771 ‘33
,.

_v-mx@xt‘u'*!‘t'

My“

.l'l‘lik

i‘x

MM“?

emu???”-

”av‘uimg‘ml

an»
m;.;........
mam-ac.

a

W.
........4......M..
”w . .A~-..I.Iv

.a—W

 

 

.,'....r.wu.... y...,....w..-4.ym,..~..,.1

.y..w.....nr,.w.

.z;::...::

 

 

._. 7 .77 7.7-. ”7. . -..¢ 777,71.

 

AWN.“

"n: n‘u‘

\Ilu

- 2.," H

 

5.22.21-

-......7..;4 ,.¢,,
....,. ,.,

 

‘- mats

 

 

 

 

 

 

 

 

 

 

 

 

iiiiiiiii‘iii

This is to certify that the

dissertation entitled

IMPROVING THE CONFIDENCE OF
MOLECULAR FLUORESCENCE MEASUREMENTS

presented by

LAWRENCE EUGENE BOWMAN

has been accepted towards fulfillment
of the requirements for

Ph.D. Chemistry

degree in

SK. 6.7% 4714.

Major professor

 

 

 

Date April 10, 1991

MSU is an Affirmative Action/Equal Opportunity Institution 0-12771

I! _ LiiEtRAFW
,Michigan State
I University

fl...—

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.

DATE DUE DATE DUE DATE DUE

 

 

 

FM . “V
4"" $3 ,5

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MSU Is An Affirmative Action/Equal Opportunity Institution
czmepmJ-pil

IIVIPROVING THE CONFIDENCE OF MOLECULAR FLUORESCENCE
MEASUREMENTS

By

Lawrence Eugene Bowman

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Chemistry

1991

 

6544 58.78

 

ABSTRACT

IMPROVING THE CONFIDENCE OF MOLECULAR FLUORESCENCE
MEASUREMENTS

By

Lawrence Eugene Bowman

Two molecular fluorescence spectrometers have been developed that correct
molecular fluorescence measurements for processes that reduce the steady-state
fluorescence intensity. One of the constructed spectrometers corrects for one of the
most common sources of reduced fluorescence intensity, collisional quenching. This
spectrometer determines the ratio of the observed intensity to the excited-state lifetime
in real time. The spectrometer constructed in this work uses a frequency domain

technique to determine the excited-state lifetime in real time. It is somewhat limited in

the range of ekcited-state lifetimes that can be determine.

In addition to the frequency domain instrument, a sin gle-photon timing (SPT)
apparatus was constructed and used to make independent measurements of excited-state
lifetimes. The constructed SPT instrument is very versatile. It can record fluorescence
decays as short as 25 ps or as long as a few hundred ns. From excited-state lifetimes
determined with SPT, accurate correction for collisional quenching is demonstrated in a

chemical system where the intensity is reduced by a factor of 30 relative to the intensity

in the absence of quenching.

 

In the second, corrected spectrometer, correction for sample absorption,
accomplished using a previously reported cell-shift method, is included in addition to
correction for collisional quenching. When the fluorescence intensity corrected for
sample absorption is ratioed to the excited-state lifetime, one obtains a quantity that is
proportional to the concentration of the fluorophore that is corrected for the reduction
caused by both collisional quenching and sample absorption. The constructed
spectrometer that uses this combined correction approach has been termed a highly
corrected molecular fluorescence spectrometer. Measurements based on this dual-
correction scheme have demonstrated near immunity to collisional quenching and
sample absorption over a limited range of quencher concentration and sample

absorption.

 

Copyfightby
Lawrence E. Bowman

1991

 

 

TABLE OF CONTENTS
List of Tables .............................................................................. x
List of Figures ............................................................................ xii
Chapter 1: Introduction to Corrected Fluorescence Measurements ............... 1
Review. ............................................................................ 3
Corection for Quenching ............................................... 3
Chemically-Based Approaches ............................... 3
Instrumentation-Based Approaches .......................... 5
Correction for Sample Absorption .................................... 9
References ........................................................................ 10

Chapter 2: Design Considerations for an Instrumentation-Based
Approach to Real—Time Correction for Errors Due to

Quenching .................................................................. 12
Criteria for the Determination of I and r .................................... 13
Review of Methods for the Measurement of I and r ....................... 17

Time Domain Methods ................................................ 17

Frequency Domain Methods..............‘ ............................ 22

Conclusions ...................................................................... 24
References ........................................................................ 27
Chapter 3: Single-Photon Timing .................................................... 29
Theory ............................................................................ 30

The Instrument Response ...................................................... 32
Instrumentation .................................................................. 33

 

 

 

Light Sources ........................................................... 33
Sample Cells ............................................................ 36
Wavelength Selection .................................................. 38
Detectors ................................................................ 41
Electronics ............................................................... 44
Cables and Connectors ........................................ 45

Amplifiers ...................................................... 46

The Constant Fraction Discriminator ....................... 47

The Time-to-Amplitude Converter .......................... 49

The Multichannel Analyzer .................................. 51

Delays and Relative Timing .......................................... 52

A Typical SPT Experiment .................................................... 55
Data Analysis .................................................................... 61
Application to Correction for Quenching .................................... 67
Experimental Details ................................................... 67
Apparatus ....................................................... 67

Reagents ........................................................ 68

Procedures ...................................................... 68

Results and Discussion ................................................ 70
References ........................................................................ 78
Chapter 4: Frequency Domain Fluorescence ....................................... 81
Thedry ............................................................................ 81
Review ............................................................................ 86
Early Work .............................................................. 89
Modern FDF ............................................................ 89
Survey of Previous Instruments ...................................... 92

vi

 

 

 

 

A Molecui
for the Efi

Ge
De

Da
Ot
In:
Pe
Applicatic

 

A Molecular Fluorescence Spectrometer that Corrects

for the Effects of Collisional Quenching .................................... 96
General Features ....................................................... 97
Detection Electronics .................................................. 99

Black Box and Mixer ......................................... 99
NIM Electronics ............................................. 100
Bandpass Filter .............................................. 104
Peak Detector ................................................ 1105
ZCD and TAG ............................................... 106
Sample—and—hold ............................................. 106
Data Acquisition ...................................................... 107
Other Components ................................................... 107
Instrument Calibration ............................................... 108
Performance ........................................................... 109

Application to Real-Time Correction for Errors due to Quenching. 112

Experimental Details ................................................. 113

Apparatus ..................................................... 1 13

Reagents ...................................................... 1 13

Procedures .................................................... 1 13

Results and Discussion .............................................. 114

References ...................................................................... 118

Chapter 5: A Highly Corrected Moleuclar Fluorescence Spectrometer ....... 121

Review .......................................................................... 122
A Fluorescence Spectrometer that Corrects

for Quenching and Absorption ............................................. 123

Fluorescence Response Surfaces ................................... 124

Experimental Details ........................................ 124

Apparatus ............................................ 124

vii

 

 

 

 

Reference
Chapter 6: Sugg
Single-Ph
Frequent}
Highly CI
Appendix l: Co:
Light Sou
Optics
Electronic
Other C01
Appendix 2: An
compote

0;

De

Constnm
Controls:

Fr

 

Reagents ............................................. 125

Procedures ........................................... 125

Results and Discussion ...................................... 126

Highly Corrected Fluorescence .................................... 128

Experimental Details ........................................ 128

Apparatus ............................................ 128

Reagents ............................................. 129

Procedures ........................................... 129

Results and Discussion ...................................... 130

References ...................................................................... 135
Chapter 6: Suggestions for Improvements and Future Work ................... 136
Single-Photon Timing ........................................................ 136
Frequency Domain Fluoresecence .......................................... 140
Highly Corrected Fluorescence ............................................. 142
Appendix 1: Components fo the Single-Photon Timing Apparatus ............ 144
Light Source and Monitoring Equipment .................................. 141

Optics ........................................................................... 145
Electronics ..................................................................... 146

Other Components ............................................................ 147
Appendix 2: An Electronically Switched Fiber Optic Delay ................... 149
Components ............................................................ ' ........ 151
Optical Fibers, Connectors, and Switches ........................ 151

Detection .............................................................. 154

Construction ................................................................... 156
Controls and Specifications .................................................. 159

Front and Rear Panel Connections and Controls ................ 159

Front Panel ................................................... 159

viii

 

 

 

Vi

Reference
Appendix 3: Al
General C

 

Rear Panel .................................................... 160

Vendors, Part Numbers, and Miscellaneous Information ...... 161

References ...................................................................... 164
Appendix 3: A User's Manual for Single-Photon Timing ...................... 165
General Considerations ....................................................... 166

The Laser System .................................................... 166

Cables and Connectors .............................................. 168

Detectors .............................................................. 168

The Temporal Reference Channel .......................................... 169

The Fluorescence Channel ................................................... 170

The TAC and MCA .......................................................... 172

More on the Light Source .................................................... 173

Data Acquisition .............................................................. 174
Complex Decays ..................................................... 176

Time Calibration ..................................................... 177

Files and Naming Concentions .............................................. 177

A Reference F luorophore .................................................... 178

A Flow Diagram for SPT .................................................... 179
Appendix 4: A Simple Frequency Synthesizer ................................... 182

 

 

 

 

 

 

 

 

TABLE 3-1.

TABLE 3-2.

TABLE 3.3,

TABLE 34.

TABLE 4-1.

TABLE 4.2.

TABLE 4.3.

 

TABLE 3-1.

TABLE 3-2.

TABLE 3-3.

TABLE 3-4.

TABLE 4-1.

TABLE 4-2.

TABLE 4-3,

 

LIST OF TABLES

TRANSIT TIME SPREAD (IT S) FROM 300 TO
700 NM IN 100 NM INCREMENT S FOR A SPEX
1681 CZERNY-TURNER MON OCHROMATOR ............... 39

FLUORESCENCE INTENSITIES, LIFETIMES
AND THE RATIO OF I/ 1' FOR NAPHTHALENE
IN CYCLOHEXANE SUCESSIVELY QUENCHED

BY OXYGEN ........................................................... 7O

FLUORESCENCE INTENSITIES, LIFETIMES
AND THE RATIO OF I/T FOR QUININE SULFATE,
10 ”M IN 0.1 M SULFURIC ACID, QUENCHED

BY CHLORIDE ION .................................................. 72

FLUORESCENCE INTENSITIES, LIFETIMES

AND THE RATIO FOR I/r FOR DANSYL-L—
PHENYLALANINE IN 2-PROPANOL, QUENCHED

BY NITROMETHANE ............................................... 74

EXCITED-STATE LIFETIMES OF STANDARD
COMPOUNDS DETERMINED BY SPT AND FDF ......... 112

FLUORESCENCE INTENSITIES, LIFETIMES ,

AND THE RATIO OF I/r FOR QUININE SULFATE,

10 11M IN 0.1 MSULFURIC ACID, QUENCHED BY
VARYING CONCENTRATIONS OF CHLORIDE ION ...... 114

FLUORESCENCE INTENSITIES, LIFETIMES,

AND THE RATIO OF I/r FOR DANSYL—L-
PHENYLALANINE , 10 ”M IN 2-PROPANOL,

QUENCHED BY VARYING

CONCENTRATTONS OF NITROMETHANE .................. 116

 

 

 

 

 

 

TAB]

TABl

TAB]

TABI

TABLE 5-1.

TABLE 5-2.

TABLE 5-3.

TABLE 5-4.

TABLE 5-5.

STEADY-STATE FLUORESCENCE INTENSITY
(ARBITRARY UNITS) FOR QUININE SULFATE
SOLUTIONS THAT ARE AFFECTED BY COLLISIONAL
QUEN CHIN G AND SAMPLE ABSORPTION ................. 130

FLUORESCENCE INTENSITY RATIOED TO
EXCITED-STATE LIFETIME FOR QUININE

SULFATE SOLUTIONS THAT ARE AFFECTED

BY COLLISIONAL QUENCHING AND

SAMPLE ABSORPTION ........................................... 131

FLUORESCENCE INTENSITY (ARBITRARY

UNITS) AS MEASURED BY THE CELL-SHIFT

METHOD FOR QUININE SULFATE SOLUTIONS

THAT ARE AFFECTED BY COLLISIONAL

QUENCHING AND SAMPLE ABSORPTION ................. 132

ABSORPTION CORRECTED FLUORESCENCE

INTENSITY RATIOED TO EXCITED-STATE

LIFETIMES FOR QUININE SULFATE SOLUTIONS

THAT ARE AFFECTED BY COLLISIONAL

QUENCHING AND SAMPLE ABSORPTION ................. 132

RELATIVE STANDARD DEVIATION OF THE DATA
OBTAINED BY EACH MEASUREMENT SCHEME ......... 133

xi

Figi
Figl

Figt

Figt

Figi

Figi

Figl

Figt
Figi

Figt
Figt

Fig];

 

 

'igure 2-1.
'igure 2-2.

'igure 2—3.
'igure 2-4.

'igure 3-1.

'igure 3-2.
igure 3-3.
gure 3-4.

gure 3-5.
gure 3-6.

gure 3-7.
gure 3-8.

ure 3-9.

LIST OF FIGURES

Block diagram of a pulse-sampling apparatus ...................... 18

Sampling sequence used in the pulse-sampling method.

The upper sequence of waveforms is a series of

fluorescence decays. The plots beneath them show

the pulse-sampling sequence and the increasing

delay between the trigger and sampling ............................. 19

Block diagram of a single—photon timing apparatus ............... 20

Excitation and emission waveforms for frequency
domain fluorescence ................................................... 24

Model of a SPEX 1681 Czemy-Tumer monochromator
showing rays that strike the left and right edges of the
grating. The wavelength of the rays is 500 nm ................... 39

Floor plan view of an American Holographic DB10-S
subtractrive dispersion double monochromator. From [25] ..... 41

Pulse height distribution of an MCP-PMT as a function
of applied voltage. The bias voltage increases from

left to right .............................................................. 43
Method of operation of a time-to-amplitude converter ........... 50
Diagram of the optimum instrumental configuration .............. 53

Fluorescence decay of naphthalene, 10 “M in
degassed cyclohexane .................................................. 59

Fluorescence decay of rose bengal, 10 I‘M in methanol,
and instrument response ............................................... 60

Fitted decay of naphthalene. The lower plot shows the
weighted residuals ...................................................... 65

Autocorrelation of weighted residuals from fit
of naphthalene decay ................................................... 66

 

 

 

Figure 3-10.

Figure 3—1 1.

Figure 3-12.

Figure 4-1.

1figure 4-2.

1:lgure 4-3.

Figure 44.

Figure 4-5.

Figure 4-6.

Flgul‘e 4:7.

Si
in:
Si;

Figure 3-10. Steady-state fluorescence intensity (circles) and intensity

Figure 3

ratioed to the excited-state lifetime (squares) for naphthalene

in cyclohexane (approx. 0.1 mM) ................................... 71
-l 1. Plot of steady state fluorescence intensity (circles) and

fluorescence intensity corrected for quenching by ratioing

to the excited-state lifetime (squares) for quinine sulfate in

0.1 M sulfuric acid quenched by chloride ion ..................... 73

Plot of steady state fluorescence intensity (circles) and
fluorescence intensity corrected for quenching by ratioing

Figure 3-12.
to the excited-state lifetime (squares) for dansyl-L-
phenylalanine in 2-propanol quenched by nitromethane .......... 74
Figure 4-1. Plot of the relative standard deviation (RSD) in an
excited-state lifetime determined from phase shift data
a function of the phase shift for an RSD in the phase

sift of 0.2 degrees ...................................................... 84
Simplified apparatus for frequency domain fluorescence ......... 86

Figure 4-2.
Calculated temporal output of a mode-locked laser (left)

Figure 4-3.

and its frequency spectrum (right). This illustration is for
a mode-locked laser in which 50 longitudinal modes are

phase—locked ............................................................ 88
Enlarged view of the temporal output of a mode-locked laser...88

Figure 4-4.
Integration scheme used in phase resolved fluorescence.

The shaded area represents the integrated signal. As the

Figure 4-5.
phase shift between F( t) and P(t) is varied from 0 degrees

Figure 4-6.

are 4-7.

or in phase (left), to slightly out of phase (middle), to 90
degrees out of phase (right), the integrated signal increases
from the maximum value (left) to an intermediate value

(middle), to zero (right) ............................................... 91
Simplified diagram of the detection electronics used in-

the instrument reported by Alcala and coworkers [24]. AMP:
amplifier, BPF: bandpass filter, INT: integrator, ZCD, zero-

crossing detector, R: rectifier ........................................ 94
Single channel of the detection electronics used in the
instrument reported by Alcala and coworkers [24]. The
signal is shown at key points to illustrate its operation.
I/ V: current-to-voltage converter, BPF: bandpass filter,
HVT: integrator, R: rectifier, ZCD: zero-crossing detector ...... 95

xiii

 

 

Figure 4-8. Di

Figure 4-9. Di

Figure 4-10. Oi

Figure 4-11. Sci
Figure 4-12. Bc

Figure 413. P11

Figure 4-14. p1.

Figure 5.1. M,

Figure 4-8.

Figure 4-9.

Figure 4-10.

Figure 4-1 1

Figure 4- 12.

Figure 4-13.

Figure 4-14.

Figure 5-1.

Figure 5-2.

Figure 5-3.

Diagram of the constructed frequency domain fluorometer
capable of directly determining the quantity I/r. SHG:

second harmonic generator, MLD: mode-locker driver,

FS: frequency synthesizer, PS: power splitter, DBM: doubly
balanced mixer, BB: black box ....................................... 98

Diagram of the circuit used to separate the AC and DC
components of the signal from the PMTs. I/V: curent-to-voltage
converter, DBM: doubly balanced mixer. .......................... 99

Overview of the circuits housed in the NIMs. I/V:
current-to-voltage converter, BPF: bandpass filter, Vpeak
to DC: Vpeak-to-DC converter, ZCD: zero-crossing
detector, TAC: time-to-amplitude converter, S&H:

sample-and-hold amplifier .......................................... 101
. Schematic diagram of the detection electronics ................... 102
Bode plot of the bandpass filter .................................... 105

Plot of steady-state fluorescence intensity (circles) and
fluorescence intensity corrected for quenching by ratioing to

the excited-state lifetime (squares) for quinine sulfate, 10 grid

in 0.1 M sulfuric acid, quenched by chloride ion ............... 115

Plot of steady-state fluorescence intensity (circles) and
fluorescence intensity corrected for quenching by ratioing

to the excited-state lifetime (squares) for dansyl-L-

phenylalanine, 10 pM in 2-propanol, quenched by

nitromethane .......................................................... 117

Methodology of the cell-shift technique reported by Lutz

and Luisi [2]. F1 and F2 are fluorescence intensities at cell
positions 1 and 2, respectively, where the optical

pathlengths are t, and (2. ............................................ 122

Response surface of fluorescence intensity as a function of
optical path length in solution for a 1.0 #M solution of
rhodamine B in ethanol. The axes units are tenths of

millimeters ............................................................ 127

Response surface of fluorescence intensity as a function

of optical path length in solution for a 100 pM solution of
rhodamine B in ethanol. The axes units are tenths of

millimeters ............................................................ 127

xiv

 

 

Figure A2-l.
Figure A2-2.
Figure A2-3.

Figure A3-1.

Figure A3-2.

Figure A4-1.

Figure A4-2.

Scl

Di.
lC\

 

 

Figure A2-l.
Figure A2-2.
Figure A2-3.

Figure A3-1.

Figure A3-2.
Figure A4-1.

Figure A4-2.

Schematic diagram of a switched fiber optic delay .............. 150
Diagram of circuit to bias the silicon avalanche photodiode . .. 155
Color coding of Optical fibers used in the switched

fiber optic delay ...................................................... 157
Diagram of the SPT apparatus emphasizing the two

data channels ......................................................... 169
Flow Diagram for SPT Experiments .............................. 179
Printed circuit board layout for MAR series amplifiers ........ 183

Diagram of the frequency synthesizer showing voltage
levels at each point in the circuit ................................... 184

XV

 

 

 

 

Molecular
ofa variety of c
selectivity and er
fluorescence assa
product of the (1.
chemical process
thus introducing

directly proportic

The fluor

expressed as;

Where E1 is the
excitation Source
immortal Cons
assumes that the
wavelengths. I“
one assumes that
set also affect th
mechanisms that.
[he fluorescencei

sample Such that 1

 

 

 

 

CHAPTER 1: INTRODUCTION TO CORRECTED FLUORESCENCE
MEASUREMENTS

Molecular fluorescence spectroscopy is a popular technique for the quantitation
of a variety of classes of compounds. This popularity is primarily due to the high
selectivity and excellent detection limits of the method. The only requirement for a
fluorescence assay is that the analyte fluoresces or can be derivatized such that the
product of the derivatization reaction is fluorescent. However, several physical and
chemical processes can affect the measured fluorescence intensity of a given sample,
thus introducing error in the assumption that the measured fluorescence intensity is

directly proportional to the concentration of the fluorophore.

The fluorescence intensity one measures with a typical fluorometer can be

expressed as:

EL = <I>¢kc (H)

where EL is the measured fluorescence intensity, <1) is the radiant power of the

excitation source, 4> is the quantum yield of the fluorophore, k is a collection of
instrumental constants, and c is the concentration of the fluorophore. This expression
assumes that the solution absorbance is negligible at the excitation and emission
wavelengths. If one directly compares fluorescence intensities from multiple solutions,
one assumes that processes influencing the fluorescence intensity of one solution in the
set also affect the other solutions to the same degree. Such processes can act via
mechanisms that depopulate the excited state of the fluorophore causing a reduction in
the fluorescence intensity, or they may involve extremes in physical parameters of the

sample such that assumptions made in deriving the expression are no longer valid.

 

 

 

 

 

If an eIe
fluorescence occr
vibrational energ
commonly referrt
intensity and the
Molecular oxygci

effective quenche

If the abs
exciting radiation
intensity will be 1
filter effect and
light prior to and
fewer molecules

This region of m

 

 

 

If an electronically excited fluorophore collides with another body before
fluorescence occurs, the energy stored in the excited fluorophore can be released as
vibrational energy in a process known as collisional quenching. This process is
commonly referred to simply as quenching, and it reduces the observed fluorescence
intensity and the excited-state lifetime. Many species can quench fluorescence.
Molecular oxygen and halide ions are among the most widely known and the most

effective quenchers.

If the absorbance of the solution is such that a large fraction of either the
exciting radiation or the emitted fluorescence is absorbed, the measured fluorescence
intensity will be less than that predicted by equation (1-1). This is known as the inner—
filter effect, and consists two types of sample absorption. Absorption of the source
light prior to and within the volume of the sample viewed results in the excitation of
fewer molecules of the fluorophore than for a solution having a lower absorbance.
This region of the solution is known as the pre-filter region, and hence a deviation
caused by such absorption is known as the pro-filter effect. Similarly, the post-filter
effect is due to absorption of the fluorescent radiation by the sample in the volume
viewed and in the region of solution between the volume viewed and the solution
boundary nearest the point of observation. It too reduces the measured fluorescence
intensity relative to a lower absorbance solution. Significant solution absorption can be
due to a high analyte concentration or to concomitant absorbing species. The term self

absorption is used to describe absorption due exclusively to the analyte.

The review of the scientific literature on correcting for these sources of error in
fluorescence measurements is presented in two parts: one devoted to correction for

quenching and one devoted to correction for sample absorption.

 

 

 

 

The theon
quenching was fr
report numerous
“POW. In .

measurements. \-

developed to mini

Quenchir
considerable time
and ”Mills have
perhaps the most
[haw Wes [5]. ;
been reported by 1
dry ice to the 33m}

meaSuffiment is n

 

 

 

REVIEW

Correction for Quenching

The theoretical basis for the reduction in fluorescence intensity due to collisional
quenching was first described in the classic work of Stern and Volmer [1]. Since that
report numerous chemical systems that exhibit collisional quenching have been
reported. In an effort to improve the accuracy of molecular fluorescence
measurements, various chemical and instrumentation based approaches have been

developed to minimize the effects of quenching.

CHEMICALLY-BASED APPROACHES

Quenching of fluorescence by molecular oxygen has been known for a
considerable time [2], and many schemes to remove oxygen from analytical samples
and solvents have been reported. Purging of samples with nitrogen [3] or argon [4] is
perhaps the most common method, but it is not as rigorous as repeated freeze-pump-
thaw cycles [5]. Recently, an incredibly simple and thorough means of degassing has
been reported by Matsuzawa et. a1. [6] in which one simply adds a small piece of clean
dry ice to the sample, and allows it to sublime completely, just before the fluorescence
measurement is made. More exotic methods such as purging while refluxing [7],
evacuation with stirring [8], purging followed by evacuation [9,10], ultrasonification
[11], and chemical scavenging [12,13] have been reported. A system for the
continuous removal of oxygen from a flowing stream has also been reported [14]. This
apparatus uses a semi-permeable membrane and an oxygen scavenging solution to
create a gradient in the oxygen concentration. The dissolved oxygen diffuses through

the solution and the membrane to the scavenging solution where it is reduced to water.

 

 

 

 

 

 

A similar schem

[15].

Another 2
shields the fluorc
them from the rr
"organized medi.
complexes such
envelop the fluor
the advantage tha

are more difficult

Several W
in the literature.

issue of the joum;

chemical Sensors,
in fluorescence i]
(benching that N:
type for in viva \
Woltbeis’ Posch,
0xlgen and haloth

 

4

A similar scheme has been used in an automated system for sample deoxygenation

[15].

Another approach for reducing quenching involves adding a compound that
shields the fluorophore from would—be quenchers to the sample, rather than removing
them from the matrix. This is one of the objectives of fluorescence spectroscopy in
"organized media", a topic of great interest in the recent literature. Inclusion
complexes such as micelles and cyclodextrins can partially surround or completely
envelop the fluorophore and protect it from collisional quenching. This approach has
the advantage that it can reduce quenching from oxygen, and from other quenchers that

are more difficult to expel from the sample.

Several reviews of organized media in fluorescence spectroscopy have appeared
in the literature. The most recent of these was in the semi-annual fundamental review

issue of the journal Analytical Chemistry [16].

As one might expect, quenching has been used to advantage; especially in
chemical sensors. Such sensors determine the quencher concentration by the decrease
in fluorescence intensity of a fluorescent dye. This decrease is of course due to
quenching that results from collisions with the quencher. An oxygen sensor of this
type for in vivo use has been reported by Peterson, Fitzgerald, and Buckhold [17].
Wolfbeis, Posch, and Kroneis [18] reported a similar sensor for the determination of

oxygen and halothane, a common anesthetic.

 

 

Attempts
measurement rat
instrumentation-b
disturbed and less
necessarily pertur

case of in situ me

The first i
was repongd by
that bOth the fit
Collisional quencl
ratio of the fluore

the Concentration

quenching SPeCies
The fluore:

where k is a pr:
mews such an
his the quantum
fluorophme. The

lifetime, 7’ t0 the]

 

5

INS HUMENT A TI ON-BASED APPROACHES

Attempts to reduce the effects of quenching by improving the basis of the
measurement rather than manipulating the matrix has been the focus of a few
instrumentation-based approaches. These have the advantage that the sample is not
disturbed and less effort is expended in sample preparation. Chemically-based methods
necessarily perturb the sample, but this may not be desirable or even possible as in the

case of in situ measurements.

The first instrumentation-based approach to correct for the effect of quenching
was reported by Hieftje and Haugen [19]. Their method takes advantage of the fact
that both the fluorescence intensity and the excited-state lifetime are reduced by
collisional quenching, and involves ratioing these quantities. They showed that the
ratio of the fluorescence intensity to the excited-state lifetime is directly proportional to
the concentration of the fluorophore and independent of the concentration of collisional

quenching species. The following derivation was presented in their paper:

The fluorescence intensity, F, can be expressed as:
F = k4>c (1-2)

where k is a proportionality constant that takes into account several instrumental

 

parameters such as the radiant power of the source, the area of the source viewed etc. ,
¢> is the quantum efficiency of the fluorophore, and c is the concentration of the

fluorophore. The quantum efficiency can be expressed as the ratio of the observed

lifetime, 7. to the lifetime in the absence of quenching, To (i.e. the natural lifetime):

as = f; (1-3)

 

 

 

 

 

Substituting this 6

If K is defined as

equation (14) her

Hence the ratio 0

and is not affecre

quantum efficient

In the orig
quenching in two
Chloride and the q
were detftrntined
determined With
Haugen were also
limited success.
time‘COI’related Si
sinchmnoufly PU]
aIlnde Pulse fate
p [W fluoreso
AlthOUgh each qr
“my, they bot]

counting techniqm
r6‘1“in for the n

6

Substituting this expression for the quantum efficiency into equation (2), one obtains:

F = kT—L— o (1—4)

If K is defined as the ratio of the constants k and To ,

K = —k— (15)
To

equation (1-4) becomes after rearrangement:
5 = Kc (1-6)
T

Hence the ratio of F/r is directly proportional to the concentration of the fluorophore
and is not affected by collisional quenching or any other process that decreases the

quantum efficiency.

In the original work [19], the Hieftje-Haugen approach was used to correct for
quenching in two chemical systems. These were the quenching of quinine sulfate by
chloride and the quenching of l-pyrene butyric acid by iodide. Fluorescence intensities
‘ were determined with a conventional fluorometer and excited—state lifetimes were
determined with a time-correlated single—photon counting apparatus. Hieftje and
> Haugen were also able to determine these quantities on a single instrument, albeit with
1 limited success. For this endeavor, they replaced the discharge lamp source of the
[time-correlated single-photon counting apparatus with a more amplitude stable
1synchronously pumped, cavity-dumped, frequency-doubled dye laser. The average
anode pulse rate integrated over a ten second time period and the integral of the
processed fluorescence decay were both used as a measure of the fluorescence intensity.
Although each quantity yielded reasonable precision as measures of fluorescence
intensity, they both suffer from limitations inherent in the time—correlated single-photon
counting technique. If the count rate of the system is increased to reduce the time

required for the measurement of intensity, a distortion in the shape of the recorded

decay results. '1
determined by in:
the intensity to th

the measurement

Alfano. F
correcting for the
m9" gated photon
the integrated are:
The temporal dec-
ernission informat
Single fluorescence
“llll an electrical
“Citing light. A

 

phOlOlede and the

were generated in

reference and 53m;

p03lll0n, AS in [i

excited.“ate decay

Demass 10m

Quenching in More:
detection Via a dual
method, the light 3

 

.. N-.._..-—.-.»

 

 

 

7

decay results. This greatly reduces the accuracy in the lifetime and in the intensity
determined by integrating the decay. Although Hieftje and Haugen showed the ratio of
the intensity to the lifetime is independent of quenching, they were not able to perform

the measurement of that ratio in real time.

Alfano, Fong, and Lytle [20] implemented the Hieftje-Haugen approach to
correcting for the effects of quenching with an instrument designed for high repetition
rate, gated photon counting. Similar to Hieftje and Haugen's work, Alfano et al. used
the integrated area of the fluorescence decay as a measure of the fluorescence intensity.
The temporal decay of the excited state was recorded by mixing the time-dependent
emission information with a temporally scanned reference. The signal generated by
single fluorescence photons impinging on a fast microchannel plate detector was mixed
with an electrical pulse generated by a fast photodiode that sampled the pulses of
exciting light. A trombone line and different lengths of coaxial cable between the
photodiode and the mixer provided the temporal scanning mechanism. Electrical pulses
were generated in the mixer and subsequently counted by a rate meter when the
reference and sample pulses were coincident at the mixer. The fluorescence decay was
obtained by plotting counts per unit time from the ratemeter versus the trombone line
position. As in the paper by Hieftje and Haugen, Alfano et al. had to curve fit the
excited-state decay data to obtain the fluorescence lifetime, and hence could not

approach a real-time measurement of the ratio of intensity to observed lifetime.

Demas, Jones, and Keller [21] reported two methods for reducing the effects of
uenching in fluorescence measurements. A modulated light source and phase sensitive
etection via a dual-channel lock-in amplifier were used in both methods. In the first
ethod, the light source was modulated faster than the period of the excited-state

'fetime. The modulation frequency was set sufficiently high that the fluorescence

 

 

 

 

 

detected at that
intensity by a si
technique of Hie
the phase shift b
the dual-channel
modulation inten

relating these qua

Here 8 is the obs
urlquenched 8am;

modulation freqm

0f the ”mil ion l

AlthOUgh

measurement [21]

modulation bands:
lifetimes of ten t
available, it is Ve
The second metho
for the analyte flt

 

fi

8

detected at that frequency occurred before collisional quenching could reduce the
intensity by a significant extent. The second method is similar to the time—domain
technique of Hieftje and Haugen [19]. In "method 2", the modulated amplitude and
the phase shift between excitation and emission were simultaneously determined with
the dual-channel lock-in amplifier, and were used to calculate the unquenched
modulation intensity at a modulation frequency of zero. The following expression
relating these quantities was given, but its derivation was not:

3
s = ° . (1-7)

[[3]] 2 + two) ] 3

Here 5 is the observed modulated amplitude and So is the modulated amplitude for an

 

unquenched sample at a modulation frequency of zero. The quantum yield, radial

modulation frequency, and natural lifetime are (b, w and 10 respectively. Fluorescence

  
  
   
 
 
 
  

of the uranyl ion quenched by chloride was studied by both methods.

Although both methods worked well and demonstrated the principle of the
measurement [21], the system studied was that of a long—lived fluorophore (1 = 170
us), and not representative of common fluorophores. The first method would require a
modulation bandwidth of gigahertz for fluorophores having more typical un-quenched
lifetimes of ten to 20 nanoseconds. Although instrumentation of. this bandwidth is
vailable, it is very expensive and not directly compatible with the reported system.
c second method requires that the natural lifetime must also be determined or known
or the analyte fluorophore in the matrix at hand. The instrumental zero-time phase

gle must also be known. The method is also somewhat computationally complex.

 

 

There is
spectroscopy at
[22], Christmann
to be completed '
in them constitu

corrected fluoresr

Yemen,
POIYCltromatic lu
based on a dual
record fluoresce;
Contained in a Sta

the work of Parke

 

 

9

Correction for Sample Absorption

There is a long history of research in absorption corrected fluorescence
spectroscopy at Michigan State University. It includes the dissertations of Holland
[22], Christmann [23], Ratzlaff [24], Adamsons [25 ,26], and Doherty [27] (and the yet
to be completed work of Victor [28]). These dissertations and the references contained
in them constitute an excellent review. This section reviews work in absorption

corrected fluorescence published since the most recent of those dissertations [27].

Yappert, Schuyler, and Ingle reported a scheme for the correction of
polychromatic luminescence signals for the inner-filter effect [29]. Their work was
based on a dual multichannel, fiber optic based instrument that can simultaneously
record fluorescence, absorption and chemiluminescence spectra from a sample
contained in a standard four cm by one cm2 cuvette [30]. In essence, they extended
the work of Parker and Barnes [3] and Holland and coworkers [31] to include situations
where the solution absorbance is not constant over the excitation and emission
bandpasses, and where optical fibers are used to collect the fluorescence. It was
assumed in previous correction procedures that the solution absorbance is constant over
these wavelength intervals. Their correction scheme was tested with a quinine sulfate -
methyl red system and subsequently used to study the lucigenin - hydrogen peroxide

chemiluminescence system [32].

l. Stem,0.;\
2. Kautsky, H.
3. Parker,C.A
4. Lund,W.;

5. Whery. E.L
Dekker: Net

6. Matsuzawa.
7. Bratin,K.;]
8. Battino,R,;
9' Norris, B.J.
hh- Fox,M.A.;
ll. Dell'ova,V
12. Rollie, M.E.

h3- MacCrehan,

 

Reim, R. An
Rollie, ME.
MCGown, L.
Pehrson, 1.1
18' W01lbeis,0g
Hieftje, GM
20. Alfano,“.g

. DCmas, J‘N.;

 

 

 

CHAPTER 1 REFERENCES

Stern, 0.; Volmer, M. Z. Phys. 1919, 20, 183.

Kautsky, H. Trans. Faraday Soc.1939, 35, 216.

Parker, C.A.; Barnes, WJ. Analyst, 1957, 82, 606.

Lund, W.; Opheim, L.-N. Anal. Chim. Acta. 1975,79,35.

Whery, E.L. in Fluorescence Theory, Instrumentaron, and Practice,Marcel
Dekker: New York, l967,.p.92. G.G. Guilbault, Ed.

Matsuzawa, S.; Wakisaka, A.; Tamura, M. Anal. Chem. 1990, 62, 2645.
Bratin, K.; Kisinger, P.T.; Brunlett, C.S. J. Liq. Chromatogr. 1981, 4, 1777.
Battino, R.; Banzhof, M.; Bogan, M.; Wilheim, E. Anal. Chem. 1971, 43, 806
Norris, B.J.; Meckstroth, M.L.; Heineman, W.R. Anal .Chem. 1971, 43, 806.
Fox, M.A.; Staley, S.W. Anal. Chem. 1976,,48, 992.

Dell'Ova, V.E.; Denton, M.B.; Burke, M.F. Anal. Chem. 1974, 46, 1365.
Rollie, M.B.; Ho, C.-N.; Warner, I.M. Anal. Chem. 1983, 55, 2445.
MacCrehan, W.A.; May, W.E. Anal. Chem. 1984, 56, 625.

Reim, R. Anal. Chem. 1983, 55, 1188.

Rollie, M.B.; Patonay, 6.; Warner, I.M. Anal. Chem. 1987, 59, 180.
McGown, L.B.; Warner, I.M. Anal. Chem. 1990, 62, 255R.

Peterson, 1.1.; Fitzgerald,R.V.; Buckhold, D.K. Anal. Chem. 1984, 56, 62.
Wolfbeis, 0.8.; Posch, H.E.; Kroneis, H.W. Anal Chem. 1985, 57, 2556.
Hieftje, G.M.; Haugen, G.R. Anal. Chim. Acta. 1981, 123, 255.

