Development of an efficient and practical route for the multikilogram manufacture of the SRC kinase inhibitor AZD0530

Article abstract of DOI:10.1021/op100163m

In a previous publication (Org. Process Res. Dev. 2010, 14, DOI:10.1021/op100161y) we described the process research and development of a manufacturing route for the potent SRC kinase inhibitor AZD0530. While the route was successfully used to manufacture 4.5 kg of AZD0530 difumarate, it was still relatively long, used two Mitsunobu couplings, and was, in our opinion, undesirable for manufacture on a larger scale. Herein we describe the research and development of a shorter, more practical synthesis of AZD0530 difumarate. The new route, which required fewer steps, scaled well to produce >80 kg of AZD0530 difumarate in an overall yield of 38%.

Full text of DOI:10.1021/op100163m

Organic Process Research & Development 2010, 14, 1088–1093
Development of an Efficient and Practical Route for the Multikilogram Manufacture
of the SRC Kinase Inhibitor AZD0530
J. Gair Ford,* Anne O’Kearney-McMullan, Simon M. Pointon, Lyn Powell, Paul S. Siedlecki, Mark Purdie, Jane Withnall, and
Phil O’Keefe Frances Wood§
Pharmaceutical DeVelopment, AstraZeneca, Charter Way, Macclesfield, SK10 2NA, U.K. and Pharmaceutical DeVelopment,
AstraZeneca, Charnwood, U.K.
Abstract:
range of starting materials such as 2,4,6-trifluorobenzonitrile.3
Following a cost of goods analysis, routes from 1 were selected
for further evaluation on the basis of the assumption that the
cost of 5,7-diflouroquinazolinone 1 would decrease as the scale
of manufacturing increased. The two principal variants of the
route are shown in Scheme 2, and following initial small-
scale studies these were evaluated further on multigram scale
in the lab.
Routes A and B use essentially the same chemistry varying
only in the order of the steps, and both were suitable for
manufacturing purposes from a synthetic point of view.
However, lab evaluation showed anilinoquinazoline 11 to have
more favorable physical properties, more specifically ease of
crystallisation and isolation, when compared to pyranyl quinazoli-
none 12. Route A was therefore selected for further develop-
ment. The third potential variant of the route in which aniline
4 is coupled last was also evaluated but found to be low yielding
(typically <10%) in the final step.
Anilinoquinazoline 11. The first step in the selected route
is the coupling of quinazolinone 1 with aniline 4 via 4-chlo-
roquinazolinone 14 (Scheme 3). Initial screening of the reaction
was carried out in toluene. While the reaction worked well, the
product was found to oil out at the end of the reaction, and
complete dissolution of the starting materials was never
achieved. Further solvents were screened in the hope of finding
a solvent that gave better solubility of difluoroquinazolinone 1
and an improved isolation of anilinoquinazoline 11. The use
of phosphoryl chloride limited the choice of solvents that could
be used safely, so the reaction was screened in acetonitrile,
propionitrile, toluene, chlorobenzene, and 1,1,1-trifluorotoluene.
None of the solvents gave complete dissolution of 1, but the
reaction did progress satisfactorily in all cases. Chlorobenzene
was selected for further investigation on the basis of reaction
profile, yield, and cost.
Having selected chlorobenzene as the solvent, the mode of
addition was investigated. Addition of solid 1 to a solution of
diisopropylethylamine (DIPEA) and phosphorylchloride
(POCl3) and addition of POCl3 to a slurry of 1 and DIPEA to
give chloroquinazoline 14 were both investigated. Both modes
of addition gave a good reaction profile and yield following
the subsequent addition of aniline 4. On the basis of results on
route B (Scheme 2) the addition of POCl3 to a slurry of 1, 4,
and DIPEA in chlorobenzene was also investigated. This mode
In a previous publication (Org. Process Res. DeW. 2010, 14, DOI:
10.1021/op100161y) we described the process research and devel-
opment of a manufacturing route for the potent SRC kinase
inhibitor AZD0530. While the route was successfully used to
manufacture 4.5 kg of AZD0530 difumarate, it was still relatively
long, used two Mitsunobu couplings, and was, in our opinion,
undesirable for manufacture on a larger scale. Herein we describe
the research and development of a shorter, more practical
synthesis of AZD0530 difumarate. The new route, which required
fewer steps, scaled well to produce >80 kg of AZD0530 difumarate
in an overall yield of 38%.
