Multinuclear NMR study of the solution structure and reactivity of tris(trimethylsilyl)methyllithium and its iodine ate complex

Article abstract of DOI:10.1021/jo802032d

The extreme steric bulk of tris(trimethylsilyl)methyl derivatives (1-X) provides interesting structural and dynamic behavior for study. Dynamic NMR studies on 1-SePh and 1-I showed restricted rotation around the C-Si bonds of each trimethylsilyl groups. A

Full text of DOI:10.1021/jo802032d

Multinuclear NMR Study of the Solution Structure and Reactivity
of Tris(trimethylsilyl)methyllithium and its Iodine Ate Complex
Hans J. Reich,* William H. Sikorski, Aaron W. Sanders, Amanda C. Jones, and
Kristin N. Plessel
Department of Chemistry, UniVersity of Wisconsin, 1101 UniVersity AVenue, Madison, Wisconsin 53706
reich@chem.wisc.edu
ReceiVed September 17, 2008
The extreme steric bulk of tris(trimethylsilyl)methyl derivatives (1-X) provides interesting structural and
dynamic behavior for study. Dynamic NMR studies on 1-SePh and 1-I showed restricted rotation around
the C-Si bonds of each trimethylsilyl groups. An extensive multinuclear NMR study of natural abundance
6
and Li and 13C enriched 1-Li revealed three species in THF-containing solvents, a dimer 1T, and two
monomers, the contact ion pair 1C, and solvent separated ion pair 1S. Observed barriers for interconversion
of 1-Li aggregates were unusually high (∆Gq ca. 9 kcal/mol for exchange of 1S and 1C, ∆Gq ) 16.4
41
kcal/mol for exchange of 1T with 1C and 1S), allowing for study of reactivity of each aggregate
individually. We can show that 1S is at least 50 times as reactive as 1C and at least 5 × 1010 times as
reactive as 1T toward MeI. The large difference in reactivity allowed further study on the mechanism of
the lithium-iodine exchange of 1-I with 1-Li and characterization of the intermediate iodine ate complex
4. Additional calibrations are presented for the sensitive yet chemically inert 13C NMR chemical shift
thermometer 1-H.
Introduction
THF,1c TMEDA1d or PMDTA1d it forms triple ions of type
+
R2Li- Li+ (1T), with counterions Li(THF)4+, Li(TMEDA)2
The tris(trimethylsilyl)methyl group as a substituent in
organometallic compounds imparts unusual chemical properties,
largely as a result of extraordinary steric effects.1,1b Here we
present our findings on the properties of tris(trimethylsilyl)-
methyl lithium (1-Li), which was first studied by the Eaborn
group.1 In the solid state, when crystallized in the presence of
and the very unusual inverse triple ion Cl(Li-PMDTA)2+ (when
formed in the presence of LiCl). Comparison of solid state
MAS2a and solution NMR spectra suggested that the triple ion
is also present in THF solution, in addition to a second species,
which was assigned the structure 1C, the contact ion pair.1e
When prepared in the absence of donor solvents by Li/Hg
exchange of [(Me3Si)3C]2Hg a four-center dimer with agostic
Li-H interactions forms.3
(1) (a) For example, bis(tris(trimethylsilyl)methyl)-zinc can be steam distilled
and does not react with either bromine or boiling concentrated hydrochloric
acid. Eaborn, C.; Retta, N.; Smith, J. D. J. Organomet. Chem. 1980, 190, 101–
106. (b) Eaborn, C.; Smith, J. D. J. Chem. Soc., Dalton Trans. 2001, 1541–
1552. (c) Eaborn, C.; Hitchcock, P. B.; Smith, J. D.; Sullivan, A. C. J. Chem.
Soc., Chem. Commun. 1983, 827–828. (d) Buttrus, N. H.; Eaborn, C.; Hitchcock,
P. B.; Smith, J. D.; Stamper, J. G.; Sullivan, A. C. J. Chem. Soc., Chem. Commun.
1986, 969–970. (e) Avent, A. G.; Eaborn, C.; Hitchcock, P. B.; Lawless, G. A.;
Lickiss, P. D.; Mallien, M.; Smith, J. D.; Webb, A. D.; Wrackmeyer, B. J. Chem.
Soc., Dalton Trans. 1993, 3259–3264. (f) Eaborn, C.; Clegg, W.; Hitchcock,
P. B.; Hopman, M.; Izod, K.; O’Shaughnessy, P. N.; Smith, J. D. Organometallics
1997, 16, 4728–4736.
We have encountered numerous triple ions in our NMR
studies of the effects of HMPA on organolithium reagents.4a-d
(2) (a) Pepels, A.; Gu¨nther, H.; Amoureux, J. P.; Fernandez, C. J. Am. Chem.
Soc. 2000, 122, 9858–9859.
10.1021/jo802032d CCC: $40.75
Published on Web 12/10/2008
2009 American Chemical Society
J. Org. Chem. 2009, 74, 719–729 719

Reich et al.
At the time we began this work,5 1T was the only localized
carbanion triple ion to be characterized (others have been found
since6a), although triple ions formed by delocalized lithium
π-complexes such as cyclopentadienyl lithiums (lithocenes),6b
and pentadienyls7a,8 as well as lithium amides,9a,b,10,11 and
ꢀ-keto12 and lithium ꢀ-imino9c enolates were known. We
undertook an NMR spectroscopic examination of 1-Li to help
better define the solution structure of this interesting system
than was possible when earlier studies were performed.1e,5 We
found that the structural characteristics and dynamic behavior
of the compound are even more unusual than previously
reported.
Results and Discussion
We have used three different THF-containing mixed solvents
for the experiments on 1-Li. This was necessary so that the
full range of temperatures (from -135 °C to +60 °C) could be
covered. Our workhorse solvent was 3:2 THF/ether, but 3:2:1
Me2O/THF/ether and 1:3 THF/Me2O were also used for very
low-temperature work. The latter mixture was especially crucial
for the rapid injection NMR (RINMR) work, since THF readily
crystallized from the other mixtures below -130 °C when the
sample was stirred (although unstirred samples gave long-lasting
supersaturated solution appropriate for standard variable tem-
perature work). We believe these three mixtures to have
comparable solvation properties for lithium reagents. For
example, the n-BuLi dimer/tetramer association constant was
66, 63, and 41 M-1 in the three solvents, and the monomer/
dimer association constant of 2-ethylphenyllithium was 0.19 M-1
in both 3:2 THF/ether and 3:2:1 Me2O/THF/ether at -140 °C.4e
The ratio of 1S, 1C and 1T to be discussed below are also very
similar in these solvent mixtures. When direct rate comparisons
were made, every effort was made to compare experiments in
identical solvent mixtures, but occasionally comparisons be-
tween different solvents were useful. The lithium reagents were
also briefly examined in diethyl ether, to determine the effects
on structure of a lower dielectric constant solvent.
Solutions of 1-Li could be prepared by the reported procedure
of metalation of 1-H with MeLi in THF at room temperature.1c
Such solutions were suitable for some NMR spectroscopic work,
and were used to prepare the selenide 1-SePh by reaction with
diphenyl diselenide and the iodide 1-I by reaction with iodine.
However, it was more convenient to prepare samples im-
mediately before use by the Li/Se exchange13a of 1-SePh with
n-BuLi, n-Bu6Li or Et6Li (half-life of ca. 40 min at -130 °C).
This technique produced clean solutions of 1-Li which were
adequate for careful spectroscopic or kinetic studies. The
BuSePh formed did not interfere with the spectroscopic work.
13C enriched 1-SePh was prepared as previously described using
[13C]formaldehyde via [13C](PhSe)2CH2,4b or by a shorter
synthesis in which [13C]chloroform was reductively silylated
to produce [13C]1-H,14 which was metalated by the literature
procedure1c and quenched with Ph2Se2.
(3) Hiller, W.; Layh, M.; Uhl, W. Angew. Chem., Int. Ed. Engl. 1991, 30,
324–326.
¨
(4) (a) Reich, H. J.; Sikorski, W. H.; Gudmundsson, B. O.; Dykstra, R. R.
