NONDISPERSWE ATOMIC ABSORPTIOR AND ATOMIC FLUORESCENCE SPECTROMETRY Dissertation for the Degree of Ph. D. MICHIGAN STATE UNIVERSITY ‘ EUGENE'FRANK PALERMO 1973 -AA._‘- State rersity WP ABSTRACT NONDISPERSIVE ATOMIC ABSORPTION AND ATOMIC FLUORESCENCE SPECTROMETRY BY Eugene Frank Palermo There are several reasons why nondispersive systems are attractive for atomic absorption and atomic fluorescence spectrometry. First of all, elimination of a conventional grating monochromator decreases the expense of the spectro- metric system and increases the radiation throughput. A second attractive feature is that multielement analyses are facilitated with a nondispersive system. In this thesis two different approaches to nondispervise atomic flame spectrometry are presented. In Part I the resonance monochromator is described as an isolation device for atomic absorption flame spectrometry and the advantages of its narrow spectral bandpass are discussed. Part II describes a nondispersive atomic fluorescence instrument which operates in the time-division multiplexed mode for multielement analysis with a flame or non-flame atomizer. Eugene Frank Palermo I. Resonance Monochromators for Atomic Absorption Spectro- metry. Theoretical expressions have been derived for the use of resonance monochromators in atomic absorption flame Spectrometry to point out the parameters which influence the output of the resonance monochromator and to compare atomic absorption sensitivities to those with conventional monochromators. Experimental growth curves for a line source and a continuum source are shown to agree with these theoretical expressions. A demountable resonance monochro- mator has been designed and evaluated. Optimum parameters, such as applied power and filler gas pressure, are deter- mined for analytical applications. Experimental results indicate that.working curves with the resonance monochro- mator and a line source are less dependent upon self- reversal in the source than are working curves with a conventional monochromator. The narrow spectral bandpass of the resonance monochromator is shown to be useful in reducing spectral interferences. Continuum source atomic absorption sensitivities are shown to be comparable to sensitivities with a line source. Finally, the spectral bandpass for a cadmium resonance monochromator has been experimentally determined to be approximately 0.029 A for 11 a cadmium concentration of 10 atoms/cc in the resonance monochromator. Eugene Frank Palermo II. Nondispersive Multielement Atomic Fluorescence Spectrometry. An automated nondispersive atomic fluorescence system employing sequentially pulsed hollow cathode lamps, a sheathed burner, and a synchronous integrator has been designed for multielement analysis in the time-division multiplexed mode. Optimum parameters such as ON time and peak current for the sources, burner position, flow rate of sheath gas, and type of sheath gas, have been determined. Detection limits for the analysis of fig, Cd, Zn, and Pb are reported. The effect of type of flame and sheath gas flow rate on detection limit are determined. A comparison between the dispersive and nondispersive systems indicates that the two systems can achieve approximately equal detection limits. A comparison is also made between the sequential and time multiplexed modes of operation, and the advantages of each are discussed. A computer-controlled atomic fluorescence system with complete software control over instrumental parameters has been developed for the flame analysis of Mg, Co, Fe, and Ni. The computerized system is also used with a platinum—rhodium (90%-10%) non- flame atomizer for the analysis of Hg, Cd, Zn, and Pb. NONDISPERSIVE ATOMIC ABSORPTION AND ATOMIC FLUORESCENCE SPECTROMETRY BY Eugene Frank Palermo A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Chemistry 1973 To Carol, Mom, and Papa Jack ii ACKNOWLEDGMENTS The author wishes to express his gratitude to Michigan State University for his research and teaching assistantships, to Mr. Keki Mistry of the Department of Chemistry Glass Shop for construction of the resonance monochromator, to Mr. Charles Hacker of the Department of Chemistry Machine Shop for construction of the sheathed burner, and to Dr. Andrew Timnick for serving as second reader and for serving his homemade wine. Many thanks also go to Mr. Jack Zynger ("there's no question about it") for providing helpful discussions, to Mr. Akbar Montaser ("the Shah's favorite son") for writing the computer program for the non-flame atomizer, and to the rest of the Crouch group, past and present, for providing many enjoyable experiences, scientific and otherwise. Last, but certainly not least, the author wishes to express his deepest gratitude to Dr. Stanley R. Crouch ("El Boss") for providing the inspiration and drive that made this all possible, for being a conscientious and help- ful research director, and finally, for frequently providing the author 33 a_. with the ubiquitous nonaqueous solvent. iii TABLE OF CONTENTS Page LIST or TABLES . . . . . . . . . . . . . vii LIST OF FIGURES. . . . . . . . . . . . . ix LIST OF SYMBOLS. . . . . . . . . . . . . xiii PART I. RESONANCE MONOCHROMATORS FOR ATOMIC ABSORPTION SPECTROMETRY I. INTRODUCTION. . . . . . . . . . . . 1 II 0 HISTORICM O I O O O O O O O I O O 5 A. Atomic Absorption Spectrometry. . . . 5 B. Resonance Radiation from Atomic Vapor . 8 C. Applications of Resonance Monochro- mators . . . . . . . . . . . 10 D. Resonance Monochromators for Multi- element Analyses. . . . . . . . 13 E. Theoretical Sensitivities . . . . . 15 III. THEORETICAL RADIANCE EXPRESSIONS . . . . . 18 A. General . . . . . . . . . . . 19 B. Atomic Absorption Spectrometry. . . . 21 C. Atomic Fluorescence Spectrometry . . . 28 D. Atomic Absorption with a Resonance Monochromator. . . . . . . . . 30 IV. RESONANCE MONOCHROMATOR AND EXPERIMENTAL SYSTEM 0 I O O O I O I O O O I O 37 A. Design. . . . . . . . . . . . 37 B. Heating Assembly . . . . . . . . 41 C. Filler Gas . . . . . . . . . . 43 D. Experimental System . . . . . . . 45 iv V. VI. VII. II. III. IV. OPTIMIZATION OF APPLIED POWER AND FILLER GAS PRESSURE O O O I O O O O O O 0 A. Line Source Growth Curves . . . . B. Continuum Source Growth Curves . . C. Optimum Conditions for Analytical Applications. . . . . . . . ANALYTICAL APPLICATIONS. . . . . . . A. Line Source Working Curves. . . . B. Spectral Interferences . . . . . C. Continuum Source Working Curves . . CONCLUSION . . . . . . . . . . . A. Summary . . . . . . . . . . B. Recommendations . . . . . . . PART II. NONDISPERSIVE MULTIELEMENT ATOMIC FLUORESCENCE SPECTROMETRY INTRODUCTION . . . . . . . . . . HISTORICAL O O O O I O O O O O O A. Atomic Fluorescence Spectrometry. . B. Single Channel Nondispersive Systems C. Multielement Spectroscopic Systems . D. Multielement Nondispersive Systems . PULSED HOLLOW CATHODE SOURCES FOR ATOMIC FLUORESCENCE SPECTROMETRY . . . . . A. Previous Work on Pulsed Hollow Cathode Lamps . . . . . . . B. Circuitry for Pulsed Hollow Cathode Sources . . . . . . . . . C. Pulsed Source Characteristics. . . SHEATHED BURNER FOR NONDISPERSIVE STUDIES. A. Previous Work on Separated Flames . B. Burner Design . . . . . . . . C. Burner Parameters. . . . . . . Page 49 49 55 56 62 62 68 74 83 83 86 89 93 93 101 109 113 115 119 125 130 131 1.34 136 Page V. INVESTIGATION OF NONDISPERSIVE MULTIELEMENT ATOMIC FLUORESCENCE FLAME SPECTROMETRIC SYSTEMS O C O O O O O O O O O O 146 A. Sequential Multielement Atomic Fluorescence Analysis . . . . . 147 B. Multielement Atomic Fluorescence Spectrometry in the Time-Division Multiplex Mode. . . . . . . . 157 VI. MULTIELEMENT ANALYSIS WITH A COMPUTER- CONTROLLED ATOMIC FLUORESCENCE SPECTROMETER I O O C O O O I O O 16 7 A. Computer-Controlled Spectrometer . . 167 B. Flame Atomic Fluorescence Studies . . 172 C. Non-Flame Atomic Fluorescence Studies. 179 VI I O CONCLUS IONS O O O O I O O O O O O 1 84 A. summary 0 O O O O O O O O O O 184 B. Recommendations . . . . . . . . 187 LIST OF REFERENCES . . . . . . . . . . . 189 VITA. O O O O O O O O O O O O O O C 20]- vi LIST OF TABLES Table Page 1. Dependence of Resonance Monochromator Radiance Terms on n and n ' . . . . . . . . . 34 o o 2. Dependence of Cadmium Vapor Pressure on Temperature. . . . . . . . . . . . 45 3. Experimental System for Atomic Absorption . . 47 4. Half-Widths of Ca 4227.6 i and Cd 2288.0 3 Resonance Lines as a Function of Lamp Current . . . . . . . . . . . . . 68 5. Theoretical De endence of AAAC on g for AAD = 0.029 O I O O O O O O O O O 81 6. Slopes Calculated for Various Lamps in DC and Pulsed Modes . . . . . . . . . . 128 7. Relative Lamp Intensity as a Function of ON Time 0 O O O I O I O I I I O I O 129 8. Experimental System for Atomic Fluorescence. . 149 9. Detection Limits Obtained With Integrator Dispersive System. . . . . . . . . . 154 10. Detection Limits Obtained With Lock-In Amplifier System . . . . . . . . . . 155 ll. Detection Limits Obtained in the Time-Division Multiplex Mode. . . . . . . . . . . 164 12. Cadmium Detection Limits Obtained in Various . Flames . . . . . . . . . . . . . 165 13. Detection Limits Obtained With Computer- Controlled Multielement Flame A.F.S. System . 177 vii Table Page 14. Fluorescence Intensity Obtained With Multi- element and Single Element Solutions . . . 178 15. Detection Limits Obtained With Non-Flame . Software System . . . . . . . . . . 183 viii LIST OF FIGURES Figure Page 1. Block Diagrams for Atomic Absorption Spectro- metry (A.A.S.), Atomic Fluorescence Spectro- metry (A.F.S.), and Atomic Absorption Spectrometry Using a Resonance Monochromator (R.D.S.) . . . . . . . . . . . . . 20 2. Diagram of Demountable Resonance Monochromator . 40 3. Temperature of Heating Element as a Function of Power Applied to Resonance Monochromator . 42 4. Cadmium Atomic Concentration as a Function of Power Applied to Resonance Monochromator . . 44 5. Block Diagram of Resonance Monochromator Atomic Absorption System. . . . . . . . 46 6. Resonance Monochromator Output as a Function of Current Applied to Heating Element . . . 50 7. Experimental Growth Curve for Zinc Resonance Monochromator With Incident Line Source; Argon Filler Gas, Pressure = 6 Torr . . . . 52 8. Experimental Growth Curve for Zinc Resonance Monochromator With Incident Line Source; Argon Filler Gas, Pressure = 12 Torr. . . . 53 9. Optimum Applied Power as a Function of Argon Filler Gas Pressure . . . . . . . . . 54 10. Experimental Growth Curves for Cadmium Resonance Monochromator With Incident Line or Continuum Source; Argon Filler Gas, Pressure = 2 Torr . . . . . . . . . . 57 ix Figure 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. Zinc Resonance Monochromator Output as a Function of Argon Filler Gas Pressure . . . Relationship of the Intensities of the Resonance Lines for Highly Volatile Metals to Inert Gas Pressures . . . . . . . . . . . Atomic Absorption Calibration Curves for Zinc With Hollow Cathode Source. . . . . . . 0 Mg 2852 A Line Profiles for Different Lamp Currents. . . . . . . . . . . . . Calculated Analytical Curves for Resonance- Shaped Lines and Different Emission and Absorption Line Half-Widths . . . . . . Spectral Interference of Copper on Lead With Multielement Hollow Cathode Source Using 2170 A Resonance Line for Grating Monochro- mator. . . . . . . . . . . . . . Dependence of Lead Absorbance on Monochromator Spectral Bandpass for Multielement Line Source . . . . . . . . . . . . . Atomic Absorption Curves for Cadmium With Line and Continuum Source. . . . . . . . . Graphical Representation of the Effect of Increasing Atomic Concentration, no, on Absorption Half-Width in Flame for Cadmium . Dependence of 100 ppm Cd Solution Absorbance on Monochromator Spectral Bandpass for Continuum Source . . . . . . . . . . Circuit Diagram for Pulsing One Hollow Cathode Lamp . . . . . . . . . . . Diagram of Circuit to Drive FET. . . . . . Logic Control Circuit . . . . . . . . . Circuit for Sequentially Pulsing Four H.C.D.L.. Page 59 60 63 64 66 71 73 75 76 79 120 121 122 123 Figure Page 25. Shift Register Sequencing Circuit . . . . . 125 26. Intensity of H.C.D.T. as a Function of DC current I O O C I O O O O O O O O 127 27. Diagram of Sheathed Burner . . . . . . . 135 28. Lead Fluorescence Intensity and Flame Background as a Function of Burner Position. . . . . 138 29. Zinc Fluorescence Intensity and Flame Background as a Function of Burner Position. . . . . 139 30. Flame Background Spectra of Air/H2 Flame. . . 141 31. Lead Fluorescence Intensity and Flame Background as a Function of Flow Rate in Dispersive MOde 0 O O O O O I O O O O O O O 1 42 32. Mercury, Zinc, and Cadmium Fluorescence Intensity as a Function of Flow Rate in Dispersive MOde O O O O O O O O C O O O O O 144 33. Cadmium Fluorescence Intensity and Flame Background as a Function of Nitrogen Flow Rate in Nondispersive Mode. . . . . . . 145 34. Block Diagram of Atomic Fluorescence System. . 148 35. Circuit Diagram for Snychronous Integrator . . 150 36. Oscilloscope Trace of Cadmium Fluorescence Signal (Lower) and Corresponding Output of Integrate and Hold Circuit. . . . . . . 150 37. Oscilloscope Traces of Peak Current for Cadmium Hollow Cathode Lamp (Upper) and Corresponding Atomic Fluorescence Signal for 5 ppm Cadmium Solution. . . . . . . 152 38. Monostable Circuit Diagram for Multiplex Operation 0 O o o o o o o o o o o 160 39. Triggering Sequence for A/D Conversions in Multiplex Mode. . . . . . . . . . . 161 xi Figure 40. 41. 42. 43. 44. 45. 46. 47. 48. Oscilloscope Trace of Hg, Cd, and Zn Fluores- cence Signals Obtained in Multiplex Operation and CorreSponding Output of Integrate and Hold Circuit . . . . . . . . . . .. Cadmium Working Curve Obtained With Time Multiplexed Mode. . . . . . . . . . Block Diagram of Computer-Controlled Atomic Fluorescence System. . . . ,. . . . . Interface Circuit to Control ON Times of Lamps and Integrator . . . . . . . . Interface Circuit to Control Hold Time of Integrator for A/D Conversion . . . . . Magnesium Fluorescence Intensity as a Function of Argon Flow Rate in Dispersive Mode. . . Iron Fluorescence Intensity as a Function of Argon Flow Rate in Dispersive Mode. . . . Cobalt Working Curve Obtained With Computer- Controlled Dispersive System. . . . . . Oscilloscope Trace of Integrator Output for One Dimensional Array of Zinc Obtained With Time Multiplexed Nondispersive Non- Flame System . . . . . . . . . . . xii Page 162 163 168 170 171 174 175 176 181 ll LIST OF SYMBOLS Absorbance for atomic absorption with a continuum source. Absorbance for atomic absorption with a line source. Absorbance for atomic absorption with a resonance monochromator and continuum source. Absorbance for atomic absorption with a resonance monochromator and line source. Atomic absorption flame spectrometry using a continuum source. Atomic absorption flame spectrometry using a line source. Atomic fluorescence flame spectrometry using a continuum source. Atomic fluorescence flame spectrometry using a line source. The a-parameter /£n2 (Ale/AID), dimensionless. Radiance of atomic fluorescence produced using a continuum source, W cm-2 sr-l' Radiance of atomic fluorescence produced using a line source, W cm’zsr-l' Spectral radiance of continuum source at wave- length 90, W cm-Zsr-lnm-l‘ Radiance of line source, W cm—zsr-l. xiii —RAC -RAL —RFC -RFL -TCA —TL £8 Radiance of atomic absorption in resonance monochromator using a continuum source, W cm"2 '1 sr Radiance of atomic absorption in resonance monochromator using a line source, W cm-zs -l. r Radiance of atomic fluorescence produced in resonance monochromator using a continuum source, W cm-zsr-l. Radiance of atomic fluorescence produced in resonance monochromator using a line source, Spectral radiance of continuum source at wave- length *0 that is not absorbed by atoms in flame, -zsr-lnm-l. W cm Radiance of line source that is not absorbed by atoms in flame, W cm-zsr-l' Self-absorption factor in flame as defined by Winefordner st 31. (30). Self-absorption factor in resonance monochromator, dimensionless. Atomic absorption coefficient in flame for pure Doppler broadening defined at 1°, cm‘l. Atomic absorption coefficient in resonance monochromator for pure Doppler broadening defined at 10, cm . Path length of atoms in flame in direction of source, cm. resonance monochromator cm. Path length of atoms in in direction of source, Path length of atoms in flame in direction of detector, cm. Path length of atoms in resonance monochromator in direction of detector, cm. Ground state concentration of analyte atoms in flame, atoms cm-3. xiv BU 6(a,'\3) 6(a,5') 6(a,3)' 6'(a,u) AAA AAD A1 ' AA Ground state concentration of analyte atoms in resonance monochromator, atoms cm’ . 0 Spectral bandpass of monochromator, A. Quantum yield for resonance absorption-resonance fluorescence in flame, dimensionless. Quantum yield for resonance fluorescence in resonance monochromator, dimensionless. Fraction of radiation absorbed when using a continuum source in atomic absorption, dimension- less. Fraction of radiation absorbed when using a line source in atomic absorption, dimensionless. Collection efficiency of resonance monochromator, accounts for any radiance which does not reach the atoms in resonance monochromator, dimension- less. Free-atom fraction, dimensionless. Voigt profile for absorption line in flame with u ranging from zero to AAS/AAA, dimensionless (30). Voigt profile for absorption line in flame with v' ranging from zero to S/AAA, dimensionless. Voigt profile for absorption line in flame with U ranging from zero to AAD'/AAA, dimensionless. Voigt profile for absorption line in resonance monochromator, dimensionless. Absorption line half-width in flame, cm. Doppler half-width of resonance line in flame, cm. Doppler half-width of resonance line in resonance monochromator, cm. Source line half-width, cm. Nebulization efficiency factor, dimensionless. XV 2/2n 2 (1-10)/AAD, dimensionless. Solid angle of radiation collected from source by entrance optics in atomic fluorescence, sr. Solid angle of radiation collected from source by entrance optics of resonance monochromator, 8r. xvi PART I RESONANCE MONOCHROMATORS FOR ATOMIC ABSORPTION SPECTROMETRY I . INTRODUCTION Resonance radiation from a hollow cathode discharge tube was first observed in 1959 by Russell and Walsh (1), who noted that an appreciable amount of the metal sputtered from the cathode was in the form of an atomic vapor. Since this atomic vapor is at low temperature and pressure, the conditions are those required for the fluorescence, or resonance radiation, of any absorbed radiation, i.e., parameters may be chosen to give a high quantum yield. In 1965 Sullivan and Walsh (2) used this principle to con- struct their first resonance detector, an apparatus designed for spectral isolation, in which an atomic vapor is produced by cathodic sputtering. Later articles reported on the design and application of resonance detectors which utilized electrical heating to produce atomic vapor clouds for the more volatile elements. There are several reasons why the resonance monochromator (resonance detector) is attractive for atomic absorption flame spectrometry. First, any unwanted flame background emission which reaches the photomultiplier transducer should be very low because the photomultiplier is positioned at right angles to the incident radiation source and does not view the flame directly. The out- standing characteristic of the resonance monochromator is that the spectral bandpass is determined by the line width of the fluorescent radiation; typically about 0.01-0.03 A. As a result of this low spectral bandpass, a second attractive feature is the virtual elimination of spectral interferences. Interferences due to the absorption of impurity or filler gas lines in the hollow cathode by atomic Species in the flame should be virtually eliminated, while interferences due to absorption by molecular species in the flame should be reduced. Another attractive feature of the resonance mono- chromator is that it cannot be easily put out of adjustment because of changes in room temperature or atmospheric pressure, and is virtually unaffected by mechanical vibrations. Thus, such monochromators may prove more useful than dispersive type monochromators under extreme conditions. There is also no adjustment corresponding to tuning in to a given line. It has already been pointed out that the effective spectral bandpass of an atomic absorption spectrometer with a resonance monochromator should be determined by conditions prevailing in the resonance monochromator. Therefore, working curves should be independent of the conditions under which the radiation source is operated. This should facilitate the maintenance of reproducible working curves in routine analytical applications. Another potential use of the resonance monochro- mator is in conjunction with a continuum radiation source. With a conventional dispersive monochromator, continuum sources have not proved extremely useful in atomic absorption because absorption sensitivities are usually lower than with line sources and dependent on the monochromator spectral bandpass, unless techniques such as selective modulation are used. Demountable resonance monochromators, or single resonance monochromators in series, on the other hand, could prove very useful with continuum sources. Finally, resonance monochromators may be useful for multi- element analyses. Because of the many potential advantages of the resonance monochromator, a complete characterizatiOn of the experimental behavior and the expected behavior based on pertinent theory is desirable. In Chapter III of Part I, the theoretical basis for the application of the resonance monochromator in atomic absorption flame spectrometry is presented. Equations for the spectral radiance output of the resonance monochromator with both an incident line source and continuum source are derived to point out its unique advantages and possible limitations when compared to a conventional dispersive monochromator. A demountable resonance monochromator with provision for rapid inter- change of elements has been designed and evaluated. Experimental data are presented for a resonance monochro— mator-line source combination and a resonance monochro- mator-continuum source combination to demonstrate how well the fluorescence signal vs. atomic population curves. qualitatively agree with growth curves predicted by theory. Operating variables, such as applied power and filler gas pressure, were varied to determine optimum operating conditions for a demountable zinc resonance monochromator. A comparison of atomic absorption working curves with a resonance monochromator and a conventional monochromator is made to test whether resonance monochromator working curves are less dependent on source operating conditions than are working curves obtained with a dispersive mono- chromator. A study of the interference of copper on the atomic absorption of lead was undertaken to demonstrate the usefulness of the resonance monochromator in reducing spectral interferences. Finally, working curves obtained with a cadmium line source—resonance monochromator and cadmium line source-grating monochromator combination are compared to a working curve obtained with a continuum source-cadmium resonance monochromator. Atomic absorption data obtained with a continuum source and grating mono- chromator are then used to determine the effective spectral bandpass of the cadmium resonance monochromator. II. HISTORICAL A. Atomic Absorption Spectrometry Atomic absorption flame spectrometry (A.A.S.) was first introduced as a method for chemical analysis in 1955 by A. Walsh (3), who pointed out that this technique has several important advantages over the then widely used atomic emission spectrometric (A.E.S.) methods. The usual atomic absorption spectrophotometer consists of a line source of the element of interest (i.e., hollow cathode tube), a flame, a monochromator, and a photomultiplier transducer. The radiation from the source is passed through the flame, into which the sample to be analyzed has been aspirated. The ground state atoms in the flame absorb part of this radiation at a wavelength character- istic of the element, and the transmitted radiation, which is isolated by the monochromator, is then incident on the transducer. The major difference between A.E.S. and A.A.S. is that A.A.S. employs selective radiational excitation of atoms via an external source, whereas A.E.S. employs collisional excitation within the flame. One of the main advantages of this atomic absorption method over flame emission is that A.A.S. is less susceptible to spectral interferences which are the result of overlap of spectral lines. L'vov (4) has shown that the probability that lines will be superimposed is on the average two orders of magnitude less in atomic absorption measurements than in emission spectroscopy under the very best conditions. The spectral interferences observed in A.E.S. arise because collisional activation in flames produces a distribution of excited-state atoms, since flames can be assumed to be in local thermodynamic equilibrium (5). Because A.A.S. normally involves selective excita— tion with a line source, only transitions which arise from the ground state are usually observed (resonance lines). The shape of the atomic absorption resonance line is determined by (a) the natural width of the line due to the finite lifetime of the excited state; (b) the Doppler contour due to the motions of atoms relative to the observer; (c) pressure broadening, either by atoms of the same kind (resonance or Holtzmark broadening) or to foreign gases (Lorentz broadening); and (d) Stark broadening due to external electric fields or to neighboring charged particles. The natural width of an atomic spectral line 0 4 A. The Doppler width, at is usually on the order of 10- o c a temperature of 2000 °K, is on the order of 0.01 A. The accurate measurement of the profile of such a line would require a very high resolution monochromator. In the past this difficulty was overcome by using the method of total absorption, a method based on measuring the absorption coefficient of the resonance line as a function of frequency. This method has the advantage that the total absorption does not depend on the resolving power of the monochromator, but suffers from the disadvantage of giving a complicated relation between n, the number of atoms, and f, the oscillator strength. A more attractive method was to measure the absorption coefficient at the center of the line, using a narrow line