M-2-012P 1 Micromixing and Mesomixing Effects on Nucleation, Agglomeration and De-agglomeration Kinetics in a Dye Precipitation Kostas Saranteas (speaker) Sepracor Inc. Gregory Botsaris Tufts University Presented at Lasentec Users’ Forum 2000 Orlando, Florida Production of sparingly soluble materials by precipitation from reactive solutions is a very important unit operation in chemical engineering practice. Precipitation involves a sequence of events including mixing, fa st reaction- diffusion, homogeneous nucleation, agglomeration, and de-agglomeration, with scale sizes ranging from molecular to macroscopic. A number of theoretical models and discussions presented in the literature in recent years have discounted the affects of micromixing on agglomeration and de- agglomeration. Particle agglomeration and breakup have been treated as slow processes when compared to homogeneous nucleation, therefore only influenced by macroscale mixing.
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
M-2-012P
1
Micromixing and Mesomixing Effects on Nucleation, Agglomeration and
De-agglomeration Kinetics in a Dye Precipitation
Kostas Saranteas (speaker)Sepracor Inc.
Gregory BotsarisTufts University
Presented at Lasentec Users’ Forum 2000Orlando, Florida
Production of sparingly soluble materials by precipitation from reactive solutions is a very important unit operation in chemical engineering practice. Precipitation involves a sequence of events including mixing, fa st reaction-diffusion, homogeneous nucleation, agglomeration, and de-agglomeration, with scale sizes ranging from molecular to macroscopic. A number of theoretical models and discussions presented in the literature in recent years have discounted the affects of micromixing on agglomeration and de-agglomeration. Particle agglomeration and breakup have been treated as slow processes when compared to homogeneous nucleation, therefore only influenced by macroscale mixing.
M-2-012P
2
Major Batch Manufacturing Cost Factors
• Raw Materials• Pollution Prevention Systems• Process Cycle
– Synthetic Reaction Step– Solids Formation Step (crystallization/ppt)– Solid-Liquid Separation Step (Filtration-Drying)
• Usually the Process Cycle Limiting Step• A strong function of Slurry PSD
Precipitation processes are used routinely in the batch manufacturing of specialty chemicals. The precipitation step follows a preparative reaction step and is in turn followed by a solids filtration step. In a typical application, precipitation starts when two liquids are brought together and solid particles are formed, either by a chemical reaction between the two liquids or by displacement of a dissolved substance from the solution.
The basic distinction between crystallization and precipitation lies in the fact that most precipitated solids have very low solubilities compared to crystallized substances. For very fast reaction-precipitation processes with semi-batch reactor operation, the reaction and precipitation does not occur throughout the reactor contents, but is highly localized. Reaction zones are typically a few centimeters in thickness. Mesomixing has been defined as the mixing between fluid elements that are coarse-scale relative to micromixing(mixing in the molecular scale), but localized relative to bulk blending ormacromixing.
M-2-012P
3
Fast Reaction(s) with a Phase Change (Precipitation)
• Precipitation– Nucleation-growth of
sparingly soluble materials via neutralization or solvent displacement of reactive or miscible phases.
– Very popular in • fertilizer mfg• catalyst mfg• pharmaceuticals mfg• proteins mfg• Photographic Chemicals
• Design and scale-up Issues– Is PSD, purity, or
morphology affected by meso-micromixing?
– If yes, how is it tested and controlled (agitation rate, feed rate, feed point?)
Mixing parameter effects on PSD (particle size distribution) have been extensively reported in literature. O'Hern and Rush (1963) found that continuous precipitation produces significantly larger particles of barium sulfate than batch processing and mixing in “rapid mixers.” Particularly at higher concentrations, the continuous stirred vessel produces much larger particles. Mulin reported (1982) that in semi-batch experiments, larger crystals of potash alum are obtained, as compared to a batch process. Tosun (1988) found that feeding the two reactant solutions simultaneously to the stirred tank (semi-batch) results in larger crystals than feeding one reactant to a stirred solution of the other.
