<%BANNER%>

Parameters Affecting Performance of Planar High-Field Asymmetric Waveform Ion Mobility Spectrometry (FAIMS)

Permanent Link: http://ufdc.ufl.edu/UFE0042321/00001

Material Information

Title: Parameters Affecting Performance of Planar High-Field Asymmetric Waveform Ion Mobility Spectrometry (FAIMS)
Physical Description: 1 online resource (183 p.)
Language: english
Creator: Rorrer, Leonard
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: faims, ion, mass, mobility, planar, solvent, spectrometry
Chemistry -- Dissertations, Academic -- UF
Genre: Chemistry thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: PARAMETERS AFFECTING PERFORMANCE OF PLANAR HIGH-FIELD ASYMMETRIC WAVEFORM ION MOBILITY SPECTROMETRY (FAIMS) By Leonard Calvin Rorrer, III December 2010 Chair: Richard A. Yost Major: Chemistry High-field asymmetric waveform ion mobility spectrometry (FAIMS) is an atmospheric-pressure gas-phase ion separation technique. FAIMS is similar to conventional ion mobility spectrometry (IMS) in that both are based on the motion of ions in a gas induced by an electric field to achieve separation. Conventional ion mobility uses low electric fields to propel ions through a drift gas and separate them based on their mobilities. However, at low electric fields, the ion s mobility is independent of applied field, which can lead to ions with similar mobilities not being resolved. At high electric fields, ion mobilities become dependent on applied field. This dependence of ion mobility on applied field is the basis of separation in FAIMS. FAIMS has been developed for both FAIMS-mass spectrometry and as a stand-alone device. Despite an increasing number of applications, many of the fundamental aspects of FAIMS are not well understood. Two classes of geometries exist for FAIMS cells: planar and curved. Planar geometries offer high resolving powers at a cost of transmission when compared to curved geometries. Our research has focused on two fundamental areas that can affect the performance of a planar geometry FAIMS cell. The first study investigated the modes of injection into the FAIMS cell. Investigations in this area led to a better understanding of the correlation between ion residence time in the cell and the tradeoffs between transmission and resolving power. A method for controlling lateral diffusional losses by means of a pneumatic focusing gas was also developed. The second study looked at the role of solvent vapor on the performance of the planar FAIMS cell. The results of this work demonstrated that controlled amounts of solvent present in the curtain/drift gas used in FAIMS yielded dramatic shifts in compensation voltages while maintaining relatively narrow peak widths yielding resolving powers up to ~300. Dramatic increases in sensitivity were also observed with solvent present, giving up to ~25x times more signal. This behavior led to large increases in resolving power for the planar FAIMS cell and increases in resolution and peak capacity for separating mixtures of analytes including isomers and explosives.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Leonard Rorrer.
Thesis: Thesis (Ph.D.)--University of Florida, 2010.
Local: Adviser: Yost, Richard A.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0042321:00001

Permanent Link: http://ufdc.ufl.edu/UFE0042321/00001

Material Information

Title: Parameters Affecting Performance of Planar High-Field Asymmetric Waveform Ion Mobility Spectrometry (FAIMS)
Physical Description: 1 online resource (183 p.)
Language: english
Creator: Rorrer, Leonard
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: faims, ion, mass, mobility, planar, solvent, spectrometry
Chemistry -- Dissertations, Academic -- UF
Genre: Chemistry thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: PARAMETERS AFFECTING PERFORMANCE OF PLANAR HIGH-FIELD ASYMMETRIC WAVEFORM ION MOBILITY SPECTROMETRY (FAIMS) By Leonard Calvin Rorrer, III December 2010 Chair: Richard A. Yost Major: Chemistry High-field asymmetric waveform ion mobility spectrometry (FAIMS) is an atmospheric-pressure gas-phase ion separation technique. FAIMS is similar to conventional ion mobility spectrometry (IMS) in that both are based on the motion of ions in a gas induced by an electric field to achieve separation. Conventional ion mobility uses low electric fields to propel ions through a drift gas and separate them based on their mobilities. However, at low electric fields, the ion s mobility is independent of applied field, which can lead to ions with similar mobilities not being resolved. At high electric fields, ion mobilities become dependent on applied field. This dependence of ion mobility on applied field is the basis of separation in FAIMS. FAIMS has been developed for both FAIMS-mass spectrometry and as a stand-alone device. Despite an increasing number of applications, many of the fundamental aspects of FAIMS are not well understood. Two classes of geometries exist for FAIMS cells: planar and curved. Planar geometries offer high resolving powers at a cost of transmission when compared to curved geometries. Our research has focused on two fundamental areas that can affect the performance of a planar geometry FAIMS cell. The first study investigated the modes of injection into the FAIMS cell. Investigations in this area led to a better understanding of the correlation between ion residence time in the cell and the tradeoffs between transmission and resolving power. A method for controlling lateral diffusional losses by means of a pneumatic focusing gas was also developed. The second study looked at the role of solvent vapor on the performance of the planar FAIMS cell. The results of this work demonstrated that controlled amounts of solvent present in the curtain/drift gas used in FAIMS yielded dramatic shifts in compensation voltages while maintaining relatively narrow peak widths yielding resolving powers up to ~300. Dramatic increases in sensitivity were also observed with solvent present, giving up to ~25x times more signal. This behavior led to large increases in resolving power for the planar FAIMS cell and increases in resolution and peak capacity for separating mixtures of analytes including isomers and explosives.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Leonard Rorrer.
Thesis: Thesis (Ph.D.)--University of Florida, 2010.
Local: Adviser: Yost, Richard A.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0042321:00001


This item has the following downloads:


Full Text

PAGE 1

1 PARAMETERS AFFECTING PERFORMANCE OF PLANAR HIGH FIELD ASYMMETRIC WAVEFORM ION MOBILITY SPECTROMETRY (FAIMS) By L EONARD CALVIN RORRER, III A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2010

PAGE 2

2 2010 Leonard C alvin Rorrer, III

PAGE 3

3 To Mom I miss you

PAGE 4

4 ACKNOWLEDGMENTS I would like to extend my sincere thanks to Dr. Richard A. Yost for his support and guidance throughout this work. I would also like to thank my committee members, Dr. Alexander Angerhofer, Dr. Renwei Mei, Dr. Nicholas Polfer, and Dr. Davi d Powell. I would like to thank Dr. Michael Belford of Thermo Fisher Scientific for providing me a FAIMS waveform generator as well as providing many helpful discussions. I would like to acknowledge the National Science Foundation CBET division for provi ding funding for the research presented in this dissertation. I would also like to extend my gratitude to the members of the Chemistry D epartment machine shop (Joe, Brian, and Todd) and the members of the Chemistry D epartment electronics shop (Steve and L arry). I would like to acknowledge my fellow group members for their friendship and valuable critical feedback. I would also like to thank my friends and family for giving me lots of support and encouragement throughout my studies. I would like to thank my mother in law, Lea Bates, for driving to our house once a week for three years so that I could finish my doctoral work. Lastly, and most importantly, I would like to thank the three most important people in my life, Becky, Lyla, and Jacob, for always b eing there for me.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ ............... 4 LIST OF TABLES ................................ ................................ ................................ ........................... 7 LIST OF FIGURES ................................ ................................ ................................ ......................... 8 LIST OF ABBREVIATIONS ................................ ................................ ................................ ........ 13 ABSTRACT ................................ ................................ ................................ ................................ ... 16 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .................. 18 Principles of FAIMS ................................ ................................ ................................ ............... 19 Basics of Operation ................................ ................................ ................................ ......... 19 History ................................ ................................ ................................ ............................. 22 Instrumentation ................................ ................................ ................................ ....................... 23 Ionization ................................ ................................ ................................ ......................... 23 Electrospray ionization ................................ ................................ ............................. 23 Atmospheric pressure chemical ionization ................................ ............................... 25 Mass Spectrometry ................................ ................................ ................................ .......... 26 TSQ 7000 ................................ ................................ ................................ ................. 26 Fundamentals of quadrupo le mass spectrometry ................................ ..................... 27 High Field Asymmetric Waveform Ion Mobility Spectrometry ................................ ..... 29 Scope of the Dissertation ................................ ................................ ................................ ........ 31 2 MODES OF INJECTION INTO PLANAR FAIMS ................................ .............................. 45 Introduction ................................ ................................ ................................ ............................. 45 Total Ion Mode ................................ ................................ ................................ ....................... 46 Methods ................................ ................................ ................................ ........................... 46 DC Voltages Only ................................ ................................ ................................ ........... 47 RF and DC Voltages ................................ ................................ ................................ ........ 48 RF only ................................ ................................ ................................ ..................... 48 RF voltage with DC voltage ................................ ................................ ..................... 49 Orthogonal and Parallel Injection ................................ ................................ ........................... 50 Curtain Plate Effects: Part I ................................ ................................ ............................ 51 Orthogonal injection ................................ ................................ ................................ 51 Parallel injection ................................ ................................ ................................ ....... 52 Residence Time ................................ ................................ ................................ ............... 53 Methods ................................ ................................ ................................ .................... 54 Results ................................ ................................ ................................ ...................... 56 Pneumatic Focusing ................................ ................................ ................................ ......... 59

PAGE 6

6 Cutain Plate Effects: Part II ................................ ................................ ............................ 61 Su mmary ................................ ................................ ................................ ................................ 62 3 SOLVENT VAPOR EFFECTS IN PLANAR FAIMS : P ART I ................................ ........... 84 Introduction ................................ ................................ ................................ ............................. 84 Results and Discussi on ................................ ................................ ................................ ........... 86 Materials and Procedures ................................ ................................ ................................ 86 Concentration Effects ................................ ................................ ................................ ...... 88 Phthalic acid isomers ................................ ................................ ................................ 88 Explosives ................................ ................................ ................................ ................ 91 Increased resolution with solvent vapor ................................ ................................ ... 92 Summary ................................ ................................ ................................ ................................ 96 4 SOLVENT VAPOR EFFECTS IN PLANAR FAIMS: P ART II ................................ ....... 111 Introduction ................................ ................................ ................................ ........................... 111 Solvent Trends ................................ ................................ ................................ ...................... 111 Solvents ................................ ................................ ................................ ......................... 112 TNT M Ion ................................ ................................ ................................ .................... 113 Solvent concentratio n ................................ ................................ ............................. 113 Field effects ................................ ................................ ................................ ............ 115 TNT [M H] Ion Versus M Ion ................................ ................................ ..................... 118 Phthalic Acid Isomers ................................ ................................ ................................ .... 121 Solvent Tr ends Summary ................................ ................................ .............................. 124 Temperature Effects ................................ ................................ ................................ .............. 125 Methods ................................ ................................ ................................ ......................... 126 Temperature Ramp ................................ ................................ ................................ ........ 127 Field Effects ................................ ................................ ................................ ................... 129 Temperature Summary ................................ ................................ ................................ .. 132 Conclusions ................................ ................................ ................................ ........................... 133 5 CONCLUSIONS AND FUTURE WORK ................................ ................................ ........... 165 Conclusions ................................ ................................ ................................ ........................... 165 Modes of Injection Summary ................................ ................................ ........................ 165 Solvent Vapor Effects Summary ................................ ................................ ................... 168 Wrap Up ................................ ................................ ................................ ........................ 173 Future Work ................................ ................................ ................................ .......................... 174 Epilogue ................................ ................................ ................................ ................................ 177 A PPENDIX: FIELD STRENGTH VERSUS TEMPERATURE ................................ ............... 178 LIST OF REFERENCES ................................ ................................ ................................ ............. 179 BIOGRAPHICAL SKETCH ................................ ................................ ................................ ....... 183

PAGE 7

7 LIST OF TABLES Table page 2 1 Summary of results of orthogonal injection residence time measurements. ..................... 77 2 2 Summary of results of parallel injection residence time measurements ............................ 77 2 3 Summary of results of studies with and without curtain plate. ................................ .......... 83 3 1 Increased resolution of [M H] ions of phthalic acid with solvent vapor. ....................... 106 3 2 Increased resolution of M and [M H] ions of four explosives with solvent vapor. ....... 108 4 1 Physical properties of solvents studied ................................ ................................ ............ 136 4 2 Maximum Solvent Concentrations ................................ ................................ .................. 136 4 3 Calculated and K h /K terms for the M ion of TNT. ................................ ................. 142 A 1 Field strengths at set dispersion voltages with differing temperature. ............................. 178

PAGE 8

8 LIST OF FIGURES Figure page 1 1 Hypothetical dependencies of ion mobility of three different ions on electric field. ........ 33 1 2 Ion motion between two plates during the application of an electric potential ................. 34 1 3 ESI process in the positive mode ................................ ................................ ....................... 35 1 4 Picture of ESI assembly used in studies involving ions generated by ESI. ....................... 36 1 5 APCI process in the positive mode ................................ ................................ .................... 37 1 6 Picture of APCI assembly used in studies i nvolving ions generated by APCI. ................. 38 1 7 Schematic showing layout of TSQ 7000 ................................ ................................ ........... 39 1 8 Schematic dia gram of quadrupole mass filter ................................ ................................ .... 40 1 9 Mathieu stability diagram for a quadrupole mass filter ................................ ..................... 40 1 10 Schematic and picture of custom built planar FAIM S cell ................................ ................ 41 1 11 Shematic of parallel and orthogonal injection ................................ ................................ ... 42 1 12 Picture of heated capillary adapter ................................ ................................ ..................... 43 1 13 Sum of sines waveforms used in FAIMS ................................ ................................ .......... 44 2 1 Structures of three positional isomers of phthalic acid ................................ ...................... 65 2 2 Comparison of TIM intensity with respect to the DC voltage applied to the FAIMS plates ................................ ................................ ................................ ................................ .. 65 2 3 Circuit schematic s used for applying RF and DC voltages on FAIMS cell plates ............ 66 2 4 Comparison of TIM intensity versus the peak to peak RF voltage applied to the FAIMS cell plates at 794 kHz ................................ ................................ ............................ 66 2 5 Comparison of TIM intensity versus the peak to peak intensity of the RF voltage applied to the FAIMS cell plates at 1.62 MHz ................................ ................................ .. 67 2 6 Comparison of TIM intensity with respect to the DC voltage applied to the FAIMS cell plate. ................................ ................................ ................................ ............................ 68 2 7 Image of curtain plate used in orthogonal injection studies ................................ .............. 69

PAGE 9

9 2 8 CV peak intensity with respect to curtain gas flow rate with three different curtain gaps ................................ ................................ ................................ ................................ .... 69 2 9 Image of dual curtain plate assembly used for parallel injection. ................................ ...... 70 2 10 CV peak intensities with respect to curtain gas flow rate at two different parallel injection curtain gaps ................................ ................................ ................................ ......... 70 2 11 Schematic of relay circuit used to gate ions in the FAIMS cell for residence time measurement studies. ................................ ................................ ................................ ......... 71 2 12 Example of residence time curve generated in residence time measurement studies ........ 71 2 13 Example of linear fit s used to measure residence time ................................ ...................... 72 2 14 Measured orthogonal and parallel residence times compared calculated average gas residence time ................................ ................................ ................................ .................... 74 2 15 Comparison of orthogonal an d parallel injection CV scans for the [M H] ion of o phthalic acid at three different residence time ................................ ................................ ... 76 2 16 Normalized FWHM and i ntensity for orthogonal and parallel injection compared to residence time ................................ ................................ ................................ .................... 78 2 17 Schematic showing position and design of porous spacers used in pneumatic focusing studies. ................................ ................................ ................................ ................. 79 2 18 Structures of explosive analytes ................................ ................................ ......................... 80 2 19 N ormalized intensity and FWHM of CV peaks with addition of pneumatic focusing gas ................................ ................................ ................................ ................................ ...... 80 2 20 Comparison of CV scans for the M ion of TNT with and without pneumatic focusing gas ................................ ................................ ................................ ....................... 82 2 21 Comparison of CV scans in orthogonal and parallel injection mode for th e [M H] ion of o phthalic acid with and without a curtain plate in place ................................ ........ 83 3 1 Schematic showing apparatus for generating solvent saturated nitrogen .......................... 98 3 2 Plots of CV FWHM, and normalized intensity with various concentrations of water vapor for the thr ee positional isomers of phthalic acid ................................ ...................... 99 3 3 Zoomed in areas of plots from Figure 3 2 showing values at low water vapor concentrati on ................................ ................................ ................................ .................... 100 3 4 Plots of CV, FWHM, and normalized intensity with various concentrations of methanol vapor for the three positional isome rs of phthalic acid ................................ .... 101

PAGE 10

10 3 5 Zoomed in areas of plots from Figure 3 4 showing values at low methanol vapor concentration ................................ ................................ ................................ .................... 102 3 6 Plots of CV, FWHM, and normalized intensity with various concentrations of water vapor for the four explosive analytes ................................ ................................ ............... 103 3 7 Zoomed in areas of plots from Figure 3 6 showing values at low water vapor concentration ................................ ................................ ................................ .................... 104 3 8 CV scans of mixture of the three position isomers of phthalic acid acquired with different solvent vapor conditions. ................................ ................................ .................. 105 3 9 CV scans of mixture of the TNT and three positional isomer of DNT (2,4, 3,4, and 2,6) acquired with different solvent vapor conditions ................................ .................... 107 3 10 Plot of CV for the two major ions of TNT with respect to various concentrations of water and methanol vapor ................................ ................................ ................................ 109 4 1 Structures of six solvents used in studies. ................................ ................................ ........ 135 4 2 CV and FWHM for the M ion of TNT with respect to increasing concentration of solvent vapor added to the drift gas ................................ ................................ ................. 137 4 3 Correlation of maximum CV shift (at 70 Td ( 3500 DV)) with molecular volume of solvent added to the drift gas. ................................ ................................ .......................... 138 4 4 CV for M ion of TNT versus percentage of carbon dioxide in the nitrogen drift gas. ... 139 4 5 CV and FWHM for the M ion of TNT with respect to increasing field strength in different drift gas compositions ................................ ................................ ....................... 140 4 6 Calculated K h /K values for the M ion of TNT with respect to increasing field strength in different drift gas compositions. ................................ ................................ .... 141 4 7 Calculated resolving powers for the M ion of TNT with respect to increasing field strength in different drift gas compositions. ................................ ................................ .... 143 4 8 Absolute and normalized CV peak intensities for the M ion of TNT with respect to increasing field strength in different drift gas compositions ................................ ........... 144 4 9 CV and FWHM for the [M H] ion of TNT with respect to increasing concentration of solvent vapor added to the carrier gas ................................ ................................ ......... 145 4 10 CV and FWHM for the [M H] ion of TNT with respect to increasing field strength in different drift gas compositions ................................ ................................ ................... 146 4 11 Calculated K h /K values for the [M H] ion of TNT with respect to increasing field strength in different drift gas compositions ................................ ................................ ..... 147

PAGE 11

11 4 12 Normalized CV peak intensity for the [M H] ion of TNT with respect to increasing field strength in different drift gas compositions ................................ ............................. 148 4 13 CV spectrum of the [M H] ion of o phthalic acid acquired at 95 Td in either dried nitrogen or with ~14,000 ppm isopropanol added to the drift gas ................................ ... 149 4 14 CV for the [M H] ion of o phthalic acid with respect to increasing field strength in different drift gas compositions ................................ ................................ ....................... 150 4 15 Calculated K h /K values for the [M H] ion of o phthalic acid with respect to increasing field stren gth in different drift gas compositions ................................ ........... 151 4 16 FWHM and normalized CV peak intensity for the [M H] ion of o phthalic acid with respect to increasing field strength in different drift gas compositions ........................... 152 4 17 CV spectrum of mixture of three positional isomer s of phthalic acid acqu ired at 70 Td in dried nitrogen and with ~14,000 ppm isopropanol added to drift gas. .................. 153 4 18 CV values for t he [M H] ions of o phthalic acid and m phthalic acid with respect to increasing temperature of the gas at the curtain plate with different drift gas compositions ................................ ................................ ................................ .................... 154 4 19 FWHM for the [M H] ions of o phthalic acid and m phthalic acid with respect to increasing temperature of the gas at the curtain plate with different drift gas compositions. ................................ ................................ ................................ ................... 155 4 20 Comparison of absolute CV peak intensity for the [M H] ion of o phthalic acid with differing heated drift gas compositions ................................ ................................ ............ 156 4 21 CV and c alculated Kh/K values for the M ion of TNT with respect to increasing field strength in different drift gas compositions and at 85 C inside the FAIMS cell .... 157 4 22 CV and c alculated Kh/K values for the [M H] ion of TNT with respect to increasing field strength in different drift gas compositions and at 85 C inside the FAIMS cell. ... 158 4 23 Comparison of K h /K values for the M and [M H] ion s of TNT at room temperature and at 85 C inside the FAIMS cell ................................ ................................ .................. 159 4 24 FWHM for the M and [M H] ion s of TNT with respect to incr easing field strength in different drift gas compositions and at 85 C inside the FAIMS cell ........................... 161 4 25 CV spectra for the M ion of TNT showing tailing of CV peaks during elevated temperature experiments ................................ ................................ ................................ .. 162 4 26 Normalized CV peak intensity for the M and [M H] ion s of TNT with respect to increasing field strength in different drift gas compositions at 85 C .............................. 163

PAGE 12

12 4 27 Comparison of K h /K values for the [M H] ion of o phthalic acid at room temperature and at 85 C inside the FAIMS cell with different compositions of drift gas .................. 164

PAGE 13

13 LIST OF ABBREVIATION S ion specific dependency on applied field in FAIMS a DC voltage stability axis of Mathieu stability diagram A c ross sectional area of FAIMS cell in mm 2 amu atomic mass units APCI atmospheric pressure chemical ionization ion specific dependence on applied field in FAIMS cm c entimeters C compensation voltage term used in equations CI chemical ionization CID collision induced dissociation CV c ompensation voltage d FAIMS analytical gap in mm D dispersion voltage term used in equations DC d irect current DNT dinitrotoluene DV d ispersion voltage e charge of ion E e lectrical fie ld strength in V/cm ESI electrospray ionization F f low rate through FAIMS cell in mm 3 /s F s e lectrical field strength in Td FAIMS h igh field asymmetric waveform ion mobility spectrometry FWHM f ull width at half maximum height of a compensation voltage peak Hz Hertz

PAGE 14

14 ID inner diameter IMS i on mobility spectrometry K l ow field mobility constant k b K f t hermodynamic equilibrium constant K h h igh field mobility constant l FAIMS path length in mm L liters m meters or mass (when referring to Mathi eu stability equations) m meta position on benzene ring min minutes mm m illimeters ms m illiseconds m/z mass to charge ratio N gas number density in molecules/cm 3 o ortho position on benzene ring P pressure in Pascals p para position on benzene ring PEEK polyether ether ketone plastic ppm parts per million PW 10% CV peak width at 10% of full height q RF voltage stability axis of Mathieu stability diagram RF r adio frequency R p resolving power R s resolution

PAGE 15

15 r o quadrupole rod spacing in mm s s econds SIM selected ion monitoring t time T temperature in Kelvin Td T ownsends TIM total ion mode TNT trinitrotoluene U DC voltage applied to quadrupole mass filter rods v velocity V v oltage (RF voltage when referencing quadrupole mass spectrometry) w/v weight per unit of volume frequency

PAGE 16

16 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy PARAMETERS AFFECTING PERFORMANCE OF PLANAR HIGH FIELD ASYMMETRIC WAVEFORM ION MOBILITY SPECTROMETRY (FAIMS) By Leonard Calvin Rorrer, III December 2010 Chair : Richard A. Yost Major: Chemistry High field asymmetric waveform ion mobility spectrometry (FAIMS) is an atmospheric pressure gas phase ion separation technique. FAIMS is similar to conventional ion mobility spectrometry (IMS) in that both are based on the motion of ions in a gas induced by an electric field to achieve separation. Conventional ion mobility uses low electric fie lds to propel ions through a drift gas and separate them based on their mobilities. However, at low electric fields, not being resolved. At high electric fie lds, ion mobilities become dependent on applied field. This dependence of ion mobility on applied field is the basis of separation in FAIMS. FAIMS has been developed for both FAIMS m ass s pectrometry and as a stand alone device. Despite an increasing numb er of applications, many of the fundamental aspects of FAIMS are not well understood. Two classes of geometries exist for FAIMS cells: planar and curved Planar geometries offer high resolving powers at a cost of transmission when compared to c urved geom etries. Our research has focused on two fundamental areas that can affect the performance of a planar geometry FAIMS cell. The first study investigated the modes of injection into the FAIMS cell.

