Upload
others
View
0
Download
0
Embed Size (px)
Citation preview
Process Development Strategies for
Energetic Materials
P. Dube, M. Jorgensen, S. Lenahan, L.J. Ping and J. S. SalanNalas Engineering Services, Incorporated, Centerbrook, CT
860-861-3691
1
What do we do?
• Transition chemical processes to the plant environment– Identify engineering challenges including heat transfer, mass transfer, and kinetics
– Evaluate chemistry in the laboratory using in situtools (IR, Raman, FBRM, PVM, heat flow)
• Evaluate pilot and production equipment. Validate processes through scale-down experiments.
• Develop low-cost chemical processes
2
Our Model
3
Process Development
Engineering
ChemistryAnalytical
• process understanding
• Scale-up support
• process modeling
• predictive experimentation
• scale-up assessments
• in-situ monitoring
• low-cost processes
• screening guidance/support
• bond forming steps
• transfer best technology available
• analytical support for reactions
• isolated intermediate and API
characterization
• in-situ support for process
monitoring and understanding
• Teamwork
• understand goals
• understand and accept
responsibilities of each group
Energetic Materials
Simply Special Chemicals
• Explosives, propellants, pyrotechnics,
reactive materials, related chemicals and
fuels, and their application in propulsion
systems and ordnance
4
Explosives
• Stimuli that can make materials react
– Impact
– Friction
– Electrostatic discharge
– Thermal
• In-situ monitoring makes sense
– Heat flow
– Spectroscopy
– Particle size analysis
– Gas evolution
5
Bond Forming Steps
Oxidations
6
Intermediates in the oxidation of an amine to the nitro
N
N N
NH2NH2
CH3
N
N N
NH2NO2
CH3
DAMT
N
N N
O2N NO2
CH3
DNMT
NAMT
Selectivity
Tailored Reactivity
Selectivity
We love nitro compounds
Selectivity
• Your toolbox…
• Kinetics
– DynoChem
• Mass Transfer
– VisiMix
• Heat Transfer
– Mettler Toledo AutoChem
7
N
N N
NH2NH2
CH3
N
N N
NH2NO2
CH3
DAMT
N
N N
O2N NO2
CH3
DNMT
NAMT
N
N N
NH2NH
CH3
OH
N
N N
NH2N
CH3
O
N
N N
NH NO2
CH3
OH
N
N N
NNO2
CH3
O
N
N N
NH2N
CH3
N
NN
NH2N
CH3
N
N N
NH2N
+
CH3
N
NN
NH2N
CH3
O-
Process Details
• Add hydrogen peroxide
• Add DAMT-HNO3
• Add sodium tungstate dihydrate
• Ramp to T=80°C over 2 hours
• Hold for 4 hours
• Notes:
– No control of pH
– All ingredients in the reactor prior to heat ramp8
Heat Flow• Potential challenges
– Hydrogen peroxide’s stability is influenced primarily by temperature, pH and impurities.
– The existing process calls for the addition of all reagents at low temperature…1. Hydrogen peroxide
2. DAMT-NO3
3. Catalyst (sodium tungstate)
Followed by a temperature ramp to T=80°C
– Having all ingredients in the reactor presents a dangerous situation; the reaction has the potential to release all heat associated with the reaction(s) at one time
– PLUS hydrogen peroxide stability is questionable9
Heat Flow • How do we typically control heat generated from a reaction?
– Dose reagent or substrate to control conversion as well as accumulation (unreacted) of starting materials
– Elevate the temperature (activation energy) to assist in getting the reagents to react faster and minimize accumulation
– Avoid temperature ramps without knowing
1. exactly how much unreacted material remains prior to ramp
2. how much heat will be generated when it is converted at the ramped temperatures (know your side reactions!)
3. Can your system handle it?
10
11
(Francis Stoessel, Thermal Safety of Chemical Processes, Risk Assessment and
Process Design. 2008, Weinheim: Wiley-VCH. ISBN: 978-3-527-31712-7)
Dec. 23, 1969. Year of Woodstock!
Increased demand for blue jeans.
