Optimize Industrial Crystallization

by Tracking Particle Size Inline

A Review of Modern Technologies

Case Studies from the

Chemical, Food Ingredient,
Mining and Minerals Industries
Benjamin Smith, Brian O’Sullivan, Mettler-Toledo AutoChem, Inc.

Industrial crystallization is an important density problems in transport and process conditions in real time to consis-
separation and purification step in the consequently particle size and quality tently achieve targeted particle size, yield
chemical, food ingredient, and mining variability in the final product. Particle and quality specifications, in the lab or
industries, yet it is often treated as an art size is also directly correlated with yield. manufacturing.
rather than a science. If misunderstood, Many of these process inconsistencies can
the thermodynamics and kinetics of be solved by understanding, optimizing, Today, inline particle characterization
crystallization can sometimes result and controlling the crystallization unit technologies are used to provide inline
in unexpected behavior such as rapid operation. particle size, shape and count measure-
precipitation or agglomeration which ments at full process concentrations
leads to variability in yield, centrifugation In the crystallizer, solvent addition rates, and in translucent or opaque slurries.
times, and other critical quality attributes. impurities, mixing conditions, seeding, Companies including the American
It has been established that the root cause and cooling rates, all impact process Crystal Sugar Company3, Dow5, Archer
of this variability is often inconsistency performance. By tracking the rate and Daniels Midland (ADM) 6, and Nestle7
in the particle size distribution1,2,3,4. degree of change to particles and particle have successfully applied best practices
Fines or agglomerates can form in the structures as they naturally exist in with inline particle size characterization
crystallizer which cause slow throughput process, scientists and engineers proac- to collect real time data and ensure true
in the centrifuge, caking, dust, or bulk tively measure, understand, and optimize statistical understanding of their process.
 Possible Process Bottlenecks Where Particles Play a Role

Raw Materials Crystallizer Centrifuge Dryer Transport Product

Potential Nucleation/ Slow Throughput/ Long Drying Caking, Yield,
Sources of Precipitation Poor Centrifuge Times Bulk Density, Particle Size,
Variability Inconsistency Rate; Yield Loss Excessive Dust Flow Properties,
Impurities

A challenge with traditional particle size measurement techniques is that most technolo-
gies are offline and require sampling, sample preparation, and remote analysis which
is time consuming. Offline samples are often prepared and altered through dilution,
dispersion, or drying, which can change or destroy particle or droplet components and
their relationship to critical quality attributions. Offline measurements also cannot
be applied to make real-time process optimization and control decisions. With estab-
lished particle characterization technology (ParticleTrack™ with FBRM® technology
and PVM® inline imaging), scientists and engineers can measure and observe crystal
behavior in situ without sampling or dilution and track the crystallization process at
full concentration and operating temperature. This enables real-time feedback and fast
decisions to improve the root cause of variability in the crystallizer and optimize the
consistency of final product particle size and yield.

Many industries apply inline particle characteriza-

tion to improve crystallization performance. This
Pharma Pulp &
paper will focus on Chemical, Food and Ingredient, Paper
and Mining and Minerals case studies:

Lactose Crystallization
(page 3) Food Consumer
Dextrose Crystallization Ingredients Product
(page 8)

L-Lysine Crystallization
Chemical Formulation
(page 6)

Lithium Carbonate
Mining and
Crystallization Biotech
(page 8)
Minerals

2
 Understand THEN Optimize (Lactose Study)
Gaining basic process understanding is the first step in optimizing crystallization.
Important steps such as measuring the solubility curve provide a fundamental under-
standing of crystallization thermodynamics and are discussed in detail in several
journal publications (see references, page 14).

