-1——, :~':‘~1‘"'11111M,W 11
'5!” ”if“ , “'yf'" ”Youhlr, 9.11 W

' ,U . “my“;fi',

V‘ “16“.
r‘b’hzv J k1
1,3,; 111171,” I" :1}. 11195113.?”

11' :1 ' ""vui‘“
11. 1.3.1111 ‘ f1.”1'.1.'!'1 “1‘11 11:11 11131131;
1 13.111.11.11, $11171 11111111: ,1‘ {£11 117:? “111.» f2}, ”13.3111” '
.. 11‘11‘31111',1""!“"9 1I'1'11H1.’11.~'I’11111.1 11111211111111 "7'11'1 11
11, TEE/f!“ .o(.q ‘1‘}: “J1 "1’5‘.|:I1:'- ‘N’T‘ I),”'{l "'1‘
.\ 1' 3111'." 1.. 11-111." ‘1'1~I.[1r1"'zr111'{'1‘J371-.1,21 '1- i 111911-11”!
- o 1-1,] ‘ ‘

€212:

        
      
  
  
 

   

 
  
   
 
 
 
  
  
 
  
  
    
  
  
   
    
   
   
     
  
 
    
   
  

 

“’11 x , 9:91;? '1 “fig!
.111 mfim¢MW 1 1M
1 1" . ‘ 1 1111,. 1111I 11.1.11“: [11" 11.1 [11,7 151%; 1111]“ Eggh‘r 3‘51 5112'1'1—11‘1‘11?’ “300' W 161};
,‘,.‘1‘1.,,'1.1',‘..1,111f:,1.111,111'l1:~1,,,1 111 1111111111111. 11511,,W,Em“, '1'111'131 111,3,11131‘7IU1P1 11:}. 1,21% - $.11 . , 1,? .,
-1 \ 1 , |1 ' " 1 ' L1:
111"" 11.": 1"‘l‘.""1 1"‘111 111111'11’111151"?’11' {“1611}!111111131'117112111’5117' '11P: ‘1
”1,111. $1.11, ,"‘1,1,'Ul$1’11111}1 (1,11! ’3'. ,1‘11.I 1,!” . - 11:1!
. .1 ”1,1,,“ “'1‘ 1"1'1’111'L1Jqui'!” . 10%,)?! I . 1‘ _. ‘ "iwwwgq! '1‘) .1)’ I fit
1 1.11.131“ 111’” 1“,“, 11111, J}? . , ’11, I, 1 1;? 3‘ 1’1" fij‘rfiflm .
1 .1 11-11. . 1 -~ 1. .1 =11
'11 11H 1 11:11.,11'15 "‘ I: 11. 1,1116}; . , .1 ,1 .11 .1 5,, V 1 .,
. “1,1,, 11“ "1"..0‘11‘1“%1fa 1.! *1 1 1 . ', I I . ‘
111, ..11 "3.1111 111111,? . ' 15 I ' '1
1
1111 11.11111" ”‘33? 1%
I ‘1'):1'1-711111'25 “5‘”: $11.,

1'”, ‘9 | h

1’]; 11'1'$J_1111$#'£i,11
1 'l‘ ..