Alfano, A.J.; Fong, F.K.; Lytle, F.E. Rev. Sci. Instrum. 1983, 54, 967.

Demas, J.N.; Jones, W.M.; Keller, R.A. Anal. Chem. 1986,58, 1717.

 

 

 

 

22.

23.

24.

26.

27.

29.

30.

32.

Holland, 1.
1971.

Christmann
MI, 1980.

Ratzlaff, E.
1982.

Adamsons,

Adamsons,
1985 .

Doherty, T,
1990.

Victor. M. A
Yappert, M.

Yemen. M.

- Holland, 1}

706.

Yappert, M.

 

 

11

Holland, J.F. Ph. D. Dissertation, Michigan State University, East Lansing, MI,
1971.

Christmann, D.R. Ph. D. Dissertation, Michigan State University, East Lansing,
MI, 1980.

Ratzlaff, E.H. Ph. D. Dissertation, Michigan State University, East Lansing, MI,
1982.

Adamsons, K. MS. Thesis, Michigan State University, East Lansing, MI, 1982.

Adamsons, K. Ph. D. Dissertation, Michigan State University, East Lansing, MI,
1985.

Doherty, T.P. Ph. D. Dissertation, Michigan State University, East Lansing, MI,
1990.

Victor, M.A. unpublished work.
Yappert, M.C.; Schuyler, M.W.; Ingle, J.D. Anal. Chem. 1989, 61, 593.
Yappert, M.C.; Ingle, J.D. Appl. Spectrosc. 1989, 43, 759.

Holland, J.F.; Teets, R.E.; Kelley, P.M.; Timnick, A. Anal. Chem. 1977, 49,
706.

Yappert, M.C.; Ingle, J.D. Appl. Spectrosc. 1989, 43, 767.

 

 

 

CHAPTERZ:
BASEDAPPR

While sat
sample absorptic
correction for the
based methods to
because the latter
time and capital 1
methods do not p
be.l'ond that need:

quenching is extre

0f the in:
Haugen method [1
fluorescence lnten

memOdS reviewed

“My Correct the

A reilletime
“Correcting for t]
lifetime and the s}
latter quantity is re

of the excited.sm
hhqllifing flUOreSCe

CHAPTER 2: DESIGN CONSIDERATIONS FOR AN INSTRUIVIENTATION-
BASED APPROACH TO REAL-TIME CORRECTION FOR ERRORS DUE TO
QUENCHIN G

While satisfactory methods exist to correct fluorescence measurements for
sample absorption, a routine, instrumentation-based technique for the real-time
correction for the effects of quenching has yet to be fully developed. Instrumentation-
based methods to correct for quenching are preferable over chemically-based methods
because the latter necessarily perturb the system under study and require investments of
time and capital in the form of additional sample preparation. Instrumentation-based
methods do not perturb the system under study, and do not require sample preparation
beyond that needed for a routine fluorescence assay. Of course in situations where

quenching is extreme, a combination of methods may be needed.

Of the instrumentally based methods to correct for quenching, the Hieftje-
Haugen method [1] is one of the best to develop. It provides an exact means to correct
fluorescence intensity for the reduction caused by collisional quenching. The other
methods reviewed in chapter one only minimize the effects of quenching. They do not

exactly correct the measured quantity.

A real-time or near real-time implementation of the Hieftje-Haugen method [1]
of correcting for the effects of quenching requires rapid acquisition of the excited-state
lifetime and the steady-state fluorescence intensity. The real-time acquisition of the
latter quantity is relatively simple. The same cannot be said for the rapid determination
of the excited-state lifetime. The criteria and choice of a measurement strategy for
acquiring fluorescence intensity and excited-state lifetime data needed for the real-time

implementation of the Hieftje-Haugen method are discussed in this chapter.

12

 

 

 

CF

In additior
lifetime, one nor

collisional quencl

Linearity
re5ponse of the t
"bending over" d
detector is a phoi
Proportional to h
PMT in absolute .
hhhoamp or less 1'
bl “’01ng a PM]
he Photocathode
detect is often lim

detection limits or

If Cofiectjr
lange is limited if
llpical Value for t
tolerated with a
memes, esoecia
not uncommOn ft
"hhgllltude in mm
intllltmems is 1'0:

 

 

13

CRITERIA FOR THE DETERMINATION OF I AND 1

In addition to typical values for the fluorescence intensity and the excited-state
lifetime, one needs to know the extent to which these quantities are reduced by

collisional quenching to implement the Hieftje-Haugen method [1] intelligently.

Linearity in measured fluorescence intensity is generally limited by the dark
response of the detector on the low end of the intensity scale and by the classical
"bending over" due to self-absorption at the high end. In modern spectrometers, the
detector is a photomultiplier tube (PMT), and the quantity measured and considered
proportional to fluorescence intensity is the PMT's anode current. The response of a
PMT in absolute darkness is referred to as the dark current, and it is on the order of a
nanoamp or less for most PMTs. A dark current of picoamps or less can be achieved
by cooling a PMT. The extent of this reduction depends on the degree of cooling and
the photocathode material. The minimum detectable concentration an instrument can
detect is often limited by the detector's dark current. State-of-the-art instruments have

detection limits on the order of 10'12 M for strongly fluorescent compounds [2].

If corrections are made for self-absorption, the high intensity end of the linear
range is limited by the PMT's maximum average anode current. One milliamp is a
typical value for the maximum average current of a PMT. Higher'light levels can be
tolerated with a photodiode. However, PMTs are superior for detecting lower
intensities, especially in the ultra-violet and the blue end of the visible spectrum. It is
not uncommon for the linear range of a fluorometer to extend over six orders of
magnitude in concentration [2]. Not surprisingly, the linear range in well-designed
instruments is roughly equal to that of a room temperature PMT: nanoamps to

milliamps.

 

 

 

 

 

Real-time
simply connect t]
digital display or
simple oscilloscc
covers the openir
lflthmt
time dependent I
applying a bias in
More elaborate 5

computer have be

huMm
for lTw-Sdlbene
Values of 1 to 1(

for anthracene [5

The exam
thmm

In this eXPh’ESSion
“Mme“

hhhcehlration 0 f '

 

 

14

Real-time acquisition of fluorescence intensity is straightforward. One may
simply connect the anode of a PMT to a common digital multimeter for a convenient
digital display or to an oscilloscope for higher bandwidth, analog display. In fact a
simple oscilloscope configuration is the suggested means of testing a PMT. One
covers the Opening in the PMT's housing with black electrical tape, and creates a light
leak in the form of a pin hole in the tape. Fluorescent room lights provide a convenient
time dependent light source. Triggering the oscilloscope from the power line and
applying a bias voltage to the PMT yields a rough 60 Hz sine wave on the oscilloscope.

More elaborate schemes for automated, high speed data acquisition using a laboratory

 

computer have become not only possible, but increasingly affordable in recent years.

In extreme cases, excited-state lifetimes can range from picoseconds (e.g. 70 ps

 

for trans-stilbene in hexane [3,4] ) or faster to milliseconds for some transition metals.
Values of 1 to 100 nanoseconds are more typical for aromatic compounds (e. g. 4.1 ms

for anthracene [5] and 96 us for naphthalene [6]).

The extent to which the fluorescence intensity and the excited state lifetime are

reduced by collisional quenching can be deduced from the Stern-Volmer equation [7]:
F0

In this expression, F0 and F are the fluorescence intensities in the absence and presence

 

of quenching, kq is the bimolecular quenching constant, To is the excited state lifetime

  
  

of the fluorophore in the absence of quenching (i.e. the natural lifetime), and [O] is the
concentration of quencher. By defining the Stern-Volmer quenching constant, K0:

K0 = quO (2-2)

 

 

 

 

the Stem-Volmer

Because collision;

lifetimes in the at

’0

Where I“ is the n

expressed as :

hhe right-hand Sit
Hence collisional

lifetime to the 53“

II is inten
constant that the
the quencher is l
COllisiOnal quenc
concenll‘ntion of r
Volmer plop, the l
810peofKD- Mu
h a Shem‘Volme,
amino acids in pa

accessible and, her

If the Steh
the ”Med redu.

15
the Stem—Volmer equation can rearranged to yield:

i=2 = 1 + KD[Q] (2-3)

Because collisional quenching is a kinetic process that depopulates the excited state, the

lifetimes in the absence (To) and presence (1) of quenching can be expressed as:

To = I‘ '1 (2-4) 1- = (P + kq[Q]) '1 (2—5)
Where I‘ is the natural rate of decay of the excited state The ratio of To to r can be

expressed as :
To
‘7— = 1 + qu°[Q] = 1 + KD[Q] (2-6)

The right-hand side of this expression is identical to that of the Stem-Volmer equation.
Hence collisional quenching reduces the fluorescence intensity and the excited-state

lifetime to the same extent.

It is interesting to note from the definition of the Stem-Volmer quenching
, constant that the fluorescence intensity is reduced by 50% when the concentration of
h the quencher is KD'1. The Stem-Volmer quenching constant and information on
‘ collisional quenching can be obtained by plotting Fo/F (or ro/r) versus the
concentration of quencher. In this classic graphical construct, known as a Stem-
Volmer plot, the plotted data are linear, intercept theordinate axis at unity, and have a
_ slope of KD. Multiple environments of the fluorophore are indicated by non-linearity
[in a Stem-Volmer plot. Such situations are often encountered in the fluorescence of
h amino acids in proteins where the amino acids on the surface of the protein are more

accessible and, hence, more readily quenched than those in the interior [8].

If the Stem-Volmer quenching constant of the quencher in question is known,

the expected reduction in fluorescence intensity due to collisional quenching can be

 

 

readily determine
containing values

the scientific liter

To get a r
excited-state lifeti
quencher. The di
cmh/sec in water
lntennediate valu
lmnslates to diffu
lifetimes of 4, j
hhhmphore is u
thlallv impossu
WWW quenc
diffusion coefficit
constant of 32,4 ‘
°Xygen of 1/(32
Woman by a fi
hhlhhlllalene, the J
bl h lacror of 3

Significant when
Wines,

me the
hu°hh50ence inter

HieftjsHaugen n

16

readily determined from the Stem-Volmer equation. Unfortunately reference volumes

containing values of kq and KD do not exist. However, such information is available in

the scientific literature for a number of fluorophore-quencher pairs.

To get a grasp on the extent that quenching reduces fluorescence intensity and
excited-state lifetime, consider quenching by oxygen, a well known and very effective
quencher. The diffusion coefficient of oxygen is quite large, ranging from 2.5 x 10'5
cmz/sec in water at 25°C [8] to 5.31 x 10'5 cmzlsec in cyclohexane at 29.6°C [9].
Intermediate values are found for solvents of intermediate polarity. For water, this
translates to diffusion distances of 44, 100 and 224 A° over the course of excited state
lifetimes of 4, 20, and 100 ns [8]. Obviously quenching is most severe if the
fluorophore is long-lived. In fact, room temperature phosphorescence (RTP) is
virtually impossible if oxygen is present [10]. The bimolecular quenching constant for
tryptophan quenched by oxygen in water is 1.2 x 1010 M" sec1 [8]. From the

diffusion coefficient for oxygen in water, one can calculate a Stem-Volmer quenching

constant of 32.4 M" for tryptophan (To = 2.7 ns) [8]. Therefore a concentration of
oxygen of 1/(32.4 M") or 0.03 M would be needed to quench fluorescence of

tryptophan by a factor of two. Assuming the same bimolecular quenching constant for
naphthalene, the same concentration of oxygen would reduce naphthalene fluorescence
by a factor of 35. For most fluorophores (i.e. 1' < 20 ns), quenching becomes
significant when the total concentration of quenching species approaches millimolar

values.

From the preceding discussions, one can specify criteria for measuring
fluorescence intensities and excited-state lifetimes which are nwded to implement the

Hieftje-Haugen method [1]. The linear range in fluorescence intensity should be

 

 

 

several orders of
wide range, and
quenching specie
lifetimes from 10
dependent, but d
determination of
fluorescenct spec:
by a factor of 10
be on the order 0

the determination

REVIEW

Fluorescer

inStrument that is
such an approach

We" apparau
information, Ea,

either a time dom.

I0 excite the Sam
detay with a devit
the advantage of

sufficient mm

17

several orders of magnitude in order to determine fluorophore concentrations over a
wide range, and correct for quenching in the presence of large concentrations of
quenching species. The instrument should also be able to determine excited—state
lifetimes from 100 ms to less than 1 ns. The excited-state lifetime is not concentration
dependent, but does vary from fluorophore to fluorophore. This range allows the
determination of excited—state lifetimes for many fluorophores, but not for all
fluorescenct species. It also allows for a maximum reduction of excited state lifetime
by a factor of 100. The time required for the measurement of these quantities should
be on the order of seconds or less. The following section reviews existing methods for

the determination of fluorescence intensity and excited-state lifetime.

REVIEW OF NIETHODS FOR THE MEASUREMENT OF I AND 1'

Fluorescence intensity and excited—state lifetime can be determined in a single
instrument that is essentially two instruments in a single housing. However pursuing
such an approach would not accomplish anything new, and would result in a needlessly
complex apparatus. Several methods exist that provide both of these pieces of
: information. Each has its own advantages and limitations, and can be classified as

either a time domain method or a frequency domain method.

Time Domain Methods

The simplest and most intuitive means of determining an excited-state lifetime is
to excite the sample with a brief pulse of light and directly record the fluorescence
decay with a device such as a camera coupled to an oscilloscope [11]. This method has
the advantage of simplicity in setup and operation. However it does not provide

sufficient accuracy for the determination of an excited-state lifetime to several

 

 

 

 

 

significant figure
nithout the use t
bandwidth in are
nanosecond. lt
before curve fittir
bra transient <
complexity; it car

is onepart in 212

An impro

block diagram of

Figure 2.1. Blc

passed throngh a
dEieCiOI' is ampl

OSCiIIOSCOPe Or a ‘-

18

significant figures. In addition, determining extremely fast lifetimes is not possible
without the use of equally fast oscilloscopes. For example, an oscilloscope having a
bandwidth in excess of a gigahertz is needed to determine a fluorescence lifetime of a
nanosecond. It is also not "user friendly", as the data must be digitized manually
before curve fitting. Replacing the conventional oscilloscope with a digital oscilloscope
or a transient digitizer removes the manual digitization step at the cost of added
complexity; it can, however, improve accuracy if the resolution of the digitizing device

is one part in 212 or higher.

An improvement over direct recording is the pulse sampling method [12]. A

block diagram of an apparatus for this method is shown in Figure 2-1. A pulsed light

PULSED LIGHT Excrrarrnu
[ SOURCE _}"‘iMnNDCHRuMATuR SAMPLE
EMISSION
REFERENCE MONOCHROMATOR
DETECTOR
SAMPLE
DETECTDR

 

               

 

SAMPLING DSCILLDSCUPE

 

BUXCAR INTEGRATEIR

 

 

Figure 2-1. Block diagram of a pulse-sampling apparatus.

source is used to excite the sample, and fluorescence from the sample is collected and
passed through a wavelength selection device prior to detection. The output of the
detector is amplified and recorded with a sampling device such as a sampling

oscilloscope or a boxcar integrator. These devices sample the detector response over a

 

 

 

 

 

 

 

narrow time win.
fluorophore and
triggered by a ti
decay by a contir

waveform. This

INTENSX TY

GATE
{—2. _. _ __ __ _

 

SAMPLING FLUORESCENCE

Figure 2.2. San
sequence of wave
Show the pulse s
sampling,

One of the
Slate Via flUOresc
The linemen“

19
narrow time window. The fluorescence decay is recorded by repetitively exciting the
fluorophore and sampling the fluorescence decay in step-wise fashion. Sampling is
triggered by a temporal reference detector, and is scanned across the fluorescence
decay by a continuously swept delay between the trigger signal and the sampling of the

waveform. This methodology is illustrated in Figure 2-2. A sampling oscilloscope

\ _\_

INTENSITY

 

 

 

 

GATE
F—

 

SAHPLING FLUURESCE NCE

TIME

Figure 2-2. Sampling sequence used in the pulse sampling method. The upper
sequence of waveforms is a series of fluorescence decays. The plots beneath them
show the pulse sampling sequence and the increasing delay between the trigger and
sampling.

, acquires a single point while a boxcar averager acquires multiple samples and returns
‘_ the mean. Because of its signal averaging nature, higher signal-to—noise data are
‘ obtained with a boxcar averager. In either case, successive decay curves are averaged
1 until the signal-to-noise ratio is sufficient for curve fitting. Computer-based curve
fitting or less sophisticated manual approaches are used to extract the excited-state
lifetime from the raw data. The fluorescence intensity is obtained by integrating the

? fluorescence decay over time.

One of the most widely used techniques for recording the decay of an excited
state via fluorescence is time-correlated single-photon counting [13, 14]. IUPAC
recently recommended the name Single Photon Timing (SPT) for this method [15].

The fluorescence decay is acquired by measuring the time between excitation of the

 

 

 

 

 

sample and the e
light sources are
excitation flash is

shown in Figure ‘

f—-_‘1

l _“_I

Figure 23- Blc

the light is dire:
comm“ Signal wi
willple. Fittores
mml SElectei
detector. After an
converted to digit
START and 3p
START/STOP pa

 

20

sample and the emission of a photon. Relatively weak, high repetition rate, pulsed
light sources are used such that the probability of detecting a fluorescence photon per
excitation flash is less than unity. A diagram of the instrumentation used in SPT is

shown in Figure 2-3. The train of exciting pulses is split such that a small portion of

PULSED LIGHT EXCITATIDN
I SOURCE l ’MDNDCHRDHATUR SAMPLE
EMISSION
REFERENCE
- DETECTUR

 

        
   
 

 
 

        
     
 

MDNUCHRDMATDR

SAMPLE
DETECTOR

 

DISCRIMINATUR DISCRIMINATDR

   

Figure 2-3. Block diagram of a single photon timing apparatus.

the light is directed toward to a reference detector, which provides a temporally
constant signal with respect to the excitation pulse, while the remainder continues to the
sample. Fluorescence photons from the sample are collected with a lens system,
Spectrally selected with a filter or monochromator, and detected with a very fast
detector. After amplification, the smooth, analog pulses generated by the detectors are
converted to digital pulses using discriminators. The edges of these pulses then provide
START and STOP inputs to a time—to-amplitude-converter (TAC). For each

START/STOP pair, the TAC provides a pulse having an amplitude that represents the

 

 

 

 

time between ST;
by a multichannel
accumulating the
photon, the med
methods, the exci
The fluorescence

or by integrating ‘

Desilets ar
state lifetimes th
fluorescence dec
Smilling oscillos
thelaser. The ar
0f the oscilloscoj
inserting a delay
oscilloscope. Th

excited-state lifet

constants, and rj

t? was recorded 1

CuWe.

21

time between START and STOP. The amplitude of pulses from the TAC is digitized
by a multichannel analyzer (MCA) and stored in a buffer memory or a computer. By
accumulating the time between the excitation pulse and the detection of a fluorescence
photon, the excited-state decay can be constructed. As with the pulse sampling
methods, the excited-state lifetime must be extracted from the raw data by curve fitting.

The fluorescence intensity can be derived by measuring the sample detector pulse rate,

‘ or by integrating the acquired decay curve.

Desilets and coworkers [16] reported a simple method for determining excited-
state lifetimes that acquires the fluorescence intensity at only two points on the
fluorescence decay. Their instrument used a nitrogen laser and a dual-channel
sampling oscilloscope. The oscilloscope was triggered by a photodiode illuminated by
the laser. The anode current from the PMT was split and sent to both of the channels
of the oscilloscope. Sampling the decay curve at two times was accomplished by
inserting a delay cable between the power splitter and one of the channels of the
oscilloscope. The ratio of the two intensities was recorded, and used to determine the

excited-state lifetime by the equation:

I M -| 5—2 +-——t“t2 27
”Itt,) ‘nk, 1 (')

where I(t,) and I(t2) are the intensities at times t1 and t2, k1 and k2 are instrumental
constants, and r is the excited-state lifetime. The ratio of the intensities at times t, and
t2 was recorded for fluorophores of known lifetime and used to construct a calibration
curve. Once the calibration curve was constructed, lifetimes could easily be
determined "on-the fly" with computer controlled acquisition and data manipulation.
Their system was used to determine the excited-state lifetime of gentisic acid and

quinine sulfate in static cells and in flowing streams. It was also used to determine

 

 

 

excited-state lifeti
method works Wt
intensity informat
intensities at the

expensive and sor
requires optimiza
splitter and the
fluorophores haw

measurement C01]

Rather th
Observing the de
by continuously
Comparing the e
utilize this pm.
The waveforms
quantities derive
page 23 in Figt
flequenCy w, -
freunIICy, bills
the finite reside
lifetime increas
mOdulated than
demodulatiOn f,

22

excited-state lifetimes for fluorescence detection in liquid chromatography [17]. This
method works well for determining excited-state lifetimes, but it does not provide
intensity information. In fact, one of the attributes of this method is that the ratio of
intensities at the two times is concentration independent. It also requires moderately
expensive and somewhat uncommon instrumentation. As presented, this method also
requires optimization, or at least changing, the length of the delay cable between the
splitter and the oscilloscope to change the temporal aperture to accommodate
fluorophores having significantly different lifetimes. However the principle of the

measurement could be implemented with other means.

Frequency Domain Methods

Rather than sequentially exciting the sample with a brief pulse of light and
observing the decay of the excited-state, the same chemical information can be derived
by continuously illuminating the sample with sinusoidally modulated light and
comparing the emission waveform with that of the exciting light. Techniques that
utilize this principle are broadly grouped as frequency domain fluorescence techniques.
The waveforms used in a basic frequency domain fluorescence experiment and the
quantities derived from them used to determine the excited—state lifetime are shown on
page 23 in Figure 2-4. The excitation waveform is modulated at radial modulation
‘ frequency w. This induces a modulation on the emitted fluorescence at the same
frequency, but shifted in phase relative to excitation. The phase shift, as, is a result of
the finite residence time of the fluorophore in the excited state and increases as the
- lifetime increases for a given modulation frequency. The emission is also less
' modulated than the excitation, and the extent of this demodulation is expressed in the

demodulation factor, m. Its value ranges from zero to unity with decreasing lifetime,

 

Fig

en

the

ey .-.a

INTENSITY

EMISSION

0 EXCITATION

 

l—>

TIME

Figure 2-4. Excitation and emission waveforms for frequency domain fluorescence.

and it is the ratio of the AC to DC amplitudes of the emission waveform relative to the
excitation waveform. The excited—state lifetime can be calculated directly from either

the phase shift or the demodulation factor [8]:

7» =tan a
0)

l l g
(2-8) Tm = Z [ ‘— ~ 1] (2-9)

m2

Modulation frequencies on the order of several tens of megahertz are required to

measure excited-state lifetimes in the nanosecond range.

Various experimental approaches have been developed to extract the phase and
demodulation information. A complete history of these is given in chapter four.
Modern instrumentation uses cross-correlation detection [18] to step down the
frequency of measurement from the high modulation frequencies used to resolve
lifetimes in the nanosecond range. The highest modulation bandwidth reported is ten

GHz [19] providing temporal resolution of 1 ps.

Almost all existing instruments use a cross-correlation frequency (i.e. the

frequency at which phase and demodulation data are measured) of 30 Hz. This allows

 

 

real-time acquisit
computer controllt
frequency is the
instrument, name

constructed by war

Of the tir
coworkers [l6], 1
is the only one m
also does not rec
the raw data. H.
Channel samplin;
Oscilloscope), an
allllruach basedc
II would have tr
fl”Citations. In
small range of
”mpuermnm

appmlll'iate dela.

The dire
flu0rescenee an

(Ismain teeItniqt
i0 eXtIaCI the e

domain techniq

24

real—time acquisition of excited-state lifetime and fluorescence intensity through
computer controlled instrumentation, The widespread use of a 30 Hz cross-correlation
frequency is the result of incorporating the detection electronics of a commercial
instrument, namely the Aminco model 4800 phase fluorometer, in instruments

constructed by various research groups.

CONCLUSIONS

Of the time-domain methods, the method developed by Desilets, Lytle and
coworkers [16], in which two points are acquired during the decay of the excited-state,
is the only one that offers a means of rapid acquisition of the excited-state lifetime. It
also does not require extensive curve fitting to extract the excited-state lifetime from
the raw data. However it requires a somewhat rare and expensive instrument, a dual-
channel sampling oscilloscope (one could also use a transient digitizer or digital
oscilloscope), and it does not provide a means of obtaining fluorescence intensity. An
approach based on their method could include acquisition of fluorescence intensity, but
it would have to be normalized to the source intensity to compensate for amplitude
fluctuations. In addition, because it uses a fixed delay line, it would be limited to a
small range of lifetimes. This could be overcome to some extent by including a
computer-controlled switch and multiple channel power splitter/combiner to select an

appropriate delay cable for a sample having a given excited-state lifetime.

The direct relationship between the parameters measured in frequency domain
fluorescence and the excited-state lifetime provides a distinct advantage over time
domain techniques. Unlike most time domain techniques, which require curve fitting
to extract the excited-state lifetime, curve fitting is not a requirement of frequency

domain techniques. However to improve confidence when examining systems that

 

 

 

exhibit multi-expc

make measuremen

Frequency
the real-time imp
effects of quencl
fluorescence was
experimentally de
time required for
most time domair
potential for real-
crosscorrelation
possible. In orc‘
different lifetime:
controlled appara

the modulation fr

Two mea
commercial freqt
and fabricated in
Commemial ente

available for this

A comp
mealSllfements w.
“8° reduce reli
St"Iflslasrds [21, 2:

25

exhibit multi-exponential lifetimes or distributions in the lifetime data, many workers

make measurements at multiple modulation frequencies and curve-fit the acquired data.

Frequency domain fluorescence was chosen as the best technique to pursue for
the real-time implementation of the Hieftje-Haugen method [1] of correcting for the
effects of quenching in molecular fluorescence spectroscopy. Frequency domain
fluorescence was chosen for several reasons. The direct relationship between the
experimentally determined quantities and the excited-state lifetime greatly reduces the
time required for the measurement and the amount of data manipulation compared to
most time domain techniques. Consequently, frequency domain fluorescence has the
potential for real-time acquisition of the excited-state lifetime. In fact, increasing the
cross-correlation frequency could make data acquisition on the stopped flow time scale
possible. In order to make accurate measurements on samples having significantly

different lifetimes, the modulation frequency must be changed. However, a computer
1 controlled apparatus that incorporates commercial frequency synthesizers could change

the modulation frequency in a few microseconds [20].

Two means are possible to implement the Hieftje-Haugen method [1]. A
commercial frequency domain fluorometer could be modified, or one could be designed
' and fabricated in-house. Either route could be reproduced by other laboratories or by
commercial enterprises. As a commercial frequency domain fluorometer was not

available for this work, the latter approach was undertaken.

A complementary method to confirm or refute excite-state lifetime
measurements would be advantageous. The availability of such instrumentation would
also reduce reliance on compounds or mixtures accepted as excited—state lifetime

standards [21, 22]. It has been shown that doing so may not be a wise practice as the

 

fluorescence dew
Because funds ar

apparatus, SPT w.

 

 

 

 

26

fluorescence decay of accepted standards may not be as quantitatively certain [23].
Because funds and equipment were available to construct a single photon timing

apparatus, SPT was chosen as the complementary method.

 

 

 

Hieftje, G.l

. Ingle, J.D.

Cliffs, NJ .

Courtney, E
Hochstrasse
Birks, I .B. :

Berlman. 1.
Academic I

. Stern, O,;‘

Lakowicz.
York, 1983

Weast. R.(
FL, 1980,

Kraus, Pi
1988.

. Demas, 1.]

1983.

Ware, W;
Lamola. a

O I Connor,
ACadeic j

Fleming ’
UfllVet‘gity

ComISSIOn

I)esilets, [
Chem. 198

' Desilets:

 

 

10.

11.

 

27

CHAPTER 2 REFERENCES

Hieftje, G.M.; Haugen, G.R. Anal. Chim. Acta. 1981, 123, 255.

Ingle, J.D.; Crouch, S.R. Spectrochemical Analysis Prentice Hall: Englewood
Cliffs, NJ, 1988.

Courtney, S.H.; Flemming, G.R. J. Chem. Phys. 1980, 83, 215.
Hochstrasser, R.M. Pure Appl. Chem. 1980, 52, 2683
Birks, J.B.; Munro, I.H. Prog. Reac. Kinetics 1967, 4, 239.

Berlman, I.B Handbook of Fluorescence Spectra of Aromatic Molecules 2nd Ed.
Academic Press: New York, 1971.

Stem, O.; Volmer, M. Z. Physik, 1919, 20, 183.

Iakowicz, J .R. Principles of Fluorescence Spectroscopy Plenum PresszNew
York, 1983.

Weast, R.C. Ed. Handbook of Chemistry and Physics CRC Press: Boca Raton,
FL, 1980, p. F-6l.

Kraus, P.R. Ph.D. Dissertation, Michigan State University, East Lansing, MI
1988.

Demas, J.N. Excited State Lifetime Measurements Academic Press: New York,
1983.

Ware, W.R. in Creation and Detection of the Excited State vol 1A, A.A.
Iamola, Ed. Marcel Dekker, Inc.: New York, 1971, chapter 5.

O'Connor, D.V.; Phillips, D. Time-Correlated Single Photon Counting
Academic Press, Inc: London, 1984.

Fleming, G.R. Chemical Applications of Ultrafast Spectroscopy Oxford
University Press: New York, 1986.

Comission on Photochemistry, Pure Appl. Chem. 1990, 62, 1632.

Desilets, D.J.; Coburn, J.T.; Iantrip, D.A.; Kissinger, P.T.; Lytle, F.E. Anal.
Chem. 1986,58, 1123.

Desilets, D.J.; Kissinger, P.T.; Lytle, F.E. Anal. Chem. 1987, 59, 1830.

 

 

oo

. Birks, 1.8.;

19. Lackzo. 6.:
J.R.Rev. Sci

30. Technical Li
21. Chen,R.F.

33. Lampert, R
Mach. S.R

J
«4.)

. . O'Connor,

 

 

 

28
Birks, J.B.; Little, W.A. Proc. Phys. Soc. 1953, 66(A), 921.

Lackzo, G.; Gryczynski, 1.; Gryczynski, Z.; Wiczk, W.; Malak, H.; Lakowicz,
J.R.Rev. Sci. Instrum. 1990, 61 2331.

Technical Literature, Programmed Test Sources, Littleton, MA.
Chen, R.F. Anal. Biochem. 1974, 5 7, 593.

Iampert, R.A.; Chewter, L.A.; Phillips, D.; O'Connor, D.V.; Roberts, A.J.;
Meech, S.R. Anal. Chem. 1983, 55, 68.

O'Connor, D.V.; Meech, S.R.; Phillips, D. Chem. Phys. Let. 1982, 88, 22.

 

 

 

 

Popularly :
Timing (SPT) is t
of electroniCélll.V ‘
described in chapt
and Thomas [3] ‘0
scintillators. Late
decay data [3.4].
instrumentation ar
excited-state deca
Information on the
devoted exclusively
fluorescence deca)
attention must be p
0n this level. Su
excited-state decays
insmiment was cor
C0"mound that has .
existing system tha
cOmPOnent Was de:

Instniment more vet

CHAPTER 3: SINGLE—PHOTON TIMING

Popularly known as time—correlated single—photon counting, Single-Photon
Timing (SPT) is the most widely used technique for recording the fluorescence decay
of electronically excited species. It is generally agreed that the technique, briefly
described in chapter two, was developed independently by Koechlin [1] and Bollinger
and Thomas [2] to study the temporal characteristics of the pulses of light generated in
scintillators. Later, biochemists used the method to acquire molecular fluorescence
decay data [3,4]. Since its introduction many improvements have been made in
instrumentation and methodology, providing a means to study fast and complex

excited-state decays. Several reviews have appeared in the literature [5-8].

 

Information on the technique can also be found in various books [9—13] including one
devoted exclusively to the technique [14]. State—of-the—art instrumentation can record a l
fluorescence decay having a lifetime of less than 20 picoseconds [15]. Careful
attention must be paid to seemingly insignificant details to achieve temporal resolution
, on this level. Such items and concerns for accurately recording multi-exponential
excited-state decays are discussed at appropriate points in this chapter. A modern SPT
i instrument was constructed. This instrument can record the fluorescence decay of a
I compound that has an excited-state lifetime less than 50 picoseconds. It is similar to an
' existing system that is restricted to recording only very fast fluorescence decays. A
3 component was designed and fabricated to remove this limitation, making the new
: instrument more versatile than many modern SPT instruments. After a brief discussion
of the theory of the measurement, the instrumentation used in state-of—the-art SPT is
described, and the restrictions imposed by various instrumental configurations are
discussed. This is followed by a description of typical excited—state lifetime

determination. Example data from the constructed instrument is presented, and

29

 

 

 

application 0f the

quenching is dis“

The basic
rationale behind l
the sample is an 2
This probability d
measurements of t
emitted photon. l
excited-state deca
These restrictions
photocathode of ti

collisions will resr

As the stat
[5, 17], only a bi

given here.

At each cy

photons win strike
distribution, and i:

Where m is the at
“am There is ti

 

 

30

application of the SPT technique to the Hieftje-Haugen method [16] of correcting for
quenching is discussed.

THEORY

The basic measurement scheme was briefly described in chapter two. The
rationale behind the method is that the probability distribution of photons emitted by
the sample is an accurate representation of the excited state decay of the fluorophore.
This probability distribution is obtained by summing the results of a large number of
measurements of the time difference between an excitation flash and the detection of an
emitted photon. Certain restrictions must be taken into consideration to assure that the
excited-state decay is accurately represented by the acquired probability distribution.
These restrictions arise from the nature of the probability of photons impinging on the
photocathode of the detector following an excitation flash, and the probability that these

collisions will result in the emission of electrons from the photocathode.

As the statistics of the SPT technique have been thoroughly discussed elsewhere
[6, 17], only a brief condensation following Pfeffer et al. [18] and Ware [9] will be

given here.

At each cycle of the excitation source, we can define the probability, P,, that n
photons will strike the photocathode of the detector. This probability follows a Poisson
distribution, and is given by:

r'nn e'm

P (3-1)

‘ - n!
here m is the average number of photons that strike the detector after an excitation

ash. There is then the probability P2 that these n photons striking the photocathode

 

 

 

 

 

sill generate x phi

where e is the c
electrons will be g

of P1 and P2 sumr

If an electron is gr

rate R for a source

R

The detector count

photocathOde per u

Hence the fluores
between the detect
Ph0tons to Source
below one percent.
distorted in Pmpor
minor diSlortion. .-
“Homscence PhOtor
upper limit of On
experiments are co

Innmm

 

31

will generate x photoelectrons. P2 follows a binomial distribution, and is given by:

n!
P2 ‘ x!(n-x)!

«bx (1-¢1""‘ (3-2)
where ¢> is the quantum efficiency of the photocathode. The probability that x
electrons will be generated following each excitation cycle is then the joint probability
of P1 and P, summed over all possible values of n:

°° (om)x e‘m¢

P = 2 P,P, = x! (3-3)

 

If an electron is generated, one can assume that a count will be recorded. The count

rate R for a source repetition rate of v is:

R=V(1'Px-O) =v(1-e"“¢l =v m¢_.(Ln.2¢l+...] (3'4)

The detector count rate will be proportional to the number of photons incident on the
photocathode per unit time when

2
map > > (Jr—“2b; (3-5)

Hence the fluorescence should be weak such that several excitation cycles occur
between the detection of two fluorescence photons. In practice the ratio of detected
photons to source flashes per unit time is the monitored parameter and is often kept
below one percent. It has been shown [7] that the recorded fluorescence decay will be
distorted in proportion to this ratio, and a ratio of up to five percent will cause only
minor distortion. As the acceptable level of distortion decreases, so must the ratio of
fluorescence photons to source flashes per unit time. It is commonly accepted that an
upper limit of one or one-half percent is suitable for most work. In practice
experiments are conducted at much lower ratios, especially when high repetition rate

lasers are used.

Before dis
considerations cor
figure by which
referred to as the
exists, it is gen
instantaneous sigr
temporal jitter of
limit of excited-st
determine a lifeti
there is some dig
COmponents of a
Aside from the if

The Ultimate dete

 

sensitivity, In 8]
sufficient data {0
efficiency at whic
Wattleters SUCha
illuminated and ‘

instmmenm detai

An aCCUra
as a means of j‘
deconvolute the d
imminent Mp0“
decay. This dis

 

 

 

32

THE INSTRUIVIENT RESPONSE

Before discussing the instrumentation used in modern SPT, a few general
considerations concerning instrument performance should be mentioned. The primary
figure by which SPT instruments are judged is the instrument response, sometimes
referred to as the instrument response function (IRF). Although some disagreement
exists, it is generally reported as the FWHM of the instrument's record of an
instantaneous signal such as Raman scattering. It is essentially the rms sum of the
temporal jitter of each component in the instrument. It defines time zero and the lower
limit of excited—state lifetimes that can be determined. With proper techniques one can
determine a lifetime that is roughly one-half of the instrument response. Here again
there is some discrepancy on this issue. Some workers claim an ability to resolve
components of a fluorescence decay one fifteenth of the instrument response [19, 20].
Aside from the instrument response, few other performance parameters are reported.
The ultimate detectability of single photons defies the common definition of the term
sensitivity. In SPT, sensitivity is essentially a measure of the time needed to acquire
sufficient data for fitting the decay of a given sample. Hence it approximates the
efficiency at which photons are collected and detected when the figure is normalized for
parameters such as source power, sample quantum yield, and the volume of the sample
illuminated and viewed. Sensitivity as such is almost never reported. Rather all

instrumental details are reported.

