Development of a practical synthesis of a p38 MAP kinase inhibitor

Article abstract of DOI:10.1021/op800250v

A practical synthesis of the phthalazine-based p38 MAP kinase inhibitor [(S)-2] was needed for an ongoing program. Vibrational circular dichroism provided the assignment of the absolute stereochemistry of the target compound. The selected synthetic route

Full text of DOI:10.1021/op800250v

Organic Process Research & Development 2009, 13, 230–241
Development of a Practical Synthesis of a p38 MAP Kinase Inhibitor
Oliver R. Thiel,*,† Michal Achmatowicz,† Charles Bernard,† Philip Wheeler,† Cecile Savarin,† Tiffany L. Correll,‡
Annie Kasparian,† Alan Allgeier,† Michael D. Bartberger,§ Helming Tan,‡ and Robert D. Larsen†
Department of Chemical Process Research and DeVelopment, Department of Molecular Structure and Design, and Department
of Analytical Research and DeVelopment, Amgen Inc., One Amgen Center DriVe, Thousand Oaks,
California 91320-1799, U.S.A.
Abstract:
A practical synthesis of the phthalazine-based p38 MAP kinase
inhibitor [(S)-2] was needed for an ongoing program. Vibrational
circular dichroism provided the assignment of the absolute
stereochemistry of the target compound. The selected synthetic
route for (S)-2 required identification of efficient reaction condi-
tions for the construction of carbon-oxygen, carbon-carbon, and
Figure 1. p38 MAP kinase inhibitors 1 and (S)-2.
carbon-nitrogen bonds to connect the key building blocks. An
efficient two-step method (chlorodehydroxylation, aromatic nu-
cleophilic substitution) for the synthesis of arylether [(S)-10] was
developed. PAT (in situ Raman spectroscopy) was utilized to
monitor and control the formation of a lithium alkoxide in this
on a phthalazine scaffold.4 A practical synthesis of phthalazine
derivative 1 was recently reported by this laboratory,5 and the
current contribution describes the process research and develop-
ment involved in the large-scale synthesis of the derivative (S)-2
(Figure 1). The two development candidates differ in the
1-substituent of the phthalazine; while the replacement of the
morpholine with the chiral trifluoroisopropanol may appear to
be only a minimal change, from a synthetic perspective, it
resulted in very different chemical and physicochemical be-
havior of the intermediates that required significant process
development efforts.
The synthesis that was employed by our medical chemistry
team is shown in Scheme 1.4 2-Chloro-6-bromophthalazine (3)6
was reacted with the sodium alkoxide derived from 1,1,1-
trifluoropropan-2-ol [(()-9] (3 equiv) in a mixture of THF and
DMF to afford ether 4 in a racemic fashion. This aryl bromide
was subjected to a Suzuki-coupling with pinacol ester 57 to
afford the target compound (()-2 in good yield. The desired
enantiomer of this compound was obtained by chiral preparative
chromatography (SFC with Chiralpak AD). This approach led
to an undetermined absolute stereochemistry of 2, making this
a key issue prior to initiation of further synthetic efforts.
This first-generation synthesis suffered from the limited
availability of the building blocks 3 and 5. Furthermore, the
need for a chromatographic resolution in the preparation of the
desired enantiomer made the yields and efficiency of this
synthesis low. To support the program we required an efficient
synthesis of (S)-2 that would enable production of kilogram
quantities.
reaction. The synthesis of (S)-2 was completed using high-yielding
Suzuki- and amide-coupling reactions. The isolation conditions for
these steps were optimized to obtain material of very high purity
without the need for any complicated workup procedures.
Introduction
p38 mitogen-activated protein (MAP) kinases are intracel-
lular serine/threonine kinases that positively regulate the produc-
tion and action of several pro-inflammatory mediators, specif-
ically the release of tumor necrosis factor-R (TNF-R) and
interleukin-1 (IL-1) in response to stress.1 These cytokines are
involved in disease states such as rheumatoid arthritis (RA),
Crohn’s disease (inflammatory bowel disease), and psoriasis.
Biological agents that sequester TNF-R show impressive clinical
efficacy in the treatment of these diseases.2 Small-molecule
inhibitors of p38 MAP kinase would be a potential alternative
for these biological agents, and the efficacy of these agents has
been demonstrated in clinical studies.3 Our medicinal chemistry
team has been pursuing the discovery of p38 MAP kinase
inhibitors for several years, and many suitable candidates have
been selected for further development. Substantial efforts have
focused on highly selective and potent inhibitors that were based
* To whom correspondence should be addressed: Telephone: 805-447-9520.
Fax: 805-375-4532. E-mail: othiel@amgen.com.
† Department of Chemical Process Research and Development.
‡ Department of Analytical Research and Development.
§ Department of Molecular Structure and Design.
(4) Herberich, B.; Cao, G.-Q.; Chakrabarti, P. P.; Falsey, J. R.; Pettus,
L.; Rzasa, R. M.; Reed, A. B.; Reichelt, A.; Sham, K.; Thaman, M.;
Wurz, R. P.; Xu, S.; Zhang, D.; Hsieh, F.; Lee, M. R.; Syed, R.; Li,
V.; Grosfeld, D.; Plant, M. H.; Henkle, B.; Sherman, L.; Middleton,
S.; Wong, L. M.; Tasker, A. S. J. Med. Chem. 2008, 51, 6271.
(5) Achmatowicz, M.; Thiel, O. R.; Wheeler, P.; Bernard, C.; Huang, J.;
Larsen, R. D.; Faul, M. M. J. Org. Chem. 2009, 74, 795.
(6) Obtained in four steps (45% overall yield) from 4-bromo-2-methyl-
benzonitrile.
(1) Chen, Z.; Gibson, T. B.; Robinson, F.; Silvestro, L.; Pearson, G.; Xu,
B.; Wright, A.; Vanderbilt, C.; Cobb, M. H. Chem. ReV. 2001, 101,
2449.
(2) Etanercept, infliximab, and adalimumab are TNF-R blockers currently
approved in the United States and elsewhere for various inflammatory
diseases.
(3) (a) Gaestal, M.; Mengel, A.; Bothe, U.; Asadullah, K. Curr. Med.
Chem. 2007, 14, 2214. (b) Peifer, C.; Wagner, G.; Laufer, S. Curr.
Top. Med. Chem. 2006, 6, 113. (c) Margutti, S.; Laufer, S. A.
ChemMedChem 2007, 2, 1116.
(7) Prepared in two steps from 3-iodo-4-methylbenzoic acid, utilizing a
palladium-catalyzed boron-carbon formation.
230
•
Vol. 13, No. 2, 2009 / Organic Process Research & Development
10.1021/op800250v CCC: $40.75
2009 American Chemical Society
Published on Web 01/15/2009

Scheme 1. Medicinal chemistry synthesis of (S)-2
Scheme 2. Retrosynthesis of (S)-2
Earlier development work for the synthesis of 1 had provided
access to building blocks 6 and 7,5 which we decided to leverage
for the synthesis of 2 (Scheme 2). We envisaged converting
6-chlorophthalazin-1-ol (6) into 1,6-dichlorophthalazine (8),
followed by displacement with the enantiomerically pure alcohol
(S)-9 to obtain the key building block (S)-10. Suzuki-coupling
with boronic acid 7 followed by amide-coupling with cyclo-
propylamine should allow access to the target compound.
Several key developmental challenges were identified in this
proposed synthetic sequence. The formation of arylether (S)-
10 would require scaleable reaction conditions, which avoid
potential side reactions that are associated with the chlo-
rophthalazine scaffold.5 The Suzuki-coupling of aryl chloride
(S)-10 was expected to require an extensive screen of reaction
conditions to obtain a high yield in the formation of this key
bond. In addition to these purely synthetic organic challenges
significant challenges in the physicochemical properties of the
intermediates were encountered. Upon switch from the racemic
to the enantiopure series a significant decrease in the solubility
of the key intermediates forced us to redesign the isolation
procedures.
to the corresponding amide. Selective ortho metalation using
methyllithium followed by quench with N,N-dimethylforma-
mide afforded 12 without incident. This hydroxyisoindolinone
was transformed to 6-chlorophthalazin-1-ol (6) using hydrazine
hydrate in acetic acid. Overall, this synthetic sequence could
be readily scaled up, and a good overall yield was obtained on
kilogram scale.5 The boronic acid 7 was obtained from 3-iodo-
4-methylbenzoic acid through a one-pot process. The carboxylic
acid functionality was deprotonated without concomitant reduc-
tion of the iodide using isopropylmagnesium chloride below
-50 °C. A second equivalent of the Grignard reagent was used
to effect a clean magnesium-iodine exchange, and the resulting
arylmagnesium species was trapped with triisopropyl borate.
Acidic hydrolysis followed by crystallization from THF/heptane
allowed access to 7 in acceptable yield.5
Results and Discussion
Determination of Absolute Stereochemistry. At the outset
of our development work, the absolute configuration of the more
active enantiomer of 2 had not been determined. Initially we
were able to procure only (()-1,1,1-trifluoropropan-2-ol [(()-
9], which gave rise to compound (()-10 and subsequent
products as mixtures of enantiomers. Thus, it was of interest to
determine the absolute stereochemistry of the active form of 2
The synthetic approach to the phthalazine core 6 leveraged
the inherent symmetry of the starting material (Scheme 3).
Reaction of 4-chlorobenzoyl chloride with tert-butylamine led
Vol. 13, No. 2, 2009 / Organic Process Research & Development
•
231

