ADAI1S 233 TEXAS A AND N UNIV COLLEGE STATION DEPT OF CHEMISTRY F/6 20/61 RECENT ADVANCES IN (RJLTICOMPONENT FLUORESCENCE ANALYSIS. (U)

APR 82 1 M WARNER- L 3 MCGOWN N0001-60-C-0703UNCLASSIFIED TA-6 NM.

m mmmmmmmmmmmEEmmmmmEE*EE

I m: mh~h~.mmmmmfflfflmmfllmmmmfIfl flllllllllf l

~ 1.8

IIIIL25 I -I 1 1 116

MICROCOPY RESOLUTIUN TEST CHART

T I NA Ho I . .Alt ' IAN[ t,

SECURI y CLASSIFICA'O!N OF TIS PAGE (W%on Dina Entered)

0 P REPORT DOCUMENTATION PAGE BEFORECOMPLETINGORM

1, REPORT -4umBER 12. GOVT ACCESSION NO. 3. RECIPIENT*S CATALOG NUMBER

4. TITLE (and Subtitle) S. TYPE OF REPORT & PERIOD COVERED

Recent Advances in Multicomponent Fluorescence Interim Technical ReportAnalysis 6. PERFORMING ORG. REPORT NUMBER

7. AUTHOR(s) 6. CONTRACT OR GRANT NUMBER(s)

Isiah M. Warner and Linda R. McGown N00014-80-C-0703

9. PERFORMING ORGANIZATICN NAME AND ADDRESS 10. PROGRAM ELEMENT. PROJECT. TASK

AREA & WORK UNIT NUMBERS

Department of ChemistryTexas A&M University NR-051-747Collpae Station. TX. 77R43

It. CONTROLLING OFFICE NAME AND ADDRESS 12. REPORT DATE

Chemistry Propram April 20, 1982Office of Naval Research 13. NUMBER OF PAGES

800 North Ouincv St. - Arlinaton. Va. 22217 69X .14. MONITORING AGENCY NAME & ADDRESS0I different from Controlling Office) IS. SECURITY CLASS. (of this report)

Martial Davonst-ONRAdministrative Contractino Office UnclassifiedRoom 582, Federal Buildinq - 300 East 8th St. OECLASSIFICATIONDOWNGRAOING

Austin, TX. 7870116. DISTRIBUTION STATEMENT (of this Report)

Approved for PUblic Release: Distribution Unlimited

17. DISTRIBUTION ST. 4ENT (of abstract entered in Block 20, if different from Report)

IS. SUPPLEMENTARY teS

Prepared for Publication in CRC Critical Reviews in Analytical Chemistry

IS, KEY WORDS (Continu, on reverse eide if necessary and identify by block-number)

Mul ticomponent Analysis, Fluorescence

C 24A ISTRACT tContinue on reverse side if necessary and Identify by block number)

C-') "lumerous advances in fluorescence spectroscopy have been made in the past fewyears. Many of these advances have come in the form of improved instrumentation

I and new data reduction strateies. Primary emphasis of these new strategies

Lis to develop new systems for multicomponent analysis. This manuscriptdescribed a number of recent advances in fluorescence spectroscopy with primaryemphasis on multicomponent techniques. Methods for increasino sensitivity offluorescence analysis are also discussed, alona with a few new ideas for use inconventional instrumentation. ,However, most of the review (Cont on Back)

DD o, 1473 82 U 0 ( U 63 \CT AIJAN O

SE[CURITY CLASSIFrICATION OPP TMIS PAGE (*%oen Dot; Entered)

OFFICE OF NAVAL RESEARCH

Contract N00014-80-C-0703

Task No. NR 051-747

TECHNICAL REPORT NO. 6

Recent Advances in Multicomoonent

Fluorescence Analysis

by

Isiah M. Warner and Linda B. McGown

Prepared for Publication

in

CRC Critical Reviews in

Analytical Chemistry

Department of ChemistryTexas A&M University

College Station, Texas 77843

April, 1982

Reproduction in whole or in part is permitted forany purpose of the United States Government

This document has been approved for public releaseand sale; its distribution is unlimited

February 1982 155

RECENT ADVANCES IN MULTICOMPONENTFLUORESCENCE ANALYSIS

Authors: Isiah M. Warner

Department of ChemistryTexas A & M UniversityCollege Station, Texas

Linda B. McGownDepartment of ChemistryCalifornia State UniversityLong Beach, California

Referee: Gary D. ChnstianDepartment of ChemistryUniversity of WashingtonSeattle. Washington

TABLE OF CONTENTS

1. Introduction

II. General Advances in Fluorescence InstrumentationA. General and Theoretical Discussions of Signal-To-Noise EnhancementB. Improvements through Modification of Instrumentation and Con-

figurationC. Improvements in Detection and Electronics

Ill. Multicomponent Fluorescence InstrumentationA. Time-Resolved TechniquesB. Phase-Resolved TechniquesC. Selective ModulationD. Total Luminescence SpectroscopyE. Synchronous Excitation FluorescenceF. Miscellaneous Instrumental Techniques

I. Double-Beam Fluorescence Spectroscopy2. Multicomponent Kinetic Studies by Fluorimetry3. Fluorescence Detectors for Liquid Chromatography

IV. Multicomponent Fluorescence Data Reduction Aeossion Fox.'

A. General AlgorithmsB. Qualitative Analysis of Fluorescence Data NU3e uAlC. Quantitative Analysis of Fluorescence Data Dnawounc e d

V. Summary and Conclusions

Acknowledgment D r.1 0D i at rl b u t i O W

References Ala()bil fry'Codes

vi l C2AVizi Ond/or

1J

156 CRC Critical Reviews in Analytical Chemistr'

I. INTRODUCTION

An increasing number of researchers are recognizing the distinct advantages offered bymolecular fluorescence spectroscopy as an analytical tool. This increase is evident fromthe large number of manuscripts and monographs published in this area. In the area offluorescence instrumentation alone, the number of referenced articles in the biennial"Fundamental Reviews" of Analytical Chemistry increased from 82 in 1976 to 156 in1978. This widespread use of fluorescence analysis arises from its selectivity andsensitivity. The selectivity of fluorescence assay is derived from the dependence of theemission intensity both on the wavelength of the exciting radiation and on thewavelength of detection for the emitted radiation. Thus, it is often possible to findoptimal spectral regions where the effects of background or interfering fluorescence areminimized. Moreover, the use of phosphorescence spectra and exploitation ofdifferences in lifetimes and polarization can provide even greater selectivity. Thesensitivity advantage of fluorescence arises from the direct detection of the emittedphotons to give an emission signal. This is in contrast to absorption spectroscopy. inwhich a small difference between two large signals is measured. Consequently, with theuse of state-of-the-art detectors, the sensitivity of fluorescence methods will oftensurpass the sensitivity of similar absorption methods by three orders of magnitude.

Numerous advances in the area of fluorescence spectroscopy have been made in thepast few years. These advances have come largely in the form of improvedinstrumentation and new data reduction strategies. A detailed account of each of theseimprovements is beyond the scope of this review. Therefore, our treatment will be limitedto those techniques which have proved to be either novel concepts or novel applicationsof established concepts. Broader, more general, coverage of many areas of fluorescencespectroscopy can be found in any number of review articles,'-' and more specificaccounts in detailed reviews by Bridges. Gibson, Guilbault.' Klainer." and others.9 -2

Although our discussion will focus on the two major areas of fluorescence spectroscopymentioned above (new developments in instrumentation and new data reductionstrategies), we do feel that the application and refinement of the more traditionalmethods of fluorimetry warrant some discussion. These advances will be discussed inSection If and will emphasize improvements in the sensitivity of fluorescence assay.Section III focuses on the development of novel instrumental methods for multi-component fluorescence analysis, including discussions of time- and phase-resolvedmethods, synchronous excitation, selective modulation, total luminescence spectroscopy,and several other techniques.

The rapid data acquisition capabilities provided by advanced fluorescence instrumen-tation have caused an increased need for new data reduction strategies. In Section IV wewill describe some of the new data reduction strategies being developed to meet this needin multicomponent fluorescence analysis.

Finally, although the major concern of this manuscript is multicomponentfluoresceitce analysis, we will also discuss multicomponent phosphorescence methods.many of which can be modified to provide multicomponent fluorescence capabilities.

U., GENERAL ADVANCES IN FLUORESCENCE INSTRUMENTATION

Many advancei in fluorescence spectroscopy have been directed towards increasingsensitivity, inoluding improvements in signal-to-noise ratio (S, N), backgroundreduction, and data acquisition and manipulation. Adaptations and modifications forimproving the sensitivity of the traditional fluorescence instrumental configuration,shown in Figure I, are discussed in this section.

February 1982 157

2) Exc:totton Woveiengrt,

:)Rodaot~on Source Se~ec~or

(filter or monochrrnaotr)"

Lens System 4)Emiss:ort Waveleingth

(fiter or monochrc.-notor)

5) Detector

FIGLRE i. Diagram ot a con,.cntional fluorimeter.

A. General and Theoretical Discussions of Signal-To-Noise EnhancementA theoretical comparison of pulsed-source excitation-gated detection with continuous

wave source excitation for both atomic and molecular luminescence spectroscopy hasbeen presented by Boutilier et al." The iniTestigators suggest that the S N advantages ofthe pulsed-source excitation-gated methods are not necessarily large enough in all casesto justify the necessarily more complicated apparatus. Mathematical expressions forsignals and for noise sources are developed for each of three measurement methods,including continuous wave excitation and measurement (cw, cw). pulsed source-gateddetection with no time resolution (p g). and pulsed source-gated detection with timeresolution (p, g,, t) (Figure 2). Noise sources considered include detector shot noise.background shot and flicker noise, scatter shot and flicker noise, source-inducedbackground shot and flicker noise, and analyte luminescence shot and flicker noise.Signal/ noise rvtios for realistic experimental conditions are then calculated using theappropriate expressions.

The investigators note that the less expensive cw, cw system is rarely used in analyticalfluorescence spectroscopy. and conclude that such systems are analytically useful only incases where source scatter, source-induced background, and nonsource-inducedbackground are small. i.e., not much greater than the analyte signal. For molecularfluorescence spectroscopy, p? g systems are useful if source-induced fluorescencebackground is small. However. as pi g, t systems become available they will prove to bemore useful due to their ability to discriminate against source-induced background.

Cooney has investigated detector and other instrumental effects on the S, N ratio."Comparisons are made between photomultiplier tubes with various image devices,including image dissector, silicon vidicon, silicon intensified target vidicon, intensifiedsilicon intensified target vidicon. and secondary electron conduction vidicon detectors.The comparison is based on theoretically developed equations for the S N ratio for eachdevice, calculating the S, N ratio for analytically important situations. His conclusionsare that the silicon vidicon is not useful for molecular luminescence spectroscopy, but theintegrating image devices are useful in the visible range, since they combine the

158 CRC Critical Reviews in Analtical Chemistry

LOa Source

iT,m.

I Det.cor Ou

tJi ino I1 Imti~f

f Tme

b. off R ,,

O' ..E1 g ,, " Lumieecenee

Time

t I

C. Off TimeOP Detecto

Off - Tim

"0 Luminescence " Tt

FIGURE 2. Schematic representation of excitation measurement systems

used in atomic and molecular luminescence. (a) Cw, cw (cw excitation with cw

detection). ib) P g (pulsed excitation with gated detection and no time delay .

In this case. the source turns on and oflat a certain repetition rate. 1. The source

on time is to (seconds). The detection gate time is t6 (seconds i. The lifetime of

the luminescence of the analyte is rt (seconds). The lifetime of the luminescence

of any source-induced fluorescence background is r, (seconds). The lifetime ofthe source scatter is r, (seconds). Recall lifetime is defined for a first-

order process (exponential decay) as the time for the intensity to de-

crease by 1, e from the initial value. The luminescence decay of allowingsource termination is actually a convolution of the decay processes hav-

ing lifetimes of ri.. rh and r,. (c? P g t (pulsed excitation with gated

detection and with time resolution). The discussion in (b) applies here except

that there is a delay o1 tj lseconds) between the termination o excitation and

the initiation of the measurement process. Key: solid line, source temporal

profile normalized to unity dashed line with dots. detector temporal prolile

normalized to unity: dashed line, luminescence profile normalized to unity

assuming steady state is reached (From Boutilier. G. D.. Bradshav,. J. D.Weeks. S. J.. and Winefordner. J D.. Appi. Speciros.. 31. 307. 1977 With

permission.)

multichannel time advantage over scanning devices with S, N ratios which approachthose of PMT tubes. These considerations will be dealt with more fully in the nextsection.

February 1982 159

FIGURE 3. Image rotation system. Exciting light emerging from exit slit I ol excitationmonochromator strikes plane mirror 2. goes up to concave mirror 3. passes through cylindricallens 4 and spherical lens 5 to cuvette 6 and retromirror 7. Fluorescence light from the cuette andretromirror 8 passes through spherical lens 9 and cylindrical lens 10 to image rotating concave I I andflat 12 mirrors and fluorescence monochromator entrance slit 13. (Reprinted with permission iromWhite. J. U.. Anal. Chem.. 48. 2089. 1976. Copyright 1976 American Chemical Society.)

B. Improvements through Modification of Instrumentation and ConfigurationLir and Miller 5 discussed the use of polarizers to improve detection limits in

fluorimetry by reducing interference due to scattered light. Such reduction is particularlyuseful in biological samples such as blood, and in solutions of macromolecules. Theinvestigators demonstrated the use of some new arrangements of polarizers that had notbeen previously studied, and suggest that their results show that the use of polarizers mayhave a wider applicability in fluorimetric analysis than previously suspected. Specifically.improved detection limits were found, using their new polarizer configurations, foraqueous solutions of quinine sulfate, riboflavin, and warfarin, and for warfarin inplasma. The improvements ranged from 10- to 100-fold in terms of pg mg -i detectionlimits. First-order Rayleigh scattered light is reduced by single, horizontally orientedpolarizers, and second-order scattering is removed using crossed polarizers, althoughthis configuration severely reduces fluorescence intensities. It is noted that the detectionlimit of warfarin in plasma using polarizers is better than the limit in pure aqueoussolution without using polarizers.

Studies by White 6 demonstrated that a 30-fold increase in sensitivity of fluorescencespectrophotometers can be achieved by rotating the monochromator slit images 900.Concave retromirrors are used behind the sample to reimage the cuvette on itself,collecting all the light passing through the sample and directing it through the sampleagain. Thus. the sample is viewed along the passing excitation slit image. and sensitivity isincreased by more efficient use of the emitted light. Figure 3 shows the image rotationsystem used for the studies. Signals can be increased by a factor of 3.8. and an overallincrease in signal/ noise of 30 was observed for a Raman spectrum of methanol.

An instrumental modification which has only recently been applied to routine (andaffordable) fluorescence analysis is double-beam spectrophotometry. Anacreon andOhnishi"' described a moderately priced commercial double-beam spectrophotometer.also discussed by Porto and Terhaar t ' in a review article. Sample fluorescence iscorrected for background scatter and unwanted fluorescence using a reference beampassing through a reference (blank) cell. Since the reference beam intensity is monitoredand then automatically subtracted from that of the sample beam instrumentally, time issaved and measurement precision is increased. Sensitivity is sacrificed due to the time-sharing nature of the electrooptical system however, this decreased Si N ratio can oftenbe tolerated in practical applications of fluorescence since the technique is inherently sosensitive. The optical paths for the sample and reference beams are equivalent, and are*1

160 CRC Critical Reviews in Analviical ChemistrY

FIGURE 4. Block diagram o1" system for smngle photon counting. IFrom Jameson. D. M..~Spencer. R. D.. and Weber. G.. Rev. Set. Instum.. 47. 1034. 1976. With permission.)

! split by an optical chopper. Both beams are measured at right angles to the exciting lightusing a photomultiplier detector. The two signals are amplified, discriminated via phasesensitive demodulation, and finally the difference between the two is taken electronically.More discussion on this technique is given in Section III.F.

C. Improvements in Detection and Electronicsi Single photon counting detection was combined with repetitive scanning to improve

S, N ratios in studies done by Jameson et al.'9 Strong fluorophores were detectable onthis system at the picomolar level. The instrumental set-up is shown in Figure 4. The

+I double monochromator arrangement is used to reduce stray light levels. Data' manipulation via computer interfacing allows background subtraction after blank and

sample data are collected (Figure 5). providing a means for extracting very small samplesignals from background for the sufficiently sensitive instrument.

