s). .. t... z 3.4 v :59. ,1: . .g . ‘3 . 9' it... 5...... _... va,.~\.>‘< n..£.'i.’ .: I. .1-.. z; «.1. we’ll-'1 do!" 6’ 0;. '53.. 33.92»: V , of»! ._ . . (3...! ‘ .ou . «c. are. I: .I (a . 1;...oall'i 3”" la. r .v pin-5...}. pol-hilt.) .1 ..I r klfi. t, a1:nt:-ha~; 9.1.1.. i. . {1-1:}... 1 r 1., I. . .c ..J.. .9iLIIJ. 39.5%.... .I 3.2.... 2:... 1.... . u .92: .. 3.3.17... .1 . A: at. ‘15:: ~51: .413... !. .. 1?. .. (I .\ a. . 2.-. a. . Yr . an. 3:. i. I'll?! . . ., l .etlrvnl...71xb . I... .555... :9: a... 2.1.... . n k avrr Iii}... A 3‘71. :0): V. 3.51.. . ~ ’34:. I 5|. «(#0. no: . . .y . lttll‘. I); . {aplhvrp‘ti 5!. .\.I 1.. . . 99:13. . ‘ . é‘t...’ :1... ...r.. sixzirl: 3.3.3.1.}... . 11:32....)9119. 5.3!...- »t...- ntfaglil. .. .fulé. . . I:‘llh....)t:a:1 LI... . zivAOvl 53...... .1.... cl.- t .2. . . . . : a 1....eivxs‘zt001 . p 3;. h 1.7 sl" 1. tlJlotbu..A..rl.. {HIV J. :6. .31... .... |.v . . . :1 r4 2 ;I..:. u. ‘1..|...... .zd. 1..le so 41>‘>}o..uirsl. lJ-s WESR SITY LIBRARIES |1111|11||1111111111111111191911 | 111| 3 1293 008859 This is to certify that the dissertation entitled Improved Compound Resolution for Chromatography With Rapid Full Mass Spectral Detection And .A Photodissociation System For Tandem Time-of-Flight Mass Spectrometry presented by Richard David McLane has been accepted towards fulfillment of the requirements for Ph. D. degree in Analytical Chemistry flfia Major professor Date Wig; [993 MS U is an Affirmative Action/Equal Opportunity Institution 0-12771 LIBRARY 1 Michigan State University PLACE IN RETURN BOX to remove this checkout from your record. TO AVOID FINES return on or before date due. _______________——————‘ DATE DUE DATE DUE . DATE DUE MSU Is An Affirmative ActionlEqual Opportunity Institution cmWMS-o.‘ _.._—- IMPROVED COMPOUND RESOLUTION FOR CHROMATOGRAPHY WITH RAPID FULL MASS SPECTRAL DETECTION AND A PHOTODISSOCIATION SYSTEM FOR TANDEM TIME-OF-FLIGHT MASS SPECTROMETRY By Richard David McLane A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Chemistry 1 993 Christie G. Enke, Advisor ABSTRACT IMPROVED COMPOUND RESOLUTION FOR CHROMATOGRAPHY WITH RAPID FULL MASS SPECTRAL DETECTION AND A PHOTODISSOCIATION SYSTEM FOR TANDEM TlME-OF-FLIGHT MASS SPECTROMETRY By Richard David McLane Combination of a time-of-flight mass spectrometer (T OFMS) with a storage ion electron ionization source [1] and time-array detection (TAD) [2] has produced an instrument capable of acquiring one hundred or more mass spectra each second with high sensitivity. This system overcomes the trade-off between mass spectral acquisition rate and sensitivity that exists for scanning mass spectrometers. These features are well suited for the detection of compounds as they elute from a capillary gas chromatography column. A deconvolution approach was developed to take advantage of small temporal differences occurring across the elution profiles provided by rapid sampling of the TOFMS/T AD instrument. The retention times of unknown compounds are found by generating a mass chromatogram peak position plot from the retention times of all mass chromatograms. A unique m/z value is determined for each coeluting compound and a pure mass spectrum for each compound is extracted using a cross-correlation approach. Deconvolution approaches can be used to either locate more compounds for a given amount of chromatographic time or to reduce the amount of time required for an analysis. The former approach was compared to two-dimensional gas chromatography with mass spectral detection (2-D GC/MS) to assess the strengths and weaknesses of each approach. The two approaches were found to be complementary. The time savings available from deconvolution was counterbalanced by the wider dynamic range afforded by the physical separation provided by 2-D GC/MS. The latter approach, called time-compressed chromatography, combines low resolution high-speed chromatography and the TOFMS/T AD system with deconvolution techniques to decrease the analysis times by a factor often or more. The technology of the TOFMS instrument has been extended to tandem mass spectrometry in a tandem time-of—flight mass spectrometer. A photodissociation system was developed and used to photofragment ions at the focal plane of the ion mirror in the first time-of-flight mass analyzer. Precursor depletion efficiencies of over 90% and photodissociation efficiencies of greater than 22% have been achieved using this system. 1. Grix, Ft.; Kutscher, Ft; Li, G.; Gruner, U.; Wollnik, H. Rapid Commun. Mass Spectrom. 1 988, 2, 83. 2. Holland, J. F.; Newcombe, 8.; Tecklenburg, R. E., Jr.; Davenport, M.; Allison, J.; Watson, J. T.; Enke, C. G. Fiev. Sci. Instrum. 1991, 62, 69. Copyright by RICHARD DAVID MCLANE 1993 ACKNOWLEDGMENTS No one ever achieves any goal in life without help from others. I am indebted to a large number of maple for the help and support they have given me over the course of many years. This is especially for all of the collaborative efforts in which I was involved. The interchange of ideas and recognition of individual strengths creates excitement and achieves results. First of all, I would like to thank my preceptor, Dr. Chris Enke, for his insight and guidance which have led to the work described in this thesis. I would also like to thank the other members of my guidance committee: Dr. Crouch, Dr. Sweeley, Dr. Babcock and Dr. Jackson for their assistance and ideas. The individuals in the time-of-flight mass spectrometry group over the past decade provided the foundation for my research. Without them, our deconvolution approach would have only limited utility and the tandem time-of- flight instmment would not have been constructed. Dr. Raimund Grix, Dr. Ron Tecklenburg, Ben Gardner, Dr. John Allison and Dr. Jack Holland have all played roles in the development of instrumentation during my "lifetime” at MSU. I would like to thank them for many useful discussions. Dr. Ron Tecklenburg was also of tremendous help in developing my understanding of photodissociation and provided a sounding-board for many of my ideas. Dr. George Yefchak, with many useful inputs from Dr. Jack Holland, worked with me on the development of our deconvolution approach. George‘s V insights and programming skills were priceless. Thanks also go to David Pinkston and Dr. Pete Rodriguez for their role in the two-dimensional gas chromatography project. Mary Puzycki/Seeterlin and Paul Vlasak have proven to be excellent coworkers during the design and construction of the tandem time-of-flight mass spectrometer. Although we have not always agreed on things, our discussions have increased my understanding in many areas. Doug Beussman has already proven his worth to the TOFfTOF project and I thank him for it. Eric Hemenway has always been a good friend and a tremendous resource for computer/programming problems. I would like to thank him for his patience while teaching me so much. Marty Rabb, Russ Geyer, Dick Menke and Sam Jackson have always provided us with tremendous service and support in designing and building components for the TOF/TOF instrument and deserve an ovation for their efforts. No one survives the ordeal of graduate school without a support network. I would like to thank the members of the Enke Group (past and present) for their moral support and assistance. Brett Quencer has proven to be a good and understanding friend who helped me to retain a shred of sanity via our summertime golf expeditions. Most importantly, I would like to thank my friends and family who have always remained by my side. Dr. Mike Baim and Dr. Pam Schofield have been VI good sources of encouragement over the past several years. My wife and love, Debbie, has been with me through all of the ups and downs; she has made the good times better and lessened the disappointments. My daughter, Alyson, always provides me with a source for joy. Thanks also to my mother and brother, Dan, for their constant encouragement and support. vii TABLE OF CONTENTS List of Tables ..................................................................................................... xiv List of Figures ........................................................................ xv Chapter 1 lntroductlon 1.1 Introduction ...................................................................................... 1 1.2 The Problem With Conventional Mass Spectrometry ....................... 5 1.3 The Michigan State University Solution ........................................... 10 1.4 Research Goals ................................................................................ 10 References ............................................................................................. 11 Chapter 2 Introduction to TIme-of-Fllght Mass Spectrometry and Deconvolutlon ................................................................................................. 1 4 2.1 Introduction ...................................................................................... 14 2.2 Detectors for Capillary Gas Chromatography .................................. 16 2.2.1 Single Channel Detectors ................................................... 17 2.2.2 Multichannel Detectors ....................................................... 17 2.2.3 Mass Spectral Array Detectors ........................................... 19 2.3 Introduction to TOFMS for Chromatographic Detection ................... 20 2.3.1 Limitations of TOFMS ......................................................... 23 2.4 The MTOF/ITR System .................................................................... 25 2.4.1 Development of the Ion Source .......................................... 25 2.4.2 lon Mirrors .......................................................................... 30 2.4.3 Time-Array-detection .......................................................... 31 2.4.4 Characteristics of the MTOF Instrument ............................. 33 2.5 Deconvolution of GC/MS Data ......................................................... 34 vlli 2.5.1 Characteristics of GC/MS Data ........................................... 35 2.5.2 Types of Deconvolution ...................................................... 36 2.6 Historical Deconvolution Approaches for Unknowns ........................ 37 2.6.1 Locating Unresolved Compounds ....................................... 38 2.6.1 Extracting Pure Mass Spectra ............................................ 40 2.7 Summary .......................................................................................... 42 References ............................................................................................. 43 Chapter 3 Deconvolutlon of GC/MS Data: Better Data Make An Old Technlque Work .............................................................................................. 49 3.1 Introduction ...................................................................................... 49 3.2 Approach .......................................................................................... 50 3.2.1 Location of Coeluting Compounds ..................................... 51 3.2.2 Determination of the Retention Time of Each Coeluting Compound ................................................................................... 55 3.2.3 Extraction of a Pure Mass Spectrum for Each Unresolved Compound ................................................................ 58 3.3 Experimental .................................................................................... 60 3.3.1 Reagents ............................................................................ 60 3.3.2 Gas Chromatography ......................................................... 61 3.3.3 Time-of-Flight Mass Spectrometer ..................................... 61 3.4 Effectiveness on Gasoline Range Hydrocarbon Mixture .................. 61 3.5 Dynamic Range Studies ................................................................... 67 3.6 Summary and Future Work .............................................................. 70 References ............................................................................................. 71 ix Chapter 4 Time-Compressed Gas Chromatography/Mass Spectrometry: Fifty-Two Compounds In Eighty Seconds ........................... 73 4.1 Introduction ...................................................................................... 73 4.2 Experimental .................................................................................... 75 4.2.1 Sample Preparation. ........................................................... 75 4.2.2 Gas Chromatography. ........................................................ 75 4.2.3 Mass Spectrometry. ............................................................ 77 4.2.4 Deconvolution Approach. ................................................... 78 4.3 Results and Discussion .................................................................... 78 4.3.1 GC/MS Analysis Optimized for Component Separation. 79 4.3.2 Analysis of the Sample by Time-Compressed Chromatography. ......................................................................... 81 4.3.3 Adequate Chromatographic Resolution. ............................. 85 4.3.4 Poor Chromatographic Resolution of Compounds with Similar Spectra. ........................................................................... 87 4.3.5 Overlap of More Than Two Compounds ............................. 89 4.3.6 Quantitation. ....................................................................... 92 4.3.7 Limitations. ......................................................................... 93 4.4 Conclusions ...................................................................................... 95 References ............................................................................................. 96 Chapter 5 Comparison of Two-Dimensional Gas Chromatography/Mass Spectrometry wIth Deconvolutlon of Gas Chromatography/ High-Speed Mass Spectrometrlc Data ............................ 97 5.1. Introduction ..................................................................................... 97 5.2 Experimental .................................................................................... 102 5.2.1 Initial GC Analysis .............................................................. 102 5.2.2 Two-Dimensional GC/MS ................................................... 102 5.2.3 GC/T OFMS/T AD with Data Deconvolution ......................... 103 5.3 Results and Discussion .................................................................... 104 5.3.1 Analysis of Region 1 ........................................................... 105 5.3.2 Analysis of Region 2 ........................................................... 113 5.4 Conclusions ...................................................................................... 123 5.5 Summary .......................................................................................... 125 References ............................................................................................. 126 Chapter 6 Photodlssoclatlon In Tandem Mass Spectrometry ..................... 128 6.1 Introduction ...................................................................................... 128 6.2 Qualities Required For MS/MS Using TOF Instmments .................. 128 6.3 Photodissociation in Mass Spectrometry ......................................... 130 6.4 ‘l'rme-of-Flight Instruments in MS/MS Analysis ................................. 136 6.5 The Tandem Time-of-Flight Mass Spectrometer .............................. 139 6.6 Theoretical Considerations .............................................................. 141 6.7 Design and Constmction of the TOF/l' OF Instrument ...................... 144 6.8 Overview of Other Areas .................................................................. 145 6.8.1 Ion Generation and Storage ............................................... 145 6.8.2 Time-of-Flight Separation of Precursor lons ....................... 147 6.8.3 Gating of the Precursor lons ............................................... 147 6.8.4 Time-of-Flight Analysis of Product lons .............................. 150 6.9 Summary and Conclusions ............................................................... 151 References ............................................................................................. 152 xi Chapter 7 System Control of the TOF/TOF Instrument ................................ 157 7.1 Introduction ...................................................................................... 157 7.2 Spatial-Temporal Considerations for the Interaction Region ........... 158 7.3 Laser Considerations ....................................................................... 158 7.4 System Control ................................................................................. 162 7.4.1 Delay Generator ................................................................. 164 7.4.2 Transient Recorder ............................................................. 165 7.4.3 Laser ................................................................................... 170 7.5 Command Flow for Data Acquisition ................................................ 173 7.6 System Control Software .................................................................. 175 7.6.1 Overview of the TOF/TOF Instrument System Control Programs ..................................................................................... 176 7.6.2 lnstmmental Characteristics Requiring Special Attention ...................................................................................... 177 7.7 Summary and Future Work .............................................................. 180 References ............................................................................................. 182 Chapter 8 Characteristics of the Ion Fragmentation Process In the TOF/T OF Instrument ....................................................................................... 183 8.1 Introduction ...................................................................................... 183 8.2 Characteristics of the Isomass Ion Packets ...................................... 184 8.3 Laser Timing and Focusing .............................................................. 189 8.3.1 Laser Beam Characteristics ................................................ 190 8.3.2 Laser Alignment .................................................................. 190 8.3.3 Laser Timing ....................................................................... 191 8.4 Photodissociation in the TOF/TOF Instrument ................................. 192 xii 8.4.1 Precursor Ion Resolution .................................................... 201 8.4.2 Photodissociation Characteristics ...................................... 201 8.4.3 Critical Timing Precision ..................................................... 205 8.5 Summary and ConclUsions ............................................................... 209 8.6 Future Work ..................................................................................... 210 References ............................................................................................. 211 Appendix: System Control Programs for the Tandem Time-of-Flight Mass Spectrometer. ........................................................................................ 213 A.1 TOF/T OF Instrument Control Program ............................................ 214 A.1.1 Include File Called by the TOF/TOF Instrument Control Program .......................................................................... 228 A12 Function Panels Called by the TOFfl' OF Instrument Control Panel ............................................................................... 230 A2 Program to Parse Data File Header for Acquisition Information ...... 236 A.3 Program to Convert LeCroy Format Binary Data to ASCII Format .................................................................................................... 238 A31 ASCII _gen Instrument Used by the Format Conversion Program ....................................................................................... 239 A32 Include File Called by the File Format Conversion. Program ....................................................................................... 247 xiii LIST OF TABLES Table 4-1. The 61 Compounds in the Test Mixture. ......................................... 76 Table 6-1. Historical Background of Photoionization/Photodissociation in MS ..................................................................................................................... 132 Table 7-1. Comparison of Excimer and NszAG Lasers .................................. 160 Table 7-2. Summing Times (Summed Averages per Second) for Different Numbers of Points at Different Frequencies ..................................................... 168 Table 8-1. Arrival times for m/z 156 and m/z 158 from acquisitions of one transient per mass spectrum using bromobenzene .......................................... 208 xiv LIST OF FIGURES Figure 1-1. The acquisition of only a few mass spectra per second can severely distort the shape of the chromatographic elution profile. Determining the number of peaks in a chromatogram is difficult under these conditions. ....................... 7 Figure 1-2. Skewing results from rapid changes in the partial pressure of the analyte in the ion source relative to the scan rate. After a scan has been acquired, all m/z values plotted in a mass spectrum are assigned to the same time. .................................................................................................................. 9 Figure 2-1. In time-of-flight mass spectrometry, ions are formed in the ion source, accelerated to mass-dependent velocities, separated and then detected ............................................................................................................ 22 Figure 2-2. The spatial location of an ion in the source and its velocity in the ion source prior to extraction cause broadening of the arrival time distribution for an isomass ion packet ............................................................................................ 24 Figure 2-3. The MTOF/ITR system for GC/MS analysis ................................... 26 Figure 2-4. (A) Conceptual diagram of a \MIey-McLaren time-of-flight mass spectrometer including the two-field ion source and the associated energy diagram showing the various voltages felt by the ions ...................................... 28 Figure 2-5. Diagram of the Wollnik electron impact ion source. Thermal electrons directed into the storage area by electron pushers (Pu) create a potential well between grids 62 and GS. .......................................................... 29 Figure 2-6. Conceptual diagram of the integrating transient recorder. Transient waveforms are captured, summed into mass spectra, converted to massfintensity pairs and stored. ............................................................................................... 32 Figure 3-1. The RTIC (solid line) does not always reveal the presence of coeluting compounds. In this case, three unresolved compounds (dotted lines) contribute to the RTIC but cannot be located by examination of the RTIC alone. .......................................................................................................................... 52 Figure 3-2. Illustration of a mass chromatogram peak position plot (MCPPP) showing ideal (A) and real (B) MCPPP plots for the three unresolved compounds from Figure 3-1. ................................................................................................ 54 Figure 3-3. Selection of the most unique mass for three coeluting compounds by inspection of mass chromatogram peak morphology (A) and by the three-point algorithm (B). .................................................................................................... 56 Figure 3-4. RTIC (A) and corresponding MCPPP (B) from the analysis of the 12- component hydrocarbon mixture. ...................................................................... 62 Figure 3-5. Unique-mass chromatograms identified for peaks 2—9 of the hydrocarbon mixture. ........................................................................................ 64 Figure 3-6. Expanded view of peak 6 from the hydrocarbon mixture showing the RTIC and unique-mass chromatograms (A). Raw mass spectra #185 (B) and #187 (C) correspond to the apices of the unique-mass chromatograms. ......... 65 Figure 3-7. Deconvolved spectra for benzene (A) and cyclohexane (B) obtained from peak 6 of the hydrocarbon mixture. Reference spectra are shown for comparison (C, D). ............................................................................................ 66 xvi Figure 3-8. Unique-mass chromatograms for the dynamic-range study ........... 68 Figure 3-9. Mass spectra of tert-butylbenzene and 1,2,4-trimethylbenzene obtained from pure samples (A,B) and from deconvolution of binary mixtures having concentration ratios of 1:5 (C,D), 1:1 (E,F), and 50:1 (G,H) .................. 