VA a E....
Anyway-“Mg
car. a
s.

E

..
.5... v
17%..
.«zmo
i

 

 

 

 

 

 

can»:
i ,r.1....x.:;
t: F. v! .i
.1

.3 ladnwflfl. Han Q...” #53.
“J ‘1' (ill.- c

 

 

swig,“

.

 

T:-«'=':!s

.932 WWWMWWWWWWW

3 1293 01812

This is to certify that the

dissertation entitled
DEVELOPMENT OF SECOND HARMONIC GENERATION
AS A TOOL FOR MONITORING CRYSTALLIZATION

presented by

DALE JERRY LECAPTAIN

has been accepted towards fulfillment
of the requirements for

Ph.D. degreein CHEMISTRL

5950. W/

Major professor

Date A; /

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

 

 

LIBRARY

Michigan State

University

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINE return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE

DATE DUE

DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

we WWW.“

Development of Second Harmonic Generation as a Tool for Monitoring

Crystallization

By

DALE JERRY LECAPTAIN

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements

for the degree of

DOCTOR OF PHILOSOPHY

Department of Chemistry

1999

ABSTRACT

DEVELOPMENT OF SECOND HARMONIC GENERATION AS A TOOL FOR

MONITORING CRYSTALLIZATION

By

DALE JERRY LECAPTAIN

Crystallization is a widely used separation and purification process. However, the
large number of variables affecting crystallization makes theoretical prediction of the
process unreliable. Therefore, to obtain a large mean size, uniform crystal size
distribution and efficient separation control of the crystallizer is necessary. This requires
in situ monitoring of the system. The studies presented demonstrate Second Harmonic
Generation (SHG) and Raman spectroscopy as useful techniques for monitoring batch
crystallization.

Raman spectroscopy is demonstrated as a useful method for monitoring the
change in solution structure, as well as the formation of crystals. Potassium dihydrogen
phosphate has a red shift in the symmetric stretch of the phosphorous hydroxyl bond
when transferring from solution to crystalline phase. Using this shift, a desupersaturation

curve can be constructed.

The non-linear optical process of second harmonic generation (SHG) is a novel
technique for this purpose. SHG is more sensitive, can be used in situ, and is less prone
to interference than turbidometric, and light scattering methods. SHG is also orders of
magnitude more sensitive than turbidity, resulting in 5-15 minutes shorter measured
induction times than aforementioned techniques. The technique is demonstrated for
measuring the induction time of an inorganic salt, a sugar, and an amino acid.

The probe configuration designed and discussed provides a practical way to
monitor the crystallization process in situ, noninvasive, and in real time using SHG.
Monitoring the crystallization using SHG provides a means for knowing when to control
it, which is demonstrated on L-lysine monohydrochloride crystallization. A slight
increase in the mean size and a drastic change in the crystal quality occur.

The SHG monitoring probe can differentiate between different polymorphs during
crystallization. The SHG selectively detects polymorph crystal formation. In situ
detection is possible of L-aspartic acid and L-glutamic acid. The former crystallizes in
small particulate form and in a larger more regular form. The latter transforms froma
small particulate form and transforms into the larger more regular form. By monitoring
the crystal formation, relative efficiency of the two polymorphs, and preliminary work
toward the incorporation of light scattering into the SHG analysis is possible.

Ease of sampling and response time of SHG makes it applicable to in situ
determination of enantiomer crystallization. L—aspartic acid crystallization can be

differentiated from DL aspartic acid crystallization using the SHG technique.

ACKNOWLEDGMENTS

First and foremost thanks are to my advisor Prof. Kris Berglund. His guidance
and numerous ideas made this work possible and always kept it interesting. Never would
a conversation end without a ‘few’ new ideas to think about. Also thanks to Prof. Gary
Blanchard who would answer any question I would have. Thanks to the rest of my
guidance committee too, Dean George Leroi and Prof. Alec Scranton.

Next up on the thanks list is Jen. Always there to hear the ‘other’ side of the
story. Mom and Dad whose support was under appreciated, but still strongest when it
counted most. My brothers and sisters, Mom and Dad ‘in-law’, the grandparents, thanks
for listening, talking, supporting, praying, helping, etc.

Number three on the list is the friends that have shared this ‘journey’ and a beer or
two. The list is long and the thanks run deep, but space only permits the ones dearest
indeed. Wendy, thank you, Matt Leanne, ditto. Eric (Killer), Karen, Bob, Tom, Richard,
Jeremy, Shannon, Daniels, LT and the list goes on. Almost forgot Art, whose hours of
counseling have meant a lot.

Last, but not least is those that most certainly need to be thanked and have had the
‘privilege’ of working with me. Dan Engebretson, for making the laser lab navigable.
Dilum and Sanjay, for a discussion or three about crystallization. Prof. Kathy Hunt and
Dr. Tom Carter for those bits of technical information. The rest of the labmates past and
present; Mike, Adam, Carinna, Hasan, Fang, Al, Anna, Parminder, Lilli, Jennifer,
Charles, Trisha, Marketta, Rob, Kitz, Borlan, Chris, Beatrice, and Laurie.

Finally, thank you for reading, I hope you read on to learn the rest of the story.

iv

TABLE OF CONTENTS

List of Tables ................................................................................ viii
List of Figures ............................................................................... ix
List of Abbreviations ...................................................................... xv
Introduction

DEVELOPMENT OF SECOND HARMONIC GENERATION AS A
TOOL FOR MONITORING CRYSTALLIZATION

......................................................................................... 1
1.1 Crystallization .................................................................. 2
1.2 Second Harmonic Generation ................................................ 12
1.3 In Situ Crystallization Studies ................................................ 15
1.4 Overview ....................................................................... 17
1.5 References ...................................................................... 17
Chapter 1

USING RAMAN SPECTROSCOPY FOR MONITORING POTASSIUM

DIHYDROGEN PHOSPHATE BATCH CRYSTALLIZATION
......................................................................................... 21
1.1 Background ..................................................................... 21
1.2 Experimental ................................................................... 22
1.3 Results and Discussion ....................................................... 24
1.4 References ...................................................................... 29

Chapter 2

THE APPLICABILITY OF SECOND HARMONIC GENERATION

FOR IN SIT U MEASUREMENT OF INDUCTION TIME OF

SELECTED CRYSTALLIZATION SYSTEMS
........................................................................................ 30
2.1 Background .................................................................... 30
2.2 Experimental .................................................................. 32
2.3 Results and Discussion ....................................................... 36
2.4 Conclusions .................................................................... 43
2.5 References ..................................................................... 47

Chapter 3

AN IN SITU SECOND HARMONIC GENERATION PROBE FOR
MONITORING BATCH CRYSTALLIZATION OF LYSINE
MONOHYDROCHLORIDE

........................................................................................ 48
3.1 System Design ................................................................. 48
3.2 Application ..................................................................... 52
3.3 Conclusions .................................................................... 61
3.4 References ..................................................................... 61
Chapter 4
DETECTION OF POLYMORPH CRYSTALLIZATION AND
TRANSFORMATION USING SHG
........................................................................................ 62
4.1 Background .................................................................... 63
4.2 Experimental .................................................................. 64
4.2 Data and Discussion .......................................................... 66
4.4 Conclusions .................................................................... 84
4.5 References ..................................................................... 93
Chapter 5
ENANTIOMERIC CRYSTAL DETERMINATION USING SHG
........................................................................................ 94
5.1 Background .................................................................... 94
5.2 Experimental .................................................................. 95
5.3 Data and Discussion .......................................................... 96
5.4 Conclusions .................................................................... 101
5.5 References ..................................................................... 101
Chapter 6
FUTURE WORK; SECOND ORDER NONLINEAR OPTICAL
CHEMICAL SENSORS
........................................................................................ 102
6.1 Introduction .................................................................... 102
6.2 Background .................................................................... 102
6.3 Experimental .................................................................. 103
6.5 References ...................................................................... 106

vi

Conclusion ................................................................................... 107
APPENDIX ................................................................................... 109

vii

LIST OF TABLES

Table A.1 Data for Figure 1.4 .......................................................... 110
Table A2 Data for Figure 2.4 ......................................................... 112
Table A. 3 Data for Figure 3.5 ......................................................... 112

viii

Figure 1.1

Figure 1.2

Figure I.3

Figure 1.4

Figure 1.1

Figure 1.2

Figure 1.3

Figure 1.4

Figure 2.1

Figure 2.2

LIST OF FIGURES

Flow chart depicting the different types of nucleation and the
corresponding subcategories.

..................................................................................... 3

Temperature versus concentration showing various zones of a solubility
curve. The solid line is the saturation limit and the dashed line is the
metastable limit.

..................................................................................... 7

Schematic showing a typical desupersaturation crystallization curve. The
induction time indicates where crystal nucleation is detected in the latent
time period. With crystal present nucleation and growth consume the
available supersaturation causing the rapid drop in concentration. Finally,
the solution returns to saturation.
..................................................................................... 9

Flow chart illustrating the interaction of between size and kinetics for
crystallization.

..................................................................................... 1 1
The experimental set-up including the Kaiser Optical HoloLab 1000
Raman Spectrophotometer and laboratory scale crystallizer.
..................................................................................... 23
The experimental set-up including the Kaiser Optical HoloLab lOOO
Raman Spectrophotometer and laboratory scale crystallizer.
..................................................................................... 26

Stacked plot of Raman spectra for the KDP crystallization. The spectra
are taken at 5 minute intervals starting at the front.

..................................................................................... 27
Peak intensity ratio of the KDP solution peak at 886 cm’1 and the
crystalline peak at 917 cm].
..................................................................................... 28
Experimental set-up used for the SHG experiments.
..................................................................................... 33

Schematic of the experimental batch crystallizer, temperature controlled
with solution pumping action. The operating solution volume is 390 mL.
..................................................................................... 35

ix

Figure 2.3

Figure 2.4

Figure 2.5

Figure 2.6

Figure 2.7

Figure 2.8

Figure 3.1

Typical Crystal Formation of potassium dihydrogen phosphate (8.9%
supersaturated KDP at 30 °C).

Region I: Supersaturated solution, no crystals

Region II: Self-nucleation, crystals starting to form (induction time)
Region III: Rapid desupersaturation, nucleation and crystal growth
Region IV: Return to saturation, no more nucleation or growth

..................................................................................... 38
Isothermal induction times for varying concentrations of KDP
supersaturated solutions. (% Supersaturated is the wt % over the
saturation value.) All measurements were recorded at 297°C with the
same S-minute cool down to supersaturation level.
..................................................................................... 39

SHG monitoring of 35.4% supersaturation a—Lactose crystal formation.
The crystallization was performed isothermally at 20 °C. The maximum
and subsequent decrease in measured intensity is due to scatter from the
large quantity of crystals.
..................................................................................... 41

SHG(O) and light scatter(X) monitoring of L-lysine-HCI. The SHG is
normalized to the amount of SHG detected, whereas the light scatter is the
amount scattered with the saturated solution being the zero measure. The
data shown is 200g lysine m-HCI dissolved into 300g water, and than
cooled to 20 °C. The addition of 30 mL of ethanol occurs at the 5-minute
mark.
..................................................................................... 42

Comparison of the induction time measurement using SHG (O) and light
scatter (X) on crystallization curve of 7.53% supersaturated solution of
KDP.

..................................................................................... 45

Expansion of figure 2.7 focusing on the early time measurement. The
intensity is increased for the two methods to near the data scatter limit.
..................................................................................... 46

General schematic showing the experimental design. A pulsed laser
provides the incident 1064 nm lights measured using a PMT. The
sampling probe is detailed in figure 3.2.
..................................................................................... 49

Figure 3.2

Figure 3.3

Figure 3.4

Figure 3.5

Figure 3.6

Figure 3.7

Figure 4.1

Figure 4.2

Figure 4.3

Laboratory scale (~400mL) crystallizer that has the SHG probe immersed
to show the details of the design. The system design section describes the
components in detail.

..................................................................................... 5 1

Cooling curve for L—lysine monohydrochloride controlled crystallization
shown in figure 3.4. The control occurs at 21 minutes when the cooling is
stopped in response to crystal detection.
..................................................................................... S4

SHG output from a typical L—lysine monohydrochloride crystallization.
Temperature induced crystallization of a 80 g. in 150 mL water/ 100 mL
ethanol solution. The cooling profile is shown in figure 3.4.
..................................................................................... 56

Crystal population size distribution for L—lysine monohydrochloride
showing slightly larger mean size for the controlled (square) experiments
versus the reference (x) crystallization.
..................................................................................... 58

Photo of the L-lysine monohydrochloride crystals sampled from the 390
micron average crystal size tray (mesh #45) of the controlled
crystallization.
..................................................................................... 59

Photo of L-lysine monohydrochloride sampled from the reference
experiment (390 micron average crystal tray). Agglomerates dominate the
sample with a small quantity of uniform crystals produced.
..................................................................................... 6O

Schematic showing the light scattering set-up and equipment.
..................................................................................... 67
SHG signal showing the Anti-solvent (methanol) crystallization of a 1.8%
wt. in water at 50°C of L—aspartic acid (Form A).

..................................................................................... 68

Photo of a sample of L-aspartic acid crystals collected after completion of
the crystallization (Form A).