111
~~~~~ 11;!" 11,711

1..

   
   
  

(1,325,; {lifiiffi}. 1: .1!" ' :1.
. 1‘1'.’ 1 '1' ‘
11": 1111’ 111111111 1

 
  
  

   
   
  
  
  

1 '1 ' i ‘11!
1211' 11.11.: 1,,1 . 1 .11 ', ', , .,,-.,.1
1' 111111 . 1.1111111.111111111171511 "-1.1 1'. ' ‘ ‘ “
‘ ‘-' ‘ :l', .‘ Pitt 1:": £..‘ ;.'.. , ,’.. . ’ :1" |
. .. ‘ , 131111111 ,1'11.,1,,,,'1 1111,121'1111‘1” 1.,.,1...,..,1 .111 .1 ,. ,,
1 ffilflw 1.1, ",1;,',J,'1;]'1‘1,1,’ 11,511,115 .' '. MH-
.1311] ‘11} ["1113 11111311” 1‘18}: .1, £11
«11 1 1 1.

  

' ‘1 111355;; {1" ‘ '

,1 . 11,1)“, 1,.1r1{.'11111:1 1 ,. 2,1 1.
7.1.1111, \,,(;YJ,fp:.'1::‘;£ 111117.]?1-31! r. 1'; :: . '21.:‘1', ; ’
1. 1 ,Ll‘l ’13:, "I,“ Fig; 11):! :1: fi:;|.1:,:1§:t;f.~rr:. . :5, ' 1.1 .
h 1") I'r‘;:}‘13"11‘"l" .‘.g -.
"1‘ .1' 1311119111.:1’

  
  
 

11 ' ,1
131,,

"'jgrg-

-~

  

  
  
      
   
  

 
    
 

      

1. 1,.
. ,(f'

.- 3-4—
' "'5: ‘
~ —‘
‘ 5%.:
1 ““2,
2-:
.5
.4,

l
1 n

. 11,1515 {-1.1 4,31,. ‘11 1.1%.» 1,
3 41‘ 81 1131,11}: ‘ ,,,J, “111.,“ ’151111313’3‘ by},
1111131"! 1,41,11,111‘2111‘1111“. ’1 1W1; 1511.111 1111 w{1—31521131111,3
1111 ”1,1,1 111»; 131’ .. 311 . .11., 1:.

i“ ”1,, [1,1113% .11 1111111311‘11111"11Ji x,
“'1' 1111‘“ '11 (21.1.1 1
1'11 ".1‘

" ';1~13fi! 31?. 1’11".)1
#INW'JM“! #151 111
111 11.,{.?.1’1",,, 1137‘"?

:11, ,I’ V‘- r‘ 1
. 1 ~11
,h’Iufi’IL g pt ’1‘“ x {.fi’: ., .3112; I; Sikh‘ : 9* ‘Q
1111““ 1111' - ~ 1 1111’. *
I .
:3- 1 o L

111111,,1 '.
1111‘ '1 .

1"

   

.....

   

  

 
 

1111111911111111111111111111111111111

00669 4958

 

LIERARY
Michigan State
University

 

 

 

This is to certify that the

dissertation entitled

PRACTICAL CONSIDERATIONS FOR THE USE OF IMMOBILIZED ENZYMES
IN FLW INJECTION ANALYSIS

presented by
Cheryl Lynn Mattson Stults

has been accepted towards fulfillment
of the requirements for

31.12. degree mm

/£Z/éwc

Mafor professor

Date—WBL—

MS U i: an Affirmative Action/Equal Opportunity Institution 0-12771

 

 

 

MSU

 

 

RETURNING MATERIALS:

 

 

 

Place in book drop to

 

 

LJBRARJES remove this checkout from
—:-—. your record. FINES will
be charged if book is
returned after the date
stamped below.
«rt-fr

 

 

PRACTICAL CONSIDERATIONS FOR THE USE OF IMMODILIZBD ENZYMES
IN FLOW INJECTION ANALYSIS

By

Cheryl Lynn Mattaon Stults

A DISSERTATION

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

DOCTOR OF PHILOSOPHY

Department of Chemistry

1987

Copyright by
CHERYL LYNN MATTSON STULTS

1987

ABSTRACT

PRACTICAL CONSIDERATIONS FOR THE USE OF MBILIZED ENZYMES
' IN FLOW INJECTION ANALYSIS

By

Cheryl Lynn Mattson Stults

Numerous practical issues concerning the use of inobilized
enzymes in flow injection analysis (FIA) have been investigated.
Inobilized glucose oxidase (GO) was used predominantly. The enzyme
reaction was coupled with a modified Trinder reaction in an FIA system.
The basic components of the apparatus were: a peristaltic pump, a
pneumatically actuated sample injection valve, single head string
reactors (SBSRs), and a miniaturized filter calorimeter. Data
acquisition, sample injection and pump speed were controlled by a
microcomputer .

The procedure for inobilizing GO on nonporous glass beads was
studied and improved. Simplex optimization was used to find an optimum
set of operating conditions for the FIA system containing a G0 SBSR.
Seven variables were specified: carrier pH, reagent pH, temperature,
flow rate, aminoantipyrine concentration, dichlorophenolsulfonate
concentration, and peroxidase concentration. The results were similar
to those obtained by univariate experiments.

Several extended applications of the GO SBSR were investigated.
The enzyme coated beads were used to make a microconduit SBSR. The
response obtained was equivalent to that of the conventional SBSR. A Go

SBSR was used to study the mutarotation of D-glucose in the presence of

phosphate. The mutarotation coefficients obtained were in good
agreement with the literature values. Galactose oxidase (GA) was
i-obilized by a procedure similar to that for glucose oxidase. A GA
SBSR was combined with a,G0 SBSR to make a parallel FIA system. The
results obtained from the analysis of real samples matched known values.

Two other issues that are related to the use of i-obilized
enzymes were also studied. First, the effect of temperature on
dispersion was examined for four classes of reactions—none, fast,
medium, and slow. A dye was used to establish the effect of
temperature. The reactions were: p—nitrophenol with sodium hydroxide,
nickel with 4-(2-pyridy1azo)resorcinol, and the GO/Trinder reaction.
The results indicate that it may be possible to calibrate FIA systems
for temperature changes. Second, a partially automated system on which
flow reversal/merging zones experiments may be carried out was designed
and characterized. The results illustrated the advantage of using a
second detector and two dimensional mapping of reaction surfaces fram a

single injection.

ACKNOWLEDGEMENTS

I’m glad for the opportunity to officially thank those who have
contributed to this achievement in my life. First and for-0st, I am
thankful for Professor Crouch’s continuous, positive encouragement and
patience. Special thanks are extended to Professor Watson fer being my
Second Reader, and to the other members of my Guidance Committee
(Professors Dye and Nocera) for their helpful suggestions and insight.

I am indebted to Dr. Adrian Wade for his friendship, gentle
pushing, and essentially "taking me under his wing". Without the help
of Paul Kraus (AKA: my cohort in crime) I might still be taking data on
a chart recorder. A big thank-you is in order for letting me use his
version of the "Bruce Bus", and for the camaraderie we’ve shared. The
friendship and support of the other members of the Crouch group is also
much appreciated.

It was a special privelege to teach for Professor Timnick,
Professor Allison, and Dr. Funkhouser. A special note of thanks to
Marty Rabb and the other members of the technical support groups in the
Chemistry Department for going out of their way to be helpful. I
also grateful for the graphics assistance provided by Bill Draper.

I owe special thanks to Dr. Peggy Dine and other teachers who have
encouraged me throughout my life to pursue goals that I mistakenly

thought were unattainable. I must also thank my mom and dad for always

believing in me and lending a hand when things got a little crazy in
this last year. Last, but certainly not least, I am deeply grateful to
my husband, John, for his constant support and taking care of the
details of our personal life so that I could complete my work; and to
my son, Michael, for providing happy moments of diversion during my

writing binges .

vi

TABLE OF CONTENTS

CHAPTER . PAGE
LIST OF TABLES ................ . .............................. ....x
LIST OF FIGURES ................................................. xi
WHAT’S IN THE BOOK ............................................... 1
BACKGROUND INFORMATION.... ....................................... 3
Introduction to I-obilized Enzymes. . ........................... .3

Methods of Enzyme Immobilization ........................... 4
Analytical Applications of Immobilized Enzymes ............. 5
Introduction to Continuous Flow Analysis. . . ...................... 7
Principles of Flow Injection Analysis ............ . ......... 7
Types of FIA Manifolds ..................................... 9
Measurement Tools in FIA .................................. 10
Use of I-obilized Enzymes in PIA ............................... 12
THE GLUCOSE OXIDASE SBSR. ....... .......... ...................... l4
limobilizntion Procedure far Glucose axidhse .................... 16
Cleaning Procedure ........................................ l7
Whisker Formation. ........................................ 19
Silanization .............................................. 20
Glutaraldehyde Attachment ................................. 21
Enzyme Attachment and SBSR Construction ............... . . . .21
Alternate Method of I-obiJization .............................. 22

vii

CHAPTER PAGE

Determination of Amount [mobilized ............................. 24
Ca talase Acti Vi ty ............................................... 27
Specificity of [mobilized Glucose Oxidase ..... . ................ 27
4 OPTIMIZATION OF THE FIA SYSTEM .................................. 31
Instrmentation.................... ....... . ..... . ..... . ......... 33
Choice of Fixed Parameters ................................ 33
Temperature Calibration .............. . ......... . . . . . ...... 39
Microcomputer Interface to FIA Syst- ..................... 39

Data Acquisition. . . . ...................................... 45 .
Improvement of System Performance ............................... 45
Univariate Experiments .................................... 47
Delay Flow Experiments .................................... 58
Simplex Optimization ..... ................ ..60
5 EXTENSIONS OF THE GLUCOSE OXIDASE FIA SYSTEM ...... ..............75

Design and Tea ting of the Microcondui t Glucose Oxidase SBSR. . . . . 7 5

Ascorbic Acid Interference Study ............. . .................. 78
Kinetic Study of D-Glucose Mutarotation. . . . . . . ............... . . . 82
Determination of Mutarotation Coefficients .............. ..87
Effect of Phosphate Concentration ....................... . .88
Effect of Galactose ....................................... 91
Parallel Determination of Sugars ...... . ...... . .......... . ..... . . 92
The Galactose Oxidase SBSR ................................ 92
Inobilized Galactose Oxidase Specificity ............ . . . . . 93
Design of the Parallel FIA Syst- ...................... . . .95
Analysis of Real Samples .................................. 96

viii

CHAPTER PAGE

6

THE EFFECT OF TEMPERATURE ON DISPERSION ........................ 101
The Apparatus for Cases 1 through 3 ............. _ .............. . 104
Case 1: No Reaction ........................................... 107
Case 2: Fast Reaction ......................................... 111
Case 3: Moderate Reaction ..................................... 113
Case4.‘ SlowReaction................ ......................... 119
DEVELOPMENT OF A FLOW REVERSALS/MERGING ZONES FIA SYSTEM ....... 124
me FIA Flow Reversals System .................................. 125
Calibration with Dye ........................................... 127
Extension to a Chemical System. ................................ 134
WHERE DO WE GO FROM HERE? ...................................... 136
LIST OF REFERENCES ............................................. 139

ix

TABLE

10

11

LIST OF TABLES

PAGE
Examples of immobilized enzymes. ................................ 13
Initial immobilization procedure ................................ l8
Glucose oxidase activity toward other sugars.... ............. ...29
Coefficients of the Steinhart-Rart equation
for different flow rates ........ . ............ . ......... . ........ 42
Range and precision of variables for simplex optimization ....... 65
Initial and optimal experimental conditions
for the FIA determination of glucose .............. . ............. 67
Effect of ascorbic acid in experiments without Brij-35. ......... 81
Mutarotation coefficients for D-glucose at 21°C ................. 90
Galactose oxidase activity toward other sugars. ................. 94
Temperature correction equations for various flow rates........106
Temperature effect on glucose oxidase/Trinder reaction ......... 122

LIST OF FIGURES

FIGURE PAGE

1

10
11
12
13

14

15
16

17

Comparison of FIA profiles obtained with an SBSR and an
open tube. Signal traces are only in the region of

interest (see text for further explanation)....... ............ ..11
Single bead string reactor....... ............................ ...15
Final immobilization procedure .................................. 23

Loss of glucose oxidase activity for routinely used

heads (a) and stored beads (A) .................................. 26
Actual (- -) and expected (--) effect of catalase

in a 20-cm SBSR (O) and a 30-cm SBSR (o). .. .. ................... 28
Glucose oxidase/Trinder reaction.............. ...... . ........ ...32
FIA system for glucose determination ............................ 34
Effect of varied glucose sample size.... ...... ........ .......... 37
Effect of flow ratio (Buffer : Carrier) on FIA profile .......... 38
T-piece containing 30-kn thermistor ............................. 40
Thermistor response curve ....................................... 41
Effect of varied flow rate ...................................... 48
Effect of AAP/DCPS ratio ............. . .......................... 49
Three dimensional surface of the FIA peak

for the Trinder reaction........ ................................ 51
Effect of peroxidase concentration .............................. 53
pH dependence of GO/Trinder reaction with Universal buffer ...... 54

pH dependence of GO/Trinder reaction with
phosphate (- -) and borax (—) buffers .......................... 55

xi

FIGURE PAGE

18
19
20
21
22
23

24

25
26

27

28

29

30

31

32

33

. Temperature response of a lO-cm GO SBSR ......................... 57
Results from delay flow experiment .............................. 59
Two dimensional simplex example ................................. 61
History of simplex optimization .............................. . . .63
Progress of the simplex ......................................... 69

Effect of experimental variables on response function:

(a) carrier pH; (b) reagent pR; (c) flow rate, ml min“;

(d) SBSR temperature, “C; (e) DCPS concentration, i4;

(f) AAP concentration, In; (g) PO concentration, mg ml‘1 ........ 71

Calibration curves for optimum operating conditions arrived at
by simplex optimization (0) and by univariate experiments (0). . .73

Side (a) and top (b) views of a microconduit SBSR ............... 77
Mutarotat ion process for D-glucose. . . . . ....... . . . . .............. 84

Effect of phosphate concentration on rate of D-glucose
mutarotation. Lines represent curves of best fit.

Upper curves were obtained with only B—D—glucose initially
present. Lower curves were obtained with only c-D-glucose
initially present. Phosphate buffer concentrations were:

(Q) 0.00 M, (O) 0.05 M, (A) 0.10 M, (I) 0.20 M .................. 89

FIA system for parallel determination of glucose and galactose. .97

FIA profiles for : (a) Pepsi, (b) Sunglo fruit punch,
(c) diluted serum, (d) deproteinized serum ...... . ............. ..98

FIA system for temperature studies ............................. 105

Effect of temperature on dispersion without reaction.
Flow rates were: (0) 0.56 ml min‘1 , (A) 1.20 ml min’1 ,
and (I) 1.88 ml min‘1 ........................... .. ............. 108

Effect of temperature on peak area without reaction (a),

peak area with fast reaction (b), peak height without

reaction (c), peak height with fast reaction (d), m

without reaction (e), and mm with fast reaction (f).

Flow rates were: (0) 0.56 ml min'1 , (A) 1.20 ml min"1 ,

and (I) 1.88 ml min“1 .......................................... 110

FIA profiles obtained for the experiment with p—nitrophenol
under the following conditions: flow rate 1.20 ml min”;
coil length 3.0 m; and temperature range 20-70‘C ............... 114

xii

FIGURE

34

35

36

37

38

39

40

41

42

Reaction progress curve for nickel (2x10'5 M)

with PAR (10" M) at pH 6.85 ................................... 116

Effect of temperature on peak area (a) and peak height (b)

for medium reaction with l-m (O) and 3-m (A) coils ............. 118

Effect of temperature on peak area(a) and peak height (h)
for slow reaction. Concentrations of standards were:
(0) 1.65 IN and (A) 3.3 nM glucose; (0) 0.14 m and

(A) 0.28 IN H202 ............................................... 120

FIA system for flow reversal/merging zones experiments ......... 126

FIA profiles for two reversals and one detector.
The three injections were: (a) 81 only; (b) 82 only;

and (c) 81 plus $2 ............................................. 129

Contour plot for two reversals and one detector ................ 130

FIA profiles for three reversals and two detectors.

The three injections were: (a) 81 only through the first
detector; (b) 81 only through the second detector; (c) 82
only through the first detector; (d) 82 only through the
second detector; (e) 81 plus 82 through the first detector;

(f) 81 plus 82 through the second detector ......... 9 ............ 132

Contour plot for three reversals through the first

detector (0) and the second detector (A). . . . . .................. 133

Six channel parallel FIA system for simultaneous
determination of glucose, galactose, maltose,

lactose, sucrose, and fructose ................................. 137

xiii

CHAPTERl

WHAT’S IN THE BOOK

The i-obilization of enzymes is no longer a novel concept.
Application of these reusable reagents has gained notoriety in the
fields of food processing, clinical chemistry, and analytical chemistry.
Research in the area of i-obilized enzymes has focused primarily on
i-obilization procedures, reactor designs, and analytical applications.
The research reported in this dissertation focuses on the practical
considerations that must be made in order to utilize i-obilized enzyme
technoloa to its fullest.

This work centered around the use of glucose oxidase (GO)
i-obilized on nonporous glass beads in the form of a single head string
reactor (SBSR). This enzyme is inexpensive, has good activity, and can
be bonded to a variety of insoluble substrates. Primary consideration
was given to the use of the SBSR in a flow injection analysis (FIA)
system. In this system the GO reactor is coupled with a reaction that
utilizes the hydrogen peroxide formed from the enzyme reaction to
produce a colored dye.

Fundamental background information is provided in Chapter 2. A
review of current i-obilized enzyme methodology is provided along with
a more specific orientation to the analytical applications of these
reagents in FIA. A brief review of the principles of flow injection
analysis is presented; the SBSR is introduced.

Chapter 3 describes the characteristics of the GO SBSR. A more

efficient i-obilization procedure is enumerated. The activity, useful

lifetime, catalase activity, and specificity of the reactor are
discussed in detail. Another matter of practical interest, optimin
operating conditions, is treated in Chapter 4. Three methods by which
the response of the FIA system can be increased are presented. Several
extended applications of the system are given in Chapter 5. The
potential for using the i-obilized enzyme beads in a microconduit is
explored. A preliminary investigation of ascorbate interference is
described. A novel use of the system, to study a batch reaction, is
presented. Finally, the knowledge gained from the work on glucose
oxidase was applied to galactose oxidase and resulted in a parallel
analyzer. The design, construction, and use of the system for the
simultaneous determination of two sugars in real samples are discussed.

Two topics of practical interest are presented following the
enzyme work. In Chapter 6 the effect of temperature on the FIA profile
is exuined. Four different reaction conditions are presented. The
design and characterization of an FIA system which codines the flow
reversal technique with merging zones is discussed in Chapter 7.. Ideas
for future projects are given in Chapter 8.

The practical information obtained in the course of this research
should prove beneficial to those who continue to work with i-obilized
enzymes in coming years. It should be possible to extend the procedures

developed here to other i-obilized enzyme syst-s.

CHAPTER 2

BACKGROUND INFORMATION

This chapter provides a condensed review of the topics with which
familiarity is assumed in the subsequent chapters. There are two major
areas of focus: i-obilized enzymes and flow injection analysis (FIA).
The first section gives the reader a perspective on the general
methodology employed with i-obilized enzymes in analytical chemistry.
More complete information on the subject of i-obilized enzyme
technology can be found in recent books (1-11) and general review
articles (12-18). Specific procedures for enzyme i-obilization can be
found in Reference 8. The second section deals with some of the
fundamental aspects of FIA which are co-on to all of the systcs in the
following chapters. Discussion of subtopics which relate only to the
material in one chapter are delayed until then. The authoritative book
on FIA by Ruzicka and Hansen was published in 1981 (19). A recent
monograph by Valcareel and Luque de Castro (20) provides a review of
current FIA technology. An excellent bibliography containing 811
references to the work accomplished in this field was published in a
recent review written by the sue authors (21). A complete bibliography
has also been published by Tecator which covers the period from 1974
through February 1985 (22).

Introduction to I-obilized Enzymes

An enzyme is a protein which acts as a biological catalyst. It

differs from an ordinary catalyst in that it has a high turnover rate at

ordinary tmeratures, is selective with respect to a particular
reaction pathway, and may be specific for one or for only a few
compounds. Enzyme solutions are useful as analytical reagents, but they
have a few major disadvantages. Some enzymes are unstable in solution,
most cannot be used in organic solvents, and all are very sensitive to
elevated temperatures. Enzymes are expensive reagents. (hoe an enzyme
is used in solution it is not easily recovered, therefore making the use
of these catalysts rather uneconomical.

The i-obilization of these biocatalysts has been a mjor factor
in the popularity which they have achieved in the past few decades. By
rendering the enzyme water-insoluble many of the disadvantages of the
soluble form are eliminated and some advantages gained. The enzyme
retains its activity over a broader range of pH and temperature. The
enzyme is reusable in this form and so provides an economical
alternative to other methods. An additional advantage of this feature
is that the enzyme is easily manipulated. It has been suggested that
there are fewer interferences with i-obilized enzymes (11); thus,

comlex samples are relatively easy to analyze.

Methods of Enzyme I-obilization

The methods for enzyme i-obilization fall into four categories:
adsorption, cross-linking, entrapment, and covalent binding. The
initial discovery by Nelson and Griffin of enzyme activity retention
upon i-obilization involved the adsorption of invertase on charcoal
(23). This type of i-obilization occurs any time an enzyme is adsorbed

on some type of insoluble support. The chemical conditions required for

this process are very mild. However, as easily as the enzyme adsorbs to
a surface it also desorbs. This is a significant problem if the
i-obilized enzyme is to be used in a system where it is exposed to a
continuous flow of solution.

Entrapment provides a sturdier environment than adsorption. A
co-on method of entrapment involves creation of a polymer in the
presence of the enzyme. The enzyme becomes enclosed in a lattice.
Polyacrylnide is often used to i-obilize enzymes in this way. Gels of
this type are co-only used in enzyme electrode applications. An
alternative type of entrapment is accomlished by encapsulation of the
enzyme in a semipermeable polymeric medarane. This usually involves a
tedious and time consuming i-obilization procedure.

A somewhat more harsh chemical method is cross-linking. The
enzyme is covalently bonded with a bi- or multifunctional reagent to
create an intermolecular cross linkage. Glutaraldehyde is co-only used
as a linking agent.

0f the four methods of i-obilization covalent binding of the
enzyme to a water-insoluble support provides the sturdiest environment.
This is the most co-only used method of i-obilization. Some type of
chemically modified glass provides an ideal inert support. The methods
of enzyme attachment to the glass are as varied as the nui>er of

applications.
Analytical Applications of I-obilized Enzymes

For a complete review of the many ways in which i-obilized

enzymes have been used in analytical chemistry the reader is referred to

the book by Guilbault (11). There are two main areas of application.
I-obilized enzymes have been coupled with amperomstric or
potentiometric electrodes in a nwer of ways. The enzyme may be bound
directly to the surface of the sensor, but is more co-only attached in
the form of a membrane. These probes have gained great popularity in
biological and clinical applications.

I-obilized enzyme reactors (IMERs) constitute the other major
area of application. Many types of such reactors have been developed.
They take the form of packed beds, open tubes, and single strings of
beads. Packed bed reactors typically contain the enzyme i-obilized on
very small particles (37-74 in) which are packed into a column. They
have been used in batch as well as continuous flow methods. The great
advantage of these reactors is the large surface area over which the
enzyme is spread. However, when used with flowing streu, these