J. Am. Chem. Soc. 1998, 120, 4035–4036. (b) The sample temperature during
NMR experiments was measured using the tris(trimethylsilyl)methane internal
13C NMR chemical shift thermometer: Sikorski, W. H.; Sanders, A. W.; Reich,
H. J. Magn. Resonan. Chem. 1998, 36, S118–S124. (c) Reich, H. J.; Goldenberg,
¨
W. S.; Gudmundsson, B. O.; Sanders, A. W.; Kulicke, K. J.; Simon, K.; Guzei,
I. A. J. Am. Chem. Soc. 2001, 123, 8067–8079. (d) Jantzi, K. L.; Puckett, C. L.;
Guzei, I. A.; Reich, H. J. J. Org. Chem. 2005, 70, 7520–7529. (e) Reich, H. J.;
Holladay, J. E.; Mason, J. D.; Sikorski, W. H. J. Am. Chem. Soc. 1995, 117,
12137–12150. (f) Reich, H. J.; Goldenberg, W. S.; Sanders, A. W.; Jantzi, K. L.;
Tzschucke, C. C. J. Am. Chem. Soc. 2003, 125, 3509–3521. (g) Reich, H. J.;
Borst, J. P.; Dykstra, R. R.; Green, D. P. J. Am. Chem. Soc. 1993, 115, 8728–
8741. (h) Reich, H. J.; Borst, J. P.; Dykstra, R. R. Tetrahedron 1994, 50, 5869–
5880. (i) Sikorski, W. H.; Reich, H. J. J. Am. Chem. Soc. 2001, 123, 6527–
6535. (j) Reich, H. J.; Dykstra, R. R. Angew. Chem., Int. Ed. Engl. 1993, 32,
1469–1470. (k) Reich, H. J.; Dykstra, R. R. J. Am. Chem. Soc. 1993, 115, 7041–
7042. (l) 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, 7201–7210. (m) Reich, H. J.; Holladay, J. E.; Walker, T. G.; Thompson,
J. L. J. Am. Chem. Soc. 1999, 121, 9769–9780. (n) Jones, A. C.; Sanders, A. W.;
Bevan, M. J.; Reich, H. J. J. Am. Chem. Soc. 2007, 129, 3492–3493. (o) Reich,
H. J.; Green, D. P.; Phillips, N. H. J. Am. Chem. Soc. 1991, 113, 1414–1416.
¨
(p) Reich, H. J.; Gudmundsson, B. O.; Green, D. P.; Bevan, M. J.; Reich, I. L.
HelV. Chim. Acta 2002, 85, 3748–3772. (q) Reich, H. J.; Phillips, N. H. Pure
Appl. Chem. 1987, 59, 1021–1026.
(5) (a) Reich, H. J.; Sikorski, W. H.; Thompson, J. L.; Sanders, A. W.; Jones,
A. C. Org. Lett. 2006, 8, 4003–4006. (b) Jones, A. C.; Sanders, A. W.; Sikorski,
W. H.; Jansen, K. L.; Reich, H. J. J. Am. Chem. Soc. 2008, 130, 6060–6061.
(6) (a) Bildmann, U. J.; Mueller, G. Organometallics 2001, 20, 1689–1691.
(b) Eiermann, M.; Hafner, K. J. Am. Chem. Soc. 1992, 114, 135. (c) Harder, S.;
Prosenc, M. H. Angew. Chem., Int. Ed. Engl. 1994, 33, 1744. (d) Zaegel, F.;
Gallucci, J. C.; Meunier, P.; Gautheron, B.; Sivik, M. R.; Paquette, L. A. J. Am.
Chem. Soc. 1994, 116, 6466–6567.
(7) (a) Fraenkel, G.; Hallden-Abberton, M. P. J. Am. Chem. Soc. 1981, 103,
5657–5664. (b) Fraenkel, G.; Chow, A.; Winchester, W. R. J. Am. Chem. Soc.
1990, 112, 6190–6198. (c) For an insightful discussion of dynamic processes
involving C-Li multiplets in monomeric aryllithium reagents, see: Fraenkel, G.;
Subramanian, S.; Chow, A. J. Am. Chem. Soc. 1995, 117, 6300–6307. (d) For
many organolithium reagents, especially aggregated ones, quadrupolar broadening
is severe enough with 7Li (92.6% natural abundance) that C-Li coupling cannot
be resolved. The 6Li analogs show little or no quadrupolar broadening and thus
couplings are more easily seen: Fraenkel, G.; Fraenkel, A. M.; Geckle, M. J.;
Schloss, F. J. Am. Chem. Soc. 1979, 101, 4745–4747.
The selenide precursors to the phenyl and isopropyl analogs
2-Li and 3-Li were prepared as shown in Scheme 1. This route
could also serve to prepare 13C labeled materials since [13C]CH-
Br3 is commercially available, but we did not find it necessary
to do so.
Dynamic Properties of Tris(trimethylsilyl)methyl Deri-
vatives. Some tris(trimethylsilyl)methyl derivatives show ro-
tational DNMR effects in the temperature region of interest for
many of the experiments reported here. Specifically, the
trimethylsilyl 13C signals of 1-SePh decoalesced into a 2:1 ratio
of signals around -130 °C (Figure 1a). Although we initially
assigned this to restricted rotation around the C-SePh bond,
which would render one of the trimethylsilyl signals different
from the other two, we observed a very similar decoalescence
(8) Hoic, D. A.; Davis, W. M.; Fu, G. C. J. Am. Chem. Soc. 1995, 117,
8480–8481.
(9) (a) Romesberg, F. E.; Gilchrist, J. H.; Harrison, A. T.; Fuller, D. J.;
Collum, D. B. J. Am. Chem. Soc. 1991, 113, 5751–5757. (b) Romesberg, F. E.;
Bernstein, M. P.; Gilchrist, J. H.; Harrison, A. T.; Fuller, D. J.; Collum, D. B.
J. Am. Chem. Soc. 1993, 115, 3475–3483. (c) Galiano-Roth, A. S.; Collum, D. B.
J. Am. Chem. Soc. 1988, 110, 3546–3553. (d) Lucht, B. L.; Bernstein, M. P.;
Remenar, J. F.; Collum, D. B. J. Am. Chem. Soc. 1996, 118, 10707–10718.
(10) (a) Jackman, L. M.; Scarmoutzos, L. M.; Porter, W. J. Am. Chem. Soc.
1987, 109, 6524–6525. (b) Jackman, L. M.; Scarmoutzos, L. M.; Smith, B. D.;
Williard, P. G. J. Am. Chem. Soc. 1988, 110, 6058–6058.
(11) Gornitzka, H.; Stalke, D. Angew. Chem., Int. Ed. Engl. 1994, 33, 693–
695.
(12) (a) Cambillau, C.; Bram, G.; Corset, J.; Riche, C. NouV. J. Chim. 1979,
3, 9–11. (b) Cambillau, C.; Ourevitch, M. J. Chem. Soc., Chem. Commun. 1981,
996–997. (c) Raban, M.; Haritos, D. P. J. Am. Chem. Soc. 1979, 101, 5178–
5182. (d) Teixidor, F.; Llobet, A.; Casabo´, J.; Solans, X.; Font-Altaba, M.; Aguilo´,
M. Inorg. Chem. 1985, 24, 2315–2317.
(13) (a) Bauer, W.; Seebach, D. HelV. Chim. Acta 1984, 67, 1972–1988. (b)
Seebach, D.; Hassig, R.; Gabriel, J. HelV. Chim. Acta 1983, 66, 308–337. (c)
Seebach, D.; Siegel, H.; Gabriel, J.; Ha¨ssig, R. HelV. Chim. Acta, 1980, 63,
2046–2053.
(14) (a) Merker, R. L.; Scott, M. J. J. Organomet. Chem. 1965, 4, 98–100.
(b) Guijarro, A.; Yus, M. T etrahedron 1996, 52, 1797–1810.
