple marine organisms, while postulations as to the origins of coal have been based largely on petrographic studies which imply the biodegradation of plant material [4, 6]. The exact structural precursors of coal are a matter of debate, although Given has presented impressive arguments supportive of 1ignin-type origins [7, 8]. Regardless of the specific coal precursors, the fact that both coal and petroleum contain carbon backbones, but differ principally in the magnitude of the hydrogen to carbon (H/C) ratio (generally, 0.7 for coal, and greater than 1.2 for petroleum), as well as, the greater number of heteroatomic moieties (present mainly as oxygen, sulfur, and " ", , .I', ," L 6J. lignin-8J. a~ 5 nitrogen functional groups) in coal [4] makes it theoretically and practically possible to derive liquid fuels from coal by either adding hydrogen through hydroliquefaction or solvent-extraction, or by re­moving carbon through pyrolysis or carbonization. Possibly the first evidence for the potential of coal-conversion to liquid fuels came as early as 1869 when Berthelot reported that coals could be substantially converted to liquids [9]. However, it was not until 1913 that the first practical methods for coal-1iquefaction were reported by Bergius [10]. In those studies, Bergius and his co­workers utilized a high-pressure, high-temperature catalytic process to produce liquid fuels by reacting pulverized coal with gaseous hydrogen [11]. As investigations continued, a coal-oil slurry was used as feedstock, particularly in later commercial operations, to reduce the severe erosion of equipment that occurred under these operating condi­tions with the dry coal [4]. For these studies, Bergius was awarded the Nobel Prize in Chemistry in 1931. The hydrogenation method utilized in the Bergius process was/is one of two general routes for the direct addition of hydrogen to coal (the other being based upon "solvent-extraction") and is somewhat loosely termed "hydroliquefaction" [11]. Bedson had earlier laid the foundation for solvent-extraction processes when he discovered in 1902 that bituminous coals are sub­stantially soluble in hot pyridine [4], The investigation into the action of solvents was expanded by Fischer and Gluud, who discovered that yields 10 to 20 times greater could be achieved by extracting coal with benzene at 250°C, rather than at the normal boiling point of benzene. Various solute/solvent mixtures at elevated temperatures 4J 9J. liquefaction lOJ. co-workers :; llJ. ( ," .. ;.' ~" 1 ; \ \" condi-tions 4J. sub-stantially 4]. , {:, ¥". 6 (above 200°C) were investigated by others, culminating in commercial 1944 11]. Broche Pfirrman hydrogen-donor" H-C Pfirrman Broche Broche H-with the Bergius process) for producing much of the fuel used by Germany during World War II [13]. also achieved from operations based on the indirect addition of hydro­gen Tropsch thermodynamically, since all carbon-hydrogen bonds first had to be broken, and then were selectively reformed [14]. Interestingly enough, the Fischer-Tropsch process forms the basis of operations for the SASOL plants in South Africa, the only large-scale coal-liquefaction facil­ities in the world [13]. The success of these plants has been due, in large part, to the aggressive position taken as early as the 1950s by the South African government toward synthetic fuels development [1]. operations during 1938-1944 in Germany [llJ. These German developments were based largely on two similar schemes; the Pott-Broche process and the Uhde-Pfirrman process [12]. Both processes employed a "hydrogen­donorll (H-donor) solvent at high pressures with temperatures of 410°C to 430°C. The Uhde-Pfirrman process, however, added gaseous hydrogen to the coal-solvent mixture and operated under higher pressure (30 MPa vs. 10-15 MPa). The resulting product was of lower boiling range when compared to extracts of the Pott-Broche process [4]. These processes (the Pott-Broche process and the Uhde-Pfirrman process), which employed the direct addition of hydrogen, were fore­runners of the hydrogen-donor (H-donor) solvent technologies under current development in the United States, and were responsible (along 13J. To a lesser degree, fuel production for the German war effort was using the Fischer-Tropsch process [4]. This method first gasified the coal before the addition of hydrogen. When compared to the appro­aches of direct hydrogenation, this process was quite inefficient 14J. Tropsch 13J. 1J. However, there are other contributing factors that are unique to that country's political situation. Apartheid policies initiated by South Africa have contributed to relatively low costs in labor. These same policies have been protested by other governments, with embargos in goods and services imposed to varying degrees. In this respect, South Africa into situation demanding somewhat similar to that experienced by Germany during World War II. Therefore, a SASOL-type (Fischer-Tropsch) coal-liquefaction plant operating outside these circumstances would not be expected to compete economically in the current international free world market. In contrast to South Africa, the United States has had more opportunity to refine and develop improved technologies for coal- 1iquefaction. However, as mentioned earlier, interest in coal-liquefaction in the United States has been cyclic, reaching in­tense levels of investigation only during periods of potential fuel shortage [3]. As a consequence, investigative efforts in the past have been frustrated because the development of new technologies has gener­ally been followed by a waning of public interest. Later, with research efforts renewed, the significance of discoveries made in the past have been diminished since the production of liquid fuels from coal must then be developed under new economic constraints. Nonethe­less, these earlier advances have allowed the U.S. to be more selective in the variety of technologies available, with the result that those reactions involved in the classic German technologies are now incor­porated under conditions of less severity [13, 16]. Some of the more promising processes currently being pursued in this country utilize the dissolution of coal in a solvent capable of r",',:,:',:',' ( 7 I' ~ ~;: t { i~ ~:, K Afr; ca has been forced ; nto a s i tuati on demand; ng energy independence :V,' type Tropsch) coal-liquefaction. in­j tense ,~ ~ ;\ ~ 3J. '~:~ i ~ gener-ally < " ; Nonethe-less, incor-porated 8 hydrogen donation. Although the slurry oils used in the Bergius proc­ess of the 1940s contained potential H-donors, these classic plant processes depended almost entirely on reaction between coal and molecular hydrogen, since the higher temperatures employed in these processes tended to discriminate against efficient hydrogen transfer between solvent and coal [17], Because current processes under develop­ment place greater reliance on hydrogen transfer from the slurry oil to the coal they can be operated under conditions of improved thermal efficiency and hydrogen utilization [11]. of undesirable components, including ash, unreacted materials, sulfur and other heteroatoms; and (3) the relative level of hydrogen con­sumption [18]. These areas are not independent, but overlap with each other and all are contributors to the key economic factor in coal- 1iquefaction: the efficiency of hydrogen utilization [9]. Therefore, data obtained basic to these areas play important roles in understanding the mechanisms of reaction involved and thus help formulate improved processes of refining and upgrading for coal liquids. General Trends in Coal Liquid Analyses In order to better understand the chemistry of conversion that takes place during liquefaction, much effort has been directed towards the chemical characterization of the molecular species involved. As a reactant in liquefaction (and other processes), coal has been the sub­ject of intense scrutiny. However, relatively few chemical or physical , 17J. llJ. Three basic areas are involved in determining the successful t , operation of any coal-liquefaction process: (1) the rate of conversion to soluble products and the maximum yield obtainable; (2) the removal t:, \',:,' " f{/ ~', F con- ~.~."".,,' sumption [18]. These areas are not independent. but overlap with each l·. a 11 contri butors economi c coa 1- ~* ' ~; r{', : ~.', ~({ . ])',' r .~\·I ~,l liquefaction: 9J. Liguid ~;. ;;:. ), i,:,·~"i,." takes place during liquefaction, much effort has been directed towards ~, sub-ject methods (e.g., X-ray diffraction [19], C-13 NMR [20], etc.) can be used to study solid coal directly [21]. Therefore, the majority of efforts to structurally characterize coal in the past have been indirect in nature. These methods have included the development of model compounds [7, 22, 23] which simulate the organic properties of coal, the study of fragments obtained from coal pyrolysis [24], and the analyses of liquid products obtained from various methods of coal dissolution. The most common of these dissolution methods has been oxidative degradation [9] . However, analyses of the corresponding products obtained through re­ductive hydrogenation, e.g., liquefaction products, are useful since the structures of these solublized species, in theory, also provide a basis for establishing the structure of the parent coal [21]. To what extent the products of dissolution represent the actual coal structure is uncertain due to the possible occurrence of non-specific secondary reactions [25]. Nonetheless, some fragments representative of coal structures are expected to remain intact. Besides providing insights into the chemical structure of coal, the analyses of coal-liquefaction products are useful in their own right. The determination of an accurate product composition for any fuel resource is desirable in order to effect an optimal adjustment of process variables. For these coal liquids, however, a detailed de­scription is deemed necessary, more so than with conventional energy resources, due to their increased level of complexity and the resultant product upgrading required [26]. Further, the solvent used in major H-donor processes for dissolution of the coal is derived from these liquefaction products [27, 28] and as such, represents a process re-actant. The composition of this "recycle solvent" directly affects the 9 19J, 20J, 21J. 23J 24J, 9J. re-ductive solub1ized 21J. specific 25J. de-scription 26J. 28J re­actant. 10 selectivity of reaction, thus determining the yield and quality of various end products and is considered the primary factor in controlling the overall behavior of these coal-1iquefaction processes [9]. Just as efforts to characterize coal are hampered by its hetero­geneity, the complex nature of coal liquids makes the task of analyzing these products a formidable one. For this reason, the characterization of coal liquids in the past has been primarily directed toward product separation into distinct fractions based on solubility characteristics [9]. The most common of these schemes yields a pentane-soluble fraction (oils), a benzene-soluble fraction (asphaltenes), and a pyridine-soluble, benzene-insoluble fraction (asphaltols) [29, 30, 31]. The two former terms originated from similar classifications in the petroleum industry, but the benzene-insoluble fraction is generally absent from petroleum feedstocks [9]. This separation scheme results in fractions containing compounds of the same or similar class, although component carry-over can and does occur to varying degrees dependent upon the species present in the original material [32]. The first two fractions have been extensively characterized by methods derived from the petro­leum industry [33]. The asphalto! fraction is concentrated with heteroatoms, making it highly polar and difficult to analyze [31]. The fact that the composition of the first two fractions is similar to those obtained from petroleum feedstocks, but the third is not and is concentrated in heteroatoms emphasizes one of the basic differences between coal and petroleum, and hence highlights a major problem in the processing of coal liquids. While a variety of solvents can solubilize coal at low temperatures [25], the successful dissolution of coal into products useful for r y. liquefaction 9J. aspha1tenes), pyridine­soluble, insoluble aspha1to1s) insoluble 'containing asphaltol ;n 11 further upgrading requires this solubilization be conducted at tempera­tures congruent to the thermal decomposition of coal and has thus been termed "extractive disintegration" [34], or "destructive distillation" [4]. Overall, these thermal reactions require high temperatures be­cause of the relatively high strengths (50 Kcal/mole) of the bonds that must be broken [35]. (An actual range of bond strength in coal appears to exist [36]). During the initial stages of liquefaction, weak linkages in the coal matrix, generally associated with aliphatic [37] and/or etheric [38] regions of the coal are broken by a thermal disruption of these bonds. This disruption results in coal particles or "micelles" that vary in size and in strength of mutual bonding [39], and it appears that a significant portion of the coal remains within these particles [36]. The initial products which result from the rupture of these labile bonds closely resemble the materials which can be extracted from coal directly with pyridine at low temperatures [40] and are thus often referred to as asphaltols [31] or pre-asphaltenes [41]. Analyses of yields from short-contact-time (SCT) liquefaction experiments have indicated that where conversions are reasonably high, the primary products obtained are asphaltenes and asphaltols [9]. While the major product constituents can vary somewhat unpredictably with coal rank [40, 42], they are always rich in asphaltols, e.g., in the case of some bituminous coals, up to 85% asphaltols [35]. The transition between short-contact and long-contact times follows a similar course for all coals, as asphaltols, which are the predominant products of coals are converted simultaneously to oils, lIextractive 34J, distillation ll mole) 35J. 36J). if matri x, ali phati c 37J etheri c ., ~. 