COMMUNICATION

pubs.acs.org/OPRD

The Safe Use of Sodium Hydride on Scale: The Process Development of

a Chloropyrimidine Displacement
Jamie M. McCabe Dunn,*, Alicia Duran-Capece, Brendan Meehan, James Ulis,|| Tetsuo Iwama,^
Guy Gloor,z George Wong, and Evan Bekos
Process Chemistry, Merck Research Laboratories, Merck Sharp & Dohme Corporation, 126 East Lincoln Avenue, Rahway, New Jersey
07065, United States

Process Chemistry, Merck Research Laboratories, Rahway, New Jersey 07065, United States

Analytical Chemistry, Merck Research Laboratories, Rahway, New Jersey 07065, United States

Analytical Chemistry in Development and Supply, Merck Manufacturing Division, Summit, New Jersey, 07901
)

Global Pharmaceutical Commercialization, Merck Manufacturing Division, Rahway, New Jersey, 07065
^
Mochida Pharmaceutical Co., Ltd., Jimba, Gotemba, Shizuoka 412-8524, Japan
z
Chesilton LLC, Colts Neck, New Jersey, 07722

bS Supporting Information
Scheme 1. Reaction of chloropyrimidine 1 with alcohol 2 to
ABSTRACT: Sodium hydride was found to be the best base provide pyrimidine 3 using sodium hyrdride as the base
for a displacement reaction on a chloropyrimidine. Due to
the insolubility of sodium hydride in all solvents and high
reactivity with atmospheric moisture, solutions dealing with
the associated safety concerns on kilogram scale are de-
scribed. The use of sodium hydride in premeasured dis-
solvable bag packaging (SecuBags), online process analytical
technology (PAT) to monitor the hydrogen o-gas, and the
development of analytical methods to monitor residual and (3) assurance of a complete quench of excess NaH prior to
polymer and potential degradation products allowed for a continuing the process.
successful scale-up and on-time delivery of a quality product.
SAFE HANDLING OF SODIUM HYDRIDE
Equipment Setup. A flush bottom drain valve reactor was
INTRODUCTION chosen to prevent collection of solid NaH at the base of the
We recently required kilogram quantities of pyrimidine 3 reactor. Dedicated cleaned and solvent-rinsed pumps and lines
starting from chloropyrimidine 1 and alcohol 2 (Scheme 1). The were used for all transfers. A pressure check of the reactor and the
initial conditions provided to the process research group by the shell side of the process condenser ensured no leaks were present
medicinal chemistry group were the use of 3.2 equiv of sodium in the condenser coil. As an added precaution, the reflux
hydride (NaH) in reuxing tetrahydrofuran (THF) for 24 h. Our condenser was set to distill mode to avoid wash back to the
rst objective for the process development of this step was to batch during the reaction. The reactor was dried and freed from
avoid the use of NaH. While some of our industry counterparts reactive solvents by refluxing with 2-methyl tetrahydrofuran
have demonstrated the use of sodium hydride on scale, we (MeTHF). The reactor contents were rigorously tested for water
historically had avoided the use of NaH on kilogram scale due content by KF and for residual methanol and acetone by GC
to its inherent safety concerns: insolubility in all solvents and prior to addition of NaH to the reactor.3 Lastly, a policy was
high reactivity to atmospheric moisture.1,2 For this reason, established that the suspension of NaH would not be transferred
alternative bases such as NaHMDS, BuLi, hexylLi, t-BuOK, into or out of the reactor prior to quenching with the exception of
t-BuONa, Cs2CO3, and CsOH were screened. Unfortunately, reaction monitoring;4 this ensured that the pumps and lines
these bases aorded low yields and unacceptable impurity would not become contaminated with sodium hydride.
proles. Because of an extremely tight delivery timeline, we Charging of Sodium Hydride. Solid NaH is typically pack-
chose to use NaH in our rst campaign. Since sodium hydride is aged as a 60% dispersion in mineral oil to minimize the risk of
insoluble in all solvents and highly reactive to moisture, we exposure to atmospheric moisture as well as improved handling
initiated the development of a protocol for the use of NaH on by decreasing the dusty nature of the dry power. Solid NaH will
scale. Three main areas needed to be addressed. (1) The physical
handling of the sodium hydride from the time it enters the Special Issue: Safety of Chemical Processes 11
reactor until it is quenched, (2) controlling the release of H2 Received: April 23, 2011
during both the deprotonation and quenching of the reaction,
Published: September 30, 2011

r 2011 American Chemical Society 1442 dx.doi.org/10.1021/op200114t | Org. Process Res. Dev. 2011, 15, 14421446
Organic Process Research & Development COMMUNICATION

