Ytterbium acetate promoted asymmetric reductive amination: Significantly enhanced stereoselectivity

Article abstract of DOI:10.1021/jo7021235

(Chemical Equation Presented) Reductive amination of prochiral unhindered 2-alkanones 1 with (R)- or (S)-α-MBA in the presence of Yb(OAc) ₃ (50-110 mol %), Raney-Ni, and hydrogen (120 psi) results in increased diastereoselectivity for the amine products 2 (80-89% de) with good yield (80-87%). The increased de is based on comparison with the best previously reported de's when using (R)- or (S)-α-MBA, regardless of the strategy employed [stepwise (isolation of ketimines) or one-pot (reductive amination)], reducing agent examined, or achiral Lewis acid or Bronsted acid examined. An in situ cis- to trans-ketimine isomerization mechanism, promoted by Yb(OAc)₃, has been proposed to account for the observed increase in diastereoselectivity and suggests a new entry into the control of ketimine geometry.

Full text of DOI:10.1021/jo7021235

Ytterbium Acetate Promoted Asymmetric Reductive Amination:
Significantly Enhanced Stereoselectivity
Thomas C. Nugent,* Mohamed El-Shazly, and Vijay N. Wakchaure
Department of Chemistry, School of Engineering and Science, Jacobs UniVersity Bremen
(formerly International UniVersity Bremen), Campus Ring 1, 28759 Bremen, Germany
t.nugent@jacobs-uniVersity.de
ReceiVed September 29, 2007
Reductive amination of prochiral unhindered 2-alkanones 1 with (R)- or (S)-R-MBA in the presence of
Yb(OAc)3 (50-110 mol %), Raney-Ni, and hydrogen (120 psi) results in increased diastereoselectivity
for the amine products 2 (80-89% de) with good yield (80-87%). The increased de is based on
comparison with the best previously reported de’s when using (R)- or (S)-R-MBA, regardless of the
strategy employed [stepwise (isolation of ketimines) or one-pot (reductive amination)], reducing agent
examined, or achiral Lewis acid or Brønsted acid examined. An in situ cis- to trans-ketimine isomerization
mechanism, promoted by Yb(OAc)3, has been proposed to account for the observed increase in
diastereoselectivity and suggests a new entry into the control of ketimine geometry.
Introduction
exemplified by the fact that at least 40% of all optically active
pharmaceutical drugs contain an R-chiral amine moiety, yet 80%
of the required syntheses rely on classical resolution methods
to obtain the enantiopure amine.5 This unmet need, especially
regarding the synthesis of alkyl-alkyl′ substituted R-chiral
amines, continues to be a major challenge and is addressed in
this manuscript.
Of the commonly explored strategies for R-chiral amine
synthesis,3 reductive amination6 holds the advantage of being
stepwise efficient7 and has been recently labeled as a key green
chemistry research area.8 Sodium cyanoborohydride and sodium
triacetoxyborohydride hold a unique role as useful hydride
agents for reductive amination and have been extensively
R-Chiral amines are useful advanced intermediates for
alkaloid natural product synthesis and have been successfully
incorporated into billion dollar drugs, e.g., several ACE inhibi-
tors and Flomax.1 Despite their increasing presence in pharma-
ceutical targets, use as ligands, and recent and extensive
exploitation as organocatalysts,2 a lack of expedient, efficient,
and simple methods for their synthesis is evident.1,3,4 This is
* To whom correspondence should be addressed. Fax: (+) 49 421 200 3229.
(1) (a) Blaser, H.-U.; Burkhardt, S.; Kirner, H. J.; Mo¨ssner, T.; Studer,
M. Synthesis 2003, 11, 1679. (b) Blaser, H.-U.; Eissen, M.; Fauquex, P. F.;
Hungerbu¨hler, K.; Schmidt, E.; Sedelmeier, G.; Studer, M. Comparison of
Four Technical Syntheses of Ethyl (R)-2-Hydroxy-4-Phenylbutyrate. In
Asymmetric Catalysis on Industrial Scale: Challenges, Approaches, and
Solutions; Blaser, H.-U., Schmidt, E., Eds.; Wiley-VCH: Weinheim,
Germany, 2004; pp 91-103. (c) Nugent, T. C.; Ghosh, A. K.; Wakchaure,
V. N.; Mohanty, R. R. AdV. Synth. Catal. 2006, 348, 1289.
(4) A noteworthy and valuable exception is the synthesis of aryl-alkyl
substituted R-chiral primary amines via a one-pot enantioselective transfer
hydrogenation process for the reductive amination of aryl alkyl ketones,
see: Kadyrov, R.; Riermeier, T. H. Angew. Chem., Int. Ed. 2003, 42, 5472.
(5) (a) Blaser, H.-U.; Spindler, F.; Studer, A. App. Catal. Gen. 2001,
221, 119. (b) Blacker, J.; Stirling, M. Spec. Chem. Mag. 2007, May, 24.
(6) The term reductive amination is sometimes incorrectly associated
with the reduction of imines and derivatives thereof. Reductive amination
is the one-pot conversion of a ketone to an amine. The term “indirect
reductive amination” can be more tersely and accurately described as “imine
reduction”. For literature pertaining to the origins and definition of reductive
amination, see: (a) Emerson, W. S. Org. React. 1948, 4, 174. (b) Moore,
M. L. Org. React. 1949, 5, 301. (c) March’s AdVanced Organic Chemistry,
5th Ed.; Smith, M. B., March, J., Eds.; John Wiley & Sons, Inc.: New
York, 2001; pp 1187-1189.
(2) (a) Connon, S. J. Chem. Eur. J. 2006, 12, 5418. (b) Taylor, M. S.;
Jacobsen, E. N. Angew. Chem., Int. Ed. 2006, 45, 1520.
(3) For recent reviews pertaining to R-chiral amine synthesis, see: (a)
Kukula, P.; Prins, R. Top. Catal. 2003, 25, 29. (b) Besson, M.; Pinel, C.
Top. Catal. 2003, 25, 43. (c) Blaser, H.-U.; Malan, C.; Pugin, B.; Spindler,
F.; Steiner, H.; Studer, M. AdV. Synth. Catal. 2003, 345, 103. (d) Ellman,
J. A. Pure Appl. Chem. 2003, 75, 39. (e) Vilaivan, T.; Bhanthumnavin,
W.; Sritana-Anant, Y. Curr. Org. Chem. 2005, 9, 1315. (f) Tararov, V. I.;
Bo¨rner, A. Synlett 2005, 203. (g) T. C. Nugent. Chiral Amine Synthesis -
Strategies, Examples, and Limitations. In Process Chemistry in the
Pharmaceutical Industry, Second Edition, Vol 2: Challenges in an Ever-
Changing Climate, Braish, T. F., Gadamasetti, K., Eds.; CRC Press-Taylor
and Francis Group: New York, 2007, pp 137-156.
(7) For a lead reference on the intermolecular direct amination of alkanes,
see: Lebel, H.; Huard, K. Org. Lett. 2007, 9, 639.
10.1021/jo7021235 CCC: $40.75 © 2008 American Chemical Society
Published on Web 01/17/2008
J. Org. Chem. 2008, 73, 1297-1305
1297

Nugent et al.
