Here, a large-scale feasible, chromatography-free process to purge triphenylphosphine oxide (TPPO) from the crude product of Mitsunobu and Wittig reactions has been developed. Divergence in physicochemical properties like polarity and solubility of TPPO against the product was utilized to precipitate TPPO directly from the reaction mixture and eliminate by simple filtration on a kilogram scale at a pilot plant with high purity of the product.
Here, a large-scale feasible, chromatography-free process to purge triphenylphosphine oxide (TPPO) from the crude product of Mitsunobu and Wittig reactions has been developed. Divergence in physicochemical properties like polarity and solubility of TPPO against the product was utilized to precipitate TPPO directly from the reaction mixture and eliminate by simple filtration on a kilogram scale at a pilot plant with high purity of the product.
Triphenylphosphine
oxide (Ph3P = O), commonly known
as TPPO, is a byproduct of very popular and widely used name reactions
such as Mitsunobu,[1] Wittig,[2] Staudinger,[3] Appel,[4] and Corey–Fuchs[5] reactions, as shown in Figure . The alkene derivatives can be prepared from the corresponding
aldehyde or ketone using triphenylphosphonium ylide (Wittig reagent)
in the Wittig reaction. The Mitsunobu reaction is the most powerful
tool for a chemist to construct C–O, C–N, C–S,
C–X, C–C, and S–S with high stereospecificity.
It involves triphenylphosphine (TPP), diethyl azodicarboxylate (DEAD),
or diisopropyl azodicarboxylate (DIAD) and an appropriate reactive
nucleophile with a broad range compatibility of functional groups.[6,7]
Figure 1
Triphenylphosphine
oxide (TPPO) as a byproduct in the renowned
popular named reactions.
Triphenylphosphine
oxide (TPPO) as a byproduct in the renowned
popular named reactions.However, the Mitsunobu
reaction constituting the significant and
persistent problem of product purification precludes its use at a
commercial scale. Product isolation is usually achieved by tedious,
expensive, and time-consuming column chromatography, which is a powerful
technique for purification of new chemical entities in almost all
medicinal chemistry laboratories; however, this purification technique
is not feasible for large-scale operations. Chromatography-free procedures
have been summarized in few reviews[8−10] such as modified phosphanes
and fluorous phosphane; modified reagents can be removed by acid or
base extraction by decomposition to volatile products or by phase-switching.
These methods suffer lack of commercial availability and large-scale
compatibility.
Results and Discussion
Given our
continued interest in the process development of NCE
of zidebactam analogues as β-lactamase inhibitors,[11,12] O-alkylated hydroxylamine is a crucial intermediate, fundamentally
obtained from the Mitsunobu reaction followed by Ing–Manske
hydrazinolysis (Figure ). We were interested in elaborating the “large-scale process-friendly”
conditions for the Mitsunobu reaction, which are used routinely by
medicinal chemists to synthesize novel molecules for functional group
transformations.
Figure 2
Retrosynthetic analysis for zidebactam analogues.
Retrosynthetic analysis for zidebactam analogues.As shown in Scheme , an attempt to obtain the nucleophilic substitution
effect of N-hydroxy phthalimide on the corresponding
tosyl 4(13) or mesyl 5(14) derivatives failed because of a substantial
low yield and an additional step. As these preliminary results were
not competitive, we did not pursue it further and decided to focus
on the Mitsunobu reaction.
