.
12.3

”fa:

 

£4”.
\

3 fit.
.1 $9.31.
i 5...}. . .

 

.an
.13 “$1.5

.iII

 

 

0.! :5

 

b.
.w. 9 5.
MW? ...
:3. “a .
4. ‘5‘:
5me .
22....

Jr

Rm.
.5?

x

3

’x .-. ¢ Z‘,
f t 7
fishnfisiv

14
#‘z

93153!

II...

.3115: 1.. n!

1‘ .I .0... :01?
-04: 1)..

1N”...l’

.. :5.

at
1 as!
villi?
it v IzaIJ‘ I
r . DER... 5.
1... 3|. .
5... (.1...th
3:511:3«8...
1’ ('11 ‘
[1.3. {1“

.2!

:5 :vv)! . “T
gimp?)
Psi..." . . J"

 

L... 3.5...
‘ Mm!

“15543

lllllllllllllllllllllllHllllllllllllllllilllllllllHlllllllli

3 1293 0155504

This is to certify that the

dissertation entitled

A Study of the Applicability of Fluorescence
Spectroscopy in the Development of A
Supersaturation Sensor and in Measuring Kinetic
Parameters in Batch Crystallization Processes

presented by

Sanjay K. Yedur

has been accepted towards fulfillment
of the requirements for

Ph . D . degree in Wineering l

 

9%on C

Major professor

DateJN’ 5;) {779

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

 

LIBRARY

University

 

 

 

PLACE II RETURN BOX to remove this checkout from your record.
To AVOID FINES rotum on or bdoto date duo.

DATE DUE DATE DUE DATE DUE

    

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MSU lo An Afflmotlvo Action/Equal Oppommlty lnotltwon
W

 

A STUDY OF THE APPLICABILITY OF FLUORESCENCE
SPECTROSCOPY IN THE DEVELOPMENT OF A SUPERSATURATION
SENSOR AND IN MEASURING KINETIC PARAMETERS IN BATCH
CRYSTALLIZATION PROCESSES

By

Sanjay Krishnamurthy Yedur

A DISSERTATION

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

DOCTOR OF PHILOSOPHY

Department of Chemical Engineering

1996

ABSTRACT

A STUDY OF THE APPLICABILITY OF FLUORESCENCE
SPECTROSCOPY IN THE DEVELOPMENT OF A SUPERSATURATION
SENSOR AND IN MEASURING KINETIC PARAMETERS IN BATCH
CRYSTALLIZATION PROCESSES

By
Sanjay Krishnamurthy Yedur

The main parameters of interest in a crystallization process are the crystal size
distribution (CSD) and the purity of the product. These parameters depend upon the
supersaturation in the crystallizer. A novel technique is presented that uses fluorescence
spectroscopy to measure the supersaturation in crystallizing solutions. In its most
fundamental terms, supersaturation is dependent on the activity of the solute in solution
which is a strong function of the solution structure. Fluorescence spectroscopy depends on
the solution structure and provides a more accurate measurement of supersaturation than

current techniques that assume ideality of the crystallizing solution.

The fluorescent properties of a probe, pyranine, were used to provide
supersaturation measurements in aqueous citric acid solutions. The solvatochromic nature
of the emission of pyranine “provides an excellent calibration curve for supersaturation
measurements. While pyranine performs very satisfactorily, its use in food and
pharmaceutical processes is limited since it is not listed as food grade. For such
applications, all dyes listed as food grade were screened for their potential use. Three of
the seven, brilliant blue, fast green, and allura red were found suitable for use.
Fluorescence spectra of these dyes in sucrose solutions are presented and calibration curves

drawn to show their ability to measure supersaturation.

For use as an in situ sensor, it is desirable to immobilize the probes onto a surface
so that an immersible sensor can be constructed. Brilliant blue is immobilized on anionic
cellulose acetate membranes for this purpose. However, the chemistry involved in the
immobilization causes some loss of sensitivity of the probes that potentially reduces their

effectiveness.

The desupersaturation curves of crystallizing solutions yield information on the
kinetic parameters involved during the nucleation and growth of the crystals. The
fluorescence of pyranine was used to generate such curves for citric acid during seeded
isothermal desupersaturation experiments. A model is described that can be used to fit

these curves and provide the kinetic constants of nucleation and growth.

Future research objectives involve the development of an apparatus that can contact
the probe with the crystallizing solution in an ex situ manner, to alleviate any concerns of
contamination of the crystals by the probe. Some basic experiments along with other

recommendations for further investigations are provided.

To my family, for their support through the years

iv

Acknowledgments

First and foremost, thanks are due to my mentor, Dr. Kris Berglund, for the
encouragement, support and guidance he has provided to me over the last few years.
My association with him provided me not only with a greater appreciation of engineering
research, but also gave me valuable insight into the workings of a university.

Thanks are also due to the members of my dissertation committee, Drs. Daina
Briedis, Alec Scranton, and Gary Blanchard for their time and the interest that they took
in my project.

Special thanks are due to my colleagues and friends Laurie Ruiz, Dilum
Dunuwila, Everson Miranda, Marketta Uusi-Penttila and numerous others present in
Kris’s lab at different times during my tenure. Out of the many, I would like to mention
some special people. Laurie Ruiz, always a great friend, provided a smooth way out of
all the red tape that] had to go through in my years here. A pleasant chat with her
would help put all the pitfalls I encountered while conducting research into proper
perspective. Dilum, my colleague and friend since I came here to graduate school, has
always been the person I call with any problem in my research, and he has never
disappointed me! And finally Everson, now a faculty member in Brazil, whose most
notable achievement, in my opinion, was to get me involved in running!

A special word about my cousin Anupama, presently at UC, San Diego. Over
the years, she has provided me with many unforgettable anecdotes from her own chaotic
life and that of those around her. Her frequent misadventures, which invariably resulted
in a frantic phone call for support and advice, provided me with much needed respites

from the tedium of graduate school. A self confessed drama queen, she constantly

courts non-traditional situations that get her into trouble and results in a great source of
entertainment for me. Anu - thanks for everything.

My friends Sam Larson and Jim Dearing deserve special mention too. Sam was
the first person thatI interacted with here at Michigan State University. She helped me
tremendously whileI was suffering through a culture shock and has been a great friend
all these years. In their household, 1 am treated as a member of the family, and l
tremendously appreciate their friendship.

Many other people influenced me in different ways. My good friends Arvind
Mathur, Himanshu Asthana, and Sanjay Padaki deserve special thanks for their help and
support wheneverl needed them. James ‘Moe’ Benda introduced me to, among other
things, the beautiful Upper Peninsula of Michigan. In my first term here, we spent
many a philosophical hour over coffee while ostensibly at work! Rik ter Veen and
Marketta Uusi-Penttila are complewa responsible for getting me interested in the game
of bridge. Shalini Mathur and Savi Dunuwila provided me with sound non-technical
advice (since they were smart enough never to be afflicted with an engineering
education) wheneverl needed it, and I am most grateful to them for that. Astrid
Baviere, Raul Tomas, Takayo Sugimoto, and Grigoris Papakostas are among the special
people I met during the Foreign Teaching Assistant Program, and remain good friends.

To all these people -a big thank you from the bottom of my heart

vi

Table of Contents

List of Tables .......................................................................................... x
List of Figures ........................................................................................ xi
Chapter 1
INTRODUCTION ................................................................................. l
1.1 Background ................................................................................. 1
1.2 The role of supersaturation in crystallization ............................................ 2
1.3 Fluorescence spectroscopy ................................................................ 9
1.4 Overview of this work ................................................................... 13
1.5 References ................................................................................ 17
Chapter 2
USE OF FLUORESCENCE SPECTROSCOPY IN CONCENTRATION AND
SUPERSATURATION MEASUREMENTS IN CITRIC ACID SOLUTIONS ......... 19
2.1 Background ............................................................................... 19
2.2 Materials and Methods ................................................................... 22
2.3 Results and Discussion ................................................................. .25
2.4 Conclusions ............................................................................... 34
2.5 References ................................................................................ 36
Chapter 3

A NOVEL ANALYTICAL USE OF FOOD GRADE COLORS: SPECTROSCOPIC
MEASUREMENT OF CONCENTRATION AND SUPERSATURATION IN

SUCROSE SOLUTIONS ...................................................................... 38
3.1 Background ............................................................................... 38
3.2 Materials and Methods ................................................................... 42

3.2.1 Absorption experiments ............................................................ 43
3.2.2 Fluorescence experiments ......................................................... 43
3.2.3 Immobilization of the probes ...................................................... 43

vii

3.3 Results and Discussion ................................................................. .44

3.3.1 Results with the ‘free’ dye in solution ........................................... 44
3.3.2 Results with the immobilized dyes ............................................... 54
3.4 Conclusions ............................................................................... 58
3.5 References ................................................................................ 58
Chapter 4
A SOL-GEL BASED FLUORESCENCE SENSOR FOR CONCENTRATION
MEASUREMENTS IN SUCROSE SOLUTIONS .......................................... 60
4.1 Background ............................................................................... 60
4.2 Materials and Methods ................................................................... 61
4.2.1 Film preparation .................................................................... 63
4.2.2 Entrapment of the chlorin ......................................................... 63
4.3 Results and Discussion ................................................................. .64
4.3.1 Sol-gel films ......................................................................... 64
4.3.2 Free chlorin in aqueous solutions ................................................. 64
4.3.3 Entrapment of the chlorin ......................................................... 64
4.4 References ................................................................................ 70
Chapter 5
ON THE APPLICATION OF FLUORESCENCE SPECTROSCOPY TO MEASURE
THE DESUPERSATURATION OF CITRIC ACID SOLUTIONS ....................... 72
5.1 Background ............................................................................... 72
5.2 Materials and Methods ................................................................... 73
5.3 Results and Discussion ................................................................. .75
5.3.1 The desupersaturation curve ....................................................... 78
5.3.2 Effect ofseed weight ............. 82
5.3.3 Effect of temperature ............................................................... 85
5.3.4 Effect of probe concentration ...................................................... 87

viii

5.3.5 Measurement of kinetic parameters ............................................... 88

5.4 Conclusions ............................................................................... 90
5.5 References ................................................................................ 92
Chapter 6
MODELS FOR THE EXTRACTION OF KINETIC PARAMETERS FROM THE
DESUPERSATURATION CURVES ......................................................... 93
6.1 Background ............................................................................... 93
6.2 Development of the model .............................................................. 96
6.3 Conclusions .............................................................................. 100
6.4 References ............................................................................... 101
Chapter 7
CONTINUING INVESTIGATIONS AND RECOMMENDATIONS FOR FUTURE
WORK ........................................................................................... 102
7.1 Background .............................................................................. 102
7.2 Continuing Investigations. . . . . .. ....................................................... 103
7.2.1 Estimation of kinetic parameters ................................................. 103
7.2.2 Ex situ monitoring of concentration ............................................. 103
7.3 Recommendations for future work .................................................... 107
7.3.1 ATR Fluorescence ................................................................ 11 1
7.4 Conclusions .............................................................................. 1 12
7.5 References ............................................................................... 1 12
Chapter 8
CONCLUSIONS ............................................................................... 114
APPENDIX ........................................................................................ 116

ix

List of Tables

Table 3. 1 Selected physical properties of FD&C colors ................................ 41
Table A1 Raw data for Figure 2.4 ...................................................... 116
Table A2 Raw data for Figure 2.5 ...................................................... 116
Table A3 Raw data for Figure 2.6 ...................................................... 1 17
Table A4 Raw data for Figure 2.7 ...................................................... 117
Table A5 Raw data for Figure 2. 8 ...................................................... 118
Table A6 Raw data for Figure 3 .6 ...................................................... 1 18
Table A7 Raw data for Figure 3.7 ...................................................... 119
Table A8 Raw data for Figure 3.9 ...................................................... 1 19
Table A9 Raw data for Figure 4. 4 ...................................................... 120
Table A10 Raw data for Figure 5.2 ...................................................... 121
TableAll Raw data for Figure 5.3 ...................................................... 123
Table A 12 Raw data for Figure 5.4 ...................................................... 125
Table A 13 Raw data for Figure 5. 5 ...................................................... 129
Table A 14 Raw data for Figure 5.6 ...................................................... 133

List of Figures

Figure 1.1 Information feedback diagram illustrating the importance of supersaturation and
the interaction between size distribution and kinetics resulting from the
constraint of the mass balance. ........................................................ 4

Figure 1.2 Schematic energy level diagram showing the potential energy wells of the
ground state and the first electronic state for a diatomic molecule. .............. 1 1

Figure 1.3 Schematic Energy Level (Jablonski) diagram for a diatomic molecule .......... 12

Figure 2.1 Molecular structure of Pyranine (trisodium salt of 8-hydroxy-1,3,6—
pyrenetrisulfonic acid) ................................................................ 21

Figure 2.2 Excited state kinetics of pyranine .................................................... 23

Figure 2.3. Emission spectra of pyranine in citric acid solutions of 10, 30, 50, and 70
weight percents; taken at 25 °C. ..................................................... 27

Figure 2.4 Peak Intensity Ratio (ratio of emission peak intensities at 440 nm and 510 nm)
as a function of citric acid concentration ............................................ 29

Figure 2.5 Peak Intensity Ratio (ratio of peak intensities at 440 nm and 510 nm) versus the
supersaturation S (defined as S=(c-c*)/c*, where c and c* are the actual

concentration and the saturation concentration respectively). .................... 30
Figure 2.6 pH of aqueous citric acid solutions as a function of citric acid concentration. . 32

Figure 2.7 Comparison of PIR's (ratio of peak intensities at 440 nm and 510 nm) obtained
from pyranine emission in citric acid solutions as a function of pH of the
solutions to the PIR's obtained from pyranine emission in HCl-NaOH solutions
of varying pH's. Closed circles: Citric acid solutions; open circles: HCl-NaOH

solutions ................................................................................ 33

Figure 2.8 Corrected PIR (corrected for pH effects) versus citric acid concentration (wt%).
The arrow indicates the solubility at 25 °C. ........................................ 35

Figure 3.1 Molecular structures of the seven FD&C colors. .................................. 40

Figure 3.2 Schematic diagram of the apparatus used to measure the fluorescence from the
immobilized sensor that is immersed in a crystallizer. 1) Source. 2) Collimating
lens. 3) Monochromators. 4) Jacketed crystallizer. 5) Immobilized probe
assembly. 6) Bifurcated fiber optic. 7) Stirrer. 8) PMT Detector. 9) Computer.45

Figure 3.3 Absorption spectra of the FD&C colors. The concentrations of each in aqueous
solution are: FD&C Blue No. 1 and FD&C Green No. 3: 5 ppm; FD&C Blue
No. 2 and FD&C Red No. 40: 10 ppm; FD&C Blue No. 2, FD&C Red No. 3,
FD&C Yellow No. 5, and FD&C Yellow No. 6: 20 ppm. ...................... 47

Figure 3.4 Emission spectra of FD&C Red No. 40 in sucrose solutions of different
concentrations (expressed as wt%). All spectra recorded at room temperature
(25 °C). Excitation wavelength: 500 nm ........................................... 49

Figure 3.5 Emission spectra of FD&C Green No. 3 in sucrose solutions of different
concentrations (expressed as wt%). All spectra recorded at room temperature
(25 °C). Excitation wavelength: 420 nm ........................................... 50

Figure 3.6 Calibration curve showing the relation between the Peak Intensity Ratio (defined
as the ratio of the peak intensities at 662 nm to that at 515 nm) and sucrose
concentration for FD&C Green No. 3. ............................................. 52

Figure 3.7 Calibration curve showing the relation between the Peak Intensity Ratio (defined
as the ratio of the peak intensities at 587 nm in each sample to the peak intensity
at the same wavelength in 0% sucrose solution) and sucrose concentration for
FD&C Red No. 40. Inset shows the parameters for an exponential fit to the
curve. ................................................................................... 53

Figure 3.8 Emission spectra of FD&C Blue No.1 in sucrose solutions of different
concentrations (expressed as wt%). The dye molecules are immobilized onto an
anion exchange membrane. All spectra recorded at room temperature.
Excitation wavelength: 400 nm. ..................................................... 56

Figure 3.9 Calibration curve showing the relation between the Peak Intensity Ratio (defined
as the ratio of the peak intensifies at 664 nm to that at 615 nm) and sucrose
concentration for the immobilized FD&C Blue No. 1. ............................ 57

Figure 4.1 Chemical structure of Octaethyl chlorin diol ........................................ 62

xii

Figure 4.2 Emission spectrum of free chlorin in solution in different sucrose solutions. All
spectra collected at room temperature. .............................................. 65

Figure 4.3 Changes in emission spectra of chlorin embedded in a titanium carboxylate thin
film in various aqueous sucrose solutions; all spectra collected at room
temperature ............................................................................. 67

Figure 4.4 Calibration curve showing a linear relationship between the Peak Intensity Ratio
(defined as the ratio of the intensities of the peaks at 646 nm and 673 nm) and
sucrose concentration. ................................................................ 69

Figure 5.1 Emission spectra of pyranine over the course of a seeded isothermal (25 °C)
citric acid batch crystallization. Ten spectra are shown here, each at an interval
of 10 minutes from t = 0 minute to t=90 minutes. These ten are culled from the
ninety spectra collected at intervals of 1 minute .................................... 77

Figure 5.2 Typical desupersaturation curve of a seeded citric acid batch crystallization
experiments. The experimental conditions were as follows: Temperature = 25
°C, initial supersaturation (c/c*) = 1.2, agitator speed = 400 rpm, probe
(pyranine) concentration = 1.0e-05 M, seeds (average size 660 microns) = 3
gms, excitation wavelength = 365 nm. ............................................. 79

Figure 5.3 Backscattered excitation light during the desupersaturation experiment as a
function of time. The experimental conditions are exactly the same as those
described in Figure 5.2. .............................................................. 81

Figure 5.4 Effect of seed amount on the desupersaturation curve of seeded citric acid batch
crystallization experiments. The ordinate are normalized so that all curves start
at the same point, since initial supersaturation (c/c* = 1.2), is the same in all
three cases. Other experimental conditions: Temperature = 25 °C, agitator speed
= 400 rpm, probe (pyranine) concentration 2 1.0e-05 M, seeds (average size
660 microns) = 3 grns, excitation wavelength = 365 nm ......................... 83

Figure 5.5 Effect of temperature on the desupersaturation curve of seeded citric acid batch
crystallization experiments. The ordinate are normalized so that all curves start
at the same point, since initial supersaturation (c/c* = 1.2), is the same in all
three cases. Other experimental conditions: agitator speed = 400 rpm, probe
(pyranine) concentration = 1.0e-05 M, seeds (average size 660 microns) = 3
gms, excitation wavelength = 365 nm. ............................................. 86

xiii

Figure 5.6 Effect of varying the probe (pyranine) concentration on the desupersaturation
curve of seeded citric acid batch crystallization experiments. The ordinate
shows the relative supersaturation {(c/c*)-1}that is the same in all three cases.
Other experimental conditions: temperature = 25 °C, agitator speed = 400 rpm,
seeds (average size 660 microns) = 3 gms, excitation wavelength = 365 nm.. 89

Figure 5.7 Control policy projected to maximize the mean size of a crystal size distribution
from a cooling batch crystallization process [7]. The policy is based on
maintaining a constant supersaturation during the course of the crystallization.
The supersaturation is defined by either of the following two equations:

Supersaturation = APIR = PIR - PIR*

APIR

Su rsaturation = o =
pe PIR*

 

where PIR is the ratio of emission peaks of a probe such as pyranine in a
supersaturated solution, and PIR* is the same ratio in a saturated solution. .. .91

Figure 7.1 Schematic of a conventional Flow Injection System. C = carrier stream, R =
reagent stream, S 2 sample, D = detector, P = pump, and W = waste stream. 104

Figure 7.2 Schematic of a reverse Flow Injection System. C = carrier stream, R = reagent
stream, S = sample, D = detector, P = pump, and W = waste stream. ........ 104

Figure 7 .3 Schematic of the apparatus used to measure the concentration of a crystallizing
solution ex situ. The top view of the flow cell is shown on the upper left hand
corner. ................................................................................. 106

Figure 7.4 Typical desupersaturation curve of a seeded citric acid batch crystallization
experiment using the apparatus shown in Figure 7.3. The experimental
conditions were as follows: Temperature = 25 °C, initial supersaturation (c/c*)
= 1.2, agitator speed 2 400 rpm, probe (pyranine) concentration = 1.0e-05 M,
seeds (average size 660 microns) = 3 gms, excitation wavelength = 365 nm.
The ordinate shows the Peak Intensity Ratio, defined as the ratio of the
intensities of the emission peaks of pyranine at 449 nm and 520 nm. ......... 108

xiv

Chapter 1

 

INTRODUCTION

1 . 1 Background

Crystallization from solution is one of the most widely used unit operations in the
chemical process industry. It ranks second behind distillation in use as a separation
process. Almost all processes that require a final purification step involve a crystallization
operation. While crystallization is quite common in many food and fine chemical
industries, it is especially important in the pharmaceutical industry, where almost all
products are crystallized. Purity of the product is the primary concern in the manufacture
of drugs. Besides separation and purification, the production of a specified crystal size

distribution is another important consideration of any crystallization operation.

To obtain the desired product purity and the specified crystal size distribution,
careful control of the whole process is essential. However, the current control schemes
used in industrial crystallizers suffer from numerous drawbacks. Control of batch
industrial crystallization process is generally done manually by operators. As with most
manually controlled processes, the success of a given batch crystallization relies on the skill
of a single individual. To overcome this drawback, there is a need to implement control
strategies that provide a quantitative output that can be either used by an operator without
any risk of error, or implemented in an automatic control scheme. A primary impetus for

improved control comes from the pharmaceutical industry wherein nearly all products are

crystallized at some point in their production. Pharmaceutical processing is usually done in
a batch mode, which is often more difficult to control than continuous processes due to the
relatively short time constants present in batch processes. Another consideration is batch to
batch fluctuations, which can cause considerable variation in the crystallization process
resulting in final product divergence. Such variation is simply unacceptable to the Food
and Drug Administration. Reworking a batch that does not meet specifications incurs

additional costs and provides an opportunity for additional contamination.

The need for control of large scale, continuous crystallizers is also a problem for the
food and chemical industries. Typical applications of crystallization include the production
of sugars (sucrose, glucose, lactose, and fructose), carboxylic acids (citric and itaconic),
and amino acids (glycine and lysine). As in pharmaceutical crystallization, often there is no
attempt at automatic control of the crystallization process. All the various products listed
and many commodity chemicals are crystallized from aqueous solution at high
concentrations and generally very high process throughput. In contrast to pharmaceutical
processing, the longer time constants of large continuous crystallization processes require
careful control of supersaturation since process upsets cause departure from steady state

that may take a long time to re-establish. During the departure from steady state the process

is probably making an off specification product that is unacceptable.

Therefore, both batch and continuous crystallization processes require the precise

measm'ement and control of the process, although for quite different reasons.

1.2 The role of supersaturation in crystallization

The most important parameter that defines the crystallization process is
supersaturation. Supersaturation is defined as the amount of solute in excess of the
equilibrium solubility that is present in solution at a given temperature and is the driving
force for the crystallization. The process of crystallization can be divided into two parts:

nucleation and growth. Nucleation refers to the birth of nuclei in solution and growth is the

 

subsequent deposition of solid mass onto the nuclei. Supersaturation is the driving force
for both crystal nucleation and growth and as such controls the overall rate of

crystallization.