The empirical knowledge provided by microscopy is perhaps the easiest way to understand
a crystallization process8. Inline imaging has been established as one of the fastest ways
100 µm
to understand crystal shape and size, as well as, identify fundamental crystallization
mechanisms such as nucleation, growth, agglomeration, and breakage9,10. Lactose, a
Figure 1. Seed crystal aggregation would have
by-product of the cheese industry derived from milk whey, is typically crystallized on a caused inconsistent surface area and secondary
massive industrial scale via cooling crystallization. Product quality and yield are highly nucleation. As the process was observed using
dependent on particle size and the crystallization process is complicated by seasonal in-process measurement, the issue was seen in
real-time and the batch was quickly reheated,
variability in milk quality, the presence of mineral impurities and slow crystal growth re-dissolved, then re-seeded to ensure consistent
kinetics - which historically require long batch times. Lactose crystallization is a perfect seed surface area.
opportunity to use inline microscopy for rapid process understanding. In the example
below, three lactose crystallization batches were processed and all image sequences
were captured inline and in real time. No sampling or dilution was necessary. It is clear
that within one batch, the crystals undergo changes including nucleation, aggregation,
secondary nucleation and growth.

By quickly comparing images from different batches at the same process temperatures,
the effect which process conditions (such as impurities, seeding, and cooling rate) have
on the final crystal properties can be observed. By measuring inline, process conditions
are quickly optimized to achieve a desired crystal size and shape and improve centrifuge
performance and process yield downstream.

Batch 1 (60 wt % deprotonated whey powder in di water; 0.1% Seed)

Inital batch

100 µm 100 µm 100 µm

25 ºC 25 ºC 25 ºC
Fast Cooling
(79°C to 25 °C over
°C
1 hr), Isothermal
Hold (8 hrs)

1.25 hrs Relative Time

Findings:
Understanding Process Implications Optimization
Batch cooled quickly, seeding Filtration was difficult and yield Apply slower cooling rate to
induced rapid nucleation and the was low due to losses in filtration next batch
final crystals were very fine and washing

3
 Batch 2 (60 wt % deprotonated whey powder in di water; 0.1 % Seed)
Adjusted cooling curve based on Batch 1 findings

100 µm 100 µm 100 µm

70 ºC 65 ºC 25 ºC
Controlled
Cooling °C
Over 8 hours
Relative Time
8 hrs

Findings:
Understanding Process Implications Optimization
Slower cooling rate was applied, Long filtration rates persisted, Apply a slower cooling rate in
however secondary nucleation was and yields were lower due to next batch
observed resulting in a fine crystal losses associated with the
distribution presence of fines

Batch 3 (60 wt % deprotonated whey powder in di water; 0.1 % Seed)

Adjust cooling curve based on Batch 1 and 2 findings

100 µm 100 µm 100 µm

70 ºC 57 ºC 25 ºC
Controlled
Cooling °C
Over 18 hours
Relative Time
18 hrs

Findings:
Understanding Process Implications Optimization
Slower cooling rate was applied Much higher yield and faster Process ready for scale-up
and ideal seed distribution is filtration rate - meets target
observed specifications

4
 In the same lactose crystallization process, 90
6000
engineers applied inline particle tracking
tools to quantify the differences in crystal 5000
80

size from batch to batch in real time. All

Temperature (°C)
70

Counts (fines)
these measurements were acquired inline 4000 Batch 2, Counts (0.5-50 µm)

at 60 wt % solids with no dilution or sample Batch 2, Temperature 60

preparation. Figure 2 tracks the number 3000

50
of fine particle counts in the range 0.5
2000
µm to 50 µm as Batch 2 progresses over 40

time. The increase in fines is typically an

1000 30
indication of secondary nucleation. It is
evident that at hour 3:00:00 and 5:30:00 00:00:00 04:00:00 08:00:00 12:00:00 16:00:00 20:00:00

the fines increase rapidly. These are most Relative Time

likely nucleation events that result from
Figure 2. Process temperature and ParticleTrack
cooling at a rate that is faster than the rate data trending the number of fine particles 0.5-
of supersaturation consumption. After 50 µm over time as nucleation and secondary
nucleation occur while process temperature
hour 12:00:00, the fines increase gradu- decreases
ally, likely due to attrition as the batch
temperature is held constant at 25 °C. Batch 1, Counts (fines) Batch 2, Counts (fines) Batch 3, Counts (fines)
6000 Batch 1, Temperature Batch 2, Temperature Batch 3, Temperature 90

