Reaction Engineering in the Pharmaceutical Industry David J. am Ende Engineering Technologies / Chemical R&D Pfizer, Inc. 13 th International Process Development Conference September 17 th -21 st , 2006 Newport, Rhode Island
Reaction Engineeringin the Pharmaceutical Industry
David J. am EndeEngineering Technologies / Chemical R&D
Pfizer, Inc.
13th International Process Development ConferenceSeptember 17th-21st , 2006
Newport, Rhode Island
Engineering Strategy• Process Characterization
– Enable the portfolio by delivering fundamental process understanding prior to scale-up through characterization of the critical rate processes.
– Develop new process characterization tools
New Technologies for Manufacturing – Increase productivity– Reduce manufacturing costs– Process Intensify
Where do Problems Occur on Scale?Numerous physical and chemical interactions exist
Kinetics
HeatTransfer
ThermodynamicEquilibrium
Mass Transfer
Mixing
Physical PropertyChanges
Scale-up is about understanding the important rates processes as they change with scale
Engineering Focus Areas
Batch
Continuous
Process Characterization
Process Design
Process Characterization & Design
• Reaction Kinetics• Heat Transfer• Mixing• Fluid Properties• Phase Equilibria• Process Safety• Process Modeling
Avoid Surprises on Scale
Right First Time• Dynochem• Visimix• Fluent• Aspen• Cosmotherm• ReactIR/ConcIRT• Athena• Numerica• Matlab• Design Expert• FusionPro
Presentation Outline
• Continuous Flow Chemistry• Characterization of Reaction Kinetics• DOE vs Kinetic Modeling• Future Horizons
Continuous Flow
Potential Advantages of Continuous FlowBenefits in R&D
– Enable High Energy Chemistry– Reduced Inventories of hazardous intermediates– Scale-Up of 2X to 4X (on pipe diameter) vs 50 to
1000X– All of the Subtrate/Reagents experience same
reaction conditions• High Intensity Mixing typically• Efficient heat exchange• Steady state
OxidationsNitrationsHalogenationsMetallationsHigh Temp Chemistry…
Benefits in Production•Reduced solvent usage•Reduced cycle times•Reduced capital cost for new equipment / expansions•Better lot-to-lot consistency
0
10
20
30
40
50
60
70
80al
kyla
tions
salt/
free
base
redu
ctio
ns
acyl
atio
ns
boc/
debo
c
hydr
olys
is
este
rific
atio
n
ethe
r for
mat
ion
pept
ide
form
atio
n
cond
ensa
tions
sulfo
natio
ns
disp
lace
men
t
deal
kyla
tion
oxid
atio
ns
addi
tions
debe
nzyl
atio
n
Grig
nard
form
atio
n
cros
scou
plin
g
rear
rang
emen
t
epox
idat
ions
met
alla
tion
nitra
tions
halo
gena
tion
deca
rbox
ylat
ion
Frie
delC
rafts
cyan
ohyd
rin
diaz
otiz
atio
n
Reloads-Out
Reloads-IN
Unique-Out
Unique-IN
Enabling Hot Chemistry via Flow
Enable Potentially Hazardous Chemistry via Flow
• Oxidations
• Nitrations
• Diazotizations
• Halogenations
• Reactions Requiring High Containment due to toxicity concerns
Estimated 10% Current Portfolio
T
T
T
T
T
Thermocouple
Solution ofProduct
to Quench/Work-up
PAT
• Plug and Play Reactor Components• Heat Transfer is 3X higher than open Tube• Pulseless syringe pumps for lab and Kilo-lab
Real-Time Data Acquisition• Flow Rates • FTIR• pH• Temperatures• FB-control capable
Equipment Configuration
8
10
12
14
16
18
20
22
0 0.5 1 1.5 2 2.5 3 3.5 4