..................................................................................... 70

xi

Figure 4.4

Figure 4.5

Figure 4.6

Figure 4.7

Figure 4.8

Figure 4.9

Figure 4.10

Figure 4.11

Figure 4.12

SHG output measured for the 1.4% wt. in water L-aspartic acid (Form B)
anti-solvent crystallization at 10 °C. Changing optical properties of the
solution causes the small fluctuation at the start. Through the course of
the rapid crystallization no SHG is produced.
..................................................................................... 7 1
Light scattering curve for L-aspartic acid crystallization (Form B)
measured simultaneously with figure 4.5. At 3 minutes methanol addition
was started and at 28 minutes the crystal formation starts.
..................................................................................... 72

Photo shows the L-aspartic acid crystals (Form B) taken at the completion
of the second experiment. The crystals appear as a very fine powder.
..................................................................................... 73

X-ray powder pattern shows the difference in the two L-aspartic acid
crystals. The gray pattern (upper) corresponds to the non—SHG active
crystal form (Form B, figure 4.6) and the black trace (lower) corresponds
to the SHG active crystal (Form A, figure 4.2).
..................................................................................... 74

SHG crystal formation curve for the anti-solvent crystallization of 2.4%
wt. in water L-glutamic acid at 20°C. Procedure A from the experimental
section was used. Samples were collected and photographed at 17.4
minutes and at the completion of the experiment.
..................................................................................... 76

Photo of (1 crystals and small particles sampled at 17.4 minutes just after
the start of the L-glutamic acid crystallization.
..................................................................................... 77

Photo of L-glutamic acid crystals, taken at the completion of the

crystallization in figure 4.8. Shows the or form of crystals.
..................................................................................... 78

Plot shows the SHG anti-solvent crystallization of 2.8% wt. in water or L-
glutamic acid. Crystal formation starts at 32 minutes. Solution samples
were taken at 21 and 41 minutes.
..................................................................................... 79

Photo of L-glutamic acid small crystals taken at 21 minutes. The
crystallization curve is shown in figure 4.7. The crystals are or form as
determined by x—ray (figure 4.22).
..................................................................................... 80

xii

Figure 4.13

Figure 4.14

Figure 4.15

Figure 4.16

Figure 4.17

Figure 4.18

Figure 4.19

Figure 4.20

Figure 4.21

Figure 4.22

Photo from the L—glutamic acid crystallization sampled at 41 minutes.
Shows predominantly B crystals.
..................................................................................... 8 1

Photo from the L—glutamic acid crystallization, sample was collected upon

completion of the experiment. Again, B crystals dominate.
..................................................................................... 82

SHG crystal formation of 3.3% wt. in water L-glutamic acid at 45°C
generated using procedure C. The monitoring was extended to allow for

transformation from the or form to the B form.
..................................................................................... 85

Photo of predominantly a L-glutamic acid crystals collected at time 69.1
minutes from the crystallization in figure 4.15.
..................................................................................... 86

L-glutamic acid crystals collected at the end of the experiment pictured in

figure 4.15. Photo shows a near equal mix of or and B crystal forms.
..................................................................................... 87

SHG (0) crystal formation and light scattering (x) plotted for L-glutamic
acid formation. The crystallization parameters are described as procedure
C in the experimental section. Methanol addition was started at 3.1
minutes and samples were collected at 28.5 and 50.5 minutes.
..................................................................................... 88

L—glutamic acid crystals sampled at 28.5 minutes. The crystals are a mix

of small particulate, 0t polymorph (prism), with a trace of the B form
(needles).
..................................................................................... 89

L-glutamic acid crystal photographed at 28.5 minutes. The crystal shape
is predominantly the or polymorph.
..................................................................................... 90

L-glutamic acid crystals collected at the completion of the experiment.
Crystal shape is the prismatic or polymorph.
..................................................................................... 91

X-ray powder patterns for L-glutamic acid. Pattern 1 is crystals collected
from crystallization of or form crystallization 4.11, pictured in figure 4.13.
Pattern 2 is prism shape (or form) and pattern 3 is the needle shape as

xiii

Figure 5.1

Figure 5.2

Figure 5.3

Figure 5.4

Figure 6.1

referenced.12 Pattern 4 is crystals collected from 4.11 crystallization and
pictured in figure 4.12.
..................................................................................... 92

The SHG crystal formation curve for L-aspartic acid. Crystal formation
starts at ~15 minutes and reaches depletion of the supersaturation at 23
minutes. Methanol as the anti solvent was used on this 1.9% wt/wt
solution crystallized at 30 °C.
..................................................................................... 97

Photograph of L-aspartic acid crystals collected after the completion of the
crystallization performed in figure 5.1. The crystals are rather small and
difficult to fully contrast in this photo reproduction.
..................................................................................... 98

Photograph of L-aspartic acid crystals collected after the completion of the
crystallization performed in figure 5.1. The crystals are rather small and
difficult to fully contrast in this photo reproduction.

..................................................................................... 99
DL-aspartic acid crystals photographed upon completion of the
crystallization. Distinct crystal shapes are present.
..................................................................................... 100
The schematic of an SHG sensor.
..................................................................................... 104

xiv

“2(1)

LIST OF ABBREVIATIONS

activity of species i
nucleation rate
supersaturation
wavenumber

nth order susceptibility
energy

second harmonic efficiency
crystal growth rate

ionic strength coefficient of species i
hertz

crystal growth rate constant
nucleation rate constant

to the ith power

second harmonic intensity
to the jth power

propagation constant

to the kth power

x-ray wavelength

potassium dihydrogen phosphate
crystal size

L-lysine monohydrochloride
molality of species i
milliliter

suspension density

change in chemical potential
standard chemical potential
chemical potential
population density
nanometer

refractive index at m
refractive index at 2(1)
polarization

peak intensity ratio

gas constant

density

supersaturation

size

second harmonic generation
supersaturation

time

temperature

volume of crystals and solution

XV

agitation rate

frequency
direction

xvi

INTRODUCTION

DEVELOPMENT OF SECOND HARMONIC GENERATION AS A
TOOL FOR MONITORING CRYSTALLIZATION

Batch crystallization is a widely used and efficient separation process for the
production and purification of pharmaceuticals, food, and fine chemicals. Uniform
crystals are desired for product consistency and for easier handling relative to fine
powders. Efficient control of the crystallization process is necessary for obtaining large
crystal mean size, narrow size distribution, and uniform purity. These characteristics
provide good flow for handling and packaging, uniform final product appearance, and an
economically feasible process. If allowed to progress in an uncontrolled manner
crystallization can result in a predominantly fine powder with a wide size distribution.
The crystallizer geometry, operating temperature, stirring rate, etc. all influence the
crystallization process. No single theory of nucleation, growth, or molecular
configuration of supersaturated solutions has a sufficient predictive value for process
design. The independent variables of the process and difficulties in modeling play a role
in the efficiency of control.

The real time monitoring of a crystallization process is critical for efficient
control. To minimize contaminating the system and ensure efficiency, monitoring needs
to be in situ and within the time frame of the phenomena. Many difficulties are

associated with removing small samples of supersaturated solution and trying to make

sensitive measurements without changes occurring making the sample unrepresentative
of the crystallizer. Perturbation to the crystallizing system must be kept to a minimum so
as not to cause unnecessary disturbances that could affect the process and ultimately the
final product. Lastly, the measurements of the crystallization properties need to be made
within the time frame of the phenomenon so that control can be exercised. A long data
collection time would give results too late for useful control.

The importance and challenges of crystallization are introduced above. At this
point a common basis in accepted crystallization principles and terms as used throughout

the remainder of the text is presented.

I.1 CRYSTALLIZATION

Crystallization proceeds by first the formation of nuclei and followed by growth.
The crystallization process is a combined kinetic, thermal, and mass transfer
phenomenon. The two primary steps of crystallization, nucleation and growth, require a
supersaturated solution. A supersaturated solution contains more than the equilibrium
amount of solute at a given temperature.

The first step in the crystallization process is the nucleation of crystals from the
solution. Illustrated in the flow chart are the two types of nucleation. Primary nucleation
is the formation of crystals form solution. Chemical potential overload of the solution
causes homogenous nucleation, whereas foreign particles, which can be in the solution or
part of the crystallizer cause heterogeneous nucleation. Secondary nucleation occurs
when new crystals form from growing crystals in the solution.1 The formation

mechanisms have been studied with no definitive model developed. Though no

Nucleation

/ \

Primary Secondary

/\

Homogenous Heterogeneous

Figure 1.1

Flow chart depicting the different types of nucleation and the
corresponding subcategories.

mechanism is accepted several correlations of nucleation kinetics have satisfactorily
described by experimental data.
B° = kN wi MR (ACk)

Where the nucleation rate (B°) is proportional to the agitation rate (W) in rpm, suspension
density (MT) in mass per volume, and the supersaturation (AC). The dependence on
agitation, suspension density, and supersaturation is realized because most nuclei in an
industrial crystallizer are generated by contact with the crystallizer environment.1

The molecular composition of the solution creates a gray area between nucleation
and growth mechanisms. The small size, randomness, and motion of the solution/slurry
in a crystallizer make direct measurement difficult. The two common theories regarding
nucleation are; l the crystals nucleate as miniature crystals and that there is a diffuse
body of molecules similar to the bulk surrounding that nucleus, 2 no clearly defined
surface or boundary exists from where the nuclei eventually forms.2‘3

The first theory would be an assembly of one, then two, three molecules etc. until
the critical nucleus size forms.4 The primary support for these studies was done using
concentration gradientss, diffusion°, and Raman7. The size of these critical nuclei has
been proposed by Larson and Garside to be clusters 2 to 10 nanometers. Alternatively,
agitation, convection, or even Brownian motion provides the energy necessary to form
critically sized nuclei.°

The second crystal nucleation theory could be considered a hydrate where the
solvent molecules are pushed out for crystal formation. Supporting this Chakraborty and

Berglund show that aqueous sucrose molecules form an envelope of solute, which

encases the water molecules. Therefore, the requirement for nucleation is the expulsion

of the water.8 An alternative way of determining this is by measuring the rotational
reorientation time using fluorescence lifetimes. Cluster size can be approximated and in
turn determine the quantity of solvent trapped. This requires a fluorescent molecule that
binds to the crystallizing species.9

The second step in the crystallization process is crystal growth. The requirements
for growth are the presence of supersaturation and stable nuclei. At the microscopic level
growth rate is irregular, but at the macro level growth rates are regular.16 The growth
process has been thoroughly for several crystal systems.10 Several models such as 2-D
growth, Burton-Cabrera-Frank Model, and the diffusion layer have been developed.
These models provide a theoretical basis for correlation of experimental data for
determination of kinetic parameters. The expression relating growth to supersaturation is

G = kg Ad

This relates the growth rate (G) to the supersaturation (C) level. The equation uses the
linear growth velocity as the basis.1

The induction time is the period of time that elapses between achievement of
supersaturation and the appearance of crystals.1 Two processes, the formation of nuclei
and the growth of crystals to a detectable size, make up this time period. The induction
time is not an experimental characteristic of the system, but rather a way of monitoring a
particular crystallization. A study by Sohnel and Mullin has shown that relative
nucleation and growth rates can be established, but caution should be exercised when
verifying the validity of such measurements. ”'17 Determination of induction time is

entirely dependent on how the crystals are detected. More efficient crystal detection

would measure a shorter induction time. The discrepancy in nucleation theories and
inability of detection make the prediction of induction times nearly impossible.

Chemical potential difference is the driving force of crystallization. Extensive
calculations have been done to relate chemical potential to the free energy of the system
and are based on the classical works of Gibbs’lz’15 The chemical potential is related to
concentration through the following for a given species i.

MT) = IM°(T) + R T In (at)

The chemical potential (m) relates to the standard chemical potential (114°) and activity
(a;). The activity is related through the ionic strength coefficient (Yr) to the molality (mi).
31 = mi * Yr

The molality though not trivial can be converted to the desired concentration units.l3
Calculation of ionic strength is impractical because of the large concentrations associated
with crystallization. Therefore supersaturation is used for reference in a crystallizing
system.

A way to implement the concept of supersaturation is consider a solubility curve
shown in figure 1.1. The saturation line divides the stable region from the metastable
region. Crystal formation cannot occur in the stable region and any crystals present will
dissolve. The upper left region is the labile, where solute must come out of the solution.
Supersaturated solutions exist in the metastable region, where crystal growth occurs. The
further left and above the saturation line, the higher the supersaturation. Factors that
affect zone width are the stirrer speed, cooling rate, evaporation rate, amount of

impurities, the size and number of seed crystals, etc. An important point to keep in mind

labile

concentration

    

stable

 

 

temperature

Figure 1.2

Temperature versus concentration showing various zones of a solubility
curve. The solid line is the saturation limit and the dashed line is the
metastable limit.l4

about the zone width is that it too is a function of the way in which the crystals are
detected.“

Traversing within the metastable region described above dictates the progress of
the crystallization. Figure 1.2 shows the crystallization curve for a batch cooling
crystallization. The latent period consists of a supersaturated solution with no detectable
nuclei present. Nucleation is the first step. The two common methods for providing
crystal surface are in an industrial batch crystallizer are to either seed the supersaturated
solution or increase the supersaturation to cause self-nucleation. Seeding is favorable
because of a lower required supersaturation. The drawbacks of seeding are that an
additional process step and guaranteeing the crystal seed quality are necessary.15 Once
nucleation or seeding has occurred growth can occur. These two processes provide the
means of desupersaturation. Crystal formation returns the solution to saturation, which
will be at equilibrium with the crystals present.

The next consideration for the crystallization is final product. Specifically,
quantifying and understanding the relationship between nucleation, growth, and
supersaturation can be done considering the population balance. For a batch crystallizer
the population balance can be reduced to the following.

5n/5t + V [5(Gn)/6L] =0
Where V is the volume of crystals and solution and is assumed to be well mixed so
population density and supersaturation are constant. G is the growth rate, n the

population density, t is time, and L the crystal size.1°

 

 

 

 

 

 

S
"3 latent period .
a saturation
8
E
o
0 \—
I
t(nucleation)
time
Figure 1.3

Schematic showing a typical desupersaturation crystallization curve. The
induction time indicates where crystal nucleation is detected in the latent
time period. With crystal present nucleation and growth consume the
available supersaturation causing the rapid drop in concentration. Finally,
the solution returns to saturation.

Tying the experimental growth kinetics and nucleation kinetics provide a means
for observing the crystallization. The growth and nucleation ultimately are a function of
the supersaturation or driving force.

G or si B° 0t 31‘
The kinetic orders for these processes i and k are system specific and typically are close
to one for i and between three and five for k. Consequentially, at high supersaturation (s)
the nucleation is favored.17

Ultimately batch crystallization can be broken down as follows. The
supersaturation is the point of external system control and the starting point for this
discussion. The common methods for changing supersaturation levels are through
changing temperature, removing solvent, changing solvent composition, or the reaction
rate. A small supersaturation (low Au) favors growth and a large supersaturation (large
Ap.) favors nucleation. The nucleation rate is kinetically higher order than growth, which
at high supersaturation drives nucleation more. Additionally, a large supersaturation
(large Au) increases the effect of system parameters such as crystallizer shape, agitation,
temperature variations, etc., has on the nucleation. The quantity of nucleation versus
growth determines the population density. The population density is the number of
crystals per unit volume of the given size L. A large population density will have a large
surface area. This large surface area permits rapid depletion of the supersaturation.
Ideally a small surface area is desired so the small number of crystals can grow. But if

the supersaturation is high, nucleation again starts to dominate.17

lO

lcontrol

 

 

 

 

supersaturation (s)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

growth nucleation crystal
rate rate surface area
G 0; si B° 0c 81‘
I i < k ]
v
population
density
1 CSD
Figure 1.4

Flow chart illustrating the interaction of between size and kinetics for
crystallization.l7

11

The premise for obtaining large mean crystal size requires keeping the
supersaturation low to limit the nucleation and promote crystal growth. As an example a
self-nucleation system needs to start with a high supersaturation for nucleation to occur.
As soon as nuclei are present the supersaturation must be lowered so the growth can

dominate. The strategy will produce larger crystals than an uncontrolled case.

1.2 SECOND HARMONIC GENERATION

Second Harmonic Generation (SHG) is the nonlinear conversion of two photons
(frequency to) to a single photon (frequency 2(1)). It is the interaction of the
electromagnetic wave dipole and the dipole of the medium. The medium has to have a
dipole within the coherence length of the electromagnetic radiation for this phenomenon
to occur. SHG is an optical method, therefore it can be used in situ and it is noninvasive
to the system being measured. A review article by Corn and Higgins illustrates the
extensive development SHG has had in the past fifteen years as an analytical tool for
probing surfaces and interfaces.18 In particular, the experimental requirements for liquid

interface studies”’20

are closely related to a batch crystallization setting. The key to the
process is the absence of inversion symmetry within the coherence length of the laser
pulse.

