Automation and the Changing Face of Process R&D in the ...

subduedlockΜηχανική

5 Νοε 2013 (πριν από 3 χρόνια και 9 μήνες)

94 εμφανίσεις

Automation and Parallel
Experimentation in Process
Research

Lori Spangler

Bristol
-
Myers Squibb

Process Research and Development

7

asdf

extending and enhancing human life



Overview / Background




Tools and Approaches




Process Chemistry Examples

Automation and the Changing
Face of Process R&D in the
Pharmaceutical Industry

asdf

extending and enhancing human life

7

Process R&D Goals


Create most efficient synthetic route


Understand each transformation in great
detail


Optimize each synthetic transformation


Demonstrate the chemistry on scale


Transition the technology to manufacturing

The primary product of Process Chemistry is

KNOWLEDGE!

Why Use Automation and
Parallel Experimentation?


Shorten cycle time for process
development


Increase productivity of each scientist


Automated reactions combined with on
-
line analysis leads to better information


Reproducibility: automation leads to
greater control over important variables

Early Era of Process R&D Field

early

development

launch

maturity

Process

Candidate

Effort

/ innovation

“learn
-
while
-
doing”

Current State of Process R&D Field

early

development

launch

maturity

Process

Candidate

“learn
-
before
-
doing”

Effort / innovation

DRUG

DISCOVERY

Early

Development

PRODUCT

LAUNCH





Candidate

Requirements

Process

Characteristic



10 g



1 kg



25 kg



300 kg

1000 kg

expedient

practical

efficient

optimal

Phase I

Phase II Phase III

Late

Development


The Process Research and
Development Challenge

Make material

Create knowledge

The Process Research and
Development Challenge

Automation Goals in Drug Discovery v.
Process R&D

Drug Discovery Chemistry

Goal:

viable drug

candidate

(Many substances,
prepared expediently)

.

(Many possible
chemistry options)

Process Chemistry

Goal:

one substance
,

prepared
well

Experimental Parameter Space

“Optimum”
Process Solution

Technical

Knowledge and

Experience

Knowledge
plus

parallel expts and
Statistical DOE


past

present

Possible Experimental Approaches

Drivers:



safety



cost



throughput



simplicity



environmental



Overview / Background




Tools and Approaches




Process Chemistry Examples

Automation and the Changing
Face of Process R&D in the
Pharmaceutical Industry

asdf

extending and enhancing human life

7

Automation is the Integration

of Modular Technologies

Experiment

Analysis

Design

Informatics

Automation Tools for Process R&D

at Bristol
-
Myers Squibb

Computer

(Design /


Analysis)

Reactor Blocks,

Liquid Handler

(Experiment Execution)

Analytical Instruments

(measured responses)

(data organization)

Network database,

Information management

Hardware: Chemistry Work Station

Robotic arm

syringe

Reactant /

Reagent rack

Reactor block

HPLC quench

vessels

Hardware: Gilson 215 Liquid
-
Handler

Tools: Reagent Libraries


Resolution
-



Chiral Acids


Chiral Amines


Lewis Acid


Inorganic Base


Organic Base


1: Tertiary amines


2: Primary/Secondary amines


PTC Catalysts


Pd Catalyst


Metals


Ligands


Scavengers


Heterogenous Catalysts


Delivered as weighed amounts in bottles or as stock solutions




Will this software let me build
the

application to fit
my business, or will I have to build my business to
fit the application?”



Sean Gallagher

Editor in Chief,
Enterprise Development

Magazine

Software: The Key Question

Software: What do I Need to Run the Gilson?

Worklist

tells the Gilson

what to do

Tray File

tells the Gilson

where your vials are

Software: Microsoft Excel
-
based

Smart Balance Technology



Functionality to:


