Pivaloylmetals (tBu-COM: M=Li, MgX, K) as equilibrium components

Article abstract of DOI:10.1002/chem.201102867

Short-lived pivaloylmetals, (H₃C)₃C-COM, were established as the reactive intermediates arising through thermal heterolytic expulsion of O=CtBu₂ from the overcrowded metal alkoxides tBuC(=O)-C(-OM)tBu₂ (M=MgX, Li, K). In all three cases, this fission step is counteracted by a faster return process, as shown through the trapping of tBu-COM by O=C(tBu)-C(CD₃)₃ with formation of the deuterated starting alkoxides. If generated in the absence of trapping agents, all three tBu-COM species "dimerize" to give the enediolates MO-C(tBu)=C(tBu)-OM along with O=CtBu₂ (2 equiv). A common-component rate depression by surplus O=CtBu₂ proves the existence of some free tBu-COM (separated from O=CtBu₂); but companion intermediates with the traits of an undissociated complex such as tBu-COM & O=CtBu₂ had to be postulated. The slow fission step generating tBu-COMgX in THF levels the overall rates of dimerization, ketone addition, and deuterium incorporation. Formed by much faster fission steps, both tBu-COLi and tBu-COK add very rapidly to ketones and dimerize somewhat slower (but still fairly fast, as shown through trapping of the emerging O=CtBu₂ by H₃CLi or PhCH₂K, respectively). At first sight surprisingly, the rapid fission, return, and dimerization steps combine to very slow overall decay rates of the precursor Li and K alkoxides in the absence of trapping agents: A detailed study revealed that the fast fission step, generating tBu-COLi in THF, is followed by a kinetic partitioning that is heavily biased toward return and against the product-forming dimerization. Both tBu-COLi and tBu-COK form tBu-CH=O with HN(SiMe₃)₃, but only tBu-COK is basic enough for being protonated by the precursor acyloin tBuC(=O)-C(-OH)tBu₂. Copyright

Full text of DOI:10.1002/chem.201102867

FULL PAPER
DOI: 10.1002/chem.201102867
Pivaloylmetals (tBu-COM: M=Li, MgX, K) as Equilibrium Components
R. Knorr,* G. Bçhrer, B. Schubert, and P. Bçhrer[a]
Dedicated to Professor Herbert Mayr on the occasion of his 65th birthday
Abstract: Short-lived pivaloylmetals,
(H3C)3C-COM, were established as the
reactive intermediates arising through
thermal heterolytic expulsion of
O=CtBu2 from the overcrowded metal
free tBu-COM (separated from O=
CtBu2); but companion intermediates
with the traits of an undissociated com-
plex such as tBu-COM & O=CtBu2
had to be postulated. The slow fission
step generating tBu-COMgX in THF
levels the overall rates of dimerization,
ketone addition, and deuterium incor-
poration. Formed by much faster fis-
sion steps, both tBu-COLi and tBu-
COK add very rapidly to ketones and
dimerize somewhat slower (but still
fairly fast, as shown through trapping
of the emerging O=CtBu2 by H3CLi
or PhCH2K, respectively). At first sight
surprisingly, the rapid fission, return,
and dimerization steps combine to very
slow overall decay rates of the precur-
sor Li and K alkoxides in the absence
of trapping agents: A detailed study re-
vealed that the fast fission step, gener-
ating tBu-COLi in THF, is followed by
a kinetic partitioning that is heavily
biased toward return and against the
product-forming dimerization. Both
tBu-COLi and tBu-COK form tBu-
alkoxides
tBuC(=O)-CACTHNUGTRENNU(G -OM)tBu2
(M=MgX, Li, K). In all three cases,
this fission step is counteracted by
a
faster return process, as shown
through the trapping of tBu-COM by
O=C(tBu)-C(CD3)3 with formation of
N
ACHTUNGTRENNUNG
the deuterated starting alkoxides. If
generated in the absence of trapping
agents, all three tBu-COM species “di-
merize” to give the enediolates MO-C-
CH=O with HNACTHNUGTRENUNG(SiMe3)3, but only
tBu-COK is basic enough for being
protonated by the precursor acyloin
tBuC(=O)-CACTHNUGTRENNU(G -OH)tBu2.