Tem
pera
ture
, o C
Reactor (Liquid) Volume , ml
Model Predictions
Tjacket
TreactorInitial Conditions
Actual Outlet temperatures
mCp ∆ T=UoA ∆Tlm
4.45 g/sec@ 21.5 oC
1.2 g/sec@ 9 oC
Tjacket
Treactor
U=1030 W/m2K
Modeling Temperature ProfilesHeat Exchange of Hot and Cold Water
15
20
25
30
35
40
45
50
0 0.5 1 1.5 2 2.5 3 3.5 4
Tem
pera
ture
s, o C
Reactor Volume, ml
PhosphoricFeed
4.45 g/sec@ 16.6 oC
1.15 g/sec@ 24.3 oC
Tj18% NaOH
1.11 g/sec@ 20 oC
15% H3PO4
Calculated Temperature Profiles
within ±1 oC
Reactor Temp
Jacket Temp
ActualTemperatures
∆ H=42.7 kJ/mol NaOH
Modeling Temperature ProfilesExothermic Fast Neutralization Reaction
Oxidation with Peracetic Acid
108 g AcOAc 128 g H2O2
substituted pyridine/ 2 vols EtOAC
30 min add40 min add
Dual Addition
-20
0
20
40
60
80
-0.02
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
-20 0 20 40 60 80 100 120 140
Hea
t Flo
w, W
atts
and
Tr,
o C
mass of dose
minutes
H2O2
acetic anhydride
Heat Flow
Tr
aq 30-50% H2O2 NO
N
substituted pyridine
AcOAc
substituted N-oxide
R2
R3
R1 R1 R2
R3O
O O+
OH
O+
Oxidation with Peracetic Acid
-20
0
20
40
60
80
-0.02
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
-20 0 20 40 60 80 100 120 140
Hea
t Flo
w, W
atts
and
Tr,
o C
mass of dose
minutes
H2O2
acetic anhydride
Heat Flow
Tr
Gut – Bill – Jorgensen – VanAlsten
O
O OH2O
OH
O2k1
OH
Ok2H2O2+
+
O
O
OHH2O+
N
k3
O
O
OH NO
OH
O+ +
R1 R2
R3
R1 R2
R3
Continuous Flow Oxidation Reactor Set-up
Pyridine Model Oxidation
• Issue of cost of substituted pyridine• Pyridine used to develop the flow platform
20
25
30
35
40
45
0 5 10 15 20
Batch run of N-oxide formation with pyridine
Tem
p [C
]
Time [min]
Temperature Profile for Batch ReactorCRC-90 Experiment (all at once addition)
N NO
OH
O+
O
O O+
H2O2
2
Flow Oxidation of substituted PyridineTemperature Profiles
20
30
40
50
60
70
80
90
100
0 20 40 60 80 100 120
Tem
p [C
]
Time [min]
Internal Tr
Jacket Tj
∆ 20 oC
Results of Oxidation Flow Experiments
27%27%7.3%93%87%2 Ethyl Pyridine
97-98%82-85%Pyridine
88%64-71%16-17%92-94%91%Substituted Pyridine
ConversionIn
Receiver
ConversionAcylation
Res.time Exit
ConversionAcylationReactor 2
Exit
ConversionOxidationRes. Time
Exit
ConversionOxidationReactor 1
Exit
DOE (Pyridine Oxidation) for Extent of Conversion
• Stoichiometry Factors:– Acetic anhydride– Hydrogen Peroxide– Water
• Current conditions:– 1.5 eq hydrogen peroxide– 1.4 eq Acetic anhydride– 2.84 eq water (50% hydrogen peroxide sol)
DOE Reactor configuration
DOE Results:Steady State
Continuous Flow Conditions
Range of Robustness
High Energy Flow Chemistry: Process Safety and Extraction of Rate Parameters
Adiabatic Calorimetry• Low Phi Factor
• All at once - Batch Mode• Temperature vs time• Pressure vs time
Hettenbach-Bill
Acetic Anhydride in MeOH
0
50
100
150
200
-20 0 20 40 60 80 100
MeasuredCalculated
Tem
pera
ture
(o C)
Time (min)
O
OOO
O
OH
O
O+
Acetic Anhydride (AA) Methanol Methyl Acetate Acetic Acid
+k
Model: r =-k[AA]1[MeOH]0
= -67.6 kJ/mol (Lit.:65-70)= 1.07
Arrhenius Parameters:E = 72.6 kJ/mol (Lit.:67-75)A=5.1010/min
= 153 oC
rH∆φ
Initial Fill:MeOH = 30.18 gAA = 49.77 g
Temperature vs. Time
adT∆
H
Hettenbach-Bill
Di-t-Butyl Peroxide in Toluene