Acquire weights


Identify chemicals
via bar
-
code system


Prepare stock
solutions


2
0

%

M
e
O
H
8
0

%

H
2
O
2
0

%

M
e
O
H
8
0

%

C
H
3
C
N
2
0

%

M
e
O
H
8
0

%

H
2
O
2
0

%

M
e
O
H
7
5

%

C
H
3
C
N
5

%


H
2
O
P
u
m
p

A
P
u
m
p

B
X
X
S
a
m
p
l
e
0
.
1

%

T
F
A
1
0

m
M

N
H
4
O
A
c
C
o
l
u
m
n

A
C
o
l
u
m
n

C
C
o
l
u
m
n

E
C
o
l
u
m
n

D

2
1
0

2
5
4
C
o
l
u
m
n

B
W
A
S
T
E
Analytics: “MeDuSA” Chromatography

Analytical Tools: HPLC systems



MeDuSA System


4 solvent/11 column


screening system


Simplified HPLC interface


with mass storage and


data reduction capability


High Throughput LC/MS


Automated Walk
-
up Prep LC

Analytics: High
-
throughput HPLC System

Analytics: “MeDuSA” Chromatography

TFA Mobile Phase 254 nm

YMC ODS
-
AQ

YMC Pro C18

Luna C8(2)

Shield RP
-
8

Luna Phenyl Hexyl

“RRT 1.34”

“RRT 1.88”

Analytics: Data Reduction / Analysis




identify impurities



study relative rates



define controls



develop effective


purification mechanisms


Model for Lab Development

Scan reaction

space (1000’s expts)

Best route
based on:



cost



safety



environmental


Develop processing

conditions (100’s expts)

Validate processing

conditions (10’s expts)

Goal:

“Identify

pathways”

“Create


knowledge”

“Verify

assumptions”



Final


“optimization”



determination of


valid parameter


ranges




Overview / Background




Tools and Approaches




Process Chemistry Examples

Automation and the Changing
Face of Process R&D in the
Pharmaceutical Industry

asdf

extending and enhancing human life

7

Statistical Design of Experiments (DOE)

Response variables (e.g. yield, impurity level,


extent of completion, %ee)



…after carrying out experiments involving...


Class variables (e.g. choice of metal, ligand,


solvent, reagent)

…and a well chosen set of...


Continuous variables (e.g. reagent equivalents,


time, temperature,


concentration)





A systematic approach to discover trends and improve
reaction performance through the analysis of
...

Two
-
level Factorial Designs
-

in a Nutshell



Each continuous variable assigned a “high” and a “low” value




Several “center
-
point” runs conducted as validation




If number of variables = x and number of center points = y,


then number of experiments = 2
x

+ y

#
temperature
concentration
solvent
catalyst
1
high
high
MeOH
AcOH
2
low
low
MeOH
AcOH
3
high
low
MeOH
AcOH
4
low
high
MeOH
AcOH
5
high
high
DMF
AcOH
6
low
low
DMF
AcOH
7
high
low
DMF
AcOH
8
low
high
DMF
AcOH
9
high
high
MeOH
TFA
10
low
low
MeOH
TFA
11
high
low
MeOH
TFA
12
low
high
MeOH
TFA
13
high
high
DMF
TFA
14
low
low
DMF
TFA
15
high
low
DMF
TFA
16
low
high
DMF
TFA
Two
-
level Factorial Designs
-

in a Nutshell

Two
-
level Factorial Designs
-

in a Nutshell



What the statistical data tells you:


-

Main effects
-

most important factors

-

Interactions
-

when main effect depends on second factor




Fractional factorial designs:


less than 2
x

runs can still provide main effects and

two
-
way interactions, but at the expense of resolution

Example 1: Use of DOE to Determine Factors That


Improve Reaction Selectivity

O
S
i
(
O
E
t
)
3
N
C
O
O
H
N
C
T
a
m
a
o
-
F
l
e
m
i
n
g
o
x
i
d
a
t
i
o
n
O
O
H
C
8

