{Rl' 1 State _ RETURNING MATERIALS: )V1ESI.} Place in book drop to LIBRARJES remove this checkout from “ your record. FINES will 7 be charged if book is returned after the date stamped below. {'9‘ng Efi CEW FUNDAMENTAL STUDIES AND ANALYTICAL APPLICATIONS OF THE DUAL LASER IONIZATION TECHNIQUE By Francis Michael Curran A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Chemistry 1983 ABSTRACT FUNDAMENTAL STUDIES AND ANALYTICAL APPLICATIONS OF THE DUAL LASER IONIZATION TECHNIQUE by Francis M. Curran Laser-enhanced ionization (LEI) is a recently developed flame spectrometric technique for the analysis of trace metals in flames. In the LEI experiment the analyte is laser-promoted into an excited state. This selectivity enhances the collisional ionization rate of the analyte, and the subsequent ionization signal is collected by a set of biased probes. Dual laser ionization (DLI) is a modification of this technique in which a separate laser is used to ionize the analyte after the resonant excitation step. Under proper conditions this can substantially reduce the collisional dependence of the ionization signal and significantly enhance signal magnitudes. In the studies reported here, a nitrogen laser is used both to pump the dye laser used to resonantly excite the analyte and to provide the ionizing beam. The pulsed nature of this laser system necessitates gated integration. Four major aspects of the DLI technique are considered in order to characterize the technique more completely. First, the energetics involved in the ionization processes in both LEI and DLI are discussed and compared in order to Francis M. Curran determine in which instances DLI can be used effectively to improve the ionization signals obtained with LEI. A close match between ionizing beam photon energy and the difference in energy between the excited state level and the ionization continuum (1.3; the energy defect) is found to be the major factor involved in signal enhancement. The dependence of the ionization signal on laser power in both LEI and DLI is examined. A simple two level model is developed to describe experimental results. It is shown that in cases where DLI enhancement is observed, further enhancement could have been obtained with a more powerful ionizing beam. The LEI technique has been shown to be adversely affected by electrical interferences caused by easily ionized components in the sample matrix. This type of interference is discussed with reSpect to the DLI technique. As in LEI, electrode positioning is found to be very important to signal recovery. Finally the application of the DLI technique to real sample analysis is described. Results consistent with those obtained in atomic absorption experiments are presented to illustrate the analytical utility of the technique. To Mom, to Dad, and to Shelley -11- ACKNOHLEDGMENTS Simple acknowledgment of Professor S.R. Crouch for his guidance as research director is quite insufficient here. In addition to being a scientist who has taught the author much, he has been a true friend. Thanks go to Profesor John Allison for his work as second reader and for his stinging wit. I am also indebted to Professors George Leroi and Paul Hunt for the many hours they spent in scientific discussion with me. The help and patience of Marty Rabb, Dr. Tom Atkinson and Russ Geyer were greatly appreciated. Thanks also go to Dick Menke, Ben Stutsman and Deak Hatters in the Machine Shop and to Ron Haas and Scott Sanderson in the Electronics Shop as they were all instrumental in the completion of this work. Margy Lynch, friend and typist. and Naomi Hack, who kept me well fed, informed and awake, were also very helpful. The author expresses his gratitude to the Department of Chemistry and to the Dow Chemical Corporation for providing financial support. In my years at MSU I have been lucky enough to meet and work with a number of pe0ple who have become my close friends. I doubt a better group could be found anywhere. -iii- Thanks to Rob and Jan, Charlie. Gene, Rytis, Lynn, John, Phil, Hugh, Scott, Pete, Bruce, Nelson, Tony, Pat, Kieth, Paul, Max and Bob. Special mention must be made of Pat Ronan, Clay Calkin, Jim Gano, Susan Alden and April Lockner; although my liver wishes I had never met you, my heart does not. Special thanks also must go to Marguerite Danna (?) and Kim Ratanathanawongs (???). I am sorry to leave you at John's mercy but at least now you are safe from flying fruit and sneak attack. Finally, I would like to thank my wife, my cats and my family for their warmth and support. The love you all have shown me has been much more than I have deserved (even if I did color well, Auntie). Bo Schembechler is also gratefully acknowledged. Four U of M victories over MSU have saved one loudmouth a lot of grief. -‘iv- Chapter II III IV TABLE OF CONTENTS LIST OF TABLES. LIST OF FIGURES. INTRODUCTION. A. 8. Overview of Dissertation. C. Historical. . . . l. The Opto- galvanic Effect in Non- flame Cells. . 2. Laser Enhanced Ionization (LEI). 3. Dual Laser Ionization (DLI) and Resonance Ionization (RIS). EXPERIMENTAL A. Overview. 8. Laser System. . . C. Flame Systems and Sample Introduction. D Signal Detection and Processing. Overview of Opto- galvanic Effect. 1. Ionization Signals. 2. Optical Signals. ENERGETIC CONSIDERATIONS IN DUAL LASER IONIZATION. . A. Introduction. 3. Experimental. 1. Apparatus. 2. Reagents. . 3. Charge Sheath Avoidance. C. Ionization Mechanisms. D. Results and Discussion. LASER PONER DEPENDENCE IN DLI AND LEI. A. Introduction. 8. Theory. 1. Ionizing Laser Sampling Functions.. 2. Steady State Case. . . 3. Non-Steady State Case. Page .xiii Chapter Page C. Experimental. . . . . . . . . . . . . . . . .70 l. Apparatus. . . . . . . . . . . . . . . . 7O 2. Reagents. . . . . . . . . . . . . 7l 3. Laser Attenuation. . . . . . . . . . . . 7l 4. Experimental Procedure. . . . . . . . . .71 D. Results and Discussion. . . .73 Effect of Ionizing Beam Power in DLI. . .74 Dye Laser Power Dependence in DLI. . . . 78 Dye Laser Power Dependence in LEI. . . . 83 Off-Resonant Photoionization. . . . . . .88 wa-i o o o e V DUAL LASER IONIZATION IN FLAMES: A SEARCH FOR IONIZATION INTERFERENCES. . . . . . . . . . . . .91 A. Introduction. . . . . . . . . . . . . . . . .9l 8. Experimental. . . . . . . . . . . . . . . . .92 1. Apparatus. . . . . . . . . . . . . . . . 92 2. Reagents. . . . . . . . . . . . . . . .92 3. Data Treatment. . . . . . . . . . . . . .93 C. Results and Discussion. . . . . . . . . . . .94 D. Conclusions. . . . . . . . . . . . . . . . .109 VI ANALYTICAL APPLICATIONS. llO A. Introduction. llO 8. Experimental. . . lll l. DLI Apparatus. . . Ill 2. Atomic Absorption Apparatus. lll 3. Reagents. . . lll 4. Samples. . ll2 5. Standard Additions Procedure. llZ C. Results. . 115 l. Hell Hater Sample Analysis. ll5 2. Blood Serum Sample Analysis. 115 VII SUMMARY AND FUTURE DIRECTIONS. l24 A. Summary. . l24 8. Future Directions. l26 REFERENCES. 130 APPENDIX A -TNO- STEP LASER ASSISTED IONIZATION OF SODIUM IN A HYDROGEN- OXYGEN- ARGON FLAME. . l39 Introduction. . . . . l39 Experimental. . . . . . . . . .l4O Results and Discussion. . . . . . . . .144 Conclusions. . . . . . . . . . . . . . 156 Appendix. . . . . . . . . . . . . . . .l58 monar- o o 0 0 .v1. Chapter APPENDIX B - DYE LASER DESCRIPTION AND ALIGNMENT PROCEDURE. . . . . . A. Introduction. . 8. Optical System Components. C. Alignment Procedure. APPENDIX C - LOCK- IN AMPLIFIER DESIGN. A. Introduction. . 8. Design Considerations. l. General. . . 2 Inputs. . . 3 Active Filters. 4. Phase Shifter. 5. Multiplier. . 6 Low Pass Filter. 7. Outputs. C. Testing. -vii- Chapter APPENDIX B - DYE LASER DESCRIPTION AND ALIGNMENT PROCEDURE. . . . . A. Introduction. . 8. Optical System Components. C. Alignment Procedure. APPENDIX C - LOCK- IN AMPLIFIER DESIGN. A. Introduction. . . 8. Design Considerations. l. General. . . 2 Inputs. . 3 Active Filters. 4. Phase Shifter. 5. Multiplier. . . 6 Low Pass Filter. 7. Outputs. C. Testing. -vii- Page .162 °.152 .163 .167 173 2173 .174 .174 '. 177 177 °.180 180 ..l81 .181 .181 LIST OF TABLES Table Page 3-1 Laser Characteristics ................... 40 3-2 Collection of Numerical Results and Useful Spectroscopic Data .................... 47 4-1 Neutral Density Filter Characteristics .......... 72 6-1 Solutions for Nell Hater Sample Experiments ........ 114 6-2 Solutions for Blood Serum Sample Experiments ....... 114 6-3 Comparison of Results of DLI and AA Analyses of Ca in Hell Water Sample .................. l18 6-4 Comparison of Results of DLI and AA Analyses of Ca in Blood Serum Sample ................. 120 C-l Frequency Ranging Data. . . . . . . . . . ........ 17g -viii- Figure 2-1 2-2 2-3 2-4 2-5 2-7 2-8 2-9 LIST OF FIGURES Page Simplified schematic of LEI/DLI system .......... 18 Block diagram of ionization signal collection circuit. . . 19 Typical time resolved ionization signal (Na as analyte) ......................... 21 Temporal overlap and pulse shapes of the N2 (I) and dye (O) lasers ................... 24 Mantle flame flowmeter calibration data (O)Ar;(A)H2;(O)02 ............... 26 Nebulizer design ..................... 28 Ionization signal collection circuitry .......... 29 Frequency response of the ionization collection circuit for (a) current amplifier configuration and (b) filter configuration. (13,) 2 M9 feedback resistor; (O) 500 kn feedback resistor; (V) 100 kg feedback resistor; (<3>) so k9 feedback resistor; (D) 10 k9 feedback resistor .......... 33 PIN photodiode circuitry ................. 37 Common ionization pathways in laser atomic spectrosc0py ....................... 43 -ix- Figure 3-2 4-1 4-2 4-3 4-4 4-5 4-6 4-7 4-8 4-9 5-2 Ionization signal versus wavelength for DLI (upper trace) and LEI (lower trace). Dye laser scanned across the Li(2s-v2p) transition at 670.7 nm. To obtain the LEI signal, the sensitivity (vertical scale) was increased by a factor of 5 and the analyte concentration by a factor of 100 ....... Simplified atomic level scheme ............ Dependence of DLI on ionizing laser power (Li as analyte) ....................... Dependence of DLI on dye laser power (Li as analyte). (O) ionizing laser at full power; (A) ionizing laser at half power .......... Dependence of DLI on dye laser power (Li as analyte; extended dye laser power range) ....... Dependence of DLI on dye laser power (Sr as analyte) ........................ Dependence of DLI on dye laser power (Ca as analyte) ....................... Dependence of LEI on dye laser power (Na as analyte) ....................... Dependence of LEI on dye laser power (K as analyte). . . . Off-resonant photoionization of Cs .......... Percent signal recovery for Na versus Cs concentra- tion, DLI (A) and LEI (0) cases. . . . . ..... Percent signal recovery for Li (A ) and Sr (0) versus Cs concentration, DLI case ........... Page 51 64 75 77 79 80 81 85 86 89 97 --100 Figure 5-3 5-4 5-5 5-6 5-7 6-1 6-2 6-3 A-1 Analytical curve for Na in the presence of 1000 ug/g Cs ...................... Percent signal recovery for Na versus K (O) and Li (A) concentration, DLI case ............ Percent signal recovery for Na versus Mg (A ) and Zn (0) concentration, DLI case, metals dissolved in HCl followed by neutralization ............ Percent signal recovery for Na versus Mg ([3) and Zn (0) concentration, DLI case, metals from hydrated sulfate salts .............. Percent signal recovery for Na versus sulfate ion concentration (from ammonium salt), DLI case ...... Standard additions analysis of Ca in well water sample (DLI) ...................... Standard additions analysis of Ca in well water sample (AA) ...................... Standard additions analysis of Ca in blood serum sample (DLI). (0) trial 1; (A) trial 2; (Cl) trial 3 ..................... Standard additions analysis of Ca in blood serum sample (AA). (A) nitrous oxide-acetylene flame; (0) air -acety1ene flame ........... Experimental arrangement for simultaneous observation of laser assisted ionization and laser induced fluorescence. a) Overall view. B, boxcar; DL, dye laser; F, flame, L, lens; M, mirror, -xi- Page 102 104 . 105 106 108 116 . . 117 . . 119 . . 122 Figure A-Z A-3 A-4 Page 0, oscillosc0pe; PL, pump laser; PM, photomulti- plier, R, x-t recorder, S, signal; T, uv rejecting filter, TL, trigger line. b) Detailed view of probes and laser beams. IF, inner flame; OF, outer flame; P, probe; Rt’ load resistor, S, signal. Dimensions are not to scale. The waist diameter of the laser beams is of the order of 0.1 mm; the distance between the probes is a few mm; the diameter of the inner flame is approxi- mately 15 m ....................... 142 Ionization signal (0) and ratio of ionization signal to sodium solution concentration (D) vs. sodium solution concentration; both lasers are irradiating the probed volume. The dye laser was tuned to the 351/2+3P3/2 (589.0 nm) transition ....... 146 Profile of ionization signal ([3) and fluorescence signal (0) obtained by scanning dye laser across the 351/2 to 3P3],2 (589.0 nm) Na transition. FNMMsOJA for fluorescence profile and 1.5A for ionization profile. Sodium solution concentration= 12 ug/mL, probe voltage= 400 V ....... 148 Dependence of ionization and fluoresence signals on Ar content of flame. The dye laser was tuned to Na(3Sy&-3P 2) transition; the sodium concen- 3’ tration was 12 pg/ml. (O) fluoresence signal -xii- Figure A-5 A-6 8-1 C-1 C-3 plotted vs. Ar flow; (0) ionization signal plotted vs. Ar flow; (A ) ratio of ionization- to-fluorescence signal plotted vs. Ar flow ........ Dependence of ionization signal on probe voltage for various sodium solution concentrations. The dye laser was tuned to the 351/2-v3P3/2 transition. A dc component in the signal, probably due to thermal (non-laser) ionization, was eliminated with a capacitor in the signal line. (O) sodium concentration= 10 ppm; (A ) sodium concentration=1 ppm; (I) sodium concentration = 100 ppb ........................ Ionization profiles obtained by scanning the dye laser across the Na(3S->4D) two-photon transition; center wavelength = 5787 A, R2= SO kn, 2% = 100 V; sodium solution concentration-100 ug/ml. a) Upper trace: N2 laser beam interrupted; lower trace: both the N2 laser beam and the dye laser beam operative. b) Detailed profile ............. Block diagram of laser system .............. Simplified circuit diagram for lock-in amplifier. . . .. Detailed circuit diagram for lock-in amplifier ...... Lock-in amplifier without noise. (a) lower trace: signal channel input; upper trace: output of tuned amplifier. (b) lower trace: output of multiplier; upper trace: output of reference -xiii- Page 152 153 157 164 175 176 Figure 0-4 Page channel. (c) lower trace: dc output with signal "blocked"; upper trace: dc output with signal "open" .......................... 184 Lock-in amplifier test with noise. (a) lower trace: signal channel input; upper trace: output of tuned amplifier. (b) lower trace: output of multiplier; upper trace: output of reference channel. (c) lower trace: dc output with signal "blocked"; upper trace: dc output with signal "open" .................. . ........ 186 -xiv- CHAPTER I INTRODUCTION A. Overview of the Opto-galvanic Effect. The irradiation of an analyte-containing sample reservoir (most commonly a flame or discharge plasma) by a light source at a wavelength corresponding to a resonant transition of the analyte may yield an observable opto- galvanic signal. The opto-galvanic effect (OGE) is pro- duced by a change in the ionization rate within the sample cell and can be detected by measuring a change in some electrical property of the cell such as its resistance, conductance or the ion current detected by biased probes. The first observation of the OGE was made by Foote and Mohler in 1925]. They used space-charge amplification in a thermionic diode to detect the ionization of C5 at reson- ant wavelengths below the photoionization threshold. Their light source was a tungsten lamp, and a monochromator was used for wavelength selection. Shortly thereafter, Penning measured the change in current through a neon dis- charge tube upon irradiation by another neon dischargez. -1- -2- Although these early experiments pointed out the existence of the phenomenon, little activity in the field of opto- galvanic spectroscopy occurred over the next 50 years because of the limited signal magnitudes obtainable with conventional light sources. This changed with the intro- duction of tunable dye lasers into the laboratory. Lasers gave spectroscopists convenient and versatile light sources with spectral irradiances capable of saturating many atomic transitions, especially in the visible region of the spectrum. The OGE, with laser excitation, was in- troduced in experiments at the National Bureau of Stan- 3.4 5’6. Observations of the dards and in the Netherlands effect in flames has been termed laser-enhanced ionization (LEI); in the recent past it has become one of the most sensitive means available for the detection of trace metals in flames7‘14. Briefly, in LEI, the absorption of resonant laser photons lowers the energy necessary for ionization and selectively increases the collisional ioniza- tion rate of the analyte. The increase can be measured by a set of biased probes placed near the laser irradiated region of the flame. Electrical detection is the major advantage of LEI; the usual limitations of conventional optical detection, such as finite solid angle of collec- tion optics, quantum efficiencies less than unity, and scattered light, are absent. Recently, we introduced a related technique called dual laser ionization (DLI) in which a second laser beam -3- provides the energy necessary for ionization from the energy level populated by a dye laser‘s. The second laser decreases the collisional dependence of ion formation by providing direct photoionization in some cases. The DLI method shows promise in both trace metal determinations 15-17. The characterization and flame diagnostic studies of the DLI technique is the major focus of this disserta- tion. B. Overview of Dissertation A detailed account of the development of LEI and other related laser-based ionization techniques (e.g. resonance ionization spectroscopy (RIS)) from their incep- tion to the middle of 1982 has been given recently by Lin18. In addition, most of the following chapters include introduction sections which detail the pertinent back- ground information. Because of this, the historical discussion that immediately follows this section centers on develOpments from 1982 to the present and reviews earlier developments only briefly. Chapter II describes the experimental system used to obtain the data presented in the main body of the disser- tation. An overview of the system points out the types of signals generated and rationalizes the equipment used. The sections that follow contain more detailed information on specific components. -4- The next four chapters describe the studies which were performed to characterize the DLI technique. Chapter III elucidates the efficacy-determining aspects of DLI ioniza- tion schemes. This begins with a review of the ionization mechanisms common to laser atomic spectroscopy and includes a presentation of pertinent rate equations. Results from experimental studies of atomic Li, Na, K, Ca, Sr, In and Cs in an HZ-Oz-Ar flame are then presented followed by a discussion of the experimental and theoretical aspects of the enhancements obtained upon addition of the ionizing beam. Strategies for optimizing the technique are also included. Chapter IV presents studies of the dependence of the ionization signal on laser power in both LEI and DLI. A simple two level model is developed to explain experimental results. Optimization of the DLI signal with respect to laser power is discussed. Chapter V presents a study of interferences in DLI. Electrical interferences caused by easily ionized matrix components have been a major problem in LEI; this chapter explores the nature of the interferences observed and the methods to avoid them. Some physical interferences which are encountered are also discussed. In Chapter VI, the DLI technique is extended to a more analytically useful air-acetylene flame. The experiments described, with Ca as an analyte, serve not only to show the utility of DLI as an analytical technique, but also to -5- test a computerized boxcar integration and data acquisi- tion system designed and built by J. Stanleng. Finally in Chapter VII, the work is summarized and recommendations for future studies are made in the hope that future workers can make the technique more generally useful. There are three appendices to the main body of the dissertation. The first involves discussion of basic DLI characteristics and presents early results obtained with C.A. wanDijk and K.C. Lin. Appendix II details the dye laser alignment procedure developed for the present exper- imental system. The final appendix contains details of the design, construction and testing of an inexpensive yet versatile lock-in amplifier, useful over the l-l0,000 Hz frequency range. This amplifier was constructed to be used with a commercial atomic absorption unit and may also find application in future atomic emission studies comple- mentary to DLI. C. Historical Two articles reintroduced the OGE with laser excita- tion in 19763’4. One of these detailed the detection of metal species in flames with biased probes. This technique has become known as laser enhanced ionization (LEI). The other paper described galvanic detection of optical absOrp- tions of laser light by species contained in a discharge tube. Since this date, research in the field of -5- opto-galvanic spectroscopy (OGS) has been Split fairly evenly between experiments in which these two types of 18 and Goldsmith and sample cells are used as both Lin Lawler20 have pointed out in their recent reviews. C-l. The Opto-galvanic Effect in Non-flame Cells 20 have noted that non-flame OGS Goldsmith and Lawler can be used in experiments requiring either moderate or high (Doppler-free) resolution. In moderate resolution experiments, typically, a single laser is operated at the resonant frequency of some species contained in a dis- charge. This produces an OGE which is then measured by observing the change in voltage across the discharge or the change in current through the discharge ballast resis- tor. Both modulated continuous wave (CN) lasers with lock-in detection and pulsed lasers with gated detection 21'23 and molecular“.25 have been used to detect atomic species. Many modifications of this basic experimental design were made between 1976 and 1982. For example, Suzuki used a pick-up coil to observe laser-induced current changes across radio frequency discharges of N2, Ar, 26 NH3 and N02 Numerous authors have used more than one laser to step sequentially from one atomic transition to another and thus to examine high lying excited state527'30. This type of excitation has also been used in LEI and will hereafter be referred to as stepwise excitation. Though early studies mainly emphasized the examination of -7- fundamental processes in the discharge, numerous practical applications were found. One of the earliest was the use of the OGE by Green, gt_gl. to frequency lock a CH dye laser31. Similarly, Smith and Moffat used the OGE to fre- quency stabilize a C02 1aser32. Keller, gt_gl.33 and Palmer g£;31,34 have