M-2-012P
4
Cyan Dye Intermediate Process Flow Diagram
1 . ClSO3H 2 . CuPc
3 . SOCl2
Product
Waste
Intermediate
Reaction PrecipitationBy quenching into water
Filtration
This study provides an extensive analysis of a reaction-precipitation system that is based on a manufacturing process for the production of a copperphthalocyanine dye intermediate through a three-step process involving synthesis, precipitation, and filtration. This figure shows the reaction scheme for the original process. The sulfonation of copper phthalocyanine withchlorosulfonic acid is followed by a chlorination step with thionyl chloride or phosphorus oxychloride that leads to the formation of the tetra-sulfonylchloride dye intermediate (Bader and Rickter, 1975). The product is isolated via a water precipitation from the acidic solution followed by a filtration operation.
M-2-012P
5
PSD Evolution from Reaction-Precipitation(In the sec time scale!!!)
Mixing Reaction’(s)
Supesaturation
NucleationGrowth
Transient Particle Population/size
Final PSD
Agglomeration
De-agglomeration
Fast events
Slow Events?
The traditional approaches for quantitative treatment of reaction-precipitation analysis have been to look at agglomeration and secondary nucleation effects as much slower processes and therefore less important than the reaction and nucleation rates for the determination of the final particle size distribution. Numerous examples have been published in recent years where proposed precipitation models ignore or neglect agglomeration when writing general population balances (Aslund and Rasmunson, 1992, Baldyga et. al., 1995, andPohorecki et. al., 1983).
M-2-012P
6
Research Strategyl Experimental system
• In-situ monitoring of PSD during Dye Precipitation
l Experimental Procedure• Measurement Of Experimental Error and Signal to
Noise Ratio for all Parameters of interest (Orthogonal Design of Experiments)
• Identify Important Parameters Affecting PSD
l Modeling Phase• Quantitative Treatment of Experimental Results
An experimental study of the semi-batch precipitation of a copper-phthalocyanine-based dyestuff through a quench neutralization of the reaction solution was done with the use of orthogonal design of experiments testing micromixing and macromixing parameter effects on particle size distribution. A Lasentec® particle size analyzer detection probe installed inside the precipitator allowed in-situ monitoring of the particle size and counts of the precipitate during and after the semi-batch feed.
The results show increased mean particle size with improved micromixing, suggesting competition between homogeneous nucleation and agglomeration. Kinetic parameters were measured for both agglomeration and de-agglomeration rates using well-established, semi-quantitative models that were based on modifications from Smoluchowski’s classical treatment of aggregation. Competing mechanisms of hydrodynamic shear versus particle-to-particle collision were also analyzed, leading to an overall mechanism of fast, collision-dominated agglomeration followed by a slower,hydrodynamically controlled de-agglomeration.
M-2-012P
7
M
TJ
Camile®Temperature
Control
Controlled Feed
1-liter Reactor
Experimental Apparatus for Precipitation Experiments
LasentecDataAcquisition
and PC
Reactionsolution
SyringePump
RTD
RTD
TR
The experimental setup used for the precipitation studies is illustrated here. A one- liter Kontes jacketed glass reactor was used for the quench studies. The batch and jacket temperatures were controlled with a temperature control loop connected to a Camile® 2000 control system and a dual temperature Neslab®
DC-25 unit. A Lightnin® model L1UO8F mixer was used to control agitation speed. The mixer shaft was coupled to a Teflon® shaft that allowed for the interchanging of different types of mixing impellers inside the reactor.
Dimensional characterization of the crystals was obtained with the use of a Lasentec M100C Particle Geometry Monitor. The Lasentec M100 monitor uses an FBRM® (Focused Beam Reflectance Measurement) technique to measure the rate and degree of change to the particle population and particle geometry (a function of the shape and dimension of the particles and agglomerates as they naturally exist in process). The Lasentec FBRM provides both a count over a fixed time and a chord length distribution, which changes with shape and dimension. Used collectively, the rate and degree of change to the particle geometry can be applied to the characterization and control of the crystal system. The reactor head was specially modified to allow the FBRM probe to be placed inside the reactor for in-situ monitoring of the quench slurries.
