.— -U...M_,.— L! 2: EM? 3’3 )3ng 2m“; MIG U331! \7’ {CE iii-CV (a if: This is to certify that the thesis entitled THE DESIGN AND CHARACTERIZATION OF A MINIATURE, NANOSECOND SPARK DISCHARGE FOR ANALYTICAL EMISSION SPECTROSCOPY presented by Gary Thomas Seng has been accepted towards fulfillment of the requirements for Ph . D . degree in C. W W/M Nljor professor Date SGDtembep 29, 1978 0.7 639 OVERDUE FINES: 25¢ per day per item RETUMING LIBRARY MATERIALS: \ Place in book return to remove charge from circulation records THE DESIGN AND CHARACTERIZATION OF A MINIATURE, NANOSECOND SPARK DISCHARGE FOR ANALYTICAL EMISSION SPECTROSCOPY By Gary Thomas Seng A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Chemistry 1978 ABSTRACT THE DESIGN AND CHARACTERIZATION OF A MINIATURE, NANOSECOND SPARK DISCHARGE FOR ANALYTICAL EMISSION SPECTROSCOPY By Gary T. Seng The design, construction, and characterization of a miniature, nanosecond spark is described. The present spark design features a coaxial, double-gap system powered by a direct current power supply. Pure argon continuously flushes the secondary gap. The analyte solution is nebulized via an ultrasonic nebulizer and desolvated in a heated chamber. The dry aerosol is swept into the analytical gap in a laminar stream of argon or helium directed either parallel to, or perpendicular to the inter-electrode axis. A quartz lens collimates the discharge radiation onto the slit of a programmable monochromator. A gated photomulti- plier tube transduces the emitted radiation into an elec- tric current. Integration of the photocurrent is performed under the control of digital time-resolution electronics with time windows adjustable in 0.1 us increments. Timing is initiated by a fast photodiode-Darlington system Gary T. Seng which monitors radiation from the secondary gap. Time-resolved studies of the excitation temperature, ionization temperature, electron density, and background emission are presented and discussed. For analytical determinations, the analyte emission line is selected under minicomputer control. Additionally, the minicomputer directs the time-resolution circuitry, acquires emission data, and performs statistical calcula- tions. The miniature Spark has been found to be an excellent source for the determination of a number of elements including Ca, Al, B, C, Si, P, Mo, Cu and others. To My Entire Family, My Parents, and Leslie 11 ACKNOWLEDGMENTS First and foremost, I wish to thank my wife, Leslie, my Parents, my Brothers, my Grandparents, my Relatives, and my Friends for their love, encouragement, and guidance through the years. I only hope that I can repay even a fraction of their past sacrifice and support in the future. During my stay at Michigan State, a number of people aided me in reaching my goals and many of them have become extremely close personal friends. To this group, I offer my deepest and sincerest thanks, best wishes, and hopes for frequent contact in the years to come. A special note of gratitude is due to my coworker, Sandra, Gene (Hero), Dan (Rock), Roy (Evil), Marty, Dave, Ed, Mike, Jim, Kathy, Rytis and the entire Crouch Group, past and present. This they richly deserve for their advice, both personal and professional, their support, and many, many indescribably good times. My thanks are sincerely extended to Dr. Stanley R. Crouch, my preceptor, and Dr. Andrew Timnick, my conscien- tious Second Reader, for their guidance and friendship. Furthermore, I owe thanks to Dr. Fred Horne and the Depart- ment of Chemistry for my teaching experience (and salary), my teaching awards, and an introduction to the world of management. The Department is also due recognition for the excellent research facilities and technical services 111 provided. More specifically, I thank Ron Haas, Len Eisele, and Jerry DeGroot of the electronics, machine, and glass shops, respectively. A final note of thanks is expressed to Michigan State University for being the great institution that it is, to Dow Chemical for the loan of needed research equipment, and to Peri-Anne Warstler for converting my scrawlings to something presentable. iv TABLE OF CONTENTS Chapter LIST OF TABLES. LIST OF FIGURES . . . . . . . . CHAPTER I - INTRODUCTION. . . CHAPTER II - HISTORICAL A. Introduction. . . . . . . . . B. Original Miniature Spark Design 1. Miniature, Nanosecond Spark Source Design . . . . 2. Spark Solution Techniques 3. Time-Resolution Spectroscopy. A. MNS Characterization and Applications. . . . . C. Modified Miniature, Nanosecond Spark O O C O O O O O O O O O O 1. General System Modifications. 2. Multielement Spectroscopic Measurement Techniques. D. Methods of Sample Injection (Introduction). . . CHAPTER III — SPARK INSTRUMENTATION . A. Introduction. . . . . . . B. MNS Description 1. MNS Housing . . . . . 2. Power Supplies and MNS Circuitry . . . . . . . . C. Injection (Introduction) Systems . . . . . . . . . . Page ix ooooooi—l CD 10 12 1A 1U 17 21 26 26 29 30 33 37 Chapter CHAPTER 2. 3. Page Modified Veillon and Margoshes Nebulizer . . . . . . . . 39 Crossed-Flow Nebulizer. . . . . . . 39 Ultrasonic Nebulizer. . . . . . . . A2 Data Acquisition Circuitry. . . . . . . AA 1. 2. 6. IV. Introduction. . . . . . . . . . . . AA Timing Circuitry. . . . . . . . . . A7 Optical Trigger . . . . . . . . A8 Delay Scaler. . . . . . . . . . 51 Integrate Scaler. . . . . . . . 55 Photomultiplier Gate Circuitry . . . . . . . . . . 58 QOU‘N Analog Electronics. . . . . . . . . 58 Minicomputer and Peripherals. . . . 65 Computer Interface. . . . . . . . . 66 a. Introduction. . . . . . . . . . 66 b. Primary Minicomputer Interface . . . . . . . . . . . 67 c. Monochromator Interface . . . . 72 Perspectives. . . . . . . . . . . . 75 MINIATURE SPARK SOFTWARE . . . . . 78 Introduction. . . . . . . . . . . . . . 78 MNS Software Structure. . . . . . . . . 80 l. 2. Perspectives. . . . . GSPARK.FA and Associated Subroutines . . . . . . . . 81 GSMLEL. FA and Associated Subroutines . . . 85 88 vi Chapter CHAPTER A. B. V. MINIATURE SPARK CHARACTERIZATION. Introduction. Determination of Electrical Properties. . . . . . . . l. Capacitance Determination 2. Breakdown Voltage Studies 3. Current Measurements. Temperature Determinations. l. Excitation Temperature. 2. Ionization Temperature and Electron Density. Introduction Systems. 1. Ultrasonic Nebulizer. 2. Modified Veillon and Margoshes, and Crossed-Flow Nebulizers 3. Nebulizer and Flow System Comparisons . Emission Studies. 1. Background Emission Studies 2. Analyte Signal—to-Noise Studies . . . . . . . . 3. Analytical Findings Multielement Studies. Real Sample Analysis. vii Page 89 89 89 89 90 92 97 99 102 105 105 108 111 113 113 115 119 12“ 126 Chapter ' Page CHAPTER VI. COMMENTARY . . . . . . . . . . . . . 128 A. General Perspectives. . . . . . . . . . . 128 B. Present MNS Capabilities. . . . . . . . . 130 APPENDIX A - Calculation of the Average Resistance and Inductance During Typical Operation of the MNS. . . . . . . . . . . . . 132 APPENDIX B - MNS Software . . . . . . . . . . . . 136 REFERENCES. . . . . . . . . . . . . . . . . . . . 1U7 viii ‘ t . Table 10 LIST OF TABLES MNS IOT Instructions . Current Measurement Results. Discharge Excitation Tempera- tures. Discharge Ionization Tempera- tures and Electron Densities Typical Operating Conditions for all Nebulizers Studied Using Axial and Side Flow Configurations MNS Detection Limits MNS Stability Information. A Comparison of the Present MNS Detection Limits to Those of Other Spectroscopic Techniques A Summary of Multielement Results. . . . . . . . Flame Retardant Analysis Results. . . . . . . . . . . . . ix Page 69 98 101 10“ 106 120 121 122 125 127 Figure mmtw 10 11 12 13 1A 15 16 17 LIST OF FIGURES A diagram of the MNS system . Photograph of the MNS system. The MNS housing . MNS high-voltage circuit. The desolvation system. Modified Veillon and Margoshes nebulizer . . . . Crossed-Flow nebulizer and aerosol chamber . Plasma-Therm ultrasonic nebulizer- crystal housing and aerosol chamber . MNS data acquisition system . Optical trigger circuitry The delay scaler circuit. The integrate sealer circuit. The analog integrate circuit. The clock/initialize/sample-and— hold circuitry. Data acquisition timing diagram . The primary minicomputer inter- face circuit. The monochromator interface circuit (scan and slew control) Page 27 28 31 35 38 A0 A1 A3 A6 “9 53 57 6O 61 62 71 73 Figure Page 18 The monochromator interface circuit (encoder monitor) . . . . . . . . 7A 19 Flow diagram of GSPARK.FA . . . . . . . . 82 20 Flow diagrams of subroutines SET.RA and ADC.RA . . . . . . . . . . . . 83 21 Software structure of the multi— element programs. . . . . . . . . . . . . 86 22 Discharge frequency effects on the spark breakdown voltage . . . . . . . 93 23 A typical voltage-time profile... . . . . 95 2“ A typical current-time profile. . . . . . 96 25 Argon flow rate effects on the observed S/N ratio and the signal- background value (using the ultra- sonic nebulizer). . . . . . . . . . . . . 107 26 Helium flow rate effects on the observed S/N ratio (using the ultrasonic nebulizer) 109 27 Argon tank pressure effects on the observed S/N ratio (using the pneumatic nebulizers) 110 28 Time-resolved spectra of the MNS background emission . 11A 29 Temporal dependence of the S/N . 