source which emits lines with a half-width much smaller than the absorption line. As sources Jones and Walsh (6) therefore proposed low-pressure hollow cathode lamps, which produce extremely narrow resonance lines of the element of interest. If such a narrow line source is used, it is no longer necessary to use a monochromator with a spectral bandpass on the same order as the half-width of the absorption line. The requirement of the monochromator now is merely to isolate the resonance line from the other lines emitted by the source. Utilizing this principle, Walsh 22 31. (7, 8) con-w structed a simple atomic absorption spectrophotometer in which a sample was aspirated into a Meker-type burner to produce the necessary ground state analyte atoms. Later, Gatehouse and Walsh (9) used a similar system in which the ground state atoms were produced by a non—flame method, cathodic sputtering. In recent years atomic absorption spectrometry has grown to become an indispensable analytical tool. Several thousand publications have dealt with this subject, and a large number of commercial instruments are now available. The technique of atomic absorption is so widely used, that several books have been written (10-17). B. Resonance Radiation From Atomic Vapor Atomic absorption spectrometry requires the use of a source which emits the spectrum of the element to be determined. The source most commonly used is the hollow cathode tube which consists of an anode and a hollow cathode (either composed of or lined with the desired material) sealed in a glass tube containing one of the rare gases. A dc power supply produces rare gas ions which strike the cathode causing free metal atoms to be sputtered off. These atoms are then excited by collision with excited rare gas atoms, thus producing the emission of an intense line spectrum from the inner region of the hollow cathode, since it is here that the metal atom concentration is high. There is also an appreciable concentration of metal atoms in the region between the electrodes and also extending several millimeters beyond the anode. Therefore, this atomic vapor can absorb part of the radiation from the inner region of the cathode and re-emit it over the entire solid angle. This re-emitted radiation, which may be viewed at right angles to the axis of the hollow cathode tube, consists primarily of resonance radiation (fluo- rescence). Therefore, an atomic vapor produced in this manner could serve to isolate the resonance line of the element of interest from an external hollow cathode tube, if it were placed after the flame, i.e., in place of the grating monochromator. Such a method would provide a resolution on the order of 0.01 A and would be highly specific. The fact that the resonance radiation is weak is compensated to some extent by the large entrance aperture and the lack of the radiation losses at mirrors and gratings. Sullivan and Walsh (2) used this principle to construct their first resonance monochromator. In their arrangement radiation from a high intensity hollow cathode lamp (18) was focussed on a resonance lamp in which the atomic vapor was produced by cathodic sputtering. Some of the resonance radiation, which is emitted in all directions, was collected through a Side arm and directed to a photo- electric detector. One application of this resonance monochromator is in atomic absorption spectrometry. For many analyses it is possible to replace the conventional monochromator with a resonance monochromator. In their preliminary investigation Sullivan and Walsh (2) employed resonance monochromators in their atomic absorption spectrophotometer to obtain analytical data for Mg and Cu. The particular advantage of this sputtering tech- nique is that it is, in principle, applicable to all 10 metallic elements, irrespective of melting point, since the atomic vapor is produced without the necessity of heating the metal. The main disadvantage of this technique is that the electrical discharge necessarily results in the emission of radiation in the visible and ultra-violet regions of the spectrum, since this is the same mechanism of excitation that occurs in a hollow cathode tube. Hence, the fluorescence observed occurs on top of emission lines which may cause interferences. C. Applications of Resonance Monochromators Since, for the low melting-point metals, the required atomic vapor can be produced by heating and with- out the generation of any appreciable radiation, Sullivan ' and Walsh (19) reported on the use of thermal resonance lamps as monochromators in the determination of calcium, magnesium, sodium, potassium, and lead in blood serum by atomic absorption flame spectrometry. Working curves obtained with the resonance monochromators were shown to be comparable to respective working curves obtained with a conventional grating monochromator. Sullivan and Walsh (20) also reported on the appli- cation of a resonance monochromator for measuring the molecular absorption of solutions at specific wavelengths, as is required, for example, in colorimetry. The resonance monochromator isolates the resonance line which passes through the solution and is then measured photo-electrically. 11 Thermal monochromators were constructed for Zn (2138 A), Cd (2288 A), Hg (2537 A), Pb (2833 A). Mg (2852 A), Tl (3776 3), Ca (4227 3), Sr (4607 3) Ba (5535 K), Na (5890, 5896 R), Li (6708 i), and x (7645, 7699 i). Cathodic sputtering resonance monochromators were used for Cu (3247, 3274 3» Ag (3231, 3383 3), Au (2423, 2676 3), Ni (2320, 3525 A), and Be (2349 A). It can be seen that the resonance lines are distributed at fairly regular intervals throughout the ultra-violet and visible regions of the spectrum, and thus are useful for making absorption measurements at these wavelengths. A resonance monochro- mator in this type of application has the advantage that the wavelength it selects is invariable regardless of the external conditions. It can also be readily combined with other resonance monochromators for simultaneous measure- ments at two or more wavelengths. An example of this type of application is the detection and determination of proteins in solution. Proteins show characteristic absorption in the region 2700-2900 A due chiefly to the presence of aromatic amino acids. The resonance line of magnesium is at 2852 A, and it has been demonstrated that proteins can be determined by isolating this line using a resonance monochromator (20). Young, Timms, and Sullivan (21) employed a nickel resonance lamp for the determination of nickel in the 0.05-2.0 per cent range in mineral samples. The nickel atomic vapor generated by cathodic sputtering afforded a 12 convenient method for isolating the resonance lines at 3415, 3515, and 3524 5. These lines can be usefully employed for atomic absorption measurements on mineral samples where the nickel concentration is too high for measurements with the more sensitive resonance line at 2320 R. Rawling and Sullivan (22) used a copper resonance monochromator in which the copper vapor was produced by cathodic sputtering. This method was employed in an atomic absorption spectrophotometer for the determination of the copper content of ore samples. Satisfactory agreement was obtained in a comparison made between a conventional and a resonance monochromator for a range of solutions in which the copper concentration varied between 4 and 92 ug/ml. Boar and Sullivan (23) used a calcium resonance monochromator in which the calcium vapor was produced by indirect electrical heating of calcium metal. Their instrument was used under routine conditions for the determination of calcium in brown coal. The values of concentration found using a resonance monochromator were not significantly different from those obtained using a conventional monochromator for the range of 0.6 to 30 ug/ml. These authors (24) then extended the use of a resonance monochromator to the determination of magnesium in brown coal. The magnesium atomic vapor was again produced by electrical heating. Working curves with a conventional 13 and a resonance monochromator were obtained for a sample concentration range of 0.5 to 25 ug/ml. A thermal lithium resonance monochromator was employed by Bowman (25) for the determination of lithium in blood serum. His results were reproducible to within i 0.2 ug/ml of lithium in the serum and the limits of sensitivity attainable for samples diluted 1:10 were 0.3 ug/ml of lithium with a conventional instrument and 0.6 ug/ml with the resonance monochromator. D. Resonance Monochromators for Multielement Analyses Sullivan and Walsh (26) have also described various applications of resonance monochromators for multielement analysis. Since the resonance radiation is emitted in all directions, it is possible in some cases to produce an atomic vapor of two or more elements in one detector, and to isolate the respective resonance line(s) for a given element at the exit windows by use of appropriate filters. The primary radiation source used with this type of resonance monochromator could be a multielement hollow cathode lamp, an array of lamps on the focal curve of a concave grating, or a continuum source. For example, if one uses a calcium- magnesium combination, the resonance radiation will consist almost entirely of the magnesium resonance line at 2852 A and the calcium resonance line at 4227 A. It is then a simple matter to isolate each of these resonance lines from the other by appropriate absorption filters. 14 Another possibility is to place resonance monochro- mators in series. It is obviously a simple matter to interchange detectors in this type of system according to the elements to be determined. However, this arrangement has the same limitations as a conventional multielement system, in that the light path through the flame is the same for all elements. Sullivan and Walsh (26) overcame this difficulty in an instrument constructed for the determination of calcium and magnesium. For example, if the concentrations of calcium and magnesium are such that it is not possible to use the same absorption path length for the two elements, the appropriate sensitivities may be obtained by using the full length of a long path length burner for calcium and using the width of the burner for magnesium. Sullivan and Walsh (27) described another way of achieving different absorption path lengths for different elements in an instrument designed for the simultaneous determination of copper, zinc, silver, nickel, and lead in ores. A circular flame was employed in which the path length through the flame was selected to give the appropri- ate sensitivity for the particular element being determined. Walsh (28) in collaboration with McDonald, Lloyd, and Sullivan designed an instrument for the simultaneous determination of copper, silver, nickel, zinc, lead, and cobalt in ores. Perforations in the burner top were arranged so as to provide the appropriate absorption path 15 length, i.e., adequate width of flame, for each element. The resonance radiation from six resonance monochromators was directed to six photomultiplier transducers. Sputtering- type monochromators were used for all the elements except zinc and lead, for which thermal—types proved superior. Their instrument was capable of providing analytical results with an accuracy of 5 per cent. For routine operation the instrument was interfaced to a computer designed to carry out an analysis for up to six elements every twelve seconds. E. Theoretical Sensitivities The simpler the resonance spectrum for a particular element the closer will be the comparison of sensitivity (ppm/1% absorption) between a resonance monochromator A.A. instrument and a conventional A.A. spectrometer. Con- versely, the more complex the resonance spectrum, the greater will be the difference between the sensitivities achieved by the two types of instruments. For example, elements such as Be, Mg, Ca, Sr, Ba, Zn, Cd, and Hg have a relatively simple spectrum. For these elements the resonance spectrum consists of one resonance line corresponding to the transition from the 1 1 Pl state to the 0 magnesium, this is the resonance line at 2852 A, and the lowest So ground state. In the case of resonance radiation, fluorescence, consists almost entirely of this line. However, for mercury, the transition which 16 o o o 0 gives rise to the line at 2537 A, 3P1 + 150, is fifty times weaker than that of the line at 1849 A, 1P1 + 180. The measured intensity of these lines is, of course, critically dependent on the spectral response of the photomultiplier transducer. For calcium, the Varian Techtron investigators (29) reported a sensitivity of 0.1 ppm with the conventional Varian Model AA-4 atomic absorption spectrophotometer. With the resonance monochromator model AR-200, a sensitivity of 0.12 ppm was obtained. For magnesium, the sensitivities were 0.01 ppm with the model AA-4, and 0.012 ppmwwith the AR-ZOO. For elements such as Li, Na, K, Cu, Rb, Ag, Cs, and Au, the bulk of the atoms are in the lowest 281/2 state and the strongest absorption lines are due to transitions 2 2 to the lowest P3/2 and Pl/2 states. Therefore, the resonance spectrum for these elements consists almost entirely of the doublet 2P3/2, 2P1/2 + 281/2, as typified by the well known sodium doublet. For the more complex spectra of lead, Varian investigators obtained sensitivities of 0.2 ppm with the AA-4, and 0.8 ppm with the AR-200. For other elements the energy level diagram becomes extremely complex because of the multiplicity of low lying levels, all of which will be populated to some degree. However, it is sometimes possible, as for example in the case of nickel, to use a simple absorption filter to restrict absorption to a given region and thus greatly simplify the resonance spectrum, e.g., the group of lines 17 in the region of 3400 A can readily be isolated. Sullivan and Walsh (26) stated that the sensitivity thus obtained for nickel is five times less than that obtained when using the most sensitive nickel line at 2320 A. Varian investi- gators obtained a sensitivity for nickel of 0.1 ppm with the AA-4 and 0.5 ppm with the AR-200. For iron, the sensitivities obtained were 0.1 ppm‘with the AA-4 and 1.0 ppm with the AR-200. As can be seen, as the fluorescence spectrum increases in resonance lines, the difference in sensitivities obtained with a resonance and conventional monochromator increases. Since the resonance monochromator produces fluorescence radiation corresponding to all the resonance lines of an element, the absorbance sensitivity for that element will depend on the relative oscillator strengths of all the resonance lines and the spectral response of the photomultiplier transducer. III. THEORETICAL RADIANCE EXPRESSIONS Atomic absorption flame spectrometry with a resonance monochromator involves normal flame absorption of a line or continuum primary radiation source. The transmitted radiation is then incident on the resonance monochromator in which absorption and re-emission, or fluorescence, of the resonance radiation takes place. This resonance radiation is then directed to a photomultiplier transducer which is positioned at right angles to the primary radiance beam. Hence, the resonance monochromator serves to isolate the resonance line of the element of interest. The radiance output of the resonance monochro- mator is dependent on both the number of ground state atoms 0 o , and the number of in the resonance monochromator, n ground state atoms in the flame, no. In this chapter, the radiance expressions for atomic absorption and atomic fluorescence are reviewed (30), and complete expressions for the radiance output of the resonance monochromator with both an incident line source and continuum source are derived. These theoretical expressions have been derived to determine the exact dependence of the radiance output on nO and no', and to facilitate optimization of experi- mental parameters. 18 19 A. General Block diagrams for the three systems that will be considered in this discussion, Atomic Absorption Spectro- metry (A.A.S.), Atomic Fluorescence Spectrometry (A.F.S.), and Atomic Absorption Spectrometry with a resonance monochromator (R.D.S.) are shown in Figure 1. In A.A.S. the radiance of a line source, 2', or the spectral radiance of a continuum source, 2C1 , is absorbed by the ground state atoms in the flame,oor atomic vapor cell, and the respective transmitted radiances,.gTL or ETCAO' are isolated by a grating monochromator. If a line source is used the monochromator need only isolate the resonance line from nearby nonresonance lines. Therefore, only a medium resolution monochromator is required, i.e., l to 10 A spectral bandpass. However, if a continuum source is used, the absorbance is dependent on the spectral band- pass of the monochromator, and therefore a high resolution monochromator is required. In A.F.S. the same absorption phenomenon occurs in the flame, but the respective fluorescent radiances, EAFL and g’ are isolated by the grating monochromator. In AFC' this case, the atoms act as their own monochromators and therefore, only a low resolution monochromator is required to reduce the flame background. Due to this low resolution requirement, it is possible to perform an atomic fluores- cence measurement without the benefit of a monochromator, 20 1 AHAJS. ,____J\____‘ LINE OR §L ATOMIC _B_.n_ CONTINUUM ' VAPOR SOURCE L Ecxo CE L g—TCAO D I A.F.S. f—_—-L————\ LINE OR _B.L ATOMIC CONTINUUM VAPOR _L‘ SOURCE 8 CELL §AFC g-AFL 2 2' ROS. , L l r I ‘ LINE OR B ATOMIC B commuum L VAPOR TL RESONANCE L' SOURCE 8 CELL B MONOCHROMATOR —c>‘0 . *‘TCXO QRFC §~RFL Figure 1. Block Diagrams for Atomic Absorption Spectro- metry (A.A.S.), Atomic Fluorescence Spectro- metry (A.F.S.), and Atomic Absorption Spectro- metry using a Resonance Monochromator (R.D.S.). 21 i.e., nondispersively, as will be shown in Part II of this thesis. If the principle that atoms act as their own monochromators is applied, then an atomic absorption. system employing a resonance monochromator results as is shown in Figure 1. This nondispersive system for atomic absorption spectrometry has a very high resolution, on the order of 0.01 A, and therefore may be used in conjunction with a continuum source. B. Atomic Absorption Spectrometry In A.A.S. there are two possible means of ex- citation of the analyte atomic vapor in the flame. One can use either a line source-monochromator system (A.A.L.) or a continuum source-monochromator system (A.A.C.). It is assumed that for A.A.L. the source line half-width, AAS, is much smaller than the absorption half-width in the flame, AAA: and for A.A.C. the spectral bandpass of the monochro- mator, g, is much larger than AKA. The total radiance absorbed, from a line EAAL ' source, with a radiance of 2., by an atomic vapor with an absorption coefficient, k is given by 0' EAAL = Ai EL [1 - exp {-k0£6(a,v)}] dA (1) S where all terms are defined in the list of symbols. Since Al is less than AAA and B is a constant, the expression S —L for EAAL may be reduced to 22 EAAL = EL [1 - exp {-ko£6(a,5)}] (2) where U of the Voigt profile is an average over the range from zero to AAS/AAA. The radiance that is not absorbed is the transmitted radiance, ETL' if scattering in the flame is negligible ETL= EL [exp {-ko£6(a,v)}] (3) For low concentrations of analyte, i.e., low values of kok, since k0 is proportional to no EAAL = EL [ko£6(a,u)] (4) while for high concentrations of analyte EAAL 2 EL (5) If a continuum source is used, the total radiance absorbed is given by EAAC = : ECAO [l - exp {-k0£5(a,v)}] dl (6) where the limits of integration are now given by the spectral bandpass of the monochromator. Since §_>> AAA, there is no dependence on the Voigt profile, 6(a,v), and the absorbed radiance is simplified to gAAC = ECAO I1 - exp (-k0£)] AAD (7) 23 Since the absorbed radiance plus the transmitted spectral radiance must equal the Spectral radiance of the source, if scattering is assumed negligible (8) s B = B + s B - -CXO -AAC —--TCXO we arrive at the following expression for the transmitted spectral radiance for a continuum source B = 33qu [§_ - AAD {1 - exp (-k0£)}] —TC>\O s (9) The measured quantity in A.A.S. is seldom the radiance absorbed, g. or B but rather the fraction of AAL —AAC' radiance absorbed, a B _ —AAL OLAAL " T— (10) -L B —AAC a = (11) AAC ECA E The absorance, A, is then defined by A E - log (1 - a) (12) Therefore, for a line source the absorbance is given by BL[1 - exp {-k0£5(a,v)}] AAAL = - log ES: (13) and for a continuum source by A0 [1 - exp (-k02)]AAD (14) A = — log AAC Eel g 0 24 Equations (13) and (14) may be simplified to _ k0£6(ar;) AAAL ‘ “—2.303 ‘15) g’- [l - exp (-ko£)]AlD AAAC = - log 3 (16) Equation (15) predicts that when a line source is used in A.A.S. the absorbance is linearly dependent on kol, and hence no. This linearity will continue until high con- centrations are reached, at which point the absorption half- width in the flame increases. When this occurs the Voigt profile factor, 6(a,3), decreases and hence the absorbance no longer increases linearly with no. Equation (15) assumes that the spectral bandpass of the monochromator isolates only the resonance line; spectral interferences are discussed in Chapter VI. Equation (16) describes the dependence of absorbance on the spectral bandpass when a continuum source is used. Only when the Spectral bandpass approaches the Doppler line half-width in the flame, AID, will absorption sensi- tivities obtained with a continuum source approach absorption sensitivities obtained with a line source. However, as §_approaches AAD, the Voigt profile may no longer be neglected as is shown in Equation (17) A _ kQ£6(a,U') AAC ‘ 2.303 (17) 25 where U' is an average over the range from zero to g/AAA. As the spectral bandpass decreases, the continuum source approaches a line source, and U' approaches 3, i.e., Equation (17) approaches Equation (15). When this con- dition is reached, absorption sensitivities with a line source will be identical to absorption sensitivities with a continuum source. For the case where a spectral continuum is used, Winefordner (31) has shown that as the spectral bandpass of the monochromator approaches the absorption line width of the atoms in the flame, the sensitivity of measurement will increase almost linearly with a decrease in the spectral bandpass. Fassel and co-workers (32, 33) and Winefordner and co-workers (34, 35) have shown that with a monochromator capable of a spectral bandpass, §_: 0.2 A, the absorption signal measured with a continumm source is not much smaller than that measured with a line source. This statement is true if the absorption half-width of the analyte atoms in the flame is on the order of 0.1 A. However, if the absorption half-width of the atoms in the flame is less than 0.1 A, then the absorption signal obtained with a line source will be greater than the absorption signal Obtained with a continuum source if the monochromator spectral bandpass is 0.2 A. In deriving expressions for atomic absorption, Vickers and Winefordner (36) noted that the sensitivity for a given determination will be greater when using a 26 very narrow line source than when using a continuum source, since 6(a,U) is generally greater than the corresponding term for a continuum source, especially if the monochro- mator spectral bandpass is greater than 0.2 A. For example, for the Cd 2288 A resonance line produced by cadmium in a Hz-O2 flame at 2650 °K, 6(a,3) with an a value of about 0.50 (38) and a U of 0.0 is 0.6 (35). If §_ is assumed to be 0.2 A, the corresponding term for the continuum source is 0.05. As long as the source line half-width, A13, is less than or equal to the absorption line half-width, and as long as the collisional half-width of the absorption line does not exceed the Doppler half-width by more than about five times, 0(a,U) will be greater than 6(a,U'), and a narrow line source will give greater sensitivity than a continuum source, even with a monochromator of medium resolution. For example, a 0.5 m Ebert spectrometer has a minimum spectral bandpass of about 0.2 A in the first order. Frank, Schrenk, and Meloan (37) also reported some- what lower sensitivities with a continuum source than with a hollow cathode line source. Furthermore, they reported somewhat greater interferences with the continuum source than with the line source because of the restrictions on the spectral bandpass of the monochromator. One method of reducing the effective spectral bandpass of the spectrometer system is to use modulation techniques such as the method of selective flame signal 27 modulation, S.F.S.M., (39, 40) or selective modulation (41-44). In S.F.S.M. only the actual analytical infor- mation of interest in the optical spectrum, i.e., in the flame, is converted to an ac signal. This technique uses a perodic piezoelectrically induced deformation of one or two fluidic channels to the solution nebulizer, or a mechanically simpler channel device which generates a regular ac ripple on the solution flow into the flame. In atomic absorption measurements with S.F.S.M. the ac signal detected is a measure of the changes in the radiance emitted by the continuum source and the radiance absorbed per half-cycle of modulation. Since non-absorbed radiation is undetected, the spectral bandpass observed is determined by the width of the absorption line in the flame. The latter technique, selective modulation, is a well known method of isolating a given region of a spectrum by means of periodic interposition at a given frequency of an absorbing medium between a radiation source and a photomultiplier transducer. This produces a modulated signal at all wavelengths that lie within the region of absorption of the medium. Bowman £5 21. (41) have recently described a method of modulating atomic resonance lines that uses a sputtering cell to produce the necessary atomic vapor. With this technique the spectral bandpass is determined by the width of the absorption line of the pulsating cloud of atomic vapor in the sputtering cell. 28 The ability of these modulation techniques to narrow effectively the spectral bandpass in absorption measurements can be advantageous in the application of low resolution monochromators to continuum source atomic. absorption spectrometry. However, the larger dc signal of the nonabsorbed spectral continuum within the resolution of the monochromotor is still impingent on the photomultiplier, and hence produces shot noise. The resonance monochromator, on the other hand, alleviates this problem but still maintains the narrow spectral bandpass, if scattering is negligible, because the photomultiplier is positioned at a 90° angle to the spectral radiance of the continuum. C. Atomic Fluorescence Spectrometry In A.F.S. there are again two possible means of exciting the analyte atomic vapor in the flame, with a continuum source (A.F.C.), and with a line source (A.F.L.). In A.F.C. it is assumed that the source half-width, A15, (or spectral bandpass of the excitation monochromator if one is used) is much greater than the absorption line half- width, AAA, and in A.F.L. it is assumed that AAS < AAA. Theoretical expressions for atomic fluorescence spectro- metry have previously been derived by Winefordner 23 31. (30, 45—52) and other authors (53-56). Equations (18) and (19) give the fluorescence radiance with a line source, B -AFL' respectively and Wlth a continuum source, EAFC' 29 B _ (“A L —AFL — T1?) (1') Y -B-AAL f5 (18) Q _ A L EAFC ’ (T1?) (I) Y EAAC f5 (19) where fS is the self-absorption factor 5 1 - e -k ms ' A f5 ___ [ fm‘XP { 41- (apV)}] d (20) o ko£6(a,v) d1 For low values of no, fs is given by fs _ w ko£6(a,y) dX = 1 (21) 6 k026(a,v) d1 and for high values of no, fs is given by 2 J? 1 f = ——-—— OI —— (22) S #01902 Jno Therefore, at low no, EAFL is proportional to no, because f = 1. However, at high n S since the self-absorption 'or factor 13 Significant, EAFL fluorescence line source growth curve passes through a a l/Vno. Therefore, an atomic maximum. At low n EAFC is again proportional to n O’ ‘01 however, at high n B is independent of no because B 0' —AFC I—AAC a Vno and fS a l/Jnd. Therefore, an atomic fluorescence continuum source growth curve reaches a plateau. The exact shape of these growth curves for atomic fluorescence will, of course, depend on the Voigt profile parameters. 