Figure 1.1 shows the role of supersaturation on a continuous crystallization process
by means of an information feedback diagram. A similar diagram with different balance
equations can be drawn for a batch crystallization process. Supersaturation affects both the
nucleation rate and the growth rate. A high supersaturation results in a larger nucleation
rate that cauSes a large number of nuclei to be formed. On the other hand, a lower
supersaturation favors a higher growth rate that results in fewer crystals that grow to a
larger size. As shown in the illustration, the final crystal size distribution also determines
the surface area of the crystals in the crystallizer that, in turn, affects the supersaturation.
Secondary nucleation, also affected by the CSD, is responsible for additional mass to come
out of solution. The specific equations describing the various processes are described in
detail in Chapter 6. The role played by supersaturation is therefore paramount in the design
and operation of the crystallizer. Thus, to develop a control strategy for any crystallizer,
either batch or continuous, an accurate measurement of supersaturation is a primary

requirement.

In spite of its importance, an accurate in siru measurement of supersaturation is still
cited as one of the most important needs of industry [1]. A number of analytical techniques
have been proposed and are in use for the measurement of solubility and supersaturation
[2,3]. These cover a broad range of techniques from simple residual weight determinations
to radioactive tracer methods. Conventional techniques that are currently used include
refractometry, interferometry, viscometry, and density calibrations. The primary difficulty
in using any of these is the procurement of a sample that accurately reflects the dynamic
conditions in the crystallizer. Generally, the separation of the crystals from the liquid phase
is a prerequisite for these techniques. While filtration and decantation are commonly used

for the separation, it is practically impossible to preserve the actual conditions that were

 

Growth

"l G(s) = kg (AC)i I'—

 

 

 

 

 

 

Supersaturation
Ac ‘ I CSD
S = 0 -L Gr ~——>
—‘l:l‘ . _ n n em I >
Ar

Nucleation

497“) = kn (Aer Mr }—
A

MT=kv°IonL3dL |<————l‘- ———————

 

 

 

 

 

 

 

 

 

 

__| AT=kalZnL2 dL J. D

Figure 1.1 Information feedback diagram illustrating the importance of supersaturation
and the interaction between size distribution and kinetics resulting from the constraint of the

mass balance.

prevailing in the crystallizer when the sample was withdrawn. Refractometry is one of the
more common techniques and involves the measurement of small differences in the
refractive indices of the solution with a change in the solute concentration. Invariably,
these differences appear in the fourth or fifth decimal place that makes the sensitivity and

reliability of this technique suspect.

Apart from these procedural complications, there is a fundamental problem with all
conventional techniques that makes the measurement of supersaturation at best an

approximation.

The fundamental driving force for crystallization is the difference between the
chemical potential of the crystallizing substance in the transferring and the transferred state
[4]. For an anhydrous solute crystallizing from a binary solution, the driving force may be

writtenas
Au=u2-ur 0-1)

where pig and M are the chemical pctentials in the crystal and in the solution respectively.
This quantity being negative in a crystallizing system, it is convenient to describe it in terms _
of an 'affinity of reaction' to make the driving force a positive quantity [4]. The 'affinity of

reaction' is defined as
=-Al1=llr'l12 (12)
The chemical porential u is defined as
u=uO-RTlna (1.3)
where 110 is the standard potential and 'a' is the solute activity.

The supersaturation S can then be expressed [5] in dimensionless terms by

s :11? = __“I ' ”2 ‘ (1.4)

RT

 

This equation can be further simplified to

S=vln [fl] (1.5)

32

=vln [EL] (1.6)

Yzcz

where the 1's are the activity coefficients, c's are a measure of the concentration, and 'v' is

the number of ions in a molecular unit.

However, due to the inherent difficulty in measuring the activities and the relative
lack of literature data on activities, three very distinct and questionable assumptions are

mMe:

a) the activity coefficients in the solution and in the crystal are assumed to be
independent of concentration, i.e., y at f(c), so that the ratios of the activity coefficients are

unity.

b) the number of ions 'v' is taken to be unity. These approximations~ reduce

equation (1.6) to

32

S=ln [fi]=1n (s, + 1) (1.7)

 

where Sc = [6‘ - CZ) (1.8)
C:

c) Sc is taken to be much less than unity in which case equation (1.7) reduces to

 

S=Sc= (0' ’ c2] (19)

 

This definition of supersaturation (equation 1.9) is the most widely used expression
because of the relative ease of the measurement of concentration as compared to the
activity.

In these previous equations, the concentrations can be expressed in any units,
although the preferred one would be that for which the activity coefficient varies the least

over the range of concentrations of interest [5].

As mentioned before, there is considerable question over the validity of each of the
assumptions. Numerous workers have attempted to estimate the errors involved and the

effects on different systems due to the three assumptions [4, 6, 7, 8, 9, 10].

Mullin and Schnel [4] have calculated errors involved in anhydrous solids when the
activity coefficients are taken to be independent of concentrations (assumption 'a'). They
found that the larger the deviation of the ratio of the activities from unity, the greater is the
error involved. They report serious errors when this ratio deviates from unity by more than
:1: 1%. The assumption is valid only for very dilute solutions which, of course, is unlikely
to occur in the majority of crystallizing solutions. Also, for crystallization of electrolytes
from aqueous solution, it is imperative not to neglect the number of ions 'v' (assumption
'b'). The errors involved due to neglecting ‘v' may be greater than that due to the use of

concentrations in place of activities.

Schnel and Mullin [6] have extended the study of the errors involved to more
complex systems like hydrates, partially dissociated electrolytes, and to systems of mixed
electrolytes. For hydrates, the errors involved due to assumption 'a' are as important as for
anhydrous solids and that the larger the deviation of the ratio of the activities from unity,
the greater is the error involved. However, they conclude that any partial dissociation of
electrolyte in solution need not be taken into account to calculate the driving force. Other
electrolytes present in the system can significantly affect the driving force calculations and

as such should be taken into consideration.

 

Sbhnel and Garside [7] have compared different methods of calculating the activity
coefficients of solutes and have concluded that irrespective of the method used, the
deviation of the ratio of activity coefficients from the true value is greater at higher
supersaturation. Van Leeuwen [8] has developed relations between theoretical and
experimental measures for the crystallization driving force. For numerous systems
including growth from aqueous solution and melt growrh, it has been shown that the ratio

of the activity coefficients is much different from unity.

Stihnel and Garside [9] have calculated errors involved by using concentrations
instead of activities and have found that for the crystallization of K2804 and KH2PO4, the
differences are comparatively small. This result is explained by the fact that for the two

particular systems under study, the activity coefficient is a weak function of concentration.

Cardew and Davey [10] have discussed various ways of calculating the
supersaturation using both activities and concentrations. This distinction is shown to have

a significant role in the determination of constants in crystal growth rate models.

All the articles cited highlight the importance of using activity and activity
coefficients in the measurements of supersaturation. However, as mentioned, for practical
reasons, it is quite impossible to obtain the activity coefficients for all systems. The
working definition of supersaturation based on concentrations (equation 1.9) is invariably

used in all techniques currently available for supersaturation measurement.

Another technique employed for the measurement of supersaturation is the use of
Diihring's rule that states that the boiling point of a solution is a linear function of the
boiling point of the pure solvent at the same pressure. This technique has been used to
measure the supersaturation of sucrose solutions and for the development of an automatic

method for controlling the degree of supersaturation [11].

The point to be noted about the supersaturation measurement techniques that are

currently in use is that all of them rely on the measurement of mass based concentration.

The errors introduced due to the neglect of activity coefficients has already been discussed.
In spite of the awareness to the problem, there is no technique yet to measure the
supersaturation based on the activities of the solutes. The practical inability to measure the
activities (that inherently depend on the molecular interactions of the solute with the
solution), is a major obstacle in the development of a sensor to measure supersaturation. In
this work, it is proposed that since activity is a strong function of solution structure,
techniques that are sensitive to solution structure will be more likely to reflect changes in

activity thereby providing a more direct method of estimation of supersaturation.

Optical spectroscopic techniques such as fluorescence and infra-red spectroscopy
offer tremendous potential in this respect. Fourier Transform InfraRed Spectroscopy has
already been proven to be well suited for the in situ measurement of supersaturation in
maleic acid [12]. In this work, the emphasis is on using the numerous advantages of
fluorescence spectroscopy to develop a sensor that can be utilized in situ in a crystallizer for

the measurement of supersatuation.

1.3 Fluorescence Spectroscopy

Optical spectroscopy is based upon the interaction of light with matter. Every
molecule possesses a series of closely spaced energy levels. By the absorption of a
discrete quantum of light, the molecule can jump from a lower energy level to a higher one.
The energy of the quanta of light has to be equal to the energy difference between the two
energy states in order for the system to obey the fundamental equation (1.10) that defines

the relation between energy and a quantum of light.

E = by = -— (1.10)

E is the energy, h is Planck’s constant, v is the frequency, c is the speed of light in

vacuum, and l is the wavelength of light. This process of absorption is depicted in Figure

 

10

1.2. Each electronic state of the molecule is composed of various vibrational states,
denoted in the figure as 0, 1, 2, and 3 for the curve depicting the first excited state. The
schematic energy level diagram, the J ablonski diagram, for a diatomic molecule is shown in
Figure 1.3. The ground state is indicated by So, the first excited electronic state by 8,, and
the first excited triplet state by T1. The multiplicity of the state, one or three, defines
whether a particular state is a singlet or a triplet. When a quantum of light impinges on a
molecule, it is absorbed in about 10'15 seconds, and a transition to a higher electronic level
takes place. This absorption of radiation is extremely specific, and radiation of a particular
wavelength of light is absorbed only by a characteristic structure of the molecule. These
ground to singlet state transitions are responsible for the absorption spectra of the

molecules.

The molecule spends an average of about 10"° seconds in the excited state, during
which time, some energy in excess of the lowest vibrational energy level is dissipated. If
the remaining energy is not further dissipated by collisions with other molecules, the
* molecule returns from the lowest singlet excited state to the ground electronic state with the
emission of energy. This phenomenon is called fluorescence. Since some of the energy is
lost in the brief period before emission can occur, the fluorescence is always of a longer

wavelength than the energy that was initially absorbed.

Another pathway for the dissipation of energy is phosphorescence, in which an
intersystem crossing from the singlet to the triplet state occurs. When the energy from this
state is released, the molecule returns to the ground state. This phenomenon appears as a
delayed response at a much longer wavelength. While the lifetime for fluorescence is in the

order of nanoseconds, it is of the order of milliseconds or longer for phophorescence.

The highly selective nature of fluorescence makes it an ideal tool to be used in an
analytical sensor. In this investigation, the focus is on using fluorescent molecules as

analytical probes of their immediate environment. The overall objective is to be able to

11

<
ll

”*fi
<
II
H N w b
\

<
ll

/
k
l m

1
Fluorescence (E

Potential Energy

Absorption

 

 

 

 

Interatom'c Distance

Figure 1.2 Schematic energy level diagram showing the potential energy wells of the

ground state and the first electronicstate for a diatomic molecule.

12

S 2 Second Excited Singlet

T
Excited triplet T 2

Lowest
Triple State T1

/

 

I

l

I

First I

51 Excited Singlet f
A

 

 

Absorption
Collisional Deactivation
Fluorescence

+_______.__

 

 

 

 

V

Ground Electronic State

 

Figure 1.3 Schematic Energy Level (Jablonski) diagram for a diatomic molecule.

13

exploit this technique for use not only as probes of supersaturation, but also to use the
fluorescent probe technique as a tool to extract kinetic parameters that describe a batch

crystallization process.

1.4 Overview of this work

In Chapter 2, results are presented on the use of a fluorescent dye pyranine,
(essentially the trisodium salt of 8-hydroxy, 1, 3, 6 -pyrene itrisulfonic acid), that is
introduced into undersaturated and supersaturated citric acid solutions. This work is an
extension of earlier work done on similar lines with various sugar solutions [13, 14]. It is
evident that the change in the fluorescence spectrum of pyranine with a change in the citric
acid solution concentration can be correlated with a highly non-linear curve that serves as a
calibration curve. It is also shown that the fluorescence speCtrum of pyranine is a function
not only of the pH of the solution, but is also strongly dependent on the solution
concentration. The high degree of non-linearity in concentrated solutions indicates the high
sensitivity of the technique in the supersaturated region, which is of interest in all
crysrallization operations. The calibration curve so generated can then be used as the

standard curve to be compared with when this technique is used in a crystallizer in situ.

The drawback of using pyranine as the trace fluorescent probe is that it is listed only
as a Drug and Cosmetic (D&C) grade color by the FDA and as such may not appropriate
for use in some food and pharmaceutical applications for fear of contamination. In an
attempt to circumvent this problem, all colors that have been approved as Food, Drug, and
Cosmetic (FD&C) grade colors were screened to test their efficacy for use as a
concentration and supersaturation sensor in food applications. There are seven colors
currently listed by the FDA as FD&C grade. The results of these investigations are
presented in-Chapter 3.

14

While the use of these probes in solution provides very satisfactory results, in some
cases, it is necessary to immobilize or contain the probe in some fashion. This necessity
may arise, for example, in pharmaceutical crystallizations, where the highest degree of
purity is desired. Having the probe freely in the crystallizing solution may not be
appropriate because of the probability of incorporation of the probe molecules into the
crystal lattice, thereby reducing the purity of the final product.

In recent years, a vast amount of research has been conducted in the area of
optically based chemical sensors. A common feature of all these sensors is the presence of
an immobilized reagent phase that reacts with an analyte to produce a change in the optical
properties. This change is picked up by appropriate sensing devices. The optical indicator
phase sensors are sometimes referred to as 'optrodes' or 'optodes' by an analogy to
electrodes [15]. An optrode is defined as a device used at the interface between an optical
fiber and a solution under study. These devices are discussed in more detail elsewhere [16,
17, 18]. Depending on the nature of the signal, optical sensors are generally classified as
fluorescence, absorbance, Raman, or IR sensors. Out of these, the fluorescent sensors are
the most widely used. Optical sensors provide a number of advantages over conventional

chemical sensors, some of which are listed below [15, 16, 17, 18, 19]:

a) Optical sensors do not require a reference signal as in potentiomeuic methods.

Also, stability is improved with respect to calibration.

b) The ease of miniaturization allows the development of light, small, and flexible

fiber sensors.

c) Since the primary signal is optical, it is not subject to elecuical interferences like

static elecuicity or surface pctentials.

d) The revolution in communications has provided optical fibers which can transmit

light to and from remote locations without large intensity losses.

15

e) Much more information can be transmitted through an optical fiber than an
electrical lead. The signals can differ in wavelength, phase, decay profile, polarization, or

intensity modulation.

f) Optical sensors can offer significant cost advantages over conventional sensors,

particularly if one spectrometer is used with several sensors.
Notwithstanding these advantages, there are certain limitations to optical sensors:

a) Interference from ambient light is probable. The sensor must therefore be used
in the dark or the signal has to be encoded so that it can be separated from the ambient

signal.

b) Sensors with immobilized indicator phases may not have long term stability due

to photobleachin g.

c) Since the indicator and the analyte are in different phases, a mass transfer

resistance is necessarily involved, thus limiting the response time of the sensor.

d) Sensors with immobilized indicator phases have a limited dynamic range when

compared to electrodes.

Optical sensors have found widespread use specially in the biomedical area where

commercial systems are already available for the measurement of pH, pCOz, and p02 in

blood. Besides these applications, these sensors have found use in an extremely wide
variety of cases, including the sensing of alkali metal ions, glucose, methane,
trinitrotoluene, antigens, enzyme activity, ammonia, halide ions, non-alkali metal ions,
chlorinated hydrocarbons, albumin, and in analytical measurements such as ionic strength,
potential, adsorption sensing, humidity, temperature, radiation, sound, magnetism,

titrations, and substrate analysis [16, 20].

In the latter half of Chapter 3, results of experiments involving the immobilization

of the FD&C colors on anion exchange membranes are presented. It is shown that

 

16

although the fluorescent spectrum is altered because of the chemistry involved in the
immobilization, the essential emission properties of the probe molecules are still maintained

and that such a membrane can be used as a concentration and supersaturation sensor.

Chapter4 deals with a new fluorescent probe - chlorin, and a new technology for
the construction of the sensor. While with FD&C colors, ion exchange membranes were
used to immobilize the dyes, here, a TiO2 sol-gel film is created on a glass surface. This
film has an extremely porous structure and allows for the incorporation of guest molecules.
Chlorin was introduced into the film and trapped inside the sol-gel matrix. When such a
modified glass strip was introduced into aqueous sucrose solutions, the fluorescence
spectra of the chlorin changed markedly with the amount of sucrose in solution. The

suitability of a 501- gel based sensor is discussed in Chapter 4.

After establishing the feasibility of a fluorescence based sensor for measurement of
concentration and supersaturation in crystallizing solutions, other possibilities for using this
technique were explored. The measurement of kinetic parameters in a crystallization
process is an important task that affects not only the operating conditions of the crystallizer,
but is quite important in establishing any kind of a control strategy for the optimal operation
‘ of the crystallizer. Isothermal desupersaturation experiments were conducted on aqueous
citric acid solutions and the change in concentration with time was tracked with a
fluorescent probe that was added to the crystallizing solution. The ability of the fluorescent

probe to measure the desupersaturation characteristics is discussed in Chapter 5.

The desupersaturation experiments provide a wealth of information about the
process of crystallization. The kinetics of nucleation and growth, including their rates and
kinetic constants can be extracted from the desupersaturation curves by solving the
appropriate balance equations. Chapter 6 deals with these issues and provides a

methodology for estimating the kinetic parameters for citric acid solutions.

 

Fl --__....":-'

17

Chapter 7 introduces the concept of Reverse Flow Injection Analysis (rFIA) and

offers some ideas for future implementation. Results of preliminary experiments with rFIA

are presented and its potential to enhance the advantages of the fluorescence probe

technique is discussed. It is expected that future research will attempt to use these

modifications to the fluorescent probe technique to make it a much more practical analytical

tool. Ideas are presented about the direction this research should take in the future and the

anticipated benefits.

1 .5 References

10.

Proceedings of the Workshop on Opportunities & Challenges in Crystallization
Research, Iowa State University, Ames, Iowa. March 4-6, (1991).

H. K. Zimmerman, Chemical Reviews, The Experimental Determinations of
Solubilities. 51, 25-65, (1952).

J. W. Mullin, In: Crystallization , Butterworth Heinemann, Oxford, England,
(1993).

J. W. Mullin, O. Sohnel, Chemical Engineering Science, Expressions of
supersaturation in crystallization studies. 32, 683-686, (1977).

J. Garside, Chemical Engineering Science, Industrial Crystallization from Solution.
40, 3-26, (1985).

O. Sohnel, J. W. Mullin, Chemical Engineering Science, Expressions of
supersaturation for systems containing hydrates, partially dissociated electrolytes
and mixtures of electrolytes. 33, 1535-1538, (1978).

O. Stihnel, J. Garside, Journal of Crystal Growth, The Thermodynamic Driving
Force for Crystallization from Solution. 46, 238-240, (1979).

C. Van Leeuwen, Journal of Crystal Growth, On the driving force for
crystallization: The growth affinity. 46, 91-95, ( 1979).

O. Schnel, J. Garside, Journal of Crystal Growth, On Supersaturation Evaluation
for Solution Growth. 54, 358-360, (1981).

P. T. Cardew, R. J. Davey, and J. Garside, Journal of Crysral Growth, Evaluation
of Supersaturation in Crystal Growth From Solution. 46, 534-538, (1979).

11.

12.

13.

14.

15.

16.

17.
18.

19.

20.

18

A. L. Holven, Industrial and Engineering Chemistry, Supersaturation in Sugar
Boiling Operations. 34, 1234-1240, (1942).

D. D. Dunuwila, An Investigation of the Feasibility of Using In Situ ATR FTIR
Spectroscopy in the Measurement of Crystallization Phenomenon for Research and
Development of Batch Crystallization Processes. PhD Thesis, Michigan State
University. (1996).

R. Chakraborty, K. A. Berglund, Journal of Crystal Growth, Steady State
Fluorescence Spectroscopy of Pyranine as a Trace Extrinsic Probe to study
structure in aqueous sugar solutions. 125, 81-96, (1992).

R. Chakraborty, K. A. Berglund, AIChE Symposium Series No. 284, The Use of
Pyranine as a Trace Fluorescent probe to Study Structure in Aqueous Sucrose
Solutions. 83, 114-123, (1991).

O. S. Wolfbeis, Trends in Analytical Chemistry, Fluorescence Optical Sensors in
Analytical Chemistry. 4, 184-188, (1985).

W. R. Seitz, CRC Critical Reviews in Analytical Chemistry, Chemical Sensors
Based on Immobilized Indicators and Fiber Optics. 19, 135-173, (1988).

S. A. Borman, Analytical Chemistry, 53:14, 1616A-1618A, (1981).

J. Janata, In: Principles of Chemical Sensors, Plenum Press, New York, 241-284,
(1989).

W. R. Seitz, Analytical Chemistry, Chemical Sensors based on Fiber Optics. 5 6,
16A-34A, (1984).

U. J. Krull, R. S. Brown, in Laser Remote Chemical Analysis, R. M. Measures,
Eds. John Wiley and Sons, Vol. 94, 505-532, (1988).

Chapter 2

 

USE OF FLUORESCENCE SPECTROSCOPY IN
CONCENTRATION AND SUPERSATURATION
MEASUREMENTS IN CITRIC ACID SOLUTIONS'

2. 1 Background

The study of solution structure is critical for a complete understanding of the
behavior of crystallizing solutions, especially in the supersaturated region. In this work,
we show that since optical techniques involving spectroscopy are dependent on solution
structure, they are well suited to reflect changes in activity, thereby providing a more direct

method of estimation of supersaturation.

Numerous workers have attempted to study solution structure using a wide variety
of techniques. Frank and Wen [1] first proposed a theory involving a cluster model for
water molecules. Since then, numerous studies have been conducted to understand the role
of the solvent molecules and their interaction with the solute molecules. The solvent-solute
and the solute-solute interactions are relatively more complex to study, especially in
concentrated solutions, and therefore have been studied mainly for undersaturated

solutions.

A variety of means have been used for studying the nature of molecular association

and solute organization. Mullin and Leci [2] found that an isothermal column of a

 

° Yedur, S. K., and K. A. Berglund. Applied Spectroscopy. In print. 1996.

19

20

supersaturated solution of citric acid generated a concentration gradient over a period of
several days. It was suggested that solute clusters were being formed and the concentration
gradient developed as a result of the density difference between the clusters and the
solution. The large degree of hydrogen bonding possible between the citric acid molecules
and the solvent water molecules was thought to be the cause of cluster formation. Cussler
[3] conducted diffusion studies in the triethylarnine-water and other binary systems and
showed that the diffusion in the systems studied could be predicted much better by using a
model that allowed for cluster diffusion as well as molecular diffusion. Myerson and co-
workers [4,5] observed the decrease in the diffusivities of urea, glycine, KCl, and NaCl
solutions which provided evidence for molecular aggregation. Hussman et al. [6] and
McMahon et al. [7] also confu'med cluster formation with Raman spectroscopy on alkali
nitrate solutions. Larson and Garside [8,9] measured density gradients in columns of citric
acid, urea, sodium nitrate, and potassium nitrate solutions, confirming the formation of
solute clusters in supersaturated solutions. Ohgaki et al. [10] have observed solute clusters

and their aggregations in citric acid solutions using scanning electron microscopy.