38J ie tr. bo n d s. IImicelles" 'i t~, r.1: ~:, ~. f f ~. I.:· .'~;'.' ,'). 39J, 36J. 40J ~. aspha1tols 41J. ~:> SeT) 9J. 42J, o 35J. ~~ .. ~. ~. ~:: );\'.: h' 12 asphaltenes, and insoluble residue [31]. In this regard, asphaltols can be considered as an intermediate product [41]. Investigations have indicated that several reaction routes are available to asphaltols, and that process conditions as well as intermediary product composition are crucial in determining end-product quality and yield [31]. Results such as these have led to the speculation that the process of coal-liquefaction proceeds through sequential stages [43], even though degrees of competition between the reactive stages seem to exist [44]. The initiation of coal-liquefaction, therefore, has been postu­lated to occur with thermal cleavage of bonds in the coal matrix followed by hydrogen transfer and resultant coal defunctionalization [4]. As bonds are thermally broken, the free radicals subsequently formed can be stabilized through at least three different pathways: (1) rearrangement and elimination, (2) addition to aromatics, usually with an increase in "average molecular weight" and char formation, and (3) the abstraction of hydrogen from one of several possible sources [9]. This hydrogen abstraction may be from the donor solvent, gaseous hydro­gen, or the coal itself [45, 46, 47]. The most efficient means for hydrogen transfer, in terms of yield of desirable end product, is via hydroaromatics present in the donor solvent [9], although at times, it appears abstraction from the coal is preferred ("intra-hydrogen trans­fer") [47]. Therefore, the stabilization of free radicals via the "preferred" route of hydrogen-transfer is affected not only by overall hydroaromatic concentration in the donor solvent, but also by the accessibility of potential H-donors to localized zones of reaction. This implies that in addition to having good chemical properties (e.g., capability of H-donation), the donor solvent should also possess good ~. 31J. ~ ~,' 41], ~. ~~. 1111 ~L II:. 1'., ;.. \ 31J. 43J, i. l~t' :i(1 44J. "i:, ~~~ . postu-lated Ilaverage 9J. hydro­{ gen, 47J. :~i ;s I", ;i: 9J, hydrogen 47J. transfer overa 11 13 physical properties (e.g., the ability to swell the coal [48], or actually break weak bonds [49]) to promote hydrogen accessibility from desired sources. In examining these solvent properties, efforts have naturally been focused upon analyses of specific compound classes to determine their role(s) in these H-donor processes. However, because of the number and variety of organic compounds present, even after separation methods have been employed, the general trend in the analysis of these liquids/- liquid fractions has been to use a combination of techniques. Many times these techniques are coupled to yield the so-called "hyphenated" [50] methods of analysis such as GC/MS, LC/MS, GC/IR, and the rela­tively new development of tandem MS (MS/MS). Results obtained have led to a better understanding during the past few years of both product composition and the related mechanisms of reaction. Motivations for Investigations of Oxygen-Containing Compounds Many of the compound classes present in coal liquids have been extensively investigated, particularly with regard to those designated as desirable end products (e.g., aliphatic and aromatic hydrocarbons) [51]. An almost equal emphasis has been directed towards investi­gations of those species which are potentially harmful and environmentally unacceptable. Of particular notoriety in this regard, are the polynuclear aromatics (suspected carcinogens) [52], and the heteroatomic moieties associated with sulfur- [53] and nitrogen-containing [54] compounds, the combustion of which, among other things, is believed to be responsible for "acid-rains" [55]. ;1" liquids/­liquid MSjr·1S). ~. Vto Containing , (. 53J nitrogen­' containing 54J 14 These areas of applied research have resulted in the production of liquid synthetic fuels from coal that are environmentally "clean" and are less polluting than either oil or coal during combustion [2]. Furthermore, the present indications are that the potential for greater degrees of monitoring, control, and confinement of possibly hazardous by-products exists with coal-liquefaction processes than with con­ventional coal use [3]. In an effort to increase operation efficiencies, other areas of research have been directed towards those species thought to play active roles in the liquefaction processes. Most notable among these are the hydroaromatic compounds present in solvents that have obvious potentials for hydrogen donation [56]. More recent has been the detailed study of species that promote hydrogen transfer ("H-shuttlers") [40], and those capable of catalytic activity (e.g., mineral matter [57]). Of all the species present in these liquids, oxygen-containing compounds have been perhaps the least well studied [58, 59]. This appears somewhat peculiar at first, since the predominant functional groups found in coal are those which contain oxygen [4]. However, in terms of synthetic fuels, the oxygen-containing compounds, of them­selves, are neither particularly desirable nor undesirable as end-products [58, 59]. It has only been with the examination of results obtained from previously dominant areas of research that specific investigations into oxygen-containing compounds have become warranted. These results, as well as those obtained from subsequent studies speci­fic for oxygen functionalities [60], have implied these compounds to be capable of a variety of roles in almost all stages of liquefaction [35]. II cl ean ll [[56J. (IIH-shuttlers ll ) 40J, ma:tter 59J. 4J. containing end­products 59J. containing a'll [35J. While the main functionalities of oxygen in coal are those asso­ciated with phenols, alcohols, ethers, carboxylic acids, and carbonyls [60], during liquefaction (bituminous coals), many of these function­alities are either lost through the evolution of CO, CO2, and/or r^O, or converted to other forms. The resultant forms principally contain phenolic groups, and to a lesser extent, etheric and carbonylic groups [25]. During this conversion, the removal of oxygen as r^O has a direct bearing on what was earlier termed the key to liquefaction [9]: the efficiency of hydrogen utilization. Ideally, the quantity of hydrogen consumed is only that amount stoichiometrically required to raise the H/C ratio of the parent coals to the level desired for quality end products [35]. However, the presence of heteroatoms in [; coal alters that ideal level, as these are often removed in reduced 1•forms (e. g., NH^, H2S, and H2O). Even when loss of hydrogen through reductive reactions is taken into account, the actual hydrogen con­sumption, in practice, is considerably greater than that which is theoretically required [9]. While the removal of oxygen and other heteroatoms is desired from the perspective of hydrogen utilization and end-product quality, some of these moieties perform necessary functions, particulary during the initial stages of liquefaction. Specifically, they affect the physical properties of the coal solvent. A good physical solvent is one that will swell the coal matrix [48], possibly rupture weak labile bonds [49], and promote the release of soluble species trapped in coal pores [36]. Studies have been made which directly associate the "potency" of a coal solvent with the existence of an unshared electron pair on either an oxygen or nitrogen atom [61]. The effect of these atoms in , 15 \' ~,~: Whi 1 e functi ona 1 iti es asso- ~:'i Ii ciated ~)~ ,{. 60J, function- ~ alities H20, ~i ) '~);;;' f 25J. H20 9J: ;s HIC 35J. coal alters that ideal level, as these are often removed in reduced . NH3, H20). ~I: theoreti ca lly requi red 9J. r:'· Whil e I,,~ !\J, :'J,I j. l', R; d~ 48J, 49J, 36J. potencyll ,\ ei ther ni trogen 61 J.' 16 the coal solvent is to increase overall polarity and, in this regard, solvent polarity is considered a primary factor in stabilizing some of the initially formed products, e.g., asphaltols, which are produced in substantial quantities by all coals [9]. Although tetralin and other hydroaromatics are often cited as "good coal solvents" [42], for these highly polyfunctional compounds, their solvating power is limited [35]. The presence of heteroatoms (specifically oxygen in phenolic form), on the other hand, have been shown to effectively stabilize the initially formed asphaltols [35, 62]. Furthermore, it has been indicated that when phenols are absent, asphaltol solubility decreases, resulting in increased levels of char formation, presumably through reactions of self-condensation. It is therefore widely agreed that phenols are a desirable, if not necessary, constituent for "recycle solvents" [9]. During the hydrogen donation process, the critical factors in determining the demand for hydrogen are dictated by the parent coal [63]. The rate of demand is greatest during the first few minutes of coal conversion, even though the total hydrogen consumed during the initial period is small [47]. During the periods when demand for hydrogen is high, selectivity is reduced and the free radicals formed will abstract hydrogen from the most proximate available source. In this respect, phenols can supplement the role of hydroaromatics, particularly when these latter compounds are present in reduced levels, by shuttling hydrogen from areas of the coal or solvent that are relatively rich in hydrogen to areas that are hydrogen deficient [62]. Phenols can even act as hydrogen donors via reactions of dimerization or through alkylation reactions with aromatic rings [9]. 9J. 42J, 35J. 62J. condensation. 