Scheme 3. Synthesis of building blocks 6 and 7
in order to facilitate selective synthesis of the more potent
stereoisomer. Single X-ray crystallography was not directly
available for assignment due to the lack of heavy-atom
containing intermediates.
HPLC analysis (Chiralpak AD-H) led to the assignment of 2
as the (S)-enantiomer.
The VCD assignment was subsequently confirmed by
chemical synthesis (Scheme 4). Racemic 2-(trifluoromethy-
l)oxirane [(()-12] was subjected to Jacobsen’s hydrolytic kinetic
resolution using (R,R)-Co(salen) as catalyst leading to enan-
tiomerically enriched (S)-2-(trifluoromethyl)oxirane [(S)-13].12,13
Reductive ring-opening with lithium aluminum hydride afforded
(S)-9 which upon reaction with 1,6-dichlorophthalazine (8),
prepared by the same method as reported in the synthesis of
1,5 led to the intermediate (S)-10, the chiral HPLC retention
time of which was identical with that of 10a, thereby cor-
roborating the predicted stereochemical assignment mentioned
Vibrational circular dichroism (VCD)8 has emerged as a
reliable and efficient method for the determination of absolute
stereochemistry of chiral molecules in solution without the need
for crystalline material as required by X-ray crystallography.
Comparison of the experimental VCD spectra of each enanti-
omer (necessarily mirror images) to those obtained via quantum
mechanical calculations9 for a defined stereochemical config-
uration provides an absolute assignment. Species 10 was chosen
for VCD determination due to its ease of chiral separation and
small size, the latter allowing fast computational spectral
determination and affording a simplified conformational land-
scape. Reaction of 1,6-dichlorophthalazine (8) with the lithium
alkoxide of (()-9 afforded (()-10 (Vide infra for more details
about the reaction conditions). After separation of the enanti-
omers by chiral preparative chromatography (Chiralpak AD-
H, hexane/2-propanol 95:5) the VCD spectra of both enanti-
omers, 10a (tR ) 5.4 min) and 10b (tR ) 6.2 min), were
acquired on a BioTools Dual PEM Chiral IR instrument.
Conformational ensembles of (R)- and (S)-10 were obtained
via stochastic search using the MMFF94 force field,10 followed
by geometry optimization and harmonic frequency and VCD
rotational strength determination with the B3LYP hybrid density
functional and 6-31G* basis set.11 The predicted VCD (line)
spectra for the three resultant, unique conformations were
convolved using a Lorentzian function, followed by Boltzmann
weighting based on the predicted relative free energy of each
conformation to yield a final predicted VCD spectrum for each
enantiomer. Comparison of the theoretical spectra with those
obtained experimentally led to the unambiguous assignment of
stereoisomer 10a (see Figure 2) as (S)-10. The use of this
compound in the remaining synthetic steps coupled with chiral
(8) (a) Stephens, P. J.; Devlin, F. J.; Pan, J. Chirality 2008, 20, 643. (b)
Freedman, T. B.; Cao, X.; Dukor, R. K.; Nafie, L. A. Chirality 2003,
15, 734.
(9) (a) Devlin, F. J.; Stephens, P. J.; Cheeseman, J. R.; Frisch, M. J. J.
Phys. Chem. A 1997, 101, 6322. (b) Devlin, F. J.; Stephens, P. J.;
Cheeseman, J. R.; Frisch, M. J. J. Am. Chem. Soc. 1996, 118, 6327.
(c) Cheeseman, J. R.; Frisch, M. J.; Devlin, F. J.; Stephens, P. J. Chem.
Phys. Lett. 1996, 252, 211.
(10) (a) Halgren, T. A. J. Comput. Chem. 1999, 20, 730. (b) The MMFF94
conformational search was performed with the Molecular Operating
EnVironment (MOE), v. 2007.09; Chemical Computing Group Inc.:
Montreal, Canada, 2007; http://www.chemcomp.com.
Figure 2. Solvent-subtracted experimental (a) and theoretical
(b) VCD spectra for (R)- and (S)-10.15
232
•
Vol. 13, No. 2, 2009 / Organic Process Research & Development

Scheme 4. Stereochemical assignment by the
enantioselective synthesis of (S)-10
Figure 3. Potential alkylating agents 14 and 15.
10; typically, no conversion was observed. At higher temper-
atures (55-60 °C), generation of multiple unidentified products,
resulting presumably from the decomposition of the alkylating
agent, were observed. Since the direct O-alkylation of 6 was
not successful, development of the SNAr route using (S)-9 and
8 as nucleophilic and electrophilic partners, respectively, was
explored (Scheme 5).
Scheme 5. SNAr route to (S)-10
above.14 In conclusion, both methods are complementary. In
this particular application the assignment using chemical
synthesis was relatively straightforward (two synthetic steps),
and it required similar resources compared to the VCD method.
In systems where intermediates of known absolute stereochem-
istry are not as easily available, however, VCD can offer distinct
advantages.
Formation of Carbon-Oxygen Bond through Aromatic
Nucleophilic Substitution. The medicinal chemistry route
involved nucleophilic aromatic substitution of 3 using the
sodium alkoxide derived from 1,1,1-trifluoropropan-2-ol [(()-
9]. A brief investigation in reversal of nucleophile and elec-
trophile was undertaken, since it would obviate the need for an
activation of 6 with phosphorous oxychloride. Tosylate 1416
and mesylate 15 were prepared as alkylating agents starting from
(R)-1,1,1-trifluoropropan-2-ol [(R)-9] (Figure 3).17 However, all
attempts to effect the O-alkylation of 6 using these substrates
under literature conditions (LHMDS in THF; K2CO3 in acetone;
NaH in DMF)18 failed to produce even a trace amount of (S)-
1,6-Dichlorophthalazine (8) had been an intermediate in the
synthesis of 1.5 The chlorodehydroxylation of 6 was performed
in chlorobenzene, followed by dilution with dichloromethane
and an aqueous NaOH quench. Initially, 8 was isolated as a
wet solid from the organic solution after solvent evaporation.
It was observed that this intermediate has some inherent stability
issues. Exposure to aqueous acidic conditions led to hydrolysis
and reversion to 6, whereas under anhydrous conditions
dimerization and oligomerization reactions took place in the
presence of Lewis or Brønsted acids. Due to these instability
concerns, a through-process was preferred in which 8 was not
isolated but carried crude to the next step (nucleophilic aromatic
substitution with morpholine). Although this through-process
was successfully executed on a kilogram scale for the synthesis
of 1, it was not optimal. The major concerns for the synthesis
of 8 that were to be addressed were as follows: (i) the potential
for a delayed exotherm during the quench, (ii) the use of
chlorinated solvents, and (iii) the lack of a control point.
Early on during development of the chlorodehydroxylation
of 6, it was noted that acetonitrile could be a viable solvent for
this transformation (complete conversion, 69% isolated yield
of 8). However, difficulties with a reproducible isolation of the
hydrolytically labile 8 from the water-miscible acetonitrile
initially discouraged further experiments. It was then conceived
that a quench and isolation from a buffered aqueous media could
mitigate the stability issues and therefore this approach war-
ranted revisiting.19 The chlorodehydroxylation reaction was
performed with 1.5 equiv of POCl3 in acetonitrile at 80 °C,
and upon complete conversion, the reaction mixture was
quenched inversely, so that the reaction mixture was transferred
into an aqueous phosphate buffer solution with concomitant
adjustment to pH 6.8-8.0 using 2 N aqueous NaOH while
maintaining the temperature below 30 °C. The product could
(11) All density functional theory calculations were performed with the
Gaussian 98 program system. Frisch, M. J.; Trucks, G. W.; Schlegel,
H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Zakrzewski,
V. G.; Montgomery, J. A., Jr.; Stratmann, R. E.; Burant, J. C.;
Dapprich, S.; Millam, J. M.; Daniels, A. D.; Kudin, K. N.; Strain,
M. C.; Farkas, O.; Tomasi, J.; Barone, V.; Cossi, M.; Cammi, R.;
Mennucci, B.; Pomelli, C.; Adamo, C.; Clifford, S.; Ochterski, J.;
Petersson, G. A.; Ayala, P. Y.; Cui, Q.; Morokuma, K.; Malick, D. K.;
Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Cioslowski, J.;
Ortiz, J. V.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.;
Komaromi, I.; Gomperts, R.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-
Laham, M. A.; Peng, C. Y.; Nanayakkara, A.; Gonzalez, C.; Challa-
combe, M.; Gill, P. M. W.; Johnson, B. G.; Chen, W.; Wong, M. W.;
Andres, J. L.; Head-Gordon, M.; Replogle, E. S.; Pople, J. A. Gaussian
98, Rev. A.11.3; Gaussian, Inc.: Pittsburgh, PA, 2002.
(12) Schaus, S. E.; Brandes, B. D.; Larrow, J. F.; Tokunaga, M.; Hansen,
K. B.; Gould, A. E.; Furrow, M. E.; Jacobsen, E. N. J. Am. Chem.
Soc. 2002, 124, 1307.
(13) The absolute stereochemistry for this substrate was assigned by
Jacobsen et al. by comparison of the optical rotation with earlier data
in Ramachandran, P. V.; Gong, B.; Brown, H. C. J. Org. Chem. 1995,
60, 41.
(14) After the completion of the synthesis of (S)-2 the absolute stereo-
chemistry was independently confirmed by a single-crystal structure
of (S)-2 in the active site of p38-kinase.
(15) Resolution of 8 cm-1 and Lorentzian half-widths for experimental and
theoretical plots, respectively. B3LYP/6-31G* frequencies scaled by
0.96.
(16) Hanack, M.; Ullman, J. J. Org. Chem. 1989, 54, 1432.
(17) The (R)-enantiomer of the chiral alcohol is also commercially available,
albeit in lower stereochemical purity (92-94% ee).
(18) (a) Cui, P.; Macdonald, T. L.; Chen, M.; Nader, J. L. Bioorg. Med.
Chem. Lett. 2006, 16, 3401. (b) Baraldi, P. G.; Tabrizi, M. A.; Preti,
D.; Bovero, A.; Romagnoli, R.; Fruttarolo, F.; Zaid, N. A.; Moorman,
A. R.; Varani, K.; Gessi, S.; Merighi, S.; Borea, P. A. J. Med. Chem.
2004, 47, 1434.
Vol. 13, No. 2, 2009 / Organic Process Research & Development
•
233

