DEVELOPMENT AND CHARACTERIZATION OF A MICROWAVE INDUCED NITROGEN DISCHARGE AT ATMOSPHERIC PRESSURE (MINDAP) FOR ELEMENTAL ANALYSIS (PLASMA GAS)код для вставкиСкачать
INFORMATION TO USERS This reproduction was m ade from a copy o f a docum ent sent to us fo r microfilming. While the m ost advanced technology has been used to photograph a n d reproduce this d o cum ent, the quality o f the reproduction is heavily dependent upon the quality o f the m aterial subm itted. The following explanation o f techniques is provided to help clarify markings or notations which may appear on this reproduction. 1. The sign or “ targ e t” for pages apparently lacking from th e docum ent photographed is “ Missing Page(s)” . If it was possible to obtain th e missing page(s) o r section, they are spliced into the film along with adjacent pages. This may have necessitated cutting through an image and duplicating adjacent pages to assure com plete continuity. 2. When an image on the film is obliterated with a round black m ark , it is an indication o f e ith e r blurred copy because o f m ovem ent during exposure, duplicate copy, o r copyrighted m aterials th at should n o t have been filmed. For blurred pages, a good image o f the page can be fo u n d in the ad jacen t frame. If copyrighted m aterials were deleted, a target note will appear listing th e pages in the adjacent fram e. 3. When a m ap, drawing o r chart, etc., is part o f the m aterial being photographed, a definite m ethod o f “sectioning” the m aterial has been follow ed. It is custom ary to begin filming at the u p p er left hand co m er o f a large sheet and to continue from left to right in equal sections with sm all overlaps. I f necessary, sectioning is continued again—beginning below the first row and continuing on until com plete. 4. F or illustrations th at cannot be satisfactorily reproduced by xerographic means, photographic prints can be purchased at additional cost and inserted into y o u r xerographic copy. These prints are available upon req u est from the Dissertations C ustom er Services D epartm ent. 5. Some pages in any docum ent m ay have indistinct print. In all cases the best available copy has been filmed. University Micrdrilnris International 300 N. Zeeb Road Ann Arbor, Ml 48106 8406789 D e u tsc h , R obert David DEVELOPMENT AND CHARACTERIZATION OF A MICROWAVE INDUCED NITROGEN DISCHARGE AT ATMOSPHERIC PRESSURE (MINDAP) FOR ELEMENTAL ANALYSIS Ph.D. Indiana University University Microfilms International 300 N. Zeeb Road, Ann Arbor, Ml 48106 1983 PLEASE NOTE: In all c a s e s this material h as been filmed in the best possible way from the available copy. Problems encountered with this docum ent have been identified here with a check mark V . 1. Glossy photographs or p a g e s ______ 2. Colored illustrations, paper or print______ 3. Photographs with dark back g ro u n d ______ 4. Illustrations a re poor c o p y ______ 5. P a g e s with black marks, not original 6. Print show s through as there is text on both sid es of page_____ 7. Indistinct, broken or small print on several pages 8. Print exceeds margin requirem ents______ 9. Tightly bound copy with print lost in spine______ 10. C om puter printout pages with indistinct print______ 11. P a g e (s )____________ lacking when material received, and not available from school or author. 12. P a g e (s)____________ seem to b e missing in numbering only a s text follows. 13. Two pages .num bered____________ . Text follows. 14. Curling and wrinkled p a g e s ______ 15. O ther_______________________________________________ ________________________ copy______ ^ University Microfilms International DEVELOPMENT AND CHARACTERIZATION OF A MICROWAVE INDUCED NITROGEN DISCHARGE AT ATMOSPHERIC PRESSURE (MINDAP) FOR ELEMENTAL ANALYSIS Robert David Deutsch Submitted to the faculty of the Graduate School in partial fulfillment of the requirements for the degree of Doctor of Philosophy in the Department of Chemistry, Indiana University November, 1983 ii T h i s is t o c e r t i f y t h a t t h e t h e s i s s u b m i t t e d by R o b e r t David D e u t s c h has been a c c e p t e d by the Ph.D. A d v i s o r y C o m m i t t e e as s a t i s f a c t o r y in p a r t i a l f u l f i l l m e n t o f t h e r e q u i r e m e n t s f o r the Ph.D. d e g r e e . rt-jLu) _ iii To the Dean and F a c u l t y of the G r a d u a t e S c h o o l : We, t h e u n d e r s i g n e d m e m b e r s o f t h e F a c u l t y o f t h e G r a d u a t e S c h o o l and m e m b e r s o f t h e P h . D . C o m m i t t e e a p p o i n t e d for the e x a m i n a t i o n of Ro b e r t D a v i d D e u t s c h e x a m i n e d h i m on N o v e m b e r 22, 1 9 8 3 . It is h e r e b y c e r t i f i e d t h a t he h a s s u c c e s s f u l l y p a s s e d t h e e x a m i n a t i o n s in m a j o r a n d m i n o r s u b j e c t s . W e r e c o m m e n d Mr. D e u t s c h f o r t h e d e g r e e D o c t o r o f Philosophy. Chai rman ^Research A d v i s e ^ ^ ' l£. ruA — ACKNOWLEDGMENTS To Margaret A. Flanagan, my wife and professional associate, 1 wish to express my love and appreciation for your encouragement, understanding and patience throughout our graduate careers at Indiana University. To my son Eric Thomas, whom I love very much, I want to thank you for your assistance and cooperation in helping with experiments. To my mother Diane, brother Miles, and dear departed father, thank you for instilling the perseverance needed to complete this degree. Your love and encouragement throughout all of my educational endeavors has been invaluable. I sincerely wish to thank Professor Gary M. Hieftje for his patience, advice and financial support throughout my graduate studies. His enthusiasm in the pursuit of different problems has broadened my awareness not only in Analytical Chemistry but in other disciplines of science. I would like to thank all the members of Dr. Hieftje's research group for their scientific assistance and friendship. Especially helpful and important has been the friendship of Jerry Keilsohn. Many thanks go to Dr. R. Mark Wightman and the members of his research group for letting me use their computers to type this thesis. A special word of appreciation goes to all the staff members of the department for their help and assistance. Special thanks is in order for the talented glass blowing of Don Fowler, and all the members of the Instrument and Machine shops. I am grateful to Dr. Ken Force for his interest and encouragement in my pursuit of this degree. DEVELOPMENT AND CHARACTERIZATION OF A MICROWAVE INDUCED NITROGEN DISCHARGE AT ATMOSPHERIC PRESSURE (MINDAP) FOR ELEMENTAL ANALYSIS ROBERT DAVID DEUTSCH INDIANA UNIVERSITY BLOOMINGTON, INDIANA ABSTRACT Emission spectroscopy is an increasingly attractive tool for the determination of trace chemical elements because of recent developments in plasma excitation sources. These sources have always had their place in analytical chemistry but with the exponential growth and interest in trace elemental analysis, plasmas are becoming the tool of choice. The two plasmas most commonly used for spectrochemical analysis are the radiofrequency inductively coupled plasma (ICP) and the microwave-induced plasma (MIP). The ICP is employed primarily for the qualitative and quantitative determination of metals in solution but has also been successfully applied to the analysis of gaseous and solid samples. Unlike the ICP, the MIP has generally been used as a detector for gas chromatography or for metals in other volatile forms. Its inability to efficiently desolvate, atomize and excite liquid samples has limited its analytical utility. This dissertation focuses on the development of a new microwave-supported plasma which overcomes these operational difficulties. The full potential of this new emission source is critically evaluated. Development of -the new MIP system was centered on using molecular nitrogen as the plasma support gas. Nitrogen is not only more economical than inert He or Ar gases used formerly, but is also more efficient at nebulizing and atomizing solution samples. To sustain a stable microwave-induced nitrogen discharge at atmospheric pressure (MINDAP), modifications were made to a microwave resonant cavity and plasma torch; the resulting plasma can easily be ignited and extends several centimeters beyond the cavity. This plasma "tail flame" was examined critically as a region for viewing atomic emission from samples introduced into the discharge. Background and analyte emission in the ultraviolet (UV) and visible spectral regions was mapped to provide information about the energy distribution in the plasma. Also, signal intensities were evaluated at different powers, flow rates, and optical viewing geometries in order to ascertain optimal analyte excitation conditions. Analytical figures-of-merit of the MINDAP system were evaluated using a pneumatic nebulizer and desolvation apparatus for sample introduction. With the plasma oriented vertically and viewed in a side-on manner, detection limits were generally in the part-per-billion (ng/mL) range and calibration curves (emission intensity vs. solution concentration) extended over a range of three-to-five orders of magnitude. Interelement (matrix) interferences, similar to those found in flame spectroscopy, were observed but could be overcome through the use of "releasing agents". The temporal stability of the plasma was investigated with the aid of noise amplitude spectra and a correlation was noted between emission intensity and microwave power fluctuations. Noise amplitude-spectrum analysis was performed using a Fast Fourier vii Transform algorithm to identify the limiting noise source of the system. These results are discussed and compared with those obtained from other emission sources. Development and characterization of the MINDAP system included fundamental studies involving the physical characteristics of the plasma. Excitation, ionization and rotational temperatures were determined and found to be approximately equal, suggesting that local thermodynamic equilibrium (LTE) might be established in the discharge. Spatially resolved excitation and ionization temperatures were comparable. Similarly, spatial emission profiles for elements of different excitation potentials closely followed those expected from a Boltzmann distribution. These results support the suggestion that excitation of analyte species in the tail flame occurs by a thermal mechanism. In the final study, a microarc atomizer was used to introduce nanogram-to-microgram quantities of atomic vapor reproducibly into the plasma. The technique of microsampling and preatomization is well suited for sample introduction into any emission source. The microarc-MINDAP combination was evaluated for the analytical figures-of-merit noted earlier. Compared to results obtained with the pneumatic nebulizer, the microarc-MINDAP system produced lower detection limits and greater freedom from classical interferences. This combination provides the economy and sensitivity needed for routine analysis. viii TABLE OF CONTENTS LIST OF TABLES xi LIST OF FIGURES xii CHAPTER 1 INTRODUCTION REFERENCES CHAPTER 2 DEVELOPMENT OF A MICROWAVE INDUCED NITROGEN DISCHARGE AT ATMOSPHERIC PRESSURE (MINDAP) _ 1 5 6 \ EXPERIMENTAL Microwave Cavity Torch Operating the Plasma Plasma Features Reagents 9 19 25 26 31 32 RESULTS Viewing Geometry Applied Power and Nitrogen Flow Viewing Position in the Tail-Flame 33 33 41 54 CONCLUSION 57 REFERENCES 61 CHAPTER 3 ANALYTICAL CHARACTERIZATION OF A MICROWAVE INDUCED NITROGEN DISCHARGE AT ATMOSPHERIC PRESSURE (MINDAP) 64 EXPERIMENTAL 65 RESULTS AND DISCUSSION Detection Limits Working Curves Precision Interferences 66 66 68 68 71 CONCLUSION 85 REFERENCES 86 ix CHAPTER 4 PHYSICAL MEASUREMENTS OF A MICROWAVE INDUCED NITROGEN DISCHARGE AT ATMOSPHERIC PRESSURE TEMPERATURES AND THERMODYNAMIC EQUILIBRIUM 88 89 EXPERIMENTAL Temperature measurements and calculations Electronic Excitation Temperature Ionization Temperature Electron Number Density Rotational Temperature 93 95 95 96 98 100 RESULTS Electronic Excitation Temperature Ion/Atom Ratios and Ionization Temperature Effect of Central-Channel Nitrogen Flow Rotational Temperature Spatial Profiles of Elemental Emission Spatially Resolved Temperatures 104 104 107 114 117 120 120 DISCUSSION 125 CONCLUSION 128 REFERENCES 130 CHAPTER 5 IDENTIFICATION OF LIMITING NOISE SOURCES IN THE MINDAP 132 EXPERIMENTAL Wavelength-Scan Flicker-Ratio Measurement Time Constant Effects on Precision Time Trace Correlation Noise Amplitude Spectral Determination 133 134 135 136 137 RESULTS Precision and Low-Frequency Noise High-Frequency Noise 141 141 147 DISCUSSION 155 CONCLUSION 168 REFERENCES 170 CHAPTER 6 MICROSAMPLE INTRODUCTION INTO THE MINDAP USING A MICROARC ATOMIZER 171 EXPERIMENTAL 173 RESULTS AND DISCUSSION Analytical Figures-of-Merit Detection Limits Working Curves Precision Interferences 181 182 182 186 186 190 X CONCLUSION 190 REFERENCES 197 CHAPTER 7 VITA CONCLUSION AND FUTURE WORK 199 CONCLUSION 199 FUTURE WORK 200 203 xi List of Tables Table Page 2-1 Specific components of the MINDAP system. 12 2-2 Interfering background band emission. 42 3-1 Detection limits of the MINDAP. 67 4-1 Temperatures for several Atmospheric Pressure MIPs. 92 4-2 Temperatures for several Atmospheric Pressure ICPs. 94 4-3 Spectroscopic data for electronic excitation temperature measurements using Fe (I) emission. 97 4-4 4-5 5-1 6-1 6-2 Ionization temperatures calculated for a range of ion/atom ratios for different electron number densities. Ill The power required to heat nitrogen to a specified temperature. 126 Initial and statistical values for each signal collected from the time tracings for 400 min. 161 Detection limits for the MINDAP microarc and nebulizer sample introduction systems. 183 Comparison of detection limits for discretesample-introduction systems. 185 xii List of Figures Figure Page 2-1 Schematic diagram of MINDAP measurement system. 11 2-2 Schematic diagram of the desolvation system. 16 2-3 Schematic diagram of uni - bipolar converter. 18 2-4 Different viewing geometries for the MINDAP system. 21 2-5 Side-on view of modified TM qjq resonant cavity. 24 2-6 Cross-sectional view of cavity, torch and plasma. 29 2-7 Comparison of viewing geometries for 190 - 360 nm spectral region. 35 Comparison of viewing geometries for 390 - 465 nm spectral region. 37 Comparison of viewing geometries for 650 - 800 nm spectral region. 39 2-10 Effect of power on plasma length. 44 2-11 Effect of powerand flow rate on Fe (I) emission. 46 2-12 Effect of power on several atom transitions. 49 2-13 Effect of power on several ion transitions. 51 2-14 Effect of power on signal-to-background noise ratio. 53 2-8 2-9 2-15 Spectral scans of plasma background and analyte emission for different vertical positions in the tai1-flame. 56 2-16 Emission spectrum for 10 ng/mL Li and 100 ng/mL K. 60 3-1 Analytical working curves for MINDAP. 70 3-2 Sodium interference on 10 ppm calcium emission. 73 3-3 Sodium interference on 10 ppm calcium emission with cesium added as an ionization suppressant. 76 3-4 Aluminum interference on 10 ppm calcium emission. 78 3-5 Phosphate interference on 10 ppm calcium emission. 80 xiii Figure 3-6 3-7 4-1 Page Phosphate interference on 10 ppm calcium emission with 0.01M EDTA added as a releasing agent. 82 Aluminum interference on 10 ppm calcium emission with 0.01M EDTA added as a releasing agent. 84 Emission spectrum of the N£+ First Negative system. 103 Spatially averaged excitation temperatures in the MINDAP tail-flame. 106 4-3 Magnesium ion/atom ratios in the MINDAP tail-flame. 109 4-4 Ionization temperatures in the MINDAP tail-flame. 113 4-5 Effect of central channel nitrogen flow rate on ionization temperatures. 116 Rotational temperatures from the discharge region of the MINDAP. 119 Spatial emission profiles of several elements in the tail-flame of the MINDAP. 122 Spatially resolved electronic excitation and ionization temperatures in the MINDAP tail-flame. 124 An illustration of the signals used for obtaining a noise - amplitude spectrum. 139 5-2 Effect of time constant on precision. 143 5-3 Noise spectrum of OH band emission over the frequency range of 0 - 5 Hz. 146 Effect of calcium concentration on emission signal and precision. 149 Noise spectrum of OH band emission over the frequency range of 0 - 500 Hz. 151 Noise spectrum of OH emission when large amounts of water were intentionally introduced into the plasma. 154 Effect of microwave power on two different impedance settings. 158 Time tracings of the emission signal, forward and reflected power for a 400-min time period. 160 4-2 4-6 4-7 4-8 5-1 5-4 5-5 5-6 5-7 5-8 xiv Figure 5-9 Page Correlation plot of forward power and emission signal. 163 Correlation plot of forward power and reflected power. 165 Correlation plot of reflected power and emission signal. 167 6-1 Microarc-MINDAP operational configuration. 175 6-2 Block diagram of the microarc-MINDAP instrumentation. 177 Microarc-MINDAP emission-time profile for 1 ng copper. 180 Calibration curves for the microarc-MINDAP system. 188 6-5 Sodium interference on 10 ng of calcium. 192 6-6 Phosphate interference on 10 ng of calcium. 194 5-10 5-11 6-3 6-4 1 CHAPTER 1 Introduction Today one of the most popular- emission techniques for the trace and ultra-trace detection of elements uses the radiofrequency (27.12 MHz) inductively coupled plasma (ICP). This atmospheric- pressure electrical discharge is used to decompose an introduced sample ' (aerosol, solid, powder, vapor) into its atomic and ionic constituents and excite them. The popularity of this plasma has arisen because it is easy to use, has few interferences, operates at atmospheric pressure and has made possible the absolute and relative detection of elements down to the sub - pg and sub-parts-per-billion limits, respectively. Although ICP atomic emission spectrometry exhibits a high degree of elemental specificity, reproducibility and sensitivity, it has drawbacks which many investigators have been diligently attempting to overcome. These liabilities are of three basic kinds; cost, size and susceptibility to interferences. Another electrical discharge is one that has been used for analytical determinations since the 1950 's (1), is operated at 2450 MHz, and is called a microwave induced plasma (MIP). The MIP was originally sustained at reduced pressure in a resonant cavity. In such a configuration, this emission source was exploited as an elementselective detector for analysis of metals and nonmetals eluting from a gas chromatograph (2-8). References 6 through 8 constitute a good review with many applications and references for the GC-MIP combination. The low-pressure MIP was also used as a source of intense monochromatic emission from the actinide and lanthanide elements 2 in what is known as an electrodeless discharge lamp (EDL) (9,10). Although practical and sensitive, the MIP was judged inconvenient because of its operation at reduced pressure. This fact, coupled with the overwhelming acceptance of the ICP, precluded the further rapid development of the low-pressure MIP. In 1973 the development and characterization of an atmosphericpressure microwave-induced plasma (11-13) rekindled interest in this emission source. Since then the MIP has been applied to a variety of determinations, not only in analytical chemistry but also in clinical biochemistry (14,15) and geology (16,17). The major advantages of this type of plasma source are its low power and gas-flow requirements, the antithesis of the ICP's failings. Unfortunately, there are two remaining features that hinder further acceptance of the atmosphericpressure MIP: its inability to efficiently decompose aerosol samples and the absence of automated systems for impedance matching (as are available for the ICP). This dissertation is devoted to the development of a practical, element selective, sensitive microwaveinduced plasma atomic emission source operated at atmospheric pressure for the detection of trace elements. Chapter 2 describes the problems that afflict most microwave plasmas and how recent advances in ICP instrumentation promoted the development of a new MIP source for the analysis of aerosol samples. This new plasma, sustained using nitrogen as the nebulizer and plasma-supporting gas, employs a modified torch and microwave cavity to enable the discharge to be easily ignited and maintained. The new plasma has a flame-like appearance and extends several centimeters beyond the resonant cavity. Termed a Microwave Induced Nitrogen 3 Discharge at Atmospheric Pressure (MINDAP), the device readily accepts aerosol from a conventional nebulizer sample-introduction system. Different viewing geometries (end-on and side-on) are assessed for utility and signal-to-background ratios. The dependence of analyte emission intensity and signal-to-noise ratio on flow rate and microwave power was determined. In Chapter 3 the MINDAP was operated under the conditions and the configuration determined in Chapter 2 to be optimal. in the subsequent Criteria in used evaluation were the detection limits of sixteen elements at thirty-three wavelengths, the susceptibility of the MINDAP to interference from aluminum, phosphate and sodium on the emission signal of calcium, linearity of calibration curves and the precision of analytical measurements. Further characterization of the MINDAP system included a study of its fundamental characteristics. In Chapter 4, excitation, ionization and rotational temperatures of the plasma were determined and found to be approximately equal, suggesting that local thermodynamic equilibrium (LTE) is approached in the discharge. Spatially resolved excitation and ionization temperatures were comparable. Similarly, spatial emission profiles for elements of different excitation potentials closely followed those dictated by a Boltzmann distribution. Origins of noise (fluctuations) in the emission signal of the MINDAP are studied in Chapter 5. Four different methods: time constant effect, wavelength-scan flicker-ratio measurement, noise amplitude spectral determination, and time-trace correlation were used to identify noise sources in the system. Noise amplitude spectral analysis was performed using a Fast Fourier Transform algorithm to 4 identify the noise frequency components generated by the nebulizer sample-introduction system and the MINDAP. Analyte emission noise spectra were acquired for desolvated and undesolvated aerosol, for varying analyte concentration, and for background emission from nitrogen species and hydroxyl radicals generated from water vapor. The system appears to be flicker-noise limited below about 1 Hz; only 60 and 120 Hz signals are detected in the frequency region from 0 to 500 Hz. In Chapter 6, a microarc atomizer was used to introduce nanogram quantities of atomic vapor reproducibly into the MINDAP. The technique of microsampling and preatomization is well suited for sample introduction into any emission source. The microarc-MINDAP combination was evaluated for the same analytical figures-of-merit employed in Chapter 3. Compared to results obtained with the pneumatic nebulizer, the microarc-MINDAP system produced lower detection limits and greater freedom from classical interferences. This combination provides the economy and sensitivity needed for routine analysis. In the final chapter of this thesis, a summary of the results are presented along with some ideas about future work needed to fully understand and develop* this new plasma into an accepted technique for the determination of trace elements. 