‘17....lflmfl

nll' vo¢1of

uh. nt: 5".
g: 3'.

. .
l O, .
u... .KNLuuupnu. A
"u“uu “mmmmmnnu
..nnumnunnuu».

: '-
ér-‘s'g.

. E

E

ii!

a;
:4; ‘3
., 5!
.3299
o"‘ _

4
I
0
L
t

l
I
.
I

5'
.5;

1:. .
5.....scT-i
viti- .

rdof
.'
2:"
at!
.
l v:
3%
.i.

if!
S
'3
E?

O

. ‘\..A.
_ v. r 15:75.0 I

hi
w
v16
25'
{a
S

.I . ['1. .;l0 :5
4.:nll.

t‘::
‘l’ I ‘I
55‘. ' '
“2:531
5!» -:
g; :33.
.42 £pr
2:2§5,15§
Sggfgsiz.

I: z!
i
z'

I

,i
-l

3i
:
x
:
5

v

6.:
fling?!
v‘ v;' a!
5}I§:§5§§§

'
a

-
.-
?

v

otatl
‘I.’

It

v. .010- U...

lbpzl " 1|! a
.‘.CIA 91.

‘11,
lei—(YO. I, .
0: 11' a .

.53) -.

V. . .r
0‘13: .a .3.

‘o'v‘:
1.7.1 l 4.|vu0
a l... I la ,

 

 

. «ems

II/i/I/i/Lmiim 7W7:

meme».
1/ x z W
3 1 93 00795 7365

 

 

This is to certify that the

thesis entitled

Design and Construction of a Multichannel

detector for Flow Injection Analysis
presented by

Edwin Josue Castellanos

has been accepted towards fulfillment
of the requirements for

wdegree in Chemistry

 

Major professor

 

Date August 7, ]992

 

0-7639 MS U is an Affirmative Action/Equal Opportunity Institution

 

 

 

 

 

r-r “w
LIBRARY 1
i Memes State
University

'\ A

 

 

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

:J_____
Vii—ji—
L__JL_ t__._
:][_Z

L4G
mm

MSU Ie An Affirmative ActioNEquel Opportunity Inethuion

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

DESIGN AND CONSTRUCTION OF A MULTICHANNEL DETECTOR
FOR FLOW INJECTION ANALYSIS

37
Edwin J 09116 Caetellanoe

A THESIS

. Submitted to
Mic an State Univere!
in partial ent of the req mente
for the degree of

MASTER OF SCIENCE

Department of Chemistry

1992

ABSTRACT

DESIGN AND CONSTRUCTION OF A MULTICHANNEL DETECTOR
FOR FLOW INJECTION ANALYSIS

By

Edwin Josue Castellanos

A novel flow-through photometer suited for Flow Injection Analysis
was constructed and tested. The new detector features a multichannel
parallel design especially suited for multideterminations in flowing
systems. The simple photometer uses an LED as light source to avoid
the need for a wavelength selection device. A reference photodiode in a
feedback loop controls the forward current for the LED, assuring a
constant level of light. The observation cell in every channel is
constructed of Delrin and Kel-F and every component is removable.

The figures of merit for the new photometer are determined by the
type of LED used in a given determination. In the present thesis. the
detector was used in the enzymatic determination of sugars in food
samples using the leucomalachite green-H202 detection reaction. The
calibration curves were linear up to absorbances of 0.9. The deviation
from linearity was mainly caused by a polychromatic light effect. The
sensitivity of the detector to variations in the flowing liquid limited the
precision of the measurements. Still. the results were satisfactory when

the new detector was used in the glucose determination in apple juice.

To my wife Kelly Anne
and my family in Guatemala.
especially my mother.

 

 

 

 

 

ICSCEL

busy

inclu.
circu
and 1
leucc
Were

Some

insr

ACKNOWLEDGDIENTS

Special thanks to Dr. Crouch for his guidance thoroughout this
research and for his promptness in correcting the manuscript despite a
busy summer of '92.

People who were of great help in completing parts of this project
include: Marty Rabb. electronic designer. for his advice on improving the
circuits of the detector: Kris Kurtz for providing the reactors (both plain
and enzymatic) for the experiments; Pavlos Aspris for the hints on the
leucomalachite green reaction; Steve Medlin for his help when things
were not so smooth with computers; and Brett Quencer for recording
some of the LED spectra with a diode array detector.

The support of my wife, my parents-in-law. and my family was key
in succesfully completing this work.

iv

 

 

List 0

 

List 01

INTRt

CHAE

CHA‘

CHA

TABLE OF CONTENTS

List of Tables
List of Figures
INTRODUCTION

CHAPTER I: Photometric Detectors for Flow Injection Analysis
A. Flow Injection Analysis and its use with
other Techniques
B. LED-based Flow-through Photometric Detectors
C. Multidetermination and Multidetection

CHAPTER 11: Overview of the Detector
A. The Sugars Analyzer
B. The Multichannel Detector

CHAPTER 111: Design and Construction of the
Flow-through Photometer
A. The Photometric Cell
I . Choice of Materials
2. Design and Construction
3. Choice of Components

PAGE

12
12
16

23
23
23
25
3O

 

 

 

 

CHM

CI

Re

B. The System Electronics
1 . Light Regulation Electronics
2. Signal Processing Electronics

CHAPTER IV: Operation and Characterization of the Detector
A. Operation Procedures
1. Initial Operation
2. Adjusting the Light Level
B. Characteristics of the Detector
l . Sensitivity to Flow Variations
2. Baseline Noise and Drift
3. Peak Broadening
4. Dynamic Range
5. Precision of Measurements
6. Detection Limit
C. Practical Application

CHAPTER V: Conclusions and Recommendations

References

PAGE
36
36
39

42
42
42
43
45
47
52
56
58
8O
88
88

92

96

Table

Table

Table

 

Table

Table 2- 1

Table 4- 1

Table 4-2

Table 5- 1

LIST OF TABLES

Descri tion of Features Included in the
Multichannel Detector to Fulfill Specific
Requirements of the Sugar Analyzer

Results of the Experiment of Repeatability of
Measurements with the LED-Based and
Patton's Photometers

Relative Standard Deviations for the Triplicate
Eeadings of the Solutions in two Calibration
urves

Experimental Results for a Single Photometer
of the Multichannel Detector

PAGE
20

85

87

95

 

 

|
Figur

 

 

 

Figur
Figur
Figu;
Figm

FigU]

Figu-

F1 gu

Figu

Figu]

Fight

Figure 1-1

Figure 1-2
Figure 2- 1
Figure 2-2
Figure 2-3

Figure 2-4

Figure 3- 1

Figure 3-2

Figure 3-3

Figure 3-4

Figure 3-5

I“igure 3:6

LIST OF FIGURES

Possibilities for Multidetection and
Multidetermination using Flow Injection
Analysis.

Schematic diagrams of FIA manifolds for
obtaining sequential multidetection.

Parallel multichannel flow in ection analyzer
for enzymatic determination o sugars.

Enzymatic methods for the determination of
the six sugars.

Schematic diagram of the six channel
detector. Channel 2 is omitted for clarity.

Schematic dia am showing the different
components 0 the photometer in every
channel of the detector.

Construction of the hotometric cell used in
every channel of the etector (drawn to scale).

Main bog of the photometric cell made of
Dealt-1?. measurements in inches (drawn to
s e .

Other components of the photometric cell: a)

hotodiode socket (Delrin); b) LED socket
lrin); c) observation cell (Kel-F). All

measurements in inches (drawn to scale).

Absorption spectrum of malachite green, the
substance detected in the enzymatic
determination of sugars.

Emission spectrum of a ultrabright type LED
from Radio Shack.

Emission spectra of two available LED's for
the detection of the malachite green.

viii

PAGE

10

14

15

18

19

27

28

29

31

33

34

 

 

Figu:

Figm
Figux

Figur

 

 

 

 

Figux

Figui

Figu:

Fi gt
F1g\
F i g1

Fig

Figure 3-7

Figure 3-8
Figure 4- 1

figure 4-2

Figure 4—3

Figure 4-4

Figure 4-5

Figure 4-6

Figure 4-7
Figure 4-8
Figure 4-9

Figure 4- 10

Figure 4- 1 1

Light re%ilating circuit. The current
produced y the reference photodiode controls
and regulates the intensity of the LED.

Signal processing circuit.

Schematic diagram of the FIA manifold used
in the riments to characterize the
multichanne detector.

Injection of a 0.04 mM solution of hydrogen
peroxide. The noise in the signal came from
the pump oscillation. The flow rate was 1 .3
ml/ min.

Stabilization device used in the carrier
channel after the pump to reduce the efiect of
its oscillations.

Injection of a 0.04 mM solution of hydrogen
peroxide using two stabilization devices: one
in the carrier channel and one in the sample
channel. The small peak at 8 seconds was
caused by the injection process.

Injection of a 0.04 mM solution of hydro en
peroxide using a lar er observation cell. c
noise in the si came from the pump
oscillation. The ow rate was 1.3 ml/min.

Injection of a 0.04 mM solution of hydrogen
peroxide using a larger observation cell and
one stabilization device in the carrier channel.

Dis rsion patterns for both detectors at low
an high sample concentrations.

Calibration curve using LMG-h ogen
peroxide with stock solution of LMG 1 . mM.

Calibration curve using LMG-hydrogen
peroxide with stock solution of LMG 3.0 mM.

Calibration curve using LMG-hydrogen
peroxide with stock solution of LMG 1 .5 mM
using two sizes of observation cells.

Calibration curve using LMG-h drogen
oxide with stock solution of LMG .0 mM.
e LED detector was used with the three
available LED's.

PAGE
37

40
46

48

50

51

53

54

57

6O

62

66

 

 

Figm

Figui

Figur

 

Figur
Figu

Figm
Figm

Figm

Figu:

Figu

F1 gt

F1 g1

Figure 4- 12

Figure 4- 13

Figure 4- 14
Figure 4- 15
Figure 4- 16
Figure 4- 17
Figure 4- 18

Figure 4- 19

Figure 4-20

Figure 4-2 1

Figure 4-22

Figure 4-23

Calibration curve using LMG-h drogen

eroxide with stock solution of LMG l .0 mM.

e LED at 610 nm was used in Patton's
detector with and without a filter.

Calibration curve using LMG-h drogen
roxide with stock solution of LMG l .0 mM.
e LED detector was used with LED forward
currents of 10 and 30 mA. .

Emission spectra for the LED at 6 10 with low
and high forward currents.

Absorption spectrum of methylene blue in
water.

Calibration curve using methylene blue at 610
and 660 nm.

FIA peaks observed with the calibration curve
using prepared malachite green.

Calibration curve using malachite green
previously prepared.

9 injections of a 0.04 mM solution of H 0
using the LED-based detector. The R88 0%
the measured absorbance is 2.63%.

9 injections of a 0.04 mM solution of H202
using Patton's detector. The RSD of the
measured absorbance is 1 .62%.

8 injections of a 0.08 mM solution of H 0
using the LED-based detector. The R513 0%
the measured absorbance is 2.25%.

8 injections of a 0.08 mM solution of H202
using Patton's detector. The RSD of the
measured absorbance is 1.95%.

Calibration curve for the glucose content
determination in apple juice.

PAGE
69

71

73

74

76

78

79

81

82

83

84

90

INTRODUCTION

Analytical chemists are always in the search of simple instruments
that can replace tedious manual methods or long instrumental
determinations and perform in a faster and more reliable way. Flow
Injection Analysis (FIA) has proven to be very effective in this sense,
making the handling of liquid samples an easier task.

The determination of sugars in food samples is an example of an
analytical procedure that is not always suited to the changes and needs
of modern industry. The ofi‘icial method of analysis dictated by the
Association of Analytical Chemists is a non-selective volumetric method
that can only determine the total amount of sugars present [1] but not
the individual concentrations of specific sugars. To accomplish the
latter. the use of High Performance Liquid Chromatography (HPLC) is
recommended, a procedure which can be very elaborate and time
consuming.

The work presented in this thesis is part of a larger research
project in Dr. Crouch's group which aims at the construction of a novel
analyzer for sugars in food samples based on immobilized enzymes and
FIA principles. The system will provide a fast and simple way of
determining up to six nutritionally important sugars in food samples
with complex matrices without prior separations. Work on the project
has been going on for several years already. with various researchers
focusing on different aspects, such as immobilization processes. sample
preparation, and construction of the FIA manifold [2-6].

1

2

My part in the project was to construct a simple and inexpensive
detector with a parallel design; that is. a detector able to monitor six
difi'erent flowing streams simultaneously. The determinations are based
on photometric measurements using LED's as light sources.

A goal of the research was to build a detector suited to the needs of
the sugars analyzer but applicable also to a FIA system in general.

The sequence of the chapters follows the way the project was
completed. The first chapter reviews the work done in the area by other
researchers. whose ideas gave origin to my project. In the second
chapter. the detector is presented as part of the complete sugars analyzer
to show how the characteristics of the final instrument influence the
requirements for the detector. Then, in the third chapter. a more
detailed description of the detector is included beginning with the design
process of the flow-cell. This chapter ends with a discussion of the
detector electronics. Chapter IV presents the experiments performed in
order to fully characterize the new detector and to compare it to the one
currently in use. The thesis ends in Chapter V with conclusions and
recommendations concerning how to attain an optimum performance

from the detector.

