Monitoring Chemical Reactions in Real Time with NMR Spectroscopy

Monitoring Chemical Reactions in Real

Time with NMR Spectroscopy

Mark Zell,1 Brian Marquez,1 Dave am Ende,1 Pascal Dube,1 Eric Gorman,2 Robert

Krull,3 Don Piroli,3 Kimberly Colson,3 Michael Fey3

1. Pfizer Global Research and Development, 445 Eastern Point Road, Groton, CT 06340

2. Department of Pharmaceutical Chemistry, University of Kansas, 2095 Constant Drive, Lawrence, KS 66047

3. Bruker Biospin, 15 Fortune Drive, Billerica, MA 01821

Entering the forest he moves not the

grass; entering the water he makes

not a ripple.

Zenrin Kushu

Introduction

• Many analytical techniques (IR, UV, Raman) are used to monitor

chemical reactions. – For many reactions these techniques show important features of the reaction process

and therefore provide a great deal of information to apply to a process.

– Some features, such as structural isomers and low level impurities are not typically

observed.

– Utilization of NMR provides additional “molecular resolution” at the process development

stage to provide many details about the process (e.g. mechanistic details, kinetics of

impurity formation, etc…)

• On-line NMR is a very powerful tool for monitoring reaction progression as well as

providing mechanistic insight into the synthesis.

• We are exploring ways to add detailed, real-time chemical information to better

assess synthetic reactions and develop a better process understanding.

• This data should be acquired in a non-invasive way to better

understand the chemistry.

Introduction

Goals

Monitor chemical reactions via NMR Spectroscopy.

• Develop a system that can provide a real-time reaction

monitoring capability with value added information for

better process understanding.

Pictures of the “first” System in our lab (~2006)

