Deuterium-labeling and NMR study of the dearomatization of N-alkyl-N-benzyldiphenylphosphinamides through anionic cyclization: Ortho and benzylic lithiation directed by complex-induced proximity effects

Article abstract of DOI:10.1021/ja039863t

The mechanism of the anionic cyclization dearomatizing reaction of N-benzyl-N-methyldiphenylphosphinamide (1) upon treatment with s-BuLi in tetrahydrofuran (THF) at -90 °C has been analyzed by deuterium-labeling and natural abundance multinuclear magnetic

Full text of DOI:10.1021/ja039863t

Published on Web 09/14/2004
Deuterium-Labeling and NMR Study of the Dearomatization of
N-Alkyl-N-benzyldiphenylphosphinamides through Anionic
Cyclization: Ortho and Benzylic Lithiation Directed by
Complex-Induced Proximity Effects
Ignacio Ferna´ndez,† Javier Gonza´lez,‡ and Fernando Lo´pez-Ortiz*,†
Contribution from the AÄ rea de Qu´ımica Orga´nica, UniVersidad de Almer´ıa, Carretera de
Sacramento s/n, Almer´ıa, Spain, and Departamento de Qu´ımica Orga´nica e Inorga´nica,
UniVersidad de OViedo, C/Julia´n ClaVer´ıa 8, 33006 OViedo, Spain
Received November 30, 2003; E-mail: flortiz@ual.es
Abstract: The mechanism of the anionic cyclization dearomatizing reaction of N-benzyl-N-methyldiphen-
ylphosphinamide (1) upon treatment with s-BuLi in tetrahydrofuran (THF) at -90 °C has been analyzed by
deuterium-labeling and natural abundance multinuclear magnetic resonance (1H, 2H, 7Li, 13C, 31P) studies.
In the absence of coordinating cosolvents such as hexamethylphosphoramide (HMPA), eight major anionic
species were identified, which allowed us to unravel the pathway of the metalation reaction. In agreement
with the complex-induced proximity effect (CIPE) mechanism, the sequence of transformations emerging
from this study involves the coordination of the lithium base to the PdO group of 1 to give four dimeric
precomplexes whose NMR data are consistent with structures Va/Vb and VIIIa/VIIIb. The diastereomers
Va/Vb are the precursors of the monomeric benzylic anion II, whereas the VIIIa/VIIIb diastereomers are
assumed to undergo ortho deprotonation leading to anions I. Translocation from the ortho anion to the
benzylic one is not observed. Intramolecular conjugate addition of anion II to the P-phenyl rings happens
in a reversible way, affording the monomeric dearomatized anions III, IV, VI, and VII. The reaction progresses
to yield a mixture containing only the species I, III, and IV. HMPA acts as a catalyst for the ortho-to-
benzylic translocation and anionic cyclization reactions. Two-dimensional (2D) 7Li,31P{1H} shift correlations
and 7Li{31P} NMR spectra proved to be crucial for the structural assignment of the anionic species. These
techniques also demonstrated the diastereotopicity of the two achiral ligands involved in a dimer with s-BuLi
(Vb) owing to the slow configuration inversion of the carbanion center.
been further manipulated to provide heterocyclic derivatives.5
A more efficient alternative to this approach is provided by the
Introduction
Dearomatization reactions of aromatic hydrocarbons have
attracted much attention for quite some time. This continued
interest is driven by the possibility of using low-cost and readily
accessible starting materials for the construction of functional-
ized carbocyclic compounds with a defined substitution pattern
and stereochemistry in a one-pot process.1 The intermolecular
conjugate addition of a nucleophile to an electron-deficient
aromatic hydrocarbon is one of the methods best suited to
breaking down the conjugated π system of an aromatic ring.2
The usefulness of this strategy has been demonstrated through
the preparation of a series of natural bioactive substances3 and
some biomimetics.4 The initial dearomatized products have also
anionic cyclization dearomatizing methodology.6 The general
procedure consists of the formation of a carbanion adjacent to
a heteroatom through a metalation reaction directed by an
electron-withdrawing group, which also activates the aromatic
ring toward the subsequent intramolecular nucleophilic attack.
Amides,7 sulfones,8 sulfonamides,9 and phosphinamides10 are
suitable functional groups for this dual role and have enabled
the introduction of nitrogen, sulfur, and phosphorus heteroatoms,
respectively, into the heterocyclic moiety of dearomatized
systems. The dearomatized compounds have been used as
(3) (a) Meyers, A. I.; Roth, G. P.; Hoyer, D.; Barner, B. A.; Laucher, D. J.
Am. Chem. Soc. 1988, 110, 4611. (b) Andrews, R. C.; Teague, S. J.; Meyers,
A. I. J. Am. Chem. Soc. 1988, 110, 7854. (c) Meyers, A. I.; Higashiyama,
K. J. Org. Chem. 1987, 52, 4592. (d) Robichaud, A. J.; Meyers, A. I. J.
Org. Chem. 1991, 56, 2607. (e) Gant, T. G.; Meyers, A. I. Tetrahedron
Lett. 1993, 34, 3707. (f) Hulme, A. N.; Meyers, A. I. J. Org. Chem. 1994,
59, 952. (g) Hulme, A. N.; Henry, S. S.; Meyers, A. I. J. Org. Chem. 1995,
60, 1265.
† Universidad de Almer´ıa.
‡ Universidad de Oviedo.
(1) (a) Mander, L. N. Synlett 1991, 134. (b) Donohoe, T. J.; Garg, R.; Stevenson,
C. A. Tetrahedron: Asymmetry 1996, 7, 317. (c) Schultz, A. Chem.
Commun. 1999, 1263. (d) Lukeman, M.; Wan, P. J. Am. Chem. Soc. 2003,
125, 1164.
(2) (a) Semmelhack, M. F. In ComprehensiVe Organic Synthesis; Trost, B.
M., Fleming, I., Eds.; Pergamon Press: Oxford, 1991; Vol. 4, pp 517-
549. (b) Harman, W. D. Chem. ReV. 1997, 97, 1953. (c) Brooks, B. C.;
Gunnoe, T. B.; Harman, W. D. Coord. Chem. ReV. 2000, 206-207, 3. (d)
Pape, A. R.; Kaliappan, K. P.; Ku¨ndig, E. P. Chem. ReV. 2000, 100, 2917.
(e) Smith, P. L.; Chordia, M. D.; Harman, W. D. Tetrahedron 2001, 57,
8203. (f) Rose-Munch, F.; Rose, E. Eur. J. Inorg. Chem. 2002, 1269.
(4) (a) Alam, M.; Bechtold, C.; Skoog, M. S.; Gant, T. G.; Patrick, A.; Colonno,
R. J.; Meyers, A. I.; Li, H.; Trimble, J.; Lin, P.-F. AntiViral Res. 1993, 22,
131. (b) Basil, L. F.; Nakano, H.; Frutos, R.; Kopach, M.; Meyers, A. I.
Synthesis 2002, 2064.
(5) Degnan, A. P.; Meyers, A. I. J. Org. Chem. 2000, 65, 3503.
(6) Clayden, J. Organolithiums: SelectiVity for Synthesis; Pergamon: Amster-
dam, 2002.
9
10.1021/ja039863t CCC: $27.50 © 2004 American Chemical Society
J. AM. CHEM. SOC. 2004, 126, 12551-12564
12551