M-2-012P
8
Typical Parameter profiles during and after semi-batch precipitation runs
0
10
20
30
40
50
60
70
80
90
0 10 20 30 40 50 60 70 80
Time, min
Ch
ord
Le
ng
th M
ea
n S
ize
,
0
2000
4000
6000
8000
10000
12000
14000
16000
18000
20000
Av
g P
art
icle
co
un
ts
Chord Length Mean Size,m
Avg Particle Counts
Fee
d en
d
Post Precipitation hold, ~ 1 hr
This figure illustrates a typical run of the particle size parameters before, during, and after the semi-batch feed was completed. The chord length mean reaches a maximum size rather quickly during the semi-batch feed and then drops fast, approaching an equilibrium size asymptotically, after the end of the quench feed. The particle counts increase fast mainly during the semi-batch feed (primary nucleation) and then slowly increase at an approximately constant rate during the post-quench period (secondary nucleation).
M-2-012P
9
Dynamic Particle Size Distributions During and After Semi-batch Feed
0
200
400
600
800
1000
0 50 100 150 200 250 300 350 400
Particle Size, µm
Par
ticle
Cou
nts/
Cyc
le1-min into FeedEnd Of FeedEnd of 1-hr Hold
This figure illustrates the particle size and count distributionat three distinct periods: 1) right after the initial feed start, 2) at the end of the semi-batch feed, and 3) at the end of the one-hour post-quench time. A significant shift of the particle size and count distribution to a smaller size and more fines during the semi-batch feed is clearly shown between the one-minute and seven-minute periods. The post-quench period results in some further shifting of the relative distributions, but at a smaller scale (even though the relative time was larger).
M-2-012P
10
Particle Size Profiles During and After Feed Addition (exp. # 2)
0
500
1000
1500
2000
0 10 20 30 40 50 60 70
Time, min
Lar
ge
and
Med
ium
Siz
e P
artic
les
0
2000
4000
6000
8000
10000
12000
Fin
e P
arti
cle
Co
un
ts
Large Particle CountsFine Particle countsMedium Particle counts
Feed 1-hr post addition hold
This figure illustrates the dynamic profile from a typical run with the 38 particle size ranges grouped together into three major sub-groups for illustration and simplicity reasons: Fine particles with chord length mean less than 37 µm, medium-size particles with a chord length mean size range between 37 and 105 µm, and large particles with particle size greater than 105 µm.
From this figure we can clearly see two major particle breakage mechanisms at work during and after semi-batch feed: The aggregate fragmentation to smaller aggregates and attrition of aggregates to generate fine particles. Aggregate fragmentation to smaller aggregates is more apparent during the semi-batch feed and is limited to aggregate sizes greater than 105 µm. At the end of the semi-batch period, some further aggregate break-up to smaller aggregates is observed, but during the post-quench period, we can conclude that the particle counts of the medium-size and large aggregates stay approximately constant. The fine particles, on the other hand, never go through a de-agglomeration phase, continuously increasing with time even after the post-quench period. This suggests a surface-attrition type of de-agglomeration mechanism.
M-2-012P
11
Possible Experimental Parameters affecting PSD
l Chemistry Parameters (unique to specific chemistry)
––– Reaction reagentsReaction reagentsReaction reagents––– ReactionReactionReaction StoichiometryStoichiometryStoichiometry––– OtherOtherOther
l Process Parameters (Of general interest)
––– Mixing Impeller typeMixing Impeller typeMixing Impeller type– Mixing Power––– Baffled Baffled Baffled vs.vs.vs. UnbaffledUnbaffledUnbaffled reactorreactorreactor––– Addition RateAddition RateAddition Rate
– Feed Point Location– Feed Tube Diameter
The experimental results reported using the three- level factorial design experiments show that both total particle counts and particle mean size are strong functions of feed point, even at the 95% confidence level. Such an effect, as was indicated in the introduction, suggests that both response parameters are micromixing controlled.