120
140
160
180
200
220
240
260
-20 0 20 40 60 80 100 120 140
MeasuredCalculated
Tem
pera
ture
(C)
Time (min)
O O O C2H6+2
Di-t-Butyl Peroxide Acetone Ethane(DTBP)
k
Model: r =-k[DTBP]1
= -191.3 kJ/mol (Lit:240-260)= 1.07
Arrhenius Parameters:E = 156.0 kJ/mol (Lit.:155-165)A=3.57x1017/min
= 113 oC
rH∆
Initial Fill:DTBP = 12.28 gToluene = 49.25 g
φ
Temperature vs. Time
adT∆
Hettenbach-Bill
Acetic Anhydride Hydrolysis
Model: r =-k[AA]1[H2O]0
= -54.9 kJ/mol (Lit.:58-62)= 1.06
Arrhenius Parameters:E = 65.3 kJ/mol (Lit.: 40-60)A=8.6x109/min
= 11 oC
rH∆
O
OO O
OH+ H2O 2
Acetic Anhydride (AA) Acetic Acid (P)Water
k
φ
Initial Fill:H2O = 50.0 gAA = 4.88 g
Temperature vs. Time
18
20
22
24
26
28
30
32
-10 0 10 20 30 40 50 60 70
Tem
pera
ture
(o C)
Time (min)
φ
φ
φ = 1.06
= 2
= 5adT∆
φ adT∆
Hettenbach-Bill
“Quality by Design”
“Designed Process Understanding (DPU)”yi=f(xi) : process mapping
• Goal of (DPU) is for improved process understanding at time of validation and will position processes for continuous improvement post launch. In addition, DPU is used for critical-process-parameter justification and to support regulatory filings.
• Quantify measured outputs as functions of input conditions…ie Process Modeling (DOE, Kinetics, etc)
ln BasesubstrateECFkdt
substrated0]][[][][ =−
Process Modeling Approaches• Traditional Approach: One Factor at a time (OFAT)
– No variable interaction– Fragmented information. No process mapping
• Process Modeling Approach :Combining a number methodologies/techniques, to maximize the information out of a limited number of experiments
• DOE/Empirical Approach (linear local approximations)– Data mining of historical information– Screening Experiments
• Efficient identification of important variables and interactions to focus experiments– Response Surface Experiments
• Quantify Interaction and Curvature• Map the design space
– No rate (impurity accumulation) information
• Physicochemical Models / Engineering Kinetic Approximations: All Batch Chemical Processes are dynamic
– Impurities are growing over time (parallel, or consecutive reactions) - rate information always important in an implicit fashion
– Time to completion affects impurity level not only throughput– Provides the starting point for PGM optimization – Fundamental knowledge with full mapping of the dynamic design space – Enhanced Process
Understanding
• Equipment Modeling / Simulation (Engineering Approximations)– Heat Transfer (exothermic reactions)– Mass Transfer (multiphase systems, heterogeneous reaction, agitation effects) Mustakis
Process Modeling and Parameter Ranges
• If nonlinearities are present, process modeling not “highly predictive”: less information
• Filing/ manufacturing: High interest to avoid amendments
• Larger number of experiments will be required
Wide ranges Narrow Ranges
• Process modeling much more predictive – Non linearities are avoided
• Filing/ manufacturing: Higher risk for amendments and restrictions on manufacturing
Design Space: Experimental design needs to balance
Watson/Mustakis
Statistical DOE• Local linear (algebraic) approximation of the experimental
space(is oblivious to chemistry or fundamental knowledge)