Introduction
AZD0530 is a potent SRC kinase inhibitor with a number
of ongoing phase II clinical trials in a variety of solid tumour
malignancies.1 In a previous communication2 we disclosed the
process research and development of a manufacturing process
for the synthesis of AZD0530 which was used to deliver 4.5
kg of AZD0530 Difumarate for use in toxicology and phase I
clinical studies (Scheme 1). Following this successful delivery
we were required to deliver a further 40 kg of AZD0530
difumarate for ongoing clinical evaluation, requiring pilot-plant
rather than kilo-lab-scale manufacture. Analysis of the route
we used for the kilo-lab-scale manufacture indicated that, while
significant improvements had been made to the medicinal
chemistry route, the resulting synthesis was still relatively long
and required two Mitsunobu couplings. While it was felt that
the chemistry could be developed for use on a limited pilot-
plant scale, we did not feel it would be suitable for longer-
term manufacture, and this directed us to investigate alternative,
potentially more efficient routes to AZD0530.
Alternative Route Evaluation. Various routes were inves-
tigated for the synthesis of AZD0530, starting from a diverse
* Author for correspondence, Telephone: +44 (0)1625 513375. E-mail:
gair.ford@astrazeneca.com.
§ Current address; Jealotts Hill International Research Station, Jeallotts Hill,
Bracknell, Berkshire, RG42 6EY, U.K.
(1) Hennequin, L. F.; Allen, J.; Breed, J.; Curwen, J.; Fennell, M.; Green,
T. P.; Lambert van der Brempt C.; Morgentin, R.; Norman, R. A.;
Olivier, A.; Otterbein, L.; Ple, P. A.; Warin, N.; Costello, G. J. Med.
Chem. 2006, 49, 6465–6488.
(2) Ford, J. G.; Pointon, S. M.; Powell, L.; Siedlecki, P. S.; Baum, J.; Chubb,
R.; Fieldhouse, R.; Muxworthy, J.; Nivlet, A.; Stenson, R.; Warwick,
E. Org. Process Res. DeV. 2010, 14, DOI: 10.1021/op100161y.
(3) Ford, J. G.; McCabe, J. F.; O’Kearney-McMullan, A.; O’Keefe, P.;
Pointon, S. M.; Powell, L.; Purdie, M.; Withnall, J. WO/2006/064217,
2006.
1088
•
Vol. 14, No. 5, 2010 / Organic Process Research & Development
10.1021/op100163m  2010 American Chemical Society
Published on Web 08/23/2010

Scheme 1. Kilo-lab synthesis of AZD0530
Scheme 2. Alternative routes to AZD0530
of addition gave a comparable reaction profile, yield, and
product quality when compared with the separate preparation
of 14 but resulted in a simpler process and shorter overall
reaction time. Factorial experimental design was used to further
optimise the addition time of POCl3, equivalents of reagents,
temperature, and concentration of the reaction. A more con-
centrated reaction was found to proceed more rapidly but
resulted in poorer filtration of the product. A typical reaction
involved addition of POCl3 (1.5 equiv) over 20 min to a slurry
of 1 (1 equiv), 4 (1.05 equiv), and DIPEA (1.2 equiv) in
chlorobenzene (10 relative volumes) at 90 °C. The reaction was
then held and cooled to 20 °C, and product 11 was isolated as
its HCl salt.
Following this optimisation of the reaction, our process safety
group examined the process using calorimetry and found
significant exotherms capable of exceeding the boiling point
of chlorobenzene. Examination of the data combined with lab
observations and examination of a typical reaction using
REACT-IR suggested that the exotherms observed were the
combination of the thermodynamics of the reaction and also a
significant component due to crystallisation of the hydrochloride
of 11. In order to control the exotherms, slower additions of
POCl3 were attempted. Addition of POCl3 (1.5 equiv) over 2 h
resulted in control of the exotherms, but there was a significant
increase in the levels of impurities formed, and the reaction
did not go to completion. Reactions could be driven to
Vol. 14, No. 5, 2010 / Organic Process Research & Development
•
1089

Scheme 3. Pilot-plant route to AZD0530
completion by increased charges of POCl3 and DIPEA. Faster
additions were found to give a good reaction profile, but the
exotherms could not be safely accommodated in our pilot plant.