Glushko et al., :° developed a method for increasing S/N ratios of weak fluorescencesignals using drift-free integration to enhance the analog signals obtained. A gain insensitivity of an order of magnitude is obtained using high resolution voltage-to-frequency conversion, followed by even counting. It is pointed out that, although weaksignals are best measured by discrete photon counting systems, appreciable advantages insensitivity for weak signals can be inexpensively obtained using the enhanced analogtechnique described, avoiding conversion to photon counting instrumentation.

Finally, another electronic approach to enhancing S/N ratios for spectrofluorimeters.described by Wehrly et al..:' utilizes a bipolar averaging circuit. One hundred signals areaveraged at variable sampling rates (a few seconds or longer for collection of the 100, orless time with modifications). The exciting light is also monitored as a reference signal foradjusting phototube dark currents and amplifier offsets. Figure 6 shows a fluorescenceemission spectrum before and after the signal averaging is applied.

III. MULTICOMPONENT FLUORESCENCE INSTRUMENTATION

The large number of variable parameters in fluorescence spectroscopy provides a

A _ -T*/STPLW

February 1982 161

A

C D

E

FIGURE 5. Extracting small sample signals from background. (A) Back-ground emission of 0.2 NV H:SO, scanned from 371 to 560 nm. Excitationat 340 nm. Raman appears at 385 nm. (B) Background plus 10" .M4 quininesulfate. (C) Spectrum B minus spectrum A: 10" .1 quinine sulfate.(D) 10"" Ml quinine sulfate. IE) Overlay of spectra C and D. (FromJameson. D. M.. Spencer. R. D.. and Weber. G.. Rev. Sc. Instrum.. 47.1034. 1976. With permission.)

wide variety of approaches to the analysis of multicomponent mixtures. For example, thefluorescence intensity can be defined as a function of the wavelength of excitation Andmonitored wavelength of emission. Mathematically, this can be expressed as

It - f (A,. A.,,) (I)

where 1, is the intensity of fluorescence and X., and X.m are the wavelengths of theexciting and emitted radiation, respectively. If one were to pulse the excitation radiation.and monitor the fluorescence intensity as a function of time after the termination of the

162 CRC Critical Reviews in Analytical Chemistry

X

1 'S

0 0

FIGURE 6. Signal-averaged spectrum ot fluorescein. -luorescence ems.ion ,pcctruof 10 "' .l l#uorescein in carbonate butler. Exciiaton wa',elngih i 480 nm Emison scannedfrom 490 to 580 nm. (A) Direct acqusition. (I) A erage ol 100 signals at each wavelength.(Reprinted with permission Irom Wehrlv. J. A.. Williams. J. F.. Jameson. D. M . andKolb. D. A.. Anal. Chem.. 48. 1424. 1976. Copyright 1976 American Chemical Society

excitation beam, an exponential decay would be observed. This is because thefluorescence decay process obeys first-order rate kinetics. Thus, the fluorescenceequation as a function of time would have the form

Hot ', (2)I,{o)

where I,(o) is the fluorescence intensity at time zero, and 1,(t) is the fluorescence intensityat a time t after termination of the excitation source. The constant k is a summation of allthe first-order rate processes which compete for deactivation of the excited molecule.Included in this constant are the rate constants for radiative decay, intersystem crossing,and collisional quenching where the number of quencher molecules is assumed to be largecompared to the number of excited molecules.

Use of the three individual parameters (A, ,,, time) and combinations of themprovide a variety of approaches to the development of novel multicomponentfluorescence instrumentation. Most of the discussion in this section will describeinstrumentation which exploits one or more of these three variables. Of course. it is to herealized that the addition of other parameters. e.g.. fluorescence polarization, couldprovide an even greater variety of instrumental approaches.

Although this section will be concerned with new developments in fluorescenceinstrumentation, we will first briefly discuss conventional fluorescence instruments.More detailed information can be obtained by consulting one of several differentmonographs.?2 As depicted previously in Figure I. the conventional fluorimeterconsists of an excitation source, an excitation wavelength selector, a sample cuvette. anemission wavelength selector, and a detector. The radiation source for most researchinstruments is the xenon arc lamp. For the excitation and emission wavelength selector, a

I

February 1982 163

grating system is usually chosen over the less expensive filter systems. Thephotomultiplier tube is the preferred detection system in emission studies because of itshigh gain, low noise and relatively broad spectral sensitivity.

A. Time-Resolved TechniquesSince emission decay rates vary from molecule to molecule, an obvious approach to

multicomponent analysis is to exploit these differences in emission lifetimes. A few yearsago, this was most easily performed using phosphorescence emission since thephosphorescence decay is usually very long lived (lifetime 5 1 msec). With suchrelatively long-lived processes, devices such as electromechanical shutters could easily beemployed as mechanical pulsers for the lamp source and for "gating out " the interferenceof the shorter-lived fluorescence signal. A number of experiments have employed suchtechniques.2" -" Although this technique is not new, it does warrant discussion since newapplications of the technique are continuing to be developed. Moreover, it serves as abackground for the more widely used nanosecond time-resolved fluorescence techniques.Perhaps the first published application of time-resolved phosphorimetry to multi-component analysis was that of Keirs et al. in 1957. '2 This early work showed that it waspossible to examine the components in a mixture by proper choice of the wavelengths ofexcitation and emission, using the decay times of the various components.

More recent work by Winefordner's group has explored the possible advantages ofusing a pulsed-source phosphorimeter for time resolution phosphorimetry. - " In one ofthese articles, Fisher and WinefordnerA provide an extensive comparison of pulsed-source and continuous operating mechanical phosphoroscope systems. A number ofadvantages of pulsed-source phosphorimetry are cited, including lower detection limitsand greater discrimination of short-lived phosphors. Figure 7 gives a block diagram oftheir pulsed-source phosphorimeter. Two types of readout methods were used to observethe emission of rapidly decaying phosphors. In the first, the phototube was operatedcontinuously, and the integrated phosphorescence intensity examined by an electronicgate. The gate could be scanned to display the decay curve. The second method employeda continuously operating phototube with a fast multichannel readout device capable ofsignal averaging. The second method was preferred for ease of data processing. Threemajor methods of processing the data were presented, all of which assume that theunknown molecules absorb and emit independently of each other. The first method iscalled the "multiple analytical curve method" and is essentially the same as that originallyused by Keirs et al.2 For a binary mixture, this method requires obtaining analyticalworking curves at two different time intervals, t, and t:. In the original work by Keirset al.. the two variables were X, and ,X., i.e., two different wavelengths of emission. Ineither case, the concentration of each component in the binary mixture can be obtainedby the simultaneous solution of two equations. Of course, this method is applicable tohigher-order mixtures. However, n equations are always required for n unknowns.

The second method employed is called an "exponential method."This method requiresprior measurement of the decay times of the components in the mixture. Thephosphorescence intensity of each component is then expressed as an exponential decayfunction of the time after termination of the exciting light. Assuming that the absorptionand emission processes of each molecule are independent, one can analyze a binarymixture at delay times t, and t 2 . Again, this provides n equations for n unknowns whichcan be solved simultaneously.

The third method used by Fisher and Winefordner is a logarithmic technique used innuclear decay studies. The technique consists of a semilogarithmic plot of decay rate vs.time. For a multicomponent sample, different linear portions will be observedcorresponding to components having different half-lives. Extrapolation of these linear

164 CRC Critwcal Reviews in .4nalvzical Chemistri

A 9 C

0

K

FIGURE Block diagram of pulsed source phos-phorimeter iA Pulse generator iModel 100.Datapulse Inc., Culser Citt. Calif.)( Bl 300- powersuppl. Model C-28 1. Lambda Electronics. Mel..ille.L, I. Y i, (C) 1400-V power supph% (Model 4113 %I.John Fluke Mig.. Co .S~attfe. Wash.. t Dl Triggercircuit. ( L) Xenon tlashtube compartmentlaboratorv constructed). I Fl Sample cell

compartmen t idcentical to amino phosphoroscopeaccessorN minus rotating phosphoroscope can).(iEmission monochromator (Model EL -'00. HeathCo.. Benton Harbor. Mich i. tH) Phototube powersuppl% Model EL-701-30. Heath Co.. BentonHarbor. Mich. I. ill RCA IP28 photomultiplier tubeOf Load resistors (500 f) to 3 Mfland last-Nwitchingdiodies. (K) Oscilloscope (.Model 545. Tektronix.Inc.. Orlando, Fla.). i(LI Preamplifier i Model 4h5A.Hewlett-Packard. Orlando. Fla.). I M) Signal averager(Biomac Model 1000. Data Laboratories. Ltd.. Lon-don. England). or Boxcar Integrator 1 Model I).Princeton Applied Research. Princeton. % J iStrip chart poteniomeitric recorder I Model T R, E HSargent. Chicago. 111.). 10) Synchroniation foroscilloscope and signal .r~erager ( Reprinted withpermission trom Fisher. R. P and Nkinetorciner.JD.. 4nal Chem.. 414. 94M. 19-2 Cop~rtght 19'2American Chemical Societ% j

portions back to time zero after activation (or excitation) provides the initial decay.intensity of each component.

All three of these data processing techniques were applied to the analysis of variousmixtures of phosphors. The results of the analysis of binary mixtures of 4-bromo-acetophenone and benzaldehyde are shown in Table L"' The results of the multipleanalytical curve method (first method) and the exponential method (second method) are

February 1982 165

Table IANALYSIS OF TWO BINARY MIXTURES BY THEMULTIPLE ANALYTICAL CURVE METHOD AND

THE EXPONENTIAL METHOD

Binary Mixtures 1: 2.5 X 10 6 M4-bromoacetophenone(A)1.5 X 10- 6M benzaldehyde (B)

Rel std ErrorMean value found dev(%) (%)

Multiple Analytical Curve Method

C. 3.09 I0X 1 4 24CR 1.30 x I 0M 15 13

Exponential Method

C 2.53 x 10" .' 2 1C1 1.98 x 10"-.1f 12 32

Binary Mixture 11: 1.0 X 10-"M4-bromoacetophenone(A)3.5 X 10-6,AM benzaldehyde (B)

Multiple Analytical Working Curve Method

C, 0.85 x 10-M 35 15CR 3.18 x 10"'.W 26 9

Exponential Method

C, 0.76 x 10-1 25 24

C8 3.45 X I0"',M 21 I

Reprinted with permission from Fisher. R. P. and Winefordner. J. D.,Anal Chem., 44. 948. 1972. Copyright 1972 American Chemical Society.

compared. Table 2 shows the results of the analysis of a ternary mixture by thelogarithmic decay method (third method)." The technique of pulsed source-timeresolved phosphorimetry was later applied to the analysis of halogenated biphenylsr ' andto the analysis of drugsi" using a more intense flash lamp source.

The rate of decay for a fluorescence signal after termination of the exciting source alsofollows first-order kinetics. However, while the phosphorescence lifetimes are usuallylarger than a millisecond, the fluorescence lifetimes are usually in the range of fractions ofmicroseconds to nanoseconds. Basically. the components of a time-resolved fluorimeterare similar to the components of a time-resolved phosphorimeter. Obviously, the muchfaster fluorescence decay rates require more sophisticated electronics. A typicalnanosecond fluorimeter has been described by Badea and Georghiou.'5 Figure 8 shows ablock diagram of their system. The system is a pulsed sampling type fluorimeter, whichuses a boxcar averager for short data acquisition times. The exciting nanosecond lightsource has a repetition rate of about 10 k Hz for an electrode gap of 0.5 mm and an appliedvoltage of 15 kV. The fluorimeter is also equipped for baseline sampling, allowingcorrection for dc drifts. Corrections are made for exciting light intensity fluctuations byincorporating a beam splitter and a second photomultiplier tube for ratioing the

166 CRC Crithcal Reie its in -lnalt-tical ChemistrY

Table 2ANALYSIS OF A TERNARY MIXTURE BY THE

LOGARITHMIC DECAY METHOD

Ternary Mixture 1: 2.3 x 10O'M4-bromoacetophenone(A).r=6.8 msec

2.3 X 10-'. benizaldehyde (B).r =2.2 msec

2.3 X 10 M benzophenone (C),

r =4.8 msec

Rel sid ErrorMvean value found dev 1%) f%)

C8 2.1 x 10, Vl 38 J0C, 3.1 XON 10'V 3 35

Reprinted with permission fronm Fisher. R. P asnd Winetordner. J D0Anal. Chem.. 44. 948. )972. ('op~right J972 AmericatnCimwiSct

BE PM

E:< 'T.ON SPLIrT~EP

I.ZiLE LENSLENS- ----* ENS

AVERAGEP Ir

F16L.RE 8 Block diaigram oi aoeoditio e In he photmultpliers are

manutactured hv Ampere The trgern nigna %%ssrthed h% aTltoni' rise-

time limiter I he input to each channe was ii a d b abou 0 n Frm Bddeis.

N1W G. and Georghiou. S . Rev SCii. /0 31rum.. 41. 314. 1976 W4ith permission.)

fluorescence signal with the exciting light signal. Although not shown in this diagram. thesystem also contained an interface to an I BM 360 computer for data analysis. Althoughmany nanosecond luorimeters do not have all of the features described for this system.the basic components are common to all nanosecond fluorimeters. Other time-resolvedfluorimeters have been described.40' 2 and comprehensive reviews of nanosecondfluorescence spectroscopy of macromolecules have been written by Yguerabide.4'Weissler. 4' and Cundill and Palmer."~

February 1982 167

Table 3CALCULATION OF LIFETIMES OF SIMULATED

CURVES BY DECONVOLUTION ANDGRAPHICAL METHODS'

Lifetime (nsec)

Calculated Error f%)of graphical

Theoretical Deconvolution Graphical method

4.00 4.00 3.92 2.03.00 2.97 3.05 2.72.00 2.00 2.15 7.51.00 1.01 1.26 25.00.50 0.50 1.07 114.0

Simulated source lifetime. 1.10 nsec; width at half-maximum.3.2 nsec: counts in peak channel, 12.000.

From Shaver, L. A. and Love. L. J. C., Appl. Specirosc.. 29. 485.1975. With permission.

A number of works can be cited which demonstrate the possible applications of time-resolved fluorimetry for multicomponent analysis."6-" However, the discussion here willbe limited to the works of Love and co-workers4 950 and a review by Love and Shaver."Shaver and Love 9 have quantitatively evaluated errors in luminescence lifetimecalculations obtained by the graphical slope method, in comparison to mathematicaldeconvolution methods which are more complex and require a digital computer withrelatively large memory capacity. This work evaluated the accuracy of each technique forcalculation of the lifetime of a fluorescence specie. Table 3 shows some typical resultsobtained in this study, 9 calculating lifetimes of computer-simulated excited fluorophoresby the deconvolution and graphical methods. The source lifetime was fixed at I. 10 nsecfor these results. As one would expect, the mote time-consuming deconvolution methodis more accurate under all conditions. However, the authors concluded that. under cer-tain conditions, the graphical slope method might be preferred since its computerrequirements are less stringent. For example, fluorophores with lifetimes of 3 to 5 nsec inthe presence of a pulsed-light source with a lifetime in the I to 3 nsec range can bedeconvoluted with an error of 5% or less. On the other hand, serious errors areencountered when graphical slope-calculated lifetimes are within 2 nsec of the pulsed-source lifetime.

Another study by Love and co-workers0 has evaluated solvent effects on thefluorescence decay times and quantum yields of atabrine and its homologues. This studywas designed to elucidate the excited state properties which allow differentiation betweensimilar species. The effects of molecular structure and microenvironment on the excitedsinglet state properties of an atabrine-based homologous series were determined. Thestructures of atabrine and related species which were studied are presented in Figure 9.The effects of solvent polarity on the important parameters of the excited molecules, suchas lifetimes, quantum yields, and transition energies, were extensively evaluated. Typicalresults are displayed in Table 4. These results seem to indicate that increase in the carbonchain length in the solvents 0. 1 N hydrochloric acid and methanol, acetic acid causes adecrease in the observed lifetime, r, with no apparent trend in the natural lifetime, r.. Ofthe solvents examined, the best, in terms of increased quantum yield for analyzing any ofthe members of the atabrine-based series, is an acetic acid, chloroform mixture.