Although many investigations have demonstrated microscopic solvent interactions
using techniques such as NMR, X-ray diffraction, and Raman [11,12,13], a novel method
using the fluorescence properties of a probe molecule was developed by Chakraborty and
Berglund[14,15]. In their work, the authors have demonstrated that relative amounts of
associated and bulk water can be estimated by monitoring the relative peak intensities of the
steady state fluorescence of a probe molecule such as trisodium 8-hydroxy-l,3,6-

pyrenetrisulfonate (pyranine). The molecular structure of pyranine is shown in Figure 2.1.

Pyranine is a non-toxic D&C dye and as such is approved for use in drug and
cosmetic applications. It is listed by the FDA as D&C Green No. 8. Pyranine has been
widely applied to the study of reverse micelles [16, 17, 18], sols [19], sol-gels [20], and as

a source of protons for pH jump experiments [21, 22, 23].

21

NaO3S CH

Figure 2.1 Molecular structure of Pyranine (trisodium salt of 8-hydroxy-1,3,6-

pyrenetrisulfonic acid)

22

The emission properties of pyranine have been well characterized [16, and
references therein]. In aqueous solution, pyranine exists with the sulfonate groups
dissociated. Upon excitation with light at a wavelength of around 355 am, it has two
excited states. One, emitting at 440 nm, is due to the protonated form of the molecule,
whereas the other, emitting at 510 nm, is due to the deprotonated form. These kinetics are
shown in Figure 2.2. Whether the observed emission is from the protonated form or from
the deprotonated form depends on the molecular environment in its immediate vicinity.
Thus, solute-solvent interactions in solution can be studied by observing the emission of

the probe molecule.

In this work, the concentration of citric acid solutions has been measured by using
pyranine, both in undersaturated and supersaturated conditions, to get an accurate

estimation of the supersaturation.

The pH of citric acid solutions decreases linearly with concentration. As has been
noted [17], the emission properties of pyranine are affected by the pH of the solution. In
this work, howeVer, we show that the pyranine molecule's response to the citric acid
solution is more than a pH effect alone and that it is possible to get an estimate of the

supersaturation by using this fluorescence technique.
2. 2 Materials and Methods

A stock solution of pyranine (obtained from Eastman Kodak, Rochester, NY) was
prepared by dissolving 0.0264 grams of pyranine in 50 mL of spectroscopic grade
methanol. Citric acid solutions of 10, 20, 30, 40, 50, 60, and 70 weight percent were
prepared by dissolving the appropriate amount of citric acid monohydrate (obtained from
Columbus Chemical Industries Inc., Columbus, WI) in distilled water. The 70%
supersaturated solution was prepared by mildly heating the solution in a water bath until all

the solute was dissolved and subsequent cooling to room temperature. The solutions were

355 nm m

 

 

 

23

 

PyOH PyO + H
440 nm 510 nm >W 355 nm
PyOH PyO' + H
Figure 2.2 Excited state kinetics of pyranine

24

used as prepared without any degassing. The pH of the solutions was measured using a

pH elecrrode with an Orion Research digital ionalyzer/501.

Ten microliters of the pyranine stock solution was taken in a vial and the methanol
allowed to evaporate. Ten milliliters of the citric acid solution was added. This procedure
was repeated for each of the citric acid solutions prepared. A 1.0"'10'6 M pyranine
concentration was thus achieved in each case. A similar protocol was followed to obtain

pyranine concentrations of 1.0"‘10'5 M and 1.0"‘10'7 M.

Fluorescence spectra of the pyranine in each of the citric acid solutions were
collected within a couple of hours of the preparation of the solutions. All spectra were
taken in a controlled environment set at 25 'C with a SPEX 1681 FLUOROLOG.
Spectrometer equipped with an Xenon lamp source and a PMT detector. A quartz cuvette
was used in all cases. Based on the absorption behavior of pyranine, an excitation
wavelength of 342 nm was used. The emission was recorded over a range from 400 to
600 nm. The quantum efficiency of pyranine is close to unity so a very small amount of
excitation light is required. Copious amounts of emitted light are generated which
prompted the excitation monochromator slits to be 0.5 mm Open while the emission
monochromator slits were opened to 1 mm. The response of the PMT detector in the range
of wavelengths studied was constant which precluded the need for corrections to be made

for PMT response.

The fluorescence spectra of pyranine in solutions of different pH's were also
obtained. Solutions with pH's of below 3 were studied. These solutions were prepared by
adding appropriate amounts of hydrochloric acid and ammonium hydroxide. A similar
procedure to that described earlier was used to prepare 1.0"‘10'6 M pyranine concentration

solutions at different pH's.

25

2.3 Results and Discussion

The quantum yield of pyranine is very close to unity under all experimental
conditions [23]. This property was one of the strongest points behind choosing pyranine
as the probe. Although it is unclear whether the yield is unity for both emitting species, the
change would be the same for both the species. This is evidenced by the fact that the
emission spectra show a very consistent relative change in the two peaks. Any possible
variation in the individual quantum efficiencies will be factored out during the ratioing of

the two peaks.

Even though the probe may be affected by oxygen quenching, in this case it is not a
potential problem as the purpose of the experiments is to monitor the performance of the
probe in a real-life crystallizing solution. For this purpose, the aqueous solutions prepared
are assumed to be saturated with oxygen and so any quenching would not adversely affect

the behavior of the probe during the course of the experiment.

From previous literature [14,15], it was known that the emission properties of
pyranine do not change from a range of 10 to 100 ppm in aqueous sucrose solutions. To
test any possible effect in citric acid solutions, experiments were conducted with three
pyranine concentrations of 1.0"‘10‘5 M, 1.O"‘10‘6 M, and 1.0"‘10‘7 M. As expected, there
are no distinctive changes in the three spectra. Two peaks at 440 nm and at 510 nm appear
with an isosbestic point. However, when the concentration is l.0"'10'S M, there appears to
be a diminution of emission intensity in highly concentrated citric acid solutions. This
might be due to some self-aggregation of the probe in the highly viscous environment.
This does not mean that there is any evidence for excimer formation, as no broadening of
the peak or bathochromic shifts are noticed. The fact that pyranine is not a planar molecule
(due to the presence of the sulfonate groups) also excludes the possibility of excimer
formation. As expected, the absolute intensities vary slightly for the lower two

concentrations, but identical peak intensity ratio curves are generated. This suggests that

26

either of these. concentrations may be used. However, in this work, we choose to use a
pyranine concentration of 1.0*10'5 M throughout, as it is a more convenient amount to
work with. The relative inexpensiveness of the probe does not justify the use of a lower

concentration that is more difficult to measure accurately.

The emission spectra of 1.0*10'6 M pyranine in citric acid solutions of four
different concentrations are shown in Figure 2.3. The spectra of the other concentrations
also followed the trend. As noted previously, the emission spectrum of pyranine in water
shows two distinct peaks, one at 440 nm and the other at 510 nm. Pyranine occurs in two
excited states in equilibrium: one in which the hydroxyl proton is associated with the
molecule and another in which it is dissociated. As is evident from the figure, the
intensities of the peaks at 440 and 510 nanometers change with citric acid concentration.
An isosbestic point is observed at 495 nm that indicates that the two fluorescing species are
in equilibrium. At least three experiments were carried out at the same concentrations and

the spectra are quite reproducible.

At low concentrations of citric acid, emission at 510 nm is more intense than that at
440 nm, indicating the presence of deprotonated pyranine in solution. At higher
concentrations, the intensity of emission at 440 nm increases with a corresponding decrease
in the intensity at 510 nm. This indicates that the equilibrium between the deprotonated and
protonated forms of pyranine shifts towards the protonated form with an increase in citric
acid concentration. Fully protonated pyranine molecules exist only when there are no
proton acceptors (water molecules) available for exchange in the immediate vicinity of the

pyranine hydroxyl proton.

The Peak Intensity Ratio (PIR), defined as the ratio of the peak intensity at 440 nm
to the intensity at 510 nm is thus a measure of the ratio of water associated with the
solute(citric acid) to that not associated with the solute. The use of this ratio also provides

an internal correction for intensity fluctuations due to possible variations in the

Figure 2.3.

Intensity (counts per second)

27

 

70% citric acid

 

 

 

 

l
400 450 500 550 600

Wavelength (nm)

Emission spectra of pyranine in citric acid solutions of 10, 30, 50, and 70

weight percents; taken at 25 °C.

28

spectrophotometer setup. Chakraborty and Berglund [14,15] have put forward a theory
based on the work of Birch et al. [24] according to which three kinds of exchangeable
protons are available in concentrated sugar solutions: bulk water, solvation water, and the
sucrose hydroxyl. Here, we propose that a similar environment occurs in the case of
concentrated citric acid solutions. Out of the available protons for exchange, the protons
associated with the citric acid molecule are unlikely to compete with the water molecules in
interacting with the pyranine molecules because the pH of the solutions are generally below
the pK's of the carboxylic acid groups. It follows that the pyranine only reacts with the
water molecules and nor with the citric acid molecules. The pyranine molecule that interacts
with a water molecule which is in contact with other water molecules will become
deprotonated and will emit at 510 nm. This suggests one microenvironment of the
pyranine to be bulk water. On the other hand, a pyranine molecule in the vicinity of
solvation water that is already associated with a citric acid molecule would not exchange
protons and would be in the protonated state emitting at 440 nm. Thus, the other
microenvironment of the pyranine molecule would be water of solvation associated with the

citric acid molecules.

The previous discussion suggests that pyranine responds to the available water in
its immediate environment. This association with water is, in turn, is related to the amount
of citric acid present in solution. Thus, the change in emission properties of pyranine can
be related to the concentration of aqueous solutions of citric acid. A graph of the PIR
versus concentration is shown in Figure 2.4. The smooth nature of the curve indicates that
it can be used to determine the concentration, and also the supersaturation of citric acid
solutions. The presence of an upward curvature indicates that the response is highly non-
linear. The same data are presented as a plot between the PIR and the supersaturation in

Figure 2.5. It has been shown elsewhere [14,15] that the supersaturation measurement

29

 

3.5 l l j I I

Peak Intensity Ratio

 

 

 

l

O l I l l
10 20 3O 4O 50 60 7O

citric acid concentration, wt%

Figure 2.4 Peak Intensity Ratio (ratio of emission peak intensities at 440 nm and 510

nm) as a function of citric acid concentration.

3O

 

      
   

3.5“; ‘
3d: cir-
o - .
t3 3 :
a: 2.54- ..
>. C I
.t.’ - .
m 1
c 2- --
m 4
H l
E .
x - cli-
m 1.5 .
a) .
a. .
1 . -_
0.5“ ._

 

 

Ora l A 1 J A l A l 4 A J 144 l A: J I 1 '4 A 1

j I I
-r -0.8 -O.6 -o.4 -o.2 o _ 0.2

 

Supersaturation of Citric Acid Solutions

Figure 2.5 Peak Intensity Ratio (ratio of peak intensities at 440 nm and 510 nm) versus
the supersaturation 8 (defined as S=(c-c*)/c*, where c and c* are the actual

concentration and the saturation concentration respectively).

31

using the PIR is more sensitive than other techniques, as it is a consequence of the activity

rather than just the concentration.

There is, however, a caveat because the pH of citric acid solutions changes
dramatically with concentration. It has been found that the effect of varying the pH on the
emission of pyranine is also striking. Results of Bardez et al. [17] suggests that lowering
of the pH also causes an increase in the emission at 440 nm. The question arises as to
whether the changes in the PIR seen in the citric acid solutions were just pH effects or were

truly the result of changes in the microenvironment of the probe molecule.

Figure 2.6 shows a plot of the pH of citric acid solutions versus the concentration.
The plot indicates that as the concentration of the aqueous solution increases, the pH
decreases linearly. At extremely high concentrations of near saturation (about 65 wt% at
30°C) and above, the pH falls very close to zero and also dips below zero. A reliable

measurement using a pH meter is not possible at these high concentrations.

Experiments were conducted by dissolving pyranine in HCl-NaOH solutions of
different pH's. At basic conditions of high pH's, pyranine cannot be excited from the
ground state due to it’s pKa. At such pH’s, it is not feasible to study the emission
properties and so the emission of pyranine was studied only in solutions of lower pH

where it can be excited. This was also the range of the pH of citric acid solutions.

Figure 2.7 shows a comparison of the PIR's obtained fiom pyranine emission in
citric acid solutions to the PIR's in HCl-NaOH solutions of different pH's. As is evident,
the emission in the presence of citric acid shows a steeper slope and greater absolute values
than the PIR in the absence of citric acid, indicating that the pyranine molecule, besides
being sensitive to the pH of the solution, also responds to the presence of the citric acid
present. Since the relationship between the concentration and the pH is linear, and that

between concentration and PIR is highly non-linear, it can be concluded that the PIR

 

1,5....J,...I. I. . .+. I

1.4-

It'tffi

12*

vrt[r

pH

V V I

0.6-

0.4-

0.2‘

UTIITTI'IVII

 

 

O LAALJ A A A A4 AAA A lAJJA l A A A A lJLJ A l AAJA

l
0 IO 20 3O 40 SO 60 7O

 

Citric acid concentration, wt %

Figure 2.6 pH of aqueous citric acid solutions as a function of citric acid concentration.

33

 

Peak Intensity Ratio

 

b

 

 

o ~dwi-in

j ' r T
O 0.2 0.4 0.6 0.8 1 l2 1.4 1.6
pH

 

A A A l A A A 1 AJ J l A A A

Figure 2.7 Comparison of PIR's (ratio of peak intensities at 440 nm and 510 nm)
obtained from pyranine emission in citric acid solutions as a function of pH
of the solutions to the PIR's obtained from pyranine emission in HCl-
NaOH solutions of varying pH's. Closed circles: Citric acid solutions;

open circles: HCl-NaOH solutions.

34

obtained can be used to measure the supersaturation of citric acid solutions. In an effort to
decouple the effect of pH from the effect of the solution on the PIR, both the curves were
fit to exponential curves. The curve fits provided significant R2 values of 0.98 and 0.97
for the overall effect and the pH effect, respectively. Using these fits and the relationship
between concentration and pH shown in Figure 2.7, the pH effect was subtracted from the
overall effect to generate a corrected PIR curve as a function of citric acid concentration.
This corrected calibration curve is shown in Figure 2.8. This plot clearly demonstrates that
the non-linear response of the PIR as a function of concentration can be used as a
calibration curve for the measurement of supersaturation as the technique works well even
in the supersaturated region. Note that the solubility at the experimental temperature of 25

'C is marked on the plot by the arrow.

2.4 Conclusions

The emission properties of a fluorescent probe molecule like pyranine in trace
amounts can be used to obtain an estimate of the supersaturation of citric acid solutions.
The method used, involving the measurement of the Peak Intensity Ratio (PIR), defined as
the ratio of the intensity of the emission peak of pyranine at 440 nm to the intensity of the
peak at 510 nm gives rise to a parameter - the PIR, that is strongly correlated to the solute
concentration. The relationship between the two is highly non-linear and extends into the
supersaturated region. Since the emission properties are a function of the solvent
microenvironment around the probe molecule, they are strongly dependent upon the activity
of the solvent. This technique holds promise for better accuracy than current mass based
techniques for supersaturation measurements. It is also shown that the response of the
pyranine is not only due to a pH change of the citric acid solution at higher concentrations,

but truly reflects the solute and solvent microenvironments.

35

 

 

 

 

3 r 1 A1 1 r 1
2.5“ "
2. ,,
E
O.
b h
8 1.5"
§
5
U
1 II P
0.5' '
0 fi r I I r I
10 20 30 40 SO 60 70

citric acid concentration (wt%)

Figure 2.8 Corrected PIR (corrected for pH effects) versus citric acid concentration

(wt%). The arrow indicates the solubility at 25 °C.

36

2.5 References

5‘5”.“

S"

10.
11.
12.

l3.
14.

15.
16.
17.

18.

I9.

W.S. Frank and W.Y. Wen, Disc. Faraday Soc. 24, 1333, (1957).
J.W. Mullin and CL. Leci, Phil Mag. 19, 1075, ( 1969).
E.L. Cussler, AIChE Journal. 26:1, 43, (1980).

Y.C. Chang and A.S. Myerson, in: Industrial Crystallization '84, SJ. Jancic and
EJ. DeJong, Eds. Elsevier, Amsterdam, 27, (1984).

LS. Sorrel and A.S. Myerson, AIChE Journal, 28:5, 772, (1982).

GA. Hussman, M.A. Larson, and K.A. Berglund, in: Industrial Crystallization
'84, SJ. Jancic and El. DeJong, Eds. Elsevier, Amsterdam, 21, (1984)

RM. McMahon, K.A. Berglund, and M.A. Larson, in: Industrial Crystallization
'84, SJ. Jancic and EJ. DeJong, Eds. Elsevier, Amsterdam, 229, (1984)

M.A. Larson and J. Garside, Chem. Eng. Sci. 41:5, 1285, (1986).

M.A. Larson and J. Garside, J. Crystal Growth. 76, 88, (1986).

K. Ohgaki, N. Hirokawa, and M. Ueda, Chem. Eng. Sci. 47:8, 1819, (1992).
M. Mathlouthi and D.V. Luu, Carbohydrate Research. 81, 203, (1980).

M. Mathlouthi, C. Luu, A.M. Meffroy-Biget, and D.V. Luu, Carbohydrate
Research. 81, 213, (1980).

M. Mathlouthi, Carbohydrate Research. 91, 1 13, (1981).

R. Chakraborty and K.A. Berglund, AIChE Symposium Series no. 284. 83, 114,
(1991).

R. Chakraborty and K.A. Berglund, J. Crystal Growth. 125, 81, (1992).
H. Kondo, I. lea, and J. Sunamoto, J. Phys. Chem. 86, 4826, (1982).

E. Bardez, B.T. Gougillon, E. Keh, and B. Valeur, J. Phys. Chem. 88, 1909,
(1984).

GD. Correll, R.N. Cheser, F. Nome, and J.H. Fendler, J. Am. Chem. Soc.
100:4, 1254, (1978).

V.R. Kaufman, D. Anvir, D. Pines-Rojanski, and D. Huppert, J. Non-Crystalline
Solids. 99, 379, (1988).

20.
21.

22.
23.
24.

37

LC. Pouxviel, B. Dunn, and J.I. Zink, J. Phys. Chem. 93, 2134, (1989).

K.K. Smith, K.J. Kaufman, D. Huppert, and M. Gutman, Chem. Phys. Letters.
64:3, 522, (1979).

E. Pines and D. Huppert, J. Phys. Chem. 87, 4471, (1983).
M.J. Politi and J.H. Fendler, J. Am. Chem. Soc. 106:2, 265, (1984).
GO. Birch, J. Grigor and W. Derbyshire, J. Solution Chem. 18:8, 795, (1989).

Chapter 3

A NOVEL ANALYTICAL USE OF FOOD GRADE
COLORS: SPECTROSCOPIC MEASUREMENT OF
CONCENTRATION AND SUPERSATURATION IN
SUCROSE SOLUTIONS

3.1 Background

The use of food colors has traditionally been only as an additive to various food
products and cosmetics along with certain very specific uses in medical devices. Colors are
applied to fruits, meats, and other foods to improve their desirability and are added to drugs
and pharmaceuticals primarily for ease of identification [1]. The utility of these colors is
limited to increasing the visual appeal of the final product. Foods and cosmetics are
colored basically in two ways - either the free dye is added to the solution, or the dye is
added as a lake. A lake is defined as a pigment that is prepared by precipitating a soluble
dye onto an approved insoluble reactive or adsorptive substratum or diluent [1]. Currently,
alumina is the only approved base for FD&C lakes. While there are a number of colors
approved for external drug and cosmetic use, there are only seven colors listed in the
United States by the Food and Drug Administration (FDA) as certified for food, drug, and
cosmetic (FD&C) use. In the following list, the official designation is as a number, the
common names are listed in brackets. The colors are FD&C Blue No. 1 (brilliant blue),
FD&C Blue No. 2 (indigo carrnine), FD&C Green No. 3 (fast green), FD&C Red No. 3
(erythrosine), FD&C Red No. 40 (allura red), FD&C Yellow No. 5 (tartrazine), and

38

39

FD&C Yellow No. 6 (sunset yellow). The chemical structures of the seven FD&C colors
are shown in Figure 3.1. Some important physical properties are listed in Table 3.1.
Additional basic information about food colors and their application has been reviewed and

compiled elsewhere [1, 2, 3].

The combined usage of these colors in all various applications is quite high. In
1990, the total amount of FD&C primary colors and lakes certified by the Food and Drug
Administration exceeded 8.5 million pounds [1]. While the use of these colors as visually
appealing materials are apparent, their use as analytical probes, however, is a novel
approach that has come out of the work done in this laboratory in the last few years on

fluorescent compounds [4, 5, 6].

The proposed research utilizes a technique involving a fluorescent probe molecule,
that is based on the sensitivity of the probe molecule placed in a crystallizing environment,
to the solution structure. It is known that the relative changes in the emission spectra of
compounds such as pyranine, carrninic acid, nile red, and pyrene butyric acid, when placed
in different solutions, can be correlated to the polarity of the solution. The photochemistry
of some fluorescent probes like pyranine [7] and carrninic acid [8, 9] arewell understood,
whereas the majority of probes are yet to be studied. The molecular structure of pyranine is
shown in Figure 2.1 while Figure 2.2 shows the kinetic diagram for the two-state excited
state proton transfer process of pyranine (PyOH). Pyranine is highly soluble in aqueous
solutions where it is completely dissociated. The protonated form emits at a wavelength
maxima of 440 nm, while the dissociated form emits at a wavelength maxima of 511 nm.
Steady-state fluorescence investigations using pyranine as a trace fluorescent probe have
demonstrated that the relative amounts of associated and bulk water can be measured by
monitoring the relative peak intensities of the steady state fluorescence [4, 5, 6]. This
result suggests that pyranine is an excellent probe for measuring bulk concentration of the

solution in which it is placed.

40

SO3Na O

/©-—N(C2H5)CHT—C§
.\ C2 5 H
Q:\<:>=N’< H >c r—QO

 

 

FD&C Blue No. 1 FD&C Blue No. 2
SO3Na
/©—N (Csz)CH2 —0 H3 OH
OH QC Na038 gqu:
3° \QN (C zlis)CH2 ‘Q

aNa Q503M

FD&C Green No. 3 FD&C Red No. 40

I . 1
NaO _ / o
‘- \ I = O I
0 COM:
FD&C Red No. 3
HO
HO —c—-N—©_so3~a
II I NaOaS —©-N=N
NaOaS-O—N— —C\C//N
COONa O3Na
FD&C Yellow No. 5 FD&C Yellow No. 6

Figure 3.1 Molecular structures of the seven FD&C colors.