By adjusting the cooling rate from one 5000

80

hour, to eight hours, then to eighteen

70

Temperature (°C)
Counts (fines)

hours, nucleation is suppressed (Figure 3). 4000

60
3000
Figure 4 shows the endpoint distribu- 50
tions measured in-process for these three 2000
40
batches. By applying this immediate
understanding of particle dimension and 1000 30
particle count, scientists and engineers can
20
“tune” crystallization parameters to “dial
00:00:00 04:00:00 08:00:00 12:00:00 16:00:00 20:00:00
in” the final particle size distribution.
Relative Time

Figure 3. Process trends tracking changes to the

fine counts and temperature over time for the three
lactose crystallization batches

200 Batch 1, endpoint

Batch 2, endpoint
Counts (Length Weight)

Batch 3, endpoint

150

100

50

0
1 10 100 1000
Chord Length (µm)

Figure 4. The counts vs. dimension, chord length

(µm) measured in process at the endpoint of
5 three batches
 Seeding to Optimize Crystallization
In more recent years, seeding has become one of the most critical steps to control
crystallization behavior and optimize the final particle size. Seed crystals introduced at Seeding too Late
(Figure 7)
the right point in the process can change a poorly behaved, inconsistent crystallization

Concentration
process, to one that is consistent and produces particles with the required particle size Seeding too Early
specification. Inconsistent yield, filtration rate, drying time, bulk density, flow property, Metastable Zone Limit (Figure 6)

and particle size distribution can also be traced back to uncontrolled crystallizations
with inconsistent nucleation or non-optimized seeding. Important parameters must be
Solubility
considered when designing a seeding strategy such as seed size, seed loading (mass)
and seed addition temperature. These are generally optimized based on process kinetics
and the desired final particle properties, and must remain consistent during scale-up Temperature

and technology transfer.

Figure 5. Solubility curve and MSZW (Metastable
Zone Width)
Monitoring the seed behavior in situ, during the crystallization itself, provides a signifi-
cant advantage during the design and manufacturing of a crystallization process. The
30 minutes following seed addition are generally the most critical period in ensuring the
effectiveness of the seed, and mapping the
seed behavior inline ensures the batch pro-
50
Seeding
ceeds as expected. Understanding seeding (add 2.25 g
unmilled seed)
Counts, 200-1000 µm

parameters means fewer experiments are Seeds disperse and

40
necessary to optimize a crystallization11. quickly dissolve
30
When to Add the Seed?
(L-Lysine Study) 20
Tracking the rate and degree of change
to inline particle size and count provides 10
insight to understand, optimize, and con-
trol seeding. If seeds are added at the inap- 0
propriate temperature, based on solubility, 01:20:00 01:30:00 01:40:00 01:50:00 02:00:00
the seeds will quickly dissolve (Figure 6). Relative Time
In other cases, the seeds are added too
late after the product has already crys- Figure 6. L-Lysine crystallization seeds are added
just above the solubility temperature. The seeds
tallized (Figure 7). This phenomenon
disperse, dissolve and disappear within 15 min.
is sometimes observed with variability
in nucleation kinetics due to changes in
Seeding
the purity of the raw material or variable (add 0.1 %
mixing and temperature gradients upon 400 milled seed) Counts (200-1000 µm)
scale-up to manufacturing.
Counts,1-100 µm

40
300
Counts (No. Wt. 1-100 µm) Seeding has
Counts (No. Wt. 200-1000 µm)
minimal effect
In either case, when seeds dissolve or when
seeds are added after primary nucleation, 200 20
the process is effectively unseeded and can
often result in a less controlled process 100

where rapid nucleation introduces fine 0

0
particles or uncontrolled and inconsistent
02:15:00 02:30:00 02:45:00 03:00:00 03:15:00 03:30:00 03:45:00
growth forms aggregates.
Relative Time

Figure 7. L-Lysine crystallization seeds added

and dispersed after particle nucleation already
occurred. The effect of seeding at this point
is minimal due to the abundance of crystals
already present.
6
 Beginning, 33 ºC Midpoint, 25 ºC Endpoint, 15 ºC

100 µm 100 µm 100 µm

Figure 8. L-Lysine images at the beginning,

midpoint, and endpoint of the unseeded process.
Formation of fine particles is observed in real time
with inline images providing insight to the exact
conditions which result in secondary nucleation.