 

siir
kee
wit

pill

Six-

the
dis

Clix
Silt
Val
to

him

CHAPTER I
PHOTOMETRIC DETECTORS FOR FLOW INJECTION ANALYSIS

One of the most attractive characteristics about FIA is the
simplicity of the instrumentation needed. Much efi'ort has gone into
keeping the system simple. including the detector. A simple photometer
with a LED as the light source has provided excellent results for this
purpose. This chapter reviews the work done to construct such flow-
through photometers. work that was used to develop the ideas for the
six-channel detector described in this thesis. A brief presentation of FIA
and its many applications with different analytical techniques precedes
the review as a means of introduction. The chapter concludes with a
discussion of the different possibilities in FIA for a multidetermination of
analytes in a sample.

A. FLOW INJECTION ANALYSIS AND ITS USE WITH OTHER

TECHNIQUES

Flow Injection Analysis (FIA) has developed throughout the past
decade as an important method of analysis due to its versatility and
simplicity [7.8]. The basic components of a FIA system (pump. injection
valve. reaction coil. and detector) can be arranged in many different ways
to adapt to applications in such varied fields as agriculture.
biochemistry. food science. and clinical chemistry. to name but a few.

3

U

 

 

to

ca

521
ha
wt
clr

C85

an:

W0:

list

(en

tire

lee]
Tea:

aha

4
Many review articles in the different areas of application are available in
the literature as well as three monographs [7-14]. The reader is referred
to those sources for a comprehensive description of the available
methodologr.

As noted by Poppe [15]. the definition of a detection system in FIA
can be ambiguous. In many cases the FIA system is much simpler than
the detection system. so that it is no longer valid to talk of a detector for
FIA: rather. it is more accurate to say that the FIA apparatus serves as a
sample introduction system for the more complex instrument. It also
happens. with techniques such as HPLC. that the detection device is the
whole flow injection system itself. with the sample being delivered by the
chromatographic separation unit instead of being injected [16]. In any
case. FIA systems have made easier the task of sample handling before a
final instrumental measurement.

FIA has been used in conjunction with a variety of detectors and
analytical techniques. It would be impossible in the short space of this
work to review even briefly the different methods that may be used in
FIA. Just to give an idea. the monograph by Ruzicka and Hansen [7]
lists seventeen spectrometric methods. eight electrochemical methods.
nine separation methods. eleven gradient techniques. kinetic methods
(enzymatic and non-enzymatic). plus a variety of less common
techniques such as enthalpimetry and viscosimetry.

Spectrophotometry is by far the most widely used detection
technique. employed in more than 40% of the published papers [7]. A
reason for this is the fact that spectrophotometry is the most common
analytical technique. so there are methods available to determine almost
any type of analyte. Also. it is very easy to adapt a spectrophotometer to

5
a FIA system provided that its optical components produce a light beam
small enough in diameter to be sent through flow cells of small
apertures. A flow cell of the Hellma type is commonly used. but the so-
called Z cell is also very well-suited [7].

Two optical techniques that are gaining popularity in the FIA field
are atomic absorption (AAS) and inductively coupled plasma
spectrometry (ICP). FIA has proven to be an excellent injection system
for these instruments. increasing their sensitivity over 100 times [1 7. 18].

Apart from the optical techniques. electrochemical methods have
also been coupled with success to FIA systems. and research is directed
now to the miniaturization of electro-sensors and the use of ion-selective

field effect transistors (ISFET‘s) [19].

B. LED-BASED FLOW-THROUGH PHOTOMETRIC DETECTORS

In a conventional design of a FIA system using spectrophotometric
detection. the flowing stream with the sample has to be taken back and
forth from the FIA manifold into the spectrophotometer adapted with a
flow-through cell. Conceptually. it is better to introduce the detector into
the manifold system rather than to pump the carrier stream into the
detector. thus avoiding extra tubing and higher dispersion. If the use of
a spectrophotometer is necessary. this can be accomplished by using
fiber optics to transport the light to the manifold [20].

In many cases. though. a full spectrophotometer is not needed. If
the FIA system is to be used only on a specific chemical determination.
the wavelength selection capability of a spectrophotometer becomes
superfluous. and the measurements can be made with a simple

photometer. If a light source with a narrow emission band is used. a

. more

 

 

   

 

6
filter to isolate a given wavelength is not needed. which simplifies the
photometer even further.

A light emitting diode (LED) emits light with a bandwidth ranging
from 20 to 50 nm which is monochromatic enough for the purposes of
most FIA determinations. A photometric measurement in the visible
region can thus be carried out using the LED as the light source and a
photodiode or phototransistor as the light detector.

The replacement of the tungsten lamp by a LED for photometric
determinations in the visible region is nowadays more feasible with the
commercial availability of ultra-bright LED's. This special type of LED
can reach luminous intensities as high as 5 candelas (about 1000 times
brighter than normal LED's). The brightest LED's available emit at 660
nm. the natural wavelength of emission for the Gallium Phosphide pn
junction. Doped Gallium Phosphide LED's are available at lower
wavelengths and lower brightnesses. typically of 0.3 candelas [2 1].

An advantage of LED's over conventional tungsten lamps is their
low power consumption. LED's typically dissipate 75 mW. compared to 3
to 5 W for miniature tungsten lamps. Despite using considerably less
power. LED's can still attain high luminous intensities due to narrow
bands of emission and high luminous efficacies.

The first applications of LED-based photometers for analytical
purposes were to static systems [22.23]. Betteridge et al. [24] applied the
idea to design a flow-through LED-based photometric detector for FIA.
The detector was extremely simple and inexpensive. having the LED and
phototransistor glued in a block of plastic with holes drilled in it to form
the flow-cell. Despite its simplicity. the detector proved to have an
adequate level of sensitivity and detection limit.

7

Sly and co-workers [25]. from the Betteridge laboratories. improved
the original detector by designing a more flexible circuit that allowed
different gains and LED currents. and by placing the LED and
phototransistor perpendicular to the flowing stream and outside the
carrying tube. The detector showed better resolution. but higher
detection limits. The sensitivity was maintained by the improved
amplification circuit.

A variation of this principle was implemented by Hooley and Dessy
[26] in a system with a multi-LED detector for kinetic studies in flow
systems. They combined the ideas of Sly [24] and Anfalt [23] et al. to
construct a detector with the light path perpendicular to the flowing
stream. and with improved circuitry: the light source was modulated to
reduce the efi'ects of ambient light. and a reference photodiode was
added in a feedback loop to monitor the light from the LED and keep its
intensity constant. A total of eight detection heads were used to monitor
the reaction mixture at different positions along the manifold.

Still another modification to the original detector of Betteridge et
al. was made by '11-ojanowicz et al. [27]. The detector built by this group
had removable holders for the LED and the photodiode. making their
replacement an easy task. This is an important feature since different
LED's have to be used for different determinations.

C. MULTEETERMINATION AND MULTIDETECTION
The flexibility of a FIA system offers the choice of several solutions
for a given problem. A good example is found in multideterminations
(the determination of two or more analytes in a sample) such as
determining six sugars in food samples. The first decision is whether the
analysis will be sequential. determining n analytes from n sequential

8
injections of the same sample. or simultaneous, doing the entire analysis
from a single injection. The simultaneous choice calls for multidetection.

Multidetection. as defined by Luque de Castro and Valcarcel [28].
refers to obtaining two or more signals from a single injected sample. To
accomplish this. the options are several [29.30]; a summary of them is
presented in Figure 1-1. From the study of Figure 1-1 and from the
preceding discussion. it can be deduced that a multidetermination can
be achieved by multidetection. but multidetection does not necessarily
produce a multidetermination. An example of the second situation is
found in the use of FIA for kinetic studies. where several signals at
different times are obtained from the same chemical system.

Figure 1-2 shows examples of FIA manifolds to obtain sequential
multidetection. The arrangement of several detectors in series (Figure 1-
2 c) is most adequate for kinetic measurements. For other types of
measurements. the choice of a single detector (Figure 1-2 a.b) or several
detectors in parallel (Figure 1-2 d.e) depends on the nature of the
chemical composition of the individual channels. If the systems are
compatible. they can be mixed at the end. and only one detector is
needed for the determination; otherwise. several detectors in parallel are
required. For our system. the option of six detectors in parallel is
preferred not for the incompatibility of chemical systems (all the
channels will carry the same reagents with slightly different pH's). but
rather for time reasons: waiting for six peaks to elute through one
detector would substantially increase the time of analysis.

A single injection (Figure 1-2 (1) will also help to keep the time of
analysis short. Low reproducibility in equally splitting a small sample

Multideterminaiion

- |
| l
Sequential Simultaneous __, Multidetection
- |
| |
Secuential Simultaneous
|
I l

Single Detector Several Detectors

In Series In Parallel
(kinetic studies) (Sugars analyzer)

 

 

(diode array)

 

Figure 1-1: Possibilities for Multidetection and Multidetermination
using Flow Injection Analisis.

10

 

 

Different—(cuter lengths
, produce mum-553ml in. & Single.

 

w detector (merging Lone)

 

Same. as a) but with

 

 

 

 

 

 

 

5
VI!
w.
P
b) 5
\FI *
R. 7 "
lag-">1 5:
P -g

. l l
‘ <12
1
e-tns
(<7
‘ ‘
S

 

_V —__h>l ‘muliltinjwution

 

 

 

Detectors in Series

 

 

tor ‘kin‘t‘tic’ShldieS‘

<7_

 

EN)!

 

 

 

 

 

—.—— _—_-.._----_ ,

Detectors in parallel

 

 

 

 

 

‘6‘ W
P" M“ “’mih‘sragh’splriimg ‘ ‘

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

_ a, \
-c ,X
R; 7 w of the Sample _ -__.-
we)... .... f ._ .__, ____. _ _.
R. ‘ . @‘1 TWO injections With
—-~———— ~-- -2. immune -
“2 ... 9 W
_-_..- __- g ____. ____.--__. _ ___ __ -
Hf) __ - Axum _____‘ _. ,
"'" Jr .n. Similar to d) bid With
at twain channel selector

 

 

 

 

 

 

4 s
-c_+___6__g_€r;ia
W for directing “16 Sample

 

C‘ ‘Cmier R ‘RtC-jent P’Pump 5‘ scrapie. D’uodector Wu Waste

8 of FIA manifolds for

Figure l -:2 Schematic
obtaining sequential multidetection.

1 1
into six portions could be problematic: a simultaneous multiple injection
may avoid this situation (Figure 1-2 e. i).

The sugars analyzer under construction in our laboratories. will
then. make a simultaneous multidetermination of sugars in food samples
from a single injection. or at least from a simultaneous multi-injection.
and the detection will be sequential with six detectors in parallel.

 

 

 

 

ot‘
ad

Sll

dc

go

Wt

Was
Pat:
Dies
anal

ofu

CHAPTER II
OVERVIEW OF THE DETECTOR

In the first part of this chapter. a brief description of the sugars
analyzer under construction in our laboratories gives background
information for introducing the multichannel detector. The ideas from
other researchers for LED-based detectors. presented in Chapter I. were
adapted to the specific detection requirements of the multichannel
sugars analyzer. This chapter presents a study of those requirements
and gives a general overview of the features of the detector. A more
detailed description of its design and construction is given in Chapter III.

A. THE SUGARS ANALYZER

Work on difi'erent aspects of continuous flow analysis has been
going on for over a decade in our group. so it is difficult to define exactly
when the first research concerning the sugars analyzer was done.

The idea of determining six sugars in food samples simultaneously
was first suggested by Stults [4) who used a FIA system designed by
Patton [3]. Previously. Thompson [2] had determined glucose using FIA.
These are probably the first projects closely related to the FIA sugars
analyzer as it is being currently used.

With the final version of the FIA sugars analyzer. the determination
of up to six sugars in food samples will be simpler and faster (as

12

13

compared to traditional methods such as HPLC) for two main reasons:
first. the highly parallel design of the system. and second. the elimination
of any prior separations. since the selectivity for each sugar is achieved
by the use of specific enzymes. An ultimate goal is to produce a
miniature analyzer that operates on battery power so that it can be taken
into the field for speedy determination of sugars in. for example. fruit
juices. in order to determine the optimum harvest time.

Stults [4.3 I] worked on the optimization of a single channel
glucose analyzer using immobilized glucose oxidase. It was suggested
later that this concept could be extended to make a parallel multichannel
sugars analyzer. as shown in Figure 2- 1. Within this scheme. each
sugar is converted to its oxidized form by the action of the enzyme
reactor containing the corresponding oxidase with the release of
hydrogen peroxide. Figure 2-2 shows the enzymatic reactions used to
determine the six sugars. Kurtz worked on the immobilization
procedures for the enzymes involved [5].

The H202 released by the oxidation reaction of sugars can be
detected by reacting it with a dye in the presence of horseradish
peroxidase. The change in color in the oxidized dye is detected
colorimetrically. Therefore. the simultaneous determination of the six
sugars can be accomplished by detecting the appearance of a single
colored substance: an oxidized dye.

There are several reactions involving the oxidation of a dye by
hydrogen peroxide. The Trinder reaction [32].

H202 + AAP + DCPS 355939918; red dye + H2804

l4

.muoflnm no :ofluecwaueuou awuoaauco non
nonmaoco newuoewcw soda “encasemenoa Heauouom firm ououmm

ES.—

    

«83:5
.aouauz
a:
=t=u

honor—co:

        

  

3333.
as... she

 

)QFII

       

a

A
o

m.
2

     
 

stag

 

a... sinus.

