Dimethyltitanocene: From millimole to kilomole

Article abstract of DOI:10.1021/op034180j

The process development of a dimethyltitanocene-mediated ester olefination is described. The synthetic challenges and stability issues involving large-scale production of dimethyltitanocene are documented, and the optimization of the ester olefination is detailed. The process described was used to make hundreds of kilograms of an advanced intermediate for aprepitant (Emend).

Full text of DOI:10.1021/op034180j

Organic Process Research & Development 2004, 8, 256−259
Dimethyltitanocene: From Millimole to Kilomole
Joseph F. Payack,*,† Mark A. Huffman,*,† Dongwei Cai,† David L. Hughes,† Paul C. Collins,‡ Brian K. Johnson,‡
Ian F. Cottrell,† and Linda D. Tuma‡
Departments of Process Research and Chemical Engineering, Merck & Co. Inc., P.O. Box 2000,
Rahway, New Jersey 07065-0900, U.S.A.
Scheme 1
Abstract:
The process development of a dimethyltitanocene-mediated
ester olefination is described. The synthetic challenges and
stability issues involving large-scale production of dimethylti-
tanocene are documented, and the optimization of the ester
olefination is detailed. The process described was used to make
hundreds of kilograms of an advanced intermediate for aprepi-
tant (Emend).
Aprepitant (Emend) is a substance P antagonist that was
recently approved in the United States as a therapy to prevent
chemotherapy-induced nausea and vomiting.1 While several
alternative syntheses have been described,2 drug supplies to
support the clinical program were produced via a route that
employed dimethyltitanocene to convert an ester into a vinyl
ether (Scheme 1).3 The large amounts of bulk drug needed
during development necessitated the production of hundreds
of kilograms of dimethyltitanocene (DMT). This contribution
details the efforts to make the preparation and use of this
reagent safe and reliable from benchtop to 100-kg scale.
Petasis4 developed dimethyltitanocene as a convenient
alternative to the Tebbe reagent5 and the Grubbs metalocy-
clobutane analogue,6 for the olefination of esters, ketones,
and amides. The advantages of DMT over the earlier reagents
were ease of preparation, absence of Lewis-acidic aluminum
byproducts, and importantly for the organic synthetic chem-
ist, air and water stability. DMT is very effective at small-
scale olefination transformations; indeed, the reaction de-
picted in Scheme 1 proceeded in near quantitative yield.
Scheme 2
* To whom correspondence should be addressed. E-mail: joseph_payack@
merck.com, mark_huffman@merck.com.
While the Petasis reagent is extremely effective at ester
† Department of Process Research.
‡ Department of Chemical Engineering Research and Development.
(1) Navari, R.; Reinhardt, R. R.; Gralla, R. J.; Kris, M. G.; Hesketh, P. J.;
Khojasteh, A.; Kindler, H.; Grote, T. H.; Pendergrass, K.; Grunberg, S.
M.; Carides, A. D.; Gertz, B. J. N. Engl. J. Med. 1999, 340, 190.
(2) Brands, K. M. J.; Payack, J. F.; Rosen, J. D.; Nelson, T. D.; Candelario,
A.; Huffman, M. A.; Zhao, M. M.; Li, J.; Craig, B.; Song, Z. J.; Tschaen,
D. M.; Hansen, K.; Devine, P. N.; Pye, P. J.; Rossen, K.; Dormer, P. G.;
Reamer, R. A.; Welch, C. J.; Mathre, D. J.; Tsou, N. N.; McNamara J. M.;
Reider, P. J. J. Am. Chem. Soc. 2002, 125, 2129. Zhao, M. M.; McNamara,
J. M.; Ho, G.-J.; Emerson, K. M.; Song, Z. J.; Tschaen, D. M.; Brands, K.
M. J.; Dolling, U.-H.; Grabowski, E. J. J.; Reider, P. J.; Cottrell, I. F.;
Ashwood, M. S.; Bishop, B. C. J. Org. Chem. 2002, 67, 6743. Pye, P. J.;