116 ratio for calcium (ion) xi Figure 30 31 Temporal dependence of the S/N ratio for atomic calcium on the gap atmosphere. Calibration curve for calcium (ion at 393BAzlog-1og plot) . . . . xii Page 118 123 \ a 14- I. INTRODUCTION A spectrochemical Source is a device which converts an analytical sample (gas, solution or solid) into an atomic vapor and provides the energy necessary to excite a certain fraction of the analyte atoms to electronic levels above the ground state. The spectrochemical source thus serves as a sample atomizer, as well as an excitation source. It can be viewed as a transducer whose output is (NA);, the number of density of excited analyte atoms in the Jth electronic energy state. The input to the source is CA, the concentration of analyte-containing molecules in the original sample. The ratio of (NA)§/CA is the transfer function of the source is generally an extremely small value. An inspection of such fundamental principles permit one to envision an "ideal" spectrochemi- cal source. The "ideal" spectrochemical source might be charac- terized in a number of ways. However, some of the more general and desirable characteristics are: 1. good sensitivity to all analytes; 2. no background emission; 3. controllable excitation energy; ability to examine samples in any form; no chemical, spectral or physical interferences; O\U1 accurate and precise results; 7. rapid and low—cost analyses; 8. simple construction and simplicity in use; 9. compactness and portability; and 10. safety for the operator. Through the years spectroscopists have repeatedly attempted to create such a source with varying degrees of success. Unfortunately, although a variety of in— geneous and useful spectrochemical techniques have been developed, no source available today can claim to approach the "ideal" source. This is substantiated through the literature where one finds a-continuing effort to develop and evaluate analytically useful sources. Modern atomic emission techniques include flames, arcs, plasma Jets, sparks, and the high frequency plasmas. Each source exhibits unique good and poor qualities when compared to the "ideal" case. The task of the emission spectroscopist is to select the optimum source for the particular analysis to be performed from the resources available. Intelligent selections are made by recogniz- ing the strengths and weaknesses of each technique. The primary advantages of conventional flames are simplicity, low cost, and good reproducibility; especially for solution analyses of easily excited atoms (alkali metals, alkaline earths, etc.). These qualities account for their extreme popularity as spectrochemical tools. On the other hand, flames suffer from several types of severe interferences which limit their utility in a number of applications. Furthermore, the flame provides a rather low excitation energy, and is limited in the extent of energy variation. To reduce both problems, a multitude of burners, fuels and oxidants have been employed. How- ever, routine users often avoid the inconvenience and danger associated with many of the varieties on the market and simply resort to other techniques when faced with problem analytes. The family of electrical discharge devices classified as arcs reduces or eliminates many of the interferences normally observed in the flame. Due to the continuous, high-temperature, high excitation—energy conditions which exist within the arc, the technique is extremely sensitive to a wide variety of elements. Furthermore, the analyst generally has some control over the excitation energy. Disadvantages include poor precision, bulky, expensive electronics, and problems associated with sample prepara- tion. Solid samples or solution residues provide the most favorable results and are almost exclusively used. The DC plasma Jet is an attempt to meld the high temperature characteristics of the arc and the repro- ducibility of the flame. Although the precision is much better than that of the arc, the detection limits do not reach those attainable by flame emission; nor do flame interferences disappear to the extent predicted. To realize the full potential of the technique, a large concentration of an easily ionizable element must be combined with the analyte solution. As a result, the DC plasma jet has not enjoyed widespread use. The classical high-voltage spark discharge preserves the high temperature and controllable excitation energy of the arc, while affording greater precision in the final results. Large continuum backgrounds, ionization interferences, and radio frequency (RF) noise generation are the major problems associated with the spark. As in the case of the arc, bulky electronics are necessary and most early work involved solid samples or residues. However, more recently, solutions have been analyzed as well. The high frequency plasmas have received a great deal of prominence in recent times. This is especially true for the RF and microwave-induced plasmas. The major reasons stem from two characteristics which are the high temperature and a flame-like nature; i424, those sought for the plasma jet. The high-frequency plasmas exhibit substantial freedom from matrix effects, respectable precision, and good detection limits for a number of ele— ments. The most serious drawbacks are the eXpense, bulk, and problems incurred when attempting to introduce solu- tions into the torch. To avoid extinguishing the plasma regularly, rather complex electronics are necessary. The performance of the introduction system is also ex- tremely critical. These requirements have limited the use of such techniques for routine analyses. The miniature, nanosecond Spark was developed at Michigan State University (1—3) as a viable addition to the sources previously described. The instrument consists of a small coaxial spark source, a Veillon and Margoshes (A) introduction system, a monochromator and photomulti- plier modular combination, a data acquisition system, and a minicomputer. The introduction system produces a dry aerosol which is swept into the spark gap using an argon carrier gas. Emission results as the spark dis- charges between two thoriated tungsten electrodes. The radiation emitted is collected by the monochromator and passed to a photomultiplier tube for conversion to the current domain. The current produced is transferred to a series of transducers which ultimately yield numerical data. Good precision and ease of sample introduction are provided by the discontinuous nature of the plasma. The spark source has been coaxially designed for maximum power, a spark of nanosecond duration, and freedom from significant RF noise. These properties permit the use of compact low-power, electronic supplies to produce the discharge, and virtually unshielded digital circuitry in the vicinity of the spark source. It is therefore pos- sible to employ boxcar integration of the emission photocurrent which drastically improves the analytical results. The spark, in essence, appears arc-like through a delayed examination of the discharge. In a comparison of overall expense and simplicity, the miniature, nano- second spark fares well against arcs, conventional sparks and the high—frequency plasmas. Furthermore, it has been found to be an excellent source for the determina— tion of a variety of elements including those which form refractory oxides. However, continued research on the system by several workers suggested that a number of modifications might be beneficial. It is readily apparent when examining today's numer- ous, complex spectrochemical instruments that the loss of analytical information through the separate components is quite large. Although it is the heart of the system, an "ideal" source is only one of the "ideal" units which are necessary to construct the "ideal" instrument. It seems appropriate to apply the cliche, "a chain is only as strong as its weakest link", to spectrochemical instrumentation. The approach taken in this research follows the "weakest link" philosophy. Each successive unit of the miniature spark was examined and modified in an attempt to produce a more "ideal” instrument. Improvements have been made on the nebulizer, the gas flow system, the spark design, the electronic triggering device, the optical system, the boxcar integrator, and the entire data acquisition system. The software and hardware developed permit the user to control the boxcar integrator, as well as collect, manipulate, store, and print the data from a console. In the chapters of this dissertation which follow, a brief survey of the recent literature relevant to the present work is first presented. In Chapter III, the instrumentation which was developed to improve the minia- ture spark is described in detail, while in Chapter IV, the minicomputer software necessary to control the instrumentation and to acquire and analyze the data is described briefly. Chapter V deals with the characteriza- tion of the improved spark source, both in terms of its physical properties, and its analytical capabilities. The final section of this dissertation is a commentary which examines the present status of the miniature spark and offers the author's views on its future. II. HISTORICAL A. Introduction To understand the use of the relevant literature in the redesign of the original miniature spark, one must first examine the key papers concerned with its concep- tion. After presentation of the major design considera- tions and the results obtained from analyses, it is then possible to discuss system modifications on the pre— viously developed miniature Spark instrumentation in- telligently. The section entitled the "Modified Minia- ture Spark Design" deals with the bases for the instru- mental changes made excluding the introduction systems and the time resolution electronics. The introduction systems are described in a separate section of this Historical, while the electronics are presented in a later chapter. B. Original Miniature Spark Design 1. Miniature, Nanosecond Spark Source Design The miniature, nanosecond spark (MNS), reported by Zynger and Crouch (1,2), was originally designed to be a short duration pulsed light source for fluorescence lifetime measurements. Such sources have been previously reported with flash durations of 20 to 350 ns (5,6), These sources produce a short, high—power, capacitative discharge between two tungsten electrodes. Various atmospheres are employed to produce the radiation required. Zynger developed the MNS to achieve lower pulse widths (7 to 15 ns) and high electrical power dissipation (on the order of 107 w) by building a coaxial capacitor directly into the source. Due to the fact that the MNS provides rather large amounts of energy over very short periods of time and retains the classical spark characteristic of good precision, it was felt that the source should also function well as a spectrochemical tool for the analysis of samples externally introduced. 