5 REFERENCES 1. H. P. Broida and M. W. Chapman, Anal. Chem., 30, 2049 (1958). 2. C. A. Bache and D. J. Liske, Anal. Chem., 39, 786 (1967). 3. J. P. J. van Dalen, P. A. de Lezenne Coulander and L. de Galan, Anal. Chim. Acta, 94, 1 (1977). 4. B. D. Quimby, P. C. Uden and R. M. Barnes, Anal. Chem., 5Q_, 2112 (1978). 5. C. A. Bache and D. J. Liske, Anal. Chem., 43, 950 (1971). 6. I. S. Krull and S. Jordan, Amer. Lab., 12_, 21 (1980). 7. A. T. Zander and G. M. Hieftje, Appl. Spectrosc., 35, 357 (1981). 8. B. L. Sharp, Sel. Ann. Rev. Anal. Sci., 4^, 37 (1976). 9. M. J. Al. Ani, R. M. Dagnall, T. S. West, Analyst, 92, 597 (1967). 10. R. Avni, and J. D. Winefordner, Spectrochim. Acta, 30B, 281 (1975). 11. C. I. M. Beenakker, Spectrochim. Acta, 31B, 483 (1976). 12. C. I. M. Beenakker and P. W. J. M. Boumans, Spectrochim. Acta, 32B, 173 (1977). 13. C. I. M. Beenakker, B. Bosman and P. W. J. M. Boumans, Spectrochim. Acta, 33B, 53 (1978). 14. H. Kawaguchi and B. L. Vallee, Anal. Chem., 47_, 1029 (1975). 15. I. Atsuya, G. M. Alter, C. Veillon, and B. L. Vallee, Anal. Biochem., 79, 202 (1977). 16. K. Govindaraju, G. Mevelle and C. Chouard, Anal. Chem., 48, 1325 . (1976). 17. J. 0. Burman, "Applications of Plasma Emission Spectrochemistry", ed. R. M. Barnes, Heyden Publishers, 1979, p.15. 6 CHAPTER 2 Development of a Microwave Induced Nitrogen Discharge at Atmospheric Pressure (MINDAP) Atomic emission spectrometry is today perhaps the most widely used method for trace elemental analysis. Of the alternative emission sources being employed, rf plasma systems have provided some of the most sensitive measurements. The rf plasmas are comprised principally of two groups: the inductively coupled plasma (ICP) which usually operates at 27.12 and 40.68 MHz and the microwave-induced plasma (MIP) which operates at 2450 MHz. For solution analysis, the ICP is by far the most widely employed because aerosol can be easily directed into it from a suitable nebulizer system. The ICP is then capable of desolvating the aerosol sample, decomposing the dried salt particles into their atomic and ionic constituents, and exciting the resulting atoms and ions. Most analytical measurements are made high in the ICP tail flame where background emission from the plasma is minimal. Although the ICP would appear to be a nearly ideal source for atomic emission measurements, its initial and operating costs are extremely high. Moreover, intense background emission, even in the plasma tail-flame, complicates the measurement of many analytical lines. As a result, high-resolution dispersion systems or exotic modulation schemes (1) are needed to minimize spectral interferences. In comparison, the MIP is inexpensive to purchase and operate. It enables the sensitive detection of both metals (2) and nonmetals (3,4) but has been limited to use with samples introduced in a vaporized or 7 atomized form. The most successful means of sample introduction into the MIP has been through use of a gas chromatograph (2-6). Other methods that have been employed include thermal and nonthermal atomizers (7-10), hydride generators (11,1-2), and laser vaporization devices (13). Recent reviews (14,15) have listed many other alternative sample-introduction techniques for the MIP. . Solution analysis is difficult and cumbersome with the MIP, mainly because of its low power (20 - 200 W), low support-gas flow rate ( < 2 L/min), small physical dimensions (4 mm in diameter), low thermal temperature (^2000°K), and an operating frequency which corresponds to an absorption band of water. Ironically, many of these same features are what make the MIP such an affordable and potentially useful emission source. Clearly, a MIP system that could directly analyze aerosol samples without the need for complicated sampleintroduction arrangements would be very appealing as a compact, easyto-use, and sensitive emission spectroscopic tool. Although the MIP is practical and sensitive, the overwhelming acceptance of the ICP has foreshadowed its development. Current trends are to improve the performance of the ICP and reduce its cost. One of the first ICP improvements was the design of a torch that controlled and stabilized the flow pattern of the plasma gas and consequently the plasma discharge. Recently, it has been found possible to reduce the instrumental and operational costs of the ICP but maintain the sensitivity, simply by reducing the plasma size. Numerous publications have demonstrated that a reduction in power consumption, support-gas flow rate, and torch size do not significantly degrade sensitivity or analytical performance (16-19). 8 Other factors that contribute to the atom-excitation ability of the ICP are its operating frequency and type of support gas. Increasing the frequency from 27.12 to 40.68 MHz has a pronounced effect on sensitivity (20). Similarly, ICPs sustained with nitrogen gas (because of its economy, availability, and ionization energy comparable to Ar) have been extensively studied (21-27). Unfortunately, the N^ ICP yielded poorer detection limits under similar operating conditions than the Ar discharge (28-31). Although in other studies, Capitelli, et al. (28), and Barnes and Nikdel (29) determined experimentally and theoretically that nitrogen ICPs are more efficient at decomposing analyte particles than argon ICPs. Other information about signal-to-background ratio, ion-to-atom ratio and applied power have suggested that the nitrogen ICP is closer to local thermodynamic equilibrium (LIE) than a comparable argon plasma (24,26). Nitrogen has been used also as a support gas for other electrical discharges. Cobine and Wilbur (30) sustained an atmospheric-pressure microwave (1000 MHz) plasma with nitrogen and other diatomic gases. These plasmas were found to possess an unusually high thermal temperature which the authors attributed to the heat of association of the molecules dissociated by the discharge. Eckert, £t^ _al_. noted that molecular gases possess more thermal energy than do atomic gases because of the energy available in the rotational, vibrational and electronic levels. In contrast, energy storage in raonatomic gases is possible only in translational or electronic states (21). Another advantage to using nitrogen as a plasma gas accrues from the possible formation of "active nitrogen" (31). Two techniques, APAN (Atmospheric Pressure Active Nitrogen) (32) and METALS (Metastable 9 Energy Transfer for Atomic Luminescence) (33,34), have exploited the high energy available in active nitrogen for exciting metal and nonmetal atoms. Interest in the microwave plasma (2450 MHz) has been revived largely because of the recent development of a resonant cavity to easily sustain either an argon or helium plasma at atmospheric pressure (5,35,36). Another contributing improvement adapted from the ICP has been the stabilization and control of support-gas flow patterns (37) through the design of a new microwave plasma torch. These developments have been utilized here to generate a stable discharge in flowing nitrogen. In this chapter, a microwave induced nitrogen discharge at atmospheric pressure (MINDAP) is sustained by a new torch assembly and evaluated with an pneumatic nebulizer and desolvated-aerosol sampleintroduction system as an analytical emission source. Operating parameters (power, flow rate, and viewing geometry) were optimized for signal-to-noise and signal-to-background ratio. A discussion of the background spectral features, their origin, and their effect on analytical measurements is presented. EXPERIMENTAL The instrumentation used in this study is shown schematically in Figure 2-1; Table 2-1 lists all optical, electrical and gas-handling equipment and the respective manufacturers. Sample solutions are introduced into the plasma with a conventional pneumatic nebulizer / spray chamber system. A concentric glass nebulizer (38) operating at a 10 Figure 2-1: Schematic diagram of the MINDAP measurement system. The broken lines indicate that the computer was remotely located from the experimental apparatus. The Tektronix 4051 terminal was used only for plotting the collected data from the PDP 11/34 computer and so its connection is designated by dashed lines. i-to-V - current to voltage converter. B.P.C. - Bipolar converter. H.V.P.S. - High Voltage Power Supply. PMT - Photomultiplier tube. 11 PIa*ma I to V AAA/V” Monochromator —/ 'onvor t«r— -5 B.P.C. — i tmp I If I«r » Samp I H.V. P.S. IntroductIon Svatam PDP 11/34 T«rmI no I or MXNC Computar Graphic Plottar TaktronIx 4051 TarmInaI Table 2-1: COMPONENT Specific componenCs of MINDAP system. TYPE/DESCRIPTION COMPANY/LOCATION Gas handling equip. Regulators/flow meters Type 603 Matheson Corp., Joliet, IL. Spherical Mirror f = 108 mm, MgF„ overcoat, d » 108 mm Oriel Corp., Stamford, Conn. Lens f = 50 mm, suprasil Diameter = 75 mm Melles Griot, Irvine, CA Microwave cavity Modified TMq ^q cavity I. U. Machine Shop, Bloomington, IN Microwave generator Model 420, 500 W @ 2450 Mhz Micro - Now Instr. Co. Chicago, IL Microwave powersupply cables RG 115 A/U Times Wire and Cable Corp., Conn. Double-stub tuner Model DS109 Weinschel Eng., Gaithersburg, MD Plasma torch Modified ICP-style torch. See Fig. 2-6. I. U. Glass Shop, Bloomington, IN NebulizerSpray chamber concentric glass conventional spray chamber I. U. Glass Shop, Bloomington, IN Aerosol condenser Model 9270 Ace Glass Inc., Vineland, NJ Monochromator EU-700, f.l. = 0.35 m Heath Co., Benton Harbor, MI PMT R928 Hamamatsu Co., Middlesex, NJ H. V. Power Supply Model 244 @ -800 V Keithley Instr., Cleveland, OH Current Amplifier Model 427 Keithley Instr., Cleveland, OH Amplifier Uni - bipolar converter driver and multiplier Author's design (Fig. 2-3) 13 Table 2-1: Specific components of MINDAP system (continued). TYPE/DESCRIPTION COMPANY/LOCATION CRT ADM 3A Lear Siegler Inc., Anaheim, CA Data Acquisition PDP 11/34; MINC-11/23 Digital Equip. Corp., Maynard, MA Graphic System Tektronix 4051 terminal Tektronix Co., Beaverton, OR Recirculating cooler Multicool system FTS Systems, Stone Ridge, NY COMPONENT 14 flow rate of 1.75 L/min of nitrogen gas aspirates sample solution at a rate of 1.8 mL/min. The resulting aerosol passes through a conventional spray chamber and desolvation system (39) en route to the MINDAP. The desolvation apparatus, shown in Figure 2-2, consists of a glass tube 20 mm o.d. x 150 mm long which was wrapped with heating tape. A Variac was used to control the voltage necessary to maintain the tube at approximately 140°C for drying the aerosol. The water vapor was then removed by two tandem spiral condensing tubes which were maintained at -20°C by a refrigerated recirculating water-methanol bath. Unlike other MIPs the MINDAP is capable of decomposing and analyzing aerosols injected directly into it; however, under these conditions both background and noise levels were higher. Consequently, desolvation was employed for all studies described below. Emission from the plasma was focussed onto the entrance slit of the monochromator using either a lens or mirror assembly. The resulting PMT photocurrent was converted to a proportional voltage by a current amplifier and then to a bipolar (-5 to +5 V) signal by a laboratory-constructed amplifier. This final signal was then compatible with the input of the analog-to-digital converter of the PDP 11/34 computer. The schematic diagram of the bipolar amplifier is shown in Figure 2-3. This amplifier consists of four stages, each implemented with an operational amplifier. The first stage (A) adds an offset voltage (-5 V) to the input signal (0 - 10 V) and provides intermediate gain in multiples of 1, 2, and 5. The second amplifier stage (B) is adjusted for unity gain and inverts the output of the first stage to produce 15 Figure 2-2: Schematic diagram of the desolvation system employed for all measurements. To Pl a s m a AerosoI Dryer Nebulizer Chamber Rec ircuI at ing Coo Iant Samp Ie So Iut ion Heat ing Tape Tandem Condenser Drai n 17 Figure 2-3: Schematic diagram for the unipolar-to-bipolar converting amplifier. 0.01 uF 0.01 uF I n put »-VSAAAA- 356 356 356 ba in -1 8 - +10 VO Its |— W M A /*— ^ OFFset 356 Output 19 the correct signal polarity. The third stage (C) is used to drive the voltmeter and to avoid loading the second stage. The final stage (D) is a voltage follower used as a driver to send signals up to 100 ft to the remote computer. To test the circuit and determine the line losses and noise, a constant-voltage source was connected to the constructed amplifier and also directly to a MINC-11/23 computer (to monitor the voltage fluctuations of the voltage supply) through four feet of cable. The signals from both computers were compared over a 24-hr period; each deviated from the actual signal by less than 1 part in 4096 (the resolution of the A/D converter). Two different viewing geometries (axial and radial) for the MINDAP system have been evaluated and are diagrammed in Figure 2-4. The axial (end-on) method of viewing is the most common geometry for MIP systems. In a recent review by Zander and Hieftje (15), such plasma systems are described. In this configuration, the emitted radiation from the plasma inside the cavity was collected with a lens, or series of lenses, and focused onto the monochromator with unity magnification. In the radial (side-on) configuration, a spherical mirror formed a 2:1 magnified image of the plasma tail-flame on the monochromator entrance slit. Microwave Cavity The Beenakker cavity described by Zander and Hieftje (8) was initially used for igniting and sustaining the nitrogen discharge. Although that configuration enabled a nitrogen plasma often to be sustained, the discharge wandered about the torch and usually resided 20 Figure 2-4: Different viewing geometries for the MINDAP system. A. Radial (side-on) configuration. B. Axial (end-on) configuration. J..M. Impedance Matcher (double-stub tuner). TaiI-Flame Monochromator Cav ity MICROWAVE P.S. Samp Ie AerosoI B Cavi ty — “l i A Monochromator --- > Samp Ie Aeroso1 V JL I. M. MICROWAVE P.S. P M T 22 on the inner walls of the open quartz tube. Moreover, it was difficult to minimize reflected power without extinguishing the plasma, and the cavity and all electrical connections became hot with use. These heating difficulties have been reported by others (15,40) and recent studies have sought to overcome them through cooling and internal tuning of the resonant cavity (41,42). Unfortunately, cooling the cavity with air did not eliminate these problems in the present study. Instead, it was necessary to modify the resonant cavity and torch design. The cavity was altered in two ways to permit its use with different plasma gases and torch diameters. The first modification permitted simplified, reproducible alignment of different diameter torches in the center of the cavity where the electric field is greatest. Copper adapters and face plates were machined to fit differently sized plasma torches. The adapters had a standard outer diameter which were slip-fitted into the cavity housing. New face plates with similarly sized holes were machined because the face plates were not thick enough to securely support a slip-fitted adapter. The second cavity modification extended the tuning range of the system. The brass tuning screws ordinarily mounted on the cavity (both axially and radially) were replaced with a double-stub tuner and quartz tuning rod as shown in Figure 2-5. The rod was positioned axially into the cavity and parallel to the plasma torch. It has been stated that the tuning range of such a system depends on the diameter and composition of the dielectric tuning rod, and on its axial and radial position from the cavity center (43). Different diameter quartz rods were tested and a 6-mm-diameter rod inserted axially halfway into the cavity was optimum for igniting and tuning the nitrogen discharge. 23 Figure 2-5: Side-on view of modified TMq j q resonant cavity. dimensions are in mm. All 24 U G — 58 Connector C o u p I ing Loop : I*-5 s s PIasma Tube s Q u a r tz T u n in g s s Rod Copper Body 25 Torch The nitrogen discharge was initially sustained in a standard 6-mm o.d., 4-mm i.d. open quartz tube. Unfortunately, even with the cavity modification, the plasma was then not always temporally and spatially stable, probably because of the uncertainty in the gas-flow pattern through the tube. This instability would occasionally cause the plasma to reside on the inner walls of the discharge tube and create a "hot spot". Etching of the tube was apparent from the resulting increased Si and Na emission; this etching eventually led to the tube's destruction. The plasma instability also influenced the degree to which the analyte vapor mixed with the discharge. These kinds of plasma instabilities have been noted by others also (15,34). To overcome these problems, a stable gas-flow pattern was produced through use of a torch similar to that employed with the ICP (44). A similar approach was taken by Bollo-Karma and Codding (37), who used a threaded quartz insert to create in a MIP the spiral-flow pattern characteristic of the ICP. The present torch is similar in concept but simpler in design; it consists of two concentric quartz tubes, an outer one of 6-mm o.d. (4-mm i.d.) and an inner one of 1-mm o.d. (0.5-ram i.d.). The inner tube is approximately 2.5 cm in length. A stable tangential gas-flow pattern is created by forcing the plasma gas to flow around the central tube. This pattern was verified using the hydrodynamic technique described by Sexton, Savage and Hieftje (45). The nitrogen flow introduced tangentially contains the analyte aerosol and sustains the plasma. The low flow through the central tube is mainly to keep the hot plasma from contacting that tube and melting 26 it. If the plasma does touch the central tube, the tube will soften and deform, changing both the tuning of the cavity and the stable spiral-flow pattern of the sustaining plasma gas. Initially, it was intended to introduce the analyte through the central tube (similar to an ICP) but the back pressure created by the small orifice of the torch made using a flow-dependent nebulizer difficult. Although a flow- independent sample-introduction system such as an ultrasonic nebulizer could conceivably be used, one was not available for use in the present study. Operating the Plasma In early trials the new MINDAP torch was mounted like that of Codding (37), where the central tube (or Codding's threaded insert) was aligned with the outside wall of the microwave cavity. Once the dis charge was ignited, a suspended plasma was then formed and tuning was accomplished through use of the axial quartz rod and double-stub tuner. Unfortunately, it was not always obvious which tuning element needed to be adjusted, since the position of each element altered the effect of the others. employed. Consequently, an iterative adjustment was Later, it was determined that the quartz rod was always located halfway into the cavity for minimum reflected power. Fixing the quartz rod at this optimal place then made lighting the MINDAP as easy as with other microwave plasmas. In the final and presently used configuration (see Figure 2-6), the axial quartz rod has been eliminated entirely; intracavity tuning is effected with the MINDAP torch itself. The torch is mounted so its 27 central tube extends into the TMQ10 cavity. The added dielectric from the the central tube serves the same purpose as the axial quartz rod did formerly. As with the quartz rod, this arrangement promotes reproducible plasma ignition because the central tube is always fixed in position. Importantly, the reflected power is even lower with this configuration than with the quartz-rod tuning element, probably because of the radially central position of the added dielectric. Plasma torches with different central-tube lengths were tested; it was found that the optimal length for plasma stability and tuning was equal to half the inside cavity thickness (approximately 1.25 cm). Whenever the torch was replaced, minor adjustments of the double-stub tuner were necessary to compensate for variations in the diameter and length of commercial quartz tubes. At present, the major obstacle in tuning the cavity is oxide formation on the solid copper housing; periodically, the cavity must be disassembled and cleaned with a household coppercleaning product. In practice, the MINDAP is ignited in the following manner. A flow of greater than 1 L/min of nitrogen gas (typically 1.75 L/min) is introduced into the side-on inlet and approximately 30-100 mL/min through the central tube of the torch. An applied power of greater than 200 W (typically 250 - 350 W) is applied to the cavity. A tungsten wire attached to a wooden match stick (for insulation) is inserted and moved about in the quartz torch within the ™ cavity. q ^q The wire is inductively heated by the field and the ejected 28 Figure 2-6: Cross-sectional close-up view of the modified torch, nitrogen discharge, and modified cavity. 