A R T I C L E S
Ferna´ndez et al.
Scheme 1a
building blocks for the synthesis of kainoids,11 natural products
analogues,8c,12 and functionalized γ-aminophosphinic acids.10,13
Recently, Clayden et al. demonstrated the feasibility of using
two different groups for aromatic ring activation and anion
stabilization in the dearomatization of oxazolylnaphthalenes
bearing an ether appendage.6,14 Upon metalation in the R
position with respect to the oxygen atom, the oxazoline Michael
acceptor15 drives an anionic cyclization reaction, yielding
dearomatized oxygen heterocycles.16
The optimization of the dearomatization-electrophilic trap-
ping reactions of N-alkyl-N-benzyl(diphenyl)phosphinamides
revealed some mechanistic hints that can be summarized as
follows:17 (1) coordinating solvents such as hexamethylphos-
phoramide (HMPA) or 1,3-dimethyl-3,4,5,6-tetrahydropyrimi-
din-2(1H)-one (DMPU) notably accelerate the reaction, the best
results being obtained with the latter; (2) significant amounts
of products derived from ortho and benzylic metalation were
formed for short metalation (t1) and electrophilic quenching (t2)
times; and (3) the anionic cyclization step is reversible, which
allows one to obtain products of either kinetic or thermodynamic
control. The general picture emerging from these experimental
results is shown in Scheme 1 for the particular case of N-benzyl-
N-methyl(diphenyl)phosphinamide (1). The products isolated (2,
3, and 4) are consistent with the participation of three possible
anions: the ortho-lithiated species I, the benzylic derivative II,
and the dearomatized intermediate III. This situation is similar
to that reported for the anionic cyclization of N-benzylaryl-
amides.18 By analogy with the mechanism proposed for the
dearomatization of these compounds, we assumed that the ortho
anion may translocate to the benzylic one. It must be pointed
out that Clayden et al. suggested that the cyclization step of
a (1) s-BuLi (2.5 equiv), DMPU (none or 6 equiv), THF, -90 °C, t1. (2)
E+, -90 °C, t2. (3) H2O.
lithiated arylamides is better described as an electrocyclic ring
closure rather than a standard Michael addition.19
The reactions that produce anions I and II upon treatment of
phosphinamide 1 with s-BuLi are examples of ortho- and
benzylic-directed metalation, respectively. Directed lithiations
are a topic of considerable debate.20 The selective deprotonation
of an ortho or benzylic position assisted by an electron-
withdrawing group bearing electron lone pairs may be explained
through the complex-induced proximity effect (CIPE) model.21
This mechanism considers the lithiation as a two-step process.
First, the coordination of the lithium cation of the base with
one Lewis basic heteroatom of the substrate results in the
formation of a complex. This complex brings the carbanionic
center of the base close to the acidic proton, thus favoring the
transfer of the proton in the second step. Ortho-directed
deprotonations have been interpreted by an alternative mech-
anism involving a one-step reaction. In this model, the meta-
lation is described as a kinetically controlled transformation for
which the term “kinetically enhanced metalation” has been
coined.22 A detailed analysis of the lithiation of phosphinamides
in THF solution in terms of these models has not been previously
performed. In this article we describe the isotopic-labeling and
NMR study of the mechanism of the reaction of N-alkyl-N-
benzyl(diphenyl)phosphinamides with s-BuLi leading to dearo-
matized products. Deuteration reactions afforded the distribution
of anionic species present in the reaction medium, and the use
of deuterated phosphinamides showed the feasibility of anion
translocation. NMR monitoring of the reaction made possible
identification of all the intervening lithiated intermediates and
(7) (a) Ahmed, A.; Clayden, J.; Yasin, S. A. Chem. Commun. 1998, 231. (b)
Ahmed, A.; Clayden, J.; Rowley, M. Chem. Commun. 1998, 297. (c)
Clayden, J.; Menet, C. J.; Mansfield, D. J. Org. Lett. 2000, 2, 4229. (d)
Clayden, J.; Tchabanenko, K.; Yasin, S. A.; Turnbull, M. D. Synlett 2001,
302. (e) Clayden, J.; Menet, C. J.; Mansfield, D. J. Chem. Commun. 2002,
38. (f) Clayden, J.; Knowles, F. E.; Menet, C. J. Synlett 2003, 1701. (g)
Clayden, J.; Knowles, F. E.; Menet, C. J. J. Am. Chem. Soc. 2003, 125,
9278.
(8) (a) Crandall, J. K.; Ayers, T. A. J. Org. Chem. 1992, 57, 2993. (b) Padwa,
A.; Filipkowski, M. A.; Kline, D. N.; Murphree, S.; Yeske, P. E. J. Org.
Chem. 1993, 58, 2061. (c) Clayden, J.; Kenworthy, M. N.; Heliwell, M.
Org. Lett. 2003, 5, 831.
(9) (a) Breternitz, H. J.; Schaumann, E.; Adiwidjaja, G. Tetrahedron Lett. 1991,
32, 1299. (b) Aggarwal, V. K.; Ferrara, M. Org. Lett. 2000, 2, 4107. (c)
Aggarwal, V. K.; Alonso, E.; Ferrara, M.; Spey, S. E. J. Org. Chem. 2002,
67, 2335. (d) Luisi, R.; Capriati, V.; Florio, S.; Ranaldo, R. Tetrahedron
Lett. 2003, 44, 2677.
(10) (a) Ferna´ndez, I.; Lo´pez-Ortiz, F.; Tejerina, B.; Garc´ıa-Granda, S. Org.
Lett. 2001, 3, 1339. (b) Ruiz-Go´mez, G.; Lo´pez Ortiz, F. Synlett 2002,
781.
(11) (a) Clayden, J.; Tchabanenko, K. Chem. Commun. 2000, 317. (b) Clayden,
J.; Menet, C. J.; Mansfield, D. J. Chem. Commun. 2002, 38. (c) Clayden,
J.; Menet, C. J.; Tchabanenko, K. Tetrahedron 2002, 58, 4727.
(12) (a) Ahmed, A.; Bragg, R. A.; Clayden, J.; Tchabanenko, K. Tetrahedron
Lett. 2001, 42, 3407. (b) Bragg, R. A.; Clayden, J.; Blandon, M.; Ichihara,
O. Tetrahedron Lett. 2001, 42, 3411.
(13) Ferna´ndez, I.; Lo´pez-Ortiz, F.; Mene´ndez-Vela´zquez, A.; Garc´ıa-Granda,
S. J. Org. Chem. 2002, 67, 3852.
(14) This is the general strategy used in anionic cyclization reactions that do
not involve the nucleophilic attack at an aromatic ring. For reviews, see:
(a) Mealy, M. J.; Bailey, W. F. J. Organomet. Chem. 2002, 646, 59. See
also: (b) Deng, K.; Bensari, A.; Cohen, T. J. Am. Chem. Soc. 2002, 124,
12106. (c) Bailey, W. F.; Daskapan, T.; Rampalli, S. J. Org. Chem. 2003,
68, 1334.
(19) Clayden, J.; Purewal, S.; Helliwell, M.; Mantell, S. J. Angew. Chem., Int.
Ed. 2002, 41, 1049.
(20) For reviews, see:(a) Klumpp, G. W. Recl. TraV. Chim. Pays-Bas 1986,
105, 1. (b) Snieckus, V. Chem. ReV. 1990, 90, 879. (c) Green, L.; Chauder,
B.; Snieckus, V. J. Heterocycl. Chem. 1999, 36, 1453. (d) Mugesh, G.;
Singh, H. B. Acc. Chem. Res. 2002, 35, 226. See also: (e) Chadwick, S.
T.; Rennels, R. A.; Rutherford, J. L.; Collum, D. B. J. Am. Chem. Soc.
2000, 122, 8640. (f) Saa´, J. M. HelV. Chim. Acta 2002, 85, 814.
(21) (a) Beak, P.; Meyers, A. I. Acc. Chem. Res. 1986, 19, 356. (b) Beak, P.;
Basu, A.; Gallagher, D. J.; Park, Y. S.; Thayumanavan, S. Acc. Chem.
Res. 1996, 29, 552.
(15) Oxazolines are one of the best activating groups for the conjugate addition
to naphthalene rings. (a) Reuman, M.; Meyers, A. I. Tetrahedron 1985,
41, 837. (b) Gant, T. G.; Meyers, A. I. Tetrahedron 1994, 50, 2297. (c)
Meyers, A. I. J. Heterocycl. Chem. 1998, 35, 991.
(16) Clayden, J.; Kenworthy, M. N. Org. Lett. 2002, 4, 787.
(17) Ferna´ndez, I.; Force´n-Acebal, A.; Lo´pez-Ortiz, F.; Garc´ıa-Granda, S. J.
Org. Chem. 2003, 68, 4472.
(18) (a) Ahmed, A.; Clayden, J.; Rowley, M. Tetrahedron Lett. 1998, 39, 6103.
(b) Bragg, R. A.; Clayden, J. Tetrahedron Lett. 1999, 40, 8327.
(22) (a) Van Eikema Hommes, N. J. R.; Schleyer, P. V. R. Angew. Chem., Int.
Ed. Engl. 1992, 31, 755. (b) Van Eikema Hommes, N. J. R.; Schleyer, P.
V. R. Tetrahedron 1994, 50, 5903. (c) Kremer, T.; Junge, M.; Schleyer, P.
V. R. Organometallics 1996, 15, 3345.
9
12552 J. AM. CHEM. SOC. VOL. 126, NO. 39, 2004

N-Alkyl-N-benzyldiphenylphosphinamides
A R T I C L E S
demonstrated the existence of an equilibrium between the
dearomatized anions corresponding to the kinetic and thermo-
dynamic products of the reaction. The study also showed that
deprotonation of 1 is preceded by a precomplexation step in
which the base and the phosphinamide are held in close
7
proximity. Li,31P shift correlations were used to establish the
connectivity between the 7Li,31P spin pairs of each species. The
role of coordinating cosolvents has been elucidated by perform-
ing the NMR measurements both in the presence and absence
of HMPA.
Results and Discussion
First, we will comment on the results obtained from deu-
teration reactions of dearomatized 1 and dearomatization of
deuterium-labeled phosphinamides. Second, an extensive NMR
study of the lithiation of 1 will be discussed. Within each section
the experiments performed in the absence of cosolvent are
treated first, and the effects of the addition of HMPA are
analyzed subsequently.
Isotopic-Labeling Studies. The general reaction conditions
for the synthesis of the benzoazaphospholes 4 (Scheme 1)
consisted of the treatment of a THF solution of 1 and 6 equiv
of DMPU (or HMPA) with 2.5 equiv of s-BuLi at -90 °C for
t1 min, followed by the addition of the appropriate electrophile
and stirring for t2 min at the same temperature. Unless otherwise
stated, the starting phosphinamide used throughout the text will
be 1. Phosphinamide 1 contains only one benzylic moiety, and
in terms of NMR, the methyl group is a good reporter of the
transformations that 1 may experience along the reaction
pathway. The reaction times t1 and t2 were optimized for each
electrophile; t1 ) t2 ) 30 min for trapping the dearomatized
species with organic protonating agents,13 whereas for aldehydes
the process was completed for reaction times as short as t1 ) t2
) 1 min.17 These differences mirror the different reaction rates
of the lithiated species present in the medium toward a proton
or a carbonyl group. The proton is so reactive that it irreversibly
quenches all anions immediately and in an indiscriminate
manner. The fact that in the sequence of dearomatization-
protonation reactions the metalation step requires 30 min to
achieve good yields of dearomatized products indicates that this
time is necessary for the evolution of the ortho and benzylic
anions to the dearomatized lithiated species. Otherwise, proto-
nation of these two types of anions will regenerate the starting
phosphinamide.23 In sharp contrast, aldehydes react very rapidly
and selectively with the dearomatized nucleophile, driving the
equilibrium among the anions to the dearomatized species,
which is consumed through reaction with the aldehyde.
To determine the anion distribution at a given metalation time,
we used D2O as quench reagent (Scheme 1). Assuming an
almost negligible isotopic effect of D2O versus H2O, deuteration
of the anions I to III will provide three different products: 2-d,
Figure 1. 2H{1H} NMR spectra (46.5 MHz) of the dearomatization-
deuteration of 1 as a function of the metalation time t1. Spectra measured
from the crude product in CDCl3 at room temperature.
Table 1. Distribution of Products (%) in the
Dearomatization-Deuteration of 1 at Increasing Time (t1) Based
on 2H NMR Measurementsa
t
1 (s-BuLi)
2-d
3-d
4-d
1 min
36
37
38
35
33
15
12
10
8
49
51
52
57
60
30 min
60 min
4 h
12 h
7
a The concentration of 1 in the 1H NMR spectra was below the detection
level.
the anions.17 The reactions were quenched with D2O/THF at
-90 °C for 30 min, and then aqueous workup gave the crude
mixtures that were analyzed through their 1H, 2H, and 31P NMR
spectra. As a result of the very small chemical shift difference
1
between 2-d and 3-d, the H and 31P NMR spectra afforded
only the ratio of dearomatized versus open-chain products
2
obtained. In contrast, H NMR spectra allowed us to readily
identify all deuterated compounds because each deuterium
absorbed in a well-separated region of the spectrum. Figure 1
2
shows the measured H NMR spectra, and Table 1 gives the
distribution of products as a function of t1 deduced from the
integrals of the respective product signals in the spectra. Previous
studies10a,13 have shown that quenching the dearomatized anions
with H2O and MeOH affords mixtures of products arising from
the protonation at the alpha (cis/trans isomers at the bridging
carbon atoms), gamma, and epsilon positions with respect to
the phosphorus atom. To prevent confusion, we grouped all
deuterated dearomatized compounds under the generic label 4-d.
The deuterium signal corresponding to the epsilon derivative
is clearly seen in all spectra at δ ) 2.2 ppm (relative yield of
2
3-d, and 4-d (in all cases E ) H without the equilibration
observed when the electrophile was an aldehyde). Thus, a series
of experiments was performed in which phosphinamide 1 was
lithiated as stated above for a time t1. No cosolvent was used
because in the presence of DMPU or HMPA the anionic
cyclization is too rapid to establish any reasonable trend between
(23) The magnitude of t2 ) 30 min represents a value optimum for most of the
protonating reagents used that takes into account solubility problems at
the temperature of -90 °C. For simple alcohols, this time can be reduced
to 1 min, as deduced from the decoloration of the reaction mixture.
2
3%). In the spectrum measured at t1 ) 12 h, the H signals of
the products having a deuterium atom at the alpha and gamma
9
J. AM. CHEM. SOC. VOL. 126, NO. 39, 2004 12553