:

- L, ,_7

—. .59. ;

 

 

 

 

 

 

 

 

15

.muomnm me ecu
mo :ofluonBuouoU on» you moozuofi cauofihwcm NIN ouomwm

 

QmOuoflhulQ + mmoofluwlfl ill 2.9382: 03020

“5305le

omens—minis + 3005319 1 cum + $322

 

.l 03330ng n

 

omoaoflowloln + mmoogwla A oSEnSoearn

 

 

amouoabla + mucus—win I A cam + omocoom

3323:: 33.8.2:

Noam + umoEScoNonlgoflamla i

loeaguo 0.32.30

 

 

Noam + ecoaoflieloaoogmin A ONE + No +

oneExo euoogo

cam + N6 + 330338

omooEMIQ

Cum + $303

‘

16
where AAP = 4-aminoantipyrine and DCPS = 3.5-dichloro-2-
hydroxyphenyl sulfonic acid. was the first option investigated. Later. the
oxidation of the leucomalachite green (LMG) reaction to form the highly
colored malachite green (MG):

H202 + LMG -£9£9§93§6MG + H20

was found to give higher sensitivity and lower detection limits. The
reaction was optimized by Aspris [6] who investigated the application of
the analyzer to real samples.

All previous research in our laboratories has been done with a FIA
system designed by Patton [3]. This system uses a simple dual-beam
photometer [33] to monitor the formation of the colored dye. The
photometer uses a miniature tungsten-halogen lamp as light source and
a bifurcated fiber optic that serves two functions: to split the beam to
produce a reference: and to transport the main beam to the observation
cell. This cell is a commercial flowcell (PN 178- 13724-02. Technicon
Instruments; Tarrytown. NY) with sapphire windows. 1 .0 cm pathlength.
and 0.05 cm internal diameter. The wavelength of interest is isolated by
interference filters that can be exchanged to suit the particular
application. This type of detector has been applied only as a
single-channel analyzer.

B. THE MULTICHANNEL DETECTOR
From the above discussion of the sugars analyzer. it can be seen
that the instrument is a typical example of the case discussed in Chapter
I in which a full spectrophotometer serving as detector is unnecessary.
No wavelength selection capability is needed since the species to be
determined is always the same.

1 7

The concept of a LED-based photometer was applied to construct
the six-channel detector schematically shown in Figure 2-3. Basically.
the instrument consists of six independent. LED-based. flow-through
photometers such as the one in Figure 2-4 placed in parallel. Detailed
drawings and specifications describing materials and components.
construction. and dimensions are found in chapters III and IV. The
remainder of this chapter is dedicated to an overall description of the
various features of the detector.

Table 2- 1 summarizes the general requirements for the detector. as
well as the features included in the instrument that satisfy these
requirements.

Each one of the individual photometers composing the
multichannel detector (Figure 2-3) is seen to have three main parts: the
measuring cell. the light regulating electronics. and the sigial processing
electronics.

The measuring cell (Figure 2-4) is made of black plastic and
houses the observation cell which consists of a cylinder made of Kel-F
with holes for the flowing stream drilled in it. The side of the observation
cell facing the LED has a lens to focus the light. and the other side
features a glass window. This cell is connected to the rest of the FIA
system by stainless steel tubes. The measuring cell has two removable
holders for the LED and two photodiodes. one of them looking at the tip
of the LED and the other one at the other side of the observation cell.

The light level from the LED is kept constant by means of the
photodiode next to it. The photocurrent produced by this reference

photodiode is converted to a voltage and then compared to a reference

18

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

   

 

 

 

 

 

 

 

 

   
 
 

OWuT Z
(en mac)

 

 

auteur j

 

 

 

 

 

 

 

 

METAL (A58

nun. um:
fl Asleeuuequuwb

110V PowEIL
60“! SUFF-

Figure2- -:3 Schemati tiocdiag‘am oefth sixchannel
detector. Channel 2 is omitted for clarity.

19

.6838 05 .«o Snooze
E96 5 Louofiouonaamwmmm Samson—Boo
Sachem—o 05 955.7. E 238058 "TN Paw—m

 

 

 

 

 

 

 

 

 

 

code—0,300.50 QUE—0.30.30
unannouoa «3.52m .200 OahaIIOuosm Osmond—ah: usomfl
OOQudo>
332m eoceuevec
housfidg
.Afiwuo
Q I
#05080
:00 39
\fl \ T808 39.5. 83.388 23 «:23 e3-
Ouozl

 

 

 

 

 

 

 

 

 

 

 

 

«3.3$|llA/_ 1N)... U = C m , m

 

 

 

 

 

 

 

 

 

 

 

 

 

 

>Al~\r'

gag \

9405."... \
(um OH

 

 

 

 

 

 

 

 

 

 

 

20

Table 2- 1:

Description of Features Included in the Multichannel Detector to Fulfill

 

 

 

Specific Requirements
for the Sugar Analyzer
Requirement Feature
of system: of detector:
1 - Fixed Wavelength Easily done with LED as light
source

2- Simple

3- Inexpensive
4- Low energr requirement
5- Portable

6- Replaceable parts

7- Rapid Response

8- Adjustable gain

9- Broad working range.
adequate sensitivity
and detection limit.

Use of LED eliminates need for
monochromator and / or filters

Can be constructed in-house
LED's suitable for battery operation

Weight less than 5 lb. (without
power supply); small size

All active components in removable
holders

Six sigrals can be monitored
simultaneously. User can choose
number of channels to use

Gain adjustable from 1 to 60
Parameters adjusted by

selecting different LED's or
by changing driving currents

2 1
voltage. The output of this comparator controls the amount of current
supplied to the LED, and will vary it to counteract any light variations.

The current from the photodiode across the observation cell. which
is linearly related to the transmittance of the liquid in the cell. is
converted to voltage and then amplified with a variable gain amplifier.
This gives the final output sigral of the detector.

As previously mentioned. using LED's as light sources is an
excellent alternative when no wavelength selection capability is needed.
In the multichannel detector. the LED is easily removed. thus providing
additional flexibility in choosing the wavelength of analysis. High
brightness LED's with peak emissions at various wavelengths in the
visible spectrum are available from difi'erent manufacturers for the
determination of analytes absorbing visible light.

The use of LED's greatly simplifies the photometer and eliminates
wavelength isolation components. which in turn reduces the price of the
detector. Likewise. a reduction in cost is achieved by replacing the
commercial miniature flow-cell. with prices as high as $200. with a flow-
cell constructed from a piece of plastic material with the necessary holes
drilled in it. Keeping the costs of individual detectors low is important
when six of them are going to be used.

Apart from simplifying the instrument. LED's require considerably
less power than tungsten lamps. This is an important factor to consider
if the instrument is operated on batteries. Battery operation capability is
a desirable feature in an instrument that might be taken to the field.
Light weight (less than 5 pounds without power supply). and small
dimensions (the six detectors are enclosed in a 12"x15" case) also

contribute to the portability of the detector.

22

The desigi and construction of the instrument provide easy
revision and maintenance of the different parts. All the active electronic
components are easily removable. as are the parts in the plastic
observation cell.

With respect to output characteristics. the parallel desigi for the
six channels will keep the time of analysis per sample similar to that for
the determination of a single sugar. The output level is adjustable by
means of a variable gain amplifier to compensate for solutions of different
absorbances.

The three important characteristics for a detector —linear range.
sensitivity. and detection limit— may be adjusted by using different
LED's and by varying the current driving them. Using the brightest LED
available. with peak emission at 660 nm and narrower bandpass. will
give the broadest dynamic ranges. but for the detection of malachite
green. with a peak absorption at 618 nm. the sensitivity will be low and
the detection limit subsequently high. The use of a LED with peak
emission closer to 618 nm with lower brightness and broad bandpass
will increase the sensitivity and lower the detection limit at the expense
of reducing the dynamic range. The brightness of the LED can be
increased by increasing the current driving it. but not without two
drawbacks: increased emission bandwidth and reduced lifetime.

CHAPTER III
DESIGN AND CONSTRUCTION OF THE FLOW-THROUGH
PHOTOMETER

We saw in the previous chapter that the multichannel detector is
simply a set of six independent photometers operated simultaneously.
This chapter includes a detailed description of the design and
construction of one of these photometers. This work was completed in
two phases: construction of the photometric cell and construction of the
electronics. This chapter is divided accordingly in two main sections.

A. THE PHOTOMETRIC CELL
The construction of the body of the calorimeter took most of the
time of the project primarily because of difficulties in machining the
difi'erent pieces. TWO designs were discarded before any tests were done
on them because of problems in construction. The observation cell

presented a special problem because of its small dimensions.

1. Choice of Materials
Factors taken into consideration in choosing the materials
included availability. mechanical and optical characteristics. and
resistance to chemical attack. Plastic materials have been the choice in

23

24
the construction of previous LED-based photometric flow-cells due to
their chemical resistance and light weight.

Trojanowicz et al. [27] used Teflon in their cell. This plastic is well-
suited for its opacity and high resistance to chemical attack. but it is
difficult to machine. especially when considering the dimensions needed
to construct a miniature flow-cell.

Plexiglass was also considered since it has been used in the
construction of components for FIA manifolds. This plastic shows
excellent mechanical and chemical properties but its transparency would
present problems concerning stray light.

The material chosen was black Delrin. a plastic from DuPont with
good mechanical properties and low cost. Delrin is readily attacked by
solutions of extreme pH (acid or base) [34], so it can only be used in the
parts of the colorimeter that will not be in direct contact with any
solution. Although the determination of sugars using the enzymatic
method discussed in the preceding chapter only involves solutions of
moderate pH. and therefore using Delrin in the composition of the cell
would pose no problem. using this material would limit the applicability
of the detector to other systems. Another problem arises regarding the
cleaning of the FIA system. which is usually done after a long period of
use by flushing with a 10% solution of nitric acid.

As will be seen later in this chapter. the only part of the
calorimeter in direct contact with the flowing solution is the observation
flow-cell. This component was made of Kel-F. a white plastic from 3M
with superior mechanical properties and great resistance to attack from
highly corrosive chemicals [35]. The material was found to be somewhat
transparent to light so the sides of the flow-cell had to be painted black.

25
2. Design and Construction

The main idea in the design of the body of the photometer is to
have all the components be replaceable. Although this makes the
assembly process slightly more complicated. the machining of the
individual pieces is greatly simplified. This proved to be very important
in the construction in one of the earlier designs. That design did not
feature the flow-cell as a separate part: the holes necessary to make the
cell were drilled directly into the main block of the detector. This
presented two problems. The walls of the flow-cell had to be of the same
material as the entire colorimeter body. in this case Delrin. This would
have limited the use of the detector to systems with no extreme pH as
discussed earlier. Secondly. it was extremely difi‘icult to drill the
miniature holes accurately (1/40" id) to make the flow cell. and if an
error were made the entire block had to be discarded.

A second factor that affected the final design of the detector was
the implementation of a reference channel to regulate the amount of light
from the LED. An earlier design attempted to use a rudimentary beam
splitter like the one reported by Hooley and Dessy [26]. It consists of a
glass rod with polished sides. with one of the ends cut at a 45’ angle.
This "beam splitter" is actually a reflecting device that produces a 90'
deviation. and the splitting of the beam comes from imperfections at the
surface that scatter some of the light. Most of the light is then
perpendicularly directed from the LED. In the design by Hooley and
Dessy [26] the main photodetector was located at a 90' angle from the
LED. thus receiving most of the light. A similar configuration was
Moult to implement in the new LED detector. so this beam splitter
design could not be used.

26

The final design of the colorimeter is illustrated in Figures 3- 1. 3-2.
and 3-3. The observation cell (Number 2 in Figure 3-1. also Figure 3-3 c)
is a separate component from the body of the colorimeter. It is made of
Kel-F and is connected to the rest of the FIA system by stainless steel
tubes that also hold it into place. It was already mentioned that the
sides of the cell facing the LED and photodiode were painted black. The
light path in the cell is 11.2 mm (0.44") long with an id of 0.635 mm
(0.025"). This gives a hold up volume of 3.54 pl. These dimensions are
similar to those of the commercial cell in the detector currently used [3].
which has proven to be very adequate.

The side walls of the observation cell were formed by pressing a
lens and a window against the sides of the plastic cell (Figure 3- 1). This
design gives rise to two dead spaces to the sides of the inlet and outlet of
the flow-cell (figure 3-3 c) because these holes are not placed right at the
end of the observation hole. If the inlet and outlet holes were drilled at
the end of the observation hole. the area that the lens and window had to
seal would be much larger. thus making the cell more subject to leakage.
These dead spaces produce some problems with the flow patterns of the
FIA system as discussed in the next chapter.

The body of the cell (Figure 3-2) as well as the holders for the
photodiodes and LED (Figure 3-3 a. b) are made of black Delrin which
prevents ambient light from reaching the photodetectors. The two
holders are tightened simultaneously to balance the pressure on the lens
and the window at the sides of the cell. This pressure has to be strong

enough to produce a leak-proof seal. but not so excessive as to break the

glass components.