Keywords: acyl anions · cleavage
reactions · reaction mechanisms ·
A
U
A
reactive intermediates
molecules
· strained
nent rate depression by surplus O=
CtBu2 proves the existence of some
Introduction
tBuLi will typically compete with tBu-COLi for these elec-
trophiles and thus may furnish unwelcome side-products as
shown. Similar problems were encountered when tBu-COLi
was generated[5] from tBuC(=O)-Te-nBu with n-butyllithi-
um. The simplest method[6] of generating tBu-COLi by de-
protonation of pivalaldehyde (tBu-CH=O) with lithium
2,2,6,6-tetramethylpiperidide (LiTMP) will probably meet
difficulties when striving for reactions of tBu-COLi other
than the addition to unreacted tBu-CH=O.
Pivaloyllithium (tBu-COLi in Scheme 1) is thought to be the
unstable reagent that can be generated[1,2] by flooding tert-
We have discovered and describe here a cleaner way of
creating pivaloylmagnesium, -lithium, and -potassium under
conditions that permit qualitative kinetic characterizations
of these transient intermediates, even though their more
direct observation (using NMR spectroscopy) has still not
become possible.[7]
Scheme 1. Traditional generation and trapping of pivaloyllithium.
butyllithium (tBuLi) with gaseous carbon monoxide; it must
be handled at below ꢀ1008C. Related reagents[3] were pre-
pared in a similar way that was shown[4] to involve electron
transfer steps. While tBu-COLi can efficiently be trapped by
electrophiles such as ketones[1,2] (Scheme 1), the parent
Results and Discussion
Pivaloylmagnesium: The deprotonation (Scheme 2) of tri-
tert-butylacyloin[8–10] (1) by (tert-butyl- or) methylmagnesium
chloride in tetrahydrofuran (THF) produced two structural-
ly undefined Mg alkoxides 5 (detected through 1H NMR
spectroscopy in situ) which approached their equilibrium
ratio (ca. 75:25) with a first half-life time (t1/2) of roughly
8 min at room temperature.[11] Both alkoxides 5 could coex-
[a] Prof. R. Knorr, Dr. G. Bçhrer, Dr. B. Schubert, P. Bçhrer
Department Chemie der Ludwig-Maximilians-Universitꢀt
Butenandtstrasse 5–13 (Haus F), 81377 Mꢁnchen (Germany)
E-mail: rhk@cup.uni-muenchen.de
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.201102867.
Chem. Eur. J. 2012, 00, 0 – 0
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
1
&
ÞÞ
These are not the final page numbers!

existence of an “external return” process[18,19] from 6 or 6&7
to 5 was established through the following experiment.
(CD3)3 ([D9]7, 0.3 equiv)[20]
The presence of O=CACTHNUTRGENN(UG tBu)-CAHCUTNGTRENNGUN
in a THF solution of 5 decaying at 608C led to deuterium
incorporation in the recovered portion of 5 (Scheme 3):
Scheme 2. Novel generation and trapping of pivaloylmagnesium: see
Ref. [16] for the meaning of &. X=Cl, -C(O)-tBu.
Scheme 3. Deuterium incorporation into the starting material 5 affords
11a (=[D9]5) through “external return” of pivaloylmagnesium (6): see
Ref. [16] for the meaning of &.
1
ist with 1 without showing indications of H NMR coales-
cence, and they slowly released di-tert-butyl ketone (O=
CtBu2, 7) at 608C (“fission” in Scheme 2). An excess
(3.3 equiv) of the deprotonating agent H3C-MgCl did not ac-
celerate this decay of 5 but reacted slowly with the emerging
O=CtBu2 rather than with the keto function of 5, affording
the known[12,13] alcohol 10; a control experiment[11] con-
Upon termination by acidification after 25 min (25% con-
version to pivaloin 3), the deuterium analyses[21] both in situ
(experiment 2a in Ref. [22]) and after aqueous workup (ex-
periment 2b in Ref. [22]) revealed 10% of 12a and roughly
5% of 12b. Obviously, the intermediate(s) had been trapped
by [D9]7 through external return to give first the alkoxide
11a and then 11b by tBu migration.[23] This result shows that
“external return” can compete with the “dimerization” step
in Scheme 2; but it does not provide an estimate of the rate
ratio of “fission” over “return”. However, “external return”
should surely be somewhat slower than the above additions
of 6 to less bulky ketones, which were shown above to be
hardly faster than the “dimerization” step. Therefore,
“return” of 7 to 5, being kinetically equivalent with the “ex-
ternal return” of [D9]7 (Scheme 3) except for a tiny isotope
effect, can be expected to take place roughly as fast as (or
slower than) the “dimerization” step. Both “return” and “di-
merization” should then be faster than the “fission” step in
view of the low steady-state concentrations of 6 and/or 6&7.