Rossen, K.; Weissman, S. A.; Maliakal, A.; Reamer, R. A.; Ball, R.; Tsou,
N. N.; Volante, R. P.; Reider, P. J. Chem. Eur. J. 2002, 8, 1372.
(3) Hale, J. J.; Mills, S. G.; MacCoss, M.; Finke, P. E.; Cascieri, M. A.;
Sadowski, S.; Ber, E.; Chicci, G. G.; Kurtz, M.; Metzger, J.; Eierman, G.;
Tsou, N. N.; Tattersall, F. D.; Rupniak, N. M. J.; Williams, A. R.; Rycroft,
W.; Hargreaves, R.; MacIntyre, D. E. J. Med. Chem. 1998, 41, 4607.
(4) Petasis N. A.; Bzowej, E. I. J. Am. Chem. Soc. 1990, 112, 6392. Petasis N.
A.; Lu, S.-P. Tetrahedron Lett. 1995, 36, 2393.
olefinations, a series of drawbacks rapidly became apparent
as scale-up was contemplated. DMT is not stable in the solid
state,7 and a crystalline mass of the material would decom-
pose, releasing heat and gas. In addition, the molecule is
intrinsically unstable at synthetically relevant temperatures
and concentrations, raising considerable safety concerns.8
Since two moles of DMT are needed per mole of ester
(Scheme 2),9 not only are large amounts of the reagent
needed for pilot-scale production, but an efficient method
for removing the titanocene residues also had to be devel-
(6) Grubbs, R. H.; Tumas, W. Science 1989, 243, 907.
(7) Alt, H. G.; di Santo, F. P.; Rausch, M. D.; Uden, P. C. J. Organomet. Chem.
1976, 107, 257.
(8) Erskine, G. J.; Wilson, D. A.; McGowan, J. D. J. Organomet. Chem. 1976,
114, 119. Erskine, G. J.; Hartgerink, J.; Weinberg, E. L.; McCowan, J. D.
J. Organomet. Chem. 1979, 170, 51.
(5) Tebbe, F. N.; Parshall, G. W.; Reddy, G. S. J. Am. Chem. Soc. 1978, 100,
3611. Pine, S. H.; Zahler, R.; Evans, D. A.; Grubbs, R. H. J. Am. Chem.
Soc. 1980, 102, 3270.
(9) Hughes, D. L.; Payack, J. F.; Cai, D.; Verhoeven, T. R.; Reider, P. J.
Organometallics 1996, 15, 663.
256
•
Vol. 8, No. 2, 2004 / Organic Process Research & Development
10.1021/op034180j CCC: $27.50 © 2004 American Chemical Society
Published on Web 03/04/2004

oped. These and other issues were addressed during the
development of the synthetic process for DMT production
and the refinement of the olefination reaction.
The improved protocol was employed in multiple kilogram-
scale runs without incident. Before introduction into pilot-
plant scale however, operational hazard testing revealed a
substantial risk of unsafe operating conditions during the
concentration of DMT. The potential for a large exotherm
with substantial pressure increase was considered unaccept-
able on the multikilogram scale. It is believed that in the
absence of a substrate, the carbene (5) formed from heating
DMT reacts with additional DMT, leading to heat and gas
evolution.8 This issue was addressed by adding the ester
substrate to the titanocene reagent before concentration, so
that any carbene formed would be trapped in the desired
reaction. The process was successfully introduced into the
pilot plant.
In contemplating longer-term development, the limitations
of the first-generation pilot-plant protocol became apparent.
Two main issues required resolution: the workup of the
dimethyltitanocene-formation reaction mixture, which gave
emulsions that lowered yields to 80%, and a relatively narrow
end of reaction window for the olefination. Additionally, the
production cost issue took on greater importance as factory
introduction approached.
The emulsion problem encountered in the DMT workup
was caused by the presence of gummy solids, which
entrained DMT-containing organic phase into the aqueous
cut. While this was not a serious issue on kilo scale, where
it was convenient to observe the phase interface in a glass
vessel, in larger vessels the separation became difficult.
The solids issue was resolved by filtering the DMT
mixture prior to phase separation. To the quenched aqueous
organic mixture of DMT was added Celite (diatomaceous
earth), and the entire batch was filtered through a filter press.