720 J. Org. Chem. Vol. 74, No. 2, 2009

Tris(trimethylsilyl)methyllithium and its Iodine Ate Complex
SCHEME 1. Synthesis of Selenide Precursors to 1-Li, 2-Li
and 3-Li
for 1-I (Figure 1b), for which this is not a tenable explanation.
Rather, it must be restricted rotation around the C-Si bonds of
each trimethylsilyl group that causes the observed dynamic
effect, and this has been supported by a careful study reported
for several other derivatives.15
Tris(trimethylsilyl)lithium in THF/Ether. Our extensive
multinuclear NMR spectroscopic examination of 1-Li both at
6
natural abundance, as well as Li and 13C singly and doubly
enriched showed that there were three species present in THF-
containing solvents, the triple ion 1T, the contact ion pair (CIP)
1C, and the separated ion pair (SIP) 1S (Figure 2). The NMR
signals for 1C and 1S coalesced at ca. -70 °C, and the signals
of 1T coalesced with this averaged signal at >50 °C (we will
discuss these experiments in more detail below). Thus all three
species are structurally closely related. Variable concentration-
experiments were complicated by this dynamic behavior, since
under conditions where all three sets of signals were well
resolved, 1T was no longer in mobile equilibrium with the other
two. We thus measured the concentration dependence of the
three species at room temperature, where only the signals of
1T and the average of 1C/1S were seen. A plot of log[1T] vs
log[1C/1S] showed a slope of 1.97, consistent with 1T being a
dimer, and 1C/1S monomers (Figure 3). At -118 °C, where
all three species could be quantitated, but where the triple ion
equilibrates slowly with the monomers, there was a consistent
deviation, such that comparison of 1T and 1S gave a slope of
1.6, whereas 1T vs 1C gave a slope of 2.3.16 The deviations
are in the direction that 1S might be slightly aggregated at low
temperature.
FIGURE 2. Multinuclear NMR study of 1-Li in 3:2 THF/Et2O. (a)
6
6
7Li NMR spectra at natural abundance. (b) Li NMR of Li enriched
6
6
1-Li. (c) Li NMR of doubly enriched (13C and Li) 1-Li (d-f) 13C
NMR of 6Li enriched, 13C enriched and doubly enriched 1-Li. (g) 29Si
1
NMR spectrum and (h) H NMR spectrum of 1-Li.
FIGURE 3. Variable concentration experiment of 1-Li at 20 °C in
3:2 THF/ether.
The structure of the dimer 1T could be firmly assigned as a
triple ion from the appearance of two lithium NMR signals,
one unusually downfield at δ 2.7, a shift effect seen for a number
of other triple ions,4a-d and the other with a chemical shift
characteristic of free solvated Li+ at δ -0.2 (Figure 2c). The
13C and 6Li double isotopically enriched compound showed the
expected coupling in the 13C (1:1:1 t, 1JC-Li ) 8.5 Hz) and 6Li
FIGURE 1. 13C NMR spectra from a variable temperature study in
3:2 THF/Et2O of: (a) 0.16 M tris(trimethylsilyl)methyl phenyl selenide
(1-SePh); (b) 0.26 M tris(trimethylsilyl)methyl iodide (1-I).
J. Org. Chem. Vol. 74, No. 2, 2009 721

Reich et al.
1
NMR spectra (1:2:1 t, JLi-C ) 8.5 Hz) arising from coupling
of one lithium to two carbons.
The assignment of 1C as the contact ion pair (CIP) was based
7
on the observation of coupling between Li and carbon in the
13C enriched material (Figure 2f), and in both the 13C (1:1:1
triplet, JC-Li ) 6.0 Hz) and Li NMR spectra in the Li/13C
doubly enriched compound (Figure 2c, 2f). The concentration-
dependent experiments supported this assignment. The third
component present in solutions of 1-Li was assigned the
separated ion pair (SIP) structure 1S. We recognize that the
observation of distinct NMR signals for CIP/SIP pairs in ethereal
solvents is unprecedented, although such structures are well-
known for [2.2.1]crypt17,18 and HMPA4f solvated species, where
the SIP and CIP have different levels of crypt/HMPA solvation
and ligand exchange is slow on the NMR time scale. In such
situations the exchange of CIP and SIP carbanions can be much
faster than interchange of differently solvated lithium
counterions.4g In other words, for these systems it is not the
breaking of the C-Li bond, but rather cleavage of the Li-HMPA
or Li-crypt association that provides the main barrier to ion pair
interconversion.
1
6
6
FIGURE 4. 13C NMR spectra from an HMPA titration of 13C- and
6Li-enriched 0.12 M tris(trimethylsilyl)methyllithium (1{13C}-6Li) in
3:2 THF/Et2O at -121 °C.
The evidence for the assignment of 1S can be summarized
as follows. The signals for 1S, 1C, and 1T grew in together
when the metalation of 1-H was followed by NMR spectros-
copy, and their ratio was consistent throughout the metalation.16
The signal(s) for 1S were present in all samples prepared by
several different methods (metalation, Li/Se exchange, Li/I
exchange), and the appropriate signals of all of the nuclei of
1S and 1C coalesced above -80 °C. All these observations show
that the signals for 1S were not an adventitious impurity. The
species is monomeric and shows no coupling between C and
Li even at the lowest temperatures accessible. The addition of
HMPA led to complete conversion to the separated ion pair
1S, as is seen for many related lithium reagents (Figure 4).4f-i
In the lithium NMR spectra, only one signal for separated Li+
species of 1T and 1S was observed, as expected for a mobile
free lithium cation. That 1S was associated with a free lithium
cation was shown by the preparation of nonequilibrium samples
(see below) containing no 1T, which still show a SIP-Li+ at
δLi -0.25 for 1S. In less polar media (ether) the signals for 1S
disappear (Figure 5).
FIGURE 5. 6Li and 13C NMR spectra of an HMPA titration of a 0.09
M solution of 1-6Li in ether at -122 °C (1C1 indicates 1C coordinated
to one HMPA molecule).
The 13C NMR spectra of doubly enriched 1-Li showed 29Si
satellites for the carbanion carbon, with 1JC-Si of 41.0, 42.6 and
64.9 Hz for 1T, 1C, and 1S, respectively. The substantially
larger C-Si coupling for 1S can be ascribed to a more planar
environment at the carbanion carbon. Larger C-Si4i and
C-H4i,19,20 couplings for SIPs vs CIPs have been reported for
related compounds.
coordination strength and dielectric constant of ether vs THF
results in suppression of the ionic species 1S and 1T. The
HMPA titration showed a dramatic difference from the behavior
in THF in that mono-HMPA (1C1) and even some bis HMPA
CIP (1C2) species were formed, accompanied by conversion to
some 1T and eventually 1S. Superscripts refer to the number
of HMPA molecules coordinated to lithium.
Tris(trimethylsilyl)methyllithium and PMDTA. The tri-
dentate cosolvent N,N,N′,N″,N″-pentamethylethylenetriamine
(PMDTA) often (but not always4c,e) converts dimeric lithium
reagents to monomers (e.g., neopentyllithium,7b PhLi,4j,e,21,22
anisyllithium,23 LiN(SiMe3)29d). Treatment of 1-Li with PM-
DTA leads to a smooth conversion to the PMDTA-complexed
1S.16 Presumably steric factors prevent this tridendate ligand
from coordinating to either 1C or 1T.
Tris(trimethylsilyl)methyllithium in Diethyl Ether. Solu-
tions of 1-Li in Et2O showed only a single species, which could
be assigned the 1C structure based on the observation of a 1:1:1
triplet (1JC-Li ) 8.0 Hz) at δ 0.5 in the 13C NMR spectrum and
6
a single peak in the Li NMR spectrum (Figure 5). The lower
(15) Casarini, D.; Foresti, E.; Lunazzi, L.; Mazzanti, A. Chem.-Eur. J. 1999,
5, 3501–3508.
(16) Spectra and additional data are shown in the Supporting Information.
