.‘3. ‘ - . vI,: 3' "3 3W333fifi '3,“ "'. 3313 333333“33 3. 33 I3M3 1 ~— ' 1 -n ' a- (‘33 w 33333333 ‘3 ,3,,333,3 ‘3 3,, , 313,3 - 33333333333333 .33 Nv' 33‘j ‘3' 3 3333333333.:-3333 ‘3 “3333333333 3.3333333 33, .33 33 3,._.,,-,33,,,3,3.,.33,.133,,,,33,133,331 3333333333. ~, :33 3 «33333 33‘3“"3333‘.33'- ' .'~3 '. . 33(,3333 3 3333' .3 333,3 3333333 33333 t333333333 3 3333333333333333333333333333 33‘” 3333333333 ”3 ,, 3I3mH33 33” .33 ' 1333“ 343h33333r333 3 333hdm3333£ . . W3fl3r3mm33 3 33 3‘HH ,,3 33w3 w, Whh£fi .Ixfl‘ .' 3" “"33““"‘ 3.33133 '3 333:3 3' 3 333 3333 3 . 3331,...31333333333 3.I ,3 ' I33 33‘ 333 3'. 33 3| 333 ‘3 '33 3- 3 3 333333 3 $9 3MW 3 ‘ 33 L 333 Q" 3 3 3 3 3333333. .. '3'33‘ 3., “3 . 3 33 333 333 .3, 33ufl3333332 3 3.” 333 N 33 33”433figgw3333um I .- 3‘33, 3 :..' v ’9‘". a... -1M oL-4s ’ 4.- ..-..~ ma "0". , h.»— - [sci . W. Wmm 2..." MM...“ ”mu—W_" Ming“... 0 «Cw-fin.” . firm”..- ~0— v . . ,- Li .A - . .n ‘ 3 3 35,3; '0 I3: I I 3,133.3:3,’l§'.f’%: :33; 33.3 1 . .~ 33‘ 1 ~33 ..33331,..- .. -. 2,3,3": 33.11 i. ,, ‘ , g 1 t _3- I . - : "3.3 '. ’ 3 ‘ 3 ‘3-3‘ .“3 93333333333333‘ .333 $333333 333333333339 3?" "~' ‘ '- ‘3 * 3333 3-3 3 35333$3§qfi3-2 .‘3513333332.3 i3? ‘8 3:33”:- 3 3 ,E 2?, J ' 3‘: 3' ,3 :33 g' g 3} .33 53‘ I L. 25' | . ,. ;.3333=33333;333.3 3.3.33:- .3...:. 3333333333333 3.3 333 “'3‘ - W333“ 333“ ‘J‘m3flnafib‘ '3'”Wu¢fipHQm ‘ 3 WE“? 3333 3333‘ 3333;, :.3- 1333,), , 3 33'; #333333 «,-1 33333333333. 3‘333313‘ ,3 ..... ,3 333%3,r3 3,fl 3333 3W3: 333% 33‘, 3m3£333 33 QH%IH33% 33,,IN a3fi;§ .I II 3 33313333333'-.:_,.. £33333 ”33313333333333 33333 3333333333; 333 233,335,193 .33 333,3, 3333 33 3 2,33,33333“ ', 3333-3 3-3 3,1333 3‘ 3,333. 3 33* 3 33-333" ‘3 1333 33333133333. 33 3 , 33 ,33 333333 333333 33 3333‘; ,33., 33‘ ,33‘333‘ 333 3 33333333333333] , .33331333‘33'33333‘73'333 333 33333333333 333-3 33 ‘ 3 ‘ 33 .3- 333- 333.333“- ‘3 ‘ 3‘33.333333333333 ‘3 -3- 3." .3'113 .L" -.\3 , ' 3333331333 3- 33333333 ‘3‘ ‘3 3:33 3333‘" ‘3‘-«3'231‘333333333313373-333. 333 3333333 3333333- 33333333“ . ,fié 3H3 3 3 "w r. 33 . 3 mi3 ’ “Wfifiayfijhfifikw1 h”" 33-33'” 393 3' 3 -;33'333333333“3 313‘ ‘ ‘3 33333 33 333333 :33. "'4-:«.3332.33".3-‘33‘333-32333‘ 331.33 3‘J333de3-3 ! W3J “Mmfim33333m,”3,3hhw ”33”?w&fi$flflflfifldh¢3 33 33d? 3 31.3333 3333 33. 3333-.33-‘3 M33“ 333-33 «'3‘ .3333 ‘33 " ‘3 33333393333333333 . 33 _; 333 “ . _-_,' "‘3 '1 133'13331‘3333333331 ‘33.;3; 33:2..‘3333333333‘3 2.,3 ,_ -‘;7¥3'3?j;3,,3,,3;3333 ,3 3133 3331 333 ,I ,3333, 3:233! 333‘” , , ,.,_ 3,,333‘ 333133 33333 3%, ,3. II ’;31'311- 33 ”yfiflTWW 3mmflw3wu H3 wafi fidhfl% 3 3 M33 £R3HH3HP . 33333333333‘33333 3,3I II 3 3x33 33333 33 3): 33‘ I 33333 333” ”31.33333 33’“ 31.33» 33%3g35331' @1313, 33311.33? .333‘ "3‘,“ 3.3931337? w “MW 3mhu3 3 3 ‘ N“w33%mm%3‘wp 333 ,33, ,3, .,3333,333 “=3. 3 33.3 .33 ,3 ‘3, 3; 333 33333 3.3 3333'3- ,,, 3333-3333333133; g, . ,3 313333 3.3), 33 33-733 3331333” 33 3333' 3333333 3333‘ 33333. 53,3 33:3,. 333 3a I 3 3333333 ,33? ”333‘?“ . 33.33%33P1g3§33° ”333%.; 33:33 :i ‘3‘3 ‘33-». 33 33 .333 33333-33333 5.33. 333334333333... 333.323 :333.3.. 3..- 3.3333333 “""333 33 :3333 31333333333£3§33$I£I 3 LIBRARY Michigan State University This is to certify that the dissertation entitled A VERSATILE EMISSION SPECTROMETER: MODIFICATIONS OF A COMMERCIAL FLUOROMETER THAT PERMIT COMPUTERIZED DATA ACQUISITION INCLUDING PHOSPHORESCENCE LIFETIME STUDIES presented by Nelson R. Herron has been accepted towards fulfillment of the requirements for Ph. D. degree in Chemistry (Wet/W Major professor Date May 15; L987 MS U is an Affirmative Action/Equal Opportunity Institution 0-12771 MSU LIBRARIES .—_—. RETURNING MATERIALS: Place in book drop to remove this checkout from your record. FINES will be charged if book is returned after the date stamped below. A VERSATILE EMISSION SPECTROMETER: MODIFICATIONS OF A COMMERCIAL FLUOROMETER THAT PERMIT COMPUTERIZED DATA ACQUISITION INCLUDING PHOSPHORESCENCE LIFETIME STUDIES BY Nelson Robert Herron A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Chemistry 1987 (/‘( j] Hz'xa 7+ ABSTRACT A VERSATILE EMISSION SPECTROMETER: MODIFICATIONS OF A COMMERCIAL FLUOROMETER THAT PERMIT COMPUTERIZED DATA ACQUISITION INCLUDING PHOSPHORESCENCE LIFETIME STUDIES BY Nelson Robert Herron This dissertation presents a review of emission spectroscopy with special emphasis on phosphorimetry. This document presents extensive modifications to a commercial spectrofluorometer which enable the instrument to perform a variety of emission experiments under computer control, including time resolved phosphorescence. A broad range of applications, including several previously unreported studies, are demonstrated. The first capability that is discussed is the ability to perform the original fluorometric experiments under computer control. This necessitated design and construction of a data acquisition board which is inserted into the instrument. This board contains an analog-to- digital converter, a sample-and-hold amplifier, and a multiplexer. These circuits latch the various control and data signals from the instrument and convert them to a digital format amenable to the computer. The board also contains a set of optical isolators that minimize multiple paths between the computer and instrument grounds to minimize ground loops. Data collected in this mode include fluorescence spectra and room temperature phosphorescence spectra. There are two major types of experimental capabilities which have been added to the computer controlled instrument. The first is the ability to collect phosphorescence spectra and quantitate phosphorescent samples while using the rotating can phosphoroscope. The second is the ability to collect phosphorescence decay data over a wide range of time domains using a pulsed light source. These features required that the original amplifier of the instrument be replaced by one capable of faster transient response. The stringent timing tasks of both of these types of experiments are performed by a versatile counter/timer that is included in the microcomputer. Data collection in the dc limit and the photon shot noise limit are demonstrated. The limit of detection of an analyte is shown to be significantly improved over that of the original analog instrument, and emission decay data are presented for experimental lifetimes ranging from 400 us to 4.3 s. The FORTH operating environment played a crucial role in developing the instrument. The advantages of the poly-FORTH system that are salient to the operation of the instrument are discussed. Certain major shortcomings of that environment are discussed in the light of another FORTH option. A board design is discussed which allows the implementation of the data system in an IBM PC compatible computer using this second FORTH option. ACKNOWLEDGEMENTS It's been a long road from my first day at MSU, meeting Frank Curran. There have been many others in the Crouch and Zabik research groups who contributed significantly to getting this far while, I think, retaining a modicum of sanity. These certainly include Professors Crouch and Zabik. It also includes the larger group of faculty who have taught me in their classes. Another unsung group who were very important were the support personnel: Marty, Ron, Tom, Scott, Keki, Manfred, Scott, and Andy who helped with designs and fixed what I broke. A graduate career includes a multitude of graduate students whom we see almost daily for several years. With luck, we will see many of these people for the rest of our lives. Thanks Jim, Jimmy, Keith, Clay, April, Tony, Pat and Margurite, Rytis, John P., Swiat, Kowaguchi, Hai- Dong, Koffi, NgaNga, Nobuo, Paul, Holly, Mary Ann, Sue, Jane, Joe deF., Anne, Ines, Gene, Lynne and John C., Hugh, Beak and Dome, Bruce, C. Mark, Chris W., Bob 8., Karlis, John, Patty, and Jake 8., Jay, John and Kate C., and everyone else who tipped a mug with me. There are a number of old professors at Central Michigan University and Alpena Community College where I took my first steps who were very helpful in getting me this far. I would like to thank Professor K. R. Lindfors at CMU and Dr. R. Moreau at ACC. I would like to acknowledge three people, especially. One is Bill Munslow who beat me out of school again; he kept me aimed ahead. iv Another is Matt Zabik; Z. convinced me finally that one can be serious without being dogmatic. The third is Deirdre O'Leary who helped and continues to help me stay moderately coherent through the most difficult but rewarding time of my life. Thanks one and all. TABLE OF CONTENTS List of Tables ...................................................... ix List of Figures ..................................................... x I. Introduction ..................................................... 1 References ....................................................... 8 II. The Absorption of Light and the Generation of Fluorescence ...... 10 A. The Absorption of Light .................................... 10 B. Fluorescence ............................................... 16 Fluorescence Kinetics and Related Processes ............. 21 Methods of Determining Lifetimes ........................ 24 References ...................................................... 27 III. Phosphorescence ................................................ 31 A. Spin Orbit Coupling ........................................ 31 B. Kinetics ................................................... 38 Quenching and the External Heavy Atom Effect ............ 41 The External Heavy Atom Effect .......................... 43 References ..................................................... 49 IV. Room Temperature Phosphorescence: Solid Surface Phosphorescence. ............................................... S3 A. Basic Mechanism ............................................ 53 B. External Heavy Atom Effect ................................. 58 C. Incidental Methods ......................................... 61 D. Applications ............................................... 62 E. Instrumentation ............................................ 64 References ........ .. ............................................ 67 V. Room Temperature Phosphorescence: Solution Phosphorescence ....... 72 A. Introduction ....... . ....................................... 72 B. Micelle-Stabilized RTP ..................................... 73 C. Cyclodextrin Enhanced RTP .................................. 82 D. Other Methods .............................................. 89 References.. ..................................................... 91 VI. Instrumentation. ................................................ 95 A. Introduction ............................................... 95 B. Original Instrument ........................................ 96 C. The Twin Bus Microcomputer ................................. 100 D. The Amplifier...........................; .................. 104 E. The Data Acquisition Board ................................. 110 Optical Isolation. ...................................... 113 F. Timing the Phosphoroscope .................................. 116 The LSI Timer/Counter ................................... 120 G. The Flashlamp Light Source for Lifetime Determinations ..... 122 Sensing the Flashlamp ................................... 123 H. Single Photon Detection and Lifetime Determinations ........ 126 I. The IBM PC Compatible FORTH Port ........................... 129 References.. ....... . ...... . ............... . ..................... 133 VII. The FORTH Operating Environment and Other Software ............. 135 A. Introduction ............................................... 135 B. Programming the AMD 9513: "ELAPSE" ......................... 136 ELAPSE with the Rotating Can.... ........................ 138 Elapse in Lifetime Determinations ....................... 138 The Efficiency of Photon Detection ...................... 140 C. The FORTH Assembler ........................................ 140 D. Control Commands ........................................... 141 E. Scanning and Quantitating Words ............................ 143 F. Lifetime Data Routines ..................................... 144 G. Data Reduction Software Used on the PDP-ll ................. 145 References ..................................................... 147 VIII. Results and Interpretations ................................... 148 A. Introduction ............................................... 148 B. Fluorometry Mode Data Acquisition .......................... 150 C. The Rotating Can Phosphorimeter ............................ 158 Quantitative Results .................................... 163 D. Long-lived Phosphors ....................................... 166 E. Rapid Phosphorescence ...................................... 176 F. Tb3+ Emission in Aqueous Solution at Room Temperature ...... 182 References .................................................... 191 IX. Conclusions and Future Prospects ................................ 192 A. The Amplifier .............................................. 192 B. The Phosphoroscope ......................................... 195 C. FORTH Systems .............................................. 197 D. Data Reduction .......................................... p... 199 References ...................................................... 201 Appendix A. A Description of the Data Acquisition Board ............. 202 A. Introduction ............................................... 202 B. The Path Description ....................................... 203 Appendix B. A Table of Chip Select Used in the Twin Bus Microcomputer ........................................... 208 viii Appendix c. The Design of the IBM PC Compatible FORTH-Port.......... 209 A. Introduction ............................................... 209 B. The Chip Select Decoder ....................... L ............ 209 C. The I/O Port ............................................... 212 Appendix D. FORTH Source Code ....................................... 215 LIST OF TABLES Table Title Page 1. Conversion times for specimens of the AD 574 A/D converter... 112 2. Comparison of the linearity of biphenyl detection using digital data collection (D.D.C.) and analog detection (A.D.). Relative standard deviations (RSD) are given .......................... 164 3. The effect of photon-counting parameters on calculated lifetime of 2,7-dichlorodioxin (20 ug/ml). Int. is the interval between data points (ms), Pts./Int. is the number of points collected in each interval for each flash of the source, SSR is the sum of . the squares of the residuals of the data about the fit, and A0 is the calculated initial intensity of the phosphorescence. S.D. is the standard deviation ........... . .......... . ....... 169 4. Tabulation of relative error (RE) for Table 3. ............... 171 5. The effect of increasing the number of experiments summed on the calculated lifetime of biphenyl.. .............. 173 6. Rate data for the analog sampling mode ....................... 184 7. Rate data for the single-photon detection mode ............... 186 8. The effect of the fitting window on the calculated values of REM and t for 1 mM Tb3+ ................ . ........... 188 9. A list of the chips associated with Figures 40 and 41 ........ 203 10. A list of the CS signals used in the computer controlled phosphorimeter. The FORTH word generating the CS is also given. A (*) designates a microprocessor support function.... 208 LIST OF FIGURES Figgre Title Page 1. 10. ll. 12. A potential energy diagram for absorption showing a shift from the 0-0 band maximum due to a change in the equilibrium inter-nuclear distance in the excited state.... 15 A simplified Jablonskii diagram for fluorescence. A is absorption and F is fluorescence ........................... 18 A potential energy diagram for fluorescence showing the mirror symmetry between absorption and fluorescence ........ 20 A Jablonskii diagram for phosphorescence and competing processes. A is absorption, F is fluorescence, and P is phosphorescence ............................................ 34 A potential energy diagram for phosphorescence showing the breakdown of the absorption-emission symmetry due to changes in the coupling between the T1 state and the so state ...... 36 Simplified conceptual representations of a) a micelle and b) an inverted micelle. The wavy lines represent the organic moiety in a surfactant molecule and the circles represent the ionic portion of the molecule .............................. 75 A simplified structural representation of a cyclodextrin molecule emphasizing the central cavity .................... 83 The rotating can phosphoroscope. There are two slits in the inner sleeve, set 180° apart. There is a third slit in the cell holder, set at 1800 to the excitation slit ............ 99 The phosphorescence spectrum of 1-bromonaphthalene. The upper spectrum was acquired with the PDP-12 based data system, and the lower one was taken in analog mode ......... 101 Spectra of 1-chlorodioxin demonstrating the lack of relia- bility in the PDP-12 based data system (Spectra a - c). A spectrum taken with theanalog data system is included for comparison (d). ....................................... ..... 102 Schematic diagram of the C/V circuit ....................... 106 Schematic diagram of the amplifier circuitry ............... 108 l3. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. Emission decay curve for lmM Tb3+ in 020 collected with an ac-coupled amplifier in the data collection system. The dip at 1 ms is due to signal charging of the coupling capacitor .................................................. 109 A pictorial diagram of the A/D portion of the data acquisition board .......................................... 114 The amplifier output a) before optical isolation and b) after optical isolation .................................... 115 A simplified block diagram of the final version of the data acquisition board .......................................... 117 Schematic of the circuit for sensing the rotating can phosphoroscope ............................................. 119 Schematic of the circuit for sensing the flashlamp ......... 125 Block diagram of the computerized phosphorimeter with extended data system ....................................... 128 Block diagram for the IBM PC compatible FORTH-port ......... 132 Fluorescence spectra of quinine sulfate collected with the digital data system showing the effect of spectral averaging (a - d). A spectrum taken in the analog mode is included for comparison (e) ............................................. 151 MS-RTP spectra of 1-bromonaphthalene taken with the digital data system (a - d). Details of these spectra, including slit widths, are discussed in the text. A spectrum taken in the analog mode is included for comparison (e) ................. 153 MS-RTP spectrum of 4-bromobiphenyl showing overlapping fluorescence band .......................................... 157 Phosphorescence spectrum of quinine sulfate at 77 K taken with the digital data system (a). A spectrum collected in the analog mode is included for comparison (b) ................. 159 Phosphorescence spectrum of 2,7-dichlorodioxin at 77 K taken with the digital data system (a). A spectrum taken in analog mode is included for comparison (b) ........................ 161 Phosphorescence spectrum of biphenyl at 77 K taken with the digital data system (a). A spectrum taken in the analog mode is included for comparison (b) ............................. 162 Plots of signal intensity vs biphenyl concentration for the digital data system (a) and the analog data system (b) ..... 165 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. Phosphorescence decay curves for 2,7-dichlorodioxin at 77 K showing the qualitative effect of decay curve summing going from 10 times (a) to 100 times (b). A solvent decay curve is shown (c) .................................................. 168 Phosphorescence decay curves for 2,7-dichlorodioxin at 77 K showing that the data is approximately randomly distributed about the fitted line for sums of 10 experiments (a), 50 experiments (b), and 100 experiments (c) ................... 170 Plot of RE vs 1 / (square root of the number of repetitions) for the phosphorescence decay of 2,7-dichlorodioxin ........ 172 Phosphorescence decay curve of 2,7-dichlorodioxin at 77 K taken as the average of 512 experiments collecting 500 data points at 53.5 us intervals. Unambiguous curve fitting is impossible .............................................. 174 Phosphorescence decay spectrum of biphenyl at 77 K from a single signal voltage sampling experiment showing photon shot noise in recorded signal ................................... 175 Phosphorescence decay curves of biphenyl at 77 K using the single photon-detection mode showing the decrease in scatter going from 1 experiment (a) to the sum of 100 experiments (b) ............................................ 177 Plot of RE vs 1 / (square root of the number of repetitions) for the phosphorescence decay of biphenyl .................. 178 Phosphorescence decay data for 4-bromobiphenyl at 77 K. Data linearized with the natural log using exponential offset, C, of 190 (a) and 200 (b) show little difference. Simplex fit is shown for comparison (c). Lifetimes from these data are discussed in text .......................................... 180 Phosphorescence decay data for 4,4'-dibromobipheny1 with a con- taminant showing single exponential simplex fit (a), biexponen tialsimplex fit (b), and single exponential fit to data excluding the first 3.5 ms (c) ............................. 181 Room temperature emission decay curve for 1.0 mM Tb3+ in 1.0 mM DETPAA collected at 545 nm (+/- 5 nm). Note the effects of amplifier saturation at time t < 2,000 us (sigmoidal curve shape) ............................................. . ....... 185 Plot of RE vs 1 / (square root of the number of repetitions) for Tb3I/DETPAA ............................................ 187 xiii 39. 40. 41. 42. 43. Room temperature emission decay data for 1.0 mM Tb3+ in water. Plots show the effect of the simplex fitting window on a single data set: a) window: 0.1 - 5.0 ms, t = 0.63 ms, b) window: 0.5 - 5.0 ms, t = 0.53 ms, c) window: 0.75 - 5.0 ms, t = 0.48 ms, d) window: 1.0 - 5.0 ms, t = 0.40 ms. Note the systematic error of fit in curves 3 and b at about 1.5 ms and poor fit to data in d at time t < 1.0 ms... ......................... 189 A pictorial representation of the control signal portion of the data acquisition board ................ . ................ 204 A pictorial representation of the data signal portion of the data acquisition board ..................................... 206 A pictorial representation for the chip select decoder (CS) portion of the IBM PC compatible FORTH-port ......... . ...... 211 A pictorial representation for the I/O port portion of the IBM PC compatible FORTH-port, including the AMD 9513 LSI Counter/Timer.. ............................................ 214 CHAPTER I INTRODUCTION Introductory Remarks Luminescence spectroscopy has gained an ever increasing sphere of application in modern analytical methodology. Luminescence methods allow greater selectivity for the species determined than do the absorption methods that pre-date them. They have lower limits of detection than the absorbtion methods due to the geometrical, and/or time relationships between the excitation beam and the analytical beam, and they have a larger linear dynamic range, because the signals from an assemblage of analyte molecules are additive. There are three major types of luminescence that are of current interest in analytical chemistry. These are fluorescence, phosphorescence, and chemiluminescence. Chemiluminescence does not depend on irradiating the analytical sample and will not be considered here. The thrust of this document is in the area of phosphorescence, but the usual treatment is to consider it as an extension of the fluorescence phenomenon. Additionally, the instrument described is designed to function as a spectrofluorometer with a minimum of conversion, so an introductory treatment of this topic will be given. This introduction provides an outline of the rest of this document, and it gives an overview of fluorescence and phosphorescence and their relationship to the research. Chapters II and III presents a more detailed analysis of fluorescence and phosphorescence, respectively. This includes discussions of the kinetics involved in these phenomena. Chapters IV and V are a literature review of modern methods of generating phosphorescence at room temperature along with related fluorescence information. Chapter VI concerns itself with the "nuts and bolts" of designing a general purpose computer-controlled phosphorimeter, and chapter VII presents some of the computer programming considerations used to collect both spectral and time domain data. Chapter VIII gives some representative results generated by this system, including information about the hydration sphere of chelated Tb3+ ion. The extended literature review format was chosen for two reasons. The first was that many of the design considerations used in developing the phosphorimeter accommodated the high speed data acquisition restrictions required by the more modern methods. The second reason was that a large body of literature was assembled in the course of this research, and a vehicle was needed to collate this information for future reference. It is hoped that this presentation will be as useful to the reader as it is to the writer. Fluorescence was the first luminescence method to achieve broad acceptance in the area of analytical chemistry. It emerged earliest because the experimental requirements are the simplest. In fluorescence methods, the analyte molecule in a fluid solution is illuminated by a nearly monochromatic source and is promoted to an excited state with no change in the total molecular spin quantum number. A fraction of the energy in the excited state is emitted by the analyte as radiant energy, and this radiation is viewed, in general, at a 90 degree angle. This means that the sample is viewed against a substantially darker background than is the case with the absorption spectrum where changes in the intensity of the exciting beam are viewed directly. A further enhancement of this dark background is provided by the Stokes shift of the emitted light which will be described more fully in the theoretical treatment of fluorescence in Chapter II. This results in a 100 to 1000 fold increase in sensitivity of the fluorometric method over the absorbence method (1). Another advantage of fluorescence in analytical chemistry is the fact that not all compounds that absorb light exhibit fluorescence. Often an analytical sample is a solution of several compounds with overlapping absorption bands, which interfere with the species of interest in absorption spectroscopy. In fluorescence spectroscopy, which depends on the presence of aromatic rings and/or extended counjugation, only those compounds having a structure which promotes fluorescence are detected. By adjusting the excitation wavelength we may also preferentially excite one compound, enhancing its fluorescence and depressing the fluorescence of an interfering species. Both of these properties enhance the selectivity of the analysis. Finally, the fluorescence signal is the sum of all the emitted photons which are collected by the detection optics and converted to an electrical signal. Therefore, the dynamic range of this technique is, in the first approximation, limited only by the solubility of the analyte in the solvent. In practice the inner filter effect has usually limited the dynamic range to 4 orders of magnitude (2) although recent efforts (3,4,5,6) have extended this range through various corrective procedures. Fluorescence lifetimes are, generally, on the order of 2 to 30 ns (7). These times are far shorter than those in which simple computerized data acquisition systems operate, so one must employ sophisticated measurement systems to evaluate these lifetimes. Until recently the time for the collection and analysis of the raw data precluded the use of fluorescence lifetimes in routine analysis. The increasing availability of sub-nanosecond lasers and the recent introduction of high speed, gated imaging detector systems (8) may change this in the relatively near future. At present the use of time-resolved fluorescence is limited to research investigations rather than routine analysis. Fluorometry has gained its greatest application in clinical chemistry (9), with a number of routine analyses being performed by this method. These include tests for phenylketonuria performed on most newborn infants, porphyrins in the urine, corticosteroids in the plasma, and estrogen, catecholamines, and bilirubin levels in serum. Other areas where the high sensitivity of fluorometry has promoted its application are in the pharmaceutical industry and in the evaluation of pesticides and environmental contaminants (9,10). Fluorometric detectors for high performance liquid chromatography (HPLC) have also been the object of very intensive development in recent years. The great sensitivity of fluorometry has been of utmost value in this area where sample concentrations are very low (11). Phosphorescence is a related emission process that was first noted four centuries ago (9). It has only recently become a method used for cannon analytical problems, especially through the prolific efforts of Winefordner's group at the University of Florida (12). When a molecule is excited some of the excited electronic states undergo a spin inversion. The electronic spins become uncoupled, and the molecular spin quantum number, S, goes from 0 to 1. When the molecule is placed in a restricted medium such as a solvent of high viscosity, a micellar environment, or a molecular caging environment, or when it is adsorbed onto an appropriate solid substrate, this new spin state, called the triplet state, is protected from deactivational processes. It can emit characteristic radiation, called phosphorescence, and return to the ground state. The traditional method of observing phosphorescence has been to freeze the sample in a solvent at liquid nitrogen temperatures producing a rigid glass. This method is time consuming, prone to fracturing of the glass, and susceptible to frost in the liquid nitrogen chamber. These factors limited the applicability of the method in routine analyses. Recently, the development of alternative methods based on molecular structuring of the environment has resulted in methods of analysis that can be performed at room temperature. They offer much greater speed and ease of analysis, and as a result phosphorimetry has undergone a surge of interest. It is the solution techniques involving micelles and cyclodextrins, methods extensively developed by Cline- Love's group at Seton Hall University, which are of greatest interest in this project (13). Interest in phosphorescence is derived from the analytical advantages of phosphorimetry which arise from the photophysics of the phosphorescence process. As previously stated, fluorescence is a very rapid process which makes evaluation of fluorescence lifetimes difficult. However, the spin inversion process which gives rise to the phosphorescence phenomenon is classically "forbidden". Once the spin inversion has occurred, the molecule is trapped in the electronic state giving rise to phosphorescence, and the lifetimes are much longer. They range from about 200 us for some room temperature systems to 10 s for for certain cryogenic systems to several minutes for phenomena occurring in plastic media such as the "glow in the dark frisbee". As a result, the exciting illumination can be removed and the phosphorescence signal can be observed against a completely dark background in a time frame easily accessible to modern data acquisition circuits. The dark background gives phosphorescence its high sensitivity. In addition, fewer compounds phosphoresce than fluoresce, which means that phosphorescence is even more selective than fluorescence. Since phosphorescence lifetimes can be readily measured, another element of selectivity is added since the observed phosphorescence lifetime is quite variant for closely related compounds. Additionally, the quantum mechanical processes which give rise to the spin inversion, collectively called intersystem crossing (ISC), are effected by the coupling of the spin angular momentum of the electron to the orbital angular momentum of the molecule. This is known as spin- orbit coupling (SOC). SOC is particularly promoted by the presence of atoms in the molecule or its environment having heavy nuclei and filled d-orbitals, a phenomenon known as the "heavy atom effect". This requirement adds another degree of selectivity in the analysis, which is valuable when considering many compounds of current concern, such as the halogenated biphenyls and the fused ring dioxins and dibenzofurans. This research is especially concerned with the development of instrumentation to investigate these phenomena. The construction of a versatile luminescence spectrometer based on a conventional monochromator type spectrofluorometer is the major portion of the work. This instrument can obtain fluorescence spectra, phosphorescence spectra, and phosphorescence lifetime data over the wide range of times found in the various phosphorescence methods. This discussion includes the development of a FORTH based computer data acquisition system which can collect the data and it includes programs on minicomputers that fit time domain data to the appropriate function to obtain a working lifetime for compound identification. The relationship of the FORTH operating environment to the design of the instrumental interface will be considered. Finally, this document presents a description of a board designed to convert the data acquisition system from the present unsupported in- house computer to the broadly accepted IBM-PC family of computers. This device will allow many other instruments in the department using this same in-house computer, to move to the powerful IBM data acquisition system. The most important features of this device are a set of latched I/O ports, a chip select decoder for controlling instruments, and a versatile timer chip for generating precise time-bases. Their functionality and relationship to FORTH programming are discussed. 10. ll. 12. CHAPTER I References Borman, S.R., Anal. Chem., 1982, 54, 327A. Winefordner, J. D.; Schulman, S. G.; O'Haver, T. C., Luminescence Spectroscopy in Analytical Chemistry, Wiley-Interscience: New York, 1972; p.293. Holland, J. F.; Teets, R. E.; Kelly, P. M.; Timnick, A., Anal. 'Chem., 1977, 49, 706. Christmann, D. R., Crouch, S. R.; Holland, J. F.; Timnick, A., Anal. Chem., 1980, 52, 291. Christmann, D. R.; Crouch, S. R.; Timnick, A., Anal. Chem., 1981, 53, 276. Adamsons, K.; Timnick, A.; Holland, J. F.; Sell, J. E., Anal. Chem., 1982, 54, 2186. Guilbault, G.G., Practical Fluorescence: Theory, Methods, and Techniques, Marcel Dekker: New York, 1973; pp. 16,17. OMA III, maufactured by EG&G: Princeton Applied Research, Princeton, N. J. O'Haver, T. C, J. Chem. Ed., 1978, 55, 423. Van Duuren, B.L.; Chan, T.