Development and Optimization
the investigative approach
PhD
IN INDUSTRIAL CHEMISTRY AND CHEMICAL ENGINEERING (CII)
Aims of Chemical Development
• Produce a cheap product (short route, high yields)
• Simplify process
• Discover robust and safe process
• Demonstrate process works well on plant
• Use available plant efficiently But Also
•
Minimize effluent•
Understand the chemistry and the mechanisms•
Use analytical expertise to quantify data•
Understand available plant and equipment•
Collaborate with other disciplinesSome Difficult Decisions
• CHOICE of synthetic route
• WHEN to scale up
• WHETHER to scale up more than one route
• WHICH route to use to make initial supplies
• WHEN to change route
It is often better to “quickly” scale up one route whilst looking for a better one
BUT
There is a danger of getting “locked in”
First 10-20 Kg are the most difficult
• If the route is NOT likely to be used in manufacturing - Carry out minimal optimization
- Improve work-ups and isolations - Ensure safety for scale-up
- Go into plant ASAP
- Make kilogram supplies
• If the route IS likely to be used in manufacturing - Take a longer term approach
- Aim to understand the process
Cost and Safety Considerations
Costs:
• If raw materials are EXPENSIVE
- Concentrate effort on yield maximization
• IF raw materials are CHEAP
- Concentrate on improving process efficiency (work-up and isolation)
• Before starting development - Review synthetic route
Safety:
• What must be changed to make the process SAFE to scale up?
- Reagents (phosgene, diborane, acetylene, diazoalkanes, LDA?) - Exotherms
Process Considerations
• Is the order of steps most appropriate for the route?
- Linear vs convergent syntheses - Selectivity (example)
- Last step - does it involve heavy metals?
Can they contaminate the final product?
Synthesis of McN-5691: Palladium-catalyzed coupling reaction
OCH3
N H3C
CH3
O I H3C
OCH3
Pd(0), CuI Solvent
+ H C C
OCH3
N H3C
CH3
C O
H3C
OCH3
C
Attempts to Reduce Palladium in McN-5691
Work-up Procedure ppm Pd
1 Extraction, filtration, treatment with borohydride in methanol
2 Hydrogenation in ethanol in presence of carbon
3 Treatment of methylene chloride solution of product with
borohydride on alumina or silica 4 Chelation with dimethylglyoxime 5 Treatment of methylene chloride
solution of product with Amborane
100-200
930-950
770-870 259
H3CO
N CH3 H3C
C O
H3C
OCH3
C Pd
+
Revised Synthesis of McN-5691
O
I
O
+ H C C Ph Pd(0), CuI O
C
O
C Ph
NaCNBH3
OCH3 OCH3 H2N
H3CO
N H H3C
C O
H3C
OCH3
C
H3CO
N CH3 H3C
C O
H3C
OCH3
C NaBH4
CH2O Et2NH
Can Steps be Easily Combined?
• Advantages
- Yields usually higher (product loss on work-up minimized)
- Eliminates isolation, purification, drying, analysis of one or more intermediates
- Usually produces cheaper product
• Disadvantages
- Impurities may carry through - Often lower product quality
- Optimum solvent for 1st step may not be same as for 2nd step - Process investigation of problems on plant or “failed“ batches
more difficult
Cyclization
Dehydrogenation
One-Step Dehydrogenation
C.G.M. van de Moesdijk, + 3 H2
N
H CH3+ 50 kcal/mol
α-Picoline gas phase
N CH3
H2C H2
C CH2 + H2O
- 10 kcal/mol
A New Synthesis of α-Picoline
H2C C
H2 C CH2
O C CH3 N
+ 3 H2
NH CH3
+ H2O - 60 kcal/mol
α-Pipecoline Ni, 120 °C
5 M Pa
liquid phase
Combination of Steps
•
Always worth trying•
Optimize reaction steps independently•
Understand both processes fully - then combine•
Try to eliminate work-up in first reaction• Old process
• New process:
Use formic acid as reducing and formilating agent
HCO2H
HN N O
NH2 O
C3H7
HN N O
NH2 O
C3H7 NH2 1) NaNO2/HCl
2) Na2S2O4
H.J. Federsel; Astra
HN O
NHCHO
HN
O H HN N
O
NaNO2, HCO2H/Pt-C
General considerations
• These questions
- may DELAY scale-up
- are unlikely to result in the process being abandoned
• Almost ANY process can be scaled up if - engineered
- safety issues are addressed - process control is excellent - process limits are understood
eg boron tribromide, thallium reagents, alkyl lithiums
Process Optimization
The most important decision is
WHICH STAGE TO EXAMINE
• Costing will help
• Need to replace reagent or raw material?
