Development of a continuous flow scale-up approach of reflux inhibitor AZD6906

Article abstract of DOI:10.1021/op200340c

Early scale-up work of a promising reflux inhibitor AZD6906 is described. Two steps of an earlier route were adapted to be performed in continuous flow to avoid issues related to batch procedures, resulting in a robust method with reduced cost of goods and improved product quality. Toxic and reactive reagents and starting materials could be handled in a flow regime, thereby allowing safer and more convenient reaction optimization and production.

Full text of DOI:10.1021/op200340c

Article
pubs.acs.org/OPRD
Development of a Continuous Flow Scale-Up Approach of Reflux
Inhibitor AZD6906
Tomas Gustafsson,† Henrik Sorensen, and Fritiof Ponten*
́
̈
Medicinal Chemistry, AstraZeneca R&D Molndal, SE-431 83 Molndal, Sweden
̈
̈
ABSTRACT: Early scale-up work of a promising reflux inhibitor AZD6906 is described. Two steps of an earlier route were
adapted to be performed in continuous flow to avoid issues related to batch procedures, resulting in a robust method with
reduced cost of goods and improved product quality. Toxic and reactive reagents and starting materials could be handled in a
flow regime, thereby allowing safer and more convenient reaction optimization and production.
formed product. Use of smaller-size continuous flow reactors
would also make optimization of reaction conditions more
efficient and allow the required stoichiometry of reagents to be
explored. The conditions found to be most suitable could then
directly be scaled up by simply prolonging the processing time,
increasing the pump rate and the size of the reactor, or a
combination thereof.
INTRODUCTION
■
In AstraZeneca’s program to develop efficient and safe
gastroesophageal reflux inhibitors, AZD3355 (1), was chosen
for late-stage clinical studies.1 AZD6906 (2) was proposed as a
follow-up to this promising GABAB receptor agonist. Both of
these substances have been developed to affect peripheral
receptors selectively, thereby avoiding CNS side effects
associated with high doses of, among others, baclofen (3)
(Figure 1).2 When considering scale-up of the synthetic route
RESULTS AND DISCUSSION
■
The proposed synthesis of 2 is shown in Scheme 1, consisting
of three steps: formation of protected phosphinate 6, acylation
of 6 with N-Boc-glycine methyl ester 7, and finally deprotection
and salt removal. While the third step involves heterogeneous
slurries and precipitations that are inherently in conflict with
the flow regime, the first two steps are known to be highly
exothermic and could therefore benefit from being performed
in small-size continuous flow reactors. We first set out to
investigate the reaction between methyl dichlorophosphine (4)
and triethyl orthoacetate (5) to form phosphinate 6. The first
step had previously been performed in a batch process, and
from these experiments the reaction was known to be highly
exothermic.4 In addition, handling of the pyrophoric and toxic
methyl dichlorophosphine reagent (4) was a cause for concern.
Three equivalents of triethyl orthoacetate was reacted with
methyl dichlorophosphine in a simple T-mixer (PTFE, 1 mm
inner diameter) and a PTFE-tube reactor (inner diameter of 1
mm and a total reactor size of 2 mL)5 set to 20 °C with a
retention time of 2 min to give a clean conversion to product
(6). The temperature increased only a few degrees during the
reaction, demonstrating the outstanding heat dissipation of a
small dimension tube reactor. Rinsing of the reactor with
toluene followed concentration in vacuo afford the desired
product in 79% yield.6 However, it was deemed that the
exotherm associated with the reaction could just as well be
safely controlled at this scale (<100 g of methyl dichlor-
ophosphine) in a batch operation by careful addition and
external cooling to below 10 °C using dry ice/acetone.7 In this
batchwise manner, 75 g of methyl dichlorophosphine 4 was
Figure 1. GABAB agonists under development or on the market.
to AZD6906 (Scheme 1) several potential problems could be
identified. The first two steps are highly exothermic, the second
of which requires strict cryogenic conditions and slow addition
of reagents to avoid degradation of the product or side
reactions. In addition, methyl dichlorophosphine (4) is very
expensive and in short supply from commercial sources. A
general problem when generating a product with acidic protons
in enolate-type chemistry is the in situ quenching of the
reacting anion during the reaction, often requiring excess base
or reagent. By working in a different kinetic regime we hoped
to be able to bypass this problem due to instant consumption of
the reacting anion. An additional challenge was to produce
AZD6906 in a quality suitable for pharmaceutical development.