-

9
%
O
H
2
N
+
Response Factor: Impurity area%

Factor
equiv.
KHCO
3
equiv.
KF
equiv.
Urea H
2
O
2
Volume
(mL)
Temperature
(deg C)
Solvent Ratio:
THF/MeOH
High
6.5
2
6.5
80
35
0.68
Center
5
1.5
5
60
25
0.58
Low
3.5
1
3.5
40
15
0.48
RUN#
Equiv.
KHCO3
Equiv.
KF
Equiv.
urea-H2O2
Volume
Solvent Ratio:
THF/MeOH
Temp
(deg C)
Impurity
amount
2
5.0
1.5
5.0
60
0.58
25
8.65
8
5.0
1.5
5.0
60
0.58
25
8.57
11
5.0
1.5
5.0
60
0.58
25
8.24
18
5.0
1.5
5.0
60
0.58
25
8.17
5
6.5
1.0
3.5
80
0.68
15
2.86
16
3.5
2.0
6.5
80
0.68
15
3.29
13
6.5
1.0
6.5
40
0.68
15
3.93
7
3.5
2.0
3.5
40
0.68
15
4.07
15
3.5
1.0
6.5
80
0.48
15
5.72
17
3.5
1.0
3.5
40
0.48
15
7.35
6
6.5
2.0
6.5
40
0.48
15
8.35
9
6.5
2.0
3.5
80
0.48
15
9.87
3
3.5
1.0
3.5
80
0.68
35
2.94
14
6.5
2.0
6.5
80
0.68
35
3.18
1
6.5
2.0
3.5
40
0.68
35
3.52
10
6.5
1.0
6.5
80
0.48
35
4.75
4
3.5
1.0
6.5
40
0.68
35
5.22
20
3.5
2.0
3.5
80
0.48
35
6.17
19
6.5
1.0
3.5
40
0.48
35
6.54
12
3.5
2.0
6.5
40
0.48
35
12.33
RUN#
Equiv.
KHCO3
Equiv.
KF
Equiv.
urea-H2O2
Volume
Solvent Ratio:
THF/MeOH
Temp
(deg C)
Impurity
amount
2
5.0
1.5
5.0
60
0.58
25
8.65
8
5.0
1.5
5.0
60
0.58
25
8.57
11
5.0
1.5
5.0
60
0.58
25
8.24
18
5.0
1.5
5.0
60
0.58
25
8.17
10
6.5
1.0
6.5
80
0.48
35
4.75
15
3.5
1.0
6.5
80
0.48
15
5.72
20
3.5
2.0
3.5
80
0.48
35
6.17
19
6.5
1.0
3.5
40
0.48
35
6.54
17
3.5
1.0
3.5
40
0.48
15
7.35
6
6.5
2.0
6.5
40
0.48
15
8.35
9
6.5
2.0
3.5
80
0.48
15
9.87
12
3.5
2.0
6.5
40
0.48
35
12.33
5
6.5
1.0
3.5
80
0.68
15
2.86
3
3.5
1.0
3.5
80
0.68
35
2.94
14
6.5
2.0
6.5
80
0.68
35
3.18
16
3.5
2.0
6.5
80
0.68
15
3.29
1
6.5
2.0
3.5
40
0.68
35
3.52
13
6.5
1.0
6.5
40
0.68
15
3.93
7
3.5
2.0
3.5
40
0.68
15
4.07
4
3.5
1.0
6.5
40
0.68
35
5.22
RUN#
Equiv.
KHCO3
Equiv.
KF
Equiv.
urea-H2O2
Volume
Solvent Ratio:
THF/MeOH
Temp
(deg C)
Impurity
amount
2
5.0
1.5
5.0
60
0.58
25
8.65
8
5.0
1.5
5.0
60
0.58
25
8.57
11
5.0
1.5
5.0
60
0.58
25
8.24
18
5.0
1.5
5.0
60
0.58
25
8.17
1
6.5
2.0
3.5
40
0.68
35
3.52
13
6.5
1.0
6.5
40
0.68
15
3.93
7
3.5
2.0
3.5
40
0.68
15
4.07
4
3.5
1.0
6.5
40
0.68
35
5.22
19
6.5
1.0
3.5
40
0.48
35
6.54
17
3.5
1.0
3.5
40
0.48
15
7.35
6
6.5
2.0
6.5
40
0.48
15
8.35
12
3.5
2.0
6.5
40
0.48
35
12.33
5
6.5
1.0
3.5
80
0.68
15
2.86
3
3.5
1.0
3.5
80
0.68
35
2.94
14
6.5
2.0
6.5
80
0.68
35
3.18
16
3.5
2.0
6.5
80
0.68
15
3.29
10
6.5
1.0
6.5
80
0.48
35
4.75
15
3.5
1.0
6.5
80
0.48
15
5.72
20
3.5
2.0
3.5
80
0.48
35
6.17
9
6.5
2.0
3.5
80
0.48
15
9.87
O
E
t
O
O
C
H
O
F
p
i
p
e
r
i
d
i
n
e
A
c
O
H
t
o
l
u
e
n
e