12
(Francis Stoessel, Thermal Safety of Chemical Processes, Risk Assessment and
Process Design. 2008, Weinheim: Wiley-VCH. ISBN: 978-3-527-31712-7)
Reactor afterwards. The lid was 700 meter away from the site. Five fatalities, 31
injured. There was a lot of pride in the city of Bassel because of the chemical
industry doing so well. There was consequences to people being killed and injured.
Heat Flow Summary
13
Qflow: heat flow through the reactor wall U: overall heat transfer coefficient of the reactor wall W/(m2 K)A: heat exchange area (wetted area) m2Tr: temperature of the reactor contents °CTj: temperature of the jacket °C
Qaccu: heat storage (accumulation) by the reaction mass Mr reaction mass (sum of all additions and dosings) gCpr specific heat capacity of the reaction mass J/(g K)dTr/dt current change in the temperature of the reaction mass K/s
Qdos: heat input due to dosing dMr/dt mass flow of all added substances g/sCp dos specific heat capacity of the added substances J/(g K)Tr current temperature of the reaction mass °CTdos temperature of the added substances °C
Qreaction =Qflow + Qaccumulation + Qdosing
Qr = U A . (Tr - Tj) + Mr . Cpr . dTr/dt + dMr/dt . Cpdos . (Tr - Tdos)
Tdos
Qdos
Qaccu
QflowTj
Tr
Calibration Calculated Enter value
Heat of reaction calculation (Qr) based on heat flow balance:
Heat Flow Summary w pH
14
Heat Flow Summary w pH
15
• Hydrogen Peroxide (35%)
Heat Flow Summary w pH
16
• Hydrogen Peroxide +
• DAMT-NO3 Salt– Slurry
• Mass transfer…what happens when things go into solution?
Heat Flow Summary w pH
17
• Hydrogen Peroxide (35%) +
• DAMT-NO3 salt +
• Sodium Tungstate
Heat Flow Summary w pH
18
• Temperature Ramp, Tr~30°C
• Solution-note heat output
FBRM
• FBRM probe optics are rotated
at a controlled high speed, and
the focused laser beams scan
across the particle system in a
circular path19
Introduce FBRM
• FBRM great way to nail down exactly when crystallization events occur.
20
Black line = FBRM counts
Heat Flow Summary
21precipitation
• As ramp progresses, gas
evolves
• Solids form
dissolution ?
N
N N
NH2NH2
CH3
N
N N
NH2NO2
CH3
DAMT
N
N N
O2N NO2
CH3
DNMT
NAMT
Selectivity
Tailored Reactivity
Selectivity
Black line = FBRM counts
Decomposition?
• Shortly after solids form, gas evolution is noticed (T~63°C)
• Gas evolution continues as ramp progresses producing a large foam cap over time.
• Notes– Gas evolution significantly influences heat flow and subsequent calculations
– Decomposition reactions generate heat. This is not produced from your desired reaction. It is very important to know which reactions contribute heat and quantify these contributions.
22
Dissolution Again!
• ~25 minutes after ramp complete, T=80°C.
• Whatever intermediate that is forming is soluble in the reaction mixture.
• Finally, solids precipitate again but are not pure. Solids are a mixture with primary component desired product. Big ‘gumball’ forms.
23
Heat Flow
• Customer: Going to the 200-gallon scale
• How much heat can the reactor remove?
24
Black line = FBRM counts
What is safe?
25
Data source: Pilot Plant Real Book
Good rule of thumb is 50-60 Watts/liter
Watts/Liter f(t)
• l
26
W/l
∆∆∆∆Tadiabatic = 278 K!
Temperature Ramp
• In the laboratory the surface area to reaction volume is high. Very easy to remove heat through jacket wall.
• In the plant, the surface area to reaction volume is low. Very difficult to remove heat through jacket wall.
• Quantifying (RC1) the heat AND profile are key to determining scaleable process.
• Suggestion: Eliminate the temperature ramp through controlled dose of hydrogen peroxide.