The resulting clear organic and aqueous phases were easily
separated, and the organic phase was worked up with
minimal losses. To improve volumetric efficiency, the initial
reaction solvent was changed to THF (doubling the concen-
tration), while the quench mixture contained toluene, Celite,
and aqueous ammonium chloride. The largest runs using this
protocol produced 235 kg (87%) of dimethyltitanocene per
batch.
The end of reaction window was more challenging. While
the olefination reaction typically went to 98% assay yield
within 5 h at 80 °C, the yield would then drop by as much
as 10% if heating continued for a further 30 min. The time
required to sample, assay, and cool a large-scale reaction
made catching the end of reaction difficult. While we
considered in situ reaction monitoring, ultimately we devel-
oped a chemical solution.
The literature procedure for making DMT10 consisted of
treating a slurry of titanocene dichloride in MTBE/toluene
with methyllithium in ether. Then, after aqueous workup,
the material was isolated as an orange crystalline solid by
evaporation to dryness. The reagent was then redissolved in
THF or toluene and used in the desired olefination reaction.
Probe-scale investigations showed that the solid DMT would
decompose in minutes upon reaching dryness, turning black
with gas evolution. A protocol was developed where the
material was kept in solution throughout workup, then the
solution was dried with sodium sulfate, and carefully solvent-
switched and concentrated to a ∼20 wt % solution. This
produced the first 1.5 kg of DMT, which was immediately
used to olefinate 1 kg of ester. The vinyl ether was isolated
via an extremely tedious trituration with hexane, then the
remaining titanocene residues were decomposed with hy-
drogen peroxide.
The problems uncovered during the first kilogram-scale
run spurred considerable efforts into improving the DMT
synthesis and ester olefination. The first issue addressed was
replacement of the hazardous methyllithium in ether with
methyl Grignard reagent in THF. The reaction was unaffected
by the change in methylating reagent; however, the aqueous
workup of the DMT formed from MeMgCl proved prob-
lematic. It was discovered that if water was added to a
mixture of DMT and MgCl2, substantial amounts of the DMT
reverted back to titanocene dichloride and methyltitanocene
chloride. The issue was addressed by doing a reverse quench
of the DMT/MgCl2 reaction mixture into buffered water.
Optimization led to the reaction being run in toluene, with
a workup employing 6% ammonium chloride. The DMT
solution was then dried by azeotropic vacuum distillation
and concentrated to 15-20 wt %.
While optimizing the preparation of DMT, we noticed
that reagent samples containing small amounts of methylti-
tanocene chloride (Cp2TiMeCl) gave cleaner and higher-
yielding olefination reactions. Even more dramatic was the
cleaner formation of the oxo-bridged titanocene dimer
byproduct 6 (Scheme 2), which would become more
important as we contemplated recycling the titanium (see
below). These observations led to a procedure in which a
few mole percent of titanocene dichloride was added to the
DMT solution, giving methyltitanocene chloride by metath-
esis, prior to the olefination reaction.
Finally, an efficient titanocene removal protocol was
needed before further scale-up could be done. It was found
that heating the spent reaction mixture with aqueous metha-
nol decomposed the titanium to an easily filterable solid.
To protect the vinyl ether from acid-catalyzed decomposition,
and to prevent polymerization of cyclopentadiene, sodium
bicarbonate was added to maintain a slightly basic environ-
ment. After removal of the Ti residues and bicarbonate, the
product was isolated from n-propanol/water as an off-white
crystalline solid.
The source of the problem was the excess DMT required
to complete the reaction within a reasonable time. When the
ester substrate is present, the titanium carbene reacts with it
in a controlled fashion. Once the ester is consumed, the
carbene reacts by other pathways which decompose the vinyl
ether product. What was needed was another reagent to take
over when the ester 1 was consumed, providing an alternative
reaction pathway. Ideally this reagent should react more
slowly than 1 so that it does not interfere with the desired
reaction.
(10) Claus, K.; Bestian, H. Ann. 1962, 654, 8.