168 CRC Critical Reviews in Analytical Chemrstrv

it-I1-c- C- -4 Ii

44ciI 4IiI IO~ 2 ([71.- 2A1-C-llCU-i C

11111-c- C12 in:

IkiCi

I

FIGURE 9 Structure oi alabrinc and related specs I .tahrinc iha 'c irmi IIhomologous series n = I to n = " Ibas Lorm). III parent compound iha ' .imtP I%homologous ,erics facid lorm). '. proposed exciled state pecc, ,jid torm,I Reprinted with permission from Los,. L J C - L pion. L M njnd Ritter A I,,4,ai. Ch'm 50. 2059. 1978 (opy right 19" American (_herncai Si, ii

Table 4LIFETIMES (nsec) AND QUANTUM YIELDS FOR SPECIES IV (see Figure 9)

IN SOLUTIONS OF DIFFERENT POLARITIES

0.1 V HCI Acetic acid /methanol Acetic acid / chlotroform

Compound r 4 To r r

III 20 1 - - 199 0865 2' 140 o 210 67

I 32 0056 2 38 005- h' 13 I )316 41II. n = 1 12 0234 54 150 o 358 42 1)" 025 40II.n-2 f? 0090 '14 -4 0135 55 12' 0340 3-I.n3 36 0075 48 40 0089 45 129 031' 41II. n=4 2.9 0021 107 33 0062 53 134 0398 34l.n=5 2.3 002' 85 2" 005" 4' 143 0510 2811,n 6 6 0013 23 .I )050 42 135 0389 35II. n 1 3 0015 j7 2.1 0068 40 14 2 0222 64

Reprinted with permission from Love. L. J C., Upton. L. M and Ritter. A W. 4nal. Chem.. 50. 2059. 1978.Copyright 1978 American Chemical Society

A number of other references should be cited, although they will not be discussed herein detail. Aaron et al.'2 have reported the applications of fluorescence and phos-phorescence lifetimes to the analytical study of important hallucinogens. A criticalevaluation of time-resolved fluorescence for the analysis of multicomponent fluorescentmixtures has been reported in a dissertation by Shaver." Finally. a number of studies in

February 1982 169

the picosecond time domain 5 '5 are providing new avenues of study for fluorescencekinetic studies.

B. Phase-Resolved TechniquesIn our previous discussion we described one method of measuring the time decay of

fluorescence and phosphorescence. An alternative method which has been previouslyemployed for fluorescence decay studies is the phase fluorimetric method. Although onemight consider phase fluorimetry and phosphorimetry as time-resolved techniques, wehave elected to discuss them separately. The phase method is based on the fact that if aluminescing sample is excited with a modulated light signal of angular frequency w, theemitted light will also modulate with the same frequency. However, the emission willsuffer a phase lag, 0, which at a given exciting light frequency, is often characteristic of theemitting molecule. The phase lag is then related to the angular frequency and lifetime r ofthe molecule by

tan 0 = w r (3)

Another useful parameter for the phase-resolved technique is the modulation factor Mwith respect to the exciting light. This parameter is also related to the phase lag andangular frequency and is described by the expression

M =Cos 0 (4)V I+w

These two equations are equally applicable to phosphorescence or fluorescence decay.The luminescence lifetime of the species can thus be calculated using the phase changeequation or the degree of modulation. The emission decay is assumed to follow a simplefirst-order relationship. Therein lies the major disadvantage of phase fluorometry sincethis assumption is not always valid. However. the use of Equations 3 and 4 incombination will demonstrate the validity (or lack of validity) of the exponential decayassumption. Consequently, phase fluorimetry (or phosphorimetry) is a widely acceptedtechnique, and a number of applications to molecular luminescence studies can be cited.For example. Spencer and Weber 7 have described the use of a cross-correlation phasefluorimeter for measurement of subnanosecond fluorescence lifetimes. The cross-correlation technique requires the mixing of the photocurrent produced by fluorescencewith a fixed high frequency close to that of the modulated exciting light signal. The phaselag is then measured in the resulting amplified difference signal. Some of the advantagesof this approach as described by Spencer and Weber are increased sensitivity, increasedS. N ratio and signal isolation, and ability to use numerical time-interval counting forincreased accuracy through averaging. The utility of the technique was demonstratedthrough the determination of the fluorescence lifetimes of NADH (nicotinamide-adeninedinucleotide). FMN (flavin mononucleotide). and FAD (flavin-adenine dinucleotide).Another study by Weber"A has described the theory and application of differential phasefluorimetry to the detection of an isotopic molecular rotation. A wide-range, high-accuracy phase fluorimeter has been described by Schurer et al.59 A number of similarinvestigations are cited by Itoh. 60 Pohoski and Zachara, *' Reseuitz and LippertA andHauser and Heidt.6J

The application of the technique to multicomponent analysis is perhaps bestexemplified by the work of Mousa and Winefordner 6 4 This study demonstrated thepossible use of the technique in analytical phosphorimetry. Using the terminology ofMousa and Winefordner. the total intensity of the sinusoidally varying luminescencesignal at any time t can be expressed as

170 CRC Critical Reviews in ,nalytical Chemistmr

p

F\ S

F.

A rC :KL

FIGURE 10. Block diagram ol instrumental system for phaseIluorimetry. A = 0-40V. 0-30A DC power supply (Harrison Model6268A): B = starter circuit: C = 0-100V. 0-0.2A DC power supply( Harrison Model 61 16A): D = summing operational amplifier andcurrent booster (Heath Model EUW-19); E = modulation circuit: F= excitation and emission monochromator: G = photomultipliertube and housing; H = high voltage powersupply: I = load resistors:J = differential amplifier ITektronix Model I A7Al: K = amplifier(optional) I PAR Model 211): L = lock-in amplifier M = strip-chartrecorder (optional): N = X-Y recorder (optional): P = xenon arclamp; S = sample compartment. (Reprinted with permission fromMousa. J. J. and Winelordner. J. D.. Anal. Chem.. 46. 1195. 1974.Copyright 1974 American Chemical Societ% 1

I, = M L k. I:' cos(wt -8,) + k, L.' cos wt (5)

where M is the degree of modulation for phosphorescence, k, = 2.3 , bc, and k = 2.3 (bbc. The terms 0. and br are the quantum efficiencies of phosphorescence andfluorescence, respectively. The parameter L,' is the amplitude of the exciting light and w isits angular frequency. In Equation 5, it is assumed that the ac component of thefluorescence is negligible since M, is unity and Of is zero over the frequency range of 2 to1000 Hz used for this study.

Figure 10 displays a block diagram of the instrumental system used by Mousa andWinefordner for this study. They employed two methods for resolving the overlappingphosphorescence spectra of a binary mixture. One method is called the -phase method".in which the frequency of modulation is held constant and the instrumental phase anglesfor standards of each of the components in the mixture is measured. Then the phase of thereference signal used for nulling out unwanted signals in the lock-in amplifier is adjustedso that it is 900 + ti, where o, is the instrumentally measured phaseangle of one of themeasured standards. The resultant output signal is then characteristic of the othercomponent (component 2) in the mixture. Similarly, a signal characteristic of componentI can be obtained by adjusting the reference phase angle to 90* + 0p2. The other methodis called the "frequency method". This method involves setting the reference signal at900 + 0,, where 0, is a selected phase angle. A standard of one of the components in themixture is run and the frequency of modulation is adjusted until the phase angle of the

February 1982 171

1.0- .. -

28I.-

o F E 0' C\ A

4-

.2-

0, 23 5 10 203050 100 200300500 100

FREQUENCY (Hz)

FIGURE II. Variation ol experimentally deterrmned degree of modulation, Ni. with the frequenc ofmodulation for several phosphors. A = anthraquinone. rp = 3.0 msec B = 4-iodobiphenyl. rp = 3.5 msec C =benzophenone. r, = 6.0 msec; D = 4.4'-dibromobiphenyl. rp = 12 msec: E = 4-bromobiphenyl. rp = 17. msec: F= 3-bromobiphenyl. rp = 55. msec. (Reprinted with permission from Mousa. J. J. and Winefordner. J. D.,Anal Chem.. 46. 1195. 1974. Copyright 1974 American Chemical Society.)

standard is equal to 0. The result is that the signal from this standard is nulled out atfrequency, ft. A similar frequency, f2, is found using the standard of the secondcomponent. The characteristic signals of component 2 and component I are obtained bydetermining the spectra of the mixture at f, and f2 , respectively, while holding the phaseangle constant.

The experimentally determined degree of modulation, M., and the phase shift angle 0,were measured for several phosphors. These experimental curves agree qualitatively withtheoretical behavior as shown in Figures II and 12. As one would predict fromEquation 4, the M values approach unity at low frequencies and zero at higherfrequencies. One can also use these data and Equation 4 to calculate the lifetimes of thesephosphors (see Table 5). These lifetime values were calculated using the M valuecorresponding to the 50 Hz frequency since this value generally corresponded to thesteepest portion of the curve. One can also use a similar computation to derive thelifetimes from the phase shift data of Figure 12.

The analytical usefulness of this technique lies mainly in its ability to deconvolutespectral mixtures. In their study, Mousa and Winefordner used a number of syntheticbinary mixtures. These mixtures were spectrally resolved using the phase method ofnulling out one of the components. One of the binary systems studied was a mixture ofbenzophenone and 4-bromobiphenyl. The results of this experiment are shown inFigure 13. Spectrum A of this figure corresponds to a spectrum of the mixture measuredat the peak output phase setting. Spectra B and C were obtained by nulling out the signalsof 4-bromobiphenyl and benzophenone, respectively. Thus, the resulting mixture spectra

172 CRC Critical Reviews in A-nalvnical Chemisirl

90so-70

*60-E D5 A

40-

(L 20-J

to0

1 2 3 5 10 20 30 50 100 200 300 500 1000

FREOUENCY (Mr)

FIGURE 12. Variation 01 e.xpcrimentalN determined phase hilt angle. 0. %ith the lrcqucnc )Imodulation lor severalI phosphors. A = a nthraq ui none. r =3.O)msec: B =4-iodobiphen~ I. -= 3 5mscc. C= benzophenone. r = 6.0 msec: D = 4.4*-dibromobiphenyl. r = 12 msec; E = 4-bromohiphenvi. r = 17msec: F = 3-bromobiphenvl. r =55. msec. (Reprinted with permission from Mousa. J J andWinefordner. J. D., Anal. Chem.. 46. 1 195. 1974, Copyright 1974 American Chemical Societ%

Table 5PHOSPHORESCENCE LIFETIMES FROM PHASE AND

MODULATION DATA'

Lifetime Imsec)'

MIolecule Phase Modulation' Time resolved'

Benzophenone 5.9 6.1 7Anthraquanone 310 2.7 3 64-lodobiphenyl 3.5 3.3 3 24-Bromobiphenyl 14 20 13-Bromobiphenyl 55 59 584.4'-Dibromobiphenyl 12 13 12

Solvent: ethanol.Relative errors in lifetimes are = IOCi.All values calculated from measurements at 50 1-1.

Data taken from References 8 and 9 of Reference 64

Reprinted with permission from Mousa. J. J. and Winefordner. J. D. Anal.Chem.. 46, 1195. 1974. Copyright 1974 American Chemical Socict.

for B and C are characteristic of benzophenone and 4-bromobiphenyl. This study alsoexplored the use of the frequency method where the reference phase angle was set at avalue 900 away from a selected phase angle 0-. The frequency of modulation was thenadjusted to null out the signal of the selected component in the mixture. Figure 14 shows

February 1982 173

60I

70 A

60o-

z50 B

- 40 f .

so

.

.,,

20- -.

0200 300 400 So0 600

WELENGTH (am)

FIGUP E 13. Phase-resolved emission spectrum ola mixture o 2.2 x!j',l benzophenone and 2.9 X 10 'M 4-bromobiphenyl at 25 Hi.FYcitation wavelength is 275 nm. A = mixture spectrum at peak phaseangle. OR = 2700 + 390; B = mixture spectrum. (bi = 0' - 26'. C =mixture spectrum. OR = 180

" + 52': (--) = spectrum of ben7ophenonestanc.rd, OR

= 2700 + 520: (---) = spectrum ol 4-bromobiphen.l

standard OR = 2700 + 260. (Reprinted with permission from Mousa.J.and Vinetordner. J. D..Anal. Chem.. 46. 1195. 1974. Cop. right 1974.nerican Chemical Society

the results "if this study. Spectrum B of this figure corresponds to the acquired spectrumof benzov. enone obtained by nulling out the signal of 4-bromobiphenyl. Similarly.Spectrum A corresponds to 4-bromobiphenyl obtained by nulling out the signalcontributed by benzophenone.

Quantitative analysis of binary mixtures was also demonstrated by obtainingcalibration curves of each component in the mixture. Figure 15 shows the results. CurvesA I and B I correspond to measurements at the peak phase angle for 4-iodobiphenyl and4-bromobiphenyl. respectively. Curves A2 and B2 were obtained by nulling out the signalof the interfering analyte. The results obtained indicated that the best calibration isobtained when the concentrations of the analytes are less than 10"9 M. Thus, inner filtereffects must be considered as is often the case in fluorescence analysis. Other quantitativedata were reported, including the determination of a weaker phosphorescence in thepresence of a larger signal. and the use of the frequency method for obtainingquantitative results.

The technique of phase-resolved phosphorimetrv as described above has theadvantage of being independent of the degree of overlap of the components. However, aserious disadvantage is that the technique requiresapriori knowledge ofthe componentsin the mixture. This is a serious deficiency since many analyses require the identificationof unknown species.

C. Selective ModulationThe technique of selective modulation fluorimetry was first introduced by O'Haver

and Parks.6 The principle of the technique can best be explained by use of the diagram

174 CRC Critical Reviews in Analyihcal Chemistry

so

70

60o-

A

40

~-30

20

200 30 400 500 600

WAVELENGTH (im)

FIGURE 14. Frequency method resolved emission spectrum of a

mixture of 2.2 x 10 1 benzophenone and 2.9 X 10 "M 4-bromobiphenvi. Excitation wavelength is 275 nm. A = emissionspectrum of mixture. frequency = 37 5 H7. Olt = 180° - 40. B =emission spectrum at mixture. frequency = 14 7 H7. , = 180

' + 44)0

( --- )= emission spectrum of benophenone standard: I-*-) = emissionspectrum of 4-bromobiphenyl standard. (Reprinted with permissionfrom Mousa. J. J. and Winelordner. J. D. Anal. Chem.. 46. 1195.1974. Copyright 1974 American Chemical Society.)

presented in Figure 16. This diagram depicts the hypothetical excitation spectra for twohypothetical molecules A and B present in a binary mixture. Clearly. selective excitationof either of the components is not possible because of the extensive spectral overlap.However, suppose that the excitation monochromator is capable of rapidly scanningback and forth (modulating) over the interval Ax. Since the emission is proportional tothe amount of light absorbed and we are rapidly varying this absorption. the emissionintensities of the components will also be modulated. However. with the spectral intervalas specified in Figure 16. it should be clear that the emission of B will be modulated

with the same frequency (IF) as the modulation frequency of the excitation mono-chromator. It should also be clear that the A component will be modulated at twicethe frequency (2F) of the excitation monochromator. Therefore. the use of a lock-inamplifier detector would allow one to electronically isolate the signal of component Afrom component B and vice versa.

O'Haver and Parks defined several general considerations for the selective modulationtechnique in their studies. These considerations are (I) the amplitude of the IFmodulated component is usually greater than that of the 2F component: (2) to spectrallyisolate A in the IF modulation mode from B requires that B have a maximum or aminimum which falls on a sloping portion of A: and (3) the conditions for obtaining apure spectrum of A in the presence of B require the determination of the exact conditionsfor nulling out B through a previously prepared standard. Then. pure A from the mixturecan be obtained by analyzing the mixture under the predetermined standard conditions.Another consideration which should be taken into account is the relationship betweenax and peak separation. The authors found that for two symmetrical overlapping peaks.

February 1982 175

r

10 AI

A2

B1 B2

aC

i,

0

0" 1o" ids 16

Concentration (moles/liter)

FIGURE 15. Analytical curses tor 4-bromobiphenl and 4-1odobiphenyl at the peakphase setting tor each molecule and it the phase setting where one component is phasedout. Al = analytical cure tor 4-iodobiphen . (

=270 -55-. A2 = analytical curse

lor 4-iodobiphenvl. (i = 0' - 125'. B I = anal% tacal curse tor 4-bromobphenyl. OR =

2700 - 12.5*0 B2 = analytical curve tr 4-bromobiphenvi. (b* = 180' = 550 (Reprintedwith permission trom Mousa. J J and Winetordner. J D. 4nal. Chem.. 46. 1195. 1974Copyright 1974 American Chemical Societ% f

the optimum A is about twice the peak separation. In addition to modulated emission

spectra, modulated excitation spectra can be acquired by modulating the emissionmonochromator and scanning the excitation monochromator.

Selective modulation has some features in common with derivative spectrometry.However. the differences as outlined by the authors are ( I ) normal spectra, not derivativespectra. are obtained by selective modulation. (2) much larger wavelength modulations(80 nm or more) can be used in selective modulation: (3) entire emission (or excitation)spectra of individual components can be recovered from a mixture: and (4) selectivemodulation is limited to luminescence spectrometry.