-L., "Fluorescence Spectrometry", Spectrochemical Methods of Analysis, Winefordner, J.D., ed., Wiley-Interscience: New York, 1971; volume 9, pp. 428-445. Weinberger, R.; Sapp, E., Am. Lab., 1984, 16(5). Berry, V., Am. Lab., 1986, 18, 36. Barman, S., Anal. Chem., 1982, 54, 327A. Hulshoff, A.; Lingeman, H., "Fluorescence Detection in Chromatography", Molecular Luminescence gagggggggggy: Methodsgggg Applications; Part 1, Schulman, S. 6., ed., Wiley-Interscience: New York, 1985; pp.654-696. In the course of reviewing the literature more than 70 papers were attributed to Professor J. D. Winefordner and his research group at the University of Florida, Gainesville, Fla. Professor Winefordner has made a very large contribution to the 13. popularization of emission spectroscopy, particularly phosphorescence. In the field of Room Temperature Phosphorescence in liquids Professor L. J. Cline-Love of Seton Hall University, South Orange, N. J. has made major contributions as well as having published more than 30 papers on the topic in the last decade. CHAPTER II The Absorption of Light and the Generation of Fluorescence The Absorption of Light When a molecule is exposed to electromagnetic radiation, it will absorb a photon of that radiation. This can happen only if the photon is sufficiently energetic to cause an electronic transition. The energy necessary to do this is dependent upon the types of bonds available in the molecule. The energy needed to promote a sigma bonding electron to an anti-bonding orbital in a C-C or C-H bond (o->o*) is on the order of 800 kJ/mol. This energy corresponds to electromagnetic radiation of a wavelength of approximately 150 nm. To promote a lone-pair electron in a sigma bonded hetero-atom, this energy (n->o*) is on the order of 600 - 650 kJ/mol and occurs when irradiated with light in the 185 - 200 nm region. For doubly bonded 0 there is a possibility that the lone pair electron may be promoted to the n* orbital (n->n*) by the absorption of electromagnetic energy in the 275 - 300 nm region. This transition is, however, a weak one (1). The most important element in the absorption of radiation in the ultra-violet and visible region, from the point of organic molecular spectroscopy, is the C=C double bond. One of the electrons in the n cloud of this bond can absorb a photon and be promoted to the n anti- 10 bonding orbital (n->n*). The energy of such a transition for an isolated double bond is around 190 nm and is not of tremendous interest. However, many molecules of analytical and biological interest contain sequences of these doubly-bonded units alternated with single bonds in a system known as extended conjugation. The electrons behave as though they are constrained in a continuous tube of length equal to the extended conjugation. The quantum mechanical solution has a much lower energy dependent on the degree of extended conjugation. There is a simple formula which may be used to calculate the approximate wavelength of the most intense absorption (2). When light is absorbed by a solution, the power of the beam is attenuated according to the Beer-Lambert law. This law has the following expression: -log (P/Po) = A = sbc (2.1) where P0 is the incident radiant power of the beam, P is the transmitted radiant power, A is the absorbence, and c is the molar absorptivity. The validity of this expression rests on several assumptions. They require that the solution is sufficiently dilute that changes in the refractive index are negligible, that the radiation is monochromatic, and that there is no stray light (3). Of particular interest in the study of luminescence spectroscopy is the case where the extended conjugation of the absorber exists in closed loops. The proto-typical molecule of this sort is benzene. There are two broad classes of compounds based on the ring. One is the class of compounds with 4n+2 electrons in the n cloud, where n is an integer. These compounds fall under HUckel's rule for aromaticity which requires that ring systems with this number of n electrons will have a net 11 reduction of energy relative to a system where each double bond is isolated (4). They are known as cata- condensed compounds. For these compounds Platt conceived the perimeter free electron orbital model (PFEO) which describes the extended conjugation with a periodic boundary condition (5). When a molecule with m rings absorbs a photon and an electron is excited, this model results in a pair of doubly degenerate molecular orbitals. Those that result from a molecular orbital with one node are termed the B states and those arising from an orbital with 2m+1 nodes are called the L states. When the periodic potential of the C atoms is introduced into the PFEO the degeneracies are removed and this results in a total of four excited states, where the members of a pair are differentiated by the presence (Ba' La), or absence (Bb, Lb) of a longitudinal node. In the simple molecular orbital (MO) treatment the spin operators may be integrated out of the wave equation, when using the dipole operator. This is because the total spin operator, $2, commutes with the dipole operator (6). As a result, the fact that the ground state must be symmetric and the dipole operator representing the electronic transition is anti-symmetric demands that the final state he anti-symmetric with respect to the ground state. This is required by the group theoretical constraint to generate the totally symmetric representation (7). This means that the parity allowed transition is from the ground state to one of the two states 31' The PFEO model has been extended to the peri- class of compounds which have 4n n electrons, where n is an integer. These compounds do not exhibit an overall aromaticity or MO stabilization which characterizes the cata— class of compounds (4,8). They do show delocalized behavior, 12 and they undergo n->n* transitions well into the analytical ultra-violet (UV) region (wavelengths greater than 200 nm). This simple Platt model shows results similar to the more rigorous HUckel treatment which is discussed in most texts on quantum mechanics and has been shown to be correspondent to the HMO approximation (9). Formally, other transitions are inaccessible. However, vibrational modes couple with the electronic wave function (vibronic coupling) to give rise to absorption in these symmetry "forbidden" bands. Thus,the observed spectrum is best evaluated by the specific interaction integrals which give the oscillator strength describing the intensity of the absorption band. As a general guideline, though, the simple group theoretical treatment is a valuable qualitative starting point. One important factor that does hold from this simple qualitative approach is the idea of spin conservation. Since the spin operator is not explicitly invoked for the solution of the dipole absorption process, the spin quantum number is conserved. Since the ground state of aromatic molecules has all electrons paired, the spin quantum number, S, is 0 and the spin state of the target electronic configuration must be 0. This condition is invoked for the emission process also. The multiplicity of a state is defined to be ZS+1. If S=0, as in the ground state, the multiplicity is one. It is called a singlet state and it has no Zeeman components. In the simple PFEO model, emission spectra must arise from a singlet-singlet transition, and this is the source of fluorescence emission. When a compound is in solution at or below room temperature, it is usually in the lowest vibronic level of its lowest electronic singlet state, called So. When it absorbs radiation in the UV or visible region, 13 it is excited to some higher singlet state, S A transition to the n' lowest vibronic level of 81 is called the 0-0 transition. This transition will be the lowest energy absorption transition. The oscillator strength induced by the coupling of various vibrational modes with the excited electronic orbital governs the observed intensity of absorptions at higher frequencies. This effect is heightened in solution by interaction of the wave functions of the solute with those of the solvent molecules. In the gas phase, it is possible to observe absorption line spectra, but in solution absorption spectra are broad featured. The most intense band need not be the 0-0 transition. The usual explanation forwarded is the Franck-Condon Principle. Electronic transitions-take place in approximately 10'15 s, and on this time scale the nuclei move very little. According to the Born-Oppenheimer (BO) approximation, the quantum mechanical electron function may be separated from and solved independently of the nuclear function. This yields solutions which are stationary states in the sense of the Ehrenfest adiabatic prinCiple (10). The Franck-Condon principle holds that 3 electronic transitions take place with no change in nuclear configuration. Since the excited electronic state is usually more diffuse spatially, the equilibrium inter-nuclear distance for the excited state is generally greater than for the ground state. This means that the potential surface for the electrons has shifted relative to that for So. For the electron in its most probable location in the SO potential surface (the center), a vertical transition in the nuclear co- ordinate will most probably yield an excited vibronic level of S1 (See Figure 1). As a result the strongest absorption band is often shifted l4 \ JV!“ V13 V81 i ll , v=o1234... E Figure 1. A potential energy diagram for absorption showing a shift from the 0-0 band maximum due to a change in the equilibrium inter—nuclear distance in the excited state. 15 from the 0—0 band, and this shift may be used to study nuclear re- orientation in the excited state (11). Fluorescence Once the molecule has absorbed a photon and been promoted to an excited state, there is a complex set of processes that occur to release the energy and return the molecule to the ground state. The electron is usually in an excited vibrational level of an excited singlet state. It can lose its excess vibrational energy through collisional interaction with the solvent, a process called vibrational relaxation (VR). Alternatively, a highly excited vibronic level of a lower singlet may overlap the occupied level and result in a radiationless transition, called internal conversion (IC). Since vibronic levels are very closely spaced, collisional deactivation is a very fast process, and the highest probability is through VR to the lowest vibronic level of the given singlet. At this point, the most probable transition is IC to excited vibrational state of the next lowest singlet and subsequent vibrational relaxation, as above, falling through the excited state manifold until it reaches the lowest vibronic level of S1. At this point the fact that the energy of excited states is quadratically distributed enters into consideration. This means that the excited states are energetically close enough that vibronic overlap with the next excited state is sufficient to provide probable IC deactivation of the upper state during its intrinsic lifetime. The time for the IC process and VR is considered to take about 10"12 s (12), so electronic relaxation to the lowest vibrational level of $1 is highly favored. 16 However, in many molecules the difference between $1 and S0 is large enough that IC does not compete, completely, with the tendency of the molecule to deactivate by emitting a photon (13). This emission is fluorescence , and the ratio of this emission to the light absorbed, which equals the sum of the IC and fluorescence components, defines the fluorescence quantum yield of the system. See Figure 2 for a simplified Jablonskii diagram representing the processes giving rise to fluorescence. Since the fluorescence originates from the lowest vibronic level of $1 and is unrestricted as to its destination in the vibronic manifold of So, the fluorescence emission is of lower energy than the excitation photon, which promotes an electron from the lowest vibrational level of $0 to an unspecified vibrational level in Si' This reduction in energy of the fluorescence bands relative to the absorption spectrum is called the Stokes shift. It is reasonable to expect this shift since a series of non-radiative energy losses were postulated between absorption and fluorescence. The only fluorescence band which should overlap an absorption band is the 0-0 band, which is nominally unchanged as it is from the same pair of formal states in both absorption and emission. However, there is a set of circumstances which causes the 0-0 band to shift. When the molecule absorbs a photon, it is initially at equilibrium with its solvent matrix. As previously discussed, absorption takes place more rapidly than the relaxation of the nuclear framework. Likewise, the solvent sheath does not relax in this period of time. The excited molecule occupies a spatially larger configuration with a different dipole vector. It finds itself in a higher energy solvent orientation than equilibrium would dictate. In the time scale of the 17 21 £— ‘31 A 1c F l L So Figure 2. A simplified Jablonskii diagram for fluorescence. A is absorption and F is fluorescence. 18 fluorescence phenomenon, the solvent is able to re-orient to a lower energy arrangement. Thus, when the molecule fludresces, each band is shifted to a slightly lower energy based on solvent re-orientation, and we see that the 0-0 bands are not superimposable (14). As with the absorption spectrum, the Franck-Condon principle is an important factor in determining the appearance of the strongest bands. Since the vibronic energy surface for a given molecular framework is not going to vary greatly with minor alterations of that framework, the potential energy surface diagrams will bear a symmetry between the th vibronic level of each lowest vibronic bands of So and 51 and the n state. In the first approximation if a given absorption has a specified Franck-Condon probability into S1 then the transition from the lowest vibronic level of 81 into the SO manifold will show the same relative probability at the nth vibronic level (See Figure 3). As a result the absorption and fluorescence spectra of a compound often bear a rough mirror symmetry to one another (15). The fluorescence spectrum has been discussed in terms of the absorption process and the resulting excited states. It should be apparent that if one selects one of the fluorescence bands and monitors the intensity of this emission as one scans the absorption bands the result should be very similar to the absorption spectrum. This is true because all fluorescence originates in the lowest vibronic level of S1 and all decay processes must pass through this state. Therefore, the population of the fluorescing state must be proportional to the excited state population produced by absorption. Experimental factors such as the changes in the source intensity and changes in the monochromator throughput must be accommodated to give a corrected spectrum (16). 19 Jm I V., ——""‘—1r—Wp—.- 5 p J V we VII V“ t:__ v-o vso V I R I! ll 11"] v=03234.. v=...4321o : Figure 3. A potential energy diagram for fluorescence showing the mirror symmetry between absorption and fluorescence. 20 Fluorescence Kinetics and Related Processes To consider the rate at which photons are absorbed, first consider a sample of molecules in a black body cell with refractive index n. Then the number of molecules leu being excited from a lower state S1 to an upper state Su separated in energy by v is determined by the relationship leu/dt = N1 * Blu * p(v) (2.2) where N1 is the number of molecules in the lower state, Blu is the Einstein coefficient of absorption, and p(v) is the density of radiation in a frequency band capable of promoting the molecule from $1 to Su. For a molecule in an excited population Nu there is an equal probability for interaction with the same photon to produce a transition from Su to 81' This is described by the Einstein coefficient for stimulated emission, Bul‘ There is also an intrinsic probability that the excited state will spontaneously decay. This rate constant is described by the Einstein A coefficient, Aul‘ For the excited state population there is an overall rate of decay to the lower state, dNul, given by dNul/dt = 'Nu { Aul + Bul * P(V) } (2-3) Under equilibrium conditions of constant illumination the rates in Equations 2.2 and 2.3 must be equal. When the equations above are combined with the blackbody radiation expression, the following relationship results Aul = sphv3n3c'313ul (2.4) where h is Planck's constant (17). 21 The radiative transition probability, Aul' is usually referred to as kR. The reciprocal of kR is called the radiative lifetime, IR, Strickler and Berg (18) have shown that tR is inversely proportional to the integrated absorbence of the band giving rise to the fluorescence. An important feature of their treatment is their conclusion that the more intense a band is, the shorter its radiative lifetime. The weaker n->n* transition should then give rise to a state with a longer lifetime than the n->n* transition. This will be of importance in phosphorescence. For 51 the radiative lifetime by this method is calculated to be between 10'9s for n->n* transitions to 10'68 for n->n* transitions. Compared to the 10’12 s for VR and IC processes in higher states, it is easily seen that emission should occur from the lowest vibrational level of the first excited electronic state according to Kasha's rule (19). If an analogous constant, kIC' is postulated for the IC process, then under constant illumination dNu/dt = A * P0 - ( kR + kIC ) Nu = 0 (2.5) and this yields the fluorescence quantum efficiency ¢F = kR / ( kR + kIC ) (2.6) If the exciting light in the system above is terminated, equation 2.7 reduces to dNu/dt = -Nu * kF = IF (2.7) where kF = kR + kIC' Now, dI/dt = 'kF * dNu/dt = -I * kF (2.8) and dI/I = -kF dt (2.9) 22 which, upon integration from t=0, yields I = IO exp( 'kF * t ) (2.10) where I is the intensity of the fluorescence. Here, kF is referred to as the first-order fluorescence rate constant, and its reciprocal is called the fluorescence lifetime, IF. In general this lifetime is characteristic of the compound. The observed lifetime is usually less than the fluorescence lifetime due to the intervention of two major processes. The first of these is diffusional collisional quenching to SO, and it is characterized by a dependence on solution viscosity, lifetime of the 81 state, and concentration of the quencher, [Q] (20). The most commonly cited quenching agent is molecular 02 which is thought to form an excited state complex, or exciplex, which then goes through its own non- radiative VR process to the ground state, where it then dissociates into the component 02 molecule and the analyte molecule (21). Similar exciplex quenching of indole fluorescence is reported by carbonyls such as acetic acid and acetone (22), by DMF (23), and by electron withdrawing amides (24). Likewise, a large body of data indicating exciplex type of quenching to So is noted for various chlorophylls (25). The effective rate constant under these circumstances becomes k = kF + K * [Q] (2.11) The observed quantum efficiency, e, is e = kR / ( kF + K * [Q] ) (2.12) and (¢p/¢)-1=K*[Q] (2.13) called the Stern-Volmer relationship (26). 23 The second major mechanism is static quenching which is not viscosity dependent. In this model the quencher forms some sort of complex with the fluorophore and reduces fluorescence by inhibiting the excitation process (20). This sort of quenching may be invoked for situations such as acetate quenching of phenols and does not follow Stern-Volmer kinetics (27). Methods of Determining Lifetimes The theoretical emission decay function after a short illumination pulse has been shown to be an exponential. If the detector and amplifier have been constructed so that there is no dark signal, then a plot of the logarithm of the signal intensity versus time will give a straight line. A least squares fit to the linearized data will yield the rate constant, k, and its reciprocal is the experimental lifetime (28,29,30,31,32,33,34,35). This same technique has been used to determine the lifetimes of the components of binary solutions with some success (36). If the lifetimes of the components differed by less than a factor of 5, significant errors were observed. If the dark current term is non-zero, a method by Guggenheim (37) can be used to accurately linearize first order kinetic data (38,39). It, however, requires sampling the data for a minimum of four lifetimes which may not be possible. There are several non-linear fitting methods which have received wide application. These include the simplex optimization algorithm (40), which is an unweighted method, and the analytical method which is 24 weighted. The method of Marquardt combines these two approaches (41,42,43,44). With the exception of picosecond laser excitation, a common problem in determining lifetimes is caused by the fact that the fluorescence signal is starting to appreciably decay in intensity before the excitation pulse has completely died. As a result deconvolution algorithms must be used to remove this contribution from the signal before determining the lifetime. This has been done in an analog fashion using a boxcar integrator (45). Least squares fitting methods have been popular for deconvoluting the lifetime data from the excitation pulse for single and multi-component systems (32,36,46,47). The phase plane method has been applied to single component systems (47). A systems- theory based method of moments has been applied to experiments where decay curves were collected at several excitation and emission wavelengths (48). Another class of lifetime determination experiments is based on a modulated light source and the lag phase of the emission signal behind the source. This method was actually the first method developed, but the advent of intense laser pulse technology has at least temporarily superseded it (49,50,51). This provides a sketch of the ideas involved in establishing a model for what might be considered the zeroth-order of molecular luminescence. That is the model where the spin and orbital angular momenta are separately conserved. In practice there are many factors which lead to the breakdown of this assumption. When this assumption and the BO approximation fail there is an appreciable development of a 25 spin inversion of the excited electron which gives rise to the triplet state and phosphorescence. This will be the topic of the next chapter. 26 CHAPTER II References 1. Burdett, J. K., "Electronic Spectra of Polyatomic Molecules", Spectroscopy, Straughan, B. P.; Walker, S., eds.; Halsted Press: New York, 1976; v.3, p.131. 2. Burdett, J. K., ibid., p.133. or Pasto, D. J.; Johnson, C. R., Organic Structure Determination, Prentice-Hall: Engelewood Cliffs, N. J., 1969; pp.92-100. 3. Cheng, K. L.; Young, V. Y., "Ultraviolet and Visible Absorption Spectroscopy", Instumental Analysis,2nd ed., Christian, G. D.; Reilly, J. E., eds., Allyn and Bacon: Boston, 1986; pp.174-181. 4. For a graph-theoretical proof of this generally accepted concept see Gutman, 1., Chem. Phys. Lett., 1977, 46, 169. 5. Birks, J. B., Photophysics of Aromatic Molecules, Wiley- Interscience: London, 1970; p.7. 6. Pilar, F. L., Elementaryyguantum Chemistry, McGraw-Hill: New York, 1968; p.83,319. or Murrel, J. N., The Theoryiof the Electronic Spectra of Organic Molecules, John Wiley & Sons: New York, 1963; p.20. 7. Birks, J. B., op. cit., p.89. or Burdett, J.K., op. cit., p.135. or Cotton, F. A., Chemical Applications of Group Theory, Wiley- Interscience: New York,1971; Chap. 5. or Joshi, A. W., Elements of Group Theory_for Physicists, 2nd ed., John Wiley & Sons: New York, 1977; pp.168-171. 8. Pilar, F. L., o . cit., 650. 9. Pilar, F. L., ibid., 655 10. The BO approximation assumes that the superposition of states which are solutions to the nuclear independent wave-equation at various nuclear co-ordinates is a valid solution to the total Hamiltonian. The molecule is supposed to oscillate adiabatically from state to 27 ll. 12. l3. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. state. The BO approximation is demonstrated to break down in low pressure gases where radiationless deactivation of excited states is observed to occur more rapidly than collision pertubation models admit. This has been an area of intense study for many years and a list of discussions in this area is included for general interest: Pilar, F. L., op.cit., pp.4l4-421. Azumi, T.; Matsuzaki, K., Photochem. Photobiol., 1977. 25, 315. Born, M.; Oppenheimer, J. R., Ann. der Phys, 1927, 84, 458. Freed, K. F., Accts. Chem. Res., 1978,11, 74. Henry, B. R.; Kasha, M., Ann. Rev. Phys. Chem., 1968, 19, 161. Heller, D. F.; Freed, K. F.; Gelbart, W. M., J. Chem. Phys., 1972, 56, 2309. Rhodes, W., J. Chim. Phys., 1970, Zewail, A. H.; Orlowski, T.T.; Jones, K. E., Proc. Natl. Acad. Sci. USA, 1977, 74, 1310. Birks, J. B., op. cit., p.53. Parker, C. A., Photoluminescence of Solutions, Elsevier: Amsterdam, 1968; p.70. Murrel, J. N, op. cit., p.289. Parker, C. A., o . cit., p.13. Parker, C. A., ibid., p.8 or Birks, J. B., p. cit., 84-87. Guilbault, G. 6., Practical Fluorescence: Theory, Methods, and Techniques,Marcel Dekker: New York, 1973; p.8. Pilar, F. L., op. cit., p.113 or Birks, J. B., op. cit., p.50 or Burdett, J. K., op. cit., pp.297-315. Strickler, S. J.; Berg, R. A., J. Chem. Phys., 1962, 17, 814. Kasha, M., Diss. Far. Soc., 1950, 9,14. Schulman, S. G., "Luminescence Spectroscopy: An Overview", Molecular Luminescence §pgg§ppgggpy, Methods and Applications: Part I, Schulman, S. 6., ed., Wiley-Interscience: New York, 1985; p.12. Birks, J. B., op. cit., 498. Ricci, R. W.; Nesta, J. M., J. Phys Chem., 1976, 80, 974. Froelich, P. M.; Gantt, D.; Paramasigamani, V., Photochem. Photobiol., 1977, 26, 639. 28 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. Froelich, P. M.; Nelson, K., J. Phys. Chem., 1978, 82, 2401. Wolfbeis, O. S., "The Fluorescence of Natural Products", Molecular Luminescence Spectroscopy, Methods and Applications: Part I, Schulman, S. G., ed., Wiley-Interscience: New York, 1985; p. 245,253. Parker, C. A., op. cit., p.72. Guilbault, G. G., op. cit., p.108. Demas, J. N., Excited State Lifetime Measurements, Academic Press:N. Y., 1983; p. 75. Horrocks, W. deW., Jr.; Sudnick, D. R., J. Am. Chem. Soc., 1979, 101, 334. O'Donnell, C. M.; Harbaugh, K. F.; Fisher, R. P.; Winefordner, J. 0., Anal. Chem., 1973, 45, 609. Harbaugh, K. F.; O'Donnell, C. M.; Winefordner, J. D., Anal. Chem., 1973, 45, 381. Shaver, L. A.; Cline Love, L. J., Appl. Spectrosc., 1975, 29, 489. Harbaugh, K. F.; O'Donnell, C. M.; Winefordner, J. D., Anal. Chem., 1974, 46, 1206. Charlton, J. L.; Henry, B. R., J. Chem. Ed., 1974, 51, 753. SPEX Technical Reference #1934C, SPEX Ind., Inc.:Edison, N. J. Cline Love, L. J.; Shaver, L. A., Anal. Chem., 1980, 52, 154. Guggenheim, E. A., Philos. Mag., 1926, 2, 538. Cline Love, L. J.; Skrilec, M., Anal. Chem., 1981, 53, 2103. Woods, R. J.; Scypinski, S.; Cline Love, L. J.; Ashworth, H. A., Anal. Chem., 1984, 56, 1395. Demas, J. N., op. cit., p. 80. Marquardt, D. W., J. Soc. Ind. App. Math., 1963, 11, 431. Wilson, R. M.; Miller, T. L., Anal. Chem., 1975, 47, 256. Nithipatikom, K.; Pollard, B. D., Appl;_§pectrosc., 1985, 39, 109. Goeringer, D. E.; Pardue, H. L., Anal. Chem., 1979, 51, 1054. Lytle, F. E., Photochem. Photobiol., 1973, 17, 75. 29 46. 47. 48. 49. 50. 51. Ware, W. R.; Doemeny, L. J.; Nemzek, T. L., J. Phys. Chem., 77, 2038. Demas, J. N., op. cit., Chap. 8. Eisenfeld, J.; Ford, C. C., Biophys. J., 1979, 26, 73. Demas, J. N., op. cit., p. 50. Mousa, J. J.; Winefordner, J. D., Anal. Chem., LueYen, E.; Winefordner, J. D., Anal. Chem., 30 1977, 49, 1263. 1973, 1974, 46, 1195. CHAPTER III Phosphorescence Spin Orbit Coupling In the simplified fluorescence model previously developed it was explicitly assumed that there is no interaction between the spin angular momentum and the orbital angular momentum. As a result the spin momentum state is preserved when the orbital momentum is perturbed, and all molecular transitions originating from a singlet state must end in a singlet state. In reality there are a number of physical circumstances that can cause a breakdown of this approximation, and they are linked together by a treatment called spin-orbit coupling (SOC). The relativistic motion of the nucleus around the electron produces a magnetic field enveloping the electron, and the charge on the electron has some spatial inhomogeneity, causing the intrinsic spin of the electron to produce a local magnetic field. It is the interaction of these two fields which results in SOC causing the electron to "flip" its spin state (1). The resulting electronic state has two un-paired electrons and S=1. The multiplicity of this state is ZS+1 = 3, and it is called the triplet state. This is the source state for phosphorescence. In a series of papers McClure developed a crude expression for the SOC operator based on a central field approach (2,3,4,5). For hydrogen- 31 like atoms with nuclear charge 2 in a many electron central field xnl e z4 / [ n3 (1+1)(1+1/2)1 ] (3.1) where n and l are the usual quantum numbers associated with these symbols (6). Although the assumptions McClure imposed are crude - a central field model supposes a point source such as an atom rather than a spatially disperse molecule - it does give useful qualitative results. This, in part, is due to the power of group theory as demonstrated by McGlyn, et al. (6). This simplistic development shows two important trends that are carried over to molecular systems. The first, due to the 24 term, is that heavier nuclei can profoundly increase SOC. This is the so-called "heavy atom effect". The second is that increasing the quantum state of the electron rapidly attenuates SOC. For a given value of n this means that bonding through the lowest angular momentum state offers the most SOC. To maximize 2 at the same time requires that the perturbing atom be near a noble gas in the periodic table. Thus SOC should be maximized when a halogen atom is bonded to the molecule via a p-orbital; and the heavier the halogen, i.e. the greater 2, the more intense the So->T1 absorption and the T1->SO phosphorescence, which is in accord with experimental data (1,7). One interpretation is the state mixing model, where phosphorescence arises when the SOC operator mixes some intensity from an allowed singlet transition into the forbidden triplet transition. In this approach oscillator strength values such as those used by Strickler and Berg (8, same as reference 2.18) to predict fluorescence lifetimes are applicable, given a symmetry correction term. Efforts to improve the correspondence of the observed oscillator strength with the Strickler- 32 Berg theory, have centered on the inclusion of sp states in the SOC Hamiltonian for aromatic hydrocarbons, which show phosphorescence and weak So->T1 absorption (9,10,11,12,13). An alternative approach is that the molecule undergoes a singlet- singlet absorption in common with the fluorescence process. It undergoes the usual VR and IC processes to arrive in the 51 state. At this point there is a competing SOC process introduced into the previous treatment used for fluorescence. This results in a spin inversion of one of the electrons. This process of radiationlessly "jumping" from S1 to T1 is called inter-system crossing (ISC). Associated with ISC is a rate constant called kISC' which shall be considered in some detail later. First the energy of the T1 state relative to 51 state and its effect on the phosphorescence spectrum will be examined. See Figure 4 for a Jablonskii diagram showing the components leading to, and competing with, phosphorescence. Relative to the S1 configuration, the T1 state is lower in energy. This is due to the fact that in the singlet state the electrons are phased so that they more compactly occupy space giving a higher repulsive interaction. In the triplet state they are no longer phased so they exhibit an avoidance tendency that results in a more diffuse spatial distribution and a lower energy (1). A practical result is that the phosphorescence signal is shifted further toward longer wavelength than is the fluorescence spectrum (14). Additionally, the transition moment is proportional to the energy separating the two states, so the closer they are the stronger the phosphorescence (1,15). The mirror relationship between SO->Sl and the fluorescence spectrum is seen also between the SO->T1 absorption and the 33 — H e/VVE; V“ s” ' } 4L. V1! - 1:. ISO __/-'="— S. I I tf--l ‘“‘ A to F P to fl, 1 i I] LS... Figure 4. A Jablonskii diagram for phosphorescence and competing processes. A is absorption, F is fluorescence, and P is phosphorescence. 34 phosphorescence spectrum for simple aromatic compounds (14). This so- called Levshin symmetry (14), does not usually hold for the phosphorescence excitation spectrum where one phosphorescence line is monitored while the absorption spectrum is scanned. Although Kasha‘s rule that emission occurs from the lowest vibronic level of the first excited electronic state (here T1) is still generally observed (16), the SO->Sl absorption dominates S0->T1. Relative to singlet-singlet absorption, the vibronic coupling between S1 and T1 is such that the B0 states of S1 which give rise to the fluorescence symmetry are quite different from the 80 states of the T1 level. Coupled with the fact that there is evidence for the breakdown of the BO approximation (12,17), this indicates little visual correlation between the absorption and phosphorescence spectra (Figure 5). Another factor that disrupts any appreciable symmetry in the phosphorescence spectra of heteroaromatic molecules has been explained by El-Sayed‘s rule (18). In such molecules there are two kinds of transitions observed. One is of the same nature as in our previous discussions and results from the promotion of one of the n electrons to a n* MO in a transition, called a (n,n*) transition. The second mode arises from the promotion of one of the non-bonding n electrons associated with the hetero-atom to the n* orbital in a (n,n*) transition (19). These states are separated by a small energy difference, and their relative positions are often the same in the singlet and triplet multiplicities. Thus, if they are ordered 1(n,n*) > 1(n,n*), then 3(n,n*) > 3(n,n*). This is not an exclusive assignment and has many exceptions, but it is used here as an illustrative device. According to El-Sayed's rule, population of the triplet state in hetero-aromatics is 35 \ l J... l V J... V v.3 V J... ‘_ vcl . V3! V92. d.— Nul— o)-—- 5.- R llJl . 41:"...4 321 7: Figure 5. A potential energy diagram for phosphores-cence showing the breakdown of the absorption-emission symmetry due changes in the coupling between the T1 state and the so state. 36 primarily through ISC, which links an (n,n*) state to a (n,n*) state. This results in the following transitions 1(n,n*) -> 3(n,n*) or 1(n,n*) -> 3(n,n*) being favored over 1(n,n*) -> 3(n,n*) or l(n,n*) -> 3(n,n*) by a factor of 102-103 (15,20,21,22). The 1(n,n*) -> 3(n,n*) is invoked for small heterocyclic molecules (21), while the latter is supposed to be the best explanation for carbonyls such as quinoline (21), steroidal ketones (22), and anthroquinones (23). If the energy of the target 3(n,n*) state is higher than the source 1(n,n*) state, then the phosphorescence intensity is returned to the fluorescence band (20). This supports El-Sayed's rule in that very little l(n,n*) -> 3(n,n*) phosphorescence is observed even though it is energetically feasible as in sym-tetrazines and 9,10-diazophenanthrene (21). Once the ISC process is complete, the molecule undergoes a VR process to drop to the companion triplet state that is lower in energy. At this point it is noted that aromatics with a 1(n,n*) 51 state show the shortest fluorescence lifetimes (10'9- 10"8 8), while heterocycles fluorescing from 1(n,n*) show the longest lifetimes (10"5-10"3 s) (20,23,24). This is in accord with the generally lower absorption intensity of the 1(n,n*) process (19) when used in the Strickler and Berg formalism (8). As a consequence these compounds often exhibit only very weak fluorescence (25). The longer lifetime allows ISC to compete more effectively with fluorescence producing stronger phosphorescence. For the triplet states this relationship is reversed by the fact that when the 1(n,n*) is highly favored by absorption, the triplet process must be highly forbidden (18). This is due to the fact that only 37 the high energy (o,n*) component rather weakly mixes the triplet and singlet character of the transition. The resultant triplet radiative lifetime of an aromatic hydrocarbon, according to the Strickler and Berg treatment, will be much longer than that of a heterocyclic compound, where there is considerable interaction between the l(n,n*) and the 3(n,n*) states (15,18,20,21). In fact, this interaction is so strong that the heavy atom effect in compounds of this type is minimal (18,26). In the case of large molecules it is possible for the energies of the two triplet states to exchange their relative positions, resulting in phosphorescence from the 3(n,n*) yielding a phosphorescence spectrum which resembles that of the parent hydrocarbon (27). Kinetics The kinetics of the triplet state are formally similar to those developed for fluorescence. In this analysis, we first redefine kF to include kISC k1:- = kR + kISC + K * [Q] (3.2) Then we define OT, the triplet formation efficiency or triplet quantum efficiency, as follows: For the triplet population we set kw = kPR + kPNR + K'* [9] + kRIsc (3-4> where kPR is the intrinsic phosphorescence radiative rate constant. It is related to the triplet state radiative lifetime tPR by the relationship 38 Here kPNR is the non-radiative decay constant for the triplet state through an ISC process to SO; kRISC is the rate of thermal conversion of the triplet to S1; and K' is the quenching constant for a triplet quencher of concentration [Q]. The small energy separation of S1 and T1 which lends intensity to ISC also means that the reverse process may become thermally driveable at room temperature, but it is usually neglected since these experiments are traditionally conducted at 770K. Under these circumstances the experimental phosphorescence rate constant is given by A kE = kPR + kPNR + kP * [Q] (3.6) Assuming Kasha's rule to be valid in both T1 and S1 for a sharply defined pulse of light, this gives d[Tl] / dt = kISC * [S1] - kE * [T1] (3.7) which under initial conditions [S1] = [5110 and [T1] = 0 at t = 0 yields [T1] = kISC* [3110* {9XP('kEt)'eXP('kpt)} / (kp‘kp) (3-3) Since kF >> kE, the above expression reduces to [T1] = ( k13c* [51] / kF ) * exP(‘kgt) (3-9) at t > l/kF. For [T110 = °T * [5110 (3.10) equation 3.8 can be simplified to the familiar pseudo first order form [T1] = [T110 * exp(-kgt) (3.11) and IP = 1 / kE (3.12) is the phosphorescence lifetime for the system (27). As with 39 fluorescence, this lifetime in a given set of conditions can be indicative of a particular compound. Additionally, let us define two other parameters commonly encountered in phosphorescence. One is the phosphorescence quantum efficiency, qP, which is the ratio of the phosphorescence photons to the total number of triplets. The other is the the phosphorescence quantum yield, ep, which is the ratio of phosphorescence photons to the total number of photons absorbed qP = kPR / kE (3.13) and op = kPR* [T1] / [51] = qP * °T (3.14) The substitution of heavy atoms onto a molecular framework can greatly modify tp relative to that of the unsubstituted molecule. These effects have been observed to indicate direct modification of kPNR through SOC and through steric disruption of the parent molecule, so that the phosphorescence is closer to that of one of the substructure molecules (28,29). With respect to the nuclear charge, Z, kE values are found to increase with increasing 2. The magnitude of this change is in accord with simple one-center spin-orbit coupling theory (4,6,28,30). The dependence of the lifetime appears to be correlated, to some extent, with the EPR determined spin density of the phenanthrene site, to which the heavy atom is attached (28,31). The fact that the effects are not additive argues against the strict simple one-center interpretation (28). It has been determined that the increase in kg, and the attendant decrease in IP, is due to an increase in both of the components in the kE' kPR and kPNR (32), so that the increase in [T1] produced by SOC is mirrored by an increase in phosphorescence intensity. This is in accord 40 with the basic premise of SOC that communication between triplets and singlets is improved. Unloading efficiency should be enhanced as well as loading. Quenching and the External Heavy Atom Effect When fluorescence quenching by 02 was discussed previously it was assumed, for reasons of simplicity, that the sink for the excited singlet state was the ground state with no concern for the