41

.2388an :58 8 8982 .533 E commie .5.“ 98 832;
=< .33: 2: E eccozcoE fl 0% some 8% 59.20%)» 5:863 2: £822: cocoomeosc 05 Sum £53 £5 Eat Ba 8:?» 82F *

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

23082 coco coco bo> Coca ham boom bo> han— bEnSm Ewfi
Sfiooflhfiwv
.0. mm 8 .533
2 8 S a cm 3 8 s 3528
E: {guinea
- - 3% 0mm moo - woo cocoomoeoim
E: {2:522
33 mmv mow Rm 03 Se Imc :o:&8n<
30:0»
588m 30:“; 5qu tom .3323». xci £55 .520 555 03m 53 BE .2520 on:
2.2 820
38:02 vac—eufizm Sacco—2 255:5“ Acacia—5&5. Eowfifi aegis—BE. EBEQC
eczema:—
won— com E ”am 65:50 mom 252
Bo:o> Homcsm ocfigrr .eom £32 oEmESEm 580 5mm 09?: 0:5 ESE—tn 58800.
9 .02 m .02 ca .02 m .02 m .02 N .02 A .02
30:»; Dad”. 32:; Down—m BM Odom—m tom Dank 520 Down—lull 02m Dome usflouwam baomoi

 

 

.2 3 $28 Dome 05 we meme—03am Rush—m taco—om

fin 033.

 

42

The drawback of using pyranine is its standing as a D&C grade color only and as
such is not appropriate for use in food and pharmaceutical processes for fear of
contamination. The present work is an attempt to circumvent this problem by screening
dyes that have been approved as FD&C grade colors to test their efficacy for use as a
concentration sensor in food and pharmaceutical systems. Although photochemical
information other than the absorption specna about the food grade dyes is not available, it
is reasonable to believe, (by a comparison of the structures and properties), that the
photochemistry is probably similar to that of pyranine. As discussed in the Results and

Discussion section, the emission spectra confirm this assertion.

In this work, we report on the use of FD&C colors as probes for the measurement
of the concentrations of the solutions in which they are placed. The impetus for this work
is also derived fiom the fact that these colors are already present during the manufacture of
the food items. For example, the FD&C colors are present during the processing of many
varieties of candies and sodas. It is possible to use the emission properties of these colors
in situ, to record concentration differences of the solutions in which they are present. The
test crystallizing solute chosen is sucrose. Sucrose is chosen due to its low cost, high
solubility, and high viscosity. Furthermore, more sucrose is crystallized worldwide than

any other compound.

3.2 Materials and Methods

All FD&C colors used in this work were obtained as samples fiom Warner -
Jenkinson Company, Inc. St. Louis, U.S.A. and had purities of 90%. They were used as
obtained to confirm the practical utility of the approach. Sucrose (reagent grade) was
obtained from Aldrich Chemical Co. Sucrose solutions of different concentrations (wt%)
were prepared by dissolving the appropriate weight of sucrose in distilled water. All

experiments were conducted at ambient temperature unless otherwise noted. In order to

43

prepare supersaturated solutions, the mixture was gently heated to a higher temperature so
that all solids were dissolved and the solution was cooled back to the ambient temperature.
3 . 2 . 1 Absorption experiments

Absorption spectra were collected on a Guided Wave Model 260 spectrophotometer
equipped with a tungsten-halogen lamp as the light source and a silicon detector.
Fluorescence emission spectra were collected on a SPEX FLUOROLOG
spectrophotometer that has a xenon lamp as a source and a PMT detector.

For the absorption spectra, color solutions were prepared to give 5 ppm solutions
of FD&C Blue No. 1 and FD&C Green No. 3; 10 ppm of FD&C Blue No. 2 and FD&C
Red No. 4; and 20 ppm of the remaining four. The desired concentration was achieved in
each case by dissolving ‘x’ grams of the color in 20 mL of water to make the stock
solution. Next, 10 uL of the stock solution was pipetted out and dissolved in 5 mL of
distilled water. This protocol gave a concentration of ‘100x’ ppm.

' 3.2.2 Fluorescence experiments

For the fluorescence spectra, the stock solutions were prepared in a similar fashion,
but in each case generating a sample sucrose solution that had a dye concentration of 1.0e-
05 M. The smaller concentration compared to absorption is a necessity of the fluorescence
experiment. A higher concentration causes undesirable effects such as the inner filter
effect. The spectra of the free dyes in solution were collected by placing the test solutions
in 4 sided 1 cm by 1 cm quartz cuvettes that were placed in the sample compartment of the
FLUOROLOG. The excitation wavelengths used were as follows: for FD&C Red No. 40:
500 nm; for FD&C Green No. 3: 420 nm; for FD&C Blue No. 1: 400 nm; for FD&C
Yellow No. 5: 433 nm; and for FD&C Yellow No. 6: 484 nm.

3.2.3 Immobilization of the probes

For the experiments with the immobilized dyes, the food colors were immobilized

on ion exchange membranes obtained from Schleicher and Schuell, Keene, NH. The

specific membrane used was the DEAE cellulose acetate NA45 anion exchange membrane

44

that had a pore size of 0.45 um. The dye was immobilized on to the membrane using the
following protocol. Five hundredths of a gram of FD&C Blue No. l was dissolved in 20
mL of distilled water. From this stock solution, 20 uL was pipetted out and dissolved in 5
mL of distilled water to give a concentrated dye solution. A 2 cm by 0.5 cm strip of the
membrane was placed in the solution and manually stirred vigorously for one minute and
then thoroughly rinsed with diStilled water. The membrane turned from white to a bright
blue color. A sensor was constructed by attaching the membranes over the closed end of a
1 cm diameter flat-bottomed glass tube. A bifurcated fiber optic set, manufactured by
SPEX, was inserted into the open end and placed flush with the closed end. One end of
the bifurcated end was connected to the light source of the FLUOROLOG via a
monochromator, whereas the other end was connected to the PMT detector. Thus, the
bifurcated fiber optic served a dual purpose by providing the exciting light and collecting
the emitted light from the probe. The emission and excitation slits were kept cpen at 2 mm
and 6 mm, respectively. The complete sensor was directly dipped into a cyrstallizer. This

arrangement is shown schematically in Figure 3.2.

3.3 Results and Discussion

3.3.1 Results with the ‘free’ dye in solution

Fluorescence spectrosc0py is an extremely sensitive technique that involves the
study of the emission spectrum of a molecule after it has absorbed energy from an
excitation source. Probe molecules absorb light of a specific wavelength and then emit the
light at longer wavelengths. If this absorption and/or emission wavelength shifts as a result
of the solvent environment that they are placed in, the probes are termed solvatochromic.
In addition, some molecules exhibit a simple change in the emission intensities when placed
in differing solution environments. Both types of change in the emission spectrum are
extremely sensitive to small changes in the solvent concentration and can be used to

monitor changes in the solution composition continuously.

45

 

 

 

 

 

 

 

 

 

 

 

 

Figure 3.2 Schematic diagram of the apparatus used to measure the fluorescence from

the immobilized sensor that is immersed in a crystallizer. 1) Source. 2)
Collimating lens. 3) Monochromators. 4) Jacketed crystallizer. 5)
Immobilized probe assembly. 6) Bifurcated fiber optic. 7) Stirrer. 8) PMT

Detector. 9) Computer.

46

The main objective of this study was to screen the seven FD&C colors for their
suitability to behave as analytical probes of the concentration of the solution in which they
were placed.

Figure 3.3 shows the absorption spectra of each of the seven FD&C dyes. The
wavelength corresponding to the maximum absorption indicates the wavelength that should
be used as the excitation wavelength for fluorescence experiments. Each dye therefore will
be excited at a wavelength corresponding to its absorption maximum.

In order to provide an internal standard for the emission measurements, it is
necessary to normalize the spectrum. As a result, there would be no variation in data due to
any instrumental artifacts. A normalized calibration curve should be the same regardless of
any instrumental variations. Normalization can be accomplished in a variety of ways. The
most common technique is to take a ratio of the intensities of two of the most intense peaks
in a spectrum. Then, a calibration curve can be generated that shows the peak intensity
ratio as a function of the solute concentration. Another possible normalization technique
involves normalizing all spectra with respect to a standard spectrum, for example, a
spectrum in pure water (0% sucrose). This latter technique is especially useful when there
are no clear multiple peaks in the spectrum for the ratioing technique to be feasible. Both
these techniques have been demonstrated here. The normalized parameter is then plotted
against the solution concentration to give rise to a calibration curve that shows the strong
dependence of one on the other.

FD&C Blue No. 2 (Indigo camrine) and FD&C Red No. 3 (Erythrosine) do not
exhibit good stability towards light. Both colors fade appreciably within a week, that
makes them unsuitable for use in a sensor. Hence, these two were not considered any
further. Results obtained with the two yellow dyes, FD&C Yellow No. 5 and FD&C
Yellow No. 6 showed no reproducible conelation to solute concentration. These,
therefore, were also not considered any further as candidates for a concentration sensor.

The lack of correlation is probably due to high pKa values of the excited state and

Figure 3.3

47

r r r r r r r J
14 ,........j—r..r..,...,...,...,.
.
.

FD&C Blue No.1

 

 

 

—O— FD&C Blue No.2

1.2 ': —o—- FD&C Green No.3 T

+ FD&C Red No.3

_ + FD&C Red No.40

1 . + FD&C Yellow No.6
- + FD&C Yellow No.5

 

 

 

Absorption Intensity (arbitrary units)

 

 

 

 

400 440 480 520 560 600 640 680
wavelength, nm

Absorption spectra of the FD&C colors. The concentrations of each in
aqueous solution are: FD&C Blue No. 1 and FD&C Green No. 3: 5 ppm;
FD&C Blue No. 2 and FD&C Red No. 40: 10 ppm; FD&C Blue No. 2,
FD&C Red No. 3, FD&C Yellow No. 5, and FD&C Yellow No. 6: 20 ppm.

48

unfavorable polarization kinetics. The emission specn'a of the remaining three dyes
showed great promise. Figure 3.4 shows the fluorescence spectrum of FD&C Red No. 40
in differing sucrose solutions of 0, 10, 30, 50, and 65 wt % concentrations while Figure
3.5 shows the same for FD&C Green No. 3 in sucrose solutions of 30, 50, and 65 wt%
concentrations. The spectrum for FD&C Blue No. 1 showed a similar response as the
FD&C Red No. 40. It is interesting to note that the fluorescence of these dyes in pure
water is close to zero but increases dramatically in the presence of the solute. This increase
is explained on the basis of the fact that fluorescence is a mechanism for the relaxation of
the energy absorbed by the molecules when exposed to an incident light. However, in an
aqueous solution, the interactions due to random motion provide a path of lesser resistance
for the relaxation to occur than fluorescence emission does. So, the amount of energy that
manifests itself in the form of fluorescence is low. However, when the amount of solute in
solution increases, there is less mobility of the dye molecules and the energy relaxation
occurs more readily by fluorescence. Thus, the increase in solute concentration is
responsible for the large increase in fluorescence.

The photochemisuy of food colors and the kinetic diagram showing the emission
properties is unknown. However, by studying these diagrams for some dyes whose
photochemistry is known completely, like pyranine (Figure 2.1, Figure 2.2), it is possible
to make some generalizations. In solutions, the dyes occur in the form of dissociated
anions. The relative amounts of the anion and cation, which is affected by the amount of
solute in solution, gives rise to the change in the emission peaks. In the case of pyranine,
the only dissociated cation that can affect the equilibrium is the proton (the dissociated
sodium ions do not play a role toward the application being considered); while in the case
of the food dyes, it is very likely that there are a number of species in equilibrium in
solution. So, while the relative protonation or deprotonation of pyranine gives rise to the
two emission peaks; in the food dyes, the combined effect of the emission of all ionic

species is responsible for the spectrum that is observed. Thus, theoretically, the broad

49

 

 

 

 

 

7 105 l l r 7 l T
5 65% sucrose
610 — _
8 5 105 - -
m
\
(n
E 5
3 4 IO - -
O
U
' 5
5x 3 IO - -
'5
c
B s
E 2 IO -
1 105 - -
Q520 - 560 600 640 680 72— 760 800
wavelength, nm
Figure 3.4 Emission spectra of FD&C Red No. 40 in sucrose solutions of different

concentrations (expressed as wt%). All spectra recorded at room temperature

(25 °C). Excitation wavelength: 500 nm.

50

 

2106 r r r r r r

 

—e— 50% sucrose

+ 30% sucrose
—0— 65% sucrose

    

1.5106

 

 

 

 

1106

Intensity, counts/s

510s

 

 

 

 

O

450 500 550 600 650 700 750 800
wavelength, nm

Figure 3.4 Emission spectra of FD&C Green No. 3 in sucrose solutions of different
concentrations (expressed as wt%). All spectra recorded at room

temperature (25 °C). Excitation wavelength: 420 nm.

51

peaks in the case of these food colors can be resolved into an appropriate number of peaks
corresponding to the number of species in solution once the solution equilibrium is known.
However, for the purpose and application in mind, the information available at this time is
sufficient to construct a complete probe assembly for measuring concentration and
supersaturation of solutions.

Figures 3.6 and 3.7 show the calibration curves obtained for the green and red dyes
respectively. In the case of FD&C Green No. 3, the normalization parameter chosen, the

Peak Intensity Ratio (PIR), is defined as

Peak Intensity at 662 nm

. (3.1)
Peak Intensrty at 515 nm

 

Peak Intensity Ratio =

and is plotted against the sucrose concentration in Figure 3.6. For FD&C Red No. 40, the
maximum intensity of each spectrum is divided by the maximum intensity of the spectrum
at 0 wt% sucrose to give the Peak Intensity Ratio. The calibration curve so generated is
shown in Figure 3.7. This calibration curve is also fit to an exponential curve with the
parameters shown in the graph. A strong correlation coefficient is obtained that shows that
this equation can very well be used for extrapolating the curve.

In both cases, a very strong non-linear dependence of the PIR on solution
concentration is apparent. This dependence extends well into the highly concentrated
region (above 65%) which is the critical region in crystallization experiments. In this
region, the non-linearity of the calibration curve makes this technique especially sensitive,
since small changes in solute concentration will be reflected by large changes in the PIR.
Compare this to a conventional technique for measuring concentrations in a supersaturated
solution such as refractometry, where changes are recorded only in the fourth or higher
decimal place of the refractive index. This relatively large gain in signal demonstrates that
the fluorescent probe technique is superior to conventional techniques in terms of

sensitivity.

Figure 3.6

 

 

 

 

6 . . . .
S- .
-.‘-3- 4- .
(U
o:
3‘
'g 3r .
Q)
E
7*. 2~ ~
0.)
C].
1+ -
O r r r r
20 30 4O 50 60 70

sucrose concentration, wt%

Calibration curve showing the relation between the Peak Intensity Ratio
(defined as the ratio of the peak intensities at 662 nm to that at 515 nm) and

sucrose concentration for FD&C Green No. 3.

Figure 3.7

53

 

 

25 I I I I I I
o
20 - Y=aebx -
a=1.103€
b=0.045
2
15 __ R =O.99 _

 

 

 

Normalized PIR

 

 

 

O l l l l 1 AL

0 10 20 30 40 50 60 70

sucrose concentration, (%w/w)

Calibration curve showing the relation between the Peak Intensity Ratio
(defined as the ratio of the peak intensities at 587 nm in each sample to the
peak intensity at the same wavelength in 0% sucrose solution) and sucrose
concentration for FD&C Red No. 40. Inset shows the parameters for an

exponential f it to the curve.

54

3.3.2 Results with the immobilized dyes

In the work described so far, the probe was dissolved in water and was present
freely in solution during the crystallization process. For use as an analytical sensor, it is
desirable that the probe be attached to a surface to avoid concerns of contamination of the
crystals by the probe. If a probe is immobilized on a surface, an immersion probe can be
constructed that can be dipped into a crystallizer for in situ monitoring of concentration
during the crystallization process. Also, in an industrial environment, remote sensing of
supersaturation would be the preferred method of measurement. It is therefore necessary to
immobilize the probe in a way such that its emission properties are preserved and at the
same time a fiber optic arrangement be used for remote sensing. Direct coupling to the
fiber optic, although a possibility, must be executed in a judicious manner in that it is likely
to change the photochemical behavior of the molecule. A better approach is to immobilize
the dye molecules on to a membrane and use the fiber cptics to collect the emission off the
probes taking care that the membrane material chosen is rigid and stable to biological and
thermal degradation.

In this work, we report on immobilizing the probes onto an ion exchange
membrane. A bifurcated fiber optic was used that served a dual purpose by shining the
incident light and collecting the emitted light from the probe. The complete sensor was

directly dipped into a cyrstallizer. This arrangement is shown schematically in Figure 3.2.

An ion exchanger is defined as a substance that is capable of reversible exchanges
of its constitutional ions with some other external ionic species bearing the same charge
[10]. All ion exchangers have a fundamental matrix that is fixed and chemically insensitive
to the surrounding electrolytes. The hydrophilic site of exchange containing ionizable
functional groups and mobile counterions is attached to this inert matrix. These
counterions can be replaced when brought into contact with a solution containing ions of

suitable charge and size.

55

Ion exchange onto a cellulosic anion exchanger is used for immobilizing the probe
as all the FD&C dyes form anions in solution. Ion exchange celluloses are aggregates of
glucosidic chains held together by interchain bonding, which provides dimensional stability
to the matrix. Cellulosic exchangers can be fibrous or rnicrogranular and are widely used
in the purification of biological macromolecules. The rnicrogranular variety is well suited
to the study of organic molecules [10]. Some of the common commercially available
cellulosic ion exchangers include AE cellulose, DEAE cellulose, TEAE cellulose, GE
cellulose, PAB cellulose etc. Here, we use the NA45 DEAE cellulose acetate membrane
which serves to immobilize all the FD&C dyes remarkably well without negatively
affecting their emission properties. There are, however, some variations in the intensities
obtained to that with the fiee dye in solution. This variation is attributed to the
immobilization procedure that is known to cause some enhancement of the signal while also
causing a significant increase in the baseline intensity. Furthermore, the presence of the
silica fiber optics also causes an increase in the baseline that reflects in a slight loss of
sensitivity. In spite of these effects, the immobilized probe shows good performance
characteristics.

Figure 3.8 shows the emission response obtained from FD&C Blue No. 1 that is
immobilized onto the anion exchange membrane. As is evident, there is a clear correlation
between sucrose concentration and the emission spectrum of FD&C Blue No. 1. As an
internal standard, we have defined a parameter called the Peak Intensity Ratio that is
defined as the ratio of the intensities at wavelengths of 664 nm and 615 nm. Figure 3.9
shows the dependence of PIR on sucrose concentration. As was the case with the fiee
dyes in solution, a marked non-linear change is observed in the PIR with sucrose
concentrations over a wide range. This calibration curve can thus be used as an analytical
probe for crystallizing sucrose solutions where the concentration is continuously changing.
Work is currently in progressin developing immersible probes with the other FD&C

5

colors.

56

 

 

 

 

 

 

 

 

 

s 1 1
7 10 h I I U U I’ I I 'fi 1 V V U i I U T j I I V 1' M ii T ‘
e '_ J.
6 10 “b .
(D 6d,. :-
a S 10 r .
t" l
C P
3 l
o _ 1
U 6
- 4 10 " "
>‘ 4
.t.’ . .
(D
C ’ . .
3 5 ’ —x— 10% sucrose \ ‘
E 3 10 d: \ “l"
. —°—40% sucrose
5 ' —I— 50% sucrose L
2 10 ‘_ f
» —Ir- 65% sucrose .
,_ J
6 P r 1 ‘
1 10 A A A A A A A A A I A A A A J A A A4 '4 AL A A A A A A A
600 650 700 750

wavelength, nm

Figure 3.8 Emission spectra of FD&C Blue No.1 in sucrose solutions of different
concentrations (eXpressed as wt%). The dye molecules are immobilized onto
an anion exchange membrane. All spectra recorded at room temperature.

Excitation wavelength: 400 nm.

57

 

l l l l I J
4.8“ ~
.9 4‘ ”
4.:
m
o:
5‘
'17, 3.2“" '
c:
a)
H
E
x 2.4“ ”
(0
en
0.
1.6“ "
I I I I I I

 

 

 

0 10 20 30 40 SO 60 70
sucrose concentration, wt%

Figure 3.9 Calibration curve showing the relation between the Peak Intensity Ratio
(defined as the ratio of the peak intensities at 664 nm to that at 615 nm) and

sucrose concentration for the immobilized FD&C Blue No.1.

58

3.4 Conclusions

The technical feasibility of a fluorescence probe sensor for the measurement of
concentration and supersaturation in crystallizing solutions using food grade colors has
been established. The use of fluorescence permits development of a sensor that directly
measures a property (Peak Intensity Ratio) that is fundamentally related to the activity rather
than the concentration of the solute. The major advantage of this technique is that the food
colors are already present during the manufacture and processing of the foods, and so can
be used in situ. Besides, the inherent non-toxicity of these compounds is a major
advantage over more conventional probes, most of which would not be approved for use in
food products. There are no other known commercial devices on the market based on this
concept. Additionally, the use of an optically based system provides many advantages
including no interference from electrical noise and low cost resulting from recent

developments in fiber optics.

3.5 References

1. D. M. Marmion, In: Handbook of U .S. Colorants. 3rd. Ed. John Wiley & Sons,
Inc. New York. pp 84, (1991).

2. J. D. Dziezak, Food Technology, Applications of Food Colorants. 41(4), 78-88,
(1987).

3. P. Markakis, Food Colors. In: Encyclopedia of Physical Science and Technology.
Vol. 5. 502-517. (1987).

4. R. Chakraborty, K. A. Berglund, Journal of Crystal Growth, Steady State
Fluorescence Spectroscopy of Pyranine as a Trace Extrinsic Probe to Study
Structure in Aqueous Sugar Solutions. 125, 81-96, (1992).

5. R. Chakraborty, K. A. Berglund, AIChE Symposium Series No. 284, The Use of
Pyranine as a Trace Fluorescent probe to Study Strucrure in Aqueous Sucrose
Solutions. 83, 114-123, (1991).

10.

11.

59

S. K. Yedur and K.A. Berglund, Use of Fluorescence Spectrosc0py in
Concentration and Supersaturation Measurements in Citric Acid Solutions.
Accepted for publication in Applied Spectroscopy. (1996).

H. Kondo, I. Miwa, and J. Sunamoto, Journal. Phys. Chem., Biphasic Structure
Model for Reversed Micelles. Depressed Acid Dissociation of Excited-State
pyranine in the Restricted Reaction Field. 86, 4826-4831, (1982).

H. Stapelfelt, H. Jun, and L. H. Skibsted, Food Chemistry, Fluorescence
properties of carrninic acid in relation to aggregation, complex formation and
oxygen activation in aqueous food models. 48, 1-1 1, (1993).

J. P. Rasimas and G. J. Blanchard, J. Phys. Chem, Understanding the Electronic
Properties of Glycosylated Chromophores Using AMI Semiempirical Calculations.
98:49, 12949-12957, (1994).

P. R. Brown and A. M. Krstulovic, In: Separation and Purification. 3rd Edition.
Eds. Edmond Perry and Arnold Weissberger. Chapter IV, John Wiley & Sons,
New York 197-255, (1978).

Certified Food Colors. Published by: Warner Jenkinson Company, Inc. St. Louis,
MO, U.S.A. (1995).