discussed the use of the OGE in wave- length calibration. In more recent studies the OGE has 35-37 38 been used to frequency lock CO2 , He-Ne and tunable diode39 lasers. Dovichi and co-workers have continued work on wavelength calibration_of pulsed lasers using the OGE in uranium hollow cathode lamps4o. Along this same line, two groups have discussed the use of the OGE in spectro- sc0pic line profile measurements and in wavelength calibration. Bachor and co-workers found opto-galvanic and absorption measurements to be similar under proper 41 experimental conditions They also found a nonlinear relation between integral measurement and oscillator strength for Ca. Piepmeier and Beenan followed up an earlier paper on line profile distortion in OGE measure- 42 with a similar paper on saturation and prefilter effects43. ments Both color center and tunable diode lasers operating in the infrared have been used quite extensively to study Rydberg transitions in atomic species“"'47 and rovibronic levels in molecules48. -3- Early mechanistic studies by Keller and Zalewski49 indicated that direct collisional ionization following resonant laser excitation is not the major mechanism involved in opto-galvanic signal production in hollow cathode lamps. Instead, they suggested that absorbed laser energy is transferred to the electron gas in the discharge and then equally distributed throughout the discharge. Recently, Dreze and co-workers have performed very elegant experiments which have supported the above hypothesisso. In these experiments emission from all species (even the Xe fill gas) was found to increase upon irradiation of a uranium hallow cathode with laser light resonant with a U(I) transition. Also, excitation of uranium ions caused intensity increases in atomic uranium emission lines. These findings also indicate that isotOpe separation using the OGE in hollow cathode lamps is likely to be impossible, a conclusion supported by further experiments described in the same work. This does not, however, rule out completely isotope enrichment in other plasmas in which this type of energy redistribution does not occur. Kushner has recently discussed this tapic for the case of Cu in a Cu-Ne positive column discharges}. Other recent mechanistic studies 52 include metastable quenching in a He discharge and collisional mixing in a Ne discharge53. Several miscellanseous experiments have also been performed. Transitions in HCO have been studied in an rf -g- discharge by Vasudev and Zare54. Electron photodetachment 55 to from I' has been studied in an iodine discharge obtain electron affinities. Also, two-color, double resonance spectroscopy has been recently applied to the study of highly excited states in the NO molecule in a gas ce1156. Intermodulation techniques are usually used in Doppler-free high resolution spectroscopy. with opto- galvanic detection this has been termed IMOGS. Typically, a single laser beam is split, and the two beams are chopped at different frequencies (f1 and f2). These beams are then directed to counter propogate in a discharge. If the wave- length is tuned across a resonance of interest and lock-in detection is used at the sum frequency (fI'tfz), the Doppler-free apto-galvanic signal is observed (cf. 20 and references therein). Rotating polarizers have also been used in place of chOpper557. Another method for obtaining Doppler-free spectra involves absorption of two photons to excite a one-photon forbidden transition. The laser beam is reflected back upon itself through the discharge so that Doppler effects cancel. This technique has been termed TOGS (3:. 20,58). Recent studies in this area have expanded these techniques. Transitions in N2 discharges have been studied by two groups. Suzuki and Kakimoto have studied Rydberg states in N2 using IMOGS with an rf 59 discharge Miyazaki, et al., used two separate CW lasers, -10- one a ring dye laser modulated at 550 Hz and the other a linear dye laser modulated at 825 Hz, to obtain the Doppler-free spectraso. Zeeman splitting has also been observed with opto-galvanic detection. In one experiment, researchers simply used the IMOGS technique with a hollow cathode discharge immersed in a magnetic fields]. In another a single laser beam modulated at one frequency was used to irradiate a discharge in a magnetic field modulated at another frequency. The opto-galvanic signal measured by a phase sensitive detector operating at the sum frequency showed the sought-after Zeeman splittingssz. C-2. Laser Enhanced Ionization (LEI) As stated previously, the first observation of the opto-galvanic effect in a flame was made in 19763. The basic principles of the LEI technique have been described in a number of publications since this time7’8’11’20’63'57. Briefly, a laser is tuned to excite at a resonant frequency of a species contained in a flame cell. This selectively increases the ionization rate of that species, and this increase is detected by a set of biased probes near the irradiated volume of the flame cell. The first study was performed using a chopped CN dye laser and Na as the analyte. Following this, three studies appeared in succes- sion, each advancing the state of the art. In each study, flashlamp pumped dye lasers were used to increase the spectral irradianceand the tuning range. The first14 -1]- indicated that excellent detection limits were obtainable if the excited state to ionization continuum energy defect was small. In this study only visible wavelengths were used. The next publication68 provided a more detailed mechanistic approach and examined both resonant and non- resonant transitions under saturated and nonsaturated conditions. The number of elements detected by LEI was also expanded through the use of a frequency doubling ‘3 a sensiti- option in the dye laser. In the third study vity factor dependent on the atomization efficiency, laser Spectral irradiance, energy defect and the Boltzman popula- tion of the lower level of the laser excited transition was developed. Once again the number of elements studied was increased. One of the major advantages of the LEI technique is its freedom from such optical interferences as scattered light. It does, however, suffer from electrical inter- ferences caused by easily ionized matrix components. These interferences were characterized by Green and co-workers with respect to matrix composition, flame composition, probe voltage and electrode shape69. One of their sugges- tions was the use of plate rather than rod cathodes for increased collection efficiency. Plate electrodes were subsequently examined in more detail by Havrilla and Green70. A later study of matrix interference reduction 71 was performed by Turk who showed that a water cooled -12- cathode near the laser irradiated volume (lg. immersed in the flame) was far superior to external cathodes. In this same time period Trask and Green reported the effects of acid matrices72. More recently it has been shown that GM excitation is effective in decreasing matrix effects73. Pulsed signal collection has also been tried, but found ineffective in matrix interference reduction74. This same study suggested that time resolution could be used to discriminate against interferences in some cases. For this to be possible the analyte must reach its steady state concentration in the flame more quickly upon nebulization than does the interferent. To improve sensitivity and selectivity in LEI further, especially for high IP elements for which single laser excitation leaves a large energy defect, stepwise excita- tion schemes have been devised and tested. In 1979 both the N85 group and researchers in Russia published results of experiments involving stepwise excitation‘o’lz. The NBS group used two dye lasers simultaneously pumped by a single nitrogen laser to produce stepwise excitation of seven elements. The Russians used a Nd:YAG laser to pump two dye lasers in a study of sodium. Later, the Russian groups published results of studies similar to the one 75,76 77 as the described above, but with Li,Cs and In analytes. The NBS group switched to NszAG pumped dye lasers with frequency doubling so that high ionization -13- potential elements with strong resonant transitions in the ultraviolet could be studied for the first timeg. Processes related to ion collection in LEI have been studied exten- sively both theoretically and experimentally. Schenck gt_gl. used a CH dye laser with an opto-acoustic modu- lator to study spatial and temporal characteristics in a hydrogen/air flame78. Another group used two nitrogen- pumped dye lasers to study electron pulse shapes in experi- ments with sodium79. In similar studies, masses and sizes of small particles in flames have been estimated from 80, and the mobilities of atomic ions mobility measurements have been studiedal. The OGE has also been used to study molecular species in flames. In the earliest studies, transitions in LaO, SeO and Y0 were examined using a nitrogen pumped dye laseraz. These experiments involved single photon excita- tion followed by collisional ionization and proved that LEI is a viable method for studying molecular transitions in flames. Further work in molecular spectroscopy has in- volved photo- rather than collisional-ionization and is discussed in the next section. C-3. Dual Laser Ionization (DLllgand Resonance Ionization (RIS) Resonance ionization spectroscopy (R15) is the name given to resonance enhanced photoionization performed in rarified atmospheres (£3. proportional counters). This -14- has been comprehensively reviewed by Hurst et al.83. In this article schemes for the ionization of almost every atom in the Periodic Table were suggested. Experimental methods were also detailed for studying the mutual diffu- sion of gases, absolute measurements of atomic excited state densities, absolute photoionization cross-sections and rare events (such as fission fragment detection). Recently, RIS schemes have been used extensively in both atomic and molecular determinations, but very few of these experiments have involved opto-galvanic detection. In most cases mass spectroscopy has been coupled with RIS. For atomic analysis this has been termed resonance ioniza- tion mass spectroscopy (RIMS) and has been used by a number of authors (31. 84 and references therein). For molecular species the term resonance enhanced multiphoton ionization or REMPI has become common. This technique has been dis- cussed in detail in a recent overviewas. In fact, the only. recent coupling of RIS and opto-galvanic detection has been in the study of trace impurities in solids. In this tech- nique, called laser ablation resonance ionization spectro- scopy (LARIS), one laser is used to vaporize part of a target sample in a proportional counter. A second laser, tuned to a resonance of the impurity is used to give the RIS signa186. Dual laser ionization is basically the extension of the RIS technique to the flame cell. This was first -15- introduced by van Dijk, et al. in 198115.. It should be noted that the stepwise ionization LEI studies performed by the Russian groups (see preceding discussion) were initially intended to be DLI experiments. In each case an ionizing beam (either the Nd:YAG fundamental or one of its harmonics) was used. No DLI enhancement was observed be- cause the stepwise dye laser excitation left the atoms under study in highly excited states from which collisional ionization dominated photoionization (see Chapter III). Lin and co-workers have recently used the DLI techni- 19 que to determine flame temperatures , and the spatial resolution available with the technique has been found to be useful in circumventing electrical interferencesIG. Also recently, rate equations have been used by two dif- ferent authors to calculate ionizatidn yieids in DLI under common experimental conditions‘8’87. Atomic species have not been the only focus of the DLI technique. The multiphoton ionization of NO, through a two photon resonance, has been observed with opto-galvanic detection88. Comparison of experimentally obtained and computer generated data led to flame temperature estima- tions. In this same publication relative concentration measurements were used to deduce flame chemistry in different areas of the flame. Two photon photoionization, through a single photon resonance, has also been observed 89 90 and the PO molecules. for both the NO -15- 91 92 Finally, both atomic oxygen and hydrogen have been observed in flames using resonant two photon excita- tion followed by one photon photoionization. CHAPTER II EXPERIMENTAL A. Overview Figure 2-1 shows a simplified schematic diagram of the experimental system used to produce DLI and LEI signals. In a typical experiment a solution containing an analyte is introduced into the flame through a pneumatic nebulizer. A nitrogen-pumped, tunable dye laser is focused into the flame to provide resonant excitation. For DLI experiments a portion of the nitrogen (N2) laser beam is used to provide the energy necessary for ionization. This is directed to the flame so that it is temporally and spatially coincident with the dye laser. The ions, formed by the laser pulses,which occur on the nanosecond (ns) time scale, are collected by a set of biased probes immersed in the flame near the laser-irradiated region. A diagram of the signal collection circuit is shown in Figure 2-2. The electric field between the probes causes the migration of the ions; this migration is mobility- controlled and ion transit times from the ion formation region to the probes are on the order of microseconds (us). Once collected, the ion signal is amplified by a fast current amplifier. The low duty cycle of the signal -17- -13- .Eoumhm an\HmA mo caumamnom wofiufiamaam .HIN unamflm ,11 mum<4 “-2?- 2805.2 comm-S. mommzfi mqmidm mmm<4 m>o comm: -19- .uasouwo cowuooaaoo Hmcmam cowuanacOA mo Eoumowo xoon . flflm mum-H. w>o .~u~ beamem mug-«z 72358232 _ 20¢... litlj 538.... .2233. e .6960 2. -20- necessitates the use of gated integration to improve the signal-to-noise ratio. The output of the current amplifier is the input to a boxcar integrator. The output of the integrator is then either taken to a strip chart recorder for display or to a microprocessor for further processing. A typical time-resolved ionization signal is shown in Figure 2-3. 8. Laser System The NZ laser (model 0.5-150, NRG, Inc., Madison, WI), used both as a pump source for the tunable dye laser and as the ionizing beam in DLI experiments, has been described in detail elsewhere18 so only its general characteristics are listed here. The rated peak power output of the N2 laser at a repetition rate of 60 Hz is 500 kW. Repetition rates from single pulse to 60 Hz are available (typically a 20 Hz repetition rate was used). The portion of the beam split off and used for ionization had a radiant power between 100 and 150 kW. The major disadvantage of the'N2 laser as a pump source is the radio frequency interference (RFI) broad- cast when breakdown of the spark gap occurs. This introduces noise to all of the associated electronics. To minimize RFI at the source, the N2 laser was placed in a Faraday cage made of 0.5 mm c0pper plating. It was also found that keeping the Spark gap clean was effective in minimizing RFI. A strict cleaning schedule was adopted. -21- &O - $0;- 4.0 -' INTENSITY ( v ) 10‘- 20 t0 &0 &O 100 AF’EFTTUFRE lDELuAY ( ‘23 ) Figure 2-3. Typical time-resolved ionization signal (Na as analyte). -22- After each 50 hours of use the electrodes were removed and replaced. Replacement electrodes were copied from the originals by the departmental machine shop. Usually the laser tube pressure was maintained at 72 Torr while the spark gap pressure varied between 22 and 28 psi; this depended upon the length of time since the last spark gap cleaning. To avoid short term variations in the tube and spark gap pressures a dual stage pressure regulator (Model 580, Airco, Inc., Murray Hill, NJ) was used. The Hansch type laser system93 along with the alignment procedure developed in our laboratory is described in Appendix B. A list of the dyes used along with their wavelength ranges and typical peak output powers is given in the next chapter. The spectral band- widths of three of the dyes were checked. Rhodamine 66 was studied by scanning the dye to obtain the fluorescence excitation profile of the Na 0 line at 589.0 nm. The atomic line width in the flame is small compared to the dye laser bandwidth so the full width-at-half-maximum (FNHM) of the profile represents the bandwidth of the laser. For both Coumarin 460 and DCM a more direct approach was taken. The dye laser output was directed to the entrance slit of a monochromator (to be described in a following section) and observed with a photomultiplier (RCA,lP28A). The entrance slit of the monochromator was fixed at a value small enough to insure that the bandpass -23- of the monochromator was small compared to the laser band- width. The wavelength counter was set for the wavelength of the analyte resonant transition typically used with that dye in DLI experiments. The dye laser was then scanned around this wavelength. In all experiments the bandwidth was approximately 0.075 nm. The temporal overlap and pulse shapes of the N2 and dye lasers are shown in Figure 2-4. They were obtained with a PIN photodiode in a circuit to be described in a later section. The output of the photodiode circuit was displayed on an oscilloscope (Model 564, Tektronix, Inc., Beaverton, OR) equipped with a Polaroid camera (Model C-12, Tektronix, Inc., Beaverton, OR). Laser power attenuation was accomplished with a set of metal neutral density (ND) filters. These are described in Chapter IV. C. Flame Systems and Sample Introduction A laminar flow, premixed, HZ-OZ-Ar flame was used in most experiments. This flame is similar to one designed by Lijnse, et al.94 and has been described in detail by both van Dijk95 and Lin‘B. Lin's description includes a full discussion of the advantages of this type of flame and a flow diagram of our particular system. Thus, these are discussed only briefly here. Both central and mantle flames were burned from a Méker burner head. The gases were controlled by dual stage regulators at the tanks, Normalized Amplitude (an) -24- lOOr K)- 1 11 _14 1 10121416 182022242628 Timolns) L 00 2'4 A 8 1 6 Figure 2-4- Temporal overlap and pulse shapes of the N2( I ) and dye ( O ) lasers. -25- followed by precision flowmeters (Series FM-lOSO, Matheson Instruments, Horsham, PA). These flowmeters were roughly calibrated by connecting the burner head to a paddle-type flowmeter (Model 1755, Rockwell Manufacturing Co., Anaheim, CA) with a 3.2 cm id rubber hose. A stapwatch was used to determine the amount of time necessary to collect either 0.5 or 1.0 L of gas at a particular setting. Typical data are shown in Figure 2-5 for the mantle flame.” Itshould be noted thatany changes in the flow system, such as changes in nebulizer adjustment, will change the calibration. In general the central and mantle Ar settings used gave a flow of approximately 3.2 L/min. The mantle flame H2 and O2 flow settings gave a stoichiometric mix of 1 L/min and 0.5 L/min respectively. The central flame H2 and O2 settings were adjusted to give maximum DLI signals. These occurred under slightly fuel-rich conditions. The original nebulizer used was based on a design 96 and has been described elsewhere97. of Donohue and Carter The only major modification was in the use of stainless steel, rather than glass, capillaries. The glass needles were easily broken and difficult to reproduce and replace. To manufacture the stainless steel capillaries, a small bead of silver solder was placed on the tip of a 2 mm outer diameter stainless steel tube. A small hole was then drilled through the center of this bead and finally the bead was turned to a conical point. -25- “ + 'IHOMW“E(UERSFIRIMMHE) N llllllllLlllllllllllJlJLIJllll o UYWITIWIUVTUTYIUT[FjiilIIIIITIFTIUVTIIIIWIIIIIIIIUII O 10 20 30 4O 5O 60 7O 80 90 100 110 FLOWETER SETTING Figure 2-5. Mantle flame flowmeter calibration data (01M; (A182: (0102. -27- Satisfactory flow rates were obtained with a 0.025 inch hole for the solution uptake capillary and a 0.026 inch hole for the crossed-flow capillary. These capillaries were found to be virtually indestructible over hundreds of hours of use. Even so, this nebulizer design displayed instabilities as the capillary sleeves could slip in their fittings and change the capillary positioning. To alleviate this problem a new nebulizer was designed (with the help of C.A. van Dijk). In this design, shown in Figure 2-6, the nebulizer body was expanded. This allowed the capillaries to be set into sleeves which could be screwed into tapped holes in the nebulizer body. The cross-flow capillary sleeve was inset in the body and the Ar brought in through another fitting (also set in a tapped hole). Once set, the capillary could not be moved by jarring the nebulizer. The solution uptake capillary was strapped to the nebulizer body to prevent movement. This arrangement proved to be very reliable. For the analytical applications described in Chapter VI, an air-acetylene flame was used. This flame is described in Chapter VI. 0. Signal Detection and Processing D-l. Ionization Signals Several collection circuits were tried before a final design was chosen. This design is shown in Figure 2-7. The configuration of the amplifier is a -23- O' ring ”Mp Nebulizer design. Figure 2-6. >94 -29- GIN 3x 000 6‘00. cho Gxo. “.500 .huufisouao coauooaaoo Hocvfim cowumuficoH .hlm ousmwm .. .» axe. m N600 Lam” ii >2.