,1

. 12’

 

u. \l .u . I . ll!
v 1 I . nu. . .
(I! tvouiuus ..‘ [H.331 :2 htfl.
. '5‘ ‘1.)1 . IID-‘I. [all .1 O 0
IH‘.l.\§.. Elyawflfil‘ .. . .llu. AJ .0

“0...... H.,.:,.u..1 . . .. .

pd- .u . .11 t . ‘ n .
.4." .I. . .4. . .

Ill ll!

 

Hm...” . Fan.

.1.
v»!

otvbl .ml. “2:! .

. barn...

. ; ,iu .r. . ..
, uunfiMtRfitluu

1.3!. 11. .-

1 3:23.. 1: :
.l" u

Pal. . Us}?

awnh {38.33% .

In: \ubtl’co..bll

7
{VI ..
DJ v~t

.
‘I ‘
laugh:
‘t‘?!’ A
I]. .0 A
.

. .
iv‘f. ' 5
.v‘lnlo .13 .h.
r A.
II .VI 0
. . .. .

«.0. v .

u ..
.

.
In

.1 rant .
..V 1.. X".

. .
all ‘7 I?! r.
Itch)“ v0“..|lll!o"u:..0hu.,
\7‘-.. ‘L.

Ill‘A‘I' I.
|vtll ’IA. .vl.v
‘ I \Iltdl.

. .. .u..n..3........,¢nu

Q\ r
. ....x..l.dn.....)n.n.wt.. :33; .
. i . 2...,

5': YD.)
:

'Oiltozi..vtn ll.
. , 1“);A\V.u.yv. .‘s

 

.Olii‘u
.. V‘}!

‘ I‘ll.
1 1"
‘I!

o I II

  

fiuhhr ll \‘ Oil. ‘n‘fol‘ 1|
.ll'10!{unt .!
l. 'Dlll‘lbl

   
  

 

4 1.1:l DD
.uIlLI 4.
.¢l¢. ’51

. .ll:‘!o..

llnnl! [UV

   

“thiuilc‘\.v III". I. .30. .33
It‘lII‘ v o hfl‘ 0|
. \‘IYI‘O‘IQ. nu lug .
1‘1." Ill| 1‘ {till ‘0‘?Iot‘t‘¥o‘ . o
\cI )‘l‘l. |I|I Q‘lln .. .Ivtui lit: :79»
.. II It! 1 lin.noip-|OOQ. ,v , 11.1..

. v ‘5’... .. .u‘ I
‘1‘. Jr.a -(‘Vl‘fil‘fi
.q _ .{9‘}...

o.'
I“ul
.unl...

pl. 1

 

. .( '.
'isiz-

I"
”'6.“
Jul“

:1

 

 

.»‘ .
new-ui

 

~. r n I" ,
. A ' 1. , . , .
' . “mull” '

 

1145316

This is to certify that the

dissertation entitled

Methods for Absorbance-Corrected
Chemiluminescence and Fluorescence

presented by

Eugene Ratzlaff

has been accepted towards fulfillment
of the requirements for

DOCtora] degree in Chemi StY‘y

 

 

 

1%:7/4 “A

)
flajor professor

DatelZ/Ze//gl

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

 

 

METHODS FOR ABSORBANCE-CORRECTED
CHEIILUIINBSCENCE AND FLUORESCENCE

By

Eugene Henry Rntzleff

A DISSERIKTION

Submitted to
Michigen Stete university
in pertiel fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Depettnent of Chenietry

1982

 

G I “EQOGCIJ

 

 

ABSTRACT

METHODS FOR ABSORBANCE-CORRECTED
QHHMILUIINESCENCB AND FLUORESCENCE

By

Eugene Henry Ratalaff

A mathematical model of the secondary inner-filter effect in
chemiluminescent solutions has been developed to account for
measurement errors due to the attenuation of radietion by absorption.
This model has allowed the derivation of a correction function based
on a determination of the ratio of chemiluminescence intensities from
two pathlengths. aninol chemiluminescence was observed from
solutions containing the interfering chromophores picrate and ferroin.
Absorption-free signals are accurately calculated for matrices with
absorbances ranging to more than 0.75. An emission spectrum
demonstrating ferroin catalysis is recovered from an apparently
quenched reaction.

A flow cell has been designed for simultaneous fluorescence and
absorbance measurements. A bifurcated fiber-optic is used for
excitation and front-surface fluorescence collection. Convolution of
mathematical descriptions for primary and secondary absorbance effects
and the fiber-optic/cell light transfer function results in a model
for an inner-filter effect correction function. The

absorbance-corrected emission of quinine sulfate is a linear function

of concentration from 0.1 ul to 400 pl. despite self-absorption values
extending to 3.0. Fluorescence signals are corrected for primary
and/or secondary_absorbances in excess of 3.0.

Iicrocomputer-interfaced instrumentation and software were
deve10ped for these luminescence experiments. A versatile.
programmable timer/counter and a' simultaneous sampling, quad input
analog-to-digital data acquisition interface were used. A low cost
photodiode detector was engineered for stability and wide dynamic
range. A hierarchical computing environment provided the software
power and hardware flexibility needed to implement these systems.

A mathematical luminescence volume-source model was deveIOped for
evaluation and development of these absorption-correction methods.
The model verifies a collimation assumption required for the
absorbance-corrected chemiluminescence method. Derivation of
equations for the absorbance-corrected fluorescence technique required
an understanding of the fiber-aptic/cell transfer function which was

evaluated with the luminescence volume-source model.

Tb Jewel.
to my parents,

and to my family

ii

ACKNO'LEDGIENTS

my gratitude and thanks go to Professor Stanley Crouch for his
support. friendship, and for many freedoms enjoyed under his guidance.
I am indebted to Professor Chris Enke for his contribution as second
reader. and to the other members of my committees, Professors Clarence
Suelter. Jack Holland. and James Dye.

[any others should be held responsible. Foremost among these is
Charlie Patton. whose companionship helped make my work creative and
my frustrations bearable. I shall not soon forget many. many others
from whom I learned and benefited. and with whom I shared friendships
- Dave Christmann. Jim. Geno. Clay Calkin. Rytis Balciunas. Rob
Thompson, Frank Curran. Pete Aiello. Paul lraus. Keith Trischan. Pat
Iiegand. John Stanley. Iarguerite lartin. Siriwan 'Iim'
Ratanathanawongs. Hark Victor. and Bob larfman. to name just a few.
Many members of the Enke group are particularly culpable. others of
you are certainly aware of your complicity.

I gratefully acknowledge many others for their invaluable expert
help, advice, and craftsmanship: Professor Andrew Timnick, Dr. Tom
Atkinson, [arty Babb. Russ Geyer. Dick lenke. Ron Haas. Scott
Sanderson, John Funkhouser, Andy Seer. Jo lotarski. and other staff

and friends.

iii

TABLE OF CONTENTS

Chapter Page

LIST w ram 0 O O O O O O O O O O O O C O O O O O O O O O O O Vii

LIST OF FIGms O O O O O O O O O O O O O O O O O O O O O O O O Viii

W I - manual“ 0 O O O O O O I O O O O O O O O O O O O 1

CHAPTER II - ABSORBANCE-CORRECTED CHENILDIINESCENCE
IEASURENENTS WITH A DUAL-PATHLENGTH
mm C O O O O O O I C O O I O O O O O O I 4

A. Introduction and Historical . . . . . . . . . . . . . . 4
1. Principles of Chemiluminescence . . . . . . . . . 4
2. Problems with Chemiluminescence lethods . . . . . 5

B. Theory . . . . . . . . . . . . . . . . . . . . . . . . 8
1. Attenuation by Inner-Filter Effect . . . . . . . . 8
2. The Dual-Path Method of Correction . . . . . . . . 11
C. Experimental . . . . . . . . . . . . . . . . . . . . . 12
1. Apparatus . . . . . . . . . . . . . . . . . . . . 12
2. Reagents . . . . . . . . . . . . . . . . . . . . . 17

3 O Pro°.dnr° O O O O O O O O O O O O O O O O 0 O O O 18

iv

TABLE OF CONTENTS

Chapter Page

LIST OF rum 0 O O O O O O O C O O O I O O O O O O O O O O O 0 vii

LIST OF FIGms O O O O O O O O O O O O O O O O O O O O O 0 O O Viii

m I - monumon O O O O O O O O O O O O O O O O O O O O 1

CHAPTER II - ABSORBANCEPCORRECTED CHEIILUIINESCENCE

IEASUREIENTS ITTH A DUAL-PATHLENGTH
SPECTROMETER . . . . . . . . . . . . . . . . . . . 4
A. Introduction and Historical . . . . . . . . . . . . . . 4
1. Principles of Chemiluminescence . . . . . . . . . 4
2. Problems with Chemiluminescence Methods . . . . . 5
8. Theory . . . . . . . . . . . . . . . . . . . . . . . . 8
l. Attenuation by Inner-Filter Effect . . . . . . . . 8
2. The Dual-Path lethod of Correction . . . . . . . . 11
C. Experimental . . . . . . . . . . . . . . . . . . . . . l2
1. Apparatus . . . . . . . . . . . . . . . . . . . . l2

2 O h".nt' O O O O 0 O O O O O O O O O O O O O O O I 17

3. PrOOOdnr. O O O O O C O O O O O O O O O O O O O O 18

iv

Chapter

E.

CHAPTER IV - INSTRDIENTATION AND SOFT'ARE FOR

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

1. Choice of Chemical System

2. Determination of Sample Absorbance . .

3. Effectiveness of Correction Procedure

Conclusions . . . . . . . . . . . . . .

III - A FIBER-OPTIC INTERFAC. m CH”. FOR
SIIULTANEOUS ABSORBANCE AND PRIMARY
INNERPFILTER EFFECT CORRECTED FRONT
SURFACE FLUORESCENCE IEASDRENENTS

Introduction and Historical . .
Theory . . . . . . . . . . . .
Experimental . . . . . . . . .
1. Reagents . . . . . . . . .
2. Apparatus and Procedure .

Results and Discussion . . . .

1. Determination of Cell Parameters

2. Effectiveness of Correction Procedure

for Primary Absorbance . . . . . . . .

3. Effectiveness of Correction Procedure

for Secondary Absorbance . . . . . . .

Conclusions . . . . . . . . . .

LUIINESCENCE IEASURENENTS . . . .

Introduction . . . . . . . . . . . .

A Hierarchical Computing System . . .

Page

19
19
21
3O

37

38
38
41
41
47
48
51

51

56

60

62

66
66

66

Chapter

C. A Hodular Hicrocomputer . . . . . . . . .

Basic Hodules . . . . . . . . . . .
Counter/Timer II Board . . . . . . .
Analog Input . . . . . . . . . . . .
Sample-and-Hold Hultiplexer Board .

General Purpose 12-Bit ADC Board . .

D. A Photodiode Detector . . . . . . . . . .

E. Hicrocomputer Software . . . . . . . . .

1.

2.

3.

F. Software for a Dual-Pathlength CL

G. Software

Choosing a Microcomputer
Language/Operating System . . . . .

SLOPS O O O O O O O O I O O O O O 0

CHAPTER V - A UNIFIED NATHENATICAL LUNINESCENCE

VOLUNE-SOURCE NOBEL

A. Posing the Problem . . . . . . . . . . .

B. The Nodel .

C. Application to the Dual-Pathiength CL Technique

D. Application to the Fiber-Optic Coupled
Fluorescence Technique . . . . . . . . .

Spectrometer

CHAPTER VI -CONCLUSIONS . . . . . . . . . . . . . . .

BIBLIOGRAPHY

vi

for an Absorbance-Corrected Fluorimeter

Page

67
68
69
70
71
76
79

81

81
83

83

86

88
88
90

94

95

97

102

LIST OF TABLES

Table Page

2-1 Comparison of pathlength values . . . . . . . . . . . . 25

vii

LIST OF FIGURES

Figure Page
2-1 Schematic cell model . . . . . . . . . . . . . . . . . 9
2-2 Top view diagram of dual-pathlength CL

cell and collimator . . . . . . . . . . . . . . . . . . 13
2-3 Schematic diagram of instrument . . . . . . . . . . . . 15
2-4 Increasing absorbance of luminol

reaction matrix with time . . . . . . . . . . . . . . . 20

2-5 Spectral distributions: luminol emission
(arbitrary units. solid line. right ordinate);
picrate absorbance (dashed line. left ordinate):
ferroin absorbance (dotted line. left ordinate) . . . . 23

2-6 Absorbance as a function of chromophore
concentration from (X) transmittance
measurements and (0) CL ratio measurements . . . . . . 28

2-7 Correlation of absorbance values determined
by transmittance and ratio measurements:
(0) in presence of ferroin: (X) in presence
of picrate. Line has ideal slope of 1 . . . . . . . . 29

2-8 CL intensity as a function of chromcphore
concentration. (I) Inner-filter effect
attenuated. (O) Absorption-corrected.
Absorbance axis is an approximate scale
for reference purposes . . . . . . . . . . . . . . . . 32

2-9 aninol emission spectra (arbitrary units.
left ordinate) and ferroin absorbance spectrum
(right ordinate). (0) Luminol emission
reference spectrum. (I) Emission spectrum
attenuated by inner-filter effect
due to ferroin absorbance . . . . . . . . . . . . . . . 36

viii

Figure

3-5

3-7

Page

Schematic cut-away diagram of bifurcated
fiber-optic fluorescence flow cell . . . . . . . . . . 43

Schematic diagram of fluorescence instrument . . . . . 49
Quinine sulfate absorbance at 365 nm . . . . . . . . . 52

Fluorescence intensity (65.535 - 1.0 nA PlT

anode current) vs. quinine sulfate concentration;
linear-linear plot. (A) Uncorrected;

(Y) Absorbance-corrected. 365 nm excitation

wavelength; 450 nm emission wavelength . . . . . . . . 54

Fluorescence intensity (65,535 I 1.0 nA PlT

anode current) vs. quinine sulfate concentration;

log-log plot. (A0 uncorrected;

(V) Absorbance-corrected. 365 nm excitation

wavelength; 450 nm emission wavelength . . . . . . . . 55

Quinine sulfate (OS) and 1.5-dihydroxybenxoic
acid (DHBA) fluorescence intensity
(65.535 . 1.0 nA PlT anode current):
(I) Primary-absorbance attenuated and
(D) absorbance-corrected fluorescence
from 20.0 pl 08 with 0.0 to 0.8 ml DHBA:
(A) Primary-absorbance attenuated and
(V7 absorbance-corrected fluorescence
from 0.0 to 0.8 ml DHBA;
(X) Absorbance-corrected fluorescence
of 20.0 ul OS in the presence of
0.0 to 0.8 ml DHBA. DHBA fluorescence
subtracted: X - D - V .
328 nm excitation wavelength:
450 nm emission wavelength . . . . . . . . . . . . . . 59

Fluorescence intensity (65.535 . 100 pA PlT

anode current) vs. total absorbance of

20. ul quinine sulfate in the presence

of 0.0 to 80.0 nl sodium fluorescein:

(A) Uncorrected; (V) Primary and secondary
absorbance-corrected. 365 nm excitation

wavelength: 438 nm emission wavelength . . . . . . . . 61

ix

Figure Page

4-1 Differential input sample-and-hold . . . . . . . . . . 72
4-2 lultiplexer - S/H output . . . . . . . . . . . . . . . 73
4-3 S/H gating and lUX channel address latch

decoder and indicator . . . . . . . . . . . . . . . . . 74
4-4 General purpose 12-bit ADC board . . . . . . . . . . . 77
4-5 A photdiode detector . . . . . . . . . . . . . . . . . 80
5-1 Geometrical model of point-source . . . . . . . . . . . 91

CHAPTER 1

INTRODUCTION

lolecular luminescence methods have gained a great deal of
popularity within the scientific community due to their versatility
and broad applicability. Luminescence methods are useful because of
their high sensitivity and excellent selectivity. Luminescence is
often more selective than absorption because two wavelength selectors.
emission and excitation are used. In addition, luminescence
polarization and luminescence lifetimes can be used to enhance
selectivity. However, there are large numbers of instrumental
difficulties as well as physical and chemical interferences. and these
problems can prevent many potential applications from being realized.
One such problem is the attenuation of luminescence signals due to the
absorption of light by the analyte itself. or by other components in
the matrix. often knovn as the inner-filter effect.

lethods of describing and correcting for inner-filter effects are
the major concern of this thesis. The ultimate objective of this
field of research is the development of techniques and instrumentation
with all of the capabilities, sensitivity, and ease of use of modern
commercial automated instruments. but with the added feature of
automated self-detection and correction of inner-filter effects. Iith
this should come new analytical procedures made possible by this

1

improved methodology.

