-‘ wfi?‘ LIBRARY Michigan State University This is to certify that the dissertation entitled The Use of Thermionic Emission Materials in Mass Spectrometry and Gas Chromatography presented by Daniel Bombick has been accepted towards fulfillment of the requirements for Ph .D. degree in ChemistLL Major professor /%,4 M Date 2?M [79” MS U is an Affirmative Action/Equal Opportunity Institution 0- 12771 MSU LIBRARIES —— . RETURNING MATERIALS: Place in book drop to remove this checkout from your record. FINES will be charged if book is returned after the date stamped below. THE USE OF THERMIONIC EMISSION MATERIALS IN MASS SPECTROMETRY AND GAS CHROMATOGRAPHY BY Daniel Bombick A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Chemistry I 986 This is dedicated to my family, especially my brother, Dave, for the understanding and cooperation they showed me while this work was being completed. ACKNOWLEDGEMENTS I would like to thank John Allison for his guidance, encouragement, and friendship. No one could ask for a better research advisor and friend. I would like to thank Matthew Zabik and J.T. Watson for their help in the writing of this thesis. The friendship and help of my fellow students will be remembered always. Tony, Barb, Pinky and Mac continue to be very special people to me. Mac made the late nights at the Chem building interesting. Tony 'opened my eyes‘ to Greek food. Pinky deserves a round of applause for his organization of our group 'picnics'. When I thought that I was being swamped with work, I would take a look at Barb and the kids and quickly decide otherwise. D.K., Rich, Paul, and Mark (my 'second research group') were excellent companions. (Rich causes me to believe I should have taken debate in high school). Karen, Kris, and Ellen (the Flamingo crew) caused my leaving to be a very difficult one. I wish everyone continued success. I would like to thank the faculty and staff of the MSU chemistry department for all the excellent help during my stay. I wish to thank the MSU/NIH Mass Spectrometry Facility for all their help. The staff at the facility were ‘i. . ‘—.~.— —_V‘—— ABSTRACT THE USE OF THERMIONIC EMISSION MATERIALS IN MASS SPECTROMETRY AND GAS CHROMATOGRAPHY BY Daniel Bombick Certain solids when heated and positively biased emit ions. Here, an alumlnosilicate matrix containing K20 is used as a source of gaseous potassium ions. A simple method is presented for introducing these emitters into the chemical ionization (Cl) source of a mass spectrometer. K" adds to most compounds containing 11- or n-donor sites to produce an (H+K)" ion. This method provides a novel means for performing K+ Cl. Some compounds, notably amines, react on the hot surface to produce ions. The behavior parallels the response of thermionic ionization gas chromatography detectors. When nonvolatile compounds are deposited on the emitter, ions representative of the compound are formed. This method is referred to as K‘IDS (potassium ionization of desorbed species). The use of the thermionic emission probe in K‘Cl is presented first. The analysis of nonvolatile compounds including polymers follows. Finally, the use of the K‘ probe in the elucidation of the mechanism of the gas chromatographic thermionic ionization detector is presented. TABLE OF CONTENTS l Page I LIST OF TABI Fs viii ‘ LIST OF FIGURFS ix , LIST OF SCHEMFS xix F CHAPTER I - INTRODUCTION I I. Introduction I I II. Thermionic Emission 4 III. Surface Ionization I9 A. Positive Surface Ionization l9 B. Negative Surface Ionization 34 IV. Final Comments 40 CHAPTER 2 - POTASSIUM ION CHEMICAL IONIZATION AND OTHER USES OF AN ALKALI THERMIONIC EMITTER IN MASS SPECTROMETRY 4| I. Introduction 4| ll. Experimental 45 III. Results and Discussion 48 , A. Other Uses of the K" Thermionic Emitter Probe in l Mass Spectrometry (3‘ Id F— CHAPTER 3 - A NEW DESORPTION/IONIZATION MASS SPECTROMETRIC TECHNIQUE FOR THE ANALYSIS OF THERMALLY LABILE COMPOUNDS BASED ON Page THERMIONIC EMISSION MATERIAI S 78 I. Introduction 78 H. Experhnental 83 III. Results and Discussion 86 A. Applications Using K‘iDs 90 I. Saccharides 90 2. Pharmaceuticals 95 3. Peptides 98 4. Steroids l03 S. Mixture Analysis I06 8. Sensitivity I I4 C. Surface Ionization of Thermally Labile Compounds....l I6 IV. Final Comments on Chapter 3 I2I CHAPTER 4 - MASS SPECTROMETRIC ANALYSIS OF POLYMERS USING K‘ips 124 L Introduction I24 H. Experhnental I27 III. Results and Discussion I28 A. General Considerations of Using K‘ios For Polymer Analysis B. Short-Lived Polymers I. 2. 3. 4. 5. C. Long-Lived Polymers IV. Final Comments on Chapter 4 General Considerations Cyanoethylmethyl Silicone (XF-l ISO) ................... Polyvinylpyrrol Idone Polyesters Polyamides CHAPTER 5 - MASS SPECTRAL AND GAS CHROMATOGRAPHIC INVESTIGATION INTO THE RESPONSE MECHANISM OF THE THERMIONIC IONIZATION DETECTOR FOR GAS CHROMATOGRAPHY I. Introduction II. Experimental III. Results and Discussion A. What Happens to an Analyte in the TID in the Absence of Any Other Gases? I. Negative Ion Formation 2. Positive Ion Formation vi Page I28 I34 I34 I39 I46 I49 I53 I55 I72 I72 I83 I83 I84 I92 3! Page 3. The GC-TID in the .“ “P of Gases I97 B. What Role Do Gases Have Between an Analyte and the Hot Surface? 198 I. The Effects of Gases in the MS-TID ....................... I98 2. The Effects of Gases in the GC—TID ....................... 204 3. Temperature of the Bead 904 D. Air in the GC-TID 709 c. The H2/Air System in the GC-TID ................... 2l8 d. Nitrogen in the GC-TID 998 C. What are the Similarities Between the MS-TID and the GC-TID 232 APPENDIX A - REPRESENTATIVE K’IDS SPECTRA ........................... 236 LIST OF REFERENCES 986 vii Table LIST OF TABLES List of Compounds Studied Using the K" Emitter Probe In the K‘CI Mode Formulas of the Polymers Used In the K‘IDS Analyses ..... Potassium Ion Adduct Series for Cyanoethylmethyl Silicone (XF-I ISO) Potassium Ion Adduct Series For Poly(Ethylene- adipate) Apparent Average Molecular Weights Using the First Ten K‘IDS Spectra For the PPG and PEG Polymers ............. Comparison of Positive Surface Ionization of Some Amines on Heated Alkali Alumlnosilicate and Oxidized Tungsten Surfaces viii Page '53 I40 I42 I52 I69 I96 Figure 2 LIST OF FIGURES Page The Thermionic Emission Experiment 3 A Free Electron in a Potential Well 8 The Energetics Involved in the Emission of Gas Phase Metal Ions From the Solid Metal 13 The Energetics Involved in the Emission of Gas Phase Metal Ions From an Aluminosilicate Surface ........................ I6 Schematic Diagram of Thermionic Emission Probe In the Mass Spectrometer's Ion Source 46 K‘Cl Spectrum of 2-Butanone 49 KTCI Spectrum of Diethyl Ether 50 Km Spectrum of Acetonltrlle 5| K’CI Spectrum of Ethyl Acetate 52 A Log-log Plot of I(M'tK)"/[I(M+K)+ + I(K’)l vs. Pressure of Acetone (I Refers to Intensity) SS K’CI (Surface Ionization) Mass Spectrum of Triethylamine. Bead Bias is Above K+ Addition Bias Window 56 Plot of I (Product Ion i)/ll(m/z I I2) + I(m/z 72) + I(m/z 58)] vs Bead Bias For Diethylamine. Pressure and Heating Current Were Constant at 650 mTorr and 2.0 A, Respectively S9 K’CI Spectrum of 2-Propanol With Bead Biased Within i<+ Addition Bias window 6| Figure Page I4 Plot or i (product Ion)/Il(m/z 98) + I(m/z 99) + I(m/z l37)] vs. Bead Bias 62 IS K‘ti Spectrum of I2-Crown-4. Bead Bias is within Addition Bias Window 64 I6 K‘CI Spectrum of Trlethylene Glycol DImethyl Ether. Bead Bias Is Within Addition Bias Window ............................ 65 I7 K‘CI Spectrum of IS-Crown-S. Bead Bias Within the Addition Window 67 l8 El Spectrum of IS-Crown-S. Bead Bias Was -23 V With Respect to the CI Volume in This Case 68 I9 The Mass Chromatograms and Total Ion Intensity Obtained From GC/ K’CI/MS Run In the Analysis of lpl of a Neat I:l Mixture of a Ketone and an Amine. Mass Spectra Were Collected From m/z 42 to ISO. The Bead Bias is Within the K‘ Addition Window 70 20 The Mass Chromatograms and Total Ion Intensity Obtained From GC/ K’CI/MS Run In the Analysis of Ipl of a H Neat Mixture of a Ketone and Amine. Conditions the Same as in Figure I9 Except Bead Bias Above K" Addition Bias Window 7| 2| K’Cl/Surface Ionization of Glycylglyclne. Heating Current Was 2.0 A 73 22 K’Cl/Surface Ionization of Glycyl PronI Alanine. Heating Current Was 2.0 A 74 23 K'Cl/Surface Ionization of Leucyl Glycyl Phenylalanine. Heating Current Was 2.0 A 75 24 Schematic Diagram of Thermionic Emission Probe In "I Figure Page 25 26 27 28 29 3o 3 i 32 33 34 3s 36 the Mass Spectrometer's Source. The Structure of the Probe Tip is Shown Enlarged 82 K'IDS Spectrum of Glucose 92 K’IDS Spectrum of Sucrose 94 K‘IDS Spectrum of Amplcillin 96 K’IDS Spectrum of Hexaglyclne. A Heating Current of 2 A Was Used. The Insert Shows the Molecular Ion Region For Which the Heating Current Was 2.5 A. The Relative Intensity In the Latter Case Was Approximately 2.5 Times That of the Previous (2 A) Spectrum ioo K’IDS Spectrum of 20 Year Old Cholesterol .......................... I04 K’IDS Spectrum of 'Fresh' Cholesterol. Conditions the Same as In Figure 29 IOS K‘IDS Spectrum of Poly(Ethylene Glycol) IOOO at 3 Heating Current of 2.5 A I08 K’lDS Spectrum of Poly(Ethylene Glycol) IOOO at a Heating Current of 2.0 A I09 Total Ion Intensity vs. Scan Number of Poly(Propylene Glycol) 725. The Scan Rate Was l.0 Scan/s ......................... I I I Total Ion Intensity vs. Scan Number of Ampicillln. The Scan Rate Was 0.6 Scan/s l I 2 KIDS Spectrum of Poly(Properne Glycol) 725. A Heating Current of 2 A Was Used I I3 Total Ion Intensity at Peak Maximum of Polyphenyl Ether Plotted vs. Amount Deposited on the Thermionic xi Figure Page 37 38 39 4I 42 43 45 Emission Material. A Heating Current of 2 A Was lined I I5 K’IDS/Surface Ionization Spectrum of Xanthlne. A Heating Current of 2.5 A Was Used I I8 K’IDS/Surface Ionization Spectrum of Theophylllne. A Heating Current of 2.5 A Was Used I I9 Negative Surface Ionization Spectrum of L-Methionyi Glycine Deposited on the Thermionic Emission Material. The Bias on the Bead Relative to the Source Was -I.0 V. A Heating Current of 2.5 A Was Used I22 Total Ion Intensity For K‘IDS Analysis of Nylon 6. Arrow Indicates When Current Was Applied ......................... I29 Total Ion Intensity for KIDS Analysis of Poly(Ethylene Glycol) 600. The Total Run Time Shown Here is Approximately 7 Minutes. Current Was Applied at Scan 7 I30 Intensity as a Function of Time for the Desorption of Species Due to Rapid Heating and the Emission of K". The Overlap Region Shows Where K’ Spectra are Produced I32 Averaged Spectrum of Nylon 6 From Scans I l-20 (See Figure 40) I36 Averaged Spectrum of Nylon 6 From Scans 20—40 (See Figure 40) I37 Averaged Spectrum of Nylon 6 From Scans I l-40 (See Figure 40).... I38 Averaged K'IDS Spectrum of XF-I 150 Ml xii r":— *3 Figure Page 47 Averaged K‘IDS Spectrum of Polyvinylpyrrolidone ............. I47 48 Averaged K‘IDS Spectrum of Poly(Ethyleneadeate) ........... ISI 49 Averaged K‘IDS Spectrum of Poly(Propylene Glycol) 725 Using Spectra Late In the K‘IDS Analysis .................... 156 50 Averaged K’IDS Spectrum of Poly(Propylene Glycol) 725 Using Spectra Early In the K’IDS Analysis .................. I57 5l Schematic Diagram of Potassium Thermionic Emission Probe In the Mass Spectrometer‘s Source Showing Various Thermal Regions I59 52 The Apparent Average Molecular Weight of Poly (Ethylene Gylcol) 400 as a Function of Scan Number ......... I6l 53 The Relative Intensities of the OIIgomer-Potassium Ion Adducts of Poly(Ethylene Glycol) 400 Plottes as a Function of Scan Number I62 54 Intensity as a Function of Time for the Desorption of Species Due to Rapid Heating; the Distillation of Species From 'Cold' Regions; and the Emission of K+ Ions. The Overlap Regions Shows Where K’IDS Spectra are Produced I64 55 The Percent of the Total Ion Intensity as a Function of Scan Number for the Potassium Ion Adducts of Decomposition Species of Poly(Ethylene Glycol) 600. The Heating Currents Used are 2.5 Amps for Curve B and 3.5 Amps for Curve A I66 56 The Two Modes of Operation In the KoIb-Blschoff TID ...... I75 57 Negative SI Spectrum of Acetonitrile Obtained Using the Alkali Aluminosilicate Bead I85 xiii r- 58 59 60 6| 62 63 64 65 66 67 68 69 Figure Page Negative SI Spectrum of Nitromethane Obtained Using the Alkali Aluminosilicate Bead l87 Negative SI Spectrum of n-Butyl Nitrite Obtained Using the Alkali Aluminosilicate Bead I88 Negative SI Spectrum of TrI-n-Butyl Phosphate Obtained Using the Alkali Aluminosilicate Bead ................ I90 Negative SI Spectrum of Hexane Obtained Using the Alkali Aluminosilicate Bead |9I Positive SI Spectrum of Triethylamine Obtained Using the Alkali Aluminosilicate Bead I94 Negative SI Spectrum of Tri-n-Butyl Phosphate Obtained Using the Alkali Aluminosilicate Bead. This Spectrum Was Obtained With a Source Pressure of I00 mTorr of Air 902 Bead Temperature as a Function of Detector Gas Flow Rate 205 A Typical Configuration of a TID 907 Plot of Log Response as a Function of Bead Current For Several Compounds at a Constant AIr Flow Rate of I60 ml/min 9| I Response as a Function of the Number of Injections For TrImethyI P‘ r‘ ‘u- 9I3 Response as a Function of Air Flow Rate For Several Test Compounds 9I6 Response as a Function of H2 Flow rate For Some Test Compounds at Constant Current and Air Flow Rate ............ 2I9 xiv r- Figure Page 70 Chromatograms of Test Mixture at Two Different H2 Flow Rates 7?] 7| Response as a Function of Air Flow Rate For Some Test Compounds at Constant Current and H2 Flow rate....223 72 Log Response as a Function of Bead Current For Several Test Compounds 794 73 Response as a Function of Bead Current For Several Test Compounds With a Constant I60 mI/min N2 Flow Rate 730 74 K‘Ios Spectrum of PhenolPhthalein 937 75 K‘los Spectrum oi Methyl Red 238 76 K'IDS Spectrum of Methylene Blue 939 .77 K’IDS Spectrum of Methyl Violet 240 78 K‘lDS Spectrum of Martius Yellow 94I 79 K’IDS Spectrum oi Murexide 942 80 K'IDS Spectrum oi NIIe Blue A 943 SI KIDS Spectrum of Magdala Red 944 82 K‘IDS Spectrum of Bromocresol Green (Na Salt) ................ 24S 83 K'IDS Spectrum of Chiorophenol Red 746 84 K'IDS Spectrum of of Phenosafranine 247 85 K’IDS Spectrum of Ouinine 948 ~ 3 Figure Page 86 K'IDS Spectrum oi Caffeine 949 87 K‘IDS Spectrum of Riboflavin 750 88 K‘IDS Spectrum of Hypoxanthlne 75I 89 K’IDS Spectrum of F-50 (Methyl Silicone) .......................... 252 90 K’IDS Spectrum of EPON IOOI 953 9i K’IDS Spectrum of Maltose 254 92 K’IDS Spectrum of Fructose 255 93 K‘IDS Spectrum oi Lactose 956 94 K'IDS Spectrum of Gum Arabic 257 95 K’lDS Spectrum oi Gum Ghatti 9% 96 K‘IDS Spectrum of Phenobarbital 759 97 K'IDS Spectrum of Penicillin G 260 98 K’IDS Spectrum of Rhamnose 96I 99 K’IDS Spectrum oi Sorbose 959 l00 K’IDS Spectrum of Ribose 263 IOI K’IDS Spectrum of Dextran 264 |02 K'IDS Spectrum of Glycogen 265 I03 K‘IDS Spectrum of Ascorbic Acid 966 xvi Q Figure Page I04 K‘IDS Spectrum of Melezitose 967 I05 K‘IDS Spectrum of Raffinose 268 l06 K‘lDS Spectrum oi Inulln 959 l07 K‘IDS Spectrum oi Amylose 970 I08 K’IDS Spectrum of Amylopectin 27I lo9 K’IDS Spectrum of Locust Bean Gum 972 I l0 K’IDS Spectrum of Allantoin 773 I II K‘ios Spectrum oi Porphyrin (MW-604) 974 | l2 K’ios Spectrum oi Triton XIOO 975 I I3 K’IDS Spectrum oi Tween 80 976 l l4 K‘lDS Spectrum oi 55-54 (Methyl Silicone) ........................ 277 I I5 K‘loS Spectrum oi Polybutadlene 978 I I6 K’IDS Spectrum of Polyoxyethylene Lauryl Ether ............... 279 I I7 K’IDS Spectrum oi Brij 30 980 I I8 K’los Spectrum oi Ouadrol 981 “9 KIDS Spectrum of Poly(Cylcohexanedimethanol Succlnate) 982 I20 K’IDS Spectrum oi w-9a 933 I2I K'IDS Spectrum oi Cupric Acetate 984 xvii '5 1-)" 9|}. V4- . .,y 0 LIST 0." St 'r‘lfii'ii.) ................... ‘. it- ‘3‘ < / ifs-(71 LIST OF SCHEMES Scheme Page I Scheme I 97 2 Scheme 2 IOI 3 Scheme 3 I20 4 Scheme 4 M4 5 Scheme 5 I48 xix LJNIBQDIICILQN This work discusses the uses of thermionic emission (TE) materials In the areas of mass spectrometry (MS) and gas chromatography (GC). Thermionic emission materials, as will be shown, are potentially useful In both of these areas. Three projects will be described and discussed which use various types of thermionic emission substances. The first project Is an investigation of thermionic emission materials for producing gaseous metal Ions as chemical Ionization (CI) reagent Ions for mass spectrometry. Secondly, these thermionic emitters can be used. In a particular manner that will be described, for the mass spectrometric analysis of nonvolatile and thermally labile Compounds. The final project discusses the N,P-GC detector which contains a thermionic emission material. The latter investigation pertains to the response mechanism of these detectors. While these topics may seem unrelated the underlying 'theme' Is the thermionic emitter. In addition, the manufacture of a thermionic emission probe for a mass spectrometer facilitated the study of all three experiments. It was found that during the course of these Investigations Information obtained from one study was useful for another. An Introduction of each project will not be made here but will precede its respective section In the manuscript. This Introduction wIII curtail a discussion of thermionic emission and surface Ionization. These I 2 principles will serve as a foundation for all three projects since they are necessary ingredients underlying the behavior of thermionic emitters. Consider the experiment, shown In Figure I which Is maintained at a low pressure. An appropriate solid, when heated and positively biased, emits positively charged species which can be collected. This process can occur for some solids that are negatively biased and results in the emission of negative charge carriers. This process, Involving the emission of charged particles (positive and negative) from a surface Is called thermionic emission. The surface Ionization (SI) experiment IS designed In a similar fashion as that shown In Figure I. Emission of negatively or positively charged species occurs when a gas phase molecule Is allowed to strike and adsorb onto the surface. This Is In contrast to thermionic emission where the species to be Ionized must first 'unfasten' Itself from the solid matrix and diffuse towards the surface. In both SI and TE, Ionization occurs when a particular species Is adsorbed onto the surface, however, the route the species took before adsorption on the surface is the fundamental difference between the two processes. It Is my opinion that this difference calls for a distinction to be made between these two processes. In this manuscript, thermionic emission will only be referred to as that process which produces emission of species which were originally a part of the solid matrix. The surface ionization process will Include only those processes In which an external 3 . ,5 a} "eC- hp.“- : ii") "Still-tit "i mealtime 5mssnon , . \ HEATING M MATERIAL —l— SUPPLY ‘eji‘y; \. - " I“; IIIASING rowan SUPPLY l Theiillermionic Emission Experiment. 4 source (not originally part of the solid matrix) of molecules directs them to a surface. This deposition can take place via the gas phase or by direct deposition of a solid onto the surface. These two different deposition modes alter those expressions characterizing surface Ionization efficiency. Here, only the gas phase case will be considered since It has been more widely studied. Much work remains to be done on the surface Ionization of solids on surfaces. Thermionic emission will first be discussed. This discussion will include basic theory and relevant work. Following thermionic emission will be a desrlption of surface Ionization outlined similarly as that of thermionic emission. W Historically, the theory of thermionic emission was developed from experiments Involving the emission of electrons from metals. Richardson (I), Schottky (2) and von Laue (3) Initially began work on thermionic emission In the early 1900's. Fowler and Nordheim (4) developed the theory, based on the supposition that electrons within a metal behave as free particles. This theory Is believed to be the cornerstone describing the emission of electrons from surfaces. A number of books and review articles concerning thermionic emission can be found (5-9). These references can be used for complete derivations of the formulae which will be presented. What Is covered here Is a brief Introduction of several fundamental equations governing 5 thermionic emission with respect to temperature, surface composition, and applied fields. The thermionic emission of both positively (cations) and negatively charged (electrons, anions) species display a temperature dependence which can be described empirically as I = f t exp(-X/kt) (I) where I Is the thermionic current (Amps), k Is the Boltzmann constant, and T Is the absolute temperature. The constant X refers to the energy needed to extract an Ion from the surface. Therefore a plot of log I against l/T should be linear and the slope equal to the value It. This linear relationship between log I and HT does exist for positive and negative thermionic emission. It Is most likely this temperature dependence which leads to the perceived similarity between surface Ionization and thermionic emission since In many cases surface Ionization and thermionic emission shows similar behavior. The actual energy required for separating an electron from a surface Is somewhat less than x and Is called the work function (0); sometimes referred to as the thermionic work function. In addition, from emission data, equation I was modified to Include a pre-exponentlal term which Is moderately dependent upon temperature. This led to the expression I = AT2exp(O/kt) (2) 6 where A Is a constant. At first It may be difficult to believe that equations I and 2 can be used to explain the same phenomenon. This Is the case since the temperature dependence on emission Is dominated by the exponential term. In fact, thermionic emission can be expressed by an equation which has the form = constant x f(T) x exp(-E/kT) (3) where f(T) Is only moderately temperature dependent and E represents the energy Involved In the emission of the charged species. Several modifications of equation 3 led to the well-known Richardson-Schottky formula which Is In terms of the emitted current density. Incorporated In this expression Is the effect of electric field and the transmission coefficient, D. This expression Is given In equation 4. J (A/cmz) = ADT2exp(-O-(e3F)'/2/kT) (4) Here A Is a universal constant which Is equal to I20 Acm‘zK'2 and referred to as the Richardson constant. F and e refer to the electric field strength and electronic charge, respectively. D, the transmission coefficient, Is a function of the reflection coefficient (r) such that D=(I-r). The reflection coefficient refers to that fraction of species which do not have sufficient thermal energies to surpass the potential barrier at the surface of the emitter. Generally, r Is approximately zero when the 7 temperature exceeds IOOOK so that D Is unity. The Importance of reflection coefficients has been discussed previously (8, IO-I I). For simplicity, In any discussion using equation 4, D will be assumed to be unity. The field strength term, (e3F)”2, can have a small contribution to the energy term In equation 4 considering typical field strengths used for thermionic emission. In the field emission of electrons (high field strengths, lower temperatures) this term becomes more Important. Consider the typical electric field (6.5xl0'5Na>K>Rb>Cs, since these metals' ionization energies decrease in the same direction. Experimentally, the emission of alkali metals follow the trend in ionization energies. l have discussed the thermionic emission of electrons from metals and introduced several fundamental equations describing this behavior. The emission of other species, both positive and negative, was similar to this behavior but required some modification to the electron emission formulae. Figures 3 and 4 can be used constructively to envision the important processes 18 involved in thermionic emission. The projects that will be described in this work rely on these thermionic emitters, however. do not involve research entailing the development of new emitters or a theoretical investigation of thermionic emission. These pursuits would be worthwhile in light of the many uses of these materials, some of which are discussed here. i have indicated why, in my opinion, a distinction should be made between thermionic emission and surface ionization. For thermionic emission to occur, species within a matrix must diffuse to the surface. In the case of condensed metals, sufficient energy (heat of sublimation) must be supplied to break the bonds between the atoms of the metal. For either case a condition must exist in which a species is adsorbed onto a surface. For surface ionization to be possible, a species must be brought to a surface and adsorb onto it. The end result is the same and therefore expressions developed for surface ionization are similar to those for thermionic emission. in the last section of this introduction these expressions will be presented. riany organic molecules have been studied in the surface ionization experiment. It is these molecular species which are of importance in this manuscript and, therefore, this description of surface ionization will primarily focus upon these. W W A number of excellent review articles can be found for positive and negative surface ionization involving theory (7, 26-28) and experimental results (7, 29-3l). This description will be divided into two sections; positive and negative surface ionization. The theory of positive surface ionization (PSI) will first be described using simple models. These models usually involve beams of atoms hitting a hot metal surface. A more complex system, as will be described, exists for molecular systems. The phenomenon of surface ionization can be divided into three distinct stages: the arrival of a species to the surface; adsorption; and evaporation (desorption). Here adsorption refers to a condition in which the adsorbed species is in charge and thermal equilibria with the surface. The evaporation of the species determines its charge. Simply, an electron competes between the surface and the desorbing atom or molecule. For atomic neutrals a simple relationship has been derived which discribes the efficiency of surface ionization. This equation is the well-known Sana-Langmuir relationship given in equation i3 (32-33). A simple and clear derivation of this expression can be found elsewhere (34). N+IN = [( l -r.)w‘/( i -r°)wolexp(O-i/kT) (l3) 20 in equation i3, r, and r0 are the reflection coefficients of the positive ion and neutral, respectively. The terms w, and w0 refer to the statistical weights of the ion and neutral respectively. When 0 and i are expressed in units of electron volts (eV) equation i3 can be expressed as follows. NJN - [( l -r,)wi/( i -r°)w°]exp[l i600(¢-i )lkT] (i4) Using equation l4 we can calculate the efficiency of ionization for several alkali metals. Consider the system