Chapter 4

A SOL-GEL BASED FLUORESCENCE SENSOR FOR
CONCENTRATION MEASUREMENTS IN SUCROSE

SOLUTIONS

4.1 Background

An ideal optical sensor should have the capability to detect fluctuations in analyte
concentration continuously; that is, the optical probe should interact with the analyte in a
reversible manner for long periods of time. The development of the sol-gel process has
enabled the construction of novel materials that are well suited to optical sensing and
various detection systems [1, 2]. In addition to providing transparent and porous
materials, the sol- gel technique offers a number of modifying techniques that can be used to
incorporate molecular probes into the polymer matrix without any loss of optical properties
[3]. The primary focus of this investigation was to study the feasibility of using sol-gel
derived thin films for sensory applications in aqueous crystallizing solutions, where
typically the concentrations are very high. Conventional techniques of measuring
concentration and supersaturation in crystallization operations like refractometry, turbidity
measurement etc., involve the use of mass concentration measurements that lack the
Sensitivity required for optimum process control. More accurate measurements can be
made by the use of the fluorescent probe technique [4]. In this work, we report on the use

of the fluorescent properties of a chlorin incorporated in a sol-gel thin film to measure the

60

61

concentration of aqueous sucrose solutions. Sucrose is chosen as the test analyte because
of its high solubility, and because of the fact that more sugar is crystallized in the world
today than any other chemical.

Under controlled conditions, solution precursors of sol-gel systems can be
prevented from forming gels. Stable solutions thus obtained can be used to make thin films
supported on various substrates using dipping, spin casting, or deposition techniques.
These thin films retain the structural integrity of porous sol-gel glasses along with high
internal surface area and favorable optical properties. Thin films that are porous, optically
transparent, mechanically stable and resistant to aqueous and organic solvents were
produced by controlled hydrolysis of titanium (IV) isopropoxide using valeric acid as the
chelating ligand. Ethanol was used for controlling the hydrolysis which is the key to make
thin sol-gel films [5]. Spin casting was used exclusively to produce the films while dip
coating was used to incorporate the probe into the matrix. The probe used was a member
of the chlorin family - octaethyl chlorin diol. The chlorins are related to the naturally
occurring porphyrin group of molecules and are inherently non-toxic. Thus the test strip
that was used as the sensor for measurement of concentrations in sucrose solutions was

composed of the chlorin entrapped within the sol-gel film.

4.2 Materials and Methods

The specific chlorin used as the molecular optical probe was octa ethyl chlorin diol
that was synthesized and provided by Chang [6]. The structure of the molecule is shown

in Figure 4.1.

Titanium(IV)isopropoxide and valeric acid were obtained from Aldrich Chemical
Company and used without further purification. Absolute ethanol was obtained from the
Quantum Chemical Corporation. VWR microscope slides that were cut to fit 1 x 1 cm
Cuvettes were used as substrates for films. An International Clinical Centrifuge (model CL

26802 M) was used for spin-casting films. Fluorescence spectra were taken with a SPEX

Figure 4.1

62

 

Chemical structrue of Octaethyl chlorin diol.

 

I. ..,,
r
‘hl'fii-

63

FLUOROLOG spectrophotometer that was attached to PMT detector and a computer for
recording and processing the data.
4.2.1 Film preparation

In a scintillation vial, 0.75 milliliters of valeric acid was added to 0.25 milliliters of
titanium isopropoxide. Twenty microliters of distilled water was added to this mixture
which was thoroughly mixed to give the film solution. Fifteen microliters of the solution
was placed on to a clean glass slide. This was attached to the arm of the centrifuge and
spun for 10 minutes. The sol-gel film thus formed on the slide was placed in a dust free
chamber. The films were calcined in an oven at 250 'C for one hour primarily to drive off
excess valeric acid that possesses an unpleasant odor. Calcination was started within one
hour of the preparation of the film. This protocol gave films that were smooth, uncracked,

and of uniform thickness.

4.2.2 Entrapment of the chlorin

Numerous screening experiments were conducted to determine the optimum
concentrations to be used as well as the experimental protocol. As a result, the chlorin
solution was prepared by dissolving 1 mg of octaethyl chlorin diol in 100 microliters of
dichloromethane in a vial and mixing well. Four milliliters of 200 proof ethanol was added
to the solution with constant stirring. Since chlorin is sensitive to ambient light, the vials
were kept wrapped in aluminum foil.

The entrapment of the chlorin within the matrix was achieved by dip coating the
titanium carboxylate thin film in the chlorin solution. The calcined films were dipped into
the chlorin solutions two at a time for 15 minutes. This time duration was sufficient for the
chlorin from the film solution to migrate into the sol-gel film. The resulting film was then
rinsed with water and air dried with compressed air. All films were wrapped in aluminum
foil and stored in a dust free environment.

The performance of the chlorin entrapped within the polymer matrix was evaluated

using emission spectroscopy. Cast films supported on cut slides were placed vertically

64 .

inside quartz cuvettes. The glass slide with the film was placed against one wall of the
cuvette such that the film was facing the incident light. Instead of the conventional right
angle arrangement for recording emission, the cuvette was placed in the spectrophorometer
cell compartment with a front face configuration (22 degree angle to incident light) to avoid
stray incident light entering the detector. An excitation wavelength of 400 nm was used for

the chlorin while emission was recorded from 600 to 750 nm.

4.3 Results and Discussion

4.3.1 Sol-gel films

After preparation using the protocol previously described, the films were calcined
in an oven at 250 'C for one hour primarily to drive off excess valeric acid that possesses
an unpleasant odor. The calcination also causes the transformation of the nature of the film

from titanium carboxylate to a titanium oxide film [7].
4.3.2 Free chlorin in aqueous solutions

Figure 4.2 shows the emission spectra of free octaethyl chlorin diol in aqueous
sucrose solutions of 10, 30, and 50 wt %. As is evident, there is a marked change in the
emission behavior of chlorin with an increase in solute concentration. The challenge was to
obtain similar changes in the emission spectra from chlorin after it had been entrapped

within the sol-gel matrix.

4.3.3 Entrapment of the chlorin

Guest molecules can be entrapped within the porous matrix of sol-gel derived thin
films either by dissolving them in the film solution prior to spin casting or by dip coating
[8, 9]. This procedure requires the chlorin to be sufficiently soluble in a solution such that
a functional amount of chlorin is available within the thin film to react with the target
analyte. The dip coating technique used in this work provided films that were tested as a

sensor with success.

65

 

  
   

 

 

 

 

 

‘l 107 r r r r r I r r 1' r r r I 1 r r I I 1 I r r r r r r 1 r

6. ——-—10% sucrose -

8 10 - +30% sucrose —

- +50% sucrose -

' l'

‘8 6106- _
g _ -
U) 6‘ ..
5 410 — —
E - .
2106- _

O I'-'-""'TI""""'I""""'_
600 650 700 750

 

 

Wavelength (nm)

Figure 4.2 Emission spectrum of free chlorin in solution in different sucrose solutions.

All spectra collected at room temperature.

66

Although chlorins demonstrate a high degree of thennodynamic stability, there are
some instability phenomenon associated with them. These are photodecomposition [10]
and instability due to aggregation [11, 12, 13]. The first problem was taken care of by
ensuring that the film and the precursor solutions were always stored in the dark. The
second problem was taken care of by using controlled hydrolysis.

The role of controlled hydrolysis to produce sol-gel derived thin films and the
important role played by modifying reagents has been reported [5]. Reagents such as
alcohol can improve important properties of the film such as optical transparency, thermal
stability, resistance to solvents etc. In the sol-gel process, alcohols are often used as the
solvent. However, unlike most other solvents that are chemically inert, alcohols take part
in reactions that take place during the sol-gel process. Therefore, alcohols are very
effective in controlling the final structural properties of sol-gel derived materials. As a
modifying reagent, ethanol, in addition to being a dispersive agent, can also participate in
alcoholysis in sol-gel systems. This two fold effectiveness of ethanol renders it an ideal
candidate for stabilization of the film.

We have demonstrated that chlorins can be entrapped within the porous matrix of
sol-gel derived thin films by immersing the sol-gel film in a chlorin solution. In effect, we
have a sensor made up of a fluorescent chlorin, which is entrapped within a titanium matrix
that itself is bound to glass. The film is stable in aqueous solutions. The fluorescence
intensity of the chlorin changes with respect to the solvent microenvironment in which it is
placed. This effect is shown in Figure 4.3 which shows the changes in the emission
spectra of chlorin in different aqueous sucrose solutions. This figure shows that the
entrapped chlorin is still optically active and is available for sensing sucrose. A comparison
with Figure 4.2 shows that while the wavelength’s match quite well, the intensities of the
peaks are quite different from that of the spectra of the free chlorin. The change in the
intensities is to be expected because of the immobilized nature of the chlorin. The

immobilization causes formation of secondary bonds within the sol-gel matrix. This

67

 

6
610 llIlIllIlIllllllllllIIIIIIIII

 

5106

+ 10% sucrose
+30% sucrose
—+—50°/o sucrose
—o—65% sucrose

 

 

 

4106

3106

Intensity (cps)

2106

I l I I l l I I I I l I l l I I 1 l l I I l

IIIIIIITIIII'IIIIIIITr

  

 

 

1106 Illllllllllllllllllllllllllll

600 650 700 750

 

Wavelength (nm)

Figure 4.3 Changes in emission spectra of chlorin embedded in a titanium carboxylate
thin film in various aqueous sucrose solutions; all spectra collected at room

temperature.

68

chemistry causes changes in the photochemical behavior of the emitting species [14] and
explains the observed change in the intensities of the emission bands of the chlorin in free
solution and in the immobilized state.

Figure 4.3 also demonstrates the effect of sucrose on the emission spectrum of the
entrapped chlorin. Even though the changes observed due to the sucrose is not too great in
absolute numbers, the change observed is quite adequate for the measurement of the
concentration in sucrose solutions. Also, as the figme shows, a change is seen in~ the
spectrum with sucrose concentration through the ranges of high concentration where
crystallization is expected. The potential of this sensor to be used in a crystallizer, where
high concentrations of the solute are to be expected, has been established. A calibration
curve can be created using the information in Figure 4.3 relating a normalized parameter
from the emission spectrum. As an internal standard, the ratio of the peak intensities at two
wavelengths is calculated that takes care of variations in the absolute intensity numbers. In
this case, the ratio is taken of the intensities of the two peaks at 646 nm and at 673 nm.
This peak intensity ratio is the parameter that is plotted against the sucrose concentration to
generate the calibration curve that is shown in Figure 4.4. A linear curve fit gives an R2
value of 0.995. Thus, the concentration of an unknown concentration of an aqueous
sucrose solution can be determined using this calibration curve.

Since the intensities of the two peaks are dependent upon the relative amounts of
water present in the vicinity of the probe molecules, the relative intensity of the two
emission peaks of chlorin can be used effectively as a sensor for measurement of the
concentration of solute in a crystallizing solution. Thus, this technique of combining sol-
gel technology with fluorescence spectroscopy can be used as a sensor to measure the

concentrations of sucrose solutions.

69

 

 

 

 

2.1
2.05 r-
.9
a 2 ~
3‘
a
g 1.95 -
E
it 1.9 -
Q)
Q.
1.85 :—
1.8 r r r r r .r

 

0 10 20 30 40 SO 60 70
sucrose, wt %

Figure 4.4 Calibration curve showing a linear relationship between the Peak Intensity
Ratio (defined as the ratio of the intensities of the peaks at 646 nm and 673

nm) and sucrose concentration.

70

4.4 References

1. a. D. D. Dunuwila, B. A. Torgerson, C. K. Chang, and K. A. Berglund, Anal.
Chem, 66, 2739, (1994).

b. K. G. Severin, J. S. Ledford, B. A. Torgerson, and K. A. Berglund, Chem.
Mater., 6, 890, (1994).

2. a. H. Reuter, Adv. Mater., 3:5, 258, (1991).

b. J. Livage, M. Henry, and C. Sanchez, Progress in Solid State Chemistry, 18,
259, (1988).

c. H. Zheng, M. W. Colby, and J. D. Mackenzie, Mat. Res. Soc. Symp. Proc.,
121, 537, (1988).

d. H. Schmidt, G. Rinn, R. Nab, and D. Sporn, Mat. Res. Soc. Symp. Proc.,
121, 743, (1988).

e. F. Babonneau, A. Leausric, and J. Livage, Mat. Res. Soc. Symp. Proc., 121,
317, (1988).

f. S. M. Melpolder and B. K. Coltrain, Mat. Res. Soc. Symp. Proc., 121, 811,
(1988).

3. a. L. M. Ellerby, C. R. Nishida, F. Nishida, S. A. Yamanaka, B. Dunn, J.
Selverstone-Valentine, and J. I. Zink, Science, 255, 1113, (1992).

b. A. Slama-Schwok, M. Ottolenghi, and D. Avnir, Nature, 355, 240, (1992).

c. I. Kuselman, B. I. Kuyavskaya, and O. Lev, Anal. Chim. Acta., 256, 65,
(1992).

d. A. Schowk, D. Avnir, and M. Ottolenghi, J. Am. Chem. Soc., 113, 3984,
(1991).

e. T. Haruvy and S. E. Webber, Chem. Mater., 3, 501, (1991).

f. R. Zusman, C. Rottman, M. Ottolenghi, and D. Avnir, J. Non-Cryst. Solids, ,
122, 107, (1990).

g. B. Dunn, E. Knobbe, J. M. McKieman, J. C. Pouxviel, and J. I. Zink, Mat.
Res. Soc. Symp. Proc., 121, 331, (1988).

h. D. Avnir, D. Levy, Reisfield, J. Phys. Chem., 88, 5956, (1984).

3“

10.

ll.
12.

13.

14.

71

a. R. Chakraborty and K. A. Berglund, AIChE Symposium Series no. 284., 83,
114, (1991).

b. R. Chakraborty and K. A. Berglund, J. Crystal Growth, 125, 81, (1992).

c. S. K. Yedur and K. A. Berglund, Accepted for publication in: Appl. Spec.,
(1996).

D. D. Dunuwila, C. D. Gagliardi, and K. A. Berglund, Chem. Mater., 6, 1556,
(1994).

C. K. Chang, Department of Chemistry, Michigan State University, personal
communication, (1993).

M. J. Waner, Characterization of the Sol-Gel and MOCVD Processes for the
Deposition of Aluminum Oxide Thin Films. MS Thesis. Michigan State
University, (1994).

R. B. Lessard, M. M. Wallace, W. A. Oertling, C. K. Chang, K. A. Berglund,
and D. G. Nocera, Mat. Res. Soc. Symp. Ser., 155, 109, (1989).

C. D. Gagliardi, D. D. Dunuwila, C. K. Chang, and K.A. Berglund, Mat. Res.
Soc. Symp. Proc., 271, 645, (1992).

J. W. Buchler, In: The Porphyrins: Structure and Synthesis, Part A, Dolphin, D.,
Ed., Academic Press: New York, NY, Vol. 1, ch. 10, (1978).

A. E. Alexander, J. Chem. Soc., 1813, (1937).

J. A. Bergeron, J. G. L. Gaines, and W. D. Bellamy, J. Colloid Interface Sci .,
25, 97, (1967).

W. 1. White, In: The Porphyrins: Physical Chemistry, Part C, Dolphin, D., Ed.,
Academic Press, New York, NY. Vol. V, ch. 7. (1978).

M. Bacci, F. Baldini, and S. Bracci, Appl. Spec., 45, 1508, (1991).

Chapter 5

 

ON THE APPLICATION OF FLUORESCENCE
SPECTROSCOPY TO MEASURE THE
DESUPERSATURATION OF CITRIC ACID SOLUTIONS

5. 1 Background

Desupersaturation studies are usually isothermal experiments conducted on seeded
or unseeded supersaturated solutions to study the effects of various parameters such as
seed size, seed surface area (or number of seeds), agitation, and supersaturation on the
crystallization process. The desupersaturation studies also indicate the induction time and
can give quantitative information about the growth and nucleation rates. The changes in the
solution that occur during crystallization and affect the final product are of considerable
interest for the design and control of a cryStallizer.

Improved crystal size distribution is one of the most important concerns in any
crystallization operation. Maintaining a constant supersaturation is recognized as the key
for obtaining improved CSD’s in batch crystallizers. In its most fundamental form,
supersaturation is defined as the difference between the chemical potential of the solute on
the crystal surface and that in solution. However, in all measurement techniques currently
used, such as refractometry, supersaturation is approximated by the concentration
difference of the solute in the supersaturated solution and its equilibrium solubility or by a
ratio of the two concentrations. This approximation involves an underlying assumption of

ideality that is strictly valid only for very dilute solutions [1]. For concentrated solutions,

72

73

such as those that might be expected in crystallization Operations, this approximation can
lead to serious errors in the measurement of supersaturation because of the high degree of
nonideality of the solution [2]. A detailed discussion of these approximations and the
errors caused by their use was presented in Chapter 2.

The nonideal behavior of a solution is reflected in the activity coefficient 7. The
activity coefficient is strongly affected by the extensive intermolecular and intramolecular
interactions between the solute and the solvent that define the solution structure of the
whole system. It follows then, that the aetivity coefficient is strongly dependent on the
solution structure at a molecular level. Thus, a technique for measuring the supersaturation
that is reflective of the solution structure would be inherently more accurate than
conventional systems that measure bulk properties that are not sensitive to the solution
structure at a molecular level.

A novel analytical tool - the fluorescence probe technique - has been developed in
our laboratory utilizing fluorescence spectroscopy to measure concentration and
supersaturation of crystallizing solutions. This technology has been demonstrated to work
exceedingly well for crystallizing solutions of various sugars and also of citric acid [3, 4,
5]. Fluorescence specrroscopy is fundamentally reflective of the intermolecular interactions
and is thus strongly dependent on the solution structure. It has been shown that the
solvatochromic nature of the fluorescent probe molecule pyranine, whose structure is
shown in Figure 2.1, can be exploited to get a calibration curve correlating the solute
concentration to a peak intensity ratio that is obtained from collecting the emission spectrum
of the probe. In this work, we report on using the solvatochromic properties of pyranine to

study the desupersaturation of citric acid.

5.2 Materials and Methods

Reagent grade anhydrous citric acid was obtained from Aldrich Chemical Co. and

was used without further purification. The fluorescence probe Pyranine (chemically, the

74

trisodium salt of 8 - hydroxy, l, 3, 6 - pyrene trisulfonic acid), was obtained from
Lancaster, and was used as obtained. Distilled water was used in all experiments.

The desupersaturation experiments were carried out in a batch jacketed crystallizer
that had a working volume of IL. The temperature of the crystallizer was controlled by
circulating water through the jacket from a water bath whose temperature was controllable
to 0.1 'C. For each experiment, 300 mL of free solvent was used. To this, an appropriate
amount of anhydrous citric acid was added to generate a supersaturation S of 1.2. Here, 8
is defined as the ratio of the amount of solute in solution to the equilibrium solubility at any
given temperature. A constant stirrer speed of 400 rpm was used in all experiments. To get
an exact amount of probe in the crystallizer for each experiment, a stock solution was
prepared by dissolving 0.015 gm of pyranine in 50 mL of water. From this stock solution,
0.3 mL was carefully added to the crystallizer. This protocol generated a probe
concentration of 1.0e-05 M in the crystallizer which is an appropriate concentration for
fluorescence experiments. In the experiments where the probe concentration was varied, a
different amount of stock solution was added to the crystallizer to get the desired
concentration.

The fluorescence spectrum of the probe in solution was recorded with the aid of a
fiber optic cable that was introduced into the crystallizer through a port. The fiber optic
was enclosed within a glass tube that was sealed at one end. The fiber tip was placed flush)
against this sealed end and was able to pick up the fluorescence emitted by the probe
molecules through the glass barrier. For the pyranine molecules to emit light, they need to
be excited at a fixed wavelength. The excitation light was provided by a hand held ultra
violet light source marketed by Cole-Partner Inc. The lamp generated light that had a
primary wavelength of 365 nm. Since pyranine can be excited by light with a wavelength
around 350 nm, this lamp was appropriate for exciting the pyranine molecules. The fiber
optic was connected to a single channel spectrophotometer manufactured by Ocean Optics,

Inc. The Spectrophotometer comes equipped with a diode anay detector that is optimized to

75

collect any light that comes through the fiber optic from a range of 225 to about 800 nm.
The spectrophotometer itself was connected to a Pentium processor based computer that
was used for data analysis.

Desupersaturation experiments at consrant temperatures were conducted by first
dissolving the appropriate amounts of anhydrous citric acid in the 300 mL of water at a
slightly elevated temperature than the temperature of the experiment to generate a
supersaturated solutions. After complete dissolution, the temperature was held at least 10
degrees higher than the saturation temperature for about 15 minutes to ensure the
dissolution of all nuclei. The temperature was then brought down to the temperature of the
experiment. All experiments were conducted isotherrnally.

Seeds were introduced into the crystallizer at time zero at which time the recording
of the fluorescence specua of pyranine commenced. The spectra were recorded at an
interval of 1 minute for either ninety minutes or two hours. The seeds were of a monosized
distribution with an average length of 780 microns unless otherwise noted. At the end of
the experiment, the crystals that had grown in the crystallizer were filtered in a Biichner
funnel, washed with a cold saturated citric acid solution and then washed again with ether
toremove all water. For experiments at 25 'C, the effect of the amount of seed used was
studied by using 3 gms, 15 gms, and 34 gms of seeds. In another series of experiments,
desupersaturation was conducred at temperatures of 20 °C, 25 ’C, and 30 'C to study the
effect of temperature. In these latter experiments, a constant seed weight of 3 gms was
used. To establish the effect of the probe concentration on the desupersaturation curve,
identical desupersaturation experiments were conducted with three probe concentrations,

1e-05 M, 5e-06 M, and 1e-O6 M.

5.3 Results and Discussion

The emission properties of pyranine have been well characterized [6]. In aqueous

solution, pyranine exists with the sulfonate groups dissociated. Upon excitation with light

 

76

at wavelength around 350 nm, it has two excited states. One, emitting at around 440 nm,
is due to the protonated form of the molecule, whereas the other, emitting at around 510
nm, is due to the deprotonated form. Whether the observed emission is from the
protonated form or from the deprotonated form of pyranine depends on the molecular
environment in its immediate vicinity. When surrounded completely by water molecules,
pyranine exists solely in the depmtonated state, as water causes complete dissociation. The
emission appears only at 510 nm. As the amount of solute in solution increases, the
dissociated form starts getting protonated, and the emission spectra shows a shift in the
peaks. Thepeak at 510 nm decreases, whereas the peak at 440 nm, which is due to the
protonated form, increases. These observations are illustrated in Figure 2.3. Thus, solute-
solvent interactions in solution can be studied by observing the relative changes in the
emission peaks of pyranine.