Figure 8 shows inline images (PVM) at

the beginning, middle, and endpoint of 30 02:36:52 04:07:19 05:32:06
a L-Lysine crystallization captured in
Mean sqr Wt 207.1 226.5 228.7
real time from within the crystalliza- 25 Counts No Wt
29.9 173.5 257.5
tion vessel. These images provide a clear <20
Counts (No. Weight)

understanding of the wide crystal size 20

Counts No Wt
200-1000
6.6 28.2 40.4

distribution. Figure 9 shows the inline

Nucleation
chord length distribution (ParticleTrack 15 (increase in fines)
data). By measuring the inline process
dynamics, the nucleation point, secondary 10
Growth
nucleation, and the condition when these (increase in coarse)
process changes occur are immediately 5
detected. This enables scientists and engi-
neers to make decisions, adjust seeding 0
1 10 100 1000
parameters, and change the real time path
Chord Length (µm)
of crystal growth.

Figure 9. (left) L-Lysine chord length distributions

By measuring the number of fine particles at the beginning, middle and endpoint of an
(<20 µm) and coarse particles (>200 µm) unseeded process. The initial nuclei grow to
large crystals >300 µm, but at the same time
along with the mean, the tails of the dis- as secondary nucleation occurs producing fine
tribution are understood and the degree crystals <50 µm.
of nucleation, growth, and agglomeration
Table 1. (right) Statistics calculated at time points
is controlled. Table 1 presents real-time corresponding to glycine images and distributions
statistics calculated for the distributions
(Figure 9).

7
 Seed Size and Seed Mass (Dextrose and Lithium Carbonate Studies)
Questions often asked in relation to seeding are “How much seed mass to add?” and
“What size should it be?”. This depends on the final particle size required or impurity
concerns associated with the crystallization. Seeds are generally added to provide the
necessary surface area for particle growth or to induce nucleation. The size and number
of seed particles added determines this available surface area. Generally seed particles
are small as this increases the number of ‘points’ for crystal growth and reduces the
chance for secondary nucleation. If large particles are required, less seeds are added
producing a fewer number of particles which grow very large in size. On the other hand,
if small particles are required, larger seed loading is necessary so that the growth is
spread across a greater number of particles, consequently growing each particle only
by some small amount. Typical seed loadings can vary from as little as 0.1 wt % to as
much as 5 wt % (and even more) depending on the requirements.

In Figure 10, Markande12 tracks the nucle-

ation kinetics of dextrose monohydrate
crystallization by tracking the number of 20,000

particle counts (#/sec) between 0.5 µm

18,000
and 112 µm over time (hrs). When large
Counts (1-112 µm)

seeds are added (220 µm and 155 µm) 16,000

they induce an increase in the number of
14,000
particle counts which can be attributed
to secondary nucleation. Typically this is 12,000
due to the limited surface area available
for growth on larger seeds. When the same 10,000
Seed Size 125 µm
mass of a smaller seed (125 µm) is added, a Seed Size 155 µm
8,000 Seed Size 220 µm
higher surface area results in less second-
ary nucleation. By minimizing secondary 6,000
nucleation, users gain better control over 0 2 4 6 8 10 12 14 16 18 20 22 24 26

the final particle size distribution and Time (hrs)

downstream centrifugation and yield.
Figure 10. Process trend for three dextrose batches
This same phenomenon is experienced in the minerals industry with lithium carbonate tracking the number of counts between 1-112 μm
over time. Seed size varied between batches12.
crystallizations for example. Sun et al13 use inline PVM images to show the effect of
insufficient seeding conditions that result in secondary nucleation of fine particles
(Figure 11).