PAGE 17

17 Investigations in this area led to a better understanding of the correlation between ion residence time in the cell and the tradeoffs between transmission and resolving power. A method for controlling lateral diffusional losses by means of a pneumatic foc using gas was also developed. The second study looked at the role of solvent vapor on the performance of the planar FAIMS cell. The results of this work demonstrated that controlled amounts of solvent present in the curtain/drift gas used in FAIMS yielde d dramatic shifts in compensation voltages while maintaining relatively narrow peak widths yielding resolving powers up to ~300. Dramatic increases in sensitivity were also observed with solvent present, giving up to ~25x times more signal. This behavior led to large increases in resolving power for the planar FAIMS cell and increases in resolution and peak capacity for separating mixtures of analytes including isomers and explosives

PAGE 18

18 CHAPTER 1 INTRODUCTION Ion mobility spectrometry (IMS) has become an important method for the detection of many compounds. Its high sensitivity an d the ease with which it can be made portable have seen it spread into many applications ranging from detection of narcotics and explosives to analyses of biomolecular compounds [1, 2] IMS uses relatively low constant electric fields, ~200 volts/cm, to propel ions for the source of ionization to the detector. The ions are separated based on their mobility, K in a drift gas which fills the IMS apparatus. Even w ith its high sensitivity, false positives often occurs due to the overlapping mobilities of compounds [1] Within the past two decades, a new ion mobility technique has been developed which is more sensitive than IMS and can reduce the false positives that can plague IMS. This new technique, which is referred to as high field asymmetric waveform ion mobility spectrometry (FAIMS) [3, 4] is similar to conventional IMS in that it operates at atmospheric pressure and use s the motion of ions produced by an applied electric field to achieve separation The major differences between the two techniques are that FAIMS operates at much higher fields, typically gr eater than ~10,000 volts/cm, that the electric field is not constant, alternating between periods of opposite polarities and that ions are moved pneumatically through the FAIMS apparatus Two major geometries of FAIMS cells exist currently, curved geometries ( cylindrical or domed type cells as well as spherical or hemispherical designs [5] ) and planar geometries (sometimes referred to as differential ion mobility or DMS). Curved geometries offer greater transmission due to an electrostatic focusing effect at a cost of low er re s olution and resolvi ng powers [6, 7] Planar geometries offer higher specificit y and resolving power at a cost of lower transmission [8] The purpose of this dissertation is to address the parameters that affect the

PAGE 19

19 performance of planar geometry FAIMS cell and develop a better understanding of how FAIMS works in general. Principles of FAIMS Ions that are subjected to an electric field in a drift gas will move along the applied field lines with a velocity equal to vKE (1-1) where v is the drift velocity of the ion, E is the applied field strength, and K is the coefficient of ion mobility. At low electric fields, ~200 volts/cm, K is independent of applied electric field [9] and ions with different K values will travel at different velocities. Conventional IMS uses these differences to achieve separation. At higher electric fields, greater than ~10,000 volts/cm, K becomes dependent on the applied field in a nonlinear manner [9]. This high-field mobility is better referred to as Kh, a high-field mobility term, which varies depending on the field strength. The dependence of Kh on the applied electric field is the basis of FAIMS. Ions are separated using their difference in mobility at high field (Kh) and mobility at low field. The high-field behavior of Kh is compounddependent, which allows for FAIMS to separate ions which may have similar low-field mobilities. The change in ion mobility at high field reflects ion size, interactions of the ion with drift gas and neutrals within the drift gas, and structural conformation and rigidity [10]. Basics of Operation The principles of operation of FAIMS were first described by Buryakov et al. [9]. The mobility of an ion under the influence of a high-field is expressed as 241hEEKKNN (1-2)

PAGE 20

20 where K h is the high field mobility, K is the low field mobility, E is the applied electric field in volts/cm, N is the gas number density in molecules/cm 3 and and are ion specific dependencies on applied f ield. The and terms are not theoretically grounded, or part of a physical or chemical model. Rather, they are simply coefficients used to fit a two term polynomial (Taylor series) empirically to the data. Figure 1 1 gives examples of the three possi ble changes in ion mobility on going to higher field strengths. In thi s figures, the mobility of type A ions increases with increasing electric field, the mobility of type C ions decreases with increasing field and the mobility of type B ions initially in creases then decreases with increasing field. The separation of ions in FAIMS is based on the change in the ions mobility as the ion experiences differing applied fields. Consider a positive ion of type A, depicted in Figure 1 1, being pneumatically prope lled between two parallel plates, as shown in Figure 1 2. One of the plates is kept at ground potential and the other has the hypothetical asymmetric waveform shown at the middle of the figure applied to it (the actual waveform is discussed later in this chapter) The asymmetric waveform consists of a high voltage component lasting for a short period of time followed by a lower voltage opposite polarity component lasting for a longer period of time. The waveform is such that the integrated voltage time p roduct applied to the plate is zero ( 1 3 ) where V 1 and V 2 are of opposite polarities. Figure 1 2 illustrates the ion path for a portion of the waveform shown as V(t) During the high voltage portion of the waveform, the ion will move away from the upper plate with velocity ( 1 4 )

PAGE 21

21 where v 1 is the velocity of the ion, is the high field mobility at applied electric field V 1 The distance traveled by the ion is described by ( 1 5 ) where t 1 is the time period of the applied high voltage, V 1 During the longer duration, opposite polarity lower voltage portion of the waveform, the ion will move towards the upper plate with velocity ( 1 6 ) where is the high field mobility at applied electric field V 2 The distance traveled by the ion is described by ( 1 7 ) where t 2 is the time period of the applied lower voltage. The distance the ion travels during the t wo different portions of the waveform can be rewritten as ( 1 8 ) and ( 1 9 ) The asymmetric waveform guarantees that the field time products, V 1 t 1 and V 2 t 2 will be equivalent, [Equation (1 3)]. If and are identical, the distances the ion travels during the two different portions of the waveform will be the same, and the ion will experience no net displacement from its starting position. However, under high fields, will differ from and the ion will experience a net displacement as shown by the dashed line in Figure 1 2. For type A ions in general, the distance traveled during the higher voltage portion of the wav eform

PAGE 22

22 will be greater than that traveled during the lower voltage portion, i.e., d 1 > d 2 The opposite is true for type C ions. The type A ion shown in Figure 1 2 will experience a net displacement away from the upper plate and will eventually strike the lo displacement, a negative direct current (DC) voltage can be applied to the upper plated to pull the ion back between the plates. The DC voltage is commonly referred to as a compensation voltage (CV) [3] Since K h and therefore the ratio K h to K is compound dependent, the magnitude of the CV needed to offset the net displacement will be different for different ions [4] Under conditions appropriate t o transmit one ion, other ions will drift toward one of the plates and be lost. The FAIMS cell is essen tially an ion filter which selectively transmits ions with the appropriate ratio of K h to K A mixture of compounds can be analyzed by scanning the CV producing a CV spectrum [9] History The first FAIMS cell was built by Buryakov et al. [9] and consisted of two flat plates similar to that described in F igure 1 2. The instrument had two ionization sources for gas phase ionization of different compounds, a source (tritium) and a surface ionization source (molybdenum wire). The compounds were ionized and then flowed into the FAIMS part of the instrument where they were subsequently separated and then detected using either an ion collector or a mass spectrometer [9] Carnahan and Tarassov [11, 12] replaced the flat plates of the earlier ins trument design with two concentric cylinders. The concentric cylinder configuration proved to be more sensitive than the flat plate design due to a n electrostatic focusing effect on the ions [7, 12] The design patented by Carnahan and Tarassov [11] was commercially introduced by Mine Safety

PAGE 23

23 Appliances Company (MSA) for trace gas analysis and was sold under the name of Field I on Spectrometer (FIS). Guevremont and Purves [7] were the first to attach the concentric cylinder FAIMS cell to a mass spectrometer to analyze gas phase ions with mass spectrometric detection. Guevremont and Purve s [13] also developed the first FAIMS cell compatible with electrospray Ionalytics Corp.(Ottawa, Canada) in 2000. Ionalytics was purchased by Thermo Fisher S cientific (San Jose, CA) and is currently the only provider of a commercially available concentric cylinder design FAIMS. Miller et al. [14] at Draper laboratories developed a micromachined FAI MS cell which was small enough to be field portable. Their instrument is of the flat plate, planar design similar to that used by Buryakov et al. [9] in the first FAIMS device and until recently was sold by SIONEX Corporation (Bedford, MA). Owlstone Nanotech Inc. (Suffern, NY) a lso currently markets a micromachined planar FAIMS cell Instrumentation Summaries of the types of instrumentation used in this dissertation are listed below. All of the studies involved the production of ions by either electrospray ionization or atmospheric pressure chemical ionization. All data were taken using a TSQ 7000 triple quadru pole mass spectrometer (Finnigan, San Jose, CA) as the ion detector Ionization Electrospray i onization Electrospray i onization, ESI, is a liquid phase technique which allows the transfer of ions from solution to the gas phase [15] The ESI process begins as a solution of analyte is passed through a metal capillary that is held at a high potential, ~3 5 kilovolts as shown in Figure 1 3. Since the electrospray tip is very narrow, ~0.1 m m inside diameter, the electric field at the tip is

PAGE 24

24 very high, ~10,000 volts/cm. The effect of the electric field as the solution emerges is to create a mist of highly charged droplets. Assuming that a positive potential is applied, as in Figure 1 3, the positive ions will accumulate at the droplet surface and be pulled down field toward the less [15] When the applied field is high enough, the cone will be drawn into a filament which produces positively charged droplets when the surface tension is exceeded by the applied electrostatic force. The charged droplets move down an electric field toward the counter el ectrode, i.e. the heated capillary. Ions become desolvated as they move toward the counter electrode. Two mechanisms have been proposed for the desolvation process: the charge residue model (CRM) first proposed by Dole et al. [15] and the ion evaporation model (IEM) first proposed by Iribarne and Thomson [16] Both models assume that the droplets undergo a sequence of Rayleigh instabilities or ntually becomes large enough to overcome the surface tension of the droplet causing it to divide. The CR M predicts ultimately produce droplets which only contain one mo lecule of solute. The solute then becomes a gas phase ion as the last solvent evaporates. The IEM assumes the same sequence of by CRM, the field on the dropl et surface becomes strong enough to overcome solvation forces and lift the solute ion from the droplet surface into the gas phase. All analytes analyzed using ESI were ionized using a custom built ESI assembly Figure 1 4. The assembly consisted of a stainless steel body with two ports at the back end for adding sheath gases to the ESI needle. A piece of stainless steel capillary ran the entire length of the assembly and protruded ~1 cm beyond the end of the larger stainless steel body. A fused silic a

PAGE 25

25 capillary sample tube passed through the stainless steel capillary and terminated ~ 1 mm past the end of the stainless steel capillary. Analyte solutions were passed through the fused silica capillary to be ionized. The ESI assembly was mou nted on an XYZ stage (Newport M 460A, Irvine, CA) to provide reproducible positioning. Voltages for the ESI process were controlled by the native TSQ 7000 software/hardware. Atmospheric p ressure c hemical i onization Atmospheric pressure chemical ionization APCI, is a soft ionization technique like ESI but differs in that it is a gas phase ionization technique. APCI is similar to chemical ionization (CI) in the type of ionization reactions that occur. Unlike CI, it is not possible to produce electrons by means of a heated filament, because APCI operates at atmospheric pressure. In the APCI source, ionization is instead initiated by electrons typically from a corona discharge [17] The APCI process starts with a solution of analyte being passed through a heated nebulizer The heated nebulizer consists of a pneumatic nebulizer to convert the liquid into small droplets and a quart z heater (vaporizer), Figure 1 5 Aft er the liquid has been nebulized and then vaporized, it is moved by a sheath gas (nitrogen in most cases) towa rds the ion formation region where a corona discharge initiates chemical ionization at atmospheric pressure. The APCI process has a high ionization efficiency due to the short mean free path at atmospheric pressure and therefore the increased number of co llisions between the sample molecules and reagent ions. APCI relies on ion molecule reactions that generally involve reagent ions derived from either the solvent or nebulizing gas. These ion molecule reactions include proton transfer, charge exchange, el ectrophilic (positive ions) or nucleophilic (negative ions) addition, and anion abstraction (positive ions) or nucleophilic displacement (negative ions). Most reagent ions are capable of participating in more than one of the listed reactions. In general, the tendencies for proton transfer reactions to occur may be assessed from knowledge of proton affinity values.

PAGE 26

26 With analyte molecules that do not contain acidic or basic sites, a major alternative to proton transfer is ionization to produce molecular io ns by charge exchange (electron transfer). Charge exchange with positive ions takes place between a reactant ion generated from a molecule having a high ionization energy and a sample molecule having a lower ionization energy Negative ions are generate d by charge exchange when the reactant ion has a lower ionization energy than the sample molecule. When molecules have lower proton affinities and are unable to undergo charge exchange, they often undergo either electrophilic addition (positive ions) or n ucleophilic attack (negative ions) to form stable addition complexes. Studies using APCI were done using a customized APCI assembly, Figure 1 6, composed of the heated vaporizer assembly and corona discharge needle holder from the TSQ 7000 APCI assembl y The vaporizer assembly and needle holder were affixed to a piece of steel to hold them in the correct position. A custom length corona discharge needle was made using a #6 sharps sewing needle. The entire assembly was mounted on a movable linear stage (N RC TSX 1A, Fountain Valley, CA). The heat and voltages for the APCI source were controlled using the native TSQ 7000 APCI controls. Mass Spectrometry TSQ 7000 All experiments described herein were performed on a Finnigan TSQ 7000 (San Jose, CA). The inst rument is a commercial triple quadrupole mass spectrometer with the capability of using different atmospheric pressure ionization techniques. Ions produced externally by ESI or APCI enter the mass spectrometer through a heated capillary interface and are sampled and focused into the first quadrupole (Q1) by a series of lenses and multipole ions guides, Figure 1 7.

PAGE 27

27 Fundamentals of quadrupole mass spectrometry A quadrupole mass filter consists of four parallel rods which have a circular or hyperbolic cross s ection arranged symmetrically to the z axis, Figure 1 8. A voltage made up of a DC potential and an RF potential is applied to the rods. Opposite rods are electrically connected and adjacent rods has voltages of opposite polarities ( 1 10 ) ( 1 11 ) where U is the DC potential, V is the magnitude of the RF voltage, and is the frequency of the applied RF voltage. Ions are accelerated into the fi lter with a small accelerating voltage, ~10 20 volts, and oscillate in the x and y directions due to the applied field. The parameters a and q are defined as ( 1 12 ) ( 1 13 ) where e is the charge on the ion, m is the mass of the ion, and 2r 0 is the rod spacing. For an ion to travel the entire length of a quadrupole mass filter, the amplitude of it oscillations in the x and y directions must not increase or, in other words, it must follow a stable trajectory. Ions whose oscillations become unstable will not make it through the mass fil t er and will not be detected. Figure 1 9 shows the stability diagram for which values of a and q are stable. Th e area of the figure under the curve indicate s the regions in which ions will have stable oscillations and will reach the detector. Ions outside this region will be unstable and be lost in the rod assembly. The stability of ions within this regions and lack of stability of ions outside this region is how the quadrupole mass filter achieves separation.

PAGE 28

28 A mass spectrum i s acquired by scanning the values of U and V while keeping the ratio U/V constant. The recorded mass, m is proportional to V so a linear increase in the value of V will correspond to a linear mass scale. The resolution of the quadrupole mass filter can be altered by changing the ratio of U/V and thereby varying the slope of the a/q line, Figure 1 9. Increasing the slope of this line will increase the resolution of the mass filter but will lower the transmission efficiency because as the separating line moves toward the apex, a narrow range of ions will be transmitted. When the DC voltage is switched off completely, RF only mode, all ions will follow stable trajectories for a small RF voltage. This feature is used frequently as a means of ion transmissi on from the point of ionization to the point of mass analysis and to provide collision chambers of high transmission in triple quadrupole instruments. Triple quadrupole instruments such as the TSQ 7000 have the ability to perform several types of mass anal yses. These include single stage full scan MS; two stage full scan MS/MS; selected ion monitoring (SIM); and selected rea ction monitoring (SRM). Single stage full scan MS can be done using either quadrupole on e or three (Q1 or Q3), with the other two qua drupoles used in RF only mode which permits the transmission of all ions. Two stage MS/MS is operated in one of three modes : a product ion scan which looks for fragment ions of one specific parent ion, a parent scan which looks at all of the ions that g enerate a certain fragment ions and neutral loss scan which looks for all parent ions that lose a certain m/z For all of the two stage MS/MS scans, collision induced dissociation (CID) occurs in quadrupole two (Q2). The quadrupole mass filter is also ve ry well suited for SIM because selected ions from any region of the mass spectrum can be monitored without altering the optimum conditions in the ion source or mass analyzer. The mass scale can be changed rapidly and is well stabilized throughout the mass range [18]

PAGE 29

29 High Field Asymmetric Waveform Ion Mobility Spectrometry High field asymmetric waveform ion mobility spectrometry (FAIMS) s tudies were done using a custom built planar FAIMS cell This planar FAIMS c ell was designed to provide high resolution similar to those described by Smith and coworkers [19] The FAIMS cell consist s of two parallel polished stainless steel plates 20 mm wide and 65 mm long in overall size house d in a plastic (PEEK) enclosure, Figure 1 10 A and B The parallel plates were held a uniform distance apart t o create the analytical gap through which the carrier gas moved the ions. The gap was 2 mm for all of the experiments described herein. Further details of the FAIMS assembly will be discussed in subsequent chapters in context with experimental results. The FAIMS cell was constructed such that ions could be injected either parallel or orthogonal to the direction of ion separation, Figure 1 11 A and B respectively. For orthogonal injection, ions first passed through a curtain plate before they enter into the analytical gap through a hole in the counter electrode of the FAIMS cell For all studies reporte d, the hole through the counter electrode used in orthogonal injection was 1.5 mm in diameter. The total path length through the FAIMS cell for ions injected orthogonally is 50 mm from injection to detection. For parallel injection, ions either passed th rough a single curtain plate or a dual curtain plate interface before they enter ed into the FAIMS cell between the two parallel plates. The total path length for ions injected parallel is 65 mm from injection to detection. Ions produced by either ESI or APCI are pulled into the cell through the curtain plate interfaces by means of voltage gradients. Voltages applied to the curtain plates were controlled using either a Bertan 205B 20R high voltage power supply (Hauppauge, NY) for voltages over 400 volts a nd/or a Kepco 400B high voltage power supply (Flushing, NY) for voltages less than 400 volts. For experiments using curtain gas, the gas was added in the region between the curtain plate and counter electrode (orthogonal injection, Figure 1 11 B ) or in the region between the dual curtain

PAGE 30

30 plates (parallel injection, Figure 1 11 A ). Nitrogen was used as both the curtain and carrier gas for all experiments. The nitrogen was obtained from boil off from a liquid nitrogen dewar and was passed through a hydrocarb on/moisture trap (Agilent, Santa Clara, CA) to remove any impurities and dry the gas. Gas flows were controlled using either MKS M100B (Andover, MA) mass flow controllers ( controlled either by a MKS 246C single channel power supply/readout or by a Thermo Fisher Scientific FAIMS waveform generator) or by various sizes of Aalborg rotameters (Orangeburg, NY) The bulk of the gas introduced in the curtain plate region exits through the curtain plate while a portion of the gas is pulled into the FAIMS cell an d becomes the drift gas. The flow of gas into the cell is controlled by the flow rate of gas out of the cell and into the mass spectrometer through the heated capillary, ~0.7 L/min. The opposing bulk flow of gas exiting the FAIMS cell aids in desolvation of ions from the ionization source and prevents neutrals in the ion source from entering into the FAIMS cell The FAIMS cell was interfaced to the TSQ 7000 using an adapter which fit s over the heated capillary of the mass spectrometer, Figure 1 12. This adapter was constructed with six ports arranged annularly around the outside to either remove or add additional gas at the exit end of the FAIMS cell The voltages necessary for sepa ration were applied to the plate opposite to the side of injection for orthogonal injection. For parallel injection, the voltages may be applied to either side. These voltages consist of the asymmetric waveform and compensation voltage, as described earl ier. The waveform generator was provided by Thermo Fisher Scientific (San Jose, CA). Figure 1 13 shows the actual waveform used in FAIMS. The waveform used in the studies described herein are not square waveforms due to the large amount of power required to produce

PAGE 31

31 such a waveform [3, 20]. The waveforms shown in Figure 1-13 are produced electronically by summing two sine waves sin1sin2aVtCfDtfDt (1-14) where C is the CV; f = 0.61; is the frequency of the waveform (750 kHz in this instance); = 90; and D is the maximum voltage of the waveform, which is referred to as the dispersion voltage (DV, Volts0-P). The DV used in FAIMS is also often expressed in terms of field strength defined as 17 / 110 sDdFN (1-15) where Fs is the field strength in Townsends (Td), D is the dispersion voltage, d is the analytical gap in cm, N is the gas number density in molecules/cm3, and 1x1017 is a conversion factor. The electronically produced waveform is amplified via a series of transformers. Dispersion voltages, tuning of the waveform generator, and compensation voltage scans were computer-controlled using Ionalytics Selectra B8 software (Ottawa, Canada). Scope of the Dissertation This dissertation presents the results obtained from a series of experimental and theoretical investigations into the parameters affecting the performance of a planar FAIMS cell. Chapter 1 has provided a brief summary of the operating principles of FAIMS as well as details of the instrumentation used in this study. Chapter 2 will present the experimental results of studies focusing on the modes of injection of ions into the planar FAIMS cell, along with a discussion of the results. Chapter 3 will focus on initial investigations into the effects of solvent vapor on the performance of the planar FAIMS cell and a brief discussion. Chapter 4 will provide a detailed continuation of the studies discussed in Chapter 3 along with some theoretical investigations into

PAGE 32

32 the experimental results obtained throughout Chapters 3 and 4. Chapter 5 will provide a summary of the results from the experimental and theoretical investigati ons. Chapter 5 will also provide suggestions for future studies that might help further the understanding of how differing parameters affect planar FAIMS performance.