reactors may develop channels and high dispersion may result.

Open tubular reactors contain the enzyme i-obilized on the inner
surface of the tube (usually about 1 - i.d.). These are co-only used
in air-segmented continuous flow analysis. The small surface area
available per length of tubing results in very low enzyme activity.
There have been various attempts to increase the surface area of the
inner wall of the tube. Onuska et a1 (24) demonstrated the use of
mania hydrogen difluoride to grow whiskers on the inside of glass
tubes. A recent report indicated that controlled pore glass particles
had been edaedded on the walls of plastic tubing (25).

The single head string reactor (SBSR) is used in flow injection

analysis applications. Such reactors are made by i-obilizing the

enzyme on porous or nonporous beads. These are packed into a tube which
has an i.d. that is 1.3—1.7 times the diameter of the bead. The
dispersion in such reactors is low while the enzyme activity is high.
More will be presented about this type of reactor in the following

section.
Introduction to Continuous Flow Analysis

Continuous flow analysis is the technique by which a sample
undergoes treatment as it is carried by a flowing stream through a
manifold made of interconnected tubing. It can be divided into two
major areas: aimegmented and non-segmented continuous flow analysis.
Skeggs is accredited with the development of the former technique (26),
which is most often referred to simply as continuous flow analysis
(CFA). In this, air segments are introduced into the analytical stream
at regular intervals so that the stream is divided into a train of
identical segments. The sample is introduced onto the manifold by
aspiration for a defined period of time. In each segment complete
mixing takes place so that the signal obtained at the output has a
rectangular shape similar to what would be expected in the ideal case of
a plug-shaped sample. This technique has been used extensively in
clinical analysis applications; the best known co-ercial CFA instrument

is the Technicon Autoanalyzer.
Principles of Flow Injection Analysis

The other branch of continuous flow analysis was developed by two

separate groups, but has since emerged as a singular technique known as

flow injection analysis (FIA). The approach of Stewart, Beecher, and
Rare (27) was a conceptual outgrowth of liquid chromatography, whereas
that of Ruzicka and Hansen (28) was founded as an alternative to CFA.
In any case, the driving force for the development of this technique was
the ability to manipulate and reproduce dispersion of the sample. This
technique involves the injection of a snple into a nonsegmented carrier
stream. Since the conditions are usually such that laminar flow is
predominant, the development of a parabolic velocity profile is
responsible for the dispersion of the sample along the axis of the tube.
This dispersion, although much greater than that found in CFA, can be
controlled by appropriate choices of tubing length and inner diameter,
flow rate, samle size, and other components such as valves and
flowcells which determine the overall volume of the reactor in the FIA
system.

Ruzicka and Hansen have proposed an empirical method by which
dispersion can be measured (19). The ‘sample has an initial
concentration of C“ as it enters the carrier stream. As the plug
travels through the manifold, axial and radial mixing take place. This
results in a predominantly Gaussian-shaped signal profile. The maxim
concentration sensed by the detector, 0'“, is only a fraction of 0".

The formula for the dispersion is:

The height of the peak obtained with the undiluted sample is H’. After
the sample has traversed the manifold a lower peak height, H, will be

obtained due to dispersion. If the two constants are equal, as in the

case of photometric detection for a system that adheres to Beer’s law,
the peak heights of the signals can be used to determine the dispersion
of the FIA cyst... This empirical value is classified as limited
(D = 1-3), medium (D = 3—10), or large (D > 10). The amount of
dispersion that can be tolerated in an FIA system depends on the
application for which it is intended. For more transportation of a
sample, limited dispersion is ideal. On the other hand, when a chemical
reaction requiring reagent addition must take place, medium or large

dispersion is desirable.
Types of FIA Manifolds

One way of manipulating dispersion is by selection of the
appropriate type of manifold. A comparison between various types bf FIA
manifolds was carried out by Ruzicka and Hansen (29). Straight open
tubes yield relatively large amounts of dispersion. Coiled tubes show
less dispersion due to the presence of secondary flow. This type of
flow is a result of the centrifugal forces which affect the flow
perpendicular to the axis of the tube. The sue phenomenon is observed
in 3-D coiled tubes (tubes with regularly spaced knots tied in them). A
relatively small amount of dispersion has been found in the use of
packed tubes. This is due to the fact that the parabolic profile is
broken up as the sample passes through the packed material. The high
pressure drop associated with tubes that have been packed with very
small diameter particles makes them difficult to use with the

peristaltic pumps normally present in an FIA system.

10

The single head string reactor (30) has gained acceptance as a
viable alternative to open tubes and those packed with small particles.
The SBSR consists of a tube packed with beads that have a diameter 603
to 803 of the tube inner diameter. This type of reactor has the
advantage of decreased dispersion due to the break up of the velocity
profile. The pressure drop is small. Therefore the SBSR can be used to
provide longer residence times without an increase in dispersion. An
example of typical signal profiles obtained with an SBSR and an open
tube of the same length and volume is shown in Figure 1. (Each length
of open tubing had the same internal diameter). Although the arrival
time of the peak at the detector did increase with reactor length, the
point of injection is not given for the sake of simplicity. For use
with i-obilized enzymes an added advantage of the SBSR is the

additional surface area available compared to that of an open tube.

Measuraent Tools in FIA

Following the development of an FIA system with the appropriate
amount of dispersion, the analyst must choose a measurement tool from a
variety of peak characteristics. The complete signal is usually
recorded either on a chart recorder or by a computer data acquisition
system. The most co-only used measurement is that of peak height. In
this case the FIA syst- may be simplified by the use of a peak
detector. In some cases the height at some other place on the peak than
the maximum is used (e.g., in gradient techniques). The full width of
the peak at half the maximum height (FWHA) can be used as a measure of

sample dispersion. Peak area may also be used as an analytical

ll

 

Assessed—Xe bonanza you 38»
005 000.395 mo sewn»: one 5 3.5 one
soon: 133m .04.... some as was mama as
:33 vocwsano 33.3.3 <2 .«o canton—loo

8mm. m2;
n&_og om.aen§ o~ o

ELL-PE

 

 

mm? zumo

 

 

mo...u<wm o<wm 025.5
.5 2. . so 9. .58

a1 a.

H mun—Gum

 

j . COCO
Woouo
W83

W88

 

 

W83
.5 o.

EDNVBHOSBV

12

measurement tool. This measure is most easily obtained by a
co-ercially available integrator or by computer processing of the
acquired signal. The time at which the peak maximum occurs has been

found to be useful in kinetic measurements (31).
Use of I-objh'zed bay-es in FIA

Hicks and Updike used a nonsegmented flowing stream with an
i-obilized enzyme column prior to the development of the technique
called, ”flow injection analysis". Glucose oxidase was immobilized in a
polyacrylamide gel and packed into a miniature chromatographic column.
Colorimetric (32) and electrochemical (33, 34) detection methods were
employed.

Cremonisi was the first to inject a sample into a nonsegmented
stream which flowed through an i-obilized enzyme reactor. (35). Glucose
oxidase was peptide-bonded to CNBr-activated cellulose and packed into a
small HPLC-type column. Since then FIA has become increasingly popular
for use with inobilized enzymes. The references listed in Table l are
just a few of the more recent examples. The works listed are mainly in
the area of sugars’ analysis since that is the thrust of the work
presented here. For a general overview of the use of immobilized

enzymes in FIA the reader may consult References 45 through 47.

l3

 

onstaxouom apnea ososhusocwmshuom ea

 

 

 

 

 

 

 

 

 

 

 

 

 

ee ousvwxo ouooamu uuulsosaus aeo~s>oo cocoosuuosuh mmmm
swoon macaw usouomeo: cu . 1;:
me ousvwxo ouoosao ecu-acouus aco~s>oo Uwauulouogeosuuomm mmnm
unsusaso
oususo>cu
uesaossdsz sauna chem vouaohaeoo an
we ousvwxo onoosuc asalaosuus u=o~s>oo concussed-Sawmogo to: museum
ousuuo>cu :::s
oesaousasz sedan ouch vauaouusoo on
~v oaavflxo oeooaao ass-nosaas aeo~s>oo Omens-ouozmosuoomm can voxonm
sedan ouch vaunouusoo 0»
cc ouuaououvazov onoosdo ass-nosuus acmas>oo cwuuoloflu:0uom pom voxoam
unsuso>cu
onsuohsusl madam when vouuouuaoo c»
an ousvwxo snousac ass-Joanne aco~s>oo oflsuQIUhQAI< pom vozosm
newsflas nsosom 0»
mm onsvwxo mucosao acolgusuus acoas>oo awuuolouomld mom moment
sow noses» as
onscaxo ouoosuo «nuisances aco~s>oo com voxosm
salsams mason oew>on d
an vasomxosom new: voxcwalnuouo cauaoIGLOQI< ovosuoouu
ussvwxouvm
\ousvwxo sewnassx
onsvfixouom Amos sea»: 0» ovohuooHo
mm \onsvwxo unoosuo acclaosuus aco~s>oo oflsuoleuoml< cahwcolwm
Wwdwguuhom gfimm venue: nowunNMEAQ-IH v0.30! newuoouo: ms scuosmm

.uomswso confluwaomla mo seamlsxm

a u4m<8

CHAPTER 3

THE GLUCOSE OXIDASE SBSR

As was reported previously (Chapter R) there is a minimal amount
of dispersion associated with an SBSR made of nonporous glass beads;
this allows longer residence times which. are desirable in enzyme
applications. Glucose oxidase was chosen as a model enzyme since it is
inexpensive, highly active, and easily obtained. The development of the
procedure used to i-obilize this enzyme on nonporous glass heads is
described. The dimensions of the SBSR assedaled with these beads are
given in Figure 2. In order to use an immobilized enzyme single bead
string reactor effectively one must first know its characteristics. In
some cases it is desirable to know the amount of enzyme actually
immobilized. A method fer such a determination is presented. Since it
is desirable to work with a reactor that retains its activity for a long
period of time, the useful lifetime of the immobilized glucose oxidase
SBSR is discussed. One problem with many enzyme preparations is that
they sometimes contain small amounts of impurities which can be
immobilized along with the enzyme. These impurities sometimes affect
the apparent activity of the enzyme. The presence of catalase in
glucose oxidase preparations can be problematic in this respect.
Results of the investigation of catalase activity in the reactor are

presented. The specificity of the reactor is also examined.

14

15

 

9000000900
/ \

0.86 mm i.d. 0.6 mm diam.
Teflon Tubing Enzyme Coated

Glass Bead

FIGURE 2 Single bead string reactor.

16
mobilisation Procedure for Glucose mjdase

Covalent attachment of enzymes to an inert matrix provides a
sturdy environment and yet allows the enzyme to be readily accessible
for reaction in continous flow reactors. Glass provides an ideal inert
support because of its low reactivity toward most compounds and
mechanical stability. Nonporous glass beads provide added surface area
without increased dispersion. For these reasons, covalent bonding of
glucose oxidase to nonporous glass beads was the method of choice for
the reactor to be incorporated into a flow injection system. Glucose
oxidase is ca-only attached to glass by glutaraldehyde linkage. The
specific conditions for this type of i-obilization as enumerated by
Thompson (43) were used initially. There are four steps in the
procedure: cleaning, whisker formation, silanization, and enzyme
attachment. The aim of this work was to minimize the amount of time
required to perform the i-obilization and to establish a procedure that
gave reproducible results.

The conditions for a given step were varied while those for the
other three steps were held constant. In each case all four steps were
carried out on a 0.5 g sample of nonporous glass beads. After the
i-obilization of the glucose oxidase the beads were put into Teflon
tubing (10 cm in length) and incorporated into the FIA system described
in Chapter 4. A calibration curve was obtained with glucose standards
in the 0 to 800 ppm range. The slope of the calibration curve was taken

as a measure of the activity of the enzyme on the beads.

l 7
Cleaning Procedure

In order to i-obilize an enzyme efficiently it is necessary‘to
eliminate any contaminants from the surface of the glass before the
procedure is begun. There are a variety of ways by which glass surfaces
can be cleaned. The most co-on chemical methods are by chromic acid,
hydrochloric or nitric acids, or alcoholic potassium hydroxide. Chromic
acid cleaning is a very harsh treatment that ensures the oxidation of
any carbonaceous material. Other acids are used to leach any heavy
metals from the surface of the glass. As shown in Table 2 the initial
procedure specified treatment of the glass beads with both chromic and
hydrochloric acids. It is known that in order for an acid treatment to
be effective it must be carried out at elevated temperatures (48).

The use of warm concentrated acids can be hazardous and is
therefore not a highly desirable method of treatment. Also, when the
beads were cleaned with chromic acid they darkened during the whisker
formation process. These two factors suggested that an alternative
cleaning process would be advantageous. Since the glass beads are not
extremely dirty it is not mandatory that chromic acid be used. As an
alternative cleaning solution, alcoholic XOR was chosen. The beads
obtained by cleaning with only the alcoholic XOR did not darken and were
found to have a level of activity similar to those treated with chromic
and hydrochloric acids. This would suggest that the beads do not have
an excess of heavy metals at the surface; these would require leaching

by RC1.

$.92

Cleaning

Rinse

Dry
Whisker
Formation

(Dry

Heat

Silanize

Decent

Heat
Glutaral-
dehyde

Enzyme

18

 

TABLE 2 Initial immobilization procedure.
Regent Time Temperature ('0)
Chromic Acid 15 min. 23
6 M 801 10 min. "
Distilled 820 "
Acetone "

"2 n
53 (w/v) 1 hr. 23
Nflsf RF/MeOR
Ne "
3 hr. 450
2% (v/v) 30 min. 23
3-aminopropyl-
triethoxysilane/
acetone
24 hr. 90
13 solution in 30 min. 23
pH 8.0 buffer
5 mg/ml glucose 24 hr. 4

oxidase in
pH 6.85 buffer

l 9
Ihisker Pornt ion

The best conditions for whisker formation were found by Onuska
(24) and were not altered. This step involves soaking the beads in
methanol saturated with a-onium hydrogen difluoride for an hour,
decanting the solution and drying the beads, followed by beating them in
a sealed glass tube at 450°C for three hours. The effectiveness of the
treatment with a-onium hydrogen difluoride was investigated. Three
experiments were conducted. The first involved treating the beads in
the specified manner. In the second the heating of the beads in a
closed tube was eliminated. The third experiment involved no treatment
with a-onium hydrogen difluoride.

It was expected that the beads from the first experiment would
have the highest activity. Since the second experiment was carried out
at room temperature it was expected at most that only a limited amount
of etching would take place; this was expected to result in only a small
increase in surface area and thus a small increase in enzyme activity
over that obtained from the beads in the third experiment. The activity
of the beads was found to vary as expected. The beads from the second
experiment exhibited only 588 of the activity of those from the first;
the beads from the third experiment had an activity that was only 49% of
that associated with those from the first. This indicates that indeed
the surface area is greatly increased by this treatment. It should be
noted, however, that the enhancement is not due to the formation of real
whiskers. Mottola has shown that the surface looks more like a series

of "volcanic crater erosions" (45). The same author points out that

20

there is not a great advantage to be gained by roughening the surface of

the tube and i-obilizing the enzyme on its surface as well.

Silanization

The silanization of the beads proved to be the most problematic
part of the immobilization procedure. It was found that heating the
beads with the 3—aminopropyltriethoxysilane in acetone caused the
coverage to be quite non-uniform. It was suspected that the rapid
evaporation of the solution was the reason for this. Therefore, it was
thought that an aqueous solution might give better results. The
silanization procedure specified by Mosbach (8) was tried. The
resulting beads had low activity.

A "trial and error" approach was used to arrive at a silanization
procedure which gave reproducible results. The concentration of the
solution, the temperature, and reaction time were varied. It was found
that a 1" (v/v) solution gave sufficient coverage without leaving an
excessive residue on the bottom of the beaker. By allowing the solution
to evaporate from the heads at room temperature in the fume hood more
uniform silane coverage was obtained. The temperature and reaction time
were decreased in order to prevent darkening of the heads (a sign of
silane decomposition). The time required for this part of the procedure
was significantly reduced from 24 hours to 15 hours.

Since the solution used to silanize the beads was made with
acetone it was assumed that washing the beads with acetone after the
treatment would be sufficient. However, the beads stuck to the bottom

of the beaker. One solution to this problem would be to use Teflon

21

beakers instead of glass. However, the more economical solution proved
to be the dropwise addition of distilled water (DH) to the acetone until
the beads were loosened. If too much water is added the beads float.
Thus, it can be concluded that this part of the procedure seals to be

much more of an "art" than a ”science".
Glutaraldehyde Attachent

The glutaraldehyde was attached to the silane group as specified
in Table 2. A 1% (v/v) solution of glutaraldehyde in pH 8.0 phosphate
buffer was allowed to react with the silanized beads. This step was not
modified. However, it was observed that the beads become salmon colored
when this part of the procedure is successful. It was noted that the
time required for the color change to take place was in the range of

0.5-2 hours.
Enzyme Attachment and SBSR Construction

The covalent binding of the enzyme was carried out under the same
conditions as in the initial procedure (Table 2). A 5 mg nul‘1 solution
of glucose oxidase (E.C.l.l.3.4, Sigma Type II from Aspergillus niger,
specific activity approximately 17,800 U g'1 at 35°C) in phosphate
buffer was allowed to react with the glutaraldehyde treated beads for
24 hr. It is desirable to immobilize the enzyme in its most active
state. The use of a 0.05 M phosphate buffer, pH 6.85, facilitates this.
After this step was complete, a single head string reactor was made by
placing a suitable portion of the beads in a length of Teflon tubing.

This task was most easily accomplished by crimping one end of the tubing

22

and inserting it into the end of a syringe. The tube was then filled by
applying auction. The beads remaining were transferred to a glass vial.
In any case the beads were stored in 0.05 M phosphate buffer at 4°C.

The final inobilization procedure is sum-arized in Figure 3.
A1 temate Method of [mobilization

While periodically retesting the SBSRs, a rapid decrease in
activity was observed during the first few days of use. This was
attributed to enzyme being adsorbed on the surface. One remedy for the
situation is to flush the SBSR with copious amounts of buffer prior to
use. An alternative is to prevent adsorption from occurring in the
first place. Therefore, the possibility of immobilizing glucose oxidase
by continuously circulating the enzyme solution was investigated. The
immobilization procedure was carried out in the usual manner up to the
point of the glutaraldehyde attachment. The beads were then put into a
suitable length of Teflon tubing which was connected to a pump tube. A
small two-channel peristaltic pump (Ismatec) circulated the enzyme
solution at the rate of 0.8 ml min'l. The resulting beads exhibited
very low activity compared to that of beads obtained by the usual
method. The trade—off of providing the beads with a continuous stream
of "fresh” solution versus the time needed for attachment at the surface
proved to be in favor of the latter. Perhaps a very low flow rate
(e.g., 0.05 ml min”) would provide a suitable environment for the

immobilization .

.835& 330333 use: a an

3:3 22.9%.. s. no.0 s 822. 8.. 5... 8.8: 3 832.8

.
o

 

 

 

.
o
. .
Esau- z.u2~:u.:u.zn.«:ozmo A 25.26. z~x+ozun.~:o.:o.zn.«:u: ...mo
. 9v 9 23 .
I m. ...
.56522. 3:35.. .m
3 , 8:: 8:3 suing... 28.0 3 332.2
2
. .
... .5 ...
ozomuzuezuémazuzmé A - 9 an ”15. + ~zz»..zu:m.-o
JV 0. 010 mo
8:63 02522235 .N
m 3.... of 28:3 3 2.32.2
.
«:zn.~zo..m-o :o
o l ”15.50%” u a +
«zznmxuzw o 92. 9 an} . :8 :z :o
m o.- 0 and a .
cozoucgm _

24
Determination of Amount I-obilized

The are two viewpoints concerning the amount of enzyme
i-obilized. The first is the literal interpretation-what quantity of
enzyme is on each bead. The second is more pragmatic in nature-what
equivalent amount of enzyme must be used in solution in order to do the
same analysis. There are a under of ways to arrive at an answer for
the first case. One might choose to do a nitrogen analysis. This would
not guarantee an exact result unless the aminosilane content were
previously known (or determined). An alternative would be quantitation
of the protein present. However, the small amounts of other enzymes
present as impurities would result in a positive error. In either case,
the enzyme is destroyed in the process. Professor Mottola’s group at
Oklahoma State University has been working on a nondestructive solution
to this problem, but has only published. a method for the determination
of aldehyde groups on glass surfaces (49). It therefore seemed
appropriate to pursue another method of enzyme determination.

The method chosen was that of an assay by difference. A glucose
oxidase solution was prepared, half of which was used for the covalent
binding step in the immobilization procedure and the other half as a
control. After the enzyme was attached to the heads the reacted GO
solution was poured into a volumetric flask. The washings were cowined
with this and the mixture was diluted to the mark with distilled water.
The final solution was used as the carrier solution in the FIA apparatus
with a plain SBSR in place of a GO SBSR. Dilutions of the control
solution were used in a similar manner to construct a calibration curve.

For each GO solution a 400 ppm glucose standard was injected and the

25

absorbance measured. The absorbence of the reacted solution was located
on the calibration curve. After correcting for the dilution factors,
the amount of glucose oxidase i-obilized per gram of beads was
determined to be 7.3 mg (approximately 1.4 ug -'2).

From a more practical standpoint (the second viewpoint mentioned),
the activity of the G0 on a 10-cm SBSR was found to be equivalent to a
solution containing 0.5 mg of enzyme per m1. Cofiining the results from
the two determinations, it would take less than two hours under
continous operation (a flow rate of 0.5 ml min”) to use the entire
amount of glucose oxidase needed for the i-obilization. These results
were based on the initial activity of the GO SBSR. In order to make a
more realistic comparison one must take into account the drop in
activity that occurs.

A typical scenario of the loss in activity is shown in Figure 4.
It can be seen that after about 150 days the activity of the stored
beads would match that of the beads which had been used for periodic
testing. The activity in this particular SBSR (one of the first
assdbled for this research) dropped to approximately 568 of the
original. The G0 SBSR used in the determination of the amount of enzyme
i-obilized was found to retain 66X of its original activity after one
year (the i-obilization method was improved). Thus, it can be
concluded that the glucose oxidase i-obilized by the modified method
presented in Figure 3 has a very long useful lifetime. This evidence

substantiates the cost advantage of enzyme imobilization.

cmammou CURVE SLOPE mo“)

1.“

 

26

 

 

 

FIGURE 4

UUUrrritf'UllI'IfirIrrIIT'UUFV'

so 100 150
TEST DAY

Loss of glucose oxidase activity for
routinely used beads (C) and stored
beads (A).

2 7
Ca talase Acti vi ty

In all ihobilizations there is a risk of i-obilizing impurities
which will interfere with the reaction of interest. Catalase is a major
probl- in the case of glucose oxidase since it destroys the peroxide