A R T I C L E S
Scheme 2 a
Ferna´ndez et al.
1 afforded a mixture of 7-d2/8-d2/9-d2 in a ratio of 1:1:2. The
product distribution was established based on the integrals of
the signals of H-3a, H-7, and H-10 in the 1H NMR spectrum of
the crude mixture (Figure 2). Compound 9-d2 is derived from
the attack of the benzylic anion on the nondeuterated P-phenyl
ring containing two equivalent ortho positions. Consequently,
twice as much 9-d2 can be formed with respect to the other
isomers. The statistical distribution of isomers obtained indicates
that, as expected, there is no isotopic effect in the anionic
cyclization step.
a (1) Ph2P(O)Cl, Et3N (2.5 equiv), toluene, -78 °C. (2) NaH (2.2 equiv),
THF. (3) MeI (1.1 equiv), rt.
positions with respect to the phosphorus atom appear at δ )
2.97 and 2.85 ppm, respectively.
The assignments deduced from the proton spectrum were
confirmed through the conventional 13C and attached proton test
(APT) NMR spectra of the crude product, the latter being
optimized for the quantitative determination of quaternary
carbons.28 The integral measured for the signals of C-3, C-10,
C-3a, and C-7 gave a ratio of 4:2:1:1. The magnitude of the
coupling constants nJCX (X ) 2H, 31P, n ) 1, 2) measured from
these spectra for the labeled carbons are summarized in Table
2. The absence of 6-d suggests that the kinetic isotope effect
inhibits the direct lithiation of 1-d2 at the benzylic position. On
the other hand, the statistical formation of 7-d2, 8-d2, and 9-d2
reveals that anion translocation I f II (Scheme 1) is indeed
possible and that the coordinating agents HMPA or DMPU
catalyze the transformation by lowering the energy barrier of
the process.
From the results in Table 1 we concluded that in the absence
of coordinating agent, 49% dearomatization takes place by t1
) 1 min. The yield of 4-d increases to 60% with increasing t1
at the expense of the 2H-benzylic derivative 3-d (i.e., the
benzylic anion precursor). Curiously, the concentration of the
ortho-deuterated compound 2-d (i.e., the ortho-lithiated inter-
mediate) remains practically constant (≈36%) within experi-
mental error. In other words, the translocation is apparently
inhibited in the absence of cosolvent. Therefore, the distribution
of deuterated products also indicates that the ratio of ortho:
benzylic lithiation of 1 achieved by s-BuLi is approximately
36:64.24 Consequently, the direct lithiation at the benzylic
position of 1 is favored by a factor of nearly 2 with respect to
the ortho lithiation. The preferred lithiation R to the nitrogen
atom of 1 is similar to the known preference for benzylic
lithiation of N-benzylbenzamides.25 In contrast, the lithiation-
alkylation of N-benzyl-N-methyl(diphenyl)thiophosphinamides
at low temperature in the presence of N,N,N′,N′-tetramethyl-
ethylenediamine (TMEDA) leads exclusively to products of
ortho lithiation.26
In the presence of cosolvent the dearomatization of 1 and
subsequent protonation with 2,6-di-tert-butyl-4-methylphenol
yields 4 (E ) H) almost quantitatively.13 This fact suggests that
either the translocation is accelerated by the action of the
coordinating cosolvent or the cosolvent exclusively favors
lithiation at the benzylic position. To check these hypotheses,
we applied the optimized dearomatization-protonation sequence
of reactions to the phosphinamide 1-d2, which is dideuterated
at the benzylic position. Phosphinamide 1-d2 was prepared in
84% overall yield in a three-step process involving the treatment
of the doubly labeled benzylic amine 5-d2 with diphenylphos-
phinoyl chloride in the presence of triethylamine (2.5 equiv) in
toluene at -78 °C, metalation of the resulting phosphinamide
with NaH in THF at room temperature, and addition of MeI
(Scheme 2).27
NMR Studies. We next attempted the structural characteriza-
tion of anions I, II, and III by monitoring their formation
reactions with NMR spectroscopy. Most of the NMR studies
were carried out on a spectrometer operating at a proton
frequency of 300 MHz. The information collected was comple-
mented by data measured on a spectrometer of higher magnetic
1
field (11.74 T, H frequency of 500 MHz) recently purchased
by the University of Almer´ıa (see Supporting Information),
which allowed us to perform triple-resonance experiments
involving the nuclei 1H, 7Li, and 31P. For sake of clarity, these
experiments will be discussed separately. The NMR samples
of approximately 0.2 M in THF-d8 were prepared under
conditions analogous to those used in the bulk (see Experimental
Section).29 The only difference was that partial evaporation of
the solvent from the s-BuLi hexane solution under reduced
pressure at -20 °C was used to reduce the intensity of the un-
wanted hexane signals. The resulting s-BuLi was then dissolved
in THF-d8 (0.25 mL) and transferred to an NMR tube containing
a solution of phosphinamide 1 in THF-d8 (0.25 mL) at -90
°C. The effect of the cosolvent was studied on samples to which
6 equiv of dry HMPA was added prior to the metalation step.
We first tackled the NMR structural study in the absence of
coordinating agent as a result of the lower rate of the
transformation of 1 into III when HMPA or DMPU is not
present. The sample was transferred to the spectrometer at -90
If the lithiation of 1-d2 took place exclusively at the benzylic
position, then only the tetrahydrobenzazaphosphol 6-d mono-
deuterated at the carbon linked to the phenyl substituent would
be formed (Scheme 3). In contrast, the dearomatized products
derived from an anionic translocation (i.e., 7-d2, 8-d2, and 9-d2)
would exhibit a scramble of the deuterium label between all
possible ortho positions with respect to the phosphorus atom
(Scheme 3). The dearomatization-protonation of 1-d2 under the
same reaction conditions used for the unlabeled phosphinamide
1
1
°C, and then a series of H, 31P, and 13C, H DEPT135 NMR
spectra were sequentially measured at increasing metalation time
(t1).30 The spectra obtained are shown in Figures 3-5.
The proton spectrum measured in the first minute of reaction
(Figure 3, t1 ) 1 min) exhibits signals corresponding to the
(24) The ratio of 30:70 has been estimated for the ortho:benzylic deprotonation
of N-benzylbenzamides. See ref 18a.
(28) One-dimensional APT 13C NMR optimized for quaternary carbons:
repetition delay of 2 s, spin-echo time of 1/(2JCH) (JCH ) 145 Hz) 3.03
ms. Patt, S. L.; Shoolery, J. N. J. Magn. Reson. 1982, 46, 535.
(29) About five samples were examined in each spectrometer. They gave
consistent and reproducible results. Among the variables analyzed, we
evaluated the effect of cosolvent, concentration of base, temperature, and
time of metalation.
(25) Fraser, R. R.; Boussard, G.; Potescu, I. D.; Whiting, J. J.; Wigfield, Y. Y.
Can. J. Chem. 1973, 51, 1109.
(26) Yoshifuji, M.; Ishizuka, T.; Choi, Y. J.; Inamoto, N. Tetrahedron Lett. 1984,
25, 553
(27) The quantitative formation of the intermediate N-benzylphosphinamide was
monitored by 31P NMR spectroscopy of aliquots of the reaction.
9
12554 J. AM. CHEM. SOC. VOL. 126, NO. 39, 2004

N-Alkyl-N-benzyldiphenylphosphinamides
A R T I C L E S
Scheme 3 a
a (1) s-BuLi (2.5 equiv), DMPU (6 equiv), THF, -90 °C, 30 min. (2) 2,6-Di-tert-butyl-4-methylphenol, THF, -90 °C, 30 min. (3) H2O.
Figure 2. 1H NMR spectrum (300.13 MHz) of the crude mixture of 7-d2,
8-d2, and 9-d2 measured in CDCl3 at room temperature.
n
Table 2. JCX (Hz) for the Deuterated Carbons of 7-d2, 8-d2, and
9-d2
compd 1JCD
2JPC
1JCD
2JPC
1JCD
2JPC
1JCD
2JPC
C-3
C-3
C-3a
C
-3a
C-7
C-7
C-10
C-10
7-d2
8-d2
9-d2
10.6
10.6
10.6
32.4
32.4
32.4
18.8
14.0
13.1
9.6
10.6
36.2
3
starting phosphinamide 1 [δ 2.46 (3H, JPH ) 11.3 Hz), 4.11
3
(2H, JPH ) 6.4 Hz), 8.18 (4H)]. At t1 ) 13 min, all previous
signals are broadened and new signals appear in the aromatic
(Hortho of P-phenyl rings, δ 7.96 and 8.34 ppm) and, most
importantly, olefinic (δ 5.73 and 6.16 ppm) regions, which
unequivocally arise from a dearomatized species. An additional
signal at δ 4.25 ppm is overlapped with that of the methylene
protons of 1 (δ 4.11 ppm). The intensities of the new signals
increase with increasing t1 at the expense of those of 1.
Therefore, the dearomatized species can be safely assigned as
III, the precursor of the product isolated in the dearomatization-
protonation of 1.31 When almost 1 h of reaction has elapsed
(Figure 3, t1 ) 54 min), 1 is practically consumed. Two new
NMe doublets appear at δ 2.23 (3JPH ) 8.7 Hz) and 2.54 (3JPH
Figure 3. 1H NMR spectrum (300.13 MHz) of the lithiation of 1 with
s-BuLi as a function of t1. General conditions: THF-d8, -90 °C, 16 scans
accumulated; total measuring time per spectrum of 1 min.
) 11.3 Hz) ppm in the ratio of 1.4:1, and the olefinic region
shows two new multiplets (δ 5.69 and 6.08 ppm) barely
discernible at the base of the more intense one (approximate
intensity ratio of 1:0.1). We assigned the latter set of new signals
to a second dearomatized intermediate IV. On the other hand,
the methylene group at δ 4.11 ppm evolves into three partially
overlapped signals: multiplets at δ 4.08 and 4.23 ppm and a
broad singlet at δ 4.17 ppm. An additional increase of the
reaction time produced only minor changes (Figure 3, t1 ) 91
min), the most interesting being the observation of a third NMe
doublet of low intensity at δ 1.92 ppm (3JPH ) 8.3 Hz) and a
broad signal at δ 3.9 ppm. Significantly, the integral ratio of
the NMe signals at δ 2.23 and 1.92 ppm is the same as that
found for the two sets of olefinic multiplets (1:0.1). To
(30) Although the quaternary carbons are missing from the DEPT135 spectra,
the repetition rate of the experiment is controlled by the relaxation time of
the protons, which is shorter than those of the carbons. Therefore, a
reasonable signal-to-noise ratio can be obtained from the DEPT in a shorter
time than from a standard 13C NMR spectrum. This is an important factor
to obtain information about the early stages of the reaction, where the
intervening species are being transformed throughout the measurement of
the spectra.
(31) We assigned to species III the same relative configuration found for 4, the
product isolated in the dearomatization-protonation reaction, on the
assumption that in the protonation the integrity of the stereogenic centers
will not be affected. This assignment as well as that of species IV will be
confirmed in the section discussing the sample prepared in the presence of
HMPA.
1
summarize, the H NMR study shows that the lithiation of 1
leads to the formation of two dearomatized intermediates III
9
J. AM. CHEM. SOC. VOL. 126, NO. 39, 2004 12555