PAGE 33

33 Figure 1 1 Hypothetical dependencies of ion mobility of three different ions on elec tric field. K h high field and K field. (Adapted from Purves et al [3] )

PAGE 34

34 Figure 1 2 Ion motion between two plates during the application of an electric potential shown a s V(t). The ion is transported pneumatically by the gas flow. Hypothetical ion mobility dependencies of the ion during different portions of the electric potential are shown on the ABC plot. (Adapted from Purves et al. [3] )

PAGE 35

35 Figure 1 3 ESI process in the positive mode (Adapted from [21] )

PAGE 36

36 Figure 1 4 Picture of ESI assembly used in studies involving ions generated by ESI

PAGE 37

37 Figure 1 5 APCI process in the positive mode. (Adapted from [21] )

PAGE 38

38 Figure 1 6 Picture of APCI assembly used in studies involving ions generated by APCI.

PAGE 39

39 Figure 1 7 A) Schematic showing layout of TSQ 7000 quadrupoles(Adapted from [22] ) B) Schematic showing layout of atmospheric pressure interface(Adapted from [21] ) A B

PAGE 40

40 Figure 1 8 Schematic diagram of quadrupol e mass filter (Adapted from [23] ) Figure 1 9 Mathieu stability diagram for a quadrupole mass filter (Adapted from [23] )

PAGE 41

41 Figure 1 10 A) Schematic showing parts of custom built planar FAIMS cell (arranged for orthogonal injection). B) Picture of assembled planar FAIMS cell (in orthogonal injection configuration). A B

PAGE 42

42 Figure 1 11 A) Shematic showing linear injection into planar FAIMS cell through a dual curtain plate interface. Gas flow is from left to right. Ion injection path shown by red arrow. B) Schematic showing orthogonal injection into planar FAIMS cell through curtain p late interface. Gas flow is from left to right. Ion injection path shown by red arrow A B

PAGE 43

43 Figure 1 12 End view ( A) and side view (B) pictures of planar FAIMS cell adapter used to attach assembly to heated capillary interface of the mass spectrometer. The ports used to add or remove gas at the back end of the FAIMS cell can be seen in (B) A B

PAGE 44

44 Figure 1 13 Sum of sines waveforms used in FAIMS. A) Waveform with negative dispersion v oltage (DV). B) Waveform with positive dispersion voltage (DV ). A B

PAGE 45

45 CHAPTER 2 MODES OF INJECTION I NTO PLANAR FAIMS Introduction Despite the increasing applications of FAIMS, many of the fundamental aspects are not yet well understood. Of interest is how different modes of injection into a planar FAIMS apparatus can affect the performance of the apparatus The mode of injection refers to the method with which ions are injected or allowed to enter into the FAIMS apparatus. The p erformance characteris tics of the planar FAIMS cell can be broken down into two main categories : sensitivity and selectivity. Sensitivity is how the signal for the analyte of interest changes with concentration and can be described either as a relative value (related to how it varies with experimental conditions ) or i n absolute value terms Increases in sensitivity can lower limits of detection and quantitaion. Selectivity can be expressed into two manners : resolving power and resolution. Resolution ( R s ) tells one how well t wo neighboring CV peaks are separated from one another and is defined by ( 2 1 ) where CV 2 and CV 1 are the compensation voltages of the two peaks of interest and and are the peak widths in volts of the two peaks of interest at 10% of thei r full height. In general two peaks in ion mobility spectrometry are considered baseline resolved if they have a resolution above one [24] More important f ro m an instrumental standpoint is the resolving power of t he planar FAIMS apparatus. Resolving power ( R p ) is defined by ( 2 2 )

PAGE 46

46 where is the absolute value of the measured compensation voltage for a particular ion and FWHM is the full width in volts at half maximum height for the CV peak. Resolving power is an instrumental parameter that estimates how we ll the instrument (and the conditions under which it is operating) can separate or resolv e two ions with similar behaviors [24, 25] Increases in selectivity enabl e the FAIMS apparatus to distinguish ions from other components including isomers and interferents. This chapter will focus on manipulating the modes of injection of the FAIMS cell to determine how different operating parameters effect and/or potentially improve performance. Total Ion Mode One aspect of a planar FAIMS cell is that the parallel plates offer a line of sight path to the detector (heated capillary interface of the mass spectrometer in the studies described herein). In concentric cylinder FAIMS, without the presence of the asymmetric waveform, it is difficult for ions to travel around the curve of the electrodes and make it to the detector. With the line of sight pathway, even witho ut the waveform ions may still make it through the apparatus to the detector and be detected for both parallel and orthogonal injection [8] This mode of operation is referred to as total ion mode or TIM. Total ion mode offers the abilit y to see all of the io ns generated at the ions source at the same time. Methods Investigations to determine the optimum conditions for maximum ion transmission in total ion mode (TIM) were conducted on the planar FAIMS cell mentioned previously. The plana r FAIMS cell was set up for ions to be injected parallel to the direction of ion separation. A single curtain plate with a 1 .5 mm diameter hole centered in the analytical gap was used as the curtain plate interface for all of the studies described here. For studies involving only DC voltages, the

PAGE 47

47 voltage was applied directly to the FAIMS plates. For studies involving both DC and radio frequency (RF) voltages, the voltage was applied by means of a RF amplifier circuit and low pa ss filter (described below) T he ortho isomer of phthalic acid (Figure 2 1A) was analyzed in all of the TIM studies described here For all studies using the isomers of phthalic acid, the isomer of interest was dissolved and diluted to a concentration of 50 ppm in 90%/10% methanol/ water with 0.2 mM ammonium acetate The phthalic acid isomers solutions were ionized using negative ESI The ions were detected and analyzed in full scan MS mode on the TSQ 7000. The mass range analyzed was m/z 50 to 450 which included the two major ions of o phthalic acid, the [M H] ion ( m/z 165) and the [2M H] ion ( m/z 331) DC Voltages O nly Experiments were conducted to determine the effects of varying the DC voltage applied to the FAIMS plates In order for any ion s to be transmitted the entire length of the cell, the plates need be set at the same DC voltage. To maintain the same voltages on both plates, they were both connected to the same power supply. The voltages applied to the plates were controlled using a Kepco 400B high voltage power supply (Flushing, NY). Voltages were varied while total ion current was monitored with the mass spectrometer The experiments were performed with two different voltages applied to the heated capillary, 1 V and 50 V. The si gnal for the total ion current was compared to the applied DC voltage on the cell. Figure 2 2 shows the results obtained from these experiments. In general, for both cases, maximum transmission was achieved within 1 to 2 volts of the voltage applied to t he heated capillary. It was also noted that as the voltages applied to the plate became more positive compared to the heated capillary voltage, signal intensity decrease rapidly. At voltages more negative than those applied to the heated capillary, signa l was lower but not to the extent seen at more positive voltages When the

PAGE 48

48 voltage is positive with respect to the heated capillary, ions are decelerated as they approach the capillary and transmission is dramatically decreased. This is not true when the voltage is negative with respect to the heated capillary. RF and DC Voltages Experiments were conducted to determine if the addition of RF voltages to the FAIMS plates improved TIM performance. To add the RF voltage to the plates, either with or without DC voltage, an RF a mplification circuit with a low pass filter was constructed, Figures 2 3A and B. The circuit consisted of an arbitrary waveform generator (Stanford Research Systems DS345, Sunnyvale, CA) used to generate the initial sine waveform to be applied to the plates. The low voltage RF waveform was then input into a pre amplifier (Amplifier Research 10A250, Souderton, PA) to provide initial amplification. This waveform was then input into an aircore style RF coil removed from a Finnigan GCQ (Sa n Jose, CA). Using the entire length of the RF coil produce d a resonant frequency of ~780 kHz. This circuit provided amplification of ~6,000 x with a maximum achievable voltage of 6,000 V PP To add a DC voltage offset to the RF voltage, a low pass filter was constructed and connected to the groun d side of the RF coil. The low pass filter consisted of a 10 k resistor and a 33 nF capacitor. The connection to the RF coil was made between the resistor and capacitor and provided a cutoff frequency of ~800 Hz. RF only The effects of adding only RF voltages to the plates were examined. The RF voltage was added to either both plates or one plate only. Figure 2 4 shows how the ion signa l w as a ffected by adding RF voltage to the plates at a frequency of 794 kHz. The black trace in Figure 2 4 depicts the signal intensity for RF voltage applied to both plates ( Sche matic, Figure 2 3A) The blue trace depicts the signal intensity for RF voltage applied to only one plate ( Schematic, Figure 2 3B) For both cases, there was a decrease in signal intensity as the peak to peak voltage

PAGE 49

49 increased With the RF voltage appli ed to both plates, there was a decrease of ~25% at the maximum RF voltage. Radio frequency voltage applied to only one plate produced an ~18% decrease in signal at the maximum RF voltage. Differences in the starting points of the two graphs are attribute d to the day to day variations of the ESI source. A dditional experiments were performed to determine the effects of increasing the frequency of the RF voltage applied to the plates. The RF coil was connected to the low pass filter circuit in the middle o f the coil decreasing the coil length and increas ing the resonant frequency to ~1.62 MHz. This produce d a maximum amplification of ~1,400x with a maximum voltage of 2,750 V PP The experiments involving RF on both plates and one plate only were re run F igure 2 5 shows the results of these experiments. For both of these experiments, there was essentially no change in signal intensity until the maximum voltage was reached. At the maximum voltage, the signal in both experiments decreased to ~0.3% of the i nitial signal. This loss of signal is attributed to an observed arcing in the cell at the maximum RF voltage. The decrease in signal with increasing RF voltage applied to the FAIMS cell plates is likely due to the oscillations of the ions induced by appl ied RF field. As the RF voltage increases, the magnitude of the oscillations will increase and the likelihood of an ion striking one of the plates becomes greater. Increasing the frequency of the waveform decreases the distance ions travel with each cycl e of the waveform so at similar peak to peak voltages ion oscillations are relatively smaller. RF voltage with DC voltage Experiments were performed to determine the effects of the addition of RF voltage and DC voltage onto the FAIMS cell plates. In this study, the applied RF voltage was kept constant at 1,000 V PP (780 kHz). The DC voltages were applied to the RF coil through the low pass filter described earlier ( Figures 2 3A and B). Figure 2 6 (blue trace) depicts the results of these

PAGE 50

50 experiments. Figure 2 6 (black trace) also shows the results discussed earlier describing the effects of adding only DC voltages to the FAIMS apparatus plates. Both of the traces illustrate the same trend, a maximum transmission within 1 to 2 V DC of the voltage applied to the heated capillary ( 1 V in this case). The addition of RF and DC voltage to the plate did result in a decrease of ~30% over the maximum signal with DC voltage only. For all further experiments involving operating in TIM, only DC vol tages within 1 to 2 V of the heated capillary voltage were applied to the plates without the addition of RF voltage Orthogonal and Parallel Injection One of the advantages that the geometry of a planar FAIMS apparatus affords is the ability to inject ions into the analytical gap either orthogonal or parallel to the direction of separation, Figures 1 11A and B. The ability to inject ions in either mode lends to the flexibility of the planar geometry. For the cylindrical geometry FAIMS cells ions are typi cally only injected orthogonally into the cell [3] Orthogonal injection into the cell offers the ability to more easily prevent neutral molecules from entering into the cell whereas parallel injection lends itself more readily to coupling of the FAIMS device to other analytical methods (i.e. i on m obility s pectrometry) [8] Of particular interest is how these two modes of injection compare in terms of performance and the advantages of operating in one mode versus the other. Studies were performed to determine how each mode of injection performed as various parameters were changed. The parameters examined included residence time, curtain plate style and spacing, and pneumatic focusing. The results of these studies are presented in the following sections. For these studies, either o phthalic acid or trinitrotoluene (TNT) were analyzed

PAGE 51

51 Curtain Plate Effects : Part I Investigations were carried out to determine how changes in the curtain plate interface for orthogonal and parallel injection affected planar FAIMS performance Studies focused on two main aspects of the curtain plate interface: gas flow and curtain gap as t hese two factors were determined to have the greatest effect on sensitivity. In general, the selectivity of the cell was not changed by manipulating the cu rtain plate interface up to the point where the gas flowing into the cell changed in composition (discussed in detail in subsequent chapters). The curtain gap refers to the distance between the exit of the curtain plate and the entrance into the analytica l gap for orthogonal injection and the distance between the dual curtain plates for parallel injection (Figure 1 11) Gas flow rates were the total flow rate of gas into the region between the curtain plate and entrance into the analytical gap for orthogo nal injection and the region between the dual curtain plates for parallel injection. A portion of the gas flow into this region is drawn into the FAIMS analytical gap and become s the drift gas (~0.7 L/min) while the bulk of the gas exits the curtain plat e to become the curtain gas. Orthogonal injection The curtain plate used for orthogonal injection was constructed with a threaded body (Figure 2 7) so that it could be threaded into the PEEK body of the FAIMS cell and be easily positioned The hole throug h the curtain plate was 2.5 mm in diameter. For all of the curtain gap studies, the dispersion voltage (DV) was held constant at 3500 DV ( 70 Td). The voltage applied to the curtain plate was held at a constant 1.45 kV. Adjustments to the curtain plate voltage seemed to not affect the performance of the FAIMS apparatus in any significant manner. For these studies, o pththalic acid was analyzed in negative ion ESI mode To determine how the gas flow rate and curtain gap affected performance, three diffe rent curtain gaps were chosen and the flow rate at those three curtain gaps was varied. The three

PAGE 52

52 curtain gaps chosen were a minimum usable gap a maximum usable gap, and a gap in between the minimum and maximum (1.25 mm, 2.5 mm, and 2.0 mm respectively). For each gap and gas flow rate combination, the compensation voltage (CV) was scanned from 8.2 0 V to 10.2 0 V to monitor the absolute intensity of the CV peak for the [M H] ion of o phthalic acid as well as the full width at half maximum (FWHM) of the pea k. Figure 2 8 illustrates the results obtained from these experiments. The gap between the curtain plate and electrode seemed to have a significant effect on signal, with ~2.5 x more signal with the largest gap when compared to the smallest gap. All three curtain gaps had maximum signal intensities around the same curtain gas flow rates ~4.0 L/min The 2.5 mm gap did produce slightly h igher FWHMs than the 2.0 mm and 1.25 mm ga ps; 0.39 0.00 4 V, 0.36 0.00 5 V, and 0.34 0.00 7 V respectively. T he narrower gaps did produce slightly higher standard errors (standard deviation of the means) in both FWHM and signal intensity All further experiments using orthogonal injection wer e performed with a 2.5 mm curtain gap to maximize sensitivity Parallel injection For parallel injection, a dual curtain plate interface was used to provide a similar type of interface into the cell as was used with orthogonal injection i.e. a 2.5 mm diam eter hole facing the ionization source and a 1.5 mm diameter hole facing analytical gap The dual plate interface consisted of two stainless steel plates separated by a PEEK spacer, Figure 2 9. The curtain plate closest to the ionization source is hereaf ter referred to as curtain plate A (2.5 mm diameter hole) The curtain plate closest to the FAIMS electrodes is hereafter referred to as curtain plate B (1.5 mm diameter hole) As in the orthogonal injection studies, the DV was held constant through out t he experiments at 3500 V (70 Td). Voltages applied to curtain plates A and B were 1,450 V

PAGE 53

53 and 25 V respectively. As was observed with the orthogonal injection studies, curtain voltage seemed to have little effect on overall performance. Two differen t curtain gaps were chosen and the flow rate was varied for each gap. The only two gaps used were 2.0 mm and 2.25 mm as larger or smaller gaps would have required a completely different design to the curtain plate interface. For each gap and gas flow rate combination, the compensation voltage (CV) was scanned from 8.00 V to 10.50V to monitor the absolute intensity of the CV peak for the [M H] ion of o phthalic acid as well as the full width at half maximum (FWHM) of the peak. Figure 2 10 illustrates the results obtained from these experiments. The gap between the two curtain plates did not appear to have a significant effect on the signal intensity The signal intensities for the two different gaps did however maximize at different flow rates ~2. 0 L/min for the 2.25 mm gap and ~2.75 L/min for the 2.0 mm gap The FWHM values for the two gaps were 0.54 0.009 V for the 2.0 mm gap and 0.54 0.007 V for the 2.25 mm As in orthogonal injection, the narrower gap did appear to create slightly higher standard errors in FWHM and signal intensity. All further parallel injection experiments were performed with a 2.0 mm curtain gap as this gap was easier to maintain and neither of the two gaps tested offered a significant improvement. Residence Time An important observation was made in the curtain plate studies. For parallel injection, the FWHM of the CV peaks was broader than the FWHM for orthogonal injection by ~30%. This result seemed to be counterintuitive. Others have reported that the peak wi dths (and resolving powers for that matter) are proportional to residence time of the ion inside the cell. [8, 19, 26] The residence time inside the cell should also be proportional to the length of the cell. If this held true, ions injected parallel should have a longer re sidence time than ions injected

PAGE 54

54 orthogonally under the same flow rate conditions since that offers a longer path length ( 65 mm versus 50 mm ). The opposite was observed. Measurement of the actual residence time and comparison of those times for the two modes of injection could provide some insight into these differences. Methods The residence time of an ion in the planar FAIMS cell can be defined as the time it takes the i on to travel from its entrance into the cell to its exit of the cell. As mentioned previously, if the DC voltages applied to the two FAIMS plates are different, ions will not make it to the exit of the cell. An electronic circuit was created, Figure 2 11 to switch the DC voltage applied to one of the FAIMS plates between two different polarities. The circuit consisted of a double pole double throw electromechanical relay (Hasco HS212 12, New Hyde Park, NY) controlled by an SRS DS345 (Sunnyvale, CA) arbi trary waveform generator. One pole of the relay was connected to a positive DC voltage while the other pole of the relay was connected to a negative DC voltage of the same magnitude 15 V for these studies Both voltages were supplied by a MG PS10AD (Ha uppauge, NY) low voltage power supply. The output s of the relay w ere passed through a low pass filter to prevent any high frequency voltage from feeding back into the DC power supply. To control the relay, a 0 10 V square waveform was sent to the coil of the relay to switch the two poles. By switching the polarity of the voltage applied to one plate while maintaining the other plate at a constant DC voltage, ions c ould be effectively gated. The frequency of the waveform controlling the relay was scanned from low frequency ( cell open for long period) to high frequency ( cell open for short period) until the frequency was too hi gh for any ions to be seen. Ions were analyzed by the TSQ 7000 in selected ion monitoring mode operating with a narrow SIM window (0.3 amu window width) and high scan speed (~30 scans

PAGE 55

55 per second). The high scan speed was used so that at the high frequencies of opening and closing of the FAIMS cell (up to 12 Hz) the mass spectrometer still acquire d several spectra during one open an d close cycle Switching of the FAIMS plate while the asymmetric waveform was turned on was not possible as this created a destabilization of th e capacitively coupled waveform; however, the FAIMS waveform generator was still used to apply DC to the non ga ted plate. For all of the residence studies, o phthalic acid was analyzed in negative ion ESI mode For each different residence time setting, a CV scan was taken at 3500 DV (70 Td) to measure the CV at the centroid of the peak, peak width resolving pow er and intensity of the CV peak for the [M H] ion of o phthalic acid. Next, the residence time was determined in total ion mode using methods discussed later in this chapter To change the residence time of the cell, addition al gas was either removed or added at the back of the FAIMS apparatus through the six ports spaced annularly around the heated capillary adapter (Figure 1 12). Adding additional gas into the adapter increased the residence time of the cell as i t slowed the flow rate through the cell; conversely, removing gas from the adapter decrease s the residence time by increasing the flow rate through the cell The rate of gas flow into the adapter was controlled by a 082 03GL rotameter (Aalborg, Orangburg, NY). To re move gas at the back of the cell through the adapter, the inlet port of the same 082 03GL rotameter was connected to the adapter while the outlet port was connected to a Varian SHO1001UNIV scroll pump (Palo Alto, CA) and the flow rate was controlled using a Nupro B 2JN needle valve (Willoughby, OH). Flow rate into the curtain plate region was set to 3.0 L/min and adjusted to maintain a constant curtain gas flow rate as the flow rate in the cell changed.

PAGE 56

56 Normal residence time in the cell is defined as the conditions where gas was neither added nor removed through the adapter. In this case, the flow rate through the cell is simply controlled by the flow rate into the mass spectrometer through the heated capillary, which is ~0.7 L/min. The measured values of residence time were compared to the average gas residence time which is calculated by ( 2 3 ) where t is the average gas residence time, l is the path length through the cell in mm, A is the cross sectional area of the cell in mm 2 and F is the gas flow rate through the cell in mm 3 /s. In this case, l equals 65 mm (parallel injection) or 50 mm (orthogonal injection), A equals 40 mm 2 and F equals 12,000 mm 3 /s if no gas is added or removed. Results At each gas flow rate through the cell curves were generated showing the ion signal (expressed as the per cent of the fully open signal intensity as a function of the time the plate was open (residence time) An example of this is shown in Figure 2 12. To get the measured residence time for each setting, each curve had two different linear lines fit to it a s shown in Figure 2 13A and B Each fit line was adjusted to give the best visual least squares fit The first line was fit to the curve at the position where the curve crossed through 50% of the fully open signal intensity, as shown in Figure 2 13A. At the 50% mark, approximately 50% of the ions should make it through the cell, which gives a good estimate of twice the residence time. The equation for each 50% fit line was used to calculate the X ( 2x residence time) value for Y=50. This value was then divided by two to give an estimate of the resi dence time. The second line was fit to the curve along the longest linear portion of the curve, as shown in Figure 2 13B. T he equation for each full scale fit was used to calculate where this line crossed the X axis (Y=0).

PAGE 57

57 This gives a good estimate of the residence time (the point at which ions can no longer travel the entire length of the cell before the voltage changes, i.e. the cell closes ) Figures 2 14A and B show the two measured ion residence time val ues with respect to various flow rates through the cell for orthogonal and parallel injection (labels correspond to residence times) The two linear fit methods gave very similar estimates of ion residence time, typically within a few percent. These plots also show the calculated average gas residence time (Equation 2 3) through the cell. In both cases, the measured ion residence time was significantly lower that the calculated average gas residence time Tables 2 1 and 2 2 The dramatic diff erence in the residence times and the deviation from the predicted residence times is believed to be due to the way gas flows into the cell. As the gas flows into the cell through a relative small diameter hole (1.5 mm in both cases), the flow velocity in creases. This increase in flow velocity entrains the ions in the gas stream and accelerates the ions into the cell. For orthogonal injection, this accelerated ion/gas stream enters at 90 with respect to the direction of separation. The ions must make a 90 turn before they travel the length of the analytical gap. This turning process negates the initial acceleration of the ions into the cell. In contrast, for parallel injection, the accelerated ion/gas stream travels straight through the cell at a vel ocity initially higher than the average gas flow. This would decrease the residence time. For both orthogonal and parallel injection, as the drift gas approaches the heated capillary, it is accelerated as the flow velocity through the capillary is much h ID capillary) vs. 0.3 m/s in the cell). This acceleration at the back of the cell may decrease residence times as well. Examples of the CV peaks for orthogonal and parallel injection at the longest res idence time, normal residence t ime and shortest residence time are shown in Figures 2 15A and B. The

PAGE 58

58 data from the CV scans at each residence time are shown in Tables 2 1 and 2 2 for orthogonal and parallel injection, respectively. For orthogonal injecti on, the CV value remained relatively unchanged as the residence time decrease d while the FWHM and intensity increased. For parallel injection, however, as the residence time decreased, the FWHM and intensity increased as well as the CV value. The increa sing CV value in parallel injection is due to the increased flow rate into and through the cell allowing some neutral solvent molecules into the cell. This cause of such CV shift s will be discussed in subsequent chapters. Increasing values of FWHM at sh orter residence times led to a decrease in resolving power values, from 40 to 18 for orthogonal and 31 to 16 for parallel as shown in Tables 2 1 and 2 2 The results also offer some insight into the results observed in the curtain plate studies showing C V peak FWHMs for parallel injection were wider than for orthogonal injection even though the path length was longer. Under normal residence conditions, the measure d ion residence time for orthogonal injection is longer than for parallel injection, 104 ms vs. 71 ms yielding narrower peak widths. Figures 2 16A and B illustrate an important aspect to the results of the residence time measurement studies. For both orthogonal and parallel injection, increasing the residence time does narrow the FWHM of the p eaks. The narrowing of the peaks, however, comes at the cost of signal intensity. For both injection modes, the rate at which signal intensity decreases i s slightly higher than the rate at which the peak narrows. Under the gas composition conditions use d in these studies, it is not possible to achieve high er resolution values (R p greater than 50 in this case) as the extremely low signal intensity at residence times needed to achieve this would be too low to measure in this experiment Another observation made was that the effect on residence time was greater on the FWHM and intensity for parallel injection than for orthogonal injection. The change in FWHM and intensity for parallel injection was ~5 fold compared to

PAGE 59

59 only ~2 fold for o rthog onal injection. This may likely be due in part to the acceleration of ions into the cell during parallel injection both reducing diffusional losses (increases in signal) and reducing the effective length of the cell at a greater rate than in orthogon al injection. Pneumatic Focusing One of the disadvantages of working with an ion separation technique at or near atmospheric pressure is that ions will undergo significant amounts of diffusion [6, 10] Diffusion along the field lines of the applied asymmetric waveform is controlled in curved geometries by the focusing effect but is actually necessary to some extent in all FAIMS cell geometries for the separation of ions by varying CV. Diffusion in a lateral direction, i.e., perpendicular to both t he direction of separation and the direction of the applied field however, is a problem as ions can potentially diffuse far enough to be lost. A series of studies w as performed using a pneumatic focusing gas to attempt to control this lateral diffusion a nd improve signal. The main focus of this study involved the addition of the pneumatic focusing gas at the lateral edges of the FAIMS apparatus (Figure 2 17A) through a porous spacer (Figure 2 17B). The porous spacer was made out of microporous polyethyle ne plastic (Interstate Specialty Products, Sutton, MA). The average pore diameter spacer ran the entire length of the analytical gap. The microporosity of the plastic allowed the pneumatic focusing gas added at the lateral edges to be evenly distributed the entire length of the cell provid ing a uniform focusing effect. Pneumatic focusing gas flow rates were controlled by a MKS M100B mass flow controller and MKS 246C single channel controller (Andover, MA). Th e total gas flow was split by means of a tee to allow equal flow rates to either side of the cell. To compensate for the additional gas added at the sides of the cell, portions of the total gas flow through the cell were removed through the adapter at the exit end of the cell in a similar manner as was described in the residence time measurements.

PAGE 60

60 For this set of studies, trinitrotoluene (TNT, Figure 2 18A) was analyzed For all studies using TNT or other explosive analytes, the analyte was dissolved an d diluted to a final concentration of 20 ppm in 65%/35% methanol/water. The analyte solutions were ionized using negative APCI. The ionization of TNT generat es both M ion s ( m/z 227) and a minor amount of [M H] ion s ( m/z 226). The CV peak intensity for the M ion was monitored for any increase or decrease as pneumatic focusing gas parameters were varied while the DV was held constant at 3800 DV (77 Td) Four different ratios of gas added through the porous spacers t o gas removed at the back were explored. Figure 2 19A illustrates the effects of the pneumatic focusing gas on CV peak intensity for ions injected orthogonally The results showed that the maximum signal intensity increase of ~2x occurred with 900 mL/min added through the porous spacers and with 20% less gas (720 mL/min) removed at the back of the cell Figure 2 20 illustrates two CV scans for the M ion of TNT with (blue trace) and without (black trace) pneumatic focusing gas applied. In this figure, it is apparent that the pneumatic focusing gas does provide improvements in signal without affecting other performance parameters (CV, FWHM, and R p ). Figure 2 19B illustrates the effects of pneumatic focusing gas on the CV peak FWHM. The FWHM for CV peaks when less gas was removed at the back of the cell remained relatively unchanged as the pneumatic focusing gas was increased, with a small amount of peak narrowing with 40% less removed at the higher flow rates For equal gas added and removed, at higher flow rates, the peaks broadened. In general, however, the FWHMs for all peaks were broader than similar experiments run without pneumatic focusing gas and the porous spacers. This is presumably due to the porous spacers decreasing the cross section al are a, A of the FAIMS cell which decreases the residence time in the cell, as shown in Equation 2 3.