The problem of secondary absorption (absorption of the emitted
radiation) in chemiluminescent solutions is discussed in the second
chapter of this thesis. A mathematical model of secondary absorption
effects is developed. and a correction factor is derived. The
instrument and chemistry developed for implementation and testing of
the model are evaluated.

The third chapter is concerned with some aspects of the primary
absorption effect (absorption of the exciting radiation) with front
surface fluorescence techniques. A difficult requirement in
determining inner-filter effect correction factors is describing the
fluorescing solution as a volume source. rather than as a point
source. The volume source becomes more amenable to mathematical
modeling if the observed light can be collimated. Unfortunately. this
creates a severe limitation in light throughput. The instrument
developed in this work uses a bifurcated fiber-optic as an
excitation-emission window with a transmittance flow cell. Absorbance
and fluorescence are simultaneously available. This information is
used with a correction model that mathematically describes this
fiber-optic/cell configuration. Because the collimation requirement
of earlier correction methods has been eliminated. the acceptance
function of this configuration is improved over earlier instruments.
vastly improved sensitivities are predicted. The theory for
development of the model and the results of testing and evaluation are
presented.

Instrumentation and software are discussed in the fourth chapter.

A description of the hierarchical computing system and microcomputer

3
used for these experiments is presented. A counting/timing module. a
sample-and-hold/multiplexer module. and an analog-to-digital converter
module were designed by the author. A photodiode detector system is
also presented here. Two microcomputing languages. SLOPS and FORTH.
are compared and illustrated as languages for these experiments.

A volume source model for luminescence methods is developed and
discussed in Chapter V. This model permits assumptions in the
absorbance-corrected chemiluminescence technique to be tested and
proven. The model is also required for the development of the
absorbance-corrected fluorimeter, as described in Chapter III.
lodeling and fitting of the fluorescent solution as a volume rather
than a point source is a key requirement of the efficiently coupled
fiber-Optic fluorescence cell.

The final chapter summarizes the current work and projects 'to
future work. lany improvements at both theoretical and instrumental
levels are possible. Hopefully. absorbance-corrected luminescence
techniques will one day be readily available tools for routine

application in many areas of chemical analysis.

CHAPTER II

ABSORBANCE-CORRECTED EHHHHLUIINESCENCE lEASURElENTS
WITH A DUAL-PATHLENGTH SPECTROIETER

A. Introduction and Higtorical

1. Principles 2; Chemiluminesgengg

Chemiluminescence (CL) is the emission of light resulting from a
chemical reaction. In general. chemiluminescence is a result of the
promotion of a molecule into the excited state during a chemical
reaction followed by subsequent decay to the ground state accompanied
by light emission. Chemiluminescence is most commonly observed in
biological systems such as fireflies and luminescent bacteria and
algae found in the sea. This special form of chemiluminescence is
appropriately called bioluminescence and is frequently an
enzyme-catalyzed process.

Early studies of bioluminescence have led to greater interest in
chemiluminescence in general among biologists. biochemists. and
chemists. Bioluminescence was first used analytically in 1889 by
Beijerink (1) for the detection of low-level oxygen with luminescent
bacteria. Since that time a few chemiluminescence assays have become
established as sensitive and selective methods for the determination
of trace metals. hydrogen peroxide. ATP. and NADH. Analytical
chemiluminescence methods are useful because they provide excellent
sensitivity. adequate selectivity. and linear response. The inherent

4

5
sensitivity of luminescence is complemented by the added simplicity
and advantage of not requiring a source for optical excitation. The
added sensitivity of enzyme multiplication is also possible with many
bioluminescence methods. Bioluminescence also has the advantage of
the specificity associated with many enzyme-catalyzed reactions. lost
commonly used chemiluminescence reactions have half-lives of less than
10 seconds. and this allows rapid analysis times and high throughput.
Iide linear response has been demonstrated for a variety of
chemiluminescence methods. Low reagent and sample consumption is also
a result of the high detection sensitivity. Complete reviews of

chemiluminescence are available in the literature (2-7).

2. Prgblgg! git; Chemilumineggepgg leghgge

Among hindrances to further analytical applications of
chemiluminescence techniques is the non-specificity of many
chemiluminescent reactions. For this reason. many methods require a
prior separation of analytes before chemiluminescence detection.
Another problem is the interference attributable to inner-filter
effects (attenuation of emitted light intensity due to absorbance by
the solution matrix). lhen chemiluminescence is detected. any
absorbance of chemiluminescent radiation within the solution. or any
variation in absorbed chemiluminescence from solution to solution
results in a difference between the amount of light actually detected.
and the amount of light anticipated or expected. If the amount of
light detected is to be correlated with the concentration of an

analyte or the spectral distribution and radiant power of emission.

6

these values will be in error. Several such problems with
inner-filter effects have been described.

lampler. et. al. (8) demonstrated a problem with the inner-filter
effect of riboflavin in a luciferin-luciferase reaction coupled to
peroxide-forming reactions of clinical and biochemical interest. Van
Dyke. et. al. (9) mention the importance of eliminating erythrocyte
contamination to avoid inner-filter effects in a blood screening
system using luminol chemiluminescence. In the chemiluminescent
determination of humic acid by reaction with permanganate. larino and
Ingle (10) have described signal attenuation due to the appreciable
absorbance of humic acid in the higher concentration range.

Inner-filter effect interferences with chemiluminescence
spectroscopy are analogous to secondary absorption interferences in
fluorescence spectroscopy (absorption of fluorescence emission).
lethods of correction of secondary absorption in fluorescence
spectroscopy have been reviewed elsewhere (11-13). These descriptions
of fluorescence secondary-absorption corrections are complicated by
and dependent on the primary absorbance of excitation radiation and
the convolution of excitation and emission geometries. lost are
concerned with the multiple reabsorption and reemission effects of a
single fluorophore. Experimental evidence. thoroughness. and clarity
are often lacking: these methods have not been found to be directly
useful for chemiluminescence inner-filter effect correction.

Although no remedy for inner-filter effect interferences with
chemiluminescence methods has been reported. an expression describing
chemiluminescence in terms of absorbance and pathlength was given by

Stieg and Nieman and applied as a criterion for the efficient design

of chemiluminescence flow cells (14). Experimental results were
superimposed on a theoretical chemiluminescence response curve for
various pathlengths. The gallic acid chemiluminescent reaction system
was observed at 643 nm at concentrations where the absorbance per unit
pathlength was 1.6. Requirements and limitations of the model's
geometry were not stressed. nor were the effects of reemission or
deviations from monochromaticity discussed. The expression of Stieg
and Nieman has not been applied for modeling a correction method for
inner-filter effects.

The following pages describe a procedure for correcting
chemiluminescence measurements that is based on a determination of the
ratio of chemiluminescence from two different cell pathlengths. A
microprocessor-controlled instrument with a unique dual-path cell is
described. The instrument automatically produces absorption-corrected
chemiluminescence intensities at a single wavelength or
absorption-corrected chemiluminescence spectra. Results of studies
with a luminol chemiluminescent reaction are presented that
demonstrate the effectiveness of the correction procedures and the
characteristics of the instrument.

The basic premise behind this method of absorption-corrected
chemiluminescence as well as that of corrected fluorescence by the
method of cell shift and other methods. is that the. emission
attenuation is a well behaved (Beer's law) function of pathlength. and
that the intensity vs. pathlength information is useful for
identifying the corrected intensity. Interestingly. this approach is
quite analogous to the method of "Recognition by Sight" as discussed

in the classic Flatleng by Edwin A. Abbott (15). In Flatland. a two

8
dimensional world lying in a single plane. all persons have the form
of an equilateral polygon and class distinction is determined by the
number of sides. A Flatlander (having been educated in the Science
and Art of Sight Recognition) is able to distinguish the shape and
thus. most importantly. the rank of another approaching Flatlander
only because of the presence of Fog. Iithont Fog. all polygons appear
as lines (recall that the observer is in the same plane). However. in
the presence of Fog a line grows dimmer as it recedes. and by careful
observation of the comparative dimness or clearness along the line the
Flatlander can infer the acuteness of angles and thus the numbers of
sides of an approaching person. The reader may find this reference

useful for an intuitive insight into absorbance-corrected luminescence

techniques.
A. Theogy

1. Attenuation by Inner-Filter Effect

Consider a thin slice of solution of infinitesimal thickness dx.
parallel to the observation window. The slice radiates monochromatic
light of unattenuated radiant power dPo. normal to the plane of the
window. as shown in Figure 1-1. If an absorber is present between the
slice and the window. the radiant power observed at the window. dP .

X

is described by Beer's law as
up, - dPo(1o“°‘) - dPo(o'2°3°3'°‘) (2-1)

where e is the molar absorptivity of the absorber. c is its

 

dP,‘ dfi’,
W

 

 

 

t

Figure 2-1. Schematic cell model.

 

10
concentration. and x is the distance of the slice from the cell

window. The equation
dPoldx I JodndA (2-2)

describes the unattenuated power per unit thickness as the product of

the unattenuated radiant emissivity Jo. the solid angle of observation

*‘m' ..-N..

d0. and the projected area dA of the radiating slice. By combining
Equations 2-1 and 2-2. the observed radiant power can be expressed as

a function of the slice thickness dx:
up: - (JodndA)(o’2°3°3“‘)dx (2-3)

The total power observed Pb . for all slices within a cell of
pathlength b is determined by integrating Equation 2-3 over all

thicknesses from zero to b:

pb - (JodudA)I:(o'2-3°3‘°‘)dx (2-4)
Equation 2-4 reduces to

Pb - (tonnes)(1-o'2-303‘b°)/(2.303.c> (2-5)

(This is similar to Equation 10 of Stieg and Nieman (14) where r I 0).
Note that in the limit of small so. the observed radiant power is
equal to the "absorption-free" radiant power P0. and Equation 2-4

becomes
lim -
sc90 Pb ' Po ‘ ”0‘94“b (2 6)

Since many chemiluminescent reactions result in chemically excited

fluorescence of product molecules. the additional requirement that

11
light absorbed must not be reemitted (no overlap of excitation and
emission spectra) must be included. Scattered light. refractive index

effects. and reflections within the cell are assumed to be negligible.

2. ‘12; Duel-Path lethod g; ngrectign

Herein is proposed a method of chemiluminescence detection that
simultaneously and conveniently determines the value of so (and thus
the absorbance A I ebc) and the corrected radiant power P0. The
method is based on the comparison of luminescence measurements
simultaneously taken from either length of a cell having
pathlengths b' and b. where b' ( b. The ratio R of these intensities

can be found from Equation 2-5. yielding
a . Pb./Pb - (1—o’2°3°3'b'°)Ici-o'2-303‘b°> (2-7)

With the known pathlength values b' and b . and the measured values of
Pb' and Pb . the value of so can be determined from Equation 2-7 with
an iterative approach. The corrected radiant power P0 is then
deteruined by substituting so and the experimentally observed P0 into

Equation 2-8. which follows from Equations 2-5 and 2-6:

Po - 2.3ospbsbc/(1-o'2-303'b°) (2-8)

 

 

 

 

 

 

 

 

12

1. Apparatus

To investigate this model. a cell and collimator were designed to
allow the measurement of the ratios of chemiluminescence from an
homogeneous solution at 2 pathlengths as shown in Figure 2-2. The
cell was made with the arbitrarily chosen nominal pathlengths of
1.00 and 0.50 cm. Black Delrin (16) was chosen as the construction
material for both the cell and collimator because of its chemical
inertness. machinability. and relatively low reflectivity. The cell
was formed by machining the desired openings from a solid Delrin
block. 45 mm wide at the front. 25 mm high. and 30 mm deep. The front
opening was 15 mm high by 26 mm wide by 5 mm and 10 mm deep for the
short and long pathlengths.respectively. cut 3 mm from the bottom of
the block. as shown to scale in Figure 2-2. in a cut-away view from
the top. A quartz microscope slide. 17 mm high by 30 mm wide. was cut
to shape. press-fit into place. and sealed with black silicone rubber
sealant (17) to form the cell window. An 8 mm diameter hole and a
3 mm diameter hole were drilled above the long and short sides of the
cell. respectively. to allow solutions to be placed into or removed
from the cell. The collimator was made by drilling a 2 mm diameter
hole into a 50 mm long black Delrin rod. This diameter-to-length
ratio yields an acceptance angle of approximately 2°. which gives a

reasonable degree of collimation as required by the model. at the

expense of some loss in radiant energy throughput.

 

 

13

h.<‘.u

use
3852505.:
2

.uou-luuuoo vme uneo AU moemlueau no Innueum Deu> ooh

.7” 8...:

 

 

 

mOF<ZfijOU

.JLJJ

 

 

 

JJMU

14

The cell and collimator were mounted in a programmable sample
chamber (18) with the collimator 5 mm from the cell window and
centered in the exit port. The sample chamber was integrated into the
optical system as shown in Figure 2-3. A 2 mm exit slit and the 2 mm
diameter of the collimator at the entrance to the monochromator (l9)
resulted in a 16 nm spectral bandpass. All measurements were made at
425 nm unless otherwise noted. The Hamamatsu R928 photomultiplier
tube (PlT) was cooled to reduce the dark current shot noise
contrubution (12.20). Dark current was reduced more than 2 orders of
magnitude from the room temperature value to a point where the dark
current and dark current shot noise were lower than the amplifier
noise and drift. Elimination of the dark current shot noise
contribution resulted in enhancement of the signal-to-noise ratio by a
factor of 5 to 10. depending upon the light level. The PlT voltage
was -560 V; the output current was converted to voltage with a current
amplifier (21) set at 10‘. V/A gain and 10 ms rise time (10$-90$).
Although output voltages were always well below 1 V. neither higher
amplifier gain nor higher electron multiplier voltages were chosen.
This allowed sufficient amplifier "headspace" for the fast-rising
current input pulses. Higher amplification would have resulted in
frequent saturation of the high gain input amplifier. resulting in
non-linear output response.

Instrument control and data acquisition were directed by an
8085-based microcomputer developed locally (22). The cell was allowed
to oscillate freely from side-to-side at approximately 2 Hz. Cell
position detection signals available from the cell compartment were

sensed and used to synchronously gate chemiluminescence measurements.

 

15

 

 

l .m] [Tame]
l I

I
LSI "/23 I-SEBN-SL—I 50‘“

 

 

 

 

 

 

 

 

 

iCELL POSITION-
. SENSE/CONTROL

 

 

l2-llt AID
51H

 

    

MONOCHROMATOR

 

 

  

 

 

 

PROGRAMMABLE COOLED
CELL Pm
Posmonen

Figure 2-3. Schematic diagram of instrument.

16
At 2 samples per oscillation and 2 oscillations per second. the
sampling rate resulted in an overall duty cycle of 47! for data
collection.

Absorbance measurements of the chemiluminescent solutions were

A... -_

also made with an absorption spectrophotometer. For purposes of

comparison. the single-beam spectrophotometer had an optical geometry"

similar in design to that of the chemiluminescence detection
instrument. The components were: a tungsten/deuterium. source. a
dichroic bandpass filter with a wavelength maximum equal to that of

the monochromator wavelength setting. a transmission sample cell of

1 cm pathlength. an H-10 monochromator with 2 mm slits (16 nm spectral

bandpass). an RCA 1P28 PlT. a current-to-voltage converter. and a
digital multimeter.