Reagents: (a) NsF, DBU, toluene;
(b) CDI, NaBH4; (c) CH3C6H4SO2Cl, TEA; (d) CH3SO2Cl, TEA; (e)
PhthNOH, base, DMF.We reviewed the literature
for the postreaction precipitation of
TPPO to purify the product. Though the reported methods possess some
advantages, there are several disadvantages such as follows: (a) the
reaction solvent [tetrahydrofuran (THF), dichloromethane (DCM), acetonitrile,
alcohol] must be removed completely before work-up, (b) the product
must be soluble in highly nonpolar solvents such as cold hexanes or
diethyl ether or a mixture of both,[15] (c)
treatment with metal chlorides (ZnCl2[16] or MgCl2[17,18]) to obtain a TPPO–metal
complex, (d) treatment with acidic reagents such as acetic acid[19] or oxalyl chloride[9] to afford immiscible TPPO-AcOH or triphenylphosphonium chloride,
respectively, (e) an acid–base extraction of the product, and
(f) using alkylative trapping on Merrifield resin.[20,21]To overcome the above-mentioned disadvantages, it was decided
to
utilize the physicochemical properties of the product and byproducts
in purification. It is known that TPPO is almost insoluble in deionized
water, cyclohexane, petroleum ether, and hexane.[22] TPPO can be removed with EtOH,[23] MeOH,[24] and isopropyl alcohol (IPA).[25] It was also evident to take the advantage of
hydrogen bond-induced effective crystallization of TPPO[26] through the formation of cocrystals with several
impurities/byproducts specifically 1,2-isopropylhydrazine dicarboxylate.[27] Solubility screening of selected solvents for 6, TPPO, and TPPO-H2DIAD is tabulated in Table .
Table 1
Solubility
Screening
solubility (mg/mL)
compound-6
TPPO-H2DIAD
TPPO
solvent
RT
0 °C
RT
0 °C
RT
0 °C
2-propanol
insoluble
insoluble
60.4
13.3
46.5
25.4
tetrahydrofuran
124.5
101.0
157.5
61.0
169.7
105.0
cyclohexane
insoluble
insoluble
insoluble
insoluble
insoluble
insoluble
toluene
46.7
42.1
23.2
15.95
44.3
42.9
Fortunately,
after preliminary screening of solvents, it was observed
that coldtoluene triggers effective precipitation of triphenylphosphine
oxide-1,2-isopropylhydrazine dicarboxylate (TPPO-H2DIAD)
and was successfully removed from the crude product at a lab scale.
It was noted that the solubility of TPPO-H2DIAD increased
with an increase in temperature. Therefore, at first, TPPO-H2DIAD was precipitated followed by crystallization of the product
in a suitable solvent. Toluene[18,23,28] was a solvent by choice: relatively inexpensive, promotes crystallization
of weakly acidic derivatives with TPPO,[27] and avoids operational time/cost of distillation. In addition, this
change afforded almost a similar result like THF along with merits
of feasibility at large-scale (easily recoverable and reusable) usage.
The rate of the reaction was accelerated by increasing the reaction
temperature from room temperature (RT).Our goal was to find
an optimal reaction condition, which would
offer significant benefit in terms of purification through direct
precipitation of TPPO-H2DIAD from the reaction mixture.
Therefore, various conditions (combinations of solvents and temperature)
were screened, as described in Table .
Table 2
Screening of the Mitsunobu Reaction
on a 50 g Scale (Scheme )
entry
solvent
volume (mL)
temperature
(°C)
TPPO-H2DIAD (dry) by filtration (g)
product isolation
yield (%)
1
THF
250
0–5
0
column
71
2
toluene
500
25–30
0
column
70
3
toluene
500
–5–0
5.5
column
70
4
toluene
200
25–30
10
column
70
5
toluene
200
–5–0
75
IPA (125 mL)
68
It was identified that the concentration of the reaction
mass plays
a very crucial role. Precipitation of TPPO-H2DIAD is inversely
proportional to the concentration of the reaction mass; a significant
drop in precipitation was observed with a diluted reaction mass. The
tuning of lower temperature seemed to be promising for removing TPPO-H2DIAD directly from the reaction. The notable observation was
that TPPO-H2DIAD could not be removed at RT or higher temperature.
Also, multiple solvents and solvent mixtures were explored to precipitate
the desired product selectively. Highly pure 6 was readily
isolated from the crude by treatment of IPA followed by simple filtration.N-Boc-4-hydroxy-l-proline methyl ester
(1) was transformed to N-boc-4-fluoro-l-proline[29] (2) and
treated with 1,1′-carbonyldiimidazole (CDI)/sodium borohydride
(NaBH4) instead of popular pyrophoric lithium aluminum
hydride (LAH) to afford reduced alcohol 3, as shown in Scheme . The Mitsunobu reaction
was carried out in toluene at 0–50 °C. Precipitation of
TPPO-H2DIAD was induced by cooling the reaction mixture
along with continuous stirring. The solid was filtered and the filtrate
was concentrated to remove toluene. Large-scale feasible IPA was used
to crystallize 6 as a white solid at a pilot scale (Scheme ). Using a simple work-up process, most of the impurities
from the Mitsunobu reaction were successfully removed (Figure ). Quantitative % content (w/w)
of compound-6 and TPPO during the work-up process was
estimated by high-performance liquid chromatography (HPLC) and is
tabulated in Table .