AZD6906, being a small and highly functionalized (amino-
ketone, zwitterionic, and phosphinic acid) molecule, had been
found to have serious stability issues, in the impure state.
It was envisioned that most of these issues could be
addressed by performing the reactions in continuous flow. With
the uniform conditions that are normally obtained inside
narrow channels, reaction temperatures and residence times
could be carefully controlled.3 This would allow convenient
examination of the effects of running the reactions at higher
temperatures with continuous quenching and removal of
Special Issue: Continuous Processes 2012
Received: November 28, 2011
Published: February 21, 2012
© 2012 American Chemical Society
925
dx.doi.org/10.1021/op200340c | Org. Process Res. Dev. 2012, 16, 925−929

Organic Process Research & Development
Article
a
Scheme 1. Synthesis of API 2
a
Reagents and conditions: a) In batch: 0 °C, addition of 4 during 4 h,
stirred for 2 h. b) In continuous flow: 25−55 °C, 4 and 5 continuously
mixed, 2 min retention time. c) In batch: THF, −78 °C, slow addition
of 6 to LDA, stirred for 1 h, dropwise addition of 7, stirred for 45 min.1
d) In continuous flow: 2-MeTHF, 35 °C, 6 and 7 continuously mixed
with LDA, 1.5 or 2 min retention time . e) i. HCl, i PrOH, rt, 14 h; ii
propylene oxide (13 equiv), MeOH, rt, 14 h; iii charcoal, MeOH, rt,
30 min; iv recrystallization from H2O/EtOH.
converted into intermediate 6. Due to difficulties transporting
the pyrophoric and highly toxic phosphine, the remainder of
the required material (an additional 700 g of methyl
dichlorophosphine) was processed, using the batch protocol,
by the supplier and was delivered as the protected intermediate
6.8
The key acylation step, in which the carbon framework is
constructed, had several limitations for batch processing. First,
according to the original procedure a 2-fold excess of the costly
phosphinate 6 was used.1,4,9 The extra equivalent of the anion
of 6 is consumed by deprotonation of the more acidic product
8. In addition, cryogenic conditions had to be used to control
the strongly exothermic reaction. Aiming at reducing the need
for the costly phosphinate 6 we set up the reaction conditions
to get instant reaction of the phosphinate anion with the
glycine substrate 7 directly followed by quenching to minimize
the time for side reactions. On the basis of conditions known
from the previously performed reaction using a batch
protocol1,4 (addition of 1 equiv of N-Boc-glycine methyl ester
to 2 equiv of phosphinate 6 deprotonated with 3.5 equiv LDA
in THF at −78 °C), we attempted the same stoichiometry but
at a higher temperature and shorter reaction times in a
continuous flow reactor. Thus, a solution of N-Boc-glycine
methyl ester (7, 1 equiv) and methyl phosphinate (6, 2 equiv)
was mixed with a solution of freshly prepared LDA (3 equiv) in
a T-mixer at room temperature and reacted in a PTFE tube
reactor (2 mL) thermostatted at 25 °C. The flow rates were
adjusted to achieve a retention time of 5 min (see Figures 2 and
3). After workup and monitoring by TLC, the reaction
appeared to have gone to completion.10 The experiment was
repeated using excess (2 equiv) of N-Boc-glycine methyl ester,
pleasingly with the same apparent result. However, when
Figure 2. Schematics of reactor. (a) first generation; (b) second
generation.
abilities of miniaturized reactors. Knowing that the reaction can
tolerate high temperatures for at least short periods of time, the
effect of the reaction temperature was examined in the range of
25−55 °C and was found to be of little consequence for the
reaction. The reactor temperature was therefore increased to 35
°C, reasoning that this might reduce problems with viscosity-
related clogging during longer processing times.11 The
retention time was examined, and it was found that the same
results were achieved after 90 s as there were after 15 min; i.e.
the reaction proceeds very quickly at these temperatures.