27'

(\Ol

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

65 65 6565625 65 6:5: 6)

I. Cell body

2. Removable observotion cell

3. Stainless steel connecting tubes
4. Lens

5. LED socket

6. Reference photodiode

7. LED

8. Gloss window

9..Pbotodiode socket

IO: Pbotodiode

FIGURE '_3'l: Construction of the

Photometric Cell used in every
channel of the detector-

Front Back

 

 

 

0.48

Side

 

FIGURE 3'3:
a) Pnotodiode

Front

(b)

Front

(c)

Diner
socket

O- 625

0.2

0.4

Back

O.l2

 

 

004‘

Enlargement of the Observation
Chamber. (Scale: 41H

ports at tire calorimeter:
(Delrin); blLED socket

(Delrin); c) Observation cell. All measurements
(Drawn to scale)-

:29

 

 

 

 

 

 

 

 

 

 

 

 

'1?
O o I I2
L L5 i-fl—
Plan
0.3l
{—1
“mm mm
~L~H~

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

L 0.5 L 0.5 L 0.54

Longitudinal Section

 

 

 

 

 

 

 

  

 

 

Isometric

 

L I. 0.625 (J-

 

 

 

|.|2

7

Front

 

I'm (J
Li?!

 

 

 

Cross Section

FIGURE 3‘2: Cell Body mode of
Delrin. All measurements in inches.

'(Dravrn to scale)

30

The use of a beam splitter in the LED to produce the reference
signal was avoided by placing the reference photodiode so that it views
the tip of the LED. The air-plastic interface at this point scatters enough
light to be detected by the photodiode to produce the reference signal
that is used to control the intensity of the LED. No glue is used to hold
the photodiodes and LED in place. so they can be easily removed and
replaced if necessary.

3. Choice of Components

1. LED:

Several electro-optical characteristics must be taken into
consideration when choosing the appropriate LED. the most important
being the peak emission wavelength. which should be as close as
possible to the wavelength of maximum absorptivity of the analyte. Also
important is the spectral line halfwidth which will determine. among
other things. the linear range of the calibration curve. The luminous
intensity is an important factor to consider given the small internal
diameter of the observation cell, which greatly restricts the amount of
light reaching the photodetector. For the same reason. it is important to
have a small beam divergence. which is measured as the "half-value”
angle. the ofi-axis angle at which the luminous intensity is half the axial
luminous intensity [36]. The last factor to be considered is the peak
forward current. which is the maximum forward current that can be
applied to the diode to give the highest luminous intensity without
burning it.

In the enzymatic determination of sugars (see discussion on page
16). the substance to be detected. malachite green. shows a maximum
absorption at 6 18 nm (Figure 3-4). Four types of LED's were considered

31

 

 

 

l l 1 1 l l J

:00 52's 550 57 600 625 650 675 700
WAVELENGTH (nm)

 

Figure 3-4: Absorption spectrum of malachite green. the
substance detected in the enzymatic determi-
nation of sugars.

32
as possible light sources for this wavelength: two from Hewlett Packard
with maxima at as 585 and 640 nm (part numbers HLMP-3850 and 3750
respectively): one from Sharp Electronics Corporation with a maximum
at ~ 610 nm (part number GL5HS40); and one from Radio Shack with a
maximum at as 660 nm (ultra-bright type). The spectrum of the LED
from Radio Shack and typical spectra from the other LED's considered
are shown in Figures 3-5 and 3-6. The ultrabright LED from Radio
Shack is ten times brighter than the other LED's. and the spectral line
halfwidth is 15 nm narrower. It would be the first choice if the overlap
with the absorption spectrum of malachite green were not so poor.
Nevertheless. it was considered as a possibility for experiments in which
a large linear range is required. and in which sensitivity is not critical.
worn the remaining three LED's, the one from Sharp had the best
overlap with the absorption spectrum of malachite green. but it also
showed the lowest luminous intensity.

The light throughput of the two light sources of the detectors used
in the sugars analyzer were compared by measuring the photocurrent
produced by the two optical systems: a miniature tungsten lamp.
bifurcated fiber optic and bandpass filter. the setup in Patton's detector
[3]. and a LED placed right in front of the glass flowcell with a forward
current of 20 mA. a setup similar to the LED detector. The LED from
Sharp gave photocurrents of the same magnitude as those of the
tungsten lamp setup; the photocurrents from the Hewlett Packard LED's
were twice as high: and the ultrabright LED from Radio Shack gave a
photocurrent too high to be measured by the detector electronics. In
order to get a measurable photocurrent, the forward current for that LED

had to be reduced to 3 mA.

33

PEAK =6“ um.

   
  
  
   

~- -—-

---S,LIT;ii.OO#-m -:

~ PHT: 100 V

- SENSIIO"_”V -.
VEchL SCALE :1 v

_ roman cusuurzlumn
(m. L29)

m.-.._

 

Figure 3-5: Emission spectrum of a ultrabright type LED
from Radio Shack.

Relative intensity

3.5

2.5

1.5

0.5

 

642
HLMP-3750 —>
.. 610
60 nm
GL5H340 ->
48 nm

 

550 575 600 625 650 675 700 725

Wavelength. nm

Figure 3-6: Emission spectra of two available LED’s
for the detection of the malachite green

 

35

b. Lens and Window:

The purpose of the lens is to concentrate the light beam that
reaches the photodiode. T‘wo difi'erent ideas were considered when
choosing the lens. One possibility was to use a lens of short focal length
and place the LED far enough away to simulate a collimated beam. so
that the light could be focused to a point close to the end of the flow-cell.
This calls for a focal length between 8 and 10 mm. which would require
an impractical distance between the LED and the lens to produce a
"collimated" beam. Only with very short focal lengths (around 3 mm)
would that distance be reasonable. The alternative considered was to
place the emission element of the LED at the focal point of the lens.
Given that the LED is far from being a point source. the divergence of the
beam was reduced. but good collimation was not obtained. A coated lens
of 6 mm in diameter and 18 mm focal length was used. This is the
distance from the lens to the LED and to the photodiode.

A glass window 6 mm in diameter and 2 mm in thickness was
made for the other wall of the flow cell.

c. Photodiode:

The photodiode selected was a low bias voltage type from EG&G
Photon Devices (model FFD-0403). Similar to other photodiodes
available. this photodiode shows a maximum responsivity in the IR
region (around 900 nm) with a spectral range from 400 to l 150 nm. The
responsivity in the region of absorption of malachite green is about 70%
of the maximum with a value of 0.4 Al“! [37].

The photodiode has a circular active area 1 mm in diameter. which
is slightly larger than the diameter of the observation cell. The smallest

active area available was used to minimize the dark current when the

36
photodiode was operated under a bias voltage. At the end. only the
reference photodiode was operated with a bias of - l 5 V. with the main
photodiode operating in the photovoltaic mode (no bias voltage).
The first set of photodiodes received had the active area slightly
off-center. but this deficiency was not observed in the second set.
Photodiodes with active areas not centered were used only for the

reference circuit.

B. THE SYSTEM ELECTRONICS

Each colorimeter has two independent circuits: one to control the
LED and one to process the signal produced by the photodiode. The six
sets of circuits operate independently of one another. and the only
component shared is the reference voltage chip. A commercial power
supply (Model SLD- 1 5-808- 1 5 from Sola) with a separate housing
provides the power for all the circuits via a bus line with 1:15 V and a
grounded common line.

The circuitry was wired using the Scotchflex system with a

punchboard.

1. Light Regulation Electronics

A diagram of the circuit that controls and regulates the light level
from the LED is shown in Figure 3—7. The current from the reference
photodiode is first converted to a voltage that is compared against a
reference voltage. The output of this comparator controls the current
driving the LED to produce a negative feedback loop that keeps the light
level constant. An offset voltage is added to the summing point of the
current-to-voltage converter to adjust its output. which in turn adjusts
the LED current and the light level. The reference voltage used by the

37

Ann:— 05 56:85 05 852.50.. 23 £95.80 ova—.83....
8:53.. 2: an coo—6o...— Eotao 2:. 4.3.5 mean—=9.— Eufi Rn «Sufi

. mos—.49.
mammwmmx

«mm o<_ o m: xflxgzoo
3.59. 30589:

 

 

 

 

“Eu 8.8. 552355 3355”.
aura nmofi. MbOJh m...
» mm- m? .
S: a... . .
355 m” A 5: mam... + I .
. i .. “9.8“:
szm3 .4 5‘ %
oFZO
doom ._. din . .. zo-
msfi. $.50

.3. 02

38
comparator is a fraction of the output of a 10 V constant voltage source.
Adjusting this fraction also adjusts the LED current. A good way to
measure this forward current is to measure the voltage drop across the
resistor limiting the current (test point A in Figure 3—7). A value of 9 V at
this point corresponds to a nominal current of 20 mA. A value of 6 V
corresponds to the maximum current recommended (30 mA).

The reference photodiode is biased at - 15 V for two reasons: the
increase in photocurrent requires a smaller resistor in the current-to-
voltage converter; and the output of this converter will have the correct
polarity to be compared to the reference voltage. which eliminates the
use of an extra component to invert the signal.

The capacitors in parallel with the resistors of the current-to-
voltage converter and the comparator reduce the bandpass to 34 Hz.
The capacitor in the reference voltage source gives a bandpass of 2 Hz.
To measure the noise that this combination of bandpasses allows in the
current of the LED. the voltage at its anode was observed with an
oscilloscope. The 9 V signal showed a 60 Hz noise component of 12 mV
p-p and also noise at higher frequencies up to 90 MHz with amplitudes
as high as 8 mV p—p.

Each channel of the detector uses one dual op-amp chip (T‘L083-
ACN from Texas Instruments). but the reference voltage source (Analog
Devices 581) is the same for all the channels. With this set up. an
increase in the reference voltage will decrease the light levels of all the
channels: an increase in the offset voltage at the current-to-voltage
converter of any channel will decrease the light level of that channel
only. This is an important feature that allows the operator to set all the
channels to the same output voltage. since individual photometric cells

39
have difierent light throughput due to small differences in construction.
and different LED's require different currents to provide the same
luminous intensities. Also. in a specific determination. the working
range of any channel can be increased if needed by increasing its light
level individually.

2. Signal Processing Electronics

The diagram of this circuit is shown in Figure 3-8. It consists of a
very sensitive current-to—voltage converter followed by a variable gain
amplifier. In this circuit the photodiode is used with no bias voltage to
minimize the dark current.

A high quality operational amplifier (OP-43E]. Precision
Monolithics Inc.) is used for the current-to-voltage converter. This op
amp combines a low bias current (5 pA max.) with a low offset voltage
(0.25 mV max.). A large resistor in the feedback loop was avoided by
using a combination of resistors that gives a high equivalent resistor to
the converter. The transfer function of the converter is 5. 1x108 V/A.
The capacitor in parallel reduces the bandpass to 16 Hz.

The current-to-voltage converter is enclosed in an aluminum
shielded box (Pomona Electronics) to eliminate oscillations observed in
the output at gains of 10 or higher. The frequency of the oscillation
observed was of 120 Hz at a gain of 10 and decreased with increasing
gain. The dependence of the oscillation on the gain suggested that the
amplifier circuit was sending some kind of positive feedback signal to the
converter. Besides eliminating the oscillation described. the shielding of
the converter also helped decrease the level of 60 Hz noise.

The variable gain amplifier uses a potentiometer that can be
adjusted externally. It gives a gain from 1 to 60. and the capacitor in

 

 

I

40

are.» «58.8.... 355 3.. 83:.

Swim

3

.1
I
i
l
I
I
I
I
I
r
I
r
I

.l

 

 

 

 

_ .
.
u u

. . .
Wldé " w-WL ““2 F u

.o
a . In
3...... _ dxu _
“.56 .... . w. .
u 3.0““. u
_ . .
dud
”In. H .32 u
.38 3.9 . u
“(5.0 . . .
. ”Ta .
— -
r IIIIIIIIIIIIII L
mam—Jay? 220 3m<§<>

«gag mom—”.99....2 £8

”E

0:00}.

\a

4 1
parallel reduces the bandpass as the gain is increased from a maximum
of 159 Hz to a minimum of 3 Hz at the highest gain. This keeps the 60
Hz noise to a maximum of 6 mV p-p with output signals of 10 V. The
output signal also shows noise at higher frequencies starting at 90 Hz
and going as high as 90 MHz. The levels of noise from these sources are
lower than 2 mV.

At low gains. the response speed of the circuit is limited by the
current-to-voltage converter. At gains of about 12. the response time
starts to be determined by the amplification stage. but even at the
highest gain. the bandpass is high enough (3 Hz) to fully characterize
any FIA peak. a signal of less than 0. 1 Hz.

CHAPTER IV
OPERATION AND CHARACTERIZATION OF THE DETECTOR

This chapter begins with a description of two procedures related to
the operation of the detector. The first one is the starting procedure that
is followed each time the detector is used after a long period of
inoperation. The second one is a less routine procedure: that of
adjusting the light levels of the different photometers.

The second section of Chapter IV describes the results of various
tests performed on the detector in order to form an idea of the
instrument's characteristics. These tests were also applied to the

detector currently used [4] in the sugars analyzer to provide a point of
comparison. This detector has given excellent results. so it is a good
standard against which the new detector may be compared.

The chapter ends with the results of a real glucose determination.

again using both detectors.