A block diagram for the selective modulation instrument used by O'Haver and Parks isshown in Figure 17. The instrument is completely computerized and has been describedpreviously."6 The programmable ramp generator depicted in this diagram providesrepetitive scanning, variable scan rate, and adjustable wavelength limits. Modulationwas accomplished by applying a IS-Hz pseudotriangular waveform to the motor. A

variable potentiometer was used to attenuate the waveform and control the A4 interval.Several examples demonstrating the utility of this approach were described by the

authors, For example. consider the diagram in Figure 18, which illustrates the overlap of

176 CRC Critical Reviews in Analttical (Yzemistrt,

EACJTAJION SPECTRA

A EMIlSSIONRPWLATED 92 F

- B EMISSION

IOUL TOI

WAVELENGTH

FIICRE l6. The principle ot selecti~e modulation. Aand Batre the excitation spectra oftwhypothetical molecules in a mixture. Modulation o1 the excitation waselcngth oser the indicatedInterval AA. at a modulation frequency o1 F Neri,. will modulate the fluorescence intcnsit'. oi 8at the same freq~uency. while the fluorescence intensit% o1 A will be modulated it twice thisfrequency (ie.. 2F-i. (Reprinted with permission from O'Haver. 1. C. and Park. W ' 4.-lnu.Chem., 46. INX6. 1974. Copyright 1974 American Chemical Societ%.t

WII

FIOL RE 17. Block diagram o1 the selcctise modulation system in the eseitatiorimodulation mode. For operation in the emission moduiwaion mode. the control line,to the torque motors are interchanged. (From O'Faser. 1 C .(ireen. Ci L . andKeppler. B. R.. Chem /natron,.. 4. 197. 1973 With permission.

the emission and excitation spectra of the compounds chrysene and benza~anthracene.Possible modulation intervals are also shown. Table 6 summartzes the possiblemodulation conditions for the analysis of binary mixtures of chrvsene and benza)anthra-cene. It should be clear from Figure 18 that determination of benz(a)anthracene in thepresence of chrysene by selective excitation is a difficult task, while selective excitation ofchrysene in benz(a)anthracene is somewhat easier. However, one can use a selective

IIFebruar% 1982 177

2000 4M WAVEFIpGrH I.00)

FIGL RE IN Fluorescence excitation and emission spectra ol I 0 ppm chrscnc and 0 2- ppmben,ioanthracene in c clohexane at 251 C The modulation intersals arc designated h% the %erticallines. , = %,,= I 0mm(S = 10nmr I Reprinted Aith permission rom O'Haer. I C and Park.% M Anal. Chem.. 46. 1886. 1974. Cop~righi 1974 American Chemical Socict\

Table 6MODULATION CONDITIONS FOR THE ANALYSIS OF

BENZ(A)ANTHRACENE-CHR YSENE MIXTURES'

Modulated Detection Center 1AAnalyte wavelength mode A Inm) (nm)

Benia)anthracene Excitation IF 2"o 30

Excitation F1 290 30

Emission IF 3SO 45Emission 2F 400 45

Chrvsene Excitation IF 290 20Excitation 2F -t0 40

Emission I F 400) 45Emisson 21- 30 45

Modulation frequenc. I H7

From O'Ha.er. I C.. Green. G. L . and Keppler. B R . Chem Instrum . 4. 19-.1973 With permission.

modulation approach to :limtnate the interference of a selected component in thepresence of the other. Consider the diagrams of Figures 19 and 20. Figure 19 correspondsto the selective modulation of benz(a)anthracene in the presence of chrysene incyclohexane solution. The diagram shows the observed signals from pure benz(a)anthra-cene. the mixture sample. and a chrysene standard. The signals were obtained using theIF detection mode shown in Table 6 and diagramed in Figure 18. Similar results wereobtained for the selective modulations of chrysene in benz(a)anthracene as depicted inFigure 20. Again. the results, shown in Figure 20. were obtained using the I F detectionmode shown in Table 6. The authors also demonstrated the effectiveness of using this

178 CRC Critical Reviews in .4nalvzical Chemistriy

UIIU

OINYSEME

200 WAVEIJ*G1G (mr) ii00 SEm0

FIGURE 19 Selective modulation of 0 27 ppm beni[alanthracene in thepresence of 1.0 ppm chrvscne in cvclohexane at 250 C W11 1. =[mm

(S = 10 nm). IF. excitation modulation mode: .1A =30 nm at 2'0 nm. Thespectra of chryscne and ben/iauanthracene alone are also given. as indicated

Reprinted with permission from O'Haver. T C and Park. W \M . nai.C'hem.. 46~. 1886. 1974 Copyright 1974 American Chemical Society

CIWSEM SINCTIA

lN~a)APYtMIUAW

20 VAVELLMATO SMi~

FIGURE 20. Selective modulation of 1 0 ppm chrywene inthe presence of 0.2" ppm hen7laianthracene in .sclohexane at250 C ( Reprinted with permission from O'Haver. T C andPark. W W.. A'sa. Che, 46. 1886, 1974 Copyright 1914American Chemical Society.)

technique to generate analytical working curves to determine the unknown concentra-

tion of a given analyte in the presence of another, The effectiveness of this approach isamply demonstrated in Figure 2 1. while Figure 22 compares the concentration effects ofchrysene on the benz(a)anthracene signal using selective modulation and selective

Februarv 1982 179

0.2 0.6 1.0svp 8ENZ~a)AINTROACNE

FIGURE 21 Selectise modulation .inahtical cur~.e lorbenzlalanthracene in c~clohexane containing 1.0 ppmchrysene. Conditions are as in Figure 19 (Reprinted withpermission from O'Haser. T C. and Park. W M.-AnalChem.. 46.1886i.1974. Copyright 1974 American ChemicalSoci t V.

(0 SELECTIVE EXITATION E.MISSION A[ 3%w.£E ' M9LTI EISS ION AT 415ie

ELECTIVE EXCIATION ERISSIONi AT 415me

AD

II T

FIGURE 22. Effect o1 the concentration of chrysene on the electiseexcitation and selective modulation anal'.tical signals of 0,27 ppmbenza tanthracene in cyciohexane. t Reprinted with permission lrom O'Haser.T. C. and Park. W. W. Anal. Chem.. 46. IM6. 1974. Copyright 1974 AmericanChemical Society.)

excitation. Clearly, in this case the selective modulation approach provides better results.The technique of seiective modulation was demonstrated to be a new and interesting

approach to analysis of simple multicomponent Mixtures. However, as is the case withphase-resolved phosphorimetry. it suffers from the disadvantage that the identities andspectral properties of the mixture components must be known before separation can be

180 CRC Critical Reviews in Analvii(al Chemistry

achieved. For additional information, a detailed study has been provided by Green andO'Haver' " on derivative luminescence spectrometry.

D. Total Luminescence SpectroscopyThe many parameters of luminescence previously discussed indicate that a large

amount of information and. consequently, greater specificity can be obtained byexploiting a number of these parameters simultaneously. One such technique is aninstrumental one, first called "total luminescence spectroscopy" by Brownrigg et al. " andGiering at Baird Corporation. "' The name "total luminescence'* is somevhat of amisnomer since, quite often, only the fluorescence at room temperature is measured.However, the name has come to be synonymous with the approach, and it is futile toattempt to change it at this point. Generally. the total luminescence techniqie involvesthe acquisition of the fluorescence intensity of a sample as a function of multipleexcitation and emission wavelengths. The acquired data set is then in matrix form withthe elements of the matrix representing relative fluorescence intensity, while the positionof each represents a given excitation and emission wavelength. Since. for most moleculesin fluid solution, the fluorescence spectrum has been found to be independent of theexcitation wavelength and the excitation spectra has been found to be independent of themonitored emission wavelength, we can represent the fluorescence matrix for a purecomponent in matrix form as

M = a )

where a is a concentration-dependent parameter. x is the observed excitation spectrumrepresented as a column vector, and v' is a row vector representing the observed emissionspectrum. In dilute solution, where synergistic effects such as energy transfer andquenching are negligible, the fluorescence matrix for multiple components is simply thesum of the contributing matrices. i.e..

F1

= l)X(II i iI7)

where n is the number of fluorescent species and t is an index of each. For illustration.consider the diagrams of Figure 23. The diagram in Figure 23a corresponds to thefluorescence matrix obtained for a hypothetical compound %4ith assumed doubleGaussian peaks corresp." ding to the emission spectrum. The wa,,elength distribution ofthis hypothetical spectrum is shown at the top of the emission axis in this diagram. Alsoshown is the Gaussian-distributed excitation spectrum with its wavelength distribution.The fluorescence matrix is shown here as a contour map with various contour linesrepresenting isointensity data points. A similar matrix is given in Figure 23b fora secondhypothetical compound with two bands each in its excitation and emission spectra. Ifthese compounds were present in a binary mixture at the intensities (concentrationsIdepicted. then the fluorescence matrix would appear as shown in Figure 23c. Of course.the data can also be represented as an isometric projection of three dimensions(Figure 23d).

It should be apparent from our discussion above that the total luminescence spectrumhas considerable utility for multicomponent fluorescence analysis. This utility has oftenbeen exploited in a simpler form through the use of selective excitation to isolate theemission of individual components."' However, several investigators have recognizedthe improved selectivity which can be obtained by acquiring and retaining large amountsof data as we have described for Figure 23. The most direct approach to acquiring the datais through a computerized conventional fluorimeter. For example. it is a simple task

February 1982 181

400 !00 600 '00 400 500 "10

EIASSON wAvE..ENGT.. EMI ~SSION AE..-

(C)

CZ

400 500 600 '0

EMISSION RAVEILENGT, .,

(d)

~5

EMIsSONI WLVEL E*4Gr"'

FIGURE 23 Fiuoireswence matrix tor j hspothetical compound. (From Warner. L M' Fogarty. M. P..and Sheily. D. C.. Anal. Chn.Atla. 109. 3'61. 1979. With permisbionh)

through computer control to fix the wavelength of excitation and obtain and store theemission spectrum. The process can be repeated at several additional excitationwavelengths until the desired fluorescence matrix is obtained. This approach has beenemployed by several investigators including Gering,69 Haugen et al..'2 and others.'!'-The problem with this approach is that even with computerized instrumentation. the task

182 CRC Critical Revze%% s in -)nali tical Chemistri

CUVETTE EXCITATION AND

IMAGE EMISSION SPECTRA

Red Emsu..on.

Orange -

Blue

BM All

Soo 500 V o 00

DISPERSED ISOMETRICI MAGE PROJECTION

C - D

HGt RE 24 Production of the emission-excitaion matisl\ Irh , poheiwai c.ompound~ i-A) 7he h 'poihciieal excitation andemission spectra B) 1the 11uore~ccnr cu~ettc aj-Iuminated h% thepokchromatic beam i0 he emissofl-excitatiuf matixl prod ucedat the exit dit aperature of ihe anaIl itng poikchromator D) Anisometric projection of the emission-excitaion matirix ofFrom Warner. 1. 'A.. Calks. JV B. Da~idson. F R. and

Christian. 6 D. C/jo Cherr.. 22. 14X3. 1976 With permisswn

of obtaining the desired data can be time consuming and tedious (and may be plaguedwith instrument drift during the lengthy measurement periods). Recentl . a com-mercially developed computerized Iluorimeter designed to acquire total luminescencespectra has been described." This system is described as requiring 1 1/ hr to record acomplete total luminescence spectrum. The commercial dev.elopment of such instrumen-tation is a further example of the usefulness of the total luminescence approach.

An alternative to the time-consuming mechanical scanning approach was firstproposed and developed by Callis and co-workers.'h" This new instrument employs anovel concept of polychromatic sample illumination to eliminate the necessit% of time-consuming mechanical scanning. Consider the diagram shown in Figure 24. InFigure 24A the dashed curve represents the emission spectrum of a hypotheticalcompound while the continuous curve represents its observed excitation spectrum. Thepolychromatic illumination can be achieved as shown in Figure 24B by exciting thecuvette simultaneously with spatially and spectrally resolved polychromatic light. Thisspatially resolved polychromatic source can be obtained by focusing a continuum sourcesuch as a xenon arc lamp into a monochromator which has been rotated on its side and

February 1982 183

1 -------------------------------------- 1SIT analyzing

OMA-Z multchannel FS uOetector pOlyChromotor

Field lens . Lens System

ExiainExcitation

monochromotor

L ---------- --- -- - --- -------- ----- -- -- -- -- -- - ---- -- -- -- -- -- --

FIGURE 25. Optical layout ot rapid scanning fluorimeter. (From Warner. I. M.. Fogarty.M. P., and Shelly. D. C.. AnaL Chim. A'ta. 109, 361. 1979. With permission.)

has its exit slit removed. 76 Alternatively, a flat field polychromator can be used as theexcitation source in a similar fashion.'t In either case. the observed curvette image will beas shown in Figure 24B. The observed bands correspond to the relative excitationspectrum of the compounds, i.e.. the fluorescence intensity as a function of the excitationwavelength. This cuvette image can then be focused onto the entrance slit of a

polychromator or monochromator with the exit slit removed. Each fluorescence bandwill then be dispersed into its component emission wavelengths while the y axis

information is spatially preserved. The resultant image (fluorescence matrix) at the exitplane of the analyzing polychromator is as shown in Figure 24C. Alternatively, the datacan be displayed as an isometric projection as depicted in Figure 24D. It should berealized that this image is acquired over a given spectral region without the need for anymechanical scanning. Therefore. the time limitation in acquiring these data is from thedetection system. The rapid acquisition of the two-dimensional spectral data is mosteasily achieved with a two-dimensional light transducer such as a multichannel detector(e.g a television camera).uOs Such a detection system will provide the acquistion of a256 < 256 fluorescence matrix in a time frame of less than 17 msec. This is a considerableimprovement over the hours required on conventional instrumentation. A more versatileversicn of this "video fluorimeter" has been developed by Warner et al." Figure 25prov'des an optical layout of their system. which was designed to eliminate muchunneeded transfer optics by use of a holographically ruled, flat field polychromator.Thu.-. a considerable improvement in spectral quality is realized. This improvement isnicely' illustrated by the isometric projection of an et'hanolic mixture of 3 X 10-" Manthracene and I X 10- Mf perylene. The detection electronics of this system is providedby the EG & G Princeton Applied Research OMA-2 system. This detection systemprovides great versatility and a wide dynamic range for spectral data acquisition.

The video fluorimeter should continue to provide new avenues for multicomponentfluorescence analysis. However. a major difficulty of this system is in the handling of thelarge data base generated. This problem will be discussed later in our section onmulticomponent data analysis.

Id

184 CRC Critical Reviews in AnalYtical Chernistri

I ¢ I-

g0DgUSSM~ WAVUDGTh, min

EMSSON WAVELENGTH, ni

FIGURE 26 loral luminescence and snchronou%

pecctrum oi h.pothetical binar% mixture.

E. Synchronous Excitation FluorescenceThe synchronous excitation method was first introduced by Lloyd in 1971."' The

technique is multicomponent in that spectral information can be acquired for most orallof the fluorescing species in a mixture. For example. consider the total luminescencecontour depicted in Figure 26. This diagram corresponds to the emission-excitationmatrix (EEIM) of the hypothetical binary mixture described in the previous section of thismanuscript. The synchronous spectrum of the mixture is shown at the bottom of thediagram. The synchronous excitation method differs from most conventionalfluorescence methods in that the excitation and emission monochromators are scannedsimultaneously with some constant interval difference "Ix". Thus the synchronousexcitation spectrum of Figure 26 corresponds to a plot of the fluorescence intensity alongthe line parallel to the scattered exciting light vs. wavelength of fluorescence. Thisspectrum shows two distinct wavelength maxima, one from each component. It shouldbe clear from this diagram that one can vary the shape and number of components of thespectrum simply by varying the interval "I'* between the emission and excitationmonochromator. It should also be apparent from the diagram that the spectralinformation provided by the synchronous spectrum is also contained in the totalluminescence spectrum. Thus, as with any information set which is a subset of another.the synchronous spectrum of a mixture is more ambiguous than the total luminescencespectrum. However, the synchronous method has the advantage that it can easily beacquired on a simple fluorimeter capable of simultaneously scanning the excitation andemission monochromator.

February 1982 185

iii

WavvAength (nm)

FIGURE 27. Emw4sion spectra (uncorrected) o aliquid chromatography traction containingbenzolk)lluoranthene. ben~ola)pvrene and perylene.(i) Excited at 308 nm. (i) Excited at 385 nim.(ii) Excited at 404 nm. The solvent is one volume ofdtethvl ether in three of 40 to 60' light petroleum.

