in small amounts (<0.5%). Compound 10 had long been
23
underestimated in the routine analysis of intermediate 3, and
anhydride 11 had not been recognized as a dominant byproduct
either.
In addition to these byproducts, evidence for the presence
of polymeric material in the xylene solution of 3 could also be
24
provided. Having now a clear picture about the nature and
abundance of the various byproducts, the lower yield in the
subsequent conversion 3 f 2 could be understood. The presence
of polymeric material and identified byproducts led to losses
applied reaction conditions. When working in aliphatic sol-
16
vents, monobromide 5 precipitates during the reaction and is
25
to the mother liquor in the crystallization of 2. Moreover, since
17
to some extent “protected” against further bromination,
allowing a higher conversion rate of 6 with minimal formation
<5%) of dibromide 7. In our hands, using NBS in cyclohexane
26
pentanoic acid anhydride is a consumer of tributyltin azide,
its presence also affected the conversion rate in this reaction
(
27
negatively, leading to a loss in yield.
and running the reaction to 92% conversion of nitrile 6 proved
To reach our manufacturing process goals of producing very
pure 3, the formation of polymers and compounds 10 and 11
had to be suppressed. Whereas compound 10 must have derived
to be the optimal combination. We did not consider using
18
bromine as brominating agent.
Either radical initiator AIBN (2.5 mol %) or 2,2′-azobis(2,4-
28
from the acylation of L-valine benzyl ester (8), a contaminant
in 4a, anhydride 11 and the polymers had their origin in the
19
dimethylvaleronitrile) (1 mol %) can be applied. According
to the new process, 4′-methyl-biphenyl-2-carbonitrile (6) and a
29
reaction conditions of the conversion 4a f 3. We felt that
N-ethyldiisopropylamine was responsible both for the formation
of the polymers and pentanoic acid anhydride. Apparently, this
base not only acted as an acid binder for hydrochloric acid but
20
catalytic amount of aqueous hydrobromic acid are heated in
cyclohexane to 67 °C. After the addition of the radical initiator,
21
a suspension of NBS in cyclohexane is added portionwise
over a period of 2 h at 64–70 °C. The obtained suspension is
stirred for 1 h, cooled to room temperature, and filtered. After
washing with cyclohexane, isopropanol, and water, the product
is discharged and dried under vacuum at 60 °C for 20 h. In
contrast to the original process, the new procedure requires only
one bromination step, thus increasing the throughput by at least
3
0
might have also induced the formation of n-propyl ketene,
31
which could have reacted to produce the polymers, compound
1, and ketene dimer 13. If this hypothesis were true,
32
33
1
replacing N-ethyldiisopropylamine by an alternative base not
capable of inducing ketene formation would solve this problem.
As to the other dominant byproduct, compound 10, its removal
had to be achieved by improving the washing conditions of
the filter cake of 4a.
40%. In addition, the isolated yield could be increased by 12%.
With the elimination of chlorobenzene, halogenated solvents
are no longer used in the whole valsartan production process.
(
(
23) At 254 nm, compound 3 has 48 times the UV absorbance of 10.
24) Determined in the evaporation residue of a sample of 3 (HPLC, external
standard technique). We did not investigate the nature of the postulated
polymeric compounds (3% relative to 3).
4
. Redesign of the Manufacturing Processes for 3 and 4a
4.1. Quality Investigation. Having recognized the impor-
tance of obtaining intermediate 3 in high purity, we started this
project with the reinvestigation of the analytics of this inter-
mediate. In Figure 1, the byproducts we have been able to
identify are summarized. Among these, the most dominant
compounds (5–8 mol % each) turned out to be (S)-3-methyl-
(25) When in the laboratory the conversion 3 f 2 was carried out with
pure 3, less of 2 was found in the mother liquor than when using 3
from routine production.
(26) Kricheldorf, H. R.; Lepert, E. Synthesis 1976, 329–330.