Initially, commercial-grade LDA12 was used. It was found that
the final concentration of LDA in the reaction mixture had to
be higher than 1 M to ensure fast reactions. At LDA
concentrations higher than 1.4 M the operation of the HPLC
pumps was unreliable due to clogging of the check valves,
leaving an optimal concentration of about 1.25 M LDA in 2-
methyltetrahydrofuran. Commercial- grade LDA was later
abandoned since the quality was found to be too unreliable
between batches.13 Instead, LDA was prepared by adding 2.5 M
n-BuLi to a 2.75 M solution of diisopropyl amine in 2-MeTHF
at a temperature below 0 °C. The LDA solution was degassed
at 10 °C, and a pressure of 350 mmHg was applied to avoid loss
of pressure in the pumps due to the in situ formed n-butane.
The reaction was continuously quenched by mixing the reactor
output solution with an aqueous solution of NH4Cl (5%), NaCl
(15%), and AcOH in an equimolar amount to LDA (3.5 equiv)
in a batch reactor set at 10 °C. This mixture, together with
using 2-methyltetrahydrofuran as the reaction solvent aided
phase separation and made further extractions or drying of the
reaction mixture unnecessary.
1
running a H NMR assay (against benzyl benzoate as internal
standard), it turned out that typical yields for the reaction using
2 equiv of N-Boc-glycine methyl ester was 40−50% based on 6.
Still this was already on par with the previously used batch
conditions based on limiting 6. We noticed that the
temperature of the reaction reached 80 °C when measured
directly after the T-mixer, but after only a few seconds after
entering the temperate reaction zone the temperature was
measured to 25 °C, again displaying the superior heat transfer
Using these parameters, a 20 mL tube reactor and a retention
time of 2 min, 1.3 kg of phosphinate 6 (∼3.4 mol) was
926
dx.doi.org/10.1021/op200340c | Org. Process Res. Dev. 2012, 16, 925−929

Organic Process Research & Development
Article
Figure 3. Photo of reactor (first generation).
processed during a total run time of 25 h. Unfortunately, the
use of excess N-Boc glycine methyl ester (7) and formation of
colored side products made chromatography necessary. It was
found that the product was appreciably labile on silica, and a
quick elution using straight EtOAc was used to minimize losses.
Isolated yields of product 8 from the reaction varied as much as
between 24 and 54% from different chromatography batches,
largely depending on how rapidly the chromatography was
performed. This resulted in insufficient amounts of product
that could be isolated and a pressing need for further
optimization. In addition to the instability of 8 on silica gel,
it was also found to be thermally unstable. Within 10 min most
of a sample had deteriorated at 55 °C.
Satisfyingly, small-scale (15 g phosphinate 6) experiments
showed that productivity could be maintained using a 1:1 molar
ratio of glycinate 7 and phosphinate 6, resulting in 58% yield.14
To simplify handling the LDA solution, an in situ production of
LDA was developed for the next run. Thus, 2.5 M solutions of
n-BuLi and DIPA were mixed in a T-mixer and allowed to react
for a total of 13 s in a 2 mL tube reactor at 35 °C. This saved
several hours of reactor time for LDA formation, made
degassing unnecessary, and reduced potential variance between
LDA batches. The LDA solution (3 equiv) was directly reacted
with a 1:1 mixture of glycine and phosphinate, again in a 20 mL
reactor but using 90 s retention time. Another 1.3 kg of
phosphinate 6 was processed in this way, delivering sufficient
amounts of product 8 to proceed with the deprotection without
the need for chromatographic purification.