NMR ProbeInlet and outlet

HPLC Pump

NMR Probe

ReturnReaction Vessel

Temperature controlSyringe Pump

NH2

NH2

HH

O

O

N

N

2 H2O

C12H24N2Mol. Wt.: 196.33

C2H2O2Mol. Wt.: 58.04

C14H22N2Mol. Wt.: 218.34

A Real Life Example: Flow NMR Semi-Batch Reaction

Profile for Alkylation of a Di-amine with Glyoxal

0 20 40 603000

4000

5000

0 20 40 600

5000

10000

0 20 40 600

5000

10000

0 20 40 600.8

1

1.2

1.4x 10

4

0 20 40 600

5000

10000

15000

0 20 40 600

1

2

3x 10

4

0 20 40 600

1

2

3x 10

4

0 20 40 600

5000

10000

15000

0 20 40 600

5000

10000

0 20 40 600

1

2

3x 10

4

0 20 40 600.5

1

1.5

2x 10

4

0 20 40 600.8

1

1.2

1.4x 10

4

0 20 40 600

1000

2000

3000

0 20 40 600

5000

10000

15000

0 20 40 600.9

1

1.1

1.2x 10

4

0 20 40 600

5000

10000

15000

0 20 40 600

5

10

15x 10

4

0 20 40 600

2

4x 10

4

0 20 40 600

5000

10000

15000

0 20 40 600

2

4x 10

4

0 20 40 600

2000

4000

0 20 40 600

1

2x 10

4

0 20 40 600

2

4

x 104

0 20 40 60-100

0

100

200

0 20 40 600

2

4

x 104

Flow NMR Flow NMR –– Automatic Analysis in Automatic Analysis in

MatlabMatlab Profiles all peaksProfiles all peaks

0 20 40 603000

4000

5000

0 20 40 600

5000

10000

0 20 40 600

5000

10000

0 20 40 600.8

1

1.2

1.4x 10

4

0 20 40 600

5000

10000

15000

0 20 40 600

1

2

3x 10

4

0 20 40 600

1

2

3x 10

4

0 20 40 600

5000

10000

15000

0 20 40 600

5000

10000

0 20 40 600

1

2

3x 10

4

0 20 40 600.5

1

1.5

2x 10

4

0 20 40 600.8

1

1.2

1.4x 10

4

0 20 40 600

1000

2000

3000

0 20 40 600

5000

10000

15000

0 20 40 600.9

1

1.1

1.2x 10

4

0 20 40 600

5000

10000

15000

0 20 40 600

5

10

15x 10

4

0 20 40 600

2

4x 10

4

0 20 40 600

5000

10000

15000

0 20 40 600

2

4x 10

4

0 20 40 600

2000

4000

0 20 40 600

1

2x 10

4

0 20 40 600

2

4

x 104

0 20 40 60-100

0

100

200

0 20 40 600

2

4

x 104



0 20 40 603000

4000

5000

0 20 40 600

5000

10000

0 20 40 600

5000

10000

0 20 40 600.8

1

1.2

1.4x 10

4

0 20 40 600

5000

10000

15000

0 20 40 600

1

2

3x 10

4

0 20 40 600

1

2

3x 10

4

0 20 40 600

5000

10000

15000

0 20 40 600

5000

10000

0 20 40 600

1

2

3x 10

4

0 20 40 600.5

1

1.5

2x 10

4

0 20 40 600.8

1

1.2

1.4x 10

4

0 20 40 600

1000

2000

3000

0 20 40 600

5000

10000

15000

0 20 40 600.9

1

1.1

1.2x 10

4

0 20 40 600

5000

10000

15000

0 20 40 600

5

10

15x 10

4

0 20 40 600

2

4x 10

4

0 20 40 600

5000

10000

15000

0 20 40 600

2

4x 10

4

0 20 40 600

2000

4000

0 20 40 600

1

2x 10

4

0 20 40 600

2

4

x 104

0 20 40 60-100

0

100

200

0 20 40 600

2

4

x 104



Starting

Material

Product

Reaction Profile -- Intermediates

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

Intermediate

Introduction

Goals

Monitor chemical reactions via NMR Spectroscopy.

• Develop a system that can provide a real-time

reaction monitoring capability with value added

information for better process understanding.

Next Generation

• Limitations with first system (HPLC/NMR flow probe):

– Fragile – 200 microliter small bore cell.

– If (When) flow cell broke had to be sent back for repair

– Flow rates had to be low to avoid pressure, easily plugged

– No temperature control to/from cell

– 1H observe only

• New Design in 2010:

Flow control unit coupled with a flow cell that will fit in a standard NMR

probe.

Easy to replace if flow cell is broken or plugged

Temperature control ~0 – 80° C (achieved -10° C to date), +/- ~2° C

Goal is to reach ~ -40° C

Standard 1H-19F-13C TXO NMR probe (low gamma probe on order)

5 Gauss Line

~1.5 m

Reactor

Thermostatted Supply

And Return Lines

Control

Unit

Fast Loop

(~ 25 ml/min)

Slow Loop

(~5 ml/min)

New and Improved!

1H, 13C, 19F Triple Resonance NMR Probe

Coming Soon!

ReactIR

Bruker IonTrap MS

New Flow Cell Design

9 8 7 6 5 4 3 2 1 ppm

First data acquired on new flow cell at Bruker. 2 mM sucrose sample.

Spectrum demonstrates good NMR lineshape even with tubing extended

Into flowcell.

5.35 ppm

Optimal Flow rate minimizes lag from

reaction vessel

Acetic Acid 1H Signal depends on the

flow rate through the NMR cell

0

1000

2000

3000

4000

0 2 4 6 8 10 12

Peak A

rea (

arb

itra

ry u

nit

s)

Flow Rate (mL/min)

Effect of Flow Rate on Peak Area

Acetic Acid in D2O

Time to Reach Equilibrium at 5 mL/min

1 mL Acetonitrile in 100 mL D2O

Void volume ~5mL

Acetic Anhydride

O

O O

OH

O

0

50

100

150

200

0 20 40 60

Co

ncen

trati

on

(m

M)

Time (min)Acetic Anhydride

Acetic Acid

NMR Flow Kinetics~20 ºC and 5 mL/min

Temp (ºC) 103 * kD (sec-1)

NMR tube 25 0.94

Literature ? 0.89

NMR flow ~20 0.82

Lit kH/kD = 2.9

O

O O D2O

OH

O

Quantitative NMR Spectroscopy

• Electronic REference To access In vivo Concentrations (ERETIC)

Method

– Prior to the ERETIC method

• Internal standards were necessary for absolute quantitation

• Potential problems included signal overlap, chemical interaction, and

errors in addition of the internal standard

– ERETIC

• Electronically generated signal injected into the spectrometer to

replace the internal standard

• Requires additional hardware and the reference signal can’t be moved

once the data is collected

– Digital ERETIC (QANTUS, Quantification by Artificial Signal, Upton,

et.al.)

• Synthetic signal added after data collection as part of the data

processing

• Reference signal can be moved anywhere in the spectrum without

recollecting the data and does not require additional hardware

Digital ERETIC Method

Analyte Spectrum (A)

ERETIC Spectrum (E)

Referenced Spectrum (A+E*x)

The ERETIC signal can be placed anywhere in

the spectrum without changing the calibration.

