Hydrogenation of 5-alkylidene-2,4-thiazolidiones on Pd/C catalysts under mild conditions: An alternative synthesis route to pioglitazone

Article abstract of DOI:10.1007/s10562-013-0976-8

The hydrogenation of 5-alkylidene-2,4-thiazolidiones was studied using Pd/C to establish an alternative process for pioglitazone synthesis. The reportedly sluggish reactivity was due to the high reaction temperature in formic acid. The conditions were improved to 1 atm at 296 K, which results in a quantitative product with a smaller amount of the catalyst. Graphical Abstract: [Figure not available: see fulltext.]

Full text of DOI:10.1007/s10562-013-0976-8

Catal Lett (2013) 143:495–500
DOI 10.1007/s10562-013-0976-8
Hydrogenation of 5-Alkylidene-2,4-Thiazolidiones on Pd/C
Catalysts Under Mild Conditions: An Alternative Synthesis Route
to Pioglitazone
Takashi Sugimura
Yasuaki Okamoto
•
Kanae Oie
Kenji Tanaka
•
Tomonori Misaki
Hiroyuki Mori
•
•
•
Received: 9 January 2013 / Accepted: 14 February 2013 / Published online: 5 March 2013
Ó Springer Science+Business Media New York 2013
Abstract The hydrogenation of 5-alkylidene-2,4-thiazo-
lidiones was studied using Pd/C to establish an alternative
process for pioglitazone synthesis. The reportedly sluggish
reactivity was due to the high reaction temperature in
formic acid. The conditions were improved to 1 atm at
to be the saturation process. In the literature, the hydro-
-
2
genation conditions for 1a are typically 50 kg cm
at
323 K in DMF, which give only 65 % yield of 2a [4]. The
catalyst employed therein was a large amount of Pd black
in a substrate/Pd with a molar ratio of 2.8 (mol/mol) [4]. A
quantitative reaction was only reported with a specially
prepared polymer-incarcerated Pd at room temperature,
although the substrate/Pd molar ratio of 3 was still low [5].
Similar results have been reported in patents [6–10],
Where the reactions were performed under pressured
2
96 K, which results in a quantitative product with a
smaller amount of the catalyst.
Keywords Pioglitazone Á Hydrogenation Á Catalyst
deactivation Á Formic acid Á Palladium catalyst
-
2
hydrogen (5–10 kg cm ) at 323–353 K with palladium
supported on activated carbon (Pd/C) or Pd black under
low substrate/Pd molar ratios of 5–16, which resulted in
1
Introduction
60–85 % yields. Homogeneous hydrogenations with an Rh
Tiazolidine analogues include effective drugs for nonin-
sulin-dependent diabetes mellitus. Among the numerous
analogues studied as medicines, currently pioglitazone (2a)
complex and NaBH reductions have also been reported
4
[11].
In this report, we address the essential problem of this
process and describe the development of a new hydroge-
nation procedure for this system using a commercial Pd
catalyst supported on activated carbon.
[
1] is commercially available, and troglitazone (2b) [2] was
available until 10 years ago. These analogues may be
obtained by the condensation of 2,4-thiazolidinedione and
the corresponding 4-alchoxybenzaldehyde followed by the
selective saturation at 5-exo-olefin (1 to 2 in the scheme);
however, this simple process has not been adopted for the
practical production of 2a [3]. The major drawback appears
T. Sugimura (&) Á K. Oie Á T. Misaki Á Y. Okamoto
Graduate School of Material Science, University of Hyogo,
3-2-1 Kohto, Kamigori, Ako-gun, Hyogo 678-1297, Japan
e-mail: sugimura@sci.u-hyogo.ac.jp
K. Tanaka Á H. Mori
Tsukuba Research Laboratories, Tokuyama Corporation,
4
0 Wadai, Tsukuba, Ibaraki 300-4247, Japan
Reduction of 5-alkylidene-2,4-thiazolidiones (1) to give 2
123