• Some good practices :– Include the right variables – PGM guidance is critical – “reasonable variable ranges”
• Limits no linear effects– Use center point of the design to check for curvature– Incorporate standard conditions – if possible as the center point– Replicate center point to extract experimental error– Always review the data – Be very careful in extrapolating outside experimental space– Always validate/verify
Mustakis
DOE and Hi-Lo (OFAT) approach :4 variable layoutTe
mpe
ratu
re
TFAA Mustakis
Designed Process Understanding Example(Quality Parameters)
Factors:•Ethylchloroformate (equiv)•Base (equiv)•Concentration of substrate in THF (vol)
Responses:• Impurity Homolog (wt%)• Substrate (wt%)• Product (wt%) • Total impurities
• 20 Experiments (central composite)• hplc Samples collected at 6, 12, 24, 36 hrs for each expt
• 80 chromatograms to analyze
NH
HN
O
OF3C N
HN
O
OF3C
Cl
O
O O O+
+ HCl
BaseTHF
k1
Time to Conversion(time to 0.5% starting material – interpolation)
23hr
Increasing Base
Standard Conditions
Mustakis
Base Base Base
Base
DOE provides Local Approximations“Kinetic” Information is Lost
One Model Per Sample PointModels are not connected
Mustakis
DOE Does not Utilize Rate Information
0 20 400
20
40
60
80
100
Time(hr)
% m
ol1
0 20 400
20
40
60
80
100
Time(hr)
% m
ol
2
0 20 400
20
40
60
80
100
Time(hr)
% m
ol
3
0 20 400
20
40
60
80
100
Time(hr)
% m
ol
4
0 20 400
20
40
60
80
100
Time(hr)
% m
ol
5
0 20 400
20
40
60
80
100
Time(hr)
% m
ol
6
0 20 400
20
40
60
80
100
Time(hr)
% m
ol7
0 20 400
20
40
60
80
100
Time(hr)
% m
ol
8
0 20 400
20
40
60
80
100
Time(hr)
% m
ol
9
0 20 400
20
40
60
80
100
Time(hr)
% m
ol
10
0 20 400
20
40
60
80
100
Time(hr)
% m
ol
11
0 20 400
20
40
60
80
100
Time(hr)
% m
ol
12
0 20 400
20
40
60
80
100
Time(hr)
% m
ol
13
0 20 400
20
40
60
80
100
Time(hr)
% m
ol
14
0 20 400
20
40
60
80
100
Time(hr)
% m
ol
15
0 20 400
20
40
60
80
100
Time(hr)
% m
ol
16
0 20 400
20
40
60
80
100
Time(hr)
% m
ol
17
0 20 400
20
40
60
80
100
Time(hr)
% m
ol
18
0 20 400
20
40
60
80
100
Time(hr)
% m
ol
19
0 20 400
20
40
60
80
100
Time(hr)
% m
ol
20
Mustakis
Kinetic Model to Apply to all 20 DOE Experiments
Parameter Estimation R2
k1 0.259 0.242 0.276n 1.470 1.379 1.561l -0.130 -0.200 -0.061
Confidence Interval ±95%
Model1, n=1, l = 0 0.8304
Model2 0.9504
k1 0.240 0.219 0.276
ln BasesubstrateECFkdt
substrated0]][[][][ =−
Simple Kinetic SchemeStarting Point 1st order to each reactant
Mustakis
NH
HN
O
OF3C N
HN
O
OF3C
Cl
O
O O O+
+ HCl
BaseTHF
k1
Fitting all 20 Experiments at once
0 20 400
20
40
60
80
100
Time(hr)
% m
ol
1
0 20 400
20
40
60
80
100
Time(hr)%
mol
2
0 20 400
20
40
60
80
100
Time(hr)
% m
ol
3
0 20 400
20
40
60
80
100
Time(hr)
% m
ol
4
0 20 400
20
40
60
80
100
Time(hr)
% m
ol
5
0 20 400
20
40
60
80
100
Time(hr)
% m
ol
6
0 20 400
20
40
60
80
100
Time(hr)
% m
ol
7
0 20 400
20
40
60
80
100
Time(hr)%
mol
8
0 20 400
20
40
60
80
100
Time(hr)
% m
ol
9
0 20 400
20
40
60
80
100
Time(hr)
% m
ol
10
0 20 400
20
40
60
80
100
Time(hr)
% m
ol
11
0 20 400
20
40
60
80
100
Time(hr)
% m
ol
12
0 20 400
20
40
60
80
100
Time(hr)
% m
ol
13
0 20 400
20
40
60
80
100
Time(hr)%
mol
14
0 20 400
20
40
60
80
100
Time(hr)
% m
ol