M-2-012P
12
Major Mixing Parameters
Feed Location
Below Impeller
AtImpeller
AboveImpeller
AboveSurface
FastRate
FastMulti-feed Rate
Feed Rate Power Distribution
Low Rate
SingleImpellerNo Baffles
SingleImpellerWith Baffles
LargeD/T Ratio
As the feed point moves from the surface wall to surface center, and eventually to below surface at the impeller tip, the mean particle size inc reases and the total particle counts decrease respectively.
M-2-012P
13
Experimental Design for 3 factors 9 experiments, L9
Experiment#
Experimentalerror
Feed tubediameter,
mm
Agitationrate, rps
Feed Point
1 - 0.3 2 Surface- wall
2 - 0.5 4 Surface- center
3 - 0.8 7 At impeller
4 - 0.3 4 At impeller
5 - 0.5 7 Surface-wall
6 - 0.8 2 Surface-center
7 - 0.3 7 Surface-center
8 - 0.5 2 At impeller
9 - 0.8 4 Surface-wall
This table summarizes the three-factor, three- level orthogonal experimental design. The impeller used was the axial type. Nine experiments were performed in the same experimental apparatus. One and a half liters of water was charged in the reactor and cooled to 0ºC. The impeller was set at the required speed and 44 cc (80 gr.) of reaction solution was pumped through the appropriate size feed tube at the location called by the particular experiment. The Lasentec FBRM particle size analyzer probe was installed at about three inches below the surface in order to continuously monitor the slurry dynamic profiles. After the feed quench precipitation was completed, the slurry was left under the same mixing conditions for an additional hour in order to study the particle dynamic profiles as affected by secondary nucleation conditions.
M-2-012P
14
Experimental Results for 9 experiments
Experiment # Scan countmean particlesize (microns)
Volume equiv.mean particlesize (microns)
Populationbalance
(total counts)
1 64.4 241.4 13232
2 63.5 284.9 121133 60.6 308.7 12236
4 88.4 316.8 8621
5 43.4 214.8 18049
6 78.0 329.7 9792
7 60.2 264.0 12853
8 77.8 338.7 9781
9 60.2 276.8 13288
• Scan count = No Wt
• Volume equiv. = Cube Wt
This table summarizes the results of the nine experiments for the average values of each of the response parameters monitored.
M-2-012P
15
ANOVA table After Pooling(Vol. Equiv. Mean as Response parameter)
Source Df Varianceratio
Variance %contribution
Feed Point 2 36.6 66.1
Mix Power(linear)
1 21.1 19.0
Feed Tube(linear)
1 12.5 11.2
Pooled error 4 3.6
Fc,0.05=7.71
This table summarizes the analysis of variance results for the nine experiments using the volume equivalent (cube wt chord) as the response parameter. The results show that the power dissipation, feed tube size, and feed point are statistically significant factors at the 95% confidence level on the mean particle size outcome. The feed point is the most significant factor tested, contributing approximately 66% of the total variance measured, with the power dissipation contributing to 19% of the total variance and the feed tube size contributing to 11% of the total variance.
M-2-012P
16
Summary of Experimental Findings• Feed Point Introduction is the most Important
Parameter Affecting PSD, Counts– Sub-Surface addition increases mean size and decreases Particle
counts
• Mixing Power has also statistically significant but less important effect on PSD, Counts– Increased Agitation Rate Decreases PSD, increases Counts
• Feed Tube Diameter affects aggregate mean size– The larger the tube diameter the larger the mean size
As the feed point moved from the surface wall to surface center, and eventually to below surface at the impeller tip, the mean particle size increased and the total particle counts decreased. Increasing the agitation rate from 120 rpm to 420 rpm decreased the mean particle size from 303 µm to 262 µm. Increasing the agitation rate over the same range also increased the total particle counts from a value of approximately 11,000 to a final value of approximately 14,500 particle counts. Finally, a statistically significant correlation resulted between the feed tube diameter and volume equivalent mean size. As the feed tube diameter increased from 0.3 mm to 0.8 mm, the mean particle size increased from a mean size of 272 µm to a mean size of 305 µm.
