A new approach to rapid parallel development of four neurokinin antagonists. Part 2. Synthesis of ZD6021 cyano acid

Article abstract of DOI:10.1021/op020065h

The manufacture of ZD6021 cyano acid (1) using a new project approach is described. Research Department processes were scaled up to 100 L if process safety anal robustness were not compromised; other factors were treated according to the new approach. By using this strategy, we were able to manufacture a key intermediate on sufficient scale to support delivery of 1 kg quantities of bulk drug within 6 months of the start of lab work.

Full text of DOI:10.1021/op020065h

Organic Process Research & Development 2003, 7, 58−66
A New Approach to Rapid Parallel Development of Four Neurokinin Antagonists.
Part 2. Synthesis of ZD6021 Cyano Acid
Jonathan D. Moseley,* William O. Moss, Matthew J. Welham, Claire L. Ancell, John Banister,† Sharon A. Bowden,
Glenn Norton,† and Maureen J. Young
AstraZeneca, Process Research and DeVelopment, AVlon Works, SeVern Road, Hallen, Bristol BS10 7ZE, UK
Abstract:
was the route we sought to develop. The route to cyano acid
used the same intermediates as the Research route2 with some
changes of reagents as shown in Scheme 1.
The manufacture of ZD6021 cyano acid (1) using a new project
approach is described. Research Department processes were
scaled up to 100 L if process safety and robustness were not
compromised; other factors were treated according to the new
approach. By using this strategy, we were able to manufacture
a key intermediate on sufficient scale to support delivery of 1
kg quantities of bulk drug within 6 months of the start of lab
work.
The simple structure of 1 disguises the fact that 1,3-
disubstituted naphthalenes are difficult to make. This sub-
stitution pattern defies the standard directing rules for
naphthalenes, namely that activating groups are ortho/para-
directing in the same ring while deactivating groups are meta-
directing in the other ring. Fortunately, 3-bromo-1-naphthoic
acid is described in the literature,5 and its synthesis relies
on the introduction of a temporary group to effect the
substitution. Naphthalic anhydride is brominated in the
3-position, and mercuric oxide-mediated decarboxylation
gives predominantly the desired 1,3-isomer.6 Conversion of
this compound to the desired cyano acid relies on more
straightforward chemistry.2,4 Given the absence of other
routes, successful preliminary scale-up of the chemistry in
the Research Department facilities, and an acceptance that
mercury waste could be disposed of reliably on a small scale,
it was decided to use this route for the first delivery.
However, given the undesirable nature of the process from
an environmental viewpoint, a project was initiated at Zeneca
Agrochemicals (Huddersfield) to investigate a longer-term
manufacturing route which is reported in this issue.7 It will
be helpful to know in the discussion that follows that we
remanufactured cyano acid for ZD2249 by the same route
as for ZD6021, on which most of the discussion is focused.
Introduction
The preceding contribution1 presented a new approach
by Zeneca Pharmaceuticals to the rapid parallel development
of several candidate drugs. An overview of the strategy was
given and illustrated by examples taken from across the
neurokinin (NK) project.2 This contribution and the next3
discuss in more detail the experiences of manufacture of key
fragments of molecules required for the NK programme and
seek to assess the application and success of this new
approach in more detail.
Results and Discussion
Bromo Anhydride Stage (3). The bromo anhydride stage
provided an early and multifaceted example of the application
of the new project approach. The process used by Research
started with 1,8-naphthoic anhydride (2) in 70% concentrated
nitric acid to which 0.75 equiv of bromine was added at 50
°C.5 A typical yield of 17% was achieved with acceptable
quality, residual starting material at the 3-4% level being
the main impurity, which was readily removed in the
following stages. Attempts to push the reaction further
invariably resulted in multiple bromination products and a
poor impurity profile.5 The process was assessed by our
Hazards Group and deemed to be safe with appropriate
precautions (mainly adequate gas disengagement capacity).
We therefore decided this was a Research-based process
The first compounds in the NK programme, ZD6021 and
ZD2249, both required the cyano acid (1) as a common
structural unit. The Research Department route built on that
of Dewar,4 and since a key aspect of the new approach was
to apply the Research-based route in all possible cases,
provided that process safety considerations permitted, this
† Large Scale Laboratory, Macclesfield.
(1) Moseley, J. D.; Moss, W. O. Org. Process Res. DeV. 2003, 7, 53-57.
(2) Bernstein, P. R.; Aharony, D.; Albert, J. S.; Andisik, D.; Barthlow, H. G.;
Bialecki, R.; Davenport, T.; Dedinas, R. F.; Dembofsky, B. T.; Koether,
G.; Kosmider, B. J.; Kirkland, K.; Ohnmacht, C. J.; Potts, W.; Rumsey, W.
L.; Shen, L.; Shenvi, A.; Sherwood, S.; Stollman, D.; Russell, K. Bioorg.
Med. Chem. Lett. 2001, 11, 2769. Albert, J. S.; Aharony, D.; Andisik, D.;
Barthlow, H.; Bernstein, P.; Bialecki, R. A.; Dedinas, R.; Dembofsky, B.
T.; Hill, D.; Kirkland, K.; Koether, G. M.; Kosmider, B. J.; Ohnmacht, C.;
Palmer, W.; Potts, W.; Rumsey, W.; Russell, K.; Shen, L.; Shenvi, A.;
Sherwood, S.; Warwick, P. J. J. Med. Chem. 2002, 45, 3972.
(5) Rule, H. G.; Thompson, S. B. J. Chem. Soc. 1937, 1764.
(6) Whitmore, F. C.; Fox, A. L. J. Am. Chem. Soc. 1929, 51, 3363.
(7) Ashworth, I.; Bowden, M. C.; Dembofsky, B.; Levin, D.; Moss, W. O.;
Robinson, E.; Szczur, N.; Virica, J. Org. Process Res. DeV. 2003, 7, 74-
81.
(3) Parker, J. S.; Bowden, S. A.; Firkin, C. R.; Moseley, J. D.; Murray, P. M.;
Welham, M. J.; Wisedale, R.; Young, M. J.; Moss, W. O. Org. Process
Res. DeV. 2003, 7, 67-73.
(4) Dewar, M. J. S.; Grisdale, P. J. J. Am. Chem. Soc. 1962, 84, 3541.
58
•
Vol. 7, No. 1, 2003 / Organic Process Research & Development
10.1021/op020065h CCC: $25.00 © 2003 American Chemical Society
Published on Web 12/17/2002

Scheme 1
Table 1. FED parameters for bromo anhydride stage
decision to accept the costs of disposal and also reasoned
that the quantities would not be prohibitive on this scale.