Vol. 8, No. 2, 2004 / Organic Process Research & Development
•
257

THF was obtained with a moisture content of <100 ppm,
or it was dried with molecular sieves before use. Ethanol
was denatured with 5% toluene. All operations were carried
out under an atmosphere of dry nitrogen.
Figure 1. Relative reactivity of esters toward 5.
(2R,3S)-4-Benzyl-2-({1-[3,5-bis(trifluoromethyl)phenyl]-
vinyl}oxy)-3-(4-fluorophenyl)morpholine (2). In a 1000-
gal glass-lined vessel, a slurry of titanocene dichloride (334
kg, 1.35 kmol) in THF (1000 L) was chilled to -5 to -10
°C, then methylmagnesium chloride (1020 kg, 3 M in THF,
3.03 kmol) was charged over 5.5 h, maintaining the tem-
perature below 5 °C. The Grignard reagent was rinsed in
with additional THF (17 L), and the mixture was stirred at
-1 to -9 °C for 1.5 h, during which time the remaining
titanocene dichloride dissolved. Completion of the reaction
Scheme 3
A series of esters with differing steric properties was
screened for this use. The overall rate of reaction is
determined by rate-limiting formation of the reactive carbene
5, and so it is the same for all the esters.9 However,
competition experiments between esters revealed relative
rates (Figure 1). Acetates of tertiary alcohols were found to
have the desired properties, reacting rapidly enough to protect
the vinyl ether, but slowly enough to avoid competing with
the substrate ester 1. From this class, the inexpensive ester
1,1-dimethyl-2-phenylethyl acetate was selected.11 An opti-
mized 0.75 equiv (relative to 1) was added to the reaction
mixture from the beginning, allowing complete conversion
of 1 to vinyl ether 2 in 5 h, and protecting 2 for as much as
24 h of additional heating.
1
was confirmed by H NMR.
In a 1500-gal vessel, a quench mixture was prepared by
charging water (1170 L), ammonium chloride (160 kg),
Celite (84 kg), and toluene (1500 L). The mixture was chilled
to 0 to -5 °C, then the dimethyltitanocene process stream
was added over 2.5 h, rinsing in with additional THF (113
L). The quench mixture was maintained below 6 °C. The
mixture was then filtered (9.2 m2 filter press), rinsing with
THF (170 L) and toluene (590 L). The biphasic system was
allowed to settle, and the lower aqueous phase was removed.
The organic phase was washed with water (1340 L). The
batch temperature was maintained at 0-10 °C throughout
the workup.
Finally, a method to recycle spent DMT was desired to
reduce raw material costs and minimize the solid waste
generated. As shown in Scheme 2, the byproduct formed
from the olefination reaction was the oxo-bridged titanocene
dimer 6. The dimer could be crystallized and recovered in
80% yield by concentrating the toluene reaction mixture and
adding heptane. Treatment of the isolated oxo-dimer with
HCl in toluene or THF converted it to crystalline titanocene
dichloride which was isolated by filtration in 94% yield.12
Since titanocene dichloride is the precursor to DMT, this
closes the recycle loop (Scheme 3.) After recovery of the
dimer, the remaining titanium in the olefination reaction
mixture was removed by decomposition as before, and the
vinyl ether 2 was crystallized from ethanol/water in 91%
yield.
A portion of the batch (∼2700 L) was transferred to a
1000-gal vessel, and ester 1 (250 kg, 474 mol) was added.
The solution was vacuum distilled with a maximum tem-
perature of 35 °C to a volume of 1200 L. The remaining
unconcentrated material was then introduced into the distil-
lation vessel followed by a toluene flush (200 L). The batch
was reconcentrated to 1200 L, and the solution was assayed
to contain 20 wt % dimethyltitanocene in a molar ratio of
2.4 to 1 vs the cis ester, for an 85% yield.
Titanocene dichloride (7.3 kg, 29 mol) was added,
followed by 1,1-dimethyl-2-phenethyl acetate (70 kg, 355
mol), and the orange solution was heated to 80 °C for 6.5 h.