concentration the HCl-salt of 2 was dissolved in methanol and
propylene oxide was added to precipitate the zwitterionic
product. The resulting precipitate was isolated by filtration and
washed with methanol before drying. The crude product was
recrystallized from water and an organic solvent such as
acetone, methanol or ethanol. Even though purity analysis
indicated a purity of 99% the product was colored and
decomposed upon standing.15 To our satisfaction this could be
resolved by treatment of an aqueous solution of the product
with charcoal16 to remove colored impurities. The colorless
product obtained after charcoal treatment and crystallization
was found to be stable for several months under high humidity
and elevated temperatures. Thus, the required amount, 400 g,
of the final API 2 could be delivered in a quality suitable for
further pharmaceutical development.
CONCLUSION
■
A short and concise route that was deemed difficult to scale-up
was adapted to a continuous flow regime. Two of the steps
were successfully implemented while the last deprotection was
well suited for batch processing. The final preparation avoided
cryogenic conditions as well as had a more favorable starting
material stoichiometry and allowed chromatography to be
omitted. Productivity with respect to the commercially limiting
starting material was maintained and the final product was
obtained in a higher quality than earlier procedures, resulting in
a more stable API.
EXPERIMENTAL SECTION
Simultaneous deprotection of both protecting groups was
done by treatment with 3 M aqueous HCl, or alternatively with
HCl in an organic solvent such as isopropyl alcohol. After
■
Ethyl 1,1-Diethoxyethyl(methyl)phosphinate (6).17
Method a (Batch Procedure). Methyl dichlorophosphine (4,
927
dx.doi.org/10.1021/op200340c | Org. Process Res. Dev. 2012, 16, 925−929

Organic Process Research & Development
Article
19.5 g, 0.167 mol) was added to triethyl orthoacetate (480 mL,
2.6 mol) at −3 °C. After an initiation time of about 4 min the
temperature started to rise up to about 10 °C. More methyl
dichlorophosphine (43.5 mL, 56.6 g) was added over a 10 min
period while controlling the exothermic reaction so that the
temperature did not rise above 22 °C. After complete addition
the mixture was kept at 20 °C for 30 min and then all volatile
components were removed at 47 °C, first at low vacuum and
then at ∼1−2 mbar to give 6 (140 g, 98%) as a colorless oil.
Method b (Continuous Flow). Methyl dichlorophosphine
(4, 19.4 g, 0.166 mol, flow rate 0.140 mL/min) and triethyl
orthoacetate (81 g, 0.498 mol, flow rate 0.860 mL/min) were
mixed in a T-mixer and led through a PTFE tube reactor (inner
diameter 1 mm, 2 mL, 2 min retention time) at 25 °C. The
reactor was rinsed with toluene (10 mL), and the mixture was
concentrated under reduced pressure to give 6 (29.4 g, 79%) as
a colorless oil.
3-Amino-2-oxopropylphosphinic Acid (2). Hydrochlor-
ic acid, 37% (1.5 L) was added to isopropanol (3.6 L) under a
nitrogen atmosphere in a 10-L reactor equipped with an
overhead stirrer, and the mixture was cooled to 15 °C. tert-
Butyl 3-((1,1-diethoxyethyl)(ethoxy)phosphoryl)-2-oxopropyl-
carbamate (8, 682 g (80% purity), 1.43 mol) was dissolved in
isopropanol (1.0 L), and the resulting solution was added in
portions (in order to evacuate gaseous byproducts) over a
period of 45 min to the mixture of hydrochloric acid and
isopropanol. The mixture was heated to 20 °C and stirred
overnight; the solvent was then removed in vacuo at 33 °C and
the residue coevaporated with methanol (1 L) twice. The
residue was dissolved in methanol (2.5 L) and cooled to −26
°C. Propylene oxide (1.3 L, 19.7 mol) was added under a
nitrogen atmosphere over a period of 20 min. During the
addition the temperature rose to −16 °C. The mixture was
allowed to heat up to 20 °C over a period of 3 h and was then
stirred overnight at 20 °C. The resulting crystals were isolated
by filtration and washed with methanol (0.9 L). Drying in
vacuo gave 2 (202 g, 103%19).