69 Figure 4-1. Reconstructed total-ion current chromatogram (RTIC) of a 61- component mixture acquired in 30 minutes on at a rate of ten spectra per second. Two hundred picograms of each component were separated on a 25-m length of 0.2-mm l. D. Ultra II fused silica column having a 0.33-um film thickness. .......................................................................................................... 80 Figure 4-2. Reconstructed total-ion current chromatogram of the same mixture acquired in 80 seconds collecting 30 spectra per second. A 3-m length of 0.1 mm I. D. DB-5 column with 0.4-um film thickness was used. ............................ 82 Figure 4-3. Fourteen seconds of the RTIC in Figure 2 with the times where individual mass chromatograms maximized indicated by the mass chromatogram peak position plot (MCPPP). The 24 compounds present in this region are indicated by their elution order number from Table 4-1 with an arrow to indicate their retention time. When two components exactly coelute, a (2) is placed by the arrow ........................................................................................................... 83 Figure 4-4. Overlay of the MCPPP on the elution profile for partially chromatographically resolved components. ...................................................... 86 Figure 4-5. An overlay of the MCPPP on the elution profile for two coeluting species—tert-butylbenzene and 1,2,4-trimethylbenzene. ................................. 88 xvii Figure 4-6. Comparison of the deconvoluted spectra for tert-butylbenzene and 1,2,4-trimethylbenzene with their respective library spectra. ............................ 90 Figure 4-7. Overlay of the MCPPP on the elution profile for bromoform, styrene and o-xylene whose mass chromatograms maximize at 24.33, 24.97 and 25.40 seconds respectively ......................................................................................... 91 Figure 4-8. Mass chromatogram of m/z 146 reveals that the chromatographic resolution between the two dichlorobenzene isomers is 0.7. Because these isomers cannot be differentiated based on their mass spectra, they must be chromatographically resolved. .......................................................................... 94 Figure 5-1. Conceptual Diagram of a 2-D GC/MS System. Following an initial GC separation is performed on the first column, a portion of the eluent is isolated and separated on the second GC column. The separated components are detected using a mass spectrometer ................................................................ 98 Figure 5-2. Reconstructed total-ion current chromatogram (RTIC) of the perfume from the GC/TOFMS analysis. Responses containing multiple components are indicated by an asterisk .................................................................................... 106 Figure 5-3. RTIC of region 1 from the GC/TOFMS analysis. The MCPPP is superimposed on the RTIC to show the retention times of the components located in this region. The shapes of characteristic mass chromatograms for each cluster of MCPPP lines confirm the presence of two compounds in this portion of the region .......................................................................................... 107 Figure 5-4. Spectra of the two compounds located and identified in region 1 by both techniques. Spectra A and C were obtained from the 2-D GC/MS while xvlii spectra B and D were extraCted via GC/T OFMS/T AD using mass spectral deconvolution. Library searches of these spectra identified the two compounds as Iinalool and phenyl ethyl alcohol, respectively ............................................. 109 Figure 5-5. Expansion of the RTIC shown in Figure 5-3 focusing on the fluctuation in the baseline near spectrum 8725. The superimposed MCPPP indicates the presence of at least one compound in this portion of region 1. Outlying lines in the MCPPP are the result of noise in the mass chromatograms. .......................................................................................................................... 110 Figure 5-6. Spectra for the compound located in Figure 5-5. The chromatographically resolved spectrum (A), spectrum at the retention time of the compound (B), background-subtracted spectrum (C) and deconvoluted spectrum (D) are shown. B is dominated by phenyl ethyl alcohol and thus gives little information about the compound of interest. Library-searches of the other spectra (A,C and D) all identify the compound as rose oxide ........................... 112 Figure 5-7. RTIC of region 2 from the GC/T OFMS/T AD analysis with the MCPPP superimposed to indicated the retention times of the compounds as determined by the deconvolution algorithm ...................................................... 114 Figure 5-8. Mass chromatograms of m/z values characteristic of each of the three clusters of MCPPP lines. Chromatograms of m/z 147, 161 and 93 characterize the areas near spectra 25725, 25735 and 25767. The arrow points out the shoulder in the elution profile of m/z147 .............................................. 116 Figure 5-9. Mass spectra for the three compounds located in region 2 by GC/T OFMS/T AD include the spectrum at the time of the shoulder in the RTIC xix elution profile (compound 1), the deconvoluted spectrum of the compound that eluted at spectrum 25735 (compound 2) and the deconvoluted spectmm of the compound that eluted at spectrum 25767 (compound 3) .................................. 118 Figure 5-10. RTIC of the 2-D GC/MS analysis of region 2. The five peaks indicate the presence of five compounds, labeled A-E, in this region ............... 120 Figure 5-11. Mass spectra of four of the five compounds located by 2-D GC/MS (Figure 5-10). The intensity of compound B was so low that no representative mass spectrum could be obtained. Because these compounds were proprietary, they were not identified via library searches ..................................................... 121 Figure 6-1. Conceptual diagram of the TOF/TOF instrument ........................... 140 Figure 6-2. Ion source for the TOF/TOF instrument. This source is similar to the Wollnik source with the exception of the use of two opposing filaments .......... 146 Figure 6-3. Graph showing the percent of ions transmitted from the gate to the interaction region as a function of the voltage difference between the two elements in the gate assembly .......................................................................... 149 Figure 7-1. System control for the tandem time-of-fiight mass spectrometer...163 Figure 7-2. Percent of transients used by the LeCroy 9450 for a 5000 point acquisition window at triggering rates between 20 and 500 Hz........... ............. 169 Figure 7-3. The signal produced by the thyratron discharge. The delay time is the time between triggering the excimer and the first response in the thyratron discharge signal ................................................................................................ 172 Figure 8-1. Portion of a bromobenzene mass spectrum acquired with detector placed at the interaction region ......................................................................... 186 XX Figure 8-2. Plot of the relative intensity of m/z 18 as an aperture was closed under normal operating conditions, to study the uniformity of the ion beam and to examine divergence of the ion beam ................................................................ 188 Figure 8-3. Region of bromobenzene mass spectrum showing no depletion of M156, 157, or 158. Laser emission occurred 50 ns before the arrival time of m/z 156 ............................................................................................................. 194 Figure 8-4. Laser triggered to deplete m/z156 ................................................ 195 Figure 8-5. Laser triggered to deplete m/z157 ................................................ 196 Figure 8-6. Laser triggered to deplete m/2158 ................................................ 197 Figure 8-7. Region containing the predicted arrival time of the m/z 77 product ion (from m/z156) at the detector (A) without laser pulse ion packet overlap and (B) with overlap of the laser pulse and selected ion packet .............................. 199 Figure 8-8. Photodissociation spectrum showing the appearance of the m/z 91 product ion and depletion of the m/z 92 parent ion ........................................... 200 Figure 8-9. Percent of the precursor ion remaining from the photodissociation of bromobenzene as a function of the laser trigger delay time ............................. 204 Figure 8-10. Single-shot mass spectrum of bromobenzene showing the arrival times of rn/z 156 and m/z 158. These spectra were acquired with the detector at the end of the ion flight-path ............................................................................. 207 Figure A-1. The system control function panel for the TOF/T OF instrument...230 Figure A-2. Function panel for the name of the file to be acquired .................. 231 Figure A-3. The repetition rate of the square wave trigger signal is entered into this function panel ............................................................................................. 232 xxi Figure A-4. Function panel to enter delay times for the LeCroy 4222 delay generator ........................................................................................................... 233 Figure A-5. Function panel for entry of conditions for the LeCroy 9450 transient recorder ............................................................................................................. 234 Figure A-6. Function panel for screen displayed graph conditions .................. 235 xxii Chapter 1 Introduction 1.1 Introduction Nearly all analytical samples are mixtures containing more than one component. In fact, mixtures obtained from natural sources may contain hundreds or even thousands of compounds. The components of these mixtures may possess structures that are very similar and have concentrations differing by many orders of magnitude. The need to provide qualitative and quantitative information about the composition of such mixtures is one of the greatest challenges faced by analytical chemistry. The most effective techniques for the analysis of such mixtures are multidimensional. Typically, these approaches first separate the components of the mixture and then detect the separated components. Chromatographic techniques provide temporal separation for the components of a mixture. Under ideal circumstances, each component elutes from a chromatographic column by itself. The use of a selective detection technique, such as mass spectrometry (MS) or infrared spectroscopy, can provide identification and/or structural information for each resolved compound. Thus, these detection techniques serve as component characterization tools and also produce data for approaches are gas chromatography/ mass spectrometry (GC/MS) and tandem mass spectrometry (MS/MS). In GC/MS, the components of thermally stable mixtures are vaporized and separated on a chromatographic column. Separation of the mixture components is determined by the relative affinities of the components for the mobile phase (carrier gas) and the stationary phase inside the column. The separated compounds are introduced into the ion source of a mass spectrometer. Ions are often formed using electron ionization. This approach uses an electron beam to ionize the sample molecules. Electron ionization (El) is a hard ionization technique that frequently produces many fragment ions, and, thus provides significant structural information about the analyte compounds. Extensive libraries of electron ionization spectra exist and can be used to identify an unknown compound based on its fragmentation pattern. The identified compound can then be quantified using the information provided by the mass spectral detector. The effectiveness of the mass spectrometer as a detector for gas chromatography in the analysis of mixtures can be described as follows. "The mass spectrometer provides the ultimate gas chromatographic detection system, provided that one avoids contemplation of the issue of whether a mass spectrometer unit is a gas chromatographic detector or a GC is a sampling device for a mass spectrometer.” [1] The success of GC/MS analysis of mixtures is also demonstrated by its widespread usage. Nearly every volatile and thermally stable mixture, and some that are not, has been analyzed by GC/MS. It has been used in the analysis of organic [2] and inorganic [3] compounds. GC/MS has been used in environmental analysis to determine the composition of air [4], water [5], and soil [6] samples. It has been used in areas as diverse as the analysis of foods and beverages [7], biochemical processes [8] and even the analysis of tobacco smoke [9,10]. When GC cannot completely separate components of a mixture, mathematical deconvolution algorithms may be used to resolve these compounds. Mass spectrometers acquire hundreds of data channels during a chromatographic analysis. Deconvolution algorithms take advantage of small temporal differences between the data channels to locate and identify compounds. They can be used to analyze samples for the presence of target compounds via a reverse-library search [11] or to analyze samples whose compositions are completely unknown [12]. Deconvolution of the mass spectral data can greatly reduce the amount of time and effort required to analyze a sample by decreasing the chromatographic separation needed. Tandem mass spectrometry can be performed to either detect individual components in a mixture without prior separation or to provide a second level of structural information about a pure component. In the latter case, the pure component may be isolated from a mixture by chromatographic separation. In direct component detection from complex mixtures, specific components are sequentially isolated using the first mass analyzer and then characterized using the second mass analysis. The most common MS/MS scan mode is product ion analysis. A chosen precursor m/z value is transmitted by the first mass analyzer while rejecting ions of all other m/z values. Dissociation of the selected precursor is then accomplished via any of a number of energizing techniques. After precursor fragmentation, the resulting product ions are separated in the second mass analyzer and detected. Although hard ionization techniques such as El can be used in MS/MS mixture analysis, soft ionization techniques that cause little or no fragmentation of the molecular ion simplify the separation process by reducing the number of interferences. Structural analysis of a pure compound uses the same product ion scan mode to determine the product ions that result from the ionization process. Structures can be most effectively understood for pure compounds when hard ionization techniques are used. Acquisition of the product ions for all possible precursor ions increases the amount of available stmctural information; setting the instrument at particular combinations of precursor and product ion masses enhances the selectivity for detecting compounds separated by GC. A large advantage of MS/MS in the analysis of mixtures lies in the time required to complete an analysis for targeted compounds. These analyses may be accomplished in a few minutes as compared to the tens of minutes required by GC/MS analysis to achieve the same results provided that the sample remains in the source throughout the entire data acquisition. The major limitation of MS/MS analysis is the need for prior MS/MS characterization of each target analyte and the absence of information produced about any other sample components. Despite being limited to target compound analysis for most mixtures, the selectivity and speed of MS/MS make it a powerful tool. Like GC/MS, tandem mass spectrometry has shown its effectiveness through its wide range of applicability. Analyses include aromatic hydrocarbons, chlorocarbons, phenols, amines and carboxylic acids in sludge [13,14], terpenes, esters, diphenylpropanoids and aromatic compounds in nutmeg [15], geoporphyrins in oil [16], and peptide sequencing of tryptic digests [17]. Virtually anything that can be introduced into the ion source of a mass spectrometer has been analyzed using MS/MS techniques. Solid, liquid and gas samples have all been analyzed using MS/MS [18]. When the components in a mixture are not resolved by GC/MS or MS/MS, these two techniques can be combined as GC/MS/MS to increase the selectivity and resolving power of the separation. This approach provides the benefits of both techniques. A significant fraction of the mixture is chromatographically resolved by GC and can be detected using MS. The remaining regions containing compounds that elute from the GC column simultaneously are analyzed using MS/MS techniques. This approach is not needed as often as GC/MS, but has found applicability in the analysis of environmental [19], and food [20] samples. The resolving power and selectivity available through GC/MS/MS are best used in the analysis of complex mixtures in complex matrices. 1 .2 The Problem With Conventional Mass Spectrometry Like all multidimensional techniques, GC/MS is dependent on the qualities of the gas chromatograph and the mass spectrometer and their compatibility. Advancements made in either GC or MS have a tremendous impact on the combined technique. The development and commercialization of capillary GC columns in the early 19805 signaled the beginning of a new era in chromatography. Peak widths were reduced from nearly 20 seconds to only a few seconds and the amount of material introduced into the mass spectrometer was reduced to picograms. This advance in the separation technology placed more stringent requirements on the mass spectrometer detector in both mass spectral generation rates and sensitivity. Most commercially available mass spectrometers are scanning instruments. These instruments obtain a mass spectrum by sequentially acquiring information about each m/z value in the range of interest. When the scan rate of the mass spectrometer is increased to provide an adequate number of data points across a chromatographic peak, less time is spent in acquiring each m value and the sensitivity is reduced. Thus, a trade-off between mass spectral acquisition rate and sensitivity exists for scanning mass spectrometers. Most GC/MS analyses are presently performed acquiring only a few scans per second. These scan rates provide the sensitivity required by capillary GC. Accurate description of the shape of the elution profile across a chromatographic peak requires at least 40 data points [21]. When the scanning mass spectrometer acquires only four to six data points across a chromatographic peak, as is the case when acquiring two scans per second across a 2-3 5 peak, two types of distortions occur in the data. The first of these is shown in Figure 1-1. Chromatographic resolution is lost when the mass intensity / . (A) Actual Chromatogram cm" 2 ' '4' '6' r8' 'ibii‘lz Time, seconds (8) Reconstructed Chromatogram 2: 75 c 9 E 0 2 4 6 8 1 0 12 Time, seconds Figure 1-1. The acquisition of only a few mass spectra per second can severely distort the shape of the chromatographic elution profile. Determining the number of peaks in a chromatogram is difficult under these conditions. spectral acquisition rate is decreased to 1 or 2 Hz. The actual reconstructed total ion current chromatogram (RTIC) of an analysis is shown in Figure 1-1A while the RTIC obtained at 1 scan per second is shown in Figure 1-1 B. The slow mass spectral acquisition rate not only distorts the shape of the elution profile, compounds present in the true RTIC cannot even be detected in the acquired data. Sequential acquisition of the information about each m/z value causes the second of these distortions. Under ideal circumstances, the partial pressure of the analyte in the ion source should remain constant while the entire mass spectrum of that compound is obtained. Unfortunately, the partial pressure of an analyte eluting from a capillary GC column changes radically over the time required to perform one scan. Thus, the intensities of the m/z values are skewed by the slow scan rate. This type of distortion is illustrated in Figure 1-2. The first spectrum (I) shows the real spectrum of hypothetical compound X. The remaining spectra, II, III and IV, show the effects of skewing on a mass spectrum. Spectrum II is acquired as the concentration of the analyte was increasing in the ion source. Spectrum III was obtained across the top of the peak and Spectrum IV resulted from a decrease in the concentration of Compound X in the ion source. Since the intensities of a m/z value is an important tool in the interpretation of a mass spectrum, skewing can effect compound identification. The low mass spectral acquisition rates and their accompanying problems can also severely limit the use of deconvolution techniques. Although skewing can be eliminated, most algorithms require much higher data sampling rates to be effective. The resolution of these algorithms is often described as 1.5 or 2 Intensity .é‘ a: c 2 E Time—> (I) Study SM- (111) Sean 2 I I I I >‘ I i I H e . . . . '6') I , . - . 1 nvz-D 8 Wk) (ll)Scan1 E (IV)S¢III3 ' . m/r’ . Figure 1-2. Skewing results from rapid changes in the partial pressure of the analyte in the ion source relative to the scan rate. After a scan has been acquired, all m values plotted in a mass spectrum are assigned to the same time. 10 data points. When only a few points are acquired over a chromatographic peak, these approaches are virtually useless. 1.3 The Michigan State University Solution A time-of-flight mass spectrometric system with mass spectral acquisition rates and sensitivities required for use as a detector for capillary G0 has been developed [22]. The combination of time-of-flight mass spectrometry (T OFMS) and time-array detection (TAD) permits the acquisition of 50 or more mass spectra per second and has femtogram detection limits. Capillary GC elution profiles can be accurately reconstructed on all data channels with this mass spectral detector. The mass spectra produced by this instrument are unskewed since all of the ions in the ion source are sampled at once. These qualities result in data that are well-suited for deconvolution techniques. 1.4 Research Goals The goals of the research reported in this thesis were: (1) to exploit the capabilities afforded by TOFMS with TAD by devising a simple deconvolution approach and applying it to capillary GC/MS data and (2) to develop a major portion of an MS/MS insthment based on the TOFMS/T AD technology. Much of the research involved in achieving these goals was collaborative in nature. This thesis will focus on my contributions to each of these goals. 11 The remainder of this thesis is organized in the following manner. The background and some of the theoretical considerations that led to the Michigan State University TOFMS/T AD system and provides historical information about the role of deconvolution approaches for GC/MS data are described in Chapter 2. Fully using the information available from GC/MS with the MTOF/ITR detection system is the focus of Chapters 35 Description of a deconvolution system developed by a group at Michigan State University is the focus of Chapter 3. Time-compressed chromatography where the analysis time required for mixtures can be reduced by a factor of ten or more over the traditional GC/MS approach is described in Chapter 4. Large reductions in analysis time are achieved without sacrificing any analytical information. The deconvolution approach is compared to two-dimensional GC/MS in Chapter 5. Chapters 6 and 7 are based on the development of the TOF/T OF instrument. Pertinent background and an overview of the tandem time-of-flight MS/MS instrument are provided in Chapter 6. The development of the hardware and software necessary to supply the critical timing in the instrument and considerations about the interaction region are the focus of Chapter 7. Experimental results pertinent to my specific research goals in developing this MS/MS instrument are also included in this chapter. References 1 . Bu nting; Thomas in Chromatographic Analysis Of The Environment, Second Edition Revised and Reexamined, R. L. Grob, Ed., Marcel Dekker: New York, 1983, pg. 52. 10. 12 McLafferty, F. W. In Interpretation of Mass Spectra, Third Edition, University Science Books: Mill Valley, CA, 1980. Schmidt, G. In Chromatographic Methods In Inorganic Analysis, W. Bertsch, W. G. Jennings and R. E. Kaiser, Eds., Dr. Alfred Huther Verlag: Heidelburg, 1981, Chapter 3. Braman, R. S. In Chromatographic Analysis of the Environment, 2nd ed revised and expanded, R. L. Gross, ed., Marcel Dekker: New York , 1983, Chapter 3. Giuliany, B. E. In Chromatographic Analysis of the Environment, 2nd ed revised and expanded, R. L. Gross, ed., Marcel Dekker: New York , 1983, Chapter 6. Gross. R. L.; Kanatharana, P. In Chromatographic Analysis of the Environment, 2nd ed revised and expanded, R. L. Gross, ed., Marcel Dekker: New York , 1983, Chapter 7. Jennings, W. G. Gas Chromatography With Glass Capillary Columns, Academic Press: New York, 1978, 107. Holland, J. F.; Ledly, J. L.; Sweeley, C. C. J. Chromatogr. 