Finally, on the basis of the calorimetry results, the heating/
cooling capacity of the plant vessels we intended to use, and
further lab trials, a 40-min addition was stipulated. This addition
time was set to give a good reaction profile while still allowing
safe operation of the process. Two batches were operated in
our pilot plant and behaved as predicted to give an overall yield
of 66% (an additional 4% remained on the pressure filter after
discharge).
Another problem with this reaction was the development of
meaningful in-process tests. It was found that during the addition
of POCl3 to the mixture a brown solution was obtained,
suggesting that 1 was soluble in the reaction mixture. As the
reaction approached completion 11 was generally out of
solution, however, making it difficult to monitor conversion by
HPLC in a conventional manner. Ultimately, it was found that
the chlorobenzene reaction solvent could be used as an internal
standard to monitor disappearance of 1, and a reasonably
accurate means of monitoring conversion was obtained by
standardizing the area counts of 1 against the area counts of
chlorobenzene.
screened, including sodium and potassium hexamethyldisilazide
(NaHMDS, KHMDS), LDA, LDA/TMEDA, and sodium tert-
pentoxide (NaOt-Pent). The LDA reactions gave none of the
desired product, and the reactions using NaHMDS or KHMDS
gave the desired reaction but with a poor reaction profile by
HPLC. With the use of NaOt-Pent the reaction was clean but
relatively slow. To try and speed up the reaction, tert-pentanol
and N-methylpyrrolidin-2-one (NMP) were screened, showing
NMP to give the faster reaction. Using NMP at 77 °C resulted
in the formation of the 5-pentoxy (16) and the corresponding
5-hydroxy (17) impurity (Figure 1), typically at levels of 2-3%.
Reducing the reaction temperature to 60 °C allowed the reaction
to proceed in an acceptable time (2 h) with adequate control of
16 and 17 (∼0.4%).
At the end of the reaction the addition of water precipitated
pyranylether 13, which was isolated by filtration. In the lab this
filtration was found to be very slow, although it was interesting
to note that subsequent water washes were reasonably quick.
To try and improve the filtration, the addition of the reaction
mass to water (reverse addition), addition of THF to the reaction,
and the use of DME instead of NMP were tried. Of these only
the use of DME gave a significantly improved filtration, but
unfortunately the rate of reaction was slower and the profile
poorer. Finally, as analysis had shown there to be minimal 13
in the liquors, a hot filtration was attempted. Interestingly, this
was also found to give a significantly improved filtration, which
we believe is due to the reduced viscosity of NMP at higher
temperature.4 Two batches were operated in our pilot plant and
isolated at 60 °C, giving an overall yield of 88% (a small residue
remained on the pressure filter after discharge).
AZD0530. The initial route screening work described above
settled on the use of potassium hydroxide in diglyme at 120
°C as being suitable conditions for the reaction of pyranyl ether
13 with alcohol 10. Further screening of conditions looked at
variations in solvent (diglyme, NMP, DMSO, 1,2-diethoxy-
ethane), base (KOH, K3PO4, NaH, K2CO3, LiOt-Bu, KOt-Bu,
KHMDS, NaHMDS, CsOH, NaOt-Pent) and temperature. From
Pyranyl Ether 13. Initial attempts at the selective displace-
ment of the 5-fluoro of 11 with tetrahydropyranol 7 used
potassium tert-butoxide (KOt-Bu) in THF at reflux. This gave
a reaction that was selective with respect to the formation of
the analogous displacement of the 7-fluoro of 11 but resulted
in significant levels of the undesired 5-butoxy impurity 15
(Figure 1). Indeed, using 5 equiv of KOt-Bu to reach reaction
completion resulted in the concomitant formation of up to 20%
(by HPLC area) of 15. A variety of alternative bases were
(4) Yang, C.; He, G.; He, Y.; Ma, P. J. Chem. Eng. Data 2008, 53, 1639–
Figure 1. Impurities in pyranyl ether 13.
1642.