Scheme 2. Depiction of developed reaction process

produce about 1 L of H2 gas per gram of NaH in the presence of a Operationally, the NaH-containing SecuBags were charged to
proton source. In addition, the associated exotherm provides the empty reactor via the manhole. The reactor was purged of air
sufficient heat to ignite the evolved H2 gas. The risk of ignition by rst pressurizing the reactor with nitrogen and then venting
during the addition of NaH to the reactor was eliminated by the vessel. This cycle was repeated ve times. For the remainder
using NaH in dissolvable bag packaging (SecuBags) from of processing, a nitrogen sweep was applied using a set ow of
Chemetall Foote Corp.5,6 The risks associated with waste nitrogen. Vacuum was not applied to eliminate the risk of the
disposal of unused NaH were also eliminated since the bags SecuBags bursting under negative pressure. The MeTHF was
were purchased in premeasured quantities; the entire bag was added, and agitation started after 10 min. The SecuBags immedi-
used in the reaction. ately dissolved upon addition of MeTHF to give an indicative gray
Reaction Process Design and Hydrogen Release during suspension. This suspension was heated to 80 C, and a solution of
the Deprotonation Stage. Our attention now turned to the alcohol 2 in MeTHF was added dropwise, followed by a suspen-
design of the reaction process. When designing our process, we sion of chloropyrimidine 1 in MeTHF. The deprotonation of
needed to incorporate the SecuBags into the reaction and control starting materials 1 and 2 at 80 C was instantaneous; this limited
the release of H2 during the deprotonation stage. SecuBags are the accumulation of the proton source and ensured that the rate of
completely soluble in cyclohexane, toluene, tert-butylmethyl- H2 release was directly controlled by the rate of the addition of
ether (MTBE), THF, MeTHF, and dimethylacetamide.7 Experi- each intermediate. The H2 released during the deprotonation was
mentally, our solvent choice was also restricted by the limited diluted with nitrogen and vented.
solubility of the chloropyrimidine 1. We determined that
THF and MeTHF were the best solvents for the displacement
reaction, and ultimately we chose MeTHF due to its higher QUENCH
boiling point and greener properties. In MeTHF at 80 C Having developed a safe protocol to deprotonate the starting
the reaction was complete in less than 15 h vs 24 h when materials, our focus turned towards how to safely quench the
THF was used. The developed reaction process is depicted excess sodium hydride upon reaction completion. The process
in Scheme 2. utilized 3.2 equiv of NaH even though experimentally, >2.4 equiv
1443 dx.doi.org/10.1021/op200114t |Org. Process Res. Dev. 2011, 15, 14421446
Organic Process Research & Development COMMUNICATION

Figure 1. Graphical plot of hydrogen o-gassing during the quench of

excess NaH. Start = the start time of wet MeTHF addition; Finished =
the time that the wet MeTHF addition nished.

was required for complete consumption of starting material.8 We

used 3.2 equiv because we did not split open the SecuBags
containing NaH. As 2 equiv of NaH is consumed during the Figure 2. Size exclusion chromatograph with UV detection at 215 nm.
deprotonation of starting materials 1 and 2, 1.2 equiv of active
NaH remained to be quenched. Typically, the remaining equiva-
lents would be quenched at ambient temperature with isopro-
panol. This would provide a safe, slow release of the evolving H2
gas. For this case, isopropanol was not a good crystallization
solvent for the desired pyrimidine product 3.9 Alternatively,
when wet MeTHF was used, which was prepared by addition
of 4% water to MeTHF, an immediate quench of the excess NaH
occurred. Further, the use of MeTHF provided consistent
isolation yields and purity verses the addition of isopropanol.
As in the deprotonation, the rate of H2 released during the
quench was easily controlled by the rate of addition of the wet
MeTHF.10
Use of Process Analytical Technology (PAT): Ensuring All
NaH Was Quenched. A major concern was ensuring that all of
the excess NaH was quenched prior to the addition of copious
amounts of water in the extraction stage. An online mass
spectrometer was placed in the vent of the reactor to monitor
the headspace contents. With a set positive flow of N2 through
the reactor and reactor vent, the mass spectrometer was tuned to
measure the ratio of N2 and H2 flowing through the vent during
the quench. The online mass spectrometer thus indicated the end
point of H2 release. The results from the first batch are depicted
in Figure 1. At the start of the addition of the 4% wet MeTHF to
the reactor, the mass spectrometer responded with an increase of Figure 3. Analysis of batch samples of product 3 using size exclusion
H2 detection. The ratio of H2 to N2 continued to increase until column to measure residual SecuBags.
the rate of addition of the wet MeTHF was slowed. At that point,
as expected, a drop in the H2 off gassing was detected. Continu- The product was isolated by filtration, while the bag polymer
ing the addition of the wet MeTHF at a constant rate, the graph remained in solution. An alternative method to remove the bag
shows a gradual increase and then decrease of the H2 off-gassing. material from the reaction mixture would be to bring the product
Since an excess of water via wet MeTHF was added with respect into the aqueous layer and separate the organic layer; however,
to the amount of active NaH remaining in the reactor, the time at our product could not be pulled into the aqueous layer. Thus,
which the slope of the plot is at 0 (approximately 2 h) indicated after quenching with the 4% wet MeTHF, the reaction was
that all the NaH was sufficiently quenched before the addition of concentrated via distillation, diluted with an antisolvent, and
wet MeTHF finished after approximately 3 h. acidified with 10% HCl solution (Scheme 2).11 MTBE was
Assuring Purity of Pyrimidine 3: Analytical Analysis. Lastly, chosen as the antisolvent for pyrimidine 3, as the SecuBag was
we needed to make sure we efficiently and effectively removed completely soluble in MTBE. The reaction mixture was filtered,
the SecuBags polymer from the isolated product. The conversion and pyrimidine 3 was washed with MTBE and water. Finally, the
of reagents to product proceeded as a slurry-to-slurry reaction. dry cake of pyrimidine 3 was tested for SecuBags content.
1444 dx.doi.org/10.1021/op200114t |Org. Process Res. Dev. 2011, 15, 14421446
Organic Process Research & Development COMMUNICATION