1986, 397, 3. Rapp, U.; Schroder, U.; Meier, 8.; Elmhorst, M. Chromatographia 1975, 8, 474. Jennings, W. G. Gas Chromatography With Glass Capillary Columns, Academic Press: New York, 1978, 107. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 13 Abramson, F. P. Anal. Chem. 1975, 47, 45. Biller, J. E.; Biemann, K. Anal. Left. 1974, 7, 515. Hunt, D. F.; Shabanowitz, J.; Harvey, T. M.; Coates, M. L. Anal. Chem. 1985, 57, 525. Hunt, D. F.; Shabanowitz, J.; Giordani, A. B. Anal. Chem. 1980, 52, 386. Davis, D. V.; Cooks, R. G. J. Agric. Food Chem. 1982, 30, 495. Johnson, J. V.; Britton, E. V.; Yost, R. A.; Quirk, J. M. E.; Cuesta, L. L. Anal. Chem. 1986, 58, 1325. Biemann, K.; Martin, S. A. Mass Spectrom. Rev. 1987, 6, 1. Busch, K. L.; Glish G. L.; McLuckey, S. A. In Mass Spectrometry/Mass Spectrometry: Techniques and Applications of Tandem Mass Spechometry, VCH Publishers: New York, 1988. Gale, B. C.; Fulford, J. E.; Thomson, B. A; Ngo, .A.; Tanner, S. D.; Davidson, W. R.; Shushan, B. l. Adv. Mass Spectrom. 1986, 10, 1467. Trehy, M. L.; Yost, R. A.; Dorsey, J. G. Anal. Chem. 1986, 58, 14. Chesler, S. N.; Cram, S. P. Anal. Chem. 1971, 43, 1922. Tecklenburg, R. E., Jr.; McLane, R. D.; Grix, R.; Sweeley, C. 0.; Allison, J.; Watson, J. T.; Holland, J. F.; Enke, C. G.; Gruner, U.; Gotz, H.; Wollnik, H. Proceedings of the 38th ASMS Conference on Mass Spectrometryand Allied Topics, Tucson, AZ, June 3-8, 1990. [5f Chapter 2 Introduction to Time-of-Flight Mass Spectrometry and Deconvolution 2.1 Introduction The development and commercialization of capillary columns for gas chromatography has made mixture analysis a reality for the masses. These high- efficiency columns have been used to analyze nearly every volatile, thermally stable mixture that can be introduced onto the column. Component analysis is achieved by separating all of the mixture components using capillary GC and then detecting the separated components using one or more detectors [1]. Capillary GC has become the technique of choice for separation of mixtures because of its separation efficiency and speed [2]. Capillary GC has the ability to separate mixtures containing hundreds of components at concentrations that vary by several orders of magnitude. Unfortunately, this ability does not always translate into reality. Many mixtures, especially natural mixtures, are composed of hundreds to even thousands of components. These components often have similar chemical structures. The available chromatographic separation space is limited even for high-efficiency capillary columns. The likelihood of two or more of the components coeluting 14 15 from the GC column increases with the number of components in the mixture [3]. In 1983, Davis and Giddings [4] showed that the more-or-less randomly-spaced elution times typically found in a complex chromatogram lead to a high occurrence of overiapping peaks as well as to large regions having no peaks at all. Twenty percent or 40 of 200 components would be expected to coelute in a one hour long capillary GC analysis without accounting for the presence of isomers in a mixture [5]. Bertsch describes a separation performed on the components of tobacco smoke in which 1000 compounds were resolved in a single chromatographic run [6]. The width of most peaks was only a few seconds. Despite these optimized conditions, data from the mass spectrometer detector indicated that the average peak resulted from the elution of two components. Davis and Giddings [7] estimated the probability that a single peak contains only one compound is less than 50% for a chromatogram filled to only 35% of the peak capacity. Analysis of such complex mixtures requires more than high-resolution chromatography, it requires a detection system capable of discriminating compounds that elute from the chromatographic column simultaneously. The detector must sample the data rapidly to accurately reconstruct the elution profile information and also must be sensitive due to the low sample capacity of capillary columns. 1 6 2.2 Detectors for Capillary Gas Chromatography Detection of compounds separated by capillary GC may be performed using either single or multichannel detectors. Single channel detectors may have narrow ranges of selectivity or may respond to nearly every compound. Each peak produced by a single channel detector is assumed to result from a single component unless the shape of the elution profile indicates the presence of more than one component. When the chromatographic eluent is introduced to a multichannel detector or a combination of independent single-channel detectors, the information about a peak is increased tremendously. The data across the channels can be compared to determine whether a peak is from a single component or a series of compounds that are not separated by the chromatographic column. This information can be a very powerful tool for understanding the composition of a mixture. Under ideal circumstances, a chromatographic detector would be a multichannel device that responded well to all compounds. It would have the ability to be selective by examining the information on a single or few channels or general by examining the information available across all the data channels. Compounds could then be identified using the combination of retention indices and multichannel data. 17 2.2.1 Single Channel Detectors Detectors commonly used for capillary GC include the flame ionization detector (FID) [8], thermal conductivity detector (T CD) [9], electron capture detector (ECD) [10], nitrogen-phosphorous detector (NPD) [11], and mass spectrometers. While the FID and TCD respond to a wide variety of compounds, the ECD and NPD are more selective. Compound identification for single channel detectors is performed based on retention indices and knowledge of the selectivity of the detector. For instance, the response for halogenated compounds such as chloroform may be intense on the ECD while the FID response may be small. The opposite is true when an alkane elutes from the column. The ECD will not respond while the FID will produce a signal. Coupling these different selectivities serves as a simple version of a multichannel detector. 2.2.2 Multichannel Detectors Most multichannel detectors used with capillary GC are either spectroscopic or mass spectrometric. These detectors can be operated in either of two modes. Multichannel detectors can be general and use the information on all data channels or they can be selective and use only a portion of the available information. Infrared detectors based on Fourier transform technology are commercially available for use with capillary GC. These detectors have sensitivities and acquisition rates that are compatible for use with modern 18 capillary columns [12]. Ultraviolet and visible detectors have found great use in high-performance liquid chromatography (HPLC). These instruments can acquire ultraviolet and visible spectra of compounds as they elute from a HPLC column. Both of these spectrosc0pic approaches are limited by the interdependence of the information across the available data channels resulting from the relatively broad liquid-phase spectral bands for these optical techniques. These techniques are useful but optical speCtra lack the distinctiveness of mass spectra. Data from each channel in a mass spectrometer are unique and have no overlap. Mass spectrometers acquire data about all of the ions formed from a sample in the ion source in the full scan mode. This mass spectral information can be used to determine the structure and ultimately the identity of a compound eluting from a chromatographic column. The mass spectrometer can also be operated as a selective detector using selected ion monitoring (SIM). In its ultimate form, ions of only one m/zvalue are detected by the mass spectrometer. This technique is often used to detect only a specific class of compounds which may be present in a complex matrix [13]. The added selectivity eliminates many possible chromatographic interferences, but bases identification of detected compounds on retention times alone. This approach has found favor with users of scanning mass spectrometers because sensitivity increases when a mass filter is set to monitor ions of only one mass. The limitations of scanning mass spectrometers have been documented in Chapter 1 of this thesis. 19 Array detectors can provide rapid data acquisition across all channels with high sensitivity. Consequently, they have found use as detectors for chromatography. Spectroscopic and mass spectrometric array detectors have been developed and used for these purposes. Diode array detectors are used to generate absorption spectra as a compound elutes from a HPLC column [14]. Unfortunately, these HPLC detectors have the same limitations as their sequential counterparts; the relatively broad spectral absorption bands of compounds in the liquid phase. 2.2.3 Mass Spectral Array Detectors Although several types of array detection have been used in mass spectrometry, two have been successfully interfaced to capillary GC. These are spatial-array and time-array detectors. Both of these two approaches can acquire data rapidly and with high sensitivity. Spatial-array detectors are electro-optical devices developed for use with magnetic sector mass spectrometers [15,16]. They take advantage of the fact that a magnetic field spatially disperses ions based on their mass. When a multichannel plate electron multiplier is connected to a photoplate and a diode array detector by optical fibers, the spatial distribution of ions can be captured. This distribution is then converted into mass and intensity information. Limitations in dynamic range, mass spectral resolution and mass range restrict the use of these detectors to low resolution analysis. From a more pragmatic 20 view, these detectors are expensive. Magnetic sector instruments are among the most expensive types of mass spectrometer and the addition of a detector that costs more than some mass spectrometers makes the price of such a system prohibitive for many potential applications. Time-array detection (TAD) [1] is used in conjunction with time-ot-fiight mass spectrometry. Ions are pulsed out of the ion source and separated temporally in TOFMS. Ions of all masses strike the same surface but at different times. Arrival time and intensity data are captured. For each ion source pulse, these data are readily converted to the traditional mass spectrum format. Since ions from the same source extraction pulse are temporally separated, this approach can be totally continuous. Information can be captured in one part of the array and emptied out of another without affecting the data quality or acquisition rate. This is not true for a spatial-array detector. Data acquisition should be terminated while the information in the array is emptied to avoid introduction of data intensity distortion. 2.3 Introduction to TOFMS for Chromatographic Detection The use of a time-of-flight mass spectrometer as a chromatographic detector is nearly as old as the idea of GC/MS itself. The first GC/MS system, built by Gohlke in 1959, used a time-of-fiight mass spectrometer as a detector for a packed-column gas chromatography column [17]. In 1963, TOFMS was used to detect the effluent from a capillary GC column [18]. Capillary GC 21 columns did not become commercially available until 15 years after this work by McFadden. The major advantage of time-of-flight mass spectrometers is their ability to produce thousands of mass spectra per second using relatively simple instrumentation over a theoretically unlimited mass range. Unfortunately, early TOFMS instruments suffered from poor resolution and inefficient sample use for many years. Other types of mass spectrometers with higher resolving power came to dominate the area of packed-column GC/MS in the 1960s. This trend has continued into the present despite the problems discussed in Chapter 1 of this thesis. Time-of-flight mass spectrometry is a fairly simple process. Ions are formed in the ion source of a mass spectrometer. They are extracted from the ion source, accelerated to a selected energy, and allowed to travel through an evacuated, field-free flight tube. The ions separate temporally into isomass ion packets based on their mass-dependent velocities. The time required for the ions to traverse the length of the flight tube IS measured and used determine the mass-to-charge ratio of an ion. The signal that results from all the ions of all m/z values extracted from the ion source in a single pulse striking the detector is called a transient. Each transient contains complete mass spectral information. This process is illustrated in Figure 2-1. The extraction rate of the ion source is only limited by the flight time of the largest m/z value. As soon as all of the ions produced in one source extraction pulse have reached the detector, another extraction pulse can be applied. Extraction rates of 5000 Hz or more are often used in TOFMS instruments with 4-m flight paths. Ion Acceleration Flight Tube Source Grid .1, Detector Separated isomass ion packets Transient Intensity m/z Figure 2-1. In time-of-fiight mass spectrometry, ions are formed in the ion source, accelerated to mass-dependent velocities, separated and then detected. 23 2.3.1 Limitations of TOFMS Under ideal circumstances, all ions of the same mass will arrive at the detector at the same time. When ions are formed from a sample introduced into the ion source as a gas or liquid, two types of effects increase the width of an isomass ion packet. These can be categorized as spatial and energy effects. Possible spatial and energy considerations affecting the arrival time distribution of an isomass ion packet at the detector are illustrated in Figure 2-2. Spatial effects are related to the position of an ion in the source when the extraction occurs. An ion may also have an initial thermal energy in any direction when it is formed in the ion source. The arrows on the ions depicted in Figure 2-2 indicate that a range of initial velocities is present in the source. Ions with an initial velocity toward or away from the detector when the contents of the ion source are extracted also contribute to poor resolution. When an ion is moving away from the detector, it must first be turned around by the applied extraction field, further increasing the spread in arrival times. When the sample is desorbed from a flat surface where these effects are minimized, the resolution of time-of-flight mass spectrometers can be increased from about 1000-3000 to 10,000 or more [19]. Time-of-flight mass spectrometry is a pulsed technique. Many early instruments with electron-impact sources only ionized the sample for a short period just prior to the extraction pulse [20]. Transients were detected using boxcar integrators that collected only a small portion of the mass range at a time. Thus, only a small fraction of the sample entering the ion source was actually 24 Ion Source *0 .—> To Detector O—+ fl ._. 41—. Figure 2-2. The spatial location of an ion in the source and its velocity in the ion source prior to extraction cause broadening of the arrival time distribution for an isomass ion packet. 25 permitted to ionize and only a small fraction of the ions present in a transient were captured by the data system. 2.4 The MTOF/ITR System A time-of—flight system has been developed for use as a detector for capillary GC [21]. The advantages of time-array detection have been combined with a high sensitivity mass spectrometer that uses sample efficiently and has unit resolution or better throughout the mass range of interest for capillary GC. This instrument (the MTOF), a modification of an instrument developed at the University of Giessen, uses the combination of a storage source and a non-linear ion mirror to deliver a large number of ions to the detector with high resolution [22]. All of the information generated by each pulse of the ion source is acquired by time-array , detection via the integrating transient recorder (ITR). This TOFMS system is shown in Figure 2-3. The following three sections describe important components of the MTOF instrument that help overcome the sensitivity and resolution problems encountered by earlier TOFMS systems. 2.4.1 Development of the Ion Source Wiley and McLaren [23] were the first investigators to significantly improve the resolution of TOFMS. In 1955, they developed equations to determine the flight times of ions from the source to the detector, along with the mathematical basis for the existence of a space-focus plane in a two-field ion 26 GC Interface Ion Source Detector ':[:IJE 1 II ' I Einzel 1.94" 3 [:1 Steering Plates ITR Grid-free Ion Mirror\* Mass M Chromatograms 55:3,3 Figure 2-3. The MTOF/ITR system for GC/MS analysis. 27 source. Space-focusing is based on the fact that an ion closer to the front of the source is accelerated less than an ion near the back of the source. Isomass ions from the back of the source will catch up to those from the front at the space- focus plane. A conceptual diagram of an instrument containing a two-field ion source and the space focus plane is shown in Figure 2-4A. Ions are extracted from thistwo-field source by raising the potential on the grid at the rear of the source. The location of the space-focus plane is the same for all masses and is dependent only on the energy ratio for the fixed dimensions of the ion source. In 1963, Studier built a continuous electron impact ionization (El) source for time-of-flight mass spectrometers [24]. This source formed ions continuously between three grids. The ions were then stored in a potential well between the three grids. This potential well is created by the potential applied to the center of the grids in the ion source. The ions were extracted by applying a shaping waveform to the grid furthest from the detector. The ions then fall down hill out of the source and into the mass analyzer. The idea of an El source capable of storing ions lay dormant until Wollnik and coworkers resurrected it in 1989 [25]. Their cylindrically symmetrical ion source is shown in Figure 2-5. Ions are formed and stored by a continuous electron beam between two grids (G2 and G3). Electron pushers (Pu) help to direct electrons generated by the filament through the slit between G2 and G3. Thermal electrons present between 62 and G3 aid in the accumulation of ions by creating a potential well [26]. This source accumulates a significant portion of 28 (A) Space-focus Plane Detector Ion Source Flight Tube Ionization Region Acceleration Region I B) Acceleration Region Space-focus Plane Ionizatiolr: Region \ Detector | l | I | I Energy Time Figure 2-4. (A) Conceptual diagram of a Wiley-McLaren time-of-flight mass spectrometer including the two-field ion source and the associated energy diagram showing the various voltages felt by the ions. 29 Filament G4 G3 (32 G1 Pu Figure 2-5. Diagram of the Wollnik electron impact ion source. Thermal electrons directed into the storage area by electron pushers (Pu) create a potential well between grids 62 and G3. the ions formed between extraction pulses although the storage efficiency is somewhat mass and pressure dependent. Extraction is achieved by raising the potential on G3 so that the thermal energies of ions in the source are negligible when compared to the extraction voltage. 2.4.2 Ion Mirrors An ion mirror may be used to provide either spatial or energy focusing. This mirror simply consists of a series of ”washers” that create a field that causes an ion to be slowed, stopped and then reflected out of the mirror. An ion travels at the same velocity when it exits the mirror as it did upon entering it. The mirror is used to reflect the distribution of ion positions at the space-focus plane onto the detector. A mirror may have linear or non-linear electric fields. While linear-field mirrors can only correct for first order effects, mirrors with non-linear fields can provide focusing for higher-order energy effects and thus improve resolution. The resolving power of energy-focusing first and second-order mirrors has been mathematically described by loanoviciu et al. [27]. The use of mirrors in TOFMS was begun by Mamyrin and others in the 1970s [28,29]. Early experiments using electron impact ionization and a two stage mirror achieved a resolving power of over 3000 for ReBra clusters with m/z 1266-1290. Mamyrin et al. [30] also built an instrument with a linear mirror that deflected the ions by 180° and returned them to an annular detector which surrounded the ion source. This instrument provided a resolving power of 1200 .‘l n' 31 for ‘99Hg‘27|+ (m/z 326) with a short flight path. Wollnik and coworkers [31] achieved a resolution of 2000 for N2+ using an instrument similar to the original Mamyrin instmment. Thus, the use of an ion mirror raised resolution into the thousands, making time-of-flight mass spectrometers suitable for use as a chromatographic detector. 2.4.3 TIme-Arrey-detectlon The sensitivity increase from an ion source capable of storing a significant fraction of the ions generated between extraction pulses solves only one of the two problems limiting the applicability of TOFMS. The other problem was that only a small portion of the information in each transient was captured by early data acquisition systems. In a process called time-array detection, an integrating transient recorder (ITR) acquires entire transients as rapidly as they are generated [32]. A conceptual diagram of the ITR used in this research is shown in Figure 2-6. Transients are first converted from analog to digital signals by a 200 MHz Flash converter. Then a selected number (10-1000) of sequential transients are summed in one of the parallel sets of emitter-couple logic (ECL) summers. The summed transients are then moved out of the summing registers and processed into massfintensity pairs. This mass spectral information is then stored. The ECL summers work in parallel; one is summing while the other is being emptied. No data are lost in this manner. The summed transients are called mass spectra. Although each transient contains full mass spectral information, the summing process uses the fact that transients can be generated IIIIII>—>——[ / I r A a (B \ 16kECL 16kECL Hi SPBBd Hi Speed Memory Memory (g J L i/ [Summed Transient Array ] (mass intensity parrs) {Processed into Spectrum ] [Mass Storage & Network ] Figure 2-6. Conceptual diagram of the integrating transient recorder. Transient waveforms are captured, summed into mass spectra, converted to mass/Intensity pairs and stored. .1. an (n I) “D (I 33 at rates exceeding the required spectrum generation rate to improve the signal- to-noise ratio. The maximum mass spectral acquisition rate is determined by the time required to transfer the massfintensity information over the data bus to mass storage. In our system, 100 or more mass spectra can be acquired per second by the ITR without sacrificing any mass spectral information. 2.4.4 Characteristics of the MTOF Instrument Ions are generated continuously and a portion of them are stored in the source. The contents of the ion source are typically sampled every 200 us. by raising the voltage on G3 (See Figure 2-5). By the application of a high field strength, the spatial dispersion of ions in the ion source is converted into an ion energy dispersion at the space focus plane. The space focus plane of the ion source is only about 10 cm from the ion source—too short a flight-time for temporal separation of different ion masses. To accomplish mass separation, the ions drift through a 1-m flight tube to a grid-less mirror. The mirror provides mass-independent energy focusing with high ion transmission by reflecting the image at the space-focus plane onto a multichannel plate detector. The ITR then captures the information in every transient to produce mass spectra at a user selected rate. This spectral generation rate is chosen to provide the optimal combination of chromatographic resolution and sensitivity. The mirror time-of- flight instrument with time-array detection yields a system with a resolving power of 1500 (50% valley definition) and is capable of detecting 780 fg of bromobenzene. Thus, this instrument is truly capable of performing GC/MS on the chromatographic time scale. 