— .. - £14.. -30- modification of a circuit described by Havrilla and Green98. The probes used to collect the ionization signal were either iridium or nichrome in composition. The former were found to be less susceptible to contamination, but the latter were much less expensive and experimentally acceptable as long as they were cleaned (sanded) or replaced frequently. The anode was held in place with a banana jack set into a translation stage attached to the burner head mount. The cathode was soldered into a quick-connect spade bit whose mate was also clamped to a translation stage set on the burner head mount. In this manner each probe could be easily replaced and could be moved in three dimensions. In almost all cases the probe diameter was 0.025" (0.64 mm). The probes were connected to the collection circuit with a subminiature BNC cable (RC-174, Amphenol North American, Oak Brook, IL) which was chosen both to provide shielding and to minimize stray capacitance. For the same reasons the leads were kept as short as possible. To reduce RFI pickup further, the cables were inserted into shielding braid which was connected by solder to the miniature BNC connector which, in turn, was grounded to the case. Also, the connectors at the probe ends were insulated with tissue paper and then wrapped by aluminum foil which was in turn strapped to the shielding braid. These measures -31- left only the actual probes unshielded and reduced RFI significantly. The operational amplifier used was an "Ultra Fast" JFET amplifier (LH0032CG, National Semiconductor, Santa Clara, CA). The slew rate for this amplifier is 500 V/us, and it has a 70 MHz bandwidth. The amplifier circuit is contained in a grounded aluminum case. The grounding was accomplished by attaching it to the burner head mount. This also helped to minimize the length of the probe cables. All connections, including amplifier supply voltage and high voltage, were made by BNC connectors as the use of banana cable was found to increase the RFI pickup dramatically. The supply voltage (:15 V) was obtained from a commercial power supply (A0902, Analog Devices, Norwood, MA) in a home- made circuit set in a shielded case. The probes were biased by a high voltage power supply (Model 3KlDB, Power Designs, Inc., Hestbury, NY). As shown in Figure 2-7, a single pole double throw (SPDT) switch on the input allows the ionization signal either to go directly to the amplifier or to pass through a high pass filter network. In the former configuration the circuit acts simply as a current-to- voltage converter. The gain was made switch selectable with nominal values of 104, 5a<104, 105, 5x105 and 2x106 V/A (with feedback resistors of 10, so, 100, 500 and 2000 k0, respectively). In the other configuration -32- the gain is frequency dependent as the input capacitor, the operational amplifier and the feedback resistor act together as an active high pass filter, and the 10 k0 resistor acts simply as a load. The frequency response curves for each of the configurations were obtained experimentally and are shown in Figure 2-8. To obtain these, sine waves of varying frequencies were fed to the input through a 10 k0 resistor. The amplitude was held constant at 0.01 V peak-to-peak so the input signal varia- tions were 1 uA (about the same as a typical ionization signal). The output amplitude was observed to measure the gain. In both the current amplifier (Figure 2-8a) and filter (Figure Z-Bb) configurations the response is seen to roll off at high frequencies. This is due to bandwidth limitations of the circuit. In the filter configuration low frequency signals(s lkHz) are strongly attenuated. The filter was found to have little effect in experiments involving the Hz-Dz-Ar flame, but was essential for the elimination of low frequency flame noise when the air-acetylene flame was used. In typical experiments either the 100 or 500 kD feedback resistor was used. The output of this circuit was processed by a boxcar integrator that has been described elsewhere18. In typical experiments the ionization signal was observed to be on the microsecond (us) time scale (see Section A). Because of this, a 5 us aperture (gate width) with a 50 us aperture delay range was normally employed. The .uoumamou xoonnoow ax A: ADV «you-mane.” xomnpomm ax om on «Houmflmou gonna-mom ex 2: A DV uMaugham-u xoonnoom ax com A 0v «Houmwmou xoonpoom a: N A 4v .cowuowsmamcoo “madam And can coaumuzmamcoo Hoamwamnfi ucounso A3 new uwsouwo acauooaaoo :oHuouwGOA on» no uncommon hocoavoum .mim ousmam Aer..- >ozu30umu 00.. Aer-C >02u20umL co.- n a A o T _ . _ _ ..o.. I 9 9 . V V M.- N ~ . \I \l A A o o .1 1 l L S S d d 3 3 a a v w w w d d x x I. I. o o . . 9 9 I\ I\ -34- percent delay was varied from experiment to experiment to achieve the best possible S/N ratios in each case, but values from 6.8 to 7.8% were common. The boxcar was triggered by a 20 volt, 20 ns pulse obtained from an output on the N2 laser. This pulse occurred when the laser fired, but a small modification in the circuit was necessary as the resistors used by the manufacturer were not rated for the voltages used. These were replaced. Best results were obtained when triggering on the negative edge of the pulse. In all experiments the exponential averaging mode was used and usually the input was AC coupled. Typical input and output time constants were 10 us and 0.1 sec, respectively. The output of the boxcar integrator was the input either to a strip chart recorder for data reduction by hand or to a microprocessor for further processing‘g. D-2. Optical Signals The optical system used for fluorescence detection has been described by Lin‘s. Briefly, a grating monochromator (either Model EU700 or Model EU700-56, GCA/McPherson, Acton, MA) was used for wavelength isolation. This was followed by a photomultiplier module (Model EU70l-30, GCA/McPherson, Acton, MA) which housed either a lP28 or a 1P28A photomultiplier tube (RCA, Inc., Beltsville, MA). The PMT was wired for fast pulse detection as recommended by the manufacturergg. The DC output of -35- this module was the input to one of two current amplifiers. For most studies a homebuilt current amplifier, based on the design of Christmann100 , was used. To improve the input circuit of this design, a precision, low power, FET electrometer amplifier (Model ADSlSJH, Analogue Devices, Norwood, MA) was used. The bandwidth of this system did not allow study of undistorted fast fluorescence signals and so was used when signal magnitude only was important. For undistorted signals a circuit based on a wide bandwidth linear amplifier (Model CA2812, TRN, Inc., Lawndale, CA) was used. The instantaneous bandwidth of this amplifier is l to 520 MHz, and the rated gain is 30 dB (this represents a linear gain of 25). The amplifier output was either integrated by the boxcar or displayed on an oscilloscope. When the boxcar was used a delay line was necessary as prompt fluorescence decays on the nano- second time scale and there is a minimum gate delay of 75 ns associated with the boxcar triggering electronics. For this, a 75' length of BNC cable was used to carry the signal. In all cases the monochromator was situated at a 90° angle to the dye laser beam; in most cases a lens was used to focus fluorescence onto the entrance slit. For laser wavelength calibrations and bandwidth studies the same optical system was used (procedure given in Section B). In these experiments the laser -35- beam was directed to the entrance slit with either a mirror or a quartz flat placed on the burner head. For thermal emission measurements the slower current amplifier was always used with display on the stripchart recorder. To observe the temporal characteristics of the laser pulses a fast PIN photodiode (Model 4220, Hewlett Packard, Palo Alto, CA) was used. The circuit for this is shown in Figure 2-9. In this circuit the load resistance was made switch-selectable so that a wide range of light intensities could be observed. In experimental situations the lower values were used if possible because these produced the least amount of signal distortion. The output of this circuit was displayed on a storage oscilloscope (described in Section B). -37- 50a 18V “M Aw, I @ - .. 10K“ ll II Figure 2-9. PIN photodiode circuitry. CHAPTER III ENERGETIC CONSIDERATIONS IN DUAL LASER IONIZATION* A. Introduction As noted in the first chapter, laser-enhanced ionization (LEI) has proven particularly useful in the analysis of trace metals in flames. Briefly, a laser is used to excite selectively a resonance of an analyte contained in a flame. This decreases the energy necessary for collisional ionization, and the resultant increase in ionization is detected opto-galvanically. DLI is a closely related technique that has also shown promise in 15,16 trace metal analysis The technique permits non- photometric flame temperature determinations‘7. In DLI a second laser is used to ionize the analyte from its dye laser excited level. Under the proper experimental conditions the DLI signal can be orders of magnitude greater than LEI signals obtained with dye laser excitation alone. To achieve such enhancements, it is necessary to choose an excitation scheme in which photoionization from the excited state dominates collisional ionization. In this chapter, a review of applicable theory is used to elucidate the efficacy-determining aspects of DLI -38- -39- ionization schemes exemplified in experiments with Li, Na, K, Ca, Sr, In and Cs. The widely varying enhancements in DLI signals over LEI signals are traced to their quantum mechanical source. 8. Experimental B-l. Apparatus The experimental system used in these studies has been completely described in Chapter II. The laser dyes used and typical laser powers are listed in Table 3-1. B-2. Reagents Most solutions were prepared from reagent grade chemicals dissolved in distilled, deionized water. The chloride salts of Na, Li, K and Ca were used as was the bromide salt of Cs and the nitrate salt of Sr. A commercial atomic absorption standard (Aldrich Chemical Co., Inc., Milwaukee, HI) was diluted for use in the experiments involving In. B-3. Charge Sheath Avoidance Charge sheath formation about the cathode has been 16 and LEI13,69-71 observed in both DLI experiments. These sheaths result in reduced ion collection efficiencies at high analyte concentrations. Care was taken in this study to use analyte concentrations which produced negligible charge sheathing effects as actual signal magnitudes were to be compared and thus collection circuit-limited signals were to be avoided. -40- TABLE 3-l. Laser-Characteristics A. Excitation beam-Hansch design, nitrogen-pumped dye laser Typical Peak Power at Analytical Line - 5 kw Approximate Wavelength Laser Dye - range (nm) PBBOa 395 -410 Stilbene 420a 410 -440 Coumarin 460a 450 -480 Rhodamine so” 565 -605 new3 630 -710 B. Ionizing beam - Nitrogen Laser (Model 0.5-150, NRG, Inc., Madison, WI) Typical Power in Ionizing Beam -100 kN Nave1ength.. 337.1 nm (29,665 cm") aDyes obtained from Exciton Chemical Co., Inc., Dayton, OH. bDye obtained from Eastman Kodak Co., Rochester, NY. -41- C. Ionization Mechanisms Ionization pathways common to laser atomic spectro- metry in flames are shown in Figure 3-1. These can be grouped into two general classes: laser-assisted collisional ionization and direct photoionization. Schemes A-D fall into the first class and have been explored in the laser-enhanced ionization experiments performed to date. Scheme E is of principal interest in DLI. A brief description of collisional processes in flames precedes consideration of the laser ionization schemes. The rate constant for collisional ionization from a given analyte energy level, 2, in a flame has been develOped 101 by Lawton and Neinburg and can be expressed as g . (Eo-E ) c _ 8kT c 1 A kit ' "XE—1111] th “DE—”IT J . (1) under the assumption that the reduced mass of the collision pair, u, can represent an average over all collision partners. In Equation III-l, n is the number density of x the collision partner and this is assumed to be constant here and in following chapters. This makes the collisional rate equations pseudo first-order. Also, T is the flame temperature, 0:2 is the partner-averaged cross-section for collisional ionization from level 1, the term (Ei-Eg) is the energy difference between the ionization continuum and the level 2 (hereafterreferred to as theenergy defect, AE), and other symbols take on their usual meanings. If charge -42- Figure 3-1. Common ionization pathways in laser atomic spectroscopy. -43- Aucocomeut 0v ease-e222: 5.05903... -\- 5:055. .20.»...00 522.90 523.5. 5:35. AUG-5:005. 5:85. 5.9:. .05.»...00 .05.»...00 005535 35.3.00 22% .5556 55:90 8:85.205. 85:90 003.84 5.0503... 56.x. 020 -02... 5.9.0103... 30:55. 3.52m 0.05m Jill ..I.. . w, L-F _-_ .- e-fl _-_ I a. .1 _ . . N . _ AXmQ \A a A _ as 1.1 1.1 + . . u u _ . . flux-qu- VVVVN VVVAVVV § \VwmeV m o u m. 4 mEmEocomS. :0_.0~_:0_ 00554158.. Au _ ASHBNB UH -44- recombination is negligible, the total rate of ionization of a specific species in the flame can be expressed as: dn. 1 _ c 7H? ' i kijnj ' (2) where kgj is the rate constant for the production of ions from level j and n. is the number density of level j. J In the absence of any optical input, if a Boltzmann distribution of level populations is assumed, it may be seen that there are two competing exponentials involved in the ionization process. Ionization from high-lying levels is exponentially favored as the energy defect is decreased. The number density of these states, however, decreases exponentially with the state energy. If a strong optical field, 3L3; that produced by an organic dye laser, is tuned to a resonant transition of the analyte which terminates in state A, such that for all . c c levels 3 ft, |k njl <mz scum wouaAsuAco mqu> :ouuquo :A moAm00:A .AwmaA. AAA .mAOuoam acosoucnzzo no :oAuaanaxo no» 0200 00m 6 .oma. ..u .: .eoochcmaz "00.0.0 ucflche- debaceo>ou .mmz-m=xnz .zheo. 0A500< 0:0 m500¢ no» moAuAAAasnoA; :oAuAmcnna 0:0 mzum:0Ao>03: .4 .u .cAuamz -.A .2 .omoAz ..: .u .mmAAnou ..1 .onaox Eon» cogs: 000:) n .Ao> can AmmaA. A. .Ao> .Aa-mA. A .Ao> ..u .0 .coomzAzaaz -ooAL00 me-ocAe: oeoeccosoc .m.: .mOw mdnapccum mo :aoASm AacoAusz 0:0 uo AmAauuAu .:on>oA chocu 0A500<: .u .U .0090: scum vocAmuno muoaszco>nz :. moAqucma A Nw.m wonnn A AA.N onAA A om.m manna 7P A mo.m oAqAn Aw A Am.n mOnOM A ow.n mnmaA 4.. mm.o momr MoA no.0 nmwm no. om.o MA:- 12A 70.: wmAm wcA «A.= ~noA 0>o A150 1000-a. .nom. uoosm 0coeoucczcu iuo>o Auuo:u oh.~ nammn co.m ”anon wA.m mm mm .Aq. uoomoa Auuocu aA.m omeA- mn.m mnem- AA.o momm- qA.m cmeAw an.m n2¢Am qm.q vomm mh.m chose mo.m Onamfi AA.o mama-w qA.m awn.- am.m nwwmA n>v A160 AaAu:0uoa :oAuauA:oA 11’} III iii nu.¢ mvmwm “.mnm 1.. iii mm.m wnnAm n.mwo Ii- -1- m0.m mwwmw o.wn- mmo.: mmo.3 AA.v Aawmm A.vo wnAo.o nAAo.o up.” neaAn m.mnw vnA:.o nioAxA.a oo.m cmnwn w.¢:v am.o . 9A.: No.m mum-u A.oAv Mm.A Ao.~ mo.m mmko “.00? nA.M n.A mm.w mmomn n.nnw mm:.: “90.3 0A.n mum A o.mwm unm.o Nom.o mn.A woo-A n.o~ 3.3.2- ...mmmm ea 7-... 53?.- usuAAAnanoea mo zumcouum souaun nomnA :0A0Am::uh noumAAAomo voquxo mo Auuocm 0A: .079mm.02 .0wfnn.AA Agnrvm.nu Ammfgw.mz Aznfno.nu Agmfm-.. Aweiam.:A Aamtmm.0m Azefmw_au Aamfmm.nz Amnfmn.AA :oAuAmcauu vouAuxo AA0:0:0moA w ucosoAu .0000 oAaoomoAuuocm Azwom: 0:0 muASmou AmoAAoezc mo :oAuooAAou .n-m 0An0b -48- continuum (i;g; the energy defect, AE), the energy in excess of this gap provided by the N2 laser (iLEL the energy of overshoot, E05) and the DLI enhancement factor at the laser powers stated in Table 3-1. The enhancement factor is the ratio of the ionization signal obtained with and without the N2 laser present. Before the results are described in detail, some background information about the nature of these enhancement factors must be provided as they are not meant to be absolute measurements. In the cases to be presented in which photoionization is the dominant mechanism of signal production, the resonant transition was found to be saturated by the dye laser system. Results of these experiments are presented elsewhere (see Chapter IV); it suffices here to note that at the focal point of the dye laser, the power density was at least one order of magnitude greater than the experimentally determined threshold for saturation. In cases where DLI enhancement was observed, the signal depended linearly on the ionizing (N2) beam power. In the experimental apparatus used, the N2 laser beam was brought to a tighter focus than was the dye laser beam. Hence the LEI signal emanated from a larger region of the flame than did the DLI signal in every individual experiment. Somewhat different flame volumes were irradiated by the 670.7 nm and 589.0 nm dye laser beams in the Li and Na experiments, whereas the N2 -49- laser irradiated the same volume. Thus the enhancement ratios are not perfectly comparable. In addition, no correction for possible variations in nitrogen laser power have been made. Thus the enhancement ratios are presented as simple orders of magnitude. At constant analyte concentration, the LEI signal can be increased only by increasing the saturated irradiated volume, but the DLI signal can be increased by increasing either the power density in the ionizing beam or the area of overlap at constant power density. Thus, the enhancement factors quoted refer to those generated by the available laser system, not to the maximum fundamental enhancement factors possible. As the data of Table 3-2 show, four experiments were performed in which a significant enhancement factor was obtained. In each of these experiments it can be seen that the match between the N2 laser photon energy (3.68 eV) and the energy defect is within 0.7 eV. For example, an enhancement factor of 100 was found with Na as the analyte and the dye laser tuned to the Na 0 line at 589.0 nm, Na(3$1&-+3P3b). Here the overshoot energy, E is moderate (0.64 eV) and the oscillator strength 05’ of the resonance transition is reasonably large (0.647). It should be noted that the Na(3$1&-+3P1b) transition produced a similar enhancement factor. In all cases where spectrally resolvable multiplets were available, the strongest line was used. It is also worth mentioning that -50- oscillator strengths, although noted herein, are essentially irrelevant to the analysis of the DLI vs. LEI enhancement; ionization from a truly saturated excited state should be independent of the ease or difficulty of the state's saturation. Similar DLI flame experiments involving Ca, Sr, and Li were performed with analogous results. The Ca(4s-r4p) and Sr(Ss-+5p) transitions, occurring at 422.7 and 460.7 nm respectively, possess large oscillator strengths and have energy defects almost equal in size to that found in Na. In both cases the enhancement factor is found to be of the order of 100. The enhancement factor found with the dye laser tuned to the Li(25-+2p) transition was the largest of any studied. The energy defect is larger in this case than in the others and this implies both a decrease in collisional ionization efficiency and a better energy match in the photoionization process (1:2; small E05), since the ionizing laser frequency was unchanged. These factors will be discussed in more detail below. As an example of the differences in signal obtained with and without the ionizing beam, the results of a typical study (with Li as the analyte) are presented in Figure 3-2. In this experiment, the sample was nebulized into the flame and the dye laser scanned across the Li(Zs-+2p) transition. In the figure, the ionization signal is plotted versus wavelength for the DLI (upper -51- hooded: o -2-4-e-e-Io Anamowsfnanherma' Figure 3-2. Ionization signal versus wavelength for DLI (upper trace) and LEI (lower trace). Dye laser scanned across the Li(Zs-*2p) transition at 670.7 nm. To obtain the LEI signal, the sensitivity (vertical scale) was increased by a factor of 5 and the analyte concentration by a factor of 100. -52- trace) and the LEI (lower trace) cases. To obtain the LEI signal, the measurement sensitivity was increased by a factor of 5 and the Li concentration was increased by a factor of 100, as indicated in the figure. The DLI experiment was attempted with In as the analyte, with the In(5p-*65) transition at 410.1 nm in resonance. Only a small enhancement factor was obtained. Non-resonant transitions were used in experiments with both Na and Ca. In both cases the ionization signal was derived strictly from the LEI process (Scheme C). With Na, the dye laser was tuned to excite the Na(3p-+5$) resonance at 616.1 nm; the 3p level was thermally populated. Here, the energy defect after the one-photon absorption step was quite small (1.03 eV) and consequently the EOS value was very large compared to those cases in which enhancements were observed. Two-photon excitation involving the Na(3s-+4d) transition at 578.7 nm and the Li(2$-r3d) transition at 639.2 nm were also examined (Scheme 0). As in the case of the non-resonant transitions described above for Na and Ca, these transitions populated electronic states with such small energy defects that no signal enhancement was observed upon addition of the ionizing laser beam. Finally, resonant transitions in both Cs and K were studied. In both the case of the Cs(6s-+7p) transition at 455.5 nm and the K(4s-+5p) transition at 404.4 nm, resonant LEI only was found (Scheme A). Once again, no -53- signal enhancement occurred when the N2 beam was added. Indeed, some direct two-photon ionization (Scheme F) was observed at high concentrations of Cs and K with the N2 beam alone. If the latter mechanism contributes at the lower concentrations used in the experiments reported in Table 3-2, decreased signal-to-noise ratios would result with the N2 beam present because this source of background ionization should fluctuate dramatically with fluctuations in the N2 laser power due to the squared power dependence. Intrepetation of the foregoing results requires consideration of two features of the ionization process which determine the efficacy of DLI in flames. As discussed below, photoionization cross-sections depend on E 68,102,103 In cases of "electrical" saturation the 05' near unit efficiency of collisional ionization from excited states precludes a significant DLI enhancement, moreover, collisional ionization may compete with photoionization for dominance even in the absence of such saturation. The general (and approximate) result that the ionization cross-section for atoms undergoing photo- absorption decreases with increasing energy overshoot, E05. has been recognized in various limited cases since at 1east 1930. As discussed by Heitler104 , this important qualitative feature can be predicted for the photoelectric effect in K-shell atoms, both with the Born approximation, -54- in which high energy electronic continuum states are represented simply as exp(ipr/h), and with accurate numerical representations of the continuum electronic 105. Several wavefunctions near the absorption edge model calculations have yielded cross-sections as functions of photoionizing frequencies for light alkali atoms in ground and excited state5106'116. A general decrease in photoionization cross-sections with increasing energies of the "final" electronic continuum states can be termed "hydrogenic" behavior; Bethe and Salpeter demonstrated monotonic cross-section 114. The physical principle under- decreases for hydrogen lying this behavior can be seen easily. Photoionization cross-sections depend upon the square moduli of radial matrix elements of the form 00 Rn£,£'(€) = L)rPn£(r)P€£.(r)dr , (6) where Pn£(r) and Pei.(r) are the radial wavefunctions of the valence electron initial and final states. The higher the electronic continuum energy, a, the more oscillatory the continuum electronic wave function P€£.