Since. by their nature. many chemiluminescent reaction mixtures
contain or produce fluorophores. the bandpass filter was inserted
between the source and sample to reduce possible sample fluorescence
interference caused by broadband source excitation. Additionally.
when measuring the absorbance at the wavelength of peak

chemiluminescence emission. the bandpass filter brings the source into

-..‘-I

better conformity with the spectral distribution of the._

chemiluminescence emission "source" used to determine the absorbance
from the chemiluminescence ratio measurements. Transmittance
measurements were determined as T I (I-B)/(Io-D). where I0 is the
blank signal. I is the sample signal. D is the dark signal. and B is
the background signal. The dark signal. D. is the sum of the dark
current and amplifier offsets. and is determined by blocking the beam

between the source and the sample with the blank in place. The

17
background. R. is the sum of the dark signal and the chemiluminescence
background. and is determined as D is. but with the sample in place.
This method compensates for any interference due to chemiluminescence
although the difference between R and D was only significant at

absorbance values of 1 or more.

2.239;;

Reagents were used without further purification. All solutions
were prepared with deionized. distilled water. and were stored in
brown glass bottles to prevent photodecomposition. The pH 9.5 buffer
was 0.1 l in sodium borate and contained 0.2 mL/L of 305 Brij 35
surfactant (23). The Brij 35 wetting agent keeps bubbles which form
during the course of the reaction from clinging to the cell walls. A
0.001 l solution of luminol (3-aminophthalhydrazide) (24) was prepared
in 0.002 l NaOH. A 0.1 ml ln(II) solution was prepared from lnClz°4H30
(23) with dilute HCl (pH 3.0). The "activator" solution was an
aqueous mixture containing 0.01 l 1.10-phenanthroline (25) with 0.1 l
sodium citrate (25). Solutions that were 0.8 ml and 3.9 ml in picrate
were prepared with 85% picric acid (26) in buffer with sufficient
1.0 l NaOH to restore the buffer's original pH. A 750 pl solution of
ferroin (1.10-phenanthroline ferrous perchlorate) (27) was prepared in

buffer.

3. Propedure

A working solution was prepared with a 2:2:1 volume ratio of the

luminol. activator. and manganate standard solutions. respectively.

18
Before each measurement the cell was thoroughly rinsed with deionized.
distilled water and aspirated to dryness. The buffer and
picrate-buffer or ferroin-buffer standard solutions were then added to
a combined total of 1.00 mL followed by the addition of a 0.50 mL
aliquot of the working solution. A dark current measurement was made.
and the reaction was initiated by the addition of a 0.20 mL aliquot of
0.1 l 3:0: into the cell. A 90 second delay followed to ensure
complete mixing and to allow the reaction to progress beyond the lag
period. Data were typically taken during a 3 to 8 minute period after
which the reaction decayed to a point where the signal-to-noise ratio
(SNR) became unfavorable. The logged data were transferred via a
serial line to a minicomputer (28) for data reduction. The ratio.
absorbance. and uncorrected and corrected intensities were reported as

time-averaged values.

D. Regults ppd Disgupgion

1. Choice pf Chemical sttem

To avoid reemission errors a luminol reaction was chosen because
the emitting product. 3-aminophthalate. exhibits low
excitation-emission spectral overlap (29). It was also desirable for
the selected luminol reaction matrix to exhibit a low absorption
profile across the wavelength region of interest. The specific
buffer. oxidant. and activators used provided good intensity while
allowing an unusually long half-life for the luminol reaction (30).
The exact reaction mechanism for this system is unknown. however.

general equations are given below (31-33):

19

u + 11,0, -9 l-H,0,

u—n,o, + on" —> 11-0,]! + 11,0

u—o,n + luminol + on" —) [l-0,-luminol] + 11,0
[l—0,-luminol] -> N, + [3-aminophthalate].
[3-aminophthalate]. -9 3-aminophthalate + hv. ln‘x = 425 nm

The long emission decay time permitted the time period in which a
single intensity ratio was measured (1/2 second) to be negligibly
small relative to the rate of change of the observed intensity. The
SNR was also enhanced through the use of signal averaging made
possible by the long reaction time. A disadvantage of the chemical
system is that during the course of the reaction a chromophore of
undetermined structure is formed which has an absorbance peak that
tails into the region of interest. At the wavelength of peak emission
(425 nm). this leads to dynamically changing absorbance values that
rise rapidly from 0.00 at the beginning of the reaction and approach
0.05 toward the end of the reaction. as shown in Figure 2-4.
Fortunately. these absorbance changes were near or below the limits of
detectability. and the dynamic aspect of this problem was therefore
inconsequential.

For purposes of testing the technique. chromophoric reagents were
added to the luminol solutions to vary the matrix absorbance in the

region of luminol emission. Picrate and ferroin were chosen for their

20

.oluu mama Nausea nouuoeou mooning no coneauoune unannouoou .vuu one-«m

653 m2;

 

 

mm on mN 0m 0. 0. 0 My
. . . L w i . 00.0
.1 _0.0
..N0.0
500
Ill )4 l 4 till .500

 

(““9317) BONVSHOSBV

21
relative chemical compatability with the reaction matrix and for the
overlap of their absorption spectra with the luminol emission
spectrum. as shown in Figure 2-5. lany other chemicals were tested.
however. none were found to have appropriate spectral and solubility
qualities at high pH. combined with inertness to hydrogen peroxide
oxidation.

In addition. other chemiluminescent chemical systems were
investigated. The peroxyoxalate reaction (34.35) resulting in a
chemically excited 9.10-disubstituted anthracene derivative was
initially used. This reaction is extremely bright; however. it is not
readily adaptable to aqueous solutions. making it inconvenient to use.
Also. many of the efficient fluorophores used in the peroxyoxalate

reaction have strongly overlapping excitation and emission spectra.

2. Determination 21 Sppplp Apsorbance

Experiments were performed to determine the extent of agreement
between absorbance values measured by a conventional transmittance
technique and those derived from the chemiluminescence ratio
instrument. This information also provided calibration curves to
evaluate the adherence to Beer's law by both techniques. Picrate and
ferroin were added in known amounts to the reaction mixture. After
chemiluminescence ratio measurements the solutions were promptly
transferred to the single-beam spectrophotometer where the
transmittance was measured. Both instruments were set at 425 um. the
wavelength of peak luminol emission. These experiments indicated some

initial disagreement between the two sets of absorbance values

22

Figure 2-5. Spectral distributions: luminol emission (arbitary units.
solid line. right ordinate): picrate absorbance (dashed
line. left ordinate); ferroin absorbance (dotted line.
left ordinate).

23

 

 

 

 

CL. HQTElsngY (All)
a
C. m
Jr— P\
U7
0 . J m
5' (‘4
Ln.
0... z
3 3
‘3. 5 j“
2.. m
.. + m
I <1"
4 0
.1
O
z .. .’
5 2"- m
J ’ ’ e. q- Cu]
v ' g ‘1.
, 'e.
I e
I a.
I J»
I
J U1
I "' l’\
.’ v?
m
'2
I
\ 5
\ "‘ 1h
fle
\
‘ ~
‘ . '° in
1 ~ f- t ; cu
I”?
‘3 (D (3 D
g) 55 e)

)JUHidHOSBV HVWCW

WAVELENGTH (nm)

Figure 2-5

24

determined by the two instruments. This was attributed to the
sensitivity of the ratio model to small errors in the pathlength
values taken for calculation of the absorbance. and an inability to
directly measure these distances accurately using conventional tools.
Useful pathlength values were subsequently calculated with a
non-linear curve-fitting routine. KINFIT4 (36.37). Values of R were
regressed on transmittance-derived sc values taken from the linear
portions of the concentration curves. in a weighted fit of Equation
2-7.

Cell pathlength values mechanically measured before mounting the
cell window and values obtained by curvefitting are given in
Table 2-1. Good agreement between fit values for the two data sets is
observed. Fit values are somewhat larger than values predicted from
the pre-assembly figures. by factors of 5.2% and 3.8i for the short
and long paths. respectively. This results in a ratio
of b' : b that is 1.0% larger than for the measured values.

Ve suggest two factors which may contribute to increases in the
predicted pathlengths. An actual increase in the cell pathlengths may
be the result of a displacement of the cell window from its seat.
caused by a thin layer of silicone sealant and elastic pressures
produced in press-fitting the window into the Delrin. Alternately.
the longer pathlength obtained from curvefitting may be a compensation
for shortcomings in the performance of the optical system. i.e.. the
collimator and monochromator. Tb test the collimation assumption. the
aperture function of the collimator was calculated in terms of its
mathematical counterpart - the cylinder. A narrow cylinder appears to

be an effective collimator in that it passes a relatively constant

25

TABLE 2-1. Comparison of pathlength values‘

M s n; L'. 95 n. 9.! 3" .r.°
pro-assembly 0.507 r 0.005 1.017 r 0.005
picrate data 11 0.536 r 0.008 1.06 t 0.02 0.992 0.9972
ferroin data 10 0.531 r 0.008 1.05 r 0.02 0.986 0.9988
combined data 21 0.533 t 0.005 1.06 t 0.01 0.989 0.9982

‘Uncertainties are r 1 standard deviation.
blultiple linear correlation coefficient from KINFIT4.
cLinear correlation coefficient resulting from linear weighted

regression. with absorbance from transmittance as the independent
variable. and absorbance from CL ratio as the dependent variable.

26

amount of light from each emitting plane normal to the cylinder axis.
irrespective of depth into the cell. According to numerical analysis
of the cylinder aperture function. the amount of light from each
emitting plane normal to the cylinder axis passing through the
collimator increases as a function of depth into the cell. For a
cylinder of dimensions equal to those of the collimator used. the
relative change from the front to the back of a 1.0 cm cell is found
to be approximately +4.0§. The change as a function of depth into the
cell is approximately linear. The collimator was also tested
experimentally by approximating an homogeneous emission plane with a
selection of diffusing plates backlighted by various sources. Results
varied. indicating that the collimation was effective to within
approximately r.2$. However. it appears possible that systematic
errors of unknown magnitude exist in this method of collimation
evaluation.

A new model was derived for the ratio of intensities by assuming
a linear change with depth of the amount of light observed through the
collimator. This model was then used for some KINFIT4 estimates.
Iith 4.0 ilcm as the change in the aperture function. new values of b
I 4.91 t 0.11 mm and b' I 9.33 t 0.24 mm were obtained. By allowing
KINFTT4 to assess the collimator deviation. the values of -8.9 $lcm
deviation in the aperture function with b' I 5.11 r 0.13 mm and b I
10.21 t 0.25 mm were obtained. Forcing the collimator deviation to
be positive. as indicated by the cylinder modeling. results in a
collfluator aperture function deviation value of +2.8 ilcm. and
pathlength values of b' I 5.26 r 0.06 mm and b I

10.26 t 0.15 mm. These data clearly indicate that small deviations

27
from ideality in the collimator function can result in substantial
changes in the pathlength derived by curve fitting. However. within
the experimental error of this study. curve fitting does provide a
means of compensation for these unknown deviations.

The cylinder model does not account for the problem of slight
internal reflections within the collimator. Reflections and
scattering can and do occur within the rest of the optical system as
well. The overall aperture function is also complicated by the
refraction of the chemiluminescence rays as they pass through the
solution/cell window/air interfaces. This effect is a result of the
differences in the refractive indices of the three media and is of
undetermined significance. Although it is difficult to predict the
combined effects and importance of these optical non-idealities. it is
possible that they cause a slight but perceptible change in R values.
resulting in modified pathlength values as derived by curve fitting.

Absorbance values were then calculated using the
chemiluminescence ratio and the transmittance data. The resulting
concentration plots are shown in Figure 2-6. Non-zero intercepts are
a result of the background absorbance produced during the reaction. as
discussed earlier. Also noteworthy are the negative Beer's law
deviations observed in the transmittance data at higher absorbance
values. These deviations are a predictable result of the 16 nm
spectral bandpass needed to ensure a useful SNR. Good correlation
occurs between the chemiluminescence ratio and transmittance data for
moderate absorbance values up to approximately 0.75. as shown in
Figure 2-7. However. in the region of high absorbance. a positive

deviation of the chemiluminescence ratio data with respect to the

28

ABSORBANCE

0.50

 

 

 

0.00 i i ' t } : § : IL , {
O 50 100 150 200 .250
PICRATE (PM)

2.00 --

1.50 -~ I /

1.00

00

ABSORBANCE

 

 

ooo :'r-:+:':.:+I1
O 50 100 150 200 250 300

FERROlN (PM)

 

Figure 2-6. Absorbance as a function of chromophore concentration
from (I) transmittance measurements and (0) CL ratio
measurements.

£2
.—
<
0:
:
c>
C!
L‘—
[J
o
E
a:
0:
c>
(f)
9
00
Figure 2-7.

I

29

 

 

1 . 1 1 1 . 1
I T r' ' I ' I
co 05 L0 L5 20

ABSORBANCE FROM TRANSMHTANCE

Correlation of absorbance values determined by
transmittance and ratio measurements: (0) in presence of

ferroin; (X) in presence of picrate. Line has ideal
slope of 1.

30
transmittance data is observed. 'e have observed negative Beer's law
deviations in the case of known reemission errors using other
chemistries and in the case of a grossly non-monochromatic stray light
situation (zero-order monochromator setting). Because of these
experiences and the studies of the collimator deviations as previously
discussed. the collimator was again considered as a possible source of
this error. Calculations of absorbance were made using the
chemiluminescence ratio equations modified to account for a small
increase in light collection with depth. as observed for ideal
cylinders. High absorbance values were recalculated with each set of
the collimator deviation. b' and b values mentioned earlier. New
values were all approximately the same or slightly higher than older
values. It appears that cylinder modeling cannot explain the positive
deviations found in the chemiluminescence absorbance calculations.
Apparently the non-ideal behavior is a different artifact of the
cptical system. which becomes more pronounced in the high absorbance

region.

3. Effectivenegs g; Correction Procedure

The conclusive test of the instrument was an evaluation of the
effectiveness of the correction algorithm. Picrate and ferroin again
served as chromophores. In Figures 2-8(a) and 2-8(b). attenuated and
corrected intensities observed at 425 nm from the long-path side of
the cell are plotted as a function of increasing picrate and ferroin
concentrations. respectively. The effectiveness of the instrument in

recovering effects otherwise masked by absorbance is demonstrated by

31

 

 

Figure 2-8. CL intensity as a function of chromophore
concentration. (I) Inner-filter effect attenuated. (0)
Absorption-corrected. Absorbance axis is an approximate
scale for reference purposes.

 

i W
Imam

32

ABSORBANCE
1.0 1.5 21)

JL-

l ' I

 

 

 

 

 

 

 

 

 

 

FERROiN (Fm)

Figure 2-8

<E
a
>—
t
m
23
m
F—
E
4
Q)
o +: ;:;.;

0 50 100 150 200 250
PICRATEQLM)
ABSORBANCE

0.0 0.5 1.0 1.5 2.0

30 { : } : § :
<E
a
>—
E
m
2:
m
}_.
Z
4
05““8
01% :%1.:.1:::
C) 50 100 150 200 250 300

-I!

 

 

 

 

 

 

33

the difference between the corrected and uncorrected curves. 'ithout
correction. ferroin would appear to be a mild chemiluminescence
activator in concentrations up to 25 ul. but a strong quencher at
concentrations above 50 pl. Such results are incongruous and
difficult -to chemically justify. however. after correction for
absorbance attenuation ferroin is demonstrated to have an increasing
activation effect with increasing concentration. Picrate. too. would
appear to be a strong quencher. but absorbance-correction shows this
effect to be much weaker than one might originally conclude. However.
the anomalously high absorbance values found above approximately 0.75
(Figures 2-6 and 2-7) translate into slightly over-corrected and
possibly misleading intensity values. Experiments at 475 nm with
concentrations up to 2 ml. but absorption values below 0.5. have
confirmed the increasing quenching of picrate with increasing
concentration. while pointing out the over-correction problem at
425 nm for picrate concentrations greater than 75 pl. as can be seen
in Figure 2-8(a). Further experiments at other wavelengths have shown
the same concentration dependence of the quenching and activation by
picrate and ferroin. respectively. and validate the wavelength
independent nature of the correction method.