1H NMR (400 MHz, CDCl3): d 1.06−1.13 (6 H, d, J = 7.1
Hz), 1.18−1.23 (3 H, t, J = 7.1 Hz), 1.32−1.40 (6 H, m), 3.49−
3.67 (4 H, m), 4.01−4.16 (2 H, m).
1H NMR (400 MHz, D2O): δ 3.01 (2 H, d, J = 18.1 Hz),
4.00 (2 H, s), 6.99 (1 H, d, J = 552 Hz) 31P NMR (161.9 MHz,
D2O): δ 16.0.
tert-Butyl 3-((1,1-diethylethyl)(ethoxy)phosphoryl)-2-
oxopropylcarbamate (8). First-Generation Synthesis. A
solution of N-Boc-glycine methyl ester (7, 1510 g, 8 mol) and
ethyl 1,1-diethoxy(methyl)phosphinate (6, 897 g, 4 mol) in 2-
methyltetrahydrofuran (1.6 L, total volume 4.0 L, i.e. 2 and 1 M
solutions, respectively, flow rate 2.86 mL/min) was continu-
ously mixed with lithium diisopropyl amine (10.0 L, 1.6 M in 2-
methyltetrahydrofuran, flow rate 7.14 mL/min) in a T-mixer
then led through a PTFE tube reactor (inner diameter 1 mm, 2
× 10 mL, 2 min retention time) at 35 °C. The reactor pressure
was controlled using a pressure regulator set at 2 bar.18 The
output solution was continuously quenched with an aqueous
solution of acetic acid (1.2 M), NaCl (15%) and NH4Cl (5%)
with a flow rate of 10.0 mL/min and collected in a reactor
thermostatted at 10 °C. The phases were separated and the
organic phase was concentrated. The crude product was
purified on silica (100% EtOAc) to give 8 (360 g, 24%).
Second-Generation Synthesis. A solution of diisopropyl-
amine (2.7 L, 19.2 mol) in 2-methyltetrahydrofuran (4.3 L,
total volume 7 L, flow rate 4.71 mL/min) was continuously
mixed with n-butyllithium (7 L, 2.5 M in hexane, flow rate 4.71
mL/min) in a T-mixer and then led through a PTFE tube
reactor (inner diameter 1 mm, 2 mL, 13 s retention time) at 35
°C. The resulting LDA solution (1.25 M, flow rate 9.42 mL/
min) was continuously mixed with a solution of N-Boc-glycine
methyl ester (7, 1046 g, 5.53 mol) and ethyl 1,1-diethoxy-
(methyl)phosphinate (6, 1292 g (96% purity), 5.530 mol) in 2-
methyltetrahydrofuran (3.2 L, total volume 5.5 L, 1.0 M
solution, flow rate 3.92 mL/min) in a T-mixer then led through
a PTFE tube reactor (inner diameter 1 mm, 20 mL, 90 s
retention time) at 35 °C. The reactor pressure was controlled
using a pressure regulator set at 2 bar.15 The output solution
was continuously quenched with an aqueous solution of acetic
acid (0.85 M), NaCl (15%) and NH4Cl (5%) with a flow rate
of 14 mL/min and collected in a reactor thermostatted at 10
°C. The phases were separated, and the organic phase was
concentrated in vacuo by coevaporation with heptane to give 8
Decolorization. A solution of raw 3-amino-2-oxopropyl-
phosphinic acid (2, 87 g, 0.63 mol) in water (350 mL) was
filtered through a prewetted column of activated carbon (70
g).16 The column was washed with water (400 mL), and the
combined filtrates were concentrated in vacuo to give a solid
residue (102 g). The material was dissolved in water (200 mL);
95% ethanol (200 mL) was added followed by some seeding
crystals. Additional ethanol (350 mL) was added dropwise over
a period of 30 min, and the resulting suspension was stirred at
15 °C for 1 h, filtered, and washed with ethanol (2 × 50 mL).