A R T I C L E S
Ferna´ndez et al.
Figure 5. DEPT-135 NMR spectra (75.5 MHz) of the lithiation of 1 with
s-BuLi as a function of t1. General conditions: THF-d8, -90 °C, 280-600
scans accumulated; total measuring time per spectrum of 3-31 min.
broad signal V begins to emerge at δ 45.38 ppm, which
apparently splits into two sharper signals of unequal intensity
at t1 ) 74 min (labeled as species Va/Vb). Finally, the steady
state is reached at t1 ) 92 min. Similar to the proton spectrum,
only species I, III, and IV remain in solution in a ratio of 1.2:
1:0.1. Their respective 31P chemical shifts are δ 32.85, 36.03,
and 41.48 ppm. The assignment of the 31P signals was readily
achieved through the H,31P gHMQC spectrum optimized for
1
Figure 4. 31P{1H} NMR spectra (121.49 MHz) of the lithiation of 1 with
s-BuLi as a function of t1. General conditions: THF-d8, -90 °C, 40-400
scans accumulated; total measuring time per spectrum of 1-11 min.
the observation of phosphorus-proton long-range couplings
(Figure S1, Supporting Information). The phosphorus of species
III and IV correlated with olefinic protons, while the signal of
I only showed correlations with aromatic and aliphatic protons.
and IV in an approximate ratio of 1:0.1 plus a third species
characterized by an NMe group at δ 2.54 and a broad singlet at
δ 4.17 ppm.
The DEPT-135 spectra helped to clarify the type of inter-
mediates formed during the process (Figure 5). As expected,
the first spectrum acquired (t1 ) 3 min) was dominated by the
signals of 1 [δC 32.76 (CH3), 52.25 (CH2)]. Nevertheless, an
incipient signal of positive phase is observed at δ 49.8 ppm.
This chemical shift is appropriate for a benzylic CH, and the
only source available for this type of carbon is the benzylic
anion II. The concentration of II increases slightly over the
next 13 min and reaches a maximum at t1 ) 77 min. This last
spectrum, measured just after the 31P spectrum at t1 ) 66 min,
afforded additional key information. Let us focus on the
relatively intense signal of negative phase at δ 52.25 ppm.
Considering that at this time the starting phosphinamide was
practically consumed, this signal cannot be derived from the
methylene carbon of 1; it must arise from the analogous
methylene carbon of an ortho-lithiated species such as inter-
mediate I.33 This conclusion implies that the carbons of the CH2
groups of 1 and I absorb at the same chemical shift under the
measuring conditions. The same arguments apply to the NMe
signals of 1a, I, and II, all of which coincide at δ 32.75 ppm.
The remaining signals of the spectrum are assigned to the highest
concentrated dearomatized species III. Again, the spectrum after
reaching the steady state (Figure 5, t1 ) 98 min) shows no
signals corresponding to compound II; only the signals for
anions I and III are observed. The concentration of the second
The 31P NMR spectra proved to be more informative in regard
to the distribution of species in the reaction mixture owing to
the higher dispersion of signals (Figure 4). In the first spectrum
acquired (t1 ) 2 min), besides the signal of the starting
phosphinamide 1 at δ 29.28 ppm, two new broad signals
corresponding to I (δ 33.15 ppm) and III (δ 36 ppm) can be
identified above the noise upon increasing the vertical scale.
These two signals became clearly visible in the spectrum
measured 12 min later. Moreover, the asymmetry of the signal
labeled as I suggests that it, in fact, corresponds to two partially
overlapped broad signals (δ 33.12 and 33.15 ppm). Additionally,
a new broad signal II appears at δ 23.13 ppm. The intensity of
the new species continues to grow with the consequent
decreasing of 1 as the reaction progresses, except that of II,
which decreases slightly. At t1 ) 25 min a new signal appears
at δ 41.44 ppm corresponding to IV. Over the next 30 min a
reequilibration between the species already present takes place
(Figure 4, t1 ) 55 min): the phosphinamide 1 has practically
disappeared, and the two overlapped signals assigned to I are
now partially resolved due to the reduction of the intensity of
the more deshielded one. Their actual chemical shifts are δ 33.1
and 33.39 ppm.32 For t1 ) 66 min they collapse again into a
broad and intense signal at δ 33.1 ppm. Interestingly, a very
(32) The slight variations of the chemical shifts are a consequence of the changes
in the concentration of the different species in the reaction mixture with
increasing t1. The 31P nucleus is very sensitive to these changes. Possible
fluctuations of the temperature are considered negligible. Tebby, J. C. In
Phosphorus-31 NMR Spectroscopy in Stereochemical Analysis. Organic
Compounds and Metal Complexes; Verkade, J. G., Quin, L. D., Eds.;
VCH: Weinheim, 1987; Chapter 1, pp 1-60.
(33) The structural elucidation and synthetic applications of ortho anions such
as I is currently under study. To exclude complications derived from
possible benzylic deprotonation we are using N,N-diisopropyldiphenylphos-
phinamide as the starting material. As suggested by a reviewer, the lithiation
of 1-d2 would be very informative also. We thank the referee for this
suggestion. These results will be reported in due course.
9
12556 J. AM. CHEM. SOC. VOL. 126, NO. 39, 2004

N-Alkyl-N-benzyldiphenylphosphinamides
A R T I C L E S
dearomatized species IV was too low to be detected, even after
the accumulation of 600 scans.
Variation of the concentrations of species I and II during
the NMR study (Figure 4) and correlations observed in the
1H,31P gHMQC spectrum indicate that the benzylic anion II
gives rise to only one signal at δP 23.13 ppm, whereas two ortho-
lithiated species I are initially formed, giving signals at δP 33.1
and 33.39 ppm that evolve into a single signal at δP 32.85. More
importantly, this study clearly shows that the benzylic and ortho
anions are formed almost simultaneously, with an apparent
preference for I (cf. 31P NMR spectra at t1 ) 2, 14, 25, and 55
min). The benzylic anion II evolves through anionic cyclization
to give the dearomatized anions III and IV, whereas the ortho-
lithiated compound I remains in solution. Once all species
equilibrated, integration of the 31P NMR spectrum afforded a
ratio of dearomatization:ortho lithiation of 57:43, which is in
reasonable agreement with the ratio of 64:36 deduced from the
deuteration reactions.34 However, the relative concentration of
I in the time period t1 )14-74 min is higher than that of the
remaining anions. Given that the translocation from I to II in
the absence of HMPA has been excluded through the labeling
studies, we interpret the comparatively high intensity of I at
intermediate t1 values as the result of the formation of mixed
aggregates of undefined structure between I and other anions
present in solution. This hypothesis is supported by the
broadening observed for the 31P signal of I in the first hour of
reaction. We tentatively assign a dimeric structure to I based
on the recent characterization of ortho-lithiated N,N-diisopro-
pylbenzamide as a dimer.35 As the reaction progresses, the
thermodynamically more stable species prevail in solution.
Figure 6. 31P NMR spectrum (121.49 MHz) of the lithiation of 1 with
s-BuLi measured at t1 > 2 h in THF-d8 at -70 °C. The FID was multiplied
by a Gaussian of LB ) -8 and GB ) 0.04 prior to the Fourier
transformation.
spectrum, the narrowing of the signals revealed the existence
of phosphorus-lithium-7 couplings. At -70 °C, the signals at
2
31
7
δ 36.03 and 41.48 ppm split into quartets of J( P Li) ) 6.4 and
7.4 Hz, respectively. This means that each phosphorus nucleus
is coupled to only one lithium nucleus, and therefore, the
lithiated dearomatized compounds III and IV are both mono-
mers (Figure 6). The detection of 2JPLi couplings is a well-known
structural tool for the identification of the aggregation state of
lithiated phosphoramides36 and phosphonamides.37 Reich et al.
developed a very efficient methodology for studying the ion-
pairing state of organolithium compounds through HMPA
titration. The degree of HMPA coordination to lithium is
2
deduced from the measured JPLi couplings.38 Our results
indicate that the PdO group of a phosphinamide also coordi-
2
31
7
nates to a lithium atom with a magnitude of J( P Li) similar to
2
31P7Li)
those found in phosphoramides. Unfortunately, no J(
coupling could be resolved for the phosphorus signal of the ortho
anion I.39
Prior to undertaking spectral assignment of compounds I, III,
and IV, we increased the temperature of the sample from -90
to -70 °C in steps of 10 °C and measured the spectra at the
new temperatures. This temperature modification was imple-
mented for two reasons: (1) narrower signals may be expected
due to the reduction of the viscosity of the sample, and (2) the
equilibrium state between the species present in solution could
be altered. To our delight, both effects were observed, thereby
providing important structural and chemical information. Thus,
increasing the temperature from -90 °C by 10 °C produced
better separation of the olefinic signals of III and IV in the
proton spectrum and an overall improvement of the resolution
of all signals. Moreover, a small but measurable increase of
the relative concentration of IV with the consequent reduction
of that of III was noted. These trends became even more
noticeable when the temperature was further increased to -70
°C. The new ratio of compound III:IV obtained by integration
of the NMe signals is 1:0.17 (cf. 1:0.1 at -90 °C). Additionally,
a doublet is clearly identified at δ 4.18 ppm (3JPH ) 5.7 Hz),
corresponding to the methylene protons of I. The existence of
the phosphorus-proton coupling was confirmed through the 1H-
{31P} spectrum.
Raising the temperature to -70 °C had no significant effect
on the 13C NMR spectrum (i.e., the concentration of IV was
still too low to be detected within a reasonable measuring time).
The complete structural assignment of anions I and III was
1
carried out at -70 °C based on the analysis of the H,13C
gHMQC, and gHMBC spectra (Figures S2 and S3, respectively,
Supporting Information). The gHMQC allowed us to identify
the most relevant C-H pairs of nuclei of both compounds. For
I these are limited to the CH3 [δH 2.56 ppm (3JPH ) 11.3 Hz);
δC 32.75 ppm] and CH2 [δH 4.18 (3JPH ) 5.7 Hz); δC 54.26
ppm] of the amine moiety and the Hortho of the P-phenyl rings
[δH 8.34 ppm; δ 133.53 ppm (2JPC ) 9.4 Hz)] (Figure S2,
Supporting Information).
A selection of NMR data of III is given in Table 3. The
dearomatized system is readily assigned through the correlations
(36) (a) Barr, D.; Clegg, W.; Mulvey, R. E.; Snaith, R. J. Chem. Soc., Chem.
Commun. 1984, 79. (b) Barr, D.; Doyle, M. J.; Mulvey, R. E.; Raithby, P.
R.; Berd, D.; Snaith, R.; Wright, D. S. J. Chem. Soc., Chem. Commun.
1989, 318. (c) Reich, H. J.; Green, D. P.; Phillips, N. H. J. Am. Chem.
Soc. 1989, 111, 3444. (d) Reich, H. J.; Green, D. P. J. Am. Chem. Soc.
1989, 111, 8729.
(37) Denmark, S. E.; Miller, P. C.; Wilson, S. R. J. Am. Chem. Soc. 1991, 113,
1468.
(38) (a) Reich, H. J.; Borst, J. P.; Dykstra, R. R.; Green, D. P. J. Am. Chem.
Soc. 1993, 115, 8728. (b) Reich, H. J.; Holladay, J. E.; Mason, J. D.;
Sikorski, W. H. J. Am. Chem. Soc. 1995, 117, 12137. (c) Reich, H. J.;
Gudmundsson, B. O. J. Am. Chem. Soc. 1996, 118, 6074. (d) Reich, H. J.;
Green, D. P.; Medina, M. A.; Goldenberg, W. S.; Gudmundsson, B. O.;
Dykstra, R. R.; Phillips, N. H. J. Am. Chem. Soc. 1998, 120, 7201. (e)
Reich, H. J.; Holladay, J. E.; Walker, T. G.; Thompson, J. L. J. Am. Chem.
Soc. 1999, 121, 9769. (f) Sikorski, W. H.; Reich, H. J. J. Am. Chem. Soc.
2001, 123, 6527.
The changes in the 31P NMR spectrum with the increase of
temperature revealed even more information. Besides the
reduction of the ratio III:IV already deduced from the 1H
(34) Some possible reasons for the difference observed are the higher concentra-
tion of the NMR sample with respect to the standard reaction and most
probably the imprecision of the integration of broad signals due to the
reduction produced by short transverse relaxation times of different nuclei.
(35) Clayden, J.; Davies, R. P.; Hendy, M. A.; Snaith, R.; Wheatley, A. E. H.
Angew. Chem., Int. Ed. 2001, 40, 1238.
(39) One may reasonably assume that in this anionic species the PdO linkage
will coordinate to the lithium cation, leading to a possible scalar coupling
7
via 2J(31 Li). However, if this coupling exists it must be smaller than the
P
resolution achieved in the experiment.
9
J. AM. CHEM. SOC. VOL. 126, NO. 39, 2004 12557