2. Spark Solution Techniques For a variety of reasons, solutions are extremely popular spectrochemical samples (1). Although arcs and Sparks are generally associated with solid samples, solu- tion methods have also been under development for some time. The earliest methods involved spark—to-bulk tech- niques whereby the Spark passes between one solid elec- trode and the surface of a liquid (7,8). These systems suffer from solution spattering problems. As a result, later researchers developed the spark—to—thin-film techniques which include the rotating disk and porous- Cup designs (9,10). Although these methods show some 10 improvement over Spark—to-bulk techniques, some spatter- ing is still observed, as well as solution uptake problems. Spark-in-spary techniques offer somewhat of a reduction in spattering by introducing the solution as a fine mist which is passed through the spark gap (11,12). However, great care must be taken to avoid contamination of the sparking region. Furthermore, the spark-in-spray method does not show any real improvement in sensitivity due to the inefficiency of particle desolvation by the dis- charge. To improve the sensitivity of a number of spectro- chemical techniques, various researchers have developed desolvation systems to remove the solvent from the analyte prior to introduction into the excitation cell (3,13,1A). Desolvation of the aerosol alleviates many of the problems incurred in the spark-in—spray method, and appears to be an attractive feature for spark solu- tion analyses. The MNS employs such a system to produce a desolvated aerosol which is then introduced directly into the discharge (1,3)- 3. Time-Resolution Spectroscopy Since 1900 it has been known that the continuous background and atmospheric lines present in a spark discharge disappear before the analyte lines become prom- inent (15). Given this fact, it seems apparent that 11 the observation of the total emission of a spark results in a large amount of analytically useless information. In a process known as time slicing, the observation of the emission is delayed for some preset time interval relative to the beginning of the discharge. Thus, early, large background-to-analyte emission ratios and the associated optical noise problems are avoided. As a result, the use of this method increases the sensitivity of the spark to virtually all analytes. Modern spark researchers have taken the timing process one step further br limiting the observation time of the discharge in C addition to providing the initial delay period. This method, known as time resolution, increases the sensi- tivity of the spark even further due to the fact that spectral information is collected only during the period of time yielding the maximum analyte signal-to—noise ratio. A great deal of spark research has been aimed at the construction of time-resolution and time-Slicing instrumentation for both photographic and electronic detectors. Optical shutters, rotating mirror and rotat- ing disk assemblies, and photomultiplier gating circuits have been studied as time-resolution devices (16-19). Recently, the photoelectric detector-electronic gating methods have become popular due to their good precision and to advances in electronic component tech- nology. Walters (20) and Piepen and Schroeder (21) 12 described a modern integrated circuit version of the vacuum tube gated integrator constructed by Steinhaus, 33 al. (19,20). Zynger (1,3) used a circuit similar to those presented previously to time resolve the MNS. The electronics, which were employed, involved a dual gating system whereby both the photomultiplier tube and the solid state integrator become active after some preset delay period. The use of this monostable timing system permitted one to sample the Spark emission over time intervals of approximately 2 to 100 us knifiidelay inter— vals over the same range. A. MNS Characterization and Applications After constructing the MNS and using it as a light source, Zynger (1,3) evaluated the system as a spectro- chemical instrument. Studies indicated that the excita- tion temperature varies between A300 °K and 3800 °K as time (following spark initiation) increases from 2 to 20 us. On the other hand, the ionization temperature ranges from 6600 °K to 5200 °K from 2 to 30 us. The differ- ence in the excitation and ionization temperatures indi- cates that thermodynamic equilibrium is not achieved during the period of observation. As is expected from the tem- poral difference between the background and analyte emis- sion, each particular element exhibits a maximum signal- to-noise ratio at a different time relative to spark l3 initiation. Zynger studied the signal-to-noise versus time profiles for a number of elements and determined the MNS detection limits for these same elements. The MNS was found to have better detection limits than either the flame or the inductively coupled plasma for silicon and phosphorus. The values for boron and carbon were found to be superior to those of flames, while calcium and molybdenum results proved to be better than those obtained using the inductively coupled plasma. Matrix effect studies were also performed and the results indi- cated that the effects of cesium and aluminum on the emission of calcium (neutral and ion) can be reduced or eliminated by properly choosing the observation time. However, Since the effects of the matrix are smallest early in time, the signal-to-noise ratio is decreased and the calcium detection limit increases. In our laboratory, Glass (22,23,71) performed a sizable amount of unpublished work which further charac- terized the MNS. The research included physical design changes to vary the spark capacitance and inductance, carrier gas studies, interelement and anion effects, and real sample analyses. His findings indicated that the MNS is a viable instrument for real sample analysis, although sample introduction problems, memory effects and an increase in background emission somewhat limited its usefulness. 1A Lantz (2A,25) reported the use of the MNS as a gas and liquid chromatographic detector. The system was found to be a successful, element-specific addition to the family of detectors now available. In addition to the ease of selectivity, the MNS provides sufficient information for the user to determine the empirical for- mula of the analyte. C. Modified Miniature, Nanosecond Spark In spite of the well-documented evidence for the fine performance of the MNS as a spectrochemical tool, the work performed by former researchers, as well as early work of the author and co-worker Sandra Koeplin, pointed out several general areas where the method might be im- proved. The suggested improvements are directly aimed at increases in the precision and overall sensitivity of the technique. 1. General System Modifications A number of factors contributed to the overall noise observed in the final analog Signal obtained from the MNS and, thus, to imprecision in the data. Optical noise appeared to be the primary source of noise. This noise is generated as a result of spark "wander", $L3;, the movement of the discharge channel away from the axis 1} 15 joining the electrodes. Some of the factors that influence the spatial stability of the spark include the stability and direction of the gas flow, the electronic circuitry, the geometry and condition of the electrodes, and the composition of the discharge region (1,26). The MNS employs needle-point electrodes to reduce "wander" due to electrode geometry. However, the gas flow is directed perpendicular to the inter—electrode axis and this, com— bined with variations in the flow rate due to the introduc- tion system was thought to be largely responsible for the observed spatial instability. Several workers have found that increased stability is obtained if a thin stream of gas is directed through a hole in one of the electrodes (21,26,27). Walters and Goldstein (28) published a classic article which discusses designing and sampling spatially-stabi- lized spark sources. They indicated that the best flow system consists of a thin laminar jet of gas directed from a small cone surrounding the anode towards the cathode. This would more readily lend itself to use with the MNS introduction system than would gas flow through the elec- trode. However, it must be pointed out that none of the authors were faced with the additional problem of using the stabilizing gas to nebulize solutions and carry par- ticulate matter into the discharge. Walters and Gold- stein also advocated the use of a high-frequency, 5'! l6 unidirectional source to stabilize the discharge further. The original MNS, being a critically-damped discharge,is unidirectional, but is definitely not high frequency. A reduction in optical noise is also possible through partial isolation of the charging circuitry from the analyte gap. This can be accomplished by adding a secon- dary gap which is generally purged by a well-stabilized, uncontaminated gas flow (29-31). The double gap circuit permits one to increase the excitation energy, while ensuring a constant breakdown potential at each particu- lar secondary gap length without disturbing the analytical gap. Rather than use a classical spark gap for control, several researchers have turned to encapsulated systems such as the hydrogen thyratron (30,32). The advantages of these tubes include the ability to initiate the spark reliably from a remote, accurate pulse source and the fact that sensitive adjustments of the gap length are unnecessary. Since imprecision in timing is another major source of noise, the hydrogen thyratron might first appear to be an attractive choice as a secondary gap for the MNS. However, the tubes are somewhat expensive and limited in their electronic capabilities. It should be more advantageous to "free-run" the spark with a direct current power supply and use inexpensive photodetectors to determine the beginning of the discharge combined with precise digital timers to perform time resolution. Electronic noise can be minimized through proper, hard- 5, l7 wired circuitry layout, proper shielding, and expeditious conversion of the signal to the digital domain (33). In this work, spatial stability and the stability of the breakdown potential were improved in the MNS through an axial, laminar flow of the carrier gas and a double- gap system powered by a direct current power supply. The correct choice of a direct current supply has per- mitted the user to increase the spark discharge rate for faster data acquisition. To retain the high power of the discharge, and to promote spatial stability, the dis- charge is somewhat over-damped. The data acquisition electronics included a fast photodetector, an accurate, precise digital timing network, a fast integrator, an analog—to-digital converter, and the electronics neces- sary to interface the system to a minicomputer or a micro— computer. Finally, improvements in multielement capabili- ties and the nebulizer are desirable and are discussed in the following sections. 