Umpolung by N-heterocyclic carbenes: Generation and reactivity of the elusive 2,2-diamino enols (breslow intermediates)

Article abstract of DOI:10.1002/anie.201205878

54 years later: Saturated imidazolidin-2-ylidenes react with aldehydes to smoothly produce the elusive 2,2-diamino enols A ( Breslow intermediates , first postulated in 1958) of carbene-catalyzed umpolung reactions. The 2,2-diamino enols A react with additional aldehyde in a cross-benzoin reaction. The methylated Breslow intermediates B are accessible by deprotonation of methoxymethyl azolium salts. Copyright

Full text of DOI:10.1002/anie.201205878

.
Angewandte
Communications
DOI: 10.1002/anie.201205878
Organocatalysis
Umpolung by N-Heterocyclic Carbenes: Generation and Reactivity of
the Elusive 2,2-Diamino Enols (Breslow Intermediates)**
Albrecht Berkessel,* Silvia Elfert, V. Reddy Yatham, Jçrg-M. Neudçrfl, Nils E. Schlçrer, and
J. Henrique Teles
Dedicated to Professor Ronald Breslow
In biochemistry, reactions such as pyruvate decarboxylation
and benzoin-type condensations that are catalyzed by vita-
min B1 dependent enzymes were considered some of the
mechanistically most “mysterious” transformations, until in
1958 Breslow proposed that all these reactions hinge on the
formation of an N-heterocyclic carbene (NHC) as the
catalytically active entity (Scheme 1).[1–3] The azolium salt I
(e.g. X = S, in thiamine) is deprotonated to form thiazolin-2-
ylidene II, the catalytically active species. Nucleophilic attack
of carbene II onto an aldehyde generates the primary
carbene–aldehyde adduct III. A subsequent protonation/
deprotonation affords amino enol IV.
Amino enol IVacts as an acyl anion equivalent, and reacts
with the carbonyl group of a second aldehyde molecule,
thereby generating addition product V. A second proton
transfer follows, and the elimination of the benzoin product
regenerates the carbene catalyst II. In honor of his ground-
breaking work, the nucleophilic amino enol IV has been
addressed as the “Breslow intermediate” ever since. It
represents the assumed chemical entity crucial for all
biochemical and organocatalytic umpolung reactions.
Dozens of excellent publications on new applications of
carbene umpolung catalysis appear each year, and they all
invoke Breslow intermediates as the crucial intermediates.[4,5]
Scheme 1. Catalytic cycle of the benzoin condensation as proposed by
Breslow.[1,2]
Despite its central importance in bio- and organocatalysis,
no unambiguous generation and characterization of Breslow
intermediates, that is, 2,2-diamino enols, has been reported to
date. Rovis and co-workers recently disclosed that the
reversible addition of triazolylidene carbenes to iminium
ions furnishes 1,2,2-triaminoethenes.[6,7] However, these aza
analogues of 2,2-diamino enols do not show the expected
reactivity of a Breslow intermediate, that is, formation of
a cross-benzoin product in the presence of an aldehyde.
Herein, we describe our own approach, which ultimately led
to the generation of a number of 2,2-diamino enols (Breslow
intermediates), their unambiguous characterization by NMR
spectroscopy, and the proof of their reactivity in cross-
benzoin reactions. Thus, we could provide evidence for the
existence of Breslow intermediates, for their accessibility
from N-heterocyclic carbenes and aldehydes, and for their
postulated acyl anion reactivity.
Our own previous studies were aimed at generating
Breslow intermediates based on 1,2,4-triazolylidene carbenes,
such as 1a, as they are readily accessible and highly reactive
umpolung catalysts (Scheme 2).[8,9] We reported earlier that in
the presence of excess aldehyde the primary carbene–
aldehyde adduct 2 is reversibly converted into the spiro-
dioxolane 2a, the resting state of the catalytic system in the
case of aliphatic aldehydes.[10] The addition of acid results in
intermediate 2 being reversibly protonated to give (isolable)
2b.[9] Alternatively, intermediate 2 can irreversibly isomerize
to ketone 3. The latter is catalytically incompetent and thus
a “dead end” in the catalytic system.[10]
[*] Prof. Dr. A. Berkessel, S. Elfert, V. R. Yatham, Dr. J.-M. Neudçrfl,
Dr. N. E. Schlçrer
Cologne University, Department of Chemistry
Greinstrasse 4, 50939 Cologne (Germany)
E-mail: berkessel@uni-koeln.de
Homepage: http://www.berkessel.de
Dr. J. H. Teles
BASF SE, GCS/X-M311
67056 Ludwigshafen (Germany)
[**] Support by the Fonds der Chemischen Industrie and by BASF SE is
gratefully acknowledged. We thank Felix Klauck for providing
samples of 13C-labeled 2,4-bis(trifluoromethyl)benzaldehyde.