of an alkali metal placed on a tungsten filament and heated to a temperature of i500K. Typical values of l for lithium, potassium, and cesium are 5.38, 4.34, and 3.90 eV, respectively. The work function of tungsten is 4.5 ev (9). The reflection coefficients r+ and r0 are zero; the ratio of statistical weight wilwo is l/2 for the alkali metals. Using these values in equation M the percent of species leaving the surface as positive ions is 0.063, 633, and 983 for lithium, potassium, and cesium respectively. Equation l3 applies only in the absence of an electric field. The thermodynamic equilibrium between an atom and the surface is altered so that equation l3 must be modified. This is similar to the case in which the Richardson-Schottky formula (equation 4) is necessary to describe the thermionic emission of an electron in the presence of an electric field. The electric field (F) reduces the energy difference between the work function and 21 the ionization energy of the atom, i.e., the field opposes the attraction of an ion to the surface. Equation l3 can then be altered to give the following approximate expression. mm as i( l -r,)w'/( i -r°)wo]exp[(¢-I+(e3F) ' ’2»le (l5) This equation is only an approximation since several terms which are dependent upon the polarizabllities of the atom and ion are neglected. if the electric new is less than 105 V/cm this approximation is valid. Thermodynamically, surface ionization of an atom can be represented as follows: EBQCESfi HERE}. i‘is -. l1 (EA, +3eV) (i6) l1-m*+e’ (+l) (i7) e" -' es" (0, -5eV) (iii) if .. "5’ (El, -3.6eV) (i9) Equations i6-l9 describe the overall process ”s —' "3+ + es- where the subscript 5 refers to the surface and the absence 0! any subscripts denotes a gas phase species. The term El is the heat of adsorption for an ion. El can be estimated using the following 22 E. = 3.60/rl (eV) (20) where r‘ is the ion radius in angstroms. Equation 20 is derived from the fact that an ion is attracted to a metal surface by an image force equal and opposite in sign to the ion. The image force obeys Coulomb's law. Equation 20 is only an approximation since it considers an ion as a sphere of radius r.. This approximation is better for atomic ions (e.g., K’) than polyatomic ions since ri's for the latter are more difficult to obtain. In equation i9 r1 is assumed to be i angstrom. From a thermodynamic standpoint the highest ionization energy (i) that the system will allow is 5.6 W. This value is above all the ionization energies for the alkali metals. To this point we have only considered the Saha-Langmuir equation for atomic species. This equation determines the ‘efficiency' of ionization of atoms adsorbed on a surface in which equilibrium has been established. A more useful expression would be an equation for the current generated due to surface ionization. We will call this value 's- Consider an incident flux of atoms "0 to the surface. Evaporating from the surface will be positive ions and neutrals characterized by the fluxes N, and N, respectively. We will define two new terms, the ionization efficiency at, and B, the ionization coefficient. These terms are defined as 23 a ' NJN (2|) 3 - mm, (22) if the atomic flux No is equal to the evaporating flux (steady state condition) then No - N t N, and o: and B are related to each other by the relation B =(l + we)“. (23) The current can then be expressed by 's - eNoBZ (24) where 2 IS the area 0' the emitting surface. Using equations '5, 23, and 24 the equation for is becomes 13 . enozm + Cexpu-o-(e3rfl’2/m] (25) where C equals (i-ro)wol(l-r,)w‘. This equation can be simplified if two cases are considered. if the ionization energy is high in relation to 0 then [l-O-(e3F)”2]>>kT and equation 25 becomes ' 2.4 is . eNoZ(l/C)exp[(0 + (e3F)'/2 - mm. (26) For the second case, where ionization is favored (0)”, the term [0+(e3F)"2-ll is much greater than kT and equation 25 becomes is s elioz (27) and 's is practically independent of T and F in their possible ranges. Up to this point only homogeneous surfaces have been considered. This is not usually the case. For most metals the surface is comprised of several different crystal faces. The work functions of each of these faces can be different, e.g., for tungsten there is a Li) W difference in work function between a (l i0) face and a (l i6) face (26). Frequently, inappropriate use of equations i3 and 25 lead to erroneous conclusions. Consider an inhomogeneous surface containing a collection of unique regions 2K each With a work fUhCUOh 0k. Equation 25 can be rewritten to account for the inhomogeneity of the surface; then is - zeuozm + Cexpui-o-(e3n"2llkm. (28) For the case when (l-0k(max)-(e3F)'/2) is much greater than kT, equation 26 becomes 25 is - eC*exp[(¢* + (e3i=)"2 - mm (29) where c* - (i/C)N°22kexp[(Ok-O*)II(T]. (30) The work function 0* is called the effective work function and lies between °k(min) and °k(max)- its value depends upon the number of regions with a work function 0". The temperature dependence of C* is much weaker than the exponential term (0*tie3F)'/2-l)lkT and a plot of is against i/T will be a straight line. This seems contradictory since C" contains an exponential term, however, the sum contains terms or opposite sign, and, thus is much less temperature dependent. When (¢k(m.n)+(e3F)'/2-i)>>kT equation 27 still holds and it is independent of temperature, field Strength, and work function. in most cases I will be at a value between [°k(min)*(°3n”2] and [9k(max)*(°3n”2l and. therefore, is can show increasing or decreasing behavior against i/T. For this case 0* becomes complex and differs from equation 30 since two summation terms must be included. One summation term includes those areas on the surface where (l-Ok-(e3F)"2)>>kT applies; the other term is for those areas in which this relation does not apply. A graph of is as a function of UT may deviate from a straight line, being 26 either convex or concave. Deviations will depend upon the temperature dependence of C*. i'iost surface ionization experiments are carried out on non-homogeneous surfaces. Often this point is neglected and erroneous conclusions are made. The summary of surface ionization results of atomic species on inhomogeneous surfaces can be summarized briefly here. The Saha-Langmuir equation (equation l3) cannot be used to describe surface ionization of atoms on inhomogeneous surfaces. The work function determined for inhomogeneous surfaces is the effective work function 0* and can be shown to be dependent on temperature, electric field strength, and ionization potential of the atom (i.e., O - f(T,F,i)). importantly. 0* does not have to agree with Oe" where 09* is the effective electronic work function. Finally, to calculate reflection coefficients, 0* and C have been confused (i i). As indicated by this description, these two quantities cannot be considered as equal. Atomic species, once they attain equilibrium with the surface, can either desorb as an ion or as the neutral. Molecular species are more complex. They can form i species on the surface each with some probability of desorbing as an ion. Consider a molecular flux v arriving on a surface with i effective fluxes v', each producing species of only one kind. The relation between v and v. is then v, - y‘v (3i) 27 where Y. is a coefficient dependent upon the molecule and emitting species. This coefficient describes what ratio of the incoming flux has been made suitable to be considered for the ionization of species i. This term is dependent upon temperature and the electric field strength. We treat the ionization of each ith species separately from any other, i.e., the surface ionization of one species is independent upon the ionization of another species. Much like before we can define the ionization efficiency as a, = iri’lv,o (32) and the ionization coefficient as p, - vi’lv'. (33) The current then can be expressed by is - evyip'z. (34) Using equations i5, 3i, 32, and 33 the equation for is becomes is = evy,2/(i + c,expi(l-o-(e3r)" 2)!le1. (35) As before, this equation can be simplified if two cases are 28 considered. if the ionization energy is high in relation to 0 then [i-O-(e3F)i/2l»kT and equation 35 becomes is~ evy'Z(i/C)exp((0 + (e3F)”2 - i)/kTi. (36) The second case, where ionization is favored (ON), the term [0+(e3F)i/2-il is much greater than kT and equation 35 becomes is . evy'Z. (37) in equation 37 is is still dependent upon the electric field strength and temperature since y, is dependent upon these parameters. As for atomic species, the surface ionization of molecular species on inhomogeneous surfaces must be expressed differently than in equation 35. The parameters 0. y', and C are replaced by the effective parameters of 0*, Y', amd C.*. The coefficient y, must be considered here since, in most cases, a molecule may 'react' differently on various patches on the surface. For the surface ionization of molecules on inhomogeneous surfaces we can simply replace "0 in equation 28 with vyf‘ and follow through with the same reasoning to describe surface ionization of atomic species on inhomogeneous surfaces. As an example, for the case (I-0-(e3F)"2)»kT equation 35 becomes 29 is . eC'*exol(0*+(e3F)"2-lille (38) where c,* = (i1c,)vy,*2(zkexpl(ok-O*)/kfll. (39) The terms in equations 38 and 39 have the same meaning as before. Equation 34 only considers the current for one species. The total surface ionization current, ist°t, is the summation of all currents from each respective ith species. The total current is then istot - Zevy'p'z. (40) Equation 40 is therefore a complex function since each neutral-ion pair will have a set of y. and 8. values whose temperature and electric field strength dependence will be different. Since this is the case, generally each neutral-ion pair is considered separately. Here, mass spectrometry can be a powerful tool since one ion can be selected and monitored independently of the others. Thermodynamically, the positive surface ionization of organic molecules can be represented by a series of equations similar to those in equations l6-i9. The overall surface ionization process 30 for a molecule AB can be represented by ABs-tAs’ + es + as. The subscripts used are the same as before (equations i6-i9). Equations leading to this result are listed as follows. ERQCEES ENERE! ABs - AB (EA(AB),+0.2 eV) (4i) AB .. A + B (D(A-B), +2 eV) (42) A .. A’ + e“ (i) (43) e" -. es' (0, -5 eV) (44) B .. as (we), -3 eV) (45) A” -» As‘ ((5.. -3.6 ell) (46) Typical values of EA and D(A-B) are used here. A molecule usually is physically adsorbed which leads to the low EA value in equation 4i. Typically, a part of the molecule (here 8) is chemisorbed on the surface, thus, the much higher value of EA in equation 45. Here a value of 3.6 ev is used for E. assuming a radius of i A (see equation 20). in this example the ionization energy of fragment A must be less than 9.4 W on this 5 eV work function surface. The thermodynamic approach just described is a simplification of positive surface ionization. The energy associated with the ionic desorption of A’ from the surface is only approximated here. in reality, for a large number of species, 3] this value is unknown. in addition, it has been shown that a correlation does not exist between the dissociation energy (D(A-B)) and surface ionization (35). This would be expected if the products are ionized after establishment of equilibrium. One cannot predict that a compound would yield more current than another compound based on these thermodynamic relationships. Some fraction of molecules striking the surface will be ineffectual, i.e., may produce an ith species but not in charge and thermal equilibrium with the surface. We have already defined a function (equation 31) which describes those species which are effective (i.e., in equilibrium with the surface). This function, v1, is determined by the rate of the chemical reaction, the surface temperature, and the electric field strength. The function V' must be determined experimentally. A number of equations and associated principles have been presented. in the preceding paragraph i-have presented some points that a simple energetic approach does not consider. This section is not meant to deter anyone from using an approach such as this, but rather to indicate possible areas where it can fail. The following is an attempt to put surface ionization theory into perspective and to indicate the types of applications where surface ionization can be useful. i.) Simple energetics give a clue as to what might be expected. The relationships between ionization energy and work function, for example, indicates the 995511111113: of forming an ion. For atomic species, ionic desorption energies can be better 32 approximated and, therefore, easier to describe energetically (equations i6-i9). in general, the surface ionization of atomic species, when compared to molecules, is much simpler system since the atom must either desorb as a neutral or an ion. A molecule can desorb, ionize, or react on the surface. 2.) When considering the work function for a particular surface, one needs to know how they were obtained and the condition of the surface. In most cases nonuniform surfaces are studied. These surfaces are best characterized by the effective work function, 0*. The effective work function is a weighted average of the work functions of different areas on the surface. How would 0* be affected if a change is made to the surface? Would certain areas be unchanged and others altered remarkedly? Questions such as these are not yet fully answered. 3). Perhaps the largest unknown in the surface ionization of molecular species is the minimal amount of knowledge of surface chemical reactions. The fate of a molecule on a hot surface determines the extent of surface ionization and what ionic species are produced. Little is known as to residence times of species on hot surfaces, a parameter needed for a kinetic treatment of polyatomic ion desorption characteristics. Presently, the effective flux v, must be determined experimentally. Perhaps with further experimental results this parameter can be calculated union. in order for this to be accomplished the chemistry of a molecule on the surface must be totally understood. 4). Even though surface ionization has been known for some 33 time the many uses of this technique have not yet reached their full potential. The experimental determination of values found in Si expressions are the basis for the physical investigation of many properties of solids, ionizable species, and fundamental characteristics of the interaction between a species and a solid. Surface ionization can be used for the determination of ionization energies, electron affinities, desorption energies of neutral and ionic species, solid phase transitions, and the study of heterogeneous chemical reactions. In addition to applications in physical chemistry, surface ionization is useful in analytical chemistry. Here the theory is sufficient to provide a foundation for the analysis of certain organic molecules and their structural elucidation. Host of the early surface ionization experiments considered only atomic species. More recently organic molecules have been investigated. These investigations include nitrogen- containing compounds (29, 35-48), phosphorus compounds (29, 49), and arsenic compounds (49). Host of these studies have been done on tungsten or oxidized tungsten surfaces. Other solids such as molybdenum, nickel, rhenium, and platinum have been used (50-52). Experimental and theoretical aspects of surface ionization began with the emission of positive ions. So far this introduction has focussed only on positive SI of atomic and molecular species. l‘iany principles of positive SI can be applied to negative surface ionization (NSI). The following is a brief discussion of NSI which will rely on the previous discussion of positive Si. 3.4 W if we consider the following equilibrium of an atom A on a hot surface A + es’ :2 A” . (47) an analogous equation for NSI can be derived which is comparable to that for positive Si (equation I3). 0 - N’m - [(i-r‘)w‘/(l-r)wllexp(A+(e3r)"Z-O'MT) (4s) . where all the terms used here are as before except they relate to the negative ion. The effective work function 0’ does not necessarily correspond to 0* for positive Si but is defined as an average work function effective for producing the negative ion. if the Schottky term ((e3F)”2) can be neglected (as usually is the case for low field strengths) and if 0' = 0, equation 48 becomes the well-known Saha-Langmuir equation for NSI. cl" = (w‘lw)exp[(A - 0)!le (49) in the derivation of equation 48 the reflection coefficients r and r’ are assumed to be zero. This assumption has not been verified either experimentally or theoretically however, for positive 35 surface ionization r+ and r are considered zero in the temperature range above i000 K. By analogy this has been carried over to NSI. The statistical weights of ions and atoms can be calculated using an expression derived by Scheer and Fine (53). As mentioned before w’lw has an approximate value of 0.5 for the alkali metals. Using the equation in reference 56, w‘lw for halogens has an approximate value of 0.25. An expression for the current, is-, produced by negative surface ionization can be obtained in a similar fashion as positive Si. The ionization coefficient 8‘ is defined as B- ‘ [i + (Ila-)l-i (50) and the current is“ can then be expressed by is‘ - eNoB‘Z. (5i) Using the equations of 48 and 49, equation 5i becomes ls‘ - eNOZ/[i + C'exp[(A + (e3F)"2 - «mall; (52) the NSI analogue of equation 25. Here C' equals (l-ro)w°/( i -r')w|. The negative surface ionization of polyatomic species is analogous to positive Si. Since this is the case the ionization 36 efficiency and ionization coefficient can be written, respectively, as or - (u'm) - (”menu-0° + (e3F)"2 + A)/I(T] (53) and n‘ = (N'm) - (mist/(l + (3’)] (54) where v. and 1‘ have the same meaning as before. The NSI current can be appropriately expressed by the equation is' - eNOB‘Z - eNovy.Z[a’/(i + (3‘)]. (55) A description of any simplifying approximations will not be considered here as was done for positive Si and for equation 35. Some authors (28) prefer to include a parameter 0 Into equation 55 which accounts for the 'sticking' probability of a molecule to the surface. It is difficult to evaluate a due to the lack of sufficient thermochemical data and often a is assumed to be unity. This is the case when the coefficient y. is determined. It seems arbritrary to set aside this parameter (0) since, in reality, y. sufficiently encompasses the process of forming a species capable of ionizing on the surface. At this point in this discussion of "Si it should be clear that 37 the principles of NSI parallel those of positive surface ionization. it is unnecessary to parallel each positive Si equation with a corresponding NSI partner. i have defined a number of work functions to this point and perhaps it is now necessary to reevaluate this parameter since it is crucial in all expressions for surface ionization and thermionic emission. Below is a list of work functions used thus far in this discussion. 0 work function (thermionic emission of electrons from a homogeneous surface) 0' work function for area i on an inhomogeneous surface 0, positive ion work function for thermionic emission 09* effective electronic work function (inhomogeneous surface) 0* effective work function for positive ion formation (inhomogeneous surface) 0' effective work function for negative ion formation (inhomogeneous surface) 09' effective electronic work function in presence of a sample on an inhomogeneous surface The energy required to remove an electron from a surface is given by the work function. Should this energy be equivalent for 38 all processes of surface ionization and thermionic emission on a particular surface? it has been determined that these work functions listed above are not equal to each other and in some cases are remarkably different. For example, when tungsten (0e* 8 4.48 eV) is exposed to tetracyanoethylene 0,; was found to be 2.70 ev (54). Consequently, as is sometimes the case, the approximation that 0e. ~ 0; is made for the calculation of the ionization efficiency or the ionization coefficient. Such an assumption can lead to a substantial error in these calculations. Recently, researchers in surface phenomena have begun investigating the role of surface processes in regards to changes in work function (55-56). it should not be suprising that all the various types of work functions are not equivalent since each is associated with a particular surface process. For NSI an electron is transferred to an adsorbed neutral. in positive surface ionization an adsorbed species 'loses' an electron to the surface. For the case of 09' and 0e* can we expect a surface to remain the same in the presence or absence of an adsorbing species? All these processes are different so one would not expect one common work function. In this text I will refer to the work function as that energy involved in the transfer of an electron to or from the surface for any system (e.g., NSI, thermionic emission, etc.). Comment 2 made earlier in this section should be reinforced here. Often, the term 'work function' is used loosely. When a 'work function“ is 39 calculated, it should be made clear as to the circumstances under which the value was obtained. An energetic treatment of NSI can be made in a similar fashion as was done for positive Si (see equations 41-46). Here, we will consider the production of a negative ion 8- in an overall surface reaction given by es + ABs —. As + Bs". (56) The series of equations representing negative surface ionization are EBQCESE ENEBEI ABs .. A8 (EAiAB), +0.2 eV) (57) AB 4 A * B (OM-B), *2 eV) (58) es 4 e (0, +5 eV) (59) e + B -* B- (A, +0.6 eV) (60) a .. as“ ((5,, -3.6 eV) (6i) A -» As (E A(A), -3.0 eV) (62) When typical values are used as before (see equations 4i-46), the minimum electron affinity calculated for ionization is 0.6 eV. Here, using the same arguments for positive Si, this simple energetic treatment can be useful in determining if a species is capable of Ionization. Recently, large complilations of electron affinities for atomic, radical, and molecular species have become 40 available (3i, 57). W815 This introduction is perhaps over-zealous in regards to the discussion of the following projects. I believe i have accomplished the objective of distinguishing the difference between thermionic emission and surface ionization. in addition, 'with regard to some of the proposed equations, the complexity of these seemingly simple processes is presented. The work function has been discussed and, perhaps from this introduction, a better understanding of this parameter can be perceived. This introduction may prove useful for those who 'follow in my footsteps‘ by aiding in the investigation of other solids for surface ionization or as thermionic emitters in the particular applications which will follow. When mass spectrometry (M5) is used for chemical analysis, molecular weight, functional group, and structural information can frequently be obtained. Of these, molecular weight information (i.e., the detection of a molecular ion, I1") is usually the most valuable, because it determines both the elemental composition and the 'molecuiar boundaries into which the structural fragments indicated in the mass spectrum must be fitted“ (58). Unfortunately, for a number of reasons, many compounds do not form detectable amounts of the molecular ion following electron impact (El) ionization (58). Chemical ionization (Cl) is another common technique available to mass spectrometrist (59). Chemical ionization is considered one of the 'soft' ionization methods, since it usually induces less fragmentation that does El, simplifying the mass spectrum. Also, the protonated parent, I‘lil’, is frequently formed - i.e., molecular weight information is produced. liunson and Field (60) were the first to demonstrate the analytical applications of Cl by using CH4 as a reagnet gas and generated spectra in which the analyte molecules reacted with Ciis+ and C2H5* (the 'reagent ions') In the gas phase. Since this work, many other species have been used as Cl reagents. Proton 4i 42 transfer reagents such as H3’ will react with most compounds since the proton affinity (PA) of H2 Is low (approximately iOI kcal/mol). lsobutene has a much higher PA (l95 kcal/mol); therefore I-C4Hlo Ci only protonates molecules with PA's greater than approximately i95 kcailmol. Thus, the proton affinity of the species used in CI is the 'working variable' which can be adjusted to provide a 'universal' or selective ionization method. Charge exchange (CE) chemical ionization is another Ci alternative. In CE, ions such as Ar‘, Nz’, or CO’ react with analyte molecules by electron transfer. As in proton transfer CI, varying the reagent varies the response-in CE, the variable is not proton affinity, but instead ionization potential. The popularity of CI is due in part to its compatibility with most NS techniques. Unlike other ionization methods such as field desorption (which usually requires a double focusing magnetic mass spectrometer) CI can be performed with most types of mass spectrometers. Also, CI is compatible with chromatographic inlet systems (i.e., gas chromatography (SC/HS, liquid chromatography LC/HS). in spite of the many advantages of Cl, there are a number of disadvantages. An ("HY ion is not always observed In Cl (as is often the case with alcohols) and a parent ion is not always observed in CE. Also, it is difficult to catalog Ci spectra because they can be strongly pressure dependent. Pressure effects arise in many ways. Frequently, more than one reactive species is 43 generated in CI sources (e.g., (ii-Is+ and C2H5* in methane CI, CO’ and (CO)2’ in carbon monoxide CE). Tile relative amounts of these reagent ions vary with pressure. Also, product ions can be collisionally stabilized at high pressures. Thus, small pressure changes can dramatically affect the number and relative intensities of ions in a Cl mass spectrum. in part, this section of the dissertation discusses the use of metal ions, here K’, as CI reagent ions. Studies presented here show that K’ reacts with most types of molecules (N) to from the 'pseudomolecular ion' (l'i+l()’; i.e., molecular weight information is easily obtained. By the technique described, metal ions can be generated (with a selected kinetic energy) in the absence of corresponding neutrals-providing the basis for a relatively low pressure Cl method. Recently, a large body of literature has appeared concerning the gas phase chemistry of metal ions with organic molecules. Both transition-metal and alkali Ions have been studied (GI-63). Transition-metal ions can be generated in the gas phase by El on organometallic compounds (64) or by laser desorption/ionization from solid surfaces (65). The chemistry of gaseous transition-metal Ions with organic molecules is rich, often producing products which provide both molecular weight and structural information. In contrast, alkali Ions generally react by simpler mechanisms (GI-63). Ions such as Li’ form electrostatic complexes with many analytes such as il-donor bases (e.g., benzene) and n-donor 44 bases (e.g., ethers). Adduct ions (also referred to as 'catlonized molecular ions') of the type (Ii+Na)’ have also been reported (GI-63). Alkali metal ions appear to be useful CI reagents for obtaining molecular weight information due to the low lE's of these metals. In the case of the (l‘is‘K)+ adduct, for example, the charge presumably remains localized on the metal. Thus, the decomposition pathway with the lowest activation energy is simply the loss of K‘, instead of the fragmentation of the attached molecule. The probability of a metal Induced reaction relative to attachment appears to decrease as the metal is varied from l.)