However, we soon discovered that the liquors were unstable
and resulted in gassing. It was necessary to dilute the liquors
2-fold with water to suppress this gassing to an acceptable
level as determined by our Hazards Group. We had also
stored some of the waste in HDPE containers, but the residual
bromine attacked the plasticisers, making the containers
brittle; thus, the waste was transferred to glass bottles with
venting caps. Given these additional considerations and the
additional quantities of acidic waste we generated, it is
doubtful that we would have repeated this process on the
pilot-plant scale (i.e., >10 kg). However, we did repeat the
synthesis of ZD6021 cyano acid (1) using this process to
make material for ZD2249, and achieved an identical yield
of 16.7% over five batches with no processing issues. We
therefore felt we had been justified in our intention to take
this process directly from Research and accommodate it in
the LSL on this scale. (For a discussion of the alternative
silver sulfate-mediated bromination, see below.)
parameter
low level
high level
temperature (°C)
50
0.75
10
15
30
65
0.95
60
20
240
bromine charge (equiv)
bromine addition time (min)
nitric acid charge (vols)
reaction time (min)
which could be operated in the Large Scale Laboratory
(LSL)8 for the following reasons: (a) process safety was
acceptable, (b) cheap and available reagents meant that the
low output could be tolerated, (c) the volume was not so
prohibitively high that efficiency would be compromised,
(d) the process was operationally simple (except for the
handling of bromine, which could be accommodated due to
the flexibility of the LSL), and (e) product quality was
satisfactory.
We performed a small factorial experimental design
(FED) to check for robustness and to determine if we could
improve the modest yield. We investigated five parameters
at two levels in eight experiments with three centre points,
giving a quarter factorial overall. The parameters investigated
with their high and low values are shown in Table 1.
A higher charge of bromine gave an apparently higher
yield, but quality was much worse. A reduction to 15 vols
was possible, but not to 10. However, we were advised by
Hazards Group not to go below 20 vols for safety reasons.
Overall, no improvements were made compared to the
Research process, but we did establish that the process was
robust within wide ranges for several key parameters. Having
met the specified key criteria1 of process safety and robust-
ness with acceptable quality, seven batches of 3 were
manufactured in the LSL for ZD6021 in an overall yield of
16.7%,9 with quality in the range 92-96%.
Bromo Acid Stage (4/6). Again, the Research method
was followed for this stage with only minor changes.6 The
process proceeded by digestion of the bromo anhydride (3)
with NaOH followed by insertion of Hg into the anhydride
with loss of one of the carboxylate groups as CO2. The
organo-Hg intermediates 9a and 9b were formed in a 3:1
ratio and could be isolated and oven dried. Acidic hydrolysis
with concentrated HCl liberated 1 mol of CO2 and yielded
the desired 3-bromonaphthoic acid (4) (ZD6021 bromo acid)
and its 6-bromo regioisomer (6), still in a 3:1 ratio.10 The
product was isolated by filtration from the cooled reaction
mixture, washed, and dried to give a quantitative yield of
the 2 isomers 4 and 6.
This process also left us with large quantities of acidic
waste containing a high level of bromine. We had taken the
(8) The Macclesfield LSL is a cGMP manufacturing facility for synthesis of
bulk drug for clinical studies and uses all-glass vessels. It is typically where
the first significant scale-up of a process occurs, and commonly delivers
tens of kilograms of intermediates and kilograms of bulk drug. It consists
of a range of glass reactors 10-100 L in scale, fully contained with other
ancillary equipment in fume cupboards. Operating ranges vary from -78
to +130 °C. Atmospheric hydrogenations can be performed, and a 20-L
rotary evaporator is available for distillations if required. Product is generally
isolated as a solid on Nutsches. AstraZeneca has several other LSLs at
different sites which operate in a similar fashion.
(9) The first batch gave poor-quality product which was not used and is not
included in the yield figures. The reasons for this failure were never
established, but poor-quality naphthoic anhydride (2) was suspected, and
this batch of input was not used again.
(10) We had to develop a separate HPLC method to determine the regioisomer
ratio at this stage. Details are given in the Experimental Section.
Vol. 7, No. 1, 2003 / Organic Process Research & Development
•
59

Table 2. FED Parameters for bromo acid stage
We made some initial improvements to the Research
process by cutting down the reaction time for the Hg insertion
reaction from 4 to 2 days. This was not easy to determine
as the organo-Hg adducts (9) had very poor solubility as
noted by Leuck,11 and so each analytical sample taken had
to be heated in concentrated HCl for several hours to liberate
the bromo acids (4/6) before analysis. However, 2 days was
found to be adequate for complete reaction in all cases. We
were reluctant to isolate the organo-Hg intermediates (9)
on the larger scale for handling and health hazard reasons,
and indeed it was found unnecessary to do so; the process
could be acidified directly after completion of the Hg
insertion reaction. This also avoided an unnecessary filtration
and drying procedure, as well as the handling issues. The
acidic hydrolysis was complete in 4 h, with most of the CO2
being driven off during heating.12
We also investigated the NaOH charge, which at 4.0 equiv
was potentially all neutralised by the glacial acetic acid used
to dissolve the HgO. An “all-in” reaction with no pretreat-
ment of the anhydride by NaOH unsurprisingly failed to go
to completion. However, overcharging the NaOH to 5 or 6
equiv quenched the reaction altogether, the vivid orange
colour of the HgO persisting through-out. We speculated
whether Hg might coordinate directly to the anhydride
without the need for base, but this reaction too failed. With
2.5 equiv, the reaction was reliable with only a brief induction
period to open up the anhydride of 2, even though not all of
it appeared to have reacted/dissolved during this period. With
these modest improvements made, we then attempted to
progress the regioisomeric mixture 4/6 through the subse-
quent stages in the hope of removing the 6-bromo acid (6)
or its derivatives by selective crystallization. This, however,
was unsuccessful, and we therefore decided to investigate
the front half of the process more thoroughly with the aim
of improving the isomer ratio in favour of the desired
3-bromo acid (4).
We decided to conduct an FED using our large Zymark
robot which would allow us extensive coverage of the
parameters investigated. We chose six parameters at two
levels and conducted 32 test tube-scale reactions (with five
centre points) which gave a half factorial overall. The
parameters investigated with their high and low values are
shown in Table 2.
All temperatures were at or just off reflux since given
the long reaction time there was no point investigating lower
values if we were to have a viable process. We took samples
at 6-, 12-, 24-, and 48-h time points in each case, and assayed
the reaction both for completion and regioisomer ratio.