T=0s T = 60 s T = 90 s

150 µm 150 µm 150 µm

Figure 11. Secondary nucleation is shown by inline

PVM images tracking the process over time13
8
 Lithium carbonate is precipitated by the 10000
reaction of lithium chloride with sodium Batch 1 (Seed Size: 401 µm)
Batch 3 (Seed Size: 225 µm)
carbonate. Varying the concentration 8000 Batch 5 (Seed Size: 21 µm)

of each component modifies the super-

Counts (#/sec)
saturation which affects the growth and 6000
nucleation kinetics. In Figure 12, three
crystallization processes are shown which 4000
had the same seed mass (0.75 g/L) added
at the same temperature and supersatu- 2000
ration (3.4), but with varying mean seed
size (401 µm, 225 µm, and 21 µm). At 0
Time=0, the crystallization batch with 0 100 200 300 400

the smallest seed size has the highest Relative Time (s)
number of particles. However, due to the
higher surface area for growth created Figure 12. Process trends for three lithium
carbonate batches tracking the total crystal counts
by these small seed particles, the rate of over time. Seed size varied. All batches were
increase in particle number is less than seeded at supersaturation of 3.413.
that observed for the batches with larger
seed size. Effectively this means that the 4000
Batch 1 (Seed Size: 401 µm)
batch with the smallest seed has resulted Batch 3 (Seed Size: 225 µm)
in less secondary nucleation. Figure 13 Batch 5 (Seed Size: 21 µm)
Counts (#/sec)

3000
presents the same lithium carbonate
seeded crystallization experiments but
2000
at a lower supersaturation (2.7). In this
example the number of particles remains
constant for all three batches for the first 1000

250 seconds, yet an increase in particle

count (secondary nucleation) eventually 0
occurs for the batches with larger seeds. By 0 100 200 300 400
measuring and optimizing the seed size, Relative Time (s)
scientists and engineers gain control over
the crystallization process and limit the Figure 13. Process trends for three lithium
degree of fine particle formation. carbonate batches tracking the total crystal counts
over time. Seed size varied. All batches were
seeded at supersaturation of 2.713.
In addition to seed size, optimizing the
seed mass is also a strategy to improve 4000
crystallization yield, centrifugation
Batch 5 (21 µm), SL=0.75 g/L
times, and particle size distribution at Batch 5 (21 µm), SL=0.25 g/L

the endpoint. In this example of lithium 3000

Counts (#/sec)

carbonate precipitation (Figure 14), Sun

et al increase the seed mass from 0.25 2000
g/L to 0.75 g/L. By tracking the system
using ParticleTrack, Sun et al measured
the reduction in nucleation for the larger 1000

seed mass and quickly optimized reaction

precipitation for improved product quality, 0
yield, and downstream performance. 0 200 400 600 800 1000 1200
Relative Time (s)

Figure 14. Process trends for two lithium carbonate

batches tracking total crystal counts over time.
Seed mass varied between batches. Seed size was
9 a constant 21 μm13.
 Measuring the Effect of Impurities 25,000
on Crystallization 4 % Impurities

Counts, 194-233 µm (#/sec)

7 % Impurities
Many industrial crystallization and 20,000
11 % Impurities

precipitation processes, whether batch

or continuous, experience variation in
15,000
the quality of raw materials (including
variation in vendor, source, or calendar
season) that could have an effect on raw 10,000
material impurity composition. This may
have an obvious effect on the purity of the 5,000
final product, but it can also significantly 0 2 4 6 8 10 12 14 16 18 20 22 24 26
impact the process growth and nucleation Time (hrs)
kinetics, the yield, and the downstream
solid-liquid separations efficiency. Figure 15. ParticleTrack data presenting the
number of the coarse particles in the range 194 µm
to 223 µm, trended over time (hours)14
Markande et al14 implemented
ParticleTrack technology to track
nucleation and growth of dextrose with 20,000
4 % Impurities
varying levels of incoming impurities.
Counts, 1-112 µm (#/sec)

7 % Impurities
Slower crystallization kinetics with 11 % Impurities

higher levels of impurities are now easily 15,000

observed. At 24 hours, it can be implied
that the batch with 11% impurities would
have a lower yield versus the batches 10,000
with lower impurities. By measuring
batch kinetics towards the endpoint in
real time, batch time can be adjusted or 5,000
upstream purification of the raw material 0 2 4 6 8 10 12 14 16 18 20 22 24 26

performed. Without data, crystallization Time (hrs)

may seem like an art. But by applying
real-time measurements, scientists and Figure 16. ParticleTrack data presenting the
number of fine particles in the range 0.5 µm to
engineers understand when and why
112 µm, trended over time (hours)14
inconsistencies occur and make informed
decisions to produce repeatable, high-
quality products.