SCHEME 1. Reductive Amination Products of 2-Alkanones with (S)-r-Methylbenzylamine
SCHEME 2. Two-Step Procedure for Producing (2S)-Aminooctane with Enhanced ee
examined and reviewed9 but cannot be considered as green
reagents. More atom efficient is the use of catalytic quantities
of a Brønsted acid in combination with hydrogen for reductive
amination, this method is routinely practiced by pharmaceutical
chemists for achiral carbon-nitrogen bond formation, but is
not often reported on. Asymmetric reductive amination studies
in these and other laboratories have demonstrated the effective-
ness of using stoichiometric quantities of a Lewis acid, e.g.,
tion of (R)- and (S)-R-MBA16 is partly based on their low cost,
as exemplified by their use on an industrial scale for the
resolution of some amino acids, e.g., (()-mandelic acid.17 We
have additionally pursued the use of these chiral ammonia
equivalents because the reductive amination products (2) can
be readily hydrogenolyzed,1c,15f allowing enantioenriched R-chiral
primary amines (3) to be isolated in high overall yield in two
steps from prochiral ketones (Scheme 2).
During our investigation of reductive amination reaction
conditions that could provide increased rates of reaction and/or
stereoselectivity, we examined many row 4 and 5 transition-
metal and lanthanide chlorides, acetates, and triflates. Of those,
only ytterbium acetate [Yb(OAc)3] allowed significantly in-
creased de with good reaction times. For example, 2-octanone
(1d), in MeOH, was fully consumed within 8 h providing the
secondary amine 2d in 82% de in the presence of Raney-Ni
(Scheme 2). When the solvent was changed to THF, the
stereoselectivity increased to 85% de, but 24 h were required
to completely consume the 2-octanone starting material. When
the binary solvent system of MeOH-THF (1:1) was examined,
an 86% de was observed with a fast reaction time of 10 h
(Scheme 2 and Table 1, entry 1). We additionally noted that
replacing THF, in the MeOH-THF solvent mixture, with
toluene, Et2O, or 1,3-dioxolane provided the same de, but with
i
M(OR)n where M ) Ti, Al, or B, and R ) Me or Pr, instead
of a Brønsted acid, for accessing alkyl-alkyl′- or aryl-alkyl-
substituted R-chiral primary amines (Scheme 1).10,11 Regarding
enantioselective variants, reductive amination employing mo-
lecular hydrogen,12 transfer hydrogenation,4 and organocatalytic
approaches13 have recently been demonstrated.
We recently determined1c that prochiral ketone substrates of
the general structure RLC(O)CH3 and RLC(O)RS allow high de’s
(87-98%),14 while substrates RMC(O)CH3 provide good to high
de’s (80-93%) when subjected to reductive amination with (R)-
or (S)-R-methylbenzylamine (R-MBA), Ti(OiPr)4, and Raney-
Ni or Pt-C. For straight-chain aliphatic ketones, RSC(O)CH3,14
the de’s were mediocre (66-74%).1c,11b Here, we report on a
new asymmetric reductive amination procedure using Yb(OAc)3
(50-110 mol %) that allows increased de (6-15% units) for
those ketone substrates, RSC(O)CH3 and RMC(O)CH3, that
previously only provided mediocre to good de. The enhanced
de is based on comparison with the best previously reported
de’s when using (R)- or (S)-R-MBA regardless of the strategy
employed, stepwise (via isolated ketimines) or one-pot (reductive
amination),6 or the reducing agent examined.15 Within the field
of reductive amination itself, no precedent for enhanced
stereoselectivity in the presence of an achiral Lewis acid or a
Brønsted acid exists to our knowledge.
(13) For organocatalytic reductive amination, Brønsted acids in combina-
tion with a Hantzsch ester, see: Storer, R. I.; Carrera, D. E.; Ni, Y.;
MacMillan, D. W. C. J. Am. Chem. Soc. 2006, 128, 84.
(14) The subscript serves as a generic reference to the steric bulk of the
substituent: RS ) small (any straight chain alkyl substituent, but not a
methyl group); RM ) medium, e.g. -CH2CH2Ph or -i-Bu; RL) e.g. -Ar,
-i-Pr, -c-hexyl.
(15) For advances in the diastereoselective reduction of (R)- or (S)-R-
MBA ketimines, see: (a) Nichols, D. E.; Barfknecht, C. F.; Rusterholz, D.
B. J. Med. Chem. 1973, 16, 480. (b) Clifton, J. E.; Collins, I.; Hallett, P.;
Hartley, D.; Lunts, L. H. C.; Wicks, P. D. J. Med. Chem. 1982, 25, 670.
(c) Eleveld, M. B.; Hogeveen, H.; Schudde, E. P. J. Org. Chem. 1986, 51,
3635-3642. (d) Bringmann, G.; Geisler, J.-P. Synthesis 1989, 608. (e) Marx,
E.; El Bouz, M.; Ce´le´rier, J. P.; Lhommet, G. Tetrahedron Lett. 1992, 33,
4307. (f) Moss, N.; Gauthier, J.; Ferland, J.-M. Synlett 1995, 142. (g)
Lauktien, G.; Volk, F.-J.; Frahm, A. W. Tetrahedron: Asymmetry 1997, 8,
3457. (h) Bisel, P.; Breitling, E.; Frahm, A. W. Eur. J. Org. Chem 1998,
729. (i) Gutman, A. L.; Etinger, M.; Nisnevich, G.; Polyak, F. Tetrahe-
dron: Asymmetry 1998, 9, 4369. (j) Cimarelli, C.; Palmieri, G. Tetrahe-
dron: Asymmetry 2000, 11, 2555. (k) Storace, L.; Anzalone, L.; Confalone,
P. N.; Davis, W. P.; Fortunak, J. M.; Giangiordano, M.; Haley, J. J., Jr.;
Kamholz, K.; Li, H.-Y.; Ma, P.; Nugent, W. A.; Parsons, R. L., Jr.; Sheeran,
P. J.; Silverman, C. E.; Waltermire, R. E.; Wood, C. C. Org. Process Res.
DeV. 2002, 6, 54.
Results and Discussion
Stoichiometric Ytterbium Acetate Promoted Reductive
Amination (Enhanced de). Our focus on and further exploita-
(8) Constable, D. J. C.; Dunn, P. J.; Hayler, J. D.; Humphrey, G. R.;
Leazer, J. L., Jr.; Linderman, R. J.; Lorenz, K.; Manley, J.; Pearlman, B.
A.; Wells, A.; Zaks, A.; Zhang, T. Y. Green Chem. 2007, 9, 411.
(9) (a) Abdel-Magid, A. F.; Mehrman, S. J. Org. Process Res. DeV. 2006,
10, 971. (b) Gribble, G. W. Org. Process Res. DeV. 2006, 10, 1062.
(10) For achiral methods, see: (a) Mattson, R. J.; Pham, K. M.; Leuck,
D. J.; Cowen, K. A. J. Org. Chem. 1990, 55, 2552. (b) Miriyala, B.;
Bhattacharyya, S.; Williamson, J. S. Tetrahedron 2004, 60, 1463. (c)
Bhattacharyya, S.; Kumpaty, H. J. Synthesis 2005, 2205.
(11) For recent titanium alkoxide based asymmetric methods, see ref.
1c and: (a) Borg, G.; Cogan, D. A.; Ellman, J. A. Tetrahedron Lett. 1999,
40, 6709. (b) Nugent, T. C.; Wakchaure, V. N.; Ghosh, A. K.; Mohanty,
R. R. Org. Lett. 2005, 7, 4967. (c) Menche, D.; Arikan, F.; Li, J.; Rudolph,
S. Org. Lett. 2007, 9, 267.
(16) For an extensive general review on the use of (R)- or (S)-R-MBA,
see: Juaristi, E.; Leo´n-Romo, J. L.; Reyes, A.; Escalante, J. Tetrahedron:
Asymmetry 1999, 10, 2441.
(17) Breuer, M.; Ditrich, K.; Habicher, T.; Hauer, B.; Keâeler, M.;
Stu¨rmer, R.; Zelinski, T. Angew. Chem., Int. Ed. 2004, 43, 788.