Despite this industry (greatly facilitated by the use of the
robot), we could not find any factors that improved the ratio
beyond 3:1. We did learn that the process was slightly more
robust with the higher-charge of HgO and benefited from
parameter
low level
high level
NaOH charge (equiv)
water charge 1 (vols)
HgO charge (equiv)
acetic acid charge (vols)
HCl charge (equiv)
2.5
6.75
1.08
1.0
10
4.0
13.25
1.15
2.0
20
12.5
water charge 2 (vols)
0
the 2-day reflux but was insensitive to the other factors. This
gave us confidence for the manufacture, and only one batch
was needed for ZD6021. This was repeated for ZD2249, and
yields of 97% were achieved in both cases with no processing
issues.
Bromo Acid RX Stage (4). After the failure of the FED
investigation, we reluctantly turned our attention to the
recrystallisation employed by Research, which we had been
hoping to avoid. This proved to be reproducible on scale-up
with no changes. A simple crystallization of the crude bromo
acids (4/6) from 10 vols of hot glacial acetic acid reliably
gave a 60-62% yield of the desired 3-bromo acid (4) with
the regiosiomer ratio >100:1 (i.e., effectively >80% yield
of available 4).10 Other impurities were also reduced to well
below 1%, being mainly 1-naphthoic acid derived from Hg-
mediated decarboxylation of unreacted 2 carried through
from the first stage. Bromo anhydride (3) proved more
difficult to remove if it carried through above 4%, but we
were able to ensure that this did not occur.
We attempted to crystallise the water-wet paste of crude
bromo acids directly (to avoid handling and drying this
material potentially contaminated with Hg residues), but the
paste retained about 2 vols of water which compromised the
crystallization process. We did investigate azeotropic drying,
but the process was not developed before it was needed, and
we did not review the situation for the repeat manufacture
for ZD2249. The first time through, two batches of bromo
acid RX (4) were manufactured in 61% overall yield with
0.7% of the 6-bromo acid (ratio 143:1) and 0.2% of
1-naphthoic acid. For ZD2249, one double-size batch was
made in 60% yield with the 6-bromo acid (6) undetected.
Mercury analysis was performed as discussed below.
Bromo Ester Stage (5). The Research process reacted
the bromo acid (4) with oxalyl chloride to form the acid
chloride in dichloromethane, which was then concentrated
to dryness and redissolved in methanol to give bromo ester
(5). The reaction was catalysed by DMF so that generation
of carcinogenic dimethylcarbamoyl chloride was a potential
issue.13 Clearly we could have accommodated this process,
since this method was used for the later amide bond-forming
reaction,3 but we felt in this case that a simpler procedure
could just as easily be developed within the time we had set
ourselves, that is, 4 weeks per stage.
Initially, crude bromo acid (4/6) was slurried in methanol
with concentrated H2SO4 and the mixture heated to reflux
overnight. The methyl esters (5/7) were smoothly formed
and, after an aqueous drown-out, extracted with a variety of
(11) Leuck, G. J.; Perkins, R. P.; Whitmore, F. C. J. Am. Chem. Soc. 1929, 51,
1831.
(12) The reaction mixture foamed alarmingly at reflux, and it was initially thought
that the mixture was effervescing constantly. However, the hazard study
showed that most of the CO2 generated was driven off during the heating
phase, and the foaming was attributed to the properties of the reaction. The
process was run off reflux at 95 °C to avoid foaming and hence material
being caked onto the vessel walls.
(13) Irving Sax, N. Dangerous Properties of Industrial Materials, 5th ed.; Van
Nostrand Reinhold Company: New York, 1979.
60
•
Vol. 7, No. 1, 2003 / Organic Process Research & Development

solvents to give 70-75% yields on concentration. The
presence of the 6-bromo ester (7) in up to 25% of the crude
product mixture thwarted the direct crystallization of the
desired bromo ester (5). However, once bromo acid RX was
available, a reverse drown-out of the screened hot methanolic
solution into water gave product of good form and quality
in >90% yields, with no extraction needed. We briefly
investigated other acids (CF3SO3H and CF3CO2H) and the
charge of H2SO4, but found no advantage. Minor impurities
present included 1-methyl naphthoate and the half ester/acids
(10a/b) formed by methanolysis of unreacted bromo anhy-
dride. All of these were removed to insignificant levels by
solubilising the cyanide anion. We investigated the use of
pyridine and 2,6-lutidine in the hope of improving the
reaction rate but were surprised to find that the reaction was,
if anything, faster with no additive.
Finally, to remove the copper, we used experience gained
on a similar process in another Zeneca project rather than
the dilute drown-out using ammonium hydroxide reported
by Dewar.4 The reaction mixture was first diluted with an
equal volume of NMP to prevent it from solidifying on
cooling to room temperature. A brine solution was added
which allowed removal of most of the copper as the soluble
cuprate complex and simultaneously precipitated the crude
product. This process typically gave a near quantitative yield
with organic purity >90% as determined by HPLC.15 Colour,
however, varied from light to dark brown, and later results
confirmed that, relative to the desired metal specification,
there was still much copper to remove. We had already
determined that a separate crystallization stage would be
required, though.
slurry-washing the wet product cake with a dilute bicarbonate
wash, which also removed any unreacted 4. Only one batch
of each was needed for ZD6021 and ZD2249, giving 97 and
95% yields, respectively, and with quality >98.5%. Overall,
this process was sufficiently simple to be developed within
our target 4-week time frame and was operationally easier
than the Research method. However, it was a rare luxury to
be able to develop a new process on this project, rather than
simply modify the Research one.
Cyano Ester RX Stage (8). No process was available
for this stage as generally intermediates were purified by
chromatography in the Research Department. The main
purpose was to allow further removal of copper residues
down to a sufficiently low level for continued processing
through to the bulk drug. The crude cyano ester was slurried
in ethyl acetate at 20 °C and washed with a brine and 0.15
M solution of EDTA disodium salt. The crude material was
not completely soluble at this temperature, and the separa-
tions were not ideal, but they could not be improved by
heating either. Three washes were planned on the lab scale,
but during manufacture up to six or seven washes were
needed until the characteristic blue colour of the copper salts
was no longer discernible. After a final brine wash, the ethyl
acetate solution was concentrated to dryness to give a 90%
yield of the cyano ester. This was redissolved in hot ethyl
acetate in a clean vessel and MTBE added to aid crystal-
lization. Cooling to 5-10 °C gave a white solid in typically
70% yield with the organic purity >98% and copper levels
in the ppm range (see below).