formed. In order to get some idea of the extent of the catalase
activity in the GO SBSR, an experiment was carried out in which three
lO-cm reactors were tested individually and then connected together to
make reactors 20 and 30-cm long. The expected absorbance was obtained
by adding the appropriate absorbances for i the two- or three-unit
reactors used. As can be seen in Figure 5 there is a significant
difference between the expected values and those actually obtained. In
an effort to eliminate some of this effect the color-forming reagent
stream was placed before the glucose oxidase SBSR. A wider range of
linearity was found, but there was still a significant deviation from

the expected results.
Specifici ty of I-obi 1 i zed G] ucose Oxidase

Since the ultimate intended use for the GO SBSR is in the analysis
of real samples it was necessary to ensure that the reactor is specific
for glucose compared to other sugars. Nine nutritionally—significant
sugars were tested by preparing 3.3 #4 standards of each and using them
in the place of B-D-glucose. The responses were then taken as a
percentage of that obtained for fl-D-glucose. The results are compared
in Table 3 to those found in the literature (50). The experimental

n n

values designated by a - indicate that the value obtained could not be

differentiated from the baseline. All values except that obtained for

ABSORBANCE

S

o ‘
g g

E

llllllllLlllllllllllllljllllljllllllllllllllllll

 

28

 

om UIVTr'I'UrriIUIIIUIIrTUUIUUU'I’T‘W‘IUrI—Ul

FIGURE 5

200 400 600

[GLUCOSE] (pg/ml)

Actual (- -) and expected (-—) effect of
catalase in a 20-cm SBSR (O) and a 30-cm
SBSR (O).

29

TABLE 3 Glucose oxidase activity toward other sugars.

Response Ratio

 

§uga£ §§§§ Literature
p—D-glucose l . 00 l . 00
(“D-glucose 0.3% 0.0064
galactose - 0.0014
fructose - na
mannose 0.011 0.0098
2-deoxyglucose 0.18 0.25
maltose 0.015 0.0019
xylose 0.0062 0.0098
glucosamine - (0.0005
sucrose - na

"-" denotes no distinguishable response
"as" means not available

30

a-D-glucose are as good as or better than those indicated in the
literature. The unusually high value for c—D—glucose was a result of
phosphate catalysis of the mutarotation of this species. A. more
detailed discussion of this is undertaken in Chapter 5.

In am, an inobilization procedure for glucose oxidase has
been improved and characteristics of single bead string reactors made
from nonporous glass beads containing the enzyme have been established.
In the next chapter some practical considerations concerning the use of

this reactor are discussed.

CHAPTER 4

OPTIMIZATION OF THE FIA SYSTm

The glucose oxidase/Trinder reaction (51) is one of the most
cmonly used methods for the colorimetric determination of glucose.
The reaction scheme originally proposed involves the reaction of glucose
and molecular oxygen in the presence of glucose oxidase (GO) to produce
hydrogen peroxide. The H202 then reacts with 4-aminoantipyrine (AAP)
(also known as 4-aminophenazone) and phenol in the presence of
peroxidase (PO) to produce a colored dye with an absorbance maximum at
505 nm. Trinder later found that when the phenol was replaced by
3,5-dichloro-Z-hydroxyphenyl sulfonic acid (DCPS) the reaction was four
times more sensitive and the absorbance maximum shifted to 510 nm (52).
In this work the latter Trinder reaction was coupled with the GO
reaction as shown in Figure 6.

Other workers have immobilized glucose oxidase and used it in flow
injection analysis systems in conjunction with the Trinder reaction (45,
53, 54). However, this work is the first to be published (55) regarding
the optimization of such systems.

In this chapter the computer controlled FIA system incorporating
the GO SBSR with the Trinder reaction is described. The salient
features of the instrumentation are enumerated. Three different

approaches toward improving system performance are discussed.

31

32

603000.. uevsuuh\ees3xo 0.836

U!

n
I
n: z u: a J
n
10.2,
come: o~zm+©gzu0 +||oa z o
o o 6

9532 $683.05

N0.4.: + o

n

101

IIO

I
O

xouzu

O:

I

 

m 953..
n34 mauo
u ZN:
Inom
+ I9©6 + «ofm
@ .
wmouaoéum
:o x
302.8 3820 NO + r“ I rm H...
0

ref...

33

Instrumentation

A diagram of the apparatus is shown in Figure 7. The temperature
of the solutions and the GO SBSR was controlled by a thermostatted
circulating water bath. The temperature of the reagents was controlled
by placing each solution in a jacketted beaker or flask. A jacket was
made for the GO SBSR. This consisted of a piece of glass tubing (that
fit snugly around the lO-cm length of Teflon tubing) fused into the open
ends of a small condenser.

The sample was injected via a pneumatically actuated 6-port sample
injection valve with a 30-u1 sample loop (Rheodyne) into the carrier
stream which was pumped by a lZ—channel peristaltic pump (Ismatec).
Flow rated pump tubing (Technicon) was used for delivery of the reagents
to the manifold. The sample first passed through the GO SBSR. It was
then joined at a tee by a solution containing PO, AAP, and DCPS. The
sample/reagent mixture then passed through an SBSR made with unmodified
glass beads (also referred to as a "plain" SBSR) in order to promote
mixing without added dispersion. After reaction the sample/reagent
mixture entered a miniaturized flow-through filter colorimeter which was
designed by Patton et a1 (56) and assembled in our laboratory. Data
acquisition, sample injection, and pump speed were controlled by a

microcomputer as discussed below.

Choice of Fixed Parueters

Initially there were a number of parameters that, for convenience,
were fixed. These were: SBSR lengths, sample size, flow ratio of the

carrier (buffer) and reagent streams. As mentioned previously

34

eczema-heal. 0803' no.“ use». 4: b 953.:

 

 

“Sign";

ego:-

 

 

 

_H ..eaoeaeml_ lull-.1, “I IIIII u
. V A A Jaeemoemu

:3... .....-..-..H...-.H-.i

 

.
u .

 

 

 

35

(Chapter 3) the effect of the catalase, as the GO SBSR is made longer,
invokes the ”law of diminishing returns”. Therefore, a lO-cm length was
chosen since its activity gave acceptable absorbance values. This also
minimized the back pressure which results when a packed reactor is used
in a flowing stream. A 40-cm SBSR consisting of unmodified glass beads
was used as the reaction manifold for the Trinder part of the reaction.
This length was necessary to allow sufficient time for reaction between
reagent addition and detection.

The placement of the PO reagent solution was also of interest.
Theoretically, it would be better to add this solution prior to the
glucose reaction so that any effect of catalase would be minimized. The
hydrogen peroxide would be i-ediately consumed by the color-forming
reaction and have less chance of coming into contact with catalase as
the reacting plug travels through the manifold. A comparison of the
calibration curves obtained with the reagent addition before and after
the GO SBSR revealed that there was no significant increase in slope or
linearity for a lO-cm length. The 20 and 30-cm lengths did exhibit a
significant increase in linearity. Thus the choice of a lO-cm length
was confirmed as the best one. The reagent solution could be added
after the glucose reaction without loss in sensitivity and without
risking contamination of the immobilized glucose oxidase.

A sample volume of 30 ul was chosen to match the internal volume
of the glucose oxidase single head string reactor. In order to
investigate the relationship of absorbance to sample size a 6-port
sample injection valve was connected in such a way that the sample size

was regulated by the amount of time the valve spent in the "inject"

36

position. The injection time multiplied by the flow rate of the sample
stream was used to determine the sample volume. The sample injection
time was varied from 1 s to 25 s in l 3 increments with a flow rate of
0.44 ml min‘l. The peak heights obtained are plotted in Figure 8 versus
sample volume. It is interesting to note that the relationship between
absorbance and peak height is linear from 7 pl to 81 ul. The greatest
sensitivity is attainable in this region and, thus, makes the 30-311
sample volume desirable. Beyond that region the response increases by
successively smaller amounts and finally approaches the maximum amount
of glucose that can be converted--approximately 74 ug (185 pl of a
400 ppm standard). In previous experiments related to this project
(43), double peaks were obtained with volumes greater than 65 ul for a
20-cm GO SBSR through which the reagent stream was passed concurrently
with the sample. This resulted in an optimum sample volume of 45 pl.
No incidence of double peaks was observed for this system which
indicates that complete mixing occurs in the manifold.

The ratio of the flow rate of the carrier stream to that of the
reagent stream was adjusted by appropriate choices of pump tubing. A
range of ratios from 4:1 to 20:1 was investigated. The overall flow
rate was kept at 0.5 ml min‘l. The resulting peaks are shown in
Figure 9. There was no sampling time advantage to be gained since the
baseline widths of all peaks are similar. The peaks obtained with the
two highest flow ratios showed some distortion. Therefore, a 10:1 ratio
was selected. This also helped minimize both the consumption of PO and

dilution of the sample.

PEAK HEIGHT (ABSORBANCE)

0.400

0.2”

IIILIJJLJIIlllllllelllllllljl

37

 

 

r T I l I r I I ' r U r r l t r r r j
0 50 100 150 200
SAMPLE SIZE (pl)
FIGURE 8 Bffect of varied glucose sample size.

ABSORBANCE

0300

0300

0200

OJOO

38

FLOW RATIO (BUFFER:REAGENT)

 

 

 

 

 

 

 

 

 

 

 

 

3 5 7 9 11 13 15 17 19 21 23
r I I U fir ' l V r t l I l V r r i r I' I

4
4 n
, fl
1
1
d
J

rrvwfirrrvwwqrrrrw1arrrv1

0 100 200

TIME (sec)
FIGURE 9 Effect of flow ratio (Buffer : Carrier) on

FIA profile.

39

TQerature Calibration

Since it was necessary to control. the temperature of the FIA
manifold, a method had to be devised to monitor the temperature. The
temperature of the solution flowing through the GO SBSR was not the same
as that of the water bath. A high precision 30-k0. thermistor was placed
in a thin piece of protective Teflon tubing which was subsequently
inserted into a glass tee (see Figure 10). The junction was sealed with
Teflon tape to minimize the dead volume and prevent leakage. This
temperature probe was calibrated by immersing it in a water bath at the
same level as a thermometer. The temperature of the bath was increased
in 2" or 3° increments and the resistance recorded after a stable value
was reached. The calibration curve is shown in Figure 11. The
relationship of resistance to temperature was fitted for 10° segments to
the Steinhart-Hart equation ('1"1 = A + 3*(1n R) + C*(ln R)3, taken from
Reference 57). The resulting coefficients are presented in Table 4.
The probe was then connected to the terminal end of the GO SBSR. Once
again the temperature of the bath was increased in small increments and
the resistance measured. After converting the resistance to
temperature, the following linear relationship between the temperature
at the end of the reactor and that of the water bath was found:
T = 0.505 It Ta + 11.1, where T is the temperature at the GO SBSR and Ta

is that of the water bath.
Microcmputer Interface to FIA Systa

Original 1y an in-house designed (58) Intel 8088-based

microcomputer (also known as the ”Bruce Bus") controlled the pump speed,

40

£35885 978 335380 83?... 2 952..

ll

\1 an..." @\ =00 603*
35200 2 +1 x f

 

 

‘

 

 

 

 

 

 

 

 

5562.22
.265

 

 

 

 

A

RESISTANCE (RD)

41

8

LLllllLlLllllJlllllllelllLll]

20

10

 

 

O

frrrtTIUIfI'rTTrrTITIUIITrUTlrUTrI

0 ' 10 20 30 40 50 00 70
TEMPERATURE (°c)

FIGURE 11 Thermistor response curve.

42

TABLE 4 Coefficients of the Steinhart-Hart equation for different
flow rates.

Equation: T*1 = A + B (In R) + C (In R)3

T = temperature

Temperature Range (°C)
21-30
31-40
41-50

51-60

0.133
0.0535
0.0358

0.0100

R = resistance

B
-0.0595
-0.0226
-0.0145

0.00112

C
0.00279
0.00159’
0.00137
0.000527

43

injection, and data acquisition. The software was written in FORTH
(Forth, Inc., Hermosa Beach, CA). Most of the "words" used in this work
were strings of "core words” which had been written or modified by Paul
Kraus. The final word, "PEAKS", combined injection timing, data
acquisition of 100 points over a specified period of time, automatic
storage of the data on a floppy, and repetition of the process for a
specified number of times. More extensive information on the
configuration of the microcomputer can be found in Reference 59.

In the latter phase of the project computer control was
relinquished to an IBM PC compatible microcomputer (Bentley-model T).
The machine was also Intel 8088—based. It was equipped with two floppy
drives, 512 Kb of memory, a color graphics board (IBM), and an RTI-815
interface board (Analog Devices). All routines were written with
QuickBASIC (Microsoft). The timing routines were written in assembly
and made use of the AM9513A counter/timer that was part of the interface
board. Control of the pneumatic actuator on the injection valve was
accomplished by connection to a digital output port on the interface
board. The data acquisition routines utilized direct memory access
(DMA). Plotting routines were developed for three different graphics
modes-~EGA, CGA, and HERCULES (the HERCULES drivers were provided by
Mark Victor, a graduate assistant in the Chemistry department). The
initial DMA and plotting routines written by Pete Wentzell, a graduate
assistant in the Chemistry department, were modified for this work.

The main program "FIADATA" is a menu-driven program. A number of
features were built into the program to make it flexible enough that

almost any flow injection analysis experiment could be carried out on

44

the equipment presently available in the laboratory. One important
aspect in this regard is that either the sample output (1 10 V) or the
”log-ratio” output (i l V) from the calorimeter can be monitored. The
gain selection option on the interface board is controlled by the
software. The general characteristics of the program are: user-
selected experimental parameters, a flow reversals experiment option,
data acquisition from one or two channels at a rate selected by the
user, immediate data storage on a floppy, screen plots of the data, and
printouts of the file containing the data. This program was modified to
allow variation of the injection time; "FIAINJ" was the result.

The need also arose for the capability to plot (on the screen) or
print data from files previously stored on a floppy. ”FIAPLOT"
incorporates a routine that will read the data regardless of its format.
The only stipulation is that the comments be prefaced with ”;" and
confined to the beginning lines of the file. By calculating the number
of data points and the number of lines of data, the number of columns of
data is empirically determined. The user selects which columns of data
are to be used for the plot-one x and one or two y values. There is
also an option in the program that allows the user to print the data
file. Eventually, the three programs may be merged into one large
generic program that will do everything but turn the computer on for
you.

An interface box was constructed in order to facilitate easy
connection to the interface board. The mounts for 50-pin and 34-pin
connectors were soldered to the back side of a solderless breadboard.

The breadboard was then mounted into an aluminum box along with 6 ENG

45

and 6 banana plug connectors. The appropriate size ribbon cables were
used to attach the box to the interface board. This arrangement
provides a versatile way of linking the computer to the analytical

system.

Data Acquisition

In either case each experiment involved injection followed by
monitoring the output voltage of the detector. Each data point shown in
the figures represents the average of the values obtained from 3-5
sequential injections. The baseline voltage was maintained at 5.0 V.
The detector voltage was converted to a digital signal by the
microcomputer’s lZ-bit analog to digital converter (ADC). For the in-
house built microcomputer the ADC counts were stored in a file on a
floppy disk. The files were later transferred to a DEC LSI 11/23
minicomputer, and the ADC values converted to absorbance units after
first calculating transmittance values. The software for the PC
compatible computer was written such that the absorbance was calculated
immediately following the data acquisition period and stored on the
floppy. In both cases the resultant data file contained time,
absorbance pairs. Further data manipulation was usually carried out on

the LSI ll/23.

Improvement of Sys ten Performance

The complete FIA system, when first assembled, was functional but
it was felt that better sensitivity and sample throughput could be

achieved. As mentioned earlier three approaches were used. The first

46

two approaches were focused mainly toward improving the sensitivity. In
essence the goal was to achieve the largest peak absorbance possible. A
third approach incorporated the achievement of higher sample throughput
as well. In the first and third cases an optimum set of operating
conditions was sought for six experimental variables; these were the
flow rate, temperature, pH, and the concentrations of PO, AAP,and DCPS.

All solutions were prepared using distilled water (DH) and
filtered as necessary. A peroxidase solution was prepared by dissolving
20.0 mg of Horseradish peroxidase (E.C.l.1l.1.7, Sigma Type II, specific
activity approximately 200 U mg'1 at 20°C) in 2.000 ml DH i-ediately
before use to avoid decomposition. The reagent solution consisted of
the desired volumes of PO solution, 10 i4 4-aminoantipyrine (Sigma), and
10 IN 3,5-dichloro-2-hydroxyphenyl sulfonic acid (Sigma) mixed together
and then diluted with buffer to 10.00 ml in a volmetric flask. The
reagent solution was prepared i-ediately before use.

Glucose (Sigma) standards were made by diluting with buffer the
appropriate quantities of a 2 g (c—D—glucose) 1‘1 stock solution that
contained 0.5 g 1‘1 benzoic acid as a preservative (the stock solution
was stored at 4°C). This stock solution was prepared at least 24 hours
in advance so that mutarotation equilibrium was reached. Unless
otherwise noted, a 400 ppm (2.2 #4) standard was used. The
concentrations of PO, AAP, and DCPS were 0.8 mg ml”, 1 i4, and 1 14,
respectively, unless otherwise stated.

Phosphate buffer (0.05 M), pH 6.85, was used in the univariate and
delay flow experiments. Universal buffer (60) was used in the simplex

experiments. Alternative buffers tested in the univariate experiments

47

were Tris and Borax. The recipes for all but the Universal buffer can
be found in the Ch-ical Rubber Co. Handbook. All experiments were
carried out at room temperature (22°C) except, of course, those in which

the temperature was varied.
Univariate Experiments

The most canon method of optimizing analytical systems is the
univariate approach. In this, one parameter is varied while all others
are held constant. The six variables mentioned previously were studied
by this method.

The flow rate was the first parameter investigated. As can be
seen in Figure 12 there is a marked decrease as the flow rate is
increased. The flow rate (FR) is related to the pump speed (PS) by the
following equation: FR = 0.0141*PS + 0.00167. A rate of 0.5 ml min"1
was chosen for subsequent experiments. This gave an absorbance that was
large enough to give very reproducible results yet small enough that
values obtained with more concentrated samples would not be beyond the
range of detection. This also allowed an acceptable sampling rate of one
per minute.

The effect of the concentrations of the two color-forming reagents
was explored. In the initial study the DCPS concentration was fixed at
1 i4 while the concentration of the AAP was varied from 0.1 IN to 2.0
id. The AAP/DCPS ratio that gave the best results was 1:2. This was
found to be the case for pH values between 6.0 and 8.6 as evidenced in
Figure 13. From the reaction stoichiametry one would assume that the

ratio ought to be 1:1. This caused suspicion as to whether or not the

ABSORBANCE

0.200

0.150

00‘”

48

 

 

0.0 0.5 1.0 1.5
FLOW RATE (ml/min)

FIGURE 12 Effect of varied flow rate.

ABSORBANCE

0.500

0.400

0.200

0.100

49

 

 

 

I .pH 7.99
‘ .pH 0.74
< .pH 5.00
I .pH 3.55
" A A 3 ¢ ¢ ; pH 9.03
I I I I I I I T r l l I l I l I I l I r I l Ifi—l
0.00 0 50 1.00 1.50 2.00 2.50
AAP/DCPS RATIO
FIGURE 13 Effect of AAP/DCPS ratio.

50

dye was really formed by the binding of one AAP molecule to one DCPS
molecule.

The possibility was entertained that the initial compound formed
might be 1:1 with later addition of another DCPS. Inspection of the
proposed dye compomd structure would suggest that this is not highly
likely since the addition of another DCPS would be sterically hindered.
To rule out this possibility the diode array detection ysstem developed
by Mark Victor (61) was connected in place of the calorimeter. Spectra
were acquired every second as the reacted sample plug passed through the
flow cell. It was assumed that a change in the structure of the complex
(formation of more double bonds) would cause a shift in the absorbance
maximum. The ends of the plug, where the greatest amount of reaction
occurs, seemed the most likely place for this effect to present itself.
However, as shown in Figure 14, there was no detectable shift in the
absorbance maximum over the entire plug.

Next, a continuous variations experiment seemed in order. The
coinined concentration of the two reagents was kept at 2.0 IN. Ratios
of 2:1 through 1:6 (AAP:DCPS) were investigated. The highest absorbance
was obtained for the 1:4 ratio. This ratio is very different from the
one obtained earlier, but the AAP concentration is similar (i.e., 0.4 ‘4
compared to 0.5 dd). Since the aminoantipyrine is considered the two-
electran donor for the peroxidase (62), its concentration should be the
critical factor. This is corroborated in these studies. The results
also imply that for a reagent solution 0.4 lid in AAP the DCPS

concentration must be at least 1.0 id. From these experiments it is

51

.8388 3.3: I: ..8
.12. 5.. I: so 88....- 5.8383: 8...: 3 use:

      

   

OOIIO

 

E l 2
a 1W”- n”, s . —~ g R
an

       

<54 3
‘9' “fix. \. ..O

    
    

     

7- . \. s,
a , - ,..“m...»... A. - a ....
- “mmHMMNN’Illll/ . 1. v.4“ N\\J \ww
- H ONOI/llll’ . QDWL..MW.O I. ,N~NNQ.

 

.'

 

‘\
\‘h‘ “““W‘\-
t \“\\\.H“‘~‘ 1- o
0

-1 .

 

- I l 1 1 7 7
0 5 0

0.0110

“0 (3’9 “‘0 .VO \\'0 ”'0 W0

esuaqsosqw

52

also apparent that there would be no great advantage in increasing the
total concentration of the reagents.

The peroxidase concentration was varied from 0.05 mg ml"1 to
1.0 mg ml”. An 800 ppm standard was used so that the results would
reflect the maximum amount of PO necessary for the range of standards
normally used for calibration. A plateau region emerged as the
concentration approached 0.8 mg ml"1 (see Figure 15). At this point a
small variation in the PO concentration would not drastically affect the
absorbance obtained. Since such small amounts of peroxidase are used on
a daily basis it is desirable to operate in that region.

The investigation of the effect of pH was twofold. First, the
type of buffer had to be determined. The four types of buffer listed
above were each made to be pH 8.00, since that is in the working range
of each. The experiment was conducted in the usual . fashion. The
phosphate and Universal buffers gave equal absorbazmm values while the
Tris and Borax buffers proved to be less than 1. Because the
Universal buffer has such a wide working range (pH 5.00-10.00) it was
used for the second part of the study. The pH was varied from 5.00 to
10.00 in 0.50 increments. The results shown in Figure 16 indicate a pH
optimum at 7.50. For the purpose of comparing this optimum to that
obtained with other buffers, another set of experiments was carried out
in which phosphate and Borax buffers were used to cover the range of
pH 6 to pH 10. The maximum response was obtained at pH 7 (see
Figure 17). It can be concluded from this study that the glucose
oxidase/Trinder reaction is quite sensitive not only to the pH of the

buffer but to the type of buffer used as well. The optimum values

ABSORBANCE

0.400

0.200

0.100

53

 

 

q
q

i

.1

q
-

I r j— r I I I— r I r f 1 I I I ' I I I I I I I 1

0.” 0.20 0.40 0.00 0.00 1 .00 1 .20

[P00] (mg/ml)

FIGURE 15 Effect of peroxidase concentration.

Absarbonce

0.150

54

 

 

q

«A

q

1

.1

d

d

.4

I I I I I r I T r l’ I r I I I r I I I I I I I I I I I I I l
4.50 0.50 0.50 10.50
pH
FIGURE 16 pH dependence of GO/Trinder reaction with

Universal buffer.

ABSORBANCE

0.300

55

 

 

-
.4 -’ \
I
I \
.4 t \
\
T b
q
1
q
IIrIlrIIIIIrIIIIrrI'rTII'IIII[IITI'IIIr1
4.00 0.00 5.00 1 0.00 12.00
pH

FIGURE 17 pH dependence of GOVTrinder reaction with
phosphate (- -) and borax (-) buffers.

56

obtained here are slightly higher than the pH optimum for glucose
oxidase in solution (63). This will be discussed below.