detailed path. Evidence indicates that a major route for this quenching is through the triplet state via ISC (34,35,36). The ground state for the oxygen molecule is a triplet state as indicated by its paramagnetic behavior. In a collision with an excited molecule, some of the triplet nature of the 02 is exchanged with the 31' and some ISC is observed in the excited molecule. There is conflicting evidence as to the nature of the interaction. A charge transfer type of interaction has been both affirmed (36) and denied (34). It is asserted that triplet exchange between 02 and the excited singlet state is not a significant factor in establishing the triplet population leading to phosphorescence (21). In contrast, the triplet nature of the 0: ground state is significant in collisional quenching of phosphorescence. Due to the long intrinsic lifetime of the triplet engendered by the spin forbidden nature of phosphorescence, there is a comparatively long period for oxygen to interact with T1. It does this fairly efficiently through a triplet-triplet annihilation interchange or triplet excimer (37,38). It is thought that the annihilation to form 102 is the predominant mechanism by 3 orders of magnitude (39). This process is so effective 41 that 02 quenching of acriflavine is observable in amounts as small as 10"11 moles (40). In general practice it is found that 02 quenching is sufficiently effective that phosphorescence in fluids at room temperature is difficult to obtain and has only recently been observed. In those solutions where room temperature phosphorescence (RTP) is observable due to high viscosity or extensive 02 purging, a Stern-Volmer quenching relationship is observed and K' can be evaluated (41). As a result of this quenching, the traditional phosphorescence experiment is carried out at 77 K in a solvent that forms a glass. This greatly reduces the ability of any dissolved O2 to interact with the solute, and de-oxygenation of the system can provide some additional lengthening of the observed lifetime. Other historical techniques include dissolving the compound of interest in a purified host crystal such as durene or 1,2,4,S-tetrachlorobenzene (42) or naphthalene (43). At this point it should be pointed out that biacetyl, benzil, and anisyl have been known to phosphoresce at room temperature in degassed fluids (42). Much of the limited work in this area centered on the use of biacetyl which shows room temperature phosphorescence in both aqueous and cyclohexane solutions. It shows no radiative rate changes going from cyclohexane to water to D20, but it does show about a three-fold increase in the rate of the non-radiative processes (44). The group from the Free University in Amsterdam has utilized biacetyl's low coefficient of absorption in the UV in various "sensitized" phosphorescence experiments, where an analyte that does absorb UV light is promoted by the various mechanisms to the T1 state. From there the excited molecule interacts with the biacetyl in the solution, to transfer its energy to the biacetyl and radiationlessly 42 decay to SO itself. The biacetyl then phosphoresces, and the signal is proportional to the concentration of sensitizing material such as halogenated biphenyls or naphthalenes (45,46). It is the high degree of SOC character associated with the low-lying (n,n*) orbital which make this transfer possible. This group also reports the analytical use of the quenching of biacetyl phosphorescence by a wide variety of compounds in hexane, water, and the acetonitrile/water azeotrope (47). They report limits of detection (LOD) for each system in the neighborhood of 10'7- 10'9 M. The sensitized phosphorescence of biacetyl is also reported in the gas phase (48,49) and in the solid phase arising from supersonic jet condensation (50). In sensitized phosphorescence experiments the spectrum is always that of biacetyl. The analytical advantages accrue from the efficiency of the triplet-triplet transfer and the sensitivity of the phosphorescence method. An anomalously strong phosphorescence for certain bridged biphenyls in liquid alcohols and alkanes has been noted (51). However, it does not seem to have been generally applicable, as no further references to this research have been found. The External Heavy Atom Effect Although 02 quenching is not considered to be significant in the loading of T1, a related process is known to alter the triplet population radically. This is the so-called the external heavy atom effect. This phenomenon was first reported by Kasha (52), where the So- >T1 band for 1-chloronaphthalene was so enhanced by the addition of ethyl iodide, that it was readily observed in solution at room 43 temperature. He attributed this to an increase in the SOC due to collisional perturbation. McGlyn and co-workers studied this phenomenon extensively using mono-halonaphthalenes and a variety of alkyl halides as the external perturber. They found that the So->Tl oscillator strength increased approximately with the square of the SOC coupling factor for the heavy atom attached to the host alkyl. The external heavy atom effect is not as strong as internal heavy atom coupling, as would be expected (53,54). The primary result of this correlation is that the greater 2, the greater the enhancement of the triplet loading. In the cryogenic phosphorescence of these systems, it was found that the increased intensity of the emission signals reflected the increases in absorption. Additionally, the phosphorescence decay curves were found to be non- exponential. This was interpreted as consistent with a weak charge- transfer (C-T) complex (55), although little or no spectral shifting was seen as in true C-T complexes (56). The decay curves were believed to represent the sum of two exponential decays, one for the unperturbed molecule and one for the C-T complex. Since the medium was frozen, no averaging of the two was expected (55). The overall SOC coupling constant was found only crudely to correspond to the sum of the external and internal constants. This was believed to be the result of the acid-base nature of the C-T complex giving rise to the SOC. The weaker the acid-base interaction, the weaker the coupling. They did not consider the inherent weakness of the localized one-center treatment due to the complexity of the system. They did conclude that the order of effect due to the external heavy atom effect was kISC>kPNRZkPR and that the fluorescence quenching associated 44 with this type of system is due to the increase in kISC (57,58,59). This contrasts with findings that kPNR is less susceptible to the external heavy atom effect than is kPR (60,61). From an analytical point of view, the preceding relationships point to an important conclusion. If both kISC and kPR are increased then, regardless of whether kPNR is affected to a marginally greater or lesser extent, the intensity of the phosphorescence signal should rise. This has been found to be true for a number of fused poly-nuclear aromatic hydrocarbons (PAH), with a corresponding reduction in the LCD (62). The effect is not universal and some PAH's show dramatic reduction in their phosphorescence signal (62). This quenching is analytically useful in that it can reduce interferences in the analysis of one compound in a multi-component mixture. One of the major disadvantages of the use of the external alkyl halides is that they begin to absorb strongly in the region from 300-350 nm. Thus, they often absorb a substantial fraction of the energy needed to promote the formation of the T1 state driving the phosphorescence. This is especially true for ethyl iodide which was extensively used in the early experiments, because it provided the most SOC (ZI=53), and it had good glass forming properties; these are important in obtaining a good, reproducible phosphorescence signal (62). Winefordner and coworkers (63) found that the inclusion of a few percent of a short chain aliphatic alcohol in water produced a finely fractured "snow" that increased the phosphorescence signal, presumably through an internal reflectance mode. The light was reflected many times at the edge of each micro-crystal, creating a much longer path length, while still remaining in the acceptance aperture of the emission 45 monochromator. This solvent system allowed the use of halide salts as the heavy atom perturbers (63), since they are soluble in the system, and there is no longer a need to control concentrations to ensure good glass formation. This second feature is a generally extensible advantage to snowed solvent systems. Glass formation requires careful cooling, while snow formation in capillary tubes (64) may be affected much more casually with a great time saving in the analysis. The halide salts do not absorb appreciably above 250 nm; this minimizes any apparent "red-shift" of the phosphorescence spectrum due to absorption of the emission by the halide perturber. Transparency in this spectral region increases the amount of radiation available from the source to drive the phosphorescence photo-physics. As a result, phosphorescence is increased which increases the sensitivity of the analysis. The sensitivity is also improved by the fact that the inorganic halides are available in much greater purity than are the organic compounds. The increased purity lowers the background emission which often has limited the LCD of samples run with ethyl iodide glasses (65). However, for experiments which used a pulsed laser light source, the residual contaminants in the halide salts still limited LOD's to poorer figures than without heavy atom enhancement (66). The external heavy atom effect for either organic or inorganic halides appears to be primarily limited to Br” and I“. A few compounds with a very high density of chlorine such as carbon tetrachloride, have demonstrated the external heavy atom effect (67). The halide ions appear to be more efficient, reaching a maximum effect at about 4% by wt. of the halide and leveling at that degree of enhancement. The external heavy atom effect is also seen to be less pronounced in compounds that 46 exhibit a stronger native phosphorescence, i.e. those with an 3(n,n*) lowest triplet (65). Almost all compounds investigated show some signal enhancement with 1' (65,68). An even greater increase in the phosphorescence signal is usually seen when the molecule is in the presence of Ag+ (68,69,70,71). This latter phenomenon has led workers to conclude that the external heavy atom effect operates through a C-T mechanism (69). This conclusion does not account for the fact that the effect is not observed at 77°K for T1+ and Pb+2 (69,70), and is no stronger for dimethyl-Hg than it is for ethyl iodide in heptane (70). As with the halides the background phosphorescence of the perturber generally degrades the LOD under pulsed laser excitation (66). For the analysis of nucleosides using conventional phosphorimetry, it is asserted that the use of I' and Ag+ improve the LOD's (68). The mechanism for the Ag+ interaction is so strong and pH dependent, that it is believed to be more like an internal heavy atom phenomenon proceeding through a true C-T complex (66). Whenever the external heavy atom effect is seen to operate, the observed lifetime, IP, is generally reduced (65,66,68,69). However, some of the results are believed to result from re-ordering of the triplet energy levels due to solvent effects, since the polar aqueous solvent mixture can interact strongly with the 3(n,n*) state. For the inorganic halide systems the lifetimes are exponential when a conventional pulse lamp is used (64). However, when a pulsed laser light source is used, two components with widely divergent lifetimes and responses to the external heavy atom effect are observed (66,69). The problem of externally enhanced SOC does not readily lend itself to a single systematic treatment. In the field of analytical 47 chemistry there are some applications which do warrant exploitation. In particular, the ability to stimulate phosphorescence in one compound while not effecting a very similar compound offers the possibility of selective phosphorescence analysis. Also the ability to attenuate IP for one compound more than another allows the possibility of temporally tailoring the analysis so that the phosphorescence sampling of one compound occurs after an interfering signal has died away (71). Although some analytical work has been done using cryogenic phosphorescence, the experimental difficulties pursuant to obtaining the information, including 02 scrubbing, working with liquid N2, and trying to obtain suitable glasses, have precluded its routine application. Because of the photophysics responsible, phosphorescence has some very attractive properties which have led to continued efforts to develop improved methods using less severe conditions. The specificity of SOC interactions imply that fewer compounds will exhibit phosphorescence so that analysis by this method will be less susceptible to interference. Another important asset is that the lifetime, tP, is in the range of 10' 4 - 10 s. This range allows relatively straight-forward data collection, and possession of the lifetime equips the chemist with another parameter through which analysis may be accomplished. These efforts have resulted in the development of several methods for obtaining phosphorescence information at room temperature. Examination of these RTP methods will form the basis for the next two chapters. 48 10. 11. 12. 13. 14. 15. l6. 17. 18. CHAPTER III References Turro, N. J., J. Chem. Ed., 1969, 46, 2. McClure, D. S., g; Chem. Phys., 1949, 17, 665. McClure, D. S., J. Chem. Phys., 1949, 17, 905. McClure, D. S., J. Chem. Phys., 1952, 20, 682. McClure, D. S., J. Chem. Phys., 1954, 22, 1668. McGlyn, S. P.; Azumi, T.; Kinoshita, M., Molecular Spectroscopy of the Triplet State, Prentice-Hall: Englewood Cliffs, N. J., 1969; Chap. 5. McGlyn, S. P.; Azumi, T.; Kinoshita, M., op. cit., p.624. Strickler, S. J.; Berg, R. A., J. Chem. Phys., 1962, 17,814. Henry, B. R.; Siebrand, W., J. Chem. Phys., 1969, 51, 2396. Henry, B, R.; Siebrand, W., J. Chem. Phys., 1971, 54, 1072. Siebrand, W., Chem. Phys. Lett., 1970, 6, 192. Fischer, S. F.; Lim, E. C., Chem. Phys. Lett., 1972, 14, 40. Masmanidis, C. A.; Jaffe, H. H.; Ellis, R. L., J. Phys. Chem., 1975, 79, 2052. McGlyn, S. P.; Azumi, T.; Kinoshita, M., op. cit., p.6. Lim, E. C., "Vibronic Interactions and Luminescence in Aromatic Molecules with Nonbonding Electrons", Excited States, Lim, E. C., ed., Academic Press: New York, 1977; p.305, ff. Chu, S.-Y.; Goodman, L., Chem. Phys. Lett., 1975, 32, 24. Breiland, W. G.; Harris, C. B., Chem. Phys. Lett., 1973, 18, 309. El-Sayed, M. A., Accts. Chem. Res., 1967, 1, 8. 49 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. Kasha, M., "The Nature and Significance of n -> n* Transitions", Light and Life, McElroy, W. D.; Glass, B., eds., The Johns Hopkins Press: 1961; pp.31 - 61. Barltrop, J. A.; Coyle, J. D., Excited States in Organic Chemistry, John Wiley & Sons: London, 1975; chap.3. Lower, S. K.; El-Sayed, M. A., Chem. Revs., 1966, 66, 199. Jones, C. R.; Kearns, D. R.; Wing, R. M., J. Chem. Phys., 1973, 58, 1370. McGlyn, S. P.; Azumi, T.; Kinoshita, M., op. cit., p.19. Parker, C. A., Photoluminescence of Solutions, Elsevier: Amsterdam, 1968; p.38. Parker, C. A., op. cit., p.376. Hamanoue, K.; Kajiwara, Y.; Miyake, T.; Nakayama, T.; Hirase, S.; Teranishi, H., Chem. Phys. Lett., 1983, 94, 276. Guilbault, G. G., Practical Fluorescence: Theory, Methods, Techniques, Marcel Dekker: New York, 1973; p.93. Birks, J. B., Photophysics of Aromatic Molecules, Wiley- Interscience: London, 1970; pp.193-195. Miller, J. C.; Meek, J. S.; Strickler, S. J., J. Am. Chem. Soc., 1977, 99, 8175. O'Donnell, C. M.; Harbaugh, K. F.; Winefordner, J. D., Spectrochimigg:Acta, 1972, 29A, 753. McGlyn, S. P.; Azumi, T.; Kinoshita, M., op. cit., pp.261-7. Miller, J. C.; Breakstone, K. U.; Meek, J. S.; Strickler, S. J., J. Am. Chem. Soc., 1977, 99, 1142. Masetti, F.; Mazzucato, U.; Galiazzo, G., J. Lumin., 1971, 4, 8. Guilbault, G. G., op. cit., p.119. Birks, J. B., op. cit., pp.499-504. Goldschmidt, C. R.; Potashnik, R.; Ottolenghi, M., J. Phys. Chem., 1971,75, 1025. Guilbault, G. G., op. cit., p.120. Birks, J. B., op. cit., pp.500-501. Kawaoka, K.; Khan, A. U.; Kearns, D. R., J. Chem. Phys., 1967, 46, 1842. 50 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. Fernandez-Gutierrez, A.; de la Pena, A. M., "Determination of Inorganic Substances by Luminescence Methods", Molecular Luminescence Spectroscopy: Methods and Applications; Part I, Schulman, S. G., ed., John Wiley & Sons: New York, 1985; p.461. Birks, J. B., p. cit., p.206,449. Latas, K. J.; Nishimura, A. M., J. Phys. Chem., 1978, 82, 491. Priestly, E. 8.; Hang, A., J. Chem. Phys., 1968, 49, 622. Almgren, M., Mol. Photochem., 1972, 4, 327. Donkerbroek, J. J.; van Eikema Hommes, N. J. R.; Gooijer, C.; Velthorst, N. H.; Frei, R. W., Chromatagraphia, 1982, 15, 2118. Donkerbroek, J. J.; Gooijer, C.; Frei, R. W.; Velthorst, N. H., in "Abstracts of Papers," 34th Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, Atlantic City, N.J., March 1983; Pittsburgh Conference: Pittsburgh, PA., 1983, Abst. 123. Donkerbroek, J. J.; Veltkamp, A. C.; Gooijer, C.; Velthorst, N. H.; Frei, R. W., Anal. Chem., 1983, 55, 1886. Ichimura, T.; Nahara, K.; Motoshige, R.; Mori, Y., Chem. Phys., 1985, 96, 453. Aizawa, K.; Igarashi, H.;Kaya, K., Chem. Phys., 1977, 23, 273. Abe, H.; Kamei, S.; Mikami, H.; Ito, M., Chem. Phys. Lett., 1984, 109, 217. Bonner, R. E.; DeArmond, M. K.; Wahl, G. H., J. Am. Chem. Soc., 1972, 94, 988. Kasha, M., J. Chem. Phys., 1952, 20, 71. McGlyn, S. P.; Sunseri, R.; Christodouleas, N., J. Chem. Phys., 1962, 37, 1818. Nag-Chaudhuri, J.; Stoessel, L.; McGlyn, S. P., J. Chem. Phys., 1963, 38, 2027. McGlyn, S. P.; Reynolds, M. J.; Daigre, G. W.; Christodouleas, N. D., J. Phys. Chem., 1962, 66, 2499. Gronkiewicz, M.; Kozankiewicz, E.; Prochorow, J., Chem. Phys. Lett., 1976, 38, 325. McGlyn, S. P.; Daigre, J.; Smith, F. J., J. Chem. Phys., 1963, 39, 675. McGlyn, S. P.; Azumi, T.; Kasha, M., J. Chem. Phys., 1964, 40, 507. 51 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. Bazhin, N. M., Chem. Phys. Lett., 1980, 69, 580. Najbar, J.; Birks, J. B.; Hamilton, T. D. S., Chem. Phys., 1977, 23, 281. Yamauchi, S.; Saigusa, H.; Azumi, T., J. Chem. Phys., 1981, 74, 5335. Hood, L. V. S.; Winefordner, J. D., Anal. Chem., 1966, 38, 1922. Lukasiewicz, R. J.; Mousa, J. J.; Winefordner, J. D.; Anal. Chem., 1972, 44, 963. Lukasiewicz, R. J.; Rozynes, P. A.; Sanders, L. B.; Winefordner, J. 0., Anal. Chem., 1972, 44, 237. Aaron, J. J.; Mousa, J. J.; Winefordner, J. D., Talanta, 1973, 20, 279. Boutillier, G. D.; Winefordner, J. D., Anal Chem., 1979, 51, 1384. Ebara, N.; Yajima, Y.; Watanabe, H., Bull. Chem. 52, 2866. Boutillier, G. D.; O'Donnell, C. M.; Rahn, R. O. 1974, 46, 1508. Soc. Jpn., 1979, , Anal. Chem., Boutillier, G. D.; Winefordner, J. 0., Anal. Chem., 1979, 51, 1391. Inman, E. L.; Jurgensen, A.; Winefordner, J. D., 29, 423. Barnes, C. G.; Winefordner, J. D., App. Spect., Additional References Talanta, 1982, 1984, 38, 214. Aaron, J.-J.; Winefordner, J. D., Talanta, 1975, 22, 707. Cline Love, L J.; Weinberger, R., Spectrochim. Acta, 1983, 388, 1421. O'Donnell, C. M.; Winefordner, J. D., Clin. Chem., 1975, 21, 285. 52 CHAPTER IV Room Temperature Phosphorescence: Solid Surface Phosphorescence Basic Mechanism The first of the room temperature phosphorescence (RTP) methods to obtain widespread acceptance were those based on immobilizing the analyte on some polar substrate. The first general reference to this phenomenon is in the work of Schulman and Walling, who noticed the phenomenon by accident while working up a reaction on TLC plates (1). Following this up they obtained emission spectra for salts of various carboxylic and sulfonic acids adsorbed onto a wide variety of polar materials. These spectra were very similar to the familiar cryogenic phosphorescence spectra for these compounds, but they were moderately insensitive to quenching by 02 once the system had been thoroughly dried. The spectra did exhibit a loss of fine structure and the lifetimes were shorter than those at 77 K. However, 2—naphthoic acid salts showed phosphorescence intense enough to be seen by the unaided eye for up to 5 s after removing the exciting light. The phosphorescence was observed only for the salts of ionic compounds on paper, silica, alumina, and anon-Nazcoa. Filter paper gave the best results. It was necessary to deposit the sample in a strongly alkaline solution in order to obtain the RTP signal. The RTP signal 53 disappeared upon subjecting the samples to a humid atmosphere, but it could be regenerated by drying. This combined with the fact that the molecules do not exhibit phosphorescence in the pure crystalline state, indicated that the adsorption process must stretch the molecule out and lock it in a rigid shape that 1) restricts the VR mediated ISC, and 2) limits the ability of the analyte to interact with a colliding 02, These latter processes usually completely quench phosphorescence at room temperature. Although not, strictly speaking, the first reference to solid surface RTP (2,3), Schulman and Walling's work quickly caught the attention of Winefordner's group and initiated a broad research effort. From this group, Paynter, Wellons, and Winefordner reported the first analytical studies (4), finding that limits of detection (LODs) for various naphthalene sulfonic acids on filter paper were competitive with other methods. However, limited linear dynamic ranges (LDRs) were found for many of the compounds. Many of the available filter papers exhibited a phosphorescence background. This background raised the LOD and lowered the LDR of most of the compounds. The ratio of the RTP signal on filter paper to that of the compound in an aqueous cryogenic matrix was found to be less than 1 in all cases, but the ratio was seen to approach unity as the number of ionic sites increased (5). The fact that sulfaguanidine and 5-acetyl- uracil exhibited some RTP was interpreted to indicate that hydrogen bonding was also an active force in immobilizing the analyte. This interpretation was reinforced by the discovery that the RTP signal was proportional to the surface concentration of -OH groups. Smooth, chemically inert surfaces such as polyethylene fiber, borosilicate 54 fiber, or silanized filter paper gave no observable phosphorescence (6). The addition of sugars to the support paper to fill in the physical surface with hydroxyl rich material also provided substantial signal enhancement and lengthened the observed lifetime, indicating that the analyte was not as vibrationally active (7). This conclusion was supported by the finding that the lifetime increased sharply when the sample was prepared from a basic solution causing it to be deposited as the more tightly bound salt. Quenching by water is commensurate with this interpretation as the water would mediate the hydrogen bonding, allowing the molecule a greater degree of vibrational freedom both for intra-molecular VR deactivation and collisional interaction with the ubiquitous triplet quencher 02. The use of sodium citrate as a barrier to moisture and 02 and a glove box for batch drying allowed the sample analysis time to be substantially reduced (8,9). The reproducibility of the physical sample placement was increased as was the sample phosphorescence intensity. However, an increase in the background phosphorescence prevented a reduction of the LCD. The samples were spotted from a neutral solution in contrast to experiments on untreated filter paper which required that the sample be delivered in a strongly basic medium (1,4,5,6,7). The problem of background luminescence has been one of the major drawbacks to solid substrate RTP. There have been experiments where wood pulp, cotton lint, and various filter papers were tried with no (10) or little (11,12) success as was the result of attempts to wash the contaminants from paper (10,13). Attempts to photo-bleach the lignins or hemi—celluloses (10) supposed to be the source of the background have met with varying success (10,14). For the determination of p- 55 aminobenzoic acid (PABA), anion exchange paper was found to be the superior type of paper, due to its ability to perform over a wider pH ' range. This paper also gave a lower LOD for PABA (12). Hurtubise and co-workers also studied the RTP of PABA in some detail using crystalline sodium acetate as the adsorbate (15,16,17). They found that the observation of RTP from this system was specific to the use of ethanol as the solvent because sample preparation apparently required a small amount of surface dissolution (15). No other substrate tried, including other acetate salts, gave rise to RTP. The LOD for the system was about 0.5 ng, and the linear range was greater than 2 orders of magnitude with a usable range of 3 orders of magnitude. The RTP signal was insensitive to the pH of the ethanolic application solution over 5 pH units (16). I The method was insensitive to interference from a wide variety of other vitamins commonly found in commercial formulations. The RTP of derivatives of PABA with non-planar substituents was also found to be drastically reduced. The general conclusion was that there were very specific steric interactions involving 2 sodium acetate molecules and 1 PABA molecule which led to RTP in the system. Once there was more than a monolayer of analyte, the RTP response was seen to decrease, indicating that the response was due to chemisorption directly to the surface, as indicated above (15), rather than to a physical adsorption (16). This was further substantiated by studies with sodium acetate/sodium chloride support matrices (17). Parker et a1. extended this method to filter paper impregnated with sodium acetate for the determination of pteridines with good success (18). 56 Hurtubise's group also experimented with silica gel plates with a small amount of polyacrylic acid (PAA) binder as a substrate for phthalic acid isomers and PABA (19) and nitrogen heterocycles (20). No RTP was observed when the samples were run on certain brands of chromatographic plates which were subsequently found to have no PAA binder (21). They found that the RTP for benzo[f]quinoline was maximized when the silica gel was mixed with this carboxyl rich binder (22), which was attributed to hydrogen bonding between the carboxyl groups and the n electrons of the planar quinoline. The solutions did not require the preparative solution to be strongly basic (19), as was the case with the filter paper RTP (1,4-7). In fact, terephthalic acid gave no RTP when spotted from a sodium hydroxide solution, and the nitrogen heterocycles gave much improved RTP when prepared from a strongly acidic solution (20,22). This indicates again that the existence of an ionized analyte which is strongly bonded to the surface is important to the generation of RTP. This conclusion is in agreement with Ford and Hurtubise' findings (22). When an acidic polymeric binder such as polyacrylic acid was employed, the binder needed to be in the acidic form to provide effective hydrogen bonding. This led to research on PAA itself as a substrate. It was found that PAA itself was ineffective in promoting RTP (21), and that only polyvinyl alcohol, polyacrylamide, and PAA mixed with salts or silica were efficacious (23). Approximately 5% of PAA with silica gel or sodium chloride produced the greatest RTP enhancement (23) after which the signal fell off markedly. Deposition of the sample from 0.1 M HCl routinely gave stronger signals than those from 0.1 M NaOH, and the signal from the PAA-MaCl were generally equal or superior to those from 57 Whatman #1 filter paper under similar circumstances (24). The reproducibility of the PAA based samples was also somewhat better (23) Other studies indicated that methanol was the best solvent due to strong enhancement in spite of a somewhat higher background when depositing M-heterocycles (25,26,27). One study showed that the strongest enhancement for benzo[f]quinoline came from the use of NaCl as the other component of the substrate. For these compounds neutralization of the PAA produced a steady decline in RTP with a break point at about 50% neutralization indicating some conformational change (25). This is also suggested by data for the RTP of hydroxyl aromatics (26). The signal for the compounds is almost completely lost at 25% neutralization, but for 4-phenylphenol and 4,4'-biphenyl it abruptly re- emerges at 75% neutralization, which indicates some change in the surface at this point. An obvious extension of impregnating filter paper with PAA as the substrate gave substantial improvements in the LOD for several compounds relative to the untreated paper (27). External Heavy Atom Effect While Hurtubise et a1. investigated the improvements in the basic substrate (15,19,20,21,23,24,27) and the photophysics of the process (16,17,22,26,28), other investigators applied the external heavy atom (HA) effect to the solid substrate RTP process with good success. Seybold and White showed that the addition of 1.0M MaI gave approximately a 50-fold increase in the RTP from naphthalene sulfonates adsorbed on filter paper, while reducing the fluorescence by about 95% (29,30). Iodide ion was found to increase the RTP of non-chlorinated 58 pesticides (31), pharmaceuticals (32), molecules of general biological importance such as purines (33) on filter paper substrate, and PABA on anion exchange paper (12) and filter paper (34). A particularly striking demonstration of the effect is seen with tryptophan where a 1.0M Mal solution added to the substrate caused a 450-fold increase in RTP (35). The analysis of theophylline containing drugs using I' enhancement was found to be particularly useful since the samples exhibited no interferences under the experimental circumstances (36). The failure in the case of the chlorinated pesticides (31) to show an increased RTP signal in the presence of an external heavy atom environment was attributed to lifetime effects. The lifetimes of these compounds are short at 77 K and even shorter at room temperature. The external HA effect shortens the lifetimes even more. It was believed that at the 500 Hz rotational speed of the rotating can phosphoroscope, the RTP signal had died before the phosphoroscope had rotated the 90° necessary for the phosphorescence to be monitored (31). Parker et al. found that I' quenched RTP on the sodium acetate impregnated filter paper (18). This negative HA effect was also observed in some of the work of Hurtubise's group. Br" gave slightly better RTP performance with PAA than Cl", but I' gave no RTP with naphthols (23). Likewise, spotting the analyte onto the PAA-silica gel plates from 0.1M HBr solution produced a sizable increase in the RTP of benzo[f]quinoline relative to 0.1M HCl (22,24) while the same procedure using 0.1M HI completely quenched the RTP (22). In another anomalous result this compound showed less RTP when the PAA was mixed with Br’ (25). Compounds such as 7-hydroxyl-1,8-naphthyridines showed substantial I' enhancement, 59 but other isomers showed the I- mediated reduction in the signal previously described (37). The variability of the external HA effect is seen in many systems. Jakovljevic reported that Pb+4 and Tl+ produced strong RTP in the anti- microbial drug cinoxacin on filter paper (38). This effect was anion- dependent with the acetate of lead and the fluoride of thallium being most effective. An LOD of 50 pg was cited. Thallium and lead were used in the analysis of such polynuclear aromatic hydrocarbons (PAH's) as fluorene, benzo[a]pyrene, 2,3-benzofluorene, fluoranthene, and dibenzothiophene. Cesium iodide and sodium bromide had lesser, but highly specific effects (39). In another study Ag+ was found to enhance the RTP of the PAH's pyrene and phenanthrene, while I' enhanced that of carbazole (40). The fact that silver ion more strongly affected the lifetimes of phenanthrene, quinine, carbazole, and 7,8-benzoflavone than did the negatively charged iodide ion was interpreted as evidence of a charge transfer interaction in the external heavy atom effect (41, same as ref 3.68). Su and Winefordner found that Tl+ and Ag+ enhanced the RTP of such PAH's as biphenyl, chrysene, coronene, and phenanthrene. Iodide enhanced privine, naphthalene acetic acid, and procaine (11). It was found that I' was superior to Tl+ for the enhancement of RTP in indoles (40). These studies are important in the environmental field as the PAH's are potential carcinogens found in cigarette smoke, tar, polluted water, and urban air pollution, including industrial particulates (42) and automotive particulates (43). The external HA effect due to Hg+2 shows a very specific pKa effect, which is potentially useful in the analysis of bases. There is a 60 tremendous increase in the RTP signal with almost 2 orders of magnitude improvement in the RTP enhancement factor going from phenazine with a pKa of about 1 to isoquinoline with a pKa of approximately 5 (44). This system also shows very strong RTP quenching for PAH's, and holds promise for selective determinations in complex mixtures (44). Incidental Methods There are a few related techniques which have been cited in the literature which have not been extensively developed. The first involves the use of a packed flow-through cell which has been filled with a mixture of paper lint and crushed quartz (45). This system was used to determine a number of PAH's and N-heterocycles. A T1+ HA effect was reported for benzidine. This system appears to have a number of advantages, but there are insufficient data to evaluate it. The other method involves the use of filter paper phosphorescence, but the sample is run liquid nitrogen temperatures (46). This system resulted in approximately a lO-fold increase in the phosphorescence and a similar decrease in the LOD for most of the compounds tried. Since the background signal limits the LOD in most filter paper work, this represents a major improvement in the technique. This is a relatively recent (1983) paper, and the group has developed a more convenient conduction cooling device for cryogenic work (47,48) so more of low temperature solid-substrate RTP research may be seen shortly. 61 Applications Solid substrate RTP has been used in a number of cases to characterize samples of a more routine analytical nature. It has been used to determine PABA in urine (34), as previously discussed with a limit of detection of 0.67 mg/L of urine. Fluorescamine was examined recently (1985) as a possible derivatizing agent for primary amines, particularly those amino acids which give an RTP response (49). In I" sensitized RTP, primary amines gave LOD's from 0.8 ng to 2.9 ng when the samples were derivatized in a separate procedure, and the filter paper RTP was done in the conventional manner. When the derivatization was done on the paper, the LOD for histidine fell from 2.9 ng to 1.3 ng showing potential for dramatic improvement. Linear dynamic ranges varied from about 70 to more than 200. The secondary amine proline showed no response as expected. This method shows promise as a post-column derivatization detection method for chromatography. The pesticide benomyl was determined at the part per million level on grapes and apples by extracting it and converting it to 2-aminobenzimidazole and spotting it on filter paper with OB'/I' for HA enhancement (50). By using selective excitation and the HA effect, binary and ternary mixtures of N-heterocycles and pyrene were quantitated to the 2.5 ng level (51). Vo-Dinh and Hooyman were able to analyze a synthetic mixture of PAH's at the nanogram level with 10% accuracy (52), by carefully choosing from a wide variety of HA enhancers including NaI, NaBr, CsI, AgMO3, Pb(OAc)2, and LiClO4. The same techniques were useful in the analysis of carboxylic pesticides (53) and carboxylic priority pollutants (54) with similar accuracy. 