r
e
f
l
u
x
9
0
-
9
5
%
+
O
N
a
H
M
D
S
F
O
O
O
E
t
O
N
H
4
O
A
c
C
u
(
O
A
c
)
2
N
F
O
E
t
O
N
F
O
O
F
O
E
t
3
0
-
3
5
%
i
s
o
l
a
t
e
d

y
i
e
l
d
+
1
0

-

2
5
%
+

m
a
n
y

o
t
h
e
r

i
m
p
u
r
i
t
i
e
s
Example 2: Reaction Development

HOAc
mL/g
Eq
NH
4
OAc
Eq. Cu
Residual
Water
Residual
THF
4
1.35
1.1
0
0
5
2.7
1.33
0
0
7.5
4
2
5
10
F
O
O
O
E
t
O
N
F
O
E
t
O
B
e
s
t

i
n
i
t
i
a
l

c
o
n
d
i
t
i
o
n
s
:
2
.
7

e
q
.

N
H
4
O
A
c
1
.
3
3

e
q
.

C
u
(
I
I
)
O
A
c
2
H
O
A
c
1
1
6

d
e
g

C
1
6

h
DOE Parameters Chosen for Study

Fractional factorial 2
5
-
1

experiment

RUN#
mL HOAc /
g S.M.
Eq
NH
4
OAc
Eq. Cu
Residual
Water
Residual
THF
Rel%
completion
Rel%
Decarb.
Yield
(area%)
1
7.5
1.35
2.00
5
0
59.6
15.52
40.24
2
4.0
4.00
1.00
5
10
89.3
6.14
71.82
3
4.0
1.35
1.00
5
0
77.5
13.79
55.86
4
4.0
1.35
1.00
0
10
83.2
11.53
62.49
5
5.0
2.70
1.33
0
0
91.0
10.78
69.81
6
7.5
4.00
2.00
0
0
93.6
9.64
73.14
7
7.5
1.35
1.00
0
0
93.9
8.65
75.11
8
7.5
4.00
1.00
5
0
53.5
7.11
40.14
9
5.0
2.70
1.33
0
0
90.4
11.10
69.63
10
7.5
4.00
2.00
5
10
91.3
9.16
70.93
11
7.5
1.35
1.00
5
10
87.1
13.26
64.47
12
7.5
4.00
1.00
0
10
86.8
8.73
68.15
13
7.5
1.35
2.00
0
10
71.4
11.70
50.35
14
5.0
2.70
1.33
0
0
91.6
12.11
69.20
15
4.0
1.35
2.00
5
10
81.0
14.14
56.82
16
4.0
4.00
1.00
0
0
99.7
5.99
82.80
17
5.0
2.70
1.33
0
0
91.6
10.94
69.84
18
4.0
4.00
2.00
0
10
99.9
5.36
83.33
19
4.0
4.00
2.00
5
0
98.8
9.43
78.43
20
4.0
1.35
2.00
0
0
84.2
13.95
60.54
Screening DOE

Overall Analysis

Main effects favor

> Ammonium acetate,

< Volume,

< Residual water,

> THF.


Effects of copper noted
in two way interactions.

Factor
Low
High
HOAc mL/g S.M.
C, Q
Eq NH4OAc
C, Q, D
Eq. Cu
d, q
c
Residual Water
C, Q, d
Residual THF
c, q
C: Higher conversion to product

Q: Improved yield (area%)

D: Less decarboxylation

Rel% Completion Interaction between
Eq. NH
4
OAc and Eq. Copper (p < .0001)
50
60
70
80
90
100
1.1
2
Eq. Copper
Rel% Completion
Eq NH4OAc=1.35
Eq NH4OAc=4
Example 3: Reaction Screening
-

Direct Reduction of Ester to Aldehyde

N
F
H
O
N
F
O
E
t
O
N
F
O
H
D
I
B
A
L

/

-
7
8

d
e
g

C
N
F
O
E
t
O
A
l
x
x
L
A
H
T
E
M
P
O

/
b
l
e
a
c
h
o
r
(
E
t
2
N
)
2
A
l
H
X

=

-
i
B
u
,

o
r

-
N
E
t
2
X
C
O
O
E
t
X
C
H
O
C
H
3
O
O
N
a
+
C
H
3
O
O
A
l
H
H
_
A
b
e
,

T
.
,

e
t

a
l
.