9J. 63J. ;s 47J. ;s 62J. t~rough 9J. 17 Besides contributing to favorable donor solvent properties, e.g., enhancing physical solvent qualities, promoting H-transfer, etc., phenols have been implicated in unfavorable reactions. Phenols are believed to be the main species involved in the regressive reactions (char formation) that occur in the latter stages of conversion [35]. The quantity of phenols present can actually determine the type of char formed under liquefaction conditions [9]. Char formation occurs to much lesser extents in the absence of substantial quantities of phenols, and can be inhibited to certain degrees by the presence of hydro­aromatics. The status of phenols, therefore, is somewhat analogous to that of the mineral matter present in coal: it is desirable for the mineral matter to be ultimately removed in order to improve end-product quality [64], however, these minerals may exhibit catalytic properties useful for H-transfer and early removal is considered detrimental to the overall process [9]. In view of the variety of roles oxygen-containing compounds have been implicated in, this study was proposed in order to investigate the presence of different types of oxygen-containing compounds in coal and to determine possible component changes indicative of active roles during the liquefaction process. Sample Collection and Process Description Distillates from two near-term commercial coal-liquefaction processes were obtained primarily for the analysis of oxygen-containing , (. transfer, 35J. 9J. ~ hydro-aromatics. , 64J, transfer ~' i' 9J. containing containing 18 process [11, 14, 29 Dried, pulverized coal (less than 0.2 mm) is slurried with process-derived oil in a ratio of two or three to one and pressurized to 21-24 MPa. Compressed hydrogen is added to the slurry and the mixture is preheated to 340°C-370°C before being fed to the base of a catalytic (a commercial Co/Mo preparation used in petroleum refining) ebullated-bed( compounds. Both processes incorporate direct methods of coal-hydrogenation, but through subtly different routes. The H-Coal process has been described as "hydroliquefaction" since it employs the cataly­tic addition of molecular hydrogen to coal [11]. By the same token, the Exxon Donor Solvent (EDS) process can be classified as "solvent-extraction" since hydrogen addition to the coal is obtained via a solvent that has been catalytically upgraded in a separate process to provide good H-donor properties [14]. The distinction between the two types of processes is not clear-cut, as coal-derived solvents and molecular hydrogen are present with the feed-coal in both liquefaction reactors. However, the H-Coal process allows for direct contact between coal and catalyst whereas the EDS process does not. While both processes allow for a variety of products, product flexibility is more apparent with the H-Coal process which can operate in two different modes, depending upon the desired end product [11]. Although dis­tillates from both processes were analyzed for oxygen-containing compounds, the main focus was placed on the distillates of the EDS process and, therefore, will be emphasized in this report. A brief description of the separate processes follows in order to permit comparison. H-Coal (hydroliquefaction) coal­hydrogenation, IIhydroliquefaction ll l1J. IIso1vent­extraction ll 14J. Coa1 11J. EOS .. process­derived 0;1 ;s Mo bedl 19 reactor operated at a temperature of 455°C. The three-phase flow of coal, process-derived oil, and molecular hydrogen passes up through a distribution tray into a "bubbly" bed of catalyst. The catalyst is suspended in the bed through the addition of internal recycle oil provided by a hot-oil recycle pump. A portion of the catalyst is withdrawn and fresh catalyst added continuously at a rate of about 0.5 kg/ton of processed coal. Residence times range from 30 min to over 1 hr, depending on whether the product desired is low-sulfur fuel oil or syncrude, respectively. Exxon Donor Solvent (solvent-extraction) process [11, 14, 15] The Exxon Donor Solvent (EDS) process operates in what might be termed a fuel oil mode, analogous to one of the two modes employed by the H-Coal process. Crushed coal is dried by mixing with a hot hydrogenated recycle solvent in a drier/mixer. Molecular hydrogen is added as this slurry is fed to a vertical, cocurrent, upward-entrained-flow liquefaction reactor operating at 13.5-17 MPa and 425°C-480°C. The optimum resi­dence times is about 3/4 hr. Part of the 200°-425°C fraction of the C4-540°C distillate is taken as recycle solvent. This solvent is hydrogenated separately in a conventional fixed-bed catalytic hydro-treater prior to its introduction into the drier/mixer. The catalysts employed for recycle solvent hydrogenation are a mixture of commer­cially available and proprietary multi-metal 1ic catalysts. The separate hydrogenation of solvent allows a more accurate monitoring of solvent quality. Noteworthy is the fact that Exxon has formulated a Solvent Quality Index (proprietary) to characterize solvent properties 1 9 ternperatlJre solvent- 11,14,15 mixer. entrained-C bed hydro­treater mixer. commer-cially metallic 20 and define the molecular composition of the preferred solvent [15]. They have also reported that the addition of molecular hydrogen to the liquefaction reactor, reduces solvent quality requirements for a given level of conversion and indicated that molecular hydrogen in the ab­sence of catalysts is indeed capable of "capping" some of the free radicals formed during liquefaction [46]. In addition, the EDS process incorporates a proprietary coking/gasification process (termed Flexi-coking) [11] to yield addiitional liquid products and a fuel gas of low calorific value. The end result is the conversion of essentially all of the feed-coal. University of Utah process In addition to the two commercial distillates, a "Whole-Tar," ZnC^-catalyzed coal liquid was obtained from the University of Utah to aid in the development of analytical techniques for the commercial distillates. The H-Coal distillates were collected, shipped, and stored under a nitrogen blanket and were derived from a Kentucky #11 feed-coal. The EDS distillates were collected and shipped under atmospheric conditions and were from a high-volatile bituminous Illinois feed-coal. Boiling ranges for the commercial distillates, as provided by the suppliers, are shown in Table I. 46J. gasification Flexi­coking) llJ ZnC1 2-catalyzed dis t i 11 a tes . 21 Coal-Liquefaction Process Process Fraction Boiling Range (°C)i EXXON DONOR SOLVENT (solvent-extracti on) Atmospheric Overhead Atmospheric Bottoms Hydrotreated Atmospheric Bottoms Vacuum Overhead 35 - 235 180 - 500 60 - 500 290 - 595 H-COAL (hydroli quefaction) Atmospheric Overhead Atmospheric Bottoms Vacuum Overhead 90 - 260 260 - 400 260 - 450 1 Boiling range data provided by sample supplier TABLE I PROCESS SAMPLE DESCRIPTIONS ANALYTICAL METHODS AND RESULTS Capillary Gas Chromatography The primary approach to the investigation and identification of oxygen-containing compounds in the coal liquid distillates was centered around the technique of mass spectrometry. Several different modes of mass spectrometry, e.g., field-ionization mass spectrometry (FIMS), high-resolution mass spectrometry (HRMS), and more recently, tandem mass spectrometry (MS/MS) have been shown to give fast qualitative analysis of multicomponent mixtures [65, 66, 67]. Because of the nature of the distillates, specifically, the possible presence of oxygen-containing compounds in only trace amounts, it was felt that "on-line" pre-separation of components by gas chromatography (GC/MS) also offered a reliable approach to the analysis of samples. This was thought to be particularly true with the advent of capillary gas chromatography. The use of mass spectrometry alone for the analysis of multi-component mixtures is complicated by the possible existence of compounds with similar spectra. (This complication is most apparent with low-resolution mass spectrometry, but is somewhat reduced with HRMS, and even more so with MS/MS, which technically can be considered a mode of on-line pre-separation.) Conversely, the use of gas chroma­tography alone has intrinsic limitations when characterizing coal liquids, since classes of compounds are often present with similar containing ·: HRMS), r~: ' " ,I:': MS/MS) gi ve qual i tative i" } 67J. f 't. ~J:.' 'r containing 1Y ~; linell MS) " t , \~", 'I.' t multi- ::\component ~ ~t • ,} r·t ~ s~paration.) ~' chroma- ~;:t ~( I,: ~/ ~'. t~· i tography boiling ranges and an eluted peak may represent a mixture of several different classes [9]. However, some compounds present in a multi-component mixture may yield similar mass spectra, but exhibit chromatographic mobilities that are quite different, and vice versa [68]. distributions between two phases, one of which is stationary and the other mobile [69]. In gas chromatography, the mobile phase is a weakly adsorbed gas (carrier gas) such as helium [70], while the stationary phase may be either solid (gas-solid chromatography; GSC) or liquid (gas-liquid chromatography; GLC) [69]. The latter method is the more versatile and selective of the two and consequently is much more commonly used [71], and was employed for this study. In GLC, then, the components are partitioned between the carrier gas and the non-volatile liquid substrate [72]. The resolution or separation of two adjacent chromatographic peaks is related to two factors: solvent efficiency and column efficiency [73]. The first factor, characteristic for a given substrate or stationary phase, is a result of solute-solvent interactions and is determined by the relative distribution coefficients of solutes in the liquid substrate at a given temperature [69]. The latter factor is a measure of the ability of a column to produce narrow peaks and is generally given in terms of theoretical plates per unit of column length N/L), by analogy to distillation column technology [73]. In determining column performance, however, the inverse of this ratio, or the height-equivalent of a theoretical 9J. multi­component : The theory of chromatography rests on the physical separation of two or more components in a mixture on the basis of their different 69J. 23 70J, 69J. 71J, 72J. 73J. 69J. (NIL), 73J. plate (HETP), is usually employed [72]. The HETP, therefore, is that length of column necessary for the attainment of solute equilibria between the mobile gas phase and the stationary substrate [69] and is a characteristic of a given column. Its values vary with the type and concentration of components present, but is most strongly affected by flow rates [74]. This strong dependency on column flow implies an optimum linear gas velocity, and is generally depicted by van Deemter plots [75] as shown in Figure 1. Each of the three terms shown in the van Deemter equation describes a physical effect [73]. The "A" term represents the increase in peak broadening that occurs due to the flow pattern of carrier gas passing through the column [69]. This term (related to eddy diffusion) is constant for a given column geometry (e.g., column packing and arrangement) over the range of flows normally used in GC analysis [74]. The "B" term is a result of the longitudinal diffusion of compon­ents in the carrier gas [69]. It is this effect that is responsible for the broadening of peaks that are eluted in later sections of the chromatogram during isothermal GC analyses [73]. The molecular diffusi vity is a property of both solute and carrier and is inversely proportional to gas density [76]. To minimize longitudinal diffusion, carrier gas of higher molecular weight, such as nitrogen or argon, should be used. Alternately, the pressure may be increased. Both of these adjustments, however, have a negative effect (with respect to column efficiency) on other parameters in the van Deemter equation. The "C" term is a measure of the resistance to mass transfer in the liquid phase [73J. Low carrier gas flow rates and low-density gases increase the rate at which components may diffuse from the 24 72J. 69J 74J. 73J. 69J. BII ;s ;n 69J. 73J. diffusi­vity 76J. a IIC" l i n e a r velocity Figure 1. van Deemter plot characteristics for two common carrier gases. ^ en HETP=A + B/v + Cv linear He N U1 26 substrate into the gaseous phase [69]. The effect of changing gas density (and flow rate) is nearly linear and results in a wider range of optimal gas velocities (broad region of low HETP values) for the gases of hydrogen and helium [74]. Because of the compressibility of the carrier gas, the linear velocity is not constant over the entire length of the column. Furthermore, when invoking temperature program­ming (changing gas density and hence velocity) a slight loss in efficiency can occur [77]. Both of these effects are minimized with low-density carrier gases. Helium was, therefore, selected as the carrier gas in this study. Packed columns are the most common type used in conventional gas chromatographic analysis 78]. However, for the analysis of multi - component mixtures, higher resolution than that usually obtained with packed columns is often desired. The application of capillary columns to GC/MS has provided (in many cases) this higher resolution. Conse­quently, capillary GC has drastically changed the role of this analytical method by enhancing identification power, especially with regard to low-resolution mass spectrometry [68]. The use of capillary columns instead of packed columns has two effects on column efficiency. The first is a reduction in eddy dif­fusion (A term) to almost negligible values and results in a slight decrease in the HETP parameter over the optimal ranges of carrier gas velocities 74]. Because the eddy diffusion is negligibly small, it is often ignored and the resulting equation for capillary GC (analogous to the van Deemter equation) is the Golay equation, named for M.J.E. Golay, who first introduced capillary columns in 1956 [79]. ~ " 69J. 74J. program-ming 77J. ;n [78J. multi­component capillary' 68J. [74J. 79J. 27 The second effect, and most important, is the lower resistance to gas flow relative to packed columns [78]. The lower pressure drop associated with capillary columns of a given length thus allows for longer columns and a corresponding increase in total theoretical plates. Although increasing column length is generally considered a poor way to improve column resolution (i.e., resolution is proportional to the square root of the number of theoretical plates, N, and hence to unit length, L), in the case of capillary columns, the substantially greater