be isolated by simple filtration from the quench mixture
(solubility of 8 in mother liquors: <0.5 mg/mL). Prolonged
heating with POCl3 (12-14 h), inverse quench at up to 40 °C,
and aging of the quenched aqueous mixture at pH 7-11 (12-14
h) had no adverse effects on the purity of 8 (>96%).
layers were separated, and the organic layer was washed with
water (5 vol). The reaction solvents were replaced by EtOH
(10 vol) through vacuum distillation. Water (10 vol) was added
to the ethanol solution slowly at 40 °C, and the resulting
suspension was kept at 20 °C for 30 min. Filtration then afforded
(()-10 in 78% yield (>98% purity).
The procedure thus developed was successfully performed
on kilogram scale. After complete conversion of 6 (<2%) was
observed after 3 h at 80 °C, the inverse quench was performed
over 4 h, during which the pH was maintained in the desired
range (6.8-8.0). The precipitated materials were collected by
filtration and rinsed with water twice. The filter cake was dried
under a stream of nitrogen to afford 8 in good yield (79%
corrected yield, 96.2% purity, 87.0 wt %). The relatively low
potency of this material could be accounted for by the presence
of water (0.2% weight loss by TGA) and inorganic salts (1.7
wt % chloride and 0.3 wt % phosphate by ion chromatography)
as well as some oligomeric and polymeric byproducts. These
impurities were efficiently removed during a polishing filtration
operation in the next reaction.
On the basis of the reaction conditions that were developed
for (()-9 further optimization was conducted using (S)-9 as a
solution in MTBE (35-64 wt %). The stoichiometry and quality
of the organolithium reagent used in the deprotonation was of
critical importance. Using excess n-butyllithium led to formation
of 1-butoxy-6-chlorophthalazine and an unidentified impurity.
The formation of 1-butoxy-6-chlorophthalazine (up to 33% in
product) was likely due to the presence of lithium n-butoxide
formed by partial oxidation of n-butyllithium through traces of
air. In contrast, excess of (S)-9 relative to the organolithium
reagent was well tolerated in the reaction, and no impact on
product purity or isolated yield was observed. In order to control
the stoichiometry between (S)-9 and n-butyllithium, catalytic
1,10-phenanthroline (0.1-0.4 mol %) was used as an internal
indicator during the preparation of the lithium alkoxide. The
colorless solution became light yellow during addition of
n-butyllithium and eventually turned dark brown at the equiva-
lence point. An additional charge of (S)-9 (0.1 equiv) and a
corresponding color change back to yellow ensured that no
excess organolithium was present in the subsequent step.
A range-finding study on the stoichiometry of the alkoxide
relative to 8 was performed. It was found that 1.1 equiv (S)-9
of was sufficient for a complete conversion of 8 possessing high
purity (isolated by crystallization from dichloromethane/hep-
tane). However, due to a relatively high level of inorganic and
polymeric content (10-15 wt %) in representative batches of
8, 1.5 equiv of (S)-9 was preferentially used to ensure complete
conversion. At least 20 wt % of inorganics (NaCl, Na2HPO4
and/or NaH2PO4) were tolerated under these conditions. In the
absence of alkoxide, 8 slowly reacted with 623 to produce a
dimeric side-product with MS 343. Two possible structures for
MS 343 were postulated: N-arylated dimer 16 or symmetric
ether 17 (Scheme 6). The structure of the byproduct was
assigned as 16 based on NMR studies and additional data.24
The stability of the alkoxide solution in MTBE/hexanes
was assessed using a quantitative GC assay with dodecane
and MTBE as internal standards. No significant degrada-
tion was detected analytically in samples that had been
aged for 4 days at room temperature. Side-by-side
etherification reactions were performed with a freshly
prepared solution of the alkoxide and the same alkoxide
The subsequent synthetic step was to substitute the chiral
fragment for the newly introduced chloride in 8. Initially only
racemic 1,1,1-trifluoropropan-2-ol (()-9 was available, but
during our development work we were able to identify Julich/
Codexis and DSM as large-scale suppliers of (S)-9. This chiral
alcohol could be obtained in high enantiomeric purity (>99.0%
ee) by reduction of 1,1,1-trifluoroacetone using alcohol dehy-
drogenases,20 whole cell yeast,21 or a chiral ruthenium catalyst.22
For ease of isolation of the low-boiling liquid (bp 81 °C), the
alcohol is obtained as a solution in an organic solvent. For our
intended use, a solution of MTBE (35-64 wt % by 1H NMR)
was deemed appropriate, since MTBE is chemically inert under
most reaction conditions that can be envisioned for a nucleo-
philic aromatic substitution. After initial development work was
completed with (()-9, the switch to (S)-9 was attempted;
however, significant changes in the physicochemical properties
(crystallinity and solubility) of the intermediates resulted, which
raised the need to redesign their isolation as discussed later.
While the medicinal chemistry synthesis had employed
sodium hydride for the deprotonation of (()-9, handling and
safety issues made the use of a soluble base desirable. The
lithium alkoxide of (()-9 could be prepared in THF by
treatment of the alcohol (1.5 equiv) with 1.4 equiv of n-
butyllithium in hexane. Aromatic nucleophilic substitution of
the chloride of 8 (limiting reagent) proceeded cleanly at room
temperature over 1-2 days (g98% conversion), affording
racemic ether (()-10. After an aqueous workup and filtration
through a short plug of silica, the product could be isolated in
reasonable yield (66-76%). Without affecting purity, the
reaction time was reduced from 1-2 days to 2-3 h by heating
to reflux. Replacing THF by MTBE led to a further improve-
ment as the reaction rate in the MTBE/hexane system turned
out to be faster, reducing the reaction time to 1.0-1.5 h at reflux
(48-56 °C). In an optimized procedure, the reaction mixture
(approximately 12 vol of 2:1 v/v of MTBE/hexanes) was
quenched with a mixture of saturated aqueous sodium chloride
and ammonium chloride (1:1 v/v, 10 vol). After filtration, the
(20) (a) Rosen, T. C.; Feldmann, R.; Du¨nkelmann, P.; Daussmann, T.
Tetrahedron. Lett. 2006, 47, 4803. (b) Doderer, K.; Gro¨ger, H.; May,
O. Method for the production of enantiomerically enriched 1,1,1-
trifluoroisopropanol. Ger. Patent DE102005054282, 2005.
(21) (a) Bucciarelli, M.; Forni, A.; Moretti, I.; Torre, G. Synthesis 1983,
897. (b) Doswald, S.; Hanlon, S. P.; Kupfer, E. Asymmetric reduction
of 1,1,1-trifluoroacetone. U.S. Patent Appl. 2007/0009999, 2007.
(22) (a) Sterk, D.; Stephan, M.; Mohar, B. Org. Lett. 2006, 8, 5935. (b)
Puentener, K.; Waldmeier, P. Asymmetric hydrogenation of 1,1,1-
trifluoroacetone. WO2008012240, 2008.
(23) Usually present in 1-4% in typical batches of 8.
(24) Two mono-Suzuki products (MS 443) resulting from a coupling of
MS 343 with 7 were detected during the reaction which later converted
into a single bis-Suzuki product (MS 543). This observation is
consistent only with the unsymmetrical dichloride 16.
(19) Frey, L. F.; Marcantonio, K.; Frantz, D. E.; Murry, J. A.; Tillyer, R. D.;
Grabowski, E. J. J.; Reider, P. J. Tetrahedron Lett. 2001, 39, 6815.
234
•
Vol. 13, No. 2, 2009 / Organic Process Research & Development