30 D. Atomic Absorption With a Resonance Monochromator When a resonance monochromator is used for atomic absorption Spectrometry, the radiance transmitted from the flame, ETL or ETCA , o of population no' in the resonance monochromator. The is absorbed by the ground state atoms radiance absorbed when a line source is used is given by ERAL = ETL B Ais [l - exp {-ko'fl'6'(a,v)}]dk' (23) where the factor B is introduced to account for any transmitted radiance which does not reach the resonance monochromator, and the prime symbol (') is used to differ- entiate terms in the resonance monochromator from terms in the flame. Any atomic fluorescence, EAFL or EAFC' from the flame is assumed to be negligible, and atomic emission is also neglected because an ac detection system is assumed. Since the source line half-width, A18, is still less than —TL Equation (23). However, since A18 is not much smaller than AAA, B from Equation (3) may be used directly in the absorption line half-width in the resonance monochro- mator, AAD', the integration must be carried out over A18 to obtain an exact expression. Equation (24) results when ETL from Equation (3) is substituted into Equation (23) ERAL = EL exp {-k026(a,UflBA{ [l - exp {-ko'k'6'(a,v)}]dk' S (24) 31 If AAD' A ARFL RFC ' of the source line half-width, AAS, and the absorption line half-width in the flame, AAA, while 5(a,3)' relates the overlap of the absorption half-width in the resonance monochromator, AAD', and the absorption half-width in the flame. If Als < AAD', Equations (33) and (34) predict that a line source working curve should have a greater sensiti- vity than a continuum source working curve, because the Voigt profile parameter is greater for the line source. IV. RESONANCE MONOCHROMATOR AND EXPERIMENTAL SYSTEM To characterize the variables which influence the radiant output of the resonance monochromator, a completely demountable model was constructed. With such a demountable model various elements could easily be interchanged such that zinc, lead, cadmium, and other resonance monochromators could be evaluated. Also, the effect of various pressures of filler gas on the radiant output could easily be determined with this versatile design. Furthermore, a demountable model allows the possibility of cleaning and reconditioning the resonance monochromator for extended use. The completely demountable resonance monochromator that was designed for this study is described in the following section. A. Design The function of the resonance monochromator is to isolate the resonance line of an element of interest. This is accomplished by producing a sufficient population of ground state atoms within the resonance monochromator, 37 38 onto which the radiation from an external source is focussed. The atoms in the vapor selectively absorb part of this radiation, and re-emit fluorescence over the entire solid angle. The radiation that is not absorbed simply paSses through the resonance monochromator. The selectivity of absorption is dependent on the energy required for a transition from the ground state to an excited electronic state. This energy (frequency, wavelength) corresponds to the resonance line which is characteristic for each element. For example, the energy required for the transition of a cadmium atom from the ground state to its first excited electronic state is 5.41 eV, which corresponds to the cadmium resonance line at 2288 A. Therefore, to isolate the cadmium resonance line, a resonance monochromator which contains cadmium atoms would have to be constructed, and this would be referred to as a cadmium resonance monochro- mator. To isolate the zinc resonance line at 2138 A, a zinc resonance monochromator, which contains zinc atoms, would have to be constructed, etc. Since a separate resonance monochromator would be required for every element, a completely demountable model was designed so that the elements could easily be interchanged. This model could also be used with several elements at the same time, i.e., multielement analysis. Furthermore, anticipated analytical applications in the area of continuum source atomic absorption dictated that the metals be readily inter- changable. 39 The resonance monochromator Shown in Figure 2 consists of two detachable sections. The first section contains the entrance window, and the Side window where the fluoresced radiation is directed to the cathode of the photomultiplier transducer. The second section consists of the heating assembly and the attachment for connection to a vacuum line. After a cylinder of the metal, or metals, of interest is placed inside the heating assembly, the two sections are connected by way of a S 50/50 ground glass joint. High temperature, 300 °C maximum, graphited glass joint grease, no. 0958, Podbielniak, Inc., Sergent-Welch, was used on the glass joint. The heating assembly was placed at right angles from the radiation beam to reduce scattering. The entrance and exit quartz windows and the Side window (quartz lens) were placed approximately three inches from the center to reduce deposition of the metal. However, when deposition of the metal did occur, this was not a problem with a demountable model, because the two sections could be separated and cleaned. A thermal resonance monochromator was chosen because many of the elements which show high sensitivity in atomic absorption and fluorescence spectrometry (Zn, Cd, Pb, etc.) are relatively volatile, and atomic vapors can be readily obtained thermally. Therefore, the resonance monochromator described here contains a completely interchangeable heating assembly in which metals or alloys may be placed. 40 3 I I | u 2 ’2 QUARTZ wmoow hv u -—-—h| QUARTZ ‘ / WINDOW 2%“ ‘ .,__ '§ 50/50 GROUND GLASS JOINT ~ Few—cl SILVER SOLDER xlNNER ALUMINA BOAT NICHROME (I .3/ VACUUM VALVE WIRE —--°gt; OUTER :k sf METAL ALUMINA i .J. TO BOAT -- —- VACUUM NICKEL -—-————- \ BALL 3 SOCKET \I JOINT n TUNGSTEN A . GLASS ' i ‘ + SEAL .. Diagram of Demountable Resonance Monochromator. Figure 2. 41 B. Heating Assembly The heating assembly serves to produce the necessary population of ground state atoms of the element of interest. The assembly is attached to the glass enclosed resonance monochromator by a tungsten-glass seal. The heating assembly consists of two alumina cylinders. Alumina (Morganite, Inc., Dunn, N.C., ARR crucibles, C.O.E.) was chosen to contain the metal to be heated because of its stability at high temperature, chemical inertness, good electrical resistivity, and high thermal conductivity. Nichrome wire, #30 gauge, was placed between the two alumina cylinders, and castable alumina was used as a sealant. This was necessary to reduce the nichrome wire emission which would otherwise introduce shot noise. The nichrome wire was then silver-soldered to a long nickel wire, which was attached to the tungsten-glass seal. The metal Of interest is placed inside the alumina boat and heated by passing dc current through the nichrome wire. As the current increases, the temperature of the alumina boat increases, which then increases the vapor pressure of the element, and hence produces the necessary atomic vapor. It was experimentally determined that the temperature of the alumina boat is nearly linearly dependent on the power applied to the nichrome wire. A plot of temperature vs. power applied is given in Figure 3. This plot is useful in determining the effect of the power applied to the resonance DJ 01 I) m: UJ D- t: 1" L33 Figure 3. 42 300‘ 200" 100’ O 1 I j T 1 1‘ O 2 4 6 8 IO 12 POWER APPLIED , W Temperature of Heating Element as a Function of Power Applied to Resonance Monochromator. 43 monochromator on the number of ground state analyte atoms in the resonance monochromator. For example, if a cadmium resonance monochromator is operated at a given applied power, the corresponding temperature of the alumina boat can be determined from Figure 3. From Table 2 the vapor pressure of cadmium can then be determined at that temper- ature along with the corresponding concentration of atoms. A plot of log concentration vs. applied power is given for cadmium in Figure 4. From this plot, one could directly determine the concentration of cadmium at a given applied power. Similar plots could be made for other elements. C. Filler Gas After a metal cylinder of the element of interest has been placed in the heating assembly, the two sections are connected by the ground glass joint. The resonance monochromator is then attached to a vacuum system by the ball and socket joint, and is evacuated with a diffusion pump to a pressure of approximately 10.3 torr. Sufficient power is applied to the nichrome wire to vaporize the outer layer of the metal. Argon filler gas is then added to the desired pressure. The purpose of the filler gas is to prevent the rapid migration of the atomic vapor to the walls of the monochromator. Optimization of the filler gas pressure and its effect on the applied power are discussed in the next chapter. 44 s LOG no ATOMS/CC I I I I I l O 2 4 o 8 IO 12 POWER APPLIED .W CADMIUM CONCENTRATION IN RESONANCE MONOCHROMATOR Figure 4. Cadmium Atomic Concentration as a Function of Power Applied to Resonance Monochromator. 45 Table 2.--Dependence of Cadmium Vapor Pressure on Temper- ature (57). # Pressure Temperature no' log no' (Torr) (°C) (atoms/cc) 10‘5 148 2.3 x 1011 11.36 10'4 180 2.1 x 1012 12.32 10‘3 220 1.95 x 1013 13.29 10'2 264 1.8 x 1014 14.26 10’1 321 1.6 x 1015 15.21 D. Experimental System A block diagram of the experimental system used for atomic absorption is shown in Figure 5, and the specific experimental system is described in Table 3. For comparison purposes, the resonance monochromator and associated power supply were frequently replaced by a conventional grating monochromator. Resonance monochromators with zinc, cadmium, or lead placed inside the alumina boat were con- structed to verify the theoretical expressions derived previously. For the case of constant no, the flame was not used, and the concentration of atoms in the resonance monochromator, no', was varied by changing the power applied to the heating assembly. For the case of constant no', the power applied was maintained at the optimum, and standard solutions were aspirated into the flame. Results obtained were then compared to a conventional grating monochromator for analytical atomic absorption applications. ATOMIC SOURCE VAPOR I I I I I I I I I I CHOPPER DRIVE A DIFFERENTIAL l TO RESONANCE L.V.P.S. MONOCHROMATOR PHOTO- . .R . 'MULTIPLIER HV 3 PRE- AMPLIFIER LOCK- IN SINGLE-ENDED CONVERTER . l AMPLIFIER I [RECORDER I I DIGITAL READOUT Figure 5. Block Diagram of Resonance Monochromator Atomic Absorption System. 47 Table 3.--Experimental System for Atomic Absorption Component Description and Type Supplier sources Source power supplies Chopper Burner Monochromators Photomulti— plier tube Photomulti- plier power supply Pre-amplifier Lock—in amplifier Readout devices Hollow cathode discharge tubes Model 111-150A, 150w Xenon arc lamp Model EU—703-70 AA, AB, AF module hollow cathode supply Model CA-lSO Xenon arc supply 30—Hz leaf shutter in Model EU-703-7O AA, AE, AF module powered by EU-703-3l photometric readout module Tri-flame, 10 cm slot premixed, air/H2 Model EU-700, 350 mm focal length f/6.8 aperture, 20 A/mm reciprocal linear dispersion Resonance monochromator as described in text powered by QB6-8 low voltage supply R 166 solar blind Model EU-701 Photomultiplier module Current-to-voltage converter constructed from SP2A premium parametric operational amplifier Model 840 Autoloc Model 3000 recorder Model EU-BOS Universal digital instrument Fisher Scientific CO. Pittsburgh, PA Illumination Industries 610 Vaqueros Sunnyvale, CA Heath Co. Benton Harbor, MI Illumination Industries Sunnyvale, CA Heath Co. Benton Harbor, MI Fisher Scientific Co. Pittsburgh, PA Heath CO. Benton Harbor, MI Raytheon Co. Sorenson Div. Manchester, NH Hamamatsu Corp. Lake Success, NY Heath Co. Benton Harbor, MI Teledyne Philbrick Dedham , MA Keithley Instruments Cleveland, OH Houston Instruments Bellaire, TX Heath Co. Benton Harbor, MI 48 Since the radiation from the source is modulated at 30 Hz, the output signal from the lock-in amplifier is due primarily to the fluorescence signal. However, any radiation that is internally reflected within the resonance monochro- mator will also add to this signal. Therefore, a solar blind photomultiplier tube was used to reduce the transducer response to this scattering of nonresonance lines. Un- modulated radiation from stray room light, emission from the flame, or flame background does not contribute to the measured signal, but can introduce shot noise. Therefore, the resonance monochromator was housed in a light-tight black box. V. OPTIMIZATION OF APPLIED POWER AND FILLER GAS PRESSURE A. Line Source Growth Curves To verify qualitatively the theoretical fluores- cence growth curve for a line source, Equation (29), the ground state atom population in the resonance monochro- mator, no', was varied by changing the current applied to the nichrome wire in the heating assembly. A plot of the resonance monochromator output as a function of the current applied is presented in Figure 6. A zinc hollow cathode tube was employed as the external source with a current of 30 mA. The zinc resonance monochromator had an argon filler gas pressure of 6 torr, and 4.5 inches of #30 gauge (6.7 9/ ft) nichrome wire was embedded in the heating assembly. The fluorescence signal, which is first observed at 2 A (10 W), reaches a maximum at 2.5 A (16 W), and then decreases because of the self-absorption factor. The maximum is clearly observed in an output signal vs. applied current plot. However, since the concentration of atoms in the resonance monochromator is dependent on the applied power (temperature a power), the same data were plotted vs. power 49 50 IO- RESONANCE MONOCHROMATOR OUTPUT. ARBITRARY UNITS _ I -'—I I 2 CURRENT APPLIED . A 00"“ Figure 6. Resonance Monochromator Output as a Function of Current Applied to Heating Element. 51 applied to the heating assembly and this is shown in Figure 7. The resonance monochromator output Obtained with an incident line source was assumed to be proportional to the fluorescence radiance, ERFL' Since log no' has been shown to be proportional to the applied power in Figure 4, a plot of log fluorescence signal vs. applied power should correspond to a plot of log ERFL vs. log no'. Table 1 predicts that at constant no, ERFL is proportional to no' at low no', and proportional to l/JEST at high no'. The atomic population in the flame, no, was assumed to be zero for these growth curves. The ratio Of the slope of the plot at low applied power (ml) to that at high applied power (m2) is minus two, as predicted in Table l for a plot of log B -RFL To determine the effect Of argon filler gas vs. log no'. pressure on the growth curve, a zinc resonance monochro- mator was prepared with an argon filler gas pressure of 12 torr. The growth curve obtained is shown in Figure 8. The ratio of the slopes ml/m2 is again a minus two, however, the maximum now occurs at an applied power of 12.5 W. More zinc resonance monochromators were prepared with varying pressures of argon filler gas. Figure 9 describes the dependence of the optimum applied power on the argon filler gas pressure. The data can be explained as follows. The maximum fluorescence occurs at a particular concentration, atoms/cc, of no' for a given filler gas 52 ”I RESONANCE MONOCHROMATOR OUTPUT. ARBITRARY UNITS ~0I F“ I T“ I I 1 10 12 I4 IO 18 2O 22 POV‘I’ER APPLIED, W Figure 7. Experimental Growth Curve for Zinc Resonance Monochromator With Incident Line Source; Argon Filler Gas, Pressure = 6 Torr. |..I RESONANCE MONOCHROMATOR OUTPUT, ARBITRARY UNITS 53 Figure 8. I x I \\\ I, ‘ ’ *0 SLOPE- m. 0 I. “T ”r I I I I 6 . 8 IO I2 l4 I6 I3 POWER APPLIED, w Experimental Growth Curve for Zinc Resonance Monochromator With Incident Line Source; Argon Filler Gas, Pressure = 12 Torr. 54 OPTIMUM APPLIED POWER , W IO I r T ' I I I I 0 2 4 6 8 I0 I? I4 FILLER GAS PRESSURE. TORR Figure 9. Optimum Applied Power as a Function of Argon Filler Gas Pressure. 55 pressure. As the filler gas pressure decreases, the diffusion of the zinc atoms increases. Therefore, to attain the same concentration of zinc atoms in the Observation window of the photomultiplier transducer, a higher power is required. When there is no filler gas present, the zinc atoms are free to migrate throughout the entire volume of the resonance monochromator. In this case the highest applied power is required to attain the concentration required for maximum fluorescence. It is interesting to note the stability of the fluorescence signal while at the optimum applied power. Approximately fifteen minutes were required for equilibrium to be attained. The signal was monitored on a strip chart recorder for a period of seven hours during which time the relative standard deviation of the signal was approximately 0.2 per cent. These measurements reflect fluctuations in the primary radiation source, in the atomic concentration in the resonance monochromator, and in the instrument response. B. Continuum Source Growth Curves To verify qualitatively the predicted growth curves for a continuum source-resonance monochromator combination, a cadmium resonance monOchromator was used with an argon filler gas pressure of 2 torr. Table 1 predicts that, at o o I I constant no, ERFC is proportional to no at low no , and is independent of no' at high no'. Plots of resonance 56 monochromator output vs. applied power are shown in Figure 10 for both a cadmium line source and a 150 W xenon arc continuum source. The line source plot has a ratio of slopes ml/m2 of minus two as was obtained for the zinc line source-resonance monochromator combination. The fluorescence signal obtained with the continuum source was assumed proportional tOIERFC° The continuum source plot has a slope, m3, equal to the line source slope, ml, at low applied power (low no'), but reaches a slope, m4, of zero at high applied power (high no') as predicted in Table l for a plot of log ERF vs. log no'. C The maximum in the cadmium line source growth curve occurred at an applied power of 2.5 W. From Figure 4 this corresponds to a cadmium concentration of 1011 atoms/cc. If the filler gas pressure were changed, the applied power would also have to be changed to maintain this cadmium concentration of 1011 atoms/cc. C. Optimum Conditions for Analytical Applications The two major parameters of the resonance monochro- mator which should be optimized are the applied power and the filler gas pressure. As shown in the previous section, there is an optimum applied power which maximizes the fluorescence output when a line source is incident on the resonance monochromator, while with a continuum source the fluorescence signal becomes independent of applied power 57 I r-------——- o a SLOPE=m,I ,’ B I. I ‘7 RESONANCE MONOCHROMATOR OUTPUT, ARBITRARY UNITS 1' I W I ' I I 2 3 POWER APPLI ED, W AJ Figure 10. Experimental Grow Curves for Cadmium Resonance Monochromator With Incident Line or Continuum Source; Argon Filler Gas, Pressure = 2 Torr. 0 Line Source E] Continuum Source 58 at high powers. With a given resonance monochromator, the applied power should be varied experimentally until the maximum fluorescence signal is Obtained or the plateau is reached. The optimum applied power when a line source is used has been shown to be dependent upon the filler gas pressure. To determine the effect of filler gas pressure on the fluorescence signal, a zinc resonance monochro- mator was used with a zinc line source and Operated at the maximum of the fluorescence signal vs. applied power plot. A plot of fluorescence signal vs. argon filler gas pressure is given in Figure 11. As can be seen, the fluorescence signal is greatest when there is no filler gas present, and decreases as the argon pressure increases. These results agree with those of L'vov (58) who determined the relationship of hollow cathode emission intensity to filler gas pressure for highly volatile elements. His results are reproduced in Figure 12. In both cases the signal is dependent on the diffusion of ground state atoms. The filler gas pressure also influences the life- time of the resonance monochromator. As the pressure is increased, the diffusion of atoms is decreased, and thus the deposition of vapor on the windows of the detector is minimized. The optimum filler gas pressure, therefore, is a compromise between intensity of the fluorescence signal and lifetime of the resonance monochromator. A pressure of 59 5': L, V I0‘ 0| 1 RESONANCE MONOCHROMATOR OUTPUT, ARBITRARY UNITS OIIIIIII 02468IOI2I4 - 7OO FILLER GAS PRESSURE, TORR Figure 11. Zinc Resonance Monochromator Output as a Function of Argon Filler Gas Pressure. 60 '7 .._.....____..._._.I ’T"“"’ -' ' ' --~. '3‘ 51 I 5“". . 3 9J I Cd I 5‘1A a: I s. ‘ ‘ "_ g 47 I I 4‘ A I E.:’ I \ ' 3 I\ \\ I c m' 2- a ' g; I Z.\ \\ E a Intensity {PPIOIIVQ unirsl '3 7 7 g 6-4 6- g: 5.1 5.. 5n §.4_ 4~ 1: 3.. 3- h \ '3 2" 21 \ . C \ \-\ o I 1'11 T l T r I T 0 I fin“! lqfi f.I T I I23456'789I0 12345678910 Pressure Itorr) Pressure (torr) Ne Xe ................ Ar ....... Ho ....... .. _.... Figure 12. Relationship of the Intensities of the Resonance Lines for Highly Volatile Metals to Inert Gas Pressures (58). 61 about 8 or 10 torr appears to be the optimum for a resonance monochromator that is to be used for a lengthy time, whereas, a pressure of 1 or 2 torr is the optimum for resonance monochromators that require a high intensity as in the case of a continuum source. Sullivan (59) stated that when he used a lead resonance detector with an argon filler gas pressure of 10 torr "deposition of vapour on the windows of the detector did not occur for at least 2000 hours." VI. ANALYTICAL APPLICATIONS A. Line Source Working Curves While at the optimum power for the zinc resonance monochromator, constant no', a working curve was determined for zinc. A hollow cathode lamp current of 40 mA was used for the zinc resonance monochromator working curve, while lamp currents of 20 and 40 mA were used for the grating monochromator working curves. These results are shown in Figure 13. Note that with a grating monochromator, the absorbance sensitivity for the 20 mA lamp current is greater than the sensitivity for the 40 mA lamp current. The reason for this, as shown by L'vov (60) and Rann (61), is broadening and self-reversal of the resonance line in the hollow cathode caused by an increase in the atomic vapor at higher lamp currents. For example, line profiles for the Mg 2852 A resonance line for different lamp currents are shown in Figure 14. As the lamp current increases, the resonance line broadens, and eventually self-reverses. The decrease in absorbance sensitivity is observed with a grating monochromator because the entire broadened line falls within the spectral bandpass of the grating monochro- mator. 62 63 I.0-I I I ' I I0 I5 20 ZINC CONCENTRATION, ppm 0 “-1 Figure 13. Atomic Absorption Calibration Curves for Zinc With Hollow Cathode Source. C) Resonance monochromator; lamp current 40 mA. [J Grating monochromator; lamp current 20 mA. A Grating monochromator; lamp current 40 mA. 64 Figure 14. Mg 2852 A Line Profiles for Different Lamp Currents (60). As stated by Sullivan and Walsh (17), "a working curve with a resonance detector should be independent of conditions under which the light [radiation] source is operated." Our experimental results have confirmed this as was shown in Figure 13. The resonance monochromator sensitivity, which remains essentially constant at various lamp currents, is greater than the sensitivity with a grating monochromator because the spectral bandpass of the grating monochromator isolates that broadened portion of the line source which does not overlap the absorption line in the flame. The absorption line-width in the resonance monochromator remains constant, and therefore, is not 65 affected by the broadening of the resonance line of the hollow cathode. Recalling Equation (15), as A18 increases, 6(a,U) decreases, which decreases the absorption sensiti- vity. This decrease is observed with a grating monochro- mator because all of the broadened resonance line falls within its spectral bandpass. However, recalling Equation (23) for the resonance monochromator, as Als increases, it will approach and eventually become greater than AAD'. When this occurs, the limits of integration must be changed to AAD' as shown in Equation (35). B =B 8 f -RAL -TL M . [l - exp {-ko'i'6'(a,v)}]dk' (35) D Thus the broadened portion of the resonance line from the source will not be included in the integrated absorption in the resonance monochromator, and therefore, changes in the line source half-width will not affect the calculated absorbance of Equation (33). Hence, the resonance mono- chromator treats the line source as if it were not broadened. Rube§ka and Svoboda (62) have calculated the effect of different emission and absorption half-widths on analytical working curves. Their results are shown in Figure 15. Again, as the emission half-width increases, the absorbance sensitivity decreases. These theoretical results agree favorably with our experimental results of Figure 13. Curve #3 of Figure 15 indicates that the Figure 15. 