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201205878.
12370
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2012, 51, 12370 –12374

Angewandte
Chemie
To our extreme delight, NMR spectroscopic analysis
revealed the clean formation—within a few minutes—of the
2,2-diamino enols 5d and 6e when the carbenes SIMes (1d)
and SIPr (1e) were combined with benzaldehyde and 2,4-
bis(trifluoromethyl)benzaldehyde, respectively, at room tem-
perature in [D8]THF under rigorous exclusion of oxygen.
Figure 1 (top) shows the 1H NMR spectrum of the 2,2-
diamino enol 6e. The use of 13C-labeled 2,4-bis(trifluoro-
methyl)benzaldehyde (13CHO) revealed the expected 13C-
OH/C couplings and thus additionally confirmed the chemical
Scheme 2. Formation of spiro-dioxolane (2a) and ketone (3) in the
reaction of a 1,2,4-triazolylidene with an aldehyde.
Our attempts to induce tautomerization of ketones 3 to
their reactive enol form 4 (by numerous acids, bases, silylating
agents) met with frustration—with decomposition or the
formation of isomeric silyl enol ethers as the typical outcome.
As gauged by DFT calculations (see the Supporting Informa-
tion), the enol form 4 (Scheme 2) is about 50 kJmolÀ1 higher
in energy than the corresponding ketone (for R = Ph). Our
calculations revealed, however, that the introduction of
electron-withdrawing substituents should make the enol
form more readily accessible (e.g. for R = 2,4-bis(trifluoro-
methyl)phenyl: DDG ca. 20 kJmolÀ1, Gibbs free energies at
298 K in the gas phase). Unfortunately, treatment of carbene
1a with 2,4-bis(trifluoromethyl)benzaldehyde still gave
ketone 3 exclusively—with no tautomerization to enol form
4. We therefore turned our attention to “tuning” the carbene
moiety. N-Heterocyclic carbenes can generally be subdivided
into those in which the heterocycle can become an aromatic
azolium ion, and those in which saturation at C4,5 precludes
aromatization (Scheme 3). In view of the expected lower
reactivity/higher stability of Breslow intermediates derived
from saturated N-heterocyclic carbenes, we focused our study
on the interaction of aldehydes with dihydro-IMes (SIMes,
1d) and the related dihydro-IPr (SIPr, 1e). Our reasoning was
also supported by an elegant kinetic study by Mayr and co-
workers,[11] who recently analyzed the reactivity of 2-benzyl-
idene derivatives (i.e. “desoxy-Breslow intermediates”, which
lack the enol-hydroxy function[12]) of IMes (1b) and SIMes
(1d), and also of the 1,2,4-triazolylidene carbene 1a, with
benzhydrylium ions as the electrophiles. These results con-
firmed a quite significantly higher (kunsat./ksat. ca. 1000)
reactivity of the IMes (1b) and triazolylidene (1a) derivatives
relative to the one derived from saturated SIMes (1d).