QNMR Potency Measurements

Compound Lot #

Theoretical

Potency

Traditional

QNMR

Digital ERETIC QNMR

Analyte Maleic acid

Benzoic acid NIST 39j 0.999996 --- 1.016 ---

Maleic acid 1296124 >0.99 0.990 --- 0.990

p-Fluorobenzoic acid NIST 2143 0.9995 0.987 1.024 0.994

PF-04971729-G3

L-Pyroglutamic acidGR02878

0.755

0.235

0.733

0.242

0.736

0.2410.996

3-TMS-1-propylsulfonic acid

sodium salt23021EB 1.000 0.917 0.868 0.941

1,4-Bis(TMS)benzene 04202KC 0.995 0.453* 0.460* 1.003

α-D-Glucose, anhydrous 12926AC 0.999 0.768 0.981 1.262

Digital ERETIC signal calibrated against the Maleic acid peaks in the Benzoic acid sample, assuming the purity determined by

the Traditional QNMR method

* = Sonicated several times but not all of the sample dissolved

Esterification reaction

O

OH

F

OHF

F

F

CDI

O

O

F

FF

F

Reaction Conditions:

First reaction:

20 mM para-fluorobenzoic acid circulating in reactor. CDI

and trifluoroethanol both added in excess. Acquired proton

data only (top spectra).

Second Reaction:

20 mM para-fluorobenzoic acid circulating in reactor. Added

CDI and allowed to stir over lunch. Added trifluoroethanol

and no reaction occurred. Added additional (non-

stoichiometric) amount of CDI and reaction proceeded.

Reaction did not go to completion. 19F (and 1H – not shown)

data acquired during this reaction (bottom spectra)

Reaction to be repeated again in the next couple of weeks

after pump variability is resolved.


Primarily Unreacted – 1H

Primarily Reacted – 1H

Methylene Protons


Primarily Unreacted – 19F

Partially Reacted – 19F

Methyl Fluorines

-10

0

10

20

30

40

50

60

0 200 400 600 800 1000 1200 1400

Time (min)

Co

nce

ntr

atio

n (

mM

)

Product

Impurity A

Impurity B

Product

Intermediate

StartingMaterialBenzoic acidMass Balance

Dilution due to additional trifluoroethanol


Applicability to Pfizer projects

• This technology is directly applicable to a wide range of internal synthetic projects.

• Previously showed the use of our early flow-NMR system to identify the presence of a transient intermediate in the synthesis of an active ingredient in a currently marketed product.

• The data shown in the following slides is from a current internal project, and is just one of many projects that could benefit from advances in the area of flow NMR technology. – Step 1: Formation of an aryl-metal species– Step 2: The addition of this aryl-metal to a cyclobutanone.

• These reactions are sensitive to moisture and require a careful control of the conditions. Our current understanding of the chemistry is limited by the use of off-line analytical tools.

Temperature • Reactor = -20 °C• Line temperature = -10 °C• NMR probe temperature = -10 °C

• The reaction mixture was circulated through the NMR probe at 5 mL/min.

• 1H, 19F, and 13C spectra were collected at various points during the reaction.

• Further steps were not investigated due to solvent-induced pump leak!


Aryl bromide

Aryl bromide + isopropylmagnesium chloride

Aryl Bromide + IsopropylmagnesiumChloride + Butyl Lithium (during addition)

Aryl Bromide + Isopropylmagnesium Chloride + Butyl Lithium (~10 min after addition)


THF-p8


Aryl bromide


Aryl Bromide + Isopropylmagnesium Chloride + Butyl Lithium (during addition)



19F Spectra (1H Decoupled)


THF-p8

Aryl bromide


Aryl Bromide + Isopropylmagnesium Chloride + Butyl Lithium (during addition)


13C Spectra


Before Addition

Time Since Cyclobutyl Ester Addition

During Addition

~5 min

~15 min

~25 min

10 min. interval

1H Spectra


Next Steps

• Reduce the lag time to 30 seconds or less.

– Use of a pump with appropriate seals is critical. (so far using

slow loop at 5 ml/min through 1/16” tubing)

– Improve fast loop operation while keeping approx 5 ml/min

thru the slow loop.

• Further characterize the system at sub-ambient conditions

and ensure adequate thermostating of the lines.

• Extend to catalytic hydrogenations, follow reaction kinetics

without disrupting the gas-liquid equilibrium.

• Continue to extend to new chemistries, unit operations

(distillations, pKa determinations, extractions etc)

• Enable as facilitated walk-up platform

Acknowledgements

• Bruker-Biospin

• Pfizer Groton and Sandwich SE groups

• Jason Mustakis

• Carrie Wager

• Steve Brown

• Geraldine Taber

• Andreas Kaerner – Eli Lilly & Company

• John Edwards – Process NMR

Documents

Monitoring Chemical Reactions in Real Time with NMR Spectroscopy