(18) When (S)-R-MBA is used as the limiting reagent [(ketone 1.2 equiv
and undried Yb(OAc)3 (1.1 equiv)], as shown in Table 1, (()-2-
aminooctanane was consistently observed at 6-7 area % (GC).
(12) Blaser, H.-U.; Buser, H.-P.; Jalett, H.-P.; Pugin, B.; Spindler, F.
Synlett 1999, 867.
1298 J. Org. Chem., Vol. 73, No. 4, 2008

Asymmetric ReductiVe Amination
TABLE 1. Identification of Yb(OAc)3 for Enhanced
Diastereoselectivity-Reductive Amination of 2-Octanone (Scheme 2,
Reaction Step 1)a
combined findings clearly establish Yb(OAc)3 as fulfilling a
greater role than that of a simple desiccant.
Useful Substrate Range (Stoichiometric Yb(OAc)3). After
further experimentation we established the optimal, in terms of
yield and de,20 protocol as stirring the ketone (limiting reagent),
amine 2d
(Scheme 2)
time (S)-R-MBAb 2-octanol yield
de
(%)
21
R-MBA (1.1 equiv), and dried Yb(OAc)3 in MeOH (1.0 M)
entry
additive
(h)
(%)
(%)
(%)
for 20-30 min followed by addition of Raney-Ni, as a slurry
in THF (final reaction molarity 0.5 M), and pressurization with
hydrogen (120 psi) at 22 °C. This general procedure was used
for all subsequently described reactions and suppressed the
formation of a previously noted byproduct, (()-2-aminooc-
tane,18,22,23 to 1-3 area % (GC) when examining 2-octanone.
Examination of the data in Table 2 shows that substrates of
the general class RSC(O)CH3, linear 2-alkanones,14 are excellent
substrates for the new ytterbium based reductive amination
method. In particular longer straight chain 2-alkanones, e.g.,
2-octanone (1d) and 2-hexanone (1e), show dramatic improve-
ments in de, 15% and 14%24 unit changes respectively, with
1
2
3
4
5
6
7
8
9
Yb(OAc)3
YbCl3
10
23
9
10
10
12
12
12
23
12
0.7
0.3
0.0
90.4
22.5
3.5
82.3
59.6
52.0
52.5
41.1
29.7
94.1
86.1c
66.2
77.4
[e]
Yb(OTf)3
-
96.4
11.4
24.4
29.3
28.8
34.6
43.3
1.0
d
Ti(OiPr)4
2.8
14.7
19
18.7
24.2
26.5
4.3
67.0
71.0
70.5
69.6
70.8
70.8
72
d
B(OiPr)3
MgSO4
4 Å M.S.
none
NaOAc
HOAc
10
a (S)-R-MBA (2.5 mmol, 1.0 equiv), 2-octanone (1.2 equiv), and an
additive (entries 1-5, 9, and 10: 1.1 equiv of additive; entry 6: 5.0 equiv;
entry 7: 4.0 wt equiv) were stirred in MeOH (1.0 M) for 30 min at rt, and
then THF (final molarity 0.5 M) and Raney-Ni were added and the reaction
was pressurized with H2 (120 psi). All data is based on GC area % analysis.
b Sum of (S)-R-MBA and ketimine remaining at the indicated time. c This
result is when Yb(OAc)3 was not vacuum-dried.18 d See ref 19. e Not
integrated, to emphasize the presence of the dominant alcohol byproduct.
(19) Note that in this initial study (Table 1), (S)-R-MBA was the limiting
reagent. The optimal procedures defined for titanium-, aluminum-, and
boron-based reductive aminations require the ketone as the limiting reagent
and reliably provides a de of 72% for 2-octanone in THF (without MeOH);
see ref 1c. Under those conditions, Ti(OiPr)4 (1.25 equiv) suppresses alcohol
formation below 3%. In general, B(OiPr)3 is less efficient at suppressing
alcohol byproduct formation; use of more equivalents of B(OiPr)3 (>1.25
equiv) is then helpful.
moderately longer reaction times, as did EtOH-THF. Before
moving forward with our study we hydrogenolyzed the reductive
amination product 2d to establish the enantiopurity of the
primary amine (Scheme 2). Later we found that vacuum drying
Yb(OAc)3 allowed the diastereoselectivity to be consistently
improved to 87%. This level of diastereoselectivity represents
a 15% unit increase in de over the best previously reported for
2-octanone with R-MBA.11b
Examination of other simple salts of ytterbium, Yb, demon-
strated the significance of having an acetate counterion. For
example, when 2-octanone (1d) was examined in MeOH, Yb-
(OTf)3 converted all of the ketone to the corresponding alcohol
within 9 h, whereas YbCl3 provided the product in 66% de, but
in only 23 area % (GC) after 24 h (Table 1, entries 2 and 3).
To probe the possibility that “free” acetate, in solution, had
constructively modified the heterogeneous metal surface, the
effect of adding NaOAc was examined (Table 1, entry 9). In
the event, gross quantities of the alcohol byproduct resulted and
a significantly lower product de resulted, making this simplified
scenario less likely. Next acetic acid, a commonly used Brønsted
acid for reductive amination, was examined, as expected it
suppressed alcohol byproduct formation, but provided the
secondary amine 2d in 72% de with the optimized solvent
(MeOH). When acetic acid (50, 100, or 200 mol %) was
intentionally added to the Yb(OAc)3 reaction described in Table
1 (entry 1), an amine product de of 72% was always noted (data
not included in Table 1).
(20) All quoted yield data is after chromatography. Alternatively, these
compounds can be isolated in near-qualitative purity (GC and 1H NMR)
without the need for flash chromatography using the acid/base workup
procedure found in the general Experimental Section. Note that during the
ammonium chloride washing, the small excess of (R)- or (S)-R-MBA
preferentially dissolves in the aqueous layer, while the desired and more
lipophilic amine product 2 preferentially dissolves in the organic layer.
Note: This has not been tested for product 2f. All amine products 2 should
be considered as volatile (<15 carbons) or semivolatile (<18 carbons) under
high vacuum drying or under prolonged rotary evaporation; HCl salt
formation (add ethereal HCl) is required for high vacuum drying.
(21) We noted after using several bottles of Yb(OAc)3, which is sold
and described as a semihydrated form (Sigma-Aldrich catalog no. 544973),
that it was sometimes free flowing while other bottles from the same lot
were not. As a consequence, 10 g quantities of Yb(OAc)3 were high vacuum
dried to constant weight at 80 oC (12 h), and Yb(OAc)3 treated in this way
was used for the optimal results shown here. In this manuscript, “dry Yb-
(OAc)3” means dried as just stated. The dried Yb(OAc)3 could be stored in
a dry screw cap glass bottle at room temperature, and this container could
be repeatedly opened to the atmosphere (at least 6 times without detrimental
effect) and the desired quantity of Yb(OAc)3 weighed out without the need
for a glovebox. In this way, constant and repeatable results were always
observed.
(22) Curiously, the noted byproduct 2-aminooctane was a racemate (GC
analysis of the trifluoroacetamide derivative), excluding the possibility that
it originated from a sequential reductive amination of (S)-R-MBA with
2-octanone followed by in situ hydrogenolysis of the product 2d. It also
did not originate from the in situ hydrogenolysis of (S)-R-MBA (forming
ammonia and ethylbenzene), followed by reductive amination of ammonia
with 2-octanone. This is known because ethylbenzene (relative to an
authentic reference standard) was never noted (GC) even when taking care
to work up the reaction at 0 oC with aqueous NaHCO3/EtOAc to avoid
o
evaporation of ethylbenezene (bp ) 136 C). While [1,3]-proton shifts of
imines are known, they are to our knowledge only accomplished under the
presence of a strong base; see ref 23. If a [1,3]-proton shift of the initially
formed imine occurred, followed by in situ hydrolysis, 2-aminooctane and
acetophenone would result. When 2-octanone (1.0 equiv), (S)-R-MBA (1.1
equiv), and Yb(OAc)3 (1.1 equiv) were added to THF-MeOH (standard
reaction conditions) and the mixture stirred in the absence of Raney-Ni
and H2, a very small quantity of a new compound (<2 area %) with very
similar retention time to acetophenone (as compared to an authentic sample)
was noted. No further studies regarding the origin of 2-aminooctane
formation were pursued and as stated earlier high vacuum drying of Yb-
(OAc)3 suppressed 2-aminooctane formation below 3 area % (GC).