+ to k’ (6i-63). As mentioned earlier, Blewett and Jones (l7) described a convenient method for producing metal ions in the gas phase. l'ietal ions are produced by heating and biasing a glass or metal oxide. The glasses are usually metal oxides in an aluminosilicate matrix. Oxides such as In203 behave Similarly. Thermionic emission sources were reported for l7 metal ions. Those for alkali metals are particularly copius emitters. This part will focus on uses of thermionic sources of K’ ions. A simple, low cost method is presented for producing variable energy metal ions for use as CI reagents. The method is adaptable to any mass spectometer with a Cl source and direct inlet probe. In addition, alternate modes of operation are presented which will provide additional dimensions of mass spectral information. 45 LLEXEEBIUENIAL All experiments were performed on an unmodified Hewlett-Packard 5985 (SC/"SIDS, equipped for manual tuning as described elsewhere (66). Thermionic emitters were mounted on a fabricated probe assembly and inserted into the Cl volume of the Ion sourve through the direct insertion probe inlet (see F igure 5). The tip of the thermionic emitter was made by inserting '0.007 in. (0.i78 mm) rhenium wire through a 0.5 In. (i.30 cm) length of two-holed ceramic rod of 0.047 in. (I.i9 cm) diameter. The ceramic rod gives mechanical stability to the probe tip and insulates the rhenium wire from the ion source. A small loop of the rhenium wire extends from one end of the ceramic rod. The two free ends of this wire are spot-welded onto insulated leads housed within the probe assembly. 'Thermionic K’ glass' has the composition IK20:iAl203:25i02. Its preparation has been described elsewhere (I7). The wire loop on the probe tip is dipped Into a molten mixture of the thermionic glass, withdrawn, and allowed to cool. This procedure produces a bead of glass (approximately 0.030 in. (0.76 mm) in diameter) on the wire loop. This bead is small enough to fit through an existing hole in the CI source volume and was sufficient for at least 40 hours of continuous operation. A 0-4 A, low-voltage power supply is used to resistively heat the thermionic glass bead on the wire loop. A 0-20 V, low-current power supply biases the bead positively with respect to the CI volume. Biases from 0 to *20 V were used in these .0958 :2 «4.3058303 «an: 2: E 2.6.... csmflem 2:28.35 3 8952: 2352.3 .m 9.52... CON + 2 Oh... AmE:_o> H8 6.522, U 2 8.8m 8. 9:23". 38 co «Em - mmmm a: i..- _ Hi. - P _ M. 2! "F Jul: Jess n AEQDW bga ~02...— ““wa 32a 3:8 as... 4.85. 6.32320 —bbbb—bbPhI—bIb—Pbih—DDPP—bbt—brb P— bP—PDDI—bE—PDII 1 .2593. mm 4v. servo. 8225 -m misusing aAiloiaa 50 .55“ 1563 .e 52.88 an. s 2%: ON. 0: 00. 0m 0m 8 om . Om Ode on Eb—p pp—ptp—Cppr-pp—n-pu—p--—-b-—phbb—-pup—‘pp— pupp—P-pp—pnpp—bp-n—nh:— pub—bub- n: r +3. 50.1.42 mm 9v. i 62:48 .95 .355 Allsualui aAllblaa 51 6.33322 3 52325 UL. .o 952“. 0m 00 Ch 00 . on 0? On ON 0. O b P.» —bp up >P>-_bbhp_bnhh—bbpb—thh—pnp b—bbbb—bbn .— pub—pbph—h-np_-bbb—bbbb—bb-p—upbn—ppp- — 8 +2 + znzuu. mm AZnINUV £755.84 Alisuaiui aAiloiaa S2 .328... :5: .e 2:388 8.3. .o 8.5: On. ON. O: 00. CO GO .ON CO on 0? On mu p—pnpp—pP-p—-Ppp—p-p—-bbh—-pup—ppnp—phnp—nppp—b:-—bpbp—~:-—b-p—phpb—nnpn—p-nn—npp.— bub—n:- f bu. +3. +~omxes an .x - .~0o I so. 228< 35.”. Allsuaiul aAuOIaa 53 Table I List of Compounds Studied Using the K+ Emitter Probe in the K’Cl Mode. ME MY DIHEBJQNE Pyridine yes no Isobutyl chloride yes no Benzene yes no cyclohexane no no hexane no no n-propanol yes (I‘i-H+2K)’ ethanol yes no acetone yes no butyraldehyde yes no acetaldehyde yes no aniline yes no N,N-dimethylaniline yes no Fe(CO)5 no no All of the above data was obtained when the K’CI thermionic emitter probe was operated within the addition bias window. 54 some way lose its excess energy (e.g., emit a photon) (67). Addition is more commonly assumed to be a three-body process, In which a neutral molecule collides with an ion-molecule complex and removes an amount of energy, stabilizing the (l‘l+K)' complex. Ketones give abundant ("40* ions in their K’CI spectra. Figure I0 shows a log-log plot of I(H+K)’/ [I(Ii+l<)‘+l(i(’)l versus pressure of acetone (in the CI volume). Apparently, the process leading to the formation of (acetone + K)’ is a reaction which is 'second order in acetone. At the higher pressures, the K‘ concentration is reduced substantially and the apparent order begins to change. Thus, at least for acetone, the reaction for K’Cl can be written as K“ + 2C3H60 -» (C3I-I60 + lo+ + C3H60. The termolecular nature of the adduct formation which was observed is consistent with the previously reported kinetics of alkali ion adduct formation (68). Based on this, K’CI would be very suitable for GCII'IS without a splitter, since no response for carrier gas molecules will be seen, and the carrier gas can act as the third body in the analyte addition reactions with K’. Amines gave richer K'Cl spectra than the other compounds studied. Figure ii is an example of a typical K’CI spectrum of triethylamine. in this particular spectrum triethylamine exhibited no (I‘i+K)+ ion. The ions at m/z i02, I00, 86, and 72 can be identified as (Mir, (l~i-ii)”, (n-cns)’, and (ri-c2H5)*, 55 I.O E ’.-_--.‘---‘--- Z I" I - A .. I I I- J I I _ I I +5 I s / H OJ : ” t: : I, X ' ‘I 4' " I .2. - ,1 i-l _ A +‘_\ I, X " I, ‘l' I .2. ,1 H 0.0l: II : I .. I _ I I - A I - I I _ [IA / I 0.00' 1 11111111 L 1 1111111 IO IOO IOOO Cl VOLUME PRESSURE (mtorr) Figure i0. A Log-log Plot of I(I'l+i()‘l[i(li+l()* + I(K’)] vs. Pressure of Acetone (l Refers to Intensity). 56 .3853 was 603-23 ev— o>oa< m— was enom 65503502... 3 523on was: .5322... 32.58 5.x .2 952“. ~\E o: oo. om co 2. cm on ow on FIFE . P-pprp:—p:“—b-:r:iP—upup—Lb-p—ppp.—.:.—::—b.:—::— PC—E: _ _ mm 2.. _ .z «.200 +zofeu .. mm .. 4x %.100 +ZSIOU . «o. oo. 343:8: $53255» LLNI "138 57 respectively. The product ions in the K‘Ci spectrum are also the major ions in the El spectrum of this compound. This fact, along with the fact that, of the compounds studied, amines had the lowest ionization energy, led us to believe that a charge transfer process was occurring (69). That is, the mechanism could be written as follows: K‘ + n "*x e K' E (li-i)‘, (Ii-CH3)‘, (n-c2H5)*, etc. Upon further investigation with a bare heated wire probe tip (no glass bead) the same product ions were produced (in contrast to, e.g., acetone, for which no ions are generated from contact with a bare heated wire). In this case gas-phase potassium ions were absent. Therefore, this suggested that the observed product ions for the amines were a result of surface ionization. Our results found here parallel those found previously (29). The formation of the (l‘i+i)’ ion can be attributed at least in part to a gas-phase ion/molecule reaction (presumably of (l‘l-i)’) since the (l‘i+i)’/(l'1-i)+ ratio increases with increasing pressure of the amine. The products observed for the amines were greatly influenced by the bias voltage applied to the glass head. The addition product (I'I'IK)+ was seen for all compounds studied at bias voltages between approximately 0.5 and 5 V. However, ions formed due to surface ionization from amines are produced at much higher values of the bead bias. This can be seen graphically 58 in the case of diethylamine in Figure i2. At low voltages the potassium ion adduct of diethylamine is the major product. With increasing bias voltages only the surface ionization products are observed. This bias-dependent behavior can be an advantage as will be shown later. Surface Ionization studies have predominantly been done on pure and oxidized metal surfaces. In our system, the hot surface consists of the thermionic glass bead (and some exposed rhenium wire). The thermionic glass we use in our probe tip is very similar to the thermionic material contained in thermionic ionization detectors (TID) for gas chromatography. Several surface ionization mechanisms have been proposed for these detectors (70, 7i) which account for the observed signal in the presence of nitrogen-containing compounds. Chapter 5 of this thesis will be, in part, a study of surface ionization product ions of N- and P-containing molecules with the hot thermionic material in an attempt to elucidate a mechanism that explains the behavior of TlD‘s. (To date, the link between TID response and the large body of literature on surface Ionization mass spectrometry has apparently not been made). Energy-variable Cl reagent ions can be a very useful tool, since it has been demonstrated that ion/molecule reactions can be made to occur (i.e., made exothermic) by the addition of kinetic energy to the reactant ion (72). In the case of K‘, however, this effect is not observed. We do, however, see dramatic changes in the K‘Ci spectrum for some molecules as the K’ emitter bead bias varies. Low biases favor adduct formation, 0140*. As the bias is 59 $32333: .< ON :5 .23.: one an 2328 0.83 29:3 9.2:: can 9.593.... 65.53505 .3“. 8:. =25 .m> =3 5.5.435 NEE—AN.— Nxfizzc :3 3.59.5 _ be ac:— .N_ 9.53“. mbo> .msm aqua $5.5 .x on 6.. on em - q u q a q u - q q u c a u 1 q u a 4 .3201 no. "on a): e +2223. .8. es .. .2256. rm: ”\e 4 2:16 mz_2<._>:E_o O. O x .N O 10¢ low low CO. (8911 +(ZDI +(Zil) I i! NOi lanOOl‘id) I 6.0 raised, the probability for this process dramatically decreases. For many compounds, ‘fast" K’CI gives no products. As was seen for amines, if surface ionization occurs, higher biases are required to remove these ions from the surface. 2-propanol was found to participate in both gas phase and surface processes. Figure I3 shows a K‘CI spectrum of 2-propanol (low bias). The predominant ion is the (li+i<)“ at mlz 99 and an (H+K-H2)' at 97. There is also an ion at higher mass, m/z I37. The m/z 97 and 99 ions decrease as bead bias increases (Figure i4). The intensity of m/z i37 increases as bead bias increases, typical of a surface process. An ion of m/z i37 cannot be formed by I(+ and 2-propanol in a simple process, since the mass of these two reactants only totals 99u. Initially, we identified this ions as (csiimomr. That is, - by some surface process, a six-carbon species was being formed from two propanol molecules. Closer inspection of isotope ratios suggested that m/z I37 contained two potassium atoms, i.e., (C3H7O+I<2)’. Presumably some surface process such as C3H70H + K _. C3H7OK + H occurs. The C3H7OK then complexes with a k’ ion emitted from the surface. Similar species have been observed in other techniques such as fast atom bombardment (FAB) ionization (73). The 'sampling‘ of species formed on the bead surface by K‘ ions 6] .3353 85 £03.32 ev. 55.3 comma coon 5.3 .ocaaok—iw so Eaguuoam Sex .n. 0.50: ~\E P b—b-bp—pp-p—PmPp—pb-p—ppbp—p pP-r-Pb— p-Pp—b-P- . 0m 00 ON 8 On 04. o: oo BEN—:z_::_3__::_ E , amxoexmo. mm :x40mzn2 hm .vAV. +00 Inc. 501mg 658.com. so Holt. Misuaiui angioiaa 62 I.O- v V .9- V .Be ;; .7r =fl g I: _ i. 5* O sin/z 99 gg I-m/z 97 g f. .5- . v Im/zi37 : s, - H 9,. .3I- I .2 - it . Bead Bios. Voiis Figure i4. Plot of l (Product Ion)/[i(m/z 98)+l(m/z 99)+I(mlz l37)i vs. Bead Bias. 6.3 which are emitted as discussed in more detail in the case of peptide analysis and is the basis of Chapter 3 of this dissertation. In summary, K‘ appears to be useful in providing molecular weight information, except with certain types of compounds, for which ionization on a hot surface is known to be a facile process. W’W 5mm Five other alternate uses of the probe will be described here. W in SOIUUOII. crown ethers have proven to be useful species for kinetic and thermodynamic investigations of ion-solvent interactions (74). In the gas phase, such species can be studied in the absence of solvent (75). Crown ethers form strong complexes with alkali ions in solution. With the K’ emitter, crown-K‘ complexes can also be formed in the gas phase. Figures I5 and I6 show the K‘CI spectra of i2-crown-4 (-(C2H4O)-) and Its linear analogue, triethylene glycol dimethyl ether CH3(OC2H4)3OCH3. Studies of competitive ligand exchange of such species may provide thermochemical insights into such metal-ligand complexes. W In the context of GC/I‘lS, the production of both El and CI mass spectra from a single sample component is advantageous since both the molecular weight information common in CI and the structural information of El are 64 .32....3 was €63.22 6.5.3 m. .35 noon .vic3cguiw. .6 Easuuoam .Uev. .m. 0.59.... CNN OmN CNN O.N Om. Oh. Om. bpprh—ppup-yupp-—-P-bP>ppPP0ifl—WbbP-pnP-Cppbppp—DP-pr-pphtbpb ma ... +3. +6.: some .... On. On. 0: 0m 0.. On on _PIPrI-PIDI-Ibbb—DPDDPPPbP-DIhb—PPDP-PFPP-PPPPhrpbb-DDDP-bDDL _ . m mm r +¥ r 3m. .2850 05.8: “..._o._. EOm "scammed. 26.14068 Yezeué lNl "138 lNl "EH 65 .3853 we... 5:83. 55.3 a. 8:. 83 .85.“. 152:8 .85 2.2.53: 3 22.8%.. a... .2 23: ~\E CNN CON CMN C_N Cm. Ch. Cm. rP-Pppb-pp—ptp—b...—-p-—.P-.hbpbp_--p—pppp—pbPan..p—pppp —_ v T ta w +3.3... some. . On. On. 0.. 0m ON 00 CM bin-p—bP-p—bppthpr—nbnp—P-nb—P-hp-php-hbhph—hbun—bpbn—p-nb I lNI "138 mm .x .. 36.. $5.50 958... So... EOON “Essen: 5.5.”. .3525 .85 823.65 lNl "138 66 available. A number of approaches have been demonstrated. Instruments utilizing dual sources, dual ion beams, and dual detectors have been used. Also, combination EI/CI sources have been used. In the latter, sequential El and CI spectra can be produced (alternating for each scan). The K‘Cl probe, when negatively biased, produces electrons. Apparently, electrons are generated from both the thermionic glass and the rhenium wire when heated. Thus, the bias can be sychronlzed to the scan flag of the mass spectrometer data system such that, for one mass scan, the bead bias is at, e.g., 2 V (K‘CI mode), and the next scan at e.g., -70 V (El mode). To demonstrate this, l5-crown-5 was chosen, since the EI mass spectra of crown ethers are very similar and characteristic of a crown ether, but with essentially no molecular ion to identify the particular crown ether under study. Figures i7 and i8 show the respective EI and Cl spectra obtained by using the K‘ emitter in the CI volume. In an El mode (here, bias- -23 V) the spectrum agrees well with the expected El spectrum (76). In the K’CI mode (bias of +2 V), the (i5-crown-5 + K)’ adduct is prominent, providing the necessary molecular weight information. This simple technique can extend the Information available from a mass spectrometer without costly redesigned sources, in fact, with no instrumental alterations at all. W When a gas chromatograph is coupled to a mass spectrometer, the 6C detector is removed in "lost cases. liany detectors, such as the thermal conductivity 67 .32....3 53.9.3 2.. 55.3 was anon .micioguim. .6 855.925 .Cev. .h. 0.59.“. CmN CCN CVN CNN CCN CC. CC. CV. , .i _ m fix. o 8 .. + (OIM . 03 On. ON. C: CC. CC CC Oh CC Cm CV Cm ON C. C .p. uppppP—urpp—ppr—ppP-ppp-Pb-pb-Pb-t—ppp-—P-pp «pup—nppp—PPbP—Ptibfil _ . F mm. fix mm H. .v. . .46.. .2250 9:86... 3.8.803 "83.3.5. .moongo. 9:29.99 .525an B .v. lNl "BB lNI "lat-i .38 2.: 5 2.5.2, 8 2.. 3 32.3.. 5.3 > MNI .63 «En Baum .mictoguim. .6 535025 .m .o. 0.59.“. 68 ~\E Om. CV. ON. GO. Om Om Ov ON .b-——-b.—»P-—p:-_L.-.-—.-.-._L.ipim——-ppb4—. I Mn. .0. - on i me - Aio mm . O JOOIV Pi 0L we . .oa< was amen aaooxm a. 0.59..— 5 ms 256 2: 22:33 65:2 :5 2.33. a be 23...: use: 7— a no .1. 3 mtg-£2 2: :— caa mtxaexxum Eek. 0053.5 5.2.3:. :2 .23 can «529.3525 «on: of .3 0.59“. we: hp» pp»Puppppprpbbp-pp-prE—ptpbbeP-pbbpubbb-prPrprbbthp-bbppppp .E. IIIIIII bL-pppbheppp-ppppnpppp inbb—rpppbhbtn pub-Pb-pb-rbb-Pbtp-P-Ppbb-nbntb- .3. +3205 3 cocoa :. 250... .oz EEooSEozu v.32 'I'I'I L.2 $563 3 88020525 anus. 10133130 asuodsaa 72 .1. --. U... -... .e... .0' o..u..- We have observed that, when a solution containing a nonvolatile/thermally labile compound is evaporated onto the K‘ glass bead, ions representative of the compound will be generated when the probe is heated. These ions usually last for lo-lS seconds after the bead is brought up to operating temperature. This technique was investigated by using di- and tri-peptides as analytes. We observe that (l) most of the ions formed contain K", (2) (WK? is almost always present and is also the highest m/z product formed, and (3) low-energy decompositions of the peptides occur, for example, H20 and (:02 loss occurs and an (WK-(:02? or (ri+l(-l120)+ is observed. While a number of possible mechanisms can be considered, a useful working model is as follows: A molecule thermally undergoes a decomposition to two (or more) smaller molecules on the surface. The i(+ being emitted from the surface beneath these species complexes with these species, 'sampling' the molecules present on the hot surface. Figures 2i-23 show representative spectra of a dipeptide and two tripeptides. Note that in each case an (H+K)’ is observed. For each fragment, bond cleavage with a H shift can explain the observed species. For example, glycylglycine (Gly-Gly) can thermally decompose to form glycine. fic‘g‘nu NH2 V I - H2N\ l 04,0300» , H’c:c=o + (Hzcoou I'vcim 73 .< oN no) 29:5 958: 2.822329 3 533.5. 32,595.: .3 8:2“. Om. OO. O! ON. 00. CO CO O¢ —>P>_mj—qVP-P—pm-b —-»L»—qb_bE—Am fl—Efi s. . e3. +3839 r mm. x N .. +8 I xi): bu. LmouivIs: . . O Zr. IOTwiNIUiiuUiNIMeS ex.z. o : ~12 . .>mm.m 2.63 259 BEEN +& so 9520 32:0 Misuaiu' anuoiaa 74 .< ed as) 20.25 9.38: 65cm? 1.8.. Zora .3 533.5. contamxaex .8 0.59”. Own OOm OON 8N ovN ONN CON 8. +A~Ooix+s= e.~oz.~.:no iv. e .2. mm. - n— pl hue—abi1ual a: A411}. 0mm rrrvlfi" . . b p . _ 8. ON. . 00. oem . om ov om nxw Io-m-:o-:z/ 0 UIO \ NIUlIU w. x a .e. to. \zAw- Io- :z..2 «:0 o I T T V r r Y I I' SO.» ”.33 come SEEM +x co 9.220 3.03 3&5 Misuaiui anilDlafl 75 .< od 83 29:5 9.38: 653.222... .5»... 32...... .o 533:3. contamxfib. .nu v.52“. COM COM Ovn Own OOm OON 8N OVN ONN ..~ou-x+s: mum . ..~om:vu-x.s: . o- 08 8. 8. o: om. oo. o_m . ow $3.5... . mm. om mm . mm. . . mm. . .8. m: w mow mm. mm . 10‘- .uzo o o a 0-5 :z-o-~:o- -.:z o- :u- :9. 5.»:0 o «I... are .>Om.~ "85 comm. SEEM. ev. co 3.5.2.62... .320 .58.. amoiaa . Ausuaiul 7.6 The spectrum of glycylglycine obtained in this way is fairly simple. A similar process can be postulated for the tripeptide glycyl-L-proyl-L-alanine (Ely-Pro-Ala), leading to the product at ill/Z I93. ‘ cu 9 xmz‘cuz 9 I 2‘04, ,M-iz H2NCH2'é-N\ I ___, HZNCHz-d-hk / /C\-CH2 ’Ic-cnz + Hflcoou °=C M H . I): x 0 ,NH CH \ H3! COOH (M-C3H7N02) Following this decomposition, i(+ complexes with these surface species to produce [(I'i-C3H7N02) + K]’ at m/z i93. A similar process generates the m/z 226 product for Leu-Gly-Phe. Note that in all three cases, cleavage of the peptide linkage >C‘H o:c: 5 L \NH / dominates the decomposition pathways. Thus, the possible use of this technique for sequencing such species is worth pursuing. Note that DL-leucylglycyl-DL-phenyalanine (Leu-Gly-Phe) has a much more complex spectrum at lower masses. "any of the products do not contain K’. Apparently, some bases are more susceptible to surface ionization. Over a dozen di- and tri-peptides were included in this initial study with successful results. in all cases, an (l"l+K)+ was observed, with cationized species indicative of low energy decompositions being common as well. 77 In some ways, this approach to analysis of nonvolatile/thermally labile compounds may be preferable to alternate methods such as FAB ionization, for a number of reasons. The most obvious advantage is cost. Secondly, the absence of a matrix (glycerol in FAB) and its associated spectral interferences is an advantage. Also, FAB frequently only produces an adduct ion of an analyte, (ii-HY, with little structural information. Presumably, if this experiment could be performed with a K‘ bead which emitted potassium ions with no heating, peptides would only show an (n+K)‘ ion. The temperature at which potassium emission occurs appears to induce low-energy decompositions of such compounds into smaller molecules. The emitted potassium ion then samples these to generate simple, useful spectra. In some ways the technique is analogous to FAB or SIMS-the major difference being that K’ is generated from the surface on which the species is deposited. The last use of the potassium thermionic emitter probe (e.g., the analysis of nonvolatile and thermally labile compounds) was further investigated using a variety of compounds ranging from pharmaceuticals to saccharides. These investigations are included in Chapter 3 of this thesis. This technique was also applied to polymeric systems and will be presented in Chapter 4 of this work. In Chapter 3 further comments on the mechanism of this process will be included. The initial investigations of this probe to the analysis of nonvolatile/thermally labile compounds is included in this part to serve as an introduction to the technique. The number of ionization techniques used in mass spectrometry is constantly growing, allowing for a wide variety of compound types to be studied. Early mass spectrometric analyses using electron impact, chemical .ionization, field ionization (Fl), and surface ionization required volatile analytes which could be introduced into the spectrometer source via a gas chromatograph or a direct insertion probe (DIP). Thermally labile compounds are frequently derivatized into volatile compounds in order to be analyzed by these methods. A variety of ionization techniques are now available which enable mass spectrometry to be used for the direct analysis of thermally labile compounds. These include field desorption (FD) (77), FAB (78), secondary ion mass spectrometry (SII'IS) (79), laser desorption (LD) (80), and plasma desorption (PD) (Bl). While these methods have greatly extended the range of chemical Species which can be analyzed by mass spectrometry, each method has some limitations and/or weaknesses. Simply, there is no universal ionization method for thermally labile compounds which will provide both molecular weight and structural information in 3" cases. Flash volatilizatlon or rapid-heating techniques have been 78 79 recently used with ionization methods such as El and CI in the analysis of thermally labile compounds. Rapid heating is useful due to the temperature dependence of evaporation and decomposition processes. It has been shown that at sufficiently high temperatures vaporization may be favored over degradation (82). Therefore if the analyte is rapidly heated, intact molecules may evaporate with little decomposition taking place. These intact molecules can then be analyzed by conventional El or Cl methods. It should be noted that the term 'thermally labile' is rapidly replacing 'nonvolatile', at least in context of mass spectrometric analysis. Compound classes such as peptides, sugars and salts were referred to as being nonvolatile because, on heating, gas phase intact molecules could not be generated due to their thermal Iability, i.e., thermal degradation occurred before vaporization if the sample temperature was slowly increased. it has been shown that rapid heating can generate gas phase molecules of these compounds, hence they can no longer be considered ‘nonvolatile‘. The emphasis on the thermal lability is now more relevent than apparent volatility. Another approach for the analysis of thermally ‘ labile compounds is pyrolysis mass spectrometry (Py-l'lS). There is ample literature documenting the methodology with which such compounds can be identified by the analysis of their pyrolyzates (83,84). Pyrolysis methods use no matrix and can be considered to be a universal method (i.e. most organic compounds thermally decompose to smaller, more volatile comounds). The weak aspect of Py-rls is the ionization step. Methods such as electron impact, 80 chemical ionization and field ionization have been utilized to ionize pyrolysis products. if, for example, a compound Is heated and generates l0 volatile products in the source of the mass spectrometer, electon impact will yield the sum of the mass spectra of each compound. Such spectra are exceedingly difficult to interpret. Other ionization methods can yield simpler spectra (83). in the plethora of methods listed above, the use of inorganic salts has introduced some new options. More specifically, techniques by which (I'l+Cat)‘ can be formed for thermally labile analytes (where cat- Na, Ll, K, etc.) have been investigated and found to frequently simplify the spectra from techniques such as FAB. As early as l97S, it was shown that if salts were deposited on a field ionization emitter, alkali ion-organic molecule adducts could be formed in the presence of a moderate electric field (85). Such techniques were used for both volatile and thermally labile compounds (field ionization/desorption) (85-89). Alkali ion attachment can also be utilized in laser desorption techniques (90). Thermal desorption of ions such as (I"i+i~la)+ from heated mixtures of alkali salts and organic compounds on metal surfaces has also been reported (9i, 92). The development of methodology leading to cationization of organic compounds as a method of ionization is largely due to the work of Rollgen and co-workers. Recently they have reported a technique using a two filament design (93). in a specially constructed ion source, two filaments are in close proximity, mounted on a common 'push rod'. The first filament is coated with a mixture of silica gel and alkali 8] salt which, when heated, emits alkali ions. Thermally labile compounds are present on the second filament. These molecules thermally desorb and form adducts with the alkali ions from the first filament. Cation-molecule adducts are apparently formed in the gas phase. This method produces simple mass spectra from which molecular weight information can be easily derived. Earlier in Chapter 2 the methodology was presented by which a thermionic emitter could be introduced into the source of any mass spectrometer equipped with a direct insertion probe (see Figure 24). The last section of Chapter 2 introduced the use of this probe in the analysis of thermally labile compounds. This section of the dissertation will continue with this investigation. As will be discussed, this method appears to be a combination of pyrolysis and K’Cl. Thus the technique is named K’IDS (potassium ionization of desorbed species). This method of desorption/ionization has a number of advantages. No matrix is involved. The technique has produced a mass spectrum from all of the compounds studied to date and has application in polymer and mixture analysis (will be presented in Chapter 4). This technique may not replace other ionization techniques, or that the compounds chosen as representative compounds cannot be identified by other ionization methods. What is presented is a simple, relatively inexpensive method which can be used to generate mass spectra from thermally labile species and mixtures. No investigation of alternate source designs were done since the goal was to develop a method which can be made available for limited cost and instrument 8.2 .3325 .52..." m. a: 2.9... 