Old Dominion University ODU Digital Commons
Physics Theses & Dissertations Physics
Spring 2003 Measurement of Proton Transfer Reaction Rates in a Microwave Cavity Discharge Flowing Afterglow George M. Brooke IV Old Dominion University
Follow this and additional works at: https://digitalcommons.odu.edu/physics_etds Part of the Atmospheric Sciences Commons, Atomic, Molecular and Optical Physics Commons, and the Physical Chemistry Commons
Recommended Citation Brooke, George M.. "Measurement of Proton Transfer Reaction Rates in a Microwave Cavity Discharge Flowing Afterglow" (2003). Doctor of Philosophy (PhD), dissertation, Physics, Old Dominion University, DOI: 10.25777/6vm5-5117 https://digitalcommons.odu.edu/physics_etds/38
This Dissertation is brought to you for free and open access by the Physics at ODU Digital Commons. It has been accepted for inclusion in Physics Theses & Dissertations by an authorized administrator of ODU Digital Commons. For more information, please contact [email protected]. MEASUREMENT OF PROTON TRANSFER REACTION RATES IN
A MICROWAVE CAVITY DISCHARGE FLOWING AFTERGLOW
By
George M. Brooke IV B.S. May 1994, Virginia Military Institute M.S. December 1997, Old Dominion University
A Dissertation Submitted to the Faculty of Old Dominion University in Partial Fulfillment of the Requirement for the Degree of
DOCTOR OF PHILOSOPHY
PHYSICS
OLD DOMINION UNIVERSITY May 2003
Approved by:
Leposava Vuskovic (Director)
Anatoly V. Radyushkin (Member)
Charles I. Sukenik (Member)
Paul E. Ulmer (Member)
Karl H. Schoenbach (Member)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ABSTRACT
MEASUREMENT OF PROTON TRANSFER REACTION RATES IN A MICROWAVE CAVITY DISCHARGE FLOWING AFTERGLOW
George M. Brooke IV Old Dominion University, 2003 Director: Dr. Leposava Vuskovic
The reaction rate coefficients between the hydronium ion and the molecules
ethene (C2 H 4), propene (C?//*), 1-butene (C*//*) and hydrogen sulfide (. H2 S) were
measured at 296 K. The measured reaction rates were compared to collision rates
calculated using average dipole orientation (ADO) theory. Reaction efficiency depends
primarily upon the proton affinity of the molecules. All the measurements were obtained
using the newly developed microwave cavity discharge flowing afterglow (MCD-FA)
apparatus. This device uses an Asmussen-type microwave cavity discharge ion source
that is spatially separated from the flow tube, eliminating many of the problems inherent
with the original FA devices. In addition to measuring reaction rate coefficients, the
MCD-FA was shown to be an effective tool for measuring trace compounds in
atmospheric air. This method has many advantages over current detection techniques
since compounds can be detected in almost real time, large mass ranges can be scanned
quickly, and repeated calibration is not required. Preliminary measurements were made
o f car exhaust and exhaled alveolar air. Car exhaust showed the presence of numerous
hydrocarbons, such as butene, benzene and toluene while the exhaled alveolar air showed
the presence o f various volatile organic compounds such as methanol and acetone.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Copyright, 2003, by George M. Brooke IV, All Rights Reserved.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. This dissertation is dedicated to my loving and ever patient wife, Erika.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. V
ACKNOWLEDGMENTS
I would like to acknowledge first and foremost my advisor, Dr. Leposava
Vuskovic whose support and guidance throughout my graduate career has been
invaluable. Her enthusiasm for physics and concern for her students provides the ideal
environment for intellectual development.
Additionally, I would like to thank Dr. Svetozar Popovic for all of the support he
has provided. His guidance in the laboratory and insight into physics, especially gas
discharge physics, has played a critical role in my research.
I would also like to acknowledge and thank Dr. Charles Sukenik for all the
encouragement and support that he has given me and numerous other students at ODU.
His selfless dedication to students and his enthusiasm towards physics is something I will
always attempt to emulate.
I also want to thank Dr. Gary Copeland for many hours of useful discussions and
entertaining lunches. His broad knowledge of physics and inquisitive nature always
made me reflect more deeply on my own research.
I also want to thank the other graduate students at ODU, especially Hugh (Trey)
Thurman. Trey’s depth of knowledge never ceased to amaze me, and his friendship is
something I will always cherish. I also want to thank Thao Dinh, Pasad Kulatunga,
Prasong Kessaratikoon, Dorin Todor, and Tobias Oed for all the discussions and
enjoyable times we shared.
Finally I want to say thank you to my family for everything they have done for
me, especially my beautiful wife Erika, whose unwavering support and dedication have
been paramount to my success. I love you.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. TABLE OF CONTENTS
Page
LIST OF TABLES...... viii
LIST OF FIGURES...... ix
Section
1. INTRODUCTION...... 1
2. ION-MOLECULE COLLISION THEORY AND
CALCULATION OF REACTION RATE COEFFICIENTS...... 5 2.1 Ion-Molecule Collisions ...... 5 2.2 Proton Transfer Reactions in the Gas Phase ...... 12 2.3 Determination of Charge Transfer Reaction Coefficients ...... 13
3. EXPERIMENT...... 17 3.1 Proton Donor Source ...... 19 3.1.1 Hollow Cathode Discharge ...... 20 3.1.2 Radio Frequency Discharge ...... 25 3.1.2.1 Experiment ...... 27 3.1.3 Microwave Cavity Discharge ...... 33 3.1.3.1 Experiment ...... 35 3.1.3.2 Properties of the Microwave Discharge Plasma...... 43 3.2 Flow Tube ...... 49 3.2.1 Buffer Gas Flow Properties ...... 50 3.2.2 Venturi Inlet ...... 55 3.2.3 Reactant Gas Injection Ports ...... 58 3.3 Detection System ...... 60 3.3.1 Ion Optics ...... 61 3.3.2 Quadrupole Mass Spectrometer ...... 65 3.3.3 Signal Detection and Data Acquisition ...... 68
4. RESULTS AND DISCUSSION...... 74 4.1 Error Analysis ...... 75 4.2 Ethene (C2 H4)...... 77 4.3 Propene (Cjifg)...... 84 4.4 1-Butene ( C ^ ) ...... 89 4.5 Hydrogen Sulfide (H2 S) ...... 95
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Section Page
5. APPLICATION OF THE MCD-FA TO TRACE GAS ANALYSIS...... 101 5.1 Pollution Monitoring...... 106 5.2 Exhaled Alveolar Air...... 112
6. CONCLUSIONS...... 117
APPENDICES ...... 120 A. Theory of the Quadrupole Mass Spectrometer— ...... 120 B. Theory of Cylindrical Resonant Cavities ...... 126 C. Observed Argon Lines in MCD ...... 136 D. Useful Unit Conversions ...... 143
REFERENCES...... 144
VITA ...... 151
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. LIST OF TABLES
Table Page
I. Temperature evaluation data ...... 45
II. Properties of ethene at STP...... 77
III. Rate coefficient for the reaction of the hydronium ion with ethene in units of (cm3/s) ...... 79
IV. Properties ofpropene at STP ...... 85
V. Rate coefficient for the reaction of the hydronium ion with propene in units of (cmVs)...... 85
VI. Properties ofbutene at STP ...... 90
VII. Rate coefficients for the reaction of the hydronium ion with butene in units of (cm3/s) ...... 91
Vni. Properties o f hydrogen sulfide at STP ...... 95
IX. Rate coefficient for the reaction of the hydronium ion with hydrogen sulfide in units of (cm3/s) ...... 97
X. Proton affinity of various organic and inorganic compounds ...... 103
XI. Aliphatic alkanes and their applications ...... 107
XII. Concentration o f several trace compounds in car exhaust ...... 110
XIII. Summary of cylindrical resonant cavity equations ...... 135
XIV. Observed argon lines in a water vapor discharge with an admixture of argon (grating 1 - blazed at 500 nm)...... 136
XV. Observed argon lines in a water vapor discharge with an admixture of argon (grating 2 - blazed at 800 nm) ...... 137
XVI. Mass flow conversion table ...... 143
XVII. Pressure conversion table ...... 143
XVm. Miscellaneous conversion factors...... 143
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ix
LIST OF FIGURES
Figure Page
1. Diagram o f the typical flowing afterglow apparatus ...... 2
2. Collision trajectories for different impact parameters, b...... 6
3. Effective potential curves for a 0.025 eV collision between C> 2+ and O2 usingEq. (4)...... 8
4. The locking constant as a function of ii/otn at 300 K ...... 11
5. Schematic diagram of MCD-FA apparatus ...... 18
6. The hollow cathode proton donor source ...... 21
7. Photographs of the two hollow cathode discharge modes ...... 23
8. The voltage-current curves for the hollow cathode discharge at different pressures...... 23
9. Sputtering in the hollow cathode discharge ...... 24
10. The magnetic and electric field configurations in a solenoid ...... 27
11. CCRF discharge experiment setup ...... 29
12. Characteristics o f the CCRF discharge ...... 32
13. Asmussen-type microwave cavity ...... 37
14. The different operating modes available to the small (top) cavity and large (bottom) cavity ...... 38
15. Magnetron power supply circuits ...... 40
16. Characteristics of the MCD discharge ...... 42
17. Rotational spectrum of the (0-0) vibrational bond in the A-X system o f hydroxyl, OH' ...... 44
18. H /3 line in the microwave cavity discharge ...... 47
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. X
Figure Page
19. Ion and buffer gas axial velocities in the flow tube as a function of buffer gas injection rate ...... 53
20. Ion modulation in the flow tube...... 54
21. Venturi inlets...... 56
22. Venturi inlet used in the MCD-FA...... 57
23. Injector port configuration...... 59
24. Injector ports gas manifold...... 59
25. Schematic diagram of detection system...... 61
26. Ion optics assembly ...... 62
27. SIMION modeling of ion traj ectories ...... 65
28. Mounted QMS assembly and Channeltron housing ...... 66
29. Channeltron counts as a function of operating voltage ...... 69
3 0. Channeltron pulses ...... 70
31. Twin-T notch filter circuit ...... 71
32. Channeltron count rate as a function of discriminator level ...... 72
33. Spectrum used to calibrate QMS ...... 73
34. Maxwell-Boltzmann distribution of velocities o f ethene molecules at 296 K .. .78
35. Mass spectrum when the injection rate o f the ethene gas mixture is 30 seem. ..81
36. Decay of the hydronium ion signal as a function of ethene number density 82
37. Product ions as a function of the ethene mixture injection rate...... 83
38. Mass spectrum when the injection rate of the propene gas mixture is 6.0 seem...... 86
39. Decay of the hydronium ion signal as a function of propene number density... 87
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure Page
40. Product ions as a function of the propene mixture inj ection rate ...... 88
41. Mass spectrum when the inj ection rate of the 1 -butene gas mixture is 6.0 seem...... 92
42. Decay o f the hydronium ion signal as a function of 1 -butene number density...... 93
43. Product ions as a function of the 1-butene mixture injection rate...... 94
44. Mass spectrum when the injection rate of the hydrogen sulfide gas mixture is 6.0 seem...... 98
45. Decay o f the hydronium ion signal as a function of hydrogen sulfide number density...... 99
46. Product ions as a function of the hydrogen sulfide mixture injection rate 100
47. Mass spectrum o f car exhaust...... I l l
48. Mass spectrum of laboratory air...... 115
49. Mass spectrum o f the breath of a healthy non-smoker ...... 115
50. Mass spectrum o f the breath of a diabetic ...... 116
51. Mass spectrum of the breath of a smoker ...... 116
52. QMS geometry...... 120
53. Stability Plot for QMS ...... 123
54. Cylindrical waveguide ...... 126
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1
Section 1
INTRODUCTION
The study o f gas-phase ion-molecule chemical kinetics using fast flow techniques
began with the development of the flowing afterglow (FA) by Ferguson, Fehsenfeld and
Schmeltekopf in 1963 [1,2]. The development of the FA was in response to a need for a
more accurate method for the determination of reaction rate coefficients at thermal
energies, which was required for accurate modeling of the chemistry o f the ionosphere.
Previous to the development o f the FA, very little data existed on the reaction rates of ion
molecule collisions at thermal energies, and the data that did exist typically had an
uncertainty of an order of magnitude. In some cases the reaction rates were extrapolated
from reaction cross sections measurements taken from higher energy beam experiments
[3] or obtained from the then recently developed time-resolved afterglow [4].
Figure 1 shows a diagram of a typical FA apparatus. A fast flow (-100 m/s) of a
carrier gas, typically helium, at a pressure of approximately 0.5 Torr is established in the
flow tube by a roots blower. After entering the flow tube, the carrier gas enters a region
containing an ionization source. A source gas containing the precursor neutrals is either
injected with the carrier gas so that precursor ions are created directly by the ionizer or
the source gas is injected after the ionizer so that ionization occurs indirectly through
interaction with the excited state carrier gas atoms. The ions then flow down the tube to
the reaction region of the flow tube where the neutral reactant gas is injected at a
measured rate. The reaction time is determined by the speed of the carrier gas, and the
This dissertation follows the style of The Physical Review.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2
reaction length of the flow tube. The ions are then sampled through a small sampling
orifice hole ( electron multiplier. Since the reaction time of the ions is known, the reaction rate can be measured. The methods used to calculate the reaction rates are discussed in greater detail in Section 2. Ionization lon Source Source uas Korc lon Detection C ham ber Flow Tube Reaction Region C arrier G as Reactant Gas Port R oots Diffusion Blower Pump FIG 1. Diagram of a typical flowing afterglow apparatus. One of the major drawbacks in using the FA to measure reaction rates is the presence of excited state carrier gas atoms in the reaction region o f the flow tube. These atoms can interfere with the reactions being studied, creating secondary ions that are not directly associated with the reaction of interest, and giving a systematic error in the reaction rates. In addition, many precursor ions could not be produced under the operating condition of the FA. Smith and Adams solved this problem by separating the ion source from the flow tube, and using a quadrupole mass spectrometer to select ions before they were injected into the flow tube [5-7]. Using the selected ion flow tube (SIFT) it became possible to create almost any ion species using various ion source configurations and ion source gas mixtures. Since the source ions are created and filtered in a separate chamber before injection into the flow tube, the problems associated with Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. excited state carrier gas atoms were eliminated. The FA and the SIFT have produced a wealth o f information on ion-molecule chemical kinetics, providing valuable information to many areas o f basic research including the study of atmospheres [8], interstellar clouds [9,10] and low temperature plasmas [11]. In the early nineteen nineties, Smith and Spanel began investigating the application of the SIFT to the study of trace compounds in atmospheric air [12,13]. By choosing precursor ions that did not react with the normal constituents o f air, and injecting air samples into the reactant port of the flow tube, it became possible to detect and quantify contaminants down to the level o f tens o f parts per billion (ppb) in almost real time. Smith and Spanel have done extensive research into the development of the SIFT as a method for the detection of diseases using the exhaled alveolar air of patients [14,15], but the SIFT technique is applicable to any area of science where real time detection and quantification o f trace compounds is required [16,17]. Accurate quantification of trace compounds in air requires accurate measurements of ion-molecule reaction rates. While many reaction rates have been measured, [18,19] there are still many measurements needed before the flow tube technique can be widely adopted as a method for trace gas analysis. The major objective of the study presented here is the measurement of the proton transfer reaction rates between the hydronium ion and ethene, propene, 1-butene, and hydrogen sulfide. The hydronium ion has the property that, although it does not react with the normal constituents o f air, it does react with most volatile organic compounds as well as most unsaturated hydrocarbons. In addition, most reactions with the hydronium ion occur by a non-dissociative proton transfer reaction. This makes the ion an ideal Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4 choice for charging contaminant molecules without breaking them apart. The neutral molecules were chosen due to their importance in atmospheric pollution and their important role in the nucleation and growth of particulate matter (PM). The reactions of the hydronium ion with ethene and propene have been measured previously, and were used as a benchmark for the flow tube built for the present work, while the reaction with 1-butene has not been previously measured. The reaction o f the hydronium ion with hydrogen sulfide has been studied previously three times, but there exist considerable differences in the measured values. All measurements were made using the microwave cavity discharge flowing afterglow (MCD-FA). To the best of our knowledge, the MCD- FA apparatus is the first of its kind. The final objective of this study was two preliminary measurements of trace contaminants in atmospheric air using the MCD-FA. The first measurement focused on the various trace compounds present in the exhaust of a gasoline automobile, while the second measurement looked at the trace compounds present in the breath of a smoker, a diabetic, and a non-smoker with no known illnesses. In Section 2, the theory of gas phase ion-molecule collision is discussed as well as the role that proton affinities play in proton transfer reactions. In addition, the theory of reaction rate coefficient measurements in flow tubes is developed. In Section 3 the MCD-FA technique is described in detail. In Section 4 the experimental results of the reaction rate coefficient measurements are presented while in Section 5 application o f the MCD-FA method to trace gas analysis is demonstrated. Finally, in Section 6, conclusions are made based upon the measurements. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5 Section 2 ION-MOLECULE COLLISION THEORY AND CALCULATION OF REACTION RATE COEFFICIENTS 2.1 Ion-Molecule Collisions In order to understand charge transfer reactions, it is important to have a clear understanding of the processes involved in ion-molecule collisions. Theories involving these types o f collisions have been developed since the early part of the twentieth century [20], so an extensive amount of literature exists [21,22]; therefore, only a brief overview of the applicable material will be provided here. Consider the collision between a point ion with charge q, and a point molecule with polarizability, a. The typical expression used to describe the interaction of these two particles is given by the polarization potential (l) where r is the separation of the particles. Using classical mechanics [23], it can be shown that the total energy of the two interacting particles is given by the expression (2) where ris the relative velocity of the two particles along the axis connecting their centers, mr is the reduced mass, and I is the classical angular momentum. The second term in the above expression is called the centrifugal term and is associated with an Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. outwardly directed force. As a result, this term is typically grouped with the potential energy term to give an effective potential 2 mrr Using the inverse fourth order polarization potential, and using the relationship I2 = m2 b2 v2 this expression becomes y r , mrb2 v 2 a q2 _ K rb2 a q2 « i j " 2r2 2 / ' r2 2 r4 W where b is the impact parameter of the collision, and Kr is the kinetic energy. FIG. 2. Collision trajectories in the center of mass frame for different impact parameters, b. The shaded circle represents the capture collision cross section. Since the molecule and the ion are considered to be point particles, a collision will occur only when the separation between the two particles is zero.. The condition required Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7 for this is given by the impact parameter, which is a function of the initial relative velocities of the particles. Figure 2 shows a diagram depicting the collision. The shaded circle represents the area through which a particle must pass in order for a collision to occur, and is given by the expression (Tc(v) = 7rb2c(y), (5) where bc(v) is the impact parameter. The collision impact parameter can be obtained by calculating the height of the centrifugal barrier in the potential energy expression given by Eq. (4). The height o f the barrier gives the maximum energy a particle can have in order for a collision to occur. To obtain the position of the maximum, one takes the first derivative of Eq. (4) and sets it equal to zero. Solving this for r gives the position of the maximum, (6) At rc, Vgff—Kr, so substituting Eq. (6) into Eq. (4) gives the collision impact parameter, (7) Substituting expression Eq. (7) into Eq. (6) yields the relationship bc = rc4 l . Substituting Eq. (7) into Eq. (5), it is now possible to write the capture collision cross section as (8 ) The collision rate is related to the collision cross section through the relation K = j a c f (v)v d v , (9) 0 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. where^v) is the speed distribution of the particles. For simplicity, if the average speed is assumed to be v, then using Eq. (8) the collision rate can be expressed as kc =vcrv =2nq . (10) )]mr Equation (8) is typically referred to as the Langevin cross section, while Eq. (10) is known as the Langevin collision rate. 0.04 7 0.03 0.02 ? 0.01 h - 0 - 0.01 - 0.02 FIG. 3. Effective potential curves for a 0.025 eV collision between 0 2+ and 0 2 using Eq. (4). Figure 3 shows the potential energy curves for various impact parameters assuming an effective potential given by Eq. (4), for a collision between C> 2 + and O2 at 0.025 eV (room temperature). If the impact parameter is zero, meaning that the collision is head on, then there is no centrifugal barrier, and any value o f kinetic energy will result Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. in a collision. As can be seen in Figure 3, when the impact parameter is increased, the height of the centrifugal barrier also increases. When b>bc, the centrifugal barrier is too large for a collision to occur. While Eq. (10) is adequate for the calculation of collision rates of some simple ion-molecule collisions, it typically underestimates the rate of ion-polar molecule collisions since it fails to take into account a permanent dipole moment. Additionally, while it is typically not a factor in thermal collisions, Eq. (10) fails to take into account the structure of the particles and thus predicts a cross section and rate coefficient of zero in the high energy limit (> 2-3 eV). This problem is overcome by including the hard core in the kinetic theory and is discussed further in the literature [24]. The simplest theory to include the dipole moment of the molecule is called the locked dipole approximation [25]. The potential used to describe the interaction is that of a dipole interacting with a point charge, V(r) = -^Y~cos where }M is the dipole moment o f the molecule and (j> is the angle o f the dipole relative to the collision axis. In order to simplify the theory, it is assumed that the interaction between the ion and the dipole causes the molecule to “lock” into a permanent orientation such that os) 2 mrr 2 r r Using the same techniques used to obtain Eq. (10), the collision rate becomes K = (13) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 10 Averaging the reaction rate over thermal velocities using the Maxwell-Boltzmann distribution, fiv), gives / 2 nq 2 k = 2* q 4a+ jud0^-dv 4a+nd (14) m r V 0 nkT Since the locked dipole approximation assumes that the ion-molecule reaction is time independent, and is at a maximum at all times, it considerably overestimates the collision rate of the particles, and is useful only in determining an upper bound of the rate. To describe more accurately the ion-molecule collision, Eq. (14) was modified to include a “locking” coefficient, c, in order to take into account the time averaged dipole field produced by the rotating molecule, t = 2rcq'\lar~ + cfid ' ^ (15) 7lkT 'r V The coefficient is the time-averaged value of cos^ ^cos^j and therefore varies between 0 and 1. This coefficient has been experimentally determined for several molecules [26], and became the basis for the average dipole orientation (ADO) theory developed primarily by Bowers and coworkers [21,26-28]. To predict the value o f c, ADO theory calculates the average value of c as a function o f r using the dipole moment and the polarizability of the molecule. ADO theory predicts that, at constant temperature, the locking constant is simply a function of Hdlct12. By plotting the value of c as a function of fMlct12 for a given temperature, it is possible to parameterize ADO theory allowing for simple calculation of collision rates. Figure 4 shows the locking constant as a function o f /^ /a 1/2 at 300 K. The points are calculations by Su and Bowers [26], and the curve is a polynomial fit to Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 11 the data. Values for the dipole moments and polarizabilities of many molecules are available in the CRC handbook [29]. 