IReprinted by permission from Lloyd. J. B. F.Nature. 231. 64. 1971. i, 1971 Macmillan JournalsLimited.)

To illustrate the potential of the synchronous excitation method, consider thediagrams of Figures 27 and 28. Figure 27 displays the uncorrected emission spectraobtained at different excitation wavelengths of a mixture from a liquid chromatographyfraction containing benzo(k)fluoranthene. benzo(a)pyrene. and perylene. The spectrumof each component of the mixture was obtained by selective excitation of the mixture,and Figure 28 shows the synchronous excitation spectra of the pure individualcomponents as well as that of the chromatography fraction of Figure 27. Clearly, thesynchronous excitation spectrum of the mixture provides information about each of thethree components.

Since its development by Lloyd. the synchronous method has received wide use for theanalysis of complex fluorescent mixtures. Most of Lloyd's applications of the techniquehave been in the area of forensic chemistry. For example. Lloyd has employed thetechnique to characterize rubbers, rubber contact traces, and tire prints."' In addition, hehas explored the use of the technique in conjunction with quenching to characterizecomplex mixtures. 16 Results of this study showed that quenching could be used incombination with synchronous excitation to provide increased structural features where

cope itrs: eut fti tdysoe htqecigcudb sdi

136 CRC Critical Reviews in Analytical Chemistry

S III

PLI

FIGURE 28. Emitsion spectra iuncorrected) syn-chronously excited at an interval ol 23 nm.it) Benzo(k)lluoranihene. ( iBen7oia)pyrene.

(nII) Perylene. fiv) The chromatographN fraction ofFigure 27. The solvent is the same as in Figure 27(Reprinted by permission from Llovd J B F..Vature. 231. 64. 1971 C 1971 Macmillan JournalsLimited.)

much spectral similarity exists without the use of a quencher. For example, considerFigure 29 which shows the synchronously excited spectra of a naphthenic oil incyclohexane and in chloroform. One notes considerable differences in the characteristicsof the two spectra. The fingerprtnttng applicability of this approach is demonstrated inFigure 30. Two solutions of gear oils with similar spectral characteristics are compared.Although little differences between the two are noted in the cyclohexane solution,considerable differences are noted in the partly quenched chloroform solution. John andSoutar have also explored the use of synchronous excitation spectrofluorimetry for theidentification of crude oils1' with some degree of success.

More recently. Lloyd and Evett" have developed a theory for predicting peakwavelengths and intensities in synchronously excited fluorescence emission spectra.Using the assumption that the excitation and emission peak maxima could berepresented as Gaussian peaks, they were able to predict the peak maxima in thesynchronous spectrum with a reasonable degree of accuracy. That is, the mean differencebetween calculated and observed maxima in the synchronous spectrum was 1.95 nm.

Vo-Dinh 19 has also developed a methodology to improve the selectivity of the

February 1982 187

C

U

~A3O & O 4 500

Wavelen th/nm

FIGURE 29 Synchronously excited Npccr130 nm) ol a naphthenic oil (I mg me ).(A) cyclohexane. (B) chlorolorm. B i at len timesgreater sensiti i y than A. (From Lloyd. J B I-.Analyst, 100. 82. 1975. With permission.)

D

0C

8

CCC

Ul.

8 18

A A300 M5 400 450 5o0

30 Wavelqh/r

FIGURE 30. Synchronously excited spectra(30 nm) of solutions (0.2 mg me ) of gear oils. Aand C are from one sample in cyclohexane andchloroform. respectivel,. B and D are rorn a secondsample. (From Lloyd. J. 8.F Anal rst. 100, 82. 1975.With permission.)

------------------------------------------------------------------------ '

188 CRC Critical Reviews in A1nali-twal Chemistri

E i EO ki h

10 1

IL RA '1 1cciicrpenaif ff~bn-arwn ic ~rne o ~m~uFrom~ 4~-~h 2 mat 5hm 0 1Q' 29~ 3o~lh I5t 6mrcl Xhm~jISu

synchronous excitato tecniqu nlmnseces)rmtv n i icsin

20' 40.

February 1982 189

:3 H I

4

z

3z

2 -4

V) 0-

450 500 450 500 450 500

WAVELENGTH (rm)

FIGURE 32. Ei lci oi ihe AaeleIngth nter.al on the , nchr,)nou,fluorescence nanuioii ietracene ijdi A = 3 nm lbi £A = 1(tnm C =45 nm. (Reprinted ith permision Irom %o-Dnh. I , -ijoul (ijPr N(

396. 1978 Coprght 19S American (hcmacii 1,occt.

about the same. Figure 31c shows how band narrowing occurs in the S'.nchronousspectrum. In this paper Vo-Dinh also demonstrated how omplification of thesynchronous spectra can be produced by variation of AA. Figure 32 Nhows how thesynchronous spectrum of tetracene varies as a function of -IA. In egard tomulticomponent analysis, it was demonstrated that the positions of S. nchronous signalscould be predicted for a group of pelynuclear aromatic compounds. This prediction isamply displayed in Figure 33 where X and A., are the wavelength of emission and thelongest wavelength of absorption. respectively The shaded areas in th;s figurecorrespond to the predicted spectral regions in the synchronous ,pectrum ofnaphthalene. phenanthrene. anthracene. perxlene, and tetracene which agreed well withdata presented by Vo-Dinh.

Another interesting factor useful for improving selecti itv in phosphorimetr. has beensuggested by Vo-Dinh and Gammage. Their approach is based on the synchronousexcitation technique. employing the energy difference between the phosphorescenceband and the absorption band of the compound."

Recently. Latz et al.4' have cautioned users on the limitations of s~nchronousspectrometry for multicomponent analysis. The% demonstrated how the loss ofinformation in synchronous spectrometry can lead to erroneous interpretation of data.

F. Miscellaneous Instrumental TechniquesA number of instrumental techniques are becoming useful or have recently been shown

to be useful for multicomponent fluorescence analysis. While a complete discussion oneach of these topics is not possible here, their usefulness as multicomponent techniques influorescence spectroscopy merits discussion. Therefore. a brief description of each isgiven here. and references to more extensive reviews are provided where applicable.

1. Double-Beam Fluorescence SpectroscopyOne of the few advantages often mentioned of absorption spectroscopy over

fluorescence spectroscopy is its ability to compensate for blank interferences. In recentyears. some investigators have recognized this inherent shortcoming of- fluorescencespectroscopy: consequently, double-beam fluorescence spectroscopy has been developed.

yers smeinesigtos av rconiedths nhret hotcmig 1 lurecem

190 CRC Critical Revtews in .Analvtical Chemistri

o ld I I

300 400 500

-00 - f

400 li

I I I I_ ,--500

(rim)

FIGURE 33. Data chart for synchronous spectra. (Reprinted withpermission from Vo-Dinh. T. Anal. Chem.. 50. 396. 197h Cop~right1978 American Chemical Society

One such double-beam system has been described by Anacreon and Ohnishi.i- Theirdesign incorporates several modes of operation into a single fluorimeter. Figure 34provides an optical diagram of their fluorimeter. In this system, the radiation source ischopped similar to a double-beam spectrometer at a frequency of 60 Hz. then is focusedonto the entrance slit of the excitation monochromator. The chopper is connected via ashaft to a sector mirror located after the exit slit of the excitation monochromator. Thissector mirror is used to alternately excite the sample and the reference blank.Consequently. the emitted fluorescence which is viewed at right angles is alsointermittently focused onto the entrance slit of the emission monochromator. Thus, thechopper allows temporal separation of both beams which are therefore alternatelydetected by the photomultiplier tube. Each of the two sigials is then discriminated usinga lock-in amplifier, and the reference signal is electronically subtracted from the samplesignal.

To demonstrate the utility of this technique. Anacreon and Ohnishi gave severalapplications. For example. Figure 35 shows two overlapping Raman bands of distilled

February 1982 191

M2 M3

I. Opt.. -4 291 i95 Mit permision

'CC9 AsLMM

SO RMi AO F12 N

M4 S

Apl Ot,142. JN9'5 MCCCerisio.

200

660

250 300 350 400

WAVELENGTH (nm)

FIGURE 35, Overlapping Raman bands of H:-Oand ot HSO.. (From Anacreon. R. E. and Ohnishi.Y . Appi. Opt.. 14. 2921. 1975. With permission.

water and 0. 1 NV H.S0 4. A second Raman spectrum of distilled water is also shown forcomparison. Figure 36 shows the Raman band of H zSO, alone obtained by subtractingthe Raman band of distilled water in the double-beam mode. The flat spectrum obtainedfor distilled water in both beams is also shown. The uttlitv of this approach is further

192 CRC Critical Reieiis in -Inalitical Cheinistri

go-

C-'N Ii 2 S04 VS DISTIL..ED

H2 0C REFERENCE

z

40 C4 TILLED WATERIN BOTH BE4MS

\E-SUSTRA c WMCDE

20 MODEL -2

2 t; 300 35C 4C-wAvELENGTH m

FH(,t RE Its [)touhlc-Njm spcc rumnoW SO. Riman h.*nj ;Frorn -njtcrern R L ano

Ohni,hi. N 41'p 1)f, ' 4 '9' 'A ini Vpri-on

so

40 Em mfT .3uSWRC MWE

-j 11110M512

20 SOLAilOF vt AP

@soum wil vs mvm4

0

FIGU RE 3V Double-beam Fluorecnc Npe irm )I an-thracene in c'.clohixane. From Anacreon. R E nd Ohnp~hi 'F

..ppl. Opt.. 1-4. 292L. 19"5 With permission

Februar\ 19x2 193

MONOCHROMR A M~PLE M00uLEI 1f201 0~ CONSTOR FILTERS IPumpI TEMP. BATHII

P TSUPLY~ T IM0NOC'IOMATOR

LAPLY O R FILTERS

PRINTER

CURRENT TOI TRIGGER/VOLTAGE .01H.- IMH&

I CONvERTERQ TIME BASE

Cu ENT TO-- --- iGTAL

CuRNT0 -- RATE METER TVOLTAGE - PINTER

-CONVERTER/i

Divih-jOcEoE

FIGURE 38 Block diagram of the general luorcscene reaction rate nstrument IReprinted %ith pcrmi,-sion from Wilson. R. L. and Ingle. J D. 4hal Chern . 49. ION). 19'- Cop.right 19" -merican ChemicalSociety )

demonstrated using a fluorescence spectrum of anthracene dissolved in impurecyclohexane. The fluorescence of the impurity is evident in spectrum A of Figure 37. Thiscorresponds to obtaining the anthracene spectrum vs. an air blank. Spectrum B showsthe same solution run in the double-beam mode vs. a solvent blank.

The work presented by Anacreon and Ohnishi adequately demonstrates the utility ofdouble-beam fluorescence for multicomponent analysis. However, one must recognizethat this instrument does not solve all fluorescence blank problems. In many analysisschemes, the source of the fluorescence interference is unknown. Therefore. no blanksolution is available for background compensation. For additional information ondouble-beam fluorescence spectroscopy. consult the review article by Porro andTerhaar."

2. Multicomponent Kinetic Studies by FluorimetrvFluorescence spectroscopy has frequently been used to monitor multicomponent

reaction rate kinetics. Wilson and Ingle 4 have recently described the design andoperation of a fluorimetric reaction rate instrument. Figure 38 provides a block diagramof their system. In their paper. Wilson and Ingle provide a critical evaluation of thesystem, including precision of the measurements. background signal. and S N ratio. Theauthors have also demonstrated the application of this system to the determination ofsilver. 9" This reaction rate method is based on the enhancement of the reaction betweenoxine-5-sulfonic acid and persulfate by silver ions. Figure 39 shows the emissionspectrum of the fluorescent product produced in the Ag'. oxine-5-sulfonic acid andpersulfate. Figure 40 shows the silver calibration curve from this study and Table "gives a

194 CRC Critical Reviews in AnalYtical ChemistrY

Table 7SUMMARY OF Ag* INTERFERENCE STUDY

DATA

Species Concentration (ppmn) Concentration (ppm')

,at l) 300Ki ) 300MgI ) 300Balm 300Sri 11) 300PbODh 300Cat I) 30 150(1:(0.95) 1Cu(Il) 30 300(1:(0.)3))Fei I 1) 3 30(1:(070)H g 11) 3 3041 :052l)CriIbI 30 150(1;(o98.0 89))Coti11) 3 304 E: .I MNtIdl 3 301 Eu) 4))

Z ni11) 3 3(E:)).))CddII) 3 30(Iob0.95,0 !'H

Mni11) 3 301Ei 1 01. 1.08))ri.1 1) 3 300,498.0.92))Vl!v) 003 0.30'(0.9f).0.80))Sn 11 0,3 NEI)SnIV) 0.3 3(EI)

AIII 1) 0.3 3) E :1 IH)Zr)IV, 0.0) 0. 111 )0.95))H fi I V) 0.) Iz( 0. 76) )Pull!) 0.3 3(1,(0.93.0,10))AudI!l) 0.03 0.3(E ( 1 26.) )so, 1) is Vf

130" 0.15 VJCIO ~ 0.)2 VtNO 10 1 0.)6 tfl0 (95.0 42) tC) 10~ V/ 0 ,4(I'(0.95.).Xl))F 0.0) (1 0.5 IIIE)).2fl

Concentration which had no effect on the rate oi the blank orAg-enhanced reaction.Concentration which increased or decreased the rate ol either theblank or Ag-enhanced reaction E, = ion enhances the Ag reac-tion more than the background reaction. E: = ion enhances boththe Ag and background reactions equally. E, = ion enhances thebackground reaction more than the Ag reaction. 1, = ion inhibitsthe Ag reaction more than the background reaction. I - = ion in-hibits both the Ag and background reactions equal)h Relatiseincrease or decrease of blank and Ag-enhanced (0 5 ppm -g*)rates are given in parentheses.

Reprinted with permission from Wilson. R. L and Ingle. J D..A4na. Chem.. 49. 1066. (977 Copyright 1977 A\merican ChemicalSociety

list of interfering ions for this method. This example involves a one-component system.but one can easily conceive of using this system for multicomponent analysis.

As an example of another fluorirnetric rate fluorimeter. Fogarty and Warner"" haverecently shown that the video fluorimeter could be used as both an initiator and probe ofphotochemical reactions. To demonstrate the utility of this approach. they chose to study

February 1982 195

10

E

6

4

2

350 400 450 500 550 600X (n m)

FIGURE 39. Ag " - OXSA - S:O, emission spectrum. [Ag'] =

I ppm. [S:O; ]= 0. 1 M. [OXSA] = 10 ppm, R,= I1 U . r= 01 sec,EP,,, = 800 V. A-, = 366 nm. scan rate = 100 nm mi. excitationslit = 2.0 nm, emission slit 0.5 nm. (Reprinted with permission fromWilson. R. L. and Ingle. J. D. Anal. Chem.. 49. 1060. 1977 Copyright1977 American Chemical Society )

5.0

0

4.0 /o

/in 0S3.0D 00/

02.0 -0

0.0

0.0oJo w .0 2th 3.o 40 5'0

FIGURE 40. Ag calibration curve. [OXSA] =ppm, [SzO 'q =0.1 M. ph = 2.2. other conditions are the same as in Figure 39.

(Reprinted with permission from Wilson. R. L. and Ingle. J. D.. Anal.Chem.. 49. 1060. 1977. Copyright 1977 American Chemical Society.)

196 CRC Critical Reviews in -nalhtical Chernistr'

W4

A

FIGURE 41. The EEMs of the reaction of chloroform :th anthraceneIA) T = 0 1B) T = 15 hr. (C) T = 30 hr. (From Warner 1. . andFogarty. M. P . ppl. Spe'iir .34. 43H. 1980 With permission.)

the photo-induced reactions of anthracene in polchlorinated alkanes. For example.consider the data displayed in Figure 41. This diagram shows the EEMs ofthe reaction ot

chloroform with anthracene alter photoinitiation ol the reaction at times t = 0. 1.5 hr.and 3 hr. Figure 42 shows a spectral deconvolution of the data presented in Figure 41.while Table 8 presents a tabulated list of the calculated rate constants of the reactions. Allreactions studied were found to be first order.

Based on the two studies presented here. it is reasonable to conclude that the use offluorescence for multicomponent kinetic measurements will continue to be a .aluabletool in the future.