An accurate measurement of the instrument response is of importance not only
as a means of judging the performance of the instrument, it is also necessary to
deconvolute the distortion of the fluorescence decay caused by the instrument. A finite
instrument response, in SPT or another technique, will distort the acquired fluorescence

decay. This distortion is minor for the case of a fluorescence decay that is long

 

 

 

 

 

 

 

compared to thC

response is not it

is discussed towa

A block
accompanying di

configuration dej

Although
have been con
research gmUps,
Often the majori
limiting compo
necessafily the t
SPT apparatus
Clear, and the n.
complete listing

given in Appenr

The hr:
the hght 80W
hOWevCr mme

resfleets. The

33

compared to the instrument response, but becomes important when the instrument
response is not negligible compared to the lifetime of the fluorophore. Deconvolution

is discussed toward the end of this chapter in a section on data analysis.

INSTRUMENTATION

A block diagram of a SPT apparatus was shown in Figure 2-3. The
accompanying discussion was intentionally not highly detailed because the instrumental

configuration depends on several factors.

Although a few commercial systems exist, most of the SPT instruments in use
have been constructed from home-built and commercial components by various
research groups. Each component affects the instrument response to some degree, but
often the majority of the temporal jitter is due to a single component. Historically the
limiting component has been either the light source or the detector. This is not
necessarily the case with modern instrumentation. In this section each component of an
SPT apparatus is described, its appropriate configuration in the instrument is made
clear, and the nature and extent of its effect on the instrument response is discussed. A
complete listing of the components of the SPT instrument constructed in this work is

given in Appendix 1.

Light Sources

The first consideration in any instrument used for fluorescence spectroscopy is
the light source. SPT is no exception. Of course a perfect light source does not exist;
however some of the light sources in use today approach ideal behavior in some

respects. The ideal light source for SPT would emit a highly collimated, controllable,

 

 

high repetition trai
frequencies of ligh
light is not as criti
the absolute tempt
the shortest pulse
instruments use :
disadvantages, an

instrument, dictatr

The flash
generate nanosecr
PulSt results fror
Other 821565 have
below atmOSpher
choice of gas an
most dependent r
the tempom p“
duration is norm;
totlowed by a 51
thorough Mice;

and Phillips [14]

34

high repetition train of infinitely reproducible light pulses that contain a broad range of
frequencies of light or can be tuned to a specific frequency. The width of the pulses of
light is not as critical as the reproducibility. In the light sources manufactured to date,
the absolute temporal jitter decreases with pulse width. Hence the interest in obtaining
the shortest pulse width sources. While early instruments used a flashlamp, modern
instruments use a laser. Each type of light source has inherent advantages and
disadvantages, and certain characteristics, when coupled with other components in the

instrument, dictate different instrumental configurations.

The flashlamps used in the past, and still used in some instruments today,
generate nanosecond pulses of light at a repetition rate in the kilohertz range. The light
pulse results from a high voltage discharge in a gas such as hydrogen or deuterium.
Other gases have also been used. Lamps have been constructed that operate above and
below atmospheric pressure. The spectral distribution of the lamp depends on the
choice of gas and pressure. The pulse width also depends on these factors, but it is
most dependent on the repetition rate. Gating has also been investigated. In general

the temporal profile of the pulses generated by these sources is not smooth. The
3 duration is normally a few nanoseconds, and it is not uncommon for the fast pulse to be
followed by a slowly decaying tail that may last for several tens of nanoseconds. A
thorough review of the flashlamps used in SPT can be found in a book by O'Connor

and Phillips [14] .

Mode-locked lasers [21] and cavity dumped, synchronously pumped dye lasers
[22] have been used in SPT. Unit picosecond or shorter pulses are obtainable with
these sources, although pulses of five picoseconds are more common. The temporal
profile is quite clean provided the cavity lengths of the pump and dye laser are exactly

matched. A mismatch can cause "satellite" pulses on both sides of the primary pulse.

 

 

 

 

As these can be a
be monitored and
rate can be chan
Because the ligl
monochromatic, 2
provided that a r
tuneable over the
dye laser. Natu
typical value f0:
wavelengths is

"frequency doub
collimation are t

sources used in S

In contra
Slam, a laSl

inStrumentation 3
required to mom
require cOtttinut

However, the C‘

 

35

As these can be a source of systematic error in an SPT measurement, the output must
be monitored and appropriate steps must be taken to eliminate satellites. The repetition
rate can be changed through the appropriate control on the cavity dumper driver.
Because the light source is a laser, the spectral output is necessarily highly
monochromatic, allowing one to detect emission within a few nanometers of excitation
provided that a monochromator of sufficient resolution is used. This output is also
tuneable over the limits of the laser and saturable absorber dye combination used in the
dye laser. Naturally this range varies with the dye combination used; however, a
typical value for this range is 40 nm. Another range of the available output
wavelengths is possible through second harmonic generation (SHG); so—called
"frequency doubling". Other properties of laser light such as the high degree of
collimation are helpful. O'Connor [14] and Fleming [12] have reviewed laser light

sources used in SPT.

In contrast to a flashlamp light source that can be purchased for less than

$10,000, a laser source can cost in excess of $150,000. In addition, the

instrumentation required to monitor a laser source is much more expensive than that
required to monitor the output of a flashlamp. Indeed, flashlamp—based systems do not
require continuous monitoring and more closely approximate a tum-key system.
However, the expense and maintenance requirements of a laser-based system can be
justified if the excited state decay of the system of interest is too fast for study with a
flashlamp system. Of course the successful study of a chemical system may require
multiple excitation wavelengths, and those needed may lie outside the tuning range of a
single laser dye/saturable absorber combination. This can necessitate changing the dyes
during the course of the experiment - a most inconvenient task with some laser systems.
Providing the needed excitation wavelengths can be a trivial matter for a flashlamp-

based SPT apparatus. However, mode-locked laser sources routinely operate at

 

 

 

 

 

repetition rates in
and below. This
laser based SPT a;

Other light
nitrogen and excir

they are rarely use

A mode-lo
SPT instrument is
time needed to acr
locked Nd:YAG 1

Specific componer

The select
made with only ti
and emission. 3]
designed Cetls, Ft
atIaCltment to v;
considemiOns ho]

resolution in the p

The Photo

response of this

 

.MQLH‘” ‘:.; _.__

36

repetition rates in the megahertz range while flashlamps operate in the kilohertz range
and below. This greatly reduces the time required for a given measurement with a

laser based SPT apparatus compared to a flashlamp based system.

Other light sources such as synchrotrons and low repetition rate lasers (e. g.
nitrogen and excimer) have been used. However they will only be mentioned here as

they are rarely used.

A mode-locked laser system was chosen for the light source in the constructed
SPT instrument because of the high time resolution possible with it, and the reduced
time needed to acquire a fluorescence decay. The specific laser system used is a mode-
locked Nd:YAG laser and a synchronously pumped, cavity dumped, dye laser. The
specific components are listed in Appendix 1.

Sample Cells

The selection of the sample cell in most fluorescence experiments is normally
made with only the criteria of high transmittance in the spectral regions of excitation
and emission. Special sample requirements are usually accommodated with specially
designed cells. For example, air tight cells fitted with ground glass joints are used for
attachment to vacuum lines for handling air sensitive compounds. The same
considerations hold for SPT, but one should consider additional factors when temporal

resolution in the picosecond range is needed.

The photon transit time through a standard one cm cuvette filled with an
aqueous solution is roughly 40 picoseconds. When one is striving for an instrument

response of this magnitude, one needs to take this into account since the detected

 

 

 

 

fluorescence photc
of the exciting lig
originate from a

effectively doubli:
collection optics St
the contribution t
reflected exciting I
to use a smaller cr
quartz, The cell

Window and is at
collected through

light. Hence, this
reduce reflected s
Minimize the pop“
the laser light awe
Whom the cell

mhimized.

Although 1'

faMilled cells. 1
0f the cell is View.

addition to cells fa

37

fluorescence photon could originate from a molecule excited anywhere along the path
of the exciting light pulse. Consider also that the detected fluorescence photon could
originate from a molecule excited by light reflected from the wall of the cuvette,
effectively doubling the contribution to the instrument response. Adjustment the
collection optics so that only a small area of the solution is viewed will help minimize
the contribution to the instrument response from this source, but the problem of
reflected exciting light remains. A solution suggested by Steiner and coworkers [13] is
to use a smaller cuvette (e. g. four mm path length) having two adjacent walls of black
quartz. The cell is oriented so that the exciting light enters through one transparent
window and is absorbed when it reaches the opposite wall. Fluorescence is then
collected through the other transparent window which is 90 degrees to the exciting
light. Hence, this approach provides a way to minimize the photon transit time and to
reduce reflected source light and sample fluorescence. One may also use masks to
minimize the volume of the solution viewed and a sample cell geometries that reflects
the laser light away from the region of the solution viewed. However it is difficult to
position the cell such that reflected source light and reflected fluorescence are

minimized.

Although the Special cuvette used by Steiner and coworkers [13] minimizes both
sources of temporal broadening, funds could not be obtained to purchase these specially
fabricated cells. Instead the collection optics were arranged so that only a small region
of the cell is viewed. This way standard 1.0 cm fluorescence cuvettes could be used in

addition to cells fabricated for use on a vacuum line.

 

 

 

 

In order ti
insure that only f
scattered laser ra
decay will be dist
be skewed towarc
be added if the or
instruments, and
filter, and a sma
hght rejection wt

of the Czemy-Tu

In the ca:
solution viewed i
fluorescent light
through the m0“
because rays tra
The differenoe it
he gating. Thi
ofa SPEX In (in
Solid and a We
gmfihg- The fiy

hindel the mono

38

Wavelength Selection

In order to record the decay of the fluorophore accurately, it is necessary to
insure that only fluorescent light reaches the detector. If stray light originating from
scattered laser radiation and/or Raman scattering reaches the detector, the recorded
decay will be distorted. If the contribution from these sources is small, the decay will
be skewed toward a shorter lifetime. A false, fast, second component to the decay will
be added if the contribution from these sources of stray light is significant. Many SPT
instruments, and the one constructed in this work, use a longpass filter, an interference
filter, and a small Czemy-Turner single monochromator to achieve the needed stray
light rejection while maintaining a reasonable throughput. The temporal characteristics

of the Czemy-Turner monochromator are, however, not ideal for SPT.

In the case of the sample cell, the photon transit time through the volume of
solution viewed increases the instrument response. When wavelength selection of the
fluorescent light is accomplished with a monochromator, the transit time spread of light
through the monochromator is the critical parameter. The transit time. spread arises
because rays traced through the monochromator have different optical path lengths.
The difference in optical path length increases as the ray's path approaches the edge of
the grating. This is illustrated on the next page in Figure 3-1. The figure is top view
of a SPEX Industries model 1681 Czemy-Turner monochromator (f/n = 3.9) with a
solid and a dashed line representing rays that strike the extreme edges of the diffraction
grating. The figure was generated with a commercial software package [23] used to
model the monochromator. As the monochromator is adjusted to pass wavelengths
from the blue to the red region of the spectrum, the grating angle increases and hence

the transit time spread increases. One means to minimize the transit time spread is to

 

 

 

Figure 3-1. Moe
strike the left and

Plateamask int
In essence one 1
Characteristics. ‘

tabulated in Table

TABLE H. Ti
INCREMENTS l

WAVELENGi

 

 

Figure 3-1. Model of a SPEX 1681 Czemy-Tumer monochromator showing rays that
strike the left and right edges of the grating. The wavelength of the rays is 500 nm.

place a mask in the monochromator such that only the central grooves are illuminated.
In essence one sacrifices the monochromator's throughput to improve its temporal
characteristics. The transit time spread of the SPEX model 1681 monochromator is

tabulated in Table 3-1 for full grating illumination and for a mask that blocks all but 10

TABLE 3—1. TRANSIT TIME SPREAD (ITS) FROM 300 TO 700 NM IN 100 nm
INCREMENTS FOR A SPEX 1681 CZERNY-TURNER MONOCHROMATOR.

 

'I'I‘S (ps) “ITS (ps)
WAVELENGTH (nm)
Full Grating Width (48mm) 10 mm Exposed Width
300 48 12
400 63 16
500 79 20
600 96 24

700 113 28

 

 

 

mm of the gratin

The mask reduces

Other me:
wavelength select
additional filters,
the Czemy-Tume
the filter approac
must be taken a
compensator pla
allllroach, and it
Optical componei
Selection of a ms

the Contribution r

In theory
Spread of zero. ‘
monochromator
Holographic m
Page 41 in Fig
aPllroach has the
$th light rejec
nuptes that st

Particulate matte

 

 

 

40

mm of the grating's width. Values are given for wavelengths from 300 to 700 nm.

The mask reduces the transit time spread by a factor of four.

Other means of reducing the contribution to the instrument response from the
wavelength selection device include removing the monochromator and replacing it with
additional filters, placing compensator plates in the monochromator [24], and replacing
the Czemy-Tumer single monochromator with a monochromator of another design. If
the filter approach is chosen, additional steps to achieve the needed stray light rejection
must be taken and a means to vary the throughput must be included. The use of
compensator plates achieves a slightly better transit time spread than the mask
approach, and it only reduces the throughput by the loss that results from the additional
optical components. It is also more difficult to implement than the mask approach.
Selection of a monochromator of another design provides a better means of minimizing

the contribution of the wavelength selection system to the instrument response.

In theory, a subtractive dispersion double monochromator has a transit time
Spread of zero. This is because the optical path length for all rays that pass through the
monochromator are identical. The Optical layout of such a device, an American
Holographic model DBlO-S subtractive dispersion monochromator [25], is shown in on
page 41 in Figure 3-2. In addition to its desirable temporal characteristics, this
approach has the advantage that additional filters are not needed to achieve the level of
stray light rejection required for most work. However, filters may be needed for

samples that strongly scatter source light, such as those that contain suspended

particulate matter.

 

 

 

 

 

 

 

Figure 3-2. F1
dispersion double

Had it be
have been chosen
unit that is suitab

that time

The instm]
hhtime 0f detect
train: time Sprem
of interest is the 1
”hhdttatioh, a 1
hanow PHD, Tl

mm“ phOIOmuI‘

 

41

 

GRRIING SERVING

T "r‘

 

__ e !

[_H__—___IIRROR MIRROR

Figure 3-2. Floor plan view of an American Holographic DBIO-S subtractive
dispersion double monochromator. From [25].

Had it been an option, a subractive dispersion double monochromator would
have been chosen for the constructed instrument. Unfortunately, the only commercial
i unit that is suitable for SPT, the American Holographic DBIO-S, was not available at

that time.
Detectors

The instrument response is often tied to the risetime of the detector signal. The
risetime of detectors that contain an electron multiplier is largely a function of the
transit time spread (ITS) of the pulse of electrons as it passes through the device. Also
of interest is the pulse height distribution (PHD) of the detector. Similar to the ITS

consideration, a narrow instrument response is obtained with a detector that has a

 

narrow PHD. The most widely used detectors in SPT instruments are conventional,

side-on photomultiplier tubes (PMTs) and microchannel plate photomultiplier tubes

 

 

 

 

 

 

 

 

 

(MCP-PMTS). 1

specific detectors

For a con
2.2 ns [26]. Th1
and is not strongl

response of one It

Although
information whe-
devices has allow
PaCkRged with a
lost, but the sign
PMT that Contain
The risetime is n

PMT are six micr

The PHD
Emmy low a;
However a peak
gain saturation [2
PHD’ and hence :
MCP‘PMT “w
”m is grater
one is Mommg 1

Other componenn

42

(MCP-PMTs). Both are available in the instrument constructed in this work, and the

specific detectors used are listed in Appendix 1.

For a conventional side-on PMT such as the Hamamatsu R928, the risetime is
2.2 ns [26]. The PHD of a conventional PMT follows an exponential function [27],
and is not strongly dependent on the applied voltage. These factors yield an instrument

response of one to two nanoseconds when a conventional PMT is used.

Although microchannel plates were originally designed for preserving spatial
information when amplifying a photocurrent, the extremely fast risetime of these
devices has allowed a significant reduction in the instrument response for SPT. When
packaged with a common anode, the spatial information of the micro-channel plate is
lost, but the signal bandwidth is preserved, yielding a risetime of 220 ps for an MCP-
PMT that contains two micro—channel plates composed of twelve micron channels [26].
The risetime is reduced further to 150 ps when the micro-channel plates in the MCP-

PMT are six microns in diameter.

The PHD of MCP-PMTs is strongly dependent on the applied voltage. At
relatively low applied voltages, the PHD is exponential like a conventional PMT.
However a peak in the PHD appears as the bias voltage is increased. This is due to
gain saturation [27, 28] and is illustrated on page 43 in Figure 3-3. To obtain a narrow
PHD, and hence a narrow instrument response, one needs to apply a bias voltage to the
MCP—PMT near the device's maximum rated voltage. Naturally the number of dark
counts is greater at higher applied voltages, but this is generally not a problem when
one is recording the decay of moderately efficient fluorophores. By carefully adjusting

other components in the instrument to benefit from the narrower PHD at high bias

 

 

 

\

#—|

CM! 31mm.

 

 

m

Figure 3-3. Puls
The bias voltage

voltages, an inst
micron MCP-Ph
response of 20 ps

The riseti:
as fast as that for
fluorescence in
efficiency in the
the SPeCtrum is s
when they are il
a‘ilanche photoc

The Circuit used 1

These phc
interest emits m
beyond the M
avalanche 0f do
”be counterparts
dark counts resu'

43

m lS/CHNDEL
ICH
CENTS/CHANG.

PlLSEl-EIG'IT mm PLLSE PEIGHT

Figure 3—3. Pulse height distribution of an MCP-PMT as a function of applied voltage.
The bias voltage increases from left to right.

voltages, an instrument response of 50 picoseconds or less is possible when a 12
micron MCP-PMT is used [27]. A six micron MCP-PMT can yield an instrument

response of 20 ps [15].

The risetime of the current generation of silicon avalanche photodiodes is almost
as fast as that for the MCP-PMT. These devices are not as widely used as detectors of
fluorescence in SPT because they are generally optimized for maximum quantum
efficiency in the infrared. However, their quantum efficiency in the visible region of
the spectrum is sufficient for them to be used as temporal reference detectors in SPT,
where they are illuminated with a comparatively large number of photons. The silicon
avalanche photodiode used in the constructed SPT apparatus is listed in Appendix 2.

, The circuit used to bias it is also described therein.

These photodiodes can also be used as sample detectors when the fluorophore of
, interest emits near the red region of the spectrum, or in the infrared [29]. When biased
' beyond the breakdown voltage, a single photon can, as the name implies, trigger an
i avalanche of electrons yielding single photon detectability. Similar to their vacuum
tube counterparts, an avalanche can also result from dark noise. In this case however,

dark counts result from leakage of an electron across the photoactive junction. This

 

 

 

noise source can

reduce dark noise

The detec
change in the ten
aPMT or MCP.
spread With a ch:
difference in kins
photons of differ
work function 0
cause the ejectior
the difference b
Hence, the ldnet
than the kinetic e
Shift in time resp
an MCP-PMT. :
color Effect of 1
Pitosecond range
”its [28].

The imp“

We]- the Wt [61'
However) thCSe

tit
ese component

 

 

 

44

noise source can be reduced by cooling the detector, analogous to cooling a PMT to

reduce dark noise.

The detectors mentioned above exhibit a color effect. This is observed as a
change in the temporal characteristics of the detector with a change in wavelength. For
a PMT or MCP-PMT, one observes a change in the transit time and/or transit time
spread with a change in wavelength. It is generally agreed that this effect is due to a
difference in kinetic energy of photoelectrons emitted by the photocathode for incident
photons of different wavelengths. When a photon having an energy greater than the
work function of the photocathode material collides with the photocathode, it may
cause the ejection of an electron. The kinetic energy of the ejected electron will equal
the difference between the photon's energy and the photocathode work function.
Hence, the kinetic energy of an electron resulting from a blue photon will be greater
than the kinetic energy of an electron resulting from a red photon. This effect causes a
shift in time response of a PMT of less than one nanosecond [10]. The effect is less for
an MCP—PMT. Silicon avalanche photodiodes also exhibit a color effect. Although the
color effect of MCP-PMTs is not well documented, it is considered to be in the
picosecond range for a wavelength shift of 100 nm, much less than that observed for

PMTs [28].

Electronics

The impressive improvements in the temporal resolution of SPT instrumentation
over the past ten years is largely a result of progress in light sources and detectors.
However, these deve10pments would not be possible Without progress in the other
components of an SPT instrument, and the careful incorporation and configuration of

these components in the instrument. Each of the components must be properly

 

 

 

 

 

 

configured and it
aspecific compo
These are discus

neglected compo

Although
instrument are he
can, if ignored,
eXpected to accu:
mate the compo]
conducted throng
Signal (9.8. one :
For example, a
frequencies in ex

anode-rocked, s

In order -
tables and 00m“
°°mp0nent of the
summit bandwj
c0nttectOrs Shoult
PM] comm m
adequtte martial
tame arises beta:

“Olin copper or

4S

configured and its placement in the instrument must be arranged such that limitations of
a specific component do not adversely affect the overall performance of the instrument.
These are discussed in this section. However the importance of simple, yet often

neglected components is discussed first.
CABLES AND CONNECTORS

Although the coaxial cables and connectors used to conduct the signals in a SPT
instrument are not active electronic components, they are part of the overall system and
can, if ignored, adversely affect the instrument response. If an electronic system is
expected to accurately measure a given quantity, the cables and the connectors used to
mate the components must have sufficient bandwidth such that no part of the signal
conducted through the system is attenuated. A significant portion of a rapidly rising
signal (e. g. one approximating a square wave) is contained in higher order harmonics.
For example, a signal having a risetime of one nanosecond can contain harmonic
frequencies in excess of one GHz. Consider that the harmonic content of the output of

a mode—locked, synchronously pumped dye laser extends beyond 50 GHz [30].

In order to maintain the integrity of the fast signals in a SPT apparatus, the
cables and connectors used must have bandwidths greater than the highest frequency
component of the signal. Common RG 58/U coaxial cable and BNC connectors have a
sufficient bandwidth for PMT based systems. However higher bandwidth cables and
connectors should be used in MCP-PMT systems. This is especially true if the MCP-
PMT contains microchannel plates composed of six micron channels. A more than
adequate coaxial cable is semi-rigid coax, which is used for microwave signals. The
name arises because rather than having a braided shield, the outer jacket of this cable is

1 solid copper or stainless steel tube. Its bandwidth extends to several GHz (e. g. 70 dB

 

 

 

 

 

 

 

per 100 feet @ l
immunity of the
0.250“ diameter
Even though it i

instrument betar

As ment
risetime system:
SMA connector

SMA connector

The sigr
must be Mpli
“Imam factc
However an ar
MCP'PMT, I
attilahle at v
inStfitment, the
The faster six
a"‘Dtit‘ier hm
bandwidth, the
exceed the inpt

Itich

re the inpl

46

per 100 feet @ 10 GHz) [31]. Also, its shielding is absolute, thus improving the noise
immunity of the system. Common sizes of semi-rigid coax are 0.085 ", 0.141", and
0.250" diameter. The bandwidth generally increases with the diameter of the cable.
Even though it is the lowest bandwidth semi-rigid cable, 0.085" cable was used in this

instrument because it is the most easily shaped.

As mentioned above, connectors for coaxial cable are also important for fast
risetime systems. An excellent connector for SPT with an MCP-PMT detector is the
SMA connector. The bandwidth closely approximates that of semi-rigid coax, and

SMA connectors are used for the semi-rigid cable in this instrument.

AMPLIFIERS

The signal from the detector in the fluorescence channel (PMT or MCP-PMT)
must be amplified before it reaches the discriminator. Again, bandwidth is an
important factor. For PMTs a bandwidth of 500 MHz is more than sufficient.
However an amplifier having a bandwidth of 2 GHz is needed for a twelve micron
MCP-PMT. Fortunately, low noise amplifiers having bandwidths of 2 GHz are
available at very reasonable cost. For example, the amplifier chosen for this
instrument, the model ZFL-2000 2 GHz amplifier from Minicircuits, is priced at $200.
The faster six micron MCP-PMTs require faster amplifiers. For these detectors, an
amplifier having a bandwidth of 4 to 6 GHz should be used. Regardless of its
bandwidth, the amplifier should be chosen such that its maximum output does not

exceed the input range of the discriminator.

It is common practice to insert a one dB fwdthrough attenuator immediately

before the input of the amplifier in the fluorescence channel. This provides a constant

 

 

 

 

 

 

 

DC leakage for

detector and the

Because
pulse produce a

not needed, and

The cor
This device ha
threshold val
NIM negative
removed, Tl
inStlument, ar
made Properly
of discriminat
ttd most imp

 

 

47

DC leakage for the capacitor that is the short length of coaxial cable between the

detector and the amplifier.

Because the silicon avalanche photodiodes used to detect the temporal reference
pulse produce a signal that is sufficient for the discriminator, an amplifier is generally

not needed, and one is not used in this instrument.

THE CONSTANT FRACTION DISCRIMINATOR

The component in the instrument following either detector is the discriminator.
This device has three functions in an SPT instrument. It not only blocks signals below
a threshold value, it also generates a NIM negative logic signal from the analog input.
NIM negative logic is essentially emitter coupled logic (ECL) with the DC offset
removed. These logic levels are recognized by the next active component in the
instrument, and the signals need to be in this form for the time measurement to be
made properly. In order to achieve high speed and low temporal jitter, a special type
of discriminator known as a constant fraction discriminator (CFD) is used. The third
and most important function of the CFD is to lessen the dependence of signal amplitude

on the temporal output.

The object of a CFD is to produce a logic pulse that is exactly timed with
respect to the analog input. The method of accomplishing this task should be
insensitive to the amplitude of the input. The constant fraction technique [32, 33]
realizes this by splitting the unipolar input pulse. The signal is attenuated and inverted
in one of the paths and delayed in the other. The delay is matched to the risetime of
the signal such that the attenuated signal is at its peak at the desired fraction of the

delayed signal. The two signals are then recombined. The zero-crossing of the

 

 

resulting bipolar
A conventional
constant fractior

the amplitude of

In order
before the com
proper length of
coaxial cable p2
risetime of the
modified. This
455. Although
300 PS. Hence
MCP-PMT. T1
0lititttize the C]

Concerning the t

The Ter
discrete, some c
been ObSCrVed V
chthnel, the M
can generate Sp
insure that inpt

Changes that thi

 

 

48

resulting bipolar pulse coincides with a constant fraction of the of the original signal.
A conventional zero-crossing detector generates a logic pulse corresponding to a
constant fraction of the original signal. The output is not strongly subject to changes in

the amplitude of the input with this technique.

In order to minimize walk (time dependence on amplitude) in a CFD, the delay
before the comparator must be carefully matched to the input pulse risetime. The
pr0per length of this delay is best determined experimentally. Most often the delay is a
coaxial cable patched between two connectors on the front panel of the CFD. If the
risetime of the signal is less than the internal delay of the CFD, the CFD must be
modified. This was the case with the CFD used in this instrument, the Tennelec model
455 . Although it is the fastest CFD commercially available, its internal delay is about
300 ps. Hence the unit from the factory must be modified if it is to be used with an
MCP-PMT. The suggested modifications from the manufacturer [34] were followed to
optimize the CFD used in this instrument for the 12 micron MCP—PMT. Information

concerning the modifications needed can also be found in the scientific literature [35].

The Tennelec model 455 has four independent channels. Although they are
discrete, some cross-talk does occur between channels [13]. This phenomenon has also
been observed with other multi—channel CFDs [15]. When a signal is received in one
channel, the walk in the others is degraded for approximately 40 nanoseconds. This
can generate spurious signals in the system if one does not take apprOpriate steps to
' sure that inputs to the CFD are sufficiently separated in time. The configuration

hanges that this imposes are fully discussed later in this chapter.

 

 

 

 

 

 

The fun
difference betw
pulse is then di;
personal compu
is to count the 5
interest. The
oscillations are
and known. A]
make, present
measure the tim

alternate measu:

As ment
having an mm
the data channe
34. A fast,
theeived at the
neEattive logic 5
and hold Empli

the "m pair of

Conside
TAGS aVailable

Concerns a Ten

49

THE TIME-TO-AMPLITUDE CONVERTER

The function of the time-to—amplitude converter (TAC) is to encode the time
difference between the two data channels as a unipolar voltage pulse. This voltage
pulse is then digitized and stored by the multichannel analyzer housed in a laboratory
personal computer. The conventional strategy by which a time difference is measured
is to count the signal from a precision oscillator during the time between the signals of
interest. The time difference can be accurately determined if a large number of
oscillations are counted and the frequency of the precision oscillator is highly constant
and known. Although this technique is one of the most accurate measurements one can
make, present technology cannot provide oscillators and counters fast enough to
measure the time difference between the data channels in an SPT instrument. Hence an

alternate measurement strategy is used.

As mentioned above, the purpose of the TAC is to generate an analog voltage
‘ having an amplitude that is proportional to the time difference between the signals in
the data channels. This is accomplished by the scheme illustrated on page 50 in Figure
3-4. A fast, linear voltage ramp is started when a NIM negative logic signal is
received at the START input. The internal voltage ramp continues until a NIM
negative logic signal is received at the STOP input. This event also triggers a sample
and hold amplifier. A brief, unipolar pulse is then generated and the TAC resets for

the next pair of input signals.

Consider the signal throughput of a TAC. Because it is representative of the
TACs available and it is the TAC used in this instrument, the following discussion

concerns a Tennelec model 862. The output voltage range of the TAC is zero to 10

 

 

 

 

START

STOP

Figure 3-4. Me

volts. The time
“leh can be va
TAC reset time
When the time
throughput of at
This 1imitation i
that uses a TAC

so
Time—to—Amplitude Converter

 

 

TAC V
—o

_L RAMP ”NEE Time
T P
S 0 GENERATOR HOLD

 

 

 

 

 

 

 

 

 

 

Figure 3-4. Method of operation of a time—to-amplitude converter.

volts. The time base can be varied continuously from 25 us to 3 ms. The output pulse
width can be varied from 1 to 10 ms and can be delayed for 250 ns to 25 ms. The
TAC reset time ranges from 0.5 its to 5 us for the fastest to the slowest time base.
When the time required for all of these tasks is summed, a best-case instrument
throughput of about 500 kHz is realized. In practice this figure is in the tens of kHz.
‘ This limitation imposes a restriction on the overall configuration of an SPT instrument

that uses a TAC.

The throughput limitation of the TAC is generally not a problem when a
, flashlamp is used. However, it does impose a constraint when the instrument uses a
1 high repetition rate laser. When this is the case, the throughput limitation of the TAC
can be obviated if the fluorescence channel provides the START input to the TAC, so-
called reversed timing [36, 37]. The high repetition rate signal in the temporal
reference channel then serves as the STOP signal. This configuration does not distort

the acquired excited-state decay. It merely reverses the direction of increasing time in

 

 

the recorded d

instrument, a co

From thi
limiting factor ll
has been shown
are functions of
tested, an 0RTl
they measured a
kHz to 100 RH.
when one is str
constraint when
decays even if

fOiiOWlTlg [his It

The ittttplitude
multichannel an
tconvehhom :
this deVice to h
addresses in m‘
received) the p
”the is here,
number of ian
“timber. After

””mber Yields

51

the recorded data. It does, however, require an additional component in the

instrument, a coaxial cable delay.

From this discussion one might assume that the TAC conversion rate is not a
limiting factor in an SPT experiment. Although this assumption is valid to an extent, it
has been shown by Hallam and Imhof [38] that the linearity and calibration of a TAC
are functions of the conversion rate. The linearity and calibration of the TAC they
tested, an ORTEC 457, degraded with an increase in conversion rate. For example,
they measured a shift of 90 picoseconds when the conversion rate was increased from 1
kHz to 100 kHz. They recommended that the conversion rate be kept below 1 kHz
when one is striving for maximum temporal resolution. One should also observe this
constraint when one is studying chemical systems that exhibit complex excited—state
decays even if the species under study is long-lived. IUPAC recently recommended

following this restriction on acquisition rate for all measurements [39].

THE M ULTTCHANNEL ANALIZER

The amplitude of the unipolar pulse generated by the TAC is digitized by a
multichannel analyzer (MCA), and stored in memory. This device operates faster than
a conventional analog-to-digital converter. One can consider the analog input range of
this device to be partitioned into some number of equally spaced "channels" - actually
addresses in memory. A channel is essentially a counter. When an input pulse is
received, the pulse amplitude is determined, and the channel corresponding to that
voltage is incremented. After some number of input pulses, each channel contains the
number of input pulses that have had a pulse height corresponding to that channel
number. After a large number of excitation flashes, a plot of count versus channel

number yields a histogram, the outline of which is the summed result of the time

 

 

 

 

 

 

between excitatic

words the fluores

The temp
divided by the r
consideration the
of the TAC. Of
This is the case
voltage range is :
thAs, has varia

the MCA.

The const
insened in both c
are Separated by
that the excited.5
accomPlishes this
storm on page 5
sinrjhr to their h
u“ “em Splitter.
0the Which prov

CFD‘ The digita

52

between excitation and emission for all of the detected fluorescence photons, in other

words the fluorescence decay.

The temporal resolution of the acquired data is the time base of the TAC
divided by the number of channels in the MCA. One may also need to take into
consideration the analog input range of the MCA compared to the analog output range
of the TAC. Often an MCA's input range is less than the output range of the TAC.
This is the case for the MCA used in this instrument, a Nucleus PCA/8K. Its input
voltage range is zero to about 8.2 volts, and it has 8,196 channels (8K). It, like many
MCAs, has variable gain, actually a variable number of channels up to the maximum of

the MCA.

Delays and Relative Timing

The constraints on the configuration of the instrument require that delays be
inserted in both of the data channels. These delays must be placed such that the signals
are separated by at least 40 ns when they reach the CFD, but arrive at the TAC such
that the excited-state decay is properly recorded. An instrumental configuration that
accomplishes this was reported by Steiner and coworkers [13]. Their configuration is
shown on page 53 in Figure 3—5. The configuration of the instrument constructed is
similar to their instrument with one important exception. Each data channel begins at
the beam splitter. Laser light in the temporal reference channel is coupled into a fiber
optic which provides a delay before the pulse of light reaches the temporal reference
detector, a silicon avalanche photodiode. The signal from the photodiode travels
through a short length of 0.085 " diameter semi-rigid coax to one of the channels of the

CFD. The digital pulse generated by the CFD is used as the STOP signal for the TAC.

 

 

LIGHT
PULSE ‘
IN

Figure 3-5. Di:

Light not defler
harmonic gener.
from the 821mm
Optics - an f/n
Pltneconvex fr
aP‘llttl'tzation a
detected with th
semirigid coax
a"‘Ptil‘rer. Th
Channel of the 1
table delay befr
TAC is digitize

 

53

     
  
  

 

LIGHT
PULSE SAMPLE
IN
CULLECTIUN
PTICS

FIBER OPTIC
DELAY

 

SI AVALANCHE
PHDTUDIUDE

 

Figure 3—5. Diagram of the optimum instrumental configuration.

Light not deflected by the beamsplitter continues to the sample. If needed, a second
harmonic generator is inserted between the beamsplitter and the sample. Fluorescence
from the sample is collected, and directed into the monochromator by the collection
optics - an f/n = 2 quartz, plano—convex collecting lens and an f/n = 3.9 quartz,
plano-convex focussing lens which matches the flu of the monochromator. Filters and
a polarization analyzer may be inserted between the lenses. Fluorescence photons are
detected with the MCP-PMT. The anode current pulse travels through a short length of
semi—rigid coax to a 1 dB attenuator connected to the input of a 2 GHz, 22 dB
amplifier. The amplified signal is conducted through semi-rigid coax to another
channel of the CFD. The output from the CFD is delayed through a switched coaxial
cable delay before reaching the START input of the TAC. The pulse generated by the
TAC is digitized and stored in the MCA.

 

 

 

 

The len;
reaches the CF
This ensures th
the fluorescenc
over the fastest
the time range
CFD and the 7
occurs near the
time difference
limit of the TA

excitation and

This cc
for acquiring
nanOScconds.
deified, a me:
increase the le
Window, but ‘
that could be
f1“forties to
SWitched fiber
fluorophore u

the SWitched r

 

 

S4

The length of the fiber optic delay is such that the temporal reference signal
reaches the CFD 80 nanoseconds after the pulse of light is split at the beam splitter.
This ensures that the temporal reference signal is sufficiently separated from pulses in
the fluorescence channel at the CFD, provided the fluorescence has decayed completely
over the fastest setting of the TAC time base. The two signals are brought to within
the time range of the TAC by adjusting the switched coaxial cable delay between the
CFD and the TAC in the fluorescence channel. This delay is set such that time zero
occurs near the upper limit of the TAC timebase. Since the timing is reversed, short
time differences between excitation and emission produce TAC outputs near the upper
limit of the TAC timebase, and the TAC output decreases with increasing time between

excitation and emission.

This configuration works well for very fast fluorescence decays, but it is limited
for acquiring the decays of fluorophores that have lifetimes greater than five
nanoseconds. Because an ability to determine longer excited-state lifetimes was
desired, a means of removing this limitation was developed. One solution could be to
increase the length of the fiber optic delay. This way one can collect data over a larger
window, but temporal resolution would be reduced, thus limiting the fastest lifetimes
that could be determined A crude way around this would be to have a collection of
fiber optics to insert as optical delays. A more "user friendly" approach was chosen - a
switched fiber optic delay. This allows the user to change the Optical delay to suit the

fluorophore under study. Complete details concerning the design and construction of

the switched fiber optic delay are given in Appendix 2.

 

 

 

 

The deu
measurements.
response, and th
performed for e
response should
is by no means i

of an experiment

A time c:
TAC in a config‘
known, stepwise
START/STOP p
START/STOP p
START and ST(
resumed, and th
comm This is
Signals such as t
mm“ generate
the signal in the,
TAC's STOP in,
for the TAC~M C
the MC A. when
on kHz as men
Climatic“ Need I
should be done “

55

A TYPICAL SPT EXPERIMENT

The determination of an excited-state lifetime by SPT consists of three
measurements. These are a time calibration, a measurement of the instrument
response, and the acquisition of the fluorescence decay. A time calibration need not be
performed for each excited-state lifetime determination, but recording the instrument
response should be a routine part of the experiment. The order of these measurements
is by no means fixed, and one may, in fact, repeat one or two of them over the course

of an experiment.