2: .o 832:5 2: 6952.. {325.583 «no: 2: c. 2.2.. 532:5 2:25.35. 3 8252: 2352.3. .3“ 0.52”. 3620) B. 2:20) .5 350m :0. 2 9:23. gum co 35 momma: _. j q I D u H : _ mm 3 a m2! .. -. p I —+ W W 2329 . fl \ d m. .M 2 383m bosom 38¢ .- «2.0m > 633 «not 2332...; 9 "WW .0000“: ”knew—q...“ 3% a: 395 2: v0.0... . momma: o... «l 55 5.8.9 2.; Eaezm \ 8.3 82:25 05:53 noon 83335820 83 alterations. The single filament thermionic emitter probe has demonstrated analytical utility because it can be used to ionize gas phase species by K’Cl and surface ionization and also be used for the analysis of thermally labile compounds. Included in this discussion is the apparent mechanism by which ions are formed in this technique, and the resultant spectra of various types of thermally labile compounds which have been targets for analysis by other desorption! ionization techniques. LEXEEBIHENIAL All experiments were performed on an unmodified Hewlett-Packard 5985 GCIl‘ISIDS as before. The description and manufacture of the probe is given in Chapter 2. All experiments were performed using the El source volume unless otherwise noted. Base pressures were typically l0“6 torr or less. The thermionic K+ glass was prepared as described in A reference l7. A slurry of this mixture is made with acetone. A few drops of slurry are placed onto the rhenium wire loop of the probe tip (see Figure 24) and allowed to dry. The probe tip containing the glass mixture is lightly flamed over a bunsen burner to evaporate any solvent. This procedure also hardens the mixture about the rhenium wire, adding stability to the mixture covered probe tip. The probe tip containing the thermionic K+ mixture is then conditioned within the mass spectrometer source for several minutes by passing 3 A through the filament. The final result is not a smooth glass as was before for K’Cl but 84 resembles a ceramic-like matrix. A thin coating of this material about the rhenium wire produces the best result. A single thermionic K+ probe tip typically can be used for the analysis of 30 to 40 compounds. Failure is usually due to oxidation of the rhenium filament. The rhenium filament is held at a potential of *3 V with respect to the ion source using a 0-20 V, low-current power supply. This bias voltage corresponds to the 'bias window'. A final filament current of approximately 2 A is used to resistively heat the thermionic i<+ ceramic material. The initial temperature of the probe was ioo °C, the temperature of the ion source housing. The final temperature is greater than that needed for thermionic emission of potassium ions (800-i000°C). Here 2 A and 4 A correspond to temperatures of 860 and l270°C, respectively. These temperatures were measured outside the mass spectrometer using a vacuum chamber and an optical pyrometer (Leeds 8: Northrup 8622-0. The temperature ramp is achieved manually and as reproducibly as possible. In general, less than 4 seconds are required before emission of potassium ions is observed. This time is a function mainly of the size of the thermionic k’ bead on the probe tip. Throughout this work the term rapid heating ls used; this term and experimental specifics require some explanation. The term “rapid heating' may be interpreted as implying that the rate of heating is most important, dT/dt. Rapid heating experiments are useful because some high, final temperaure can be rapidly achieved. in this experiment, we need to rapidly achieve a 8.5 temperature at which desorption rates are high. We wish to spend a minimal time at lower temperatures such that, when the maximum temperature ls achieved, most of the analyte is still present on the probe tip. The mechanics of this experiment is as follows: The power supply which heats the emitter is set such that, when the supply is turned on, it rapidly attains the preset current value. The time delay between turning on the supply and observing ions is approximately 4 seconds, independent of the final current (i.e., temperature) chosen. Therefore, when higher currents are chosen, the heating rate and final temperature achieved are higher. At this time, we have no accurate measurement of heating rate; the actual rate is not critical, although the experiment works best when the rate is on the order of l50-200°C/sec or greater. Solutions containing the analyte were prepared in an appropriate solvent depending upon the analyte. A drop of the solution was placed directly onto the thermionic K" ceramic and allowed to dry. The sample was localized on the thermionic K’ ceramic tip In order to minimize spectral differences which may occur due to sample deposition at other locations on the probe shaft. Using this procedure Several micrograms of the analyte were typically loaded onto the probe tip. All (commonly available) chemicals were ACS reagent grade and used without further purification. The peptides, xanthines, and cholesterol were obtained from the Sigma Chemical Co. Sucrose and glucose were obtained from Hallinckrodt, Inc. Polyethylene glycol was obtained from the Aldrich Chemical Co. 86 Ampicillin was generously supplied by John L. Bower (Beecham Pharmaceuticals, Surrey, mm. W W The first part of this discussion will focus briefly on the mechanism by which ions are formed in this technique. In this work, we consider a thermally labile compound deposited on the surface of the I<+ emission material, and heated to a temperature sufficient for thermionic emission of potassium Ions. A proposed mechanism must explain the following observations. Usually, all the ions observed in the resulting mass spectrum contain K‘. In most cases, the ion of greatest mlz is the analyte-potassium ion adduct (l‘l+K)’. The appearance of the mass spectrum is dependent on the final temperature to which the thermionic K‘ ceramic is raised. Frequently at some temperature a large (l"l+K)+ ion is observed, while lower mlz potassium-containing ions may dominate at other temperatures. Spectral lifetimes are dependent upon the analyte and usually do not last longer than approximately 30 seconds, although in some cases they last several minutes. K’lDS apparently involves two basic steps of desorption and ionization. Desorption of species is governed by the process of 'rapid heating'. Beuhler and co-workers (82) have used this process to enhance volatility. They have found that high heating rates favor vaporization of a thermally labile compound. In K‘iDS the final temperature to which the K’ ceramic is stepped 8.7 apparently determines the heating rate. This is consistent with the observation that higher applied currents produce (I‘l+K)* ions having higher relative intensities. In competition with vaporization is thermal degradation. Thermal degradation is thermodynamically a lower energy process than vaporization for the compounds studied here. Decomposition usually occurs via reactions such as the production of two molecules from one (frequently through l,2 elimination reactions), loss of small neutral molecules (e.g., H20, H2, C02), and cyclization reactions. The desorption process, the first mechanistic step in K‘IDS, gives a mixture of species (intact analyte and decomposition products) in the gas phase; the distribution of the species depends upon the temperature. It is believed that the next step, ionization, takes place in the gas phase. Earlier gas phase studies in Chapter 2 revealed that addition of potassium ions to molecules in the gas phase was strongly energy dependent. if the potassium ions have too much kinetic energy, adduct formation does not occur. For this early work, a low bias on the emitter was used. The bias used in Chapter 2 was between 0.5 and 5 V. The dependence of bias volatage in these desorption experiments parallels the earlier work with gas phase analytes. The similarities between these two techniques strongly supports the concept that gas phase adduct formation is occurring in this desorption/ionization method. If adduct formation occurred on the surface of the thermionic material the ions could be extracted with any applied potential. This is not the case here. Gas phase adduct formation is assumed to be the dominant ionization mechanism in other desorption/ionization methods. Cooks (94) has developed the concept of 'selvedge' to explain cationized molecular ion formation in SlI‘IS. Selvedge is the relatively high pressure region just above the surface. This concept is also useful for the understanding of other desorption methods. Regarding the concept of selvedge, adduct formation is generally accepted to be a termolecular process (from Chapter 2 and reference 68). More accurately, when K‘ and I1 react bimolecularly to form the excited adduct (i~i+kl**, the adduct must dispose of some energy or dissociate to reactants. Energy may be lost on collision with a third body or by infrared emission. This has been discussed for bimolecular adduct formation involving Li‘ and polar molecules by Beauchamp et al. (95). In K‘IDS a relatively high pressure region probably exists above the thermionic material and therefore adduct formation should be a tertiary process. The high pressure or selvedge region in K‘IDS would stabilize (I‘IsKY. As will be seen shortly, adding a colliSlon gas to the ion source increases the total ion intensity In this experiment. This behavior is consistent with gas phase, termolecular adduct formation. Similar experiments have been described in which adduct formation appears to occur bimolecularly without collisional stabilization of the adduct (93, 96-97). These complexes may be stabilized by emission of an infrared photon (95). A bimolecular reaction would be consistent with these low-pressure experiments. There are a number of advantages in using potassium ions here 89 although it is not suggested that potassium is the optimum cation which can be used for such work (93). Allison and Ridge (6i) have shown the capacity of an alkali ion to induce fragmentation following complexation decreases when going down the column in the periodic table. For potassium ions the dissociation energy [D(rl+l()’l is intermediate relative to other alkali metals. Therefore, a stable potassium adduct is produced with most polar organic compounds; it is highly improbable that this adduct will fragment to produce the other ions seen in the spectra resulting from this technique. , A summary of the mechanism of K‘IDS is as follows. intact molecules and/or thermal decomposition products of the analyte desorb into the gas phase above the rapidly heated potassium thermionic ceramic. The thermal decomposition products produced are consistent with known thermal degradation pathways. Emitted potassium ions then 'sample' these species in the gas phase to produce adducts. Another possible ionization process which can occur in this experiment is surface Ionization. Surface Ionization occurs when an analyte or a decomposition product of the analyte has a sufficiently low ionization energy. Ions formed by SI do not contain potassium. Surface ionization has been well documented (28-3 I). Previous surface ionization experiments have been conducted using organic vapors, however, here the analyte Is deposited directly onto the surface. This should make no difference If surface ionization Is to occur. The potassium thermionic ceramic used in K‘lDS is very similar to materials 90 used In previous surface ionization experiments (3i). Later in this discussion a situation will be presented in which surface ionization does occur. In these cases it Is easily determined which ions are the result of SI since they do not contain potassium. It should again be pointed out that this method is not optimized for source configuration and that the mechanism by which ions are formed may be complicated by other factors such as decomposition of described species on the ion source walls. However, this method does produce useful spectra from the analytes studied to date. K’iDS appears to be analytically useful, inexpensive to utilize, and simple to perform. To demonstrate the utility of this ionization method, the following discussion presents K’IDS spectra of several thermally labile compounds. W105 We; The study of thermal decomposition reactions of saccharides continues to be an active area of research. A large amount of data has been collected on thermal decomposition of saccharides since they are present in foods, paper, and a number of other products (98, 99). it is not suggested that the saccharides shown here are difficult to analyze mass spectrometrically and in fact have been analyzed using other desorption/ionization methods (77, 90, 92-93). Saccharides were chosen for analysis by K'IDS to compare previous thermal decomposition studies with the spectra obtained by this method. 9.1 It must be made clear that the conditions for thermolysis must be considered when comparing results. Early studies (98) identified over 70 products when glucose was subjected to high temperatures for a period of several hours. Curie-point pyrolysis of glucose however, showed 5 products (i00) which are thought to arise from the decomposition of i,6-anhydro-B-D-glucopyranose (levoglucosan). The variables which must be considered when comparing results include the amount of sample used, heating rate, heating time, and atmospheric conditions (composition and pressure). When saccharides are deposited on the potassium thermionic emitter and heated, simple spectra are produced. The spectrum obtained for glucose (H=C6H|205, l‘IW--l80) is shown in Figure 25. The predominant peak in the spectrum is the potassium ion adduct of glucose at mlz 2i9 ((l'l+l<)’). The loss of water from glucose produces the largest fragment ion at mlz 20I, [(I'l-H20)+Kl*. The fragment ion is most likely an anhydro sugar. Anhydro sugars are major decomposition products from the thermolysis of saccharides and polysaccharides (i0i). These thermal decomposition intermediates lead to the formation of lower molecular weight decomposition products (iOI, l02). The peak intensities in the spectrum in Figure 25 are not unexpected since rapid heating is employed. If the heating rate were lowered, the Intensity of the decomposition adduct, [H-H20)+K]’, would be expected to increase relative to (I‘I+K)’. The high desorption rates of both I1 and [II-H20] may account for 92 .2388 3 E2325 web. .mw 952“. 000 0'0 000 000 0¢~ 00a 00¢ 00¢ 00¢ 00¢ ...-—--.—.-~ -———-—-—-Pb— Dunn—bb-p—bh-—#b-n—bnpn—PDb-—P-b ¢oa Ionzu T IOOI IOOI 100: 1001 r 010 WI ' I92! 93 the small amount of further decomposition products. If this were not true, low mass fragment adducts would be seen as suggested by some pyrolytic methods. Host decomposition products from sugars are produced from an intermediate such as an anhydro sugar, however this may not always be the case. The low intensity ions at mlz 189 [(I‘i-CH20)~~K]+ and mlz 159 [(I‘l-C2H402)+Kl’ arise from neutral losses from glucose. This has been observed previously (l02) and is also substantiated by the presence of formaldehyde (Cl-IZOD) and hydroxyacetaldehyde (C2H402) in the pyrolysis of saccharides (l02). Figure 26 is a spectrum obtained when sucrose (I‘I=C'2H220, ', l1W=342l is deposited on the potassium thermionic emitter. The cationized molecule (I'l+l<)“ is observed at mlz 38i. Fragment IOIIS at III/Z 363, 345, and 327 are adducts Of the dehydration products of the sugar, i.e., [(H-H20)+Kl+, [(l‘l-2H20)+K]*, and [(ri-3l-I20)+I(l*, respectively. In addition to dehydration, thermal decomposition of polysaccharides involves the cleavage of the glycosidlc linkage in a l,2 elimination process and has been well documented (I03). The result of the cleavage is two products with the structures C6Hl2°6 and C6H1005. This cleavage is evident here by the presence of ions at m/z 2i9 and 20i which are [C6H'206*KI* and [C6H1005+KI’, respectively. Cleavage of the glycosidic linkage is an important process required for the sequencing of polysaccharides. From Figure 26 the molecular 94 ¢Ifl 050 .om0buam ho Sagan-00m ma: +¥ .ON 0.55....— 10.16 ”I 0¢I 00¢ 05¢ NE». on» can 2b can a P—PrP-l—nb-n—quP—ppbp Iiiiaili l 441 :o :o 0: oz 0 :o 0 10:6 10:6 Pb —b PP—t-n— «iii . P by «4 'WI 'lall 95 weight of the sugar is obtained and, in addition, one can easily conclude that the molecule is comprised of two hexose units. Further work has been started to utilize this method for more complex systems such as gums, starches, and higher order polysaccharides. Winks, A substantial amount of work has been done on the analysis of pharmaceuticals using pyrolysis techniques and desorption! ionization methods. Host of these compounds are thermally labile. As an example, the B-lactam antibiotics have been studied by Py-I'iS (i04) and by various DI methods (IOS, i06). Pyrolysis methods almost always lead to low molecular weight products and consequently compound identification relies upon matching pyrolysis mass spectra to known spectra. Desorption/ionization techniques have become the more popular tool for the analysis of these compounds, however, as yet there appears to be no optimum method for their analysis. For this reason the study of penicillins using K’IDS was undertaken. Figure 27 shows a K‘IDS mass spectrum of ampicillin (ClGH'9N3O4S, I'm-349). The largest peak in the spectrum is that of the cationized molecular ion (I‘I+K)+ at mlz 388, which is not the case in the spectrum of this compound produced by other desorption methods (e.g., 'In-beam' ionization (IOS)). In addition to the molecular adduction ion a large number of cationized fragments are present from the thermal decomposition of the ampicillin molecule. Iiajor fragments can be identified by the cleavage of the lactam and thiazolidine ring as can be shown in 62:89-2 .3 53.5025 meow. SN 0.59..— CE 00' 000 000 0'0 000 000 000 000 0'0 000 000 00¢ 00¢ 0'0 0¢ 0 'WI 1°21 9.7 IOOO IIIU nzu / \o/ \n_9 8. ca. w mom m I _ 2.35m km A“ +¥+/ 0mm ~\E mNN ~\E +x+/ \ +x+ NVN N\ E mm. ~\E 98 Scheme l. The fragment adducts observed provide structural information on the molecule. Two points can be made concerning Figure 27. The number of thermal decomposition products seen in the spectrum is substantially greater than that of the saccharides and account for a large proportion of the total ion intensity. Here the rate of decomposition on the thermionic material must be larger than the rate of desorption. Secondly, as will be shown later, cleavage at the amide functionality was expected. Cleavage of the strained lactam ring, however, appears to take precedence. This is substantiated by other desorption/ionization techniques in which Intense fragment ions are observed corresponding to protonated species arising from the cleavage of the B-lactam ring of the penicillin (l06). The free acid form of the penicillins studied to date by other DI techniques have given results comparable to those In Figure 27. However, also of interest are the salts of these compounds. in this case a cationized molecular ion is not obtained for the salt, however, adequate information is present for a structure analysis. Further results on such species will be the subject of future research. 3% The sequencing of the amino acids in peptides has been of great interest in mass spectrometry. The thermal lability of the peptides has required extensive derivatization before a mass spectrometric analysis can be performed. Ionization methods requiring no derivatization, therefore, have been of considerable interest (l07, i08). Pyrolysis processes of 9.9 amino acids have been studied and are well understood (l09). Pyrolysis of polypeptides are not very well understood. A typical K’lDS mass spectrum of a polypeptide, hexaglycine, is shown in Figure 28. The spectrum of hexaglycine (I1=C'2H20N607, HW=360) is dominated by potassium adducts of thermal degradation products, however, an (I'I+K)* ion of small intensity is present at mlz 399. We can increase the relative intensity of (li+l<)* by increasing the temperature to which the thermionic ceramic is stepped; the result is shown in the insert of Figure 28. Increasing the relative intensity of (I'l+K)“ Is done, however, at the expense of some of the decomposition adducts. A compromise is therefore needed and the result is the full spectrum in Figure 28. Scheme 2 shows the formation of the potassium adduct sequence ions arising from thermal degradation of hexaglycine. The ma jorlty of the adduct ions observed are a result of the l,2-elimination reaction at the peptide bonds. This cleavage Is a low energy process. This point can be illustrated using the decomposition of glycylglycine as an example. Blycylglycine -’ glycine + NHz-CH=C=O Using group equivalent values (i IO) this process is approximately 5 kcal/mol exothermic. This process is analogous to the cleavage of glycosidlc linkages by a l,2-elimination which was discussed earlier for the disaccharide. In addition, Scheme 2 shows the 100 .523on 2 a. 325.5 2: S :2: «2:: md >_3mE.xo.=E< an) 33 .333 2: s 3.22... 3:22. 2: .< m4. 83 22.3 2:8: as 523 .2: .832. :2 52322.. 2: «323.. tom... 2: .33 mm) < N .3 32:3 9.38: < 6592588: B 53302.. web. .8 252“. owe 00¢ can on» com own com 0mm com com man mmm . mom 3% mam . 8.. cos om» . fl mm» 8m .. o- com om. cm. oz 8. oo. oo om oo mNN mm. mm. m: 3%“. IOI Scheme 2 mlz I53 -H0K’ M342 ml: 285 ml: 228 W009 l30 2309 ”hi" ’HOK' C: 73 ear (3 HzN-CHz-C NH CHz-C' so :02 us i0? I73 245 I72 I08 eHeK’ mlz '70 oi-IeK’ mlz 'Il3 m/z 227 NH CHi'C NH CHz-C eHeK' ml: 324 -I-I0K’ mlz I7I oHoK’ 0 244 us 0 ml: "4 /H ok’ so so 9 NH CH2- c—OH 29 ml! I NH CH5 CT I3I 286 eHeK‘ m/z 284 74 343 IT oHOK' 3M I02 rearrangements leading I20 the IOW intensity adducts Observed from the cleavage of the RNH-CH2R bond. ThIS process is thermodynamically less favorable and this is reflected in the abundance of these ions. Consider as an example, the sixth skeletal bond from the N-termlnus in Scheme 2. If this C—N bond was cleaved, 2 radicals of masses IIS and 245u would be formed. However, as the skeletal bond is cleaved, a H shifts (l,2-elimination) to form two. neutrals of masses I i4 and 246u. The K‘ adducts of each are observed at m/z iS3 and 285. Here, most ions are a result of adduct formation from fragments arising from the C-terminal end. Ions at mlz 96, 2l0, and 267 should also appear in the spectrum. A more likely possibility is that the adduct at mlz IS3 is the result of a cyclization reaction to give Eflz-C(O)-NH-CH2-C(O)-NH]. Cyclization reactions were proposed previously from experiments in which depeptides were sublimed in a mass spectrometer to obtain El spectra (ill). The other ion at m/z 324 can be a dehydration product from the species at m/z 342. The reason for forming fragments from only the N-terminal end is not yet fully understood. Possibilities can include the particular interactions between the peptide and the surface of the thermionic emitter or the spatial configuration (conformation) of the peptide. Further research needs to be done on polypeptides to provide insights Into this behavior. Finally, this desorption/ionization method is potentially useful for sequencing peptides and also provides I03 Insights into their thermolysis. This information has not yet been obtained with other pyrolysis techniques. Aim Steroids constitute a large class of compounds which continue to receive much attention due to their Important role In plant and animal systems. In particular, cholesterol has been the steroid studied most extensively since it has been linked to human diseases such as atherosclerosis and cancer (I i2, l l3). Methods In the analysis of cholesterol include wet chemical (l i3), chromatographic (l l4), and mass spectrometric techniques (i I3). The latter technique involves derivatization if used in conjunction with gas chromatography since sterols are often thermally unstable. Cholesterol will be used as a representative steroid. When cholesterol (I'I=C27H460, rIW=3B6.6) was deposited onto the potassium thermionic material and heated the spectrum which resulted is shown in Figure 29. Potassium adducts of a large number of species appear at a higher mass than (I1+K)’ ion of cholesterol. The spectrum of Figure 29 was obtained from a cholesterol sample (U.S.P. grade) which was more than 20 years old. The K‘IDS mass spectrum of cholesterol from a newly opened bottle (Type CH-K, Sigma Chemical Co.) a much simpler spectrum resulted as shown In Figure 30. Here the peak at m/z 423 is the potassium adduct of dehydrogenated cholesterol I(ri-H2)*I ON no 53.5925 meox dw 0.59“— 0¢0 000 00' 00' 05' 00' 00' 0" 00' 0fl' 0¢' 00' 000 000 0B" 000 000 0'0 000 0000¢0 _ :_._. l 3 g _ , be_h Z! _: _ A flfl' AlisuaiuI anllblaa I05 .on 0.59”. 5 mm 25m 2.. 983.280 ..o..3mo.2.u .52.“... .6 82.33 no.3. .on 25...“. 0¢0 000 00' 00' 05' 00' 00' 0" 00' 00' 0¢' 00' 000 000 050 000 000 0'0 000 0000¢0 _-bb-—b-h_-bhn—bth—PbbP—DPPb—P-bh—-p~b—P 4 i 00' uhn—Dnhb—bbhb—hb-b—PPFF—-pb-—-hb—bhbb—bhhb—bbb 1‘ I‘I‘ .‘ 1' I . I I‘ll Q. s I. AllsuaiuI angibiaa 106 concerning the autoxidation of sterols and in particular cholesterol (I i3). Nearly forty oxidation products have been isolated from cholesterol samples which have been merely stored in contact with air. The products include predomiantly (:27 compounds however (:24, (:22, C21, (:20, and (2,9 products were also isolated (i 13). Two points can be made concerning the spectra in Figures 29 and 30. This ionization method presents a fast (less than five minutes) and simple method for determination of sample purity. In the case of cholesterol the 'pure' sample is eassily recognized. The second point concerns the analysis of mixtures. The sample which lead to Figure 29 is actually a mixture of cholesterol and cholesterol derivatives. Mixture analysis often requires long and complex separation processes, however, here the analysis was performed rapidly using this method. mecLAnajxsjs, Procedures for mixture analysis are often as complex as the mixtures to be analyzed. A particular class of compounds which can be considered in mixture analyses is polymers. Polymers can be categorized as synthetic or biopolymers and until recently (83) have received little attention in mass spectrometric analysis. Pyrolytic methods have been previously used however complex spectra often resulted, which required the use of pattern recognition or spectral matching for polymer identification. Figure 3i shows a typical K’IDS spectrum of poly(ethylene glycol) having an average molecular weight of l000 (PEG i000). It should be emphasized that poly(ethylene 107 glycol) is a mixture of molecules having the formula H(0CH20H2)nOH. The average molecular weight of the polymer is determined by the distribution of the molecules in the mixture. This spectrum contains a wealth of information concerning the polymer but it must also be realized that the spectrum represents the analysis of over 70 different species or decomposition products derived from PEG i000. Four series of ions in the spectrum for this polymer include [H(0CH2CH2)00H+K]‘. lcu3cuzlocnzcuzlnon+l 3.23:. co. .38. .nn 95...: 000 000 000 Dhu 00« Dau 00¢ 0h 00 00 thPb—bb-h—nPbb—PPbP—-be—p-nb—DP-n—bbPh_b-P —bbbp an» . .eeee luoo« 112 .353 06 me} Be: ceom e5 .c.._.o.oE< .o 3.3:: ceom .e> 3.9.3... no. .3: .3” 959: C. 0. NH». . 