column lengths (factor of 5) makes this objection nominal [73]. A major disadvantage with capillary columns is the lower sample load allowable. This disadvantage has been reduced with the develop­ment of support-coated open-tubular (SCOT) columns. SCOT columns contain a layer of porous material upon which the stationary phase is coated [74]. This increased surface area permits more liquid substrate per unit length and hence, larger sample loadings may be used without fear of overloading the column (i.e., washout of stationary phase). To test various substrates, three SCOT columns with different liquid phases were acquired. These phases and some of their properties are listed in Table II [69, 80]. TABLE II GC LIQUID SUBSTRATE DESCRIPTION Column Liquid Phase Characteristics Max. OV-101 methyl silicone non-polar 280°C-300°C 0V-17 50% methyl-50% phenyl silicone intermediate 280°C-300°C Carbowax-20M (poly)ethylene glycol polar 225°C-250°C 7BJ. 73J. SCOT) 74J. BOJ. Li qui d ~1ax. Temp. 10l s il i cone 2BO°OV-SO% 2BO°s il i cone 20M 22soC-2S0°C glyco 1 It was felt this collection of liquid phases would provide an adequate range for satisfactory analysis as described by Delley and Friedrich [81]. Furthermore, it was felt that the use of SCOT columns would minimize the disadvantages associated with capillary columns (as cited above). Besides the SCOT columns, a "non-conventional" technique for splitless injection onto the capillary columns was employed to further reduce the restrictions on sample size by increasing the ratio of solute to solvent actually deposited on the GC column. This technique will be described in the next section. Curie-point Flash Vaporization Device for Splitless Injection Rather than employing standard GC syringe injection methods, a Curie-point flash vaporization device was used for splitless injection onto the capillary columns. This device, originally designed for direct pyrolysis by Meuzelaar et al. [82], was modified to provide for flash vaporization rather than pyrolysis of sample [58]. The reasons for the selection of this technique were several. For example, repro­ducibility is often a problem when using conventional inlet splitters for capillary column sample injection. In addition, when analyzing for trace components in complex mixtures, the use of splitless injection is deemed a necessity [83]. Further, because of the polar nature of these samples, it was felt that inlet contamination from heavy residue deposition was probable with conventional splitless techniques. Use of the Curie-point device, on the other hand, offered several advan­tages as described by de Leeuw et al. [84], including; (a) positive pressure protection of the capillary column from atmospheric gases, (b) prevention of sample backflush, (c) potentially superior 28 81J. F'~rthermore, IJse IInon-conventional ll aZ. repro-ducibility advan-tages aZ. 84J, 29 resolution and reproducibility, and (d) the confinement of sample in glass reaction tubes - which were removed after each analysis to avoid reaction zone contamination. Finally, temperature profile reproduci­bility between different filaments of the same Curie-point temperature is high, thus ensuring uniform intersample heating environments [85]. Basis of Curie-point technique The Curie-point method is based on the inductive heating of a ferromagnetic filament and the resulting phase change that occurs as filament temperature rises. Power supplied to an induction coil re­sults in a magnetic field [86], and if an object is placed in this field, magnetic moments are induced in that object. The object material is termed paramagnetic if the induced moments are parallel and small when compared to the external field, ferromagnetic if the moments are parallel and large, and diamagnetic if they are anti-parallel, in which case the moments are always small [87]. These properties cannot be explained by classical physics, but are predicted by electron spin moments using quantum mechanics [86], The characteristic parameter associated with the magnetic proper­ties of a particular material is its magnetic susceptibility, equal to the ratio of the intensity of object magnetization, I, and the magnitude of the field, H. Most materials are paramagnetic, and Pierre Curie showed that "paramagnetic" susceptibilities are inversely de­pendent upon temperature [88]. This discovery, known as Curie's Law, was later modified by Weiss to correct for small diamagnetic effects (Curie-Weiss Law) and is the underlying principle of the Curie-point technique. )~ \'1 I: ~ , " I~. t -- r: r ~ Finallys reproduci-bility r, f 85J. rt: !. (, ~: f f,': ~; ferromagneti c fil ament resul ti ng ~;. ~! I. if ~ . ~, ~:. ~I' l II ·I~.~.; ~f {~ '~;' \.', re-sults 86J, 87J. 86J. Sf' proper-i: ! ties 1j;, ,f IIparamagneticll susceptibil ities 88J. Weiss 30 When plots are prepared depicting the temperature dependence of magnetic susceptibilities for five elements; iron, Fe; cobalt, Co; nickel, Ni; gadolinium, Gd; and dysprosium, Dy (as well as alloys of these and other elements), the presence of discontinuities at specific temperatures, called Curie-point temperatures, is evident 86]. At temperatures greater than its Curie-point, each of these substances obeys Curie's Law, or the Curie-Weiss Law (i.e., as temperature rise, magnetic susceptibilities fall) because the increased thermal agitation reduces the effective alignment of induced moments, and the result is simple paramagnetic behavior. At the Curie-point temperature, the magnitudes of the interionic actions are comparable to the thermal energies, and the materials exist in equilibrium between two phases. As temperatures are lowered below the Curie-point, the rate of increase in the magnetic susceptibilities for these materials is much greater than that predicted by the aforementioned laws and the tendency toward alignment becomes controlling [89]. This principle is also basic to the operation of electromagnets, i.e., it is because of this complex temperature dependency exhibited by ferromagnetic materials that cooling systems are required for most electromagnets in order to remove energy accumulated from Joule heating and thus maintain magnetic strength [90]. For the flash vaporization device, if no cooling is provided to the ferromagnetic filament centered in an induction coil, it will rapidly heat to its Curie-point temperature. At this temperature, the transi­tion of ferrogmagnetism to paramagnetism results in a corresponding decrease in the intensity of magnetization (somewhat analogous to removing the iron core from an electromagnet) and hence, a substantial [86J. i .equil"ibr;um 89J. 90J. 31 decrease in the energy absorbed from the high frequency (hf) field in the coil 91]. At this point, the wire temperature is stabilized due to the balance between energy gains and losses 92]. The heating of the filament is extremely fast, typically on the order of 0.1 sec. Temperature rise time is determined by the wire diameter, the composition of the alloy, the strength of the hf field and the field frequency 92]. Examples of temperature-time profile for sample filaments under standard operating conditions are shown in Figure 2. Steps in the Curie-point method The Curie-point flash vaporization technique involves several steps before the sample is actually vaporized and swept onto the capillary column. The initial step requires the preparation of a solution or suspension of sample in an appropriate solvent. The choice of solvent is fairly important. The solvent should possess qualities such that it does not interfere with identification of the sample components (e.g., fairly simple structure and low relative molecular weight). Ideally, the sample should be soluble in the solvent or, at the very least, should not interact with the solvent to form non-uniform suspensions -- inadequate for representative component analysis. It is also desirable for the solvent to be volatile in order to avoid prolonged drying times [85] and potential loss or contamina­tion of sample components. Typical concentrations of the solutions or suspensions used for sample coating are 1 to 2 mg/ml. The solution or suspension is applied in measured amounts as depicted in Figure 3. Application involves the [91J. [92J. [92J. 85J FiglJre 32 1200-1 Tc 1128 °C 800 Fe(Tc =770°C) f Tcr358 °C) POWER OFF 0.2 0.4 Figure 2. Temperature-time profiles of three Curie-point filaments under typical operating conditions. - . 1200 Co(Tc =1128°C) 400 Ni(T. =358°e) 0.2 I I POWER OFF I I I 0.4 0.6 0.8 --4.. .. t (sec) 1.0 1.2 t"ime 33 COAL GLASS SUSPENSION REACTION TUBE 6 - ^ //. /•/ v. '» FERROMAGNETIC FI LAMENT S A n P L E D R 0 P Figure 3. Method of sample application to Curie-point filament. FILAMENT SAMPLE DROP 34 use of a micropipette or microsyringe to deposit 5 yl drops close to the tip of the filament. With the abovementioned concentrations, each drop then contains 5 to 10 yg of sample. Ten to twenty yg of total sample is standard per analysis [58]. The solvent is then allowed to evaporate while the wire, still protruding from the glass reaction tube, is slowly rotated. This provides a uniform distribution of sample on the wire. It is during solvent evaporation that the loss of high volatile components may [58, 59], After drying, the wire is retracted into the reaction tube, which also serves as protective cover for the sample. cleaned after received from the manufacturer and before use. They are usually cleaned ultrasonically in a series of organic solvents and then oven dried. Care should be taken not to oxidize the outer surface of the wire, since oxide layers may result in catalytic activity and/or change the emissivity of the filament surface [85]. The tube (and wire) are removed from the reaction-zone after each analysis. This eliminates possible contamination of the reaction-zone, inlet manifold and column with heavy residues. The borosilicate glass reaction tubes are not disposable and must be cleaned after each analysis. This is accomplished by immersing the tubes in hot "Chro-merge" for several hours, followed by rinsing in water, sonication in methanol and then pentane, and subsequent oven drying. As mentioned, the method was originally designed for pyrolysis (of biological samples) [82]. Because of the nature of the coal liquid distillates and the fact that flash vaporization was the objective ~l ~g ~g 58J. occur; this is a potential disadvantage of the Curie-point technique 59J. The ferromagnetic wires are disposable; however, they should be use." 85J. zone, This;s ;n 82J. 35 rather than pyrolysis, the reaction tubes were modified [58]. A com­parison of the original reaction tube with the modified version is shown in Figure 4. The incorporated modifications minimize the space in the vaporization zone while lengthening the neck of the tube. Besides reducing "dead-volume," the extended neck permits a longer time of contact in the vaporization zone between carrier gas and hot filament. The carrier gas thus has more opportunity to heat up as it flows through the narrow passage between the hot filament and the inside tube wall before entering the zone of vaporization. While the indications are that few compounds that can be eluted from GC columns are actually lost in the injector [58], it was felt that the effects of this modification would minimize sample losses from residual con­densation of high-boiling components. Once the wire has been coated and situated in the reaction tube, the tube, guided by a ceramic sleeve, is inserted into the flash vaporization device shown in Figure 5. The base of the reaction tube (sample end) is sealed against a "Vespel" seat by a spring-loaded plunger. Under typical operating conditions of constant pressure, entering carrier gas flows into the device and through slits in the ceramic sleeve, at which point the flow is split. Approximately 10% of the total flow is forced down the neck of the reaction tube, into the vaporization-zone and then into a 1/16-inch outer diameter (0D) glass-lined, stainless steel tube connected to the capillary column. The extremely fast heating of the filament, combined with the low dead-volume of the device, provides a narrow band of sample entering the capillary column, a necessary condition for optimal column performance [83]. 58J. 58J, 00) dead­volume 83J. 36 Figure 4. Comparison of original Curie-point reaction tube designed for pyrolysis of biological samples with version modified for coal liquid distillate flash vaporization. 1) standard reaction tube; 2) modified reaction tube; 3) 358°C Curie-point filament; 4) 510°C Curie-point filament; 5) high frequency (hf) coil; 6) carrier gas flow path; 7) sample location. I ........ ~ - - ---- -- ~ - -- /// // 5 7 2 4 /// ,:~ ~ ••••••••• Curie­point 36 37 Figure 5. Cross-sectional drawing of Curie-point flash vaporization device. 