The final variable examined was the temperature of the GO SBSR.
It should be noted here that the temperature of the colorforming
reaction was varied over the same range as that used for the glucose
oxidase reaction and no significant improvement was attained. This
demonstrates that the reaction is fast comared to the time required for
the reacted sample to traverse the manifold. Since it is well known
that enzymes are prone to denaturation above 37°C the operational
temperature range was from room temperature (21°-22°C) to a few degrees
beyond this upper limit. The bath temperature was adjusted and the
components were allowed to come to thermal equilibrium before the sample
was injected. The response obtained increased with temperature as
illustrated by the plot in Figure 18. Between 31°C and 38°C a plateau
region exists, where the response is at its maximum value. Beyond 38°C
the response decreases rather rapidly because of enzyme denaturation.

A complete set of univariate experiments would have involved going
back to each variable systematically in order to determine a true set of
optimum conditions. If such an approach were used to evaluate the
optimum to within 20% in each of seven dimensions (an additional
variable is created when the pH of the carrier stream is separated from
that of the reagent stream), for this FIA system a_ total of 175
experiments would have been required. Since a great amount of time was
expended in this small set of experiments, an alternative to this method

was sought .

ABSORBANCE

57

 

 

002m -
1
' s
q
.. O
a. ‘ m -
q
-
q
as” I r I f I I
I I I r I I r I 1
20 30 40 50

TEMPERATURE (°C)

FIGURE 18 Tuperature response of a 10 — cm
GO SBSR.

58

Delay Flaw Experiments

This second approach focused solely on increasing the absorbance
obtained with a 400 ppm standard. Based on the results of the
univariate experiments a standard set of conditions was defined:
0.5 ml min”, 22°C, pH 6.85, 1 fl AAP, l m DCPS, and 0.8 mg ml"1 PO.
By integrating the FIA peak it was found that the total absorbance
obtained corresponded to conversion of 7.3% of the total amount of
glucose in the sample. It was therefore suspected that significant
increases could be attained by stopping the flow after the entire sample
plug occupied the GO SBSR. It was thought that a delay time would allow
as much reaction to occur as possible before the flow was resumed.

A preliminary study was used to determine the time at which the
pump was to be stopped. The sample was injected and the pump was
stopped at various times, while the delay time was kept constant at
10 s. Then the pump was restarted. The greatest response was obtained
when the pump was stopped after 10 s.

The stop time was then set at 10 s and the delay time varied.
Data acquisition times were adjusted to ensure complete recording of the
signal. Figure 19 shows that the maximum response can be obtained when
the flow is delayed for 10 s. The decrease in absorbance for longer
delay times is probably due to the destruction of hydrogen peroxide; as
the sample is in the reactor for a longer period of time there is a
greater probability that the peroxide will come into contact with
catalase and, thus, be reduced. This method only served to increase the

sensitivity of the system and actually decreased the possibility of

Emmi-.m-

ABSORBANCE

0.100

 

59

 

 

 

T U r f ' T T I T T r r I U

r
1 O 20 30
TIME (sec)

FIGURE 19 Results from delay flow experiment.

60

greater sample throughput. Therefore, an alternate method of

optimization was sought.
Simlex Optimization

The underlying assumption in a set of univariate experiments is
that there is no interaction between variables. If there is indeed some
interaction a false set of optimum conditions may be found.
Alternatively, a pattern search could have been used to determine the
response surface for the seven variables investigated here. Under these
conditions 78,125 experiments would have been required. A better
alternative than these methods has been shown to be a simplex
optimization (64), which allows simultaneous variation of all
parameters.

The simplex method is widely applied and accepted in analytical
chemistry (65). A simple two dimensional surface, as pictured in
Figure 20, can be used to illustrate the principles employed. The x and
y axes represent the two parameters to be varied and each concentric
circle represents combinations of those two which give the same
response. This surface can be thought of as a topographical map; as the
circles get smaller the response increases in magnitude. The simplex is
generated initially by choosing a set of experimental conditions which
are known to be suboptimal. After the response from that experiment is
obtained another set of conditions is specified. This process is
repeated until the geometric shape made of n+1 vertices is obtained,
where n is the number of parameters. In the two-parameter case, three

experiments are required and the simplex takes the shape of a triangle.

62

The response at each vertex is ranked based on its magnitude—the
largest is taken as best, the next largest as next best, etc. In the
triangular case, the point that gives the worst response (point 1 in the
figure) is reflected an equal distance through the line between the
other two points (points 2 & 3). A new set of conditions is specified,
the response obtained, and the responses are again ranked. In this
example, the response surface is shown so that the movement of the
simplex can be understood. Normally, the response surface is unknown
and becomes defined by the movement of the simplex. As the optimum set
of conditions is reached the simplex may begin to oscillate. Such
behavior can be thwarted by reflecting the next best point rather than
the worst.

The history of the simplex methodology (adapted from Reference 66)
is su-arized in Figure 21. Since the pioneering work of Morgan and
D-ing (67), many modifications have been suggested which attempt to
improve upon the speed and reliability of the method. These have been
reviewed elsewhere (68) . For this work, the method chosen was the
Composite Modified Simplex (69). This method has additional features
such as expansion when the optimum is far away and smaller movements
when the optimum is near. The composite modified simplex program
(OPTIMA—-written in BASICA) was provided by Betteridge and Wade (70).
Simplex optimization was previously attempted on the original Trinder
reaction in a static system (71), however, no optimum was found.

It has been shown that a judicious choice of a response function
can expedite the optimization process (72). Such a response function

allows a number of FIA parameters to be optimized simultaneously, and

63

 

 

 

 

T... ..I
mv. rem s maven-m m t m
(REFLECTION. next 10 m1 mun-omens.
amount n'um hem:
"“ names a. nun max «new
EXPANSION. mill-“Yuan. no to? "Rut!!!“ or
10 WORST names as m mu. mass
was mummy |
7m ERNST
m m common I
m IMHO.“ I
T... I
m an a! nouns
I mums m
mam; am;
“7‘ museum ‘——————
menus cum
1072 I—MELILAU
(Y'EAY‘N' G
museums»
magnum ol- moms:
cowsmeo me: mm
mm» snncm
an pun-no s ammo '
(LARGE mun. mm

 

1.7. KING, mm 6 WM ’0!“ 6 scam
(REINTRODUCE NEXT 70 (Ilium K01 M
m REJECTIONI

 

 

' 1
1070 ”3 mmmm a. neuron

MRWPED m.
mm m, rum Fm

 

 

 

 

   
  

1000 J RYAN, m & tom um on WIEL
(CONTROLLED WEIGHT!!! (ESYINATED l)
ammo macs: «muss»: Fm

(WW A)" (WONTED “HMO!
WEIGHTED cam Mason:

10.: has 3 commas ammo: a m
m a wow I» «emu: m
mx utmost

 

1”: wanna.

1.“ “In! t 70“!
(MAIN! m

FIGURE 21 History of simlex optimization.

 

I.

64

takes the form of a mathematical equation that codaines two or more
system parueters (peak height, cost, throughput, etc.) in an
appropriately weighted fashion.

The response function chosen was based on three considerations.
First, the activity of the enzymes involved changes over long periods of
time; second, the achievement of a sampling rate of at least 60 hr‘1
was desired; finally, the sampling rate factor should contribute no more

than 30% to the overall response. The function chosen was:

 

 

80 3 Ann»
Time Corrected Absorbance Ratio = x --
60 8 + tp Absss
where Asxp = Absorbance at peak maximum for a set of experimental

conditions, Abase = Absorbance at peak maximum for the baseline set of
conditions, and tp = time (sec) after injection at which the peak
maximum occured.

The initial simplex was obtained by entering the information
listed in Table 5 into the composite modified simplex program. The
simplex program was run on an IBM PC compatible microcomputer (PC
Designs) while data were acquired with the Bruce Bus microcomputer.
From the univariate experiments that were done prior to this
optimization, acceptable ranges for each of the parameters were
identified. The settings listed under "Reverse" are those which gave
low absorbance values and those under "Forward" gave high absorbance
values.

Some other considerations played a role in this decision making
process as well. The minimum pump setting (flow rate) indicates that a

minimum sampling rate of 24 per hour was required. The bath temperature

65

TABLE 5 Range and precision of variables for simplex optimization.
Experimental Foward Reverse
Variable Boundary Boundary Precision
Pump Setting 99 20 1
Bath Temperature (°C) 21.0 51.4 0.2
[POD] (mg ml‘l) 0.05 1.0 0.02
[AAP] (mM) 0.1 2.5 0.01
[DCPS] (mM) 0.1 2.5 0.01
pH Carrier, Sample 10.00 5.00 0.02
pH Reagent 10.00 5.00 0.02

66

was bounded at the low end by room temperature and at the high end by
37 ‘C-the maximum temperature shown to be advisable for use with this
enzyme. The high end of the PO concentration was set at 0.8 mg ml‘1
since it was desirable to limit the cost of the experiments and this
concentration was in the plateau region described previously. The upper
limit for the AAP and DCPS concentrations was such that if both were
used in their highest concentrations the reagent solution composition
would be no more than 25% (v/v) in each. The pH of the carrier and
sample was made independent of that of the reagent since it was unknown
whether the two reactions would have different optimal pH values. The
range for each was selected on the basis that the working range of the
Universal buffer is between pH 5.00 and pH 10.00.

The initial (baseline) conditions listed in Table 6 were known to
be suboptimal, but exhibited acceptable reproducibility (18 RSD). It
was found that a 400 ppm glucose standard, when used with a new GO SBSR,
gave a maximum absorbance value that was in the middle of the detectable
range and was, therefore, suitable for use in the optimization.

Since there were seven variables, eight experiments were needed to
form the initial simplex. These were performed and the results entered.
The first point of the simplex was the response obtained from the
baseline experiment with the initial conditions. Thereafter, each time
the simplex program specified a set of experimental conditions, this
experiment was performed and the time-corrected absorbance ratio was
calculated and entered. In reality two experiments were done in order
to obtain the response for one set of conditions: one at the specified

conditions and the other at the baseline conditions.

67

¢.N

N.HN

~m.c

m¢.m

Nc.m
Nm.m
mm.m
mm.m
N¢.m
o~.m

oo.m

acumen:
mm

ov.o

5%
CD

QQVQN
Heme-:08
OD bfiOl‘bCD‘D

8.

sewswmo
mm

m.N

m.~m

b.~m
m.mN
m.mm
m.mN
m.~m
m.Nm

N.NN

Gov

.osoosau mo cameosflILanv <Hb
om» mom mmowuwvmoo Hausa-whomxo Helwuao use Howummm

no.9

Humwl HI
can:
30mm

m mnm<a

@ no» mom
.>oa .vam

m may mom

5.: ado
oumao><

ob
mm

mm
mm m mos

A ammum

Caz
.mxm

68

The simplex rapidly improved the response at first. This was
followed by a more gradual increase and then a plateau region as
diplayed in Figure 22. This same pattern has been observed and
explained elsewhere (73). After twenty—seven experiments, the data
suggested that a "false maximum" had been found in experiment nine. To
check this, that particular experiment was repeated and was found to be
in error. The simplex was restarted with the original values for the
first eight experiments and the new value for the ninth. After the
thirtieth experiment with the new simplex it was intuitively felt that
the rate of improvement might be enhanced by use of a simplex that was
larger in the dimensions of temperature and PO concentration. Further
optimization was carried out using the current simplex points, but
replacing two of those experiments with two experiments at higher
temperatures and/or POD concentrations.

The optimization process was terminated when an optimum set of
conditions was deemed to have been reached. Since the composite
modified simplex method is a search procedure in which contraction
around a maximum will take place, it is likely that there will exist
several sets of conditions in the final data set that are near a true
surface maximum. By taking the average of the values for each of the
parameters for the experimental conditions that gave the six largest
time corrected absorbance ratios (i.e., those yielding a response that
was 758 or more of the largest value found), a rough measure of the
width of the optimum in each dimension was possible. These results are

given in Table 6.

69

qfiqfidJ—qqddqulddddddedddddlfiqdd

nw

w

o o

m. m.

02.5. uoz<mmomm< ouhoummoo m3...

000

20

10

o

EXPERIMENT NUMBER

FIGURE 22 Progress of the siqlex.

70

The values for the response function obtained for each of the
seventy experiments are plotted against each of the seven variables
individually. The diagr-s in Figure 25 show the results. These
diagrams should not be viewed as univariate experiments, because they
represent experiments in which all variables are changed. For this
reason, they are often called "scatter diagrams". From the conditions
giving best responses and the trends in the scatter diagrams several
observations can be made.

The pH of the carrier and sample showed a maximum around pH 7
(Figure 23a). This result agreed quite well with previous univariate
experiments. The discrepancy between the optimum for the i-sobilized
species and that for the soluble species may have been due to the
difference between the pH of the bulk solution and that seen by the
enzyme. Since the flow ratio of the carrier to the reagent stream was
10:1, it is suspected that the carrier pH would predominate and could
therefore be reasonably compared to the one pH used in the univariate
experiments. The optimum reagent solution pH (Figure 23b) was also
found to be higher than that considered optimum for P0 in solution (74),
however, it is known that that the optimum is dependent upon the
particular substrate and type of buffer used (75). The dilution factor
may also have played a role in this. Some of the mystery concerning the
real value of the pH in the manifold could be cleared in the future by
inserting a micro-electrode at various points in the stream.

By inspection of the plot for the variation of flow rate
(Figure 23¢), a maximum response is obtained at 0.56 ml min-1. This

trend is not like the steadily decreasing response found in the

71

P
m

“mm“
I
1 AAA
0
s
e
vs:
. s
'0
s
9
"WOW“
9
1
s
'nJ
A
s
s

 

 

 

 

 

 

 

 

 

: O s "
‘ . : O .5
1 ’ ... . 4 . .
-1 ‘s s s‘ ‘ ..O ome-1 {11“
d ‘ .0
.1 s‘ . ‘ “s
1 I ‘ ‘ s.s s
‘ 0‘s 4
. fi V V V'fi'T' fiTf' '1 a Y hi 'Y I I
«- mm s. m s. s. IO. m s. s. m s. s-
fllm sum
ms ms
s 3 s
i . , .
ms A I
m . s: . . .. .
. ..s 'sO. s
. .. .. O .
. s
O s
‘- ‘ . .. :A
5.. s ‘. ‘c... E .x“; ‘
.C‘ s . . .. s...0
i ‘ - . . x .. .
s. r ‘s
s s s O s . so sgs‘ O.
D vvvvv u u I l"' f* f t *fir
a. s.- “ mm s. .. us mm mm Is no
man (nu/m) ..qu
‘ I.
. E 5 . F
1 O s
ms .’ ms O Q. .
.)Q. ‘s 3
O . . . .O
‘ ‘ ‘ ‘ :5 ‘
fl .' .’ ‘ ‘0 . . O ‘5
‘ a. A . . g x
’ s : . :‘r I :. ‘ 0
fi . . O A . ‘
g 5‘“ ‘0‘: 5
.U—V—r"'"r""fi'Tr"'V r'I ”.rf 'IV—fiVU'Vfijf V['r
I. 0. I. II I. I. O. C. I. h. I.
m m- M m
I.
s
5 .
ms ‘ As‘
an ‘ . s
.. O A.
. O
s
I s...‘ ..
u .. . O“ . . .
3 r. “
. sfl
! . .
2 I ‘
as 3 O ‘
u I I 'I"'_" """""" I" I
I. ' I. I. I I. ' O. I.
M (av-o

FIGURE 23 Effect of experimental variables on
response function: (a) carrier pE; (b)
reagent pB; (c) flow rate, ml mm”; (d)
SBSR taperature, 'C; (e) DCPS
concentration, ‘4; (f) AAP concentration,
i4; (g) PO concentration, mg ml".

72

univariate experiments. The nature of the response function accounts
for this since a time factor was included.

The behavior of the response as a function of temperature
(Figure 23d) did not differ markedly from what was expected and was very
similar to the univariate results; there was a steady rise followed by a
plateau region. A rise in temperature would be expected to increase the
rate of the reaction. Modelling studies have suggested that it may also
decrease the dispersion (73) thus giving higher peak heights (this issue
is addressed in Chapter 6). There was no evidence of major interaction
between temperature and other variables in the system.

In this case also the effect of changing the concentrations of the
DCPS (Figure 23c) and AAP (Figure 23f) was simificant. The results
showed that the optimum ratio of the two was approximately 3:1. Once
again the optimum AAP concentration was found to be between 0.4 mM and
0.5 mM and the DCPS concentration was slightly greater than 1.0 mM. The
P0 concentration was also found to enhance the response (Figure 23g) in
a manner similar to that found in the univariate case, and exhibited a
typical Michaelis-Menten shaped curve.

The performance of the system (as measured by the magnitude of the
response function) was improved by a.factor of 22.5 (over the initial
conditions). A calibration curve obtained at the optimum conditions was
linear from 0.1 to 3.3 mM glucose and usable from 0.1 to 5.5 mM glucose
with the reagent concentrations specified. A comparison of the
calibration curve obtained with the simplex optimum conditions is
compared to that obtained with the univariate optimum conditions in

Figure 24w The slope for the simplex case is higher, but the linear

ABSORBANCE

0.100

0.500

0.400

0.200

0.100

73

 

TT'Ul’rrrrTrfirr'rf'V[T'fir'trfifi'

0.0 1.0

FIGURE 24

2.0 3.0 4.0 5.0 0.0
[GLUCOSE] (mu)

Calibration curves for Optill. operating
conditions arrived at by simplex
optimization (Q) and by univariate
experiments (0).

74

range for the univariate case is larger. It becomes a trade-off between
the capacity to determine the concentration with high precision and the
alternative capacity to analyze samples that have a wide range of
concentrations. For the analysis of food samples the latter choice is
appropriate. The former choice would be appropriate for clinical

determinations where a narrow range of glucose concentrations is found.

CHAPTERS

EXTENSIONS OF THE GLUCOSE OXIDASE FIA SYSTDG

After having gained a considerable amount of knowledge concerning
the optimum conditions under which i-obilized glucose oxidase could be
used most efficiently, several possibilities for the extension of that
knowledge were explored. Originally, it was hoped that the system could
be miniaturized to allow greater versatility in field applications. In
an effort toward that end, a microconduit was designed and tested. A
second area of interest was the interference of ascorbic acid in the
glucose oxidase/Trinder reaction. The problem was found to be
nontrivial. Finally, the feasibility of using the glucose oxidase
reactor for the analysis of samples containing sugar mixtures was
investigated. An interesting offshoot of that venture resulted in the
use of the reactor for monitoring the mutarotation of D-glucose in
phosphate solutions. A return to the main event resulted in the
assembly of a parallel system for the simultaneous determination of
glucose and galactose. Each of these extensions is discussed in detail

in the paragraphs that follow.
Design and Tea ting of the Mi crocanduj t Glucose 0x1“ dase SBSR

In 1984, when this work was begun, the premier article on the
construction and use of microconduits in flow injection analysis had
just been published (29). Since determinations with enzymes are used
routinely in various fields, it was felt that such a miniature system

would be of great utility. A microconduit was created in the original

75

76

work by engraving a "snake like" channel onto a small block of PVC (70 x
45 x 10 -) which was then covered with a transparent plate and sealed
in place with a pressure sensitive polymeric glue. In the work by
Ruzicka and Hansen the microchannel was a semicircular cross-section
with a typical area of 0.8 n3. Flow cells were constructed within the
block so as to make a self-contained unit that only had to be attached
to the detector via a fiber optic at one end and to the pump at the
other. More recently, there was a report of the construction of a
versatile injection valve within the block (76).

A sketch of the microconduit designed for this work is shown in
Figure 25. The shape of the channel was designed such that two
identical plates could be used. The cross-section of the channel on
each PVC plate was semicircular. By inverting and rotating one plate on
top of the other a circular cross-section (with an area of 0.8 mm?) was
obtained. Thus the channel created was similar to that found in a l--
i.d. tube. The creation of a channel with such a small radius (0.5 -)
is not easily accomplished with ordinary tools. In this case a very
smooth, uniform channel was obtained by use of a Computer Numeric
Controlled milling machine (manufactured by Charnoa). The machine was
prograned and operated by the workers at Mattson Tool 8: Die, Inc.
(Belmont, MI). The length of the channel is approximately 10 cm.

Two stainless steel plates were also machined and used to sandwich
the two PVC plates together. By tightening the screws at each corner
enough pressure could be applied to effectively seal the reactor; this
eliminated the need for any type of glue. Each metal plate had the

center cut out to facilitate connection of the ends of the channel to

77

 

FIGURE 25 Side (a) and top (b) views of a
microconduit SBSR.

78

external tubing. Holes were drilled perpendicular to the channel at
each of its ends to accomodate short lengths of Teflon tubing which were
used for the connection. Provided that the external surfaces are also
very smooth and flat another set of plates could be stacked atop the
first set without the need for any connecting tubing between the two
sets. By increasing the length of the screws used to connect the two
outer metal plates numerous reactors could be stacked together.

The microchannel reactor was packed with nonporous glass beads
containing immobilized glucose oxidase. This novel single head string
reactor was used in place of the usual lO-cm GO SBSR and a calibration
curve obtained. The operating conditions were the standard set adopted
earlier (Chapter 4). The curve was linear to 800 ppm glucose and had a
slope similar to that found for a 10-cm SBSR made from the same beads.
These results make the future very bright for i-obilized enzyme
microconduits.

Although the size of the channel employed here is not really
considered to be at the "micro” level by this author, it is identical to
channel sizes in other microconduits (29). The use of a 0.5 :- i.d.
cross section would fit the description more appropriately. Such

miniaturization has not yet come to fruition.
Ascorbic ch d Interference Study

Ascorbic acid is known to interfere with the peroxidase catalyzed
reaction. The nature of the interference is that the ascorbate acts as
an alternate oxygen acceptor for the peroxide attached to the

enzyme (77). In the case of the Trinder reaction, the sulfonate group

79

normally performs this function as it is eliminated from the DCPS. The
prenture abstraction of oxygen from the hydrogen peroxide may force the
3,5-dichloro-2-hydroxyphenyl sulfonic acid to obtain oxygen by another
pathway or not at all. A recent report by Thompson (78) provides
evidence that the former is the case. Other workers have also reported
negative errors in the Trinder reaction due to ascorbate (71). It was
therefore assumed that an investigation of the effect of ascorbate would
be routine. That proved not to be the case.

In studies of the effect of ascorbate, the same PO reagent
concentrations were used as in the delay flow experiments. A stock
solution of ascorbic acid (Sigma) was prepared fresh daily by
transferring 176.1 mg to a 1 liter volumetric flask and diluting to the
mark with distilled water. Ascorbate standards were prepared by
appropriate dilution of the stock solution. When not in use this stock
solution was stored in the dark since it is light sensitive. Glucose
standards were made from the stock solution described previously
(Chapter 4). Six glucose standards in the range of 0 id to 5.5 dd were
used. A complete set of experiments was carried out using five sets of
standards which were spiked with 0, 20, 40, 60, and 80 1M ascorbic acid,
respectively. All standards were prepared with 0.05 M phosphate buffer.

In the first round of experiments an increase in peak height was
concurrent with an increase in ascorbate concentration. The slope of
the calibration curve decreased slightly while the intercept increased.