27
Reactions
• Desired-
• Undesired-
• Undesired-28
N
N N
NH2NH2
CH3
N
N N
NH2NO2
CH3
DAMT
N
N N
O2N NO2
CH3
DNMT
NAMT
Selectivity
Tailored Reactivity
Selectivity
2 H2O2 → 2 H2O + O2
2 H2O2 + catalyst → 2 H2O + O2 + catalyst
Hydrogen Peroxide
29
Hydrogen Peroxide
+ Catalyst
30
Emergency program initiated
Overlay Heat
• Competitive reactions, little control during scale-up
31
Reaction
Hydrogen Peroxide
Hydrogen Peroxide
& Catalyst
• As temperature rises, kinetics faster
• As temperature rises, hydrogen
peroxide consumed!
Pics Hydrogen Peroxide
+ Catalyst
• As the temperature progresses more gas is generated until finally the material ejects from the reactor.
32
Video Hydrogen Peroxide
+ Catalyst
33
SummaryMto this point
• Reaction is highly exothermic and includes temperature ramp
• The 200-gallon Pfaudler will not adequately remove heat generated from this reaction.
• Decomposition of hydrogen peroxide is contributing to some degree. Catalyst exasperates the decomposition.
• The degree of decomposition depends on ramp rate and kinetics of reactions– As hydrogen peroxide is consumed to produce intermediates, products and impurities, it is no longer available to participate in decomposition reaction
– For this reason, the amount of heat generated from decomposition is highly dependent on temperature ramp rate (temperature) and kinetics.
• Scaleable?34
Analogous System
C C
N
O
N
N N+
CC
N
O
N
NN
H
H H
H
O
C C
N
O
N
N NO2
H
H
C C
N
O
N
N N
H
H H
H
C C
N
O
N
N NHOH
H
H
C C
N
O
N
N N O
H
H
C C
N
O
N
N N
CC
N
O
N
NN
H
H H
H
3,4 diaminofurazan 3-amino,4-nitrosofurazan3-amino-4-(hydroxyamino)furazan
diaminoazoxyfurazan
aminonitrofurazan
diaminoazofurazan
DAAzF
DAF ANF
DAAF
35
EasyMax Screening Results
pH Lot Number
Area Percent (Isolated Solids)
DAF ANF DAAF DAAzF16.03
min
16.51
min
Isolated
Yield
DSC
Onset
(°C)
Varies BATCH 1.52 0.13 94.89 0.44 1.23 0.96 N/A 253.34
6 NAL-02-111 1.40 0.06 91.25 1.17 3.50 2.63 82.76% 250.19
6.5 NAL-02-114 1.28 0.08 93.58 1.20 2.13 1.73 80.88% 252.19
6.8 NAL-02-134 1.25 0.08 95.51 0.42 1.37 0.99 75.41% 254.47
7 NAL-02-110 1.23 0.08 95.77 1.05 0.95 0.92 57.16% 254.48
Thermal stability increases with increasing pH
Yield decreases with increasing pH
36
N+
N NO
NH2N
NO
NH2
N
N+
N N
O
N
O- O
-
N
N NO
NH2N
+
NO
NH2
N
N+
N N
O
N
O-
O-
Batch Experiments• Challenge: charge reagents as fast as possible while maintaining pH.
• Approach: Using pH control feedback loop, charge oxidizer quickly. Quench ENTIRE reaction at different time points. Dilute entire reaction mixture as ‘sample’ for HPLC analysis. Repeat as necessary.
• Results: Input concentration profiles to DynoChem for fitting.
Solution.DAAzF (Exp) (mol)
Solution.DAAF (Exp) (mol)
Solution.ANF (Exp) (mol)
Solution.DAF (Exp) (mol)
Solution.DAAzF (mol)
Solution.ANF (mol)
Solution.DAAF (mol)
Solution.DAF (mol)
easymax eight reactions at pH6.5 25C
Time (min)
Pro
ce
ss p
rofil
e (
se
e le
ge
nd)
0.0 30.0 60.0 90.0 120.0 150.00.0
0.0012
0.0024
0.0036
0.0048
0.006
37
About Mixing
• Mixing is often neglected until there is a problem at the plant scale. Why?