29 AFTERGLOW CGreen Sheath) Blue Discharge ™oio Cavi ty Pink Discharge Copper Adapter Modified Torch m ■n2 * „Dri®d Samp Ie / N. 30 electrons seed the flowing gas to initiate the plasma. The plasma is then tuned with the double-stub tuner for minimum reflected power. Interestingly, when the plasma is not tuned properly there is an audible hiss from the discharge. Elimination of this noise corresponds with minimum reflected power of about 5 - 20 W with 250 W of applied power. The MINDAP was not studied at power levels exceeding 250 W because of limitations of the cables, tuning stubs and electrical connections. At higher power settings the tuning stubs had to be carefully set or large amounts of power could be reflected from the cavity, causing it, the cables, and tuning stubs to become hot. Another problem at such high power levels was the tendency of the plasma to contact the walls of the torch and etch them when sudden changes in tuning occurred. This situation not only led to spurious signals but also made the plasma spatially unstable. In the present configuration the cavity is equipped with holes for air cooling, although they were not needed for the power levels studied. A plasma has been operated for long continuous periods of time (18 hours) and the cavity temperature has never exceeded 50°C. This increase in temperature is gradual over the entire period of operation. With extended use, there is no noticeable change in the appearance of the plasma or in the emission intensity from the background and analyte. 31 Plasma Features After ignition, the suspended plasma uniformly fills the discharge tube and extends outside the cavity in a flame-like manner (Figure 2-6). The resulting plasma contains four distinct regions which are distinguishable by color. Two of the regions arise from the electrical breakdown of the gas within the cavity and consist of a blue zone centered in the discharge tube and a pink glow that surrounds it (46). Their emission spectrum is comprised of the ^ First Positive, ^ Second Positive and Nj+ First Negative band systems (46-49). These pink and blue regions have been described as having a high degree of ionizational and vibrational excitation and have been thought to be where local heating of the support gas occurs (49). These afterglows extend through the cavity and slightly beyond it. The other two discharge regions exist principally in the extended tail flame of the discharge and are characterized by green and yellow afterglows. The green afterglow forms a sheath around the tail flame, largely dictates its color, and reportedly results from the excitation of impurities such as water or oxygen (47). Common names applied to this afterglow are "airglow", "aurora", and "air afterglow" (47,50). The fourth zone is the characteristic yellow emission of the Lewis-Rayleigh bands. These bands are not always visible in the‘tail flame but they can be optically detected. The yellow emission was first thought to be sodium etched from the torch walls, although careful inspection proved otherwise. These bands signal the existence of "active nitrogen", so called because of its high degree of chemical reactivity. The length and intensity of each of the four discharge 32 regions is affected by gas purity, flow rate, pressure, electrical power, temperature and also the length by which the discharge tube extends beyond the cavity (48,49). Under the conditions described above for igniting the nitrogen discharge, a "suspended" plasma is formed (37). However, it is possible also to produce an ICP-like annular plasma by changing the flow rate through the central tube and by altering the central tube's axial position in the cavity. This alternative plasma operating mode is not extraordinary for this type of torch (4). However, the annular MINDAP can be sustained without residing on the inner walls of the torch as reported earlier (37). The results presented in this chapter are for a discharge operating in a suspended rather than annular mode because of the flow-dependent sample-introduction system that was used. It should be possible also to introduce sample aerosol into the central tube and to employ an annular plasma if either a flow independent or low-flow nebulizer is employed. Reagents All stock solutions were prepared as outlined by Dean and Rains (51) using distilled-deionized water and reagent-grade acids, metals and salts. solutions. Analytical standards were made by dilution of these stock 33 RESULTS The MINDAP system was optimized for those conditions that produced maximum analyte signal and signal-to-noise ratio. The operational conditions that were evaluated are applied microwave power, gas flow rate through the sample-introduction system (which also sustains the plasma), optical viewing geometry and spatial position in the plasma. Viewing Geometry The first parameter to be optimized was the viewing geometry of the plasma. Spectral scans were obtained for both radial (side-on) and axial (end-on) viewing arrangements (Fig. 2-4) for three different wavelength regions and with 10 yg/mL of several analytes present in each: (Li,K). 190 - 360 nm (Zn,Mg,Cu); 390 - 465 nm (Ca,Sr); and 650 - 800 nm In these scans (see Fig. 2-7 to 2-9) the most intense peak represents full-scale deflection of the readout system. In the radial configuration, the emission from the first 6 mm of the plasma tailflame was optically detected. Operating conditions for the recording of these spectra were an applied microwave power of 250 W, reflected power of 20 W, 1.75 L/min of nitrogen gas sustaining the plasma and 0.03 L/min nitrogen through the central channel of the torch. Spectra for the two viewing geometries are similar in shape over the region 190 - 360 nm (Fig. 2-7) but the respective intensities are different: the axially obtained spectrum was approximately five times more intense than that in the radial viewing configuration. Emission from Zn, Mg and Cu can be seen, but only with difficulty above a background consisting 34 Figure 2-7: Comparison of optical viewing geometries for the spectral region 190 - 360 nm in the MINDAP plasma. Analyte aerosol introduced into the plasma with desolvation was 10 yg/ml of Cu, Mg, and Zn. Applied power was 250 W with a nebulizer flow rate of 1.75 L/min. radially. B. MINDAP viewed axially. A. Plasma viewed » V' 190 Zn I Zn I 230 PIacma 270 Mg II CnnO Mgl Ax 1a I Iy Mg II Wavelength V Iowed O)- Mg I 310 Cu I Cu I 350 tn — Cu I Cu I LO Ul 36 Figure 2-8: Comparison of optical viewing geometries for the spectral region 390 - 465 nm in the MINDAP plasma. Analyte aerosol introduced into the plasma with desolvation was 10 yg/ml of Ca and Sr. Conditions as in Fig. 2-7. Plasma viewed radially. B. MINDAP viewed axially. A. > 390 U) (0—1 G> Ca II Ca II Ca II Ca II 405 cn Sr II Sr I " Wavelength 420 S X - L L Ca I Sr II Ca I 435 Cnm) 450 Sr I Sr I 465 38 Figure 2-9: Comparison of optical viewing geometries for the spectral region of 650 - 800 nm in the MINDAP plasma. Analyte aerosol introduced into the plasma with desolvation was 10 pg/ml of Li, and K. Conditions as in Fig. 2-6. Plasma viewed radially. B. MINDAP viewed axially. A. 39 •H ►J A. Plasma Vlawad Radially r” 650 1 680 710 740 1- 1--- 1--- 1 770 800 Wavelength CnnO H i-l ►4 B. PIamma V iewed Ax Ia IIy •*< 650 680 710 740 Wavelength CnnO 770 803 40 mostly of the NO, OH, NH, and N^ Second Positive band systems. Both zinc and magnesium are in spectral regions containing intense bands from NO and OH. Although copper emission occurs within the OH band, the background level is relatively low at those wavelengths. In the region between 390 - 465 nm, there is a distinct difference between optical viewing geometries (Figure 2-8). During axial viewing, intense band emission from the N^+ First Negative and N^ Second Positive systems dominates the spectrum and interferes with analyte emission. These bands are produced during ionization of nitrogen inside the cavity; the strongest emission is from the N^+ (0,0) bandhead at 391.4 nm. Interestingly, the amount of energy required to ionize and excite this nitrogen system is 18.74 eV (52). Only two monatomic gases that have been used to sustain MIPs have higher ionization energies: helium (24.48 eV) and neon (21.56 eV). The labels in Figure 2-8 indicate only the approximate position where each element should emit; specific lines could not be confidently located above the background. In comparison, the radially viewed plasma shows no significant contribution from the First Negative system for on-scale measurements of analyte lines at the same solution concentration. Viewed in this manner, the MINDAP produces a high signal level and signal-to-noise ratio. In the region from 650 - 800 nm (Fig. 2-9), the axially obtained spectrum contains relatively weak band emission from the N^ First Positive system, which interferes with the lithium (670.8 nm) and potassium (766.5 and 769.9 nm) lines. In contrast, radial viewing of the plasma shows no significant interfering bands. 41 The superiority of side-on viewing is clearly evident from the spectral scans from 390 nm to 800 nm (Figs. 2-8, 2-9). Table 2-2 lists all molecular bands that spectrally interfere with analyte emission in either viewing arrangement. The most troublesome bands that affect the side-on geometry will be those in the ultraviolet spectral region (190 - 360 nm). Applied Power and Nitrogen Flow The operating conditions under which the plasma is sustained have a marked effect on the length of the discharge tail-flame. An increase in applied power had the most noticeable effect (Fig. 2-10). This behavior can be explained by an increase in the excited-state population of nitrogen which occurs at higher powers. This increased population would excite to a greater extent atmospheric constituents beyond the cavity and thereby extend the length of the air afterglow. Although Figure 2-10 was obtained when no sample aerosol was being sent into the discharge, the presence of the aerosol does not appreciably affect the tail-flame length. Also, varying the nitrogen flow rate does not affect the plasma length as greatly as does applied power. The effects of applied microwave power and nitrogen flow rate on iron atom emission at 371.9 nm are shown in Figure 2-11. At all flow, rates, the background-corrected signal increases with power and reaches a plateau around 225 W. However, at a flow rate of 2.1 L/min of nitrogen, the signal intensity is only half that at lower flows, presumably because of the shorter analyte residence time in the cavity, a decrease in efficiency of the nebulizer, a change in the 42 Table 2-2: aRef. 53 Interfering background band emission. Element Wavelength (nm) Zn (I) 213.8 NO System Mg (II) 279.6 OH System Mg (I) 285.2 OH System Cu (I) 324.7 OH and NH Systems Ca (II) 393.7 N^+ First Negative N^ Second Positive Ca (I) 422.7 N^+ First Negative N^ Second Positive Sr (II) 407.8 N£+ First Negative N£ Second Positive Sr (I) 460.7 N 2 + First Negative N 2 Second Positive Li (I) 670.8 N 2 First Positive K 766.5 N 2 First Positive (I) Interfering species3 43 Figure 2-10: Effect of applied microwave power on plasma tail-flame length (distance from cavity) with a flow rate of 1.75 L/min and 30 ml/min of nitrogen in the support and central-channel torch ports, respectively. Ccm) Length Tail-FI am© 1 S 0 200 AppIied Power CVO 250 300 45 Figure 2-11: Effect of applied power and nitrogen flow rate on iron atom emission at 371.9 nm. All measurements are background- corrected. * - 1.35 L/min nitrogen gas. A - 1.75 L/min nitrogen gas. o-2.1 L/min nitrogen gas. 20 00 150 200 A p p 1ied Power CVO 250 300 47 concentration of exciting nitrogen species and a cooling of the plasma. These results indicate that the maximum signal occurs at a flow rate of 1.75 L/min for this type of sample-introduction system. Other elements that were studied showed the same trends apparent in Figure 2-11. Intensity measurements were made for several analyte atom and ion transitions at applied power levels between 120 - 250 W and a nitrogen flow rate of 1.75 L/min. For the atom lines (Fig. 2-12) as with Fe (I) (above), signals increase to a plateau with increasing power. The power level at which each element reaches a plateau appears to be a function of the element's excitation potential: sodium (excitation energy = 2.08 eV) reaches a plateau near 170 W, whereas calcium (2.92 eV) levels off at 220 W. Magnesium (4.33 eV) does not reach a plateau within the power range of the present system, although a plateau power of 300 W would be expected if the change with excitation energy is linear. These results are similar to those observed with a 7 MHz ICP with nitrogen as the coolant gas (26). In contrast, ion emission (Fig 2-13) does not appear to level off at applied powers within the tested range. Moreover, ion-line intensity increases rapidly only at input powers above approximately 170 W. This behavior indicates that the MINDAP should be operated at power levels greater than 170 W for analytical measurements, especially if ion lines are to be determined. Signal-to-background-noise ratios were measured for power levels of 120 - 250 W and are shown for atom and ion lines in Figure 2-14. With the largest observable signal-to-background-noise ratio and signal intensity occurring at 250 W for both atom and ion lines, the optimum 48 Figure 2-12: Effect of applied microwave power on emission intensity of several atom transitions in the MINDAP. not drawn to the same scale. 1.75 Curves are Nitrogen flow rates were L/min and 30 ml/min as the support and central-channel gases respectively, o - 10pg/ml Ca (I) 422.67 nm * - 10yg/ml Mg (I) 285.21 nm + - 10yg/ml Na (I) 588.99 nm Intensity 0 Line 0. Atom 0. 0. 1 0 0 200 Applied Power CUD 250 300 50 Figure 2-13: Effect of applied microwave power on the intensity of two ion transitions in the MINDAP. the same scale. Curves are not drawn to The nitrogen flow rates were 1.75 L/min and 30 ml/min as the support and central-channel gases respectively. o - 10 yg/ml Ca (II) 393.36 nm * - 10 yg/ml Mg (II) 280.27 nm Ion Line I n t e n s i t y — © © © cn AppIi ed © Power i\) © © CVO I\) 01 © 15 <S> © <S> <s> l\) -k 0) 00 52 Figure 2-14: Effect of applied microwave power oh the emission signal-to-background noise ratio in the MINDAP. o - 10yg/ml Ca (I) 422.67 nm * - 10yg/ml Mg (II) 279.55 nm + - 10yg/ml Na (I) 588.99 nm 5000 Background N oise 6000 ignaI / 2000 1000 GO 1 0 0 150 A p p I 250 200 i e d P o w e r ( W ) 300 54 operating conditions for elemental determinations using this system are: a suspended plasma viewed in the side-on configuration with an applied power of 250 W and a nitrogen flow rate of 1.75 L/min. Viewing Position in Plasma Spectral scans from 190 - 350 nm were acquired at the optimum power and flow rate at three different vertical viewing positions in the plasma tail flame: (Figure 2-15). 0 mm (top of the cavity), 5 mm, and 15 mm At 0 mm, there is strong background emission from the NO, NH, Ng, and OH band systems, as well as from the analyte introduced. At 5 mm, no change in spectral features is apparent except that the overall intensity has diminished by a factor of five. At 15 mm, the OH band becomes the dominant spectral feature and interferes with the emission of both magnesium and copper. There are basically four changes that occur as the viewing region moves higher in the plasma tail flame: overall emission intensity decreases; NO, NH, N^ band intensity decreases; analyte emission intensity decreases more rapidly than the background band emission; and the OH intensity increases relative to the other components of the spectrum. These changes can be understood by recognizing that all observed species but OH, and therefore their emission bands, originate in the primary discharge region. Moving away from this area and into the afterglow (tail-flame) then reduces the magnitude of each band. In contrast, OH bands arise principally from the interaction of the plasma with atmospheric water vapor (the sample aerosol has been desolvated). Consequently, OH emission is maximal in the upper region of the tail 55 Figure 2-15: Spectral scans of plasma background and analyte emission for different vertical positions in the plasma tail flame. Analyte aerosol introduced into the plasma with desolvation was 10 pg/mL of Zn, Cu, and Mg. viewed in a side-on fashion (Fig. 2-4). C are not on the same scale. A. 0.0 mm B. 5.0 mm C. 15.0 mm (top of the cavity) Plasma Scans A, B, and 190 Haight 5 .0 230 270 Uav«i«nflbh CnnO <3 Mg II Mg II Mg I 310 s Cu I Cu I Cu I Cu I Cu I 350 Ln O' 57 flame. In the studies described later in this thesis, the integrated emission was collected from the first 6 mm of the tail-flame in order to maximize signal-to-background ratio and to smooth out any fluctuations caused by tail-flame waver. CONCLUSION The MINDAP has many characteristics that are attractive in an emission source for elemental analysis. The gas-flow pattern is stable and centered by a torch similar in design to the ICP concentric-tube construction. The new plasma is easily lit and aerosol sample introduction is conveniently accomplished with a pneumatic nebulizer. The optimized microwave power and gas-flow rates used to sustain the plasma are slightly higher than those generally employed in argon and helium MIPs, but are far lower than required for the ICP. Impedance matching of the power supply to the cavity for these high powers was easily attained with the added dielectric from the modified torch and tuning stubs. The appearance of this new plasma is unusual compared to other atmospheric-pressure MIPs. The nitrogen discharge extends outside the cavity to lengths determined by the operational conditions. For example, under the recommended applied power of 250 W and a nitrogen flow rate of 1.75 L/min, the discharge extends 10 cm beyond the cavity. This added discharge length provides longer analyte residence times in the plasma and the consequent possibility of more complete decomposition of the introduced analyte. This long tail-flame also offers an alternative viewing configuration for analyte emission. The 58 plasma can be viewed either axially (as is common with most MIP systems) or radially (as with an ICP system). Based on background emission spectra, the side-on arrangement was judged superior. The potential of this system for chemical analysis is readily seen in Figure 2-16. Under the optimum operating and optical viewing conditions, 10 ng/ml of lithium and 100 ng/ml of potassium are easily detected. The zero (0) level marker indicates the magnitude of the dark-current contribution from the PMT. This spectrum shows also that there is essentially no background continuum emission to affect measurements at these wavelengths. In the following chapter the MINDAP system will be critically evaluated from its analytical figure-of-merits as a new, inexpensive, compact, sensitive and versatile emission source for elemental analysis. 59 Figure 2-16: Emission spectrum from 650 - 785 nm of the MINDAP tail flame viewed radially. Analyte aerosol introduced into the plasma with desolvation was 10 ng/ml of Li and 100 ng/ml of potassium. Applied power was 250 W with a nebulizer flow rate of 1.75 L/min. discussion of zero (0) marker. See text for 0 WwUiyA h i T r r f i i i i i ii "f ~ i 660 680 700 720 i r i~ ii ~ ~ t~ riri 740 760 780 i i W a velength Cnm) 61 References 1. S. W. Downey and G. M. Hieftje, Anal. Chim. Acta, 141, 193 (1982). 2. C. A. Bache and D. J. Lisk, Anal. Chem., 43, 950 (1971). 3. A. J. McCormack, S. C. Tong, and W. D. Cooke, Anal. Chem., 1470 (1965). 4. W. R. McLean, D. L. Stanton and G. E. Penketh, Analyst, 98, 432 (1973). 5. C. I. M. Beenakker, Spectrochim. Acta, 32B, 173 (1977). 6. H. A. Dingjan and H. J. Dejong, Spectrochim. Acta, 36B, 325 (1981). 7. L. R. Layman and G. M. Hieftje, Anal. Chem., 47, 194 (1975). 8. A. T. Zander and G. M. Hieftje, Anal. Chem., 50, 1257 (1978). 9. F. L. Fricke, 0. Rose, Jr., and J. A. Caruso, Talanta, 23, 317 (1975). 37, 10. J. H. Runnels and J. H. Gibson, Anal. Chem., 39, 1399 (1967). 11. W. B. Robbins, J. A. Caruso and F. L. Fricke, Analyst, 104, 35 (1979). 12. P. Barett and T. R. Copeland, "Applications of Plasma Emission Spectroscopy", p.139, R. M. Barnes ed., Heyden Publishers, 1979. 13. T. Ishizuka and Y. Uwamino, Anal. Chem., 51_t 125 (1980). 14. I. S. Krull and S. Jordan, Amer. Lab., 12^, 21 (1980). 15. A. T. Zander and G. M. Hieftje, Appl. Spectrosc., 35, 357 (1981). 16. R. N. Savage and G. M. Hieftje, Anal. Chem., 51, 408 (1979). 17. A. D. Weiss, R. N. Savage and G. M. Hieftje, Anal. Chim. Acta, 124, 245 (1981). 18. R. Rezaaiyaan, G. M. Hieftje, H. Anderson, H. Kaiser and B. Meddings, Appl. Spectrosc., 36, 627 (1982). 19. L. Ebdon, D. J. Mowthorpe and M. R. Cave, Anal. Chim. Acta, 115, 171 (1980). 20. B. Capelle, J. M. Mermet and J. Robin, submitted to Appl. Spectrosc., 1982. 62 21. H. U. Eckert, F. L. Kelly and H. N. Olsen, J. Appl. Phys., 39, 1846 (1968). 22. S. Greenfield and P. B. Smith, Anal. Chim. Acta, 59, 341 (1972). 23. H. P. Freeman andJ.D.Chase, J. Appl. Phys., 39^, 180 (1968). 24. A. Montaser and J. Mortazavi, Anal. Chem., 52, 255 (1980). 25. A. Montaser, 292 (1981). 26. S. Greenfield and D. T. Burns, Anal. Chim. Acta, 113, 205 (1980). 27. R. M. Barnes and G. A. Meyers, Anal. Chem., 521, 1523 (1980). 28. M. Capitelli, F. Cramarossa, L. Triolo, and E. Molinari, Combust. Flame, 15* 23 (1970). V. A. Fassel and J. Zalewski, Appl. Spectrosc., 35, 29. R. M. Barnes and S. Nikdel, Appl. Spectrosc., _30_, 310 (1976). 30. J. D. Cobine and D. A. Wilbur, J. Appl. Phys., 22^, 835 (1951). 31. A. N. Wright and C. A. Winkler, "Active Nitrogen", Academic Press, N. Y., 1968. 32. A. P. D'Silva, G. W. Rice and V. A. Fassel, Appl. Spectrosc., 34, 578 (1980). 33. W. B. Dodge III, andR.0. Allen, Anal. Chem., 53, 1279 (1981). 34. G. A. Capelle (1978). andD.G. Sutton, Rev. Sci. Instrum., 49, 1124 35. C. I. M. Beenakker, B. Bosman, and P. W. J. M. Boumans, Spectrochim. Acta, 33B, 373 (1978). 36. C. I. M. Beenakker, Spectrochim. Acta, 3IB, 483 (1976). 37. A. Bollo-Kamara and E. G. Codding, Spectrochim. Acta, 36B, 973 (1981). 38. J. E. Meinhard, ICP Inf. Newsl. 2., 163 (1976). 39. C. Veillon and M. Margoshes, Spectrochim. Acta, 23B, 553 (1968). 40. B. L. Sharp, Sel. Ann. Rev. Anal. Sci., j4, 37 (1976). 41. C. B. Boss, personal communication, 1983. 42. D. L. Haas, J. W. Carnahan, and J. A. Caruso, Appl. Spectrosc., 37., 82 (1983). 43. J. P. J. van Dalen, P. A. DeLezenne Coulander, and L. de Galan, Spectrochim. Acta, 33B, 545 (1978). 63 44. T. B. Reed, J. Appl. Phys., 32, 2534 (1961). 45. E. Sexton, R. N. Savage and G. M. Hieftje, Appl. Spectrosc., 33, 643 (1979). 46. U. H. Kurzweg and H. P. Broida, J. Mol. Spec., _3, 388 (1959). 47. U. H. Kurzweg, A. M. Bass and H. P. Broida, J. Mol. Spec., _1, 184 (1957). 48. G. E. Beale Jr. and H. P. Broida, J. Chem. Phys., 31, 1030 (1959). 49. J. Kishman, E. Barish and R. Allen, Preprint 1982. 50. J. W. Chamberlain, "Physics of the Aurora and Airglows", Academic Press, New York, 1961. 51. J. A. Dean and T. C. Rains, "Flame Emission and Atomic Absorption Spectrometry", Vol. 2, M. Decker, New York, 1971, Ch. 13. 52. J. P. Robin, Proc. Analyt. Atom. Spectrosc., _5, 79 (1982). 53. R. W. B. Pearse and A. G. Gaydon, "The Identification of Molecular Spectra", John Wiley and Sons, Inc., New York, 1976. 