A time calibration is performed by sending pairs of START/STOP pulses to the
TAC in a configuration where the time between START and STOP can be adjusted in a
known, stepwise fashion. The TAC output is recorded with the MCA for a series of
START/STOP pulses having a constant time difference. After a sufficient number of
START/STOP pulses have been acquired, the MCA is paused and the delay between
START and STOP is incremented by a known quantity. Acquisition by the MCA is
resumed, and the process is repeated until the temporal range of interest has been
covered. This is most easily accomplished by splitting a source of NIM negative logic
signals such as the output of one of the channels of the CFD or a properly adjusted
function generator. One leg of the split leads directly to the TAC's START input, and
the signal in the other leg is passed through a switched delay card before it reaches the
TAC's STOP input. If the time increment of the delay is known, the time calibration
for the TAC-MCA settings used is determined by a linear least squares fit of the data in
the MCA. When performing a time calibration, one should use a TAC conversion rate
of 1 kHz as mentioned in the discussion on the TAC. As mentioned above, a time
calibration need not be carried out for each excited-state lifetime determination, but it

should be done whenever the TAC time base is changed. Even if the TAC time base

 

 

has not changer
proper laborator

The acq
follow a few gt
the delays, the ‘
the data chasm:
time zero. Wl
range of the N.
delays are set,
decay is bracl
Preceding the
Channels at the
five lifetimes.
atte“utiltg to l
MCA 83in net
than simple d
lower Channel
can aChieve s

The e
kept below c
Section on thi
accuracy P08
dam“ Hon

addiiiOnal “n

56

has not changed, a time calibration should be done every few days, consistent with

proper laboratory practice.

The acquisition of a fluorescence decay is straightforward, but one should
follow a few guidelines when adjusting instrumental parameters such as the setting of
the delays, the TAC timebase, MCA gain and offset, and the count rate. The delays in
the data channels should be set such that the TAC is near the end of its time base at
time zero. When setting the delays one must also keep in mind that the input voltage
range of the MCA may be less than the output voltage range of the TAC. Once the
delays are set, the TAC timebase and MCA parameters should be adjusted such that the
decay is bracketed by dark noise. That is, the number of counts in the channels
preceding the decay should be roughly the same as the number of counts in the
channels at the extreme tail of the decay. Hence, one should acquire data over four or
five lifetimes. This is not absolutely necessary, but it is important when one is
attempting to fit a fluorescence decay to a bi-exponential or more complex model. The
MCA gain need not exceed 512 or 1024 channels, except if one is striving to fit other
than simple decays. The MCA offset allows one to block a selectable range of the
lower channels (TAC pulse amplitudes) while maintaining a high gain. In this way one
can achieve very high time resolution (about 2.7 ps/channel) over a small window

(e.g., 2 ns) for the acquisition of very fast fluorescence decays.

The count rate, the number of valid TAC conversions per second, should be
kept below one kHz to maintain integrity in the acquired data as mentioned in the
section on the TAC. This is of greatest concern when one is striving for the highest
accuracy possible and when attempting to characterize multi-exponential excited-state

decays. However, one can operate at higher count rates if one is willing to tolerate the

additional uncertainty in the acquired data.

 

 

 

 

 

If it is m
that excites the
parameters in ti
the monochron

instrument resp

Recordi
accomplished '
containing cycl
sample compa
filters to pass
instrument res]
Of course the
fluorescence “

This effect car

The to
method is She
Scatterer, Be
imilllfities in 1
“‘6 ulna-vi.
by Passing the

Ilithe

an intense 81a

method Work

57

If it is necessary, the best means to reduce the count rate is to attenuate the light
that excites the sample (e.g. by narrowing an iris). This is preferred to changing
parameters in the optical path between the sample and the detector such as narrowing
the monochromator slits or inserting neutral density filters. These will change the

instrument response.

Recording the instrument response is straightforward and is normally
accomplished within one minute. A convenient procedure is to place a cuvette
containing cyclohexane or another compound that is an efficient Raman scatterer in the
sample compartment. One then resets the monochromator and selects appropriate
filters to pass Raman scattered light from the sample. One should collect the
instrument response using the same parameters used to collect the fluorescence decay.
Of course the instrument response will be shifted relative to what it would be at the
fluorescence wavelength due to the color effect of the detector and the monochromator.

This effect can be taken into account and corrected for during the data analysis.

The major source of difficulty in obtaining an instrument response by this
method is short—lived fluorescence from impurities in the solvent used as a Raman
scatterer. Because SPT is so sensitive, one can readily detect fluorescence from
impurities in spectrophotometric grade solvents. This is especially true with excitation
in the ultra-violet. Fluorescence from impurities in cyclohexane can be reduced greatly

by passing the cyclohexane through a column of silica gel and basic alumina [40].

In the past a common means of obtaining an instrument response has been to use
an intense elastic scattering solution such as Ludox or diluted dairy milk. Although this

method works to a point, it produces intense after pulses in the instrument response.

 

 

 

These are prim
used because 1i

satisfactory wl

Now ti
is appropriate
of these is the
This is showr
was degassed
excited at 29(
is not shown
acquired in t]
this diSplay.
IUPAC room
the Complexi
A more com

also that the

58

These are primarily due to reflections. This method is attractive when a weak source is
used because little time is required to obtain the instrument response. However it is not

satisfactory when one is using a laser source.

Now that the configuration of the instrument has been thoroughly discussed, it
is appropriate to present examples of excited-state decays acquired by SPT. The first
of these is the decay of naphthalene obtained with a conventional photomultiplier tube.
This is shown in Figure 3-6 on page 59. The sample is 10 #M in cyclohexane, and
was degassed using the method suggested by Matsuzawa et a1. [41]. The sample was
excited at 290 nm, and fluorescence was detected at 350 nm. The instrument response
is not shown because it is much shorter than the fluorescence decay. The data were
acquired in the reversed timing configuration, and the time axis was reversed prior to
this display. Note that the data are presented as log counts versus time following the
IUPAC recommendation [39]. Presenting the data in this fashion aids in establishing
the complexity of the decay. A straight line is indicative of a single exponential decay.
A more complex decay mechanism is probable if the slope of the decay changes. Note

also that the complete fluorescence decay is recorded.

As evidenced by the time interval in the abscissa in Figure 3-6, the excited-state
‘ of naphthalene is a relatively long-lived. A molecule that decays much faster is rose
bengal. Its fluorescence decay and the instrument response function are shown in on
page 60 Figure 3-7. The sample was excited at 570 nm and fluorescence was detected
at 585 nm. The instrument response, the narrow "peak" on the left-hand side of the
figure, was obtained by collecting Raman scattered light at 677 nm from cyclohexane.
Like naphthalene, the decay of rose bengal appears to follow a single exponential.

Note the difference in the time axis. The decay of rose bengal is complete in less than

 

 

 

 

10’

ts
n

u

(:0

 

Figure 3-6. F‘

59

 

10

«25*
A A AAJ

V ‘Vv—V"
.l‘.
I

e, .

> n.&. 1
I ¢. 0
.g‘.‘

b .1.- ‘

e 3' .-
2 h 3".” "
10 , '3. 3.2:. a
..
s " 3‘33 1
a a. I
O
’ ..'. 0".- .. ‘
o. '0 . “ a
m P . :14 ...O. 1
O o.o c
e 0 V .0 .-
H P a '0‘ '9. ~\ . .. 4
o r. o
‘3 3'53. "r"
= b n i‘. I C . I n 7
a .-. . 0.20.0... .0... .0. a o
g ’ .Clfl-Q ~.' . I O. O i
I -0. .. C .C C -
o o 0 an o -0. o o 0 on o
c ; .70. .... a” .0. o to... .
. o o o 0.0 on... o. C
. u o N .- 0.00. 0 IO --
. I O O. .0 - O .
’ o .- ... o - o
‘o a n 0 ‘
- c a a 0.. o {

fivvv'
.0.
O

 

‘0 L l l _L

0 so an an an
Time/10" sec

 

 

3t
a
3

Figure 3-6. Fluorescence decay of naphthalene, 10 pM in degassed cyclohexane.

 

Counts

 

 

Filure 3-7. f
reslouse.

 

60

 

 

 

 

, I r r I T l I I I
s- l
10 . at» l
h :: $ 4
D J t»: t
p .0 0“ 1
t s 1
. 3%, .
. ' a ,
a.
as
D v . ‘
aa-
1%.:
x ' R‘.
. 231:...
a p S ~o'y.‘ cl
*0 » ~ :v‘... .
t e ' 55"} ‘
b ’ ‘15-'8- 4
' '8: s .
r {33$}: 4
a f o 2’ 0‘ z 1
4.0 b 115‘s?- 1
g I; . :1"? '
f ' . .0..‘:’;‘&. o. i
= - .3313. '
a '.=:r.~'-'-' °
(.3 t 2:: "" " "5:1”: i
1 o 3.. I: :0. e o 12...... :1 1
b - ~ o a. co...- o e o. . o
‘0 b -0 I C o .0000 o a. o. o o ‘1
b o... o o o e u - -0. I. o 1
P cc. 0 - e - « o 0 0...... me 1
p - o .- o co . 0.0 o 0 00"“ fl- O .0”’0 1
L g... u go‘- . :.... . . ... . . o. cannon-0--....oeoo-o 1
Paco-coo. ‘0‘- .00 do .. o . o... o c .0. o . so .0 -0- won't-.0001
yd—O --—-oooo 0...... a... oo-ouuooo - 90 0o. .0. 0-....OOOO_.— I
L _.- O I‘D-n-- m—--.- .‘-—---I—._.- a-o ‘0-0.--o..‘ {
o _ 0
i0 1 l . 4 L l L J 'l

 

0 5 so 15 20 25 fl 5 40

Time/104°sec

Figure 3-7. Fluorescence decay of rose bengal, 10 pM in methanol, and instrument
response.

 

 

four nanoseconds.
FWHM, and is fol

reflections from the

Naphthalene
different decay tim
on most laser-base
fiber optic delay
measurements easi

» pIOper data analys:

Once the i
acquired, one det
amode1(e,g. a s
the fluorescence
task, the method
tteIation of the fi

Where i is the C
experimental day

follows a POissc

61

 

four nanoseconds. The instrument response is quite narrow, about 50 picoseconds
FWHM, and is followed by a bit of a tail. This is not fluorescence, but is due to

reflections from the cuvette walls as mentioned earlier.

Naphthalene and rose bengal were used as examples because of their vastly

different decay times. Acquisition of the fluorescence decays of these two compounds

on most laser-based SPT instruments would be quite difficult. However, the switched
fiber optic delay discussed in Appendix 2 allows one to perform both of these
measurements easily. Yet the ease of data acquisition is of no consequence without

proper data analysis. This topic is discussed in the next section.
DATA ANALYSIS

Once the instrument response and the fluorescence decay of a sample have been
acquired, one determines the excited-state lifetime by comparing the obtained decay to
a model (e. g. a single exponential). The model is successively refined until it matches
the fluorescence decay. Although many methods have been used to accomplish this
task, the method of least squares [11, 42, 43] is the most widely used. With each
iteration of the fitting procedure, the parameter X’ is determined:

2
X 2 = : [warm] (M)
where i is the channel number (the natural unit of time in SPT), Y(i) values are the
experimental data, Mti) values are those predicted by the model, and the a(i) values are
the standard deviations in the experimental data. Since the uncertainty in SPT data

follows a Poisson distribution, a(i) can be estimated from the experimental data:
1

a (i) a (Y(i)); (3-7)

 

 

The value of x2 at e
exist for optimizin
advantages and dis
commercial softwa

variable step size gi

In addition
account for the d
important when th
lifetime. The acqv
fluorescence decay

related by the conv

hurt and Hit) a
inStrument respons
have been used [1]
[46’ 471- In recc
model before com
apPYOaCh than are
a“hired data.

N0 matter
acqlfired data: ODE

data, when least

“fit" 33 it is con

to the number 0f d

62

The value of X2 at each iteration is used to guide the search procedure. Many strategies
exist for optimizing the model to find a minimum in x2, and each strategy has
advantages and disadvantages, especially with respect to speed and stability. The
commercial software package [44] purchased for use with this instrument uses a

variable step size grid search followed by parabolic interpolation.

In addition to fitting a model to the fluorescence decay, one also needs to
account for the distortion of the decay caused by the instrument. This is most
important when the instrument response is significant compared to the excited-state
lifetime. The acquired fluorescence decay, R(t) is actually a convolution of the true
fluorescence decay, F(t), and the instrument response, L(t). These three functions are

related by the convolution integral [7, 10]:
R(t) = f L(t') F(t-t') dt (3—8)
0

If L(t) and R(t) are known, the true fluorescence decay, four nanoseconds. The
instrument response is quite F(t) can be determined. Although a variety of techniques
have been used [11, 39, 45], the method of iterative reconvolution is the most common
[46, 47]. In reconvolution analysis, the instrument response is convolved with the
model before comparing the model to the acquired decay. This is a more stable
approach than attempting to deconvolve the effect of the instrument response on the

acquired data.

No matter which technique is used to adjust the model so that it matches the
acquired data, one must have objective means to judge how well the model fits the
data. When least squares is used, the most basic means of determining the "goodness
of fit", as it is commonly referred to, is the reduced value of )8. This is X2 normalized

to the number of degrees of freedom (i.e., the number of data points less the number of

 

 

 

fitted parameters)
the confidence in
values above 1.5.
This is the ratio o
the expected diffe
weighting factor
indicates a systen
residuals is neces
is not obvious ir
subjective insPect
One such test of
(i.e., are half 0:
stittttthrd deviatiq
Parameter [48,49

Where Tits) is th
Chosen for the ex
dete’mme trends
(DU and DL) for
I)tlrbin and Wat:
fitted parameters
one must eth'apo

this parameter to

 

63

fitted parameters). A value of unity for the reduced x2 is indicative of a good fit, and
the confidence in the fit decreases with an increase in X2 . Most workers seldom accept
values above 1.5. After X2, it is common to examine a plot of the weighted residuals.
This is the ratio of the actual difference between the experimental data and the model to
the expected difference (standard deviation of the data). Since the data are Poisson, the
weighting factor is determined directly from the data. A non-randomness in this plot
indicates a systematic difference between the data and the model. Because a plot of the
residuals is necessarily noisy, an autocorrelation of the residuals can reveal a trend that
is not obvious in the plot of the residuals. As both of these are visual, and hence
subjective inspections, one should also test the randomness of the data by other means.
One such test of the residuals is to see if they are distributed in a Gaussian fashion
(i.e., are half of the residuals positive, are 68 percent of the residuals within one
standard deviation, etc.). Beyond this, one can determine the Durbin-Watson

parameter [48,49]:
n2

2
El rite) ' r(ti-1))

DW = f=n1 +1

n2 2
X (rttiii

i=n1

 

where r(t,) is the weighted residual, and n1 and n2 are the first and last channels
chosen for the curve fit.. The Durbin-Watson parameter was originally developed to
determine trends in econometric data, and the authors generated upper and lower limits
(DU and DL) for data sets of varying size and parameters. In their original work [48],
Durbin and Watson reported limits for DW for data sets of up to 100 points and five
fitted parameters. As the number of channels in a SPT experiment often exceed 100,
one must extrapolate values for DU and DL from the original work if one wishes to use

this parameter to judge the residuals from a curve fit. Although some deem it a more

'

sensitive judgmen

Watson parameter

Other me:
of skew [50] anc‘
distributed about

the analysis of ST

Figure 3-1
naphthalene deca
model (i.e. A +
= 0.49 us) was
values of 107 ns
ihStlument ICSpm
seems to be succ
“my, appear to
residuals are neg:
Standttfd deviatio:

Ah autoc
AlthOugh no majr
“about 13 ns. .

driver whiCh ope

64

sensitive judgment of the goodness of fit than X2 [14], few workers report the Durbin—

Watson parameter for SPT data, and IUPAC does not yet recommend its use [39].

Other means of judging randomness in the weighted residuals are the measures
of skew [50] and kurtosis [51]. The former indicates deviations that are not evenly
distributed about zero while the latter indicates convexity. Neither is widely used in

the analysis of SPT data.

Figure 3—8 on page 65 shows the results of a least squares curve fit of the
naphthalene decay shown in Figure 3—6. The decay was fitted to a single exponential
model (i.e. A + Bexp-(i/T)) over 855 channels (0.708 ns/ch). A lifetime of 110 ns (0
= 0.49 ns) was determined. This is in excellent agreement with modern literature
values of 107 ns [41] and 100 ns [52]. Reconvolution was not necessary because the
instrument response was negligible compared to the fluorescence decay. The curve fit

‘ seems to be successful because x2 is nearly unity. The residuals, plotted beneath the
decay, appear to be random and evenly distributed about zero. In fact, 45.8% of the
residuals are negative, 65.8% are within one standard deviation, 94.7% are within two

standard deviations, and 99.5% are within three standard deviations.

An autocorrelation of the residuals is shown on page 66 in Figure 3-9.
Although no major trends are evident in this plot, there is a small oscillation at a period

of about 13 ns. This corresponds to radio frequency interference from the mode locker

driver which operates at 38.25 MHz.

 

ll
4
2
. it
-2
‘4
F1gure3 8

“8:390

 

 

 

 

 

65

 

A A AAAA‘

Counts

 

 

 

 

 

 

 

 

Std.Dev

 

Figure 3-8. Fitted decay of naphthalene. The lower plot shows the weighted residuals.

 

4..
t
Figure 39. Autt

 

 

o
4 2 n. a... a.‘ 4- .1 a...
a
.

10’

ab

bu
teaxeeaun~ueeau

 

12'

 

66

 

 

 

 

 

 

 

42’ ..
10" ‘
I" .4
'l‘ 5h M II 4
a ! |
: 4* l i ‘ 't
I t on
z y n it y] it 1, “'l y 6'!
2 -2" l ' ;_ ‘
.4- .
-5' i
-3» ..
'10' L L 1 h l 14
0 5 10 15 20 a I)

Time/10* see

Figure 3-9. Autocorrelation of weighted residuals from fit of naphthalene decay.

 

Since the ii
bengal, one should
bengal was fit tr
reconvolution to SI
carried out over 1.
each case. Withox
Avalue of 549 p
Values from the 1

distribution of the
API

Steady-stat
SPT was used to
of correcting for
Data from three t
Oxygen, quinine :
bl nitromethane.

i161 and nitromet

Fluoregc‘
model LS.5 fluC

67

Since the instrument response is not negligible compared to the decay of rose
bengal, one should include reconvolution during the data analysis. The decay of rose
bengal was fit to a single exponential model, and analyzed with and without
reconvolution to show the difference in the determined lifetime. Both curve fits were
carried out over 1400 channels (2.77 ps/ch), and the value of X2 was less than 1.1 in
each case. Without reconvolution, a lifetime of 564 ps (or = 1.8 ps) was determined.
A value of 549 ps (0 = 1.7 ps) was determined when reconvolution was included.
Values from the literature are 549 ps [53] and 543 ps [54]. For both curve fits, the

distribution of the residuals was similar to that seen in the fit of the naphthalene data.

APPLICATION TO CORRECTION FOR QUENCHING

Steady-state fluorescence intensity was measured on a common fluorometer and
SPT was used to determine excited-state lifetimes by the Hieftje-Haugen method [16]
of correcting for the reduction in fluorescence intensity due to collisional quenching.
Data from three chemical systems are presented. These are naphthalene quenched by
oxygen, quinine sulfate quenched by chloride ion and dansyl-L-phenylalanine quenched
by nitromethane. Oxygen is a well known collisional quencher [55], and chloride ion

[16] and nitromethane [56] are also documented collisional quenchers.

Experimental Details

APPARATUS

Fluorescence intensities were measured with a Perkin-Elmer (Norwalk, CT)

model LS-5 fluorometer. Excited-state decays were obtained with the SPT apparatus

 

 

described in this cl
pacloge (199M soft

Certified gr
of roughly equal qt
was recrystallized I
from the certified
Purified cyclohexai

A 1.0 mM
in0.1 M sulfuric
(Malinckrodt) in o
Dansl’l'L-phenylal
was Prepared in
potaSSi“1h Chloridt
(Malinmdt) was

A degasse
fouowmg manner

inavalved, qua
fluorescenCe in ten

atmosphere brief]

68

described in this chapter and lifetimes were determined with a commercial software

package ( 199M software library, IBH Consultants Ltd., Edinburgh, UK).

REAGENT S

Certified grade cyclohexane (Fisher) was purified by passing through a column
of roughly equal quantities of silica gel and basic alumina. Naphthalene (Malinckrodt)
was recrystallized twice from cyclohexane. The first crop of crystals was recrystallized
from the certified cyclohexane, and the second crop was recrystallized from the

purified cyclohexane.

A 1.0 mM stock solution of quinine sulfate monohydrate (Aldrich) was prepared
in 0.1 M sulfuric acid (prepared by diluting 5.5 ml of concentrated sulfuric acid
(Malinckrodt) in one liter of distilled and deionized water (Corning millipure system)).
Dansyl-L-phenylalanine (Sigma) was used as received, and a 1.0 m_ls_/I_ stock solution
was prepared in 2-propanol (Burdick and Jackson). A 1.0 M stock solution of
potassium chloride (J .T. Baker) was prepared in 0.1 M sulfuric acid. Nitromethane

(Malinckrodt) was used as received.

PROCEDURES

A degassed solution of naphthalene in cyclohexane was prepared in the
following manner. A small crystal of naphthalene (approximately 0.07 mg) was placed
in a valved, quartz cuvette designed for use on a vacuum line. Cyclohexane was
degassed with several freeze-pump—thaw cycles, and transferred to the cuvette. The
fluorescence intensity and lifetime were measured. The cuvette was exposed to the

atmosphere briefly, and the measurements were repeated. This sequence was carried

 

out six times, and 1

measurements.

For the qun
systems, working s<
fluorophore and ap
volumetric flask, as
solutions were prep

quencher in these St

For the fit
wavelengths were 1
nm for the quinir
phenylalanine solu
mensurements. A
uncertainty in the
Stthe wavelengths

Wavelength was 28

Various dej
SPT meitStlrement

peak channel WCI

chttnnels. The N.

1481) Slit widths
m°“°°hromat or a

A cutoff filter (0

 

69

out six times, and oxygen was briefly bubbled through the solution for the last two

measurements.

For the qunine sulfate - chloride and dansyl-L-phenylalanine - nitromethane
systems, working solutions were prepared by combining 1.0 m1 of the stock solution of
fluorophore and appropriate volumes of the stock solution of quencher in a 100 m1
volumetric flask, and diluting to the mark with the appropriate solvent. Nine working
solutions were prepared for each fluorophore-quencher system. The concentration of

quencher in these solutions was 0, 0.005, 0.001, 0.05, 0.01, 0.5, 0.1, 0.5, and 1.0 M.

For the fluorescence intensity measurements, the excitation and emission
wavelengths were 290 nm and 330 nm for the naphthalene solution, 282 nm and 450
nm for the quinine sulfate solutions and 350 nm and 505 nm for the dansyl-L-
phenylalanine solutions. The excitation and emission bandpasses were 3 nm for all
measurements. A single intensity measurement was made on each solution, and the

uncertainty in the measurement was estimated from fluctuations in the readout. The

 

same wavelengths were used for the lifetime determinations, except that the excitation

wavelength was 282 nm for the dansyl-L-phenylalanine solutions.

Various delays, TAC time base settings, and MCA parameters were used in the
SPT measurements. Three decay curves having approximately 1,000 counts in the
peak channel were obtained for each sample. All decays were acquired over 1024
channels. The MCP-PMT was biased at -3,100 volts. The monochromator (SPEX
1481) slit widths were 0.5 mm for all measurements. A mask was placed in the
monochromator at the grating surface to block all but the central ten mm of grooves.

A cutoff filter (Oriel KV418 or WG305) and a bandpass filter (Oriel 57560 or 57540)

 

 

 

 

were included in

minutes per 501“

Each do:
was not include
The poorest fit
and the dansyl-
were randomly

the weighted n

Thef

lifetime deter

TABLE 3-2.
I/r OF NAI
OXYGEN.

ixusm
cOmplet
degass

l
2

3

70

were included in the collection optics. Acquisition times were on the order of a few

minutes per solution.

Each decay curve was fit to a single exponential model. Reconvolution analysis
was not included. The reduced X2 for all but'four of the curve fits was less than 1.1.
The poorest fits were obtained for the quinine sulfate solution that was 1.0 M in KCl
and the dansyl-L-phenylalanine solution that was 1.0 M in nitromethane. The residuals
were randomly distributed for all of the curve fits as judged by inspections of plots of

the weighted residuals.

Results and Discussion

The fluorescence intensity measurements and the results of the excited-state
lifetime determinations for naphthalene quenched by oxygen are given in Table 3-2.
TABLE 3-2. FLUORESCENCE INTENSITIES, LIFETIMES AND THE RATIO OF

I/r OF NAPHTHALENE IN CYCLOHEXANE SUCCESSIVELY QUENCHED BY
OXYGEN.

 

firmsure Number Intensity Lifetime jugs) Intensity/Lifetime
completely
degassed 8970 109.9 81.6
1 3010 39.9 75.4
2 1590 20.7 76.2
3 1530 18.9 81.0
4 1370 16.3 84.1
5 510 5.80 87.9
6 330 3.68 89.7

 

 

.gfio -m .
.....ds._._.___—.W____._V

l
l
l
v

 

As one would expe
with the number 01
oxygen to enter the
quenching by ratio
3-10. The stead;

_.

Normalized Fluorescence Intensl‘ty

Fights: $10. St
excited-state life
my

intensities by (11‘
bars on the cor
deViation in hot]
intensity data be
The ”nation i
trend with inc:

71

As one would expect, the fluorescence intensity and the excited state lifetime decrease
with the number of exposures to the atmosphere. These exposures presumably allow
oxygen to enter the solution. The steady-state intensity and the intensity corrected for
quenching by ratioing to the excited-state lifetime are compared graphically in Figure

3-10. The steady-state fluorescence intensities were normalized to the corrected

 

 

> 100-

‘t

2 I I

°’ -- I

E 80-"-- I I I

m

u

S

U 60~

m

01

L

O

3 40‘

LL.

'5 .

OJ

2‘ 20—

d o a

c .

L

2° ° -
o I I I I I I

O 1 2 3 4 5 6

Exposure Number

Figure 3-10. Steady-state fluorescence intensity (circles) and intensity ratioed to the
excited-state lifetime (squares) for naphthalene in degassed cyclohexane (approx 0.1

mM).

intensities by dividing all values by the unquenched lifetime of naphthalene. The error
bars on the corrected intensities indicate propagated uncertainty for one standard
deviation in both lifetime and intensity. Error bars are not shown for the steady-state
intensity data because they would be on the order of the size of the plotted characters.
The correction is almost ideal within the limits of the error bars. The general upward
trend with increasing exposure is probably due to evaporation of the cyclohexane

during exposure to the atmosphere. This process will increase the concentration of

F

naphthalene in the
this deviation is g

solution.

The floors
lifetime determin:

3-3. The normah

TABLE 3-3. FL'

I/r 0F QUININ
VARYING CON

._ [CT]
0.0
0.005
0.01
0.05
0.1

0.5

Wing t0 the s
11' AS in Figuy
by dividing an,
SYSlem, the flu
concenmm or
independent on

concentration 0

72

naphthalene in the solution, and hence, increase the intensity. As one might expect,
this deviation is greatest for the exposures in which oxygen was bubbled through the

solution.

The fluorescence intensity measurements and the results of the excited-state
lifetime determinations for the quinine sulfate - chloride ion system are given in Table

3-3. The normalized steady-state intensity and the intensity corrected for quenching by

TABLE 3-3. FLUORESCENCE INTENSITIES, LIFETIMES, AND THE RATIO OF
I/T OF QUININE SULFATE, 10 ”M 0.1 M SULFURIC ACID, QUENCHED BY
VARYING CONCENTRATIONS OF CHLORIDE ION.

 

[Cl'] Intensity Lifetime (ns) Intensity/Lifetime
0.0 8980 18.7 481
0.005 5070 10.9 466
0.01 3600 8.0 449
0.05 1025 2.5 418
0. l 550 1.4 404
0.5 1 18 0.31 385

ratioing to the excited-state lifetime are compared graphically on page 73 in Figure 3-
11. As in Figure 3-10, the fluorescence intensities were normalized to the ratio of Ur
by dividing all of the values by the unquenched lifetime. Like the naphthalene-oxygen
system, the fluorescence intensity and the excited state lifetime decrease as the
concentration of quencher increases. But unlike that system, the ratio of 1/7 is not
independent on the concentration of quencher. In fact, this quantity decreases as the

concentration of quencher increases. However, the corrected intensity does not

 

 

“by
In'tensl

l eel Fluorescence

Norma Iz

Figure 3-11. P,
intensity correcte
quinine sulfate in

discussed later in

The data 1
74 in Table 3.4_
decrease as the ‘
like the quinine
qUencher 1mm
in I‘igure 3.12
““luenched 1,1,

 

 

73
>. 500
t. :8:
2 I
m
E 400 I I I
m
U
5 300
U
U] I
m
L
O
3 200 .
Ln.
'6
m
2 100
d
E I
L
O a
2 0 JL C
III 1 . '.vn., . fi‘vmq fl rfivtm
(101 0.1 1

[Cl—l

igure 3-11. Plot of steady—state fluorescence intensity (circles) and fluorescence
tensity corrected for quenching by ratioing to the excited-state lifetime (squares) for
rinine sulfate in 0.1 M sulfuric acid quenched by chloride ion.

ecrease as rapidly as the uncorrected intensity. Possible reasons for this trend are

scussed later in this chapter.

The data for the dansyl-L—phenylalanine - nitromethane system is shown on page
in Table 3-4. Again, both the fluorescence intensity and the excited-state lifetime
crease as the concentration of quencher, nitromethane in this case, increases. And
e the quinine sulfate system, the ratio of 1/7' also decreases as the concentration of
encher increases, but more rapidly. These data are presented graphically on page 74
Figure 3-12. The fluorescence intensities here are also normalized to the

quenched I/1' value.

TABLE 34. FLU
n or DANSYL-
uu VARYING cc
[canon
0.0
0.005

Normalized Fluorescence Intensity

Figure 3.12‘ P
Syl‘L‘l’heilylzy

 

74

TABLE 34. FLUORESCENCE INTENSITIES, LIFETIMES, AND THE RATIO OF
I/r OF DANSYL-L-PHENYLALANINE, 10 I‘M IN 2-PROPANOL, QUENCHED
BY VARYING CONCENTRATIONS OF NITROMETHANE.

 

[CH3N07] IntensitL Lifetime (ns) Intensity/Lifetime
0.0 8980 15 .0 598
0.005 6520 11. 8 554
0.01 5280 10.3 513
0.05 3755 7.7 487
0. 1 1830 4.4 417
0.5 120 0.76 158
1.0 37 0.44 84

600::

~ I
y l

400- I
300-
200*

100- I

O- g n
I I I Yttt‘tr] t fTYYVYtI v YjYWYYYI

0 001 OJ
[CHJNoz]

NornolIzed Fluorescence Intenslty

 

 

Figure 3-12. Plot of steady-state fluorescence intensity (circles) and fluorescence
intensity corrected for quenching by ratioing to the excited-state lifetime (squares) for
dansyl-L-phenylalanine in 2-propanol quenched by nitromethane.

 

 

The Hieft
degree, but it d
correction it prr

quenching mech:

The nap
Although the ac
from the steady
occurring. The
by oxygen is w
above, the com

of an accurate 1

The oth
mechanisms 0
Hieftje-Haugey
quencher is h
concentration

quinine sultan

In bo
concentration
radiative me
disagreement
constant is I

Hence, the .

intenshy wit

75

The Hieftje-Haugen method [16] corrects for quenching in these systems to a
degree, but it does not work well with all of the chemical systems studied. The
correction it provides is most accurate when collisional quenching is the dominant

quenching mechanism.

The naphthalene-oxygen system is the most well-behaved system studied.
Although the actual concentration of oxygen in the solution is not known, it is evident
from the steady-state intensity and the excited-state lifetime data that quenching is
occurring. The most likely mechanism is collisional quenching as collisional quenching
by oxygen is well known. A Stem-Volmer plot would verify this. However, as stated
above, the concentration of oxygen in the solution is not known, making construction

of an accurate Stern-Volmer plot impossible.

The other fluorophore—quencher systems tested are cases where other quenching
mechanisms or other factors that affect the quantum yield are possible. Here the
Hieftje-Haugen method is not highly accurate, especially when the concentration of
quencher is high. The decrease in the ratio of Ur with an increase in quencher
concentration is greater in the dansyl-L-phenylalanine - nitromethane system than in the

quinine sulfate - chloride ion system. This could be due to several phenomena.

In both systems, the refractive index of the solution increases with the
concentration of quencher. This also affects the excited-state lifetime because the
radiative rate constant is dependent on the refractive index. Although there is some
disagreement on this in the literature, it is generally thought that the radiative rate
constant is proportional to the square of the refractive index of the solution [14].
Hence, the excited—state lifetime should decrease as 112 increases. The fluorescence

intensity will be coupled to this because the quantum yield, Q, is also dependent on the

 

 

radiative rate v

intersystem crc

lf kNR remain
quantum yield
obviously dep
ratio of Ur to
explain a pos

observed for f

If, for
quencher incn
the ratio of L

quenching.

Even
quenchers, a
formation of
Static quenchi
be reduced, 1
from 00th st
°°ncehtration
fluoroPhore v
wens“? mor
tiansy1.y,phe
SysteIn 0Winy

76

radiative rate constant, kR, and the rate constant for all non-radiative processes (e.g.

intersystem crossing, internal conversion etc...), kNR:
kR

Q=kR+kNR

(3-9)

If kNR remains constant, an increase in the refractive index will also increase the
quantum yield, and hence, the fluorescence intensity. The extent of the increase
obviously depends on the relative values of kR and kNR. One would expect that the
ratio of Ur to increase as the refractive index of the solution increases. This would
explain a positive deviation in the ratio of Ur, but not a negative deviation as is

observed for these systems.

If, for some reason, kNR increases faster than kR as the concentration of
quencher increases, the quantum yield will decrease. This will cause the intensity and
the ratio of Ur to decrease by an amount greater than that due to simple, collisional

quenching.

Even though chloride ion and nitromethane are documented collisional
quenchers, a possible explanation for the observed deviation is static quenching: the
formation of a non-fluorescent "dark complex" at high concentrations of quencher. If

_ static quenching were the only quenching mechanism, the fluorescence intensity would
be reduced, but the excited-state lifetime would not be affected. If quenching occurs
from both static and dynamic quenching, the ratio of I/r should decrease as the
concentration of quencher increases. This is due to the fact that a larger fraction of the
fluorophore will be complexed at high concentrations of quencher, thus reducing the

intensity more than the extent due to collisional quenching. This is more likely for the

 

dansyl-L phcu,‘ ‘ ' " ‘L system than for the quinine sulfate - chloride ion

system owing to the strong electron withdrawing nature of the nitro group. Indeed the

 

 

deviation is g
nitromethane.

Stem-Volmer 1

Another
fluorophore m
quencher is m

instantaneous q

Perhaps
asemi-quantita
quenching is .
quencher will l
Quenching mec
would constru
methanism 0ft
0f quencher j
demonstrated
qufincher is m

method,

77

deviation is greater for the 1r electron rich dansyl-L-phenylalanine quenched by
nitromethane. One could determine the extent of static quenching by constructing

Stem—Volmer plots of both FOIF and rolr at different temperatures [10].

Another possible explanation for the observed deviation is that some of the
fluorophore molecules are quenched immediately at high concentrations. That is a
quencher is more likely to be an immediate neighbor to a fluorophore, resulting in

instantaneous quenching that may appear to be static quenching [10].

Perhaps another application of the Heiftje-Haugen measurement scheme [16] is

a semi-quantitative method of determining the mechanism of quenching. If collisional
quenching is the dominant mechanism, a plot of Ur versus the concentration of
quencher will be a straight line with a slope of zero. A deviation will indicate other
quenching mechanisms such as static quenching or energy transfer. Traditionally one

3 would construct Stem—Volmer plots at different temperatures to establish the
mechanism of quenching. However one needs accurate knowledge of the concentration
l of quencher in the solution to construct meaningful Stem-Volmer plots. As
idemonstrated with the naphthalene-oxygen study, the absolute concentration of
quencher is not needed to interpret measurements made using the Hieftje-Haugen

it method.

 

 

10.

12.

Koechlin1
Bollinger
Wahl, Ph

Blumberg

. Birks, 1.1

Morton, t

Yguerabii

. Knight, A

Ware, W
A.A., Ed

Lakowicz
York, 191

Demas, J
1983.

Flenling,
University

Steiner, I
Complexes
1:Ohtissier,

O'COItnor
Academic

' DI. Gal-y

Communiv
Hieftje. G
Wahl, Pr

Pfeffer, G

10.

11.

 

78
CHAPTER THREE REFERENCES

Koechlin, Y. Ph.D. Dissertation, University of Paris, 1961
Bollinger, L.M.; Thomas, G.E. Rev. Sci. Instrum. 1961, 32, 1044.

Wahl, Ph.; Lami, H. Biochim. Biophys. Acta, 1967, I33, 233.

 

Blumberg, W.E.; Eisinger, J.; Nonen, G. Biophys. J. Soc. Abs. 1968, 12, A106.
Birks, J.B.; Munro, I.H. Prog. React. Kinetics 1967, 4, 239.

Morton, G.A. Appl. Optics, 1968, 7, l.