0.3 0.25 0.2 O 0.15 0.05 0 0.5 1 1.5 2 Hdla 112 (Debye/Angstrom372) FIG. 4. The locking constant as a function of jj/< x V2 at 300 K. The data points are theoretical calculation of c using ADO theory [26] while the line is a sixth-order polynomial fit to the data. Parameterized ADO theory provides a simple method for calculating collision rates for ion-polar collisions at thermal energies. In addition, since exothermic proton transfer reactions occur at approximately the collision rate, ADO theory provides a useful method for estimating the reaction rate. While this theory is typically accurate to within 30% for ion-molecule collisions at thermal energies, at energies greater than a few eV, it becomes critical to include the structure o f the particles in the collisions. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 12 2.2 Proton Transfer Reactions in the Gas Phase The general expression for a proton transfer reaction is AH++B±>A + B H \ (16) where A and B are atoms or molecules, and i f is a proton. The reaction can be described by the Bronsted-Lowry acid-base definition [30]. In this definition, an acid is a species that donates a proton, called the proton donor, while a base is a species that accepts a proton, called the proton acceptor. Consider, for example, the reaction of the hydronium ion with ammonia, H 30 + + NH3 NH +4 + H 20 (17) In this reaction, the hydronium ion is considered the acid, while the ammonia molecule is considered the base. For every acid, there is a conjugate base; likewise forevery base there is a conjugate acid. In reaction (17) the acid, H^Cf has the conjugate base HzO, while the base NH3 has the conjugate acid NHf. The direction in which reaction (16) occurs is determined primarily by the proton affinity of the molecules. Formally, the proton affinity is defined as the difference in the enthalpy of formation of a protonated molecule, B i t , and its neutral counterpart, PA = -&rxnH°=&fH°{B) + KfH°{H+)-lifH°{BH+) (18) The reaction direction is determined by whichever molecule has the larger proton affinity. Using reaction (16), if B has a proton affinity larger than A then the reaction will occur from left to right. In reaction (17), water has a proton affinity of 7.2 eV, while ammonia has a proton affinity of 8.8 eV meaning that the reaction will proceed from left to right. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 13 Bohme et. al. [31] and Lindinger et al. [32] have done extensive studies on exothermic proton transfer reactions and found that the vast majority o f these reactions occur at, or close to, the collision rate. It was shown [31] that the Gibbs free energy (AG) of the proton transfer reaction determined the overall efficiency. The change in Gibbs free energy o f a reaction is defined as AG = AH-TAS, where AH is the change in enthalpy, T is the temperature o f the system, and AS is the change in entropy of the system. It should be noted that in the definition of the Gibbs free energy, the temperature and pressure of the system are constants. For a change in the Gibbs free energy (at atmospheric pressure and 298 K) o f AG"9g <-10 kcal/mol (-0.4 eV), krlkc~ 1, while reaction efficiencies dropped significantly for AG^g% >-10 kcal/mol. Using efficient proton transfer reactions in conjunction with various flow tube techniques has provided a method to test ADO theory as a means for calculating collision rates. 2.3 Determination of Charge Transfer Reaction Rate Coefficients The technique used to calculate the reaction rate coefficient for charge transfer reactions is similar for all flow/drift tube experiments. In general, a charge transfer reaction has the form A++b X c ++D, (19) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 14 where kr is the reaction rate o f the charge transfer reaction. The upper limit of the reaction rate is given by the collision rate, kc, where every collision between A+ and B results in charge transfer. The efficiency of the reaction is then defined by the expression s = ^-. (20) K In most flow/drift tube experiments the determination o f the reaction rate coefficient is calculated by measuring the decay o f the precursor ion concentration, [A+], as the reactant gas number density is increased. The decay of the hydronium in the reaction region of the flow tube can be described by ■ = v (r) 5M I!_ A d (~dtA+~^r- + ■ Di^^-k{B}[£]n 5 2 U +] (21) dt dz r dr V d r J dz '-----»-----' 1 where v,{r) is the ion velocity as a function o f radial displacement and k is the reaction rate coefficient [2]. In Eq. (21), term 1 is the diffusive contribution of the precursor ions due to radial diffusion, term 2 is the diffusive contribution due to axial diffusion, and term 3 arises from reactive losses with molecule B. Depending on the type of flow tube being utilized, A is either the ambipolar diffusion constant or the free diffusion constant. In the standard FA, where a gas discharge exists in the flow tube, the ambipolar diffusion between the positive ions and the free electrons is the primary source o f diffusive losses. On the other hand, in the SIFT and the MCD-FA, the ions are produced in a separate chamber and then introduced into the flow tube. As a result, free diffusion of the ions is the predominant source o f losses. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 15 Bolden et al. [33] and Adams et al. [34] have studied solutions of Eq. (21) and found that if the axial diffusion term is neglected (resulting in an error of -2% at 0.5 Torr), then it is possible to write the analytic solution as ^L+TklBjY [A+] = [A+ ]0e^a (22) where A is a factor related to the slip coefficient between the gas and the walls of the flow tube, Y=v0/vu a is the radius of the flow tube, and v0 is the bulk gas flow velocity. The bulk gas velocity is given by the expression v „ = - T - > (23) na p where Q throughput of the buffer gas and p is the pressure in the flow tube. Smith et al. [35] studied the ion density in a flow tube as a function o f axial position, and verified the functional dependence given by Eq. (22). Using Eq. (22) it is possible to measure the reaction rate between A+ and B by fixing the reaction length z-l, and varying the number density o f the reactant gas in the flow tube. Since z is now a constant, the diffusion term can be absorbed into [A+] 0 so that Eq. (22) becomes k[B]l [A+] = [A+]0e ' . (24) Solving this equation for the rate coefficient, k = i- L ln M l (25) l)[B] where the term in brackets is the exponential decay constant of the ion number density as a function of reactant gas number density. Since the count rate /, measured in the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 16 detection chamber of the flow tube, is directly proportional to the number density of the ions in the flow tube, the reaction rate can be written as ( & = -—i— In (26) l\[B] Using Eq. (26), the reaction rate is obtained by measuring the decay constant of the precursor ion [A+] as the reactant gas number density is increased in the flow tube. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 17 Section 3 EXPERIMENT As discussed in Section 1, there are several flow tube techniques available to study charge transfer reaction kinetics. The type of technique used depends on the reaction being studied. The original flowing afterglow (FA) devices [2] produced the discharge in the flow tube resulting in a wide spectrum of ion species as well as buffer gas metastables that entered the reaction region. As a result, the first order rate equation that was assumed in Section 2 was, in many cases, incomplete. To solve this problem, Smith and Adams [6] separated the ion source from the flow tube, and filtered the ions before they were injected into the flow tube. While this technique is effective, the addition of a selected ion flow tube (SIFT) chamber does have two major drawbacks: (1) the cost, both technical and financial, of the addition of the SIFT chamber, and (2) the loss of ions in the filtering process. The microwave cavity discharge flowing afterglow (MCD-FA) technique was developed for the present studies of hydronium ions interacting with various other organic and inorganic molecules. The advantage of using the MCD-FA over the standard FA and SIFT for these types of reactions is that it retains the simplicity of the FA device but separates the ionization source from the flow tube. Figure 5 shows a schematic diagram of the MCD-FA apparatus. Hydronium ions are created in an ion source, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. H-* 00 Detection QMS Product Ion Ion Optics Ion Reaction Region Reaction Flow Tube FIG. 5. FIG. ofMCD-FA apparatus. Schematic diagram Gas Cartier Cartier Donor Magnetron Source Charge Tube QuartzDischarge Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 19 separate from the flow tube, and are allowed to effuse through a small hole into the flow tube. Since the gas discharge remains separated from the flow tube, the production of unwanted metastables and ion species is significantly reduced. To eliminate the need for the SIFT chamber, a microwave cavity discharge (MCD) ion source was developed that produces predominantly hydronium ions. The MCD-FA apparatus is the first of its kind, to the best of our knowledge. 3.1 Proton Donor Source Production o f a stable, intense and clean source of precursor ions is critical to the study of charge transfer reactions. Numerous types of sources have been successfully applied to the production of many different types of ions. The ideal characteristic of a charge donor source is that it produces a large supply of one particular species of ion, since other ion species in the reaction region of the flow tube make the reaction being studied difficult to interpret. In many cases the discharge chemistry does not produce a predominant ion, so techniques such as the selected ion flow tube were developed to address this issue. A major innovation of our research was the development o f a clean source of hydronium ions that did not need an upstream chamber for ion selection. Gas discharges in water vapor under the right conditions have been shown to be intense source of Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 20 hydronium ions [36,37]. We have studied several different ion sources in an effort to optimize hydronium ion production while reducing the presence of contaminant ions. 3.1.1 Hollow Cathode Discharge The hollow cathode discharge was the first source studied by our group [38] since previous work had shown it to be an effective source of H O 4- ions [36,37]. In addition, the source is simple to build and requires only a single high voltage DC power supply with a current limiting ballast resistor to operate. Figure 6 shows a schematic diagram and photograph o f the hollow cathode discharge chamber. The discharge produced between the anode and the hollow cathode is very similar to the DC glow discharge produced between two parallel plates. The breakdown mechanism and the structure of self-sustaining DC glow discharges is well understood and is described in many books [39,40], therefore only a brief summary will be given here. As the voltage between the anode and the cathode is increased, free electrons, liberated by cosmic rays or other sources o f external ionization, begin to accelerate away from the cathode. If the voltage difference is large enough, each electron will gain enough energy to ionize an atom during a collision, thereby producing an additional electron. This buildup of electrons, called the avalanche process, is what triggers the start of the self-sustaining discharge. Once the discharge begins, a steady state is quickly reached with free electrons drifting away from the cathode and positively charged ions accelerating towards the cathode and striking the surface. This bombardment o f the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 21 Power Supply View Port Connection F la n g e Flange Viton View O-Rings Port Gas Input Hollow Cathode Anode Teflon Vacuum Insulator (a) b. Photograph of discharge chamber. (b) FIG. 6. The hollow cathode proton donor source: (a) Schematic diagram of hollow cathode discharge chamber, (b) Photograph of hollow cathode discharge chamber. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 22 cathode is an important source of secondary electrons, and therefore the material from which the cathode is made is an important factor in the breakdown characteristics of the discharge. Figure 7(a) shows a normal DC hollow cathode discharge in argon at 450 mTorr with -2% water vapor. The bright spot in the center of the picture is the negative glow and is a plasma, containing approximately equal amounts of negative charges and positive charges. The dark space between the negative glow and the cathode surface is called the cathode fall and is where most of the potential drop between the anode and the cathode occurs. As the current in the discharge is increased, the diameter of the negative glow will also increase. Since the size of the negative glow is limited by the size of the hollow cathode, increasing the current beyond a certain value will cause the discharge to transition into the hollow cathode mode. This mode is characterized by a diffuse glow that is distributed evenly inside the cathode and is made up primarily o f thermalized free electrons [41]. Figure 7(b) shows a photograph of the hollow cathode discharge mode. Figure 8 shows the I-V characteristics of the hollow cathode discharge. The discontinuity in the curves is due to the transition from the normal glow to the hollow cathode mode. While the hollow cathode is a good source of ions its one major drawback is cathode sputtering. As discussed earlier, when ions are formed due to electron impact ionization, they accelerate and strike the surface of the cathode, producing secondary electrons. In addition, this ion bombardment of the cathode also liberates metal atoms from the cathode surface. Once free, these atoms migrate throughout the discharge chamber and are deposited on exposed surfaces. This process, called sputtering, causes cathode material deposits to develop throughout the chamber. Figure 9(a) shows the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. FIG. 7. Photographs of the two hollow cathode discharge modes: (a) Negative Glow Mode, (b) Hollow Cathode Mode. ♦ 275 mTorr ■ 490 mTorr A 525 mTorr o O 450 = 400 4, ♦ ♦ ♦ 0.0 0.5 1.0 1.5 2.0 2.5 Current (mA) FIG. 8. The voltage-current curves for the hollow cathode discharge at different pressures. The discontinuity in the curve is the transition from the normal glow to the hollow cathde mode. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 24 inside surface o f the gas input flange and Teflon insulator with copper deposits after many hours of operation, while Fig. 9(b) shows the spectral lines due to copper cathode in the emission spectrum of negative glow. 40.0 2 30.0 ■ .t : -2 20.0 - 15.0 320 321 322 323 324 325 326 327 328 329 330 ‘ ____ Wavelength (nm) _ (b) FIG. 9. Sputtering in the hollow cathode discharge: (a) Copper deposits on the gas injection flange and Teflon insulator due to sputtering, (b) Observed spectral lines due to copper in the negative glow of the discharge (324.8 nm and 327.4 nm) Development of the hollow cathode discharge was stopped after these studies due to the problems with sputtering. While a hollow cathode discharge has been successfully applied to H 3 0 + production for drift tube studies [36,37] by using an intermediate chamber for ion thermalization and sputtering protection of the tube, other concerns prevented its application to the MCD-FA. Not only would the cathode have to be replaced on a regular basis, the discharge chamber itself would have to be disassembled and cleaned of all copper deposits to prevent arcing. In addition, there were concerns that over long periods of time, copper deposits would begin to form on the inside of the flow tube. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 25 3.1.2 Radio Frequency Discharge In order to eliminate the problems associated with cathode sputtering, radio frequency discharges were considered as a possible source of hydronium ions. There are two basic types of RF discharges, capacitively coupled and inductively coupled discharges. The details of these discharges are quite complex, and many books are available which describe the theory [42,43]; therefore, only a brief, qualitative overview of each type will be given here. Consider the field produced between two parallel plates. If one electrode is electrically grounded and an RF signal is applied to the other, then free electrons between the plates will oscillate 180° out of phase with the applied field. If the electrons do not undergo any collisions then the average power absorbed by the system will be zero. On the other hand, if the electrons undergo random inelastic collisions with gas molecules between the plates, then the kinetic energy of the electrons will be transferred to the gas molecules. This process will occur as long as there are free electrons in the system. As discussed in Section 3.1.1, a self-sustaining discharge cannot be produced unless there is at least one free electron created for every electron-atom collision. If the amplitude of the RF field generated between the two parallel plates is large enough, then each electron will have enough energy, on average, to ionize a gas molecule. This process produces the electron avalanche required for a self-sustaining discharge. Since the discharge is produced by the electric field between two electrodes, this type of RF discharge is commonly referred to as capacitively coupled radio frequency (CCRF) discharge. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 26 In contrast to the CCRF discharge which is produced by a time-dependant potential difference between two electrodes, the inductively coupled discharge is produced by a time dependent current. Consider a solenoid carrying a current. For a steady state current, the magnetic field generated along the axis of the solenoid is B = ju0nlz , (27) where fig is the permittivity of free space, n is the number o f turns per unit length, I is the current in the wire and z is a unit vector along the axis o f the solenoid. Assume now that the current varies sinusoidally in time so that the time dependent current can be written as B = j i j i l sin(im t)z, (28) where ® is the angular frequency of the applied current. We know from Maxwell’s equations that a time dependent magnetic field (with no current sources) produces a time dependent electric field, Vxl = —. (29) dt Substituting Eq. (28) into Eq. (29) and solving the differential equation we obtain the electric field, E = cos (at}[yx - xy\. (30) The electric field produced by the magnetic field oscillates 90° out of phase with the magnetic field, but has the same frequency. If the current is large enough, then the induced electric field will produce a gas discharge in the same manner as described for the capacitively coupled discharge. Figure 10 shows the field configuration in a solenoid. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 27 y B-Fieid (a) (b) FIG. 10. The magnetic and electric field configuration in a solenoid: (a) A side profile of an infinitely long solenoid, and the magnetic field generated with a steady state current, (b) The electric field generated by a time varying magnetic field. Note that the diagram illustrates the fields at one instant in time. In order to generate large enough fields to induce breakdown of the gas, it is necessary to produce large RF currents in the solenoid. Since the electronic components required to produce the RF currents for an inductively coupled discharge were not readily available and the components for the CCRF discharge were, the CCRF discharge was selected for study as a proton donor source. Although the inductively coupled discharge was not studied in the present work, future work may evaluate it as a possible source. 3.1.2.1 Experiment The electrode configuration used to produce the CCRF discharge was a coil of magnet wire wrapped around a 0.75” (1.9 cm) OD quartz tube. A Mini-Circuits ZOS-50 voltage controlled oscillator (VCO), with a fixed output power o f +12 dBm, and tuned to 25 MHz was used to generate the RF signal. An RF amplifier was used to amplify the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 28 signal coming from the VCO. Through trial and error it was found that a forward power of 2 watts was needed to maintain the discharge at 300 mTorr. Power was adjusted using fixed attenuators between the VCO and the amplifier. Measurements were also made at 4 watts in order to determine how power affected ion output. An impedance matching circuit was used to optimize RF power coupling to the discharge. When optimally matched, less than 0.5 watts of power was reflected. Before the CCRF discharge was connected to the flow tube, the ion output was studied using a detection system mounted directly next to the discharge. Figure 11 shows a schematic diagram and photograph of the experimental setup. Ions were sampled from the discharge through a 350 pm hole drilled in the center o f a stainless steel VCR gasket obtained from Lenox Laser, Inc. In order to allow optimization of the ion signal, the gasket was electrically isolated from ground using an alumina spacer and held in place using Torr-Seal. Once sampled, the ions were focused and accelerated using a set of ion optics. The sampling orifice and ion optics plates were optimized by monitoring the ion signal from the detector while adjusting the voltage on each plate. Each plate was biased using a voltage divider in conjunction with a ground referenced battery. Sampled ions were filtered using a quadrupole mass spectrometer (QMS) housed inside a stainless steel vacuum chamber. The theory o f operation o f the QMS is well documented and will not be discussed here, but an overview is given in Appendix A. The rods o f the QMS used in the detection chamber had a diameter of 0.5”, a length of 8” and were made from stainless steel. Filtered ions were detected using a Galileo 4772 Channeltron biased at -2600 V. Output current from the Channeltron was amplified using an Ortec 109PC preamplifier before being sent to a computer for data acquisition. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 29 ii r \ i j ' Pyrex Water Needle Reservoir Channeltron Valve w Housing Sampling Orifice Ion °Ptics Antenna I uuc /WWWVWV\- 'UMJlMUr s Thermocouple Vacuum Sensor Channeltron Quadrupole Mass Spectrometer Mechanical Pump Diffusion VCO Amjplii® (a) FIG. 11. CCRF discharge experimental setup: (a) Schematic of CCRF discharge apparatus, (b) Photograph of CCRF discharge apparatus. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 30 A Lab view program was written to aquire the Channeltron signal while scanning the masses. In order to simplify alignment of all the components, the Channeltron was housed in a separate chamber that was attached to the quadrupole chamber by means of an o-ring seal. Alignment could easily be checked by removing the Channeltron housing and looking down axis of the quadrupole and adjusting the ion optics at the front end of the chamber. The system was pumped by a Varian HS-2 diffusion pump, with a pumping capacity of 285 1/s, and was backed by an Alcatel 2008A mechanical pump. To prevent oil from back-streaming into the vacuum chamber, a molecular sieve trap was placed between the mechanical pump and the diffusion pump, and a pneumatic gate valve with a cold-water trap was placed between the diffusion pump and the quadrupole vacuum chamber. The pneumatic gate valve interlocked to the cooling water of the diffusion pump and the mains power. If water pressure or power was lost, the gate valve would be closed and the Channeltron disabled. Unlike the SIFT, the MCD-FA used no upstream quadrupole mass spectrometer for selection of the precursor ions. This meant that our source had to supply a large number of hydronium ions with the smallest number of contaminants. Lindinger and coworkers [44] have found that hydronium ions are the dominant ions produced in a hollow cathode discharge in water vapor. Based upon this observation, it was decided to use water vapor as the discharge gas of the CCRF discharge. The water vapor was supplied to the discharge tube from a Pyrex reservoir that had been filled with deionized Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 31 water. To eliminate dissolved atmospheric gases in the water, the reservoir was simultaneously heated and evacuated for approximately 1 hour prior to each experiment. To optimize the discharge for hydronium ion production (see Fig. 12), we monitored the ion spectrum as the power and pressure were changed. Figure 12(a) shows the peak ratios for the largest peaks observed in the spectra. As the pressure in the discharge tube was varied it was observed that relative ion concentrations also varied. At lower pressures the signals from the different ions were observed to be approximately equal. As pressure in the discharge tube was increased, the hydronium ion signal increased significantly relative to the other ions. The location of the coil relative to the sampling orifice also had a noticeable effect on the output signal of the ions. While the relative concentrations o f the various ions changed very little, the signal intensity changed dramatically. It was initially thought that the closer the coil was to the sampling orifice, the stronger the signal would be, but it was found that the maximum ion signal was obtained when the coil was several centimeters from the sampling orifice. The reason for this maximum is due to the properties of the CCRF discharge. The electric field that generates the discharge couples to the electrical ground of the vacuum chambers and experiment supports. The field structure was optimized when the coil was placed 2.5 cm from the sampling orifice. Figure 12(b) shows the mass spectrum of the CCRF discharge when all the parameters were optimized for H 3 0 + production. While it was not possible to completely eliminate the production of the various contaminate ions, their signal could be reduced sufficiently so that the spectrum of contaminant precursor ions closely mirrored the contaminant ions in other SIFT experiments [12]. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 32 (H30+)/(H20+) 4 W (H30+)/(02+) 4 W 50 - - (H30+)/(H20+) 2 W (H30+)/(02+) 2 W c 30 I 20 0 50 100 150 200 250 350 400300 Pressure (mTorr) (a) - "M 1 1 1 1 1 T l”| ”, ”1 T i’TH'T TTT, , „ rrTf rl 0 10 20 30 40 50 60 70 80 90 Mass (amu) (b) FIG 12. Characteristics of the CCRF discharge: (a) Peak height ratios, (b) Mass spectrum optimized for H30 +. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 33 3.1.3 Microwave Cavity Discharge The microwave cavity discharge (MCD) was also considered as a possible source of hydronium ions. Other groups have successfully used microwave discharges as charge donor sources for flow tube experiments [45], but these groups have typically used gas mixtures and an upstream selection chamber to filter the desired ions for injection into the flow tube. As mentioned earlier, Lindinger and coworkers demonstrated that a hollow cathode discharge in water vapor produces a clean source of hydronium ions that does not need to be filtered before injection into the flow tube, hi order to eliminate the cathode sputtering caused by the hollow cathode, a MCD in water vapor was built as a cleaner alternative to Lindinger’s design. Smith and Spanel [13] have been successfully applying the Evenson-type microwave cavity [46] for production of H 3 0 +, Oi and NO+ for their SIFT experiments. The discharge gas used was argon with a small admixture of water vapor (less than 5%) and produced high concentrations of several ion species. This allows the selection of several precursor ions by simply adjusting the upstream quadrupole mass spectrometer. This experimental arrangement has the advantage that the gas discharge chemistry does not have to produce one predominant ion and reactions with multiple precursor ions can be done without changing the discharge chemistry. Since our experiments required only the production of the hydronium ion as the precursor ion, the requirement of using an upstream quadrupole was eliminated by using water vapor as the discharge gas. Although the Evenson-type cavity provides a stable source o f ions, coupling of the microwave radiation to the source requires an external Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 34 microwave generator, waveguides, and a coupling antenna. Although this equipment is available it is generally quite expensive. As an alternative to using an external microwave generator with an Evenson-type cavity, a microwave oven magnetron mounted directly to an Asmussen-type [47,48] cylindrical cavity was built. This arrangement had the advantage that microwave ovens are very inexpensive and quite impervious to destruction by reflected microwave power. In addition, construction of the microwave cavity can be built quite easily. Asmussen- type cavities have been used for many applications such as thin film deposition [49] and space plasma research [50] but, to the best of our knowledge, have not been applied to the production of precursor ions for flow tube experiments. Since previous measurements have shown microwave cavity discharges to have quite high ion densities [50] it was believed that the Asmussen-type cavity would provide a very dense source of hydronium ions for flow tube studies. As with the gas discharges discussed in the previous sections, the MCD requires a sufficiently large electric field to produce an electron avalanche and, subsequently a discharge. Production of these large fields is accomplished by designing the cavity so that the microwave radiation emitted by the magnetron is in resonance with the cavity. The primary reason that microwaves are used instead of another band of the electromagnetic spectrum is that the wavelength of microwave is on the order of centimeters making the design and construction of the cavity very simple. A full derivation of the field structure and resonant frequencies o f a cylindrical resonant cavity is given in Appendix B. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 35 3.1.3.1 Experiment The proton donor source built for our experiment was an Asmussen-type microwave cavity. In this configuration, a quartz tube containing the low pressure discharge gas extended along the center axis of a cylindrical cavity. Tuning is typically accomplished by making one of the end caps adjustable so that the length can be varied. Microwaves are introduced into the cavity by inserting the magnetron antenna through a hole drilled in the side of the resonant cavity. Although microwave generators, amplifiers and waveguides are available, the commercial microwave oven industry, by far, produces the most abundant and inexpensive supply of microwave parts. The output powers of the ovens typically vary from 600 to 1700 watts, providing more than sufficient power to induce breakdown in the discharge tube. In addition, all commercial microwave ovens operate at 2.45 GHz allowing the design of the cavity to be done with almost no concern of manufacturer. With the microwave frequency fixed, the radius and cavity length were the factors that determined the cavity mode. Two different cavities, each with a different inside diameter, were built as prototypes for a proton donor ion source. Both cavities were constructed from large diameter aluminum tubing, and machined so that the inside surface was smooth. Although other groups have used rolled brass sheet wrapped around cylindrical forms as the cavity to make construction simple [49], it was decided that the aluminum tubing would provide a more rigid cavity, and thus a more stable discharge. The end caps were machined from solid aluminum bar stock with holes being drilled in the center to allow for the insertion of the quartz tube. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 36 Figure 13(a) shows a schematic diagram of the cavity’s general design while Figure 13(b) shows a photograph of the larger cavity. Both cavities were very similar in overall design. The first cavity constructed had an inside radius of 4.6 cm, and a wall thickness of 5 mm, and the second cavity had an inside radius of 6.4 cm and a wall thickness of 5 mm. For tuning, the cavity length was adjusted by rotating a knob on one of the cavity’s end plates. This knob was threaded around a piece of threaded aluminum tubing that attached to the adjustable wall. To monitor the discharge, 0.25” holes were machined into the side of the cavity. One o f the main concerns in constructing the microwave cavity was that the large fields produced inside the cavity would cause arcing at sharp edges. To prevent this all of the edges were slightly rounded with sandpaper while being machined. Using the cavity radius and the driving frequency of 2.45 GHz, the cavity lengths were calculated using the equations summarized in Table XIII in Appendix B. It is important to point out that the calculated mode lengths are accurate for determining the location of the adjustable wall that will induce breakdown o f the particular mode. Once breakdown occurs, further tuning of the cavity is required since the plasma acts as a conductive surface, thus changing the boundary conditions used in calculating the resonant modes. Figure 14 shows the relative sizes o f the cavities and the accessible resonant modes and the length at which the cavity was primarily operated. For a given frequency, the distance between modes decreases as the radius increases. For the smaller cavity the only accessible length dependent ( p * 0 ) mode was TEni but, as can be seen in the figure, the operating length was over 2 cm shorter than this. In addition, when operating Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 37 Magnetron Magnetron Power Supply Connector Threaded Tuning Knob Adjustable Wall g Sampling Vacuum Orifice Sick _ J Pump Side Quartz Tube Knob Retaining Ring (a) (b) FIG. 13. Asmussen-type microwave cavity: (a) Schematic of microwave cavity, (b) Photograph of large diameter cavity. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 38 the smaller cavity, it was observed that a discharge could be maintained over a broad range o f cavity lengths. The most likely explanation for this is that the cavity was operating both the TMoio and the TEm modes. The operating length was determined by maximizing the ion counts. The modes of the larger cavity were much better defined. Breakdown would occur at all the resonant cavity lengths calculated, and would occur over a very narrow range. Although the cavity operated in several length dependant modes, the TMon was chosen as the operating mode since the ion output was greater than at any other. In addition, this mode appeared to be the most stable. E • | ° 1 { _ 2 1 j o OS B * loo S £?* Icn O) £ JZ2 -J * 1 . 4as— tO< co i* !*- jt- tPV 2 {LU < *C * f U i 1 1 1 1 1 E ; l 1 1 0 «$ £ 1 1 £ t £ (D Is £ o i = 1 1 <-> as r o l l 0 hw i 1 S r". 1 tri ! r»' s 5 * aj it ! _J ■ £2 - £ 4-4ro m l CN | to aiHy- I;; 1 rv Z 5 = : i o *CN 1 5 toiss 5 Sluu 1 LU h- | f * i r I t - a * ® 1 i O;* 1 I u i s St 1 i FIG 14. The different operating modes available to the small (top) cavity and large (bottom) cavity. Dotted grey lines represent calculated resonant modes based on the theory described in Appendix B. The dotted black line shows the normal operating length. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 39 The magnetron used to generate the microwaves was a 1000 watt 2M246 manufactured by LG, and removed from a Panasonic NN-S431WL microwave oven. Coupling o f the microwave radiation was accomplished by simply inserting the magnetron antenna through a hole machined through the side of the cavities. All commercial magnetrons used in microwave ovens have a soft, conductive mesh surrounding the antenna that acts as a seal to prevent the leakage of microwave radiation. Using brackets attached to the cavity, the mesh was firmly seated against the cavity wall. Although mounting the magnetron in this configuration prevented the measurement of forward and reflected power, it greatly simplified the construction of the cavity. For cooling, a muffin fan continually forced air through the heat sink fins of the magnetron. The discharge was produced in a quartz tube that ran along the axis of the cavity. The tube had an outside diameter of 0.75” and an inside diameter o f 0.625”. The vacuum system used for the MCD discharge was identical to the CCRF discharge except that the antenna coil was replaced by the discharge cavity. Initial tests used Pyrex instead of quartz for the discharge tube, but it was found that at typical operating powers the Pyrex tube would fracture due to thermal expansion, and at even higher powers the tube would melt. Power to the magnetron was supplied by a modified version o f the standard half wave voltage doubling circuit used in most microwave ovens. Figure 15(a) shows the standard microwave oven magnetron circuit. In order to operate, magnetrons require a low voltage, high current supply (usually 3 V, 10 A) for the filament, and a high voltage low current supply to bias the filament relative to ground (usually 4000 V, -400 mA). To make operating costs low, most manufacturers o f microwave ovens use a single Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 40 transformer with a 120 VAC input and dual outputs. One of the outputs is the high current supply for the magnetron filament and the other is the high voltage supply used for the filament bias. The high voltage output from the transformer is typically around 2000 V and is further stepped up to approximately 4000 V using the half wave-voltage doubling circuit, seen in Fig. 15(a). Since the output is not filtered, the microwave output o f the magnetron is pulsed at 60 Hz. Using the circuit shown in Fig. 15(a), the microwave output from the magnetron is fixed. To control the power delivered to the food during the cooking process, the high voltage to the magnetron is typically cycled on and off at different duty cycles. Although this circuit works well for cooking food, it is not effective for power regulation in a discharge cavity. Magnetron Magnetron Variac High Voltage Diode High Voltage Diode (a) (b) FIG. 15. Magnetron power supply circuits: (a) Typical microwave oven circuit, (b) Modified microwave oven circuit used for varying the peak microwave output of a magnetron. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 41 Meiners and Alford [49] found that by varying the input to the high voltage transformer, they could vary the output power of the magnetron. To accomplish this, a second transformer, whose input is controlled by a Variac, is added to the circuit. Figure 15(b) shows the schematic of this circuit. To adjust the forward power of the magnetron all that is required is the adjustment o f the Variac output voltage. Walker et al. [50] have developed a circuit that full wave rectifies the high-voltage going to the magnetron thereby increasing the pulse frequency to 120 Hz. Recently, the commercial microwave oven industry has started producing microwave ovens using switching power supplies to control the magnetrons [51]. These circuits actually vary the microwave power by adjusting the biasing high voltage of the magnetron filament and not by varying the on/off duty cycle of the magnetron. One of these ovens was obtained and the circuit was modified to allow it to operate the discharge cavity. Although the discharge would ignite, the output was found to be unstable. The most probable reason was due to an incorrect modification of the microprocessor and circuitry used for power regulation of the switching supply. Application of this circuit with a stable output would allow for continuous operation o f the discharge and shows promise for future application to the MCD-FA. Using the same experimental scheme used for the CCRF discharge, the ion output of the MCD discharge was measured as a function of water vapor pressure and forward power. The forward power was estimated from the Variac voltage and the power conversion efficiency of the magnetron. Figure 16(a) shows the peak height ratios for the major ion species in the discharge, while Figure 16(b) shows the optimized mass spectrum. It was found that after optimizing the water vapor pressure inside the flow Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 42 — (H3O+/H20+) 650 W ~«~(H30+)/(02+) 650 W (H3O+/H20+) 685 W - (H30+)/(02+) 685 W - C 3 0 50 100 150 200 250 350 400300 Pressure (mTorr) (a) 0.6 0.5 ■| 0.4 0 10 20 30 40 50 60 70 80 90 Mass (amu) (b) FIG. 16. Characteristics of the MCD discharge: (a) MCD peak height ratios, (b) Optimized mass spectrum. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 43 tube, repositioning of the cavity frequently increased the ion output of the discharge without affecting the peak intensity ratios. While monitoring the H 3 0 + signal in the detection end of the flow tube, the discharge cavity was moved forward and backward until a position was reached which optimized the production of the H 3 0 + ion. Moving the cavity provided a means of fine-tuning the coupling of the microwave radiation to the cavity. While both microwave cavities produced large numbers of ions, the large cavity was more stable in operation and therefore used as the ion source for the MCD- FA. The most likely explanation for this stability was that the modes were much better defined for the larger cavity. It was found that extremely close tolerances are required between the movable wall and the inside surface of the cavity. If the air gap was too large, the large fields present caused arcing between the surfaces that, in turn, caused instabilities in the discharge. 3.13.2 Properties of the Microwave Discharge Plasma In addition to studying the ion output of the MCD plasma, basic parameters of the discharge were also evaluated using optical emission spectroscopy. Radiation from one of the observation holes in the side of the cavity was focused through a one inch UV quartz lens (f=7.1 cm), and an image of the discharge tube was produced on the entrance slit of an Acton Spectra Pro 500i 0.5-meter spectrograph. The image produced at the exit port of the spectrograph was then focused into an Apogee SPH5 CCD camera. Images Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 44 from the camera were downloaded to a computer using Cyanogen Maxim DL 3.0 digital imaging software. The gas temperature of the discharge was calculated using the 0-0 vibrational band at 306.4 nm in the A-X system of the OH' radical. Although the literature provides methods for calculating gas temperatures from numerous molecules, OH' was chosen since water vapor was the major constituent o f our discharge gas. In addition, it was assumed that the rotational temperature of the OH' radical was approximately the same as the gas. This is assumed because the width of the microwave pulse is approximately 5 ms which is more than sufficient time for heavy particle thermalization. The discharge gas was a mixture o f argon and water at approximately equal concentrations at a combined pressure of 150 mTorr. 35000 30000 'E 25000 => 2 20000 I ^ 15000 & (A g 10000 c 5000 305 306 307 308 309 310 311 312 313 314 315 Wavelength (nm) FIG. 17. Rotational spectrum of the (0-0) vibrational band in the A-X system of hydroxyl, Off. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 45 The method used to calculate the rotational temperature is an iso-intensity technique based on the Boltzmann plot of the R 2 branch lines [52]. Figure 17 shows the emission spectrum of the OH' band at 306.4 nm. The technique for calculating the temperature requires that we select the base line (R 2B, B=l,2,3,4) at the beginning of the branch. Using the intensity of this base line we find the two lines in the tail that this intensity falls between. Then using Table 8 from Ref. [52], we can determine the rotational temperature range of the OH' radical and thus the temperature of the gas. Table I shows the results of six measurements made over two separate days. Based on the data we can conclude that the temperature of the gas is7 5 0 ^ ± 1 0 0 if. The most likely reason for the temperature variation was the inability to exactly repeat the pressure inside o f the quartz tube. Since pressure affects the loading of the resonant cavity, small tuning changes in the cavity as well as small changes in magnetron power were required, thereby affecting the temperature o f the discharge. TABLE I. Temperature evaluation, data. Trial T (baseR2l) T (base R2 2 ) T (base R2 3 ) The electron density can be approximately determined by measuring the Stark broadening o f the Hp line in the discharge. Stark broadening of the Hp is caused by the interaction of free electrons with hydrogen atoms such that the higher the electron Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 46 density, the more the line is broadened. The standard technique used for evaluating the electron density uses the empirical formula [53-56] (31) where, (32) (33) (34) Here Tg is the gas temperature in Kelvin, M is the atomic weight in atomic units, A = 4861.33 A, Ws is the Stark half-width, Wd is the Doppler half-width, Wm is the measured half-width, and Wj is the instrumental half-width. All o f the half-widths are measured in A. Figure 18 shows the Hp line measurement in the microwave cavity discharge. The half-width o f the line was measured to be 0.24 A. Using the gas temperature o f 750 K, the half-width due to doppler broadening was calculated to be 0.05 A. The half-width due to instrumental broadening was calculated to be 0.20 A. Using these half widths, and Eqs. (31)-(34), the calculated electron density is Ne = 1.6xl013cm-3. Estimated accuracy of this value is not very high since Stark broadening theory used in this density range is quite inaccurate. Estimated errors for the electron density are as high as a factor of 2 so the electron density is in the range (0 .8 -3 )x l0 13cw-3. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 47 8000 7000 - £ 6000 - £ 5000 - X* 4000 - £• 3000 - S 2 0 0 0 - 1000 - 485.8 485.9 486 486.1 486.2 486.3 486.5486.4 Wavelength (nm) FIG. 18. Hp line in the microwave cavity discharge. The electron temperature was obtained from the ratio o f the intensities between the Arl line at 842.46 nm and the Aril line at 506.2 nm. It should be noted that the electron temperature being calculated is a position and energy averaged value. Before the electron temperature can be evaluated, the ratio of the upper atomic and ionic levels corresponding to the above transitions must first be calculated. This is given by the expression [57] N (35) N + v 8 j where I is the measured line intensity, X is the line wavelength, A is the transition probability, g is the degeneracy of the upper state and s is the instrument spectral efficiency. The spectral efficiency is defined predominately by the relative efficiency of Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 48 the grating, which was blazed to 500 nm. The ratio of the instrument spectral efficiency was 20 and was obtained by a calibration of the spectrograph, grating and detector using a calibrated blackbody source. The other ratios may be obtained from the table of spectral lines in Appendix C. The calculated ratio of populations is — = 1.6xl06. (36) N + Assuming chemical equilibrium of Ar neutrals and ions, and using the Saha- Boltmann equation we may obtain the expression [58] \ | ( E:o„+EL :) N_JK k,T. (37) N + [ 2 \ S + j 2 nmekBTe; where Ne is the electron density, h is Planck’s constant, me is the electron mass, hs is Boltzmann’s constant, Te is the electron temperature, Eion=2.66 eV is the ionization energy of the electron in the upper state in the Arl transition and Eexc~ 19.26 eV is the excitation level of the upper ionic level. Solving this equation for Te gives the result Te= 7500K . This low electron temperature value is consistent with the observation of the low intensity of the Aril lines compared to the Arl lines and the almost nonexistence of Arlll lines. Appendix C shows the relative intensities of the various Ar lines observed. It should be noted that the accuracy of this result is quite sensitive to the electron density and line intensities. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 49 3.2 Flow Tube The flow tube (see Fig. 5) consists of a 304 stainless steel tube, approximately 1 m long with an inner diameter of 5.33 cm and an outer diameter of 6.05 cm. Connections to the detection chamber and ion source were made with ASA-type flanges using viton o- rings. Pressure in the flow tube was measured using a 0-10 Torr MKS Baratron in the reaction region of the flow tube, and a Hastings DV-6 thermocouple tube in the buffer gas relaxation region. All side connections to the flow tube were made using 1/8” NPT threaded nipples welded to the side of the flow tube. The flow tube was carefully cleaned employing vacuum cleaning procedures before it was connected to the vacuum system. It was important to remove all contaminants from the inside surface o f the flow tube to prevent reactions o f the hydronium ions and reactant gases with contaminant molecules. In addition, surface contamination from oils or oxides could interfere with ion diffusion in the tube. The tube was initially cleaned with Alconox detergent to remove any oils and dirt that may have been on the inside surface. It was then soaked for several hours in a warm solution of Citranox detergent to remove any surface oxides. Finally, the flow tube was rinsed with deionized water to remove all detergent residues. The flow tube evacuation chamber, which separated the flow tube from the detection chamber, was used to connect the vacuum pump to the flow tube. The chamber, which was mounted using an o-ring connection to the 8” conflat flange o f the detection chamber, had an inside diameter of 7.8 cm and was approximately 7.8 cm long. On either side o f the chamber, a NW40 quick-flange (QF) nipple was welded for Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 50 connecting to the vacuum pumps. Rigid 3” PVC tubing was used for the vacuum lines and PVC NW40 QF connectors were used for connecting the vacuum lines to the stainless QF connectors on the vacuum chamber. A Stokes model 310-401 roots blower, with a maximum pumping speed of 189 1/s, was used to pump the buffer gas. Backing the roots blower were two Varian SD-700 rotary vane pumps, each with a maximum pumping speed of 13 1/s. A gate valve was also mounted to the roots blower so that the buffer gas flow speed could be regulated. To prevent gas expansion from reducing the ion signal as the gas exited the flow tube and entered the evacuation chamber, the flow tube was extended into the chamber so that the exit o f the flow tube was approximately 1 cm from the sampling orifice. The sampling orifice assembly was also mounted to this chamber, and the electrical feed through, used to bias the sampling orifice was placed in the top of the chamber. 3.2.1 Buffer Gas Flow Properties As shown in Fig. 5, the flow tube can be broken into two separate regions, the buffer gas relaxation region, and the reaction region. The buffer gas relaxation region is required due to the initial turbulent flow caused by the high velocity injection of the buffer gas from the Venturi inlet. Previous studies have shown [59] that the turbulent flow will relax into laminar flow after approximately 0.5 m. Laminar flow in the reaction region is required in order to obtain constant ion velocity, and thus constant reaction times. Evidence of turbulent flow typically manifests itself as erratic or non-exponential Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 51 hydronium ion decay curves when measuring reaction rate coefficients as well as large pressure differentials between the ends of the tube. The primary feature of any buffer gas is that it is non-reactive with the ions and molecules present in the flow tube. In most experiments [2,6] helium was used to prevent collisional dissociation between the buffer gas molecules and the ions during injection from the ion source because of its small mass. The energy in the center of mass frame is given by Ecm=EL--^ ~ , (38) mi +mb where E l is the energy in the lab frame, mb is the mass o f the buffer gas molecule, and m* is the mass o f the ion [60]. From this relationship we can see that the larger the mass of the buffer gas atoms, the larger the energy will be in the center of mass frame. The use of helium is especially important in the SIFT [6] where the ions are being injected into the flow tube at energies typically greater than 10 eV. The buffer gas used for all reactions discussed presently was argon. Helium was tested in the MCD-FA, but back-streaming o f small amounts o f helium into the microwave discharge ion source caused the discharge to run unstably at safe magnetron powers. The unstable operation of the discharge was due to two factors: (1) the high ionization energy of helium (Eion=24 eV), and (2) the discharge was run in a pulsed (60 Hz) mode. Because the ionization energy of helium is so high, not every pulse of microwave energy resulted in a breakdown of the gas. The use o f argon (Ei0n—16 eV) as the buffer gas eliminated the instabilities in the discharge. Since the injection of the ions into the flow tube o f the MCD-FA occurs as a result of thermal diffusion, collisional dissociation was not expected and it was not observed. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 52 As discussed previously, the pressure in the flow tube was measured using a Baratron in the reaction region and a thermocouple in the relaxation region. The thermocouple tube was calibrated using the Baratron by turning off the roots blower (the backing pumps were left on), throttling the gate valve, and injecting argon through a reactant gas injector port, and not the Venturi inlet. In this way the buffer gas velocity was sufficiently small so that no measurable pressure differential existed between the ends of the flow tube. At 340 mTorr, the typical flow tube operating pressure, the pressure differential between the two regions was approximately 15 mTorr. The buffer gas flow rate was controlled using an Aalborg GFC mass flow controller. Using the measured rate it was possible to calculate the buffer gas flow velocity in the flow tube using the relation v ,=pA 4 > <39) where Q is the flow rate in Torr cm3/s (1 standard liter per minute (slm) = 12,700 Torr cm3/s), p is the flow tube pressure in Torr, and A is the crosssectional area of the flow tube (A=22.35 cm2). Figure 19 shows a plot of the buffer gas velocity as a function of flow rate in the reaction region of the flow tube. At the standard operating pressure of approximately 350 mTorr the buffer gas velocity is approximately 1600 cm/s. The estimated error on the buffer gas velocity is 5% and is due to the uncertainty in the mass flow measurement (±0.23 slm) and the uncertainty in the pressure measurement (± 3%). The line through the points is a best fit of the data. It should be noted that this flow velocity is much smaller than the flow velocities of other FA and SIFT experiments due to the small capacity roots blower used. The primary advantage o f using larger flow velocities is that losses of the hydronium ion, due to radial diffusion, is reduced. On the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 53 other hand, larger flow velocities require higher flow rates. This can produce considerably more turbulent flow and even shock waves [60] from supersonic injection velocities from the Venturi inlet. 5000.0 4000.0 2? 3000.0 a Ion Velocity • Buffer Gas Velocity — Linear (Ion Velocity) Linear (Buffer Gas Velocity) 2000.0 1000.0 0.80 0.90 1.00 1.10 1.20 1.30 1.40 1.50 1.60 Buffer Gas Flow Rate (slm) FIG. 19. Ion and buffer gas axial velocities in the flow tube as a function o f buffer gas injection rate. In addition to the flow velocity of the buffer gas, the flow velocity of the hydronium ions was also measured. Two stainless steel probes, with diameters of 1.27 mm, and separated by 45.0 cm, were inserted into the flow tube. A square wave pulse, oscillating between 0 volts and -15 V was then used to modulate the hydronium ion signal as measured in the detection chamber. Using a digital oscilloscope, and triggering off of the square wave, a time delay could be observed between the pulse edge and the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 54 hydronium ion signal. Figure 20 shows an example of the signal modulation. The time difference between the rising edge of the function generator signal and the rising edge of the ion signal is the time it took for the ions to reach the detector. To obtain the average ion velocity, the difference in pulse delays at two points, separated by 45.0 cm, was measured. Figure 19 shows the measured ion velocity as a function of buffer gas flow rate, and how it compares to the buffer gas velocity. Due to the uncertainty of the rising edge of the ion pulse, an estimated error of ±10% was placed on each velocity measurement. The uncertainty in the mass flow measurement was specified by the manufacturer to be ±0.23 slm and applies to all points on the graph. The lines on the plots are a best fit line to the data. i. l j ... ‘i.4,:. i i ; •jM-'.-r • s -i....- fkit", Hfttt, i, : 4 '■''tlilljiif n » 1i ■ m i ' !m : i Mr'" — : : : : : & :::::: S.OOmV ^C h2f s.'o o v M l0 .0 ins W idth " C h ' 2 ISP s.oomv ^ch2j 5,oo v Ivfio.oms! ; width ch2 (a) (b) FIG. 20. Ion modulation in the flow tube: (a) Modulation of the ion signal due to probe 1, (b) Modulation of the ion signal due to probe 2. The flow tube typically operated at a pressure of approximately 350 mTorr. The operating pressure o f the flow tube was based upon several factors. It has been found in other flow tube experiments that at pressures less than -150 mTorr the gas flow becomes Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 55 molecular (mean free path « flow tube diameter), causing large pressure differentials in the flow tube [34]. This molecular flow cannot be accurately characterized as having an average flow velocity which is critical to the accurate measurement of reaction rates. In addition, to obtain accurate reaction rate coefficients, the ions exiting from the ion source must undergo a sufficient number of collisions for thermalization with the buffer gas. At higher flow tube pressures (larger than 0.4 Torr), the detection chamber pressure becomes too high (larger than 10‘5 Torr) for the QMS and Channeltron to operate safely. 3.2.2 Venturi Inlet Injection of the argon buffer gas into the flow tube occurs through a specially designed inlet, which is typically referred to as the Venturi inlet. The Venturi inlet design, as used in flow tube experiments, was originally developed by Smith and Adams [6] to prevent the back-streaming of the buffer gas into the SIFT chamber of the original SIFT apparatus. For the standard flowing afterglow (FA) apparatus back-streaming is not a consideration since the discharge that is used to produce the ions is generated in the flow tube. In the SIFT and the MCD-FA device described here, the ions are produced in a separate chamber and back-streaming must be kept to a minimum to prevent swamping the discharge region (MCD-FA) or SIFT chamber with the buffer gas. Since the flow dynamics of the buffer gas are critical to the proper operation of the flow tube, considerable study has been done on the optimal design of the Venturi inlet [61-63]. There are two major designs, shown in Fig. 21, used as injection ports: (a) the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 56 hole array port, and (b) the annular slit port. Both designs inject the buffer gas on, or at a small angle relative to the axis of the flow tube. The hole-type injector consists of a series o f holes surrounding the ion entrance, whereas the annular-type injector has an annular slit surrounding the ion injection entrance. In addition, both designs inject the gas through a port(s) that surround the hole through which ions enter from the ion source. Inside Surface ofFlowtube / X / \ Ion Entrance Hole Ion Entrance Hole © Gas Injection Gas Injection Hole Array Annulus / (a) (b) FIG. 21. Venturi inlets: (a) Hole-type injector, (b) Annular-type injector. The studies on Venturi inlet designs have shown that, in general, the annular type inlet provides slightly better differential pumping between the chambers. On the other hand, this slight performance increase must be weighed against the extremely difficult construction and alignment of the annular slit design, as compared to the simple design of the hole injector. Recent work has concluded [62,63] that there were no major advantages to using the annular injector except in cases where the pressure differential between the source chamber and the flow tube is a critical factor. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 57 o \ ^ Buffer Gas (Entrance Tube ,Viton 0-R ings Ion Injection Hole Buffer Gas Injection Holes FIG. 22. Venturi inlet used in the MCD-FA. Since small amounts o f the argon buffer gas had no measurable effect on the ion output of the microwave cavity discharge, the hole injector design was used as the Venturi inlet. Figure 22 shows a diagram of the Venturi inlet used in the MCD-FA. The inlet was made in two parts from 304 stainless steel using viton o-rings to maintain vacuum integrity. The ion injection hole in the center o f the inlet was 1.0 mm. The gas injection holes were 0.35 mm in diameter and the hole pattern had a radius of 4.3 mm. The gas was injected into the flange from a 0.25” stainless steel tube welded onto the side o f the flange. Swagelok connection was used to connect to the stainless tubing to the mass flow controller. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 58 3.2.3 Reactant Gas Injection Ports Injection of the reactant gas into the flow tube occurs along a series of injection ports located along the side of the flow tube. Figure 23 shows a diagram of the injection port configuration. The reactant gas was injected using a 1/8” stainless steel tube that extended to the center axis of the flow tube. The end of the tube was beveled in an attempt to more evenly distribute the reactant gas radially into the buffer gas. The injection ports were held in place using Ultra-Torr™ connectors that were connected to the flow tube with NPT threads with Teflon thread tape. Connection to the reactant gas injectors were made using Swagelok connectors, which in turn were connected to the reactant gas input manifold. The manifold, shown in Fig. 24, used a series of valves to select the desired injection port, and was connected to the mass flow controller used to control the reactant gas injection flow rate. To prevent the back-streaming of oil from the mechanical pumps, an air purge metering valve was used to bleed air into the flow tube when the reactant gas mass flow controller was turned off. Injection port 3, located 23 cm from the sampling orifice, was used for all the proton transfer reaction rates measured presently. As discussed in Section 2, the proton transfer reaction rates are obtained by measuring the decay o f the hydronium ion as a function of reactant gas number density. If the port is located too far from the sampling orifice, then the reaction times will be too long and the decay curve will drop too fast to be measured. On the other hand, if the injection port is too close, the signal will not have decayed sufficiently (preferably 2 orders of magnitude) to accurately measure the decay. It was found empirically that port 3 provided the best decay curves. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 59 Injector Port Ultra-Torr Connector To NPT Thread Flow Tube Buffer Gas Flow FIG. 23. Injector port configuration. Flow Tube Buffer G as Flow Air Purge Plug Metering Valve Valve Plug Valve M ass Flow Reactant Controller G as Plug Valve FIG. 24. Injector ports gas manifold. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 60 3.3 Detection System The vacuum system housing the product ion detection system (see Figure 5) consisted of a stainless steel chamber with an inside diameter o f approximately 15 cm and a length of approximately 50 cm. The system was pumped by an NRC 4” diffusion pump with a maximum pumping speed of 1200 1/s. Neovac Sy fluid was used in the diffusion pump. The diffusion pump was backed by a Welch 1397 mechanical pump which had a maximum pumping speed o f 8.4 1/s. To prevent back-streaming of oil into the detection region a foreline trap was placed between the mechanical pump and the diffusion pump, and a Varian 334 water cooled baffle was placed between the diffusion pump and detection chamber. A gate valve was place between the cold water baffle and the detection chamber. The valve was partially closed when the experiment was not running to help reduce possible oil contamination of the detection system. Pressure in the vacuum chamber was measured using a Huntington IK-100 ion gauge tube with a Varian 843 controller. Figure 25 shows a schematic of the detection system. Ions in the flow tube are first sampled by a biased sampling orifice plate at the end o f the flow tube. Once sampled the ions are accelerated and focused into the QMS which then filters the ions according to their mass to charge ratio. The filtered ions are then detected using an off axis Channeltron detector. In order for the quadrupole mass spectrometer (QMS) to operate correctly, as well as to prevent damage to the Channeltron detector, the pressure in the detection chamber had to be maintained below ~10'5 Torr. The pressure Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 61 differential between the flow tube and the detection chamber was maintained by a 350 /an sampling orifice. Roots 8" Conflat Flange Blower ion Channeltron Housing Optics Channeltron Sampling QMS Housing Orifice Roots Blower FIG. 25. Schematic diagram of detection system. 3.3.1 Ion Optics Before the ions entered the quadrupole mass spectrometer (QMS) they were accelerated and focused using a set of custom built ion optics. The first optical element in the assembly was the biased sampling orifice used to maintain the pressure difference between the flow tube and the detection chamber. The orifice, purchased from Lenox Laser Corporation and originally designed to be used as a controlled leak VCR gasket, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 62 was made from electro-polished 316 stainless steel.. It had a diameter of 20 mm and a thickness of 0.7 mm. At the very center of the orifice the stainless steel had been etched away so that the sampling hole of 100 pm was laser drilled through a thickness of approximately 130 pm. Although the gasket was originally purchased with a 100 p m hole, it was found that a 350 pm hole, drilled with a #80 drill bit provided a larger signal and was able to maintain the required pressure difference between the flow tube and the detection chamber. - ■ Top Hat E lectrode A lum ina S p a c e rs F o cu sin g ' ; Electrode A lum ina - Support Rod Ion Injection »- Alumina Direction ; S p a c e rs E lectrode FIG. 26. Ion optics assembly. To optimize the signal, the sampling orifice was electrically isolated and could be biased relative to ground. The orifice was isolated from ground using an alumina spacer, and the entire assembly was held together using Torr-Seal epoxy resin, which also provided a vacuum seal between the two chambers. Because the alignment of the sampling orifice hole was critical to the effectiveness of the entire ion optics assembly, a holder was built to secure the pieces together as the Torr-Seal cured. Once curing was complete, a 316 stainless steel wire, used for biasing, was spot welded to the outside edge Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 63 of the orifice optic and was connected to an electrical feed-through in the flow tube evacuation chamber. The wire was carefully positioned so as not to interfere with ion collection o f the orifice. The biasing potential for the sampling orifice was provided by a 0-25 V, Hewlett-Packard 6216A DC power supply. Once the ions entered the detection chamber through the sampling orifice, they were accelerated and focused using three electrostatic lenses. Each optic plate was machined from 304 stainless steel and was designed to collect as many ions entering the chamber as possible. Figure 26 shows a photograph and exploded drawing of the lens assembly. It has been found [6] that by placing the first element as close as possible to the sampling orifice the signal can be significantly increased. Due to the limited space available behind the sampling orifice, this optical element must be designed in such a way that it is typically referred to as the “top hat” electrode. The middle optical element in the assembly is a planar element whose primary purpose is to focus the ions. As seen in Figure 26 the focusing lens has a cylindrical extension which acts to shield the ions from the supporting alumina rods and spacers. The third element in the assembly, called the injector lens, accelerates the ions before they enter the QMS. To minimize the effects of the QMS fringing fields, this optical element was designed as a planar optic with a cylindrical extension that protrudes approximately 5 mm down the axis of the QMS. hi this configuration the curvature of the QMS fringing fields is minimized so that the trajectories of the ions are perturbed as little as possible as they enter the filter and is biased relative to ground using a 316 stainless steel wire which has been spot welded to the outside edge o f the lens. Each wire extends through fiberglass insulation to an electrical feed-through in the detection chamber. The ion optics biasing circuit was a set Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 64 of potentiometers connected in parallel to a Hewlett Packard 0-100Y 6212A DC power supply. Before the optic assembly was built, the design was modeled on a computer using SIMION 4.0. This program allows the user to graphically input the electrode surfaces, and then model ion and electron trajectories based upon user specified initial conditions. Although the program is limited to two dimensional modeling of the electrodes, it can render ion trajectories in three dimensions as long as the electrodes are symmetric (either cylindrical or planar) around the ion trajectory axis. Ion trajectories are calculated using a finite grid technique to define all points in space. The largest allowable grid allowed by SIMION 4.0 for drawing electrode surfaces is 16000 points, although the grid used for trajectory calculations can be considerably finer so that trajectories can be accurately modeled. In order to obtain the greatest resolution for the electrodes, the area modeled was limited to 3.78 cm wide by 1.59 cm tall, with an electrode grid point spacing of 0.194 mm/point. Since cylindrical symmetry was specified when defining the array, the actual area of the modeled region was 3.78 cm wide by 3.18 cm tall. Trajectories for several ions were calculated assuming that they entered the detection system at room temperature (0.025 eV) and at angles ranging from -30° to +30° relative to the symmetry axis. With the initial conditions o f the ion trajectories established, the electrode potentials were then varied until the ions appeared to be focused into the QMS. Once the optics assembly was built and tested, it was found that the potentials modeled provided only a weak signal. The potentials used for all rate measurements were obtained by setting the QMS to filter only the hydronium ions, then adjusting the potentials until the maximum signal was obtained. Figure 27(a) shows the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 65 modeled trajectories o f the ions when the electrodes were set to the initially modeled potentials and Fig. 27(b) shows the trajectories when the electrodes were set to the optimized potentials. It is immediately apparent that the trajectories are much more tightly focused for the experimentally optimized potentials. Focusing F ocusing Electrode (-40.0 V) Top Hat Injector Electrode Electrode (-20.0 V) (-35.6 V) jSampling Sampling I Ormce OrlSce I (0.0 V) (-7.7 V) (a) (b) FIG 27. SIMION modeling of ion trajectories: (a) Initial ion optic potential modeled on SIMION, (b) Experimentally optimized ion optic potential modeled on SIMION. 