(27) In the manufacture of 2, prolonged heating at reflux in xylene has to
be avoided due to a decomposition tendency of 2 under these rather
harsh reaction conditions. Obviously, when using a constant excess
of tributyl tin azide, the presence of 11 must lead to a somewhat longer
reaction time (for the same conversion rate of 3) and therefore to a
lower yield.
22
2-pentanoylamino-butyric acid benzyl ester (10) and pentanoic
acid anhydride (11). The compounds 12–15 were only present
(
28) To completely remove L-valine benzyl ester 8, efficient washing of
the filter cake of hydrochloride 4a was required.
(
(
16) (a) Katsura T.; Shiratani H. Eur. Pat. Spec. EP 0709369 B1, 1999.
b) Morita A.; Kobayashi Y.; Tamura K. Jap. Pat. Appl. JP 09176103
(
(29) Hydrochloride 4a did not contain any polymeric material.
(30) Hill, C. M.; Hill, M. E.; Schofield, H. I.; Haynes, L. J. Am. Chem.
Soc. 1952, 73, 166.
A2, 1997.
17) Compared to the homogenous conditions in chlorobenzene, about half
of 7 is formed under the heterogeneous conditions in cyclohexane (at
the same conversion rate of 6). Dibromide 7 remains to a large extent
in the mother liquor in the crystallization of bromide 5 (see
Experimental Section).
18) For the bromination of 6 with bromine in cyclohexane see: Haber, S.
Patentschrift DE 19712339 C1, 1998.
19) (a) Half-life for AIBN at 65 °C ) 10 h. Talât-Erben, M.; Bywater, S.
J. Am. Chem. Soc. 1955, 77, 3712. (b) Half-life time for 2,2′-
azobis(2,4-dimethylvaleronitrile at 68 °C ) 1 h. Overberger, C.G.;
O’Shaughnessy, M.T.; Shalit, H. J. Am. Chem. Soc. 1949, 71, 2661.
20) By addition of catalytic amounts of aqueous hydrobromic acid, the
start of the bromination reaction can be significantly accelerated.
21) At plant scale, the suspension of NBS was added in 8 equal portions.
22) For this investigation compound 3 from routine production (50%
solution in xylene) was used. The structures of the compounds 10
and 12–15 were derived by LC–MS, and the structure of 11 by GC–
MS. In addition, compounds 10 and 13 were independently synthesized
(31) Competitive oligomerization appeared to be a limitation in intramo-
lecular cycloaddition reactions of ketenes. Markó, I.; Ronsmans, B.;
Hesbain-Frisque, A.-M.; Dumas, St.; Ghosez, L.; Ernst, B.; Greuter,
H. J. Am. Chem. Soc. 1985, 107, 2192–2194.
(
(32) The formation of anhydride 11 is thought to proceed via addition of
pentanoic acid (formed by hydrolysis of pentanoic acid chloride) to
n-propyl ketene or by reaction of pentanoic acid with pentanoic acid
chloride. In either case, the presence of water is required. Since
hydrochloride 4a was not dried, the presence of small amounts of
water could not be excluded.
(
(
(33) (a) No effort was made to determine the geometry of the double bond
in 13. However, in the preparation of structurally related ketene dimers
using also N-ethyldiisopropylamine, exclusively the (Z)-isomers were
obtained. Purohit, V. C.; Richardson, R. D.; Smith, J. W.; Romo, D.
J. Org. Chem 2006, 71 (12), 4549–4558. (b) However, in the
preparation of structurally related ketene dimers using also
N-ethyldiisopropylamine, exclusively the (Z)-isomers were obtained.
Calter, M. A.; Orr, R. K.; Song, W. Org. Lett. 2003, 5 (24), 4745–
4748.
(
(
(
see Experimental Section), and anhydride 11 was purchased from
Fluka.
8
94
•
Vol. 11, No. 5, 2007 / Organic Process Research & Development