Figure 1. Top : 1H NMR spectrum ([D8]THF, 600 MHz) of the reaction
of SIPr (1e) with 2,4-bis(trifluoromethyl)benzaldehyde (1 equiv) which
results in the formation of the 2,2-diamino enol 6e. Bottom: 2,2-
diamino enol 6e, 13C label at C6.
shift assignments of enol 6e: d = 4.40 ppm (d, 2JCOH = 3.23 Hz,
see Figure 1, bottom); d = 145.8 ppm (d, 1JC6-C2 = 106.4 Hz,
1
C2), d = 142.5 ppm (d, JC6-Car = 69.8 Hz, Car). The labeled C
atom (C6) itself resonates at d = 109.5 ppm (see the Support-
ing Information for the 13C NMR spectra). Most character-
istic is the resonance at d = 4.40 ppm which is assigned to the
hydroxy proton of 2,2-diamino enol 6e, supported by its rapid
H/D exchange upon addition of [D4]MeOH (see the Support-
ing Information for spectra). Another characteristic feature
of the enol OH proton is the pronounced temperature
dependency of its chemical shift (Figure 2). The variable-
temperature 1H NMR spectra shown in Figure 2 also revealed
coalescence of the enolꢀs imidazolidine (C4,C5) protons (d ca.
3.7–4.0 ppm at 258C) at approximately À558C. The under-
lying conformational behavior of the 2,2-diamino enol 6e will
1
be the subject of future studies. Note that identical H NMR
spectra were recorded before and after cooling the solution to
À958C (Figure 2, top and bottom red traces).
By simple combination of a saturated carbene with one
equivalent of aldehyde, we were able to generate cleanly the
2,2-diamino enols 5d, 5e–10e shown in Figure 3. NMR
spectroscopy did not indicate any degradation of the 2,2-
diamino enols over several hours under rigorous exclusion of
oxygen. Instantaneous decomposition occurs upon admission
of air. As exemplarily tested with 13C-labeled enol 5e (label at
C6), exposure of the mixture to acid (acetic acid, 2 equiv)
Scheme 3. 4,5-Unsaturated (1b,c) and saturated (1d,e) N-heterocyclic
carbenes.
Angew. Chem. Int. Ed. 2012, 51, 12370 –12374
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 12371

.
Angewandte
Communications
Figure 2. Variable-temperature 1H NMR spectra ([D8]THF, 600 MHz,
258C to À958C) of the 2,2-diamino enol 6e. At T=258C, the OH
proton of the enol resonates at d=4.40 ppm.
Figure 4. Formation of the benzoin-azolium salt 11 from SIPr (1e) and
2,4-bis(trifluoromethyl)benzaldehyde, and its X-ray crystal structure.[13]
tion, however, allows benzoin formation to proceed
(Figure 4).
With this in mind, we examined the reaction of the
benzaldehyde-derived and presumably more nucleophilic 2,2-
diamino enol 5e. Traces of homobenzoin were at best formed
in the presence of excess benzaldehyde (10 equiv). However,
when diamino enol 5e was exposed to the more electrophilic
2,4-bis(trifluoromethyl)benzaldehyde, NMR spectroscopic
analysis first of all revealed a comparatively rapid equilibra-
tion between the benzaldehyde-derived diamino enol 5e and
the 2,2-diamino enol 6e derived from 2,4-bis(trifluoro-
methyl)benzaldehyde (Figure 5), thus proving the reversibil-
ity of the 2,2-diamino enol formation. Simultaneously, the
expected cross-benzoin product 12 was produced slowly from
the 2,2-diamino enol 5e. The identity of the cross-benzoin
product 12 was proven by independent synthesis (see the
Supporting Information). Neither its isomer 13, nor signifi-
cant amounts of the two possible homobenzoin products were
formed.
Figure 3. 13C NMR shifts [ppm] of C2/C6 of the 2,2-diamino enols 5d,
5e–10e ([D8]THF, 258C); 5d: R=2,4,6-trimethylphenyl, 5e–10e:
R=2,6-bis(2-propyl)phenyl.
Clearly, the reaction manifold shown in Figure 5 can also
be accessed from compound 6e: The formation of 12 was also
observed when 6e was generated from carbene 1e and
13CHO-labeled 2,4-bis(trifluoromethyl)benzaldehyde [that is,
13C label at C6 (C-OH) in 6e] and exposed to benzaldehyde
(1:1).