(23) (a) Cainelli, G.; Giacomini, D.; Trere`, A.; Boyl, P. P. J. Org. Chem.
1996, 61, 5134. (b) Willems, J. G. H.; de Vries, J. G.; Nolte, R. J. M.;
Zwanenburg, B. Tetrahedron Lett. 1995, 36, 3917. (c) Soloshonok, V. A.;
Kirilenko, A. G.; Galushko, S. V.; Kukhar, V. P. Tetrahedron Lett. 1994,
35, 5063.
Past speculation regarding the role of Lewis acids in Lewis
acid promoted reductive amination has generally relegated them
to the level of efficient desiccants for in situ ketimine formation.
To study this effect more closely, we examined some traditional
desiccants. When Yb(OAc)3 (1.1 equiv) is replaced by MgSO4
(5.0 equiv) or 4 Å molecular sieves (4.0 wt equiv), both vacuum
oven dried at 150 °C for 15 h before use, not only are low
diastereoselectivities observed (Table 1, compare entries 1, 6,
7), but large quantities of 2-octanone are reduced to the alcohol
byproduct. In relation to these results, alcohol byproduct
formation could be suppressed when employing Ti(OiPr)4 (1.25
equiv), but the de remained low (Table 1, entry 4).19 These
J. Org. Chem, Vol. 73, No. 4, 2008 1299

Nugent et al.
TABLE 2. Ketone Substrate Breadth for Enhanced Diastereoselectivitya
a Ketone (2.5 mmol, 1.0 equiv), dried Yb(OAc)3 (1.1 equiv),21 (S)-R-methylbenzylamine (1.1 equiv), Raney Ni, H2 (120 psi), MeOH-THF (1:1) 0.50 M,
22 °C, 12 h. b Isolated yield of both diastereomers after chromatography. c Determined by GC analysis of crude product 2. d Percent unit change in de
compared to the best previously reported de.1c,11b,15b e Pinacolone is a Pt-C substrate and requires 22 h at T ) 50 °C.
good isolated yield (Table 2, entries 4 and 5) vs the best
previously reported methods.1c For this study, we additionally
synthesized and isolated the N-(S)-R-MBA ketimine of 2-oc-
tanone, its reduction with Raney-Ni/H2 in THF-MeOH (1:1)
provided 2d in 64% de. The short-chain 2-butanone (1f) showed
a smaller but consistent and significant 6% unit increase in de
vs the best previously reported result (Ti(OiPr)4/CH2Cl2/Raney-
Ni: 74% de).1c
Comparison of substrates that can be loosely qualified as
having a medium-sized R substituent residing on the 2-alkanone
carbonyl, RMC(O)CH3,14 help define the boundary substrates
for enhanced stereoselectivity. For example when the γ-branched
benzylacetone (1c) was examined, a 9% unit increase in de was
observed vs the previous best reported methods (Table 2, entry
3).1c,15b For the â-branched isobutyl methyl ketone (1a), no
improvement in stereoselectivity (1% increase) was noted (Table
2, entry 1). When R-branched 2-alkanones, e.g., isopropyl
methyl ketone or cyclohexyl methyl ketone, are subjected to
reductive amination with Ti(OiPr)4 the product is formed with
high de (>98% de);1c no improvement was noted when using
Yb(OAc)3.25,26
acid have not been previously reported for promoting efficient
reductive amination, we further investigated this point. When
2-octanone was examined, the use of 80 mol % of Yb(OAc)3
allowed the same 87% de to be achieved as when 110 mol %
was used, and interestingly, the use of 50 mol % of Yb(OAc)3
was equally effective (86% de). Lowering the mol % of Yb-
(OAc)3 below 40% led to a rapid deterioration in the diaste-
reoselectivity, such that when 10 mol % of Yb(OAc)3 was used
no enhancement in stereoselectivity could be expected. At 20
mol % loading, a 79% de was observed, and repetition of this
experiment in the presence of 4 Å molecular sieves (4 wt equiv)
provided no additional benefit.
When less than 10 mol % of Yb(OAc)3 was employed, greater
than 3 area % (GC) of the alcohol byproduct was noted; thus,
we chose 10 mol % to determine the isolated yields for a broad
range of substrates (Table 3).27 Before doing so, we investigated
a large variety of transition-metal, lanthanide, and metalloid
halides, acetates, alkoxides, and sulfonates for their ability to
allow fast and high yielding reductive amination reactions to
occur. By doing so, we were able to further identify Ce(OAc)3
(26) Examination of aryl alkyl ketones, e.g., acyclic acetophenone or
cyclic benzosuberone, or a non-2-alkanone, e.g., isopropyl n-propyl ketone,
proved problematic. Acetophenone required higher temperature, 60 °C with
120 psi of H2, and produced the product in 92% de but with large amounts
of the corresponding alcohol noted (∼20 area %, GC). For benzosuberone
and isopropyl n-propyl ketone, repeated attempts to obtain the intended
product by heating and/or increasing the hydrogen pressure failed. For these
substrates, the desired products can be produced in good yield and de using
Ti(O-i-Pr)4 instead of Yb(OAc)3; see ref 1c.
Catalytic Lewis Acid-Promoted Reductive Amination.
Encouraged by our results with stoichiometric quantities of Yb-
(OAc)3 and aware of the fact that catalytic quantities of a Lewis
(24) 2-Hexanone is quoted as having a 14% unit increase in de (71% vs
85% de); this is based on comparison with entry 9 of Table 3 of this
manuscript. If compared to our earlier findings, ref 1c, the increase would
be 19% (66% vs 85 % de). In ref 1c, we speculate that the low de of
2-hexanone is not explainable based on the 2-octanone and 2-butanone
results shown therein; thus, we believe 14% more accurately reflects the
true difference.
(25) Ketone substrates with an R-quaternary carbon do not undergo
reductive amination with Raney-Ni, even under forcing conditions; instead,
Pt-C is the catalyst of choice for this class of prochiral ketones. Examination
of pinacolone (1b) with Yb(OAc)3/Pt-C/H2 provided a consistent 92% de
(Table 2). For this Pt substrate, there is no change in de verses the previously
best reported method; see ref 15f.
(27) Regarding aryl alkyl ketones, acetophenone (Table 3, entry 3) was
sluggish to react even at 50 °C and 432 psi (30 bar) of hydrogen, with
isolated yields varying between 55-65% and concomitant alcohol by-
product formation always noted. Examination of 1-phenylbutanone at
50 oC and 432 psi (30 bar) of hydrogen only allowed ∼20 area % (GC) of
the expected product to form after 24 h. Benzosuberone (cyclic aryl alkyl
ketone) and isopropyl n-propyl ketone, under similar forcing conditions
o
(50 C, 580 psi (40 bar) H2, >24 h), showed these sterically challenging
substrates could not be converted to the desired product with catalytic
quantities of Yb(OAc)3.