66 Calculated Analytical Curves for Resonance- Shaped Lines and Different Emission and Absorption Line Half-Widths (62). 1. A15 = 0 AAA = 5 2. A18 = 1.5 AAA ; s 3. A18 = 5 AAA = 5 4. A18 = 15 AAA = 5 67 emission and absorption half-widths are equal. This contradicts the assumption made in Chapter III. If this is the case, then Equation (1) must be used to obtain an exact expression for 2AAL' Various authors have experimentally determined emission and absorption half-widths. For example, Yasuda (63) and L'vov (64) have determined the half-widths of the 4226.7 A calcium resonance line at various lamp currents. Their results are given in Table 4. Kirkbright _e_t_ a_]_.. (65-67) have determined the absorption half-widths for calcium atoms in flames at various concentrations. For example, the absorption half-width of the Ca 4226.7 3 line for a 100 ppm solution in an air-acetylene flame is 0.055 A and in a hydrogen-nitrogen flame is 0.036 A. Therefore, the absorption half-width for calcium is not that much greater than the emission half-width. Bazhov and Zherebenko (68) determined the absorption half-width of the Cd 2288.0 A line for a 1.5 ppm cadmium solution in an air- propane flame to be 0.0098 A. From Table 4, this is lggg than the emission half-width of the cadmium hollow cathode. Therefore, the absorption and emission half-widths are similar, even though the absorption half-width increases slightly with concentration. Consequently, our results in Figure 13 indicate that the emission line width of the Zn hollow cathode at 40 mA is comparable to the zinc absorption half-width in the air-hydrogen flame. These results also indicate that the absorption half-width in the resonance 68 Table 4.--Half-Widths of Ca 4227.6 3 and Cd 2288.0 A Resonance Lines as a Function of Lamp Currents. Element . Current Half-Width Reference (mA) (A) Ca 25 0.024 (64) 40 0.027 (63) 60 0.031 (63) 80 0.036 (63) 90 0.041 (63) Cd . 10 0.018 (64) 10 0a0185 (68) monochromator is less than the absorption half-width in the flame since a 40 mA lamp current was used for both working curves. Hence, working curves Obtained with a resonance monochromator are less affected by broadening in the hollow cathode lamp, and should have greater sensitivities and more reproducibility than working curves obtained with a grating monochromator. B. Spectral Interferences Another interesting feature of the resonance monochromator is the potential elimination of spectral interferences because of the narrow spectral bandpass. Interferences in flame absorption spectrometry can be classified into blank interferences and radiation inter- ferences. Blank interference may be attributed to two 69 sources; cross-sensitivity and background. There are two conditions that can occur in flame absorption that could be classified as cross-sensitivity: (l) overlap Of an absorption line Of a concomitant, i.e., any element I radical or solute other than the analyte, with the absorption line of the element of interest, and (2) the appearance Of a nonabsorbing source line within the spectral bandpass of the monochromator. The first case, spectral line interference, may occur in atomic absorption spectrometry when there is significant overlap of an emission line profile of the primary source and the absorption profile of any inter- fering species in the flame. Fassell, gg_gl. (69) have shown that spectral line interferences can occur in atomic absorption spectrometry. Jowrowski and Weberling (70) reported the interference of copper on the determination of lead when employing the 2170 A lead resonance line from a hollow cathode lamp containing both elements. Hall and Woodward (71) also reported interferences of copper on lead from copper impurities in a lead hollow cathode lamp. This interference is due to copper resonance lines at 2165 A and 2179 A. If a grating monochromator with a spectral band- pass of about 4 A is used, this interference will not be significant because only the 2170 A lead resonance line is isolated. However, if a grating monochromator with a O spectral bandpass greater than 4 A is used, the copper 7O resonance lines will introduce an interference if copper is present in the solution. As the spectral bandpass is further increased, more of the copper resonance lines are passed by the monochromator, and hence more of an inter- ference is introduced. Figure 16 shows the dependence of the absorbance obtained for a 5 ppm lead solution, con- taining various concentrations of copper, on the monochro- mator spectral bandpass. The 2170 5 lead resonance line from a Pb, Cu, Zn, Ag hollow cathode lamp was employed in this study. The reciprocal linear dispersion of the grating monochromator was 20 R/mm. Thus the four slit widths correspond to nominal spectral bandpasses of 4 i, 6 i, 8 i, and 10 3. With a spectral bandpass of 4 i, very little copper interference is observed. However, as the monochromator bandpass is increased, the absorbance increases because more of the copper lines are passed. The lead resonance monochromator, on the other hand, virtually eliminates the copper spectral interference because of its narrow spectral bandpass as may be seen in Figure 16. The second case of a cross-sensitivity interference is the appearance of a nonabsorbing source line within the spectral bandpass of the monochromator. If the monochro- mator bandpass isolates only the resonance line from the line source, absorption sensitivities should be independent of monochromator slit width as long as stray light is negligible. However, if a nonabsorbing interfering line 71 .3C)~ .2C)- _ <—> 4 e l I l 25 50 I00 COPPER CONCENTRATION, ppm Figure 16. Spectral Interference of Copper on Lead with Multielement Hollow Cathode Source Using 2170 A Resonance Line for Grating Monodhromator. Each Solution Contains 5 ppm of Lead. E] 500 um slit.width ‘7 200 Um slit.width C) 400 um slit width <> Lead resonance monochromator [5 300 um slit width 72 is also passed by the monochromator, absorption sensiti- vities will decrease with increasing monochromator spectral bandpass. This is illustrated in Figure 17 for the atomic absorption of a 5 ppm lead solution employing a Pb, Cu, Zn, Ag multielement hollow cathode lamp. As can be seen, when the 2833 A lead resonance line is employed, the observed absorbance remains constant until the monochromator band- pass is sufficiently wide, 9 A, to pass a nonabsorbing Cu line at 2824 A. When the 2170 A lead resonance line is employed, the observed absorbance follows the same behavior. The absorbance is constant at low monochromator bandpasses but decreases as the bandpass increases to pass the non- absorbing copper lines at 2165 A and 2179 A (no copper was present in the solution). Due to the very low spectral bandpass of the resonance monochromator, nonabsorbing lines should not cause an interference. The observed absorbance for a 5 ppm lead solution is also shown in Figure 17 and falls between the two limiting absorbance values for the 2170 A resonance line and the 2833 A resonance line. Since the resonance monochromator produces fluorescence radiation at both of these wavelengths, the photomultiplier transducer will respond to them because the solar blind has a spectral response which extends from 1800 A to 3200 A. Therefore, this behavior is expected at low atomic concentrations because the observed absorbance with the resonance 73 J0- -00 I I I I r I O 2 4 6 8 IO l2 MONOCHROMATOR SPECTRAL BANDPASS, K Figure 17. Dependence of Lead Absorbance on Monochromator Spectral Bandpass for Multielement Line Source. Resonance line isolated Interferinggline (s) isolated (3 2170 A; grating monochromator 2165 A and 2179 A E] 2833 A; grating monochromator 2824 A <> 2170 A and 2833 A; resonance - monochromator 74 monochromator will depend on the relative oscillator strengths of the two resonance lines. C. Continuum Source Working Curves Because of its narrow spectral bandpass, the resonance monochromator is useful not only with an incident line source, but also with a continuum source. To obtain a resonance monochromator-continuum source working curve a cadmium resonance monochromator was employed at the plateau of the growth curve. Atomic absorption working curves were obtained with a cadmium resonance monochromator with both a cadmium line source and a xenon arc continuum source. These results are shown in Figure 18 along with a working curve obtained with a cadmium line source and a grating monochromator. A lamp current of 10 mA was used for the cadmium line source to avoid self-reversal. Note the improved linearity obtained with the continuwm source- resonance monochromator combination. This phenomenon may be explained by recalling Equations (33) and (34) - AX ARFL = komm'v) U = 0 to FTS' (33) 2.563 A .. I ko£5(a,v)' _ Aka ARFC = -—-2-.-3-5-§— U = 0 t0 AXA (34) For the case of the line source, as the concentration of atoms in the flame increases, the fraction absorbed increases, but the absorption half-width does not initially broaden as is shown in Figure l9.a. Recall that 75 L04 ° A 0131 I I I . I I fit 0 IO 25 5O 75 I00 CADMIUM CONCENTRATION, ppm Figure 18. Atomic Absorption Curves for Cadmium With Line and Continuum Source. D Resonance monochromator and line source. A Grating monochromator and line source. C) Resonance monochromator and continuum source. *n—“—--—— 0 “hr- “’-v, ‘- x.“ a. <-- Linc source half-width {3N A3330: . '7'? (t C"? s. . 3 ‘3". [AC [l k-I no increasing m). 4—-—Resonancc monochromator . ‘ . obsorphon haH‘wu’Ifh Figure 19. Graphical Representation of the Effect of Increasing Atomic Concentration, no, on Absorption Half-Width in Flame for Cadmium (72). 77 the source line half-width for the Cd 2288 A resonance line in a hollow cathode operated at 10 mA is 0.018 A (64, 68), and that the absorption half-width for a 1.5 ppm Cd solution in a flame is 0.01 A. This trend continues until the fraction absorbed reaches one, at which point the absorption half-width begins to broaden (72). Since the spectral bandpass of the resonance monochromator-line source com- bination is equal to the source line half-width then, according to Equation (33), the absorbance should increase linearly with n until the absorption half-width in the o flame becomes greater than the source line half-width. At this point, bending will be observed, as indeed is the case in Figure 18 at a cadmium concentration of approximately 25 ppm. The spectral bandpass of the resonance monochro- mator-continuum source combination, on the other hand, is equal to the half-width of the absorption line in the resonance monochromator. In this case, the concentration of analyte atoms in the flame may be increased further, as in Figure 19.b, before bending is observed.. Since at a cadmium concentration of 100 ppm, the absorbance obtained with the line source-resonance monochromator combination, A is equal to the absorbance obtained with a continuum RFL' source-resonance monochromator combination, ARFC' the Voigt profile factors must be equal. Therefore, the absorption half-width for a 100 ppm Cd solution in the 78 flame is equal to the absorption half-width in the resonance monochromator, or the spectral bandpass. For a continuum source-conventional grating monochromator combination, the absorbance should increase as the spectral bandpass decreases to a limiting value equal to the Doppler half-width in the flame, a bandpass unapproachable with conventional monochromators. The exact dependence is given by Equation (16) for large Spectral bandpasses, 2 - {l - exp (-k°2)}AAD AAAC = - log 3 (16) and by Equation (17) for low spectral bandpasses ko£6(a,U') _' ' AAAC = -_—7735§__ v = 0 to AAD /AAA (17) For a 100 ppm Cd solution the absorbance obtained from Figure 18 was 1.25. Since at this concentration AAD' = AAA, the v parameter is equal to l, and, assuming that a = 0.5, 6(a,59 = 0.35 (26). Therefore, koi is equal to about 8. If AAAC is now obtained for various values of g, the absorption half-width in the flame may be determined for that concentration from Equation (16). Atomic absorption data for a 100 ppm cadmium solution were obtained with a 150 W xenon arc source and a conventional grating monochro- mator at several slit widths. A plot of observed absorbance vs. monochromator spectral bandpass is shown in Figure 20. Theoretical values of AAAC as function of g were then 79 —A— .A “L M“ o-‘nfi‘AhL .14- I- I H .01 .1 1 MONOCHROMATORO SPECTRAL BANDPASS A Figure 20. Dependence of 100 ppm Cd Solution Absorbance on Monochromator Spectral Bandpass for Con- tinuum Source. 0 ——- Theoretical for AXD = 0.029 A C) Experimental, grating monochromator E] Experimental, resonance monochromator 80 determined at various absorption half-widths. The best fit was obtained with an absorption half-width of 0.029 A. The theoretical results are tabulated in Table 5, and are also shown in Figure 20. Therefore, we may conclude that the spectral bandpass of the cadmium resonance monochro- mator is approximately 0.029 A. For the cadmium working curves of Figure 18, the line source-resonance monochromator combination has a greater sensitivity than the continuum source-resonance monochromator combination because of the narrow half-width of the line source, 0.018 A. However, the range of linearity for the line source-resonance monochromator is not as great as that for the continuum source-resonance monochromator. The continuum source-resonance monochro- mator combination, on the other hand, has a decreased I sensitivity, but improved linearity because its wider spectral bandpass, 0.029 A, does not respond to broadening in the flame until a higher concentration is reached. Therefore, as the spectral bandpass or limiting spectral bandpass increases, the sensitivity decreases and linearity increases because of the Voigt profile factor. This theory could be extrapolated to its limit when we have the case of a very wide spectral bandpass which would be independent of the Voigt profile. The sensitivity would be very low, but the range of linearity would be extremely high as long as the concentration of analyte atoms in the flame remained Table 5.--Theore§ical Dependence of A 81 on‘g for AAD = 0.029 A. “C Spectrai)bandpass AAAC (theoretical) AAA (experimental) 0.03 1.47 1.25* 0.04 0.56 0.05 0.37 0.06 0.29 0.07 0.23 0.08 0.19 0.09 0.17 0.1 0.15 0.2 0.068 0.064** 0.3 0.044 0.4 0.032 0.032** 0.5 0.026 0.6 0.021 0.7 0.018 0.8 0.016 0.017** 0.9 0.014 1.0 0.013 *Resonance monochromator **Grating monochromator 82 proportional to the concentration of the analyte in solution. This is the case, of course, of a continuum source-grating monochromator combination which was covered in Equation (16) where the spectral bandpass of the monochromator was assumed to be much larger than the absorption half-width in the flame. VII. CONCLUSIONS A. Summary The resonance monochromator has been shown to have several advantages over a conventional dispersive monochro- mator for atomic absorption flame spectrometry. The narrow spectral bandpass of the resonance monochromator, 0.01 A- 0.03 A, is perhaps the most advantageous feature. With the resonance monochromator arrangement, the width of the lines of the resonance spectrum are determined solely by the physical conditions such as temperature and pressure prevailing in the resonance monochromator. This contrasts with the situation that prevails when using a conventional atomic absorption spectrometer in which the spectral bandpass of the output signal due to the resonance line isolated by the monochromator is determined by the width of the resonance line emitted by the radiation source and not by the optical resolution of the monochromator. It is also important to note that in a conventional atomic absorption spectrometer, the measured atomic absorption coefficient will depend critically on the width of the resonance line emitted by the radiation source and, there- fore, on the current at which the source is operated. 83 84 However, when a resonance monochromator is used, the absorption measurement is determined predominantly by the width of the absorption line in the resonance monochromator, and is largely independent of the conditions prevailing in the radiation source. Consequently, working curves obtained with a line source are relatively independent of source broadening and thus should remain constant from day to day. A further advantage of the resonance monochromator is that random noise due to radiation emitted by the flame can only be produced by that portion of the radiation that falls within that region of the spectrum that can be absorbed by the atomic vapor in the resonance monochromator. Therefore, this noise may be less than in a conventional spectrometer, where random noise can be produced by any radiation from the flame that passes through the monochro- mator, the spectral bandpass of which is many times greater than the width of the lines in the absorption spectrum of the atomic vapor. Also due to the narrow spectral bandpass, many troublesome spectral interferences can be avoided with the resonance monochromator. In addition, the resonance monochromator has been shown to be of use with a continuum source for atomic absorption analyses. Other advantages, such as the absence of tuning in to a given resonance line, compactness and durability, and a lower cost have previ- ously been summarized in Chapter I. 85 There are, however, several disadvantages of the resonance monochromator system described here. Thermal monochromators may be constructed only for relatively volatile elements, whereas sputtering type monochromators are required for other elements. This precludes the use of one simple demountable monochromator design with a readily interchangeable assembly for a large number of elements. Also, a resonance monochromator has a limited lifetime during which its performance may vary. For some elements the maximum sensitivity obtainable will be considerably less than with a conventional atomic absorption spectro- meter. The absorption sensitivities have been shown to be a function of the oscillator strengths of all resonance lines viewed by the photomultiplier transducer. This loss in sensitivity becomes more appreciable with increasing complexity of the resonance radiation spectrum. Also a resonance monochromator requires about fifteen minutes of warm-up time before stability is reached. A final dis- advantage of the resonance monochromator described here is that it is specific for only one element. In summary, the resonance monochromator has two major advantages. First, the spectral bandpass is on the order of 0.01 to 0.03 A. The spectral bandpass of the cadmium resonance monochromator was experimentally deter- mined to be 0.029 A at a concentration of 1011 atoms/cc. Second, a demountable resonance monochromator may be cleaned and re-conditioned for a variety of elements. 86 B. Recommendations The resonance monochromator offers attractive possibilities in the field of multielement analysis. This is made possible with a demountable model by placing more than one element in the alumina boat. If line sources are used, the fluorescence radiation for each element could be isolated by adjustment of the power applied to the heating element. For example, the optimum power for the cadmium resonance monochromator was 2.5 W, the zinc resonance monochromator 12 W, and the lead resonance monochromator 28 W. The optimum power, of course, is a function of the filler gas pressure. If the optimum powers for two elements are relatively close, the fluorescence radiation for each element could be isolated by appropriate filters. If a continuum source is used, the power could be adjusted until the plateau in the fluorescence vs. power curve is reached for all elements. Appropriate filters could then be used to isolate the respective fluorescent radiances. Sputtering type resonance monochromators could be constructed for multielement analysis thus widening its range of usefulness. Cathodic sputtering in an electrical discharge can produce atomic vapor from a wide variety of metals since this technique can be used irrespective of the melting point of the element and without the necessity of using high temperatures. In this technique, however, it is inevitable that the production of atomic vapor is 87 accompanied by the emission of radiation, and it is essential to isolate this from the desired resonance radiation. This may be achieved by modulating the source radiation via a mechanical chopper or by modulating the power supply to the hollow cathode lamp. One attractive possibility is to use pulsed hollow cathode lamps, which are described in Part II of this thesis. This method has two advantages: first, the primary radiation is modulated, and second, the intensity of the primary radiation is increased, which increases the signal- to-noise ratio of the measurement. The power supply for producing the atomic vapor in the resonance mono- chromator is not modulated, and thus an ac detection system tuned to the modulation frequency of the source would be used. The output from the detection system will then consist entirely of resonance radiation. Any radiation emitted by the atomic vapor in the resonance monochromator is not modulated and thus produces no signal at the output of the amplifier, but can introduce shot noise. Once the atomic vapors for the various elements have been produced by cathodic sputtering in the resonance monochromator, pulsed hollow cathodes could be sequentially operated,.and- the respective atomic absorption measurements sequentially determined. Another possibility would be to have all of the hollow cathodes operated simultaneously, with the respective ON times out of phase with each other. Since the aperture of the resonance monochromator is large, it should not be difficult to simultaneously focus the radiation from the sources such that they pass through the flame and enter the resonance monochromator. With either system a filter would £22,be required because the resonance fluorescence radiances for the respective elements are separated in time. Simultaneously operated pulsed hollow cathode lamps for atomic fluorescence spectrometry are described in Part II of this thesis. PART II NONDISPERSIVE MULTIELEMENT ATOMIC FLUORESCENCE SPECTROMETRY I. INTRODUCTION The technique of atomic fluorescence spectrometry (A.F.S.) is closely related to that of atomic absorption spectrometry (A.A.S.). However, since A.A.S. is independent of the source intensity (except for cases discussed in Part I), the instrumentation employed usually has low light-gathering characteristics. A.F.S. on the other hand, requires a high light-gathering detector to be used. Jenkins (54) suggested that a good detector for atomic fluorescence should have high transmission and large aperture optics, a high photocathode sensitivity and a narrow bandwidth. Therefore, it is not necessary to use a monochromator, because it can be arranged that only one fluorescing species is stimulated at one particular time, with the result that only its set of resonance lines will emerge from the atom reservoir at that particular instant. This possibility of eliminating the monochromator from A.F.S. measurements when using narrow line sources makes feasible a multielement nondispersive A.F.S. technique with only one photomultiplier-amplifier system. 89 90 Atomic fluorescence spectrometry has many features which make it extremely useful for nearly simultaneous multielement analysis. In A.F.S., as in A.A.S., the signal is dependent on the number of atoms in the ground state. Also as in A.A.S., A.F.S. is relatively free from spectral interferences (except in cases covered in Part I), and the desired fluorescence radiation can be electronically separated from flame radiation by using a modulated line source with an ac detection system. A.F.S. has several other properties which are of particular importance in multielement analysis: a. at low atomic concentrations the intensity of the fluorescence radiation is proportional to the intensity of the exciting radiation; b. there are few optical problems in simultaneously focussing a number of radiation sources on the atom reservoir; c. fluorescence radiation is viewed at an angle to the exciting radiation, and hence non-resonance lines from the source do not interfere with the measurement process; d. fluorescence signals from each element can be independently switched on and off, by switching the excitation sources on and off; e. A.F.S. is particularly well suited to pulse excitation techniques. The electronic detection 91 system can be OFF between pulses, thus rejecting most of the noise from background radiation; optical filters are also suitable for use with atomic fluorescence systems, since they permit very efficient observation of the irradiated area of the atom reservoir and very wide apertures; finally, with the advent of very low background flames, a completely nondispersive instrument may be employed which can yield higher signal-to-noise ratios than instruments which employ a monochro- mator. The principal benefits to be derived from such a nondispersive system for atomic fluorescence are: 1. 