10
 “Without data, crystallization may
seem like an art. But by applying
Conclusions real-time measurements, scientists
The desire to produce products with a specific particle size distribution and powder and engineers understand when
handling properties continues to drive crystallization research. The need for a targeted and why inconsistencies occur and
particle size is also a driver to reduce batch time, improve yield, and solid-liquid
separation rates. In certain bulk chemical processes large particles are produced to
make informed decisions to produce
ensure fast separation times - maximizing the economic efficiency of the process. On repeatable, high quality products.”
the other hand, in some processes small particles are required to meet final customer
requirements. Regardless of the product’s critical quality attribute, there is a requirement
to target a pre-defined particle size distribution that if not met can result in additional
processing costs or, in extreme cases, loss of product.

Fundamental understanding of crystallization has improved dramatically over the past

30 years yet this single unit operation still poses some of the greatest challenges when
designing and developing new chemical processes. With established in situ monitoring
techniques, such as ParticleTrack with FBRM Technology and PVM, scientists and
engineers have the ability to monitor and control crystallization processes in real
time. The enhanced level of understanding and control these technologies provide has
significantly improved yield, throughput, and repeatability in manufacturing. Rather
than the traditional approach of sampling at the process endpoint, when it is too late,
in situ technologies monitor critical steps during the process so process deviations or
upsets can be observed in real time and quickly corrected.

White Paper Spotlight

Best Practice for Inline Particle Size Characterization
This White Paper shows how scientists and engineers apply inline
particle size and count measurements to troubleshoot and improve
process performance and product quality. By implementing inline
technologies, companies avoid offline sampling and sample preparation
errors as well as enable the real-time optimization of processes.

Several applications are discussed including:

- Measuring Fines Precipitation to Improve Solid/Liquid Separations
- Improving Product Stability During Formulation of a Liquid Emulsion
- Consistently Meeting Particle Size Specifications

www.mt.com/wp-E25

11
 Appendix A:
Focused Beam Reflectance Measurement
(FBRM®)
Measurement for optimization in real time – FBRM is a highly precise
and sensitive technology which tracks changes to particle dimension,
particle shape, and particle count. Over a wide detection range from
0.5 to 2000µm, measurements are acquired in real time while particles
are forming and can still be modified enabling process optimization
and control. No sampling or sample preparation is required – even in
highly concentrated (70% and higher) and opaque suspensions. 

Figure c. Chord Length

Distributions
Laser Source

Laser Return

Optics Module
1 2 3 4

Figure b.

Figure a.

Sapphire Window Figure d. Trended

Statistics

How does FBRM work?

The FBRM probe is immersed into a dilute particle structure to another edge. Thou-
or concentrated flowing slurry, droplet sands of individual chord lengths are typ-
emulsion, or fluidized particle system. A ically measured each second to produce
scanning laser is focused to a fine spot at the Chord Length Distribution (CLD)
the sapphire window interface (Figure a). (Figure c). The CLD is a “fingerprint” of
A magnified view shows individual par- the particle system, and provides statistics
ticle structures will backscatter the laser to detect and monitor changes in particle
light back to the probe (Figure b). These dimension and particle count in real time
pulses of backscattered light are detected (Figure d).
by the probe and translated into Chord
Lengths based on the simple calculation Unlike other particle analysis techniques,
of the scan speed (velocity) multiplied by FBRM measurement makes no assump-
the pulse width (time). A chord length (a tion of particle shape. This allows the
fundamental measurement of particle di- fundamental measurement to be used to
mension) is simply defined as the straight directly track changes in the particle di-
line distance from one edge of a particle or mension, shape, and count.
12
 Appendix B:
Particle Vision Microscope (PVM®)

Vision for understanding and optimization – PVM is a real-time probe

based vision tool which provides instant critical insight into crystal,
particle, and droplet systems. PVM enables chemists and engineers
to detect and understand process changes that could take months to
discover with traditional offline microscopy techniques.