• Mixing DOES NOT SCALE
38
Round bottom flask
About Mixing
• Round bottom flask-choice of chemist
• OptiMax-choice for process development
• Batch Plant Reactor-choice for manufacturing
• What ties this equipment together during scale-up?
39
Round bottom flask
Batch OxoneTM Process
• Charge water to reactor
• Charge 3,4 diaminofurazan to reactor
• Charge sodium bicarbonate to reactor
• Charge oxidizer slowly to reactor (gas generation)
• 2nd charge sodium bicarbonate to reactor (adjust pH)
• 2nd charge oxidizer to reactor
• Filter solids
• Good stuff!
40
Nalas developed a continuous process and demonstrated
it in the laboratory producing 300-grams high quality product
Proof of Concept
• Produced over 300-grams
– 2-liter and 10-liter in series
• Particle size distribution controlled with continuous process
41
2-liter
10-liter
Filtration
1. DAF solution
2. Oxone Solution
3. Sodium Carbonate
Solution
1 2 3
Aft
er
2nd
Do
se
Oxo
ne
ho
ld p
eri
od
Results, 2-liter/10-liter
Batch
Continuous
2-liter to 10-liter run
In-situ PVM images
Design of 50-liter CSTR
Reactor 4100
Reactor 4200
ph1
ph2
Scale 1/DAF
Scale 2/oxone
Scale 3/sodium
carbonate
Pump 3
VFD: 3
Pump 1
VFD: 6
Pump 2
VFD: 5
Pump 5/DAAF slurry
Duplex Head
VFD: 4
DAAF Slurry
E-33Scale 4/sodium
carbonate (Optional)
Pump 4
No VFD
VFD: 1
VFD: 2 Buchner Filter
Process Water Drum
Diaphragm
Pump
43
MIXING MODEL NOT COMPLETE AT THIS TIME
Schedule = Go Go Go!!!!!
Pilot Scale-Two 50-liter CSTRs in Series
44
• Operated from control room
• Minimal operator ‘hands-on’ requirements
• Product had impurity that influenced thermal
stability including initial melt.
• What is the best pH?
Particle Comparison
45
1st 50-liter continuous campaign
“spherical DAAF”
2-liter to 10-liter run
Reactor Design
• Kinetic Model
– Used DynoChem
– Levenspiel plots
– Validation
• Mixing Model
– VisiMix
– Validation
46
Precipitation and Solids Dispersion
• Product is not soluble in water (solvent)
• Precipitation occurs immediately upon formation of product
• If crystals are present, growth and/or agglomeration may
dominate over nucleation
• Poor dispersion may lead
to some particles residence
time being longer than others
47
Solid-Liquid Mixing
• Dispersion of solids-physical process where solid particles or aggregates are suspended and dispersed by the action of the agitator in a fluid to achieve a uniform suspension or slurry.
• What about a continuous process?– Batch vs. continuous crystallization
– Scale-up
48What does our system look like?
poor better Ideal!
Scale-down Experiments
• Mixing parameters can be matched between EasyMax, OptiMax,
RC1, lab-scale equipment, pilot-scale equipment and production
equipment
• Mixing challenges can be identified and resolved early in
development
• It is pointless to generate data in the lab that is not representative
of data that will be generated in the plant. A mixing sensitivity
can make lab data useless and any resulting kinetic model
generated from lab data.
Reactor Geometry
• Mixing does not scale
• Use VisiMix to model conditions in the plant environment
• Use VisiMix to match these conditions in the laboratory
• Gain confidence in your process!
50
VisiMix Model
• Identify methods to control morphology and particle size
51
rpm
impeller
Definitions
• Maximum degree of non-uniformity (axial, radial)– This parameter characterizes the maximum deviation in local concentrations of the solid phase
from the average concentration in the tank. Complete uniformity of the solid phase distribution is
not always necessary. However, if actual non-uniformity is higher than 25-30%, partial settling or
flotation may occur.