Scheme 2
Pilot-Plant Synthesis of 6
Figure 3
Flowchart and the HPLC chromatograph of the work-up process for 6.
Table 3
Quantitative % Content
(w/w) of Compound-6 and TPPO by HPLC
% content (w/w)
compound-6
TPPO
reaction mixture
20.55
29.33
TPPO- H2DIAD
0.14
toluene
residue
24.86
6.25
pure compound-6
0.05
IPA residue
13.21
10.97
Flowchart and the HPLC chromatograph of the work-up process for 6.Notwithstanding
the popularity of the Wittig reaction, a similar
purification process was also investigated to prepare the alkene derivative.
TPPO was efficiently removed without column chromatography in the
Wittig reaction for the synthesis of vitamin A on a large scale. The
pure vitamin A acetate product could then easily be obtained via crystallization
after removing some of the heptane and TPPO separated from the MeOH/water
phase.[30] Relatively inexpensive cyclohexane
was used instead of n-heptane, THF, or DCM to isolate TPPO directly
from the reaction mass without utilizing alcoholic phase separation.
The keto derivative 8(31) was
obtained from hydroxyl derivative 7. However, the large-scale
validated Corey–Kim oxidation condition[32] developed in our lab at elevated temperature (−5
to −10 °C) was chosen instead of the routine cryogenic
(≥−30 °C) condition with an appropriate ratio of
toluene/DCM instead of TCCA/TEMPO. Potassium tert-butoxide was added to the suspension of the Wittig reagent in cyclohexane
followed by the addition of 8 at RT. The reaction mass
was cooled, filtered to eliminate TPPO, and the filtrate was washed
with brine. A cyclohexane layer was concentrated and treated with
hexane to remove the small amount of TPPO invariably present after
filtration. A hexane layer was concentrated under reduced pressure
to afford methylene derivative 9 at a pilot scale (Scheme ).
Reagents: (a) NCS/DMS/TEA; (b)
PPh3CH3Br, KOBu;
(c) 1M TBAF, THF; (d) PhthNOH, PPh3, DIAD.The described methodology to remove TPPO directly from
the reaction
mass may not be suitable for each and every product of these types
of reactions; indeed the concept is effective and will help chemists
to remove maximum TPPO directly from the reaction mass by additional
manipulations. It may be possible to envisage the current process
to purify products of Appel, Staudinger, and Corey–Fuchs reactions.
The utilization of this concept with other TPPO generating reactions
has not yet been tested, but it is anticipated that the described
methodology has potential for achieving a practical solution to remove
TPPO without column chromatography. In this paper, we have described,
along with full experimental details, the preparation of 6, 10 (Mitsunobu reaction), and 9 (Wittig
reaction).
Conclusions
The present method illustrates one of the
ideal approaches to isolate
products with the highest purity by implementing an appropriate solvent,
reaction concentration, and temperature depending on the polarity
and solubility of products to improve the process development of NCE.
However, precise conditions truly depend on the specific substrate
at hand. The process showcased product isolation at a pilot scale
without chromatography of two renowned named reactions wherein TPPO
removal is a bottleneck. We anticipate that the present process will
provide an initial isolation tool for other reactions that generate
TPPO as a byproduct.
A suspension of 2 (5.0
kg, 21.44 mol) in THF (25L) and CDI (5.3 kg 32.68 mol) was
added in 7 equal portions and agitated for 3 h at room temperature.