15
0 20 400
20
40
60
80
100
Time(hr)
% m
ol
16
0 20 400
20
40
60
80
100
Time(hr)
% m
ol
17
0 20 400
20
40
60
80
100
Time(hr)
% m
ol
18
0 20 400
20
40
60
80
100
Time(hr)
% m
ol
19
0 20 400
20
40
60
80
100
Time(hr)%
mol
20
Mustakis
DOE + Kinetics Model Predictions
0.5
0.5
0.5
0.5
1
1
1
1
55
5
1010
10
2020
20
4040
4050
5050
6060
6080
8080
Time (hr)
Eth
yl C
hlor
ofor
mat
e (e
q)
0 10 20 30 402
2.5
3
3.5
4
4.5
5
5.5
6
6.5
7
0
10
20
30
40
50
60
70
80
Mustakis
0.5
0.5
0.5
1
1
1
5
5
5
5
10
10
10
20
20
20
40
40
40
50
5050
6060
6080
8080
Time (hr)
THF
(lt/K
gr)
0 10 20 30 40
4
6
8
10
12
14
0
10
20
30
40
50
60
70
80
Effect of ECF Effect of Dilution
KineticModels
DOERegressionB
ase
DOE-Kinetic “Designed Process Understanding” Summary
• Kinetic modeling provides more direct and accurate description of the time dependence functions for those responses that can be accurately modeled– Ie conversion, starting material, product, major side products
etc– A single kinetic model replaced 4 separate polynomial
regression equations at 6, 12, 24, 36 hr• All 20 phosphate experiments were simultaneously fit
to second order kinetics via multivariate non-linear regression analysis MATLAB– Significantly reduced the number of parameters required to
describe the responses• DOE Response surface analysis was still required for
low level impurities (say only measurable at 36 hr or extended time points) or when the complexity is too high to model Kinetically
Kinetic ToolsCurrently Used Tools
• ALR’s , RC-1, Multimax• Calorimetry• HPLC• FTIR• UV-Vis• Raman• ConcIRt algorithms
Leveraging Multiple in-situ Analytics(The Current Challenges)
• Operational Integration (of multiple probes)
• Software Integration• Closed Software-exporting ReactIR spectra• ConcIRt beyond midIR
– Heat Flow + ConcIRt– Multiple Spectral Set e.g UV + MidIR+ Raman
• Analytical Specificity• Facilities (space, Liq N2 for IR, chiller issues)• Material Intensive
Next Generation tools need to give us more and evenbetter information per experiment….while consumingEven less material
Typical InSitu Analytics:Specificity, Sensitivity, and Cost
10
100
1000
Cos
t ($K
)Technology
Spe
cific
ity
SensitivityLow Med High
Low
M
ed
Hig
h
FTIR
UV, NIRRaman NIR
UV
FTIRRaman
NMR
NMR
We need higher specificity and better structural elucidation in real timefor routine reaction characterization…NMR appears to be well suited
Functional groups
proton, carbon, etc
Integration of Chemical Reactors and Real-Time NMR
Feed
Heating/Cooling
SampleLoop
am Ende, Marquez, Mustakis
hplc pump
Numerous References :Maiwald et al, J. Mag. Resonance, 166 (2004) 135-146Hasse, Albert, et al, Chem. Eng, and Processing, 44 (2005) 653-660Horvath et al, Chem. Rev., 1991 (91) 1339-1351…
Initial NMR Kinetic ExperimentsNMR Tube Kinetics
Inject Reagent, Shake NMR Tube, & Insert in NMR
am Ende, Marquez Sept, 2005
0
0.5
1
1.5
2
0 10 20 30 40 50 60 70
Hydrolysis of Acetic Anhydride in D2O @ 25 oC
Composition vs time via NMR
Con
cent
ratio
n [M
]
Minutes
AcOD
AcOAc
1st order kinetics: Ca=Caoexp(-k*t)
measured kD=0.94 * 10-3 (sec-1)
lit. kD=0.89 * 10-3 (sec-1)
reported isotope effect =koH/koD=2.9Batts & Gold, J. Chem. Soc. A, (6), 984, 1969.