The second round of experiments gave decreasing peak heights with
increasing ascorbate concentration and the calibration curves became

nonlinear. The only difference between the two rounds was that Brij-35

80

(a wetting agent) was added to the buffer that was used in the second
round. A third round of experiments was carried out with buffer that
did not contain Brij-35. Once again the intercept of 'the calibration
curve increased in direct proportion to the concentration of ascorbate
present. The slopes, intercepts, and correlation coefficients for the
first and third experiments are given in Table 7 . These data suggest
that the addition of the wetting agent has an adverse affect on the
Trinder reaction. Brij-35 is a 303 aqueous solution of
polyoxyethylene(23)lauryl sulfate. It appears that this inert
surfactant has some effect on the reaction.

In order to investigate the effect of the further, a standard
containing 2.2 dd glucose and 50 M ascorbate was prepared. Injections
from the solution were made at five minute intervals. The initial
absorbance was lower than that obtained with a non-spiked glucose
standard. However, the absorbance of the spiked standard gradually
increased in each subsequent injection. Since the standard was shaken
vigorously in order to obtain complete mixing the ascorbic acid may have
depleted some of the oxygen in the solution, and limited the amount
available for the glucose oxidase reaction. Since the container was
left open to the air, some of the oxygen in the solution may have been
gradually replenished.

The increase in the y—intercept merits some discussion. This
trend suggests that a chromophore of some type is formed when only
ascorbic acid is present. It may be that the ascorbate binds directly
to the peroxidase, thus creating a complex that absorbs at 510 nm. The

other possibility is that the ascorbate in some way facilitates the

 

81

TABLE 7 Effect of ascorbic acid in experiments
without Brij-35.

 

 

 

Expt. [Ascorbate] slope corr.
No. (uM)x10‘ x103 intercept coef.
1 0 4.31 0.78 0.99997
20 4.26 6.1 0.99999
40 4.25 13.0 0.99999
60 4.08 27.8 0.9998
3 0 6.11 -5.7 0.9991
20 6.08 8.4 0.998
40 6.19 23.1 0.9996
60 6.24 40.2 0.9996
80 6.15 57.2 0.9995

82

coupling of the MP to DCPS. By using the diode array detection system
one could look at the spectrum of the flowing stream just prior to and
while the sample plug containing ascorbic acid passes through the flow
cell. A decrease at any of the wavelengths associated with P0
absorbance coinined with an increase at 510 nm in the absence of
peroxide would confirm the formation of an ascorbate-P0 complex. A
second set of spectra could be taken with a reagent stream consisting of
only MP and DCPS to ascertain whether or not there is any direct
interaction between the ascorbate and the two color-forming reagents.
The effect of ascorbic acid on the Trinder reaction has proved to
be a nontrivial matter. Since this FIA system could be used ultimately
to determine the sugar content in food this interference is important.
Further investigation should yield concrete solutions to the problems
identified in this work. Because this was a substantial deviation from
the main goals of this project it was decided to leave this

investigation for the future.
Kinetic Study of D-Glucose Mutarotation

The major use of i-obilized enzyme systems has been for the
determination of specific substrates in various types of samples (11).
Chibata (7) gives an excellent review of other ways in which immobilized
enzymes can be used (e.g. structural analysis of biopolymers and the
elucidation of enzyme mechanisms). A novel way of implementing the
i-obilized enzyme reactor is by using it to follow the kinetics of a
batch reaction which produces a substrate for which the enzyme is

specific.

83

Previously described kinetic studies by FIA are predominantly
those of reactions that occur in the manifold. In one method the flow
is stopped and the absorbance of a species monitored as it reacts within
the flow cell (19). In another, the sample plug circulates continuously
in a loop that contains the detector (79). Painton and Mottola recently
showed that the rate constant for a reaction in a single sample plug can
be determined by comparison of the signal profile with those from
numerical simulations (80). FIA has only recently been used to study
the kinetics of relatively slow batch reactions (81-83). This is
accomplished by injecting a sample from the solution in the batch
reactor onto the FIA manifold at prescribed time intervals. The
previously discussed FIA syst- containing a GO SBSR was employed. to
study the mutarotation of D-glucose.

This reaction is well-characterized (84, 85) and therefore can be
easily used as a model since the kinetic parameters are well known. The
reaction (see Figure 26) proceeds by a first order reversible process
until equilibrium is reached. The integrated form of the rate equation

for this process is:

(F ‘ K's)
-ln = (R1 + k2)t

(‘0 " 3“!)

where 8), h, and fig are the concentrations of the B anomer initially,

 

at time t, and at equilibrium, respectively. The term, k1 + k2, is
known as the mutarotation coefficient. It is more couon to use the
optical rotation of the solution in place of the concentration of the p

anomer (86). However, since glucose oxidase is specific for the p

34

a-.. .8
. 080:

3320 .. o 1 Q

 

as:
.3 sod venous
ssooo

9.

3.

mu 95lo

u 6
3320-0

 

85

anomer of D-glucose the absorbance from the product of the Trinder
reaction can be used.

A set of experiments was designed in which the apparatus
(described in Chapter 4) was used to follow the mutarotation kinetics of
D-glucose in the presence of various concentrations of phosphate. The
mutarotation process was also followed for D-glucose in the presence of
D-galactose. Since D-galactose also mutarotates, the conventional
method for this type of study (polarimetry) would be more complex than
the method described here.

A 0.2 M phosphate buffer stock solution was prepared by adding
13.609 g of 161sz (Mallinckrodt) and 14.196 g of Noam (Fisher
Scientific Co.) to a 1 liter flask and diluting to the mark with D".
Additional buffer solutions were made by dilution of the stock solution.
In each case the pH was adjusted to 6.84 by addition of 0.2 M neon (made
with pellets from B.M. Science).

The reagent solution for the Trinder reaction was prepared in a
volumetric flask immediately before use. It was composed of 0.8 mg ml"1
PO, 1 mM AAP and 1 mM DCPS, diluted to the mark with 0.05 M phosphate
buffer.

Glucose solutions were prepared at the time of use by measuring
15.0 mg of the appropriate anomer into a 25.0 ml volumetric flask and
diluting to the mark with the appropriate solution. For the experiments
in which galactose was also in solution, 15.0 mg of D-galactose was
added to the 15.0 mg of glucose before transfer to the volumetric flask.

The overall flow rate was increased to 1 m1 min'l. Four different

phosphate concentrations were used: 0.00, 0.05, 0.10, and 0.20 M. Only

86

the 0.05 M phosphate buffer was used in the case where galactose was
added. The mutarotation process was followed in both the forward and
reverse directions, starting with the pure 1 and 3 forms, respectively.
For each anomer in each concentration of phosphate, the following
experiment was repeated 2-4 times. The ancner was transferred to the
volumetric flask and the clock was started. The sample was diluted as
described above and allowed to dissolve completely. The tube leading to
the sample injection valve was inserted into the flask and the pump
started. After 90 s the solution had had sufficient time to occupy the
sample loop and the first injection was made. Subsequent injections
were 60 s apart. Each injection was followed by monitoring the output
voltage from the detector.

Mutarotation coefficients were obtained by means of a simplex
fitting routine which was based on the method of Nelder and Mead (87)
and written by Pete Wentzell. It was necessary to provide the program
with an equation describing the relationship of the x and y values to
the constants that result from the fit. The following model was used to

find the best values of the three parameters—(Ro-Boq, k1+k2, and fig):

Rene = Foq+(Bo-Koq)*EXP(-t*(k1+k2))
This is simply a different version of the previous equation in which all
of the 3 values represent peak absorbance rather than concentration of
the 3 anomer. The calculated 3 value, Beale, was obtained during the
fit. All points were weighted equally. Initial values of the three
parameters were supplied by the user. In the case of the a anomer, Bo
was assumed to be zero. For the B anomer, Bo could not be measured

directly, and therefore the peak absorbance vs. time data were

87

extrapolated back to t = 0. Initial In values were calculated from the
average of the last two peak absorbance values, except in the case of
the water solution. It is known from polarimetric measurements (86)
that at room temperature 643 of the D-glucose is in the 3 form at
equilibrium and thus 3" for the water solution was estimated by
appropriate multiplication of the 30 value. Initial estimates of k1+k2
were obtained by calculating the initial slope of the curve for each set
of data points.

The initial simplex consisted of four vertices and was calculated
from initial estimates and an appropriately chosen step size for each of
the three parameters. At each vertex the following procedure was
performed. For each value of t in the data set a value of Rule was
computed. The squares of the differences between this value and the
experimentally obtained value (3) were sued for all values of t using
the function: R = 2 (f-BcalcF. The magnitude of this sum of the
squares of the residuals, R, was a measure of how closely the current
estimates for the three parameters matched the experimental behavior.
The simplex moved through successive iterations as it tried to minimize
R (i.e., to obtain the closest match possible). Convergence was assumed
when the sum of the squares of the residuals did not decrease after

further iterations.
Determination of Mutarotation Coefficints

Convergence of the simplex was usually obtained in less than 100
iterations and was not more than 140 in any instance. Complete

separation of Ba and flu values in the equation used for the fitting

88

procedure was investigated. This did not decrease the nuiber of
iterations necessary nor did it improve the final R value obtained. The
experimental points (shown in Figure 27) are connected by the
corresponding fitted curves. The fitted curves show good agreement with
the experimental data. The standard error of the estimate was less than
103 and the between run deviation was at most 53. The mutarotation
coefficients obtained with a and E anomers in various phosphate
concentrations are listed in Table 8.

The generally accepted value for the mutarotation coefficient of
D-glucose in water at 20°C is 0.015 min"1 (50, 88). The value of
0.018 min“1 at 21°C obtained here is in good agreement with the
literature value. The small deviation can be accounted for by the

slight temperature difference and the standard error of the measurement.
Effect of Phosphate Concentration

It is well known that phosphate ion catalyzes the mutarotation of
D-glucose (84). Preliminary experiments indicated that the phosphate
concentration had no affect on the GO/Trinder reaction for an
equilibrium mixture of a- and fl-D-glucose. Therefore, the increase in
exponential character of the curves (Figure 27) is attributed to the
catalytic effect of the phosphate ion on the batch reaction alone. The
results, shown in Table 8, indicate the mutarotation coefficient to be
linear with phosphate concentration. This same trend was observed in
the polarimetric studies of Murschhauser (89). The slopes, intercepts,
and correlation coefficients are given in Table 8. The mutarotation

coefficient obtained when the D-glucose is initially in the a

0.400 -

89

0.300.: W

M

g .

m 0.200 -

O

a 1
0.100 -

 

‘ °-°°° ‘WWWWWWWWM

0 300

FIGURE 27

TIME (sec)

Effect of phosphate concentration on rate
of B-glucose mtarotat ion. Lines
represent curves of best fit. Upper
curves were obtained with only fi-D—glucose
initially present. Lower curves were
obtained with only e-D-glucose initially
present. Phosphate buffer concentrations
were: (Q) 0.00 M, (O) 0.05 M, (s) 0.l0 M,
(I) 0.20 M.

TABLE 8

Phosphate Buffer
Concentration (M)

90

Mutarotation coefficients for D-glucose at 2190.

Coefficient x 103(s‘1)

only a initially

only } initially

 

0.00 0.3 0.3
0.05 1.9 1.5
0.10 3.3 2.6
0.20 6.9 5.0
slope (Mrls‘l) 0.033 0.024
intercept (s‘l) 0.00016 0.00027
correlation coeff. 0.998 0.999

91

conformation is larger than that for the 3 conformation by approximately
30%. The values for the y~intercepts in the equations relating
phosphate concentration to the mutarotation coefficient indicate that
the coefficient in water ought to be 0.009—0.018 min”, which compares

favorably with literature values.

Effect of Galactose

Since the apparatus employed here could be used to investigate the
mutarotation of D-glucose in the presence of other sugars and
mutarotase, it was desirable to demonstrate the selectivity of the GO
reactor. The main interference envisioned was that of galactose since
its mutarotation is catalyzed at a rate similar to that for glucose.
The mutarotation coefficients for D-glucose in the presence of an equal
amount of galactose were found to be 1.9 x 10"3 s‘1 and 1.7 x 10'3 s’1
for the a and 3 forms, respectively. These values agree with those
obtained for the solutions containing glucose alone, and thus
demonstrate the selectivity of the glucose oxidase reactor. This
further substantiates the claims of specificity made in Chapter 3.

Since the enzyme is specific for a single sugar, this method
facilitates the study of the mutarotation process for that sugar in a
matrix of other optically active species (e.g., a mixture of sugars
several of which mutarotate). The mutarotation of some sugars is
complex since, when in solution, furanose as well as pyranose structures
are present (84). It may be possible to investigate separately the slow
reactions of the pyranose form of other sugars with an apparatus similar

to that described here.

92

Parallel Determine ti on of Sugars

Galactose oxidase was chosen as a complementary enzyme to glucose
oxidase. The interest in the determination of galactose arises from
both the nutritional aspect as well as the clinical standpoint. It is
considered one of the six nutritionally-significant sugars (90). The
reaction of galactose with galactose oxidase also produces hydrogen
peroxide. From the clinical standpoint the condition of galactosemia
sometimes found in infants causes elevated levels of galactose in the
urine. The conventional method of diagnosis is based on a negative
glucose test and a positive test for reducing substances, followed by
identification of the reducing substances (91). This procedure could be
greatly simplified by determining the galactose level directly. Since
galactose oxidase is more expensive than glucose oxidase, immobilization

of the enzyme was thought to be cost effective.
The Galactose Oxichse SBSR

The procedure used to immobilize glucose oxidase was also used for
galactose oxidase. The critical parameters of pH and enzyme
concentration were pondered. Assay procedures employing galactose
oxidase specify a pH between 6 and 7 (92, 93). This, would suggest that
the enzyme is most active in that pH range. Therefore, since it is
desirable to i-obilize the enzyme in its most active state the same
phosphate buffer was used. The concentration of the enzyme solution was
determined by calculating the amount of enzyme necessary to give an
equivalent nuiaer of Units of activity as that obtained for glucose

'oxidase. The appropriate portion of galactose oxidase (E.C.l.l.3.9,

93

Sigma Type V from Dactylium dendroides, specific activity approximately
70 U mg'1 at 25°C) was diluted with 5 m1 of phosphate buffer. This
solution was used in the place of the glucose oxidase solution in the
i-obilization procedure. Other workers have i-obilized galactose
oxidase on porous glass (94, 95), but this is the first time nonporous
glass heads have been used.

The nonporous glass beads containing the enzyme were put into a
50-cm piece of Teflon tubing since it was suspected that the activity
would be less than that available from the GO SBSR. A galactose stock
solution was prepared by transferring 200 mg to a 100 ml volumetric
flask and dilutin to the mark with DW. Galactose standards were
prepared from the stock by appropriate dilution with phosphate buffer.
Although the actual absorbance values obtained were much less than those

for CO, the calibration curve was linear from 0.13 to 3.3 ‘4 galactose.
I-obilized Galactose Oxidase Specificity

The galactose (GA) SBSR was tested for specificity against the
other five nutritionally-important sugars: maltose, lactose, glucose,
sucrose, and fructose. Standard solutions (3.3 IN) of each sugar were
prepared by transferring the appropriate amount of powder to a
volumetric flask and diluting to the mark with phosphate buffer. The
results are given in Table 9 as a percentage of the response obtained
with galactose. An interesting phenomenon occured with the sucrose and
maltose samples: negative peaks were produced. This was probably not a
result of a reaction, but rather a change in refractive index which

caused an apparent decrease in absorbance. Significant absorbances were

94

TABLE 9 Galactose oxidase activity toward other sugars.
m 3.12:2
galactose 1.00
glucose 0.12
fructose -
sucrose -
lactose 0.28
maltose -

”—” indicates no detectable response

95

obtained with glucose and lactose. This may be the result of a slight
glucose oxidase impurity in the enzyme preparation. The relatively high
response to lactose was expected. Other workers have used GA reactors
for the determination of lactose in the absence of galactose in
substances such as milk (94, 95). Lundback (95) devised a procedure by
which the sszyme is further purified prior to the i-obilization
procedure. This could be implemented in the future to obtain greater
activity per gram of beads.

Galactose oxidase requires more care than glucose oxidase. It is
recs-ended that the i-obilized version be stored in a copper
sulfate/EDTA/Tris buffer solution in order to maintain the activity of
the enzyme (95). The copper(III) in the enzyme changes to copper(I) as
it catalyzes the reaction. Via a side reaction it sometimes reaches an
inactive state (copper(II)) which can be restored by use of
hexacyanoferrate. For future experiments these procedures should become
routine. Consideration must also be given to inactivation of the enzyme

by high galactose concentrations.
Design of the Parallel FIA System

The FIA system was modified to accomodate the two i-obilized
enzyme SBSRs. The sample size was increased to approximately 70 ul and
a tee placed i-ediately after the injection valve. The GO SBSR was
connected to one arm of the tee and the GA SBSR to the other. In order
to accomplish even splitting of the stream, a 40-cm plain SBSR was added
between the tee and the GO SBSR. This compensated for the pressure drop

associated with the 50-cm GA SBSR. Pump tubing and speed were adjusted

96

so that the overall flow rate was maintained at 0.5 ml min‘1 and a flow
ratio of 10:1 preserved. The manifolds were identical following each
i-obilized enzyme reactor-the same as that used for one reactor. The
parallel syst- is shown in Figure 28.

Alizarin red (0.00025 M) was injected into a stream of 0.05 M
phosphate buffer to determine the actual percentage of sample going to
each side. Plain 8888s of the same dimensions replaced the respective
enzyme SBSRs. Each peak obtained was integrated and the area taken as
representative of the total amount of sample reaching that reactor. The

GA SBSR was recipient of 50.63 of the injected sample.
Analysis of Real Samles

No sets of calibration standards were prepared from stock
solutions of glucose and galactose. One set contained glucose and the
other set contained galactose. The range of standards was 0.0 to
5.5 114. The calibration curve for glucose was linear from 0.1 m to
5.5 id. The linear range for galactose was from 0.1 nM to 3.3 IN. The
glucose oxidase did exhibit a small, but perceptible response to the
galactose standards. As mentioned earlier, there was also a small
response to glucose by the galactose oxidase. As a preliminary check on
the systsm’s ability to perform accurate determinations of glucose and
galactose simultaneously, a mixture containing known amounts of the two
sugars was prepared. The amounts determined were found to be within 10%
of known values.

A few real samples were analyzed with this system. The peaks

obtained are shown in Figure 29. Pepsi was found to have 4.92X glucose

97

689816 was 030qu
ho nausea-L31. “0:93.— ..oh Bolas en..— QN HBO:

 

 

ssasalso

 

 

 

 

..sassasmlr

 

l

 

ssassasa

 

 

fi. 9.0.8

 

 

 

 

 

 

 

 

 

 

J1 _ all .32....

8.0 G. ‘0 Q. '0 0’

mama om slam ll use—.30

Was» «a so as

 

 

 

 

 

 

ABSORBANCE

0.150 -

 

0.000 '4

 

Li.

 

FIGURE 29

98

i. -- from GO snsa
u, - from GA sasn

LC -0

 

:—

 

 

 

 

 

 

 

L] ..J

l I I r r l I I U I l

0 200
TIME (sec)

 

FIA profiles for : (a) Pepsi, (b) Sunglo
fruit punch, (c) diluted serum,
(d) deproteinized serum.»

1!“ n - "tr-11'“

99

and no galactose. This result agrees well with the value (4.95%)
determined by a different method (42) and the nominal value reported in
that same work (4.3%). Sunglo fruit punch was found to contain 3.48%
glucose and no galactose. Reconstituted serum samples (SeraCh- by
Fisher) were also tested. The first peak was obtained from a 1:5
dilution of the serum with DW. The glucose concentration was found to
be 77 mg dl‘l, compared to 78 i 10 mg dl'1 given in the manufacturer’s
data sheet. The initial negative peak obtained from the glucose oxidase
reactor was probably due the difference in viscosity between the carrier
and the sample. This effect has been observed by other workers (96).

To correct for the baseline interference in the GO reactor the
serum sample was deproteinized. To 1.0 m1 of serum in a centrifuge tube
was added consecutively 1.5 m1 Ba(OH)2 solution (20 g 1’1) and 1.4 m1
ZnSOs solution (20 g 1‘1). After the solution was thoroughly mixed, it
was centrifuged for three minutes. The supernatant was used directly
for the determination and found to contain 74 mg (11"1 glucose. In
addition to the good agreement with the published value, the FIA profile
was much improved. The presence of protein did not affect the
determination (i.e., the two values were equivalent within the
experimental precision of this system--an 880 of 5:: for three replicate
determinations); this substantiates an earlier report (71). The
activities of the reactors were not adversely affected by reaction with
a sample containing protein—-the slopes of the calibration curves
remained constant.

The large response of the galactose oxidase reactor to the serum

sample was beyond the range of the calibration curve. The data sheet

100

did not list any substances which might have caused such a high
response. Each successive injection yielded an increased response.
This indicates that some type of activation was taking place and
provides solid evidence that further precautions need to be taken when
working with this enzyme.

From these results it can be concluded that parallel determination
of sugars is quite feasible. A major improvement in the systsm
performance could be attained with a sample selector. This would allow

the individual optimization of each channel.

CHAPTERS

THE EFFECT OF TWERATURE ON DISPERSION

Considerable attention has been given to the study of dispersion
in flow injection analysis (19, 73, 97-101). Flow injection brought to
analytical chemistry a concept that was entirely new, that of
"manipulated dispersion” (102). Prior to this, complete mixing of
sample and reagent solutions was considered a prerequisite for
performing a chemical analysis. In most cases it is desirable to
control dispersion rather than minimize it, since some amount of
dispersion is usually required to facilitate a chemical reaction, dilute
the sample, or produce a gradient. Sample throughput may be dispersion-
limited for systems that merely transport a sample, and in these minimum
dispersion is advantageous.

In an attempt to quantify dispersion a numerical technique was
used to solve the diffusion-convection equation (97). Elsewhere, Reijn
et al indicated the additive dispersive effects of the processes of
injection, transport, and detection (98). Other workers have studied
the various aspects of system design and experimental parameters which
affect these. Ruzicka and Hansen recognized the importance of
dispersion when they included sample volume, tube length, flow rate, and
the presence of a mixing chamber into seven rules for developing FIA
systsms (19). Vanderslice et a1 (99) indicated that the diffusion
coefficient and tube radius play a role in dispersion. A further
