2584 J . Org. Chem., Vol. 65, No. 8, 2000
Notes
Ta ble 1
soluble in the alcohols used in this study, failed to yield
3 under these conditions.
time to
Other PTAs that were investigated were TBAH as a
40% solution in methanol (TBAH-2), benzyltriethylam-
monium hydroxide as a 40% solution in methanol (BTAH),
tris[2-(2-methoxyethoxy)ethylamine (TMEA), and tetra-
n-butylphosphonium bromide (TBPB). Because it is a
nonbasic PTA, reactions involving TBPB required ad-
ditional base (method II). As shown in Table 1 (entries
6-10), all PTAs investigated gave good yields of 3,
inferring that the nature of the PTA is of little conse-
quence in this procedure.
We also investigated the effect of other radical initia-
tors. In this study, 2ME was used as solvent, with or
without TBPB as the PTA. We again observed that AIBN
and ACVA, which are both soluble in 2ME, gave good
yields of 3 in the absence of TBPB. Addition of the PTA
nonetheless accelerated the reaction, cutting completion
time in half (entries 11-13). On the other hand, azobis-
(cyclohexanecarbonitrile) (ACCN) and 2,2′-azobis[2-meth-
yl-N-(2-hydroxyethyl)propionamide) (AMHP), which have
very little solubilities in 2ME, required addition of PTA
to yield tractable reactions in reasonable time (entries
15 and 16). The attempted deoxygenation with benzoyl
peroxide in 2ME was unsuccessful.
We conclude that a versatile but robust procedure has
been developed for deoxygenations, which can be used
economically on a large scale. Thus, method III was used
to convert 15 kg (17.2 mol) of 4 to 3 (30 min, 90% yield
after crystallization from MeOH/H2O), toward prepara-
tion of 1. Furthermore, the process may be of general
utility in the deoxygenation of compounds bearing acid
sensitive functionalities. As an example, 1,2:5,6-di-O-iso-
pr opylidin e-3-O-(pen t a flu or oph en oxy)t h ion oca r bon -
ylglucofuranose was efficiently deoxygenated (10 min,
69% yield) in 2ME, using TBAH as the PTA.
completion yield
entry solvent
PTA
initiator method
(min)
(%)
1
2
EtOH
PrOH
TBAH
TBAH
ACVA
ACVA
ACVA
ACVA
ACVA
III
III
III
III
III
III
III
III
III
III
I
120
50
120
20
10
70
70
60
70
60
40
20
20
10
60
60
70.7
94.0
92.4
80.0
95.1
83.4
89.3
90.7
87.8
90.1
83.6
95.0
77.2
75.3
71.5
86.1
3
i-PrOH TBAH
4
2ME
5
6
7
8
2ME
PrOH
PrOH
PrOH
PrOH
PrOH
2ME
2ME
2ME
2ME
2ME
2ME
TBAH
TBAH-2 ACVA
TBAH
BTAH
TMEA
TBPB
ACVA
ACVA
ACVA
ACVA
ACVA
ACVA
AIBN
9
10
11
12
13
14
15
16
TBPB
II
I
TBPB
TBPB
TBPB
AIBN
II
ACCN
AMHP
II
II
radical initiator were attained, it may be possible to
develop an inexpensive and robust process for the prepa-
ration of 3. We therefore studied the effect of solvents
and phase-transfer agents on the conversion of 4 to 3
using NaH2PO2 and different radical initiators. We
hereby report the results of our studies, which have
yielded an efficient and robust process for the synthesis
of 3 on a multi-kilogram scale.
Resu lts a n d Discu ssion
The effect of solvents was first studied using ACVA
with or without the addition of 1 M tetrabutylammonium
hydroxide in methanol (TBAH) as the phase-transfer
agent (PTA). In EtOH, PrOH, and i-PrOH in which NaH2-
PO2 was only sparingly soluble, reactions without the
PTA were quite sluggish (>3 h completion time), yielded
a number of byproducts, and were, in some cases,
intractable. In the presence of TBAH, however, these
reactions proceeded smoothly to afford 71 to 94% yields
of 3 (entries 1-3, Table 1). The reaction was significantly
faster in the higher boiling PrOH. 2-Methoxyethanol (2-
ME) gave a more homogeneous reaction mixture and
cleanly yielded 3 (20 min, 80%) in the absence of TBAH
(entry 4). In the presence of TBAH, the reaction in 2ME
was accelerated (10 min) and afforded a 95% yield (entry
5). This observation suggests that solubility of reactants
in the solvent or homogeneity of the reaction mixture is
an important parameter in this reaction. Furthermore,
polar protic solvents are required, as aprotic solvents,
such as dioxane, propyl actetate, acetonitrile, and toluene
failed to give any useful reactions, with or without PTAs.
Exp er im en ta l Section
Compound 4 was prepared as previously described.1 All
solvents were reagent grade. ACVA and AMHP were from Wako
Chemicals USA, Inc. (Richmond, VA), while AIBN and ACCN
were obtained from Aldrich Chemical Co., Inc. (Milwaukee, WI).
TBAH-2, BTAH, and TBPB were obtained from Lancaster
Chemicals (Windham, NH), while TBAH and TMEA were
purchased from Aldrich. Sodium hypophosphite monohydrate
was purchased from Aldrich/Fluka Chemicals (Milwaukee, WI)
and from Occidental Chemical Corp. Reactions were monitored
by HPLC for the disappearance of 4. Unless otherwise stated,
the yield of 3 was determined by HPLC of the crude product
obtained after deacetylation in MeOH. Microanalyses were
performed by Robertson Microlit Laboratories, Inc. (Madison,
NJ ).
Meth od I. Exa m p le of a Rea ction w ith ou t a P h a se-
Tr a n sfer Agen t. 4,4′-Azobis-(4-cyanovaleric acid) (1.61 g, 5.75
mmol) was dissolved in cold 2-methoxyethanol (10 mL) and the
pH adjusted to 8.0 with Et3N. A portion (2/3) of the above
solution was added, under nitrogen, to a refluxing mixture of
NaH2PO2 (3.05 g, 28.75 mmol) in the solvent (50 mL). A warm
(45 °C) solution of 4 (5 g, 5.75 mmol) in the desired solvent (15
mL) was added to the refluxing mixture. The remainder of the
initiator solution was added portionwise to the reaction mixture
over 40 min and the reaction monitored to completion. The
reaction mixture was cooled to 45 °C, pH was adjusted to 8 by
addition of 10% aqueous NaHCO3, and solvents were removed
in vacuo. The residue was partitioned with EtOAc (50 mL) and
H2O (50 mL). The organic layer was separated and further
washed with H2O (50 mL) and evaporated in vacuo to afford a
residue that was redissolved in MeOH (25 mL). The solution
was heated to 50 °C for 8-10 h and MeOH removed to yield
crude 3.
The reaction in PrOH also gave a new thionocarbamate
byproduct, which we characterized as the n-propyl de-
rivative (6), obviously a product of thionoimidazole sol-
volysis. We later synthesized 2′-O-acetyl-4′′-O-(1-pro-
pionoxy)thionocarbonyl erythromycin B (7) and demon-
strated that it did not undergo deoxygenation at the 4′′-
position, under the conditions of these experiments.
Hence, the alcoholic solvents did not participate in these
reactions, but merely served to attain the solubility of
the reactants and stability of the radical anion necessary
for the deoxygenation with the inorganic salt. A further
observation is that whereas NaH2PO2 may not be the only
inorganic hypophosphite salt that can effect this reaction,
it is important that the salt be, at least, sparingly soluble
in the alcoholic solvent. Thus, MnHPO2, which is in-