4
96
T. Sugimura et al.
2
Experimental
initial hydrogenation rate was determined around 15–25 %
conversion. The hydrogenation of 1c was performed in a
manner similar to that of 1a.
2
.1 Materials
Substrate 1a was obtained from Tokuyama Corp
mp = 443.9–444.6 K) and used for most runs. In some
cases, 1a was used after being recrystallized
mp = 445.8–446.1 K). Substrate 1c was prepared from
,4-thiazolidione and p-anisaldehyde and was recrystallized
(
3 Results and Discussion
(
Pd/C is a widely applicable and popular catalyst for
hydrogenation. Highly dispersed Pd metal particles on
carbon supports with extremely high surface areas show
prominent characteristics such as high activities, repro-
ducible high turnover numbers, and mostly undetectable
metal leaching into the product solution. In particular, Pd
metal particles of STD-type Pd/C catalyst are highly dis-
persed, with a mean diameter of 1.4 nm.
2
from a mixture of THF and toluene following the reported
method [6]; 1c: 92.7 % yield, pale-yellow solid,
1
mp = 493.5–493.6 K. H NMR data were identical to the
reported values [12]. In this study, the 5 % Pd/C employed
was 53 % wet STD-type (N.E. Chemcat, Japan) and was
stored at 280 K until use. The activated carbon support had a
2
-1
BET surface area of 1,007 m g , and the Pd surface area
Because 1a is poorly soluble in many solvents, except in
hot and highly polar solvents, systematic studies on the
effects of solvent and temperature for the hydrogenation of
1a are rare. Thus, we first studied a simpler and highly
soluble analogue 1c that is essentially identical to 1a at the
hydrogenation site. With the model substrate 1c, the effect
of the hydrogenation conditions could be investigated
without the solubility problem inherent to 1a. The hydro-
genation of 1c was performed in various solvents starting at
room temperature. First, the reaction was performed with
1c/(Pd/C)/solvent = 235 mg (1.0 mmol)/43 mg Pd/C (Pd:
10 lmol)/10 mL; the results are shown in Fig. 1. The
reactions, except for those performed in ethanol, were slow
but proceeded smoothly at 296 K. This result is somewhat
surprising because all the reported hydrogenations of 1a
used elevated temperature of 323–353 K [6–10]. However,
the hydrogenation in ethanol was an exception, which was
rapid in the initial stage, although the conversion of 1c
could not exceed 10 % accompanying with the character-
istic smell of hydrogen sulfide. These preliminary experi-
ments suggest that hydrogenation potentially proceeds at
room temperature irrespective of the solvent polarities;
however, as observed in the case of ethanol, suppression of
the catalyst deactivation during the reaction is a key factor
for achieving a good yield from the overall reaction.
Considering poor dissolving ability of toluene, 1,4-diox-
anes showing the moderate dissolving power and accept-
able reactivity was selected for the following reactions.
The temperature dependence of the hydrogenation of 1c
was studied in 1,4-dioxane, and the results obtained at
296–353 K are shown in Fig. 2. At the highest temperature
of 353 K, the initial rate was the fastest; however, the
hydrogenation quickly became sluggish and was almost
interrupted after 25 h, with a conversion less than 25 %.
The interruption was less pronounced at 333 K; however,
the rapid reaction did not continue after the conversion
reached 20 %, and the total conversion after 48 h was less
than that achieved at 313 K, which reached 40 %. The
2
-1
was 339 m g . Palladium was distributed uniformly
throughout the carbon particles (size = 23 lm) with 76 %
dispersion. TheotherPd/C catalysts were obtainedfrom N.E.
Chemcat (AER type, 1 % STD type, Pd/Al O , and ASCA)
2
3
and from Kishida Chemicals, Japan (10 %). The ASCA type
is known as a Pd–Pt (4.5 ? 0.5 %) bimetallic catalyst.
2
.2 Apparatus
Hydrogenation was performed under atmospheric pressure
using a temperature-controlled water bath, an efficient
magnetic stirrer (1,200 rpm), and a proper burette to measure
1
the consumption of hydrogen gas. H NMR data was recor-
ded on a JEOL ECA-600 using CDCl as a solvent and as an
3
internal standard (d 7.24). HPLC analysis was conducted
with an ODS column (YMC-ODS-A, 4.6 mm i.d. 9
1
50 mm), and 1.0 mL gradient elution was performed with
CH CN/0.1 M NH OAc/AcOH = 100/100/4 to 120/80/3.
3
4
Amounts of 1 and 2 were determined at 269 nm, and were
calibrated. Other side products were estimated to have sim-
ilar absorption coefficients to 2, and were not calibrated.
Conversion (%) was calculated as 100 9 2/(1 ? 2 ?
others).
2
.3 Hydrogenation
Wet 5 % Pd/C (46.4 mg, 22 mg of the actual amount, 1 mg
as Pd) and solvent (4 mL) was placed in a flat bottom flask
(
50 mL). The solvent used was toluene, 1,4-dioxane, eth-
anol, or HCOOH. Hydrogen was charged, and the reaction
mixture was then stirred for 15 min at 296 K. In some
cases, the flask was heated at 353 K for 30 min and sub-
sequently cooled to 296 K. A solution (6 mL) of the sub-
strate (typically 1.0 mmol, 356 mg 1a) was charged into
the flask ([subst]init = 0.1 M), and the reaction course was
monitored by HPLC, of which results were confirmed by
1
H NMR and hydrogen consumption measurements. The
1
23