A. OPERATION PROCEDURES

1. Initial Operation:
When the detector is used for the first time during the day. the

normal warm-up period of 15 minutes for the electronics should be

42

43
allowed. It was found that after a long time of inoperation the light
throughput of the flowcell is greatly reduced when the cell is stored with

water. thus giving a very low signal. The signal level starts to increase
when the cell is flushed with fresh water. This continues for a period of

approximately 1 5 minutes after which the signal reaches a stable point.
The initial reduction in signal is attributed to some deposition of

minerals on the surfaces of the cell including the lens and the window.
The time needed to wash off these residues is greatly increased because
of the geometry of the cell which has dead spaces at the sides of the inlet
and outlet holes (see Figure 3-3). The flow inside the cell is apparently

not turbulent enough to reach these points effectively.

If. on the other hand. the cell is rinsed with plenty of distilled water
after use and stored with no water. the remaining drops inside the cell
evaporate. The signal in the dry cell is very strong. about 5 times
stronger than the signal of the cell with water running through it:
nevertheless. the light intensity drops immediately when the liquid
begins to flow. When this is the case. the signal takes about 10 minutes

to stabilize.
The procedure recommended. then. is to rinse the cell with

distilled water after use. but to empty the cell for storage. The first time
it is used after storage. the system should be warmed up for 15 minutes

with water flowing through the cell.

2. Adjusting the Light Level:
As mentioned in the previous chapter. there are two controls that

afi'ect the light levels of the photometers. The reference voltage level
changes the current in all the LED's simultaneously. In addition. each

44
channel has an individual control to adjust the current driving its
particular LED.

There are two modes of operating the six photometers
simultaneously. One is to adjust all the currents for the different LED's
to the same value and to use the difierent gain controls to balance all the
output levels. To set this constant current mode. the offset voltage of
every channel has to be set by adjusting the corresponding trimmer to a
value that gives the same current for all the LED's (the same voltage at
each test point A in Figure 3-7). Once the currents are all equal. they

can be brought back to 20 mA (or higher) by adjusting the reference

voltage level. This mode has the disadvantage of giving a lower

signal-to-noise ratio for the channels with less light reaching the

photodiodes.
The second mode of operation is to adjust the forward currents of

each channel in order to equalize all the output signals. The offset
controls should be adjusted to give the same output levels when the gain
is the same for every channel. Again. the reference voltage is used to
bring back the currents to a value of approximately 20 mA. In this
constant signal mode some LED's will have higher forward currents. and
will thus result in slightly broader peaks and shorter lifetimes. As will be
seen later. the dependency of the FWHM of the LED's emission spectrum
on the forward current is very small: also. the price of the LED's is so low

that replacing them more frequently would not substantially increase the

price of operation of the detector. These considerations indicate that

operating the detector in the constant signal mode should give better

results for most applications.

45
It is important to remember that whenever an adjustment is being
made to the light level of any channel. the forward current of the LED
should be monitored in order to avoid values over the maximum rating of
30 mA (a minimum of 6 V at test point A in Figure 3-7).

B. CHARACTERISTICS OF THE DETECTOR

In order to determine the difi’erent figures of merit for the new
detector. a series of measurements and experiments were performed
using a single channel of the instrument (later referred to as LED-
detector). The same experiments were performed with the detector
constructed by Patton and Crouch [33]. (later referred to as Patton's
detector). Except for the experiments with methylene blue. the results
presented in this section were obtained using the FIA manifold shown in
Figure 4- 1 . This arrangement is equivalent to a single channel of the

sugars analyzer depicted in Figure 2- 1 .

The FIA apparatus consisted of a 12-channel peristaltic pump
(Ismatec model 7618-50; Glattbrugg. Switzerland) with flow-rated pump
tubing (Technicon Instruments: Tarrytown. NY): a pneumatically
activated injection valve (Rheodyne lnc.: Cotati. CA): either a digital
multi-meter (Fluke model 8000A: John Fluke MFG. Co. Inc.. Seattle, WA)
or a strip chart recorder for the repeatability experiments (Model SR-255

A/B. Heath Company): an IBM PC compatible microcomputer equipped
with an RT‘I-815 interface board (Analog Devices: Norwood. MA). The
computer controlled the sample injection and data acquisition via
software previously written for the FIA system. TYansmittance values
were record ed by the computer. and the corresponding absorbance was
calculated from the 100% transmittance reading. Teflon tubing of 0.86
mm i.d. (Benton-Dickinson: Parsipanny. NJ) was used throughout the

uBoSo—o Ends—02:8 o5 magma—o 8 888%
oases: 2.5585 usagesfieofiw "$8.55

 

 

 

 

 

 

 

 

 

 

 

 

 

coucoooa
, arena
_I.o..oEo.6ca
, 20000.".
283 T.\|/ 5:50
0688

 

 

 

 

 

 

 

 

 

47
manifold. The single head string reactors (SBSR) were previously
prepared using 0.6 mm diameter non-porous glass beads (Propper MFG.

00.; L]. City. NY).

1. Sensitivity to Flow Variations:
The photometers of the LED-detector show an unusual sensitivity

to any disturbance in the flow pattern of the liquid under observation.

The oscillations of the peristaltic pump are especially problematic. The
FIA peak in Figure 4-2 shows the extent of this noise. The noise
frequency depends on the flow rate setting of the pump. and is generally
quite low (between 1 and 3 Hz). Experiments showed that the noise
increases with increasing pressure of the pump rollers on the plastic
tubing. Increasing the LED brightness also increases the noise.

The shape of the peak generated by the detector response to the
oscillations of the pump (as observed with an oscilloscope) suggests that
it is caused by a refractive index effect also observed in the detector
designed by Betteridge et al. [24]. and more recently described in detail
by Evans et al. [38]. This effect makes the flowing stream behave like a
dynamic lens that first focuses and then diverges the light beam. The
pressing action of the pump's rollers on the tubing would seem to create
regions of higher and lower pressure within the stream. which may
change the refractive index of the liquid just enough to produce the
observed response.

The frequency of the pump oscillations is too low to be efi'ectively
filtered without affecting the sensitivity to the FIA peak. Possible ways of
reducing the noise include the use of a difi‘erent pump. The pump
currently used in the sugars analyzer is relatively new and produces very

intense oscillations. Tests with older pumps showed that as the pump

 

0.40

0.35 '-

0.25 '

 

0.15 '

Absorbonce
O
B

0.10 '

0.05 - I

5 10 15 20 25 30 35 40 45 50

 

 

 

Time, seconds

Figure 4—2: ' 'on of a 0.04 mM solution of genperouride.
noise the signal came from higmprrmp osdllation.

in
T'heflawratewas 1.3m1/min.

49

gets older. the pressure exerted by the rollers is decreased. which
decreases the noise in the detector signal. Using compressed gas [39] or
even gravity to pump the liquid produces a completely pulse-free flow.
but the compressed gas set-up is clumsy. and the use of gravity might
not provide enough pressure to push the liquid through string bead or
packed bed reactors. Still. these options may be implemented in extreme
cases. If a peristaltic pump is to be used. a pulse dampener should be
placed immediately after the pump. The bead reactors are in fact pulse
dampeners. but they did not reduce the pulsations effectively. unless
they were irnpractically long. A small stabilization device reported by
Bergamin et al. [40] (Figure 4-3) proved to be effective in reducing the
pulsations when placed right after the pump in the carrier channel. The
addition of one of these devices to the sample or reagent channel did not
reduce the noise. however. and produced an unwanted peak caused by
the sudden change in pressure during the injection process (Figure 4-4).

A problem with using a pulse dampener like the one described
above arises when the flowing liquid is changed. for example. from water
to a reagent. Every time the glass bulb is emptied and filled with a new
liquid. the baseline at the output takes about 15 minutes to stabilize.
This of course represents a longer waiting period and some waste of
reagent. With our particular sugars analyzer. however. the reagents are
never changed. so this process would have to be carried out only when
the detector is used with water before the experiments are run on a given
day. This can be avoided if the same carrier that will be used in the
experiment is used in the initial stabilization period.

The question remains of why the detector is so sensitive to the

small changes in the refractive index of the flowing liquid. Reports on

 

50

—> glass bulb

   

—> connector

 

from pump —> —> to manifold

i

metal tube

Figure 4-3: Stabilization device used in the carrier channel
after the pump to reduce the effect of its
oscillations.

Absorbonce

51

 

0.40

0.35

0.30

0.25

0.20

0.15

0.10

0.05

 

 

 

 

l I P 1 l l I 1,
10 15 20 25 30 35 40 45 50

Time. seconds

Figure 44: Injection of a 0.04 mM solution of hydrogen peroxide

using two stabilization devises: one in the carrier
channelandoneinthesamplechannel. Thesmall
peakatSseoondswascausedbyflieiniecfionprooess.

 

52

other photometric detectors for FIA. including the detector currently used
in the sugars analyzer. have not mentioned such a high sensitivity to
changes in the refractive index of the observed liquid. Three causes are
suggested: 1) the very small internal diameter of the flow cell (all other
LED-based detectors had cell dimensions at least twice as big as those in
the multichannel detector): 2) the position of the lens and window might
not be exactly parallel due to uneven pressure from the holders (this
would change the angle of incidence of the beam); and 3) the active area
of the photodiode is too small (if the active area is much larger than the
aperture stop defined by the internal diameter of the flow cell. the

dynamic lens effect is minimized).

To further study the first suggested cause for the problem. a new
observation cell was constructed with a larger internal diameter (1 .5 mm
instead of 0.6 mm). The peaks produced by this cell (Figure 4-5) did not
show any reduction in the baseline pulsations. In fact. these pulsations

were increased to the point that the pulse dampener was not able to
completely eliminate them (Figure 4-6). Also. the bigger cell produced
longer elution times. This is expected since the cell's internal diameter is
almost twice as large as the rest of the FIA tubing used.

No further attempts were made to reduce the detector's sensitivity
to flow variations. since the pulse dampener placed in the carrier
channel gave excellent results. All the experiments reported in the

following sections were run using this device.

2. Baseline Noise and Drift:
The baseline electrical noise was measured with the aid of an
oscilloscope. 'IVVO different types of noise were observed. One type is due

to oscillations from the peristaltic pump. These oscillations were greatly

 

0.40

0.30 -

0.25 '-

0.20

0.15

Absorbonce

0.10

0.05

 

Figured-5:

 

 

 

10 15 20 25 30 35 40 45 50
Time. seconds

Injection ofa 0.04 mM solution ofhydrogen peroxide
usinga obwvafioncenThenoiseinthesignal

larger
camefiomthepumposcillation. Theflowratewas
1.3m1/min.

Absorbonce

 

0.40

0.35 P

0.30 '-

0.25 h

0.20 -

0.15 -

0.10 -

0.05 -

 

 

 

 

5 lo 15 20' 25‘ 30' 35' 46 45' 55
Time. seconds
Figure 4—6: Injection of a 0.04 mM solution of hydrogen peroxide

using a observation cell and one stabilization
device in e carrier channel.

55

decreased by the use of the pulse dampener. but not totally eliminated.
Because of its low frequency. this noise cannot be effectively reduced by
the RC filter of the variable gain amplifier. and its amplitude increases
with increasing gain. At the minimum gain the noise can hardly be
detected. but at the gain needed to reach an output of 5 V (approximately
10) it has an amplitude of 10 mV. This represents a baseline noise of
0.001 Absorbance Units (A.U.) near zero absorbance.

The second type of baseline noise comes from the ac power line.
Noise at 60 and 90 Hz and higher harmonics is present with an
amplitude lower than the pump noise. This fixed-frequency electrical
noise is effectively filtered at higher gains so that the amplitude remains
approximately constant. The 60 Hz noise was about 6 mV p-p and the
higher frequency noise was 2 mV p-p or less.

Three different types of drifts in the baseline can be defined. The
first drift is due to the electronics. This was measured with no liquid
flowing through the cell and monitoring the output with the aid of a
chart recorder. A drift of 0.001 A.U./hr. was observed. A much higher
drift was observed when there was liquid flowing through the cell.
especially when the cell was used for the first time during the day. as
previously explained. It was suggested that a 15 minute period of
conditioning the cell with the reagents flowing was enough to obtain a
reasonable drift. Under these conditions the measured drift is 0.002
A.U./min. normally increasing the baseline signal. a fairly high value but
still low enough to accurately record an FIA peak (which takes less than
one minute to go from baseline to maximum). After one hour of
operation the initial drift due to the flow was reduced to 0.001 A.U./ min.
This third type of drift was observed in both directions -decreasing or

56
increasing the baseline signal— and it was also observed in the tests
with Patton's detector. The drift may be caused by very small variations
in flow rates. probably due to small variations in the voltage supplied to
the pump.

3. Peak Broadening:

There are three factors that contribute to the broadening of the
peak produced by a flow-through detector [41]: 1) dispersion in the
connecting tubes: 2) a finite effective measuring volume: and 3) the
dynamic behavior of the transducer and signal processing electronics.
The first factor seemed to be negligible for the LED detector since the
connecting tubes are less than 2 inches long. Nevertheless. it was
observed that the peaks recorded with the LED detector had a delay of
about 0.6 seconds with respect to the peaks recorded with Patton's
detector (Figure 4-7). This is a result of the liquid having to travel a
longer distance through the connecting tubes. It can also be observed in
Figure 4-7 that the peaks of the LED detector do not fall as fast as those
obtained with Patton's detector. This higher dispersion of the LED
detector is produced in part by the effect of the laminar flow in the open
connecting tubes. There should also be a contribution from the second
factor mentioned above. that is. the cell in the LED detector must behave
more like a mixing chamber than the cell in Patton's detector. This is
expected since the drilled plastic walls in the LED cell must have a more
uneven surface than the glass ones in Patton's commercial cell. To
estimate the contribution to dispersion from the second factor. we can
assume that the flow cell in the LED detector behaves completely as a
small mixing chamber (the worst case). Then. the volume standard
deviation is the cell volume Vc. and the time standard deviation from this

Absorbonce

57

 

 

 

 

 

 

 

 

2.0
L.
1.5 '
: > Patton high cone.
1.0 -
.. \ 5 LED high cone.
" \- J) Patton low cone.
. \\
0.5 :- \ —> LED low conc.
r k
0 P . l l . ...... i"

 

 

 

5 8 11 14 17 20 23 26 29

Time. seconds

Figure 4-7: man patterns for . both demctors at low and
sample concentrations.