2.5 Deconvolution of GC/MS Data Compounds with overlapping chromatographic elution profiles can often be distinguished using the information available across all data channels. Deconvolution approaches for GC/MS data have been developed to locate and identify coeluting compounds based on differences across the data channels. This deconvolution process can be very effective when the data on each channel are completely independent of the data on any other channel. The presence of many data channels increases the probability of locating differences across the data set. One common feature of all mathematical algorithms is that their effectiveness is determined by the quality of the data on which they operate. Two types of problems occur in GC/MS data when relatively slow scan rates are used: the elution profile is undersampled and the resulting mass spectra are skewed [1]. When the elution profile is undersampled, its shape is not well defined and chromatographic resolution is lost. Many deconvolution approaches can resolve unknown compounds whose retention times differ by two or more sampling intervals. These algorithms have limited utility when there are a limited number of samples (often 3-6) across a three-second wide chromatographic peak. While acquiring these few mass spectra, spectral skewing occurs as the 35 concentration of an analyte in the ion source changes dramatically during the mass scan of the mass spectrometer. The relative intensities of the m/z values in each scan are distorted when the time axis of a GC/MS run is plotted as ”scan number“ rather than the actual time at which information about each m/zvalue is obtained. Problems with data quality, especially for the narrow elution profiles obtained with capillary GC/MS, has resulted in little development or use of deconvolution since its introduction over a decade ago. The MTOF/ITR system produces the quality data required by deconvolution algorithms. Spectra are generated at high rates and without mass scanning. Data from this instrument (T OFMSfI’ AD) allow the power of deconvolution approaches to be realized for capillary GC/MS. 2.5.1 Characteristics of GC/MS Data All deconvolution algorithms for unknown samples make one assumption: the total intensity for any m/z value is assumed to be a linear combination of the responses from its contributors. This may be mathematically expressed as ii=iciEi where l. is the intensity of the response on a given data channel, n is the number of compounds contributing to the response on the selected channel, C. is the concentration of a compound and E. is the efficiency for the compound forming of an ion with a m/z value on the selected channel. This is tme for every m/z 36 value. The only information contained in the GC/MS data is the total response for all compounds of a given m/z value eluting at a chosen time. 2.5.2 Types of Deconvolutlon The deconvolution approach is determined by the amount of information known about the sample and the results desired by the analyst. If the data are examined to determine the presence and concentration of a specific compound, a mass spectmm of that compound is compared to the elution profile in a reverse library-search [33-35]. This process is also called ”target" analysis since the identities, retention times and mass spectra of the desired compounds are known prior to the analysis. Under these conditions, the target compound should be separated from other compounds with very similar retention behavior and mass spectra to prevent detection of interfering species. Thus, target compound analysis is most effective when the analyte and matrix are well characterized. The opposite circumstance arises when the composition of the sample is unknown. The number of compounds present in a chromatographic peak, their retention times, identities and concentrations all need to be determined by the deconvolution approach. The ideal deconvolution approach should not only determine all of this information about coeluting compounds, it should be able resolve coeluting compounds as they elute from the column. Deconvolution of unknown samples has tremendous potential in dramatically increasing the 37 amount of information available from a single GC/MS mn without increasing the analysis time. This area has become the focus of our deconvolution effort. 2.6 Historical Deconvolutlon Approaches for Unknowns The basic steps required to deconvolute coelutions in a sample of unknown composition are: (1) determine the number of components present in each chromatographic peak, (2) determine the retention time of each coeluting compound, (3) identify the m/z values associated with each component and (4) extract a pure spectrum for each compound present. Identification and quantitation can be performed on the sample once the retention time and mass spectrum of a compound have been determined [36]. A single mathematical algorithm cannot accomplish all of these steps. Therefore, two or more algorithms are typically required to deconvolute coeluting compounds. These algorithms may be grouped by function. One class of algorithms are used to determine the number and retention times of coeluting species. Another set is used to extract mass spectra for unresolved compounds. These approaches may be related for some approaches. For example when principal component analysis is used to determine the number of compounds present in a coelution, factor analysis is often used to extract pure mass spectra. 38 2.6.1 Locating Unresolved Compounds Locating and determining the retention times of coeluting compounds is typically the first step in their deconvolution. A variety of approaches have been applied to locate coeluting compounds. Some of the eartiest of these extended deconvolution methods that determined the number of components present in a region by making assumptions about the shape of the elution profile from single- channel data to multichannel mass spectral data (37,38). This approach was able to find the number of coeluting compounds in simulated data but failed when applied to real data. This failure is due to the difficulty in accurately modeling the shapes of real peaks. Chromatographic peaks often contain features which are not easily modeled without any prior knowledge about their composition. Compound structures may affect peak shapes. Polar compounds often tail when analyzed on columns with non-polar stationary phases. Peak widths and shapes depend on the concentrations and amount of time a compound is retained on the chromatographic column. One of the most common approaches used for locating coelutions is based on the observation of Biller and Biemann [39] that the intensities of ions corresponding to any given compound eluting from a chromatographic column will rise and fall synchronously with the partial pressure of the analyte in the ion source [40,41]. All mass chromatograms across the many data channels with the same temporal profiles are related to the elution of the same compound. Thus, examination of the mass chromatograms can reveal the number of compounds 39 eluting in a specific region and their retention times. The only difficulty associated with this approach occurs when coeluting compounds have fragment ions with the same m/z values. Shared m/z values will reach their maximum intensity at some time between the retention times of the coeluting compounds. In this case, shared m values must be discriminated from those which are present only in one compound. Even under best circumstances, the Biller- Biemann approach is accurate to one spectral acquisition in determining the retention time of a compound. When a compound has a retention time between two adjacent mass spectra, mass chromatograms related to only one compound may reach their maximum intensity over two sampling periods. Ghosh and Anderegg [42] used derivatives to more accurately determine the retention times of compounds. Their approach allows interpolation of the true retention time of a compound from the acquired data for each mass chromatogram. Another more computer-intensive approach is to use factor analysis to determine the number of compounds present in a coelution [43-50]. Factor analysis can detect compounds that coelute completely if there is some distinctive feature for each coeluting compound. This includes differences in elution profile shape such as slopes of rising and falling edges. The unfortunate aspect of using factor analysis to determine the number of compounds present in a coelution is the amount of time required to process the quantity of data collected in a GC/MS run. Factor analysis approaches often rely on some assumption to reduce the processing time required to determine the number of components and their retention times. Knorr et al. [51,52] determined the 40 number of coeluting compounds by minimizing a function while assuming that several different numbers of compounds were chromatographically unresolved. The number of compounds present at this minimum was assumed to represent the number of coeluting species. When this approach was applied to real data, the selected functions were not always effective in determining the number of compounds, especially where large differences in relative abundance were present. These iterative approaches are limited by the time required for data analysis to post-run processing. 2.6.1 Extracting Pure Mass Spectra Once the number of compounds present in a coelution has been determined, the next step is to extract a pure mass spectrum for each coeluting compound. The simplest approach for extraction of a pure mass spectrum relies on the presence of one data channel that is characteristic of one of the coeluting species [53-56]. Features of these unique m/zvalues such as retention time and the shape of the elution profile can be assumed to be characteristic of that compound. This information allows the intensities of shared m/z values to be properly assigned across all of the data channels. The spectral channel for the pure compound often has the narrowest response in the elution window of interest [57]. When factor analysis has been used to determine the number of compounds present in a coelution, least-squares [58,59], factor analysis [49,51] 41 or other approaches [48.50.60] have all been successfully applied to simulated and real packed-column GC/MS data. The least squares approach is dependent on a model for the elution peak shape [61]. This model must be fairly complex to permit accurate deconvolution of the overlapping components. The elution profiles of unknown compounds are difficult to portray accurately since the quality of the chromatography is dependent on the character of the analyte. Factor analysis is most easily performed when a characteristic mass can be located for each coeluting compound. Despite the increased analysis time required when no unique m/z value is present, factor analysis is a powerful approach for resolving coeluting compounds. Numerous curve-resolution approaches have been applied with good success [37,48,49,56]. These approaches cannot be performed in real-time but they have tremendous power when a difficult separation is being analyzed. All of the techniques used to extract pure mass spectra for coeluting compounds have a dependence on sampling rates, the degree of chromatographic separation between the unresolved species and the relative concentrations of the compounds. As compounds elute closer together, they become more difficult to discriminate. A wide range of intensities can further complicate the situation by appearing to be noise or part of the major component's spectrum rather than another compound present at low concentration. Rapid sampling provides the best chance for resolving the coeluting species because small differences in retention time are more apparent. This does not always simplify the extraction process to obtain a pure mass 42 spectrum. Thus, an investigator may be able to determine that more than one compound is present in a peak but not be able to resolve them. Unfortunately, this valuable piece of information does not resolve the compounds. The resolution must be accomplished by altering the chromatographic conditions. Despite the success of these approaches, only a few simple versions of peak finding algorithms are commercially available [62]. These Biller-Biemann based algorithms locate regions containing chromatographically unresolved compounds but do not extract pure mass spectra for the coeluting species. Thus, deconvolution has not been commonly used despite its tremendous potential for improving the compound resolution of GC/MS data. 2.7 Summary Advancements in the field of time-of-flight mass spectrometry have resulted in a detection system capable of acquiring full MS information on the capillary GC/MS time scale. The MTOF/ITR system has the sensitivity and resolution needed to acquire MS data from a narrow-bore capillary column. These qualities provide the instrumental basis for the use of deconvolution techniques to determine the composition of unknown mixtures. Deconvolution techniques have been applied to packed-column GC/MS data with some success but lost favor with the commercialization of capillary GC. This difficulty was a result of the trade-off between mass spectral acquisition rate I I‘ and sensitivity for scanning instruments. As a result of these problems, deconvolution techniques have been studied in less stringent environments such as HPLC with diode array detection [63,64] and analyzing HPLC fractions with Raman detection [65]. The capabilities of the MTOF/ITR system now allow the use of deconvolution techniques for data from capillary GC/MS analyses. References 1. Holland, J. F.; Enke, C. G.; Allison, J.; Stults, J. T.; Pinkston, J. D.; Newcomb, B.; Watson, J. T. Anal. Chem. 1983, 55, 997A. 2. Ettre, L. S. Introduction to Open Tubular Columns, Perkin-Elmer Corporation: Nonrvalk, Connecticut, 1979.. I 3. Guiochon, G.; Gonnard, M. F.; Zakaria, M.; Beaver, L. A.; Siouffl, A. M., Chromatographia 1 983, 17, 121. 4. Davis, J. M.; Giddings, J. C. Anal. Chem 1983, 55, 418. 5. Rosenthal, D. Anal. Chem 1982, 54, 63. 6. Bertsch, W. in Recent Advances in Capillary Gas Chromatography, Bertsch, W., Jennings, W.G., Kaiser, R. E., Eds., Dr. Alfred Huthig Verlag: New York, 1981, Chapter 1. 7. Davis,J. M.; Giddings, J. C., Anal. Chem. 1983, 55, 418. 8. McWiIliam, l. G.; Dewar, R. A. Anal. Chem. 1957, 29, 925. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. Ashbury, c. K.; Davies, A. J.; Drinkwater, J. w. Anal. Chem. 1957, 29, 918. Karmen, A. J. Chromatogr. Sci. 1969, 7, 541. Lovelock, J. E. Chemtech 1981, 11, 531. Hewlett PackardCompany, Avondale, PA 19311. Sweeley, C. 0.; Elliot, W. H.; Fries, I. Ryhage, R. Anal. Chem. 1966, 38, 1549. Dessy, R. E.; Reynolds, W. D.; Nunn, W. G.; Titus, C. A.; Moler, G. F. J. Chromatogr. 1 976, 126, 347. Hedfjall, B.; Ryhage, R. Anal. Chem. 1 979, 51, 1687. Hedfjall, B.; Ryhage, R. Anal. Chem. 1981, 53, 1641. Gohlke, R. S. Anal. Chem. 1959, 31, 535. McFadden, W. H.; Teranishi, R.; Black, D. R.; Day, J. C. J. Food Sci. 1963, 28, 316. Walter, K.; Boesl, U.; Schlag, E. W. Int. J. Mass Spectrom. lon Processes 1986, 71, 309. CVC 2000, CVC Product Inc., Rochester, New York. Tecklenburg, R. E., Jr.; McLane, R. D.; Grix, R.; Sweeley, C. 0.; Allison, J.; Watson, J. T.; Holland, J. F.; Enke, C. G.; Gruner, U.; Gotz, H.; 22. 23. 24. 25. 26. 27. 28. 29. 30. 45 Wollnik, H., Presented at the 38th ASMS Conference on Mass Spectrometry and Allied Topics, Tucson, AZ, June 6, 1990. Grix, R.; Gruner, U.; Li, G.; Stroh, H.; Wollnik, H. Int. J. Mass Spectrom. Ion Proc. 1 989, 93, 323. \Mley, W. C.; McLaren, l. H. Rev. Sci. Instrum. 1955, 36, 1150. Studier, M. H. Rev. Sci. Instrum. 1963, 34, 1367. Grix, R.; Gruner, U.; Li, G.; Stroh, H.; Wollnik, H. Int. J. Mass Spectrom. Ion Proc. 1989, 93, 323. Yefchak, G. E.; Puzycki, M. A.; Allison, J.; Enke, C. G.; Grix, R.; Holland, J. F.; Li, G.; Wang, Y.; Wollnik, H., Presented at the 38th ASMS Conference on Mass Spectrometry and Allied Topics, Tucson, AZ, June 6, 1990. loanoviciu, D.; Yefchak, G. E.; Enke, C. G. Int. J. Mass Spectrom. Ion Proc.1989, 94, 281. Karataev, V. I.; Mamyrin, B. A.; Shmikk, D. V. Sov. Phys. Tech. Phys. 1972, 37, 45. Mamyrin, B. A.; Karataev, V. I.; Shmikk, D. V.; Zagulin, V. A. Sov. Phys. JET P, 1973, 49, 762. Mamyrin, B. A.; Shmikk, D. V. Sov. Phys. JET P, 1979, 49, 762. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 46 Gohl, W.; Kutscher, R.; Laue, H.; Wollnik, H. Int. J. Mass Spectrom. Ion Phys. 1983, 48, 411. Holland, J. F.; Newcombe, B.; Tecklenburg, R. E., Jr.; Davenport, M.; Allison, J.; Watson, J. T.; Enke, C. G. Rev. Sci. Instrum. 1991, 62, 69. Abramson, F. P. Anal. Chem. 1975, 47, 45-49. McLafferty, F. W.; Hertel, R. H.; Villiwock, R. D. Org. Mass Spectrom. 1974, 9, 690. Blaisdell, B. E.; Sweeley, C. C. Anal. Chim. Acta 1970, 42, 21. Sharaf, M. A.; Kowalski, B. R. Anal. Chem. 1982, 54, 1291. Allen, G. C.; McMeeking, R. F. Anal. Chim. Acta Comp. Tech. Opt. 1978, 103,73. Grushka, E.; Myers, M. N.; Giddings, J. C. Anal. Chem. 1970, 42, 21. Biller, J. E.; Biemann, K. Anal. Left. 1974, 7, 515. Dromey, R. G.; Stefik, M. J.; Rindfieisch, T. C.; Duffield, Anal. Chem. 1976,48, 1368. Colby, B. N. J. Am. Soc. Mass Spectrom. 1992, 3, 558. Ghosh, A.; Anderegg, R. J. Anal. Chem. 1989, 61, 73. Davis, J. E.; Shephard, A.; Stanford, N.; Rogers, L. B. Anal. Chem. 1974, 46, 821. 45. 46. 47. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 47 Justice, J. B., Jr.; Isenhour, T. L. Anal. Chem. 1975, 47, 2286. Ritter, G. L.; Lowry, S. R.; Isenhour, T. L.; Wilkins, C. L. Anal. Chem. 1976, 48, 591. Mallinowski, E. R. Anal. Chem. 1977, 49, 612. Mallinowski, E. R.; Howery, D. G. Factor Analysis in Chemistry, erey:New York 1980. Sharaf, M. A.; Kowalski, B. R. Anal. Chem. 1981, 53, 518. Sharaf, M. A. Anal. Chem. 1986, 58, 3084. Vanslyke, S. J.; Wentzell, P. D. Anal. Chem. 1991, 53, 2512. Knorr, F. J.; Futrell, J. H. Anal. Chem. 1979, 51, 1236. Knorr, F. J.; Thorsheim, H. R.; Harris, J. M. Anal. Chem. 1981, 53, 621. Halket, J. M. J. Chromatog. 1979, 186, 443. Mallinowski, E. R. Anal. Chim. Acta 1 982, 134, 129. Mallinowski, E. R. Anal. Chem. 1984, 56, 778. Wrndig, W; Liebman, S. A.; Wasserman, M. B.; Snyder, A. P. Anal. Chem. 1988, 60, 1503. Hites, R. A.; Biemann, K. Anal. Chem. 1970, 42, 855. Allen, G. C.; McMeeking, R. F. Anal. Chim. Acta 1978, 103, 73. 59. 60. 61. 62. 63. 64. 65. 48 Knorr, F. J.; Thorsheim, J. M.; Harris, J. M. Anal. Chem. 1981, 53, 821. Sharaf, M. A.; Kowalski, B. R. Anal. Chem. 1982, 54, 1291. Osten, D. W.; Kowalski, B. R. Anal. Chem. 1984, 56, 991. Frnnegan MAT SanJose, CA 95134. Lindberg, W.; Ohman, J.; Wolde, S. Anal. Chem. 1986, 58, 299. Genitsen, M. J. P.; Tanis, H.; Vandeginste, B. G. M.; Kateman, G. Anal. Chem. 1992, 64, 2042. Mallinowski, E. R.; Cox, R. A.; Haldna, U. L. Anal. Chem. 1984, 56, 778. Chapter 3 Deconvolution of GC/MS Data: Better Data Make An Old Technique Work 3.1 Introduction The difficulties caused by overlapping chromatographic elution profiles in which two or more distinct chemical components are present, are well known. Analyses in such situations are improved to some degree by the use of two- dimensional detectors such as mass spectrometers. These detectors reduce the temporal separation needed for the identification of components with distinctive response patterns among the detector channels. Although complete component separation is not achieved for all samples, deconvolution techniques combined with two-dimensional detection can greatly enhance the “effective” resolution of chromatographic systems. Ideally this process could be performed on GC/MS data in real time. When the mass spectral acquisition rate and data quality are sufficient, even a simple approach will allow the deconvolution of data acquired by capillary GC/MS. The problems with commercial GC/MS systems as chromatographic detectors and our solution have been described in Chapters 1 and 2 of this thesis. The MTOF/ITR system overcomes all of the previous 49 50 sampling limitations that restricted the use of deconvolution techniques in the analysis of capillary GC/MS data. A relatively simple deconvolution approach that will locate and extract pure mass spectra for compounds in unknowns was developed to demonstrate the capabilities of deconvolution in GC/MS analysis when adequate mass spectral acquisition rates are used. This approach was applied to two different mixtures to demonstrate its power. The deconvolution approach described in this chapter was developed in conjunction with Dr. George Yefchak. Our approach to peak-finding culminated from a series of discussions. George developed the cross-correlation algorithm to extract the pure mass spectra and wrote all the software necessary to apply our deconvolution approach to real data. 3.2 Approach The goals of any deconvolution technique applied to unknown samples are to determine the number of unresolved components and their mass spectra. Mass spectrometers are often viewed as detectors for GC/MS in the same manner as flame ionization detectors or other one dimensional detectors. They are used to detect and identify the ”separated” components as they elute from a GC column. Unseparated compounds are usually visible as shoulders or anomalous shapes in the elution profile. The hypothetical elution profiles for three poorly-resolved compounds shown in Figure 3-1 illustrate the weakness of this approach. The reconstructed total ion current chromatogram (RTIC) is the sum of the intensities I. NU 4V:- 3. 51 for all of the data channels at any time. Nothing in the shape of the RTIC reveals that more than two compounds elute in this profile. The three dotted lines show the three compounds that contribute to the RTIC to form the overlapped peak. Chromatographically overlapped compounds can often be located and identified using the information on the mass axis. Acquisition of GC/MS data can either be viewed as the generation of one hundred or more mass chromatograms or the acquisition of a series of mass spectra during the chromatographic run. When the information contained in these mass chromatograms is fully used by deconvolution techniques, the three compounds can in this example can easily be located and identified. A variety of approaches have been successfully applied to GC/MS data, but most of them follow the same steps: (1) determine the number of components present in each chromatographic peak, (2) determine the retention time of each eluting compound, (3) identify the m/z values associated with each component and (4) extract the "tme" spectrum for each compound present. A single algorithm cannot accomplish all of these steps. Therefore, combinations of two or more algorithms are typically required to deconvolute coeluting compounds. 3.2.1 Location of Coeluting Compounds As Biller and Biemann observed [1], the intensities of ions corresponding to a given compound eluting from a chromatographic column will rise and fall 52 Time Figure 3-1. The RTIC (solid line) does not always reveal the presence of coeluting compounds. In this case. three unresolved compounds (dotted lines) contribute to the RTIC but cannot be located by examination of the RTIC alone. 53 synchronously as the partial pressure of that compound changes in the ion source of a mass spectrometer. If the intensities of different sets of masses have different temporal profiles, the presence of more than one compound is indicated. If all of the mass chromatograms are examined to determine the points at which they reach their maximum intensities during a peak, a mass chromatogram peak position plot (MCPPP) may be generated. Each mass chromatogram is first smoothed in the manner of Savitzky and Golay [2]. Our peak-finding algorithm then searches the smoothed mass chromatograms for any region containing intensities that nearly-monotonically increase and then decrease in the same manner. The acquisition time corresponding to the point of highest intensity within this region is taken to be the peak position of that m/z value. When this procedure is performed for every m/z value, a MCPPP may be generated to show the accumulated intensity (or some other related quantity) that maximizes during each acquired mass spectrum. This approach will effectively locate the retention time of a compound to within one spectmm generation, but does not interpolate between spectrum generation intervals. Under ideal circumstances (Figure 3-2A), the MCPPP will show a single response for each compound. Two difficulties in generating the MCPPP usually prevent this from being tme. These are illustrated in Figure 3-2B. One is the presence of noise which may shift the time of maximum intensity of individual mass chromatograms away from their true position. This difficulty results in a cluster of lines around the correct retention time. The second is the presence of m/zvalues shared by more than one of the coeluting compounds. The maximum (A) Time (3) Time Figure 3-2. Illustration of a mass chromatogram peak position plot (MCPPP) showing ideal (A) and real (B) MCPPP plots for the three unresolved conpounds from Figure 3—1. 55 intensity of mass chromatograms containing contributions from more than one compound occurs at some time between the retention times of the coeluting compounds. The exact position of this maximum intensity is dependent on the relative concentrations of the coeluting compounds and the relative abundance of the shared m/z value in the mass spectrum of each compound. Thus, mass chromatograms of shared m/z values may yield a series of responses in the MCPPP between the retention times of the pure compounds. 