1) spring; 2) plunger; 3) 0-ring; 4) 0-ring; 5) ceramic sleeve; 6) ferromagnetic filament; 7) borosilicate reaction tube; 8) 0-rinq; 9) hf coil; 10) cooling chamber; 11) Vespel seat; 12) lead washer; 13) removable reactor base; 14) capillary column; 15) sample; 16) Vespel or graphite ferrule; 17) 1/16-inch low dead-volume Swagelok nut; 18) heating mantle; 19) GC inlet port. ... CARRIER ~~~~ GAS ~ ~'h~~~ sectional O-O-ring; sieeve; O-r;n9; 19) 38 The body of the device shown in Figure 5 is of stainless steel and cooling water. A heating mantle, attached to the base of the device, allows base temperatures to be in excess of 300°C and thus helps avoid condensation of vaporization products on otherwise potentially cool surfaces. In addition, the reactor base was designed to permit direct gas chromatograph. The remainder of the carrier gas flow (typically 20-25 ml/min for capillary Operating Conditions for Curie-point GC Analyses The whole liquid distillates from both processes (EDS and H-Coal) were either soluble or formed adequate suspensions in benzene. Con­centrations of 2 mg/ml were prepared and two 5 ul drops were applied to iron/chromium/nickel (Fe/Cr/NI) alloy filaments (Curie-point tempera­ture, 510°C [58]). The reactor induction coil was connected to a 1.5 kW, 1.1 MHz Fischer Labortechnik high frequency power supply. Temper­ature rise time to Curie-point was 150 milliseconds after which the ferromagnetic filament stabilized at its Curie-point temperature as long as the hf coil remained on [91]. Total heating time was 10 seconds. Helium was used as carrier gas at a pressure of 5 psig so as to yield flows of approximately 3 ml/min through the 30 meter long, 0.5 millimeter inner diameter (ID) support-coated open-tubular (SCOT) capillary column coated with 0V-101 substrate, purchased from Scienti­fic Glass Engineering, Inc. (SGE). A vented gas flow rate of 20 ml/min is protected from excessive temperatures by a circulation system of interfacing with the injection port of a Hewlett-Packard 7620A research m1/columns) is vented through the device cap and into a meter which allows additional control and monitoring of flow conditions. Con-centrations ~1 chromium/nickel Cr/NI) 58J). 91J. SCOT) OV-10l 39 was maintained to facilitate stable pressure control as prescribed 59]. chromato-graph was set at 280°C with the reactor base heated to 260°C. Actual column flow was measured at 3.4 ml/min at the end of the column and temperature programming was as follows: 80°C isothermal (1 min), 4°C/min to 250°C, 10°C/min to 260°C, and 260°C isothermal (10 min). Preliminary GC Results Results of the GC analysis of the EDS atmospheric bottoms are shown in Figure 6. These results were obtained using an OV-101 coated SCOT column in combination with the Curie-point injector and a flame ionization detector (FID). The SCOT column with the OV-101 liquid substrate separates components primarily on the basis of individual boiling points [80]. The test chromatograms supplied by SGE showed a column efficiency of 29,100 total plates for this particular column. Two observations are evident from Figure 6. The first is a demonstration of the high reproducibility and peak resolution that can be obtained using the Curie-point device, in agreement with de Leeuw's results [84]. This high degree of reproducibility helps articulate the second observation: the effect (or apparent lack of reactivity) of mild oxidative conditions upon the EDS atmospheric bottoms. A test of stability under oxidative conditions was performed by bubbling oxygen gas through a portion of the distillate at room temperature for up to seven days. A comparison of the chromatograms shows no appreciable change in the GC profiles. However, as mentioned earlier, the EDS distillates were shipped under atmospheric conditions. Therefore, the possibility existed that the full extent of oxidation under an oxygen elsewhere [59J. The injection port temperature of the gas chromato­graph programmi ng i sotherma 1 mi n) , 1,4°C/min lOoC/i EOS 10l FlD). 10l 80J. 84J. EOS profil es. menti oned earl ier, EOS 40 t ( m i n ) Figure 6. Comparison of FID chromatograms of the EDS atmospheric bottoms distillate (a) as received and (b) after having oxygen bubbled identical solvent peak. (a) (b) .--t. --:,' i--r -~i -', "-i~- -.. . _- j- ---r------- ---~'--~--r--~--r--.. ! , 1-- --- - - J,. -- -- .. i,- . ----,_._- 1_-- _ --'------1 ___ : __ I , --< ---:-- - --:--:--i---j--- : : I I I - -:--_. -; ---i----!---:- . , . . - ... ,-----!---'-- --'-'-:'- -_.- ···_-t---:---,--· • J • • --.- -- -.--;'--_. ,- . , . ----,-.-:-- ._; ----,- -- .. -:._--_ .. --- t (min) 10 20 ", , .' _._ .. _.~_! __ . __ ; _____ ' ________ t -~t= ~f~---- ~ > ! ---i- -'-'- _+--1-,_:_,,_. ,~ i~ !=,=t~~~f :-~,: ~-- ---r- i I; . ~. ___ . __ ~_. -, -1- o. I i _. i ___ : -- -- -----'-___- -r=;-_. -. -i! _ --- i 30 ! ; i - '.- j ._-,-- 1 ': 40 FlD through for seven days at room temperature. Note that the two chromat­ograms are nearly identica.l in every detail except for the initial or oxygen-containing atmosphere had already been reached. Furthermore, it has been reported that (alkyl)phenols can act as anti-oxidants when present in coal liquids [35], and, as was later discovered, (alkyl)- phenols constituted the major class of heteroatomic compounds in this distillate, although, their overall relative concentration was quite low. Once it was shown that reliable results could be obtained, other distillates were analyzed using these same conditions. Figure 7 com­pares the FID chromatograms obtained from the EDS raw solvent and its hydrotreated counterpart. The two chromatograms show both qualitative and quantitative differences. The hydrotreated distillate is dominated by high-volatile components as evidenced by the confinement of larger peaks to the earlier portion of the chromatogram. This domination of high-volatiles was thought to be due principally to the conversion of aromatic components to perhydroaromatic species during hydrotreatment [58]. These analyses performed on the whole liquids, while of interest, were not suitable, since the primary objective was to investigate the oxygen-containing compounds present, and the whole liquids appeared to be dominated by hydrocarbon species (as later confirmed by GC/MS). Therefore, it was proposed that other separation techniques be used to increase the relative concentrations of heteroatomic species (specifi­cally, oxygen-containing compounds) in these samples. Pre-separation Methods Two fractionation approaches were considered: (1) separation into component classes based on solubility characteristics, e.g., 41 containing a1ky1)35J, alkyl)­phenols FlO EOS 58J. containing GCjMS). containing 42 Figure 7. Comparison of FID chromatograms of (a) the EDS atmospheric bottoms and the (b) EDS hydrotreated distillates on an 0V-101 SCOT column. Numbered peaks have been identified by GC/MS analysis (courtesy of David L. Pope) as: 1) tetralin; 2) naphthalene; 3) methyltetralin; 4) methyltetralin; 5) 2-methylnaphthalene; 6) 1- methylnaphthalene; 7) C2-tetralin; 8) acenaphthene; 9) C2-naphthalene; 10) methyl-diphenyl methane; 11) dibenzofuran; 12) fluorene; 13) phenanthrene. : . : -~- --:" ----.---,- . I 1 t -- I ----:--:--- ---:---.--:--.--- -i-.-- ; __ .~. ~--:- .- . . --- i --'j . , .... -j ·-1- _ .. ---~-.- I i L-- I I 1 .. ' ._1 __ . - ' -- -----,-:-,-.-! --' ._g-----,----: I - , I' - ._. -1--, __ :._ I I =- '. '-'I=~i=:i=-T~:3 7' - - - --, - ~-t-~'~~.'::- : ~ ~:~ -=:=R=t,!--, -: . .s _' '~.~_~~:::=t:::'==== ,-, :1_- ___I _ ,-_'_:__ I .' .. -+- i .. (b) ~:~~r~'~\~>: =!: ;~!~~t~~(t -- -' -. ---~--: .-.; --, _:._~' _i_~_' ___ _ - -.. ---,._! --: .".- ~!, ~ -' .. --; .... ' .- I 10 20 t(mln) 30 40 FlO EOS EOS OV-10l methylnaphtha1ene; methylnaphtha1ene; tetra1in; naphtha1ene; diphenylmethane; 43 traditional solvent-extraction techniques, and (2) liquid chromato­graphic (LC) column separation. The latter approach was viewed with apprehension because of the potential loss of highly polar compounds. Because the structure and hence behavior of the oxygen-containing compounds in these distillates was unknown, it was felt that efforts should be directed towards preserving the integrity of the overall sample composition. For this reason, a solvent-extraction technique was initiated. Solvent-extraction technique The solvent-extraction technique developed by Schweighardt and Thames at the Pittsburgh Energy Research Center (PERC) [93]. Sample amounts of 3-4 g were ground by hand for 15-20 minutes under liquid nitrogen in order to obtain fine suspensions and fully expose the sample to solvent action. While still under liquid nitro­gen, the suspension was placed in an ultrasonic bath and 240 ml of n-pentane for 10 minutes and, after evaporation of the liquid nitrogen, decanting of the supernatent liquids. Three or four pentane washes followed, each subjected to sonication and centrifuging before multiple washes with pesticide-grade benzene, tetrahydrofuran, and methanol in like manner. This procedure varied from the more common solvent-extraction sequence of pentane, benzene, and pyridine [94]; however, results obtained with the zinc chloride (ZnC^-catalyzed Whole-Tar sample showed total dissolution into sub-fractions corresponding to extraction with the four solvents mentioned. technigue used was originally n­pentane (pecticide-grade) was added. This was followed by centrifuging 94J; ZnC1 2)-cata1yzed 44 When this separation technique was applied to the EDS atmospheric bottoms distillate, as much as 95% of the sample went into the pentane sub-fraction with the residue being nearly completely soluble in methanol. However, applying this technique to the corresponding hydrotreated distillate resulted in greater than 99% dissolution of the sample in pentane. These results suggested that a strong decrease in the polar (oxygen-containing) compounds occurred as a result of the solvent hydrogenation process. However, this was later shown through GC/MS studies to be false [58]. On hindsight, this apparent contra­diction was not unreasonable, since the solubility of a substance in a mixture is not only a function of its structure and chemical function­ality, but also depends on interactions with other substances present which may act as co-solvents [32]. In this case, the higher percentage of hydroaromatics appeared to promote the solubility of polar compounds in the pentane sub-fraction. It was therefore concluded that solvent-extraction techniques alone were inadequate for classification of these samples, particularly with regard to the component classes present in relatively small amounts. Analysis of the polar sub-fraction (pentane-residue) obtained from the EDS atmospheric bottoms under GC conditions, identical to those employed earlier for the whole-liquid analysis, resulted in chromato-grams that were unsatisfactory. This pentane-insoluble sub-fraction proved to be too polar for adequate resolution of the major components by the OV-101 coated SCOT column. In view of this, an effort was made to obtain satisfactory chromatograms by using other SCOT columns coated with more polar stationary phases. A SCOT column coated with OV-17 was installed and operated under conditions similar to those employed EOS contra-diction solvent­extraction EOS liquid chromato­grams 10l 45 earlier. Figure 8 compares the EDS atmospheric bottoms pentane-residue sub-fraction chromatogram obtained with an OV-101 coating to that obtained with the OV-17 substrate. The major difference appears to be a slight reduction in baseline with only minor improvement in peak resolution. At this point, the scope of study was expanded. The pursuit of SCOT column GC analysis was continued, but using a column coated with a highly polar substrate, Carbowax-20M (CBW-20M). In addition, two other alternatives were considered. The first was to pursue LC column fractionation, in spite of the potential hazards of irreversible component adsorption. The second approach was to utilize direct GC/MS analysis of the whole liquids. Open-column liquid chromatographic technique An open-column (gravity elution) LC technique was used to separate the samples into four sub-fractions. These sub-fractions were in­creasingly polar in nature corresponding to sequential elution with the following solvent/solvent mixtures: pure hexane, benzene/hexane (1:8), benzene/ethyl ether (4:1), and benzene/methanol (1:1). The mixed solvents were adapted from Rubin et al., [95] where they were applied to the neutral fractions (insoluble in aqueous acid and base solutions) of coal liquids rather than to the whole-liquids as used here. Silica gel (60/120 mesh, Baker Analyzed Reagent) was used as substrate after being washed with methanol and dried at 175°C for 1-2 hours. The use of silica gel as substrate minimized the potential loss of sample since fewer of the substances in coal liquids are irrevers­ibly adsorbed by silica gel than by other adsorbants [9]. Two grams of EOS 20M CBW-MS technigue 1 :1 :et aZ., 95J liquids 120 irrevers-ibly 9J. 46 T i i 1 i J • ' i i | • i » t | i i i i \ t i i t y 50 t(min) Figure 8. Comparison of the FID chromatograms of the EDS atmospheric bottoms pentane-insoluble sub-fraction on (a) OV-101 coated and (b) OV-17 coated SCOT columns. (a) __ ,_~ __ -, __ f .. _ .. _ .. --- --.... -,~ --.