Yguerabide, J. Meth. Enzymol. 1972, 26, 498,

Knight, A.E.W.; Selinger, B.K. Aust. J. Chem. 1973, 26, l.

Ware, W.R. in Creation and Detection of the Excited State vol 1A, Iamola,
A.A., Ed. Marcel Dekker, Inc.: New York, 1971, chapter 5.

Iakowicz, J.R. Principles of Fluorescence Spectroscopy Plenum Press: New
York, 1983.

Demas, J.N. Excited State Lifetime Measurements Academic Press: New York,
1983.

Fleming, G.R. Chemical Applications of Ultrafast Spectroscopy Oxford
University Press: New York, 1986.

Steiner, R.F.; Holtom, G.H.; Kubota, Y. Laser Spectroscopy of Nucleic Acid
Complexes in Lasers in Polymer Science and Technology: Applications, vol 4.
Fouassier, LP. and Rabex, J.F. Eds. CRC Press: Boca Raton, FL, 1989

O'Connor, D.V.; Phillips, D. Time-Correlated Single Photon Counting
Academic Press: London, 1984

Dr. Gary Holtom, Pacific Northwest Laboratories, Richland, WA. Personal
Communication, December 1990.

Hieftje, G.M.; Haugen, G.R. Anal. Chim. Acta. 1981, 123, 255.
Wahl, P.R. New Tech. Biophys. Cell. Biol. 2, 233. 1975

Pfeffer, G.; Lami, H.; Laustriat, G.; Coche, A. Compt. Rena. 1962, 4, 1035.

 

 

20.
21.
22.

23.
24.
25.
26.

27.

28.

29.
30.
31.
32.
33.

34.

35.
36.
37.
38.
39.

Cramer, l
Grinwald,
Kinoshita,

Phillips, I
1985, 14,

BEAM4,
Bebelaar,
Technical
Catalog 0

Applicatii
Related 1

February

Characte
RES-079

19.
20.
21.
22.

23.
24.
25.
26.
27.

28.

79
Cramer, L.E.; Spears, K.G. J. Am. Chem. Soc. 1978, 100, 221.

Grinwald, A. Anal. Biochem. 1976, 75, 260.
Kinoshita, S.; Ohta, H.; Kushida, T. Rev. Sci. Instrum. 1981, 52, 572.

Phillips, D.; Drake, R.C. O'Connor, D.V.; Christensen, R.L. Anal. Instrum.
1985, I4, 267.

BEAM4, Stellar Software, Berkeley, CA.

Bebelaar, D. Rev. Sci. Instrum. 1986, 5 7, 1686.

Technical Literature, American Holographic, Littleton, MA, 1990.

Catalog of Photomultipliers, Hamamatsu Photonics K.K., Japan, January 1988.

Application of MCP-PMTs to Time Correlated Single Photon Counting and
Related Precautions Technical Literature, Hamamatsu Photonics K.K., Japan,
February 1987.

Characteristics and Applications of Microchannel Plates Technical Manual
RES-0795. Hamamatsu Photonics K.K., Japan, January 1987.

Lacaita, A.; Cova, S.; Ghioni, M Rev. Sci. Instrum. 1988, 59, 1115.
Lakowicz, J.R.; Laczko, G.; Gryczynski, I; Rev. Sci. Instrum. 1986, 5 7, 2499.
Technical Literature, Precision Tube Company, North Wales, PA, 1988.
Gedcke, D.A.; McDonald, W. J. Nucl. Instr. Meth. 1968, 58, 253.

Leskovar, B.; Lo, C.C.; Hartig, P.R.; Sauer, K. Rev. Sci. Instrum. 1976, 47,
1113.

Graves, W.; Coyne, B. In-house Correspondence, Tennelec, Inc. Oak Ridge,
TN. September 26, 1986.

Anfinrud, P.A.; Struve, W.S. Rev. Sci. Instrum. 1989, 60, 800.

Spears, K.G.; Cramer, L.E. Hofland, L.D. Rev. Sci. Instrum. 1978, 49, 255.
Haugen, G.R.; Wallin, B.W. Lytle, F.E. Rev. Sci. Instrum. 1979, 50, 64.
Hallam, A.; Imhof, R.E. J. Phys. E. 1980, I3, 520.

Comission on Photochemistry, Pure Appl. Chem. 1990, 62, 1632.

 

 

40.

42.

43.

44.
45.
46.
47.
48.
49.
50.
51.
52.
53.
54.
55.
S6.

Lantpert,
Meech, S

. Matsuzau

Bevingtor
McGraw

Press, W
Recipies t

199M So:
Inpez-Dt
O'Conno:
McKinno
Durbin, 1
Durbin, 1
Irvin, D.
Pearson,
Comissio
Ware, W
Cramer,
Kautsky,

Dreeskm

40.

41.

42.
43.

44.
45.
46.
47.
48.
49.
50.
51.
52.
53.
54.
3' 55.
‘ 56.

 

80

Lampert, R. A.; Chewter, L. A; Phillips, D.; O' Connor, D. V.; Roberts, A. J;
Meech, S. R. Anal. Chem. 1983, 55, 68.

Matsuzawa, S.; Wakisaka, A.; Tamura, M. Anal. Chem. 1990, 62, 2645.

Bevington, P.R. Data Reduction and Error Analysis for the Physical Sciences
McGraw Hill: New York, NY 1969.

Press, W.H.; Flannery, B.P.; Teukolsky, S.A.; Vetterling, W.T. Numerical
Recipies Cambridge University Press: New York, NY, 1986. chapter 14.

199M Software Library, IBH Consultants Ltd., Edinburgh, Scotland, UK.
Lopez-Delgado, R.; Tramer, A, Munro, I.H. Chem. Phys. 1974, 5, 72.
O'Connor, D.V.; Ware, W.R.; Andre, J.C. J. Phys. Chem. 1979, 83, 1333.
McKinnon, A.E.; Szabo, A.G.; Miller, D.R. J. Phys. Chem. 1977, 81, 1564.
Durbin, J .; Watson, G.S. Biometrika 1950, 37, 409.

Durbin, J.; Watson, G.S. Biometrika 1951, 38, 159.

Irvin, D.J.G.; Livingston, A.E. Comp. Phys. Commun. 1974, 7. 95.
Pearson, K. Biometrika, 1905, I V, 181.

Comission on Photochemistry, Pure Appl. Chem. 1988, 60, 1108.

Ware, W.R.; Pratinidhi, M.; Bauer, R.K. Rev. Sci. Instrum. 1983, 54, 1148.
Cramer, L.E.; Spears, K.G. J. Am. Chem. Soc. 1978, 100, 221.

Kautsky, H. Trans. Faraday Soc. 1939, 35, 216.

Dreeskamp, H.; Koch, E Z. Naturforch. 1975, 30, 1311.

 

Fre
[1], and v
excited mt
not as int
FDF is n
nature, th

time-dom.

followed
described
excited-s
froquenc;
and the I
at a sing
determin
method |

CHAPTER 4: FREQUENCY DOMAIN FLUORESCENCE

Frequency domain fluorescence (FDF) was first reported by Gaviola in 1921
[1], and was the first technique capable of measuring the decay of an electronically
excited molecule with nanosecond resolution. Briefly described in chapter two, FDF is
not as intuitive as the various time—domain methods such as pulse sampling. Hence
FDF is not as widely used as the time—domain methods. Despite its non—intuitive
nature, the capabilities of modern FDF instrumentation are comparable to those of the

time-domain techniques.

This chapter begins with a brief treatment of the basic theory of FDF. This is
followed by a historical survey of the technique. A frequency domain fluorometer is
described which is designed to simultaneously acquire fluorescence intensity and
. excited-state lifetime in near real time. This fluorometer is similar to existing
frequency domain fluorometers, and differs primarily in the cross-correlation frequency
and the method used to measure the phase shift. The constructed fluorometer operates
at a single modulation frequency and, hence, the range of excited-state lifetimes it can
determine is rather narrow. Application of this instrument to the Hieftje-Haugen

method [2] of correcting for the effects of collisional quenching is discussed.
THEORY

The waveforms used in FDF and quantities derived from them were illustrated
in Figure 2-4. The excited state lifetime can be determined from either the
demodulation factor, m, or the phase shift, 4:, and several routes exist to derive the
expressions that relate the excited-state lifetime to these experimental parameters. The

discussion on the theory of FDF presented here is logically straightforward,
' 81

 

algebraically o

additional treat:

Conside
initial populatir

from the initial

where y and k

dependent inte

where a and
modulation fr:
the emission 1
and has redu

Population of

A alld B rep“
Shift bemem

modlflated fig

81117811011101} (

The Empora]

82

algebraically complex, and follows that of Iakowicz [3]. Readers interested in

additional treatments are referred to those by Ware [4] and Demas [5].

Consider a fluorophore excited by an infinitely short pulse of light generating an
initial population of the excited state. The population of the excited-state will decrease

from the initial value following:

L289 = 47 + k) MI) (44)

where 7 and k are the respective rates of non-radiative and radiative decay. The time
dependent intensity of a sinusoidally modulated source is given by:
flt) = a+b sin wt (4-2)
where a and b indicate the degree of modulation (Figure 2-4) and w is the radial
modulation frequency. A sinusoidally modulated source used to excite a sample causes
the emission to be modulated at the same frequency, but the emission is phase-shifted
and has reduced modulation. When such a source is used, we can express the
population of the excited-state as:
‘ N(t) = A + B sin(wt - as) (4-3)
, A and 8 represent modulation parameters of the emission waveform and ¢> is the phase
‘ shift between excitation and emission. When the sample is excited with sinusoidally
1 modulated light, the rate of decay of the excited state becomes:
1
1
‘ %i” = - } Nlt) + f(t) A (4-4)
Substitution of (4-3) and (4-2) into (4-4) yields:

‘The temporal dependence of the excited-state can also be expressed as:

%§tl = (08 coslwt - a) (4-6)

 

 

After expandir

and the consta

The relation 1

(4-8) and sub:

That for mod
yield:

From (4.7), l

Orin directa

Hence the ex

01 the demod

Onea

the ElatiVe

fieunncy do

 

83

After expanding the sine and cosine functions, equating coefficients of sin wt, cos wt

and the constant terms, we obtain from (4-5) and (4-6):

1
alflA-o do
1 .
w 0034: - : sm¢~ = 0 (4-8)
. 1 b
w srn¢ + : cos <13 = E (4-9)

The relation between excited-state lifetime and phase shift is obtained by rearranging
(4-8) and substituting:
sin¢>
cos¢

 

= temp = (or, (4-10)

That for modulation is obtained after squaring (4-8) and (4—9) followed by addition to

yield:
2 12-22
to + [T — 8 (4-11)
From (4-7), A = ar and (4-11) can be expressed as:
_ fl _ 2 2 1
m-b/a—(1+wrm)2 (4—12)
Or in direct terms of rm:
1
1 1 3
Tm = ;[—n—‘IE- 1] (4-13)

Hence the excited-state lifetime can be determined directly from either the phase shift

or the demodulation factor.
One additional topic that should be mentioned here is the fundamental limit on

he relative standard deviation (RSD) in an excited-state lifetime determined with

"requenc domain techni ues. A simple relationship between the RSD of an excited-
y q

1—._

 

 

 

 

state lifetime

derived as fol]

Writte:

The relative

variables d ar

After taking

Multiplying "

approaches (

Taking s‘lua

 

84

state lifetime determined from phase shift data and the absolute phase shift can be

derived as follows.

Written directly in terms of the excited-state lifetime, (4-10) becomes

tand:
r = —

p w (4-14)

The relative uncertainty in 1,, can be determined by propagation of error in the

variables d> and w:

 

6 tamb] 2 6 [tanqb] 2 2
—l— A 2 + — — Aw
A07 2 = (NS w ¢ do) w 4_15
2 2 ( )
7p [tango]
to

After taking partial derivatives the equation becomes:

4
sec (b tan2¢
713(1):! + Tsz

 

 

 

= tan 245 (4-16)
0,2
Multiplying by (oz/0:2, we have:
4 tanqu 2
sec ¢ A¢2 + 2 Aw
= w (4'17)
tan2¢>

This expression can be further simplified because as w approaches co , (tanzo / (02)
approaches 0 and l -

= W (4-18)

Taking square roots to obtain the RSD:

v m (4-19)

O'T-

Tp tan 45

 

 

And from the (

Upon inspectit
One can also

graphically in

IM‘

Figure 4-1,
determined
an RSD in 1

figul'e, One

shift is be“

A s
dependenct

85

And from the definitions of secant and tangent, we can finally write:

EE_J__ _
1p _ sin d> cos «b (4 20)

Upon inspection we find that as d) approaches 0° or 90°, the RSD in 1,, approaches 00.
One can also estimate the RSD in 1,, for a known 045,. Such an estimation is shown

graphically in Figure 4-1 for a typical value of 0.2 degrees for ties. According to this

 

 

121
81
A
55
o
tn
0:
4 _
1
O I l l I If *I
O 15 30 45 60 75 90

Phase Shift (deg)

Figure 4—1. Plot of the relative standard deviation (RSD) in an excited-state lifetime
determined from phase shift data as a function of the phase shift. This plot is based on
an RSD in the phase shift of 0.2 degrees.

figure, one should arrange the experimental parameters such that the measured phase
1 shift is between 10 and 80 degrees. to obtain an RSD of two percent or less
A similar approach has been attempted for the RSD in the excited-state lifetime

.dependence on the demodulation factor, but a simple relationship has not been found.

 

However, plot

similar to Fig

A sim]

 

1781116 4-2, :

dependence ,
medulator d;
Wavelength s
is sPlit with
”811118 light

 

86

However, plots of the RSD in 1 as a function of m have been constructed and they are

similar to Figure 4-1 with respect to shape and relative value of the RSD in 1'.
REVIEW
A simplified apparatus for FDF experiments is shown in Figure 4—2. A time

MODULATOR SAMPLE

 

cw LIGHT SOURCE A

 

 

 

 

 

OSCILLATOR
WAVELENGTH

SELECTOR

 

 

 

 

 

REFERENCE

V‘_‘

 
   

DETECTION
ELECTRONICS

 

 

 

Figure 4-2. Simplified apparatus for frequency domain fluorescence.

dependence is impressed on the output of a continuous wave light source by a
modulator driven at a radial modulation frequency w by an appropriate oscillator. A
wavelength selector may or may not be used in the excitation path. The exciting light
38 split with a portion directed toward a reference detector while the remainder of the
exciting light continues to the sample. Fluorescence from the sample is collected with

Lppropriate collection optics and passed through a wavelength selector before reaching

 

 

the sample dr
signals by the

A nee:
modulated s01
Pockels cells
optic (A0) r
modulators [
devices, esp
frequencies 1
end of the vi
modulators I
value for 11g
modulation

modulated 3

One
be generate
3171118041 we
fu“(laments
Although 11
pulses 0ij
high repet
°°mp0nent
modes are

3- Noted

87

the sample detector. The frequency domain parameters are derived from the two

signals by the detection electronics.

A necessary component of an FDF instrument is a modulator or an intrinsically
modulated source. Various electro—optic (E0) modulators such as Kerr cells [6,7] and
Pockels cells [6,7] have been used to modulate a continuous wave source. Acousto-
optic (A0) modulators such as those of the Debye-Sears [7,8] type and solid state
modulators [9,10] have also been used for this purpose. Although some of these
devices, especially those of the 130 family, have been used to modulate light at
frequencies of several GHz [6], they are somewhat restricted to light toward the red
end of the visible spectrum and the near infra-red. The modulation bandwidth for those
modulators that do not significantly absorb light at ca. 400 nm is much less. A typical
value for light of this wavelength is 200 MHz. A means of achieving a much higher
modulation bandwidth for many regions of the spectrum is to use an intrinsically

modulated source.

One of the basic concepts of Fourier theory is that any complex waveform can
be generated by summing a sufficiently large number of harmonic components. Hence
a pulsed waveform contains frequency components that are higher in frequency than the
fundamental frequency of the waveform. Consider the output of a mode-locked laser.
Although they are expensive and not widely available, mode-locked lasers can generate
pulses of light that are less than 100 ps FWHM. These pulses are generated at a very
high repetition rate (e. g., 100 MHz) and contain a large number of harmonic
components. The temporal output of a mode-locked laser in which 50 longitudinal
modes are phase-locked and its frequency spectrum are shown on page 88 in Figure 4-

3. Note that the temporal output consists of very narrow pulses and that the frequency

Intensity

 

 

Figure 4
Spectrurr
longitudf
Spectrun
longitud
the "peg
01 the t

Figure

 

 

 

 

 

42‘
g Q 1111
E 5 '11‘11
0— ‘ '11 .
' ' lllllllllllllllh
Time Frequency

Figure 4-3. Calculated temporal output of a mode-locked laser (left) and its frequency
spectrum (right). These waveforms are for a mode-locked laser in which 50
longitudinal modes are phase-locked.

 

spectrum shows a large number of harmonic components, 50 to be exact - one for each
longitudinal mode. One can also see evidence of the higher harmonics at the base of t
the "peaks" in the temporal output. To emphasize this, an enlarged view of this region

of the temporal output is shown in Figure 4-4. The higher harmonics are readily

 

 

 

Figure 4-4. Enlarged View of the temporal output of a mode-locked laser.

 

 

 

apparent in
greater than .

several GHz.

Folio
Dushinsky
Szymanowsl
In that year,
accuracy of
excited-state
factor. The
frequency d
instruments

number of 1

cofrelation
Shift and d1
shift betwt
fluorophort
0f 45 deg“

fre‘Illency

hetemdynl

89

apparent in this illustration. Quite often the number of longitudinal modes is much
greater than 50, and the useful harmonic content of a mode-locked laser can extend to
several GHz.

Early Work

Following the work of Gaviola [l], the theory of FDF was refined by
Dushinsky [11] and minor improvements in instrumentation were made by
Szymanowsky [12]. Modest improvements were made by various workers until 1953.
In that year, Schmillen [13] reported a crude heterodyne technique which improved the
accuracy of phase shift measurements and Birks and Little [14] reported the first
excited-state lifetime measurements from both the phase shift and the demodulation
factor. The first review of FDF was that of Bonch~Bruevich [15] in 1956. By 1961
frequency domain parameters could be measured directly at 10 MHz [16], but FDF
instruments were limited to measurements at a single frequency or one of a small

number of frequencies.
Modem FDF

In 1969, Spencer and Weber [17] published a landmark paper on cross-
correlation detection. This method greatly reduces the frequency at which the phase
shift and demodulation factor are measured. It is difficult to directly measure a phase
shift between two signals at the high frequencies used in FDF. Consider that a
fluorophore that has an excited-state lifetime of ten nanoseconds will have a phase shift
[of 45 degrees when the modulation frequency is 16 MHz. Cross—correlation is actually
frequency mixing of two different frequencies. This process is also known as

heterodyning. In cross-correlation detection, the gain of the detector is modulated at a

 

 

 

II.

frequency slit
detector esser
light source,
and different
exciting ligh
same source
present in b
signal can is
easier to me
frequencies

correlation,

Ano
Their worl
resolved“ 0
in the liter;
the techniq
rt‘vsolution.
P111, is mr
integrated.

¢D, is val-i;

Where A i
“Citation
moduhtio

when Flt)

90

frequency slightly higher than that of the exciting light. The resulting signal from the
detector essentially contains four frequencies: one at the modulation frequency of the
light source, one at the modulation frequency of the detector, and one at each the sum
and difference of the two modulation frequencies. The signals used to modulate the
exciting light and the gain of the detector must be phase locked or derived from the
same source to eliminate phase. The phase shift and modulation information are
present in both the sum and difference frequency. Simple low pass filtering of this
signal can isolate the difference frequency which can be as low as 30 Hz. It is much
easier to measure the phase shift and demodulation factor at 30 Hz than it is at the high
frequencies used in FDF. Because of the greater accuracy and simplicity of cross-

correlation, almost all FDF instruments reported since its introduction use it.

Another major advance was initiated in 1970 by Veselova and coworkers [18].
Their work marked the beginning of a detection methodology known as "phase
resolved" or "phase sensitive" fluorescence. The two names have used synonymously
in the literature, and it has been recommended [19] that the former be used to refer to
the technique and to spectra, and the latter to the detection used to accomplish phase
resolution. In phase resolved fluorescence, the AC portion of the detector response,
P(t), is multiplied by a digital, periodic waveform of the same frequency, F(t), and
integrated. The phase angle between the periodic waveform and the detector response,
1’0. is variable from 0 to 360 degrees, and the integrated signal follows:

F(daD) = A m“ m cosupD - <4) (4-21)
where A is the observed steady-state emission intensity, e is the phase shift between
excitation and emission, mex is the excitation modulation, and m is the emission
modulation. The integrated signal reaches a maximum when ¢> equals (to. That is

when F(t) is exactly in phase with Pit). The methodology is illustrated in Figure 4-5.

 

 

 

P(t)

F(t) -

 

Figure 4-5.
represents t1
0 degrees 0
phase (right
intermediate

' By adjustiny
fluorophore
that detecto

CXponential

Sint
113], numt
include ins
instrument
9’ 0f pols
television
Optics [28
exPollentia
371 have
fre(Illency
resolved [

91

P(t) __I_——I__ _IT—I_ _I——_I_

W111: but pen

Figure 4-5. Integration scheme used in phase resolved fluorescence. The shaded area
represents the integrated signal. As the phase shift between F(t) and P(t) is varied from
0 degrees or in phase (left), to slightly out of phase (middle), to 90 degrees out of
phase (right), the integrated signal decreases from the maximum value (left) to an
intermediate value (middle) to zero (right).

By adjusting the phase shift between F(t) and P(t) to 90 degrees beyond that of a given
fluorophore, the signal from that fluorophore can be eliminated from the total signal at
that detector phase angle. One can use this means of "phase nulling" to resolve multi-

exponential fluorescence decays or mixtures of fluorophores.

Since the introduction of cross-correlation [17] and phase resolved detection
[l8], numerous improvements in instrumentation have been reported. Some of these
include instruments capable of a continuously variable modulation frequency [20-23],
instruments that use the harmonic content of the output of a. mode-locked laser [24, 25]
or of pulses of light generated by a synchrotron [22]. Instruments based on a UHF
television tuner [26], gain-modulated silicon avalanche photodiodes [27], and fiber
optics [28] have also been reported. Several reports of the resolution of multi-
exponential fluorescence decays [29-33] and of complex mixtures of fluorophores [34-
37] have appeared in the literature. At present, the highest reported modulation
frequency is 10 GHz [38], and mixtures containing up to six fluorophores have been

resolved [37]. The most recent review of FDF was that of McGown and Bright [19].

 

 

The
FDF provid
figures in re
chosen for
instrument
quenching 1
specific ins
the literatur
Hieftje-Hat

As
state lifetir
this decisi
insontanet
Capable of
Possible, 1
been fully

92

The direct relationship between the phase shift and the demodulation factor in
FDF provides a means to determine an excited—state lifetime to two or three significant
figures in real or near real time. Because of this unique ability, FDF was the technique
chosen for determining excited-state lifetimes in the instrument described here. This
instrument corrects for the reduction in fluorescence intensity caused by collisional
quenching using the Hieftje—Haugen method [2]. As there are choices with respect to
specific instrumental configuration and detection methods, specific instruments from
the literature are surveyed in the next section with an emphasis on application to the

Hieftje—Haugen method [2].

Survey of Previous Instruments

As mentioned above, FDF was chosen as the method for determining excited—
state lifetimes with the Hieftje-Haugen method [2]. Although the primary criterion for
this decision is the ability of FDF to determine an excited state lifetime virtually
instantaneously, not all molecular fluorescence spectrometers designed for FDF are
capable of this. While some workers have strived for the highest modulation frequency
possible, other improvements in the technique such as rapid data acquisition have not

been fully developed.

Before considering specific instruments, a choice needs to be' made concerning
the detection methodology, phase resolution or direct measurement of the phase shift
and demodulation factor. The chosen method must be consistent with the goal of
determining the excited state lifetime instantaneously. Although the method of phase
resolution is very powerful for the determination of multi-exponential fluorescence
decays and mixtures of fluorophores, the requirement of scanning the detector phase

angle restricts this detection method from making excited—state lifetime measurements

 

 

 

in real time

automated, 1

Alth
common cc
shift and d1
FDF instru:
are lamp ar
uses two pi
and the oth
only Opera
cross-cone
frequency
noise, Ar
integer nut
during on:
actonlplisl
dYHOde. f
01 mOdula
an LED d

of the n“
modulator
frequenci.
PMTs, .

W0rkers 1

93

in real time. Since direct measurements of phase shift and demodulation can be

automated, this means of detection was chosen.

Although it is not capable of determining an excited-state lifetime in real time, a
common commercial frequency domain fluorometer that directly measures the phase
shift and demodulation factor is the Aminco 4800 phase fluorometer. Many modern
FDF instruments are in fact upgraded Aminco 4800s. This spectrometer uses a xenon
arc lamp and Debye-Sears modulators [8] to generate intensity modulated light. It also
uses two phase-locked frequency synthesizers, one to drive the Debye—Sears modulators
and the other for cross-correlation detection [17]. Although the frequency synthesizers
only operate at three frequencies, they are stable and of sufficient resolution that the
cross-correlation frequency in this instrument is 30 Hz. Setting the cross-correlation
frequency so that it is an integer fraction of 60 Hz helps minimize AC power line
noise. Any effect the power line has will exactly cancel at, say, 30 Hz because an
integer number of cycles of the AC line frequency, two in this case, will have occurred
during one period of the cross-correlation frequency. This method of detection is
accomplished within the PMTs by adding a modulating voltage to the bias at the second
dynode. This instrument uses a ratio digital voltmeter to determine intensity and depth
of modulation and a phase meter to measure the phase shift. The output is presented on

an LED display, and the user then calculates the excited state lifetime.

Many workers have replaced one or two, or many parts of this instrument. One
of the most common modifications is replacing the xenon arc lamp and Debye—Sears
modulators with a laser and either an A0 or E0 modulator. The number of modulation
frequencies can be increased by replacing the frequency synthesizers and re-wiring the
PMTs. While these modifications can increase the modulation bandwidth, very few

workers have attempted to improve the detection electronics. Indeed, where workers

 

 

 

have replac
controllable

process.

Altl
and coworl
capabilities
and the (11
reported bj
mode-loch

detection e

I REFER

@

Figure 4.
reported
lntegrator

the figm-|
sample

9

deSCribe

94

have replaced the frequency synthesizers with completely variable and computer
controllable commercial units, few have bothered to automate the measurement

process.

Although it was reported a few years ago, the instrument described by Alcala
and coworkers [24] is the most advanced frequency domain fluorometer with respect to
capabilities for rapid data acquisition. With the exception of the modulated light source
and the detection electronics, it is similar to the Aminco 4800. The instrument
reported by Alcala and coworkers [24] uses the harmonic content of the output of a
mode-locked laser as an intensity-modulated light source. A simplified version of the

detection electronics used in their instrument is shown in figure 4-6. As is evident in

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

INT
{> V I Intensity
“r in
REFERENCE PMT Jl> R r/ v . n
START S
U NCGJNTS
IPF 2CD N ¢ ¢
STOP T
E
R
M? INT
[ SAMPLE PMT } 41> R 41> v . n
INT
‘ {> v «Intensity

Figure 4-6. Simplified diagram of the detection electronics used in the instrument
reported by Alcala and coworkers [24]. AMP: amplifier, BPF: bandpass filter, INT:
integrator, ZCD: zero-crossing detector, R: rectifier.

the figure, the instrument has two identical channels, one each for reference and the
sample, for deriving the frequency domain information. In order to more clearly

describe the operation of the detection electronics, one of the channels and the signal at

 

various por

 

 

in
A

Figure 4-1
by Alcala
oPeration.
rectifier, 2

Current W]
the figure
determine
and the 1
integration
connected
signal fl’ol
a co“liter
crossingI

SIOpPing .

 

95

various points in it is shown in Figure 4—7. The signal from the detector, a PMT, is a

INT V or intensity
I\ g
V v r—

/ o T V or modulation
> w—e—o

 

 

 

 

 

 

 

R INT v
BPF o—-—-
—1>4v W T
O
v /V\/ To Counter
r
o r ZCD
v
I\ o
r/ . T

Figure 4-7. Single channel of the detection electronics used in the instrument reported
by Alcala and coworkers [24]. The signal is shown at key points to illustrate its
operation. I/V: current-to—voltage converter, BPF: bandpass filter, INT: integrator, R:
rectifier, ZCD: zero-crossing detector.

current which is converted to a voltage by a current-to-voltage converter. As shown in
the figure, the signal contains an AC and a DC component. The DC voltage is
determined by integrating the signal. The AC signal is isolated with a bandpass filter,
and the depth of modulation is determined by rectifying the signal followed by
integration. The determination of the phase shift is accomplished with a counter
connected to both channels. A zero-crossing detector is used to generate a digital
signal from the filtered AC signal. These digital signals provide start and stop inputs to
a counter (Figure 4-6). With the edges of the digital signal corresponding to the zero
crossing point of the AC waveform and the reference and sample signals starting and

stopping the counter, the phase shift is proportional to the number of counts. The

 

 

 

 

counter cou
measured p

are acquire

Aftt
determinati
analysis. J
modulatior
the cell 111
. lifetime.
availabler
Haugen [2

A MO?

quantity 1
assemble
these are
resolutio:
radiofre
One alst
Althougl
all, oft

collld n(

96

counter counts oscillations from a 10 MHz oscillator providing high accuracy in the
measured phase shift. All four of the analog voltages and the contents of the counter

are acquired with a computer, automating the measurement and calculation processes.

After calibration of the instrument, this design allows a virtually instantaneous
determination of an excited state lifetime through computer data acquisition and
analysis. Since the reference is used to provide a continuous signal of the excitation
modulation in addition to the phase reference, one merely needs to place a sample in
the cell holder and "push a button" to obtain a direct readout of the excited-state
lifetime. Because the information needed to directly determine the ratio of HT is
available with this design, it was chosen for real-time implementation of the Hieftje-

Haugen [2] method of correcting for the effects of collisional quenching.

A MOLECULAR FLUORESCENCE SPECTROMETER THAT CORRECTS
FOR THE EFFECTS OF COLLISIONAL QUENCHING

In order to construct an instrument capable of a real—time, direct readout of the
quantity I/r using the overall design reported by Alcala and coworkers [24], one must
assemble several expensive components not found in a typical fluorometer. Among
these are an intensity modulated light source, a frequency synthesizer having 0.1 Hz

resolution and capable of synchronization to the modulation of the light source, and a
I radio-frequency amplifier capable of delivering several watts of power to a modulator.
One also needs access to test and measurement equipment to aid in construction.
Although funds could not be obtained to purchase the needed equipment, most, but not
all, of the above-mentioned components were acquired. The need for one item that

could not be obtained, the frequency synthesizer, was satisfied by constructing a crude,

 

 

 

 

single fro

synthesizer

Th.
many moc‘
harmonic
dumped d
reported 1
accomplis
crosscorn
coworkers
detector is
of mixing
modulatio
correlatioi
811111-th
r91Illire a
iHStIumen
it Grated,
c(il'feltttior
f1'e‘Illency

measuring

97

single frequency device. The limitations imposed by the constructed frequency

synthesizer result in a less accurate method of measuring the phase shift.

General Features

The instrument constructed to directly determine the quantity HT is similar to
many modern frequency domain fluorometers. A dual channel instrument, it uses the
harmonic content of the output of a synchronously pumped, mode-locked, cavity
dumped dye laser as an intensity modulated source, and is similar to the instrument
reported by Alcala and coworkers [24]. Unlike their instrument, cross-correlation is
accomplished external to the PMT with a doubly balanced mixer. This method of
cross-correlation detection was used in an instrument reported by Lakowicz and
coworkers [25]. In general, cross-correlation accomplished by modulating the gain of a
detector is the preferred method, because one can realize almost theoretical efficiency
of mixing when a PMT used. This is because one can achieve nearly perfect
modulation with a PMT. The use of a doubly balanced mixer to generated the cross-
correlation signal is less desirable because these devices are less efficient mixers than
gain-modulated PMTs. External cross-correlation was chosen because it does not
require a high power RF amplifier. Although such an amplifier was obtained for this
instrument, it was not suitable because of ground loop problems and other interferences
it created. The instrument described here is unique in that it uses a very high cross-
correlation frequency of 325 kHz and is restricted to operation at a single modulation

frequency, 76.5 MHz. Because of the high cross-correlation frequency, a method for

measuring the phase shift was developed.

 

 

 

Ad

Mode

 

 

 

98

A diagram of the instrument is shown in Figure 4-8. As mentioned earlier, a

 

 

 

 

~ Mode—Locked NszAG Laser SHG Dye Laser

 

 

 

 

 

 

 

 

 

Reference SHG

DBM BBH PMT Mono :1

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MLD FS PS

 

 

 

 

 

oafingH PMT <11>fJ

Mono Sample

 

 

 

 

 

 

Personal Analog and
Computer Digital Electronics

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 4-8. Diagram of the constructed frequency domain fluorometer capable of
directly determining the quantity I/r. SHG: second harmonic generator, MLD: mode-
locker driver, FS: frequency synthesizer, PS: power splitter, DBM: doubly balanced
mixer, BB: black box.

mode-locked, synchronously pumped, cavity dumped dye laser is used to provide an
intensity modulated source. This is the same laser used in the SPT apparatus described
in Chapter three. Light from the dye laser is split with a portion directed to a
scattering solution and the remainder to the sample. If excitation in the ultra-violet is
needed, the laser light can be frequency doubled with a second harmonic generator.
Scattered laser light is filtered with a small monochromator and detected with the
reference PMT. Fluorescence from the sample is collected and passed through a
monochromator before reaching the sample PMT. The outputs from the PMTs contain
DC and AC components. These are separated in a black box with the DC portion of
the signal fed directly to the detection electronics and the AC portion fed to the RF
input of a doubly balanced mixer. The frequency that is mixed with the signal from the
PMT is synthesized in the frequency synthesizer which derives its frequency reference

 

from the 11
power spli'
generated 1
the analog
coworkers

five DC Vt

presented
The frequt

AC and]:

Figure 4.5
Slgnal fro
mlxer.

 

99

from the mode-locker driver. The output from the frequency synthesizer is split by a
power splitter and fed to the LO port of the DBM. The cross-correlation frequency is
generated in the DBM and accessed at the DBM's IF port. This signal is analyzed with
the analog and digital electronics. As in the instrument reported by Alcala and
coworkers [24], the intensity and the frequency domain information are contained in

five DC voltages that are acquired with a laboratory personal computer.
Detection Electronics
The detection electronics are discussed in two stages. A general overview is
presented first, and this is followed by a more detailed description of the circuits used.
The frequency synthesizer is fully described in Appendix 3.

BLACK BOX AND MMR

The contents of the "black box" indicated in Figure 4-8 and used to separate the

AC and DC portions of the signal from the PMTs are shown in Figure 4-9. The AC

 

 

 

 

 

_ 76.8243 MHz
BLACK 80)? from F5
470 pF RF Amp LO
1 m I—W ..
1 mH DC

 

1/V

—->— To Computer

Figure 4-9. Diagram of the circuit used to separate the AC and DC components of the
signal from the PMl‘s. I/V: current—to-voltage converter, DBM: doubly balanced
mixer.

 

 

 

 

and DC 81,
capacitor (1
the laser or
tovoltage 1
acquisition
amplified ‘
mentioned
the signal

the light is

. laser.

Th
extracted

nuclear i;
NIMS are
converted
correlatio
1efined a
does not
circuit th
The Pha.r
describe:
by a 29,

100

and DC signals from the PMT are separated with an inductor (passes DC) ar
capacitor (passes AC). The DC signal represents the intensity of light, either the
the laser or of fluorescence, and is converted to a voltage and amplified with a curr
to-voltage converter. This voltage is acquired with a personal computer in which a
acquisition board is installed. The AC portion of the signal is high frequency an
amplified with an RF amplifier before being fed to the doubly balanced mixer.

mentioned previously, the cross-correlation frequency is generated in the mixer f
the signal from the frequency synthesizer, 76.8243 MHz, and the frequency at w
the light is modulated, 76.5 MHz - a harmonic of the fundamental repetition rate of

laser.

NIM ELECHRONICS

The output of the mixer is filtered and the frequency domain informatio
extracted by the rest to the detection electronics which are housed in two single-w
nuclear instrumentation modules (NIMs) situated in a NIM bin. The contents of
NIMs are in shown on page 101 in Figure 4-10. The signals from the mixers a.
converted to voltages and amplified with current-to—voltage converters. The cross
correlation signal is isolated with a bandpass filter that is perhaps more appropriatel
termed a tuned amplifier, as it provides gain over a narrow band of frequencies
does not attenuate outside the passband. The depth of modulation is derived fror
circuit that provides a DC voltage at the peak voltage of the cross-correlation sig
The phase shift is measured using a time-to-amplitude converter. This device
described in Chapter three. The cross-correlation signal is converted to a digital sig
by a zero-crossing detector [39] such that the rising edge of the digital sig

corresponds to the - to + transition of the cross-correlation signal. The refere

 

REFE

Figure 4-1
converter,
crossing d1

signal pros
The heigh
136an tl
them. Th:
and-hold :

equipped y

Mt
have cros:
fre‘lllency
this instru
“804 in an

the DC 81

101

 

 

 

 

 

 

 

 

 

 

Vpeak to DC
{>f = Va mLaur
l/V BPF
REFERENCE —>—->—
ZCD
11> sma
SIGNALS T 5““
FROM A V a phase
MIXERS ZCD C shift
p
l/V BPF
SAMPLE ———[>—>—
Vpeak to DC
Jj> 3 V a m Fluoruconco

 

 

 

Figure 4-10. Overview of the circuits housed in the NIMs. I/V: current-to-voltage
converter, BPF: bandpass filter, Vpeak to DC: Vpeak-to-DC converter, ZCD: zero-
crossing detector, TAC: time-to-amplitude converter, S&H: sample-and-hold amplifier.

signal provides the start input to the TAC and the signal input provides the st0p input.
The height of the pulses generated by the TAC is proportional to the time difference
between the start and stop inputs and hence is proportional to the phase shift between
them. The pulses generated by the TAC are converted to a DC voltage with a sample-
and-hold amplifier. The three DC voltages are acquired with a personal computer

equipped with a data acquisition board.