Cyano Ester Stage (8). The Research process heated
bromo ester (5) and copper (I) cyanide in NMP at 180 °C to
achieve displacement of the bromide by cyanide.4 Obvious
concerns with this process were the ability to handle cyanide
on the large scale, the high temperature required, and the
need to remove copper from the product. The use of cyanide
seemed unavoidable, and the copper could be dealt with, but
the high temperature reaction could not be achieved in the
LSL except in a heating mantle. This was deemed unac-
ceptable for safety reasons in combination with cyanide,
although we did later successfully make use of a heating
mantle for the S-thiocarbamate stage of ZD2249.14 We were
therefore limited to 130 °C in this case, which consequently
extended the reaction time from 4 to 5 h for the Research
process to 48 h at 130 °C. Initial lab reactions also revealed
that an impurity identified as the amide ester (11a) was being
produced in up to 25%, presumably by adventitious moisture
hydrolyzing the nitrile group under the forcing conditions.
Rigorous exclusion of water from the apparatus and the use
of anhydrous reagents reduced this to a tolerable level of
4%. However, unreacted bromo ester (5) also remained at
4% which was too high for subsequent processing. A slight
under-charge (0.98 equiv) of CuCN had been used initially
to ensure that no cyanide remained during the workup. With
a slight over-charge (1.02 equiv), less than 2% 5 remained
after 48 h which was acceptable. “Catalytic” pyridine had
also been used previously,4 presumably the idea being to
accelerate the reaction by binding to the copper and thus
The yield for the cyano ester and cyano ester RX stages
combined was 69% overall for ZD6021 manufacture. A slight
mischarge of the crystallization solvents during the second
manufacture led to only a 59% yield for ZD2249, with a
small second crop being isolated in addition. Attempts to
avoid concentrating to dryness or to use a single solvent
throughout (toluene and MTBE were investigated) could not
be made to work in the short time available. Had time
(15) In addition to unreacted 5 and hydrolysis product 11a, there were other
minor impurities (<1% each). None was identified, although the debromi-
nated reduction product would be a possible candidate. The levels were
not troublesome to further processing, and no further work was needed.
(14) Bowden, S. A.; Burke, J. N.; Gray, F.; McKown, S.; Moseley, J. D.; Moss,
W. O.; Murray, P. M.; Welham, M. J. Org. Process Res. DeV. Manuscript
in preparation.
Vol. 7, No. 1, 2003 / Organic Process Research & Development
•
61

Table 3. Hg and Cu analysisa
allowed, the ethyl acetate/MTBE ratio would have been
further optimised to improve the recovery.
stage
Hg by XRF
Hg by AFS
Cu by XRF
Cyano Acid Stage (1). Cyano ester was hydrolyzed with
LiOH in water/THF at 20 °C overnight. The use of LiOH
was made to minimise hydrolysis of the cyano group, but
even with an under-charge (0.98 equiv), typically 2% of the
amide acid (11b) was formed. Acidification with HCl
precipitated generally good yields (90-95%) of product with
93-95% purity by LC. The solvent ratio of water:THF
(5:1) was found to be important for good recovery; at a ratio
of 1:1 the crystallization failed completely. A volume of
methanol equal to that of THF was also found to be
advantageous, in this case for suppressing additional hy-
drolysis of the nitrile group. More was not added in case of
perturbing the crystallization. The reaction could also be run
faster at a higher temperature, but predictably levels of the
amide acid (11b) also increased. By this stage in the
manufacturing sequence, development time was very short,
and the process was judged fit-for-purpose without further
work. This was a good example of an unoptimised stage
which required minimal effort and was achieved within the
4-week time frame.
Overall, the process performed smoothly in the LSL for
both campaigns, giving 88 and 89% yields with quality
∼94%. The amide acid was controlled at 2%, but on scale-
up, 3% of another impurity was detected which could not
be identified by LC-MS at the time. Fortunately it processed
out in the later stages without complication, and we did not
return to this issue.
Cyano Acid RX Stage (1). Due to the low final purity
achieved for ZD6021 (94.8%), it was decided for the next
compound (ZD2249) to maximise the input quality of the
three key intermediates prior to their coupling.1,3 Cyano acid
quality was ∼94%, and it was felt that this could be
improved. A series of 16 test tube-scale scale crystallizations
was conducted, covering a range of solvents and solvent
ratios. A blended batch of cyano acid (93.6% purity by LC)
was used throughout, but quality was only improved by 2%
in most cases (i.e., almost within experimental error), with
recoveries in the range 50-90%. An IPA/isohexane com-
bination (1:1) gave a marked improvement to 98.3% by LC,
although recovery was disappointing at 55%. Further adjust-
ments of the solvent ratio to 2:1 and cooling to 0 °C for
several hours improved the recovery to 78% with quality
reliably around 99%.
bromo acid RX
bromo ester
cyano ester
1100
850
31
6
1057
846
31
6
9
n.d.
n.d.c
n.d.
n.d.
175
55
cyano ester RX
cyano acidb
9
0
ZD2249 pure
1
a All figures in ppm; see ref 16 also. b Mean for two batches. c n.d. ) not
determined.
tolerated in the manufacture. This in turn required, first, their
removal from intermediates and bulk drug (as discussed
above for Cu) and, second, analysis to prove that they had
been removed to acceptable levels in the final bulk drug.
For ZD6021, the manufacture out-ran our ability to perform
the appropriate analysis in time, and thus most results were
performed retrospectively. Final figures for Hg by XRF were
2 ppm at cyano ester RX and 1 ppm at cyano acid for
ZD6021 manufacture, with Cu not determined. A fuller set
of data for both Cu and Hg was available for the second
manufacture supporting ZD2249, and this is collected in
Table 3.
Mercury analysis was conducted by our analytical depart-
ment at Macclesfield using XRF and locally by Avlon Works
Environmental Group using AFS, and the results showed a
remarkable level of agreement, given the experimental error
expected on these techniques.16 Crude bromo acid (4/6) itself
was not assayed, but the level in bromo acid RX after the
acetic acid crystallisation was far too high at 1100 ppm.
There was effectively little reduction during the bromo ester
stage, but the aqueous drown-out of the NMP solution at
the crude cyano ester stage gave an impressive reduction
(nearly 30-fold). The EDTA washes in the recrystallisation
stage appeared to bring the level down a further 5-fold, but
there was no improvement beyond this.