Hydrogenation of 5-Alkylidene-2,4-Thiazolidiones
497
solvent ratio was varied while the substrate/Pd molar ratio
was maintained at 20, and the substrate ratio was varied
while the catalyst and solvent amounts were fixed at
10 lmol/10 mL (both at 296 K). Figure 3 shows the initial
rate and conversion after 24 h for the hydrogenation of 1c
with a constant substrate/Pd molar ratio. As long as the
substrate/Pd molar ratio was kept constant, the initial rate
was not significantly affected by the amount of the solvent.
Figure 4 shows the initial rate and conversion after 72 h for
the reaction with varied amounts of 1c. The initial reaction
-
1
-1
rate of 5 mmol g
h
at a substrate/Pd molar ratio of 10
greatly decreased as the amount of the substrate was
increased, and the rate calculated as a function of the
amount of catalyst appeared to become constant at a low
substrate/Pd molar ratio of 100 (0.1 M). The conversion
was also drastically affected, and 100 % conversion of 1c
was attained at a low substrate concentration (\0.02 M), as
shown in Fig. 4. We concluded that a full conversion was
not achievable even after 72 h, when the substrate/Pd
molar ratio was greater than 40.
Fig. 1 Conversion/% of the hydrogenation of 1c under the standard
conditions (1c/Pd/solvent = 1.0 mmol/10 lmol/10 mL at 296 K) in
different solvents
catalyst during the hydrogenation should be poisoned by a
compound generated from the reactant. The poisonous
compound must be some sulfur compound, but MS anal-
A comparison of the results in Figs. 3 and 4 reveals that
the substrate/Pd molar ratio is a critical factor for the
hydrogenation of 1c. This result is not attributable simply
to reaction kinetics, such as the Langmuir–Hinshelwood
mechanism, that assumes the strong adsorption of 1c onto a
Pd metal surface, as expected from the results in Fig. 4 (a
negative reaction order with 1c), because the reaction rate
is relatively high and almost constant with a constant
concentration of 1c in the same concentration range (0th
order with 1c, Fig. 3). As previously suggested, sulfur
poisoning of the Pd metal surface due to the partial
decomposition of 1c might cause the deactivation of the
catalyst even at 296 K and become critical at a substrate/
1
ysis as well al H NMR analysis of the reaction mixture did
not suggest any structural information during and after the
hydrogenation in ethanol and in dioxane at elevated tem-
perature. Overall, the decomposition problem became
apparent even at 333 K, and a reaction temperature of
3
13 K resulted in the fastest kinetics for the reaction of 1c,
-
1
-1
g ) was
even though the initial rate (r = 0.6 mmol h
-
slower than that at 333 K (r = 1.5 mmol h
1
-1
g
).
A high concentration of the substrate, which indicates
the use of a smaller amount of solvent, may also be
desirable in industrial applications. We studied the con-
centration effects with 1c in dioxane in two ways: the
Fig. 3 Initial reaction rate (open circle, left axis) and conversion
(filled circle, right axis, after 24 h) for varied amounts of the solvent
1,4-dioxane, which ranged from 2.5 to 20 mL, under fixed amounts of
Fig. 2 Temperature dependence of the hydrogenation of 1c in
dioxane (1c/Pd/dioxane = 1.0 mmol/10 lmol/10 mL)
1c (0.2 mmol) and the catalyst (Pd: 10 lmol) at 296 K
123