64 CHAPTER 3 Analytical Characterization of the Microwave Induced Nitrogen Discharge at Atmospheric Pressure (MINDAP) In Chapter 2 the microwave induced nitrogen discharge at atmospheric pressure (MINDAP) was shown to possess many attractive features for atomic emission spectrometry. The initial expense of a microwave power supply, impedance matching devices (tuning stubs), cavity and cables are quite low (ca. $5000). Also, because the plasma is sustained using nitrogen at less than 2 L/min, the operating costs are less than with either helium or argon which are respectively 2 and 3 times more expensive. The instrumentation is compact because of the low microwave power required ( 500 W) and the small size of a resonant cavity operated at a frequency of 2450 MHz. Importantly, the MINDAP extends beyond the cavity to a length of 10 cm when the input microwave power is 250 W. This length enables side-on viewing (similar to the inductively coupled plasma) and increases the analyte residence time within the plasma. Side-on viewing reduces strongly the background emission from the discharge whereas a long analyte residence time provides more complete decomposition of an introduced aerosol sample, and enables more efficient signal collection. Sample introduction is straightforward in the MINDAP system, unlike that in most other microwave plasmas. Nitrogen has good nebulization qualities and the flow rate used to sustain the plasma is compatible with most common nebulizers. Moreover, the applied power is slightly higher than ordinarily used with other MIPs, making the MINDAP more tolerant of water and large sample volumes. '65 In this chapter the MINDAP is critically evaluated for use as an atomic emission source for analytical measurements. Results show that the plasma exhibits detection limits (ng/mL) comparable to those of other more elaborate and costly emission sources. Calibration curves are linear over a range of 3 to 5 orders of magnitude, typical of most plasma sources. In contrast,, concomitant interferences are similar to those observed in flame spectrometry, and can be overcome in much the same way with ionization suppressants 'and releasing agents. A comparison with other emission sources is made and the analytical potential of the MINDAP assessed. EXPERIMENTAL The data collection and operating conditions for the MINDAP systemare identical to those described in Chapter 2. The detection limit is defined here as the concentration of analyte that produces a background-corrected signal that is 1.96 (95% confidence interval) times the standard deviation of the blank signal. The mean signal and background levels were obtained from fifteen consecutive integrations. 1 -s The limits of detection were obtained by extrapolation from analyte solution concentrations of at least an order of magnitude greater than those calculated. 66 RESULTS AND DISCUSSION Detection Limits Detection limits for sixteen elements were determined at thirty-three different wavelengths and are compared in Table 3-1 with those obtained from other microwave and inductively coupled plasmas. Overall, the MINDAP provides sensitivity comparable to that of competitive sources. The lowest MINDAP detection limits arise from atomic transitions; in fact, only five atom lines are more sensitive in the argon MIP and ICP: Cd, Cu, Mg, Pb, and Zn. In contrast, ion-line detection limits in the MINDAP are ordinarily worse than those of the other techniques listed. Ion lines such as those from Cd(II) and Fe(II) were superimposed on strong background band emission from the NO system, making their detection even more difficult. Detection limits for the easily ionizable elements were among the best found, perhaps because the MINDAP is a more nearly thermal source than other plasmas. From Table 3-1, it is apparent that spectral lines of lower excitation potential exhibit better detection limits than those of higher potential. This feature suggests that analyte excitation in the MINDAP occurs partially by a thermal mechanism, similar to that in a chemical flame. If so, and if the MINDAP is in local thermodynamic equilibrium (LTE), the excited-state population would follow a Boltzmann distribution. No other microwave plasma yet reported has been shown to be in LTE, although Busch and Vickers (1) postulated that plasmas at atmospheric pressure should attain LTE because of their high particle density. 67 Table 3-1: Detection Limit Comparison for Atomic Emission Atmospheric-Pressure Plasma Sources (ng/mL) Element Wavelength (nm) MINDAP Ar-ICPa A1 (I) Ba (II) (II) (I) (II) Ca (II) (II) (I) Cd (II) (I) Cu (I) (I) Fe (II) (I) (I) K (I) (I) Li (I) Mg (II) (II) (I) Na (I) Ni (I) (I) Pb (I) (I) Pd (I) (I) Sr (II) (II) (I) V (I) Zn (I) 396.15 455.40 493.40 553.55 614.17 393.36 396.85 422.67 214.44 228.80 324.75 327.39 238.20 371.99 373.48 766.49 769.89 670.78 279.55 280.27 285.21 588.99 341.47 352.45 368.35 405.78 340.45 363.47 407.77 421.55 460.73 437.92 213.85 13. 15. 40. 19. 99. 2.3 7.5 19. 0.87 1.5 61. f a b c d e f g h i j Ref. Ref. Ref. Ref. Ref. Ref. Ref. Ref. Ref. Ref. 2, unless otherwise indicated 3 4 5 6 7 8 9 10 11 0.13 0.33 6.7 1.7 1.9 3.6 6.5 3.1 73. f 32. 30. b 1 . 2 410. 28. 4.4 2.3 280. 36. 98. 5.4 6.5 0 . 2 2 13. 1 1 . 1 2 . 0.29 1 0 . 5.3 84. 91. 45. 24. 16. 2 1 . 3.1 47. 1 2 0 . . Ar-MIP 60. c 1 0 . i e 0 . 6 . 1 . 9. 1 0 . 400. 2 0 1.9 1 0 . 1 0 0 . 8 h e j e j - j e 0 . 2 0 1 . 1 19. 32. 30. 230. 180. 29. 36. 0.28 0.51 45. 77. f 1 . 2 5. j 80. j . j 0 . 8 e 1 . 0 g 2 0 0 5. j 1 0 . j 80. d 0.5 g 68 Working Curves Calibration curves for four elements, shown in Figure 3-1, are linear over three-to-five orders of magnitude and are representative of those obtained for other elements. Working curves were extended to 100 ng/mL, which is between two and three orders of magnitude above the detection limits. The dynamic range is expected to extend at least another two orders of magnitude to the detection limit. Similarly, working curves have been extended to 5000 yg/mL, above which they begin to bend downward. This rolloff at higher concentrations is presumably caused by a decrease in nebulizer efficiency. At such high salt concentrations, memory effects also were noted from analyte residue on the inside of the quartz torch and the desolvation apparatus. For concentrations of 1000 yg/mL or less, no such effects were encountered. Precision Relative standard deviations were determined for a series of ten consecutive measurements of analyte signal and background at concen trations two orders of magnitude and greater above the detection limit. The %RSD values remained constant for all concentrations and were 0.3% for the blank and between 0.5% - 1.8% for the analyte; this precision is among the best obtained for MIP systems (11) used for the analysis of aerosol samples. 69 Figure 3-1: Analytical working curves for the MINDAP. o - Cu (I) 324.75 nm * - K (I) 766.49 nm + - Sr (I) 460.73 nm □ - Ca (II) 393.36 nm Re Iat ive Intens ity IQ 10* 0 .1 10 100 1000 10000 **sj o Concentration Cjug/m I) 71 Interferences Classical interferents are those such as phosphate or aluminum that depress analyte emission or those such as the alkali metals that enhance analyte signals. These types of interferences are well documented in flame spectrometry literature (12-14). Unfortunately, interference effects in the microwave plasma are not as consistent or well understood as those in flame work. Skogerboe and Coleman (6 ) and Lichte and Skogerboe (5) concluded that interfer ences caused by the presence of easily ionizable species (e.g., sodium) were not present in their plasma. In contrast, many other workers have documented strong interferences caused by sodium (10,11,15-19). There seems to be a general agreement, however, on the depressing effect of phosphate on calcium emission from a MIF. In assessing the analytical potential of the MINDAP, interelement interferences were examined that involve vaporization (Al, P0^) and ionization (Na). Figure 3-2 shows that both calcium atom and ion emission are enhanced as sodium concentration increases to 1%. This behavior is quite different from that observed in flames and is more like that associated with an ICP or MIP. In flame studies, the increased concentration of electrons caused by sodium addition is known to shift the calcium ionization equilibrium, enhancing the atom line and decreasing the ion line. £n contrast, the effect of sodium in other plasmas is to change excitation behavior and thereby enhance both atom and ion emission. Although a non-thermal mechanism would therfore seem to be operating in the MINDAP tail-flame, the behavior might also be caused by a change in the plasma dimensions. When sodium is added 72 Figure 3-2: Sodium interference on 10 yg/ml calcium atom emission at 422.67 nm. 0.5 Emission 0.4 0.3 Ca Intensity 0.6 0. 1 0.2 0 0 10 0 0 2000 3000 4000 5000 6000 lo Sodium Concentration Cjug/m L) 74 to the MINDAP, it appears wider, much as has been observed in other discharges (20,21). A wider plasma would fill the discharge tube more completely and thereby increase the interaction of the analyte with the plasma. This situation increases atomization efficiency and produces a net increase in the emission signal. In practice the effect of sodium or another easily ionized element can be reduced or eliminated by the addition to both standard and blank solutions of an excess amount of another easily ionizable species. ability of 1 0 0 0 The yg/ml cesium to reduce the ionization interference is shown in Figure 3-3. Even at this high cesium concentration, the effect of sodium is still apparent, although less than when cesium is absent. At a concentration of 2500 yg/mL cesium, the enhancement was able to be eliminated (see Fig. 3-3). Aluminum and phosphate both depress calcium emission signals as shown in Figures 3-4 and 3-5. Although not presented in Figure 3-5, the effect of phosphate on calcium emission intensity is constant to a phosphate/calcium molar ratio of 60. The behavior in Figures 3-4 and 3-5 is similar to that often observed in flames and other MIPs. The depression is probably caused by the formation of refractory phosphatecalcium and aluminum-calcium species, which reduce the rate at which free calcium can vaporize (12-14). Suppression of the formation of those refractory materials is ordinarily accomplished by the addition of a "releasing agent" such as EDTA or lanthanum (14,22,23). plotted in Figures 3-6 and 3-7, the interference can be the addition of EDTA as the ammonium salt. As eliminated by 75 Figure 3-3: Sodium interference on 10 yg/ml calcium emission with 1 0 0 0 yg/ml of cesium added as an ionization suppressant. Dashed line represents the addition of 2500 yg/mL of cesium. * - Ca (I) 422.67 nm o - Ca(II) 393.36 nm > -4-> X U (0 C <D C H 0.15 c 0 » 0.1 (0 E LU 200 Sodium 400 600 Concentration 800 Cjug/mLD 1000 77 Figure 3-4 Aluminum interference on 10 yg/ml calcium emission, o - Ca (I) 422.67 nm * - Ca(II) 393.36 nm ity .2 Re Iat ive .6 Intens 2 . 8 .4 0 0 200 400 600 800 1000 Co A I Concentration (jug/mL) 79 Figure 3-5 Phosphate interference on 10 yg/ml calcium emission. * - Ca (I) 422.67 nm o - Ca(ll) 393.36 nm Re Iat ive Intens ity 2 5 5 0 oo o P04 / Ca Molar Ratio 81 Figure 3-6: Overcoming phosphate interference on 10 pg/ml calcium emission by the addition of 0.01M EDTA as a releasing agent. Phosphate added as H 3 PO4 and EDTA as the NH^ salt. * - Ca (I) 422.7 nm o - C a (II) 393.36 nm Relative P04 / Ca Molar Ratio Z8 Intensity 83 Figure 3-7: Use of 0.01M EDTA to overcome the interference of aluminum on 1 0 yg/ml calcium emission. * - Ca (I) 422.67 nm o - Ca (II) 393.36 nm 2 Calcium Emission Intensity 3 0 250 500 750 Aluminum Concentration C^ig/mL!) 1000 85 CONCLUSION The MINDAP is a promising tool for atomic-emission analysis of solutions. low cost, sensitive The plasma is easy to operate, adapts conveniently to common nebulizer systems and uses inexpensive nitrogen to operate. Detection limits are low for most elements and comparable to those offered by other plasma systems. Working curves are linear over three-to-five'orders of magnitude and the available precision is excellent (0.5% - 1.8%). Interelement interferences exist but can be eliminated with the aid of ionization suppressants and releasing agents as in flame spectrometry. The MINDAP has many features that are more characteristic of flames than plasmas. Atom line emission is more intense than that of ions, elements of lower excitation potential have better detection limits, and similar effects from concomitants are observed. results are interesting because the current MIP do not include a thermal mechanism. These theories of excitation in a In the following chapter information concerning the plasma's physical characteristics are described and used to explain the MINDAP's successful operation with an aerosol sample-introduction system. 86 REFERENCES 1. K. W. Busch and T. J. Vickers, Spectrochim. Acta, 28B, 85 (1973). 2. P. W. J. M. Boumans, 3. P. W. J. M. Boumans and R. M. Barnes, ICP Inf. Newsl., 3^ 445 (1978). 4. R. K. Skogerboe and G. N. Coleman, Anal. Chem., 48, 611A (1976). 5. F. E. Lichte and R. K. Skogerboe, Anal. Chem., 45, 399 (1973). 6 . Spectrochim. Acta, 36B, 169 (1981). R. K. Skogerboe and G. N. Coleman, Appl. Spectrosc., 30, 504 (1976). 7. N. Furuta, C. W. McLeod, H. Haraguchi, and K. Fuwa, Appl. Spectrosc., 34, 211 (1980). 8 . R. K. Skogerboe, D. L. Dick, D. A. Pavlica, and F. E. Lichte, Anal. Chem., 47, 568 (1975). 9. K. Fallgatter, V. Svoboda, and J. D. Winefordner, Appl. Spectrosc., 25, 347 (1971). 10. H. Kawaguchi, M. Hasegawa, and A. Mizuike, Spectrochim. Acta, 27B, 205 (1972). 11. C. I. M. Beenakker, B. Bosman and P. W. J. M. Boumans, Spectrochim. Acta, 33B, 373 (1978). 12. C. Th. J. Alkemade and R. Herrmann, "Fundamentals of Analytical Flame Spectroscopy", Ch. 9, John Wiley and Sons, New York, 1979. 13. R. Mavrodineanu and H. Boiteux, "Flame Spectroscopy", Ch. 12, John Wiley and Sons, New York, 1965. 14. D. G. Peters, J. M. Hayes and G. M. Hieftje, "Chemical Separations and Measurements", Ch. 20, W. B. Saunders Co., Philadelphia, 1965. 15. C. Veillon and M. Margoshes, Spectrochim. Acta, 23B, 503 (1969). 16. G. Pforr and K. Langner, Z. Chem., 5_, 115 (1965). 17. G. Pforr and V. Kapicha, Collect. Czech. Chem. Coramun., 31,4710 (1969). 18. S. Murayama, Spectrochim. Acta, 25B, 553 (1969). 19. I. Kleinmann and V. Svoboda, Anal. Chem., 41^, 1029 (1969). 20. W. E. Rippetoe, E. R. Johnson and T. J. Vickers, Anal. Chem., 436 (1975). 47, 87 21. P. W. J. M. Boumans, "Theory of Spectrochemical Excitation", Hilger and Watts, London, 1966. 22. A. C. West and W. D. Cooke, Anal. Chem., 32, 1471 (1960). 23. J. I. Dinnin, Anal. Chem., 32, 1475 (1960). 88 CHAPTER 4 Physical Measurements of the Microwave Induced Nitrogen Discharge at Atmospheric Pressure In the previous two chapters several interesting observations were made concerning the MINDAP's excitation ability. In Chapter 2, a direct correlation was observed between the excitation potential of an element and the applied microwave power at which emission intensity peaked. Similarly, detection limits are found to be better for elements of low excitation potential and Chapter 3 indicated that atomic lines are more sensitive than their respective ion lines. These conditions suggest that a "thermal" mechanism is possibly responsible for analyte excitation in the MINDAP. In this chapter physical measurements were made in an attempt to explain the phenomena by which analyte is decomposed and excited in the MINDAP. As stated previously, one of the problems encountered during the analysis of solutions in a MIP is its low thermal energy. From results obtained with the nitrogen ICP, it would be anticipated that the MINDAP should be closer to LTE than a conventional MIP. Evidence to support this supposition can be found in analyte excitation behavior discussed in previous chapters. In the present chapter, a quantitative determination of several temperatures in the MINDAP is undertaken. There are two regions of the plasma that are of particular importance for spectroscopic measurements: the primary discharge zone where decomposition of the sample occurs and the tail-flame where analyte excitation and emission are observed. In the discharge region, thermal (rotational) temperatures were measured for various operating 89 conditions and found to be between 4500 and 5000°K. The tail-flame also was extensively studied for its excitation ability using spectroscopic temperatures, ion/atom ratios and spatially resolved emission profiles as figures-of-merit. A tail-flame excitation temperature of approximately 6000°K was measured, whereas ion-to-atom ratios for magnesium lines were typically less than 0.5. The spatial profiles reveal that temperatures decrease as the distance from the discharge region increases. Similarly, analyte emission profiles appear to depend on an element's excitation potential. A comparison with temperatures of other microwave and inductively coupled plasmas is also presented. TEMPERATURES AND THERMODYNAMIC EQUILIBRIUM IN LABORATORY SOURCES For a system to be in thermodynamic equilibrium, it must meet several criteria: the velocity distribution of all particles in the system must follow that dictated by Maxwell's equations; the population of excited states must conform to a Boltzmann distribution; the atom-ion equilibria must behave according to the Saha-Eggert equation; and the distribution of the electromagnetic radiation must be in agreement with Planck's law (1). Simply stated, a system is in thermodynamic equilibrium if it can be described by a single temperature. "Temperature" is in this sense a descriptive term that relates to the magnitude of each type of energy associated with the source. Various temperatures describe the distribution of kinetic energy of the electrons (electron temperature), of atoms and molecules (translational, gas, or kinetic temperature), and the population of 90 atoms in different energy states (excitation or spectroscopic temperature). Ordinarily, it is assumed that lower-energy excitation temperatures (rotational,vibrational) are the same as that which governs the translation of large particles such as atoms and ions. Most laboratory flames and plasmas are not in complete thermodynamic equilibrium for a number of reasons. One is that radiation emitted by a source cannot be adequately described by Planck's law because the source is transparent over large spectral regions and, at atmospheric pressure, collisionally induced transitions are far more frequent than radiative ones (1,2). Another reason is that thermal gradients are established as the gases leave the primary reaction zone and mix with cooler atmospheric gases. The situation that exists when thermal equilibrium is established in particular spatial locations in the source is termed "local thermodynamic equilibrium" (LTE). Laboratory flames and plasmas are generally described qualitatively and quantitatively in terms of their deviation from LTE. At atmospheric pressure, LTE would be expected to prevail in flames and plasmas because of their high particle density (1,3-5). This suggestion appears to be valid for most chemical flames, but appears not to hold for high-frequency electrical plasmas (MIP and ICP). In flame systems, two different situations exist; in the inner cone (primary reaction zone) LTE seems not to be established but in the secondary combustion zone (hot flame gases), which is a few centimeters above the reaction zone, thermal equilibrium is achieved (6,7). conclusions are derived from localized temperature measurements. the downstream flame gases, temperatures describing excitation, These In 91 ionization, and translation are equal and range between 2600°K to 3300°K for different types of flames and operating conditions. In contrast, no single temperature can be defined in the primary reaction zone, although a rotational temperature of 5600°K for an acetylene-air and acetylene-oxygen flame contrast sharply with the spectroscopic temperature of only 2500°K (6). This disparity suggests that the exchange of rotational energy through collisions is more effective than the exchange between electronic and rotational energies (6). Clearly, deviation from LTE in this region is considerable. The degree to which a particular plasma approaches LTE depends strongly on its physical nature and operating conditions. Microwave induced plasmas (MIP) operating at 2450 MHz on low power and flow rates of either argon or helium gas tend to be compact discharges that remain inside the resonant cavity that supports them. Because the MIP is small, spatially averaged temperatures are usually reported, some of which are listed in Table 4-1-. The inequality of the various tempera tures indicates a substantial departure from LTE. The low thermal temperature explains why analysis of aerosols directly introduced into the MIP is difficult and also why volatilization and atomization problems occur (8). Compared to a MIP, the inductively coupled plasma (ICP) operates at lower frequencies (27.12 and 40.68 MHz), at higher powers and flow rates, and yields a much larger discharge. Although this larger discharge ('vlOO mm in length) enables it to be mapped spatially, few local regions appear to approach LTE. Kalnicky (9), Mermet (10) and Kornblum (11) all agree that under their experimental operating 92 Table 4-1: Temperatures Reported for Several AtmosphericPressure Microwave-Induced Plasmas. __________________ Temperatures (°K)__________________ Support Gas Electronic Excitation Ionization Rotational Ar 5000 (a,b,c) 5000 (a) 2000 (a) He 7250 (d), 8000 (c) -- 2100 (e) a Ref. 12 b Ref. 13 c Ref. 14 d Ref. 15 e Ref. 8 93 conditions the Ar-ICP is not in LTE. In contrast, Weiss (16) has shown through extensive spatial mapping that there are regions in the Ar-ICP where LTE is approached as well as regions that depart from LTE, as claimed by others (9-11). A helium ICP has been sustained and temperature measurements indicate that it too is not in LTE (8 ). A comparison of these temperatures is presented in Table 4-2 for ICPs supported in several gases. Recently, ICPs have been described that use nitrogen as an inexpensive diluent or replacement for the argon support gas (17-19). Temperatures measured for such argon-nitrogen and pure nitrogen plasmas are included in Table 4-2. Argon plasmas were found to have higher excitation temperatures (2 0 ,2 1 ) but nitrogen-doped plasmas decomposed particles more rapidly and efficiently (18,21,22). Similarly, lower ion-to-atom ratios and lower total continuum radiation is reported for the Ar-N^ and ^ plasmas (18). These results suggest that nitrogen introduced into the ICP increases its thermal temperature and results in a plasma that is closer to LTE. EXPERIMENTAL The MINDAP was sustained as a suspended plasma and viewed radially for all measurements reported in this chapter. Except for those involving rotational temperatures, all measurements were made with the spectrometric system discussed in Chapter 2. Rotational temperatures were determined using a high-resolution monochromator (Model HR1000, I.S.A. Inc., Metuchen, N. J.) with a slit width of 10 pm (0.04 A° resolution). 