- .-.---- - .. -- -- _._-' ----j-- ~--.-- ---- -.--.--- -,. __. - - -.... _- ... _---.. '-. - - -- -_ .. (b) 10 20 30 40 50 t (min) FlO EOS 10l 47 sample were placed on top of a column (1.5 X 30 cm) packed with 6-10 grams of silica gel. The quantities of solvent/solvent mixture used were 150 ml each and all solvents were of pesticide-grade or better. The extraction solvents were removed from the LC sub-fractions with a rotary evaporator (Buchi, Rotavapor-R) using a nitrogen flush. Since rotary evaporation was of limited efficiency in cleanly sepa­rating the extraction solvents from the volatile components, McClennen [59] introduced an infrared monochromator (Wilks, Mi ran 1A) on line to monitor solvent evaporation. Approximately 5% of the vapor leaving the rotating solution flask was continuously sampled through a 1/16-inch 0D Teflon tube and monitored in a 4 cm pathlength gas cell evacuated by a rotary pressure-vacuum pump (Gast, model 0211). Condensation problems in the gas cell were eliminated by the dilution effect of the nitrogen flush. While the potential loss of volatile components remained, this procedure provided for more accurately determined evaporation endpoints, and thus, a greater degree of repro­ducibility. The LC fractionation yields obtained for the different distillates are shown in Table III. Only the yields for the atmospheric overheads were poor, due principally to the loss of the high-volatile components. Significant quantities of (alkyl)phenols were thought to be present in both the benzene/ethyl ether (B/E) sub-fraction and in the pentane-insoluble sub-fraction of the EDS atmospheric bottoms. Standard solutions of (alkyl)phenols (Aldrich Chemical Co.) were pre­pared and subsequent GC analysis of the sub-fractions and standard solutions seemed to verify these assumptions. Results of these analyses are depicted in Figure 9 and nearly identical profiles for plated !TIl pes t i ci de-R) - Miran lA) 00 em vacuum repro­duci bi 1 i ty . phenols EOS phenols TABLE III LC FRACTIONATION YIELDS LC Fractions recovered)1 Process Fraction Atmospheric Overhead DONOR SOLVENT Atmospheric Bottoms Hydrotreated Atmospheric Bottoms Vacuum Overhead Hexane Fraction 38.7 80.0 87.8 3.7 Hexane/- Benzene Fraction 0.1 2.4 13.3 Benzene/- Ether Fraction 5.8 73.1 Benzene/- Methanol Fraction 0.4 16.5 Total 59.42 97.9 96.4 106.5 H-C0AL Atmospheric Overhead Atmospheric Bottoms Vacuum Overhead 55.0 78.4 68.0 0.3 6.2 11.7 11.5 12.4 17.5 0.5 2.2 3.0 67.52 99.2 100.2 1 Most cases are an average of two determinations 2 Loss of highly volatile components. II I (% l Process 3B.7 O. 1 13.5 7.5 59.42 EXXON ABtmototosmphse ric BO.O 1.5 14.9 1.5 SDOOLNVOERN T HAytdmrootsrpehaetreidc 87.B 11. 5 67.52 COAL ABtmototosmphse ric 7B.4 6B.0 11. 7 determinations. ~ 00 (a) J min) Figure 9. Comparison of FID chromatograms obtained with a Carbowax-20M coated SCOT column of (a) the EDS atmospheric bottoms B/E sub-fraction; (b) a mixture of (alkyl)phenol standards and (c) the pentane-insoluble sub-fraction of the EDS atmospheric bottoms distillate. " " , , , ,, --....-"---------________ ~..J>..._J''______'~" (b) (c) 20 40 • , t(mln) I: O-CIlESDt. I""", II-CllfSOt ,,," II-(:R£SOL PHENOL 60 I "l' " " " " "" "" " ..""'' " " " " , ol. .. ·XFLENOI. " ,,, :2~J.5-TRI-tlrrHrL PHENOL 80 I 100 120 I , FlO 20M EOS phenol pentane­insoluble several (alkyl)phenols are exhibited with regard to retention times and relative peak shape. This figure also illustrates the improved per­formance of the CBW-20M substrate over other stationary phases earlier employed, even though temperature programming of the column was more restricted, e.g., maximum CBW-20M column temperature was 225°C versus 280°C for the OV-101 coated column. Actual operating conditions were as follows: carrier gas flow, 3.1 ml/min; column dimensions, 25m x 0.5 mm ID; injection port temperature, 220°C; reactor base temper­ature, 240°G; temperature programming, 80°C isothermal (1 min), l°C/min to 160°C, 2°C/min to 200°C, 4°C/min to 220°C, and 220°C isothermal (25 min). As a result of these analyses, the CBW-20M coated SCOT column was selected for further GC/MS studies. Description of GC/MS System Having achieved satisfactory FID chromatograms of the phenol-rich sub-fractions (pentane-insoluble and B/E sub-fractions) with a CBW-20M coated SCOT column, the Curie-point flash vaporization device was transferred to a Hewlett-Packard (HP) 7620A gas chromatograph inter­faced to an HP 5930A mass spectrometer. This mode of operating conditions, with separation at least partially achieved by the gas chromatograph, uses the ion source of the mass spectrometer as detector for the column effluent. This results in a total ion current (TIC) chromatogram, analogous to chromatograms obtained by FID as shown in Figure 10. The traces shown are not continuous, but are composed of discrete segments, each segment corresponding to an actual mass spectrum of the compound(s) eluted. In addition, this mode of 50 phenols 10l 10; C; 1°FlO TIC) FlO MS en c .C..D. C TOTAL ION CURRENT SELECTED ~--- ION / TRACES ~ t / SPECTRA Figure 10. Schematic of GC/MS analysis mode. Mass spectrometer ion source serves as detector for the GC column effluent. U1 --' 52 operation permits the monitoring of selected ion traces as an aid in locating compounds of interest. Because this GC/MS system was originally designed for use with packed columns, the GC/MS interface had to be modified in order to retain the improved resolution and sensitivity achieved with the SCOT capillary columns. Accordingly, the original GC/MS interface, a Biemann-Watson molecular separator, was removed and other interfaces with lower dead volumes investigated. Of the several coupling methods tried, the one finally adopted was an "open-splitter" interface fabricated from a section of borosilicate glass capillary tube. The borosilicate capillary was drawn to form a narrow restriction. One end of the capillary tube was inserted directly into the ion source via a direct-probe inlet. The other end of the capillary tube was connected to the outlet of the SCOT column by a one-meter length of flexible fused silica capillary (0.3 mm I.D.). The restriction in the capillary allowed 0.5 ml/min, out of approximately 3 ml/min total column effluent, to enter the ion source of the mass spectrometer. Ion source pressure was maintained at 5 x 10"6 torr. This system caused no apparent loss in peak resolution when compared to FID chromatograms and thus, was deemed satisfactory. A schematic view of this GC/MS inter­face system is shown in Figure 11. GC/MS Analysis of Whole Liquid Distillates Total ion current (TIC) chromatograms obtained with this system and the CBW-20M coated SCOT column for the EDS atmospheric overhead, EDS atmospheric bottoms, and the hydrotreated counterpart of the atmospheric bottoms are shown in Figure 12. GC operating conditions splitter" probe 1.0.). 10 -6 FlO inter-face Oi sti 11 ates Figure 11. Schematic view of GC/MS "open splitter" interface. 1) Curie-point flash vaporization device; 2) glass-lined stainless steel tubing; 3) GC oven; 4) glass SCOT column; 5) fused silica capillary; 6) heated interface area (250°C); 7) open-type glass splitter; 8) restriction (.5 ml/min., atm. vacuum); 9) ion source; 10) vacuum chamber; 11) quadrupole rods. ^ CO ~ 01 W Figure 12. Total ion chromatograms from the GC/MS analysis of (a) the EDS atmospheric overhead; (b) the EDS atmospheric bottoms; and (c) the EDS hydrotreated distillate. (a) (b) (c) •••• I • 20 t(min) 00 .. 00 "0'" 00 00- ceo ..0 '"" ' 6.. 0 "..6 <."." r ..~ §. .. 0- 6 " I •• , •••• I ' , • , ., ••• I i •• , • I , , • I • i •• i." I" i i' .70 80 90 100 110 EOS EOS 55 were identical to those previously described for the CBW-20M coated SCOT column FID analyses. The operating conditions for the mass spectrometer were as follows: mass range scanned, 46-210 m/z; inter­face and source temperature, 250°C; and scan rate, 50 spectra/min. This relatively high scanning rate allowed up to 10 scans per GC peak, thus permitted good peak definition, and took maximum advantage of SCOT column resolution. Over 4000 mass spectra per run were subsequently recorded and stored on disk (HP 2100 mini-computer). The compound identities shown were based on comparisons with standard library spectra. For the range of retention times shown, the EDS atmospheric overhead chromatogram was completely dominated by (alkyl)phenol peaks, the more volatile hydrocarbon constituents (mainly "oils") having eluted during the first few minutes of the analysis. In contrast, the EDS atmospheric bottoms chromatogram was dom­inated by two- and three-ring (alkyl)aromatic peaks while that of the hydrotreated distillate was dominated by the corresponding hydro-aromatic compound peaks. For these latter two distillates, the only oxygen-containing compounds identified which exhibited well-defined peaks were dibenzofuran (retention time at approximately 91 min) and a few alkylphenols (retention times at approximately 79, 84, and 88 min). It is noteworthy that both of these distillates contained significant quantities of those classes of compounds deemed "necessary" for good solvent quality, namely; (1) polynuclear aromatics and (alkyl)phenols, and (2) potential hydrogen-donors [40, 62]. The former promote coal solubility and help stabilize the initial products of conversion (i.e., asphaltols) [35]. In addition, there is evidence that these species, along with some heterocyclic aromatics, e.g., dibenzofuran, can act as ; ,\ ~ { ,", r: ;;. FlO min. phenol dom- ~, inated alkyl ),~;, lY'', (, I. '. containing phenols, 62J. 35J. 56 H-shuttlers to transfer hydrogen from areas in the coal-solvent mixture that are relatively rich in hydrogen to other areas relatively lean; this activity is apparently enhanced by reduced levels of hydro­aromatics [9]. Note that the hydrotreated distillate contained significant quantities of hydroaromatics other than tetralin, the "classic" H-donor. In this regard, it has been reported that even when tetralin content is low, the presence of alternate hydroaromatics in large quantites can cause conversions as high or higher than those achieved when tetralin is used as the only H-donor [35]. The complex nature of these distillates is readily apparent from these TIC chromatograms. Even with GC separation and resolution retained, the sheer number of chemical species present results in a formidable problem of identification. However, examination of selected ion chromatograms (e.g., monitoring those ion traces characteristic of a specific compound or class of compounds) can be used to isolate areas of a chromatogram where a particular class of compounds might elute. Ion traces characteristic of certain compounds/compound classes developed from the GC/MS analysis of the atmospheric overhead are shown in Figure 13. The various compound classes shown are confined to specific segments of the overall chromatogram. By focusing on a specific segment, the problems of identification for a particular class are significantly reduced. Again, it is noteworthy that although the (alkyl)phenols in the EDS atmospheric bottoms and hydrotreated dis­tillates appear in only trace amounts, the selected ion trace characteristic of alkylphenols (m/z 107) indicates a whole series of alkylphenols to be present. shuttlers 9J. tetra1in 35J. phenols EOS characteristic of alkylphenols (m/z 107) indicates a whole series of a1kylphenols to be present. m/z t(min) Figure 13. Selected ion chromatograms for major fragment ions of specific chemical classes from the EDS atmospheric bottoms GC/MS analysis. Ion chromatograms were derived from the GC/MS analysis shown in Figure 12 (b). Ion series are chracteristic of alkanes (m/z 85); tetralins (m/z 131, 145); naphthalenes (*: m/z 128, 141, 155); phenols (m/z 107); dibenzofuran (m/z 168) and dimethyl-biphenyls (m/z 182). Note the retention of oxygen containing compounds (e.g., alkylphenols and dibenzofuran) away from the predominant aromatic and hydroaromatic compounds which elute earlier on the polar Carbowax-20M SCOT column. ~~~-= -~:::~.= =:.-. ----~---~-~--------~------------ 145 MAJ... ____ * . ~ 1281 --__ 114411//LI.....-----,,-..