A practical understanding of light propagation through a dielectric medium is
necessary for determining applicability of SHG. As an electromagnetic wave, described

by Maxwell’s equations, propagates through a dielectric medium a linear polarization is

induced in the medium:21

12

P = x“) E
Where P is the polarization of the medium, x") is the susceptibility, and E is the energy of
the electromagnetic wave. When the intensity of the incident electromagnetic wave is

large enough a nonlinear response occurs in the medium. The resultant power series
expansion of the polarization is shown below.
P = x“) E + x“) E*E + x‘” EE*E +

Inherently, the magnitude of )0 is much greater than 75(2) , which is much greater then x“).
Therefore, in large electromagnetic field intensities the x0) response can be detected.

Substitution of the equation for an electric field (E = E0 cos (wt - kz)) illustrates
the harmonic frequencies at 2(1), 3(1) as well as (u of the incident wave. Where (1) is the
frequency, t is time, k the propagation constant, and z the direction.

P = 76" E0 cos (wt - kz) + 1/2 x”) E02 [1 + cos (th - 2kz)] + higher order

The second order term has two components, a DC response and a frequency doubling
response. The doubled frequency (2(0) is the component of interest for optical second
harmonic generation.22

The conservation of energy requires that the second order (962)) response occur
only in anisotropic media. A symmetric medium has the same x”) response for a positive
and a negative electric field, therefore no net polarization occurs. Conversely, a non-
symmetric medium has a different x0) response for a positive and negative E-field,
producing a x”) response. A )6” response is evidenced by the generation of the doubled
light frequency.22

The intensity of the second harmonic is dependent on several factors. The

inherent properties of the crystal are constant. The second harmonic intensity is 120,, p is

13

the density of the crystal, and S is the size of the crystal. The Efficiency of the crystal is
a function of the phase matching and the wavelength being used.

hwapSEfi
A phase matching crystal system has the propagation constant at the second harmonic
equal to the propagation constant at the incident wavelength. This can be related to the

refractive indexes. Therefore when refractive index of the medium at (r) (110,) equals the

refractive index of the medium at 2(1) (mm) the system is phase matched. Each face of a
crystal has a different refractive index, which makes the intensity dependent on the face
of the crystal being illuminated. The case where 1]“, does not equal 1120, is the non-phase
matching condition. These may still be SHG active, but now depend heavily on the
coherence length of the electromagnetic field. 23 The two alternative non-phase matching
methods that exist in literature are Cerendov radiation and quasi-phase-matching type I
and H respectively. Type I uses lithium nibate waveguides where the end-faces are
coated for high reflectivity of both incident and second harmonic and the temperature is

24

tuned to form a resonator at both frequencies. Type II involves a media where the

refractive index is large enough so the second harmonic is radiated into the bounding
medium.2°‘2°

The number and the size of particles also contribute to the SHG intensity. The
polarization and incident energy are related quadratically, therefore the increase in size
will go with the square of the intensity. The intensity of the second harmonic increases
linearly with the number of crystals.31

A nanosecond or picosecond Nd/Y AG laser produce an intense enough electric

field for the second harmonic process to occur. The 1064 nm line is easiest to obtain and

14

therefore most often chosen, but is not required. Additionally, because of the frequency
difference between the fundamental and second harmonic, filtering is quite efficient.
Also the response of a conventional PMT is more efficient at 532 nm than at 1064 nm.
The data collection necessary is the change in intensity of the doubled frequency.

There has been extensive work done on non-linear optical materials for the
purpose of laser optical material development. As the availability and use of laser
expanded so did the drive for understanding non-linear optical materials.27'28’29’3° The
original work by Kurtz and Perry demonstrated SHG from crystal powder systems.31
This was expanded upon in the development of a screening method of potential crystal
systems for laser optics.32'33 A review of the nonlinear efficiencies of organic crystals by
Oudar and Zyss emphasizes this point.”35 This powder technique was extended for use
as a screening tool for determining the presence or absence of symmetry as a prelude to

x-ray crystallography calculations.”3°

1.3 IN SITU CRYSTALLIZATION DETECTION

In situ as defined by the Webster Dictionary is “in the natural or original
position”.37 This common definition does not provide adequate detail for specific ideas
relevant to industrial batch crystallization, therefore it is expanded for this work. When
using the term throughout the text the following definition needs to be kept in mind. The
basic requirements that needs to be met are contact with the system should be kept to a
minimum, no sample removal or excessive handling, and the response needs to be

instantaneous relative to the experiment time scale. Additionally, from a practical

standpoint the technique should be robust and sensitive to handle industrial conditions.

15

Monitoring crystallization can be accomplished through a variety of methods.
Measuring the exothermic response due to the heat of crystallization is impractical due to
the scale factor and measuring refractive index is impractical on slurries. Optical
methods are non-intrusive and can offer a higher degree of reproducibility. A sampling
of these studies include FT IR-ATR3 8, fluorescence”, and Raman spectroscopy.40 These
techniques may require various chemometric methods as they measure the
supersaturation level. This direct link to the driving force provides intricate information
about the crystallization. The variety of methods available lends themselves to
potentially different systems, as development of a variety of techniques is necessary since
no single technique works for all systems.

Instead of measuring the supersaturation level the appearance of crystals can be
measured. The intuitive methods for measuring the crystals are visual and light
scattering.“ Visual monitoring is analytically inefficient and light scattering suffers from
interference from bubbles or other foreign particles in the system. These problems cause
lower sensitivity and incorrect measurements.

Conventional methods for monitoring the onset of nucleation are based on light
scattering, diffraction, or reflection. These techniques are prone to interference from dirt,
air bubbles, changing solution consistency or a combination therein, which are very
prevalent in industrial crystallizers. The physical differences of compounds and different
types of crystallizers42 also make a universal technique for monitoring the process

difficult.

l6

1.4 OVERVIEW

Monitoring the formation of crystals can provide a way to refine models and aid
in the control of crystallization. The novel proposed application of SHG provides a
potentially better way for monitoring crystallization.

Laser optic research forms the basis for SHG monitoring of crystals. The focus
there was to find materials for development of laser materials using large single crystal
systems or well graded and prepared powders in refractive index controlled laboratory
settings. The initial groundwork here was the design and set-up of systems that could be
used in an industrial crystallization. Viscous and slurry solutions needed to be controlled
for exposure with the laser light necessary for the second harmonic process.

The work is laid out roughly in the order of development. The exception is
chapter 1, which deals with monitoring crystallization, but employs the use of Raman
spectroscopy instead of SHG. Chapter 2 is the initial demonstration and proof of the
technique applicability. In Chapter 3, refinements are presented and demonstrations of
the technique in control are shown. Chapters 4 and 5 demonstrate the versatility of SHG
for monitoring crystallization as polymorphic systems are studied as well as enantiomeric
crystal systems. The final chapter is proposed research using SHG as an analytical sensor

employing some of the groundwork laid out here.

1.5 REFERENCES

I Myerson, A. S. and Ginde, R., H_andbool;of Industrial Crystallization, Edt. Myerson,
A. S, Butterworth-Heinemann, Boston, 1993

ZJense, A. H. and De Jong, E. J., Trans. IChemE., 1978, 56, 187

17

3Mullin, J. w. and Jancic, S.J., Trans. IChemE., 1979, 57, 188

4 Larson, M. A., Advances in Industrial Crystallization, Garside, J., Davey, R. J., and
Jones, A. G.: edt., Butterworth-Heinemann Ltd., Oxford, 1991,

SMullin, J. w. and Leci, c. L., Correspondence, 1969, 1075; Larson, M. A. and
Garside,J., J. of Cryst. Growth, 1986, 76, 88

°Larson, M. A. and Garside, J., Chem. Eng. Science, 1986, 41, 1285

7Frost, R. L. and James, D. W., J. Chem. Soc., Faraday Trans. 1, 1982, 72, 3235;
Rusli, I. T., Shcrader, G. L., and Larson, M. A., J. of Cryst. Growth, 1989, 97, 345;
Cerreta, M. K. and Berglund, K. A., J. Cryst. Growth, 1987, 84, 577

°Chakraborty, R. and Berglund, K. A., J. ofCryst. Growth, 1992, 125, 81

9 Rasimas, J. P., Berglund, K. A., and Blanchard, G. J., J. Phys. Chem, 1996, 100: (42)
17034

1° Baird, J. K., Hill, S. C., Clunie, J .C., J. of Cryst. Growth, 1999, 196, 220; Him, S.
J.and Myerson A.S., Indust. Eng. Chem. Res., 1996, 35, 1078; Mydlarz, J. and Jones,
A. G., Chem. Eng. J. Bioch. Eng., 1994, 55, 69; Garside, J ., Gilbilaro, L. G., and
Tavare, N. 8., Chem. Eng. Sci, 1982, 37, 1625; Joshi, M. S., Antony, A.V., J. of
Phys. DAppl. Phys., 1977, 17, 2393

”Sbhnel, o. and Mullin, J. W., J. ofColloid and Interface Science, 1988, 123, 43

'2 Gibbs, J. W., Collected WorksfiVolume 1, Thermodynfl'cs, Yale University Press,
New Haven, 1948

13 Myerson, A. S., Harndbook of Industrial Crystallization, Edt. Myerson, A. S,
Butterworth-Heinemann, Boston, 1993

M Mersmann, A., Crystallization Technology Hgdbook, Marcel Deker Inc., New York,
1995, Chaps. 1&2

15 Mullin, J. W., Cgstallization, Butterworth-Heinemann, Oxford, 1993, Chaps. 1-5

1° Wankat, P. C., Rate Controlled Separations, Elsevier Science Publisher, London,
1990, Chaps. 3&4

’7 Berglund, K. A., findbook of Indusfll Crystallization, Edt. Myerson, A. S,
Butterworth-Heinemann, Boston, 1993

18

18Com, R. M. and Higgins D. A., Chem. Rev., 1994, 94, 107

‘9 Conboy, J. C.,Dashcbach, J. L., and Richmond, G. L., J. Phys. Chem, 1994, 98,
9688

2" Kott, K. L., Higgins, D. A., McMahon, R. J., and Corn, R. M., J. Am. Chem. Soc.,
1993, 115, 5342

2’ Shen, Y. R., The Principles of Nonlinear Optics, John Wiley & Sons, Inc., New
York, 1984, pp. 1-11

22 Prasad, P. N. and Williams, D. J ., Introduction to Nonlinear Optical Effects in
Molecules & Polymers, John Wiley & Sons Inc., New York, 1991

23 Agrawal, G. P. and Boyd, R. W., Contemporary Nonlinear Optics, Academic Press
Inc., San Diego, 1992, Chaps 1 & 2

2" Taniuchi, T., and Yamamoto, K., CLEO Digest, Paper WR3, 1986
25 Lim, E. J., Fefer, M. M., and Byer, R. 1..., Electron. 1m, 1989, 25, 174
2° Webjorn, J ., Laurel], F., and Arvidsson, G., IEEE J. Lightwave Tech, 1989, 7, 1597

27 Edited by Chemla, D.S. and Zyss, J ., Nonlinear Optical Properties of Organ—i0
Molecules m Crystals Volume 1, Orlando, Academic Press, 1987

28 Wagniere, G. H., Linear and Nonlinear Optical Properties of Molecules, Weinheim,
New York, 1993

29 Marder, S. R., Editor Sohn, J. E., and Stucky, G. D., Materials for Nonlinear Optics
Chemical Perspectives, Washington DC, 1991

 

3° Prasad, P. N. and Williams, D. J., Introduction to Nonlinear Optical Effects in
Molecules and Polymers, New York, Wiley, 1991

3‘ Kurtz, s. K. and Perry, T. T., J. oprpl. Phys., 1968, 39, 3798
32 Dougherty, J. P. and Kurtz, s. K., J. Appl. Cryst., 1976, 9, 145
33 Coda, A. and Pandarese, F., J. Appl. Cryst., 1976, 9, 193

3" Zyss, J. and Oudar, J. L., Phys. Rev. A, 1982, 26, 2016

19

35 Oudar, J. L. and Zyss, J., Phys. Rev. A, 1982, 26, 2028
3“ Doughery J. P. and Kurtz, s. K., J. Appl. Cryst., 1976, 9, 145

37 Webster’s Ninth New Collegiate Dictionary, Merriam-Webster Inc., Springfield,
1983

3“ Dunwilla, D. 13., Carroll, L. 13., and Berglund, K. A., J. Cryst. Growth, 1994, 137,
561

3" Chakraborty, R., and Berglund, K. A., J. Cryst. Growth, 1992, 125, 81
40Cerreta, M. K. and Berglund, K. A., J. Cryst. Growth, 1990, 102, 869-876
4' Chrisman, R. W., McLachlan, R. D., and Hamer, R. S., US Patent # 4,672,218, 1997

42 Bennett, R. C., Handbook of Industrial Crystallization, Edt. Myerson, A. S,
Butterworth-Heinemann, Boston, 1993

20

Chapter 1

USING RAMAN SPECTROSCOPY FOR MONITORING POTASSIUM
DIHYDROGEN PHOSPHATE BATCH CRYSTALLIZATION

Raman spectroscopy is a useful method for monitoring batch crystallization.
Practical monitoring of an industrial process, such as crystallization, requires the
measurements be in-situ, done in real time, and are nonintrusive to the process. The
measurement needs to be easy to obtain where no special sample preparation is necessary
which would introduce error and delay the results. Recent advances in Raman
instrumentation enable the in situ study of crystallizing systems while meeting the
aforementioned requirements. These developments include the use of transmission

gratings, the availability of high quality notch filters, and sensitive CCD cameras.

1.1 BACKGROUND

Raman spectroscopy is an optical technique that relies on inelastic light scattering
from molecules. The resultant spectra are a measure of the polarizable vibrational
energies of the particular molecules involved. Additional information discussing the
theory and practice of using Raman spectroscopy can be found elsewhere.1 The other
major topic of this work is crystallization, which is the separation of crystalline material
from a solution of that particular species. Here again the theories and principles of the

process can be found in the introduction and in greater detail elsewhere.2

21

The application of this technique is demonstrated on potassium dihydrogen
phosphate (KDP). Previous work has been done within the research group3 and

elsewhere"'5 ’°'7

on understanding the growth kinetics of dihydrogen orthophosphates.
This work demonstrates the ability to monitor the KDP crystallization using Raman

spectroscopy. The system can be monitored in real-time, in situ, and without

perturbation.

1.2 EXPERIMENTAL

The experimental set-up is illustrated in figure 1.1. The key features in
monitoring the crystallization are the instrument used and the crystallizer. The
instrument permitted rapid and sensitive spectra collection and the batch crystallizer is
laboratory scale with a sighting window.