1090
•
Vol. 14, No. 5, 2010 / Organic Process Research & Development

Figure 2. Impurities in AZD0530.
this screen it was clear that KOH and the alkoxide bases gave
the best conversions with the optimum solvents being diglyme
and 1,2-diethoxyethane (DEE). NaOt-Pent in DEE at 120 °C
appeared to give the best reaction although significant formation
(∼12%) of impurity 18 (Figure 2) was observed. This impurity
was presumably the result of a low level of 2-ethoxyethanol in
the DEE, and it was found that reducing the dilution of the
reaction from 30 relative volumes of DEE to 6 relative volumes
(rel vol) significantly reduced its formation as did the use of
higher purity solvent.
by the addition of aqueous HCl before extracting AZD0530
into ethyl acetate (EtOAc). Separation of the aqueous layer
and further water washes removed residual alcohol 10
before concentration by distillation and the addition of
water induced crystallisation of the trihydrate. Two batches
were operated in our pilot plant with the reaction
proceeding as expected. However, a lower than expected
yield of 56%, was obtained. While it was found that
the first batch operated as expected and filtered well, the
filtration of the second batch was extremely poor. The
residue left on the pressure filter after discharge was
subsequently found to be unexpectedly high for the volume
of the filter, equivalent to 12% overall yield which in
combination with the isolated yield brings the performance
of the stage within expectations. We believe the poor
filtration of the second batch and the large heel to be
indicative that the trihydrate is highly compressible and
that discharge of the first batch from the filter led to a
highly compacted cake being left on the filter between
batches which retarded the second filtration.
AZD0530 Difumarate. Finally, a difumarate salt forma-
tion process incorporating a polishing filtration was
developed. This process involved addition of a solution
of AZD0530 free base in aqueous isopropanol to a solution
of fumaric acid in aqueous isopropanol. This process
produced material of the desired quality and was trans-
ferred to our pilot plant where we intended to manufacture
three batches. Disappointingly we found that filtration and
washing times were extremely long during the first batch.
The process was changed such that the solution of fumaric
acid was added to the solution of AZD0530 free base with
the addition of seed partway through the addition.
Considerable improvements were made to the filtration
and drying times as a result. Using both processes, three
batches of AZD0530 difumarate were produced in our
pilot plant with an overall yield of 90%.
Serendipity. Further optimization of the reaction condi-
tions led to the use of 3 equiv of NaOt-Pent and 3-4
equiv of alcohol 10 at 86 °C. Although it would be
expected that 2 equiv of base would be required for the
reaction (due to deprotonation of both 13 and 10), the
higher stoichiometry of both base and 10 is thought to
be required to drive the reaction to completion in a
reasonable time. While the levels of 18 were now well
controlled, impurities 19 and 20 (Figure 2) were typically
observed at a combined level of ∼8% in the reaction
mixture. During this work, alcohol 10 was obtained from
two separate suppliers and while both samples were of
good strength (>95% w/w), one sample was found to
contain water (∼4% w/w, equivalent to ∼0.3 equiv). We
initially considered rejecting this material as being unsuit-
able for use as we expected the water could lead to the
analogous phenol impurity in the reaction but as KOH
had previously been shown (see above) to mediate the
desired reaction we decided to user trial the material. To
our delight we found that the reaction proceeded well and
surprisingly, combined levels of 19 and 20 were signifi-
cantly reduced to 2.5%. Clearly the presence of water was
having a beneficial effect on the selectivity of the reaction.
Further investigation revealed the presence of 1.9-3 equiv
of water to be optimal, with higher levels of water such
as 4 equiv also resulting in a more selective reaction but
taking longer to reach completion.
Development of a work-up and isolation was our next
challenge, principally due to the need to remove the excess
alcohol 10 and the poor physical properties of AZD0530
which was known to be amorphous. During the develop-
ment of the work-up we made the discovery that a
crystalline trihydrate of AZD0530 could be formed and
had favorable physical properties when compared to the
amorphous anhydrous material prepared previously. Fol-
lowing a satisfactory end of reaction check, the bulk of
the DEE was removed from the reaction by distillation.
The pH of the crude mixture was then adjusted to ∼7
Conclusions
• The previous synthesis which required six synthetic
steps from difluoroquinazolinone 1 has been signifi-
cantly reduced to a more efficient and direct, three
synthetic step process. The overall yield from 1 has been
improved from 25% to 38%.
• These processes scaled well into our pilot plant,
enabling us to produce >80 kg of AZD0530 difuma-
rate.