58

factor is the cell volume divided by the flow rate. at=Vc/fr. For a typical
glucose determination. the total flow rate is set at around 9 uL/s: the
LED photometer has an observation cell with a measuring volume of 3.54
uL. This gives a peak broadening due to the cell volume of 0.40 s. The
contribution of the third factor can be estimated from the time constants
in the amplifier chain. With a typical gain value of 10. the time constant
of the UV converter is of the same magiitude as the one from the
amplifier: 10 ms. This would give a total contribution of 14 ms. which is
negligible compared to the contribution of the other two factors.

In the determinations done with the multichannel detector. typical
peak widths were on the order of 20 s. Following the guidelines given by
Ruzicka and Hansen [7] the detector should contribute a mardmum of
5% to the overall peak width. This permits a contribution from the
detector up to l s. Even with the increased dispersion. the LED detector
still falls within the recommended limits set by the FIA system as used in
the sugars analyzer.

4. Dynamic Range:

The main objective of the experiments done in this area was to
determine the photometric working range for the LED detector and the
linear part of that working range. Possible factors affecting the deviation
from linearity were also studied.

Four major factors were seen to contribute to deviations from
linearity: two coming from the detector: stray light and non-
monochromatic light: one coming from the FIA manifold: insufi‘icient
dispersion of concentrated samples: and one from the chemical system:
the detection reaction in the sugars determination is an equilibrium one.

59
The following experiments give some guidelines to understand the effect
of these factors.

a. Experiment 1: Calibration curve using LMG-H202. Solutions
of difi'erent concentrations of H202 were used to study the detection
reaction for the enzymatic determination of sugars. The procedure is the
one suggested by Aspris [6] and Kurtz [5]. The stock solutions needed for
the experiment were: Hydrogen Perordde 0.01 M to prepare the different
samples of concentrations 0.02 to 0.20 mM: Leucomalachite Green
(LMG) 0.5 g/L (1.5 mM) in Acetic acid 37%: phosphate buffer (0.1 M. pH
6); and acetate buffer (0. l M. pH 4). The reagent solution was prepared
by dissolving 8 mg of horseradish peroxidase in 2.4 ml of pH 6 bufi'er.
adding 0.6 ml of stock LMG. and diluting to 10 ml with pH 4 buffer. The
final concentration of LMG in this solution is 0.090 mM. This solution
had to be prepared just before the determination as it is unstable to light
and atmospheric oxygen. The phosphate buffer pH 6 was used as a
carrier and also to prepare all the H202 solutions. The flow rates in the
carrier. sample. and reagent channels were 0.51. 0.40. and 0.06 ml/min
respectively (pump setting of 30). A 30 1.1L sample was injected. each
sample being injected in triplicate for this and the rest of the
experiments. A 10 cm plain SBSR was used in place of the glucose
oxidase reactor. and a 30 cm plain SBSR was used for the detection
reaction. Patton's detector was used with a filter centered at 620 and a
bandpass of 20 nm. The LED detector was used with the LED from
Sharp Electronics emitting at 610 nm and with a forward current of 30
mA.

The results are presented in Figure 4-8. The curves for both

detectors show the same tendency to level ofi' at absorbances of 0.75 for

 

2.00

l .80
6 LED detector

1 .
60 * Patton's detector

1 .40
1 .20
1.0)
0.80
0.60
0.40

Absorbonce

0.20

0 0.02 0.06 0.10 0.14 0.18 0.22 0.26 0.30

TIIIIIIIIIUIIIITIII

 

 

l l l l I l l

 

Concentration of hydrogen peroxide. mM

Figure 4—8: Calibration curve 11 ° IMG- gen peroxide with
stock solution of 1M 1.5 .

 

6 l

the LED detector and l .3 for Patton's detector. The deviation from
linearity in both cases starts at concentrations above 0. 16 mM. The
slopes of the linear ranges are: Patton detector. 6.56 A/mM; LED
detector. 4.28 A/mM. The response of the first detector is 1.5 times
higher than that of the second one. The similar behavior in the curves
from both detectors suggests that the reagent LMG might be limiting the
reaction at concentrations of H202 above 0. 16 mM. This is very possible
since the effective concentration of LMG in the carrier stream is 0.010
mM.

1:. Experiment 2: Effect of the concentration of LMG. The same
experiment is repeated keeping all the conditions constant but increasing
the concentration of the stock solution of LMG to 1 g/l (3 mM). This
gives a final concentration of LMG in the reagent solution of 0. 180 mM.
The problem with increasing the concentration of LMG in the reagent
solution is that its instability increases. so the solution has a shorter
usable period of time. The standard solutions of H202 are prepared
from 0.05 mM to 0.40 mM to have a wider range of observation.

The results of this experiment are shown in Figure 4-9. The
general shape of the curves is similar the curves in Figure 4-8. with two
important differences: The deviations from linearity now start at higher
concentrations (above 0.25 mM): and the values of absorbances where
the curves level off are consequently higher. The slopes for the linear
part of the curves are: Patton's. 7.36 A/mM; LED's, 3.82 A/mM. The
possibility of a limiting reagent is still present since doubling the
concentration of reagent roughly doubled the concentration of H202
where the deviation started. Of course. there is also the possibility of

other factors affecting the curves. since there were high concentrations of

Absorbonce

2.0

 

1.8
1.6
1.4

0 LED detector
* Patton‘s detector

1 .2
1 .0
0.8
0.6
0.4
0.2

0 0.10 0.20

 

TIIUIIIIIIIUUTIIUUI

 

l l l

0.30 0.40

 

Concentration of hydrogen peroxide. mM

Figure 4—9: Calibration curve u ° IMG-h gen peroxide with
stock solution of LM 3.0 vadro

ll

 

63
H202 involved and high absorbance values. at least in the case of
Patton's detector.

c. Experiment 3: Effect of the light throughput and stray light.
The conditions in experiment 2 were used to compare the performance of
two observation cells of different diameters in the LED detector. If the
observed deviations from linearity were caused by stray light. increasing
the light throughput in the observation cell should decrease the
percentage of stray light reaching the photodiode. A regular diameter cell
(0.6 mm) and a large diameter cell (1.5 mm) were used with the added
feature of having the side walls of the cells facing the LED and the
photodiode covered with a sealant and painted completely black to block
any possible stray light.

The results are presented in Figure 4- 10. The two calibration
curves are practically the same as the ones presented in Figure 4-9 for
the LED detector. The slopes for the linear part of the curves are: large
cell. 4.31 A/mM: small cell. 3.92 A/mM (compare to 3.82 A/mM for this
cell in the previous experiment). The efforts to reduce the stray light in
the LED detector did not improve its response. The effect of increasing
the amount of light reaching the photodiode through the observation cell
had little effect on the general shape of the calibration curve.

(1. Experiment 4: Effect of the concentration of LMG. The
conditions of the detection reaction were completely changed to try to
improve the working range. The new conditions were: stock solution of
LMG 3 g/l (9 mM) in acetic acid 37%: and phosphate buffer (0.05 M. pH
6.86) used to prepare the reagent solution. the H202 solutions. and as a
carrier. The reagent solution was prepared as follows: 8.0 mg of

perorddase dissolved in 5 ml of phosphate buffer: 0.5 ml of LMG stock

 

2.0
1.8

a
1. 6 Large cell

1.4 + Normal cell

1.2
1.0
0.8
0.6

Absorbonce

0.4
0.2
0 l

lIlllIllllllIlIIle

 

 

l l l l l l

0. 10 0.20 0.30 0.40

 

Concentration of hydrogen peroxide. mM

Figure 440: Calibration curve 11 ° Dag-human peroxide with
stock solution of LM 1.5 using two sizes of
observation cells.

65

solution added and diluted to 10 ml with the same phosphate bufi'er.
The final concentration of LMG in the reagent solution was 0.450 mM.
The volume injected was increased to 50 uL. The flow rates were
increased to 0.70. 0.53. and 0.08 ml/min for the carrier. the sample. and
the reagent respectively (pump setting of 40). The plain SBSR reactor for
the detection reaction was increased to 40 cm. The calibration curve was
done with the LED detector using the three available LED's (see Figure 3-
5) to determine if any important variations in the shape of the calibration
curve could be observed by using LED's with difierent bandwidths of
emission.

The results are presented in Figure 4- 1 1 . The response was now
much higher for both detectors to the extent that only concentrations up
to 0.08 mM H202 could be read with Patton's detector. The slopes for
the two detectors were now 4 times higher than in previous experiments
(32.9 A/mM for Patton's detector and 14.7 A/mM for the LED detector at
610 nm). This change in slope indicates that the LMG was in fact
limiting the reaction. and at lower concentrations of LMG. the reaction
was not reaching completion.

As expected. the sensitivity of the curves obtained with the LED's
at 640 and 660 nm was much lower. The slopes of the linear parts are
7.3 A/mM for the LED at 640 nm and 6.7 A/mM for the LED at 660 nm.

The curve from Patton's detector shows a negative deviation from
linearity starting at 0.08 mM H202 and completely levels ofi' at 0. 10 mM
H202. A similar deviation is observed in the three curves for the LED
detector. The strong deviation in Patton's detector may be explained in
terms of stray light. since this change was observed at very high values
of absorbances. Because the deviations observed in the LED detector

2.6
2.4 P
2.2 -
2.0 -
1.8 -
1.6 -
1.4 -
1.2 -
1.0 - , - m
0.3 -
0.6 - 4
0.4 - ’3. 1
0.2 - ' /

 

Absorbonce

 

 

0 l l l l l l J l

0.02 0.04 0.06 0.08 0. 10

Concentration of hydrogen peroxide. mM

3 Patton's detector 2 LED at 610 9 LED at 640 ‘ LED at 660

Figure 4-11: Calibration curve using mag-3dr? peroxide with
stock solution of LMG 9.0 . e LED detector
was used with the three available LED's.

67

started at the same concentration as in Patton's detector. it is difficult to
conclude that the deviations are due to effects coming from the detector
and not from another chemical effect hidden under the deviation fi'om
stray light in Patton's detector. This becomes more feasible when one
realizes that the effective concentration of LMG in the carrier stream is
0.05 mM due to the difi'erent flow rates of carrier and reagent channels.
and this is the concentration of H202 where the deviations occurred.

e. Experiment 5: Effect of non-monochromatic light. In the past
four experiments. the response of Patton's detector has been higher than
that of the LED detector. This is easily explained by the fact that the
filter used in Patton's detector was centered very close to the wavelength
of maximum absorbance of malachite geen with a bandpass of 20 nm.
On the other hand. the LED detector used a source with emission
centered on a shoulder of the absorption spectrum of MG and with a
broader bandpass. If the bandpass of the LED could be narrowed. the
sensitivity of the detector would be increased by increasing the slope of
the calibration curve. The linear range should also be increased if the
deviation observed so far is due mainly to a non-monochromatic light
effect.

One way to reduce the bandpass of the emission from the LED is
to use a filter in front of it. In the LED detector this is not so easy to
accomplish. but this can be done if the LED is used as the light source in
Patton's detector. T\vo calibration curves were done in this way with and
without a filter in Patton's detector. The conditions of the experiment
were identical to the ones in Experiment 4. except for the filter used. If a
620 nm filter was used with a 610 nm LED the light throughput was too
low to have good detection. so a 610 nm filter was used instead. Still.

68
the light throughput of the LED-filter combination was much lower
(about 10 times) than with the regular tungsten lamp. On the other
hand, the LED alone gave a good signal.

The results are shown in Figure 4- 12. The calibration curves had
slopes of 26.1 A/mM and 18.4 A/mM for the LED-filter combination and
the LED alone respectively. This proves that the sensitivity can be
increased in the LED detector by decreasing the bandpass of the
emission from the LED. On the other hand. the question of the main
source of deviation from linearity when a LED is used alone still remains
unanswered from the data of this experiment.

There is a small increase in the linear range when a filter is used in
front of the LED. meaning that in fact there is some contribution from
polychromatic light effects to the deviation observed when using the LED.

Furthermore. the same gaph gives evidence that there is another
factor causing the deviation. The response of the LED-filter combination
is lower than that of tungsten lamp-filter combination. This is explained
by the fact that in the first situation. a filter with a center farther from
the absorbance maximum of the analyte was used. With the decreased
signal. the samples of 0.08 and 0.10 mM H202 are no longer out of
range as they were in the previous experiment. and now they show a
similar deviation to the one Observed using the LED alone as light
source. This could mean that this deviation is actually due to a chemical
effect as was suggested previously. On the other hand. there is also the
possibility of this deviation being caused in part by stray light. Stray
light in Patton's detector comes mainly from ambient light. By using an

LED-filter combination the source radiance was decreased tenfold. and

 

 

 

 

 

1.8
.- __.II
1.6 r a with filter
1.4 E * without fitter
a) 1.2 h
g 0.8 t
<
0.6 -
0.4 -
0.2 -
0 l l J j l l l I l l
0.02 0.04 0.06 0.08 0. 10
Concentration of hydrogen peroxide. mM
Figure 412: Calibration curve using WG- peroxide with

stock solution of LMG 10.0 mM. LED at
610 nm was used in Patton’s detector with and
without a filter.