The emission spectra in Figure 2.3 show that over the range from 0% citric acid to
well above saturation, the two emission peaks change a lot. However, in an actual
crystallization operation, the solution has to remain supersaturated. As soon as the solution
concentration falls to saturation, crystallization is complete. Thus, under crystallization
conditions, the relative change in emission peaks will not be as dramatic as is apparent in
Figure 2.3. In fact, the peak at 510 nm, caused by the deprotonated state of pyranine, will
hardly be visible as the solution concentration is very high. Although present, it will be
dwarfed by the emission caused by the protonated form of pyranine that fluoresces at
around 440 nm. As mentioned earlier, the use of fiber optics can cause a bathochromic
shift in emission peaks. This is borne out in Figure 5.1 which shows the spectra obtained
from a desupersaturation experiment. The emission peak appears at 449 nm. The
experiment was carried out under the following conditions. Temperature = 25 °C, initial
supersaturation (c/c*) = 1.2, agitator speed = 400 rpm, probe (pyranine) concentration =
1.0e-05 M, seeds (average size 660 microns) = 3 gms. The excitation light for the

pyranine molecules was provided from a UV light source that had a primary output

77

 

2000 IUIIITIIU:UUIIIIIII:IIIIIUIIU

       
  

t=O minute i

I 500--

\

t=90 minutes \

1 000‘

500; i j-

 

Emission Intensity, counts/s

 

O AAAAAAlLllLlllllllJlJllllAll.

400 450 500 550
wavelength, nm

Figure 5.1 Emission spectra of pyranine over the course of a seeded isothermal (25 'C)
citric acid batch crystallization. Ten spectra are shown here, each at an
interval of 10 minutes from t = 0 minute to t=90 minutes. These ten are

culled from the ninety spectra collected at intervals of 1 minute.

78

wavelength of 365 nm. The emission spectra was collected over the whole range of the
detector, i.e., from 225 to 800 nm. The spectrum was collected every minute for 90
minutes at which time the experiment was deemed complete. In this figure, only ten
spectra at intervals of ten minutes each are shown for clarity. Note that a number of
spectra, especially at the end of the experiment, fall on top of each other. There is an
intense peak at 449 nm, but the peak at 510 nm is almost buried in the tail of the other.
However, from a knowledge of the excited state kinetics of pyranine, we can infer that the
peak at 510 nm, albeit extremely small, is present.

The absolute intensity of the emission peaks is influenced, among other things, by
the excitation light intensity and the sensitivity of the spectrophotometer. To compare
results of different experiments, it is more convenient to define and use a parameter that is
not affected by variations in the instrument used. As an internal standard, then, we take a
ratio of the two most intense peaks of pyranine and use that parameter, defined as the Peak
Intensity Ratio (PIR), as the parameter that changes with time. In the experiments
described, the PIR was defined as the ratio of the peak at 449 nm to that of the peak at 510
nm.

5.3.1 The desupersaturation curve

From each of the spectrum obtained, a ratio of the peaks at 449 and 510 nm was
calculated. This ratio was plotted against the time of the experiment and the result is shown
in Figure 5.2. The shape of the curve matches a conventional desupersaturation curve
remarkably well. In the initial, relatively flat part of the curve, the supersaturation
decreases very slowly. In this region, the desupersaturation is dominated by the growth of
the seed crystals. There is also a possibility of secondary nucleation, but this is not
detectable in this initial portion. After an induction time, ‘t, secondary nucleation causes the
surface area of the crystals to reach a critical size that is capable of consuming all the excess
mass in solution very rapidly. This phenomenon is reflected in a sudden increase in the

desupersaturation rate. This increased desupersaturation rate continues to deplete all the

79

 

w
—|
UT

1

.l
.1—
n—th—

.l
cut-—

.4

ul

‘1

.l

to
L.
l

2.85:

o 3 (so 3
._ . o .
E 3.05 ‘: 0Q, i”
ii . 2 2
. o .
.4: 3 “f 00 T”
‘3 : °,, 1
G) - o .
+5 2.95 ‘: QDQ; '“
x : ‘32.
8 2.9 “f R: '"
o- ..

 

 

2.8 J l A % l 1 MI I A l l l l J l m l

40 60 80 100
time, min

 

O
N
0

Figure 5.2 Typical desupersaturation curve of a seeded citric acid batch crystallization
experiments. The experimental conditions were as follows: Temperature =
25 'C, initial supersaturation (c/c*) = 1.2, agitator speed 2 400 rpm, probe
(pyranine) concentration = 1.0e-05 M, seeds (average size 660 microns) = 3

gms, excitation wavelength = 365 nm.

80

supersaturation and the solution eventually stabilizes to its saturation value. This end point
is represented by the flat portion of the curve at the tail end. In this experiment, this point
is reached in about 90 minutes.

The comment that growth of the seeds dominates in the relative absence of
nucleation in the first flat part of the curve is borne out by observation, both visual and
electronic. Visually, it is easy to observe that during the initial growth phase, the number
of crystals (same as the number of seeds added) is almost constant and there is no
appreciable change in the turbidity of the solution in the crystallizer, but after the induction
period, a slew of nuclei appear, making the solution quite turbid. The cloudiness due to the
new crystals steadily increases till the equilibrium saturation is reached.

Electronically, this phenomenon of nucleation is reflected by the backscattering of
the incident light. The light from the UV lamp that is used to excite the probe molecules
also provides information on nucleation. Before seeding, when no crystals are present in
solution, none of this light is reflected, or ‘backscattered’ into the detector. However, after
seeding, this incident light is picked up by the detector as a small peak at a wavelength of
370 nm. Again, the small shift in this wavelength from 365 nm to 370 nm is due to the use
of the fiber optic that causes a small bathochromic shift. The intensity of this peak is
dependent only on the amount of backscattered light which is very strongly dependent on
the number of crystals in the crystallizer. Figure 5.3 shows a plot of the peak intensity at
370 nm as a function of time. Initially, the amount of backscattered light is relatively
constant. After some time, which corresponds exactly to the induction time in Figure 5.2,
the intensity increases rapidly. This increase in intensity is due to the formation of new
crystals by secondary nucleation. This process of secondary nucleation continues until the
supersaturation is totally depleted. Corresponding to this, the backscattering intensity also
increases and reaches a saturation value after which there is no increase in intensity,
indicating that nucleation is complete at this point. Thus, a comparison of Figures 5.2 and

5.3 serves to confirm the observation made that, in the desupersaturation experiment, there

81

 

 

 

800 L I I I : I T I '1 I I I } M fi I % I I I
. sides
" O
.. o O
1'; 700 “f o 5%“ “f
H l" (9 -
.— _ o O 4
s - . -
600 1' ° 1"
b _ CD
to C 0%
b 500 “j ‘b j"
._ L .
g : 00w 3
v 400 '1: o 1"
a?! : 0°
2 300 "_" 0 -~
3 ; (9"
E 200 -- . --
haw
1 00 A J l g 4 l l : l l l % l l l g l A L

 

0 20 40 60 80 100
time, min

Figure 5.3 Backscattered excitation light during the desupersaturation experiment as a
function of time. The experimental conditions are exactly the same as those

described in Figure 5.2.

82

is an initial period of growth of the seeds, during which there is no detectable nucleation.
After the induction period, secondary nucleation starts and the crystals, both old and new,
continue to grow until the supersaturation is completely depleted and the system comes to
equilibrium at the temperature of the experiment.

The combination of fluorescence and backscattering information that is
simultaneously collected here makes this technique especially powerful. The hardware that
makes this possible is the diode anay detector that is part of the spectrophotometer used.
The salient feature of this detector is that it is able to record the number of photons being
entering the fiber optic over its entire operating wavelength range of 225 nm to 800 nm
without discriminating between the backscattered incident light and the fluorescent light
emitted by the probe molecules. As the incident light is known to be of a particular
wavelength and the fluorescence is known to occur at longer wavelengths, it is easy to
distinguish between the backscattered incident light intensity and the peaks due to
fluorescence. Note that the backscattering information obtained does not require the
presence of the probe at all. It simply gives a measure of the number of crystals reflecting
the incident light back to the detector.

5.3.2 Effect of seed weight

Figure 5.4 shows the desupersaturation curves for citric acid obtained with different
seed weights of 3 gms, 15 gms, and 36 gms. All three curves were generated under
isothermal conditions of 25 'C, and the initial supersaturation of 1.2 was identical for each
case. Other experimental conditions, such as agitator speed (400 rpm), were also constant.
The seeds used were of a monodisperse size with an average length of 780 micrometers.
The seeds were generated by simply seiving a batch of anhydrous citric acid that was
obtained from the manufacturer and were used without any pretreatment. The probe
(pyranine) concentration was 1e-06 M for the runs with seed weights of 15 gms and 36
gms but was higher (1e-05 M) for the experiment with 3 gms of seeds. Although this is an

order of magnitude difference in the probe concentration, we have demonstrated that this

83

 

315' l JIIIIITTIIjrlTfiIJTIT
. r I r I I ,

  
   

 

° 3 gms S
U 15 gms
0 36 gms

l
I

 

 

AlAAAA

 

A

AL

AAJ

Normalized Peak Intensity Ratio

  

 

 

 

275 llllllJlllJllllllllllllllLJ
e I I I I I j I

0 20 40 60 80 100120140
time, min

Figure 5.4 Effect of seed amount on the desupersaturation curve of seeded citric acid
batch crystallization experiments. The ordinate are normalized so that all
curves srart at the same point, since initial supersaturation (c/c* = 1.2), is
the same in all three cases. Other experimental conditions: Temperature = 25
'C, agitator speed = 400 rpm, probe (pyranine) concentration = 1.0e-05 M,

seeds (average size 660 microns) = 3 gms, excitation wavelength = 365 nm.

84

does not cause any appreciable change in the nature of the desupersaturation curve except
for a very small increase in the overall intensity. This anomaly is corrected for by
normalizing all curves so as to have the same initial value since the initial value has to be the
same for all three cases. This normalization is discussed further in Section 5.3.4. Since
the average size of the seeds is constant, it follows that the different seed weights used
essentially translate to the different number of nuclei used.

The observed steep dr0p in supersaturation after a certain interval of time indicates
the point where secondary nucleation causes the surface area of crystals in the crystallizer to
exceed a critical value, at which point there is enough area available to rapidly
desupersaturate the whole system. This drop in supersaturation continues till the solution
comes near to its equilibrium saturation. A tail is observed in the desupersaturation curve at
which time the crystallization is deemed complete. This was confirmed in all experiments
by collecting all the crystals from the crystallizer at the end of the run. The dried crystals
were weighed to confirm that the amount of solute in solution that was in excess of the
solubility had indeed come out as crystals.

A standard protocol was used to obtain the final crystalline product. At the end of
the experiment, the crystals were separated from the mother liquor by passing the slurry
through a filter paper placed in a Biichner funnel that was attached to a water aspirator. The
crystal mass was further washed with a cold saturated solution of citric acid to remove the
minuscule amount of the probe that could possibly be sticking to the surface of the crystals.
This washing step was repeated at least two times. The cold saturated solutions serves also
to remove any mother liquor sticking to the crystals, while at the same time ensuring that
none of the crystals dissolve in the wash solvent. After washing with the saturated
solution, the crystals were washed with ether to remove any water that might be present.
The crystals were then air dried at room temperature. While this protocol ensured the
complete removal of solvent, it was not adequate to get dried crystals for CSD

measurements. The problem with this protocol was the air drying step which caused the

85

agglomeration of crystals. Room temperature (about 25 °C) is not enough to properly dry
the crystals. However, the use of high temperature (above 35 'C) is also not possible as
that would drive away the water of crystallization associated with the citric acid product. In
fact, prolonged exposure to room temperature, has been known to cause the same problem.
When that happens, fusion of crystals occur that renders the whole exercise of CSD
measurement futile. The possible solution to this problem is to dry the crystals in an oven
carefully controlled under 35 °C for as short a time as is necessary. Future experiments,
where an attempt will be made to predict the CSD using the data presented in this paper,
will involve using some such technique to solve this problem.
5.3.3 Effect of temperature

The effect of temperature on the desupersaturation experiments was studied by
conducting the experiment at temperatures of 20 °C, 25 °C, and 30 °C. In these three
experiments, all other parameters were kept constant. The results are presented in Figure
5.5 as a scatter plot. The initial supersaturation in each case was 1.2, which led us to
normalize the data at the initial point. The interesting observation here is the time taken for
the surface area of the crystals to reach the critical size at which rapid desupersaturation
takes place. At 20 °C, it takes almost an hour after seeding for the rapid drop to
commence, whereas it is correspondingly lesser at the higher temperatures. This decrease
in induction time with temperature can be explained by an inspection of the kinetic
parameters which are a function of the temperature. The nucleation rate is usually higher at
higher temperatures which causes nuclei to appear earlier. So, at the higher temperature,
the critical surface area would be achieved earlier, causing the desupersaturation rate to
increase earlier. This conclusion can be proven by obtaining the kinetic information from
these plots. This exercise is the subject of Chapter 6.

One other point to note in this set of experiments is the small initial rise in the Peak
Intensity Ratio for the experiment at 20 'C. The observation is almost certainly due to an

' experimental artifact in that the strong UV light source used to excite the probe molecules

 

 

 

 

 

 

 

3.15 I l l IT! I l I I I l : I I 1 % I I I 1' l I I i I I
3.1-~ °W° v :'
access? “is W :
.g D “fig 0 20 C -
g 3 05- 23% n 25 c «-
>. ‘i, 0 30 C .
t o ‘6 ‘
m 3-- o -.
5 '3’: s. m '
E on ° ‘23
; 295- a ’~, e. --
m a 9.. o
11’ 5': ’s “o :
'o 2.9- ‘5 ~ ‘5 --
o la, s 6!; .
.5 fi .: .
Tu “ q"
E 2.85-- ”in “t1. --
L O
O 00
z
2.8-- -.
2.75...:L..%..._:...1...:L1.§.1.‘

 

O 20 40 60 80 100 120 140

time, minutes

Figure 5.5 Effect of temperature on the desupersaturation curve of seeded citric acid
batch crystallization experiments. The ordinate are normalized so that all
curves start at the same point, since initial supersaturation (c/c* = 1.2), is
the same in all three cases. Other experimental conditions: agitator speed =
400 rpm, probe (pyranine) concentration = 1.0e-05 M, seeds (average size

660 microns) = 3 gms, excitation wavelength = 365 nm.

87

must have dissolved some of the fine nuclei at the beginning of the experiment causing the
initial supersaturation to go up a little bit and then stabilizing. The normalization done (to
match the initial supersaturation) serves only to match theory with experiment and should
be viewed as such.

5.3.4 Effect of probe concentration

A valid question with the use of this technique is the effect of the probe on the
system. The probe is usually added at an extremely small concentration of the order of le-
05 mols/liter. While this is a reasonably small amount of impurity, the quest for zero effect
on crystal purity prompted us to check the effect of this small amount on the purity of the
product. Two questions were posed: 1. Does the probe enter the cryStal lattice of the citric
acid product? 2. Does the concentration of probe have any effect on the results of the
desupersaturation experiments?

The first question was answered by analyzing the crystal product obtained. A
simple experiment was conducted by redissolving the washed crystals in water and taking
the fluorescence spectrum of the solution. The spectrum obtained had absolutely no
detectable fluorescence, which indicated that no appreciable amount of the probe enters the
crystal lattice and that all the probe molecules stay in the mother liquor. This conclusion
was further confirmed by analyzing the final crystal product by FAB-mass spectroscopy
that looks for specific masses of fragments. if any amount of pyranine had entered the
crystal lattice, it would be detected by this technique. Three samples of product crystals
were analyzed. The results of the mass spectrometry corroborated the negative
fluorescence result in two cases but showed a trace amount of pyranine in the third. This
amount was probably left on the surface of the crystal due to poor washing and could
definitely be removed by another washing of the crystal product with a saturated citric acid

solution. These experiments conclusively prove that the pyranine does not enter the

88

crystalline product and that loss of crystal purity is nor a concern when using this
technique.

The second question, that of any effect Of the concentration of probe on the
desupersaturation experiments, was answered by conducting identical experiments with
three different probe concentrations. The results are shown in Figure 5.6 by plotting the
relative supersaturation, 6, against time. The relative supersaturation, c, is defined by
equation (5.1). The three probe concenu'ations used were le-OS M, Se-06 M, and 1e-O6
M, covering an order of magnitude. As is evident in the figure, there is no appreciable
effect on the shape of the desupersaturation curve. In the raw data, the only difference is a
slight increase in the ordinate numbers, which is consistent through the whole length of the
experiment. This issue can be resolved very simply by normalizing each plot with respect
to the initial point, knowing that the initial supersaturation was constant in all experiments
and thus should have the same value ( if all other parameters are the same).

5.3.5 Measurement of kinetic parameters

The preceding sections demonsu'ated the ability of the fluorescent probe technique
as a viable tool for the in situ measurement of supersaturation. Further, its utility in the
measurement of desupersaturation experiments of citric acid was shown. In this section, a
brief discussion is presented on the usefulness of these measurements in a control scheme
and in measuring the kinetic parameters of a batch experiment.

It has been shown elsewhere [3, 4, 5] that the PIR can be correlated to the solution
concentration with a high degree of confidence. The PIR is a parameter that is reflective of
the solution structure and is inherently related to the activity. 'Ihus, supersaturation in the
solution can be alternatively defined in terms of this new parameter, the PIR. If PIR“ is
defined as the Peak Intensity Ratio obtained at equilibrium saturation and PIR is the ratio at
any other concentration, then the supersaturation can be defined either as a difference or as

a ratio as follows:

Relative Supersaturation

.
_O.02‘ll:l“:-‘l:“l:I-‘:r“l“4

Figure 5.6

 

89

 

 

 

 

 

 

 

 

I
O 20 4O 60 80 100 120 140

time, minutes

Effect of varying the probe (pyranine) concentration on the
desupersaturation curve of seeded citric acid batch crystallization
experiments. The ordinate shows the relative supersaturation {(c/c*)- 1 }that
is the same in all three cases. Other experimental conditions: temperature =
25 °C, agitator speed = 400 rpm, seeds (average size 660 microns) = 3

gms, excitation wavelength = 365 nm.

9O

Supersaturation = APIR = PIR - PIR *

(5.1)
APIR
PIR *

 

Supersaturation = o =

Since PIR is a parameter that reflects the fundamental structure of the solution, APIR and a
will most likely contribute to a more accurate measurement of supersaturation. Maintaining
a constant supersaturation is a requirement for attaining a CSD with a large average size in
batch crystallizations. For this purpose, a set point for a control scheme that is designed to
obtain such a CSD can be defined [7] as:

APIR = constant , or

o = constant

This idea is expressed picrorially in Figure 5.7. This definition is also a direct
measurement of supersaturation that does not require calibrations and therefore is more
practical as a set point for use in a control scheme. Thus, the accurate measurement of
supersaturation that is defined in terms of the PIR can be an extremely valuable tool for use

in the control of crystallizers.

5.4 Conclusions

The usefulness of the fluorescent probe technique to monitor the isothermal
desupersaturation characteristics of a seeded batch crystallizer has been demonstrated. The
probe molecule (pyranine) was added to the crystallizing mixture at the start of the
experiment and its emission spectrum monitored over time. A ratio of two emission peaks
was plotted against time to give rise to the desupersaturation curve. The effect of varying
the seed weight on the desupersaturation curve was shown. With increasing seed size, the
time required for secondary nucleation and the rapid desupersaturation to start decreases.
An increasing temperature of experiment also caused the same effect. It was demonstrated

that the relative amount of the probe in solution does not significantly alter the

 

91

 

 

)

Supersaturation

 

 

 

 

 

 

A .5
E
3 _. ___________ .
(U
a ”l
8.
: A(P|R) or o
(I)

 

 

 

 

Figure 5.7

.0
0

Temperature
>

Time 1-0
>

 

 

Control policy projected to maximize the mean size of a crystal size
distribution from a cooling batch crystallization process [7]. The policy is
based on maintaining a constant supersaturation during the course of the
crystallization. The supersaturation is defined by either of the following
two equations:

Supersaturation = APIR = PIR - PIR*

APIR
PIR *

 

Supersaturation = o =

where PIR is the ratio of emission peaks of a probe such as pyranine in a

supersaturated solution, and PIR“ is the same ratio in a saturated solution.

92

desupersaturation curve. Furthermore, the presence of the probe in the crystallizing

solution does not affect the cryStal purity.

5.5

References

.1. Garside, Chemical Engineering Science, Industrial Crystallization from Solution.
40, 3-26, (1985).

J. W. Mullin, O. Stihnel, Chemical Engineering Science, Expressions of
supersaturation in crystallization studies. 32, 683-686, (1977).

R. Chakraborty, K. A. Berglund, Journal of Crystal Growth, Steady State
Fluorescence Spectrosc0py of Pyranine as a Trace Extrinsic Probe to study
structure in aqueous sugar solutions. 125, 81-96, (1992).

R. Chakraborty, K. A. Berglund, AIChE Symposium Series No. 284 , The Use of
Pyranine as a Trace Fluorescent probe to Study Structure in Aqueous Sucrose
Solutions. 83, 114-123, (1991).

S. K. Yedur and K. A. Berglund, Accepted for publication in: Appl. Spec.,
(1996).

H. Kondo, I. lea, and J. Sunamoto, J. Phys. Chem. 86, 4826, (1982).

D. D. Dunuwila, An Investigation of the Feasibility of Using In Situ ATR FI‘IR

Spectroscopy in the Measurement of Crystallization Phenomenon for Research and
Development of Batch Crystallization Processes. PhD Thesis, Michigan State
University. (1996).

Chapter 6'

MODELS FOR THE EXTRACTION OF KINETIC
PARAMETERS FROM THE DESUPERSATURATION
CURVES

6.1 Background

In the previous chapter, the fluorescent probe was used to obtain information on
the desupersaturation of ciuic acid solutions during isothermal batch crystallization
experiments. The information obtained was in the form of curves that showed a parameter
defined as the Peak Intensity Ratio changing during the time course of the experiment. A
representative desupersaturation curve is shown in Figure 5.2. A number of parameters
were varied to test their effect on the nature of the desupersaturation curves. These effects
were shown in Figures 5.4, 5.5, and 5.6 and discussed in detail in Chapter 5.

The desupersaturation curves are ideal for studying the kinetics of batch
crystallization parameters. These experiments are almost always conducted isothermally
since the kinetic constants are temperature dependent. The process of crystallization can be
divided into two parts: nucleation and growth. Nucleation refers to the birth of nuclei in
solution and growth is the subsequent deposition of solid mass onto the nuclei. The

nucleation and growth rates are expressed as

G = kg 3“ (6.1a)

B = kn s'n (6.1b)

93

94

where G and B are the growth and nucleation rates respectively; kg and k1' are the grOWth
and nucleation rate constants, S is the supersaturation, and n and m are the ‘order’ to which
the supersaturation is raised. These orders do nor imply a reaction order, they are simply
empirical constants that indicate the power to which the supersaturation is raised. Thus, the
order has no fundamental significance. Typically, m>n, indicating that the nucleation rate
is a much stronger function of supersaturation. For aqueous solutions, it is usually
between 1 and 2, while m is usually between 2 and 5 [1]. A higher supersaturation will
favor more nucleation and less growth whereas a lower supersaturation will favor the
growth process with little nucleation. The relationship between all these parameters for a
mixed suspension, mixed product removal (MSMPR) crystallizer is shown in Figure 1.1.

The expressions in equations 6.1a and 6.1b are rather simplistic. Each term is
dependent on a number of parameters. Both the rate constants are dependent upon the
temperature, the agitation rate (for batch crystallizers), and any impurities that might be
present in the solution. Specifically, k8 may also be dependent on the crystal size L, and on
the suspension voidage in the crystallizer, while kn is also dependent on the suspension
density MT. The‘supersaturation S can be described in any convenient units that will in turn
decide the units of the growth and nucleation rate. In the previous chapter, equations 5.1
gave two possible definitions for supersaturation.