Free O=CtBu2 (7, added or emerging as the byproduct)
should accelerate the return of free pivaloylmagnesium (6)
to 5 (or first to 6&7), so that the overall decay rate of 5 to
give the enediolate 2 might decrease. The detection of this
common-component[24] (or “mass-law”)[25] rate depression
would provide unambiguous[18] evidence for 6 as a free inter-
mediate. In fact, the overall decay rate of 5 became roughly
halved (second[14] t1/2 ca. 95 min) in the presence of excess
O=CtBu2 (3.6 equiv), showing[18] that the proposed inter-
mediate 6 was generated at least partially as a free species
separated from 7. (This does not exclude the presence of
a second reactive[19] intermediate, such as a complex 6&7,
which cannot be repressed by free O=CtBu2.) A rather
modest common-component retardation appeared to show
up also in cases without O=CtBu2 addition, when the very
firmed the surprisingly sluggish formation of 10 (first t1/2
=
19 min at 608C) from 7 with H3C-MgCl (2.9 equiv) in THF.
The remaining fragment pivaloylmagnesium (6) is thought
to “dimerize” with formation of the enediolate 2, which
1
could not be detected by H NMR in situ but was identified
through protonation or deuterolysis (with H3COD), produc-
ing pivaloin (3) or [4D]pivaloin (4), respectively. The ap-
proximately measurable overall decay rate (corresponding
to a second[14] t1/2 of ca. 45 min at 608C) of the two alkoxides
5 was hardly increased (less than doubled) in the presence
of ketones (Scheme 2), affording small portions of pivaloin
(3) upon aqueous workup and mainly the adducts 8 from
adamantan-2-one (0.8 equiv), 9a from benzophenone
(1.6 equiv), or 9b[15] from pivalophenone (2.0 equiv). The
observed product mixtures indicated that the product-form-
ing steps from 6 to 8, 9a, and 9b are hardly faster than the
“dimerization” step to give 3 (from 2). Moreover, the over-
all reaction rates of these three ketones are hardly greater
than that of the “dimerizing” decay of the Mg alkoxides 5,
which suggested that the proposed pre-equilibrium, generat-
ing 6 from 5, provided a common upper limit for these simi-
lar overall rate values. All of these trapping modes may be
envisioned as operating either on the separated species 6+7
or on a transient complex 6&7 (which is postulated here in
anticipation of the evidence presented in Section B).[16] Nei-
ther of these two intermediates was detected through NMR
in situ; therefore, they appeared to be present in steady-
state concentrations below the NMR detection level due to
their rapid consumption by “dimerization” and/or by a fast
back reaction such as the “return” step in Scheme 2.[17] The
&
2
&
www.chemeurj.org
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 0000, 00, 0 – 0
ÝÝ
These are not the final page numbers!

Pivaloylmetals as Equilibrium Components
FULL PAPER
first decay phase could be observed: Due to the initially tiny
O=CtBu2 concentrations (hence initially insignificant back
reaction), the overall decay rate of 5 at 608C often seemed
to be abnormally high (albeit not great enough for more
precise measurements at room temperature) and dropped
soon with the production of O=CtBu2, so that the first t1/2
values generally became ill-defined and the second or
later[26] t1/2 data[14] had to be used for semiquantitative speci-
fications.
The results of this section support the mechanistic propos-
al of Scheme 2, with the rate sequencing of “return”ꢁ“di-
merization”>“fission” during a proceeding decay of the Mg
alkoxides 5.
practically finished within 4 min at 218C and furnished 12a
(31%), 12b (7%), and 12c (3%) upon acidification (experi-
ments 3a,b in Ref. [22]). Obviously, the deuterated Li alkox-
ides 15a and 15b were generated through “external return”
with subsequent tBu migration,[23] followed by the fission of
15b to give LiOC-CACTHNUTRGNEUN(G CD3)3 ([D9]13) and a subsequent
second external return that converted [D9]13 into 15c. In
terms of the reaction sequence proposed in Scheme 4, these
“external return” events generating 15a and 15c are mecha-
nistically equivalent with return of 13 to 14 (except for
a tiny isotope effect), so that all of these “return” processes
must be faster (because successful already at 218C) than the
“dimerization” step forming enediolate 2. In addition, these
return processes are also substantially faster than the “fis-
sion” step of 14, because 13 and/or 13&7 (both undetected)
do not accumulate.[17] Therefore, the fast “fission” step
limits the rate of deuterium incorporation and should
become a common kinetic “bottleneck” in the cases of other
reagents (Scheme 5) which capture 7 or 13 with significantly
higher rates than those of the “dimerization” and “return”
processes, as explored in the sequel.