PAGE 61

61 Similar studies of pneumatic focusing gas on parallel injected ions showed no improvement on signal A t flow rates observed to give increases for orthogonally injected ions, there was a significant decrease in signal for parallel injected ions This lack of pneumatic focusing effect i s believed to be due to the insignificant lateral diffusion of ions in parallel injection. As discussed in the resi dence time studies, parallel injected ions are accelerated into the FAIMS cell and are likely to stay in a relatively tight area of the cell around the axis they were injected upon ; thus, addition of pneumatic focusing gas should offer little or no improve ment in signal. Cutain Plate Effects: Part II To determine the amount of ion loss through the curtain plate interface, a series of studies were performed to compare operating the planar FAIMS apparatus with and without the curtain plate in place. For bot h orthogonal and parallel injection, TIM scans and CV scans ( 3500 DV, 70 Td) were taken initially with the curtain plate (curtain plate A for parallel) in place. Then the curtain plate was removed and the same TIM and CV experiments were rerun. For thes e studies, o phthalic acid was analyzed in negative ion mode and t he ESI source was a djusted in each case to provide the most intense signal. Table 2 3 summarizes the results of these studies. Figures 2 21A and B show the CV scans under each condition f or orthogonal and parallel injection, respectively. In orthogonal injection mode, the TIM signal was slightly less without the curtain plate and the CV scan signal without the curtain plate was ~20% of the signal with the curtain plate. In parallel injec tion mode, the effect on signal was opposite that of orthogonal injection. The TIM signal was ~10 x higher without the curtain plate and the CV signal was ~33x higher without the curtain plate. The significant loss of signal during the CV scan in orthogo nal injection may have been due partially due to disturbances in ESI electric fields caused by the asymmetric waveform. The

PAGE 62

62 quite significant increases in signal for parallel injection without the presence of the curtain plate (curtain A) may be indicativ e that, at least for parallel injection, the transition through the curtain plate/curtain gas region creates a fairly significant loss of ions. While the signal effects were substantial, another quite dramatic result was observed. In both injection modes, without the presence of the curtain plate, and therefore no curtain gas, the CV value for the [M H] ion of o phthalic acid increased dramaticaly For ESI, the sample was which is ~3 mL/min of solvent vapor (90% methanol, 10 % water) being present at the entrance of the cell. Without the presence of a curtain gas, the solvent vapor generated in the ionization process along with air (~78% nitrogen, ~21% oxygen) is allowed to enter the cell. The air/solvent vapor mixture becom es the drift gas. The CV shifts are most likely due to clustering interactions of the ions inside the cell with the solvent vapor present in the drift gas. Further experiments involving the manipulation of this cluster process by controlling the concentr ation of solvent inside the cell will be explored in Chapters 3 and 4. Summary The studies discussed in this chapter focused on the modes of injection and some of the parameters that can affect the performance in those modes. The results of the total ion mode ( TIM ) studies showed that the maximum signal was achieved when DC voltages within ~1 to 2 V of the voltage applied to the heated capillary interface were applied to the planar FAIMS apparatus plates. The addition of RF voltages offered no improvements. This 1 to 2 V range also holds for operating the planar FAIMS cell with the asymmetric waveform on. A DC voltage bias must be applied to the FAIMS plates to keep them within that range. Studies focusin g on the curtain plate interface showed that the curtain plate gap was an important factor for orthogonal injection but had little effect for parallel injection (at least at the gaps studied). For all further orthogonal injection studies, a gap of 2.5 mm was used. Parallel

PAGE 63

63 injection studies with a dual curtain plate interface had a 2.0 mm gap between the two plates. Without a curtain plate in place, sensitivity was increased significantly for parallel injection but was decreased for orthogonal injection. The results of the curtain plate studies (Part II) showed that without a curtain plate present and with no curtain gas present, solvent vapor was allowed to enter the cell. This produced very dramatic CV shifts and will be explored in further detail in Chapters 3 and 4. Residence time measurements showed a direct correlation between the resolving power of the cell and the sensitivity of the cell to the residence time. The increases in selectivity (increases in R p and decreases in FWHM) came at the cost of sensitivity. The rate of loss of sensitivity was also somewhat higher than the increases in sensitivity. Th ese correlation s are important when determin ing which factor is more important from an analytical problem standpoint. If sensitivity is not imp ortant, higher selectivity analyses may be run and vice versa. Initial studies using a pneumatic focusing gas showed that in orthogonal injection mode, the addition of a pneumatic focusing gas through porous spacers placed at the lateral edge of the planar FAIMS cell can improve signal up to ~2x. At the conditions giving the maximum signal improvement there was also no decrease in selectivity. The inability of pneumatic focusing gas to have any significant effect on parallel injection sensitivity along with the shorter measured ion residence times contribute to the idea of ions being accelerated into the FAIMS cell in parallel injection. The acceleration into the cell both lowers the residence time and keeps ions from diffusing laterally. Generally s peaking, the orthogonal injection mode for the planar FAIMS cell used in these studies is more selective. Peaks were narrower than peaks from similar experiments in parallel injection mode. As the flow rate into the cell increased (to lower resolution an d increase

PAGE 64

64 sensitivity ), orthogonal injection provided a greater ability to prevent neutral molecules from affecting the performance, as was observed in the CV shifts for parallel injected ions at low residence times. Parallel injection did prove to be mo re sensitive, especially when operated without the presence of the curtain plate (Curtain A) and a curtain gas.

PAGE 65

65 Figure 2 1. Structures of three positional isomers of phthalic acid. (A) o isomer (phthalic acid) (B) m isomer (isophthalic acid) (C) p isomer (terephthalic acid). Figure 2 2. Comparison of TIM intensity with respect to the DC voltage applied to the FAIMS plates. The blue trace ( ) is with 50 V applied to the heated capillary and the black trace ( ) is with 1 V applied to the heated capillary.

PAGE 66

66 Figure 2 3. A) Circuit schematic for applying RF and DC voltages to both FAIMS cell plates. B) Circuit schematic for applying RF voltage to one plate and DC voltages on both plates Figure 2 4. Comparison of TIM intensit y versus the peak to peak RF voltage applied to the FAIMS cell plates at 794 kHz A B

PAGE 67

67 Figure 2 5. Comparison of TIM int ensity versus the peak to peak intensity of the RF voltage is for RF voltage

PAGE 68

68 Figure 2 6. Comparison of TIM intensity with respect to the DC voltage applied to the FAIMS p p RF ates.

PAGE 69

69 Figure 2 7. Image of curtain plate used in orthogonal injection studies. A) Top view. B) Side view. The grid has 5 mm spacing. Figure 2 8. CV peak intensity with respect to curtain gas flow rate with three different curtain gaps Black trace ( ) is for 2.5 mm curtain gap. Blue trace ( ) is for 2.0 mm curtain gap. Red trace ( ) is for 1.25 mm curtain gap. A B

PAGE 70

70 Figure 2 9. Image of dual curtain plate assembly used for parallel injection. Figure 2 10. CV peak intensities with respect to curtain gas flow rate at two different parallel injection curtain gaps. Black trace ( ) is for 2.0 mm curtain gap and blue trace ( ) is for 2.25 mm curtain gap.

PAGE 71

71 Figure 2 11. Schematic of relay circuit used to gate ions in the FAIMS cell fo r residence time measurement studies Figure 2 12. Example of residence time curve generated in residence time measurement studies. Plot is percent of full y open signal versus residence time in seconds for orthogonal normal residence time.

PAGE 72

72 Figure 2 13. A) Example of linear fit through 50% mark of residence time curve. The linear fit is shown in blue. B) Example of linear fit through longest linear portion of residence time curve. The linear fit is shown in blue. A

PAGE 73

73 Figure 2 13. Continued B

PAGE 74

74 Figure 2 14. A) Measured orthogonal residence times calculated from residence time curves using either 50% fit (black trace ( )) or full scale fit (blue trace ( )) with respect to the total flow rate of the drift gas through the cell. Also shown are the calculated average gas residence times (red trace ( )). B) Measured parallel residence times calculated from residence time curves using either 50% fit (black trace ( )) or full scale fit (blue trace ( )) with respect to the tota l flow rate of the drift gas through the cell. Also shown are the calculated average gas residence times (red trace ( )). A

PAGE 75

75 Figure 2 14. Continued B

PAGE 76

76 Figure 2 15. A) Comparison of orthogonal injection CV scans for the [M H] ion of o phthalic acid at three different residence times: 68 ms (red trace), 103 ms (black trace, normal residence), and 147 ms (blue trace). B) Comparison of parallel injection CV scans for the [M H] ion of o phthalic acid at three different residence times: 49 ms (red trace), 71 ms (black trace, normal residence), and 102 ms (blue trace). A B

PAGE 77

77 Table 2 1. Summary of results of orthogonal injection residence time measurements. Total Flow Calculated Gas Residence Residence Time ( 50% Fit ) Residence T ime ( Full Scale Fit ) CV FWHM Resolving Power Intensity 0.45 L/min 267 ms 144 ms 149 ms 9.33 0.03 V 0.24 0.01 V 40 32,000 0.58 L/min 209 ms 114 ms 113 ms 9.31 0.03 V 0.32 0.01 V 30 56, 000 0.70 L/min 171 ms 104 ms 102 ms 9.39 0.01 V 0.38 0.01 V 25 8 4,000 0.90 L/min 133 ms 86 ms 86 ms 9.31 0.01 V 0.41 0.01 V 23 9 3,000 1.10 L/min 109 ms 81 ms 77 ms 9.35 0.01 V 0.45 0.00 V 21 11 7,000 1.30 L/min 92 ms 74 ms 73 ms 9.41 0.04 V 0.50 0.01 V 19 117, 000 1.50 L/min 80 ms 66 ms 69 ms 9.53 0.03 V 0.55 0.01 V 18 124, 000 Standard deviation of the mean Table 2 2 Summary of results of parallel injection residence time measurements Total Flow Calculated Gas Residence Residence Time ( 50% Fit ) Residence Time ( Full Scale Fit ) CV FWHM Resolving Power Intensity 0.45 L/min 347 ms 103 ms 100 ms 9.49 0.01 V 0.31 0.02V 31 50,000 0.58 L/min 271 ms 87 ms 87 ms 9.30 0.05 V 0.38 0.02 V 24 91 000 0.70 L/min 222 ms 70 ms 71 ms 9.36 0.01 V 0.45 0.02 V 21 11 4,000 0.90 L/min 173 ms 63 ms 60 ms 9.77 0.02 V 0.58 0.02 V 17 13 5,000 1.10 L/min 141 ms 57 ms 52 ms 11.45 0.04 V 0.68 0.03 V 17 16 1 0 00 1.30 L/min 120 ms 52 ms 48 ms 13.00 0.02V 0.90 0.04 V 15 24 4 000 1.50 L/min 104 ms 49 ms 48 ms 13.93 0.03 V 0.87 0.02 V 16 28 8 000 Standard deviation of the mean

PAGE 78

78 Figure 2 orthogonal injection compared to residence time. B) Normalized FWHM (black trace Values in (A) and (B) are normalized to normal residence time values A B

PAGE 79

79 Figure 2 17. A) Schematic showing position for porous spacers in planar FAIMS apparatus used in these studies. B) Schematic showing porous spacer used in pneumatic focusing studies. A B

PAGE 80

80 Figure 2 18. Structures of explosive analytes. Trinitrotoluene (TNT, (A)). 3,4 dinitrotoluene (3,4 DNT, (B)). 2,6 dinitrotoluene (2,6 DNT, (C)). 2,4 dinitrotoluene (2,4 DNT, (D)). Figure 2 19. A) Normalized int ensity of CV peaks with addition of pneumatic focusing gas with four different ratios of gas removed at back: equal in and out (black trace ( )), 20% less removed out back (blue trace ( )), 30% less removed (red trace ( )), and 40% less removed (purple tr ace ( )). Values normalized to no pneumatic focusing gas values. B) FWHM of CV peaks with addition of pneumatic focusing gas with four different ratios of gas removed at back. A

PAGE 81

81 Figure 2 19. Continued B

PAGE 82

82 Figure 2 20. Comparison of CV scans for the M ion of TNT with (blue trace) and without (black trace) pneumatic focusing gas. Pneumatic focusing gas was applied at 900 mL/min with 720 mL/min removed at back.

PAGE 83

83 Table 2 3. Summary of results of studies wit h and without curtain plate. Orthogonal Injection Parallel Injection TIM Intensity CV FWHM CV Intensity R p TIM Intensity CV FWHM CV Intensity R p With curtain plate With curtain plate 72,100 9.31 V 0.36 V 62,000 26 145,000 9.31 V 0.43 V 131,000 21 Without curtain plate Without curtain plate 65,500 40.86 V 0.57 V 1,150 72 1,980,000 44.65 V 2.00 V 4,380,000 22 Figure 2 21. A) Comparison of CV scans in orthogonal injection mode for the [M H] ion of o phthalic acid with and without a curtain plate in place. B) Comparison of CV scans in parallel injection mode for the [M H] ion of o phthalic acid with and without a curtain plate in place. A B

PAGE 84

84 CHAPTER 3 SOLVENT VAPOR EFFECT S IN PLANAR FAIMS: PART I Introduction One of the interesting characteristics of planar FAIMS observed in the studies dis cussed in Chapter 2 was the dramatic shift in compensation voltage (CV) when the curtain gas was turned off allowing solvent to enter into the cell and become incorporated into the drift gas Under these conditions, the concentration and composition off the drift gas are not well controlled. Investigations into the effects of solvent vapor in the planar FAIMS cell, as described in this and the next chapter, dictate that this concentration and composition should be controlled. As was discussed in Chapter 1 (Equation 1 2), the dependence of K h on high field is dependent on two ion specific terms, and The reasons why different ions have different calculated and terms are not fully understood. Of particular interest is how ion chemistry can affect way ions behave in the FAIMS apparatus and change those terms Previous studies ha ve shown that when water vapor concentrations or contamination in the FAIMS drift gas is high, there are very apparent chemical effects [3, 27 30] When water or other vapors are present in the drift gas, an ion may collide with those neutral molecules and form complexes. These complexes (or solvated ions) in turn may have different mobilites at the t wo different fields used in FAIMS as compared to the bare ion. In FAIMS cells using curv ed geometry, these complexed ions will not be effectively resolved, or may be l os t all together from the FAIMS cell due to the focusing effect [3, 6, 7, 13] Indeed, it is critical that dry gases be used with curved plate geometries. Otherwise, ion CV peaks broaden and signal is often lost altogether [3, 27] Due to this demand for dry gas in curved geometry FAIMS cells it has been shown that careful selection of ionization and injection c onditions at the entrance of the FAIMS cell is critical for

PAGE 85

85 good performance. Particularly important are the flow rate of the solution into the ionization source, the solvent composition of the analyte solution, and the flow rate of the drift gas [27] Eiceman and coworkers showed that concentrations of water in the drift gas above approximately 50 ppm caused CV shifts of organophosphorus ions using a micromachined planar FAIMS apparatus [28] These shifts were explained by a formation of water ion complexes during the lower voltage portion of the asymmetric waveform which were dissoc iated to bare ions during the higher voltage portion of the waveform. The complexation of the ions was enough to change their mobility so that t he difference between the lower field mobility and higher field mobility changed [28, 3 1] This group also reported the addition of dopant compounds to the carrier gas to assist in the detection of explosives [30] Others groups have shown separation o f the [M H] ions of the three positional isomers of phthalic acid (Figure 2 1) by the addition of carbon dioxide to the carrier gas [32, 33] Much of the work exploring the effects of solvent vapors or contaminants has concentrated on either curved FAIMS spectrometers coupled with mass spectrometry or micromachined planar FAIMS spectrometers. Bo th of these geometries offer some interesting insights into the effects of solvent vapor on performance of the FAIMS spectrometer. For curved geometries, detailed investigations into solvent vapor effects using high concentrations are not possible, as the focusing effect causes a dramatic decrease in ion signal or a complete loss of ion signal all together [3, 27, 28] The main focus for research on the effects of solvent vapor in curved geometry has therefore been how to limit and control the levels of vapors allowed to enter the spectrometer. Ion signal in a planar geometry FAIMS spectrometer is not as affected by so lvent vapor due to the lack of ion focusing effect. Micromachined planar geometries however do not offer very high resolving powers due to the small size and short ion residence time. As was

PAGE 86

86 discussed in Chapter 2, the resolving power in planar geometr y FAIMS cell s increases as the residence time inside the cell increases. The use of high resolving power FAIMS cell s, like the one used in these studies, to monitor the effects on performance with the addition of solvent vapor can provide additional insig ht into how solvent vapor affects ions as they move through the FAIMS cell. The studies described here focus on initial investigations to monitor the effects on ions when solvent vapors are added to the cell in controlled amounts. Ions generated by either ESI or APCI were analyzed and the CV value, peak width and ion signal were monitored as the concentration of solvent vapor (methanol and water in these studies) was changed. The effect on CV value, peak width and ion signal as a function of solvent vapor concentration can lead to a better understanding of the ion chemistry inside the FAIMS apparatus. The use of a high resolving power FAIMS apparatus provides a greater opportunity to monitor small changes in CV peaks to help elucidate the progression of s olvent effects. Initial investigations into the analytical potential and practicality of addition of solvent vapor are also explored. The following chapter will expand on the observations made here. Results and Discussion Materials and Procedures Two dif ferent sets of analytes were studied. The three positional isomers of phthalic acid (Figure 2 1) were studied, as previous work has shown that the isomer ic ions exhibit different behaviors in FAIMS and can be separated [33, 34] These isomer s cannot be distinguished using mass spectrometry alone or using conventional IMS [35] Four explosives (Figure 2 18), trinitrotoluene (TNT) a nd three positional isomers of dinitrotoluene (2,4, 2,6, and 3,4) were also studied. Previous work has shown that FAIMS can provide high levels of sensitivity and selectivity for the detection of explosives [30, 36]

PAGE 87

87 The solv ent used in these studies, HPLC grade water and methanol were obtained from Fisher Scientific (Waltham, MA). The phthalic acid isomers were dissolved and diluted to a final concentration of 50 ppm each in 90%/10% methanol/water with 0.2 mM ammonium acetate. The individual isomer solutions were ionized by nega tive ESI, producing [M H] ions ( m/z 165). The individual explosive analytes were dissolved and diluted to a final concentration of 20 ppm each in 65%/35% methanol/water and ionized using negative APCI, producing both M and [M H] ions ( m/z 227 and 226 f or TNT and m/z 182 and 181 for the DNT isomers). Nitrogen was used as both the curtain and drift gas for all experiments. Each analyte was analyzed individually using either dried nitrogen or dried nitrogen with various concentrations of solvent vapor add ed to the curtain/drift gas. To obtain the various concentrations of solvent vapor, dried nitrogen was sparged through a volume of the solvent of interest in a sealed HPLC bottle kept at room temperature (~22 C), Figure 3 1. The solvent saturated headsp ace of this bottle was then taken and mixed with additional dried nitrogen to obtain the final desired concentration. The amount of solvent vapor present in the gas was measured by determining the amount of weight lost after set volume s of gas were passed through the solvent. The setup used here provided solvent saturated gas at the flow rates used in the studies (3.5 3,000 mL/min). The maximum concentration was 17,500 ppm (w/v) for water and 123,000 ppm (w/v) for methanol. To produce various different concentrations of solvent vapor in the drift gas, the solvent saturated gas was diluted with additional dried nitrogen to the concentration of interest. For each experimental condition, the CV was scanned over the range of interest. The mass spectrometer was set to acquire data in selected ion monitoring mode for the ions of interest. The changes in CV value at the centroid of the analyte peak, peak width at

PAGE 88

88 FWHM, and the ion signal intensity were monitored to determine the effects of solvent vapor on be havior in the FAIMS apparatus. Concentration Effects Phthalic acid isomers Figure s 3 2A, B and C show plots of the CV value, FWHM, and normalized signal intensity (normalized to value in dried nitrogen) for the [M H] ions of the t hree isomers of phthalic acid as a function of water vapor concentration in the nitrogen curtain/ drift gas. Figures 3 3A, B, and C show zoomed in plots for the regions at low water vapor concentrations for Figure 3 2. Figures 3 4A, B and C show the same plots for the three phtha lic acid isomers versus concentration of methanol vapor. Figures 3 5A, B and C show zoomed in plots for the regions at low methanol vapor concentrations for Figure 3 4. The three isomers showed different CVs, peak widths and intensities as the concentration of water or methanol increased It is not known if the decrease in signal for all three isomers is a function of the presence of solvent vapor inside the FAIMS spectrometer or due to the effect of curtain gas containing solvent vapor on the electrospray ionization process. The ESI configuration used in these experiments did not use a sheath gas to aid in desolvation of the ions so the presence of solvent in the curtain gas, which is used to aid desolvation, may have affected the intensities of the [M H] ions Intensity effects will be discussed in more detail in Chapter 4. The [M H] ions of the m and p isomers showed similar behavior, especially in the changes in FWHM at lower solvent concentrations and CV shift For all three isomers, there was initially peak broadening at low concentrations. As the concentrations increased, the peak widths returned to approximately the valu e measured in dried nitrogen. The CV values for the isomers all increased with increasing concentration before beginning to level off at the highest

PAGE 89

89 concentrations achievable. The concentration levels at which peak widths return to approximately the original values and CV values b egin to level off are similar The observations made here are most likely due to clu stering interactions between the ions and solvent vapor inside the cell. Eiceman and coworkers stated that concentrations of vapors or contaminants above approximately 50 ppm can affect the performanc e of the FAIMS cell [28] As ions are moved through the FAIMS cell by the drift gas, they are being oscillated in opposite the ions, and asymmetry in the waveform causes the ions to experience two different ion temperatures. During the high er low er field portion of the waveform, clustered with solvent molecules forming clustered ion neutral species. These clustered species will tend to have lower mobilities in the drift gas, as they are larger than the unclustered spec ies. This decrease in mobility will increase the net difference between the high field and low field mobilities. As mentioned in Chapter 1 the CV is used to offset net displacements inside the cell caused by differences in high er and low er field mobili ty. The increase in the mobility difference means that additional voltage will be needed to compensate for the amount of net displacement inside the FAIMS cell increasing the CV. The observation of initial peak broadening and then narrowing supports the theory of clustering of solvent molecules around the ions and suggests a possible mechanism. As solvent vapor is first introduced into the FAIMS cell and the concentrations are low, some ions may have more solvent molecules attached than other ions. Th is creates a broad distribution in the extent of solvation. This distribution also creates a distribution in the mobilities during the low er field

PAGE 90

90 portions of the waveform and therefore a distribution in net displacements for the ions of interest. This r ange of net displacements requires a broad range of CV values to transmit ions effecti vely and creates a broad peak. As more solvent is added, the ion solvent clusters begin to all approach a maximum size and no more solvent molecules can be added [37 39] er field means that the ions no longer exhibit a distribution of mobi lities but have a single low er field mobility. At this stage there should be a single net displacement, as there is only one low er field mobility and one high er field mobility. The peak width returning to approximately the value in dried nitrogen at the se higher concentrations adds support for this theory. Presumably in dried nitrogen, ions would exist with two distinct mobilities as there would be little to no solvent vapor present to create clusters. Other support for this theory is seen in the plots of CV value. As mentioned above, the clustering of solvent with the ions at low er field leads to the increase in CV er field, the CV value should reach a maximum. As the concentration of so lvent vapor increases, there is an initial large increase in CV value. At higher solvent vapor concentrations, the CV values begin to level off, reaching a er field. The plot for the CV va lue for o phthalic acid with methanol vapor concentration also showed a slight decrease in CV value at the highest concentrations achievable. One possible cause of this is that at the high concentrations present, not only is the low solvated, some of the high field ions start to become solvated. This would decrease the net displacement in the FAIMS spectrometer lowering the CV value needed to transmit the ions. Note that no peak broadening was observed at high methanol concentrations. The phthalic acid

PAGE 91

91 the effects on the isomer ions behavior with the addition of car bon dioxide to the drift gas [33] Specifically similar effects were observed in the increase in the CV value as the concentration of carbon dioxide increased and the similar behavior of the m and p isomers Explosives Figures 3 6A, B and C show the plots of CV value, FWHM, and normalized signal intensity (relative to value in dried nitrogen) for the M and [M H] ions of four explosives as a function of the water vapor concentration in the nitrogen curtain/drift gas Figures 3 7A, B and C show zoomed in plots for the regions at low concentrations for Figure 3 6 The M ions of the four explosives showed similar trends in CV value as were observed for the [M H] ions of the isomers of phthalic acid. There was a large increase in CV value as the concentrations of water vapor increased before beginning to level off and approaching a maximum at the highest vapor con centrations achievable. Differences in the behavior of the explosive analytes and phthalic acid isomers were observed in which ionic species reacted. Negative APCI of the explosive analytes produced the M ion as the major ion for TNT, 3,4 DNT and 2,6 D NT, with an [M H] ion at only 0.25 to 2 % relative abundance. In contrast, APCI of 2,4 DNT generated both an M ion and an [M H] ion, with the latter being the 5 to 10 times more abundant than M The behavior of the [M H] ion of 2,4 DNT was very dif ferent from the M ion. As the concentration of water vapor increased, there was a very gradual increase in CV for the [M H] ion as compared to the M ion. It was also observed that as the concentration of water vapor increased, there was an initial lar ge increase in the signal intensity for the M ions before a decrease at higher concentrations. The [M H] ion had very little if any signal increase before decreasing at higher water vapor concentrations. The reasons for the changes in signal and sensit ivity were not investigated.