Drying in vacuo overnight gave 2 as off-white crystals (61.99 g,
71%) in two crops.
AUTHOR INFORMATION
Corresponding Author
*Telephone: +46 31 776 16 65. Email: fritiof.ponten@
astrazeneca.com.
■
Present Address
†FOI, Swedish Defence Research Agency, Cementvagen 20,
̈
SE-901 82 Umea, Sweden
̊
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
We thank Anna Stenemyr, Fabrice Odille, Johan Malmberg
(Chemical Science, AstraZeneca Sodertalje) and Anna Tivesten
(Medical Evaluation, AstraZeneca Molndal).
■
̈
̈
̈
REFERENCES
■
(1) Christer Alstermark, C.; Amin, K.; Dinn, S. R.; Elebring, T.;
Fjellstrom, O.; Fitzpatrick, K.; Geiss, W. B.; Gottfries, J.; Guzzo, P. R.;
̈
Harding, J. P.; Holmen
́
, A.; Kothare, M.; Lehmann, A.; Mattsson, J. P.;
Nilsson, K.; Sunden, G.; Swanson, M.; von Unge, S.; Woo, A. M.;
́
Wyle, M. J.; Zheng, X. J. Med. Chem. 2008, 51, 4315−4320.
(2) (a) Korenkov, A. I.; Niendorf, W. R.; Darwish, N.; Glaeser, E.;
Gaab, M. R. Neurosurg. Rev. 2002, 228−230. (b) Raphael, J.; Leo, R. J.;
Baer, D. Psychosomatics 2005, 503−507.
(3) For recent general reviews on continuous flow chemistry, see:
(a) Wirth, T. Microreactors in Organic Synthesis and Catalysis; Wiley-
VCH: Weinheim, 2008; (b) Hartman, R. L.; Jensen, K. F. Lab Chip
1
(682 g with purity of 80%; based on H NMR assay some
heptane and ethyl acetate remain in the material, 26%).
1H NMR (400 MHz, CDCl3): δ 1.14 (6 H, dt, J = 7.0, 1.9
Hz), 1.28 (3 H, t, J = 7.1 Hz), 1.38 (9 H, s), 1.46 (3 H, d, J =
12.5 Hz), 3.01 (1 H, dd, J = 16.9, 13.2 Hz), 3.17 (1 H, t, J =
14.2 Hz), 3.48−3.76 (4 H, m), 4.02−4.25 (4 H, m).
928
dx.doi.org/10.1021/op200340c | Org. Process Res. Dev. 2012, 16, 925−929

Organic Process Research & Development
Article
2009, 9, 2495−2507. (c) Mason, B. P.; Price, K. E.; Steinbacher, J. L.;
Bogdan, A. R.; McQuade, D. T. Chem. Rev. 2007, 107, 2300−2318.
(d) Geyer, K.; Gustafsson, T.; Seeberger, P. H. Synlett 2009, 15,
2382−2391. (e) Baumann, M.; Baxendale, I. R.; Ley, S. V. Mol. Divers.
A1, Priority Date Jan 5, 2006. (b) Fouger
́
e, C.; Erwann, G.; Hardouin,
J.; Lecouvey, M. Eur. J. Org. Chem. 2009, 34, 6048−54.
(18) Back Pressure Controller from Syrris FRX system. For further
information, see http://www.syrris.com.
(19) The higher than theoretical yield is presumably due to an error
in estimating the purity assay of 8 from the previous step.
2011, 15, 613−630. (f) Jahnisch, K; Hessel, V.; Lowe, H.; Baerns, M.
̈
̈
Angew. Chem., Int. Ed. 2004, 43, 406−446. (g) Luis, S. V.; Garcia-
Verdugo, E. RSC Green Chemistry No. 5: Chemical Reactions and
Processes under Flow Conditions; The Royal Society of Chemistry:
London, 2010. (h) Wiles, C., Watts, P. Micro Reaction Technology In
Organic Synthesis; CRC Press: Boca Raton, FL, U.S.A. 2011.