000' IR00« H3 .oem: on) < N .o Zeta 9.33: < .men :8»... «Eager: .o 52.88 2.... .mn 05...“. N\ E 000.“ 000 own 000 0'0 000 00s. ONB 000 010 000 000 0ND 00' 01‘ 00' 00" 0mm 00m 01m 00“ 'Wl 'lall 114 [H(0(Zl‘l2C(CH3)HCH2)“0H~‘K]+ ions. From the spectrum shown the average W is calculated to be 7l0. This ionization method appears to be a fast and simple method for determining the mean molecular weight in such mixtures. Unfortunately the mass range of the mass spectrometer used in these studies does not permit such analyses of higher molecular weight polymers. This ionization technique has been applied to a wide range of synthetic polymers and biopolymers. These results will be presented in Chapter 4 of this thesis. W For polymer analysis the amount of sample used is usually large compared to trace analysis. Sensitivity of this Di technique however is of concern particularly in regard to trace components in complex mixtures. in Figure 36 the total ion intensity is plotted versus an amount of polyphenyl ether (H=C36H2605, Hw=538) deposited on the potassium thermionic material. Polyphenyl ether (6 ring) is a compound which does not thermally decompose readily and therfore the spectrum produced from this ether using this technique consits of only the (li+K)’ ion at mlz 577. A solution containing 239 nglpl of the ether in acetone was Prepared. The curve was obtained by applying the appropriate amount of solution on the probe tip, drying, and then heating within the mass spectrometer source. Curve A in Figure 36 was obtained at the base pressure of the H5 ISOOF I l200 900- Tli (counts) 600- 300- L l 0 500 l000 EKK) 2000 Amount Polyphenyl ether (rig) Flgure 36. Total ion Intensity at Peak Maximum of Polyphenyl Ether Plottes vs. Amount Deposited on the Thermionic Emission "aterial. A Heating Current of 2 A Was Used. 116 mass spectrometer (2xi0"6 torr). Here 500 ng of sample gave a peak ion intensity of 200 counts at a signal-to-noise (SIN) ratio of 20. Presently, attempts are being made to improve the sensitivity of this ionization method. If K‘iDS involves gas phase adduct formation the sensitivity would be enhanced by the addition of a neutral collision gas. Asseen from earlier studies in Chapter 2 increasing the pressure facilitates adduct ion formation since this process is termolecular. Curve 8 in Figure 36 shows this experiment repeated at a pressure of ixi0’4 torr 0' N2 With") the mass spectrometer source. In this case approximately 250 ng of sample are required to obtain a peak ion count of 200 at an equivalent SIN ratio which is a factor of 2 improvement from the low pressure case. Further work is needed in this direction to improve sensitivity. Polyphenyl ether was chosen for this study primarily for its simple spectrum. Other compounds may prove more or less sensitive using this ionization method however, for most analyses presently performed, sample sizes used were in the low microgram range. To this point all of the spectra presented contain K‘ adducts of the analyte molecule or l<+ adducts of thermal decomposition products derived from the analyte. However, some organic compounds when analyzed using this ionization technique produce ions which do not contain potassium. Here two possibilities "7 exist: i) direct thermal desorption of ions (as seen for quaternary salts (9i, 92)) or 2) surface ionization. Thermal desorption has generally been accepted to occur for species which exists as 'pre-formed' ions prior to the application of heat. Thermal desorption will not be applicable in the following discussion. For surface ionization to occur the analyte molecule or a decomposition product of the analyte must have a low ionization energy. Amines generally have ionization energies less than 9 eV and have been observed to interact with heated surfaces (characterized by high work functions) to form positive ions (29). ionization energies of 9 eV have been calculated for the thermally labile xanthines (l i5). Presently, however, only one study has determined these ionization energies. These authors find the ionization energies for the class of compounds containing the xanthines are greater than 9 W (i i5). Figures 37 and 38 show the spectra of xanthine (f1=C5H4N402, i'lw=i52) and theophylline (C7H8N402, riw=l80) respectively. Here an abundance of ions which do not containe K+ are formed including the molecular ions of each compound. A mechanism for understanding the ions formed from theophylline is shown in Scheme 3. This alternative ionization process is useful in this ionization method as can be seen from both spectra in Figures 37 and 38. Molecular weight information from the m/z of the l‘i+ and the (i'l+i()’ ions is obtained as well as a number of fragment ions indicative of the molecule‘s structure. If the temperature of the 118 one: no} < We 2 29:5 9:22. < .2228. .o 8322.... 22.322 oneromxmcfv. .nn 95...“. 0mm 0mm 0¢N 00m 00¢ 00¢ 05¢ 00—¢ 00_¢ 0¢¢ 0M¢ 0N¢ 0¢¢ 00¢ 0—0 0—0 05 00 00 —p:P—E_Ppbp—bbpn—-pm:_P— b— 11111 it 1—1—1 mn¢ I Z/ \\O 2 U F H p: z u\ I .. 0 00¢ 33— :: _PP1.1._-_{.IP.EFP_LLLL._—.t.— al.-.175.— ttFLL-r—ri 00¢ WI 138 119 .oen: we} < me. 2 22:3 9.28.. < 2....282: 2 89.225 .3332... ooe...:m\mo_.+v. .em 95...... N\E one one one ooe on. on. on. on. on. oe. on. on. o3 oo. on on on on on _ . a .i new no i no. one no i on nxw .. z zxooo F H ,__ - Z \ I I w #8 . on. o "“l 192! 120 Scheme 3 1—?er flffi” 1 CH3 c'H3 C“ 3 ml: 95 m" '30 mlz l23 -HCN V 1' °CH3 0 ‘1' NEG-CENH <—— CHs-N-C-CE NH m/z 53 m/z 68 121 thermionic emitter is low enough to permit surface ionization a selective ionization process exists, that is, only species with low ionization energies will be ionized. Such a technique would be helpful in selectively ionizing components in a complex mixture. in addition, the thermionic material can be used in a negative ion mode. Figure 39 shows the negative ion spectrum of L-methionylglycine (ri=c7llI 4N203S, riw=206) when deposited on a 00510!“ 11110111110010 emitter and heated. The (301110031th11 Of the cesium thermionic emitter is 25l02:Al203:C520. Holecular weight information is obtained from the (ll-H)" ion at mlz 205. The largest ion in the spectrum is that of CN' which is not surprising due to its high electron affinity (a requirement for negative surface ionization). A cesium thermionic emitter was used since negative surface ionization is favored on low work function materials (30). For the alkali oxide thermionic material used in these studies the work function of alkali-containing material decreases as one goes down that column in the periodic table. Here, again, the use of negative surface ionization would offer selectivity in an analysis since only species with sufficiently high, positive electron affinities would be ionized. W The use of the thermionic emission probe has proven to be useful in the analysis of thermally labile compounds. 122 .32. we) < Wu 2 29:3 9:22. < .> c... we) 3.5.... e... 3 3.2.3. .22. e... no we... 2: 2.32: 522.5. 2.3.5.2: e... co 3:895 23>... .2252... .o c.3225 co.2~2o. 32...... 2:29.: .on 959“. N}: one ooe o... o... . 9.. on. 8. 8 oo 9. on F . _ _ _ _ _. _ . _ 13.4.-. - 44:1 n all a. .. n I 0101 II —U-U-m-U non reluinrulzzelzulzf mx1_ = = O O on WI lea 123 Particularly enlightening was the success achieved for polymeric systems. This success led to further characterization of K’iDS in polymer analysis which is the subject of the next part of this thesis. A number of other compounds have been tried using K’IDS but have not been fully evaluated. The spectra for these compounds can be found in Appendix A of this work. Conventional analysis by mass spectromety involves the ionization and subsequent analysis of gaseous analytes. Polymeric systems, due to their nonvolatllity, require either pyrolytic methods or desorption/ionization mass spectrometric techniques for their analysis. Pyrolysis methods are most often used for the MS analysis of polymers (83). These techniques generally involve thermally degrading‘the polymer followed by ionization using electron ionization, chemical ionization, or field ionization of the pyrolyzates. Pyrolysis can be achieved using resistively heated filaments, lasers, or by heating a ferromagnetic wire to its Curie point using a radio frequency field (Curie-point pyrolysis). Pyrolytic methods of analysis can be divided into the two categories of pyrolysis-gas chromatography-mass spectrometry (Py-GC-l'iS) and pyrolysis-mass spectrometry (Py-l'iS). in the former thermal degradation products from a sample can be chromatographically separated and each qualitatively analyzed by conventional mass spectrometry. The mass spectrum from a Py-riS experiment can be used to identify degradation products or to produce a 'fingerprint' spectrum that may be interpreted using pattern recognition techniques. Pyrolytic techniques in the analysis of polymers can suffer 124 125 from several disadvantages. Secondary reaction products can be formed in the pyrolysis zone. These products are not desirable since they complicate the spectra and diminish the quantity of primary degradation products. The heating rate and final equilibrium temperature used are important factors since thermal decomposition processes of polymers are time and temperature dependent. The last disadvantage concerns the nature of pyrolysis products. Often, only low mass species are observed. These species are helpful when determining the structural subunits of a polymer, however, no molecular weight information can be derived. These disadvantages suggest the need for alternate mass spectrometric methods for the analysis of polymers. The optimum method would be rapid and direct (i.e. the pyrolysis step and the MS analysis would be simultaneous). riost polymers are a mixture of oligomers consisting of a specific repeating unit (monomer). For the analysis of such mixtures, average molecular weight, the oligomer distribution profile, and structural information is desired. The advent of new ionization modes in mass spectrometry for nonvolatile/thermally labile compounds have made the analysis of some polymers possible. Various desorption/ionization techniques have been used for a number of polymeric systems with varied results. Dl methods which have been used for polymer analysis include desorption Cl (ii6), rapid-heating methods (i i7), laser desorption (80, li6) plasma desorption (l l8), secondary ion mass spectrometry (i i9), and eiectrospray (l20). in most cases these DI methods produce 126 spectra reflecting the oligomer species and, when applicable, can make it possible to determine average molecular weights of polymeric mixtures. The mass difference between two consecutive oligomer ions in a mass spectrum allows the mass of the monomeric unit to be determined, however some of these Dl methods only yield information of the mass of the monomeric unit. When the distribution of oligomer species is beyond the mass range of the mass spectrometer only structural information can be obtained. In this case pyrolysis techniques would be advantageous. K‘lDS offers the advantages of other Di techniques and pyrolytic methods for the analysis of polymeric systems. The potential of this method has been demonstrated in Chapter 3. KIDS offers a number of advantages over pyrolysis methods and some Di techniques, due to the fact that the method utilizes rapid heating. Rapid heating allows for the rapid attainment of temperatures at which desorption rates are greater than competitive degradation rates. in polymer systems this is useful if molecular weight information is needed. in addition, rapid heating may tend to favor the production of larger mass fragments from the polymer chain. This type of information is often lacking in pyrolytic methods. For many polymers, the average molecular weight falls outside the mass range of current mass spectrometers. In these cases, the goal is to obtain structural information to identify the polymer. By adjusting the heating rate used in K’iDS, spectra containing potassium adduct ions of thermal decomposition products are produced. Therefore 127 K‘iDS may be used in structural studies of such large polymers. Another advantage of K‘iDS is that the ionization process is in close proximity to the site of desorption. In general this feature is inherent in most DI methods but can be lacking in pyrolytic methods. It would appear that K‘IDS has some of the advantages of both pyrolytic and DI techniques. It should be noted that the use of alkali ions in the analysis of thermally labile compounds has been exploited in a number of related Di methods as pointed out in Chapter 3. Examples of various polymeric systems analyzed by K‘lDS will be presented in this chapter. It is apparent that various polymers behave differently when analyzed by this technique. In the first section polymers with average molecular weights outside the mass range of the mass spectrometer will be discussed. Here K’iDS provides structural information useful in identifying the polymer. The final section focuses on relatively low molecular weight polymers to highlight the utility of K‘IDS for molecular weight determinations. LEXEEBHIIENIAL The polymer analyses were performed on the same GCIriS/DS system as before. This desorption/ionization technique and the operation parameters have been described in Chapters 2 and 3. The thermionic probe was the same used in the earlier studies. The temperature to which the thermionic probe was stepped determines the heating rate. In this case higher applied currents 128 produced higher heating rates. The mass spectrometer was set to begin scanning above mlz 42. Appropriate solvents were used to dissolve the polymers. Several drops of dilute solution were applied to the thermionic K+ ceramic and the solvent evaporated off. Typically, several micrograms of polymer were placed on the probe tip in this way. All (commonly available) chemical were ACS reagent grade. Nylon 6, polyvinylpyrrolidone, poly(ethylene glycol)s and poly(propylene glycol)s were obtained from Aldrich Chemical Co., and XF-i lSO was obtained from Applied Science. W Wrmimmmmam A particular polymer's behavior in K’iDS will fall into one of two categories. These categories are differentiated solely by the length of time in which spectra characteristic of the polymer are produced during the K’iDS experiment. The first class of polymers generates total ion intensity curves similar to that shown in Figure 40. The duration of useful K‘IDS spectra usually is less than 30 seconds. The second category of polymers show Tli curves such as that shown in Figure 4i. Here the duration of useful spectra often is longer than five minutes. Polymers in the . former group will be referred to as 'short-lived' and polymers in the latter group will be called 'long-lived'. The short-lived polymers are viscous, high average molecular weight polymers, 129 62.92 .23 29:3 22.3 8222. 3o.:< .e :2»: .e 22.22 8.... 8. 5.23... .3. .23 .8. 2...... HHP .0¢00 IR00¢ 130 .n seem 2 3:92 we... 29:3 .325... n 20282932 2 one: 2323.. e2: .5: 22h 2; .25 23:6 2.0155on .o «2222 web. .8“. 3.982. co. 2.: ..v 959... coon own 000 D5“ 00“ can 000 05¢ 00¢ nm¢ 00¢ 05 00 DN 0 HHP Pub—ihh-b—bP-h—PbbP—Pb-b—-bbh_thb—-th—Ph-h—lbbpb_bbhp_hbhn—-h .1- .NDD'¢ I 0¢ 131 usually gums or solids. For these polymers, K’lDS provides useful structural information. Long-lived polymers are of low viscosity, usually liquids. These polymers were chosen since their average molecular weights (riw) fell within the mass range of the mass spectrometer used here. In this case K‘lDS can be evaluated for providing molecular weight information. Polymers representative of both groups will be discussed. For both cases suggestions on how to use the resulting K‘iDS spectra will be presented, that is, one purpose of this research is to provide the methodology for the analysis of polymers using K‘lDS. Once this methodology is established, polymer analysis can be performed, as will be demonstrated here. _ The rapid heating of a thermally labile compound is an Important aspect of K’lDS. A short discussion of this topic is relevent here. Consider a case in which a thermally labile compound is deposited on the material and heated to a temperature at which potassium ions are produced (Figure 42). Until the emission temperature is reached, species desorb from the surface of the thermionic ceramic and no ions are formed. At the onset of potassium ion emission adducts will form with whatever is in the gas phase above the emitter. From Figure 42, spectra of the analyte will be formed only when the two curves overlap. Rapid heating leads to both vaporization of the sample, and the prompt production of potassium ions before the sample extensively decomposes. Pictorially, this leads to a greater overlap of the two curves in Figure 42. Figure 42 only applies to short-lived polymers and most other thermally labile compounds 132 63.69... 92 9.22... more. 9.2.3 2.2.... :23: 2.225 of C. 2 .329...“ e... .22 9:22.. Bee: 3 .95 9.225 2 222.395 2.. no. 0...: 2 222.5“. e no 3.932. .9 95...“. USE. M I m N - i. .4. N B I..- IA 20.32.. 4/ . . .m. t. 9.22.. Boom. / .3 39630 ne_oeom 133 which have been studied previously. Desorption of thermally labile compounds by rapid heating has been performed using electron or chemical ionization (1 l7). Friedman and co-workers (82) have previously found that desorption is favored at higher temperatures over decomposition. Simply, the heating rate and final temperature determine the extent of fragmentation. The K‘IDS spectra may be 'adjusted' by altering the heating rate. Thus, the heating rate affects the cross section of the two curves in Figure 42, and also the types of neutrals which are being desorbed. One factor which affects the heating rate is the amount of thermionic emission material on the rhenium filament. The amount determines the rate of heat transfer from the rhenium filament to the sample. A thin coating of the thermionic material on the filament allows for a sufficiently large heating rate. This is another variable in K’lDS; the “amount of thermionic material needed will depend upon the type of information (structural or molecular weight) which can be obtained for a polymer. Rapid heating of the probe applies to both short-lived and long-lived polymers. in short-lived polymers the heating rate is adjusted to maximize the overlap of desorbed species and potassium emission (F igure 42). However, due to the hlgh molecular weights of these polymers, decomposition is required to provide structural information. Therefore the heating rate must not be too high, so that decomposition of the polymer will occur. What follows is a more detailed account of the short-lived polymers. This will then be followed by a discussion of long-lived polymers where 134 molecular weight information may be obtained using K‘lDS. WERE; LW A typical example of a short-lived polymer's total ion intensity curve is given in Figure 40. Nylon 6 ([NHiCH2)5CO-ln) will be used as a 'model' polymer in explaining how the results of the K’lDS experiment should be interpreted. For this figure the mass spectrometer scan rate was set such that each scan took 0.8 5. Past scan 40 the total ion intensity drops considerably and, in addition, pyrolysis proceeds to a point in which useful spectra are no longer obtained. Therefore there is little difficulty in deciding the useful 'cut-off' point in the TH plot. Here, the cut-off point refers to the point in the curve in which information from the spectra is not useful for analyzing the polymer. uAfter the cut-off point is determined the next step is to decide which spectrum to choose for analyzing the polymer (in this case Nylon 6). Several questions need to be addressed before this decision can be made. Do the spectra change with time? if so, what should be considered as the K‘lDS spectrum of this compound? From Figure 40 the Til curve can be divided into two regions. One region is located between scans ii and 20 and the other from scans 20 to 40. The first region encompasses the 'peak' that is usually observed after the potassium thermionic 135 material on which the polymer is deposited is rapidly heated. The second region contains that part in the K’lDS run where the total ion intensity is decreasing. Figures 43 and 44 are averaged spectra from each of these two regions in the Til curve of Figure 40. Only small differences exist between the two averaged spectra; they are very similar. Therefore, interpretation of polymers belonging to the short-lived group would begin by averaging all of the spectra up to the cut-off point. The total averaged spectrum for Nylon 6 is shown in Figure 45. This averaged. spectrum is not markedly different .than the averaged spectra from the two regions in the Tll curve (Figures 44 and 45). For short-lived polymers the procedure for selecting a spectrum for interpretation would involve: l. Choosing the point in the Til plot where intensity diminishes to an unusable limit or to a point where potassium adducts of pyrolysis products create a 'peak at every mass' situation. Usually the usable region of the Til curve for short-lived polymers is 30 seconds or less. 2. The averaged spectra from each region in the Tll curve should be compared for gross differences. In most cases, for short-lived polymers there is little difference between these regions. 3. The K‘lDS spectrum is, then, an average over both of these regions. This final spectrum is suitable for interpretation. I36 .3 0.5.... now. ow... . 28m :3... o :2»: .o Estevan. coo-to: H... 95...“. ~\E 000 005 005 000 DVD OD. 000 can 00' 0" 00' on” can 0.1.. 0'“ on“ 00¢ bhhb—nbbh—r-nhb—n-bb—bpbb—~ ‘ “ ' ‘ ‘i‘l " _ 1 : — IBM AJJSNBLNI EALLV'TSH 137 3.. 2...... com. ownow 23m Ea... e :2»... .o E:...uo..m coma»: .vv 95...... ~\E 000 can can 0.0 0'0 000 000 can 00' 0" 00' con Gun O.“ 0'“ can 00¢ pbb—hbnn—tpp up. _nh-—--b bb—p hn—hhbb ‘ ‘i..11 ‘ 1'q:‘l {I 1111‘- ‘1“‘11II{‘ Ohm ALISNBLNI EALLV'IEH 138 3.. 2...... com. 9?. . 28m Eu... 9 5...: 3 82.925 399.22 .mv 2.6.“. ~\E 000 005 can 000 0'0 000 000 can a... 0" 00' can can on“ 0'“ com 00¢ bet—hbtb-Pbb— bb _biPPh—hhnb (1 (4.1141 11. Ihfl ALISNBLNI HAIR/138 139 Using this procedure a number of polymers which fall in the short-lived class will be presented here. The average spectrum of each polymer will be used to interpret the results. Several polymers will be presented which represent some of the more common polymer classes. W The structure of cyanoethylmethy silicone is shown in Table 2. Siloxane polymers have been used extensively as lubricants, chromatographic stationary phases, and in vacuum pump technology (l2i-l23). In most cases these polymers are used in high temperature applications due to their nonvolatility and thermal stability. A number of thermal degradation studies have been reported for the silicones (i24-l26). These polymers were chosen for KIDS analysis so the results can be compared with earlier degradation literature. Figure 46 shows an averaged K’IDS spectrum of XF-i lSO obtained in the manner which was described for short-lived polymers. Two ion series dominate the spectrum of this polymer. These series are shown in Table 3. The average molecular weight ("w) of this polymer is unknown, however, in comparing l’iw values of other silicones it would be safe to say that the weights of the cyanoethylmethyl silicone oligomers are outside the mass range (i000 amu) of the mass spectrometer used here. The major thermal degradation pathway in an inert environment (vacuum) is 140 Table 2 Formulas of the Polymers Used in the K‘lDS Analyses £93225; Structure XF-llSO 9H3 r (EH3 C'H: H3C-Si-OTSi Si—CH3 i l i CH:5 942 n CH.5 9*: CEN I "3'10" 5 {NH-é-(CHQSE- n r 0. S? Poly(ethyleneadipete) _; CH2- (3.2-0- t - (CH2 )(C‘O h. Polyvinylpyrrolidone (PVP) H25 —CI:H2 H2C\ /C-O N I —E— CH —CH23;- 141 Good .2... T... 3 82.8.5 25. coat}. .3. 0.52. 00s. ~\E 0" '00 00' 00¢ ALISNBLNI BALLV'IEH 142 Table 3 Potassium ion Adduct Series for Cyanoethylmethyl Silicone (XF‘I '50) Series Proposed Structure of Degadotion Product (M) ml: of (DHKV A (CH3)3SiO[Si(CHg)(C3H4N)Oln5i(CH3)3 540 um). 653 mm. 766 (n=4). 879 (n=5). 992 (n=6) 8 [Si(CH3)(C3H4N)OIn 378 (n=3l, 491 (n=4). 604 (n=5). 717 (n=5) 143 the formation of cyclic siloxanes and is believed to take place in either of the two ways shown in Scheme 4. The first possibility in Scheme 4 involves the formation of cyclic siloxanes through degradation at the terminus of the polymeric chain. This has been shown for hydroxy-terminated polymers (l25) however here a much larger terminal group (Si(CH3)3) is involved. in this case the second possibility in Scheme 4 is favored. This mechanism proposes that formation of cycloslloxanes occurs in the middle of the chain. This has been suggested by Grassie and r‘lacfarlane (i25). Either mechanism, however, explains the appearance of K’ adducts of a series of decomposition products, [telly-551051(cn3ic3n4molnsucu3i3 + K‘], in the K‘ios spectrum. it has been shown that the average molecular weight of silicone polymers decreases linearly due to the formation of such decomposition products (l27). in addition to potassium adducts of these decomposition species, K’ adducts of‘ the cyclic siloxanes are observed in the K’lDS spectrum. Their abundance is greater than that of the previous series of decomposition products. This is not surprising since, qualitatively, more cyclic siloxanes would be expected when a long silicone chain is degraded. These two series are labelled in Figure 46. Note that these species are all potassium ion adducts. it is interesting to note that there is freedom of rotation about the silicone backbone allowing these polymers to coil helically (l24). Substitution of larger groups on the backbone of the silicone affects the ability of the polymer to coil (i24). 144 Scheme4 R CH R CH R \/ 3 \( 3 I WSi-O-Si-O-Si-CH3 CH3.” I R: -CH2 -CH2 -CEN H3C-SIDO-ISl-O CH3 R CH3 R CH O \./ 3 . R\ ./ \ ./R M SI ’O‘SHCH3)3 + ,Si Si\ H3C l | CH3 ' O 0 / \Si /\ R CH3 V35" Rim” MMSi-CID-Si-T MQD'Si-O—?'