(17) (a) Cahen, Y. M.; Dye, J. L.; Popov, A. I. J. Chem. Phys. 1975, 79,
1289–92. (b) Cahen, Y. M.; Dye, J. L.; Popov, A. I. J. Phys. Chem. 1975, 79,
1292–1295. (c) Shamsipur, M.; Popov, A. I. J. Phys. Chem. 1986, 90, 5997–
5999.
(21) Bauer, W.; Winchester, W. R.; Schleyer, P. v. R. Organometallics 1987,
6, 2371–2379.
(18) Shamsipur, M.; Karkhaneei, E.; Afkhami, A. Polyhedron 1998, 17,
3809–3815.
(19) Schade, S.; Boche, G. J. Organomet. Chem. 1998, 550, 381–395.
(20) Rabideau, P. W.; Wetzel, D. M.; Husted, C. A.; Lawrence, J. R.
Tetrahedron Lett. 1984, 25, 31–34.
(22) Schu¨mann, U.; Kopf, J.; Weiss, E. Angew. Chem. 1985, 97, 222.
Schu¨mann, U.; Kopf, J.; Weiss, E. Angew. Chem., Int. Ed. Engl. 1985, 24, 215–
216.
(23) Harder, S.; Boersma, J.; Brandsma, L.; van Mier, G. P. M.; Kanters,
J. A. J. Organomet. Chem. 1989, 364, 1–15.
722 J. Org. Chem. Vol. 74, No. 2, 2009

Tris(trimethylsilyl)methyllithium and its Iodine Ate Complex
FIGURE 7. Variable temperature 13C NMR study of [13C]1-Li in 3:2
THF/Et2O.
quartet. The signal at δ 1.9 for the triple ion 1T at -136 °C, on
the other hand, is a broad doublet, characteristic of a nucleus
coupled to a spin 3/2 7Li nucleus undergoing quadrupole induced
T1 relaxation at a rate comparable to JCLi. T1 is estimated at 11
ms at -135 °C from line-shape simulations using published
equations.7c,24 The sharp 1:1:1 triplet from the natural abundance
6
of Li is superimposed on the broad doublet (6Li has a much
smaller quadrupole moment, and hence longer T17d). Similar
multiplets have been reported for other C-Li13b,21 and P-Li
signals.4k Triple ions generally show broad peaks in their Li
7
NMR spectra, thus leading to poor resolution of 7Li-13C
coupling as seen here. It is usually necessary to observe 6Li to
detect them easily.4a,i,7d The larger size of the triple ion molecule
results in longer correlation times, and the linear structure with
two carbanions bonded to lithium1c presumably provides higher
electric field gradients than the normal tetrahedral coordination
of lithium. Both effects lead to more effective quadrupolar
relaxation for 1T compared to 1S and 1C.
FIGURE 6. Variable temperature 13C NMR study of 13C- and 6Li-
labeled 0.12 M tris(trimethylsilyl)methyllithium (1{13C}-6Li) in 3:2
THF/Et2O generated at low temperature under conditions where the
triple ion does not form. Note that there is scale shift at -45 °C in the
right panel.
DNMR Studies of 1-Li. A number of variable temperature
multinuclear NMR experiments were performed to establish the
dynamic behavior of the species 1T, 1S and 1C. Figure 6 gives
an overview of this behavior using 13C NMR spectroscopy with
As the temperature was raised, the methyl and carbanion
signals of 1S and 1C coalesced between -70 and -80 °C
(Figure 7). The signals for 1T showed no signs of dynamic
chemical exchange in this temperature range, since the peaks
for the natural abundance C-6Li signal remain sharp. However,
they do show some interesting relaxation effects. As the
temperature was raised, the underlying C-7Li signal collapsed
6
13C and Li doubly enriched material. There are two coales-
cences: First the carbanion and methyl signals for 1C and 1S
coalesce around -70 °C. The barrier is nearly identical in the
two principal solvent mixtures used in this study (∆Gq
)
-80
7
to a broad singlet; evidently the Li T1 is becoming shorter at
9.6 kcal/mol in 3:2 THF/ether and 9.4 kcal/mol in 1:3 THF/
Me2O). Above 40 °C signals for the triple ion 1T and the
averaged 1C/1S signals start to broaden, but coalescence can
barely be reached even for the relatively close signals of the
SiMe3 carbons because of solvent limitations. Similar behavior
higher temperatures (about 4 ms at -90 °C). This is the opposite
trend to what is usually observed.25 Interestingly, above -90
°C the signal broadens again, and becomes a broad doublet at
-47 °C with a 7Li T1 of 10 ms.16 Now T1 is showing the normal
increase with increasing temperature. A reasonable explanation
for this behavior is that the average molecular correlation time
for 1T is close to the Larmor precession frequency at ca. -90
was seen in variable temperature NMR experiments by observ-
16
1
7
6
ing the H, 29Si, Li and Li NMR signals.
Figure 7 shows 13C NMR spectra of the 13C enriched
compounds with Li at natural abundance between -135 and
-47 °C. At low temperature the signal for 1C is the expected
1:1:1:1 quartet from coupling to the spin 3/2 7Li nucleus (1JC-Li.
) 15.6 Hz). Careful examination reveals the natural abundance
6Li coupled 1:1:1 triplet between the central two peaks of the
7
°C, leading to optimal quadrupole relaxation of Li, with less
effective T1 relaxation at both higher and lower temperatures.26
(24) Bacon, J.; Gillespie, R. J.; Quail, J. W. Can. J. Chem. 1963, 41, 3063–
3069.
(25) Hartwell, G. E.; Allerhand, A. J. Am. Chem. Soc. 1971, 93, 4415–4418.
J. Org. Chem. Vol. 74, No. 2, 2009 723

Reich et al.
encumbered base such as an electron-rich benzaldehyde does
accelerate it.5b
Extrapolation of the dissociation rate of the triple ion
measured by DNMR to low temperatures (Figure 8) shows that
one can expect laboratory scale lifetimes at -100 °C. Thus it
should be possible to directly address the reactivity of the triple
ion in such solutions using in situ NMR experiments. The
DNMR data predict that even the interconversion of the two
ion pairs 1C and 1S will become slow enough below -130 °C
that reactivity experiments might be performed on them
individually. We will address these issues below. This exciting
possibility encouraged us to examine the analogs 2-Li and 3-Li
to see whether these might have even higher barriers which
would simplify the rapid injection NMR experiments.
Bis(trimethylsilyl)(phenyldimethylsilyl)methyllithium (2-
Li). Smith and Eaborn examined the solid state structure of 2-Li
crystallized from ether.1f The structure was monomeric, with a
single molecule of ether coordinated to lithium. The Si-Ph group
was also coordinated to the lithium. We find that in 3:2 THF/
ether solution 2-Li is almost completely a SIP. Thus the 13C
NMR signals were essentially unchanged as HMPA was added,
and the lithium signals went through the series of lithium-HMPA
complexes characteristic of SIPs.4f,16 In ether 2-Li is entirely a
CIP, as can be seen from the observation of C-Li coupling
(1JC-(6)Li ) 7.3 Hz) and the HMPA titration, which produces
some triple ion, 2C1 and 2C2 CIP species, as well as the various
HMPA-solvated SIP species.16
FIGURE 8. Rate data for the interconversion of 1T, 1C and 1S in 3:2
THF/ether.
We performed DNMR line shape analysis of the Me3Si
signals in several variable temperature experiments analogous
to the one in Figure 6, but with unlabeled material. As part of
the simulation we determined the concentrations of 1S, 1C and
1T as a function of temperature.16 There is only a very slight
increase in the fraction of 1T at low temperature. Apparently,
the opposing effects of increasing the number of particles upon
dissociation of 1T to monomers and reducing the number by
additional solvation of 1S and 1C are nearly balanced for this
system.
Bis(trimethylsilyl)(isopropyldimethylsilyl)methyllithium
(3-Li). To test whether the increased formation of SIP in 2-Li
was a consequence of the inductive effect of the phenyl
substituent we prepared the sterically similar isopropyl analog
3-Li. This was also a SIP in THF/ether, and a CIP in ether,
with behavior very similar to that observed for 2-Li.16 Thus
the increased tendency of 2-Li and 3-Li to form SIPs compared
to 1-Li appears to be largely a steric effect (steric hindrance to
solvation of the CIP). Neither 2-Li nor 3-Li allowed investiga-
tion of the triple ion, CIP and SIP reactivity in a single substrate
like that possible for 1-Li.