The mechanisms by which picrate and ferroin affect this reaction
have not been deduced. It is known that picric acid forms
monopicrates with 1.10-phenanthroline and most of its known
derivatives. Since the ln(II)-phenanthroline complex is important in
the catalysis of this reaction. addition of picrate would be expected
to quench or otherwise perturb the catalytic mechanism. As regards

the ferroin perchlorate activation. the chemical system was found to

34

be unaffected by the presence of the perchlorate anion alone. It was
also found that ferroin had no enhancement effect in the absence of
ln(II). This suggests that the enhancement effect of the ferroin
addition is due in some way to an interaction of the ln(II) and Fe(II)
phenanthroline complexes.

A corrected spectrum of the luminol emission in the presence of
90 ul ferroin was reproduced by measuring the chemiluminescence of
solutions with and without ferroin at one wavelength. and similarly
repeating this procedure at other wavelengths. A reference spectrum
of the luminol emission. corrected for the small matrix-generated
absorbance. is shown in Figure 2-9 with the uncorrected and corrected
spectra of luminol emission in the presence of ferroin. Also shown is
the absorption spectrum of ferroin and the matrix. generated by the
correction algorithm from the same data. The uncorrected emission
spectrum is misshapen and diminished in intensity by a non-constant.
wavelength-dependent factor in comparison to the reference spectrum.
However. the absorption-corrected spectrum has a spectral distribution
identical to that of the reference spectrum. and is. by a constant

factor. more intense due to activation or catalysis of the reaction.

E. Conclusions

Although the current theory requires collimation of the observed
light. this constraint was initially chosen for mathematical and
conceptual simplicity. A method which reduces or eliminates the

collimation requirement. allowing more efficient light detection.

could be similarly effective. unfortunately. the mathematical

 

 

 

35

Figure 2-9. Luminol emission spectra (arbitrary units. left ordinate)
and ferroin absorbance spectrum (right ordinate). (0)
Luminol emission reference spectrum. (1) Emission
spectrum attenuated by inner-filter effect due to ferroin
absorbance. (#) Absorption-corrected emission
spectrum. (A) Ferroin absorption spectrum from R values.

36

ABSORBANCE

n80

0%
O

«EC

‘9
C)

mAu

O.F

m.e

a." .uauom

 

 

thv :ooznnn><;

flax“ 3mm aziw nxv Afixy mmv fizgv nxm. can

_ _ _ _ _ _ _ O

_ _ _ _ . _ _

x;
If [I coom
II II Ocov
11 11 coco
II II coom
LI 1: ocoor

 

 

 

OOONP

ad
I}

M”

(nv) AUSNEL

37
solution for such a model is much more complex. A procedure for the
evaluation of detector geometries has been suggested as a basis for
extension to include self-absorption effects (38).

Further instrumental work in this area should be directed toward
increasing the SN! by increasing the optical speed. devising methods
for faster measurements. and studying the effects of various
bandpasses and Beer's law deviation conditions. It would also be
useful to incorporate a chemical system which has no background
absorbance. and to find and apply chromophores which do not otherwise

quench. activate. or interact with the chemiluminescence.

CHAPTER.III

A FIBER-OPTIC INTERFACED FLO! CELL FOR
SIlULTANEOUS ABSORBANCE AND PRIlAR!
INNERPFILIER EFFECT CORRECTED FRONT

SURFACE FLUORESCENCE lEASURElENTS

A. Introduction and Historicgl

Fluorescence techniques have become popular in analytical
chemistry principally because of their sensitivity. selectivity. and
wide dynamic range. These features depend on the absorbance and
ensuing emission of photons at characteristic wavelengths. and with
characteristic probabilities. i.e. excitation and emission spectral
distributions. molar absorptivity. and quantum efficiency are all
constant. As the concentration of a given analyte changes. the
aforementioned characteristics should be experimentally apparent. as
long as the solution absorbance at the wavelength of interest remains
diminishingly small. However. if the solution absorbance at the
excitation or emission wavelength does become appreciable. the
excitation or detection probabilities is correspondingly diminished.
which results in errors. Such errors are referred to as inner-filter
effects. The inner-filter effect at excitation wavelengths is due to
'primary' absorption and likewise. 'secondary' absorption occurs at
the emission wavelength.

Although fluorhmetry is useful for the determination of a vast
number of analytes such as blood and urine components of clinical

interest. trace metals and pollutants in natural waters. and

38

39

polyaromatic hydrocarbons in petroleum products. inner—filter effects
often preclude the direct use of fluorescence methods without prior
separation. dilution. or derivatixation steps. Especially
disconcerting is the fact that many inner-filter effect errors go
unnoticed because most fluorescence instruments cannot correct or even
diagnose this error automatically.

various methods have been proposed to reduce the errors caused by
inner-filter effects. but many lack general applicability or suffer
from various limitations. The most popular method of reducing the
error is by sample dilution (39). Absorbance is reduced by dilution.
but fluorescence sensitivity is also reduced. Dilution can also cause
various indeterminate chemical effects. lethods which reduce the
effective pathlength. thus reducing the effective absorbance. include
front surface geometry (40-41). microcell (42). and altered cell
positioning (42). These methods do not eliminate the errors; they
only serve to lessen them to some degree. Appropriate selection of
the excitation and emission wavelengths can be used to reduce or
eliminate inner-filter effects (42). unfortunately. this generally
causes decreased sensitivity. For certain analyses. two-photon
excitation eliminates primary absorbance errors (43). however.
expensive. highrpower lasers are required. because the two-photon
absorption cross-section is extremely small. until such lasers become
routinely available. this solution will not become generally
applicable.

Correction methods for inner-filter effects and associated
problems in fluorimetry have been addressed in a variety of ways for

several years and continue to be of interest to many. Two excellent

40
references to the historical development and current state-of-the-art
in correction methods can be found in the work of Christmann and
Adamsons (11-13. 44-45). The only other recent addition to the
literature of notable interest has been a semi-empirical method of
correcting for primary inner-filter effects in synchronous
fluorescence spectrosopy (46).

The major portion of the work on absorption-corrected
fluorescence has centered on methods for right-angle instruments.
since most commercial instruments use right-angle detection. lost of
the remaining methods are for the determination of quantum
efficiencies. for which a variety of configurations have been
suggested. No attention has been directed to applications involving
flowing streams. and little interest in small volume techniques has
been expressed. This technological gap has led to the development of
the fiber cptic coupled. small volume flow cell for front surface.
primary absorption-corrected fluorescence. as presented here.

litchell. Garden. and Aldous (40) introduced a front-surface
fluorescence instrument which uses a bifurcated fiber-optic to achieve
a front surface fluorescence geometry. Smith. Jackson. and Aldous
(48) went on to develop a flow cell based on this technology. The
"Y"-shaped bifurcated fiber optic is used to direct the excitation
beam into one arm of the "I" to excite fluorophores at the common end.
Emitted light returns from.the common end through the second arm of
the "Y" where it may be detected. In this way. the bifurcated fiber
optic simplifies front surface (zero degree) fluorescence detection.
Errors due to inner-filter effects are reduced with front surface

fluorescence as compared to right angle fluorescence because the

41

effective pathlengths are shorter. Since the common end of a
bifurcated fiber optic may be used as the front cell window (or
immersed in the solution) an efficient. large solid angle is formed
for both excitation and for observation of emission. and the
reflection of excitation light from the cell window back to the
detector. a problem with other 0' techniques. is eliminated. In
addition. the fiber optic input/output characteristics are almost
independent of length. causing the input and output to appear
cptically as though they were spatially the same. i.e.. as though the
fiber optic was vanishingly short. This simplifies optical layout and
alignment. and optimizes the efficiency of light transport.

The balance of this chapter describes the theoretical development
of a correction model for the bifurcated fiber optic cell. It is
shown how the unique characeristics of the fiber optic as a cell
window are incorporated into a volume source model of a cylindrical
flow cell. The flow cell also allows transmittance measurements to be
made of the fluorescence excitation beam. This information is used to

calculate the inner-filter effect correction.

8. Theory

An excellent theoretical basis for quantitative signal vs.
concentration or signal vs. absorbance relationships in fluorimetry is
available in the work of van Slageren. et al. (47). For convenience.
and because some unique differences occur for the cell modeled here. a

complete derivation is given.

42
Consider a fluorescence cell initially having the following

characteristics and conditions. as shown schematically in Figure 3-1:

1. A cylinder of known dimensions defines the volume of the cell.

2. The excitation-emission window is located at one end of the
cylinder. perpendicular to the cylinder axis. It is formed by the
optical face of the common leg of a bifurcated fiber optic
containing the randomized fibers of both fiber bundles.

3. A transparent window at the opposite end of the cell affords exit
for the excitation beam. thus allowing transmittance measurements
to be made.

4. The cell's windows and cylindrical wall are non-reflective.

5. Excitation and observed emission is monochromatic. homogeneous.
and parallel to the cell axis.

6. Scattered light. refractive index effects and reflections within
the cell are assumed to be negligible.

7. Reemission is negligible.

The radiant power Px' at any point in the cell a distance x from

the excitation window is given by Beer's law as
Px I Pollo-{(‘°)exx}] I POIe-2'3°3{('°)ex‘}] (3-1)

where (sc)ox represents a summation of the absorptivities per unit
pathlength at the excitation wavelength. and P0 is the unattenuated
power of the excitation beam at the window. The fluorescent radiant
power generated at this point dFdx’ is a function of the amount of

light absorbed by the fluorophore as given by

.uuoo Iona comececuoe—u
c«uaoluoa«u ecu-cucuma no leumeav heaeluno emu-lemon

 

 

 

 

 

 

 

.Tn 23.: .
2E0. duo 45.3235
20.3.5 . 282;,
\ We moz<E2mz<fi
f N 38;; zo_mm_2u\zo_5_oxm.o
9.50
20.25%

a¢
who

02.28“. mmwmi QMN_EOoz<m

 

 

 

 

 

P
30...“.

44

dFdx I QP‘(sc)pdx I QPole-2'3°3{(‘°)exx}l(sc)Fdx (3-2)

where Q is the fluorophore quantum efficiency. and (sc)F represents a
summation of the absorptivities per unit pathlength of the
fluorophores. The fluorescent radiant power observed from this plane

dF is a function of absorptivity per unit pathlength at the emission

x,

wavelength:

de _ T(x)dFdx[.-2.303((sc)°-x}]

= T(x)P0[e-2’3°3{(‘°)ex + (3°)em}x](sc)pdx (3-3)

where (sc)°. represents a summation of the absorptivities per unit
pathlength at the emission wavelength and T(x) is a transfer function
that incorporates quantum efficiency. wavelength dependence and
efficiencies of optics and detectors. and the fraction of light
transmitted through the fiber optics as a function of depth into the
cell. The total fluorescence observed Fb' is, determined by

integrating Equation 3-3 over the length of the cell. b:

F1. - pound: r<x)o'2-3°3“'°’.x + (“h-ha; <3-4)

If the excitation and emission beams are parallel to the cell
axis. T(x) may in general be considered constant and Equation 3-4 is
solved simply. unfortunately. a parallel beam requirement implies a
very collimated beam and thus low excitation and emission light
throughput efficiency. 0n the other hand. the fiber optic allows
facile excitation/emission coupling with good excitation/emission

efficiency. Since the fiber optic transmittance is dependent upon the

45
angle of incidence of the input light. the transfer function T(x) for
a fiber optic must be explicitly incorporated into Equation 3-4.

Cell modeling incorporating the fiber optic angular transmittance
function based on methods discussed in Chapter V indicates that the
transfer function follows an exponential decay with increasing
distance into the cell. This modeling assumes infinitesimally small
fibers and perfectly random distribution of the excitation and
emission fibers. In practice. the finite size of the individual
fibers and the non-random grouping of the fibers causes the transfer
function to be significantly different than that modeled for short
distances from the fiber optic face. Each individual excitation fiber
emits light into a volume which can be described as an inverted
frustum of a cone. Likewise. each individual emission fiber collects
light from a conically shaped volume. Additionally. as the angle of
incidence of a transmitted ray increases from normal to the fiber
face. the transmittance of the fiber decreases. The net effect
results in discrete volumes being excited near the fiber optic surface
from which emission cannot be observed because the cones of excitation
and the cones of observation do not yet overlap. No emission is
observed from the solution at the surface of the fiber optic. but as
'the distance from the the fiber optic face increases. the overlap of
the excitation and collection cones increases until this effect
becomes negligibly small. Although this effect has not been modeled
on more than an intuitive level. it would seem fair to approximate
such an effect roughly as an exponential increase approaching zero
lepe at infinite distance. By convoluting this exponential increase

with distance with the exponential decay previously modeled. the

 

46
transfer function can be understood to be characterized by an initial
increase from zero followed by an exponential decay from a maximum
value.

Experimental evidence supporting this mathematical/intuitive
description of the transfer function has been provided by Smith.
Jackson. and Aldous (48). These workers measured (Figure 5. reference
48) radiation from the bifurcated fiber-optic excitation arm that
backscattered from a reflective surface and was collected and detected
through the emission arm. The detected scattered light intensity was
found to increase with increasing distance of the reflective surface
with an intensity minimum of about 10$ of maximum at zero scattering
distance. laximum scatter intensity was observed with a scattering
distance of 3 mm. Scattering then descreased exponentially to a low
of 5%»of maximum with the reflective surface 30 mm distant from the
optic. An empirical mathematical description of such a function that

can be exactly integrated within Equation 3-4 is
T(x) - ("o-n: - in) (3-5)

where G is a normalization constant. K describes the exponential
increase. and H describes the exponential decay. Each of these values
can be determined by curve fitting as described later.

By substituting Equation 3-5 into Equation 3-4 and integrating.
the total fluorescence observed can be described in terms of the

respective absorbances within the solution by

Fb I P0(sc )FG[i(1-e-ub) - §(1-e-Nb)] . (3-6)

47

where

l I H + 2.303((sc).x + (sc)..)

N I I + 2.303{(I:c).x + (sc)..)
The limiting form of Equation 3-6 for low absorbance becomes
120 =- pouupcflu-o'm’) -f<1-o"")] (3-7)

where F0 is the unattenuated fluorescence observed.
If a correction factor CF. is defined such that
F0 . CF'Fb
then by substitution of Equations 3-6 and 3-7 into Equation 3-8. and

rearrangement. the correction factor is

 

CF _ iU-e—m’) - (l-e-n’)

fiu-o'“) - *( l-e-Nb)

(3-9)

C. figperimenta;

1. Rgagentg

All chemicals were commercially available reagents and were used
without further purification. unless otherwise noted. All solutions
were prepared with 0.2 N sto‘. A stock solution of 10.0 ml quinine
sulfate (OS) (24) was used to prepare solutions in the range of 0.1 ul
to 0.4 ml. A stock solution of 40.0 ml 2.5-dihydroxybenzoic acid
(DHBA) (24) (recrystallized from water) with 20.0 pl 08 was used to
prepare a series of DHRA solutions ranging from 0.1 to 0.8 ml with a

constant concentration of 20.0 ul OS. A series of 0.1 ml to 0.8 ml

48
DHHA solutions was also prepared. A solution of 0.0500 g/L K,Cr.0, in
0.01 N H1804 was prepared for the determination of the cell pathlength
(49). Nine solutions with sodium fluorescein concentrations ranging
from 0.0 ul to 80 pl were made with a constant 20.0 ul OS
concentration. All solutions were transferred to hard polyethylene
bottles immediately after preparation to prevent adsorption losses.

All solutions were stored in the dark to avoid photodecompostion.