The crux of this work is the spectrometer that was used. Manufactured by Kaiser
Optical (Ann Arbor, MI) the HoloLab 1000 uses a 30 rnilliwatt HeNe laser, a 7 fiber (1
excite, 6 collection) fiber optic probe, and a holographic transmission grating. The
resolution is approximately 8 cm'1 and the focal length of the probe optic is 6 inches.

The crystallizer is a laboratory scale (400 mL) water-jacketed glass reactor vessel.
The optical window is 3 cm in diameter. A marine style impeller provided the agitation
of the solution of Aldrich ACS grade KDP which was 30% wat in in-house reverse
osmosis water. The crystallizations were done isotherrnally at 30°C with 1 second

exposure time for 25 scans.

22

 

Crystallizer

Figure 1.1

The experimental set-up including the Kaiser Optical HoloLab 1000
Rarnan Spectrophotometer and laboratory scale crystallizer.

23

1.3 RESULTS AND DISCUS SION

KDP was chosen as an example system because of prior knowledge of it Raman
spectrum.3 By using the differences in the crystalline and solution Raman spectra a
collection of the changing peaks can be ratioed. This provides a measure of the
crystallization process as it occurs.

The monitoring of KDP crystallization centers on the spectra of both the crystal
and solution phases (figure 1.2). The solution spectrum has a water band around 3400
cm". Two prominent features in this spectrum are the P-(OH)2 symmetric stretch at 886
cm’1 and the O=P=O symmetric stretch at 1086 cm". The interesting feature of the
crystal spectrum is the P-(OH)2 symmetric stretch at 917 cm'l. When going from
solution to crystalline phase the P-(OH)2 symmetric stretch shifts 31 cm'1 and the O=P=O
symmetric stretch disappears. A combination of these changes provides a tool to monitor
the decrease in solution concentration and an increase in the presence of crystals. The
increasing crystalline peak and decreasing solution peak are illustrated in figure 1.3.
Shown is the region centering on the peaks discussed above. The spectra are plotted
along a time axis of approximately 5 minute intervals.

By ratioing the solution KDP peak at 886 cm'1 and the crystalline peak at 917
cm'1 a plot is constructed. The ratio of these peaks as a function of the crystallization
time produces a "fingerprint" of that particular batch. This "fingerprint" measures the
induction time, rate of concentration decrease, the rate of crystal formation, and the

ending point of the crystallization. Additionally, Raman spectroscopy can measure both

24

the solution phase and crystal phase. Providing a means for monitoring the solution
concentration, while simultaneously collecting information about the crystals forming.

The discussion would not be complete without considering the limitations of
using Raman for monitoring crystallization. The dominant limit is the particular crystal
system and its ability to Raman scatter. The Raman spectrum of the crystallizing species
needs to be clear of solvent Raman scatter and have a high enough efficiency. A way to
circumvent this would be using resonance Raman and employing chemometric data
analysis. In comparison to the SHG technique for monitoring induction time, Raman
spectroscopy is expected to be less efficient. As will be demonstrated in the next chapter,
SHG has a lower detection limit than Raylaigh scattering. Therefore, SHG would be
expected to have lower detection limits than Raman spectroscopy too.

Raman spectroscopy can be used to monitor KDP crystallization. Ratioing the
intensity of the solvated molecule and crystalline molecule provides a “fingerprint” of the
particular batch crystallization. This “fingerprint” contains the induction time, relative
concentration, solid characteristics, and an ending point of the batch crystallization being

monitored.

25

Intensity

450

 

 

 

 

 

 

 

400 ~
350 ~
300 -
250 ~ saturated
solution
200 -
150 . crystals
100 -
‘ m.»
50 -
o 1 1 1 1 1 1 1
0 500 1000 1 500 2000 2500 3000 3500

Raman Shift (cm-1)

Figure 1.2

Raman spectra of KDP in crystalline and saturated solution taken using
the instrument in figure 1.1.

26

4000

Intensity

 

mg”..- '
:3. _

 

 

 

 

0
-§
I . [*4
. . I“. a
‘Ilw Xvi '11 /‘K‘
f. w jig, ill V 5‘.de
\ fl . I.
1\ f‘
I / J ‘\
//\\\ - fj/ ‘\ / \\f\\V~M,_,..W_e-Ir”f// \m
I I I
5 00 1 000 1 5 00

Raman shift (cm")

Figure 1.3

Stacked plot of Raman spectra for the KDP crystallization. The spectra
are taken at 5 minute intervals starting in the front.

27

 

PIR (886cm-1 /917cm-1)

0.8

0.6

0.4

0.2

 

 

 

 

 

 

 

 

» O

-80300008g8.;°9.;8888 A

0 ° 0 o o ‘

b O . O O

_ - . o ‘
. 5
0.0000 “
O ‘ .

P .“ ..aaai
10 1 5 20 25 30 35 40 45 50 55 60
Time (minutes)

Figure 1.4

Peak intensity ratio of the solution KDP peak at 886 cm'1 and the
crystalline peak at 917 cm}. Inset shows repetitive runs of KDP
crystallization.

28

1.4 REFERENCES

l Colthup, C. P., Daly, L. H., and Wiberley, S. E., Introduction to Infrared aid Raman
Spectroscopy Third Edition, Academic Press, San Diego, 1990

2 Mullin, J. W., gystallizafion 3'° Edition, Butterworth-Heinemann, Oxford, 1993

3 Cerreta, M. K. and Berglund, K. A., "The Structure of Aqueous Solutions of Some
Dihydrogen Orthophosphates by Laser Raman Spectroscopy", J. Cryst. Growth, 1987,
84, 577

4 Van Enckevort, W. J. P., Rosmalen, R. J., and Van Der Linden, W. H., J. Cryst,
Growth, 1980, 49, 502

sBarata, P. A. and Serrano, M. L., J. Cryst. Growth, 1996, 160, 361
°Mullin, J. w. and Amatavivadhana, A., J. Appl. Chem, 1967, 17, 151

7Paul,B. K. and Joshi,M. s., Ind. J. Pure &ApplPhys., 1976, 14, 544

29

Chapter 2

THE APPLICABILITY OF SECOND HARMONIC GENERATION
FOR IN SIT U MEASUREMENT OF INDUCTION TIME OF
SELECTED CRYSTALLIZATION SYSTEMS
Second Harmonic Generation (SHG) is proposed for measuring crystal formation
in supersaturated solutions. SHG is able to detect some crystalline material in a liquid

medium. The focus of this paper is to describe the in situ and noninvasive technique for

measuring the induction time of batch crystallization process as using SHG.1

2.1 BACKGROUND

Crystallization is the formation of crystalline material (solid phase) from solution
(liquid phase). Crystallization is composed of nucleation and crystal growth. The
chemical potential of a supersaturated solution is the driving force, where a small
supersaturation favors growth and a large supersaturation favors nucleation. The optimal
crystallization is to have an appropriate number of nuclei present to permit crystal
growth. The relationship between these two processes is much more complex and a
detailed description appears elsewhere.1 The two common methods for introducing

nuclei in an industrial batch crystallizer are to either seed the supersaturated solution or

1 This chapter was adapted from previously published work; LeCaptain, D. J. and

Berglund, K. A., J. of Cryst. Growth, 203, 1999, 561
30

increase the supersaturation to cause self-nucleation. Seeding is favorable because of the
lower supersaturation required, therefore favoring growth and ultimately larger mean
crystal sizes. The drawbacks of seeding are that an additional step is necessary and
guaranteeing the crystal seed quality is necessary. Conversely, a large supersaturation is
necessary to cause self-nucleation, but this large chemical potential increase the effect of
system parameters such as crystallizer shape, agitation, temperature variations, etc., has
on the nucleation. Significant batch variations result, making it very difficult to predict
the nucleation.2

Nucleation is the formation of crystalline species from a supersaturated solution.
The time from a supersaturation level to formation of the first nuclei is the true induction
time. The experimental or measured induction time is the true induction time plus the
time necessary for the crystals to grow to a detectable size. Therefore the method of
detection plays an important role in the induction time measured.3 The more sensitive the
technique the closer to the measured time is to the true induction time. Additionally,
nucleation is very dependent on the system conditions, therefore induction time is not
constant for different batches. The importance of induction time relates back to the
previous paragraph on nucleation versus growth. Ultimately the in situ measuring of
induction time aids in the efficient control of the crystallizer and the better technique
would be more sensitive.

Induction time measurements can be done using the non-linear optical technique
of SHG. Examples of the previous studies on powders suggest the potential of using
SHG for monitoring crystallization.4’5’° Since the advent of lasers, the efficiency of non-

centrosymmetric crystals as frequency doublers has been thoroughly studied. A review

31

of the nonlinear efficiencies, how well a species responds in the second order process, of
organic crystals by Oudar and Zyss emphasizes this point.7’8 In particular studies of the
second harmonic from powders demonstrate the possibility of use in solutions.4 SHG is
the nonlinear conversion of two photons (frequency (1)) to a single photon (frequency 2(1)).
It is the interaction of the electromagnetic wave dipole and the dipole of the medium.
The medium has to have a dipole within the coherence length of the laser pulse for this
phenomenon to occur. N anosecond and shorter laser pulses are required for high energy
density, which are necessary for the non-linear response to occur. The key to the process
is size and rigidity of the dipole species. Foremost, a dipole is necessary, meaning a
centrosymmetric structure. The dipole species also has to be large enough and properly
structured to couple with the light wave to produce the second harmonic. The
randomness of the bulk solution causes it to be SHG inactive.

The noncentrosymmetric requirement for the SHG process limits the number of
crystal systems. It is necessary to have a noncentrosymmetric crystal structure. Large
portions of batch crystallizations are performed on pharmaceuticals and fine chemicals,
which are generally biological or biologically related molecules. These types of
molecules tend to be noncentrosymmetric, producing noncentrosymmetric crystal

structures and therefore ideal for SHG detection.

2.2 EXPERIMENTAL
The Experimental set-up is diagrammed in figure 2.1. A Quanta-Ray DCR laser
was used to produce an average 12 to 14 millijoules per pulse energy at 1064 nanometers

(nm). The pulse width was approximately fifteen nanoseconds at a frequency of 10 Hz.

32

 

Nd/Y AG 1064 nm \

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

laser \
dichroic
monochromator /[|J
532 nm nl ”
PMT U! V:
sam 1e
/l§ - p
filter
Figure 2.1

Experimental set-up used for the SHG experiments.

33

The turbidity measurements were done by averaging the transmitted intensity at 490 nm
from a Guided Wave fiber optic absorbance instrument.

The crystallizer was designed to mimic an industrial crystallizer with agitation in
a forced circular configuration and temperature control to 103°C by a NesLab water
circulator. A 1-inch-impeller turning at 650 rpm pumped the solution. A high power
dichroic mirror and glass filter were used to filter the incident light (1064 nm) from the
generated second harmonic light (532 nm). The second harmonic was measured using a
Spex monochromator and 1106 Hammatsu PMT. A SRS Gated Boxcar Averager
averaged the collected second harmonic intensity over either 3 to 10 seconds. Collection
optics and PMT settings were held constant. The data were collected every 15 to 30
seconds by a short program in National Instruments LabView software.

The potassium dihydrogen phosphate (KDP), potassium chloride (KCl), Citric
Acid, chromium chloride (CI'CI3) L—lysine monohydrochloride (lysine m-HCl), and were
used as received from Aldrich Chemical Company. KDP was chosen for the majority of
the studies because of availability and good second harmonic efficiency. Lysine m-HC]
and 0t lactose were chosen to demonstrate the technique on an amino acid and a soluble
organic compound. The solution concentrations were measured by percent mass in
water. The water used was the building tap, reverse-osmosis water. The percent

supersaturation (G) was calculated as follows:

6 = (wt % of solution / wt % @ saturation - l)*100

The weight percent saturation was extrapolated from solubility data. 2

34

 

 

 

 

 

 

 

 

     
 

 

 

L
thermistor ‘ , -\
I ‘. .
\; “d’ : impeller
i‘ ._. /e/
water i '1’ “fi :
jacket 5 : : E
\E j i E
1 ' ' I
I ' : 1
I i : . I
,--J E : OLD 5
I I ' ,
" "1 I i I.--
‘\\ ' .1 ‘- T ' 3"
\ I‘ '1
\ l
1 ’1
quartz I,”
cell
(1cm x 1cm)

Figure 2.2

Schematic of the experimental batch crystallizer, temperature controlled
with solution pumping action. The operating solution volume is 390 mL.

35

All data collection was performed under isothermal conditions. The samples were
weighed and then dissolved in the crystallizer by elevating the temperature to 45°C
(above the saturation level) for approximately 1 hour. The solution was cooled at the
maximum rate possible taking approximately 3-5 minutes to reach 300°C 1 O.2°C which
was maintained for the duration of the experiment. The induction time measurement was
started when the solution temperature reached 301°C. The lysine m-HCl was

crystallized by ethanol anti-solvent crystallization.

2.3 RESULTS AND DISCUS SION

This technique can measure the presence of crystals in a liquid medium by
detecting the produced second harmonic. Nucleation and growth of the crystals causes
the increase in second harmonic light produced. Detection in bulk solution is possible
because of the non-centrosymmetric symmetry requirement for SHG. This symmetry
requirement also eliminates interference from the bulk solution; foreign particles such as
air bubbles or impurities, which are not SHG active. The crystals are the medium
necessary for SHG, therefore no probe molecules or special sample treatments are
necessary, which could potentially perturb the system. Another advantage of SHG is
detection over a dark background, allowing for detection that is more sensitive.

A brief explanation of specific SHG theory indicates why this process works.
SHG requires a non-centrosymmetric medium within the coherence length of the incident
wavelength, otherwise phase cancellation of the second harmonic will occur. When the
particle is larger than the coherence length, surface SHG effects can be measured.9 The

particle (crystals in this case) are smaller than the coherence length, therefore the second-

36

order light scattering process called hyper-Rayleigh scattering is a more likely
explanation. Hyper—Rayleigh scattering occurs when density and orientation fluctuations
of the crystal species in bulk solution disrupt the phase cancellation previously
discussed.10

The crystal formation of KDP monitored using SHG is shown in figure 2.3. The
results show the second harmonic intensity increasing from zero baseline (region I) to the
normalized full-scale detection limit of the apparatus (region IV). Region H shows the
first change in signal. The left is the measured induction time for the process. Region
III occurs after formation of adequate surface area (adequate number of nuclei) and is the
desupersaturation of the solution. This is shown by the steeply sloped region and is due
to a combination of growth and nucleation. Finally, after substantial crystal formation
the supersaturation is depleted and the solution reaches its saturation level again (region
IV), with no additional crystal growth or nucleation.