3.3.2 Quadrupole Mass Spectrometer Ion selection was accomplished using a quadrupole mass spectrometer (QMS). The theory of the QMS is well developed [64] and will not be discussed here, but a brief overview of the operation is discussed in Appendix A. While mass spectrometers, such as magnetic sector types, may have higher resolutions, the QMS was chosen due to its ability to scan large mass ranges quickly, its mechanical simplicity and its small size. Figure 28 shows the custom built, flange mounted QMS assembly with the Channeltron housing on the back. For a detailed description of the QMS assembly see Figure 25. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 66 FIG. 28. Mounted QMS assembly and Channeltron housing. The QMS used in the ion detection system was taken from a Hewlett-Packard 5970 MSD Gas-Chromatograph Mass Spectrometer. Each rod in the assembly is 8.000” long, has a diameter of 0.500”, and is made of stainless steel. The four rods are symmetrically mounted around the ion beam axis using two alumina holders. The entire rod assembly was secured inside of a stainless steel tube that in turn was mounted to an 8” conflate flange. Since the optics assembly is attached directly to the QMS mount, the entire system is self aligning with only minor adjustments needed once the detection assembly is put together. Electrical connections to the QMS were made with stainless steel wires that were spot welded to the outside surface of the rods. These wires run through alumina insulation tubes to a high voltage electrical feed-through in the top of the detection chamber. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 67 Power to the QMS is supplied by an Extranuclear Laboratories QMS Power Supply. The QMS power supply is divided into three separate units: the quadrupole control, the radio frequency (RF) power supply and the high-Q head. The quadrupole control contains the electronics for mass scanning as well as the feedback amplifier used to stabilize the output o f the RF power supply. The RF power supply generates the RF signal using a resonant LC circuit signal which is then amplified using a pentode in an electron-coupled oscillator circuit. The output o f the RF power supply is then sent to the high-Q head. The high-Q head is a high-Q, tunable transformer which steps up the voltage to a suitable level to drive the QMS. As discussed in Appendix A, the mass range o f the QMS is limited by the size of the QMS, the operating frequency, and the maximum output voltage. The ion mass that the quadrupole will transmit is given by the expression V_ m ~ T T > (40) 1219S>-> r\ 2 r- J1 2 5 v 7 where m is the mass of the ion in amu, V is the peak RF voltage in volts, r0 is the radius of a circle passing through the center of the QMS poles in cm, and / is the frequency of the RF signal in MHz. Using Eq. (40) it is possible to calculate the approximate mass range of the QMS. The maximum output voltage of the QMS power supply is limited by the high-Q head model used, and the capacitance of the QMS. In the operating manual o f the QMS power supply, the operating frequency o f the different high-Q heads as a function o f capacitance is shown. In addition, the operating frequency of the high-Q as a function o f capacitance is also shown, so it possible to measure both the capacitance and the peak RF voltage by simply measuring the operating frequency o f the QMS. Using an oscilloscope and an Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 68 antenna next to the QMS output an operating frequency of 2.6 MHz was measured. For the D1 High-Q head used here, this corresponds to a capacitance of approximately 65 pF and a peak RF voltage o f approximately 3000 V. Using this value for the peak voltage and rQ = 0.55 cm in Eq. (40) gives a maximum mass of 200 amu. Because of the large gas load of the detection chamber, as well as the confined space o f the QMS and ion optics assembly, there was concern of large pressure gradients and therefore of the arcing between surfaces. As a precaution, the maximum mass was typically limited to less than 150 amu. 3.3.3 Signal Detection and Data Acquisition The ions were detected using a Galileo model 4772 Channeltron. This Channeltron was specifically designed to be used as a detector for QMS systems, as the detection cone is mounted off axis inside the alumina assembly. This configuration prevents background counts from the UV radiation from the ion source. Power to the Channeltron was supplied by a Keithley 247 power supply. Since the 4772 Channeltron was run in the pulse counting mode, it was necessary to measure the pulse counts as a function of operating voltage, with the ion current in the flow tube held constant. If the voltage was too low then the electron multiplication efficiency would be too low and not all ions hitting the surface of the Channeltron would be counted. On the other hand, if the voltage was too high then electron multiplication would be too high and one ion could be counted as multiple pulses. The manual states Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 69 that the Channeltron should operate approximately 25 V beyond the first elbow in the voltage-gain curve, in the “plateau” region of the curve. Figure 29 shows the voltage gain curve measured for the Channeltron. Based on this curve, an operating voltage of -2600 V was used. - 140000 - 130000 - 120000 - 110000 -3T V* - 100000 B c 3 - 90000 O - 80000 - 70000 i - 60000 -2700 -2600 -2500 -2400 -2300 -2200 -2100 -2000 Channeltron Voltage (volts) FIG 29. Channeltron counts as a function of operating voltage. To prevent false counts caused by the ion gauge, the Channeltron was mounted inside a stainless steel housing that was connected directly to the back of the QMS housing. A 0.188” hole, for the filtered ions to pass through, was drilled through the front of the assembly. As with the ion optics, the Channeltron housing was self aligning and required only minimal adjustments once it was attached. The Channeltron required three electrical connections to operate: a high voltage connection for power, a Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 70 signal output, and a deflection plate biasing connection used to steer the ions into the Channeltron. Electrical feed-throughs in the Channeltron housing were drilled through the back and alumina rods were used as insulation. Stainless steel wire was run from the Channeltron housing to electrical feed-throughs in the rear o f the detection chamber. Connected directly to the signal output of the detection chamber feed-through was an Ortec model 109 PC preamplifier. The output of this preamplifier was initially sent directly into counting electronics, but had to be modified due to the noise generated by the high-voltage RF of the QMS. Figure 30(a) shows the output of the preamplifier sent directly to the 50 Q input of an oscilloscope. Figure 30(b) shows a pulse once the QMS was turned on. In many cases single pulses were counted as multiple pulses due to the RF signal riding on top of this pulse. In order to eliminate this problem, a twin-T notch filter circuit was designed and built, which eliminated the RF noise. Figure 31(a) shows the circuit, while Figure 31(b) shows the calculated and measured filtering properties. To compensate for the impedance mismatch caused by the filter, an additional preamplifier (Ortec 113) was used before the signal was sent to the counting electronics. Figure 30(c) shows the signal with the QMS on, after passing through the filtering circuit. 'jffSniir (a) (b) (c) FIG. 30. Channeltron pulses: (a) Pulse with QMS off and no filter, (b) Pulse with QMS on and no filter, (c) Pulse with QMS on and filter on. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 71 1000 10000 100000 1000000 10000000 4.39 kfi 4.39 kCt -w v AAA ------10 ( i— • 12 pF ♦ Measured Frequency Response — Calculated Frequency Response Frequency (Hz) (a) (b) FIG 31. Twin-T notch filter circuit: (a) Ciruit, (b) Calculated and measured frequency response of the notch filter. Once the RF noise had been removed from the signal, it was sent to an Ortec 421 integral discriminator to eliminate ground noise and condition the pulse for digital counting. The integral discriminator output was a 50 ns TTL pulse when the input pulse reached a preset discrimination pulse amplitude. In order to set the discriminator level, the QMS was set to transmit hydronium ions only, and then the output of the integral discriminator was measured as a function of the discriminator level. Figure 32 shows the resulting plot. Based upon this, the discriminator was set to 0.4 V. Data acquisition and mass scanning of the QMS was controlled using a National Instruments PCI-6035E data acquisition card (DAQ) and a program written in Lab View. To control the QMS scanning properties, one of the analog outputs of the DAQ card was used on the external command input of the QMS controller. The user inputs the step width that controls the resolution of the scan, the starting voltage that defines at what mass the QMS will start scanning, and the number of steps that defines the scan width. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 72 Counting of Channeltron pulses was accomplished using one of the onboard counters on the DAQ card. The user inputs the counting time per step. Longer counting times allow better averaging o f the signal, but required a longer time per scan. Once the scan was complete, the program would output the data to a file. 1000000 900000 800000 700000 m 600000 ^4-J 500000 C O 400000 300000 200000 100000 0 0.2 0.4 0.6 0.8 1 Discriminator Level (Volts) FIG. 32. Channeltron count rate as a function of discriminator level. The Lab View program was not capable of calibrating the data to the mass spectrum, so the generated file had to be exported to a separate program (Microsoft Excel), and calibrated using a known set of peaks. The peaks chosen, due to their large separations, were the hydronium ion (H 3 OQ, molecular oxygen ( 0 2 +), and protonated acetone (C 3H7 OQ. The quadrupole controller remained linear over the entire scan width. Figure 33 shows a calibration spectrum. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 73 o o o CT» I o co 00 O + ” 0 CO 1 o h“ + Oc~ o X CD co o +o 3 co o X CM E CO X ^ 8 o + O LO oco m X (/) ( 8 o - CM o X Tj- +o CO X + QMS. the used to calibrate FIG. 33. Spectrum CM o o CO + O o co CM X O O o o o o o O o o o o h- CD lO CO CM (s 2)/s*unoo Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 74 Section 4 RESULTS AND DISCUSSION Gas phase charge transfer reactions play an extremely important role in the chemistry o f atmospheres, gas discharges, and interstellar clouds. In order to correctly model these phenomena it is crucial to have accurate measurements of the rate coefficients o f the reactions involved. Using the techniques described in Section 2, the proton transfer reaction rate coefficients were measured for the reaction between the hydronium ion and the three simplest alkenes: ethene, propene, and 1-butene. In addition, the proton transfer reaction rate coefficient o f the hydronium ion and hydrogen sulfide were measured. These molecules were chosen due to the important role they play as atmospheric pollutants. The rate coefficients of ethene and propene have been measured previously, and were used to test the MCD-FA. The reaction of the hydronium ion with 1-butene has not been previously measured. The hydrogen sulfide reaction has been measured two times previously, but a large discrepancy existed between these measurements. In addition to the measurement of reaction rate coefficients, the MCD-FA has the potential to become a powerful tool in the detection and quantification of trace compounds in atmospheric air. To demonstrate the effectiveness o f this method, car exhaust and exhaled alveolar air were sampled and introduced into the MCD-FA. The mass spectra from these samples show a large number of trace compounds. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 75 4.1 Error Analysis As discussed in Section 2, the reaction rate coefficient can be obtained by assuming that the decay of hydronium ions in the flow tube is governed by Eq. (21) yielding the solution, (41) where v,- is the hydronium ion axial flow velocity, R is the decay rate of the hydronium ion as a function o f the reactant gas number density, and I is the flow tube reaction length. For each reaction, ten separate measurements o f the decay rate as a function of reactant gas number density were made and the average of these measurements was used as the decay rate for the calculation of the reaction rate. The calculated error in the decay constant was obtained by taking the standard deviation o f the ten measurements using the expression (42) where N is the total number of measurements and x is the mean value of the rate. This represents the standard deviation o f each measurement. The uncertainty in the average decay rate was taken to be the standard deviation of the mean, (43) Vat ' The uncertainty in the measured decay rate arises from several sources o f error. The largest source of error in the decay curves are due to the uncertainty in the flow rate Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 76 o f the injected reactant gas, which was specified by the manufacturer to be ±0.75 seem. In addition, minor ion velocity variations due to fluctuations in the pumping speeds of the flow tube (±4 m/s) also had an observable affect on the decay curves. While small compared to the systematic sources of error, the statistical error in the ion counts ( 4 n ) also contributed to the uncertainty in the decay constant. The uncertainty in the flow velocity and the reaction length were taken to be the same for all the measured reactions. The ion flow velocity, discussed in Section 3, was measured to be 37 m/s with uncertainly of Svt = 4 m/s. The reaction length was measured to be 23 cm with an uncertainty in the measurement of approximately Sz — 0.5 cm. The temperature in the laboratory was monitored during each measurement using a standard laboratory mercury thermometer and was found to be 23 ± 1 °C. In most cases, the calculated uncertainty in the present reaction rate coefficient measurements, using the previously discussed uncertainties, is approximately 10%. As a conservative estimate, many measurements in the literature are quoted to have an uncertainty o f ± 2 0 % to take into account effects such as ion diffusion in the reaction region of the flow tube as well as uncertainty in the flow dynamics [ 6 ]. Since the effects of diffusion and flow dynamics could not be accurately measured in the present work, the estimated error in the reaction rate coefficients will be given as ± 2 0 %. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 77 4.2 Ethene (C 2H 4) Ethene is the simplest of the alkenes, consisting of 2 carbon atoms, joined by a double bond, and 4 hydrogen atoms. Table II summarizes the properties o f ethene.The units shown in Table II are most frequently used for the calculation of collision rates (see Section 2), but conversion to SI units is possible using the factors provided in Appendix D. Due to the simplicity of the ethene molecule there are only two channels available for the reaction with the hydronium ion, H30 + + C2H4 -> C2H; + H20 (44) H30+ + C2H4 +A r—> C2H4 • H30 + + Ar (45) The proton affinity of the ethene molecule is smaller than the proton affinity of the hydronium ion so the proton transfer reaction (44) is endothermic and will occur at much lower rate than the collision rate. In addition to the proton transfer reaction, the three body recombination reaction (45) was also observed. TABLE II. Properties of ethene at STP. Dipole Proton Melting Boiling Polarizability Density Structure Moment Affinity Point Point (xlO'24 cm3) (g/ml) (D) (eV) (°C) (°Q -o/X / o X- II 4.252 0.015 7.05 -169 -103.7 0.5678 Although the proton transfer reaction is endothermic, the difference between the proton affinities is only 0 . 1 eV so the reaction occurs due to the thermal distribution of Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 78 kinetic energies of the molecules. Figure 34 shows the velocity distribution of the ethene molecules at a temperature of 296 K. This minimum activation energy corresponds to a speed of 730 m/s for the ethene molecules. Integrating the Maxwell-Boltzmann distribution over all speeds greater than this gives the probability of a particles having a speed greater than 730 m/s \ 2 L*J m v — \ v 2e™ dv = 0.05, (46) 73 0 where T—296 K, k is the Boltzmann constant, and m is the mass of the ethene molecule. Using the ADO theory discussed in Section 2, the calculated collision rate is &c=1.4xl0 ' 9 cm3/s. If it is assumed that the proton transfer reaction will occur for every collision with sufficient kinetic energy then the estimated reaction rate is, kr = p-kc = 1 x\0~ucm3 / s . 8 7 6 ■ r 5 o E 4 3 2 1 0 0 100 200 300 400 500 600 700 800 900 1000 1100 1200 velocity (m/s) FIG 34. Maxwell-Boltzmann velocity distributiony(v) of the ethene molecules at 296 K. The hatched portion of the distribution shows the molecules with velocities corresponding to the activation energy of the proton transfer reaction (~5%). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 79 In order to measure the proton transfer reaction rate, a mixture of 0.5040% (5040 ppm) ethene in nitrogen was used. The gas mixture was injected into the flow tube at rates between zero and 50 standard ml/min, which corresponded to a number density in the flow tube between zero and 2.5xl0 12 molecules/cm3. Figure 35 shows the mass spectrum in the flow tube when the reactant gas flow rate was 30 seem. A reaction rate coefficient of £,.=(6.9±1.4)xl0'u cm3/s was measured using an average of 10 measurements over a period of two days at a temperature of 296±1 K. The error on this measurement is primarily a result of the uncertainty in the ion velocity in the flow tube (-10%). Figure 36 shows the decay of hydronium ions as the ethene flow is increased into the flow tube, and Table III summarizes the results. It is important to point out that the measured reaction rate is a combination of both the three-body reaction rate as well as the proton transfer reaction rate, H30+ + C2H4 -> Products, (47) where Products respresents both reaction (45) and (46). The proton transfer reaction rate o f kr = (6.3±1.6)xl0'n cm 3/s, measured by Bohme and Mackay [65], was obtained by measuring the reaction rate coefficient for reaction (47) at various flow tube pressures and extrapolating the curve back to zero pressure to obtain the rate coefficient for the proton transfer reaction only. This was not possible in our experiment due to the limited pumping range of our roots blower. The collision rate was calculated using ADO theory. TABLE III. Rate coefficients for the reaction of hydronium with ethene in units of (cm3/s). Reaction Rate Coefficient Proton Transfer Reaction Rate Collision Rate Coefficient (present) Coefficient (T651) (calculated) (6.9±1.4)xl0'u (6.3 ± 1.6)xl0'n 1.4x1 O'9 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 80 Figure 37 shows the measured product ions in the flow tube as a function of the injected reactant gas. The three-body reaction (45) appears to dominate at higher flow rates while at low reactant gas inputs the proton transfer reaction (44) appears to dominate. This is a result of the reaction of C2 H5 * with the other ethene molecules producing the secondary ions observed in the mass spectrum, c 2h ; + c 2h 4 c 2h ; + c h 4 (48) c 2h ; + c 2h 4-> c 4h ;. m The secondary ion formed in reaction (48) then reacts with ethene to form the tertiary product CsH^, C3H; + C2H4->CsH;. (50) The peak at 28 amu is due to the reaction of ethene with C> 2 +. o ; + c 2h 4 -> c 2h ; +o2. (51) The source of C>2 + in flow tube experiments has been discussed in Ref. [13]. It has been found that contaminant O 2 in the buffer gas is photoionized by UV from the gas discharge ion source. It was found by separate measurements that the C> 2+ peak could be minimized in the flow tube by running the gas discharge in water vapor at pressures greater than 100 mTorr. However, at pressures greater than 150 mTorr the measured density of water cluster ions started to become a problem. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 81 seem. 30 is mixture gas ethene the of rate injection the when spectrum Mass 35. FIG. (amu) Mass Mass o o o a> o oo o o co o o CO o CM 1 1 o K K LO <- /f X o If) 5 1 X X ( o' + + OO o in CM m X X o o CM CO X o £ o o Mr* CM o in o in CM i)i^unoo (s CO o o o o o o co oo o IO o o o o o o O O iO "ft c% X o ooos Reproduced with permission ofthe copyright owner. Further reproduction prohibited without permission. to oo 1 8 1 4 1 6 12 (x1011 cm'3) (x1011 10 8 Gas Number Density 6 Reactant FIG. FIG. 36. Decay of ionhydronium signal as a function of ethene density. number Note the that statistical error on the ionpoint. counts error is the data than density smaller of number in the ethene is The uncertainty due to points. data to all the applies and measurement flow in the controller uncertainty the 100000 120000 1 6 0 0 0 0 1 4 0 0 0 0 1 8 0 0 0 0 220000 200000 2 4 0 0 0 0 2 6 0 0 0 0 3 o n o Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 83 10000 1000 100 t I 10 ♦ H30+ (19 amu) m C2H4+ (28 amu) A C2H5+ (29amu) x C3H5+ (41 amu) X C4H9+ (45 amu) • C2H4:H30+ (47 amu) + C2H4:C2H5+ (57 amu) a C5H9+ (69 amu) 10 15 20 25 30 Reactant Flow Rate (seem) FIG. 37. Product ions as a function of the ethene mixture injection rate. The uncertainty in the flow is ±0.75 seem. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 84 4.3 Propene ( C 3H 6) There are three channels available for the reaction between the hydronium ion and propene, H30+ + C3H6 -> C3H; + H 20 , (52) H30+ + C3H6 + Ar-> H30+ ■ C3H6 + A r, (53) H30+ + C3H6 - »C3H+5 +H 2+ H20 . (54) Reaction (52) is the proton transfer reaction. Unlike ethene, the proton affinity of propene is larger than water’s affinity and the reaction is exothermic. As discussed previously, exothermic proton transfer reactions are extremely efficient and typically occur at approximately the collision rate. Reaction (53) is a three-body recombination reaction, with the third body being an argon atom. Reaction (54) is a dissociative reaction that is common to alkenes and becomes more probable as the alkene size increases [17]. The propene molecule consists of an ethene molecule with one methyl group replacing a hydrogen atom. Like ethene, the simplicity of the propene molecule does not allow any structural isomers. Table IV lists the molecular properties. A mixture of 0.4900% (4900 ppm) of propene in nitrogen, prepared by BOC Gases, was used for all measurements. The gas mixture was injected into the flow tube at rates between zero and 10 standard ml/min. This corresponded to a maximum propene number density inside the flow tube o f approximately 5 x l0 n molecules/cm3. Injection rates were limited by the count rates in the detection chamber. If the counts became too few, longer counting times were required to get better statistical certainty. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 85 TABLE IV. Properties of propene at STP. Polarizabilily “<>* “elti”8 Denaify Structure . „ „ „ 2 4 3, Moment Affinity Point Point , , n ______(x 1 0 cm ) (p) (eV/ (oc) (oc) (g/ml) H H \c—c / / \ 6.26 0.366 7.79 -185.2 -47.6 0.505 H H \ H Figure 38 shows the mass spectrum in the flow tube when the reactant gas flow rate was 6.0 seem while Figure 39 shows a representative decay curve for the hydronium ion as a function of the propene number density. The rate coefficient was measured to be 0 l kr = (1.3±0.1)xl0' cm /s using an average of 10 measurements, taken over a period of 2 days at a temperature of 296±1 K. Table V shows the measured rate coefficient results and comparison to a previous measurement by Mackay et al. [6 6 ] using a flowing afterglow device with He and H 2 as the buffer gases. In addition, the table shows the collision rate as calculated using ADO theory. TABLE V. Rate coefficients for the reaction of hydronium with propene in units of (cm3/s). Reaction Rate Coefficient Reaction Rate Coefficient Collision Rate Coefficient (present) ([66]) (calculated) (1.3±0.3)xl0 "9 (1.5±0.3)xl0 ‘9 1 .8 x 1 O' 9 Figure 40 shows the various reaction product ions as a function of the number density of propene. The most predominant product ion was the protonated propene molecule, produced from reaction (52), but several other species are observed. The product ion H 3 0 + CbH6 (61 amu) is a result of the three body recombination reaction (53). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 86 O © CO © o h < o CO Mass Mass (amu) CO FIG. 38. Mass spectrum 38. is FIG. gas mixture 6.0 propene of seem. the rate when the injection Mass spectrum o o o © © © o © © © O © © to O in o CO CM CM ■*” (s O/siunoo Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. oo -j 6 5 4 3 Gas Number Density (x1011 cm"3) 2 Reactant 1 FIG. FIG. 39. Decay of ionhydronium signal as a function of propene density. number Note the that statistical error on the ion counts is smaller than the data points. points. on the ion the counts data density is of than error in the smaller number uncertainty The is propene due to the uncertainty in the flow controller measurement and applies to all the data points. to applies data all the and measurement the flowin controller uncertainty to the 0 1000 10000 100000 1000000 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. without prohibited reproduction Further owner. copyright the of permission with Reproduced Counts/(30 s) 10000 1000 100 10 flow is ±0.75 seem. is ±0.75 flow 1 o.o FIG. 40. Product FIG.40. as function ionsathe propene of mixture The uncertaintyinjection rate. in the 1.0 I 2.0 . 40 5.0 4.0 3.0 i I # Reactant Flow Rate (seem) Rate Flow Reactant m m • (H30+)(C3H6) (61 amu) amu) (61 (H30+)(C3H6) • amu) (57 C4H9+ x + C5H9+ (69 amu) (69 C5H9+ + A amu) (19 H30+ ♦ x C3H8 (44 amu) amu) (44 C3H8 x C3H6+ (42 amu) (42 C3H6+ C3H7 (43 amu) amu) (43 C3H7 6.0 . 8.0 7.0 88 89 The peak at 44 amu is most likely due to a reaction between molecular oxygen ions, and contaminant propane from the propene cylinder, o ; + c 3h s -»c 3h ; + o2. (55) Reaction (55) has been found to be extremely efficient and occurs at approximately the collision rate [18]. Commercial propene is typically created as a byproduct of ethene production. In addition the production of propene, propane is also created, and cannot be completely separated from the propene. The other peaks are most likely due to contaminant ethene in the propene cylinder yielding the following reactions, H30+ + C2HA -»C2H5+ + H20 , (56) c 3h ; + c 2h a -> c 5h ;, (57) c2h ; + c2h a -> cah ;. (58) The peak at 46 amu is a due to the reaction of the propene molecules with C> 2+. 4.4 1-Butene (CVffg) The channels available for the reaction of 1-butene with hydronium are very similar to propene. Because the reaction is exothermic, the dominant channel is the proton transfer reaction given by reaction (59). In addition, the three-body recombination reaction (60) and a disassociative reaction (61) are also observed. H30+ + CaH% -> Cah ; + H20 (59) H30+ + CaH% +A r-> H30+ • CaH s + Ar (60) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 90 H ,0+ + C4H% C4H; +H 2+ H20 (61) The butene molecule is the simplest alkene to have multiple structural isomers, as shown in Table VI. The difference between 1-butene and 2-butene is the location of the double bond. In addition, 2-butene consists of two structures, defined by the location of the two methyl groups relative to the double bond. Trans-2-butene has the methyl groups opposite to each other, while cis- 2 -butene has the two methyl-groups on the same side of the double bond. Although the difference between these three molecules seems minor, these structural isomers possess different properties which are summarized in Table VI. A mixture of 0.4920% (4920 ppm) 1-butene in nitrogen, prepared by BOC Gases, was used for all measurements. The injection rate for the gas mixture varied from zero to 1 0 seem, corresponding to a maximum number density in the flow tube of approximately i i i 5x10 molecules/cm . Since the reaction between the hydronium ion and the 1-butene molecule occurs at the collision rate, injection rates greater than this reduced the hydronium signal to a point where longer sampling times were needed to obtain better counting statistics. TABLE VI. Properties of butene at STP. Proton Dipole Melting Boiling Isomer Polarizability Density Structure Affinity Moment Point Point Name (xlO'24 cm3) (g/cm3) (eV) (D) (°C) (°C) , ch3 1-Butene ,CH / 8.25 N/A 0.438 -185.3 -6.2 0.588 h2c ^ ^ ch2 CH, trans-2- CH=CH 8.49 7.74 N/A -105.5 0.8 0.599 Butene / h3c cis-2- C H =C H / \ N/A N/A 0.253 -138.9 3.7 0.616 Butene h3c c h 3 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 91 Figure 41 shows the mass spectrum in the flow tube when the reactant gas flow rate was 6.0 seem, while Figure 42 shows a representative decay curve of the hydronium ions as a function of the 1-butene number density. A total o f 10 measurements were made over a period of 2 days at a temperature of 296±1 K yielding an average reaction rate coefficient kr = (1.4±0.1)xl0"9 cm3/s. Table VII shows the result of this measurement and a comparison to a previous measurement [67] of trans-2-butene of the reaction of F^O* with trans-2-butene and the collision rate calculated from ADO theory. There is no previous reaction rate coefficient measurement for the hydronium ion with 1- butene. TABLE VII. Rates coefficients for reaction of hydronium ion with butene in units of (cm3/s). Reaction Rate Coefficient for Reaction Rate Coefficient for Collision Rate Coefficient 1-butene trans-2-butene ([67]) (calculated) (present) (1.4±0.3)xl0'9 (1.77±0.35)xl0'9 2.0xl0'9 Figure 43 shows the various product ions observed in the flow tube as the injected 1-butene gas mixture was increased. The most abundant product ion is the protonated 1- butene molecule at 57 amu but numerous other species are observed. The large peak at 71 amu is most likely due to contaminant pentane in the reactant gas cylinder. Since the hydronium ion does not readily react with saturated hydrocarbons, this peak is most likely due to the reaction of O2 in the flow tube with pentane, 0;+C5Hn ^ C 5H+n+02. (62) The O2 also reacts with 1-butene, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. without prohibited reproduction Further owner. copyright the of permission with Reproduced 2500 s )/ o o n p /s i) (s FIG. 41. Mass spectrum when the injection rate of the butene gas mixture is 6.0 seem. 92 u> vo 6 4 5 3 2 Reactant Gas Number Density (x1011 cm'3) 1 FIG. FIG. 42. Decay of ionhydronium signal as a function of density. number 1-butene Note the that statistical error on the ion point. counts error is the data than smaller density ofin number the The uncertainty is 1-butene due points. data to the all applies and flow inthe measurement controller uncertainty to the 0 1000 10000 100000 1000000 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. without prohibited reproduction Further owner. copyright the of permission with Reproduced counts/(1 s) 10000 1000 100 uncertainty in the flow is ±0.75 seem. uncertainty ±0.75 is flow inthe I.4. rdc osa ucino h -ueemxueijcinrt. The Product1-buteneasathefunction ions of mixture injection rate.FIG. 43. 0 4. .0 5 .0 4 .0 3 Reactant Flow Rate (seem) Rate Flow Reactant X(H30+)(C4H8) (75 amu) (75 X(H30+)(C4H8) amu) (71 C5H12+ + A amu) (19 H30+ ♦ C4H8+ (56 amu) (56 C4H8+ C4H9+ (57 amu) (57 C4H9+ 94 95 o ; +c 4h s-> c4h ; +o2. (63) The mass peak at 55 amu is a combination of C 4H7 * from reaction (61) and the hydrated hydronium ion, HiCf -HzO. The small peak at 75 amu is due to the three-body recombination reaction, (60). 4.5 Hydrogen Sulfide (.H 2S ) The only reaction channel for the hydronium ion with hydrogen sulfide is proton transfer, H30+ + H2S -> H3S+ + H20 . (64) The proton affinity o f hydrogen sulfide is larger than water so this reaction is exothermic and is expected to occur at approximately the collision rate. Unlike the other reactions studied, three-body recombination is not observed in this reaction. Table VIII lists the properties of hydrogen sulfide. TABLE VIII. Properties of hydrogen sulfide at STP. Dipole Proton Melting Boiling Polarizability Density Structure Moment Affinity Point Point (xlO'24 cm3) (g/1) (D) (eV) (°C) (°C) S—H / 3.78 0.97 7.307 -85.5 -59.55 1.39 H Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 96 A mixture of 0.500% (5000 ppm) of hydrogen sulfide in nitrogen, prepared by BOC Gases, was used for all measurements. The gas mixture was injected into the flow tube at rates between zero and 1 0 standard ml/min, corresponding to a maximum flow tube number density o f 5x l0 n molecules/cm3. Figure 44 shows the mass spectrum in the flow tube when the reactant gas flow rate was 6.0 seem while Figure 45 shows a representative decay curve for the hydronium ion as a function o f the hydrogen sulfide number density. Q 1 A rate coefficient of (l.l±0.2)xl0' cm /s was obtained from the average of 10 measurements taken over a period of two days at 296±1 K. As can be seen in Table IX, several previous measurements have been made with quite different results. It is interesting to note that the two experiments yielding 1.9xl0 "9 cm3/s both used a standard flowing afterglow apparatus [68,69], while the experiment employing a SIFT obtained a result of 1.4x1 O ' 9 cm3/s [70]. The reaction rate measured in the present study corresponds, within the experimental error, to the SIFT measurement. The reaction rates obtained using the FA were likely artificially high due to reactions of the buffer gas metastables with the hydronium ions, and/or fragmentation of the hydrogen sulfide molecules by metastable atoms, causing an increase in the decay rate of the hydronium ion. The collision rate was calculated using ADO theory. Figure 46 shows the various product ions as a function of the reactant gas injection rate. The most abundant product ions produced in the flow tube is the protonated hydrogen sulfide. There is also a strong H 2 S+ signal due to the reaction, O; + H2S -> H2S+ + 0 2. (65) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 97 As discussed previously, the molecular oxygen ion comes from contaminant oxygen in the buffer gas that is photoionized by UV radiation from the microwave discharge. TABLE IX. Rates coefficients for the reaction of hydronium ion with hydrogen sulfide in units =fl£i2i!ifA^======_==_=__===_=====_====___^^ Reaction Rate Coefficient Reaction Rate Coefficient Collision Rate Coefficient ______(present)______([68-70]) ______(calculated)______(l.l±0.2)xl0-9 1.5x1 O'9 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 98 a>O ooo o CD (amu) Mass Mass o o CO CM FIG. 44. Mass spectrum 44. FIG. sulfide is hydrogen of gas mixture 6.0 the rate seem. when injection the Mass spectrum O O o o o o o o o O o o o o o © § o o o o 00 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. vo vo 6 4 5 3 2 Reactant Gas Number Density (x1011 cm-3) 1 FIG. FIG. 45. Decay of hydronium ion signal as a function of hydrogen sulfide density. number Note the that statistical error on the ion counts is smaller than the data point. point. density on ofthe the ion data number ethene counts isin the than smaller error uncertainty The statistical points. the data to all applies and measurement flow the in controller uncertainty is to the due 0 1000 10000 100000 1000000 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. without prohibited reproduction Further owner. copyright the of permission with Reproduced m *) 10000 1000 100 10 uncertainty in the flow is ±0.75 seem. uncertainty ±0.75 is in theflow ( ) 1 . 10 . 30 . 50 . 70 . 9.0 8.0 7.0 6.0 5.0 4.0 3.0 2.0 1.0 0.0 I.4. rdc osa ucino h yrgnsliemxueijcinrt. The Productasathefunction mixturehydrogen ions sulfide of injection FIG.rate. 46...... i I '1 T I i I m r n * m Reactant Flow Rate Rate Flow Reactant x ...... 1 ...... ! * * m k ...... (seem) [ ...... 1 1 i ■ ...... XH3S+ (35 amu) (35 XH3S+ amu) (34 XH2S+ amu) (19 H30+ • 1 m 1 1 100 101 Section 5 APPLICATION OF THE MCD-FA METHOD TO TRACE GAS ANALYSIS The MCD-FA method provides the ability to measure reaction rate coefficients of charge transfer reactions, but it can also be applied as a new and powerful method of trace gas analysis. Methods currently used such as gas-chromatograph mass spectrometry (GC-MS) are limited in their overall ability to detect various types of molecules accurately, and in real time. Since virtually all GC-MS units use electron impact ionization to ionize the trace molecules, fragmentation is a considerable problem in interpreting the results. In addition, the GC-MS is not capable o f real time analysis due to the necessity o f using capillary tubes to separate the various molecules in the sample. While some techniques are available to detect trace molecules in atmospheric air in real time, they are typically limited to specific contaminants and therefore only have a specific application. The MCD-FA method is able to “soft-ionize” molecules by simply protonating the contaminant molecules. In many instances, addition of the proton to the molecule does not cause dissociation, so the mass of the molecules is immediately known. While the identification of the contaminant molecules is usually straightforward, molecules whose masses are similar can be further identified by injecting various reactant gases and studying the reaction products. Different molecules will typically give a signature reaction product spectrum. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 102 Arguably the most powerful feature of the MCD-FA is its ability to measure trace contaminants in real time. The only waiting time required is the reaction time of the molecules in the flow tube, which is typically tens o f milliseconds. This aspect o f the MCD-FA could make it a very powerful tool for real time monitoring o f atmospheric pollution, trace compounds in a person’s breath during the breathing cycle, as well as real time detection of the volatile compounds used in bombs and chemical weapons. Since the MCD-FA technique monitors the entire mass spectrum, and is limited only by the range of the mass spectrometer, it is capable of detecting numerous contaminants. There are several considerations that must be taken into account when selecting the primary ion. It is critical that the primary ion does not react with the normal constituents of air; otherwise interpretation of the resulting mass spectra would be extremely difficult. While it is important that the primary ion does not react with the normal constituents of air, it is equally important that the ions react very efficiently ( kr « kc) with the contaminant molecules. The hydronium ion is the perfect candidate for the study o f many volatile organic compounds (VOCs) as well as many other common pollutants. As shown in Table X, the proton affinity o f water is larger that the proton affinity of all the normal constituents of air. Furthermore, since reaction of the hydronium ion with many VOCs is an exothermic proton transfer reaction, the reactions will occur at approximately unit efficiency. As discussed in Section 2, the loss of hydronium ions in the flow tube can be accurately described by Eq. (21). Neglecting the axial diffusion, and fixing the reaction length, the solution, given by Eq. (24) is simply the solution to the first order rate equation, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 103 TABLE X. Proton affinity of various organic and inorganic compounds. Proton Affinity Molecule Formula KJ/mol eV Helium He 177.8 1.84 Argon Ar 369.2 3.83 Oxygen 02 421 4.37 Hydrogen h2 422.3 4.38 Nitrogen n 2 493.8 5.12 Carbon Dioxide co2 540.5 5.61 Methane c h 4 543.5 5.64 Nitrous Oxide n2 o 549.8 5.70 Nitroden Dioxide n o 2 591 6.13 Carbon Monoxide CO 594 6.16 Ethane c 2h6 596.3 6.19 Propane c 3h8 625.7 6.49 Ethene c 2 h4 680.5 7.06 Water h2o 691 7.17 Hydrogen Sulfide h2s 705 7.31 Formaldehyde c h 2o 712.9 7.40 Hydrogen Cyanide HCN 712.9 7.40 Benzene c 6h6 750.4 7.78 Propene c 3h6 751.6 7.80 Methanol c h 4o 754.3 7.82 Acetaldehyde c 2 h4o 768.5 7.97 Ethanol c 2h6o 776.4 8.05 Acetonitrile c 2 h3n 779.2 8.08 Toluene C7Hs 784 8.13 Propanol c 3h8o 786.5 8.16 Acetone c 3h6o 812 8.42 o X Isoprene ) V CO 826.4 8.57 Ammonia n h 3 853.6 8.85 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 104 d[Hf ' ] = -{H,0']k[A\, (66) at where [/7?<9+] is the number density of the hydronium ion in the flow tube, [A] is the number density of the reactant gas molecules, k is the measured reaction rate coefficient, and t is the reaction time t=l/vi. Likewise, the growth o f protonated reactant gas molecules can be described by the differential equation, = (67) at For the measurement o f rate coefficients, the experiment operates under the condition [A] » [HiO+] so that reactions (66) and (67) can be solved for the rate constant. On the other hand, trace gas analysis typically results in the situation where [Hi( ? ] »[A] so that the solution used previously does not apply. If the reaction times are sufficiently small so that [A] remains approximately constant, then Eq. (67) can immediately be integrated to give the solution, [A+] =k[A]tL J = -v ^ , (68) where IA and IHQ+ are the count rates o f the reactant gas and hydronium ion, respectively. Rewriting Eq. (68), the number density [A] in the flow tube is given by 1 J * v kl V where I is the reaction length of the flow tube and v* is the ion velocity. If there are multiple peaks associated with several reactions, this equation can be generalized to M = t IpA + + I p2k2 + ... ' (70) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 105 where / +1, / + , etc. are the counts of the different product ions, Ip], IP2 , etc. are the counts of the different precursor ions, and kj, h.2 etc. are the reaction rate coefficients of the various reactions. Using Eq. (69) it is now possible to calculate the concentration of the contaminant in the injected gas in the atmosphere. The number density o f the contaminant in the flow tube can be written as ®»+®, ^•3-54xl0'S' <71> where is the flow of the pollutant gas, is the flow of the argon buffer gas, $ / is the flow of the injected gas mixture,/*/is the pressure in the flow tube in Torr, and 3.54xl016 is the conversion factor for pressure in Torr to number density in cm-3 at room temperature. Using Eqs. (71) and (69) the partial pressure of the contaminant gas in the atmosphere is given by pA^ . lmTorr^ ------® ^ . 7 6 0 J W ® , iv ,(3.54x10“ ) 0 , pf It is important to note that an absolute partial pressure o f the contaminant is measured using relative intensities of the hydronium ion and the protonated contaminant. Unlike other trace gas techniques like GC-MS, calibration gases are not required to obtain absolute measurements using the MCD-FA method. The absolute sensitivity o f the MCD-FA method is determined by several factors. The greater the numbers of hydronium ions in the flow tube the higher the probability that there will be a charge transfer to the contaminant molecules. Additionally, since the detection of the protonated contaminant ions is done by counting pulses on a channeltron, the longer the counting is at each mass value, the better the statistical averaging will be. If the MCD-FA is to scan mass spectra quickly so that data may be obtained in almost Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 106 real time, then long counting times cannot be used and therefore large numbers of hydronium ions are required. In situations where real time data acquisition is not required, such as the study of bag sampled gases, overall sensitivity can be increased by increasing the pulse counting time. 5.1 Pollution Monitoring The effect of emissions from man-made as well as natural sources on the environment is currently the source of considerable study [71,72]. Due to the complex chemistry of the atmosphere, correct modeling requires a detailed knowledge of the concentrations o f the various molecules present as well as the rates for the various reactions. One of the major sources of atmospheric pollution, especially in large urban areas, is from the exhaust of automobiles. Both gasoline and diesel engines bum hydrocarbons, fractionally distilled from crude petroleum, to generate the energy required for motion. The large fractions of these hydrocarbons are aliphatic alkanes. Table XI shows the hydrocarbon sizes and their applications. Under ideal conditions, the burning of hydrocarbons produces only carbon dioxide and water, but as the size o f the hydrocarbons increases and the molecular structure gets more complex, complete burning in an internal combustion engine becomes impossible. In addition, the high heats and large pressures generated during the burning cycle are capable of producing various complex molecules including smaller alkenes and aromatic hydrocarbons such as Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 107 benzene, toluene, and xylene. Fuels also contain various additives as well as sulfer, which is present naturally in crude oil. As a result, a complex mixture of hydrocarbons, water and various other organic and inorganic molecules are emitted from the tailpipe. Once these compounds enter the atmosphere they can directly, and indirectly, interfere with the natural chemistry [71,72]. In addition, many o f the emitted compounds are also known to present severe health risks. TABLE XL Aliphatic alkanes and their applications. Carbon Atoms U ses C to C4 Flammable gases C5 to C7 Naphthas, used for solvents and dry cleaning fluids C7 to C n Mixed to form gasoline for automobiles C12 to C15 Mixed to form kerosene Ci6 Cetane, major component of diesel fuel C17 tO C19 Heavy fuel oils and lubricating oils In addition to the direct health risks posed by the various molecules emitted in the exhaust of automobiles, many of the molecules emitted form the building blocks of particulate matter. Particulate matter consisits of particles that range in size from approximately 1 nm to tens of microns. It is composed primarily of hydrocarbons and sulfur based compounds. The Environmental Protection Agency (EPA) has created two major categories for particulate matter: PM10 consists of particles that range from 2.5 (an to 10 frn, and PM2.5 consists o f particles smaller than 2.5 fm. The greatest health risks come from the PM2.5 particles due to their small size and ability to penetrate into the alveolar sacs of the lungs. Several studies have documented the effects of PM on health, incuding a link to both respiratory and cardiovascular illness [73]. In addition, the atmospheric lifetimes of particles ranging from 200 nm to 2 /an has been found to be Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 108 approximately 10 days as compared to the much shorter lifetimes for smaller and larger particles [74]. While all internal combustion engines produce particulate matter, diesel engines have been found, depending on many factors, to emit as much as 200 times more particulate mass than similar sized catalytically equipped gasoline engines [75], with approximately 90% o f the emitted particles being less than 1.0 pan in size [76]. The nucleation and chemistry of particulate matter formation is not well understood, but studies have shown that larger concentrations of sulfuric acid (H2SO4) as well as other sulfur based compounds in the exhaust increases the nucleation rate and therefore the concentration o f particulate matter [77]. While current techniques used for studying automobile exhaust allow for the characterization of particle size and