2. Apparatus and Procedure

A schematic diagram of the fluorescence instrument is shown in
Figure 3-2. To construct the sample cell. a cylinder with an inside
diameter of 2.0 mm was formed from a black Delrin rod (16) having a
6.4 mm outer diameter. This was press-fit into a stainless steel
block which was then machined to allow the inlet and outlet of
solution at the ends of the cylinder via 2.0 mm holes perpendicular to
the cell axis. The block was also machined for the mounting of the
bifurcated quartz fiber optic (50) and a 2.0 mm diameter by 50 mm
quartz rod (51). which act as cell windows. A 1.0 mm transmittance
path is formed. creating a 32 microliter cell volume. The entrance
slit of an f/3.5 UV grating monochromator (52) with a 10 nm/mm
reciprocal linear dispersion is positioned to collect light from the
cell through the cell's quartz window/light pipe. This monochromator
is required to reject stray light and fluorescence which interfere
with measurements of the transmittance of the excitation beam. A 20
um bandpass was permitted through the use of 2.0 mm slits. A silicon

photocell housing with integral amplifier is used for light detection.

.uneleuuen« eonocoeuonnu no lea-emu can-lemon .«Im one-«m

—|..I.

 

 

 

 

 

 

 

 

SHONE mg ’33— I

 

\
W‘

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0.1800 ¢E§o§uozox T,
: 324E522» 7 2515...!
. fin _ _
on... pro... C530
.223 W , El _ T, .l mg...
.83.: . 13.»;
r 50.5 .13 _
. guinazc
4(5-m >\-
mum ,
9
4 «02282820: 203285028.
ht 298.5 29258
_ 5
_ Ow. .AI/ \ 8.2835
mum: to...
C33 Beam 7
muzumuuum,

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

musom

 

 

 

 

 

 

Ebuo use 28..“

 

 

 

 

 

 

 

 

   

50
Solutions are continuously pumped or drawn from the cell by means of a
peristaltic pump at the rate of approximately 1 mL/minute.

The excitation source is a high pressure xenon short are lamp
(150 V) driven by a constant-power lamp supply (53). The source
radiation is focussed by an optically fast projection lens onto the
entrance slit of an f/3 monochromator (l6 nm/mm reciprocal linear
dispersion) (19). The excitation arm of the bifurcated fiber-optic is
placed at the exit slit. A single quartz light fiber is placed in
close proximity to the excitation optic to transmit a reference beam
to a silicon photocell detector (54). The reference signal required
to compensate for source drift. The emission arm of the bifurcated
fiber optic is centered in front of the entrance slit of a second f/3
monochromator. identical to the emission monochromator. The FIT
fluorescence detector current output is converted to voltage and
amplified (21) by 10' to 101. VIA. The PlT supply voltage ranged from
-500 to -650 volts.

The voltage outputs from the three detectors are simultaneously
sampled at 5 millisecond intervals by synchronously triggered.
differential input sample-and-hold (S/H) amplifiers. The sampled
signals are then successively transferred via an analog multiplexer to
a 12-bit ADC with a 35 microsecond conversion time. Average signal
values are accumulated from 2000 samples and displayed in tabular form

along with reference-corrected transmittance and fluorescence values.

51

I). MEAN
1. Detgrminagion‘gf Qell Parameters

The cell pathlength was determined by measuring the absorbance of
0.0500 g/L K,Cr,0, at 350 nm. The absorbance was found to be
0.5351 t 0.0003. The absorptivity was taken as 10.71 L/cm g (48) from
which the cell pathlength was fortuitously found to be 1.000 cm.

Having obtained the cell pathlength. the only other parameters
required for the correction factor equation are the empirical transfer
function values for H and K of Equation 3-5. In addition. freedom
from stray light and fluorescence interference in the transmittance
measurements must be confirmed by experimentally verifying absorbance
linearity as a function of concentration. With experimental
fluorescence and absorbance concentration curves of quinine sulfate
(08). the instrument may be checked for adherence to Beer's law while
providing fluorescence information to test the suitability of
Equation 3-5 by evaluating the goodness of (fit of the data to
Equation 3-6. Fluorescence and absorbance values were determined for
solutions of 08 ranging from 0.1 ul to 400 ul. A 4 nm bandpass
(0.5 mm slits) centered at 365 nm (excitation maximum) was used for
excitation and .' 16 um bandpass (2.0 mm slits) centered at 450 nm
(emission maximum) was chosen for emission. The PlT voltage was -600 V
and the fluorescence current sensitivity was 10’ V/A.

Excellent absorbance linearity to greater than 3 absorbance units
was observed. as shown in Figure 3-2. A weighted least squares linear
regression of absorbance as a function of concentration resulted in

these figures of merit:

 

 

ABSORBANCE (.385 n m)

52

2.00 “'- .1

1.00 .0- /

/

“mi ? i 1‘ 11— 'r } 1‘ —1‘
'3.

100. 200. 300. 400.

 

 

oumme SULFATE (12M)

Figure 3-3. Quinine sulfate absorbance at 365 um.

53
slope. (sb) - 7566. t 3. u"

intercept I -0.0015 t 0.0004

standard error I 0.0019

correlation coefficient. (r) I 0.9999905
Iorst case linearity extends to an absorbance of 1.0 with a 16 nm
excitation bandpass. Large Beer's law deviations were observed for
conditions of high fluorophore concentration with highly absorbing
solutions when the transmittance monochromator was removed. This
problem results from the fluorescence becoming a significant fraction
of the light detected by the transmittance detector. Although the
need for post-cell transmittance filtering was not anticipated. the
20 nm bandpass of the transmittance monochromator was found to
adequately reject fluorescence. Earlier. a broadband filter was found
to be sufficient; however. the convenience of tunability with the
monochromator is much more desirable.

Values of H and K were determined by regression of the
fluorescence vs. absorbance ((‘°)ex) data on Equation 3-6 ((sc)"I I 0)
using IINFIT4 (37.38). Excellent results were obtained:

H I 1.108 t 0.052

K I 9.60 t 0.33
The average relative error of the calculated estimates was
0.05%. Vith data from a second. repeated experiment and these
estimates of the parameters H and I. primary absorbance corrected
fluorescence values were calculated using Equations 3-8 and 3-9.
Figures 3-4 and 3-5 show the uncorrected and corrected fluorescence
data as a function of concentration in linear and logarithmic form.

The corrected fluorescence is linear over nearly 4 decades. Weighted

54

500000. II!-

10m}
5
S
\

\

300000. "F' /

200000. "' '- /

FLUORESCENCE INTENSITY (55535

100000. dr- ,V

 

  

 

 

 

g l L I J J l I
- F r r I F 1 i
a so 100 150 zoo 250 300 3.50 400

QUININE SULFA'IE {pm

Figure 3-4. Fluorescence intensity (65.535 I 1.0 nA Pl'l‘ anode
current) vs. quinine sulfate concentration; linear-linear
plot. (A) Uncorrected; (V) Absorbance-corrected. 365 nm
excitation wavelength: 450 nm emission wavelength.

FLUORESCENCE INTENSIW (65535 = 1.0 M.)

55

IOUWXI

//
/
1000.
100.
I/’
,/
l
10. x’
1 .
0.01 0.10 1.00 10.00 100-00 1000.00

QUININE SULFATE {001)

Figure 3-5. Fluorescence intensity (65.535 I 1.0 nA FIT anode

current) vs. quinine sulfate concentration; log-log
plot. (A) Uncorrected; (Y) Absorbance-corrected. 365 nm
excitation wavelength; 450 nm emission wavelength.

56

linear regression of corrected fluorescence provides the following
summary:

slope - [1.02609 : 0.00028] - 10’ u“‘

intercept I -80 t 5

standard error I 406

correlation coefficient. (r) I 0.99975

The unique prOperties of the bifurcated fiber-optic coupling

result in a decrease in the fluorescence signal at absorbances greater
than 1.5. As the absorbance increases. there is a movement toward the
fiber-optic of the volume in which most of the excitation beam is
absorbed. and thus. from which most of the fluorescence is emitted.
As this emission volume moves very close to the fiber-Optic. the
emission-collection fibers are spatially precluded from efficient

emission collection. as earlier predicted.

2. Effegtivengsglgf Correctign Progegures

To evaluate the effectiveness of the primary absorbance
correction. 2.5-dihydroxybenzoic acid (DHBA) was used as an
interfering chromophore. The absorption spectrum of DHBA in
0.2 N H380‘ has a peak at 329 nm which falls off to zero above
400 nm. At pH values higher than 2.0. DHHA fluoresces strongly in the
blue region. However. in 0.2 N 3,30. (pH 2 1.0) the quantum efficiency
decreases dramatically (55). Fortunately. DHBA does not quench the 08
fluorescence. nor does the fluorescence spectrum of DHRA overlap
appreciably with the excitation spectrum of 08; thus any error due to

reexcitation and reemission via this pathway is precluded.

57

An experiment was devised (by which the as fluorescence at
constant concentration was to be determined in the presence of varying
amounts of the primary-absorbing and fluorescing DHEA. One set of
solutions was prepared with 20.0 ul OS and 0.0 ml to 0.8 ml DHRA in
0.2 H1804. Another set of solutions was prepared with matching
concentrations of DHHA in 0.30.. but without 08. Using a 4 nm
excitation bandpass centered at 329 nm. the DHRA absorbance maximum.
and a 16 nm emission bandpass centered at 450 nm. the OS emission
maximum. both sets of solutions were examined. The PlT voltage was
-610 V. and the current-to-voltage converter was maintained at 10' VIA
gain. Absorbance values extended to 3.25 with good linearity. The
absorbance-corrected fluorescence intensity from OS was determined by
subtracting the corrected fluorescence contribution of the DHRA. as
determined from the DHBA solutions. from the corrected fluorescence
intensity of the mixed OS and DHHA solutions. The results are shown
in Figure 3-6. amply demonstrating that the technique accurately
corrects for primary-absorbance inner-filter effects. The average
relative standard deviation of the mean 08 fluorescence is 0.39%.
while the maximum error is less than 1%.

Since the OS also contributes appreciable primary absorbance
(A I 0.17). the experiment demonstrates that the correction technique
is independent of the number of fluorophores or the relative
distribution of absorbance between principle fluorophore and other
chromophores. while validating the additivity of corrected

fluorescence.

 

 

Figure 3-6.

58

Quinine sulfate (08) and 1.5-dihydroxybenzoic acid (DHBA)
fluorescence intensity (65.535 I 1.0 nA Pll' anode
current):

(I) Primary-absorbance attenuated and

(D) absorbance-corrected fluorescence from 20.0 pl 08
with 0.0 to 0.8 ml DHBA:

(A) Primary-absorbance attenuated and

(V) absorbance-corrected fluorescence from 0.0 to

0.8 all DEBA:

(X) Absorbance-corrected fluorescence of 20.0 Ill 08 in
the presence of 0.0 to 0.8 ml DHBA. DEBA fluorescence
subtracted: I I D -V .

328 nm excitation wavelength; 450 nm emission wavelength.

1.0 m}

FLUORESCENCE INTENSIW (55535

70090.0

59

 

 

 

 

_ n 4——--EI
.. : U
00000.0 .. 4: 0: X x )1. 3:
50000.0
40000.0 -
00000.0 1
20000.0 -1
10000.0 ..
0.0 -
"9090-9 l f 1 r I T T *I
0.0 1.0 2.0 3.0 4.0 0.0 0.0 7.0 an

DHBA {mM x 10)

Figure 3-6

60

3- M10 the o 9.1__.19._LC°“ ti nmmm

Although this instrument was not initially designed or intended
for correction of secondary absorbance inner-filter effects. the
method. as derived. is theoretically capable of including both primary
and secondary absorbance in the correction factor. To test this
hypothesis. the secondary absorbancefcorrected fluorescence of 20.0 pl
03 was determined in the presence of 0.0 ul to 80.0 ul sodium
fluorescein in 0.2 N H,SO.. The excitation and emission wavelengths
were 365 nm and 438 nm (fluorescein absorbance maximum). respectively.
and both monochromators had 4 nm bandpasses. The narrow excitation
bandpass is necessary to maintain Beer's law linearity for absorbance
measurements and the emission bandpass is adjusted to match the
bandpass at which the absorbance is determined. The excitation and
transmittance monochromators were first set at the emission wavelength
of 438 nm and the secondary absorption was measured for all solutions.
The excitation and transmittance monochromators were then reset to the
excitation wavelength. The primary absorbance at 365 nm and
fluorescence at 438 nm were then measured for all solutions. The
primary absorbance ranged from 0.16 to 0.33 while the secondary
absorbance ranged from 0.00014 to 3.06. with good linearity. The
fluorescence values were corrected for both primary and secondary
absorbance inner-filter effects. Results are shown in Figure 3-7.
The relative standard deviation of the mean corrected fluorescence is
1.5%. However. all of the results appear to be high relative to the
first solution which had negligible secondary inner-filter

attenuation. since it contained no fluorescein. Relative to the first

 

 

61

600010 HI-

50000-0

40000.0

30003.0

20000.0

FLUORESCENCE INTENSITY (65535 = 100 M}

10000.0

 

 

 

139 l I l l l l .J
- l r r I T 1 ‘I
0.0 0.0 1 .0 1 .0 2.0 2.0 3.0 3.0

TOTAL ABSORBM‘JCE (355 nm & 438 nm)

Figure 3-7. Fluorescence intensity (65.535 I 100 pA PlT anode
current) vs. total absorbance of 20.0 pl quinine sulfate
in the presence of 0.0 to 80.0 ul sodium fluorescein:
(AD uncorrected: 0?) Primary and secondary
absorbance-corrected. 365 nm excitation wavelength;

438 nm emission wavelength.

62
solution. the mean error is 3.1%. and the largest error is 6%. which
is much improved compared to the range of errors of 41%1to 89% before
absorbance correction.

Several potential problems could contribute to the positive
errors. The wavelength setting accuracy of the excitation and
emission monochromators is :1 nm and the setting reproducibility is
t 0.4 nm. A significant discrepancy between the wavelength at which
the absorbance was actually measured and the wavelength at which the
fluorescence was measured could have contributed to an error.
Secondly. although the quantum efficiency of fluorescein in acid is
low. a small overlap of the fluorescein fluorescence spectrum at
438 nm could cause a slight additional fluorescence contribution.
This possibility seems unlikely. but it is difficult to evaluate.
There may also be some small discrepancies in the cell modeling which
are compensated for in the primary absorbance correction by the curve
fitting. but which are not symmetrical with respect to secondary
absorbance and thus go uncorrected. Such asymmetric errors could
include reflection of the excitation beam from the quartz light pipe.
effects of the inlet and outlet ports of the cell. refractive index
effects. and changes in the fiber optic transfer function which are a

function of wavelength.

E. Qonglusiogg

The fiber-optic coupled front surface fluorescence flow cell
appears to function quite well for primary-absorption inner-filter

effect corrections. It is low volume. tolerates vibrations and cell

63

movement. and is easily optically coupled to the excitation and
emission monchromators. The transmittance monochromator could also be
similarly coupled. making this cell ideal for a variety of flow
methods. Since the primary absorbance and fluorescence can be
simultaneously measured on the millisecond time scale. this instrument
is readily adaptable to kinetic methods. including stopped-flow mixing
systems. Reaction monitoring can be based on either absorbance or
fluorescence alone. or absorbance information could be used to flag
for inner-filter effect errors and to correct them. Alternatively.
one species could be monitored by absorbance while another was
simultaneously monitored by fluorescence.

A principle disadvantage of this instrument is the need for 3
light filters. Although these would ideally be matched monchromators.
in certain dedicated applications bandpass filters could be used.
This particular cell also has an appreciable problem with a high
fluorescence blank signal which may be attributable to the innate
fluorescence of cell materials or adsorbed materials. or a result of
stray light from the back-scattered excitation beam. Poor wash
characteristics are also apparent. which could be a result of
adsorption or trapping in the cell. Future cells should probably not
be constructed with the black Delrin used. since chemical degradation
of the black Delrin has been observed in this lab. Graphite filled
Teflon or white Teflon (a modified theory may be required) would
perhaps be preferred.

Vith matched. stepper-motor driven monochromators.
secondary-absorbance corrections could be automated. Stepper-motor

coupling should improve the wavelength accuracy and setting

64
reproducibility. thereby improving correction accuracy. The
excitation and emission monochromators could be programmed to
repeatedly slew between excitation and emission wavelengths to
determine the absorbance at both wavelengths for the same solution.
In addition. a microcomputer controlled emission monochromator would
permit corrected excitation and emission spectra to be measured.