Scheme 6. Formation of side product 16 as a result of the
undercharge of alkoxide
solution except for being aged at room temperature for a
day. The assay yield of (S)-10 in this set of experiments
was within experimental error (88-91%). Additional 19F
NMR experiments with R,R,R-trifluorotoluene as an
internal standard showed a partial degradation of the
alkoxide solution in the presence of 10 mol % excess
n-butyllithium over 24 h. In summary, the alkoxide
solution appears to be sufficiently thermally stable at room
temperature provided that excess n-butyllithium is promptly
destroyed using excess (S)-9.
Figure 4. An overlay of raw in-process spectra acquired to
monitor the conversion of (S)-9 into its lithium alkoxide during
kilogram-scale manufacturing.
PAT-Application for Monitoring Alkoxide Formation.
As previously described, the deprotonation of (S)-9 to generate
the lithium alkoxide was monitored at bench scale by the use
of an internal colorimetric indicator (1,10-phenanthroline).
However, concerns regarding contamination with phenanthro-
line decomposition products25 led to pursuit of an in situ
spectroscopic method which also served to eliminate the need
to titrate the n-BuLi. In situ Raman spectroscopy was ultimately
used to monitor the deprotonation of (S)-9 and concomitant
formation of the corresponding lithium alkoxide. On the basis
of reference spectra, features indicative of C-O stretching or
deformation at ∼780 cm-1 and ∼792 cm-1 were chosen as
strong markers for the alkoxide and alcohol, respectively. These
features were free from spectral interference by the reaction
solvent (MTBE). At bench scale, in situ spectra were acquired
by the use of a dispersive Raman spectrometer (Raman Rxn 1
Analyzer, Kaiser Optical Systems, Ann Arbor, MI) with incident
radiation at 785 nm coupled to a filtered probe head and
Hastelloy immersion optic. A custom-built Pilot E probe in
conjunction with a Raman Rxn 3 analyzer (Kaiser Optical
Systems, Ann Arbor, MI) was used for in-process control at a
kilogram scale. Reaction completion was determined by moni-
toring the running relative standard deviation of spectra acquired
at 1-min intervals. Figures 4 and 5 show the spectral region of
interest recorded as n-butyllithium was being charged. The
deprotonation of the alcohol as evidenced in the spectra at ∼792
cm-1 is kinetically fast, and no delay in the spectral changes
relative to addition was observed. Since an excess of n-
butyllithium resulted in side reactions in the subsequent step
and excess (S)-9 was well tolerated in the subsequent chemistry,
a small recharge of (S)-9 was performed to ensure that no excess
alkyllithium was present. The results obtained by in situ
Figure 5. Typical waterfall plot monitored during in situ
spectroscopic in process control.
Figure 6. Melting point of 10 depending on its enantiomeric
composition.
spectroscopy were ascertained by performing an off-line 1,10-
phenanthroline check on an aliquot of the alkoxide solution.
A revised workup and isolation protocol had to be developed
for the isolation of (S)-10, since it displayed a much lower
solubility compared to that of (()-10.26 Aqueous ethanol was
selected as the solvent for the isolation of 10 as good recovery
(>90% yield) and rejection of impurities (>99.0% purity) could
be ensured. In the optimized procedure, the alkoxide solution
was prepared at 10-25 °C by addition of 2.5 M n-butyllithium
in hexanes to a solution of (S)-9 in MTBE (1.45 equiv). The
deprotonation was monitored by Raman spectroscopy as shown
in Figure 4, and after reaching the equivalence point, an
additional amount of (S)-9 (0.10 equiv) was added. The
preformed alkoxide solution was added to a suspension of 8 in
(25) Decomposition of the charge-transfer complex into multiple unidenti-
fied products was detected.
(26) Solubilities of (S)-10 vs (()-10 at rt (respectively): 68 vs 133 mg/mL
in MTBE and 9 vs 14 mg/mL in EtOH/water (2:1 v/v).
Vol. 13, No. 2, 2009 / Organic Process Research & Development
•
235

Figure 7. XRPD-pattern of (()-10 and single enantiomers of 10.
MTBE, and the reaction mixture was heated to reflux with
complete conversion being observed within 2 h (<0.2% 8). The
reaction mixture was cooled to room temperature and diluted
with THF to assist complete dissolution of the product. After
quenching with an aqueous ammonium chloride solution, a
polish filtration was performed through a Celite pad to remove
insoluble inorganic and polymeric impurities. A crude organic
solution of (S)-10 (>95% assay yield) was obtained after phase
cut. A distillative solvent switch to ethanol was performed (<5.0
wt % THF, <1.0 wt % MTBE), and the product was crystallized
by addition of water at 55-65 °C followed by cooling to rt.
On filtration, (S)-10 was isolated in high purity (loss to mother
liquors: ∼ 8 mg/mL; 90-95% isolated corrected yield; >99.5%
purity).
Characterization of Solid-State Properties of 10. Although
(S)-9 of high enantiomeric purity had been procured, we
explored the possibility of using material of lower enantiomeric
excess for subsequent campaigns. The decreased solubility of
(S)-10 as compared to that of (()-10 offered a potential for an
ee upgrade at this stage. Applying the standard reaction and
isolation conditions, we were able to obtain (S)-10 of relatively
high purity (97.4% ee) from (S)-9 of 88.4% ee.
Since the procurement of (S)-9 of lower ee would be of
interest from a cost-of-goods perspective, we were interested
in defining the type of racemate (conglomerate, racemic
compound, or solid solution) that 10 forms.27 Solid samples
with different enantiomeric excesses were prepared by dissolu-
tion of varying amounts of (R)- and (S)-10 in ethanol followed
by evaporation. The recovered solids were analyzed by dif-
ferential scanning calorimetry and chiral HPLC. The melting
point curve (Figure 6) indicates a eutectic point close to a mole
fraction (x) of 0.5 (ee ) 0%). Interestingly, the melting point
of 10 shows only very small variation (<2.0 °C) over a
relatively wide range of x (0.37-0.62). (()-10 shows a distinct
Figure 8. Ternary phase diagram of enantiomers of 10 in
EtOH/water (2:1 v/v).
X-ray powder diffraction pattern (XRPD) compared to the
enantiopure compounds, (S)-10 and (R)-10 (Figure 7). Together
this data suggest that the racemate is either a conglomerate with
a polymorphic form different from the enantiomers of 10 or a
racemic compound with a eutectic point that is relatively close
to x ) 0.5. To differentiate between these two possibilities,
conversion experiments in which mixtures of (S)-10 and (()-
10 were slurried at elevated temperature followed by XRPD
analysis of the recovered solids were performed. The patterns
corresponded to those recorded for simple mixtures of the two
starting materials, suggesting that the racemate crystallizes as
a racemic compound. To confirm this hypothesis, a ternary
phase diagram (TPD) for the enantiomers of 10 in ethanol/water
(2:1) was constructed. Different mixtures of (R)- and (S)-10
were equilibrated in that solution for 24 h at room temperature,
and the solubility for the individual enantiomers was determined
by chiral HPLC as plotted in Figure 8. While deviations from
the expected perfect symmetry can be explained by the
variability of the solubility assay using the high-throughput
(27) (a) Chen, A. M.; Wang, Y.; Wenslow, R. M. Org. Process Res. DeV.
2008, 12, 271. (b) Coquerel, G. Enantiomers 2000, 5, 481. (c) Collet,
A. Enantiomers 1999, 4, 157.
236
•
Vol. 13, No. 2, 2009 / Organic Process Research & Development