M-2-012P
17
Fundamental Questions
• What does Feed Point Location mean in a fast reaction-precipitation system?
• Why introduction into a more turbulent region increases particle mean size and decreases particle counts while increasing agitation rate has the opposite effect?
Experiment results clearly show that the selectivity parameter is much more affected by the feed point selection than by the agitation rate.
M-2-012P
18
A AA
B
BB
B
BB
B
BB
A A AA A
λκ
Macro- and Micromixing in Reactive Mixture
A + B C
λk = { }1/4ν 3
Ε
Macromixing: Mixing due to shear forces on the scaleof the whole system
B
BB
Micromixing: Mixing due to molecular diffusion on thescale of dimensions less than λκ
B
Existence of a feed point effect indicates a mesomixing (localized) or micromixing (molecular diffusion) controlled precipitation mechanism. Introduction of a feed into a more turbulent region significantly increases the nucleation rate, but also (due to the localized nature of the micromixing) dramatically increases the particle concentration in the neighborhood of the reaction zone. Macroscale de-agglomeration is a much slower process and is predominantly controlled by the macroscale mixing characteristics of the mixing environment.
M-2-012P
19
QAA
A A A
A AA A
A
A
AB
B
BB
BB B
C
C CC
CC
C
Mesoscale Mixing (for very fast reactions)
- Coarse mixing relative to Micromixing
- Localized relative to Macroscale mixing
B
B
B
B
A + B C
Mesomixing is defined as the mixing between fluid elements that are coarse-scale relative to micromixing (mixing in the molecular scale), but localized relative to bulk blending or macromixing.
M-2-012P
20
A B
C
3.58
0.26
0.06 0.07
0.29
0.03
1090
513
1693
PartecProbe
Energy distribution factor Φ (r,z) for L9 Experimental DesignE r z E
E N N D
loc avg
avg P
= ⋅
= ⋅ ⋅
Φ( , )3 2
The selectivity difference can be used as a basis for estimating the localized energy dissipation difference. The results clearly show that the re is dramatic difference between average and local energy dissipation in the reactor system tested, with the subsurface feed region two to three orders of magnitude greater than the average values.
M-2-012P
21
Feed Point Effect on PSDSub-surface additionImproved Micromixing
(Eloc Increased)
Smaller λKSmaller Diffusion path length
Faster Reaction RateHigher Localized supersaturation
Faster Nucleation Rate
Agglomeration /De-agglomerationRate much slower than nucleation rate
Smaller ParticleMean Size
Agglomeration De-agglomerationRate not slower than nucleation rate
?
Agglomeration resulting from precipitation reactions can be an extremely fast process competing with, rather than following, the micromixing reaction and primary nucleation events. Agglomeration is enhanced by improvedmicromixing due to the high concentration of stable nuclei formation and their interaction in the localized precipitation zone. Agglomeration is much faster than the de-agglomeration process. Agglomeration is also controlled by a particle-to-particle collision mechanism. De-agglomeration is also affected by the local conditions in the reaction zone. Initial de-agglomeration is primarily controlled by hydrodyamic shear, with some effect also caused by the aggregate collision.
M-2-012P
22
Influence of Better Mixing on Precipitation Processes*
The effect of feed point in a semi-batch reaction-precipitation has also received significant attention in the recent literature (Baldyga et. al., 1992) regarding precipitation processes. B. Marcant and R. David (1991) report that increasing stirring rate with the same feed point not only enhances primary nucleation, but also enhances secondary nucleation and agglomeration rates. Therefore, it is difficult to predict the effect on particle size or number of crystals.
However, the same authors, when discussing the addition of the feed in a more turbulent zone with the same overall stirring rate, say it becomes a test for the presence of micromixing effects in precipitation. In an attempt to explain the effect of feed point introduction on particle mean size, they speculate that feeding in a more turbulent region only enhances primary nucleation. Therefore, according to them, because of the increased nucleation rate, we should expect increased number of crystals and decreased mean size with enhanced micromixing.