(r) will be, and cancellation may be expected to reduce the radial matrix element and hence the cross-section. In actuality, when both the discrete and continuum states are oscillatory, it is possible that the oscillations may coincide for some a values, thus giving rise to structure in the- -55- photoionization cross-section. For example, the calculated photoionization cross-section for Cs(5d) shows two minimalls; more simply, the 4p and 5p cross-sections for K show maxima displaced from the ionization threshold113. Thus in practical analytical applications, the simplest approach to optimization of the DLI signal is an ionization scheme which minimizes E05. It should, however, be recognized that photoionization cross-sections may possess structure and thus permit additional optimization. The competition of collisional and photo-absorption routes to ionization enters the interpretation of the DLI experimental results reported above. The analysis is aided substantially by consideration of various theoreti- 107,108,110,113,114 117,118 cal and experimental cross- sections for these processes given in the literature. The discussion that follows employs comparisons of DLI enhancements ratios, and the caveats presented in the previous section are applicable. The largest enhancement ratio reported in Table 3-2 is that for Li (Zs-+2p-+IC). It is instructive to compare the three order of magnitude enhancement realized in the Li experiment with the factor of 100 achieved with Na(35-+3p +IC). (In both cases the excited states were observed 117 to be saturated). There exist both experimental and 113 theoretical results suggesting a Na(3p-+IC) photoioniza- tion cross-section of approximately 4 Mbarns at the N2 118 laser wavelength. Similar experimental and -55- ”3 results for Li(2p+IC) indicate an theoretical approximate cross-section of 17 Mbarns at 337.1 nm. The ratio of these cross-sections is consistent with the ratio of observed DLI enhancements; moreover, consideration of the relative collisional ionization cross-sections suggests a larger relative enhancement for Li than that predicted simply by the ratio of photoionization cross- 119 have calculated the average sections. Kelley and Padley collisional cross-sections for a thermal population (2100-2800 K) of Na and Li atoms in collision with Ar and H20, the principal collisional partners in the flame under study. The values om,H20 =11,ooo A2,oNa,Ar=27OO A2, °Li,H20 =3500 A2, and °Li,Ar = 1200 A2 may be considered approximations to the corresponding cross-sections specific to the Na 3p and Li 2p states. If collisional ionization is more probable in Na after dye laser excitation than in Li, then the DLI to LEI ratio for Na will be smaller than for Li. Inspection of the results obtained with the dye laser tuned to the Na(3p) and Na(Ss) transitions illustrates the potential value of Optimization of the ionizing beam energetics and also helps to explain the remainder of the results shown in Table 3-2. For the Na(3p) case the photoionization dominates collisional ionization by two orders of magnitude and the cross-section for absorption of the N2 beam, referred to previously, is 4 Mbarns. For Na(55) at the 337.1 nm excitation wavelength the cross- -57- 116 and hence the section is approximately 0.005 Mbarns photoionization contribution to the rate would be significantly smaller. Even if the energy of the ionizing beam were selected to Optimize photoionization (the ionization cross-section at threshold from Na(Ss) is 2.7 Mbarn5116) , collisional contributions would be expected to dominate ionization from Na(55). With the conservative assumption that the collisional cross-section is unaffected by excitation, the ratio of the collisional ionization rate constants can be approximated by the ratio of Arrhenius factors from Equation III-l, kc AE -AE kSS = exp[ 3pkT 55] , can P which in this case is approximately equal to 66,500. Thus the collisional ionization from the 55 state would be roughly two orders of magnitude greater than the rate of photoionization. Indeed it has been suggested that the efficiency of collisional ionization from high-lying excited states is near unity103. This would represent the "electrical" saturation discussed previously. The validity of the argument presented above is borne out in results of experiments in which Li and Na were used as analytes. Both Zorov, et al.76 75 and Matveev, et a1. attempted three-photon photoionization of Li in a flame. In these experiments two dye lasers, pumped by a single Nd:YAG laser, provided sequential (stepwise) excitation -58- of Li to the 3d state (AE =l.49 eV). No enhancement in ionization signal was observed when photoionization from this level was attempted with the second harmonic of the 10 Nd:YAG laser (E =0.84 eV). Gonchakov, et al. performed 05 a similar study on Na in which the two dye lasers were used to excite Na to its 4d state (AE =0.86 eV) and photoioniza- tion was attempted with the Nd:YAG fundamental at 1.06 pm. The ionizing laser beam was both well matched to the energy defect (EOS =0.3 eV) and powerful, but still no enhancement was observed. On the other hand, photoionization from the 4d level of Na was demonstrated in a study in which the Na was electrothermally atomized in a pure Ar 120. Under these conditions atmosphere at room temperature collisional ionization is less efficient than in a flame cell, and enhancement was observed upon the addition of the ionizing beam. The argument above explains why ionization from the K(5p) level, with a photoionization cross-section of about 3 Mbarns, but a AE of only 1.28 eV, is predominantly collisional. The same argument might be made in the cases of Ca(7p) and Cs(7p), were the applicable cross-sections available. On the other hand, In(6$) appears to be a good candidate for DLI study with appropriate optimization, although the energy defect is slightly smaller than those in the cases in which significant enhancement was achieved. -59- Also apparent is the fact that a more versatile DLI system could be easily designed. For example, a Nd:YAG- pumped dye laser system, equipped with a frequency doubling option on the dye laser and frequency doubling, tripling, and quadrupling options for the fundamental, would provide considerable flexibility, especially in the study of high ionization potential elements. Another natural extension of the DLI technique is its application to more common analytical flames. Preliminary experiments have been carried out with Ca as the analyte in an air-acetylene flame; enhancement factors for this element were similar to those obtained with the Hz-Oz-Ar flame. Studies on additional elements are in progress. There are no naturally occurring elements for which the lowest excited state energy defect is less than 1.5 eV, and hence with the exception of a few elements with very high ionization potentials, 343;_rare gases, the majority of the periodic table appears attractive for study. The incorporation of infrared lasers into the DLI technique would be necessary in instances of very low ionization potentials. For example, excitation of Cs to its first excited state requires a photon wavelength of 849 nm. In addition to resonant excitation considerations, apprOpriate attention to the effect of the ionizing laser wavelength will advance the DLI technique significantly. As new measurements and theoretical calculations of photoionization -50- cross-sections become available, highly optimized DLI schemes will become easier to design. In the interim, minimization of EOS appears the simplest and most profitable approach to DLI experimental design. *Obtained jointly with K.C. Lin, G.E. Leroi and P.M. Hunt. P.M. Hunt is particularly acknowledged for his input on the discussion of quantum mechanics. CHAPTER IV LASER PONER DEPENDENCE IN DLI AND LEI A. Introduction A clear understanding of the effect of laser power on signal generation is necessary for elucidating the fundamental processes involved in laser spectroscopic techniques and for systematically optimizing the methods. In this chapter the effects of laser power on signals obtained in DLI and LEI experiments are examined. The validity of a "rate equations" approach to systems involving resonant laser excitation has been 121’122. These studies discussed in the literature indicate that this approach, rather than the more complex density matrix formulation, can be used as long as no pulsations in the level populations occur at the charac- teristic Rabi frequency of the system. The experimental 18 and system used in this study has been examined by Lin found to be amenable to the "rate equations" treatment. Lin also used a three level system to describe DLI processes. Although complete, this system is more compli- cated than necessary if it can be assumed that ionization processes, both collisional and Optical, do not -51- -52- significantly affect the populations in the dye laser- coupled levels. This assumption allows the use of a simple two level atomic model similar to the one derived for laser excited atomic fluorescence (LEAF) by DeOliveras‘23. In the next section this type of model is developed. Following this, the validity of the abovementioned assumption is studied. Data from both DLI and LEI experiments are then explained in terms of the model. 8. Theory A simplified atomic level scheme is shown in Figure 4-1. The rate processes pertinent to DLI are included and labeled a-i. A description of each process is presented below. In this discussion, n1, 112 and "i are the population densities of the three levels (cm‘3), Ed is the 2 spectral irradiance of the dye laser (N cm' Hz’]) and E1 is the photon irradiance of the ionizing beam (photons cm"2 5"). a) Stimulated excitation rate per unit volume; n1b21Ed (B12 =Einstein coefficient for stimulated absorption (cm2 HzH-l 5-1)). b) Stimulated emission rate per unit volume; n2821Ed (82] =Einstein coefficient for stimulated emission 1 5-1)). c) Spontaneous emission rate per unit volume; "2A21 (cm2 Hz N- (A2] =Einstein coefficient for spontaneous emission in (sec'])). d) e) f) i) -53- Collisional excitation rate per unit volume; n1k12 (k12 =collisional rate constant for the process in which an atom in level 1 is collisionally excited to level 2 (sec'1)). Collisional deexcitation rate per unit volume; n2k2] (k2, =collisional rate constant for the process in which an atom in level 2 is collisionally deexcited to level 1 (sec'])). Collisional ionization rate per unit volume; "2k2i (k21 =collisional ionization rate constant (sec'])). Photoionization rate per unit volume; nzoE1 (o is the cross-section for absorption of the ionizing laser beam (cm2)). Recombination rate per unit volume to the excited state; niki2(kizis the rate constant.forthe process in which an ion and an electron combine to produce an atom in excited state 2 (sec'1)). Recombination rate per unit volume to the ground state; nikil (kil is the rate constant for the process in which an ion and an electron combine to produce an atom in the ground state (sec-1)). This description assumes that multiphoton ionization from the ground and excited levels is negligible as is collisional ionization from level 1. Also, the electron density in the flame is assumed to be constant making the recombination rate equations pseudo first-order. These are reasonable assumptions for most atomic systems in flames. -54- 1 1.111.... 11 III...I... 111111 4 _ i 2 Simplified atomic level scheme. Figure 4-1. -55- From the above the following rate equations are obtained: dn1 _—F'= "2(321Ed+A21+k21) -n](B]2Ed+k12) +"ikil (1) dn2 __f'= ”1(812Ed+k12) -n2(821Ed+A21+k2]+k21+oEi) + nik].2 (2) dni At this point some simplifying assumptions will be made. These are justified for the DLI case under study in the results and discussion section to follow. Assumptions: 1) with The photoionization rate in cases where significant DLI enhancement is observed is greater than the collisional ionization rate (i_._e__._oE1 >>k21)' Neither photoionization nor collsional ionization drain the excited state (level 2) to a significant extent (iLEL 821Ed+A21+k21 >>°Ei+k21)° Collisional excitation is negligible compared to absorption (i.e. 812Ed >>k12). Recombination is negligible (this follows from assumption 1) (j_._e_._ nikiZEO and n.k. 20). 1 11 these the rate equations (IV-l)-(IV-3) are reduced to: dn1 _t - "2(321Ed+a)-n]B]2Ed (4) dn2 dt ‘ nlBlZEd '"2(3215d+a) (5) dni —dt = "ZGEi a: 0 (6) where the excited state depopulation processes, A2] and k2], have been summed into the parameter, a. These assumptions and rate equations have two important consequences. First, levels 1 and 2 then comprise an isolated system as in the case of resonance fluorescence. Second, the instantaneous ionization signal is an indicator of the excited state population just as resonance fluorescence is. If levels 1 and 2 are isolated the total atomic population density, "T’ is given by The two level rate equations approach, taken by DeOliveras, is followed closely in the develOpment here. In the following the dye laser pulse is assumed to be a rectangular (boxcar) function so E O.St.$to 0 t‘>to where to is the duration of the laser pulse. If Equation (IV-7) is substituted into Equation (IV-5), the rate -57- equation becomes dn2 dt ’ "TBlZEd '"2(a+bEd) (8) in which b has been substituted for Blz'+321' Either one of two possible cases will be experimentally encountered. One case occurs if the population in level 2 attains a steady state value (iLQ; dnz/dt.=0) and the other if it does not. These cases will be discussed in detail after the nitrogen laser sampling function is examined. B-l. Ionizing_Laser Samplinngunction As a useful first approximation, the nitrogen laser pulse is assumed to be a rectangular function similar to that used to describe the dye laser pulse. Thus the instantaneous ionization signal should be directly proportional to the excited state population just as the instantaneous fluorescence is in laser-excited atomic fluorescence (LEAF). For signal-to-noise enhancement, gated integration is typically used in LEAF and thus the quantity measured is the average fluorescence signal during the gate duration. Similarly, the total ion current measured in DLI is proportional to the average excited state population during the nitrogen laser pulse. One important difference between DLI and LEAF should be noted here. In LEAF with short excitation pulses, very short integrator gate widths must be used because of the short excited state lifetimes of atoms. As discussed -68- previously (see Chapter II) the ion transit time from the formation region is on the order of microseconds. Thus the boxcar gatewidth can be approximately three orders of magnitude larger in DLI experiments than in LEAF and this simplifies the instrumental approach greatly. B-2. Steady State Case If the population in level 2 attains a steady state value, n35, Equation (IV-8) can be solved for 112 to give 8 E 55 12 d n - n (9) 2 T b [Ed+Es) where ES =a/b. From Equation (IV-9) it can be seen that, at low irradiances (Ed <a, Equation (IV-13) becomes (l-expE-bEdt ]) "T'B‘ 1' bE t o d o (16) At very high powers, the second term in parentheses goes to zero, and the average value of 112 reaches a limit (as in the steady state case) of F = n 92 . (17) 2 T g1*92 For the non-steady state case the saturation power density, where Ed =ES, is given by E E I S bto C. Experimental C-l. Apparatus All power dependence measurements described were obtained with the experimental arrangement described in Chapter II. The Hz-OZ-Ar flame was used in all cases. The laser dyes used have been described previously (Chapter III, Table 3-1). -71- C-2. Reagents Solutions were prepared as described in Chapter III. C-3. Laser Power Attenuation Laser power attenuation was accomplished with a set of metallic neutral density (ND) filters (Model 5023, Oriel Corp., Stamford, CT). Their nominal absorbances, as supplied by the manufacturer, are given in Table 4-1 along with the absorbances measured at the wavelengths used in the DLI studies. These were obtained through absorbance measurements performed with a commercial UV-VIS spectrophotometer (Model Cary 15, Varian Assoc., Inc., Palo Alto, CA). C-4. Experimental Procedure In actual experiments the lasers were first aligned so their focal points were coincident. The dye laser focusing lens was then moved to maximize the DLI signal. In the most recent studies (involving Li, Ca and Sr) it was found that defocusing the dye laser to a spot size of 0.5 cm diameter at the point of focus of the N2 laser was most effective. For both the Na and In studies, a tightly focused dye laser beam produced the best results. Laser power varied from day to day and alignment to alignment. In fact, small variations could be expected during the course of any experiment. Because of this, measurements were taken in bracketed sets to assure validity. That is, for each set of measurements the magnitude of the ionization signal under a standard set -72- cc.~ aA.N mm.~ mm.~ om.~ Am.~ Am.N .1 m om.A cm.. oo.~ co.~ oo.~ co.~ co.~ m..~ N Nom.o mma.o mma.o omo.o mma.o Aqm.o A-a.o Acm.o A NAm.o AAm.o cmm.o owm.o mAm.o AAm.o aAm.o mmm.o m.o mAm.o MAm.o .0m.o o0m.o emm.o Amm.o aem.o Ao¢.o ¢.o Aom.o Nom.o emm.o mm~.o .m~.o am~.o mm~.o Amm.o m.o AAA.o AAA.o om..o om_.o mm..o NmA.o NmA.o mc~.o ~.o .mo.o aAo.o «mo.o mmo.o Amo.o mmo.o mmo.o ch.o A.o Mdem mdwmw Mammw mammm. Mdmwm mimam «dflmw Hammm aoean-omae Ac:_20z 11--- AE:. :um:0Am>03 umwaum an 00:0ncomn< Ao=u0< -1--1- .mo.bm.-a0oe-a;0 ca0._- Ab.m=ao .a-bzoz ._-a m.m4p; 404.4 nm), Cs(65-r7p; 455.5 nm) and In(5p-r6$; 410.1 nm)). The results obtained in the K experiment were typical and are shown in Figure 4-8. In each case the log-log plot of the data was linear with a slope of almost one. This indicates no optical or ionization saturation was occurring. The results of these studies might have been predicted from the data given in Table 3-2. Both the Einstein A and B coefficients are oscillator-strength dependent as A21: E52%;)" Inc).O 2 and B21 ‘ 8__11h A21 In the above, f represents the oscillator strength of the transition and the other terms take on their usual meanings. As noted previously, the saturation power -85- 2.5 q) A 3 a 11- V 2.0 —1 4 < z 4’ o m G I .J 151- o o .J to I I 1 0.5 1.0 1.5 2.0 LOG PERCENT MAXIMUM DYE LASER POWER Figure 4-7. Dependence of LEI on dye laser power (Na as analyte). -86- 23'- A 3 O V J - ( 1.5 2 o m I] J 1.0 - o o J 0’5 I I I 05 1.0 1.5 2.0 LOG PERCENT MAXIMUM DYE LASER POWER Figure 4-8. Dependence of LEI on dye laser power (K as analyte). -87- density in the case in which steady is reached is given by E = a/b = A21"‘21 S LT+T91/927812] This is more commonly expressed as A N d in which 9 has replaced (1+g]/g2) for simplicity and Y is the fluorescence efficiency factor given by A21 A 21+k21 Y = From the above it can be seen that the saturation power density becomes independent of oscillator strength, in the steady state case, only when the fluorescence efficiency factor approaches unity(i;g; when collisional deactivation becomes unimportant). This is not a common' occurrance. In the non-steady state case the saturation power density is approximated by l E a ——— s btO and so is always dependent on oscillator strength. For each of the abovementioned analytes the oscillator strength for the laser-excited transitions is considerably lower than those seen in the cases in which saturation was observed. -88- The non-resonant transition of Ca(3d-+7p; 424.0 nm) gave the smallest energy defect studied so ionization saturation would be most likely here. The log-log plot was found to be linear, however, ruling this out. The slope of the line was found to be higher than those observed previously (i;§;_~l.5) and this indicates that multiphoton processes may be occurring. Finally the two photon transition in Na at 578.7 nm (3s-+4d) was studied. Here, the log-log plot was found to be linear and had a slope of two as expected. The low cross-section expected for two photon absorption makes saturation improbable and none was in evidence in the study. 0-4. Off-Resonant Photoionization Background ionization signals caused solely by the nitrogen laser were first observed in studies of electrical interferences (see Chapter V). Both K and C5 are common matrix components in real samples and both were found to produce background signals when high concentrations were nebulized into the flame. Nitrogen laser power dependence studies were performed on both. The results of the Cs experiment are presented in Figure 4-9. Here the plot is seen to be linear and the slope is nearly two (~l.8). Similar results were obtained with K as the analyte. These results indicate that the signal arises from off-resonant two photon photoionization of the Species. No saturation -39- A 2.0 " 3 O V .J < Z O .- 15 A m 2 O 5. < N 2 0 L0- 0 p. O I o. 0 O .I 0.5 1 I 1.0 1.5 2.0 LOG PERCENT MAXIMUM IONIZING LASER POWER Figure 4-9. Off-resonant photoionization of Cs. -90- effects are expected (due to the low cross-sections two photon transitions normally have), and none were observed. CHAPTER V DUAL LASER IDNIZATION IN FLAMES: A SEARCH FOR IONIZATION INTERFERENCES A. Introduction Almost since its inception, LEI has been known to suffer from electrical interferences caused by easily 13’69’70. The nature of this interference has been explored in detail by Turk7]. ionized matrix components His findings indicate that thermal ionization of the matrix causes an increase in the positive space charge near the cathode. The applied potential falls rapidly 125 across this charge sheath , which decreases the collec- tion efficiency. Turk also tested a water cooled cathode 7]. He found that that could be immersed in the flame keeping the laser irradiated region near the cathode (isfis within the cathode fall) was effective in reducing matrix interferences as in this region the ions "see" the electric field. More recently, pulsed signal collec- tion has been tried and found ineffective in reducing matrix interferences74; it was suggested that time resolution may be effective,as in some cases analyte and matrix component concentration deveTopment rates differ. -91- -92- Also, LEI with cw laser excitation has been found to suffer much less from electrical interferences because of the radically different nature of the charge collection processy3. As DLI is closely related to LEI (with pulsed laser sources) it was of interest to study the effects of common matrix components on the DLI signal and to compare these results with those obtained from LEI experiments. This chapter presents the results of those studies. 