4
98
T. Sugimura et al.
Table 1 Initial hydrogenation rate and reaction conversion for the
reaction of 1a, 1c, and their mixture
Run
1a
(
1c
(mmol)
r
(1a)
0
Conv.
(1a) (%)
r
0
Conv.
(1c) (%)
a
a
mmol)
(1c)
b
1
2
3
4
5
6
–
0.14
–
–
1.6
–
[99
b
0.14
0.14
(0.14)
0.28
(0.28)
[184
73
99
–
c
c
0.14
0.14
0.14
0.14
90
0.14
0.02
0.42
0.06
30
d
d
c
–
–
49
c
c
[184
–
84
45
c
–
34
Hydrogenation was performed at a substrate/Pd/HCOOH ratio of
0
.14 mmol/25 lmol/10 mL at 296 K
a
-1 -1 1
h ) determined by H NMR
Initial hydrogenation rate (mmol g
b
c
d
After 1 h
After 6 h
Fig. 4 Initial reaction rate (open circle, left axis) and conversion
filled circle, right axis, after 72 h) for varied amounts of 1c, which
ranged from 0.1 to 1 mmol, under fixed amounts of the catalyst (Pd:
0 lmol) and 1,4-dioxane (10 mL) at 296 K
The product 2a was added to the hydrogenation of 1c
(
1
catalyst molar ratio greater than 40. Another possibility is
that this result was due to coking by the decomposition of
1
c during hydrogenation.
With the basic properties of 1c in hand, the target mol-
ecule 1a was studied next. To clarify the difference between
a and 1c in the reactivity, the hydrogenations of 1a and 1c
1
were compared under a high catalyst-ratio condition of
substrate/Pd/HCOOH = 0.14 mmol/10 lmol/10 mL; the
results are summarized in Table 1. Notably, 1a and product
2
a were not fully soluble in formic acid, even under such
low substrate-concentration conditions. Surprisingly, the
reactivity of the hydrogenation of 1a on Pd/C was high
compared with that of 1c ([100 times), even though the
reaction sites were similar. Because the purity of the sub-
strate can affect the initial reaction rates, the difference was
confirmed with a 1:1 mixture (total 0.28 mmol, run 3) and a
Fig. 5 Initial reaction rate (open circle, left axis) and conversion
filled circle, right axis, after 6 h) for the hydrogenation of 1a at a 1a/
Pd/HCOOH ratio of 0.14–1.4 mmol/10 lmol/10 mL at 296 K
(
2
:1 mixture (total 0.42 mmol, run 5) of 1a and 1c. The
reaction of 1a was still rapid; however, the reaction of 1c
was suppressed to one-tenth or one-fourth the rate of 1a,
which indicated that 1a and/or product 2a adsorbed onto the
catalyst more strongly than 1c. More precisely, the sup-
pression of the hydrogenation of 1c was greater with 2a (1/
times faster than that of 1c at the initial stage (Table 1).
However, the decrease in the rate was dramatic, and the
complete conversion of 1a with a substrate/catalyst ratio
greater than 50 was difficult to achieve. Apparently, the
characteristics of the hydrogenation of 1a include the rapid
formation of 2a and its rapid deactivation. Notably, the
disubstituted pyridininium in 1a was inert for the hydro-
genation in all cases.
3
0–1/80) than with 1a. We concluded that the existence of
the 2,5-disubstituted pyridinium unit increased the adsorp-
tion strength and interrupted the hydrogenation of the
5
-alkylidene of the 2,4-thiazolidinedione of 1c and that the
degree of this disruption was much stronger with 2a than
with 1a; i.e. the adsorption strength decreases in the order
In addition to the 5 % STD-type Pd/C, several com-
mercial Pd catalysts were examined for the hydrogenation
of 1a. As shown in Table 2, 5 % STD-type Pd/C exhibited
a distinctly rapid initial hydrogenation rate and an optimal
conversion after 6 h. However, at least under these con-
ditions, a very rapid initial hydrogenation rate did not lead
to a high conversion, which suggests that more active
hydrogenation catalysts deactivate more rapidly. Overall,
2a [ 1a [ 1c, and thus the hydrogenation of 1a proceeds
by overcoming the interruption by 2a.
Figure 5 shows the result of the hydrogenation of 1a as a
function of the initial substrate concentration. At a low
concentration of 0.014 M, the hydrogenation of 1a was 100
1
23