2. greater energy throughput, simultaneous collection of multiple lines for elements with a complex fluorescence spectrum, convenience for multielement analysis, simplicity and ruggedness of instrumentation, reduced cost. In this section of the thesis an automated non- dispersive atomic fluorescence system for multielement analysis will be described. The instrument employs a single channel detection system and operates in the time multiplexed mode. The integral components of this system include a method for operating hollow cathode lamps in an intermittent, high current mode, a sheathed burner, a 92 synchronous integrator, and a computer interface for data acquisition and statistical analyses. Optimum experimental parameters for the sources and burner have been determined. A comparison will be made between the dispersive and non- dispersive methods, and between the sequential and the time-division multiplexing modes. Finally, a computer- controlled system will be described which incorporates a non-flame atomizer. The advantages and disadvantages of the various systems will also be discussed. II. HISTORICAL The phenomenon of fluorescence by free atoms was first observed in sodium vapor above molten sodium by Wood (73) in 1905. Atomic fluorescence in flames was first demonstrated for Be, Ca, Na, Li, and Sr by Nichols and Howes (74) in 1924 and later by Badger (75) in 1929. Robinson (76) observed a fluorescence signal for magnesium in an oxy-hydrogen flame using a conventional hollow cathode lamp in 1961. Alkemade (77) suggested in 1962 that atomic fluorescence spectrometry might have considerable analytical utility. A. Atomic Fluorescence Spectrometry Atomic fluorescence spectrometry was first described as a means of chemical analysis by Vickers and Winefordner (45), and first applied by Staab and Winefordner (78, 79), and Mansfield, Veillon and Winefordner (80). The original work by Winefordner and his group was soon extended by T. S. West (81) at Imperial College London. West and co- workers published a series of articles on applications of atomic fluorescence spectrometry (82—87). The technique 93 94 was pioneered vigorously by both groups originally and now occupies the attention of a large number of researchers in countries all over the world. The implications of its specificity, sensitivity, and simplicity for nondispersive multielement analysis are gradually being recognized and will be demonstrated in this thesis. 1. Types of Fluorescence There are six main types of atomic fluorescence; resonance, direct line, stepwise, thermally assisted, sensitized, and multiphoton. a. Resonance fluorescence occurs when radiation at a given wavelength excites an electron from its ground state energy level E0 to an excited state E j' Subsequently the excited electron returns from Ej to E0 and emits radiation of the same wavelength as was absorbed. Thus the absorbed and the fluoresced energy have the same wavelength. The most common form of resonance fluorescence is between the first excited state and the ground state. Resonance fluorescence is the most common type observed in flame atomizers. b. Direct line fluorescence involves a radiative transfer between a radiatively excited state Ej and an intermediate state Ei above the ground state. Quite commonly Ei may be the first excited state. In this process the wavelength of the 95 fluoresced line radiation is longer than that of the excitation line radiation. This is particularly useful when dealing with analyte solutions that contain matrix material which can scatter the source radiation when nebulized into a flame. Stepwise fluorescence occurs when an atom is radiatively excited to a level Ej from which the electron passes to an intermediate level Bi by nonradiative processes. The return to the ground state Eo from level E1 is accompanied by the emission of radiation. The nonradiative processes that take the electron from Ej to Ei commonly involve deactivation by collision with other species in the atom reservoir or intersystem crossing. If this nonradiative process involves a singlet-triplet transition, Alder and West (88) have suggested that this more appropriately be called atomic phosphorescence rather than stepwise fluorescence. Once again since the wavelength of the fluoresced line is longer than that of the excitation line, scatter problems may be avoided. Thermally assisted fluorescence involves a non- radiative component in the excitation process. For example, an electron is promoted to an excited state E. by absorption of radiation. The electron J is then further promoted to a nearby intermediate 96 level Ej+ by thermal excitation from the flame. It then drops back to a lower excited level Ei by direct line fluorescence or to the ground state by resonance fluorescence. The phenomenon of thermally assisted direct line fluorescence was first observed in 1967 by West and co-workers (86). e. Sensitized fluorescence occurs when one atomic species is excited at its appropriate wavelength and transmits its energy to a second chemical species which thus becomes excited and subsequently emits its own characteristic radiation. Sensitized fluorescence is not very common in flame media. f. Finally, multiphoton fluorescence (89) occurs when two or more identical photons excite an atomic species which then radiatively relaxes. 2. Theoretical Considerations The many articles describing the theoretical basis for atomic fluorescence spectrometry have already been discussed in Part 1. Equation (18) describes the fluores- cence radiance, EAFL' obtained with a line excitation source of radiance, EL. If only dilute solutions are considered, this expression may be reduced to the following _ Q EAFL — (fi)Y EL kol (36) 97 I Therefore, the fluorescence radiance is directly proportional to the number of atoms in the atom reservoir per cc of flame gases, n , since ko a n . When a flame is o o employed an equilibrium value of n is attained ifa o reproducible flame is used to yield a reasonably constant population of atoms in the observation window. The popu- lation of ground state neutral atoms of element E in the observation window is related to the concentration of element in solution CE through an aspiration efficiency factor, a, and an atomization efficiency factor, 8'. n0 0 eB'CE (37) Over a large range of concentration, 58' will remain con— stant. However, at very low CE’ 8' will usually decrease because of increased ionization of analyte atoms, and at very high CE, 8 will decrease because of a decreased efficiency of introduction of submicroscopic forms of the analyte into the atomizer (52). The value of the atomiza- tion efficiency will also vary from one flame to another. Various authors (90-96) have discussed the determination of the degree of atomization in various flames. When a non-flame atomizer is employed, the efficiency of solution transfer to the atomizer, the aspiration efficiency, 6, and the atomization efficiency, 8', are greater than with a flame atomizer. The solvent may be initially evaporated and the resulting salt particles 98 are, therefore, in direct contact with the atomizer. However, an equilibrium value of n0 is not attained as in a flame. Since a transient signal is observed, an inte- gration should be performed to obtain an output proportional to concentration. The non-flame atomizer employed for this study is described in Chapter VI. B. Single Channel Nondispersive Systems A nondispersive system for atomic fluorescence was first proposed by Jenkins (54), who noted that a filter could replace the monochromator to isolate the resonance line of the element of interest. West and Williams (97) also noted that for the purposes of analytical atomic fluorescence where a chemically specific source such as a high intensity hollow cathode lamp is used, the most sensitive results might be obtained by simple use of an efficient narrow bandpass or interference filter instead of a monochromator in which considerable intensity attenuation can occur. Walsh 33.31. (98) suggested the application of solar-blind photomultiplier detectors for atomic fluores- cence flame measurements. Since a monochromator had been used to restrict the noise due to radiation emitted by the flame, most of which appears at wavelengths longer than 3000 A, it is possible to allow the fluorescence radiation to fall directly on the solar-blind photomultiplier and still avoid flame background noise. Such a technique 99 permits the collection of much more fluorescence radiation than is possible when using a monochromator and also has the advantage of observing the fluorescence of all lines lying within the spectral response of the photomultiplier, which enhances the sensitivity of the method. Vickers and Vaught (99) described such an instrument for observing the atomic fluorescence of Cd, Zn, and Hg. This instrument made use of the inherent spectral resolution of atomic fluorescence and did not employ either a filter or a monochromator, but rather a solar-blind photomulti- plier. Warr (100) compared results obtained with an inter- ference filter-photomultiplier combination with those from another nondispersive detection system, incorporating a solar-blind photomultiplier. Also, these results from the nondispersive systems were compared with those obtained from a conventional dispersive system incorporating a monochromator. Improvements of approximately 700-fold and lO-fold respectively in the limits of detection for zinc and mercury resulted from replacing the monochromator with the filter, while results with the filter were similar to those obtained with a solar-blind photomultiplier. Vickers, Slevin, Muscat and Farias (101) described a nondispersive atomic fluorescence system which responded to radiation of wavelengths shorter than 2800 A. Their system incorporated a chlorine cutoff filter with a solar- blind photomultiplier tube. The effects of the high energy 100 throughput and broad spectral response of the system on the signal-to-noise ratio of atomic fluorescence measurements with line excitation sources were discussed. They con- cluded that when a flame atom reservoir is employed, if the spectral bandpass of the measurement system is increased, the signal-to-noise ratio decreases. Therefore, for this case, a nondispersive system, which may be considered as the limiting case of increasing the monochromator spectral bandpass, would 22; be advantageous. As the spectral bandpass is increased, the effect of flame background emission becomes more prominent. If more than one fluo- rescence line falls within the spectral bandpass of the detection system, the situation is more complex, and a non- dispersive system ggy.improve the signal-to-noise ratio.‘ Elser and Winefordner (102) described a simple monochromatorless atomic fluorescence flame spectrometer. Since the sensitivity of atomic fluorescence increases in' proportion to the primary source irradiance, the sensitivity may be improved by using optics which subtend as large a solid angle as possible. For this reason, the authors chose to use mirrors to provide maximum optical speed at lowest cost. Their Czerny-Turner type of arrangement minimized scattered source radiation from the flame come partment. Interference filters were employed along with a solar-blind photomultiplier tube. The authors concluded that the selectivity and sensitivity of their system should 101 be adequate for the analysis of trace metals in polluted air and water. Larkins (103) described a single channel non- dispersive atomic fluorescence spectrometer which employs a solar-blind photomultiplier, and a sheathed flame. The author examined the fluorescence of 24 elements with resonance lines in the 2000-3000 A spectral region, and found that his system gave better detection limits for nine elements compared to atomic absorption flame spectrometry. He found the greatest improvement for the nondispersive system compared to the monochromator system when a number of lines falling within the spectral response of the photomultiplier could be detected simultaneously. C. Multielement Spectrometric Systems Atomic spectrometric methods of analysis have proven extremely useful in trace metal analyses. However, most commercial instruments which employ these methods consist of only a single channel detection system. Although this type of system is very efficient for single element analyses, multielement analyses may be performed only by successive measurements, and this is possible only by changing the detection system. It would, however, be more convenient, and certainly more economical, if simultaneous or near simultaneous multielement analyses could be performed. Various systems striving toward this end have been designed and will now be reviewed. 102 Haagen-Smit.and Ramirez-Munoz (104) described a multichannel integrating flame photometer. Solutions con- taining sodium, potassium, and calcium with a lithium internal standard were aspirated into an air-propane flame. A high speed optical chopping and demodulating system was described which allowed the integration of signal and back- ground. A rotating wheel was employed with filters that passed the resonance lines of the elements to be analyzed. Dawson, Ellis and Milner (105) constructed an automated high speed scanning multichannel spectrophoto- meter for the analysis of Na, K, Ca, and Mg in clinical samples by simultaneous emission and absorption flame spectrometry. An electro-optical wavelength scale generator was developed to ensure precise identification of spectral lines. It was shown that it is possible to obtain precise high speed scanning (18,000 A/sec) over a wide spectral range and to use this facility for multichannel, simul- taneous emission and absorption flame spectrometry. Multielement analysis by means of atomic absorption with a time-resolved spark was reported by Strasheim and Human (106). Standard solutions were aspirated into the spark, which was employed as the primary radiation source. The advantages of this method are the ease with which the resonance radiation of several elements can be obtained simultaneously for use as radiation sources for multielement atomic absorption by controlling the composition of the 103 solution aspirated into the spark, and the ease with which the working range can be altered by varying the concen- tration of the spark solution. A multichannel spectrometer for simultaneous atomic absorption and flame emission analyses has been described by Mavrodineanu and Hughes (107). The desired characteristic lines from hollow (cathode tubes are combined into a single, collimated polychromated beam that is passed through an absorbing medium, is then spatially resolved into its components, and each component is directed to a detector. The sen- sitivity and precision of analysis obtained by this method were found to be equivalent to the results of conventional single element determinations. West and co-workers (108-112) have published a series of articles on multielement atomic fluorescence systems which employ multielement electrodeless discharge lamps (E.D.L.). The construction of a multielement E.D.L. source for Bi-Hg-Se-Te and dual element lamps for Cd-Zn, Ga-In, As-Sb, Ni-Co and many others were described. The performance and fluorescence detection limits obtained with these multielement lamps were of the same order as those obtained with the individual single element lamps. No spectral interference from the presence of 100-fold excesses of the co-elements in the analyte solution were found. Cresser and West (110) manufactured a dual-element E.D.L. which contained elements of quite widely differing 104 volatility, by placing the volatile compound or metal in an outer jacket and the less volatile one in the inner jacket, which was then protected from cooling effects. Cordos and Malmstadt (113) described an automated atomic fluorescence spectrometer for rapid multielement determinations. Their instrument incorporated sequentially pulsed hollow cathode lamps (114, 115), a programmable monochromator for high speed wavelength isolation (116), and a dual channel synchronous integration measurement system (117). The process of producing, isolating, and accurately measuring the fluorescence radiation for one specific element could be completed in one or two seconds. The optimum conditions for the next elements were then preset automatically within an average time of about three seconds. Their instrument may be used to determine up to 12 elements sequentially. The application of a wavelength scanning technique to multielement determinations by atomic fluorescence spectrometry has been described by Norris and West (118). Two dual-element E.D.L.'s were employed, to provide intense resonance radiation for four elements. The atomic fluo- rescence was rapidly measured by scanning over the appro- priate wavelength range. Results obtained for the sequential multielement determination of zinc, cadmium, nickel, and cobalt showed that the sensitivity and selectivity were the same by the scanning technique as by 105 conventional A.F.S. Wavelength scans were also given for nickel, cobalt, iron and manganese, and for selenium, tellurium, nickel and cobalt combinations. The time taken for these measurements was about 100-150 seconds for each scan. Walsh (28) investigated various new techniques for simultaneous multielement analysis. The multichannel spectrophotometer which employs resonance monochromators was previously described in Part I of this thesis. Walsh also described a multichannel spectrophotometer which employs selective modulation. One of the problems associ- ated with these two techniques for the isolation of resonance lines is that the resultant signal may be due to several lines of different oscillator strength, which can result in curvature of working curves. Walsh also dis- cussed the possibility of developing an A.F.S. instrument for multielement analyses. The instrument would incorporate high-intensity lamps, which are switched on sequentially, with pulse durations of 1 msec and pulse separation of 125 msec. Walsh claimed that this approach should improve performances, but did not include any experimental results. Walsh g£_gl. (155) also described a multichannel spectro- photometer for the direct fluorescence spectrometric analysis of metals and alloys by cathodic sputtering. One of the main attractions of this technique is that it offers the possibility of direct analysis of metals and alloys 106 without any chemical treatment. Early experiments indi- cated that it was possible to carry out analyses by the cathodic sputtering technique but there were many practical problems which had to be solved before the method could be considered suitable for routine analyses. Boumans and Boer (119) reported on their preliminary investigations of a nitrous oxide-acetylene flame and an argon plasma torch emission source for simultaneous multi- element analysis. The authors reported that the appli- cation of a nitrous oxide-acetylene flame as an emission source is limited to 40 elements, and for complete coverage of the common group of 70 "spectroscopic" elements, flame atomic fluorescence must be used to complement flame emission. The argon plasma torch was shown to be a suitable excitation source with very low detection limits (.02-10 ng/ml). Simultaneous multielement trace analysis could be performed with only a few compromises. However, due to the high excitation temperature, the plasma torch produces spectra that are rich in lines, and thus requires a high resolution monochromator. Also, matrix effects, such as ionization suppression in the plasma and influences of the matrix on the efficiency of wet and dry aerosal production, do occur. The simultaneous multielement analysis of major cations with a 10,000 °K argon plasma jet was reported by Corcoran of Spectrometrics Inc. (120). The Spectraspan Model 210 echelle spectrometer utilized in 107 conjunction with the argon plasma Spectra-Jet employs a single detector and provides simultaneous measurements for up to ten elements. This combination of a high temperature plasma capable of providing rapid and efficient excitation of liquid samples and a direct reading spectrometer capable of simultaneous measurements could be a valuable tool for analyzing large numbers of samples in a minimal time. Boumans and Brouwer (121) performed studies on photodiodes and phototransistors as detection devices for multichannel emission spectrometry. The response of the phototransistors exceeded that of photodiodes by a factor of about 200. The signal-to-noise ratio of the transistors was at least a factor of 10 higher. It was shown that a one-dimensional array of silicon phototransistors could be used for spectral measurements at closely spaced wavelengths (line and background) in the focal plane of a spectrometer. The spectral response of the phototransistors ranged from 2100 A to 10,000 A. However, fairly high intensity levels were required for these transistors to be useful in flame emission. Horlick and Codding (122) used self-scanning linear silicon photodiode arrays for the detection of spectral information. Linear arrays of up to half an inch in length, and with array densities of about 1000 diodes per inch, were reported. These photodiode arrays have the desired characteristics necessary for an all electronic replacement of the photographic plate as a detector in a 108 multichannel spectrometer. At present, however, they are limited in the wavelength range which can be covered. With 0 a conventional grating monochromator about 150 A could be simultaneously detected (122). Fassel 35 El‘ (123) have described a direct-reading optical emission spectrometer system in which an image- dissector photomultiplier tube serves the dual purpose as a light detector and as a non-mechanical, spectral scanning device. No exit slits are required and programming the instrument for different sets of spectral lines is greatly simplified. These tubes are capable of providing a non- mechanical, linear-scanning system with a wide range of linear intensity response. In a multichannel direct- reading system based on utilizing these detectors, each exit slit and phototube assembly is replaced by an image- dissector tube. Thus, each tube normally serves as the detector for a single spectral line, although one tube can be used to observe two or more closely spaced spectral lines. Effectively then, each image-dissector tube scans a narrow wavelength region in the vicinity of the spectral lines of interest. The image-dissector detection system could also be directly applicable to atomic absorption instruments which utilize either hollow cathode line sources or spectral continuum sources. A television camera tube, which offers a com- bination of the properties of photographic and photoelectric 109 detection, was suggested by Margoshes (124). The camera tube combines two advantages of the photographic emulsion; recording of an entire image, and integration during the exposure time. It also combines many desirable features of the photomultiplier tube, such as having a high sensitivity and direct conversion of the radiation signal to an elec- trical analog signal. A multichannel spectrometer employing such a camera tube would have several advantages. Mitchell and Aldous (125) employed such a tube, a vidicon detector, in a multichannel spectrometer. A comparison was made between this multichannel system and a single channel photomultiplier system in terms of spectral response and sensitivity. Detection limits and working curves were reported for the analysis of trace elements in single and multielement samples by atomic emission and fluorescence. D. Multielement Nondispersive Systems Various instruments have been used for simultaneous multielement analysis, such as X-ray fluorescence spectro- metry (X.R.F.), spark and arc emission spectrometry, spark source mass spectrometry, and thermal neutron activation analysis. Each of these techniques has its own advantages and disadvantages. The detection limits that may be attained improve in the order the techniques have been arranged. However, there is a corresponding increase in the cost and complexity of the instrumentation. These techniques have been complemented by flame emission, flame 110 absorption and flame fluorescence, and the various attempts to employ multichannel instrumentation for these techniques have been reviewed in the previous section. The conventional design of these multichannel spectrometers is a multiple slit-multiple detector system. However, instruments of this type are limited in the number and location of the wavelengths at which intensities can be measured simul- taneously. Therefore, many of the techniques employ a single detector-scanning method to alleviate this problem, or other methods described earlier. Another interesting technique of performing simultaneous multielement analyses is a completely non- dispersive method. This method has been used for X-ray fluorescence by employing a Si (Li) detector instead of the conventional wavelength dispersive goniometer. The major advantage of this energy dispersive, or nondispersive, xeray fluorescence method is rapid qualitative and semi- quantitative analysis. The method can determine all the elements from Na to U. An actual scan time is 200 seconds as compared to 26 minutes for conventional X.R.F. However, the detection limits achievable with this method range from 10-50 ppm. The nondispersive method also has the dis- advantage of lower resolution, but it does not suffer from interference of higher order lines characteristic of crystal spectrometers. Both x-ray methods suffer limitations in quantitative analysis due to the presence of strong matrix effects. 