A R T I C L E S
Ferna´ndez et al.
Table 3. Selection of NMR Data of III Measured in THF-d8 at -70
°Ca
C/H
δH
nJPH
δC
nJPC
3
3a
4
5
6
7
7a
9
10
13
4.05
4.05
4.12
5.69
4.24
6.14
75.89
56.19
98.28
129.90
93.66
133.85
67.32
143.40
134.82
145.75
17.9
17.4
5.2
b
4.6
s
7.7
8.4
15.1
16.0
143.9
133.8
7.5
7.94
b
10.4
a δ in ppm, J in Hz. b Not measured due to partial signal overlap.
of the olefinic methyne groups. As expected, the sp3 carbon
C3a absorbs at the highest field. The delocalization of the
negative charge through carbons C4 and C6 produced a large
upfield shift [∆δ(III-4) ) -24.5 ppm for C4] (4, R1 ) Me, E
) H, Scheme 1), whereas carbons C5 and C7 are much less
affected [∆δ(III-4) ) 4.5 ppm for C5 and -1.7 ppm for C7].
The quaternary carbon C7a was identified based on the
correlations observed in the gHMBC spectrum with H3a (Figure
S3, Supporting Information). For this correlation four cross
peaks are obtained due to the passive coupling to 31P either
during the evolution (1JPC) or acquisition time (2JPH). Moreover,
the negative slopes of the cross peaks indicate that both
couplings are of opposite relative sign. Considering that in P(V)
Figure 7. NMR spectra of the earlier stages in the lithiation of 1 with
s-BuLi at -110 °C in THF-d8. (a) 31P{1H} (202.46 MHz), including
expansions of III and VI; (b) 7Li{1H} NMR (194.37 MHz), including
expansions of V; and (c) 7Li{31P,1H}. In all cases Gaussian multiplication
of the FID (LB ) -2, GB ) 0.14 for 31P and LB) -4, GB) 0.24 for 7Li)
was performed prior to the Fourier transformation.
II, the monomeric dearomatized compounds III and IV, and
the unknown species V. No translocation from I to II was
observed, in agreement with the labeling studies. The anions I
and III were fully characterized. In the first case, only one
isomer persisted when the steady state was reached at -70 °C.
Similar to III, compound IV also forms a monomeric contact
ion pair. However, the low concentration of IV, and in the case
of II its short lifetime, impeded complete structural character-
ization. No information about the aggregation state of benzylic
anion II and the species labeled as V could be obtained.
compounds JPC is always positive,40 it follows that JPH is
negative. On the other hand, the correlation of H3 with C4
facilitates the unequivocal assignment of C4 and, by default,
C6. Regarding the stereochemistry of III, we tentatively
assumed that the stereochemistry of the product of the de-
aromatization-protonation sequence 4 reflects the stereochem-
istry of the major component of the mixture (i.e., III). Although
in light of the Curtin-Hammett principle this is not necessarily
true, this assumption was later confirmed (see below). Signifi-
1
2
cantly, carbon C6 shows the widest 13C NMR signal (W1/2
)
7Li,31P Shift Correlation. Triple-resonance experiments were
carried out on a NMR spectrometer working at a proton
frequency of 500 MHz. A fresh 0.2 M sample in THF-d8 was
46 Hz); this is the position generally attacked by most of the
electrophiles used to trap the dearomatized species of 1. The
increased signal width may be explained by the proximity of
the quadrupolar lithium-7 nucleus, which causes an increase in
the transverse relaxation rate41 This feature suggests that the
lithium cation is most probably η3-coordinated to the C6-C7-
C7a moiety of the dearomatized six-membered ring anion. The
dissymmetry of the coordination would be forced by the
coordination of the PdO group to the lithium atom, as deduced
from the phosphorus-lithium-7 coupling constant measured.
Also noteworthy is the large increment of 1JPC experienced by
7
prepared, and the 31P and Li NMR spectra were measured at
-90 °C. Due to the larger spread of frequencies available,
species that at 300 MHz are in the fast-exchange regime on the
NMR time scale (giving rise to an average signal) might now
be observed as individual components at the higher field of 500
MHz. This proved to be the case as can be seen in the 31P NMR
spectrum shown in Figure 7a (cf. Figures 4 and S4, Supporting
Information).32 The excess of s-BuLi is clearly established in
1
the H NMR spectrum by the signal of the lithiated methine
1
C7a [∆1JPC(III-4) ) 23.2 Hz]. Such effects on JPC upon
group at δ -0.89 ppm (Figure S5, Supporting Information).
The signals assigned to the ortho anion I (probably mixed
aggregates and/or stereoisomeric dimers) are very broad and of
metalation have been previously noted in other phosphorus-
stabilized anions.42
To summarize the NMR results discussed above, the study
of the lithiation of 1 in the absence of coordinating cosolvent
showed the presence of at least five intermediate species,
I-V: the ortho anion I (presumably a dimer), the benzylic anion
(42) (a) Bottin-Strzalko, T.; Seyden-Penne, J.; Simonnin, M.-P. J. Chem. Soc.,
Chem. Commun. 1976, 905. (b) Bottin-Strzalko, T.; Seyden-Penne, J.;
Froment, F.; Corset, J.; Simonnin, M.-P. J. Chem. Soc., Perkin Trans. 2
1987, 783. (c) Denmark, S. E.; Dorow, R. L. J. Am. Chem. Soc. 1990,
112, 864. (d) Denmark, S. E.; Swiss, K. A.; Wilson, S. R. J. Am. Chem.
Soc. 1993, 115, 3826. (e) Lo´pez-Ortiz, F.; Pela´ez-Arango, E. J.; Tejerina,
B.; Pe´rez-Carren˜o, E.; Garc´ıa-Granda, S. J. Am. Chem. Soc. 1995, 117,
9972. (f) Davies, J. E.; Davies, R. P.; Dunbar, L.; Raithby, P. R.; Russell,
M. G.; Snaith, R.; Warren, S.; Wheatley, A. E. H. Angew. Chem., Int. Ed.
Engl. 1997, 36, 2334. (g) Hitchcock, P. B.; Lappert, M. F.; Uiterweerd, P.
G. H.; Wang, Z. X. J. Chem. Soc., Dalton Trans. 1999, 3413. (h) Ferna´ndez,
I.; AÄ lvarez-Gutie´rrez, J. M.; Kocher, N.; Leusser, D.; Stalke, D.; Gonza´lez,
J.; Lo´pez-Ortiz, F. J. Am. Chem. Soc. 2002, 124, 15184.
(40) (a) Kalinowski, H. O.; Berger, S.; Braun, S. 13C NMR Spektroskopie; George
Thieme: Stuttgart, 1984; p 420. (b) Lo´pez-Ortiz, F.; Pela´ez-Arango, E.;
Palacios, F.; Barluenga, J.; Garc´ıa-Granda, S.; Tejerina, B.; Garc´ıa-
Ferna´ndez, A. J. Org. Chem. 1994, 59, 1984.
(41) (a) Bauer, W.; Schleyer, P. V. R. AdV. Carbanion Chem. 1992, 1, 89. (b)
Gu¨nther, H. In AdVanced Applications of NMR to Organometallic
Chemistry; Gielen, M., Willem, R., Wrackmeyer, B., Eds.; John Wiley:
New York, 1996; Chapter 9, pp 247-290.
9
12558 J. AM. CHEM. SOC. VOL. 126, NO. 39, 2004