Most frequency domain instruments that have been reported in the literature
have cross-correlation frequencies near 30 Hz. While working with circuits for this
frequency is straightforward, constructing circuits for the cross-correlation frequency in
this instrument is much more challenging. Indeed, many of the operational amplifiers

used in this instrument were introduced by manufacturers in the last three years.

A schematic diagram of the circuits housed in the NIM modules and those for

the DC signals that are housed separately is shown on page 102 in Figure 4-11. The

Fig

 

.Lotuaacu .3323. a: 3 a tad notuuaacu Eng-u u: Ed .- 5..- bansu Luann on». so not; .32. to: 389521 0.3
uvfiufiu 13.9.03... .2 .159. unit-cue nun—5 unvon< unqun 00.22 9:. uvlugfi 33.953: .130: twigzro Inc.:: Utes-u one «.32. .333 052.1 0.9:.
U; .95qu (also 0.3 25.... Uncut: ugotuanou coinage“. 39:5 to non?) L930 mega... utovufiou i: .38: a... 2 to I393)... :23 .- toga-c

“who:

“Math «09.5
9. op

cl» >u+
“ n Q4 can I

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1/ dddd
o 05 tons
a .20!
H E... _ .n E. 8. S
0.5.5.3303 no .1
u: _ <€>¢2 - nan) . who i.
29.
2 {In unit I «30) . 8
m (can .3 .3 Eu

(m
.C
m
U
m
m
m
e
d
e
.m
f
0
a! 3 m
.m:
d
c
.U
a
m
e
h
C
S
L
1..
4
m
u
g
.1
F

 

 

jllll

/— 14: 6! Nu
9. 3 “.1
A a
n «35.3.
a! an A 693 68 cu W

 

 

can I;
u¢h 0h

D
D
D
’
AL
VV" >

 

 

V'V'

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Ti. Z - E. .xouhai.
dddd 0
«85.5 S... _ _ S. — / mm:— 1:9...9.
160n> I #3) 0L <><p<><> “T 38 u Edd
09.... “v 49... ha 8 T
8. am “"5. n rill.
. 5:: «9:32..-
1.1-did
Ti 9. a
.2:
dbdbdbd’
e. 2

 

 

 

 

Lot->80 Blevins-an)

 

 

 

 

current-to-
left comer
through a
an inverti
feedback 1
the input (
07, is not
W [40].

a measure

the top of
"black bo:
also used
gain-band
[40]. Thi
first 31npl
the nextc
amplifiers
8458. Th
Power bar
[41]. Its

are not qu

103

current-to-voltage converter and amplifier for the DC signals are shown in the lower
left corner of the figure. The DC current from the "black box” is shunted to ground
through a 10 k9 resistor. The voltage developed across this resistor is amplified with
an inverting amplifier having a gain of 1,000. A capacitor is placed across the
feedback resistor to attenuate the high frequency response of the amplifier. Trimming
the input offset voltage of this amplifier, a Precision Monolithics (Santa Clara, CA) OP
07, is not necessary because of this amplifier's extremely low input offset voltage of 25
pV [40]. This circuit generates 10 mV per nA of input current, and is used to provide

a measure of the intensity of the sample or of the reference.

The circuit for deriving the depth of modulation of the AC signal is shown at
the top of the figure. Like the circuit for the DC signal, the input current from the
"black box" is shunted to ground through a 10 k0 resistor. An inverter with gain is
also used to amplify the voltage created by the IR drop, but the gain is five because the
gain-bandwidth product of this amplifier, a Precision Monolithics OP 27, is eight MHz
[40]. This amplifier is followed by an inverter of unity gain. It is needed because the
first amplifier does not have sufficient power to drive the active bandpass filter which is
the next component in the circuit. The amplifiers used for the inverter and all other
amplifiers except the first one in the circuit are Analog Devices (Norwood, MA) AD
8458. This amplifier was chosen because of its excellent AC characteristics. The full
power bandwidth of the AD 845 is 16 MHz and it is stable when operated at unity gain
[41]. It was not used as the first amplifier in the circuit because its DC characteristics

are not quite as good as those of the OP 27 .

To
with a dasl
from the t
necessarily
large numl
that the cc
signal afte
searching
Daryanani
negative a
and a pea
design of
that the g;
and cutof
was foun.
capacitor:
Variable 1
Optimize
285 0 an
Figure 4.

 

 

 

104

BANDPASS FILTER

To aid in identifying it in the figure, the bandpass filter is labeled and outlined
with a dashed line. A bandpass filter is necessary to isolate the cross-correlation signal
from the other frequencies generated in the mixer. The signal from the mixer is
necessarily rich in frequency components since the signal from the PMT contains a
large number of harmonics. The filter must isolate the cross-correlation signal such
that the contribution from the other frequencies comprises only a small fraction of the
signal after the filter. A filter was developed to satisfy this need after a great deal of
searching and experimentation. The filter design chosen is described in a book by
Daryanani [42]. It is a second-order bandpass filter with gain derived from both
negative and positive feedback. The filter was designed with a quality factor, Q, of 20
and a peak gain of 40 dB. A circuit simulation program [43] was used to aid in the
design of this filter. Although amplifiers used in active filters are usually chosen such
that the gain-bandwidth (GBW) product it at least ten times the product of the filter's Q
and cutoff frequency [42] (a GBW of 65 MHz would be needed according to this), it
was found that the AD 845, which has a GBW of 16 MHZ was sufficient. Mica
capacitors were used because of their nearly ideal behavior at high frequencies. The
variable resistors were initially set to values predicted by theory, and then adjusted to
optimize the performance of the filter. The final values of the variable resistors were

285 0 and 14.26 k0. A Bode plot of the constructed filter is shown on page 105 in

Figure 4-12.

 

 

 

Figure 4-

T
convene
enclosed
is quite :
t0 the nr
the orig
Phase 01
the non.
voltage
reslions
Convert.

measure

105
50—

4o—

30.

20-

Gain (dB)

 

 

Y I ' I ' I r I . I ' r *fi
0 100 200 300 400 500 600 700
Frequency (kHz)

Figure 4-12. Bode plot of the bandpass filter.

PEAK DETECTOR

The signal from the active filter is fed to two circuits. These are a VPm-to-DC

converter and a zero-crossing detector. The V -to—DC converter is also labeled and

put
enclosed with dashed lines. This circuit was described in a book by Hufault [44], and
is quite simple. The diodes in the feedback loop of the first amplifier limit the output
to the negative half of the waveform. The output from this amplifier is combined with
the original waveform in a summing integrator. Since the first amplifier shifts the
phase of the waveform by 180 degrees, its output exactly cancels the positive half of
the non-limited waveform. The value of the variable resistor is set so that the output
voltage equals the peak voltage of the original signal. The circuit works well, but its
response is slightly dependent on the signal level. Because of this, each Vpuk-to-DC
converter was calibrated, and the results of the calibration were used to correct

measurements made with this circuit.

The
upper port
a compara
comparato

of the con

Th
generated
signals an
to its mar
the analo
sample-a1

output pp

A
Was user
order to
least 25.
condition
The cirr
S&H, a

(:0um

106

ZCD AND TAC

The zero-crossing detector is the only other part of the circuit shown in the
upper portion of Figure 4-11 that has not been discussed. It consists of a log amp and
a comparator [39]. The log amp provides a nearly constant amplitude waveform to the
comparator and minimizes saturation and time shift with signal amplitude. The output

of the comparator is used to provide start and stop inputs to the TAC.

The TAC was described in the chapter three. The amplitude of the pulses
generated by the TAC represents the time difference between the reference and sample
signals and hence is indicative of the phase shift. The TAC output pulse width was set
to its maximum value of 10 us. Since this is much faster than the conversion time of
the analog-to-digital converter on the data acquisition board housed in the computer, a
sample-and-hold (S&H) amplifier was used. The circuit used to sample and hold the

output pulses from the TAC is shown in the lower right portion of Figure 4-12.

SAM PLE-AND—H OLD

A fast S&H amplifier, a National Semiconductor (Santa Clara, CA) LF 398A
was used. The minimum acquisition time of this S&H is four as [45]. However, in
order to operate properly, the signal applied to its logic input must return to ground at
least 250 ns before the analog signal level changes [45]. In order to satisfy this
condition, a circuit was designed to provide an appropriate logic signal to the S&H.
The circuit consists of a comparator and a monostable. In addition to being fed to the
S&H, the signal from the TAC is fed to through a 1 k9 resistor to the input of a fast

comparator, a National Semiconductor LM360. The input to the comparator is shunted

to ground
reduces th
height. I
minimize
monostabl

the S&H.

Tr
informatit
ratio I/I' i
can have
two IBM
cOmpatib
and mod-
the S&H
Common
acquires
This is '

interrupt

 

 

107

to ground through a diode. This limits the signals that reach the comparator input and
reduces the dependence of temporal response of the comparator on the input pulse
height. The threshold voltage is set slightly above the noise level in the circuit to
minimize spurious signals. The output from the comparator is used to trigger a
monostable (74121) which is configured to generate a 7 p.S pulse which then triggers
the S&H.

Data Acquisition

The five analog voltages derived from these circuits represent all of the
information needed to measure fluorescence intensity and excited—state lifetime. The
ratio HT is easily calculated by the data acquisition software. The five analog voltages
can have values from zero to about 11 volts. All of these voltages are acquired with
two IBM Data Acquisition and Control Adapters (IBM DACAs) housed in an IBM-
compatible personal computer (Zenith Data Systems). While the voltages for intensity
and modulation are essentially constant and can be acquired at will, the voltage from
the S&H decays slightly with time from acquisition - a behavior termed droop that is
common with S&H amplifiers. To reduce this source of error, the output voltage is
acquired immediately after S&H acquisition with an interrupt service routine [45].
This is triggered by the Q output of the monostable which is used to generate an

interrupt request pulse on the PC's bus.

Other Components

Before discussing the operation of the instrument, the other components used should be

mentioned. Several of the components used in this instrument were also used in the

SPT apparatus described in Chapter three. These include the light source and the

 

 

'7

equipment
2000). T
Johnson.
(NeWPort,
(Metucher
signal, an
fluorescen
IP28. Tl
biased usi

. coworker:

was accor
Boca Rat-
Systems,

(Redmonr
5.1. 'n
fluorescer
for COITec

the subjec

and refer.
monochrC
tonghly e
With exci

detectOI's,

108

equipment used to monitor it, the TAC, and two RF amplifiers (Minicircuits ZFL—
2000). The doubly balanced mixers (Minicircuits ZAD-l) were donated by Kermit
Johnson. The collection optics consisted of two achromatic doublets, f/n = 2 and 6.8
(Newport, Fountain Valley, CA). The monochromators were an Instruments SA
(Metuchen, NJ) model H10, used to filter the scattered laser light for the reference
signal, and a Heath-McPherson (Benton Harbor, MI) model EU-700, used to isolate
fluorescence from the sample. The PMTs were Hamamatsu (Bridgewater, NJ) model
1P28. These were housed in simple aluminum boxes with shutters. The PMTs were
biased using a circuit similar to that used in the instrument reported by Alcala and
coworkers [24] and a dual high voltage supply (Bertan model 362). Data acquisition
was accomplished with two IBM data acquisition and control adapters (IBM DACAs,
Boca Raton, FL) installed in an IBM-compatible personal computer (Zenith Data
Systems, Benton Harbor, MI). The data acquisition software was written in Microsoft
(Redmond, WA) QuickBASIC, version 4.0 and Microsoft Macro Assembler, version
5.1. The software is menu driven and provides instantaneous readout of the
fluorescence intensity, excited-state lifetime, and the ratio I/r. It also contains routines
for correcting for sample absorption in combination with QuickBASIC for quenching -

the subject of the next chapter.

Instrument Calibration

The instrument is calibrated by placing scattering solutions in both the sample
and reference cell holders. Instrumental parameters such as PMT bias voltages and
monochromator settings are adjusted so that the DC voltages from both detectors are
roughly equivalent. Neutral density filters may also be used and are generally needed
with excitation in the ultra-violet The DC and AC signals are acquired for both

detectors, and these values are used to normalize the modulation measurements.

Calibration
from the T
voltage he
uses the r
establishes
completed
monochrOI
presses the
immediate
demodulaI
modulatiu
signal av
averaging

V01tage is

B4
excitedsi
0f 10 to :
is 370 ps

noise in
PIOducir
DC Corr
RFI at E

 

109

Calibration of the phase shift is also carried out with scattering solutions. The voltage
from the TAC is measured, a known delay is inserted in one of the channels, and the
voltage from the TAC is measured again. The software acquires these voltages and
uses the delay, entered by the user, to generate a calibration factor. This also
establishes the voltage for zero phase shift. Once these calibrations have been
completed, the instrument is ready for use. One simply tunes the sample
monochromator to the desired emission wavelength, places a sample in the holder, and
presses the ENTER key from the appropriate menu option. The user is provided nearly
immediately with fluorescence intensity, excited state lifetime from both phase and
demodulation, and two values of the ratio I/r, determined from both phase and
modulation. The time required for the measurement is a function of the extent of
signal averaging. The measurements are virtually instantaneous with no signal
averaging and require about ten seconds when the mean of 100 acquisitions of each

voltage is determined.

Because the instrument is limited to a single modulation frequency, the range of
excited-state lifetimes one may determine with it is limited. With the phase shift range
of 10 to 80 degrees suggested earlier in this chapter, the range of excited-state lifetimes

is 370 ps to 11.8 ns.
Performance

Despite a great deal of effort, under "ideal” conditions the dominant source of

noise in the instrument is radio frequency interference (RFI). With the PMTs
producing only dark current, a signal of about three volts was observed at the VM-to—

DC converter. After extensive testing, it was determined that this noise resulted from

RFI at 38.25 MHz being picked up in the cables that carry the signals from the PMTs

and from the freq
mixed with the I
correlation freque
interferes with r
operates at 38.25
the mechanism l
' driver was leaki‘
cable that carrie
the building A
immediately aft
radiated RFI w
[46]. This cab
38.25 MHz ra
shielded and ac
With semi-rigir
housed preven
once with U]
clamped to th
The thus clad
SYStem sits wi

at the 011tpnt I

The "
the itVailable
mOdulated 1]
570 nm to ‘

thlilable, '

 

110

and from the frequency synthesizer. Like the signal from the detectors, this noise is
mixed with the output of the frequency synthesizer generating noise at the cross-
correlation frequency. Since this noise is at the cross-correlation frequency, it directly
interferes with measurements on the signal of interest. Because the mode locker
operates at 38.25 MHz, it was logically the source of the RFI. It was determined that

the mechanism of the interference was two-fold. RF generated in the mode—locker
' driver was leaking into the building AC power system, and it was radiating from the
cable that carries the mode lock signal to the laser. The propagation of RFI through
the building AC power system was greatly reduced by inserting a line filter
immediately after the mode locker driver in the mode locker's power cord. The
radiated RFI was reduced by adding shielding around the cable as suggested by Ott
[46]. This cable is a 60 inch length of RG 58C/U coax that carries ca. four Watts of
38.25 MHz radiation from the mode locker driver to the laser. It is insufficiently
shielded and acts as a transmitter at 38.25 MHz. Ideally, one would replace this cable
with semi—rigid coax, but other considerations in the laboratory in which the laser is
housed prevented this. Instead, the cable was wrapped twice with aluminum foil and
once with 1/ 16th inch low inductance braid. The braid and aluminum foil were
clamped to the BNC connectors at each end of the cable with common hose clamps.
The thus clad cable was also electrically tied to the optical table upon which the laser
system sits with braid. This reduced to the level of RFI to about 100 mV as measured

at the output of the V -to-DC converter.

peak

The "ideal" conditions alluded to above concern the light source. In terms of
the available ranges of wavelengths, the laser system that is used as the intrinsically
modulated light source is quite good. From two cavity dumped dye lasers, light from
570 nm to 630 nm (rhodamine 6G dye) and 420 nm to 470 nm (stilbene 3 dye) is

available. Excitation in the ultra-violet can be obtained by frequency doubling the

 

 

output of
mode loc'
infra-red .
of the las
of the fre
the Nd:Y
dye laser
doubled .
generatio
infra-red
the case I
to pump
turbulent
cavity, 3

the S/N:

the S/N
excitatio
useftll lit
are less
this insh
a1ighmel
noise,

Perform

111

output of the rhodamine dye laser and frequency tripling the infra—red output of the
mode locked Nd:YAG laser. Light at 532 nm generated by frequency doubling the
infra—red output of this laser can also be used. However, the signal-to-noise ratio (S/N)
of the laser light used depends on several factors. From a fundamental standpoint, all
of the frequencies of light available have a S/N less than that of the infra-red output of
the Nd:YAG laser from which all of the useful frequencies are obtained. Both of the
dye lasers are pumped with light derived from frequency mixing. The noise in the
doubled or tripled light from the Nd:YAG laser follows the power of the harmonic
generation [47]. Hence the noise in the 532 nm light is the square of the noise in the
infra-red and the noise in the 355 nm light is the cube of the noise in the infra-red. In
the case of light from the dye lasers, the S/N of the output follows that of the light used
to pump the dye laser and is further degraded because the jet of dye is not free of
turbulence. The cavity dumper also adds to the noise level as does dust in the laser
cavity. Because the fluorescence intensity is directly proportional to the source power,

the S/N in the fluorescence measurements follows the S/N in the laser light used.

In practice the S/N of the fluorescence measurements is directly proportional to
the S/N of the laser light used, and adequate S/N laser light is rarely obtained. The
excitation source for the Nd:YAG laser is a pair of xenon arc lamps. The typical
useful life of these is about 400 hours as judged by the needs of other experiments that
are less sensitive to the S/N of the laser light (e. g. SPT). Adequate S/N laser light for
this instrument is obtained only during the first 50 or so hours of lamp life. Also, the
alignment of the Nd:YAG laser and the dye laser must be nearly perfect to minimize
noise. Because of these factors that affect the signal-to-noise ratio, satisfactory

performance has only rarely been obtained with this instrument.

 

 

 

 

 

Th
checked v
earlier. '.

TABLE
DETERh

the liter
from th
uncertajr
Perform:

measure
chemic;

Phenyla

 

112

The accuracy of excited state lifetimes determined with the instrument was
checked with a few compounds that have excited state lifetimes in the range determined

earlier. The results are shown in Table 4—1 with values determined by S171“ and from

TABLE 4-1. EXCITED-STATE LIFETIMES OF STANDARD COMPOUNDS
DETERMINED BY SPT AND FDF.

 

Literature
Compound/Solvent Value SPT FDF
Rose Bengal / methanol 549 ps [48] 549 ps 350 j; 150 ps
Rhodamine B / ethanol 2.85 ns [49] 2.78 ns 4.1 i 2 ns

the literature. The concentration of all solutions was 10 #M- The lifetimes calculated
from the phase shift and the demodulation factor agree to within experimental
uncertainty. The uncertainties in the lifetimes determined by FDF reflect the

performance of the laser system at the time the measurements were made.

APPLICATION TO REAL-TINIE CORRECTION FOR ERRORS DUE TO
QUENCHIN G

The instrument described in this chapter was used to correct fluorescence
measurements for the reduction in intensity caused by collisional quenching. The
chemical systems studied were quinine sulfate quenched by chloride ion and dansyl-L-

phenylalanine quenched by nitromethane.

113

Experimental Details

APPARATUS

The instrument described in this chapter was used for all measurements.
Excitation was from the frequency doubled output of a cavity dumped dye laser. The
laser dye was Rhodamine 6G, and the cavity dumper was set to divide by 10.. The
wavelength was 293 nm and the power was ca. five mW (estimated from the dye laser

power of 110 mW and SHG efficiency of five percent).

REAGENT S

These chemical systems were also studied in Chapter three, and details
concerning reagents and solution preparation can be found there. In addition to those
solutions, a scattering solution was prepared by dissolving ca. one gram of non-dairy

coffee creamer (Quality Dairy, Lansing, MI) in a few milliliters of distilled water.

PROCEDURES

The instrument was calibrated for both phase and modulation measurements
using scattering solutions as described in this chapter. Triplicate measurements were
made on each solution, where each measurement is the mean of 100 analog-to-digital

conversions.

 

114

Results and Discussion

The results for the quinine sulfate - chloride ion system are listed in Table 4-2.

TABLE 4-2. FLUORESCENCE INTENSITIES, LIFETIMES, AND THE RATIO 1/ 1'
FOR QUININE SULFATE, 10 pM IN 0.1 M SULFURIC ACID, QUENCHED BY
VARYING CONCENTRATIONS OF CHLORIDE ION.

 

[Cl'] Intensity Lifetime (ns) Intensity/ Lifetime
0.0 3695 23 160
0.005 1798 8.3 216
0.01 1304 5.4 241
0.05 510 2. l 243

The lifetimes shown are mean values of the lifetime determined from the phase shift
and the demodulation factor. Lifetimes determined from these quantities agreed to
within experimental uncertainty. The steady-state intensity and the intensity corrected
for quenching by ratioing to the excited-state lifetime are compared graphically on page
115 in Figure 4-13. The fluorescence intensities are normalized to the unquenched
value ofI/r. As was the case with the SPT data in Chapter three, both the intensity and
the lifetime decrease as the concentration of chloride ion increases. Note also that
values are not given for solutions where the concentration of quencher is greater than
0.05 M. This is because the intensity from these solutions is below the detection limit
for an excited-state lifetime. That is, the signal level is on the order of the signal
generated from RFI as discussed earlier in this chapter. Unlike the SPT data (Figure 3-
10), the ratio of Ur increases as the concentration of quencher increases. This is

because the noise resulting from RFI which is in phase with the exciting light becomes

115

 

 

3‘

E

3 ‘t

E

a zoo-

C -u-

0

O

m u

e a

O «-

3

E

'0 100‘

0

.5 °

"0 .

E

L.

O

2 o
0 T T r I I l
O 00 0.01 0 02 0.03 O 04 O 05

[OFT

Figure 4-13. Plot of steady-state fluorescence intensity (circles) and fluorescence
intensity corrected for quenching by ratioing to the excited state lifetime (squares) for
quinine sulfate, 10 #M in 0.1 M sulfuric acid, quenched by chloride ion.

dominant as the cross-correlation signal decreases. This causes the determined excited-
state lifetime to be less than the true value. Because the intensity is not affected by the
RFI, the determined ratio of U1 is greater than the true value. This source of error

becomes significant as the fluorescence intensity decreases.

The data for the dansyl-L—phenylalanine - nitromethane system is shown on page
116 in Table 4-3. Note that values are not given for solutions where the concentration
of quencher is greater than 0.10 M. This is because the intensity from these solutions
is below the detection limit for an excited state lifetime. Again, the steady-state
intensity and the excited-state lifetime decrease as the concentration of quencher

increases. The lifetimes listed here are also mean values of the lifetime determined

116

TABLE 4-3. FLUORESCENCE INTENSITIES, LIFETIMES, AND THE RATIO I/r
FOR DANSYL-L—PHENYLALANINE, 10 ”M IN 2-PROPANOL QUENCHED BY
VARIOUS CONCENTRATIONS OF NITROMETHANE.

 

[CH3N02] Intensity Lifetime (ns) Intensity/ Lifetime
0. 0 3508 12 292
0. 005 2816 8 . l 347
0.01 2207 6.9 319
0. 05 1418 4 . 6 308
0. 10 627 1 .9 330

from the phase shift and the demodulation factor, and the lifetimes determined from
these quantities agreed to within experimental uncertainty. The steady-state intensity
and the intensity corrected for quenching by ratioing to the excited-state lifetime are
compared graphically on page 117 in Figure 4-14. The fluorescence intensities are
normalized to the unquenched value of 1/1. As mentioned above, the signal level for
these solutions is on the order of the signal that results from RFI. These data show an
apparent correction for quenching. There is a dominant downward trend in the SPT
data for this system (Figure 3-12) that is not evident here. This is a case of one
systematic error cancelling the effect of another. The excited-state lifetime is skewed
toward lower than true values, and the measured fluorescence intensity appears to be

reduced by processes other than collisional quenching as discussed in Chapter three.

In summary, the precision in excited state lifetimes determined with this
instrument is strongly dependent on the noise in the output of the laser system. The
accuracy of this measurement degrades as the fluorescence intensity is reduced. This is

due to a signal that results from RFI which is in phase with the exciting light and

117
4001

300: I
200—

100-

Normallzed Fluorescence Intensity

 

o Y I Y j— W I Y I
0.00 0.02 0.04 0.06 0.08 0.10

[CHaNOZ]

 

Figure 4-14. Plot of steady-state fluorescence intensity (circles) and fluorescence
intensity corrected for quenching by ratioing to the excited state lifetime (squares) of
dansyl-L-phenylalanine, 10 am in 2-propanol, quenched by nitromethane.

causes the lifetime to be reduced relative to the true value. The degree of this
distortion increases as the analyte signal level decreases to a point where it is
comparable to the RFI-generated signal. These sources of error greatly limit the ability
of this instrument to correct for the effects of quenching using the Hieftje-Haugen
method [2]. Suggestions for reducing these sources of error are discussed in Chapter

six.

10.
11.
12.
l3.
14.
15.
16.
17.

18.

19.
20.

118

CHAPTER 4 REFERENCES

Gaviola, E. Z. Phys. 1927, 42,852.
Hieftje, G.M.; Haugen, G.R. Anal. Chim. Acta.1981, 123, 255.

Lakowicz, J .R Principles of Fluorescence Spectroscopy Plenum Press: New
York, 1983

Ware, W.R. in Creation and Detection of the Excited State,vol 1A A.A.
Lamola, Ed. Marcel Dekker, Inc.:New York 1971.

Demas, J.N. Ercited State Lifetime Measurements Academic Press: New York,
1983.

Hecht, B.; Zajac, A. Optics Addison-Wesley: Reading MA, 1979

Ingle, J.D.; Crouch, S.R. Spectrochemical Analysis Prentice Hall: Englewood
Cliffs, NJ 1988.

Debye, P.; Sears, F.W.; Proc. Nat. Acad. Sci. 1932, 18, 409.
Maydan, D. IEEE J. Quant. Elec. 1970, QE-6, 15.

Lekavich, J. Lasers Appl. 1986, 4, 59.

Dushinsky, F, Z. Phys. 1933, 81, 7.

Szymanowsky, W. Z. Phys. 1936, 95, 440.

Schmillen, A. Z. Phys. 1953, 135, 294.

Birks, J.B.; Little, W.A. Proc. Phys. Soc. A, 1953, 66, 921.
Bonch-Bruevich, A.M. Izvest. Akad. Nauk. S. S.S.R. 1956, 20, 591.
Birks, J.B.; Dyson, DJ. J. Sci. Instr. 1961, 38, 282.

Spencer, R.D.; Weber, G. Ann. N.Y. Acad. Sci. 1969, 158, 361.

Veselova, T.V.; Cherkasov, A.S.; Shirokov, V.I. Opt. Spectrosc. 1970, 29,
617.

McGown, L.B.; Bright, F.V. CRC Crit. Rev. 1987, 18, 245.

Gratton, B.; Limkeman, M. Biophys. J. 1983, 44, 315.

21.

22.

23.

24.

25.
26.
27.
28.
29.
30.
31.
32.

33.

34.

35.

36.

37.
38.

39.

40.

119

Gratton, B.; Jameson, D.M.; Hall, R.D.; Ann. Rev. Biophys. Bioeng. 1984, 13,
105.

Gratton, B.; Jameson, D.M.; Rosato, N. Weber, 6.; Rev. Sci. Instrum. 1984,
55 , 486.

Lakowicz, J.R.; Maliwal, B.P. Biophys. Chem. 1985, 21, 61.

Alcala, J.R.; Gratton, B.; Jameson, D.M. Anal. Instrum. 1985, I4, 225.
Lakowicz, J.R.; Lackzo, G. Gryczynski, 1. Rev. Sci. Instrum. 1986, 5 7, 2499.
Bright, F.V.; Wilson, D.A.; Hieftje, G.M. Appl. Spec. 1986, 40, 683.
Bemdt, K. Opt. Commun. 1985, 56, 30.

Bright, F.V. Appl. Spec. 1988, 42, 1531.

Weber, G. J. Phys. Chem. 1981, 85, 949.

Balter, A. Opt. Commun. 1982, 42, 407.

Lakowicz, J .R.; Cherek, H. Maliwal, B. J. Lumin. 1984, 31, 699.

Lakowicz, J.R.; Lackzo, G. Cherek, H.; Gratton, B.; Limkeman, M. Biophys.
J. 1984, 46, 463.

Lakowicz, J.R.; Maliwal, B.P. Anal. Instrum. 1985, 14, 193.

Gratton, B.; Limkeman, M.; Lakokwicz, J .R.; Maliwal, B.P.; Cherek, H.
Lackzo, G. Biophys. J. 1984, 46, 479.

Lakowicz, J .R.; Gryczynski, 1.; Cherek, H.; Lackzo, G.; Joshi, N. Trend. Anal.
Chem. 1986, 5, 257.

Keating-Nakamoto, S.M.;Cherek, H. Lakowicz, J.R. Anal. Chem. 1987, 59,
271.

Nithipatikon, K; McGown, L.B. Appl. Spec. 1987, 41, 395.

Lackzo, G.; Gryczynski, 1.; Gryczynski, Z.; Wiczk, W.; Malak, H. Lakowicz,
J.R. Rev. Sci. Instrum. 1990, 61, 2331.

Malmstadt, H.V.; Enke, C.G.; Crouch, S.R. Electronics and Instumentation for
Scientists, Benjamin/Cummings: Menlo Park, CA. 1981.

Analog Integrated Circuit Data Book, Vol 10. 1990, Precision Monolithics Inc.,
Santa Clara, CA.

41.

42.

43.

44.

45.

45.

46.

47.

48.

49.

120
Linear Products Databook, 1990-91, Analog Devices, Norwood, MA.

Daryanani, G. Principles of Active Network Synthesis and Design Wiley: New
York. 1976.

Micro-CAP 11, student edition, release 5. Addison-Wesley and Benjamin
CummingszNew York, NY, 1988.

Hufault, J.R. 0p Amp Network Design, Wiley: New York, NY, 1986.

Data Acquisition Linear Devices Databook, 1989, National Semiconductor
Corporation, Santa Clara, CA.

Bowman, L.E.; Victor, M.A.; Crouch, S.R. Trend. Anal. Chem. 1990, 9, 111.

Ott, H.W. Noise Reduction Techniques in Electronic Systems, 2nd ed. Wiley:
New York, NY, 1988.

Fleming, G.R. Chemical Applications of Ultrafast Spectroscopy, Oxford
University Press: New York, NY, 1986.

Ware, W.R.; Pratinidhi, M.; Bauer, R.K. Rev. Sci. Instrum. 1983, 54, 1148.

Comission on Photochemistry, Pure Appl. Chem. 1988, 60, 1108.

 

 

CHAPTER 5: A HIGHLY CORRECTED MOLECULAR FLUORESCENCE
SPECTROIVIETER

When they occur during a fluorescence assay, collisional quenching and sample
absorption will cause the measured fluorescence intensity to be less than the value in
the absence of these processes. By combining methods that correct fluorescence
measurements for the reduction in intensity caused by these processes, one can improve
the accuracy of fluorescence measurements. An instrumentation-based approach to
realize a fluorescence measurement that is corrected for both collisional quenching and
sample absorption is described in this chapter. Because the constructed instrument
corrects fluorescence measurements for both of these sources of error, it has been

termed a "highly corrected" molecular fluorescence spectrometer.

The method used to correct for collisional quenching is that reported by Hieftje
and Haugen [1]. One measures both the steady-state fluorescence intensity, I, and the
excited-state lifetime, 1, and determines the ratio I/r. Because both of these quantities
decrease as collisional quenching increases, this ratio is in principle insensitive to
collisional quenching and dependent only on the concentration of the fluorophore. The
instrument constructed to directly determine the ratio 1/7 is described in Chapter four.
Correction for sample absorption is accomplished with the cell-shift method as
described by Lutz and Luisi [2]. The cell-shift method was chosen because all of the

measurements needed for it are fluorescence measurements.

In this chapter, the implementation of the cell shift method and the instrument
constructed to accomplish it are described. This is followed by a description of the
highly corrected measurement and its application to analytical situations where

uncorrected molecular fluorescence would not normally be used.

121

 

122

REVIEW

Although several instruments and methods have been developed at Michigan

State University (see Chapter one and references cited therein), the implementation of
the cell~shift method reported by Lutz and Luisi [2] was chosen. While intuitive
approaches that correct for sample absorption by directly measuring the absorbance of
the solution have been developed [3-5], direct knowledge of the solution absorbance is
not needed in the method reported by Lutz and Luisi [2]. Their method is one in which
the cell is moved [6-8], and measurements of the fluorescence intensity at different

locations are recorded. In order to correct for absorption of the emitted fluorescence in

addition to absorption of the exciting light, Lutz and Luisi [2] considered the case of a

cell shifted along its diagonal as shown in figure 5-1. They derived the following

Io

 

Figure 5-1. Methodology of the cell-shift technique reported by Lutz and Luisi [2]. F1
and F2 are fluorescence intensities at cell positions 1 and 2, respectively, where the
optical pathlengths are l, and £2.

 

 

 

 

 

 

123
relation for the corrected fluorescence intensity, F0, in terms of F1 and F2 , the

fluorescence intensities at positions 1 and 2, and a geometric term, a:

F1 8 (1
Fa = F1 — (5-1) a = —-, a (5-2)
2 ' 1

Because the excitation and emission paths change by the same amount when the cell is
shifted along its diagonal, the difference in the fluorescence intensity at the two
locations reflects the solution absorbance at the excitation and emission wavelengths.
Hence, one can correct for both primary and secondary absorption with only two
fluorescence measurements. The only assumption made in deriving the above
expression that affects the measurement is that the detected fluorescence is collimated.
If it were not, the value of the pathlength would not be equal to l, (or L2). Collimation
is readily accomplished with appropriate masks in the collection optics of the

fluorometer.

A FLUORESCENCE SPECTROMETER THAT CORRECTS FOR QUENCHIN G
AND ABSORPTION

A molecular fluorescence spectrometer was developed to correct for the
reduction in fluorescence intensity caused by collisional quenching and sample
absorption. A cell holder was designed and fabricated for the cell-shift method and
used with the single-photon timing apparatus described in Chapter two and the
frequency domain fluorometer described in Chapter four. The cell—shift is
accomplished by mounting the cell on two micrometer-driven stages moved by stepper
motors. This instrument was used to map the fluorescence response surface of the cell

and to make highly corrected fluorescence measurements.

124

Fluorescence Response Surfaces

EXPERIMENTAL DETAILS

Apparatus

The frequency—doubled output of the Nd:YAG laser used in the SPT system
described in Chapter three was used to excite the sample. The laser power was reduced
by splitting the beam with a quartz plate and attenuating the reflected beam with neutral

density filters such that the power at the sample was about 15 mW.

A holder for positioning the cell was fabricated from two micrometer driven
translation stages (Newport, Fountain Valley, CA). The stages were driven by stepper
motors connected to the adjustment handles of the micrometer drives with shaft drive
assemblies constructed in-house. The stages were mounted orthogonally so that the cell
position with respect to the excitation beam and the collection optics could be varied in
two directions. The stepper motors were controlled with a stepper motor controller
designed and constructed in-house and an IBM-compatible personal computer (Zenith
Data Systems) equipped with an IBM Data Acquisition and Control Adapter.
Computer control of the stepper motors was accomplished by adding subroutines to the
software used in the frequency domain fluorometer described in Chapter four. The
combined resolution of the stepper motors and translation stages was 0.0014 mm per

step.

The fluorescence measurements were made with the frequency domain

fluorometer described in Chapter four. A variable iris was adjusted to an aperture of

 

 

125

approximately one millimeter and placed between the lenses of the collection optics to
insure that only collimated fluorescence reached the detector. The imaging resolution
of the optical system was about 0.3 mm2 when the monochromator slits were set to
1000 um. The waist of the exciting laser beam was reduced with plano-convex lenses

to about this value to optimize the fluorescence signal.

A standard 1 cm2 x 4 cm quartz fluorescence cuvette was used for all

measurements .

Reagents

A 0.001 M stock solution of rhodamine B (Aldrich) was prepared in 95%
ethanol (Department of Chemistry, Michigan State University). Aliquots of the stock

solution were diluted with ethanol to prepare 1.0 I‘M and 100 #M working solutions.

Procedures

Both solutions were excited at 532 nm and fluorescence was detected at 600 nm.
The monochromator slit width was 1000 um. The voltage applied to the
photomultiplier tube was adjusted to yield the maximum signal at the output of the
current-to-voltage converter without exceeding the input voltage range of the analog-to—

digital converter.

The fluorescence response surface of the cell was mapped using the computer
controlled stepper motor driven translation stages and the frequency domain

fluorometer described in chapter four. A response surface of 144 points was recorded

126

for both of the rhodamine B working solutions. The interval between points was one

millimeter. Each point was the mean of 100 analog-to-digital conversions.

RESULTS AND DISCUSSION

The concentrations of the rhodamine B solutions were chosen to illustrate
minimal and extreme sample absorption. The absorbance of the 1.0 pM solution was
0.007 at the excitation wavelength and not measurable at the emission wavelength. The
absorbance of the 100 #181 solution was 1.07 at the excitation wavelength and 0.016 at
the emission wavelength. Each response surface is illustrated from two different
viewpoints so that no part of the surface is obstructed. In these illustrations, the X and
Y axes are respectively orthogonal and parallel to the monochromator entrance slit with
the distance in the solution increasing with X and Y. Hence, the X axis indicates
emission distance, and the Y axis indicates excitation distance. The axes are in units of
tenths of millimeters. Position 0,0 was set at approximately the center of the

intersection of the cell walls.