After completion of the reaction, CDI adduct was cooled at −5
to 0 °C. In another flask, NaBH4 (0.82 kg, 21.68 mol)
was added in 3 equal portions to water (15 L) at 0 to 5 °C under
agitation; the above chilled CDI adduct solution was added dropwise
at 5–10 °C and stirred for 3 h at room temperature. After
completion of the reaction, it was acidified with 50% HCl in water
(19 L) till pH ∼ 3–4. DCM (25 L) was charged and stirred
for 15 min. The DCM layer was then separated and washed with 5% aq
sodium bicarbonate (10 L) followed by water (15 L). The organic layer
was concentrated under reduced pressure to afford 3 (4.25
kg, 90%) as a pale yellow low melting solid. 1H NMR (a
mixture of rotamers, 400 MHz, DMSO-d6):
δ 5.31–5.18 (m, 1H), 4.77 (s, 1H), 3.77 (bs, 1H), 3.66–3.58
(m, 1H), 3.56–3.39 (m, 3H), 3.17–3.16 (s,1H), 2.27–2.15
(m, 2H), 1.40–1.36 (s, 9H); 13C NMR (125 MHz, DMSO-d6): 153.79–153.55, 94.25, 93.44, 92.88,
92.06, 78.74, 62.12–61.93, 57.90, 53.33–53.15, 33.78–33.08,
28.06; ESMS: 219.9 (M + H); HPLC: ∼94.77%; SOR [α]D20: −34.56
(c 0.5, MeOH).
To a stirred solution of 3 (4.0 kg, 18.26
mol) in toluene (16 L) in a jacketed flask, N-hydroxy
phthalimide (2.98 kg, 18.26 mol) and TPP (7.17 kg, 27.39 mol) were
charged at room temperature and cooled to 0–5 °C. DIAD
(5.53 kg, 27.39 mol) [Caution: DIAD is a health hazard
and causes acute toxicity; must be handled with appropriate precautions!]
was added dropwise at below 25 °C and stirred for 30 min. The
reaction mass was heated for 3 h at 50 to 55 °C. After completion
of the reaction, the reaction mass was cooled at −5 to 0 °C.
TPPO-H2DIAD (9.13 kg, wet) was filtered over PNF (Pressure
Nutsche Filter) and washed with chilled toluene (4 L). The organic
layer was distilled out under reduced pressure to get an oily mass
and treated with IPA (10 L) at 50–55 °C for 30 min. It
was then cooled, filtered, and washed with chilled IPA (3 L) followed
by drying under vacuum at 50–55 °C in a rotocone vacuum
dryer (RCVD) for 6 h to afford 6 (4.52 kg, 68%) as a
white powder. 1H NMR (a mixture of rotamers, 400 MHz, CDCl3): δ 7.83–7.75 (bs, 4H), 5.33–5.23 (m,
1H), 4.54–4.46 (m, 1H), 4.32–4.18 (m, 1H), 4.11–4.09
(m, 1H), 3.71–3.54 (m, 2H), 2.82–2.75 (s, 1H), 2.29–2.21
(s, 1H), 1.43–1.39 (s, 9H); 13C NMR (125 MHz, CDC13): δ 163.59, 163.38, 154.11–153.88, 134.54–134.42,
128.96–128.77, 123.49, 94.01, 93.23, 92.61, 91.84, 80.37–80.12,
54.42,53.74–53.11, 35.34–35.18, 34.33–34.17,
28.31–28.18; ESMS: 264.9 (M + H); HPLC: ∼99.50%; SOR
[α]D20: −57.18 (c 0.5, DCM).
To a stirred solution of 9 (50 g, 0.23
mol) in toluene (200 mL), N-hydroxy phthalimide (38.24
g, 0.23 mol) and TPP (90.39 g, 0.345 mol) were charged at room temperature
and cooled to 0–5 °C. DIAD (69.75 g, 0.345 mol) [Caution: DIAD is a health hazard and causes acute toxicity;
must be handled with appropriate precautions!] was added dropwise
at below 25 °C and stirred for 30 min. The reaction mass was
heated for 3 h at 50 to 55 °C. After completion of the reaction,
the reaction mass cooled at −5 to 0 °C. TPPO-H2DIAD (90.0 g, wet) was filtered and washed with chilled toluene (50
mL). The organic layer was distilled under reduced pressure to obtain
an oily mass (HPLC: 83.68%). Crude was purified by silica gel column
purification to afford 10 (58.17 g, 70%) as colorless
oil, which turned to a white solid upon extended exposure to high
vacuum. 1H NMR (400 MHz, CDC13): δ 7.83–7.75
(bs, 4H), 5.13–5.07 (m, 2H), 4.33–4.30 (m, 2H), 4.05–3.87
(m, 3H), 2.95–2.89 (m, 2H), 3.71–3.54 (m, 2H), 1.44–1.40
(s, 9H); 13C NMR (125 MHz, CDC13): δ 163.31,
154.06–153.67, 144.27, 143.03, 134.51, 128.90–128.77,
123.48, 108.74, 107.93, 80.02–79.8, 54.95, 50.97, 50.29, 35.52,
34.51, 28.27, 21.91; ESMS: 359.1 (M + H); HPLC: 95.08%.