8. Experimental B-l. Apparatus The experimental system used in these studies has been described completely in Chapter II. B-2. Reagents All solutions were prepared from reagent grade chemicals. When possible these were dissolved directly in distilled, deionized water. The chloride salts of Na, Li, and K were used as was the bromide salt of Cs and the nitrate salt of Sr. Solutions for studies involving Zn and Mg were prepared both with the hydrated sulfates of these metals and by dissolution of the pure metals in HCl followed by neutralization with NH4OH. For studies of anionic interferences, ammonium salts of various anions were used. All solutions, unless otherwise noted, contained an analyte concentration of 10 ppm. The following laser dyes were obtained commercially and used without further purification: Rhodamine 6G -93- 3 (Eastman Kodak Co.; 7.5 XlO' M in ethanol) for Na analysis; Coumarin 460 (Exciton Chemical Co., Inc.; 1 x10'2 M in ethanol) for Sr analysis; and DCM (Exciton Chemical Co., 3 Inc., 5 XlO' M in DMSO) for Li analysis. B-3. Data Treatment In most cases data were collected with the boxcar integrator in the exponential averaging mode and the dye laser tuned to the resonance line of the analyte of interest (iLEL, 589.0 nm for Na, 670.8 nm for Li and 460.7 nm for Sr). The interference studies are reported as percent signal recovery versus concentration of the potential interferent. Percent signal recovery is defined in Equation (V-l) as % Recovery = (SA/SA) X100 (1) where SA is the analytical signal in the presence of the potential interference, and SA is the analytical signal obtained from a solution containing no added interferent. Because of the presence of background ionization of some interferents, in most cases due to off-resonant Nz-laser- induced multiphoton ionization, background subtraction was necessary to obtain 5; and SA. Two methods were found to give similar results. In the first method, the dye laser was scanned across the resonance line 0f the analyte, and the analytical signal was obtained as the height of the resulting peak relative to the baseline away from resonance. When the matrix constituents are unknown, as in a practical -94- analysis, this method of background subtraction is necessary. In the second method the analytical signal SA in the presence of the interferent, was obtained as the total signal ST in the presence of the interferent minus the signal SB from the interferent alone T.58. (2) Likewise, the analytical signal in the absence of interferent, SA’ was obtained as the total signal ST minus that of a distilled, deionized water blank, SB 8 . (3) Beecause this latter method required no change in dye laser wavelength for a given analyte, it was considerably more rapid than the former method and was used in all the studies reported here. C. Results and Discussion Experiments were performed to demonstrate the effects of common matrix constituents on the DLI signals of various elements. The study concentrates on a search for electrical interferences. Also reported are two chemical interferences that appear to be different from those encountered in routine flame emission studies. Non- specific (physical) interferences are not discussed as they should be identical to those found in conventional spectroscopic techniques. -95- Electrical interferences in DLI experiments can be broken down into two categories. One type is those interferences caused by the ionization of matrix components via thermal processes in the flame. These have been 71 described for LEI by Turk and this description was based, in turn, on the extensive studies of electrical phenomena in combustion systems by Lawton and WeinburgIzs. Briefly, the extraction rate of electrons by the anode is much greater than that of cations due to the greater mobility of electrons in the flame. This leads to excess positive charge in the flame which builds up as a sheath around the cathode and essentially negates, or drops, the potential difference between the probes. Turk found that signal recovery could be expected only when the laser irradiated region (ionization region) was within this sheath. Results from this study indicate similar phenomena as signal recovery is highly dependent on cathode/laser beam positioning. The DLI experiment also suffers from laser-related (spectro-electrical) interference, due mainly to off-resonant multiphoton ionization of easily ionized matrix components. Oscilloscope tracings showed these signals to be combined with the signal from the analyte in a single ionization pulse. They can be compensated for, in part, by various methods (cf. section B of this chapter), but do add noise to the signal because the off-resonant multiphoton processes have extremely low cross sections (i.e. non-saturated -95- phenomenon) and consequently are strongly affected by pulse-to-pulse variations in laser power. The low cross sections, however, do make minimization of these effects possible, without significant signal loss, through slight defocusing of the laser beams, especially if there is saturation of the analyte transition. Cesium was chosen as the interferent in initial studies because of its low ionization potential (3.9 eV). Cesium has no atomic transitions in the region of the analytical lines for Na or Li, but the 460.7 nm transition of Sr is quite close to the 455.5 nm transition of Cs from which large LEI signals have been obtained (see Chapters III and IV). In Figure 5-1 the DLI and LEI percent recoveries are shown for solutions containing 10 ppm Na and varying amounts of C5. In the DLI plot, the decrease in signal recovery with increasing Cs concentration is very gradual and does not exceed 10% even at the maximum interferent to analyte ratio of 100:1. Higher Cs concentrations were not used because flame emission studies showed that they decreased atomization efficiency significantly indicating, in all likelihood, a physical type of interference. Although our system is by no means optimal for the performance of LEI experiments (as we use a nanosecond pulsed dye laser rather than the typical flashlamp pumped, microsecond pulsed tunable dye laser),by blocking the path of the N2 laser to the flame, LEI experiments can -97- 130- 120‘ )- m “110'- > ‘3 0100- N u 4 m - 90- i I T .J < z 80- 0 W 70‘— 1... Z 60-“ LIJ 0 0: 50- LL] 0. 40‘ 3° I I I I -l 0 l 2 3 LOG INTERFERENCE CONCENTRATION (PPM) Figure 5-1. Percent signal recovery for Na versus Cs concentration, DLI (A) and LEI (0) cases. [Na concentration: 10 ppm ( A ); 100 ppm (0 )l . -98- be performed. Because the LEI signal was approximately two orders of magnitude lower than the signal obtained with 0L1” , the Na concentration was increased to 100 ppm and the flame was made less reducing in order to obtain a workable signal level. The results of this study are also noted in Figure 5-1. Here signal reduction is much more pronounced and similar in appearance to those seen 13’69’70. It should also be in early LEI experiments understood that the maximum interferent to analyte ratio used in this study was only 10:1. In a complementary study a high-lying transition (3s-f4d) in Na was reached via two photon excitation (two photons of wavelength 678.7 nm). It has previously been observed that ionization from this level is almost completely collisional because the addition of the N2 laser beam to the excitation region results in no signifi- cant signal enhancementls. Signal recovery on the ninety percent level was observed from the solution with the highest interferent-to-analyte ratio used in the single photon LEI experiment. Two factors contribute to this result. First, the smaller energy defect in the two photon experiment (0.84 eV as opposed to 3.02 eV) leads to decreased collisional dependence. The higher collisional efficiency should lessen the effect of ionization interferences. Also, cross-sections for two photon transitions are very low compared to those for single photon resonances, and previous results have shown that the -99- LEI signal from the Na(3$-+3p) transition is not fully saturated with respect to dye laser power (see Chapter IV). Thus it is reasonable to assume that in the two photon case the majority of the signal arises from a small region about the focal point where the photon flux is maximized whereas in the single photon case the signal arises from a larger region. Thus in the two photon case the collection probe could be placed near the region from which most of the ionization signal arises where collection efficiency should not be much affected by charge sheathing effects. Figure 5-2 shows the results of DLI studies with Sr and Li as analytes and Cs as interferent. The Li data show little signal loss with increasing Cs concentration and, in fact, that there is some signal enhancement at the extremes.(It should be noted here that in all the studies performed, exact percent signal recovery varied from day to day with probe configuration, nebulizer characteristics, etc. The results reported here are typical). The Sr curve shows no significant interference out to an interferent-to-analyte ratio of 10:1. After this, however, significant signal decrease is seen. Closer examination of the background ionization signals found with the dye laser tuned to the Sr resonance at 460.7 nm gave a possible explanation for this behavior. Multiphoton ionization of Cs via two N2 laser photons, two dye laser photons and via a combination of the two lasers was found -100- 150 - 14-0 - 130 -' 120 ‘- N _l_ 100 - (P ND 90" 80- 70- 60- 50- PERCENT SIGNAL RECOVERY 4o... 3° I 1 I T -1 o 1 2 3 LOG INTERFERENT CONCENTRATION (PPM) Figure 5-2. Percent signal recovery for Li (A ) and Sr (0) versus Cs concentration, DLI case. -101- to be occurring to a significant extent. At all other dye laser wavelengths studied, dye laser related Cs ionization was negligible compared to the photoionization produced by the N2 laser. In this instance Cs can be looked upon as truly a "worst case" interferent. To study the analytical utility of the DLI technique further, an analytical curve was constructed for Na in the presence of 1000 ppm Cs. This curve is shown in Figure 5-3 and is quite similar to the one obtained previously in the absence of an interferentls. As before the lowest Na concentration detected was 100 ppb (note here the interferent-to-analyte ratio of 10,000:l). The curve is linear for more than two and one-half orders of magnitude; a slight negative deviation is apparent at high concentra- tions. This deviation, however, is not as drastic and appears at higher Na concentrations than noted previouslyls. The data shown here seem to rule out recombination as an explanation for the deviation from linearity; the increased free electron concentration produced by C5 would be expected to enhance the effects of recombination and make the negative deviation more pronounced than that observed earlierls. The data are, however, consistent with charge sheathing. At high Cs concentrations, Cs should be the major contributor to the charge sheath until extremely high Na concentrations are reached. The presence of Cs should prevent Na from influencing the charge sheath (at -102- 4.0 - 313'- :LS - 145'! OAS- LOG NORMALIZED SIGNAL (A.U.) O b l I .0 In I I I I 1 -2 -1 o 1 2 3 LOG SODIUM CONCENTRATION (PPM) Figure 5-3. Analytical curve for Na in the presence of 1000 UQ/g Cs. I 4 -103- reasonable Na concentrations) and extend the range of linearity. LEI experiments have shown that electrical interfer- ences decrease with increasing ionization potential of 69. Thus, the effects of cations the interfering species with ionization potentials greater than that of Cs were studied. Figure 5-4 illustrates the effects of K and Li as interferents in Na analysis. Here it can be seen that K has no significant effect on the ionization signal. Lithium produces a slight signal enhancement and this varies quite gradually across the range of interference concentra- tions. A similar study performed with Zn and Mg is shown in Figure 5-5. The curves are nearly coincident and decline gradually at the high concentration end. Comple- mentary flame emission studies showed that the sodium emission signal also decreases in a similar manner; this small signal reduction is thus physical or chemical, not electrical, in nature. The data in Figure 5-5 were obtained from the metals dissolved in HCl as described in section B of this chapter. Results were also obtained with Mg and Zn solutions prepared from the hydrated sulfates, and percent signal recovery studies for these solutions are shown in Figure 5-6. The Na DLI signal from the Zn containing solutions falls off more rapidly than before, and that from Mg- containing solutions actually shows a trough with a minimum percent recovery at a Mg concentration of -104- 60- 1“)- 13)- mo— ‘H0— 100 - 9 90- 80— 70- PERCENT SIGNAL RECOVERY 60- 5° I I I I 1 -1 o l 2 3 4 LOG INTERFERENT CONCENTRATION (PPM) Figure 5-4. Percent signal recovery for Na versus K (O) and Li (A ) concentration, DLI case. -105- 150 -' 140-- 130 -* 120 - 110 '- 100 — 90— -* 80- 70- PERCENT SIGNAL RECOVERY 60- 50 I I I I 1 -l O l 2 3 4 LOG INTERFERENCE CONCENTRATION (PPM) Figure 5-5. Percent signal recovery for Na versus Mg (A ) and Zn (0 ) concentration, DLI case, metals dissolved in HCl followed by neutralization. -106- 130'- 120‘- 110'- 100‘- 90‘- BO‘fi 70‘- 60...- 50‘- PERCENT SIGNAL RECOVERY 4o- 30 I I l I n -1 o 1 2 3 4 LOG INTERFERENT CONCENTRATION (PPM) Figure 5-6. Percent signal recovery for Na versus Mg (D) and Zn (0) concentration, DLI case, metals from hydrated sulfate salts. -107- approximately 50 ppm. Studies of Na flame emission in the presence of Mg showed no similar effect. However, flame emission is not directly comparable to DLI because a much larger area of the flame is viewed in the emission experiment. In order to test if the results shown in Figure 5-6 could be explained by a sulfate ion effect, the Na DLI was measured in the presence of varying concentrations of (NH4)2504, and the results are shown in Figure 5-7. The gradual decrease in the Na DLI signal and a similar decrease in the Na flame emission in the presence of ZnSO4 lead us to conclude that the interference of ZnSO4 is due to a small sulfate chemical interference, a slight decrease in nebulizer efficiency or a combination of the two. The interference of M9504 is, however. intriguing, because conventional flame spectroscopic interferences usually display either an enhancing or a depressing effect, but not both. Exceptions, however, can be found in the literature. For example, Firman, in 1965, found that Fe interfered in the determination of Mg in a similar manner‘ZG. It was suggested that this behavior was due to the formation of a series of Fe-Mg compounds of varying volatilities. In many conventional flame spectroscopic experiments, interferences of this type have been eliminated or reduced through the use of releasing agents that function by replacing the analyte atom in the non-volatile compound. For example, excess Sr, La, Nd, Sm and Yb remove the depressive effects of sulfate, -108- 130- 120 - 100 J 90‘- 80 -' 70'- PERCENT SIGNAL RECOVERY 60‘- 50 I I f I o 1 2 3 4 LOG INTERFERENT CONCENTRATION (PPM) Figure 5-7. Percent signal recovery for Na versus sulfate ion concentration (from ammonium salt), DLI salt. -109- 127. We tested phosphate and aluminate ions in Ca emission this hypothesis by obtaining Na DLI signals from solutions of Na and Mg concentrations corresponding to the minimum in Figure 5-6, but varying in K concentration. The percent signal recovery improved with increasing K concentration from less than 40% to a maximum of approximately 70%. We thus conclude that the interference of M9504 is due to the formation of non-volatile species containing Na and not to any ionization interference, per se. 0. Conclusions In the present study the DLI technique has been extended to analytes in samples containing easily ionized matrix components. Reasonable signal recovery can be expected under normal conditions as long as the point of ionization is near the cathode. High concentrations of easily ionizable matrix components appear to extend the linearity of analytical curves, but can lead to higher noise levels because of photoionization of the matrix components. CHAPTER VI ANALYTICAL APPLICATIONS A. Introduction The ultimate goal of anyone developing a new analyti- cal technique is to demonstrate a successful application by analysis of real samples. In early DLI experiments an analytical curve was obtained for Na in distilled, deionized water using the HZ-Oz-Ar fIame‘S. This showed that the DLI . signal was linear with concentration over several orders of magnitude. The HZ-Oz-Ar flame did not permit detection of signal from solutions with Na concentrations much below 100 ug/L. Further studies with Li and Ca as analytes in the same flame produced similar results. In addition, a linear calibration curve was obtained for Na in a matrix that also contained a large concentration of an easily ionized background interferentls. Attempts to determine Ca in both an unsoftened well water sample and in a blood serum sample, however, were unsuccessful when the HZ-Oz-Ar flame was used. Because of this, an air-acetylene flame was incorporated into the DLI instrumental arrangement and the experiments were repeated. The air-acetylene flame -110- -111- (2400-2650 K) is much hotter than the Hz-Oz-Ar flame (1800-2100 K) and so provided better analyte atomization. Results were obtained and are presented in this chapter along with the results of complementary atomic absorption studies performed as an independent means of verification. 8. Experimental B-l. DLI Apparatus The instrumental arrangement used in the DLI studies was essentially the same as that described in Chapter II except for the flame. For this the burner mount was modi- fied to accommodate a slot burner taken from the atomic absorption unit used in the verification experiments. (Model AAS, Varian Techtron Pty. Ltd., Springvale, Australia). This burned a premixed air-acetylene flame. Sample introduction was accomplished with a pneumatic nebulizer driven by the air stream. Actual flame condi- tions are discussed in the forthcoming results section. B-2. Atomic Absorption Apparatus A commerical atomic absorption (AA) unit (Model AA5, Varian Techtron Pty. Ltd., Springvale, Australia) was used in all experiments performed to verify the results of the DLI studies of the well water and blood serum samples. B-3. Reagents The concentrated Ca stock solution used to make standard additions was made from reagent grade Ca(NO3)2- 4H20 (J.T. Baker Chemical Co., Phillipsburg, NJ). This -112- solution had a nominal concentration of 10,000 ug/mL Ca. Standardization with EDTA revealed an actual concentration of 9370 ug/mL. All dilutions were made with distilled, deionized water. 8-4. Samples The unsoftened well water sample was obtained from P. Wiegand. The blood serum sample was obtained from a commerical source (Moni-Trol.ES Level II Chemistry Control and Carbonate Diluent II, Lot no. PTS-103, Diluent X502- 132, American Dode, Miami, FL). This was reconstituted as suggested by the manufacturer. B-5. Standard Additions Procedure The standard additions method was chosen for the analyses to minimize the effects of physical and chemical interferences. The well water sample was known to be approximately 100 ug/mL Ca. Preliminary studies showed that both the AA and DLI instruments produced the best results when Ca concentrations in the range of 0.5 to 10 pg/mL were used. Because of this, each of the solutions in the series made for the standard additions method was prepared by adding a 1 mL aliquot of the well water sample and a 0L quantity (between 0 and 100 pL) of the stock solution to a 100 mL volumetric flask and diluting to the mark with distilled, deionized water. A description of this series of solutions -113- is given in Table 6-1. In this, X stands for the amount of Ca (in pg/mL) in the original well water sample divided by 100. As with the well water samples, the blood serum samples were known to contain slightly more than 100 ug/mL Ca. To make each of the solutions for the series in this case a 500 DL aliquot of the serum sample and a 0L quantity of the stock solutions (between 0 and 50 pL) were added to a 50 mL volumetric flask and diluted to the mark with distilled, deionized water. A description of this series is given in Table 6-2. As before, X stands for the amount of Ca (in pg/mL) in the blood serum sample divided by 100. Class A glassware was used for all solution prepara~ tion. The pL additions noted above were all made with automatic pipettes (Eppendorf, Brinkmann Instruments, Inc., Westburg, NY). With the standard additions method the amount of analyte originally present in solution can be obtained by plotting the analytical signal versus the volume of stock solution added. The X-intercept of the plot then gives the negative of the number of pL that would have to be added to a "clean" sample to get the response given by the solution to which no addition was made. Concentration is then found with a simple calculation which also takes dilution factors into account. -114- TABLE 6-1. Solutions for Well Water Sample Experiments. Sample Standard Ca Concentrations Number Addition (0L) (pg/mL) l 0.0 X 2 10.0 0.937 + X 3 20.0 1.87 + X 4 30.0 2.81 + X 5 50.0 4.68 + X 6 100.0 9.37 + X TABLE 6-2. Solutions for Blood Serum Sample Experiments. Sample Standard Ca Concentrations Number Addition (0L) (ug/mL) l 0 X 2 5 0.937 + X" 3 15 2.81 + X 4 25 4.68 + X 5 50 9.37 + X -115- C. Results C-l. Well Water Sample Analysis In the DLI and AA experiments five replicate determina- tions were performed on the series of solutions described in the last section. The averaged DLI results are shown graphically in Figure 6-1. The plot is linear for solutions up to and including the 50 DL addition. Because of this, a linear regression analysis was performed on the first four points in each data set. The x-intercept for each set was calculated in order to determine the Ca concentration of the original sample. The averaged results of the AA experiment are shown graphically in Figure 6-2. The same linear range is observed as in the DLI experiments. As before, a linear regression analysis was performed on each data set and the x-intercept in each case was calculated. A comparison of the results obtained with both methods is given in Table 6-3. The student "t" test shows that the difference between these means is not significant at the 95% confidence level. C-2. Blood Serum Samples The complex nature of the blood serum sample presented a difficult test for the DLI technique. This matrix con- tains high concentrations of easily thermally ionized components (Egg; the Na concentration alone is estimated to be approximately ten times greater than the Ca -116- 200. l 80. 160. 140. 120. 100. 80.0 DLISIGNAL(mrn) 60.0 40.0 20.0 |llIlLll'lllllllllfllllllllllLLL'lellll] 0.000 W 0123456789101112 VOLUME Co STOCK ADDED ( pL) I N I d Figure 6-1. Standard additions analysis of Ca in well water sample (DLI). -117- 0.600 q - i 0.500 - 1 .1 0.400 - U -I Q a i g 0.300 -- 8 A 9 - 0.200 - 0.100 - 1 -2-10123456789101112 VOLUME Co STOCK ADDED ( m. ) Figure 6-2. Standard additions analysis of Ca in well water sample (AA). -118- concentration). Preliminary results showed that the DLI signal was highly dependent on flame conditions. Optimal results were obtained when the flame was adjusted to burn in a slightly fuel-lean condition with a high air flow rate. This is probably caused by a number of factors. The air drives the nebulizer and so the higher rate means better sample nebulization. Also, excess 02 in the flame leads to more complete combustion and a lessening of electrode fouling. TABLE 6-3 Comparison of Results of DLI and AA Analysis of Ca in Well Water Sample Average Value of Ca Method Average X-intercept Concentrat1on 1" Original Sample (HL) (Hg/mL) DLI 12.3 1 0.23 115 i 2.2 AA 11.9 + 0.21 112 + 2.0 Five replicate DLI determinations were performed on the series of solutions. For graphical clarity, the results of only the first three trials are shown in Figure 6-3, the other two trials gave similar results and are discussed in the text. The figure shows that the signal dropped steadily as the experiment progressed. The signals ob- tained in the fourth and fifth trials, however, were slightly larger than those found in the third. This -119- 160. 140. l 20. D l 00. 