Pivaloyllithium: Two structurally undefined Li alkoxides 14
(population ratio ca. 20:80 by 1H NMR in situ) resulted
from deprotonation (Scheme 4) of tri-tert-butyl acyloin[8,9,10]
Scheme 4. Decay of 14, and deuterium incorporation through “external
return” of pivalolyllithium (13): see Ref. [16] for the meaning of &.
(1) in THF by methyllithium, whereas only one kind of Li
alkoxide 14 was formed from 1 in hydrocarbon solution with
n-butyllithium (nBuLi).[11] The cleavage of these alkoxides
(“fission” in Scheme 4) will generate pivaloyllithium[7] (13)
or its complex 13&7 with O=CtBu2 (7). In analogy with
Scheme 2, the entailing “dimerization” is thought to form
the NMR-silent enediolate 2 and therefrom (after acidifica-
tion) pivaloin (3, 0.5 equiv) along with O=CtBu2 (7).
Scheme 5. Trapping of the fission products 7 and 13 of Li alkoxide 14:
see Ref. [16] for the meaning of &.
Crude estimates of the second[14,26] t1/2 values for this over-
all decay process, starting with comparable initial concentra-
tions (typically 0.2–0.3m), were found to be 15 min in THF
at 608C but 105 h at 218C, and in benzene solution 44 min
at 608C but roughly 14 h at 218C. In striking contrast to this
The Li alkoxide 14 vanished in the presence of benzophe-
none (1.9 equiv) in THF at 218C (9a isolated after aqueous
workup) with a second[14,26] t1/2 of ca. 3 min, to be compared
with the above 105 h in the absence of trapping agents;
a similar run with benzophenone in benzene solution also
was complete in <12 min. A mixture of 14 and its parent
substance 1 (ca. 1:9), generated with only 0.1 equiv of
sluggish decay, the very fast incorporation of O=CCAHTUNGTRENNUNG
[20]
ACHTUNGTRENNUNG
Chem. Eur. J. 2012, 00, 0 – 0
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
&
3
&
ÞÞ
These are not the final page numbers!

Rudolf Knorr et al.
H3CLi, produced very little O=CtBu2 (7) at 218C within
4.4 h, showing that the acyloin 1 did not intercept 13 or
13&7 by protonation (no acceleration of the overall decay).
However, the subsequent addition of adamantan-2-one
(1 equiv) completed the consumption of both 14 and
1 within <16 min, affording the adduct 8 after aqueous
workup; this implies fast proton transfer from 1 to the prod-
uct alkoxide of 8, so that more of 14 became available for
continuing the decay process. Di-isopropyl ketone (2 equiv)
consumed 14 in benzene solution at 298C with a second[14,26]
t1/2 of ca. 6 min; in the solvent THF, this reaction did not
occur at ꢀ458C but proceeded as usual at 208C, furnishing
the adduct 16 after aqueous workup. However, cyclohexa-
none did not capture 13 in benzene or in THF, presumably
due to fast proton transfer to 14 with re-generation of 1, al-
though the expected ketone adduct had been obtained[1]
with 13 as generated from tBuLi and carbon monoxide at
ꢀ1108C. This failure of our system dictated the choice of
two small, non-enolizable ketones as trapping agents: dicy-
clopropyl ketone (1 equiv) and Li alkoxide 14 in pentane or
in THF afforded O=CtBu2 and the adduct 17a after aque-
ous workup, providing no evidence of ring-opened products
that would have been expected from a radical mechanism.