111 The nondispersive technique for multielement' analysis can also be very useful for atomic fluorescence. The working range for nondispersive X.R.F. is from ppm to 100 per cent, whereas the working range for nondispersive atomic fluorescence is ppb to ppm. Therefore, the two methods may complement each other. A nondispersive atomic fluorescence method for multielement analysis was first described by Mitchell and Johansson (127, 128). Their instrument employed four sequentially pulsed hollow cathode lamps, a rotating filter wheel, and a synchronous inte- gration system. The instrument was used for the deter- mination of Ag, Cu, Fe, and Mg using an air-hydrogen flame, and their results were quite comparable with those obtained by others who worked with conventional monochromator instruments. A six-channel atomic fluorescence spectro- meter based on this system was manufactured by Technicon Corp., Tarrytown, N.Y. West and covworkers used this AFS-6 instrument for the determination of metals in soil extracts (129), in aluminum alloys (130), and with pre- concentration by an automated solvent extraction procedure (131). The AFB-6 instrument claims to perform a Eiflfllf taneous multielement analysis. However, by the nature of its rotating filter wheel, it is rather a sequential multielement analysis. By definition, the scanning tech- nique of Norris and West, the programmable monochromator 112 technique of Cordos and Malmstadt, and the rotating filter wheel of Mitchell and Johansson are all sequential methods of analysis because one element is determined 2££g£_the previous element. In Part II of this thesis an automated nondispersive atomic fluorescence spectrometer is described which performs nearly simultaneous multielement analyses, in the time-division multiplexed mode. The pulsed hollow cathode lamps for increased intensity and time separation, the sheathed burner for flame background reduction, the synchronous integration system, and the computer interface for data acquisition are all described in the subsequent chapters. III. PULSED HOLLOW CATHODE SOURCES FOR ATOMIC FLUORESCENCE SPECTROMETRY The fluorescence radiance observed in an atomic fluorescence measurement is directly proportional to the radiance of the excitation source as is shown by Equation (36). Therefore, to improve the sensitivity of the atomic fluorescence technique, various sources have been modified and/or constructed which have an increased output intensity. Sullivan and Walsh (18) reported the use of a high inten- sity hollow cathode lamp which employs one electrical discharge to produce the necessary atomic vapor by cathodic sputtering, and a second electrical discharge to produce the necessary excitation. A modified version of this lamp, in which the secondary discharge passes through the center of the open-ended cylindrical hollow cathode, was described by Lowe (132) as a source for flame atomic fluorescence. His results showed that this modified form emits resonance lines with considerably greater intensity than the con- ventional high-intensity hollow cathode lamp. Dinnin (133, 134) employed a demountable hot hollow cathode lamp as an 113 114 excitation source for A.F.S. The cathode is a graphite rod -with a cavity at one end to hold a small amount of the element of interest. The intensity of the hot cathode lamp was comparable to that obtained by commercially available high intensity hollow cathode lamps for gold and silver, and 10 times more intense than a 150 W xenon arc lamp for gold, silver and nickel. However, operation of the hot cathode for low boiling elements waslnot as convenient as with the more refractory metals. Winefordner 22,21. (135, 136) have used a 150 W xenon arc continuum source for A.F.S. measurements. The advantage of the continuum source is that only one source is required for all elements. Furthermore, the continuum source does not require stringent restrictions on the spectral bandpass of the monochromator as in A.A.S. However, the intensity of the source over a narrow bandwidth (spectral radiance) is not as great as that of a hollow cathode. This is especially true in the range from 2000-3000 A where A.F.S. is most useful. E.D.L.'s (135-137) have also been used for A.F.S. for single element and multielement analysis. However, the cost of (the microwave power supplies can be quite large if more than one lamp is to be employed in a simultaneous mode. Winefordner gtflal. (138) and Piepmeier (139, 140) have used a pulsed, tunable dye laser to excite atomic fluorescence. The advantage of the increased intensity of the laser is that a nearly saturated population provides greater 115 regulation of the fluorescent radiation with respect to changes in laser power density and quenching. However, the dye laser has the disadvantage of a very high cost and a lower wavelength limit of approximately 2500 A at present. The spectral source which appears to be highly suitable for atomic fluorescence with respect to cost, availability, wavelength coverage, selectivity and intensity is the hollow cathode lamp operated in a pulsed mode. Previous work performed with pulsed hollow cathode lamps will first be reviewed. The circuitry for the pulsed hollow cathode lamps employed in this study will then be described. Finally, optimum experimental parameters for the pulsed sources obtained for multielement atomic fluorescence will be discussed. A. PreVious Work on Pulsed Hollow ACEEHSHE_Lamps ‘ ' The requirement of a high intensity source for atomic fluorescence has led to new modes of operation and design for the standard hollow cathode lamps. Some authors (141) have merely operated these lamps at higher currents than previously suggested by the manufacturer. This increased current does produce an increased output inten- sity. However, as discussed in Part I, the increased current also produces increased line broadening and can lead to self-reversal. Furthermore, lamps operated at high currents have a shorter lifetime due to decomposition 116 of the cathode material and plating onto the walls of the tube. This problem was alleviated by Dawson and Ellis (142), who described the operation of hollow cathode lamps in a pulsed mode. By passing large currents (300-600 mA) of short duration (15-40 usec.) repetitively through con- ventional hollow cathode lamps, the authors obtained resonance line emission many hundred times more intense than that obtained under equivalent steady current oper- ation. The authors also reported that the resonance lines were not significantly broadened in the pulsed mode. Willis (143) also employed pulsed hollow cathode sources, but reported that the large increase in emission intensity was achieved at the expense of an increase in broadening of the resonance lines emitted. This increase in line-width was demonstrated very clearly by the use of a resonance mono- chromator in place of a conventional monochromator. The fluorescence signal from the resonance monochromator excited by a pulsed copper hollow cathode lamp was only about five times that obtained by operating the lamp in the dc mode at the same average current. Since the resonance monochromator has a spectral bandpass of about 0.01 A, this experiment showed that the increase in intensity at the center of the line obtained with pulsed operation is not as large as found by Dawson and Ellis (142) for the integrated intensity over the whole line. This increase in broadening seems reasonable if one considers the 117 mechanism of excitation in a hollow cathode lamp operated at high currents. Bfiger and Fink (144) have claimed that the discharge filler gas is excited or ionized by electron impact. The discharge ions then collide with the cathode walls where they are neutralized or release a neutral atom of the cathode material. Due to this cathodic sputtering, these neutral metal atoms are brought into the plasma where they can be excited by electron impact or by col- lisions of the second kind. At higher currents more dis- charge gas is ionized and, hence, more neutral atoms are formed. These neutral atoms can either form a deppsit at the cathode or leave the discharge by diffusion. Thus, in a hollow cathode discharge there are two separate regions in the tube. Inside the hollow cathode there is a plasma emitting and partly absorbing the resonance emission. Outside the electrode there is a slab (145, 146) containing metal vapor diffusing from the discharge, which is absorbing only. If the concentration of absorbing atoms between the emitting atoms and the lamp window is significant, additional broadening will be experienced by the resonance lines (147). Therefore, it is not unreasonable that hollow cathode lamps operated in a high current pulsed mode do indeed show effects of self-reversal due to the increased concentration of neutral atomic vapor. Despite the increase in emission line width, pulsed hollow cathode sources are still very useful for atomic fluorescence. The increase 118 in intensity over conventional dc operation, although not several hundred times as reported by Dawson and Ellis, is still one or two orders of magnitude larger. The exact intensity increase depends on the element and filler gas. Furthermore, pulsed hollow cathodes allow the possibility of sequential pulsing for multielement analyses. Pulsed hollow cathode sources were also used by Kielkopf (148) who operated them at peak currents between 300 and 1500 mA. During the current pulse the concentration of the vaporized cathode material inside the hollow cathode 14 atoms/cc. With this pulsed hollow was on the order of 10 cathode lamp, excitation of up to the third spectra of iron and aluminum was observed, which corresponds to an electron temperature of 15,000 °K. The temperature decreased with increasing time and decreasing peak current. A time- resolved study indicated that radiation could still be observed 700 usec after the discharge. Osten and Piepmeier (149) used pulsed hollow cathode lamps by atomic absorption measurements on material vaporized by a Q~switched laser. Barnett and Kahn (150) employed pulsed hollow cathode sources for a comparison of atomic fluorescence with atomic absorption. Weide and Parsons (151) used pulsed hollow cathode lamps for atomic fluorescence flame spectrometry. Finally, the pulsed hollow cathode sources employed by Mitchell and Johansson (127, 128) and Cordos and Malmstadt (113, 116, 117) have previously been cited. 119 B. Circuit for Pulsed Hollow Catfiode Sources For the atomic fluorescence measurements reported in this thesis, the pulsing circuitry described below was employed. The circuit for operating one hollow cathode discharge lamp in the pulsed mode is shown in Figure 21. An operational amplifier, OA, controls the voltage applied to the base of a driver transistor. The feedback control, for current regulation, is obtained by connecting the feedback resistor, Rf, to the emitter of the transistor. The current through the lamp, iL, is given by the expression in Figure 21, and can be adjusted by varying the input resistor, Rin’ A field effect transistor, FET, enables the power supply to be operated in an intermittent mode by switching the reference voltage, er, ON and OFF. The FET requires a separate driving circuit which is shown in Figure 22. TTL logic may then be used to control the ON and OFF times. For example, an astable multivibrator was employed to control the frequency of the intermittent operation as is shown in Figure 23.a. This produces a square-wave output whose period is determined by the RC time constant. Frequencies of ten to twenty Hz were employed. The output of the astable multivibrator is directed to a monostable multivibrator, Figure 23.b, which is triggered on a "l" to "0" transition. The pulse width, or ON time of the lamp, was typically 5 msec as determined by the RC time constant. The TTL output of the monostable 120 it: 2.3.; R5 Fifi! ' Figure 21. Circuit Diagram for Pulsing One Hollow Cathode Lamp. 121 .‘ISV 10k 390pF 'SOpF ON- F—-j * ' ‘ ~ OFF ‘7 Control ' ' ' Signal 100k ‘5 r. 100k IN- ‘ oour ‘ FETM 0 '*5V Figure 22. Diagram of Circuit to Drive FET. was used to drive the PET in the OA circuit and to provide a reference signal for the measurement circuitry. To operate four hollow cathode lamps sequentially, FETs were placed between the OA output and the bases of four driver transistors as is shown in Figure 24. FETs were also introduced in the input circuit so that the peak current for each of the four hollow cathodes could be individually controlled. The sequencing of the four hollow cathode lamps was controlled by a shift register as is shown in Figure 25. When the sequencing circuit is activated, the first hollow cathode lamp is operated in an intermittant mode for two seconds, during which time the peak current is controlled by the first input resistor. After two seconds the sequencing circuit shifts to the second hollow cathode lamp with its preset peak current. This process 122 Figure 23. Logic Control Circuit. (a) Astable Multi- vibrator; (b) Monostable Multivibrator. 123 .A.o.u.m usom mcwmasm kHHmwucmsomm uom baseman :vm musmwm Oanu um egos... am cxoom+ousm >«u to 124 SEQUENC ING CIRCUIT a] N Y 7R 0: A C P82 o—J (3 I J. <3 I J (}-—JL—-J CD~OE> l C ‘ —I-—I< O K O' unlo— .4 PULSE wuom = 2 SECONDS Figure 25. Shift Register Sequencing Circuit. PB 2 is Push Button Switch to Activate Circuit. 125 continues until all four lamps have been sequentially operated for two seconds each. Therefore, the corresponding atomic fluorescence that is excited by these lamps may be measured directly by a photomultiplier transducer while still maintaining a separation in time. Circuitry for operating the lamps in the time-multiplexed mode is described in Chapter V. C. Pulsed Source Characteristics One of the advantages of using pulsed hollow cathode sources is that the increase in current during the ON time is accompanied by a corresponding increase in emission intensity, while the average current over a full cycle is still maintained at a safe operating level. The exact relationship between the intensity, I, and the current, i, has been given by L'vov (152), I = ai (38) where a and n are specific constants for each combination of cathode material and filler gas. The power index n has been determined by Crosswhite gtflgl. (153) and was found to be 1.3, 1.9, 2.5, and 2.9 for krypton, argon, neon, and helium respectively. If self-absorption occurs in a lamp, there will be a deviation in the relationship between I and i, which will appear as a deflection of the slope toward the horizontal axis in a plot of log I vs. log i. For example, the relationship between intensity and current in 126 the dc mode was determined for zinc, cadmium, and magnesium hollow cathode lamps, and the results are shown in Figure 26. The slopes of these curves are given in Table 6. The zinc and cadmium hollow cathode lamps do not show appre- ciable self-absorption even at dc currents of 50 mA. The magnesium lamp, on the other hand, shows the effects of self-absorption at 35 mA. Recall the profile of the' 2852 A Mg resonance line in Figure 14 at varying dc currents, which demonstrated self-reversal at 35 mA. When operated in the pulsed mode, the exponent n (i.e., the slope of the line) depends on the particular lamps being investigated. For example, L'vov reported n values of 1.8 for Ni and 3.0 for Pd. The value of n for the zinc lamp operated in the pulsed mode was experimentally determined here to be 2.42. Once the n value for a lamp has been calculated, the anticipated improvement in inten- sity, for the same average lamp current, may be determined from the following equation. I pulsed. _ .1 pulsed “ do do For the experimental data presented, peak currents of 200 to 300 mA.were employed. Since duty cycles of 1/10 to 1/30 were used, this corresponds to mean currents of 8 to 25 mA. Since the atomic fluorescence sensitivity, and hence the detection limit, is directly proportional to the 127 INTENSITY, A d 0| o l 10" — 1'0 100 DC. LAMP CURRENT.A Figure 26. Intensity of H.C.D.T. as a Function of DC Current. 0 Zinc E] Magensium o Cadmium 128 Table 6.--Slopes Calculated for Various Lamps in DC and Pulsed Modes. Element Filler Gas Slope = n Mode Zn Ne 2.30 dc Cd Ne 2.34 do Mg Ne 2.28 do Hg Ar 1.78 do Zn Ne 2.42 pulsed intensity of the source, maximum currents were used in all cases, without arriving at the point of self-reversal. The effect of pulse width on relative intensity was also investigated. Pulse widths of l to 5 msec were employed and the results are listed in Table 7. These data were obtained with the computer-controlled system described in Chapter VI. Except for the pulse width of l msec, the intensity appears to be independent of pulse width. These results agree with those of Muscat (156) who determined the effect of pulse width on detection limit for a Ni hollow cathode lamp. The detection limit for Ni decreased (intensity increased) until a pulse width of 300 usec was attained, after which it remained essentially constant. Muscat used the pulsing circuit described by Dawson and Ellis (142) with a peak current of 400 mA and a modulation frequency of 1000 Hz. Cordos and Malmstadt (115), who employed a different pulsing circuit, similar to the circuit 129 Table 7.--Relative Lamp Intensity as a Function of ON Time. ON Time Relative Intensity, Arbitrary Units 1 msec 200 2 msec 245 3 msec 265 4 msec 265 5 msec 274 used in this work, reported lamp response and power supply response problems at ON times less than 2 msec. For the experimental work reported in Chapter V, an ON time of 5 msec was employed. The final characteristic of the lamps investigated was the stability of the output radiance. This was deter- mined by introducing a metal cylinder between the sources and photomultiplier transducer to produce a scattering signal. A pulse width of 5 msec, a duty cycle of l/l6, and 20 total pulses were employed. The relative standard deviation of the scattering signal ranged from 0.67 to 0.85 per cent. IV. SHEATHED BURNER FOR NONDISPERSIVE STUDIES In A.F.S., as in A.A.S., the flame functions mainly to produce a population of ground state atoms. However, the flame used for A.F.S. should also have low background emission, a low concentration of quenchers, and a long residence time of analyte atoms in the observation window. The low flame background is not as important in A.A.S., since the radiance of the source is many times greater than the radiance of the flame. However, in A.F.S., the radiance of the flame is comparable to the fluorescence radiance, especially at the detection limit, and therefore, may be the limiting factor. Various flames have been used for A.F.S. in an attempt to reduce background emission. Winefordner £2 21. (157) discovered that the fluorescence intensity of Mg could be increased by adding argon to the oxygen supplied to the burner. The fluorescence intensity increased with increasing argon concentration until pure argon was employed with only entrained air maintaining combustion. However, this cooler flame has been shown to suffer from increased interferences (158). Bratzel and Winefordner (159) have 130 131 determined the influence of the type of turbulent flame on limits of detection in atomic fluorescence flame spectro- metry. The limits of detection were found to be critically dependent on flame type, but were not strongly dependent on the fuel and aspirant flow rates. Winefordner gtflgl. evaluated premixed flames produced by using a total con- sumption burner (160) and premixed air-hydrogen flames (161) for atomic fluorescence spectrometry. The major breakthrough in the reduction of the flame background was the introduction of the separated flame first described by Kirkbright and West (162). The application of separated flames will be reviewed, and the sheathed burner designed for this study will be described. A. Previous Work on Separated Flames In a conventional flame the background emission comes from both the primary cone, where the fuel gas and oxident undergo reaction, and from the secondary zone, where the unburned residues of the primary cone combust by reaction with atmospheric oxygen which diffuses in toward the center of the flame (163). Virtually all of the hydrocarbons are broken down in the primary reaction zone and carbon monoxide and hydrogen are the main unburned products which burn in the secondary zone. The most intense emission from the secondary zone comes from the OH band system in the region of 3100 A. If atmospheric oxygen is denied access to the base of the flame by sheathing it with 132 an inert gas, this secondary combustion is prevented. Therefore, the atomic fluorescence measurement may be made in the secondary zone where background emission is extremely low. Separated flames of this type were used by West 25,31. (164-169) for atomic emission, atomic absorption, and atomic fluorescence measurements. Jenkins (90) used a shielded flame to determine the effect of flame composition on the fluorescence yield in the absence of air entrainment. He also demonstrated the efficiency of the shielding flame in reducing the rate of quenching excited atoms. Larkins and Willis (170) described the use of a nitrogen sheathed, water cooled burner for nitrous oxide-acetylene and nitrous oxide-hydrogen flames. These flames in conjunction with a high intensity hollow cathode lamp and a nondispersive system were useful for the determination of metals forming refractory oxides and having their resonance lines below 3000 A. A separated air-acetylene flame was utilized by Browner and Manning (171) for atomic fluorescence spectro- metry. The argon separation provided a ten-fold reduction in background emission, with a corresponding reduction in the background noise of about 3 times. Slevin, Muscat and Vickers (172) evaluated a premixed Maker-type flame with provision for inert gas sheathing. These authors reported that with the sheathed flame, involatile compound formation was not as serious a problem as with a standard Maker-type burner. The atomization efficiency was greater, and the 133 atomic concentration in the flame was more uniform. These factors make the choice of the viewing region of the flame less critical and could lead to enhanced sensitivity when coupled with a nondispersive system. The separated flame does, however, have one dis- advantage, i.e., the temperature of the primary cone decreases with increasing sheath gas flow rate. Kirkbright and Vetter (173) have determined the effect of inert gas sheathing on temperature profiles in premixed flames. The extent to which the analytically useful interconal region of the flame is cooled on sheathing with inert gas depends on the flame composition, height of observation and whether argon or nitrogen is used as the sheath gas. The sheathed fuel-rich flame is cooler than the unsheathed flame by 40 to 180 °C in the region between 2 and 25 mm above the primary cone. In general, nitrogen cooled the flame to a greater extent than argon. Despite the disadvantage of a cooler temperature, the sheathed flame is of tremendous importance in atomic fluorescence spectrometry. This is especially true in a nondispersive system for multielement analysis, since the photomultiplier transducer views the flame directly. Therefore, a sheathed burner was designed for our system and is described in the next section. 134 B. Burner Design The burner designed for this study is shown in Figure 27. Two stainless steel plates were fabricated to mount directly on a Jarrell Ash pre-mixed burner. Fuel, oxidizer, and sample droplets pass through the inner circular array of holes in the top plate. Sheathing gas, nitrogen or argon, enters the bottom plate through tygon tubing, mixes in the canal, and passes through the outer circular array of holes in the top plate. As the flow rate of the sheath gas increases, the flame separates into its primary cone and its secondary zone. The effect of the sheath gas is described in the next section. The dimensions of the bottom plate are 14 x 5 x 1.5 cm. The large inner hole has a diameter of 2.2 cm. The outside diameter of the canal is removed by 0.2 cm from the inner hole. The diameter of the canal is 0.4 cm and the depth is 1.0 cm. The sides of the canal are extended, as shown in Figure 27, to achieve efficient circulation of the sheathing gas. The top section has dimensions of 14 x 5 x 0.5 cm. The outer circular array consists of 16 holes, each 0.8 mm in diameter. These holes are positioned directly above the canal. The inner circular array consists of concentric circles of 16, 16, 16, 8, and 1 holes, each 0.8 mm in diameter. All of these holes are directly above the large hole in the bottom plate. With this design the sheath gas 135 .umcnnm omnummnm mo Emummflo . 200m< fl. (//1\l\ "OZ—mad. ZOO>._. > wdeO _ .— . _ H _ _ _ _ . _ . . .. . . b a .. :5. \U 7. .14 ,. u u .. u 1 n 1 __ u a «p \m« s . 4H5 mwwszzpw . em enema 136 surrounds the flame without coming in direct contact with the fuel, oxydizer, or sample droplets. C. Burner Parameters The burner parameters that were optimized were the type of sheath gas, the position of the burner relative to the observation window, and the flow rate of the sheath gas. The sheath gas has many effects on the measurement of the fluorescent intensity. Sheathing can a. reduce quenching by atmospheric nitrogen, b. constrict and therefore control the height of the visible portion of the flame, c. cool the primary cone by about 100 °C, d. increase the atomization efficiency, e. render the concentration of atoms in the flame more uniform, f. separate the flame into its primary cone and secondary zone by preventing the unburned hydrogen from reacting with diffused atmospheric oxygen. The sheath gases that were employed for this study were argon and nitrogen. The effect of the type of sheath gas on detection limits is presented in Chapter V. A comparison between argon sheathing and nitrogen sheathing was also made by West (174) for various elements. He obtained an improvement of 1.1 to 2.0 when using argon sheathing for the various elements. This is probably due to the fact that nitrogen sheathing cools the flame to a 137 greater extent than argon sheathing. In atomic fluorescence spectrometry the major noise source at the limit of detection is the flame background, even after the flame has been separated. This is especially true in the non- dispersive mode. To reduce this noise, the tip of the primary cone should be lowered relative to the observation window. However, as the flame is lowered a point is reached where the analyte fluorescence decreases even more drastically than does the noise due to the flame background. This is evident in Figure 28 where the fluorescence signal 1:1?) viii?! | for lead and the flame background at 2833 A (spectral bandpass of 20 A) are plotted as a function of the burner position relative to the observation window. These data were taken at a constant flow rate of argon sheath gas. The optimum position for lead is with the burner as close to the observation window as possible. The fluorescence signal for lead decreases drastically as the burner is lowered because the temperature decreases. Since the boiling point of lead (b.p. 1744 °C) is rather high, a decrease in the flame temperature can be detrimental. The temperature of the air-H2 flame after separation is about 2000 °C at the tip of the primary cone. In the case of zinc, the decrease in fluorescence signal is not as severe as is shown in Figure 29. The boiling point of zinc is 907 °C, therefore, the decrease in flame temperature is not as important as in the case of lead. The profiles 138 I2 A Z (I) :3 1: ° 2 32 :3 4 :6 ‘1 a: a: .3- ~43 :5 a" o. 5 9. ”I- D 2 e l-' (5 .55 ,2- —»2 2‘ .0 O 0‘) < .m In 0: IL! 0 2 D ... S .u. ‘ _,| u. o ."‘ . as ...l I I I O 2 . ' 4 6 BURNER POSITION. INCHES Figure 28. Lead Fluorescence Intensity and Flame Background as a Function of Burner Position.* Error Bars Represent One Standard Deviation. [:1 Lead fluorescence 0 Flame background *Distance between top of burner and bottom of observation window. 