The signal-to-noise ratio (SNR) characteristics of this
instrument also appear to be better than those for the Christmann cell
shift instrument. This is probably and predictably attributable to
the elimination of the collimation requirement of the method of cell
shift. As a comparison. the cell shift instrument of Christmann
achieves a maximum absorbance SNR (signal/s.d.) of about 1000 at
A I 1.5. (12) whereas this instrument achieves a SNR of about 5000 at
A I 1.5. The use of both pre- and post-cell monochromators also
provides excellent absorbance linearity and better immunity to
reemission errors as occasionally observed in the absorbance
measurements with the method of cell shift. For the maximum
fluorescence signal observed with the cell shift instrument. 5 nA. the
SNR was approximately 170. For this method a SNR of 900 is achieved
under similar conditions. Similarly. for 20 pl as excited at 365 nm
on the Hg line of a 150 I Hg-Xe arc and observed at 436 nm with 4 um
bandpasses. the Christmann corrected fluorescence SNR was
approximately 60. For this instrument under identical conditions
except for 438 nm emission and excitation with a 150 I Xe arc the SNR
exceeds 500. despite identical bandpasses but .narrower slits and

without the advantage of the higher Hg-Xe arc flux at 365 nm.

65

Two other practical problems. though surmountable. remain with
the present instrument. First. Xenon arc intensity variations of 10%
to 20% occur over the course of most experiments. Despite the use of
a reference detector. such large drifts cannot be adequately
corrected. causing inaccuracies in the data. An optical feedback
stabilized arc could minimize both the short term variations and long
term drift to less than 1.0%. latching of the reference and
transmittance detection circuits. in the same housing if photodiodes
are used. would also likely reduce drift. Lastly. bifurcation of the
end of the excitation fiber optic and randomization of these fibers
would permit a more accurate and homogeneous sampling of the
excitation beam. by the reference detector. Currently. the reference
and excitation beams are slightly different with respect to both
sampling geometry and wavelength. and both of the photodiodes and
their amplifying circuits are different. The other major problem in
this instrument is 1/f noise. Thus far. this problem has been the
overwhelming source of error. particularly at low light levels. where
the dark current 1/f contribution becomes significant. Further useful
studies of precision. SNR. reproducibility. linearity. and accuracy
cannot be completed until this 1/f noise is vastly reduced. This
problem could be eliminated by chopping the source beam. A simple
mechanical chopper could easily be used in conjunction with the

microcomputer at chopping rates between 100 and 500 Hz.

CHAPTER IV

INSTRDIENTNTION AND SOFTWARE FOR
LHlINESCENCE lEASURElENTS

A. Introduction

The development of the research instrumentation used in the
experiments outlined in this thesis required a system. which
incorporated intelligent data acquisition and parameter control.
Because both flexibility and adaptability are desirable. a computer
system is the obvious choice. allowing evolution of instrument
hardware or applications to be accomodated by software. A computer
system also facilitates accurate data storage. transfer. and
retrieval. giving the chemist greater power to make decisions and draw
conclusions through the use of graphic display and statistical

evaluation.

B. A Hierarchical Cogputing System

A hierarchical computing sytem was used. This system links a low
cost and flexible Intel 8085-based (56) microcomputer system with a
more powerful L81 11/23 (28) "mini"-computer. The 8085 microprocessor
and its associated family of support chips form an efficient system
which can be readily adapted to many laboratory applications. The
interrupt structure of the 8085 system gives quick and efficient
response to the needs of peripherals and interface. Serial
communication beween the 8085 and the 11/23 gives the user the

66

67

advantages of custom interface design for data acquisition and control
of the 8085 and the power of the 11/23 for editing. calculating.
cross-assembly. compilation and mass storage. The 8085 also has
greater portability or can be affordably dedicated to a single
instrument. This computer networking system allows several
microcomputer-based instruments to share the software and hardware
facilities of the 11/23.

The chief drawback of the networked system is the lack of
high-speed access to the 11/23 upon which the instrument is dependent.
Because the serial communication is relatively slow. program
downloading or data uploading and the requirements in writing the
software to facilitate these transfers are time absorbing and often
difficult. Total portability is also limited due both to the
requirement of a serial link and the real or apparent dependence of
the user on the powerful facilities of the 11/23 RSI-ill environment.
However. local mass storage could overcome the physical dependence

upon the 11/23.

C. .A lodular licrogogputgr

The 8085 microcomputer used for these (and other) instruments was
a versatile system designed by Bruce Newcome within the research group
of Dr. C. G. Enke (22). Its general design attributes include
modularity and flexibility in adapting and expanding its peripherals.
Various capabilities can be added to the computer via small boards of
selected functionality. Individual functions can be easily upgraded

as more powerful integrated circuits become available. Reliability

68
and compactness are enhanced through selection of large scale
integrated circuits. wherever possible. The microcomputer's
usefulness was 'further extended by constructing it within a single
portable case. with standard BNC connectors for analog input and
output (I/O) and a standard "D” type connector for parallel digital
I/O. The instrument has available 4 parallel analog inputs. 2
parallel analog outputs. 22 programmable digital I/O lines. and 2

serial RS-232 links.

1. Basic lodules

The microcomputer hardware capability is comprised of the
following modular subunits:

1. CPU - 8085A.microprocessor and associated bus transceivers and
buffers. status indicator LEDs. and hardware RESET switch.

2. RDl/RAl - up to 32! of any combination of the designated 2! read
only memory (ROI) or random access memory (RAl) chips. A total of
4! of ROI and 12! of RAl was used for the chemiluminescence
experiments. while 10R of Rfll and 14! of RAl was used for the
fluorescence experiments.

3. Programmable Interrupt Controller - The PIC intelligently expands
the processors interrupt capacity and capabilities..-

4. Dual USART - Tho RS-232 communications ports are available. One
port is assigned to the user's terminal and the other is usually
used for I/O with the 11/23.

5. Chip Select - The addressing of most devices other than the

ROl/RAl and CPU boards is accomplished with the aid of this single

69
decoding module. unnecessary redundancy of decoding circuits is

thereby eliminated at a savings of time. space. and expense.

Four other specialized circuits were used to customize this basic
microcomputer to realize its intended purposes. A dual-DAC board
employing 2 8-bit digital-to-analog converters was added to provide
triggered I-t or 14! oscilloscope display of data or X-Y/X-t recorder
plots. The three remaining circuits were designed by the author. Tho
of these circuits (timer and ADC) are currently being used in other

microcomputers. while the third (S/H-lHX) is perhaps the most novel.

2. CoggterlTimer I; Board

The Counter/Timer II board supercedes an earlier board based on
the Intel 8253-5 device with the Advanced licro Devices 9513 System
Timing Controller. This board is simple in external design. belying
an extremely powerful microprocessor designed for counting.
sequencing. and timing applications. To fully appreciate the
usefulness. flexibility and complexity of this device. the reader is
referred to more complete documentation (57). The 9513 contains five.
16-bit. individually programmable counters. Applications of interest
include time-of-day clocking. real-time program independent clocking.
count down/up. polled or interrupt timing for data collection
sequencing or peripheral servicing. programmable waveform synthesis.
and event count accumulation.

The interface of the 9513 to the microcomputer bus is
straightforward. Bus lines D0—7 are interfaced to the 9513 I/0 pins

DB0_7 through the 74L8245 tri-state transceiver. NOT WRITE. NOT CHIP

70
SELECT. and NOT READ are interconnected between bus and 9513
(pins 9-11. respectively). and the NOT READ and NOT CHIP SELECT bus
lines are also connected to 74L8245 lines SEND/NOT RECEIVE (pin 1) and
NOT CHIP ENABLE (pin 19). respectively. Bus line A0 is connected to
9513 input CONTROL/NOT DATA (pin 8). Ten kilohm pullup resistors are
connected beween power and 9513 lines DB8_15 and SOORCE1_5. An
external crystal-controlled TTLrlevel output device is the preferred
timing base for the 9513 since the internal oscillator was found to be
unreliable. Such a device can be added to the board in the

multi-purpose wire-wrap area provided.

3. Analog Input

The analog input section of the microcomputer is comprised of 4
sample-and-hold (S/H) inputs multiplexed into a 12-bit
analog-to-digital converter (ADC). This circuit allows simultaneously
gated sampling of up to 4 analog signals. followed by sequential
analog-to-digital conversion of each signal sample. Analog sampling
and multiplexing and the associated bus interface are located in a
single module appropriately called the sample-and-hold multiplexer
(S/HrlUX) board. The ADC and associated curcuits are on a separate
card which is one of several modules that form the set of Optional

functions commonly available for other microcomputers of this type.

71

4. Sppple-and-Holg lultipleger Board

The SIB-HUI is described here as 3 sub-units: the differential.
input S/H. the multiplexer-SIB output. and the S/H gating and l0!
channel addressing decoder. latch. and indicator circuit. as shown in

Figures 4-1. 4-2. and 4-3.

(a) Differential Input Sample-and-Hold

Four identical differential input inverting S/H circuits form the
first stage of the analog input circuit. The difference input (58)
S/H. is based on a commonly available monolithic circuit SIH (Datel
Intersil SHl-IC-l; Analog Devices AD583; Harris HA-2520/2425) (59-60)
modified for improved differential input characteristics. as shown in
Figure 4-1. RN60-type metal-film 100 kilohm resistors (R1-R‘) were
carefully matched to better than 0.02% using a portable bridge. This
provides excellent common mode rejection (measured to be better than
85 dB). while providing immunity to ground loops between the analog
input commons. An alternate differential input circuit considered
would have required 4 instrumentation amplifiers at considerably
higher cost and/or complexity. A 10.000 pF polystyrene capacitor was
used to provide S/H characteristics consistent with the overall analog
input design requirements: 12 microsecond acquisition time. 5 mV per
second droop. 0.002%1ho1d-mode feedthrough. 2 mV sample-to-hold offset
error. and 0.005% sample-to-hold gain error. Each of the 4 Sle has
an independent digital control permitting on-board (described later)
or external mode control for flexible operation initiated by software

or asynchronous events.

72

u—omlunelenmmee «anew ueuuneneuuaa .nlv enough

 

 

 

 

£>

10

 

 

xazlld

 

 

 

 

 

 

73

.000000 mxa 1 cone—anunam .u14 cannon

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

:
_IHO G18— HNZ N _
_(
G u .1 (U 0
O .30
T
Tl. ZwQO
mm 1111..
0+ H. 0.010
.mhfixsa
Na .
yew

 

 

 

 

 

‘9

74

.uoueomunm one neuooee moo-n neeumoe memoemo Hal one unuuem nxm .nlv.enemum

10%

Ace

 

 

 

 

e9

 

 

3.00
r

my?

 

 

 

 

 

 

Teawmmw yQJQ‘mmw
a on
a 1 R
1 _ A. _ n
~o_zozwn1~zm .N _<n o.
z.
.0
00 .0 0 no we .0
k\ Rt 0‘ ti NE. _/_.
<0 <0
X22

 

75

(b) lultiplexer - S/H Output

All of the _difference-input Sle are multiplexed by a ClOS
8-channel multiplexer (59). The lX-808 multiplexer was chosen because
the inputs have overvoltage protection and break-before-make switching
is used. This monolithic device is also popular and reasonably
priced. Channel decoding is software programmable via a decoded latch
(discussed later). The NH! output is buffered by a simple LF351 (61)
follower circuit to maintain transfer accuracy despite the finite l0!
load‘ impedance. The buffered output of the HE! is directed into an

inverting SIH used as the ADC input buffer as shown in Figure 4-2.

(c) S/H Gating and HOT Channel Address Latch Decoder and

Indicator

A 74LS75 quadruple latch and a 74L800. as shown in Figure 4-3.
are the integrated circuits used to provide a variety of S/H gating
and NH! channel selection nodes. A chip select (EERUI) pulse at the
multiplexer enable inputs (EN¢_3) latches the status of bus data lines
DO-l' to provide l0! channel selection via 03 and 02. respectively.
LED lamps tied to outputs 05 and 03 give the microcomputer operator
visual indication/diagnostic of the analog channel(s) chosen.
Similarly. a second chip select (CSs/n) pulse at enable inputs (ENO_1)
latches DO-l at 00-1. respectively. Along with latched outputs 00_1
reflecting data lines DO-l' a latched value for D0°D1 is also
available for gating S/Hs. These three outputs provide a variety of
jumper-selectable S/H modes for simultaneous or asynchronous

software-controlled sampling.

 

 

76

5. General Pugpgge 13-31; A29 Heard

The ADC board was designed to be a relatively low cost. general
purpose. medium speed. l2-bit ADC interface. A basic functional
schematic is shown in Figure 4-4. The circuit is based on the AD574
(60). a successive approximation l2-bit analog-to-digital converter
with a 35 us conversion time. The AD574 (1C1) has an internal voltage
reference. internal clock. and provides for a 10 V or 20 V range
unipolar or bipolar analog input. The choice between unipolar or
bipolar cperation is provided at building time by the proper selection
of components and jumpers. as described below:
unipolar (0 to 10 V or 0 to 20 V):

Pl I 100 ohm potentiometer

P2 I 100 ohm potentiometer

R1 I 100 kilohm resistor. J1 to J7

R2 I 100 ohm resistor. J8 to J9

Jump J1 to J2
Bipolar (-5 to +5 V or -10 to +10 V):

P1 I 100 ohm potentiometer

P2 I 100 ohm potentiometer

Jump J2 to J3

Jump J4 to J5

Jump J5 to J7
Space is also provided for a S/H amplifier to be used as a
non-inverting gain-of-one input buffer.

The digital interface is completed by a 74LS245 octal tri-state

buffer (1C2). and a 74L8368 or 8T98 tri-state hex inverter (1C3). The

77

.oueoe 9:4 «manna eeonnnm ueucnec

.414 canned

 

 

 

 

 

 

 

 

 

 

 

 

E 0..
i i i i M...
- -w , , 0..
. L. w». o<..z_>o~z_>o_ a. 2..
V
Wm... 0M a
- m8. .1 no Em as 3.0%....00 Ni
51¢ . ~10
0AM Tan“ :5me m. N...
0100A“. mg mg: _o_ H... d:
zlmm
Em mm mo m5
F n+3 + n
J ET 0. 0
_mb ....

 

«.19

 

 

78

octal buffer drives the ADC output onto the bus under the direction of
'CS1. When reading. A0 LO forces data bits DR4_11 (lSBs) onto the data
bus. D0-7' With'Ao HI. DB0_3 are forced onto D0-3' and a binary 0000
is forced onto D4_7. The conversion cycle is initiated by writing to
the memory area of 001. During conversion. the status output line.
STS. goes HI. until the conversion cycle is completed. This line is
polled by the microprocessor by reading from.the memory space of 'CS2.
where D7 LO indicates that the ADC is busy. The STS line may also be
used to asynchronously interrupt the processor at the end of the cycle
and/or to trigger the S/H buffer to hold the analog input steady
during the conversion cycle.

The first version of this board is still being used in the
microcomputer described in this thesis. This board was originally
found to be missing codes in the two or three least significant bits.
This problem disappears when the bus is halted during the conversion
cycle (the SIS line is used to generate a microprocessor "wake-up"
interrupt). indicating that the missing codes may be symptomatic of an
undersized ground plane shielding the ADC from electromagnetic
interference generated on the bus. This practice is still being used
for this board; however. the ground plane has since been expanded on
later boards. Additionally. an area has been added for an optional
:15 V dual-tracking voltage regulator. for those microcomputers whose

busses do not already meet this need.