In our work, we define the supersaturation as:

a = PIR - PIR (6.2)

pm"

 

Here, PIR is the Peak Intensity Ratio (as defined in Chapter 5) at any concentration and
PIR“ is the same ratio at equilibrium conditions.

The kinetic constants together with the growth and nucleation rates describe the
crystallization process. If these parameters are known for a particular system, the complete

CSD and various crystal product properties such as total crystal area and mass can be

95

predicted. There have been numerous works that address this issue of obtaining the kinetic
data from experiment.

The particle number-size distribution theory introduced by Randolph and Larson [2]
and the fundamental supersaturation balance discussed by Mullin and Nyvlt [3, 4, 5]
provide powerful models to evaluate crystallization processes. The model derived from the
particle number-size distribution theory is based on the p0pulation balance (number of
particles per unit volume per length range) and takes nucleation and growth of crystals into
account. The fundamental supersaturation balance is derived taking into consideration that
the driving force for all crystallization phenomena such as nucleation and growth is
supersaturation. Thus, kinetic parameters for nucleation and growth are nesred in these
models giving them both predictive and descriptive capabilities. Another valuable set of
equations commonly used in crystallization analysis is the moments of the particle number-
size distribution [2]. The moments have been utilized to circumvent a basic dimensional
incompatibility between the population balance and the transport equations for mass,
momentum and energy.

The particle phase space consists of a set of internal coordinates in addition to the
three spatial dimensions. This incompatibility complicates the simultaneous solution of the
population balance with the transport equations. The most significant internal coordinate in
the particle phase space is the particle size. Therefore, the population density distribution is
integrated over the particle size internal coordinate reducing the dimensionality of the
population balance to that of the transport equations. The moments ‘m’ that give the total
particle properties upon integrating the population balance over the particle size coordinate

are given by

mj(t)=jn(L,r)LJ'dL, j = o, 1, 2, (6.3)
0

96

where n is the population density disrribution and L is the particle size. The first moment
gives the length, the second gives the area, and the third moment gives the mass of the
crystals.

The application of these models in engineering simulations and process analysis and
control has been limited by the unavailability of a reliable and sensitive technique to obtain
kinetic data in situ. In crystallization processes, in situ measurement is extremely important
since the hydrodynamic conditions set by the crystallizer dimensions, flow patterns induced
by the propeller and the suspended crystal slurry can significantly alter kinetic parameters
from that obtained under controlled laboratory experiments [6]. Theoretically, kinetic
parameters can be extracted from the desupersaturation rate of the solution monitored by in

situ fluorescence spectroscopy.

6.2 Development of the model

To obtain kinetic information from the desupersaturation curves that were generated
by the fluorescent probe technique, we have modified the differential approach proposed by
Mullin and Nyvlt [3]. The original model was intended to calculate the programmed
cooling curve for a batch crystallizer. Here, the general model is presented as is, with
appropriate substitutions where necessary.

Consider a batch crystallizer that contains a supersaturated solution that has 1 kg of
free solvent. This solution is seeded with a mass W, of seeds of a uniform size. The
crystallizer is agitated at a constant speed and crystallization is accomplished by cooling the
crystallizer.

The fundamental supersaturation balance is defined as:

‘31—? = s- kng" — knom (6-4)

where the derivative is with respect to time and A is the crystal surface area. The first term

on the right hand side, 5, is the supersaturation rate that corresponds to the creation of

97

supersaturation in the system, while the second and third terms correspond to the
desupersaturation due to growth and nucleation respectively.

The supersaturation rate, 3, can be expressed as

.. i- 9519 (,5,
dt d6 dt °

As the temperature, 6, goes down in a cooling crystallizer, equation (6.4) may be rewritten

as

—-(:i—(: = pdit-i-kgAOn-lrkndm (6-6)

This equation provides us with a general equation that can be used to fit the
desupersaturation data. The experiment provides the left hand side of equation (6.6) while
a fit can provide the relevant parameters on the right hand side of the equation.

As mentioned before, the terms on the right hand side are dependent on a number of
parameters and any fit has to taken into account some of these. As with any model, it is
desirable to fit the data with a minimum number of parameters, and that is the goal here.
The following analysis considers the whole process from a fundamental viewpoint. We
start with a mass of seeds and track their growth along with the formation of new nuclei
due to nucleation over a fixed time interval. These growing nuclei and the newborn nuclei
consume mass, and that is reflected in the desupersaturation rate. Thus, a relationship can
be derived between the desupersaturation rate and the growth and birth of nuclei.

To obtain the relation between the cooling temperature 0 on time, the following
analysis is made. The total number of crystals present at any given time N(t), per kg of
free solvent present in solution, is comprised of the number of seed crystals added at the
start of the process and the number of crystals that are generated due to nucleation.

N(t) = N8 + Nn

lsfl. + kn(60)[6(60)1mt

 

(6.7)

 

98

where W is the weight of the crystals, 01 is the volume shape factor, pc is the density of the
crystals, and L is the characteristic length of a crystal. The subscript ‘5’ refers to the seed
crystals, whereas ‘a’ refers to the nuclei formed.

All crystals that are present in the supersaturated solution grow in size according to

the following equations:
t-l

1.3(1) = 1.80 + 2gi(t)At (6.8)
i=0
t-l

Ln(t) = Lno + Zgi(t)At (6.9)

i=1,
where to denotes the time at which an individual nucleus first appeared and At is the time

interval of growth. gi is the linear crystal growth rate and is defined as

, k 9-,L- c“
g, = _8( . 1’5 (6.10)
309‘;

 

where B is the surface shape factor. The crystal growth rate constant, k', is temperature
dependent. It may also depend on the crystal size, the concentration of any impmity, and
the voidage of the crystal mass.

The crystal surface area at a time ‘t’ in a unit mass of solution is given by

l
A(t) = BNSLZSO) + BXN,L%(:)A1 (6.11)
l

Combining equations (6.7), (6.8), (6.9), and (6.11), we obtain

1.1 2 1—1 r-1 2
A(t) = BNs[Ls0 + 28(9i’Lsi)At] + BNn Z[Ln0 +28(91.Ln(i-j))At] (6.12)
i=0 i=0 j=1

Substituting these equations into equation (6.4) and solving for the temperature drop with

time, we obtain

99

 

 

 

 

2
-09 = 3Wso gum-91) 1+ Eg(eirLsi)m
dl 29. 4. d_° L50 i=0 L50
d6 d6

“1k e as m e.L- At '4 ge.r. 1112
——, ‘ kn(00)[c(90)]m + 32" "(MI ( 0)] g“ 1.1.1) +2 “J W”)
dc + .d_0 i=0 Lno j=l Lno
d6 d6

(6.13)

This equation represents the theoretical cooling curve for a batch cooling crystallizer.
For the isothermal desupersaturation experiments, this equation can be modified
very simply by moving all temperature derivatives to the left hand side and combining the

terms. Since the temperature is constant throughout, dc*/dt = 0. These manipulations

result in the following equation:

2
_d_o = 3Wsog<Lst.61>[l+ 2L_-_<9nl«sn>m] +

dt LsO i=0 L50

2
+3an H60)[c(60)] g(9,,lL nit)At[1 + §g(9j,Ln(j-i))At

kn(90)[°(9o)im L o
n

(6.14)

This complex equation provides a basis for obtaining the kinetic parameters if the
desupersaturation profile (the left hand side of equation 6.14) is known. The unknowns on
the right hand side are g, kg, kn, n, and m. Note that L is calculated from equations (6.8)
and (6.9) using ‘g’ from equation (6.10). Once these are all known, the nucleation rate can

be calculated by using a power model such as

100

0 1 [ n n
B = kn, 0' v + kmsM 0' ] (6.15)
uchiIO p

 

where B0 is the nucleation rate, kn. is the primary nucleation rate constant, km is the

p
secondary nucleation rate constant, tip is the order of primary nucleation, and n, is the order
of secondary nucleation.

The first term on the right hand side of the equation (6.14) describes
desupersaturation that is only due to growth of the seeds that are added, without any
nucleation. From the desupersaturation experiments that were described in Chapter 5, (see
for example Figure 5.2), this region of growth of seeds and no nucleation is given by the
first flat part of the curve. In this region, the second term on the right hand side of equation
(6.14) does not apply. By fitting the data to the reduced equation, the growth kinetics are
obtained. With this information in hand, the entire data set can be fitted to the equation
using the known growth parameters to provide the nucleation rates and rate constants.
Thus, a successful fit will provide all the kinetic parameters of the crystallization such as
the nucleation and growth rates and the respective rate constants.

Once the kinetic parameters are known, it is possible to predict the crystal size

distribution at any time interval by simply integrating the equations for the length of the

crystal at the desired time. These further analyses are part of the ongoing work in this area.

6.3 Conclusions

The desupersaturation experiments using the fluorescent probe technique are
uniquely capable of providing kinetic information about the crystallization process. The
utility of the fluorescent probe is in its ability to measure the continually changing
conditions in the crystallizer with great sensitivity. While conventional techniques usually
require the removal of a sample from the crystallizer to analyze it, this technique provides
an in situ measurement tool of the crystallizing solution. Furthermore, the high sensitivity

offered provides information of the changing conditions in the crystallizer within extremely

101

small time intervals. In the experiments conducted, the time interval chosen was one
minute, but this can very easily be reduced to as low as one second. The high sensitivity
obtained results in a larger number of data points that can be fit to the model with a much
greater accuracy. Thus the fluorescent probe technique is convenient not only to measure
the concentration and supersaturation in a crystallizer, but also to obtain valuable

information on the crystallization kinetics.

6.4 References

1. P. C. Wankat. In: Rate Controlled Separations. Elsevier Applied Science. pp88.
(1990).

2. A. D. Randolph and M. A. Larson. Theory of Particulate Processes, Academic
Press, Inc., New York, USA, ch. 3, (1988).

J. W. Mullin and J. Nyvlt. Chem. Eng. Sci., 26, 369, (1971).
J. Nyvlt. Colln. Czech. Chem. Commun., 30, 2269, (1965).
J. Nyvlt and J. W. Mullin. Chem. Eng. Sci., 25, 131, (1970).

O‘MAUJ

N. S. Tavare, Separation and Purification Methods, 22:2, 93, (1993).

Chapter 7

CONTINUING INVESTIGATIONS AND
RECOMMENDATIONS FOR FUTURE WORK

7. 1 Background

In chapters 2, 3, and 4, the utility of the fluorescent probes as a sensor for
concentration and supersaturation measurements was demonstrated. In Chapter 5, the
fluorescent probe pyranine was used to generate desupersaturation curves for seeded citric
acid solutions in a batch crystallizer. Chapter 6 provided a predictive model to estimate the
kinetic parameters of the crystallization of citric acid monohydrate in a batch crystallizer
based on the desupersaturation curves. In most of these applications, the probe was
allowed to be present fieely in solution. The exceptions were the results presented with the
immobilized probe in Chapters 3 and 4. These experiments with the immobilized probe
showed that although the sensory properties of the probe can be preserved, the chemistry
involved in the immobilization cause significant changes rendering the use of immobilized
probes unsatisfactory. The problem of the probe serving as an interfering impurity remains
to be addressed. Although Fast Atomic Bombardment (FAB) experiments on the citric acid
monohydrates that was crystallized showed only trace amounts of residual probe in the
final product, in many cases, it will still be unacceptable to add fluorescent probes of
unknown toxicity into a crystallizer. Even if the probes are inherently safe, as is the case
with the FD&C dyes, the possible inclusion of these probes into the crystals, thereby
affecting the purity would undoubtedly cause reluctance on the part of current

crystallization operations to use this technique. In the case where crystal purity is not a

102

103

serious concern, the question whether these probes affect the crystal habit would have to be
answered. Although there is no indication that the habit is affected in any way by these
probes; at this time, we are not in a position to make such a claim categorically.

For these myriad reasons, it is desirable that an alternate technology be developed
that would use the same principles of fluorescent probes, but would be able to lay to rest
any fears of possible contamination or habit modification. For this to be possible, it is
necessary to contact the probe with the crystallizing solution ex situ. Such a configuration
would involve drawing out a sample solution stream from the crystallizer, preferably with
the crystals, contacting the probe with the sample in a flow cell where the fluorescence
spectrum can be recorded, and then either returning the sample stream to the crystallizer

after removing the probe, or by simply sending the sample stream to waste.

7.2 Continuing Investigations

7.2.1 Estimation of kinetic parameters

Based on the model provided in Chapter 6, work in ongoing to fit the experimental
data to the model. A successful fit will provide the citric acid growth and nucleation rate,
as well as the various kinetic parameters. These can potentially be used to predict the
crystal size distribution.
7 .2.2 Ex situ monitoring of concentration

This proposed configuration of the experimental setup to solve the problems
discussed in the section 7.1 is very similar to manifolds used in conventional Flow
Injection Analysis (FIA). FIA is defined as a non-chromatographic flow analysis technique
for quantitative analysis, performed by reproducibly manipulating sample and reagent
zones in a flow stream under thermodynamically non-equilibrated conditions [1]. The
basic components of FIA system and a variation of the conventional - reverse Flow
Injection Analysis (rFIA) system are shown in Figures 7.1 and 7.2. There arecertain basic

differences between a conventional FIA system and an rFIA system. The most suiking of

104

 

 

 

 

 

 

C > //\/\—‘___—’W

 

 

 

Figure 7.1 Schematic of a conventional Flow Injection System. C = carrier stream, R

= reagent Stream, S = sample, D = detecror, P = pump, and W = waSte

 

 

 

 

stream.
R
C ‘—"_"l r
s > / D

 

 

Figure 7.2 Schematic of a reverse Flow Injection System. C = carrier stream, R =
reagent stream, S = sample, D = detector, P = pump, and W = waste

stream.

105

these is in the nature of the mobile phase. In conventional FIA, the reagent (probe) stream
moves continuously through the system along with the carrier while the sample is injected
at fixed intervals. On the other hand, in rFIA, the sample and the carrier move
continuously, while the reagent is injected intermittently. FIA techniques are used for a
wide variety of applications [1, 2]. In crystallization too, FIA has been used for the
determination of d- and l- glutamic acid [3], and for the determination of phylate and phytic
acid during the crystallization of calcium oxalate [4, 5]. rFIA is generally of use in
applications where there is an abundance of sample. An example where rFIA has been
used is in the examination of ocean water chemistry aboard a ship [6]. In our case, we
have a similar situation with an abundance of sample in the crystallizer making the use of
rFIA quite appropriate. Furthermore, the use of a carrier sueam is obviated in the proposed
configuration.

Preliminary experiments with a working protorype of such a system were
conducted in a batch crystallizer. Desupersaturation experiments with citric acid similar to
those described in Chapter 5 were conducted. A schematic of the system used is shown in
Figure 7.3. A supersaturated solution of citric acid was made by dissolving solute in
excess of solubility at 25 'C. Dissolution was carried out at 35 ‘C and the resulting
solution was cooled down to 25 °C. A supersaturation of 1.2 (=c/c*) was generated by
this protocol. Pyranine was added to this solution to generate an overall probe
concentration of 1.0e-05 M in the crystallizer. A constant stirring rate of 400 rpm was used
throughout. A sample stream was drawn out from the crystallizer with the aid of a
peristaltic pump. The sample was directed through a flow cell. The flow cell was
essentially a cylindrical disc with an OD. of 2.54 cm and a height of 1 cm equipped with
inlet and outlet port at its sides. A schematic depiction of the flow cell is presented in
Figure 7.3. A fiber optic end was placed flush with the top of this flow cell to record the

fluorescence spectrum of pyranine. After passing through the flow cell, the sample stream

 

flow cell

 

 

  
   
 

    

 

 

UV source

 

 

 

 

probe solution

 

 

 

:3

J_| crystallizer
l_l

 

 

 

peristaltic pump

Figure 7.3 Schematic of the apparatus used to measure the concentration of a
crystallizing solution ex situ. The top view of the flow cell is shown on the

upper left hand comer.

107

was returned to the crysrallizer. For the desupersaturation experiment, 3 grams of seeds
(average size 780 mm) were added to the crystallizer at time zero.

The emission spectrum of the pyranine in the sample stream was recorded every
minute for 2 hours. As expected, at the conditions maintained, the ciuic acid monohydrate
crystals have an induction time of about 30 minutes, during which the seeds grow. After
the induction time, secondary nucleation occurs that continues till all the excess
supersaturation is depleted and the system comes to equilibrium at 25 'C. The ratio of the
emission peaks of pyranine, as a function of time, is shown in Figure 7.4. As is evident,
this sampling technique works very reasonably and the curve compares well with the one
shown in Figure 5.2.

The important disrinction with conventional techniques is that here, there is no
separation of the solid phase from the solution necessary. The emission spectrum is
collected in the presence of crystals and gives remarkably good results. The scattering of
data points observed during the initial part of the curve is due to the changing number of
crystals that are passing in front of the fiber optic end. As secondary nucleation
commences and the number of crystals increases significantly, the flow cell is almost
completely filled with crystals flowing past the fiber tip. Beyond a certain number, the
emission specu'um is nor affected by the number of crystals and there is a substantial
reduction in the scattering of the data points. Work in ongoing in this area to improve the

design so as to reduce this scattering.

7.3 Recommendations for future work

The utility of the experiment described in the previous section is manifold. It
demonstrates that the emission spectrum of the probe can be collected ex situ by
withdrawing a sample stream from the crystallizer without having to separate out the
crystals. While in this experiment, the probe was added to the crystallizer, it would be

relatively simple to change the experimental setup such that the probe is injected into the

108

 

w
H
U)
or
.1
u
——
q
d
.1
—u—
-1
.1
d
-—1—
q
.-
.1
-U—
.1
'1
-(
—1—
cl
.1
4
—h—
.1
.1
.1

rJirrror
(p
O
O

3.2 00 --
.9 o o .
+4 - 9’ 0 0 Q '
W if 00 t9 ‘
m 3.1 ‘1; 0 c Q; 1"
a . s
171 C 9% i
C 3 -- __
cu - 0% .
H b 0 091° Q: o
c .
_ om 3W
x 2 9‘: 0060b 0 ~ 60 '-
m ' _ 0 <58anch .
Q) - o .
D. r .
l- o «1
2.8 r % .
L .

 

 

2.7 l 11:11J:_Jl 1%1 1 1:1 1 1%1 j l : l l l

O 20 4O 60 80 100120140
time, min

 

Figure 7.4 Typical desupersaturation curve of a seeded citric acid batch crystallization
experiment using the apparatus shown in Figure 7.3. The experimental
conditions were as follows: Temperature = 25 'C, initial supersaturation
(c/c*) = 1.2, agitator speed = 400 rpm, probe (pyranine) concentration =
1.0e-05 M, seeds (average size 660 microns) = 3 gms, excitation
wavelength = 365 nm. The ordinate shows the Peak Intensity Ratio,
defined as the ratio of the intensities of the emission peaks of pyranine at

449 nm and 520 nm.

109

flow cell directly by means of a syringe pump. This modification will be a primary focus
of the continuing work in this area. Modifications in the design of the flow cell will also be
required to incorporate an injection port for the probe solution. The design will also have
to ensure adequate mixing of the probe with the sample stream before the errrission
spectrum is taken. This might require the probe to be injected first into some kind of a
mixing chamber, either a mixing tee or a larger chamber with provision for physical
agitation, that is placed before the flow cell. The overall objective being to incorporate this
analytical technique into the conu'ol scheme for the crystallizer, it is important to miniaturize
the whole technique as much as possible to shorten the lag times of the response. The
miniaturization would require shortening the length of all the tubing to a minimum. The
size of the tubing will have to be optimized very carefully, taking care that blockage of the
lines fi‘om the crystals does not occur. Once all these parameters are optimized, it will be
possible to incorporate this technique for measuring the concentration into a control scheme
for the crystallizer.

The question of the further role of the sample stream exiting the flow cell has to be
answered. Should the sample stream be returned to the crystallizer, or should it be sent to
waste? The answer to this would depend on the scale of the crystallizer and the econorrrics
involved. In the case of a large scale crystallizer, where the sample volume is very small
relative to the working volume, it may be preferable to send the sample to a waste stream.
However, if the crystallizer is very small, it would not be possible to remove sample
continuously without affecting the concentration inside the crystallizer. Also, if the
compound being crystallized is quite expensive, the economics of the whole operation
comes into the picture, and the sample stream cannot be discarded.

In these latter cases where the sample sueam cannot be discarded, treatment of the
sample stream to remove the probe molecules would be essential. The removal of the
probe from the sample stream can be accomplished quite easily with the aid of ion exchange

surfaces. As a pretreatment step, the sample stream should be heated so that all the crystals

110

dissolve. The dissolution can be achieved by passing the stream through a heater coil of
sufficient length. The solution obtained has the probes in their anionic form. By passing
this solution through a bed of anion exchange resins (such as those manufactured by Rohm
and Haas under the trade name series IRA), the probe molecules can be exchanged onto the
resin. The length of the bed and the residence time of the solution in the bed would have to
be optimized keeping in mind the ion exchange capacity of the resin. The nature of the
solute, the viscosity of the solution and the amount of probe in the sample stream would be
other significant parameters to be considered for the design of the ion exchange bed. The
only drawback to this technique would be the fact that the cationic sodium ions would still
remain in solution (in the case of pyranine). If this is considered a problem, similar cation
exchange material can be placed in the path of the stream to eliminate the cations. After this
step, passing the sample stream through a heat exchanger to lower the temperature back to
that in the crystallizer would complete the sample sueam treatment necessary. The treated
sample can be returned to the cryStallizer with no effect other than the loss of a few
crystals. As noted earlier, miniaturization of the whole process will significantly reduce
such effects.

Once such a system is up and running, numerous experiments can be conducted
that would be of interest in the field of crystallization. Preliminary experiments to expand
the utility of this system would involve screening potential probe molecules for different
crystallizing solutes. It has been our experience that there is no single fluorescent molecule
that can be used for all solutes that might be of interest. It is essential to find probes that
show a solvatochromic response in their emission spectra in the presence of the solution of
the particular solute that is of interest. There is an extremely large number of fluorescent
melecules that are used today for different applications. Only a few of those, though, will
be suited any specific application. A comprehensive effort at screening this large number
of fluorescent probes would be a necessary first step. It is envisaged that a list could be

made of probes that are suited to be used with a particular class of compounds such as

111

sugars, carboxylic acids, amino acids, etc. Once such a list is available, it would be a
simple matter for future researchers to pick a fluorescent probe that will work with a given
solute.

An important area of interest is to study the effect of impurities on the crystallization
process. When imptuities are present in the mother liquor, it is well known that the
kinetics of the process change that can result in an unspecified crystal size distribution. It is
also quite likely that the habit of the resulting crystals change [7]. Studying the kinetics of
such an impurity-affected system, along the lines of that described in Chapter 5 and 6,
would give quantitative indications of the effect of the impurities.