Vol. 14, No. 5, 2010 / Organic Process Research & Development
•
1091

• The raw material costs ($/kg AZD0530) required to
produce AZD0530 using the new route was ∼13% of
the cost using the previous route.2,5
a further 3 h at 60 °C. During the addition a sticky solid formed
on the vessel walls but as the addition was continued this was
converted to a free-flowing crystalline solid. The product was
filtered at 60 °C, washed with water (2 × 229 mL, 3 rel vol),
and dried to constant weight to give 13 (68.6 g, 0.164 mol,
95% w/w, 80% yield). This process was operated in our pilot
plant in two batches using an input of 50 kg of 11. Spectroscopic
analysis was in agreement with the reported data.3
• The serendipitous discovery of the benefit of water
in the final fluoride displacement had a significant
impact on the development of our processes and
highlights the benefit of performing experiments which
might reasonably have been expected to fail.
• The discovery of a crystalline trihydrate had a signifi-
cant impact on our ability to develop an isolation of
AZD0530 free base.
N-(5-Chloro-1,3-benzodioxol-4-yl)-7-[2-(4-methylpiperazin-
1-yl)ethoxy]-5-(tetrahydro-2H-pyran-4-yloxy)quinazolin-4-
amine Trihydrate (AZD0530). Alcohol 10 (13.79 g, 3.19
mol equiv) was added to a mixture of 13 (12.51 g, 30
mmol) and sodium pentoxide (9.87 g, 3.0 mol equiv) in
DEE (37.5 mL, 3.0 rel vol). Water (1.34 mL, 2.48 mol
equiv) was added followed by a DEE (25 mL, 2.0 rel vol)
line wash before heating the reaction mixture to 86 °C
and holding at this temperature for 18 h. The temperature
was reduced to 50 °C before setting the vessel for
distillation. Distillate (∼50 mL, 4 rel vol) was removed
at a pressure of ∼60 mbar. A solution of hydrochloric
acid (10 mL of a 36% w/w solution, 3.88 mol equiv) in
water (84 mL, 6.72 rel vol) was added slowly. The
addition was exothermic, and the rate of addition was
controlled to prevent the reaction temperature from
exceeding 60 °C. The pH was checked (expected to be
between 7.0-7.6) and adjusted if necessary. Ethyl acetate
(225 mL, 18 rel vol) was added and the mixture heated
to 60 °C before separating the organic layer and extracting
it with water (50 mL, 4 rel vol then 25 mL, 2 rel vol).
The volume of the organic solution was reduced by
atmospheric distillation to leave a solution (∼140 mL of
distillate), and the resulting solution (∼100 mL, 8 rel vol)
was then cooled to 65 °C before adding water (4 mL, 0.32
rel vol). The mixture was then cooled to 45 °C over 1 h
and held at this temperature for 2 h during which time
crystallisation occurred. The mixture was reheated to 55
°C and held at this temperature for 5 min before cooling
to 18 °C over 4 h and holding for a further hour. This
temperature cycling was included to aid removal of
product from the vessel walls. The product was filtered
and washed with water (17 mL, 1.36 rel vol) and tert-
butylmethylether (17 mL, 1.36 rel vol) before drying to
give AZD0530 trihydrate (12.5 g, 20.3 mmol, 88% w/w,
68% yield). This process was operated in our pilot plant
in two batches using an input of 49 kg of 13. Spectroscopic
analysis was in agreement with the reported data.3
Experimental Section
General. Starting materials, reagent, and solvents were
obtained from commercial suppliers and used without further
purification. HPLC analyses were performed with an Agilent
1100 instrument. Intermediate purities were typically measured
by quantitative NMR using an internal standard of known purity,
typically maleic acid, benzyl benzoate, or 2,3,5,6-tetrachloroni-
trobenzene. Yields quoted are corrected for purity.
The processes described below are taken from the laboratory
process descriptions used as the basis for plant manufacture.
An indication of the scale on which these processes were
operated in our pilot plant is also given for each stage.
5,7-Difluoro-N-(5-chloro-1,3-benzodioxol-4-yl)quinazolin-
4-amine (11). Diisopropylethylamine (7.47 mL, 1.2 mol
equiv) was added to a slurry of 1 (6.49 g, 99.8% w/w,
35.62 mmol) and 4 (6.72 g, 1.1 mol equiv) in chloroben-
zene (64.9 mL, 10 rel vol). The resulting slurry was heated
to 95 °C before adding phosphoryl chloride (4.96 mL,
1.5 mol equiv) at a constant rate over 40 min. When ∼0.35
equiv had been added, a solution was obtained; when
∼0.85 equiv had been added, product started to crystallise.