79

D. A Photodiode Detector

Three light-detectors were used for the fluorescence experiments:
a PIT for fluorescence detection. a photodiode detector for an
excitation intensity reference and a photodiode for excitation
transmittance detection. Photodiodes were used because of their
durability and immunity to vibration. because they are not adversely
affected by a very high photon flux. and because they are inexpensive.
have a wide dynamic range. and are stable and devoid of hysterisis or
memory effects. Additionally. they do not require a high voltage
power supply. They can be easily integrated into a umall package with
the required power supply and current-to-voltage converter and
amplifier. The reference photodiode detector was designed by Patton
(54). A functional diagram of the device designed by the author is
shown in Figure 4-5. A Hamamatsu S780-5BQ silicon photocell (62)
develops a current which is converted to a negative voltage by the
first op-amp. The feedback amplifier is switch selectable with the
feedback resistance ranging from 100 R0 to 1000 l0. The feedback
capacitors were adjusted to maintain a 10 us time constant. The
second stage amplifier is a voltage follower with gain variable from 1
to 20. The third op-amp is an inverting. active filter with
switch-selectable capacitors ranging from 100 pF to 10 uF for rise
times from 10 us to 1 s. All of the op-amps are LF351s. Power is
internally supplied by an Ll326 :12 V dual-tracking voltage regulator.
The regulator is in turn supplied by an external :15 V supply via a
shielded-twisted pair. Since very little noise is introduced into the

metal enclosure (case common) by the power cord. and a very stable

.ucuoeueu ooouuouomm < .nlv enough

 

 

 

 

% 41.
s
41

 

 

 

 

 

 

81

power supply is maintained by the dual-tracking regulator. the
detector circuit maintains a high quality. low drift output.

unfortunately. silicon detectors have a very low quantum
efficiency in the UV. Due to this inefficiency. higher gains are
required. resulting in correspondingly greater amplifier noise and
drift problems. This problem may well be reduced if the LF351
general-purpose op-amps were replaced with op-amps that had lower
noise and drift characteristics. Alternatively. if studies below
370 nm continue. these detectors could be replaced by photomultiplier
tubes which have much better quantum efficiencies in the uV. Lastly.
users of photodiodes should be aware that the high IR quantum
efficiencies may require an IR blocking filter to be used in
conjunction with broadband cutoff. bandpass. or dichroic interference

filters. as these filters often pass some IR.wavelengths.

E. licrocopputer Software

1. Choosing p licrocogputgr Lpnguggglgperating stgem

The choice of programing language used with this
microcomputerized instrumentation was based on several factors. The
decision must be made within the context of the experimental and
hardware requirements. costs. ease of installation. availability. and
strengths and weaknesses of a given language. A major requirement for
the language of choice is that it permit maximum use of all the power
and speed of the microcomputer while minimizing the difficulty of
programming at both high and low levels. It should be relatively easy

to learn and facilities for self-tutorial or other instructional

82
opportunity should be available. Documentation should be both broad
in scope for the novice. and specific. accurate. and thorough for the
expert. The language should facilitate computing at both the machine
language or assembly level as well at a high level. At the low level.
the programmer should be able to use all possible processor
instruction codes to implement any task required for a given set of
hardware. This code should also be easily merged with other high
level programming. System constants and variables should be mutually
accessible at both levels. High level programming should facilitate
”housekeeping" of memory management for arrays. variables. constants.
pointers and vectors. High level commands should be available for
comparisons. jumps. loops. returns. and conditionals. Tbrminal and
mass storage I/0 tasks should be readily implemented. Ideally. the
language might have a continuum of commands to cover programming from
very low to very high levels. A facility for efficient and rapid high
level mathematics should be included. (Although this usually takes
the form of a floating point math package. fixed point math packages.
as discussed later. are also viable options). The ideal microcomputer
operating system environment would also provide facile editing. and
rapid compilation and assembly. Error messages and program
development techniques should reduce debugging time and frustration to
a minimum. System firmware and user programs should be compact.
lultitasking performance is an often desired option. while multiuser
capability is of little probable value for a hierarchial environment.
A file-structured system orientation is highly desirable. Ultimate
power and flexibility in software with currently popular conventional

languages (such as BASIC) often implies the maximum in development

83
time. maintainence time. expense. complexity. memory and execution

time. Other options are worth considering.

2.01.02:

Two languages were used for the experiments described within this
thesis. The chemiluminescence experiments were conducted with the
language SLOPS. written by Hugh Gregg. SLOPS is a minimum system
which contains a kernel linking several very basic subroutines and
providing for communications between the 8085 microcomputer and the
11/23 minicomputer host. Programs were written in 8085 assembly
language mnemonics on the 11I23 and then cross-assembled and down
loaded into 8085 memory for execution. This system has the advantages
of using the powerful text editors and file-structure of the 11/23.
but suffers from lengthy cross-assembly and serial communications
times. High level computing was virtually unavailable on the

microcomputer.

LEE

FORTH. a rising star in microcomputer control systems languages.
was used for the fluorescence instrument. FORTH is a readily
exstensible language and features total system capability. Editing.
compiling. assembly. math packages. and I/O are all available in
compact firmware. A unique feature of FORTH is that it does not
actually compile. but rather links programs or subroutines into new
programs or subroutines. Programmers begin with a subset of useful

subroutines called "words" in FORTH lingo. Programming entails

84
creating new words by linking previously defined words or assembly
language commands. FORTH links/compiles quickly and is compact. Near
ultimate speed ii available only through linear programing in assembly
language which obviates many of the advantages of FORTH. Since FORTH
programs are ultimately a collection of highly nested subroutines.
considerable time is spent by the FORTH executive in jumps and
returns. For high level mathematics calculations. however. FORTH
excels in speed because of its use of fixed point rather than floating
point math. as has been convincingly argued elsewhere (63). Other
criticisms of FORTH are its lack of a file structure. difficulties in
documenting or reading FORTH programing . and the need for uniqueness
of "word" naming. lultitasking and multiuser capability are also

features which are potentially useful though not required.

F- §2££13£3 £2£.£ 223l:2££2l£nl£h EL finsesrsaetes

All of the chemiluminescence experiments were run within the
SLOPS microcomputer operating system/language. Programs were
downloaded from the 11l23 via the serial link using a protocol in
which the 8085 requested that the 11/23 download an object file and
the 11/23 responded by downloading the file preceded by starting
address and length. Since the 8085 response time was quicker than the
transfer rate determined by the baud rate of 2400. no complicated
handshaking was required.

The voltage from the current amplifier was sampled by a
sample-and-hold amplifier and converted to digital form by the ADC.

Signal integration over integral multiples of 16.67 ms (1 period of a

85

60 Hz signal) was employed to eliminate a significant 60 Hz noise
component of the signal; a result of the very high gain
characteristics of the instrument. For both cell sides. at each
oscillation. 1006 AID conversions were acquired by an interrupt-driven
subroutine at 116 us intervals. as determined by the Counter/Timer II
board. These data were accumulated and rounded-off to form a l6-bit
sum representing the signal integrated over seven 60 Hz periods
(116.7 ms). At 2 samples per oscillation and 2 oscillations per
second. this sampling rate resulted in an overall duty cycle of 41%
for data collection. After subtraction of the dark current value.
determined in a similar way just before each run. the signal values
were stored in memory for later processing.

Data were stored as successive intensity pairs from alternate
sides of the cell. After each run. data were uploaded to the 11I23
into a binary file. A more complicated protocol requiring a "receipt
acknowledged" character from the 11I23 to the 8085 was required to
allow for the slower 11I23 response time. Data were post-processed on
the 11I23. Individual ratios were calculated for each intensity pair.
Ten pairs were averaged and standard deviations for each mean were
calculated. A weighted average was then calculated using all means.
The integrated light intensity was similarly calculated by using the
intensities from the long pathlength side of the cell. Using these
ratio and intensity values. the absorbance. correction factor. and
corrected intensities were all calculated and reported in tabular

form.

_.. AEtL-‘l

86

G. ngtware fpg‘pp Apsorpange-Correctgg Fluorippper

The FORTH operating system is used for all of the fluorometry
experiments. The 11I23 is used to transfer ASCII "blocks" into RAl
buffers for subsequent linking into FORTH code. Such transfers are
implemented on the 11I23 by FPIP (64). In conjunction with vectored
routines in the microcomputer firmware. FPIP allows the 11I23 to
appear to be a disk mass-storage device. Transfer protocols are
similar to those with SLOPS. except that checksums are monitored to
assure accurate serial transfer.

The voltage outputs from the three light detector circuits are
simultaneously sampled at 5 ms intervals. Average signal values are
accumulated from 2000 samples. Three FORTH words are used to conduct
the fluorescence experiments. "DARRGET" obtains and stores the dark
current values from the reference and transmittance detectors with the
source blocked. With the source unblocked. "OGET" obtains a reference
value from the reference detector. a 100% T value from the
transmittance detector. and a blank background level from the
fluorescence detector. "SPIN" executes an infinite loop which obtains
current light values from the detectors in 10 s intervals and then
calculates and displays the data in tabular form. The table contains
the dark. reference. and light levels. the light levels corrected for
dark current or background. the ratio of the current to the initial
reference levels. the transmittance and reference-corrected
transmittance. and the fluorescence and reference-corrected
fluorescence. This tabular display format allows the user to examine

the data in real-time. assessing drift and precision. as well as the

87
experimental trends. Currently. data are manually recorded from these
tables for subsequent calculation of the absorbance. correction

factor. and corrected fluorescence using a programmable calculator.

CHAPTER V

A UNIFIED lNTHElKTICAL
LUlINESCENCE VOLUlE-SOURCE lODEL

A. igppipg the Propleg

A basic premise of the techniques presented in this thesis. as
well as. other absorbance-corrected luminescence techniques. is that
the attenuation of the emission signal is a predictable phenomenon
governed by geometry and Beer's law. Beer's law is simple. familiar.
and easily applied. However. all to often. correction schemes have
incorporated a falsely simplistic approach to the geometric aspects of
the problem. or do not have clearly stated underlying presuppositions
or supporting proofs. This chapter is an attempt to provide a partial
proof or test of some of the geometrical assumptions applied in the
absorbance-corrected techniques of Chapters II and III. This work may
also prove useful in extending or modifying these techniques in new
ways. or for comparison to other approaches at geometrical
descriptions.

A luminescence signal represents the integrated intensity from
many planes of emitters. each affected by attenuation of light as a
function of its respective depth according to Beer's law. A primary
difficulty in modeling Beer's law attenuation arises from an
additional geometry complication. Each of these planes has a
particular optical efficiency with which it is excited or observed.

and even at different locations on the plane. this efficiency function
88

89

may vary significantly. Since modeling the luminescence signal
requires mathematical integration over all of the emitting volume. an
accurate model "of luminescence as a function of absorbance cannot be
obtained without careful consideration of the geometry of the
excitation and emission light paths. A mathematical model is
presented here that considers the excitation and emission geometry at
any point in the active volume of interest. Integration of this
point-source model over the entire planar area of emission produces an
expression that can be used to evaluate changes in excitation and
emission cross-section as a function of depth. independent of changes
which are a function of Beer's law attenuation.

In the absorbance-corrected chemiluminescence technique described
in Chapter II. an underlying assumption implicit in Equation 2-2 is
that the cross-section of detected light is constant with increasing
depth (independent of absorbance attenuation). Experimentally. this
restriction was approximated by using a collimator. This collimation
approximation permits simplified expressions to be used for the
derivation and correction of the inner-filter effect. Later in this
chapter. this experimental configuration is tested to validate its
effectiveness by a comparison at various depths of emission
cross-sections determined by this model.

For the fiber-optic front-surface fluorescence cell. the
point-source. excitation/emission model is linearly combined with the
fiber-optic angular acceptance function (AAF). which describes the
decrease in transmittance of the fiber-optic with a decrease in the
incidence angle measured with respect to the fiber axis (65). For UV

fiber-optics. the AAF is nearly Gaussian in distribution. typically

90
having a half-width of 20° to 30. at half-height (66). The sine of
the angle at which the light input or output falls to 50% of its
maximum is refered to as the numerical aperture (NA) and is a measure
of the light-gathering power of the fiber optic. The NA decreases
with decreasing fiber length and decreasing wavelength of light.
Since ‘the length of the fiber-optic does not change. and the
wavelength range is limited. the effects of figer length and
wavelength may be neglected here. The AAF is integrated with the
point source model over all angles of emission. and all point sources
are integrated over each plane of emission normal to the cell axis. to

determine the transfer function referred to in Chapter III.

B. Th; lodgl

For a point source emitting light into 43 steradians (all
directions) with equal intensity. the fraction of light which passes
through a hole in a plane of given dimensions. orientation and.
distance from that point source can be calculated.

Assume that the point source is at the origin (0.0.0) (see

Figure 5-1). and that the hole is described by a circle centered at:

with a radius of R.

91

 

.ccuooelanuom no novel ueomuuemoem

 

.«1n cannon

 

 

W

 

 

92

The equations of this circle are (Cartesian coordinates):

z I d (5-1)

(x - r)3 + y’ I R3 (5-2)

To transform to spherical coordinates:

H
II

psin(0)cos(0)

psin(0)sin(0)

'4
I

pcos(0)
0 I arcos(zIp)

0 I arctan(y/x)
Transforming Equation 5-1 into spherical coordinates:

pcos(0) I d (5-3)
Transforming Equation 5-2 into spherical coordinates:

p'sin'(0)- 20rsin(0)cos(0) I R2 - ra (5-4)

The fraction of light through this circle is given by the solid
angle formed between the point source and the circle. divided by 4n

steradians. The element of solid angle is
do I sin(0) dOdO (5-5)

Thus. the solid angle is obtained by integrating over the limits of 0

and 0:

6 - I” To .me) dOdO (5-6)

93
There are two cases for integration:
i. For r S R
0 extends from O to arctan((R - r)ld) where

O I 2n

ii. for r > R

0 extends from arctan((r - R)Id) to arctan((r + R)Id)

Equation 5-4 is then solved for 0 using Equation 5-3:

0 I rarcos(c) (5-7)

where

c - (a’t.n‘<e) - 2' + r’)I(2drtan(0)) (s-s)

Putting the limits of integration into Equation 5-6 for cases i and

ii:
a I In J2“ sin(0) dOdO +
0 O
arcos(c)
jg I+ sin(0) dOdO (5-9)
7 -arcos(c)
where

o I arctan((R - r)ld))
8 I arctan((R + r)ld))

1 I arctan((r - R)Id))

and c is given by Equation 5-8.

94

Equation 5-9 simplifies to
u . 2n[1 - d/(d' + (r - 31’11’2] +

arctan((R + r)ld)
2 sin(0)arcos(c) d0 (5-10)
arctan((r - R)Id)

Equation 5-10 can be numerically integrated using Simpson's rule. The

fraction of light is then

fraction I ul4n (5-11)

C. Application £2 Q_ Techigue

To determine the fraction of light which is emitted from a point
through a cylindrical collimator. Equation 5-10 is numerically
integrated using the circle defined by the collimator exit to
determine the values of R. r. and d for the limits on the first part
of the equation. For the second part of Equation 5-10 the low 0 limit
for r 2 R is defined by the collimator entrance; for r S R the low 0
limit is always defined by the collimator exit. The large 0 limit is
determined by the collimator exit. These limits are apparent from the
geometry (Note: The use of a long cylinder such as the leftover
cardboard tube from a roll of paper towels is useful for visualizing
this problem). Arcos(c) is the smaller of the values taken for both
the exit and entrance. (This yields the limiting value.)

The fraction of light from a plane can be calculated by numerical
integration of Equation 5-11 over all points of the active plane. and

is analogous to the integration of the area of a circle:

95

cross-section I I:° fraction(d)-rc drc (5-12)

where rc is the radius of the cross-sectional area. and is defined by
the intersection with the plane of interest of a ray intersecting the
collimator axis and both circles formed by the ends of the collimator
(see Figure 5-1).

Equation 5-11 was numerically integrated using Simpson's rule
with the FORTRAN program POINT on a PDP LSI 11I23. To reduce
potential roundoff errors. all calculations. constants. and variables
were double-precision. POINT was compared to the program CALLOBS
which uses a completely different approach employing elliptical
integrals. CALLOBS is based on an independent derivation by Dr. Gale
Harris (67). Both programs produce self-consistent results for a
variety of combinations of the variables d. r. and R. POINT was then
used as a subroutine to numerically integrate Equation 5-12 in the
program SLICE. I Results of SLICE have shown that the relative change
in the amount of light observed from each plane from the front to the
back of the chemiluminescence cell is approximately +4%. agreeing with
experimental evaluations within experimental error. The effect of
this increasing cross-section of detected light with increasing depth
was discussed in Chapter II. and it appears to be a negligible effect.