The next step in demonstrating this technique is reproduction of the induction
time measurement. Figure 2.4 illustrates the induction time measurement of various KDP
supersaturation that were isothermally self-nucleated. The induction time is defined here
as the change in signal exceeding the 95% confidence of the baseline in the crystal
formation curve. Clearly, a mathematically more rigorous method could be employed,
but the inherent discrepancies in induction time measurement do not make it necessary at
this time. As expected, the measured induction time increases as the percent
supersaturation of the solution decreases. The increase in the range of induction times as
the percent supersaturation is decreased demonstrates the need for in situ monitoring.

This variation is expected and parallels similar published data on KDP induction times.11

37

SHG Intensity
g)
.5

(18

.9
o~

 

 

 

 

 

 

 

 

O O O
q9329 Q? 95% ‘3 ‘b<2
0 o o 0 0 cu;
m “9,0%%% $0 0% 000 m
0
«9° °o
I [I 12?
"is
m0
0 0
Q10
O
0%
O
O
‘b
[H IV
10 20 30 40 50 60
'Thne(nfinufi5)
Figure 2.3

Typical Crystal Formation of potassium dihydrogen phosphate (8.9%
supersaturated KDP at 30 °C).
Region I: Supersaturated solution, no crystals
Region H: Self-nucleation, crystals starting to form (induction
time)
Region IH: Rapid desupersaturation, nucleation and crystal growth
Region IV: Return to saturation, no more nucleation or growth

38

70

Time (minutes)

80

20

 

 

 

 

 

 

 

5 6 7 8 9 10 ll 12 l3 l4

% Supersaturation

Figure 2.4

Isothermal induction times for varying concentrations of KDP super saturated
solutions. (% Supersaturated is the wt % over the saturation value.) All
measurements were done at 297°C with the same 5-minute cool down to

supersaturation level.

39

A concern of this technique that needs to be addressed is the effect of the laser
light used. A second harmonic intensity from a saturated solution containing crystals was
measured by varying the laser power from complete signal attenuation (0.9 mJ/pulse) to
the threshold of the filtering optics (1.5 mJ/pulse). The change in signal varied linearly
with the measured signal through this limited range that is well below the energy used in
the laser crystal induction studies cited above. Additionally at high laser light energies
perturbation of the system could be a concern. Garetz and coworkers have demonstrated
static urea crystal nucleation induced by laser light.12 Coinciding with this would be
sample degradation. To address this concern KDP crystallization was run under twice the
laser energy used for the measurements and no perturbation of the system was observed.
This can be justified by the large sampling volume used and the motion of the solution.
The combination of those two negates the concerns of localized heat build up or sample
degradation.

Additional crystal systems were studied to demonstrate the versatility of this
technique. Figure 2.5 and 2.6 are the SHG measurements of or lactose and lysine m-HCl
respectively. Lactose a very soluble organic material, lysine an amino acid, and KDP an
inorganic salt are a sampling of the range of materials this technique can be used for.
Note as the lactose is crystallizing a maximum intensity is reached and than begins to
decrease. This is due to the large amount of crystalline material that scatters both the
incident laser beam and SHG produced. SHG (O) and light scatter (X) monitoring of
lysine m-HCl. The SHG is normalized to the amount of SHG detected, whereas the light
scatter is the amount scattered with the saturated solution being the zero measure. The

data shown is 200 g. lysine m-HCl dissolved into 300 g. water, and than cooled to 20 °C.

40

SHG Intensity

 

%
g s
. ‘31.
0.8 ~ 8 is
19
o it
3’ 8
O 0
° to
0.6 . 5’ 8
4’ 8
O
8 0
a B
O
04 ~ 3
' 8
§
0
O
8

0.2 7 i k

l 1 I 1 1 1 1

 

0 20 40 60 80 100 120 140

Time (minutes)

Figure 2.5

SHG monitoring of 35.4% supersaturation a—Lactose crystal formation.

The crystallization was performed isothermally at 20 °C. The maximum

and subsequent decrease in measured intensity is due to scatter from the
large quantity of crystals.

41

 

 

 

.3

Q
I
O
O

Intensity
C

 

 

 

Time (minutes)

Figure 2.6

SHG(O) and light scatter(X) monitoring of L-lysine-HCl. The SHG is
normalized to the amount of SHG detected, whereas the light scatter is the
amount scattered with the saturated solution being the zero measure. The

data shown is 200g lysine m-HCl dissolved into 300g water, and than
cooled to 20 0C. The addition of 30 mL of ethanol occurs at the 5-minute

mark.

42

The addition of 30 ml. of ethanol occurs at the 5-minute mark. It can be seen that a large
amount of light is scattered due to the intermixed phases of concentrated water and the
fresh ethanol, but no crystalline species are present until approximately 7 minutes later.

An area to be addressed is the lack of symmetry present at crystal to solution
interfaces. To demonstrate that SHG in not produced by the solid to liquid phase
separation and is indeed only the crystal the crystallization of KCl and citric acid were
monitored. Both have centrosymmetric crystal structures that won't produce the second
harmonic. Since no second harmonic was observed, it is concluded that the boundary
region also does not produce the second harmonic.

The final area studied with this technique is in comparing it to light scattering
measurements. SHG and light scatter were simultaneously used to monitor KDP
crystallization. The light scatter instrumentation is described previously and measures
percent transmittance. By subtracting the percent transmitted from 100% T of the under
saturated solution the percent of light scattered is obtained and this is what is plotted.
Figure 2.7 shows the KDP crystallization curve being monitored by both techniques. It
appears the techniques are very comparable, but the increased sensitivity available from
the SHG apparatus can be seen in Figure 2.8. By expanding the axis and looking at the
induction time, it is apparent that the SHG is more sensitive to the crystals forming.

Therefore, the SHG method measures a shorter induction time.

2.4 CONCLUSIONS
Second harmonic generation is demonstrated as a technique for monitoring crystal

formation in supersaturated solutions. This method is advantageous over turbidometric

43

methods as it able to measure the crystal formation in situ and shows a more sensitive
response. The availability of laser systems and advances in laser design make this

method viable for numerous industrial applications.

Intensity

 

 

 

 

 

80

Time (minutes)

Figure 2.7
Comparison of the induction time measurement using SHG (O) and light

scatter (X) on crystallization curve of 7.53% supersaturated solution of
KDP.

45

Intensity

0.001

 

0.0009 -
0.0008 .
0.0007 -
0.0006 —
0.0005 ~ 916’
0.0004 - 0
0.0003 — O ‘26 °
0.0002 ~ x

0.0001

 

 

 

 

0 5 10 15 20 25 30
Time (minutes)

Figure 2.8
The early time frame expanded from figure 2.7. The intensity change for

SHG is much greater than the light scattering and the SHG has an earlier
detection of the induction time.

46

2.5 REFERERENCES

' Berglund, K. A., Edited Myerson, A. S., Handbook of Industrial Crystallization,
Butterworth-Heinemann, Boston, 1993, 89-101

2Mullin, J. W., Crystallization, Butterworth—Heinemann, 1993

3 Otakar, s. and Mullin, J. W., J. Coll. Int. Science, 1988, 123, 43
4Kurtz, s. K. and Perry, r. T., J. oprpl. Phys., 1968, 39, 3798
SDougherty, J. P. and Kurtz, s. K., J. Appl. Cryst., 1976, 9, 145
°Coda, A. and Pandarese,F., J. Appl. Cryst., 1976, 9, 193
7Zyss, J. and Oudar, J. L., Phys. Rev. A, 1982, 26, 2016
°Oudar, J. L. and Zyss, J., Phys. Rev. A, 1982, 26, 2028

9 Wang, H., Yan, E. C. Y., Borguet, E., and Eisenthal, K. B., Chem. Phys. Lett., 1996,
259, 15

1° Terhune, R. w., Maker, P. 1)., and Savage, C. M., Phys. Rev. Lett., 1963, 14, 681;
Clays, K. and Parsoons, A., Phys. Rev. Lett., 1991, 66, 2980

11 Barata, P. A. and Serrano, M. L., J. Cryst, Growth, 1996, 160, 361

’2 Garetz B. A., Aber, J. E., Goddard, R. G., Young, R. G., and Myerson, A. S., Phys.
Rev. Lett. L, 1996, 77, 3457

47

Chapter 3

AN IN SITU SECOND HARMONIC GENERATION PROBE FOR
MONITORING AND CONTROLLH‘IG BATCH CRYSTALLIZATION

The probe for measurement of SHG is described for immersion into a crystallizer.
The demonstration of the SHG technique for monitoring crystal formation appears in the
previous chapter and elsewhere.1 The present work demonstrates that the SHG probe can
be used in-situ and in real time during crystallization, thus enabling control to be
exercised. The probe is described in detail and control of L-lysine-monohydrochloride

(Lmhc) crystallization is presented.

3.1 SYSTEM DESIGN

The SHG probe described is designed for in situ, real time measurements. The
system was constructed from standard modules and equipment to demonstrate the
feasibility of the technique and the system.

A schematic of the system is shown in figure 3.1. A Coherent Infinity Laser
produces the incident beam (1064 nm). The pulses are approximately 4-6 nm wide at a
repetition rate of 25 pulses per second. The pulse energy was measured to be 2 to 3
millijoules in front of the probe. Second harmonic light detection is accomplished using

a Heath Photomultiplier Module, model EU-701-30. The PMT voltage was 900V and

48

 

detector

 

 

 

 

 

 

 

 

 

 

NdlYag _ dichroic
laser ' mirror
sampling
probe

Figure 3.1
General schematic shows the experimental design. A pulsed laser

provides the incident 1064 nm lights measured using a PMT. The
sampling probe is detailed in figure 3.2.

49

was sampled with a SRS BoxCar Averager. Each datum is the back weighted average of
100 laser pulses. The average PMT output signal is transferred to LabView software on a
PC where the intensity is stored and plotted versus the time from experiment start.

The probe design enables SHG measurements to be made in a batch crystallizer.
Figure 3.2 is a diagram of the probe and how it operates in a crystallizer. The 400 mL
crystallizer has a water jacket for iO.3°C temperature control by a NesLab water
circulator. The laboratory scale crystallizer uses a l-inch marine style impeller turning at
650 rpm for agitation. The SHG probe transmits the 1064 nm incident light into the
sample and separates the generated second harmonic (532 nm) from the incident for
transmission to the PMT. The high power dichroic mirror reflects the incident laser beam
into the glass waveguide (1cm x 15 cm glass rod). The dichroic mirror is reflective to
1064 nm light, but transmissive to 532 nm light. The glass waveguide channels the
incident laser beam into the bulk of the solution, avoiding the surface or edges of the
crystallizer where perturbations or inconsistencies tend to occur. The path length or area
being sampled is the distance between the end of the waveguide and hypotenuse of the
leg-mirrored prism. The incident beam (1064 nm) and the majority of the generated
second harmonic (532 nm) are co-linear, therefore are reflected by the mirrored prism.
The legs of the reflecting prism are mirrored because the similarities in refractive index
of the solution and the glass prism prevent efficient reflection using a regular prism.
Additionally, the mirrored surfaces of the prism are not exposed to the solution, which
prevents scratching. Transmissivity differences of the solutions can be accounted for by
adjusting the path length. The glass waveguide transmits the generated second harmonic

either reflected from the first pass through the solution or upon the second pass to the

50

collection

   
 

 

 

 

 

t 1‘
r I
lass ._ , ,
‘ 1‘
. r s
W V ' : 2 cm
. 3 .
a egu1 e t .
3 ’ . .1 . ._ - -_ . .
. -.o.,~.-.-¢_ -Er'~. (5,;
" l . ' ‘ ' I .l. | . .‘m
A‘ - v . .. . . . 1r ‘Q. . w,
I ‘ tit-.159." 1,. T' I -4 p 111.. ,7
r' ‘2. '1”... 1 z. ..
a - v r .. 4 n g . .1:
In? , . " ‘ a . ‘Iz‘d ‘ ‘ 'r‘ l
X» “1‘1" Q I . l ’
7» ‘ u or. I
. a .

(15cm x 1.2 cm) 1Impeller

 

 

 

 

Figure 3.2

Laboratory scale (~400mL) crystallizer that has the SHG probe immersed
to show the details of the design. The system design section describes the
components in detail.

51

dichroic mirror. The dichroic mirror transmits the 532nm light and reflects the residual
incident beam (1064 nm) into a beam dump. The transmitted second harmonic light is
focused into an Oriel Liquid Light Guide and transmitted to the PMT. The separation
efficiency of the second harmonic from the incident by the dichroic mirror is not 100%
but the residual 1064 nm light is filtered by the liquid light guide. The liquid light guide
is very efficient in the transmission of 532 nm light and very inefficient for the 1064 nm
light. The light guide has a large acceptance angle allowing for easy alignment and

flexibility similar to fiber optic cable allowing for versatility.

3.2 APPLICATION

The design and ultimately the application of the SHG probe encompasses crystal
systems that lack inversion symmetry. Pharmaceuticals and fine chemicals tend to be
chiral and complex molecules that in turn tend to form crystals that meet this symmetry
requirement. The potential of the technique can be realized by considering the numerous
crystal species that have been quantitatively determined in powder for'rn.2’3 The main
consideration is whether the SHG efficiency of a particular species is great enough to be
measured. The refractive index of the crystal and solution at both the incident
wavelength and second harmonic wavelength determine the efficiency of the second
harmonic. From a practical point of view the testing of each crystal system will likely
prove the easiest and most reliable way to determine the viability of the SHG probe. The
SHG probe described above works for or lactose, potassium dihydrogen phosphate, urea,
d-Sorbitol, and L-glutamic acid, but this by no means is an inclusive list. The SHG

powder efficiency of these molecules is in the relative magnitude of KDP as previously

52

reported.3 The crystallization of these molecules is not shown, rather are mentioned to
emphasize the broader applications possible.

By monitoring the Lmhc (L lysine monohydrochloride) crystallization the mean
crystal size and the size distribution can be improved. A Lmhc solution was cooled in an
uncontrolled manner (~ 2°C / min.) to serve as a reference to the controlled experiments.
The controlled crystallization was done the same except the cooling was changed in
response to the presence of crystals. Obtaining larger crystals requires favoring crystal
growth over nucleation, thus a low supersaturation is desired. Growth cannot occur
without first some crystal nucleation, but nucleation needs a high supersaturation.
Therefore as soon as crystals are present stopping the increase in supersaturation (stop the
cooling) will allow more crystal growth and less nucleation.

The similarities for the two experiments consisted of dissolving 80 grams of
Lmhc into a mixed solvent of 150 mL water and 100 mL ethanol. The solution was
heated to 80°C and than rapidly cooled to 50°C where the time was started. For the
reference experiments the cooling was continued until the temp reached 5°C. The
cooling rate was approximately 2 degrees centigrade per minute. The slurry was filtered,
dried, and sieved. The filtering, drying and sieving methods remained constant for all
samples analyzed. The filtering consisted of Whatman number 2 paper in a Buchner
funnel. The crystals were rinsed with cold ethanol and allowed to dry in room conditions
for 24 hours. The sieves were standard Fisher Brand Scientific 8 inch sieves, mesh
numbers 16, 25, 30, 35, 40, 45, 60, and 80 with a sieving time of 25 minutes.