3.2.2 Determination of the Retention Time of Each Coeluting Compound The next step in our deconvolution approach is the identification of individual mass chromatograms having responses due to only one of the compounds with overlapping elution profiles. These corresponding m/z values are termed ”unique masses". The retention times for these unique m/z values are those of the pure compounds. Our simple approach is shown in Figure 33 Consider the task of identifying the ion that is unique for compound II. From inspection of the reconstructed mass chromatograms in Figure 3-3A, mass m is clearly the correct choice; our goal for the unique-mass algorithm is to quickly determine this without visual inspection. Note that in Figure 3-3A the mass chromatograms have been normalized to unit maximum intensity. In Figure 3-3B, the normalized intensity values have been depicted for each mass only at the scans corresponding to each retention time. The intensity of the unique mass for compound ll (”12) is low at the retention times of compound I, but high at that of compound II. To find the unique ion for compound II, the algorithm computes for 56 Figure 3-3. Selection of the most unique mass for three coeluting compounds by inspection of mass chromatogram peak morphology (A) and by the three-point algorithm (B). 57 each mass the quantity h = a + b, where a and b are defined as the intensity changes shown in Figure 3-3B; the mass corresponding to the largest value of h is chosen as most likely to be unique. The value of h is normalized, since otherwise a non-unique but high-intensity mass could yield a higher h score than a unique but low-intensity mass. Very low-intensity mass peaks, however, may be poorly defined and may even be due to background noise (i.e., no eluting component present at all). To avoid these problems, we have found that multiplying the h scores by the logarithm of the raw peak intensity discriminates against low-intensity noise spikes. Representing the raw intensities for a given mass m in the scans prior to, at, and following the scan time of interest by lm,-1, Imp, and lm,+1. respectively, the value of H = h logIm,0 for that mass is therefore calculated as 2! - I - +I Hm: [mgiiiliflm'i 1 '°g ""9 where the function max(-,-,-) yields the largest of the three raw intensities. When determining unique masses for either the first or last component in an overlapped set, intensities of zero are used for all masses in the absent (i.e., preceding or following) scan. Peaks in the mass chromatograms identified as unique masses are confirmed by examination of the mass chromatographic profiles for the tabulated MCPPP results. Once unique masses have been located for all of the unresolved compounds, the number of coeluting species and their retention times are known. 3.2.3 Extraction of a Pure Mass Spectrum for Each Unresolved Compound Once the number of compounds present in a coelution has been determined, a pure mass spectrum is extracted for each coeluting compound. When a unique mass is known for each coeluting compound, intensities of shared m values to be properly assigned by comparison of mass chromatograms. We assume that the spectral patterns observed during the elution of overlapping peaks are linear sums of the pure-component spectra. The shape of the elution profile for each unique mass is taken to represent the true elution profile for the corresponding compound. Thus, the degree to which the temporal profile for the unique mass of a given compound is matched by mass chromatograms for the other masses should correspond, in some way, to the intensity of those masses in the desired spectrum. Cross-correlations between the normalized mass chromatograms for a unique mass with those of other masses can be used to assign intensities to all mass chromatograms associated with the elution of a compound. To extract the pure-component spectra each mass chromatogram is first normalized to unit vector length over the region of interest according to the formula 59 Rm: [w Nm5= where Nm,s is the normalized intensity for mass m at scan 5, Rm; is the raw intensity, and a and b are the first and last scan numbers, respectively, for the desired scan range. Cross-correlations between pairs of normalized mass chromatograms are calculated according the formula b Corr(p,q)=2NpJ.NqJ [=0 Note that, in general, Corr(or, [3) = Corr(B. or) and Corr(or, or) = 1. The cross-correlation function is used to generate a factor which, when multiplied by the observed intensity of a given mass in a scan acquired at one of the elution times, yields an approximation of the true intensity of that mass in the spectmm of the corresponding compound. Since the mass chromatograms for unique masses from two different compounds represent independent elution profiles, the cross-correlation between these mass chromatograms represents the degree of overlap. The raw cross-correlation values will then go from zero, for completely separated peaks to unity for exactly coeluting compounds. We obtain the desired factor by linearly re-mapping the raw values so that the cross- correlation between unique masses for two adjacently-eluting compounds is mapped to zero. That is, we identify the cross-correlation between the two given unique masses as Crow and re-map each value according to the formula 60 Corr( -.-)' Clow I - Clow COfr'( °, 0: Values for these re-mapped cross-correlations are multiplied by the corresponding raw intensities to yield the extracted spectra. Software for the deconvolution algorithms was written in C and operated under Unix System V on a Motorola MVME147-A1 computer (Motorola, Inc.). Initial development of the algorithms was performed using the spreadsheet program Wingz (lntorrnix Software, Inc.) on a Macintosh llsi computer (Apple Computer, Inc.). 3.3 Experimental 3.3.1 Reagents The gasoline-range hydrocarbon mixture was prepared from a ThetaKit TK-102 sample set (Theta Corp.) by adding 5 pL of each component to 5 pl. reagent-grade dodecane in a 7 mL vial. The vial was heated to 60°C, and a 0.2 pL headspace sample was withdrawn for injection into the gas chromatograph. The compounds tert-butylbenzene and 1,2,4-trimethylbenzene were obtained from Chem Service, Inc. and dissolved in reagent-grade methanol. Chromatographic conditions were adjusted provide a large degree of overlap in the elution profiles for these two compounds. ‘.‘. “a. .U 61 3.3.2 Gas Chromatography Chromatography and sample introduction to the mass spectrometer were accomplished with a HP-5890A (Hewlett-Packard, Inc.) gas chromatograph using a 2 m x 100 pm 035 column having a 0.4 pm film thickness (J&W Scientific, Inc.). The helium flow rate was adjusted to obtain optimum peak shape; resulting in a linear velocity of approximately 85 cm/s. The injector and the transfer line to the mass spectrometer were both heated to 200°C. 3.3.3 11me-of-Fllght Mass Spectrometer The MTOF instmment was operated with an electron energy of 70 eV in the ion source and an extraction frequency of 3 kHz. Mass spectra were acquired at a rates of 20 and 30 spectra per second by summing 150 and 100 transients for the respective analyses. 3.4 Effectiveness on Gasoline Range Hydrocarbon Mixture A mixture of twelve gasoline-range hydrocarbons was prepared and partially separated over a total elution time of ten seconds by high-speed gas chromatography. The reconstructed total-ion chromatogram (RTIC), shown in Figure 3-4A, reveals only ten peaks; two coelutions (at peaks 6 and 7) are indicated, however, by the MCPPP shown in Figure 3-4B. Thus the number of components indicated by the MCPPP is correct. The algorithm used to obtain Intensity Relative 62 GA) Eb” A g" a. :00 23*“ :=—-*~I 0 10 o 100 ' 230 300 Spectrum Number CED :- ‘3 .5 .5 g .2 o tho. zoo- soc Spectrum Number Home 34. RTIC (A) lid MCPPP from the W (3) mum-12- 63 this MCPPP identified any region of a mass chromatogram having at least 5 consecutively-increasing points followed by at least 5 consecutively-decreasing points as a peak. One noise spike was tolerated within both the rising and falling windows, and the data were filtered by a 3-point Savitsky-Golay smooth prior to analysis. Unique masses were determined for peaks 2-9; mass chromatograms for these are shown in Figure 3-5. An expanded view of the RTIC for peak 6 is shown in Figure 3-6A, together with unique-mass mass chromatograms for the two overlapping components. The raw spectra obtained at the apices of the two mass chromatograms (at 9.25s and 9.35s, respectively) are shown in Figures 3-6B-C. Since the mass chromatogram maxima are separated by only two spectral acquisitions, or 0.01 5, each of these raw spectra are nearly 1:1 combinations of the actual spectra for the separate components. Spectra extracted for the peaks by deconvolution are shown in Figures 3-7A-B. Even though the elution profiles are almost completely overlapping, the algorithm yields completely acceptable spectra, as shown by comparison with the reference library spectra [3] in Frgures 3-7C-D. Although their retention times are too similar to allow their determination by scanning mass spectrometers, major differences in the mass spectra of benzene and cyclohexane allow these compounds to be easily identified. Extraction of the pure mass spectra for these compounds is simplified by their lack of shared m/z values. This study reveals the capabilities of our ml 57 4 ’ z I l A? r l e I v I/f\%/'\A v 1 ml 84 3 z ‘W ' f/l' ' ' v I r; mlz42 Q2 ' T ' r 120 140 rec too 200 220 240 Spectrum Number Figure 3-5. Unique-mass chromatograms identified for peaks 2—9 of the hydrocarbon mixture. 65 RTE a (a) 3 6 3 5 O .2 E 0 ml: 84 m 30 50 7O 90 ml: (C) 5‘ ml: 78 3 O 2.5 O .2 E O , a: I .- . 170 180 190 200 30 50 7O 90 Spectrum Number m/z (A) PlgunS-O.Expandedviewofpeak8homflrehydrocarbonnixhrreshowhgflte RTIC and unicpe-mass chromatograms (A). Raw mass spectra #185 (B) and #187 (C)correspondtotheapicesof the unique-mass chromatograms. Relative intensity (a) 0 Relative Intensity (A) 50 70 ml: (C) m/z 0 - O (B) 50 70 ml: (D) ml: Figure 3-7. Deconvolved spectra for benzene (A) and cyclohexane (B) obtained frompedteotthehydrocarbonmixture.Reterencespectraareshownfor corrparison (C, D). 67 deconvolution approach for locating unresolved components with distinctive spectral features at nearly equal concentrations. 3.5 Dynamic Range Studies In order to explore the practical limits for deconvolution of minor component spectra from those of major components, a series of binary mixtures having different concentration ratios were analyzed. Mixtures of tert-butylbenzene and 1,2,4-trimethylbenzene were prepared in methanol at relative concentration ratios of 1:5, 1:1, and 50:1 and M through the GC under conditions where the two analytes nearty coelute. The mass chromatograms obtained for masses unique to these two compounds are shown in Figures 3-8A-C. Note that the peak for M 134 has an intensity of only about 20% in the tert-butylbenzene spectrum, but M105 is the base peak of the 1,2,4-trimethylbenzene spectrum. Assuming similar ionization and fragmentation efficiencies, the unique-mass intensity ratios thus range from 1:25 to 10:1 as the concentration ratios range from 1 :5 to 50:1. Spectra for the two compounds were obtained by deconvolution at each of the three concentration ratios. The resulting spectra are shown in Figures 3-9A-H together with spectra obtained from injection of the pure compounds. Acceptable spectra were obtained over the entire concentration- ratio range, except for the loss of the m/z 119 ion in the 50-fold dilution of 1,2,4- trimethylbenzene. (A) > (B) m/z 105 mlz 105 ml: 134 .‘ m’z ‘34 g ' . I .A Figure 3-8. Unique-mass chromatograms for the dynamic-range study. (A) (B) 40 so ,m 120 160 40‘ so W, 120 160 (C) (D) 40 so mar o tic 40 8 mar 0 16b (E) (F) 40 so W, 120 16c 40 so W, 120 160 (G) (H) 40 so W, 120 16c '40 so W, 120 160 Figure 3-9. Mass spectra of tert-butybenzene and 1,2,4~trirnethybenzene obtained from pure samples (A,B) and from deconvolution of binary mbrtures having concemration ratios of 1:5 (C,D), 1:1 (E,F), and 50:1 (G,H). 70 This study was a rigorous test of the deconvolution approach. The algorithms had to not only locate unresolved species and allocate the intensities of many shared anvalues over a significant range of concentrations. The use of real experimental data shows the capabilities of the deconvolution approach in relatively difficult circumstances. 3.6 Summary and Future Work Mass spectral deconvolution of chromatographically unresolved components has been attempted by many researchers following the seminal paper of Biller and Biemann. Despite these studies, however, automated spectral deconvolution in GC/MS has not become routine. The promising results shown here result not from any fundamental breakthrough in deconvolution algorithms but rather by the abundance and fidelity of data provided by TOFMS/T AD. Although this approach does not represent the pinnacle of deconvolution, it shows the power of deconvolution approaches and their potential role in chromatography/MS analysis. Although this deconvolution approach is powerful, it also has several areas that should be improved to make real-time deconvolution a reality. The first of these is the method used for MCPPP generation. The Biller-Biemann approach is effective, but has resolution of :I:1 scan. A derivative approach such as that used by Anderegg [4] can improve the compound resolution to a fraction of a scan. The derivative approach is also more amenable to use with digital .v-. 71 signal processors (DSPs) than the present approach. This would permit generation of the MCPPP in real-time. MCPPP interpretation is the most difficult step in the deconvolution of unknowns. Each MCPPP response must initially be presumed to result from a different compound. A variety of factors can be examined to determine the best means of locating shared and pure m values. These include the peak widths and shapes of mass chromatograms. Another limitation of this approach is the use of the cross-correlations to generate factors which are, in turn, used to generate the mass spectrum of pure compounds. This cross-correlation approach is a simple trick whose mathematical basis is questionable. The success of this mass spectral extraction technique lies partially in the algorithms used for library-searches. The presence of a m/z value in a mass spectrum is considered more important than its intensity by many library-search algorithms [5]. Thus, properly assigning a m value to the mass spectrum of a compound is more important than having the correct intensity. Other mass spectral extraction approaches such as factor analysis should be investigated to improve the accuracy of the ”pure" mass spectra generated by the deconvolution approach. References 1. Biller, J. E.; Biemann, K. Anal. Left. 1974, 7, 515-528. 2 Savitzky, A.; Golay, M. Anal. Chem. 1964, 36, 1627-1639. 3 NIST/EPA/MSDC Mass Spectral Database, PC Version 3.0, National Institute of Standards and Technology. Gaithersburg, Maryland. 72 4. Ghosh, A.; Anderegg, Fl. J. Anal. Chem. 1989, 61, 73. 5. Crawford, L. R.; Morrison, J. D. Anal. Chem. 1968, 40, 1464. Chapter 4 Time-Compressed Gas Chromatography/Mass Spectrometry: Fifty-Two Compounds in Eighty Seconds 4.1 Introduction The basis of high resolution gas chromatography/mass spectrometry is the chromatographic separation of components in a mixture. The speed of analysis is determined by the time required for the optimum chromatographic separation; the mass spectrometer functions simply as a detector. In practice, the ability to accurately identify unknown components, having overlapped elution profiles, by means of deconvolution techniques actually decreases the need for optimum chromatographic separation if many mass spectra are acquired across the elution profile of a compound. As long as a component possesses at least one distinctive data channel in its mass spectrum, it may be identified and accurately quantified. Chromatographic analysis times can often be greatly reduced while still providing enough separation to meet this criterion. We call this approach time-compressed chromatography. Time-compression of GC/MS data is attained by combining ”high-speed chromatography" with high mass spectral acquisition rates. The use of deconvolution techniques requires 73 74 differences on only one of the several hundred available data channels to discern the presence of even an unknown component. The greatest reductions in analysis time are achieved by eliminating excess chromatographic resolution via a shorter length of column and/or faster carrier gas linear velocity. Taking advantage of the vacuum outlet of the column, the chromatographic conditions can be adjusted to provide the best separation per unit of time [1,2]. Because shorter columns produce narrower elution profiles [3], rapid mass spectral acquisition rates are needed to provide a minimum of ten data points across the elution profile of each component. The contribution of inlet variance which becomes greater when operating at high linear velocities can be decreased by using specialized injection techniques to further reduce peak widths [4-7]. For this study, we chose to treat the mixture as an unknown and use no specialized techniques for reducing injection variance. Thus, we separated the sample using low-resolution high-speed chromatography with rapid mass spectral detection. Deconvolution techniques were then employed to resolve compounds that were not separated by the chromatography. Further decreases in analysis time should be attained if the sample is analyzed using target compound analysis and better sample introduction methods. As demonstrated here, analysis times can be reduced by an order of magnitude or more without a reduction in the quantity or quality of information provided even for this worse case scenario. 75 4.2 Experimental 4.2.1 Sample Preparation. A 61-component mixture of volatile organic compounds (Ultra Scientific) was used in all of these analyses. The concentration of each component (see Table 4-1) in the methanol solvent was 200 ug/mL. Although this sample was designed to be used as a standard for a purge-and-trap method (EPA Method 524.2), the sample was injected directly into the gas chromatograph. 4.2.2 Gas Chromatography. A 5890A gas chromatograph (Hewlett-Packard, Inc.) equipped with a 5% phenyl dimethyl silicone capillary column was used for all analyses. Two columns of different physical dimensions were used in these experiments. In either case the column was interfaced directly into the ion source of the mass spectrometer. The injector and transfer line temperatures were always held at 200°C. The split ratio of the gas chromatograph and injection volume were adjusted so that about 100 pg of each component were placed onto the chromatographic column. The 30-minute analysis was performed using a 25 m x 0.2 mm id. Ultra II column with a 0.33-um film (Hewlett-Packard, Inc.). The linear velocity of the helium carrier gas was 26 cm/s. Oven conditions were optimized to 76 Table 4-1. The 61 Compounds in the Test Mixture. 1. Dichlorodifluoromethane 2. Chloromethane 3. Trichlorofluoromethane 4. 1,1-Dichloroethene 5. Vinyl Chloride 6. Brornomethane 7. Chloroethane 8. Dichloromethane 9. 1,1-Dichloroethane 10. sis-1 ,1 -Dichloroethene 11 . trans-1,1-Dichloroethene 12. Bromochloromethane 1 3. 2,2-Dichloropropane 1 4. Chloroforrn 1 5. 1,1 ,1 -Trichloroethane 1 6. 1 ,2-Dichloroethane 1 7. 1 ,1-Dichloropropene 1 8. Carbon Tetrachloride 1 9. Benzene 20. 1.2-Dichloropropane 21 . 1 ,3-Dichloropropane 22. Trichloroethene 23. Dbromomethane 24. Bromodichloromethane 25. cis-1,3~Dichloropropene 26. trans-1,3-Dichloropropene 27. Toluene 28. 1 ,1 ,2-Trichloroethane 29. Ethylbenzene 30. Chlorodibrorrromethane 31 . 1 ,2-Dibromomethane 32. Tetrachloroethene 33. Chlorobenzene 34. 1,1 ,1 ,2-Tetrachloroethane 35. m-Xylene 36. p-Xylene 37. Bromoform 38. Styrene 39. o-Xylene 40. 1,1 ,2,2-Tetrachloroethane 41 . 1 ,2,3-Trichloropropane 42. Isopropylbenzene 43. Bromobenzene 44. 2-Chlorotoluene 45. n-Propylbenzene 46. 4-Chlorotoluene 47. 1,3,5-Trimethylbenzene 48. tert-Butylbenzene 49. 1,2,4-Trimethylbenzene 50. 1,3-Dichlorobenzene 51 . 1 ,2-Dichlorobenzene 52. sec-Butylbenzene 53. 4-Isopropyltoluene 54. 1,4-Dichlorobenzene 55. n-Butylbenzene 56. 1,2-Dibromo-3-Chloropropane 57. 1,2,4-Trichlorobenzene 58. Napthalene 59. 1,3,5-Trichlorobenzene 60. Hexachlorobutadiene 61 . 1,2,3-Trichlorobenzene chromatographically resolve as many of the 61 components as possible using the chosen stationary phase. In this case, the oven temperature was initially held at 10°C for 3 minutes and then programmed to 130°C at 4°C/minute. The 80-second analysis was performed using a 3 m x 0.1 mm id DB-5 column with a film thickness of 0.4 pm (J&W Scientific, Inc.). Oven conditions for these analyses were optimized to provide the maximum chromatographic resolution in the run without increasing the analysis time. The linear velocity of the helium carrier gas was set to 88 cm/s and the oven temperature was ramped from 60°C to 130°C at 50°C/minute. 4.2.3 Mass Spectrometry The time-of-flight mass spectrometer used for this work has been described in Chapter 2 of this thesis. Analyte ions are accumulated in the source during the period between ejection pulses. Every 333 us, the ions are pulsed out of the source. The ions are accelerated into the field-free flight tube, reflected by the mirror, and detected by a dual multichannel plate detector. Trme-array detection is performed using an integrating transient recorder [8]. For this analysis, 300 and 100 consecutive transients were summed to produce 10 and 30 spectra per second for the initial and time-compressed analyses, respectively. 78 4.2.4 Deconvolution Approach The deconvolution algorithm used in this analysis is described in Chapter 3 of this thesis. A mass chromatogram peak position plot (MCPPP) was generated by determining the retention time for each mass chromatogram. The information contained in the MCPPP and individual mass chromatograms was used to determine the number of components present in the region of interest. A characteristic m/z value, shared by no other compound, was used to determine the mass spectrum of each coeluting compound via the cross-correlation approach. 4.3 Results and Discussion We chose to test the concept of time-compressed chromatography and the capability of our instrumentation to perform it with a commonly used commercial test mixture of 61 volatile compounds (Table 4-1). The mixture of volatile organic compounds contains over twenty pairs of isomers with similar or identical mass spectra. Many of the compounds in this mixture, including several sets of these isomers, are difficult to resolve chromatographically. This mixture was analyzed with and without time-compression to determine the minimum attainable analysis time and to determine the limitations imposed by the presence of so many isomers. 79 4.3.1 GC/MS Analysis Optimized for Component Separation The mixture was first analyzed under typical high-resolution GC/MS conditions to provide a basis for comparison. The linear velocity of the carrier gas and the temperature program were adjusted to yield optimum chromatographic separation of the components in the mixture. A data acquisition rate of 10 spectra per second was used to obtain at least ten mass spectra for each eluting component. The results of this 30-minute analysis are shown in the reconstmcted total-ion current (RTIC) chromatogram contained in Figure 4-1. The high data acquisition rate better defines the elution profiles and allows more accuracy in determining component retention times and areas than is attainable using the slower data acquisition rates available on commercial mass spectrometers. Even under these conditions, not all of the components in the mixture are chromatographically resolved. Methanol, present as a solvent in the mixture, prevents the identification of components 1-7 in Table 4-1. Several other components have overlapping elution profiles (compounds 9-10,11-12,18-19, 21-22, 33-34, 35-36 and 38-39 in Table 4-1). Some of these coeluting species may be resolved by unique ion identification aided by spectral subtraction since the spectrum obtained from any point in the elution profile is unskewed (for example, compounds 31 and 32 in Table 4-1). Other coeluting compounds, such as the xylenes (compounds 35, 36 and 39 in Table 4-1), are isomers and Intensity Relative o 5 1o 15 I I 15 20 25 so Time (Minutes) Figure 4-1. Reconstructed total-Ion current chromatogram (RTIC) of a 61- componentmixtureacquiredinao minutesonatarateoftenspectraper second.TwolundredpicogramsoteachcomponeMwereseparatedona25-m length of 0.2-mm I. D. Ultra II fused silica column having a 0.33pm film thickness. 81 possess similar mass spectra. These compounds must be chromatographically resolved to be successfully identified. 