2. Multielement Spectroscopic Measurement Techniques Multielement analysis techniques have become increas- ingly important and popular as the number and complexity of analytical samples has increased. With the present demand for such techniques, the development of instru- mentation has increased dramatically. Winefordner, g: 31. (3“) divided the multielement detection systems into l8 temporal, spatial, and multiplex devices. Temporal devices are the most common and include the rotating filter systems, the scanning or programmed mono— chromators, and the image dissector photomultiplier. Each of these methods requires some type of sequential wavelength selection, either through the wavelength isola- tion device, as in the case of the monochromators and the rotating filter system or the detector itself, as in the case of the image dissector (35). The rotating filter assemblies eliminate the need for the expense of a mono— chromator, and in addition, offer a larger spectral through- put. However, high background problems and poor resolu- tion have generally limited their usage to atomic fluores- cence. Both monochromator methods reduce background and spectral overlap problems, but they require some form of wavelength control. The scanning technique is limited by the response time of the detector and the measurement time allotted to each individual analysis line (36). The sequentially programmed technique is generally more complex, but is superior in that the spectral areas con- taining no relevant information are quickly passed by slewing through these regions. The image dissector in- volves an electronic scanning system which passes an electron image of the analyte line to an aperature. The image is electronically swept into the aperature and is then amplified using the classical dynode chain. 19 This has been found to be an excellent detector for atomic absorption and atomic emission (37,38). The major draw- back is the cost of the apparatus. The spatial devices, which perform true simultaneous multielement determinations, include polychromators com— bined with photomultiplier tubes and slits (direct read- ers), or with silicon vidicon detectors, or with photo- diode arrays. The direct reader is the oldest and most sensitive of the three, but is expensive and limited to a number of specific lines by space considerations. The photodiode arrays provide the means by which one may examine a complete portion of the spectrum (39). The major drawback is the low sensitivity. Codding (“0) used a linear diode array as a detector for the MNS and found that the detection limits increased by a factor of at least 200 over those observed with a photomultiplier tube. Although the conventional silicon vidicon tube possesses approximately the same sensitivity as the photo- diode array, recent developments in vidicon technology incorporate image intensifiers into the detector which greatly enhance the signal power and make the device much more attractive (35). Multiplex techniques, such as Hadamond transform spectroscopy and Fourier transform spectroscopy, use a single channel detector, yet simultaneously record several spectral components by encoding the information in the 20 frequency domain (3h). Both methods require computer facilities capable of handling the sophisticated soft— ware. In general, for ultraviolet and visible work, the techniques are at a disadvantage when compared to spatial and temporal techniques. Busch and Morrison (35) concluded that at the present state of technology, the programmable monochromator and the intensified vidicon tubes show the most promise as multielement tools. Winefordner, _t _l. (“1) agree to a certain extent, but feel that the choice should be based on fundamental considerations, such as signal-to-noise characteristics and the complexity of the sample. In another earlier work, Winefordner and co-workers (#2) concluded that at the present "state—of—the-art", the vidicons and diode arrays would find only limited use for atomic emission. Since the previous MNS equipment was easily inter- faced to a programmable monochromator system, and since the single channel photomultiplier detector is less expensive and more sensitive than other detectors, it was felt that this method of multielement analysis would prove to be the most advantageous. The design of such a system necessitates the use of a dependable means of wavelength determination (35). Dawson and co-workers (“3) have constructed an instrument which employs an electrooptical device to determine the wavelength ac- curately. More recently, Malmstadt and Cordos (U4,U5) _-'- ’V-~..___. m¢‘._~——_ ,, 21 described a multielement system which employs an encoder to track the monochromator. The encoder is a compact device which fits easily inside a commercial monochromator and produces electronic pulses as the drive chain is moved. Due to the fact that the monochromator described in the Malmstadt paper is the same make and model as that now in use in the MNS, an updated version of the Malmstadt design was found to be a viable method for MNS multielement work. D. Methods of Sample Injection (Introduction) Although sample injection is known to be extremely important in achieving good analytical results, it has not received the amount of attention that sources or techniques have received. Obviously, poor injection will result in low sensitivity and may provide a large source 0 cl unwanted noise. With recent advances in sources, wavelength isolation techniques, background correction techniques, and data acquisition systems, atomic spec- troscopy may rapidly be approaching the point at which the sensitivity of a determination is limited by the process of sample injection. Therefore, it is important to examine the methods of sample introduction for possible incorporation into the MNS system. Zynger (1,3) initially used a Beckman total-consump- tion burner to create a liquid aerosol which was directed a," 22 into a Veillon and Morgoshes (A) desolvation chamber. This nebulizer was relatively inefficient and was re- placed with a modified design. In spite of the improve- ment observed in the emission, pneumatic nebulizers of this type are inefficient, require large operating gas flow rates and pressures, and in the case of the MNS, partially cause optical noise problems (23,A6). For the MNS, other introduction systems may prove more advantag- eous. Electrothermal atomizers, such as the various furnaces, carbon rods, and loops or strips, have been extensively employed in atomic absorption and atomic fluorescence spectroscopy (#8). They offer the advantages of requir- ing small sample volumes and performing pre—analysis ashing and desolvation (47). In the case of the MNS, the atomization step would reduce problems associated with particle loss through the introduction system and remove one energy requirement of the discharge. However, due to the transitory nature of the atomized gas and the spark discharge itself, as well as the large dilution effects of the carrier gas, the inherent sensitivity of these techniques would be low. A crude electrothermal graphite braid atomizer has been studied as an introduction system on the MNS with the predicted results (23). Sample boats, sampling rods and microprobes are gen- erally associated with flame analyses (U8). As is the 23 case with the electrothermal atomizers, small volumes of both solid and liquid samples can be analyzed by placement of the sample container directly into the flame. Unfortunately, the use of these injection systems with the MNS would introduce the imprecision due to sample placement and flame fluctuations, in addition to the problems encountered with the electrothermal atomizers. Chemical vaporization methods have been reported by a number of authors (AB—51). The process is based on volatile compound formation by elements in groups IVA, VA and VIA, as they are exposed to suitable reducing agents. The gases generated are swept into the source by a carrier gas. In the case of the MNS, the technique may suffer from problems similar to those of the electrothermal atom— izers, and furthermore is limited to a few elements. On the other hand, it provides an interesting way to determine a group of elements which are generally difficult to de- termine. To reduce the time-consuming work involved in solution preparation of certain difficult samples (geological, metal, 233.), several researchers have developed rather unique methods for solid sample introduction. Gilbert (52) and Willis (53) suspended finely ground samples in solvent mixtures for aspiration into flames. With finely— ground samples, other workers utilized vibrations from pulsating pistons or ultrasonic generators to create a .' u?" .l‘. i. i 1 2U cloud of material which is carried into the source (U8). Jones, gt al. (5A) and Fassel, _t_al. (55) reported a rapidly moving arc sampler for use with solid metal sam- ples. With the exception of the arc sampler, the solid samplers would be difficult to interface to the MNS. A large portion of everyday solution analysis is performed with nebulizers which generate an aerosol from the sample. In addition to that previously explained as a part of the original MNS, two other nebulizers have recently gained prominence. The crossed—flow nebulizer described by Davies, §£.§l° (56) consists of two adjust- able channels placed perpendicular to one another. As gas is fed through the venturi and over the perpendicular capillary, liquid is drawn into the gas stream and is converted to a fine mist. The design permits easy inter— change of various sizes of capillaries and venturis. Valente and Schrenk (57) modified the design by using stainless steel hypodermic syringes for gas and liquid flow. Fassel and co-workers (58) simplified the crossed— flow system and introduced glass capillaries in place of the hypodermic syringes. They found that the flow rate of gas necessary to operate the nebulizer efficiently was as low as 0.8 i/min with a pressure of 25 psi. Re- cently, Donahue and Carter (59) have further simplified the use of the nebulizer by constructing the Fassel design so that the capillaries are easily adjustable. 