The very slow decay of 14 in benzene at 208C was accelerat-
ed to a time-domain[14] of about 2 min through the injection
of bis(a-methylcyclopropyl) ketone[27] (1.2 equiv) whose
adduct 17b to 13 was obtained as the only product along
with O=CtBu2 after aqueous workup. A further simple and
instructive mode of suppressing the “return” reaction uti-
lized the rapid removal of O=CtBu2 (7): Li alkoxide 14,
prepared from 1 at ꢀ708C with an excess of methyllithium
in THF and warmed quickly to 238C, vanished through trap-
ping of 7 by residual H3CLi (2.1 equiv) with a third[14,26] t1/2
of ca. 4 min; the subsequent acidification revealed the for-
mation of adduct 10 and pivaloin (3) in situ. Since the piva-
loyllithium (13) left behind on consumption of 7 cannot
return in the absence of 7 but nevertheless still does not ac-
cumulate, this experiment confirms that the “dimerization”
step toward 2 and 3 can proceed faster (as if stimulated by
the interception of 7) than the “fission” of 14. In practice,
any excess over the stoichiometric amount of H3CLi at am-
bient temperature should be avoided through careful titra-
tion of the acyloin 1 with H3CLi (measuring CH4 in a gas
burette, for instance) at or below ꢀ408C, lest the rapid for-
mation of the alkoxide 18 of adduct 10 consume 14, frustrat-
ing a different application of 13. Besides, the conceivable
possibility (Scheme 5) of a cleavage of 18 in THF to give
tBuLi was excluded through prolonged warming of an au-
thentic sample of 18 which remained intact at 558C for
more than 90 min.
lows. The Li alkoxide 14 was prepared from 1 with H3CLi
(3.0 equiv) with magnetic stirring at ꢀ408C (CH4 evolution)
and then warmed quickly up to 38C under dry argon gas.
Three samples were withdrawn after reaction periods of 2,
12, and 240 min and quenched with iced HCl (1m). Extrac-
tions with Et2O furnished three clean product mixtures (no
O=CtBu2, no tBuCH=O) of starting material 1, product
10, and pivaloin (3) in the molar ratios of 97:3:1, 94:6:4, and
0:65:35, respectively, all in acceptable agreement with the
expected 10/3 ratio of 2:1. This low reaction rate at 38C ap-
pears to be compatible with a time domain (and t1/2) nar-
rowed down to approximately 3 min at ambient temperature
for the “fission” step as a kinetic “bottleneck” of the overall
decay rates. Reverting to the very much slower overall
decay (t1/2 ca. 105 h) with “dimerization” (upper part of
Scheme 4), this attenuation by a factor of roughly 3 min/
105 hꢁ1/2000 in the absence of efficient trapping agents
(such as ketones) suggests that the rapid initial “fission”
event of 14 is followed by a 1/2000 partitioning between “di-
merization” and “return”.[28] This information does not
define absolute time domains for “dimerization” and
“return”, both of which were classified above as being sub-
stantially faster than the “fission” step.
If the complex 13&7 were the only “dimerizing” inter-
mediate that produces the enediolate 2, the overall decay
rate would not depend on the concentration of free O=
CtBu2 (7). On the other hand, the free intermediate 13 (sep-
arated from 7) alone would be recognized through an in-
verse linear dependence of the overall decay rate on the
concentration of free 7 under the present condition of
a mobile pre-equilibrium (“return” being faster than “fis-
sion” and “dimerization”). However, the common-compo-
nent[24] rate depression expected for this case was again
found to be rather small: In the presence of a substantial
excess (3.8 equiv) of O=CtBu2, the overall decay rate of 14
was only halved with a second[14,26] t1/2 of ca. 30 min THF at
608C. This rate depression provides unambiguous[18] evi-
dence for at least some free pivaloyllithium (13); its low
magnitude points to the existence of a second pathway that
consumes 14 and can generate the same products as free 13
but cannot be retarded by free O=CtBu2 (7). This suggests
that both 13 and 13&7 may indeed be present as transient
intermediates (corresponding to the kinetic system[18] of
a free ion and its ion pair). The surprisingly similar decay
rates of 14 in both THF and hydrocarbons may have to do
with a temporary stabilization of 13 in 13&7 by the interac-
tion of the Li center with O=CtBu2 in either solvent.
Pivaloyllithium (13) is a weaker base than LiTMP (pKa =
37.0 for TMP in THF[29]), as shown by the deprotonation[6]
of pivalaldehyde (tBuCH=O) with LiTMP mentioned in
the Introduction. Thus, a pKa value in the range of 27–36
can be expected for the aldehyde proton in tBuCH=O, be-
cause the following experiment shows that 13 is a stronger
All of the above crude estimates of time domains agree
with an upper limit of few minutes (or a corresponding
lower limit of the reaction rates) for the observed overall
decay, pointing to a rate-limiting “fission” step of the Li alk-
oxide 14 at room temperature; but considerably higher over-
all decay rates could not always be excluded by such crude
timing. However, the latter possibility was dismissed as fol-
base than LiN
ACHTUNGTERN(NUNG SiMe3)2 (LiHMDS) in THF. The injection of
HN
AHCTUNGTRENNUNG
a very slowly decaying THF solution of the two Li alkoxides
14 initiated a vivid consumption of 14 (95% conversion
&
4
&
www.chemeurj.org
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 0000, 00, 0 – 0
ÝÝ
These are not the final page numbers!