62 Using different HA enhancers and the synchronous scanning technique, Vo-Dinh and co-workers have analyzed PAH's extracted from 1) workplace air by XAD-2 resin and then re-extracted by liquid chromatography (39), and 2) in the synthetic fuel Synthoil (55). In synchronous scanning the excitation and the emission monochromators are scanned in unison separated by some fixed wavelength or energy (56) to increase the selectivity of the spectral analysis. The results for these analyses were very similar to those obtained using synchronous fluorometry. These workers have used the selective HA perturbation technique to analyze mixtures of N-heterocycles and PAH's (44) and they have used the selective HA effect coupled with synchronous scanning and second-derivative techniques (57) to directly estimate benzoquinolines in a coal tar fraction (58). This second-derivative method has also been applied to PAH's in synthoil (56). To improve the precision and accuracy of RTP determinations, Warren and co-workers have suggested that the classical internal standard and standard additions methods be employed (59). The RTP spectra for the analyte, the analyte with the internal standard, and the analyte with the internal standard and added standard analyte are collected. These spectra are obtained at each excitation wavelength applicable to the absorption spectrum of the system given the bandpass of the excitation monochromator. This set of RTP spectra are assembled to give the excitation-emission matrix spectrum of the system. Factor analysis is then used to strip the spectra of the components from the composite spectrum mathematically, and the analyte is ratioed to the internal standard to give a normalized intensity. The study indicated 63 that relative standard deviations of 1%-3% are possible with relative errors for salicylic acid in serum of about 5%. Instrumentation The first solid substrate RTP signals were obtained by inserting the filter paper strip into the empty Dewar flask ordinarily used for cryogenic phosphorescence and manually rotating the strip to give the optimum signal (1). Subsequently a tip was designed to hold the small (1/4") filter paper disks and replace the optical finger of the Dewar flask (4). A simple modification of the phosphoroscope allowed much easier sample introduction (60). A continuous filter paper tape was spotted, passed through a dryer, and then over the opening in the phosphoroscope where the optical tip of the Dewar flask was ordinarily inserted. A small rotating mirror was mounted at 45°. When it passed the excitation slit, light was reflected upward to an analytical spot directly over the opening. As the mirror rotated this light was cut off and the phosphorescence signal began to illuminate the mirror. When the mirror rotated past the emission slit the RTP signal was measured. It was difficult to measure lifetimes with the rotating mirror apparatus as the duty cycle was fixed and the delay time was limited by the range of rotational speeds available through the mirror spinner. Yen(Bower) et a1. (61) reported a simple phosphoroscope with a variable time base for sampling the photomultiplier output after a chopper had blocked the excitation source. To use this improved phosphoroscope for RTP, these researchers designed a simple sample holder canted at 45° to both the excitation and emission slits (62). The filter paper tape 64 previously described (60) was pulled through the holder vertically and the photomultiplier output was sampled at various time intervals after the excitation beam had been terminated. This system was applicable to both time-resolved and spectral analysis. Goeringer and Pardue (63) described the application of an imaging vidicon camera in conjunction with a flashlamp to obtain time-resolved RTP spectra. They were able to collect a complete spectrum in 8 ms and they were able to resolve three component mixtures into the constituent spectra. However, no data as to the LOD of this system were presented. Ward et al. (64) presented an improved version of the device described by Paynter et al. (4). It consisted of a pair of slotted guides for holding a long bar. It was secured by an indent and spring loaded ball which held the bar in any of several vertical positions. Each of these positions corresponded to positioning a different 1/4" filter paper disk at a 45° angle to both the excitation and emission beams in a conventional rotating can phosphoroscope. No improvement in the relative standard deviations (RSD's) of sets of measurements was realized, but the means of handling batches of samples improved throughput. Scharf and coworkers (65) described a micrometer adjustable modification to this system which served to improve the observed RTP intensity. They found micro-positioning to be essential because a variation of 0.400 mm could cause a 50% change in the signal. Su's group (53) reported a circular carousel type device that can hold up to 20 filter paper disks. It had a co-axial high-speed rotating disk chopper which allowed time resolution between 0.6 and 3 ms. It had the added feature of a liquid N2 reservoir under the disk allowing the option for low temperature solid surface phosphorescence studies. 65 It was reported by Vo-Dinh et a1. (60) and by Yen-Bower and Winefordner (62) that there appeared to be non—uniformity of the analyte on the spot, and it was speculated that these inhomogeneities might be the source of the usually poor reproducibility of RTP experiments. McAleese and Dunlap (66) reported a modification of the optics of their instrument to produce a larger image of the arc source on the analyte sample. In concert with smaller 1/8" support disks they reported intensity increases of 600% for PABA and 300% for 3-biphenylcarboxylic acid when going from the 1/4" disk to the 1/8" one and opening the illumination slit completely. The RSD for PABA showed 85% reduction to 1.3% and for the carboxylic acid 80% reduction yielded an RSD of 1%. At this juncture solid substrate RTP is recognized to be an analytical method with a good deal of potential. The external HA effect offers some very useful advantages by its selectivity. However, the irregularity of the external HA effect forces the experimenter to find a highly individualized method for a given well-defined problem. Such problems can be found in quality control laboratories. One application has been described (67) for brodifacoum bacteria in dog food. The recent improvements in instrumentation and methodology coupled with the liberation of phosphorescence from liquid nitrogen and the fragile quartz Dewar flask should portend continued development of the area. 66 10. 11. 12. 13. 14. 15. 16. 17. CHAPTER IV References Schulman, E. M.; Walling, C., J. Phys Chem., 1973, 77, 902. LLoyd, J. B. F.; Miller, J. N., Talanta, 1980, 26, 180. Roth, M., J. Chromatog,, 1967, 30, 276. Paynter, R. A.; Wellons, S. L.; Winefordner, J. D., Anal. Chem., 1974, 46, 736. Wellons, S. L.; Paynter, R. A.; Winefordner, J. D., Spectrochim. Acta, 1974, 30A, 2133. Schulman, E. M.; Parker, R. T., J. Phys. Chem., 1977, 81, 1932. Niday, G. J.; Seybold, P. G., Anal. Chem., 1977, 50, 1577. McAleese, D. L.; Dunlap, R. 8., Anal. Chim. Acta, 1984, 162, 431. McAleese, D. L.; Freedlander, R. S.; Dunlap, R. B., Anal. Chem., 1980, 52, 2443. Bateh, P. P.; Winefordner, J. D., Talanta, 1982, 29, 713. Su, S. Y.; Winefordner, J. D., Can. J. Spect., 1983, 28, 21. Karnes, H. T.; Schulman, S. G.; Winefordner, J. D., Anal. Chim. Acta, 1984, 164, 257. Ward, J. L.; Bower, E. L. Y.; Winefordner, J. D., Talanta, 1981, 28, 119. McAleese, D. L.; Dunlap, R. B., Anal. Chem., 1984, 56, 600. von Wandruszka, R. M.; Hurtubise, R. J., Anal. Chem., 1976, 48, 1784. von Wandruszka, R. M.; Hurtubise, R. J., Anal. Chem., 1977, 49, 2164. Senthilnathan, V. P.; Hurtubise, R. J., Anal. Chem., 1985, 57, 1227. 67 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. '29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. Parker, R. T.; Freedlander, R. S.; Schulman, E. M.; Dunlap, R. B., Anal. Chem., 1979, 51, 1921. Ford, C. D.; Hurtubise, R. J., Anal. Chem., 1978, 50, 610. Ford, C. D.; Hurtubise, R. J., Anal. Chem., 1979, 51, 659. Hurtubise, R. J.; Smith, G. A., Anal. Chim. Acta, 1982, 139, 315. Ford, C. D.; Hurtubise, R. J., Anal. Chem., 1980, 52, 656. Dalterio, R. A.; Hurtubise, R. J., Anal. Chem., 1982, 54, 224. Ramasamy, S. M.; Hurtubise, R. J., Anal. Chem. 1982, 54, 1644. Ramasamy, S. M.; Hurtubise, R. J., Anal. Chem., 1982, 54, 2477. Dalterio, R. A.; Hurtubise, R. J., Anal. Chem., 1983, 55, 1084. Senthilnathan, V. P.; Ramasamy, S. M.; Hurtubise, R. J., Anal. Chim. Acta, 1984, 157, 203. Dalterio, R. A.; Hurtubise, R. J., Anal. Chem., 1984, 56, 336. Seybold, P. G.; White, W., Anal. Chem., 1975, 47, 1199. White, W.; Seybold, P. G., J. Phys. Chem., 1977, 81, 2035. Aaron, J.-J.; Winefordner, J. D., Analusis, 1979, 7, 168. Bower, E. L. Y.; Winefordner, J. D., Anal. Chim. Acta, 1978, 101, 319. Vo-Dinh, T.; Yen(Bower), E. L.; Winefordner, J. D., Anal. Chem., 1976, 48, 1186. Karnes, H. T.; Bateh, R. P.; Winefordner, J. D.; Schulman, S. G., Clin. Chem., 1984, 30, 1565. Meyers, M. L.; Seybold, P. G., Anal. Chem., 1979, 51, 1609. Bateh, R. P.; Winefordner, J. D., Anal. Lett., 1982, 15(b4), 373. de Lima, C. G.; de Nicola, E. M., Anal. Chem., 1978, 50, 1658. Jakovljevic, I. M., Anal. Chem., 1977, 49, 2048. Vo-Dinh, T.; Gammage, R. B.; Martinez, P. R., Anal. Chem., 1981, 53, 253. Aaron, J.-J.; Andino, M.; Winefordner, J. D., Anal. Chim. Acta, 1984, 160, 171. 68 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. Boutillier, G. D.; Winefordner, J. D., Anal. Chem., 1979, 51, 1391. Vo-Dinh, T.; Yen(Bower), E. L.; Winefordner, J. D., Talanta, 1977, 24, 146. Cadle, S. H.; Nebel, G. J., "Control of Automotive Emissions", Introduction to Environmental Toxicology, Guthrie, F. E.; Perry, J. J., eds; Elsevier North Holland: New York, 1980; 428. Abbott, D. W.; Vo-Dinh, T., Anal. Chem., 1985, 57, 41. Lloyd, J. B. F., Analy§t(London), 1978, 103, 775. McCall, S. L.; Winefordner, J. 0., Anal. Chem., 1983, 55, 391. Ward, R. P.; Bateh, R. P.; Winefordner, J. D., Appl. Spect., 1980, 34, 15. Ward, J. L.; Walden, G. L.; Bateh, R. P.; Winefordner, J. D., Appl. Spect., 1980, 34, 348. Long, W. L.; Su, S. Y., Anal. Lett., 1985, 18(85), 543. Vannelli, J. J.; Schulman, E. M., Anal. Chem., 1984, 56, 1033. Senthilnathan, V. P.; Hurtubise, R. J., Anal. Chem., 1984, 56, 913. Vo-Dinh, T.; Hooyman, J. R., Anal. Chem., 1979, 51, 1915. Su, S. Y.; Asafu-adjaye, E.(B.); Ocak, S., Analyst(Londoh), 1984, 109, 1019. Asafu-Adjaye, E. B.; Jung, 1. Y.; Su, S. Y., Anal. Chem., 1985, 57, 904. Vo-Dinh, T.; Gammage, R. B.; Martinez, P. R., Anal. Chim. Acta, 1980, 118, 313. Vo—Dinh, T.; Gammage, R. B., Anal. Chem., 1978, 50, 2054. Vo-Dinh, T.; Gammage, R. B., Anal. Chim. Acta, 1979, 107, 261. Vo-Dinh, T.; Miller, G. H.; Abbott, D. W.; Moody, R. L.; Ma, C. Y.; Ho, C.-H., Anal. Chim. Acta, 1985, 175, 181. Warren, M. W., II; Avery, J. P.; Malmstadt, H. V., Anal. Chem., 1982, 54, 1853. Vo-Dinh, T.; Walden, G. L.; Winefordner, J. 0., Anal. Chem., 1977, 49, 1126. 69 61. 62. 63. 64. 65. 66. 67. Texts: Yen(Bower), E. L.; Boutilier, G. D.; Winefordner, J. D., Can. J. Spect., 1977, 22, 120. Yen-Bower, E. L.; Winefordner, J. D., Appl. Spect., 1979, 33, 9. Goeringer, D. E.; Pardue, H. L., Anal. Chem., 1979, 51, 1054. Ward, J. L.; Bateh, R. P.; Winefordner, J. D., Analyst(London), 1982, 107, 335. Scharf, G.; Smith, B. W.; Winefordner, J. D., Anal. Chem., 1985, 57, 1230. McAleese, D. L.; Dunlap, R. B., Anal. Chem., 1984, 56, 836. Freedlander, R. S.; Parker, J. J., in "Abstract of Papers," 34th Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, Atlantic City, N. J., March, 1983; Pittsburgh Conference: Pittsburgh, Pa., 1983; Abstr. 293. Additional References Hurtubise, R. J., Solid Surface Luminescence Analysis: Theory, Instrumentation, Applications, Marcel Dekker: New York, 1981. Vo-Dinh, T.; Room Temperature Phosphorimetry for Chemical Analysis, Wiley-Interscience: New York, 1984. Reviews: Aaron, J.-J.; Winefordner, J. D., Analusis, 1982, 10, 299. Hurtubise, R. J., Anal. Chem., 1983, 55, 669A. Miller, J. N., Trends Anal. Chem., 1981, 1, 31. Parker, R. T.; Freedlander, R. S.; Dunlap, R. 8., Anal. Chim. Acta, 1980, 119, 189. Parker, R. T.; Freedlander, R. S.; Dunlap, R. B., Anal. Chim. Acta, 1980, 120, 1. Vo-Dinh, T., Appl. Spect., 1977, 13, 261. 70 Ward, J. L.; Walden, G. L.; Winefordner, J. D., Talanta, 1981, 28, 201. Yen-Bower, E. L.; Ward, J. L.; Walden, G.; Winefordner, J. D., Talanta, 1980, 27, 380. 71 CHAPTER V Room.Temperature Phosphorescence: Solution Phosphorescence Introduction With the development of phosphorescence spectrometry in the early 1970's, many new areas were explored with this technique. Biochemists were interested in investigating protein conformation by these methods. They surmised that the dense folded structure of the protein might isolate the phosphorescent moiety from the deactivational effects of 02. In this vein Saviotti and Galley first reported tryptophan phosphorescence in a pair of enzymes at room temperature (1). From these results and their own studies of surfactant micelles as membrane emulators, Kalyanasundaram, Grieser, and Thomas reported the first observation of RTP in surfactant micelles (2). At this time biochemists were also investigating enzyme models based on the cyclic polysaccharides called cyclodextrins. The cyclodextrins mimiced certain enzyme functions by surrounding a portion of the target molecule and restricting its movement. In the course of these works, the spectroscopic behavior of the cyclodextrin-based systems were investigated, and it was found that the cyclodextrins shielded a portion of the included target molecule from the quenching effects of 02. This led to successful efforts to observe cyclodextrin- 72 mediated RTP. This chapter focuses on the literature of the solution- based RTP methods, of which, these two methods are the most important examples. Micelle-Stabilized RTP In this section, a short review of the properties of surface- active agents (surfactants) is presented first, in order to lead into a discussion of micelle-stabilized RTP (MS-RTP). A surfactant is a compound with a non-polar, hydrophobic backbone with a localized, highly polar, hydrophilic region. The prototypical compound used for MS-RTP is the sodium dodecyl sulfate molecule which is also known widely as sodium lauryl sulfate. It is a linear twelve-carbon molecule with an ionic sulfate group attached to one end. The ionic portion interacts with water molecules to solubilize the compound. At low concentrations the surfactant solute exists as a monomer, but due to its extended organic component, it distorts the structure of the water, thus increasing the free energy of the solution. At this stage the surfactant molecules tend to concentrate at the solution interfaces. When the concentration reaches a specific value, called the critical micelle concentration (CMC), there is a large entropic component favoring aggregation of the non—polar moieties with the polar heads protruding into the solvent. In water these aggregates generally consist of from 25 to 400 molecules. For sodium dodecyl sulfate ($08) the aggregation number has been reported to be 71. Apparently there is an increase in the freedom of association of both the solvent and the organic molecules (3). 73 There are several different types of micelles such as tubular micelles, lamellar sheet micelles with layers of water separating the sheets, and inverted micelles in organic solvents. In the latter, the polar heads are pointed inward around an aqueous nucleus. However, for MS-RTP the primary vehicle is the roughly spherical micelle of SDS. Figure 6 shows a simplified representation of two of these structures. Kalyanasundaram et al. (2) reported MS-RTP for several fused aromatics and biphenyls after degassing by bubbling oxygen-free nitrogen gas through the samples for 30 min. They reported phosphorescence in several ionic micelle systems with lifetimes in the millisecond range. They also found that cupric ion reduced both the lifetime and the yield of phosphorescence, while thallous ion decreased the lifetime while increasing the yield. This was especially apparent in weakly emitting parent compounds which have long lifetimes and are more subject to quenching. In the halogenated molecules the effect was less pronounced, which indicates that there is a distinct SOC effect operating to increase the radiative coupling of the T1 state with the ground state. Humphrey-Baker and co-workers (4) and Turro's group (5) reported similar results shortly thereafter. Humphrey-Baker reported that the silver salt of dodecyl sulfate produced dramatic increases in the phosphorescence of pyrene and naphthalene indicating a strong external heavy atom effect. The Turro reference (5) also reported RTP in deoxygenated solutions of these compounds in organic solvents such as acetonitrile, especially in the presence of the heavy atom perturber ethylene dibromide. However, this group did not follow this research extensively, so it does not seem to have been a general phenomenon. 74 .oaoooHOE ecu uo :ofiuuom vacoH on» nucomoumou odouao may one oaooeaoa acouomuuom 6 cu muowoe oacomuo on» acououmou mocaa >>oz one .oaaoofie pouuo>c« on An poo oHHoOHE 0 As no ocoHucucomoumou Hmoumwocoo peduwamEHm .w ouzofim Leg fish g a as 75 Almgren, Grieser and Thomas (6) used the MS-RTP of biphenyl, naphthalene, and 1-methylnaphthalene to study the dynamics of solute exchange between micelles and the bulk solvent. These experiments set a lower limit for the exit rate of these compounds from SDS micelles at 5x104 s'l. For 1-bromonaphthalene, they found the exit rate to be 2.5x104 5'1, and the diffusion-controlled entrance rate to be 5.8x109 5’ 1. This value they considered to be a representative value for all of the other compounds. Turro and Aikawa (7) reported MS-RTP and delayed fluorescence for 1-chloronaphthalene. Their analysis showed the MS-RTP decreased and the delayed fluorescence increased with an increase in the mean occupancy number of the luminophore in the micelle. This indicated that the delayed fluorescence arose from intra-micellar triplet-triplet annihilation. Their exit rate data showed a strong correlation with that of Almgren et al. (6). Nitrite ion was also demonstrated to be an efficient MS-RTP quencher. In 1980 Cline Love's group published the first in a series of papers dealing with analytical uses of MS-RTP (8). For biphenyl, pyrene, and naphthalene in the uM concentration range they found 2 orders of linear dynamic range (LDR) with a severe inner filter fall off of the signal at about 10'4 M. T1+ was found to be superior to Ag+ as a heavy atom ion by a factor of 50%, with the maximum enhancement occurring at a ratio of TlI/Na+ equal to 0.2 in a total dodecyl sulfate concentration of 0.15 M. The average precision of their measurements was 6%. There was also a rapid decrease in the measured intensity as the temperature was varied from 20° to 35° C. It varied from 40% for biphenyl with a silver 76 ion heavy atom, to 80 % for the same system with the thallous ion. Naphthalene exhibited a 60% decrease in both of the solvent systems. Cline Love's group subsequently reported MS-RTP data for a variety of substituted compounds based on the previous parent molecules (9). They found limits of detection (LODs) in the range of 0.1 - 10 uM using the thallous ion as an external heavy atom perturber. These data compared favorably with cryogenic phosphorescence and solid surface RTP for these compounds. They also found that, generally, 75% - 99% of the fluorescence was quenched. Lifetimes were found to range from 0.25 to 1 ms. It was also noted that carbonyl compounds showed weak to non- existent MS-RTP due perhaps to poor inclusion of these relatively polar compounds into the isolating interior of the micelles. Cline Love et al. (10) determined that the primary contributors to the observed lifetime were the rate constants for exit of the molecule from the micelle and intra-micelle deactivation under conditions of constant surfactant concentration. The re-entry rate of the molecule in SDS solutions in the range 2x10"2 to 2x10"3 M is such that extra- micellar deactivation is negligible. They resolved the lifetimes of pyrene and naphthalene in a 1:2 molar concentration ratio with an accuracy of 25% using a Guggenheim fit to the logarithm of the data. These are very good data considering that the single-component lifetimes were in the ratio of only 2:1 (pyrene, 0.93 ms; naphthalene, 0.45 ms). Oxygen quenching of MS-RTP was used to provide background correction for interfering fluorescence (11). Phenanthrene in the presence of 2-naphthaldehyde shows a weak MS-RTP peak on the side of the fluorescence band of the 2-naphthaldehyde. By recording the MS-RTP spectrum and then subtracting the residual spectrum obtained upon 77 aeration, a limited linear working range was established. Similar methods were used to extract the MS-RTP spectrum from the total luminescence spectrum of n-(2-chloroethyl)carbazole without having to use a source chopper (11). Subsequent studies of this and related carbazoles using this MS-RTP quenching methodology showed LOD's in the range from 0.5 - 100 nM. Generally 90+% of the fluorescence was quenched in the presence of 30% thallous counter-ion in the surfactant (12). The RTP of thioketones in de-oxygenated acetonitrile under the conditions outlined by Turro (5) was found to decease in intensity very rapidly in the 0.1 mM concentration range (13). In solutions containing SDS or Triton X-100 the decrease in the signal was much more gradual. This indicated that the diffusion-controlled triplet-triplet annihilation process was mediated by the isolation of the phosphor in the micelle (13). The fact that the exit rate of the molecule from the micelle is much less than the diffusion controlled entrance rate causes the molecule to remain in an isolated environment longer. This results in a higher probability of phosphorescence than if the molecule is in a homogeneous solution where the higher diffusion rate constant gives a higher probability of entering the annihilation radius of another excited molecule. Mediation of triplet-triplet annihilation is limited to concentrations where multiple occupancy of a micelle is negligible. At concentrations yielding multiple occupancy, the molecules is held in closer proximdty, and the probability of annihilation increases. In contrast to the earlier findings that ketones did not phosphoresce in TII/Na+ dodecyl sulfate (9), Leigh and Scaiano found that benzophenone and related compounds exhibited RTP in micelles and in solutions of acetonitrile and alkane solvents (14). They used the 78 selective 02 quenching method (12) to obtain the fluorescence background which they subtracted by an unspecified method. They used a pulsed xenon source to obtain their results, and suggested that earlier works had been subject to photo-bleaching of the solution because a high-powered xenon arc lamp was used. To demonstrate, they irradiated a sample for 1 min with a 150 W xenon lamp and found that the signal had been eliminated. Woods and Cline Love (15) found that the pyridinic molecules phenazine and acridine showed greater RTP enhancement with 50:50 AgI/Na+ dodecyl sulfate (0.08 M) than with the usual thallium solution. This is similar to an effect observed in solid surface RTP (Reference 41; Chapter IV). For this type of compound the n electron density is localized on the nitrogen. Thus, there is a strong donor-acceptor relationship between the silver and the nitrogen portion of the molecule, and the SOC is enhanced more to the level of an internal heavy atom. The weaker fluorescence quenching reflects the higher degree of SOC provided by the nitrogen lone pair. Since the S1 state is already highly perturbed toward the T1 state an external heavy atom effect will be less pronounced. Cline Love's group also demonstrated the use of synchronous scanning MS-RTP for the simultaneous determination of mixtures. Second derivative treatment of the data was demonstrated to improve resolution. They found that at total analyte concentrations exceeding 0.5x10"5 M, triplet transfer causes serious errors. This is due to the fact that in spite of the isolating nature of micelles there is enough diffusion at these concentrations for significant analyte-analyte encounters suitable for such transfer (16). 79 An obvious problem to anyone who blew soap bubbles as a child is the problem of deoxygenation of micelle containing solutions. They readily form bubbles that fill the sample cell and flow out the top spilling the solution. Early methods relied on careful control of the purge gas flow rate, which introduces a considerable time element into the experiment. Kalyanasundaram et al. specified 30 min. of degassing(2). One solution to the problem was to construct the cell with a long neck with a ground glass joint. A 60° funnel with a ground glass joint was inserted into the tube and oxygen-free nitrogen was bubbled through the solution via a long teflon tube. The resulting bubbles flowed up the tube and broke as they spread into the funnel, with the teflon tube acting as a wick to return the solution to the cell. This resulted in a 60%-75% time reduction-for de-aeration (17). Recently, a chemical de-oxygenation scheme has been reported in the literature (18). Sulfite anion reacts with molecular oxygen to form sulfate. The fact that the sulfite ion has the same charge as the dodecyl sulfate micelle caused the formation of a layer depleted of sulfite around the micelle. Thus, there was an induction time for oxygen scavenging from this layer. Since this is the region where most quenching occurs, the solution was allowed to sit for about 10 min before MS-RTP was measured. The ease of handling of these solutions relative to bubbling nitrogen through a soap solution should make this method worth further investigation. Due to their amphiphilic nature the application of micelles in chemistry is rapidly increasing. It is natural that researchers would try them as modifiers in high performance chromatography (HPLC) solvent systems (19). Several studies have been presented to describe the 3- 80 phase partitioning between the stationary phase, the bulk mobile phase, and the micelles (20,21,22,23). The use of this mobile phase leads to the idea that MS-RTP may be done directly on the HPLC eluent. Armstrong and co-workers first reported using this technique to separate and detect naphthalene, pyrene, and biphenyl (24). Cline Love's group later reported similar results for these compounds with LODs in the 5-20 ng range and linear dynamic ranges (LDRs) of 3 orders of magnitude (25). There are several addenda to the discussion of MS-RTP which provide more insight to the photo-physical possibilities of micelles. Fluorophores in micelles have been included in micelles and have demonstrated dramatic fluorescence enhancements relative to homogeneous solutions (24,26,27). Detection limits for biological quinones, vitamins, and amino acids have been improved by factors ranging from 2- 10 (27). Non-rigid polar compounds with phenyl groups directly attached have shown strong increases in their fluorescence signals (26). This was attributed to being surface adsorbed on the micelle, which would give the molecule an increased rigidity and decrease the number of degrees of freedom available for radiationless deactivation. The fluorescence lifetime of naphthalene was found to increase greatly when incorporated into a micelle (28). The addition of a quenching counter ion then revealed a fast decaying component due to the naphthalene being partitioned into the aqueous phase. This offered a means to study the partitioning of naphthalene between water and the micelles. Bolt and Turro (29) synthesized detergents tagged with phosphorescent species. These compounds have been used to study the micellar entrance and exit rates of aqueous soluble quenchers such as 81 cobalt(III) hexamine. These studies were used to examine the micelle lifetime. Turro and co-workers (30) also reported the synthesis of perfluorinated surfactants, and their studies of MS-RTP of Ru(bpy)32+ indicated an aggregation number of 7, a very small "mini" micelle. Cyclodextrin Enhanced MS-RTP Biochemists were the first to study the compounds known as cyclodextrins (CD's). These molecules result from the degradation of starches by the enzyme cyclodextrin glycosyl transferase, a type of amylase, found in bacilli. Starches are polysaccharides formed into left-handed helices containing 6 glucose units per turn. The glycosyl transferases detach a turn from the starch helix and cyclize it. The result is a toroidal polysaccharide containing 6-12 glucose units. The most common forms are alpha, beta, and gamma CD having 6, 7, or 8 glucose monomers linked together. The hollow interior, varying in diameter from 5 A for the alpha-CD to 8 A for the gamma-CD, is quite hydrophobic and gives these compounds peculiar properties in aqueous solutions. Figure 7 shows a representation of a CD. The diameter of the hole in these doughnut-like molecules is small enough that water molecules that enter have a very constrained, high energy, low entropy configuration. When small organic molecules are placed into a solution of these compounds, they can enter the cavity, reducing their own relatively high energy of solution and, by dispelling the water into the bulk phase, lower its energy, and increase its entropy. This is analogous to the formation of micelles where the water is expelled from the micelle increasing the entropy of the water. The 82 Figure 7. A simplified structural representation of a cyclodextrin molecule emphasizing the central cavity. 83 fact that the formation of CD complexes with organic "guest" molecules has been found to depend on the size of the guest and that the association constant for the complex is independent of its chemical nature has been cited as evidence (31). Early research was often concerned with using the hollow cavity of CD's to model enzymatic cleavage sites, and from there spread to the general field of using them as catalysts. A few representative examples are presented. Cramer et al. studied the use of CD's to catalyze the fission of pyrophosphates as an enzyme model (32) and the effect of dye- substitution patterns on inclusion kinetics and thermodynamics (33). Bender and co-workers (34,35,36) examined the use of alpha-CD on the cleavage of phenyl esters. Tabushi's group at Kyoto University (37) has examined a number of reduction and hydrolysis reactions using CD's. They have also examined "capping" one end of the CD with one or more functional groups, to increase the enzymatic type of specificity of these reagents. This same group has reported one step syntheses of vitamin K analogs using CD's (38). Breslow has described accelerated Dials-Alder condensations in CD's (39) and selective chlorination of anisole in alpha-CD solutions (40). Early spectroscopic studies of compounds included in the cyclodextrin cavity revealed enhanced circular dichroism spectra indicating that substitution patterns could radically change the orientation of the guest molecule in the cavity (41). At the same time, the fluorescence spectra of the included compounds were observed to be significantly enhanced (42,43,44), as was seen in micelles. The development of a naphthylacetate excimer band in solutions of gamma-CD 84 and sodium naphthylacetate indicated that it was possible to generate 2:1 guest/host complexes (45,46). Cline Love's group has used this enhanced fluorescence to examine CD's as an analytical tool (48). They have found that the size constraints of the CD cavity can be used to discriminate fluorometrically against polychlorinated biphenyls, as they are excluded from the cavity (47). They also found that the LOD' of various fluorometrically determined drugs may be lowered by using CD inclusion complexes. Fluorescence has been used to study the packing of the guest molecule. Naphthols are tightly packed in the beta-CD, while rather loosely included in the gamma-CD. The fact that the dissociation constant of the naphthol was sharply reduced indicated that the hydroxyl moiety was inside the cavity (49). Cline Love and coworkers have also shown the existence of two different inclusion complexes for coumarins. One complex with the phenyl portion of the molecule protected from the bulk solvent showed fluorescence enhancement while the other, having this portion protruding into the solvent phase, exhibited fluorescence quenching (50). A study by Turro's group of central importance to phosphorimetry involved the fluorescence of inclusion complexes of 1,3-bichromophoric propanes (51). When the size relationship between the CD and the guest caused the 2 ring systems to be squeezed into an eclipsed configuration intramolecular excimer emission was observed. Additionally, the emission was insensitive to oxygen quenching, even under high pressure. This led him to investigate whether CD's might be used to shield molecules from this ubiquitous phosphorescence quencher so that RTP could be observed. 85 The first reports of cyclodextrin-enhanced RTP (CD-RTP) appeared in 1982. Turro's laboratories reported that CD-RTP of 1-chloro and 1- bromonaphthalene were observed in nitrogen purged CD solutions (52). They showed that the emission was substantially shielded from nitrite quenching, and lifetimes were enhanced by the addition of 10% acetonitrile. A short time later they reported the CD-RTP spectrum of the 1-(4-bromo)naphthoyl group in aqueous gamma-CD solutions under 1 atm of oxygen (53). These compounds were found to form 2 types of complexes. One complex, which was insensitive to oxygen quenching, had a lifetime of 3.5 ms, and the other, which was completely quenched, had a lifetime of 600 us. Cline Love's group has followed this lead with an extensive evaluation of CD-RTP for analytical applications (54,55,56). They found that the CD-RTP signal for numerous polynuclear aromatic compounds (PNAs). They found that the strongest signal was obtained when the analyte was included in beta-CD in a solution containing the external heavy atom ethylene dibromide (EDB) in large excess. Apparently the inclusion complex contains one or more EDB molecules in addition to the analyte. They also found that the system was only moderately insensitive to oxygen quenching. They demonstrated an LDR over 4 orders of magnitude. Interference from the CD and the EDB necessitated background correction below 1 nM, but by using localized calibration curves derived through the correction technique, they were able to extend their LOD's to the pH range. The rigidity of the CD cavity was found to inhibit the relaxation of the biphenyl molecule to the planar shape necessary for efficient phosphorescence. 86 Cline Love and coworkers reported CD-RTP for 2- and 3-ring nitrogen heterocycles and bridged biphenyl heterocycles such as dibenzofuran (58). In this case the bridging atom forces the biphenyl system into planarity so that the phosphorescence is favored. Quenching by oxygen was observed to be related to the solubility of the compound in the bulk solvent (57). Efforts to introduce the heavy atom into the CD torus by bromination of the primary hydroxyls of the CD resulted in lower phosphorescence intensities than the previously described system using EDB. A combination of the CD-RTP and MS-RTP techniques by Cline Love's group indicated that the system phenanthrene-beta-CD goes through three states (59). The first state, at low surfactant concentrations, shows RTP that is primarily CD mediated. At higher concentrations, but still lower than the CMC, showed approximately a 50% increase in the intensity of the CD-RTP. At concentrations above the CMC, which is raised to 0.1 M by the presence of the CD, MS-RTP is observed. This is characterized by a 7 nm red-shift in the spectrum and a loss of fine structure. Biacetyl (2,3-butanedione) is a compound with a low energy T1 state, which Gooijer and co-workers have studied extensively (60,61; References 44,45,and 46 in Chapter III). By exciting higher energy phosphors in oxygen-purged solutions of biacetyl, the phosphorescence of the biacetyl is observed in fluid solutions. The low-lying triplet of the biacetyl is populated by triplet-triplet transfer from the analyte. The narrow excitation spectral band (20 nm FWHM) of biacetyl at 250 nm precluded direct excitation when most analytes were excitation optimized. Typical LOD's were in the range of 10 nM for benzophenones and biphenyls. Cline Love's group reported efforts to improve these 87 results by incorporating micelles and CD's into the biacetyl solutions (62). Their findings showed that both of these modifiers improved the LOD's for various PNA's. Beta-CD's generally improved the LOD's by an order of magnitude, while micelles improved them 2-3 orders of magnitude. These effects were attributed to the ability of these additives to organize the solution, bringing the analyte and the biacetyl closer together to enhance the triplet-triplet exchange. This technique indicates a good potential for PNA screening since the observed signal is solely that of the biacetyl, but intensity is proportional to the total concentration of all PNA's participating in the triplet transfer. In a comparative study of CD-RTP, MS-RTP, and sensitized phosphorescence of biacetyl, several generalities were pointed out. Among them were, while CD's can show lead to phosphorescence enhancement, this effect is very dependent upon the ability of the molecule to fit into the CD, and the extreme hydrophobicity of the cavity often results in unfavorable competition with the bulk solvent for even slightly aqueous soluble analytes. Both of these factors can lead to no observed RTP for molecules that readily show MS-RTP. Cyclodextrins also often demonstrate'limited water solubility. Micelles do not demonstrate the size dependency, and their dynamic nature makes the hydrophobicity issue less important. Cline Love‘s group also point out that sensitized phosphorescence shows marked enhancement going from CDs to micelles (63). 88 Other Methods The recognition that molecular organization is an important driving force for the observation of RTP has led experimenters to examine other media. Weinberger has examined the use of colloidal/microcrystalline suspensions (64,65). These systems, produced by directly injecting a stock solution of the analyte into water, showed strong RTP. The self-organizing effect of the crystal was the vehicle for RTP enhancement. The concentration dependence of RTP in these samples was found to be non-linear. The solutions were not purged so that any residual RTP from impurities would be suppressed. Scaiano's group at the National Research Council of Canada has experimented with the RTP of aromatic ketones in silicalite (66,67,68). Silicalite is a hydrophobic zeolite with a mixture of circular channels which zig-zag through the particle and elliptical channels which penetrate directly through the structure. They have found that these ketones show that probe molecules enter the channels of the zeolite and are constrained in a geometry which precludes the usual intra-molecular de-activation. Their studies show two different molecular configurations in the substrate with differing susceptibilities to oxygen quenching. These workers have also experimented with the RTP of arylpropiophenones in lyophilized CD inclusion complexes. Their findings have shown strong RTP with some loss of the fine structure found in the aqueous solutions of these systems. They also found that the intensity varied widely as the relative size of the guest versus the diameter of the cavity varied. The alpha-CD gave the poorest results in all cases (69). These are not solution-based RTP emission, but the molecular organization is much more 89 specific than that found in solid surface RTP and reflects more closely those techniques used in solution RTP. The idea that a solution can be microscopically organized is just beginning to make itself felt in chemistry. The applications of other systems to this task are being followed in our laboratories. Among those topics which may bear fruit are investigations of RTP mediated by aza- paracyclophanes and the dendrite macromolecules. 