94 Table 4-2: Temperatures Measured in Several AtmosphericPressure Inductively Coupled Plasmas. Temperatures_(°K)________________ Support Gas Ionization Rotational Ar 6000-9000 He 4100 (a) 2400 (a) Ar-N2 4500-5100 5000-7000 (b,c) n2 a Ref. Electronic Excitation 8 b Ref. 20 c Ref. 23 d Ref. 24 e Ref. 11 f Ref. 25 7000-8000 (f) 2100, 4000 (e) 6000-8000 (d) 95 Spatial emission profiles for several elements of different excitation potential were collected in the tail flame of the MINDAP. 1 A -nnn circular aperture was positioned in front of the entrance slit of the Heath monochromator upon which radiation from the plasma was focussed. The spatial resolution for these profiles was 0.5 mm vertically and 50 ym horizontally of the central portion of the tail-flame. The microwave cavity and torch assembly were translated vertically by means of a programmable stepper motor (Denco Inc., Tucson, Ar.); a distance of 5 cm for this travel was sufficient to acquire the entire profile for each element. Data for the temperature measurements were not Abel inverted for radial spatial information. Similarly, all intensity measurements were corrected for the photo metric response of the measurement system. Temperature Measurements and Calculations Equations used to describe the various temperatures of the plasma necessarily assume that energy distribution is thermal and that an observed emission intensity can be predicted from the Boltzmann relation. However, complete LTE is not required for the definition of a particular kind of temperature (e.g., ionization) or for its measurement; it is necessary to assume only that equilibrium is established among the levels described by that temperature. Electronic Excitation Temperature. The spatially integrated excitation temperature of a 50 ym-wide section from the central portion of the first 6 mm of the MINDAP tail flame was determined using iron as 96 a thermometric species. An iron solution of 1000 yg/raL was introduced into the plasma and the relative radiances of six atom lines in the spectral region of 355 - 385 nm were measured. The excitation (spectroscopic) temperatures at different applied microwave powers and nitrogen flow rates through the side-on inlet of the torch were determined from a logarithmic form of the Boltzmann equation (26,27) (eq. 4-1). in where I - V (kIE> (4-1) = relative intensity of specific iron lines = wavelength of the observed iron transition g^ = statistical weight of the upper state A ^ = transition probability k = Boltzmann constant E^ = excitation energy of the excited state TE = electronic excitation temperature A plot of the left hand side of equation 1 against the excitation energy of the excited state (E^) yields a straight line with a slope inversely proportional to the excitation temperature [-l/(kT_)]. fe Table 4-3 summarizes the transitions, statistical weights, transition probabilities and excitation energies used for these measurements. Ionization Temperature. Spatially integrated intensity measurements of magnesium atom (285.4 nm) and ion (279.55 nm) lines were collected with the same spatial position and resolution as used 97 Table 4-3: Spectroscopic data for Electronic Excitation Temperature measurements using Fe (I) emission lines3 . Wavelength (nm) Eki(em_1) 358.120 34844 13 1.03 360.886 35856 5 0.797 371.99 26875 11 0.163 373.487 33695 5 0.886 381.584 38175 7 0.948 382.043 33096 7 0.638 aRef. 2 with excitation temperatures. The ion-atom ratios were calculated from these values and substituted into the Saha-Eggert equation (27) (eq. 4-2), I_ I 4.83 x 1015 n, e A where hi ( 9k Ak1 \ / \ \ 9k Ak1 / A Xk1 /I 5040(V, + V, \ (4-2) ” i°n/atom intensity ratio ng a electron number density * ionization temperature Vj = ionization energy (7.65 eV for Mg) V^j. = excitation energy for the ion line (4.43 eV) a excitation energy for the atom line (4.35 eV) from which ionization temperatures were calculated by an iterative process. Because of the form of the equation, no explict expression for temperature is obtained. Consequently, temperature values are substituted into the Saha-Eggert equation and a resultant ion/atom ratio is obtained. determined one. This ratio is compared to the experimentally The temperature is varied until both experimental and calculated ion/atom ratios are equal. These temperatures were obtained for several applied powers and nitrogen flow rates through both inlet ports of the plasma torch. Electron Number Density. Two methods were employed in an attempt to determine electron number densities in the MINDAP: the series-limit 99 line-merging technique described by Montaser, £ £ (28) and the method involving Stark-broadening of the hydrogen Balmer line described by Griem (29). The line-merging technique used by Montaser and described earlier by Pannekock (30) and then by Inglis and Teller (31) takes advantage of the fact that the principal quantum number of the last discernible line in a series depends on the electron number density according to equation (4-3): log (n) = 23.26 - 7.51og ( n j (4-3) In eq. 4-3 n^ is the principal quantum number at which the merging of the series lines occurs and n is the sum of the electron and ion densities. In an electrically neutral plasma, such as those used for elemental analysis, the electron and ion densities are equivalent, so the electron density is just half the value (n) obtained from eq. 4-3. Experimentally, a solution containing a single element, either Al, Li, or Ca, at a concentration between 2500 ppm and 5000 ppm is introduced into the plasma. The emission spectrum from the transitions (see ref. 28) is then scanned; 2 P - 2 D The principal quantum number from the last discernible transition, where the lines merge into a continuum, is then used for the calculation of electron number density (eq. 4-3). This technique is easy to employ but is criticized because of the difficulty involved inidentifying the last "discernible" line in the series. In addition, transitions must be chosen in spectral regions where plasma background emission will not interfere with them. 100 Electron densities can be obtained also by accurately measuring the full-width at half maximum (FWHM) of the Stark-broadened H p emission line at 481.6 nm. After deconvolution of the line profile, to account for instrumental, collisional and Doppler broadening, the electron density can be related to the FWHM by eq.(4-4): «e - C < v V where AAg <*-*> S )3/2 = FWHM of the Hg line C(ne>Te) = coefficient that depends on temperature and electron density These C values are compiled in Griem's book (29) for different values of electron temperature and number density. In practice, a spectral scan of the H line is all that is P necessary to determine the FWHM of the line because of its considerable Stark broadening ( 1 - 5 A0) (4,9). Broadening from the effects mentioned above are ordinarily small and therefore can be neglected. Introduction of a solution, water, or even hydrogen gas enhances the H^ emission signal and makes the determination less sensitive to electronic and plasma background noise. Rotational Temperatures. Rotational temperatures were calculated from the relative intensities of lines in the R and P branches of the N£+ First Negative band system. These measurements were made in the discharge region of the plasma because the intensity of the lines was much stronger than in the tail-flame. Calculations of the 101 rotational temperatures from the OH band in the tail-flame were attempted, but assignment of each line was nearly impossible. A spectral scan of the 2 £ - 2 £ transition, taken in the primary discharge region, is shown in Figure 4-1. Individual lines were assigned on the basis of the data of Childs (32). The intensity of each line in the R and F branch can be described from eqs. (4-3) and (4-6) (33). Ij = 4 (4-5) 4 where A a C em V /Q = constant because the wavelength range is r small. Cgm ® emission constant depending on the dipole moment, V is the optical frequency of the transition and Qr is the number of molecules in the rotational states. K1 = quantum number of the upper state K" = quantum number of the lower state 2 B' = rotational constant of the upper Ip = line intensity of the R branch Ip = line intensity of the P branch h = Planck's constant c = speed of light k = Boltzmann constant £ state = rotational temperature Combining equations (4-5) and (4-6) and assuming a Boltzmann 102 Figure 4-1: Vibrational-rotational emission spectrum of the 2 E transition of the + 9 E - First Negative band system viewed axially in the MINDAP with a high-resolution monochromator. Operating conditions were an applied microwave power of 250 W, reflected power 20 W, nitrogen flow rate of 1.75 L/min and 0.1 L/min through the side-on and central-channel of the plasma torch, respectively. eoi R E L A T IV E IN T E N S IT Y in o G> (0 -0 ) O) to oo -G> ocd------- •s -j ad O) co ■> in 9C4 ------- m — 1 00 i -<S ? O) c CO ® ® > o (O -CJ) 00 CO j -O ) oo CO 104 distribution among rotational levels, permits the rotational temperature to be computed [equation (4-7)]. In CaI/(K'+K"+l)) where 3 A - B'K'(K"+l)hc/(kTr) (4-7) B'hc/k = 2.983 a = 1 for even values of K' of the R branch a = 2 for odd values of K' of the R branch A semilogarithmic plot of In (ai/(K'+K"+l)) vs. K'(K"+1) in eq. (4-7) will yield a straight line with a slope of -B'hc/(kTr)) from which the rotational temperature can be determined. This method has the advantage of requiring only relative intensities. RESULTS Electronic Excitation Temperatures Spectroscopic temperatures were determined for a range of applied microwave power levels and nitrogen gas-flow rates (through the side-on inlet of the torch). Interestingly, a linear relationship was observed between applied microwave power and temperature (Fig. 4-2). For the various nitrogen flow rates, the temperature varies from an average value of 4100°K for an applied power of 150 W to 5800°K at 250 W. Not surprisingly, excitation temperatures were least for the highest nitrogen flow rate which was employed (2.1 L/min). As more gas is introduced into the plasma, more of the gas molecules have to be heated 105 Figure 4-2: Excitation temperature integrated over the first the MINDAP tail flame. 6 mm of An iron solution of 1000 yg/mL was introduced as a thermometric species. * - 1.35 L/min nitrogen flow in the side-on port of the MINDAP torch. + - 1.75 L/min nitrogen flow in the side-on port of the MINDAP torch. o-2.1 L/min nitrogen flow in the side-on port of the MINDAP torch. Exc itat ion Temperature CIO 6000 5500 5000 4500 4000 3500 150 250 200 AppIied Power CW) 300 107 and the lower the temperature will be for a given power input. Strangely enough, temperatures for 1.35 and 1.75 L/min are quite similar. The same dependence on flow rate was noticed for the emission intensity of iron atoms in Chapter 2 (Fig. 2-11). These experimentally obtained temperatures also support the optimum operating conditions selected in Chapter 2. It is not surprising that at 250 W of applied power the MINDAP compares favorably with the argon MIP and ICP. A comparison of excitation temperatures for these different plasmas (Table 4-1, 4-2) indicates that the MINDAP, Ar-MIP and ICP should all have similar excitation abilities. The detection limit comparison of Chapter 3 (Table 3-1) also supports this conclusion. Ion/Atom Ratios and Ionization Temperatures Figure 4-3 shows the magnesium ion-to-atom line intensity ratio as a function of applied power and nitrogen flow rate. As the power increased so did the ratio in a 1/(A+Bx) fashion where A and B are the intercept and slope from the best-fit curve to the data. This increasing ion/atom ratio is consistent with the results obtained in Chapter 2, in which atom-line intensities reached a plateau whereas ion signals increased steadily with applied microwave power. The effect of increasing the nitrogen flow rate from 1.35 L/min to 2.1 L/min was less than 5% at any power level. The magnesium ion/atom ratio obtained here is much smaller than that commonly reported in the Ar-ICP, which for most elements is typically greater than one and for magnesium approximately 11 (34,35). This "ion line advantage" in the ICP is one which does not exist in 108 Figure 4-3: Ion-atom line-intensity ratios for 10 yg/mL magnesium introduced into the plasma in three different flow rates of nitrogen: 2.1 L/min, 1.75 L/min, and 1.35 L/min. There is less than a 5% deviation among the curves obtained at different flow rates. - Ratio Ion / Atom 0.3 1 0 0 150 200 250 300 109 AppIied Microwave Power CIO 110 either flames or MIPs. When nitrogen was introduced into the Ar-ICP a decrease in ion-to-atom ratio was observed, suggesting that in the presence of nitrogen the ICP is closer to LTE (18,19). Ionization temperatures can be calculated from the ion/atom line ratios of Fig. 4-3 if electron densities are known. Unfortunately, Stark broadening of the hydrogen Balmer line could not be used because it was not visible in the emission spectrum. Addition of molecular hydrogen or water vapor into the MINDAP served only to increase the intensity of the molecular (NH,N0,0H) bands; no atomic emission from either hydrogen or nitrogen (which can also be used for electrondensity measurements) was observed. The line-merging technique for electron-density determination was also tried but to no avail. The weak transitions could not be observed for aluminum, lithium, calcium or potassium even though their convergent series lay in different spectral regions. Therefore, in order to calculate the ionization temperatures, an electron density was tentatively assumed. For atmospheric-pressure plasmas this number is often between 1 0 ^ - 1 0 ^ cm ^ (4). In Table 4-4 ionization temperatures for the MINDAP are calculated for several different assumed electron number densities. Depending on the ion/atom ratio, a temperature difference of 1300 - 1600°K is obtained for a 100-fold increase in electron density. For the calculation of ionization temperatures, shown in Figure 4-4, an electron density of 5 x 10 cm 3 was selected; Zander (36) and Fallgatter, et al. (12) determined electron number densities to be between 10 15 1 0 ^ and -3 cm for He and Ar MIPs operating at atmospheric pressure. Ill Table 4-4: Ionization temperatures calculated for a range of ion/atom ratios observed in the MINDAP (see Fig. 4-3) and for different assumed electron number densities (n ). e Ion/Atom Ratio n e 0 . 1 0 . 2 0.3 0.4 1.0E14 4683 4849 4951 5026 2.5E14 4904 5086 5198 5280 5.0E14 5085 5280 5401 5490 7.5E14 5198 5401 5527 5620 1.0E15 5280 5490 5620 5716 2.5E15 5561 5793 5937 6043 5.0E15 5793 6043 6200 6316 7.5E15 5937 6200 6365 6486 1.0E16 6043 6316 6486 6613 112 Figure 4-4: Ionization temperatures calculated using the ion/atom ratios o£ Fig. 4-3 and an assumed electron density of 5x10 14 -3 cm . The central-channel nitrogen flow rate was 0.1 L/min. * - 1.35 L/min nitrogen flow in the side-on port of the MINDAP torch. o - 1.75 L/min nitrogen flow in the side-on port of the MINDAP torch. + - 2.1 L/min nitrogen flow in the side-on port of the MINDAP torch. CIO Temperature Ionization 6000 5500 5000 4500 1 0 0 150 250 200 300 113 AppIied Power CIO 114 Four basic conclusions can be drawn from a comparison of calculated ionization and excitation temperatures in the MINDAP tail-flame. First, ionization temperatures increase linearly with applied power as do excitation temperatures (Fig. 4-2). Secondly and most practically, high temperatures (5000 - 6000°K) suggest the ability of the tail-flame to excite and ionize analyte. Third, although excitation and ionization temperatures are similar, they differ by as much as 20% at some applied powers. This difference is not surprising, since only a single electron density was used for calculating ionization temperatures at all operating conditions. In MIPs (40) the electron number density appears to be slightly dependent on the applied power; at low powers the electron density is lower than for higher powers. This effect would decrease the calculated ionization temperature at the lower powers and increase them at the higher applied powers, making them closer to measured excitation temperatures. Fourth, excitation temperature is dependent on flow rate whereas ionization temperature is not. These latter two differences strongly suggest a departure from LTE in the MINDAP tail-flame, although this deviation appears to be less dramatic than in most other plasma sources. Effect of Central-Channel Nitrogen Flow Central-channel nitrogen flow rates have ranged over 0.03 0.1 L/min during various stages in the MINDAP's development. Ionization temperatures were calculated as above for central-channel flow rates between 20 - 800 mL/min and at the optimum operating 115 Figure 4-5: The effect of central-channel nitrogen flow on the ionization temperature calculated from measured ion/atom line ratios and an assummed electron density of 5x10 14 -3 cm . Operating conditions were an applied microwave power of 250 W, reflected power of 20 W and a flow rate of 1.75 L/min through the side-on inlet of the MINDAP torch. CIO 5600 Ionization 5700 Temperature e» 5500 5400 5300 5200 100 200 300 400 500 600 700 800 900 1000 Central Channel Flo w Ra t e CmL/min) 116 0 117 conditions; values are shown in Figure 4-5. At a central-channel flow rate of 20 mL/min, the plasma is positioned immediately beyond the central tube. Consequently, some energy is lost to the torch that would ordinarily be available to ionize and excite the introduced aerosol. As the flow rate increases the temperature also increases because the plasma moves away from the tube and is cooled less by it. At higher flow rates ( > 500 mL/min), a decrease in temperature occurs because a greater volume of support gas must be heated. Also, such a high gas flow moves the discharge off center and at the same time begins to punch a hole in the base of the discharge (similar to an ICP). For later studies, the central-channel flow rate was maintained at 0.1 L/min to conserve gas, promote plasma stability and provide high temperatures. •Rotational Temperatures Rotational temperatures measured axially in the primary discharge region are plotted in Figure 4-6 for a range of operating powers and nitrogen flow rates. These temperatures, measured with no aerosol introduced into the MINDAP, correspond to the thermal energy available in the discharge region and are therefore an indication of the plasma1s ability to decompose aerosol samples into their atomic constituents. As with the other measured temperatures (ionization, excitation), a linear increase with applied power was noted. In contrast, higher flow rates of nitrogen gas produced higher temperatures. Importantly, rotational temperatures were somewhat lower (^500°K) under a given set of conditions than either excitation or ionization temperatures 118 Figure 4-6: Rotational temperatures calculated from R and P branches of the 2 E - 2 E + First Negative band system. The plasma was viewed axially and no aerosol was introduced into it. The central channel flow ratewas 0.1 L/min. 0 - 2 . 1 L/min nitrogen flow in the side-on port of the MINDAP torch. * - 1.75 L/min nitrogen flow in the side-on port of the MINDAP torch. + - 1.35 L/min nitrogen flow in the side-on port of the MINDAP torch. r\ 5500 5200 <D C L £ 4900 <D + I— _ 4600 + 0 c 0 -+J 4300 0 -4-> 0 q : 200 Microwave 250 Power CWD 120 even though they were obtained in the absence of aerosol introduction and in the center of the MINDAP discharge, presumably its hottest zone. Of course, it would not be surprising if this hot zone were well removed from LTE and that the excitation and ionization temperatures in it were much higher than the rotational temperatures. Nonetheless, the discharge center is clearly a thermally hot (5000°K or greater) environment and one well suited for atomization of aerosol samples. Spatial Profiles of Elemental Emission Spatial emission profiles for four elements shown in Figure 4-7 reveal that the greatest signal is generated just above the tip of the MINDAP torch. Interestingly, the vertical position where analyte emission becomes relatively constant depends at least partially on the element's excitation potential. Sodium, with an excitation energy of 2.08 eV, levels off near 38 mm above the cavity whereas zinc (5.75 eV) emission approaches a constant value much lower in the tail flame ( 1 2 mm). Spatially Resolved Temperatures Spatially resolved excitation and ionization temperatures were measured over the first 18 mm of the MINDAP tail-flame and are plotted in Figure 4-8. The substantial difference between the temperatures close to the cavity ( 0 - 3 mm) argues that LTE is not approached there presumably because the main discharge (which originates inside the cavity and is assumed to be well removed from LTE) extends outside the 121 Figure 4-7: Spatial emission profiles of several elements in the tail flame of the MINDAP. Operating conditions were an applied power of 250 W, reflected power 20 W, flow rate of 1.75 L/min and 0.1 L/min through the side-on and central channel of the plasma torch. A - 10yg/mL A1 (I) 396.2 nm B - 10yg/mLZn (I) 213.8 nm C - 10yg/mL Sr (I) 460.7 nm D - 10yg/mL Na (I) 588.9 nm 122 B HZI - 0.9 •.0 - 0.8 - 9.7 9.7 • 9.6 8.8 - 9.8 - 9.4 - 9.3 - 9.2 - 9 .1 - 9 4 3 2 I 8 4 3 M * . © M in 9 Plo R-lol t 123 Figure 4-8: Spatially resolved electronic excitation and ionization temperatures in the tail flame of the MINDAP. conditions are 250 W applied power, 2 0 Operating W reflected power, with flow rates of 1.75 L/min and 0.1 L/min through the side-on and central channel inlets of the plasma torch. * - excitation temperature o - ionization temperature Temperature CIO 7500 6500 5500 4500 3500 0 3 6 9 12 15 Height Above Cavity (mm) 18 21 125 cavity. Higher in the tail-flame the two temperatures are much closer, indicating a different, possibly thermal excitation mechanism. DISCUSSION The MINDAP discharge region and tail-flame have been described in this chapter in terms of the energy available for decomposing aerosols and exciting the resulting atoms and ions. Applied power has been the parameter that affects most greatly each of the temperatures measured. It is interesting to calculate the gas temperature which would be expected as a result of simple Joule heating at each applied power. This temperature can be calculated from eq. 4-8: W = cs_Mfr (T - 273) G G__________ (4-8) fSTP where W = power required to heat the gas to a given temperature (watts). c «*conversion factor (4.19 joules/cal). Sg = specific heat of the gas (cal/g). M = molecular weight of the gas (g/mol). £„ = flow rate of the gas (L/sec). ^STP = ^ *414 L/mol of gas at standard temperature and pressure. T = temperature (°K). Table 4-5 summarizes the results from this computation for nitrogen gas at flows used in this study. 