N-Alkyl-N-benzyldiphenylphosphinamides
A R T I C L E S
relatively low intensity, whereas the concentration of the two
species Va/Vb is rather high. The signal intensities and line
widths of Va and Vb are different. In addition, two new species,
VI and VII, characterized by 1:1:1:1 quartets at δ 34.9 [2J(
31P7Li)
2
31
7
) 4.4 Hz] and 32.7 ppm [ J(
) 7.7 Hz], respectively, are
P Li)
observed (Figure 7a). This multiplicity, together with the new
signals appearing in the olefinic region of the 1H NMR spectrum
(Figure S5, Supporting Information), suggests that VI and VII
are two new monomeric dearomatized anion stereoisomers of
III and IV. Considering that in the dearomatization of 1 three
new stereogenic centers are formed, the dearomatized anions
III, IV, VI, and VII represent all possible diastereomers that
could be obtained in the reaction. Compounds VI and VII are
relatively unstable, and with time they evolve into the more
stable isomers III/IV. It is noteworthy that in the dearomati-
zation-electrophilic trapping of 1a small amounts of products
derived from anions other than III and IV have been occasion-
ally isolated.17 The time and temperature evolution of the sample
(Figure S4, Supporting Information) shows the same features
observed in the sample previously measured at lower frequency
(Figure 4). Thus, at -70 °C only anions I, III, and IV are
present in solution.
Figure 8. 2D 7Li,31P{1H} HMQC NMR (194.37 MHz) of the earlier stages
in the lithiation of 1 with s-BuLi at -110 °C in THF-d8. The inset represents
the correlation detected for the species Va/Vb. Experiment started at a
metalation time of 2 min. Total measuring time: 2 h 40 min.
Table 4. 31P (202.46 MHz) and 7Li (194.37 MHz) NMR Data of
Anions I-VII Measured in THF-d8 at -110 °Ca
7
The Li NMR study was conducted on a freshly prepared
data
Ib
II
III
IV
Va
Vb
VII
0.2 M sample. For simplicity, in this case an equimolecular ratio
of base and substrate was used. In the earliest stages of the
deprotonation reaction, a significant amount of free phos-
phinamide 1 (δ 28.22 ppm) is still visible in the 31P NMR
δ (31P) 31.92, 32.22 21.21 34.97 40.32 44.12
δ (7Li) 1.07
0.01 -0.13 1.92 2.36
2JPLi bsc
4.5 6.9 6.6 4.0, 5.5
43.16 32.66
2.43 -0.04
5.0
7.8
7
a δ in ppm, J in Hz. b Mixed aggregate and/or stereoisomeric dimers
spectrum (Figure S6, Supporting Information). The Li NMR
(see text). c Broad singlet.
spectrum measured at -90 °C exhibits up to seven signals: three
doublets, two poorly resolved triplets, and two very broad
singlets. Lowering the temperature to -110 °C allowed us to
takes place exclusively at the benzylic position, and the resulting
anion is stabilized by forming dimers containing either (LiC)2
or (LiO)2 four-membered rings.43 Not unexpectedly, the broad
7Li signal at δ 1.07 ppm is the only one that does not show any
31P correlation. Its large signal width implies rapid transverse
relaxation, and thus, the transverse magnetization generated at
the beginning of the pulse sequence is completely lost during
the delays involved in the experiment. It can be assumed that
this lithium signal corresponds to the ortho anions I exhibiting
broad singlets in the 31P NMR spectrum. At -70 °C, the
temperature at which the ortho anion produces a sharp 31P
singlet, all lithium signals collapse to an uninformative broad
singlet. Further work is in progress to elucidate the structure of
diphenylphosphinamides exclusively lithiated at the ortho posi-
tion.33
resolve all multiplets except a broad signal at 1.07 ppm (Figure
2
31P7Li)
7b). The doublets at δ -0.13 ppm, J(
) 6.9 Hz; -0.04
2
2
31
7
31 7
ppm, J( P Li) ) 7.8 Hz; 0.01, J( P Li) ) 4.5 Hz; and 1.92 ppm,
J( P Li) ) 6.6 Hz are clearly observed. The multiplets at δ 2.36
2
31
7
and 2.43 ppm were identified, respectively, as a double doublet
2
2
2
31
7
31
7
31 7
[ J( P Li) ) 4.0, J( P Li) ) 5.5 Hz] and a triplet [ J( P Li) ) 5.0
Hz] from the Li{31P, H} spectrum (Figure 7c).
7
1
The connectivity between the Li,31P spin pairs was readily
7
obtained through the Li,31P{1H} HMQC spectrum43 recorded
at -110 °C (Figure 8).44 The assignments made are given in
7
Table 4. The correlations arising from the dearomatized mono-
7
meric species indicate that the pairs of 31P and Li signals
corresponding to anions III and VI are completely overlapped.
In contrast, those derived from IV and VII are well separated.
Most importantly, the 31P signal of the benzylic anion (δ 21.21
ppm) correlates with the lithium doublet at δ 0.01 ppm, a
multiplicity consistent with a monomeric structure. Monomeric
lithium benzylic anions adjacent to a nitrogen atom, which are
dipole-stabilized by a carbonyl moiety, have been characterized
both in the solid state45 and solution.46 We recently demonstrated
that lithiation of phosphinamide 1 with s-BuLi in diethyl ether
7
The correlations of Va (δ 43.16 ppm) with the Li triplet at
δ 2.43 ppm and Vb (δ 44.12 ppm) with the 7Li double doublet
at δ 2.36 ppm are especially interesting. The reaction pathway
for the transformation of 1 into compounds 2, 3, and 4 (Scheme
1) involves ortho, benzylic, and dearomatized anions that have
already been identified. On the other hand, the presence of mixed
aggregates resulting from coordination of the starting material
1 to any Li+1- species can be ruled out because the 31P NMR
spectrum should show two very distinct signals of equal
intensity, which is not the case. Consequently, any other
intervening anionic species should precede the deprotonation
reaction; in other words, the phosphorus signals Va and Vb
can be considered to arise from two pre-lithiation complexes
formed between the phosphinamide 1 and s-BuLi. Taking into
account that the organolithium base used has a stereogenic
center, the multiplicity of the lithium signals reveals that the
(43) Ferna´ndez, I.; Lo´pez-Ortiz, F. Chem. Commun. 2004, 1142.
(44) At this temperature the intensity of the signals corresponding to species I
decreased so much that they are only barely distinguishable above the noise
in the resolution-enhanced 31P NMR spectrum (Figure S6, Supporting
Information). Additionally, the signals of the dearomatized anions III and
VI are completely overlapped.
(45) Boche, G.; Marsch, M.; Harbach, J.; Harms, K.; Ledig, B.; Schubert, F.;
Lohrenz, J. C. W.; Ahlbrecht, H. Chem. Ber. 1993, 126, 1887.
(46) (a) Faibish, N. C.; Park, Y. S.; Lee, S.; Beak, P. J. Am. Chem. Soc. 1997,
119, 11561. (b) Gran˜a, P.; Paleo, M. R.; Sardina, F. J. J. Am. Chem. Soc.
2002, 124, 12511.
9
J. AM. CHEM. SOC. VOL. 126, NO. 39, 2004 12559