Scheme 7. PTC conditions for etherification of 8 with (S)-9
Scheme 8. Suzuki-coupling and isolation of (()-11
method,28 the TPD shows that the eutectic point for 10 is at x
) 0.43, corresponding to an ee of 14%. This result clearly
demonstrates that the racemate of 10 is a racemic compound,
and it highlights that the TPD provides a more sensitive method
to differentiate conglomerate from racemic compound, com-
pared to the melting point diagram. The solid-state properties
of (S)-10 have made it an ideal candidate for upgrade of its
enantiomeric purity by crystallization as any mixture of 10 with
>14% ee can be upgraded to enantiopure material by proper
choice of crystallization conditions.
The ether formation under the predefined conditions per-
formed well on scale, but interest remained to conduct the
reaction without the use of an alkyllithium reagent. An attractive
alternative to the current process is the phase transfer catalysis
(PTC) etherification using alcohol (S)-9 and aqueous base. The
PTC conditions were successfully applied on laboratory scale
using 8, (S)-9 (1.3 equiv), and cat. Aliquat 100-S (aq tetra-n-
butylammonium bromide) in a biphasic mixture of MTBE and
10 N aq NaOH (Scheme 7). After 1.5 h at 55 °C the reaction
mixture showed >99.5% conversion of 8. Aqueous workup
followed by solvent switch from MTBE to ethanol and
crystallization from ethanol/water led to the isolation of (S)-10
in good yield (74%) and purity (99.5% purity). This procedure
provided an operationally simpler alternative to the originally
developed conditions.
Formation of Carbon-Carbon Bond through Suzuki-
Coupling. The Suzuki cross-coupling between aryl chloride
[(()-10] and boronic acid 7 was initially performed using
conditions closely resembling those of literature procedures.29
Compared to the synthesis of 1, this coupling reaction appeared
to be more challenging, and the newest generation of biphe-
nylphosphine ligands had to be utilized for successful results.
Complete conversion was observed using Pd2(dba)3/S-Phos and
K3PO4 in dioxane at reflux. The product (()-11 was obtained
in >95% yield. Subsequently, a number of bases (NaOH,
Na2CO3, K2CO3, K3PO4, KHCO3, dicyclohexylamine) and
solvents (mixtures of ethanol, 2-MeTHF, n-butanol, or 2-pro-
panol with water) were screened in order to (i) eliminate the
use of dioxane and (ii) minimize the protodeboronation of 7
leading to p-toluic acid. All inorganic bases led to substantial
protodeboronation, and deposition of palladium black on the
walls was observed prior to achieving complete conversion. In
contrast, reactions performed in the presence of dicyclohexy-
lamine were devoid of these two issues. Complete conversion
could be obtained within 4 h at 80 °C in a mixture of ethanol/
water (4:1 v/v) with minimal protodeboronation (<1%) and
palladium black formation. Product isolation was accomplished
by formation of a solution of sodium carboxylate using aq
NaOH, a polishing filtration, washing of the filtrate with IPAc/
heptane to remove dicyclohexylamine, and acidification of the
aqueous layer to afford crystalline (()-11 (98.4% purity, 93.4
wt %) in 92% yield (Scheme 8).
As was the case with intermediate (10), a much lower
solubility was observed for (S)-11 compared to (()-11.30
Moreover, the corresponding sodium salt [(S)-18] also had poor
aqueous solubility.31 The success of the previously developed
cross-coupling conditions for the racemic series [Pd2(dba)3,
S-Phos, dicyclohexylamine, aq EtOH] hinged upon an effective
two-stage workup: (i) filtration of insoluble byproducts followed
by (ii) extractive removal of the dicyclohexylamine freebase.
In particular this procedure relied upon a good aqueous
solubility of the sodium salt (()-18, which was not preserved
for (S)-18. A modification of the workup by using concentrated
NaOH (10 N) was attempted, allowing a complete dissolution
of the sodium carboxylate (S)-18 in the ethanol-rich aqueous
mixture.31 This solution was then extracted with heptane to
remove dicyclohexylamine. Partial partitioning of ethanol into
the heptane/dicyclohexylamine amine layer resulted in super-
saturation of the carboxylate and partial product precipitation,
which complicated the phase separations. Despite the very
favorable reaction profile with dicyclohexylamine as base, the
issues in the isolation of (S)-18 required a reevaluation of the
base for the Suzuki reaction.
A more extensive screen focused on base (Na2CO3, K2CO3,
Cs2CO3, NaHCO3, KHCO3, NaOAc, KOAc, K3PO4, K2HPO4,
NaOtBu, KOtBu, dicyclohexylamine)32 and solvent (ethanol,
1-pentanol, DMF, DMAc, NMP, toluene, methyl isobutyl
ketone)33 while using the same ligand and palladium source
[0.25 mol % Pd2(dba)3, 0.60 mol % S-Phos] as in the previous
studies. Only sodium carbonate in aq EtOH performed relatively
well when compared to the established dicyclohexylamine-
promoted conditions, with protodeboronation being the major
side reaction (complete conversion and up to 15 mol % p-toluic
(30) Solubilities of (S)-11 at rt: 6 mg/mL in EtOH, 55 mg/mL in THF, 0.5
mg/mL in water.
(31) Solubilities of the sodium carboxylate [(S)-18] at rt: >90 mg/mL
(EtOH), 61 mg/mL [EtOH/water 1:1 v/v], 3 mg/mL (water).
(32) The same amount of base was used in every reaction (4 equiv).
Dicyclohexylamine was used as a control.
(28) The complete data set for the solubility experiments can be found in
the Supporting Information.
(29) Barder, T. E.; Walker, S. D.; Martinelli, J. R.; Buchwald, S. L. J. Am.
Chem. Soc. 2005, 127, 4685.
(33) Organic solvent (8 vol) with water (2-3 vol).
Vol. 13, No. 2, 2009 / Organic Process Research & Development
•
237