126 Table 4-5: Applied power required to heat nitrogen to specified temperatures assuming constant heat capacity. 8 M 0.2477 cal/g M =* 28 g/mol Flow Rate (L/min) Temperature 1.35 1.75 2.1 6000 167 217 260 5500 160 198 237 5000 138 180 215 4500 123 160 192 4000 109 141 169 127 At an applied power of 250 W, with a reflected power of 20 W and a flow rate of 1.75 L/min, the amount of power calculated to heat nitrogen up to 5500°K is almost 200 W. This requirement leaves 50 W unaccounted, of which 20 W can be ascribed to reflected power. The remaining 30 W is probably dissipated as heat in the cavity, cables, tuning stubs, electrical connections, and also as radiation losses and from absorption of the microwave power by water. Unfortunately, the calculated values in Table 4-5 and the experimentally determined temperatures contradict each other. The rotational temperatures increase with flow rate at any given power, excitation temperatures are highest for 1.75 L/min and ionization temperatures are found to be independent of flow rate. This disparity is not too surprising since the values in Table 4-5 are based on the assumption of constant heat capacity and thermal conductivity. In fact, these parameters change strongly with temperature, especially for a diatomic gas like nitrogen. Barnes and Nikdel (37) have described mathematically nitrogen's heat capacity and its electrical and thermal conductivities for several temperature ranges below 2 0 ,0 0 0 °K. Between 6500 and 9000°K the thermal conductivity behaves anomolously, dropping to a minimum and then increasing with temperature. Even with corrections for these changes, it is unlikely that the calculated gas temperatures would agree with those that describe ionization and electronic excitation temperatures. Clearly, the deviation between rotational and the other temperatures indicates that the plasma is not entirely in LTE. A useful comparison can be made between a flame system and the MINDAP. Both sources emit more intense atom, rather than ion, lines, 128 both exhibit interelement interferences which can be overcome through use of releasing agents, and both possess distinct spatial regions where non-thermal and thermal mechanisms appear to prevail. Each source is best viewed from its side and is limited in sensitivity by background spectral interferences in the UV. However, the MINDAP has more energy available for dissociating and exciting analyte. In the helium and argon microwave plasmas, Busch and Vickers (5) noted that the ratio of the excitation temperatures was similar to the ratio of ionization potentials of the support gases. Ar-MIP and the MINDAP, similar results are found. approximately equal at 1.02. In comparing the Here the ratios are This evidence supports the theory that the mean electron energy in a MIP depends on the ionization potential of the support gas (5). CONCLUSION Temperature measurements in the MINDAP indicate that the plasma core is thermally hot, more so than flames and other MIPs, and that the MINDAP is comparable to argon MIPs and ICPs in its excitation ability. The spatially averaged temperatures calculated from the Boltzmann and Saha-Eggert equations indicate that excitation and ionization in the tail-flame are more closely related to a thermal mechanism than one involving Penning ionization or any of the others proposed for MIPs and ICPs. Additional evidence that supports a thermal mechanism includes the relationship between position of maximum analyte emission and an element's excitation potential, and the similarity of spatially resolved ionization and excitation temperatures high in the tail- 129 flame. These measurements support the hypothesis in ICP literature that it is possible to achieve LTE with the use of nitrogen as a supporting plasma gas. Unfortunately, without knowledge of tail-flame rotational temperatures, accurate electron number densities and the role that other excitation processes play in the tail-flame, the existence of LTE cannot be conclusively confirmed or rejected. The previous two chapters, combined with the findings here, reveal the potential importance of this new microwave emission source for analytical and physical measurements. The plasma has two well defined regions, one in the primary discharge zone which is highly energetic (at least 18.74 eV) and which sustains the discharge and the other downstream where background bands and continuum radiation are much less. The thermal temperature in the primary discharge is much higher than in other MIPs, making the MINDAP more tolerant of aerosol samples introduced into it and more efficient at decomposing them into their atomic constituents. Because the MINDAP is more thermal in character than most plasma sources, the emission behavior of elements is predictable, atom lines are more intense than ion lines, and lines of lower excitation energy are more sensitive. 130 REFERENCES 1. I. Reif, V. A. Fassel and R. N. Kniseley, Spectrochim. Acta, 28B, 105 (1973). 2. I. Reif, V. A. Fassel and R. N. Kniseley, Spectrochim. Acta, 33B, 807 (1978). 3. S. Greenfield, H. McD. McGeachin and P. B. Smith, Talanta, 22, 1 (1975). 4. A. T. Zander and G. M. Hieftje, Appl. Spectrosc., 35, 357 (1981). 5. K. W. Busch and T. J. Vickers, Spectrochim. Acta, 28B, 85 (1973). 6 . R. Mavrodineanu and H. Boiteux, "Flame Spectroscopy", John Wiley and Sons, New York, 1965. 7. C. Th. J. Alkemade and R. Herrmann, "Fundamentals of Analytical Flame Spectroscopy", John Wiley and Sons, New York, 1979. 8 . M. H. Abdullah and J. M. Mermet, Spectrochim. Acta, 37B, 391 (1982). 9. D. J. Kalnicky, V. A. Fassel and R. N. Kniseley, Appl. Spectrosc., 31_, 137 (1977). 10. J. M. Mermet, Spectrochim. Acta, 30B, 383 (1975). 11. G. R. Kornblum and L. deGalan, Spectrochim. Acta, 32B, 71 (1977). 12. K. Fallgatter, V. Svoboda and J. D. Winefordner, Appl. Spectrosc., 25, 347 (1971). 13. A. T. Zander, R. K. Williams and G. M. Hieftje, Anal. Chem., 49, 2372 (1977). 14. R. D. Deutsch, unpublished results. 15. A. T. Zander and G. M. Hieftje, Anal. Chem., 50, 1257 (1978). 16. A. D. Weiss, M. S. Thesis, Bloomington, In., 1980. 17. A. Montaser and J. Mortazavi, Anal. Chem., 52, 255 (1980). 18. R. M. Barnes and G. A. Meyer, Anal. Chem., 52, 1523 (1980). 19. A. Montaser, V. A. Fassel and J. Zalewski, Appl. Spectrosc., 35, 292 (1981). 20. M. H. Abdullah and J. M. Mermet, J. Quant. Spectrosc. Radiat. Transf., 19, 83 (1978). 21. R. M. Barnes and S. Nikdel, Appl. Spectrosc., 310 (1976). 131 22. M. Capitelli, F. Cramarossa, L. Triolo and E. Molinari, Combust. Flame, 15, 23 (1970). 23. F. Cramarossa and G. Ferraro, J. Quant. Spectrosc. Radiat. Transf., 14, 159 (1974). 24. P. B. Zeeman, S. P. Terblanche, K. Visser and F. H. Hamm, Appl. Spectrosc., 32, 572 (1978). 25. J. F. Alder, R. M. Bombelka and G. F. Kirkbright, Spectrochim. Acta, 35B, 163 (1980). 26. R. H. Tourin, "Spectroscopic Gas Temperature Measurements", Elsevier, New York, 1962, p.47. 27. P. W. J. M. Boumans, "The Theory of Spectrochemical Excitation", Hilger and Watts, London, 1966. 28. A. Montaser, V. A. Fassel and G. Larsen, Appl. Spectrosc., 35, 385 (1981). 29. H. Griem, "Plasma Spectroscopy", McGraw-Hill, New York, 1964. 30. A. Pannekock, Mon. Not. R. Astr. Soc., J98, 694 (1938). 31. D. Inglis and E. Teller, Astrophys. J., j)0, 439 (1939). 32. W. H. J. Childs, Proc. Roy. Soc., A137, 641 (1932). 33. G. Herzberg, "Spectra of Diatomic Molecules", Van Nostrand, New Jersey, 1950. 34. P. W. J. M. Boumans and F. J. De Boer, Spectrochim. Acta, 32B, 365 (1977). 35. N. Furuta and G. Horlick, Spectrochim. Acta, 37B, 53 (1982). 36. A. T. Zander, personal communication, 1981. 37. RJ M. Barnes and S. Nikdel, J. Appl. Phys., 47, 3929 (1976). 132 CHAPTER 5 Identification of Limiting Noise Sources in the MINDAP Accuracy, precision and the limits of detection of any analytical method depend on the magnitude of extraneous fluctuations imposed on the measured signal. These fluctuations, also called noise, are apparent as a variation about the mean signal value. In order to improve the performance of a measurement system, each noise component and its origin must be identified and, ideally, reduced or eliminated. The noise features of several spectrochemical sources have been studied in the past, particularly the inductively coupled plasma (1-3), microwave-induced plasma (4) and various analytical flames (5-10). The general conclusion is that, at concentrations near the detection limit, background noise is the precision-limiting factor, whereas above the detection limit analyte flicker is dominant. Similarly, noise power spectra reveal that the ICP is flicker-limited at frequencies below about 10 Hz (1), indicating that integration times greater than 0.1 s will not provide any improvement in signal-to-noise ratio (3). Talmi, £t _al. (4) examined the frequency composition of the noise from an argon microwave-induced plasma and found the noise spectrum to be essentially white (i.e., flat). This behavior is important because the S/N ratio then improves proportionately with the square root of the signal measurement period. In this chapter the noise associated with the emission signal from the MINDAP is critically evaluated. Precision measurements obtained for various signal integration periods and noise spectra both indicate 133 that the MINDAP is flicker-noise limited at frequencies below 0.2 0.5 Hz. This behavior is independent of desolvation, sample introduction method and analyte concentration, suggesting it to be characteristic of the MINDAP itself. Importantly, a dominant noise source was found to be the power supply, showing that MINDAP signals could be improved through better filtering and stabilization of the microwave source. A discussion of various noise sources and their effects is presented. EXPERIMENTAL The MINDAP was operated as a suspended plasma (cf. Ch. 2) at an applied power of 250 W and reflected power 20 W. The plasma support and central-channel nitrogen flow rates were 1.75 L/min and 0.1 L/min, respectively. The detection electronics were the same as in Chapter 2, except that the MINC 11/23 computer was exclusively used for data collection and computation. All programs were written using Ver. 2.0 of MINC BASIC. Four different experimental methods were used to identify the noise sources which were associated with the signal. were: These techniques wavelength-scan flicker-ratio measurement, time-constant effect, time-trace correlation and noise amplitude spectral determination. Below is a description of each technique, how it was used and the information that it provides. 134 Wavelength-Scan Flicker-Ratio Measurement It has been shown by Winefordner (6 ) that the relative amounts of white (shot) and flicker noise produced by an emission source can be determined by means of a particular form of wavelength scanning. In this method, a monochromator scans a spectral region containing either background or analyte'emission; at each wavelength, the emission signal is collected at a known sampling rate. The mean signal and standard deviation at each wavelength are then computed and used for determining the magnitude of each kind of noise. The total amount of noise on the signal at any wavelength is, by definition, equal to the standard deviation of the signal. In turn, the total noise (N^) is just the quadratic sum of the shot (Ng) and flicker (N^) components (equation 5-1). N t - t(N s ) 2 + (N, ) 2 r Because shot noise follows ] 1 / 2 (5-1) Poission statistics, its value is equal to the square root of the average signal (S) (eq. 5-2). From the measured total noise and (5-2) calculated shot-noise magnitudes, the flicker contribution can be calculated from eq. 5-3. At each wavelength in the spectrum, Nf = [(Nt ) 2 - S] 1 / 2 (5-3) 135 a ratio is calculated from the flicker- and shot-noise values. If this ratio is greater than two, flicker (either signal or source) is considered to be the dominant noise affecting the measurement at that particular sampling frequency. For this application 30 data points were collected for each wavelength over a spectral region of frequency of either 42 or 1 Hz. 1 0 to 2 0 nm, at a sampling The average, standard deviation and flicker ratio were calculated at each wavelength. From each flicker ratio, the dominant noise (shot or flicker) was identified. Time Constant Effect on Precision Precision indicates the reproducibility of a signal over a particular measurement time period. The MINDAP emission signal was monitored using time constants that ranged from 1 ms to 10 s. The average, standard deviation and relative standard deviation were computed for an observation period equal to 1 0 0 time constants. The limiting noise from these results can be identified from a logarithmic plot of precision (%RSD) vs. time constant. In a region where the dominant noise is white, a square-root relationship exists between the noise amplitude and bandwidth of the measurement system (11,12). Therefore, increasing the time constant will yield a proportionately smaller bandwidth, and thereby reduce the noise but not the signal. This reduced noise will result in an increase in precision and appear as a straight line with a slope of -0.5 on the logarithmic plot. However, if the noise is flicker-limited, its amplitude is proportional to the reciprocal of the frequency (1 /f) and no 136 improvement in precision is gained by increasing the time constant (zero slope on logarithmic plot). where n > 1 Finally, if the noise is' l/fn , , increasing the time constant will decrease the precision of the measurement. This effect will produce an upward shift of the logarithmic curve. Time-Trace Correlation Time tracings of emission can be used efficiently to identify common noise sources in an instrument or technique. If there is a common source of noise (i.e., plasma tail-flame waver) that affects analyte and background emission, a similarity will be observed between the simultaneous time tracings of the two signals. The degree of this coherence can be quantified from a correlation plot of the two parameters. Time correlation plots were constructed by monitoring the emission, forward-power and reflected-power signals from the MINDAP for a 400-minute time period. In order to record the power signals, a reflected power meter and BNC connections for each meter were installed on the power supply. All signals were collected simultaneously using three A/D converter channels on the MINC 11/23 computer. This correlation experiment was performed over a two-day period. On the first day, the power supply was turned off and remained off for 24 hours. This "off time" was necessary in order to determine the short-term fluctuations that arise from the warmup of the power supply. Data were collected only after the power supply had been on for three minutes. Within this three-minute time period, 90 s were 137 needed for a thermal relay inside the power supply to engage and transfer the high voltage to the magnetron tube, after which the plasma was lit and quickly tuned for minimum reflected power. For the remainder of the experiment, neither the tuning nor the power was adjusted. Noise Amplitude Spectral Determination Noise amplitude spectral measurements reveal the frequency composition of the noise present on a signal. In the present study, noise amplitude spectra were acquired for desolvated and undesolvated aerosol, for varying analyte concentration, and for both analyte and plasma background emission features of the MINDAP. The procedure used for the acquisition of noise amplitude spectra is illustrated in Figure 5-1. For selected emission features, both a background (A) and signal (B) trace were digitized. Initially, fluctuations in the plasma background (A) are detected, after which a sample is introduced into the MINDAP and an increase in the occurs (B). d.c. level The fluctuations associated with either A or B levels are collected at a sampling rate at least twice that of the highest desired frequency in the noise spectrum, thereby satisfying the Nyquist theorem. In acquiring a noise spectrum, accurate control of the measurement bandwidth was necessary to reduce the effects of aliasing (1). For these measurements an active low-pass filter (Krohn - Hite model 3342) with a very sharp roll-off (-96 dB/octave) was inserted between the fast current amplifier and the computer's A/D converter. 138 Figure 5-1. Schematic diagram of the signals used here for obtaining a noise-amplitude spectrum. A. Background emission from the plasma. B. Analyte emission. C. Real frequency component from the Fourier Transformation. D. Imaginary frequency component from the Fourier Transformation. E. Resulting noise-amplitude spectrum for low-frequency flicker-limited noise. 139 ffl < 140 Noise amplitude spectral data were acquired for the frequency ranges of 0 - 5 and 0 - 500 Hz at sampling rates of 14 and 1400 Hz, respectively. The frequency cut-off ( - 6 dB point) of the low-pass filter was set to the Nyquist frequency (7 and 700 Hz) to ensure that the contribution from aliased signals at 5 and 500 Hz was negligible (3). Five sets of time-domain data, each consisting of 1024 points, were acquired at each sampling rate. The resulting data sets were then transformed to the frequency domain by a Fast Fourier Transform (FFT) algorithm supplied with MINC BASIC. The resulting five noise spectra were averaged to yield the spectrum which is reported (1 ). In the process of implementing the FFT, both real (C) and imaginary (D) spectral components are produced; each contains positive and negative frequencies over the sampled ranges (i.e., -7 to +7 Hz). The peak in the middle of the real frequency spectrum (0 Hz) corresponds to the d.c. contribution. The amplitude spectrum (E) of the positive noise frequencies is calculated as the square root of the sum of the squares of the real (C) and imaginary (D) frequency components. The resulting spectrum was plotted on a Tektronix 4662 plotter as frequency (Hz) vs. root-meansquare current (nA). The bandwidth or resolution (R) of the noise-spectrum measurement system is calculated from equation 5-4. E - 2^ In eq. 5-4, N is the total (5-4) number of data points (1024) and fmflx is the maximum frequency which is properly sampled (7 or 700 Hz). The factor 2 indicates that only 141 half the number of data points collected is unique (i.e., -7 to 0 Hz frequency range contains the same information as 0 to 7 Hz). The measurement bandwidths for 7 and 700 Hz frequency ranges were 0.0137 and 1.37 Hz, respectively. Noise spectra are often displayed in power rather than amplitude terms. Noise amplitude and power can be related to each other by the square of the noise amplitude per unit bandwidth of the measurement system, as shown mathematically in equation 5-5. Np = (Na )2/R (5-5) Here, Np is the noise power, N^ is the noise amplitude and R is the bandwidth of the measurement (defined above). RESULTS Precision and Low-Frequency Noise The effect of the measurement time constant on relative standard deviation is often determined as a first step in improving the precision of any analytical measurement. The relationship between these parameters for the emission of a 10 pg/mL calcium solution in the MINDAP is shown in Figure 5-2. As the time constant is increased from 1 ms to 1 s, the %RSD decreases steadily. indicates that flicker (l/fn where n « not contribute significantly. 