-R 1 CH3 (EH3 R\ /O\ /R mSi-CH3 ,Si 5‘ I + H3C I I\CH3 \Si/ 145 Larger groups hinder coiling more than small groups. The distribution of the potassium adducts of the cycloslloxanes in the K’IDS spectrum may be a reflection of this coiling ability. That is, for the larger backbone groups such as cyanoethyl, only larger coils are permitted leading to larger fragments such as tetramers and pentamers (n-4 and n-5 for cyanoethylmethyl silicone in Table 2). The large amount of the trimer found in thermal degradation studies of dimethylsiloxanes would also support this reasoning. Further work on other silicones will aid in the investigation of this behavior. K‘iDS provides useful structural information for this silicone polymer. The potassium adducts of a series of decomposition products can be seen at mlz 540, 653, 766, 879, and 992 (see structure in Table 2; n-2 to 6). The potassium adducts of the cyclic. siloxanes (mlz 378, 452, 604, 7i7) are indicative of silicone polymers. For both series the mass of the repeating unit of the polymer can be obtained. The large number of products seen in the low mass region of Figure 46 (below mlz 300) may arise from impurities and/or residual catalyst as is often the case in such polymers (l22). Therefore one possible use of K‘iDS is as a screening method for silicones used in chromatography. Chromatographic applications of these polymers impart severe demands on the polymer purity since small amounts of impurities can dramatically affect the chromatographic behavior. K‘iDS can also be used for thermal decomposition studies of the silicones. Such studies are valuable due to the high temperature applications of the polymer class. 146 mm Figure 47 shows an averaged K‘lDS spectrum of polyvinylpyrrolidone (PVP) which also is in the short-lived category. it has an average molecular weight of l0,000. The structure for this polymer is given in Table 2. A number of series are observed for PVP, all of which are separated by lli mass units, the repeating monomeric unit ([-CH(C4H6NO)CH2-l). The major series in the spectrum contains the potassium adducts of the dimer (n=2), trimer (n=3), up to the octamer (n=-8) at mlz 927. The value of n refers to the number of monomeric units in the structure of PVP (see Table 2). This series is labelled with. an H in Figure 47. The oligomer series can arise from either low mass oligomers already present in the polymer or from the chain scission of the polymer backbone. Here the end groups of the PVP polymer are unknown and thus we are uncertain of the first possibility. From prior pyrolytic studies of polyolef ins, however, it seems likely that the second possibility does occur (i28). A likely mechanism proposed for the thermal degradation of polyolef ins and polymers such as polystyrene involves a hydrogen transfer to the site of the chain scission yielding a saturated and an unsaturated (l28). This is shown for PVP in Scheme 5. This intramolecular hydrogen transfer to the site of the scission can explain all but one of the series in the K‘IDS spectrum. Another possibility is an internal cyclization mechanism similar to that which accounted for the formation of cyclic siloxanes from the silicones. For this case end groups on the polymer are irrelevent 147 .o..o....o.........>...>>.£. .o 53.8.... no... 63983. .NV 95.... N\E 0'0« 000“ 80 on. O.- O'. OO- Olh dab 000 0'0 000 5 2 2 D x 000 000 can 00' 0" on? 000 can bid ova CON 00¢ T E S aka 1 AllSNHlNI BALLV'IBH 148 Scheme 5 *3 Miami: H H. We-e-i-i’re-iw R' Hie—E”; R H R H R H 2 \1" 149 to the production of decomposition series. A potassium adduct of the monomer of PVP is not evident in the K’iDS spectrum. it should be noted that a free radical mechanism is a requirement for monomer (128); little evidence for radical mechanisms are present in K‘IDS. The less intense series of K‘ adducts of decomposition products at mlz 287, 398, 509, 620, 73i, 842, and 953 ([HZC=CH[CH(R)-CH2]n-C(R)=CH2 + «1*, R=C4H6NO, n=l-7) may result from scission at the side chain. The loss of the side chain in PVP can be compared to the formation of benzene from polystyrene. Polystyrene is structurally similar to PVP, however the side chain is a phenyl group. Benzene was found to be a thermal decomposition product for polystyrene (i28). For polystyrene intramolecular hydrogen transfer has also been considered for the formation of benzene. in a similar fashion this process may explain the previously mentioned decomposition series for PVP. The literature is rich with pyrolytic studies of polystyrene and polyolefins. Here the K‘lDS spectrum appears to be consistent with these earlier findings. Mm The next group of short-lived polymers reported here is polyesters. Polyesters are a major ingredient in paints, varnishes, and textiles. They are frequently used as polar stationary phases in gas chromatography. Thermal decomposition 150 studies of polyesters have been previously reported (l29). From these studies it has been shown that the carboxyl group -C(O)-O- is the reactive center for the initiation of thermal degradation. Figure 48 shows an averaged K‘iDS spectrum for poly(ethyleneadipate). In Table 4, the desorbed species which lead to the observed potassium adducts are listed. in one particular case a decomposition series has two proposed isomeric structures. The two structures cannot be distinguished since both are possible decomposition products. in addition to those products listed in Table 4 a number of unidentified peaks exist. These species are, like those identified, separated by i72 mass units which is the repeating unit of the polymer. The end groups for these decomposition species are consistent with those found in the Py-El-riS study reported for this polymer (i30). A comparison of the K’IDS spectrum with other analyses of poly(ethyleneadipate) using FD, FAB, and negative FAB (i3l) is possible. Not one of these ionization techniques parallel the K’IDS results, however, certain series from each of these methods can be found in K’iDS. For example, oligomeric carboxylates (H') of the type HotcmCH2)4c02(CH2)201,,cmxcazi4c02‘ (n = 1-7) appear in the negative FAB spectrum. in K’IDS a similar series appears as potassium adducts of the protonated form (H'+H’+K‘) and corresponds to the ions at m/z l85, 357, 529, 70l, and 873 (h-o-4). Other correlations exist between the k’ios and the FD 151 .33....5222313. .3 5:38.... m9... 638...: .3 2...... ~\E 96¢ coca 000 can can a... no. 005 ans. one ova coo _*DbP-I4IIb—5UDPI_II 1 .l in a n x I can can can one o: 2... can can can own com on. 4. “4.40 . . ._ 1 .l r- t .3 T ALISNELNI SALLY/732:1 152 Table 4 Potassium ion Adduct Series For Poly(Ethyleneadipate) We" W (M) mlz of (may; 1110(CH2)2002(CH2)4C(O)]nO(CH212011 273 (n=1), 445 (n=2) HOHCHZ)2COz(CH2)4CO1n(CH2)2C02(CH2)3CH=C=O 211 (n=0). 383 (n=1). HZC=C(H)C02(CH2)4C(O)[O(Cflg)2C02(CH2)4COanH 555 mm, 727 (h=3i. 899 (n=4) H10(CH2)2COz(CHg)4C(O)1n0(CH2)2COg(CH2)4C02H 229 (n=0). 401 (n=l). 573 (n=2) H0[C(O)(CH2)4C02(CH2)201nC(0)(CH214C02H 185 (n=0). 357 (n=i). 529 (n=2), 701 (n=3), 873 (n24) 153 spectra. From earlier results obtained for FAB and FD the two ionization methods appear to be complimentary and are useful in characterizing the polymer by identifying diagnostic degradation products (i3i). K‘IDS appears to compliment these two ionization techniques and can also be used to characterize this polymer by identifying diagnostic thermal decomposition products. mm The K‘iDS spectrum of a polyamide (Nylon 6) is shown in Figure 45. The Nylon 6 sample was used for characterizing K‘lDS; it also demonstrates the utility of the method for the analysis of aliphatic polyamides. Aliphatic polyamides such as Nylon 6 have been studied previously by mass spectrometry (i32-i34). From Py-i‘iS it has been shown that the main thermal decomposition pathways of polyamides are the cleavage of the amide bonds and the elimination of methylene units (i34). At lower temperatures cyclic oligomers are formed. in a recent SIMS study (i3S) six nylons were analyzed in which high mass adducts were seen with Na‘, K“, and Ag‘. Here potassium adducts are observed for the dimer ([-NH-C(O)-(CH2)5])n, n=2) to the hexamer (n=6). in the SIMS study the highest fragment ion containing potassium was that for the tetramer, while for Na‘ it was the 24-mer. it was proposed that Na+ stabilizes long chains of polymers more effectively than potassium or silver, as suggested by Rdllgen and co-workers (93). initial investigations from this laboratory 154 using sodium emitters for KIDS type work suggest that sodium cationization (Na‘iDS) may be advantageous in mixture analysis. in the case of Nylon 6, we cannot be sure of the site of cleavage along the polymer chain. From previous work with peptides it would seem a l,2- elimination at the amide bond WONG be favorable. However we have also shown that NH‘CHZ and CHZO bonds cleave as well for the peptides. As can be seen from Figure 46, Nylon 6 undergoes extensive fragmentation. This is consistent with earlier studies (l35) in which cleavage of C-C, C-C(O)-, and C-N bonds occur. Essentially every bond in the monomeric unit may be broken down giving rise to two different species. For example, equation 63 shows cleavage of CH2-CO; equation 64 shows cleavage of Nil-CH2, and equation 65 shows cleavage of CHZ-CHZ bonds. ~CH2-CH2-C(O)-NH-CH2~ .. CHz-CH3 + O=C=N-CH2~ (63) ~Ci-i2-Cl-i2-C(O)-NH-CH2-Cl-l2~ -+ ~CH2-CH2-C(O)-NH2 + CH2=CH (64) ~01 ‘CH -CH *C(0)-NH~ -’ ~CH=CH + H C-C(O)-NH~ (65) 2 2 2 2 3 An important question is, why do the oligomer series stop at the hexamer in K’iDS? A number of possibilites exists which include cationization efficiency of potassium ions, intermolecular hydrogen bonding (l35), heating rate, and sample 155 introduction (the sample was evaporated onto the potassium thermionic probe tip from solution using a heat gun). More work is needed to answer the above question. W The last section of this discussion focuses on 'long-lived' polymers; those polymers are characterized by long lasting total ion currents in excess of approximately 45 seconds. Long-lived polymers exhibit characteristics different than those of short-lived polymers. The two largest differences are the duration of spectra, and spectral differences which arise during various portions in the Tii curve. The two polymer types which are included here are the poly(ethylene glycol)s (PEG) and the poly(propylene glycol)s (PPG). Their behavior is illustrated in Figures 49 and 50 using PPG 725 as an example. PPG 725 has an average molecular weight of 725. In Figure 49 the spectrum shows a distribution of oligomer-potassium ion adducts which yields an average molecular weight of 7i0. Here only the oligomer-potassium ion adducts, (HOICH2CH201nH+K)*, are used for the calculation of molecular weight. This averaged spectrum was obtained late in the Til curve. in Figure 50 an averaged molecular weight is shown early in the Tii curve. The apparent average molecular weight in this case is much lower. in contrast to short-lived polymers, these spectra vary with time. Since this is the case an averaged spectrum should not be taken over the entire Tii curve. if this were done the spectrum would not be 156 6.3.2... wary. 2.. c. 3.... 9.32.... 9...»: mwh 28>... 0:22.252... .3 6332.... we. .6. 92.9.0.2 .3 2...... ~\E onau 09¢ OMB can can 000 OM! 000 on“ can —rP%Pb bP-i-_ -.- —b-pb Pb h—beP—n d i. d 4- “‘1-‘i¢ dd 44.4 I ALISNBLNI BAILV‘IEH 157 .2235. 8:0. 05 s as. 2.008 9...... was 23>... 2.0.2.232... .0 53.02.... mo. .0. 62.0.0.2 .9... 0.5.... ~\E 000« 000 00. 00h OD. 000 00' OD” 000 00¢ —-b-—-Pb- Pb—P- bin—ph- « —1- . << . d 1— ‘ d «a 1— -. —b-b- “ Osh- AilSNEiNI HAHN-133 158 truly representative of the polymer's behavior in the K‘lDS experiment. To explain this behavior we propose the following mechanism. in the K‘iDS experiment, immediately after current is applied to the thermionic probe tip, a number of thermal regions are produced. These are labelled A, B, and C in Figure 5i which is a schematic of the thermionic probe inside the mass spectrometer. Region A is the hottest and reaches some final temperature (Tmax) rapidly. it is this region in which species are quickly desorbed into the gas phase. As a consequence it is this region where analyte is depleted first. Regions 8 (the probe shaft) and C (the source walls) are colder than A. Here region 8 is continually proceeding to Tmax and the temperature of region C, while experimentally set, slowly increases when the probe is hot. (in most cases the source temperature was lOO°C). Gas phase species originating from regions 8 and C would be formed over longer time periods since these regions are at lower temperatures. This behavior accounts for the long-lived total ion currents seen for the PEG and PPG polymers. The following is a discussion how this behavior affects the analysis of these polymers, in obtaining both structural and molecular weight information. What region of the Til curve should be analyzed for polymers which produce continuously varying spectra during the K’iDS analysis? Similarly, is there a spectrum which we can consider as 'the K’IDS spectrum' of the polymer? To answer these 159 2530.. .252.» 2.2.5.. 9:523 005.... m...0.2.......02.m 2.8.. 2.. c. 2.6.... 5.8:... 2.5.5.2... 55.32.... .6 8252.. 2.2.2.8 ..m 0.5.... .0526) Hm. 2.5.o> U 0053.. e... 2 02.6.0”. team .5 «Em nmmnaz . .1 . 18.. i m _ . _ p r!7 . - . i .+ .I i . m... .52... _ _ m. 2.0m .60... 09 w 2 3.5.6 .03.... a .. £03.. 08... 23.820; gammm memo“... “flung .2... a: 303/ .0 8%.... o. uuuunuuuumunu t food. 3.3 Eazocmx \ J. 2.3 c.5205. 05.0.60 .50.. 0.3...8...E...o 13"" 9310314 160 questions Figure 52 shows a curve of apparent average molecular weight as a function of time (scan number) for poly(ethylene glycol) 400 (PEG 400). it appears from this curve that three regions exist in the K’IDS analysis. The minimum in this curve corresponds to approximately the correct molecular weight for PEG 400, fortuitously. Figure 53 provides a means of investigating the behavior of these long-lived polymers during K‘iDS analysis. Here the relative intensities of the oligomer-potassium ion adducts are plotted as a function of scan number for PEG 400. Each ion is shown to behave similarly during this experiment, that is, early in the K‘iDS run a maximum value for each ion is reached. At some time later in the same run the ion reaches another maximum value. Apparently two discrete sources of desorbed neutrals are responsible for this behavior. These two regions occur from a temperature gradient which is produced immediately after current is passed through the filament which contains the thermionic K’ ceramic. This behavior is analogous to the proposed mechanism which identifies the cationizaion process in laser desorption (9i, 97). investigators (9i, 97) have concluded that emission of alkali metal ions occurs from the region directly irradiated by the laser. Cooler regions, away from direct laser irradiation, are responsible for analyte desorption. Cationization is a gas phase process between the emitted alkali metal ions and the desorbed species. Similarly in K’iDS, although the K+ thermionic ceramic is resistively heated, a range of temperatures exists along the emitter probe. it should be noted that the laser in LD appears to 161 ...0....52 zoom .0 .5325. 0 ms 8.. =8»... 2.0.2.522. .o 2.203 02:00.0... 2.2.03. ...0..2...< 2. p N... 052. $9232 Z «6+ 005 .01....— 0..=.0..2.E0... 000......m 28...... S. 0......2 3.5-.. Kiisuaiul aniteiau 189 formed in thermal degradation pathways. The response from phosphorus compounds in the TID has been neglected. Speculation has led to the popular belief that P02“ ls the charge carrier causing the response in the TID due to P-containing compounds (M4). Figure 60 shows the negative ions produced from tri-n-butyl phosphate on the hot alkali aluminosilicate bead. The most intense ion in the spectrum is at mlz 79, mi. The P02” anion is not observed. The phosphate presumably adsorbs on the surface via the lone electron pairs of the oxygen atoms or the electron donating capibilities of the surface. Similar behavior has been observed for the adsorption of organophosphates, organophosphites, and organophosphines on metal oxides (lSO). Hydrocarbons elicit a response in the TID which is often 3 or 4 orders of magnitude lower than that of N- and P-containing molecules. While the response of the hydrocarbons is small in the TID, there is a response. Figure 6i shows the negative charge carriers produced when hexane comes into contact with the hot alkali aluminosilicate bead. The ion series of Call“ stands out in this mass spectrum. Hydrocarbon decomposition studies on metals and metal oxides have shown that a carbonaceous layer (so called 'surface carbide') is produced at high temperatures which is severely dehydrogenated (56). CnH fragments have also been shown to have high electron affinities (57). These observations would support the appearance of the hexane SI spectrum in Figure 190 icon 3092395352 28.: 05 as»: toe—3.5 3232... 339-..-..» 3 52825 .m 2582. .8 95a... .7 ’ ’b ’ b i’ b P’ _ . . ....~._-..;.~_C.Em.........~_i...:..~.r-...n.~ :fi. t..;~. caricmriam: S: be... :2 max L L ’ > b b a: . $5.: . EM: . car 2 .327 .cwm _ l .=.—.: i .2: cm. . . cm. as. t . cm I p . h D ’ Dr il‘ .5. 2. zmsuawl aAuBIaH Mgsuaiul annatau 191 6009 3002395892 28.: 2: 9.25 353.5 2.88: 3 E3525 8 2.3302 .3 2:2... o: on. on on on on on as on on o. o .—-b-—-b-—b--—b-h-—-nb-—-b-—pp-p_-P#—. p—npr—PbL- . . — _ ___.—- ._ _ _ no ifnown L28. 9. r 5....» is"... r on 8.5 9.0 9.0on 192 6i. Our experience is that, in our NSI experiment the overall process of ion formation for alkanes is loo-i000 times less efficient than that for N- and P-containing compounds. This behavior is consistent with the selectivity of the TID and could support a surface-only mechanism. Won A TID mechanism must account for both types of detectors, i.e., those that collect positive species as well as negative charge carriers. The results that are presented here are consistent with a surface ionization mechanism for the positive ion mode of the TID. For positive surface ionization the ionization efficiency is largest when the difference between the work function of the surface and the ionization potential of the adsorbed species is small, i.e. when a high work function solid interacts with an atom/molecule fragment with a low ionization energy. When nitrogen and phsophorus compounds are used with favorable work function materials this condition can exist. High temperatures also favor ion formation in two ways, because of the temperature dependence in those expressions for negative surface ionization efficiency (see Chapter i) and since thermal degradation products (having sufficiently low ionization energies) are formed at elevated temperatures. Triethylamine is used here to demonstrate the possible routes that a molecule adsorbed on the hot alkali aluminsilicate bead can 193 take. A typical positive SI mass spectrum of triethylamine using our experimental approach is shown in Figure 62. I'iolecules are in equilibrium between the gas phase and adsorption sites on the surface. Triethylamine (and in general most N-containing molecules) can adsorb on the surface via the lone electron pair on the nitrogen atom. A similar process can be visualized for P-containing molecules. Once the species are adsorbed decomposition products can form. These newly formed species are also adsorbed on the surface and may exist in equilibrium with their precursors. An induced partial positive charge can cause a weakening of the B-bonds relative to the nitrogen atom. Evidence exists for this behavior (l5i). Thus for triethyamine, the radicals [H‘*(C2H5)2N-CH’CH3] and ['CH3*(C2H5)2N’CH2] are produced. Strong bonds between the radicals and surface atoms can make this a low energy process. The adsorbed species can follow three possible pathways: ( i) an ion can be formed; (2) the neutral product can desorb from the surface; and (3) the species can further decompose. For an ion to form the adsorbed neutral must have a low IP. The selectivity of the TID can be explained in this way. "any compounds decompose on hot surfaces (e.g. hydrocarbons) however, these decomposition products have relatively high ionization energies, leading to undetectably small signals. An example of an adsorbed neutral undergoing further decomposition is shown in equation 66. CH3‘CH'N(C2H5)2(S) " (C2H5)NH‘CH'CH3(S) " C2H4(S) (66) 194 0: no. 00. pnp—pp doom gang—3.35:5: 20x2 05 9.3: 92:33 2.25352.» 3 52.8.5 .m 25.8.. No 959.. nm om mm 0O 05 O ppbb ub-b—ppp _b-b—-.-—pupp—np-phb-bp— ‘ . ‘ 5 mm 00 no on 0? O? Pbbb—nb-h—nbbb_bbbb .- 1 — u:.:. 00. m7... mm a .32... _ a» 195 Similar reactions have been reported to be the intermediate steps leading to neutral products from amines on metal surfaces (i52). From equation 66 the new decomposition products can form the ion [(Czi-is)th===CH-CH3]+ (mlz 72) or perhaps decompose further. Table 6 shows the parallels between these results of positive surface ionization on alkali metal aluminosilicates with prior surface ionization results obtained on oxidized tungsten. The surface temperatures in both methods are comparable. Clearly our results are in good agreement with those reported earlier with only small variations. We do not necessarily expect the surface ionization products in this experiment to precisely match those previously reported since the two systems are chemically different in a variety of ways. The previous discussion supports surface ionization as the dominant ionization mechanism in the TID. Decomposition can occur on the hot alkali aluminosilicate. in a surface-only mechanism, decomposition on the surface is highly important for the production of species with low ionization energies. A large body of literature exists for positive Si of organic compounds and has been recently reviewed (see references cited in Chapter i). in most cases this literature refers to studies involving metal or metal oxide surfaces. Results using the alkali aluminosilicate thermionic bead closely parallel positive SI experiments on metal and metal oxides. it would appear from the surface ionization studies (both postive and negative) on the alkali aluminosilicate bead a surface-only mechanism is a possibility. 196 3.2.3.9.. 1a.... u 5.-....2 grins-=2 u N..... .......N..-..z u ...,...n..... 2.... \ N.. . u 2 .fl . / 9...... .3... " 2.3.5.. 3.... n. .\ ...:.. .... . z. I /....~... 13...... n 2......NC: 1:23.33... 1...... n ...-.-........ ._....~...-..z u N:3. .3... - :c ._z - F. 3.... ._..._~......z - .....-....... ._~..z~.....~,.._ 1...: u .....-........ 1.2.2.328. 222:... 3.238.. né a... a... 2... Q4. =3. 8.8.... 2388.. .o 2 a a .2 1.... 27..-... 1...... z. .2... ,2. QA=.=V on—_o—‘~ .23...-2. ..-.a-p¢~ .272. 1...... .272. .222. no vv 0... up ca ma 3.: .6:— vv 2.... 2:. m... 1. vv av 5.2.5. .. 63.2.5 a}. u so 998...... 2o. .2 12.5-... 1..-.....-.-n=uo-z. 1.....«0-2. 1.3-...0-2. 125-2. 1...... 1..-... +2.3... 1......0-2. 195-... 1..-... 1...... 1.7.... 1...... $98— 0.I£._§.c.:-< 35.83.... .8 v0.5...— n.8- nan}: 0...: _=< 2:152 0565.223... 2.4.... 2:522:22 mvuzz 05:... 2:”. 9:56.38 «32...... 5.35; 352.6 ecu 33:32:52 :32 3.2.: .3 «2.2.2 2:3 .6 533...... 83...... 2.3.2... .o 52.2.83 9 03o... 197 W The compounds studied in the above surface ionization experiments on the alkali aluminosilicate material can behave differently when introduced in the GC-TID. up reponse is observed for these compounds when the (SC-TID is operated using no detector gases (other than 30 mllmin He) and high current settings. it would seem that if a surface-only mechanism is operative a response should be seen in the TID for these compounds. This may not necessarily be the case since detector gases not only influence the gaseous environment surrounding the bead but can also influence the surface. Next we will attempt to answer the next question: What is the role of detector gases in the TID? We will discuss the influence of several gases on the response of N- and P-containing compounds in the riS-TID. The low pressures used in the mass spectrometric experiments would not likely support a 'reactive zone' near the hot bead. (However, high flow rates in the TID may not allow a 'static' reactive zone to be maintained either). The mass spectrometric studies would be useful in identifying what surface effects these detector gases have on the hot thermionic bead. This behavior then will be compared to results using the 6C-TiD. The lack of response in the 6C-TID in the absence of other gases (other than the analyte) suggest that the gases are extremely important to the response mechanism. The following investigation in this reaction will be divided into the categories of mass spectrometry (HS-TID) and gas chromatography (EC-TID). WELD. initially the effects of introducing air with the organic vapor into the ion source containing a hot alkali aluminosilicate bead was studied. The introduction of air in the mass spectral experiment had three effects on the results: lichanges in the total ion intensity (increase or decrease of product ions); 2) change in the product iondistribution; and 3) a 'renewal' of the surface ionization characteristics of the thermionic bead. In general, the compounds giving the largest increase in total ion intensity were halogenated compounds. These compounds generally give responses 20 to 30 times higher when the pressure in the Cl volume was doubled by using air while keeping the pressure of the organic compound constant. The increased response shows an increase of each ion signal in the spectrum without any change in the relative abundances. Other compounds such as acetonitrile and triethylamine yielded reduced responses 199 on addition of air. This behavior at higher pressures may be in part due to the cooling of the alkali aluminosilicate bead since the bead is heated under constant current conditions. To investigate what role an additive gas (e.g. air) has on a surface-only mechanism, consider the simple model which is represented by equations 67-69. Mg) + H 4 NH (67) NH -» 811 + on (68) a-n, c-n -» 3*(9) + 6(9) (69) Equation 67 considers the adsorption of molecule A to some surface site l‘i located on the thermionic ceramic. in 68 decomposition of A on the surface forms adsorbed decomposition products (811 and (:11). if the electron affinities of the adsorbed decomposition products are sufficiently high these products can surface ionize. When the thermionic emitter is correctly biased gas phase ions are observed. Using this simple model the effects of adding air to the mass spectrometer will be investigated particularly focusing on each step in this process. In a surface-only mechanism the addition of air may influence the surface by changing the work function. in our model the work function would be only important in equation 69. it has been shown that on metals (e.g. tungsten) the addition of 02 increases the work function of the surface (7). Some positive Si methods in mass spectrometry are performed with the addition of 02 into the 200 source (35). Here the addition of air appears to lower the work function of the surface which would favor the production of negative ions. This apparent reversal in behavior upon the addition of air is likely due to differences between the surface characteristics of metals and of the alkali aluminosilicates. if the addition of air to the Si experiment discussed here only served the purpose of lowering the work function of the surface we would expect to enhance the response for compounds such as triethylamine. This is not the case, triethylamine has a reduced response upon the addition of air. Evidently air can influence the surface in a manner other than changing the work function. The adsorption 0f 02 (air) on the surface can affect the adsorption of other species. This behavior pertains to equation 67 in our model. For surface ionization to occur the neutral species (molecule A) must be adsorbed onto the surface. Adsorption of two different neutrals can influence the adsorptive behavior (ref). This behavior results in either interactions between the two coadsorbed species or one species blocking adsorptive sites (l‘i) thus decreasing the concentration of the other adsorbate on the surface. Coadsorption is one phenomenon which can inf iuence the adsorption behavior of a molecule on the thermionic ceramic. Other factors are the thermodynamics of adsorption (heats of adsorption) and types of adsorptive behavior (molecular or dissociative). Additive gases such as 02 may alter the surface here as well to change these adsorbate-surface interactions. 