Pivaloylmetals as Equilibrium Components
FULL PAPER
imine 19 (0.23 equiv), supplemented by the adduct
8
(0.51 equiv) and a little[33] pivaloin (3) but no tBuCH=O
(20) after acidification. Thus, the 19/8 ratio of 0.45 indicates
that both HMDS and adamantan-2-one intercepted pivaloyl-
lithium (13) with similar efficiency and that LiHMDS is,
therefore, not always recommendable for generating 13.
The rate sequencing as deduced in this section for the ki-
netic setup in Scheme 4 and Scheme 5 is “return”@ “dimeri-
zation”> “fission” during a proceeding decay of the Li alk-
oxides 14.
Scheme 6. Protonation of pivaloyllithium (13) by HNACTHNUGTRNEUNG(SiMe3)2 (HMDS):
see Ref. [16] for the meaning of &.
Pivaloylpotassium: Benzylpotassium (PhCH2K, Scheme 7) is
a suitable base for generating pivaloylpotassium (23) from
during 20 min at 238C), affording (Scheme 6) O=CtBu2 (7)
[11,31]
and tBuCH=N-SiMe3
(19) as the only derivatives of 14
1
observed through H NMR in situ. Evidently, the fast proto-
nation of 13 by HMDS generated tBuCH=O (20) and the
weaker base LiHMDS; a separate experiment confirmed
that these two primary products furnished the imine 19
quickly. The latter condensation reaction to give 19 is faster
than the fast deprotonation of acyloin 1 (Scheme 6) by
LiHMDS: A catalytic amount of LiHMDS (0.2 equiv) in
THF containing HMDS (1.1 equiv) initiated a moderately
accelerated consumption of 1 (67% conversion in 90 min),
producing O=CtBu2 (7) and tBuCH=O (20) along with
some short-lived imine 19, but neither the pivaloin alkoxide
21 nor pivaloin (3). Recalling the hesitance of acyloin 1 to
protonate 13, it is clear that 13 became protonated by the
acid HMDS with regeneration of LiHMDS, so that the over-
all decay from 1 to 7 and tBuCH=O (20) was temporarily
catalyzed by LiHMDS. However, this catalyst was soon
poisoned through its faster condensation with the accumu-
lating 20 to give 19, such that the initially fast overall decay
of 1 discontinued at 70% conversion. Obviously, the remain-
ing bases LiOSiMe3 and HMDS were unsuited for deproto-
nating the last 30% of 1. The interception of 13 by HMDS
was fast enough to prevent 13 from adding to the accumu-
lated 20 with formation of the pivaloin alkoxide 21, although
this addition is known to occur readily[1] even at ꢀ1358C
and hence should be quite fast at 238C. Besides, small
amounts of tBu-CH2OH and the Tishchenko ester tBuCO2-
CH2tBu[32] were observed in situ after several d at 208C.
With a greater portion of LiHMDS (1.2 equiv) and less
HMDS (0.55 equiv), the faster consumption of acyloin
1 (85% conversion in 5 min at 218C) furnished only O=
CtBu2 (7) and the imine[11,31] 19 in place of 20, but again no
pivaloin alkoxide 21; separate runs established that
LiHMDS does not deprotonate 21 (to generate enediolate
2) in THF at 218C. For a somewhat sharper estimate of the
rapidity of trapping 13 with HMDS, the acyloin 1 was added
Scheme 7. Formation and decay of the potassium alkoxide 22: see
Ref. [16] for the meaning of &.