The response surface of the 1.0 I‘M solution is shown on page 127 in Figure 5-
2. Because the absorbance of the solution at the excitation and emission wavelengths is
small, the response is essentially flat. The slightly higher intensity near the cell walls
that are orthogonal to the plane of the entrance slit is due to a slight blurring of the
image as the distance in the solution increases. The effective focal length of the
collecting lens changes as the cuvette is moved toward the lens due to the difference
between the refractive index of the solution and the refractive index of air. The slight

texture in the surface reflects noise in the laser.

 

 

 

     
    

127

       
 
 

  

  
        
 

3::
:0"
:..

’O:.;wo \

-
9:0::0.0" I \\‘

' "ti

III!!! I“;

”I”, “\\\

    
 

)«\\\

  

   

Figure 5 2 Response surface of fluorescence intensrty as a function of optical path
length in solution for a 01 0 pM solution of rhodamine B in ethanol The axes umts

are tenths of millimeters
The response surface of the 100 I‘M solutlon 18 shown in Figure 5 3 Note that

   

  

\l“
\“l‘

LL“! 6‘“

      
  
   

          

"slit/II

Figure 5 3 Response surface of fluorescence intensrty as a function of optical path
length 1n solution for a 100 pM solution of rhodamine B in ethanol The axes umts are

    
 

tenths of millimeters

 

128

the fluorescence intensity decreases dramatically with increasing distance in the
direction of excitation (Y). This is due to the high absorbance of the solution at the
excitation wavelength. The absorbance at the emission wavelength causes a slight
decrease in the fluorescence intensity as the depth of solution through which the

fluorescence is viewed increases (X).
Highly Corrected Fluorescence

Fluorescence measurements corrected for the reduction in intensity caused by
sample absorption and collisional quenching were made. The fluorophore chosen for
study was quinine sulfate in 0.1 M sulfuric acid. Chloride ion was used to quench
fluorescence, and the acid-base indicator bromocresol green was used to increase the

absorbance of the solution at both the excitation and emission wavelength.
EXPERIMENTAL DETAILS
Apparatus

The frequency domain fluorometer described in Chapter four, the cell
positioning apparatus described earlier in this chapter, and the single-photon timing
apparatus described in Chapter three were used to correct for the reduction in
fluorescence caused by collisional quenching and sample absorption. For comparison,
steady-state fluorescence measurements were made with a Perkin-Elmer model LS-S
fluorometer. A Perkin-Elmer ”Lambda3" UV/VIS absorption spectrometer was used

for absorbance measurements.

129

Reagents

A 1.0 mM stock solution of quinine sulfate monohydrate (Aldrich) was prepared
in 0.1 M sulfuric acid (prepared by diluting 5 .5 ml of concentrated sulfuric acid
(Malinckrodt) in one liter of distilled and deionized water (Corning millipure system)).
A 1.0 M stock solution of sodium chloride (J .T. Baker) was prepared in 0.1 M sulfuric
acid. A 1.0 mM stock solution of bromocresol green (Aldrich) was prepared by
dissolving the solid in a minimum amount of 95% ethanol (Department of Chemistry,

Michigan State University) and diluting with 0.1 M sulfuric acid. A 100 uM stock

solution of bromocresol green was prepared in 0.1 M

sulfuric acid by diluting an
aliquot of the previous solution. A total of twelve working solutions were prepared.
Each working solution was prepared by diluting to 100 ml a mixture composed of 1.0
m1 of the quinine sulfate stock solution, 0.0, 1.0, or 10.0 ml of the sodium chloride
stock solution, and 0.0, 10.0, or 50.0 ml of the 1.0 mM bromocresol green stock

solution or 0.0, 1.0, or 10.0 ml of the 100 pM bromocresol green stock solution.
Procedures

The steady-state fluorescence intensity of the working solutions was measured
on the Perkin—Elmer fluorometer. The solutions were excited at 350 nm and
fluorescence was detected at 450 nm. The bandpass for both excitation and emission

was 3 nm.

To obtain the highest accuracy, the excited-state lifetimes were determined with
the single-photon timing apparatus. The solutions were excited at 290 nm, and
fluorescence was detected at 450 nm with a photomultiplier tube. The fluorescence

decay of each sample was acquired over 1024 channels and fit to a single exponential

 

 

130

model. The value of reduced X2 was less than 1.2 for all of the curve fits except those
for the solutions that were 0.1 M in sodium chloride. A X2 of 3 was typical for the

curve fits for these solutions.

Fluorescence intensities corrected for sample absorption were measured with the

cell-shift method and the frequency domain fluorometer.
RESULTS AND DISCUSSION

Four tables of fluorescence intensity are presented in this section. A table is
given for steady-state measurements of fluorescence intensity and for each of the three
correction schemes. The data in each table reflects the ability of each measurement
strategy to correct for the reduction in intensity caused by collisional quenching and

sample absorption.

The steady-state fluorescence intensity measurements are shown in Table 5-1.

TABLE 5-1. STEADY-STATE FLUORESCENCE INTENSITY (ARBITRARY
UNITS) FOR QUININE SULFATE SOLUTIONS THAT ARE AFFECTED BY
COLLISIONAL QUENCHIN G AND SAMPLE ABSORPTION.

 

ABS @ 450 nm
0.0 0.03 0.3 1.0 2.3
[NaCl]
0.0 9000 7920 3510 900 85
0.01 3835 3510 1600 389 30
0.1 576 522 240 59 6

 

 

 

 

 

 

131

The columns are labeled with the absorbance of the solution at the emission
wavelength. Sample absorption increases across a row and collisional quenching
increases down a column. Because the fluorescence intensities shown in this table were
obtained with steady-state measurements, the intensity decreases both across a row and
down a column. Note that the fluorescence intensity varies over three orders of

magnitude even though the concentration of the fluorophore is constant.

The fluorescence intensity corrected for quenching by the Hieftje-Haugen

method [1] is shown for these solutions in Table 5-2. Note that the ratios listed in

TABLE 5-2. FLUORESCENCE INTENSITY RATIOED TO EXCITED-STATE
LIFETIME FOR QUININE SULFATE SOLUTIONS THAT ARE AFFECTED BY
COLLISIONAL QUENCHING AND SAMPLE ABSORPTION.

 

ABS @ 450 nm
0.0 0.03 0.3 1.0 2.3
[NaCl]
0.0 523 460 204 52 5
0.01 498 456 207 51 4
0.1 427 387 178 44 5

 

this table are essentially constant in each column but decrease across each row. The
ratio of fluorescence intensity to excited-state lifetime corrects for collisional quenching

but not sample absorption.

The results of the cell-shift measurements are shown on page 132 in Table 5-3.

Here, the reported quantity is almost constant across a row but decreases down a

132

TABLE 5-3 FLUORESCENCE INTENSITY (ARBITRARY UNITS) AS
MEASURED BY THE CELL—SHIFT METHOD FOR QUININE SULFATE
SOLUTIONS THAT ARE AFFECTED BY COLLISIONAL QUENCHING AND
SAMPLE ABSORPTION.

 

ABS @ 450 nm
0.0 0.03 0.3 1.0 2.3
[NaCl]
0.0 196 227 203 189 105
0.01 91 113 98 123 448
0.1 14 18 15 14 5

column. The cell-shift method corrects for sample absorption but not for collisional

quenching.

The fluorescence intensities shown in Table 5-4 are the cell-shift measurements

TABLE 5-4. ABSORPTION CORRECTED FLUORESCENCE INTENSITY
RATIOED TO EXCITED—STATE LIFETIME FOR QUININE SULFATE
SOLUTIONS THAT ARE AFFECTED BY COLLISIONAL QUENCHING AND
SAMPLE ABSORPTION.

 

ABS @ 450 nm
0.0 0.03 0.3 1.0 2.3
[NaCl]
0.0 11.4 13.2 11.8 10.0 6.1
0.01 11.8 14.7 11.8 15.9 6.2
0.1 10.0 13.5 10.6 9.9 3.9

from Table 5-3 ratioed to the excited-state lifetime. Here, the reported quantity is
almost independent of both collisional quenching and sample absorption. This "highly

' ' ’ wn a column.
corrected" fluorescence rntensrty rs nearly constant across a row and do

 

133

Consider that the extreme values of this data set vary by a factor of four while the
range of observed steady-state fluorescence intensity exceeds three orders of magnitude

for the same solutions.
The relative standard deviation of the data obtained using each correction
scheme is shown in Table 5-5. The highly corrected measurements are not nearly as

TABLE 5-5. RELATIVE STANDARD DEVIATION OF THE DATA OBTAINED
BY EACH MEASUREMENT SCHEME.

 

. Highly
Steady-State I/ 1' Cell-Shift Corrected
ABS Oto 1.0 114% 64% 75% 16%
All solutions 135 % 87 % 95 % 31 %

widely scattered as the results of the other methods. The correction provided is
especially satisfactory if one does not include the solutions that have absorbances
greater than 1.0. Although both the ratio of UT and the cell-shift schemes are better
that the steady-state values, neither comes close to the correction provided by the

highly corrected measurement.

Each of the correction schemes provides the best compensation for those
solutions that are neither highly quenched nor strongly absorbed. This is primarily due
to the very low fluorescence intensity of these solutions. In addition to absorbing the
quinine sulfate fluorescence, bromocresol green also absorbs the exciting light. This
greatly reduces the fluorescence intensity relative to an absorbing species that would

only absorb the emitted fluorescence.

134

In addition to providing a fluorescence-based measurement that is corrected for
both of these sources of uncertainty, the highly corrected measurement concept could
allow fluorescence to be more useful in diagnostic work. If changes in the highly
corrected fluorescence intensity occur while the concentration of the fluorophore is

constant, it could be an indication that other processes such as static quenching or

energy transfer are occurring.

 

 

 

 

 

135

CHAPTER 5 REFERENCES

Hieftje, G.M.; Haugen, G.R. Anal. Chim. Acta. 1981, 123, 255.
Lutz, H.P.; Luisi, P.L. Helv. Chim.Acta. 1983, 66, 1929.

Parker, C.A.; Barnes, W.J. Analyst 1957, 82, 606.

Holland, J.F.; Teets, R.E.; Timnick, A. Anal. Chem. 1973, 45, 145.

Holland, J.F.; Teets, R.E.; Kelley, P.M.; Timnick, A. Anal. Chem. 1977, 49,
706.

Britten, A.; Archer-Hall, J.; Lockwood, G. Analyst 1978, 103, 928.

Christmann, D.R.; Crouch, S.R.; Holland, J.F.; Timnick, A. Anal. Chem.
1980, 52, 291.

Christmann, D.R.; Crouch, S.R.; Timnick, A. Anal. Chem. 1981, 53, 276.

 

CHAPTER 6: SUGGESTIONS FOR INIPROVEIWENTS AND FUTURE WORK

Most of what has been written in this dissertation concerns instrumentation for
determining excited state lifetimes. Because the capabilities of modern technology are
constantly improving, no developments described in this document can remain state-of-
the-art for long. As the Single-Photon Timing (SPT) apparatus described in Chapter
three will probably be used by other students, some comments concerning
improvements to the constructed instrument are discussed in this chapter. Suggestions
for improving the frequency domain fluorometer are also made. If the means could be
found to improve the frequency domain fluorometer along the lines described here, it
could find more realistic analytical applications. Some of these are also described in

this chapter.
SINGLE-PHOTON TIMING

Of the SPT instruments that have been described to date, the apparatus
constructed in this work and discussed in Chapter three is probably one of the most
versatile. While other instruments have slightly shorter instrument response functions,
many of these are not capable of acquiring fluorescence decay data over a continuous
temporal window from about two nanoseconds to a microsecond. This capability was
made possible with the switched fiber optic delay described in Appendix two. If it is
needed, the instrument response can be reduced from its present value of about 45
picoseconds to slightly less than 20 picoseconds. To realize a narrower instrument

response, a few of the present components must be replaced.

The first items that should be replaced are obviously those that are presently

limiting the instrument. However replacing one component will not make much of a

136

M‘—

 

 

 

137

difference. Four principal items would be needed. These are a subtractive dispersion
double monochromator, a six micron MCP-PMT, a 6 GHz amplifier, and a biased
TAC. The transit time spread of the present monochromator has been reduced by
placing a mask at the diffraction grating. While this reduced the monochromator's
contribution to the instrument response to an acceptable level for a 12 micron MCP-
PMT, it also reduced the instrument's sensitivity. This is not a problem with efficient
fluorophores, but it requires long acquisition times for weakly emitting species. A
subractive dispersion double monochromator has no temporal dispersion and would not
contribute to the instrument response. If the present monochromator is replaced with a
subtractive dispersion double monochromator, the instrument response may be reduced
by a few picoseconds. This will be most noticeable in measurements made toward the
red end of the visible spectrum. The extent of any improvement will reflect the affect
of the masked, Czemy-Turner monochromator that it presently used. Even if there is
no discernable reduction in the instrument response, the instrument's sensitivity will
certainly be improved. At present, only one subtractive dispersion double
monochromator suitable for SPT exists, the model DB—lOS, manufactured by American
Holographic. It costs slightly under four thousand dollars when ordered with ion

milled gratings.

A faster detector or one with a smaller transit time spread would also be needed
to reduce the instrument response function. At present the detector that has the
smallest transit time spread that is suitable for SPT is the Hamamatsu R2809. This is
an MCP-PMT that contains two microchannel plates composed of six micron channels.
It has been used in other SPT instruments to yield an instrument response below 20

picoseconds.

 

138

If a faster detector such as the one mentioned above is obtained, a higher
bandwidth amplifier and attenuator would also be needed. For this six micron MCP-
PMT, a amplifier having a bandwidth of at least 6 GHz is recommended. A 10 GHz
bandwidth amplifier would be preferred. The semi-rigid coax that is already in place is
sufficient for this detector, but the present attenuator would need to be replaced with a

higher bandwidth device.

With these faster components, a faster time-to—amplitude converter would be
needed to accurately record the instrument response and very fast fluorescence decays.
The best device presently available is the Tennelec model 864 TAC, which has a built-
in bias amplifier. This allows one to amplify a narrow region of the time base. In this
way the normal output voltage range can represent a time range of only a few
nanoseconds. While this can improve the temporal resolution of the TAC-MCA
measurement system to about 0.2 picoseconds per channel, the bias amplifier can
degrade the linearity of the TAC response. This distortion is minor and can be
minimized by carefully adjusting the delay so that only the most linear region of the

TAC is used.

With a higher bandwidth detector and amplifier and an improved temporal
resolution of the measurement system, one small, passive component may be limiting
the instrument response. This is the switched coaxial cable delay. The coaxial cable
used in it is RG—174 which is not really adequate for the bandwidths mentioned above.
An improvement would be a delay constructed with semi-rigid coax and high
bandwidth switches, such as those used for microwave signals. Unfortunately such a
device is not commercially available, but it could be constructed in-house. The major
difficulty would be fitting the semi-rigid coax in a convenient housing such as a NIM.

Because semi-rigid coax is manufactured in five foot lengths, it would also be a bit

 

Ii
1
I
l
l

139

tedious to splice together pieces of semi-rigid cable to obtain the longer delays needed.
Of course one could special-order a given length of semi-rigid coax coiled so as to fit

inaNIM.

With these improvements, an instrument response of less than 20 picoseconds
should be obtainable. Of course one would have to carefully adjust the discriminator
thresholds and change the shaping delay in the CFD to optimize its performance with

faster risetime signals.

One of the items that presently limits the ability to resolve fast and complex
decays is the sample cell. Reflections from the cell walls are evident in the instrument
response. In addition, a distortion caused by them can sometimes be observed in the
early part of a fluorescence decay. The best means of removing this source of
interference is to use the special cuvettes mentioned in Chapter four. Although they are
expensive and must be specially made at present, they are simple to use and are proven

to reduce the interference caused by wall reflections to a tolerable level.

Beyond the instrument response considerations, it would be worthwhile to
develop the hardware and software needed to acquire time—resolved emission spectra.
This is most easily accomplished by operating the MCA as a single channel analyzer
(SCA). One sets upper and lower level discriminators and scans the analyzer in time.
By simultaneously scanning the monochromator, one collects in each channel the
number of photons emitted in the set time window over the wavelength interval
corresponding to the width of a channel. One can signal-average spectra by collecting
several SCA and averaging over the same temporal window. A complete time resolved
emission profile can be obtained by recording SCA scans for several temporal

windows.

 

140

If there is serious interest in having this capability as a routine research tool,
one would need to control the monochromator with the computer. At present this is
not possible with the manually controlled SPEX 1681. It would be trivial to drive the
American Holographic DB-lOS with a stepper motor as was done for the cell-shift
measurements described in Chapter five. Also, an MCA that is capable of software
control of the discriminator levels would greatly reduce tedium in this experiment.
Such an MCA has recently become available. It is the model 5000 MCA manufactured
by Viking (Madison, WI).

FREQUENCY-DOMAIN FLUORESCENCE

Changes in instrumentation are certainly needed if the accuracy and precision of
the frequency domain fluorometer are to be improved. Two major changes are needed
to realize these improvements. These are a more stable light source and a commercial
frequency synthesizer. Aside from the RFI problems associated with the laser system
used, it is not sufficiently stable for frequency domain fluorescence. This is one of the
major advantages of SPT; it is most insensitive to fluctuations in the intensity of the
light source. SPT is also more sensitive - one of the reasons that it is more widely used
than the various frequency domain techniques. The RFI could be reduced to a more
reasonable level by replacing the cables between the mode locker driver and the mode
locker crystal with semi-rigid or twinaxial coax. As mentioned in Chapter four, the
RFI was reduced slightly by adding shielding to the length of coax between the mode
locker driver and the laser. However there is another cable between the end bezel of
the laser and the mode—locker crystal. This cable is only partially shielded and is
probably the primary source of radiation. Not only would this greatly reduce the RFI

in the frequency domain measurements, it would probably remove the 38 MHz noise

 

141

that is observed in plots of the autocorrelation of residuals in SPT experiments on long-

lived species.

To obtain higher accuracy and precision, it would probably be better to use a cw
laser and an external modulator than to continue to use the mode-locked Nd:YAG laser
system. By using a He-Cd, Ar+, or Kr+ laser, one could obtain much higher power

at analytically useful wavelengths, in addition to more stable modulation.

High resolution commercial frequency synthesizers would allow the cross—
correlation frequency to be reduced to a more reasonable value. The availability of
these and the low noise amplifiers needed to drive a modulator and to modulate the
gain of a PMT, would enable not only more effective utilization the capabilities of
FDF, but also greatly improve the accuracy of the method. For some time, this author
has been interested in studying the signal-to-noise ratio of excited state lifetimes
determined by frequency domain techniques as a function of the cross-correlation
frequency. It would be straightforward to make such a study in the frequency range of
a few Hz to 100 kHz. Rather than designing and constructing active filters for specific
frequencies as was required in this work, one could use complete integrated circuit
filters such as the National Semiconductor MFS universal monolithic switched capacitor
filter. One only needs to provide an oscillator at the cutoff or bandpass frequency and
analog and digital voltages to set the filter's configuration and quality factor. With
computer control of the frequency synthesizers and the filter, acquiring the data for this
study could be completed in less than one day. The information derived from it could
greatly benefit the state-of-the-art of frequency domain fluorescence. The cross-
correlation frequency of 30 Hz used in other instruments is almost blindly followed by
workers in the field. The attitude of "if it isn't broken, don't fix it" has limited

progress in the area. While the time required for an SPT experiment can be as little as

 

 

142

one minute, the best frequency domain fluorometers require about one—half hour to
acquire data to determine the same information. The phase shift and demodulation are
acquired at each of several modulation frequencies. The time required for this could be
greatly reduced by going to a higher cross—correlation frequency, which would
minimize the time required for signal averaging. It may also improve the accuracy of

the acquired data.

A suitably improved frequency domain fluorometer could be applied to
determinations based on flowing streams such as flow injection analysis (FIA) and
liquid chromatography (LC). The present apparatus does not possess sufficient

precision over the time window needed to profile these systems.

HIGHLY CORRECTED MOLECULAR FLUORESCENCE

The work discussed in Chapter 5 demonstrated the potential of this combined
correction strategy. Since the highly corrected measurement scheme requires an
accurate determination of the excited-state lifetime, the improvements suggested earlier
in this chapter for the frequency domain fluorometer would need to be implemented
before much meaningful work in highly corrected fluorescence could be attempted.
Moreover, an intense light source is crucial for absorption correction. One must have
an intense source so that there is sufficient fluorescence intensity to measure in the
presence of strongly absorbing solutions and high concentrations of quenchers. A laser
capable of generating one Watt at the excitation wavelength could be very useful in an

apparatus similar to the one described in chapter five.

Regardless of whether an intense laser is obtained, the cell-shift scheme could

be automated more intelligently than in this work. If the solution absorbance is great,

 

143

one could program the controlling computer to use optimum cell positions for the
measurements. That is one would use less of a path length difference when the solution
is great. One could also make measurements at other locations in the cell to better

correct for extreme primary or secondary sample absorption.

The cell-shift method was chosen to provide absorption correction because all of
the measurements needed for it are fluorescence measurements. This and the ease with
which it could be automated and incorporated with the frequency domain fluorometer
and the SPT apparatus made it a logical choice for this work. However, there is no
reason that absorption correction could not be accomplished using fiber optics or
another approach that would make highly corrected measurements possible in flowing

streams.

 

 

 

APPENDIX 1: COMPONENTS OF THE SINGLE PHOTON TIIVIING

APPARATUS

This appendix contains a listing of the components of the SPT instrument, their
part numbers, and names and addresses of vendors. Common items such as flexible
coaxial cable, mirrors, and ordinary optical mounts are not included.

LIGHT SOURCE AND MONITORING EQUIPMENT

Component

Mode Locked
Nd:YAG Laser
w/ High Power
Second Harmonic
Generator

Third Harmonic
Generator and
Wavelength
Separator

Dye Laser

w/ Cavity Dumper
Rhodamine 6G dye
(570-630 nm)

Dye Laser

w/ Cavity Dumper
Stilbene 3 dye
(420-470 nm)

Cavity Dumper Drivers
(TWO)

Ultra-fast PIN
Photodiode

Mgfiel Number/Dgsriptign

768

7950

702

702

7220

 

Vendor/ Address

Coherent, Inc.
3210 Porter Dr.
Palo Alto, CA 94303

Antel Optronics, Inc.
3225B Mainway
Burlington, ON
L7M 1A6

Canada

145

 

LIGHT SOURCE AND MONITORING EQUIPMENT, CONTINUED

Component

Sampling Oscilloscope

Conventional
Oscilloscopes
(TWO)

Autocorrelators

(TWO)

Commnent

Pellicle Beamsplitter

Plano-convex lenses
(25.4 mm dia. and
fused silica unless
indicated otherwise)

Connectorized
Fiber Optic

Model Number/Dgsription
7904A Mainframe

71‘11A Sampling time base
7S11 Sampling Unit

S-4 Sampling Head

2225

514-A

w/KDP crystal #530-080
for Rhodamine 6G dye
514-B

w/BBO crystal #530-040
for Stilbene 3 dye

OPTICS

Model Number/Dfisription
0501000

01LQF028, 50 mm f.1.

01LQF136 150 mm f.1.

SPX016, 50.2 mm f.1.
SPX022, 100.0 mm f.1.
2m, 100/140 G.I. fiber

w/SMA on one end and an
ST connector on the other

Vendor/ Address

Tektronix, Inc.
P.O. Box 500
Beaverton, OR 97077

INRAD, Inc.
161 LeGrand Ave.
Northvale, NJ 07647

Vendor/ Address

ESCO Products, Inc.
171 Oak Ridge Rd.
Oak Ridge, NJ 07438

Melles Griot
1770 Kettering St.
Irvine, CA 92714

Oriel Corporation
250 Longbeach Blvd.
Stratford, CT 06497

Boston Electronics
72 Kent St.
Brookline, MA 02146

Commnent

Fiber optic positioner ‘

with chuck

Comp_onent

NIM Bin and
Power Supply

Quad Constant
Fraction Discriminator

Switched Delay

Time to Amplitude
Converter

Multichannel
Analyzer

High Voltage
Power Supply
(0 to SKV)

2 GHz Amplifier

1 dB Attenuator

Microwave cable
assembly
(mounted inside
chilled housing
for MCP—PMT)

146
OPTICS, CONTINUED

Model Number/Decsription

FPR—Cl
FPH-CA3

ELECTRONICS
Model Number/Decsription

TB3/TC911—6

TC 455

TC 412A

TC 862

PCA/AT (8K)

305

ZFL 2000

SAT- 1

GZZEKR52007.0

Vendor/ Address

Newport Corporation
18235 Mt Baldy Circle
Fountain Valley, CA
92728

Vendor/ Address

Tennelec, Inc.
601 Oak Ridge Tpke.
Oak Ridge, TN 37831

The Nucleus Inc.
P.O. Box 2561
Oak Ridge, TN 37830

Bertan Associates, Inc.
121 New South Rd.
Hicksville, NY 11801

Mini-Circuits
PO. Box 350166
Brooklyn, NY 11235

W.L. Gore and
Associates, Inc.
1901 Barksdale Rd.
Newark, DE 19711

mnn

0.085” semi-rigid
coaxial cable

SMA connectors
for semi-rigid coax

SMA Connectors
for R6 174 coax

LEMO Conectors

Emmet

Second Harmonic
Generator
in Gimbal Mount

w/ Plano-Convex
Lenses, (25 .4 mm dia)

Czemy-Turner
Monochromator

147

ELECTRONICS, CONTINUED

ngiel Number/Dmsriptign

2001-7885-00

554-90/010

FFA.00.250.CTAC22

(for 0.085" semi-rigid coax)

FFA.00.250.CTAC27
(for R6 174 coax)

OTHER COMPONENTS

Model Nummr/Dmsriptign

KDP-R6G

40390, 250 mm f.1.
fused silica

41360, 100 mm f.1.
borosilicate glass

1681

Vendor/ Address

Omni-Spectra
21 Continental Blvd.
Merrimack, NH 03054

Amphenol Corp.
RF/ Microwave Div.
3237 Rt. 112 Suite 2
Medford, NY 11763

LEMO USA, Inc.
335 Tesconi Circle
Santa Rosa, CA 95406

Writes;

INRAD, Inc.
161 LeGrand Ave.
Northvale, NJ 07647

Oriel

Spex Industries
3880 Park Ave.
Edison, NJ 08820

McPherson
530 Main St.
Acton, MA 01720

 

mm
Photomultiplier Tube

Microchannel Plate
Photomultiplier Tube

Cooled Housing
for MCP-PMT

Switched Fiber
Optic Delay
w/ Detector

148

OTHER CONIPONENT S, CONTINUED

Mgflel Number/Dgsriptign
R928

R1564U/11
serial # PD553

TE-321/RF

Vendor/Address
Hamamatsu Corp.
350 Foothill Rd.
Bridgewater, NJ
08807

Products for Research
88 Holten St.
Danvers, MA 01923

See Appendix 2

APPENDIX 2: AN ELECTRONICALLY SWITCHED FIBER OPTIC DELAY

The use of a high repetition rate light source in a Single Photon Timing (SPT)
instrument imposes a restriction on the time over which data can be acquired. This
restriction is due to the requirement that the data channel that contains the temporal
reference signal be connected to the STOP input of the time-to-amplitude converter
(TAC). This, coupled with the fact that signals in the two data channels must be
separated by at least 40 ns when they pass through the constant fraction discriminator
(CFD), necessitates the insertion of delays in each of the data channels. Because the
only reasonable means to delay the fluorescence channel is to insert an electronic delay
(i.e., a coaxial cable) between the CFD and the TAC, one must insert a delay in the
temporal reference channel before the CFD. Placement of a coaxial cable delay
between the detector and the CFD is not desirable as the risetime of the waveform
would not be preserved after passing through the needed length of coaxial cable. The
best solution is to optically delay the temporal reference pulse between the point at
which the pulse train is sampled and the detector. This is most easily accomplished
with an optical fiber. However, a fixed length of optical fiber imposes a restriction on
the time over which data can be acquired. Consequently most TCSPC instruments that
use a high repetition rate laser source are limited to recording very fast fluorescence
decays. This appendix discusses in detail the design and fabrication of a switched fiber
optic delay that allows the user of a TCSPC instrument to easily change the time over

which data is acquired. The finished device is housed in a triple-width NIM module.

As the name implies, the switched fiber optic delay consists of different lengths
of fiber optics connected with fiber optic switches. The delay depends on the total
length of fiber optic between the input and the detector. Different delay times are

obtained by directing the pulses of light via fiber optic switches through different
149

 

 

150

lengths of fiber optic. A schematic drawing of the device is shown below in Figure

A2-1.

    
 

r.o. new is

”a.” _ DETECTOR
PULSE IN

D INUUTB A FER OPTIC GONNETOR

FIGURE A2-l. Schematic diagram of a switched fiber optic delay.

The length of the shortest delay to be inserted in the temporal reference channel
must be chosen such that the instrument's ability to record very fast decays is not
limited by the length of the fiber. If this fiber is too long, time zero will lie outside the
temporal window of the TAC. An optical fiber that is too short will not provide
sufficient temporal separation of the two data channels at the CFD. In order to
determine the correct length of this fiber optic, one must carefully consider the transit
time of photons and electrons in both data channels beginning at the point at which the
laser pulse train is sampled. One should choose the length of this fiber such that the
propagation time through the temporal reference channel equals the minimum
propagation time for the fluorescence channel less the narrowest temporal window of
the TAC. As the maximum scan rate of the TAC combined with the input voltage
range of the multichannel analyzer (MCA) used in our instrument yields a window that
is approximately 20 ns wide, one must accurately account for every portion of each

data channel's propagation time. Of course one must consider the velocity of

 

 

151

propagation through the coaxial cables used in the instrument as well as the refractive

index of the fiber optic when calculating the propagation time for each data channel.

A length of 14 m was chosen for the first fiber optic. This length of optical
fiber delays the temporal reference signal just enough such that time zero lies near the
upper limit of the TAC-MCA window, thus providing optimum conditions for
recording very fast decays. In addition, this length of fiber optic provides adequate
separation of the two data channels at the CFD. The second fiber optic is 20 m in
length, which when added to the 14 m fiber optic allows data to be recorded over a 100
ns window. To acquire data over a longer time period, an external fiber optic can be
inserted in the optical path (by means of two connectors on the rear panel of the NIM
module). At present, this may be a 50 m and/or a 100 m fiber optic. These allow the

user to easily configure the instrument to acquire data for up to a microsecond.

CONIPONENTS

This section contains a description of the major components of the switched
fiber optic delay. Where appropriate, the rationale behind the choices made is
discussed. A complete description of vendors, part numbers, and other information

necessary for the purchase of the components is also given.

Optical Fibers, Connectors, and Switches

Aside from the length of the optical fibers used, several choices must be made

with respect to the type of optical fibers, the style of connectors, and the mechanism

and type of the fiber optic switches used.

152

Optical fibers have been in widespread use, especially in the communications
industry, for several years. Many types of optical fibers are available offering a broad
range of bandwidths, spectral characteristics, package ruggedization, and ease of
handling. General information on optical fibers can be found in catalogs from

manufacturers [l] or resellers [2] of optical fibers.

For this application, high bandwidth, low attenuation from 400 to 700 nm, and
ease of handling are the desired attributes for the fiber optics used. Of the
commercially available fiber optics, single mode fibers possess the greatest bandwidth.
However they are difficult to work with and special equipment is required to mate two
fiber optics. Multi-mode fiber optics, on the other hand, are very easy to handle and
mate, and their bandwidth is sufficient for this application. Of the family of multimode
fiber optics, the least expensive and most common is the so called "100/ 140 G1" fiber.
This fiber optic has a core diameter of 100 um and a cladding thickness of 20 pm. Its
construction is slightly different than the step index model that is commonly used for
explaining the total internal reflection mechanism by which light propagates through the
fiber optic. The refractive index of the cladding increases with distance from the core.
This gradient in the refractive index increases the bandwidth of these "graded index"

fibers relative to step index fibers.

Although the transmission of the 100/ 140 GI fiber is optimized at 1300 nm,
losses in the visible region of the spectrum are tolerable over the lengths needed for a
switched fiber optic delay for TCSPC. This fiber would not be suitable for conducting
light below 400 nm or for conducting visible light over several kilometers. However,
in any fiber optic application one should reCOgnize that the bulk of the loss of the
overall system will probably occur where light is coupled to the fiber. Consider that

over ten percent of the light will be lost as it is coupled to the fiber when a simple lens

 

153

is used to focus collimated light onto the end of a fiber (approximately four percent loss

for each surface of the lens and four percent for the air-fiber interface).

Since the 100/140 GI fiber is widely used commercially, the least expensive and
most readily available package for it is more rugged than is needed in a typical
laboratory. The most common package for 100/ 140 GI fiber consists of the fiber
proper and an additional millimeter or two of strength members enclosed in a PVC
jacket. Fibers covered only with a thin PVC jacket are also available, but they are not
as widely used and hence more expensive than the more rugged version. However, the
near—bare fibers require less storage volume for an equal length of fiber. Jacketed fiber

cables were used in the constructed device because of their lower cost.

Many types of connectors are available for fiber optics. The more advanced
designs enable an ease of interconnection of fiber optics similar to that for coaxial
cables fitted with BNC connectors. In fact one of the better connectors for fiber optics,
the "ST" connector, uses a bayonet mating mechanism similar to the BNC connector.
Double female ST connectors analogous to the BNC "barrel" provide convenient
interconnection of two fibers. Chassis mount connectors designed to mate an ST
connectorized fiber with a TO-l8 "can" package also exist. This provides a very
convenient means of coupling detectors or sources mounted in this popular package to

a fiber optic.

The heart of the switched fiber optic delay is of course a fiber optic switch.
Although the principle is quite simple, this author had a great deal of difficulty in
locating a manufacturer of such a device. Essentially two manufacturers exist: DiCon
Fiberoptics (Berkeley, CA) and JDS Optics (Nepean, ON, Canada). DiCon

manufactures electronically actuated switches, and JDS manufactures both

 

154

electronically actuated and manually actuated switches. Both types of fiber optic switch
operate by a moving fiber mechanism. They differ primarily in the method of
actuation. One by two (1x2) and one by four (1x4) switches are available from both
vendors. For the design shown in Figure Al-l, electronically actuated 1x2 switches

from DiCon were chosen. Cost was the primary criterion in the choice of vendor.

The last component prior to the detector is listed in Figure A1-1 as a "fiber
optic splitter/combiner". The actual component used in the switched fiber optic delay
is perhaps more correctly termed a "multi-leg fiber optic assembly”. Commercially
available, bidirectional splitter/combiners are either 1x2 or 1x4 devices, and cost in
excess of $250 for a 1x2 device and $400 for a 1x4 device. Rather than using an off-
the-shelf device, a less expensive, custom made three fiber optic to one ST connector
combiner was purchased. This device, termed by the manufacturer (Boston
Electronics, Brookline, MA) as a multi-leg assembly, is simply three short fiberoptics
with ST connectors on one end that are brought together at the other end in a single,
drilled out ST connector. Since this connector is intended to mate with what is a
comparatively broad area detector, this means of construction accomplishes the task at
hand while costing less than half the price of a conventional fiber optic

splitter/ combiner needed for this application.

Detection

The last component in the optical path of the switched fiber optic delay is the
detector. Of the relatively inexpensive, fast detectors used for TCPSC, silicon
avalanche photodiodes are the most widely used. The risetime of these devices is less
than 500 ps, and is only surpassed by the fastest PIN type photodiodes which are about

25 times more expensive. Of the commercially available silicon avalanche

 

 

155

photodiodes, those manufactured by RCA (RCA Inc, Electro Optics Division,
Vaudreuil, QU, Canada) have the fastest risetime and the greatest gain. The
photodiode used in the switched fiber optic delay was RCA part # C30902E (TO—18
package).

Since the moderately high DC voltage needed to bias the photodiode is not
directly available from the NIM bin power supply, the circuit shown in Figure A2-2

was developed to provide the necessary bias voltage.

100!)
1/2 w 56 roll 66 k0

Out
DC to DC
converter 0-001 Si Avalanche
; g ! “F ! photodiode

 

-24 V

1000 “FI 1N4746
M2.5A

  
 
 
 

 

 

 

Figure A2-2. Diagram of circuit employed to bias the silicon avalanche photodiode.

The circuit is designed around a DC-to-DC converter (Endicott Research
Group, Endicott, NY) that is rated to provide an unregulated 225 volt output when
supplied with a 24 volt input. Two modifications were needed beyond a simple
insertion of this device between the NIM power supply and the photodiode. The first
was a requirement to reduce the output voltage to less than -195 volts, the breakdown
voltage of the photodiode. According to the manufacturer, the output voltage of the
converter is proportional to the input voltage from the specified input to about 20
percent of that value. Hence a simple Zener shunt regulator [3] was inserted to
regulate the DC-to-DC converter's input voltage to a value that would produce the
required output voltage. The fabrication of the needed regulator was accomplished
with a Zener diode having a breakdown voltage of 18.4 volts and a one Watt power

diode. The second modification was the incorporation of a ”pi" filter [4] at the output

156

of the DC-to—DC converter. The unfiltered output of the DC-to-DC converter
contained an unacceptable level of switching noise at about 25 kHz. The constructed pi

filter did not eliminate the switching noise, but did reduce it to an acceptable level.