201 Figure 63 shows the effect of adding air with tri-n-butyl phosphate. The rich chemistry of phosphorus compounds in general causes difficulty in certain peak assignments, however it is important to note the change in the relative intensity of each ion in the spectrum. Changes in the adsorptive behavior or work function of the surface would only serve to change the intensity of the ionic species. Compounds such as CCl4, a halogenated hydrocarbon, in the presence of air shows an increased total ion intensity without any changes in relative intensities of the observed ions (Cl‘, Ci2’, CCl2', CCl3', CCl4"). The change for this P-compound would seem to indicate a process other than those implicated in equations 67 and 69. it would appear that the environment at the surface of the head has changed so that it is taking a chemically active role in the production of ionic species. This process would be described best by equation 68. The effect of surface oxidation or catalytic combustion is well documented (lS3). This process is often facilitated on surfaces by the addition of air. Catalytic combustion 6C detectors were commercially available during the i960's (lS3-l54). At the time their ionization mechanism was a subject of debate. it is beyond the scope of this work to discuss in detail these detectors, however their behavior parallels our results in the mass spectrometric experiments involving the addition of air. The addition of air appears to affect a variety of processes-adsorption, decomposition and surface ionization. It may not be possible only to get a change in work function without 202 ....< .6 top... 8. .6 9.3.3.... 3.2.... a 5.3 2.533 83 5:38am 2.... .23.. 33585.32 :22 2: 9...»: BESS 3238.... 3.3.1:: .3 8:533 .m 2.336: .3 2.5.“. :2 a... EN .3 S. a? a: E. .2; w 2.. a... a... Tr.lr.lr.....li..r.r...:girlq .Il.r».l_....r..-..xir. m 1.0.... I :2. .. £..--.....:..m.:,a..._.:......_.-;..£. 2.. 2F 3. j 2. a. 1.0.. . S .. Kusuaiul aAnatag Ausuawl annals}; 203 a corresponding change in the adsorptive behavior of the surface to a compound. Qualitativeiy however, the simple model as suggested by equations 67-69 does well to explain our findings. The 6C-TID is most widely used with hydrogen and air as detector gases. These gases, by themselves, only produce an ion current at very high bead currents (>5 A) within the mass spectromter. The lifetime of the thermionic probe tip housing the alkali aluminosilicate is short at this current setting and no useful information can be obtained. The ion current at this high bead current consists solely of OH' ions. The background current in the TID has been surmized to be due to the hydroxyl ion and our results would seem to support this view. When N-, P-, and chlorinated compounds are introduced with these two gases the behavior is identical to that obtained upon the introduction of air. Phosphorus and chlorinated compounds show an enhanced response and any spectral changes (e.g., trimethyl phosphate) are the same as with the addition of air. At no time was the 0H“ ion seen with the other product ions in the spectra. The lifetime of the bead is short at high currents so that no conclusions can be made at higher temperatures. The observations here are nothing more than those obtained for air. Apparently the effect on the response of any compound in the . presence of radical species of hydrogen and oxygen cannot be studied here at these pressures. While this is discouraging the importance of pressure becomes evident. Pressures greater than I t0" are needed for the COII'IDUSUOI'I 0' H2 at reasonable bead 204 currents (<5 A) and as will be shown later the combustion of H2 must in some way take part in the production of charge carriers in the 6C-TID. Next we will characterize the GC-TIO in terms of detector gases. This is necessary in order to make a comparison between the HS- and 6C-TlD's. W Before discussing the role of various detector gas systems the role of any gas on the bead temperature will be discussed. This can be done here since the gases' affect on the temperature of the bead was independent of the gas used as will be seen. in addition, detector gas-bead temperature phenomenon will be needed in later separate discussion of detector gas systems. Wilma. The detector block in all the 6C-TiD experiments is 300°C. as suggested in the operation manual supplied with the TID. Figure 64 shows the effects of changing flow rates of various detector gases on the temperature of the bead. The temperature was measured with an infrared thermometer and occasionally checked with an optical pyrometer when applicable (above 750°C). The data used in Figure 64 varied by no more than i0°C. Figure 64 shows that the temperature of the bead changes with flow rate independent of the nature of the gas used. Only the top of the TID head is visable and it was this location from which the temperature was measured. This would suggest that this phenomenon is physical and not chemical. in 205 °2 ir soo- “2 33 S s E . . .- i 600 - 260 460 640—0 850 jo'oO Flow Rate (mi/min) Figure 64. Bead Temperature as a Function of Detector Gas Flow Rate. 206 addition, when various pressures 07 H2 are used With air or 02 no change is observed for either curve. This is not true when sufficient H2 is added to bring the H2-02 system above the explosion limit. At this point a true flame is established. As can be seen from Figure 64 the temperature of the beade is below 600°C. when no detector gases are used (except for ~30 mllmin He as carrier gas). Under these conditions no response is observed from all the test compounds used here. These observations would suggest that, in part, one role of the detector gas appears to in some manner increase the bead temperature. A physical explanation which would support our observations involves laminar flow of the detector gases around the TID bead. Consider the configuration of the TID used in these studies (see Figure 65). The TID bead is located within the detector cylinder. The TID head is directly above the flame tip and separated by a distance of approximately 3 mm. The sample, carrier gas, and H2 are introduced through the flame tip. Other detector gases (e.g. 02, N2, and air) are introduced through small holes at the base of the flame tip. A schematic of the TiO used here is shown in Figure 65. The flow of any gas into the detector via the flame tip has no influence on the bead temperature. This is not true however, when the flow exceeds 80 mllmin. (The head temperature decreases). it is the flow of gases introduced at the base of the flame tip which affects the temperature. For this case laminar f low would produce areas of low pressure around the 207 e-coLLECTOR THERmomc asap —’ 3 3 mm K P (— FLAHE TIP p4— —¢q w AIR OR N2 AIR OR N2 H2 AND CARRIER GAS Figure 65. A Typical Configuration of a TID. 208 bead. Coliisional cooling of the bead would produce areas of low pressure around the bead. Coliisional cooling of the bead would diminish in these areas of low pressure which results in an increase in temperature. The detector configuration does not permit observation of the entire bead. One would expect however, the bottom of the bead would be cooler due to its bombardment by the effluent from the flame tip. This would suggest that the temperature at the surface of the bead is not uniform and may vary by several hundred degrees Celsium. How can these different temperature regions affect the TID response? The adsorption, decomposition, and surface ionization of neutral species are all temperature dependent in a surface-only mechanism. Varying temperature zones could also affect a gas phase decomposition mechanism by dictating the concentration and composition of the reactive gas phase boundary surrounding the bead. it will be shown shortly that the TID response is dependent upon the flow rates of the detector gases. in addition to changes in bead temperature, flow rates can alter the composition of atmosphere surrounding the bead. Both these factors (temperature and composition) must be jointly discussed with changing flow rates of the detector gases since they cannot be separated. It will be shown that in some cases the temperature of the bead or the composition of the atmosphere surrounding the bead is more important to the detector response. Several experiments and definitions must be clarified before 209 discussing the behavior of the TID with respect to different detector gases. in these experiments the response will be reported as peak height (Amps) per mole compound injected. The selectivity (response of one compound relative to another) will be determined by the ratios of the response of the two compounds in question. Generally the selectivity of the TID is calculated with respect to a hydrocarbon. Here the units of selectivity will be a mole ratio. There are three types of experiments which will be used in these behavioral studies: l) response as a function of flow rate keeping bead current constant; 2) response as a function of bead current keeping flow rate constant; and 3) response as a function of injection number. The first two types of experiments are self explanatory; the third involves injecting an analyte repeatedly with constant time intervals in between each injection (typically l-2 minutes). These intervals are timed from the point after elution of a compound to the next injection. in addition to these three experiments a compound may show irregular behavior in regard to peak shape. These experiments will serve to characterize the TID response for various compounds with respect to different detector gases. The first detector gas which will be discussed is air. W Throughout these studies several model compounds are used typically having one functional group which contains the N or P atom. l'iolecules such as malathion, methyl parathion, and azobenzene have been used previously to characterize the TID (7i). Here simpler molecules are used. For 210 some of these simpler molecules data such as electron affinities and decomposition (surface and gas phase) are more likely to be found. In Figure 66 the behavior of several test compounds in the constant flow rate study is shown. Here the air flow rate is a constant l60 mllmin. No other gases are used. A plot of response versus current, when plotted logarithmically, is linear. It is important to note that heating current is used instead of temperature due to the likely existence of a temperature gradient along the bead. Therefore the relationship between current and bead temperature is unclear. What is known is that bead current is proportional to bead temperature is unclear. What is known is that bead current is proportional to head temperature; further increasing the current increases the average temperature of the entire bead. Several observations can be made from the curves in Figure 66. The slope of the lines for the functionally similar compounds (p-nitrophenol, p-nitrotoluene and Cn-Cn) are nearly the same. This would suggest the role of air as a detector gas is dependent upon the functional groups on the molecule. . As a consequence each compound with a different functional group should have a different slope as is the case here. The different slopes are responsible for the change in selectivities over the current range shown. For example, p-nitrophenol has a higher response at lower currents than tributylamine; after 3.6 A this behavior is reversed. The influence of temperature can be large as is the case for tributylamine where the response changes over 2" Propylcuorue . Thai-3mm Cymide " o-Nitrophmoi E - mum 3 -. 17 0 / I V -a . . v f} —3'! 3'! 41.5 BEAD CURRENT (A) Figure 66. Plot of Log Response as a Function of Bead Current For Several Compounds at a Constant Air Flow Rate of l60 mllmin. 212 4 orders of magnitude. As shown temperature has a profound effect on response. One test compound, trimethyl phosphate, could not be plotted in Figure 66 since the response varied upon repetitive injections of the compound. The peak shape of this compound was symmetrical over a large range of sample sizes. it was concluded that this behavior is not a column effect but is detector related. Response for trimethyl phosphate is plotted as a function of injection number in Figure 67. These results are obtained at 3.2 A bead current and a i60 mllmin air flow rate into the detector. The elution of trimethyl phosphate is equally spaced by approximately 2 minutes. The increase in response for this compound is nearly linear up to l4 injections where a plateau is reached. After 3.6 A bead current trimethyl phosphate shows no dependence upon injection number. We believe this behavior is a surface phenomenon. The possibility that reactive gas phase species (created when the compound is introduced into the detector) could remain in the detector for several minutes is small. At this flow rate (i60 mllmin) and considering the design of the TID, a species remaining in the gas phase after eluting from the 6C column would reside within the detector for no longer than 5 seconds. The time interval between each introduction of trimethyl phosphate into the detector was 2 minutes. The modification of oxide surfaces by phosphorus compounds has been recently studied 050). it was observed that 213 RESPONSE (A/MOLE x 105) TRIMETHYLPHOSFHATE Figure 67. Response as a Function of the Number of in jections For Trimethyl Phosphate. 214 decomposition of these P-compounds on the surface led to gas phase species such as ethers and hydrocarbons. No gas phase P-containing molecules were found. Decomposition therfore, led to a modification of the surface by various phosporus compounds which are mostly oxides. Above a particular bead current, repetitive injections have no enhancing effect. Apparently high surface temperatures serve to keep the surface 'ciean' by not permitting any build-up of P-containing species. The modified surface has an effect on the response of other compounds. Compounds eluting after trimethyl phosphate show a diminished response. Conditioning the bead at higher bead currents apparently 'rejuvenates' the surface to the test solutes‘ original responses. The chlorinated compound used here (propyl chloride) also shows behavior similar to trimethyl phosphate. Below 3 A bead current the response increases with repetitive injections similar to that shown in Figure 67. in contrast to trimethyl phosphate the other test solutes show an enhanced response on elution after propyl chloride. With air as the only detector gas the use of chlorinated solvents has been suggested as a response enhancer in the TID (l4B). increasing the bead current for this case also rejuvenates the surface to its original condition. The opposite effects that propyl chloride and trimethyl phosphate have on sample response is most likely a reflection upon the various ways a modified surface can change a compounds response in the TID. For example, a decrease in response would suggest the 215 obstruction of adsorptive surface sites or an increase in work function. An increase would suggest a decrease in work function. in light of these results, care must be taken in quantitative analyses of a singular component from mixtures where a caiibaration curve is used. it is advisable to use internal standardization. In Figure 68 several test compounds are plotted against air flow rate at a constant 3.6 A bead current. The dependence of response on flow rate can be categorized into three groups. Tributylamine and nitrobenzene produce curves which have a definite maximum. Trimethyl phosphate shows an exponential decrease in response with flow rate. Not shown but similar to trimethyl phosphate is propyl chloride. Oodecane exhibits the third type of behavior in which response is nearly linear with flow rate over an extended range. (Response for dodecane falls rapidly at very low or high flow rates). in general, the degree to which the sensitivity changes with flow rate is smaller than with bead current. The curves for tributylamine and nitrotoiuene in Figure 68 would suggest a temperature dependence. The shape of these curves parallels the temperature dependence on air flow rate shown in Figure 64. The response for all compounds studied depend upon temperature. For instance, below approximately 50 mllmin air no response (i.e. response is less than leO'6 Almole) is observed. The response for trimethyl phosphate and propyle chloride would appear to depend upon more than the temperature 216 § BESPONSElA/MOLE x 105) .g/ L Nibotoiuene / \cu 1,, w W w —:3r—— AIR FLOW RATE (mi/min) Figure 68. Response as a Function of Air Flow Rate For Several Test Compounds. 217 of the bead. What this suggests is that trimethyl phosphate and propyl chloride's response is dependent upon the concentration of detector gas (most likely 02). For the phosphate, as stated earlier, oxides of phosphorus can form. The chloride's response may be increased by a reactive mechanism such as that for the phosphate or perhaps 02 can increase the sticking probability of this compound on the surface. in the case of dodecane it is unclear what effect causes the behavior in Figure 68. The steep fall-off region suggests temperature is important. it also appears once a particular bead temperature is reached the dodecane response is independent on air flow rate. Several important points can be made in summarizing the characteristics of the TID when air is used as the only detector gas. Temperature appears to have more of an effect on the test compounds' responses than the flow rate of air. Selectivity can vary greatly with bead current (bead temperature) with a reversal of selectivity observed for some compound pairs. it is also apparent that the TID response can be dependent upon the surface species on the bead. Trimethyl phosphate and propyl chloride when injected can change its own or other compounds responses in subsequent injections. This behavior diminishes with increasing bead currents therefore one role of temperature may be a cleansing mechanism which maintains a constant response. Finally, the sensitivity of all these compounds are much higher when a mixture of H2 and air is used as the detector gas. The 218 H2] air system W111 now be 015008580. MW The TID is MOSt Often used with a mixture of air and H2 as detector gases. in the detector used here H2 is introduced to the thermionic bead via the flame tip (see Figure 65). if H2 is introduced along with air into the detector at the base of the flame tip a much different response is obtained. For example, a higher H2 flow rate is needed for optimum response. These experiments involve some changes in the 'normal' TID design. These experiments will not be discussed here however, since they are incomplete. These experiments are justified since such changes do bring about different responses and can be used to aid in elucidating the TID mechanism. Here we are concerned only with an unmodified commercial detector. The optimum response of the H2/air system at a given flow rate and bead current is strongly H2 dependent. Figure 69 shows the response of a number of test compounds plotted versus H2 flow rate at a constant bead current and air flow rate. Very small changes in H2 dramatically change the sensitivity of the detector. it appears that for the compounds studied here there are two different optimum H2 flow rates that depend upon whether the compound contains nitrogen or phosphorus. The nitrogen containing compounds reach a maximum value and then 219 “1 e17 xzooo Wm —mm 5... E m Ak mum, 4.0M. 2 —4L é 5 i, nan-um Figure 69. Response as a Function of H2 Flow Rate For Some Test Compounds at Constant Current and Air Flow rate. 220 decrease. in contrast to this behavior the phosphorus, hydrocarbon, and chloride compounds show a plateau or no maximum at all. All the test compounds show a substantial increase after a small addition of H2 into the detector. Above 8 mllmin hydrogen all compounds begin to increase in response. With the additon of H2 above this limit a true flame is produced and the background of the detector increases to a point at which useful information cannot be obtained. in addition to an increase in sensitivity H2 can influence the peak shape of particular compounds. Figure 70 shows two Chromatograms which are obtained under identical condition except for different H2 flow rates. Once again only a small difference in hydrogen flow rate has a profound effect on the results. Here higher flow rates favor symmetrical peak shapes for the two compounds containing phosphorus. After i mllmin H2 the peak shapes for these two compounds become symmetrical and baseline resolution is obtained. No effect is seen on azobenzene or the hydrocarbon when changing the flow rate of hydrogen. The same type of behavior, comparable to the P-compounds, is observed for propyl chloride (not shown). Column effects are unlikely since the chromatographic conditions are identical. Apparently at particular Hzlair mixtures P-containing molecules and chlorinated compounds interact at the surface of the bead which causes the tailing in their respective peaks. It would seem 221 Methyl Parathion Malathion VK .1 E 1.2 mi/min H2 4.4mi/rnin H2 Figure 70. Chromatograms of Test l‘iixture at Two Different H2 Flow Rates. 222 that one role of hydrogen/air is to keep the surface clean and this can only occur at the appropriate ratio 0t H2 to air. Changing the air flow rate while keeping H2 and bead current constant does have a small effect on the sensitivity of the test compounds (Figure 7i). As for air in the absence of H2 the flow rate would appear to follow the temperature of the bead shown in Figure 64. Unlike before, when air alone was used, all compounds show the same behavior. The response curves for P- and Cl-containing compounds are similar to those of the N-compounds. This behavior can be observed for all H2 flow rates and bead currents used here. The presence of H2 therefore, radically Changes the response of P" and Cl-compounds in the T"). The optimum air flow rate in the presence of H2 approximately corresponds to the maximum temperature observed (see Figure 64). Bead current has the largest effect on all the test compounds responses as can be seen in Figure 72. The response of a compound using the H2/air detector gas mixture can vary by over 2 orders of magnitude when the bead current is increased. From Figure 72 it is apparent that hydrogen enhances the TiD's response by several orders of magnitude when compared to the air-only case. Two distinct differences between H2/air and air are evident from Figure 72. For hydrogen/air the curves are no longer linear but are curved. As suggested by Patterson et al. RESPONSE (A/MOLE x103) 223 flrmefififlflfime Nitrotoluene Tribmviamine c" x 10" l 100 1150 200 250 AIR FIOW RATE (mi/min) Figure 71. Response as a Function of Air Flow Rate For Some Test compounds at constant current and H2 Flow rate. 224 2 - Trimethylphospime l Malathion ‘— Meihyl Paratfion —-Tributyhmhe Azobenzene Log RESPONSE 553 470 BEAD CURRENT (A) Figure 72. Log Response as a Function of Bead Current For Several Test Compounds. 