dispersive effect during the transport process has been attributed to

chemical reaction (19, 31, 73, 103). Painton and Mottola (104) also

101

102

point out that the usefulness of the ”practical dispersion nuiaer” can
be invalidated or obscured when ch-ical reactions take place. The
nature of the reaction manifold used (straight tubes, coiled tubes,
single head string reactors, etc.) has been shown to be a key factor
(105). Dispersion introduced during detection was recently partially
attributed to the flow cell configuration (106). A recent article by
Stone and Tyson (107) provides a model for the dispersion process that
takes into account the flow rate, tube length, tube inner diameter, and
method of injection. The complex interactions of all these factors in
the ever widening range of flow injection analyses are still not well
characterized. It is obvious that deeper theoretical and practical
studies of the processes that cause a sample plug to become dispersed as
it travels through a flow injection analyzer are necessary (102).

In all of the discussion concerning various sources of dispersion
there has been little mention of the effect of temperature. A
hypothesis can be made regarding the relation ship of dispersion and
temperature from the equation derived by Ruzicka and Hansen that
describes the relationship of dispersion, D, to various experimental

parameters. It is as follows:

Vr L0 1‘
D= 'Ir2
Sv 120-

where Vr is reactor volume, Sv is sample volume, L is tube length, 0 is

 

flow rate, and r is tube radius. The molecular diffusion coefficient,
BI, is proportional to temperature. This implies that dispersion is

inversely proportional to temperature.

103

. No experimental studies have been published until now (108) that
substantiate this hypothesis, although the case for such is strong in
light of the circumstantial evidence available. For example, (i) in FIA
viscometric studies temperature control was necessary to obtain the
required reproducibility of peak height measurements (103); (ii)
simulation studies indicated that an increase in temperature (and the
resultant decrease in viscosity) would promote radial mixing, decrease
longitudinal dispersion and increase peak height (73, 106); (iii) peak
shapes changed as the temperature of one flow injection system was
increased (109). The reaction kinetics which take place in flow
injection analyzers are often favored by higher temperatures. For slow
reactions this can be used to achieve greater sensitivity or decreased
time per analysis, since the "optimum" experimental conditions will
change.

An investigation was launched in an effort to validate the
aforementioned hypothesis. Four types of reactions were studied in
order to ascertain the effect of temperature on the on the FIA peak
profile. The first type involved the injection of a dye into a non-
reactive carrier stream. The results from this are compared to those
obtained for a fast reaction, that of p-nitrophenol with sodium
hydroxide. The complexation reaction of nickel and PAR
(4-(2-pyridylazo)resorcinol) was used for the investigation of a
moderate speed reaction. The same apparatus configuration was used for
all three reactions and is therefore described first. And last, but
certainly not least, the effect of temperature on the glucose

oxidase/Trinder reaction was used as a model of a slow reaction.

104

The Apparatus for Cases 1 through 3

The FIA system for this set of experiments was asselwled from the
basic components described earlier (Chapter 4) and is shown in
Figure 30. The manifold in this case was very simple. The sample was
injected into a carrier stream which flowed through 1 m of coiled 0.8 -
i.d. microline tubing (2.0 cm coil diameter) to the detector.

The sample container, reagent container, and manifold coil were
immersed in a water bath. This consisted of a large pyrex
recrystallization dish filled with water and placed on a hotplate which
had an adjustable heating control. A thermometer was immersed in the
water at the level of the coil and containers.

Accurate compensation for the slight heat loss that occurred as
the solutions travelled from the thermostatted vessels to the detector
flow cell was attained by measuring the temperature of the stream as it
entered the flow cell. The precision 30-“). thermistor was used as
described earlier (Chapter 4) to obtain temperature calibration curves
at various flow rates. Distilled water was pumped through the systsm
for this process. Table 10 shows the slope, intercept and correlation
coefficient for the linear relationships between bath temperature (Tb)
and manifold temperature (Tm) at each flow rate.

Data acquisition was accomplished with the Bruce Bus
microcomputer. A data set was obtained from five replicate injections
for each 5° increment of temperature. The files for each of the
replicates were then transferred to the DEC L81 11/23 minicomputer where
the counts were converted to absorbance units and plots of each peak

were obtained with the aid of a two pen plotter. A baseline correction

105

.325: assassin. sou 'ashs 4H.— 8 go:

 

 

 

 

 

whm<3 Ob

 

.20.. om...de m4m2<m

 

 

106

TABLE 10 Temperature correction equations for various
flow rates.

Flow Rate
(ml min‘l)

0.56
1.20
1.88

1.00

Equation:

0.947
0.952
0.860
0.650

Th

= ath + b
Correlation
Coefficient
.17 0.9999
.02 0.9999
.25 0.9997
.46 0.9999

107

for each plot was obtained by measuring the height of each peak by hand.
For convsmience, the W (full width at half the maximum height) was
also measured by hand. A program written in FORTRAN was used to obtain
the area of each peak by integration with Simpson’s rule. In each case

the average of the five replicates was taken as the "true" value.
case 1: Nb fibection

A solution of alizarin red (0.00048 M) was prepared on the day of
use by adding 33.1 mg of the powder to a 200 m1 volumetric flask and
diluting to the mark with 0.05 M phosphate buffer. The solution was
stirred until dissolution was complete.

The dispersion characteristics of the single line manifold were
investigated for a temperature range of 20-70°C at flow rates of 0.56,
1.20, and 1.88 ml min"1 . Before injection, the sample solution, carrier
solution, and reaction manifold were allowed to come to temperature
equilibrium with the water bath. A 30-111 sample of alizarin red
solution (0.00048 M) was injected into a stream of phosphate buffer
(0.05 M) which flowed through the coiled manifold. The absorbance of
the stream was monitored at 510 nm.

In an additional set of experiments, the peak height for the
undiluted dye, flux, was measured. The same procedure was used as
above, except that the sample ’injection’ was a continuous stream of the
dye, rather than a finite plug.

The effect of temperature on dispersion for the system without
reaction is shown in Figure 31. The dispersion values were calculated

by dividing Ihax by H. For each of the flow rates studied, the same

108

.73. a! 84 Alv
mas .Tswm ml 84 Adv .Tsul ml 8.0 A3
”one: seas.- ..o: .8300?- ago—.33

Seasons? .8 83209-3 so Boots 3 552..

CL wmabummazmk
2. om on 9. on

Preps-bps.-—-bps—-h-b-ssnp

 

om
. con
r
moon
r
W
[cos
H.09.
_. fl

 

c3
0.
un

(H/"°‘“H) Nousaadsm

109

trend was seen; increased temperature resulted in decreased dispersion.
Since no change in the total amount of dye present was expected, it was
initially concluded that the change in dispersion was due to
temperature. Over the temperature range of 26°C to 70°C, for the flow
rates of 0.56, 1.2 and 1.88 ml min‘l, the dispersion, D, decreased by
24.5, 21.1 and 21.1% respectively.

travelled through the manifold, the effect of temperature on peak area
was investigated. The horizontal lines obtained (Figure 32a) indicated
that, indeed, the peak areas did not change with temperature. The
"areas" shown for the three different flow rates indicate that the
integration of peak height was originally carried out over time, not
volume. When these areas were multiplied by the appropriate flow rates
they became equal within the limits of experimental error.

Figure 32c shows that the peak height increased with increased
temperature: and decreased flow rates. The. experimental flow rate
dependence has been reported elsewhere (109) and is in accordance with
trends predicted by theory (19, 73, 106).

Figure 32e shows that the peaks became narrower as the temperature
increased. Also, the steepness of the initial slope decreased as the
flow rate increased. This was thought to be due to action of the flow
rate and temperature upon the dispersion process. Theoretical
predictions and interpretations given elsewhere (73, 106) indicate that
decreased longitudinal dispersion may be directly attributed to an

increase in radial mixing.

PEAK HEIGHT (inches) PEAK AREA

FWHM (inches)

110

 

 

 

 

 

 

 

 

uz.oo-. . , , . , . , , , uz.oo-
. d . 0 . I 0 . . I
‘ I0.00-u 0' ° °
I000.4 A ‘4 B
8.00‘ 8.00"
smi. ‘ 4 O s s A 4 4 4 600:. . . ‘ . . . s O s A
4.001. s s s s s s s s m s 400+ . . . - s s s ' ' 0
2.00 'V'I'VV'lr'r'r'v'frvrrfi 2.00
20 30 40 50 60 70 20 30 40 50 60 70
4.50- 3.15:0q E
I C . D . ' l
1 . . 1 . . E
‘ . ‘ s u:
9 . s s 3.00? . . s . .
. ° s
3.50. . ‘ . . j . . .
4 . O a . . 1 . . .
.1. . . : m I 250:. s . . I
I d I I
‘4 m I * Is '
2.50 2. kvvarvr tvvvr'vvvv
20 30 40 50 60 70 20 30 40 50 60 70
|.00 0.80
s . E 0 . F
. ' s .
0.80 ° .
‘ s . 0.60 ' ' s .
o s s ' s
0.60
s
. A 0.40 ‘ .
0.40 “ ‘.... “‘..
' . ' . I I s s s s . . ‘
. . . . . ' ' I . I m I
ozc.----,....,.rrr,n.., . .., 0.20. rr.,ffi.,.. ..
20 30 40 50 60 70 20 30 40 50 60 70

TEMPERATURE (°C)

FIGURE 32

TEMPERATURE (°C)

Effect of temperature on peak area without
reaction (a), peak area with fast reaction
(b), peak height without reaction (c),
peak height with fast reaction (d), F"!!!
without reaction (e), and mm with fast
reaction (f). Flow rates were: (I) 0.56
ml min”, (A) 1.20 ml min”, and
(I) 1.88 ml min“.

l 1 1
Case 2: Fast Reaction

A p-nitrophenol stock solution (0.01 M) was prepared by adding
0.6958 g p-nitrophenol to a 500 m1 volumetric flask and diluting to the
mark with DW. The solution was stirred until dissolution was complete
and then transferred to an amber bottle and stored at 4°C. On the day
of use a 0.0002M solution of p-nitrophenol was made by appropriate
dilution of stock solution with DW. A sodium hydroxide stock solution
(0.2 M) was prepared by diluting 12 m1 of 50 X (w/v) aqueous sodium
hydroxide to 1 liter with DW. The other desired sodium hydroxide
solutions were made by subsequent dilutions of the stock solution with
DW. The sodium hydroxide solutions were then standardized by titration
against anhydrous potassium hydrogen phthalate with phenolphthalein as
the indicator (110).

Since the interaction of p-nitrophenol with NaOH is a simple acid-
base reaction, it was assumed to be fast. Prior to the FIA experiments
a stopped-flow experiment was carried out to confirm this. A
p-nitrophenol solution (0.0008 M) was instantaeously mixed with NaOH
solution (0.01 M). The data acquisition rate was set at 1 ms'1 (the
fastest available on the custom built apparatus (111, 112)). A nonzero
absorbance in the range expected for the completed reaction was observed
immediately--no change was observed for a period of 100 ms. This showed
that the reaction rate is very rapid compared to the time that would be
required for the sample to travel through the FIA manifold.

The experimental parameters employed in this set of experiments
were the same as those for Case 1. A 30-111 sample of p-nitrophenol

solution (0.0002 M) was injected into a stream of sodium hydroxide

112

(0.01 M) which flowed through the coiled manifold. The absorbance of
the stream was monitored at 420 nm.

The slight positive slope below 50°C for the areas of the
p-nitrophenol curves (Figure 32b) indicated that the peak areas (and
hence, total absorbance of product) were increased by temperature. It
was thought that this effect might be due to a shift in the chemical
equilibrium, rather than a reaction rate or mixing phenomenon.
Therefore, static experiments were carried out to see if product
absorbance increased with temperature, irrespective of any sample
dispersion effect. Since the sample would see a range of hydroxide
concentrations in the flow injection systu, a range of hydroxide
concentrations was used in these confirmatory experiments. Three
premixed solutions containing 0.0002 M p-nitrophenol and, respectively,
0.005 M, 0.01 M, and 0.05 M sodium hydroxide were used. These solutions
were passed sequentially through the FIA apparatus without injection.
The absorbance was monitored at 420 nm relative to 091 over the same
temperature range as above, except that 15°C temperature increments were
used. No change in absorbance with temperature was observed. Thus, the
change in peak area observed could not have been due to a shift in
equilibrium and another avenue of investigation was pursued.

The reaction of p-nitrophenol with sodium hydroxide is very fast
compared to the time the sample takes to traverse the manifold
(milliseconds compared to seconds). Thus, the most likely cause of the
observed effect was one of temperature on the sample dispersion, i.e.,
in this flow injection system, at temperatures below 50°C, insufficient

mixing occurred to facilitate complete reaction. This was corroborated

113

by calculation of the number of tanks, N, for this system. Here N was
approximately 5 for all three flow rates. This low value indicated
incomplete mixing. An additional set of experiments was carried out
with a 3.0 m coil in place of the 1.0 m coil. It was found that under
these conditions the peak area did not increase with temperature. Thus,
incomplete mixing was confirmed as the cause of the artifact in the
original data set.

The effect of temperature on peak height for the reaction
(Figure 32d) was similar to that observed in the system without
reaction. An exception was noted for the 1.20 ml Illin‘1 flow rate; the
peak height was lower than expected for temperatures below 40°C. This
deviation could not be attributed to experimental error, and does not
lend itself to an easy explanation. The trends for the values of the
mm (Figure 32f) were the same as those in Case 1.

A typical set of peaks obtained at different temperatures for this
fast reaction is displayed in Figure 33. It is of interest to note that
there is an increasing amount of noise in the baseline as the
temperature increases. This could have a significant effect on the
limit of detection at high temperatures. The nature of this artifact

would make an interesting topic for future investigation.
Case 3: Moderate Reaction

A nicke1(II) stock solution (10‘2 M) was prepared by measuring
0.2908 g Ni(N03)2 61-120 into a 100 ml volumetric flask and diluting to
the mark with DW. A stock solution of PAR was prepared by transferring

25.5 mg to a 100 ml volumetric flask followed by dilution to the mark

ABSORBANCE

0.700
0.000
0.500
0.400
0.300
0.200

0.100

 

LLJLlIAAJJlLlLJllJlLlLJilJlLl

22°C

 

 

FIGURE 33

J JLJLMMJL.

114

69°C
59°C
50°C R fl
40°C n
3I°C n

 

 

 

 

 

 

 

 

 

 

 

r U T +1 TT r I r I
0 100 200

TIME (sec)

FIA profiles obtained for the experiment
with p-nitrophenol under the following
conditions: flow rate 1.20 ml min“; coil
length 3.0 m; and teqerature range
20-70'C.

115

with DH. The nickel and PAH reagent solutions were made on the day of
use and were 2 x 10'5 M and 10“ M, respectively. Since the cmplex is
1:2 a l to 5 reagent concentration ratio was deemed sufficient. Since
the complexation reaction is pH-sensitive, a 0.05 M phosphate buffer was
used to make the solutions used for the experiment. The pH of this
buffer (6.85) is well within the pH range over which complexation takes
place, and chosen to be less than pH 8 so that no precipitation would
occur.

As in the case for the fast reaction, a stopped-flow experiment
was carried out to determine the length of time required for the
reaction to reach completion. The reagent solutions were prepared as
indicated above. A data acquisition rate of one point per 800 ms was
used to obtain the reaction curve shown in Figure 34. The nonzero
absorbance present at the initiation of the run was due to the
background absorbance of PAH at 510 nm. The progress of the reaction
approaches a plateau region after 80 s. This information was utilized
in the choice of coil length used in the study.

The apparatus as described earlier was modified slightly for the
nickel/PAR experiments. The jacketed reagent vessels connected to the
thermostatted water bath (previously used in the enzyme studies)
replaced the make-shift water bath on the hotplate. The reaction coil
was innersed in a 250-ml jacketed flask (also connected to the water
bath) filled with water. As before, compensation was made for the loss
in heat by calibration with the precision thermistor. The flow rate of
the carrier stream (PAR reagent solution) was 1 ml min”. Nickel

reagent solution (30 111) was injected into the carrier stream, passed

ABSORBANCE

116

 

 

FIGURE 34 React ion

TIME (soc)

progress curve

for

nickel

(21(10'5 M) with PAR (10“ M) at pH 6.$.

117

through the reaction coil to the detector where the absorbance was
monitored at 510 nm. Two different lengths of reaction coils were used.
A 1.0-m coil was selected based on the fact that it would provide a
reaction time (30 s) which was much less than that required for the
reaction to reach completion. A 3.0-m coil was chosen to provide a
reaction time (90 s) longer than that required for the complete
reaction.

The effect of temperature on the peak area for the two different
coil lengths is depicted in Figure 35a. The trend for the shorter coil
is as expected; as the temperature is increased there is a rapid
increase in total absorbance created by an increased amount of reaction.
The data points obtained with the longer coil do not exhibit the ideal
straight-line behavior seen in the previous case studies, but the trend
is similar. The scatter of the points was probably a result of
integrating a noisier than normal signal.

Inspection of the effect of temperature on peak height
(Figure 35b) reveals an increase for both reaction times. The increase
with temperature is much more marked for the shorter reaction time.
This is reasonable since the dispersion of the sample plug undergoing a
shorter reaction time is affected both physically and chemically. In
the longer coil the effect of chemical reaction on plug dispersion
becomes a constant and the physical process predominates.

The random walk theory for FIA shows that an increase in
temperature increases the physical process of radial diffusion and
therefore enhances mixing. On a short time scale more complete reaction

is predicted (73, 106). The nickel/PAR experiments lend credence to

118

Pm AREA (mom-see)
I:
l A J
b
O

 

 

10“ tiriftsrrffvtttfrr'rrrtrrvrr'rrffl

20 no 40 so
TEMPERATURE (°C)

PEAK HEIGHT (W05)
9

 

 

a.“ TI‘V'YTTUrVrfYIUUrilrrrrt'TVT'IU‘U"
20 30 40 $0

TEMPERATURE (°C)

FIGURE 35 Effect of tewerature on peak area (a) and
peak height (b) for mediu- raaction with
1-m (a) and 3-m (A) coils.

119

these predictions. The chemical enhancement observed here may be due to
better mixing or an actual increase in the reaction rate, or to a
combination of these effects.

The theory also predicts that a temperature-induced reaction rate
enhancement will affect slower reactions more noticeably. The last case
presented was undertaken in an effort to substantiate this. An enzyme
systsm provides a classic example for temperature-induced rate effects
since a very small change in temperature can bring about a large change
in reaction rate. The rapid increase in product absorbance with

increased temperature noted earlier (Chapter 4) illustrates this effect.
Case 4: Slow Reaction

The FIA syst- for the glucose oxidase/Trinder reaction was used
as described earlier (Chapter 4). The operating conditions were the
standard set specified by the univariate experiments. Peroxide
standards were prepared by appropriate dilution of BOX hydrogen
peroxide. The stock solution was titrated with standardized thiosulfate
on the day of use (110). Two calibration curves—one for peroxide, the
other for glucose-were prepared in order to determine the peroxide
concentration that would give an absorbance equivalent to that obtain
with a 300 ppm glucose standard.

Data were taken at four different temperatures between 20°C and
40°C. Two glucose standards and their equivalent peroxide standards
were used. The peak heights obtained at each temperature are plotted
versus temperature in Figure 36b. The values obtained for the peroxide

standards at room temperature are a bit higher than those for the

§

5"
A L 14 l A L A LJ

PEAK AREA (ABSORBANCEOSEC)
“2’
L

 

120

 

 

 

 

0.” mrswrm'vvarfTvr‘
20 25 30 35 40
TEMPERATUREPC)

0.400-

PEAK HEIGHT (ABSORBANCE)

 

 

 

0.200 -
‘ _D
< U
0.100 - .
0.000 I . r . . I . . . . l . . . . l
20 2: 30 3: w
TEMPERATURE (°C)
FIGURE 36 Effect of taperature on peak area(a) and

peak height (b) for slow reaction.
Concentrations of standards were:
(a) 1.65 i4 and (s) 3.3 ‘1 glucose;
(0) 0.14 i4 and (A) 0.28 i4 820:.

first.“ we ‘0‘“ him-

121

glucose standards due to a slight mismatch of concentrations. This
results in a crossover at the second temperature. A relatively small
increase in absorbance takes place when peroxide is used compared to the
large increase for the glucose standards. The effect of temperature on
the peroxide part of the reaction is due to the physical factor and
possibly the reaction as well. By integration of the peaks, the cause
of the increase in peak height can be ascertained.

If it is the case that only the physical factor predominates, then
one would expect to see a horizontal line as in Cases 2 and 3. If the
chemical reaction is also responsible, an increase in peak area would be
expected. The plot of 'peak area versus temperature (Figure 36a)
suggests that the latter is the case, but only to a very small extent.
This was expected since earlier results indicated that the peroxidase
reaction is not drastically affected by changes in temperature. The
temperature effect on the peak area for the glucose standards is much
more severe. This is corroborated by the knowledge that the glucose
oxidase reaction is slower than the peroxidase reaction. I

The percent increase in response for each glucose standard over
that obtained for the respective peroxide standard was calculated. The
values can be found in Table 11. The percentage increase is linearly
related to the temperature. This relationship could be used to
"correct" for tuperature the response obtained from the enzyme FIA
system. In a "field" situation this would be a very useful technique
since the temperature may vary by a few degrees over a short period of
time. As illustrated here, that small temperature variation could

result in a rather large measurement error.

3 ‘ 4hr-“

five-n. A

122

TABLE 11 Temperature effect on glucose oxidase/
Trinder reaction.

 

 

Glucose X incr. in peak area
Standard Temperature due to Reaction
1.65 d! 27.0°C 35

32.1 54

37.0 72
3.3 mM 27.0 28

32.1 43

37.0 SI

123

In su-ary, this experimental study shows that sample dispersion
in flow injection analyzers decreases with increasing temperature and
substantiates the theoretical predictions of Betteridge et a] (73, 106).
This effect is especially important when high precision results are
required as, for example, in viscometry. Since the magnitude of the
effect can readily be quantified, the experimentalist is in a better
position to decide whether rigorous temperature control is necessary for
a particular flow injection analysis application.

For well-characterized reactions it may be possible to "correct" a
system for temperature changes as demonstrated in Case 4. Improved
repeatability without temperature control could be obtained.
Temperature sensitive devices could be incorporated into FIA manifolds
routinely. Results which are independent of temperature could then be
obtained by applying an appropriate multiplier to the peak height
obtained, or by carrying out, some transform on the peak shape. Such an
approach could be of considerable utility for process control since a
wider range of temperature conditions is more prevalent there than in
the laboratory. One specific case in which this might be a useful
technique is in food process control where an automated i-obilized

enzyme FIA system is used on-line.

CHAPTER 7

DEVELOPMENT OF A FLOW REVERSALS/MERGING ZONES FIA SYSTEM

It has been shown by Wade et a] (113) that the effective length of
the FIA manifold can be extended by reversing the flow direction of the
stream after its first pass through the detector. Several reversals
allow multidetection to be accomplished by this technique with only one
detector. The FIA profile is different after each reversal. The peak
shape becomes more Gaussian and the width at the base increases as the
sample plug is allowed more time to disperse. In the case where a
reaction occurs between the injected sample and the carrier stream, an
increased amount of product (for moderate or slow reactions) would be