We furthermore succeeded in preparing 2,2-diamino enol
ethers of type 15. As shown in Scheme 4, we reasoned that
these methyl enol ethers should be accessible by alkylation of
N-heterocyclic carbenes such as IMes (1b), IPr (1c) or SIMes
(1d) with MOM chloride (MOM: methoxymethyl), followed
by deprotonation. The alkylation of the carbene (1b and 1c)
azolium salts with tBuOK and MOM-Cl proceeded smoothly
in THF at room temperature, thereby affording the salts 14b
and 14c in almost quantitative yields (see the Supporting
Information for experimental details and the X-ray crystal
structure of the azolium salt 14c).[13]
results in clean fission to the corresponding aldehyde and
azolium salt.
The smooth formation of 2,2-diamino enols proves the
accessibility of the Breslow intermediate from aldehydes and
N-heterocyclic carbenes. By using Ph-C2HO and 1H/2H NMR
spectroscopy we could show that the source of the hydroxylic
proton/deuteron in the 2,2-diamino enol product is indeed the
aldehydic H/2H atom (see the Supporting Information for 1H
and 2H NMR spectra).
NMR spectroscopic studies in [D8]THF revealed that SIPr
(1e) still formed the 2,2-diamino enol 6e when treated with
an excess (10 equiv) of 2,4-bis(trifluoromethyl)benzaldehyde.
The use of 13C-labeled aldehyde (13CHO) confirmed that no
homobenzoin formation took place. However, crystallization
of the benzoin-azolium salt 11 (Figure 4) occured when n-
pentane was used as the solvent (1 equiv of aldehyde). In
other words, the 2,2-diamino enol is the energetically most
favorable state of the equilibrating catalytic system in
homogeneous solution. Shifting the equilibria by crystalliza-
Treatment of suspensions of salts 14b,c in [D8]THF with
tBuOK (1.1 equiv) resulted in deprotonation, as evident by
rapid dissolution. NMR spectroscopic examination of the
12372 www.angewandte.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2012, 51, 12370 –12374

Angewandte
Chemie
(C6) ppm. It is noteworthy that no 2,2-diamino enol could
be generated cleanly from the unsaturated carbene IMes (1b)
and aldehyde, whereas the O-methylated (i.e. protected) enol
15b was readily obtained by the route outlined in Scheme 4.
Upon removal of the THF solvent, both enol ethers 15b
and 15d could be crystallized from n-pentane, and their
crystal structures are shown in Figure 6. The crystal structures
Figure 5. Top: Reaction manifold involving the 2,2-diamino enols 5e
and 6e, benzaldehyde, and 2,4-bis(trifluoromethyl)benzaldehyde.
Bottom: Time course (1H NMR spectroscopy in [D8]THF) of the
reaction of the 2,2-diamino enol 5e with 2,4-bis(trifluoromethyl)benz-
aldehyde (1 equiv), which results in the rapid formation of the diamino
enol 6e and the slow production of the cross-benzoin 12. The first
measurement was at 3 min; the signal integral of 2,2-diamino enol 5e
is set to 1.0.
Figure 6. X-ray crystal structures of the methylated Breslow intermedi-
ates 15b (top) and 15d (bottom).[13]
of the methyl enol ethers 15b and 15d clearly reflect the
saturated (15d) versus unsaturated (15b) character of the
=
heterocycles. The C C bond of the 2,2-diamino enol ether
moiety, however, is virtually the same length in both
compounds. Consequently, a zwitterionic resonance form
involving an aromatic azolium does not appear to contribute
significantly to the structural features of the 2,2-diamino enol
ether 15b. This is in line with X-ray data obtained by Mayr
and co-workers on the related 2-benzylidene IMes (dC
=
=
C
1.361 ꢁ) and 2-benzylidene SIMes (dC C = 1.354 ꢁ).[11]
=
The 2,2-diamino enol ethers 15b,c are readily alkylated at
C6 by methyl iodide (see the Supporting Information for
NMR spectra). No reaction was observed when 15c was
exposed, at room temperature, to aldehydes such as benz-
aldehyde or 2,4-bis(trifluoromethyl)benzaldehyde. The
methylated Breslow intermediates 15b–d were almost indef-
initely stable in solution under rigorous exclusion of oxygen.
The admission of air, however, results in rapid oxidation to
afford the corresponding heterocyclic ureas (see the Support-
ing Information for the X-ray crystal structure of the IMes-
and the IPr-urea).[13]
In summary, we have reported the generation and NMR
spectroscopic characterization of 2,2-diamino enols (Breslow
intermediates) from N-heterocyclic carbenes and aldehydes.