7.3.1 ATR Fluorescence

Attenuated Total Reflectance (ATR) is a concept that is widely used with Infra Red
and UV-Vis spectroscopies. Recently, ATR-FTIR spectroscopy has been very
successfully used for the in situ measurement of supersaturation in maleic acid [8]. The
use of ATR in fluorescence however, is quite limited. This is due to the inherent nature of
fluorescence that requires the emitted light to be spatially separated from the incident light.
This necessitates the introduction of a new light wave to carry the emitted light that has to
cross a system boundary and enter the medium in which the internally reflected beam is
passing through. Thus total internal reflection (that occurs in a single medium), which is
the basis of ATR spectroscopy is not strictly. possible. However, a slight variation has
been used in which fluorescent molecules are coated onto the core of a fiber optic cable that
is placed in the sample solution [9, 10]. The evanescent wave generated excites the
fluorescent molecules and the emission is picked up either by the same fiber or by a
different fiber placed alongside. This configuration has its advantages in that the probe is
immobilized and the development of an analytical test kit of sorts is simpler. However, the
immobilization procedure has to be optimized very carefully, as it is relatively easy to lose

or alter the fluorescent properties of a molecule when it is immobilized. This aspect was

112

discussed in detail in Chapters 3 and 4. In spite of these concerns, ATR fluorescence holds

tremendous promise in future research efforts.

7.4 Conclusions

With recent advances in fiber optic and spectrophotometer technologies, the
development of an analytical sensor based on fluorescence for the measurement of
concenuation and supersaturation in a crystallizer has become a reality. Not only is the
fluorescence technique capable of measuring the solution concentration, the kinetics of
growth and nucleation can also be predicted. Once the kinetic parameters are known, the
crystal size distribution can also be predicted using the theories of population balance. This
predictive nature of the fluorescent probe technique makes it an extremely versatile tool in
improving current crystallization operations and designing better control strategies for the

future.

7.5 References

1. Z. Fang, In: Flow Injection Separation and Concentration. VCH, Weinheim,
Germany. (1993).

2. B. Karlberg and G. E. Pacey. In: Flow Injection Analysis A Practical Guide.
Elsevier, Amsterdam. (1989).

E. Ballesteros, M. Gallego, and M. Valcarcel. Anal Chem, 68, 322-326, (1996).
F. Grases, A. Costa-Bauza, and J. G. March. Analysis, 21, 95-99, (1993).

F. Grases and P. March. Analytica Chimica Acta, 219, 89-95, (1989).

J. Thomsen, K. S. Johnson, and R. L. Petty. Anal. Chem, 55, 2378, (1983).

\IO‘UIAUO

D. L. Klug. In: Handbook of Industrial Crystallization. Ed. Allan S. Myerson.
Butterworth-Heinemann, USA. (1993).

8. D. D. Dunuwila, An Investigation of the Feasibility of Using In Situ ATR FTIR
Spectroscopy in the Measurement of Crystallization Phenomenon for Research and
Development of Batch Crystallization Processes. PhD Thesis, Michigan State
University. (1996).

10.

113

Cheng, Y. L., Lok, B. K., and Robertson, C. R. In: Surface and Interfacial
Aspects of Biomedical Polymers. Ed. J. D. Andrade. Vol 2., Plenum Press, New

York. (1985).

Hlady, V., Van Wagenan, R. A., and Andrade, J. D. In: Surface and Interfacial
Aspects of Biomedical Polymers. Ed. J. D. Andrade. Vol 2., Plenum Press, New
York. (1985).

Chapter 8

 

CONCLUSIONS

The fluorescence probe technique is shown to be well suited for accurately
measuring the supersaturation in aqueous solutions of citric acid and in sugar solutions.
The measurement is more sensitive than those obtained by conventional analytical
techniques such as refractometry. The nature of the measurement also facilitates the in situ
measurement of supersaturation in a crystallizer without separation of the solid crystals
from the solution. The relative supersaturation is defined in terms of a parameter( the Peak
Intensity Ratio), that is generated by monitoring the fluorescence of probe molecules placed
in contact with the crystallizing solution.

Pyranine is currently the best choice for use as a fluorescent probe. This
recommendation is based partly on its strong solvatochromic behavior and partly on its
high quantum efficiency and long shelf life.

Three of the seven colors listed by the FDA as food grade are well suited for use as
supersaturation sensors in sucrose solutions. These are FD&C Blue No. 1, FD&C Red
No. 40, and FD&C Green No. 3.

Anionic ion exchange membranes can be used to immobilize these probe molecules
for use in immersible sensors. However, the immobilization procedure causes a reduction
in sensitivity and performance.

The kinetic parameters of crystal nucleation and growth can be estimated by

modeling the desupersaturation curve that is generated by monitoring the fluorescence of

114

115

pyranine during the course of an isothermal batch crystallization experiment. The changing
supersaturation in the crystallizer is reflected by the change in the fluorescence behavior of
pyranine. The desupersaturation curve generated is then modeled to fit theoretical
equations of crystal nucleation and growth to give the various kinetic parameters such as
growth and nucleation rate constants and the growth and nucleation orders.

Although the presence of the probe in contact with the solution does not appear to
affect the crystallization process in general, it is still desirable to contact the probe with the
crystallizing solution in an ex situ manner, so that no contamination of the crystals is
possible. A system is set up based on the principle of Reverse Flow Injection Analysis,
that is well suited to make the necessary measurements by drawing out a sample of the
slurry from a crystallizer and contacting the probe with it in an ex situ manner. Results
obtained show a very close correspondence with the experiments where in the
measurements were made in situ. Future research efforts will attempt to modify and perfect

this new setup as a first step in expanding the scope of the fluorescence probe technique.

APPENDIX

Appendix

 

Table A1 Raw data for Figure 2.4

 

 

 

 

 

Citric acid Peak Intensity
concentration Ratio

Wto/o

10 0.34835
20 0.58491

30 0.90729
40 1.3249

50 1.9628

65 3.1494

 

Table A2 Raw data for Figure 2.5

 

 

 

 

 

Supersaturation Peak Intensity
Ratio
0.081917 3.93
-0.072643 3.02
-0.2272 1.99
-0.38176 1.35
-0.53632 0.903
-0.69088 0.589
-0.84544 0.354
-1 0.0512

 

 

116

 

 

 

 

 

 

117

 

 

Table A3 Raw data for Figure 2.6
Citric Acid pH
Concentration
(wt%)
10 1.51
20 1.24
30 1.06
40 0.78
50 0.51
65 0.09
Table A4 Raw data for Figure 2.7
pH of HCl-NaOH Peak Intensity pH of citric acid Peak Intensity
solutions Ratio solutions Ratio
0.12 1.3364 1.54 0.34835
0.3 0.90244 1.28 0.58491
0.5 0.70972 1 .06 0.90729
0.74 0.60671 0.78 1.3249
1 0.4647 0.51 1.9628
1.26 0.28647 0.09 3.1494
1.5 0.19378

 

 

 

 

 

 

118

Table A5 Raw data for Figure 2.8

 

 

Citric Acid Peak Intensity
concentration Ratio
wt%
10 1.54
20 1.28
30 1.06
40 0.78
50 0.51
65 0.09

 

 

 

 

Table A6 Raw data for Figure 3.6

 

Sucrose Peak Intensity
Concentration, Ratio
wt%

   

30 0.26825
50 1.0721
65 5.5909

 

 

 

 

 

119

Table A7 Raw data for Figure 3.7

 

Sucrose Peak Intensity
Concentration, wt% Ratio
0 1
10 1.9709
30 4.4627
50 9.4343
65 22.281

 

 

 

 

Table A8 Raw data for Figure 3. 9

 

Sucrose Peak Intensity
Concentration, wt% Ratio
10 1.6064
30 2.3278
50 3.21 14
65 4.5053

 

 

 

 

120

Table A9 Raw data for Figure 4.4

 

Sucrose Peak Intensity
Concentration, wt% Ratio
10 1.825
30 1.925
50 1.99
65 2.08

 

 

 

 

 

Table A 10

Raw data for Figure 5.2

121

 

 

Time (minutes) Peak Intensity

—I—I
doomwmmaw

12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
3S

 

Ratio

3.1094
3.1104
3.1096
3.1069
3.1044
3.1051
3.1074
3.1022
3.1033
3.099
3.1031
3.1003
3.0969
3.0985
3.0993
3.0973
3.0969
3.0964
3.0937
3.0942
3.0934
3.0899
3.0863
3.0903
3.087
3.0881
3.0893
3.0845
3.0812
3.0803
3.0736
3.0689
3.0703

 

Time (minutes)

36
37
38
39
4o
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
6o
61
62
63
64
65
66
67
68

 

Peak Intensity
Ratio contd.

3.0642
3.0559
3.0485
3.0401
3.038
3.0297
3.0157
3.0066
2.9946
2.9877
2.9794
2.9686
2.9608
2.9523
2.9483
2.9358
2.9322
2.9238
2.9168
2.9116
2.9037
2.8991
2.8947
2.8853
2.8839
2.88
2.8743
2.869
2.8638
2.8617
2.8555
2.8556
2.8498

 

 

Table A 10 Continued

69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
9O
91

 

2.8455
2.8451
2.84
2.8345
2.8341
2.8315
2.8288
2.8279
2.8252
2.8248
2.8205
2.8187
2.8147
2.8122
2.8116
2.8146
2.8073
2.8082
2.8052
2.8044
2.803
2.8021
2.8021

 

 

122

 

123

Table A1 1 Raw data for Figure 5.3

 

 

 

 

 

 

Time (minutes) Peak Intensity Time (minutes) Peak Intensity

(contd.)
3 157.66 36 253.11
4 170.87 37 289.93
5 165.76 38 315.91
6 168.41 39 344.63
7 155.07 40 387.39
8 168.33 41 414.74
9 171.12 42 426.07
10 164.46 43 439.59
11 165.06 44 442.67
12 159.94 45 498.79
13 169.08 46 489.39
14 165.8 47 521.19
15 155.34 48 533.19
16 160.37 49 527.12
17 164.55 50 569.66
18 157.77 51 570.59
19 159.69 52 598.47
20 164.41 53 613.42
21 159.6 54 639.81
22 170.12 55 654.97
23 171.34 56 657.33
24 165.27 57 662.58
25 159.62 58 676.91
26 164.06 59 644.88
27 171.19 60 720.57
28 177.51 61 708.97
29 182.05 62 707.28
30 184.91 63 696.55
31 200.43 64 697.01
32 199.85 65 720.27
33 214.55 66 700.43
34 222.9 67 703.04
35 248.56 68 736.31

 

 

Table A1 1 Continued

69
7O
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
9O
91

 

703.48
718.1 1
717.17
706.62
723.05
713.09
713.4
755.45
756.91
778.23
722.21
772.34
743.05
759.7
770.6
780.55
751.93
777.68
785.93
766.53
754.09
770.1

770.1

 

 

124

125

Table A 12 Raw data for Figure 5.4

 

 

 

 

 

 

Time 3 gms 15 gms 36 gms
(minutes)

1 3.096 3.096 3.096
2 3.097 3.081 3.0994
3 3.0962 3.0874 3.1085
4 3.0935 3.083 3. 0996
5 3.091 3.0847 3.106
6 3.0917 3.0888 3.0955
7 3.094 3.0837 3.0864
8 3.0889 3.0803 3.0737
9 3.0899 3.0767 3. 0827
10 3.0856 3.0829 3. 0883
11 3.0898 3.0864 3. 0799
12 3.087 3.0799 3. 08
13 3.0836 3.0769 3.0883
14 3.0852 3.0731 3.0728
15 3.0859 3.0738 3.074
16 3.084 3.0727 3.0751
17 3.0835 3.0652 3.076
18 3.083 3.0771 3. 0571
19 3.0804 3.0554 3. 0601
20 3.0809 3.059 3. 0479
21 3.0801 3.0581 3.0506
22 3.0766 3.0699 3.0366
23 3.073 3.0693 3.0336
24 3.077 3.0562 3.0203
25 3.0737 3.0553 3.0159
26 3.0748 3.0493 3. 0093
27 3.076 3.0609 3. 004
28 3.0712 3.0307 2.9978
29 3.0679 3.0394 2.993
30 3.067 3.0208 2.9842
31 3.0604 3.015 2.966
32 3.0557 3.0051 2.9658

 

 

Table A 12 continued

33
34
35
36
37
38
39
4o
41
42
43
44
4s
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67

 

3.057
3.051
3.0427
3.0354
3.027
3.0249
3.0166
3.0027
2.9936
2.9817
2.9748
2.9666
2.9558
2.948
2.9396
2.9356
2.9232
2.9196
2.9112
2.9043
2.899
2.8911
2.8866
2.8822
2.8729
2.8714
2.8676
2.862
2.8566
2.8515
2.8493
2.8432
2.8433
2.8375
2.8332

 

126

3.01
2.9845
2.9749
2.9707
2.9612
2.9466
2.9528
2.9338

2.939
2.928
2.9279
2.9203
2.9031
2.9092
2.8973
2.883
2.8956
2.881 1
2.8878
2.8746
2.8807
2.8691
2.8708
2.862
2.856
2.8515
2.854
2.8519
2.8483
2.8483

2.84
2.8403
2.8403
2.8243
2.8421

 

2.9675
2.963
2.9429
2.942
2.9376
2.9351
2.9349
2.933
2.911
2.9103
2.9084
2.9013
2.8965
2.8855
2.893
2.8845
2.8753
2.8816
2.8762
2.8768
2.8651
2.8592
2.8574
2.8488
2.845
2.8474
2.8542
2.8542
2.8409
2.8356
2.842
2.8384
2.8401
2.8313
2.8408

 

 

Table A12 continued

68
69
7O
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102

 

2.8329
2.8278
2.8223
2.8219
2.8193
2.8166
2.8157
2.813
2.8126
2.8084
2.8066
2.8025
2.8001
2.7995
2.8025
2.7952
2.7961
2.7931
2.7923
2.791
2.79
2.79

 

127

2.8237
2.8331
2.8225
2.8254
2.8229
2.8259
2.8184
2.8129
2.8177
2.8287
2.8173
2.8205
2.8138
2.8172
2.8148
2.8217
2.8058
2.8117
2.8158
2.7998
2.813
2.7976
2.8067
2.8035
2.8127
2.8146
2.8098
2.8106
2.8086
2.8098
2.8129
2.7987
2.8009
2.8157
2.7983

 

2.8403
2.8359
2.8252
2.837
2.8422
2.8236
2.8248
2.8299
2.8189
2.8246
2.8271
2.8301
2.8308
2.8266
2.8185
2.8285
2.8246
2.8232
2.8256
2.8189
2.8159
2.8189
2.8305
2.8169
2.8171
2.8158
2.824
2.8156
2.8136
2.8261
2.8216
2.8098
2.8258
2.8175
2.8202

 

 

Table A 12 continued

103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121

 

 

128

2.7954
2.806
2.8065
2.7991
2.8007
2.8031
2.8053
2.8158
2.8082
2.805
2.792
2.7976
2.8089
2.8103
2.8073
2.8014
2.8265
2.8109
2.8109

 

2.8125
2.8114
2.8122
2.825
2.8179
2.8292
2.8261
2.8145
2.8242
2.8222
2.8372
2.8161
2.8215
2.8129
2.8118
2.8184
2.8259
2.837
2.837

 

 

Table A13

Raw data for Figure 5.5

129

 

Time (minutes)

 

wwwWNNNNNNNNNNd—I—i—i-‘d-‘d—l—I ,

 

 

 

  
   

 

 

 

Table A 13 continued

34
35
36
37
38
39
4o
41
42
43
44
45
46
47
48
49
5o
51
52
53 ‘
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68

 

3.0957
3.0988
3.098
3.1028
3.101 1
3.0934
3.0937
3.0964
3.0919
3.0942
3.0984
3.0932
3.0895
3.0998
3.0974
3.1006
3.0902
3.0963
3.0968
3.0955
3.0943
3.0924
3.09
3.0907
3.0944
3.0881
3.0887
3.0851
3.0861
3.0837
3.0786
3.0726
3.0725
3.0743
3.0742

 

130

3.0221
3.0235
3.0175
3.0093
3.0021
2.9938
2.9917
2.9835
2.9697
2.9608
2.9489
2.9422
2.934

2.9234
2.9156
2.9073
2.9033
2.891 1
2.8875
2.8792
2.8724
2.8672
2.8594
2.8549
2.8505
2.8414
2.8399
2.8361
2.8305
2.8252
2.8202
2.818

2.812

2.8121
2.8063

 

3.0279
3.0223
3.0193
3.0206
3.0153
3.015
3.0106
3.0126
3.0101
3.0054
2.9972
3.0005
2.9947

' 2.9866

2.9862
2.9808
2.9756
2.9658
2.9594
2.9542
2.9524
2.9441
2.9374
2.9291
2.9285
2.9221
2.9151
2.9124
2.9055
2.9019
2.9004
2.8912
2.8898
2.8877
2.8815

 

 

 

Table A 13 continued

69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
9O
91
92
93
94
95
96
97
98
99
100
101
102
103

 

3.07
3.0698
3.0619
3.0559
3.0595
3.0524
3.0451
3.0403
3.0373

3.034
3.0274
3.0289
3.0198
3.0092
3.0047
2.9947
2.9942
2.9851
2.9843

2.972
2.9693

2.96E+00

2.95E+00
2.9562
2.9515
2.9323
2.9484
2.9202
2.9194
2.9192
2.9144
2.9058
2.9037
2.8916
2.8963

 

131

2.8021
2.8018
2.7967
2.7913
2.7909
2.7883
2.7857
2.7848
2.7821
2.7817
2.7775
2.7758
2.7717
2.7693
2.7687
2.7717
2.7645
2.7654
2.7624
2.7617
2.7603
2.7594
2.7594

 

2.8815
2.8701
2.8753
2.8663
2.8643
2.8631
2.8568
2.86
2.8567
2.8559
2.8531
2.8516
2.8489
2.8451
2.8434
2.8426
2.8381
2.8356
2.8374
2.8341
2.836
2.8309
2.8291
2.8304
2.8273
2.8218
2.8229
2.8248
2.8221
2.8194
2.8236
2.8164
2.8147
2.8149
2.8184

 

 

Table A13 continued

104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121

 

2.8909
2.8834
2.8715
2.8682
2.8661
2.8639
2.8638
2.8546
2.8534
2.8525
2.8459
2.8536
2.8407
2.8264
2.846
2.8366
2.8255
2.8255

 

132

 

2.813
2.8165
2.8123
2.8121
2.8086
2.8079
2.8063

2.808.
2.8064
2.8054
2.8053
2.8023
2.8031

2.803
2.8043
2.8048

2.80E+00
2.80E+00

 

 

Table A 14

Raw data for Figure 5.6

133

 

 

Time (minutes)

 

wwwNNNNNNNNNN-J—I—I—i-I—I—I—I—I—I
N-uomooxicrmawN-aomcoxiarmawNaommVO‘U‘PwN‘

Probe

0.10967
0.10968
0.1 1003
0.10975
0.10878
0.10787
0.10814
0.10894
0.10712
0.10749
0.10594
0.10743
0.10643
0.10521
0.10578
0.10605
0.10535
0.1052
0.10502
0.10407
0.10425
0.10395
0.1027
0.10141
0.10285
0.10166
0.10208
0.10251
0.10077
0.099596
0.099271

 

Concentration

 

Probe

Concentration

0.10913
0.10914
0.10941
0.10784
0.1 1007
0.1077
0.10789
0.10709
0.10628
0.10899
0.10418
0.1068
0.10502
0.10399
0.10404
0.10248
0.10327
0.10599
0.1023
0.10077
0.10337
0.10026
0.098061
0.097969
0.098122
0.09956
0.094771
0.093509
0.094063
0.091 146
0.091321

 

Probe

Concentration
Se-OS M

0.1077
0.10688
0.10673
0.10586
0.10539
0.10442
0.10388
0.10332
0.10263
0.10275
0.10233

. 0.10148

0.10192
0.1009
0.10036
0.10054
0.099034
0.098181
0.099819
0.098767
0.098087
0.098167
0.098093
0.097186
0.095737
0.09571 3
0.095962
0.095597
0.095885
0.094662
0.093684
0.095483

 

 

Table A 14 continued

33
34
35
36
37
38
39
4o
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67

 

0.09689
0.095207
0.095703
0.093525

0.09057
0.087942
0.084946
0.084185
0.081217
0.076223
0.072975
0.068687
0.066235
0.063271
0.059427
0.056619
0.053612
0.052169
0.047728
0.046432
0.043414
0.040943

0.03906
0.036241

0.03461
0.033037
0.029708
0.029182
0.027818
0.025784
0.023862
0.022024
0.021257
0.019074
0.019086

134

 

0.088795
0.087261
0.083709
0.080677
0.077522
0.076777
0.071783
0.068865
0.061886
0.063919
0.060907
0.055243
0.054482
0.049691
0.047704
0.043834
0.043321
0.042043
0.036495
0.03601 1
0.03331 5
0.030909
0.029658
0.028249
0.028183
0.026749
0.023613
0.022186
0.021666
0.017177
0.01 5301

~ 0.014004

0.01578
0.016397
0.012273

 

0.093888
0.092571

0.092878

0.092145
0.089994
0.090251
0.08973
0.088926
0.088048
0.087402
0.085479
0.085241
0.083549
0.08136
0.079779
0.079439
0.077496
0.074042
0.07231 5
0.069782
0.068222
0.065834
0.063187
0.060506
0.058478
0.056458
0.052827
0.0508
0.049971
0.04718
0.045373
0.043529
0.041 518
0.039014

' 0.038187

 

 

Table A 14 continued

68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102

 

0.017013
0.015476
0.015356
0.01354
0.01 1573
0.01 1417
0.010477
0.0095228
0.0091971
0.0082284
0.0081023
0.006582
0.0059395
0.0044793
0.0036083
0.0033799
0.0044541
0.0018657
0.0021626
0.0010999

0.00082795
0.00033224

1.68E-06
1.68E-06

135

 

0.013063
0.01 21 75
0.0097865
0.0095951
0.0071082
0.0089512
0.0078827
0.0045987
0.0081414
0.0052404
0.0053052
0.0054979
0.0027778
0.002612
0.004661 5
0.004621 5
0.0036372
0.0002051
0.0032405
0.0013043
0.00031779
0.0041843
-0.001 3255
0.0020082
-0.001 3985
-0.00092 809
0.00090257
-7.98E-05
-0.0009081 7

-0.0058984 ,

-0.0039394
-0.0033395
-0.00071 599
-0.00073313
-0.0021225

 

0.036181
0.03357
0.033309
0.030608
0.029635
0.028983
0.028515
0.025755
0.02443
0.022689
0.021888
0.021205
0.019377
0.01911
0.017858
0.017196
0.016076
0.014865
0.014629
0.013513
0.014106
0.012281
0.011 172
0.01 1556
0.0090085
0.010012
0.010171
0.010284
0.0091739
0.0078122
0.0078776
0.0072616
0.0058647
0.0048532
0.0056013

 

 

Table A 14 continued

103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121

 

136

 

-0.0080443
-0.0018527
-0.0055465
-0.0060348
-0.00151
-0.0040607

-0.00066716

-0.0025145
0.0012106
-0.0046223
-0.0022702
-0.0013974
-0.0046406
-0.0014245
-0.0050638

-0.00068145

-0.0058775
-7.84E-06
-7.84E-06

 

0.0060486
0.0054515
0.0047842
0.0049872
0.0037122
0.0029977
0.0029724
0.0024399
0.001629
0.0021 135
0.0034755
0.0013731
0.00241 17
0.0017937
0.00055298
0.0020173
0.00013278
2.23E-06
2.23E-06

 

 

ST TE UNIV LIBRQRIE

llllllllllllllllIllllllllllllllll