PAGE 92

92 The increases in the CV values of the ions is likely due to clustering of solvent molecules around the ions during the low-field portion of the asymmetric RF waveform, as was theorized for the phthalic acid isomers above. However, the lack of any significant peak broadening as was seen with the phthalic acid isomers, specifically the mand pisomer, may indicate differences in the way solvent molecules cluster around the ions. As mentioned above, peak broadening may be indicative of a broad distribution of solvation levels at low field. In general terms, some ions may have one solvent molecule attached whereas others may have two or more solvent molecules attached. For the explosive ions, there may be a much narrower distribution of solvation levels. This narrower distribution may in turn yield an ion with two distinct high and low field mobilities, and thus the CV peak width would stay relatively unchanged. Equation 3-1 represents the formation equilibrium for a negative ion and a negative ion f n K ionionsolvent (3-1) clustered with n solvent molecules with a formation constant of Kf. Differences in ion chemistry and clustering mechanisms will yield different Kf values for different compounds and even for different ionic species of individual compounds. Further experiments to help understand this formation equilibrium and potentially take advantage of the behaviors of ions with the addition of solvent vapor are important in the development of FAIMS as an analytical technique. Chapter 4 will focus on some of these. Increased resolution with solvent vapor These experimental results suggest that there may be practical analytical benefits from solvent vapor in FAIMS. Looking at the graphs for the shift in CV with addition of solvent vapor, one can see that in dried nitrogen there is often overlap of analytes. As the solvent vapor concentrations increase, the individual analytes shift at different rates and generally move farther

PAGE 93

93 from one another. Note that a greater difference in CV values (the differences between the centroids of the peaks) does not guarantee improved resolution of A from peak B, since that is also affected by peak width. To clarify the effect of solvent vapo r on resolution, the analytes were analyzed in dried nitrogen and those results were compared to results obtained using a set concentration of solvent vapor added to the curtain/carrier gas. Concentrations were chosen by picking points on the graphs were the analytes showed significant differences in CV as well as reasonable signal intensities and peak widths that were approximately back to the values in dried nitrogen. The values used were ~7,000 ppm (w/v) for water vapor and ~15,000 ppm (w/v) for methan ol vapor. Figure s 3 8A, B and C show three CV spectra for the isomers of phthalic acid acquired under thr ee different solvent conditions, dried nitrogen, water vapor added, and methanol vapor added respectively. Traces for the [M H] ion of individual analytes are shown to aid in distinguishing one compound from another ; however a mixture of analytes did exhibit the same overall CV spectrum. For dried nitrogen (Figure 3 8A ), the m isomer was separated and baseline resolved from the o and p isomers. The o and p isomers showed some separation but were not resolved from one another, and in a mixture would appear as a single peak with two maxima. With the addition of ~7,000 ppm water vapor, the [M H] ions of the three isomers mo ved away from their dried nitrogen CV values by different amounts, significantly increasing their separation. The o isomer moved much more than the other two isomers allowing all three isomers to be resolved. With the addition of ~15,000 ppm methanol v apor, the [M H] ions of the three isomers moved even further from their initial CV positions in dried nitrogen, further increasing the resolution between the individual peaks. Table 3 1 summarizes the CV values, resolving power and resolution for the ind ividual CV spectra. The data shown in the table are

PAGE 94

94 the averages from three scans plus/minus the standard error. The first column in each section lists the CV value at the centroid of each individual peak. The second column lists the resolving power at each peak, as calculated by Equation 2 2 The third column lists the resolution between neighboring CV peaks, as calculated by Equation 2 1 If two peaks have a resolution value above one, the y are considered to be baseline resolved [24] For the dried nitrogen case, all three phthalate ions exhibited different CV values. However, as is seen in Figure 3 8A the o and p isomers are not fully resolved, with a resolution value less than one (~0.67). The maximum re solving power for the FAIMS spectrometer is ~27. With the addition of ~7,000 ppm water, the CV values for all three isomers increased. The order in which the isomers appeared in the spectrum also changed, as the o and p isomer switched positions. The calculated resolution between peaks also increased, with values now well above one. Maximum resolving power increased to ~89 due to an increase in CV value while maintaining relatively narrow peak widths. With the addition of ~15,000 ppm methanol vapor, there were further increases in CV values, although the order of the isomeric ions stayed the same as seen with water vapor added. The calculated resolution between peaks decreased slightly for the m and p isomers but was still well above the baseline r esolution threshold of one. The calculated resolution between the o and p isomer increased dramatically over the value seen with water vapor added due to the very large increase in CV value for the o isomer. The maximum resolving power for the FAIMS spectrometer was ~141, once again due to the large increase in CV value while maintaining relatively narrow peak widths. Figures 3 9A and B show two CV spectra for four explosives acquired under different solvent conditions: dried nitrogen and ~7,000 ppm w ater vapor added to the curtain/drift gas. Individual traces for the M ion of the fours analytes (and for the [M H] ion of 2,4 DNT) are

PAGE 95

95 shown to aid in distinguishing one compound from another. For dried nitrogen (Figure 3 9A ), the M ions of TNT, 3,4 DNT and 2,4 DNT were separated from one another. The [M H] ion of 2,4 DNT, however, completely covered the peak for M ion of 2,6 DNT. This means that in a mixture, 2,6 DNT would not easily be detected without mass spectrometric assistance (i.e. the pea ks are not separated in CV but the M ion of 2,6 DNT is m/z 182 and the [M H] of 2,4 DNT is m/z 181). With the addition of ~7,000 ppm water vapor (Figure 3 9B ), however, all of the species in the dried nitrogen CV spectrum (Figure 3 9A ) shifted to new po sitions and were fully resolved from one another. The order of the ions also changed due to different levels of behavior wit h the addition of water vapor. Table 3 2 summarizes the data for the individual CV spectra shown in Figure 3 9 For CV spectra acq uired using dried nitrogen, all of the ion CV peak pairs noted showed resolutions higher than one (i.e. baseline resolved) except for the [M H] ion of 2,4 DNT and M ion of 2,6 DNT which were overlapping, as noted in Figure 3 9A The maximum resolving p ower the FAIMS spectrometer achieved with dried nitrogen was ~30. For the CV spectra acquired with ~7,000 ppm water vapor added, all of the CV peak pairs showed resolutions well above one and are all baseline resolved. The maximum resolving power the FAI MS spectrometer achieved in these conditions was ~82, due to the increase in CV values while maintaining relatively narrow peak widths. Most interesting, as was seen in Figure 3 6A is the dramatic difference in CV shifts of the [M H] ion and M ion of 2, 4 DNT. With addition of less than one percent water vapor to the carrier gas, these ions went from barely being separated to being separated by more than 17 volts. This same trend was observed for TNT as well. Figures 3 10A and B show plots of the CV va lue both the M ion and [M H] ion of TNT with either water vapor or methanol vapor

PAGE 96

96 added. The TNT M ion and [M H] ion behave quite differently with the M ion showing the greatest amount of CV shift. With ~15,000 ppm methanol vapor added (Figure 3 10B), the two ions are separated by almost 30 V. Differences in the behavior of the two ions will be discussed further in Chapter 4. In general, there were substantial analytical gains made by the addition of solvent vapor to the curtain/carrie r gas used for the FAIMS spectrometer. For the three phthalic acid isomers, adding solvent vapor made it possible to fully resolve the ions from one another. Separation of these ions can also be achieved by also increasing the dispersion voltage or by ad ding carbon dioxide to the carrier gas, but neither approach yields such high resolut ions or resolving powers [30, 33] For the explosives analyzed, not only was there a dramatic increase in the separation of the analytes, there was an added benefit of being able to separate individual ionic species for an individual analyte. There was a 10x decrease in s ignal intensity for the phthalic acid isomers under the solvent vapor conditions used to acquire the CV spectra, but as mentioned above, it is not yet known if the signal decrease is due to FAIMS behavior or ESI behavior. Signal intensities for the explos ive analytes under the conditions used to acquire the CV spectra were approximately the same as the signal obtained using dried nitrogen only. The effects of solvent vapor on signal intensity will be explored in Chapter 4. Summary These studies have shown the changes in behavior of two different groups of ions with the addition of solvent v apor to the curtain/carrier gas in planar FAIMS apparatus The use of a high resolving power FAIMS cell enabled us to better monitor position and shape of the CV peaks for individual ions. In general, the ions showed dramatic shifts to larger CV values with the addition of solvent vapor. The shift to larger CV values supports the theory that ions will cluster with solvent and other molecules inside the FAIMS spectromet er at low fields and then

PAGE 97

97 decluster at the high fields of the asymmetric waveform cycle. By monitoring the peak shape and the peak width, it is theorized that different ions may cluster with solvent molecules by different mechanisms. There are substantia l analytical benefits provided by adding solvent vapor, although the generality of these results remains to be studied. Further investigations into these observed behaviors may lead to a better understanding of and further development of FAIMS.

PAGE 98

98 Figur e 3 1. Schematic showing a pparatus for generating solvent saturated nitrogen. Dried nitrogen comes into HPLC bottle through a sparge tube to generate a headspace containing solve nt saturated nitrogen. Solvent saturated nitrogen is diluted to appropriate concentration with additional dried nitrogen.

PAGE 99

99 Figure 3 2. Plots of CV (A), FWHM (B), and normalized intensity (C) with various concentrations of water vapor for the three positional isomers of phthalic acid: o phthalic (black trace ( )), m phthalic (blue trace ( )), and p phthalic (red trace ( )) A B C

PAGE 100

100 Figure 3 3. Zoomed in areas of plots from Figure 3 2 showing values at low water vapor concentration. o phthalic (black trace ( )), m phthalic (blue trace ( )), and p phthlaic (red trace ( )) A B C

PAGE 101

101 Figure 3 4. Plots of CV (A), FWHM (B), and normalized intensity (C) with various concentrations of methanol vapor for the three positional isomers of phthalic acid: o phthalic (black trace ( )), m phthalic (blue trace ( )), and p phthalic (red trace ( )). A B C

PAGE 102

102 Figure 3 5. Zoomed in areas of plots from Figure 3 4 showing values at low methanol vapor concentration. o phthalic (black trace ( )), m phthalic (blue trace ( )), and p phthlaic (red trace ( )) A B C

PAGE 103

103 A B C Figure 3 6. Plots of CV (A), FWHM (B), and normalized intensity (C) with various concentrations of water vapor for the four explosive analytes : TNT ( M black trace ( )), 3,4 DNT ( M blue trace ( )), 2,4 DNT ( [M H] red trace ( )), 2,4 DNT (M purple trace ( )), and 2,6 DNT (M green trace ( )).

PAGE 104

104 Figure 3 7. Zoomed in areas of plots from Figure 3 6 showing values at low water vapor concentration. TNT (M black trace ( )), 3,4 DNT (M blue trace ( )), 2,4 DNT ([M H] red trace ( )), 2,4 DNT (M purple trace ( )), and 2,6 DNT (M green trace ( )) A B C

PAGE 105

105 Figure 3 8. CV scans of mixture of the three position isomers of phthalic acid acquired with different solvent vapor conditions. A) Dried n itrogen. B) ~7,000 ppm water vapor. C) ~15,000 ppm methanol vapor A B C

PAGE 106

106 Table 3 1. Increased resolution of [M H] ions of phthalic acid with solvent vapor. Dried Nitrogen ~7,000 ppm Water ~15,000 ppm Methanol CV R p R s CV R p R s CV R p R s m 7.37 0.02 V 21 m 29.65 0.02 V 72 m 43.79 0.09 V 114 2.42 4.17 3.4 0 o 8.99 0.02 V 22 p 32.58 0.03 V 89 p 46.53 0.07 V 114 0.67 3.98 18.3 0 p 9.45 0.02 V 27 o 35.56 0.03 V 86 o 62.18 0.10 V 141 Standard deviation of the mean

PAGE 107

107 Figure 3 9 CV scans of mixture of the TNT and three positional isomer of DNT (2,4, 3,4, and 2,6) acquired with different solvent vapor conditions. A) Dried nitrogen. B) ~7,000 ppm water vapor A B

PAGE 108

10 8 Table 3 2. Increased resolution of M and [M H] ions of four explosives with solvent vapor D ried Nitrogen ~7,000 ppm Water CV R p R s CV R p R s TNT (M ) 4.46 0.01 V 14 2,4 DNT ([M H] ) 15.10 0.02 V 40 3.07 4.60 3,4 DNT (M ) 6.29 0.02 V 19 TNT (M ) 18.70 0.03 V 38 1.82 11.34 2,4 DNT (M ) 7.40 0.02 V 28 2,6 DNT (M ) 27.76 0.02 V 79 1.45 1.88 2,4 DNT ([M H] ) 8.25 0.02 V 30 3,4 DNT (M ) 29.18 0.01 V 70 0.20 2.70 2,6 DNT (M ) 8.37 0.01 V 26 2,4 DNT (M ) 31.25 0.01 V 82 Standard deviation of mean

PAGE 109

109 Figure 3 10. A) Plot of CV for the two major ions of TNT with respect to various concentrations of water vapor. TNT M (black trace ( )) and TNT [M H] (blue trace ( )). B) Plot of CV for the two major ions of TNT with respect to various concentration s of methanol vapor. TNT M (black trace ( )) and TNT [M H] (blue trace ( )). A

PAGE 110

110 Figure 3 10. Continued B

PAGE 111

111 CHAPTER 4 SOLVENT VAPOR EFFECT S IN PLANAR FAIMS: PART II Introduction The results show n in the previous chapter illustrate d how the introduction of solvent vapor into the curtain/drift gas can affect the performance of planar FAIMS. The se results a dd ed support to the idea that ions are undergoing a clustering/declustering process as they are oscillated by the asymmetric waveform. A fundamental understanding of this process would provide insight into how ions interact with solvent/neutrals present inside the FAIMS cell. This could also lead to guidelines on how best to implement addition of solvent vapor to the curtain/drift gas in planar FAIMS for improvements in performance To improve this fundamenta l understanding, further investigations were carried out into the role of solvent vapor on performance characte ristics of planar FAIMS. These investigations were focused on two main areas: solvent trends and temperature effects The work shown here help s to better characterize the effects of solvent vapor in planar FAIMS and lead to extension s of the technique as a whole All of the experimental data presented in this chapter were obtained using the planar FAIMS cell in orthogonal injection mode as it has been shown to provide better resolution and resolving powers Solvent Trends T he investigations described in Chapter 3 demonstrated that the use of methanol produced a larger change in both compensation voltage (CV) and peak width at full width half ma x (FWHM) than did water. To determine if the increase in the effects was due to the increase in the size of the solvent molecule, a series of similar experiments was conducted with different solvents of increasing size. The hypothesis was that as the si ze of the solvent increased, the magnitude of the effect on CV and FWHM would increase as well since a larger solvent

PAGE 112

112 field portion of the field mobility more and create a greater difference in mobilities than smaller solvents. Other effects of molecular size could also include steric effects (impacting the number of molecules that could fit around the ion) as well as strength of clustering association (related to hydrogen bonding and other effects). Solvents A series of six solvents were chosen to determine if there w ere any oberservable trend s corresponding with increasing solvent size. These solvents were water, methanol, ethanol, isopropanol, tert butanol (t butanol) and 1 butanol (Figure 4 1) The six solvents used were chosen based on two factors: increasing size relative to one another and similar molecular structure (they all have the same functional group, OH ) Using solvents with similar functional groups meant that the interaction s between the ions in the FAIMS cell and the neutral solvent molecules would likely be of a similar nat ure. Table 4 1 gives some of the physical properties of the solvents including molecular weight, surface area, volume, dipole moment relative polarity (to water) melting point, boiling point and vapor pressure at room temperature. The surface areas, vo lumes and dipoles were calculated using HyperChem 8.0 (Hypercube, Gainesville, FL) emperical method. All six solvents were introduced in the same manner as described in Chapter 3 (Figure 3 1). Table 4 2 lists the maximum concentrations achievable for the solvents. The maximum concentrations were obtained in a similar manner as was discussed in Chapter 3. With the exception of t butanol, all of the solvents were used at room temperature (~2 5 C). Du e to t butanol being a solid at normal room temperature (melting point is 25.2 C), experiments using this solvent were conducted

PAGE 113

113 during a period when the laboratory temperature was elevated due to an air conditioning failure (~30 C). TNT M Ion Solvent con centration Investigations were carried out focusing on the effect of solvent concentration on the behavior of the ions of TNT in the planar FAIMS cell with a fixed field strength (70 Td, 3500 DV) Figures 4 2A and B show the CV and FWHM for the M ion of TNT with respect to increasing concentration of solvent vapor added to the curtain/drift gas. For both the CV and the FWHM, there was a definite trend with respect to the solvent used. This trend fits well with the idea of the clustering model. As the size of the solvent increases, the amount of voltage need to compensate for the difference between the high field and low field mobilities should increase as well if the ions are interacting with the solvents in a similar manner. Figure 4 2A demons trates this trend. As the solvent size increases, the maximum compensation voltage increases as well, from ~25 V for water to ~80 V for 1 butanol. The maximum CV correlates better with molecular volume (Figure 4 3) than with any other parameter in Table 4 1. The FWHM for the CV peaks, Figure 4 2B, also supports the clustering model described in Chapter 3. As the solvent vapor is initial ly introduced, the concentration is likely not sufficient to completely solvate the ions during the lower field portion of the waveform. This incomplete solvation will create a distribution of ion/neutral clusters. For smaller solvents, i.e. water, the difference between clusters sizes will not be very significant and the lower field mobilitie s, and therefore mobility diff erence, will not be large. Small sizes differences would not require a large range of compensation voltages to filter those ions and the CV peak would remain narrow. As the size of the solvent increases, however, the distribution in sizes between cluster s will increase as well. This increase in size distribution will increase the range of CV

PAGE 114

114 values needed to pass these ions, thereby broadening the peak. Going from water to 1 butanol, the maximum FWHM for the CV peaks increased from ~0.4 V to ~2.8 V. As the concentration of the solvent increased, the FWHM of the CV peaks returned to approximately the value observed in dried nitrogen ; water is an exception, presumably because the maximum concentration that can be achieved at room temperature (~17,000 ppm) is too low to reach this point, given the small molecular size. The concentration at which the peaks returned to nearly relative values corresponds roughly to the concentrations at which the CV va lues maximize. Maximization of CV values indicates that a n upper limit of mobility difference has been achieved. In other words, the ions have reached a limit in extent solvation at low field At this point of maximum solvation, only two distinct ion mobilities should be present narrowing the range of CV valu es needed to transmit the ions, i.e. FWHM returns to near dried nitrogen values. An interesting feature of the plots shown in Figure 4 2A is the slow decrease in CV value for the M ion of TNT after the initial large increase as the concentration of solven t vapor increases This observation correlates somewhat with previous work done with mixtures of carbon dioxide and nitrogen used in FAIMS [33, 40] In this work mixtures of the two gases were studied varying from 100% nitrogen to 100% carbon dioxide. The work showed that as the ratio of the two gases changed the CV value for a specific ion did not change in a linear manner. There was an initial increase and t hen a decrease as the concentration increased further. Figure 4 4 illustrates this shift in CV with the addition of carbon dioxide for the M ion of TNT. The f ion mobility in gas mixtures is based on the first approximation of the momentum transfer theory for the mobility and diffusion in gases, and is given by ( 4 1 )

PAGE 115

115 where K mix is the mobility of the ion in a mixture of j different gases, X j is the mole fraction of each individual gas used, and K j is the mobility of the ion in th e pure gas [41] Equation 4 1 implies that the mobility of an ion should change in a linear manner as the mole fraction of the gas is changed. A plot of 1/ K mix versus the mole fraction of the gas should give a linear positively or negatively sloped line for a binary mixture of gases dependi ng on which of the two [41] The fields explored in these studies reach a maximum of approximately 120 Td. Deviations from linearity of up to 20 percent, especially at h igh fields, can occur if the two gases used in the system differ significantly in the molecular weight (i.e. N 2 and He mixture, where N 2 is seven times heavier than He) [41] The deviations from linearity observed with the addition of carbon dioxide are much larger than would be predicted by molecular weight differences alone. If a nalyses could be done using 100% solvent vapor (water or alcohol) as the drift gas, it is likely that the deviations from linearity would be more than predicted by molecular weight difference alone as well. Larger deviations from linearity indicate inter actions (clustering) with one or both of the gases used in the system s [41 43] The maximum deviation from linearity is indicative of the concentration at which there is the greatest amount of interaction, i.e. largest difference in cluster sizes during the asymmetri c waveform. are in agreement with the model (hypothesis) of ion clustering with solvent molecules in FAIMS. Field effects Investigations into the effect of increasing field strength were performed with a fixed concentration of solvent vapor added to the curtain/drift gas. For all of the solvents investigated, the concentration s were fixed based on the poin t at which there was a maximum shift in CV value (Figure 4 2A) Table 4 3 The concentration was held constant while the field strength was increased. Figures 4 5 A and B show the effect of increasing field strength on CV value and

PAGE 116

116 FWHM for the M ion of TNT. For all of the drift gas compositions, the increase in field strength also led to an inc rease in CV. The trend in the magnitude of CV shift ob served as a function of solvent size was comparable to what was observed with the concentration effect investigations (Figure 4 2A) with water causing the least shift and 1 butanol causing the greates t shift. To determine the exten t of the effect of field strength on the ion behavior, t he and terms for each condition were calculated using Equation 4 2 : ( 4 2 ) where C is the compensation voltage, D is the dispersion voltage, and d is the analytical gap size in cm [44] The calculated and terms were used to calculate the K h /K values (Equation 1 3) for the maximum field strength examined. The se calculated values are shown in Tabl e 4 3. Figure 4 6 shows a plot of the K h /K value s for each solvent condition with respect to the increasing field strength. The inset in this figure is for the K h /K value for the M ion of TNT in only dried nitrogen. These plots illustrate that i n dried nitrogen, the ion has B type behavior, where there is an initial increase in mobility as field increases before beginning to over ( K h /K is no longer increasing) With water, the ion still exhi bits some B type behavior As the size of the solvent increases, th e ion no longer exhibits this B type behavior. This loss of B type behavior is attributed as an enhancement of the A type behavior with the addition of solvent. The effect of increasing field on FWHM (Figure 4 5 B) illustrated an interesting p henomenon. There i s still a trend correlating with amount of effect and solvent size. More important to note i s that for all of the solvents investigated, the increasing field strength increase s the FWHM. The reasoning behind this is not fully understoo d. There are two possible