HPLC assay showed complete consumption of the starting
ester. The mixture was cooled to 25 °C and then vacuum
concentrated to a volume of 780 L, keeping the temperature
below 30 °C. The crystallized titanium byproduct (Cp2TiMe)2O
(6) was further precipitated by adding heptanes (723 kg) over
2 h and then collected by filtration and washed with heptanes
(795 kg).
The filtrate was concentrated to 875 L via vacuum
distillation at 10-20 °C. The remaining titanium residues
were then quenched by adding sodium bicarbonate (60 kg),
ethanol (415 kg), and water (48 kg), and then heating to 60
°C for 6 h. The volatile organic gases generated in the quench
were efficiently scrubbed via a thermal oxidizing unit. The
slurry was cooled to 20 °C, and the inorganic residues were
removed by filtration, rinsing with toluene (260 kg).
The filtrate was then vacuum concentrated at 20-25 °C
to a volume of 750 L and then the solvent was switched to
ethanol via a constant volume distillation. When the solution
contained <2% toluene by volume, the distillation was
The technical challenges of the large-scale synthesis and
use of DMT were overcome by a team effort involving
chemists, engineers, and safety experts. This work demon-
strates that the production and use of technically challenging
organometallic reagents is feasible in pharmaceutical manu-
facturing.
Experimental Section
A laboratory-scale procedure for the synthesis of dim-
ethyltitanocene13 and a titration method for its quantitative
analysis14 have been published. All reagents were purchased
from commercial sources and were used without purification.
(11) Huffman, M. A. U.S. Patent 6,255,550, 2001.
(12) Huffman, M. A.; Payack, J. F., U.S. Patent 6,063,950, 2000.
(13) Payack, J. F.; Hughes, D. L.; Dongwei, C.; Cottrell, I. F.; Verhoeven, T. R.
Org. Synth. 2002, 79, 19.
(14) Vailaya, A.; Wang, T.; Chen, Y.; Huffman, M. A. J. Pharm. Biomed. Anal.
2001, 25, 577.
258
•
Vol. 8, No. 2, 2004 / Organic Process Research & Development

stopped. Further ethanol was added (286 kg) and then the
vinyl ether was further crystallized by the addition over 1 h
of 50% aqueous ethanol (prepared from 432 kg of ethanol
and 550 kg of water). The product was collected in a
centrifuge filter, with a 380-L wash of 2:1 ethanol water,
then dried under vacuum at 45 °C. Yield 227 kg of vinyl
96.8, 115.5 (d, JCF ) 20), 121.9 (m), 123.5 (q, JCF ) 272),
125.8, 127.5, 128.4, 128.7, 130.6, 131.2 (q, JCF ) 40), 133.4,
137.7, 138.6, 155.0, 162.2 (d, JCF ) 241). Anal. (C27H22F7NO2)
Calc: C, 61.72; H, 4.22; N, 2.67; F, 25.31. Found: C, 61.79;
H, 4.10; N, 2.65; F, 25.27.
Acknowledgment
1
ether (91%): H NMR (400 MHz, CDCl3) 2.42 (dt, J ) 3.6,
We thank Mr. Don Bachert for conducting the operational
hazard evaluations. We are also grateful for the support of
the pilot plant staff at the Merck Rahway site.
12.0, 1H), 2.90 (d, J ) 12.0, 1H), 2.91 and 3.94 (ABq, J )
13.6, 1H), 3.62-3.66 (m, 1H), 3.72 (d, J ) 2.6, 1H), 4.09
(dt, J ) 2.4, 12.0, 1H), 4.75 (d, J ) 3.2, 1H), 4.82 (d, J )
3.2, 1H), 5.32 (d, J ) 2.6, 1H), 7.09 (t, J ) 8.8, 2H), 7.24-
7.33 (m, 5H), 7.58-7.62 (m, 2H), 7.80 (s, 1H), 7.90 (s, 2H).
13C NMR (100.6 MHz, CDCl3) 51.4, 59.5, 60.2, 69.4, 89.8,
Received for review December 3, 2003.
OP034180J
Vol. 8, No. 2, 2004 / Organic Process Research & Development
•
259