1300 J. Org. Chem., Vol. 73, No. 4, 2008

Asymmetric ReductiVe Amination
TABLE 3. 10 mol % Yb(OAc)3 Catalyzed Reductive Amination: Substrate Breadtha
a Ketone (2.5 mmol), dried Yb(OAc)3 (10 mol %),21 (S)-R-methylbenzylamine (1.1 equiv), Raney Ni, 0.50 M (MeOH-THF, 1:1), 12 h. Entries 1-4: T
) 50 °C, H2 pressure ) 290 psi (20 bar); entry 5: T ) 50 °C, H2 pressure ) 120 psi (8.3 bar), entries 6-9: T) 22 °C, H2 pressure ) 120 psi (8.3 bar).
b Isolated yield of both diastereomers after chromatography; also see ref 20. c Determined by GC analysis of the crude product. d Pt-C was used instead of
Raney Ni, T ) 50 °C, H2 pressure ) 120 psi. e Non-Yb(OAc)3 based methods provide 74% de.
(15 mol %) and Y(OAc)3 (15 mol %)28 as consistently providing
results similar to Yb(OAc)3 (10 mol %), regarding reaction time,
yield, and diastereoselectivity for the reductive amination of
2-octanone (1d) with (S)-R-MBA.29 Use of stoichiometric Ce-
(OAc)3 or Y(OAc)3 did not provide enhanced de or any other
added benefit over those reactions examined at the 15 mol %
level.
Finally, in contrast to earlier industrial findings regarding the
beneficial use of 5 mol % of thionyl chloride for promoting
ketimine formation,30 replacement of the Lewis acid by thionyl
chloride led to the amine product 2d in 50% de. The use of
phosphorus oxychloride has not been previously reported on
and we found its use to be beneficial at 10 mol % (72% de),
but upon completion of the reductive amination, the alcohol
byproduct was observed in ∼4 area % (GC).31
Stoichiometric and Catalytic Brønsted Acid-Promoted
Reductive Amination. Brønsted acids are well-known for
promoting reductive amination, but an overview of useful
Brønsted acids is, to our knowledge, not available in the primary
literature. The Table 4 data shows that 20 mol % of acetic acid,
trichloroacetic acid, or formic acid is optimal for the reductive
amination of 2-octanone with R-MBA. The use of 5 mol % of
AcOH had the detrimental effect of allowing significant alcohol
byproduct formation (>5 area %, GC). The use of stoichiometric
quantities of acetic acid, trichloroacetic acid, or formic acid had
no detrimental effect on the reaction rate or yield (GC) and
consistently provided secondary amine 2d in 72% de (Table
4), in contrast to Yb(OAc)3 (Table 2, 87% de).
(28) Y(OAc)3 and Ce(OAc)3 were dried as indicated in ref 21. Sigma-
Aldrich catalog nos.: Y(OAc)3 326046, Ce(OAc)3 529559.
(29) A second and larger group of Lewis acids, In(OAc)3, Sc(OAc)3,
CuOAc, Er(OAc)3, Gd(OAc)3, Dy(OAc)3, AgOAc, Zn(OAc)2, and Cd-
(OAc)2, emerged as being useful (15 mol %), but always provided larger
quantities of the alcohol byproduct, typically 5-15 area % by GC analysis.
Use of the corresponding chlorides or triflates of the same elements (15
mol %) proved detrimental, with >25 area % of the alcohol byproduct
forming in most instances. Exceptions to this general observation were noted
for Bi(OTf)3, AgCl, ScCl3, and scandium hexafluoroacetylacetone, which
provided 5-15 area % of the alcohol byproduct and/or observably longer
reaction times than Yb(OAc)3 (10 mol %), Y(OAc)3 (15 mol %), or Ce-
(OAc)3 (15 mol %). Initially, 15 mol % of the metalloids Bi(OAc)3 and
Sb(OAc)3 were included as useful as Y(OAc)3 or Ce(OAc)3, but they smelled
of AcOH and were thus dried to a constant weight. Upon reexamination,
dried Bi(OAc)3 and Sb(OAc)3 proved uninteresting, implying that coexisting
AcOH had catalyzed the earlier reactions.
(30) Farina, V.; Grozinger, K.; Mu¨ller-Bo¨tticher, H.; Roth, G. P.
Ontazolast: The Evolution of a Process. In Process Chemistry in the
Pharmaceutical Industry; Gadamasetti, K. G., Ed.; Marcel Dekker, Inc.:
New York, 1999; pp 107-124.
(31) An isolated yield was not determined for the reductive amination
reaction catalyzed by phosphorous oxychloride.
J. Org. Chem, Vol. 73, No. 4, 2008 1301

Nugent et al.
FIGURE 1. Nitrogen-benzylic carbon bond rotamers responsible for hydrogen addition to cis- and trans-N-R-MBA ketimines.
TABLE 4. Brønsted Acid Based Reductive Amination of
1d (83% isolated yield, 72% de, T ) 22 °C, 120 psi (8.0 bar)
H2), isobutyl methyl ketone 1a (80% isolated yield, 92% de, T
) 50 °C, 120 psi (8.0 bar) H2), cyclohexyl methyl ketone 1g
(82% isolated yield, 98% de, T ) 50 °C, 290 psi (20 bar) H2),
and acetophenone 1i (55% isolated yield, 93% de, T ) 50 °C,
435 psi (30 bar) H2) provided very similar reaction rates, isolated
yield, and the same de data as the optimal catalytic Lewis acids
[Yb(OAc)3 (10 mol %), Y(OAc)3 (15 mol %), and Ce(OAc)3
(15 mol %)] but never showed the enhanced diastereoselectivity
possible when using 50-110 mol % Yb(OAc)3 (Tables 2 and
3).34
2-Octanone with (S)-r-MBAa
2-octanone
Brønsted acid
(20 mol %)
remaining
(area %)
alcohol formed
(area %)b
entry
deb
1
2
3
4
5
6
7
acetic acid
1
1
2
72
71
72
71
73
26
70
trifluoroacetic acid
trichloroacetic acid
formic acid
oxalic acid
thiophenol
22.3
2.1
2.7
24
3.5
29
1.7
51.8
phenol
a 2-Octanone (2.5 mmol), Brønsted acid (20 mol %), (S)-R-methylben-
zylamine (2.75 mmol), Raney Ni, 0.50 M (MeOH), 12 h, T ) 22 °C, H2
pressure ) 120 psi (see the Supporting Information). b Determined by GC
analysis at 12 h.
Basic Stereochemical Considerations. Figure 1 provides a
working model of the low energy conformers of the cis- and
trans-ketimines influencing facial selectivity during the addition
of hydrogen to N-R-MBA ketimines. The models are consistent
with the major amine diastereomer formed (either S,S or R,R
depending on the starting enantiomer of R-MBA) and are in
accordance with the well-established concept of allylic 1,3-
strain35 and earlier proposed models.15c,i Contrary to this model,
it has been previously suggested that the reductive amination
of an R-keto ester with R-MBA may involve a rotamer (about
the nitrogen-benzylic carbon bond) with the phenyl ring of
R-MBA coplanar to the ketimine double bond.36 The idea is
appealing because of the well described affinity of π bonds for
heterogeneous hydrogenation catalyst surfaces,37 but examina-
tion of the two possible trans-ketimine rotamers (not shown),
with a coplanar phenyl group relative to the ketimine double
bond, shows one suffering from high allylic 1,3-strain35
(prominent phenyl group steric crowding with the methyl group
attached to the carbonyl carbon of the ketimine) while the other
rotamer is less encumbered (1,2-strain from the electron pair
A different product profile was noted for strong mineral acids
and the very acidic trifluoroacetic acid. For example, when using
stoichiometric or catalytic (5 or 10 mol %) quantities of 12 N
HCl or 18 M H2SO4, p-TsOH, or trifluoroacetic acid in MeOH,
high alcohol byproduct formation was always noted (15-30
area %, GC).32 Regardless of the low yield of 2d, when using
these strong Brønsted acids, the product de ranged from 70 to
72% for 2-octanone 1d.