3. Fluorescence Detectors for Liquid ChromatographyPerhaps one of the most useful aspects of fluorescence spectroscop for multi-

component analysis is as a detector for separation methods such as liquid chromatog-raphy. This usefulness arises from the selectivity and sensitivity of fluorescence describedearlier in our introductory section. The variation in designs for fluorescence detectors forliquid chromatography ranges from conventional fluorimeters to less conventional typessuch as the video fluortmeter. A number of examples ol conventional detection can becited and will not be discussed in detail here. For example. Johnson et al.4" andSlavin et al." have described conventional fluorimeters as detectors in high performanceliquid chromatography. Less conventional designs have included the free-falling dropdetector developed by Martin et al.." which was designed to minimize the instrumentallimitations of conventional flow-through fluorimeters such as large dead volumes. A

a

February 1982 197

HI-ILh 41B

TIME T=3.0 (HR5;

I I

FI(LRE 41(

diagram of this detector is shown in Figure 43. The detection chamber is spherical indesign. The eluent enters this chamber in the form of a drop from a Teflono needleconnected to the chromatographic column. The drops are collected at a hole in the

198 CRC Critical Reviews in A4nalvzical ChemistrY

(B

-00

-0

o 0

5-0.

L(C) (0)4

o CA

I0

0 C!

7Ci'A.319.C) lI.L6981 L t.3 73 "ID)

EXIAINWVLNGH(M MSIN AEEah IM

FIUR 42 hcairehd1cnvltinoth AofFg 1B A nhrceeectto pcrm

zB znhaeeeiso pcrm C htpoutectto pcrm D htpouteisospctum (Fo anr .'.adFgry .P pl. pcrs 4 3.)8 ihpriso.

TabeCALULTE RTECOSTNT

Ant*cn dea routomto

LUl, = n l )- lUtI( - -I )- Iil i

CCI D.M 3 ) 2.26 318 ".23 2 .4 97 109 '.0 3.9 1 0.93 S 7.3 56.'

spcrm From Foarer. . . and Woarnr. 1M. . App. Spcrosc.. 34. 43 8 190 With riso.

permission. Podetromaio

February 1982 199

Needle Emission filter

Ptotomultipher I Photomultipher 2

Light

light entrance

FIGURE 43. Free-failing drop detector. (From Martin. F..Maine. J.. Swelley, C.. and Holland. J. F.. Clin. Chem.. 22. 1434.1976. With permission.)

11O

75 375 75 075

Sucrose (nmoles)

FIGURE 44. Typical vity of free-fallingdrop detector. The indicated amounts of sucrose wereapplied to the anion exchange column 86 mmol Rborate buffer and eluted isocratically with 0.86 mol qborate buffer (pH 8.8). (From Martin. F.. Maine. J,Swelley. C.. and Holland. J. F.. Clin. Chem.. 22,1434. 1976. With permission.)

bottom of the sphere which leads to a waste receptacle. A drop ofeluent passing throughthe sphere scatters exciting light which is unfiltered and therefore corresponds to totallight within the integration sphere. The output from photomultiplier I is used as a timingsignal to process the fluorescence signal measured by photomultiplier 2. The lightstriking photomultiplier 2 is filtered to admit wavelengths in the range 320 to 395 nm.Figure 44 shows the typical sensitivity of this detector, and Figure 45 shows thechromatogram of a reference mixture of saccharides.

An objective of many detectors designed for liquid chromatography is to capture as

200 CRIC Critical Reviews in Analvtical Chemistri

$me"ns (3OPgI

0 2 4 6 10 12 14 4 i

Negrs

FIGURE 45. Chromatogram of a reference mixture of saccharides. The sample contained '75 nmol each ofsucrose. lactose. globoside tetrasaccharidle. G L3 trisaccharide. and 150 nmol of xylose and glucose inO0. 15 micof86 ninat Q ("107%) borate buffer. The gradient maker initially contained 101-r borate buffer in chamber 1.3W,, borate buffer in chamber 2. 5W,~ borate buffer in chambers 3 and 4. and IOO- borate buffer in chambers 5through 9. Each chamber contained 20 msi of solution. The solvent programmer was set to withdraw solventfrom the gradient maker for 16 hr and then from the limit buffer reservoir for 6 hr (From Martin. F.. Maine. JSwelley. C.. and Holland. J. FV. C/in. C/hern.,.22. 1434. 1976. With permission.)

(A PERKiN ELMER2 OMA NWP 2A

LUEX BANDWIDTH4 10 E X AND EM BANDWIDTH4S.EM BANDWIDTH4 0 5 rwi nn

nm 41 6mMDa

FIGURE 46. Fluorescence emission spectra of Ovalene (plAtic sidi Ex A =3503 nn. (Reprinted withpermission from Jadamec. J. R.. Sauer. W A.. and Talit. Y.. Anal. Chem.. 49. 1316. 1977 Copyright 197'American Chemical Society.)

much information as is possible in real time. Consequently, it is expected that rapidscanning multichannel detectors would find applications in liquid chromatography. Onesuch system has been described by Jadamec et al. "~ This system uses a Princeton AppliedResearch Corporation (PARC) Optical Multichannel Analyzer (OMA. Model 1205A)to control an intensified vidicon detector (PARC Model 120'5D) interfaced to a FarrandOptical Mark I S pectrofluoro meter. The spectral coverage of the detector system was110 nm. Figure 46 provides a spectral comparison of this system with a Perkin Elmer

February 1982 201

6

55

+ 234

~O. 13

~cl, 171

19 222?. 1 12

MIinNUTE

21

8 26~ 31 32

0 20 40

MINUTES

FIGURE 47. Liquid chromatogram of #2 fuel oil. (Reprinted with permission from Jadamec. J. R..Sauer. W. A.. and Talmi. Y., Anal. Chern.. 49, 1316. 1977. Copyright 1977 American Chemical Societ.

MPF-2A spectrofluorimeter. and Figure 47 shows the chromatogram of a #2 fuel oil.Figure 48 shows the emission spectra captured in real time using the OMA detectionsystem.

There are obvious advantages of capturing complete emission spectra of eluents on-the-fly. An even greater advantage would be to capture complete emission and excitationspectra on-the-fly. Thus. one can conceive of a video fluorimeter as a detector for H PLC.Two such approaches have been described, one by Herschberger et al. 1o'and the other byShelly et al. 'o In the latter approach, a flow cell (Precision Cells, Inc.. P N 8830) with aflow cavity of square cross-section. and a 20 -j 1 sample volume is coupled at the end ofla

reverse-phase chromatographic column. Data are then captured by using the videofluorimeter" as an eluant detector (Figure 49). Figure 50 shows a conventionalabsorption chromatogram for six polynuclear aromatic compounds, and Figure 51shows the EE Ms (background subtracted) captured on-the-fly. The quality of the spectrais noted to be as good or better than that obtained in a conventional cuvette.

The examples of fluorescence detectors for HPLC that combine fluorescence with aseparation method show a very powerful approach to multicomponent analysis.Indicative of this realization is the increasing number of articles found on fluorimetricderivtization methods in HPLC.10 3

.10 4

IV. MULTICOMPONENT FLUORESCENCE DATA REDUCTION

Up to this point, we have discussed instrumental techniques for multicomponentfluorescence analysis. Some of these techniques are inherently designed for spectraldeconvolution such as th,: wavelength modulation technique of O'Haver and Parks."

202 CRC Critical Reviews in Ana/lt.ial Chemistr.

=

'.-

~k77

- .,.-. &, !

*3.. .-

al~~Ej ... e i ...

February 1982 203

A

--------------D E

FIGURE 49. Video t11.orimeter for chromatographic eluantdetection. (A) Altex microprocessor controlled H PLC. B) Column:15 x 250 mm preparative or 4.6 X 250 mm analytical (cationexchange). (C) Light source. (D) Excitation polychromator.(E) Microfluorescence flow cell. (F) Emission polychromator.(G) SIT detector (intensified vidicconO. (H) QMA-2 optical multi-channel analyzer (1) HP9945T calculator. (From Warner. 1. M..Fogarty. M.P.and Shelly. D.C.. Anal. C/urn. Acta. 109,361. (979.With permission.)

0 F~gure I

So-- 9.10.dimethyganthraeoe CaibaZote

450-

FIUE5.Sprtoi(aomtchdoabnnihU detctornhn

204 CRC Critical Reviews in Analtzical Chemistrt

.... .

,.,. , . . ..; " -, , '. -

'16

. . .

* ' I•• " . .. . .-. - s\'"-,2. * ...

t A

FIGURE 51 Three-dimensional fluorescent scans of aromatic hydrocarbons.

Others require algorithms to aid in spectral interpretation. This section will discuss datareduction algorithms common to most fluorescence techniques as well as novelalgorithms developed in recent years to complement the development of novelinstrumentation such as the video fluorimeter. " "'

A. General AlgorithmsOne of the most common algorithms used for fluorescence instrumentation is for

spectral correction. Two types of spectral corrections are required. one for distortions inthe excitation spectra due to the nonlinear output of the excitation source-opticscombination, and the other for distortions in the emission spectra due to the detector-optics ombination. Parker" has identified several approaches to spectral correction

February 1982 205

COMfECTON

LION! PaT

EXCITATION EEILSlON

OTTIECPOETC

MOOLE

----- 6-E ISO

AMP LFE REAM

EXCITATIOTN¢OSAMPLE

RA--,,EE TI O.

FIGURE 52. Block diagram of corrected excitation and emission system. (F~rom Porro, T. J. Anacreon,

R. E.. Flandreau. P S.. and Fagerson. I. S., I. Assoc. Off Anal. Chern.. 56. 608. 1972. With permission.)which we will not discuss in detail here since we are mainly concerned with theimplementation of algorithms to perform the spectral correction. A paper by Porro etal. 0. has identified and illustrated a number of principal uses of corrected spectral datausing several polynuclear compounds as examples. Basically, a general diagram for acorrected-spectra fluorimeter is shown in Figure 52. A comparison of the corrected anduncorrected excitation spectra of anthracene ls provided in Figure 53. Other examples ofspectral correction instruments are cited b Shephard,.'5 Tuan and Wild.' " Budde andIn addition to spectral correction. it is often necessar., to compensate for attenuation ofthe excitation beam due to strong absorpti,n b, the analvte. Recently. Holland et al.''0developed an algorithm for this task. The authors defined the absorption correctionfactor as

a =

(8)InT IW - t

SIT'a

lTER

where T is the transmittance and W1 and W: are defined as the fractional distances acrossthe cell at points X , and X2. respectively. Most methods which correct for absorption bythe analyte will generally use an equation of the form defined above. For detailedinformation on correction of inner-filter effects via right angle and front-surfaceillumination, a recent paper by Leese and Wehry''' can be consulted.l

The simplest approach to qualitative multicomponent fluorescence data reductioniiivolves a file searching system where the spectrum of an unknown compound iscompared to spectra from a library of known compounds. Several systems have beendescribed. More recently. Miller and Faulkner' 2 developed a computer-assisted

206 CRC Critical Reviews in .-1naiv~ucal Chemistriy

100-

s0- UNCORRECTED

CORRECTED EXCITATIONEXCITATIONz

W60-z

z

V) 40-

020

0-

240 320 400240 320 '#00

WAVELENGTH tim)

FIGURE 53. Anthracene in ethanol. corrected vs uncorrected excitation spectra (From Porro. TJAnacreon. R. E.. Flandreau. P S.. and Fhgerson. I S .I 4ssoi Off 4nal. Chtem.. 56. 608. 1972. Withperission.)

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

SOLVENT I.0. NUMBER

o 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15MAWeR OF NUMBER OF

EMISSION EXCITATION LOCATION OF EXCITATIONPEAKS PEAKSMNIU

0 1 2 3 4 5 6 7 8 9 10 1 12 13 14 15

RELATIVE INTENSITY ORWIDTH AT HALF HEIGHT LOCATION OF PEAK

FIGURE 54. Packing of spectral information into 6-bit computer words.(Reprinted with permission from Miller, T C and Faulkner. L R . 4nal. Chtemn48. 2083. 1976. Copyright 1976 Amertcan Chemical Society.)

structural interpretation program for fluorescence file seacching. Their scheme employeda series of 16 bit computer words to store spectral information (Figure 54). A basic flowchart for their file searching program is depicted in Figure 55. Table 9 provides the results

February 1982 207

TO A

READ IN1.8 CT INPU T

ACUR CTO ,

FILE SPO T / FiE

OBTAIN SPICRMI FL

EXCITATION APPLY CORRSPECTRUM FACTORS ANO)SHIFT

AIDETERMINE

TAIONLOC ANDISSI NREL INT

SPECTRUMOF PEAJKS

SA PLOSPECTRUM EI

PON

W SCTV ARIABES YES

FIGURE 55. Flow chart for the file searching program. LB = library; CT = cassette.

(Reprinted with permission from Miller. T. C. and Faulkner. L. R. Anal. Chiem.. 48.2083. 1976. Copyright 1976 American Chemical Society.I

of several searches on this system. The index column in this table refers to the degree ofcertainty in the spectral match. A value of 0" indicates a perfect match and less thanperfect matches can extend to values as high as 65.000. Another paper from Faulknersgroup'" outlined the advantages of data compression of fluorescence spectral files byFourier transformation.

An interesting algorithm has been developed by the research group of Buck''4 andapplied to the analysis of binary fluorescent mixtures. This algorithm assumes that thefluorescence spectra of the components can be described by a series of Gaussian orasymmetric Gaussian curves. One example of qualitative analysis of binary mixtures wasa study of single and binary component mixtures of anthracene, rubrene naphthalene.and naphthacene in cyclohexane. These systems were chosen because of extensivespectral overlaps. The spectrum shown in Figure 56 is typical of the data obtained in thisstudy. It was assumed that each binary mixture could be represented as a series of tenGaussian peaks. Table 10 lists the peak parameters obtained for the individual standardcompounds, and Table I l provides a listing and assignment of the seven most dominantpeaks determined by the algorithm for the mixture spectra. The agreement of the mixturedata with standard data demonstrates the feasibility of this approach.

Since the development of the video fluorimeter which we described earlier, a number ofalgorithms have evolved to aid in qualitative analysis of data. Although a completedescription is not practical in this paper, a brief synopsis is warranted. We previouslystated that the data acquired on the video fluorimeter for a multicomponent mixture can

........ TP

. ..... . ~ ~ iT ., . - L B II

208 CRC Critial Revtiws in 4nalhtwal Cheintirv

Table 9RESULTS FROM SEARCH

Positionof coTTiect

comupoundin list of

Compound Solvent' results Index

iAnthracene C 169. 10.Di bromoanthracene C 1 59-Mecthylanthracene C 1 17

9. 10-Diphenvianthracene C 1 49Diphenylamine C 16Diphentylamine M 1 I

'-Hydroxvcoumarin %I hi1-Hydroxycoumarin W 1 3Quinine Sulfate Sm-Methylanisoic C 23.5-Xvlenol 51 -h

3.S-Xvlenol 9 I9Salicylamide \43Ip-Toluidine 'A 3 41P-Cresol M 4 IS

C = cyclohexane: %A methanol. S =0 1 % H -SO,. W =water

A perfect match has an index 01 iero

Reprinted with permission from MAil~er. T C and Faulkner. L R . -malChem.. 48. 2083. 1976 Copyright 1976 American Chemical Societ%.

RESCALEO SPEC7Rum

CALCULATED SPECrRUm 8Ir-rAUSSIPN COP9QNENTS)

$00 - 66 -5 N 116 1

A B

FIGURE 56. Spectra of binary mixtures. (A) Ant hracene -naphthalene binary mixture spectrum.j

(B) Component peaks. Peaks with maxims at 364.5. 379.8. and 393.8 nm are attributed to noise. iFromRechsteiner. C. E.. God .SadBc.R .Ad hm ca 5 1977 With permission.