product. The collaborative eort between process chemistry,

pilot-plant operations, early development engineering and pro-
cess safety, and early development analytical research allowed for
a successful scale-up of pyrimidine 3.

EXPERIMENTAL SECTION
General Procedure. A 60% dispersion in oil of sodium
hydride in MeTHF soluble SecuBags (166 mol, 3.2 equiv) was
charged to a dry reactor, and MeTHF (70 L) was added. The
resulting gray suspension was heated to 75 C (internal
temperature), and a solution of alcohol 2 (52 mol, 1 equiv) in
MeTHF (70 L) was slowly added to the reactor followed by the
addition of a suspension of chloropyrimidine 1 (52 mol, 1 equiv)
in MeTHF (140 L). The batch was held at 75 C for 15 h and
then cooled to 25 C. A solution of 4% wet MeTHF (42 L) was
slowly added to the reactor. After complete addition of the wet
MeTHF the batch was concentrated under reduced pressure until
70 L of MeTHF remained in the reactor. Then MTBE (98 L),
water (28 L), and 3 M HCl (42 L) were sequentially added to the
reactor, and the batch was cooled to 10 C and agitated for 3 h.
The biphasic mixture was filtered, and the filter cake was washed
Figure 4. Analysis of batch samples of product 3 using headspace gas with MTBE (56 L) and then water (56 L). After drying in a
chromatography with FID detection to monitor residual styrene. vacuum oven at 60 C for 15 h, pyrimidine 3 was obtained in
75% yield.
The SecuBag is made of styrene butadiene styrene (SBS)
block copolymers. To ensure the SecuBag was suciently re- ASSOCIATED CONTENT
moved during the ltration, analytical methods were developed to
test for residual SecuBag and styrene in the isolated product. bS Supporting Information. Comments to a reviewers
An analytical method that was developed to monitor residual question. This material is available free of charge via the Internet
SecuBags polymer used size exclusion chromatography with UV at http://pubs.acs.org.
detection at 215 nm. The method resolves SecuBags polymer
from the starting materials 1 and 2, the pyrimidine 3, Marcol AUTHOR INFORMATION
82,12 and the diluent (Figure 2). The batch samples were
analyzed at a concentration of 0.025 g/mL for pyrimidine 3 with Corresponding Author
a detection limit of 30 ppm. Analysis of each batch sample of [email protected].
product 3 is depicted in Figure 3. The batch samples contained
SecuBags polymer in the range of 120 250 ppm with respect to
total weight of sample analyzed, which was well below the ICH ACKNOWLEDGMENT
recommended level of 1000 ppm. We thank David Andrews for guidance and support during the
Residual Styrene in Product. We also evaluated the isolated process development and Jennifer Albaneze-Walker, Kevin
product 3 for the presence of residual styrene as a result of Campos, Ephraim Bassan, and David Tschaen for reviewing.
polymer degradation. Residual styrene would negatively impact
the quality of the product, since the ICH allowable level for REFERENCES
styrene was 125 ppm based on our projected human dose. (1) Shimizu, H.; Shimizu, K.; Kubodera, N.; Tetsuhiro Mikami, T.;
Residual styrene was monitored using headspace gas chroma- Tsuzaki, K.; Suwa, H.; Harada, K.; Hiraide, A.; Shimizu, M.; Koyama, K.;
tography with FID detection. This method was capable of Ichikawa, Y.; Hirasawa, D.; Kito, Y.; Kobayashi, M.; Kigawa, M.; Kato,
resolving styrene from ethanol, reaction solvents (MeTHF and M.; Kozono, T.; Tanaka, H.; Tanabe, M.; Iguchi, M.; Yoshida, M. Org.
MTBE), and the diluent. Analysis at a concentration of 0.1 g/mL Process Res. Dev. 2005, 9, 278. Ford, J. G.; Pointon, S. M.; Powell, L.;
for pyrimidine 3 with a detection limit of 5 ppm styrene (0.5 g/ Siedlecki, P. S. Org. Process Res. Dev. 2010, 14, 1078.
mL) demonstrated that styrene was not detectable in any of the (2) Urben, P. G. Metal Hydrides. Brethericks Handbook of Reactive
four batches produced by the described process (Figure 4). Chemical Hazards, 7th ed.; Elsevier: Burlington, MA, 2007; Vol. 2, pp
226 227.
(3) Methanol and acetone are generally used to clean the reactors
CONCLUSION after a batch run; therefore, trace amounts of each could be present in a
Sodium hydride was safely utilized on kilogram scale by using a clean reactor.
(4) Reaction completion was monitored by sampling through the
variety of controls: charging the sodium hydride in the dissol-
reactor dip-tube into a small, sealed glass jar. After analysis, the sample
vable SecuBags for safe handling, use of online process analytical was disposed in the lab by quenching with isopropyl alcohol.
technology to monitor the hydrogen o-gassing in the process (5) Fritz, S.; Feldmann, R. Soluble lm for packaging air-sensitive
vent during the reaction and workup, and development of and/or moisture-sensitive hydride reagents. German Patent 196 21 661,
analytical methods to monitor residual polymer and potential Aug. 28, 1997. SecuBags is a registered trademark. At the time of the
degradation products from the dissolvable bag in the isolated batch NaH in SecuBags was only available in 1 kg or 5 kg quantities;