PAGE 117

117 explanations for this result. As the field strength increases, the amount of kinetic energy that is imparted to the ion and the ion/neutral cluster increases. The increase in kinetic energy, and therefore thermal energy, may be gin to prevent some of the ions from reaching a fully solvated state. As was mentioned in the solvent concentration discussion, this lack of fully solvated states potentially creates a distribution in ion mobilities at the lower field portion of the wavef orm. It would be assumed that if there was a perturbation of full solvation, there would also be a decrease in CV shift. There is no observed decrease in CV, but there is a decrease in the ratio of CV to field strength (slope of lines in Figure 4 5A). T his may be indicative of lessening of the clustering process. O ne other possibility for this increase in FWHM is that as the thermal energy increases, there may be changes in the orientations or arrangements of the neutral s clustered around the ion Ions that are fully solvated but have different cross sections due to orientation differences would produce slightly different CV values, broadening the CV peak. For both of these theories, the effect would be more pronounced with a larger solvent, as is obse rved in the trend. This increase in FWHM with increasing field coupled with the increasing CV values generates resolving power (R p ) values that initial ly increase and then decrease, as shown in Figure 4 7 Figures 4 8 A and B illustrate a field and solve nt vapor effect that was not predicted based on observations from previous work. As the field strength increases there i s a dramatic increase in the signal intensity. Figure 4 8 A illustrates the signal intensity with increasing field strength compared to the signal with no FAIMS cell present (dashed line). The signal without FAIMS present was acquired by removing the FAIMS cell and performing APCI in a nitrogen environment which most closely match es the conditions at the curtain plate interf ace. The figure demonstrates that under some conditions, the signal with FAIMS is actually greater than

PAGE 118

118 that without FAIMS present. However, t his comparison has some uncertainty, since signal reproducibility can vary widely. Figure 4 8 B shows that the i ntensity data normalized to the intensity at the lowest field, supports the solvent trend idea. This figure shows the intensity of the CV peak for the M ion of TNT normalized to the intensity value a t 50 Td with respect to increasing field strength. Th e smaller solvents generate less of an increase than the larger ones. With the addition of either butanol (green and orange trace s), there i s a maximum of > 60x more signal with increasing field stren gth. The reasons for the dramatic increase in signal int ensity with the addition of solvent and increas ing field are not fully understood. One potential explanation is based on the occurrence of collision induced dissociation (CID). The use of the asymmetric waveform oscillates ions in the FAIMS cell between two velocities as described previously. As the ions move rapid ly through the drift gas at high field they are undergoing multiple collisions which can impart small amount s of energy to the ion. As the ions continue to oscillate through the drift gas at low field some of this energy is lost through further collisions If the ions retain some of this energy and the total becomes high enough, the ion may undergo some CID, losing the lowest energy fragment. In FAIMS, if a parent ion fragments the fragment ion will now have a different CV value and be lost. This would cause a decrease in signal for that parent ion. With the addition of solvent vapor and the formation of clusters at the lower field, the energy that is imparted upon the ion may eith er distributed amongst the solvent cluster molecules or the CID c ould simply remove solvent cluster molecules This would allow the parent ion to remain intact, increasing the ion signal. TNT [M H] Ion Versus M Ion Atmospheric pressure chemical ioniza tion of TNT produces an M ion as the major ion. Th is ionization process also produces a small amount of an [M H] ion as well. Using dried

PAGE 119

119 nitrogen as the drift gas for FAIMS, these two ions typically have approximately the same CV value except at the highest fields achievable. Even at these high fields, the two ions are separate d by less than 1 V ( CV= 8.2 V (M ) vs. 7.4 V ([M H] )). During the initial solvent vapor investigations, it was observed that with the addition of solvent vapor, these two ions trended away from one another in CV (Figures 3 10 A and B). Investigations were performed to determine how the [M H] ion was affected by solvent vapor behavior differed from the M Figures 4 9 A and B illustrate the behavior of the [M H] ion of TNT with respect to increasing concentrations of solvent vapor. For all of the solvents investigated, the maximum CV value for the [M H] ion Figure 4 9 A, was only as high as the maximum CV value for the M i on of TNT with water vapor added (dashed line in figure). The CV values also are still increasing at the maximum solvent concentrations in contrast to the trends for the M ion (Figure 4 2A) The FWHM for the CV peaks, Figure 4 9 he initial increase then decrease observed for the M i on. There was also a lack of an observed trend corresponding to solvent size. This still increasing movement of the peak width and CV value may indicate that these [M H] ions have not reached their ion. Figures 4 10 A and B illustrate the effect of increasing field strength on the CV and FWHM of the [M H] ion of TNT measured under the same conditions as the field experiments for the M ion of TNT. Figure 4 1 1 shows the calculated K h /K values for this ion with respect to increasing field strength. The data for the CV values show that the organic alcohol s all produced similar amounts of CV shift. The plot of K h /K values demonstrates that unlike the M ion (Figure 4 6) the [M H] ion has a B type ion behavior under all of the solvent conditions. The effects on intensity, Figure 4 12, are also quite different than what was observed for the M ion (Figure 4

PAGE 120

120 8B). For the [M H] ion, t here is only a 3x increase in intensity over the lowest field strength compared to >60x increase for the M ion. The discontinuity for the dried nitrogen trace in Figure 4 12 is due to the [M H] ion overlapping with the M ion at lower fields before bein g separated at higher fields. The investigations into the behavior of the [M H] ion of TNT indicate that it behaves quite differently from the M ion. These differences arise from different types of interactions between the ions and the solvent molecules present in the drift gas. All of the results show that the amount of CV shift, FHWM changes, and intensities increases are all smaller for the [M H] ion. The B type behavior of the ion gives some insight into what might be happening. It is th eorized t hat A type ion behavior is due to interactions between the ions and the drift gas along with neutrals present in the gas. The transition of A type behavior to the B type behavior is thought to be due to increasing field strength further heating the ions reducing interactions with the drift gas [44 46] The presence of the B type behavior for the [M H] ion and suggests that the interactions between the solvent and the [M H] ion are weaker than with the M ion. The lower magnitude of CV shift with increasing concentration also supports this idea. With a weaker interaction between the ion and solvent, smaller effects should be observed. The weaker interactions of the [M H] ion with solvent vapor compared to the M ion may be attributed to observations made concerning how the ions are produced in APCI. Previous studies have shown that the M ion of TN T is generated by a transfer of an electron from the reagent ions in the APCI source [47] Th ese studies go on to show that the production of the [M H] ion occurs when oxygen is present. The mechanism of the proton abstraction goes through an intermediate [M O 2 ] before generating the [M H] io n It is hypothesized that t he differences between the two TNT ions with the addition of solvent vapor may be due to the ability of the M ion to complex

PAGE 121

121 with oxygen containing molecules ( OH group or water) whereas the [M H] ion may interact less with oxygen containing molecules yielding less effects in the FAIMS cell. The plot of the CV values for the [M H] ion with respect to increasing field (Figure 4 10A) showed that all of the organic alcohols produced similar CV shifts. If fact, at the highest f ield strength, there was only ~3 V difference between the five alcohols, compared with an ~70 V difference for the M ion (Figure 4 5A) The similarity of the alcohol da ta on CV shift along with the B type behavior may be indicative of something more than clustering affecting the behavior of the ions. The weak interactions with the solvent may be changing the way the ions experience the applied field rather than changing the ions size through cluster formation. In general though, it is quite apparent tha t the two ions of TNT (M and [M H] ) are affected very differently by the addition of solvent vapor to the drift gas in planar FAIMS. Phthalic Acid Isomers As was discussed in Chapter 3, the addition of solvent vapor can also be used with ions generated b y electrospray ionization (ESI) to yield performance enhancements. Specifically, the [M H] ions of the positional isomers of phthalic acid were investigated. Building on what was learned from the experiments looking at solvent trend effects with the ion s of TNT, parallel investigations were carried out using the o phthalic ac id isomer to see if there were similar trends. For the studies with o phthalic acid, we encountered an unforeseen barrier to getting a complete set of data for the range of alcohol s used in the TNT studies. Figure 4 1 3 illustrates why the full range of solvents was not investigated. This figure show s the CV spectrum for the [M H] ion of o phthalic acid acquired at 95 Td ( 4700 DV) with either dried nitrogen (left side of plot ) or dried nitrogen with ~14,000 ppm isopropanol added to the drift gas (right side of plot ). At this field strength, the addition of isopropanol produces a ten fold shift in CV, moving

PAGE 122

122 the CV value from ~17 V to ~170 V. The large shift in CV also increases the R p value from ~45 in dried nitrogen to ~304 with isopropanol added. Increasing the field strength further to 97 Td (the highest achievable), moves the CV value to ~191 V. The issue encountered was that the maximum CV achievable with the waveform gen erator used in these studies was 200 V When experiments using 1 butanol were attempted, the CV was above this 200 V maximum The data that were acquired for the [M H] ion of o phthalic acid does however, illustrate a similar trend as was observed for t he M ion of TNT Figure 4 1 4 shows the CV shift for the [M H] ion of o phthalic acid with respect to increasing field strength at fixed concentrations of solvent vapor added to the carrier gas (same concentrations used for the TNT studies). Although on ly four solvents were investigated, they did show the same trend as the TNT M ion (Figure 4 5A) with respect to increasing solvent size increasing CV shift. The similarities between the M ion of TNT and the [M H] ion of o phthalic acid are also illustr ated in the plot of K h /K versus the dispersion field, Figure 4 1 5 (Compared to Figure 4 6) There was the same A type behavior enhancement with increase solvent size. This is where the similarities end To start, the nature of the two ions is very different with the o phthalic acid isomer producing a [M H] ion compared to the M ion for TNT. T he plots of FHWM and normalized intensity for the [M H] ion of o phthalic acid, Figures 4 1 6 A and B respectively, were quite different than what was observ ed for the M ion of TNT (Figures 4 5B and 4 8B) Whereas there were definite trends in increasing intensity and increasing FWHM for the TNT M ion there was not any discernable trend for FWHM of the [M H] ion of o phthalic acid relating to solvent size For the FWHM, Figure 4 1 6 A, the width stayed relatively constant over the dispersion field range, with the larger solvents having very slightly larger peak widths. The lack of any broadening in the peak widths lead to significant increases

PAGE 123

123 in resolving powers. With the addition of 14,000 ppm isopropanol, the R p for the [M H] ion of o phthalic acid increased from ~50 in dried nitrogen to ~370 at 97 Td ( 4900 DV). In contrast to the huge increases in intensity observed for the M ion of TNT (Figure 4 8 B ), the intensity for the [M H] ion of o phthalic acid, Figure 4 1 6 B, decreased with respect to increasing field strength. T he addition of solvent vapor to the drift gas caused the decrease in intensity to be more rapid than with no solvent vapor present. This increased rate of intensity loss showed no trend related to solvent size, with all the solvents examined producing very s imilar results. The loss of signal as the field strength increases is generally what is predicted in planar FAIMS due to increases in di f fusional losses at the higher fields [7, 48] This predicted decrease in intensity with field is why the increases observed for the M ion of TNT were quite unexpected. While there was a loss of intensity with increasing field strength for o phthalic acid, the absolute intensity at lower fields was actually higher (i.e. 140,000 counts in dried nitrogen versus 187,000 counts with isopropanol added at 50Td) A t 70 Td, with the addition of solvent vapor, the [M H] ion signal had decrease d more compared to lower field strengths but the signal was still higher than with no dried nitrogen added. Figure 4 1 7 illustrates CV scan s of a mixture of the [M H] ions of the three positional isomers acquired in dried nitrogen (left of line) and the same mixture analyzed with ~14,000 ppm isopropan ol added to the drift gas (right of line). Both CV scans were taken at 70 Td ( 3500 DV). The signal w ith the solvent added is ~2x higher. With the addition of isopropanol to the drift gas, the CV values for the for the [M H] ion of m phthalic acid incr eased from 8.83 V to 41.12 V and the CV values of the [M H] ions of p phthalic acid and o phthalic acid increased from 10.02 V (overlapping peaks) to 45.17 V and 75.82 V respectively. This led to increases in both R p (maximum of ~30 in dried nitrogen ver sus

PAGE 124

124 maximum of ~145 with isopropanol) and increases in R s specifically for the [M H] ions of the p and o isomers (overlapping peaks to R s >32). Solvent Trends Summary To support the idea of clustering interactions between the ions in the FAIMS cell and any solvent molecules or other neutral molecules present, a series of solvents were examined. The solvents were of similar nature, all containing a OH functional group. There was variation in the molecular size of the solvent ranging from water to 1 butanol. The results of these experiments showed that, at least for the M ion of TNT and the [M H] ion of o phthalic acid, there was a trend in CV shift correlating to t he size of the solvent molecule. The larger solvent molecules produced a large CV shift. This trend fits well with the idea of solvent molecules clustering to the ions at the lower field portion of the waveform and then declustering in the higher field portion An ion with larger solvent molecules clustered around it when compared to the same ion with smaller solvent molecules (i.e. 1 butanol vs. water), would have a larger depression in the lower field mobility. This would result in a larger CV needed to offset the differences in mobility. For the M ion of TNT, there was also a trend correlating to solvent size in both the FWHM broadening and intensity with increasing field strength. When comparing the three ions discussed here, there were differen ces observed in the way the ions behaved. Specifically, the three ions had different trends in intensity, K h /K values, and FWHM. When looking at only CV shifts, it is apparent that the addition of solvent vapor to the drift gas definitely affects the ion Based on deviations from deviation s from linearity are indicative of ion/neutral interactions. The differences in the other the neutral solvent molecules are different in nature.

PAGE 125

125 Temperature Effects As has be en discussed previously, it is hypothesized that ions are experiencing a clustering and declustering process in the FAIMS cell on the timescale of the asymmetric waveform. Th e cluster/decluster process is believed to be dependent on the instantaneous temperature of the ion generate d by the applied field. As the ions alternated between two different applied field strengths, they experience two different velocities and kinetic energies and therefore temperatures. Equation 4 2 shows the theoretical equilibrium between an ( 4 3 ) ion and an ion clustered with n solvent molecules. The formation equilibrium constant of the cluster complex, K f can vary depending on the ion of interest and th e solvent used. This variation is illustrated b y the different effects of varying solvent and difference between similar ions (isomers). Th e formation constant is a thermodynamic property and should therefore be subject to temperature changes. In fact, as mentioned above, temperature changes are the reason for the cluster/decluster process. Being able to better explain how this process is occurring from a temperature (and thermodynamic) standpoint is important to the development of the application of solvent vapor for improved performance in planar F AIMS. To develop some initial understanding of what effect temperature has on the behavior of ions in the FAIMS cell, studies were conducted focusing on varying the temperature inside the cell itself. The initial hypothesis was that at some temperature, i nteractions with the solvent vapor present would no longer be favorable and the ions would behave similar ly to their behavior in dried nitrogen. This ob servation was not made at the temperatures achievable in these experiments. However, there were some very valuable insights into the ion/ neutral chemistry going on inside the FAIMS cell.

PAGE 126

126 Methods To raise the temperature inside the FAIMS cell is not as simple as one would think. Due to the application of the high field to the cell, heating the cell by ele ctrical means is not possible, or very difficult at the least. There have been discussions focusing on heating the cylindrical style FAIMS cell to impro ve the performance [49] To heat the cylindrical FAIMS cell used in th o se experiments, the authors pass hot gas th rough the cores of the electrodes The purpose of heating in th o se experiments is to modify the field profile between the FAIMS plates to affect the performance of the cell. The commercial Thermo FAIMS instrument (San Jose, CA) incorporates this type of heating. For the planar FAIMS cell used in the work described here, passing gas through the core of the electrodes was not possible as the electrodes are fairly thin (~4.5 mm) and drilling gas passages through them might distort their planarity. It was decided that heating of the planar FAIMS cell would be done by means of heating the curtain/drift gas. To heat the curtain/drift gas, the gas was passed through either a process heater (Omega AHP 756, Stamford, CT) or through a coil of copper tubing wrapped with heating tape (Briskheat, Columbus, OH). The temperature of the process heater and the heating tape w as controlled using a vari able transformer (Powerstat 116, Bristol, CT). To minimize heat loss, the gas traveled only ~10 cm from the exit of the heater before entering the curtain plate region of the cell. The temperature of the gas was measured at the exit of the curtain plate and inside the FAIMS cell. The maximum temperature achievable was ~135 C at the curtain plate which corresponded to ~85 C inside the cell. For all temperatures investigated, the temperature of the cell was held at the temperature of interest for 30 minut es before experiments were performed. All of the studies using solvent vapor added to the curtain/drift gas used the same fixed

PAGE 127

127 concentration of solvent vapor used in the field effect experiments discussed previously in this chapter. Temperature Ramp One important aspect of FAIMS that must be considered when focusing on temperature inside the cell is the field strength applied in the cell. As was mentioned in Chapter 1 (Equation 1-15), the field inside the FAIMS cell is dependent not only on applied voltage but also on the gas number density, N (molecules/cm2). Gas number density, N, is defined by: 6 110 bPNTk (4-4) where P is the pressure in Pascals of the experimental setup, T is the temperature in Kelvin, and kb is Boltzmans constant. As temperature increases, the gas number density decreases and the applied field strength increases. Appendix A shows a list of dispersion voltage values and their corresponding field strength in Townsend at room temperature (~25C) and at ~85C in the FAIMS cell. The effect of increasing temperature of the cell on the changes in performance with addition of solvent vapor focused on the behavior of the [M-H]ions of m-phthalic acid and o-phthalic acid. The pisomer is not shown but exhibits similar behaviors to the m-isomer. All of the investigations with increasing temperature were done at -3500 DV. The increasing temperature meant that the actual applied field increased from ~70 Td at room temperature (normal operating temperature) to ~85 Td at the highest achievable temperature. Figure 4-18 illustrates the change in CV with respect to increasing temperature of the gas at the curtain plate (i.e. temperature in cell) for the two isomers in either dried nitrogen, water added to drift gas or methanol added to drift gas. For all three solvent conditions, there was a decrease in CV with increase in temperature. In dried nitrogen, the CV shift was similar for both

PAGE 128

128 isomers. There were differences noted with the addition of the two solvents. Both isomers showed different amounts of change. The m isomer (dashed lines) showed a smaller d ecrease with respect to increasing temperature, particular ly with methanol added. Figure 4 1 9 shows the FWHM for the two isomers with respect to increasing temperature. The FWHM for the two isomers in general show ed some amount of increase with increasing temperature. However, the amount of increase in FWHM was different for the two isomers. Specifically, with methanol added, the FWHM for o p hthalic acid was ~1.33 V at 120 C whereas the FWHM for m phthalic acid was only ~0.47 V. heat and the differences in FWHM, it becomes apparent that the two isomers have different interaction potentials. The lesser effect of temperature ob served for the [M H] ion of the m isomer may indicate that the clustering interaction is stronger than for the [M H] ion of the o isomer, i.e. solvents more tightly bind to the ion. At the temperatures studied the CV s did not decrease back to the val ues observed in dried nitrogen. This implies that at the temperatures studied, there are still interactions going on between the solvent and the ions in the FAIMS cell. One interesting observation was made during the temperature studies. As the temperatu re was increased, the re was an initial increase in signal intensity when solvent was present, before decreasing at the highest temperatures studied. Figure 4 20 illustrates the maximum intensity achieved for the o isomer and the peak data for th e se condit ions. It is unclear at this point if the increases observed are due to increase s in ionization efficiency due to better desolvation with the addition of heat or are strictly due to heating of the FAIMS cell alone.

PAGE 129

129 Field Effects As has been discussed previ ously, the asymmetric waveform generates two different velocities for ions due to the two different ion mobilities. These two different velocities equate to two different ion temperatures and is the principle behind why ions undergo a cluster/decluster process. As the strength of the field increases, the ion velocit ies increase as well, leading to higher effective ion temperatures. This increasing temperature with higher field strength i s generally believed to be why ions that are initially A type ions become B type ions [4, 44] The A type behavior is presumably due to interactions between ions in th e FAIMS cell and neutral molecules in the drift gas. As mentioned previously in this chapter, the addition of neutral molecules into the drift gas enhance s the A type behavior. The transition to B type behavior with increased field strength is thought to be due to increasing ion temperatures at the high fields preventing or limiting interactio ns of the ions with neutrals. Heating of the FAIMS cell at a fixed field showed a shift to lower CV values indicating a change in the extents of interaction. To determine how temperature in conjunction with increasing field affected the behavior of ions in the planar FAIMS cell, the field effect studies of TNT (both M and [M H] ions) were repeated at an elevated temperature. The temperature inside the FAIMS cel l for these experiments was ~85 C (~135 C at the curtain plate). The changes in CV, FHWM, and intensity were monitored as the field strength inside the cell was increased. Only five of the solvents used in the room temperature field effect studies were i nvestigated at elevated temperature. The use of t butanol was not possible as lab temperatures had returned to normal and is was n ot possible to generate solvent saturated nitrogen. It should be noted that the maximum applied dispersion voltage was lowe r at elevated tempera tures due to observed voltage breakdown at the highest

PAGE 130

130 voltages. Taking this into account, the applied field strengths were still higher due to decreased number density at the elevated temperature (Equation 4 4) with a maximum of ~12 0 Td compared to ~ 97 Td at room temperature Figure 4 2 1 A illustrates the CV for the M ion of TNT at a FAIMS cell temperature of 85 C with respect to increasing field strength. Figure 4 2 1 B illustrates the calculated K h /K values for the six solvent conditions. Figure 4 2 2 A and B show the same plots for the [M H] ion of TNT. From the plots of CV (Figures 4 2 1 A and 4 2 2 A) it is apparent that the addition of heat lessens the amount of CV shift for all systems studied. For the M ion, the maximum C V observed at 85 C was ~35 V with the addition of 1 butanol compared to the value at room temperature of ~145 V. The CV values for the [M H] ion at 85 C actually showed a decrease below dried nitrogen values at high field strengths for dried nitrogen and water. All other solvents conditions showed a decrease at high fields as well for this ion. When looking at the K h /K plots for these two ions (Figures 4 2 1 B and 4 21B), it is quite apparent that under these conditions, both ions a re exhibiting B type b ehavior. It was also noted that in dried nitrogen and at the highest field achievable, the K h /K value for the [M H] ion goes below one indicating more of a C type ion behavior, where the higher field mobility is less than the lower field mobility. Figures 4 2 3 A and B show a comparison of the K h /K values for the M and [M H] ions respectively, of TNT with respect to increasing field strength for both room temperature (dashed line, reproduced from Figures 4 6 and 4 1 1 ) and elevated temperature. Loo king at the se comparison s the elevated temperature has a significant effec t on the ions behavior. At ~85 C, there is no longer an enhancement of the A type ion behavior, but t here is now what could be considered an enhancement of the B type behavior.

PAGE 131

131 F igures 4 2 4 A and B show the FWHM of the CV peaks with increasing field strength for the M and [M H] ion s of TNT. The FWHM values for the M ion were comparable to those measured at room temperature. The values for the [M H] ion were only slightly high er t han those at room temperature. The observations on peak width are not necessarily a real representation of behavior inside the cell. It was observed that the CV peaks at 85 C showed a fair amount of tailing towards higher CV values, Figure 4 2 5 Th i s tailing broadens the peaks but is not readily apparent when measuring peak width at half height (FWHM) Currently, it is unclear if the tailing of the peaks towards higher CV values is due to a real phenomenon occurring in the FAIMS cell or is due to the method of heating the FAIMS cell. By heating the FAIMS cell with the curtain/drift gas, the entrance of the cell is slightly hotter than the exit end of the cell. The actual temperature differential over the length of the cell is not kn own. Measurements of the external temperature of the cell show an approximately 15 to 20 C difference between the ends of the cell. The cooler temperature at the exit end may lead to an increased amount of clustering changing the CV values and broaden in g the peaks. Figures 4 2 6 A and B show the normalized intensity (normalized to the intensity at 61 Td ) for the M and [M H] ions at 85 C with respect to increasing field strength. The intensity change for the M ion with respect to increasing field streng th was significantly less than was observed at room temperature (Figure 4 8B) The intensity change for the [M H] ion was larger at elevated temperatures than at room temperature (Figure 4 12) The absolute intensities for both ions were both lower than observed at room temperature. The reasons for the changes in intensity are not well understood. As was discussed earlier in the chapter, changes in intensity due to addition of solv ent might be related to decreased fragmentation of the ions.