4.3.2 Analysis of the Sample by Time-Compressed Chromatography The time required for the chromatographic separation was reduced by using a short length of column and a rapid GC temperature-program. The RTIC chromatogram obtained when the 30-minute run was reduced to 80 seconds is shown in Figure 4-2. Thirty spectra were acquired each second to ensure the acquisition of at least 10 mass spectra while each component elutes from the column. The resolving power for this analysis is comparable to that of a packed- column GC. The nature of the data obtained and the method of analysis employed is illustrated by the 14-second segment of the 80-second run is shown in Figure 4 - 3. The retention times of individual components are determined from the overlaid Mass Chromatogram Maximum Peak Position Plot (MCPPP) [11]. This plot shows the sum of the peak intensities for m/z values that maximize during each acquired spectrum. Since the MCPPP is simply a plot of the retention times of all of the mass chromatograms, each MCPPP line indicates the possible elution of a component. Examination of the MCPPP reveals multiple responses rather than a single response for each eluting compound. These responses result from the effects of noise and the presence of peaks at m/z values that are common to the mass spectra of two or more compounds with overlapping elution Relative Intensity 82 J 1 luiulld LJ 1 l I I o 20 . 40 so ' so Time, seconds Figure 4-2. Reconstructed total-ion current chromatogram of the same mixture acquired in 80 seconds collecting 30 spectra per second. A 3-m length of 0.1 mm l. D. 085 column with 0.4-um film thickness was used. Relative Intensity 10,11 (2 -Figure4~3. Fourteensecondsofthe RTIC in F1gure2withthetimeswtnre maximized indicated by the mass chromatogram peak position plot (MCPPP). The 24 conpounds present in this region are indicated by their elution order water from Table 4-1 with an arrow to indicate their retention time. Where two conpounds exactly coelute, a (2) Is 13 12 14 18,19 (2 20.21 (2) 17 15, 16 (2) 4;;- 25 22 /23 24 456789 Time, seconds #29 26 27 31 32 3° I r I 1011 individual mass chromatograms placedbythearrow. 12 13 14 15 16 17 profiles. Noise effects manifest themselves by splitting the response for a single compound over more than one ”scan" or via small responses that do not correlate with the retention time of any compound. These can be located by examining mass chromatograms at the appropriate retention times. When coeluting compounds contain ions with the same m/z value, the shared m/z values will have a retention time that is between that of the coeluting compounds. The actual retention time of each shared mass is determined by the relative concentrations of the coeluting compounds and the intensity of the response for a given shared m/z value in each compound. Once the correct mass chromatograms have been examined, the MCPPP interpretation was corrected for these artifacts. The actual retention times of the mixture's components were known. Their mass spectra were determined using mass spectral deconvolution strategies. The 24 compounds that elute in the 14- second region shown in Figure 4-3 are numbers 8-32 in Table 4-1. In all, 52 of the 61 components in the mixture were identified. Compounds 1-7 were not observed because they coeluted with the solvent. If the mixture was analyzed using the purge-and-trap technique specified by the EPA protocol, these components would be easily identified since there are significant differences in their mass spectra. Four pairs of compounds, benzene/carbon tetrachloride, 2-chlorotoluene/4-chlorotolue ne, 1 ,2-dichloropropane/1 ,3- dichloropropane and ethylbenzene/xylene eluted simultaneously, which prevented their distinction by direct deconvolution; our approach requires the presence of a unique m/z value for each coeluting compound. The spectra 85 obtained during their elution were those of a mixture of the two compounds. The latter three pairs of coeluting compounds must be chromatographically resolved due to the similarity of their mass spectra. Spectral matching software and principal component analysis can, in some cases, unravel simple mixtures from their mixed spectra [9]. This task is, of course, greatly simplified if the list of compounds expected or sought is limited [targeted analysis). While identifying 52 of the 61 compounds, a number of distinct and interesting situations were encountered. These are discussed separately in the following sections. 4.3.3 Adequate Chromatographic Resolution The first and simplest case occurs when the GC separation chromatographically resolves one component from all others in the sample. This is illustrated in the 4.67-second segment of the 80-second analysis shown in Figure 4-4. Since 1,1,2,2-tetrachloroethane, 1,2,3-trichloropropane, isopropylbenzene and bromobenzene are resolved chromatographically, the data do not require deconvolution. For these compounds, there was no difference between a single spectrum acquired at any point in the elution profile and a spectrum obtained via deconvolution. All spectra were readily recognized when compared with library spectra. The overlaid MCPPP in Figure 4-4 shows the retention time of each component and reveals that there are no other components present in this region. Spurious MCPPP responses (adjacent to the major responses) resulted Intensity Relative A 4A Isopropylbenzene 1,2,3-Trichloropropane 29.33 5 28.26 s \ \ Bromobenzene 30.23 s 1,1 ,2,2-Tetrachlcro- ethane 27.33 s _,_LL .1 .. ‘ I I l I 27 28 29 30 31 Time, Seconds Figure 4-4. Overlay of the MCPPP on the elution profile for partially chromatographically resolved components. 87 from the method used to obtain the peak positions in the individual mass chromatograms, noise in the mass chromatographic data and instrument anomalies. Component peaks that maximize at times between consecutive spectra often result in individual mass chromatograms that maximize at the spectrum either before or after the actual elution maxima. Examination of the appropriate mass chromatograms reveals these effects. An intensity change in the RTIC of the isopropylbenzene response resulting from a data point lacking intensity values for a portion of the mass range was responsible for the most intense, spurious MCPPP response at 29.2 sin Figure 4-4. 4.3.4 Poor Chromatographic Resolution of Compounds with Similar Spectra A second and more complex situation exists with the coelution of two components with similar retention times and mass spectra. The RTIC chromatogram of the 2.3-second region where tert-butylbenzene and 1,2,4- trimethylbenzene elute is shown in Figure 4-5. These two compounds, possessing similar mass spectra, have retention times that differ by only 0.13 seconds. These retention times are so close that nothing in the shape of the RTIC elution profile would indicate that this band represents more than one component. However, each compound has an intense mass spectral peak at an m/z value not shared by the other. The molecular ion of tert-butylbenzene (m/z 134) and the [M—1 51+ fragment of 1,2,4-trimethylbenzene (m/z 105) are unique to the mass spectrum of each compound. Although the MCPPP is complicated by Relative Intensity 1 ,2,4-Tri methylbenzene tert-Butylbenzene 39.23 s 39.37 s I 1 r T I 38 39 40 Time, seconds Figure 4-5. An overlay of the MCPPP on the elution profile for two coeluting species—tert-butybenzene and 1,2,4-trimethybenzene. the many shared m/z values and the effects of noise, the time difference at which the mass chromatograms of m/z 134 and m/2105 maximized was greater than experimental error, suggesting the presence of multiple components with similar mass spectra. This is confirmed by comparing the library spectra in Figure 4-6. The two mass spectra share many similarities. Among these are the presence of many fragments resulting from the aromatic ring and the loss of a methyl group. Although the spectra obtained by deconvolution of the overlapped peaks compare favorably to the library spectra (Figure 4-6), the effects of noise on the deconvolution algorithm are apparent in the loss of several low intensity peaks. Using slightly different chromatographic conditions, this same pair of components was resolved and a pure matchable spectrum of each component was obtained when the retention times of the two compounds were only two spectra apart (0.07 seconds). These two compounds were identified and could be quantitated in the presence of each other with a chromatographic resolution of less than 0.2 as compared to the 0.7 normally required to quantity on the basis of height when using non-selective detectors with GC [10]. 4.3.5 Overlap of More Than Two Compounds. Although the majority of the observed elution bands are composed of one or two components, a few regions contain three components. In the 2.67-second region shown in Figure 4-7, bromoform appears as a slight shoulder on the tert-Butylbenzene DGCOI'IVOIUied Relative Intensity II TIIIF’FII II 40 60 80 100 120 140 m/z Library Relative Intensity deal. 40 60 80 100 120 140 ml: 1 ,2,4-Trimethylbenzene Deconvoluted 40 so so 100 120 140 m/z Library 4O 60 80 100 120 140 m/z Flguu4-6.Conpanscncfflredecorwoutedspecuafortertbtnyberueneand 1,2,4~trimethybenzene with their respective library spectra. Relative Intensity 91 Styrene 24.97 s o-Xylenel 25.40 s Bromofonn 24.27 s ill i II T l r ‘ ' 24 25 26 Time, seconds —[ Figure 4-7. Overlay of the MCPPP on the elution profile for brcmofcrrn, styrene and c-xylene whose mass chromatograms maximize at 24.33, 24.97 and 25.40 seconds respectively. 92 leading edge of the elution profile. Despite the low-intensity response, the MCPPP clearly indicates the presence of a component. The remaining compounds in this region, styrene and o-xylene, are easily differentiated on the basis of their mass spectra. Although the deconvolution is capable of resolving more than three components, three was the most encountered in this analysis. 4.3.6 Quantitation Once a compound has been identified by high resolution GC/MS, it can be quantified in either of two ways. If the compound is chromatographically baseline-resolved from all other compounds, quantitation can be based on the RTIC profile. If the sample is not completely resolved from other species, the mass chromatogram for a unique molecular or fragment ion is typically the basis for quantitation. When time-compressed chromatography is performed, one of the steps in the generation of the MCPPP is the determination of those m values that are not shared among the coelutants. The intensities from the mass chromatograms for all those m/z values unique to a given compound can then be summed to more accurately determine a response related to that compound. Quantitation may then be performed by comparing the unknown response with that of an internal or external standard. 93 4.3.7 Limitations Time-compressed GC/MS possesses most of the same insthmental limitations as normal GC/MS. The maximum mass spectral acquisition rate allowed by the mass spectrometer and data system is a major limiting factor. Deconvolution, when using our cross-correlation approach, requires that the elution profiles of two coeluting species do not maximize at the same spectrum number. Therefore, a greater mass spectral acquisition rate coupled with faster chromatography (narrower elution profiles) can further reduce the required analysis time. In our case, for a mass range of m/z 40—200, (due to current ITR limitations) the greatest acquisition rate available for 80 seconds is 30 spectra per second. Thus, we could resolve closely eluting compounds with retention time differences of 0.07 seconds or more. Another hardware limitation is the maximum rate at which the oven of a gas chromatograph can increase temperature. Although isothermal GC analysis is capable of providing a much faster separation, a complex mixture, particularly one whose components possess a wide range of volatilities, may be separated more efficiently when a temperature program is used. Our approach to deconvolution of mass spectra also places limitations on the attainable speed of analysis. As the degree of similarity in the mass spectra of two compounds increases, the chromatographic resolution also must be increased. The resolution achieved for the separation of two dichlorobenzenes is shown in Figure 4-8. The chromatographic resolution between these two isomers Relative Intensity 94 1 ,3-Dichlorcbenzene 1,2-Dichlorobenzene 41.13 s 42.20 s ' r f 40 41 42 43 Time, seconds Figure 4-8. Mass chromatogram of m/z 146 reveals that the chromatographic resolution between the two dichlorobenzene isomers is 0.7. Because than isomers cannot be differentiated based on their mass spectra, they must be chromatographically resolved. 95 is 0.7—adequate for identification and quantitation on the basis of peak height [20]. In this example time-compression cannot be significantly extended beyond this point without loss of resolving power. 4.4 Conclusions Time-compressed chromatography is a powerful tool for greatly reducing the time required for analysis of complex mixtures. The combination of fast data acquisition of unskewed mass spectra and spectral deconvolution allows for dramatic reductions in analysis time using a normal capillary gas chromatograph. The major limitations on the analysis time reduction factor obtainable are the maximum mass spectral acquisition rate at the required detection limits and the necessary chromatographic resolution of mass spectra that contain no unshared m/zvalues. In most of the above data analysis and discussion, we have assumed that the complex mixture was of completely unknown composition. However, the aim of many analytical procedures is the quantitation of targeted compounds in samples with a generally predictable list of components. In these situations, time-compression could potentially reduce the required analysis times by even larger factors. With an accurate library of mass spectra and approximate retention times of the sought compounds, overlapping components can be resolved by reverse-search methods and even exactly coeluting components quantitated by principal component analysis. 96 References 10. Giddings, J. C. Anal. Chem. 1962, 34, 314-319. Cramers, C.A.; Schutjes, C. P. M.; Leclercq, P. A. J. Chromatogr. 1981, 203, 207-216. Yost, R. A.; Fetterolf, D. D.; Hass, J. R.; Harvan, D. J.; Weston, A. F.; Skotnicki, P. A.; Simon, N. A. Anal. Chem. 1984, 56, 2223-2228. Gaspar, G.; Arpino, P.; Guiochon, G. J. Chrom. Sci. 1977, 15, 256- Wade, R. L.; Cram, S. P. Anal. Chem. 1972, 44, 131. Schutjes, C. P. M.; Cramers, C. A.; Vidal-Madjar, C.; Guiochon, G. in Proc. Fifth Intl. Symposium on on Capillary Chromatography, Rijks, J.; ed., Elsevier: Amsterdam, 1983, 304. Desty, D. H.; Goldup, Swanton, W. T. in Gas Chromaography, Brenner, N.; Gallen, J. E.; Weiss, M. 0.; ads, Academic Press: New York, 1962, 105. Holland, J. F.; Newcombe, B.; Tecklenburg, R. E., Jr.; Davenport, M.; Watson, J. T.; Enke, C. G. Rev. Sci. Instrum. 1991, 62, 69-76. Sharaf, M. A.; Kowalski, B. R. Anal. Chem. 1982, 54, 1291-1296. Snyder, L. R. J. Chrom. Sci. 1972, 10, zoo-212. (7 [A (I I) I) I. ll Chapter 5 Comparison of Two-Dimensional Gas Chromatography/Mass Spectrometry with Deconvolution of Gas Chromatography/ High- Speed Mass Spectrometric Data 5.1. Introduction Resolving the components of complex mixtures is a difficult challenge for high-resolution gas chromatography/mass spectrometry (GC/MS). These mixtures may contain hundreds of components whose concentrations may differ by several orders of magnitude. Since the separation space in a chromatogram is limited, the likelihood of two or more of the components coeluting from the GC column increases with the number of components in the mixture [1]. Davis and Giddings [2] estimated the probability that a single peak contains only one compound is less than 50% for a chromatogram filled to only 35% of the peak capacity. One approach to resolving coeluting components is two-dimensional gas chromatography with mass spectral detection (2-D GC/MS). A conceptual diagram of a 2-D GC/MS instrument is shown in Figure 5-1. In a procedure called heart-cutting, regions of the chromatogram containing coeluting species 97 98 T f Injector Vent Lirrogs er 65 V Oven I Oven 2 Moss Spectrometer Figure 5-1. Conceptual Diagram of a 2-D GC/MS System. Following an initial GC separation performed on the first column, a portion of the eluent is isolated and separated on the second GC column. The separated components are detected using a mass spectrometer. are isolated and then separated on a second chromatographic column with a different selectivity from that of the original column. The separated components are detected using a mass spectrometer. While this approach can effectively resolve many coeluting species, it is relatively time-consuming. Regions that potentially contain chromatographically unresolved components must first be somehow identified. The regions selected (for further separation are, in subsequent runs, sequentially isolated onto the second column. The second column is placed in its own oven and operated under a separate set of conditions. Because the second stage of GC analysis can only be performed on one selected heartcut region at a time, multiple chromatographic runs are generally required to examine all of the selected regions that are suspected of containing coeluting species. Heartcutting relies on a pressure controlled valve which either directs the eluent onto the second GC column or vents it. The process of venting the column eluent is often performed via a detector such as a flame ionization detector. A mass spectrometer or any other type of GC detector can be used to examine the results of the initial GC separation. Column selection is the basis of any chromatographic separation and is especially important in 2-D GC. Two sets of factors determine column choice. The choice of stationary phase is based on the nature of the sample to be analyzed. A non-polar stationary phase is often used for the preliminary separation while the final separation is performed on a more polar column. This 100 combination of selectivities increases the resolving power of the chromatography to a far greater extent than would the coupling of two columns with similar polarities. The second set of considerations relate to the concentrations of components in the mixture to be analyzed. 2-D GC is often used to isolate trace components that coelute with compounds present at higher concentrations. Thus, a thick-film column with its large sample capacity is often selected for the preliminary separation while the final separation is performed on a thin-film column. The first column has a large sample capacity, but less resolving power than a thin film column. The final column is selected to provide the best possible resolving power for the material isolated by the heart-cut. An alternative approach is to analyze the sample using gas chromatography/high-speed mass spectrometry and then to examine the data from the regions that contain coeluting compounds by data deconvolution. Rapid acquisition of mass spectral information across the elution profile allows accurate reconstruction of the elution profile for each m/z value. Mathematical deconvolution algorithms, such as the approach described in Chapter 2 of this thesis, can effectively use the small temporal differences present in the mass chromatograms to determine the number of compounds present in a peak and provide a mass spectrum for each compound in an unknown mixture. Although deconvolution algorithms have been around since the 1960's [3], these algorithms have not been in general use for several reasons. Algorithms used to deconvolute unknown samples function most effectively when 101 the elution profile of each mass chromatogram is known accurately. This requires 20 or more data points across each chromatographic peak. The narrow peaks produced by capillary columns (a few seconds in width) thus require mass spectral acquisition rates of 5 to 25 spectra per second. One system capable of simultaneous acquisition of all masses is the combination of time-of-fiight mass spectrometry with time-array detection (1' OFMS/T AD) which can generate 25 or more mass spectra per second with subpicogram detection limits [3]. Analysis of unknown mixtures by either 2-D GC/MS or deconvolution of GC/high-speed MS data requires the same series of steps. First, regions ' containing chromatographically unresolved compounds must be identified. The number of components present in each coelution is then determined. Frnally, a pure mass spectrum of each unresolved compound is obtained to allow identification of that compound. These capabilities were compared for the two techniques using a test perfume sample whose composition was unknown to the analysts. The sample for this collaborative effort was selected in conjunction with David Pinkston of The Procter and Gamble Company (P&G). Pedro Rodriguez and other collaborators at P&G performed the 2-D GC/MS analysis, while the TOFMS/TAD/Deconvolution analysis was performed by me. 102 5.2 Experimental 5.2.1 Initial GC Analysis The sample was screened by gas chromatography (Hewlett-Packard, Inc., Model 5880A.) on a 30-m methyl silicone column (J8W Scientific). The linear velocity of the helium carrier gas was set to 30 cm, sec" for the 0.32-mm i.d. column which was coated with a 1.0 pm film. One microliter of the neat mixture was injected at a split ratio of ca. 100:1. The oven was initially held at 40°C for two minutes. It was then ramped to 250°C at 5°C min“, and the final temperature was held for 15 minutes. The column effluent was split with half going to a flame ionization detector and half to a sniff-port. The temperature of the injector and detectors was 250°C. 5.2.2 Two-Dimensional GCIIIIIS Two columns, a 30-m x 0.32-mm i.d. Rtx-t column with a 3.0-um film and a 60-m x 0.32-mm i.d. Stabilwax column with a 0.5-um film (Restek) were installed in the first and second ovens of a 2-D gas chromatograph (ES Industries, Siemens SiChromat, Model 2-8). The injector temperature was set to 250°C. The linear velocity of the first column was adjusted to 17 cm sec‘ while the second column had a linear velocity of 38 cm sec‘. Following injection of 1.0 pL of the neat oil at a split ratio of ca. 100:1, the first column was held at 50°C for 4 minutes and then ramped to 250°C at 5°C per minute. When a cut was 103 initiated, the temperature of the second oven was held at 70°C for 2 minutes and then programmed to 220°C at 10°C per minute. The second column was then held at 220°C until all of the components in the cut eluted from the column. The column effluent was split equally between three detectors: a flame ionization detector, a mass-selective detector (Hewlett-Packard, Inc. Model 5790A) and a matrix-isolation FT IR detector (4). The quadrupole mass spectrometer was scanned from M 14-300 at 1.5 scans per second. For the purposes of this paper we considered only the range between m/z 40-200. These conditions provided 4-5 mass spectra over chromatographic peaks that were approximately six seconds wide at the baseline. 5.2.3 GC/TOFMS/TAD wIth Data Deconvolutlon A 25-m x 0.20-mm i.d. HP—5 column with a 0.33-um film (Hewlett-Packard, Inc.) was installed in a gas chromatograph (Hewlett-Packard, Inc., Model 5890A). This instmment was operated under the same conditions that were used in the initial analysis of the sample with the following exceptions. The sample was analyzed twice. The split ratio was adjusted to 30:1 for the first run and 250:1 for the second, while 0.1 uL of the neat oil was injected for each analysis. The time-of-flight mass spectrometer used for this work was the MTOF instrument described in Chapter 2 of this thesis. Ions, continuously formed and stored in the ion source via electron ionization, were extracted every 200 us. Two hundred successive transients were summed by an integrating transient 104 recorder (ITR) [5] to produce 25 spectra per second for this analysis. These conditions resulted in the generation of about 50 spectra across chromatographic peaks that were approximately two seconds wide at the baseline. The deconvolution algorithm used in this analysis is described in Chapter 3 of this thesis. A mass chromatogram peak position plot (MCPPP) was generated by determining the retention time for each mass chromatogram. The information contained in the MCPPP and individual mass chromatograms was used to determine the number of components present in the region of interest. A characteristic m/z value, shared by no other compound, was used to determine the mass spectrum of each coeluting compound via the cross-correlation approach. 