detected after each reversal. The situation would become more
complicated if the product were degraded in a period of time that
corresponded to the timescale of the reversals. This is a rather simple
example in which a single concentration gradient is present.

In order to exploit the use of flow reversals fully, a more
complex case was considered. A complexation reaction which is sensitive
to pH was selected. By overlapping a concentration gradient with a pH
gradient, and than allowing the extent of overlap to increase with time,
a two dimensional chemical surface is generated. It was thought that
with a flow reversal experiment this surface could be mapped using the
information obtained from a single injection. Computer simulation

studies (114) indicated that this is a real possibility.

124

.0”th fl

125

The FIA F1 on Reversals System

The goal of this project was to assemble a semi-automated flow
injection analysis system which could be used to obtain this two
dimensional information for the complexation of metals with PAR. This
task was performed under the supervision of Adrian Wade, a postdoctoral
research associate in the Chemistry department. The FIA apparatus was
put together as shown in Figure 37. The same pump, light source, and
detector were used as in the studies presented in the previous chapters.
The two sample injection valves, V1 and V2, were manually operated six-
port sample injection valves with 70-u1 sample loops (Rheodyne). The
computer-controlled pneumatic actuator was attached to a third six—port
valve, V3. The connections on this were rearranged so that the f law
could be reversed by changing the position of the valve from ”fill" to
"inject" and vice versa. The sample passed through a detector into a
5-m holding coil. A slider valve (Alltech), V4, was used during the
reversal to bring a second 5-m holding coil into line. The coils were
0.8 mm i.d. Microline tubing. Most of the connections were made with
HPLC type threaded fittings joined with 0.8 mm i.d. Teflon tubing.

To ensure precise simultaneous injection of the two sample plugs,
81 and 82, the pump was stopped, thevalves switched to the "inject”
position, and the pump restarted. The two plugs were then merged at a
tee to create the overlap. This technique is coumnonly known as "merging
zones" (19). Asymmetric overlap was obtained by insertion of a short
piece of tubing, Cl, into one arm of the tee as shown in the figure.
The two merged gradients were then joined with the reagent stream, R, at

another tee. The flow of the stream was reversed by changing the

126

\usasogs

 

 

.33.“..me cocoa Samuel

30:

so»

gamma

<Hh

R museum

 

 

 

 

 

 

 

 

 

 

 

 

Nm
N40

.40
_m

 

 

 

HIM I]

127

position of V3. The holding coil was simultaneously brought into line
by manually switching the position of V4; this prevented loss of the
sample plug to waste upon the first reversal.

Computer control of the flow reversal was of paramount importance.
The time at which each reversal takes place must be known and totally
reproducible. This is because dye samples are used to determine the
shape of the sample plug each time it passes through the detector. For
a select chemical system it was presumed that there would be only
specific regions within the plug where product would be detected. The
absorbance would result from the convoluted effect of the concentration
and pH gradients. To ensure that the same plug shape was produced for
each set of reversals on each injection, computer control was necessary.
This facilitated accurate mapping of the surface.

The 8088-based microcomputer (Bentley-Model T) was prong as
mentioned previously (Chapter 4). The BASIC program, FIADATA, was
supplemented with a few extra lines of code in order to perform the
reversals. The reversal times were included in a DATA statement. On
each occasion that a set of A/D conversions was averaged, the current
time was calculated. This time was compared with the reversal time; if
it was equal to or greater than the reversal time, V3 was switched.
Data acquisition was the same as described earlier. A sampling rate of

Zszasused.

Calibration wi tn Dye

A solution of Alizarin Red (0.00048 M or 0.00025 M) in 0.05 M

phosphate buffer was prepared. Three experiments were performed. In

-._._.. as“;

128

all cases the reagent stream, R, and the two carrier streams, CA1 and
CA2, were 0.05 M phosphate buffer, and the overall flow rate was 1.0 ml
min”. The dye was first injected as 81 into CA1, then as 82 into CA2,
and then simultaneously into both carrier streams (i.e., 81 and 82 were
merged). In each case the absorbance of the stream flowing through the
detector was monitored continuously. This set of experiments was done
for both configurations of the system, i.e., both with and without the
delay coil in place. This technique was adopted because it was
previously shown that this type of characterization can be used to
predict the shape of a pH gradient in an FIA system (115). The flow was
reversed every 80 s in these experiments.

To investigate the location of experimental points on the two
dimensional surface referred to earlier, 82 absorbance is plotted
against 81 absorbance. The curves shown in Figure 38 were obtained for
the system without the delay coil. The peak maximum for 82 (Figure 38b)
occurs slightly later than that for 81 (Figure 38a). This indicated
that the overlap was not complete (i.e., there was some asymmetry
present). Figure 38c illustrates the small amount of I peak broadening
that occurs as a result of the asymmetry when the two samples are
injected simultaneously. If the overlap of the two plugs is complete
and the plugs are identical, a straight line of 45° slope is expected
when 82 absorbance is plotted versus 81 absorbance. However, due to
slight differences in sample volume and dispersion characteristics of
the plugs (a result of inhomogeneity in the apparatus) the straight
line becomes an ellipse as shown in Figure 39. This is because the

absorbance of SI initially rises faster than that of 82, then the $2

a. 4"» d... 1

 

129

 

 

 

 

 

 

 

 

04:00 1 A
mm-
3 u
mum -u
a.” jv 1 I V I
0 ms ms 300
M (000)
0.” G B
1
we -
2 4
CAN-I
1
a.“ [ fl T r T r f ‘r l
0 100 ms sum
M (sec)
0.“ q C
: .
J
4
0400;
a 1
0.13 d
0.000 f r r , fir r f u
0 um :00 :00
me (see)
FIGURE 38 FIA profiles for two reversals and one

detector. The three
(a) 81 only; (b)
(c) 81 plus 82.

injections were:
82 only; and

ABSORBANCE FOR 52

0.250

0.200

0.150

0.1”

0.050

 

0.000 0.050

FIGURE 39

130

0.100
ABSORBANCE FOR 51

Contour plot
one detector .

0.150

for

two

0.200

reversals

0.250

and

131

absorbance decreases slower than that of 81. If the sample volumes were
different, but the dispersion characteristics identical, a single curved
line would be expected.

By comparison of the FIA profiles shown in Figure 40 the effect of
the delay coil can be seen. The sample plug, 82, arrives at the
detector noticeably later than 81. This design allows 81 to experience
the chemical environment of CA1 completely befOre a gradient is imposed
by merging with $2 in CA2. For this set of experiments two detectors
were used in series; the first detector was separated from the second by
50 cm of Microline tubing. By the time the two plugs reach the second
detector it is evident that more dispersion has taken place.

A nuuflaer of interesting observations can be made about the 82
versus 81 plot in Figure 41. The asymmetric nature of the outermost
ellipse is due to the asymmetric nature of the peaks. The distribution
of the sample becomes more Gaussian after each reversal. This is
illustrated by the increased symmetry of the ellipse. With each
reversal. the ellipse contracts since the sample undergoes more
longitudinal dispersion and is better mixed with the carrier, which
makes the gradient less steep. As can be seen from the diagram, as this
lobe contracts a greater density of points defines the lobe. By
changing the data acquisition frequency, one could adjust the number of
points so that enough are taken to define the shape and yet not so many
that the information becomes redundant. Otherwise, a surface fitting
routine with continuously adjustable weighting factors may be required

in order to generate a good fit over the entire surface.

 

 

 

 

 

 

 

 

 

 

 

 

 

&-- ems!
1
1 A. u 8.
1 1
1 J
u q
1 4
1 1
1 1
1 1
tIflcl CAI-1
1 1
4 1
1 1
1 1
4 «a
1 1
1 1
1 1
J
a. + u rf r u m ' m I ' ' v—T ' me I
0 I. am ” 0 i. - fl
mm (ass) III (In-I
“'1 mg
‘ 1
(C. p D.
‘ u
‘ 1
1 1
1 1
i 1
‘ 4
C'.‘ .J.‘
‘ 1
‘ 1
i 1
‘ 1
" q
1 1
1
1
m T‘f'fl'Y'fi—l mrv.,',v,f',fifr1
M (sea) 1" (.3)
m 0.8.
T E. . F
1 1
1
1
m: m‘
1 1
1 1
4
0.1.4 £0”.
4 1
4 1
J 1
4 4
0.. ' * T m I I m f ' I '—"' r r ~ 1
o no a no 0 IO am as
mm (ass) M (at)
FIGURE 40 FIA profiles for three reversals and two

detectors. The three injections were:
(a) 81 only through the first detector;
(b) 81 only through the second detector;
(c) 82 only through the first detector;
(d) 82 only through the second detector;
(e) 81 plus 82 through the first detector;
(f) 81 plum 82 through the second
detector.

ABSORBANCE FOR 52

0.150

0.100

0.050

lIllLUILLJLIlllllllll'llLLlJLllJ
C. O

 

FIGURE 41

133

0.050 0.1 00

 

0.150

ABSORBANCE FOR 51

Contour plot for three reversals through

the first detector (0)
detector (A).

and the second

TIII'UT'V'rTIT'

I‘- 9 ‘I Q.

134

It is also apparent that without the second detector there would
be a large gap in information between that obtained in the initial pass
through the detector and that after the first reversal. This
illustrates how the use of a second detector can greatly contribute to
the characterization of the dispersion behavior as the samples pass
through longer lengths of tubing. The simulation showed that by use of
as many as eight detectors, excellent surface coverage would be expected
(114).

In Figures 39 and 41 the diagonal lines represent the contours of
the surface. These were generated for the two sets of data with a
fitting routine (based on the Kalman Filter) which was written by Pete
Wentzell (116). The results from the injection of $1 and 82
individually were used as the base response. The results from the
injection of $1 and $2 simultaneously were matched to this model. Since
dye was used it would be expected that the match would be exact and thus
generate diagonal straight lines. The diagonal nearest the origin
represents the lowest iso-response area. The subsequent diagonals
represent areas of increasing response which reflect increasing total
dye concentration. By matching the results obtained from a chemical
reaction within the two plugs to the non-reactive model of the two dye
plugs, one would expect to obtain curved contours which would describe

the two dimensional chemical surface.
Extension to a Cbemi cal System

The complexation of various metals with PAR was chosen as the

reaction to be implemented in this apparatus. The extent of

135

complexation for each metal is governed by pH. Metals such as copper
and/or vanadium would be interesting examples. Copper forms two
complexes with PAR, one begins to form at pH 2 while the other is formed
above pH 9. Vanadium forms a comlex between pH 6 and pH 8 but not
above pH 9. Injection of metal, 81, into a stream of citric acid (low
pH) and PAR followed by merging with 82, a plug of hydroxide (high pH),
should provide the ideal environment to map the reaction surface for the
two complexes.

Another consideration for the experiment with reaction is that PAR
also absorbs at the same wavelength (480 nm) as either of the two
complexes. Therefore it will be necessary to use the proper dilution of
the reagent stream in the reference channel of the calorimeter. When
PAR is mixed with hydroxide its absorbance increases. Therefore, a
baseline experiment will be necessary. The baseline experiment with
hydroxide must be subtracted from the other results. A number of static
experiments should also be done in order to correlate complex absorbance
and pH.

The apparatus constructed here is rather makeshift but has shown
in the dye experiments that it should be possible to obtain meaningful
results for the chemical syst- of interest. The technique of flow
reversals, in general, could be very useful in other areas as well. It
is a unique alternative to closed loop recycling (117) or multiple
detectors (118, 119). For the immobilized enzyme work presented in the
bulk of this work this methodology could be used to optimize reactor
lengths in an automated system and may also be useful in determination

of kinetic parameters .

CEAPTERB

WHEREDOHEGOFRMHERE?

There are number of unfinished projects which have surfaced as a
result of this research. The development of an on-line correction
method for ascorbic acid interference with the Trinder reaction would be
useful. Such a method could be implemented with other types of
reactors. This would be especially useful in clinical applications
since ascorbic acid is the only remaining interferent in serum after
deproteinization.

Increased interest in the past few years concerning
miniaturization of FIA systems indicates that there is a future for the
extension of immobilized enzymes to microconduits. Studies with smaller
tubing would provide the necessary information concerning lengths and
inner diameters of channels that could ‘be used with the pumps currently
available. Immobilization of the enzyme on smaller particles should not
be a problem since many workers have used controlled pore glass for many
years. 1

A longer term project involves the construction of a parallel
sugars analyzer. Glucose is one of six nutritionally-important sugars
(aldohexoses) (90). There are enzyme reactions for each of the sugars
which generate hydrogen peroxide and could therefore be coupled to the
Trinder reaction. Enzyme reactors to determine glucose, galactose,
maltose, sucrose, fructose, and lactose could each be incorporated into
one channel of a six channel systm. A possible configuration for this

system is shown in Figure 42. The sample would be injected into one of

136

137

Uhn<§
0.

.oeouushm nae
.eeoausa .eeaneu .oeofiel .eeouoeuem
.ueoosAm .«o awesomeness—o 300.335.:

mom loge cu..— «ozeuea mom-:30 Xmm Ne mun—0H.-

 

CUB-“0P0.

eczeeem eseaoeem
est-tam eEmwsu
3.00

     

 

   

 

“bud.

 

cohouaum 0 Ea... —

HOCLCUhZ-

    

lapzim

 

=on

  

    

u3<u¢hm

b2u0(u¢
sec

32.330

138

the six channels by a selector. A simpler detection system could be
used for each channel. The circuit designed by Betteridge et al (120)
could be used effectively. This system could be miniaturized by
constructing individual microconduits. A further enhancement would be
provided by the use of an optrode that is sensitive to hydrogen
peroxide. This could incorporate i-obilized peroxidase. The results
obtained with glucose and galactose indicate that development of a
parallel analyzer employing SBSRs is feasible.

There a few more general FIA studies that could result from the
insights gained here. As was noticed in the temperature study, the
noise in the baseline increases with temperature. This merits further
investigation since it could affect the reproducibility of results
obtained by FIA. The source of this "ripple" needs to be identified.

The flow reversals work is far from complete. The data obtained
from experiments with dye on any merging zones system could be used to
determine the nature of the overlap. The 32 versus 81 plots are much
more revealing than a simple look at the FIA profile. The speciation of
metals using the apparatus discussed in Chapter 7 needs to be undertaken
as well.

This research has opened new avenues for exploration. It is hoped
that the practical issues addressed in this work will aid others in the
pursuit of greater achievements in the use of immobilized enzymes and

flow inject ion analysis.

LIST OF REFERENCES

10.

11.

12.

13.

l4.

15.

LIST OF REFERENCES

0. Zaborsky, I-ohilized Enzymes, CRC Press: Cleveland, 1973.

M. Salmons, C. Saronio, S Garattini, Insolubilized fizymes, Raven
Press: New York, 1975

R.A. Messing, Ed. , I-obilized bzymes for Industrial Reactors,
Academic Press: New York, 1975.

11.11. Weetall, Ed., I-obilized hzymes, Antigens, Antibodies, and
Peptides, Marcel Dekker: New York, 1975.

L.B. Wingard, E. Katchalski-Katzir, and L. Goldstein, Eds.,
Applied Biochemistry and Bioengineering, Vol . l, Acadmic Press:
New York, 1976.

T.M.S. Chang, [mobilized figures and Proteins, Vols. 1 and 2,
Plenum Press: New York, 1977.

I. Chibata, Ed., I-obilized Enzymes: Research and Development,
John Wiley & Sons: New York, 1978.

R. Mosbach in Immobilized Enzymes (S.P. Kolowick and N.O. Kaplan,
Eds.) Methods in Enznology, Vol. 44, Academic Press: New York,
1978.

M.D. Trevin, Immobilized Enzymes, John Wiley & Sons: New York,
1980.

P.W. Carr and L.D. Bowers, I-obilized Enzymes in Analytical and
Clinical Chemistry, John Wiley & Sons: New York, 19%.

6.0. Guilbault, Analytical Uses of I-obilized Enzymes, Marcel
Dekker: New York, 1984.

L.D. Bowers and P.W. Carr, Anal. Chen. 1976, 48, 544A.

D.N. Gray, M.H. Keyes, and B. Watson, Anal. Chem. 1977, 4.9, 1067A.

M.M. Fishman, Anal. Chem. 1978, 50, 261R.

L.D. Bowers, P.W. Carr, Adv. Eiochem. Eng. 1980, 15, 89.

139

Hi.- '99-‘fisr-u

16.
17.
18.

19.

20.

21.

22.

23.

24.

25.

26.
27.
28.
29.

30.

31.
32.

33.

35.

36.

37.

140

M.M. Fistuan, Anal. Chem. 1980, 52, 185R.
L.D. Bowers, fiends Anal. men. 1%, 1, 191.
L.D. Bowers, Anal. Chem. 1%, 58, 513A.

J. Ruzicka and 8.11. Hansen, Flow Injection Analysis, John Wiley &
Sons: New York, lml.

M. Valcarcel and M.D. Luque de Castro, Flow Injection Analysis:
Principles and Applications, John Wiley 8: Sons: New York, 1&7.

J. Ruzicka and E.H. Hansen, Anal. Chim. Acta 1%, 179, 1.

Flow Injection Analysis Bibliography, Tecator AB: Hoganas,
Sweden, 1985.

J.M. Nelson and 8.0. Griffin, J. Am. Chem. Soc. 1916, 38, 1109.

F.I. Onuska, M.E. Codaa, T. Bistricki, and R.J. Wilkinson, J.
Chras. 1977, 142, 117.

M.C. Gosnell, R.E. Snelling, and R.A. Mottola, Anal. Chen. 1986,
58, 1585.

L. Skeggs, Am. J. Clin. Pathol. 1&7, 13, 451.

R.E. Stewart, G.R. Beecher, P.E. Hare, Fed. Proc. 1974, 33, 1439.
J. Ruzicka and E.H. Hansen, Anal. Chim. Acta 1975, 78, 145.

J. Ruzicka and 8.8. Hansen, Anal. Chim. Acta 1984, 161, 1.

J.M. Reign, W.E. van der Linden, and H. Poppe, Anal. Chim. Acta
1981, 123, 229.

0.0. Painton and H.A. Mottola, Anal. Chim. Acta 1984, 158, 67.
G.P. Hicks and S.J. Updike, Anal. Chen. 1%6, 38, 726.

S.J. Updike and G.P. Hicks, Nature (London) 1%7, 214, 986.
S.J. Updike and G.P. Hicks, Science 1%7, 158, 270.

P. Cr-onisi and R. Bovara, Biotech. Bioeng. 1976, 18, 1487.

G.J. Moody, G.S. Sanghera, and J.D.R. Thomas, Analyst 1987,
112, 65.

T. Yao, M. Sato, Y. Kobayashi and T. Wasa, Anal. Chim. Acta 1984,
165, 291.

141
38. B. Watson, D.N. Stifel, and P.E. Selersk'y, Anal. Chim. Acta 1979,
106, 233.
39. M. Mason and A. Townshend, Anal. Chim. Acta 135, 171, 185.

40. R. Appelqvist, G. Marko-Varga, L. Gorton, A. Torstensson, and G.
Johansson, Anal. Chin. Acta 1985, 16.9, 237.

41. B. Olsson, B. Stalbon, and G. Johansson, Anal. Chim. Acta 1986,
179, 203.

42. C.A. Koerner and T.A. Nienan, Anal. Chem. 19%, 58, 116.

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

44. J.N. Miller, B.F. Rocks, and D.T. Burns, Anal. Chim. Acta 1977,
93, 353.

45. R.A. Mottola, Anal. Chim. Acta 1983, 145, 27.

46. G. Johansson, L. Ogren, and B. Olsson, Anal. Chim. Acta 1983,
I45, 71.

47. M.H. Ho and M.U. Asouzu, Use of I-obilized bzymes in Flow
Injection Analysis, Ann. ”.7. Acad. Sci. 1984, 434, 526.

48. 1.. Holland, Me Properties of Class Surfaces, John Wiley 8: Sons:
New York, 1%4.

49. M. C. Gosnell and R.A. Mottola, Anal. Chem. 1986, 58, 631.
50. D. Keilin and E.F. Hartree, Biochen. J. 1952, 50, 331.
51. P. Trinder, Ann. Clin. Biochem. 1%9, 6, 24.

52. D. Barhan, and P. Trinder, Analyst 1972, 97, 142.

53. P.J. Worsfold, Anal. Chim. Acta 1983, 145, 117.

54. L. Gorton and L. Ogren, Anal. Chim. Acta 1981, 130, 45.

55. C.L.M. Stults, A.P. Wade, and S.R. Crouch, Anal. Chim. Acta 1987,
1.92, 155.

56. C.J. Patton and S.R. Crouch, Anal. Chim. Acta 1986, 179, 189.

57. YSI Precision Thermistors, Yellow Springs Instrunents, Yellow
Springs, Ohio, 1977.

58. B.H. Newcone and 0.6. Enke, Rev. Sci. Instm. 1984, 55, 2017.

59.

m.
61.
62.

63.

66.

67.

69.

70.

71.

72.

73.

74.

75.

76.

77.
78.

79.

' 80.

142

P.E. Kraus, Ph.D. in progress, Michigan State University.
H.T.S. Britton and R.A. Robinson, J. Chem. Soc. 1931, 1456.
M. Victor, Ph.D. in progress, Michigan State University.
B.W. Griffin and P.L. Ting, Biochemistry 1978, 17, 2206.

D. Keilin and E.F. Hartree, Biochem. J. 1948, 42, 221.

D. Betteridge, T.J. Sly, A.P. Wade, and J.E.W. Tillman, Anal.
Chen. 1983, 55, 1292.

S.N. Deming and 8.1.. Morgan, Anal. Chim. Acta 1-, 150, 183.
A.P. Wade, Ph.D. Dissertation, University of Wales, 1&5.

S.N. Daing and 8.1.. Morgan, Anal. Chen. 1974, 46, 1170.

D. Betteridge, A.P. Wade, and A.G. Howard, Talanta 1&5, 32, 723.
D. Betteridge, A.P. Wade, and A.G. Howard, hlanta 1%, 32, 709.

D. Betteridge and A.P. Wade, British Petroleun Research Centre,
Sunbury- on-Thales, UK, Personal commication, February 1986.

J.A. Lott and E. Turner, Clin. Chen. 1975, 2l, 1754.

D. Betteridge, A.P. Taylor, and A.P. Wade, Anal. Proc. 1&4,
21, 373.

D. Betteridge, C.Z. Marczewski, and A.P. Wade, Anal. Chim. Acta
1984, 165, 227.

H.U. Bergneyer, Ed. , Methods of bzymatic Analysis; Verlag Che-ie:

Weinheil, 1%5; p. 990.

K. Zeile in Die Methoden der Fermentforschung (E. Banann and K.
Myrback, Eds. ); reproduced by Academic Press: New York, 1941;
p. 2601.

B.C. Erickson, B.R. Kowalski, and J. Ruzicka, Anal. Chen. 1m7,
59, 1246.

B. Chance, J. Biol. Chen. 1943, 151, 553.
B.C. Thompson, Anal. Chen. 1987, 59, 1119.

A. Rios, M.D. Luque de Castro, and M. Valcarcel, Anal. Chim. Acta
1986, 179, 463.

C.C. Painton and R.A. Mottola, Anal. Chim. Acta 1984, 158, 67.

tit-12‘?- Q'W ”2|

81.

82.

83.

85.

86.
87 .

88.

89.

90.

91.

92.

93.

97.

98.

99.

143
M.A. Houpparis, 2nd Int ’1 Symp. on Kinetics in Analytical
Chemistry, Preveza, Greece, Septemer, 19%.

H. Hirano, Y. Baba, N. Yoza, and S. Ohashi, Anal. Chim. Acta 1%,
179, 209.

C.L.M. Stults, A.P. Wade, and S.R. Crouch, Anal. Chen, accepted
for publication.

W. Pimn and H.S. Isbell, Advan. Carbohyd. Chem. 1%8, 23, 11.

H.S. Isbell and W. Pigman, Advan. Carbohyd. Chem. Biochem. 1%9,
24, 14.

0.8. Hudson and J.K. Dale, J. Am. Chen. Soc. 1917, 39, 320.
J.A. Nelder and R. Mead, Comput. J. 1%5, 7, 308.

G. Livingston, F. Franks, and L.J. Aspinall, J. Solution Chem.
1977, 6, 203.

H. Murschhauser, Biochem. Z. 1920, 110, 181.

G.P. Stewart and M.A. Amerine, Eds., Introduction to Food Science
and Technology, and Ed.; Acad-ic Press, NY, 1982,. p. 137.

Faulkner, W.R., Ed. in Chief Selected Methods in Clinical
Chesistry, 1982, 9, 379.

Galactose Oxidase Product information Sheet: Sigma Chemical Co. ,
St. Louis, W, 1987.

H.U. Bergmeyer, Methods of Enzymatic Analysis, Volume 2, Verlag
Chsmie: Weinheim, 1983.

S.R. Dahodwala, M.K. Weibel, and A.E. Humphrey, Biotech. Bioeng.
1976, 18, 1679.

B. Olsson, H. Lundback, and G. Johansson, Anal. Chim. Acta 1985,
167, 123.

D. Betteridge, Talanta 1976, 23, 409.

V. Ananthakrishnan, W.N. Gill, and A.J. Barduhn, A. I. Ch. E. J.
1%5, 11, 1063.

J.M. Reijn, W.E. van der Linden, and H. Poppe, Anal. Chim. Acta
1981, 126, l.

J.T. Vanderslice, K.K. Stewart, K.K., A.G. Rosenfeld, and D.J.
Higgs, Talanta 1981, 28, 11.

100 .

101.
102.

103.

104.
105.

106 .

107.

108.

109.

110.

111.

112.

113.

114.

115.
116.P.

117.

118.

119.

120 .

144
H. Wada, S. Hiroaka, A. Yuchi, and G. Nakagawa, Anal. Chim. Acta
1%, 179, 181.
J.F. Tyson, Anal. Chim. Acta 1986, 179, 131.
J. Ruzicka and E.H. Hansen, Anal. Chim. Acta 1M), 114, 19.

D. Betteridge, W.C. Cheng, E.L. Dagless, P. David, and T.B. Goad,
Analyst 1983, 108, 17.

C.C. Painton and H.A. Mottola, Anal. Chem. 1981, 53, 1713.
D.C. Stone and J.F. Tyson, Anal. Proc. (London) 19%, 23, 23.

C.D. Crowe, H.W. Levin, D. Betteridge, and A.P. Wade, Anal. Chim.
Acta, in press.

D.C. Stone and J.F. Tyson, Analyst 1987, 112, 515.

C.L.M. Stults, A.P. Wade, and S.R. Crouch, Anal. Chin. Acta 1987,
192, 301.

A.P. Wade, unpublished results, 1983.

I.M. Kolthoff, E.B. Sandell, E.J. Meehan, and S. Bruckenstein,
Quantitative Chemical Analysis, 4th Ed. ,' Macmillan: NY, 1%9.

P.M. Beckwith, Ph.D. Dissertation, Michigan State
University, 1972.

R.G. Harfmann, Ph.D. in progress, Michigan State University.

D. Betteridge, P.E. Oates, and A.P. Wade, Anal. Chem. 1987,
59, 1236.

A.P. Wade, C.L.M. Stults, S.R. Crouch, and D. Betteridge, 13th
Annual EACSS Meeting, St. Louis, W), October 2, 1986.

D. Betteridge and B. Fields, Anal. Chim. Acta lml, 132, 139.
D. Wentzell, Ph.D. in progress, Michigan State University.

J. Ruz, A. Rois, M.D. Luque de Castro, and M. Valcarcel, Jalanta
1986, 33, 199.

D. Betteridge, W.C. Cheng, E.L. Dagless, P. David, and T.B. Goad,
Analyst 1983, 108, l.

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

T.J. Sly, D. Betteridge, D. Webberley, and D.C. Porter, J.
Automated Chem. 1982, 4, 186.

"’TI’LfililfljflifliflmfliLWififliflT