70
this would be the equivalent to increasing the stray light percentage ten
times.

Comparing the curves obtained using Patton's detector with the
LED and no filter (from Figure 4-12) and the curve using the LED
detector (from Figure 4- 1 1). a higher response is still observed with
Patton's detector. This could be due to a lower dispersion in the sample
zone produced by the cell in Patton's detector as previously discussed.

f. Experiment 6: Effect of the LED forward current. Reducing the
forward current of the LED is an easy way to reduce the width of the
emission band from the LED. This was tested in an experiment to see if
working at low forward currents would improve the performance of the
LED detector. Once more. calibration curves using LMG-H202 were
obtained with the LED detector. The conditions were the same as in
Experiment 5. and the LED in the detector was first used at 30 mA. the
maximum forward current recommended. and then at 10 mA. The
results are presented in Figure 4- 13.

At the two currents used. there was no noticeable change in the
performance of the detector. In fact. the two curves are so similar in the
lower points that the experiment could be used to show the repeatability
of measurements. The upper points show a variation. but this could be
attributed to changes in the reagent solution. which after a period of time
starts to degade.

The possibility of stray light variations at the difi'erent forward
currents does not agee with the data. In the LED detector the stray
light comes mainly from the light passing through the imperfections in
the sealant and black paint on the side walls of the observation cell. This
means that the stray light should be proportional to the total radiance of

 

 

 

 

 

1.2
o
i
i
0 l I l l l g l l l l
0.02 0.04 0.06 0.08 0. IO
Concentration of hydrogen peroxide. mM
Figure 4-13: Calibration curve using LMG- peroxide with

stock solution of LMG 10.0 mM. LED detector
wasusedwithlEDforwardcurrentsoflOandmmA

72
the LED. so that at higher forward currents a stronger negative deviation
should occur. The opposite is observed in Figure 4- 13.

The spectra of the LED at two difi'erent currents is shown in Figure
4- 14. The calculated FWHM is 8 nm larger when the current is
increased from 1 1 .6 mA to 20.3 mA. Apparently. the FWHM is already
so large that a difi'erence of 8 nm does not have any sigiificant effect on
the detector's response.

g. Experiment 7: Calibration curve using a Dye. From the
previous experiments. the question of whether a chemical effect or a
stray light effect was causing the observed deviation from linearity could
not be answered with confidence. Injecting a solution of pure dye
reduces the chance of a chemical effect. and. if the dye is chosen
correctly. the non-monochromatic light efi'ect can also be reduced.
Several pH indicators were tested looking for a relatively flat absorption
spectrum in the 6 10 nm area. The one that meets that condition very
well is methylene blue with peaks at 612 nm and 660 nm (Figure 4- 15).
Unfortunately. methylene blue forms a dimer at high concentrations.
The dimer has a higher absorptivity at 600 nm. which produces a
positive deviation from Beer's law. and the monomer absorbs stronger at
660 nm producing a negative deviation [42]. Another problem with the
dye is that it adheres strongly to glass. and the SBSR's were retaining a
lot of the color.

A difierent manifold was then used which included an Open Teflon
tube (30 cm) instead of the two SBSR's in the carrier channel. No
reagent channel was used. The flow rate was 1 ml] min (pump setting
60). A stock solution 0.2 mM of the dye in pure water was used to
prepare injection samples 0.02 mM to 0.2 mM. Calibration curves were

Relative intensity

2.8

2.4

1.6

1.2

0.8

0.4

 

 

 

550 575 6(1) 625 650 675 7(1)

Wavelength. nm

Figure4~14zEmissionspectrafortheLEDat610with
lowandhighforwardcurrents

 

74

 

 

 

575 600 625 650 675 700
WAVELENGTH (nm)

 

Figure 4- 1 5: Absorption spectrum of methylene blue in water.

75
prepared at 610 and 660 nm for both detectors. The results are
presented in Figure 4- 16.

The curve from the LED detector at 610 nm shows a deviation from
linearity at absorbances above 1 . The curve does not level off as fast as
the curves involving LMG-H202. suggesting that in fact. there was a
chemical limitation producing the deviation in those curves. The slope
for the linear part of the curve is 10.9 A/nm.

The curve for Patton's detector at 610 nm is linear up to
absorbances of 2.4. which confirms the results obtained for the LMG-
H202 reaction. The slope of the regession line is 13.0 A/nm. or 1.2
times higher than the curve with the LED detector. The positive
deviation that is observed at 600 nm was not seen at 610 nm. at least for
the concentrations studied. perhaps because this wavelength is closer to
the isobestic point which is expected at a wavelength between 600 and
660 nm. With methylene blue having a similar absorptivity between 6 10
and 625 nm. the chances of 610 nm being close to the isobestic point
increase. At 660 nm. the negative deviation from linearity starts at low
concentrations for both detectors. as expected.

Even with the dimer formation affecting the curve at 660 nm. some
interesting conclusions can be drawn from that curve. Comparing the
curves at 610 and 660 nm for the LED detector. it is observed that the
curve at 610 levels off faster even though the amount of stray light in the
detector is expected to be lower at 610 nm because that LED is at least
10 times dimmer. This indicates that the negative deviation at 610 nm
might have a stronger effect from non-monochromatic light. Even with
the flat area of absorptivity around 615 nm. the bandwidth of emission is
so broad for the LED that it seems to be the dominant cause of

 

Absorbonce

76

 

2.4 -

2.0 -

I. "if —¢_—

1.6 P
1.2 '-
0.8 h
0.4 - '
r' /

0 L l L l l l 1 L l l l l

0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18 0.20

 

 

 

Concentration of hydrogen peroxide. mM

a PaflonblOnm. Paiton660nrn .LEDOlOnm ‘LED660nm

Figure 4-16: Calibration curve using methylene blue with the
twodetectorsattwowavelengths.

77
non-linearity. In the LMG-H202 reaction the efi'ect of non-
monochromatic light should be stronger and definitely dominant since
the measurement is taken on a shoulder of the absorption spectrum.

h. Experiment 8: Calibration curve injecting prepared malachite
geen. To prove that there was a chemical limitation in the previous
LMG-H202 curves. the reaction was carried out in flasks using a
stoichiometric excess of LMG. but still keeping the conditions as close as
possible to the ones used in the FIA system. To do this. the proper
amount of a stock solution of H202 (0.256 mM) was added to volumetric
flasks to give final H202 concentrations in the range of 0.01 to 0.15 mM.
50 ml of a reagent solution was prepared by dissolving 40 mg of
peroxidase in 20 ml of pH 6.86 phosphate buffer 0.05 M. 5 ml of a 10
mM stock solution of LMG in 37% acetic acid were then added. and the
mixture was diluted to 50 ml with phosphate buffer. pH 6.86. giving a
final concentration of LMG in the reagent solution of 1 mM. This
solution was added to the flasks containing the H202 in the same
proportion to give a dilution of the LMG of 1 to 5. The dark blue-geen
color appeared immediately and the solutions were then taken to volume
with phosphate buffer of pH 6.86. The final solutions were slightly more
acidic than the equivalent mixtures in the FIA system due to the
increased amount of LMG in acetic acid added. The solutions were then
injected in the FIA system using the same manifold and settings of
experiment 5.

Figure 4— 1 7 shows a set of peaks obtained with the LED detector
for this experiment. Figure 4- 18 shows the calibration curves for both
detectors. The curve using Patton's detector is again linear to an
absorbance of 2. as previously reported. The curve for the LED detector

Absorbonce

1.0

0.8

0.6

0.4

0.2

78

 

 

0.102 mM

0.077 mM

0.051 mM —/
0.031 mM —/

0.010 mM /,

 

 

—

I l I l l l l

5 101520253035404550—5560657075

Time. seconds

Figure «iv-17: FIA peaks Observed with the calibration curve

usingpreparedmalachitegreen.

 

Absorbonce

2.0

1.6

1.2

0.8

0.4

79

 

 

 
  
   

n Patton's detector
+ LED detector

  
  

0.02 0.04 0.06 0.08 0. 10 0. 12

   

Concentration of hydrogen peroxide. mM

Figure 4—18: Cah’bration curve using malachite green
previously

 

will: i: “q

80

shows a smoother deviation from linearity than in previous LMG-H202
curves indicating that in fact. the chemical system was the limitation in
those curves. The deviation observed should be mainly due to the non-
monochromatic light effect. and it is comparable to the deviation
observed for the methylene blue at 610 nm (Figure 4- 16).

The slopes of the regession lines are 19.7 A/nm for Patton's
detector and 14.0 A/nm for the LED detector. These slopes keep the
same ratio of 1.4 that was observed in previous LMG—H202 curves.

5. Precision of Measurements:

Any measurement taken by the detector must be taken with an FIA
system attached to it. which means that the photometric precision will be
affected. and probably limited. by variables from that system. If different
measurements taken with any two detectors result in similar
repeatability. it could be inferred that the FIA system is limiting the
precision. rather than the detectors themselves. or that the two detectors
actually have similar repeatability. T‘wo solutions of H202 were
repeatedly injected and measured with the two detectors in question
using the conditions described in Experiment 2. A solution of 0.04 mM
H202 was injected 9 times in the FIA system for both detectors (See
Figures 4- 19 and 4-20). A second solution twice as concentrated was
injected 8 times in the FIA system also for both detectors (Figures 4-2 1
and 4-22). The results are summarized in Table 4- 1. Patton's detector
shows a lower relative standard deviation (RSD) for both samples. The
dependance of the RSD with concentration of H202 is not clear from
these data. An additional source of variation seemed to afi'ect these
results: the instability of the reagent solution. It was observed that the
last measurements of the set of injections produced lower values than

81

 

Figure 4-19 9 injections of a 0.04nM solution of H202 using the LED-hm
detector. The RSD of the measured absorance is 2.63%.

82

_s_.._' .

4-..... _. .1... -..-

--'..-o._.-- .

 

Figure 4-20 9 injections of a 0.04 mM solution of H202 using Paton 's detector.
The RSD of the measure absorance is 1.62%

 

Figure 4-21 8 injections of a 0.08 MM solution of H2)2 using the LED-based
dectector. The RSD of the measured absorbance is 2.25%.

 

Figure 4—22: 8 injections of a 0.08 mM solution H 0 using
Patton‘s detector. The RSD of the mgaszured
absorbance is 1.95%.

85

Table 4- 1:

Results of the experiment of repeatability of measurements with the

 

LED-based and Patton's photometers

 

 

 

 

 

 

 

Solution Solution Average
0.04 mM H202 0.08 mM H202 RSD %
avg. 0.143 AU.‘ avg. 0.335 A.U.
LED SD 0.0038 A.U. SD 0.0075 A.U. 2.44
Detector RSD 2.63% RSD 2.25%
n=9 =8
avg. 0.204 A.U. avg. 0.617 A.U.
Current SD 0.0033 A.U. SD 0.012 AU. 1.79
Detector RSD 1.62% RSD l .95%
n=9 n=8
‘A.U. = absorbance units

86
the initial ones. This may come from some degradation of the reagent
solution during the time required to take the measurements.

More information regarding the precision of measurements with
both detectors can be obtained from the calibration curves of the
experiments in the previous section. Each point in every curve was
measured in triplicate so that a RSD can be calculated for each sample.
Table 4-2 shows the RSD observed for the points in two calibration
curves for both detectors. In both cases the average RSD for the
injections with the LED detector was roughly twice that of the injections
using Patton's detector. In all the curves, the RSD tends to decrease
with higher absorbance readings, an unexpected trend. Normally. if the
precision of the absorbance reading is limited by any intrinsic noise from
the detector (except for flicker noise). the RSD should increase with
higher absorbance values. The observed values then suggest that the
precision in the readings may be limited by noise coming from a source
external to the detector.

The noise in the output signals of both detectors measured with an
oscilloscope is on the order of 4 mV for an output of 1 V when no liquid
is flowing through. When there is a flowing liquid present. both detectors
show higher noise in the signals due to variations in the flow patterns.
As previously discussed. the LED detector is highly sensitive to these
variations. and Patton's detector is sensitive as well but to a lower
degree. Even when a pulse dampener is used. the pulsations coming
from the pump increase the signal noise in both detectors to values
higher than 4 mV. This means that in the 100% T baseline reading. the
noise is limited by the pump pulsations. However, when a highly
absorbing species was present. it was observed that the variation in the

87

Table 4-2:

Relative Standard Deviations for the triplicate readings

of the solutions in two calibration curves

 

Calibration curve with LMG-H202

 

Concentration (mM) 0.01 0.02 0.03 0.04 0.05 0.06 0. 10

RSD of measurements 2.2 1.7 0.3 1.3 0.7 0. l 0.3
with Patton's det.

RSD of measurements 3.0 3.8 0.8 1.9 0.5 2.6 1.9
with LED det.

 

Calibration curve with Methylene Blue

 

Concentration (M 0.01 0.03 0.05 0.08 0. 10

RSD of measurements 1.3 0.9 1.8 0.5 0.1
with Patton's det.