Flow NMR KineticsReactor Integrated to Flow NMR and run SemiBatch mode
am EndeMarquezMustakis
April 7, 2006
0
0.5
1
1.5
2
0 10 20 30 40 50 60
Hydrolysis of Acetic Anhydride in D2O @ 25 oC
Composition vs time via NMR flow cell
Con
cent
ratio
n [M
]
Minutes
AcOD
AcOAc
1st order kinetics: Ca=Caoexp(-k*t)
measured kD=0.84 * 10-3 (sec-1)
20 ml AcOAc dosed over 20 minutesinto 200 ml D
2O
NMR flow = 3 ml/min, Gain=40
103 * k (sec-1)• NMR tube 0.94• Lit 0.89• NMR flow 0.84
Spectra collected during the
20 min dose
NMR monitoring of a Semibatch Reaction:Alkylation of a Di-amine with Glyoxal
Feed
Heating/Cooling
ppm
Intensity
minutes
SampleLoop
NH2NH2
N
COCF3
O
O
HH
NCOCF3
NNOH
+
am Ende, Marquez, Mustakis
D2O/IPO-d8
NH2NH2
N
COCF3
O
O
HH
NCOCF3
NN
H2O
OH
+ + 2 0 20 40 600
2
4
x 104
0 20 40 600
2000
4000
0
1
2
0 20 40 600
1
2x 104
StartingMaterial
Product
0 10 20 30 40 50 600
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
NH2NH2
N
COCF3
O
O
HH
NCOCF3
NNOH
+
H2N N
NCOCF3
HN NH
NCOCF3
HO HO
H H
4 4.2 4.4 4.6 4.8 50
20
40
60
-100
0
100
200
300
400
500
5 5.2 5.4 5.6 5.8 60
20
40
60-1000
0
1000
2000
3000
7 7.2 7.4 7.6 7.8 8 8.2 8.4 8.6 8.8 90
20
40
60
-200
0
200
400
600
800
1000
1200
1400
1600
38Numerica Technology
Kinetic mechanism superstructure
The superstructure may be created by
merging several smaller mechanismsintroducing hypothesized reactions
A family of mechanism is created from the superstructure by removing various subsets of species and reactions
The figure on the right illustrates a superstructure
The boxes denote chemical species The arrows denote reactions
Automatic mechanism selection
The global dynamic optimization capabilities of JACOBIAN can be used to select simultaneously a mechanism from a superstructure while fitting the rate constants against experimental data
Merging Platform Reactor Technology with “Lab” NMR
FTIR
Reactor
Calorimetry
UV Flow Cell
Recycle Solution Phase
Powerful Tool for Elucidation:•Reaction Mechanism•Structure•Kinetics
1H, 13C, 19F, 31P
Summary• Continuous Flow
– Process intensification opportunities– Portfolio Enabling– PlugFlow reactors are fed stoichiometrically
• equivalent to all at once dosing– Highly exothermic reactions pose challenges for
acquisition of isothermal kinetics• Adiabatic calorimetry one option for extracting rate information
– Engineering Technology working to:• Developing plug flow platform for gmp installation – hot
reactions• Develop “small-scale” continuous slurry reaction,
crystallization, isolation platforms as well.• Develop holistic continuous flow processes on selected existing
processes in manufacturing – raw materials to API
Summary-Continued• Kinetic Understanding is Key for
– Scale-Up – time scale analysis for Mixing Effects• Time scale of impurity kinetics vs mixing times
– Process Understanding (Quality by Design) Mapping• Leverage Kinetic models when possible vs DOE models
– Continuous Flow Reactor Design• Rate Equations to model conversion vs length or res. time
• Next generation platform tools needed to help elucidate pathways even faster– Hardware needed to:
• Allow multiple analytic (UV, FTIR, RAMAN, etc) sensors on 10-15 ml scale
• Provide heat flow directly• Make online Quantitative NMR routine and easy
– Software needed to:• Simultaneously analyze multiple analytic data sets - FAST
– UV, FTIR, RAMAN, NMR, Tr, Qr– With constraints – Ao, Bo, dose time, Tr, etc
• Rapidly assess kinetic models• Global optimization of parameter estimation and model discrimination
for complex kinetics
Acknowledgments
• Kevin Hettenbach• Matt Jorgensen• Brian Marquez• Jason Mustakis• Geraldine Taber• Tim Watson• Eric Dias (Symyx)
Mettler-Toledo
Back-Up Lithiation Slides
Kinetic CharacterizationMetallation Example
A Simple Reaction
• Lithiation Reaction performed in RC-1 calorimeter at -50ºC.– During nBuLi addition, reaction went from colorless-orange-
green-black.– Large volume of solids produced, forming a “cap” at top of
solution.– Subsequent runs at -65ºC produced similar results, although
color change was slower.
• Comparison with bench-top experiments– Reaction always proceeded as expected in RBF
experiments from 0.1 – 22L scale.