We furthermore showed that these 2,2-diamino enols can
react as acyl anion equivalents with additional aldehydes to
provide benzoins. In addition, we have generated O-methy-
Scheme 4. Preparation of the methylated Breslow intermediates
15b–d. NMR data: [D8]THF, 500/400 MHz.
resulting solutions revealed the clean formation of the enol
ethers 15b and 15c, respectively, as the sole products. The
1
most characteristic H and 13C NMR shifts are included in
Scheme 4 (see Supporting Information for one- and two-
dimensional NMR spectra of 15b,c). The imidazolidine-
derived enol ether 15d was prepared simply by treating
MOM-Cl with two equivalents of SIMes (1d). As in the case
of the 2,2-diamino enols 5d and 5e–10e (Figure 3), the 13C
resonances of the 2,2-diamino enol ether moiety are all found
in the typical regions of d = 138–146 (C2) and 100–115
Angew. Chem. Int. Ed. 2012, 51, 12370 –12374
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 12373

.
Angewandte
Communications
lated 2,2-diamino enols and characterized them, inter alia, by
X-ray crystallography.
[1] R. Breslow, J. Am. Chem. Soc. 1957, 79, 1762 – 1763.
[2] R. Breslow, J. Am. Chem. Soc. 1958, 80, 3719 – 3726.
[3] Review: R. Kluger, K. Tittmann, Chem. Rev. 2008, 108, 1797 –
1833.
Experimental Section
6e: In a glovebox, an NMR tube was charged with SIPr 1e (20 mg,
51 mmol, 1.0 equiv) in [D8]THF and sealed with a septum. 2,4-
Bis(trifluoromethyl)benzaldehyde
[4] Recent reviews: a) A. Grossmann, D. Enders, Angew. Chem.
2012, 124, 320 – 332; Angew. Chem. Int. Ed. 2012, 51, 314 – 325;
b) X. Bugaut, F. Glorius, Chem. Soc. Rev. 2012, 41, 3511 – 3522;
c) T. Drçge, F. Glorius, Angew. Chem. 2010, 122, 7094 – 7107;
Angew. Chem. Int. Ed. 2010, 49, 6940 – 6952; d) J. Douglas, G.
Churchill, A. D. Smith, Synthesis 2012, 2295 – 2309.
[5] For the characterization of acyl azolium intermediates in
conjugate umpolung reactions, see J. Mahatthananchai, P.
Zheng, J. W. Bode, Angew. Chem. 2011, 123, 1711 – 1715;
Angew. Chem. Int. Ed. 2011, 50, 1673 – 1677.
[6] D. A. DiRocco, K. M. Oberg, T. Rovis, J. Am. Chem. Soc. 2012,
134, 6143 – 6145.
[7] D. A. DiRocco, T. Rovis, Angew. Chem. 2012, 124, 6006 – 6008;
Angew. Chem. Int. Ed. 2012, 51, 5904 – 5906.
[8] D. Enders, K. Breuer, G. Raabe, J. Runsink, J. H. Teles, J.-P.
Melder, K. Ebel, S. Brode, Angew. Chem. 1995, 107, 1119 – 1122;
Angew. Chem. Int. Ed. Engl. 1995, 34, 1021 – 1023.
[9] J. H. Teles, J.-P. Melder, K. Ebel, R. Schneider, E. Gehrer, W.
Harder, S. Brode, D. Enders, K. Breuer, G. Raabe, Helv. Chim.
Acta 1996, 79, 61 – 83.