Reactivity of 1T, 1C, and 1S. Extrapolation of the DNMR
data of Figure 8 to low temperatures suggests that Curtin-
Hammett limitations should be easily avoidable for intercon-
version of 1T with the monomers, and might be possible for
the 1S/1C interconversion. We report here several “proof-of-
concept” experiments which demonstrate that the reactivity of
different aggregates can be measured. For these experiments
we used three techniques for low temperature mixing: (1) To
the sample of the substrate in a 10 mm NMR at -138 °C tube
was carefully added the reagent, the sample was frozen in liquid
N2, the NMR tube was placed in the spectrometer at the desired
temperature (e.g -135 or -85 °C) and allowed to melt. This
experiment provides only approximate temperature control, and
a time resolution of several minutes. The experiment of Figure
6 was carried out this way (1-SePh as substrate and n-BuLi as
reagent). (2) A sample containing 1-Li was placed in the NMR
probe cooled to the desired temperature, and an electrophile
was carefully added by syringe using a long nylon needle. A
long thin glass rod with a bent end was inserted into the tube,
the solution was stirred and spectra were obtained. This
Figure 8 shows the Eyring plot and activation parameters
found for the low temperature (interconversion of 1C and 1S)
and high temperature (1T and 1S/1C) processes. The barriers
for interconversion of 1S and 1C (∆Gq ca. 9 kcal/mol) are
substantially higher than other barriers of this type. For
comparison, barriers of 5.3 kcal/mol for a bis(arylthio)-
methyllithium,5b and 2-4 kcal/mol for arene radical anions have
been reported.27 The much higher barriers for 1-Li must have
steric origins. We speculate that the lithium of 1C may already
be poorly solvated in THF solution for steric reasons, so that
breaking of the lithium-carbon coordination of 1C can not be
effectively assisted by solvent, requiring separation of a severely
undersolvated lithium cation. Conversely, in the conversion of
1S to 1C the normally tetrahedrally solvated free lithium cation
may have to shed one or more of its associated solvents before
close approach to the carbanion center is possible. The free
+
energies of dissociation of THF from Li(THF)4 and from
+
Li(THF)3 in the gas phase have been estimated at 14 and 24
kcal/mol.28 Although gas phase dissociation energies of ionic
species will always be higher than solution phase ones, these
numbers provide some sense of the energetic cost of unassisted
solvent dissociation from a lithium cation.
Conversion of 1T to 1S and 1C likely encounters even more
severe solvation problems, leading to the very high barriers
observed (∆Gq ) 16.4 kcal/mol). The central lithium is
41
probably unsolvated, as judged from the crystal structure.1c Thus
the C-Li bond must be broken with little or no stabilization of
the C-Li fragment by solvation. Consistent with this is the
observation that the powerful donor solvent HMPA has no effect
on the rate of dissociation of 1T, although a less sterically
(26) The observation of a minimum T1 at-90 °C for 1T in THF/ether
((Me3Si)3CLiC(SiMe3)3 Li(THF)4, MW 470 or 765 (depending on whether the
cation and anion move as a unit) and 7Li Larmor frequency of 140 MHz) is
qualitatively consistent with other observations. For example, ReH7(PPh3)2 (MW
718 shows T1 (min) at-73 °C for the hydride signals at 500 MHz in toluene-d8:
Hamilton, D. G.; Crabtree, R. H. J. Am. Chem. Soc. 1988, 110, 4126–4133.
(27) The barrier for interconversion of the SIP/CIP of lithium naphthalide
in THF-ether is Gq ) 3.5 kcal/mol at-50 °C: Hirota, N.; Carraway, R.; Schook,
W. J. Am. Chem. Soc. 1968, 90, 3611–3618.
(28) Jarek, R. L.; Miles, T. D.; Trester, M. L.; Denson, S. C.; Shin, S. K. J.
Phys. Chem. A 2000, 104, 2230–2237.
724 J. Org. Chem. Vol. 74, No. 2, 2009

Tris(trimethylsilyl)methyllithium and its Iodine Ate Complex
Reaction of 1T with Methyl Iodide. When methyl iodide
was injected into a solution of 1-Li in 3:2 THF/ether at -85
°C the NMR signals of 1S and 1C disappeared immediately
(<1 s) to form (Me3Si)3CMe (1-Me), leaving only 1T in
solution. The latter reacted slowly over a period of 1-2 h
following first order kinetics to form additional 1-Me. In a series
of such experiments where 1 to 8 equiv of MeI were added the
rates were all identical within experimental error. Thus the
reaction of 1T with MeI is first order in 1T and zero order in
MeI. The triple ion does not react with MeI directly, but must
first dissociate to the monomers, one or both of which then react
quickly with MeI. The results of a variable temperature RINMR
study in the 1:3 THF/Me2O solvent mixture are reported in
Figure 10. The rates are indistinguishable from those measured
in 3:2 THF/ether. Extrapolation to -132 °C gave a value for
the rate of dissociation of 1T as 4·10-10 s-1. The pseudofirst
order rate of reaction of MeI with 1T must be slower than this.
Reaction of 1T with Diethyl Disulfide. The rate of reaction
of 1-Li with diethyl disulfide was examined between -62 and
-92 °C. The behavior was nearly identical to that of MeI. The
monomers 1S and 1C disappeared immediately to form
(Me3Si)3CSEt, followed by a slow first-order reaction with 1T
whose rate was independent of disulfide concentration. The rate
of the reaction of 1T with the disulfide was identical to that
with MeI (Figure 10b), confirming the same rate-limiting step
for both, i.e. dissociation of 1T to 1S and 1C.
FIGURE 9. 6Li (a) and 13C (b) NMR spectra at -121 °C acquired
before and after thermal equilibration of 0.12 M tris(trimethylsilyl)m-
ethyllithium (1{13C}-6Li) in 3:2 THF/Et2O generated by low-temper-
ature Li-Se cleavage of 1{13C}-SePh by n-Bu6Li.
There was a small discrepancy between the DNMR rates
extrapolated from temperatures above 30 °C to the region of
the RINMR rate experiments between -60 and -90 °C (Figure
8). One possible origin of this was suggested by a careful
analysis of the 13C chemical shifts of the C-Li and Me3Si
carbons of 1C/1S. If the spectra of Figure 6 are aligned
appropriately,16 a dramatic chemical shift change in both the
Me3Si and C-Li carbon for 1C/1S can be detected between
-60 and 30 °C, suggesting that a change in structure of one or
both of the component species (possibly loss of a solvent
molecule from 1C) occurred in this temperature range. This
would mean the high-temperature DNMR rates for dissociation
of 1T and the low temperature RINMR rates were performed
on solutions where one or more of the components are
structurally different.
Reaction of 1T with Benzaldehydes. Of the electrophiles
tried, only benzaldehydes reacted with 1T at a rate faster or
competitive with the dissociation to 1S and 1C. Even here, it
turned out that the role of the aldehyde was to catalyze
dissociation of 1T to the monomers, and there was no indication
of a chemical reaction with 1T. These results have been reported
in some detail, and will not be reiterated here.5b
FIGURE 10. (a) RINMR kinetic study of the reaction of 1T with MeI
in 1:3 THF/Me2O at 3 temperatures. (b) Activation parameters for the
reaction compared with other trapping experiments.
experiment provides a time resolution of ca. 5-20 s; (3) The
automated RINMR apparatus described recently4l became
available toward the end of this research, and was used for a
few of the experiments. This apparatus provides excellent
temperature control and a time resolution of 1 s or less at
temperatures as low as -135 °C. The experiment in Figure 10
was performed this way.