139 (’5 t: 18" Z ' " A CD 3 I: g g “’- .m‘ t a: ('75 .16- 5 z . . I‘-’ ‘73 g —.05 a E a: ' In C) L) a: . m 34‘ 0 m < 8 In In fig ~93 E; o :3. E .m— N “‘ —.OI .10 I I I o 2 4 a BURNER POSITION. INCHES Figure 29. Zinc Fluorescent Intensity and Flame Background as a Function of Burner Position. E] Zinc fluorescence () Flame background 140 for cadmium (b.p. 765 °C) and mercury (b.p. 357 °C) were similar to that of zinc. The optimum signal-to-noise ratio for the dispersive case, therefore, occurred when the tip of the primary cone was just below the observation window. Under these conditions, the flame background spectrum before and after separation, is as shown in Figure 30. Profiles could not be determined for the nondispersive case because the flame background.was too intense in the higher positions. Burner position profiles obtained in the dis- persive mode dictated that the burner should be placed as close to the observation window as possible for maximum fluorescence. However, in the nondispersive mode, the burner should be positioned as far away as possible from the observation window to reduce the flame background. There- fore, a compromise had to be made. All the data presented for the nondispersive case were Obtained with the burner positioned as near to the observation window as possible without overloading the measurement system. To determine the effect of sheath gas flow rate on the fluorescent intensity and the background signals, data were taken at a constant burner position in the dis- persive mode. The results for lead with both argon and nitrogen sheathing are shown in Figure 31. Note that the fluorescent intensity decreases in both cases with increasing sheath gas flow rate because the temperature decreases. Note also that the fluorescent intensity 141 FLAME BACKGROUND (ARBITRARY UNITS) l L l l 2600 2300 2600 2900 $200 WAVELENGTH, A Figure 30. Flame Background Spectra of Air-H2.Flame. a. No sheath gas b. Argon sheath gas 142 8 I: 2! as D Z (a m. .3—4 t: a: 2! :5 :3 ' mi .>_' - a: cg s LIJ - D '"Z" 2 _. . 5 :3 k- ,24 . M C) 2: a: . u; o . g2 .8 -- -.. o ‘ UJ a .I ‘I 6 ° °° a -.. g .“- -- 3 C3 J'_ LL :5 -.2 .J -J I I _ I 10 2O 30 How RATE, L/MIN. Figure 31. Lead Fluorescence Intensity and Flame Background as a Function of Flow Rate in Dispersive Mode. o-u () Lead fluorescence; argon sheathing [3 Lead fluorescence; nitrogen sheathing 0 Flame background 143 decreases more for nitrogen sheathing because of the enhanced temperature decrease. The background is signifi- cant at 2833 A, and also decreases with increasing sheath gas flow rate. Flow rate profiles for mercury, zinc and cadmium are shown in Figure 32. The fluorescent intensity decreases in the case of mercury, while results for zinc and cadmium showed a slight increase followed by a decrease with corresponding sheath gas flow rate. The temperature decrease is not as important for these three elements as it is for lead. Flow rate profiles were also obtained in the non- diapersive mode. However, measurements could not be made at low flow rates, because the flame was not effectively separated, and hence the flame background was too great. The flow rate profile with nitrogen sheathing for cadmium is shown in Figure 33. Notice now the tremendous decrease in background signal with increasing flow rate; whereas the flow rate has only a slight effect on the fluorescence signal. Similar results were obtained for the other three elements. One can now see the tremendous importance of the sheath gas flow rate when a nondispersive system is employed. 144 museumouosam ggmo I oocmommuogm Osaa . mocoomouosam Soyuz ‘ .mpoz w>amuommfia H." ovum spam no sounuossh n ma. ramsmucH oosoommuosah gflgmo can open $9588: .23? . E51 26.: .2092 cu . on o _ _ Ind I o... ( SlINn AWHIISHV)A1ISN31NI lNaOSBHOITH .Nm musmflm 145 .4- W4 " CD I: Z 3 '16 OK .3- -.3 S d 2 8 I: 3 .2- -.2 0 <[ m . Lu - 2 < .1 u. .I-I CADMIUM FLUORESCENT INTENSITY, (ARB. UNITS) I r I 40 30- NITROGEN FLOW RATE,L/MIN. Figure 33. Cadmium Fluorescence Intensity and Flame Back- ground as a Function of Nitrogen Flow Rate in Nondispersive Mode. l] Cadmium fluorescence 0 Flame background V. INVESTIGATION OF NONDISPERSIVE MULTIELEMENT ATOMIC FLUORESCENCE FLAME SPECTROMETRIC SYSTEMS To perform a multielement atomic fluorescence measurement, the fluorescent intensities of the various elements must be separated. This separation can be made in time, as with a rotating filter wheel, scanning mono- chromator, or sequentially programmed monochromator, or it can be made in space, as with a direct-reading spectro- meter, a photodiode array, or a vidicon detector. The former method employs a single channel detection system and the measurement is performed sequentially. The latter method employs a multichannel detection system and the measurement is performed simultaneously. In this chapter a new method of performing a multielement analysis is introduced, the time-division multiplex mode. This method employs a single channel detection system to perform a nearly simultaneous measurement. Various sequential methods are first described, and their results are then compared to those obtained with the time-division multi- plex mode. 146 147 W Atom c uorescence Anaky§_§. l. Instrumentation A block diagram of the complete system used for this study is shown in Figure 34, and the specific experi- mental system is described in Table 8. For comparison purposes the synchronous integrator was replaced by a lock- in amplifier. Also, a conventional grating monochromator was introduced into the system so that dispersive and non- dispersive modes of operation could be compared. To measure the fluorescent radiances during the ON times of the lamps, a synchronous integrator was employed as is shown in Figure 35. The fluorescent radiance is integrated only during the ON time of the lamp and held for a preset time during which an analog-to-digital (A/D) conversion is made. The capacitor is then discharged and the integrator waits for the next pulse. For example, Figure 36 is an oscilloscope trace of the fluorescent intensity obtained for a 5 ppm Cd solution and the corre- sponding output of the integrate and hold circuit. In sequential operation the first hollow cathode lamp is operated in an intermittent mode for a certain time (i.e., 2 sec.), then the second lamp, etc. The respective operation times of the lamps are controlled manually by switches for the dispersive mode, or auto- matically by a shift register for the nondispersive mode. 148 BLOCK DIAGRAM OF SYSTEM I POWER SUPPLY . ‘ ’ - LOGIC “33$?” __ CONTROL CLOCK CIRCUIT L I- .\ \[HOLLOW CATHODE] LAMPS - LAB 8/E 0 L31: SYNCHRONOUS COMPUTE \‘ ' s l \ l \ .' A PRE- I .~ “M'T- AMP.- INTEGRATOR [ATOMIC] . VAPOR Block Diagram of Atomic Fluorescence System. Figure 34. 149 Table 8.--Experimental System for Atomic Fluorescence. Component Sources Source power supply Burner Lenses Monochromator Photomulti- plier tube Photomulti- plier power supply Pre- amplifier Synchronous integrator Lock-in amplifier Logic control and sequencing circuit Computer Description and Type Hollow cathode discharge lamps Model EUB7OB-7O hollow cathode power supply, modified (114) Premixed, air/H2 modified (Figure 27) One inch diameter, plano- convex and biconvex quartz lenses Model ED 700, 350 mm focal len th, f/6.8 aperture 20 /mm reciprocal linear dispersion R 166 Solar blind Model EU-42A high voltage power supply Model 427 current amplifier See Figure 35 Model 840 Autoloc Model EU 801-11, 12, 13 analog digital designer PDP Lab 8/E Supplier Fisher Scientific CO. Pittsburgh, PA Heath Co. Benton Harbor, MI "5 Fisher Scientific Co. ‘fl Pittsburgh, PA Esco Optics Products 171 Oak Ridge Oak Ridge, NJ W; Heath Co. Benton Harbor, MI Hamamatsu Corp. Lake Success, NY Heath Co. Benton Harbor, MI Keithley Instruments Cleveland , OH Keithley Instruments Cleveland, OH Heath Co. Benton Harbor, MI Digital Equipment Corp. Maynard, MA 150 INPUT c L—oro A/D CONVERTER Figure 35. Circuit Diagram of Synchronous Integrator. D Controls the ON Time and E Controls the Hold Time. 9. J. I-n IIVTUI IIAIALLAIAILAAAIIIIIII TWUI'IIUIIIIIIIWVIIJV IIUIUII if: ,L 1 VOLT/ DIV r 20 MSEC/‘DIV L I l Figure 36. Oscilloscope Trace of Cadmium Fluorescence Signal (Lower) and Corresponding Output of Integrate and Hold Circuit. {é "mu «a. "I 151 The system employed in the sequential dispersive mode is similar to that described by Cordos and Malmstadt (113), who employed a programmable monochromator, and by Norris and West (118), who employed a scanning technique. The major advantage of this sequential operation is that parameters may be optimized for each specific element during the measurement time. The frequency of the intermittent mode is determined by the astable multivibrator, and the ON time by the monostable multivibrator. The peak currents for the H9: Cd, Zn, and Pb lamps during their respective ON times were 265, 200, 215, and 250 mA as determined by measuring the voltage e shown in Figure 21. For example, the upper s traces in Figure 37.a and 37.b represent the peak currents for the Cd lamp, and the lower traces represent the corre- sponding cadmium fluorescence obtained for a 5 ppm cadmium solution. The duty cycle for this case was 1/30. In the sequential dispersive mode, when the first hollow cathode lamp is manually turned on, the computer stores the data in the first dimensional array. An A/D conversion is made during the hold time of the synchronous integrator determined by the TTL logic of the sequencing circuit to the clock input. The analysis time for one element is dependent on the number of pulses, e.g., 25 pulses would correspond to three seconds. [At the end of the measurement time the wavelength is adjusted for the 152 I I lIll n lLlLllll ' .U'V'U'II'VIUUVUU ' (a) VVVVVVVVVVVVVVVVVVVVVVVV 2 VOLTS I DIV E g 1 I /DIV Tb)” 2 VOLT I I 1 l l L l I 20 MSEC/ DIV Figure 37. Oscilloscope Traces of Peak Current for Cadmium Hollow Cathode Lamp (Upper) and Corresponding Atomic Fluorescence Signal for 5 ppm Cadmium Solution. 153 second element, the second lamp turned on, and the data stored in a second dimensional array. This process is continued until fluorescence measurements have been made for all four lamps. The entire process is then repeated for a total of five times. At the end of the last measure- ment, the program enters a calculation routine in which the average and standard deviation are determined. For the data presented, the average and standard deviation for five water blanks were stored in core. The signal-to-noise ratio for five fluorescence measurements were then calculated according to the following formula § (§)F= fi'an 2 F 5 (40) JCT + OB JOF + 20B S = fluorescence signal N = total noise T = average total signal (fluorescence + blank) 5 = average blank signal §F = average fluorescence signal 0T2 = variance of total signal (OF2 + 032) 032 = variance of blank signal OFZ = variance of fluorescence signal The detection limits reported in the following section were determined on the basis of a signal-to-noise ratio of two. 154 2. Results Working curves were obtained in the sequential dispersive mode using the synchronous integrator. The detection limits obtained are given in Table 9. The burner was positioned two inches below the entrance slit of the monochromator. Argon was employed as the sheath gas at the given flow rates. Note that the mercury detection limit f]! is improved by a factor of two when no sheath gas is used. Recall the flow rate profile for mercury in Figure 32 in which the fluorescent intensity decreased with increasing L] sheath gas flow rate. Table 9.--Detection Limits Obtained With Integrator Dis- persive System. Element Ar Flow Rate Detection Limit (l/min) (PPm) Cd 20 i I 0.02 Zn 20 0.5 Hg 20 10 Hg 0 5.0 Pb 0 20 For comparison purposes, working curves for cadmium, zinc and mercury were obtained using a lock-in amplifier. This is feasible with a sequential system because all of the lamps are operated at the same frequency. However, a lower duty cycle, e.g., 1/10, must be employed. The 155 monostable multivibrator of the logic control circuit that pulsed the lamps in an intermittent mode was used as the reference frequency. The dc output of the lock-in was monitOred under computer control. A series of 50 A/D conversions were made every 20 msec for a total analysis time of one second. The burner was placed in the same position. Argon and nitrogen sheath gases were used at flow rates of 20 l/min. Detection limits obtained with this system are given in Table 10. A slight decrease in detection limit was observed for all three elements when argon was employed as the sheath gas. These results agree with those obtained by West (174). Table 10.--Detection Limits Obtained With Lock-In Amplifier System. Nondispersive Dispersive (ppm) (ppm) Element Argon Argon Nitrogen (30 l/min) (20 Z/min) (20 i/min) Hg 13 20 30 Cd 0.03 0.04 0.06 Zn 0.5 1.0 2.0 The lock-in amplifier was also used for the sequential nondispersive mode. The major advantage of the nondispersive mode for multielement analysis is that the monochromator setting does not have to be changed for each be “1.1 » 156 element. Mitchell and Johansson (127) also used a non- dispersive system for simultaneous multielement analysis. However, their system incorporated a rotating filter wheel. Therefore, their method would have to be termed sequential, gg£_simultaneous. Our system with the lock-in amplifier is also a sequential system, however, it does ggE_employ a rotating filter wheel. The fluorescence radiation is allowed to fall directly on the photomultiplier transducer. The operation times of the lamps are now controlled automatically by the shift register. When the shift register is activated, the first lamp is operated in an intermittent mode for two seconds while the computer stores the data in the first dimensional array similar to the dispersive mode. When the sequencing circuit shifts to the second lamp, the computer immediately stores the data in a second dimensional array, etc. The total analysis time has, therefore, been reduced because of the elimination of the monochromator. Working curves were then obtained for Hg, Cd, and Zn in the nondispersive mode. Since the energy throughput for a nondispersive system is increased, the flame back- ground is also increased. Therefore the burner had to be lowered relative to the observation window. The detection limits obtained with the nondispersive system using argon sheathing are given in Table 10. A flow rate of 30 E/min of argon was used. Although the energy throughput for the 157 nondispersive system is greater than that for the dispersive system, the S/N enhancement or depression will depend on the particular flame employed. For example, Muscat (175) has reported that mercury has a 48-fold signal throughput enhancement when a nondispersive system replaces the dispersive system (this figure of course is dependent on ‘ the monochromator used for the comparison). Therefore the Fa increase in total noise throughput for the nondispersive system would have to be less than 48 for an increase in S/N to occur. For most flames, air-CZHZ, NZO-Hz, NZO-C2 2, EJ 02-H2, this will ggE_be the case. Therefore, the non- dispersive system can be disadvantageous, especially for elements with only one resonance line. For elements with a complex fluorescence spectrum the situation can possibly be improved. However, this improvement is highly dependent on the element and location of its fluorescence lines with respect to the spectral response of the detector. Since the detection limits for the three elements decreased slightly with the nondispersive system, this indicates that the increase in noise throughput for the separated air-H2 flame is not as great as the increase in energy throughput for the fluorescence signal. B. Multielement Atomic Fluorescence Spectrometry in the TIES:Division Multiplex Mode The data presented in the previous section were obtained in a sequential manner, i.e., one lamp was 158 operated 2225; another lamp. To perform a simultaneous or nearly simultaneous multielement analysis all of the lamps must be operated at the game time. However, there must still be a mechanism of distinguishing the fluorescence signals of the various elements if a single channel detection system is employed. This may be accomplished in one of three ways. First of all, the lamps could be operated at the same time, but at different frequencies. The respective fluorescent intensities could then be determined by Fourier transformation methods. This would transform the amplitude vs. time spectrum to an amplitude vs. frequency spectrum. Another possibility would be to operate all of the lamps at the same time and at the same frequency, but with the ON times of respective hollow cathode lamps out of phase. This shall be referred to here as the time-division multiplex mode. A final possibility would be to operate all of the lamps at different frequencies and use the lock- in amplifier to separate the frequencies. However, this would still be an inherently sequential method. Because of its attractiveness for rapid multielement analysis, the time-division multiplex mode was chosen. 1. Instrumentation To perform the nearly simultaneous analysis non- dispersively, the hollow cathode lamps were operated in the time-division multiplex mode. This was accomplished by means of a series of monostable multivibrators as is shown 159 in Figure 38 for three lamps. This method was made possible by use of the inherent long OFF time of the lamp during any one cycle. The outputs of the first, third, and fifth monostables control the 5 msec ON times of their respective lamps, while the second and fourth monostables introduce a 15 msec delay time between the pulses. The monostable outputs also Control the gating and hold for the integrator circuit and the triggering sequence for A/D conversions as is shown in Figure 39 for one cycle. When the monostable circuit is activated, A/D conversions are made during the consecutive hold times of a cycle and stored in dimensional arrays. The timing for A/D conversion is controlled by the real time clock of the lab 8/E com- puter. Under computer control the entire process is repeated for a designated number of cycles, usually twenty. The total analysis time for all the elements has therefore been reduced by l/N, where N is the number of elements. Even if 16 lamps were operated in this manner, the total analysis time would still be about the same because the delay time is variable. An example of the data obtained with this nearly simultaneous operation is shown in Figure 40. The lower trace is the fluorescent intensity obtained for a solution containing 500 ppm Hg, 5 ppm Cd, and 50 ppm Zn respectively for one cycle. The upper trace represents the corresponding output of the integrate and hold circuit. 160 .uouaumuucH can «mama ouofifio 3030a mo no.5.“ "8 Houucou 95990 O was .m J .coaumuomo Mung—Baa: wow Emu—93o Mason—do magnumocoz .mm mun—mam 82833.: n "86;: :55, #5.. r c. U c c m H < mam via mum «is _% k h h ._. ._.I|Ilo xUOLU O .|._|.O O 90 0 161 A c 3 c #50 TO REAL TIME c c— CLOCK INPUT . E D .u _-_E: -_-i__--__-_E: OUTPUT __..__.._..__.L: Figure 39. Triggering Sequence for A/D Conversions in Multiplex Mode. 162 'I'IU'I'V" "‘\____:"' Jallllh 'TT" ‘1 laaaaalaaaaaanaa UUVIIV'VI'TU'IU' mllllllnlnljllljllll 'I'UUVV'UIIVU‘UIIV‘Y‘V 1"111nn u I. an :m d. I VOLT / DIV lnllillillllllllll Io MSEC/ DIV Q" Figure 40. Oscilloscope Trace of Hg, Cd, and Zn Fluores- cence Signals Obtained in Multiplex Operation and Corresponding Output of Integrate and Hold Circuit. 2. Results Working curves were obtained for four elements with this time-division multiplex system in the nondispersive mode. A typical working curve for Cd with a sheath gas flow rate of 30 i/min is shown in Figure 41. A least square fit of the data gave a slope of 0.99. Detection limits for all four elements at various flow rates are summarized in Table ll. The optimum argon flow rate for all four elements was determined to be 30 2/min. At 20 l/min the S/N is lower because the noise from the flame background has increased. At 40 2/min the S/N is lower because the fluorescence signal has decreased due to the decrease in temperature. The resulting detection limits 163 oop SEQ . 20 o— _ l .uuoz vuxoamfiuasz aw vocwsuno o>usu mCMxHoz EswEumu Z._<._ O._. “H .¢MIoomIDm ..00 canon .cnmo sauna .uousumwucH can mmfimq mo mmfiwa zo Houusoo on uwsouwo wusmuoucH .mw musmfim 9 83mm SUI—m . ‘ . mosmo. 53mm 30um IEOH . .833 0'1 83mm 358 . .o|‘ Fonz. C>Obmw ZO mOkdm—OMHZ_ IIII FDQHDO 4| Dm<0 10.5...— 171 ., Bo: moEmoEi/z .CMIoomIam ..oo spasm .oumo roams .cofimum>soo a\¢ you uoumummuaH no mass mace Houucoo on unsouao momuumuaH .se shaman : 85am . mosmo mmoEm. oll a no H 2 83% 35.3 P) . 83mm . mo_>mo :52. PI rotam C mo... SEDO Ilia A , _ ‘l 920 102.. _ CC . 172 variable with the hardware control system. However, the ON time was fixed by the RC time constant of the monostable circuit. The length of the ON time is not critical when a flame is used for atomization because an equilibrium popu- lation of atoms is achieved. Therefore, the hardware control system is suitable for this purpose. However, when a non-flame system is employed, the population of atoms in the observation window varies with time. In this case, the ON time is critical, and hence its optimization is essential. Software control over the lamp ON time allows this optimi- zation to be performed with no changes in circuitry. B. Flame Atomic Fluorescence Studies For comparison purposes, the computer-controlled multielement atomic fluorescence spectrometer was first used with a flame atomizer. An Arr-H2 flame was employed with an argon sheath gas. To test the operation of the lamps, the computer-controlled system was used in the dispersive sequential mode. Detection limits were obtained for cadmium and mercury and were found to be identical with those obtained with the hardware-controlled system. Next the computerized system was used for the flame analysis of Mg, Fe, Co, and Ni. Flow rate profiles were determined, working curves were obtained at the optimum flow rates, and finally, interelement effects were studied. 173 1. Flow Rate Profiles The argon flow rate profile obtained for Mg is shown in Figure 45. The burner was positioned two inches below the entrance slit of the monochromator. The optimum flow rate for this case was 15 2/min. The flow rate profile Obtained for Fe is shown in Figure 46. The flow rate profiles Obtained for Co and Ni were similar to that of Fe. For these three cases, the optimum flow rate was determined to be 20 l/min. The fluorescence intensity I . increases as the flow rate initially increases because the sheath gas reduces the quenching effect of atmospheric nitrogen. However, as the flow rate increases further, the temperature decreases and hence the fluorescence intensity decreases. 2. Working Curves WOrking curves were Obtained for the four elements, and the resulting detection limits are summarized in Table 13. An example of the working curve obtained for Co is presented in Figure 47. The detection limits obtained for Co, Ni and Fe are not unreasonable considering that these are high boiling point elements, 2900 °C, 2732 °C, and 3000 °C respectively. However, the detection limit for Mg appears quite high considering the fact that its boiling point is only 1107 °C. The discrepancy may be explained by the fact that the increase in intensity of the Mg hollow 174 A 00 L': o :2) T 5 .. Q ' 0 g o < V >. L': 0) 2 El 10 - ’ o g I I- Z UJ O 00 UJ [E O a s -‘ U. 9 U) E E E. 337,.” n ; ,. .,,. ”I.” It. “Mum. rfll.‘m:‘;jg‘gm.‘rm .u. .rn.nvv.'.-_vvou-.mx.ara- in . .rxuaw-umw «tau-co.“ O 10 20 30 40 . z A I W " \ a; n ,P. I" i [ '- 5 3.! [,1 1 \ A-” Cr‘ ‘3‘ l {MC/'1 12‘ \* {xi 1‘ \v I L... ’ J. i 1 h{’ K V Figure 45. Magnesium Fluorescent Intensity as a Function of Argon Flow Rate in Dispersive Mode. 175 C.) O I M O l H~e Fe FLUORESCENT INTENSITY,(ARB. UNITS) O L fl I I I 0 I0 20 ' 310. 40 ARGON FLOW RATE,L/MIN.' Figure 46. Iron Fluorescent Intensity as a Function of Argon Flow Rate in DiSpersive Mode. 100- TO“ FLUORESCENT INTENSITY. (ARB. UNITS I 176 Figure 47. I ‘ I ' 'fi TO 100 1000 COBALT CONCENTRATION, PPM I Cobalt Working Curve Obtained With Computer- Controlled Dispersive System. 177 Table 13.--Detection Limits Obtained With Computer- Controlled Multielement Flame A.F.S. System. Peak Optimium Ar Detection Elements Current Flow Rate Limit ‘ (l/min) (ppm) Hg , 265 0 2 Cd 200 20 .02 Mg 200 15 1 Ni 275 20 10 Co 290 20 2 Fe 330 20 20 cathode lamp is only 12 when operated at a peak current of 220 mA compared to dc operation as reported by Cordos and Malmstadt (115). This lack of intensity increase is probably due to the ease of self-reversal in the case of Mg. 3. Interelement Effects The solutions that were used for the working curves were composed of mixtures of the four elements. To deter- mine if any interelement effeCts were present, the fluores- cent signals obtained for the mixture solutions were compared to fluorescence signals obtained for solutions containing only the single element. All solutions were made from the chloride salts. The data obtained are presented in Table 14. As can be seen only slight inter- element effects were obtained for these elements. 178 Table l4.--Fluorescence Signals Obtained With Multielement and Single Element Solutions. Average A.F. Signal, Arbitrary Units Element Concentration Mixture Single (Ppm) Ni 50 123 130 200 326 287 1000 662 657 Co 50 311 419 200 600 718 1000 1402 1317 Fe 50 400 386 200 719 600 1000 1728 1540 Mg 50 550 403 200 1271 1232 1000 2230 2557 179 ' C. Non-Flame Atomic Fluorescence Studies The application of non-flame atomizers in atomic absorption and atomic fluorescence spectrometry has been reviewed by Kirkbright (176), and more recently by Wine- fordner and Vickers (177). Non-flame atomizers require small sample sizes, e.g., l to 6 pi, respond to low con- centrations of analytes, e.g., ppm to ppb, and produce extremely low background emission. Since sheathing is used, fluorescence quenching can be made very small by appropriate choice of the sheath gas. Thus, non-flame atomizers seem to be ideally suited for atomic fluorescence spectrometry. The types of non-flame atomizers that have been recently employed include the graphite rod (178-181), the carbon-filament atom reservoir (182-183), and the graphite furnace (184). A simpler atomizer, the hot-wire loop atomizer made from platimum or tungsten, has been introduced by Winefordner and co-workers (185-186). Crouch and co- workers reported improvements with a platinum-rhodium (90%-10%) loop (187). The latter is the atomizer that was employed for this study. 