---------------- 107 ______ ~ __ ~~ _____________ ~ ______________________ ~L _.N~~A~~~~ _____________ _ 168 182 TIC 20 t(min) 30 40 50 60 70 80 100 110 EOS 20M 58 The monitoring of additional ion traces characteristic of (alkyl)- phenols, specifically, the (M-l)+ ions, indicated a variety of homologs present in each of these distillates. These traces are shown in Figures 14, 15 and 16. Whereas, the distribution of (alkyl)phenols in the EDS atmospheric overhead is primarily of phenol, cresols, and C^- phenols, those of the two heavier EDS distillates are dominated by C2-, C^- and higher alkyl homologs. In the identification of these (alkyl)phenols, retention times obtained from both literature references and analyses of standard compounds were used to compliment library spectra in the determination of the various isomers. One unexpected result from these GC/MS experiments was the dis­covery that the unhydrotreated and hydrotreated atmospheric bottoms exhibited almost identical alkylphenol distributions as shown by these selected ion traces. This indicated that the relative distributions of alkylphenols in both samples were nearly identical. This was sur­prising, since it has been reported in the literature [9] that the EDS process incorporates the catalytic removal of (alkyl)phenols. Either the alkylphenols were little affected or not at all, by the hydro-treatment, or else they were reduced in amount along lines of constant relative concentration. Figure 17 compares the TIC chromatograms and selected ion traces for the atmospheric bottoms with the hydrotreated distillate. Note that the expansion factors of the m/z 107 and m/z 121 ion traces indi­cate the total alkylphenol concentration in the hydrotreated distillate to be several times lower than in the atmospheric bottoms. In view of the expected different chemical reactivity of the various isomeric and a"lkyl)­phenols, a1kyl)pheno1s EOS C2- EOS C2-. C3- phenols, atmospheri'c distr-ibutions a1kylphenols 9J EOS phenols" hydro­treatment, Figure'17 m/z i-i-•-i-i-i-i I i i i § i-i i 'i'" i |-i-i-i-i i i-i-i-i-|-i-i-i-i-i-i-i i i-I-i-i i i-i-i-i-i i I i i i i i i i i i 50 60 70 80 90 t(min) Figure 14. Selected ion chromatograms from the GC/MS analysis of the EDS atmospheric overhead representing the major (alkyl)phenol fragments and parent ions. The m/z traces of 94, 108, 122 and 136 indicate the parent ions of phenol, methyl phenols, dimethyl- or ethyl phenols and C3~phenols, respectively. The m/z 136 trace indicates an absence of C3-phenols. Shaded peaks have tentatively been identified as molecular ions of: 1) phenol; 2) o-cresol; 3) p-cresol; 4) m-cresol; 5) 2,6-xylenol; 6) o-ethylphenol; 7) 2,5-xylenol; 8) 2,4-xylenol; 9) 2,3-xylenol; 10) 3,5-xylenol; 11) p-ethylphenols; 12) 3,4-xylenol; 13) m-ethylphenol. 1 3 6 • i i ii i ii HII i II Hi II Mil Uu i i lii i i u i M n i l i i z 136 122 121 108 107 94 TIC t (min) "blllllllM U I j tI I II! I II I II II , IDS phenol methylphenols, ethylphenols C3-cresol; ethy1phenol; 8) xy1enol; xyleno1; ethy1phenols; ethy1phenol. Figure 15. Selected ion chromatograms representing the major (alkyl)phenol fragments and parent ions from the total ion chromatogram of the EDS atmospheric bottoms. Shaded peaks refer to molecular ions of phenolic compounds as identified in Figure 14. The peaks numbered 14 and 15 refer to the molecular ions of 2-isopropylphenol and 2,3,5-trimethylphenol, respectively. 150 m/z . j ---------~~ .. -- - ~-,...----- .!;....,.-." ............ .....""...-.-..-.-,.= .. ~ '- 14 136 "----""--- .. ~ ____"_ = ____ - 121 =~,- : 107 94 --,.,.·.F . .~ ,_,..,.,. ..., ... ". . ,. ... .............,.. ...,. .._ ..0 "'1~'.".'.- . -.,..,._,.. ... ., ......,.. .......... -t.,... .. ad, , ~ .... .,..,......,.,. .... , ....... ., __ a.a ...... zctt .... ,~I~;j;=;~:;·,i,~i,:.,.--~,·;;7'7,'j:;j' 60 70 80 90 100 t(mln) phenol EOS 0"1 a m/z 1 5 0 a . . A- . 1 4 9 fljAndkAfrtmii. - n . . i ... . j JL » uutaj L 1 3 6 14 15 1 3 5 1 2 2 ,0- 1' 6. - - - -* - III! « Mil 111 MJ.Wll i.Wl HltO.1 . K». 1 2 1 - a 7 8 , U * \ l j 2 , 13 9 4 I r rir.i-S.it Miw.fi - i n - - 1 * - - - *1 i^ith rS i - - tV - t n*™lrt li - v A - * 1 * * " * * 1 " 1 " ' " TIC Figure 16. Selected ion chromatograms representing the major (alkyl)phenol fragments and parent ions from the total ion chromatogram of the EDS hydrotreated distillate. Shaded peaks refer to molecular ions and peak numbers correspond to those species identified in Figures 14 and 15. m/z 150 ~: .. ". .... " .•. : .:.,,:= : .• : :::.~ 136 14 15 _____________ ~ ........ ~_""'~~~ _ ___'._..JA ..... "" .... ___ .... ,.""' ___ _ 135 ---A __________________________ ~ __ .~. ______ ~~~ •• ~~ ________ _ 122 ...... "'.d' h -.. ' .................. ' .... 7 8 .. 6J( 121 108 ~ •• ' •• ""h't". ,.,e.t ... 1It l """ •• ,. , A" ••. 1 .. 107 ..... ~ 94 TIC .,.. ii' .'\ =; , .. : , , , i , , ,=, , , : i q; At , u75Plfb i q , I 1 , , ""'~""!'=-r:"" 60 70 80 90 100 t(mln) phenol en ----' CO" m/z133 R I / Z 1 2 1 _A A A- » -* m/z107 X 18 TIC co CO'" co- X 235 z121 * * <W>OoO 30 t(min) 40 50 60 70 80 90 100 X 82 X 96 110 Figure 17. GC/MS profiles of (a) the EDS atmospheric bottoms and (b) the corresponding hydrotreated dis­tillate obtained with a CBW-20M coated SCOT column. Major classes of aromatic and hydroaromatic compound dominate both chromatograms while the alkylphenol and (alkylJindanol series appear in almost trace amounts Both series are present, however, as indicated by the ion traces of m/z 107, 121 and m/z 133. Expansion factors at the end of each of these selected ion traces are relative to the parent distillate from which they were derived. X 48 m/z121· X 17 zl07 00 00" TIC 00 m/z133 m/z 121 0) m/zl07 . 98 "y • 00- 00'" w I~ cx9 i Ii . I •• I'" 30 t(mln) EOS compounds alkyl)indanol amounts. mjz end of each of these selected ion traces are relative to the parent distillate from which 0'1 N homologous alkylphenols, it is difficult to envisage plausible chemical events which could have lowered the total concentration of the alkyl­phenols (particularly along lines of constant relative concentration) during the hydrotreatment process [59]. Therefore, a dilution effect, rather than the results of direct involvement in hydrogenation was postulated. However, since information regarding any intermediate catalytic or other type of processing that may have occurred was un­available, the possibility of whole class reduction cannot be ruled out. The ion traces at m/z 133 were indicative of a second class of oxygen-containing compounds that were discovered and will be discussed in the next section. GC/MS and Direct-Probe, Low-Voltage MS Analyses of Whole Li qui? Distillate Sub-fractions While the analyses of the whole liquids were useful and informa­tive, particularly in regard to the dominating hydroaromatic/aromatics present, they were, in general, inadequate for oxygen-containing com­pound characterization. Even with the use of retention times and reference spectra, it can be seen (as shown from the ion traces of Figure 13) that interferences resulting from the co-elution of other compounds in this segment of the chromatogram can mask the presence of the alkylphenols and yield ambiguous compound identification. Therefore, having successfully achieved nearly complete sample recovery from the open-column LC separation technique, the sub-fractions of the distillates were analyzed under GC/MS conditions identical to those employed for the whole-liquids. The TIC 63 59J. infonnation mlz containing Liquid part~cularly containing com-pound sub­fractions liquids. 64 chromatograms of the EDS atmospheric bottoms B/E sub-fraction and parent distillate are shown in Figure 18 along with characteristic alkylphenol ion traces (i.e., m/z 107, 121). It can be seen that these ion traces are nearly identical for both analyses, thus implying a quantitative recovery of the alkylphenols in the B/E sub-fraction. Similar analyses of the hexane/benzene (H/B) and benzene/methanol (B/M) sub-fractions revealed only minor traces of alkylphenols. In fact, more than 95% of the alkylphenols appeared to be recovered in the B/E sub-fraction [59]. As mentioned earlier, the ion traces at m/z 133 of Figure 17 were indicative of a second series of oxygen-containing compounds. This series was first discovered by examination of the EDS atmospheric bottoms B/E sub-fraction chromatograms. While alkylphenols were re­sponsible for many of the peaks of the B/E sub-fraction chromatogram shown in Figure 18, a significant number of peaks resulted from the presence of other compounds. Further investigations indicated (alkyl)- indanols to be the second most abundant class of oxygen-containing compounds (after the alkylphenols) in this sub-fraction. At this point, a re-examination of the whole-liquid distillates showed that (alkyl)indanols were indeed present, but in very low amounts as indi­cated by the expansion factors for the m/z 133 ion traces shown earlier in Figure 17. To validate the presence of oxygen-containing species other than alkylphenols, in this sub-fraction, and to quickly examine other sub-fractions, a direct-probe, low-voltage MS Curie-point technique was employed. These analyses were performed on a commercially available Py-MS Curie-point instrument (Extranuclear Model 5000-1) under the a1ky1phenols 8/S) 8/M) alky1pheno1s. alky1pheno1s 59J. containing alkyl)­indanols liquid containing alkyl phenols, sub­fractions, probe, MS 65 70 80 90 100 110 t ( m i n ) Figure 18. Total Ion Current (TIC) chromatograms obtained with a CBW-20M coated SCOT column of the EDS atmospheric bottoms and the B/E sub-fraction derived from this distillate. The selected ion traces for both analyses are very nearly identical, although the parent distillate is dominated by aromatic and hydroaromatic species whereas the B/E sub-fraction is composed almost entirely of (alkyl)indanols and alkylphenols. A) WHOLE LI QU 10 m/z107 B) BENZENE/ETHER FRACTION "--- OH «? ..l ;. ~ Oel 6a& l&. '.&. :Jt" ~ TIC £15 110 t(m.") alkyl phenols. 66 following operating conditions: sample size, 25 yg; Curie-point, 510°C (Fe/Cr/Ni ferromagnetic filament); rise-time to Curie-point, 6 seconds; total heating time, 10 seconds; electron energy 12 eV; emission current, 1.75 mA; ion energy, 6.7 eV; mass range scanned, 40-240 m/z; scan rate, 1000 amu/second; total scan time, 20 seconds; and background pressure, 2 x 10"^ torr (measured at the oil diffusion pump). Mass scans were summed by computer with a real-time signal averaging program called PYR0. The resulting integrated mass spectrum for each sample was stored on disk (HP 2100 mini-computer). The computer disk was then transferred to a HP 21-MX computer where the spectra were processed with a separate program (MASSA) for peak detection, identification and integration. Both the PYR0 and MASSA programs were developed at the Biomaterials Profiling Center, University of Utah [59]. A solution of standard reference compounds was prepared for com­parison with the sub-fractions. This solution, using pesticide-grade benzene as solvent, consisted of the following reagent grade standards: phenol (100 ng/ml), p-cresol (200 ng/ml), p-ethylphenol (400 ng/ml), o-isopropylphenol (200 ng/ml), 1,3-dihydroxybenzene (200 ng/ml), 4- methyl-1,2-dihydroxybenzene (200 ng/ml), 5-indanol (200 ng/ml), 1-naphthol (200 ng/ml), and 2-hydroxybiphenyl (100 ng/ml). The results of these direct-probe low-voltage MS analyses are shown in Figure 19. The use of low electron energy results in spectra dominated by molecular ions. The mass spectrum of the standard mixture appears relatively simple when compared to those of the various dis­tillate sub-fractions. Each of these distillate sub-fraction spectra contains complex patterns indicative of several series of oxygen-containing compounds. In contrast to the spectra of the other ~g; 10-7 PYRO. MX MASSA) PYRO 59J. grade ethy1phenol dihydroxybenzene l,2-dihydroxybenzene l-naphthol probe 67 101 5 - STAN0AR0 MIXTURE 94 / .-1 no (a) 134 144 ,170 i i i 1 r PENTANE INSOL. rovdu* [EDS Atm. Bottom*] 110 5 - 94. 1 , ,.1,11111,, (b) i[iiiiiiiiijiiiiiiiiijiiiiiiiiiiin.i,».|., UJ 35 z s hi Z o ui 10- > < -J Ui OC 5 - BENZENE/ETHER fraction [EDS Atm. Bottom*] JMllli ,111111111, BENZENE/ETHER fraction [EDS Hydrotrootod Bottom*] 5 - BENZENE/ETHER fraction [h-COAL Atm. Bottom*] M / Z Figure 19. Direct-probe, low-voltage mass spectra of a standard mixture and several polar coal liquid distillate sub-fractions. Homologous ion series are thought to represent (alkyl)phenols (be­ginning m/z alkyl)naphthalenes m/z 128), indanols at naphthols alkyl)(alkyl)hydroxybiphenyls/hydroxyacenaphthenes alkyljanthracenes/phenanthrenes at a series of unknowns possibly including (alkyl)hydroxy-tetrahydroacenaphthenes z in arbitrary units.) 10 -; .! 10 fI) 1&1 t-in ffi !I t­Z 10 STANDARD MIXTURE 8ENZENE/ETHER f'IIC/lolt {EDS Aim. 8IIttll""] / / ./ 8ENZENE/ETHER f'lIc/lon { EDS Hyd,o" •• ,.d 811_] . / / / 8ENZENE/ETHER frr,etlan [H-COAl AI",. /Jotto",.] 