PAGE 132

132 To illust rate that increased temperature in conjunction with increasing field strength affects other ions produced by other ionization methods, o phthalic acid was analyzed at both room temperature and ~85 C in the cell. Figure 4 2 7 shows the K h /K for the [M H] ion of o phthalic acid with respect to increasing field strength at both room temperature (dashed lines) and at the elevated temperature with either ethanol or isopropanol added to the drift gas As with the M ion of TNT, with the addition of heat, the [M H] ion of o phthalic acid went from an enhancement of A type behavior to a B type behavior. Temperature Summary The premise for increasing the heat inside the FAIM cell was to see if above some set temperature, the behavior of the ion s in the FAIMS c ell would be similar to the ir behavior in dried nitrogen. This idea arose out of the hypothesis that as ion s travel through the FAIMS cell they repeatedly undergo a cluster/decluster process with solvent or neutrals present in the cell W ith the additio n of heat, the probability of ions interacting with the ne utrals present would be significantly reduced The temperature ramp studies illustrated that there were decreases in CV values with solvent added as the temperature increased inside the cell. T he field effects studies showed that with the addition of heat, the ions now had a definite B type behavior under all conditions studied. From this work it is apparent the addition of heat to the cell modifies the behavior and interactions of the ions wi th solvent or neutral molecules present inside the FAIMS cell. The behaviors of the ions in dried nitrogen showed similarities to the behaviors with solvent vapors added (i.e. decrease in CV with ramp, enhancement of B type behavior). Although the effe cts were less than with solvent present, i t was evident that there may well be

PAGE 133

133 i on interactions with dried nitrogen This may be due to a n incomplete dryness of the nitrogen or may be due to interactions of the ions with the nitrogen molecules. One thing that was not as clear as was observed at room temperature was the trend in results corresponding to the trend in solvent size. The CV data for the field effect studies did show a trend corresponding to solvent size for the M ion of TNT. The CV data for the [M H] ion of TNT did also have similar pattern of CV values as was observed at room. For the FHWM values and intensities, there w ere, however no definable trend. As mentioned above, the heating of the cell by means of hot curtain/drift gas does create some temperature differential across the length of the cell. It is thought that this might contribute somewhat to FWHM values. Conclusions The work presented in this chapter has offered some insight into the effect of a ddition of solvent vapor to the planar FAIMS drift gas. To aid support for the hypothesis that ions and solvents undergo a clustering/declustering process in FAIMS, a series of solvents of varying size were investigated. The results for the three ions st udies in the solvent trend investigations all showed a direct relationship between the size of the solvent used and the amount of CV shift and FWHM changes. For two of the ions, the M ion of TNT and the [M H] ion of o phthalic acid, the addition of solv ent vapor to the carrier gas increased the ions A type behavior. The [M H] ion of TNT showed B type behavior under all of the solvent conditions. It was i nteresting to note that two ions of the same molecule exhibited quite different behaviors. These differences are due to different interaction potentials the ions have with neutral solvent molecules present inside the FAIMS cell. Initial work focusing on efforts to characterize the thermodynamic properties of ion behavior inside the cell focused on inc reasing the temperature inside the FAIMS cell. The results of this work showed as temperature increased, the CV values for all of the solvent

PAGE 134

134 conditions investigate decreased. These decreases are indicative of a lessening of the interaction between the i ons and the neutrals present. Simply put as the temperature increased, the size of the clusters being formed decrease s, thereby lowering the difference between the higher field mobility and lower field mobility. Studies of increasing field at high tempe rature showed that th e ions that had an enhancement of A type behavior at room temperature, now had B type behavior. The point at which the ions begin to transition to the B type behavior may indicate where the amount of heat ( imparted upon the ion by bot h heating the cell and heat added through kinetic energy ) is enough to begin to shift the equilibrium for the cluster/decluster process (Equation 4 3).

PAGE 135

135 Figure 4 1. Structures of six solvents used in studies.

PAGE 136

136 Table 4 1. Physical properties of solvents studied Solvent Molecular Weight (g/mol) Surface Area ( 2 ) Volume ( 3 ) Dipole (Debyes) Relative Polarity Melting Point ( C) Boiling Point ( C) Vapor Pressure (torr) @ 25 C 1 Butanol 74.1 250 347 1.49 0.51 89.5 117.6 8.4 tert Butanol 74.1 243 336 1.59 0.39 25.5 82.2 54.7 Isopropanol 60.1 219 290 1.57 0.55 88.5 82.4 58.9 Ethanol 46.1 190 239 1.53 0.65 114.1 78.5 78.7 Methanol 32.0 156 182 1.61 0.76 98.0 64.6 170.7 Water 18.0 115 117 1.91 1.00 0.0 100.0 17.5 *Relative to water Table 4 2. Maximum Solvent Concentrations Solvent Maximum Concentration (ppm w/v) 1 Butanol ~19,000 tert Butanol ~160,000 Isopropanol ~91,000 Ethanol ~99,000 Methanol ~123,000 Water ~17,000

PAGE 137

137 Figure 4 2 A) Compensation voltage (CV) for the M ion of TNT with respect to increasing concentration of solvent vapor added to the drift gas: water (black trace ( ) ), methanol (blue trace ( ) ), ethanol (brown trace ( ) ), isopropanol (purple trace ( ) ), t butanol (orange trace ( ) ), and 1 butanol (green trace ( ) ). B) FWHM for the M ion of TNT with respect to increasing concentration of solvent vapor added to the drift gas. A B

PAGE 138

138 Figure 4 3. Correlation of maximum CV s hift (at 70 Td ( 3500 DV)) with molecular volume of solvent added to the drift gas.

PAGE 139

139 Figure 4 4 CV for M ion of TNT versus percentage of carbon dioxide in the nitrogen drift gas.

PAGE 140

140 Figure 4 5 A) CV for the M ion of TNT with respect to increasing field strength in different drift gas compositions: dried nitrogen (black trace ( ) ), ~7,000 ppm water (blue trace ( ) ), ~15,000 ppm methanol (red trace ( ) ), ~20,000 ppm ethanol (brown trace ( ) ), ~18,000 ppm isopropanol (purple trace ( ) ), ~20,000 ppm t butanol (orange trace ( ) ), and ~9,500 pppm 1 butanol (green trace ( ) ). B) FWHM for the M ion of TNT with repect to increasing field strength in different drift gas compositions. A B

PAGE 141

141 Figure 4 6 Calcu lated K h /K values for the M ion of TNT with respect to increasing field strength in different drift gas compositions: dried nitrogen (black trace, ( )), ~7,000 ppm water (blue trace, ( )), ~15,000 ppm methanol (red trace, ( )), ~20,000 ppm ethanol (brown trace, ( )), ~18,000 ppm isopropanol (purple trace, ( )), ~20,000 ppm t butanol (orange trace, ( )), and ~9,500 pppm 1 butanol (green trace, ( )) Inset depicts trace zoomed in area of trace for dried nitrogen (black trace ( )).

PAGE 142

142 Table 4 3. Calculated and K h /K terms for the M ion of TNT Solvent Concentration (ppm) (Td 2 ) (Td 4 ) K h /K at 97 Td 1 Butanol 9,500 4.981 x 10 5 5.548 x 10 10 1.39 tert Butanol 20,000 4.439 x 10 5 1.085 x 10 9 1.32 Isopropanol 1 4 ,000 4.170 x 10 5 1.192 x 10 9 1.27 Ethanol 20,000 3.510 x 10 5 7.905 x 10 10 1.24 Methanol 15,000 3.072 x 10 5 1.297 x 10 9 1.17 Water 7,000 1.613 x 10 5 1.110 x 10 9 1.05 Dried Nitrogen N/A 3.248 x 10 6 1.692 x 10 9 1.02

PAGE 143

143 Figure 4 7 Calculated resolving powers for the M ion of TNT with respect to increasing field strength in different drift gas compositions: dried nitrogen (black trace, ( )), ~7,000 ppm water (blue trace, ( )), ~15,000 ppm methanol (red trace, ( )), ~20,000 ppm ethanol (brown trace, ( ) ), ~18,000 ppm isopropanol (purple trace, ( )), ~20,000 ppm t butanol (orange trace, ( )), and ~9,500 pppm 1 butanol (green trace, ( )).

PAGE 144

144 Figure 4 8 A) Absolute CV peak intensity for the M ion of TNT with respect to increasing field strength in different drift gas compositions: dried nitrogen (black trace, ( )), ~7,000 ppm water (blue trace, ( )), ~15,000 ppm methanol (red trace, ( )), ~20,000 ppm ethanol (brown trace, ( )), ~18,000 ppm is opropanol (purple trace, ( )), ~20,000 ppm t butanol (orange trace, ( )), and ~9,500 pppm 1 butanol (green trace, ( )). The dotted line depicts the signal for the M ion without the FAIMS cell present. B) Normalized CV peak intensity for the M ion of TNT with respect to increasing field strength in different drift gas compositions. Intensity is normalized to intensity at 50 Td. A B

PAGE 145

145 Figure 4 9 A) Compensation voltage (CV) for the [M H] ion of TNT with respect to increasing concentration of solvent vapor added to the carrier gas: water (black trace, ( ) ), methanol (blue trace, ( ) ), ethanol (brown trace, ( ) ), isopropanol (purple trace, ( ) ), t butanol (orange trace, ( )), and 1 butanol (green trace, ( )). The CV for the M ion of TNT with water added is also shown for comparison (black dashed trace ( ) ). B) FWHM for the [M H] ion of TNT with respect to increasing concentration of solvent vapor added to the carrier gas. A B

PAGE 146

146 Figure 4 10 A) CV for the [M H] ion of TNT with respect to increasing field strength in different drift gas compositions: dried nitrogen (black trace, ( )), ~7,000 ppm water (blue trace, ( )), ~15,000 ppm methanol (red trace, ( )), ~20,000 ppm ethanol (brown trace, ( )), ~18,000 ppm isopropanol (purple trace, ( )), ~20,000 ppm t butanol (orange trace, ( )), and ~9,500 pppm 1 butanol (green trace, ( )). B) FWHM for the [M H] ion of TNT with repect to increasing field strength in different drift gas compositions A B

PAGE 147

147 Figure 4 1 1 Calculated K h /K values for the [M H] ion of TNT with respect to increasing field strength in different drift gas compositions: dried ni trogen (black trace, ( )), ~7,000 ppm water (blue trace, ( )), ~15,000 ppm methanol (red trace, ( )), ~20,000 ppm ethanol (brown trace, ( )), ~18,000 ppm isopropanol (purple trace, ( )), ~20,000 ppm t butanol (orange trace, ( )), and ~9,500 pppm 1 butanol (green trace, ( )).

PAGE 148

148 Figure 4 1 2 Normalized CV peak intensity for the [ M H] ion of TNT with respect to increasing field strength in different drift gas compositions: dried nitrogen (black, ), ~7,000 ppm water (blue, ), ~15,000 ppm methanol (red, ), ~20,000 ppm ethanol (brown, ), ~18,000 ppm isopropanol (purple, ), ~20,000 ppm t butanol (orange, ), and ~9,500 pppm 1 butanol (green, ). Intensity is normalized to intensity at 50 Td.

PAGE 149

149 Figure 4 1 3 CV spectrum of the [M H] ion of o phthalic acid acquired at 95 Td in either dried nitrogen (left side of figure) or with ~14,000 ppm isopropanol added to the drift gas (right side of figure).

PAGE 150

150 Figure 4 1 4 CV for the [M H] ion of o phthalic acid with respect to increasing fi eld strength in different drift gas compositions: dried nitrogen (black trace ( ) ), ~7,000 ppm water (blue trace ( ) ), ~15,000 ppm methanol (red trace ( ) ), ethanol (brown trace ( ) ), and ~14,000 ppm isopropanol (purple trace ( ) ).

PAGE 151

151 Figure 4 1 5 C alculated K h /K values for the [M H] ion of o phthalic acid with respect to increasing field strength in different drift gas compositions: dried nitrogen (black trace, ( )), ~7,000 ppm water (blue trace, ( )), ~15,000 ppm methanol (red trace, ( )), ethanol (brown trace, ( )), and ~14,000 ppm isopropanol (purple trace, ( )).

PAGE 152

152 Figure 4 1 6 A) FWHM for the [M H] ion of o phthalic acid with respect to increasing field strength in different drift gas compositions: dried nitrogen (black trace, ( )), ~7,000 ppm water (blue trace, ( )), ~15,000 ppm methanol (red trace, ( )), ethanol (brown trace, ( )), and ~14,000 ppm isopropanol (purple trace, ( )). B) Normalized CV peak intensity for the [M H] ion of o phthalic acid with respect to increasing field strength in different drift gas compositions. A B

PAGE 153

153 Figure 4 1 7 A) CV spectrum of mixture of three positional isomers of phthalic acid acquired at 70 Td in dried nitrogen. B) CV spectrum of mixture of three position isomers of phthalic acid acquired at 70 Td with ~14,000 ppm isopropanol added to drift gas. A B o and p m m p o

PAGE 154

154 Figure 4 1 8 CV values for the [M H] ions of o phthalic acid (solid traces ) and m phthalic acid (dashed traces ) with respect to increasing temperature of the gas at the curtain plate with different drift gas compositions: dried nitrogen (black traces ( ) ), ~7,000 ppm water (blue traces ( ) ), and ~15,000 ppm methanol (red traces ( ) ).

PAGE 155

155 Figure 4 1 9 FWHM fo r the [M H] ions of o phthalic acid (solid traces ) and m phthalic acid (dashed traces ) with respect to increasing temperature of the gas at the curtain plate with different drift gas compositions dried nitrogen (black traces, ( )), ~7,000 ppm water (blue traces, ( )), and ~15,000 ppm methanol (red traces, ( )).

PAGE 156

156 Figure 4 20 Comparison of absolute CV peak intensity for the [M H] ion of o phthalic acid with differing heated drift gas compositions: dried nitrogen (gray box 90 C), ~7,000 ppm water (blue box 90 C), and ~15,000 ppm methanol (red box 50 C). Values represent the maximum measured intensities throughout the temperature ramp studies.

PAGE 157

157 Figure 4 2 1 A) CV for the M ion of TNT with respect to increasing field strength in different drift gas compositions and at 85 C inside the FAIMS cell: dried nitrogen (black trace ( ) ), ~7,000 ppm water (blue trace ( ) ), ~15,000 ppm methanol (red trace ( ) ), ~20,000 ppm ethanol (brown trace ( ) ), ~18,000 ppm isopropanol (purple trace ( ) ), and ~9,500 pppm 1 butanol (green trace ( ) ). B) Calculated K h /K values for the M ion of TNT with respect to increasing field strength in different drift gas compositions and at 85 C inside the FAIMS cell. A B

PAGE 158

158 Figure 4 2 2 A) CV for the [M H] ion of TNT with respect to increasing field strength in different drift gas compositions and at 85 C inside the FAIMS cell: dried nitrogen (black trace, ( )), ~7,000 ppm water (blue trace, ( )), ~15,000 ppm methanol (red trace, ( )), ~20,000 ppm ethanol (brown trace, ( )), ~18,000 ppm isopropanol (purple trace, ( )), and ~9,500 pppm 1 butanol (green trace, ( )). B) Calculated K h /K values for the [M H] ion of TNT with respect to increasing field strength in different drift gas compositions and at 85 C inside the FAIMS cell A B

PAGE 159

159 Figure 4 2 3 A) Comparison of K h /K values for the M ion of TNT at room temperature (dashed lines) to K h /K values at 85 C inside the FAIMS cell (solid lines) in different drift gas compositions: dried nitrogen (black trace, ( )), ~7,000 ppm water (blue trace, ( )), ~15,000 ppm methanol (red trace, ( )), ~20,000 ppm ethanol (brown trace, ( )), ~18,000 ppm isopropanol (purple trace, ( )), and ~9,500 pppm 1 buta nol (green trace, ( )). B) Comparison of K h /K values for the [M H] ion of TNT at room temperature (dashed lines) to K h /K values at 85 C inside the FAIMS cell in different drift gas compositions. A

PAGE 160

160 Figure 4 2 3 Continued. B

PAGE 161

161 Figure 4 2 4 A) FWHM for the M ion of TNT with respect to increasing field strength in different drift gas compositions and at 85 C inside the FAIMS cell: dried nitrogen (black trace, ( )), ~7,000 ppm water (blue trace, ( )), ~15,000 ppm methanol (red trace, ( )), ~20,000 ppm ethanol (brown trace, ( )), ~18,000 ppm isopropanol (purple trace, ( )), and ~9,500 pppm 1 butanol (green trace, ( )). B) FWHM for the [M H] ion of TNT with respect to increasing field strength in different drift gas compositions and at 85 C inside the FAIMS cell. A B

PAGE 162

162 Figure 4 2 5 Two CV spectra for the M ion of TNT showing tailing of CV peaks during elevated temperature experiments Both s pectra were acquired with the FAIMS cell heated to 85 C (internal temperature) and with ~9,500 ppm 1 butanol added to the drift gas. The CV spectrum on the left was acquired at ~61 Td ( 2500 DV) and the spectrum on the r ight was acquired at ~115 Td (~4600 DV)

PAGE 163

163 Figure 4 2 6 A) Normalized CV peak intensity for the M ion of TNT with respect to increasing field strength in different drift gas compositions at 85 C : dried nitrogen (black trace, ( )), ~7,000 ppm water (blue trace, ( )), ~15,000 ppm methanol (red trace, ( )), ~20,000 ppm ethanol (brown trace, ( )), ~18,000 ppm isopropanol (purple trace, ( )), and ~9,500 pppm 1 butanol (green trace, ( )). Intensity is normalized to intensity at 61 Td. B) Normalized CV peak intensity for t he [ M H] ion of TNT with respect to increasing field strength in different drift gas compositions. Intensity is normalized to intensi ty at 61 Td. A B

PAGE 164

164 Figure 4 2 7 Comparison of K h /K values for the [M H] ion of o phthalic acid at room temperature (dashe d traces ) and at 85 C inside the FAIMS cell with different compositions of drift gas: ~20,000 ppm ethanol (black trace ( ) ) and ~14,000 ppm isopropanol (red trace ( ) ).

PAGE 165

165 CHAPTER 5 CONCLUSIONS AND FUTU RE WORK Conclusions This dissertation has presented a series of studies which help to better understand some of the param eters that affect the performance of planar geometry high field asymmetric waveform ion mobility spectrometry, FAIMS. The studies were focused on two main areas which can affect how planar FAIMS cells behave. The first area of focus was on the mode of in jection into planar FAIMS. The second area of focus was on the role of solvent vapor in FAIMS behavior. The studies presented provided quite interesting and often unexpected results, some of which are not yet fully understood. Nonetheless, the observati ons made have helped to develop a better understanding of factors that can change how planar geometry FAIMS performs. The observations also have helped to develop some fundamental aspects of how ions behave inside the FAIMS cell as they travel through it. Modes of Injection Summary The work presented focusing on modes of injection investigated how ions are injected into the FAIMS cell and how they travel through the cell to the detector. The primary analyte ions studied were the [M H] ion of o phthalic a cid and the M ion of trinitrotoluene (TNT). The first aspect explored took advantage of the line of sight from ionization source to detector that planar geometry FAIMS cells offer (in contrast to the cylindrical geometries which do not provide a straight path from ionization to detector ) With this line of sight, planar geometry FAIMS cells can operate in a t otal i on m ode (TIM). Studies of improvements in overall signal of the TIM were presented focusing on the voltages applied to the FAIMS cell plates. V oltages explored were either direct current (DC), radio frequency (RF), or a combination of both. The results of these studies showed that the highest signal s w ere achieved when the DC voltage applied to the

PAGE 166

166 plates was within 1 to 2 V of the vo ltage applied to the heated capillary. The addition of RF voltages only served to lower the signal. The next area studied explored curtain plate effects in orthogonal and parallel injection. In planar geometry FAIMS cells, ions can be injected into the cell either orthogona l l y or parallel with respect to the direction of separation. For both types of injection, the use of curtain plates is necessary to both provide a source of dry (or known composition) drift gas and a curtain gas The curtain plate ca n also serve as a counter electrode for the ionization source in the case of APCI and ESI. A series of studies focused on the effect of the gap between the exit of the curtain plate and the entrance into the cell (through the counter electrode in orthogon al injection and curtain plate B for parallel injection). The results of this work showed that in orthogonal injection, the largest practicable gap provide d better signal with more reproducibility. In parallel injection, the gaps studied showed no discer nable difference in signal but once again the larger gap did appear to offer more reproducibility. Throughout these studies, a couple of interesting observations were made. It appeared that parallel injection provide more absolute signal intensity but provided lower resolving powers (R p ). The lower R p values in parallel inj ection were due to broader full width at half maximum (FWHM) peak widths of the compensation voltage (CV) peaks. This broader peak width in parallel injection compared to orthogona l injection (when all things else are held equal) is not necessarily what theory predicts. It is commonly held that the longer an ion s residence time inside the planar FAIMS cell, the higher resolving power the cell should provide [48] Longer residence times should also decrease signal intensities due to increased amounts of diffusional losses. As ions are propelled through the FAIMS cell by gas flow, if the gas flow an d cross sectional area of the cell are the same, the only other thing that would affect residence time in the

PAGE 167

167 cell is the path length. Based on this, it would be predicted that parallel injected ions sh ould have a longer residence time on this cell since the path length is 15 mm longer than for orthogonal injection. A series of studies to measure the actual ion residence time inside the cell were conducted to determine what if any were the differences in residence time between the two methods of inj ection. Residence times were measured by switching the voltage polarity of one FAIMS cell plate while maintaining the other voltage constant and then varying the rate a t which this switching occurred. This process effectively creat es a gate to turn ion c urrent on and off. Several residence time s inside the cell were investigated by varying the gas flow at the back of the cell. The results of this work showed that at all flow rates, parallel injected ions had a lower residence time than orthogonally inj ected ions ( i.e. 70 ms vs. 103 ms for normal residence time). P arallel injected ions are believed to have a shorter residence time due to how the drift gas enters into the cell through the curtain plates. In parallel injection the gas flow into the cel l goes through a 1.5 mm hole in curtain plate B aimed directly through the center of the cell. The g as flow through this small hole accelerates ions into the cell, effectively shortening the path length for parallel injection. Orthogonally injected ions are also accelerated into the cell but the 90 turn towards the exit of the cell negates th is acceleration. The residence time studies also showed that there was a tradeoff between improvements in resolving power and loss of signal. For both orthogonal and parallel injection, increasing the residence time inside the cell did improve the resolving power of the cell but there was a significant decrease in signal. Th is rate of the signal decrease was greater than the rate at which resolving power increas ed. The inverse was true for shorter residence times. Another aspect investigated was a method to control losses of ions due to lateral diffusion inside the cell (orthogonal to both the direction of separation and direction of applied field). The

PAGE 168

168 studies concentrate d on the use of a pneumatic focusing gas applied through porous spacers placed at the lateral edges of the planar FAIMS cell. The results of these studies showed up to a 2 fold improvement in signal with no other discernable loss in performanc e, i.e. no change in FWHM or CV. The final area investigated in the modes of injection examined how the presence of the curtain plate affected the performance of the planar FAIMS cell. In these studies, data were compared with the curtain plates in plac e and without the curtain plates in place ( curtain plate A for parallel injection). The lack of the curtain plate also meant that the composition of the drift gas inside the cell was no longer controlled and therefore likely contained solvent vapor from t he ionization source. The results for this work showed quite different results for orthogonal and parallel injection. For orthogonal injection, the lack of the curtain plate did not affect the TIM signal but the CV peak signal was significantly lower. This CV peak signal loss may be attributed to poor performance from the ion source with the presence of the high field asymmetric waveform. For parallel injection, the lack of the curtain plate has the opposite effect. Signal s for both TIM and the CV pea ks were significantly higher. It appeared that for parallel injection, the presence of the curtain plate had a significant effect on the signal and for orthogonal the curtain plate was not necessarily the biggest detriment A commonality between both mo des of injection was a significant shift in CV value without the curtain plate present (and therefore with uncontrolled drift gas composition). The shifts were believed to be due to interactions between the ions and solvent vapors that were being allowed inside the cell. These CV shift observations led to the second main area of focus in this dissertation. Solvent Vapor Effects Summary Without the curtain plate present, one is unable to effectively control the dryness and composition of the drift gas in the cell. Under these uncontrolled conditions, dramatic shifts in

PAGE 169

169 CV values were observed. Based on the se observations, studies were conducted to determine how the prese nce of solvent vapor affects the performance of the planar FAIMS cell. These studies focused on controlling the amount of solvent vapor present in the drift gas and measuring the effects on overall performance. A solvent vapor generator was set up to gen erate solvent saturated nitrogen which could then be diluted to the desired concentration with additional dried nitrogen. The solvent vapor studies were initially focused on an exploration of the effects of controlled amounts of solve nt vapor added to the drift gas. Based on observations made in the initial studies, a more th o rough exploration was undertaken of the solvent trends and the effects of temperature on solvent vapor effects. The initial studies looked at how the concentrat ions of water and methanol vapors in the drift gas affected the performance of the planar FAIMS cell. All of these studies were conducted with the field fixed at ~70 Td ( 3500 DV) to eliminate the effects of varying field. For these studies, ions formed from the three positional isomers of phthalic acid as well as a series of four explosives were analyzed. The result of these studies showed that as the concentration of solvent vapor added to the cell increased, there were dramatic shifts to higher CV val ues when compared to dried nitrogen values. As the concentration increased, the FWHM initially increased, before decreasing to approximately dried nitrogen values as concentration continued to increase. This decrease in FWHM to approximate dried nitrogen values with large increases in CV values meant that the resolving power of the cell increased dramatically, from a maximum of ~30 in dried nitrogen to a maximum of ~140 with methanol vapor added to the drift gas. Compensation voltage shifts are believed t o be caused by ion neutral interactions occurring as the ions transverse the length of the cell towards the detector. What is though t to be

PAGE 170

170 happening is that the asymmetric field used in FAIMS caus es the ions to experience two different effective ion temp enough to limit or preclude interactions with the neutral solvent molecules present in the drift temperatures decrease to where ion neutral interactions occur. The interactions are believed to be of a clustering nature. This clustering action creates a larger ion size which depresses the the difference field and low field mobilities, if the lower field mobility is depressed, the difference between the two mobilites increases requiring a larger CV to maintain a stable trajectory for that ion through the FAIMS cell. Anot her obs er vation made in these initial studies was that all of the analyte ions investigated showed differing amounts of CV shift with the addition of solvent vapor Based on these observations mixtures of the analytes were analyzed in dried nitrogen and t hen with different amounts of solvent vapor added to the drift gas. The results for the phthalic acids showed that in dried nitrogen the CV peak for one of the isomers, m isomer, was separated from the other two isomers whereas the other isomers o and p wer e not separated With the addition of ~7,000 ppm water or ~15,000 ppm methanol vapor to the drift gas, the CV peaks for all three isomers were well separated. The results for the explosives showed that in dried nitrogen, the CV peaks for th e M ions of TNT, 3,4 DNT, and 2,4 DNT were well separated from one another. The CV peaks for the [M H] ion of 2,4 DNT and the M ion of 2,6 DNT were lying on one another and could not be distinguished without mass spectrometric help. With the addition of ~7,000 ppm water, the CV peaks for all of the ions were now well separated.