1 The slope of the line ) noise is present but does At time constants between 1 - 5 s, the %RSD reaches a minimum (n = 1), indicating the best signal-to-noise 142 Figure 5-2 Effect of measurement time constant on precision from a 10 yg/mL calcium soulution introduced into the MINDAP. Calcium measured as atomic emission at 422.7 nm. 10 RSD 8 6 4 2 0 0.001 0.01 0.1 10 1 0 0 143 Time Constant (sec) 144 ratio that can be obtained under the present operating conditions ( 1% RSD). Above 5 s the system is susceptible to long-term drift (n > 1 ), as is evident from the upward shift of the curve. The wavelength-scan flicker-ratio method described by Winefordner (6 ) supports the conclusions derived from Fig. 5-2. At both sampling frequencies (42 and 1 Hz) the flicker-ratio method yielded a flickerto-shot noise ratio less than two, indicating that at these frequencies flicker is not the dominant noise arising from either the plasma background (OH band from 300 - 310 nm and N^+ bands from 385 - 395 nm) or analyte emission (Ca, Mg, Na) in the MINDAP. The exact sampling rate at which flicker noise became dominant was not established with this method since it can be more easily found from a noise spectrum. The different types of noise (shot, flicker and interference) can easily be distinguished in a noise-amplitude spectrum. Figure 5-3 is a noise amplitude spectrum of OH emission at 306.4 nm over the frequency range of 0 - 5 Hz. In this spectrum two types of noise can be positively identified: a white-noise component, which is well removed from 0 Hz and has a noise amplitude that is flat with frequency, and a flicker-noise component which has a noise amplitude proportional to the reciprocal of the frequency. The frequency where the 1/f curve merges with the white noise, 0.2 - 0.5 Hz, defines the sampling rate for maximum attainable precision. spectrum for Interestingly, the noise amplitude emission at 391.4 nm exhibits the same shape as Fig. 5-3, suggesting a noise source common to the two. These noise amplitude spectra complement the precision and flicker-ratio measurements but are more accurate for identifying each type of noise 145 Figure 5-3. Noise spectrum o£ OH band emission at 306.4 nm over the frequency range 0 - 5 Hz. 0.35 Root Mean Squar• Currant CnA3 0.7 ___________ I___________ I___________ I___________ I___________ 0 1 2 3 5 146 Frequency (Hz) 4 147 present, the relative magnitude of each and its frequency composition. Analyte emission-noise spectra for desolvated solutions of calcium and sodium (0.1 - 1000 pg/mL) were similar to that in Figure 5-3 with a 1/f behavior becoming prominent between 0.2 - 0.5 Hz. As the analyte concentration increased, for the 0 - 5 Hz spectrum, the mean signal and noise amplitudes increased proportionately (see Fig. 5-4). These results support the hypothesis that the MINDAP is limited by multiplicative noise and that analyte flicker is dominant for concentrations well above the detection limit. The effect of introducing into the plasma undesolvated aerosols was to increase both the mean signal and its associated noise. The shape of the noise amplitude spectrum remained similar to that in Fig. 5-3, although the relative standard deviation for the 0 - 5 Hz spectrum increased from 2.5% (for desolvated aerosol) to 4%. These precision values were obtained independently of analyte concentration over the range 0.1 to 1000 pg/mL. These findings support the observation made in Chapter 2 that the MINDAP can handle aerosols directly, although the S/N ratio is better when desolvation is employed. High-Frequency Noise A noise amplitude spectrum between 0 - 500 Hz (Fig. 5-5) from the emission of atmospheric water vapor (OH) reveals three distinct features: a low-frequency flicker component (A), and peaks at 60 Hz (B) and 120 Hz (C). The 60 and 120 Hz spikes do not arise from interference noise associated with the detection electronics, since a 148 Figure 5-4. Effect of calcium concentration on emission signal and precision. o - Calcium atom emission for various concentrations of analyte (slope'll). The noise associated with each emission signal is smaller than the circle repre senting the emission intensity. * - %RSD for Ca (I) emission at each concentration. This is drawn on a linear scale ranging from 2.5% - 3.5%. Re Iat ive Intens ity 10 10* RSD 10 1 0 0 1000 10000 149 Analyte Concentration Gug/mL) 150 Figure 5-5. Noise spectrum of OH band emission at 306.4 nm over the frequency range 0 - 500 Hz. A - low frequency flicker of rms amplitude 0.5 nA. B - 60 Hz peak of rms amplitude 4 nA. C - 120 Hz peak of rms amplitude 3 nA. Frequency (Hz) 152 spectrum of similar amplitude taken with a flashlight did not exhibit these peaks. Instead, this interference noise probably originates in the microwave power supply. The power supply basically consists of a high- voltage transformer, a full-wave rectifier, some filtering capacitors and the magnetron tube (source of 2450 MHz). Ripple noise from the filtering circuit superimposed on the output high-frequency waveform is believed to couple to the plasma and to modulate it. The resulting modulation is then viewed as strong peaks at the modulating frequency (60 Hz) and its harmonics. This hypothesis is supported by the behavior of the plasma when even more water vapor is introduced into it. Additional vapor was intentionally swept into the MINDAP by passing the nitrogen support gas through a gas dispersion tube, immersed in distilled water. The resulting plasma became visibly unstable and could be seen to pulsate. Moreover, the noise spectrum of the OH 306.4 nm emission band contained exaggerated contributions from the power supply ripple and its harmonics (Fig. 5-6). This peculiar behavior can be qualitatively attributed to the strong water-vapor absorption band at the microwave operating frequency. Presumably, the 60 and 120 Hz ripple on power transferred into the plasma couples with the water vapor directly, causing it to amplify the plasma modulation. This modulation then becomes so strong that the plasma is nearly extinguished during each modulation cycle. The resulting nonlinear modulation generates the range of 60 Hz harmonics apparent in Fig. 5-6. 153 Figure 5-6. Noise spectrum of the OH 306.4 nm band when a large flow of water vapor is intentionally passed into the MINDAP. CnA3 Currant Root M*an Squara 12.5 100 200 300 500 154 Frequency (Hz) 155 DISCUSSION There are two major kinds of noise: additive and multiplicative. Additive noise arises independently of a measured signal (emission). It is easier to understand and overcome in a measurement, although it is usually not dominant at high analyte concentrations. Examples of additive noise in a spectrometric system are stray light variations, detector offset drift and amplifier baseline noise. Methods often used for minimizing additive noise are blank subtraction and signal modulation (10,13,14). In contrast, multiplicative noise is introduced simultaneously with the signal and is more difficult to correct (15). As its name implies, this kind of noise multiplies the signal, in effect, making the resulting fluctuations proportional to both the noise source magnitude and the signal itself. Multiplicative noise arises from fluctuations in source temperature, sample introduction systems, power transfer, impedance tuning, source flicker, gas dynamics (flow rate and pressure) and analyte concentration. The only way to overcome multiplicative noise is to ratio the signal of interest to a reference that is related to the original noise source. A common form of such correction involves the use of an internal standard. It appears that the MINDAP system is most greatly affected by multiplicative noise. The similarity of noise spectra obtained from analyte emission and various background spectral features argues for a common, multiplicative source, as does the independence of precision with analyte concentration (Fig. 5-4). Several possible origins of multiplicative noise exist; in addition to those mentioned above, 156 microwave power-supply fluctuations would be suspected. Figure 5-7 shows the dependence of OH emission (306.4 nm) on applied microwave power for two different impedance-matcher tuning settings. At a constant tuning setting, emission intensity varies linearly with power. However, changing the tuning, but not reflected power, establishes a different proportionality between power and emission intensity. From Fig. 5-7,,any fluctuations in applied power should produce proportional variations in an emission signal. Moreover, Figures 5-5 and 5-6 suggest the possible importance of power-supply variations as a noise source. To determine the importance of such fluctuations, simultaneous time tracings were obtained of plasma background emission (N£+ 1st Negative system at 391.4 nm) and of applied and reflected microwave power. These tracings are reproduced in Fig. 5-8 and are plotted as the deviation of the signal from its normalized mean. Table 5-1 lists the initial values and statistical features of each of the three signals. A careful examination of the three traces reveals a slight similarity between those for forward power (A) and background emission (B). This similarity suggests that power-supply noise is a strong, if not dominant, source of variation in the MINDAP. However, Table 5-1 shows that the relative standard deviation of the background . trace is considerably greater than that of forward power, indicating that other significant noise sources exist. Correlation plots (Figures 5-9 to 5-11) of the three sets of data verify the foregoing qualitative observations. The only plot that shows no correlation is that of the background emission signal and reflected power. The correlation coefficient for the forward power and 157 Figure 5-7 The effect of microwave power and two different impedance-matcher settings on the emission intensity of OH at 306.4 nm. 2 Re Iat ive Intens ity 3 170 230 210 A p p 1 ied P o w e r CW) 250 158 190 159 Figure 5-8. Time tracings of the relative fluctuations from the mean. A. Forward power. B. N^+ First Negative System. C. Reflected power. Relative Amplitude Time Cm in) 09T 161 Table 5-1: Initial and statistical values for each signal collected from time tracings over a period of 400 min (see Fig. 5-8). N2+ Signal Forward Power Reflected Power Initial 1.618 261 25.9 Max 1.729 274 27.9 Mean 1.576 266 25.4 Min 1.457 257 21.8 Std. dev. 0.65 3.32 1.32 % RSD 4.12 1.25 4 162 Figure 5-9. Correlation plot of the forward power with the emission signal from the plasma background. Powor Forward 163 Emission Signal 164 Figure 5-10. Correlation plot of the forward microwave power with reflected power. F o r w a r d P ower Reflected Power £91 166 Figure 5-11. Correlation plot of the reflected power with emission signal from N£+ plasma background. Pow«r R«fl«ct«d 167 Emission Signal 168 emission signal is 0.55. 0 . 6 8 and between forward and reflected power is This degree of correlation supports the argument that emission intensity is influenced by fluctuations in the power applied to the plasma. As suggested above, additional factors contribute to signal instability in the MINDAP. One such factor is air turbulence. The plasma's tail flame, which extends ten centimeters from the cavity and has a diameter of approximately surrounding air drafts. 6 mm, is easily perturbed by In fact, in the last eleven minutes of the time-tracings the air currents in the room markedly affected the plasma's stability. Accounting for this pertubation (by discarding those data points), the correlation coefficients from Figs. 5-9 to 5-11 were recalculated. Only the correlation coefficient for the signal with forward power was affected, increasing from 0.68 to 0.74. There are other factors that will multiplicatively influence emission intensity which might not appear as a change in microwave power. These additional parameters are: instability of the magnetron tube's output frequency, changes of either the amount or type of dielectric introduced into the cavity, changes of impedance matching (as noted in Fig. 5-7), and the efficiency with which each electrical connection transfers the applied power. CONCLUSION The MINDAP appears limited at high analyte concentration by multiplicative noise arising at least in part from the power supply, impedance matcher, analyte concentration, and source flicker. A 169 flicker-limiting frequency of 0.2 to 0.5 Hz was determined for the MINDAP and found to be independent of analyte concentration, element excitation potential and the existence of desolvation. At low frequencies, noise spectra retained the same shape but increased in amplitude when aerosol was introduced into the plasma without desolvation. This finding supports the observation made in Chapter 2 that desolvation of an introduced aerosol increases the signal-to-noise ratio. . At present, a significant and perhaps dominant source of multiplicative noise arises from the power supply and power transfer systems. Nonetheless, after seven hours of free running, the relative standard deviation for the power and signal fluctuations was less than 5%. Similarly, the precision for an analytical determination, lasting typically 30 s, was also less than 5%. The results in this chapter have provided information about several factors that affect the precision and sensitivity of the MINDAP system. Other factors which need to be considered to totally characterize this plasma source are the effects of gas flow rate and pressure on signal stability. In the next chapter, sample is introduced into the plasma in a completely atomized form, and should therefore indicate the full potential of the MINDAP as an analytical emission source. 170 REFERENCES 1. R. M. Belchamber and G. Horlick, Spectrochim. Acta, 37B, 17 (1982). 2. G. L. Walden, J. N. Bower, S. Nikdel, D. L. Bolton and J. D. Winefordner, Spectrochim. Acta, 35B, 535 (1980). 3. R. H. Belchamber and G. Horlick, Spectrochim. Acta, 37B, 71 (1982). 4. Y. Talmi, T. Crosmun, and N. M. Larson, Anal. Chem., 48, 326 (1976). 5. G. M. Hieftje and R. I. Bystroff, Spectrochim. Acta, 30B, 187 (1975). . K. Fujiwara, A. H. Ullman, J. 0. Bradshaw, B. D. Pollard and J. D. Winefordner, Spectrochim. Acta, 34B, 137 (1979). 7. C. Th. J. Alkemade; Tj. Hollander, K. E. J. Honings, H. A. Koenders and R. J. J. Zijlstra, Spectrochim. Acta, 34B, 85 (1979). . C. Th. J. Alkemade, H. P. Hooymayers, P. L. Lijnse and T. J. M. J. Vierbergen, Spectrochim. Acta, 27B, 149 (1972). 9. C. Th. J. Alkemade, Tj. Hollander, H. Snipe and R. J. J. Zijlstra, Spectrochim. Acta, 36B, 77 (1981). 6 8 10. M. Markinkovic and T. J. Vickers, Anal. Chem., 42^, 1613 (1970). 11. G. M. Hieftje, Anal. Chem., 44, 81A (1972). 12. G. M. Hieftje, Anal. Chem., 44, 69A (1972). 13. C. Th. J. Alkemade, W. Snelleman, G. D. Boutilier, B.D. Pollard, J. D. Winefordner, T. L. Chester and N. Omenetto, Spectrochim. Acta, 33B, 383 (1978). 14. G. D. Boutilier, B. D. Pollard, J. D. Winefordner, T. L. Chester and N. Omenetto, Spectrochim. Acta, 33B, 401 (1978). 15. C. Th. J. Alkemade, W. Snelleman, G. D. Boutilier and J. D. Winefordner, Spectrochim. Acta, 35B, 261 (1980). 171 CHAPTER 6 Microsample Introduction into the MINDAP Using a Microarc Atomizer In the preceding chapters, the microwave-induced nitrogen discharge at atmospheric pressure (MINDAP) has been used with a driedaerosol sample-introduction system. exhibits In this configuration, the MINDAP high temperatures (Chap. 4), low detection limits (Chap. 3), and matrix interferences which can be overcome in a manner similar to that employed in flame spectrometry. In this chapter the MINDAP system is investigated as an analytical atomic emission source for microsampling analysis using a microarc atomizer. The (1 ) as a microarc atomizerwas developed in 1974 by Layman and Hieftje device to convert discrete microquantities of liquid sample into atomic vapor for emission analysis in a microwave plasma. The microarc is a high-voltage, low-current discharge that sequentially and efficiently desolvates, vaporizes and atomizes sample volumes from to 40 yL (1). 0 . 1 This concept of separate atomization and excitation was discussed also by Falk, et al. (2) who showed how it could enhance the sensitivity of other atomic emission measurements. The microarc has been successfully combined with several plasma emission sources (1,3-5). In conjunction with the inductively coupled plasma (ICP) (3,4), the microarc yielded lower detection limits than other microsampling techniques applied to the ICP (6-9). Argon and helium microwave-induced plasmas (MIP), when coupled to the microarc (1,5), offer the sensitivity and freedom from interferences that has 172 been associated with the ICP. Yet their physical size, instrumentation and operating requirements are more compact and economical. Elemental analysis using a microwave plasma as the excitation source has been limited to samples introduced as a vapor effluent from gas chromatography (10,11), thermal atomizers (12,13), hydride generators (14,15), and laser vaporization devices (16). Nebulizer systems have also been employed (17) but to a lesser extent and less successfully. Analyte is preferably introduced as a vapor because the MIP lacks the thermal energy needed to decompose the sample. Another limitation of the MIP is its small physical size which restricts the amount of sample that can be introduced before overloading occurs. The MINDAP system overcomes many of the inconveniences ordinarily associated with microwave plasmas: it readily accepts aerosol samples, possesses the high thermal energy needed for sample decomposition, and is relatively unaffected by high analyte concentrations. The MINDAP system has previously been evaluated for continuous solution analysis, and is now the subject of investigation with discrete microvolume aliquots from the microarc sample-introduction technique. This is the first time that the microarc has been used in a molecular-gas atmosphere (i.e., nitrogen). Its operating characteristics are therefore slightly different from those in an inert monatomic gas atmosphere and a qualitative description of its behavior is included. The microarc-MINDAP combination yields pg - fg detection limits, a broad linear dynamic range, good precision (3 - 7% rsd), and essentially no interference from either sodium or phosphate. 173 EXPERIMENTAL Connection of the microarc atomizer to the MINDAP was straight forward and is detailed.in Figure 6-1. Initially it was attempted to introduce the microarc-atomized sample through the central channel of the plasma torch. However, in this configuration the suspended MINDAP plasma was perturbed when the arc was struck. Consequently, in this investigation the microarc was connected to the side-on gas inlet of the torch (Figure 6-1). Timing of the microarc and data collection were controlled by a laboratory computer in a manner similar to that described by Keilsohn, Deutsch and Hieftje (3). shown in Figure 6-2. A block diagram of the experimental set-up is The computer controls the duration of the desolvation period, the ignition and duration of the arc, and the collection and processing of the emitted-radiation signals. The MINDAP plasma was sustained using the conditions determined in Chapter 2 to be optimal for the nebulizer-based system. These conditions produced a suspended plasma configured vertically, with an applied microwave power of 250 W and reflected power of 20 W. The gas flow rates through the central channel of the torch and through the microarc assembly were 0.1 L/min and 1.75 L/min, respectively. Analyte emission was collected in a radial viewing configuration from the first 6 m m of the plasma tail flame. The microarc electrodes (30-gauge tungsten wire as the cathode and 24-gauge stainless-steel syringe needle for the anode) were positioned to support a stable and reproducible discharge in the flowing nitrogen atmosphere; an interelectrode spacing of approximately Figure 6-1: Microarc-MINDAP operational configuration. M I N D A P h” ^ \ aaaa > discharge Viewing direction m MINDAP torch TEMqio Anode 176 Figure 6-2: Block diagram of the instrumentation used to collect and analyze the transient emission signal from the microarc-MINDAP system. PMT =■ R928 Photomultiplier tube i-to-V Amp = Keithley 427 current amplifier H.V.P.S. = Keithley 244 High voltage power supply asma Monochromator uWAVE Power Su p p I y Constant Current Sour ce ower SuppIy MINC Arc T im e r 1I Computer 177 178 0.5 mm vas employed. The current and voltage needed to sustain the microarc discharge in the flowing nitrogen atmosphere were 28 ma and 1800 Vp_pi which were higher than those required in the flowing argon systems (1,3). Analyte desolvation was accomplished by ohmically heating the cathode from a constant current supply of 4V at 2A (typically 30 8 for 1 pL of solution) before initiating the burn. The appearance of the microarc discharge in the nitrogen atmosphere was similar to that noted in other gases. The stable arc emits a bluish- colored plume, characteristic of the electrical breakdown of nitrogen. When first struck, the arc anchors to the tip of the hairpin-shaped cathode and then uniformly surrounds it. Because of the rather small separation between the microarc electrodes, care was required to dispense the microvolume sample onto only the cathode. In trials when the anode was contaminated by sample solution, the arc was difficult to initiate and, once struck, generated broadband emission from the electrode material. The duration of the arc was empirically optimized by monitoring both signal and background emission traces, as described previously (3). For arc times beyond one second there was no detectable analyte emission; consequently, in later determinations the arc duration was one second and data were collected for two seconds. This additional time enabled the computer to establish a baseline, allowed for analyte transport to the plasma and encompassed total decay of the emission signal. A typical emission-time profile accumulated under these operating conditions is displayed in Figure 6-3. 179 Figure 6-3: Emission-time profile for 1 ng (1 yL of 1 yg/mL) copper at 324.7 nm. A - Analyte emission B - Background emission Re Iat ive Intens ity S> T ime Csec. l\> 081 181 RESULTS AND DISCUSSION Desolvation Technique A factor that limited reproducibility and signal stability in initial microarc-MINDAP trials was uncertainty in the duration of the desolvation process. If desolvation were incomplete, initiating the arc resulted in intense broadband emission from the MINDAP, caused presumably by tungsten oxide liberated from the sample electrode (18). On the other hand, waiting a sufficient period of time to ensure complete desolvation decreased the throughput of sample analysis. It was decided that the best way to overcome this difficulty would be to monitor the desolvation process. The voltage-derivative method developed by Layman and Hieftje (19) records the desolvation process electronically by monitoring changes in the microarc filament resistance as the solvent evaporates. Unfortu nately this technique was not successful when incorporated into the microarc-MINDAP system. Erratic derivative signals were produced during the sample desolvation, caused probably by the high flow rate of gas and by the consequent instability in the solvent evaporation rate. Instead, it was found that desolvation could be monitored independently of gas flow and solvent evaporation rates by observing the emission of OH at 306.4 nm. During desolvation of an aqueous sample, water vapor is continuously swept into the MINDAP to yield strong emission at 306.4 nm; a pronounced reduction in this signal indicates the end of desolvation. This spectroscopic monitor was found to be as sensitive and accurate as the voltage-derivative technique for 182 determining the end of solvent evaporation but is more compatible with the flowing-nitrogen system. Compared to waiting a fixed length of time for desolvation, the spectroscopic technique increased sample throughput by as much as 50%. Analytical Figures-of-Merit Detection Limits. Limits of detection for the microarc-MINDAP system were determined at the 95% confidence level for eight elements of varying excitation potential. The method used is similar to that previously described (3) and is based on a noise value equal to the standard deviation obtained from thirty background traces (N=30, a=*0.05, t=2.045) (20). The values reported in Table 6-1 are the averages of at least five separate determinations of the detection limit. For comparison, Table 6-1 includes detection limits obtained with the MINDAP coupled with a nebulizer sample-introduction system. The values in Table 6-1 are consistent with the general trend that elements with higher excitation potential have higher detection limits than those of lower potential. The results indicate also an improvement in detection limits when the analyte is introduced in a preatomized form by the microarc. Similar improvement has been reported (1,3,4,20) when the microarc was used for sample introduction into other plasma systems. A comparison of detection limits offered Table 6-1: Detection limits (ng/mL) for the MINDAP obtained with microarc and nebulization/desolvation sample-introduction systems. Element Ca(I) Q Wavelength(nm) 422.7 Excitation Energy (eV) 2.92 Detection Limits Nebulizer1' Microarc" 1 . 