A suspension of N-chlorosuccinimide
(3.63 kg, 27.19 mol) in DCM (9 L) and toluene (9 L) was agitated and
cooled to −15 °C. A solution of dimethyl sulfide (1.96
kg, 31.71 mol) in toluene (3 L) was added dropwise at below −5
°C. After 30 min, a solution of 7 (3 kg, 9.06 moles)
in toluene (24 L) was added dropwise at −5 to −10 °C
and stirred for 2 h. Triethylamine (3.21 kg, 31.71 mol) was added
at −5 to −10 °C by controlling the addition rate
and stirred for 30 min. After completion of the reaction, saturated
aq NaHCO3 (30 L) was added at −5 °C and stirred
for 30 min. The reaction was allowed to warm to room temperature.
The organic layer was separated and washed with 15% aq NaCl (20 L).
The organic phase was concentrated under reduced pressure to afford 8 (2 kg, 67%) as a pale yellow liquid. 1H NMR (a
mixture of rotamers, 400 MHz, CDCl3): δ 4.47–4.46
(m, 1H), 3.95–3.65 (m, 2H), 3.58–3.50 (m, 1H), 3.41–3.32
(m, 1H), 3.21–3.18 (m, 1H), 2.07–2.05 (m, 1H), 1.89–1.86
(m, 1H), 1.47–1.44 (s, 10H), 0.86 (s, 19H), 0.04 (s, 12 H); 13C NMR (125 MHz, CDC13): δ 154.58, 78.95,
70.46, 69.75, 64.43, 63.58, 57.67, 55.04–54.69, 38.16, 37.23,
28.50, 27.55, 25.84–25.74, 18.16–17.97; SOR [α]D20 + 3.2 (c 1.0,
CHCl3).
To a stirred suspension of methyltriphenylphosphonium
bromide (2.38 kg, 6.66 mol) in cyclohexane (10 L), potassium tert-butoxide (0.75 kg, 6.66 mol) was added in 3 equal portions
(slight exotherm observed) to obtain a bright yellow color suspension
and stirred for 1 h at RT. A solution of 8 (2 kg, 6.06
moles) in cyclohexane (6 L) was added dropwise at below 30 °C
and stirred for 4 h. After completion of the reaction, the reaction
mass was cooled to 0 °C, filtered through celite, and washed
with cyclohexane (2 L). The filtrate was washed with 15% aq NaCl (2
× 10 L). The organic phase was concentrated under reduced pressure
and agitated with hexane (3 L) for 30 min. It was then filtered and
washed with hexane (1 L). The obtained filtrate was concentrated under
reduced pressure to afford 9 (1.45 kg, 73%) as a pale
yellow liquid. 1H NMR (a mixture of rotamers, 400 MHz,
CDC13): δ 4.95–4.89 (m, 2H), 4.05–3.91
(m, 2H), 3.82–3.79 (m, 1H), 3.63–3.57 (m,1H), 3.37–3.36
(m,1H), 2.57–2.54 (m, 2H), 1.45–1.43 (s, 11H), 0.86
(s, 12 H), 0.01 (s, 8H); 13C NMR (125 MHz, CDC13): δ 154.59, 154.02, 145.80, 144.49, 107.07, 106.22, 79.41–79.18,
70.44, 69.73,64.43, 36.57, 63.25, 58.30–57.68, 55.04–54.73,
51.43, 50.75, 38.15, 37.27, 34.87, 34.19, 28.49, 25.84–25.72,
18.15–17.95; Mass: 328.0 (M + H); SOR [α]D20: −22.0
(c 1.0, CHCl3).