Preparation of Nonequilibrium Ratios of 1C, 1S and
1T. When 1-SePh was treated with n-BuLi below -130 °C,
the sample showed mainly 1C, 1S and a trace of 1T; the ratio
of 1C+1S to 1T was 96:4 (Figures 6 and 9). Significant
conversion to the equilibrium mixture did not occur until the
temperature was raised above -90 °C. When the equilibrated
sample was cooled back down to -121 °C, this ratio was 47:
53. This experiment shows that laboratory time experiments can
be performed on 1T under non-Curtin-Hammett conditions.
Several such experiments are described below.
Conversely, when any one of several electrophiles (MeI,
Et2S2, Ph2Se2) was injected into solution of 1-Li at temperatures
below -100 °C, the 1C and 1S disappeared within seconds,
leaving a solution containing only 1T. This illustrates that the
interconversion of 1T with the monomers is slow at these
temperatures, and that 1T does not react with these electrophiles
on a time scale of many minutes.
Reactions of 1S and 1C with Electrophiles. Extrapolation
of the DNMR data between -65 and -100 °C for intercon-
version of 1S and 1C (Figures 6 and 8) to -130 °C, the lowest
practical temperature for RINMR experiments in this system,
predicted a half-life of 22 s in 3:2 THF/ether. The measured
half-life of 20 s using experiments described below in the similar
solvent system 1:3 THF/Me2O was very close this value, and
is too fast to allow manual techniques for low-temperature
mixing to address the question of the reactivity of 1S and 1C.
These experiments were handled using the automated RINMR
apparatus.4l Two 1-s interval RINMR experiments of the
injection of MeI into solutions of 1-Li at -131 °C in 3:1 Me2O/
1
THF are shown in Figure 11. In this experiment the H NMR
signals of 1S disappeared during the 1 s mixing period (kobs. g
J. Org. Chem. Vol. 74, No. 2, 2009 725

Reich et al.
FIGURE 12. 13C RINMR experiment of the addition of 1-I-[13C] (0.26
mmol) to 0.22 M n-Bu6Li (0.8 mmol) in 3:2:1 Me2O/THF/ether at -136
°C. All of the major signals are those of the C-X carbon.
FIGURE 11. RINMR kinetic study of the reaction of 1C with MeI at
-131 °C in 1:3 THF/Me2O at two concentrations of MeI.
(SE1 reaction) was the primary product. In practice, the addition
of 1-I into a solution of n-Bu6Li at -136 °C did lead to the
immediate (10 s) disappearance of the n-BuLi dimer signal at
SCHEME 2. Pathways for Conversion of Ate Complexes to
Lithium Reagents
6
δ 1.8 in the Li NMR spectra (as with most other reactions of
n-BuLi,4l the tetramer was unreactive), and the appearance of a
6
7
separated lithium species in the Li and Li NMR spectra.
However, examination of the 13C spectra in a similar RINMR
experiment (Figure 12) performed on a 10 s time scale on 13C
enriched 1-I revealed conversion of about 30% of the iodide to
1S or 1C, in what appeared to be their approx 1:3 equilibrium
ratio. The major product was a new species, which we have
identified as the iodine ate complex 4. Apparently any 1-Li
initially formed reacts rapidly with 1-I to form 4, and it is the
SIP lithium cation of 4 that was principally detected in the
lithium NMR spectra, and not 1S. At longer reaction times
conversion of 4 to 1S and 1C, as well as formation of the
butylation product 1-Bu was detected. No 1T was formed.
Several experiments support the assignment of structure 4.
Addition of one equiv of 1-I to a solution of 1-Li at -78 °C
led to almost complete disappearance of all four starting
components (1-I, 1T, 1S and 1C) with formation of a new
species (4) (Figure 13c). This species had only a single peak in
the 13C NMR spectrum, but this was shown to be the result of
a coincidence of the Me3Si and C-I carbons at δ 6.1 by
comparison of solutions of 4 prepared from natural abundance
and 13C enriched samples, each of which showed the major peak
at δ 6.1.
2 s-1). The disappearance of 1C had a half-life of ca. 20 s, and
could be readily followed. The rate was independent of MeI
concentration over a 5-fold range. Thus 1C does not react with
MeI, but first dissociates to 1S.
From analysis of these data we can show that 1S is at least
50 times as reactive as 1C. Comparison with the extrapolated
rate of reaction with 1T gives lower limit of 5 × 1010 as the
relative rate of reaction of 1S versus 1T toward MeI. However,
since the 1S rate could actually be much faster (2 s-1 is a lower
limit) and the 1T rate much slower (there is no indication that
MeI reacts with 1T in competition with its dissociation), the
actual number is probably much larger that this.
Benzaldehyde behaved identically to MeI: 1S reacted com-
pletely within the first second after injection at -131 °C,
whereas 1C disappeared at the known rate of its dissociation
to 1S.5b
The Lithium-Iodine Exchange. Reactions of Tris(tri-
methylsilyl)methyl Iodide (1-I). The ability to individually
detect and measure the CIP and SIP structures of 1-Li suggested
that we could perform an experiment that might shed some light
on the mechanism of the lithium-iodine exchange, in which
we have had a long-term interest. There is substantial evidence
that such reactions proceed through intermediate iodine ate
complexes (R-I-R- Li+), as first suggested by Wittig,29 and
several such complexes have been spectroscopically char-
acterized.4m,30-33 An open question is the mechanism for
conversion of the ate complex intermediate to the product
lithium reagents. Spectroscopic studies have shown that iodine
ate complexes, as well as those formed from ArLi and
Ar2Te4m,n,30 and ArSnR34o are SIPs. Two limiting mechanisms
can be envisioned for conversion to the lithium reagents
(Scheme 2): (1) the ate complex dissociates to a SIP R-//Li+
and R-I (SE1 mechanism) followed by ion combination to form
RLi (CIP), or (2) the ate complex is attacked by the lithium
cation to form the CIP RLi directly (SE2 mechanism).
We reasoned that if a Li/I exchange were performed on 1-I
it might be possible to detect whether 1C (SE2 reaction) or 1S
Injection of MeI into a solution of 4 at -125 °C led to the
slow (t1/2 ca. 20 min) disappearance of 4, increase in the
concentration of 1-Me, and the reappearance of the signals for
1-I (the latter complicated by a near coincidence of the Me3Si
carbon signal with those of 1-H, and by dynamic broadening
(30) The evidence for this statement for Ph2ILi and Ph3HgLi has not been
fully published previously, and we include some NMR spectroscopic studies of
HMPA titrations in Supporting Information.
(31) Farnham, W. B.; Calabrese, J. C. J. Am. Chem. Soc. 1986, 108, 2449–
2451.
(32) Schulze, V.; Bro¨nstrup, M.; Bo¨hm, V. P. W.; Schwerdtfeger, P.;
Schimeczek, M.; Hoffmann, R. W. Angew. Chem., Int. Ed. Engl. 1998, 37, 824–
826.
(29) Wittig, G.; Schollkopf, U. Tetrahedron 1958, 3, 91–93.
(33) Bailey, W. F.; Patricia, J. J. J. Organomet. Chem. 1988, 352, 1–46.
726 J. Org. Chem. Vol. 74, No. 2, 2009

Tris(trimethylsilyl)methyllithium and its Iodine Ate Complex
FIGURE 13. 13C NMR spectra (Me3Si signals) in 3:2 THF/ether. (a)
1-I at -103 °C. (b) 1-Li at -125 °C. (c) Addition of 1 equiv of 1-I to
1-Li - formation of 4. (d) Injection of 2 equiv of MeI. The broad peak
at δ 3.3 is the partially coalesced (see Figure 1) trimethylsilyl signal
of 1-I.
FIGURE 14. 1H NMR spectra of a RINMR experiment in which 2 eq
of 1-I were injected into a 0.08 M solution of 1-Li in 3:1 Me2O/THF
at -130 °C.
of the trimethylsilyl signals of 1-I due to restricted rotation
around the C-Si bond, see Figure 115). The reaction is
essentially first order in MeI (2.9 fold rate increase for 4 fold
increase in concentration). Either the ate complex 4 is reacting
directly with the MeI, or the ate complex is in rapid equilibrium
with a small amount of 1S, and this is the reactive species. We
favor the latter explanation. The activation barrier obtained from
the pseudofirst order rate constant for the experiment with 8 eq
of MeI was ∆Gq-125 ) 10.3 kcal/mol. A 13C DNMR coalescence
experiment in 3:2 THF/ether on a solution of 4 containing excess
1-Li showed that interconversion of 4 with 1-Li had ∆Gq-69 of
∆Gq
) 10.4 kcal/mol), which is immediately converted to
-130
more 1S/1C by the excess (BuLi)2 present.