0 1 . 2 0.13 4.4 0.4 5.4 0 . 2 0 . 2 2 Cu(I) 324.7 3.80 K (I) 766.5 • 1.61 Li(I) 670.8 1.84 Mg(l) 285.2 4.33 0.5 Na(I) 589.0 2.09 0.072 (c) Pb(I) 405.8 3.04 56 Zn(l) 213.8 5.77 166 • • • all volumes were 1 pL, unless otherwise indicated. ^Chapter 2 (c) 0.5 pL sample volume. 1 2 0.29 82 1 2 0 184 by several discrete-sample-introduction / multielement-analysis systems is offered in Table 6-2. The microarc-MINDAP combination compares favorably with the far more elaborate and expensive systems. The increased sensitivity offered by the microarc over the nebulizer system is not surprising. The microarc concentrates the sample introduced into the plasma as a "plug" of analyte vapor much as a carbon furnace does in AAS. plasma is nearly 1 0 0 In addition, sample delivery to the % efficient with the microarc whereas common nebulizers are at best 5% efficient. One of the major advantages of using the microarc-MINDAP is that it is mass-sensitive rather than concentration-sensitive. Therefore, preconcentration of sample solutions onto the microarc filament should improve detection limits even further. Finally, the separation of sample desolvation and atomization by the microarc promotes MINDAP stability and analyte excitation. When water vapor is introduced into the MINDAP, the plasma background and noise level increase, as shown in Chapter 5. This shift in background level was noticed as early as 1917 by Strutt, who reported an increase in the NO band systems in the nitrogen afterglow upon the addition of water vapor (22). It has recently been suggested that the NH bands in the background spectrum are derived from the reaction of nitrogen with the dissociation products of water (23,24). Introduction of water vapor into the MINDAP might have the added consequence of quenching the production of excited nitrogen species necessary to sustain the nitrogen afterglow and which might be related 185 Table 6-2: Comparison of detection limits (pg) offered by several discrete-sample-introduction / multielement-analysis systems. Element Ca (I) microarc MINDAP 0.05 microarc Ar-MIPa 1.0 Ca(Il) ETA ICPC - AAS 1.0 0.06 Cu (I) 0.13 K (I) 0.4 Li (I) 0.2 - Mg (I) 0.5 0.045 0.16 3.9 - - Mg(II) Na (I) microarc Conv-ICPb - 0.87 0.036 0.01 Pb (I) 56 0.38 Zn (I) 166 0.92 - 0.4 - 1.0 - - 0.4 - 3.6 280 0.12 0.5 100 20 0.02 0.4 0.4 0.7 1000 ^icroarc coupled to an argon-supported Microwave-Induced Plasma (1). bMicroarc coupled to an Inductively Coupled Plasma (ICP) (3). g Electrothermal atomization into an ICP (7). ^Carbon-furnace Atomic-Absorption Spectroscopy (21). 186 to analyte excitation (25,26,27). Of the elements listed in Table 6-1, lithium and sodium should not have been affected significantly by the influence of water vapor on background emission, since both lie in a low-background spectral region. Yet, the presence of water vapor hardly affected the detection limit for lithium, but reduced that for sodium by almost an order of magnitude. This change might be due to a transfer of energy to sodium from nitrogen in the first positive system of the Lewis-Rayleigh afterglow (28). Starr has demonstrated that vibrationally excited nitrogen molecules can transfer energy to the electronic levels of sodium by a collisional process (29). Working Curves. Figure 6-4 shows calibration curves for zinc (213.8 nm), copper (324.7 nm), lead (405.8 nm), and lithium (670.8 nm) with lines drawn to an order of magnitude above their respective detection limits. These curves have slopes of approximately unity over a concentration range of three-to-five orders of magnitude which are typical for most elements studied. The measurements at 3000 Wg/ml were performed using the preconcentration technique (4); three 1 ML aliquots of 1000 Mg/rnL solution were dispensed onto the cathode before the arc was struck. Precision. The precision of the microarc-MINDAP combination was ascertained from five successive determinations of each of five different days. 1 0 ng of copper on The relative standard deviation on each day was between 3 and 7% whereas the reproducibility among days varied between 1 and 4%. At least three factors appeared to affect the precision of the measurement: the electrode spacing and relative 187 Figure 6-4 Calibration curves for the microarc-MINDAP system. Curves for copper and lithium are superimposed. A - Zn(I) 213.8nm o - Cu(I) 324.7nm * - Pb(l) 405.8nm * - Li(I) 670.8nm 10 Re Iat ive Intens ity 10 10 -2 1 0 -3'-------- L— 1 .0E-3 0.01 10 100 1000 10000 188 Analyte Concentration CMg/mL) 189 geometry, the position of the sample droplet on the cathode and the surface characteristics of the sample electrode. These factors have been excellently characterized by Bystroff, et al., who operated the microarc under different conditions than presented here (18). As a result, the qualitative discussion below will dwell on the effects the flowing nitrogen atmosphere has on these parameters. The electrode spacing affected the microarc characteristics -greatly. If the electrodes were too close ( <0.5 mm), anode contamination was common, and the breakdown of nitrogen and the decomposition of the sample affected. When the electrodes were too far apart ( > 1 mm), the breakdown was not reproducible. The relative geometry of the electrodes influenced arc stability by controlling the position where the arc contacted the cathode. Greatest precision was found empirically when the electrodes were aligned so the arc was directed at the tip of the hairpin-shaped cathode (cf. Figure 6-1). The position of the sample droplet on the cathode determined both precision and sample vaporization efficiency. The more centralized the droplet was at the tip, the more complete became the vaporization step. When a determination was attempted before the solvent was completely evaporated, two effects occurred: broadband emission from the electrode material and difficulty of the sample aliquot adhering to the cathode. The droplet would bead-up on the wire and be blown off by the flowing nitrogen gas. It took between five and ten one-second strikes of the arc before the sample would adhere reproducibly to the cathode. An investigation of the electrode surface might be able to explain these effects. 190 Interferences Two classical interferences on calcium emission were examined: the ionization effect caused by the addition of sodium (Fig. 6-5) and the vaporization effect from the addition of phosphate (Fig. 6 -6 ). The absence of interferences is evident from the constant calcium atom emission intensity as the interferent concentration is increased. The ionization interference is essentially absent even at a concomitant solution concentration of 0.1%. Similarly, vaporization effects are not present even at a phosphate/calcium molar ratio of 60. In comparison to the nebulizer system, which was severely influenced by the solution matrix (Chap. 3), the microarc exhibited no interference even in the absence of releasing agents. These results are consistent with previous work using the microarc (1 ) and are attributable to its ability to separate and efficiently perform the processes involved in decomposing the sample into its atomic constituents. CONCLUSION The microarc-MINDAP system is an economical and sensitive spectroscopic tool for microvolume sample analysis. between 0.1-10 pL can be easily analyzed. Sample volumes Importantly, only the volume dispensed onto the electrode is analyzed, unlike the situation with nebulizer systems which are seldom greater than 5% efficient. Because the microarc- MINDAP combination responds to sample mass and exhibits minimal matrix interferences, calibration curves can be prepared using only a single solution standard. A broad linear dynamic range of 191 Figure 6-5: Sodium interference on 10 ng (1 yL of 10 yg/mL) Calcium at 422.7 nm. 300 1 ... 1 ! - - S igna 250 200 }- CD Ca ------ g ----- o 150 - 1 0 0 - . 1 1 10 100 Sodium ... Concentration _,l ... 1000 Cjug/mL) 10000 193 Figure 6 -6 : Phosphate interference on 10 ng (1 yL of 10 yg/mL) Calcium at 422.7 nm. 300 200 Ca CD S i gna 250 150 - 1 0 0 0.01 1 0 0 P04 / Ca M o l a r Ratio 195 three-to-five orders of magnitude is typical for these curves. Unlike the argon (1) and helium (5) MIPs, the microarc-MINDAP system can handle sample masses up to at least 3 yg. This discrete sample-introduction system exhibits detection limits which are comparable to those produced by other microwave-induced plasmas (1,5), inductively coupled plasmas (3-4), and flameless atomic absorption (21) techniques. t The system is virtually free from matrix (phosphate and sodium) interferences, unlike the ICP-microarc combination and nebulizer-MINDAP systems. The overall precision is good and ranges between 3 and 7%. The operational features of this new combination make it convenient to use. The sample introduction and emission systems are inexpensive to construct and operate because of their low power requirements and the use of nitrogen for the plasma and sample-delivery gas. Both the microarc and detection electronics can be easily automated (1,3) for accurate and precise measurements. analysis time is short - approximately 1 Moreover, minute from dispensing the sample until the computer finishes calculating and writing the results on a storage disk. Although inherent sensitivity and economy make this new combination attractive, it is by nature a discrete-sampling technique. Continuous sample introduction is therefore impossible and signal averaging is time-consuming. Fortunately, the MINDAP is dominated by low-frequency noise and, as indicated in Chapter 5, time constants longer than approximately two seconds do not improve precision. Because the MINDAP is a used in an emission mode, simultaneous multielement detection is possible even for discrete sample analysis. 196 Other minor drawbacks could be minimized. The precision and sample throughput rate could be improved further by measuring automatically the mass of the analyte dispensed onto the cathode using the described OH-band monitoring technique. The microarc-MINDAP combination could be improved also by optimizing the applied power and nitrogen flow rate for greatest signal-to-background ratio. 197 References 1. L. Layman and G. M. Hieftje, Anal. Chem., 4 7 , 194(1975). 2. H. Falk, E. Hoffmann, 1. Jaekel and Ch. Ludke, Spectrochim. Acta, 34B, 333 (1979). 3. J. P. Keilsohn, R. D. Deutsch and G. M. Hieftje, Appl.Spectrosc., 37, 101 (1983). 4. R. D. Deutsch and G. M. Hieftje, "Microarc Microsampling for a Mini ICP", paper #179, IX Annual Meeting of the Federation of Analytical Chemistry and Spectroscopy Societies, Philadelphia, PA. 1982. 5. A. T. Zander and G. M. Hieftje, Anal. Chem., 50, 1257 (1978). 6 . D. E. Nixon, V. A. Fassel, R. N. Knisely, Anal. Chem., 46, 210 (1974). 7. A. M. Gunn, D. L. Millard, and G. F. Kirkbright, Analyst, 103, 1066 (1978). 8 . R. L. Dahlquist, J. W. Knoll, and R. E. Hoyt, Paper #341, 26th Pittsburgh Conference, 1975. 9. A. Aziz, J. A. C. Broekaert, and F. Leis, Spectrochim. Acta, 36B, 251 (1981). 10. C. I. M. Beenakker, Spectrochim. Acta, 32B, 173 (1978). 11. B. D. Quimby, P. C. Uden, and R. M. Barnes, Anal. Chem., 5£, 2112 (1978). 12. F. L. Fricke, 0. Rose, and J. A. Caruso, Talanta, 23, 317 (1975). 13. J. H. Runnels, and J. H. Gibson, Anal. Chem., 39, 1399 (1967). 14. W. B. Robbins, J. A. Caruso, and F. L. Fricke, Analyst, 104, 35 (1979). 15. P. Barett and T. R. Copeland, "Applications of Plasma Emission Spectroscopy", 139, R. M. Barnes ed., Heyden Publishers, 1979. 16. T. Ishizuka and Y. Uwamino, Anal. Chem., 52, 125 (1980). 17. F. L. Lichte and R. K. Skogerboe, Anal. Chem., 45, 399 (1973). 18. R. I. Bystroff, L. R. Layman, and G. M. Hieftje, Appl. Spectrosc., 33, 230 (1979). 19. L. Layman and G. M. Hieftje, Anal. Chem., 46, 332 (1974). 198 20. J. D. Winefordner and T. J. Vickers, Anal. Chem., 36, 1939 (1964). 21. Instrumentation Laboratories, Publication AID 91, Mass. Wilmington, 22. R. J. Strutt, Proc. Roy. Soc., A93, 254 (1917). 23. P. Goudmand, G. Pannetier, 0. Dessaux, and L. Marsigny, Compt. Rend., 25£, 422 (1963). 24. G. Pannetier, P. Goudmand, 0. Dessaux, and N. Tavernier, J. Chim. Phys., 61, 395 (1964). 25. E. P. Lewis, Phys. Rev., 18, 125 (1904). 26. R. J. Strutt, Proc. Roy. Soc., A85, 219 (1911). 27. R. J. Strutt, Proc. Roy. Soc., A88, 539 (1913). 28. A. N. Wright and C. A. Winkler, "Active Nitrogen", Academic Press, N. Y., 1968. 29. W. L. Starr, J. Chem. Phys., 43, 73 (1965). 199 CHAPTER 7 Conclusion and Future Work As has hopefully been demonstrated in this thesis, the MINDAP has excellent potential as an inexpensive, compact, and sensitive elementselective detector for routine analytical measurements. Other compa rable methods are either very expensive, require special procedures for introducing samples, do not have high excitation temperatures or complicate the analysis with intense background spectral features. The MINDAP system possesses many of the qualities of an ideal emission source for routine trace elemental determinations: it is sustained through a low flow ( < 2 L/min) of replenishable and economical molecular nitrogen gas. The principal component of the power supply used to generate the plasma (a magnetron tube) is inexpensive and commercially available, being used in both medical diathermy units and microwave ovens. instrumentation and operation are low. Consequently, costs of both The plasma has both high thermal and excitation temperatures, suggesting its ability to decompose and excite the analyte. As a result, it provides low detection limits (ppb) and linear working curves for analyte introduced in the form of either an aerosol or atomic vapor. Available precision ( < 2%) is comparable to that of other more sophisticated and expensive techniques. Other qualities of the new plasma system include its ability to analyze both macro and micro aliquots of analyte and its relative freedom from interfering matrix effects. Although these characteristics are somewhat dependent on the operating conditions, the 200 MINDAP exhibits also dominant noise at low frequencies. For analytical measurements this feature means that longer integration times need not be used to obtain useful precision and sensitivity. In the development and characterization of a new technique, many new and interesting questions arise and a host of further research opportunities appear. In the following section, a few of these questions and ideas are listed, along with a brief description of their importance for the development and understanding of the nitrogen discharge in the presence of a microwave field. FUTURE WORK The plasma should be sustained and characterized using a smaller diameter torch (4 mm o.d.) which has already been constructed. A decrease in the size of the torch should increase its interaction with the high electric field centered within the microwave cavity. Such a smaller torch would be expected to produce an increased energy density in the plasma at any desired applied power or, alternatively, an analytically useful plasma at lower applied powers and nitrogen flow rates. A thorough investigation of the smaller plasma's analytical and physical figures-of-merit might be a significant scientific contribution. The MINDAP should be coupled to a flow-independent nebulizer (i.e., ultrasonic). This combination would enable the optimum operating conditions of the MINDAP to be determined because the introduction of analyte would then be independent of gas flow rates. 201 Such a gas-flow race independent nebulizer would also enable Che MINDAP Co be sustained as an annular plasma and Che performance of Che resulting system compared to work presented in this thesis. Another aid to understanding the plasma would be to isolate the energetic species in the plasma and determine the extent to which each contributes to the excitation of analyte. In afterglow studies two species have been found to be principally responsible for .excitation: vibrationally excited nitrogen molecules (N£ ) and ground-state nitrogen atoms (N). Chemical separation of either of these species in the discharge is possible through the addition of N^O (to remove ) or NO (to remove N) gas. Results from these experiments should help to identify and explain a mechanism by which analyte is excited in the plasma. An interesting study would be to look at the spatial-timedependent emission signal from the microarc atomizer in the MINDAP tail-flame with a gated vidicon detector. This investigation would provide spatial and temporal information about the distribution of atoms in the tail flame. possibly Results from this type of an experiment could suggest better methods for introducing and concentrating sample into the plasma. Identifying and determining the magnitude and effect of other noise sources will be essential for optimal operation of the MINDAP. Cross-correlation studies involving signal noise, from either background or analyte emission, with fluctuations from such operational 202 parameters as gas pressure, flow rate, and solution nebulization are necessary for understanding the origins of multiplicative noise observed in the MINDAP signals. A practical application of the MINDAP would be to couple a GC, capillary GC (through the central tube), or even an LC or HPLC system to it. The ease with which the two could be interfaced make this an attractive, inexpensive, yet sensitive detector for chromatographic separations. It would be extremely interesting and helpful to elucidate the characteristics of the plasma within the microwave cavity itself. Such measurements could be performed by viewing through one of the cooling holes drilled in the side of the cavity. Temperatures and analytical figures-of-merit could be obtained using a "remote sensing" fiber optic bundle to transmit the radiation from inside the cavity. This plasma system probably will not get the Hieftje "Seal of Approval" if a droplet generator system is not some how incorporated into it. Droplets, separated in time, could either be introduced into one of the gas flows entering the plasma torch and experience the breakdown region or be shot into the MINDAP tail flame and encounter the energy transferred from the main discharge region. Understanding atomization and vaporization processes in the MINDAP would be aided with this approach. 203 VITA Robert David Deutsch Born: August 16, 1956, Far Rockaway, New York. Undergraduate Education: University of Rhode Island Kingston, Rhode Island (1974-1978) Professional Societies: Society of Applied Spectroscopy Optical Society of America Positions: Research Associate, Indiana Univ. (1980-1983) Associate Instructor, Indiana Univ. (1978-1980) Undergraduate Research Associate, URI (1976-1978) Undergraduate Teaching Assistant, URI (1977-1978) PRESENTATIONS "Instrumental and Operational Characteristics of an AtraosphericPressure Microwave Induced Nitrogen Plasma," 8th annual Federation of Analytical Chemistry and Spectroscopy Societies meeting (Philadelphia, PA, September 1981) R. D. Deutsch and G. M. Hieftje. "Analytical Characteristics of an Atmospheric-Pressure Microwave Induced Nitrogen Plasma," 8th annual Federation of Analytical Chemistry and Spectroscopy Societies meeting (Philadelphia, PA, September 1981) R. D. Deutsch, J. P. Keilsohn and G. M. Hieftje. Near Infrared Emission from Nonmetal Atoms in a Helium Microwave Induced Plasma," 8th annual Federation of Analytical Chemistry and Spectroscopy Societies meeting (Philadelphia, PA, September 1981) J. E. Freeman, R. D. Deutsch and G. M. Hieftje. "Time Resolved Atomic Fluorescence in a Microwave Induced Plasma," Winter Conference on Plasma Spectrochemistry (Orlando, FL, January 1982) G. M. Hieftje, J. P. Keilsohn and R. D. Deutsch. "The Use of a Microarc Atomizer for Sample Introduction into the Inductively Coupled Plasma," Pittsburgh Conference (Atlantic City, NJ, March 1982 J. P. Keilsohn, R. D. Deutsch and G. M. Hieftje. "Interelement Interferences and Temperature Measurements in a Microwave-Induced Nitrogen Discharge at Atmospheric Pressure (MINDAP)," 24th Rocky Mountain Conference (Denver, CO, August 1982) R. D. Deutsch and G. M. Hieftje. 204 "Microarc Microsampling for a Mini ICP," 9th annual Federation of Analytical Chemistry and Spectroscopy Societies meeting (Philadelphia, PA, September 1982) R. D. Deutsch and G. M. Hieftje. "Microarc Sample Introduction for the Microwave Induced Nitrogen Discharge at Atmospheric Pressure (MINDAP)," 9th annual Federation of Analytical Chemistry and Spectroscopy Societies meeting (Philadelphia, PA, September 1982) R. D. Deutsch and G. M. Hieftje. "New Techniques and Instrumentation for Time-Resolved Fluorimetry," Pittsburgh Conference (Atlantic City, NJ, March 1982) G. M. Hieftje, G. R. Haugen, E. E. Vogelstein, R. E. Russo, R. D. Deutsch and J. P. Keilsohn. "Analysis of Noise in the MINDAP (Microwave Induced Nitrogen Discharge at Atmospheric Pressure)," 10th annual Federation of Analytical Chemistry and Spectroscopy Societies meeting (Philadelphia, PA, September 1983) R. D. Deutsch and G. M. Hieftje. "Vidicon Based Detection Systems for Spatially Resolved Studies of Atomic Sources," 10th annual Federation of Analytical Chemistry and Spectroscopy Societies meeting (Philadelphia, PA, September 1983) J. W. Olesik, R. D. Deutsch, G. M. Hieftje and J. P. Walters. "Analysis of Noise in the MINDAP (Microwave Induced Nitrogen Discharge at Atmospheric Pressure)," (Vancouver, Canada, October 1983) R. D. Deutsch and G. M. Hieftje. PUBLICATIONS J. P. Keilsohn, R. D. Deutsch and G. M. Hieftje, "The Use of a Microarc Atomizer for Sample Introduction into the Inductively Coupled Plasma," Appl. Spectrosc., 37_, 101 (1983). R. D. Deutsch and G. M. Hieftje, "Development and Characterization of a Microwave Induced Nitrogen Discharge at Atmospheric Pressure (MINDAP)," submitted to Applied Spectroscopy. R. D. Deutsch, J. P. Keilsohn, and G. M. Hieftje, "Analytical Characterization of a Microwave Induced Nitrogen Discharge at Atmospheric Pressure (MINDAP)," submitted to Applied Spectroscopy. R. D. Deutsch and G. M. Hieftje, "Temperature Measurements in the MINDAP System," submitted to Applied Spectroscopy. R. D. Deutsch and G. M. Hieftje, "Microarc Microsampling for a Mini ICP," in preparation. R. D. Deutsch and G. M. Hieftje, "Microarc Sample Introduction for the Microwave Induced Nitrogen Discharge at Atmospheric Pressure (MINDAP)," submitted to Applied Spectroscopy. 205 R. D« Deutsch and G. M. Hieftje, "Analysis of the Noise Components in the MINDAP System," submitted to Applied Spectroscopy.