Angewandte
Chemie
through dual iron catalysis. These iron catalysts are not only
responsible for the in situ regeneration of NAD(P)H models
with molecular hydrogen (transition-metal (TM) catalysis),
but also act as Lewis acids to promote the transfer hydro-
genation of carbonyl and imine groups (LA catalysis;
Scheme 2, right).
To probe the feasibility of the above proposal, we chose a-
ketoesters as substrates because of the significance of the a-
hydroxyester products and the reactivity of the substrates. For
these reasons, a-ketoesters are used frequently in biomimetic
reductions of NAD(P)H and its models.[10] We first tested
a series of representative NAD(P)+ models, such as diethyl
2,6-dimethylpyridine-3,5-dicarboxylate (3a), phenanthridine
(3b), acridine (3c), 2-phenylbenzo[d]thiazole (3d), and N-
methylphenanthridinium iodide (3e), with ethyl 2-oxo-2-
phenylacetate (1a) as the model substrate. Only phen-
anthridine (3b) showed minor catalytic activity in toluene
under H2 (50 bar) at 658C in the presence of [Fe3(CO)12]
(3.33 mol%; Table 1, entries 1–5). Although many different
iron–carbonyl complexes are known to catalyze the reduction
of carbonyl compounds,[11] only [Fe3(CO)12] promoted the
desired reaction of 1a in the presence of 3b (Table 1, entries 2
=
and 6–8), in agreement with its specific reduction of C N
groups.[11d]
Next, we conducted stoichiometric reactions of NAD(P)+
model 3b and its reduced form H2-3b with [Fe3(CO)12] as the
catalyst under H2 (50 bar) or Ar (50 bar), respectively. The
results in Equation (1) imply that NAD(P)H model H2–3b is
readily regenerated from its oxidized form 3b through iron-
catalyzed hydrogenation, although the reverse process occurs
with the same catalyst in the absence of H2. We therefore
considered that the low yield of the desired a-hydroxyester 2a
might be the result of the low Lewis acidity of the iron–
carbonyl complexes. When different higher-valent iron salts
were added to the reaction mixture as Lewis acids (Table 1,
entries 9–12), the product yield improved slightly, especially
with Fe(OTf)2 (Table 1, entry 12, 15% yield). Moreover,
ligand 4a also had a positive effect on the reactivity in toluene
(Table 1, entry 13, 28% yield). To further improve the
reaction efficiency, we tested different solvents and found
1,4-dioxane to be the most suitable (Table 1, entry 15, 30%
yield). However, in contrast to its effect in toluene, ligand 4a
had an adverse influence on the reaction in 1,4-dioxane
(Table 1, entry 14, 10% yield). Finally, the yield was funda-
mentally improved simply by increasing the catalyst loading
(Table 1, entry 16).[12] To verify our concept, we performed
several control experiments (Table 1, entries 17–21). We
could conclude that: 1) [Fe3(CO)12] is able to catalytically
hydrogenate the NAD(P)+ model 3b and also possesses some
Lewis acidity to promote the transfer hydrogenation of a-
ketoester 1a (Table 1, entry 17). On the other hand,
[Fe3(CO)12] failed to directly hydrogenate substrate 1a
(Table 1, entry 18). 2) NAD(P)+ model 3b is indispensable
for this biomimetic reduction process (Table 1, entry 21).
3) The higher-valent iron salt Fe(OTf)2 is unable to catalyze
the hydrogenation of NAD(P)+ model 3b and a-ketoester 1a
(Table 1, entries 19 and 20).
Having optimized the reaction conditions, we examined
the scope of the reaction and found that a variety of a-
ketoesters are suitable substrates for this biomimetic trans-
formation under dual iron catalysis (Table 2). Comparison of
the results with various aryl glyoxylic acid esters showed the
electronic effects of different substituents (Table 2, entries 1–
12). In general, better yields were observed when electron-
deficient substrates were used. In particular, in the case of the
highly electron rich substrate 1l, a prolonged reaction time
was needed for good conversion (81% yield of the isolated
product; Table 2, entry 12). Nevertheless, all aromatic sub-
strates tested were reduced to the corresponding a-hydroxy-
esters in good to excellent yields. Notably, this double iron-
Table 1: Iron-catalyzed biomimetic reduction of ethyl phenylglyoxylate in
the presence of different NAD(P)+ models.[a]
Entry
Fe carbonyl
Fe salt
NAD(P)+
model
Yield
[%][b]
1
2
3
4
5
6
7
8
9
10
11
12
[Fe3(CO)12]
[Fe3(CO)12]
[Fe3(CO)12]
[Fe3(CO)12]
[Fe3(CO)12]
[Fe2(CO)9]
[Fe(CO)5]
[NEt3H][HFe3(CO)11]
[Fe3(CO)12]
[Fe3(CO)12]
[Fe3(CO)12]
[Fe3(CO)12]
[Fe3(CO)12]
[Fe3(CO)12]
[Fe3(CO)12]
[Fe3(CO)12]
[Fe3(CO)12]
[Fe3(CO)12]
–
–
–
–
–
–
–
–
–
3a
3b
3c
3d
3e
3b
3b
3b
3b
3b
3b
3b
3b
3b
3b
3b
3b
–
–
5
–
–
–
–
–
–
4
7
4
15
28
10
FeCl3
FeCl2
Fe(OTf)3
Fe(OTf)2
Fe(OTf)2
Fe(OTf)2
Fe(OTf)2
Fe(OTf)2
–
13[c]
14[c,d]
15[d]
16[d,e]
17[d,e]
18[d,e]
19[d,e]
20[d,e]
21[d,e]
30
93(89)[f]
28
–
–
–
–
–
Fe(OTf)2
Fe(OTf)2
Fe(OTf)2
3b
–
–
–
[Fe3(CO)12]
[a] Reaction conditions: 1a (0.25 mmol), iron carbonyl (10 mol%, based
on iron), iron salt (3.6 mol%), NAD(P)+ model 3 (10 mol%), toluene
(0.5 mL). [b] The yield was determined by GC. [c] Diamine 4a
(3.6 mol%) was added as a ligand. [d] 1,4-Dioxane was used as the
solvent. [e] [Fe3(CO)12] (6.67 mol%), Fe(OTf)2 (7.2 mol%), and 3b
(20 mol%) were used. [f] The yield of the isolated product is given in
parentheses.
Angew. Chem. Int. Ed. 2013, 52, 8382 –8386
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
8383