composition, as well as the monitoring of some hydrocarbons, nitrous oxides and sulfur-based compounds, no instrument is capable of measuring a complete mass spectrum of the automobile exhaust in real time. Application of MCD-FA has the potential to allow the monitoring concentration of many compounds in almost real time to better understand the nucleation and growth of particulate matter. A clearer insight into the formation of these particles may prove helpful in their reduction. To demonstrate the effectiveness of the MCD-FA technique to studying atmospheric pollutants, as well as the potential for the technique to particulate matter formation, the exhaust o f an automobile was sampled and analyzed. The type of automobile used was a 1993 Saturn SL2 with a 4 cylinder, gasoline engine. Samples of the exhaust were obtained using a sealable polyethylene bag with an attached 0.25” (0.64 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 109 cm) poly tube and a Swagelock connector for attaching the bag to the flow tube. During sampling the connector was sealed with a cap to prevent loss o f the exhaust. The car was started and allowed to run for five minutes before the sample was taken. To obtain the sample, the bag was completely collapsed, then the seal was opened just enough to allow the tailpipe to be place in the bag. The bag was kept in place approximately 15 seconds to allow the bag to fill, and then it was sealed. Samples were then taken to the MCD-FA for analysis. The time between obtaining the sample and injection into the MCD-FA was approximately 10 minutes. Figure 47 shows the mass spectra obtained from the sampled car exhaust. The mass range shown was split into 500 bins, resulting in a resolution of 0.24 amu/bin. For each bin the pulse accumulation time was 2 seconds. The flow tube was operated at a pressure of 340 mTorr and a flow rate of 1.00 slm. The exhaust was injected into the flow tube at a rate o f 20.0 seem. The mass spectrum shows a wide variety o f species generated in the combustion process. The most abundant ions in the spectra are the hydronium ion and the hydrated hydronium ions H 3 0 +(H2 0 )n (n=l,2) at 19 amu, 37 amu and 55 amu which are evidence of the large amount of water created during combustion. Other than the small peaks associated with contaminant 0 2 + created from photo-ionization of the oxygen in the system (Epj = 12.07 eV), the majority of the ions shown in the spectra are various aliphatic and aromatic hydrocarbons produced from the cracking o f the larger aliphatic hydrocarbons that make up gasoline. Although larger molecules have smaller concentrations in the exhaust, peaks showing the presence of the protonated aromatic Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 110 hydrocarbons benzene (79 amu), toluene (93 amu), and xylene (107 amu) can be seen in the data. The most obvious features missing from the car exhaust mass spectrum are the peaks associated with the various oxides of nitrogen. The simple reason for this is that the proton affinity of all of these molecules is less than water so the proton transfer reaction is endothermic and is unlikely to occur. A small peak can be seen at 30 amu showing the presence o f NO, but this is a result of the reaction with the contaminant ion O2 . The peak at 28 amu is most likely not due to the nitrogen molecule, but is instead protonated ethene. TABLE XII. Concentration of several trace compounds in car exhaust. Compound kr for reaction with H 3 0 + Atmospheric Concentration (cm3/s) (ppm) Propene (C 3H6 ) 1.4 x 10'9 [Present Work] 40 Butene (C4H8 ) 1.5 x 10'9 [Present Work] 120 Benzene (CeHe) 1.9 x 10‘9 [78] 0.8 Toluene (CyHg) 2.2 x 10'9 [78] 0.5 Using the measured reaction rate coefficients, the concentration of several gases was calculated using Eq. (72). Table XII shows the calculated concentrations. Unfortunately, no data for comparison was available in the literature. It should be pointed out that this is desired ratios between the hydronium ion and the contaminant ions have thus far not been completely optimized for trace gas analysis using the MCD-FA, so the calculated concentrations are preliminary. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. o o O O o o o o O o o Lfi o ID o 500 CO CM CM T“ X— (s z)/s}unoQ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 112 5.2 Exhaled Alveolar Air The exhaled alveolar air of an individual contains many molecules which can provide a wealth o f information about that person’s health [79], The concentration of these compounds is typically in the low ppm to ppb range which can make them challenging to detect, especially in real time. The MCD-FA provides a method that makes real time detection of these molecules possible and provides a non-invasive alternative to disease diagnosis. Current methods for detecting trace molecules on a person’s breath suffer from several drawbacks. Although gas chromatograph (GC-MS) techniques can be used for detecting the trace volatile organic compounds (VOCs) [79-81], the sample typically must be concentrated using a cryogenic or absorption trap. In addition, although the method provides detection o f the VOCs, quantification o f the concentrations is not accurately known due to the laborious procedure required to collect the sample. Finally, as with all GC studies of trace gases, fragmentation of the trace molecules due to the electron impact source can make interpretation of the resulting mass spectrum extremely difficult. The use of rare isotopes used in conjunction with GC-MS techniques has also been applied to the detection of diseases [82]. Although this method is effective in detecting a specific disease, there is considerable cost associated with detection due to the price of the rare isotope. The application of the MCD-FA has the possibility of overcoming virtually all of these difficulties. Using the techniques described in the introduction of Section 5 it is possible to detect and quantify trace compounds on a person’s breath without the use of Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 113 rare isotopes. Since the detection time is on the order o f milliseconds, the compounds can be detected in almost real time allowing the measurement of concentrations during the normal breathing cycle. Since the technique employs proton transfer to “soft-ionize” the molecules, few of the reactions will result in molecular cracking thereby simplifying the interpretation of the spectra. Finally, since the MCD-FA can be employed over a large mass spectrum, the device will not be limited to single disease detection, but could be used to identify many illnesses by looking for “finger print” molecules of a particular disease. Identifying molecules associated with a particular disease would simply require building a database of compounds by sampling the breath of many individuals. To demonstrate the effectiveness of the MCD-FA method to detecting trace compounds in exhaled alveolar air, the breath of three individuals was sampled and analyzed. One individual was a regular smoker, one individual was a diabetic, and one individual was a nonsmoker suffering from no known illnesses. The breath of smokers is known to contain numerous complex molecules due to the sticking o f these molecules to the surface of the lungs during the inhalation of cigarette smoke [83], while the breath of diabetics is known to contain elevated levels of acetone [81]. The healthy nonsmoker was used as a control. To obtain the sample the individual exhaled into a polyethylene bag which was immediately sealed and attached to the flow tube injection port using previously connected poly tubing and a Swagelok connector. The breath was then injected into the flow tube at a rate of 30 seem. The flow tube was operated at 340 mTorr with an argon flow of 1.0 slm. All three runs were done consecutively on the same day. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 114 Figures 48 through 51 show the mass spectrum of the lab air, and the spectrum obtained from the breath sample of each individual. It should be noted that the measured acetone level (59 amu) is elevated due to a small contamination in the flow tube, most likely from a contaminated o-ring. The laboratory air contains numerous compounds such as methanol (33 amu 69 amu) and acetone (59 amu and 77 amu). Several peaks can be seen in the exhaled breath of all the individuals that are not seen in the lab air. Most notable is the presence of isoprene (the contributor to the peak at 69 amu), increased levels of acetone and methanol. Although only slight differences can be seen between the spectra o f the three individuals, the diabetic has an increased level of acetone (59 amu) while the smoker has an elevated peak at 69 amu which is more than likely due to an increased level of isoprene. Additionally, the acetonitrile (42 amu), can be observed in the breath of the smoker and the diabetic. The peaks at 20 amu and 21 amu are the isotopes of the hydronium ion, H 3 180 + and H2DO+ respectively, while the peaks at 37 amu, 55 amu, and 73 amu are the various hydrated hydronium ions, H 3 0 +(H2 0 )n (n=l,2,3). The mass spectra shown in Figures 15-18 demonstrate the ability to use the MCD- FA as a tool to detect trace compounds on a person breath, but increased sensitivity will be required before it can be used to accurately quantify these compounds. To increase the detection sensitivity, it will be necessary to increase the number o f ions reaching the interaction region o f the flow tube. The major loss of proton donors is due to the diffusion of the ions to the walls o f the flow tube. A larger roots blower, providing a much faster ion flow velocity would increase the number o f ions reaching the reaction region as well as facilitating a greater flexibility in reaction times. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. without prohibited reproduction Further owner. copyright the of permission with Reproduced Counts/30 s Counts/30 s 100000 100000 10000 10000 1000 1000 100 100 10 1 2 3 4 5 6 7 8 9 100 90 80 70 60 50 40 30 20 10 0 K ■ 0 t t t t t t t t t t t t t t FIG. 49. Mass spectrum Mass FIG.the 49. of breatha healthy of non-smoker. 10 / h FIG. 48. Mass spectrum Mass laboratory of FIG. air.48. 20 3 o + NO’ NO+ ) P W . T O 30 05 60 50 40 Mass (amu) Mass Mass (amu) Mass h 3 o CH +( h s c +(H20) 0 2 5 o h )2 9+ 70 H30 +(H20)3 C3H70 +(H20 ) 80 90 100 115 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. without prohibited reproduction Further owner. copyright the of permission with Reproduced Counts/30 s Counts/30 s 100000 100000 10000 10000 1000 1000 100 100 10 1 2 3 4 5 6 7 8 9 100 90 80 70 60 50 40 30 20 10 0 0 10 FIG. 50. Mass spectrum FIG.Mass50. the of breathadiabetic. of FIG. 51. Mass spectrum of spectrumMass FIG.51.the breath of smoker.a of h 3 20 o + r 7 i L li,fir {'f'tf'ti rL "iTt'tYri, ,-?-i^ V i i i i‘ t i rrn r r i t i r - n r - n - i r t ' I r n r r - n r r - ‘n i i r f i i i i i V ^ i - ? - ), l f r i ,i i r Y t ' t T i '" Ll ir r if t ' f t ' ‘{''>f r , t ' l rw i f i*, il V / u i - ILf T f - f r r 'ir NO' NO+ 02 30 3 +(H20)H30 40 H aC m O ) ) O m aC H Mass (amu) Mass Mass (amu) Mass 50 h 0+(H2030 )2 070 60 CH s c +(H20)0 5 h 9+ 3 +(H20H30 )3 C3H70 +(H20 ) C3H70 +(H20) 9080 100 116 117 Section 6 CONCLUSIONS The reaction rates between the hydronium ion and ethene (C2H4), propene (C3H6), 1-butene (C4H8), and hydrogen sulfide (H2S) were measured. Measurements were then compared to collision rates that were calculated using average dipole orientation (ADO) theory. For exothermic proton transfer reactions the reaction rate is expected to be approximately equal to the collision rate. The reaction rate between the hydronium ion and ethene was measured to be (6.9±1.4)x 10'11 cm3/s, while the collision rate was calculated to be 1.4x1 O'9 cm3/s. Since the reaction is endothermic, the activation energy required for the reaction to occur must be supplied by the translational kinetic energy of the gas molecules. A calculation of the reaction rate coefficient based upon the collision rate and a Maxwell-Boltzmann distribution o f molecule velocities yielded a rate of 7xl0'n cm3/s, which agreed with the measured result within the error of the measurement. The reaction rate agreed with a previous proton transfer reaction rate measurement to within the experimental error. The proton transfer reaction rate between the hydronium ion and propene was measured to be (1.3±0.3)xl0'9 cm3/s and the collision rate was calculated to be 1.8xl0'9 cm3/s. Since the reaction is exothermic, the reaction rate was expected to be approximately equal to the collision rate. The difference between the two values is approximately 30%. This measurement agreed with a previous measurement o f propene to within the experimental error. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 118 The proton transfer reaction between the hydronium ion and 1-butene was measured to be (1.4±0.3)xl0'9 cm3/s and the collision rate was calculated to be 2.0x1 O'9 cm3/s. The measured rate was found to be slightly smaller than a previously measured reaction rate of (1.77±0.35)xl0'9 cm3/s between the trans-2-butene isomer and the hydronium ion. Since the reaction is exothermic, the reaction rate was expected to be approximately equal to the collision rate. The difference between the two values is approximately 35%. The proton transfer reaction between the hydronium ion and hydrogen sulfide was measured to be (l.l±0.3)xl0'9 cm3/s and the collision rate was calculated to be 1.5x1 O'9 cm3/s. Since the reaction is exothermic, the reaction rate was expected to be approximately equal to the collision rate. The difference between the two values is approximately 30%. Two previous measurements of the proton transfer reaction rate 0 1 using a standard flowing afterglow gave a result of 1.9x10' cm /s, which is higher than the calculated collision rate. The most likely explanation for this is that buffer gas metastables in the FA caused fragmentation of the hydrogen sulfide molecules, increasing the number density of reactant molecules. The resulting reactions would cause an increase in the decay rate o f the hydronium ion and yield an artificially high reaction rate. Another previous measurement of the reaction using a SIFT device measured a value of (1.4±0.3)xl0'9 cm3/s which is within the experimental error o f the present measurement. All of the measurements were made using a novel microwave cavity discharge flowing afterglow (MCD-FA). In order to eliminate the errors associated with the conventional flowing afterglow device, the ion source was separated from the flow tube. Initial development of the flow tube focused on the application of a hollow cathode Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 119 discharge as the ion source, but this source was eliminated due to problems with sputtering. A capacitively coupled RF discharge was also studied, but was also eliminated due to the low ion currents supplied to the reaction region of the flow tube. The microwave cavity discharge in water vapor was chosen since it did not create sputtered products and provided a clean, intense source of hydronium ions. In addition to studying the ion output of the MCD, optical emission spectroscopy was used to better characterize the operating parameters o f the discharge. The gas temperature of the discharge was measured using the 0-0 vibrational band of the OH' radical located at 306.4 nm. The temperature was measured to be 750±100 K. The electron density of the discharge was determined by measuring the Stark broadening of the H$ line in the discharge. The electron density was measured to be 1.6x10 cm ', but the estimated error on the measurement could be as high as 100% since Stark broadening theory in this density range is inaccurate. The electron temperature was measured using ratio of intensities between two different argon ion lines, and was determined to be 7500 K. The MCD-FA was also shown to be an effective tool for real time trace gas analysis of atmospheric air. The MCD-FA has several advantages over the conventional GC-MS technique of trace gas analysis in that measurements are obtained in almost real time, and ionization of contaminants occurs predominantly through simple protonation instead of electron impact ionization which typically causes molecular dissociation. Using the MCD-FA technique, the presence of various compounds was detected in car exhaust and exhaled alveolar air. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 120 APPENDIX A THEORY OF THE QUADRUPOLE MASS SPECTROMETER The quadrupole mass spectrometer (QMS), developed by Paul in the fifties [84], has become an invaluable tool in many different scientific fields. Due to the popularity of the device as a scientific instrument, an extensive amount o f literature exists discussing the theory of operation [64,85-87]. The information presented in this appendix is only a brief overview of the basic theory of operation of the QMS, and the interested reader is directed to the provided references for greater detail on the subject. .. / m / m Ion Injection / \ m X m -m (a) (b) FIG. 52. QMS geometry: (a) Hyperbolic electrode geometry of the QMS, (b) Configuration of the QMS for ion filtering. Ideally, the QMS consists of four hyperbolic electrodes, running parallel to a common central axis as shown in Figure 52. The electrodes that are opposite one another Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 121 are connected together at a potential (f(t), while the two adjacent rods are connected to an opposite potential, -(fit). The resulting field between the rods is given by the expression, (x2 -y2) m = - 2— ^(0 . (73) ro where r0 is the distance of the central axis to the surface of the electrodes. If the potential on each electrode is a DC biased sinusoidal field ( potential between the electrodes becomes, 0 ( 0 = (U + V0 cos cot). (74) The electric field between the plates is then given by / A a\ ~2xi + 2 yj £ (x ,y ,0 = -V O = {U+ V0 cos cot), (75) so that the equations of motion for particle o f mass m, and charge e between the electrodes are, a x + —%{U + V coscot)x = 0 , (76) mr0 2e * \ y T(U + V0 coscot)y = Q. (77) mr By making the substitutions, cot = 20, (78) 8 e U sn n \ - (79> mrBco . 8eF /om 2q = — r T, (80) mrco Eqs. (76) and (77) may be rewritten, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 122 ^ * + (a + 2#cos2#)x = 0, (81) dd1 d2y [a + 2q cos 20 )y = 0, (82) d02 which are in the form of the well known Mathieu equations. There are two classes of solutions to these equations, a stable, oscillatory solution, where x (or y) remains finite as #-»oo(r—»a)5 and an unstable solution where x-»ooas 0-»oo. It has been shown [88] that the solutions are stable in the x direction if a is less than the value given by the expression 8 64 1536 In the y direction, stable solutions exist if a is less than the value given by the expression, ^y+o(?‘)- (84) Figure 53 shows a plot of these functions. The shaded region under represents the region o f stability, where the ion will have stable oscillations in both the x and y direction. It should be noted that there exists other regions o f stability in the a-q plane, but virtually all QMSs operate in the region closest to the origin shown in Figure 53. Mass scanning in a QMS is accomplished by defining a line of constant slope on the a-q plane that crosses through the region of stability. From Eqs. (79) and (80), the ratio of the a and q coefficients is given by ^ = (85) <1 K As can be seen in Figure 54, the range of masses that the QMS is capable o f scanning over is defined by the slope of the a/q ratio, and the initial and final masses are determined by the intersection of the a/q line with the ax and ay curves. Using Eq. (79) or Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 123 Eq. (80), it can be seen that there are two possible methods for scanning over different mass values. The first and most common method is to fix the RF frequency, and vary the peak RF voltage and DC offset voltages on the rod while keeping their ratio constant. The second method is to fix the voltages on the rods while varying the RF frequency. B.2BD80 4 Constant a/q 0 1 0,3 0.4 0,6 s.r 0.8 FIG. 53. Stability plot for QMS. In order for mass separation to occur, the ions must remain in the quadrupole field long enough to undergo a sufficient number of RF cycles. The resolution is thus primarily controlled by three factors: the RF frequency, the length of the QMS, and the injection velocity o f the ion into the QM S. The resolution o f any mass spectrometer is defined by Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 124 R = -^~, (86) Am where m is the mass of the ion and Am is the full width half maximum of the peak. It has been shown in Ref. [85] that the minimum number o f oscillations required to obtain a resolution R is given by the expression « = 3.5Vr . (87) Using this in conjunction with Eq. (79), the maximum energy a particle can have when injected into the QMS is given by the expression Ema ~ 2(3.5f R 5 (88) where I is the length of the QMS, and/is the operating frequency. In addition to the factors just discussed, there are numerous other factors affecting resolution that are beyond the scope of this Appendix. A more detailed discussion o f the various parameters affecting resolution can be found in the literature [64,89]. While the ideal QMS is constructed with hyperbolic electrode surfaces, in practice precision machining of these types of surfaces is extremely difficult. Dayton et al. [90] showed that cylindrical rods could be used, providing a field geometry very similar to that produced by hyperbolic rods. The rod spacing found to produce the largest region of quasi-hyperbolic fields was found to be given when the ratio of the roddiameter to the field radius r0 was 1.16. Due to the ease o f precision machining cylindrical rods, virtually all modem commercial QMSs use cylindrical electrodes instead of hyperbolic electrodes. The effect of using cylindrical rods instead o f hyperbolic electrodes is discussed in more detail in Ref. [64]. In general, the larger the diameter the rods used to Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 125 construct the QMS, the higher the resolution R, while the smaller the diameter used, the larger the scannable range of masses. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 126 APPENDIX B THEORY OF CYLINDRICAL RESONANT CAVITIES The general outline for this derivation was taken from the book Classical Dynamics by Jackson [91]. Figure 52 shows a cylindrical, hollow waveguide with a radius, R. FIG. 54. Cylindrical waveguide Let us assume that the walls of the cylinder are made of a perfect conductor and the interior space is filled with a non-dissipative material with a permittivity, e , and permeability, f i . Let us also assume that the electromagnetic waves propagating inside the cylinder possess a sinusoidal time dependence so that they may be written E = E {x,y,z)e i B = B(x,y,z)e ~iM. (90) If no current or charge sources exist inside the cylinder then, using the above definitions for the fields, Maxwell’s equations have the form Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 127 V x £ = icoB, (91) V x 5 = -ico/j-sE, (92) V*£ = 0, (93) V»5 = 0. (94) If we take the curl o f Eqs. (91) and (92) and use vector identities, then we get the wave equations (v2 + fflV)I = 0, (95) (V2+® V )B = 0. (96) If we take advantage o f the cylindrical geometry of the of cavity, then we can write the field in terms of an axial component, z and transverse components, x andy, E = E (x,y)e±ikz-im, (97) B = B {x,y)e±ih-ia,t. (98) Substituting these expressions back into Eqs. (95) and (96) we get the wave equations, (v2+(/W))i = 0, (99) (v2+(/i«y2))i =0 , (100) where V2 =V2- —dz2 . We can also write the fields in terms of an axial and transverse component, E = Et + Ez (101) where Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Here we use z to denote the unit vector in the z (axial) direction. The magnetic fields have the identical form. With the above definitions, we now wish to rewrite Maxwell’s equations in terms of transverse components and axial components. Let us first rewrite the vector product zx : ( V x i) , ps z X (V X e ) = V (z«E) - Z«VE = VEZ E 1 z dz dz dz So, z x (V x E ] = VtEz - — Et . (103) v ' dz Taking the cross product of z with Eq. (91) and applying Eq. (103) we get, VtEz =iG>(zxB) + — Et . (104) v ’ dz Taking the dot product of z with Eq. (91) z.