Regarding the optimal solvent for Brønsted acid based
reductive amination, MeOH and EtOH allowed fast reactions,
e.g., with AcOH (20 mol %) only 8 h was required. Using the
combination of THF-MeOH (which was optimal for the earlier
discussed Lewis acids) slowed down the reaction (12 h). Use
of THF as the sole solvent for AcOH (20 mol %) further
suppressed the reaction rate such that 30-45% of the ketone
consistently remained after 24 h of reaction (22 °C, 120 psi
H2).33 By contrast, the stoichiometric and catalytic Lewis acid
promoted reactions in THF were complete at 24 h under
otherwise identical reaction conditions. This noted difference
in protic vs aprotic reaction solvent profile may allow the future
use of substrates containing acid labile functional groups or
having restricted solubility.
(34) When using 20 mol % AcOH with the sterically hindered ketone
substrates benzosuberone, 1-phenylbutanone, or isopropyl n-propyl ketone,
the effect of temperature (22-50 oC) and hydrogen pressure [120-580 psi
(8-40 bar)] were independently and then concurrently examined; only
1-phenylbutanone provided the desired product and then only in very low
yield (20-25 area %, GC) after >24 h of reaction. For these types of
substrates, the only effective reductive amination conditions are those
employing stoichiometric quantities of Ti(OiPr)4; see ref 1c.
(35) (a) Hoffmann, R. W. Chem. ReV. 1989, 89, 1841. (b) Lee, K. W.;
Hwang, S. Y.; Kim, C. R.; Nam, D. H.; Chang, J. H.; Choi, S. C.; Choi, B.
S.; Choi, H.-W.; Lee, K. K.; So, B.; Cho, S. W.; Shin, H. Org. Process
Res. DeV. 2003, 7, 839.
(36) Siedlaczek, G.; Schwickardi, M.; Kolb, U.; Bogdanovic, B.;
Blackmond, D. G. Catal. Lett. 1998, 55, 67.
(37) (a) Kraynov, A.; Suchopar, A.; D’Souza, L.; Richards, R. Phys.
Chem. Chem. Phys. 2006, 8, 1321. (b) Bu¨rgi, T.; Baiker, A. Acc. Chem.
Res. 2004, 37, 909. (c) Studer, M.; Blaser, H.-U.; Exner, C. AdV. Synth.
Catal. 2003, 345, 45.
Finally, acetic acid (20 mol %) was examined for its ability
to affect efficient reductive amination in dry MeOH, and
representative ketones from Table 3 were chosen. 2-Octanone
(32) It should be noted that the strong Brønsted acids mentioned here
may act as good acid catalysts for the reductive amination of ketone/amine
combinations that are less sterically congested as compared to the ketone/
R-MBA combinations shown here.
(33) It is important to note that little or no alcohol byproduct formation
was noted during these incomplete reactions in THF.
1302 J. Org. Chem., Vol. 73, No. 4, 2008

Asymmetric ReductiVe Amination
TABLE 5. Early Reaction Progress: Stereochemical Outcome for the Reductive Amination of 2-Octanone (1d)a
amine product 2d
entry
1
additive
time (h)
(S)-R-MBAb (%)
2-octanone (%)
2-octanol (%)
yield (%)
de (%)
Yb(OAc)3
1c
3d
1
3
1
18.1
12.5
43.9
8.5
47.1
29.2
6.5
2.5
7.9
4.0
10.5
1.1
0.5
1.0
0.63
2.1
16.6
27.6
73.9
81.5
47.6
85.4
25.8
42.2
87.1
87.2
67.0
67.0
72.0
72.1
2
3
Ti(OiPr)4
none
3
a 2-Octanone (2.5 mmol, 1.0 equiv), (S)-R-MBA (1.1 equiv), and Yb(OAc)3 or Ti(OiPr)4 (1.1 equiv) were stirred in MeOH (1.0 M) for 90 min at rt, and
then THF (final molarity 0.5 M) and Ra-Ni (100 wt %) were added and the reaction was pressurized with H2 (8.3 bar/120 psi). All data is based on
GC area % analysis. b Sum of (S)-R-MBA and ketimine remaining. c 1.0 area % of (()-2-aminooctane was noted. d 2.5 area % of (()-2-aminooctane was
noted.
TABLE 6. In Situ Ketimine Formation Study: 2-Octanone and
with Yb(OAc)3 made it more susceptible to hydrolysis upon
workup, even under the mild conditions of aq NaHCO3/EtOAc.
To experimentally support this, we synthesized the N-R-MBA
ketimine of 2-octanone (Dean-Stark trap, see the Experimental
Section), added MeOH (anhydrous)39 and an equivimolar
quantity of Yb(OAc)3 (110 mol %). After 30 min, the mixture
was worked up (aq NaHCO3/EtOAc), and again no ketimine
(6 area %, GC) was noted, instead the ketone and amine were
found. This result combined with the fact that reductive
amination is possible in the presence of Yb(OAc)3 strongly
suggests that ketimine formation can and does occur in the
presence of Yb(OAc)3. Unfortunately, the labile nature of the
ketimine in the presence of Yb(OAc)3 rules out the possibility
of determining the origin of the enhanced de through comparison
of the cis/trans ketimine ratios when no additive or Ti(OiPr)4 is
present vs when Yb(OAc)3 is present.40
(S)-r-MBAa
ketimine area % (GC analysis)
time (min) no additive Ti(OiPr)4 (1.25 equiv) Yb(OAc)3 (1.1 equiv)
30
90
16
33
38
40
<3
<3
a 2-octanone (1.0 equiv), (S)-R-MBA (1.1 equiv), and the indicated
additive were added to anhydrous MeOH (1.0 M).39 All aliquots were
worked-up with saturated aq NaHCO3/EtOAc at the indicated times.
of the ketimine nitrogen and one of the ortho hydrogens of the
phenyl ring) but leads to the formation of the wrong amine
product diastereomer. It is therefore unlikely that the phenyl
ring of an N-R-MBA ketimine is adsorbed to the surface of a
heterogeneous hydrogenation catalyst at the time of hydrogen
addition.
Taking the above facts into consideration, it seemed probably
to us that the in situ generated trans- and cis-ketimine mixture
was being enriched to one having greater trans-ketimine
representation before hydrogenation occurred. This would be
possible if Yb(OAc)3 was capable of isomerizing some of the
cis-ketimine to the trans-ketimine. To examine this possibility,
we dissolved the N-(S)-MBA ketimine of 2-octanone, isolated
after Dean-Stark trap synthesis, in MeOH (anhydrous),39 and
stirred it for 30 min41 in the presence of 110 mol % of Yb-
(OAc)3 at 22 °C before adding Raney-Ni (THF slurry) and
pressurizing with hydrogen (120 psi). At 12 h, no ketimine
remained, the reaction profile was clean, and an amine product
de of 86% was noted for amine 2d. This high de for ketimine
reduction correlates very closely with that observed during the
reductive amination of 2-octanone with (S)-R-MBA in the
presence of Yb(OAc)3 (87% de) and strengthens the possibility
that an in situ cis- to trans-ketimine isomerization is occurring.42
Treatment of the same ketimine under identical reaction
Mechanistic Considerations: Enhanced Stereoselectivity
with Yb(OAc)3. Regarding reductive amination in general and
the reduction of isolated N-R-MBA ketimines, no precedent
exists for obtaining enhanced product diastereoselectivity
through the use of an achiral Lewis acid or Brønsted acid.
Experiments were thus designed to probe the origin of the
enhanced diastereoselectivity.