Copper analysis was determined using XRF alone, and
was only relevant from the cyano ester stage, although again
not determined for crude 8. The EDTA washes had already
reduced the level to 175 ppm in the crystallised cyano ester
RX, and there was a further drop to 55 ppm in the cyano
acid stage. The Hg specification was to be set at 5 or 10
ppm in the bulk drug, and for Cu at possibly up to 50 ppm
for initial toxicity testing. Given the molecular weight
increase from the cyano acid to the bulk drug, no further
clean up was actually required. However, for both metals,
further reductions were observed during downstream pro-
cessing, leading to final levels in ZD2249 bulk drug of Hg
as not detected and Cu at 1 ppm (similar results were
achieved for ZD6021). Our decision to use Research-based
chemistry more reliant on stoichiometric heavy metals had
therefore paid off, although it did reveal that we needed more
analytical resources to support these activities using this
strategy.
This process however was never used in the LSL. Use of
cyano acid RX in the following amide alcohol coupling
stage3 gave only a marginal improvement in overall quality
of 2%, and there was no improvement in yield, allowing for
the different input strengths. By this stage of ZD2249
manufacture, the yield loss on crystallising cyano acid in
78% far out-weighed the marginal 2% improvement in
quality obtained and could not be tolerated. Cyano acid was
judged to be fit-for-purpose at 94% and was used accord-
ingly, with no issues.
(16) The error range for the XRF results (allowing for calibration of the apparatus,
sample matrix, and sample collection parameters) was estimated to be (2
ppm at the 10 ppm level for heavy metals such as Hg and Pd, and (5 ppm
in this range for first-row transition metals such as Cu, Fe, and Zn, in non-
halogen-containing samples.
Metals Analysis. A consequence of the new project
approach was the acceptance that more metals, and probably
more toxic heavy metals such as Hg, would have to be
62
•
Vol. 7, No. 1, 2003 / Organic Process Research & Development

Alternative Bromination Reaction. Given the very low
yield of bromination of naphthoic anhydride, we naturally
considered other processes. Another process in the literature
for this conversion was that of Mitchell,17 which used
bromine in concentrated H2SO4 with 0.5 equiv of silver
sulfate (presumably to precipitate AgBr and hence drive the
reaction to completion). Mitchell quoted an 81% yield and
in our hands the yield was apparently quantitative. Unfor-
tunately, the quality was obviously poor, with typically up
to 15% unreacted 2 contaminating the product. The melting
point was also significantly depressed against crystallised
material by nearly 30 °C. A further determination of strength
by NMR showed the level to be about 60%, giving an overall
yield of typically 60%. This was much improved on 16%,
but there were significant processing problems also.
First, the AgBr salt had to be screened off from
concentrated H2SO4, which was judged to be inoperable in
the LSL because the solvent would have destroyed the paper
filters. The acid solution was then diluted with a vast quantity
of water, which was highly exothermic. This also limited
the output, even allowing for the higher yield per batch. The
form of the product was very poor, leading to extended
filtration and drying times. Finally, the product was only
about 60% strength, which meant the impurities were carried
into the next stage. A separate crystallization stage would
almost certainly have been needed. Overall, this process was
judged to have too many problems to be worth pursuing in
the time available, and thus we focused our efforts on the
existing process described above. The cost of the silver
sulfate was not considered on this scale.
undesirable (bromo acid and cyano ester). Interestingly, the
overall yield was only modestly improved over the Research
route from 4.0% to around 5.4%, but this had not been our
primary goal. We made use of the robot for a large screening
study on the bromo acid stage and performed two FEDs for
the first two stages. We did expend some effort later in trying
to improve stages (bromo anhydride and cyano acid RX)
since 1 was an intermediate common to two compounds.
Most stages were developed sufficiently to pass our safety
and robustness criteria within 4-6 weeks as intended; bromo
ester and cyano acid required only 3.
We also gained much valuable learning from this project.
In particular, we learnt that we needed more analytical
support, especially LC-MS and metals analysis, to support
product development at the rate of 4 weeks per stage, and
although we managed to handle the SHE issues of some
unpleasant reagents and demanding conditions, we realised
that more effort was required to accommodate these less well
developed stages. This is likely to be a continuing trend
within the industry.
In summary, ZD6021 cyano acid manufacture provided
challenging examples to our new project strategy in almost
every aspect. Even so, we delivered the desired quality and
quantity of material to manufacture bulk drug within the
demanding 6-month deadline. That we were able to re-
manufacture a second campaign for a 1-kg delivery of bulk
drug without incident supports the conclusion that this
approach was both safe and robust.
Experimental Section
General Procedures. Reaction mixtures and products
were analysed by reverse phase HPLC on Hewlett-Packard
1050 or 1100 instruments according to the following condi-
tions. Method A (general): column, HiChrom (or Waters
Spherisorb) S5ODS-1, 250 mm × 4.6 mm i.d.; eluent,
550:450 acetonitrile:water with 0.1% v/v TFA; flow rate 1.0
mL/min; wavelength 230 nm; injection volume 10 µL.
Typical retention times (tr’s) were: 1,8-naphthalic anhydride,
6.0; 1-naphthoic acid, 5.2; bromo anhydride, 10.1; 3- and
6-bromo acids, 8.6; 3-bromo ester, 28.4; 6-bromo ester, 27.3;
cyano ester, 10.4; amide ester, 3.9; 1-methyl naphthoate,
12.9; half acid-esters, 5.5; cyano amide, 2.8; cyano acid,
4.6 min. Method B (regioisomer ratio): column, Zorbax
SB-Phenyl, 250 mm × 4.6 mm i.d.; eluent, 700:200:100
water:acetonitrile:THF with 0.1% v/v TFA; flow rate 1.5 mL/
min; wavelength 230 nm; injection volume 10 µL. Typical
retention times (tr’s) were: bromo anhydride, 19.2; 6-bromo
acid, 20.3; 3-bromo acid, 21.1 min. HPLC purities/strengths
are area % normalised, except where noted otherwise.
Melting points were determined using a Griffin melting point
apparatus (aluminium heating block) and are uncorrected.
1H and 13C NMR spectra were recorded on a Varian Inova
400 spectrometer at 400 and 100.6 MHz, respectively, with
chemical shifts given in ppm relative to TMS at δ ) 0.
Electrospray (ES+) mass spectra were determined on a
Micromass LCT with time-of-flight and electron impact (EI)
mass spectra were determined on a Micromass Autospec.
Analytical TLC was carried out on commercially prepared
plates coated with 0.25 mm of self-indicating Merck Kiesel-
We did briefly review the decision before the start of
ZD2249 manufacture, but decided that this process was still
very problematic. The existing process, despite its 16% yield,
was robust and operable on this scale with reasonable output.