(V t xE t) = icoBz . (105) Repeating this procedure with Eq. (92) ps VtBz = -iju£co(z xEt) + — Bt, (106) v ' dz z«(V; x Bt j = -inecoEz . (107) Equations (93) and (94) may be rewritten in terms o f axial and transverse components by simply substituting in the definitions (101) and Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. This gives the separated equations, (108) (109) Equations (104) though (109) are simply Maxwell’s equations written in terms of axial and transverse components. We can now apply these equations to the propagation of the electric and magnetic fields propagating inside of the cylinder. Let us begin by considering the boundary condition imposed on the fields by the interior surfaces. For perfectly conducting walls, the boundary conditions are hxE = 0 surface (110) = 0. I surface Writing the first boundary condition in terms of axial and transverse components we get hxE = hxEt + hxEz = 0 . (ill) surface surface Both the normal vector and the transverse component of the electric field lay in the x-y plane. This implies that the cross product of the vectors lies along the z direction. The cross product between the normal vector and the axial component of the electric field lies in the x-y plane. Since the resulting cross products in the above relations are orthogonal to one another, they must both be zero independent o f one another. Since n and Ez are orthogonal to one another, due to the geometry of the problem, this requires that Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 130 E \ f = 0 , (112) 2 1 surface Consider next the second boundary condition of (110). Take the dot product of (106) with the normal at the surface, h»VtBz =- an •— 9 sStr ■iHSCoh»zxEt (113) surface surface dz surface The second boundary condition of (110) is n»B =n»Bt +n°Bz = 0 . (114) surface surface surface Since the z-componant of the magnetic field is perpendicular to the normal by the geometry o f the problem, the second term in the above expression is zero. This immediately implies that the first term must also be zero, which implies the first term on the right side of Eq. (113) must also be zero. Consider now the second term on the right side of Eq. (113). n»zxE,\ = z»E. x«| =-z*nxE , = 0 . I surface I surface surface So Eq. (113) simply reduces to the result of thr second boundary condition ».VA = — B z = 0 . (115) surface Qn surface Consider now the propagation of the waves inside of the cylinder. For TM waves BZ= H Z= 0 so Eq. (106) becomes Ht = ± — z x E (116) k If we take the cross product of this result with z then , .SCO , „ 0 , SCO z x H = ± — z x z x E , = ± — \z-Et)z-z-zEt] = T^-Et Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 131 Using this in Eq. (104) gives, ±1M(+1^ L £(=V,£„ k E = +- — 2 Vv ^E ’ (117) where, y 2 = fisco2 - . (118) The wave equation for Ez is (119) For TE waves E = 0 so (104) becomes 2 x H = + — Et . (120) fXCO But, z x z x H t — {^z»Ht^z-z»zHt= -H . (121) So, H = ± z x E t . (122) fJLCO Using this in Eq. (106) (123) r The wave equation for/7zis (v;+rj)^=o, (124) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 132 Assume now that the cylinder has perfectly conducting end caps such that it has a inside length of d. Since we know that the electromagnetic waves will reflect from the end caps, the solution to the axial fields will take on the form Asm{kz} + Bcos{kz). (125) It is evident from the above solution that in order for the fields to satisfy the boundary conditions (110) on the end caps, it is necessary for the wave vector to take on the discrete values, * = = 0,1, 2,3,... (126) d Let us begin by considering the TM modes of the cavity where the axial component of the magnetic field, Bz, is zero. The boundary condition on the electric field at the end caps requires z x E = z x E t = 0. (127) surface surface On the end caps the unit vector z is always perpendicular to Et . This then requires Et\ = 0 . (128) \ surface In order to satisfy this boundary condition the transverse electric field must have the form Et(x ,y ,z) = Et(x,y) s i n f ^ p ) . (129) V « Using this and the modified Maxwell equation (108), it can be seen that the axial component o f the electric field must look like Ez = w {x ,y ) cos . (130) I d Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 133 For TE modes the axial electric field is zero, and the boundary condition on the magnetic field is n*B\ = z»B, + z»B \ = 0, \ surface surface surface B, = 0 , (131) surface From this boundary condition to hold true we require that the axial magnetic field have the form ^ pn z^ Hz = ^(x,y)sin (132) where we have used the definition, B = juH. Using Eqs. (116), (117) and (130) the transverse components for TM waves are 5 pn . f p7tz E - S m - — ' d r 2 { d / \ (133) ' PKZ ' HU = lSC0 cos z x V ^ , 7 where y/ is the solution to the wave equation (119). Likewise, using Eqs. (120), (123) and (132) the expressions for the transverse components for the TE waves are r pKZ^ EB -i& y—sin P . zxVty/, r pnz (134) H.B = — Pn rcos d f where y/ is the solution to the wave Eq. (124). The eigenfrequencies for the fields, given by Eq. (118), become _ 1_ p n (135) <*>X p = r l + eju Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 134 Let us now consider the solution to the wave equation for TM fields. Since our cavity is cylindrical, write the wave equation in cylindrical coordinates, 1 d 1 d2 2 . p —- w +— — f + r f = o. (136) p d p { dp p 2 d Our boundary conditions require that y/ = 0 when p = R . The solution to this equation that remains regular at the origin is (137) where, rmn = ~ ’ m = 0,1,2,...;« = 1,2,3... (138) In the above equations, Jm is the first order Bessel function and xmn is the ntil root of Jm. The eigenfrequencies are given by *mn , P * (139) mmnp yfps r z \l v R2d 2 + " ^ 2 The TE fields are obtained by solving the equation d 1 d2 2 n 1 J L P-r-W + —r — ry/ + y y / = 0 , (140) p d p d p ’ ) p 2 dtp2 subject to the boundary condition, dyr = 0 . dp surface The resonant frequencies are given by 2—2 1 Xmn , P * 6), + - (141) mnp yJJlS V R2 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 135 where xmn is the nth root of Jm (x) = 0. The solutions that remain regular at the origin are given by w{p, Table XIII summarizes the results obtained which can be used to calculate the resonant frequencies and field structures of cylindrical resonant cavities. The roots to the Bessel functions,3 xm„tnn and x' mn can be found in numerous math tables [921. »• J TABLE XIII. Summary of cylindrical resonant cavity equations. TM Fields TE Fields 2 2 2 -_2 2 1 Xmn,PX p n co.mnp -+ Resonant yfps V R1 d 2 Frequencies m = 0,1,...; « =1 ,2,...; £> = 0,1,. m = 0,1,...; » = 1,2,...; p = l,2,. r p n z ' ' pn z Ez = y/(p,(p) cos =V(P’ r pnz^ s - ia p . r pnz^ Et5 = -Pn ^ n • is, = 7-sm zxW ty/, C T j y j Transverse Fields \ p n z ^ u Pn pn z os z x V ty/. H, = — r-COS v ^ . r ' Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 136 APPENDIX C OBSERVED ARGON LINES IN MCD TABLE XIV. Observed argon lines in water vapor discharge with an admixture of argon (grating 1 - blazed at 500 nm) Species Intensity 337.0915 AE 153 337.347 AI 472 337.6436 An 99 338.8531 All 131 339.373 AI 751 339.7895 All 103 340.6180 AI 215 354.5595 All 91 354.5843 All 88 355.4305 AI 3518 355.9508 All 51 356.103 All 105 356.2191 All 49 356.3286 AI 600 554 356.5029 All 52 356.7656 AI 1939 1765 357.2295 AI 636 598 357.6615 All 111 128 358.1608 All 46 358.2354 All 77 358.844 All 87 121 360.6522 AI 2850 363.2683 AI 786 363.446 AI 1095 364.3116 AI 421 364.9832 AI 1388 372.9308 AH 115 376.527 All 100 376.6118 All 59 377.0369 AI 695 378.084? An 340? Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 137 Wavelength Species Intensity (nm) 389.466 AI 141 390.0626? An 277? 392.5719 All 33 392.8623 All 71 394.6097 AH 30 394.7505 AI 300 394.8979 AI 1500 TABLE XV. Observed argon lines in water vapor discharge with an admixture of argon (grating 2 - blazed at 800 nm) Wavelength Species Intensity 383.4678 AI 5560 389.466 AI 1570 394.7505 AI 10240 394.8979 AI 28570 396.8359? A n 3040? 397.9355 All 400 399.4792 All 130 401.3856 All 250 403.3809 All 420 404.2893 All 450 404.4418 AI 49000 404.4418 AI 56500 404.5965 AI 4670 405.4526 AI 2580 405.4526 AI 2400 410.3912 All 560 410.3912 An 420 413.723 a u 600 413.723 A ll 620 415.8591 AI 325100 415.8591 AI 260550 416.418 AI 36580 416.418 AI 34800 416.418 AI 38290 418.1884 AI 51700 418.1884 AI 56120 418.1884 AI 48730 419.0713 AI 92430 419.0713 AI 82010 419.0713 AI 87680 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 138 Wavelength Species Intensity (nm)______419.8317 AI 210000 419.8317 AI 198710 420.0675 AI 282600 420.0675 AI 307170 422.8158 An 470 422.8158 a h 470 423.7219 All 350 423.7219 All 220 425.1185 AI 21900 425.1185 AI 22020 425.1185 AI 22740 425.9362 AI 190000 425.9362 AI 216320 426.6287 AI 72000 426.6287 AI 57200 426.6287 AI 67610 427.2169 AI 127900 427.2169 AI 127770 427.2169 AI 107220 427.7528 All 1290 427.7528 An 1150 427.7528 All 1210 430.0101 AI 94750 433.1199 All 980 433.1199 a h 960 433.1199 All 840 433.3561 AI 111000 433.5338 AI 33650 434.5168 AI 37750 434.8064 An 4000 434.8064 All 4050 434.8064 All 3010 436.3795 AI 2400 436.3795 AI 3330 436.3795 AI 2770 436.7832 All 4850 437.0753 A ll 2150 437.0753 An 2570 437.0753 All 1880 437.5954 All 230 437.5954 A ll 210 437.9666 A ll 710 437.9666 A ll 690 440.0096 A ll 370 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 139 Wavelength Species Intensity (nm) 440.0096 All 290 440.0986 All 1170 440.0986 All 1090 442.3994 AI 1470 442.3994 AI 1300 442.6001 All 2310 442.6001 All 2310 443.0189 AH 800 443.0189 AH 720 443.0996 AH 450 443.0996 A ll 360 447.4759 a h 270 447.4759 All 250 448.181 A ll 560 451.0733 AI 103000 452.2323 AI 11710? 452.2323 AI 22800 4 5 4 . 4 7 4 6 AI/AII 2200 457.9349 All 1400 458.4956 AI 290 458.6611 AI 200 458.7209 AI 140 458.9289 AI 1040 458.9898 All 1600 495.6750 AI 1820 1840 496.5079 A ll 1500 1580 497.2160 An 120 150 498.9943 AI 780 735 500.9334 All 835 820 501.7163 A ll 775 795 503.2024 AI 920 890 504.8811 AI 4780 505.4176 AI 3860 505.6529 AI 1290 1300 506.0079 AI 5260 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 140 Wavelength Species Intensity (nm)______506.2037 All 670 670 507.3076 AI 1580 1670 508.7084 AI 1550 1550 511.8202 AI 3000 514.1783 An 990 800 514.5308 A n 460 440 515.1391 AI 11680 516.2285 AI >40000 539.3979 AI 1220 541.0475 AI 2400 541.72 Continuum 760 542.1351 AI 11600 544.2246 AI 4050 545.1652 AI 29000 545.7416 AI 4890 545.9651 AI 23750 546.716 AI 2500 547.3451 AI 3940 549.012 AI 2020 549.208 AI 970 549.5873 AI 54626? 550.6112 AI 11800 11690 552.4957 AI 7380 552.8961 AI 220 7880? 553.4479 AI 1020 554.0860 AI 260 555.8702 AI 60760? 88600 555.9676 AI 6700 6210 557.2541 AI 32470 35320 558.1871 AI 2050 2430 558.872 AI 4570 6150 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Wavelength Species Intensity 559.7476 AI 6030 6350 560.6733 AI 93150 561.8014 AI 3420 562.0921 AI 1480 562.377 AI 2080 563.5576 AI 2810 563.734 AI 1000 563.912 AI 1770 564.1376 AI 4720 564.2428 A ll 470 564.8687 AI 5210 565.0704 AI 52570 565.9127 AI 5700 568.1849 AI 6200 568.373 AI 960 570.0872 AI 2075 <658-662 Continuum Wing 660.4853 AI 35600 663.2084 AI 3400 663.8220 All 800 663.9740 A ll 550 664.3697 All 480 665.6725 AI 1035 666.0677 AI 19650 666.4051 AI 22150 666.6359 All 250 667.7281 AI 182050 668.4788 AI 3450 669.8874 AI 14450 671.9219 AI 15000 672.2890 AI 2500 675.2834 AI >300000? 675.4372 AI 7100 675.6163 AI 22500 676.6611 AI 41150 693.7664 AI 335000 696.5430 AI 11710000 699.2212 AI 23290 703.0251 AI 670000 491610 706.7217 AI 15600000 706.8735 AI 488000 298000 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 142 ^ ( nm)1^ 1 Species Intensity 708.6705 AI 20800 710.7477 AI 131000 70400 712.5820 AI 98200 49730 714.7041 AI 1320000 1270000 715.8839 AI 60600 720.6980 AI 160000 166000 722.9940 AI 9700 726.5172 AI 34100 727.0664 AI 36260 727.2935 AI 2550000 731.1716 AI 134100 139700 731.6005 AI 71500 83090 735.0814 AI 20460 44060 735.3293 AI 268500 342400 737.2117 AI 977600 738.0423 A ll 12600 738.3980 AI 20920000 739.2980 AI 152300 71380 741.2337 AI 86240 49560 742.2312 AI 29600 742.5294 AI 149800 50700 743.5368 AI 211500 179000 743.6297 AI 54100 46700 747.1164 AI 33360 748.433 AI 54390 750.3868 AI >770000 21900000 751.0408 AI 47700 751.4651 AI 25400000 763.5105 AI 47060000 772.3760 AI 13840000 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 143 Wavelength Species Intensity (nm) 772.4207 AI 12300000 794.1876 AI 14900000 800.6156 AI 13260000 801.4785 AI 29520000 810.3692 AI 24335000 811.5311 AI 89860000 840.8209 AI 16815000 842.4647 AI 23366000 25000000 24500000 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 144 APPENDIX D USEFUL UNIT CONVERSIONS TABLE XVI. Mass flow conversion table. Pa m3/s mbar 1/s Torr 1/s Atm cm3/s pi/s seem slm Pa m3/s 1 10 7.5 9.87 7.5x103 592 0.592 mbar 1/s 0.1 1 0.75 0.987 750 59.2 5.92xl0'2 Torr 1/s 0.133 1.33 1 1.32 1000 78.9 7.89xl0‘2 Atm 0.101 1.01 0.76 1 760 59.8 5.98xl0'2 cm3/s p 1/s 1.33xl0'4 1.33xl03 10'3 1.32xl0'3 1 7.89xl0'2 7.89xl0'5 seem 1.69xl0'3 1.69xl0'2 1.27xl0'2 1.67xl0'2 12.7 1 10'3 slm 1.69 16.9 12.7 16.7 1.27xl04 1000 1 TABLE XVII. Pressure conversion table. Pa pbar Torr Micron PSI bar atm (N/m2) (dyn/cm2) (mmHg) (mTorr) (lb/in2) Pa 1 lxlO'5 10 7.5xl0'3 7.5 9.87xl0'6 1.45x1 O'4 (N/m2) bar lxlO5 1 10s 750 7.5x10s 0.987 14.5 pbar 0.1 lxlO"6 1 7.5xl0'4 0.75 9.87xl0"7 1.45x10‘5 (dyn/cm2) Torr 1.33xl02 1.33xl02 1330 1 1000 1.32x1 O'3 1.93xl0'2 (mm Hg) Micron 0.133 1.33x10‘6 1.33 10'3 1 1.32xl0'6 1.93xl0'5 (mTorr) atm 1.01x10s 1.013 l.OlxlQ6 760 7.6xl05 1 14.7 PSI 6.89xl03 6.89xl0'2 6.89xl04 51.71 5.17xl04 6.8xl0'2 1 (lb/in2) TABLE XVIII. Miscillaneous conversion factors. Unit Type Unit Name Abbreviation Relation to S.I. Units Energy 1 erg erg 10~7J Energy 1 electron volt eV 1.602x10'19 J Energy 1 calorie cal 4.184 J Energy 1 kilocalorie/mole Kcal/mol 4184 J/mol Electric Charge 1 statcoulomb (esu) statC (esu) 3.33564xlQ'10 C Dipole Moment 1 debye D 3.33564x1 O'30 Cm Force 1 dyne dyne 10‘5N Mass 1 atomic mass unit amu 1.661x1 O'27 kg 3 Polarizability 1 cubic centimeter cm 1.113xl0"16 Cm/V Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 145 REFERENCES [1] D. K. Bohme, Int. J. Mass Spectrom. 200, 97 (2000). [2] E. E. Ferguson, F. C. Fehsenfeld, and A. L. Schmeltekopf, Advances in Atomic and Molecular Physics, 5 ed. by D. R. Bates and I. Estermann, 1 (Academic Press, New York, 1969). [3] M. Henchman, Ion-Molecule Reactions, 1 ed. by J. L. Franklin, 101 (Plenum Press, New York, 1972). [4] P. H. G. Dickenson and J. Sayers, Proc. Phys. Soc. 7 6 , 137 (1960). [5] N. G. Adams and D. Smith, Int. J. Mass Spectrom. Ion Phys. 21, 349 (1976). [6] D. Smith and N. G. Adams, Advances in Atomic and Molecular Physics, 24 ed. by D. R. Bates and B. Bederson, 1 (Academic Press, New York, 1988). [7] D. Smith and N. G. Adams, Gas Phase Ion Chemistry, 1 ed. by M. I. Bowers, 1 (Academic Press, New York, 1979). [8] D. Smith, N. G. Adams, and E. Alge, Planet. Space Sci. 29, 449 (1981). [9] D. Smith and P. Spanel, Acc. Chem. Res. 2 5 ,414 (1992). [10] D. Smith, Chem. Rev. 92,1473 (1992). [11] W. Lindinger, Gaseous Ion Chemistry and Mass Spectrometry, ed. by J. H. Futrell, 237 (John Wiley & Sons, New York, 1986). [12] P. Spanel, M. Pavlik, and D. Smith, Int. J. Mass Spectrom. Ion Proc. 145,177 (1995). [13] D. Smith and P. Spanel, Int. Rev. Phys. Chem. 1 5 ,231 (1996). [14] R. F. Service, Science 281, 1431 (1998). [15] P. Spanel, P. Rolfe, B. Raj an, and D. Smith, Ann. Occup. Hyg. 40, 615 (1996). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 146 [16] W. Lindinger, A. Hansel, and A. Jordan, Int. J. Mass Spectrom. Ion Proc. 17 3 , 191 (1998). [17] D. Smith and P. Spanel, Rapid Commun. Mass Spectrom. 1 0 ,1183 (1996). [18] Y. Dcezoe, S. Matsuoka, M. Takebe, and A. Viggiano, Gas Phase Ion-Molecule Reaction Rate Constants Through 1986 (Maruzen Company, Japan, 1987). [19] V. G. Anicich, J. Phys. Chem. Ref. Data 22,1469 (1993). [20] H. Eyring, J. O. Hirschfelder, and H. S. Taylor,J. Chem. Phys. 4,479 (1936). [21] T. Su and M. T. Bowers, Gas Phase Ion Chemistry, 1 ed. by M. T. Bowers (Academic Press, New York, 1979). [22] R. B. Shirts, Gaseous Ion Chemistry and Mass Spectrometry, ed. by J. H. Futrell (John Wiley & Sons, New York, 1986). [23] J. B. Marion and S. T. Thornton, Classical Dynamics of Particles & Systems, Third Edition (Harcourt Brace Jovanovich College Publishers, Texas, 1988). [24] N. Boelnijk and W. H. Hamill, J. Am. Chem. Soc. 84, 529 (1962). [25] T. F. Moran and W. H. Hamill, J. Chem. Phys. 64, 2753 (1963). [26] T. Su and M. T. Bowers, Int. J. Mass Spectrom. Ion Phys. 12, 347 (1973). [27] T. Su and W. J. Chesnavich, J. Chem. Phys.76, 5183 (1982). [28] W. J. Chesnavich, T. Su, and M. T. Bowers, J. Chem. Phys.72, 2641 (1980). [29] D. R. Lide, CRC Handbook of Chemistry and Physics, 80th Edition (CRC Press, New York, 1999). [30] J-F Gal, P-C Maria, and E. D. Raczynska, J. Mass Spectrom.36, 699 (2001). [31] D. K. Bohme, G. I. Mackay, and H. I. Schiff, J. Chem. Phys. 7 3 , 4976 (1980). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 147 [32] W. Lindinger, D. L. Albritton, F. C. Fehsenfeld, A. L. Schmeltekopf, and E. E. Ferguson, J. Chem. Phys. 62, 3549 (1975). [33] R. C. Bolden, R. S. Hemsworth, M. J. Shaw, and N. D. Twiddy, J. Phys. B. 3, 45 (1970). [34] N. G. Adams, M. J. Church, and D. Smith,J. Phys. D 8,1409 (1975). [35] D. Smith, N. G. Adams, A. G. Dean, and M. J. Church, J. Phys. D8 , 141 (1975). [36] W. Lindinger, Physical Review A 7, 328 (1973). [37] F. Howorka, W. Lindinger, and M. Pahl, Int. J. Mass Spectrom. Ion Phys. 12, 67 (1973). [38] G. Brooke, Direct Current Hollow Cathode Discharge for Proton Donor Production (unpublished, M.S. Thesis, Old Dominion University, Norfolk, Virginia, 1997). [39] Y. P. Raizer, Gas Discharge Physics, ed. by J. E. Allen (Springer, Berlin, 1991). [40] S. C. Brown, Introduction to Electrical Discharges in Gases (Wiley, New York, 1966). [41] M. T. Ngo, The Breakdown Mechanism, o f Hollow Cathode Discharge (unpublished, M.S. Thesis, Old Dominion University, Norfolk, Virginia, 1991). [42] Y. P. Raizer, M. N. Shneider, and N. A. Yatsenko, Radio-Frequency Capacitive Discharges (CRC Press, New York, 1995). [43] S. J. Hill,Inductively Coupled Plasma Spectrometry and its Applications (CRC Press, New York, 2000). [44] A. Hansel, A. Jordan, R. Holzinger, P. Prazeller, W. Vogel, and W. Lindinger, Int. J. Mass Spectrom. Ion Proc. 149/150, 609 (1995). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 148 [45] S. T. Groul and R. R. Squires, Mass Spectrom. Rev. 7 , 263 (1988). [46] F. C. Fehsenfeld, K. M. Evenson, and H. P. Broida, Rev. Sci. Instr. 3 6 ,294 (1965). [47] J. Asmussen, Jr., R. Mallavarpu, J. R. Hamann, and H. C. Park, Proceedings of the IEEE 62, 109 (1974). [48] R. Mallavarpu, J. Asmussen, and M. C. Hawley, IEEE Trans, on Plasma Sci. PS-6, 341 (1978). [49] L. G. Meiners and D. B. Alford, Rev. Sci. Instrum. 57, 164 (1986). [50] D. N. Walker, D. Duncan, J. A. Stracka, J. H. Bowles, C. L. Siefring, M. M. Baumback, and P. Rodriguez, Rev. Sci. Instrum. 65, 661 (1994). [51] H. Kako, T. Nakagawa, and R. Narita, IEEE Trans, on Cons. Elec. 37, 611 (1991). [52] G. H. Dieke and H. M. Crosswhite, J. Quant. Spectrosc. Radiat. Transfer 2, 97 (1962). [53] Kelleher, J. Quant. Spectrosc. Radiat. Transfer 2 5 ,191 (1981). [54] S. R. Goode and J. P. Deaver, Spectrochim. Acta 39B, 813 (1984). [55] C. Thomsen and V. Helbig, Spectrochim. Acta 46B, 1215 (1991). [56] S. Jovicevic, M. Ivkovic, Z. Pavlovic, and N. Konjevic, Spectrochim. Acta 55B, 1879(2000). [57] I. M. Podgomiy, Topics in Plasma Diagnostics (Plenum Press, New York, 1971). [58] A. Rodero, M. C. Garcia, M. C. Quintero, A. Sola, and A. Gamero, J. Phys. D: Appl. Phys. 29,681 (1996). [59] T. M. Miller, R. E. Wetterskog, and J. F. Paulson, J. Chem. Phys. 80,4922 (1984). [60] E. W. McDaniel, Collision Phenomena in Ionized Gases (John Wiley & Sons, New York, 1964). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 149 [61] G. Dupeyrat, B. R. Rowe, D. W. Fahey, and D. L. Albritton, Int. J. Mass Spectrom. Ion Phys. 44,1 (1982). [62] V. N. Fishman and J. J. Grabowski, Int.J. Mass Spectrom. 177, 175 (1998). [63] D. B. Milligan, D. A. Fairley, C. G. Freeman, and M. J. McEwan, Int. J. Mass Spectrom. 202, 351 (2000). [64] P. H. Dawson, Quadruple Mass Spectrometry and its Applications (Elsevier Scientific Publishing Company, New York, 1976). [65] D. K. Bohme and G. I. Mackay, J. Am. Chem. Soc. 103,2173 (1981). [66] G. I. Mackay, M. H. Lien, A. C. Hopkinson, and D. K. Bohme, Can. J. Chem. 56, 131 (1978). [67] P. Ausloos and S. G. Lias, Int. J. Mass Spectrom. Ion Proc. 5 8 , 165 (1984). [68] D. K. Bohme, G. I. Mackay, and S. D. Tanner, J. Am. Chem. Soc. 101, 3724 (1979). [69] K. Tanaka, G. I. Mackay, and D. K. Bohme, Can. J. Chem. 56, 193 (1978). [70] R. J. Shul, R. Passarella, L. T. DiFazio, Jr., R. G. Keesee, and A. W. Castleman, Jr., J. Phys. Chem. 92,4947 (1988). [71] P. Brimblecombe, Air Composition and Chemistry (Cambridge University Press, Cambridge, 1995). [72] E. Meszaros, Global and Regional Changes in Atmospheric Composition (Lewis Publishers, New York, 1993) [73] J. M. Samet, F. Dominici, F. C. Curriero, I. Coursac, and S. L. Zeger, N. Engl. J. Med. 343, 1742 (2000). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 150 [74] R. Atkinson, Air Pollution, the Automobile, and Public Health, 99 (National Academy Press, Washington, D. C., 1988). [75] G. M. Solomon, T. R. Campbell, T. Carmichael, G. Ruderman Feuer, and J. S. Hathaway, Exhausted By Diesel (Report for NRDC, 1998). [76] S. T. Bagley, K. J. Baumgard, L. D. Gratz, J. J. Johnson, and D. G. Leddy, Characterization of Fuel and Aftertreatment Device Effects o f Diesel Emissions, Research Report Number 76 (Health Effects Institute, Topsfield, Massachusetts, September 1996). [77] J. P. Shi and R. M. Harrison, Environ. Sci. Technol. 33, 3730 (1999). [78] P. Spanel and D. Smith, Int. J. Mass Spectrom. 18 1 , 1 (1999). [79] M. Phillips, Sci. Amer. 74 (1992). [80] A. Manolis, Clin. Chem. 29, 5 (1983). [81] M. Phillips and J. Greenberg, Clin. Chem. 3 8 , 60 (1992). [82] M. W. Felz, G. J. Burke, and B. M. Schuman, JABFP10, 385 (1997). [83] P. Spanel and D. Smith, Med. Biol. Eng. Comput. 34,409 (1996). [84] P. W. Steinwedel, Z. Naturforsch. 8A, 448 (1953). [85] E. Mathieson and T. J. Harris, Am. J. Phys.3 7 , 1054 (1969). [86] E. W. Blauth, Dynamic Mass Spectrometers (Elsevier Publishing Company, Amsterdam, 1966). [87] J. Roboz, Introduction to Mass Spectrometry Instrumentation and Techniques (Interscience Publishers, New York, 1968). [88] N. W. McLachlan, Theory and Application ofMathica Functions (Dover Publications, New York, 1964). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 151 [89] W. Paul, H. P. Reinhard, and U. von Zahn, Z. Physik 152,143 (1958). [90] I. E. Dayton, F. C. Shoemaker, and R. F. Mozley, Rev. Sci. Instr. 25, 485 (1954). [91] J. D. Jackson,Classical Electrodynamics (John Wiley & Sons, New York, 1962). [92] M. Abramowitz and I. A. Stegirn, Handbook of Mathematical Functions with Formulas, Graphs, and Mathematical Tables (Dover Publications, New York, 1965). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 152 VITA NAME George Mercer Brooke IV ADDRESS United States Air Force Academy, Department of Physics, Laser and Optics Research Center, USAFA, CO 80840 EDUCATION May 2003 Ph.D., Physics, Department of Physics, Old Dominion University, 4600 Elkhom Ave, Norfolk, VA 23529 Dissertation: Measurement of Proton TransferReaction Rates in a Microwave Cavity Discharge Flowing Afterglow. December 1997 M.S., Physics, Department of Physics, Old Dominion University, 4600 Elkhom Ave, Norfolk, VA 23529 Thesis: Direct Current Hollow Cathode Discharge for Proton Donor Production May 1994 B.S., Physics, Department of Physics, Virginia Military Institute, Lexington, VA 24450 HONORS AND AWARDS 1999-2002 Graduate Teaching Fellowship, Virginia Military Institute, Lexington VA. 1998 Teaching Assistant of the Year, Department o f Physics, Old Dominion University, Norfolk VA. 1994 Stonewall Jackson Award for Physics, Virginia Military Institute, Lexington VA. PROFESSIONAL MEMBERSHIP American Physical Society Sigma Xi Scientific Research Society Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.