Reductive amination of 2-octanone with Ti(OiPr)4 or Yb-
(OAc)3 requires 12 h for complete consumption of the ketone
to occur, and examination of the reaction at 1 and 3 h showed
the diastereoselectivity of the product to be fully consistent with
that found at the end of the reaction (Table 5).38 These results
suggest that one mechanism is operating throughout the reaction
period but did not provide further insight into the origin of the
enhanced diastereoselectivity.
Because the cis/trans ratio of diastereomeric N-R-MBA
ketimines is the major parameter influencing the de of the
product, we next examined the progress of in situ ketimine
formation. GC analysis of a mixture of 2-octanone and (S)-R-
MBA after 30 min with no additive or with Ti(OiPr)4 (1.25
equiv) provided a similar area % of the ketimine, but interest-
ingly Yb(OAc)3 never showed appreciable amounts (<3 area
%) of the ketimine (Table 6). Attempts to force greater quantities
of the ketimine to form in the presence of Yb(OAc)3, e.g., by
extending the reaction time to 12 h, failed.
(39) Note the optimal reaction conditions call for stirring the ketone,
amine, and Yb(OAc)3 for 20-30 min in MeOH (1.0 M), followed by the
addition of Raney-Ni (as a slurry in THF, final reaction molarity 0.5) and
immediate pressurization with hydrogen.
(40) In our hands, 1H NMR experiments of the ketone, R-MBA, and
Yb(OAc)3 in CD3OH proved unhelpful due to the broadening of resonance
patterns. The broadening is likely due to the influence of an NMR active
isotope of Yb which is paramagnetic. It should be noted that these mixtures
are turbid and attempts to filter them and then record the NMR spectrum
also failed to provide useful spectroscopic data.
The lack of ketimine accumulation in the presence of Yb-
(OAc)3 raised the possibility that the N-R-MBA ketimine was
forming in situ, but subsequent Lewis acid-base pair formation
(41) Extending the stirring time from 30 min to 12 h, before the onset
of hydrogenation, resulted in the same de.
(42) We also examined the effect of replacing MeOH with THF, but the
de dropped to 77%. This can be readily explained by the low observed
solubility of Yb(OAc)3 in the THF mixture containing the N-(S)-MBA
ketimine of 2-octanone, allowing the back round reaction to occur and
showing the importance of having MeOH as a cosolvent.
(38) The Table 5 study employed the optimal reaction conditions (ketone
limiting reagent), when the same study was performed, albeit using (S)-R-
MBA as the limiting reagent with excess ketone (1.2 equiv), trivial changes
in the Table 5 area % numbers were noted.
J. Org. Chem, Vol. 73, No. 4, 2008 1303

Nugent et al.
SCHEME 3. Proposed Mechanism for in Situ Ketimine
Isomerization during Reductive Amination
FIGURE 2. Speculative relative free energy differences for branched
and nonbranched 2-alkanones.
substituent of the former carbonyl carbon (Figure 2, anti-6, RR
or Râ ) alkyl). Enhanced diastereoselectivity is noted for
γ-branched 2-alkanones (Figure 2, RR and Râ ) H) and
nonbranched 2-alkanones (Figure 2, RR, Râ, Rγ ) H) suggesting
that for these ketone substrates Newman projection gauche-6
is higher in energy than anti-6. This relative free energy
difference would be expected to drive cis- to trans-ketimine
isomerization in the presence of Yb(OAc)3.43-45
Key Findings for Reductive Amination with R-MBA. This
study, and our earlier ones, complete a body of research
regarding the reductive amination of prochiral ketones with (R)-
or (S)-R-MBA (1.1 equiv) under the influence of an optimal
achiral Lewis acid or Brønsted acid. Comparison of the de’s of
the reductive amination products 2 when using Ti(OiPr)4
(stoichiometric) or catalytic Lewis acids [Yb(OAc)3 (10 mol
%), Y(OAc)3 (15 mol %), or Ce(OAc)3 (15 mol %)], or Brønsted
acids, show them to be equal. Furthermore, if (R)- or (S)-R-
MBA ketimines are synthesized, isolated, and reduced, the same
de is noted as when the above-noted Lewis or Brønsted acid
catalysts are used for reductive amination of the same ketone
conditions, albeit without Yb(OAc)3, resulted in an amine
product de of 64%.
Outlined in Scheme 3 is a proposed mechanism for ketimine
isomerization occurring during reductive amination. It relies on
in situ ketimine formation (not shown) followed by Lewis acid-
base pair formation between the cis- and trans-ketimines and
Yb(OAc)3 (Scheme 3, cis-5). Intramolecular attack at the
electrophilic carbonyl carbon of the iminium ion by the acetate
ligand of ytterbium would proceed via a six-membered transition
state forming an oxygen-acetylated hemiaminal (gauche-6).
Pyrimidal inversion at nitrogen is a low energy process which
readily occurs at room temperature and for gauche-6 would be
further driven by relief of the steric crowding arising from the
gauche relationship between the “R-methylbenzyl” substituent
on the nitrogen atom and the “R” substituent of the former
carbonyl carbon of the ketimine. Nitrogen inversion of gauche-6
would provide anti-6 (Scheme 3), allowing an anti-relationship
to exist between the “R-methylbenzyl” substituent of nitrogen
and the “R” substituent of the former carbonyl carbon; this
conformation also allows an antiperiplanar arrangement to exist
between the nitrogen lone pair and the acetate leaving group,
allowing facile elimination of acetate and trans-ketimine forma-
tion (trans-5).
An explanation for why straight-chain 2-alkanones (e.g.,
2-octanone 1d) and γ-branched 2-alkanones (e.g., benzylacetone
1c) are good substrates for the method, while R- and â-branched
2-alkanones are not, can be rationalized by more closely
examining the “R” substituent of the two Newman projections
(gauche-6 and anti-6) shown in Scheme 3. Experimentally, it
is known that no improvement in diastereoselectivity is noted
for R- or â-branched 2-alkanones when using Yb(OAc)3. This
implies that there is not a significant energetic difference
between the gauche steric crowding arising from the “R-
methylbenzyl” substituent on the nitrogen atom and the “R”
substituent of the former carbonyl carbon (Figure 2, gauche-6,
RR or Râ ) alkyl) vs the steric crowding experienced when the
“ytterbium” substituent of nitrogen is gauche to the “R”
(43) The origin of the difference in relative free energy for the Newman
projections can be further rationalized by noting that the “R-methylbenzyl”
substituent of nitrogen produces a very large steric volume in the immediate
vicinity of nitrogen, while the ytterbium-nitrogen bond will be expected
to be longer than the benzylic carbon-nitrogen bond of the “R-methyl-
benzyl” substituent and thereby reduce the immediate steric volume next
to nitrogen. The conclusion, based on the observed substrate de, is that
regardless of the level of branching within the “R” substituent (Scheme 3),
there is always a high degree of steric crowding when the “R-methylbenzyl”
substituent is gauche to it (Figures 2, gauche-6). This would not always be
the case when comparing the steric crowding of the “R” substituent with a
gauche positioned ytterbium. Thus “R” substituents having only γ-branching
might be expected to have medium steric crowding with a gauche ytterbium
atom (Figure 2, compare gauche-6 and anti-6 when RR and Râ ) H), while
nonbranched “R” substituents would have low (relative) steric crowding in
relation to a gauche positioned ytterbium atom (Figure 2, compare gauche-6
and anti-6 when RR, Râ, Rγ ) H). These considerations would support the
possibility of cis-ketimine to trans-ketimine isomerization for γ-branched
and straight-chain 2-alkanones and thereby their enhanced product de, while
at the same time explain why R- and â-branched 2-alkanones would not
provide enhanced product de.