The output voltage of the DC-to-DC converter after the pi filter is -187 volts. It

contains 400 mVp_p of 25 kHz switching noise, as measured with a conventional 1 M0
oscilloscope. This does not appear to affect the performance of the photodiode as used
in a SPT experiment, but the dark response of the device contains about 150 mVW of
switching noise as measured with a 1 M9 oscilloscope. This noise level is not

observable in a 50 (1 system.
CONSTRUCTION

A description of the construction and layout of the switched fiber optic delay is
in order should it require repair or modification. As previously mentioned, the device
is housed in a triple-width NIM module. All of the components related to the fiber
optic switches and the photodiode bias supply are mounted on a large piece of perf
board which is mounted on the right hand side of the NIM module. Removal of the
right-hand panel of the NIM module will reveal this board. The only components
mounted on the bottom are the diodes used in the Zener shunt regulator. These
components are the most likely suspects should a problem develop in the photodiode

signal.

The first two fiber optics in the optical delay path are coiled and mounted
between pieces of perf board (and double sided pc board) which are firmly connected
and mounted in "layer cake" fashion with standoffs and long bolts. The fiber optic

assembly can be removed by removing the left hand panel of the NIM module, and

 

157

removing the four brass nuts visible on the bottom of the board exposed when the right

hand panel was removed.

The delay fiber optics are orange and can be readily identified as the length of
each fiber optic is written on the sheath of the connector on either end of the fiber.
The input and output fibers of the switches are also color coded, and the manner of
connection of the fiber optics is shown in Figure A2-3. The figure also shows the

position of each fiber optic switch when power is applied to it.

F.P. Switch
(m)

1
l
0
+
O
I
t

  

+sv
ORANGE/WHITE

  
   

UGHT
PULSE IN

E INDKMTES A F35! omc MOE.

Figure A2-3. Color-coding of optical fibers used in the switched fiber optic delay.

The switches are identical with respect to color coding of the attached fibers.
To aid in identifying the fibers from the fiber optic switches, the fiber optics for the
first switch were lightly painted with "white out" near the fiber's connector. These

fibers are indicated by a color/white label (e. g., red/white) in Figure A2-3.

 

 

 

 

158

The power lines to both of the fiber optic switches are bypassed at the solder

contact for the switch with a 1,000 uF capacitor.

 

The silicon avalanche photodiode and current limiting resistor are contained in a
small "black box" (Pomona Electronics, Pomona, CA) mounted inside the front panel
of the NIM module. The photodiode is mounted in a panel receptacle for an ST
connectorized fiber optic designed to mate the fiber optic with a TO-18 package. The
output from the photodiode is accessed via an SMA connector on the front of the NIM l
module labeled "Fast Out". The output attenuated by a factor of five (2000 series
resistor), is also available on the front panel via an SMA connector labeled "Monitor",

following a suggestion by Steiner, Holtom, and Kubota [5].

 

 

 

 

 

159

CONTROLS AND SPECIFICATIONS

Detailed information concerning the connectors and controls on the front and

rear panels of the NIM module and other pertinent items are presented in this section.

Input

Optical Delay

Outputs

Front and Rear Panel Connections and Controls

FRONT PANEL

ST fiber optic connector, located near the bottom of the front panel.

Three—position switch, allows the user to select different delay times
via selecting different optical paths. Values of 80 and 180 ns are
available with the fiber optics mounted inside the NIM module.
Additional delay times are possible if an external fiber optic is
connected at the rear panel. At present two fiber optics are available;
50m and 100m. These provide delay times of 430 ns (50m) and 680
ns (100m). A delay of 930 ns is possible if these fiber optics are
connected in series.

Two male SMA connectors, labeled "Fast Out" and "Monitor"
provide a very fast signal N ns after a picosecond pulse of light is
applied at the input, where N is the setting of the switch labeled
"Optical Delay". The signal is a negative, Lorentzian-like pulse
having risetime of less than 500 ps and a total width of approximately
one ns. The signal is the output of a fast, silicon avalanche
photodiode, RCA part # C30902E. The full photodiode signal is
available at the connector labeled "Fast Out", whereas the signal at
the connector labeled "Monitor" is the photodiode signal attenuated

by a factor of five.

 

Fuses

External Delay

+ 12 V output

160

REAR PANEL

Two switched fuse holders, labeled F.O switches, 0.5 Amp (for the
Fiber Optic Switches) and Detector, 0.1 Amp. Both are slow blow
fuses.

Two fiber optic ST connectors for the insertion of an additional fiber
optic to increase the length of the longest available delay time. These
connectors should be capped while not in use to prevent the accidental
emission of laser light and contamination of the connectors. At
present, two fiber optics are available for this purpose. These fiber
optics are 50 m and 100 m in length, and provide total delays of 430
ns and 680 ns respectively. A delay of 930 us is possible by
connecting these two fiber optics in series before connecting them to
the NIM module.

BNC connector providing +12 volt power. This is intended to
provide power for the amplifier in the data channel of the TCSPC
instrument (Minicircuits model ZFL 2000). WARNING - _Th_is mwer
s_ceour i_s n_ot M.

 

 

161

ADDITIONAL INFORMATION
Power Requirements
360 mA, +6 volts
50 mA, -24 volts

Vendors, Part Numbers, and Miscellaneous Information

A complete listing of vendors and part numbers for the components in the electronic

switched fiber optic delay is given below. Suggestions for alternate sources are

included for some of the components.

 

Company/ Address Item/Pg Number Approximate 99g
NIM modules

EG&G Ortec Triple-width NIM module $90

100 Midland Rd.

Oak Ridge, TN 37831
(615) 482-4411
(800) 251-9750

Tennelec, Inc. Triple width NIM module $80
601 Oak Ridge Turnpike (alternate source)

Oak Ridge, TN 37831

(615) 483-8405

(800) 255-1978

 

DiCon Fiberoptics
950-C Gilman St.
Berkeley, CA 94710
(415) 528-0427

JDS Optics
P.O. Box 5245
Nepean, ON K2C 3H5

Canada
(613) 727-1303

Boston Electronics

72 Kent St.

Brookline, MA 02146
(617) 566-3821

(800) 347-5445

Light Control Systems
2 Bridgeview Circle
Tyngsboro, MA 01879
(508) 649-9870

OFTI

5 Fortune Dr.
Billerica, MA 01821
(508) 663-6629

162
Fiber Optic Switches

1x2 fiber optic switch
part # SlZ-N—lOO-ST-S

alternate source for
fiber optic switches

Fiber Optics and Connectors

100/ 140 G1 fiber optics
with stainless steel ST
connectors

Multileg assembly

alternate source for
fiber optics

double female ST
fiber optic connector
part # 300-48TC-1002

ST/TO-l8 receptacle
part # 300-48TC-0001

also a source for fibers
and connectors

DC-to-DC converters

Endicott Research Group, Inc. DC-to-DC converter

2601 Wayne St.
Endicott, NY 13760
(607) 754-9187

E512 series, part #
varies according to
availablity

$400

$1.30 per m plus
$25 per connector

$125

$25

 

RCA , Inc.
Electo-Optics

PO. Box 900
Vaudreuil, QU J7V 7X3
Canada

AEG Corporation
PO. Box 3800
Somerville, NJ 08876
(201) 722-9800

 

163
Silicon Avalanche Photodiodes

Silicon avalanche
photodiode
part # C30902E

alternate source for
silicon avalanche
photodiodes

part # BPW28B

$80

$300

 

164

APPENDIX 2 REFERENCES

Technical Literature, Corning Glass Corporation, Corning NY
Newport Corporation, Fountain Valley, CA, 1990 Catalog

Malmstat, H.V; Enke, C.G.; Crouch, S.R. Electronics and Instrumentation for
Scientists Benjamin/Cummings: Menlo Park, CA, 1981, p.62.

The ARRL Handbook for the Radio Amateur, 64th ed. , American Radio Relay
League: Newington, CT , 1987, p. 11-8.

Steiner, R.F.; Holtom, G.; Kubota, Y. in "Laser Spectroscopy of Nucleic Acid
Complexes " in Lasers in Polymer Science and Technology: Applications " vol IV,
CRC Press: Boca Raton, FL, 1989, J .P. Fouassier and J.F. Rabex Eds.

 

 

APPENDIX 3: A USER'S MANUAL FOR SINGLE PHOTON TIMING
EXPERIMENTS

In this Appendix, instructions are given for acquiring a fluorescence decay with
the single-photon timing (SPT) instrument described in Chapter 3. Cautions
concerning some of the more delicate components are also raised. And as they are
non-trivial from the standpoint of curve fitting, suggestions are made for the
determination of multi-exponential decays and the decay of anisotr0py. Specific
acquisition parameters are given for acquiring the excited-state decay of a " standard
compound", rose bengal, in methanol. It is assumed that the reader is thoroughly
familiar with the SPT technique, the operation of the laser, and the acquisition and
analysis software. It is also assumed that the reader has previously read the manuals

for the major components.

This Appendix is not intended to serve as a step-by-step set of instructions for
obtaining a fluorescence decay with the SPT instrument described in Chapter 3; that
would be extraordinarily lengthy. To aid novice users, a flow chart is given, but it is
not intended to serve as a recipe. In this author's opinion, SPT is a technique: a
methodology for recording the decay of excited-states. An SPT apparatus is a tool for
recording this information, not a ”black box” in which a sample is inserted and a
"magic" number is printed out. The importance of a thorough understanding of the
technique, and the components from which the instrument is assembled, cannot be

stressed too much.

Because some of the acquisition parameters must be optimized for the excited-
state decay under study, one essentially adjusts these parameters to suit the decay

under investigation. Since the user obviously does not have a priori knowledge of the

165

 

 

 

 

 

166

particular excited-state lifetime, the adjustments are made on-the-fly. Making these
adjustments properly and efficiently is the key to the efficient use of one's time and the
resource that is the SPT instrument. Once the instrument is fully prepared, one simply

collects data until the desired signal-to—noise ratio is obtained.

GENERAL CONSIDERATIONS

In this Appendix, and elsewhere in this dissertation, the desireability of a
narrow, symmetric, and smooth instrument response function (IRF) is stressed. As
some readers may not fully appreciate the value of these characteristics, a few words
concerning their importance is in order. Although the effects may not be significant, a
measurement scheme in which the system under study is perturbed distorts the quantity
that is determined. This is certainly true for the acquisition of a fluorescence decay.
An understanding of exactly the extent to which the instrument, be it an SPT apparatus
or one based on another technique, distorts the true fluorescence decay is crucial to
correcting the determined excited—state lifetime. An accurate record of the IRF is
needed for the numerical techniques by which the lifetime is extracted from the
experimentalk data. The quality of the IRF affects the ability of these algorithms to
provide an accurate correction. Consider also that an instrument having a short but
skewed IRF will produce less accurate and reliable data than an instrument with a
slightly longer but perfectly symmetric IRF. This is true regardless of the difference

between the excited-state lifetime and the IRF.

The Laser System

Unlike the other experiments that use the output of one of the synchronously

pumped, mode-locked, cavity-dumped dye lasers, SPT is relatively insensitive to

 

 

 

 

167

fluctuations in intensity. However, obtaining a reproducible and narrow instrument
response function (IRF) is strongly dependent on proper adjustment of the dye laser
cavity length and the phase and delay adjustments on the cavity dumper driver. If
satellite pulses are present in the dye laser pulse train, the IRF will not only be
needlessly broadened, it also may not be smooth because the IRF can be fast enough to
record the individual satellites. Such an IRF will reduce the accuracy of reconvolution

analysis. It will also cause additional distortion in the acquired fluorescence decay.

The proper adjustment of the cavity dumper driver can result in contamination
of the pulse train. If the RF pulse to the Bragg cell is not temporally centered with
respect to a pulse in the dye laser, part of an adjacent pulse can be ejected from the dye
laser cavity in addition to the main pulse. If this is the case, a small afterpulse may be
evident 13 ns following the IRF. This corresponds to the interpulse spacing of the

mode-locked pulse train.

Aside from proper alignment and adjustment, one should be aware of another
characteristic of dye lasers. Similar to the plasma background of ion lasers, the
spectral output of the dye laser is contaminated with broad-band fluorescence from the
laser dye. This can cause a distortion if the laser dye fluorescence is significant at the
sample's emission wavelength. This is generally not a problem if one is frequency
doubling the dye laser's output, because only the sharp laser pulse is frequency
doubled.

 

 

 

168

Cables and Connectors

The fast signals in an SPT instrument require high bandwidth coaxial cable,
such as semi-rigid coax and high bandwidth connectors, e. g. SMAs. Although semi—
rigid coax appears rugged, it should be treated with care. Because its shield is a solid
copper tube, semi-rigid coax is formed to the desired shape before the connectors are
attached. If the finished cable is distorted, the connection to the center contact of the
SMA connector can be broken. At the very least, the 50 9 impedance connection will

be altered, which causes a reduction in the cable assembly's bandwidth.

Detectors

There are three detectors in the SPT instrument. They are a silicon avalanche
photodiode, a conventional side-on PMT, and a 12 um MCP-PMT. The photodiode is
housed and biased such that the user need not do anything other than turn on the power
to the switched fiber optic delay. As PMTs are widely used, no cautions will be given
beyond those normally associated with these devices. However, MCP-PMTs are not as
widely used and are more fragile than conventional PMTs. One should handle them
with a great deal of care as they are especially sensitive to physical shocks. Apply and
remove the bias voltage slowly. The instrument response has been optimized for
operation at 3,100 volts. Dp npt exceed the maximum rated voltage of 3,400 volts.
The chiller in which the MCP-PMT is installed need not be operating unless dark
counts are a significant. But since the MCP-PMT can be warmed through ohmic

heating, it is a good practice to circulate water through the housing whenever the MCP—

PMT is biased.

 

 

 

l
I
l
l

169

THE TEMPORAL REFERENCE CHANNEL

A diagram of the instrument that emphasizes the two data channels is shown in

Figure A3-l. The temporal reference channel begins with laser light reflected by the

FIBER UPTIC
DELAY

SI AVALANCHE
PHDTDDIDDE

 

 

STUP START

 

Figure A3-1. Diagram of tthe SPT apparatus emphasizing the two data channels.

pellicle beamsplitter. This light is conducted to the switched fiber optic delay with a
two meter fiber optic. A fiber optic positioner has been modified to facilitate coupling
the laser light into the fiber optic. It has been modified by mounting a small plano-
convex lens its focal length from the tip of the fiber optic. The f—number of the lens
was chosen to match the numerical aperture of the fiber optic. The fiber optic

positioner allows the user to translate the fiber in the X, Y, and Z directions relative to

 

 

170

the point of focus of the laser light. A good method for the alignment of the positioner
involves removal the other end of the fiber optic from the switched fiber optic delay.
This allows one to directly view the intensity of the light conducted through the fiber
optic and more readily optimize the adjustments for the fiber optic positioner. Once
the other end of the fiber optic has been reconnected to the switched fiber optic delay,

the preparation of the temporal reference channel is nearly complete.

The SMA connector labeled "fast out" on the switched fiber optic delay is
connected to one of the channels of the CFD with a semi-rigid coaxial cable that is
about six inches long. This cable has an SMA connector at one end (to mate to the
switched fiber optic delay) and a LEMO connector on the other end (to mate with the
CFD). The SMA connector which mates to the other SMA connector on the switched
fiber optic delay is labeled "monitor" and is terminated at the connector with a 50 Q
terminator. This termination ensures that the signal presented to the CFD is the same
as the signal that was viewed on the oscilloscope when the CFD was adjusted. A delay
cable for this channel of the CFD was constructed from a very short (ca. 2 cm) length
of semi-rigid coax. Optimum values for the walk and threshold adjustments were

0.501 volts and 0.028 volts, respectively.

The output of the CFD is connected to the STOP input of the TAC with
RGl74/U coax. The cable is ca. eight inches long and is connectorized to mate with
both the CFD and the TAC.

THE FLUORESCENCE CHANNEL

The fluorescence channel also begins at the pellicle beamsplitter. Light not

reflected by the beamsplitter continues to the sample. Proper alignment with respect to

 

 

171

the collection optics is achieved by adjusting the mirrors prior to the sample so that the
laser light is centered on the two alignment irises. As discussed in Chapter 3, the
collection optics were chosen to minimize the volume of the sample viewed. This
reduces the effect of the transit time of laser light through the cuvette. If filters are
needed to improve stray light rejection, they should be inserted in the collection optics
between the two lenses. A polarization analyzer can also be placed here. At present,
the LASER lab has two polarization analyzers for the SPT instrument: one for the near
infrared and one for ca. 400 to 700 nm. These are mounted in appropriate holders and

stored in the sample compartment of the SPT instrument.

Either of the two available monochromators may be employed. The SPEX
monochromator is used with the MCP-PMT, and has a mask at the grating as discussed
in Chapter 3. Two gratings are available for this monochromator. The one presently
installed is blazed at 500 nm and the other is blazed at 700 nm. The grating blazed for
the near IR has not yet been used, but may be needed if one is working with a weak,
near-IR emitting fluorophore. Both gratings have 1,200 grooves per mm. Two PMTs
are available for the McPherson monochromator. They are both manufactured by
Hamamatsu, and are model numbers R928 and R406. They have maximum
efficiencies at 400 nm and 730 nm, respectively. Each is installed in a shuttered
housing, and one can easily exchange them as they are mounted with four socket-head

machine screws.

The MCP-PMT is installed in a refrigerated housing. If this detector is chilled,
the user should make certain that the cooling water is flowing before the Peltier cooler
is powered. About two hours are required for the temperature of the detector to
stabilize. The equilibrium temperature inside the housing is about 40 degrees Celsius

below the temperature of the cooling water.

 

172

The signal from either detector is conducted through a short length of coaxial
cable to a 1 dB feedthrough attenuator. This attenuator is connected to the input side of
a 2GHz amplifier. The amplified signal is fed directly to a CFD. Again, semi-rigid
coax is used to conduct the signal from the MCP-PMT, and RG174/U flexible, coaxial
cable is used for the PMT. The CFD channels for these detectors have been optimized
for the respective signals. A short length of RG174/U is used for the delay for the
PMT, and, as discussed in Chapter 3, the CFD for the MCP-PMT was modified to
accommodate the very fast risetime of the MCP-PMT signal. The output of the CFD is
passed through the switched coaxial cable delay before being fed to the START input
of the TAC. As discussed in Chapter 3, the switched coaxial cable delay is needed to
eliminate cross-talk between channels of the CFD, and is set so that “time zero" is near
the end of the TAC's timebase. The delay setting is most easily optimized while

recording an IRF.
THE TAC AND MCA

The two data channels converge at the TAC. The time base of the TAC can be
continuously varied from 25 ns to 3 ms. The MCA response is sharpest when the TAC
output width is ca. 3 us. At maximum temporal resolution (TAC set to its maximum
sweep rate and the MCA set to maximum gain) setting the switched coaxial cable delay

to 53 us will locate the IRF near the middle of the 15th 16th MCA memory group.

Data acquisition over just a few nanoseconds is accomplished with a non-zero
MCA offset and selection of the appropriate memory group. The MCA input voltage
range is resolved into a number of channels equal to the set MCA gain. The offset

inserts a pedestal voltage equal to the selected number of channels of offset. The

 

 

 

173

memory group is the fraction of the selected gain that is displayed on the screen.
Hence, to acquire only the 512 channels of the memory group mentioned above, one
would set the offset to 7,684 and select the 15th 16th memory group. A smaller offset
and larger memory group will, of course, increase the time window over which data

are acquired while keeping the temporal resolution constant.

MORE ON THE LIGHT SOURCE

Light from the lasers is vertically polarized. Two half-wave plates are available
for polarization rotation. They are Newport "zero-order" half-wave plates for 300 and

585 nm. Each is mounted in an appropriate holder.

Excitation in the UV can be realized with the second harmonic generator
(SHG). It consists of two lenses, a KDP crystal, and a bandpass filter for the UV (see
Appendix 1). It is useful for doubling the output of the rhodamine 66 dye laser. If it
is utilized, it should be inserted between the two alignment irises. One should keep in
mind that the polarization of the frequency doubled light is rotated 90 degrees relative
to the visible laser light. Hence the UV is horizontally polarized.

The period of the dump frequency of the cavity dumper driver should be set to
at least five times the anticipated lifetime. Although it can be set to 4, the minimum
working value of the divide by N setting is 10. The stability of the dye laser generally
increases with the value of this setting, and 10 is a good compromise between laser
stability and tuning range, and acquisition time. Obviously the divide by N setting
must be increased if one is recording the decay of a long-lived molecule. A special
trigger box has been constructed to allow the repetition rate to be reduced further than
the minimum rate of the cavity dumper driver (148 kHz for N = 259). It is fed by the

 

 

 

174

38.25 MHz signal from the mode-locker driver and its output is connected to the cavity
dumper driver's "EXT PLS IN" input. The appropriate front panel control can be set
to allow the repetition rate to be selected by the frequency of the external signal.
However, the 38.25 MHz signal from the mode-locker driver should not be connected

to the cavity dumper driver's 38 MHz input.
DATA ACQUISITION

Once all of the components are powered and ready, acquisition can commence.
It is generally best to start with the IRF. This ensures that all of the components are
functioning properly, and it takes just a few minutes. One of the best means of
acquiring an IRF is to monitor Raman scattering from cyclohexane. The strongest
Raman band occurs ca. 2850 cm'1 from the laser line. Unless the cyclohexane has been
purified (see Chapter 3), fluorescence from impurities will be observed. One should
use constant and narrow monochromator slits (i.e. 0.5 mm) to restrict the volume of
the sample that is viewed. One should narrow the alignment irises If one needs to
reduce the TAC valid convert rate. This attenuates the fluorescence intensity without

changing the IRF.

After checking the IRF, one can begin acquiring the fluorescence decay. In
order to maintain the integrity of the IRF, one should keep as many of the acquisition
parameters constant as possible. This includes such seemingly trivial items as the
number and type of filters used in the collection optics. Clearly, the IRF measurement
must be repeated if parameters such as the TAC time base are changed. Remember
that the decay should be acquired such that it is bracketed by dark counts. It should be
immediately evident if the time over which the decay is recorded should be increased or

decreased. The first step to increase the temporal window over which the fluorescence

 

175

decay is acquired is to decrease the TAC sweep rate. Because the TAC has a 0.5 to
1.5 X multiplier in addition to a step-wise time base setting, one can "fine tune" the
timebase for the fluorophore at hand. This will work to a point, but a limit willquickly

be reached.

This limit is due to the reversed timing configuration. Because the STOP signal
must reach the TAC after the decay is complete, one must increase the delay in the
temporal reference channel to acquire more of the fluorescence decay. (This is not
possible in some SPT instruments, but it is easily accomplished with the one
constructed in this work.) The delay for the temporal reference channel can be
increased with the switched fiber optic delay (see Appendix 2). Unlike the TAC
timebase setting, the switched fiber optic delay cannot be varied in a continuous
fashion. However, by optimizing the TAC time base, the MCA offset, and the two
delays, one can acquire the fluorescence decay over a time window that is reasonably
close to a time corresponding to 57'. Also consider that the region of data for curve
fitting can be specified with the analysis software. Hence, one can acquire the decay
over a longer than needed time window. The only disadvantage to this procedure is

that the resolution suffers slightly.

If any of the TAC or MCA parameters have been changed, another IRF with
the same settings as those used for the sample must be acquired. However, if the decay
time is long in comparison to the IRF, one does not need to include reconvolution in

the data analysis.

176

Complex Decays

The complexity of the decay can be estimated by viewing the data as log counts
versus channel number. This is easily accomplished with the MCA software. If a
complex decay is evident, one should acquire data until the peak channel contains on
the order of 10,000 counts. Although 1,000 counts in the peak channel is sufficient for
simple decays, a higher signal-to-noise ratio is needed to confidently curve fit with a
more complex model. Even though more counts are preferred for complex decays, it is

good practice to save the data at 1,000 counts in the peak channel.

This is more important for the decay of anisotropy. Because the quantity of
interest is dependent on the dzfierence between multiple decay curves, a high signal-to-
noise ratio is absolutely necessary. In fact, the number of dark counts should be
roughly equal in each of the polarization-dependent decays to achieve the greatest

confidence in the determined quantities.

For complex decays, the challenge is not in acquiring the data, rather it is in
obtaining a good fit. This is generally achieved only after expending some effort.
However, keep in mind what makes kinetic sense for the compound under study.
Consider that the acquired decay may be contaminated with fluorescence from an
impurity. The single-photon sensitivity of the technique requires that the sample be as

pure as reasonably possible. This includes the solvent and the sample proper.

 

177

Time Calibration

The procedure for a time-calibration was discussed in Chapter 3. The only
point that will be reinforced here is that the TAC valid convert rate should not exceed 1

kHz.

FILES AND NANIING CONVENTIONS

One generates several files over the course of an excited-state lifetime
determination. Because their identity can easily be forgotten, it is good practice to
follow simple conventions for the file names. While the file name itself should be
chosen to reflect the sample under investigation, the extension should be chosen to

indicate the type of file. Consider these suggestions:

Name.EXT Identigr

SAMPLEDAT MCA data file

SAMPLE.199 data file in 199 format

INS@710.DAT IRF @ wavelength (MCA)

SAMPLEFIT single exp fit

SAMPLE.2FT double exp fit

SAMPLERES residuals file

SAMPLEAUT autocorrelation file

2NSTAC.DAT time calibration w/ 2 ns steps

SAMPLE.1MP impulse response functions
(when global analysis is used)

SAM-HH.199

SAM-HV.199 to indicate excitation and

SAM-VH.199 emission polarizations for

SAM-VV.199 anisotropy measurements

 

178

A REFERENCE FLUOROPHORE

For future reference, the parameters for acquiring the decay of rose bengal in
methanol are described in detail. The compound was from Aldrich and the solution
was 10 uM in B&J Brand methanol. The sample was excited at ca. 570 nm with 5 ps
FWHM pulses of light. The laser power at the sample was too low to be measurable
with the Coherent model 210 power meter. The repetition rate of the laser was 3.8
MHz (cavity dumper divide by 10). Fluorescence was detected at 600 nm, and the
entrance and exit slits were 0.5 mm. No filters were included in the collection optics.
The MCP-PMT was biased at 3,100 volts. The coaxial cable delay was set to 53 ns.
The switched fiber optic delay was set to 80 ns. The TAC timebase was 25 ns. The
MCA parameters were: gain = 8192, offset = 6144, and the memory group was the
fourth quarter. The time calibration was 2.77 ps/ch with these parameters. The TAC
valid convert rate was adjusted to ca. 1,200 by narrowing the two alignment irises.
The TAC valid convert rate for dark counts was about 25 counts per second. The
acquisition was stopped when ca. 1,000 counts were acquired in the peak channel
(channel 250). This occurred after 4.58 minutes of data collection. The fluorescence

decay was indistinguishable from the background by channel 1,400.

An IRF was also acquired. This was accomplished by monitoring Raman
scattering from cyclohexane. Except for the scattering wavelength of 682 nm and the
TAC valid convert rate of ca. 500, all of the other acquisition parameters were the
same as those for the fluorescence decay of rose bengal. The peak channel of the IRF

was channel 234. Acquisition of the IRF was complete after 0.4 minutes.

The fluorescence decay of rose bengal in methanol and the IRF are listed in a

log book of standard fluorescence decays.

 

 

179

A FLOW DIAGRAM FOR SPT

To facilitate the training of novice users, a flow diagram of the basic sequence
of an SPT experiment is given in Figure A3-2. It is assumed that the laser is on and
has been properly adjusted for SPT. If the LED in the temporal reference CFD is on

and has a constant intensity, the temporal reference signal is probably functioning

 

| LASER OPERATING AND TUNED UH

 

 

 

IS TEMPORAL REF. VORKING PROPERLY? IF NUT CHECK ALIGNMENT OF FIBER
(CHECK LED ON THE CF D) OPTIC POSITIONER

 

 

 

 

 

 

 

 

[ CHECK THE IRF ]

l

RETUNE THE MONO.
ADJUST FILTERS.
AND INSERT THE SAMPLE

l

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

[

IS A LONGER TIME IF SO
1 1) INCREASE THE TAc TIMEBASE

VINDUV NEEDED? 2) INCREASE THE FIBER DPTIc DELAY

L 3) USE CDAxrAL CABLE DELAY TD MAKE

MINOR ADJUSTMENTS
TSDRJE’DJcEHTEHELi/GANLIDTcirRrixs/ERT 4) DECREASE THE LASER DUMP RATE

RATE TD 1 kHz. I

 

 

 

 

 

l

COLLECT DATA UNTIL THERE
ARE AT LEAST 1.000 COUNTS IN

THE PEAK CHANNEL

1

STOP THE ACQUISITION
VHEN A SUFFICIENT
NUMBER OF COUNTS HAVE
BEEN ACQUIRED

 

 

 

 

 

 

 

 

 

[_SAVE THE DATA IN 199 FORMAT

J

IF NEEDED. REPEAT THE
MEASUREMENT OF Tl-E IRF

 

 

 

 

 

 

 

[PERFORM A TIME CALIBRATION]

 

 

[CURVE FIT THE AcoUIRED DECAY<Srj

Figure A3-2. Flow diagram for SPT Experiments.

 

 

 

 

 

180

properly. If it is off or flickering, the fiber optic positioner probably needs
realignment. It is best to begin with an IRF - a good choice is to use Raman scattering
from cyclohexane. The first-time user should set the MCA gain to 8192, use no offset,
and examine the full MCA memory (memory group F). The TAC should be set to its
maximum sweep rate of 0 to 10 volts in 25 ns. The switched fiber optic delay should
be set to the minimum delay (80 ns) and the coaxial cable delay should be set to 53 ns.
Excitation can be from either of the dye lasers or the frequency doubled output of the
rhodamine 6G dye laser. The wavelength of the dye laser can be estimated to within a
nm with the calibration plots that are kept near the dye lasers. The laser power at the
sample should be reduced to ca. 10 mW by reducing the aperture of the alignment
irises. The strongest Raman signal is 2850 cm'1 from the exciting laser line. The
monochromator slits should be set to 0.5 mm. The user may or may not place
appropriate filters in the collection optics. If the MCP-PMT is used, it should be
biased at 3,100 volts. The high voltage Should be added in small steps with the shutter
closed. One should start to see dark counts at ca. 2,600 volts. These are manifested
by brief flashes from the LED in the CFD for the MCP-PMT. The number of dark
counts that are generated during the selected time window can be monitored at the TAC
valid conv. out BNC. After the MCA has been set to ACQUIRE mode and the shutter
has been opened, one should see an increase in the number of valid TAC conversions
per second, and the emergence of a sharp peak should be observed on the MCA
display. The alignment irises should be adjusted to yield a TAC valid convert rate of
ca. 1 kHz. Because the Raman signal is concentrated over a very narrow time window,
only a few seconds of acquisition are needed to obtain 1,000 counts in the peak channel
when the TAC valid convert rate is 1,000 Hz. At this point, the user may wish to
move the temporal location of the IRF slightly by toggling the lns or the 2ns delay
switch on the coaxial cable delay. After the delay has been set such that the IRF is

 

181

located near the highest MCA channel and a good IRF has been acquired, the MCA
should be paused and the data should be saved. It is generally good practice to save the
data as both an MCA *.DAT file and an IBH *.199 file. After the cyclohexane has
been replaced with the sample (a novice may want to start with a reference compound
such as rose bengal or rhodamine B) and the monochromator has been retuned to the
sample's emission wavlength, one can start to acquire a fluorescence decay. For the
case of rose bengal, the excited-state lifetime is very short, and one should not have a
problem acquiring the complete decay. Such a molecule is also helpful for acquainting
the user with the MCA acquisition parameters of gain, offset, and memory group.
Blindly adjusting these parameters can be very frustrating, so READ THE MCA
MANUAL FIRST 1 A new user should have no problem reproducing the decay of rose

bengal in methanol as shown in the log book of reference fluorescence decays.

If a longer-lived compound is being studied, one will probably have to increase
the width of the temporal window over which a decay is recorded. The first step is to
increase the TAC timebase. But as mentioned earlier, one may also have to increase
the delay in the temporal reference channel. It may also be necessary to decrease the
cavity dumper repetition rate. Again the coaxial cable delay can be used to shift "time
zero". By adjusting these parameters and those of the MCA, one can acquire

fluorescence decay data from an upper limit of ca. 1 us down to the limit imposed by

the IRF.

 

 

 

 

 

 

 

 

APPENDIX 4: A SIMPLE FREQUENCY SYNTHESIZER

As funds to purchase a commercial frequency synthesizer could not be obtained,
a simple, single-frequency device was constructed. The basic design behind this device
is rather simple. The RF frequency which is used to step down the frequency of
measurement to the cross-correlation frequency is generated by heterodyning. The
cross-correlation frequency is mixed in a double-balanced mixer with a radio frequency
(RF) that matches the modulation frequency of the laser. The output of the mixer
contains frequency components of RF :1; n6 where RF is the radio frequency, 6 is the
cross-correlation frequency, and n = 1,2,3 The desired frequency is RF + 6
(n=1), and it is isolated from the comb of frequencies generated by the mixer with a
crystal filter. This frequency, 76.8243 MHz, is then amplified and fed to the mixers

used to heterodyne the signal from the photomultiplier tube.

Relatively low gain, high bandwidth RF amplifiers are needed to amplify the
signals in the frequency synthesizer and the output of the crystal filter. These
amplifiers were built in-house using commercially available RF amplifiers constructed
from two RF transistors. The amplifiers used were from the MAR series manufactured
by Minicircuits (Brooklyn, NY). Each amplifier is built on a small (5 cm by 7 cm)
piece of 50 0 epoxy glass "GlO" printed circuit (PC) board. The circuit is shown in
the PC board layout illustrated on page 183 in Figure A4—l. Facilities to properly
fabricate PC boards were not available, so an alternate method was developed. It was
discovered that by carefully cutting and lifting the copper foil off of one side of new,
double sided PC board, the unwanted copper could be removed. Although far from
perfect, this technique sufficed to prepare PC boards with nearly 50 0 traces. A trace
that is 0.108 inch wide on 610 PC board will have an impedance of 50

182

 

 

183
+V

RF m RF out

 

[:1 Surface mount 'chlp' capacitor
—+— MAR series RF amplifier

0 Pseudo plated-through hole

Figure A4-l. Printed circuit board layout for MAR series amplifiers

0. The traces made in this manner were close enough to 50 0 for the low gain
amplifiers in the MAR series, but not the high gain amplifiers. Plated-through holes
were simulated by drilling a small hole in the PC board and inserting a small piece of
22 gauge wire. The wire was folded over and soldered to both sides of the PC board.
The power supply scheme suggested by the manufacturer was followed. Power is
supplied through a current limiting resistor to a small pad on the PC board. The
current path is from this pad to the output side of the amplifier in the 50 0 line through
an inductor (100 pH to 1 mH, Dale IL-2). This arrangement supplies power to the
amplifier while keeping the RF from leaving the 50 0 line. Two bypass capacitors (0.1
pF tantalum and 0.01 “F ceramic) are placed between the pad and ground to provide a
low impedance path to ground for any RF that does leak through the inductor. The RF
at the input side of amplifier travels through a surface mount "chip” capacitor (0.1 or
0.01 pF) before reaching the amplifier proper. This acts as a DC block to prevent the
DC power from traveling down the 50 0 line. A chip capacitor is also inserted in the

output 50 0 line for the same purpose. The traces are brought to a point at the legs of

184

the amplifier to minimize step discontinuities which can cause reflections at high
frequency. The PC board is mounted in an RF shielded enclosure (Pomona "black
box"), and the input and output signals are conducted through BNC connectors. The
shield of each BNC connector is soldered to the ground plane of the PC board on both
sides of the 50 9 line. These provide the only path to ground in the circuit. Amplifiers
constructed in this fashion provided gains within a few percent of the manufacturer's

specification.

A diagram of the completed synthesizer is shown in Figure A4-2.

 

(500 an—p
E 76.5 MHz)
l (40 an-p. RF
370 nV - LED
38.25 MHz In FD—e RF P P
(750 an-p)

    
  
  

 

 

1
325 KHz m
(630 nV dlgltol) L“

(50 an-p)

Crystal
Fflter

(200 an-p RF
2 Vp-p LO)

 

 
 

76.824 MHz out
(960 an-p)

   

l

__ (375 an-p RF
(250 nVP D) 34 vp-p LU)

Figure A4-2. Diagram of the completed synthesizer showing voltage levels at each
point in the circuit.

There are two inputs to the synthesizer. These are a digital signal at the cross-
correlation frequency and an RF signal at exactly half of the fundamental modulation
frequency of the laser. Both signals are derived from a cavity dumper driver which is
synchronized to the mode locker driver. The former is obtained by setting the divide

by N setting on the cavity dumper driver to 117. This provides a digital signal at the

 

185

sync output equal to the RF frequency (38.25 MHz) divided by 117. This is the cross-
correlation frequency (325 KHz). The RF Signal originates in the ovenized crystal
oscillator in the mode locker driver, and it is accessed at the 38 MHz out connector on
the cavity dumper driver. The RF is exactly half of the mode locked frequency. This
is because there are two nodes, and hence two loss points, along one wavelength. The
RF is doubled in a frequency doubler (Minicircuits FD-2) and amplified before being
fed into the RF port of a double balanced mixer (Mincircuits ZAD-l). The 325 KHz
signal is also amplified before the mixer, and this signal is connected to the LO port.
The frequency comb generated in the mixer and accessed at the IF port is amplified
twice prior to the crystal filter (Four pole bandpass filter, K&L Microwave, Phoenix,
AZ). The output of the crystal filter is amplified twice such that the power level is

sufficient to drive the two mixers that follow the two PMTs.

The components of the frequency synthesizer are mounted on a 12" by 32"
piece of double sided PC board to which several pseudo plated-though holes have been
added. Fifteen-volt power is supplied from a single supply (Power One) from the
underside of the large PC board. A line filter was placed at the connection to the 110
VAC system to further reduce the chance that residual RF would leak into the AC

power system. The large PC board and the power supply are mounted in a wooden

box for stability.

 

 

 

 

-»-,
-u..-..

“1...
u)..-
u....,.v.
“In

 

 

 

I‘ikt