Scheme 9. Suzuki-coupling and isolation of (S)-11
Scheme 10. Final amide bond formation for the synthesis
of (S)-2
acid). However, under these reaction conditions the poor
aqueous solubility of the sodium carboxylate [(S)-18] could be
used advantageously in the isolation. In fact (S)-18 was obtained
directly from the postreaction mixture by simple filtration and
p-toluic acid was rejected to the aqueous ethanol filtrates
(Scheme 9). In an optimized protocol, (S)-10 was coupled with
7 (1.2 equiv) in the presence of Pd2(dba)3 (0.25 mol %), 0.60
mol % S-Phos and Na2CO3 (4 equiv) in a mixture of ethanol
and water (4:1 v/v). Complete conversion was obtained after
15 h at 78 °C. Precipitation of the sodium salt [(S)-18] as a
crystalline solid was observed during cooling of the postreaction
mixture. Additional water was added to increase the recovery,
leading to isolation of (S)-18 in good yield and purity (100%
purity, 89 wt %, 94% yield). Isolation from water/ethanol (2:1
v/v) provided a good combination of impurity rejection (includ-
ing p-toluic acid and residual palladium) with good product
recovery (5-10 mg/mL in supernatant). As a result of a large
excess of a common countercation (Na+), the effective solubility
of (S)-18 was significantly lower than the solubility of (S)-18
in pure water/ethanol (2:1 v/v) (34 mg/mL). Meanwhile aryl
chloride (S)-10 was only sparingly rejected34 by this workup;
therefore, a high conversion [e0.5% (S)-10] in the Suzuki-
coupling was critical to achieve a high purity of the cross-
coupling product. Due to a significantly higher solubility of (()-
18 compared to that of (S)-18, the possibility for an upgrade of
the enantiomeric purity during the isolation from aq ethanol
was investigated. Starting from 10 of 97.4% ee, a significantly
upgraded carboxylate product [(S)-18, (99.8% ee)] was obtained
as expected.
acid, pyridinium hydrochloride, or trifluoroacetic acid) did not
lead to any significant improvement. Closer examination of the
solid-state properties of (S)-18 showed that it was isolated as a
hydrate with varying water content. Subjection of a batch of
(S)-18 containing 3.8 wt % water (∼1 equiv) to 1.05 equiv of
sulfuric acid and 2 equiv of CDI led to complete conversion,
showing that the moisture in (S)-18 was responsible for the
initially observed low conversion.
Considering the propensity of (S)-18 to form hydrates with
varying water content, we decided to introduce an additional
isolation step into the synthetic sequence (Scheme 10). In
contrast to its sodium salt, free acid (S)-11 was a crystalline
high-melting solid (mp 267-268 °C) with little propensity to
form hydrates or solvates during the crystallization. While direct
isolation of (S)-11 at the end of the Suzuki-coupling had been
challenging, the conversion of the sodium salt (S)-18 to the free
acid (S)-11 was straightforward. Dissolution of (S)-18 in ethanol
was followed by a polishing filtration, and crystallization by
addition of hydrochloric acid (1.1 equiv) led to a clean formation
of (S)-11. Dilute hydrochloric acid (0.5 N) was used for this
purpose to obtain a final solvent ratio that ensured high recovery
(>99% yield, >99.8% purity).35 No hydrolysis of the acid-
sensitive ether linkage in (S)-11 was observed under these
conditions. The introduction of this additional isolation point
also allowed a significant reduction in the palladium content.
Typical levels of palladium observed in (S)-18 directly after
the Suzuki reaction were 300-600 ppm, whereas (S)-11 was
isolated with 40-80 ppm Pd. This finding allowed avoiding
Formation of Carbon-Nitrogen Bond and Isolation of
(S)-2. All activation attempts through acid chloride formation
led to incomplete conversions when applied to the synthesis of
1. Activation with 1,1-carbonyldiimidazole (CDI) had, however,
been very successful, and we therefore refocused our develop-
ment work on the use of CDI for the amide-coupling in the
synthesis of (S)-2. Direct subjection of sodium carboxylate (S)-
18 to CDI activation (1.5 equiv) led to incomplete conversion
to the acylimidazolide, and addition of acidic additives (sulfuric
(34) A reaction at 75 °C reached 95% conversion [5% (S)-10] after 24 h.
No significant rejection of the starting material was observed during
the isolation of (S)-18.
(35) The solubility of 12 in a 1:1 v/v mixture of ethanol/water is 0.2 mg/
mL.
238
•
Vol. 13, No. 2, 2009 / Organic Process Research & Development

Scheme 11. Optimized synthetic sequence for the synthesis of (S)-2
the use of expensive scavenging reagents for the removal of
palladium from the API.
Summary
A concise synthesis of the p38 MAP kinase inhibitor (S)-2
was developed. The overall sequence involves four synthetic
steps from 6-chlorophthalazin-1-ol (6) and proceeds in 45%
overall yield (Scheme 11). This compares favorable with the
33% overall yield from the more advanced building block
2-chloro-6-bromophthlazine (3) in the original synthesis. The
developed sequence involves highly efficient reactions for the
key constructions of carbon-oxygen, carbon-carbon and
carbon-nitrogen bonds. VCD was utilized to correctly assign
(S)-configuration to the desired enantiomer of 2 prior to initiation
of synthetic efforts. PAT (in situ Raman spectroscopy) was
implemented to monitor the formation of a lithium alkoxide
on large scale.
Activation of (S)-11 with CDI (1.2 equiv) in THF proceeded
without issue. The intermediate acylimidazolide [(S)-19] was
hydrolytically unstable under HPLC conditions (0.1% TFA in
acetonitrile/water), but the activation step was monitored by
quenching aliquots of the reaction mixture with benzylamine
and analyzing quantitatively the resulting benzylamide [(S)-20].
The activation proceeded well at rt (>95% conversion), but
heating at 50 °C for 1 h was required to achieve complete
conversion of (S)-11 [<0.5% relative to (S)-20]. Reaction of
the activated intermediate with cyclopropylamine was facile,
and complete conversion to (S)-2 [<0.5% (S)-11] was observed
within 1 h at rt. The isolation of (S)-2 was attempted by addition
of an antisolvent, but was complicated by the propensity of the
system for formation of long cylindrical needles and occasion-
ally that of cotton-like agglomerates. Prolonged time for stirring
of the reaction mixture after crystallization led to breaking of
the needles, thereby significantly improving the agitation of the
mixture. Nonetheless, a balance between product recovery and
mechanical properties of the crystallization mixture had to be
found. Under optimized conditions, the final solvent composi-
tion of tetrahydrofuran/2-butanol and water in a 7:1:6 (v/v/v)
ratio was utilized. (S)-2 could be isolated from this mixture in
acceptable yield (72%, 22.8 mg/mL loss to mother liquors) and
very high purity (>99.9%). More in-depth analysis of the
solubility data of (S)-2 in the ternary mixture of tetrahydrofuran/
2-butanol and water showed that a moderate decrease in THF
content in the final mixture could potentially be used to
significantly improve the recovery (up to 20%).36 This procedure
was not adopted for the initial delivery due to concern about
the stirrability of the crystallization mixture, but it could
potentially be implemented in the design of a more controlled
crystallization.37
Experimental Section
1,6-Dichlorophthalazine (8). 6-Chlorophthalazin-1-ol (6)
(2.50 kg, 13.9 mol) and anhydrous acetonitrile (<300 ppm
water, 5.8 L) were charged to a 100 L reactor (“reaction
vessel”). The jacket temperature was set at 20 °C, and agitation,
at 76 rpm. Phosphorus oxychloride (1.94 L, 20.8 mol, 1.50
equiv) was charged to the reaction vessel, and the addition
equipment was rinsed with acetonitrile (1.7 L). Reactor contents
were then heated to 80 ( 5 °C. After 2 h, a sample was analyzed
for reaction completion showing 1.0 HPLC area % of 6
remaining (target e2.0%). The reaction vessel contents were
cooled to 20 °C over 1.5 h. Sodium phosphate monobasic
monohydrate (8.0 g) and sodium phosphate dibasic anhydrous
(21.1 g) were dissolved in purified water (15.5 L), and the
solution was charged into a 100 L reactor/receiver (“quench
vessel”) fitted with a pH probe. The reaction mixture was slowly
transferred from the reaction vessel into the quench vessel via
peristaltic pump, while maintaining the pH at 6.8-8.0 and the
batch temperature e30 °C. pH adjustment was achieved by
simultaneously charging the 2 N NaOH solution during the
reaction mixture transfer. A total of 37.0 L of the 2 N NaOH
solution was charged over 1.25 h; the final pH was 7.6.
Acetonitrile (5.0 L) was charged to the reaction vessel as a rinse
and then transferred into the quench vessel. An additional 2.5
(36) Solubilities of 2: 23.6 mg/mL in THF/2-butanol/water (50.0/7.2/42.8
v/v/v) and 3.7 mg/mL in THF/2-butanol/water (37.5/8.9/53.6 v/v/v).
For additional solubility data in this solvent system see Supporting
Information.
(37) A crystallization combined with wet milling should improve the
handling properties of (S)-2.
Vol. 13, No. 2, 2009 / Organic Process Research & Development
•
239