A R T I C L E S
Ferna´ndez et al.
Chart 1. Structure of Precomplexes Preceding the Deprotonation
Reactions of 1
Chart 2. Chiral Lithium Chelates
uration will be diastereotopic, and those of the diastereomer
VIIIb bearing unlike s-Bu groups will be enantiotopic. There-
7
fore, three different Li signals should be observed, which is
not the case. The preference for the formation of Va/Vb versus
isomers VIIIa/VIIIb may be assigned to steric interactions. The
vicinal substituents of the four-membered ring of Va/Vb are in
a trans configuration, and the ring may adopt a puckered
conformation that minimizes the through-space interaction
between substituents in relative positions 1 and 3. In contrast,
the bulky phosphorus-containing and s-Bu moieties of VIIIa/
VIIIb are in a cis configuration; thus, steric repulsions would
disfavor this type of dimer.50
rate of stereochemical inversion at the carbanion center is slow
on the NMR time scale.47 The observation of slow inversion at
the chiral center of s-BuLi is in accord with previous work
showing that s-BuLi exists in cyclopentane solution as a mixture
of diastereomeric dimers, tetramers, and hexamers that invert
slowly at the lithiated carbon.48 The splitting of the 7Li signals
of Va/Vb is consistent with a dimeric structure most probably
constructed around an (LiO)2 core in which the carbanionic and
phosphorus-bearing substituents occupy opposite faces of the
Li-O-Li-O four-membered ring (Chart 1).49 Thus, the 31P
Complex Vb is an anionic system with diastereotopic 31P
nuclei of two phosphorus(V) Lewis base ligands owing to a
slowly inverting stereogenic carbanion center. The only related
examples are the chelate organolithium compounds 10-12
reported by Reich et al. in which diastereotopic HMPA groups
have been detected by NMR (Chart 2).51 One mixed s-BuLi/
lithium N-isopropyl-O-methyl valinol aggregate consisting of
1 equiv of s-BuLi and 2 equiv of lithium amide has been
characterized in the solid state.52 Mixed aggregates of s-BuLi
with 3-methoxypropyllithium (tetramers)53 and LiBusMe2SiO
[xLiBus, (6-x)LiBusMe2SiO; (x ) 1-4)]54 have been identified
in solution by NMR spectroscopy. Evidence for the involvement
of precomplexes in the deprotonation of organic compounds
by organolithium bases directed by functional groups bearing
lone pairs (CIPE) is also limited.55 This is the first time that
pre-lithiation complexes have been structurally identified in the
directed deprotonation of a phosphorus-bearing substrate. As
the metalation progresses, complexes Va and Vb are consumed.
Analogous to the sample measured in the spectrometer of lower
magnetic field, when the temperature is raised to -70 °C a
steady state is reached in which the only anionic species
remaining in solution are I, III, and IV in a ratio of 1:0.77:
0.04.
7
and Li nuclei of Va, in which the two slowly inverting s-Bu
moieties are of like configuration, are homotopic, and accord-
7
ingly, a triplet is observed in the Li NMR spectrum. On the
other hand, under slow configurational inversion conditions, the
31P nuclei of dimer Vb with two s-Bu moieties of unlike
7
configuration are diastereotopic, whereas the Li nuclei are
enantiotopic. Consequently, the 7Li NMR spectrum should
consist of one signal that could exhibit two different 31P,7Li
couplings. Indeed, this is the case, as demonstrated by the double
doublet at δ 2.36 that collapses to a singlet upon 31P decoupling
(Figure 7c). Not unexpectedly, the 31P NMR spectrum shows
only a broad 31P signal for each dimer. At the temperature of
the measurement, two slowly exchanging 31P signals of very
similar chemical shift that are additionally split into 1:2:3:4:3:
2:1 septets due to coupling to two 7Li nuclei would give rise to
an unresolved broad signal. It is important to point out that the
The results of the NMR study allow us to unravel the details
of the mechanism of the anionic dearomatization of 1 in the
7
acquisition of the Li spectrum under simultaneous 31P decou-
pling proved to be crucial to identifying the multiplicity of the
7Li signal of Vb. Although the difference between the resonance
frequencies of the two nuclei is only 8 MHz, the experiment
can be routinely performed by using appropriate selective
frequency filters. Furthermore, this single datum allowed us to
distinguish Va and Vb from the corresponding stereoisomers
having a trans arrangement for the pair of carbanionic and
phosphinamide substituents VIIIa/VIIIb (Chart 1). The 31P and
7Li nuclei of complex VIIIa with s-Bu groups of like config-
(50) The presence of these types of precomplexes cannot be completely
discarded. The 31P NMR spectrum shows a few signals that could not be
assigned due to their very low intensity.
(51) Reich, H. J.; Kulicke, K. J. J. Am. Chem. Soc. 1996, 118, 273.
(52) Williard, P. G.; Sun, C. J. Am. Chem. Soc. 1997, 119, 11693.
(53) Schmitz, R. F.; de Kanter, F. J. J.; Schakel, M.; Scmitz, R. F.; Klumpp, G.
W. Tetrahedron 1994, 50, 5933.
(54) Zundel, T.; Zune, C.; Teyssie´, P.; Je´roˆme, R. Macromolecules 1998, 31,
4089.
(55) (a) Al-Aseer, M.; Beak, P.; Hay, D.; Kempf, D. J.; Mills, S.; Smith, S. G.
J. Am. Chem. Soc. 1983, 105, 2080. (b) Meyers, A. I.; Funetes, L. M.;
Reiker, W. F. J. Am. Chem. Soc. 1983, 105, 2082. (c) Meyers, A. I.;
Dickman, D. A. J. Am. Chem. Soc. 1987, 109, 1263. (d) Hay, D. R.; Song,
Z.; Smith, S. G.; Beak, P. J. Am. Chem. Soc. 1988, 110, 8145. (e) Bauer,
W.; Schleyer, P. v. R. J. Am. Chem. Soc. 1989, 111, 7191. (f) Resek, J. E.;
Beak, P. J. Am. Chem. Soc. 1994, 116, 405. (g) Luitjes, H.; de Kanter, F.
J. J.; Schakel, M.; Scmitz, R. F.; Klumpp, G. W. J. Am. Chem. Soc. 1995,
117, 4179. (h) Anderson, D. R.; Faibish, N. C.; Beak, P. J. Am. Chem.
Soc. 1999, 121, 7553. (i) Gross, K. M. B.; Beak, P. J. Am. Chem. Soc.
2001, 123, 315. (j) Arvidsson, P. I.; Hilmersson, G.; Davidsson, O. HelV.
Chim. Acta 2002, 85, 3814. (k) Bailey, W. F.; Beak, P.; Kerrick, S. T.;
Ma, S. H.; Wiberg, K. B. J. Am. Chem. Soc. 2002, 124, 1889. (l) Krow, G.
R.; Herzon, S. B.; Lin, G.; Qiu, F.; Sonnet, P. E. Org. Lett. 2002, 4, 3151.
(m) Gohier, F.; Castanet, A.-S.; Mortier, J. Org. Lett. 2003, 5, 1919.
(47) Basu, A.; Thayumanawan, S. Angew. Chem., Int. Ed. 2002, 41, 716.
(48) Fraenkel, G.; Henrichs, M.; Hewitt, M.; Su, B. M. J. Am. Chem. Soc. 1984,
106, 254.
(49) Alternative structures based on dimers of s-BuLi with a C2Li2 core in which
the lithium atom is additionally coordinated to phosphinamide 1 can be
reasonably discarded. The similar magnitude of the 31P,7Li couplings of
Va/Vb is consistent with the existence of static rather than fluxional
aggregates. Therefore, species containing one and two molecules of 1 should
7
be present to explain the observed Li multiplicity. For such a mixture of
aggregates a larger number of 31P signals than those experimentally observed
would be expected.
9
12560 J. AM. CHEM. SOC. VOL. 126, NO. 39, 2004

N-Alkyl-N-benzyldiphenylphosphinamides
A R T I C L E S
Scheme 4. General Mechanisms of the Anionic Dearomatization of 1
absence of coordinating cosolvents (Scheme 4). The first step
is the precomplexation of the base by the phosphinamide, which
leads to the formation of two diastereomeric dimers, Va and
Vb, in a ratio of 3:1. In these complexes, the proximity of the
carbanion center of s-BuLi to the benzylic protons of the
phosphinamide ligand promotes the deprotonation to give
monomer II. This new anion is relatively unstable and undergoes
intramolecular attack at the ortho position of each P-phenyl ring
to afford all four possible stereoisomers of the dearomatized
species III, IV, VI, and VII in a ratio of 1:0.07:0.13:0.25. These
anions are in equilibrium, and as the reaction progresses the
equilibrium shifts in favor of the thermodynamically more stable
species III and IV (in the ratio of 10:0.06). On the other hand,
ortho-lithiated anions I (presumably dimers) are present in the
reaction right from the first measurements. Although they coexist
with significant amounts of precomplexes Va and Vb, their
concentration is not time dependent. Consequently, complexes
Va and Vb do not participate in the ortho-deprotonation reaction.
Considering that the surface containing the Li-O-Li-O four-
membered ring and the phosphorus atoms of Va and Vb should
be rather flattened, the evolution of these precomplexes
exclusively toward anion II indicates that they adopt a preferred
conformation where the benzylic protons are proximate to the
carbanion center (Chart 1). Apparently this feature suggests that,
in sharp contrast to carboxamides,55c the ortho lithiation of 1 is
a kinetically driven process22 that does not involve a CIPE
mechanism. A more reasonable explanation is to assume that
the ortho anions I are formed through the intermediate precom-
plexes VIIIa and VIIIb. As already mentioned, these complexes
suffer from unfavorable steric interactions due to the cis
arrangement of the s-Bu moiety and the phosphorus-bearing
substituent of the (LiO)2 ring. Molecular models show that to
minimize steric effects, the s-Bu and N-benzylic fragments
should be kept away from each other (Chart 1). In this
conformation the carbanion center lies very close to the P-phenyl
rings, enabling the deprotonation. Complexes VIIIa and VIIIb
are thermodynamically disfavored, but they could be generated
under kinetic control. Once formed, they would be too reactive
to be detected, leading immediately to the ortho-metalated
species I. The fact that the concentration of I remains practically
constant implies that complexes Va and Vb do not equilibrate
with VIIIa and VIIIb by virtue of a large energy barrier of
interconversion. Complex-induced proximity effects in benzylic
lithiation directed by amide-type functional groups have been
reported previously.55f,h,56 To our knowledge, this is the first
time that the operation of the CIPE model has been demonstrated
in the ortho- and benzylic-directed deprotonation of a phosphi-
namide. Unfortunately, all attempts to grow crystals of these
anions suitable for X-ray studies were unsuccessful.
Deprotonation of 1 in the Presence of HMPA. We
performed an analogous NMR study on the sample containing
an excess of HMPA. As expected, the additive accelerates the
transformation of 1 into dearomatized species. The first 31P
NMR spectrum acquired at -90 °C (t1 ) 2 min) showed only
four signals. The signals corresponding to the two most
deshielded phosphorus atoms were readily assigned to the
dearomatized species III (δ 35.04 ppm) and IV (δ 39.8 ppm)
and occurred with a ratio of 1:0.6. The most intense signal results
from free HMPA (δ 25.18 ppm), and a broad one centered at
δ 25.7 ppm is assigned to the HMPA coordinated to the lithium
cations of III and IV (Figure 9). The splitting of the phosphorus
signal of each anion into a quartet of intensity 1:1:1:1 again
provides evidence for their monomeric nature.57 The increase
of the concentration of IV with respect to the sample prepared
in the absence of HMPA allowed its spectroscopic characteriza-
tion. For that purpose, we acquired the same set of NMR spectra
(56) Rein, K.; Goicoechea-Pappas, M.; Anklekar, T. V.; Hart, G. C.; Smith, G.
A.; Gawley, R. E. J. Am. Chem. Soc. 1989, 111, 2211.
(57) In our first report (see ref 10a) we assigned the lowest-field signal to a
solvent-separated ion pair based (SIP) on the 31P NMR spectrum (121.44
MHz) of a 0.158 M sample measured in THF-d8 at -90 °C. Under these
2
31
7
conditions the J( P Li) could not be resolved. Clearly, the lack of multiplicity
is not a consequence of SIP structure but a concentration effect.
9
J. AM. CHEM. SOC. VOL. 126, NO. 39, 2004 12561