60 (a) (b) (e) (d) (e) phenols (be­~ inning at 117), (a1kyl)naphtha1enes (beginning at (alkyl)quinolines (beginning at m/z 129), (alkyl)indanols (beginning at m/z 134), (alkyl)naphtho1s (beginning at m/z 144), (a1ky1)carbazoles (beginning at m/z 167), (alkyl)hydroxybiphenyls/hydroxyacenaphthenes (beginning at m/z 170), (alky1)anthracenes/phenanthrenes (beginning at m/z 178) and a series of unknowns possibly including (alkyl)hydroxy­tetrahydroacenaphthenes (beginning at m/z 174). (Intensities are in indanols dihydroxy-benzenes) in this sub-fraction. This was possibly due to the extraction of these compounds into the pentane-residue sub-fraction by the action of other species present which may have influenced solute/- solvent m/z alkyl)alkyl)phenanthrene/respectively) spilled over" insoluble sub-fraction. These results further emphasize the problems associated with class characterization by solvent-extraction [96]. EDS Z^- C^-m/same conditions when coal is present [9]. They also exhibit low volatility and were not observed in GC/MS studies. The peak series indanols alkyl)hydroxybiphenyls/- hydroxyacenaphthenes (beginning at m/z 170) are present in both EDS B/E sub-fractions, but show obvious differences. The peak intensities of the (alkyl)indanols are lower for the first few members of this series in the hydrotreated sub-fraction, but increase for the apparent 68 sub-fractions, the spectrum of the pentane-insoluble residue (dissolved in methanol) indicates the alkylphenols and (alkyl)indanols to be present, but in lesser amounts relative to other series (e.g. dihydroxy­benzenes) solute/­solvent parameters, as earlier mentioned. Moreover, the series beginning at 128 and m/z 178 indicate that some aromatic hydro­carbons ((alkyl )naphthalene and (a1kyl)phenanthrene/anthracene, respecti ve1y) have "spi 11 ed over II into the pentane-i nso1 ub1 e sub­fraction. 96J. Spectra for both EOS B/E sub-fractions exhibit similar alkylphenol distributions with maxima in peak intensities occurring at the C2- or C3-phenol molecular ions (m/z 122 and m/z 136, respectively). The dihydroxybenzene and naphthol series appear to be entirely absent from the hydrotreated distillate as indicated by the lack of characteristic peak series (beginning at m/z 110 and m/z 144, respectively). While these two classes of compounds are stable under liquefaction conditions in the absence of coal, they are known to rapidly decompose under the 9J. MS characteristic of the (alkyl)indanols and (alkyl )hydroxybiphenyls/­hydroxyacenaphthenes EOS alkyl )69 homologs of higher molecular weight. Whether these differences repre­sent a shift in the average molecular weight of the (alkyl)indanols in the hydrotreated distillate or are a result of contributions from other ion series has not as yet been determined. This increase in peak intensity at high molecular weight is also prominent for the series beginning at m/z 170 mentioned above. Further, the presence of a relatively strong homologous ion series apparently beginning at m/z 174 in the hydrotreated sample spectrum may well represent a series of hydroaromatic compounds formed during the hydrotreatment process [59]. The spectrum of the H-Coal B/E sub-fraction indicates (alkyl)- indanols to be the dominant series of oxygen-containing compounds. The presence of several other homologous ion series significantly com­plicates this spectrum and is believed to be due to indoles (m/z 117) and quinolines (m/z 129). The latter of these two has been implicated in the hydrogenation process as a possible catalytic agent in hydrogen-transfer mechanisms [97]. Whether the presence of these nitrogen-containing compounds is characteristic of the H-Coal process, the result of coal feedstocks different than those used in the EDS process, or the result of different conditions for sample collection, storage and shipment (as mentioned earlier), is uncertain. In light of these discoveries, it was felt that the investigation of these sub-fractions should be pursued in greater detail. One of the obvious choices was further characterization via GC/MS studies. While GC/MS analysis posed obvious disadvantages with regard to compound series of low volatility, it offered the opportunity for a detailed characterization of the (alkyl)indanols. A detailed investigation of this compound class was felt to be particularly significant for several alkyl )59J. 8/alkyl)­indanols containing hydrogen­transfer 97J. nitrogen­containing EOS 70 reasons. Besides appearing to be chemically reactive during hydro-treatment (as indicated by the peak intensity differences of Figure 19), it was either the first or second most abundant class present in the various B/E sub-fractions. The CBW-20M coated SCOT column used earlier for GC/MS studies had been selected because of its ability to separate polar compounds from most of the aromatic and hydroaromatic hydrocarbons. However, the open-column LC technique had resulted in four sub-fractions separated according to varying degrees of polarity. Because the B/E sub-fractions were found by direct-probe low-voltage MS studies to be composed primarily of oxygen-containing species (presumably of similar polarity), it was felt that the characteristics of the CBW-20M substrate would be marginal for optimal analysis of these B/E sub-fractions. Therefore, it seemed that the best separations could be achieved on the basis of differences in individual boiling points and an 0V-101 coated SCOT column was selected for these analyses. The use of the 0V-101 substrate also permitted higher maximum operating temper­atures with a corresponding reduction in problems of elution for higher-boiling components. Results of the GC/MS analyses of the EDS atmospheric bottoms and hydrotreated distillate B/E sub-fractions are shown in Figures 20 and 21, respectively. MS operating conditions were similar to those em­ployed earlier. Operating conditions for the 0V-101 SCOT column were also similar to those used earlier for FID analyses with this column, but with modifications in temperature programming as follows: 70°C isothermal (1 min), 2°C/min to 160°C, 4°C/min to 200°C, 6°C/min to 250°C, and 250°C isothermal (30 min). hydro­treatment sub­fractions probe containing sub­fractions. OV-10l OV-10l OV-10l 70 0e 71 m/z 162 m/z 1 48 x 4 1 .5 x 1 8 .5 m/z 134 x 5 I I I I 1 1 1 - I - I I I I IT i i I i i i i-r-r-r*~r-t 4 3 # 52 I I I I 1 I I ! I I I I I I I I I I I I I I I i I I ' 1 I T T m / z 164 m / z 150 m / z 136 m / z 122 JtJLk x 2 6 .5 x 10 x 2 .5 m/z 108 x 2 x 3 .5 1 0 • i i t | i . i • i . i 2 0 | > • • • • i i i i | t i * i * i ' i • 3 0 40 t (min) Figure 20. GC/MS profiles of the B/E sub-fraction of the EDS atmo­spheric bottoms distillate obtained with an OV-101 coated SCOT column. Shaded peaks in the selected ion traces indicate molecular ions. The lower series of ion traces are indicative of alkylphenols, while the upper series are indicative of the parent ions of (alkyl)indanols. Numbers on the TIC chromatogram refer to tentative identities as listed in Table IV. x 41.5 148 x 18.5 mli x 5 11 1 • TIC m/z 164 x 26.5 m/z 150 x 10 m/z 136 )( 2.5 m/z 122 x ·2 x 3.5 10 20 30 40 E 10l alkyl phenols, alkyl )IV. 72 m/z 162 It , 1 , L X III 85 m/z 148 i 111 U X 33 m/z 134 X 1 3 TIC 6 J • ? I L | L | • I • r r 111 i 5 i 7 25 2 • . 1 . . i • 133 I' so • m/z 164 ' 1 ' ' 1 ' • 1' X 3 1 m/z 150 X 13 m/z 136 L A X 3 m/z 122 i X 2 m/z 108 X 4 . ... 1 . . . . . . 10 20 ' 1 ' 30 111 40 t (min) Figure 21. GC/MS profiles of the B/E sub-fraction of the EDS hydro-treated distillate obtained with an 0V-101 coated SCOT column. The lower series of selected ion traces are indicative of molecular ions of alkylphenols and are very nearly identical to the corresponding series of Figure 20. Some differences are noted with respect to the lower weight alkylphenols (i.e., m/z 108 and m/z 122) most probably as a result of early evaporation of these more volatile components during insertion of the Curie-point filament into the reactor. The upper series of selected ion traces show substantial differences when com­pared to the analogous series of Figure 20. These differences are reflected by the arrows shown on the TIC chromatogram, and are thought to correspond to the loss of 5-indanol and its higher alkyl homologs. • m/z 162 m/z 148 lC 33 m/z 134 lC 13 ! 1.7 .7 TIC .I.S 2: 3J 13 69 m/z 164 x :I m/z I~O x 13 m/z 136 x 3 m/z 122 x 2 m/z 108 )( , , .. , . , , 10 20 30 40 t< min) hydro­treated OV-10l a1ky1pheno1s 73 The proposed chemical identities of the peaks shown in the TIC chromatograms along with the measured retention indices of these peaks are listed in Table IV [59]. The ion chromatograms at m/z 108, 122, 136, 150, and 164 represent molecular ions (shaded in black) of C-j- to Cg- alkylphenols, as well as some fragment ions. Of these, only the m/z 108 (cresols) and m/z 122 {C^- alkylphenols) ion traces show appre­ciable intensity differences between the two samples. These differences were probably a result of sample loss by evaporation from the Curie-point filament prior to flash vaporization. The remaining alkylphenol ion traces for the two samples are essentially identical, in agreement with results obtained using the CBW-20M substrate. The TIC chromatograms obtained from these two analyses, however, exhibit marked differences, as indicated by arrows on the hydrotreated distillate B/E sub-fraction TIC chromatogram (Figure 21). These arrows depict an absence of peaks when compared to the chromatogram of the atmospheric bottoms B/E sub-fraction (Figure 20). The loss of these peaks is reflected in the molecular ion traces of m/z 134, 148, and 162. The two peaks shown in the m/z 134 ion trace of the atmospheric bottoms B/E sub-fraction were tentatively identified as the molecular ions of 4-indanol and 5-indanol (TIC chromatogram peaks #25 and #29, respectively). In the corresponding ion trace of the hydrotreated sub-fraction, one of these peaks (TIC chromatogram peak #29) is absent. Other peak losses can be noted in the m/z 148 and m/z 162 molecular ion traces of this sub-fraction and were thought to correspond to the alkyl homologs of 5-indanol. The presumed loss of 5-indanol and its higher homologs is the apparent reason for the lower peak intensities 59J. Cl - C5- alkyl phenols, (C2- sub­fraction, 74 TABLE IV HYDROXYAROMATIC COMPOUNDS IDENTIFIED IN BENZENE/ETHER FRACTION OF EDS ATMOSPHERIC BOTTOMS AND HYDROTREATED ATMOSPHERIC BOTTOMS Proposed Chemical Identity phenol impurity o-cresol m-cresol, p-cresol 2,6-xylenol o-ethylphenol 2,4-xylenol m-ethylphenol, p-ethylphenol 3,5 xylenol, 2,3 xylenol, C3 phenol 2-isopropylphenol o-n-propylphenol C3-phenol C3-phenol C3-phenol m-n-propylphenol," p-n-propylphenol methyl ethyl-and/or i sopropylphenol C3- and C4-phenols C3- and C4-phenols C4-phenol C4-phenol C^-phenol C4-phenol indanol, C4-phenol C4-phenol C4-phenol C4-phenol 5-indanol methylindanol methyl-3-n-propylphenol methylindanol, C5-phenol methylindanol, 3-n-butyl-and/or isobutylphenol C4-phenol methylindanol, Cs-phenol methylindanol, C2-indanol C2-indanol, Cs-phenol C5-phenol C2-indanol, methylindanol, C5-phenol methylindanol, Cs-phenol C2-indanol, C5-phenol methylindanol, C2-indanol, Cs-phenol Cs-phenol Peak Retention Kovats Index2 1 978 2 11.5 3 12.1 1037 4 13.3 1060 5 1085 6 16.1 1117 7 16.4 1128 8 1152 9 1157 10 1173 11 1180 12 20.9 1200 13 21.2 1206 14 21.5 1209 15 21.8 1217 16 22.1 1220 17 1241 18 23.7 1247 19 24.0 1250 20 1258 21 25.1 1267 22 25.5 1373 23 25.9 1280 24 1286 25 26.4 1291 26 1297 27 27.5 1305 28 27.8 1310 28.1 1316 30 28.6 1324 31 1327 32 29.1 1332 33 29.6 1338 34 29.9 1344 35 30.6 1356 36 31.1 1364 37 1367 38 31.9 1376 39 32.4 1387 1392 41 33.1 1398 42 33.5 1403 43 34.0 1410 IV EDS #1 Time Index2 (min) 9.2 l1.S 12. 1 cresol, S 14.3 xy1eno1 16. 1 ethy1pheno1 xyleno1 17.9 l1S2 ethylphenol, ethylpheno1 18.3 l1S7 3,S xyleno1, xyleno1, (. 19.4 3,4 xylenol i 19.7 2-;sopropylphenol '\ propy1pheno1 ~, 21. 2 {} 21. 5 C3-phenol ~ 21. 8 C3-phenol C3-phenol 23.3 n-propy1phenol,. propy1phenol methy1ethyl-isopropyl phenol \;' I 24.5 C3- and C4-phenols '~>,, I 2S.l ?t)' " 2S.5 C4-phenol t C4-26.3 C4-phenol :{:., 4-indano1, pheno1 1\' ~ , 26.9 C4-phenol pheno1 pheno1 29 28. 1 5-indanol 28.8 Cs-phenol isobutylpheno1 C4-phenol 31. 1 31.3 C5-31. 9 Cs-phenol Cs-40 32.8 33. 1 CS-33.S methy1indanol, indano1, CS-TABLE IV (continued) Peak l 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 Retention Time (min) 34.7 35.0 35.4 35.8 36.0 37.1 37.3 37.7 38.2 38.7 39.3 39.9 40.6 40.6 41.1 41.9 43.1 43.5 44.1 44.6 45.1 45.7 Kovats Index2 1423 1427 1434 1439 1444 1460 1466 1471 1482 1490 1494 1507 1514 1517 1541 1560 1569 1576 1586 Proposed Chemical Identity methylindanol C2-indanol, C5-phenol ethylindanol C2-indanol C2- indanol C2- indanol C2-indanol C2-indanol hydroxybiphenyl3, C2~indanol, C4»phenol methylhydroxybi phenyl3 napthol + Cs phenol C2- and C3~indano