• Step unlikely to be suitable for plant?
• Elimination of chromatography
Optimization - Yield Improvement
• Correct choice of reaction conditions
• Attention to detail
• Understanding the chemistry
- Mechanism - Byproducts
- Competing processes
• Good analytical methods
- Weight-based assays
- Standards
The Investigative Approach
• Obtain weight balance (prep HPLC?)
- Accurately determine isolated yield
• Assess yield after reaction but before work-up -
Is yield lost because reaction is poor?- Is product lost during work-up?
• Follow reaction quantitatively - e.g. using hplc?
- Does yield peak at any stage?
• Was starting material pure?
- Check manufacturers’ figures
- Does recryst. of starting material improve yield?
Has the reaction gone to completion?
• Is there starting material left?
- increase time
- increase amount of reagent
- (may be reaction or complexing with product)
• Has starting material reacted
- with more than one mole of reagent?
- with solvent or adventitious water?
• Is the order of addition sensible for the process?
- Is it appropriate for the plant?
- (e.g. alkylation of primary amines with alkyl halides)
• Are the reagents of known purity?
Was product formed but reacted further?
• Examine effect of extended reaction time - Exaggerate the effect
- Isolate impurities
• If secondary reaction is a problem - Use deficiency of reagent
- Stop reaction at earlier stage
- Investigate effect of temperature
ArCHO + NaCN I
OPr OMe Ar =
DMF ArCHO ArCO2H
14 % 6 %
OH Ar O
Ar 60 %
OCOAr Ar
O
Ar 14 %
CO2But
OCOAr Ar
O
Ar 7 % ArCO2H
Ar 9 %
O CO2But 51 %
A.S. Thompson et al. J. Org. Chem. 1992, 7044-52
Synthesis of PAF Antagonist MK-287
Byproducts from Solvents
• Acetone in isopropanol
• Aldol reactions with aldehydes
• Dimethylamine in DMF
• Reacts with active molecules
• Methanol in ethanol
• Transesterification
• Dichloromethane
• Far more reactive than is appreciated
• e.g. with amines, sodium azide, ..
• Polyaromatic in benzene/toluene
Competing Processes
• Isolate and characterise ALL by-products
- Recrystallise product, then evaporate liquors
- Chromatography (prep. TLC, column, prep. HPLC) - Structure gives clues to mechanism
• Synthesise impurities in large amounts may be useful
- for analysts
- for taking through synthesis
• Examine effect of changing reaction conditions on impurity
level
Competing Processes
t-BuOH conc. H2SO4
urea
+ N N
Cl HO N N
Cl HO
t-Bu
t-Bu
N N Cl HO
t-Bu
N N Cl HO
t-Bu By-product formed on scale-up caused by extended reaction time on plant
Byproducts in Reactions -
Baylis-Hillman - Revised Process
CH
2O, H
2O CO
2Bu
tquinuclidinol 10 min. 80 °C
CO
2Bu
tCH
2OH
Byproducts
ButO2C
O CO2But
ButO2C
O CO2But nO
ButO2C
O nOBut
(CH2O)n CO2But
dipolar solvent
CO2But CH2OH quinuclidine
ButO2C
O CO2But nO
in the presence of water
byproducst
Final Process:
Add CH2O, CH2=CHCO2But, to quinuclidine, water and cosolvent.
heat, then add toluene, cool, separate
Baylis-Hillman - Revised Process
Understand Reaction Mechanism
Process for Manufacturing of Morpholine
42% Ni on Al OH
180-220°C DEGA
HO O
+ NH3 / H2O
Ni
NH2 HO O
NH O
+ H2O
C.A. Cooper, Chemtech, 1991, 378.
D.D. Dixon, U.S. Patent 4,645,834, 1987
Possible Byproducts
HO O
N H HO O
N
O
O OH
H
2N O N
O
N O N
O O
Other byproducts arise from oxidation
of alcohols
RCH
2OH RCHO + H Ni
2(slow step: control kinetics)
RCHO + NH
2R RCH=NHR’ + H
2O
or RCH-NHR’
Reduction or
OH
hydrogenolysis
RCH=NHR’ RCH H
2 2NHR’ + H
2O
Note: Amines also undergo nickel catalysed dehydrogenation
Mechanism of Ni catalysed amination
Selective Preparation of o-Hydroxybenzaldehydes
CHO CH2O
OH
Mg(OMe)2
OH CH2O
CH2OH OH
< 50 ppm p-hydroxy!!