20.0 DLISIGNAl-(mm) . . .8 O [LJJleLlllllllllllllellJlllllll 00000 . IIIIIIIIUTIIIIIIIIIFIIIIIITIII[IUIUIIFITITIIIIIUIII"IIITIIIIIIUII -3-2-10123456789101112 VOLUME Ca STOCK ADDED ( uL ) Figure 6-3. Standard additions analysis of Ca in blood serum sample (DLI) . (O ) trial 1; (A ) trial 2; ([3) trial 3. -120- behavior is probably caused by electrode fouling. In earlier experiments blood serum samples were found to de- posit carbon on the electrodes. These deposits can in- fluence the electric field between the probes and decrease the collection efficiency. Before beginning the experi- ments described herein the electrodes were sanded clean. Carbon deposits were seen to form during the experiments. Also particulates were observed to dislodge from the probes the course of the experiments. This probably explains the slight increase in signal described above. In any event, the signal was seen to deviate from linearity after the third point so only the first three points in each data set were used in the regression analysis to find the x-inter- cept and thus the Ca concentration in the original sample. The results are reported in Table 6-4. fi *— w v_._ w—— '— TABLE 6-4. Comparison of Results of DLI and AA Analysis of Ca in Blood Serum Sample Average Value of Ca Method Average X-intercept Concentration in (pL) Original Sample (uglmL) DLI 14.5 i 0.75 135 _+_ 7.0 AA1 14.9 1 0.40 140 _+_ 3.8 AA2 14.4 1 0.71 135 i 5.7 1 air-acetylene flame 2 nitrous oxide-acetylene flame -121- Complementary AA experiments were performed with both a nitrous oxide-acetylene flame and the air-acetylene flame. The experiment was performed nine times with the air-acetylene flame, and the averaged data are shown in Figure 6-4 along with the averaged results of four replicates taken using the nitrous oxide—acetylene flame. If any serious chemical interferences were biasing the results obtained with the air-acetylene flame different results would be expected with the nitrous oxide-acetylene flame. The figure shows that this is not the case. While the absorbance values obtained using the nitrous oxide- acetylene flame were higher than those obtained using the air-acetylene flame, due to the higher atomization efficiency in the nitrous oxide-acetylene flame, the x- intercepts (which translate to the sample concentrations) are nearly identical. The results of both of the AA analyses are given along with the DLI results in Table 6-4. The student "t" test shows that the differences between the means given in the table are not statistically significant at the 95% confidence level. Thus the two methods produce essentially equivalent results. The value of the Ca concentration in the reconstituted blood serum sample reported by the supplier is 127 i 2 09/ mL Ca. This value was obtained by a well-established method (in this case AA). Other methods yielded results from 113 to 138 pg/mL. Thus the values obtained herein lie 1 .00 0.800 0.200 0.000 _ . . . _ . IN I .. . . . . . . . Co . / -2-10 Figure 6-4. -122- To 123456789101112 VOLUME Co STOCK ADDED ( 11L ) Standard additions analysis of Ca in blood serum sample (AA). ( A ) nitrous oxide- acetylene flame; (O ) air-acetylene flame. -123- on the high side of the expected range. The consistency between the AA and the DLI results suggests that this discrepancy is due to some systematic error, such as a slight miscalibration in the automatic pipette or the standard solution, rather than any fundamental limitation of the DLI technique. CHAPTER VII SUMMARY AND FUTURE DIRECTIONS A. Summary Four aspects of DLI have been examined in the preceding chapters in order to characterize the basic processes involved in the technique and to illustrate its potential as a useful analytical method. First, the energetics of the various processes which affect signal generation in the DLI experiment were studied (Chapter III). Basic ionization pathways in laser atomic flame Spectroscopy were examined both from a theoretical and from an experimental standpoint. The photoionization pathway was found to dominate under the proper conditions, the most important being a close match between the energy defect and the photon energy of the ionizing laser beam. Various optimization strategies were then presented based upon this result. The next chapter treated the dependence of ionization signals upon laser power in both LEI and DLI. Dye laser power dependencies in single photon resonant and non- resonant LEI and in two photon LEI were examined. Similar studies were described for resonant DLI and the signal -124- -125- dependence on ionizing beam power was studied and discussed. The most important results indicated that the DLI signal was not saturated with respect to ionizing beam power and that the actual area from which a majority of the ionization signal arises is quite small. A search for ionization interferences in the DLI technique was detailed in Chapter V. The major contributer to signal degradation was found to be the formation of a charge sheath around the ion-collecting cathode. This sheath essentially negates the potential difference between the probes and was found to be most severe when easily ionized components were contained in the sample matrix. Minimization of the interference was shown to be possible by keeping the ionization region close to the cathode (igg; within the boundary of the sheath). Off- resonant photoionization of matrix components as well as some chemical interferences were also discussed. Finally, analytical applications of the DLI technique were examined in Chapter VI. Here an air-acetylene flame was used for the first time in the DLI experiment. Several experiments were described in which various sample matrices were tested for calcium content. The results were consis- tent with those obtained by other means. The DLI technique has a promising future as a sensitive and selective technique for the analysis of trace metals in flames and also as a tool in combustion diagnostics. There are many areas in which further study should -126- improve both the fundamental understanding and the general utility of the technique. Some Of these are discussed in the next section. 8. Future Directions AlthOugh the nitrogen pumped dye laser system was a good starting point it is clear that a different laser system could improve the DLI technique substantially. The present system suffers from three major shortcomings. First, there is not enough power in the dye laser beam to allow frequency doubling. This means that the ultraviolet wavelengths necessary for the study of high ionization potential elements are inaccessible. Although the power output of the present dye laser could probably be significantly improved by replacing the grating now in use with one having a more optimal blaze angle, it is doubtful that a reliable frequency doubled beam of reasonable power could be obtained in a convenient and inexpensive fashion. The single fixed wavelength at limited power in the ionizing laser beam is also a draw- back. Finally, the radio frequency noise generated by the breakdown of the spark gap in the nitrogen laser is a major limitation as it prevents the examination of very rapid signals. The Nd:YAG pumped system available in the departmental laser facility would be an excellent choice as a replace- ment. It has a number of advantages over the present -127- system. First, the dye laser can be frequency doubled to provide an ultraviolet excitation beam. For preliminary studies a moderate amount of selectibility in the choice of ionizing wavelength could be derived by simply doubling, tripling or quadrupling the infrared fundamental (10641A). The non-linear crystals necessary for this are already available so the experimenter could choose between 532 nm, 355 nm and 266 nm (2.33 eV, 3.49 eV and 4.66 eV, respective- ly). As an example, Ni could be used as the analyte in initial studies. It has a strong resonant transition at 232 nm (oscillator strength, f =0.68) which could be reached by frequency doubling the output of the strong, blue Coumarin 460 laser dye. The energy of overshoot (see Chapter III) with a 532 nm ionizing beam would be only 0.04 eV. As this is extremely close to threshold 8 reasonable DLI signal would be expected. If preliminary experiments such as the one described above are successful, the experiment could then be expanded by assembling a second dye laser to provide a truly tunable ionizing source. This would make possible a number of experiments. First, Optimization of analytical signals would be possible. In addition, the relative magnitudes of cross-sections for ionization with respect to the photon energy (wavelength) in the ionizing beam could be obtained and compared to values from theoretical calculations. Population redistribution upon laser excitation could be studied using electrical detection -128- once confidence in cross-section values was obtained. This would eliminate the experimental complexities of fluorescence detection which are presently involved in this type of experiment. The inherent Spatial resolution available through the use of crossed, rather than colinear, beams can be put to use in combustion diagnostics. With proper focusing and alignment the region from which the majority of the ionization signal arises could easily be made to be as small as one cubic millimeter. A stationary probe/laser configuration in conjunction with a movable flame could easily be used to map atom distribution or flame temperature in three dimensions. As there are now many motorized positioning mounts available commercially, this type of experiment could be placed under complete computer control. The analytical utility of the technique has barely been explored. In addition to enabling the study of high ionization potential elements, the move to the Nd:YAG laser system should greatly improve sensitivity. The reasons for this are two-fold. First, the power available for use in the ionizing beam will be higher than that available with the nitrogen laser system. As the ionization signal has been found to be linear with ionizing beam power, absolute signal magnitudes should increase. Secondly, the fact that the Nd:YAG rods are flashlamp -129- pumped will eliminate many of the noise problems related to the nitrogen laser spark gap. Finally, the use of off-resonant transitions in the DLI process could prove to be an important and interesting area of study. The large LEI signal obtained from an obscure, off-resonant Ca transition, as detailed in Chapter III, at least justifies further study as many elements have strong transitions originating from low lying excited states. The selectivity of the technique could be greatly enhanced if reasonable Signals could be obtained from these. A particularly interesting case for study is the Fe transition at 227.7 nm. The oscillator strength for this transition is about 2 and this is approximately four times larger than that of the major analytical line at 248.8 nm. A comparison of the DLI signals from these lines should prove quite informative. LIST OF REFERENCES 10. ll. 12. 13. LIST OF REFERENCES Foote, P.D.; Mohler, F.L. Phys. Rev. 1925, 26, 195. Penning, F.M. Physica 1928, 8, 137. Green, R.B.; Keller, R.A.; Schenck, P.K.; Travis, J.C.; Luther, 0.0. J. Am. Chem. Soc. 1976, 98, 8517- 8518. Green, R.B.; Keller, R.A.; Luther, G.C.; Schenck, P.K.; Travis, J.C. Appl. Phys. Lett. 1976, 29, 727- 729. Alkemade, C. Th. J. "Proc. 20th Coll. Spectrosc. Int. and 7th Int. Conf. on Atomic Spectrosc." Prague, Sbornik, VSCHT, 1977, 93-104. van Dijk,C.A.; Alkemade, C.Th.J. Combustion and Flame 1980, 37-49. . Travis, J.C. J. Chem. Ed. 1982, 59, 909-914. Travis, J.C.; Turk, G.C.; Green, R.B. Anal. Chem. 1982. 54, lOO6A-1018A. Turk, G.C.; DeVoe, J.R.; Travis, J.C. Anal. Chem. 1982, 54, 643-645. Gonchakov, A.S.; Zorov, N.B.; Kyzyakov, Y.Y.; Matveen, 0.1. Anal. Lett. 1979, 12, 1037-1048. Travis, J.C.; Turk, G.C.; Green, R.B. "New Applica- tions of Lasers to Chemistry", Hieftje, G.M., ed., ACS Symposium Series 85, American Chemical Society: Washington, D.C., 1978, Chapter 6. Turk, G.C.; Mallard, W.G.; Schenck, P.K.; Smyth, K.C. Anal. Chem. 1979, 51, 2408-2410. Turk, G.C.; Travis, J.C.; DeVoe, J.R.; O'Haver, T.C. Anal. Chem. 1979, 51, 1890-1896. -130- 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. -l31- Turk, G.C.; Travis, J.C.; DeVoe, J.R.; O'Haver, T.C. Anal. Chem. 1978, 50, 817-820. van Dijk,C.A.; Curran, F.M.; Lin, K.C.; Crouch, S.R. Anal. Chem. 1981, 53, 1275-1279. Curran, F.M.; van Dijk,C.A.; Crouch, S.R. Appl. Spec- trosc. 1983, 37, 385-389. Lin, K.C.; Hunt, P.M.; Crouch, S.R. Chem. Phys. Lett. 1982, 90, 111-116. Lin, K.C., Ph.D. Dissertation, Michigan State Univer- sity, 1982. Stanley, J.D. Masters Thesis, Michigan State Univer- sity, 1982. Goldsmith, J.E.M.; Lowler, J.E. Contemp. Phys. 1981, 22, 235-248. Keller, R.A.; Engleman, R., Jr.; Zalewski, E.F. J. Opt. Soc. Am. 1979, 69, 738-742. Rosenfield, A.; Mory, S.; Konig, R. Opt. Comm. 1979, 30, 394-396. Zalewski, E.F.; Keller, R.A.; Engleman, R., Jr. J. Chem. Phys. 1979, 70, 1015-1026. Feldman, 0. Opt. Comm. 1979, 29, 67-72. Rettner, C.T. ; Webster, C.R.; Zare, R.N. J. Phys. Chem. 1981. 85, 1105-1107. Suzuki, T. Opt. Comm. 1981, 38, 364-268. Wakata, S.; 38, 271-272. Shuker, R.; Ben-Amar, A.; Ercz, G. Opt. Comm. 1981, 39, 51-540 Saikan, S.; Kimura, M. Opt. Comm. 1981, Bjorkland, G.C.; Ausschnitt, C.P.; Freeman, R.R.; Storz, R.H. Appl. Phys. Lett. 1978, 33, 54-57. Ausschnitt, C.P.; Bjorkland, G.C.; Freeman, R.R. Appl. Phys. Lett. 1978, 33, 851-853. Green, R.B.; Keller, R.A.; Luther, G.G.; Schenck, P.K.; Travis, J.C. IEEE-J. Quant. Electron. 1977, OE-13, 63-64. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. -l32- Smith, A.L.S.; Moffatt, S. Opt. Comm. 1979, 30, 213-218. Keller, R.A.; Zalewski, E.F. Appl. Optics 1980, 19, 836-837. Pa1mer, B.A.; Keller, R.A.; Kowalski, F.V.; Hall, J.L. J. Opt. Soc. Am. 1981, 71, 948-952. Jirman, J.; Durr, N. Electron. Lett. 1982, 18, 69-71. Kavaya, M.J.; Menzies, R.T.; Oppenheimer, U.P. IEEE- J. Quant. Electron. 1982, QE-18, 19-21. Kavaya, M.J.; Menzies, R.T.; Oppenheimer, U.P. IEEE- J. Quant. E1ectron. 1982, 18, 807. Apostol, 0.; Blanaru, C.; Ionescu, A.; Popescu, G.; Popescu, II; Vasilui, V.; Rev. Roum. Phys. 1982, 27, 581-585. Yamaguchi, S.; Suzuki, M. Appl. Phys. Lett. 1982, 41, 51-98. Dovichi, N.J.; Moore, 0.5.; Keller, R.A. Appl. Optics 1982, 21, 1468-1473. Bachor, H.A.; Manson, P.J.; Sandeman, R.J. Opt. Comm. 1982, 43, 337-342. Beenan, G.J.; Piepmeier, E.H. Anal. Chem. 1981, 53, 239-242. Piepmeier, E.H.; Beenan, G.J. Appl. Spectrosc. 1982, 36, 235-240. Begemann, M.H.; Saykally, R.J. Opt. Comm.1982, 40, 277-282. Labastie, P. J. Phys. B. 1982, 15, 2595-2603. Roesch, L.P. Opt. Comm. 1983, 44, 259-261. Alchalaki, S.; Stewart, R.S.; I11ingwo, R.; Ruddock, I.S. J. Phys. B. 1983, 16, 115-124. Webster, C.R.; Menzies, R.T. J. Chem. Phys. 1983, 78, 2121-2128. Keller, R.A.; Zalewski, E.F. Appl. Optics 1980, 19, 3301-3304. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. -133- Dreze, C.; Demers, Y.; Gagne, J.M. J. Opt. Soc. Am. 1982, 72, 912-917. Kushner, M.J. Appl. Optics 1983, 22, 1970-1975. Tam, A.C.; IEEE Plas. S. 1983, 10, 252-256._ Uchitomi, N.; Nakajima, T.; Mauda, S.; Hirose, C.; Opt. Comm. 1983. 44, 154-158. Vasudev, R.; Zare, R.N. J. Chem. Phys. 1982, 76, 5267-5270. Webster, C.R.; McDermid, I.S. and Rettner, C.T. J. Chem. Phys. 1983, 78, 646-651. Ebata, T.; Mikami, M.; Ito, M. J. Chem. Phys. 1983, 78, 1132-1139. Lyons, O.R.; Schawlow, A.L.; Yan, G.-Y. Opt. Comm. 1981, 38, 35-38. Goldsmith, J.E.M.; Ferguson, A.I.; Lawler, J.E.; Schawlow, A.L. Opt. Lett. 1979, 4, 230-232. Suzuki, T.; Kakimoto, M. J. Molec. Spect. 1982, 93, 423-432. Miyazaki, K.; Scheiner, H.; Vidal, C.R. Phys. Rev. Lett. 1983, 50, 1046-1049. Beverini, N.; Galli, M.; Inguscio, M.; Strumia, F., Bionducci, G. Opt. Comm., 1982, 43, 261-264. Hannaford, P.; Series, G.W. Phys. Rev. Lett. 1982, 48, 1326-1329. Schenck, P.K.; King, 0.5.; Smyth, K.C.; Travis, J.C.° Turk, G.C. "Lasers in Chemistry", West, M.A., ed., Elsevier: New York, 1977, 431-435. ‘ King, 0.5.; Schenck, P.K. Laser Focus 1978, 14, 50’630 . Schenck, P.K.; Hastie, J.W. Opt. Eng. 1981, 20, 522-528. Travis, J.C.; DeVoe, J.R. "Lasers in Chemical Analysis", Hieftje, G.M., Travis, J.C. and Lytle, F.E., eds., Humana Press: Clifton, New Jersey, 1981, Chapter 5. 9 67. 68. 69. 70. 71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. -134- Zorov, N.B.; Kuzyakov, Y.Y.; Matveen, O.I. J. Anal. Chem. USSR 1982, 37, 400-411. Travis, J.C.; Schenck, P.K.; Turk, Mallard, W.G. Anal. Chem. 1979, 51, 1516-1520. Green, R.B.; Harvilla, G.J.; Trask, T.O. Appl. Spectrosc. 1980, 34, 561-569. Havrilla, G.J.; Green, R.B. Anal. Chem. 1980, 52, 2376-2383. Turk, G.C. Ana1. Chem. 1981, 53, 1187-1190. Trask, T.O.; Green, R.B. Anal. Chem. 1981, 53, 320-324. Havrilla, G.J.; Weeks, S.J.; Travis, J.C. Anal. Chem. 1982, 54, 2566-2570. Nippoldt, H.A.; Green, R.B. Anal. Chem. 1983, 55, 554-557. Matveev, O.I.; Zorov, N.B.; Kuzykov, Y.Y. Talanta 1980, 27, 907-908. Zorov, N.B.; Kuzykov, Y.Y.; Matveev, O.I.; Chap1ygin, V.I. J. Anal. Chem. USSR 1980, 35, 1108-1112. Salado-Torrez, L.E.; Zorov, N.B.; Kuzyakov, Y.Y. Schenck, P.K.; Travis, J.C.; Turk, G.C.; O'Haver, T.C. J. Phys. Chem. 1981, 85, 2547-2557. Berthoud, Th.; Lipinski, J.W.; Camus, P.; Stehle, J.L. Anal. Chem. 1983, 55, 959-963. Smyth, K.C.; Mallard, W.G. Combust. Sci. Tech. 1981, 26. 35-41. Mallard, W.G.; Smyth, K.C. Combust. Flame 1982, 44, 61-70. Schenck, P.K.; Mallard, W.G.; Travis, J.C.; Smyth, K.C. J. Chem. Phys. 1978, 69, 5147-5150. Hurst, G.S.; Pyane, M.G.; Kramer, 5.0.; Young, J.P. Rev. Mod. Phys. 1979, 51, 767-819. Donohue, 0.L.; Young, J.P. Anal. Chem. 1983, 55, 378-379. 85. 86. 87. 88. 89. 90. 91. 92. 93. 94. 95. 96. 97. 98. 99. 100. 101. 102. -135- Parker, 0.H. "Ultrasensitive Laser Spectroscopy", Kliger, 0.5., ed., Academic Press: New York, NY, 1983, Chapter 4. Mayo, 5.; Lucatonto, T.B.; Luther, G.G. Anal. Chem. 1982, 54, 553-556. Miller, C.M.; Nogar, N.S. Anal. Chem. 1983, 55, 481-488. Rockney, E.H.; Cool, T.A.; Grant, E.R. Chem. Phys. Lett. 1982, 83, 141-144. Mallard, W.G.; Miller, J.H.; Smyth, K.C. J. Chem. Phys. 1982, 76, 3483-3492. Smyth, K.C.; Mallard, W.G. J. Chem. Phys. 1982, 77, 779-787. Goldsmith, J.E.M. J. Chem. Phys. 1983, 78, 1810-1811. Goldsmith, J.E.M. Opt. Lett. 1982, 7, 437-439. Hansch, T.W. Appl. Opt. 1972, 11, 895-898. Lijnse, P.L.; Elsenaar, R.J. J. Quant. Spectrosc. Radiat. Transfer 1972, 12, 1115-1125. van Dijk,C.A. Ph.0. Dissertation, University of Utrecht, 1978. Donohue, 0.L.; Carter, J.A. Anal. Chem. 1978, 50, 686-687. Caulkin, C. Ph.D. Dissertation, Michigan State University, 1981. Havrilla, G.J.; Green, R.B. Chem. Biomed. Env. Instrum. 1981, 11, 273-280. Technical Series PT-61 "Photomultiplier Manual", RCA Corp.: Harrison, NJ, 1970, 107-108. Christmann, 0. Ph.D. Dissertation, Michigan State University, 1980. Lawton, J.; Weinberg, F.J. "Electrical Aspects of Combustion", lst Ed., Clarendon Press: Oxford, 1969, 231. Turk, G.C. Ph.0. Dissertation, University of Maryland, 1978. 103. 104. 105. 106. 107. 108. 109. 110. 111. 112. 113. 114. 115. 116. 117. 118. 119. 120. -136- Smyth, K.C.; Schenck, P.K.; Mallard, W.G. "Laser Probes in Combustion Chemistry", Crosley, 0.R., Ed.; ACS Symposium Series 134, American Chemical Society: Washington, D.C., 1980, 175-181. Heitler, W. "The Quantum Theory of Radiation", 3rd, ed.: Clarendon Press: Oxford, 1970, 204-209. Stobbe, M. Ann. d. Phys. 1930, 7, 661. Burgess, A.; Seaton, M.J. Mon. Not. R. Astron. Soc., 1960, 120, 121-151. Moskvin, Y.V. Opt. Spectrosc. 1963, 15, 316-317. Ya'akobi, B. Proc. Phys. Soc. 1967, 82, 100-106. Rudkjobing, M. Pb. Kbh. Obs. 1940, 18, 1-15. Gesalov, Kh.B.; Ivanova, A.V. High Temp.-High Press. 1968. 400-404. Caves, T.C.; Dalgarno, A. J. Quant. Spectrosc. Radiat. Transfer 1972, 12, 1539-1552. Weisheit, J.C. J. Quant. Spectrosc. Radiat. Transfer 1972, 12, 1241-1248. Aymar, M.; Luc-Koen E.; Farnoux, F.C. J. Phys. B.: Atom. Molec. Phys. , 1'9. 1976 9, 1279-1291. Bethe, H.-A.; Salpeter, E.E. "Quantum Mechanics of One and Two Electron Atoms", Springer Verlag: Berlin, 1957. Msezane, A.; Manson, S.T. Phys. Rev. Lett. 1975, 35, 364-366. Aymar, M. J. Phys. 8.: Atom. Molec. Phys. 1978, 11, 1413-1423. Rothe, 0.E. J. Quant. Spectrosc. Radiat. Transfer 1969, 9, 49-62. Rothe, 0.E. J. Quant. Spectrosc. Radiat. Transfer 1971, 11, 335-365. Kelly, R.; Padley, P.J. Proc. R. Soc. Lond. A. 1972, 327, 345-366. Gonchakov, A.S.; Zorov, N.B.; Kuzyakov, Y.Y.; Matveev, O.I. J. Anal. Chem. USSR 1980, 34, 1792- 1795. 121. 122. 123. 124. 125. 126. 127. 128. 129. 130. 131. 132. 133. 134. 135. 136. 137. 138. -137- Ackerholt, J.R.; Eberly, J.H. Phys. Rev. A 1976, 14, 1705-1717. Daily, J.W. Appl. Opt. 1977, 16, 2322-2323. DeOliveras, 0.R. Ph.D. Dissertation, Indiana University, 1976. vanCalcar, R.A.: VandeVen, M.J.M.; vanUitert, B.K.; Biewenga, K.J.; Hollander, Tj.; Alkemade, C.Th.J. J. Quant. Spectrosc. Radiat. Transfer 1979, 21, 11-18. Lawton, J.; Weinberg, F.J. "Electrical Aspects of Combustion", lst ed., Clarendon Press: Oxford, 1969, 319-322. Firman, R.J. Spectrochimica Acta 1965, 21, 341-343. Binnen, J.I. Anal. Chem. 1960, 32, 1475-1480. Lucatorto, T.B.; McIlrath, T.J. Phys. Rev. Lett. 1976, 37, 428-431. Collins, C.B.; Curry, S.M.; Johnson, B.W.; Mirza, M.Y.; Chellehmalzadeh, M.A.; Anderson, J.A.; Popescu, 0.; Popescu, I. Phys. Rev. A. 1976, 14, 1662-1671. Bridges, W.B. J. Opt. Soc. Am. 1977, 67, 1417. Vidal, C.R. Optics Letters 1980, 5, 158. Penock, R.; Freeman, R.R.; White, J.C.; Storz, R.H. Optics Letters 1980, 5, 160-162. Alkemade, C.Th.J. Plenary Lecture, VI Annual FACSS Meeting, Philadelphia, PA, September 17, 1979. Bellum, J.C.; George, T.F. J. Chem. Phys. 1978, 68, 134-144. Omenetto, N.; Bower, J.; Bradshaw, J.; vanDijk, C.A.; Winefordner, J.0. J. Quant. Spectrosc. Radiat. Transfer 1980, 24, 147-158. Hellfeld, A.V.; Caddick, J.; Weiner J. Phys. Rev. Lett. 1978, 40, 1369-1373. Bearman, G.H.; Leventhal, J.J. Phys. Rev. Lett. 1978, 41, 1227-12300 Behmenburg, W. J. Quant. Spectrosc. Radiat. Transfer 1964, 4, 177. 139. 140. 141. 142. 143. 144. 145. 