Since the interconversion of 1C and 1S (k ) 3.5·10-2 sec-1
in 1:3 THF/ether at -131 °C, ∆Gq
) 9.1 kcal/mol, Figure
-130
8) is faster than the rate of dissociation of 4, it is not possible
to directly answer the question we posed on whether 1C, 1S or
both are kinetic products of the ate complex dissociation. There
1
is, however, an indirect approach. A RINMR H experiment
1
10.3 kcal/mol. Similarly, H and 13C DNMR experiments of 4
was performed in which an excess of 1-I was injected into a
solution of 1-Li at -130 °C (Figure 14). In this experiment, 1S
disappeared in under 2 s, but 1C disappeared at the same rate
as the normal 1C to 1S conversion in the absence of 1-I (k )
0.04 s-1). In other words, 1-I does not react with 1C directly
on a time scale of ca. 10 s. In the aforementioned injection of
1-I into a solution of n-BuLi, no 1C was detected in the first
10 s of the reaction. If 1C were a kinetic product of the mixed
ate complex dissociation (SE2 mechanism in Scheme 2) it would
have built up during the first 10 s of the experiment. That it
does not suggests only 1S is the kinetic product of the exchange.
There is one additional point of interest in the NMR spectra
of 4. Other derivatives of tris(trimethylsilyl)methane show
decoalescence of trimethylsilyl 13C NMR signals below -100
°C due to slow rotation around the C-SiMe3 bond (Figure 1).
Compound 4 shows no such behavior down to -135 °C. Of
course it could simply be that the chemical shifts of the two
types of methyls are accidentally coincident, but more likely
the much longer bond length of the hypervalent C-I bond in 4
compared to that in 1-I considerably reduces steric effects and
lowers the barrier to trimethylsilyl rotation. For example, the
C-I bond length in lithium bis(pentafluorophenyl)iodinate is
240 pm,30 whereas a typical Ar-I bond length is 209 pm.34
13C Chemical Shift Thermometer. In all of the (Me3Si)3C-X
compounds the 13C NMR signal of the C-X carbon undergoes
a steady downfield drift of ca. 1 Hz/degree as the temperature
was raised (see Figures 6 and 7). This effect is very pronounced
for 1-H, and we have performed appropriate calibrations which
allow the difference between the 13C chemical shifts of the C-H
with excess 1-I coalesced in the same temperature range, and
gave a similar ∆Gq
of 10.0 kcal/mol for coalescence of 4
-69
and 1-I.16 Although these activation energies were measured at
different temperatures, they suggest that dissociation of 4 to
1S is at least comparable in rate to the reaction with MeI.
Two additional RINMR experiments with approximately 1-s
sampling rates were performed in 1:3 THF/Me2O at -130 °C.
Solutions of the iodide 1-I were injected into solutions of a
sufficiently large excess of n-BuLi (5 equiv.) that there was
more n-BuLi dimer than iodide present (BuLi tetramer is
unreactive under these conditions). In the first experiment the
reaction was followed by 7Li, in the second by 1H NMR
7
spectroscopy. The Li experiment confirmed that there was
excess (BuLi)2 present after the injection, and that during the
7
first 20 s only a SIP Li signal was observed (the Li NMR
signals of 1S and 4 cannot be distinguished). After this 1C
started to appear. In the 1H experiment the signal corresponding
to 4 was already present after 2 s, and there was no detectable
starting iodide 1-I left, and no detectable 1C, 1S or 1T was
initially formed. Small amounts of both 1S and 1C were
detectable after 20 s in approximately their 1:1 equilibrium ratio,
and these continued to increase over a period of several minutes,
accompanied by the slow formation of the butylation product
1-Bu from reaction of 1S with BuI. Although these experiments
are qualitative in nature and rate details are not fully interpret-
able, they are consistent with a reaction scheme in which 1S
and/or 1C are formed in the first few seconds of the reaction
(presumably via fast dissociation of the mixed ate complex
Me3.Si)3C-I-Bu- Li+) and are then rapidly captured by
remaining 1-I to form 4 (k > 2 s-1). The ate complex 4 then
(34) Chaplot, S. L.; McIntyre, G. J.; Mierzejewski, A.; Pawley, G. S. Acta
Crystallogr., Sect. B 1981, B37, 2210–2214.
slowly dissociates to 1S/1C and 1-I (k ) 4.6·10-4 sec-1
,
J. Org. Chem. Vol. 74, No. 2, 2009 727

Reich et al.
calibration curve for 3:2 THF/ether produced temperatures
approximately 2 degrees higher than when an equation calcu-
lated from a 3:2 combination of the two pure solvent equations
was used. We estimate that for pure solvents or specifically
calibrated solvent mixtures the temperatures are accurate to
within 1 degree.
Conclusions
Tris(trimethylsilyl)methyllithium (1-Li) exists as a mixture
of three species in solvents containing a significant fraction of
THFsthe separated ion pair 1S, the contact ion pair 1C and
the triple ion 1T. In diethyl ether solution only 1C is detected,
and addition of HMPA converts 1-Li completely 1S. Analogs
in which one trimethylsilyl group was replaced by phenyldim-
ethylsilyl (2) or isopropyldimethylsilyl (3) are all SIP in THF,
and CIP in ether.
The triple ion 1S interconverts with the two monomers slowly
on the NMR time scale up to about 40 °C, and slowly on the
laboratory time scale below -60 °C, with activation barriers
from 14 to 16 kcal/mol. The monomers 1C and 1S interconvert
slowly on the NMR time scale up to -100 °C, and laboratory
time scale experiments on individual ion pair structures are
possible below -130 °C using RINMR spectroscopy. Activation
barriers are ca. 9 kcal/mol. These exceptionally high barriers
for interconversion can be ascribed to steric effects which hinder
solvent participation in ion pair interconversion process.
Nonequilibrium ratios of the 1-Li species could be formed
as follows: if 1-Li was formed by Li/Se exchange of 1-SePh
below -100 °C then solutions of only 1S and 1C were formed,
with only traces of 1T. On the other hand, if equilibrated
solutions of all three species were treated with MeI below -100
°C, the monomers could be selectively removed, leaving
solutions containing only 1T. Injection of MeI into solution of
1C and 1S at -130 °C removed 1S, leaving only 1C (and any
1T that might have been present).
Reactivity experiments using RINMR spectroscopy showed
that 1T did not react directly with several electrophiles (MeI,
EtSSEt), rates of reaction were identical to the dissociation rate
of 1T to the reactive monomeric species. We showed that 1T
is at least 10 orders of magnitude less reactive than 1S.
Reactivity comparisons of 1C and 1S had a much more limited
dynamic range, since their interconversion had a half-life of
ca. 20 s at -130 °C, but we were able to show that several
electrophiles (MeI, benzaldehyde, and 1-I) did not react with
1C, but only with 1S.
FIGURE 15. Temperature-chemical shift calibrations of 1-H in various
solvents. ∆ν ) ν(CH) - ν(CH3).