M-2-012P
23
DISRUPTION AGGREGATION
Surfaceattrition
Breakage
Single celladdition
Aggregateaglomeration
Schematic representation of aggregate formation and disruption
According to the empirical models in aggregate formation-disruption, a finite aggregate concentration is not reached. Brown and Glatz (1987) introduced a modification in the first and second order disruption processes by assuming that the concentration of aggregates to be broken up is equal to the difference between the actual aggregate concentration and the equilibrium aggregate concentration. Because an equilibrium value is also achieved in the formation process, the same assumption is extended to the formation process.
M-2-012P
24
Agglomeration/De-agglomeration Mechanismsand Kinetic Models
I. Particle-Particle Collision Mechanism
II. Hydrodynamic Shear Mechanism
Agglomeration/De -agglomeration (for equal size particles)
( )
[ ( )
,
dNdt
K N
NN
NeqN
e
f d N
eq
N K teq N
≈ − ⋅
=− − ⋅ − ⋅ ⋅
22
0
1 1 2
De-Agglomeration
( )
( )
dNdt
K N
N N N e N
d N
eqK t
eqN
≈ − ⋅
= − ⋅ +− ⋅
1
01
These equations can assist in defining the controlling aggregation/break-up mechanism for both the micromixing controlled period during semi-batch feed and the macromixing controlled period after the end of the feed. For aggregate particles of the same size, the hydrodynamic shear based contribution is described by a linear dependence of rate of change of particle counts on actual counts. The particle-particle collision mechanism calls for a quadratic dependence of rate of change on actual particles of the same size.
M-2-012P
25
Rate of change of 250 m size particles during and after
semibatch feed (experiment # 6)
-30
-20
-10
0
10
20
30
40
50
60
70
80
0 50 100 150 200
Particle counts of size 250, N
d N
/ d
t Agglomeration
De-Agglomeration
Particle count profiles for average size of 250 µ
0
20
40
60
80
100
120
140
160
180
200
0 10 20 30 40 50 60 70
Time, min
Par
ticle
cou
nts
, N
End of
semibatch feed
These figures illustrate a typical count-versus-time and rate-versus-counts profile for aggregate size of 250 microns. The choice of this particular size was based on the fact that it belongs in the size range where significant agglomeration and breakup occurs during and after the semi-batch feed, making it a good choice for studying the proposed models.
In the first figure, the number increases rapidly at the start, suggesting fast agglomeration until the number of aggregates reaches about 200. De-agglomeration starts during the semi-batch phase, but at a much lower rate. The rate-versus-counts plot in the second figure shows a distinct parabolic profile during the agglomeration phase. The de-agglomeration rate profiles do not show any distinct pattern and are characterized by a significant noise-to-signal ratio, making it difficult to draw any conclusions.
M-2-012P
26
Counts Based Agglomeration Regression Summary for 250 µm size aggregates (experiment # 6)
Quadratic Regression Summaryy = -0.0078x 2 + 1.5604x - 1.3191
R2 = 0.937
KN2,f = .0078 min -1
Linear Regressiony = -0.0154x + 44.165
R2 = 0.0017
0
10
20
30
40
50
60
70
80
0 20 40 60 80 100 120 140 160 180 200
Particle counts at 250 µ m average size, N
d N
/ d
t , m
in-1
The parabolic profile of the agglomeration rate is reproducible for all experiments, suggesting a particle-to-particle collision mechanism rather than shear- induced agglomeration. This figure illustrates a first- and second-order regression analysis for one of the experiments (#6), clearly showing a second-order dependence of agglomeration rate on particle counts. From the regression results, the agglomeration kinetic constant can be estimated.
M-2-012P
27
Summary-Conclusions• In Reaction-Precipitation systems agglomeration rate is an
extremely fast process competing with rather than following reaction/nucleation events.
• Existence of a feed point (Eloc) effect on Particle Size suggests meso-micromixing controlled precipitation mechanism. The resulting PSD is controlled by the slowest step in the precipitation process.