The controlled experiments were cooled until evidence of crystal formation

appeared in the SHG at which point the cooling was stopped. Because of signal to noise

53

Temperature (deg. C)

50

 

 

 

I I 1 1 l

 

10 20 30 4O 50

Time (minutes)

Figure 3.3
Cooling curve for L-lysine monohydrochloride controlled crystallization

shown in figure 3.4. The control occurs at 21 minutes when the cooling is
stopped in response to crystal detection.

54

 

considerations an approximate 20% increase in the SHG signal was the criteria used for
determining the presence of crystals. Figure 3.4 shows a typical cooling curve. A slight
temperature elevation occurs due to the heat of crystallization. After stabilization of the
SHG signal occurred cooling was again started and continued to 5°C when the slurry was
filtered, dried, and sieved. The SHG data are shown in figure 3.5. Note the appearance
of the second harmonic at 22 minutes, which corresponds to the point when cooling was
halted and the temperature slightly increased. The large quantity of crystalline material
present after nucleation causes the SHG photons to be scattered and the SHG signal lost.
The preceding procedure was repeated and used for obtaining the crystal size distribution
data shown in figure 3.5. The population distribution is calculated from the following
equation.4
N=W/(kv*p*L3)

Where N is the number of crystals, W the mass of crystals, k. is the volume shape factor,
p is the density, and L is the average crystal size. The volume shape factor
approximation for a sphere or cube is 6.5 The average size between the upper and lower
sieves of a tray approximates the crystal size for that tray. Figure 3.6 shows the crystal
size distribution for the controlled and reference crystallizations of Lmhc. Note, the
controlled and reference crystal mean size data points are offset for clarity of the figure.
The controlled experiments (square) have a slightly larger mean crystal size (316 microns
versus 303 microns) than the reference (x) experiments. Also, the deviation for each size
is smaller for the controlled crystallizations.

Visual inspection of the crystals shows the importance of the control of the

crystallization. Figure 3.7 is a photograph of the crystals produced in the controlled

55

SHG Intensity (arbitrary)

0.8

 

 

 

 

0.7 r
0.6 ~
0.5 — 3’ °
crystal °
0-4 " detection
0 o
0.3 r o
O
°o
0.2 r :00 0
° % OM98»
‘b W $0 0.0:w?%000%°° :03 30%‘2183; qgfih 090 0 0°00 :9
0.1 09950:? 000 vipe‘gpo o o 9
° 0
0 1 1 1 1
0 10 20 30 40 50

Time (minutes)

Figure 3.4

SHG output from a typical L-lysine monohydrochloride crystallization.
Cooling crystallization of 80 g in 150 mL water/ 100 mL ethanol solution.
The cooling profile is shown in figure 3.4.

56

experiments and figure 3.8 is a photograph of the reference experiment. The samples
photographed were collected from corresponding sieve sizes (#45 mesh number) and
handled as described above. Controlling the crystallization produces crystals with a
uniform shape. The reference, uncontrolled, experiment photo shows a mix of crystals
and agglomerates (figure 3.7). Recalling that agglomerates are assemblies of smaller
crystals the results of the size distribution need to be interpreted differently. The
reference experiments have a represented larger mean size than actually exists because of
the agglomerates. The lack in crystal uniformity will be more evident when the
agglomerates easily break down with handling.

The relative efficiency of this configuration versus the previously described
system (Chapter 2) needs to be addressed. Experimentally the probe is more versatile.
The optical path is not interfered with when samples are changed, but the additional
optics necessary cause an estimated 25% decrease in the second harmonic collection.
Modifications to the probe would permit a more efficient, robust, flexible, and cost
effective device. In replacement of the glass waveguide that is submerged into the
solution a bifurcated fiber optic could be employed to transmit the incident 1064 and then
collect the generated 532 nm light. Focusing optics would need to be incorporated at the
end of the fiber bundle. Either a PMT or a high speed photodiode could handle
measurement of the generated second harmonic. Practical consideration of the laser
indicates that the research grade laser used here exceeds the requirements for the probe.
A lower power, air cooled laser would be more economically feasible and convenient.

Thus, the entire apparatus could be substantially reduced in size and increased in rigidity.

57

Number of Crystals (relative)

3 .0E+05

 

2.5E+05 a T

2.0E+05 a

1.5E+05 ~ EZI

 

 

1.0E+05 1 ‘11 '

5.0E+04 . <

 

 

 

 

0.0E+00 ‘ 1 I I % 31 13: 1 1 13:
0 100 200 300 400 500 600 700 800 900 1000

Crystal size (microns)

 

Figure 3.5
Crystal population size distribution for L-lysine monohydrochloride

showing slightly larger mean size for the controlled (square) experiments
versus the reference (x) crystallizations.

58

 

Figure 3.6

Photo of the L-lysine monohydrochloride crystals sampled from the 390
micron average crystal size tray (mesh #45) of the controlled
crystallization.

59

 

Figure 3.7

Photo of L—lysine monohydrochloride sampled from the reference
experiment (390 micron average crystal tray). Agglomerates dominate the
sample with a small quantity of uniform crystals produced.

60

3.3 CONCLUSIONS

The probe configuration discussed provides a practical way to monitor a
crystallization process in situ and in real time. Considering the cited potential crystal
systems highlights the applicability of this approach. Monitoring the crystallization
provides a means for knowing how to control it. Control is demonstrated for Lmhc
where a modest increase in the mean size and a drastic improvement in the crystal quality

OCCUI'.

3.4 REFERENCES

'LeCaptain, D. J., and Berglund, K. A., J. ofCryst. Growth, 203, 1999, 561
2Delfino,M., Mol. Cryst. Liq. Cryst., 1979, 52, 271-284

3 Edited by D.S. Chemla, J. Zyss, Nonlinear Optical Properties of Organic Molecules
and Crystals Volume H, Orlando : Academic Press, 1987

4 Randolph, A. D., and Larson, M.A., Theory of particufilate processes 2°° edition,
Acedemic Press Inc., San Diego, 1988

5 Mullin, J. W., Cgstallization, Butterworth-Heinemann, 1993

61

Chapter 4

DETECTION OF POLYMORPH CRYSTALLIZATION AND
TRANSFORMATION USING SHG

Polymorphic crystalline materials pose a challenge to industrial processing,
particularly in the pharmaceutical, food, and fine chemical industries, characterizing. A
particular polymorph can be formed during crystallization and subsequently transform to
a different polymorph, usually through solvent mediation. The nonlinear optical process
of second harmonic generation (SHG) can be used to monitor polymorph formation and
transformation in situ because each polymorph has a different second harmonic
efficiency and if one polymorph has inversion symmetry no signal will be obtained at all.
Ostwald’s law of stages states that the least stable form appears first, thus making the
transformation a concern.1

SHG can detect crystal formation and has been demonstrated in chapters 2 and 3.
Polymorph discrimination of crystal detection is used for L-aspartic and L-glutamic acid
crystallization. The experiments show formation of an agglomerated crystal polymorph
that is not SHG active whereas the other crystalline form is SHG active. Data are

presented for polymorphic formation and transformation of crystalline L—glutamic acid.

62

4.1 BACKGROUND

Polymorphism started to receive extensive consideration in the 1950’s with the
advent of modern drug chemistry. Caution needs to be exercised when discussing
polymorphs because of the tendency to mistake precipitation with polymorphism.
Discussions of both are provided to avoid this misconception.

Polymorphs are identical in liquid, solution, and gas states, but differ in the
crystalline solid state. These different crystal structures often have different physical
properties such as melting point, density, shape, etc.2 Polymorphs are of great concern in
the drug industry where tableting, solubility, bioavailability, stability in suspension and
overall stability are affected.3

A problematic area of polymorphism is phase transitions. The metastable form
may be desired, but often transforms to a stable phase. Conversely, the stable phase may
be desired in which case the transformation is needed.4’5‘° The transformation can occur
in either the solid phase7 or in the solution phase. The latter is solvent-mediated phase
transformation where the metastable form is able to dissolve and nucleate as the stable
forrn.°

Precipitation is another method for obtaining solid material from a solution system.
Unlike crystallization where a large mean crystal size is desired, precipitation results in a
fine powder, which may not be crystalline. The process tends to involve relatively
insoluble material, which can result in highly supersaturated solutions. This high
supersaturation will favor nucleation and produce a large number of particles.
Precipitation is important for finished products that need the characteristics of fine

particle size. Examples include dyes and paints.9

63

Differentiation of these two separate processes is possible through a variety of
methods. If the crystals are large enough to see with magnification they will have distinct
straight edges and clear crystalline shapes. Additionally, crystals can diffract x-rays
where a precipitate or amorphous material will not.9

Final consideration is given to the applicability of SHG for monitoring polymorph
systems. Delfino gives a sampling of the number of biological systems that are SHG
active.1° likewise, a publication by Borka and Haleblien lists thousands of compounds
that have been reported in literature having polymorphs.11 Biological molecules tend to
be complex in terms of low symmetry properties, which in turn tend to crystallize without
inversion symmetry. Lack of inversion symmetry is the requirement for being SHG

active.

4.2 EXPERIMENTAL

Two chemical compounds, L-aspartic acid and L-glutamic acid, were chosen for
this study. Both were used as purchased from Aldrich (98%) and Sigma (99%),
respectfully. The measurement of SHG employs the SHG probe discussed previously in
chapter 3. It is used here with no adaptations except designation of a standardization
method as described in the following. Additionally, the conditions and procedure for the
individual experiments used for this study are detailed in this section.

A method for standardizing the probe apparatus was developed. PMT voltage was
set to 900V with 156 mW of 100 Hz. 1064 nm light measured in front of the probe.
Using 62.5 grams of Spectrum KDP in 250 ml of water the output intensity was

measured. Stirring rate and temperature (20°C) were held constant.

64

The procedure used for crystallizing L-aspartic acid is as follows. (Form B) A
1.4% wt. in water was heated to 70°C, and then cooled to 10°C. Upon reaching 10°C the
time and the addition of methanol to the solution were started. The methanol addition
rate was approximately 4 mL/min. The crystallization of L-aspartic acid (Form A) was
accomplished by heating a 1.8% solution in water to 80°C and than cooling to 50°C. It
was held at 50°C for 30 minutes and then methanol was added (~4 ijmin.) during the
following time intervals; 5.4 — 6.4 min., 8.4 — 9.4 min., and 11.4 — 32.7 min.

The L-glutamic acid crystallization loosely follows previous work done by
Kitamura.12 His work depicted two different polymorphs, potential conditions for
forming each, and a pathway for solvent mediated transformation. These two
polymorphs were produced.

The following four procedures were used for crystallizing and transformation of the
L-glutamic acid. Procedure A; a 2.4% wat in water solution was heated to 70°C and
than cooled to 20°C when the timer was started. At 5 minutes the addition of methanol
was started at a rate of 4 mL/min. Procedure B involved heating a 2.8% solution in water
to 80°C and then cooling to 45°C. The timing was started immediately and at 4.1 minutes
methanol addition (~4 mL/min.) was started. Procedure C; a 3.3% wt/wt solution in water
was heated to 80°C and than cooled to 45°C when the time was started. Procedure D is
identical to procedure C except, the concentration was 2.8% and at time 3.1 minutes
methanol was added (rate of 4 ml/min.).

The measurement of light scatter was incorporated into the experiment for
comparison. Averaging the scatter from a Metrologic Neon Laser at 632.8 nm was done

using the fiber optic detection of a Guided Wave Absorbence Instrument. The

65

crystallizer was equipped with a cell window in the side through which the HeNe laser
was shown. The intensity of scattered light was measured above the solution surface
using the aforementioned instrument.

The x-ray powder patterns were collected on a Ragiku instrument. The x-ray
source was a copper rotating anode using the ka wavelength. The values were collected
at reflection 0 and 20. Instrument settings were x-rays 100 microamps and 45 millivolts

with all spectra collected at room temperature.

4.3 DATA AND DISCUSSIO N

Two amino acids previously known to be SHG active13 were chosen for the current
study. The first, L-aspartic acid is crystallized in two separate experiments. The second,
L-glutamic acid is crystallized using the four procedures listed above. These experiments
deal with monitoring L-glutamic acid crystal formation, relative efficiency for the two
polymorphs, and preliminary work towards incorporating light scattering in the analysis.

First, the SHG monitoring of L-aspartic acid crystallization was performed. Figure
4.2 shows the SHG monitoring of L-aspartic acid crystal formation, crystallized using the
procedure listed in the experimental section. The crystallization follows the typical
crystal formation curve when using the SHG technique as described in chapter 3. Upon
completion of the experiment the solution was filtered and the crystals dried. A
photograph of the resulting crystals (Figure 4.3) and x-ray data (Figure 4.8) verify that
the solid is indeed crystalline.

Next, L-aspartic acid is crystallized under a second set of conditions (Form B),

which results in a different polymorph. Figure 4.4 shows the second harmonic output

66

 

 

 

 

 

 

 

 

 

crystallizer

Figure 4.1

Schematic shows the light scattering set-up and equipment.

67

SHG Intensity

0.3

0.25

0.2

0.15

.0
p—

0.05

-0.05

 

 

 

 

. oo o
o o
0 o
_ ’9 . o
o co . 000
0’0
5 o
o
P °vooo.
o’o.oo.
o
o co
c .~o’~o
o .00..
o 0
° 0 so. 009’. O.
1.0’ o .s.°.~.~ ~°~ °co° ‘0 ' °
0
0 5 10 15 20 25 30 35

Time (minutes)

Figure 4.2

SHG signal showing the Anti-solvent (methanol) crystallization of a 1.8%

wt. in water at 50°C of L-aspartic acid (Form A).

68

40

from this crystallization experiment. No discernible second harmonic signal is detected.
Reference to the light scatter of a HeNe laser provides qualification of the increasing
presence of solid in the solution. Light scattering (figure 4.5) shows the initial solution
clouding occurs with the start of methanol addition. It then records the increase in solid
formation, which starts at 25 minutes. Photograph (figure 4.6) and x-ray diffraction
shown in figure 4.7 verify the difference in crystal structure. This particular crystal form
does not frequency double. Either the change in crystal structure or extreme small size of
the crystals results in the rapidly crystallizing form (Form B) not to be SHG active.
Comparison of figure 4.3 and 4.5 proves SHG can be used for selectively detecting a
particular polymorph. The crystallization of the second form (Form B) is not SHG
active, therefore the difference between the two crystallizations can be detected.

The second portion of this chapter addresses polymorphic determination of L-
glutamic acid polymorphs. The first experiment demonstrates using the SHG technique
on L—glutamic acid crystallization. The next two experiments monitor transformations
from one crystalline form to the other. The final crystallization is dedicated to
groundwork in the combination of SHG monitoring with light scattering.