TABLE 1. Chemical Shift Calibrations for the 1-H Chemical Shift
Thermometer (∆δ ) δ(CH) - δ(Si-CH3)
Pentane
C6D6
Toluene
CDCl3
CH2Cl2
3:2 THF/Et2O
MeOH
Et2O
THF
Me2O
T (°C) ) 83.231 (∆δ) - 86.6 (-124 to 26 °C)
T (°C) ) 82.249 (∆δ) - 19.9 (9 to 69 °C)
T (°C) ) 77.931 (∆δ) - 24.9 (-89 to 68 °C)
T (°C) ) 84.711 (∆δ) - 36.5 (-56 to 49 °C)
T (°C) ) 81.729 (∆δ) - 44.5 (-89 to 33 °C)
T (°C) ) 78.675 (∆δ) - 58.1 (-139 to 60 °C)
T (°C) ) 83.925 (∆δ) - 83.9 (-80 to 42 °C)
T (°C) ) 75.72 (∆δ) - 79.6 (-120 to -56 °C)
T (°C) ) 77.75 (∆δ) - 50.2 (-108 to 8 °C)
T (°C) ) 65.66 (∆δ) - 71.8 (-146 to -70 °C)
T (°C) ) 69.28 (∆δ) - 47.2 (-136 to -32 °C)
T (°C) ) 84.31 (∆δ) - 65.05 (-92 to -18 °C)
T (°C) ) 59.83 (∆δ) - 47.59 (-132 to -53 °C)
2,5-Me2THF
1,3-Dioxolane
2-MeTHF
3:2:1 Me2O/THF/Et2O T (°C) ) 68.73 (∆δ) - 66.9 (-153 to -73 °C)
3:1 Me2O/THF
1:1 THF/HMPA
3:1 THF/HMPA
HMPA
T (°C) ) 71.56 (∆δ) - 66.46 (-150 to -86 °C)
T (°C) ) 78.66 (∆δ) - 22.12 (-30 to -18 °C)
T (°C) ) 79.47 (∆δ) - 38.87 (-60 to -19 °C)
T (°C) ) 94.18 (∆δ) + 4.42
and Me3Si carbons (∆δ ) δ(CH) - δ(CH3)) to be used as an
effective internal 13C chemical shift thermometer with many
advantages over conventional tube-substitution methods.4b We
now routinely add approximately 0.1% by volume of 10% 13C
enriched 1-H to NMR samples for convenient temperature
measurement simply by running the 13C NMR spectrum. The
temperature dependence of the chemical shift (slope) varies only
slightly with solvent, but the calibration line (intercept) is offset
for each solvent. Since our initial report and in the course of
this and related projects we have constructed additional calibra-
tion curves for a number of solvents of interest in organometallic
chemistry. Figure 15 shows the new calibrations, and Table 1
shows the equations used to calculate chemical shifts, both for
the original calibrations4b and the new data. The data for HMPA
were not measured on a pure solvent, but were calculated from
the calibrations for two HMPA/THF mixtures, by assuming
additivity of effects.
A RINMR study of the lithium-iodine exchange reaction of
1-I at -130 °C showed that the main first detected product on
treatment of 1-I with n-BuLi was the iodine ate complex 4,
formed by reaction of 1-Li with 1-I. It could be shown that the
primary product of the lithium-iodine exchange was 1S, and
not 1C, so conversion of presumed intermediate ate complex
-
Li+ Bu-I-C(SiMe3)3 to the lithium reagent proceeds by an
SE1 mechanism, and not by electrophilic attack of lithium cation
on the ate complex (SE2 mechanism), which would give 1C as
the primary product.
Experimental Section
It was shown that the calibration curves for mixed solvents
can be estimated from a weighted average of those of the pure
solvents,4b and we have made extensive use of this technique.
However, such temperatures are likely to be less accurate
because of deviations from ideal mixing behavior. Thus the
General. All reactions requiring a dry atmosphere were per-
formed in glassware flame-dried or dried overnight in a 110 °C
oven, sealed with septa and flushed with dry N2. Tetrahydrofuran
(THF) and diethyl ether (ether) were freshly distilled from sodium
benzophenone ketyl before use. Dimethyl ether (Me2O) was distilled
728 J. Org. Chem. Vol. 74, No. 2, 2009

Tris(trimethylsilyl)methyllithium and its Iodine Ate Complex
via cannula into the NMR tube from a graduated conical cylinder
at -78 °C containing n-BuLi (for drying). Glassware was placed
overnight in a 110 °C oven or flame-dried before purging with N2
to remove moisture. Common lithium reagents were handled with
septum and syringe-based techniques and were titrated using
n-propanol in THF with 1,10-phenanthroline as an indicator.35
spectra were observed through the decoupler channel tuned to
360.132 MHz. Spectra were obtained in nondeuterated ether
solvents with the spectrometer unlocked and were referenced
internally to the trimethylsilyl peak of Me3SiPh (δ 0.44). The
temperature of the RINMR sample was checked at the beginning
and the end of an experiment. For many experiments, the probe
temperature was raised a few degrees ca. 10 s prior to the injection.
This offsets the warming that results from the injection itself. The
size of the offset has been determined in a number of control cases4l
and depends on the injection volume. There is approximately a 1
°C temperature jump for a 0.05 mL injection.
1
NMR Spectroscopy. Routine H and 13C NMR spectra were
acquired on a 300 MHz spectrometer with CDCl3 as the solvent
and tetramethylsilane as the internal standard.
All multinuclear NMR experiments were performed in 10 mm
NMR tubes using a wide-bore AM-360 spectrometer at 360.15 MHz
(1H), 52.00 MHz (6Li), 139.96 MHz (7Li), 90.56 MHz (13C), 71.54
MHz (29Si), or 145.78 MHz (31P). 13C NMR spectra were referenced
internally to the C-O carbon of THF (δ 67.96), Et2O (δ 66.57) or
Me2O (δ 60.25), and Lorentzian multiplication (LB) of 2-3 Hz
Tris(trimethylsilyl)methyllithium (1-6Li) - Typical Proce-
dure for Preparation of 1-6Li from 1-SePh. To a dried, N2-purged,
10 mm NMR tube fitted with a septum and maintained under
positive N2 pressure were added 0.204 g (0.526 mmol) of 1-SePh,
7 µL of the 13C chemical shift thermometer4b (1:10 ratio of 13C-
labeled and unlabeled 1-H), 1.8 mL of THF, and 1.2 mL of Et2O.
The solution was cooled to -78 °C and 0.34 mL (0.524 mmol) of
1.54 M n-Bu6Li in pentane were added. The sample was then ready
for detailed, addition of cosolvents, or RINMR injections. Com-
prehensive details are provided in the Supporting Information.
6
7
was applied. Li and Li spectra were referenced externally to 0.3
M LiCl in MeOH (δ 0.00) or internally to Li+(HMPA)4 (δ -0.40).
31P NMR spectra were referenced externally to 1.0 M PPh3 in THF
(δ -6.00) or internally to free HMPA (δ 26.40). 7Li and 31P spectra
were generally transformed with Gaussian multiplication, with the
GB parameter set to the fractional duration of the FID and the LB
parameter set to -(digital resolution)/GB.
Acknowledgment. We thank the National Science Founda-
tion for financial support of this research (CHE-0074657, CHE-
0349635, CHE-0717954). The spectrometers are funded by the
NSF (CHF 8306121) and the National Institute of Health (NIH
1 S10 RR02388). We thank Dr. Joe Borst for the HMPA
titrations of the iodine and mercury ate complexes.
The lithium reagent samples were prepared in 10 mm thin-walled
NMR tubes which were oven-dried, fitted with a septum (9 mm
i.d.), and N2-flushed. Since nondeuterated solvents were used, the
spectrometer was run unlocked, and shimming was performed on
the 13C FID of C-3 of THF or other solvent peak. Temperatures
were measured using either a thermocouple submerged in a second
NMR tube containing the same solvent mixture or the internal 13
C
chemical shift thermometer (Me3Si)3CH.4b
Supporting Information Available: Experimental proce-
dures for the preparation of compounds and NMR samples, 1H,
RINMR Experiments. The apparatus and techniques for per-
forming RINMR experiments have been described previously.4l
Low-temperature RINMR experiments were performed on a 360
6
29Si, Li, 13C and 31P NMR spectra of NMR experiments not
presented in the main body of this paper. This material is
available free of charge via the Internet at http://pubs.acs.org.
1
MHz spectrometer with a 10 mm broadband probe. H RINMR
(35) Watson, S. C.; Eastham, J. F. J. Organomet. Chem. 1967, 9, 165–168.
JO802032D
J. Org. Chem. Vol. 74, No. 2, 2009 729