L of the 2 N NaOH solution was charged over 15 min; the
final pH was 7.5. The quench vessel contents were cooled to
20 °C, and agitation was increased to 120 rpm. The contents
of the quench vessel were filtered through a 25 µm polypro-
pylene filter cloth, and the cake was rinsed twice with purified
water (2 × 25.0 L). The cake was left under nitrogen blanket
and slight vacuum for 4 h. The wet cake was transferred to a
vacuum oven and was dried under vacuum with a dry nitrogen
sweep at 25 °C until loss on drying (LOD) was e1.0% between
weighings to afford 8 as an orange crystalline material (2.59
kg, 83.0 wt %, 96.7% purity, 78% yield). The analytical data
were in agreement with published data.5
final batch volume was ∼20 L. Ethanol (21.5 L) was charged
to the reactor, and the reactor jacket temperature was increased
to 80 °C. Twenty-two liters of the contents was distilled off
(4.75 psia), and the final batch volume was ∼21 L. Ethanol
(10.5 L) was charged to the reactor, and 12 L was distilled (3.5
psia) off to a final batch volume of ∼20 L. The reactor contents
were heated to 64 °C. Purified water (10.5 L) was slowly added
over 30 min while maintaining a batch temperature of 60 ( 5
°C. Reactor contents were agitated at 250 rpm for 10 min and
then cooled to 21 °C over 4 h. After ∼8 h at 21 °C, the contents
of the reactor were filtered through a 25 µm polypropylene filter
cloth. The cake was rinsed twice with a mixture of ethanol (2.1
L) and purified water (8.4 L). The cake was left under nitrogen
blanket and slight vacuum for 5 h. The wet cake was transferred
to a vacuum oven and dried under vacuum and nitrogen sweep
at 55 °C for ∼48 h to afford 10 as yellow crystals (2.73 kg,
94.8 wt %, 99.6% purity, 90% yield). Mp 135-137 °C; IR
1616, 1583, 1552, 1474, 1457, 1425, 1415, 1398, 1387, 1330,
1274, 1215, 1179, 1170, 1146, 1113, 1072, 1026, 918, 896,
848, 818, 757, 687, 655 cm-1; 1H NMR (DMSO-d6, 400 MHz)
δ 9.42 (1H, s), 8.34 (1H, d, J ) 1.9 Hz), 8.16 (1H, d, J ) 8.8
Hz), 8.07 (1H, d, J ) 8.8, 1.9 Hz), 6.35-6.22 (1H, m), 1.63
(3H, d, J ) 6.4 Hz); 13C NMR (DMSO-d6, 101 MHz) δ 157.8,
148.3, 137.8, 133.7, 129.7, 125.8, 124.5, 124.4 (1JCF ) 281
Hz), 117.0, 68.9 (2JCF ) 32 Hz), 13.2; HRMS (m/z): [M +
H+] calcd for (C11H9ClF3N2O): 277.0350; Found: 277.0347.
Anal. Calcd for C11H8ClF3N2O: C, 47.76; H, 2.91; N, 10.13.
Found: C, 47.66; H, 2.85; N, 10.14.
Sodium (S)-4-Methyl-3-[1-(2,2,2-trifluoro-1-methylethox-
y)phthalazin-6-yl]benzoate [(S)-(18)]. (S)-6-Chloro-1-(1,1,1-
trifluoropropan-2-yloxy)phthalazine [(S)-10] (94.8 wt %, 2.73
kg, 9.37 mol), 3-carboxy-6-methylphenylboronic acid (7) (92.9
wt %, 2.36 kg, 12.2 mol), sodium carbonate (3.97 kg, 37.5 mol),
and ethanol (21.0 L) were charged to a 100 L reactor. Purified
water (5.2 L) was charged, and the reactor was vacuum purged
twice by evacuating the vessel to ∼11 psia and then bringing
the pressure up to ∼15.5 psia with nitrogen. 2-Dicyclohexy-
lphosphino-2′,6′-dimethoxy-1,1′-biphenyl (S-Phos, 23.1 g, 0.056
mol) and tris(dibenzylideneacetone)dipalladium(0) (21.4 g,
0.023 mol) were charged as slurries in ethanol (0.75 L each) to
the reactor. The reactor was vacuum purged twice by evacuating
the vessel to ∼11 psia and then bringing the pressure up to
∼15.5 psia with nitrogen. The reaction mixture was heated to
80 ( 10 °C and agitated at 133 rpm. After ∼17 h, a sample
was analyzed for reaction completion showing e0.05 HPLC
area % of (S)-10 remaining (target e0.50%). The reactor
contents were cooled to 65 °C and purified water (37.0 L) was
charged to the reactor while maintaining a batch temperature
g48 °C. The reactor contents were then cooled to 24 °C over
2.5 h with agitation at 180 rpm. The contents of the reactor
were filtered through a 25 µm polypropylene filter cloth. The
cake was rinsed twice with purified water (2 × 13.0 L). The
cake was left under nitrogen blanket and slight vacuum for 4.5
days to afford (S)-18 as an off-white crystalline material (4.98
kg, 67.1 wt %, 99.9% purity, 90% yield). Mp 90-92 °C; IR
3288, 1612, 1582, 1543, 1437, 1390, 1333, 1169, 1153, 1110,
1081, 1027, 909,7, 848, 785, 677 cm-1; 1H NMR (DMSO-d6,
400 MHz) δ 9.48 (1H, s), 8.20 (1H, d, J ) 8.4 Hz), 8.16 (1H,
(S)-6-Chloro-1-(2,2,2-trifluoro-1-methylethoxy)phthala-
zine [(S)-(10)]. (S)-1,1,1-Trifluoropropan-2-ol [(S)-9] in MTBE
(91.3 wt %, 1.92 kg, 15.4 mol) and anhydrous MTBE (21.1 L)
were charged to a 100 L reactor (alkoxide vessel). The reaction
mixture was cooled to 10 ( 5 °C and agitated at 104 rpm.
n-Butyllithium in hexanes (2.5 M, 23 wt %) was slowly added,
while maintaining a batch temperature e25 °C (maximum temp
18 °C). During the n-butyllithium addition, the reaction was
monitored by in situ Raman spectroscopy. Complete disap-
pearance of the characteristic alcohol band was observed after
addition of 4.00 kg of 23 wt % solution of n-butyllithium in
hexanes. As confirmation, a small sample was subsequently
taken and added into a solution of 1,10-phenanthroline in THF,
resulting in a brown indicator solution (excess n-butyllithium
detected). Additional (S)-9 (91.3 wt %, 196 g, 1.57 mol) was
charged to quench any excess n-butyllithium in the system. After
the second charge, the alcohol peak was again visible by Raman
spectroscopy, and the 1,10-phenanthroline test resulted in a
colorless indicator solution (no n-butyllithium detected). The
resulting alkoxide solution was left stirring in the reactor at e10
°C. 1,6-Dichlorophthalazine (8) (83.0 wt %, 2.54 g, 10.6 mol)
and anhydrous MTBE (8.5 L) were charged to a second 100 L
reactor (reaction vessel). The reactor jacket temperature was
set at 20 °C, and agitation was set at 64 rpm. The alkoxide
solution was then transferred from the alkoxide vessel to the
reaction vessel over 15 min; final temperature after the addition
was 15 °C. Anhydrous MTBE (1.0 L) was charged as a rinse
to the alkoxide vessel and was subsequently transferred to the
reaction vessel. Once the alkoxide addition was complete, the
reactor contents were heated to 45 °C and stirred at 154 rpm.
At 1.5 h of elapsed reaction time, a sample analyzed for reaction
completion showed 0.29 HPLC area % of starting material 8
remaining (target e0.50%). Reactor contents were cooled to
24 °C, and anhydrous THF (10.5 L) was charged to the reactor.
A solution of ammonium chloride (0.57 kg) in purified water
(4.2 L) was slowly charged to the reactor in small increments
over 6 min to control concomitant NH3 gas evolution. The
reactor contents were agitated at 156 rpm for 13 min and then
allowed to separate overnight. The contents of the reaction
vessel were filtered through a tightly packed pad of Celite 521
(2.0 kg). The pad was rinsed with THF (2.0 L). The biphasic
filtrate was collected in a second reactor, heated to 40 °C,
agitated for 10 min, and allowed to separate over 30 min. The
aqueous layer and the rag layer were discarded. The reactor
jacket temperature was set to 70 °C, and ∼30 L of organic
distillate was collected under reduced pressure (9.75 psia); the
240
•
Vol. 13, No. 2, 2009 / Organic Process Research & Development