25 In contrast to the pneumatic nebulizer which is based on gas pressure, ultrasonic nebulizers generate liquid aerosols through ultrasonics. Modern ultrasonic systems employ radiofrequency generators in the range of 20 KHz to 6 MHz to supply a piezoelectric transducer with suf- ficient energy to create an aerosol from liquids placed on the surface (A8). Although ultrasonic nebulizers are much more complex than are pneumatic nebulizers, they enjoy such advantages as producing smaller, more uniform droplets of controllable size, and independence of aerosol generation from the pressure and flow rate of the carrier gas (EU). A number of ultrasonic nebulizer designs have been reported in the literature (60—66). In addition, several researchers have studied the performance of such systems (67-70). Comparisons of ultrasonic and pneumatic nebulizers have shown the former to be superior in all cases. Recently, Fassel, _t _l. (A6) developed a well— designed ultrasonic system for use with an inductively- coupled plasma. This system is now available commercially as a product of Plasma—Therm, Inc. Due to their excellent performance and the ease of interfacing the crossed-flow and ultrasonic nebulizers to the MNS, these nebulizers have been examined as pos— sible alternatives to the original MNS nebulizer. III. SPARK INSTRUMENTATION A. Introduction The general, overall configuration of the present MNS system is outlined in Figure 1, while Figure 2 is a photo- graph of the instrument. The analyte solution is drawn up to one of three possible nebulizers either through pres- sure differentials within the nebulizer or by an external peristaltic pump. The nebulizer produces a fine mist which is directed into a heated chamber similar to that of Veillon and Margoshes (A). There desolvation occurs and the result- ing mixture is passed through a modified Friedrichs con- denser to remove the solvent vapor. The dry salt particles are swept through the analyte cone in the MNS housing and along the inter—electrode axis in a laminar flow of argon or helium. The secondary gap is continuously flushed with tank argon. A high voltage power supply charges the coaxial capacitor located at the top of the spark housing to the breakdown potential of the system. Spark discharge radiation is monitored by a photodiode-Darlington trigger in the secondary gap and a spectrometer in the analyte gap. The trigger pulse signifies the beginning of the dis- charge to the gated integrator (or sample—and-hold) cir- cuitry which is employed to time resolve the emission from the MNS. In the analytical gap, the spark atomizes the 26 rduima Pmo KNPDQZOU—ZZ MUS—awFZ. .m mOPdmomkz _ cm .20 2 $3085 , w .4 nwf on. mo Empucwp < p oh "L 4‘4 1 r . 1...»- ybrmlr.:L .Lu.r._ — fl com... . 596 I 55285me a; zmkmrm 1' h comb J 29.2” ommo mmNEDmmz mummy—Eb woo—OOPOIQ hamDm ”Egon woSUO> IQI 28 .Eopmzm m2: opp no nowLwOposm .m mudwfim 29 salt particles and excites the atomic cloud produced. Emission from the discharge is collimated to fill the en- trance slit of a programmable monochromator and passed to a gated photomultiplier tube. The signal obtained by integrating or sampling the photocurrent is available for direct output use by a recorder and oscilloscope, or as an input to an analog-to—digital converter. In the latter case, the digitized data are accessible to a mini- computer. The minicomputer stores the data, performs simple statistical calculations, and controls the mono- chromator and the time-resolution electronics. Manual control of the timing circuitry is also provided. A detailed description of the MNS design is presented in the following sections. Separate discussions are apportioned to the introduction system, the spark housing and elec— tronics, the data acquisition circuitry, and the mini- computer. Optimum Operating oonditions and physical MNS determinations are primarily left to Chapter V. B. MNS Description The present MNS source design was developed to increase the spatial stability of the discharge, and to reduce background emission fluctuations by decreasing the in- fluence of the analytical gap on the breakdown potential (28,31). An explanation of the MNS source can most 3O conveniently be treated by dividing the system into two sections. The physical description of the housing is first discussed followed by the power supply and the spark circuitry. l. MNS Housing Preliminary work on the design of a simple prototype was performed by co-worker, S. Koeplin and myself. The present design, shown in Figure 3, was independently developed by S. Koeplin (72). The housing can easily be separated into four main sections. They are from top to bottom, the capacitor housing, the secondary gap housing, and the analytical gap housing, which consists of two pieces. The capacitor housing is similar to that developed by Zynger (I). It is constructed from two concentric copper tubes connected by a brass plate on one end. The inner tube runs the length of the housing and provides the ground connection to the bottom electrode of the analytical gap. A solid aluminum cylinder covered by a 0.02 in polyethylene sheath (Cadillac Plastics) fits inside the inner copper tube to form the coaxial capacitor. A high-voltage connector screws into the top of the alumin- um cylinder to provide the electronic connection to the power supply. The most recent design is 3-3/A in. longer than that shown in the diagram and features a removable capacitor for simplified maintenance and for quick 31 Figu re 3 . The MNS h ousin g. 32 interchange of power supplies. The brass secondary gap housing supports a teflon chamber with quartz windows and is similar in design to that reported by Lantz (2A). Argon, monitored by a Gilmont flow meter, continuously passes through the secon- dary gap region. The discharge is formed between two 2% thoriated tungsten electrodes (Union Carbide, Linde Div.). A light pipe (Edmund Scientific), used to conduct emission from the secondary gap to an optical trigger system, passes through a teflon support in the brass housing, and butts up against one of the quartz windows. The brass analytical gap housing supports a rather complex teflon gas flow system. The analyte stream from the heated chamber passes into this system via one of the male Swagelock fittings on the side of the unit. The analyte stream is guided into the cone surrounding the upper electrode and along the inter-electrode axis (note bottom view of flow system). A port on the opposite side of the unit is available for a sheath gas. If employed, the sheath gas flow parallels the inter-electrode axis. The length of the analytical gap is easily adjusted over a 1.5 cm range by twisting the threaded brass rod which extends from the bottom of the housing. A quartz lens of 1.0 in focal length is set 1.0 in from the center of the discharge to collimate the radiation produced. The vertical height of the lens is adjustable and it is 33 aligned with the gap following gap length changes. The lengths of both gaps are accurately determined using a reticle (Edmund Scientific) and an optical coupler develop- ed by S. Koeplin (72). 2. Power Supplies and MNS Circuitry To reduce the overvolting problems found in the original MNS, a direct current (DC) power supply (Spell- man, UHRlOPlOO, 10 mA at 10 kV) is employed in the present system. To current limit the supply during the discharge, a 1.0 Mg, 100 W carbon film resistor (American Component Co.) has been introduced into the charging circuit. For safety and RF noise reduction, the resistor is totally enclosed in a plastic container, which in turn is sur- rounded by a Faraday cage. Current is delivered to the resistor and ultimately to the MNS coaxial capacitor via coaxial cable (Beldon, RG 8/u) and Amphenol connectors (82-8U3, 82-320). The DC supply has been found to produce sparks at repetition rate more than an order of magnitude higher than that found to be possible using the original MNS pulsed supply. It is also possible to use the original pulsed supply (Xenon Corp., A73A Nanopulser) by simply changing the coaxial capacitor system to that compatible with the supply. The pulsed supply needs no current limiting resistor and uses RG 58 c/u (Beldon) coaxial cable with appropriate Amphenol connectors (29100, 27025). 3A A simplified schematic diagram of the MNS is present- ed in Figure A. To obtain the power necessary for high atomization efficiency and excitation energy, and yet use a small-sized power supply, as well as to simplify the time resolution process, the spark discharge should be nearly critically damped. Equation 1 indicates the rela- tionship between the equivalent circuit resistance, induc- tance, and capacitance for such a discharge (76). R 2/L/c (1) For a more complete understanding of the present MNS dis- charge, a semiquantitative solution to this equation is useful. The calculation of MNS values for R and L are presented in Appendix A. They are 210 Q and 0.3 pH, respectively for a typical gap combination (secondary = 3.0 mm, analytical - 3.0 mm) in an argon atmosphere. The capacitance of the system has been found to be 350 pF (Chapter V, Section Bl.). Using the previously calculated quantities, the right- hand side of Equation 1 becomes 60 9. Since R was de- termined to be 210 Q, the present MNS discharge appears to be somewhat overdamped. However, these semiquantita- tive calculations assume that the gap atmosphere is pure argon, while in reality, the analytical gap contains water vapor and some type of analyte(s) or analyte(s) and matrix 35 .ufiSULHo wwwufio>lnwfin mzz .: omsmfim l .._ m l..l... m .o .m lid 9 ....... 583m # 0 III! mmson. ._ a .Hulu $459 00 o .J l 10.: p m GS: 36 which definitely change, and probably lower R and L (78,79). Therefore, from the calculations presented, it is dif— ficult to exactly determine the extent of overdamping quantitatively. The true character of the discharge lies somewhere between the overdamped discharge described, and a critically damped discharge. Obviously, the use of other gases in the gap will greatly affect the discharge character by changing R and L. The bridging resistor, RB (10 M0 or greater), is often employed in the MNS circuit. It decreases the influence of the analytical gap on the breakdown potential of the system (31). With this resistor connected, the bottom electrode of the secondary gap becomes a true ground. Gap breakdown occurs when the voltage on the capacitor reaches that amount necessary to initiate the discharge in this gap alone. Upon discharge formation, current passing through the resistor increases the voltage across the analytical gap quickly and, finally, this gap begins to conduct. During the period of time that the analytical gap is conducting, RB presents a resistance which is too large to affect the discharge appreciably. Bridging resistors of this type are generally employed in classi- cal double-gap sparks to stabilize the breakdown potential and, thus, the total spark energy. 37 C. Injection (Introduction) Systems The modified Veillon and Margoshes (A) pneumatic nebulizer performed well in use with the original MNS system. On the other hand, coaxial-flow nebulizers are, as is the case with most pneumatic nebulizers, somewhat inefficient (A6). Furthermore, the large operating pres- sure and flow rate become problems when interfacing this nebulizer to the present MNS flow system. Walters (73) stated that for flow systems similar to that employed in the present MNS, stability decreased for flow rates above 0.6 A/min (a rather low flow rate to realize with any nebulizer). However, the crossed-flow and ultrasonic nebulizers described in this section perform efficiently at flow rates much nearer those required for good sta- bility. A desolvation system developed by Veillon and Mar- goshes (A) is employed with all three nebulizers to de- solvate the sample. The apparatus, shown in Figure 5, is constructed from a cylindrical, glass,heated chamber and a modified Friedrichs condenser. The heat is supplied by heating tape (Biskeat, 768W, Brisco Mfg. Co.) which encompasses the chamber. At a Powerstat (Superior Electric Co.) setting of 55 V, under nebulizing conditions, the inside temperature of the chamber reaches approximately 160 °C. In the water-cooled condenser, 98% of the solvent (water) is removed (1). The dry aerosol is carried to 38 .Empmmw coapw>aomoo one .m opsmfim Ammkqsmmoza 2.55 $5.53 + 54m R #5 93 .11 D $.53 » u monomzzoo n L Basam 28m: m 9.333: m 83.00% m mumZmozoo Eon: n AmIomomEu 1 E9245 8.31 .vouI All EoON v Fm? Ow... 39 the spark housing through large bore tygon tubing. A tapped teflon connector couples the tubing to the male Swagelock fitting on the MNS housing. 