A R T I C L E S
Ferna´ndez et al.
Figure 9. 31P{1H} NMR (121.49 MHz) spectrum of the lithiation of 1
with s-BuLi in the presence of HMPA, measured at t1 ) 2 min and -90
°C. The FID was multiplied by a Gaussian of LB ) -6 and GB ) 0.08
prior to the Fourier transformation.
Table 5. Selection of NMR Data of IV Measured in THF-d8 at -70
°C in the Presence of HMPAa
C/H
δH
δC
nJPC
3
3a
4
5
6
7
7a
9
10
13
3.90
3.76
3.90
5.64
4.19
5.97
81.29
52.36
95.71
129.38
92.50
134.47
62.43
141.36
136.01
145.53
14.7
19.3
5.6
S
14.7
12.7
164.8
114.5
7.1
Figure 10. 2D 7Li,31P{1H} HMQC spectrum(194.37 MHz) of a sample
with a ratio s-BuLi:1:HMPA of 1.5:1:1, measured at -110 °C in THF-d8.
7.99
6.6
HMPA ratio of 1.5:1:1 was prepared. The 31P spectrum
measured at -110 °C consisted of a mixture of dearomatized
anions III, IV, and VII and at least four HMPA complexes
absorbing in a narrow region (24.1-24.6 ppm). Two 1:1:1:1
quartets resulting from 7Li,31P couplings were clearly observed
in this region. However, broadening effects and partial overlap
impeded the determination of the precise number of species
present (Figure S8a, Supporting Information). Fortunately the
a δ in ppm, J in Hz.
used for the structural assignment of I and III. The results of
the analysis of these spectra are given in Table 5. Comparison
of the NMR data of III and IV indicates that both have very
similar structures. Significantly, the largest differences cor-
respond to the magnitudes of 1JPC. Thus, the phosphorus-carbon
coupling of the ipso carbon C9 is 20 Hz larger in III than in
IV, whereas the opposite occurs for C7a, ∆1JPC(III-IV) ) -21
Hz. These findings suggest that compounds III and IV may
differ in the relative configuration of the phosphorus atom. This
is a reasonable suspicion considering that the two P-phenyl rings
of the benzylic anion II are diastereotopic, and therefore, the
anionic cyclization by attack at the pro-R or pro-S phenyl ring
would provide two dearomatized species epimeric at the
phosphorus center.
This hypothesis was supported by the 2D ROESY spectrum
measured at -70 °C (Figure S7, Supporting Information). In
both compounds a correlation is observed between H3a and the
ortho protons of the phenyl ring adjacent to the nitrogen,
indicative of their syn arrangement. Unfortunately the correla-
tions involving the P-phenyl rings could not be unraveled due
to signal overlaps. If the relative configuration of C3 and C3a
is the same in both compounds, the difference must arise from
the configuration of the phosphorus atom. However, the problem
of assigning the correct P-configuration of III and IV remains
unresolved. The ROESY spectrum also confirmed the coordina-
tion of HMPA molecules to the lithium through the correlations
of the methyl protons with H3, H5, H7, and aromatic protons.
Additionally, the lower intensity of the correlation with H5
relative to that of H7 supports the η3 coordination of the lithium
to the dearomatized ring proposed above.
7Li spectrum did not suffer from these effects. The major signals
2
were readily identified as three triplets (δ -0.51 ppm, JPLi
)
)
2
2
10.8 Hz; δ -0.13 ppm, JPLi ) 7.6 Hz; δ -0.02 ppm, JPLi
2
8.8 Hz) and one quartet (δ -0.43 ppm, JPLi ) 9.4 Hz). The
assignment of the multiplicity was assisted by the Li{31P,1H}
7
spectrum (Figure S8b,c, Supporting Information). The 7Li,31P-
{1H} HMQC spectrum showed that each one of the two lowest-
field lithium triplets (δ -0.13 and -0.02 ppm) correlates with
a pair of phosphorus nuclei corresponding to one HMPA
molecule and one dearomatized anion; that is, the coordination
sphere of lithium in the dearomatized anions is completed
through binding to one HMPA molecule (Figure 10). The
remaining 7Li triplet (δ -0.51 ppm) and quartet (δ -0.43 ppm)
correlate exclusively with HMPA signals, indicating that these
multiplets originate from the respective coordination of two and
three molecules of the phosphorus ligand to the lithium cation.
Curiously, the 31P signals of these complexes appear completely
overlapped.
The unequivocal assignment of the structures III and IV was
achieved in two ways. We reasoned that if the concentration of
IV in the sample prepared in the absence of cosolvent increased
with increasing temperature, a similar effect could occur when
HMPA is present. The temperature dependence of the relative
ratio of III:IV could be anticipated based on the known
reversibility of the anionic cyclization step and the possibility
of obtaining products of kinetic or thermodynamic control.17
Indeed, the ratio III:IV practically equalizes when the temper-
ature of the sample is raised from -90 to -10 °C, and this
product distribution remains unaltered when the temperature is
Additional evidence of HMPA coordination to the dearoma-
tized complexes was obtained through 7Li,31P shift correlation
spectroscopy. To avoid the presence of the intense signal
produced by the excess HMPA, a sample with an s-BuLi:1:
9
12562 J. AM. CHEM. SOC. VOL. 126, NO. 39, 2004

N-Alkyl-N-benzyldiphenylphosphinamides
A R T I C L E S
Scheme 5 a
a (1) s-BuLi (2.5 equiv), HMPA (6 equiv), THF, -90 °C, 30 min. (2) 60 min, -30 °C. (3) 2,6-Di-tert-butyl-4-methylphenol, -90 °C, 30 min. (4) H2O.
decreased again. Interestingly, at temperatures close to 0 °C
the anions revert quantitatively to the starting phosphinamide
1, most probably by deprotonation of the solvent. A similar
concentration variation is observed when the sample without
additives is heated from -80 to -10 °C. The final ratio III:IV
obtained is 55:45 (Figure S9, Supporting Information). These
results indicate that III is the precursor of the kinetic product
4 (E ) H, Scheme 4), and IV is the precursor of the
thermodynamic product 13 (Scheme 5). On this basis, it can be
safely affirmed that the configuration of the phosphorus atom
of III must be the same as that in 4. Consequently, the P-phenyl
ring of IV is arranged anti to H3a and the phenyl substituent
linked to C3.
The equilibration between III and IV suggested the possibility
of isolating both diastereoisomers. With this aim in mind, we
performed the dearomatization of 1 in the usual way (addition
of 2.5 equiv of s-BuLi in THF at -90 °C in the presence of 6
equiv of HMPA for 30 min), and the reaction mixture was then
stirred for 1 h at -30 °C to establish the thermodynamic
equilibrium prior to quenching with 2,6-di-tert-butyl-4-meth-
ylphenol13 at -90 °C (Scheme 5). Under these conditions,
compounds 4 and 13 were obtained in a ratio of 55:45 in 88%
yield (Table S2, Supporting Information). The separation of both
isomers by flash column chromatography allowed the spectro-
scopic characterization of 13 (see Supporting Information). One-
dimensional gNOESY experiments confirmed the structural
assignment previously made: the selective excitation of the
ortho protons of the P-phenyl ring (HP°) and of H3a produced
NOEs on H3 and the ortho protons of the phenyl ring (H°)
linked to C3, respectively (Figure S10, Supporting Information).
II is a monomer of relatively short half-life (<90 min) evolving
through intramolecular attack of the carbanion center at each
of the diastereotopic P-phenyl rings. As a result, the four
monomeric dearomatized anions III, IV, VI, and VII (initial
ratio of 1:0.07:0.13:0.25) are obtained. These four anionic
species are in equilibrium (i.e., the anionic cyclization is a
reversible process). On the other hand, a significant amount of
ortho deprotonation leading to I takes place in the initial
moments of the reaction. The ortho anion I is assumed to be a
dimer of undefined structure arising from two highly reactive
precomplexes VIIIa and VIIIb (CIPE mechanism), isomers of
Va and Vb, that could not be detected. They do not equilibrate
with Va and Vb under the conditions used. Once formed, VIIIa
and VIIIb immediately progress to the ortho anion I, whose
concentration remains constant from that instant. When the
reaction reaches a steady state, only anions I, III, and IV are
observed in solution (relative ratio of 1.2:1:0.1). The dearoma-
tized anions III and IV represent the precursors of the products
of kinetic and thermodynamic control, respectively. The depro-
tonation of 1 in the presence of HMPA directly gives a mixture
of III and IV in a ratio of 1:0.6. In this case the lithium atom
is coordinated to one molecule of HMPA. Thus, the cosolvent
catalyzes the translocation of the ortho anions I to the benzylic
species II as well as the dearomatizing cyclization reaction. One-
dimensional 7Li{31P,1H} and 2D 7Li,31P{1H} HMQC NMR
experiments proved to be excellent tools for the structural
assignment of lithiated organophosphorus compounds containing
a PdO group. The fact that anion IV predominates at temper-
atures above -30 °C is synthetically meaningful because it gives
access to a new family of dearomatized compounds with a
different configuration at the phosphorus center with respect to
those usually isolated in the anionic dearomatizing-electrophilic
alkylation reactions of phosphinamides.
Conclusions
A detailed picture of the lithiation of N-methyl-N-benzyl-
diphenylphosphinamides (1) with s-BuLi in THF has been
obtained by deuterium labeling and multinuclear magnetic
resonance studies. In the absence of coordinating cosolvents
such as HMPA or DMPU, the lithiation of 1 and subsequent
trapping reaction with D2O revealed that the phosphinamide
group is capable of directing the metalation to the ortho and
benzylic positions, with a preference for the latter by a factor
of almost 2. Ortho-to-benzylic anion translocation occurs only
when HMPA or DMPU is present in the reaction medium, as
evidenced by the scrambling of deuterium in the anionic
dearomatization of a derivative of 1 dideuterated at the benzylic
position. NMR monitoring of the reaction allowed us to unravel
the detailed reaction pathway. First, in the absence of HMPA,
two diastereomeric s-BuLi-phosphinamide precomplexes, Va
and Vb, are formed. These are dimeric precursors for the
deprotonation at the benzylic position (CIPE mechanism).
Moreover, for the Vb dimer a slow inversion at the carbanion
center of the sec-butyllithium was observed. The benzylic anion
Acknowledgment. Financial support by the Ministerio de
Ciencia y Tecnolog´ıa (Project PPQ2001-3266) is gratefully
acknowledged. We are indebted to Dr. M. G. Davidson (Bath)
and Dr. D. Stalke (Wu¨rzburg) for their guidance in the attempted
X-ray studies of the dearomatized anions. I.F. thanks Junta de
Andaluc´ıa for a Ph.D. fellowship.
Supporting Information Available: Experimental details
including NMR sample preparation, relevant spectrometer
information, synthetic procedures, and characterization of 1-d2,
7-d2, 8-d2, 9-d2, 4, and 13. 2D 1H,31P gHMQC, 1H,13C-gHMQC,
and 1H,13C-gHMBC NMR spectra of the lithiation of 1 at -90
1
°C in THF-d8. Evolution of H and 31P{1H} NMR spectra of
the lithiation of 1 as a function of metalation time (t1). 31P{1H}
NMR spectrum at t1 ) 2 min of lithiation at -90 °C in THF-d8
with and without resolution enhancement (LB ) 2). 2D
gROESY spectrum of the lithiation of 1 in the presence of
9
J. AM. CHEM. SOC. VOL. 126, NO. 39, 2004 12563

A R T I C L E S
Ferna´ndez et al.
7
7
HMPA. 31P{1H}, Li{1H}, and Li{31P,1H } NMR spectra of
excitation of 13. Product distribution in the dearomatization-
protonation reaction of 1. This material is available free of
charge via the Internet at http://pubs.acs.org.
the lithiation of 1 at -110 °C in the presence of HMPA.
1
1
Temperature effect on the olefinic region of the H and H-
{31P} NMR spectra of the mixture I, III, and IV in the absence
of HMPA. 1D gNOESY rows corresponding to the selective
JA039863T
9
12564 J. AM. CHEM. SOC. VOL. 126, NO. 39, 2004