CCD Camera

Objective Lens

Illumination Lens

Sapphire Window

How does PVM work?

PVM uses a high resolution
CCD camera and internal illumination
source to obtain high quality images
even in dark and concentrated suspen-
sions or emulsions. With no calibration
needed and easy data interpretation, PVM
quickly provides critical knowledge of
crystal, particle, and droplet behavior.

200 µm

Left: PVM inline image;

Right: Offline microscope

13
References
1. T. A. Bell, Powder Technology, 150 (2): 60-71 (2004).
2. Kim S. et al. Organic Process Research & Development, 9, 894-901 (2005).
3. Sobolik, J. In situ Partcle Size Analysis, American Crystal Sugar Company, USA (2002).
4. Mousaw, P., Saranteas, K. & Prytko, B., Organic Process Research & Development,
12(2), 243-248 (2008).
5. Larsen, Paul (The Dow Chemical Company). Tips and Tools for Solution Crystallization www.mt.com/crystallization
Process Design, AIChE Process Development Symposium (2013)
Benjamin Smith
6. Archer Daniels Midland (ADM). De Zaan Cocoa Manual, 123-124 (2009). http://www.
adm.com/en-US/products/food/cocoa/Documents/deZaan-Cocoa-Manual.pdf
[email protected]

7. Rousset, P. et al. Fat crystallization in w/o emulsions controlled by shear and cooling
conditions, Nestle (2006).
8. McCrone, Walter C. The Particle Atlas. Elsevier Science & Technology Books, (1979).
Our People
METTLER TOLEDO has a global
9. Duffy, D., Cremin, N., Napier, M., Robinson, S., Barrett, M., Hao, H., & Glennon, B.
Chemical Engineering Science, 77, 112–121. (2012). network of Technology and Application
Consultants with extensive research and
10. Barrett, M., Hao, H., Maher, A., Hodnett, B. K., Glennon, B., & Croker, D. Organic Process
Research & Development, 15(3), 681–687. (2011). industry experience supporting inline
particle characterization applications.
11. Bakar, M. R. A., et al. Organic Process Research & Development, 13(6), 1343–1356
(2009).
Email: [email protected]
12. Markande, A., Fitzpatrick, J., Nezzal, A., Aerts, L., & Redl, A. Application of inline
monitoring for aiding interpretation and control of dextrose monohydrate crystallization. Phone: 410-910-8500
Journal of Food Engineering, 114(1), 8–13. (2013). http://www.sciencedirect.com/
science/article/pii/S0260877412003639 Blog
13. Sun, Yu-zhu. et al. Seeded Induction Period and Secondary Nucleation of Lithium Chemical Research, Development and
Carbonate. The Chinese Journal of Process Engineering. Vol.9 No.4. Aug. (2009). Scale-up is our Blog highlighting the
http://www.jproeng.com/qikan/manage/wenzhang/209138.pdf
latest publications and providing expert
14. Markande, A., Nezzal, A., Fitzpatrick, J., Aerts, L., & Redl, A. (2012). Influence of impuri- commentary from our own internal
ties on the crystallization of dextrose monohydrate. Journal of Crystal Growth, 353(1),
145–151. experts and from academic and indus-
try professionals.
15. Barrett, P. and B. Glennon, Characterizing the metastable zone width and solubility
curve using Lasentec FBRM and PVM, Chemical Engineering Research & Design 80
(A7): 799-805 (2002). Customer Community
Our Customer Community Site provides
owners and users of our technologies
with free access to archived citation
lists, application reports, case stud-
ies, and extensive training materials
– including immediate access to all of
our on-demand webinars.

Social Media
Get real-time updates through Facebook
and Twitter on the latest developments
in chemical synthesis, chemical engi-
Internet: http://www.mt.com/autochem neering and scale-up.
Subject to technical changes
© 06/2013 Mettler-Toledo AutoChem, Inc.
7075 Samuel Morse Drive
Columbia, MD 21046 USA
Telephone +1 410 910 8500
Fax +1 410 910 8600
Email [email protected]