• Hydrodynamic Shear (mixing) can enhance agglomeration through increased aggregate or particle to particle interaction in the localized precipitation zone
• Hydrodynamic shear (mixing) can enhance de -agglomerationthrough:– particle to particle collisions in the localized precipitation zone– Surface attrition primarily in the localized mixing region and secondarily in
the macro-scale mixing environment
• More Fundamental studies needed to generalize conclusions
Agglomeration resulting from precipitation reactions can be an extremely fast process competing with, rather than following, the micromixing, reaction, and primary nucleation events. Existence of a feed point effect indicates a mesomixing (localized mixing) or micromixing (molecular diffusion) controlled precipitation mechanism. Introduction of a feed into a more turbulent region significantly increases the nucleation rate, but also (due to the localized nature of the micromixing) dramatically increases the particle concentration in the neighborhood of the reaction zone. Agitation rate can enhance agglomeration rate through increased particle-particle collision frequency during the localized precipitation phase. Agitation rate can also decrease the agglomerate particle size, primarily through a hydrodynamic shear mechanism, but the rate of de-agglomeration is significantly slower than the corresponding agglomeration. A reaction-precipitation general population balance model should not ignore the agglomeration effect as a slower event without experimental testing. Although such assumption simplifies the mathematical analysis, it can also become a source of significant error.
M-2-012P
28
Acknowledgements
• Polaroid Corporation • Tufts University• Sepracor Inc.• Lasentec Inc.• Others
The authors would like to thank Polaroid Corporation for the continuous support of this research. Our appreciation also goes out to Tufts Universiy,Sepracor Inc., Lasentec Inc., and others who made this work possible.
M-2-012P
29
Questions and Answers
Q: I might be quibbling but, if we look at slide 21, does it necessarily mean you have a higher localized supersaturation when you improve the mixing? I would have thought it would have been the other way?
KS: If you have increased localized mixing rate, what happens is that your micromixing scale will decrease. So your diffusion length between that pathway has to be reacted, or it will automatically increase your reaction rate, because it’s a thinner film now.
The point is that it’s a very localized event. The reaction is completed within a few milliseconds after entering the point. If you look at the reaction kinetic between neutralization, you’re talking tenths of seconds or an extremely fast process.
What you’re aiming for is: How can I improve micromixing? I can introduce into a region where I can get more reaction rate by minimizing that thickness layer, the diffusion layer called molecular diffusion, and that is what we think is happening there. The introduction at that feed point accelerates the reaction rate, which by definition generates higher supersaturation in the product.
M-2-012P
30
Questions and Answers
Q: So by increasing mixing you are increasing supersaturation?
KS: Let me go through the steps. Suppose you introduce in a more turbulent region. You remember the micromixing scale? It’s a function of the local energy and the kinetic disbursement. As you increase the micromixing, you reduce the diffusion length, and you have a complete molecular diffusion. By reducing that, the reaction rate is accelerated. Your reaction rate is accelerated, you generate the precipitate, which in this case is your product, and that happens in a very localized region. So what happens is you have stable nuclei in the reaction zone, which is a very small zone, but they interact. After interaction, before they migrate to the bottom of the reactor, they collapse together and form large agglomerates. That’s why I can explain why the agitation rate almost has the opposite effect, but why the introduction to the more turbulent region does not.
M-2-012P
31
Questions and Answers
Q: What is the size of the primary particle increase? Are they different under different circumstances?
KS: Yes, we have got an extremely small particle size. If you look at it as a real agglomerate, the agglomerates can look as high as 500 microns, and individually it’s a few micron grains. So you can see the very small nuclei collapse together to form an agglomerate after precipitation.
M-2-012P
32
Questions and Answers
Q: Can you explain your larger particle size? You have very small nuclei, so the typical particle size is larger, and the agglomeration is faster. Instead of making larger particles, that’s why you have very high saturation?
KS: The best that I can describe it is that you form this precipitate, these stable nuclei in the smaller form, and they have to interact. They are large, so there is a lot of colliding and they collapse into agglomerates.
M-2-012P
33
Return to Table of Contents
Related Documents