146. -138- Muller III, 0.H.; Steinberg, M.; Schofield, K. "Laser Probes for Combustion Chemistry", Crosley, 0.R., Ed.; ACS Symposium Series 134, American Chemical Society: Washington, D.C., 1980; 189-194. Ashton, A.F.; Hayhurst, A.N. Trans. Far. Soc. 1970, 66, 833. Wilson, H.A. Rev. Mod. Phys. 1931, 3, 156. McDaniel, E.W. and Mason, E. "The Mobility and Dif- fusion of Ions in Gases", Wiley: New York, 1973, p. 17. Kra11, M.A.; Trivelpiece, A.N. "Principles of Plasma Physics", McGraw-Hill: New York, 1973, p. 71. Malmstadt, H.V.; Enke, C.G.; Crouch, S.R. "Electronics and Instrumentation for Scientists", Benjamin/Cum- mings: Menlo Park, CA, 1981. Holler, F.J.; Avery, J.P.; Crouch, S.R.; Enke, G.G. "Experiments in Electronics, Instrumentation, and Microcomputers", Benjamin/Cummings: Menlo Park, CA, 1982; Unit 8. Jung, W.G. “IC OP-Amp Cookbook", Howard W. Sams and Co.: Indianapolis, IN, 1974; p. 351. APPENDICES APPENDIX A APPENDIX A TWO-STEP LASER ASSISTED IONIZATION OF SODIUM IN A HYOROGEN-OXYGEN-ARGON FLAME* A. Introduction Laser enhanced ionization (LEI) in flames and plasmas has been observed with various experimental designs6’21’68’128'132. In the usual LEI experiment, ionization results from the absorption of a resonant laser photon followed presumably by collisions of the excited atom with flame gas combustion product56’68. Absorption of a resonant laser photon with an energy E reduces the effective ionization energy and increases the ionization rate6’68 by a factor of exp (E/kT), where T is the flame temperature; this factor can easily be on the order of 104. In this paper we report a relatively simple and inexpensive method for producing laser-assisted ionization in a flame. The method utilizes a two-step process, as suggested by 83’133. The first step is the absorption various authors of a resonant laser photon by an atom in the ground state, followed by the absorption of an off-resonant photon by the atom in the excited state. The sum of the energies of -139- -140- the two photons is such that it exceeds the ionization energy of the atom. The fact that two laser photons are being used to ionize the atom does not necessarily mean that the ioniza- tion proceeds through a straightforward two-step process. It has recently been proposed that photoionization might be collision assisted134. Since collisions are frequent in a flame at atmospheric pressure, this latter possibi- lity was investigated, and some experiments are presented to elucidate the predominant ionization mechanism. B. Experimental The experimental set-up is shown in Fig. A-l. relatively cool (~1800 K) H2-02-Ar flame is burned from a circular Meker burner and shielded by a mantle flame of the same composition to prevent the entrainment of air. Sodium is introduced into the inner flame by nebulizing a sodium salt solution with a pneumatic nebulizer driven by the argon stream. A homemade tunable dye laser of the 93, using Rhodamine 66 as a dye, is pumped Hansch design by a Nz-laser (Model 0.5-150, NRG, Inc., Madison, WI). The pulse duration is $5 ns for both lasers. The dye laser beam is focused in the center of the flame, approxi- mately 2 cm above the burner head, well away from the com- bustion region. The width of the spectral profile of the O dye laser at wavelength of 5890 A was measured using the fluorescence excitation profile as discussed by Omenetto, -141- Figure A-l. Experimental arrangement for simultaneous observation of laser assisted ionization and laser induced fluorescence. a) Overall view. B, boxcar; Dl, dye laser; F, flame, L, lens; M, mirror, 0, oscilloscope; PL, pump laser; PM, photomultiplier, R, x-t recorder, 8, signal; T, uv rejecting filter, TL, trigger line. b) Detailed view of probes and laser beams. IF, inner flame; OF, outer flame; P, probe; R1, load resistor, S, signal. Dimensions are not to scale. The waist diameter of the laser beams is of the order of 0.1 mm; the distance between the probes is a few mm; the diameter of the inner flame is approximately 15 mm. -142- o.) b.) -143- g£_gl.135 and found to be (0.7 i 0.1) A. The wavelength of the N2 laser is 337 nm. Output powers were measured with a pyroelectric power meter (Model J3-O5, Molectron Corp., Sunnyvale CA). Part of the N2 laser beam was split off from the pumping beam and also directed into the flame, where it was focused with a quartz lens. Positioning devices carrying mirrors and lenses facilitate in maximiz- ing the spatial overlap of the two laser beams. The degree of temporal overlap between the two laser pulses can be varied by changing the optical path lengths of one or both lasers. The path lengths were adjusted in these experi- ments for maximum temporal overlap of the two lasers at the burner. The ionization signal was detected with a pair of biased nichrome wires, which had a diameter of 0.7 mm. These probes were mounted on micrometer-driven translation stages and located in the immediate neighborhood of the irradiated region, but were not irradiated themselves. Ionization signals were measured across an interchangeable load resistor in the probe circuit. The signals from the probes were usually processed by a boxcar averager with gated integrator (Model 162-164, Princeton Applied Research, Princeton, NJ) and read out by an x-t recorder. A photomultiplier (RCA 1P28) wired for fast pulse process- ing monitored the fluorescence from the 3P-35 transitions of sodium. Both the ionization and the fluorescence signal can be observed on a storage oscilloscope (model 564 -144- Tektronix, Inc., Beaverton, OR). Appropriate shielding prevents stray light due to the dye laser from entering the photomultiplier. A UV rejecting filter in front of the photomultiplier was used to cut out stray light due to the Nz-laser. Two r.f. chokes, consisting of several wire turns around a ferrite core, were soldered into the H.V. power line of the Nz-laser; both the laser and the chokes were put inside a grounded Faraday cage, consisting of 0.5 mm copper plating. These measures substantially decreased the r.f. noise from the Nz-laser discharge. C. Results and Discussion Experiments were performed to measure the enhancement of the ionization signal with both the dye laser and the N2 laser irradiating the flame over the signal obtained with dye laser alone. For these experiments, the dye laser was tuned to the Na(351/2 + 3P3/2) transition and a Na concentration of 12 pg/ml was nebulized into the flame. The laser peak powers were of the order of 40 kW and 100 kW for the dye laser and the split off part of the N2 laser beam, respectively. The ionization signal was integrated over time. With the dye laser only, an ionization signal of 9.0 mV was obtained; with both lasers the ionization signal increased to 1.3 V. No signal was obtained with the N2 laser alone. The enhancement factor was thus 144 for these experiments. The exact enhancement factor depends on the configuration of the beams and the probes, -145- the intensities of the beams and the composition of the flame. We observed that this factor can be improved by moving the probes closer to the irradiated region. The latter manipulation, however, tends to affect the repro- ducibility of the experiment because of the more critical alignment. Moreover, when the lasers irradiate probe surfaces, non-specific ionization occurs, which is possi- bly due to electrons being liberated from the probe material by laser photons or due to off-resonant multi- photon ionization of atoms and molecules in particle sheaths which might envelOp the probes. To investigate the analytical utility of the two-step ionization technique, the ionization signal and the degree of ionization were measured as a function of the sodium solution concentration. The results are given in Fig. A-2. The dye laser was tuned to the Na(BSU2 + 3P3/2) transi- tion, the N2 laser irradiated the same flame volume, and the time-integrated ionization signal was measured. The analytical curve shows a nearly constant slope of one over more than two decades of solution concentration, and the lowest concentration detected was 0.1 ug/ml. From the fact that a slope of one occurs it can be concluded that associative ionization is not present to any significant extent, since two sodium atoms are required (2Na* + hv + + e-136,137 Na2 + and a slope of two would result. The non-linearity of the curve in Fig. A-2 may be due to a -l46- 100.0 7 .5 J‘_o LL w Normalized Signal (an) '6 , I l 1 'OJ LO 10.0 100.0 Na Concentration ( [1.9/le Figure A-2. Ionization signal (0 ) and ratio of ionization signal to sodium solution concentration (D) vs. sodium solution concentration; both lasers are irradiating the probed volume. The dye laser was tuned to the BSEé-*3P%é (589.0 nm) transition. -147- decrease in the ion collection efficiency caused by the 13 or by ion-electron recombina- formation of charge sheaths tion. A plot of the ratio of the ionization signal to the sodium solution concentration versus the sodium concentra- tion has been previously related to the degree of ioniza- tion6. Such a plot, shown in Fig. A-2,also exhibits non- linearity. This phenomenon has been observed with a different ionization scheme, but in virtually the same f1ame6. That the deviation from linearity in Fig. A-2 was not due to nebulizer characteristics was checked by plotting the thermal emission from the Na(3P + 3S) transitions against sodium solution concentration. The latter plot showed deviations from linearity in a concen- tration range well above that of Fig. A-2,due to the combined effect of self absorption and decreasing atomiza- tion efficiency with increasing concentration. Profiles of the ionization- and fluorescence signal were obtained by tuning the dye laser across the Na(BS”2 + 3P3/2) transition, and are shown in Fig. A-3 The sodium solution concentration nebulized into the flame was 12 ug/ml. The difference in the full widths at half maximum (FWHM) which is evident from the profiles of Fig. A-3,has been observed in a related experiment by 6 van Dijk,et al. where a flashlamp laser was used. n J u . 3 r2... -1: .. .. :1 .> oovu ammuaokr onoum :HE\mn NH" Godumuucmocoo :oflusHOm EsficOm .oHflm—oum :oHumecofi How < m...” can oawmoum mocoomouosam HON < bonanza ~\ ~\ .cofluwmcmwu oz AS: c.mmm mmm ou dmm on» mmouom momma can unaccmom can pocfimuno A O. Hmcmwm mocommuoaam can AD V Hmcmfim :OHHMchoH mo oawmoum .ml< gunman. .825on .228 SE. n :2 9.2.28 0N ON 0.. 0.. no 0 mo- _. 0... ON- - . _ 1 . . _ q _ _ I g o -148- L (710) 1011593 9341M esseésesa -149- The fluorescence excitation profile in Fig. A-3 essen- tially represents the spectral laser profile because the width of the Na(3P + 35) transitions is known to be approximately 70 mA in our flame138, which is small com- pared to the observed width. Saturation of the Na(3P”2 + 3P3/2) transition hardly affects the latter conclusion, since the broadening of the excitation profile depends on the square root of the natural logarithm of the dye laser power135 and is therefore a weak function of this power. The measured FWHM of 0.7 A of the fluorescence excitation profile is consistent with the value which we obtained when using a monochromator to resolve the spectral laser profile. Broadening of the ionization profile might be related to saturation of the laser-excited level, space charge 13, and possibly recombination6. effects On a us time scale the onset of the recombination might occur and this might partially account for the dif- ference in widths observed in the profiles of Fig. A-36. As noted by Muller, gt_gl,139, appreciable charge densities are needed to give rise to noticeable recombina- tion on a us time scale. We offer the following hypothesis to account for recombination under our experimental condi- tions: the possible existence of charge sheaths around the probes might partially mullify the electric field between the probes and lead to an increase in the contribu- tion of diffusion to the motion of the charges in the -150- 'field-free' region. Therefore, a significant number of ions might be able to overcome the repulsive field at the positive probe. These ions have a higher probability of being neutralized because of the electron sheath envelop- ing the latter probe. An analogous reasoning applies to the electrons. As the charge densities in the sheath are higher than in the surrounding plasma, the recombination rate will also be higher near the sheaths. The fact that space charges diminish the field might also decrease the collection efficiency of the probes and give rise to broadening of the ionization profile. We investigated the dependence of the ionization and fluorescence signals on small variations in flame gas composition. The influence of changing the oxygen content on the signals is quite noticeable, but different for the excited state population and the ionization. However, the ratio of the ionization signal to the fluorescence signal shows only a weak dependence on the 02 concentration. The ratio of the ionization signal to the fluorescence signal is proportional to the degree of ionization 8, provided 8 is small and the 35”2 + 3P3/2 transition is saturated. s a I~a*I/(I~a*1 + [Na]) = [Na*I/I~a*I . i/F (I) * O 0 where [Na ] is the density of the exc1ted sodium atoms; i = ionization signal and F = fluorescence signal. -151- In an analogous experiment, where we varied the argon content of the flame, a similar weak dependence was observed, as is evident from the curves in Fig. A-4. From these experiments one can conclude that if the degree of ionization is low, the discrete to continuum step is pre- dominately photon absorption and thus affected only slightly by collisions with 02 or Ar. On the other hand, if the degree of ionization B is high, this conclusion does not necessarily hold because a variation in the ionization rate ' ', [fl "Jw \rL.! will have little influence on B for 8 close to unity. Rough calculations indicate that the degree of ionization within [r .. the laser irradiated volume may indeed be quite high. When the discrete to continuum step is brought about solely by collisions, the dependence of i/F on the 02 concentration in the flame is much more pronounced6’124. Changing the voltage across the probes results in a change of ionization signal, as is shown in Fig. A-S. For this experiment we used various sodium solution concentra- tions, and the dye laser was tuned to the 3S”2 to 3P3/2 transition. With increasing voltage the curves are seen to level off, especially those og the higher concentra- tions. The decreasing slopes indicate that the current drawn from the plasma approaches saturation, which means that with high voltages eventually all the available charge carriers in the probed region are being collected. -152- Normalized Signal (an) lm 1— _io :r’ ‘ . m _ 2" : A . A 80 - . pr... 70 ' . (SC,- . : 50 ,_ . 40 " I 30; “L" CI 1 l l l l I 60 7.0 8.0 9.0 10.0 i 1.0 Argon Flow (L/minutel Figure A-4. Dependence of ionization and fluorescence signals on Ar content of flame. The dye laser was tuned to Na(BSBé-r3Paé) transition; the sodium concentration was 12 g/ml. (O)f1uorescence signal plotted vs. Ar flow; (D ) ionization signal plotted vs. Ar flow; (A ) ratio of ionization-to-fluorescence signal plotted vs. Ar flow. -153- 10000—- O :000— ' ‘3 3 :00— '8 c 5? m 4p 1 (1- 10b 1 1 1 l 1 l 1 J l 2 5 IO 20 50 100 200 500 1000 Probe Voltage Figure A-S. Dependence of ionization signal on probe voltage for various sodium solution concentrations. The dye laser was tuned to the 3Sllé-r3P3/2 transition. A dc component in the signal, probably due to thermal (non-laser) ionization, was eliminated with a capacitor in the signal line. (.) sodium concentration = 10 ppm; (‘ ) sodium concentration =1 ppm; ( I ) sodium concentration = 100 ppb. 454- A check as to the transport mechanism can be obtained by plotting the measured decay time T of the ionization signal versus the reciprocal of the applied probe voltage 2gp. According to eqn. (A13) of the appendix, such a plot is expected to yield a straight line: indeed we found this to be the case. From the slope of this plot we cal- culate, using eqn. (A13) with 2L - 2.8 mm, that the ionic 2 mobility u] is on the order of 20 cm V'ls'].' When we assume that the additional decay time I introduced by the e circuitry, especially the load-resistor (= 100 k0), is simply added to the characteristic decay time Ti, we find from the intercept of the plot of decay time versus inverse probe voltage that Te = 20 us. Hence the stray capacitance of the circuitry is m 200 pF, in reasonable agreement with estimates based on cable length and input capacitance of the apparatus. Using the relation pi = qD/(kT), where q = elementary Boltzmann's con- charge, 0 = diffusion coefficient, k stant and T = flame temperature, we calculated the order of magnitude of pi from measurements of the diffusion coefficient in flames by Ashton and Hayhurst140 and found our result to be within the range of the latter authors. Electronic mobilities are expected to be much larger be- cause of the difference in mass between the ions and the electrons‘4]. When we calculate the decay time under the assumption that diffusion is the predominant transport 142 mode, we find, using Tdiff = (2L/n)2/0 , and -155- D = pikT/q, Tdiff is of the order of several milliseconds and is therefore much larger than our observed values. Two conclusions pertaining to the particular conditions of our experiment can be drawn from these experiments: 1. The predominant transport mechanism is drift due to the electric field of the probes. 2. The current is limited by the ions. It follows from our measurements and from eqn. (A13) that the actual decay time of the charge cloud is several microseconds. In order to eliminate the contribution of the circuitry to the decay time constant, we plotted this contrast 13° the value of the load resistor. These latter measurements support our observation that T] is of the order of a few us, since for low values of R5Z the observed time constant becomes independent of RI' The temporal dependence of the ionization signal was measured with a load resistor R2 of 5 k0. The peak ampli- tude Vo of this signal was 12 mV and the decay was close to exponential with a time constant T of 7.9 i 0.2 us. An estimate of the total number N of ions collected in this pulse follows from (I) -t/t _ V0. é’e dt - Nqu (2) where q is the elementary charge. From these results N is found to be approximately 108. Note that the exponential -156- decay is consistent with eqn. (A4) of the appendix. By tuning the dye laser across the 35-40 two-photon transition of sodium and using a 100 ug/ml sodium solution concentration, the profile shown in Fig. 6 was obtained. The signal from this transition is weaker than the signal of a 3P-3S transition when referred to the same solution concentration. The peak observed with the dye laser only is probably due to collisional ionization proceeding from the 4D levels’lz. The fact that very little enhancement of the latter peak occurs when the Nz-laser is present might mean that the ionization induced by the dye laser and the collisions is nearly complete, or alternatively, that the cross-section for photoionization from the 4D- 1eve1 to the continuum is small. The continuum signal observed with both lasers irradiating the flame disappears when the dye laser is blocked or when the N2 laser is blocked. This indicates that the continuum signal is due to off-resonant two-photon ionization by both lasers. 0. Conclusions We have shown that ionization of seed atoms in the flame can be achieved relatively simply by irradiating the atoms simultaneously with one resonant and one off- resonant laser pulse, both of which have a duration of a few nanoseconds. From the experiments conducted we conclude that the predominant mechanism for ionization proceeding from the Na(3P) level is two step .mawmoum ooawmumo an .o>«umummo Emma momma m>p any can soon nomad «2 may anon "mocha um3oH «noumsuwmucw Emma nomad mz “oomwu Homo: Am .aE\mn ooanucoHuawucmocoo coflusHom EaHpOm u> coauumem a .cx own" m .<.>mhmn"£umcoao>m3 woucmo «coHuflmcmuu :ouosmlo3u Anvi.mmvmz may mmonom nomad who on» mcwccmom an omcwmuno moawmoum coHumNHcoH £4 95:28 3.: 05:28 A. z. A o 4:- a 1.. A o 2-8.0 .ml4 owsmwm O -157- '1 (I l ‘0. O -158- photoionization; in the case of population of the Na(4D) level, collisions cause the excited atom to ionize. The described method to measure ion mobilities has been tested and been found to yield a value consistent with previous results. Although the detection limits by the two-step method are not as low as those obtained by normal LEI12 it is expected that perfection of this method will improve these If: limits. Feasible improvements are: using a slot burner and extended probes running parallel to the laser beam for several inches to increase the collection efficiency of the probes; using an ultrasonic nebulizer instead of our L“ pneumatic nebulizer to increase nebulization efficiency; narrowing the bandwidth of the dye laser to increase its spectral irradiance. E. Appendix In this appendix we develop expressions for the mobility-controlled ion density in a flame. As an approxi- mation to our experimental conditions we consider two parallel plane probes at a distance 2L from each other, with a voltage difference 2gp applied to them by an external source. The probes are located inside the flame and the charge density is assumed to be uniform throughout the space between the probes when the applied voltage is zero. -159- We are interested in the case in which the distribu- tion of the ions is primarily determined by the electric field; consequently, we neglect here diffusion, the over- all movement of the flame gases, recombination, and ion production. The continuity equation for the ion density then reads .__1 = - H.V.(n.E), (A1) where ui denotes the ionic mobility and E denotes the electric field. If the contribution of the laser induced charges to the field E is small, then E is nearly constant during a laser pulse. The latter assumption allows us to separate variables as follows "1(XSYszat) = Ni(x,y,z)fi(t). (A2) Substituting enq. (A2) into eqn. (Al) gives 0) "h 'C 1 -—1-= —iv.(NiE). (A3) fi(t) 3t N1 Since the left hand side of eqn. (A3) depends on t only and the right hand side on the spatial variables only, we set both sides equal to a constant -1/Ti. The solution for fi(t) is found to be fi(t) = e‘t/Ti (A4) The equation for the steady state can be written as -160— V.(aA) = aV.A + A.Va (A6) where a and A represent differentiable, but otherwise arbitrary scalar (a) and vector (A) functions of the Spatial coordinates, eqn. (A5) becomes ~ ~ NiV.E + E.WNi - Ni/(“iTi) = 0 (A7) If we assume that the field E is nearly uniform throughout the probed region, then E = - ¢p/L. (A8) If our attention is restricted to the one-dimensional case, then substitution of eqn. (A8) in (A7) gives — = - NlL/(¢pu1T1)° (A9) Hence "i = n epr-xL/(4puiri11 (A10) where n is a constant of integration. Thus, the ion density is highest near the negative probe and decreases in the direction of the positive probe (located at x = + L). Identifying Ti with the characteristic time it takes for an ion to drift from x = O to x = - L L = -V.i‘l'i, (A11) where vi is the average drift velocity. In the limit of a -161- low, uniform field V. = u.E. (A12) Combining eqns. (A8), (A11) and (A12) r, = Lz/(u,¢p) (A13) establishes a relation between the characteristic decay time and the ionic mobility in terms of the distance be- tween the probes and the applied potential. Using eqns. (A10) and (A13) Ni = n exp(-x/L), (-L < x < L) (A14) which agrees with the solution found for the so-called plasma-capacitor in the limit of low charge density143. Due to the absence of the diffusion term in eqn. (A5), the charge density does not obey the boundary conditions Ni(j'_L) = 0. *Obtained jointly with K.C. Lin and C.A. van Dijk. APPENDIX B APPENDIX 8 DYE LASER DESCRIPTION AND ALIGNMENT PROCEDURE A. Introduction A tunable organic dye laser system of the design first described by Hansch93 was used in all the experiments. In this design a laser dye solution in a quartz cuvette is incorporated into a short Optical cavity formed between a diffraction grating and a partially reflecting output coupler. The dye molecules, which function as the active material, are optically pumped by a repetitively pulsed nitrogen laser placed at a 90° angle to the cavity. Laser dyes are large multiringed organic structures which contain conjugated double bonds and absorb strongly in the ultraviolet. The energy levels of these molecules are 1 to 2000 cm'], because line wide, typically 1000 cm' broadening processes in solution smear the rotational lines to a continuum around the vibrational lines composing the individual electronic states. Thus the fluorescence emission is broadband in nature, which allows the laser to be tuned. The wavelength fed back through the active region to initiate the laser action can be selected with the grating. The laser output consists of a train of short pulses of high intensity. Repetition rates up to -162- -163- 60 Hz are possible. Extremely low bandwidths can be obtained with the inclusion of additional wavelength selective elements (g;g; a Fabry-Perot etalon) if necessary. The large gain bandwidth also permits the generation of ultrashort (picosecond) pulses through modi-locking. The wavelength region from about 350 nm to 800 nm is accessible with currently available, high-gain laser dyes. The remainder of this Appendix consists of a descrip- We tion of the components of our optical system and their § functions followed by the alignment procedure for the dye 5 laser. 