PAGE 171

171 To determine some of the more fundamental aspects of the effects of solvent vapor the effect of solvent size and temperature inside the cell were investigated. To support the premise of solvent molecules clustering with the ions inside the cell and creating the dramatic shifts in CV value, a study of the trend of solvent size was done. The hypothesis was that if clustering of solvent with the ions during the lower field porti on of the asymmetric waveform was creating a depression in mobility, larger solvents should create a greater effect. Six solvents were chosen based on increasing size relative to one another and similarity in functional group s in order to maintain similar type s of ion neutral interactions. These studies focused on the behavior of the M and [M H] ions of TNT and the [M H] ion of o phthalic acid. The results of these studies showed a definite trend relating to solvent size for the M ion of TNT and [M H] ion of o phthalic acid With increase s in solvent vapor concentration and applied field, both of these ions exhibited larger CV shifts, increases in FWHM and increases in signal as the size of the solvent increased. The [M H] ion of TNT did not exhi bit a definite trend related to solvent size and in gen eral behaved quite differently than the M ion of TNT. This difference in behavior is likely due to differences in the mechanisms of interaction with solvent vapor between the two ions. In general l arger solvents did seem to generate large r solvent vapor effects when compared to dried nitrogen alone supporting the idea of ions clustering with the solvent and depress their mobility. A couple of interesting observations were made in these studies. The addition of solvent vapor to the cell changed the ABC type ion behavior of the ions. It appeared that for ions exhibiting large CV shifts, the solvent vapor was enhancing the A type behavior of the ion, and that enhancement increased with increasing s olvent size. The second interesting, and completely unexpected, observation was that the addition of solvent vapor increased the signal for the ions

PAGE 172

172 of interest up to ~20x higher than in dried nitrogen. Indeed, s ignals with solvent vapor present were com parable to signals without the FAIMS cell present. The idea of the clustering/declustering interaction inside the FAIMS cell is based in part on the changes in temperature ions experience as they are oscillated by the asymmetric waveform. Equation 5 1 sho ws a theoretical equilibrium between an ion and an ion clustered with n solvent ( 5 1 ) molecules. The K f term is the thermodynamic equilibrium constant for the formation of the ion neutral clusters. A series of studies were performed focus ing on changing this thermodynamic equilibrium. To induc e changes in this clustering equilibrium, the FAIMS cell was heated by means of heating the curtain/drift gas. The hypothesis was that above some temperature, ions would no longer be able to interact with the solvent vapor (i.e. the equilibrium in Equati on 5 1 shift ed to the left). The studies focused on looking at how increasing the temperature inside the cell changed the performance and how increasing field strength at elevated temperatures affected performance. The results of this work did show decr eases in CV values as the temperature inside the cell increased. The values did not approach the values in dried nitrogen and in fact, the values in dried nitrogen decreased as well. There were also increases in the FWHM for the CV peaks of the ions inv estigated when solvent vapor was present. The decrease in CV values and the increase in FWHM indicated that as the temperature inside the cell increased, the extent of interactions between the ions and solvent vapors were decreasing i.e. the equilibrium in Equation 5 1 moving to the left With respect to increasing field strength at elevated temperatures, there was no longer an enhancement of A type behavior. All of the ions studied showed B type behavior and in one case, an ion exhibited C type behavior. The combination of

PAGE 173

173 heat and increasi ng field strength offers insight into the thermodynamic properties of the ion neutral solvent molecule interactions. Wrap Up The results presented in this dissertation show some of the parameters that can affect the performance characteristics of planar geometry FAIMS cells. The investigations into modes of injection have shown that operating the FAIMS cell in a total ion mode requires that the DC voltage applied to th e cell need be within 1 2 V of the heated capillary. S tudies of the differences in orthogonal and parallel injection have shown that p arallel injection offer s more sensitivity but lower resolving powers compared to orthogonal injection. Ion residence tim e measurements support these observations as well as show the trade offs between improvements in resolving powers and loss of signal. Studies of controlling lateral diffusion losses by means of pneumatic ion focusing demonstrate up to a two fold increase in signal without loss of resolving power or change in CV. The modes of injection studies demonstrated that taking away dry drift gas from the cell dramatically increased CV values. This led to detailed studies of the effects of controlled amounts of s olvent vapor inside the planar FAIMS cell. The results of the studies showed that solvent vapor inside the cell dramatically increased the CV values of ions while maintaining relatively narrow peaks, drastically increasing both resolving power and resolu tion for mixtures. Further studies of solvent size effects showed that increases in solvent size generally increased the observed effects. This supported the theory that ions were clustering with the solvent vapor. Studies of temperature effects with th e presence of solvent vapor demonstrated changes in behavior that offer insight into the thermodynamic properties of the ion chemistry taking place inside the FAIMS cell.

PAGE 174

174 Future Work As with all aspects of research, the results obtained from experimentatio n often times lead to other areas of research and sometimes generate more questions than were answered. The results presented in this dissertation provide some valuable insights into what affects the performance of planar geometry FAIMS cells. Some of th e observations made in this work have lead to other paths of experimentation to explain unexpected results. For instance, the discrepancy between parallel and o rthogonal injection (with orthogonal injection providing higher resolving powers when theory sa ys parallel should ) led to the measurements of the ion residence times. The dramatic shift in CV when the curtain plates were removed led to detailed studies of the effects of solvent vapor in the planar FAIMS cell. There are however, still some questi ons that still remain unanswered. One aspect of the FAIMS technique that needs to be explored in greater detail is the gas flow through the cell. As the ions in FAIMS are propelled through the FAIMS cell by gas flow, being able to understand how and whe re the gas is moving is of importance. One method for determining this is by using gas flow vector modeling. This type of modeling generates a series of vectors which show the direction and speed of the gas inside a system. Being able to see how gas is moving in the FAIMS cells w ould lead to improvements in cell design to generate optimal flows through the cell. Flow vector modeling would also be beneficial in supporting the hypothesis concerning the acceleration of ions into the cell as the drift gas f lows through the small hole in either the counter electrode plate (orthogonal injection) or curtain plate B (parallel injection). The studies of solvent vapor effects provided some quite dramatic results concerning improvements in resolving power, signal, and resolution. Some of the fundamental ion chemistry involving interactions between the ions in the cell and the solvent vapor neutrals

PAGE 175

175 present was investigated Even though the results of this work provide a good initial understanding of what is happen ing inside the cell, the re are still some aspects that sh ould be investigated in more depth. One of these aspects is the thermodynamic characteristics of the ion neutral solvent molecule interactions. It is currently believed that ions are undergoing a c luster/decluster process as they are moved back and forth between the two applied fields in the FAIMS cell. The cluster/decluster process is also thought to be dependent upon the heating and cooling of the ions as they oscillate. The temperature studies presented in this dissertation were focused on trying to determine some of those thermodynamic aspects. While the results of these studies did show interesting results and hinted at some of those aspects, a true empirical thermodynamic understanding could not be reached. This was most likely due to the temperature limitations of the current cell design. Refinement of the cell design to incorporate higher and more controlled temperatures might help to further the thermodynamic understanding. Another aspec t of the ion neutral solvent molecules that is not well understood at this point are the kinetic characteristics of the interactions. It is hypothesized that the ion solvent clusters are being formed and destroyed on the time scale of the asymmetric waveform used to generate the high fields. If this is the case, varying the frequency of the waveform would have a large effect on the behavior of the ions. As the frequency is increased, there may likely be a point a t which the switching of fields inside the cell would be faster than ions and solvents could cluster or decluster This observation could aid in determining a kinetic rate constant for the formation and destruction of the ion neutral solvent molecule clus ters. Increasing the frequency of the waveform does, however, not come without some drawbacks. The first is that current waveform generators are built and tuned such that the y will only work at one frequency. To change the frequency, one would need to make a completely

PAGE 176

176 new waveform generator. The second drawback is that increasing the frequency of the waveform could be detrimental to the ions themselves. Currently, it is thought that ions do not undergo much collision induced dissociation as they trav el through the cell due to losses of energy to collisions being greater than gains. As the frequency of the waveform increases, the increased rate of the oscillations of the ions could cause the energy of the ions to build to the point at which they fragm ent and are lost, especially so called fragile ions [50] To go along with the ther modynamic and kinetic experimentation, it is useful to develop a model system to help understand how things learned from experimentation might apply to other similar systems. Theoretical modeling can often be beneficial as it can provide preliminary resul ts without the time/effort/expense of setting up experiments that may or may not work. The difficulty with theoretical modeling is often determining what should be modeled. As has been discussed in this dissertation, ions are thought to be clustering wit h the solvent vapor present in the cell. How the solvent molecules and ions are interacting is not known nor is the number of solvent molecules that may interact with one ion. Molecular modeling of ion /solvent clusters may lead to some insights in this area as it may show distinctions between similar ions that correlate to experimental result differences. One aspect of modeling that presents some difficulty is that the processes and interactions occurring inside the FAIMS cell are all happening at or n ear atmospheric pressure. Developing a model for the interactions at atmospheric pressure will likely be difficult as m uch of the published work looking at the formation and destruction of clusters of ions and neutral molecules has been done at reduced p ressures [51, 52] One final aspect that needs investigat ion and improvement is the method of introduction of solvent vapor into the drift gas of the planar FAIMS cell. The work presented in this dissertation used a solvent vapor generator comprised of a sparge system inside a sealed bottle. While this

PAGE 177

177 system did provide relatively reproducible results, as a whole, the system is quite bulky and the actual concentration of the so lvent vapor present inside the gas is not known accurately (~ 500 ppm). In order for the addition of solvent vapor to the drift gas in FAIMS to be more user friendly, a simple and more compact method of vapor generation would be preferable. Being able t o more accurately control the concentration of solvent vapor in the FAIMS drift gas would be beneficial for both application development and for fundamental studies. For the fundamental studies, previous work has shown that at reduced pressures (~20 torr) clustering i nteractions exhibit second order kinetics [52] If this observation applies at pressures at or near atmospheric pressure, it is imp ortant to know very exactly what the concentration of the solvent vapor is in order to develop a kinetic understanding of the interactions. Epilogue From the studies presented in this dissertation, it is clear that there is a variety of factors that can affect the performance of planar geometry FAIMS cells The studies of modes of injection illustrated several factors that can influence performance including the voltage applied to the plates of the FAIMS cell, the method wi th which ions are injected into the cell, and the residence time inside the cell. Manipulation of these factors will allow a user to select the best performance characteristics needed for the application of interest. The studies of the effects of solvent vapor demonstrated that when the concentration is controlled, there were often dramatic improvements in performance of the planar FAIMS cell. These improvements from solvent vapor addition increase the separation ability of FAIMS and may lead to a wide range of applications. It is the hope that future work will provide support for the proposed theories and possibly lead to their refinement and application.

PAGE 178

178 APPENDIX FIELD STRENGTH VERSU S TEMPERATURE Table A 1. Field strengths at set dispersion v oltages with differing temperature. Dispersion Voltage (Volts) Td (at 25 C) Td (at 85 C) 2500 50.76 60.97 2600 52.79 63.41 2700 54.82 65.85 2800 56.85 68.29 2900 58.88 70.73 3000 60.91 73.17 3100 62.94 75.61 3200 64.97 78.04 3300 67.00 80.49 3400 69.03 82.92 3500 71.03 85.36 3600 73.09 87.80 3700 75.12 90.24 3800 77.15 92.68 3900 79.18 95.12 4000 81.21 97.56 4100 83.25 100.00 4200 85.28 102.44 4300 87.31 104.88 4400 89.34 107.31 4500 91.37 109.75 4600 93.40 112.19 4700 95.43 114.63 4800 97.45 117.07 4900 99.49 119.51 5000 101.52 121.95

PAGE 179

179 LIST OF REFERENCES [1] G.A. Eiceman, Z. Karpas, Ion Mobility Spectrometry, CRC Press, Boca Raton, Florida, 2004. [2] R.G. Ewing, D.A. Atkinson, G.A. Eiceman, G.J. Ewing, A critical review of ion mobility spectrometry for the detection of explosives and explosive related compounds, Ta lanta, 54 (2001) 515 529. [3] R.W. Purves, R. Guevremont, S. Day, C.W. Pipich, M. Matyjaszczyk, Mass spectrometric characterization of a high field asymmetric waveform ion mobility spectrometer, Rev. Sci. Instrum., 69 (1998) 4094 4105. [4] R. Guevremont, H igh field asymmetric waveform ion mobility spectrometry: a new tool for mass spectrometry, J. Chromatogr. A, 1058 (2004) 3 19. [5] J.G. Bryant, M. Prieto, T.A. Prox, R.A. Yost, Design and evaluation of a novel hemispherical FAIMS cell, Int. J. Mass Spectr om., (2010). [6] E.V. Krylov, A method for reducing diffusion losses in a drift spectrometer, Tech. Phys., 44 (1999) 113 116. [7] R. Guevremont, R.W. Purves, Atmospheric pressure ion focusing in a high field asymmetric waveform ion mobility spectrometer, R ev. Sci. Instrum., 70 (1999) 1370 1383. [8] E.V. Krylov, Comparision of planar and coaxial field asymmetric ion mobility spectrometer (FAIMS), Int. J. Mass Spectrom., 225 (2003) 39 51. [9] I.A. Buryakov, E.V. Krylov, E.G. Nazarov, E.K. Rasulev, A new metho d of separation of multi atomic ions by mobility at atmospheric pressure using a high frequency amplitude asymmetric strong electric field, Int. J. Mass Spectrom. Ion Processes, 128 (1993) 143 148. [10] E.A. Mason, E.W. McDaniel, Transport Properties of Io ns in Gases, John Wiley & Sons, New York, New York, 1988. [11] B. Carnahan, A. Tarassov, Ion mobility spectrometer, in, U.S., 1995. [12] B. Carnahan, S. Day, V. Kouznetsov, M. Matyjaszczyk, A. Tarassov, Proceedings of the 41st Annual ISA Division Symposium in, Framingham, MA, 1996, pp. 85 94. [13] R. Guevremont, R.W. Purves, Electrospray ionization high field asymmetric waveform ion mobility spectrometry mass spectrometry, Anal. Chem., 71 (1999) 2346 2357. [14] R.A. Miller, G.A. Eiceman, E.G. Nazarov, A.T. King, A novel micromachined high field asymmetric waveform ion mobility spectrometer, Sens. Actuators, B, 67 (2000) 300 306. [15] M. Dole, L. Mack, R. Hines, R. Mobley, L. Ferguson, M. Alice, Molecular beams of macroions, J. Chem. Phys., 49 (1968) 2240 22 49.

PAGE 180

180 [16] J. Iribarne, B. Thomson, Evaporation of small Ions from charged droplets, J. Chem. Phys., 64 (1976) 2287 2294. [17] A.P. Bruins, Mass spectrometry with ion sources operating at atmospheric pressure, Mass Spectrom. Rev., 10 (1991) 53 77. [18] J.T. Watson, Introduction to Mass Spectrometry, Raven Press, New York, New York, 1985. [19] A.A. Shvartsburg, F. Li, K. Tang, R.D. Smith, High resolution field asymmetric waveform ion mobility spectrometry using new planar analyzers, Anal. Chem., 78 (2006) 3706 3714. [20] E.V. Krylov, Pulses of special shapes formed on a capacitive load, Instrum. Exp. Tech., 40 (1997) 47 50. [21] Finnigan MAT TSQ/SSQ 7000 Series: Atmospheric Pressure Ionization Operator's and Service Manual, San Jose, CA. [22] Finnigan MAT TSQ/ SSQ 7000 Series: Schematics and Reference Manual, San Jose, CA. [23] J.R. Chapman, Practical Organic Mass Spectrometry, John Wiley & Sons, Chichester, England, 1993. [24] S. Rokushika, H. Hatano, M.A. Baim, H.H. Hill, Jr., Resolution measurement for ion mo bility spectrometry, Anal. Chem., 57 (1985) 1902 1907. [25] W.F. Siems, C. Wu, E.E. Tarver, H.H. Hill, Jr., P.R. Larsen, D.G. McMinn, Measuring the resolving power of ion mobility spectrometers, Anal. Chem., (1994) 4195 4201. [26] A.A. Shvartsburg, R.D. Sm ith, Scaling of the resolving power and and sensitivity for planar FAIMS and mobility based discrimination in flow and field driven analyzers, J. Am. Soc. Mass. Spectrom., 18 (2007) 1672 1681. [27] B.M. Kolakowski, M.A. McCooeye, Z. Mester, Compensation v oltage shifting in high field asymmetric waveform ion mobility spectrometry mass spectrometry, Rapid Commun. Mass Spectrom., 20 (2006) 3319 3329. [28] G.A. Eiceman, N. Krylova, E. Krylov, J.A. Stone, Field dependence of mobility for gas phase ions of organ ophosphorus compounds at atmospheric pressure with differential mobility spectrometry and effects of moisture: insights into a model of positive alpha dependence, Int. J. Ion Mobility Spectrom., 6 (2003) 43 47. [29] N. Krylova, E.V. Krylov, G.A. Eiceman, E ffect of moisture on high field dependence of mobility for gas phase ions at atmospheric pressure: organophosphorus compounds, J. Chem. Phys., 19 (2003) 3648 3654. [30] G.A. Eiceman, E. Krylov, N. Krylova, E.G. Nazarov, R.A. Miller, Separation of ions fro m explosives in differential mobility spectrometry by vapor modified drift gas, Anal. Chem., 76 (2004) 4937 4944.

PAGE 181

181 [31] B.B. Schneider, T.R. Covey, S.L. Coy, E.V. Krylov, E.G. Nazarov, Control of chemical effects in the separation process of a differential mobility mass spectrometer system, Eur. J. Mass Spectrom., 16 (2010) 57 71. [32] L.C. Rorrer, III, R. Guevremont, D.A. Barnett, R.A. Yost, The role of clustering on ion behavior in FAIMS, in: 50th ASMS Conference on Mass Spectrometry and Allied Topics, Or lando, FL, May 31 June 4, 2002. [33] D.A. Barnett, R.W. Purves, B. Ells, R. Guevremont, Separation of o m and p phthalic acids by high field asymmetric waveform ion mobility spectrometry, J. Mass Spectrom., 35 (2000) 976 980. [34] L.C. Rorrer, III, M. Prieto, R.A. Yost, Evaluation of linear injection and orthogonal injection into planar FAIMS MS, in: 56th ASMS Conference on Mass Spectrometry and Allied Topics, Denver, CO, June 1 5, 2008. [35] F.W. Karasek, S.H. Kim, Identification of isomeric phthalic acids by mobility and mass spectra, Anal. Chem., 47 (1975) 1166 1168. [36] B.M. Kolakowski, Z. Mester, Review of applications of high field asymmetric waveform ion mobility spectrometry (FAIMS) and differential mobility spectrometry (DMS), Analyst 132 (200 7) 842 864. [37] D. Zhan, J. Rosell, J.B. Fenn, Solvation studies of electrospray ions method and early results, J. Am. Soc. Mass. Spectrom., 9 (1998) 1241 1247. [38] P. Kebarle, M. Arshadi, J. Scarborough, Hydration of negative ions in the gas phase, J. Chem. Phys., 49 (1968) 817 822. [39] S. Chowdhury, E.P. Grimsrud, P. Kebarle, Bonding of charge delocalized anions to protic and dipolar aprotic solvent molecules, J. Phys. Chem., 91 (1987) 2551 2556. [40] L.C. Rorrer III, The role of clustering on ion beh avior in high field asymmetric waveform ion mobility spectrometry, in: Chemistry, University of Florida, Gainesville, FL, 2002, pp. 108. [41] M.A. Blanc, Recherches sur les mobilites des ions dans les gaz, J. Phys., 7 (1908) 825 839. [42] L.B. Loeb, The n ature of gaseous ions from it study of mobilities in mixtures, Phys. Rev., 31 (1928) 1115. [43] L.B. Loeb, Recent light on the nature of gaseous ions, Phys. Rev., 32 (1928) 81 96. [44] R. Guevremont, D.A. Barnett, R.W. Purves, L.A. Viehland, Calculation of ion mobilities from electrospray ionization high field asymmetric waveform ion mobility spectrometry mass spectrometry, J. Chem. Phys., 114 (2001) 10270 10277.

PAGE 182

182 [45] A.A. Shvartsburg, K. Tang, R.D. Smith, Differential ion mobility separations of peptides w ith resolving power exceeding 50, Anal. Chem., 82 (2010) 32 35. [46] A.A. Shvartsburg, W.F. Danielson, R.D. Smith, High resolution differential ion mobility separations using helium rich gases, Anal. Chem., 82 (2010) 2456 2462. [47] K.A. Daum, D.A. Atkinso n, R.G. Ewing, The role of oxygen in the formation of TNT product ions in ion mobility spectrometry, Int. J. Mass Spectrom., 214 (2002) 257 267. [48] A.A. Shvartsburg, K. Tang, R.D. Smith, Modeling the resolution and sensitivity of FAIMS analyses, J. Am. S oc. Mass. Spectrom., 15 (2004) 1487 1498. [49] D.A. Barnett, M. Belford, J. Dunyach, R.W. Purves, Characterization of a temperature controlled FAIMS system, J. Am. Soc. Mass. Spectrom., 18 (2007) 1653 1663. [50] J.E. McClellan, J.P. Murphy, III, J.J. Mulho lland, R.A. Yost, Effects of fragile ions on mass resolution and on isolation for tandem mass spectrometry in the quadrupole ion trap mass spectrometer, Anal. Chem., 74 (2002) 402 412. [51] A.W. Castleman Jr., R.G. Keesee, Small cluster: aerosol precursors Aerosol Sci. Technol., 2 (1982) 145 152. [52] A.W. Castleman Jr., P.M. Holland, R.G. Keesee, Ion association processes and ion clustering: elucidating transitions from the gaseous to the condensed phase, Radiat. Phys. Chem., 20 (1982) 57 74.

PAGE 183

183 BIOGRAPHICAL SKETCH Leonard Calvin Rorrer, III, was born in Salem, Virginia, in 1977. After graduating from Floyd County High School in 1995, he went on to Washington and Lee University in Lexington, Virginia. At Washington and Lee University, he worked as a research assistant in the area of organic and inorganic synthesis involving macrocyclic compounds. During his time there, he received two Robert E. Lee Scholar grants for summer undergraduate research. Le onard received his Bachelor of Science degree in June, 1999. In August of 1999, he entered the Department of Chemistry at the University of Florida to pursue his Master of Science degree under the supervision of Dr. Richard A. Yost in the area of mass spectrometry and ion mobility. Upon completion of this graduate work, he worked at U.S. Lithium Energetics, LLC in Alachua, FL as manager of Quality Control and Quality Assurance. After leaving U.S. Lithium Energetics, Leonard returned to the Departmen t of Chemistry at the University of Florida to pursue his d octora l degree under the supervision of Dr. Richard A. Yost in the area of high field asymmetric waveform ion mobility spectrometry. While working on his doctora l degree, he received the Proctor & Gamble Award for Excellence in Research. Upon completion of his graduate work, he plans to work in the area of analytical instrument design and development.