We felt the effluent could be managed on this scale and
therefore ran the manufacture again with no problems for
ZD2249.
Conclusions
ZD6021 cyano acid manufacture proved to be an excellent
example of our intended new project strategy,1 in particular
our revised approach to traditional long-term manufacturing
factors such as environmental, health, manufacturability, and
output/yield criteria. Although we did manage to avoid the
chromatography option for quality targets, we did need two
separate crystallizations to deal with specific cases (the
refractory regioisomer ratio of the bromo acid stage, and
removal of copper at the cyano ester RX stage). As planned,
safety was not compromised in any way, and robustness was
also demonstrated in that only one batch had to be quaran-
tined from both manufactures.9
We achieved our aim of demonstrating that a Research-
type route could be successfully operated on up to 100-L
scale with minimal development to deliver 1 kg of bulk drug.
This included one process (bromo anhydride) that we would
have judged inoperable for a larger-scale delivery (and later
did),7 and two others which would have been highly
(17) Mitchell, W. J.; Topsom, R. D.; Vaughan, J. J. Chem. Soc. 1962, 2526.
Vol. 7, No. 1, 2003 / Organic Process Research & Development
•
63

gel 60 F254 and visualised by UV light at 254 nm. Prepara-
tive-scale silica gel flash chromatography (for lab work) was
carried out by standard procedures using Merck Kieselgel
60 (230-400 mesh). Where not stated otherwise, assume
standard practices have been applied. Note well: Yields and
strengths given in the text above are those which can be
expected from the processes operated under normal condi-
tions; figures below are quoted for individual experiments
and may vary slightly from those quoted in the text.
(100 mL each) and dried in vacuo at 60 °C to constant weight
to yield the title compound as an white solid (15.5 g, 111%
uncorrected for strength). HPLC purity 72% by area, with
14-16% unreacted naphthoic anhydride remaining. 1H NMR
strength against maleic acid internal standard, 56%. Corrected
yield using NMR strength, 8.7 g, 62%. Mp (crude) 215-
217 °C (lit.,15 246-252 °C). Other data as recorded above.
Preparation of ZD6021 Bromo Acid (4/6). The preced-
ing bromo anhydride (22.25 g, 80.3 mmol) was slurried in
sodium hydroxide solution (201 mL of 1.0 M solution, 201
mmol, 2.5 equiv) and heated to 50 °C with agitation for 15
min. During this period most of the bromo anhydride
dissolved, but complete solution is not required for successful
reaction. In a second vessel, yellow mercury (II) oxide (18.91
g, 87.0 mmol, 1.09 equiv) was charged followed by water
(57.3 mL) and then glacial acetic acid (19.1 mL). The
contents are heated to 50 °C with gentle agitation during
which time the mercury (II) oxide readily dissolved to give
a colourless solution. (Note well: The order of addition of
the solvents, water then acetic acid, affects the ease of
solution.) This freshly prepared mercury (II) acetate solution
was transferred to a heated addition funnel at 50 °C and
added dropwise over 15 min to the bromo anhydride solution.
(Note well: The mercury acetate will crystallise if not kept
warm.) A milky white suspension forms immediately in a
pale orange solution, the colour of which fades to white
during heating. The reaction mixture was heated at reflux
for 2 days, or just off reflux (95 °C) on the larger scale if
preferred, and then cooled to 80 °C. Concentrated HCl (158
mL, 1.98 mol, 24.6 equiv) was added dropwise over 10 min
and the reaction mixture reheated to reflux for a further 4 h.
The reaction mixture was cooled to 20 °C and the resulting
solid isolated by filtration. The wet cake is slurry-washed
three times with water (225 mL each) and then dried in vacuo
at 70 °C to yield a 3:1 mixture of the 3- and 6-bromo acids
as a white solid (18.7 g, 93% corrected for strength for the
combined bromo acids). HPLC (Method A: tR ) 8.6 min
for both isomers, strength 89%); mp (crude) 195-198 °C.
Other data reported in the bromo acid RX section below.
Preparation of ZD6021 Bromo Acid RX (4). The
preceding crude ZD6021 bromo acid (4.98 g, 19.8 mmol)
was slurried in glacial acetic acid (50 mL) and heated to
reflux with stirring. The reaction mixture was held at reflux
for 3 h to ensure that all the solid had dissolved; it was then
cooled evenly at 10 °C/h to 20 °C and stirred overnight at
this temperature. The resulting solid was isolated by filtration
and dried in vacuo at 70 °C to yield the title compound as
a white solid (3.22 g, 58% corrected for strength). HPLC
(Method A: tR 8.6 min, strength 90%; Method B: 6-bromo
acid, tR 20.3 min; 3-bromo acid, tR 21.1 min; isomer ratio of
3:6-bromo acid ) typically >100:1). Data on pure 3-bromo
acid (4): mp 230-231 °C (lit.,5 231-232 °C); 1H NMR (400
MHz, d6-DMSO) δ 13.55 (1H, br s), 8.82 (1H, d, J ) 8.4
Hz), 8.48 (1H, d, J ) 1.2 Hz), 8.18 (1H, J ) 2.0 Hz), 8.02
(1H, d J ) 7.6 Hz), 7.63-7.72 (2H, m); 13C NMR (100.6
MHz, d6-DMSO) δ 167.25, 134.78, 134.37, 131.96, 130.05,
129.13, 128.08, 127.88, 127.36, 125.58, 117.57. Data on
pure 6-bromo acid (6) (obtained by base hydrolysis of an
Preparation of ZD6021 BromoAnhydride (3). 1,8-
Naphthalic anhydride (50.0 g, 252 mmol) was slurried in
concentrated nitric acid (1000 mL of 70% solution, sp gr
1.42) and heated to 50 °C. Bromine (9.7 mL, 189 mmol,
0.75 equiv) was added evenly over 10 min, and the resulting
brown solution was held at 50 °C for 4 h before cooling to
20 °C. A cream-coloured solid precipitated during this
cooling period which was isolated by filtration at a water
pump. (Note well: Acidic gases present!) The filter cake
was slurry washed with water three times (130 mL each) or
until the final water wash was neutral as determined by pH
papers. (Note well: The damp product is unstable if dried
in the presence of residual acidic liquors.) The isolated
product was dried in vacuo at 60 °C to yield the title
compound as a white-to-pale cream-coloured crystalline solid
(14.2 g, 19.1% corrected for strength (based on naphthalic
anhydride)). HPLC (Method A: tR 6.0 min, strength 94%);
1
mp 238-240 °C (lit.,5 242-243 °C); H NMR (400 MHz,
d6-DMSO) δ 8.85 (1H, s), 8.53 (1H, d, J ) 7.6 Hz), 8.48
(2H, d, J ) 7.6 Hz), 7.95 (1H, t, J ) 7.8 Hz); 13C NMR
(100.6 MHz, d6-DMSO) δ 160.12, 159.57, 136.73, 134.31,
133.88, 132.65, 132.61, 128.68, 128.32, 121.33, 120.07,
119.38; MS (EI+) 276/278 (M+, 1:1, 94%), 232/234 ((M -
CO2)+, 1:1, 100%). (Caution: waste disposal. The liquors
from this reaction are acidic and are known to decompose
with gassing on standing. It was found necessary to dilute
the liquors at least 2-fold with water before further disposal.