3 8. Optical System Components A block diagram of the entire laser system is given in Figure B-l. The laser cavity is formed by a diffraction grating and a quartz output coupler. The grating was mounted backwards on the sine-bar of a Czerny-Turner grating monochromator (Model EU-700, GCA/McPherson Instruments, Acton, MA) to provide for the rotation needed for tuning. The monochromator sits on an aluminum table equipped with three independently adjustable legs. The table is positioned so that it is directly above the Optical rail that holds the rest of the dye laser cavity components. The precision plane grating replica used was the standard grating available with the monochromator (48 square millimeter ruled area with 1180 lines/mm). The blaze wavelength was 250 nm. This grating Was adjusted on the -l64- . .Emumam momma mo Emumawo xoon .Hlm ousmwm 23818 1 . J . $3 5.9550211 0 a I198 Pg 3:5 523 .s. F L 3.2.: I! .53 .2. ~.§..J\ 11.35.. event» 9!.6 E23 .32 cz8zz -165- sine-bar so as to operate in the second order (see alignment procedure section-step 12). A 4% reflective quartz flat (1 inch diameter) was used as an output coupler for all of the experiments performed. This flat was mounted in a three point adjustable holder attached to a pole that was fixed in a mount riding on the optical rail. The distance between the ends of the cavity varied from alignment to alignment but was always between 30 and 35 cm. The dye laser cuvette had four quartz sides with single layer antireflection coatings. The two sides perpendicular to the cavity were offset at a slight angle to prevent them from acting as cavity end mirrors. This cuvette was obtained from a commercial source (Type 509, Precision Cells, Inc., Hicksville, NY) and had a path- 1ength of 8 mm. It was mounted in a Delrin holder on an aluminum frame and attached to the optical rail in a fashion similar to the output coupler. The frame slides on a dovetail to give adjustability across the rail. Also. the frame is fitted with a small DC motor with a magnetic stirring bar attached to the shaft. The bar can spin freely under the Delrin holder. This arrangement was constructed to provide mixing of the dye solution with a miniature magnetic stirring bar immersed in the dye solution. Studies indicated, however, that stirring causes vibrations which can misalign the cuvette; thus use has been discontinued. The motor controller circuit has been described elsewhereIOO. No further attempt at mixing was -166- made because the output stability of the dye laser was deemed sufficient for the experiments performed. The final cavity element is the inverting telesc0pe (Model 1592, Oriel Corp., Stamford, CN), which collimates and expands the incoming beam so that it nearly fills the grating. The Galilean design incorporates a diverging input lens (in the final configuration this has a focal length of -4 mm) and a converging output lens with a focal length of 160 mm. The mount, attached directly to the Optical rail, provides adjustment both horizontally and vertically (across the rail). Also micrometer drives provide “tilt" (iLQ; angle tuning) in both the horizontal and vertical planes. A 5 mW randomly polarized helium neon laser (Madel 05-LHR-151, Melles Griot, Irvine, CA) is mounted on an optical rail at 90° to the laser for cavity alignment. The mount is a nine inch V-groove to accommodate easy removal and replacement. The beam is brought to the cavity with a commercially available, two mirror, laser beam aligner (Model 6650, Oriel Corp., Stamford, CN). To direct the beam down the cavity a flat mirror held in a 3-point adjustable mount is placed on the Optical rail containing the cavity components. To bring the nitrogen laser pumping beam to the dye cuvette a quartz optical flat (2 inch diameter) and a cylindrical lens are used. The flat changes the beam spread to the cylindrical lens and is a coarse adjustment -167- only. The cylindrical lens is set in a circular Teflon holder and this is then placed in a three-point gimbal mount. The Teflon holder can be rotated 360°. This entire arrangement is mounted to provide motion in three dimension. C. Alignment Procedure The following procedure assumes that all the dye laser components are completely misaligned, including those involved in directing the nitrogen beam to the dye cuvette. A number of the steps (denoted by asterisks) can be omitted if some components are known to be aligned. All components are set in mounts (many atop poles) that are free to slide on the Optical rails. The target referred to is a square card, with measured markings, attached to a pole set in a rail mount. 1. Remove all components from the dye laser Optical rail. *2. Start the nitrogen laser and adjust (either with nitro en laser table legs or with cylindrical mirror mount? the beam direction so that the beam strikes the cylindrical lens in the center. 3. Place the target on the optical rail and turn the face 90° so that the beam (nitrogen laser) strikes it. Adjust the height of the target so the cylindrical beam, when focused, falls on the horizontal cross hair. This will be the height of the center of the cavity. 4. Turn off the nitrogen laser, rotate the target face back and turn on the He-Ne laser. *5. Make sure that the He-Ne laser beam is striking the lower mirror of the laser beam alignment unit. -168- 6. Place the reflecting mirror on the dye laser optical rail so that it forms roughly a 45° angle with the rail and so that the alightment beam strikes it. If the mirror is at 45°, the beam should be directed down the rail (roughly-finer adjustments will follow). 7. Move the target to a position approximately 3 inches from the 45° mirror, and use the laser beam aligner to adjust the beam so that it strikes the target in the center. 8. Move the target to the other end of the cavity (about 2 inches in front of the grating). 9. Use a screwdriver to adjust the mirror mount so the ,- beam is centered once again on the target (this may take more than one attempt as the 45° mirror may have to be moved slightly or twisted). Then return . the target back to its earlier position near the 45° ] mirror to make sure the spot is still centered. If i not, repeat steps 7-9. ‘ Upon completion of step 9 the He-Ne laser is directed E straight down the rail and the dye laser components can now be aligned with it. 10. Remove the target from the rail and adjust the position of the grating (by sliding the monochromator on its table) so that the beam strikes it roughly in the center. At this point the grating must be aligned in the second order with the wavelength selector set to allow a sufficient tuning range. With the grating reversed on the sine-bar and operating in the second order, a change of approximately 2.5,A on the wavelength counter equals a change of l A in the laser output wavelength. Since the He-Ne laser line at 6327 A is used, the cavity will be aligned at this wavelength at the end of the procedure. For the laser to lase at 4000.A, approximately the lowest wavelength normally used, the wavelength counter must be free to move, -169- in a downward direction, a number of Angstroms given by (6327-4000) x2.5 = 5818A Similarly to lase at the approximate upper limit of 7500 A the counter must be free to travel about 2930.A upwards. As the full counter range is 10,000 units, an initial wavelength selector setting of 7000.A will cover the entire useful range. 11. Set the wavelength selector to 7000 A. *12. Set the grating angle so it is Operating in the second order. Loosen the screw on the sine-bar (near the grating holder) so the grating is free to pivot. Twist the grating so the second order reflection is directed down the cavity (to be sure, count the orders: the zero order reflection will be stopped by the stray light shield between the grating and the folding mirror. The first . order reflection will miss this and can be viewed either on the wall or on a card held near the grating. The next reflection is the one wanted for cavity alignment). *13. Tighten the holding screw. *14. Loosen the monochromator holder and translate (slide) the monochromator on the table so that the alignment beam is hitting the grating in the center. 15. Make minor table leg adjustments so the second order reflection is collinear with the incoming beam. The grating is now aligned with the He-Ne beam. 16. Attach the telesc0pe mount to the rail so that the telescope output lens is about 5 cm from the center of the grating. 17. Use the course (non-micrometer) adjustments on the telesc0pe mount to position the telescope so that the alignment beam strikes the 4 mm input lens; then remove this lens. -170- 18. With the lights out, two diffuse (blurry) spots will be seen in the far field (i.e. a card near the 45° mirror). Use the micrometer drives to bring these together. 19. Insert the 16 mm input lens. A card held behind the 45° mirror will Show a large spot of light. Slide the output lens sleeve so that this focuses to a pinpoint. A bright area, 2-3 cm in diameter, should now be visible on the grating. Use the course adjustments to move this to the grating center. and then use the micrometer drives to direct the pinpoint so it falls on the alignment beam spot on the 45° mirror. 20. Replace the 16 mm lens with the 4 mm lens. The spot on the grating should be larger and dimmer than before (in fact, will more than cover the grating). A card behind the 45° mirror will once again show a large spot which should be focused as much as possible (although this will not focus to a pinpoint as before). Once again, position the spot, with the micrometer drives, to hit the reflection of the alignment beam on the 45° mirror. {‘5’} 'C - “L's?"‘v'i l max-Hum”! The telescope is now aligned as much as possible at this point. Minor adjustments will be needed later. (Note: the diffuse reflected beam found with the 4 mm lens is sometimes difficult to find behind the 45° mirror. It is because of this that the 16 mm lens is used for "rough" initial alignment. If there is still difficulty, there is an 8 mm input lens that can be inserted after the 16 mm lens is removed (i;g; repeat step 20 but with this 8 mm lens). This will make the step down easier but more tedious.) 21. Place the output coupler on the rail 30 cm to 35 cm from the grating. Adjust the mount so that the reflection lines up with the incoming beam. 22. Place the cuvette holder on the rail so that the cuvette is in front of the cylindrical lens that directs the nitrogen laser beam into the cavity. -171- 23. As the nitrogen beam only excites a small volume near the first quartz face that it strikes, it is necessary that the He-Ne beam go through this area. Move (slide) the cuvette across the rail so that the He-Ne beam is as close to this face as possible. The cavity should now be aligned at the He-Ne laser wavelength. Set the wavelength counter to a reading appropriate for the dye in use. For example, if you want to use Stilbene 420 which, in methanol, has a maximum in its laser gain band at 425 nm, the wavelength counter F7 setting is given (approximately) by 7000 -(6327-4250)(2.5) and thus should be 1808.A. For other dyes the start up )- setting is obtained simply by replacing 4250 in the above equation with the wavelength of maximum gain for the dye of interest. 24. After settin the wavelength counter to the appropri- ate reading Isee above), start up the N2 laser and hope for the best. The laser will probably not lase at once and small cavity adjustments may be necessary. First make sure the cylindrical lens is focused near the inside wall of the cuvette. 25. A card held in front of the input lens of the telescope will show a bright spot from the dye solution. This should strike the 4 mm lens (i.e. fall on the same area as the alignment beam). If it doesn't use the mount of the cylindrical lens and/or twist the dye cuvette slightly so that it does. 26. If the laser still is not lasing use the micrometer on the underside of the telesc0pe to adjust the horizontal tilt angle. A card held behind the output coupler may show a reflection that is not a spot but a dim line. This should be brought down so it is even with the alignment laser. At this point laser action is almost always achieved. Once laser action is found it can be improved by 27. -172- making minon adjustments to the end mirror, telescope, etc. The pure, sometimes grainy, reflection on the grating (at the dye laser wavelength) can be seen and centered. Reflected laser light may appear as a spot near the main laser spot. This can be removed (if necessary) by placing a diaphragm in the cavity. This is usually taped to the output coupler mount. 'L—J‘J- 1 re 1'. APPENDIX C APPENDIX C LOCK-IN AMPLIFIER DESIGN A. Introduction A lock-in amplifier is often used to enhance signal- to-noise ratios (S/N) in Spectrometric measurements. Typically, the slowly varying dc signal generated by the experiment is used to modulate the amplitude of a carrier wave which has a frequency well removed from dc where 1/f noise can predominate. The lock-in amplifier then selectively amplifies the signal, synchronously demodulates it by multiplication with a bipolar reference square wave of equal frequency and phase and finally low pass filters it. The output, then, is an amplified version of the input. Lock-in amplifiers have been described in more detail by Malmstadt, Enke and Crouch144. A lock-in amplifier system, which is a modification of a design given by Holler, gt_al.145, is described in this Appendix. The design is simple and inexpensive as it involves only three integrated circuits other than common Operational amplifiers. Still, the lock-in amplifier is versatile enough to handle signals generated in all common laboratory experiments such as atomic -173- -174- absorption, thermal emission, etc. Carrier frequencies between 10 and 10,000 Hz can be used. The unit is intended for use with the atomic absorption spectra- photometer described in chapter VI. This atomic absorption instrument uses a carrier frequency of 285 Hz. Because of this, the tests of the lock-in amplifier described in this Appendix were performed at this frequency. B. Design Considerations B-l. General The simplified circuit for the lock-in amplifier is shown in Figure C-1 and the detailed schematic is shown in Figure C-2. The basic design consists of a signal channel for tuned amplification of the modulated carrier wave, a reference channel to provide a bipolar, phase- shiftable square wave, a multiplier for synchronous demodulation and a low pass filter to provide a (amplified) dc output. All of the operation amplifiers (op amps) used were common single IC dual op amps (either Model TL082, Texas Instruments, Inc., Dallas, TX or Model LF353, National Semiconductor Corp., Santa Clara, CA). Both the reference and signal channels employ AFlOO universal active filters (National Semiconductor Corp., Santa Clara, CA). These are configured similarly and described together. -175- .Amflmhaaea cflixooa pom snowman baseman emanaaaehm .Huo masons SO sotim omoca a 3:.“— 3.3222 c2025 .. v m 82.36.. 593 00 . 3.32:: So 9:4 . 2:5... 532.5 omega A. 8.82.58 3...... .uuonucom , 02.04 53:... , «mango . o>=u< hum buts—m u. S . .255 -176- .HOMMflHQEm :Hixoofl MOM Eaumawp ufisoufio pwafimuoo .mlo owsmflm =5 Hog-gun”. ochuiu Had—t Jug uuzucunuc 42.33 .38.» . 30.43.th -177- B-2. Inputs The modulated carrier is the input to the signal channel through a simple voltage follower buffer. The reference signal is connected to a comparator followed by a voltage divider. With this type of input even low level signals can be used for the reference since the comparator (with the non-inverting input grounded) will produce a bipolar square wave of approximately 115 V. The voltage divider (3:1) then reduces this to a usable :5 V. B-3. Active Filters The active filter in the signal channel acts as the tuned amplifier to selectively amplify the modulated carrier wave. In the reference channel the active filter converts the reference square wave to a sine wave. This is necessary as the phase-shifting circuit works only on sine waves. The AF 100 filters are identically configured and operated in the band pass mode. To tune the center frequency of the filter two resistors are used, one between pins 7 and 15 and one between 14 and 16. The tuning range of the basic filter is 200 to 10,000 Hz. Lower frequencies (to dc) can be obtained by inserting capacitors between pins 7 and 8 and 15 and 16. To provide continuous variability in the filters, variable potentiometers were used in parallel with switch- selectable fixed resistors. This arrangement was used to maximize the frequency selectivity as for each range of frequencies almost the entire range of the variable -178- resistor is used. Since the frequencies of the AF lOO filters in each channel must tune simultaneously, a four level modular potentiometer was used. This was assembled from two modular potentiometers purchased from a commercial source (Series 70 "Mad Pot", Allen-Bradley Co., Milwaukee, WI). Each of the four modules has a 250 k0 tuning range and the unit was equipped with a 16 turn verneir drive for separability. The parallel, “Ln- -_q fixed resistance was chosen with a 4 pole, 4 section, 11 position rotary switch (Series 1400, Centralab, Inc., Menomonee Falls, WI). The fixed resistors were matched. "I." ' -II. The switch positions, frequency ranges and fixed resistance values are listed in Table C-l. For ranges below 200 Hz a 4 pole double throw switch was used to connect the capacitors (.022 uF). For frequency ranges above 200 Hz the center frequency of the filter is given by r0 = 50.33 x 106/R and below 200 Hz by 6+1 XlO'g) _h ll 0.05033/R(0.022 x10- where R is the value of the tuning resistance. The quality factor of the filter, 0, determines the band- pass of the filter. 0 is determined by the value of the input resistor (RIN) and the Q tuning resistor (RQ). These values were chosen to be 5 k0 and 1 k0 respectively. -179- TABLE C-l. Frequency Ranging Data. Switch Frequency Fixed Position Range (Hz) Resistance (k0) 1 10-20 none 2 20-40 200 3 40-80 75 4 80-160 32.4 5 160-200 14.7 6 200-400 none 7 400-800 261 8 800-1600 90.9 9 1600-3200 38.3 10 3200-6400 16.8 11 6400-10,000 8.25 -180- in the filter configuration used this gives a O of approximately 35. At the 285 Hz test frequency, the band- pass (from upper to lower 3dB points) with this 0 is 281-289 Hz. B-4. Phase Shifter The circuit used to adjust the phase of the reference wave to match that of the signal is a constant-amplitude phase shifter which has been described elsewhere146. In F this a ten turn potentiometer (100 k0 maximum value) is varied to determine the amount of shift. To function properly the input must be a sine wave; the output of the circuit is also a sine wave. In order to regenerate r a square wave, which produces optimal results upon multiplication with the signal channel output, the output of the phase shifter is directed to a comparator followed by a 3:1 voltage divider. 8-5. Multiplier The outputs of the signal and reference channels are input to a four quadrant multiplier (AD 534, Analog Devices, Norwood, MA). This provides the synchronous demodulation of the carrier wave through cross-correlation. This strongly discriminates against random noise which is not in phase with the signal and reference waves. The output of the multiplier is a full wave rectified sine wave whose amplitude is proportional to the input signal. -181- B-6. Low Pass Filter The output of the multiplier was fed through a low filter to produce a "smoothed" dc level prOportional to the input signal. The time constant of this filter was chosen to be 1 second for the tests to be described in section C. The final version of the amplifier (now under construction) will have a variable output time constant. 8-7. Outputs As this lock-in amplifier was intended for use in an instructional laboratory situation, a number of outputs in addition to the normal dc output were provided. The outputs of the tuned amplifier, the multiplier and 11““ the reference channel were all buffered and taken to BNC connectors on the case. C. Testing To test the lock-in amplifier a signal was generated by pulsing an LED with the TTL output of a function generator. This was detected by a photodiode whose output was fed to a current-to-voltage (I/V) converter. To add noise to the signal output of a noise generator circuit based on the 5837 digital noise generator IC was used. This type of signal source has been described in more detail elsewhere145. The output of this circuit was connected to the signal channel input. The reference wave was obtained from the bipolar square wave output of the function generator. The reference channel was -182- found to work as expected for input peak-to-peak square wave amplitude as low as 20 mV. For initial tests the function generator was set to 285 Hz, and the noise generator circuit output was disconnected. A series of photographs of the various inputs and outputs of the lock-in under these conditions are presented in Figure C-3. Figure C-3a shows the actual input signal (lower trace) and the output of the F tuned amplifier (upper trace). The fluctuations in the ' input signal were caused by the room lights. These were strongly attenuated by the tuned amplifier. The 1?- u. _w____ data also show a linear gain in this step of approximately 25 (30 dB). The upper trace in Figure C-3b shows the output of the reference channel and the lower trace shows the output of the multiplier (in-phase condition). Finally, the actual dc output is shown by the upper trace of Figure C-3c. The lower trace is the baseline or "signal blcked" condition. Shown in the series of photographs in Figure C-4 are the same series of measurements with the noise generator connected. The lower trace of Figure C-4a shows the signal buried in random noise (~0.8 V p-p). As the original input signal was on the order of 0.1 V the input S/N ratio was about 0.125. The upper trace shows the output of the tuned amplifier. The reference wave is once again Shown in Figure C-4b along with the -183- Figure C-3. Lock-in amplifier without noise. (a) lower trace: signal channel input; upper trace: output of tuned amplifier. (b) lower trace: output of multiplier; upper trace: output of reference channel. (c) lower trace: dc output with signal “blocked"; upper trace: dc output with signal "open". pilfiiiliu- .‘n-MAI-un Nlfllfllll. . fillflflllh ,nuuugll! I'Iill'Iili I! — ‘ _— \ -—-~— , ., b) “IIIIII . NAME-1M7“ V Ell-“III"; l .lE .‘ i a Figure C-3. -185- Figure C-4. Lock-in amplifier test with noise. (a) lower trace: signal channel input; upper trace: output of tuned amplifier. (b) lower trace: output of multiplier; upper trace: output of reference channel. (c) lower trace: dc output with signal "blocked"; upper trace: dc output with signal "open". a) b) IIIIIIIIII IIIIIIIIII LII-III]...- C) Figure C-4. -187- output of the multiplier. Finally, in Figure C-4c the dc output is shown. From this, the signal was measured as 5.6 V. The peak-to-peak noise on both the upper and lower traces is approximately 0.4 V so the actual noise in the measurement can be estimated as 0.56 V. The S/N, then, has been increased to 10 so the total S/N enhancement by the lock-in amplifier was 80.