the acyloin 1: Its preparation[34] from KOtBu, toluene, and
nBuLi in saturated hydrocarbons at room temperature is
easy and safe; the brick-red PhCH2K precipitates from the
hydrocarbon solution, can be purified through washing with
pentane or cyclopentane, and is stable as a powder in an
inert atmosphere. Its dark red solution in anhydrous THF is
relatively unstable above ꢀ208C but will partially survive
1
a rapidly performed H NMR[35] analysis at ambient temper-
ature under argon cover gas. Its titration at ꢀ708C by
adding small batches of the acyloin 1 up to decolorization
afforded toluene and the K alkoxide[11] 22 which may be
a rapidly (on the NMR time scale) interconverting pair of
alkoxides (in contrast to the slow interconversion of the Li
alkoxides 14 as described in Section B). Observed in a tightly
closed NMR tube under argon gas, 22 decayed very slowly
(second[14,26] t1/2 ca. 8 h) at 208C with formation of O=CtBu2
(7); no other product became visible until an acidification
disclosed pivaloin (3, 0.51 equiv relative to 7 as required),
which in this case can derive only from the NMR-silent ene-
to
a THF solution of HMDS (0.34 equiv), LiHMDS
(1.2 equiv), and adamantan-2-one (2.4 equiv) at ꢀ788C. The
1H NMR spectrum in situ, recorded 4 min after a quick
warm-up, showed residual Li alkoxide 14 (only the major
component with dH =1.01 and 1.28 ppm, as always with
LiHMDS as the base), O=CtBu2 (50% conversion), and
Chem. Eur. J. 2012, 00, 0 – 0
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
&
5
&
ÞÞ
These are not the final page numbers!

Rudolf Knorr et al.
diolate 2 as formed through “dimerization” of pivaloylpotas-
sium (23) in analogy with the behavior of pivaloyllithium
(13), except for the new side-product[11,36] 26 (0.07 equiv) as
generated from pivalaldehyde (20) with PhCH2K. At 608C,
22 decayed again much faster (within 1 h) than the Li alkox-
ide 14, but all further runs with 22 were conducted at or
below ambient temperature.
Most of the properties of the system 22/23 can be under-
stood on the basis of lessons learned in Section B: In the
presence of surplus (1.0 equiv) benzylpotassium, the decay
of K alkoxide 22 ran down with a dramatically shortened
second[14,26] t1/2 of 4 min (instead of 8 h) at 208C. NMR spec-
tra in situ revealed that the K alkoxide[11] 25 was present in
place of O=CtBu2 (7) whereas pivaloin (3) and its K alkox-
ide[11] 24 were absent; clearly, the rapid conversion of 7 to
25 had overcome the “return” process of pivaloylpotassium
(23) with 7 toward 22 and thus brought the fast “fission”
step of 22 to 23 into focus. After aqueous workup, the main
product[11,37] 27 was accompanied by pivaloin (3,
0.38 equiv)[33] and by the adduct 26 (0.32 equiv)[11] of
PhCH2K to 20. These products indicate that pivaloylpotassi-
um (23), being restrained from returning to 22, had rapidly
ture by fast deprotonation of 1, so to avoid the side-reac-
tions described above. The suitable base PhCH2K may not
be replaced by KH.11 Similar accelerations up to the limiting
rate of the common “fission” step (the “kinetic bottle-
neck”) were found with two ketones as trapping agents for
23 or 23&7: The addition of benzophenone (Ph2C=O,
0.4 equiv) accelerated a slowly decaying solution of 22 in
THF and consumed an equivalent amount of 22 in less than
6 min at 228C,[11] yielding the adduct 9a after aqueous
workup. Similarly, pivalophenone (tBuCO-Ph, 1.0 equiv)
consumed 22 in THF at 208C within <9 min, affording O=
CtBu2 (7) together with the K alkoxide[11] of 9b and there-
from after aqueous workup the pure adduct 9b (85% yield).