90 10. 11. 12. 13. 14. 15. l6. 17. 18. CHAPTER V References Saviotti, M. L.; Galley, W. C., Proc. Nat. Acad. Sci, USA., 1974, 71, 4154. Kalyanasundaram, K.; Grieser, P.; Thomas, J. K., Chem. Phys. Lett., 1977, 51, 501. Rosen, M., Surfactants and Interfacial Phenomena, Wiley Interscience: New York, 1978, Chap. 3. Humphrey-Baker, R.; Moroi, Y.; Gratzel, M., Chem. Phys. Lett., 1978, 58, 207. Turro, N. J.; Liu, K-C.; Chow, M-F.; Lee, P., Photochem. Photobiol., 1978, 27, 523. Almgren, M.; Grieser, F.; Thomas, J. K., J. Am. Chem. Soc., 1979, 101, 279. Turro, N. J.; Aikawa, M., J. Am. Chem. Soc., 11980, 102, 4866. Cline Love L. C.; Skrilec, M.; Habarta, J. G., Anal. Chem., 1980, 52, 754. Skrilec, M.; Cline Love, L. J., Anal. Chem., 1980, 52, 1559. Cline Love, L. J.; Habarta, J. G.; Skrilec, M., Anal. Chem., 1981, 53, 437. Cline Love, L. J.; Skrilec, M., Anal. Chem., 1981, 53, 1872. Skrilec, M.; Cline Love, L. J., Anal. Chem., 1981, 53, 2047. Ramesh, V.; Ramamurthy, V., J. Photochem., 1982, 20, 47. Leigh, W. J.; Scaiano, J. C., Chem. Phys. Lett., 1983, 96, 429. Woods, R.; Cline Love, L. J., Spectrochim. Acta, 1984, 40A, 643. Femia, R. A.; Cline Love, L. J., Anal. Chem., 1984, 56, 327. Buell, S. L.; Demas, J. N., Rev. Sci. Instrum., 1982, 53, 1298. Garcia, M. E. D.; Sanz-Medel, A., Anal. Chem., 1986, S8, 1436. 91 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. Armstrong, 0. W.; Henry, S. J., J. Lig, Chromatogr., 1980, 3, 657. Dorsey, J. G.; DeEchegaray, M. T.; Landy, J. S., Anal. Chem., 1983, 55, 924. Armstrong, D. W.; Nome, F., Anal. Chem., 1981, 53, 1662. Arunyanart, M.; Cline Love, L. J., Anal. Chem., 1984, 56, 1557. Yarmchuck, P.; Weinberger, R.; Hersch, R. F.; Cline Love, L. J., Anal. Chem., 1982, 54, 2233. Armstrong, 0. W.; Hinze, W. L.; Bui, K. H.; Singh, H. N.; Anal. Lett., 1981, 14(A19), 1659. Weinberger, R.; Yarmchuck, P.; Cline Love, L. J., Anal. Chem., 1982, 54, 1552. Grases, F.; Genestar, C.; Gil, J. J., Microchem. J., 1985, 31, 44. Hinze, W. L.; Singh, H. N.; Baba, Y.; Prevlic, V., in "Abstracts of Papers," 34th Pittsburgh Conference on Analytical Chemistry and Applied SpectroscopY, Atlantic City, N.J., March, 1983; Pittsburgh Conference: Pittsburgh, PA., 1983, Abst. 297. Van Bockstaele, M.; Gelan, J.; Martens, H.; Put, J.; Dederen, J. C.; Boens, N.; De Schrijver, F. C., Chem. Phys. Lett., 1978, 58, 211. Bolt, J. D.; Turro, N. J., J. Phys. Chem., 1981, 85, 4029. Turro, N. J.; Lee, P. C. C., J. Phys. Chem., 1982, 86, 3367. Saenger, W., Angew. Chem. Int. Ed. Engl., 1980, 19, 344. Hennrich, N.; Cramer, F.; J. Am. Chem. Soc., 1965, 87, 1121. Cramer, F.; Saenger, W.; Spatz, H.-Ch., J. Am. Chem. Soc., 1967, 89, 14. Bender, M. L.; VanEtten, R. L.; Clowes, G. A., J. Am. Chem. Soc., 1966, 88, 2319. VanEtten R. L.; Clowes, G. A.; Sebastian, J. F.; Bender, M. L., g; Am. Chem. Soc., 1967, 89, 3253. VanEtten, R. L.; Sebastian, J. F.; Clowes, G. A.; Bender, M. L., J. Am. Chem. Soc., 1967, 89, 3242. Tabushi, I., Acc. Chem. Res., 1982, 15, 66. Tabushi, 1.; Yamamura, K.; Fujita, K.; Hiromu, K., J. Am. Chem. Soc., 1979, 101, 1019. 92 39. 40. .41. 42. 43. 44. 45. 46. 47. 48. 49. 50. ' 51. 52. 53. 54. 55. 56. 57. 58. 59. Rideout, D. C.; Breslow, R., J. Am. Chem. Soc., 1980, 102, 7816. .Breslow, R.; Campbell, P., J. Am. Chem. Soc., 1969, 91, 3085. Harata, K.; Uedaira, H., Bull. Chem. Soc. Jpn., 1975, 48, 375. Tabushi, I.; Shimokaea, K.; Shimizu, N.; Shirakata, H.; Fujita, K., J. Am. Chem. Soc., 1976, 98, 7855. Ueno, A.; Takahashi, K.; Hino, Y.; Osa, T., J. Chem. Soc. Chem. Comm., 1981, 194. Hoshino, M.; Imamura, M.; Ikehara, K.; Hama, Y., J. Phys. Chem., 1981, 85, 1820. Ueno, A.; Takahashi, K.; Osa, T., J. Chem. Soc. Chem. Comm., 1980, 1921. Yorozu, T.; Hoshino, M.; Imamura, M., J. Phys. Chem., 1982, 86, 4426. Femia, R. A.; Scypinski, S.; Cline Love, L. J., Environ. Sci. Technol., 1985, 19, 155. Jules, 0.; Scypinski, S.; Cline Love, L. J., Anal. Chim. Acta, 1985, 169, 355. Yorozu, T.; Hoshino, M.; Imamura, M.; Shizuka, H., J. Phys. Chem., 1982, 86, 4422. Scypinski, S.; Drake, J. M., J. Phys. Chem., 1985, 89, 2432. Turro, N. J.; Okubo, T.; Weed, G. C., Photochem. Photobiol., 1982, 35, 325. Turro, N. J.; Bolt, J. D.; Kuroda, Y.; Tabushi, I., Photochem. Photobiol., 1982, 35, 69. Turro, N. J.; Cox, G. S.; Li, X., Photochem. Photobiol., 1983, 37, 149. Scypinski, S.; Cline Love, L. J., Anal. Chem., 1984, 56, 322. Scypinski, S.; Cline Love, L. J., American Laboratory, 1984, 16(3), 55. Technical Note F7, SPEX Industries, Edison, N. J. Scypinski, S.; Cline Love, L. J., Anal. Chem., 1984, 56, 3331. Femia, R. A.; Cline Love, L. J., J. Phys. Chem,, 1985, 89, 1897. Femia, R. A.; Cline Love, L. J., J. Coll. Interface Sci;, 1985, 108, 271. 93 60. Donlerbroek, J. J.; Gooijer, C.; Velthorst, N. H.; Frei, R. W., Anal. Chem., 1982, 54, 891. 61. Donkerbroek, J. J.; Elzas, J. J.; Frei, R. W.; Velthorst, N. H., Talanta, 1981, 28, 717. 62. DeLuccia, F. J.; Cline Love, L. J., Anal. Chem., 1984, 56, 2811. 63. Cline Love, L. J.; Grayeski, M. L.; Noroski, J.; Weinberger, R., Anal. Chim. Acta, 1985, 170, 3. 64. Weinberger, R.; Cline Love, L. J., Spectrochim. Acta, 1984, 40A, 49. 65. Weinberger, R.; Cline Love, L. J., Appl. Spectrosc.. 1985, 39, 516. 66. Casal, H. L.; Scaiano, J. C., Can. J. Chem., 1984, 62, 628. 67. Casal, H. L.; Scaiano, J. C., Can. J. Chem., 1985, 63, 1308. 68. Scaiano, J. C.; Casal, H. L.; Netto-Ferreira, J. C., "Intrazeolite Photochemistry: Use of Beta-Phenylpropiophenone and Its Derivatives as Probes for Cavity Dimensions and Mobility," ACS Symp. Ser. 278(Org. Phototransform. Non-Homogeneous Media), Fox, M., ed., ACS: Washington, D. C, 1985; 211. 69. Casal, H. L.; Netto-Ferreira, J. C.; Scaiano, J. C., J. Incl. Phenom., 1985, 3, 395. Additional References Cline Love, L. J.; Weinberger, R., Spectrochim. Acta, 1983, 38B, 1421. Cline Love, L. J.; Habarta, J. G.; Dorsey, J. G., Anal. Chem., 1984, 56, 1132A. Kuboyama, A.; Matsuzaki, S. Y., J. Incl. Phenom., 1984, 2, 755. Lochmuller, C. H.; Marshall, D. B., Anal. Chim. Acta, 1982, 142, 63. Martens, F. M.; Verhoeven, J. W., J. Phys. Chem., 1981, 85, 1773. Pelizetti, E.; Pramauro, E., Anal. Chim. Acta, 1985, 169, 1. Potanay, G.; Rollie, M. E.; Warner, I. M., Anal. Chem., 1985, 57, 569. Tabushi, I.; Fujita, K.; Yuan, L. C., Tet. Lett., 1977, 29, 2503. 94 Chapter VI Instrumentation Introduction The original intent of this research was to construct and evaluate an intensified diode array spectrometer for the investigation of RTP phenomena. Some development work was done in this area (1). However, it was decided that a single channel phosphorimeter should be constructed first. This instrument was needed to build a base of experimental knowledge in the problems associated with collecting phosphorescence decay information. In particular, the lifetimes determined in MS-RTP are dependent on the experimental parameters, and it was determined that a conventional instrument would be necessary to characterize these experiments before more sophisticated experiments were attempted. Therefore, this document will examine the development of the single channel, microcomputer-controlled phosphorimeter. The development of the computer-controlled phosphorimeter went through many stages. The first efforts based on a PDP-12 Instrumentation Computer (Digital Equipment Corp., Maynard, MA.) are briefly described, and a few representative results are demonstrated. Data derived from the basic analog phosphorimeter are included for comparison. After this the development of the current version of the instrument is presented. The discussion focuses on two aspects of this instrument: the amplifier 95 modifications, which made the existing instrument compatible with computerized data acquisition and phosphorescence lifetime measurements, and the data acquisition board, which converted the output of this amplifier to a digital record. The use of a timer chip and associated trigger circuitry to improve the detection of the phosphorescence signal is examined at some length. The features of the data acquisition computer that were of particular importance in controlling the phosphorimeter are described in some detail. Finally, a board which was made to port the most useful features of the computer to an IBM PC (International Business Machines Corp., Endicott, N.Y.) compatible computer are described. The Original Instrument The phosphorimeter platform on which the experiments were performed was an SPF-500 Ratio Spectrofluorometer (American Instrument Co., Silver Spring, Md. acquired by SLM Instruments, Inc., Urbana, 11.) which was equipped with a rotating can phosphorimeter (SLM Instruments, Inc., Urbana, 11.). This was a dual photomultiplier tube (PMT) system with ratioing electronics capable of source correction in the fluorescence mode. This was accomplished by passing a fraction of the excitation beam through a quantum counter solution. The fluorescence from the quantum counter solution was passed to the second PMT. The relatively long rise time of cryogenic phosphorescence emission (2, same as Reference 3.70) made this facility impracticable in phosphorimetry. The long rise time of the phosphorescence signal is due to the longer time scale of the photophysics of phosphorescence. A 96 source variation detected by the quantum counter would not appear in the phosphorescence of the sample for several seconds. Thus, any source flicker correction made by the ratioing electronics would only increase the noise level. The amplifier in this system was a single, high-gain, operational amplifier current-to-voltage converter with variable resistance feedback to control the gain. The system was equipped with an output damping circuit to smooth the signal prior to the strip-chart recorder. There were several design features of this system which simplified the instrument, but which rendered the circuit unsuitable for computerized data acquisition. The large resistance feedback resistor (500 M9) and the long lead from the PMT were excellent antennas for radio-frequency interference (RFI). Using an oscilloscope, one observed intense RFI noise at the input stage of the amplifier which decreased after sunset. The 870 kHz noise was traced to the campus radio station which went off the air at sunset. The operational amplifier (AD 506 K, Analog Devices, Norwood, MA.) was a low noise device that had a maximum frequency response of 1 MHz and a full power response of 100 kHz (3). This slow response, coupled with the secondary damping constant, was sufficient to integrate high frequency noise to zero. The amplifier output with a single photon input, monitored with an oscilloscope, proved to be in the range of 50 mV with a transient width of between 5 and 6 ms. A 300 mV, 60 Hz component was also observed at the output of the amplifier, and this, too, was integrated to zero by the external time constant. For the purposes of computerized spectral scanning, this amplifier with the external damping constant had many good features, including a 97 high degree of stability. However, there were two reasons that a redesign of the circuitry was required. The foremost reason was that it was desired to make lifetime measurements with this system. For many species of interest, such as halogenated biphenyls (4,5) where the lifetimes are on the order of 10 ms, the observed signal would be the convolution of the phosphorescence signal and the amplifier response function. For RTP experiments where the lifetimes have been seen to be on the order of hundreds of microseconds (Chap. V), this problem would be exacerbated. As in fluorometry, complex deconvolution algorithms would have to be invoked (6,7,8). The second problem with the existing amplifier revolved around the use of the rotating can phosphoroscope accessory in a spectrofluorometer chassis. With this device the sample was placed in a quartz tube (5 mm i.d.) and immersed in liquid nitrogen in a quartz Dewar flask. This assembly was mounted in a double-walled cylindrical holder with three holes drilled in it. Two of the holes were co-linear with the excitation" beam and the third was at 90°, pointing toward the emission monochromator. A second co-axial, motor-driven cylinder with diametrically opposed holes in its sides rotated in the space between the two walls of the holder. When the holes in the rotating sleeve lined up with those in the holder along the excitation axis, the sample was illuminated. When the inner portion rotated 90°, the excitation source was cut off and the phosphorescence was permitted to illuminate the PMT. Figure 8 shows a simplified diagram of the rotating can phosphoroscope. The first experiments in computerized data acquisition were performed using this amplifier. Data collection was accomplished with a PDP-12 computer. This was essentially a PDP-8 computer (Digital Equip. 98 .oaae eoaoouaoxo one on come up new .oooaoz Home oz» ea seam when» a ma moose .oeoao come now .o>eon mecca ea» c« muwam 03» one muons .emoomouocmwocm coo mcaumuou one .m apnoea mozmommmommmomm ZOHBmmqm in . UZHHuaafiocaaou no xoma ecu mcwuouum:05op saxo«pouoHcouH mo muuoomm .0“ shaman E 3.. a tank-.500: h2— due 102 construction of a floppy disk controller (10) so that the system would have bulk storage capacity to ease programming and data storage. The version of the twin bus microcomputer implemented in this laboratory was based on the Intel 8085A microprocessor (Intel Corp., Santa Clara, CA.). It was designed to operate under the poly-FORTH operating system/programming language (FORTH, Inc., Hermosa Beach, CA.), which contained an assembler for the 8085 instruction set. The FORTH system was configured to use a VT-52 terminal format (Digital Equip. Corp.). The system was equipped with two 8251A Programmable Communication Interfaces (USART or UART, Intel Corp.) on a single board. One of these was for programmer terminal communication and one was for communication with a PDP-ll computer operating under the RSX-llM/M+ operating system (Digital Equip. Corp.) for permanent storage of data records, including programs. Protocols for this transfer, including the program FORTHPIP, were developed in the laboratories of Professor C. G. Enke by P. Hoffman (11). Interrupt processing for the computer was handled by an 8259A-2 Programmable Interrupt Controller (Intel Corp.). The ability to mask or redirect interrupt service routines easily using the Forth operating system was used extensively in this research. Memory in this system was partitioned at the time the system was stored in the Read Only Memory (ROM), the actual system used to boot the computer. The final FORTH configuration consisted of 2 kilo-bytes hexadecimal (2Kh) of ROM and 6Kh of Random Access Memory (RAM). Most references to memory in this document will be in base 16 since it is easiest to use in the FORTH environment. A digit equals the contents of a given byte. 103 The entire computer system was tailored so that it might be integrated into the intensified diode array experiments. With that in mind it was equipped with a bus multiplexer (9). This allowed two 4Kh RAM boards to be swapped into the computer data space. This was done so that, during the anticipated high data acquisition rates from the photodiode array, one bank of RAM could be collecting data under direct memory access control (DMA), while the other board was being interrogated by the processor. Although this functionality was never exploited, it did provide for a 4Kh data buffer which was separated from direct system management. Thus, using explicit data addressing, programs were able to store data contiguously in a large buffer without concern for system parameters. This simplified data acquisition significantly. It also prompted a somewhat different peripheral addressing scheme when compared with other systems in the department. Since the 4Kh RAM board could not overlap any of the 4Kh internal boundaries for proper debugging with available software (a 16-bit address bus is capable of addressing 64 K decimal which equals 16Kh giving 3 of these 4Kh boundaries), the data buffer was assigned a starting address of C000h, the top 4Kh. As a result the UARTs were assigned addresses of BEOOh and BE40h, the disc controller was placed at BFCOh, and the interrupt controller was at BE80h. A complete list of CS addresses used in the twin bus microcomputer are shown in Table 9 in Appendix B. The Amplifier When the twin bus microcomputer was installed, it was decided to change the amplifier system in the spectrofluorometer in order to make 104 the system capable of following faster transients, as well as do spectral scanning and quantitative analysis. The amplifier system was constructed using LF 351 op-amps. A current-to-voltage converter (C/V) was constructed by connecting a 100 k0 resistor between the PMT anode and ground. The voltage drop across the resistor was buffered with an op-amp voltage follower (12) (See Figure 11). In this case an AD 506K op-amp was used because the signal was generally quite small, and pulse broadening was not a problem. This configuration was chosen since it produced a minimum of ringing. This C/V produced an 8 mV pulse, 2 us in duration corresponding to a photon striking the photocathode. The C/V was attached to the 4" twin-axial cable extending from the PMT socket. The C/V was equipped with a second twin-axial socket with grounded inputs so that the remaining connector to the original amplifier could be connected and grounded when not in use. By simply disconnecting the two twin-axial cables from the C/V and re-connecting them, the original analog instrument functions could be accessed for purposes such as academic instruction. The C/V output signal was passed through a co-axial cable to the LF 351-based amplifier. At the amplifier the signal was connected to ground through a 50 Q resistor to minimize reflections in the transmission line. The amplifier consisted of two direct-coupled inverting amplifiers with feedback (13). In the first stage the input resistance was 1 kc and the feedback resistor was 10 k0. The second stage had a 1 kn input resistor and the feedback resistor was 100 kn. The photon pulse output from this amplifier was 5 V with a duration of 12 us. 105 + ‘5 V. 100k Ohm Figure 11. Schematic diagram of the C/V circuit. 106 The amplifier saturated easily in the spectral scanning, direct current mode while displaying baseline output 20 us later. This was observed with an oscilloscope many times. The high gain and wide bandwidth of the amplifier caused this behavior. A 3 ms low pass filter was constructed at the output of the first stage of the amplifier to rectify this problem. It was made with a 47 0 resistor and a 10 uF capacitor. The output of the filter was buffered through an LP 351 voltage follower to the input resistor of the second stage (See Figure 12). The filter was designed so that it could be switched into the circuit between the two stages of the amplifier with a single switch. With this filter in place, the output of the amplifier for a photon pulse input was seen to be on the order of 150 mV with a duration of approximately 600 us. When a phosphorescence signal was observed at the output of the amplifier using the rotating can to modulate the excitation, a smooth pulse was seen and its width was only slightly greater than the width of the emission window as detected by a photodiode monitoring the excitation window. The amplifier exhibited a residual 100 mV p-p 60 Hz noise when connected to the C/V. This noise was associated with ripple in the PMT high voltage power supply. Attempts to capacitively couple the input of the final stage of the amplifier to remove this noise were unsuccessful because the coupling capacitor saturated at the high gain used. Efforts to shunt the DC component at the capacitor were also unsuccessful. This capacitor charging effect produced some very pronounced effects in the emission decay experiments (See Figure 13). The final d-c amplifier design had an unstable offset and required continual adjustment to keep the output near ground. 107 1 Oh Ohm he. PIT A ._- 4. . LP 381 so on... } 1h Ohm _ 100k Ohm TW— .__._. <+ vaw->l ~4P_' I LP 381 47 Ohm Figure 12. Schematic diagram of the amplifier circuitry. 108 scum: mmsmr ( A-D ums) : a 0° :3 a. ’e’ 2 t ‘33.. Figure 13. Emission decay curve for 1mM Tb3+ in D20 collected with an ac-coupled amplifier in the data collection system. The dip at 1 ms is due to signal charging of the coupling capacitor. 109 Efforts to find the source of a 1-2 MHz noise component with a 150 mV p-p amplitude, or to shield the amplifier from it, also met with failure. The changing offset was the most serious problem with the amplifier. The two noise components described were effectively averaged out of the data. The Data Acquisition Board The control of the spectrofluorometer/phosphorimeter demanded that a number of control pulses be sent to the instrument. These included setting and releasing the stepper motor clutches for both monochromators, setting the scan direction, and stepping the stepper motor. Two strategies presented themselves. One was to send instructions to the instrument through a parallel port on the microcomputer and to decode them at the phosphorimeter with discrete logic. The alternative was to use the chip-select (CS) board functionality designed for the modular twin bus microcomputer (9). Since most of the control signals were used to initiate a mechanical action that took a period of time to affect, simultaneous decoding of several instructions from one 8-bit word coming over the parallel port would leave the system in an ambiguous mechanical state. Therefore, the parallel port offered few advantages, and ease of implementation dictated the use of the CS board. The CS boards generated a negative going pulse 500 ns wide whenever a dummy value was written into the address space that the unit occupied. In application this board could be programmed to execute multiple instructions sequentially at a rate of 330 kHz when speed was important. The CS signal was generally inverted and used to either set 110 or reset a 74LS74 dual D flip-flop to generate the appropriate level- sensitive signal for mechanical control. The signals from the microcomputer were sent to a point-to-point solder board installed in the spectrofluorometer where the level conditioning was implemented. In addition to the electro-mechanical functions, the A/D functions were mounted on this board to minimize the distance between the amplifier and the A/D converter. There were three signals which needed to be digitized, the signal arising from the PMT and two DC voltages, ranging from 0.0-10.0 V, which were proportional to the wavelength of the excitation and emission monochromators over the range 200.0-1,200.0 nm. These latter two voltages were used for computerized setting of the approximate starting parameters of the spectral experiments. These three signals were passed through an MX 808 (Datel/Intersil, Mansfield, MA.) 8-channel analog multiplexer to a sample-and-hold amplifier (SHA), AD 582 (Analog Devices). From the SHA circuit the signal traveled to an AD 574 A/D converter (Analog Devices). All PMT signals passed through shielded cable when the distances were greater than 1 cm. The A/D converter was set to work in 8-bit bus mode (pin 2 to gnd), and it was always selected (pin 3 to gnd) and enabled (pin 6 to +5 V). In this configuration a CS pulse would set a 74LS74 dual D flip-flop to generate a hold signal for the AD 582 and initiate conversion. When the conversion was complete, the A/D converter STATUS signal would go low triggering a free-running 74L8121 monostable multivibrator which generated a short reset signal to the flip-flop controlling the SHA. The data were read from the A/D converter 8 bits at a time under the control of the lowest address line, A0 which was connected to pin 4 of the A/D converter. When A0 was low, the 8 most significant bits (M88) 111 were presented to the data bus, and when it was high, the 4 least significant bits were bused with 4 trailing zeros (LSB). This was made especially easy by the fact that a given CS signal was available over a range of 64 successive addresses. Thus, the lowest address for the data latch cycle (BBFBh) selected the M88 and the next address (BBF9h) obtained the LSB. The number was then divided by 10h to give the datum. The conversion times of different specimens of the AD 574 were found to be quite different. The results of conversion time tests performed in the data acquisition board are shown in Table 1. Table 1. Conversion times for specimens of the AD 574 A/D converter Specimen number Conversion time (us) 1 28 2 34 3 18 4 32 5 24 Average 27 Unfortunately, specimen 3 malfunctioned part way through this research, and the data reported were derived from specimen 5. The timing delay for A/D conversion was achieved through execution of dummy assembly language instructions. Control of the multiplexer was affected through a 74LS174 hex D flip-flop. The lowest three hits of a data word were written to this chip. The multiplexer decoded them, and selected the one of eight possible channels. Channel 7 was used for the PMT signal, channel 6 for the excitation monochromator, channel 5 for the emission monochromator, 112 and the remainder were unused. Appendix A contains a list of different hardware addresses used in the twin bus microcomputer phosphorimeter. Figure 14 shows the essential features of the A/D portion of the data acquisition board. Optical Isolation The data acquisition board, when originally installed, showed a large 60 Hz signal, even when its input was grounded, whenever the microprocessor was connected to the phosphorimeter. All efforts to establish single ground links between the various portions of the instrument were unsuccessful at eliminating this problem. Following the lead of Fontaine (14), it was decided to insert 6N137 optical isolators (Hewlett-Packard, Palo Alto, CA.), or their equivalents, in all of the lines between the microcomputer and the instrument. Since the optical isolator is uni-directional, data lines going to the 74LS174 had to be buffered separately from those coming from the A/D converter. Upon insertion of the optical isolators, the 60 Hz signal was no longer observable when the amplifier input was grounded (See Figure 15). However, the 6N137 device had a propagation delay of 60 ns (15), and this resulted in data timing errors in the read cycle of data collection. Data could not be collected under any circumstances, and it was impossible to set the 74LSl74 to select the correct channel of the multiplexer. It was decided to latch all data and the A0 line with 74LS373 octal latches. The computer could then be instructed to perform a read or write instruction twice, and on the second execution of the instruction, the necessary parameters would already be established at 113 .ouooo coHuHmfioooo sumo on» no cofluuom Q\¢ ecu no Emummfio HoMuOuowm 4 .vH oucmwm sec-Sesaeecel 530.30.300.32 gazeteuu Bozo—EN .0893 o( QOQIXI raced us..e:e°leoaese p thpudoh thmdoh pflpfldvh :3... 8.5 abandon a» p x38 eh thfldvh p Ornath towed (In 5308.51.33 Eat A. 00 00v a .93- 3. .3230 .2228 an...» 114 A E I P O . 0.. .'~ 0 ‘ e i fito‘ '. .3 .n - I -. . v d p . ~ ‘. o d )- l " ' h . e O ... I . O. O . d b ' e 0’ \ \ .' ‘5 .0“ ' d P .' e o o . O m A L _1 1 l '4 _ .1 L I i 1 v T T v 1 v w I 1” use. ” - r- d b autumn—om) M(W) Figure 15. The amplifier output a) before optical isolation and b) after optical isolation. 115 the inputs to the particular chip. When time was of the essence this could be done in assembly language instructions with a 3 us overhead; the strategy proved to be highly reliable. Figure 16 shows a block diagram of the final data acquisition system. Appendix A contains a complete point to point listing of the signals on this board. Timing the Phosphoroscope The rotating can phosphoroscope provided an inherent challenge in the form of timing the data collection process. When phosphorescence data were collected at random, the signal collected would more probably be the amplifier null since the periods when the phosphorescence fell on the PMT were of shorter duration than were the periods when no illumination fell on the detector. The average signal obtained by random collection was much lower than the peak signal. The amplifier noise was dominated by two noise sources, 60 Hz and high frequency, which were independent of the signal amplitude. Therefore, averaging the true signal with the baseline had the effect of raisinging the LOD. Thus, a mechanism was needed to correlate the detection process with the appearance of the phosphorescence signal at the PMT. The solution used is called time correlated data collection. Data were only collected when the phosphorescence illuminated the PMT. Two related objections prohibited the use of the phosphorescence in conjunction with an optical detector to trigger the data collection. First, it was found that the intensity was usually too low to provide a dependable signal when a photodiode sensor was used. Second, when the 116 .oueoc cOuuumHoooe eueo ecu mo coaoue> Hecum ecu mo Ecumewo cooHc oeauwamaum < .mH museum lanesT lease l I T)? I ”a I can..- 2h= Odom Eco SIN 119 illuminated the sample, a portion of the radiant energy would pass completely through the cell and illuminate the photodiode. The LSI Timer/Counter The photodiode signal was the same frequency as the appearance of the phosphorescence at the PMT (the time when the phosphorescence illuminates the PMT will also be called the emission window). The time that the phosphorescence started to illuminate the PMT needed to be determined, but the frequency was not explicitly known, beforehand. In terms of collecting data, this meant that there was an unknown time between the positive going edge of the photodiode signal and the start of the emission window at the PMT. The phase angle was known to be 90° so an AMD 9513 LSI timer/counter (Advanced Micro Devices, Inc., Sunnyvale, CA.) was chosen to determine the timing parameters needed to optimize the data collection process. The characteristics of this chip and its applications to laboratory problems have been previously described (16) so only a short description will be given here. It is a large scale integration (LSI) chip containing 5 counter/timer (C/T) circuits which can be cascaded. Each C/T has a mode register where counting parameters such as internal or external counting sources, count up/down, count repetitively or singly in binary or binary-coded-decimal (BCD), reload source, and terminal count characteristics can be specified. Each has a load register, where a starting number can be loaded, and a hold register, which can be used similarly to the load register so that alternating load sources may be used for variable duty cycle counting. The hold register can be used to store the contents of 120 the counter upon receipt of a save command. There is a master mode register which controls the chip as a whole. Functions such as binary/BCD division of the clock input frequency, 8/16 bit bus, the source for the output oscillator (including binary or BCD divisors of the input crystal), scalers from 1:1 to 1:16 of the chosen source, and time-of day functions are controlled by this register. It comes in a standard 40-pin package, and a 1 MHz oscillator was installed as the clock standard. The simplest approach to timing the data acquisition system was to connect one of the C/T's, which had been configured to count up on the rising edge of each pulse, to the signal from the photodiode circuit. A second counter was set to count down for 20 8, both counters were armed for counting, and the computer was placed in a halt state. When the count-down C/T reached zero it sent a positive going terminal count (TC) to the interrupt controller which restarted the computer. The first action after the interrupt had been serviced was to disarm the C/T and save its contents in the hold register. This gave the total number of excitation pulses that had occurred in 20 8. Since the usual number of counts was in the range of 1200 and the time base was accurate to microseconds, 20 5 divided by the number of counts gave an average rotation time that was accurate to about 0.1%. Direct observation of the photodiode signal with an oscilloscope revealed that there was approximately a 5% variation of the rotation speed with a maximum pulse- to-pulse variation of about 1%. For a rotation of 90° this corresponds to a 0.5% variation from excitation to emission. At 6 ms per pulse this variation translates to 30 us. The time between the excitation and emission windows was then taken to be one-half of the inter-pulse time, 121 and a safety factor of 100 us was added. This method, although it mandated a 20 8 dead time at the start of each experiment, was found to be the easiest and most reliable since all signals were processed by the AMD 9513 LSI Counter/Timer without external hardware. Data collection was performed by an assembly language routine which collected 8, 12-bit data points from a given emission window in 490 us. These values were then averaged, and the average was returned to memory. Data for 512 emission windows were averaged to yield the spectral value when scanning. The data acquisition software is considered in detail in the next chapter, along with uses of the AMD 9513 LSI Counter/Timer in evaluating software execution times. The Flashlamp Light Source for Lifetime Determinations One of the primary reasons for modifying the amplifier was to construct an amplifier that could faithfully track transient signals at lifetimes as short as 100 - 200 us. In order to perform such experiments, it was necessary to obtain a light source that could deliver a high power, short duration , ultraviolet pulse. The system used was the 437A Nanopulser with an N-789B source (Xenon Corp., Wilmington, MA.). This was an air-gap, high-voltage flash-lamp which was rated at 500 kW, peak power, in a 20 ns pulse. The pulse lamp showed a broad emission band centered at 320 nm which was suitable for exciting molecules of interest. Although the pulse lamp power supply was equipped with a facility for remote triggering by "a +3-7 volt signal (risetime 2.0 us or less )" (17), computer-controlled firing of the power supply with the flashlamp 122 attached did not prove easy. Two lamp power supplies supplied by the same manufacturer (Xenon Corp.) were used, but the current demand for the trigger was greater than the 40 mA provided by the most powerful transistor-transistor logic (TTL) circuits available. Although one of the lamp power supplies would fire the lamp intermittently from a TTL signal; the other would not do so under any circumstances. The first attempted solution was to use the TTL signal to drive the base of a 2N3053 npn transistor connected to the +5 V supply of the computer. The signal was connected by a reverse biased diode to ground to shunt any negative transients that might be given off by the high-voltage discharge. This signal would fire both power supplies. However, one appeared to have a weak SCR switch and would quit firing after it warmed up under heavy use, while the second would reset the computer upon firing. An improved solution was to connect the output of the transistor to a pulse transformer. Although this ostensibly isolated the grounds between the two systems, both problems remained. It was determined that the lamp power supply that caused the system to crash repeatedly had a defective pulse transformer. It eventually could be heard to make a snapping sound when it fired. The final solution was to use the supply with the defective SCR and keep the repetition rate low enough that the defective part did not overheat. Then, when it did fail, a manual firing pulse from the front panel would usually restore its operation. Sensing the Flashlamp The intense UV light pulse from the flash lamp was assumed to be strong enough to disrupt the structure of the UV-100-BQ photodiode (18) 123 so a cheaper photo-conductive photodiode found in the laboratory was used. In an effort to increase the performance of this sensor with the hope of substituting it for the other sensor and reducing the part count in the confined quarters of the sample cell, a discrete transistor amplifier was constructed. A complementary pair of transistors, a 2N3904 transistor (npn) and a 2N3906 transistor (pnp), were connected as shown in Figure 18, and the output was passed through two stages of Schmitt triggering (7414) to eliminate false signals due to ringing. The two sensing circuits were mounted on a single card and attached to a self-locking plastic ring which slipped over the Dewar flask holder. It could be rotated so that the appropriate detector was located over the appropriate hole. For spectral scanning and quantitative work, the first circuit was rotated to observe the beam coming from the excitation monochromator, and the line to the data acquisition board was attached to the 7414 Schmitt trigger. For lifetime studies the ring was rotated approximately 180°, and the line to the data acquisition board was attached to the 74LS14 chip. The rotating inner sleeve was removed from the phosphoroscope attachment. The flashlamp was mounted in a "V"-shaped, adjustable mounting bracket so that the same holes used for excitation in the conventional phosphorimeter could be employed. In all experiments the sample compartment was shrouded with a baffled cardboard housing. A visual inspection of the amplifier output with an oscilloscope showed no difference in the output whether the entire system was enclosed in a blackened box. With the cardboard housing in place during PMT cooling experiments, periods as long as 5 s were seen to be free of thermal or 124 .meeucmeaw ecu ocuecem uOu uwsouwo ecu mo oaueEecom .ma euomem 9 pm...» 5&0 Oh? 5&0 0.. Count“ 6&0 can 2.3 o— Coons“ 5‘0 8°09 UGO-905.03.