1445 dx.doi.org/10.1021/op200114t |Org. Process Res. Dev. 2011, 15, 14421446

Organic Process Research & Development COMMUNICATION

however, NaH in SecuBags is currently available in 50 g and 100 g

quantities from Aldrich.
(6) The exotherm to which we are referring to is associated with the
reaction of solid NaH with moist air or alcohol vapor. Given that the
auto-ignition temperature of hydrogen in air is high (585 C), there is no
likelihood of the reaction mass heating to the extent it would ignite the
hydrogen, since the solvent heat capacity and heat of vaporization would
absorb the energy of either the desired or an undesired reaction prior to
reaching this temperature. The risk is that individual particles of solid,
nely divided NaH falling through humid air during charging (for
example, while being scooped from a drum and dropped into a funnel)
can reach a temperature sucient to ignite either the liberated hydrogen
or solvent vapor.
(7) Chemetall Lithium home page: http://www.chemetalllithium.
com (accessed on April 19, 2011).
(8) In the event that the project would move forward, we would
request from Chemetall Foote smaller quantities of NaH so that there
would not be a large excess of NaH left at the end of the reaction. Since
this was for a rst delivery and it was known that excess NaH did not
degrade pyrimidine 3, we did not request special packaging.
(9) Distillation of the mixed solvent system (isopropanol/MeTHF/
water) gave inconsistent isolated yields.
(10) A reviewer asked how we determined what the safe rate of
released hydrogen gas was with respect to the batch temperature; our
comment can be found in the Supporting Information.
(11) Typically, one would acidify and remove the aqueous layer
prior to concentration of the batch; however, in this case pyrimidine 3
was a slurry in MeTHF/water and would not separate from the aqueous
layer. Therefore, we could not remove the aqueous layer from the
suspended pyrimidine 3. In an eort to have better control of our solvent
ratio for crystallization and having determined via a stress test that our
product was stable to the resultant NaOH from the quench, the batch
was concentrated prior to the addition of MTBE and aqueous HCl.
(12) Marcol 82 is the mineral oil used to suspend the sodium
hydride, and the certicate of analysis states that it meets the EU/USP
and FDA purity requirements.Cheplen, M. Chemetall Foote Corp.,
Kings Mountain, NC, U.S.A. Personal Communication and COA
Attachment, 2009.

1446 dx.doi.org/10.1021/op200114t |Org. Process Res. Dev. 2011, 15, 14421446