s), 8.05 (1H, dd, J ) 8.4, 1.5 Hz), 7.87 (1H, dd, J ) 8.0, 1.9
Hz), 7.84 (1H, s), 7.27 (1H, d, J ) 8.2 Hz), 6.37-6.25 (1H,
m), 2.26 (3H, s), 1.64 (3H, d, J ) 6.4 Hz); 13C NMR (DMSO-
d6, 101 MHz) δ 169.1, 158.0, 149.3, 146.8, 138.9, 138.2, 135.0,
134.6, 130.6, 129.6, 129.1, 128.9, 126.2, 124.6 (1JCF ) 281
Hz), 121.7, 117.1, 68.7 (2JCF ) 32 Hz), 19.9, 13.3; HRMS (m/
z): [M + H+] calcd for (C19H15F3NaN2O3): 399.0927; Found:
399.0917.
(3.0 L) was charged to the reactor as a rinse and then transferred
through the same 5 µm polypropylene filter into the second
reactor. Reactor agitation was set at 131 rpm and jacket
temperature at 20 °C. Cyclopropylamine (1.12 L, 16.2 mol)
was charged to the reactor over 18 min while maintaining a
batch temperature e30 °C. After 1 h a sample analyzed for
reaction completion showed 0.26 HPLC area % of (S)-11
remaining (target e0.50%). A solution of 2-butanol (3.0 L) and
purified water (18.0 L) were charged to the reactor over 30
min; agitation was set at 180 rpm. The contents of the reactor
were filtered through a 10 µm polypropylene filter cloth. The
wet cake was washed with three portions of a THF/water
mixture (1/3 v/v, 28.0 L total). The cake was left under nitrogen
blanket and slight vacuum for 19 h. The wet cake was
transferred to a vacuum oven and dried under vacuum at 50
°C for ∼48 h to afford (S)-2 as crystalline white solid (2.37
kg, 100.1 wt %, 100.0% purity, 72% yield). Mp 200-202 °C;
IR 1651, 1539, 1492, 1432, 1392, 1268, 1184, 1153, 1127,
1109, 1081, 1024, 849, 763, 677 cm-1; 1H NMR (DMSO-d6,
400 MHz) δ 9.48 (1H, s), 8.46 (1H, d, J ) 3.9 Hz), 8.25-8.20
(2H, m), 8.09 (1H, dd, J ) 8.4, 1.4 Hz), 7.84 (1H, dd, J ) 7.8,
1.6 Hz), 7.81 (1H, s), 7.46 (1H, d, J ) 8.0 Hz), 6.38-6.25
(1H, m), 2.89-2.81 (1H, m), 2.30 (3H, s), 1.64 (3H, d, J )
6.5 Hz), 0.73-0.66 (m, 2H), 0.59-0.54 (m, 2H); 13C NMR
(DMSO-d6, 101 MHz) δ 166.8, 158.0, 149.2, 145.5, 139.3,
138.3, 134.5, 132.3, 130.6, 128.8, 128.4, 127.1, 126.5, 124.6
(1JCF ) 281 Hz), 121.8, 117.3, 68.7 (2JCF ) 32 Hz), 23.0, 20.0,
13.3, 5.7; HRMS (m/z): [M + H+] calcd for (C22H21F3N3O2):
416.1580; Found: 416.1574.
(S)-4-Methyl-3-[1-(2,2,2-trifluoro-1-methylethoxy)ph-
thalazin-6-yl]benzoic acid [(S)-(11)]. Sodium (S)-4-methyl-
3-(1-(1,1,1-trifluoropropan-2-yloxy)phthalazin-6-yl)benzoate [(S)-
18] (4.98 kg) and ethanol (25.0 L) were charged to a reactor.
Agitation at 100 rpm was applied until complete dissolution
was achieved. A sample of the solution analyzed for the net
content of (S)-18 showed 3.34 kg (8.38 mol) of (S)-18 in
solution. The direct determination of the potency of solid (S)-
18 was challenging due to sampling heterogeneity; the acid
charge was therefore based on the assay in solution. The solution
was filtered through a 5 µm polypropylene filter into a second
reactor. The reactor jacket temperature was set at 20 °C and
agitation at 158 rpm. While the batch temperature was
maintained at e30 °C, 0.5 N HCl (18.4 L, 9.20 mol) was slowly
charged to the reactor. The reactor contents were then cooled
to 20 ( 5 °C and stirred at 156 rpm. After aging for 1 h the
reactor contents were filtered through a 25 µm polypropylene
filter cloth. The cake was rinsed three times with a mixture of
ethanol/water (1:1, 13.0 L). The cake was left under nitrogen
blanket and slight vacuum for 16 h. The wet cake was
transferred to a vacuum oven and dried under vacuum and
nitrogen sweep at 55 °C for ∼48 h to afford (S)-11 as a
crystalline white solid (3.18 kg, 100.8 wt %, 100.0% purity,
100% yield). Mp 268-270 °C; IR 3352, 1627, 1574, 1402,
1390, 1300, 1268, 1257, 1238, 1182, 1156, 1077, 1028, 842,
763, 703, 679 cm-1; 1H NMR (DMSO-d6, 400 MHz) δ 13.01
(1H, s), 9.47 (1H, s), 8.24-8.20 (2H, m), 8.08 (1H, dd, J )
8.6, 1.4 Hz), 7.95 (1H, dd, J ) 8.0, 1.8 Hz), 7.87 (1H, s), 7.52
(1H, d, J ) 8.0 Hz), 6.36-6.25 (1H, m), 2.36 (3H, s), 1.64
(3H, d, J ) 6.5 Hz); 13C NMR (DMSO-d6, 101 MHz) δ 166.9,
158.0, 149.2, 145.1, 140.4, 139.7, 134.3, 131.0, 130.4, 129.1,
128.9, 126.5, 125.1 (1JCF ) 281 Hz), 121.9, 117.4, 68.7 (2JCF
) 32 Hz), 20.2, 13.2; HRMS (m/z): [M + H+] calcd for
(C19H16F3N2O3): 377.1108; Found: 377.1098.
(S)-N-Cyclopropyl-4-methyl-3-[1-(2,2,2-trifluoro-1-meth-
ylethoxy)phthalazin-6-yl]benzamide [(S)-(2)]. (S)-4-Methyl-
3-(1-(1,1,1-trifluoropropan-2-yloxy)phthalazin-6-yl)benzoic acid
[(S)-(11)] (3.01 kg, 7.99 mol), imidazole (274 g, 4.02 mol) and
THF (18.0 L) were charged to a reactor. Jacket temperature
was set at 20 °C and agitation at 122 rpm. 1,1′-Carbonyldiimi-
dazole (CDI, 1.56 kg, 9.64 mol) was slowly charged to the
reactor in portions to control gas evolution, and the reactor
contents were then heated to 50 ( 5 °C. After 1 h at
temperature, a sample was quenched with benzylamine and
analyzed for reaction completion, showing 0.30 HPLC area %
of (S)-11 remaining (target e0.50%). The reactor contents were
cooled to 22 °C, and the reaction solution was transferred
through a 5 µm polypropylene filter into a second reactor. THF
Acknowledgment
We thank Dr. M. J. Martinelli, Dr. M. M. Faul, Dr. A. S.
Tasker, Dr. D. Zhang, and Dr. L. Pettus for valuable discussions
on the synthesis of the target compound. Dr. J. Preston, Dr. T.
Yan, J. Chen, C. Scardino, Dr. P. Grandsard, Dr. T. Wang, B.
Shaw, Dr. B. Shen, Dr. J. Ostovic, M. Petkovic, M. Ronk, Dr.
K. Turney, and M. Wacker are thanked for analytical support.
Dr. P. Schnier and R. Stanton are gratefully acknowledged for
providing the spectral convolution routine. Dr. A. Gore and Dr.
K. Nagapudi are thanked for help in solid-state characterization.
J. Manley, J. Clare, Dr. K. McRae, J. Tvetan, J. Tomaskevitch,
S. Huggins, B. Southern, and G. Sukay are thanked for support
and valuable input during scale-up.
Supporting Information Available
1
Copies of H NMR and 13C NMR spectra for compounds
(S)-2, (S)-10, (S)-11, 16, (S)-18 and (S)-20; additional experi-
mental procedures and characterization data for compounds (S)-
13, (S)-9, (S)-10, and (S)-20; additional details concerning the
VCD experiments, the experiments for generation of the ternary
phase diagram of 10, and solubility data for (S)-2. This material
is available free of charge via the Internet at http://pubs.acs.org.
Received for review October 1, 2008.
OP800250V
Vol. 13, No. 2, 2009 / Organic Process Research & Development
•
241