L-glutamic acid has two polymorphs.12 The or form pictured in figures 4.9 and 4.10
was crystallized using procedure A from the experimental section. Figure 4.8 shows the
crystal formation curve. At 17 minutes the crystal formation starts and then finishes
around 25 minutes. Figure 4.9 was collected at the onset of crystallization (17.4
minutes). The photo is a mix of CI crystals and other smaller particulate. Figure 4.10,
collected at the completion of the crystallization shows the crystals have grown in size

and the finer crystals are no longer present. The crystallization progresses from a mix of

69

 

Figure 4.3

Photo of a sample of L—aspartic acid crystals collected after completion of
the crystallization (Form A).

70

SHG Intensity

 

0.3

0.25 r

0.2 ~

0.15 r

p
.—
I

0.05 _

o MW-‘WMWW

 

 

1

 

_005 1 1 1 1 1
0 10 20 30 40 50 60 70

Time (minutes)
Figure 4.4
SHG output measured for the 1.4% wt. in water L-aspartic acid (Form B)
anti-solvent crystallization at 10 °C. Changing optical properties of the

solution causes the small fluctuation at the start. Through the course of
the rapid crystallization no SHG is produced.

71

Intensity Light Scattered

 

 

 

 

 

/./
O
O.
I
F O
.0
.0
O
1. O
O
.0
.06
i” ’ \
mo 0
0 10 20 30 40 50 60 70

Time (minutes)
Figure 4.5
Light scattering curve for L-aspartic acid crystallization (Form B)

measured simultaneously with figure 4.5. At 3 minutes methanol addition
was started and at 28 minutes the crystal formation starts.

72

 

500 um

l—i

Figure 4.6

Photo shows the L-aspartic acid crystals (Form B) taken at the completion
of the second experiment. The crystals appear as a very fine powder.

73

Intensity

 

a

 

 

 

 

 

 

40

A I 1‘ j 1 i r ‘ Ii 1
“WMHIIMIIHIHIH
. * 2 * 1 I
5 10 15 20 25 30 35
x-ray diffraction

0

Figure 4.7

X-ray powder pattern shows the difference in the two L-aspartic acid

crystals. The gray pattern (upper) corresponds to the non-SHG active
crystal form (Form B, figure 4.6) and the black trace (lower) corresponds

to the SHG active crystal (Form A, figure 4.2).

74

small particles and crystals to just crystals upon completion. The particle formation is
scrutinized more closely in the following crystallization.

The importance of monitoring the transformation from one crystal form to another
needs to be stressed. Most solution based techniques that measure supersaturation cannot
detect transformations, nor will light scattering or visual techniques differentiate these
crystals.

Two samples were collected from different times during the crystallization
(procedure B), filtered, and than photographed. The final solution was also filtered and
the solid photographed. Photos of the samples are shown in figures 4.12, 4.13, and 4.14
respectively. The SHG monitoring of the system is shown in figure 4.11. Note, the
crystal formation occurs at approximately 35 minutes.

The first sample (50 mL) was collected at 21 minutes, figure 4.12 shows the solid
that resulted after filtering and drying. A small particulate crystalline material is the only
solid present, which is not SHG active (figure 4.11). The x-ray powder pattern (Figure
4.23) indicates this is the or form. At 32 minutes into the experiment crystal formation is
evidenced by the SHG increase. Figure 4.13 taken at 41 minutes verifies the presence of
crystals. Upon completion, the solution was filtered, dried, and photographed (figure
4.14). Like figure 4.13, this photo shows the presence of crystals. The dominant crystal
form is B. The small particulate or form is not SHG active but the larger B form is.

The formation of the B polymorph of L-glutamic acid is demonstrated in Figure
4.15. Procedure C was used for this experiment. The crystal formation starts around 60
minutes. The photo shown in Figure 4.16 taken from the sample collected at 69.1

minutes shows predominantly polymorph or with slightly different shape characteristics

75

 

 

(14
0.35 . '

O O. 9‘

0.3 “ .0

(125 r ’

SIIGPInhmmfity
c:
as
o

(115 r cs

0.1

(105 r

 

 

 

0 5 10 15 20 25 30 35 40
11nme(nfinuues)

Figure 4.8
SHG crystal formation curve for the anti-solvent crystallization of 2.4%
wt. in water L-glutarnic acid at 20°C. Procedure A from the experimental

section was used. Samples were collected and photographed at 17.4
minutes and at the completion of the experiment.

76

 

Figure 4.9

Photo of or crystals and small particles sampled at 17.4 minutes just after
the start of the L-glutamic acid crystallization.

77

 

ISOOumI

Figure 4.10

Photo of L—glutamic acid crystals, taken at the completion of the
crystallization in figure 4.8. Shows the or form of crystals.

78

SHG Intensity

0.4

0.35

0.3

0.25

.o
t0

0.15

0.1

0.05

-0.05

 

 

 

 

e o
o
00
o
” o ’ 00’.
0 ° 0 o
00 '.Q
o
._ 0" . O... . 0
o 0 °
9 o
.0
r o
‘00
. .00
.00
o
E o
6‘
e 0’
b. o .0”
° I
P ’den. o w”
0 10 20 30 40 50

Time (minutes)

Figure 4.11

Plot shows the SHG anti-solvent crystallization of 2.8% wt. in water a L-
glutamic acid. Crystal formation starts at 32 minutes. Solution samples

were taken at 21 and 41 minutes.

79

60

 

Figure 4.12

Photo of L-glutamic acid small crystals taken at 21 minutes. The
crystallization curve is shown in figure 4.7. The crystals are or form as
determined by x-ray (figure 4.22).

80

 

Figure 4.13

Photo from the L-glutamic acid crystallization (Figure 4.11) sampled at 41
minutes. Shows predominantly B crystals.

81

 

Figure 4.14

Photo from the L—glutamic acid crystallization, sample was collected upon
completion of the experiment (Figure 4.11). Again, B crystals dominate.

82

than the previous photo. The solution was then monitored for extended time to allow
solvent mediated transformation to the B form of crystals (figure 4.17). The photo is
predominantly the B form. The SHG decreases over time as the crystals transform from
the Otto the B form. From Kitamura’s work, the on form has a higher solubility than B.
What can be concluded is the or form has a higher SHG efficiency. The lower the
solubility of B means more B crystals are formed than with the or form yet the SHG
decreased. Confirmation of the crystal morphology was done by x-ray powder
diffraction (figure 4.22) and parallels Kitamura’s results. 12

The experiment indicates the polymorphs have a different SHG efficiency. The
two polymorphs have identical crystal symmetry and differ on in the spacing within the
crystal lattice. 14‘”

The final crystallization procedure (D) brings to light the difficulties trying to
quantify the SHG technique for two polymorphs. The crystals formed are not the clean
prism shape of the low temperature anti-solvent crystallization. Rather, they are a rough
hexagonal shape, pictured in figures 4.19 and 4.20. The SHG monitoring of the
crystallization and simultaneous light scattering monitoring is shown in figure 4.18.
Approximately 50 mL of sample were collected at 28.5 minutes and 50.5 minutes for
photos 4.19 and 4.20 respectively. Figure 4.21 shows the resulting solid sampled at
completion of the experiment.

These results highlight the difficulty in simultaneous monitoring with light scatter
and SHG. As each sample is drawn for verification of the solution, the light scatter drops

in intensity due change in solution conditions. SHG on the other hand is not affected as

83

much by the changes. This highlights the benefit of SHG being less prone to solution

interference.

4.4 CONCLUSIONS

SHG can be used to differentiate between crystal polymorphs. In situ detection is
possible for L-aspartic acid and L-glutamic acid crystallization. The former is
crystallized separately as the two different forms. One form is SHG active while the
other is not. The L-glutamic acid or crystal formation and transformation to B form,
relative efficiency of the two polymorphs, and preliminary work toward the incorporation

of light scattering into the SHG analysis is performed.

84

SHG Intensity

 

1.2

O
00
9‘1»
00.
‘3?
o . ‘0
3.2“ *
13”
9""
2’
o.
1":
l

 

 

 

°B . ° ”0 00 “fig
06 - co 0 o 0 ~ . O.‘ . ’ 0‘.
O
t A. .
0.4 ~ : s
i
0.2 ‘ r o
O O
0 M
O O
_O.2 1 1 1 1 1
O 50 100 150 200 250

Time (minutes)

Figure 4.15

SHG crystal formation of 3.3% wt. in water L—glutamic acid at 45°C
generated using procedure C. The monitoring was extended to allow for
transformation from the or form to the B form.

85

 

Figure 4.16

Photo of predominantly a L-glutamic acid crystals collected at time 69.1
minutes from the crystallization in figure 4.15.

86

 

Figure 4.17

L-glutamic acid crystals collected at the end of the experiment pictured in
figure 4.15. The photo shows a near equal mix of or and B crystal forms.

87

SHG (intensity)

 

 

 

0.14
0.12 —
0.1 ~
0.08 .
O 00x
0.06 00 code x
000% 00 8%‘296’ x
cow-om” o x
X
004 - ffmxr
0.02 0 fit;

I l 1 I I 1 l 1

 

 

0 10 20 30 4O 50 6O 70 80
Time (minutes)

Figure 4.18

SHG (0) crystal formation and light scattering (x) plotted for L-glutamic
acid formation. The crystallization parameters are described as procedure
C in the experimental section. Methanol addition was started at 3.1
minutes and samples were collected at 28.5 and 50.5 minutes.

88

900

800

700

600

500

300

200

100

Light Scattered (intensity)

 

Figure 4.19

L—glutamic acid crystals sampled at 28.5 minutes. The crystals are a mix
of small particulate and 0t polymorph (prism) with a small amount of the B
form (needles).

89

 

Figure 4.20

L—glutamic acid crystal photographed at 28.5 minutes. The crystal shape
is predominantly the or polymorph.

90

 

Figure 4.21

L-glutamic acid crystals collected at the completion of the experiment.
Crystal shape is the prismatic 0t polymorph.

91

Intensity

 

 

 

 

 

 

 

 

 

     

0 5 IO 15 20 25 30 35 40

Difraction angle

Figure 4.22

X-ray powder patterns for L-glutamic acid. Pattern 1 is crystals collected
from crystallizations of or form crystallization 4.11, pictured in figure
4.13. Pattern 2 is prism shape (or form) and pattern 3 is the needle shape
as referenced.12 Pattern 4 is crystals collected from 4.11 crystallization
and pictured in figure 4.12.

92

4.5 REFERENCES

I Ostwald, W., Grundiss der Algemeinen Chemie, W. Engelmann, Leipzig, 1899

2 Haleblian, J. and McCrone, w. J.Pharm. Science, 1969, 58, 911

3Haleblian,J., J. Pharm. Science, 1975, 64, 1269

4Davey, R. J. and Richards, J., J. ofCryst. Growth, 1985, 71, 597

5 Matsuda, K. et. al., JofPharm. And Phanncology, 1992, 44, 678

°Davey, R. J., Guy, P. D., and Ruddick, A. J., J. ofcoll. and Inter. Sci., 1984, 73, 173
7 Kitaigorodsky, A. I. et. al., J. Phys. Chem. ofSolids, 1965, 26, 463

°Cardrew, P. T. and Davey, R. J., Proc. Royal Soc. London A, 1985, 398, 911

9 Sohnel, O. and Garside, J., Precipitation Ba_sic Principles and Industrial Applictflms,
Butterworth-Heinemann LTD., Oxford, 1992

1ODelano, M., Mol. Cryst. Liq. Cryst., 1979, 52, 271

“ Borka, L. and Haleblian, J., Acta. Pharm. Jugosl., 1990, 40, 71
‘2 Kitamura, M., J. Crystal Growth, 1989, 96, 541

'3 Riechoff, K. E. and Peticolas, w. L., science, 1964, 147, 610
”Hirayama, N. et. al., Bull. Chem. Soc. Jpn., 1980, 53, 30

‘5 Lehmann, M. s., Koetzle, T. F.,and Hamilton, w. c., J. Cryst, Mol. Struct, 1972, 2,
225

93

Chapter 5

ENANTIOMERIC CRYSTAL DETERMINATION USING SHG

Mixed crystallization systems, where more than one enantiomer is present are
important in the production of chiral compounds. The similarity between enantiomers is
very important particularly in pharmaceuticals. The motivation is in situ determination in
production of chiral chemicals for pharmaceuticals. This chapter focuses on using SHG

to differentiate between a single enantiomer and conglomerate crystallization.

5.1 BACKGROUND

The term “enantiomer” describes the relationship between two different
molecules. The molecules are identical in atoms present and the bonding of these atoms.
The difference is in the physical arrangement or physical structure. Enantiomers are
mirror images of each other, which are not superimposable.l The existence of
enantiomeric molecules has been recognized since the early days of organic chemistry.
Enantiomers crystallized such that both are in the same crystal lattice are a conglomerate.
Pasteur recognized the first example of a conglomerate formation in ammonium tartrate
crystals.2

Naturally occurring molecules tend to be exclusively one enantiomer whereas
synthetic production tends to produce racemic mixtures. Separation of these mixtures is

possible using crystallization. By seeding a supersaturated solution with a specific

94

enantiomer, that enantiomer may crystallize preferentially.3 A number of compounds
have been studied to determine mechanisms for conglomerate formation.4'5’°
Characteristics of general crystal structures has been attempted with general trends seen
in the hydrogen bonding.7 Though no clear mechanistic theory applies. A variety of
physical methods have been used for determination of crystal form. Fractional

crystallization using a chiral resolving agent is the most common, along with FTIR,

Raman, and physical data such as optical rotation and melting point. 3’8

5.2 EXPERIMENTAL

The L-aspartic acid (98% purity, Aldrich) and DL-aspartic acid (98% purity,
Aldrich) were used as received from the supplier. The SHG apparatus used is described
in chapter 3 with the modifications described in chapter 4.

Crystallization of the L-aspartic acid was carried out as follows. A 1.9% wat
solution in water was heated to 80°C, cooled to 30°C when the time was started.
Methanol was added as the anti-solvent at a rate of ~2 mIJmin. between 9.7 and 11.7
minutes. The flow rate was increased to ~4 lemin and was added from 29.1 minute to
31.1 minutes.

The DL-aspartic acid crystallization was done by taking a 3.1% wat in water
solution, heating it to 80°C, and then cooling to 50°C when the timer was started.
Methanol was again added as the anti-solvent from 5.1 — 13 minutes (rate of ~2 mIJmin.)

and 13 — 37 minutes (rate of ~ 4 mIJmin.)

95

5.3 DATA AND DISCUSSION

The two investigations carried out here focus on enantiomeric and polymorphic
detection. L-aspartic and DL-aspartic acid are crystallized. The experimental section
underscores the complexity of the crystallization procedures. As the complexity of the
crystallization increases efficient SHG monitoring becomes now useful.