— 125 photonic emission. These experiments demonstrated that the cardboard shroud was effectively "light-tight". Single Photon-Detection and Lifetime Determinations Timing the data collection was primarily software controlled with the exception of very long-lived phosphors such as biphenyl and 2,7- dichlorodioxin. The details of the software timing and their relationship to the use of the AMD 9513 chip in the FORTH operating environment are discussed in detail in the next chapter. However, one timing aspect of collecting lifetime data for the long-lived phosphors was more properly a function of the utilization of the AMD 9513 as stand alone hardware and is considered here. The decay curve for long-lived phosphors in the lifetime experiments was indistinguishable from the background. Even with the 3 ms filter in place, the signal for biphenyl was seen, on an oscilloscope, to fall below 250 mV within 20 ms. One solution was to include a gated integrator and electronically sum the signal for a fixed period of time. Due to the high gain of the amplifier used, a stable integrator could not be constructed. A relatively quick and inexpensive alternative to the integrator was to put in a simple single photon-detection circuit that could be switched into the data acquisition circuit. This was done by passing the signal through an LP 311 comparator with the (-) input set at 2 V. This eliminated system response to most of the PMT dynode thermal emission, and it produced a 5 V square pulse for those signals that were passed. The width of a signal from a single photon was 12 us. This pulse was 126 used to set a 74LS74 flip-flop. The A/D circuit then sampled the state of the flip-flop, and the change in the STATUS signal from the A/D converter was used to generate a reset signal to the flip-flop. This 74L5121 monostable multivibrator circuit which conditioned the reset signal was discussed with the sample-and-hold amplifier in the section describing the data acquisition system. The mechanism for timing the data acquisition software and the method for calculating the collection efficiency are reserved for the next chapter. The AMD 9513 was used to establish computer interrupt signals to start the data acquisition process at regular intervals during the phosphorescence decay of long-lived phosphors. This was necessary because allowing the data acquisition software to run for 8 s would have required 800 kilobytes of data memory. Eight seconds was required to sample the phosphorescence decay of biphenyl for two lifetimes. The interrupt-driven data acquisition strategy was easily implemented with the AMD 9513 LSI Counter/Timer chip by simply setting the first C/T to count down for the prescribed number of milliseconds at the start of each data acquisition cycle. The software collected and stored the raw data and then set the processor in a halt state. The C/T was programmed to generate a positive terminal count pulse (TC) when the allotted time had expired. The TC signal went to the interrupt controller. The interrupt controller chip had been pre-progranmed to provide the shortest service routine possible, and when it returned to the program in 20 us the process repeated itself until the software had determined that the experiment had ended. Figure 19 shows a block diagram of the complete system. This includes the peripheral devices needed to store and plot the data. These 127 .Eeumaw euep pepceuxe cue: ”mica xwa >925“. hctoaxn .299 _ cabanacolozoi dono- anhz. — wloxoi an. ueuesuuocmmocm penuueuomeoo ecu mo Ecumewp xoon .mH ensues cute: u socuzoo 22:33 so; Sued... i=8 9: , paeuao guac. 132 10. 11. 12. 13. 14. 15. Chapter VI References Herron, N. R.; Zabik, M. J.; Crouch, S. R., in "Abstracts of Papers," 35th Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, Atlantic City, N.J., March, 1984; Pittsburgh Conference: Pittsburgh, PA., 1984; Abstract No. 204. Barnes, C. G.; Winefordner, J. D., Appl. Spectro., 1984, 38, 214. Data Acquisition Book 1982: Integrated Circuits, Analog Devices: Norwood, Mass., 1982; 429. O'Donnell, C. M.; Harbaugh, K. F.; Winefordner, J. D., Spectrochim. Acta, 1973, 29A, 753. O'Donnell, C. M.; Harbaugh, K. F.; Fisher, R. P.; Winefordner, J. D., Anal. Chem., 1973, 45, 609. Cline Love, L. J.; Shaver, L. A., Anal. Chem., 1976, 48, 365A. Shaver, L. A.; Cline Love, L. J., Appl. Spectrosc., 1975, 29, 485. Ware, W. R.; Doemeny, L. J.; Nemzek, T. L., J. Phys. Chem., 1973, 77, 2038. Newcome, B. H.; Enke, C. G., Rev. Sci. Instrum., 1984, 55, 2017. Thiim, R., M.S. Thesis, Michigan State University, E. Lansing, Mi., 1982. Hoffman, P., Program FORTHPIP, Michigan State University, E. Lansing, MI., 1982. Malmstadt, H. V.; Enke, C. G.; Crouch, S. R., Electronics and Instrumentation for Scientist, Benjamin/Cummings Pub. Co.:Reading, MA. 1981; p. 113. Malmstadt, H. V.; Enke, C. G.; Crouch, S. R., ibid., 118. Fontaine, D., Ph.D. Dissertation, Michigan State University, E. Lansing, Mi., 1983. Application Note 939: High Speed Optically Coupled Oscillators, Hewlett Packard: Palo Alto, CA., 1972. 133 16. 17. 18. Wiegand, P. M.; Trischan, K. K.; Crouch, S. R., Anal. Instrum., 1985, 14, 127. Model 437A Nanopulser Operating Instructions. Xenon Corp., Wilmington, MA. Private communication with Mr. Marty Raab, electronics specialist in the Chemistry Department, Michigan State University, E. Lansing, MI., Jan., 1986. 134 Chapter VII The FORTH Operating Environment and Other Software Introduction This chapter considers the application of the FORTH operating environment to the collection of phosphorimetric data. FORTH is a stack oriented, threaded interpreter whose general advantages as an operating environment for analytical instrumentation have been described (1). Advantages particular to the experiments described in this thesis are presented in this chapter. The use of the AMD 9513 LSI Counter/Timer chip in the FORTH operating environment is examined. A powerful symbiotic relationship between the two was used to time data acquisition procedures and establish an accurate time base for the various experiments. Application of the FORTH assembler to data acquisition, and the advantage of the FORTH assembler over conventional assemblers are described. The various FORTH "WORDS" used to operate the phosphorimeter are defined to provide the reader with a framework to implement understanding of the complete code package. All data and procedures in FORTH are called "WORDS", and they are capitalized because the FORTH interpreter only recognizes the upper case. 135 A brief discussion of a pair of software packages used on the PDP— 11 minicomputer under the RSX-llM/M+ operating systems (Digital Equipment Corp., Maynard, MA.) is also given. One program converted the unformatted binary data in files generated by the program FORTHPIP into ASCII files that could be readily manipulated. The other program used a modified simplex algorithm to fit an exponential curve to lifetime decay data. Programming the AMD 9513: "ELAPSE" The FORTH operating environment used in the instrumentation computer was an important link in the collection of phosphorescence data. Timing data collection was of paramount importance, both in the scanning/quantitation mode and in the phosphorescence decay mode. The timing strategy utilizing the AMD 9513 as stand-alone timing generator was described in Chapter VI. An additional constraint in data collection with the rotating can phosphorimeter was to ensure that the data acquisition cycle had been completed during the emission window. Since the AD 574 A/D converter could digitize a piece of data in 24 us, and data storage was executed in 30 us, it was an inefficient strategy to collect one piece of data per emission window. It was better to collect several data points from each opening of the rotating can at the emission monochromator. Thus, it became important to be able to time a data acquisition routine that did multiple sampling in each emission window to ensure that the sampling did not continue after the phosphorescence had been cut off by the rotating can. 136 When lifetime data were collected for short-lived phosphors, the direct use of the AMD 9513 to generate an interrupt to trigger the data acquisition cycle was a time- wasting procedure. The interrupt service routine was a process of unknown duration that was necessarily interposed between the end of the timing interval and the re- initialization of the timer for the next interval. Repetitive use of this interrupt routine would introduce a cumulative error into the time co-ordinate assigned to the data. This interval also lowered the data acquisition rate. In this scenario it was decided to utilize the large (4Kh byte) data buffer to alleviate these problems by collecting data continuously and storing them in the buffer. This still left the problem of evaluating the execution time of the interrupt service routine and of each data acquisition loop in the reiterative, assembly language data collection software. It was also of interest to time the execution of various assembly language commands, so that approximate effects of introducing these commands into routines might be assessed ahead of time. The solution to all of these problems was found in the word ELAPSE (2). This word loads and arms one of the counter/timers (C/T) before issuing the EXECUTE command for a word whose address has already been stored on the top of the stack. It then saves the contents of the C/T upon finishing the execution of the word. Executing ELAPSE with the EXECUTE command removed generates the system overhead time. This value is then subtracted from the C/T value to give the execution time of a word. 137 ELAPSE with the Rotating Can The word ELAPSE was used to devise a strategy to increase the collection efficiency of the rotating can data acquisition software. The data collection routine was written so that the acquisition cycle collected 8, 12-bit data words and stored them in successive memory locations. This data acquisition cycle was found to execute in 510 us using the timing routine. The average width of an emission window was measured on an oscilloscope to be approximately 1.2 ms under experimental conditions. With a 100 us safety offset added to the rotating can phasing time (See chapter 6), the data acquisition period extended from 100 to 700 us into the emission window. This ensured that the data were representative of the phosphorescence signal. The number of data points could have been increased by 50% without fear of sampling into the background period, but 8 data points, which was arbitrarily chosen, gave good results. ELAPSE in Lifetime Determinations In the analysis of the lifetime experiments, the AMD 9513 LSI Counter/Timer and ELAPSE proved useful in 3 areas. As stated previously, the execution time for the generic interrupt service routine was not known. This problem was solved easily using ELAPSE. One of the C/T's was programmed to count down for 1 us and then generate a terminal count pulse (TC) to the interrupt controller. After the C/T was loaded and armed, the processor was halted. ELAPSE timed the execution of this word to be 94 us. The execution time of a similar word with no halt state and 138 no TC was 65 us giving an execution time for the interrupt service routine of 29 us. The same process, when the interrupt controller had been re-programmed to service requests more efficiently, yielded a service time of 20 us. This interrupt service routine was used in all lifetime studies. In photon-detection experiments with long-lived phosphors where the interrupt routine was repeatedly accessed, these data assured that an error of less than 2 ms would be generated by 100 interrupts. Since these experiments ranged from 1-10 8, this resulted in a relative error in time assignments of 0.2% or less. There were two types of lifetime experiments used in this research when the luminophore had a lifetime of 20 ms or less. These methods were described in Chapter VI. The software timing routine ELAPSE was used to establish a time base for each. For experiments where analog voltages were measured, 2 bytes of data were collected from the analog-to-digital (A/D) converter in rapid succession and stored in adjacent memory locations on each data conversion cycle. The conversion cycles were repeated non-stop until a counter was decremented to 0. By experimenting with a number of initial counter values (different number of data points collected) a 53.5 us execution time was determined. This enabled the assignment of a precise time base for each data point. This was equal to the data point number times 53.5 us, with 20 us added for the interrupt service routine to enter the data acquisition cycle upon receipt of the signal from the flashlamp sensor. For single photon-detection experiments only the most significant byte of data was sampled and stored. This routine was determined to take 41.3 pa per sample. Unfortunately, the A/D converter was not set to short-cycle the conversion process. 139 The Efficiency of Photon Detection When the data acquisition cycle time was determined it was then possible to calculate an efficiency for detection of single photons. The A/D converter used took 24 us to convert the data before resetting the flip-flop so this meant there was a 17.3 us period when the flip-flop was active in the data collection cycle. When combined with the 12 us pulse width from the discriminator this gave a 70% collection efficiency for photons under photon-counting conditions. Inclusion of a short cycle switch in the A/D circuit would cut the conversion time of this A/D converter to 16 us, resulting in a nearly 90% collection efficiency. It would also have made detection of "pulse pile-up" more certain. The FORTH Assembler The FORTH environment, as implemented, had an integral assembler which allowed the construction and execution of assembly code directly. This abrogated the cumbersome compilation and linking procedures used with the macro assembler on the IBM PC, for instance. All data collection words were written with the assembler. This was the only way to achieve the collection speed needed for lifetime studies or multiple data acquisitions in a single emission window. The fact that words written in assembly code could be immediately executed like any higher level FORTH word expedited debugging of the code. A data collection word could be executed and the contents of the memory immediately examined to see if the collected data corresponded to the input signal. This cut 140 many hours from the code debugging process. Additionally, the ability to dump and examine the memory is lost if a compiled language is used and the compiler must be reloaded. In this respect, the fact that this twin bus microcomputer had a separate, system-independent data buffer was also indispensable. If the system crashed during the execution of a word, re-booting it usually left the contents of the data buffer intact. The contents could then be used to determine how much of the word had executed before the system crashed. Control Commands The software developed for the instrument fell into two classes, instrument control and data acquisition. These were then utilized by 7 major experiment drivers. This section describes the FORTH commands which performed basic hardware control functions. The words HIGHER and LOWER set a flip-flop on the acquisition board to control the monochromator scan direction. The words EMSCAN, EXSCAN, and SYNCHSET set the monochromator clutches to scan the emission, excitation, and both monochromators simultaneously, and CLUTCHRELEAS freed both clutches. WAIT generated a 1 ms delay. STEP sent a single pulse to the flip-flop driving the stepper motor and 2 such commands alternated with a WAIT command caused the monochromator to step 0.05 nm. The WAIT state was necessary for both the stepper motor and the clutch controls to permit these mechanical devices to completely actuate. The numbers 5,6, and 7 used in conjunction with CHSET programmed the multiplexer to select the emission or excitation monochromators or the PMT signal as the input for the A/D converter, respectively. An 141 additional word HALT halted the processor and awaited keyboard input. The HALT state was particularly useful in permitting manual fine- adjustment of the monochromator before starting an experiment. Manual adjustment was necessary since the computer-generated setting was invariably off by a small amount. The monochromators were then stepped a fixed number of times to give the experimental wavelength increment in nanometers. The wavelength increment was specified at the time the experimental parameters were defined at the start of the experiment. The word DELF calculated the number of step pulses to be issued, and it was found to be much faster and more accurate than calculating each setting from an A/D converter reading. The A/D commands and I/O commands for the data collection words were explicitly programmed in assembly language using the CS's. However, words were written to perform these functions in high level FORTH so that the system could be debugged. The word CONVERSION initiated an A/D converter conversion, and the words M58 and LSB returned the values of the most significant and least significant bytes of the A/D converter output, respectively. Due to timing problems, all data read commands needed to be executed twice and the first value obtained was discarded. This was done automatically by the word LAMBDA which used the word DROP. The word LAMBDA proved particularly useful in checking whether the amplifier baseline had drifted. In the assembly language data acquisition words, the read cycle timing problem was circumvented by simply loading the accumulator twice from the same address. In all cases the data were properly collected on the second execution. Table 9 in Appendix B provides a catalogue of all of the CS signals used by the 142 twin bus microcomputer driven phosphorimeter. This includes the system support facilities. Scanning and Quantitating Words The word 90DEGREEWAIT timed the rotating can and programmed the AMD 9513 to generate an interrupt at the appropriate time for the word 8CAPTUREDATA to collect 8 data points when the phosphorescence illuminated the PMT. The data were averaged by the word 8DATA. A similar set of words 8FCAPTURE and 8FDATA performed the same task in fluorometric data acquisition mode. The fluorometer words were not interrupt-driven. Both sets of words stored the 8, 2-byte words of data in 16 contiguous bytes of memory starting at FFAOh and returned the average to FFBOh. BIGDATA averaged 512 of these interrupt driven phosphorimetric averages and SAMPLE averaged 1000 of the fluorometric data points. These words were used by the 2 scanning drivers TAKEPHOSSPEC and TAKEFLUOROSPEC and by the 2 quantitation words QPHOS and QFLUO. The scanning drivers interrogated the experimenter to collect the scan parameters, including the desired location, in memory, for the spectrum. They set the approximate starting wavelengths, released the clutches, and placed the computer in a HALT state while the experimenter did a final adjustment of the monochromators. A keyboard entry then initiated the scan using either BIGDATA or SAMPLE to collect the data. For quantitation it was decided that manual setting of the monochromators was the most efficient method. When they had been set, QPHOS timed the rotating can and then executed 8DATA 1000 times. It then 143 averaged the results. QFLUO followed the same procedure using 8FDATA. Both routines returned the data to the screen. Lifetime Data Routines The 3 lifetime collection routines were called RATEXP, SLOWRATEXP, and FOTONCOUNTRATEXP. Each used an interrupt driven data collection core routine. They were FLCONV, SLOFLCONV, and PCFLCONV, respectively. The first was a PMT analog voltage sampling routine and collected 2 bytes of data during each cycle. The latter two were single photon-detection routines which only collected the M88 of the A/D converter output as it digitized the output of a flip-flop (Chapter VI). SLOFLCONV and PCFLCONV were nearly identical with the exception of an extra HLT command in SLOFLCONV. This command was inserted so that multiple observation windows for long-lived phosphors could be triggered by the AMD 9513. As a result, the data collection loops in SLOFLCONV and PCFLCONV had identical execution times. The timing of all of these routines was discussed earlier under the topic "ELAPSE in Lifetime Determinations". Due to the structure of the averaging algorithm used, RATEXP permitted numbers of repetitions that were powers of 8. This limited algorithm was used because the mixed arithmetic functions in the version of FORTH implemented in the read-only-memory chips (ROM) of this twin bus microcomputer seemed to have some commands missing. Code that worked on other twin bus computers at Michigan State University, using ostensibly the same ROM's, would cause the machine used in this research to crash. 144 Since the single photon-detection routines used a simpler summing procedure, they permitted any number of repetitions. Some experiments where the flashlamp pulse was passed through the excitation monochromator were performed 256,000 times (unfortunately, these experiments often gave poor results). All of the lifetime data routines updated the count of pulse lamp flashes that had been made to the screen through the word LOKATOR. No discussion of FORTH should end without a warning to the reader that the structureless nature of FORTH often means the code is difficult to read. It is full of arcane mnemonics that have an intuitive meaning to the programmer, only. It is, however, a compact and powerful programming tool for instrumentation. It is hoped that the outline of the data system provided will help the reader understand the full package. A complete code listing is provided in Appendix B. Data Reduction Software Used on the PDP-ll Two major pieces of software that were used in this research have been amply documented in the Chemistry Department of MSU. One was the file transfer program FORTHPIP from Professor Enke's research group (3), and the other was the graphics package MULPLOT (4). MULPLOT (or MULPLT) has been the topic of instruction modules for the CEM 838: Scientific Instrumentation course. There were two other pieces of software written in FORTRAN by members of Professor Crouch's research group that were used extensively. One was a short program that read the un-formatted binary numbers in an RSX-ll file produced by FORTHPIP and converted them to a MULPLOT 145 compatible file that could be plotted (5). This program queried the operator for an abscissa value for the first datum in the file it read and the increment for each succeeding data point. These values could be times or wavelengths. It wrote the data to a new file in pairs where the first value was the calculated abscissa value and the second value, the ordinate, was the data value from the initial file. These numbers were in standard ASCII format. The other package was a modified simplex curve-fitting algorithm which could accommodate data files containing up to 500 data points (6). The version used was called EXPFIT since the curve to be fit in lifetime experiments was an exponential. The exponential function to be approximated was of the form Y = Aoexp(-kt) + C (7.1) This equation was entered into the sub-routine YFUNC. This program optimized the fit of the curve to the data by minimizing the sum of the squares of the residuals of the data from the fitted curve. This program could fit up to 5 parameters. It could, thus, accommodate a double exponential decay of the form Y = Aoexp(-klt) + Boexp(-k2t) + C (7.2) This was done to analyze the cryogenic phosphorescence decay curves of compounds such as 4-bromobiphenyl, where a residual, short-lived phosphor in the solvent was present. A convenient modification of this program (7), which allowed the experimenter to analyze only a portion of the lifetime data made this unnecessary for the purposes of this research. It did, however, demonstrate the applicability of the package to the resolution of component lifetimes in binary mixtures. 146 CHAPTER VII References Wiegand, P. M., Ph.D. Dissertation, Michigan State University, E. Lansing, MI., 1985. Ratzlaff, E., ELAPSE: a FORTH word, Michigan State University, E. Lansing, MI., 1982 - 1983. Hoffman, P., Program FORTHPIP, Michigan State University, E. Lansing, MI. 1982. Atkinson, T. V., and Gregg, H. L., "MULPLT: A Multiple Data Set, File-Based Data Plotting Program", Michigan State University, E. Lansing, MI., 1982. Kraus, P., X: a FORTRAN program to read FORTHPIP files, Michigan State University, E. Lansing, MI.,1984 - 1985. Wentzell, P., XYFIT: a FORTRAN curve fitting program using the Simplex algorithm, Michigan State University, E. Lansing, MI., 1984 - 1985. Kraus, P., Modification of FORTRAN program XYFIT allowing fitting of selected data in the file, Michigan State University, E. Lansing, MI., 1986. 147 CHAPTER VIII RESULTS and INTERPRETATIONS Introduction The data collected in this research fell into 4 classes. Three of those classes used a unique mode of operation of the phosphorimeter, while the fourth class could be performed in either of 2 modes and provided a platform for comparison of performance between the two. Two of the classes involved the spectral scanning capacity of the instrument. The first was for compounds whose experimental lifetimes were too short for their spectra to be measured using the rotating can phosphoroscope, which takes about 2.0-5.0 ms to rotate 90°. These included conventional fluorometry, MS-RTP spectra of organic compounds, and the excitation spectra of room temperature emission of Tb3+ in both HZO and D20. These experiments were run with the rotating sleeve removed from the phosphoroscope and the sample cuvette simply placed in the holder. Data acquisition was done by sampling the A/D at a randomly convenient time. This is called the fluorometric mode of data acquisition. The second type of experiment was for spectral scanning, at 77 K, of longer-lived phosphors. These ranged from brominated biphenyls with lifetimes of about 10 ms to biphenyl with a lifetime of 4.3 ms. These experiments used an interrupt driven mode of data acquisition where the 148 interrupt was generated by the AMD 9513 LSI Counter/Timer chip. This device was used to determine the rotation time of the rotating can from the excitation window to the emission window, so that the interrupt for sampling the analog-to-digital (A/D) converter appeared when the phosphorescence illuminated the PMT. These experiments are also referred to as time-correlated data acquisition. The third method collected the phosphorescence decay curves of long-lived phosphors such as biphenyl and 2,7-dichlorodioxin. These experiments were performed in a single photon-detection mode, since the analog signal was too small to be measured in a voltage sampling mode. They used repetitive retriggering of the data collection software by the AMD 9513 LSI chip, so that an 8 s spectrum could be fit into a 4Kh data buffer. The last class of experiments was used for short-lived phosphors. The data acquisition software was allowed to collect data as rapidly as possible for a specified number of data points after each flash of the flash lamp. This will also be referred to as "free-running" data acquisition. The time co-ordinate was determined by adding an interrupt offset to the product of the ordinal number of the data point and the execution time of the data acquisition loop. These experiments were run in the voltage sampling mode and the single photon-detection mode. The AMD 9513 LSI Counter/Timer was used to time the execution of the data acquisition loop. In the last 2 classes of experiment, where the flash lamp was used, the rotating sleeve was removed from the phosphoroscope. The Dewar flask holding the small cryogenic sample cell was placed in the remaining holder for low temperature experiments, and the larger RTP cell was inserted directly into the holder for room temperature work. 149 Fluorometry Mode Data Acquisition One casual method of comparing the performance of the digital data acquisition system with the original analog data acquisition system is to visually inspect spectra acquired by both systems. Figure 21 shows the effect of spectral averaging on the digitally collected fluorescence spectrum of a 1 ug/ml solution of quinine sulfate in 0.1 N 82804 (Regis Chemical. Co., Morton Grove, Ill., prepared by 10:1 dilution from stock). The excitation monochromator was set at 350 nm (+/- 1 nm), and the emission slit was 1 nm. The driver for the data system in these experiments collected only one data point at a time. The amplifier output was unfiltered so the system was operating in the photon shot- noise limit. One can readily see that as the the number of points averaged increased, the spectrum more closely resembled the analog spectrum taken with a 2.0 s damping filter. At 512 data points averaged per spectral value, the spectrum closely resembled the spectrum taken in the analog mode. However, a residual noisy appearance remained that reflects the shot noise of the amplifier output. The Nyquist sampling theorem gives the experimenter a more quantitative point of analysis. It states that sampling at a rate twice the highest frequency component in a signal will allow accurate reconstruction of the signal (1). For a dc signal this is a trivial criterion. For the decaying phosphorescence signal it will be of much greater importance. If one samples a dc signal more than once, noise components whose frequencies meet the Nyquist criterion established by the sampling rate 150 RILM(A-DWB) mm(A-Dm) mm(»om} mutation-DWI!) Figure 21. Fluorescence spectra of quinine sulfate collected with the digital data System showing the effect of spectral averaging (a - d). A spectrum taken in the analog mode is included for comparison (e). 151 are sampled with the data. Higher frequency components are also aliased into the sampling frequency domain (1). A major problem in digital data collection is to filter out high frequency noise components from an essentially dc signal. This is usually done by averaging many data points. To compare data collected by an analog data system with that collected by a digital data system with averaging, the analysis becomes more complicated. For the analog system the equivalent bandwidth is at 1/(4 RC) where RC is the time constant of the output filter. For the output filter with a two second time constant which was used to collect the previous data, the equivalent bandwidth is 0.125 Hz. The amplifier in digital data system used in these experiments produced a pulse approximately 125 us wide. If one considers this to be the equivalent time constant, then 512 samples averaged gives an equivalent bandwidth of 16 Hz (2). This gives a signal-to-noise (S/N) advantage of approximately 10 to the analog system in this set of experiments. This would seem to agree with the visual inspection. Figures 22a and 22b show the MS-RTP excitation and emission spectra of 1-bromonaphthalene with no electronic filtering, and Figures 22c and 22d show the same spectra with a 3.5 ms low pass filter inserted between the two amplifier stages. These spectra were taken of a 20 ug/ml solution of the analyte in 0.15 M sodium dodecyl sulfate (SDS) (prepared by Soxhlet extraction with ethanol and re-crystallization from ethanol of bulk technical grade SDS, source unknown). The solution was de-gassed for 30 min with high-purity N2 (Airco Industrial Gases, Riverton, N. J.) in cells of Suprasil quartz (Ace Glass Co., Vineland, N. J.) which were fabricated by the glass shop 152 ”If! 4” see ”0 C“ Figure 22. MS-RTP spectra of 1-bromonaphthalene taken with the digital data system (a - d). Details of these spectra, including slit widths, are discussed in the text. A spectrum taken in the analog mode is included for comparison (e). 153 of the Chemistry Department, MSU. They were equipped with stopcocks so that samples could be maintained indefinitely after de-gassing. The emission slits were 2 nm for both of the emission spectra (Figures 22b and 22d), and the excitation wavelength was 293 nm for each. For the unfiltered emission spectrum (Figure 22b) the excitation band-pass was 10 nm, and for the filtered spectrum (Figure 22d) it was 20 nm. For each of the excitation spectra (Figures 22a and 22c) the excitation slit was set at 2 nm. For the unfiltered excitation spectrum (Figure 22a) the emission monochromator was set at 500 nm (+/- 5 nm), while the filtered one (Figure 22c) was at 500 nm (+/- 20 nm). These spectra were taken with the final version of the data system, so 8,000 data points were averaged at each wavelength. Figure 22a shows an emission spectrum taken with the instrument in its original analog configuration. The emission spectra collected with the digital data system (Figures 22b and 22d) and the literature spectrum (3, same as Ref. 5.2) show a more pointed maximum than does the spectrum taken in the analog mode (Figure 22e). The source of this deviation in the analog spectrum is believed to be due to a loss of linear response in the chart recorder used to collect the analog results. The data from the digital system illustrate an important consideration when using a high gain, wide band-width amplifier under conditions of high photon flux. If several photons are coincident and the resultant photoelectron pulses are amplified simultaneously, the amplifier saturates, and the output signal is no longer proportional to the number of photons. Instead, it is a fixed value, the A/D converter maximum. Because the emission of photons is a random process, the next 154 data collection cycle may detect no signal, and the output is proportional to the photon flux at that time. The average of the two signals is artificially weighted toward the amplifier zero. To minimize this it was necessary to adjust the slits at the wavelength of maximum emission intensity, while monitoring the amplifier output using an oscilloscope, so that little evidence of saturation was seen (no flat-topped pulses). With the systems shown, the emission slits were 2 nm in all cases. For the experiments employing the amplifier with no external filtering, the excitation slits were 10 nm. For the experiments where the 3 ms filter was inserted between the two amplifier stages, the excitation slits were 20 nm. Since the excitation band is nearly flat at the excitation wavelength used, 310 nm (Figures 22a and 22c), it is possible to assume that the intensities of the 2 experiments is in the ratio of the square of the slit width ratio (4). This is what is observed, since a 2:1 slit ratio gives approximately a 4:1 signal ratio. This indicates that the amplifier was truly in its linear operating range. Shot-noise is again seen to be an important contributor to the spectral appearance. An interesting photophysical point is brought out by the excitation spectra. For most cryogenic phosphorescence there is an induction time for the phosphorescence signal. This is due to the fact that once the triplet state is populated, the transition to the ground singlet state, So, is "forbidden” and the rate of the transition is low. It takes a period of time before phosphorescence equilibrium is attained. This ranges from several milliseconds for halogenated systems and 3(n,n*) systems to several seconds for 3(n,n*) systems. For a conventional analog scanning instrument, phosphorescence excitation 155 spectroscopy is impossible. By the time phosphorescence from an excitation band is contributing to the amplifier signal, the recording apparatus indicates another excitation wavelength is the contributor. In MS-RTP the exchange dynamics of the analyte between the micelles and the bulk solvent reduce the experimental lifetime to the millisecond time scale or less. As a result the excitation spectra can be obtained in the same fashion as fluorometric excitation spectra. The use of a computer controlled scanning instrument equipped with an AMD 9513 LSI Counter/Timer can also be used to circumvent this problem in cryogenic spectroscopy. The computer can set the wavelength of the monochromator and go into a halt state. One of the 4 unused C/T's of the timer can count down an appropriate equilibration period and then trigger the data collection software. This is a somewhat time consuming process, but it does open the door for synchronous scanning of long- lived phosphors without induction errors. The experimental MS-RTP lifetimes are all very short, generally less than 1-2 ms (5,6,7). Because of the relatively long time (1.5 ms, minimum) it takes the rotating can phosphoroscope to rotate 90°, the short-lived MS-RTP signal died away before the emission window opened. As a result, the spectra were collected in the fluorometric mode. Fortunately, 1-bromonaphthalene does not show any fluorescence in micellar solutions, so the MS-RTP spectrum shows no fluorescence background. This is a problem, however, in the MS-RTP spectrum of 4- bromobiphenyl shown in Figure 23. The MS-RTP appears as a shoulder on the fluorescence spectrum in the fluorometric mode of data collection. This fluorescence background could be alleviated by the use of a rotating disk chopper phosphoroscope which allows signal sampling within 156 lilkflMElNflDflflnf(.k-O UN": ) wAvamcm ( hm. ) Figure 23. MS-RTP spectrum of 4-bromobiphenyl showing overlapping fluorescence band. 157 the time limitations of the computer interrupt service routine. On the twin bus microcomputer this time was 20 us. The Rotating Can Phosphorimeter The first rotating can timing scheme used a complex set of external flip-flops and latches to enable and then save the contents of the AMD 9513 LSI Counter/Timer on successive signals from the photodiode circuitry. This circuit timed the interval between two successive excitation pulses. The computer averaged 100 of these times, divided them by 2, and added a 100 us safety factor offset. The Counter/Timer was programmed to generate a terminal count (TC) pulse after one of the individual counter/timers (C/T) had counted down this number of microseconds. This TC pulse was used as an interrupt signal by the twin bus microcomputer to start the data acquisition software after the 90° rotation from excitation to emission had been completed. Figure 24 compares the digitally collected phosphorescence spectrum of quinine sulfate with that obtained using the analog phosphorimeter. The sample contained 1.2 ug/ml of quinine sulfate (Baker, Baker grade, Phillipsburg, N. J.) dissolved in 95% ethanol. The sample was placed in a cylindrical quartz cuvette and immersed in liquid nitrogen in a quartz Dewar. The excitation monochromator was set at 350 nm (+/- 2 nm), and the emission slit was set at 1 nm. The lack of fine structure in the spectrum is due to the 5 nm wavelength increment used. The spectrum shows that the system does collect spectra quite accurately relative to the analog standard, even with a less than ideal amplifier. 