1304 J. Org. Chem., Vol. 73, No. 4, 2008

Asymmetric ReductiVe Amination
the ketone and ketimine intermediate remained, GC), the reaction
mixture was diluted with MeOH (10 mL) and filtered through a
bed of Celite and the Celite subsequently washed with MeOH (3
× 10 mL). The combined filtrate was then evaporated to dryness
(rotary evaporate at e25 °C for short periods due to product
volatility), and CHCl3 (20 mL) and aqueous NaOH (15 mL, 1.0
M) were added. After being stirred for 60-90 min, the biphasic
solution was transferred to a separatory funnel, the CHCl3 layer
was removed, and the aqueous layer was further extracted with
CHCl3 (3 × 15 mL). The combined CHCl3 extracts were filtered
through a short bed of Celite (removes turbidity), and the Celite
was subsequently washed with CHCl3 (2 × 15 mL). The filtrate
was then washed with saturated NH4Cl (2 × 20 mL) [removes
residual R-MBA] and then with brine (1 × 20 mL), dried over
MgSO4, filtered, and evaporated to dryness (rotary evaporate at
e25 °C for short periods) to obtain the crude product (this material
is used to determine the de). Purification by silica gel flash
chromatography provides the mixture of diastereomers as a colorless
viscous liquid (rotary evaporate at e25 °C for short periods);
treatment with etheral HCl allows hydrochloride salt formation.
Note that amine products 2 are considered to be semivolatile and
converted to their HCl salts to enable high vacuum drying (defined
as 0.5-2.0 mm Hg) to constant weight (generally for g12 h) for
yield determination.
General Procedure: Catalytic Yb(OAc)3, Y(OAc)3, or Ce-
(OAc)3 (Normal de). In a reaction vessel, the Lewis acid [Yb-
(OAc)3 10 mol %, Y(OAc)3 (15 mol %) or Ce(OAc)3 (15 mol %)]21
was added and subsequently evacuated under high vacuum for 5
min before flooding with nitrogen. To the vessel were added
anhydrous methanol (2.5 mL, 1.0 M), ketone (2.5 mmol, 1.0 equiv),
and (S)-R-methylbenzylamine (0.35 mL, 2.75 mmol, 1.1 equiv) and
the mixture stirred for 20-30 min at the temperature at which the
hydrogenation was performed (22 or 50 °C). A THF slurry of
Raney-Ni (100 wt % based on the ketone, triturated with EtOH
(×3) and then with anhydrous THF (×3) before addition) was
transferred to the reaction mixture using 2.5 mL of anhydrous THF
(final molarity of reaction solution is 0.5 M). The vessel was then
pressurized to the indicated pressure 120-290 psi (8-20 bar) of
hydrogen and stirred at room temperature or at 50 °C as indicated.
At 12 h (<3 area % of ketone and intermediate ketimine by GC),
the reaction mixture was worked-up as delineated in the above
section (General Procedure: Stoichiometric Yb(OAc)3 (Enhanced
de)).
starting material with (R)- or (S)-R-MBA. In contrast to these
stereochemical trends γ-branched and nonbranched 2-alkanones,
e.g., benzylacetone (1c) and 2-octanone (1d), respectively,
undergo reductive amination in the presence of Yb(OAc)3 (50-
110 mol %) providing enhanced diastereoselectivity with good
yield. Finally, failure to use the correct acid (Lewis or Brønsted)
during reductive amination, or no acid at all, results in high
alcohol byproduct formation.
Conclusion
Strategies for R-chiral amine synthesis employing isolated
ketimines or enamines are stepwise long and can suffer from
lower overall yield because of mediocre ketimine or enamine
yield forming steps, as previously commented on.46 This
problem can be alleviated by using a reductive amination
strategy as outlined here, and when followed by hydrogenolysis
allows enantioenriched primary amine synthesis in good to high
overall yield in two reaction steps from prochiral ketones.
To our knowledge ytterbium acetate has not been previously
investigated for influencing the double bond geometry of imines
or enamines and the use of a ytterbium salt for reductive
amination has no prior precedent. The speculative mechanism
elaborated on here suggests that a Yb(OAc)3-promoted cis- to
trans-ketimine isomerization pathway may be operative and
responsible for the enhanced product de noted for γ-branched
and straight-chain 2-alkanones.
Experimental Section
General Procedure: Stoichiometric Yb(OAc)3 (Enhanced de).
In a dry reaction vessel, Yb(OAc)3 (0.96 g, 2.75 mmol, 1.1 equiv)21
was added and subsequently evacuated under high vacuum for 5
min before flooding with nitrogen, anhydrous MeOH (2.5 mL, 1.0
M) was then added. To this solution a prochiral ketone 1 (2.5 mmol,
1.0 equiv) and (S)-R-methylbenzylamine (0.35 mL, 2.75 mmol, 1.1
equiv) were added and subsequently stirred at room temperature
for 20-30 min. A THF slurry of Raney-Ni (100 wt % based on
the ketone, triturated with EtOH (×3) and then with anhydrous
THF (×3) before addition) was transferred to the reaction mixture
using 2.5 mL of anhydrous THF (final molarity of reaction solution
is 0.5 M) and the reaction vessel pressurized at 120 psi (8.3 bar)
of hydrogen. After being stirred for 12 h at 22 °C, <3 area % of
General Procedure: Brønsted Acids (Normal de). The reaction
vessel was evacuated under high vacuum for 5 min before flooding
with nitrogen. To the vessel were added anhydrous methanol (2.5
mL, 1.0 M), acetic acid (20 mol %), ketone (2.5 mmol, 1.0 equiv)
(1), and (S)-R-methylbenzylamine (0.35 mL, 2.75 mmol, 1.1 equiv),
and the mixture was stirred for 20-30 min at the temperature at
which the hydrogenation was performed (22 or 50 °C). The
remaining procedural details should be followed as in the above
section (General Procedure: Catalytic Yb(OAc)3, Y(OAc)3, or Ce-
(OAc)3 (Normal de)).
(44) The mechanism presented here (Scheme 3) is more likely than those
in which simple ligation of Yb(OAc)3 to the in situ formed cis- and trans-
ketimine mixture allows improved facial selectivity for the addition of
hydrogen. This is supported by the fact that no other lanthanide or transition
metal acetate, chloride, or triflate (over 30 individual metals were examined)
was identified as capable of providing enhanced de; and importantly other
ytterbium salts, e.g., YbCl3 and Yb(OTf)3, did not allow enhanced
diastereoselectivity (Table 1). Furthermore, the rate of reaction is unchanged
in the presence Yb(OAc)3, implying that the ligated ketimine is not the
species being reduced but instead the free ketimine. Finally, scenarios in
which only one of the ketimine diastereomers is ligated to Yb(OAc)3 are
unlikely because catalytic quantities of Yb(OAc)3 do not allow enhanced
diastereoselectivity.
Acknowledgment. We are grateful for financial support from
Jacobs University Bremen. This work has been performed within
the graduate program of Nanomolecular Science.
(45) Yb(OTf)3 could in theory undergo a similar ketimine isomerization
process via its sulfonate oxygen but as noted previously only provided the
alcohol byproduct under the reductive amination conditions noted here
(Table 1), implying it is too Lewis acidic. Regardless, when we synthesized
and then treated the N-R-MBA ketimine of 2-octanone with Yb(OTf)3 (110
mol %) the desired product 2d was noted, albeit without enhanced
stereoselectivity.
Supporting Information Available: Detailed experimental
procedures are provided for each amine (2) synthesized, and all
compounds have been previously and fully characterized.1c 1H and
13C NMR data for amines showing enhanced de (2a-f) is provided;
GC chromatograms are provided for all amines 2. This material is
available free of charge via the Internet at http://pubs.acs.org.
(46) For references regarding imine formation in general, see ref 1c, p
1291.
JO7021235
J. Org. Chem, Vol. 73, No. 4, 2008 1305