SHG can distinguish enantiomericly pure L-aspartic acid crystals from DL-aspartic
acid crystals. The SHG crystallization monitoring for L-aspartic acid is shown in Figure
5.1. Crystals start forming at 15 minutes and continue until 20 minutes, when the SHG
no longer increases. The photo in figure 5.2 shows a sample of the crystals after filtering
and drying. i

Next, the racemic mixture of DL aspartic acid was crystallized. Figure 5.3 shows

 

the SHG monitoring. The DL aspartic acid is not SHG active, therefore no second
harmonic is generated. For proof, after completion of the experiment, the crystals were
filtered and dried. The crystal size is again fairly small (Figure 5.4), but notably larger
from the L form crystallized above. The photo shows very distinct crystal shapes.
Comparison of the previous two crystallizations demonstrates how SHG can distinguish
between a crystalline racemic mixture and chiraly pure form of aspartic acid. For
example synthetic pathways to manufacturing enantiomeric material often times does not
selectively produce a pure enantiomer. SHG could be used in situ to determine the
particular crystal forming. The enantiomerically pure crystal and the conglomerate have
identical crystal symmetries except for the number of molecules in the lattice. DL
aspartic acid has 8 whereas L aspartic acid only has 4.9’1° This is supported by the data of

the L form being SHG active and DL form is not.

96

SHG Intensity

 

0.1

 

 

 

O O
°.°
0.08 e O .0. O. .O.
O. O . ”O .0 O O. O
O. .0... ..O O . .00. O. 0
”.0 0.. O O 0.. .Q..
0061- O O. 0....0....O O. 0..
' O O
O
O
0.04 ~ .
.0
.0.
O
0.02 r
.-
O .0.
O O . O
0 -0 cm. ‘.
ex. 3 o:
O
_002 1 4 1 L I
0 10 20 30 40 50

Time (minutes)

Figure 5.1

The SHG crystal formation curve for L-aspartic acid. Crystal formation
starts at ~15 minutes and reaches depletion of the supersaturation at 23
minutes. Methanol as the anti solvent was used on this 1.9% wt/wt
solution crystallized at 30 °C.

97

 

Figure 5.2

Photograph of L-aspartic acid crystals collected after the completion of the
crystallization performed in figure 5.1. The crystals are rather small and
difficult to fully contrast in this photo reproduction.

SHG Intensity

0.1

0.08

0.06

0.04

0.02

-0.02

 

 

 

 

.. O
O
O. O O O 0

cm.“ 0.00.0.0... O O 00 00.. Q. . .OO0.0.0..O. 0 ~ 00 O O O.
00 O O O OO0.0~O.~ 0.0 O O. O O 0.00 O O O O“ O. O 0 0.00...

O O O O O O O O

O
l A l 1 1 L l 1
O 5 10 15 20 25 3O 35 40 45
Time (minutes)
Figure 5.3

SHG monitoring of 3.1% wat DL-aspartic acid methanol anti solvent
crystallization at 50 °C. No second harmonic is detected, but crystals are

present as evidenced by the photo in figure 5.4.

99

 

Figure 5.4

DL—aspartic acid crystals photographed upon completion of the
crystallization. Distinct crystal shapes are present.

5.4 CONCLUSIONS

In situ determination of enantiomer or polymorph is crucial for efficient batch
crystallization. Ease of sampling and response time of SHG makes it applicable to such
experiments. L-aspartic acid can be differentiated from DL aspartic acid crystallization

using the SHG technique.

5.5 REFERENCES

lCarey, F. A., Organic Chemistry, McGrew-Hill Book Company, New York, 1987
2Pasteur, L. C., Acad. Sci, 1848, 26, 535

3 Jacques, J., Collet, A., and Wilen, S. H., Enantiomers Racemates. and Resolutions,
Krieger Publishing Co., Malabar, 1994

 

4 Kosma, D., Boeskei, 2., Simon, K., and Foggasy, E., J. Chem. Soc., Perkin Trans.,
1994, 2, 1883

5Larson, 8. and Marthi,K., Acta. Crystallogr., 1995, 351, 338
°Ung, A. T. et. al., J. Am. Chem. Soc., 1995, 117, 8745
7Kazushi,K.,et. al., J. Am. Chem. Soc., 1996, 118, 3441

8 Kozma, D., Maria, A., and Eemer, F., Tetrahedron, 1994, 50, 6907
9Rao,S.T., Acta Cryst., 1973, B29, 1718

10 Derissen, J. L., endeman, H. J ., and Peerdeman, A. F., Acta. Cryst., 1968, B24, 1349

101

 

 

Chapter 6

FUTURE WORK; SECOND ORDER NONLHVEAR OPTICAL
CHEMICAL SENSORS

6.1 INTRODUCTION

There are many different types of chemical sensors, ranging from pH meters to
therrnistors. Sensors are small, sometimes robust devices capable of responding to a
change in analyte concentration or system condition. Most common spectrochemical
sensors rely on absorbance, reflectance, or luminescence.1 Sensitivity and selectivity are
of key importance. The continued work could use second harmonic generation (SHG) in

the construction of a more sensitive and selective chemical sensor.

6.2 BACKGROUND

Sensors are composed of two separate components, the transducer and the sensing
region. The transducer translates the change in "signal" to a meaningful value.
Commercial modules exist for this portion making it inexpensive, small, and very robust.
The second area is the sensing region, which needs to be small, robust, and chemically
selective. The size and robustness for this specific sensor type are obtainable because of

current fiber optic and light detection technology.‘

102

The applicability of SHG for chemical sensing relates directly to the number of
interfaces and surfaces available for study. An extensive review by Corn and Higgins
examines SHG measurements from liquid — liquid, gas — solid, and liquid - solid

interfaces.2

The chemical sensor would implement an interface for the sensing area. For
example, immobilizing an antibody onto a substrate would make that surface selective to
a particular antigen. The change at the surface due to this binding could than be
measured using SHG. The surface sensitivity and specificity of the measurement would
be the advantages of the sensor. The large number of interface applications being studied
creates a large variety of potential sensor applications.

The SHG sensor has certain equipment considerations. SHG is a nonlinear
optical technique, therefore it requires a monochromatic, high intensity light source.
N anosecond or shorter laser pulse widths are adequate. Laser technology and availability
has substantially increased in recent years making small relatively inexpensive diode

pumped lasers available. The remaining portion of the sensor (see diagram), optics,

detector, and electronics for the transducer are available commercially.

6.3 EXPERIMENTAL

This section presents two pathways for investigating the SHG chemical sensor.
The first relies on a liquid-solid interface for the sensing region to differentiate
enantiomers. The second would use SHG circular dichroism to measure enantiomeric or
potentially polymorphic differences at a liquid — air interface.
Eiantiomeric detection
This slight difference in orientation of enantiomers is known to inhibit crystal formation
and crystal growth differently for similar chemical compounds. Gases and March have

103

 

 

to
computer

 

1‘
858°C

 

 

 

 

 

 

   
 

bifurcated
fiber optic

Figure 6.1

The schematic of an SHG sensor.

104

demonstrated this phenomenon on the inhibition of L-glutamic acid crystal formation by
L-lysine. D-lysine does not inhibit the L-glutamic acid crystallization. These studies
were done in bulk solutions, measuring the crystal formation by using a light scattering

technique.3

However, crystal formation is a very complicated process that is difficult to
control. Crystal growth on the other hand is better understood. By immobilizing the L-
glutamic acid on the substrate, the presence of L-lysine would alter the crystal growth,
changing the crystal structure, producing a change in SHG. The quantity of L—lysine
would be measured without interference from the D form. Clearly, the opportunity to
investigate other enantiomeric combinations exists.
Polymorphic determination

A second project is detection of enantiomeric and polymorphic differences.
Polymorphs are identical chemically, differing only in the crystal structure. By using
circular difference SHG, selectivity of the enantiomers and polymorphs is achievable.
SHG circular difference is analogous to linear circular dichroism, but is more sensitive.
Verbiest and coworkers have demonstrated this SHG circular dichroism on

bacteriorhodopsoin at an air - water interface.4 Here again, a vast number of polymorphs

and enantiomers exist for potential sensing applications.

6.4 CONCLUSION

Potential applications in the fields of industrial process control and environmental
sensing exist. Validation of this proposed technique would be necessary. Raman
spectroscopy and X-ray diffraction would be the primary methods of verifying and

calibration of this technique.

105

6.5 REFERENCES

lCattrall, R. W., Chemical Sensors, Oxford Science Publications, Oxford, 1997,
Chaps. 1-3

2Corn, R. M., and Higgins, D. A., Chem. Rev., 1994, 94, 107
3 Grases, F. and March, J. G., Trends in Analytical Chemistry, 1996, 10, 190

4 Verbiest, V., Kauranen, M., Persoons, A., Ikonen, M., Kurkela, J ., and Lemmetyinen,
H., J. Am. Chem. Soc., 1994, 116, 9203

106

CONCLUSION

DEVELOPMENT OF SECOND HARMONIC GENERATION AS A
TOOL FOR MONHORMG CRYSTALLIZATION

Monitoring crystallization provides a means for knowing how to control it. The
goal of this research project was to develop an in situ probe for monitoring
crystallization. The ability to monitor crystal formation enables better control of the
crystallization process. The development of SHG as a way of monitoring the
crystallization process is an excellent complementary tool to the methods already
developed.

The work is laid out roughly in the order of development. The exception is
Chapter 1, which dealt with monitoring crystallization, but employs the use of Raman
spectroscopy instead of SHG. By considering the practicality, benefits, and limits of this
technique on the laboratory scale, potential application to industrial scale can be realized.
Raman spectroscopy is an optical technique that can be performed in-Sl'tu and within the
time frame of the crystallization phenomena of selected chemical systems.

The initial demonstration and proof of the applicability of the SHG technique as a
method for monitoring crystal formation in supersaturated solutions is demonstrated. The
advantages over turbidometric methods are shown. Refinements to the sampling

technique are made and demonstrated for enhancing the crystallization of lysine

107

monohydrochloride. A modest increase in the mean crystal size and a drastic change in
the crystal quality occurs when controlling the crystallization.

The versatility of the SHG monitoring technique is shown for polymorphic and
enantiomeric detection. SHG can be used in situ for L aspartic acid and L-glutamic acid
polymorph crystallizations and transformations for the latter. Additionally, SHG can
distinguish the enantiomerically pure crystallization versus conglomerate formation of

aspartic acid.

108

APPENDIX

 

109

 

Table A.1

 

 

induction 886 917 886/917 normalized
time anion/crystal
0.41 194.00 58.00 3.34 0.98
1.02 194.00 58.00 3.34 0.98
1.63 194.00 58.00 3.34 0.98
2.25 194.00 58.00 3.34 0.98
2.86 197.00 60.00 3.34 0.98
3.47 197.00 60.00 3.28 0.96
4.09 197.00 60.00 3.28 0.96
4.70 197.00 60.00 3.28 0.96
5.31 197.00 60.00 3.28 0.96
5.93 197.00 60.00 3.28 0.96
6.54 197.00 60.00 3.28 0.96
7.15 197.00 60.00 3.28 0.96
7.77 197.00 60.00 3.28 0.96
8.38 197.00 60.00 3.28 0.96
9.00 200.00 61.00 3.28 0.96
9.61 200.00 61.00 3 .28 0.96
10.22 200.00 61 .00 3.28 0.96
10.84 200.00 61 .00 3.28 0.96
l 1.45 200.00 61 .00 3.28 0.96
12.06 200.00 61 .00 3.28 0.96
12.68 200.00 61 .00 3.28 0.96
13.29 200.00 61 .00 3.28 0.96
13.90 200.00 61 .00 3.28 0.96
14.52 200.00 61 .00 3.28 0.96
15.13 201 .00 61 .00 3.28 0.96
15.74 201.00 61 .00 3.30 0.96
16.36 201.00 61 .00 3.30 0.96
16.97 201.00 61 .00 3.30 0.96
17.58 201.00 61.00 3.30 0.96
18.20 201.00 61 .00 3.30 0.96
18.81 201.00 61 .00 3.30 0.96
19.42 201.00 61 .00 3.30 0.96
20.04 201.00 61 .00 3.30 0.96
20.65 201.00 61 .00 3.30 0.96
21.27 201.00 61 .00 3.30 0.96
21.88 201.00 61 .00 3.30 0.96
22.49 201.00 61 .00 3.30 0.96
23.11 201.00 61 .00 3.30 0.96
23.72 201.00 61 .00 3.30 0.96

 

 

110

 

 

 

 

Table A 1 (continued)

 

 

 

 

111

 

 

wavenumber
induction 886 917 886/917 normalized
time anion/crystal
24.33 201.00 61 .00 3.30 0.96
24.95 201.00 61 .00 3.30 0.96
25.56 201.00 61 .00 3.30 0.96
26.17 201.00 61 .00 3.30 0.96
26.79 201.00 61 .00 3.30 0.96
27.40 200.00 61 .00 3.30 0.96
28.01 185.00 56.00 3.28 0.96
28.63 154.00 45.00 3.30 0.97
29.24 122.00 37.00 3.42 1.00
29.85 101.00 33.00 3.30 0.96
30.47 83.00 28.00 3.06 0.89
31.08 69.00 25.00 2.96 0.87
31.69 56.00 22.00 2.76 0.81
32.31 45.00 19.00 2.55 0.74
32.92 38.00 19.00 2.37 0.69
33.54 32.00 17.00 2.00 0.58
34.15 28.00 17.00 1.88 0.55
34.76 26.00 16.00 1.65 0.48
35.38 24.00 16.00 1.63 0.47
35.99 22.00 15.00 1.50 0.44
36.60 22.00 16.00 1.47 0.43
37.22 21.00 15.00 1.38 0.40
37.83 21.00 15.00 1.40 0.41
38.44 20.00 15.00 1.40 0.41
39.06 19.00 15 .00 1.33 0.39
39.67 19.00 14.00 1.27 0.37
40.28 18.00 14.00 1.36 0.40
40.90 18.00 14.00 1.29 0.38
41.51 18.00 14.00 1.29 0.38
42.12 17.00 13.00 1.29 0.38
42.74 18.00 14.00 1.31 0.38

 

 

Table A2

 

 

 

 

 

 

 

 

 

% supersaturation KDP Induction times
13.26 1.50 4.20 4.33 3.34
12.00 6.67 6.67 13.67 9.00
10.40 6.17 11.17 9.17 8.83
9.00 10.17 18.67 16.17 15.00
7.96 9.67 42.67 32.17 28.17
6.58 31.17 40.17 16.17 29.17
5.55 47.17 40.17 86.17 57.83
Table A. 3
number of crystals
Crystal cont. uncont.
size-5 size +5 ave. std dev ave. std dev
922.5 932.5 734.45115 297.84759 223.0356 203.9142
622.5 632.5 1548.7506 293.0588 721.2872 234.1405
542.5 552.5 3832.3789 377.89957 2191.088 951.4569
457.5 467.5 10144.979 557.20486 7283.248 5110.062
384.5 394.5 55507.45 28933.435 48959.4 54214.4
299.5 309.5 149451.13 54496152 1878223 90831.53
208.5 218.5 52900.897 23373.993 70139.98 183585.8

 

 

 

112

 

 

 

"lllllllll111111111711“