158 am -If- 3 b a n u A : i n l l n I a e- V I 3 m M «0.00 m m 00.00 m I E 2 [-4 2 H ) M l .1 .1L 4* 4 l A A A J u 'r 350 «so 550 WAVELENGTH (on) Figure 24. Phosphorescence spectrum of quinine sulfate at 77 K taken with the digital data system (a). A spectrum collected in the analog mode is included for comparison (b). 159 The decreased chip count, increased reliability, and intuitive simplicity of the second method for timing the rotating can (see Chapter VI) made it preferable to the method which timed each interval between excitation and emission. Figures 25 and 26 show the correspondence between the time-correlated spectrum and the analog spectrum for 2,7- dichlorodioxin (Analabs, N. Haven, Conn.) and biphenyl (Baker, Baker grade), respectively, using this second method. These samples each contained 10 ug/ml of the reagent dissolved in 3-methylpentane (3-MP) (Phillips Petroleum Co., Bartlesville, Ok., purified by stirring over HNO3/HZSO4 (1:1) and distilling over Na ribbon). Both sets of spectra were collected with the emission slit width set to 1 nm. The excitation wavelength for biphenyl was 263 nm (+/- 20 nm). The wide slit was necessary in order to get sufficient radiation to the sensing photodiode to produce a reliable signal. The dioxin spectrum was run with an excitation wavelength of 303 nm (+/- 7.5 nm). Both spectra were taken using the 3 ms filter interposed between the 2 amplifier stages. These data were collected at 5 nm intervals for the dioxin and at 2 nm intervals for biphenyl. The spectrum of the dioxin acquired in the digital mode shows poor fine structure due to the 5 nm interval. This points out an important fact about computerized data acquisition for determining spectra. It is easy to overly simplify a spectrum by the inappropriate choice of wavelength interval. This is a problem that can be described as Nyquist sampling problem in the wavelength domain; the wavelength resolution is limited to twice the wavelength sampling increment. A similar problem of oversimplification of a chromatogram in GC-MS has been examined by Holland et al. (8). The GC-MS problem is one of inadequate time-domain sampling of the time- 160 INTENSITY INTENSITY __J l l 1 r T fi 450 550 650 WAVELENGTH (nm) Figure 25. Phosphorescence spectrum of 2,7-dichlorodioxin at 77 K taken with the digital data data system (a). A spectrum taken in analog mode is included for comparison (b). 161 INTENSITY wmnimmuatmn INTENSITY I l I I J. 450 500 550 600 WAVELENGTH (nm) Figure 26. Phosphorescence spectrum of biphenyl at 77 K taken with the digital data system (a). A spectrum taken in the analog mode is included for comparison (b). 162 varying GC effluent, but both problems arise from the limited bandpass of the detection system. The problem of inadequate sampling of spectra is especially apparent in the context of the limited data acquisition bandpass of the twin bus microcomputer. There is the temptation to speed spectral acquisition by using wider spectral intervals. As more sophisticated computers are applied to the task and averaging on the fly becomes possible, this will be less of a problem. Quantitative Results A better evaluation of the performance capabilities of the computerized data system is provided by a comparison of LDR's and LOD‘s of the computer controlled phosphorimeter with those obtained on the same instrument using the original amplifier and a strip chart recorder. Biphenyl was chosen as the test compound, and the samples were prepared by serial dilution from a 28.6 ug/ml stock solution with 3MP. The excitation wavelength was 263 nm (+/- 20 nm) and the emission wavelength was 504 nm (+/- 1 nm) for all experiments. A 3 ms low pass filter was used to smooth the amplifier signal in the digital system, while a 0.5 s time constant was employed with the analog system. The computer performed the QPHOS routine 5 times and, the results were averaged giving a mean value for 5,000 samples. The analog recording was sampled at 1245 intervals for 1 min, and these values were averaged. Both systems gave excellent linearity with a least squares fit correlation coefficient of 0.99999. The data are compiled in Table 2. 163 Table 2. Comparison of the linearity of biphenyl detec- tion using digital data collection (D.D.C.) and analog detection (A.D.). Relative standard deviations (RSD) are given. Conc (mg/ml), D.D.C. RSD A.D. RSD 0.0 40 4.5 4.16 6.3 26 78 1.4 6.28 6.4 104 165 1.5 12.58 7.6 520 733 0.5 45.56 4.8 2600 3512 0.4 Ave. RSD 1.7% 6.2% The calculated LOD (3 times the standard deviation of the blank) for data acquired in the interrupt driven, or time-correlated, mode was found to be 4.0 ng/ml . Data obtained with the analog instrument gave an LOD of 10 ng/ml. The LOD for data obtained with the time-correlated data system shows that this system had a linear dynamic range (LDR) greater than 650. This approaches the limit of the 12 bit A/D converter which can only resolve 4096 successive values. Figure 27 shows a graphical representation of the data (the error bars are too small to be shown). The reduction in the LOD is in accord with shot-noise theory (2). It should be emphasized that the LDR determined above is for the data collection system only. The phosphorescence signal is often linear over four or five orders of magnitude. It would have been necessary to construct a precision programmable gain amplifier to verify the LDR of the phosphorescence experiment. The LDR for the analog system, assuming a maximum measurable signal of 100 corresponding to the top of the recording chart, would extend from 10 ng/ml to 1100 ng/ml. This limited LDR of two orders of magnitude is a result of the limit imposed by a fixed gain amplifier and recorder. In the case where the amplifiers for each data collection system could be precisely varied and the highest 164 Relative Intensity ( A/D units ) Relative Intensity ( arbitrary units ) Figure 27. Plots of signal intensity vs biphenyl 4000 30009 501K} I ' I ' I ' l . 0 0.40 0.80 1.20 1.60 1 T 1 v I 2TH) 1140 ‘ Biphenyl Concentration ( micrograms/ml ) 40.004 30.00- 2000-1 10.004 i (100 0.00 0.'10 0.'20 0.30 0.30 0.50 0.60 Biphenyl Concentration ( micrograms/ml ) concentration for the digital data system (a) and the analog data system (b). 165 concentration of analyte that could be linearly detected under the constraints of the inner filter effect, the LDR for the experiment would be governed by the LOD. The LDR of the digital data collection system would be 2.5 times that of the analog data collection system given the LOD's determined. Long-lived phosphors The first issue in lifetime experiments to be addressed is the lifetime limit that can be adequately sampled by the at the Nyquist frequency of the digital data acquisition system. If one takes the Fourier transform of the predicted exponentially decaying phosphorescence function, one obtains an expression for a Lorentzian' function: a/(a2 + £2) where a is the reciprocal of the lifetime and f is the frequency component of the signal (9,10). This function is characterized by a width at half maximum so the function is adequately sampled at any frequency greater than 2f. For the voltage sampling mode lifetimes longer than 110 us can be accommodated while the single photon-detection mode could characterize lifetimes as short as 85 us. All of the phosphorescing systems studied had lifetimes in excess of 450 us so the Nyquist criterion is met. However, the fact that the electronic pulse resulting from a photon was only about 10 us in duration implied that a large amount of high frequency noise would be upredictably aliased into the observed signal. The first class of phosphorescence decay curves to be considered are those for species with lifetimes greater than 50 ms. If the free- running data collection system were used for these studies, so many data 166 points would be obtained that the buffer memory would overflow in less than 2 lifetimes (2 lifetimes is used, if possible, as a benchmark for accurately determining lifetimes). In order to circumvent this problem the data were collected in the single photon-detection mode so that only 1 byte was used for each data point (0 or 1). Additionally the data collection system was interrupt driven via the AMD 9513 LSI Counter/Timer so that up to 100 data intervals could be sampled in a given decay curve, and the times for the these sampling windows could be evenly distributed over the desired portion of the decay process. The number of intervals was always fewer than 100 to limit the timing error introduced by the interrupt service routine. Lifetimes were determined parametrically by using the modified simplex algorithm to fit a single exponential curve (See Equation 7.1) to the emission decay data. The decay curve of 2,7-dichlorodioxin (20 ug/ml) in 3-MP served as test system to evaluate the effects of various data collection parameters on the calculated lifetime. The experiments were performed by placing the flashlamp next to the sample holder with the rotating sleeve removed from the phosphoroscope. Representative decay curves are given in Figure 28. These show the qualitative effect of increasing the number of repetitions of the experiment from 10 to 100. Both of these spectra were collected at 20 ms intervals with 250 points sampled per interval (collection time for 250 points was 10.3 ms) An emission wavelength of 500 nm (+/-20 nm) was used. A solvent blank curve is included for reference (1 flash, 20 ms intervals, 250 samples per interval). The dioxin system was used to determine the effect of summing different numbers of experiments before fitting the experimental value of t. This system was also used to characterize quantitatively the 167 em m 00.00 econ m m TIME (ms) m u»- A 1 1p m .. 1» l. I) run m- db 4? 1’ m... -—.—+—.—+—.+.—|—.—[——+—.—q—.-4—.—+—.—4 em M M can use. an TIME (ms) '[ new ~- 0 w -- TIME (ms) Figure 28. Phosphorescence decay curves for 2,7- dichlorodioxin at 77 K showing the qualitative effect of decay curve summing going from 10 times (a) to 100 times (b). A solvent decay curve is shown (c). 168 effect of the length of the time interval and the number of points collected per interval on the calculated lifetime. These data are presented in Table 3. The calculated initial intensity of the phosphorescence, A0 (see Eq. 7.1), is also presented. Table 3. The effect of photon-counting parameters on calculated lifetime of 2,7-dichlorodioxin (20ug/ml). Int. is the interval between data points in ms, Pts./Int. is the number of points collected in each interval for each flash of the source, SSR is the sum of the squares of the residuals of the data about the fit, and A0 is the calculated initial intensity of the phosphorescence. S.D. is the standard deviation. No. Sums Int. Pts./Int. 1 (ms) SSR A0 10 10 125 417 11,200 313 " " " 461 10,800 329 50 " " 437 66,400 1626 100 " " 435 93,600 3190 10 20 250 440 16,500 513 50 " " 422 27,600 2641 Ave. 435 (S.D. = 15) There is no statistically significant variation of the lifetime in these data, based on either of the parameters, if 10 or more decay curves are summed. The simplest comparison of the errors that can be easily computed is the relative average error at time t=0. In this treatment the sum of the squares of the residuals (SSR) is divided by the number of data points and raised to the 1/2 power to give the average error per point (AE). By inspecting the accompanying Figure 29, one can see that the data points in each case are approximately randomly distributed about the best fit. This adds validity to this method. If one then divides this number by A0, one obtains a relative error of the experiments (RE) at time t = 0. These values are collected in Table 4. 169 mu 0" runner ammo mu 0' PMTGIS COUNTED manormm Figure 29. Phosphorescence decay curves for 2,7- dichlorodioxin at 77 K showing that the data is approximately randomly distributed about the fitted line for sums of 10 experiments (a), 50 experiments (b), and 100 experhmente (e). 170 Table 4. Tabulation of relative error (RE) for Table 3. No. Sums Int. Pts./Int. AE RE (%) 10 10 125 10.6 3.4 10 " " 10.4 3.2 50 " " 25.6 1.6 100 " " 30.6 1.0 10 20 250 18.2 3.5 50 " " 23.5 0.9 As shown the RE function decreases with the inverse of the square root of the number of data curves summed for the experiments using 10 ms intervals. The RE function is a simple, inverted signal-to-noise ratio (SNR). Since the SNR increases linearly with the inverse of the square root of the number of values averaged when the experiment is white noise limited (2), it is reasonable to expect the RE function to decrease linearly with the reciprocal of the square root of the number of experiments averaged. This is only an approximate treatment since it assumes that the root mean square (rms) noise will be uniformly distributed about the observed signal. Shot noise theory predicts that the noise signal will be proportional to the square root of the signal (11). A plot of these data is shown in Figure 30. Good linearity is demonstrated (Correlation coefficient, R = .999), as expected in single photon-detection. Similar results will be shown later for the voltage sampling mode of Operation. The RE function indicates that longer sampling intervals with more points summed per interval give somewhat better statistics; however, these data are not conclusive. 171 .ceonpouoHcououh.~ no wooep eoceoweuocmeocm ecu new Acheueuemeu no sense: ecu no uOOu eueooev \ u m> mm mo u0uo .om museum AmZO_._._Euwm LO mmmEDZ LO boom “HE/30$ \ _. 8.8 83 83 coco oooo — _ — — DAV The. (:4) as iAHn 172 Figure 31 shows a representative example of data for 2,7- dichlorodioxin obtained with the free-running data system which sampled the amplifier voltage. It collectied the maximum number of data points allowed by the curve fitting routine (500). The slight rise at short times is due to a contaminant in the sample. It is seen by inspection that the calculated lifetime of the fitted curve can vary widely when using data collected in this type of experiment. For biphenyl, (10 ug/ml), the lifetime results were not as good for small numbers of runs summed. As the number of experiments summed reached 100, however, the value for t (4.35 s) equaled the literature value of 4.3 s (12). All experiments were done with 200 ms intervals and in each interval 300 data points were sampled. The emission wavelength was 500 nm (+/- 20 nm). Table 5 presents these data. Figure 32 is included to show the result of collecting data for 30 ms from a single flash in the free-running mode. It demonstrates the photon shot-noise nature of the phosphorescence signal for long-lived phosphors using this amplifier in an analog sampling mode. Table 5. The effect of increasing the number of experiments summed on the calculated lifetime of biphenyl. # Sums t (5) A0 AE RE (%) 1 3.05 25 3.6 14.4 10 5.88 266 12.5 4.7 50 4.15 776 17.7 2.3 100 4.35 1395 21.6 1.6 Ave. 4.37 173 e>uou nooomacfieco .eHe>ueucu m1 m.mm ue euceom emeue>e ecu we :exeu M be ue auxOACOuOHcoepuh.~ no A9.“ 85. .euouoooosa we oeauuao euep com mouuoeaaoo mucefiuuemxe Nam mo e>u=o meoep eoceomeuocmmocm .Hm enamem .Hoooo .2000. .ooovu .oooou .0000— .0008, _ r 1 d L - P b [1| - J u d N 1- «I. Ch ul- q— uh q- dr- 2 I l g; ALISNHLNI amr..:fiooo 174 .Hecmum oepuooeu cu ewuoc uocm couocm ocwzoce uceEeuemxe ocuumsew emeuuo> eumcfim e scum M or ue Hacecmuc mo e>uoo aeoep eoceoeeuocmmocm .Nm euooem . 2:32.... . a... .88... .88« .88. 88... . . . y r n a 883 133i ___. .Hcoop .Hooou ——_————— l mu H ) uismu numl Qdéon .T 38. 175 Although the average of the lifetimes is very close to the accepted value of 4.3 s, it is obvious that the use of fewer than 50 repetitions of the experiment is not sure to give good results. Figure 33 shows the change in the scatter of the data going from 1 to 100 iterations of the experiment. The high degree of scatter in unaveraged data shows how an improper lifetime assignment might easily be made. Even though the values of t vary widely, the RE function shows the proper form (a 100-fold increase in the number of experiments gives a 10-fold decrease in the RE). The graph of these data (Figure 34) again shows good linearity (Correlation coefficient, R = .999). Rapid Phosphorescence The short-lived lumiphor experiments fell into 2 categories. The first group was a set of simple phosphorescence lifetime experiments performed on brominated biphenyls to test the most basic mode of lifetime data collection. This mode sampled the voltage output of the phosphorescence signal. The samples of 4-bromo and 4,4'-dibromobiphenyl were prepared at 10 ug/ml in 3MP. The emission monochromator was set at 510 nm (4-bromo, +/- 1 nm; 4,4'-dibromo, +/- 4 nm). The instrument collected data continuously for a pre-determined number of points after each flash. In this early version of the system, a software delay was available to adjust the data cycle execution time to 122 us. These experiments illustrated several important points about data handling. The experiments using 4-bromobiphenyl were linearized by taking the natural logarithm after subtracting the offset term, C (see 176 some-- 8 4emm«-- I: la: 5 C ‘2 c Jump [- 2 a. II- C a: 20.000 M an a 2 teen OJNNO : crumb zone 53 tune I— U in: t: D U) z O E; sump a. II- D 8 n E some 1 _L l j I A I I comma r’ . l a s I . r r . . - , menu: ammo «moo some some Figure 33. Phosphorescence decay curves of biphenyl at 77 K using the single photon-detection mode showing the decrease in scatter going from 1 experiment (a) to the sum of 100 experiments (b). 177 n.5- .Hhconman mo wuomc wocmommuonmmo:m wzu you Augeauaummwu mo amass: ecu mo uOOu muosvmv \ H m> mm no aon .vm ousmam AmzoEmnmm .._o Emzaz [.6 Box mm§omv \ P oowd oomd oovd oowd 000.0 P‘ p \b _ . h 00 L b P (z) 33 ofi P 178 Equation 7.1) according to a treatment by Mangelsdorf (13). As can be seen in the data shown in Figure BBC, the value for C could only be approximated from the data. Values of C were chosen to be 200 and 190, and the resulting data were linearized and fit to a straight line using an un-weighted least squares routine. The resulting lifetimes were 18.1 and 19.7 ms, respectively. These data are shown in Figures 35a and 35b, respectively. This demonstrates the problem of calculating lifetimes from linearized data - small variations in the offset can lead to significant errors in t (14). A solution to this problem was to apply the modified simplex algorithm described in Chapter VII. The fitted curve is shown superimposed on the data in Figure 35c. This program gave a lifetime of 16.1 ms which is in good agreement with the literature value of 17 ms (15). The statistical parameters were not recorded for these experiments. A problem of fitting the data to a simple single exponential curve is illustrated by the data for 4,4'-dibromobiphenyl. The solvent is believed to have become contaminated with a very fast decaying phosphor (the Pesticide Research Center is infested with small ants which crawl into everything, and several were seen in the bottle later). This fast phosphor is apparent in the spectrum (Figures 36a, b). Attempts to fit the data using a single exponential model (Equation 7.1) resulted in a lifetime of 7.7 ms. The simulated curve in Figure 36a readily shows that the program weighted the initial values very strongly in doing the fit. This resulted in a lifetime much shorter than the literature value of 12 ms (15). 179 museum-cum) mm'mvu-om) ‘" 1 l 1 I L I s l a l L I ' v ' U ‘ T j I m m m t. n. n ma. TIE(W) I ' I ' I Figure 35. Phosphorescence decay data for 4-bromo-biphenyl at 77 K. Data linearized with the natural log using exponential offset, C, of 190 (a) and 200 (b) show little difference. Simplex fit is shown for comparison (c). Lifetimes from these data are discussed in text. 180 REL INTENSIYY (AID UMTS) m” was mncnoszcouos) ”“ REL. INTENSITY (AID UNITS) l I L l J l I 1 t " V T V ' TME (IICROSECONDS) mammals-I!) Figure 36. Lifetime decay data for 4,4'-dibromobiphenyl with a contaminant showing single exponential simplex fit (a), biexponential simplex fit (b), and single exponential fit to data excluding the first 3.5 ms (c). 181 By using a biexponential decay function (Equation. 7.2) in the fitting program, a much improved fit was obtained (Figure 36b). This program reported a lifetime of 10.1 ms which is in much closer agreement with the literature value. It also reported a lifetime of 0.54 ms for the faster decaying component. An alternative data treatment was to discard the points in the first 3.5 ms from the fitting process and use the single exponential fitting program. This resulted in Figure 36c, and the calculated lifetime was 11.0 ms. This is in good agreement with the literature. Tb3+ Emission in Aqueous Solution at Room Temperature Trivalent terbium ion in aqueous solutions gives a phosphorescence-like emission. The lowest excited state is the 504 state, and the strongest emission is to the 7F5 level in the ground state manifold. Since the transition is not from the triplet state to the ground singlet state, it is not phosphorescence in the sense that was developed for molecular species. However, the phrase "phosphorescence-like" is used to indicate that there is a multiplicity change of 2 and that the emission lifetime is in the same time frame as phosphorescence (greater than 100 us). The terbium system was interesting to consider for two reasons. The first was that the observed ion decay constant was inversely proportional to the number of water molecules in the co-ordination sphere. The excited ion can deactivate through the -OB overtones of water. However, the OD overtones of 020 are inaccessible. By varying the mole fraction of 020, it has been found that the emission decay 182 constant, kEM' ranges linearly from 0.28 ms"1 (I = 3.57 ms) for pure D20 to 2.45 ms"1 (t = 0.410 ms) for pure water where there are 9 molecules in the ionic hydration sphere (16). Duplication of this research would provide a good test of the capabilities of the instrument to characterize very short-lived lumiphors accurately. It would also be a good system to do comparisons of the voltage sampling mode with the single photon-detection mode. Interest in this ion also lay in the use of Tb3+ as a probe of proteins, in particular, the calcium binding protein calmodulin. The calcium ion is non-luminescent, but Tb3+ may be inserted into the protein, and its luminescence may be used as a monitor of the protein structure. One method of doing this is to excite the T133+ and study the luminescence lifetime. In aqueous solutions the luminescence decay will be proportional to the number of water molecules in its co-ordination sphere. This allows the researcher to infer the ability of water to penetrate the binding site and, hence, its structure. The Tb3+ chelating agent diethylenetriaminepentaacetic acid (Aldrich Chemical. Co., Milwaukee, Wi.) was selected as the calmodulin model for evaluation. It was relatively inexpensive and available in the laboratory. It has 5 chelating sites arranged around a flexible organic backbone. It was thought that it would be a good model for the folded structure of a protein site, and it would permit us to determine whether the digital phosphorimeter could produce results for an unknown chemical system that would be correlatable to the existing body of knowledge. All of the time-dependent emission signals for the Tb3+ systems were collected at 545 nm. The width of the slit was varied to minimize amplifier saturation. The slit widths were in the range of 1 - 10 nm. 183 Outside this range, significant amplifier saturation resulted. This is indicated by the sigmoidally shaped decay curve of 1.0 mM Tb3+ in DETPAA shown in Figure 37. The samples were all 1.0 mM in Tb3+, and the solutions containing DETPAA were 1.0 mM in that reagent, also. Table 6 compiles the data for the voltage sampling mode, and Table 7 contains the data for the single photon-detection mode. Aqueous solutions are denoted H20, deuterated solutions are marked 020, and aqueous solutions containing DETPAA are labeled DETPAA. Table 6. Rate data for the analog sampling mode. File Solvent No. Iterations k (msil) t (ms) RE (%) 450 020 512 0.283 3.5 2.4 419 " " 0.345 2.9 1.4 418 " " 0.3346 2.9 2.9 454 H20 " 2.4 0.42 2.4 412 " " 2.23 0.45 3.5 411 DETPAA " 0.55 1.83 0.5 410 " 64 0.55 1.83 1.3 408 " 8 0.55 1.83 3.4 409 " 1 0.50 2.00 13.8 422 " 512 0.55 1.83 1.4 453 " " 0.55 1.83 0.9 Inspection of files 418 and 419 shows some indication of amplifier saturation. In particular, file 419 has a high initial value of about 2300 A/D units. When the amplifier is operated without the 3 ms filter, values this high are usually indicative that it is saturated much of the time. These effects can be visually quite subtle, and no reliable method to detect amplifier saturation effects has been found, other than careful inspection of the graphical data. This problem and its effects on the reported lifetime are considered in more detail, shortly. 184 .Aommsm w>ucu ampfioemwmv m: ooom v u use» an coeueu500w uwfiuflamew mo avenue 0:» ouoz .AE: m s\+v E: mvm we pwuooafioo <uco recon cmemHEm mucumuwmeou Boom .nm ousofim 33 ms: LLISNHLNI 185 Table 7. Rate data for the single photon-detection mode. File Solvent No. Iterations k (mail) tyLms) RE (%) 451 020 500 0.29 3.4 3.3 420 " 300 0.285 3.5 3.8 455 820 500 2.2 0.46 3.4 456 " 1000 2.4 0.42 1.1 452 DETPAA 500 0.51 1.95 3.3 The data for curves 409,408, 410, and 411 in Table 6 indicate that the RE function was roughly inversely proportional to the square root of n, the number of times the experiment was repeated. This indicates that the experiment is in the photon shot-noise limit. Figure 38 shows a plot of the RE function versus the inverse of the square root of the number of times the experiments was repeated. This graph shows good linearity when the data point arising from 8 repetitions is discarded from the fit (Correlation coefficient, R = .996). Two additional decay curves were obtained for the deuterated solution under different experimental conditions which give similar decay constants. One experiment was performed with a Corning CSO-54 290 nm cut-off filter (Corning Glass Works, Corning, N. Y.) interposed between the flashlamp and the sample. The other was performed by mounting the flashlamp cradle at the entrance slit of the excitation monochromator and passing the radiation through the monochromator. The monochromator was set at 350 nm (+/- 10 nm). Both experiments were done at 545 nm with the emission slit set at 40 nm. In each case the experiment was repeated 256,000 times in the single photon-detection mode. The experiment using the excitation monochromator yielded a decay 186 .4dmemo\+mnfi ~00 AchwuHuwmom mo gonad: ecu no uOOu onwawmv \ H m> mm 00 an0 .mm AmzoEEnmE no mmmzaz no 89. mm§omv \ P 000.0 000.0 r L P - 00_v.o 00-N0 000.0 new aouu pevusomav acaoe much O 0.0 069 omnmwm (%) 38 187 constant of 0.294 ms'1 (t = 3.4 ms), and the filter experiment gave k = 0.286 ms’1 (t = 3.5 ms ). The problem of establishing the proper fitting window is of critical importance. Table 8 shows the effect of varying the fitting window for the data from a single experiment on the decay constant and lifetime of Tb3+ in aqueous solution. Table 8. The effect of the fitting window on the calculated values of REM and t for 1.0 mM Tb3+. Lower Lim.(ms) Upper Lim.(ms) k (msli) 1 (ms) RE(§) 100 5000 1.6 0.63 3.0 500 5000 1.9 0.53 0.8 750 5000 2.1 0.48 0.6 1000 5000 2.5 0.40 0.3 Figure 39 shows the plots of the data used to generate Table 8. It demonstrates the need to visually inspect the curve fitted to the experimental data in order to verify the fit. In this case it reveals that, although the lifetime reported for the window extending from 1,000 to 5,000 us is closest to the literature data, it is not the best fit for the experiment. The curve obviously does not fit the high intensity data although the RE function, which is based on the residuals of the fit, indicates that it is the best fit. The values for the decay constant of the solutions containing 1.0 mM DETPAA are remarkably consistent. Based on the previously mentioned linear relationship between k and the number of water molecules in the hydration sphere, the average of the data in the 2 tables suggests that Tb3+ chelated with DETPAA experiences the field of 1.1 molecules of water in its co-ordination sphere. If one uses just the data from the 188 .m: 000." v u OEHH uu 0 cu cusp cu uuu moon one we m.~ uaone us A use 0 ne>h=u cu uuu no uoauo ouuueoueau ecu euoz .6... 3.6 u p .2. 9m .. o.“ 8 as 2.6 u .p .2. 9m - 2.6 3 as and u p is 6.... - m6 3 as no.0 a w .eE o.m u «.0 "soccer as ”use cusp suntan e :0 300cm: mcnuuwu Nounede ecu uo uueuue ecu sous nuo~m .ueues cu +nAH :5 o." no aeoev caeeeuse eucueuoméeu zoom .mn eusmum “ISM! : “I’m-II .a “ISM! so “I”! no 189 single photon-detection experiments these results indicate that 1.0 water molecule is in the co-ordination sphere of the ion. Terbium has been found to have 2.8 water molecules in its hydration sphere with the related chelating agent EDTA (16). DETPAA has 2 acetic acid chelating moieties on each end of a 7 atom chain and 1 in the middle giving it 1 more than EDTA. This, coupled with the flexibility of the skeleton, would allow DETPAA the molecule to fold around the Tb3+ ion and effectively shield it from the bulk solvent, so this is a reasonable interpretation of the data. 190 10. ll. 12. 13. 14. 15. 16. CHAPTER VIII Malmstadt, H. V.; Enke, C. G.; Crouch, S. R., Electronics and Instrumentation for Scientists, Benjamin/Cummings Pub. Co.:Reading, MA. 1981; p. 406 -407. Malmstadt, H. V.; Enke, C. G.; Crouch, C. G., ibid, 415 - 416. Kalyanasundaram, K.; Grieser, F.; Thomas, J. K., Chem. Phys. Lett., 1977, 51, 501. Lecture notes, Professor S. R. Crouch, Chem. 835, Michigan State University, E. Lansing, MI., 1981. Cline Love, L. J.; Skrilec, M., Anal. Chem., 1980, 52, 1559. Cline Love, L. J.; Habarta, J. G.; Skrilec, M., Anal. Chem., 1981, 53, 437. Skrilec, M.; Cline Love, L. J., J. Phys. Chem., 1981, 85, 2047. Holland, J. F.; Enke, C. G.; Allison, J.; Stults, J. T.; Pinkston, J. D.; Newcome, B.; Watson, J. T., Anal. Chem., 1983, 55, 997A. Bracewell, R. N., The Fourier Transform and Its Applications, McGraw—Hill Book Co.:N. Y., 1978; p. 392. Champeney, D. C., Fourier Transforms and their Physical Applications, Academic PresszLondon, 1973; p. 22. Malmstadt, H. V.; Enke, C. G.; Crouch, S. R.; Horlick, G., Instrumentation for Scientists: Optimization of Electronic Measurements, Module 4, W. A. Benjamin, Inc.:Menlo Park, CA., 1974; p. 14. Harbaugh, K. F.; O‘Donnell, C. M.; Winefordner, J. 0., Anal. Chem., 1973, 45, 39. Mangelsdorf, P. C., J. Appl. Phys., 1959, 30, 1443. Woods, R. J.; Scypinski, S.; Cline Love, L. J.; Ashworth, H. A., Anal. Chem., 1984, 56, 1395. O'Donnell, C. M; Harbaugh, K. F.; Fisher, R. P.; Winefordner, J. 0., Anal. Chem., 1973, 45, 609. Horrocks, W. deW., Jr.; Sudnick, D. R., J. Am. Chem. Soc., 1979, 101, 334. 191 CHAPTER IX Conclusions and Future Prospects There are two general areas of evaluation into which the results of this research fall. The first is a review of the hardware that was developed and some proposals for their improvement. This focuses on 2 sub-divisions, the amplifier and the phosphoroscope. This latter topic also considers various difficulties encountered with the radiation sources. The second major area of discussion is a commentary on the data acquisition computer and data processing software. This includes a discussion of the shortcomings of the poly-FORTH system, possible advantages accruing to the use of the HSFORTH operating environment, especially on an IBM PC type of computer. It includes a proposal for improvements in the simplex curve fitting routine, particularly with reference to an evaluation of the statistics of the fit. The Amplifier The intended goal of this research was to take a commercial analog spectrofluorometer/phosphorimeter and develop a flexible, modular amplifier/data acquisition system. This would allow computer control of the instrument and computerized data collection. This system was to be constructed so that it could be quickly removed, and the original functionality of the instrument could be restored so that it could be used for teaching purposes. 192 These goals were met and their success has been documented in Chapter VIII. The instrument has successfully collected fluorescence and MS-RTP spectra in a fluorometric mode, where data collection was performed randomly in time. Cryogenic phosphorescence spectra were collected in a time- correlated mode, where the AMD 9513 LSI Timer/Counter chip was used to time the rotating can phosphoroscope and then to generate an interrupt pulse. The latter synchronized data collection with the presence of the phosphorescence at the PMT. The efficacy of this strategy was demonstrated by the improvement of the LOD of biphenyl, relative to the original analog detection system. The application of the faster amplifier to both short and long- lived phosphors, using a simple, box-car type of single photon-detection strategy was also successful. By simply choosing the appropriate software driver lifetimes over 4 orders of magnitude were accurately characterized. A similar box-car strategy of sampling the signal from short-lived lumiphors (t greater than 200 us) also proved to be effective. There are some serious problems with the system which do need to be addressed. Most of these revolve around the improper matching of the bandwidth and gain of the amplifier to the data acquisition properties of the twin bus microcomputer and the AD 574 analog-to-digital (A/D) converter. For voltage sampling of short-lived phosphors, the simultaneous amplification of photoelectrons from more than 2 photons at the gain used resulted in amplifier saturation. The narrow pulse width of the photoelectron signal often meant that the sampled signal was at the amplifier null, even under fairly intense illumination. This resulted in low dynamic range of the amplifier and gave artificially low - 193 average intensities. The high gain of the last stage also made the amplifier subject to severe baseline drift. The proto-type design of the amplifier also resulted in poor ground shielding which led to significant 60 Hz noise. Finally, the need to place the amplifier in close proximity to the data acquisition board, which was housed in the control module of the instrument, dictated that the C/V circuit be separated from the amplifier. This necessitated the use of more chips, both to drive the signal over that distance and to condition the signal to the proper polarity for the A/D converter. There are several modifications of the existing system that would alleviate many of these problems and simplify the overall design. They consist of integrating the C/V, amplifier, and data acquisition system onto one board that could be placed in a well shielded box under the instrument. The control signals could then be ported by a single cable to the control module of the instrument. This would facilitate the use of a solid common-ground plane to minimize the 60 Hz noise. Reduction of this noise component might permit as much as an 80% reduction in the gain of the amplifier while still giving a good single photon signal. Such a modification would enable the discriminator to differentiate photon pulses from the background, a capability necessary for single photon-detection of long-lived phosphors, while increasing the dynamic range of the amplifier, which is important in sampling the voltage of the signal. This strategy would lower the chip count in the amplifier section and improve reliability. The larger box, housing a unified system, would permit the use of more precise wire-wound trim-pots allowing finer and more stable trimming of the amplifiers. Finally the use of a DC-servo 194 feedback loop might be considered to stabilize the direct current component of the amplifier output (1). The amplifier should also be given a longer time constant, so that the photon event more nearly corresponds to the sampling window of the data acquisition package. The data collection rate of the system can be precisely determined, and a low pass filter of the appropriate time constant inserted at the input of the second stage of the amplifier. If single photon-detection can be compared to photon-counting, the ratio of the pulse width to the width of the sampling window should be adjusted between 0.5 for low signal strength and 0.99 for strong signals (2). Since most time decay curves vary from a high signal to background ratio (SBR) at the onset of data acquisition cycle to a low SBR at the end, a pulse width to data acquisition time ratio of 0.8 - 0.9 would be a good general purpose value (the pulse width in this case is the time that the signal voltage is greater than the discriminator voltage). This would enable the sampling mode to still operate in the photon shot noise limit at low light levels with no loss of signal intensity. At higher signal levels the amplifier would be less susceptible to saturation. The Phosphorosc0pe This research has shown the ability of the system to collect time- correlated data with good precision, for long-lived phosphors using the rotating can phosphoroscope. However, for MS-RTP spectroscopy and other solution-based RTP methods, the lifetime of the phosphor is so short that the signal has completely Or nearly completely decayed by the time the rotating can has rotated 90°. This necessitated the collection of 195 MS-RTP spectra in the fluorometric mode with an unchopped source. As a result, the MS-RTP spectra were often superimposed on the trailing shoulder of a fluorescence band. The use of a rotating disk phosphoroscope would alleviate this problem. The data collection cycle could start immediately upon closure of the excitation window. This would allow a higher signal-to-noise ratio to be easily established for scanning and quantitative work, since the signal is stronger the sooner it is sampled after the excitation process has been terminated. It would also make it easier to collect time decay curves for short-lived lumiphors since the noisy, unreliable pulse lamp could be eliminated. A variable duty cycle could easily be accommodated by the simple expedient of inserting a different mask with a different number of slits. This could allow the experimenter to access different lifetimes. Data collection would become limited by the data handling procedures, rather than the present duty cycle limitations of the pulse lamp power supply. The fluorometer upon which this research was based, was equipped with a beam splitter, a sample cell for holding a quantum counter solution, and a second PMT. This system could be effectively implemented into a source compensator. Xenon arc wander is a common source of flicker noise in these systems. The are lamp in this fluorometer also showed a sizable 120 Hz noise component. This was apparently due to poor regulation of the lamp power supply. Correction for these noise sources in the spectroscopy of short-lived phosphors would greatly enhance noise reduction and improve the LOD's of experiments under these conditions. In those cases where the lifetime is long, a good mechanical shutter would substitute well for the flashlamp. A shutter has the advantage that the sample can be exposed for an extended period of time, 196 so that the sample is in phosphorescence equilibrium before the shutter is closed. This provides a stronger signal. It was seen by inspection that the flashlamp was unable to excite the samples appreciably. These samples showed strong phosphorescence for several seconds when excited to phosphorescence equilibrium using the rotating can phosphorimeter. In all cases the Xenon Corp. flash lamp system was seen to be a poor light source for molecular phosphorescence spectroscopy. It was a constant source of extraneous signals which caused the data system to crash. It had a low repetition rate which limited the rate of collection of lifetime data for short-lived phosphors. It was a bulky system which intruded into the operating environment of the sample compartment. A final note on the advantages of the use of the chopper over the rotating can phosphorOSCOpe is that it allows the use of a sealed fluorescence cell for RTP spectroscopy. Since the chemical de- oxygenation scheme described in Chapter V works well in such a cell plugged with a simple teflon stopper, the cumbersome cells presently used could be discarded. This would greatly ease the problem of cleaning the cell which was a constant problem with the current cell. Finally, the quartz in commercial cells made from Suprasil silica shows none of the residual, short-lived emission found in generic quartz glass. This residual emission was found to be strong enough to interfere with the determination of the MS-RTP lifetime studies. FORTH Systems Two of the largest drawbacks of the final data system were the unreliability of the floppy disk drive controller used with the twin bus 197 microcomputer and the block structured file system used by poly-FORTH. Disk failures resulted in many hours of data lost due to disk crashes. The block structured poly-FORTH files resulted in a plethora of floppy disks where data were stored in a fashion which was not immediately decipherable using the operating system. Poly-FORTH does not support named files. A viable solution would be to transport the system to an IBM PC or PC-AT type of computer using the HSFORTH operating environment. The disk systems on these computers are highly reliable, and HSFORTH supports file structured storage. A number of graphics packages are currently available in Professor Crouch's research group which permit on-line visual analysis of the data. The richer register structure of the microprocessors in these machines enables the programmer to use more sophisticated averaging schemes, and the faster clock used in these computers facilitates greater data acquisition rates. Finally, a FORTRAN program has been written which uses the simplex algorithm to extract an experimental lifetime from the decay data (2). The computer that collected the data could perform the data reduction. This would greatly increase the data throughput over the present cumbersome system where data must be sent via a serial line to a minicomputer for reduction. This port of the system has already been essentially implemented in the FORTH-Port board described in Chapter VI. The use of a PC compatible computer would also make the presentation of hard copies of the data easier. The software to drive the HP 7475 plotter from a PC has already been written in this laboratory, and it would no longer be necessary to transfer data files between computers. 