• Average concentration of solid phase in continuous flow– This parameter is related to continuous flow processes. The concentration of the suspension in
the outlet point is assumed to be equal to the concentration in the feed flow. The difference
between the actual average concentration in the tank and the concentration in the flow depends on
the degree of uniformity of axial and radial distributions and on the outlet position.
• Relative residence time of solid phase– This parameter is related to continuous flow processes, and is estimated as a relation of the mean
residence time of the solid phase to the mean residence time of the suspension. Mean residence
time of the solid phase depends on the relation of local concentration in the outlet point to the
average concentration of the solid in the tank. The higher is the degree of uniformity of
distribution, the smaller is the difference in the residence times for the solid and liquid phases.
This difference can also be reduced by a correct choice of the outlet position.
52
Turbulence Parameters150 rpms 230 rpms
Energy dissipation-maximum value (W/kg) 12.7 45.8
Energy dissipation-average value (W/kg) 0.182 0.655
Characteristic time of micromixing (s) 4.00 2.11
150 rpms 230 rpms
Maximum degree of non-uniformity-axial, % 298 166
Maximum degree of non-uniformity - radial, % 13.6 10.5
Average concentration of solid phase in continuous flow (kg/m3) 4.44 6.31
Relative residence time of solid phase (ratio) 1.80 1.27
Solid-Liquid Mixing Parameters
Average concentration of solid phase in continuous
flow approaches ideality (8.0 kg/m3)
rpm
Single paddle
impeller 230 rpms
Single pitched
impeller 230 rpms
Dual pitched
impeller 230 rpms
Max degree of non-
uniformity –axial, %
35.4 39.0 97.0
Average concentration of
solid phase in continuous
flow (kg/m3)
6.32 6.08 6.86
Relative residence time
(ratio)
1.27 1.32 1.17
Characteristic time of
micro-mixing (sec)
4.63 3.79 2.79
54Best for pH control
impeller
Flow Patterns• Dual impeller-second impeller actually interferes with flow pattern, resulting in higher degree of non-uniformity
55
39% non-uniformity
97% non-uniformity2nd impeller causes local
high degree of non-uniformity
Comparison
56
2-liter/10-liter Setup Initial 50-liter setup Revised 50-liter setup
poor better Ideal!
You can always lower your dip tube!
Cost Avoidance?• Producing the material initially included the following estimated costs:– Labor ~$60K (processing, drying, packaging)
– Materials ~$50K (DAF starting material alone at $1,000/kg plus 8 week lead time)
• Not only as the material expensive but will set the project timeline further to the right due to long lead times
• Some pharma raw ingredients can cost 10Xs as much
– Analytical ~$35K
– Subtotal ~$145K (one step)
– Remember, this is ONE synthetic step! (i.e. 10 steps ~$1.5M)
• Modeling effort cost~$30K– Model reactors/process (dimensions, material properties)
– Scale-down experiments
– Analytical
– Report
• Modeling effort schedule~2 weeks total
57
Scale up Thoughts
• Producing chemistry is a challenge all by itself. Pharma targets can have 10-15 steps in Phases I/II.
• Modeling allows the user to quickly and efficiently (low cost!) evaluate how your process will run in the exact equipment to which you scale.
• Synthesis Workstations (EasyMax/OptiMax) provide a consistent and affordable solution for model validations (scale-down experiments)
58
Scale up Thoughts• Utilizing the proper tools (models and automated reactors) allows the chemical engineer to reduce risk upon scale up, provide the project a higher chance of success with fastest possible path to market.
• Chemical Engineering is not magic! It is a science that assists the team in producing processes that work in the equipment to which you will produce your product.
59
Final Summary• VisiMix allows for quick and efficient investigations related to mixing that minimize risk upon scale-up
• VisiMix allowed us to evaluate a specific reactor configuration, make appropriate changes and produce quantities of high quality material for our customer
• New focus is on reducing cost of starting material (DAF)
• Optimizing equipment trains for commercial manufacturer of product (DAAF)
• Having and knowing how to use the right tools makes the difference between SUCCESS and FAILURE
• Modeling is proven to save development time and MONEY
60