T
e
t
r
a
h
e
d
r
o
n

5
7
,

2
0
0
1
,

2
7
0
1
-
2
7
1
0
N
H
K
O
t
B
u


0
.
2

m
o
l
T
H
F
,

1
0

d
e
g

C
2
.
0

m
o
l
2
.
1

m
o
l
>

9
0
%
+
X
C
H
2
O
H
<

5
%
T
w
o

i
n
t
e
r
m
e
d
i
a
t
e
s
p
r
o
p
o
s
e
d

b
a
s
e
d
o
n

m
e
c
h
a
n
i
s
t
i
c

w
o
r
k
:
X
N
H
O
X
H
O
E
t
O
(
R
O
)
2
A
l
(
R
O
)
2
A
l
N
_
_
N
a
+
N
a
+
H
2
O
Literature Lead

b

RUN#
Solvent
Temp
Eqs Pyrr
Eqs RedAl
Ratio
aldehyde/alcohol
1
MTBE
20
2
4
4.69
2
THF
10
3
3
3.11
3
THF
0
2
2
1.57
4
Toluene
0
2
4
10.62
5
THF
20
2
4
1.10
6
MTBE
0
2
2
0.00
7
THF
10
3
3
2.95
8
Toluene
0
4
2
10.55
9
THF
10
3
3
3.41
10
MTBE
20
4
2
1.09
11
THF
0
4
4
4.47
12
Toluene
20
2
2
8.00
13
THF
10
3
3
3.37
14
MTBE
0
4
4
2.34
15
Toluene
20
4
4
10.88
16
THF
20
4
2
2.73
N
F
H
O
N
F
O
E
t
O
N
F
O
H
R
e
d
A
l
-
P
X
Kinetic Study:

Stability of Active Acyl Species




Six well
reactor block

HPLC
autosampler
vials

N
C
O
C
l
N
O
O
S
C
O
P
h
N
C
O
C
l
N
O
O
S
C
O
P
h
0
0.005
0.01
0.015
0.02
0.025
0.03
0.035
0.04
0
500
1000
1500
2000
2500
3000
3500
Time (sec)
Apparent Molarity Enantiomer
(HPLC)
-5 C
2 C
10 C
16 C
Kinetic Study: Results and Data Reduction

Kinetic Study: Results and Data Reduction

-4.5
-4
-3.5
-3
-2.5
-2
0
500
1000
1500
2000
2500
3000
Time (sec)
ln(2S-Si+Ri)
-5 C
2 C
10 C
16 C
First Order Plot

9/29/2003
Bristol-Myers Squibb Co.
54
Window of Opportunity
2% Threshold
Conversion
Acid Chloride
racemization

1h 2h 3h 4h

Reaction

Screening

Solubility

measurements

Solution

stability

measurements

Reaction Mixture

stability

measurements

Crystallization

method screening

Kinetics

measurements

Reaction

Development



Parameter

Range

Studies

Partition

coefficient

determination

Automation Technology
-

Multiple Functions

Process

Knowledge

intuition

inspiration

art


linear approaches,


of limited scope

based on existing knowledge


handcrafted solutions,

phenomenological

basis


a test of the design


rational

methodic

analytical


parallel approaches,

based on knowledge but

poised for discovery


systematically
-
derived,

integrated solutions,

knowledge
-
based


validation of what is known


…basis

of their field

and work


…approach

to route scouting



...approach to

overall process

design


…approach to

scaleup

Values / Beliefs of Process Organic Chemists

Past Future

Automated and parallel experimentation
assists process chemists with the
generation of
process knowledge



Shorten cycle time for process
development


Increase productivity of each scientist


Automated reactions combined with on
-
line analysis leads to better information


asdf

extending and enhancing human life

7

Acknowledgements

Process R&D Automation Group:

Erik Rubin, Victor Rosso, Jun Qiu, Anne Song, Greg Lane, John
Venit


Tamao
-
Fleming Reaction Development:

Kishta Katipally, Victor Rosso, Wen
-
sen Li


Pyridine Intermediate Reaction Development:

Jun Qiu, Rama Chidambaram, J. Zhu, Victor Rosso, Joydeep
Kant


Ester Reduction Reaction Screening:

Bin Zheng, Joydeep Kant


Acid chloride Racemization Kinetics:

Eric Rubin, Victor Rosso


This talk developed and presented at the National Organic
Symposium, 2003 by:


Ed Delaney