5.3 Results and Discussion We chose a perfume sample with which to compare the resolving powers of 2-D GC/MS and GC/T OFMS/T AD with deconvolution. Perfumes are typically prepared by combining natural extracts and distillation fractions. Consequently, they often contain a large number of compounds with similar structures and chromatographic behavior. Many chromatographic regions were expected to contain coeluting compounds when the unknown sample was screened under standard chromatographic conditions. The results of this analysis are shown in the reconstructed total-ion current chromatogram (RTIC) from the 105 GC/TOFMS/TAD data contained in Figure 5-2. Although this chromatogram appears to be relatively simple in comparison to those obtained for many complex mixtures, several of the peaks in the RTIC actually result from the coelution of two or more compounds. Two of these regions were selected for further study based upon the sniff-port data. These regions are indicated in Figure 5-2. Two-dimensional GC/MS and GC/high-speed MS with deconvolution were used to examine each of these regions. 5.3.1 Analysis of Region 1 The first of the two regions to be analyzed was the area eluting at about 13 minutes (first starred peak in Figure 5-2). An expanded view of this region is shown in Figure 5-3, the RTIC from the GC/T OFMS/T AD analysis. The shoulder on the leading edge of the RTIC (near spectrum 8075) indicates that at least two components elute in this region. Because the detector was saturated for some m/z values when using the 30:1 split ratio, these data were obtained using the 250:1 split. The MCPPP is superimposed on the RTIC in this figure to show the retention times of any compounds in this region. The shape of the elution profile and MCPPP lines indicated that two partially resolved components were present in this response. Mass chromatograms of two m/z values characteristic of the clusters of MCPPP responses are also shown in Figure 5-3. The retention times and peak shapes of these two masses account for the shape of the RTIC. Sniff- port analysis of this same region also located two components. Rel. Intensity 106 T . D d- N r I l r I r i i i i i i i 15 18 21 24 27 Time, Minutes Figure 5-2. Reconstructed total-ion currerx chromatogram (RTIC) of the perfume from the GC/TOFMS analysis. Responses containing muk'ple ccrrponents are indicated by an asterisk. 107 j Rel. Intensity RTIC ....,....,....jf...,...er....,...a, 8100 8200 8300 8400 8500 8600 8700 8800 ’ SpectrumNumber Figure 5-3. RTIC of region 1 from the GCI‘I’OFMS analysis. The MCPPP ls supennpcsedonflieRTlCtoshcwflteretentlontimesofthecorrponems heated in this region. The shapes of characteristic mass chromatograms for each cluster of MCPPP lines confirm the presence of two corrpounds In this portion of the region. 108 Because the chromatographic separation of these two components was significant, 2-D GC/MS was not necessary to obtain a clean spectrum of each compound. Although spectra obtained from the GC/MS data are library- searchable, the physical separation provided by 2-D GC/MS yields spectra containing less chemical noise. Spectra from the 2-D GC/MS data are shown in Frgure 5-4 (A and C). Deconvolution of the GC/high-speed MS data was not required for this region. However, it could be employed to produce the same clean spectra that were available from the 2-D GC/MS analysis, as shown in Figure 5-4 (B and D). The deconvolution approach eliminates noise from the spectra, but may also have difficulty in discriminating low intensity response from the background. The only other notable difference between the two sets of spectra is the slight change in relative intensities of some m/z values. These variations in relative intensity were too small to affect their identification. Library searches of the spectra using a NIST database identified, with good agreement, the compounds as Iinalool and phenyl ethyl alcohol, respectively. One of the major challenges for either of these techniques is to determine the presence of minor components in the mixture. The small fluctuation in the baseline near specthm 8725 is the focus of Figure 5-5. The MCPPP has been overlaid on the RTIC to show the retention times of individual mass chromatograms. The small cluster of MCPPP lines at about spectrum 8725 indicated that at least one compound with low intensity was eluting at this time. The two outlying responses at spectrum 8708 and 8769, upon inspection, were assigned to noise, being anomalous jumps in the intensity of an apparently 109 g:- 71 g, 55 93 (A) 9.! £1 _' 121 a 1071116 40 so so 10612 140' 160 180 200 1%) 5 71 c 7 9 (8) 0,1 g‘ ' . 5‘. 121 a] _ 197 I 136 4c 60 so 100 1,39 140 160 180 200 Z‘ Z 'a‘ 91 5‘ (C) E —t‘ 122 o, 65 E 51 .l 27 .. - I 40 6o so 100 1319 140 160 180 200 Z ,3 91 8 (D) 21 .5. -" 1 ric’ 5,11 T77 , 212, , , - - 4o 60 so 100 13g 140 160 180 200 FIgureMSpectraoffltetwcccrrpoundsbcatedandideMifiedinregionf by bothtechniques.SpectraAandeerecbtainedfromth82-DGCIMSthe spectra 8 and D were extracted via GC/TOFMS/TAD using mass spectral deconvolution. Lbrary searchesotthesespectra identified thetwoconpcunds as Iinalool and phenyl ethyl alcohol, respectively. 110 RTIC i 8800 8700 l 8600 £825 ._.8m 8900 Spectrum Number 111 unrelated m/z value. The remaining peaks resulted from a single compound where noise at the low-intensity levels interfered with exact peak position assignment. Thus, only one compound was present in this region. Once a compound has been located, a clean mass spectrum is sought for compound identification. Figure 5-6 shows a series of mass spectra obtained for the compound that elutes at about spectmm 8725 in the GC/high-speed MS analysis. The first spectrum (A) was obtained from the 2-D GC/MS data. The large signal-to-background ratio for the 2-D GC/MS analysis results from the combination of increased sample concentration in the GC/high-speed MS analysis and chromatographic resolution of this compound from all other components (is. reducing the chemical noise). Thus, this spectrum is an accurate spectrum of the compound of interest. The remaining three spectra were all obtained from the GC/high-speed MS data. Spectrum B shows the mass spectrum at the retention time indicated by the MCPPP. The base peak in this spectrum is m 91 which is not even present in the spectrum A. Since this low- intensity compound eluted on the tail of phenyl ethyl alcohol, whose base peak is m/z 91, the spectrum at the retention time of the compound is dominated by the intense background and is not really representative of the compound of interest. Subtraction of an average background from the average signal across the peak yielded spectrum C. This spectrum contains most of the same m/z values as are found in spectrum A. However, the presence of several responses at masses greater than m/z 154 shows that a significant amount of noise remains when this approach is used. The final spectrum (D) was produced using 112 >, .g‘ 69 139 g (A) ”41 g. 55 fi] .~ ...1 - i (:40 60 80 100 120 140 160 180' 200 >‘ Iii/Z u] 91 g, (B) G) H E‘ ‘ 55 69 ll jlflj .Id .. g 104 32,1319 170184 (:40 so so 100 120 140 160 180 200 >‘ W2 ,1 69 34155 (C) 3 5‘ 139 . 8f 3'] “.1 . 99198 .1 I. 1.79114. 1:40 60 80 100 [31% 140 160 180 200 >‘ 41 59 I H 2 139 (D) 8 T 51 7;- - 33- - - - ; 154. $40 60 60 100 120 140 160 180 200 m/z Figure 5-6. Spectre-Icrthe-ccrmcund located it Figure 5-5. The cflcmetcgrmhbalynscNedepecwm(A),spectrunatfltereteMbnfimecfhs ccrrpcund (B). Elma-situated spectrum (C)anddeccnvolrted (maresimaiedcnmetedbyphenylethytelcotnlendflusglveeltb ettcrnrationmmeccnpourtdotflmenry-ceuchesofflfednr spectra(A.CandD)aliderIilytheccnpomdesrceecxide. ‘ 113 the deconvolution algorithm. This spectrum contains every m/z value that has a relative intensity greater than 20% in spectrum A, without the noise observed in spectrum C. The M values with relative intensities less than 20% were not extracted by the cross-correlation algorithm due to their! similarity to the baseline. Library searches of the NIST database for spectra A, C and D all selected rose oxide as the first choice for this compound. 5.3.2 Analysis of Region 2 The second of the two regions selected for detailed study eluted at about 20 minutes into the chromatogram contained in Figure 5-2 (second starred region). The most notable feature of this region was the asymmetric shape of the elution profile as shown in the RTIC of the GC/high-speed MS data in Figure 5- 7. One explanation for the fronting observed in this response is column overload. However, other responses of similar intensity would also be expected to show fronting. The peaks occurring at about 18 minutes and in region 1 are of similar intensity and still symmetrical. Close examination of the RTIC reveals a small shoulder on the leading edge of the elution profile (spectmm 25590). Based on these features, more than one compound was suspected of eluting in this region. GC/TOFMSIT AD with Deconvolutlon Analysis. The data obtained by the GC/T OFMS/T AD analysis are shown in Figure 5-7. The MCPPP is overlaid on the RTIC to indicate the retention times of compounds detected by the deconvolution algorithm. These lines cluster into three groups. They are 114 RTIC J 3) '75 C 93. E . 35' a: 25500 25600 25700 25800 T Spectrum Number T T Compounds 1 2 3 FIgureS-T.RTICofregion21romtheGC/TOFMSITADanaIysiswihthe MCPPP superimposed to indicated the retention times of the corrpounds as determined by the deconvckrtion algorithm. 115 centered at spectra 25725, 25735 and 25767. Three groups of MCPPP lines might be interpreted as indicative of three compounds, but even three compounds eluting at the times suggested by the clusters do not explain the shape of the elution profile. However, if two or more compounds that share all M values (such as isomers) coeluted in region 2, the MCPPP would have responses only at the times when the broad, individual mass chromatograms reach their maximum intensity. When the component eluting last in this region has the largest intensity, the MCPPP lines will indicate the retention time of this compound and not give any retention information about any other isomers. Mass chromatograms representing each of the three clusters are shown in Figure 5-8. We selected m/z147, 161 and 93 to represent the clusters of MCPPP responses at spectra 25725, 25735 and 25767, respectively. (The elution profiles of these mass chromatograms illustrate the presence of more than one compound better than other m/z values even though they have more intense responses.) The elution profiles of m/z 147 and 161 are very similar in shape. They both exhibit the fronting observed in the RTIC of region 2. The shoulder at spectrum 25590 in the RTIC is apparent only in the responses characterized by m/z 147 (indicated by the arrow in Figure 5-7). As expected from the combination of elution profile and lack of a MCPPP response, the intensity of m/z 147 increases until about spectrum 25725. The series of MCPPP responses near spectrum 25735 occurs at the retention time of the most abundant of these compounds, while the series of MCPPP lines around spectrum 25725 probably results from shared m/z values whose relative intensities are larger for a compound other than the most abundant one. Thus, at least two compounds that share all 116 m/2161 Rel. Intensity m/z 147 . . , . . . 25600 25700 25800 Spectrum Number FIguuS-auesectlumbgrmcfmmchamcterlstfccteechctu- three clusters of MCPPP lines. Chromatograms of m 147, 161 and 93 characterize the areas near spectra 25725, 25735 and 25767. The arrow palms outthe shoulderintheelrtion profileof M147. 117 abundant m/z values are present in this region. Additional isomers may be present in this region, but the MCPPP and chromatographic information were not adequate to locate and identify them. Changes in the chromatographic conditions could provide the resolution required for their location and identification. The third set of MCPPP responses are very different from those of the isomers. The lack of response around spectrum 25767 by the other mass chromatograms indicates that the compound eluting at this time differs greatly in structure from other compounds in this region. The narrow peak width and symmetrical shape of the mass chromatogram for m/z 93 indicated the presence of a single component. The tight cluster of MCPPP responses at this time confirms this conclusion. Mass spectra of the three compounds in region 2 obtained by deconvolution of GC/T OFMS/l’ AD data are shown in Figure 5-9. The spectrum of the first of the two isomers was obtained from a spectrum that occurred at the time of the shoulder in the elution profile because our deconvolution algorithm requires a m/z value unique to each coeluting compound. The two remaining mass spectra were generated using the deconvolution algorithm. The mass spectra of the two isomers differ only in relative intensity for all significant m/z values. This supports the MCPPP results and explains the shape of the elution profile of the RTIC and mass chromatograms in region 2. 118 Compound 1 19°) 131 41 1 147 57 117 l . 7E l ‘52 , 4o 60 80 100 3120 140' 160 180 200 z Rel. Intensity Compound 2 4., ‘39 2.- 'a 41 9 511,711l g 57 s 77 j c: - - l 40 60 80 100 [InZP 140 160 180 200 93 121 Compound3 Rel. Intensity 4455-717. 41 T550166 194 40' 60 80' 100 r‘lnP 140 160 180 200 elution profile (cormwnd 1), tl'redeccrwchrtedspectrumoftheccrrpcundu eMedatspectnun25735(ccmpound2)andflfedeccrmutedspectnunof hect- ccrmcundthstehrtedetspectrum25767(ccrmound3). 119 2-D GC/MS Analysis The RTIC shown in Figure 5-10 was obtained when region 2 was isolated onto a second GC column and the components separated. Since the goal of 2-D GC/MS is to resolve all of the components chromatographically, each peak should represent a single compound. Five peaks, labeled A-E, were found in this region. Compounds A, D and E were baseline resolved from all other components and, thus, could be easily distinguished from other components. Compounds B and C were not completely chromatographically resolved, but the shape of the elution profile revealed the presence of at least two components in this peak. Mass spectra obtained from region 2 by 2-D GC/MS are shown in Figure 5-11. While no identifiable mass spectrum was obtained for compound B due to the extremely small response at all m/z values and interference from compound C, the mass spectral data at the elution time of compound B contain the same key masses as compound D (is. m/z 135, 150, 163 and 191). The background- subtracted spectra of the other four components are clean and distinctive. Scmtiny of the mass chromatograms across the asymmetrical elution profile of compound E revealed no changes in the relative intensities of any m/z values. Region 2: Summary and Conclusions Substantial differences exist in the results produced by the two techniques in region 2. The most striking of these differences is that the deconvolution approach located only three compounds in region 2 while the chromatographic approach found five compounds. Examination of the mass spectra in Figures 5-9 and 5-11 shows J 2509! not”; :_fl£e7r .Steit .4297‘ .28071 .OCO?‘ .IEOSf .BE451 .ec461 .BCOB 120 .5 20.8 29.3 21.8 21.3 22.0 Tim. turn.) ;_ 22.3 23.. Figures-10. RTlCctthez-DGCIMSenereiect IrtdicetehepreeerrceofflverebeledA-Eh region-2. Theflvepedu tiieregicn. 121 a ._‘ 93 121 g CcmpoundA- .5: 135 E 4‘ 55 150 a; 107 191 - 7.7 4 l‘ 40 60 80 100 "1130 140 160 180 200 57 .- Compound—c." 5 s _‘ 1 E} 93 07 134 157 191 40 so so 100 $2,140 160 180 200 a. 4157 CompoundD .-. 1 g s 31 35149163 g 93 109123 _. I 178 m 191 m in I I I [A 4o 60 100120140 160 18 m o 200 z.‘ Compound E 189 17:" 9. 5. 147 354‘ 57 91 117131 mo 77' 159 00 40 60 80 mar: 00 Meatlheespmcefcmctthcivemm . m ' ("in 510)- M«mmywmelmmm.mml mmmmmt cored m. mummy-M 122 only two pairs of spectra that agree in composition and intensity. The spectra of compounds 2 and 3 from the deconvolution data match those of compounds E and A from the 2-D GC/MS data, respectively. The inversion .in the retention order is a result of the selectivity of the second column in the 2-D GC/MS analysis. The intensities of compounds C and D comprise less than 1% of the total mass contained in region 2 and were probably present at levels near the detection limits of the TOFMS/TAD system. Unfortunately, simply increasing the on-column sample concentration would not permit detection of compounds C and D by the deconvolution approach. The peak-finding algorithm requires that the intensities of all m/z values lie within the dynamic range of the mass spectrometer. The detector was saturated in region 2 for some m/z values when the sample concentration was increased to a suitable amount. The inability of 2- D GC/MS analysis to find the first compound located by the deconvolution approach reveals its intrinsic weakness. Compounds must be chromatographically resolved from any compound in the region, especially those with similar mass spectra. The most likely elution time for this compound was nearly the same as that of compound E. Coelution of these two isomers explains the asymmetrical shape of the elution profile for peak E, but the large degree of similarity between the mass spectra of the two compounds disguises the presence of a second compound. Combining the results of the two approaches, a total of six compounds were located in this region. Only two of these six compounds were identified by both approaches. The dynamic range of the TOFMS/T AD system restricted the 123 deconvolution approach to locating and identifying the three compounds with the largest intensities. Five of the six compounds were located by 2-D GC/MS including the three compounds with the lowest intensities. Spectra for all of these compounds could be obtained with little difficulty. However, 2-D GC/MS failed to locate the compound with second largest intensity of those in region 2. Although this compound was partially resolved from its isomer by the first GC column, the selectivity of the second column caused the retention times of the two isomers to converge. 5.4 Conclusions Each of the two techniques increases the amount of information available from chromatographically coeluting regions of complex mixtures over the traditional GC/MS approach. When these two techniques were applied to the two regions of interest, the results were sometimes complementary, revealing the strengths and weaknesses of each approach. A great advantage of GC/high-speed MS with data deconvolution is the reduced analysis time. The deconvolution approach operates on data acquired in a single chromatographic mn, no matter how many regions are selected for detailed analysis. For 2-D GC/MS, separate mns are required for each region suspected of needing further resolution. The MCPPP lines provide information about the degree of structural similarity and the retention time of coeluting compounds. The use of high-speed MS detection also provides more information 124 about the shape of the elution profiles for all masses, allowing the location of compounds not visible by traditional GC/MS. Since our present deconvolution approach is based on the presence of a unique m/z value for each coeluting compound, chromatographically unseparated compounds with very similar mass spectra will not be located or identified. A compound will also not be located if the detector is saturated for any significant m/z value because the MCPPP locates the point of maximum intensity for each mass in a chromatographic peak. Thus, the dynamic range of this approach is limited by the need to maintain the integrity of the chromatographic profile for all mass chromatograms. The major strength of 2-D GC/MS is its capability to physically separate compounds with similar mass spectra that coelute in the primary GC analysis. The dynamic range of this approach is larger because intense responses by compounds present at high concentrations can saturate the detector without affecting regions of the chromatogram containing smaller responses. The presence of trace compounds can be determined by overloading the first chromatographic column to pass a detectable amount of material into the second chromatographic dimension. Chromatographic resolution of these trace compounds reduces the background and permits their detection at very small concentrations. The intrinsic limitation of 2-D GC/MS is the need to chromatographically resolve all of the components placed onto the second column by the differing selectivity of the chromatographic columns. Sometimes, as occurred in region 2, this approach backfires and the retention times of compounds become more similar. When this occurs, santiny of the shape of the 125 elution profile may reveal the presence of multiple components. These compounds are not otherwise detectable. These two approaches, designed for the same purpose, may reveal complementary information about the same coelution. Use of both techniques is desirable for this reason. The deconvolution approach can provide a large fraction of the information about the regions containing coeluting compounds in a relatively small amount of time. Deconvolution data, through MCPPP and elution profile information, and sniff-port results can be used to locate those regions requiring more chromatographic resolution or dynamic range. The structural information available from the data used to generate the MCPPP can also be used as a tool in selecting the conditions for 2-D GC/MS analysis. Increased information about coeluting species and the need to analyze fewer regions via 2-D GC/MS can significantly reduce the time required to characterize a complex mixture of unknown composition. 5.5 Summary Compounds that coelute when analyzed by GC/MS may be analyzed using 2-D 60 with mass spectrometric detection or by deconvoluting GC/MS data acquired using the MTOF/ ITR system. While 2-D GC/MS instrumentation is commercially available, this approach is time-consuming due to the need to physically isolate coeluting compounds onto the analytical column. The deconvolution approach has the potential to reduce the time without sacrificing 126 any information. The perfume sample of unknown composition used for this study revealed strengths and weaknesses in each technique. Combining these two techniques could reduce the time required to analyze complex mixtures and use the strengths of each technique. Analysis of complex natural mixtures is difficult because of the presence of many isomeric species that possess similar retention and spectral characteristics. This problem is exacerbated when a sample of unknown composition contains hundreds of components. Chromatographic resolution of all of these compounds often cannot be achieved in the limited separation space available with even capillary column GC. Compounds with very similar mass spectra are difficult to locate when they coelute from a chromatographic column. References 1. Guiochon, 6.; Gonnard, M. F.; Zakaria, M.; Beaver, L. A.; Siouffi, A. M., Chromatographia 1 983, 17, 121. 2. Davis,J. M.; Giddings, J. C., Anal. Chem. 1983, 55, 418. 3. Tecklenburg, R. E., Jr.; McLane, R. D.; Grix, R.; Sweeley, C. 0.; Allison, J.; Watson, J. T.; Holland, J. F.; Enke, C. G.; Gruner, U.; Gotz, H.; Wollnik, l-l., Proceedings of the 38th ASMS Conference on Mass Spectrometry and Allied Topics , Tucson, AZ, June 3-8, 1990. 1 27 Rodriguez, P. A; Eddy, C. L.; Marcott, C.; Fey, M. L.; Anast, J. J. Microcol. Sep.1991, 3, 289-301. Holland, J. F.; Newcome, B.; Tecklenburg, R. E., Jr.; Davenport, M.; Watson, J. T.; Enke, C. G. Rev. Sci. Instrum.1991, 62, 69. Chapter 6 Photodissociation in Tandem Mass Spectrometry 6.1 Introduction The same problems that have limited GC/MS analysis are present in tandem mass spectrometry (MS/MS). Several factors unique to tandem mass spectrometry increase the time required to acquire complete product spectra of all precursors. These factors priman‘ly result from the processes necessary to isolate the precursor ion of interest, fragment the selected precursor and acquire the product ion spectrum. The same trade-off between scan speed and intensity present in sector and quadrupole instruments plays a role in the time needed for precursor ion selection and product ion analysis. Acquisition of a single product ion spectmm requires setting the conditions for the mass analyzer used for precursor ion selection and scanning the mass analyzer used for product ion analysis. Commercial MS/MS instruments often require several minutes to acquire complete MS/MS data sets (all products of all precursors). 