RSD of measurements 3.7 2.4 2. l 0.9 0.4
with LED det.

 

88
output reading decreased for both detectors. This behavior explains the
decrease in RSD at high absorbances. A full explanation of the behavior
of the flow-through photometers with respect to flow variations is beyond
the scope of the present thesis.

In conclusion. the precision of the measurements taken with
Patton's detector seems to be twice as high as the precision of
measurement with the LED detector. and this is mainly due to a higher
sensitivity of the latter to flow variations and not to intrinsic noise within
the detector.

6. Detection Limit:

The formula DL=ksbkl m [42] permits the calculation of the
detection limit of a determination using a calibration curve. with a
confidence level determined by k (k=2 gives a 97.7 confidence level). from
the values for Sbk- the standard deviation for a blank injection, and m.
the slope of the calibration curve. The relative values for these quantities
are lmown for both detectors from the data in the preceding sections so
that a relative detection limit can be estimated. In the LED detector. the
standard deviation of the blank should be two times that of Patton's
detector; and the slope of the calibration curve using the LMG detection
reaction should be 0.67 times the slope from Patton's detector. This
indicates that the LED detector should have a detection limit three times
higher than the detection limit of Patton's detector.

C. PRACTICAL APPLICATION
As a final test for the new LED detector. a real glucose
determination was carried out. The procedure suggested by Aspris [6]

was followed to determine the glucose content of a sample of commercial

89

100% apple juice. No special pre-treatment was performed on the
sample. 20.0 grams of the juice were diluted to 500 ml using distilled
water. 1 ml of this solution was transferred to a 50 ml flask. 1 mg of
ascorbate oxidase (1700 u / mg) was dissolved in 0. 1 ml phosphate buffer
0.1 M pH 5.5 and the solution was added to the 50 ml flask with the
sample to destroy the interfering ascorbic acid. The solution was taken
to volume with phosphate buffer 0.05 M pH 6.86. Before injection. the
solution was filtered with a No. 1 Whatman filter paper.

Standard glucose solutions 0.04 mM to 0.25 mM were prepared
from a 20 mM stock solution using phosphate buffer (0.05 M, pH 6.86)
for all the dilutions. 1 0 ml of the LMG reagent solution were prepared
using 8.0 mg of peroxidase and 0.5 ml of 10 mM LMG stock solution and
the reagents were diluted with phosphate buffer (0.05 M, pH 6.86). A 10
cm glucose oxidase SBSR was used in the manifold shown in Figure 4-1.
50 pl samples were injected in triplicate. Both detectors were used at a
wavelength of 610 nm. The calibration curves are shown in Figure 4-23
and a description of the results follows.

In the case of the LED detector, 4 points were included in the
calculation of the regression line giving a slope of 3.79 :i: 0. 13 A/nm; the
average deviation from the regression was 0.02 1 . The reading for the
unknown gave a glucose concentration of 0.083 mM with a standard
deviation due to the regression of 0.004 mM (4.9% RSD). The last
standard point (0.25 mM) was not included in the regression because it
showed a definite negative deviation. The average RSD from the
triplicate injections was 1.5%. The calculated glucose content for the
apple juice was 1.9 :i: 0.1 % w/w.

Absorbonce

1.6

1.4

1.0

0.8

0.6

0.4

0.2

 

,.
1.2 '-

 

u Patton's detector
+ LED detector

I I I I I I I I I I I I

 

0.04 0.08 0.12 0.16 0.20 0.24

Concentration of hydrogen peroxide. mM

Figure 4-23: Cah’braiion curve for the glucose content

determination in apple juice.

 

9 1

In the case of Patton's detector, 5 points were included in the
calculation of the regression line giving a slope of 5.55 :l: 0. l2 A/nm; the
average deviation from the regression was 0.025. The reading for the
unknown gave a glucose concentration of 0.08 1 mM with a standard
deviation due to the regression of 0.003 mM (4. 1% RSD). The average
RSD from the triplicate injections was 1 .5%. The calculated glucose
content for the apple juice was 1.8 :i: 0. l % w/w.

The difference in the two results is not significant with a
confidence level of 99%. The sensitivity of the determination as
measured from the slopes of the calibration curves was 1 .46 times higher
in Patton's detector. mom the absorbance values observed in the LMG-
glucose curve and in the LMG- H202 one. it can be estimated that a
H202 solution is producing the same absorption as a glucose solution
four times as concentrated. meaning that the glucose oxidase reactor is
converting 25% of the glucose present. This activity in the reactor is
similar to values previously observed [6]. even though the enzyme reactor

was more than one year old.

CHAPTER V
CONCLUSIONS AND RECOMMENDATIONS

The LED-based photometer constructed offers a simple. low cost
option for detection in flow injection analysis. especially for parallel
multidetection. The main requirement necessary to be able to use the
detector in a given determination is the availability of an LED that emits
at a wavelength where the analyte absorbs. In recent times. stronger
emitting LED's at many wavelengths in the visible and IR region have
become available. making the detector suitable for more applications.

For the sugars analyzer under construction in our laboratories. the
detector provides an adequate response within its limitations. when the
malachite green reaction is used for the detection. The glucose content
of the commercial apple juice determined in the last experiment
presented in the previous chapter was in excellent agreement for both
the currently used and the LED-based detectors. giving strong evidence
of the applicability of the new detector.

The main limitation of the LED detector. when used in a sugar
determination with the malachite green reaction. is the low linear range
observed. The measurements have to be limited to absorbances below
0.9 if a linear calibration curve is expected. It was concluded from the
experiments presented in Chapter IV that the main source of this

92

93
deviation was the polychromatic light effect. LED's emit with a relatively
broad emission spectrum. and the effect was increased because the
measurements were taken on a shoulder of the absorption band of
malachite green.

The observed linear range should be suitable for most applications.
and an even broader range can be used if a polynomial regression is
used. One way of increasing the linear range of the LED detector is to
place a filter between the LED and the photodiode. A filter could be used
instead of the window in front of the photodiode: also. there are
photodiodes available with built-in filters [37]. These options can be
implemented if a larger linear range is necessary, and if low operation
cost is not so important. It is also expected that new technology will
bring improved LED's with narrower emission spectra (the ultrabright
LED from Radio Shack is a good example). Laser diodes have very
narrow bands of emission; if one emitting in the vicinity of 610 nm is
available. it will present another option to improve the linear range of the
LED detector.

The other consequence of having a broad bandwidth of emission
from the LED is decreased sensitivity. The LED detector always gave
calibration curves with slopes 0.7 times the slopes observed with Patton's
detector when the malachite green reaction was used.

When compared with the detector currently used. the LED detector
showed a lower precision (about half of the other detector) mainly due to
a higher sensitivity of the LED detector to the variations in flow in the
observed liquid. This lower precision became less important in the
enzymatic determination of glucose. because the method itself presented
a larger variation, as observed from the calibration curves in Figure 4-23.

 

94
and the deviation values from the regression analysis. The
determination using the LED detector gave a relative error of 5% against
4% in the determination with the other detector.

A possible way to reduce the sensitivity of the LED detector to the
variations in flow. thus increasing its precision. is to use pulsed LED's
with synchronous detection.

It was not possible to determine a figure for the percentage of stray
light in the LED detector. mainly because the effect of polychromatic
light on linearity could not be separated from the stray light effect. The
amount of stray light should be low if care is taken in sealing and
painting the sides of the observation cell. Special care is needed when
the cell is taken apart for cleaning purposes because some of the paint
may come off when the lens or the window are removed. If this happens,
the cell should be painted again before use.

A summary of some of the figures of merit for the LED detector. as
determined from the experiments presented in Chapter N, is included in
Table 5- l.

Table 5- 1:

Experimental Results of a Single Photometer

 

 

 

of the Multichannel Detector
Specification Value Comments
Wavelength of 6 10 nm Other wavelengths available
analysis by exchanging LED's.
Spectral bandpass 50 nm Small variations from LED
to LED.
Photometric 2.0% RSD of repeated
precision measurements.
Photometrichnear 0 to 0.9 with the LMG detection
range (A.U.) reaction.
Stray light 0. 1% or a definite value could
less not be determined.
Baseline noise 0.002 with a flowing liquid.
level (A.U.)
Baseline drift 0.001 After 15 min. warm-up.
(electrical)
(AU. / hr.)
Baseline drift 0.001 After 1 hr. of use.
(due to flow)
(A.U. / min.)

'A.U. = absorbance units

10.
ll.

12.

13.

14.

15.

REFERENCES

AQAQ Qffigigl Mgmggs Qf Anahgis. 13th ed. Association of Oficial
Analytical Chemists. Washington D.C.. 1980.

R. Thompson. Ph.D. Dissertation. Michigan State University. East
Lansing. MI, 1982.

C. J. Patton. Ph.D. Dissertation, Michigan State University. East
Lansing. MI, 1982.

C. L. M. Stults. Ph.D. Dissertation, Michigan State University. East
Lansing. MI, 1987.

K. Kurtz. Ph.D. Dissertation, Michigan State University, East
Lansing. MI. 1992.

{MAT 1918' MS. Thesis. Michigan State University. East Lansing,

J. Ruzicka and E. H. Hansen, Flow Injection Analysis 2nd Edition.
John Wiley and Sons. New York (19 ).
M. Valcarcel and M. D. Luque de Castro. 3% Injggggn Analysis;

Prgé i l A li . John Wiley and Sons. New York
1 7 .

B. Karlber , Anal. Chim. Acta 180. 16 (1986). (In fact. the entire
volume 18 of Anal. Chim. Acta is dedicated to FIA).

J. Ruzicka. fiesenius Z. Anal. Chem 125;, 745 (1986).

F. J. Krug. H. Bergamin F9, and E. A. G. Zagatto. Anal. Chim. Acta
122. 103 (1986).

M. Gallego, M. D. Luque de Castro. and M. Valcarcel. At. Spectrosc.
5. 16 (1985).

H. Matschiner and P. Sivers. Wiss. Fortschr. 35, 142 (1985). [in
German].

B. Karlberg and G. E. Pacey. Fl In An i
guide, Elsevier. Amsterdam (1989).

H. Poppe. Anal. Chim. Acta 114,. 59 (1980).

96

16.
17.

18.

19.

20.
21.

22.

23.

24.

25.

26.
27.

28.

29.
30.

31.

32.
33.
34.

35.

97
Y. Baba. N. Yoza. and S. Ohashi. J. Chromatogr. .318. 319 (1985).

(ZigPéafig. J. Ruzicka. and E. H. Hansen. Anal. Chim. Acta 1%. 23

S. D. Hartenstein. J. Ruzicka. and C. D. Christian. Anal. Chem 51.
21 (1985).

8.9%.619400dy. J. M. Slater, and J. D. R. Thomas, Anal. Proc. 23, 287

J. Ruzicka and E. H. Hansen. Anal. Chim. Adam. 3 (1985).

flag]; Light Emitti_ng Diodes. Sharp Technical Data Sheet. Ref.
No IH'I‘504D. 1987.

fiéglgschka. C. McKeitken. and R. Barnes. Anal. Lett. 6. 585

EWL A. Graneli. and M. Strandberg. Anal. Chem 48. 357

D. Betteridge. E.L. Dagless. B. Fields. and N.F. Graves. Analyst
193. 897 (1978).

T.J . Sly. D. Betteridge. D. Wibberley. and D.G. Porter. J. Autom
Chem 4, 186 (1982).

D. J. Hooley and R. E. Dessy. Anal. Chem 55. 313 (1983).

M. TYo anowicz. W. Augustyniak, and A. Hulanicki. Mikrochim Acta
2. 17 1984).

(“926)IMque de Castro and M. Valcarcel. Rev. Anal. Chem 5. 71
1 .

M. D. Luque de Castro and M. Valcarcel. Analyst 153, 413 (1984).

M. Valcarcel, M. D. Lug ue de Castro. F. Lazaro and A. Rios. Anal.
Chim. Acta 216. 275 (1! 89).

C. L. M. Stults. A. P. Wade. and S. R. Crouch, AnaL Chim. Acta
122.. 155 (1987).

P. T‘rinder, Ann. Clln. Biochem 6, 24 (1969).
C. J. Patton and S. R. Crouch, Anal. Chlm Acta 179. 189 (1986).

J. Agranoff (ed.). Modern Plasggs Engglgpegia. McGraw-Hill. New
York (1983).

M. 1. Ash. compiler. Eng clo ediafllastigs. Poi ers. and Reins
Vglumg II, Chemical Pu 115 wing Co.. New York (1982).

 

 

 

 

 

 

 

 

 

 

36.

37.

38.

39.

40.

41.
42.

98
ElgggQ-thics Handbook. Technical Series EOH-l 1. RCA Corp..
Lancaster. PA.. 1974.
Photodiode . Technical Data Sheet, EG&G Photon Devices. Salem
MA.
C. E. Evans. J. G. Shabushnig. and V. L. McGumn. J. Chromatogr.
452. 1 19 (1988).

.(JiQLaoovsm. R. Koile. and D. Johnson. Anal. Chim. Acta 116. 33

fibgggamin. B. Reis. and E. Zagatto. Anal. Chim. Acta 92. 427

H. Poppe. Anal. Chim. Acta £15. 17 (1983).

J .D. Ingle and SR. Crouch, Spggggchemigal Anahzsis. Prentice
Hall. New Jersey (1988).