This validates the collimation assumption used in the derivation of

the correction method.

D. Appliga;ionp§p,php Fibep:gp§ic Coupled Fluorescepgg Teghnigue

To determine the transfer function. T(x). discussed in Chapter

III. the fiber-optic angular acceptance function (AAF) is linearly

96
combined with Equation 5-9 before integration over 0. The AAF was
evaluated by measuring the relative optical output of the fiber-optic
as a function of the angle of incidence of a collimated light beam as
described by lath (68). The data were found to fit the empirical

equation
1.6
mm - o"”°-35’ (5-13)

where O is given in radians. Equation 5-13 was linearly combined with
Equation 5-9 which in turn was integrated over all angles and points
as with Equations 5-11 and 5-12. resulting in several
excitation/emission cross-sections determined for various depths.
Results show that the fiber-optic cell excitation/emission
cross-section decreases exponentially with increasing distance from
the fiber-optic into the cell. This model was used to derive a
somewhat empirical transfer function based on this prediction of an
exponentical decay. as discussed in Chapter III. The resultant
transfer function based on this model is combined with theoretical
fluorescence equations to derive a mathematical correction factor

which accurately calculates absorbance-corrected fluorescence.

CHAPTER VI

CONCLUSIONS

A dual-pathlength method for absorbance-corrected
chemiluminescence has been demonstrated. Experiments have shown that
correction is possible to greater than 0.75 absorbance units.
Extension of the correction procedure to higher absorbance values is
predicted if the collimator artifacts are eliminated. the SNR
improved. and the emission bandwidth narrowed.

Further studies should be initially directed toward increasing
SNRs. This should be possible through increased light collection
efficiency by reducing or eliminating the collimation requirement and
extending the theory to account for these changes. The theory might
be extended through the use of luminescence volume-source models. such
as the model presented in Chapter V. Fiber optic coupling could be
used for light collection from a cylindrical cell similar to the
fluorescence cell described in Chapter III. Pathlength could be
adjusted by means of a piston alternately moving toward and away from
the fiber-optic. or conversely by moving the fiber-optic as a piston
within the cylinder. Curve-fitting of pathlength vs. signal
intensity information for a solution where A I 0.00 could be used to
develop an accurate light collection transfer function to convolute

with the function describing Beer's law attenuation.

97

98

Other obvious improvements could include the incorporation of
photon-counting detection and/or chopping of the light signal to
reduce 1/f noise. A thermo-electrically cooled PlT housing would
increase the reliability and convenience of cooling. Additionally.
since the sensitivity of the FIT is a function of temperature. a more
reliably thermostatted housing should reduce temperature induced drift
errors.

A multichannel detector such as an intensified diode array would
be useful for multiple wavelength studies. Simultaneous correction at
all wavelengths is possible. provided that the SNR is sufficient at
each wavelength. With improved instrumentation and SNR. further
theoretical and chemical investigations could be pursued.
Absorbance-corrected chemiluminescence with a narrow wavelength
bandpass should be tested at high absorbance values to see if
currently observed deviations are stray-light induced due to
nonmonochromaticity. as was suggested.

With the deve10pment of methods for faster measurements many
other chemiluminescent reaction systems could be studied. For
example. the rapid hypochlorite-luminol reaction might have the
advantage of negligible background absorbance. Applications in
clinical chemistry could be investigated. where inner-filter effects
have been noted for such interferents as riboflavin and erythrocyte
contamination. as mentioned in Chapter III.

Finally. a most valuable but equally or more difficult problem to
deal with is reemission. Since most chemiluminescence systems do
exhibit reemission. a correction method which permits or accounts for

reemission would be very valuable.

99

An absorbance-corrected fluorescence method for a flow cell has
also been demonstrated. The primary absorbance correction has been
demonstrated to. be accurate to within 1% to an absorbance of
3.0. Secondary-absorbance correction to an absorbance of 3.0 is
accurate. with all errors below 10%: however. this figure might easily
be reduced with improvements in technique brought about by
monochromator automation. as discussed in Chapter IV.

A variety of other possible instrumental improvements have
already been discussed in Chapter III. particularly methods for
reducing 1If noise. such as source stabilization and chopping.
Reduction of 1/f noise and the fluorescence background signal would
also extend the fluorescence detection limit. a figure which cannot
now be reasonably estimated because of 1If noise.

As discussed earlier. photon-counting. photomultiplier cooling.
or intensified array fluorescence detection might prove useful.
Alternate methods of secondary-absorbance correction. such as the
moving piston method for chemiluminescence as suggested above. might
permit primary and secondary absorbance corrections in near real-time.
Another option might be to replace the quartz light pipe with another
bifurcated fiber-optic. which could be used for simultaneous
transmittance and 180' fluorescence ,detection. The 0° and 180'
fluorescence signals will vary in significantly different ways as a
function of secondary absorbance. With proper modeling and curve
fitting. a secondary-absorbance function could likely be derived based
on the ratio of 0' and 180' fluorescence. Such a cell could be used
for primary- and secondary-absorbance corrected fluorescence on the

millisecond time scale.

100'

Excitation efficiency could also be improved by replacing the
current crude lamp and lens assembly with a collimated Eimac lamp and
achromatic beam condenser to direct a more intense. narrow beam into
the excitation monochromator. Computer-controlled slit programming of
all monochromators could be used to Ooptimize the slit width for
varying absorbance conditions - narrow at high absorbance. wide at low
absorbance.

Software development of the current instrument could also be done
to provide total calculation capabilities of absorbance values and
correction factors. as well as collection and transfer of data files
to the 11I23 for hardcopy or graphic output.

As with chemiluminescence. reemission errors are a dominating
limitation of absorbance-corrected luminescence methods. Development
of instrumentation which could correlate concentration and
luminescence signal. despite inner-filter or reemission effects. would
expand the number of applications of luminescence techniques
immensely. With the absorbance-corrected fluorescence flow cell. an
exponential dilution technique might prove useful for diagnosing or
evaluating reemission effects by comparison of the dilution vs
corrected fluorescence curves of negligible reemission solutions. to
the curves of solutions which are suspected of having reemission.

unfortunately. this fluorescence system has not been sufficiently
developed nor has time allowed for an investigation of several
analytical applicaitons of interest. This instrument is ideally
suited for kinetic. low volume. primary absorbance-corrected
fluorescence methods. Applications in clinical chemistry are of

interest. since several homogenous assays of blood plasma suffer from

101
high primary absorbance (69). As an example. many Homogenous Enzyme
lultiplied Immunoassay Technique (ElIT) (70.71) methods determine NADH
by its rate of formation in the presence of blood plasma. Since
absorbance detection is used. longer reaction times are required than
might be required for an absorbance-corrected fluorescence technique.
Fluorescence Immunoassay Analysis (FIA) (69) has similarly become
popular in clinical laboratories. but occasionally suffers
inner-filter effect errors. Other clinical methods could also benefit

from absorbance-corrected fluorescence instrumentation (72.73).

BIBLIOGRAPHY

(1)
(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9)

(10)

(11)

(l2)

(13)

(14)

(15)

BIBLIOGRAPHY

36116111111: 11. v. Argh, Negrl, 3611, 1889 3;. 416-427.

Rauhut. l. l. girk-Othmer EnczclI ChemI Teghnolngpgggl“t 1979 5.
416-450. l. Graysen and D. Echroth (Eds.). Wiley. New York.

Seitz. W. R.: Neary. l. P. AnalI Chgm, 1974 16. 188A-202A.
Cormier. l. J.; Hercules. D. l. 3 Lee. J.. Eds.. I

"Chemiluminescence and Bioluminescence." Plenum Press. New York.
N. 1.. 1973. f

 

Ericka. Larry J.; Carter. Timothy J. N.. Eds.. "Clinical and
Biochemical Luminescence". Clinical pp; Biochemigal ApalzsisI
v01.‘;;. Dekker. New York. 1982.

Seitz. W. R.; Neary. l. P. leghogs BipghemI Anal. 1976 21.
161-188.

Whitehead. Thomas P.; Ericka. Larry J.; Carter. Timothy J. N.;
Thorpe. Gary H. G. Clin, Chem, 1979 25. 1531-1546.

anpler. John E.; lulkerrin. lichael G.; Rich. 8. Edwin. Jr.
Clin, Chem, 1979 15. 1628-1634.

Van Dyke. Knox; Van Dyke. Cynthia; Udeinya. Joseph; Brister.
Calvin; Wilson. lark Clin, Cheg, 1979 25. 1655-1661.

larino. D. F.; Ingle. J. 0.. Jr. Apalzgica Clinic; Agta 1981 124.
23-30.

Christmann. D. R.: Crouch. S. R.; Holland. J. F.3Timnick. A.
Applp ChemI 1980 5;. 291-295.

Christmann. D. R.; Ph.D. Thesis. lichigan State University. East
Lansing. II. 1980.

Christmann. D. R.; Crouch. S. R.; Timnick. A. Appl; Chem, 1981
at 276.2 80 e

Stieg. Scott; Nieman. Timothy A. Anal, Chem, 1978 59. 401-404.

Abbott. Edwin A. "Flatland". Dover. New York. 1952.

(16) Cadillac Plastics and Chemical Co.. Flint. lI.

102

(17)
(18)
(19)
(20)
(21)

(22)

(23)
(24)
(25)
(26)
(27)
(28)
(29)

(30)

(31)

(32)

(33)
(34)
(35)
(36)

(37)

(38)

(39)

103
Silicon Rubber Sealant. Dow Corning Corporation. lidland. l1.
lodel Eu-721-11. GCA lcPherson. Acton. IA.
lodel H-lO.‘Instruments SA. letuchen. NJ.
Hoyt. S. D.; Ingle. J. D. mp. 239...; 1976 41. 232-234.
lodel 427. Reithley Instruments. Cleveland. OH.

Newcome. Bruce; Enke. C. G. . lichigan State university. private
communication.

Fisher Scientific Compay. Fairlawn. NJ.

Aldrich Chemical Company. lilwaukee. WI.

lallinokrodt. St. Louis. l0.

latheson Coleman and Bell. Norwood. OH.

G. Frederick Smith Chemical Co.. Columbus. OH.

LSI 11I23. Digital Equipment Corp.. laynard. lA.

Lee. John; Seliger. H. H. Phoppghgm, Phopopiol. 1970 1;. 247-258.

Pilipenko. A. T.; Barovski. V. A.; Ralininchenko. J. E. Eh; Apal,
Rhim, 1978 3;. 1880-1884.

Burdo. T. G.; Seitz. W. R. ApalII Chem, 1975. 11. 1639-1643.

lacDonald. A. ; Chan. R. W. ; Nieman. T. A. Apal Chem, 1979. 5;,
2077-2082.

Chang. C. A.; Patterson. H. A. ApalI ghpg‘ 1980 52, 653-656.
Rauhut. l. l. App; Chem, gpp‘ 1969 g, 80.

Scott. 6.; Seitz. W. R. Apalztiga Chin, Agta 1980 115. 221-228.
Dye. J. L.; Nicely..V. A. lp‘ghpgp Ed; 1971 18. 443-448.

Dye. J. L.; Nicely. V. A.: Tbhan. F.; Cochran. R.; Leckey. J..
"RINFIT4 - A General Non-Linear Curvefitting Program". 1977.
Chemistry Department. lichigan State University. E. Lansing. lI.

Bullock. D. 0.; Bunce. R. A.; Carter. T. J. N. Anpl,‘Lp;;p 1979
11. 841-854 e

Guilbault. G. G. "Practical Fluorescence. Theory. lethods. and
Techniques." Dekker. New York. 1973.

(40)

(41)

(42)
(43)

(44)

(45)

(46)

(47)

(48)

(49)

(50)
(51)
(52)
(53)

(54)

(55)

(56)

(57)

(58)

(59)

104

litchell. Douglas 0.; Garden. J. S. ; Aldous. R. l. Applp.§hpgp
1976 48. 2275-2277.

Tiffany; T. 0.; Burtis. C. A.; lailen. J. C.; Thackery. L. H.
Anal, ChemI 1973 15. 1716-1723.

Chen. R. F.; Hayes. J. E. Applp‘flipppppp 1965 11. 523-529.
71:11, l. 1.; 1.116, F. a. 51.1, 011-. 1977 52, 2054-2057.

Adamsons. Iarlis. l.S. Thesis. lichigan State university. East
Lansing. II. 1982.

Adamsons. Rarlis; Timnick. Andrew; Holland. John F.; Sell. John
E. Appl; Chem, 1982 54, 2186-2190.

Lloyd. J. B. F. 12511.1 1980 1_;. 97-109.

van Slageren. R.; den Boef. G.; van der Linden. W. E. Iplgpgg
1973 22. 501-512.

Smith. R. l.; Jackson. R. W.; Aldous. R. l. m, M 1977 12,.
2051-2053.

Tietz. Norbert W.. Ed. "Fundamentals of Clinical Chemistry."
Saunders. Philadelphia. PA. 1976.

laxlight Optical Waveguides. Inc.. Phoenix. Arizona.
Wilmad Glass Co.. Inc.. Buena. NJ.

lodel 91087. Farrand Optical Co.. Inc.. Valhalla. NT.

lodel CA-150. Illumination Industries. Inc.. Sunnyvale. CA.

Patton. Charles . Ph.D. Thesis. lichigan State university. E.
Lansing. II. 1982.

Stallings. lerian 8.; Schulman. Stephen G. AnalI Chin. Apt; 1975
38 483-486 0

Intel Corporation. Santa Clara. CA.

lOS/LSI Data Book. Advanced licro Devices. Inc.. Sunnyvale. CA.
1980.

lalmstadt. Howard V.; Enke. Christie 0.; Crouch. Stanley R.
"Electronics and Instrumentation for Scientists."
Benjamin/Cummings. lenlo Park. CA. 1981.

"Data Acquisition Component Handbook." Datel-Intersil. lansfield.
lA.

(60)

(61)

(62)

‘ (63)

(64)

(65)

(66)

(67)

(68)

(69)

(70)

(71)
(72)

(73)

105

"Data Acquisition Components and Subsystems." Analog Devices.
Norwood. IA. 1980.

"Linear Databook." National Semiconductor Corporation. Santa
Clara. CA. 1978.

"Silicon Photocells. PIN Silicon Photodiodes. GaAsP Photodiodes.”
Hamamatsu Corporation. liddlesex. NJ. 1980.

Brodie. Leo . Starting FORTH. Prentice-Hall. Now York. 1981.

Hoffman. Phillip. lichigan State University. E. Lansing. NI.
1982. -

Siegmund. Walter P.. ”Fiber Optics." in "Handbook of Optics."

Walter V. Driscoll. Ed.. chraw-Hill. Now York. 1978. 13-1 to
13-26 e

"Light-Conducting Fibres." Ipfogpption Fipre Qppigp 32; 1998112.
JENA GLASWERK SHOTT GEN.. lainz. West Germany.

Dr. Gale Harris. Associate Professor. Department of Radiology and
Department of Physics-Astronomy. personal communication. 1979.
lichigan State University. E. Lansing. lI.

lath. Irwin. "Simple fiber measurements.” Lpppp Focus 1979
Septepper. 75.

Soini. Erkki; Hemmila. Illka Clin, Chem, 1979 21, 353-361.

Scharpe. Simon L.; Cooremann. W. l.; Blomme. W. J.; Lackeman. G.
M. Olin, ChemI 1976 2;. 733-738.

Wisdom. G. Brian Clin, ChemI 1976 22. 1243-1255.
Froehlich. Peter Applp Spec, Rev, 1976 1;. 83-129.

Rubin. lartinquz‘ 91in, ghpgp 1970 11. 163-269.