Used for the manufacture of:
CH=NOH OH
Byproduct is: OH
R
OH
R
Work-Up
• Many low-yielding processes result from poor isolation technique
- examine the aqueous stream
• Study chemical properties of product
- Especially solubility
• Exploit differences between product and by-products
- Solubilities
- Hydrophilic/hydrophobic nature - pKa
- Molecular weight
• Material balance
• DESIGN the work-up for each process
Work-Up & Optimisation
• During optimisation studies work-up is also a variable
• Work-up may vary with solution yield
• Better NOT to work-up
- Saves time
- Relies on good analytical procedures
• Design work-up later
Remember: Murphy’s Law 2
The more innocuous a process change appears, the further
its influence will extend
Streamlining the Process/Optimization
Further development to improve
• Cost
• Ease of scale-up
• Work-up
• Process simplification
• Yield
• Change of Process Variables
• Optimum conditions will rarely be reached by one-step-at-a-time (OVAT) variations
• Optimum for one parameter (e.g. stoichiometry) will probably change as another parameter (e.g. concentration) is varied
• Changes in solvent are CERTAIN to change other parameters
OVAT Variations
20 40 60 80 100
Conc. (g/g)
20 40 60 80 100 Temperature °C
40 60 Path 1 80
Path 2
50 70 90
Change of Reagent
• To improve yield
• To improve selectivity
• To reduce cost
• To control effluent
• Example
- Investigative development in the Cefoxitin process
L. Weinstock, Chem. & Ind., 1986, 86
N S
TsCl
MeOCH2Cl CH2OCONH2
CO2H O
N OMe O
NH2 CO2H
H
N S
CH2OCONH2 CO2-
O
N OMe O
NHTs CO2-
H
N S
CH2OCONH2 CO2CH2OMe
O
N OMe O
NH2
CO2CH2OMe H
N S
CH2OCONH2 CO2H
O
N OMe O
H
S
Cephamycin C
Cefoxitin
i) TMS Me-carbamate or 4A molecular sieve ii) H+
S CH2COCl
Cefoxitin Process
R. Tyson, US Patent 2,168.699, 1988; Chem. & Ind., 1988, 119
O N O N
H2N CH
H
S CH3 CH3
O
O OCO2Et H3C
Ph
• Effective orally -well-adsorbed
• Ethoxycarbonyl group metabolised to ethanol and carbon dioxide
• 1 -Chlorodiethyl carbonate available in bulk BUT
Astra Bacampicillin Process
Astra Bacampicillin Process
SOLUTION: Use more reactive reagent - bromodiethyl carbonate PROBLEM: Not commercially available
SOLUTION: Contract out synthesis
Initially
Subsequently
Process scaled up by Palmer Research 35 t/a
H3C H C O Cl
CO2Et H3C H C O Br
CO2Et
H3C H C O Br
CO2Et hν
Br2 (EtO2)2CO
Rate & Order of Addition of Reagent or Catalyst
• Order of addition MAY be changed to facilitate scale-up
• Rate of addition WILL change on scale-up -the effect should be studied
- Exotherms - Yield variation
- By-product formation
• Therefore RECORD rate of addition in ALL lab and pilot
experiments
Importance of Controlling Addition Rate
Org Syn Process -
add all reagents and heat to 80°C 30 - 35% yield on scale-up*Better for scale-up -
malonic acid O
CO2H piperidine
pyridine
CO2H CO2H CO2H
CO2H
OH
Stoichiometry
• Often requires very careful control - e.g. Cefoxitin
• Changes during reagent addition
• Not always as expected from mechanistic reasoning - Friedel-Crafts with aluminium trichloride
- Alkylations with sodium hydride
• Require accurate assay methods for reagents and starting materials
Stoichiometry Differences:
• Overall
• IN solution at any one time
• Affected by rate of addition AND rate of reaction (and vice versa)
Synthesis of Benzazepines
solvent CH2Cl
X
Mg CH2MgCl
X
H2 C
X
CH2
O NR
X
X
NR HO
X
NR
Use of Organometallic Reagents
To get good yields in Grignard reaction
•
Grignard must react as soon as formed•
Cannot make 1 mole of Grignard first - it couples to bibenzyl•
Separately but simultaneously add benzyl chloride and oxazolidine to magnesium - good yields•
Yield depends on ability to control rates•
If oxazolidine added to quickly - reaction killed•
Cannot mix benzyl chloride and oxazolidine - slow quaterniz.Scale-Up of Butyl Lithium Chemistry
•
On scale-up, prefer to charge butyl lithium first•
Affects anion vs. dianion formation•
Work-up gives evolution of butaneA
50% hydrogen peroxidesodium tungstate cat.