1. System Design The computer-controlled non-flame atomic fluores- cence spectrometer used for this study has been described by Montaser and Crouch (188). An assembly language (Pal III) program controls the operation of the pulsed 180 hollow cathode sources, the two-step temperature of the atomizer, the synchronous integrator, and the data aquisition and treatment. A 4 ui sample is placed on the platinum-rhodium loop by syringe or by an automatic sampler (188). A low current is applied to the loop to vaporize the solvent. This is followed by a high current which atomizes the sample. The lamps are operated only during the atomization step. Since the atomic vapor of the elements is in the observation window for onlya very short time, e.g., 1-2 seconds, a sequential mode of operation could not possibly be used for multielement A.F.S. measure- ments. The data for the four elements are stored in their respective dimensional arrays, and the areas under the peaks, which are proportional to the fluorescence, are determined. Once the data are in core each dimensional array can be sequentially displayed on an oscilloscope for visual observation. For example, Figure 48 is an oscillos- copic trace of the zinc data from one array. This A.F.S. peak was detected in the presence of mercury, cadmium, and lead with all four lamps operated in the time-division multiplex mode described in the previous chapter. Under keyboard control the data for the other three elements in their respective arrays can be displayed. The points represent the A/D conversions that were made during the hold time. As can be seen the peak is Gaussian and the full width of the peak is about 1.5 seconds. During these 1.5 seconds, eighteen data points were taken. It can be 181 JI‘ It. .p Figure 48. Oscilloscope Trace of Integrator Output for One Dimensional Array for Zinc Obtained With Time Multiplex Nondispersive Non-Flame System. seen now why the ON time of the lamps is important. For example, the data shown in Figure 48 were obtained with an ON time of S msec, and a total period time of 80 msec. For 18 cycles, this corresponds to 1.44 seconds. If the ON time is reduced to 2 msec, period time of 32 msec, 45 data points would be included during the 1.5 seconds that the atoms are in the observation window. The optimum ON time and number of cycles were determined for each element. The combination of ON time and cycle number, of course, deter- mines the atomization time and hence the operation time of the four lamps. The actual time, relative to the start of atomi- zation, that the atoms for each element were in the observation window varied according to the boiling point of the element. Mercury was observed earliest in time, 182 followed by cadmium, zinc and lead. However, there was an overlap of the peaks which would rule out the possibility of sequential operation. 2. Working Curves Working curves were determined for these four elements, Cd, Hg, Zn and Pb in the dispersive mode. For these working curves the atomizer was positioned directly below the entrance slit of the monochromator and the radiation from the source was focussed directly above the atomizer. The detection limits obtained with the loop atomizer are shown in Table 15. For Zn, Cd and Hg the detection limits were decreased when compared to the flame system described in Chapter V. This decrease in detection limit is due to the fact that background noise from the non-flame atomizer is less than the background noise from the flame atomizer. The detection limit for Pb, on the other hand, has increased because of the lower temperature of the loop atomizer. When the graphite braid atomizer (189) was employed, a tremendous decrease in the Pb detection limit was Obtained because of the increased atomization efficiency of the braid. The reproducibility of the fluorescence signal was 8-10 per cent when a syringe was used to place the sample on the loop or braid. This may be increased to 2-3 per cent when an automatic sampler is employed (188) . 183 Table 15.--Detection Limits Obtained With Non-Flame Software Systems. Element Atomizer Dispersive Nondispersive (ppm) (ppm) Cd Pt loop 0.008 0.005 Hg Pt loop 0.5 0.5 Zn Pt loop 0.1 2.0 Pb Pt loop 50 Pb Braid .005 Detection limits were then Obtained for Cd, Hg and Zn with the non-flame nondispersive system. These detection limits are also shown in Table 15. The detection limits for Cd and Hg are Comparable to those obtained in the dispersive mode. However, the Zn detection limit is higher. This increase is due to the fact that the atomizer position had to be lowered relative to the observation window because of the radiation from the atomizer and because of scattering of the primary sources. As the loop is lowered, the temperature decreases drastically. Hence the atomic population decreases. The non-flame nondis- persive system may be used for multielement analysis. However, the detection limits obtained for certain elements will be inferior to those obtained for the single element dispersive case. VI I . CONCLUSIONS A. Summary The three flame spectrometric methods, A.E.S., A.A.S., and A.F.S., have developed primarily as single- element methods, which are optimized to determine a given element with high accuracy and precision. The various attempts to employ these methods for multielement analyses. have been reviewed. Multielement methods are particularly valuable for analyses where simultaneous information on a large number of elements is desired. However, a compromise in operating parameters is required to encompass the different behaviors of some of the elements. In Part II of this thesis, an automated nondis- persive atomic fluorescence spectrometer has been described for nearly simultaneous multielement analysis. The benefits to be derived from such a nondispersive technique include an increase in energy throughput, simultaneous measurement of multiple lines, simplicity, ruggedness and reduction of cost. The sensitivity obtainable with this atomic fluorescence technique, especially when using a non-flame atomizer, challenges, and sometimes surpasses those of 184 185 x-ray fluorescence, spark source mass spectrometry, and activation analysis with instrumentation which cost approximately one-tenth as much and operate at a fraction of their maintenance and running costs (56). Various sources have been employed for multielement atomic fluorescence. These include multielement electrode- less discharge lamps, a bank of single element electrode- less discharge lamps, a bank of pulsed hollow cathode lamps, a tunable dye laser, and a continuum source. Of these, the pulsed hollow cathodes seem to be the most promising because of their lower cost, availability, and intensity. For the nondispersive system, the choice of flame is critical. The argon separated air—H2 flame described in this thesis has low background emission, low concentra- tion of quenchers, good atomization efficiency, long residence time in the observation window, and low scattering of excitation radiation. Various detection systems have been employed for multielement atomic fluorescence. These include rotating filters, scanning monochromator, sequentially programmed monochromator, image-dissecting photomultiplier, photo- diodes and phototransistors, direct reading spectrometers, and vidicon detectors. However, all of these systems that have been employed either incorporate one detector which sequentially determines various elements, or many detectors 186 which simultaneously determine various elements. The non- dispersive, time-division multiplexing, atomic fluorescence system described in Chapter V is the first technique which employs 222 detector to determine various elements in a multiplexed mode. The system can also be used in the sequential mode I if desired. The principle advantage of the sequential mode .1. of operation is that it allows the operator to change parameters between determinations to assure that optimum Tum-r“ ' i. conditions for each element are attained. The multiplexing mode, on the other hand, has the advantage of a reduced analysis time even when many elements are being determined. However, a compromise has to be made as to the parameters chosen. This will be true for any multielement technique used. For the multiplexing system, a synchronous inte- grator was employed. This method has the advantage that it is ON only during the time that the fluorescence radiation is being emitted. For both the lock-in amplifier and direct integration systems, the nondispersive method resulted in detection limits comparable to the diSpersive method only for the separated air-H2 and Ar-H2 flames. For other flames, the nondispersive method is disadvantage- ous because the increase in energy throughput is accompanied by a greater increase in noise throughput. 187 The full advantage of the nondispersive system was achieved when a non-flame atomizer was employed. The same increase in energy throughput is attained. However, the background emission is much less with the non-flame atomizer. We feel that the software-controlled pulsed hollow cathode lamps for excitation, the non-flame method of atomization, coupled with a nondispersive detection !_1 system employing a synchronous integrator is an ideal method for simultaneous multielement atomic fluorescence analyses. B. Recommendations A number of attempts have been made by various investigators to apply multielement analyses to one of the three atomic spectrometric methods, atomic emission, atomic absorption, and atomic fluorescence. Each of these methods, however, has its own advantages and disadvantages for analysing specific elements. Flame emission (190, 191) is the most easily adapted method to multielement analysis because all elements are excited simultaneously. Also, no primary excitation sources are required. However, A.E.S. is not a sensitive technique for elements with resonance lines below 3500 A. This situation may be improved with the use of an induction-coupled plasma (192-194). Atomic absorption can be used for multielement analysis (195) but the optical arrangements required to produce multicomponent radiation beams is rather complicated. Atomic fluorescence 188 has been shown to be useful for multielement analysis. However this technique is not very sensitive for elements with resonance lines above 3500 A. Therefore, no one atomic spectrometric technique can be used fOr the analysis of a wide variety of elements. Since atomic absorption offers no advantages over atomic fluorescence, the best method for multielement analysis appears to be a hybrid instrument for atomic emission and atomic fluorescence. Such a method was used by Winefordner gt El (196) for the analysis of trace wear metals in jet engine oils. Morrison and Busch (197) have recently reviewed various methods used for multielement flame spectroscopy. They concluded that the important considerations in a multielement technique include speed of analysis, wide dynamic range and linearity, sensitivity, scope, sample type and size, simplicity of operation, and finally, cost. We feel that the multiplexed atomic fluorescence spectro- meter described in this thesis incorporates all of these criteria except for scope. Since a solar blind photo- multiplier tube is used, only elements with resonance lines below 3200 A may be analysed. This situation, however, may be improved with the use of a hybrid atomic fluorescence— atomic emission spectrometer for multielement analysis. LIST OF REFERENCES 10. ll. 12. J. . V. LIST OF REFERENCES Russell and A. Walsh, Spectrochim. Acta, 10, 883 (1959). Sullivan and A. Walsh, Spectrochim. Acta, 21, 727 (1965). Walsh, Spectrochim. Acta, 1, 108 (1955). O V. L'vov, "Atomic Absorption Spectrochemical Analysis," American Elsevier Publishing Company, New York, p. 34, 1970. L'vov, "Atomic Absorption Spectrochemical Analyais," American Elsevier Publishing Co. Inc., New York, p. 146, 1970. Jones and A. Walsh, Spectrochim. Acta, 16, 249 (1960). Russell, J. P. Shelton, and A. Walsh, Spectro- chim. Acta, 8, 317 (1957). Box and A. Walsh, Spectrochim. Acta, 16, 255 (1960). Gatehouse and A. Walsh, Spectrochim. Acta, 16, 602 (1960). I Robinson, "Atomic Absorption Spectroscopy," Marcel Dekker, Inc., New York, 1966. Elwell and J. A. F. Gidley, "Atomic-Absorption Spectrophotometry, Pergamon Press, Oxford, 1966. Ramirez-Munoz, "Atomic-Absorption Spectroscopy," Elsevier Publishing CO., Amsterdam, 1968. 189 13. 14. 15. l6. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 190 W. Slavin, "Atomic Absorption Spectroscopy," Inter- science Publishers, New York, 1968. I. Rubeska and B. Moldan, "Atomic Absorption Spectro- photometry," Iliffe Books Ltd., London, 1969. G. D. Christian, "Atomic Absorption Spectroscopy,"l Wiley-Interscience, New York, 1970. R. Reynolds, "Atomic Absorption Spectroscopy," Griffin, London, 1970. B. V. L'vov, "Atomic Absorption Spectrochemical Analysis," American Elsevier Publishing Co. Inc., New York, 1970. See Reference 2, 2.}. 721 (1965). Ibid., 22, 1843 (1966). Ibid., 23B, 131 (1967). P. A. Young, A. B. Timms, and J. V. Sullivan, Proc. Aust. Inst. Min. Metall., 226, 31 (1968). B. S. Rawling and J. V. Sullivan, Trans. Instn. Min. Metall.,_London, 16, 238 (1567). P. L. Boar and J. V. Sullivan, Fuel, London, 46, 47 (1967). Ibid., 230 (1967). J. A. Bowman, Anal. Chim. Acta, 31, 465 (1967). J. V. Sullivan and A. Walsh, Applied Optics, 1, 1276 (1968). J. V. Sullivan and A. Walsh, Application of Resonance Monochromators to the Simultaneous Deter- mination of Several Elements, Sixth Australian Spectroscopy Conference, Brisbane, 1967. A. Walsh, Pure and Applied Chemistry, 22, l (1968). Varian Techtron, Bulletin 604, "Model AR-200 Atomic Absorption Spectrophotometer Utilizing the Resonance Detector Principle," 1968. J. D. Winefordner, V. Svoboda, and L. J. Cline, CRC Critical Reviews in Analytical ChemistryL‘i, 233 (I970). 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 191 D. Winefordner, Appl. SpectrOSC-I $1. 109 (1963)- A. Fassel and V. G. Mossotti, Anal. Chem., 3_ 252 (1963). A. Fassel, V. G. Mossotti, W. E. Grossman and R. N. Kniseley, Spectrochim. Acta, 23, 347 (1966). W. McGee and J. D. Winefordner, Anal. Chim. Acta, 37, 429 (1967) De Galan, W. W. McGee and J. D. Winefordner, Anal. Chim. Acta, 31, 436 (1967). J. Vickers and J. D. Winefordner, "Analytical Emission Spectroscopy," part II, p. 348, edited by E. L. Grove, Marcell Dekker, Inc., New York, 1972. W. Frank, W. G. Schrenk, and C. E. Meloan, Anal. Chem., 39, 535 (1967). W. McGee and J. D. Winefordner, J. Quant. Spectry. Radiative Transfer, _J 261 (1967 7) G. Mossotti, F. N. Abercrombie, and J. A. Eakin, Appl. Spectrosc., 25, 331 (1971). GojoviE and A. AntiE-JovanoviE, Spectrochim. Acta, 273, 385 (1972). A. Bowman, J. V. Sullivan, and A. Walsh, Spectro- chim. Acta, 22, 205 (1966). M. Lowe, Spectrochim. Acta, 24B, 191 (1969). D. Lloyd and R. M. Lowe, Spectrochim. Acta, 27B, 23 (1972). A. Sebestyen, Spectrochim. Acta. 253, 261 (1970). D. Winefordner and T. J. Vickers, Anal. Chem., 36, 161 (1964). D. Winefordner, M. L. Parsons, J. M. Mansfield, and W. J. McCarthy, Spectrochim. Acta, 23B, 37 (1967). L. Parsons, W. J. McCarthy, and J. D. Winefordner, J. Chem. Ed., 44, 214 (1967). 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. B. C. J. J. C. 192 Winefordner, M. L. Parsons, J. M. Mansfield, and W. J. McCarthy, Anal. Chem., 29, 436 (1967). Th. Zeegers, R. Smith, and J. D. Winefordner, Anal. Chem., 40513 , 26A, (1968). Th. Zeegers and J. D. Winefordner, Spectrochim. Acta, 268, 161 (1971). Elser and J. D. Winefordner, Anal. Chem., 43(4), 24A (1971). Svoboda, R. F. Bowner, and J. D. Winefordner, Appl. Spectrosc., 36, 505 (1972). Th. J. Alkemade, Pure and Applied Chemistry, 33, V. S. 73 (1968). Jenkins, Spectrochim. Acta, 23B, 167 (1967). Hooymayers, Spectrochim. Acta, 23B, 567 (1960). West, Appl. Spectrosc. Reviews, 7(1), 79 (1973). C. Handbook of Chemistry and Physics, "The Chemical Rubber Company," 47th Edition, D-108 (1967). L'vov, "Atomic Absorption Spectrochemical Analysis," American Elsevier Publishing Company, New York, p. 42, 1970. Sullivan, C. S. I. R. 0., Clayton, Victoria, Australia 3186, personal communication, March, 1972. L'vov, op, cit., p. 50, 1970. Rann, Spectrochim. Acta, 23B, 827 (1967). Rubeska and V. Svoboda, Ana1.+Chim. Acta, 33, 253 (1965). Yasuda, Anal. Chem., 38, 592 (1966). V. F. L'vov, 22, cit., p. 48, 1970. Kirkbright, O. E. Troccoli, and S. Vetter, Spectrochim. Acta, 28B, 1 (1973). Ibid., 33 (1973). 67. 68. 69. 70. 71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. J. C. 193 Alger, G. F. Kirkbright, and O. E. Troccoli, Appl. S. A. Spectrosc., 31, 177 (1973). Gazhov and A. V. Zherebenko, Ehurnal Prikadnoi Spektroskopii, 12(3), 403 (1970). Fassel, J. O. Rasmuson, and T. G. Cowley, Spectrochim. Acta, 233, 579 (1967). Jaworowki and R. P. Weberling, Atomic Absorption M. M. W. Newsletter, 5, 125 (1966). Hall and C. Woodward, Spectrosc. Lett., 2, 113 (1969). '— Winefordner, W. W. McGee, J. M. Mansfield, M. L. Parsons, and K. E. Zacha, Anal. Chim. Acta, ‘36, 25 (1966). Wood, Phil. Mag., lQ' 513 (1905). Nichols and H. L. Howes, Phys. Rev., 23, 472 (1924). Badger, Z. Phys., 55, 56 (1929). Robinson, Anal. Chim. Acta, 23, 254 (1961). Th. J. Alkemade, in Prox. x Collog. Spectro. Internat., Spartan Books, Washington,—5.C., T—l 63, p. 143.. Staab and J. D. Winefordner, Anal. Chem., _§J 165 (1964). ' , 1367 (1964). Mansfield, C. Veillon, and J. D. Winefordner, Anal. Chem., 31, 1051 (1965). West, Analyst, 21, 69 (1966). Dagnall, P. Young, and T. S. West, Talanta, 12, 803 (1966). Dagnall, K. C. Thompson, and T. S. West, Anal. Chim. Acta, 26, 269 (1966). Ibid., Talanta, 14, 551 (1967). Ibid., Talanta, $2! 557 (1967). 86. 87. 88. 89. 90. 91. 92. 93. 94. 95. 96. 97. 98. 99. 100. 101. 102. 103. 104. 194 Ibid., Talanta, 14, 1151 (1967). Ibid., Talanta, 14, 1467 (1967). J. F. Alder and T. s. West, Anal. Chim. Acta, §_1_, 365 (1970). ' Omenetto and J. D. Winefordner, Appl. Spectrosc., 36, 555 (1972). . R. Jenkins, spectrochim. Acta, 253, 47 (1970). B. Willis, Spectrochim. Acta, 253, 487 (1970). R. Koirtyohann and E. E. Pickett, Spectrochim. Acta, 263, 349 (1971). S. Smyly, W. P. Townsend, P. J. Th. Zeegers, and J. D. Winefordner, Spectrochim. Acta, 263, 531 (1971). G. C. Human and A. Strasheim, Spectrochim. Acta, 273, 503 (1972). Stevens and T. S. WeSt, Spectrochim. Acta, 273, 515 (1972). 7' Reif, V. A. Fassel, R. N. Kniseley, Spectrochim. Acta, 283, 105 (1973). S. West and X. K. Williams, Anal. Chem., 42, 335 (1968). L. Larkins, R. M. Lowe, J. V. Sullivan, and A. Walsh, Spectrochim. Acta, 243, 187 (1969). J. Vickers and R. M. Vaught, Anal. Chem., 41, 1476 (1969). D. Warr, Talanta, 11, 543 (1970). J. Vickers, P. J. Sleirn, V. I. Muscat, and L. T. Farias, Anal. Chem., 44, 930 (1972). C. Elser and J. D. Winefordner, Appl. Spectrosc., 35, 345 (1971). L. Larkins, Spectrochim. Acta, 263, 477 (1971). W. Haagen-Smit and J. Ramirez-Munoz, Anal. Chim. Acta, 36, 469 (1966). 105. 106. 107. 108. 109. 110. 111. 112. 113. 114. 115. 116. 117. 118. 119. 120. 121. 122. 123. J. 195 B. Dawson, P. J. Ellis, and R. Milner, Spectrochim. Acta, 233, 695 (1968). Strasheim and H. G. C. Human, Spectrochim. Acta, 233, 265 (1968). Mavrodineanu and R. C. Hughes, Appl.ggptics, 7, 1281 (1968) . " B. Marshall and T. S. West, Anal Chim. Acta, 51, 179 (1970) . "" Fulton, K. C. Thompson, and T. S. West, Anal. Chim. Acta, 51, 373 (1970). S. Cresser and T. S. West, Anal. Chim. Acta, 51, 530 (1970). D. Norris and T. S. West, Anal. Chim. Acta, fig, 359 (1971). Ibid., §__9_, 474 (1972). H. V. Malmstadt and E. Cordos, Am. Lab., p. 35, August (1972). Ibid., Anal. Chem., 44, 2407 (1972). Ibid., 45 Ibid., 45, 4: 27 (1973). 425 (1973). Ibid., 4 2277 (1972). D. Norris and T. S. West, Anal. Chem., 45, 226 (1973). w. J. M. Boumans and F. J. de Boer, Spectrochim. Acta, 273, 391 (1972). . L. Corcoran, Jr., Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, paper no. 100, 1972. w. J. M. Boumans and G. Brouwer, Spectrochim. Acta, 27B, 247 (1972). H. Horlick and E. G. Codding. Anal. Chem., 45, 1490 (1973). . W. Golightly, R. N. Kniseley, and V. A. Fassel, Spectrochim. Acta, 253, 451 (1970). 124. 125. 126. 127. 128. 129. 130. 131. 132. 133. 134. 135. 136. 137. 138. 139. 140. 141. 196 M. Margoshes, Spectrochim. Acta, 253, 113 (1970). K. M. Aldous and D. G. Mitchell, Pittsburgh Con- ference on Analytical Chemistry and Applied Spectroscopy, paper no. 83 (1973). L. Papouchado, E. I. du Pont de Nemours, Experimental Station, Wilmington, Delaware, personal communication, 1973. A. G. Mitchell and A. Johansson, Spectrochim. Acta, 253, 175 (1970). Ibid., 263, 677 (1971). R. M. Dagnall, G. F. Kirkbright, T. S. West, and R. Wood, Anal. Chem., 43 , 1765 (1971). R. M. Dagnall, G. F. Kirkbright, T. S. West, and R. Wood, Analyst, 91, 245 (1972). M. Jones, G. F. Kirkbright, L. Ranson, and T. S. West, Anal. Chim. Acta, 63, 210 (1973). R. M. Lowe, Spectrochim. Acta, 263, 201 (1971). J. I. Dinnin and R. W. Helz, Anal. Chem., 22, 1489 (1967). J. I. Dinnin, Anal. Chem., 1491 (1967). P. M. Dagnall and T. S. West, Appli. Optics, 1, 1287 (1968). R. F. Browner, B. M. Patel, and J. D. Winefordner, Anal. Chem., 44, 1272 (1972). R. M. Dagnall, M. D. Silvester, and T. S. West, Spectrochim. Acta, 283, 51 (1973). N. Omenetto, N. N. Hatch, L. M. Fraser, and J. D. Winefordner, Spectrochim. Acta, 283, 65 (1973). E. H. Piepimeier, Spectrochim. Acta, 273, 431 (1972). Ibid., 445 (1972). D. C. Manning and P. Heneage, At. Abs. Newsletter, 7(4), 80 (1968). 142. 143. 144. 145. 146. 147. 148. 149. 150. 151. 152. 153. 154. 155. 156. 157. 158. 159. 160. 197 B. Dawson and P. J. Ellis, Spectrochim. Acta, 23A, 565 (1967). B. Willis, Rev. Pure and App1.Chem., 17, 111 (1967). A. Bfiger and W. Fink, Spectrochim. Acta, 1359 (1970). V. Gelder, Spectrochim. Acta, 253, 669 (1970). van Trigt, Phys. Rev., 181(1), 97 (1969). F. Bruce and P. Hannaford, Spectrochim. Acta, 263, 207 (1971). F. Kielkopf, Spectrochim. Acta, 263, 371 (1971). E. Osten and E. H. Piepmeier, Appl. 115 (1973). Spectrosc., 2 B. Barnett and H. L. Kahn, Anal. Chem., 44, 935 (1972). O. Weide and M. L. Parsons, Anal. Lett., 5(6), 363 (1972). V. L'vov, "Atomic Absorption Spectrochemical Analysis," American Elsevin Publishing Co. Inc., New York, p. 46 (1970). N. Crosswhite, J. Opt. Soc. 270 (1955). Am., 45, V. L'vov, op, cit., p. 62 (1970). S. Gough, P. Hannaford and A. Walsh, Spectrochlm. Acta, 283, 197 (1973). J. Muscat, "Studies in Nondispersive Atomic Fluorescence Spectrometry," Ph.D. Thesis, Florida State Univ., 1972. J. M. Mansfield, M. L. Parsons, and J. D. 204 (1966). Veillon, Winefordner, Anal. Chem., 28, Smith, C. M. Stafford, Anal. Chim. Acta, 43, and J. D. Winefordner, 523 (1968). P. Bratzel and J. D. Winefordner, Anal. Lett., 1, 43 (1967). P. Bratzel, R. M. Dagnall, and J. D. Winefordner, Anal. Chem., 41, 1527 (1969). 161. 162. 163. 164. 165. 166. 167. 168. 169. 170. 171. 172. 173. 174. 175. 176. 177. 178. 179. 198 Ibid., 713 (1969). G. F. Kirkbright and T. S. West, Appl. Optics, 1, G. F. 1305 (1968). Kirkbright, A. Semb, and T. S. West, Talanta, 11, 1011 (1967). Hingle, G. F. Kirkbright, and T. S. West, Talanta, 12, 199 (1968). Kirkbright, A. Semb, and T. S. West, Talanta, ‘15, 441 (1968). Hobbs, G. F. Kirkbright, M. Sargent, and T. S. West, Talanta, 12, 987 (1968). Hingle, G. F. Kirkbright, and T. S. West, Analxst, 2;, 522 (1968). Dagnall, G. F. Kirkbright, T. S. West, and R. Wood, Anal. Chem., 42, 1029 (1970). Aldous, R. F. Browner, R. M. Dagnall, and T. S. West, Anal. Chem., g_2_, 939 (1970). Larkins and J. B. Willis, Spectrochim. Acta, 26B, 491 (1971). Browner and D. C. Manning, Anal. Chem., 22., 843 (1972). Slevin, V. I. Muscat, and T. J. Vickers, Appl. Spectrosc., 26, 296 (1972). Kirkbright and S. Vetter, Spectrochim. Acta, 26B, 505 (1971). West, Pure Appl. Chem., 23, 99 (1970). Muscat, Ph.D. Thesis, p. 115 (1972). Kirkbright, Analyst, 26, 609 (1971). Winefordner and T. J. Vickers, Anal. Chem., 44, 150R (1972). Reeves, C. J. Molnar, and J. D. Winefordner, Anal. Chem., 44 , 1915 (1972). Reeves, B. M. Patel, C. J. Molnar, and J. D. Winefordner, Anal. Chem., 246 (1973). 180. 181. 182. 183. 184. 185. 186. 187. 188. 189. 190. 191. 192. 193. 194. 195. 196. 199 J. D. Winefordner, 32 (1973). 1., Anal. Chem., 45, 203 J. D. Winefordner, gg'gl., Appl. Spectrosc., 31, 171 (1973). R. G. Anderson, I. S. Maines, and T. S. West, Anal. Chim. Acta, 51, 355 (1970). J. Aggett and T. S. West, Anal. Chim. Acta, 21, 15 (1971). D. Clark, R. N. Dagnal, and T. S. West, Anal. Chim. Acta, 62, 11 (1973). M. P. Bratzel, R. M. Dagnal, and J. D. Winefordner, Anal. Chim. Acta, 48, 197 (1969). Ibid., Appl. Spectrosc., 24, 518 (1970). S. R. Goode, Akbar Montaser, and S. R. Crouch, to be published in Appl. Spectrosc. Akbar Montaser and S. R. Crouch, Pittsburgh Conference on Analytical Chemistry and Applied Spec- troscopy,.paper no. 119 (1973). Akbar Montaser and S. R. Crouch, in preparation for Anal. Chem. V. A. Fassel and D. W. Golightly, Anal. Chem., 19.. 466 (1967). E. E. Pickett and S. R. Koirtyohann, Anal. Chem., ‘41 (14), 28A (1969). R. H. Wendt and V. A. Fassel, Anal. Chem., 920 (1965). 37’ C. Veillon and M. Margoshes, Spectrochim. Acta, 23B, 503 (1968). G. W. Dickinson and V. A. Fassel, Anal. Chem., 1021 (1969). 41, L. R. P. Butler and A. Strasheim, Spectrochim Acta, 31, 1207 (1965). R. L. Miller, L. M. Frazer, and J. D. Winefordner, Appl. Spectrosc., 22, 477 (1971). 197. 198. 199. 200. 200 W. Busch and G. H. Morrison, Anal. Chem., 45 (8), 722A (1973). Zynger and S. R. Crouch, Appl. Spectrosc., 36, 631 (1972). F. Palermo and S. R. Crouch, Anal. Chem., 45, August Issue (1973). F. Palermo, Akbar Montaser, and S. R. Crouch, in preparation for Anal. Chem. VI TA VITA Eugene F. Palermo was born on September 2, 1947 in Morgantown, West Virginia. He lived in Pennsylvania for four years, in Italy for two years, in Florida for six years, in Minnesota for ten years, and in Michigan for four years. He received his high school diploma from St. Thomas Military Academy in St. Paul, Minnesota in June, 1965, where he received the Bausch and Lomb Science award. He entered the College of St. Thomas in September, 1965, and received his B.A. degree in Chemistry in June, 1969. In September, 1969, he entered Michigan State University, where he studied Analytical Chemistry under the direction of Dr. Stanley R. Crouch. He received the Ph.D. degree in August, 1973. He is a member of the American Chemical Society and the Society for Applied Spectroscopy. He will commence employment with E. I. du Pont de Nemours & Company, Experimental Station, Plastics Department, Wilmington, Delaware in September, 1973, under the direction of Mr. John Mitchell. 201