Neither diluted nor undiluted liquors should be stored in
plastic containers even in the short term, as the residual
bromine can attack the plasticisers making the containers
brittle. Glass containers with a venting cap were used in this
case.)
Preparation of ZD6021 Bromo Anhydride (3) by Silver
Sulfate Method.17 1,8-Naphthalic anhydride (10.0 g, 50.5
mmol) and silver sulfate (7.9 g, 25.25 mmol, 0.50 equiv)
were slurried in concentrated sulfuric acid (200 mL) at
ambient temperature to give a dark yellow/green slurry.
Bromine (3.2 mL, 63.0 mmol, 1.26 equiv) was added over
30 min, and then heated to 60 °C for 6-12 h, before cooling
back to 20 °C. The solid silver bromide byproduct was
filtered off at a water pump to give a clear orange solution
(Note well: acidic gases present!). This was added dropwise
over 45 min to a second vessel containing water (1600 mL)
and standing in an ice bath. (Caution: a substantial exotherm
occurs on this scale!) An off-white solid precipitates during
the addition. This was isolated by filtration either im-
mediately or after several hours, but in both cases, the
filtration rate was slow. The filter cake was washed once by
displacement with water (50 mL) and twice with cold ethanol
64
•
Vol. 7, No. 1, 2003 / Organic Process Research & Development

analytically pure sample of 6-bromo ester): mp 180-181
completion of the reaction as determined by HPLC, the
reaction mixture was cooled to 30-40 °C and diluted further
with NMP (40 mL) to prevent the product from crystallising
prematurely. The solution was cooled to 25 °C and a
saturated solution of brine (200 mL) added dropwise over 1
h. The resulting slurry was stirred for 3 h at 20 °C and the
solid isolated by filtration. The filter cake was washed
sequentially, once by displacement with saturated brine (40
mL) and twice with water (40 mL each), and dried in vacuo
at 50 °C to yield the crude title compound as a light to mid-
brown solid (15.24 g, 87% corrected for strength). HPLC
(Method A: tR 10.4 min, strength 91%); mp 101-103 °C.
Other data reported in cyano ester RX section below.
Preparation of ZD6021 Cyano Ester RX (8). The
preceding crude cyano ester (5.0 g, 23.7 mmol) was slurried
in ethyl acetate (60 mL) at 20 °C. Saturated brine solution
(20 mL) and EDTA solution (20 mL of 0.15 M EDTA-
disodium salt) were added to the ethyl acetate slurry of cyano
ester and stirred vigorously at 20 °C for 1 h. The layers were
allowed to settle, and the blue-coloured lower aqueous layer
was separated. This procedure was repeated once more, or
on the larger scale, until the aqueous layer no longer retained
the characteristic blue colour of solvated copper ions (up to
six times). The ethyl acetate solution was washed with a
saturated brine solution (20 mL), and the ethyl acetate layer
concentrated under reduced pressure to give cyano ester as
a brown solid. In a clean vessel, this solid was redissolved
in ethyl acetate (4.3 mL) and MTBE (17.2 mL) with stirring
by heating to 60-65 °C for 1 h to achieve a solution. The
reaction mixture was then cooled at 14 °C/h over 4 h down
to 5-10 °C, and stirred at this temperature for 1.5 h. The
resulting solid was isolated by filtration, washed twice by
displacement with cold MTBE (8.6 mL each), and dried in
vacuo at 50 °C to yield the title compound as a white solid
(2.77 g, 58% on this scale). HPLC (Method A: tR 10.4 min,
1
°C (lit.,4 185-186 °C); H NMR (400 MHz, d6-DMSO) δ
8.82 (1H, d, J ) 9.2 Hz), 8.31 (1H, d, J ) 2.0 Hz), 8.18
(1H,d, J ) 7.2 Hz), 8.14 (1H, d, J ) 8.0 Hz), 7.76 (1H, dd,
J ) 9.2, 2.0 Hz), 7.64 (1H, t, J ) 7.6 Hz); 13C NMR (100.6
MHz, d6-DMSO) δ 168.47, 135.03, 132.36, 130.67, 130.63,
130.56, 129.45, 128.10, 128.06, 126.43, 119.75.
Preparation of ZD6021 Bromo Ester (5). The preceding
bromo acid RX (100 g, 0.40 mol) was slurried in methanol
(1200 mL) at 20 °C with mechanical agitation, and concen-
trated sulfuric acid (10.9 mL of 98% strength, 0.20 mol, 0.5
equiv) was added dropwise to the slurry over 4 min. The
reaction mixture was heated to reflux (65 °C) for 18-24 h.
The resulting solution was cooled to 60 °C and filtered
through preheated apparatus to remove any fines. The filter
apparatus was rinsed with methanol (50 mL) and warmed
to 60 °C, and the wash was combined with the main reaction
solution in a jacketed addition funnel at 55 °C. The warm
solution of bromo ester was added dropwise over 1 h to a
second vessel containing water (1500 mL) and stirred at 25
°C which precipitated the bromo ester product. (Note well:
The drown-out vessel may need additional cooling on this
scale.) The addition funnel was washed with warm methanol
(50 mL) and the washings were added to the drown-out
vessel. The resulting white precipitate was stirred at 25 °C
for 1 h and then isolated by filtration. The damp cake was
slurry-washed sequentially with a dilute solution of NaHCO3
(300 mL of 0.05 M solution) and then twice with water (300
mL each), pulled as dry as possible, and dried in vacuo at
30 °C (Note well: low-melting solid) to yield the title
compound as a white solid (103.7 g, 96.3% corrected for
strength). HPLC (Method A: tR 28.4 min, strength 98.5%;
6-bromo ester, tR 27.3 min; the isomer ratio can also be
determined at this stage using Method A at 295 nm if
required). Data on pure 3-Bromo Ester (5): mp 58-59 °C
1
1
(lit.,5 59 °C); H NMR (400 MHz, d6-DMSO) δ 8.69 (1H,
strength 100%); mp 109-110 °C (lit.,4 108-109 °C); H
d, J ) 8.4 Hz), 8.51 (1H, s), 8.16 (1H, d, J ) 2.0 Hz), 8.04
(1H, d, J ) 8.0 Hz), 7.60-7.73 (2H, m), 3.96 (3H, s); 13C
NMR (100.6 MHz, d6-DMSO) δ 165.94, 134.78, 134.69,
132.03, 128.90, 128.86, 128.36, 127.98, 127.55, 125.21,
117.49, 52.58; MS (ES+) 306/308 (M + CH3CNH+, 100%),
265/267 (M + H+, 61%), 233/235 (M - CH3OH+, 48%).