B
•
Process- Aqueous solution of sodium tungstate pre-treated with 50%
hydrogen peroxide (3 moles/mole A)
- Added to aqueous solution of A over 90 minutes
•
Yields- In development lab > 90%
- In safety lab (automated dosing) 9%
•
Experimental conclusion- Add 10% of solution all at once, followed by dosing of rest of solution - Yield 97-100%
Effect of Rate of Addition
Dropping Funnel Test
0 30 60 90
0 0.2 0.4 0.6 0.8 1.0
Time (min)
W. Fr. Charge Delivered
Linear 90 min. Rate
3.0 3.5
Variation of Temperature
• Changes reaction rate (ca. × 2 every 10°C)
• Alters selectivity
• Exothermic reactions may be best carried out at HIGHER temperature to prevent accumulation
• Must control temperature accurately, especially during exothermic processes
• Study effect of overheating on process
- Decomposition?
- Runaway?
- Low yield?
- Loss of selectivity?
Decarboxylation Reaction
aq. base NH
N
CO2Et O
NH N
CO2H O
NH N
Dowterm A O
or PhOPh 250°C
0 10 20 30 40 50 60
0 50 100 150 200 250 300
Yield (%)
OH OH HO
OH OH HO
COOH
OH HO
(aq)
, t KHCO3
∆
Temperature Control Can Be Critical
•
Equilibrium mixture of cis & trans isomers produced•
In lab, with slow addition of hexane, only trans isomer isolated•
cis Isomer is oil, so if isolated with trans a slow-filtering sticky product obtained•
Process requires hexane addition at 50°C over several hours•
Process worked well on 2000 L scale•
At 10,000 L scale, sticky crystals obtained at 50°C•
Good quality product if hexane added at 55°CDr S. Bone, “Scale-Up of Chemical Processes”, Brighton, Sept 1994 (conference organised by Scientific Update)
Temperature Control Can Be Critical
NaH, HCO2Et R
O
OEt O
solvent, 0°C R O
OEt O
OH 50°C NO PRODUCT
(CO + NaOEt)
Minor Change of Intermediate
• Change in protecting group
• Change in ester
- To increase/decrease rate of reaction - To improve selectivity
• Change in salt form of intermediate
- To improve isolation
• Change in leaving group
- To change rate of reaction - To change selectivity
- To reduce cost
Variation of Pressure
• Usually only varied in gaseous reactions
- Catalytic hydrogenation - Ammonia reactions
- Carbonylation
• Very high pressures (10-20 kbar) will affect solution-phase reactions
- with high negative activation volumes - e.g. cycloaddition reactions
Time
• Time costs money
• Reduce plant occupation by simplifying reactions
• Increase rate by increasing temperature, concentration, pressure
• Kinetic studies assist chemical engineer
Concentration/Volume Efficiency
• Traditional research methods use dilute solutions
• Plant methods require the process to be as concentrated as possible
• Need to study effect of concentration on
- Yield
- Exotherm hazards - Rate of reaction
- By-product formation
• Need to minimise work-up volumes
Optimisation - Quality of Materials
• Use raw materials of comparable quality to those to be used on plant
• Ideally optimise using intermediate/raw material from I large batch of TYPICAL quality
• Periodically check quality of intermediates, raw materials and catalysts
- Do they deteriorate on storage?
- Do they pick up moisture?
Optimisation - Quality of Materials
• Check quality of solvents, particularly water content
• Carry out use-tests on
- raw materials from new suppliers - batches of in-house intermediates - new batches of critical materials
- (the supplier may be developing his process too, and quality may vary within the specification)
• Ask suppliers what impurities are in their raw materials
- Get a material balance
References
1. Optimization of Chemical Processes, 2nd Ed., T.F. Edgar, D.M.
Himmelblau and L.S. Lasdon, McGraw-Hill, Boston, 2001
2. Ming Ge, Qing-Guo Wang, Min-Sen Chiu, Tong-Heng Lee, Chang- Chieh Hang, Kim-Hock Teo, An Effective Technique for Batch
Process Optimization with Application to Crystallization, Chemical Engineering Research and Design, Volume 78, 2000, 99-106.
3. Babu, B.V., Process Plant Simulation, Oxford University Press (2004).
4. Kalyanmoy, D., Optimization for Engineering Design, Prentice Hall (1998).