Br Br n-BuLi
Br LiTHF
-78 to -40oC
BrR1
R2
OHR1 R2
O
Problem….why are we having problems running in the RC-1
Br Br n-BuLi
Br LiTHF
-78 to -40oC
Br Br
THF
Br Li
775 and 665 cm-1
Br Br
17.4 g of 2.5 M n-BuLi in Hexanes
(0.062 mols dosed in 2 minutes)
14.7 g C6H4Br2 = 0.062 molsIn 400 ml THF
-65 oC
Preparation ofLithium Bromobenzene
Under diluteconditions
-20
0
20
40
60
80
100
120
-120
-110
-100
-90
-80
-70
-60
-50
-200 -100 0 100 200 300 400
Hea
t Flo
w, W
atts
Mas
s n-
BuLi
/Hex
anes
add
ed, g R
eactor Temperature, oC
Seconds
Tr
Preparation of Lithium BromobenzeneReaction Calorimetry under Dilute Conditions
teflon tubeinserted
sub surfacenear reactor
wall
Dosing Started
Heat Flow, W = Q=Qflow + Qaccum + Qdos
∆ Hrxn = -112 kJ/mol = -26.8 kcal/mol
Br Br n-BuLi
Br LiTHF
-78 to -40oC
Qdos
0.15 M
Kinetic Pathways
Br BrLi + Fast
k0Br Li Br+
Br Br Br Li+ Br Br
Li
Br Br
Li
+ Br
+Br LiBr
k1
k2
Br Br + Br Br
Li
+Br
Li
BrBr
Br
k4
Br LiBr +fast
Br
Li
Br
k3
k5
Calorimetry Results
0
5
10
15
20
25
30
35
0 1 2 3 4 5 6
-45 C-50 C-55 C-65 C
Hea
t Flo
w (W
)
T im e (hr)
∆ Hrxn = -289 kJ/mol = -69 kcal/mol
Br Br + Br Li BrBr
+ LiBr
Br Br
14.8 g (0.062 mols) of dibromobenzene+ 25 ml THF
(dosed in 5 minutes)
0.062 mols of LiBrbenzeneIn 400 ml THF
-65, -55, -50, -45 oC
Reaction of Lithium Bromobenzene
and 1,4 Dibromobenzene
ReactIR FTIR w/ Dicomp:anion solution taken as background
BrBr
Br Br
Li
Br Li
Br
740 cm-1
Br
Kinetic Modeling of the Undesired Reactions
0
2
4
6
8
10
12
0 25 50 75 100 125 150
Hea
t Flo
w (W
)
Minutes
RC-1 Data
Kinetic Model
Br Br
Br Li
Br Br
Li
BrBr
Li
0
0.01
0.02
0.03
0.04
0.05
0.06
0 25 50 75 100 125 150
Mol
s
Minutes
Br Br Br Li+ Br Br
Li
Br Br
Li
+ Br
+ LiBrBr Li+ Br
Li
Br
slow
slow
k1
k2
-55 oC
14.8 g (0.062 mols) of dibromobenzene+ 25 ml THF
0.062 mols of LiBrbenzeneIn 400 ml THF
Experimental MeODQuenched samples
0 5 10 15 20
minutes
0.0
0.1
0.2
0.3
0.4
0.5
@ t=0 Anion =0.5 MolarDibromobenzene=0.5 Molar
-75-70
-65
-60
-55
Effect of Temperature on Rate of the Undesired Reaction
Conditions Tr (°C) Yield (%)
500 rpm -65 44
Pre-chill nBuLi -65 68
Tj = -70°C -60 72
10X dilution -65 87
Results from RC-1 Lithiation/Quench
Rate of Side Reaction isSignificant
At these Temperatures
So Long add timesResults in more side
Reactions
Ani
on (m
ols)
25 min30 min
Add time
• Hot spots and competing kinetics significantly impacted yield in RC-1 experiments
• Undesired reaction highly temperature dependent• Take Home Messages:
– Understand the kinetics of competing side reactions– Run more dilute 10X – will reduce Tad from 89 to 9 oC– Run colder –75 oC– Minimize time for Anion to “see” dibromobenzene such as in a
flow system (proposal)• Use a pre-prechilled feeds in jacketed static mixer
Conclusions of Lithiation Study
n-Butyl Lithium in Hexanes
Bromobenzene/THF
Ketone/THFcoolant
Conclusions (Cont.)• Understanding the kinetics of the Undesired Reaction
was Key to Understanding this process and how to improve it.
• Heat of reactions were measured for:– anion formation (lithium bromobenzene) = –112 kJ/mole– Biphenyl formation via exothermic benzyne route = –280
kJ/mol of anion• Activation energies were estimated = 12-13 kcal/mol
• Need to characterize undesired reaction pathways (Kinetics) to fully understand the process