Hydrogenation of 5-Alkylidene-2,4-Thiazolidiones
499
Table 2 Hydrogenation of 1a on Pd catalysts
-1
-1
a
Conversion (%)
Entry
Catalyst
r
0
(mmol g
h
)
1
2
3
4
5
6
5 % STD-type Pd/C (50 mg)
5 % AER-type Pd/C (50 mg)
58
10
82
75
69
57
77
61
Pd/Al
O
2 3
(50 mg)
4.8
7.0
6.4
7.5
10 % Pd/C (25 mg)
1 % STD-type Pd/C (250 mg)
5 % ASCA-2-type Pd–Pt/C (50 mg)
Hydrogenation was performed at 296 K in formic acid on the substrate 1a/Pd (or Pd ? Pt for ASCA-2)/solvent = 0.42 mmol/25 lmol/10 mL
a
After 6 h
temperature. However, the contribution of the poisoning
may not be dominant factor in the hydrogenation of 1a in
formic acid; i.e. when Pd/C in formic acid was warmed at
353 K under hydrogen for 30 min and then cooled and
used for the hydrogenation of 1a at 296 K, the catalytic
activity decreased by less than 1/100 (closed circles). The
catalyst heat treatment in dioxane did not deactivate the
catalysis so much (open squares vs. closed squares). These
findings clearly indicate that Pd/C is drastically deactivated
by formic acid when heated at 353 K, and this decrease is
probably a major reason for the slow hydrogenation of 1a
after the heat treatment in HCOOH at 353 K.
4
Conclusions
Fig. 6 Conversion of 1a for hydrogenation under a 1a/Pd/solvent
ratio of 0.42 mmol/10 lmol/10 mL at 296 K (a) in HCOOH (10 mL)
In the present study, the hydrogenation conditions of
-alkylidene-2,4-thiazolidiones (1a) were established using
5
(
3
3
open circle), (b) hydrogen pretreatment in HCOOH (10 mL) at
53 K (filled circle), (c) after a pretreatment in dioxane (2 mL) at
53 K and the subsequent HCOOH (8 mL) addition at 296 K (open
Pd/C for an alternative process for pioglitazone (2a) syn-
thesis. The optimized conditions (1 atm at 296 K) result in
a quantitative product 2a with a smaller amount of the
catalyst than that reported until now. 1a is poorly soluble in
many solvents, and warm formic acid is the only applicable
solvent if a biologically safe and properly volatile solvent
is desired. The slow and sluggish kinetics of reported
examples are mainly due to the deactivation of the Pd
catalyst that results from the heating of Pd in formic acid.
The formation of poisonous compounds by partial
decomposition of the reactant and/or product may be
responsible for the slow hydrogenation at high substrate/
catalyst ratios. As a result, we established a synthesis
process for 2a in a quantitative yield via the hydrogenation
of 1a.
square), (d) in dioxane (2 mL) without the pretreatment and then
HCOOH (8 mL) (filled square) at 296 K, and in a mixture of dioxane
(
2 mL) and HCOOH (8 mL) (open triangle) at 296 K
the type of Pd catalyst does not significantly affect the
reaction conversion.
The reaction conditions were further tuned in some
extent. The reaction of 1a at 5–10 MPa in an autoclave
increased the hydrogenation rate, but the degree is within
twofold. The recrystallization of 1a may lead to a higher
initial hydrogenation rate to some extent (30–60 %), but
the improvement in the reaction rate did not last for a
prolonged period, and the rates eventually became the
same during the reaction.
Lastly, time profile of the hydrogenation of 1a in formic
acid at room temperature (296 K) was given in Fig. 6
(
open circles). It is seen that the hydrogenation of 1a is
References
smooth under the optimized conditions. As suggested for
the hydrogenation of 1c, decomposition of 1a/2a is a
possible catalyst poisoning mechanism at elevated
1. Sohda T, Momose Y, Meguro K, Kawamatsu Y, Sugiyama Y,
Ikeda H (1990) Arzneim–Forsch. Drug Res 40:37–42
123

5
00
T. Sugimura et al.
2
3
. Nolan JJ, Ludvik B, Beerdsen P, Joyce M, Olefsky J (1994) J
Med 331:1188–1193
. Madivada LR, Anumala RR, Gillam G, Alla S, Charagondla K,
Kagga M, Bhattacharya A, Bandichlor R (2009) An improved
process for pioglitazone and its pharmaceutically acceptable salt.
Org Process Res Dev 13:1190–1194
6. Sakhardande R, Patent: WO2010/82212 A2, 2010
7. Patent to Cadila Healthcare Comp., WO2004/7490 (2004)
8. Patent to Takeda Chem. Ind., EP257781 (1991)
9. Patent to Takeda Chem. Ind., EP257781 (1991)
10. Patent to Takeda Chem. Ind., EP257781 (1991)
11. Bodil M, Mohanned S, Andrew A. WO11/15868 (2011)
12. Hu Y, Xie T, Fu K, Kang H, Wei P, Huang H (2009) Hetero-
cycles 78:757–761
4
5
. Momoshe Y, Meguro K, Ikeda H, Hatanaka C, Oi S, Shoda T
(
1991) Chem Pharm Bull 39:1440–1446
. Okamoto K, Akiyama R, Kobayashi S (2004) J Org Chem
9:2871–2873
6
1
23