1. Modified Veillon and Margoshes Nebulizer For all MNS studies, Zynger (l) employed the modified Veillon and Margoshes (A) nebulizer as diagrammed in Figure 6. A teflon-coated stainless steel ball-joint fits over the nebulizer and serves to seal it to the heated chamber. Solution is directly aspirated into the heated chamber. For original MNS use, normal operating conditions were a tank pressure of 50 psi (3.5 kg/cm2), an argon flow rate of 3.8 z/min, and a solution uptake rate of 2.5 mQ/min. 2. Crossed—Flow Nebulizer Recently a crossed—flow pneumatic nebulizer has been reported by Fassel, et a1. (58). A modified version of that nebulizer,which was used in this work, is shown in Figure 7a. Argon is directed through the horizontal capillary (0.23 mm tip orifice) at flow rates of 0.8 z/min or greater (9a. 25 psi). Solution is drawn through the vertical capillary (5 pi pipet, 0.28 mm 1.0.) and nebulized into the aerosol chamber diagrammed in part b of Figure 7. Although, the solution capillary used is 140 CARRIER GAS INLET * ‘2 SOLUTION { INLET | CI ’32 Figure 6. Modified Veillon and Margoshes nebulizer. We j- \ H—SLEEVE k . s (1) Lugs: 'LLARY SWAGELOK INLET -I- TO HEATED 1- CHAMBER . N . hagglLlARY DRAIN—> INLET (b) Figure 7. Crossed-flow nebulizer and aerosol chamber. N2 convenient and simple to adjust, the solution uptake rate is low (the original design employed a 0.25 mm orifice diameter). To Obtain good results in the present design, it is better to pump the solution through the capillary. A peristaltic pump (Ismatec) Operating at 2.3 mA/min is used to deliver the solution to the nebulizer. The capillaries are secured by compressing two 0- rings which are seated between a brass insert and a teflon insert—Swagelock combination. Pressure is exerted on the teflon insert by tightening the external female Swagelock fitting. Optimum nebulization is Observed to occur when the capillary to capillary distance is approximately 0.1 to 0.15 mm. The aerosol chamber (Figure 7b) design has been des- cribed by Fassel and co—workers (A6) for use with an ultra- sonic nebulization system they developed. The design minimizes the number of large droplets which reach the heated chamber. The vaporization of such droplets cause gas flow pulsations which can lead to source instability. The aerosol chamber is easily interfaced to the heated chamber via an 18/9 male-tO-male elbow. 3, Ultrasonic Nebulizer A Plasma-Therm ultrasonic nebulizer (Model UNSl) has also been examined as a nebulizer for the present MNS. The design, presented in Figure 8, is extremely similar “3 .meEmco HOmomow 0cm moamzon prmmmoIpmNfiadnm: OHCOmmLpH: EamanwEmem .w mmswfim Ir— _I _I 56%. I_ 2.4% i 2. m3 mmmmqo/ C 2. Susan AWW+ HI n/I\ z. I fix... \ I “3&3 0222 , 1 > EEhm 2. am IIT 4 fl . &.+Ladfim$hfiv m 9m. 50 o Ilv \ LF oo EEov mum—245 ow...de O... an to that described by Fassel, _t_al. (A6). The piezo— electric crystal is seated in a cylindrical teflon housing which permits a constant flow Of cooling water to play over the back face. Electrical connections to the crystal are made through this chamber. Power is supplied by a 50w radiofrequency generator of variable frequency (l350-1A00 kHz, see Figure 2). The analyte solution is pumped to the crystal through tygon tubing, followed by 1.0 mm (i.d.) polyethylene tubing, by a peristaltic pump (Ismatec). The delivery tube is positioned extremely close to, but not touching, the upper-center portion of the crystal face. The carrier gas is introduced through a side-arm in the drain tube to avoid flow obstruction by the condensed aerosol. Normal Operating conditions include an incident power of 15W, 1.0 to 1.25 Q/min argon flow rate, and a solution delivery rate Of 2.3 mi/min. However, unlike the nebulizers previously discussed, the ultrasonic nebulizer functions acceptably at argon flow rates as low as 0.3 z/min or with helium as the carrier gas. D. Data Acquisition Circuitry 1. Introduction The MNS data acquisition circuitry consists of a fast trigger, gated integrateaand-hold electronics (or sample- and-hold), a computer interface, and a monochromator “5 interface. The heart of the system is housed in a modi- fied rack mount (Varipak, Elco Corp.) as six printed cir- cuit boards which plug into a backplane bus. The six mainframe boards are the delay scalar, the integrate scaler, the analog integrate, the clock/initialize/sample— and-hold, the primary computer interface, and the mono— chromator interface. The Optical trigger, photomultiplier gate circuitry, and a portion of the monochromator elec- tronics are external to the mainframe. Power is supplied by a commercial power supply (Power-One, Inc., HCBB-75W), which provides +5 and :15 volts. Figure 9 is a functional outline of the data acquisi- tion system. As the spark discharges, emitted background radiation from the secondary gap begins the timing sequence when it strikes the Optical trigger. A pulse from the trigger circuitry enters the delay scaler and gates on a 20 MHz clock. The delay time required can be varied in 0.1 us increments and is set either manually using rotary switches (0.1 to 99.9 us), or via the minicomputer (0.1 to A09.6 us). Upon the completion of the delay period, the photomultiplier gate returns all dynodes Of the photo- multiplier (PM) tube to their full operating tube voltages while either the sample-and-hold (S/H) or the gated inte- grate-and-hold (GI/H) begin to receive the photocurrent. If the S/H is selected on the front panel, the end-of- delay pulse initiates the sample function, whereas if the .Empmmm coaufiwfisvom mumo m2: .m unawam muSazoon >385 mmomoomm Bomuwflww So can I Eu szoE A an __ - a T! «2.11524 r $.28 «Scam SEES. I oo< 4|, o 3458.; 4 >38 _ r r ".9285. H L H A «02.368 mofizomxoozos 3:092 4596 $.er #53 3.552208% ~12 ON .2230 2 A7 GI/H is selected, integration begins under the super- vision of the integrate scaler. User control of the integrate scaler is equivalent to that of the delay scaler. Amplification of the integrated signal can be manually adjusted from a gain Of l to 100 in stages. Both the S/H and the GI/H pass their analog signals to the analog—to—digital converter (ADC). The GI/H signal is also available from a remote monitoring connector. The digital output of the ADC is accepted by the minicomputer through the interface. Spectral scans or fast multiele- ment data collection is also possible under computer con- trol via the monochromator interface. To understand the capabilities and limitations Of the data acquisition system more fully, a detailed examina- tion of the circuitry is warranted. The remainder of this chapter is devoted to such an examination. 2. Timing Circuitry TO accurately characterize the discharge, Observe early spark emission, and achieve optimum analytical results, a fast trigger and fast photocurrent measurement is neces- sary. The Optical trigger, delay scaler, integrate scaler, PM gate, I/H and S/H, combine to form a fast data collec- tion network capable of examining the MNS. Since timing is one of the central functions of the electronics and is extremely critical to the quality Of the analytical A8 results Obtained, the optical trigger, delay scaler, in- tegrate scaler, clock, and PM gate are presented first. The entire system was constructed within the sequencer framework described by Crouch, 32 a1. (7“). a. Optical Trigger Studies of the MNS early in time require that the timing circuitry consistently detect the spark immediately after the onset of the emission. The original MNS relied on high-frequency noise to initiate timing. This was somewhat unreliable due to inconsistencies in the actual time of formation of the discharge. Glass (23) used a phototransistor (Til6A, Texas Instruments) to determine the time of formation Optically in the manner shown in Figure 10a. Although the phototransistor is a step in the right direction, it is moderately slow (1.5 Us rise time) when compared to the sub—microsecond time scale in which theremaindercm‘the MNS electronics Operates. Of the remaining optical triggers available, it was found that only the photodiode was sufficiently fast for possible application (1.0 ns rise time). These devices have been employed in laser research for some time. Un- fortunately, they possess low spectral sensitivities and in stand-alone usage with the MNS, initial studies indicated that the photodiode could not supply the necessary current and voltage for direct interface to digital circuitry. A9 +5v *5" O O h” I.5K.Q (G) 5“ A MRD ooo IOKQ h r ' +5v O IOK 4.7!