198 Data Reduction The simplex data fitting routine used in this research proved to be a powerful method of determining phosphorescence decay constants and lifetimes. This was especially true when the data exhibited a baseline which made linearization difficult. However, at present, this routine does not return a truly useful statistical analysis parameter. Since the instrument often works in the shot noise limit, it would be useful to modify the program to calculate the noise envelope of the data based on this noise model. This could allow a quick visual inspection of the data to ascertain whether a given fit was in a time window that was in the shot noise limit. This modification might be easily implemented from the sum of the squares of the residuals (SSR) value which the program currently returns. Since inverse of the square of the residual is proportional to the signal intensity in the shot noise limit (3), an exponentially weighted statistic could be generated to provide a noise envelope. Extrapolation of this envelope to time t = 0 would give a more accurate RE function. The envelope could also be used to verify visually whether any other noise factors, such as source flicker or 60 Hz interference noise, were major contributors to the goodness of the fit. The spectrometer proved to be versatile and effective. The computerized data acquisition system was removed, and the original analog detection with strip chart recording was restored in about 2 min. The data derived from the digital data system showed good correlation with data obtained through a number of other experimental methods. 199 Finally, the experiences gained in this research point to a number of inexpensive but powerful improvements that can be easily made. The system should then be capable of routinely performing a broad range of luminescence experiments. 200 CHAPTER IX References 1. Private communication with Mr. Kevin Blair, electronics designer and owner, Buggtussel Electronics, 2299 Knob Hill, #17, Okemos, MI. Nov., 1986. 2. Newsham, M., SIMPLEX, a FORTRAN curve fitting program based on the simplex algorithm which is designed to run on an IBM PC compatible microcomputer, Michigan State University, E. Lansing, MI., 1985. 3. Darland, E. J.; Leroi, G. E.; Enke, C. G., Anal. Chem., 1980, 52, 714. Additional References (Photon-Counting) Candy, 8. H., Rev. Sci. Instrum., 1985, 56, 183. Candy, 8. H., Rev. Sci. Instrum., 1985, 56, 194. Darland, E. J.; Leroi, G. E.; Enke, C. G., Anal. Chem., 1979, 51, 240. Hayes, J. M.; Schoeller, D. A., Anal. Chem., 1977, 49, 306. Wittig, B.; Rohrer, F.; Zetzsch, C., Rev. Sci. Instrum., 1984, 55, 375. Jameson, D. M.; Spencer, R. D.; Weber, G., Rev. Sci. Instrum., 1976, 47, 1034. Meade, M. L., J. Phys. E: Sci. Instrum.,1981, 14, 909. Niemczyk, T. M.; Ettinger, D. G., Appl. Spectros., 1978, 32, 450. Niemczyk, T. M.; Ettinger, D. G.; Barnhart, S. G., Anal. Chem., 1979, 51, 2001. Stuart, D. C.; Kirk, A. 0., Rev. Sci. Instrum., 1977, 48, 186. 201 APPENDIGS APPENDIX A A Description of the Data.Acquisition Board Introduction The description of the data acquisition board is divided into 2 sections. The first covers those signals which originate at the computer and control the instrument. The second group describes the paths of signals which originate in the spectrometer and go to the computer. These are the data lines and the signal from the photodiode sensor. Each of these sections is accompanied by a pictorial diagram. Every signal passes through an optical isolator (6N137, Hewlett- Packard, Palo Alto, CA.). The input to each of these devices is at pin 2 and the output is at pin 6. Each of the optically isolated outputs is connected to a 470 Q resistor. The description will only specify which isolator is used in a given sub-circuit. All TTL chips are of the LS variety. Table 9 gives a description of all of the chips used used to process the signal into a digital format. The function of each signal will be given with its FORTH name and hexadecimal value in parentheses when applicable. Computer sourced signals begin at the pin number on the edge connector bringing the signal from the computer. Data descriptions will start at the appropriate pin on the analog-to-digital converter (AD 574, Analog Devices). When a signal is processed through a chip, the input pin and 202 the output signal pin will be given in succession separated by a comma as in the following example: 026 p.2, p.18 The logic performed on the signal will not be indicated, but it can be determined by consulting any standard reference on TTL logic. Table 9. A list of the chips associated with Figures 40 and 41. Pictorial Designation Chip Type 01 - 020 6N137 025, U27, U28, U31, U34, 039 74LSO4 029, U30, U38, U45 74LS74 032, 040, 046 74LS373 037, 044 74L8121 026 74LSZ45 033 74LS14 035 74L332 036 74LSl74 041 MK808 MUX 042 AD 582 SHA 043 AD 574 A/D The Path Description Computer Generated Signals (See Figure 40): Emission Clutch Set(EMSET:BDCO); m80 -> 027 p.13, p.12 -> 011 p.2, p.6 -> 029 p.1, p.5 Scan higher(HIGHER:BDC8); m81 -> 027 p.11, p.10 -> 014 p.2, p.6 -> 029 p.13, p.9 Scan lower(LOWER:BDDO); 180 -> 027 p.9, p.8 -> 016 p.2, p.6 -> 029 p.11, p.9 Excitation clutch set(EMSET:BDD8); 181 -> 027 p.1, p.2 -> 08 p.2, p.6 -> 030 p.13, p.9 Excitation Clutch release(EXREL:BDEO); k80 -> 027 p.3, p.4 -> 010 p.2, p.6 -> 030 p.10, p.9 203 .cueon :oflufimfiovow cusp ecu uo cenuuom Hmcmwm Houucoo ecu no :ofiumucwmmumeu HmHMOuUem < .ov unseen .. o. C O C D O I 3 S. a... 2.. 2. 2.. .S 2. =9 .2. o u . _ _ . . _ . _ . . II M o u- q 3 .- O . .- .- a 3 e- .- .- 8 -~ .- 3 - n u- 3 .- O .- 5 — o. is 3 .- IUuaau an I i _ 4 5:34:32: r — 89:30 la A 3: 353.3236... 2.22:1. Y'lll'll'l'. neuzsc.xuoxoa .u . coeeaossuoses In E a; 4:80; L . a'a'.'. - 204 Emission Clutch Release(EMREL:BDE8); k81 -> 027 p.5, p.6 -> 013 p.2, p.6 -> 029 p.4, p.5 Step stepper motor 1/2 step(STEP:BBDO); f81 -> 025 p.5, p.6 -> 02 p.2, p.6 -> 033 p.3, p.4 -> 038 p.11, p.10 Flash the pulse 1amp(POLSE:BBFO); f80 -> 05 p.2, p.6 (Not used) A0; h81 -> 040 p.3, p.2 -> 025 p.1, p.2 -> 01 p.2, p.6 -> 043 p.4 (High/low byte) The following 3 signals serve multiple functions. Set multiplexer channel(CHSET:BDFO); j80 -> 025 p.13, p.12 -> 046, p2 (Latch data used as multiplexer inputs) j80 -> 025 p.13, p.12 -> 06 p.2, p.6 -> 033 p.5, p.6 -> 044 p.3, p.5 -> 036 p.9 (Select channel determined by latch outputs) A/D convert(CONVERSION:BDF8); j81 -> 025 p.11, p.10 -> 03 p.2, p.6 -> 030 p.5, p.6 -> 042 p.3 (Sets SHA) j81 -> 025 p.11, p.10 -> 03 -> 043 p.5 (Convert) j81 -> 025 p.11, p.10 -> 035 p.2, p.3, -> 040 p.11 (Permits short cycle conversions on BDF9) Read data(MSB, LSB: BBF8, BBF9); h80 -> 026 p.19 (Drive data on the computer bus) h80 -> 034 p.1,p.2 -> 07 p.2, p.6 -> 033 p.9, p.8 -> 032 p.11 (Latch A/D output) h80 -> 035 p.1, p.3 -> 040 p.11 (Latch A0 so that high or low byte is presented by A/D) The following 3 signals are data inputs to the multiplexer used to set the analog input channel (See Figure 41). DO; n80 -> 046 p.3, p.2 -> 034 p.13, p.12 -> 017 p.2, p.6 -> 036 p.11, p.10 -> 041 p.15 205 . pueon cowuumesvoe euep ecu mo coHuuom Heaven euec ecu mo :oHueucemeumeu HeeuOuoem< .av eusmfim :8 -d-. I 4d . mun. — 1 DPFDFI 141“ a: - 1 ’P’F ‘1.“‘ 0'. r» 1— dd‘d‘dd 206 D1; n81 -> 046 p.4, p.5 -> 034 p.11, p.10 -> 020 p.2, p.6 -> 036 p.13, p.12 -> 041 p.16 DZ; q80 -> 046 p.7, p.6 -> 034 p.9, p.8 -> 019 p.2, p.6 -> 036 p.14, p.15 -> 041 p.1 The Signals From The Spectrometer (See Figure 41): DO; 043 p.20 -> 032 p.18, p.19 -> 031 p.13, p.12 -> 09 p.2, p.6 -> 026 p.2, p.18 -> n80 01; 043 p.21 -> 032 p.17, p.16 -> 031 p.11, p.10 -> 012 p.2, p.6 -> 026 p.4, p.16 -> n81 02; 043 p.22 -> 032 p.14, p.15 -> 031 p.9, p.8 -> 015 p.2, p.6 -> 026 p.4, p.16 -> q80 D3; 043 p.23 -> 032 p.13, p.12 -> 031 p.1, p.2 -> 018 p.2, p.6 -> 026 p.5, p.15 -> q81 D4; 043 p.17,25 -> 032 p.7, p.6 -> 031 p.5, p.6 -> 022 p.2, p.6 -> 026 p.7, p.13 -> r81 05; 043 p.16,24 -> 032 p.8, p.9 -> 031, p.3, p.4 -> 021 p.2, p.6 -> 026 p.6, p.14 -> r80 06; 043 p.18,26 -> 032 p.4, p.5 -> 033 p.13, p.12 -> 023 p.2, p.6 -> 026 p.8, p.12 -> 881 D7; 043 p.19,27 -> 032 p.3, p.5 -> 033 p.11, p.10 -> 024 p.2, p.6 -> 026 p.9, p.11 -> $80 Photodiode; 04 -> 034 p.3, p.4 -> e81 207 APPENDIX B A Table of Chip Selects Used in the Twin Bus Microcomputer A total of 22 chip select signals (CS) were used in the phospho- rimeter. Fourteen of the CS's were used by the data collection hardware, and eight were used for microcomputer system support. Table 10 gives a list of these signals with a brief description of their functions. Table 10. A list of the CS signals used in the computer controlled phosphorimeter. The FORTH word generating the CS is also given. A (*) designates a microcomputer system support function. CS FORTH WORD Functigg BBDO STEP Stepper Motor Movement BBFO Pulse Pulse to Flash Lamp Trigger BBF8 MSB Latch Most Sig. Byte from A/D BBF9 LSB Latch Least Sig. Byte from A/D BDCO 2RES Set Emission Clutch BDC8 HIGHER Scan Selected Monochromator to Longer Wavelength BDDO LOWER Scan Selected Monochromator to Shorter Wavelength BDD8 3RES Set Excitation Clutch BDEO BSET Release Excitation Clutch BDE8 ZSET Release Emission Clutch BDFO CHSET Select Multiplexer Channel BDF8 CONVERSION Start A/D converter Conversion BFFO COMAND AMD 9513 Command Register BFFl DATAREG AMD 9513 Data Register BEOO(*) UODATA Terminal UART Data Register BE01(*) UOCMD Terminal UART Command Register BE40(*) UlDATA PDP-ll UART Data Register BE41(*) 01CMD PDP-ll UART Conmand Register BE80(*) INTCONT Interrupt Controller Command Register BE81(*) IMSK Interrupt Controller Data Register BFCO(*) <1771-CS> Floppy Disk Controller Enable BFC8(*) (DRIVE-CS) Floppy DIsk Data Latch 208 APPENDIX C The Design of the IBM PC Cbmpatible FORTH-Port Introduction The Design of the IBM PC Compatible FORTH-port contained an address decoder which generated a negative going chip-select (CS) pulse whenever an input/output (I/O) command was sent to that address. The FORTH-port was equipped with three sets of 16-bit I/O ports which could be selected as two 8-bit ports. Each of these features will be discussed in some detail. The third major component of the board was an AMD 9513 LSI Counter/Timer chip with an 8-bit port. This chip was included in the same format as implemented on the twin bus microcomputer and will not be discusssed in great detail. The Chip Select Decoder The IBM PC compatible FORTH-port decoded the 20h addresses reserved for development work into the same number of chip select (CS) signals. Signals that are normally high will be denoted (NH) in this Appendix, and signals that are normally low will be designated (ML). The CS signals were decoded in a manner analogous to that used on the twin bus microcomputer (Reference 9, Chapter VIII) using 74136 Exclusive-OR gates. The CS (NH) signals were read/write enabled by the exclusive-OR 209 of the IOR (NH) and IOW (NH) signals from the PC bus. The 13 most significant address lines, A5 (NL) through A17 (NL), were also decoded using the 74136 exclusive-OR gate. Address lines A8 and A9 were decoded as the exclusive-OR of the signal and ground. The remainder of these address lines were decoded as the exclusive-OR with 5 v. These decoded addresses were collected in common with the output of the read/write signal, and the resultant signal (NH) was used as one of the enable signals, E0 (NH), of each of the 74154 1-16 Decoder chips. The other enable, E1 (NH), of 74154 chip #2 was connected to the exclusive-OR of address line A4 with 5 v (NL), and the second enable of 74154 #1 was connected to the NOT of this signal (NH). Whenever the address 03XXh (x is a don't care value) appeared on the address bus, both decoder chips were enabled at E0. Under those circumstances, A4 could be high enabling chip #1 (base address 300h)' or could be low enabling decoder chip #2 (base address 310h). The four lowest address lines, A0 through A3, were connected directly to each of the 74154 chips. When either of the two chips were enabled, the contents of these four lines was decoded to provide a CS signal. When 74LSlS4 #1 was enabled the CS would correspond to one of the addresses 3°0h to 30Fh, and when 74LSlS4 #2 was enabled the CS signal would be derived from an address in the range 310h to 31Fh. A pictorial diagram of this portion of the board is shown in Figure 42. Address lines are not shown coming from the edge connector. They are shown at the location on the 74136 where they are utilized. 210 ' .uuomumsmom eHcHummEOU um zmH ecu mo caeuuom .mu. nepooep uoeaem mace ecu ecu coflumuceweumeu Heauouoam d we encode op< op< .—.._.......s.................2..........123.228.8. rehomzzoo menu 30. I. .5 new < new m n. I m m 10--.--. L-;--.L--.:. I... a. a... 3. 63:: i 2:: a. of: n. 337-. <::H...... HkmMuwfil Husk—am...» K m K m uonloon mo hunlo.n mo 1:: 2...: 2 2 2. 2 .1 3 S. .1 S. 9. E on 211 The Port Three l6-bit ports were included and each consisted of 2 74373 octal latches with sequential addresses. One set (74373 #‘s 1 & 2) was constructed to port data from the board, and the enable pins (NL) had addresses 318h (NH) and 319h (NH) which had been conditioned to the proper, (NL), state by a 7404 hex INVERTER chip. This was a write only facility whose output was always enabled. Another port (74373 #‘s 3 & 4) was identical except that the output enable signal could be strobed by an external signal or another CS. Data were sent to these latches using CS's 31Ch (NH) and 310h (NH) which had been conditioned to the proper (NL) state using a 7404 Inverter chip. The final port (74373 #‘s 5 & 6) was an input port. Data could be strobed into the latches by the PC, using the CS signals 315h (NH) and 314h (NH) which had also been inverted to the correct logic level using a 7404 Inverter chip., or by an external signal, such as an end of convert pulse from an A/D converter or a TC pulse from an AMD 9513. The computer could then activate the output enable pin of the 74373 (NH) using CS's 316h (NH) and 317h (NH). The AMD 9513 LSI Counter/Timer chip was mounted next to a socket for a crystal oscillator. The socket allowed easy change of the crystal so that the time base could be varied. Data were ported to and from the AMD 9513 chip through a single 74245 octal transceiver. The read/write (read = low) function of the 245 was controlled by the IOR (NH) pulse from the computer. The chip was enabled by the OR function of the NOT functions of CS's 31Eh (NH) and 31Fh (NH). This same signal was used to enable the AMD 9513 LSI Counter/Timer I/O port. These two addresses have 212 the A0 line low and high, respectively. Thus, data written with CS 31Fh is sent data to the Command Register of the AMD 9513 chip and data written with CS 31Eh is sent to the Data Register of this chip. A pictorial diagram of the port structure for the IBM PC compatible FORTH port is shown in Figure 43. The CS signals which are unmodified come directly from the 74154 chips, and the complete connections are not shown. The paths for the CS signals which must be inverted are shown starting at the 7404 hex INVERTER where the logic conditioning was done. Data paths are not shown. 213 .uesee\ueucaoo Hm; mumm axe ecu ocucsuocu .uuoe-=emoc eacfiuemeoo om :mH ecu mo cOuuuom uuom 0\H ecu mow cOwueucemeumeu HeHAOuon <.mv euomwm athOLuz m. x wzh >420 Om x—noo mmwz00< ozeh¢1?! ! 28 WORD HUBER; (KEYED. INPUT) 11 (USEAS"N"DD.IN. ITWILLACCEPTNCRARSPOLWEDN/.) 12 :JIMPBNAIT STEP WAIT STEP: (ICGIPLETE STEPPER STEP) 14:.11MPESODOJWBLOOP: (DONSTEPSOFSTEPPERWTOR) 15 DECIMAL 218 Block Nuiaer: 14 0 ( PLUORGIETER SETUP CONT. ) HEX I 2 : LAMBDA CONVERSION 16B DROP "SB 10 ‘1 LSB DROP LSB 10 / OR ; 3 4 ( INITIATES CONVERSION, COLLECTS MSB. DISCARDS FIRST ERRONEOUS 5 VALUE, ADDS 4 BITS. OOLLECTS LSB SAME WAY. STRIPS TBAILIM 0’s 6 COMBINES RESULTS INTO ONE WORD ) 7 8 DECIMAL 9 10:NAVEZOO-DUPDUP4tSWAP10/+SW0250/-: 11 12:NOSTEPSDUPDUP58$WAPB/-SWAP128/+; 18 14 ( ROUTINES TO CALCULATE WAVELENGTH USING 10V NAME ON nono- 15 CERCMAIOR AND 4096 VALUES m A/D ) Block Number: 15 0 ( BASICS 0? AMD 9513 PROGRAMING ) HEX 1 BPFI CONSTANT CGIAND ( ADDRESS OF 9513 COMAND PORT ) 2 3 BFPO CONSTANT DATAREG ( ADDRESS OF 9513 DATA PORT ) 4 5 :OJD(N—>)CMANDC!;(STORESCQHANDSIN9513) 6 7 :M—M17CMD; (CALLSUPM—MREGISTER) B 9 :DRG(N—>)DATAREGC!; (STORESDATAINQSIS) 10 11 :(DRG(->N)DATAREGCODATARE009256$+; 12 13 ( 2 BYTES man 9513 DATA REG. W BYTE FIRST. INTO ONE WED ) 14 DECIHAL 15 Block Nulber: 16 0 ( BASICS OF AMD 9513 PROGRAMING ) HEX 2 :ZBYTES DUP 0< IF DUP 32767AND256/128+ELSE DUP256/ 3 TEEN SWAP 255 AND ; g ( BREAKS DATA WORD INTO TWO BYTES, LSB FIRST, MSB SECOND ) g : >DRC ZBYTES DRC DRG ; g ( ONE 16 BIT WORD INTO DRG AS TWO 8 BIT WORDS, LSB FIRST ) 11:RESETPPCHD5PQID17GID: I: ( RESETS AMD 9513 ) i2 nacnuu. 219 CDDQG’UO-waU-IOU mmqmmthu-oow F'F'hi:;hfihi (II-DU HO lock Nu-ber: 17 ( GENERATE PIIOSPNOROSCOPE DELAY ) HEX CODE HALT ELT NEXT JMP ( MAKES KEYBOARD SEVICEABLE HALT IN HIGHER FORTH ) VARIABLE ROLDR : ZOSECSET 17 CHD CIEO >DRG 9 CMD 4E20 )DRG 1 CMD 601 >DRG 0ACMDO>DRG2GID708)DRG; ( SETS UP 9513 TIMER 1 IN 20 SEC COUNTDOWN ADDE, TIMER 2 TO COUNT PULSES mom ROTATING CAN ) DECIMAL lock Nulber: 18 ( GENERATE PROSPROROSCOPE DELAY ) HEX : TIMEBASE ZOSECSET 63 CMD RALT 82 (MD 12 (MD (DRG ROLDR ! ; ( LLA'S TIMERS 1&2, RALTS NR TIMER 1 TOC PULSE, SAVES TIMER 2 ON RESTART, STORES COUNT VALUE IN IDLDR ) : DEELAY TIMEBASE 3E8 2710 Mt ROLDR O M/ 100 + $131! 2 ; ( CONVERTS O PULSES TO TIME/PULSE, / BY 2 AND + 256 MICROSEC ) : 90DEGREEWAIT DEELAY 2 (MD CB01 >DRG 0A CMD ROLDR C was : (PROGRAMS 9513 TO GEN. TOC AFTER DEELAY TIME) DECIMAL lock Nufier: 19 ( PNOSPHORIMETER BASIC DATA COLLECTION WORDS ) REX CODE 8CAPTUREDATA 628AMOVBFF1 STA(L&ACTR2) OEOAFDV FFFO STAOOOAMV FFFl STA ( SETUPEXIDOP) AOOAMVFFFZSTA FF‘AMVFFFESTA ( DATABUFF. OFFAO) FFF2 LRLD XCRG ( BUFE. ADDR. TO D-E REG. PR. ) HLT ( WAIT FOR MISSION WINDOW ) BEGIN BDFBSTA FFFOLRLDRDCXRAMVEAMVRAMV FFFO SRLDLORAARFDV (CONV., CHE. FOR 8TH TIME IN R-L PR., mm H ADDY TILL END OF CONV., STORE MARKER REG. R & IN FFFO IN CASE MT DONE) BBF9 LDA BBF9 LDA D STAX D INN ( FEB TO DUFF. 0: “DR. ADDR. ) BBF8 LDA BBF8 LDA D STAX D IN! ( LSB TO DUFF. I: IMR. ADDR. ) R A nov 0= END ( FINALLY REALLY CHECK FOR 8TH THE) NEXT JR? 220 Block Number: 20 QUNIU’O‘JIUNl-‘O 10 12 ( PROSPRORIMETER DATA AVERAGINC WORD ) REX : 8DATA 0 FFBO 2 BCAPTUREDATA 8 0 DO EPBO O PPAO I 23 + O 10 /MOD SWAP DROP + PFBO ! LOOP FFBO O 8 /MOD SWAP DROP DECIMAL ( PICKS UP 8 SEQUENTIAL DATA WORDS A STRIPS TRAILING O’S & ADDS TO BUFP. SUM IN PPBO. THEN TAKES SUM AND DIVIDES BY 8 ) Block Nulber: 21 (DNQOUhUMt-io ( FLUOROMETER BASIC DATA.COLLECTION ).HEX CODE BECAPTUREDATA 08 O A MOV FFFO STA 00 O A MOV FFFI STA A0 8 A MOV FFF2 STA FF 8 A.MOV FFF3 STA FFEZ LHLD XCNG BEGIN 1 O A MOV BDF8 STA FFFO LELD R DCX E A MOV R A MOV R A MOV FFFO SRLD L ORA A RIMOV BBF9 LDA BBF9 LDA D STAX D INX BBF8 LDA BBF8 LDA D STA! D INX D INX R A MOV 0= END NEXT’JMP DECIMAL ( SAME AS BLOCK 13 BUT NO RALTS OR INTERRUPTS: SIMPLE FLUOROMETRY ) Block Number: 22 DmdmmwaO-‘O ( FLUOROMETER SETUP ) REX VARIABLE VALU 4 ALLOT VARIABLE CNAN.NO. : 8FDATA O EFBO 2 BFCAPTUREDATA 8 0 DO EFBO O FFAO I 2t + O 10 /MOD SWAP DROP + PPBO ! LOOP PFBO C 8 [MOD SWAP DROP ; DECIMAL ( SAME AS BLOCK 14 BUT FOR INTERRUPTLESS BFCAPTUREDATA ) 221 Block Nmer: 23 @QQGOIDNNHO ”18, -E ( FLIDRGIETER BDDE DATA COLLN WORD - LARGE ) : SAMPLE 0 0 VALU 2! 1000 0 DO CHAN.NO. O CHSET VALU 2. 8FDATA 0 0 VALU 2! 1000 0 DO CHANJD. O CHSET VALU 29 8FDATA M+ VALU 2! WP VALU 2. 1000 M/ 3 ( DOUBLE PRECISION: SET CHANNEL I: COLLECT 8 PT AVERAGE 1000):, STORE IN VALU ) Mr: 24 DATA COLLECTION WORDS ) MEDIIMDATA O O VALU 2! 64 0 8DATA M+ VALU 2! LOOP VA LU BIGDATA 0 0 VALU 2! 512 0 DO 8DATA M+ VALU 2! LOOP VALU :648AMPLEOOVALU22640DOCHAN. M+VALU 2! LOOP VALU2064M/VALU! - o a 4'3 § 8 E w 5 E :256800VALU2!256ODOCHAN. 8DATA M+ VALU 2! LOOP VALU 20 ( VARIETY OF DIFF. DATA COLLN. W O M/ VALU ! ; ON SAME PRINCIPLES ) 388 Mer: 25 MNOCHRMATOR SETTING WORDS ) :ISETWAVE DUP64$AMPLE VALU. DUPMTSWAP< IFIWER SWAP - NOSTEPS JIMPE ELSE HIGHER - NOSTEPS JIHPE TEEN ; ( READS MNO. AND SETS ACCORDING TO INPUT PARAMETER-WARM ) : WISSET 5 CHANJD. ! WAN ISET ; EXCITSET 6 CHAN. no. ! EXSCAN ISET' , ( GIVE WAVELENGTH I: INSTRIM. WILL SET ITSELF-mm OR LESS ) ( THIS ROUTINE WAS NEVER SATISFACTORY FOR ANY BORE THAN ROWE SETTING, SO THE PROCESSOR WOULD HALT AFTER THIS AND RELEASE TTIE CLUTCHES SO THE MONOCHRMATORS COULD BE HAND ADJUSTED. ANY KEY- STROKE WOULD THEN RSTART THE DATA COLLECTION. DURING SCANNING ALL WAVELENGTRS WERE ASSIGNED ON THE BASIS OF 0.05 III/STEP ) 222 (ODQQUI-EUNHOU p O 11 12 13 14 15 'DDQG’OIJDIANHOQ lock “or: 26 ( FLIDRGIETER CONTROLLER VDRDS In PARADETER wanna ROUTINES ) VARIABLE ADDR VARIABLE STEPPE VARIABLE F1 VARIABLE MDALLE VARIABLE OTHVAR VARIABLE DIR VARIABLE NO.STEPPES : READ 7 CRAN.NO. ! SAMPLE ; (FLUORO. COLLECTS 1000 E-PI'. AYES.) : PHOSREAD 7 CHAN.NO. ! BIGDATA 3 (PROS. COLLS. 512 E-PT AVES.) : UPDWN DIR O 0 > IF HIGHER ELSE LOWER THEN ; (SCAN DIR.) :mm.sm CR HEX ." THE LAST m4. LOG. US. WAS " ADDR O U. CR ." ENTER THE NEXT LOC. TO BE USED ( HEX, 5 DIGITS IICLl. ) " 5 NO.IN ADDR ! DECIMAL ; :DIRIOD CR ." ENTER 1. IF SCAN IS TO LONGER WAVELEIKITR " CR ." ELSE ENTER 0. " 2 NO.IN DIR! ; :OSTEPPES CR ." HOWMANY WAVELENGTHS ARE TO BE SAMPLED ? " ." (3 DIGITS W/. ) " 4 NO.IN NO.STEPPES ! ; lock Nulber: 27 ( F LUORGIETER CONTROLLER WORDS - FLUORCMETER BRIDE ) HEX :DELF STEPPE C 20 * JIMPES; ( .05 DIR/STEP, SO 20 t LENGTH OF INTERVAL = NO. OF ms ) : EMISSCAN UPDWN NO.STEPPES O 0 DO READ ADDR C 2+ DUP ADDR ! ! DBCAN WAIT DELF LOOP READADDR02+DUPADDR!!1388ADDR€2+!; : EXCITSCAN UPDWN NO.STEPPES C 0 00 READ ADDR C 2+ DUP ADDR ! ! EXSCAN WAIT DELF WP READADDRO2+DUPADDR! !1388ADDRO2+! ; DECIMAL ( THIS BIDCH IS FOR FLUORGIETER MDE; NO INTERRUPTS ) ( EACH DATA COLLN DRIVER MARKS END OF BUFF. W/ 5000:1388hex SO PDP-11 DATA FORMATTING SOFTWARE WILL RECOGNIZE END OF DATA ) lock Number: 28 ( PIKJSPIDRIMETER CONTROLLER WORDS - INTERRUPT BDDE ) REX : PHOSfliISSCAN UPDWN NO.STEPPES O 0 DO PROSREAD ADDR 9 2+ DUP ADDR ! ! HISCAN WAIT DELF LOOP PHOSREAD ADDR C 2+ DUP ADDR ! ! 1388 ADDR C 2+ ! ; : PHOSEXCITSCAN UPDWN NO.STEPPES C 0 DO PMSREAD ADDR C 2+ DUP ADDR ! ! EXSCAN WAIT DELF LOOP PIDSREAD ADDR C 2+ DUP ADDR ! ! 1388 ADDR C 2+ ! ; :I'MASE 80 BEEI C! ; : MASERESTORE 0 BE81 C! ; ( MASKS OUT INT #8 USED WITH FLASHLAMP, I: RESTORES IT. ROTATIM CAN WILL GENERATE SO MANY INTERRUPTS OVER THIS LINE THAT MACHINE FAILS TO OPERATE. ALTERNATIVELY, DISCONNECT INTERRUPT JIHPER ) DECIMAL 223 Block Nxfler: 29 0 t FLUORCHETER PARAMETER LOADING ROUTINES ) 1 2 : FILD CR ." STARTING FREQ. ( 3 DIGITS WI. ) " DECIMAL 3 4 NO.IN F1 ! ; 4 5 : STPLD CR ." LENGTH OF STEP ( 2 DIGITS W/ . ) " DECIMAL 6 3 NO.IN STEPPE ! ; 7 8 : MDDE CR ." ENTER -1. FOR SYNCRRONOUS SCAN MDE. 00. FOR " 9 CR ." EMISSION SCAN MDE, OR 01. FOR EXCITATION SCAN MDE " 3 10 NO.IN MDALLE ! ; 11 12 13 14 15 Block Mar: 30 0 ( DDR3 PARADETER LOADING ROUTINES ) 1 : VARLOD F1LD STPLD ; 2 3 : EX.FREO. CR ." EXCITATION FREQ. ( 3 DIGITS Nl. ) " 4 DECIMAL4NO.IN OTEVAR ! OTRVARO; 5 6 : ENJREO. CR ." mISSION FREQ. ( 3 DIGITS Wl. ) " 7 DECIMAL 4 NO.IN OTRVAR ! OTRVAR O ; 8 9 10 11 12 13 14 15 Block Met: 31 0 ( PHOSPBORIMETER DRIVER ROUTINES INTERRUPT noon ) 1 : TAKEPROSSPEC RESET 2 SODEGREEWAIT 3 NDDDE mDALLE O 4 0: IF CR ." EMISSION SCAN ADDR " VARLOD 5 MEN.SET EXJ‘REO. EXCITSET 6 F1 6 EMISSET CLUTCNRELEAS HALT WSSCAN 7 ELSE 8 MDALLE O 0( IF CR ." SYNCERONOUS SCAN noun " CR 9 ." SCANNIM IN TRIS MDE IS BASED ON THE DAISSION H O BONCRRCMATOR SE‘I'TDKS " VARLOD M.SET EXJ‘REO. 11 EXCITSET F1 0 MISSET SYNCSET PmSEMISSCAN ELSE CR ." EXCITATION SCAN MDE " VARLOD m. SET BLFREO. NISSET F1 0 EXCITSET 15 CLUTCERELEAS HALT PROSEXCITSCAN TEEN THEN ; HHH hUN 224 Block Ntfler: 32 ”04010-5”ch HHo-n NHO 13 14 15 DQQQOchwND-‘OU WUQO’U‘IwaO-OOU ( FLLORCMETER CONTROLLER ROUTINES - FLUORGAETER MDE ) : TAKEFLUOROSPEC BDDDE WDALLE O 08 IF CR ." EMISSION SCAN MDE " VARLOD MDLSET EX.FREO. EXCITSET F1 0 NISSET CLUTCHRELEAS HALT NISSCAN ELSE MDALLE 9 0< IF CR ." SYNCHRONOUS SCAN MOE " CR ." SCANNING IN THIS none IS BASED ON THE MISSION mncm'roa SETTIN " VARLOD EX.FREO. HELSET EXCITSET F1 9 EMISSET CLUTCHRELEAS HALT SYNCSET WISSCAN ELSE CR ." EXCITATION SCAN FDDE " VARLOD HELSET BLFREO. MISSET F1 9 EXCITSET CLUT‘CHRELEAS HALT EXCITSCAN THEN THEN ; 10:): Mar: 33 ( PULSE WDE DATA COLLECTION - CORE WED ) HEX CODE FLCONV FF30 LHLD XCHG BBFO STA ( PULSE LAMP ) HLT ( WAIT FOR FLASH ) 7OAMV BDFO STA (SET HUX) BEGIN ( BEGIN DATA COLLN. LOOP ) BDF8 STA ( CONVERT ) FFOOLHLDHDCXHAMVHAMVHAMVFFOOSHLDLORAAHBDV (LOOPCOUNTERFRGAFFOO, DECR.,CF. W/O, RETURNTOFFOOI-E) BBF8 LDA BBF8 LDA D STAX D INX (COLL. I: STORE BBB. DER. ADDR.) BBF9 LDA BBF9 LDA D STAX D IN! (COLL. & STORE LSB, DER. ADDR.) HAMVO=END (TESTFOREND) nnmvsLmvmosnw (RETUID'FINALADDRTOFF30) NEXT JMP DECIMAL lock Nunbcr: 34 ( PULSE BDDE PARAMETER SETUP ) VARIABLE STORIDC VARIABLE DPS VARIABLE DELAY : EMISSFREO CR .°' THE EMISSION WAVELENGTH IS ( 3 DIGITS W/. ) " DECIMAL 4 m.IN F2 ! ; :REPEATS.”HOWMANYTIMES IS THIS EXPERIMENTTOBE REPEATED ?" CR."1.=1TIME, 2.=ETII\BS. 3.864TIMES,AND4.8512TI MES”2NO.INOTHVAR!; :DATAPOINTS ." WMANYDATA POINTS ARETOBE COLLECTED 7 " CR ." 3 DIGITS W/. " DECIMAL 4 NO.IN DECIMAL DPS 2 ; : PSTORAGE CR ." IN WHICH STORAGE LOCATION DO YOU WISH THE FIRST DAT!!! " HEX 5 NO.IN DECIMAL STORLOC ! ; 225 Block Nulber: 35 0 ( SETUP AVERAGING BUFFERS I. POINTERS ) HEX 1 VARIABLE 2NDDEST VARIABLE 3RDDEST VARIABLE FINALDEST 2 : PULSE I REED C! ; 3 4 :DIDATAMANIPFFODOODOFF30011+2t-1+0010/FF300 5 11+23-001080RFF300I1+2t-2LO0P: 6 7 :2NDESTDPSGZ¥IO+S‘IORLOCO+2NDDEST!; 8 . 9 :OZNDESTDPSOODOOZNDDESTOIZC‘v-QLOOP; 10 11:3RDESTDPSO4320+STORLOCO+3RDDEST2; 13 :O3RDESTDPSQOD003RDDEST9123-t ! LOOP; 15 : 4THDEST DPS G 6 t 30 + STORLOC 6 + FINALDEST ! ; DECIMAL Block Nader: 36 ( AVERAGING ROUTINES - DONE E WORDS AT A TIME DUE TO PROBLDS WITH FORTH ) :18TAVDPSOODOZNDDESTOI2¥+98MDSWAPDROP 3RDDESTOIZ¥+DUPOROT+SWAP2LOOP; : DATACOLLECTION STORLOC O FF30 ! 7 CHSET FLCONV DPS O FFDO ! D1DATAMANIP ; (DOQOOI-DuNO-‘O :ISTXFERDPSOODOSTORLOCOIZ 2NDDESTOI28+DUPOROT+SWAP 0 LOOP 2NDEST ; H O 0-. 4' HH NH :2NDAVDPSOOD03RDDESTOI2t+OEflDDSWAPDROP 13 FINALDESTOIZ:+DUPOROT+SWAP!LOOP; o-n-o 0|ch DECIMAL Block Mar: 37 0 ( INTERRUPTS ) REX 1 2 : FLASHVEC 203E 3050 ! 32F3 3052 2 DEED 3054 ! C9?! 3056 ! ; 3 : INTRESTORE C315 3050 ! 10 3052 ! C3I‘5 3054 ! 10 3056 ! ; 4 DECIMAL 5 ( THIS INTERRUPT ROUTINE OVERLAYS TEE SPECIFIC ROUTINE IOR m 6 FLARLAMP INTERRUPT ON AN EXISTING VECTORTABLE: VECTOR TRROWR 7 THE WORD FLASNVEC AND RESTORES THE TABLE To DEFAULTS 081m 8 INTRESTORE. THE SPECIFIC CODE REPRESENTS THE comm) STEM: 20 9 O A mV-ooi inst-DI REED STA-tend 001 to pic-RI RET ) 10 11 VARIABLECOWNTER 12 :IOKATOR13EMITCGVNTEROI+DUP.CGVNTER!; 226 Block Nuflnr: 38 0 ( NIGER LEVEL AVERAGING WORDS ) HEX 1 2 :ZEMESODOOCOOOIZC+!I.OOP: 3 4 : BXDATA 8 0 DO 7 CRSET DATACOLLECTION 18TXFER LOEATOR LOOP ; 5 6 : 6420ATA B 0 DO BXDATA ISTAV 02NDEST LOOP ; 7 8 : HEEDR CR ." TRIS EXPERIMENT COLLECTS MISSION DECAY DATA " ; 9 10 11 12 13 14 15 Block Met: 39 0 ( PDOSPRORIMETER PULSE MODE DRIVER CODE ) HEX 1 : IRATEXP REEDR 2 EMISSFREO P2 9 EMISSET CR DATAPOINTS CR REPEATS 3 PSTORAGE 2NDEST 3RDEST 4TEDEST 0 COWNTER ! CR 4 DPS e FPOO ! - 5 FLASRVEC CLUTCRRELEAS HALT (WAIT FOR FNL END SET A KEYED INPUT) 6 OTl-IVAR O 7 1 8 IF 8 7 CHSET DATAOOLLECTION lsTXFER LOEATOR 9 2NDDEST O DPS 0 2* + 1388 SWAP 2 10 ELSE OTEVAR O 11 2 = IF 12 BXDATA ISTAV 1388 13 3RDDESTODP3023+ 2 ELSETEENTBN; 14 15 Block Met: 40 0 (WREOFTREDRIVER)HEX 1 : ZRATEXP OTHVAR 0 2 3 8 IF 3 64XDATA 4 DPS e 0 DO 5 3RDDEST012¥+DUP08MD SWAP DROP SWAP 2 LOOP 6 1388 3RDDEST 0 DPS C 23 + 2 7 ELSE OTHVAR O 8 4 8 IF 9 8 0 DO 64XDATA 10 2NDAV 03RDEST LOOP DPS O 0 D0 FINALDEST O 11 DUP G 8 MD SNAP DROP SWAP DUP 2+ FINALDEST I ! IDOP 12 1388 FINALDESTO!ELSETBNTHEN; 13 : RATXP 1RATEXP ZRATEXP : : RATEXP 1000 ZEIDES RATXP ; 14 DECIMAL 15 227 mmammauvoo—ou HH HO HHHH 0&9)” ODOQOUIbNNF-‘OB! CDQQOOIbQNO-‘OU lock Nu-ber: 41 ( FIDTON-COUNTING CIDCK SET-UP ) HEX : LONGSAMSETUP 17 (MD C180 >DRG ; ( 1m cwcx ) : SAMPLINGRATE 1 (MD 601 >DRC 9 (MD >DRG 41 OD ; (COUNTDOWNMDEFRWINPUTPAMTER) DECIMAL loch Met: 42 ( PHOTON-COUNTIN ODDS - CORE WORD ) HEX CODE SLOFIDONV mo LELD X036 1 O A ADV DBFO STA (ADDR TO D-E REG, FLASH W) HLT ( MAIT FLASH ) 610AMVDFF18TAHLT( LIACTRITOCNT-DWN, HLTNETN) BEGIN 61 8 A MV EFFI STA FFOO LELD ”‘02 SHLD D PUSH ( RESTART CTR 1. STORE OSAMPLES IN TEMP ADDR m2. PUSH INDEX ) BEGIN ( DATA COLLN AS IN FLCONV, EXCEPT ONLY ESE SAVED ) EDI-'8 STA WV 11 A WV 11 A WV STAX D INX ( COLL. DSD ONLY, IND. ADDR. ) DCXNABDVFFO‘ISHLDLOMHLTO8END FALLTNEDINSUAVEEEENCOLLECTED) MVPBOSHLDNEXTJMP lock Nubcr: 43 ( WEE SW RATE STUFF - PARAMETER IDADING ) VARIABLE “SAMPLES VARIABLE IOPOINTS : SWEATE CR ." ENTER THE SAMPLIM RATE FOR THIS EXPERIMENT " CR ." EACH UNIT IS EQUAL T0 1.0 MS ( 3 DIGITS Nl. ) " 4 IDJN SAMPLINGRATE; :OSAKPLESCR."ENTERTHEM.OFSAMPLESTODEOOLLECTEDATEAC EPT.”CR."(3DIGITSW/.)"4ID.IN ICSAMPLES23 :OPOINTSCR ." “RETURN. OF DATA POINTSTODE COLLECTEDTIII STIME"CR ." WERNOTTOOVERFWMRY ( 3DIGITSN/.) " 4 NO.IN lOPOINTS ! ; 228 mmqmubwcowou DDQOOQGNHO' UOQQO‘MNHOU lock Nuflnr: 44 ( moron—comma DATA REDWTION ) HEX VARIABLE BASEADDR VARIABLE CTRDEST 2' ALLOT VARIABLE STARTADDR :SETCOWTIOSWLESOODOBASEADDROI+GFDAIIDO=IE ELSECTRDESTODUPOI+SWAP2TEENWP; ( CF’S DEB W/ F0. IF NOT 0 THEN FETCHES In INCR'S CURR. INDEX IN CTRDEST ) : OEXPCOUNT IOPOINTS O 0 DO STARTADDR O IOSAMPLES C I t + BASEADDR 2 FEOO I 28 + CTRDEST ! SETOOUNT LOOP ; ( LOOPS TNRU SETOOUNT I: INCR CTRDEST TO TOTE ALL DATA BINS ) :ZERROCOODOOFD401+C2LOOP; DECIMAL lock Number: 45 ( WEE SETUP ) VARIABLE BEGINNING :OTIMESCR."ENTERTREN1NBEROFTIMESTREEXPERIMENTISTOBE REPEATED(3DIGITSW/. ) "4m.momvm2; REX :FEZEROIODODOOFEOOI+C2LOOP; DECIMAL lock Number: 46 (THE DRIVER) : SLOWRATEXP DECIMAL IDNGSAMSETUP OPOINTS OSAMPLES #TIMES SLOWRATE PEZERO 0 COWNTER 2 CR EX OTIWAR O 0 D0 0000 ”30 2 0000 STARTADDR 2 "SAMPLES 0 mo 2 ”POINTS 0 HM 2 SIOFLOONV OEXPOOUNT LOEATOR LOOP CTRDESTO4+1SBBSWAP 2 DECIMAL: 229 DOQGGDHNHOU 10 11 mmqmmbunwou lock Mar: 47 ( SET UP FOR SRORT-LIVED PROTON-OOUNTIM ) REX : FORMATXPER IOPOINTS e 0 DO FEOO I 4 t + 2+ 0 BEGINNING 0 I 28 + 2 LOOP ; DECIMAL :OPCTIMESCR.”ENTERTREN0.0FTIMESTEEEXP.ISTOBEREPEAT ED"CR.”(5DIGITSW/.)"GNOJNOTRVAR!; :PCSETCOUNT IOPOINTSOO DOBASEADDRO I +OFOAND O= IF ELSECTRDEST0123+DUP01+SWAP2TRENLOOP; lock Ntnber: 48 ( SHORT-LIVED PHOTON-COUNTING - CORE WED ) REX CODE PCFLCONV ”'30 LRLD XCRG BBFO STA BBFO STA ( PULSE LAMP ) HLT 7 O A WV BDFO STA BEGIN BDF8 STA FFOOLHLDRDCXEADDVRAMVRAMV FEOOSNLDLORAARMV BBF8 LDA BBF8 LDA D STAX D INX 11 A WV 0: END DRMVELMV mo SRLD NEXT IMP DECIMAL Block Nulber: 49 IDDQOUMDGNHO Hy HO HHHH (ll-bu” ( SHORT-LIVED PROSPROR PROTON~COUNTIND DRIVER ) HEX : POTONCOUNTRATEXP REX 2000 ZEROES DECIMAL OPOINTS #PCTIMES 0 COWNTER 2 CR HEX F000 CTRDEST 2 OTRVAR O 0 DO IOPOINTS 0 FFOO 2 C000 FP30 2 C000 DASEADDR 2 7 CHSET PCFLCONV PCSETCOUNT LOEATOR L00? 1388 lOPOINTS 0 2! P000 + 2 DECIMAL ; DECIMAL 230 mmqmmautowow QOQOODQNHOU IDOQUOOthNr-‘OU lock Nunber: 50 ( DATA OUANTITATION ROUTINES ) : PHOSPBOROUANTDATA 90DEGREENAIT 7 CNANJD. 2 0 U VALU 22 1000 0 DO CNAN.NO. G CHSET VALU 20 8DATA M+ VALU 22 LOOP VALU 20 1000 HI VALU 2 ; : OPIDS PMSPHOROUANTDATA CR VALU O . ; : FLUOOUANTDATA 7 CRAN.NO. 2 O O VALU 22 4096 0 DO VALU 20 LAMBDA 14+ VALU 22 LOOP VALU 20 4096 M/ VALU 2 ; : QFLUO FLUOOUANTDATA CR VALU O . : lock Number: 51 ( WHY TO DISK ) VARIABLE BIDKNIN VARIABLE BADRES VARIABLE MADRES VARIABLE KOUNT ( MADRES = WHY ADDRESS OF DATLM. BADRES = THE ADRESS IN TEE VIRTUAL BLOCK TO NHICR THE DATIM IS SENT ) CODE CHER BEGIN MADRES LHLD XCNG D LDAX D INN XCNG MADRES SNLD BADRES LRLD 12086 D STAX D INX XCRG BADRES SNLD KOUNTLNLDRDCXKOUNTSHLDLAMVRORAO=ENDNEXTJMP ( TRIS ASSEMBLY CODE TAKES A DAM F2104 THE DATA HEIDI" AND PUTS IT IN THE ACCLMULATOR AND THEN TRANSFERS IT '10 THE APPROPRIATE VIRTUAL BLOCK ADRESS WHILE INCREMENTING THE ADRESS POINTERS AND DECREMENTING THE BLOCK FULL POINTER KOUNT : 400-8 = 1024 -10 ) DECIMAL lock Huber: 52 ( MR? TO DISK ) REX : ADINIT CR ." ENTER THE STARTING ADDRESS OF THE HEIDRV TO BE TRANSFERRED " CR ." ADDRESS IS IN REX ( 4 DIGITS N]. ) " REX 5 m.IN MADRES 2 DECIMAL ; : BIDCKNLHBERE CR ." WNICH BLOCK IS THE DATA TO STD" BE IN (3 DIGITS W/ . ) " CR 4 NO.IN BLOKNIM 2 ; : BIDCKSTORE ADINIT BLOCKNIMBERE BLOW O BIDCE BADRES 2 400 KOUNT 2 CEKR BLOKNIH C BLOCK UPDATE FLUSH DROP ; : IBLOGKSTORE BLOKNIH O BLOCK BADRES 2 400 EOUNT 2 CHER BLOKNIM C BLOCK UPDATE FLUSH DROP ; DECIMAL 231 CDQQQOI&UNHOH lock Nmer: 53 ( MRY '10 DISK CONTINUED 1 ( TESTER ESTABLISNES THE STARTING ADRESS OF THE VIRTUAL BLOCK BY CALLING THE APPRORIATE BLOCK WITH THE VARIABLE BLOKNIH. IT SETS THE BLOCK FULL KOUNTER AT 400 HEX AND THEN STARTS TEE TRANSFER ROUTINE NRICR CYCLES UNTIL ONE FULL BLMK OF 1024 was HAS BEEN TRANSFERRED. IT THEN FLUSRES TRE VIRTUAL BLOCK TO A DISK BIDCK. ) EEK : BLKDRV IBLOCKSTORE BIDKNIM 0 1+ BLOKNIH 2 i : MELXFER C000 MADRES 2 BIDCKNIHBERE 10 0 DO BLKDRV LOOP ; ( DATA MRY MISES 16 BIDCKS - 4-K HEX - OF RAM MET. AFTER A BLOCK HAS BEEN TRANSFERRED THE BLOCK POINTER BIDKNIH IS INCREMENTED IN BLKDRV. MFER THEN CAUSES ALL 16 BLOCKS TO BE TRANSFERRED. ) DECIMAL 232 "7271211227"fifimflfilflifitflffl'fljflflmfiflflflmms 129