(8.4 mL, 51 mmol, 1.0 equiv) was added by
means of a syringe and the reaction was
followed by NMR spectroscopy. The
1H NMR spectrum recorded 3 min after
addition of the aldehyde revealed full
conversion of the carbene 1e into the 2,2-
diamino enol 6e. 1H NMR (600 MHz,
[D8]THF): d = 7.44 (brs, 1H; H25), 7.18
(t, 3JHH = 7.3 Hz, 1H; H17), 7.14 (d, 3JHH
=
3
7.3 Hz, 2H; H16, H16’), 6.98 (dd, JHH
=
8.2 Hz, 4JHH = 1.2 Hz, 1H; H23), 6.95 (t,
3
3JHH = 7.0 Hz, 1H; H10), 6.92 (d, JHH
=
3
8.2 Hz, 1H; H22), 6.92 (d, JHH = 7.0 Hz,
2H; H9, H9’), 4.40 (s, 1H; OH), 3.93 (t,
3
3JHH = 7.6 Hz, 2H; H5), 3.77 (t, JHH
=
=
3
7.6 Hz, 2H; H4), 3.64 (sept, JHH
6.8 Hz, 2H; H18, H18’), 3.48 (sept, 3JHH = 6.8 Hz, 2H; H11, H11’),
1.32 (d, 3JHH = 6.8 Hz, 6H; H19, H19’), 1.30 (d, 3JHH = 6.8 Hz, 6H;
H20, H20’), 1.26 (d, 3JHH = 6.8 Hz, 6H; H13, H13’), 1.18 ppm (d,
3JHH = 6.8 Hz, 6H; H12, H12’); 13C NMR (150 MHz, [D8]THF): d =
148.3 (2C; C15, C15’), 147.4 (2C; C8, C8’), 145.8 (1C; C2), 142.5 (1C;
C21), 141.9 (1C; C14), 140.7 (1C; C7), 131.6 (1C; C22), 127.9 (1C;
C10), 127.7 (1C; C17), 126.4 (1C; C26), 126.3 (1C; C23), 125.0 (1C;
C27), 124.9 (1C; C28), 124.5 (2C; C9, C9’), 124.4 (1C; C24), 124.3
(2C; C16, C16’), 124.1 (1C; C25), 109.5 (1C; C6), 54.1 (1C; C5), 52.4
(1C; C4), 29.5 (2C; C18, C18’), 28.9 (2C; C11, C11’), 26.9 (2C; C13,
C13’), 25.2 (2C; C19, C19’), 24.0 (2C; C20, C20’), 22.9 ppm (2C; C12,
C12’); 19F NMR (376 MHz, [D8]THF): d = À63.25 (s, 3F; C27-F3),
À60.15 ppm (s, 3F; C28-F3).
[10] a) A. Berkessel, S. Elfert, K. Etzenbach-Effers, J. H. Teles,
Angew. Chem. 2010, 122, 7275 – 7279; Angew. Chem. Int. Ed.
2010, 49, 7120 – 7124; b) for a subsequent computational study,
see O. Hollꢂczki, Z. Kelemen, L. Nyulꢃszi, J. Org. Chem. 2012,
77, 6014 – 6022.
[11] a) B. Maji, M. Horn, H. Mayr, Angew. Chem. 2012, 124, 6335 –
6339; Angew. Chem. Int. Ed. 2012, 51, 6231 – 6235; b) after
acceptance of this paper,
a related publication appeared,
describing the generation and reactivity of other methyl
diaminoenol ethers: B. Maji, H. Mayr, Angew. Chem. 2012,
124, 10554 – 10558; Angew. Chem. Int. Ed. 2012, 51, 10408 –
10421.
[12] a) C. E. I. Knappke, J.-M. Neudçrfl, A. Jacobi von Wangelin,
Org. Biomol. Chem. 2010, 8, 1695 – 1705; b) C. E. I. Knappke,
A. J. Arduengo III, H. Jiao, J.-M. Neudçrfl, A. Jacobi von Wan-
gelin, Synthesis 2011, 3784 – 3795.
Received: July 23, 2012
Revised: August 31, 2012
[13] CCDC 892932 ()11, 892933 (14c), 892934 (15b), 892935 (15d),
892936 (IMes-urea), 892937 (IPr-urea) contain the supplemen-
tary crystallographic data for this paper. These data can be
obtained free of charge from The Cambridge Crystallographic
Data Centre via www.ccdc.cam.ac.uk/data_request/cif.
Published online: October 18, 2012
Keywords: Breslow intermediates · carbenes ·
.
NMR spectroscopy · reaction mechanisms · umpolung
12374 www.angewandte.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2012, 51, 12370 –12374