225 (MB) under these conditions the response should have a Amp"2 dependence upon current, i.e., a plot of response (detector current) versus i/(amp)2 (bead current) should be linear. This was suggested when the heat transfer properties of the TID was compared to the thermal conductivity detector. Their findings are based upon the bead being of uniform temperature and also that current showing a nonlinear relationship with temperature. Our results (when H2/air is used) is in agreement with Patterson's. With air only, the response is linear with current. Adding H2 does not affect the bead temperature (measured at the top of the bead) and the same temperature dependence (at the top of the bead) on current is observed. From the two kinds of curves produced it would seem that two different processes can occur depending upon the composition of the detector gases. We cannot say presently that this is evidence supporting decomposition on the surface of the bead or in the gas phase but that more than heat transfer properties must be considered in the TID. The second major difference in the constant flow rate experiment between the two detector gas systems concerns selectivity. When biz/air is used the selectivity remains nearly constant with bead current. This occurence is due to all compounds having similar response curves. The selectivity is much different when only air is used due to each compound (with different functional groups) having different slopes. (Tributylamine (NR3) behaves differently than benzyl cyanide 226 (RCN) with air but their response in hydrogen/air is nearly the same). There is also a dependence on flow rates on selectivity with both detector gas systems. For air only the P-compound's selectivity (mole trimethyl phosphate/mole C17) continually decreases contrary to the N-compounds. With H2/air all compounds show a maximum value for selectivity (compared to the hydrocarbon). This behavior along with the other experiments from both detector gas systems suggest a very important point concerning the behavior of the TID. When air is used the response appears to be very compound (i.e. functional group) related, i.e., the environment in which the N or P atom is located in the molecule determines the TID response. For the H2/air system this structure dependence no longer exists. All P- or N-compounds behave nearly in the same fashion. In summarizing the H2/air detector gas system we must consider the fate of these gases in the no when they come near- or in contact with the hot surface of the thermionic bead. The production of radicals from these gases occurs. This simple verification that water is produced can be performed by holding a cold finger over the exit port of the TiD. in order for water to be formed the combustion of H2 must take place. The combustion of H2 involves the production of a number of radicals such as H, 0, and 0H. Among these radical species the 0H radical has a sufficiently high electron affinity to surface ionize on the bead. 227 Even at low pressures in the mass spectrometric studies the OH' species was observed. The background current is highest in the TID when the H2/air system is used which may suggest that it is OFF responsible for this background emission. The influence of these radical species on any analyte introduced into the detector is unclear. The decomposition of any compound usually begins by the formation of a radical species. Decomposition can take place by pyrolysis (strictly a thermal process), combustion (gas phase or surface) or by a catalytic process on an appropriate surface. These radicals can influence any of these processes and as yet one process cannot be singled out to explain the TID response. We have shown that the H2/air mixture can apparently influence the surface to affect peak shapes for P- and Cl-containing species. While this is true we cannot claim that decomposition of these compounds takes place solely on the surface. Coupling these possibilities with a thermal gradient along the bead complicates any explanation of the TID response. Low or high temperature processes can occur in both the gas phase or the surface. For example, the gas phase process of autoxidation (which leads to the production of radical intermediates) can occur at temperatures below 300°C. Surface processes such as pyrolysis or catalysis can occur at temperatures even below this, e.g., the pyrolysis of nitromethane which produces N02 (EA. 4.0 eV) can occur at 200°C (:49). At best, what can be said about the hydrogen/air system is that 228 decomposition is likely a combination of all three processes (catalysis, combustion, and pyrolysis). it should be said that regardless of the decomposition mechanism, the production of electronegative species in the H2/air system is also more efficient than when N2 is used as a detector gas as will be seen from the following discussion of this system. WM Consider the role of the earlier detector gas systems. Air and H2/air can produce chemically reactive species or modify the surface of the TID bead. The gases, regardless of the action on the response via the surface or the gas phase, appear to actively participate in the response mechanism. Nitrogen, contrary to these gases, is nearly inert and no such participation can be envisioned. The role of nitrogen is to maintain an inert environment on and around the TID bead. In this case, the behavior a compound exhibits is mainly due to its interaction on the various thermal regions of the thermionic bead. Here the pyrolytic behavior and the compounds electron affinity would be extremely important. The inertness of nitrogen would also suggest that the behavior of the GC-TID should parallel our results obtained in the HS-TID (no added gases). As will be shown from the following discussion this is typically the case. Under constant current conditions the results using N2 as a detector gas were similar to that of H2/air. Each compound studied had a maximum value which nearly coincided with the maximum temperature obtained (Figure 64). As would be expected 229 no dependence on the gas is observed. Here, temperature should be the dominant factor determining the response of the TID using nitrogen. Figure 73 shows the dependence of several compounds under constant flow rate conditions. The optimum flow rate corresponding the maximum in Figure 64 was chosen. it should be noted that the curves are linear similar to the air-only case discussed previously. in addition the magnitude of the response is on the same order as that of the GC-TID when air is used. At low currents the response parallels the electron affinity. That is tetracyanoethylene>dinitrobenzene>chloronitrobenzene which have respective electron affinities of 66.5, 43.5, and 27.5 kcallmole. The response of these compounds also parallel the total ion intensities obtained in the i'iS-TID. For the most part, at low currents with the mass spectrometer, molecular anions or ionic species which have not undergone substantial decomposition (ii-H, i'l-NO, and ri-Oii) are observed. When the current is increased this no longer holds true and species such as CN" begin to increase in the mass spectra. Similarly in the GC-TID when the current is increased the response may no longer depend upon the electron affinity as can be seen for chloronitrobenzene. At higher currents this compound shows a greater response than dinitrobenzene. The selectivity using the nitrogen mode of the GC-TID depends upon which current is used. At low currents response parallels the electron affinity of the molecule. At high currents response depends upon how decomposition takes place to 230 'lbiracymoeihylene Tribufyiamine -3} Chlomnitrobenzene g DiMMene fl) w -4 - O 3 .5 p l l 1 14 3.6 *5'3 (IF BEADCURREN'HA) Figure 73. Response as a Function of Bead Current For Several Test Compounds With a Constant i60 mllmin N2 Flow Rate. 231 produce fragments with high electron affinities. Hydrocarbons can elicit a response at very high currents and it has been shown that two isomers of octane (n-octane and iso-octane) behave differently in this mode of the GC-TID. The nitrogen only mode of the GC-TID can be used to increase the selectivity between various analytes. At low currents responses are only obtained for compounds which have a high molecular electron affinity. in this case the selectivity of the GC-TID can be very large (e.g. substituted nitrobenzenes compared to hydrocarbons). When the current is increased the molecular structure and any functional groups on the molecule determine how it will decompose and to what species. if these fragments have a favorable electron affinity they will surface ionize and give a response. Therefore the response is decomposition dependent. Hydrocarbons can have an observable response here since supposedly upon decomposition can form species such as c" and CnH which have high EA's. We have discussed our results in the two systems (the mass spectrometer and the 6C) under a number of various conditions. The following question attempts to coordinate these results and make conclusions as to the mechanism of the gas chromatographic thermionic ionization detector. 232 ‘ i I _ . .. . .. . - . - .ii- . l n . . . i i I all i We believe that presently there would be no argument as to the ionization process within the TID. it appears that surface ionization can be the only possibility over any proposed gas phase ionization mechanism. Thermodynamically, gas phase ionization is improbable at the normal operating temperatures of the GC-TID. The pertinent question is how a molecule behaves prior to ionization? Does the molecule decompose and if it does where; the gas phase or the surface? The following is a summary of the results obtained here. A concern between the two studies is the pressure differential of the mass spectrometer and the 6C. The pressure within the mass spectrometer (290: 5° E—CHUOQ” ”A: +2 .ON- 6530.....— 001 0" ON' 00' 000 can 0V0 0N0 DON DON ODN OVN ONN OON 00¢ 00« 0'.“ ON« _..b._-b.-—-mm._.-p.—-..P— Dnn VNM OBN VON unb-_-Pb—-:-i—Pnhb_hnnp —h-hb—b-bb i‘ II 1 (di it (it 240 .33; 152.. .o 52625 more. .R 95...: can cm? on! 0" ON! 001 can can ovn ONO can OON ODN OVN ONN OON 00¢ 00.“ O'« —-hpr-pn—E- —-b-— up —--h—- P—nhbh—bb-b —:b- ——--—b-hb—--b—b:-—bb-b . l l ' ll i 11 l 11‘ ‘ #‘ VON man can I Nd' 241 .32.: 25.5: no 52325 mofx .3. 0.59“. can omN DON OhN OQN DON OVN OMN ONN OdN OON 00¢ 00« 05¢ 00¢ 00“ OVd d “‘ t t . t ‘ ‘4‘ ‘t t‘ t ‘d “{ t-“ 1‘“ t‘ 1 I‘. —--—pnP-—>Pmmm—-bl:—ppbw—--—-b-—-\= VNN NON n..-—->-in—hbl-im—-<-i-—-- Eh...» NG« “‘.i J ‘1 t‘.‘ 242 62.8.5: .3 Eacuuoam web. .05 0.59“. OON OVN ONN h'n «an OOM OON MON OON O'N ONN buPPn-nn—h-np—h-bL th OON OO« 00« OVd __ _ _ 243 OOV VON .< 0:3 0:2 .3 53.5025 morv. .OO 0.59..— OOM OON OVN non OVN ONN OON ONN OON OVN ONN OON OD“ 00¢ OVn thh—PbPh—np-b—nP-b—P-nb—-bb- _ . ‘ (Id! «((1 . it (I 1! Km «on .. 244 .3: £88: 3 223on more. .8 8:3“. OOV OOV OVV ONV —b-P-—-bn——-p . i l 4 J ONV OdV OOV OON OON OVN ONN OON OON OON OVN ONN OON ON“ OOu OVa ~—P---_--—nbbP—-b-—-bpn—pbbb—P-pb—pppb—pnnp—pnbn—nppn_--—b-- . .- i i i i l 1 i 245 .Au—Om NZV OOOL@ —OOOLUOEOLD 5O EOLHUOOW “A: +v. .N0 0530.“. ONO OOh OVN OON. OOO ONO OOO OVO OOO OOV ONV OOO OVO OOO OON ONN OO¢ «OV ONO OVO «OD 246 OON— —OCO.£OOLO—£U hO Eur—HUGO” ”GP-V. .MQ 0530.“. OuV OOO ONO OOO ONN OqO OON ONN OON ONN OuN OOu ON.“ OOn NON l OON NON no“ 1 n3 n0 OON l «NO 247 .OC—CNLhNOOCOp-Q bO EOE—OOO” WQ-+¥ .V0 0530—“. OdO OOV ONV OOV ONV OdV OOO ONO OOO ONO OuO OON ONN OON ONN OuN OOu ONw OOu tun—nbbn—hhbh_-b-P—-Ppb—-b:t_-b-—- bbP-n_hnnp—-P-n—bnhn_D-nb—nb-p—P-nn—b-nb .‘ d‘illli 1“ . y. I 4 \- NON NVO 248 655.5 .6 53.50on web. .mc 0.59“. OuO OOV ONV OOV ONV OuV OON ONN OON ONO OdO OON ONN OON ONN OuN OOd ONd OO¢ —bb-—-bp-—-pb—--p—h-b—-Pn—;bb— Put—b-bn—nh-n—nph-—b-phi—bbnp—P-bP—D-n-_Eh—Ppnb , 1411 1 44 i 41 1, «I 11 i i 4 VON 249 .050th so 55.5025 marv- .OO 0.59“. ONO OOV OVV OOV OON ONN OON OVN OON _-.._.b-b—...._..LF—L..._.p.._...._ OOd .— w......,._ A _ ON« Jh b J‘ - OVn NO“ OOu OO r 250 .5532: 3 5:36ch 8C. .3 2:9“. ON-V OO_V OOV OdV OOO ONN OON ON-O O«_O OON ON_N OO-N ONN OuN OO¢ ON« OOu :nbp- PF—P :hbbp bb— DP: _b-h—bbbh—-P l- (1:1— ii 44-“ (1.11 lui 1i‘1144141- I OONuON OOV OuV 251 6555.23»: .3 83325 wary. .cm 0.59“. OON ONN OuN OOu —--—--—_-h—br . . u- - ON« ONu OO« ONv _ _ ONd Ouw OO« OO ON OO ON ..P._p.-P 252 dance—2O i539: omiu— no 35.5025 marv- .OO 0.59...— OnO ONN ONN OOO OOO OuO ONO OOO OOV OOV OvV ONN ONO OON OON O«N ON« ON“ ON —- b—b--_nb-—P-bn—nnnu— .l y. __ _ .. _. ONO «OV .ON «OV «VO 1 ON «ON T «NN ”X l NVd 253 ._OO— 20am .3 53.5025 OED. .OO 0.59“— OOO OOV OOV OVV ONV OOV OON OOO OVO ONO OOO OON OON OVN ONN OON OOn OOw OVfi NON ONN avqann NON NNN ONu 254 .ONOu—N—L 5O EOE—006m Mun—ev- JG 0245—“. OOV OOO OOO OVO ONO OOO OON OON OVN ONN OON OO« __. ___~__ , .. _ ._!1_ .n , L. _ .. . I ._. VON NNN ONN « mm «mm «0“ «NO OVN NON O«N OO« «ON OO« OV« OV« 255 OON OON _b-pL.—- 4- 41 O«N .3363“. be 853925 no.2. .Nm 6.59“. O«N «ON ON« ON« OO« OO« O«« ON ON OO OO Pub.—P-b-—P-nb—-—?-—-b-_bbLL—hb-n—nnLn 256 OON O«N .3333 3 $3325 web. .no 6.5a.“— O«N «ON ON« ON« OO« OO« O«« ON ON OO --—- _--—--—-- i. i ‘ _::_:: ON 1 .‘i-uil'l— OV« 257 .6392 8.5 3 Estevan web. .96 0.59“. OOO OOV OOV OVV ONV OOV OOO OOO OVO ONO OOO OON OON OVN ONN OON OO« OO« OV« ON« —-b-—P-b i (1 -4 can own 353 «ON O«N N«N «ON 258 .5“ch 5.5 .3 35:00.5 NO.+¥ .NO 0.52.... OOV ONV O«V OOO ONO OOO OOO O«O OON ONN OON OON O«N OO« ON« OO« pr-PP — Em. _ _ _ 11—41 11¢ «nu momma .. «ON N«N OVN «ON NON « NN NON 259 Antennae-.25 .3 55.502..." mO_+v. .OO 6.59“. ONO OOO OOO O«N OON ONN OON ONN O«N OO« ON« OO« OO« O«« OO burn—Thl-nn—TL- h—Pbbr_bbbh ‘Ill 1" 14 1 11 —hb:—hP-P—-bb—nui-_b-P-—-pb—.D.-—b-«b_-—hb— «NN 260 .O 5:3an .3 55.5025 OO.+¥ .NO 0.50: OOV ONV OOV ONV O«V OOO ONN OOO ONO O«N OON ONN OON OON O«N OO« ON« OO« OO« O«« OO Pump._-m-L.i.m_wm:—im:p_.wb - :p—:L::_::—Pw:_::—.: —.:i:—p l 1 i .4. i - l. «Room i own nunnon . ohm i and f L on x i 63 .. 261 .omecsncm 3 533625 no.5. do 953“. OVN OON ONN ONN ONN O«N O«N OON OON OO« OO« OO« OO« ON« ON« OO« OO« OO« OO« OV« OV« Pbbn—nhhh—ppun—p-pn—npbP—hb-P nP- - PWJ—pbe—P-np—-p-—-h-—-:-—P P —Pb-P—thb—-hhn—bb-b—-p-—-- :11 (1111 iiiiii 31 141 (1lqtiiiti41‘ 3 (ti 111111111 1.11 1111 144(11 11¢ 11 1144 ((1 (11‘41 t I r I I OON 262 .omoacom 3 52303 met. do 6.59“. OVN OON OON ONN ONNO O«N OON OON OO« OO« OO« OO« ON« ON« OO« OO« OO« OO« OV« OV« —...-—.wwm_.wb._.m..“«"31mmpfl._m-.._q__ _mm1p_-pbh__: m_-b-b_~b-_wumw—.-p_.p.__"-pb_-.bp_.-.._-... «ON I O«N 263 .2..an .3 83.5025 wary. .OO— 6.59... OVN ONN ONN ONN ONN O«N O«N OON OON OO« OO« OO« OO« ON« ON« OO« OO« OO« OO« OV« OV« _-pp._-.-m—bbmp_...m_mmm._m..-_.b-m_...m_...._.. nun—m-b_--m_ : nau« m.-m—.m-._.m.._--._.b-._...b_-.b 264 ONO OON OVN OON OOO ONO OOO OVO OOO OOV ONV OOO OVO OOO OON ONN OO« OV« —-:P—r—y- ph—bb —-n-—-bb— .55on .3 £3325 wary. .3. 6.59“. ‘ d ‘ 1 1“. nth I d ‘ “ '1 ‘1 NOV«NV OON NNN OVN «ON «ON «ON 265 365329 .3 c.8325 me?! No. 6.59“. ONO OON OVN OON OOO ONO OOO OVO OOO OOV ONV OOO OVO OOO OON ONN OO« OV« ONO r O«x i «ON 266 63¢ o.a._oum< .6 Eaguuoam mun-+2 .MO— 0.59”. OON OON OVN ONN ONN —CPp-P...1_...1._..1.._. 3m 68 63 as as on. on“ a: np—nnup—b-pn—nbbn—nn—n_-1-in-—-L-p—plnh- 111 1 111 1 4 1 1J1q1 1 O«N «1—d 267 OOO OOO —-L-—b It‘{‘ 1 1} .OOOuuNO—Oz .—O EOE—OOO” WD-+¥ .VO— OLOQ—n— ONO OO-V _...- 1.1.1 OVV OOV OON ONN OON OVN OON ._::_ OVO 1“ ‘1‘1 - A 11 111‘: q 4 NOV NON « ON son 268 dag—Chm: so 35.5025 wary. .mO_ 0.59..— ONO OON OVN OON OOO ONO OOO OVO OOO OOV ONV OOO OVO OOO OON ONN OO« OV« _::_.:-rip—PPPp—beT—h:P—E-P—.P:_-.:_: ONO VON NNN l «NN «ON O«N 269 OOO ONO .53.: .3 53.5625 OO_+¥ .OO. 0.59..— OOV OVV OOV OOV OON ONO n34UNN OON OVN OON OO« ON« «ON O«N «ON 270 .omo_>E< 3 Escuuoam web. No. 0.59“. ONO OON OON O«N ONO ONO OOO OOO O«O ONV ONV OOO OOO O«N ONN OON OO« NON OOO « NO L «ON «ON 27] 3.32.2.3? 3 53303 we?! 60. 0.59“. ONO OON OON O«N ONO ONO OOO OOO O«O ONV ONV OOO OON O«O ONN ONN OO« OOV ONO «ON «ON 272 .E:@ anon 3:66.. 3 63303 mofx .oo. 6.59“. ONO OON OVN OON OOO ONO OOO OVO OOO OOV ONV OON OVN OON OON ONN OO« OV« 0 m0 0'” .l OOO . ONN «ON «ON 273 .O—OucN— —( 5O 3...: 38m more. .2. 2%: OON OON OVN ONN ONN O«N OON OO« OO« ON« OO« OO« OV« ON« ON« O«« OO« OO phbh-ppbh—pbbP —-PP—-h-—Pb- 1 1 .- 1 NO« duWId «I. 11 ON« 274 #8133 £5.22. 3 5:56ch more. ._ Z 2%: O«N ONO OOO OOO OOO —-L-—- I OVO O OVO O«O To O«O ONV ONV OOO OOO O«O ONN OON OO« n—bppb—bb-n—hb ~1—1-nb1- .1 1‘ 511‘ 1‘ 1 d E... 275 do; .33.; no 25.5025 NEL- .N— — 0.55: O«O ONO OOO OON OON O«N ONO ONO OOO OOO O«O ONV ONV OOO r... __ PD.- 4, ‘41 . NO OOO .- OOO NNNNON NO OOO «OO ««O ONN «co nun NOO 276 OOO OVO ONO OOO .oo coat; 3 £33on web. .3 . 0.59“. OON OON OVN ONN OON OOO OOO OVO ONO OOO «VO OOO NOO NOO «VOON OOO OOO OVO ONO j ONO OOO OOV OOV OVV ONV OOV OON OOO OVO ONO OOO OON OON OVN ONN OON OO« NOV ONV NOV NNV OOV OVN «NO I 277 3.36:5 15053.4.“ 3 5:383 8.2. .3. 2:2“. coon coo can can own can on» can can own can can r O r man l «me an» non ban «on «no son 1 r nun man 1 can own oov ovv ooe can own cum ovm com on“ om« vn aov ,avm i onv «mm 1 non «am on r can now s cc an n own T can 5v. . 278 68682.22. .3 52525 8.2. .m: 2.6: OON ONN OON OON O«N OOO ONO OOO ONO O«O OOO ONO OOO ONO O«O OOV ONV OOV ONV O«VOOO hunk—pup- .1 mun r no» «on v .50 non men own . «me man uvn nun r sun can r «on 279 .35”.— ..Cafi oce.»50>xo>3¢ 3 533on no.2. .0. _ 959... ONO OON OVN OON OOO ONO OOO OVO OOO OOV cos «co «NO NNO OOV N«O NOV ONV OOO OVN OON OON ONN OO« OV« 280 .3 :5 .3 523on more. .2. 8:2“. zlald I‘D. . 3*. (Ma? OOO ONO OOO ONO O«O OOO ONO OOO ONO O«O OOV ONV OOV ONV O«V OON ONN OON ONO O«O OON «mnuou NNO OVO NOO OOV ONV OVV NOV ONV «OV n—nb-n —P--—_-- d1 1- 281 .3..um:O .3 Easuoaam OO_+¥ .O— _ 0.52“. OON OVN ONO OOO OON OON OVN ONN OON OO« OO« OV« Fb-bp—i- 1111 q 1 NON O«N «ON b—nbnn— 11. 1‘ ‘ NN« OO«NV« 282 ng? 1Hu1whfy ".3 ii 43236.5 3:23.:365333.333; .3 835625 wary. .o: 95...: OVO« OVO OVO OVN OVO OVO OVV OVN OVN OV« _b-PP—- q 4 .- “l- —-n-—-Dbb 151:. , _ N«N av .. one man so NOV OOV 283 OVO« .313 .o 5:588 mac. .8. 8%: OVO OVO OVN OVO OVO OVV OVN ONN OOV OON «NO OOV NOO OVN NON OV« NV« 284 .oun36< 2.3.5 3 $3325 web. ..N. 6.59“. OOO OOV OOV OVV ONV OOV OOO OOO OVO ONN OON OON OON OVN ONN OON OO« OO« OV« VVO l «OV NO« ONN 285 r rfiohfimflg mist-liar .53 :2“. >22. 3 .5388 more. Na. 2%: ONO OOO OVO OOO OON ONN OOO OVO OOO OOO ONO OOV OVV OOV OOO ONO OON OVN OON OO« OON NOO nmv son an «ON «ON O«N wow l0. ll. l2. l3. I4. IS. 16. I7. 18. LIST OF REF ERENCES Richardson, 0.W. Proc. Cambridge Phil. Soc. I902, LL, 286. Schottky, W. Physik Z. I9I4, _L5_, 872. von Loue, l1, Jr. Radakt. Elektr. I9I8, L5, 205. Fowler, R.H.; Nordheim, LW. Proc. Roy. Soc. (London) I928, A112, I73. Richardson, O.W. 'The Emission of Electricity From Hot Bodies', Longmans, Green and Co., London, 2nd Ed., I934. Reimann, A.L. 'Thermionic Emission”, John Wiley & Sons, New York, I934. Page, F11; Goode, 6.C. 'Negative Ions and the rlagnetron', Wiley Interscience, New York, I969. I'Iadinos, A. Surf. Sci. I982, _LI_5_, 469. Fomenko, V.S.; Samsonov, 6.V. 'Handbook of Thermionic Properties', Plenum Press, New York, I966. Taylor, J.8.; Langmuir, I. Phys. Rev. I933, 45, 423. E" Copley, ri.J.; Phipps, T.E. Phys. Rev. I934, 35, 344. Shelton, H. Phys. Rev. I957, 191, I553. 3 Child, C.D. Phys. Rev. I9I I, 12, 498. L Langmuir, I.; Blodgett, K.8. Phys. Rev. I924, 25, 49. Chow, K.I(.; i'laddlx, H.S.; Chorney, P. AppI. Phys. Letts. I967, 19, 256. Weber, R.E.; Cordes, L.F. Rev. Sci. Instr. I966, 31, I I2. Blewett, J.P.; Jones, E.J. Phys. Rev. I936, 59,, 464. Futamoto, I'I.; Nakazawa, rt; Kawabe, U. Surf. Sci. I980, 286 I 9. 20. 2 I. 22. 23. 24. 25. 26. 27. 28. 29. 30. 3I. 32. 33. 34. 35. 36. 287 .LQQ, 470. I'Ioon, P.8.; Oliphant, I'LL. Proc. Roy. Soc.(A) I932, L31, 463. Smith, LP. Phys. Rev. I929, 33, I496. Smith, L.P. Phys. Rev. I930, 15, 38I. Scheer, ".D. J. Res. I970, 14A, 37. Hartenson, BIL; Wilhelmsson, 5.0.; Int. J. of Mass Spectrom. Ion Proc. I985, 51. I79. I‘IeIton, C.E. 'Principles of Mass Spectrometry and Negative lons', Marcel Dekker, New York, I970, p.68. Kelso, J.F.; Pantano, C.G.; Garofalini, S.H. Surf. Sci. I983, 133, L543. Zandherg, E.Ya.; lonov, N.I. Sov. Phys. Uspehki I959, 61, 255. Sroubek, 2. Int. J. Mass Spectrom. Ion Phys. I983, 51, 289. Kawano, H.; Page, F.i1.; Int. J. Mass Spectrom. Ion Phys. I983, 5_Q, I. Zandberg, E.Ya.; Rasulev, U. Kh. Russ. Chem. Rev. I982, 51, 8I9. Kawano, H.; Hidaka, Y.; Page, F.i‘I. Int. J. Mass Spectrom. Ion Phys. I983, 59, 35. Kawano, I-I.; Hidaka, Y.; Suga, II; Page, Fri. Int. J. Mass Spectrom. lon Phys. I983,_50, 77. Kingdon, K.Ii.; Langmuir, I. Phys. Rev. I923, 2J_, 380. Kingdon, K.H.; Langmuir, I. Proc. Roy. Soc(A) I925, .191, 6i. Reference 24, pp 64-66. Fujii, T. Int. J. Mass Spectrom. Ion Proc. I984, 51, 63. Zandberg, E.Ya.; Rasulev, U.Kh. Izv. Akad. Nauk SSSR Ser. 37. 38. 39. 4I. 42. 43. 45. 47. 49. 50. 288 Fiz. I966,AQ, I56I. Zandberg, E.Ya.; Rasulev, U.Kh. Dokl. Akad. Nauk SSSR I968, 116. 327. Zandherg, E.Ya.; Rasulev, U.Kh.; Sharapudinov, I1.R. Teoretich. I Eksper. Khim. I970, 6, 328. Zandberg, E.Ya.; Rasulev, U.Kh. Teoretlch i Eksper. Khim. I972, Q, II4. Zandberg, E.Ya.; Rasulev, U.Kh. Teoretich. I Eksper. Khlm. I972, 8, 658. Failes, R.L.; Joyce, J.T.; Watlon, E.C. J. Phys. 8: Atom. I‘IoIec. Phys. I979, 12, 2379. Rasulev, U.Kh.; Khalikov, Sh.l'l. Zh. Org. Khim. I976, 12, 479. ‘ Rasulev, U.Kh. Zh. Org. Khim. I973, 9, I573. Zandberg, E.Ya.; Rasulev, U.Kh.; Khalikov, Sh.l'1. Zh. Org. Khim. I976, 12, II43. Zandberg, E.Ya.; Rasulev, U.Kh.; Takhistov, u.v.; Sharapudinov, I1.R. Teoretich. i Eksper. Khim. I970, 6, 776. Zandberg, E.Ya.; Rasulev, U.Kh.; Sharapudinov, I1.R. Izv. Akad. Nauk SSSR Ser. FIz. I97I, 35,, 334. Zandberg, E.Ya.; Rasulev, U.Kh. Zh. Anal. Khim. I972, 21, 2459. Zandherg, E.Ya.; Rasulev, U.Kh.; Shustrov, B.N. Dokl. Akad. Nauk SSSR I967, 112, 885. Rasulev, U.Kh.; Zandberg, E.Ya.; Bogolyubov, 6.". Zh. Obshchei Khil'n. I974, 55, '2I98. Fomin, O.K.; lehomirov, I1.V. Russ. J. Phys. Chem. I964, SI. 52. S3. 54. 55. 56. S7. 58. 59. 60. 6|. 62. 63. 64. 65. 66. 289 3.8, 444. Fomin, O.I(.; lehomirov, I1.V. Russ. J. Phys. Chem. I964, 38., 383. . Persky, A.; Greene, E.I-'.; Kuppermann, A. J. Chem. Phys. I968, 59, 2347. Scheer, ".D.; Fine, J. Phys. Rev. Lett. I966, 11, 283. Lyons, L.E.; Palmer, L.D. Int. J. Mass Spectrom. Ion Phys. I975, _L6, 431. Horan-Lopez, J.L.; Ten Bosch, A. Surf. Sci. I977, 68, 377. SomorjaI, 6A Science I985, 221, 902. Bartmess, J.E. J. Phys. Chem. Ref. (in press) HcLafferty, F.W. 'Interpretation of "ass Spectra', 3rd Ed., University Science Books: I'IIII Valley, CA, I980; Chapter 3. I'Iunson, H.S.8. Anal. Chem. I977, 49, 772A. I'Iunson, "5.8.; Field, F.H. J. Am. Chem. Soc. I966, 88, 262i. Allison, J.; Ridge, D.P. J. Am. Chem. Soc. I979, ,l_9_1_, 4998. Hodges, R.V.; Beauchamp, J.L. Anal. Chem. I976, 48, 825. ' Burnier, R.D.; Byrd, G.D.; Freiser, 8.5. Anal. Chem. I980, 32, I64I. Weddle, 6.H.; Allison, J.; Ridge, D.P. J. Am. Chem. Soc. I977, 29, I05. Cody, R.8.; Burnier, R.D.; Reents, W.D. Jr.; Carlin, T.J.; I'IcCrery, D.A.; Lengel, R.K.; Freiser, 8.S. Int. J. Mass Spectrom. Ion Phys. I980, 33, 37. I‘lusselman, 8.0.; Davenport, r1.J.; Holland, J.F.; Dolnikowski, 6. 30th Annual Conference on Mass Spectrometry 8: Allied “$43.54;“. 1| « -_ 67. 68. 69. 70. 7I. 72. 73. 74. 75. 76. 77. 78. 79. 80. 290 Topics, Honolulu, HI, I982; p 7I5. Freiser, 8.5.; Staley, R.H.; Beauchamp, J.L. Chem. Phys. Lett. I976, 39,, 49. Spears, K.G.; Ferguson, HE. J. Chem. Phys. I973, 39, 4I74. Ackermann, 8.; Allison, J.; rlusselman, 8.; Pinkston, D.; Bombick, D.; Dolnikowski, 6.; Tsarbopoulos, A.; Watson, J.T. 3Ist Annual Conference on Mass Spectrometry 8: Allied Topics, Boston, MA, I983; p 600. Olah, K; Szdke, A.; Vajta, 2.5. J. Chromatogr. Sci. I979, 11, 497. w . Patterson, P.L.; Batten, R.A.; Ontiveros, G. J. Chromatogr. Sci. I982, 29,, 97. Armentrout, P.B.; Halle, L.F.; Beauchamp, J.L. J. Chem. Phys. I982, 16, 2449. Barber, I'I.; Bordoll, R.S.; Elliot, E.J.; Sedgwick, R.D.; Tyler, A.N. Anal. Chem. I982, 54, 645A. Popov, A.I. Pure Appl. Chem. I979, 1L, IOI. Huang, S.K.; Allison, J. Organometallics I983, 2, 883. Lee, Y.C.; Popov, A.I.; Allison, J. Int. J. Mass Spectrom. Ion Phys. I983, 51., 267. Beckey, H.D. Int. J. Mass Spectrom. Ion Phys. I969, 2, 500. Barber, I1; Bordoll, R.S.; Sedgwick, R.D.; Tyler, A.N. J. Chem. Soc., Chem. Commun. i98i, 325. Benninghoven, A.; Sichtermann, W. Anal. Chem. I978, 50, I I80. Posthumus, HA; Kistemaker, P.G.; I'Ieuzelaar, H.L.C.; Ten Noever de Brauw, I1.C. Anal. Chem. I978, 50, 985. 8I. 82. 83. 84. 85. 86. 87. 89. 90. 9I. 92. 93. 94. 95. 96. 291 I'lacfarlane, R.D. Acc. Chem. Res. I982, 15, 268. Beuhler, R.J.; F lanigan, E.; Greene, L.J.; Friedman, L. J. Am. Chem. Soc. I974, 96, 3990. Schulten, H.-R.; Lattimer, R.P. Mass Spectrom. Rev. I984, 3, 23I. Grassie, N. Pure Appl. Chem. I982, 53, 337. ROIIgen, F.W.; Schulten, H.-R. Z. Naturforsch. I975, 36A, I685. ROIIgen, F.W.; Schulten, H.-R. Org. Mass Spectrom. I975, 10, 660. Giessmann, U.; RGIIgen, F.W. Org. Mass Spectrom. I976, 11. I094. Borchers, F.; Giessmann, U.; Rbllgen, F.W. Org. Mass Spectrom. I977, 12, 539. ROIlgen, F.W.; Giessmann, U.; Borchers, F.; Levsen, K. Org. Mass Spectrom. I978, 13, 459. Van der Peyl, G.J.0.; Isa, K.; Haverkamp, J.; Kistemaker, P.G. Org. Mass Spectrom. I98I, 16, 4I6. StoII, R.; ROIlgen, F.W. Org. I‘Iass Spectrom. I98I, 16, 72. Cotter, R.J.; Yergey, A.L. Anal. Chem. I98I, 63, I306. Schmelzeisen-Redeker, G.; Giessmann, U.; ROIlgen, F.W. Org. I'Iass Spectrom. I985, 29, 305. Day, R.J.; Unger, S.E.; Cooks, R.G. Anal. Chem. I980, 52, 557A. Woodin, R.L.; Beauchamp, J.L. Chem. Phys. I979, 31, I. RdIIgen, F.W.; Biessmann, U.; StoIl, R. Nucl. Instrum. Methods I982, 198., 93. 97. 98. 99. I00. IOI. I 02. I 03. I 04. I 05. I 06. I 07. I 08. I 09. HO. III. II2. 292 StoII, R.; Rbllgen, F.W. Z. Naturforsch. I982, 31A, 9. F agerson, I.S. J. Agric. Food Chem. I969, 11, 747. Higman, E.B.; Schmeltz, I.; Schlotzhauer, W.S. J. Agric. Food Chem. I978, 18, 636. Ohnishi, A.; Kato, K.; Takagi, E. Polym. J. I975, 1, 43I. Irwin, W.J. “Analytical Pyrolysis A Comprehensive Guide; I‘Iarcel Dekker: New York, I982; Chapter 7.3,:pp 339-352. Schulten, H.-R.; Gortz, W. Anal. Chem. I978, 59,, 428. Shafizadeh, F. Adv. Carbohyd. Chem. I968, 23, 4I9. Muller, ”.D.; Seibl, J.; Simon, W. Anal. ChIm, Acta I978, 169, 263. Gower, J.L.; Beaugrand, C.; saIIot, C. Biomed. Mass I98I, 6, 36. Barber, I'I; Bordoll, R.S.; Sedgwick, R.D.; Tyler, A.N.; Greene, I982, Spectrom. 8.N.; Parr, Y.C.; Gower, J.L. Biomed. Mass Spectrom. 2, II. Heerma, W.; Kamerling, J.P.; SIotboom, AJ.; van Scharrenburg, G.J.; Greene, 8.N.; Lewis, I.R.S. Biomed. I'Iass I983, 10, I3. Westmore, J.B.; Ens, W.; Standing, K.G. Biomed. Mass Spectrom. I982, 9, II9. Ratcllff, I'I.A., Jr.; Medley, E.E.; Simmonds, P.G. J. Org. Chem. I974, 39, I48I. Franklin, J.L. Ind. Eng. Chem. I949, 51, l070. Svec, H.J.; Junk, G.A. J. Am. Chem. Soc. I964, 66, 2278. Lin, v.v.; Low, C.-E.; Smith, L.L. J. Steroid biochem. I98I, 14, 563. Spectrom. II3. II4. IIS. II6. II7. II9. I 20. I2I. I 22. I 23. I 24. I 25. I 26. I 27. 293 Smith. L.L. 'Cholesterol Autoxidation', Plenum, New York, I98I. KruII, U.J.; Thompson, n; Arya, A. Talanta I984, 31, 489. Lifshitz, C.; Bergmann, E.D.; Pullman, 8. Tetrahedron Lett. I967, 46, 4587. Cotter, R.J. Anal. Chem. I980, 52, I589A Udseth, H.R.; Friedman, L. Anal. Chem. I98I .53, 29. Chait, 8.T.; Field, F.H. 3Ist Annual Conference on Mass Spectrometry 8: Allied Topics, Boston, Hay 8-I3, I983, p. 6I9. Gardella, J.A., Jr.; Hercules, 0.". Anal. Chem. I980, 52, 226. Dole, n; Cox, H.L., Jr.; Glemic, J. Adv. Chem. Ser. I973, 125, 73. Trash, C.R. J. Chromatogr. Sci. I973, 11, I96. Baiulescu, G.E.; IIIe, ILA. 'Stationary Phases In Gas Chromatography:, Pergamon Press: Oxford, England, I975; pp. I23-I38. Billmeyer, F.W., Jr. 'Textbook of Polymer Science', 2nd ed.; Wiley: New York, I97I ; p.482. Fox, H.W.; Taylor, P.W.; Zisman, W.A. Ind. Eng. Chem. I947, 39, MM. Grassie, N.; flacfarlane, I.G. Eur. Poly. J. I978, 13, 875. Grassie, N.; I‘Iacfarlane, I.G.; Francey, K.F. Eur. Poly. J. I979, 15, 4I5. Grassie, N. 'Encyclopedla of Polymer Science and Technology', Wiley: New York, I966; Vol. 4, p. 647. I 28. I 29. I30. I3I. I32. I33. I34. I 35. I36. I37. I38. I39. ' I40. I4I. I42. I 43. 294 I'Iadorsky, S.L. 'Thermal Degradation of Organic Polymers', Wiley: New York, I964, Chapters 3-4. Conley, R.T. 'Thermal Stability of Polymers', Marcel Dekker: New York, I970, Vol. I, Chapter I0. I976, 111, 2093. Doerr, I'I.; Luderwald, I.; Schulten, _H.-R. Fresenlus 2. Anal. Chem. I984, 318, 339. - Luderwald, I.; I'Ierz, F.; Rothe, I1. Angew. I'Iakromol. Chem. I978, 62, I93. Adams, R.E. Anal. Chem. I983, 55, 4I4. Bahr, U.; Luderwald, I.; I'Iuller, R.; Schulten, H.-R. Angew. I984, 129, I63.‘ Bletsos, I.U.; Hercules, D.I‘I.; Greifendorf, D.; Benninghoven, A Anal. Chem. I985, 51, 2384. Ettre, L.S. J. Chromatogr. Sci. I978, 16, 396. Farwell, S.D.; Gage, D.R.; Kagel, R.A. J. Chromatogr. Sci. I98I, 19, 358. Aue, WA J. Chromatogr. Sci. I975, 13, 329. Bonati, IL; Castelli, D.; Latinl, R.; Garattini, S. J. Chromatogr. I979, 164, I09. Cochrane, W.P.; Greenhalgh, R. Chromatographla I976, 9, 255. I'leola, J.I‘i. Chromatogr. Newsletter I977, 5, I. Vandemark, F.L.; Adams, R.F. Chromatogr. Newsletter I977, 5, 4. Ames, I'I.I'l.; Pavls, G. J. Chromatogr. I979, 114, 245. Luderwald, I.; Urrutia, H. Hakromol. I‘Iakromol. Chem. I44. I 45. I 46. I 47. I 48. I 49. I50. I 5 I . I 52. I53. I54. 295 Brazhnikov, V.V.; Gur‘ev, Ii.v.; Sakodynsky, K.I. Chromatographic Rev. I970, 12, I. Karmen, A.; Guiffrida, L.E. Nature I964, 291, I204. Kolb, 8.; Bischoff, J. J. Chromatogr. Sci. I974, 12, 625. Burgett, C.A.; Smith, D.H.; Bente, H.8. J. Chromatogr. I977, 134,, 57. Patterson, P.L. J. Chromatogr. I978, 161, 38I. 8amford, C.H.; Tipper, C.F.H., Eds. 'Comprehensive Chemical Kinetics'; Elsevier: New York, I972; Vol. 5. Lin, S.-T.; Klabunde, K.J. Langmuir I985, 1, 600. Leclerk, J.C.; Lorquet, J.C. J. Phys. Chem. I967, LI, 787. Walker, B.W.; Stair, P.C. Surf. Sci. I98I, 193, 3I5. Fujii, T.; Arimoto, H. Anal. Chem. I985, 51, 2625. Folmer, O.F.; Yang, K.; Perkins, G. Anal. Chem. I963, 35, 454.