6.2 Qualities Required For MS/MS Using TOF Instruments The keys to MS/MS analysis in a TOF instrument are high resolution for selection of the precursor m/z value, effective fragmentation of the selected 128 129 precursor, transmission of a large fraction of the product ions formed to the detector and at least unit resolution of product ions. Since resolution in a TOF instrument is related to the flight time of an ion, both precursor and product ion resolution increase with the length of the flight path. lon packets have their narrowest spatial distribution at the space-focus plane(s) of a TOF instrument. Thus, precursor ion selection is best achieved at a space-focus plane a significant distance from the ion source. Effective fragmentation and high transmission of the precursor ion are dependent on the dissociation technique. A dissociation technique must introduce energy into the precursor ions under conditions that cause a significant fraction of them to fragment and reach the detector. These requirements have led to the use of SID and PID as the two major dissociation techniques for TOF instruments. When PID is used, the laser power must be sufficient to insure that a significant fraction of the selected ions in the packet absorb a photon and fragment. The laser pulse rate should be the same as the extraction rate of the ion source to provide maximum sensitivity for the production of product ions. After fragmentation has occurred, the newly-formed product ions travel at the same velocity as their precursor. These product ions have a fraction of the kinetic energy distribution of their precursor as well as the energy acquired in the photodissociation process. Acceleration to a higher energy just after fragmentation occurs imparts mass-dependent velocities to the product ions. Mass separation now occurs in a manner similar to the first TOF mass analyzer 130 except that the product ions are not monoenergetic. Ions have the energy received in the acceleration region plus the fraction of precursor ion energy retained after fragmentation. For an ion mirror to achieve perfect energy focusing, all of the ions should have the same approximate energy. Since the second mirror is optimized for the energy of the precursor ions, the product ions will not be perfectly focused. The best product ion resolution can be achieved as the energies of the precursor and fragment ions become more similar. Thus, as the acceleration voltage at the interaction region is increased and as the initial energy of the precursor ion is decreased, a wider range of product ion masses will be focused. Unit mass resolution should be achieved for all product ions of precursors compatible with 60 analysis. 6.3 Photodissociation in Mass Spectrometry Photodissociation (PlD) as a means of inducing the fragmentation of ions has been used with several different types of mass spectrometers and light sources (see Table 6-1). Photodissociation efficiency plays a major role in the selection of the choice of mass spectrometer and light source to be used in an analysis. If the photon and ion densities are uniform and overlap completely when laser excitation occurs, photodissociation efficiency can be described by the relation Efficiency = —'r::—°— = (pa I 131 where Nc is the number of fragmentation-inducing collisions occurring between photons and ions in one laser pulse, N. is the number of ions in the irradiated area, it is the photon density of the laser pulse in the region populated by ions, and o is the photon interaction cross-section for the selected ion at a given wavelength. Thus photodissociation efficiency can be improved by increasing the number of photons given the opportunity to interact with an ion, increasing the number of ions present in the laser beam and choosing the radiation wavelength judiciously. These factors all play roles in the approaches used to perform photodissociation in mass spectrometers. The orientation of the laser to the ion beam is dependent on a variety of factors including ion densities and type of mass spectrometer used. Many researchers have sought to maximize the ion exposure time to the laser beam by irradiating the sample for up to 3 s in an ion cyclotron resonance mass spectrometer (lCR) or by making the laser beam coaxial to the ion beam in sector or quadrupole instruments. Ion sources in sector and quadrupole mass spectrometers usually form and analyze ions continuously. The ion densities in these instrument are relatively small. McGilvery and Morrison [1], and Krailler and Russell [2,3] have directed the laser perpendicularly to the low-density ion beam in quadrupole and sector mass spectrometers, respectively. The sensitivity problems that they experienced forced them to adopt a coaxial geometry that increased the number of ions intercepted by the laser light path. Trapped ions in lCRs can be exposed to the laser for relatively long periods of time (up to 3 5) allowing single and multiphoton PID processes to occur. But 132 Table 6-1. Historical Background of Photoionization/Photodissociation in MS. investigator: instrument: Light Source: Wavelength: Energy/ Photon: Pulse Width: Power Photons Required: °/o Fragment. Photon Morrison‘ TOMS (axial) DyeLaser 606 nm (vis) 2.0 eV 1 us 30 mW/cm2 = 5001.0 ) frequency = 10.0; if( duty_fraction ==- 0.0 ) duty_fraction = 0.1; if( duty_fraction >= 1.0) duty_fraction = 0.9; /' Initialize timer and start pulsing */ lnit_DA_Brds (3, &board); CTR_Rate (frequency, duty_fraction, &timebase, &H|_period, &LO_period); CTR_Square (3, 1, timebase, Hl_period, LO _period); delay(0.1); RemovePopup (0); break; } l 220 /' This function sets the number of points to be displayed in the popup graph. */ void Get_Graph_lnfo() I int graph_event; GetPopupEvent (1, &graph_event); switch (graph_event) { case graph _pts_to _graph: break; case graph_retum: GetCterai (graph_handle, graph _pts_to _graph, &pts_to _graph); GetCteral (graph_handle, graph_first_pt, &first_pt); if (first _pt < 0) first __pt = 0; if (pts_to_graph > 8192) pts_to_graph = 8192; RemovePopup (0); break; case graph_fi rst_pt: break; case graph_regraph: RemovePopup (0); GetCteral (file_handle, name_file, file_name); GetCteral (trans_handle, trans_start_point, &start_pt); GetCteraI (trans_handle, trans_number_pts, &number_pts); GetCtrIVai (trans_handle, trans _pts_to_array, &pts_to_array); GetCteral (graph_handle, graph _pts_to _graph, &pts_to _graph); GetCtriVal (graph_handle, graph_first_pt, &first_pt); if (first _pt < 0) first _pt = 0; if (pts_to_graph > 8192) pts_to_graph = 8192; WaveformGraphPopup (spec, pts_to_graph, 4, 1.0, 0.0, first _pt, 1.0); return; break; 221 l r This functions sets the delay values to be used by the LeCroy 4222 delay generator. '/ void Get_Deiay_lnfo(void) l int delay_event; int temp; GetPopupEvent (1, &delay_event); if (delay_event== hes_return) { GetCteral (delay_handle, hes_hest, &delay1); GetCteral (delay_handle, hes_hes2, &delay2); GetCteral (delay_handle, hes_hes3, &delay3); GetCteraI (delay_handle, hes_hes4, &delay4); if (delay1 < 170){ delay1 = 170; MessagePopup (”Deiay1 set to smallest legal value"); i if (delay2 < 170) { delay2 = 170; MessagePopup ("Delay2 set to smallest legal value"); } if (delay3 < 170) { delay3 a 170; MessagePopup ("Delay3 set to smallest legal value”); } if (delay4 < 170) { delay4 = 170; MessagePopup (”Delay4 set to smallest legal value”); i if (delay1 > 16777385){ delay1 = 16777385; MessagePopup (”Deiayt set to largest legal value"); 222 if (delay2 > 16777385){ delay2 = 16777385; MessagePopup (”Delay2 set to largest legal value”); } if (delay3 > 16777385){ delay3 = 16777385; MessagePopup ("Delay3 set to largest legal value”); } if (delay4 > 16777385){ delay4 :- 16777385; MessagePopup ("Delay4 set to largest legal value"); } RemovePopup (0); return; } l I. This function sets the values for components of the transient recorder. */ void Get_Trans_lnfo () 1 int trans_event; GetPopupEvent (1, &trans_event); if (trans_event == trans_return){ GetCtrtVal (trans_handle, trans_instmm_prog, &instrum_prog); GetCteral (trans_handle, trans_source, &sig_source); GetCteral (trans_handle, trans_start_point, &start_pt); GetCteral (trans_handle, trans_number_pts, &number_pts); GetCteraI (trans_handle, trans _pts_to_array, &pts_to_array); if (instmm_prog < 0) MessagePopup (”Error in Instrument Program Selections”); if (sig_source < 0) MessagePopup (”Error in Signal Source Selections"); it (start _pt < 0) start _pt = 0; 223 it (start _pt > 50000){ start _pt = 50000; MessagePopup ("The starting point has been set to 50000.”); i if (number _pts <= 0) number _pts = 5000; if (number _pts > 50000) { number _pts = 50000; MessagePopup (”T he number of points to be acquired has been set to the legal limit"); } if (pts_to_array <= 0) pts_to_array = 5000; if (pts_to_array > 8192) pts_to_array = 8192; RemovePopup (0); return; i i /" This function changes the name of the file to be acquired. ‘/ void Change_l'-”rle_Name () { int num; int temp; GetPopupEvent (1, &num); if (num == name_retum) { GetCtrIVal (file_handle, name_file, file_name); RemovePopup (0); return; } 224 [I This function acquires and graphs the data using the entries into all of the user interface panels. '/ int Acquire_Data() int delay_status; int num_pts; double interval; 0'80; Setup_Run(); /' The LeCroy 8901 CAMAC controller is initialized at GPIB address 5.The LeCroy 4222 delay generator is initialized and the delay times are set to appropriate values. The delay generator is then triggered and the CAMAC controller closed.*/ I‘ I lecCAMAC_init (5); lec4222_set_module_location (5, 1); lec4222_disable_trigger (); lec4222_set_delay_times (delay1, delay2, delay3, delay4); lec4222_enable_trigger (); delay_status = lec4222_get_status (); lecCAMAC_close 0; } ‘l I‘ Data are acquired on the LeCroy 9450 digital oscilloscope. The instrument is initialized at GPIB address 4. The default analysis conditions are loaded by recalling the contents of Setup Panel 4 and the data are acquired. The acquired data are written to a file. The contents of the file are then writtento an array called "spec”. (Both the file and the array contain 5000 elements.)*/ error = lec94xx_init (4, 0); if (error l= 0){ 225 MessagePopup ("Transient Recorder Initialization Error"); FmtOut("Transient Recorder Initialization Error:\t%d\n", error); return (-1); 1 error - lec94xx_save_load_setup_inst (1, instrum_prog); if (error l= 0){ MessagePopup (”Error Loading Instrument Program "); FmtOut("lnstrument Program Loading Error:\t%d\n", error); return (-1); } finished_summing 0; error = iec94xx_read_wvfm_to_file (sig_source, file_name, 1, start _pt, number _pts); if (error l= 0){ MessagePopup (”Error Writing Waveform to File"); Frr1tOut("CouId not write waveform to file. Error:\t%d\n", error); return (-1); l error = lec94xx_read_wvfm_file_arr (file_name, pts_to_array, spec, &interval, &num_pts); if (error I= 0){ MessagePopup (”Error Writing File to Array”); FmtOut("Could not waveform to array. Error:\t%d\n", error); return (-1); error = lec94xx_close 0; if (error l= 0){ MessagePopup (”Error Closing Transient Recorder"); FmtOut("Error closing Transient Recorder:\t%d\n", error); return (-1); } 226 I. This function gets the values for all variables used by the Acquire_Data function. The number of points contained in the file, graph and array are compared to produce the most appropriate values for display. ‘/ void Setup_Run () { GetCtrIVal (file_handle, name_file, file_name); GetCteral (delay_handle, hes_hesf, &delayt); ' GetCteral (delay_handle, hes_hesZ, &delay2); GetCteral (delay_handle, hes_hesS, &delay3); GetCtrIVal (delay_handle, hes_hes4, &delay4); GetCteraI (trans_handle, trans_instnrm_prog, &instrum_prog); GetCtriVal (trans_handle, trans_source, &sig_source); GetCteral (trans_handle, trans_start_point, &start_pt); GetCterai (trans_handle, trans_number_pts, &number_pts); GetCtriVal (trans_handle, trans _pts_to_array, &pts_to_array); GetCtriVal (graph_handle, graph _pts_to _graph, &pts_to _graph); GetCtrIVal (graph_handle, graph_first_pt, &first_pt); GetCteraI (pulse_handle, rate_frequency, &frequency); GetCtrIVal (pulse_handle, rate_duty_fraction, &duty_fractlon); if (pts_to_g raph > pts_to_array) pts_to_g raph = pts_to_array; } /‘ This function queries the LeCroy 945010 determine whether or not the procedures performed by functions E and F (particularly averaging) are complete. When this procedure fails, a negative number is returned. Success is indicated by a return value 010. */ int finished_summing () I 227 int user_req; int error; int comm_err; int exec_err; int dev__spec; int inr_vai; int event_status; int status_byte; inr_val = 0; GetCtriVal (trans_handle, trans_source, &sig_source); I. If using Fcn E or Fcn F, but not both functions simultaneously, the following section of code should be used. The program will not work properly if both functions are used together. *I I. if (sig_source == 10){ while (inr_val l= 2049){ error = lec94xx_read_stat (&status_byte, &event_status, &inr_vai, &dev_spec, &exec__err, &comm_err, &user_req); if (sig_source -- 9){ while (inr_vai l= 1025){ error a: lec94xx_read_stat (&status_byte, &event_status, &inr_val, &dev_spec, &exec_err, &comm_err, &user_req); } 'l I. If using Function F of the LeCroy 9450 via a signal processed by Function E, use the following special routine. This routine should not be commented and the preceding section MUST be commented for the program to run properly. */ if (sig_source == 10){ while (inr_val l= 3073){ error = lec94xx_read_stat (&status_byte, &event_status, &inr_val, &dev_spec, &exec_err, &comm_err, &user_req); } } 228 A.1.1 Include File Called by the TOF/TOF Instrument Control Program This program defines the response to buttons used by the TOF/T OF instument control program. */ */ /‘ LabWindows User interface Resource (UIR) Include File /‘ Copyright (c) National Instruments 1991. All Rights Reserved. /" WARNING: Do not add to, delete from, or otherwise modify the contents */ l’ of this include file. */ #define #define #define #define #define #define #define #define #define #define #define #define #define #define #define #define #define #define #deflne #define #define #define #define #define tofpan 0 tofpan_power 0 tofpan_hes 1 tofpan_trans 2 tofpan_graph 3 tofpan_start 4 tofpan_name 5 tofpan_led 6 tofpan_rate 7 trans 1 trans_instnrm_prog 0 trans_source 1 trans_start_point 2 trans_number_pts 3 trans _pts_to_array 4 trans_return 5 trans_messageI 6 trans_message2 7 trans_message3 8 has 2 hes_hes1 0 hes_hes2 1 hes_hes3 2 hes_hes4 3 #define #define #define #define #define #define #define #define #define #define #define #define #define #define #define 229 hes_retum 4 hes_times 5 name 3 name_file 0 name_retum 1 name_file_info 2 graph 4 graph _pts_to _g raph 0 graph_retu rn 1 graph_first_pt 2 graph_regraph 3 nus 5 nae_fiequency 0 rate_duty_fraction 1 rate_retu rn 2 230 A.1.2 Function Panels Called by the TOF/TOF instrument Control Panel Essen... “BEE 2. a. .28 Sec... .228 Ease 9:. .3 28.“. _co~¢ww~awuc anaem— _wcamwwucau seesa— —seuneuex acowwcexw_ , _o~mm encam— w=e_8m cam _2s 3:— fiOhfip—DU flcgaflwzH haHOH 231 3.88. 8 2 a... 2. .o 2.2 o... a. .55.. 58.5... .«.< 2.5.“. _:u=uo=_ ._o=em m_:p gem / an woweuamom on was: mo_aopuou_= _ «em. dome/5533— ; Wm; noe_:&uc an op ofiwm ego: o_—m 232 .28 395. .2. as 85.5 a. .29» :8... 25: 3:8 2.. .o as 5.88. 2: .3 23: _ 6550i _ a.s_ _ aa.=sm_ T3395 :91: =H_co.,:ro._,u . opus o._=a access 233 5.828 8.2 «8' .203 2:2 82.. 58 .22. 2 .22 5.2:“. .7... 2.6... _:n=wo=_ w: =_ moau_— one wan—o: _ s__ _v e_.=_ _ 5:_ _~ am.2: as. _m ..2...: _ ssm_ .mc.cn mam—o: wow 234 SE82 .2322. 8.3 >203 a... a. 2338 .0 Eco 8. .23.. 29.95“. .m.< 23E .ocmasesasm m=_=:=mruo:m_:_m osu =_ oweu cap me :=_uhom _a_uomm one on: .133? m 5”— 3: "— E."— mfiwa an E __________fl use: an eagmem m gum m cum h TEE— : lo: @ magma _ use c H; m an”. The... .2 .35.: new M "a.“ N :0 N no.3...— ! z. _. a... E E 55:95 anagbwaw seepage: pgoungeah 235 22.28 .32.. 858.... 528 a. .28 5.22 .3. 2.5.“. _amauu assuag— _=s:ao:_ r sfl _ L :gcgw :Ame:_am L: cones: _amacu aw ac_am vacuu— e=e_a_s=eo sauna 236 A2 Program to Parse Data File Header for Acquisition Information I' This file parses the header of a binary file acquired by the LeCroy 9450 to determine pertinent information for further data processing. The only line in this program that should be edited by a user is the constant ”FILE_NAME". This value should be changed to an appropriate name prior to runnimg the program. *I #define FILE_NAME "F:\\lw\\data\\brb6.dat” char ou10[15]; char out1[15]; char out2[15]; char out3[15]; char out4[15]; char out5[15]; char out6[15]; char out7[15]; main () { ASCII_header (FILE_NAME, 0, outO); ASCII_header (FILE_NAME, 1, out1); ASCII_header (FILE_NAME, 2, out2); ASCII_header (FILE_NAME, 3, out3); ASCII_header (FILE_NAME, 4, out4); ASCII_header (FILE_NAME, 5, out5); ASCII_header (FILE_NAME, 6, out6); l' ASCII_header (FILE_NAME, 7, out7); */ c'80; FmtOut ("The name of the file is: \t%s\n", FILE_NAME); FmtOut("The acquisition time was:\t%s\n", outO); FmtOut("The acquisition date was:\t%s\n", out1); FmtOut("The number of points was:\t%s\n", out2); FmtOut(”The sampling interval was:\t%s\n", out3); FmtOut("The number of points per screen was:\t%s\n",out4); FmtOut("The first point was:\t%s\n”, out5); FmtOut("T he probe attenuation was:\t%s\n", out6); 237 /' FmtOut(“The channel coupling was:\t%s\n", out7); '/ l 238 A.3 Program to Convert LeCroy Format Binary Data to ASCII Format This program converts a binary data stored in the format used by the LeCroy 9450 transient recorder to a tab delimited ASCII file. static int error; static int num_of_pts; static double sample_int; static double spec[8192]; main () ASCII _gen ("F:\\lw\\data\\brb5.dat", 5000, spec, &sample_int, &num_of_pts, error, "F:\\lw\\xldata\\brb5.dat”); } 239 A31 ASCII _gen Instrument Used by the Format Conversion Program This program is a modification of a format conversion program used by the LeCroy 9450. The major feature change is that this program does not require the 9450 to be turned on or initialized before the file format is transformed. static char buffer [17000]; static int Iec9450_invaiid_integer_range (int, int, int, int); static int lec9450_file_exists (char *); static int lec9450_err; static char *channel_cp[5]; I. *l I' This function reads the information from a waveform header file. *I I. ’/ int ASCII_header (file_name, info_type, out_str) char *file_name; int info_type; char *out_str; { int handle; intpad; int temp_time[2]; int year; int coup; intsp; int temp_int; double temp_doub; double sec; long temp_long; /‘ if (lecQ450_device_closed () l= 0) return lec9450_err; ‘/ if (Ifile_name[0]) { lec9450_err= -2; 240 return lec9450_err; } if (Ilec9450_file_exists (file_name)) { lec9450_err = 310; return lec9450_err; l handle -—- OpenFile (file_name, 1, 0, 1); if (ReadFile (handle, buffer, 500) < 0) { CloseFile (handle); lec9450_err = 315; return Iec9450_err; } if (CloseFile (handle) < 0) { lec9450_err = 316; return lec9450_err; } pad = FindPattem (buffer, 0, -1, "WAVEDESC", 0, 0); switch (info_type) ( case 0: pad += 296; if (Scan (buffer, "%1f[i‘z]>%f", pad, &sec) I: 1) { Iec9450_err = 236; return Iec9450_err; l pad += 8; if (Scan (buffer, "%2d[b1zi*]>%2d", pad, temp_time) l= 1) { lec9450_err = 236; return lec9450_err; } Fmt (out_str, "%s<%d[p0w2]:°/od[p0w2]:%f[p2]", temp_time[1], temp_time[0], sec» break; case 1: pad += 306; if (Scan (buffer, "%2d[b12i*]>%2d", pad, temp_time) l= 1) { lec9450_err = 236; return lec9450_err; } pad 4»: 2; if (Scan (buffer, "%1d[zi’]>%d", pad, &year) I= 1) { lec9450_err = 236; return lec9450_err; } 241 Fmt (out_str, "°/os<%d[p0w2]/°/od[p0w2]/%d", temp_time[t], temp_time[0], year): break; case 2: pad += 116; if (Scan (buffer, "%1d[i*b42]>%d[b4]”, pad, &temp_long) l= 1) { lec9450_err = 236; return lec9450_err; I Fmt (out_str, "%s<%d[b4]", temp_long); break; case 3: pad += 136; if (Scan (buffer, "%1d[i‘b4z]>%d", pad, &sp) l= 1) { Iec9450_err = 236; return Iec9450_err; pad += 40; if (Scan (buffer, "%1f[i*b4z]>%f", pad, &temp_doub) l= 1) { lec9450_err = 236; return Iec9450_err; } temp_doub *= (double)sp; Fmt (out_str, "%s<%f", temp_doub); break; case 4: pad += 120; if (Scan (buffer, "%1d[i"b42]>%d[b4]", pad, &temp_iong) l= 1) { Iec9450_err = 236; return lec9450_err; } Fmt (out_str, ”%s<%d[b4]”, temp_long); break; case 5: pad += 132; if (Scan (buffer, "%1d[i*b4z]>%d[b4]", pad, &temp_long) l= 1) { Iec9450_err = 236; return lec9450_err; l Fmt (out_str, ”%s<%d[b4]", temp_long); break; case 6: pad += 328; if (Scan (buffer, "%11[b4zi*]>°/od", pad, &temp_int) l: 1) { 242 lec9450_err = 236; return le09450_err; I Fmt (out_str, "%s<%d", temp_int); break; /‘ case 7: pad += 326; if (Scan (buffer, "%1d[zi']>%d", pad, &coup) I: 1) { lec9450_err = 236; return Iec9450_err; l Fmt (out_str, "%s<%s", channel_cp[coupl): break; *I l I' return Iec9450_err; ‘/ 1 I. *I /" This function reads a waveform file into an array.(from the LeCroy 9450 file) *I /‘ */ int ASCII _gen (file_name, num_of_pts, wvfm, x_incr, acq_pts, lec9450_err, output_file) char *file_name; char output_file[35]; int num_of_pts; double wvfm[8192]; double ‘x_incr; int *acq_pts; I int pad; int handle; int 1; int output; int count; int num_of_bytes; int num_read; long wav_arr_cnt; long sp; long new_sp; long wv_desc_len; 243 long usr_txt_len; double vert_gain; double vert_offset; double val; /" if (lec9450_device_closed () l= 0) return lec9450_err; */ if(ifile_name[0]) { lec9450_err= -2; return lec9450_err; i if (lec9450_invalid_integer_range (num_of_pts, 0, 8192, -2) I= 0) return lec9450_err; if (llec9450_file_exists (file_name)) { lec9450_err = 310; return Iec9450_err; } handle - OpenFile (file_name, 1, 0, 1); if (ReadFile (handle, buffer, 500) < 0) { CloseFile (handle); lec9450_err = 315; return lec9450_err; i if (CloseFile (handle) < 0) { Iec9450_err = 316; return Iec9450_err; I pad = FindPattem (buffer, 0, -1, "WAVEDESC", 0, 0); pad += 36; if (Scan (buffer, "%1d[i‘b42]>%d[b4]", pad, &wv_desc_len) l= 1) { lec9450_err = 236; return lec9450_err; l pad += 4; if (Scan (buffer, "%1d[i"’b4z]>%d[b4]", pad, &usr_txt_len) l= 1) { lec9450_err = 236; return lec9450_err; pad += 76; if (Scan (buffer, "%1d[i*b4z]>%d[b4]", pad, &wav_arr_cnt) l: 1) { Iec9450_err = 236; return lec9450_err; } pad += 20; 244 if (Scan (buffer, "%1d[i*b42]>%d[b4]", pad, &sp) I: 1) { lec9450_err = 236; return lec9450_err; } pad += 20; if (Scan (buffer, "°/o1f[i'b4z]>%f", pad, &vert_gain) I: 1) { lec9450_err = 236; return Iec9450_err; } pad += 4; if (Scan (buffer, "%1f[i*b4z]>%f", pad, &vert_offset) i=1) { lec9450_err = 236; return lecO450_err; l pad += 16; if (Scan (buffer, "%1f[i*b42]>%f", pad, x_incr) l= 1) { lec9450_err = 236; return lec9450_err; } new_sp = wav_arr_cnt / (long)num_of_pts; if (new_sp == 0L) { *acq_pts = (i nt)wav_arr_cnt; ‘x_incr *= (double)sp; num_of_bytes = 2; } else ( 'acq_pts .. num_of_pts; ‘x_incr = *x_lncr * (double)sp * (double)new__sp; num_of_bytes = (int)(2L * new_sp); } handle = OpenFile (file_name, 1, 0, 0); SetFilePtr (handle, wv_desc_len + usr_txt_len + (long)pad - 176L, 0); for (count = 0; count < *acq_pts; count++) { num_read = ReadFile (handle, buffer, num_of_bytes); if (num_read <= 0) { Closer-"fle (handle); lec9450_err = 315; return lec9450_err; } if (Scan (buffer, "%1d[z]>%f", &val) l= 1) { CioseFile (handle); lec9450_err = 236; return lec9450_err; } 245 wvfm[count] = val; } if (CloseFile (handle) < 0) { Iec9450_err = 316; return lec9450_err; } LinEv1 D (wvfm, *acq_pts, vert_gain, -vert_offset, wvfm); it I The ASCII file is written from the array. ’/ output = OpenFile (output_file, 2, 0, 1); for (i = 0; i < num_of_pts; i++){ Fthile (output, "%s<%f[w15] \t \n", wvfm[ij); } Closel-‘rle (output); return lec9450_err; } I. ‘/ I‘ This function checks an integer to see if it lies between a minimum */ l‘ and maximum value. if the value is out of range, set the global error */ /‘ variable to the value err_code. The return value is equal 0 for *l /' success and -1 for error. ’I It *1 int lec9450_invalid__integer_range (val, min, max, err_code) intvak int min; int max; int err_code; { if (val < min || val > max) ( Iec94xx_err = err_code; return -1; 246 1 return 0; } It */ I‘ This function checks to see if a file exists on disk I. */ int lecO450_file_exists (file_name) char ‘file_name; int file_handle; file_handie = Opeane (file_name, 1, 1, 1); CloseFile (file_handle); if (file_handle <= 0) return 0; else return 1; return lec9450_err; 247 A.3.2 Include File Called by the File Format Conversion Program This program defines the response to buttons used by the TOF/T OF instument control program. mags: l /‘ --------- LabWindows Generated Code: Mon Aug 10 21:27:11 1992 --------- *I l' = Lecroy 9410/14/20/9424/30/9450 DSO'S Include File m .l l' = GLOBAL FUNCTION DECLARATIONS *l int lec94xx_init (int, int); int Iec94xx_config_vert (int, double, int, double); int lec94xx_config_horiz (int, int, int); int Iec94xx_config_probe (int, int); int lec94xx_config_disp (int, int, int, int, int, int, int, int, int, int, int); int lec94xx_auto_setup (void); int Iec94xx_config_stand_trig (int, int, int, int, double, double); int lec94xx_config_smart_trig (int, int, int, double, int, int, int, int); int lec94xx_read_wvfm_arr (int, int, long, double [8192], double ‘, int '); int lec94xx_read_wvfm_to_file (int, char *, int, long, long); int Iec94xx_write_wvfm_to_inst (int, char *); int lec94xx_read_wvfm_inst_mem (int, int); int lec94xx_parse_wvfm_header (char *, Int, char *); int lec94xx_read_wvfm_file_arr (char *, int, double [8192], double *, int *); int Iec94xx_trig (void); int Iec94xx_read_stat (int *, int *, int *, int *, int *, int *, int *); int le094xx_save_ioad_setup_fiie (char *, int, int); int lec94xx_save_load_setup_inst (int, int); int lec94xx_hard_copy_setup (int, int, int, int, int, int); int lec94xx_hard_copy (int); int lec94xx_close (void); /‘ = GLOBAL VARIABLE DECLARATIONS ’/ /‘ Global error variable for the instrument module */ int lec94xx_err;