During the addition the batch temperature was seen to
rise to ∼100 °C. The reaction was held at 95 °C for 10 h
before cooling to 18 °C and holding for a further 30 min.
The product was filtered and washed with chlorobenzene
(2 × 22.8 mL, 3.5 rel vol) before drying under reduced
pressure at 45 °C to constant mass to give 11 (8.90 g,
23.1 mmol, 97% w/w, 67% yield) as its hydrochloride.
This process was operated in our pilot plant in two batches
using an input of 39 kg of 1. Spectroscopic analysis was
in agreement with the reported data.3
7-Fluoro-N-(5-chloro-1,3-benzodioxol-4-yl)-5-(tetrahydro-
2H-pyran-4-yloxy)quinazolin-4-amine (13). 11 (76.42 g,
95.4% w/w, 0.205 mol) was added portionwise to a solution
of sodium tert-pentoxide (85.69 g, 3.80 mol equiv) in NMP
(500 mL, 6.54 rel vol) over 20-30 min. This addition is highly
exothermic with a 40 K temperature rise possible. The exotherm
can be controlled by the rate of addition. A line wash of NMP
(20 mL, 0.26 rel vol) was added before adding 7 (23.52 mL,
1.20 mol equiv) followed by a further line wash of NMP (15
mL, 0.20 rel vol). The reaction was heated to 60 °C for 2.5 h
before analyzing by HPLC for completion. Water (764 mL,
10 rel vol) was added over 3 h before holding the mixture for
N-(5-Chloro-1,3-benzodioxol-4-yl)-7-[2-(4-methylpiperazin-
1-yl)ethoxy]-5-(tetrahydro-2H-pyran-4-yloxy)quinazolin-4-
amine Difumarate (AZD0530 Difumarate). A solution of
AZD0530 (27.1 g, 50 mmol) in isopropanol (210 mL, 7.75
rel vol) and water (30 mL, 1.11 rel vol) was filtered at 40
°C through Celite into a reaction vessel. A line wash of
isopropanol (20 mL, 0.74 rel vol) was passed through the
filter before heating the solution of AZD0530 free base
to 75 °C. A solution of fumaric acid (12.77 g, 2.2 mol
equiv) in isopropanol (200 mL, 7.38 rel vol) and water
(20 mL, 0.74 rel vol) at 70 °C was filtered into a second
vessel, and half of the resulting solution (∼110 mL) was
(5) No allowances have been made for economies of scale; thus, the real
reduction in costs will have been less than this. The new route produced
>80 kg of AZD0530 difumarate, whereas the old route was used to
produce 4.5 kg.
1092
•
Vol. 14, No. 5, 2010 / Organic Process Research & Development

added to the solution of AZD0530 free base, maintaining
the temperature between 70-75 °C. AZD0530 difumarate
(20 mg, 0.07 wt %) was added to seed the crystallisation,
and the stirred mixture was held at 75 °C for one hour
after crystallisation had begun. The remainder of the
solution (∼110 mL) was added over 1 h to the crystalli-
sation vessel followed by an isopropanol (20 mL, 0.74
rel vol) line wash. The mixture was stirred at 75 °C for
14 h before cooling to 20 °C over at least 2 h. After
stirring for a further 30 min the product was filtered,
washed twice with a 10% solution of water in isopropanol
(50 mL, 1.85 rel vol then 100 mL, 3.69 rel vol), and dried
to constant mass at 45 °C under reduced pressure to give
AZD0530 difumarate (35.8 g, 46 mmol, 98% w/w, 92%
yield). This process was operated in our pilot plant in three
batches using an input of 23 kg of AZD0530. Spectro-
scopic analysis was in agreement with the reported data.3
Acknowledgment
We thank Amjad Khan for analytical assistance. We thank
Steve Hallam and his colleagues from our process safety group
for evaluations and expert advice. We thank Frank Harkness
and his colleagues from our Pilot Plant and Ruth Moyes for
contributing to the successful scale-up of the processes. We also
thank Graham Robinson for many helpful discussions.
Received for review June 10, 2010.
OP100163M
Vol. 14, No. 5, 2010 / Organic Process Research & Development
•
1093