Februar, 1982 209

Table 10COMPONENT PEAK PARAMETERS FOR PURE REFERENCE

MATERIALS

Waveiength(nm) Intensity Peak parameter (I" Peak parameter(-

Anthracene 399 1 6.f) 8.82 ".99

85 Ae 422.1 2.80 10.05 10.014478 0.86 915 10.23384.3 027 4.21 4.21475.1 0.17 12.91 14.33

Naphthacene 473.3 9.10 ..9 8.38

(I ,Ag mq" ) 505.0 6.40 10.51 964541.0 1.1) 1780 13 59

341.9 1)93 6.49 10.143603 0.61 '69 36.32

Naphthalene 325.0 6.00 8.79 9.89

i 136ag mi") 342.4 3.20 8.54 11.81363.0 0.68 13.62 19.07319.6 0.26 2.28 2.28335.4 0.75 3.76 4.46

Rubrene 551.2 1.37 14.21 16.89(33 , g mi-:) 588.0 0.75 18.14 21.38

633.4 0.13 20.03 25.66472.3 0.03 6.64 7.43537.2 0.15 13 43 10.53

From Rechstemner, C. E.. Gold. H S.. and Buck. R P . .4nal. Chim. .4 cta. 95. 51. 197 ". With permission.

be represented byn

ci E ali) xli) vlil) 191

where n is the number of fluorescent species and i is an index of each. Implicit in thisequation is the assumption that the emission spectrum of each component is independentof the excitation wavelength and that the excitation spectrum of each component isindependent of the monitored emission wavelength. This assumption has been found tobe valid at low total absorbance (<0.01) where synergistic effects such as energy transferand inner filter effects are negligible. In terms of linear algebra. the analysis scheme issimplified by these assumptions since for a pure component the expected lineardependence of the rows and columns would indicate a rank I "matrix. That is. each row(column) is some linear multiple of another row (column). Additionally, the rank of amixture matrix is a lower bound to the number of fluorescent components contributingto that matrix. However, the presence of noise in the data matrix complicates theanalysis. Neglecting noise, it is reasonable to assume that a set of n vectors which form abasis for M are related to the n desired spectral vectors. This approach to qualitativeanalysis of the EEM has been used with success.'' '' '" Consider the positive definite squarematrices MM' and M' M. It is a simple algebraic procedure to show that the rows andcolumns of MMr are linear combinations of the columns of M and that the rows andcolumns of MrM are linear combinations of the rows of %M. Consequently, the neigenvectors of M MT form a basis for the columns of M. Similarly. the n eigenvectors ofM M form a basis for the rows of M. Thus, the set of n basis vectors for the columns androws of M are easily obtained. However, it is well known that most eigenanalysisprocedures will provide eigenvectors which form an orthogonal basis set. Thus, the

210 CRC Critical Reviews in Analtical Chemistry

Table I ICOMPONENT PEAK PARAMETERS FOR BINARY MIXTURES

ContributingWavelength compound

(nm) Intensity Parameter (+) Parameter (-) label'

Naphthalene 3994 5.33 .59 7.22 Aanthracene 421.6 2.42 10.00 9.52 A

384.3 1.84 7.02 6.42 A325.0 0.89 11.04 9.49 NI447.5 0.62 9.72 10.43 A345.6 0.21 4.43 5.98 NI475.3 0.09 10.68 10.68 A

Naphthalene 325.2 452 87,8 999 NInaphthacene 341.2 2.69 7 49 10 19 NI

473.0 1 83 " 62 94 Nc

504.9 1 27 949 9.71 Nc360.4 0.89 10.54 1466 NI541.3 0.33 18.02 13.04 Nc

334.7 0.35 3 26 3.26 NI

Naphthacene 473.3 2.31 7 60 7,80 Ncrubrene 545.7 1.96 15.77 16.53 R

505.1 1.59 981 9 111 Nc585.7 0.85 19.42 21.41 R

342.5 0.28 7 21 11 19 Nc360.9 0.18 7 65 12.65 Nc630.5 0.13 17 39 23 22 R

A = anthracene: NI = naphthalene; Nc = naphthacene: R = rubrene

From Rechsteiner. C. E.. Gold. H. S.. and Buck. R. P. .4nal. Chim. .4ca. 95. 51. 197" With permtission

eigenvectors for more than one component will usually contain negative eciements. Truespectral vectors are defined to be nonnegative. For more than one component. we thenneed to transform the eigenvectors to a nonnegative set of basis tspectral) vectors. Such a

transformation has been described.'' 6 We will not discuss this transformation here, buttwo general observations can be made.

First, the transformation has only been applied to binary mixtures. Second. thetransformation to a nonnegative basis set can be ambiguous depending on the degree ofspectral overlap. As an example, consider the data presented in Figure 57. The spectrumderived from a mixture of anthracene and perylene (Figure 57A) is unambiguous due to aregion of partial overlap between the emission and excitation spectra of the two com-pounds. Similar observations about the spectral overlap in Figure 57B and C predict thata mixture of anthracene and pyrene and a mixture of pyrene and fluoranthene would haveone ambiguous spectrum each. Applications of this algorithm to a mixture of twoporphyrin compounds also proved fruitful as shown by the deconvoluted spectraof Figure 58.

Another algorithm for qualitative analysis of the EEM has been developed by Fogartyand Warner. ' This algorithm is similar to one previously developed by Hirschfeld fordeconvolution of infrared data.' In application to fluorescence data, a majorassumption of this algorithm is that each component in the mixture has some spectralregion where :t is a lone emitter. Consider another form of Equation 9:

n

C., 4.(10)

X, 'A IN PECRUM3PERUFebruary 1982 211

*;x24~N ~(VUP CI53r3

QEI6u. SPECTRUM -EISO sp"u

.90, 801

H ~70f

40

~30 30

20 2

';-31367 3948 444 459 495 520 S46 2036416739241 44649SZS6

.ExciTriom SPECTRUM4 -fMISSION SPECTRUJM +EXCITATION SPECTRUM -EMI5tlON SPEC7RUMp

90 9

t70 70

so. 60

40, :740

30 30

20 20

10 10

Ki363137392 418 444 469 495 S2 56903631 367 392 418 444 469 495 S20 S46W9 3634 VELENGrTide B WAEEN .Mi

E(XCITATION SPECTRUM9 -EMISSION SPECTRMq +ErXCITAT[ON SPCTRUM -EMISSION SPECTRUMi

70r a0~

~301V 30201 / 20

i90 316 '341 367 392 416 444 469 495 520 346 90316 341 367 392 410 444 469 495 S20 S46.VELENO M.4. C IVELENGTM.N

FIG URE 57. Ambiguities of binary mixtures of aromatic hydrocarbons. (A) Anthracene (left) and perylene(right). (B) Anthracene (left) and pyrene (right). (C) Fluoranthene (left) and pyrene (right). (Reprinted withpermission from Warner. 1. M.. Davidson, E. R., and Christian, G. D.. Anal. Chep'. 49. 564. 1977.Copyright 1977 American Chemical Society.)

mo I ,'

-i ~ 4'~~ loom-

212 CRC Critical Reviewvs in A nalvtical Cheinistri-

'109

4s-4

54 591 621 644 668 691 '15 738A -ISSKN WPVELENG7HNM1

+EXCITATION SPECTRUM -EMISSION SPECTRUMI

100'

90'

80'

S70

Z60

so

4a

~3020

10

40 48 SOS 539 568 S98 628 657 687 716 746

B I4AVELENGTHM

FIGUiRE 58. Re'.oI~cd two-dimensionalj spcctra iA,Contiour plot ot datla ol experimuntall d..qulred 1~,-zomponent mixture o1 ZnOEI and H:OEP (8) Rc-Nolked two-dimensional spectra ol component I iZnOEP)iC) Resolved tvo-dimcnionai spetra 11 component(H:OEP). fReprnted with permitssion lrm 'Aarncr. I MDiaiij,on. IF R-. and Christian. G. D.. -inm (hern. 49564. 1977. Cop~right 197" American (hemnical Noeiti

where N, is an arbitrary standard matrix whose spectral characteristics dre not kno%%nand C, is the concentration of the ih component in M divided by its coniccntrationin N,. If we can obtain aset ofn matrices containing then components of Equatio-i 10. wewil have a set of n equations in n unknowns. In matrix form, this is expressec as

Y A N iii1* 11

where A is the set of C, coefficients. M *is the series of mixture matrice-,. and N

February 1982 213

+EXCITRTION SPECTRUM -EMSSIONi SPECTRUM

'OOr

-80!

, ", jill rl'Z 3C lI

4',) 4Afl ']C, 9 S'V ,F 598 628 657 60? '16 7?46C: P1Vf LENGTH .NI'

FIGURE 58(

represents the series of standard component matrices. Then. if A is invertible, thesolution to Equation I I is

N* = A M (12)

Fogarty and Warner have shown that estimates of C,, can be obtained in regions wherethe components are lone emitters by obtaining a ratio of one of the mixture matrices andthe other n - I matrices. The authors have also shown that quenching can be an aid in thisratio deconvolution. Consider the example shown in Figure 59A which corresponds to amixture of anthracene and 9,10-dimethylanthracene. Figure 59B corresponds to theaddition of the quencher methyliodide, while Figure 59C is the matrix obtained byratioing Figure 59A and B. The resulting deconvolution is shown in Figure 60.

Both of the algorithms which we have described for the analysis of data from a videofluorimeter are limited to mixtures of only a few components. The eigenanalysis methodhas only been applied to binary mixtures while the ratio method has been applied totertiary mixtures. The ratio method has the advantage of requiring less computer timefor convergence.

C. Quantitative Analysis of Fluorescence DataThe determination of the concentration of known components in a fluorescent mixture

is a problem routinely confronting the analyst. These routine determinations will moreoften than not involve the use of standard solutions of the known compounds. Such is theapproach used for quantitative evaluation of the emission-excitation matrix. Recall thatwe stated that Equation 9 could be rewritten as Equation 10 or. in another notation.we can write

nm c +(m ) . 013)

i=I

where C, and (M), are the previously defined C,, and N,. respectively. The best leastsquares approximation to M would require a minimization of the weighted square of theelements of the error matrix E. i.e..

L ! .. . .. .. ..

214 CRC Critical Reviewvs in Analytical ChemistrY

4 04

IS 2026

354 11

3 1 i)3

32 39934

*-". 5?VISO3;,'zZ~~ZS

t 0 13-

413g

AL

<'41

FIGURE 59. (A) Three-dimenbional 5pectrum 0, .anthracene mixed with 9. 0-dimethylanhhracent. (B) Quenched with methyl iodide. I C Ratio of A aind 0. (Reprintedwith permission from Fogarty. Ms. P. and Warner. 1. %M. Ana. Chem.. 53. 259. 1981.Copyright 1981 American Chemical Society.)

ST%, Ek, F . [M, -~C.(MAl.' = min (14)

where Tk represents weighting factors which we shall assume are unity for simplicity.Derivation of the least squares equation gives

B ( i(M (M.,]1C, -= Ml) k (I5)j i kI

February 1982 215

114as

2835?

1073,.

W"

36a 4 ~ ~ * 4

0z- 7.19 -

~ IS

'41 Dass .5

413 ,%-A 1

04'32 1 441

~ZB5357

FIGUJRE 60. Deconvoluted spectra ol Figure 59C (A) Ratio deconvoluted spectrum ofanthracene. (B) Ratio deconvoluted spectrum 0f 9.lO-dimethli anthracene. IC) Eigen-analysis dcconvolution 01 anthracene. (D) Eagenanalysi! decons olution o1 9. lO-dimeth~ Ianthracene. tReprinted with permission from Fogart'. M P And W~arner, 1. M.. .4na.Chem.. 53. 259. 1981. Copyright 1981 American Chemical Socict%.

216 CRC Critical Reviewvs in Anal irical ChemistrY

la

Fla

1 40

F-IGURE 61. Quantitative analysis ot EEM. (a) Rank one H:OEPstandard EEM. (b) Rank one ZnOEP standard EEM. (c) Rankone SnC2 QEP standard EEM. (d) Mixture with scatteredlight removed (data collected at 4-nm increments). (Reprintedwith permission from Warner, 1. M.. Davidson. E. R.. andChristian. G. D.. Anal Chem., 49. 2155. 1977. Copyright 1977American Chemical Society.

where Mw1 is an element of M and (Mi), and (Mki), are elements of i"' and j'f standardmatrices. An alternative approach uses linear programming to maximize the function

n

A, C,(16)

February 1982 217

FI1L RE MDE)

Table 12

LEAST SQUARES AND LINEAR PROGRAMMING ANALYSIS OFEXPERIMENTALLY ACQUIRED TERNARY MIXTURE OF

H 2OEP. SnCi 2OEP, AND ZnOEP

True Least squares Simple%concentration concentration Estimated concentration

Component 10 1 !W0" 1-m std dev tO'W

HzOEP 2.42 2.46 0.02 2.38ZnOEP 21.4 21.6 0.5 21.0SnCI:OEP 3.04 - 3.44 0.23 3.22

Calculated using the spectrum of a 4.83 x 10-. M H ,OEP solution, a 2.14 X 10"' W ZnOEPsolution. and a 6.07 x 10" .14 SnCl:OEP solution as standards.

Reprinted with permission from Warner, I. M., Davidson. E. R.. and Christian, G. D 4na.-"Chem.. 49, 2155, 1977. Copyright 1977 American Chemical Society.

subject to the constraints

Ek, Mt-, C.(M"). a 0 (17)

where

A, (M( )kl

Both of these methods have been applied to quantitative analysis of the EEM withsuccess.'" As an example, Figure 61 shows typical data in this study, while Table 12displays the input and computed results of the analyses. This study demonstrated that theleast squares method was satisfactory when all of the components are known. However.the linear programming method is preferred over the least squares method when all of thecomponents are not specified.An alternative to the least squares and linear programming method has been proposed

and applied by Ho et al. 120. " This method is termed the"method of rank annihilation"and is more useful than either the least squares or linear programming method when all ofthe components are not specified. Briefly, one can approximate M by a matrix M' formedfrom the eigenvalues fk and eigenvectors Uk and Vk. I.e.,

S

' 1 f , U Vr (19)k-I

- . . _ _ _ _ . . ...

218 CRC Critical Rei ;e%% s in .4 na/u tical Chemistri

Table 13RESULTS OF LEAST SQUARES AND RANK ANNIHILATION FOR A

ONE-COMPONENT PERYLENE SOLUTION

Reltive concenthations'

20.0 10.0 2.0014 0.200 0.100 0.020

Least squares 1999 1007 2.23 ).250 0,114 0.022Annihilation 20.03 1009 2.23 0249 0.11IS 0023annihilation 20.07 10 10 2.23 o_150 4).14 0.022

b, 0.9978 0.9969 09987 0.9998 0.9941 09500a0999' o.9984 0 9998 0 .9999 41 Y ' 0 9852

The concentrations are reltiv t:o the ; stnar6 0 10 w 1per2e soio. For thistndr

Reprinted with permission from Ho. C N . Christian. G. D,. and Davidson. E. R . 4nal. Chem.. 50.110l8l. 197S,.Coprigt 178 mercanChemical Society.

where s :5 n. If the test compound is present, then the vectors x and v' (Equation 9) and ukand vkr should satisfy the conditions

n

aE 4 U = I =1(20)

andn

Fbt = (v~yvV = I (2t)

Bessel's inequality requires that

~ a~ ~ I22)

and thus the nearness of

(Ela)k b) 1231

to unity is the estimate of whether the test compound is present. If Equations 20 and 21

are satisfied, then the matrix

E' = M - O3N (24)

will have rank n - I when evaluated at /3=C C:'where C. is the concentration of the testcompound in the mixture and C, is the concentration of its standard. Simply stated, therank annihilation method measures the value of 0 at which the test component is exactlyeliminated from the mixture matrix. Table 13 shows the results of the least squares and

i I rank annihilation methods for a one-component system. ''o Table 14 shows the results ofrank annihilation for perylene in this mixture while Figure 62 is an eigenvalue plot of thesame data.'-" It is apparent from these results that the rank annihilation method showsthe greatest promise for quantitative analysis of the EEM, when the identities ol allcomponents are not specified.

Februar% 1982 219

Table 14RESULTS OF THE METHOD OF RANK

ANNIHILATION USING THE EXTENDED RANKPR OCED URE

Relative perilene concentrations'

r Eigenvalue* 20.00 10.00 2.00 1.00 0.20 0.10

2 20.04 008 2.11 1 0 85 0 -S3 2 20.04 10,08 2.05 1.29 0 9 t) 1

3 14.66 9.18 2.11 099 0.34 0264 2 20.04 1008 2.05 1.28 0 5 068

3 14.66 9 19 2.10 096 o.20 0 106 3 0.20 0 0

Solutions and standards are the Name as in Table 13*Dimension ol e matrix betore extension.Index ot eigensalue minimiiced.

Reprinted with permission from Ho. C. N .Christian. G. D . andDavidson. E. R.. Anal. Chem.. 52. 1071. 1980 Copyright 1980 AmericanChemical SocietN.

2

5

2

Z2

2

.1 108 0.7

RELATIVE CONCENTRATIONS

F-IU RE t62. Plot ot cigenvalues %s relative concentrationPlot ot the cigensalues u with R = 6 against the reiatiseconcentration (0) at the standard subtracted lor anthracene inSample I The relatisec concentration plotted must be multipliedb,. 2.5 (gain tactort to obtain the actual relatise concentration.