(i) Kirschning, A.; Solodenko, W.; Mennecke, K. Chem.Eur. J.
2006, 12, 5972−5990. (j) Wiles, C.; Watts, P. Eur. J. Org. Chem. 2008,
10, 1655−1671. (k) Hartman, R. L.; McMullen, J. P.; Jensen, K. F.
Angew. Chem., Int. Ed. 2011, 50, 7502−7519.
(4) Elebring, T.; Guzzo, P.; Holmen, A.; Olsson, T.; Swanson, M.;
von Unge, S. New Aminopropylphosphinic Acids. PCT Int. Appl. WO
0142252, Priority Date Dec 9, 1999.
(5) A Vapourtec R4/R2+ equipment with two or three pumps was
used. For further information, see http://www.vapourtec.co.uk/.
(6) The product 6 is somewhat volatile, and the slightly lower yield in
comparison to that from the batch process is presumably due to
coevaporation with toluene. For physical properties of 6, see: Zhang,
D.; Yuan, C. Chem.Eur. J. 2009, 15, 4088−4101.
(7) If the addition rate was too high, the temperature could be
controlled only with difficulty. At one point, we experienced a
temperature increase from 0 to 45 °C in approximately three seconds
before the reaction could be stabilized by cooling in a dry ice/acetone
bath.
(8) Strem Chemicals, Inc., 7 Mulliken Way, Newburyport, MA
01950-4098 U.S.A.
(9) In reference 1 the yield for the acylation step is given as 85%.
However, this yield is based on glycine ester 7, and if recalculated to
reflect use of phosphinate 6, the yield is 42.5%. It is known from later
batchwise scale-up work that the yield decreases further on larger scale
(unpublished results).
(10) LC or LC−MS techniques were not helpful since the product
was neither UV-absorbing nor prone to ionize (positively or
negatively) in the mass spectrometer.
(11) During processing, a continuous buildup of the pressure could
sometimes be observed. This could be avoided by rinsing the reactor
with 2-methyltetrahydrofuran when the reactor pressure approached
the set cutoff limit (20 bar).
(12) A 2 M solution in tetrahydrofuran/heptane/ethylbenzene,
Sigma-Aldrich, product number 361798.
(13) For the generation of 8, other conditions such as phase transfer
catalyst using sodium hydroxide/tetrabutylammonium hydrogen
sulfate in dichloromethane/water or potassium hydride as base instead
of LDA were tried. In our hands these experiments were unsuccessful.
By using KHMDS, NaHMDS, and LiHMDS (3.5 equiv) and 7 (2
equiv), yields of 9, 12, and 15% (by 1H NMR assay), respectively, were
obtained.
(14) The experimental setup that was used worked well in trial
experiments, up to 15 g of 6 in a 20 min run, but during prolonged
processing the isolated yields dropped. It can be speculated that the
thermal instability of 8 in combination with the strongly exothermic
reaction generating up to 80 °C on the outer surface of the T-mixer,
even though exposure was very brief, is accountable for this difference.
(15) No additional signals could be seen in 1H NMR purity assays for
either charcoal treated or untreated material. Since the product is a
zwitterion and contains no chromophores, HPLC analysis was difficult.
It was noticed that charcoal-treated product was in the form of white
crystals that were stable at 75% relative moisture at 45 °C for several
months, whereas untreated product was slightly tan and formed a dark
oil after approximately one week at 75% moisture and 45 °C. Thus, the
1
color, in addition to H NMR assays, served as a fair indicator of
purity.
(16) Darco G-60, Sigma-Aldrich, product number 242276.
(17) (a) Chebib, M.; Kumar, R. J.; Johnston, G.; Hanrahan, J.
Neurologically active compounds. PCT Int. Appl., WO 2006000043
929
dx.doi.org/10.1021/op200340c | Org. Process Res. Dev. 2012, 16, 925−929