“? To DELAY SCALER 2N2369 +5v (b) IOKQ 1 TO DELAY SCALER hu % TIL 64 Figure 10. Optical trigger circuitry. x} 50 To increase the photodiode signal strengths to workable levels, fast video amplifiers are often combined with photo- diodes. These combination units are commercially avail— able, but are somewhat expensive and are easily ruined by high—frequency noise such as that created by the MNS. All attempts to construct a combination from discrete parts and use it as an optical trigger failed in spite of heavy shielding and other noise-reducing precautions. The combination type Optical trigger was abandoned in favor of the photodiode-Darlington circuit shown in Fig- ure 10b. Emission from the secondary gap of the spark is transmitted to the photodiode via an optical light pipe. The A700 0 resistor to ground was chosen as a compromise between the RC time constant of the circuit and the vol— tage developed at the base of the first amplifying tran- sistor. The use of greater resistance results in large rise times due to the ApF capacitance associated with the photodiode, while smaller resistances result in vol- tages below the transistor threshold. In practice, the photodiode-Darlington circuit produces negative-going Spikes with fall times of less than 75 ns duration and rise times of 200 ns. It has been found to perform well with either argon or helium in the analyte gap and for all gap distances examined. However, it fails when helium is used in the secondary gap due to the low background emitted by the discharge. In this case, the previously 51 presented phototransistor circuit must be employed. b. Delay Scaler The delay scaler (Figure 11) and the clock (Figure 1A) combine to form the timing circuitry necessary to delay integration or sampling of the photocurrent following spark initiation. Figure 15 summarizes the timing func- tions of the MNS electronics. The magnitude of the delay time can either be set manually using front panel rotary switches or via the minicomputer under interactive user control. To ensure that the delay scaler is capable of handling the high-frequency clock and that the propagation delay through the timing chips is within reasonable limits on the 0.1 us time scale, the timing circuitry has been constructed from Schottky TTL integrated circuits. The timing sequence begins as the external trigger signal from the optical trigger signal sets an RS flip- flop (7ASOO, gates 3 and A). This, in turn, gates on the external 20 MHz clock via JK flip-flop l (7ASllZ-l). The output of this flip-flop supplies the cascaded counters (7A3161, 1-3) with 10 MHz clock pulses. The magnitude of the delay selected by the user is contained in three 7Al6l binary counters. As the 7ASl6l counters are clocked, the outputs Of both sets Of binary counters (7ASl6l and 7A16l, 1-3) are continuously compared by three 7A885 magnitude comparators. Upon completion of the delay 52 Figure 11. The delay scaler circuit. 53 .HH mhzmfim mu~< m»:< mu.< m>r< r 1 o NIN—umth uu< Hz— 4L 4L ham 41 oum< A: Dum< o v. 5: a: a: oum< o h mau _ _ 2 m [fizz . :6: h - . names _ _ o- _ _ us<¢ompz. j up; .2. 3. C u. .3 WV: n. _ _ I/\/\/\(n+ .>z. m. a» u. nu.u ~<«d. u: .< n. s..a.ln..u¢...2§ I_I._I a. , goo n. 61 .mppfisoafio paonlochOHQEmm\ONfiHmfiuficfi\xOOHo O£B .:H mmswfim ”O .--ua.o. I NI: ow .hr II » .IHH can «smo.x <4¢hx wmssa uu rllllln. u~.4<~»~z_ no xoonu wNH4<4uo comes OJOI I ~1 usez z 0 3 O Q :5 u < E m 2.0 1» 1.5 e a c c : e : e L : c 0.0 1.0 2.0 3.0 “.0 5.0 Frequency (kHz) Figure 22. Discharge frequency effects on the spark breakdown voltage. 9A with a fast sampling module and a camera. Voltage-time profiles were recorded on film, enlarged, and digitized. An example of such a curve is presented in Figure 23. The current-time profiles were generated by S. Koeplin (72) using a Runge-Kutta (83,8A) solution to the first order differential equation which relates the voltage-time profile to the current-time profile. This relationship is expressed by the following equation: V(t) = L(dI(t)/dt) + I(t)R + I(t)t/C, (2) where I is the spark current and L, R, and C are the nichrome wire inductance, resistance, and capacitance, respectively. The capacitance was estimated to be in the pico-Farad range (85). The resistance was accurately Obtained by measuring the voltage drop across the wire from a standard current. The value Obtained was 0.0792 9. The inductance is more difficult to determine and is important because it affects the results rather dramati- cally. Indications are that the value lies somewhere between 1 nH and 10 nH (86,87). A value in the center of these extremes was chosen due to the fact that the current profiles Obtained were consistent with the cir- cuit calculations presented in Appendix A. Figure 2A shows a typical current-time profile. Although the profiles Obtained are probably close to the "true" 95 0.0207— ""' u t t c t ‘- .__ t * A i ‘ (D 5.. 1: ._-| t O .— > 1- V .- 1; DJ 0 < " 1- |- “T: _l a O > t l —— I! t a: "u 1- .9". __ 1- “t ‘t.tt'*. -0.015 I llllL_l Llll T l I I l I l l l T I 0.000 TIME (NANOSECONDS) x10 ' Figure 23. A typical voltage-time profile. 96 2.400" .1» . o ¢ ¢ 1 l —-- I t ‘ o t z; __. c u: o C: :5 - " I". up (:I v “F“ 4. I— E: ‘ ‘ .2. x * a: t m: e :3 -L ', L’ o I . c I ‘A # ...“ ‘ ‘5‘“...‘it.’ I O . Li J J l I l l_4l l l aegeflflllllITlIl++i| 0.000 7.500 TIME (NANOSECONDS) ~ 1 X].0 Figure 2A. A typical current-time profile. 97 profiles, it is best to use the results presented for relative comparisons only. Table 2 summarizes the results of the studies. The same dependence on the frequency (DC supply voltage) found in the breakdown voltage studies are suggested by the results of the helium trails. Furthermore, it is evident that helium produces lower currents than argon under similar conditions. The data also demonstrate gap length effects and gap atmosphere effects. It appears that water vapor increases the cur- rent, as has been found to be the case with sparks in air (78). It also appears that Glass (23) was correct in his studies involving current observations in high analyte concentration solutions. He stated that the water vapor acted as a buffer which prevented large cur- rent changes with analyte introduction. Finally, although current profiles were not determined for the original MNS, a similar one-gap system using the pulsed supply yielded voltage-time profiles which appeared to be oscillatory and which had an initial peak 9a. three times that found for the largest peak presented in Table 2. C. Temperature Determinations The plasma temperature(s) bear a more direct rela— tionship to the observable physical properties of the spark than the electrical parameters previously discussed. 98 Table 2. Current Measurement Results. Gap* Time of DC Length— Maximum Maximum Supply Sec.,Anal. Current Current Voltage (mm) Atmosphere (amps) (ns) kV 3.“,2.0 He 17.8 26.5 D.O 3.h,2.0 He 23.7 22.9 3.0 3.0,2.0 Ar “0.1 30.1 U.O 3.0,3.0 Ar 52.5 26.5 D.O M.O,2.0 Ar 57.0 32.5 “.0 3.0,2.0 Ar “6.9 24.1 U.3 3.0,2 0 Ar, H2O vapor 55.0 26.5 “.3 3.0,2.0 Ar, H20 vapor 56.5 26.5 U.3 500 ppm Ca * No bridging resistor was employed in any trials reported. 99 In fact, the temperature is the most vital parameter governing the characteristics of a plasma (76). Four possible temperatures may be used to describe a monatomic gas plasma (143;, the electron, gas, excitation, and the ionization temperatures). In the sections which follow, the excitation and ionization temperatures are examined for the present MNS. l. Excitation Temperature The excitation temperature describes the population of various energy levels in the gaseous system. For a Maxwell-Boltzmann distribution of atoms at some tempera- ture, T, the intensity, I, of a spectral line produced by transitions from energy level E2 to level E1 is given by: Cg2A2+ I = _-—T—_l exp (-E2/kT), (3) where C is a constant, g2 is the statistical weight of state 2, A2+l is the transition probability (Einstein Coefficient), A is the wavelength of emission, and k is the Boltzmann constant (1,76,78). For determination of the excitation temperature, Equation 3 can be expressed in a more convenient form. E IA 2 ' ‘ 2.303kT (u) 100 A plot of the left hand side of this equation versus E2 should produce a straight line with a slope that is pro- portional to the excitation temperature. This method is applicable with the constraints that self—absorption is negligible, accurate transition probabilities are avail- able, and the spectrometer exhibits a uniform spectral response over the wavelength region used (88). The first requirement is true for argon plasmas below 65003, and the latter two are true if the correct series of argon lines are chosen. The slopes of the lines fitted to the data were cal- culated by a least squares routine. Transition probabili- ties for the spectral lines employed were reported by Adcock and Plumtree (89). Table 3 lists the results of the determinations, the results found for the original MNS, and the spectral lines employed. Temperatures were determined for several gap lengths with and without the bridging resistor, and over various time intervals. It is apparent from the values obtained, that the bridging resistor does at least partially remove the analytical gap from the circuit during the charging period. A comparison of the data with those obtained by Zynger (l) and Lantz (2U) supports the previous findings of a lower discharge energy for the present system. Furthermore, the temperature appears to decrease more quickly than was observed by Zynger and Lantz. It is therefore expected lOl Table 3. Discharge Excitation Temperatures. Gap Orig. Length- Time Spark Sec.,Anal. Period Temperature Results (mm) (us) (°K) (°K) 3.0,3.0 0.5-2.0 M8001UOO --—— 1.0—3.0 HMOOtuOO ——-- 2.0-5.0 uOOOinO M300 5.0-l0.0 3600:400 H700 10.0—15.0 3000iUOO “550 15.0-20.0 -------- 3800 3.5.3.0a 0.5-2.0 u500:u00 2.0-5.0 3700:“00 5.0—l0.0 3000:UOO 14.5.3.0a 0.5-2.0 M6001u00 2.0—5.0 3800iu00 5.0-l0.0 3300:“00 aBridging resistor employed. b Single gap, pulsed supply. 102 that the time window of maximum signal-to-noise for each element will occur earlier in time and for a shorter period of time than was found by previous workers. Zynger (1) pointed out that the MNS resembles the micro- wave plasma in argon reported by Taylor 33 al. (90), as far as the excitation temperature is concerned. 2. Ionization Temperature and Electron Density The ionization temperature describes ionization equilibria and is mathematically expressed by the Saha equation (76). The following equation, derived from the Saha relationship, relates the intensity ratio of an ion-atom line pair of one element to the ionization temperature. The equation states that: g A A log (I+/I) = - log Fe + log 2+1, gA2+lx SOHO + _ _ where "+" designates the ions, Vij is the apparent ioniza— tion potential, which contains the partition functions, Vq is the energy level from which emission occurs, and Fe is the electron pressure (76,91). The electron pres- sure is defined as: N T P = e , (6) e 7.3ux1021 103 where Ne is the electron density (76). In turn, the electron density can be obtained from the half width of the M8613 HB line which is subject to a linear Stark effect (92). The relationship is: N8 = 0(Ne,T)m3/2 . (7) The coefficient, C, is slightly dependent on the electron density and the electron temperature, and therefore, the final value employed is an average over electron densities previously reported for the MNS and over an electron tem- perature range of 10,000 to 30,000 °K. The half widths, AA, were obtained by scanning slowly over the HB line and monitoring the emission intensity on a strip chart re- corder. Ne values, determined under various conditions, are presented in Table H. With the electron density and the intensities of the Ca ion (3933i) and Ca neutral (M227K) lines, the ioniza— tion temperatures were determined through an iterative technique. The results of calculations are presented in Table A, together with other pertinent information. As is the case with the original MNS, the ionization tempera- tures are larger than the excitation temperatures de- termined under the same conditions. However, the tempera- ture differences observed in the present system are much smaller than those found previously, which suggests a 10% Table M. Discharge Ionization Temperatures and Electron Densities. Gap Orig. Sparkb Length- Time Ne Ioniz. Ne Sec.,Ana1 Period 3 Temp. 3 Temp. (mm) (0880) (e/cm ) (°K) (e/cm ) (°K) 3.0,3.0 0.5-2.0 1.11x1016 55001600 .6x1016 --—— 1.0-3.0 9.ux1015 5U00t600 --__ --_- 2.0—5.0 6.7x1015 5200:600 .ux10l6 —--- a 16 3.5,3.0 0.5-2.0 1.3xlO 51001600 —--— _-__ 1.0-3.0 8.5x1015 ........ ---_ _--- 14.5.3.0a 0.5-2.0 1.14x10l6 5200:600 -—-— -_-- 1.0-3.0 9.1x1015 ———————— ---- _-__ 2.0-5.0 5.7x1015 ........ ---- _-_- 2-15 6600 15-30 5150 5-10 6.1x1016 ---— aBridging resistor employed. bSingle gap, pulsed supply. 105 plasma that is somewhat closer to a state of equilibrium during the time intervals chosen. The large difference between present and past findings is a further indication that the present discharge is less energetic. D. Introduction Systems Analytical findings using the three nebulizers pre- viously discussed are sufficiently unique to the present flow systems and discharge to warrant a detailed explana- tion. Each nebulizer will be briefly discussed followed by a general comparison of the three. In addition to the axial aerosol injection system described in Chapter III, side introduction, as was employed in the original MNS system, was also used. Normal operating conditions for the nebulizers with both flow configurations are presented in Table 5. l. Ultrasonic Nebulizer As was expected, of the three nebulizers studied, the ultrasonic nebulizer produced the best results. The detection limits found while using each nebulizer are presented later in this chapter. For axial gas flow, signal—to—noise (S/N) ratio studies at various argon flow rates indicated that a maximum occurred in the curve at ca. 1.25 fi/min, as is shown in Figure 25. 106 Table 5. Typical Operating Conditions for all Nebulizers Studied Using Axial and Side Flow Configura- tions. Flow Flow Rate Nebulizer Configuration or Tank Pressure Ultrasonica Side 2.0 l/min " Axial 1.25 l/min Crossed-Flow Side 35 psib " Axial 30 psi Modified V and M Side 50 psi " Axial 25 psi aIncident power equals 15 w, cooling H2O equals 2 gal/hr Auxiliary gas flow of 1.0 R/min used. 107 9.0. ) Ca+ (10 ppm) at 3933 3 1b 8.0+, 7.; ., .p :3 :3 7.0” a p m I) p :3 a 6.0:) .3 Sig.-Bgrd. 5 Q 4 E O 5.0 «y m (I) x J o r E. J: 14.0” <3: 2 o H 1' U) E; 3.0.. 2 E3 4. S/N 2.0<' 4r 1.0 2 a ;. 2: c t c e : : : 4 0.0 1.0 2.0 ARGON FLOW RATE (l/min) Figure 25. Argon flow rate effects on the observed S/N ratio and the signal—background value (using the ultrasonic nebulizer). 108 On the other hand, a plot of the signal minus background value (data scaled for plotting convenience), as a func- tion of the flow rate does not reach a maximum until 9a. 1.75 i/min argon. These facts, combined with detection limit studies (which yielded poorer results than was found for the original MNS system), suggested that a less restricted flow configuration might prove beneficial. The most convenient and well-researched design was the side introduction system employed by Glass (23), Lantz (2A), and Zynger (1). It was therefore decided to com- pare the results of side introduction to those obtained using axial flow. Helium was also employed as a carrier gas in the ultrasonic system. Figure 26 shows that the maximum S/N value for helium occurs at 2.3 i/min. The increased optimum flow rate for helium over argon is not surpris- ing considering the difference in density between argon and helium. 2. Modified Veillon and Margoshes, and Crossed—Flow Nebulizers As Figure 27 clearly shows, both the Veillon and Mar- goshes, and the crossed-flow nebulizers also exhibit a S/N dependence on the carrier gas flow rate (a pressure). For the Veillon and Margoshes nebulizer, the maximum at 25 psi represents a pressure of about one-half that S/N RATIO 3.0 1.0 Figure 109 {I 1! ‘ v V 1! A 4 k A V V V t 0.0 1.0 2.0 3.0 1) HELIUM FLOW RATE (2/min) 26. Helium flow rate effects on the observed S/N ratio (using the ultrasonic nebulizer). 110 8'0 '1 0a+ (10 ppm) at 3933 X S/N RATIO C-F (aux. flow) L 0 D p O 10 20 30 “O 50 ARGON TANK PRESSURE (psi) Figure 27. Argon tank pressure effects on the observed S/N ratio (using the pneumatic nebulizers). 111 used for side introduction (past and present). Although the average droplet size is only slightly affected by this reduction, the amount of aerosol produced per unit time is definitely smaller. For the crossed—flow nebulizer, the maximum occurs at 30 psi. At low pressures, an auxiliary gas flow (Figure 7b) improves the S/N ratio. However, at high pressure, the added flow destabilizes the discharges and reduces the S/N ratio. 3. Nebulizer and Flow System Comparisons It is readily apparent from the detection limit data presented later in this chapter that the ultrasonic nebulizer produces superior results in both flow con- figurations. However, the ultimate decision on which nebulizer is to be used must take into account the sub- stantial increase in complexity of the construction and control of the ultrasonic nebulizer as compared to the two pneumatic nebulizers. In spite of the increased discharge stability observed for the axial flow con- figuration, side introduction yields better overall results. The ratio of axial to side introduction detec- tion limits is approximately A.5:l. This difference is partially attributed to the less restricted flow design of the side introduction system. A redesign of the present axial flow system, eliminating the sheath gas system (since it is only desirable at low gas flow rates), and 112 increasing the entrance and cone orifices, may improve axial results. However, there may always be an inherent advantage to side introduction where particulate matter is concerned. In the case of the two pneumatic nebulizers, which have efficiencies at least partially dependent on the tank pressure, the differences observed between the two flow systems are more dramatic. For the modified Veillon and Margoshes nebulizer, the axial to side ratio averages 7.“:1. This ratio physically demonstrates the previously discussed decrease in aerosol production per unit time at the lower pressures required for axial flow. Initially, it may seem surprising that the crossed—flow nebulizer (axial to side ratio is 5.0:1) does not fare much better than the modified Veillon and Margoshes nebulizer, especially at the lower tank pressures (axial flow) where it should be more efficient. The reason probably lies in the present design. A smaller aerosol chamber, an improved capillary adjustment system, a smaller diameter solution capillary, and a better aerosol chamber to heated chamber interface should improve the relative results. 113 E. Emission Studies 1. Background Emission Studies The background radiation emitted by the spark, like the temperature, provides some direct insight into the nature of discharge. Furthermore, the time profile of any optical noise in the background radiation emitted primarily determines the ultimate detection limits observed. Figure 28 is a time-resolved view of the emis— sion of the present MNS using tank argon over a spectral range extending from 2000A to 50003. From the diagram it is relatively easy to observe the increase and de- crease of emission from carbon (2A7BA - impurity in tank argon), hydroxyl radicals (band head at 30643 — from water impurity), the second positive band system of nitrogen (band heads at 39773, 33713, 35761, 38051, u0591 - im- purity in tank), oxygen (33708, 35173, 373ufi — impurity in tank), and the group of argon lines used in the excita— tion temperature determinations (A1503 - A3003) (93,9A). In comparison to the original MNS, the findings here again indicate a less energetic plasma. Very little continuum radiation is observed in the present dis- charge, while Zynger (1,3) observed a fairly strong continuum which reached a maximum intensity in the 0.5-2 us time period. Furthermore, hydroxyl band emission in the present system is observed to be present during the 11A 1 X10 INTENSITY 2-5 Inset 1 INTENSITY x10 5-20 Issac INTENSITY _.-‘L .J A4 A _. ‘1‘ 1 A UJH 2.099 ' ' ' j NAVELENGJH (A) X10 O O O 0 55.906 Figure 28. Time—resolved spectra of the MNS background emission. 115 0.5-2.0 us time interval with maximum emission occurring between 2.0 and 5.0 us, whereas in the original spark, the time periods were found to be 5-10 us and 30—100 us, respectively. It appears that for the present system, maximum elemental emission should occur in the 2.0-5.0 us time period. 2. Analyte Signal-to-Noise Studies The maximum signal-to-noise observed during any analytical determination is dependent on a number of variables, several of which are unique to the MNS system. The effect of the delay period on the S/N ratio has been previously explained. Figure 29 is an example of typical curves. The figure also demonstrates one possible ef- fect of an added matrix. The curve shown is that of the calcium ion S/N ratio versus time with a five-fold excess of aluminum added. Aluminum is a known calcium emission depressant, and is thought to physically reduce the free calcium concentration in plasmas (95). As would be pre- dicted, the calcium ion curve in the presence of aluminum is shifted to earlier times where the discharge is more energetic. As is expected, any gap changes which effect the breakdown voltage produce a shift in the value and tem- poral position of the maximum S/N ratio. For example, calcium ion S/N ratios were simultaneously monitored I". 116 0.: 1D .AcoHV Ezfiofimo pom ofivmu 2\m on» no mocmocmamp Amav mzHe ~aommklmsae .A.opucfi mofimv q .A.oppcfi oofimv do OHCOmmppH: ucmmop Q ofiCmeppH: on Umpcmmmaa do no ofipmmo 0 00 no 022 Hpcfimfito no panama a 0.0m IIII o.H IIIIIII IIII so oo.om IIII m.o IIIIIII IIII AHHVmo < 2 new > m.0 IIII mH.o I IIIIII IIII so 00.0 IIII 00.0 IIIIIII IIII AHvao m z 0:0 > 0.0m IIII o.H IIIIIII III! 50 00.00 IIII m0.0 uuuuuuu IIII AHHveo < 00 o.oH IIII om.o IIIIIII IIII so 00.0 IIII 00.0 IIIIIII IIII AHvao m 00 0.: IIII wo.o o.:Io.m 02mm so 00.: IIII 00.0 0.mI0.m mmmm AHvao a .eppas m.m 00.0 00.0 0.:I0.m 000m 00 m.m hm.o m.o m.mIm.m mmmm m H.m nmm.o mH.o m.mlm.m Hmmm H< 0.0 000.0 H.0 0.mI0.m 0000 m H.H wo.o no.0 w.mlw.a mfimm Hm m.H H.o wo.o m.:Im.m 0mm: Avao mm.o omoo.o Ho.o o.mIo.m mmmm AHHVmo m .mppap 00000 A5000 40 A2000 A000 A00 peeeefim m0fimum empfifisemz somfiamoEoo xpwom an 2\m .xmz mcfig Hmfix