1 Reprinted with permission trom Ho. C. % . Christian. Gi. D . andDastdson. E. R.. Anal. ('hem.. 52. 1071. 19K0 Copyright 1980American Chemical Socitt~

220 CRC Critial Revieis in A nalincaI Che'mis.iri

V. SUMMARY AND CONCLUSIONS

In this manuscript, we ha,,e presented se~eral examples of the recent advances inmulticomponent fluorescence anal~sis. Our discussion has concentrated mainly onde~elopments in instrumental methods. althoug we ha~e also pro~ided a brief're~iew of'the increasingly important area of multicomponent data reduction. No single criticalreview article can reasonably co'.er all possible aspects of a gi~en topic. and for thosetopics not included, we have referred the reader to other re% ie% articles. One such topicwhich we feel is noteworthy is matrix isolation ,pectroscopv. In this technique, theconstituents of a multicomponent mixture are anal,.ied at low temperatures afterisolation through ' aporizavion and large dilution bN an inert gas. The spectra obtainedare comparable in sharpness to those obtained bN the low temperature Shpol'skiimethod.~ A number of" applications of the use of this technique in fluorescencespectroscopy are cited by Wehry and co-workers.' A general re, ew on matrixisolation spectroscopy has been given by Wehry, and Mamanto%.

Another important area which we should mention here is the application of lasers tomulticomponent fluorescence analN sis. Several ad antages accrue from the use of thesehigh intensity, narrow line excitation sources, including increased s.ensitioitv. andselectivity. A recent review was written by Wright and Wirth on the use of lasers formulticomponent fluorescence analysis." Applications in fluorescence line narrowingspectrometry are becoming increasingly important.'-'

Finally, we would like to emphasize that multicomponent fluorescence analysis is avery new field experiencing rapid growth in recent years. New instrumental techniquesand data reduction strategies are being developed at a fast pace. and we look forward totheir increased application in chemical analysis.

ACKNOWLEDGMENT

1. M. Warner is grateful for support from the Office of Naxal Research and theDepartment of Energy (DE-AS05-8OE'v 10404) during the preparation of thismanuscript.

REFERENCES

1. O'Donnell, C. M. and Solil. T. N.. mitt. Chem.. 50. I?(9R 1 1978)2. O'Donnell. C. M. and Solil. T. N.. -foal Chem-, .1K. 1 '5R 1 1976i3. Aaron. J. J. and Winefordner. .J. D.. Talantia 22. 707 1191514 Karasek. F. W.. Res/ Dev. 27. 28 1i19'76).5. Bridges. J. W.. Wethudol. Dev. Bunthem.. 5. 25 (1166 Gibson. E. P.. J Forensic Sci.. 22. 403 (1i9-7)7. Guiibauit. G. G.. Phototchem. Photobtoi.. 25. 403 I 1917)8. Klainer. S. M.. Opt. Eng.. 15. SR14. 1197619 Lloyd. J. B. F., Chern. Br.. 11. 442 11975).

$10. Haddad. P. R.. Ta/anta. 24. 1 I 9771I I Krause. L.. 4dv Chern. Phi-s.. 28. 267, 1975111. Richardson. F. S. and Riehi. J. P.. Chern. Rev. -7. 773 11f971)13. Douilier. G. D.. Bradshaw. J. D.. Weeks. S. J., and Winefordner. J. D.. 4pp/. Spettrosc. 11.

307 (1977).14. Cooney. R. P.. Detector and Other Insirumental Effects on Signal-To-Notse Ratio in Gas and Liquid

Phase Fluorescence. Dissertation. University of Florida. Gainesville. 197"I5. Limt. C. S. and Miller. J. N., A4na. Chim. ,4cra. 100. 235 1978)16. While. J. U., .4na. Chem.. 48. 21089 119761.17 Anacreon. R. E. and Ohnishi. Y.. App. Opt.. 14. 2921 11975118. Porro. T. J. and Teehaar. D. A., 4na. C/hern.. 48. 1 103A 119761

Februar\ 19X2 221

19 Jameson. D. M.. Spencer. R. D., and Weber, G.. Rev Sc'. In.sirum.. 47. 1034 11 9"h.:0. Glushko. V., Caley. R.. and Karp, C.. 4nal. Chem.. 48. 20- 119'0121. Wehrly, J. A., Williams. J. F.. Jameson. D. 1.. and Koib. D. A., Anal. Chem. 48. 1424 19Th,22. Bowen. E. J.. Luminescence in Chemisirv. Van Nostrand. London. 196823. Parker. C. A., Photoluminescen'e of Solutions. Elsevier. New York. 1968.24. Wehry, E. L.. Alodern fluresrente Specirucopi. ' ols I and 2. Plenum Pres . \c" 'ork. 19-h25. Guilbault, G. G.. Practical Fluorescence. Marcel Dekker. New York. 197326. Lewis, G. N. and Kasba. M.. J. Am. Chem. So(.. 66. 2100 (19441.27. St. John, P. A. and Winefordner, J. D.. Anal. Chem.. 39. 500 I 19h').28. Freed. S. and Salmre. W.. Science. 128. 1341 1958).29 Freed. S. and Vise, M. H.. Anal. Biochem.. 5. 338 11963130. Winefordner. J. D. and Latz. H. W.. Anal. Chem.. 35. 151- 11963).31. Hollifield. H. C. and Winefordner, J. D.. Chem. Instrurm.. 1. 341 119h9l32. Keirs. R. J.. Britt. R. D.. and Wentworth. W. E.. 4nal. Chem . 29. 202 195')33. O'Haver. T. C. and Winefordner. J. D., anal. Chem. 3X, t)02 196,t34. O'Haver. T. C. and Winefordner, J. D.. 4nal. Chem., 31X. 125h 19i35. Winefordner. J. D.. -cc. Chem. Res.. 2. 361 (1969).36. Fisher. R. P. and Winefordner. J. D.. Anal. Chem , 44. 94h1 i19'2)37 O'Donnell. C. M.. Harbaugh. K. F.. Fisher, R. P., and ,,inefordner. J. D.. -naL Chem 45

609 (1973).38 Harbaugh, K. F.. O'Donnell. C... and Winefordner, J. D., 4nal. Chem. 46. 120h i 19.439. Badea. M. G. and Georghiou, S,, Rev. Sci. Instrum.. 47. 314 11976).40. Steinograber. 0. J. and Berlman, I. B.. Rev. Sci. Instrum.. 34. 524 11963)41 Hazan, G.. Grinvald, A.. Maytal. M.. and Steinberg, I. Z.. Re%. St In. trum.. 45. 161)2 19"4i42. Thomaz. M. F.. Barrades. I., and Ferreira, J. A.. J Luminesc. II. 55 (1975.43. Yguerabide, J., .Methods in En:r'nolog'. Vol 26. Part C. Colowick, S. P and Kaplan. N D . Fd .

Academic Press. New York. 191Z.44. Weissler, W., Anal. Chem.. 46. 500R (1974).

45. Cundill, R. B. and Palmer. T. F., Annual Reports on the Progress of Chemivirt. Vol. '(. Sec ..The Chemical Society. London. 1973. p. 31

46. Loper, G. L. and Lee, E. K. C.. Chem. Phys. Left.. 13. 140 f 1972)47 Yoshihara, K.. Kasuym. T., Inove. A., and Nagakura, S., Chem. Phii Lett. 9. 4(,9 (191)48. Deinum. T., Werkhoven. C. J.. Langelaar. J.. Pettschnick. R. P.. and van Voosot. J. D. %.., Chwm

Phis. lei,.. 12. 189 11971).49. Shaver, L. A. and Love. L. J. C.. 4ppl. Spectrosc.. 29. 4185 (1975)50. Love. L. J. C.. Upton. L. M.. and Ritter. A. W., 4nal. Chem . 50, 2059 (19"))51. Love. L. J. C. and Shaver, L. A., Anal Chem.. 48. 364A (1976).

52. Aaron, J. J., Sanders, L. B.. and Winefordner, J. D.. Clin. Chtim. Ata. 45. 3"5 I19'3153 Shaver, L. A., Critical Evaluation of Time-Resolved Fluorescence for the .- nal.si, ot Multicomponcnt

Fluorescent Mixtures. Dissertation, Scion Hall Umversitv. Scuth Orange. N J . 19"Th54 Busch, G. E. and Rentzepeis, P. M.. Science. 194. 276 (1976).55 Flemming, G. R.. Morris, J. M., and Robinson, G. W.. Alust. J Chem. 30. 2337 1197-1.56. Porter, G.. Sadkowski. P. J.. and Tredruell. C. J.. Chemn. Phv l.eft. 49. 416 119,157 Spencer, R. D. and Weber, G.. 4nn. V. Y 4(ad. Sci.. 156, 361 119691,58. Weber. G.. J. Chem. Phys.. 66. 41)81 t 197").59. Schurer. K., Ploegaert. P. G. F., and Wennekes. P. C. M.J. Phi E. 9. X21 I 197660. Iltoh. K. B., Gifu Daigaku Igakuhu Kivo. 25. 97 (197')61. Pohoski. R. and Zachara. S.. Zes.. Vauk. Wid:. .lat.. Fi:. Chau.. L niv Gdanski. Ft:. 1. lIo 11972)62. Reseuitz. E. P. and Lippert. E- Ber. Bunsenges Phis. Chem.. ",. 1227 (19,4,63. Hauser, M. and Heidt, G.. Rev. Sct. Instrum.. 46. 470 (19751.64 Mousa. J. J. and Winefordner. J. D.. Anal. Chem.. 46. 1195 11974).65 O'Haver. T. C. and Parks. W. W.. Anal. Chem.. 46. 1886 11974166. OHaver. T. C., Green. G. L.. and Keppler. B. R.. Chem. Instrum.. 4. 197 (1971)67. Green. G. L. and OHaver. T. C., Anal. Chem.. 46. 2191 (1974)68 Brownrigg. J. T., Hornig, A. W., and Coleman. H. J., Paper 39h presented at the 28th Pittsburgh

Conference on Analytical Chemistry and Applied Spectroscop.. March 1969. Giefrng. L. P., Ind. Res.. 20. 134 119781.70. Parker. C. A.. Photolummnescence of Solutions. Elsevier. New York. 1968. p 217L. Guilbaut. G. G., Practical Fluores,enie: Theory tfethods and Techniques. Marcel Dekker.

New York, 1973.72. Hauten , G. H.. Raby, G. A., and Rigdon, L. P.. Chem In. t. 6. 205 11975)

'3 Bentz. A. P., Anal. Chem., 48. 455A 11976).

222 CRC Crittcal Revtews in Anahvtcal Chemtstrl

"4. Adlard. E. R., J Inst. P~trol. 58. 63 1192).

-5 Gaugh. T. H.. Science. 203. 1330 1 1979)."6 Johnson. D. %.. Gladden. J. A.. Callis. J. B., and Christian. G. D.. Rev. Sci. Instrum.. 49. 54 119-8f-- %arner. I. M.. Callis. J. B.. Davidson. E. R.. and Christian. G. D.. Chn. Chem.. 22. 1483 11976).-4 Johnson. D. W.. Callis, J. B.. and Christian. G. D.. Anal. Chem.. 49. '47A 1!977).-9 L FS--00 Flat Field Spectrograph. Product of Instruments SA. Inc.. 08840, Metuchen. N J.SO Talmi. Y.. Anal. Chen.. 47. 697A (1975).S1 Talmi. Y.. 4nal. Chem.. 47. 685A 11975).S2 Talui. Y., Baker. D. C., Jadamec. J. R.. and Saver. W. A.. Anal. Chem.. 50. 936A (1978).

3 Warner. 1. M., Fogarty. M. P.. and Shelly, D. C.. Anal. Chim. Acta. 109. 361 t1979)14 Lloyd. J. B. F., Nat. Phys. Sct.. 231. 64 11971).85 Lloyd, J. B. F.. J Aorensics Sc. Soc.. 11. 83 (1971).86 Lloyd. J. B. F., 4nal'st. 100. 82 t1975): 99. 729 (1974).8' John, P. and Soutar. I.. Anal. Chem.. 48. 520 (1976).,8 Lloyd. J. B. F. and Evett, I. W.. Anal. Chem.. 49. 1710 11977).

89 Vo-Dinh. T.. Anal. Chem.. 50, 396 (1978).90, Vo-Dinh. T. and Gammage, R. B.. Anal. Chem.. 50. 2054 11978)91 Latz. H. W. Ullman, A. H.. Winefordner. J. D.. Anal. Chem.. 50. 2148 11978).92. Anacreon. R. E. and Ohnishi. Y.. Appl. Opt.. 14. 2921 (1975)93 Porro. T. J. and Terhuar, D. A.. Anal. Chem.. 48. 1103A (1976).94 Wilson, R. L. and Ingle, J. D., Anal. Chem.. 49. 1060 (1977).95 Wilson, R. L. and Ingle. J. D., Anal. Chem.. 49. 1066 (1977).96. Fogarly. M. P. and Warner. I. M., Appl. Spectrosc.. 34, 438 (1980).97 Johnson. E.. Abu-Shumays. A.. and Abbott. S. R., J. Chromatolr. 134. 107 (1977)98. Slavin0 W.. Rhip Williams, A. T., and Adams, R. F., J. Chromatogr., 134. 121 (1977)99 Martin. F.. Maine. J., Sweiley, C.. and Holland, J. F., Clin. Chem.. 22. 1434 11976)

100. Jadamec. J. R., Sauer. W. A., and Talmi. Y.. Anal. Chem.. 49. 1316 (1977).101 Hershberger. L., Cailis. J., and Christian. G.. Anal. Chem.. 51. 1444 (1979).102. Shelly, D. C.. tIger. W. A., Fogarty, M. P.. and Warner 1. M.. Alltex Chromatogram. 311).

November 1979103. Lawrence, J. F.. J. Chrom. Sc,., 17. 147 (1979).104. Dunges. W., Meyer, A.. Muller. K., Muller. M., Pietschmann. R.. Plachetta. C.. Sehr, R.. and

Tus, H.. Z. Anal, Chem.. 288. 361 11977)105. Porro, T. J.. Anacreon, R. E.. Flandreau. P. S.. and Fagerson. i. S.. J. Assoc. Off Anal. Chem..

56. 608 (1972).106. Shepherd. T. M.. Chem. Ind., No. 7. 332 (1973).107. Tuan, V. D. and Wild. U. P.. Appl. Opt.. 13. 2899 11974).108. Budde. W. and Dodd, C. X.. Appl. Opt., 12. 2607 11971).109. Pailthorpe, M. T.. J. Phvs. E: Sci. Instrum.. 8. 1975).110. Holland. J. F., Telts, R. E., Kelly, P. M.. and Timnick. A., Anal. Chem.. 49. 706 (1977).Ill. Leese. R. A. and Wehry. E. L., Anal. Chem.. 50. 1193 )1978).112. Miller, T. C. and Faulkner. L. R.. Anal. Chem.. 48. 2083 (1976).113. Yim, K. W., Miller, T. C.. and Faulkner. L. R., Anal. Chem.. 49. 209 11977).

114. Rechsteiner. C. E.. Gold, H. S., and Buck. R. P.. Anal. Chm. Acta. 95. 51 (1977).115. Weber. G., Vature (London). 190. 27 (1961).116. Warner. I. M., Christian. G. D.. Davidson. E. R., and Callis. J. B., Anal. Chem.. 49, 564 11977).117. Fogarty, M. P. and Warner. I. M., Anal. Chem.. 53. 259 (1981).19. Hirschfeld. T., Anal. Chem.. 48. 721 (1976).

119 Warner. I. M., Davidson. E. Rt., and Christian, G. D., Anal. Chem.. 49. 2155 (19771.120 Ho. C. N.. Christian, G. D.. and Davidson. E. R., Anal. Chem.. 50. 1108 11978).121 Ho, C. N.. Christian, G. D.. and Davidson, E. R., Anal Chem.. 52. 1071 (19801.122. Kirkbright. G. F. and De Lima. C. G., Analyst. 99. 338 (1974).123 Stroupe, R. C., Tokousbaiides, P., Kickinson. R. B.. Wehry, E. L.. and Mamantov. G.. Anal. Chem..

49. 701 (1977).124. Tokousballdes. P., Hinton, E. R.. Dickinson. R. B.. Belolta, P. V.. Wehry. E. L.. and Mamantov. G..

Anal. Chem.. 50. 1189 (1978),125 Dickinson. R. B. and Wehry. E. L.. Anal. Chem.. 51. 780 (1979)126. Wehry, E. L. and Mamantov, G.. 4nal Chem., 51. 643A 11979).127. Wright. J. C. and Wirth. M. J.. Anal. Chem.. 52. 989A (1980).128. Personov. R. I. and Kharlamov. B. M.. Opt. Commun.. 7. 417 (1973).129 Eberly, J. H., McGolgin. W. C.. Kawooka, K.. and Marchetli. A. P.. Nature, 251. 215 11974).130. Brown. J. C.. Edelson. C., and Small. G. J., Anal. Chem.. 50. 1394 (1978).

me