Data on pure 6-bromo ester (7) (obtained by flash
chromatography of a crude mixture of the 3- and 6-bromo
esters and eluting with 5% MTBE in isohexane): mp 62-
NMR (400 MHz, d6-DMSO) δ 8.84 (1H, s), 8.75 (1H, d, J
) 8.8 Hz), 8.29 (1H, s), 8.17 (1H, d, J ) 8.4 Hz), 7.87 (1H,
t, J ) 7.8 Hz), 7.77 (1H, t, J ) 7.6 Hz), 3.98 (3H, s); 13C
NMR (100.6 MHz, d6-DMSO) δ 165.88, 138.93, 132.53,
131.37, 131.04, 129.72, 129.48, 128.42, 128.04, 125. 27,
118.14, 107.67, 52.69; MS (ES+) 253 (M + CH3CNH+,
100%) 211 (M+, 20%).
Preparation of ZD6021 Cyano Acid (1). The preceding
cyano ester RX (2.0 g, 9.47 mmol) was dissolved in THF
(10 mL) to give a coloured solution. Lithium hydroxide
monohydrate (389 mg, 9.28 mmol, 0.98 equiv) was dissolved
in water (40 mL) to give a clear solution which was added
to the cyano ester RX over several minutes. A pale precipitate
formed which disappeared as the reaction proceeded. A line
wash of water (10 mL) was added followed by methanol
(10 mL) and the reaction mixture stirred at 20 °C overnight
(16-18 h). The solution was screened through a filter to
remove any residual solid before the addition of dilute
aqueous HCl (2.0 M, 9.5 mL, 18.9 mmol, 2.0 equiv). A solid
crystallised during this addition and was isolated by filtration,
slurry-washed twice with water (20 mL each), and dried in
vacuo at 60 °C to yield the title compound as a white solid
1
63 °C (lit.,4 63-64 °C); H NMR (400 MHz, d6-DMSO) δ
8.67 (1H, d, J ) 9.2 Hz), 8.30 (1H, d, J ) 2.0 Hz), 8.15
(2H, m), 7.75 (1H, dd, J ) 9.4, 2.2 Hz), 7.63 (1H, t, J ) 7.8
Hz), 3.92 (3H, s); 13C NMR (100.6 MHz, d6-DMSO) δ
166.86, 134.71, 132.52, 130.72, 130.43, 130.42, 128.91,
127.47, 126.73, 126.18, 119.71, 52.35.
Preparation of ZD6021 Cyano Ester (8). The preceding
bromo ester (20.0 g, 75.4 mmol), copper (I) cyanide (6.89
g, 77.0 mmol, 1.02 equiv) and anhydrous NMP (40 mL) were
charged to a nitrogen-inerted vessel with stirring at 20 °C.
The resulting mixture was heated to 130 °C for 48 h under
a gentle stream of nitrogen. A yellow solution was formed
which became progressively darker during this period. On
Vol. 7, No. 1, 2003 / Organic Process Research & Development
•
65

(1.44 g, 72% on this scale but typically 90-95%). HPLC
(Method A: tR 4.6 min, strength 94%); mp 209-210 °C (lit.,4
210-212 °C); 1H NMR (400 MHz, d6-DMSO) δ 8.88 (1H,
d, J ) 8.4 Hz), 8.81 (1H, s), 8.29 (1H, s), 8.16 (1H, d, J )
8.4 Hz), 7.85 (1H, t, J ) 7.6 Hz), 7.76 (1H, t, J ) 7.4 Hz);
13C NMR (100.6 MHz, d6-DMSO) δ 167.33, 138.37, 132.61,
131.71, 130.70, 130.11, 129.39, 129.36, 127.82, 125.72,
118.35, 107.67; MS (ES+) 198 (M + H+, 100%), 154 ((M
- CO2)+, 45%).
Preparation of ZD6021 Cyano Acid RX (1). The
preceding crude cyano acid (4.0 g crude, 19.0 mmol corrected
for strength) was slurried in 2-propanol (28 mL) with stirring
and heated to reflux (82 °C) to achieve a complete solution.
The solution was cooled back to 60 °C and isohexane (14
mL) added dropwise from an addition funnel over 10 min.
The solution was cooled to 20 °C evenly at 20 °C/h and
stirred at 20 °C overnight (for convenience). The product
crystallised in the range 40-50 °C. The slurry was further
cooled to 0 °C and stirred for 2 h, and the solid was isolated
by filtration. The filter cake was washed by displacement
twice with cold isohexane (10 mL each) and dried in vacuo
at 40 °C to yield the title compound as a white solid (2.92
g, 78% corrected for strength). HPLC (Method A: tR 4.6
min, strength 99%); mp 215-216 °C (lit.,4 210-212 °C).
Other data as noted above.
Acknowledgment
We thank Steve Hallam and Paul Gillespie (Process Safety
Group, Blackley, now at Macclesfield) for sterling work on
the hazard studies in support of this project; Bob Osborne
(PR&D Avlon) for the use of the Zymark robot and both
him and Rob Shaw (Macclesfield) for help with the FEDs
and statistical analysis; Jo Brenton (Avlon Works Environ-
mental Group) and Gordon Plummer (Analytical Department,
Macclesfield) for Cu and Hg analyses; Steve Lake (PR&D
Avlon) for mass spectra; and Caroline Wright and Roger
Crosby for additional experimental work on the bromo
anhydride stage. We also benefited from excellent coopera-
tion with our Research colleagues at Wilmington, DE, in
particular, Jim Hulsizer and his team.
Received for review July 22, 2002.
OP020065H
66
•
Vol. 7, No. 1, 2003 / Organic Process Research & Development