These time domains[14] agree with the above 4 min estimate
for the “fission” step, whereas the actual addition rates of 23
to the ketones were certainly much higher: Namely, these
rates should be higher than that of the corresponding addi-
tion of 23 to the more shielded (hence less reactive) O=
CtBu2 (7) which is the “return” reaction to 22 and must al-
ready be much faster than the “fission” step, since 23 did
not accumulate.[17] A direct visualization of this “return”
step and the limiting “bottle-neck” rate was again achieved
with the isotopologue[20] [D9]7 of O=CtBu2 as follows. A
slowly decaying THF solution of 22 containing just only 3%
of 7 was mixed with [D9]7 (0.6 equiv) at ꢀ708C, then
warmed up quickly to 238C and quenched after 6 min with
“dimerized” to generate the NMR-silent enediolate
2
ACHTUNGTRENNUNG
ciency to furnish 20 that formed the K alkoxide of 26 with
surplus PhCH2K. In this strongly basic milieu,[38] a portion
of the initially present acyloin 1 must have served as the
only available acid for proton transfer to the rapidly emerg-
ing pivaloylpotassium (23) in preference to protonating the
enediolate 2 or the more basic PhCH2K. This fast proton
transfer from 1 to 23 constitutes a prominent difference of
pivaloylpotassium (23) from pivaloyllithium (13) and was
1
F3C-CO2H. Upon aqueous workup, H and 13C analyses[21]
revealed the production of additional O=CtBu2 (total
0.37 equiv) along with the deuterated starting material as
formed through the following reaction sequence (Scheme 8):
confirmed as follows. Employing
a
substoichiometric
amount of PhCH2K (0.27 equiv only), the overall decay of
K alkoxide 22 was again accelerated rather than slowed
down: After the initial consumption of PhCH2K by 1 had
furnished 22 and plenty of O=CtBu2 (but no 25) in the first
4 min (immediate decolorization), the decay of surplus
1 was fast (96% conversion in 25 min) at 218C due to the
partial suppression of the “return” reaction through trap-
ping of 23 by proton transfer from 1 with formation of piva-
laldehyde (20). However, 20 could not be detected by
1H NMR in situ because of its interception by 23 (instead of
PhCH2K) to give the K alkoxide 24 and because of the con-
comitant transformation of 20 into the Tishchenko ester[32]
tBuCO2-CH2tBu which began to emerge already after
12 min. In this absence of detectable amounts of 20, 24 is
a diagnostic in-situ indicator of the partial protonation of 23
to give 20, as opposed to the “dimerization” of 23 affording
the NMR-silent enediolate 2 (which was shown above not to
be protonated by 1); on aqueous workup, however, both 2
and 24 (Scheme 7) are converted to pivaloin (3), of course.
The unpleasant high decay rate of the K alkoxide 22 caused
by surplus amounts of either acyloin 1 (protonating 23) or
PhCH2K (trapping 7) is limited by the initial “fission step”
of 22 which could be frozen by cooling to ꢀ278C. Therefore,
preparations of 22 should be performed at a lower tempera-
Scheme 8. Trapping of pivaloylpotassium (23), and deuterium incorpora-
tion into the starting material (28a=[D9]22) through “external return” of
23: see Ref. [16] for the meaning of &.
&
6
&
www.chemeurj.org
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 0000, 00, 0 – 0
ÝÝ
These are not the final page numbers!

Pivaloylmetals as Equilibrium Components
FULL PAPER
Capture of 23 by [D9]7 to give the labeled K alkoxide 28a
(converted into 36% of 12a through acidification), tBu mi-
gration[23] in 28a to produce 28b (!10% of 12b), fission of
28b to generate [D9]23 whose capture by [D9]7 afforded 28c
(!6% of 12c), and (D3C)3C migration in 28c to produce
28d (!3% of 12d), as analyzed in experiments no. 4–6 in
Ref. [22] Clearly, this deuterium incorporation occurred
again in a time domain[14] of some minutes as imposed by
the “fission bottle-neck” from 22 to 23. Considering that
both free tBu-COK (23 separated from 7) and its complex
23&7 might take part in the D incorporation, we sought un-
ambiguous[18] evidence for at least some free 23 through an
assessment of the common-component[24] rate depression of
22: The overall decay rate of 22 became strongly reduced in
the presence of free O=CtBu2 (7, 2.0 equiv) and was reani-
mated through subsequent heating at 608C, which led to
completion of the run within 55 min. Unfortunately, the
qualitatively obvious rate depression could not be quantified
due to the poor numerical reproducibility as caused by the
above-mentioned strong acceleration effects of varying 1/
PhCH2K ratios. It appeared probable that free tBu-COK
(23) contributes with a substantial portion, while a contribu-
tion of the undissociated complex 23&7 could not be exclud-
ed.
ed from O=CtBu2) but apparently also as activated species
exhibiting the traits expected for undissociated complexes
such as tBu-COM & O=CtBu2.[16] The latter formulation
disregards the possibility of molecular aggregation that is
quite usual for polar organometallics (and perhaps also fa-
vorable in the unknown course of the “dimerization” step
producing the enediolate 2).
The precursor alkoxide 5 of tBu-COMgX (6) is most
easily prepared because it is stable at room temperature in
THF, presumably owing to its slow initial fission step. How-
ever, the subsequent “dimerization” of 6 can successfully
compete with the addition to ketones, so that product mix-
tures may be obtained. The precursor alkoxide 14 of piva-
loyllithium (13) is fairly stable at 238C in the absence of sur-
plus organolithium compounds that can capture O=CtBu2
from the mobile pre-equilibrium. The fast “fission” step of
14 is slower than the “dimerization” and “return” steps of
13, although its high rate is not immediately evident from
the very sluggish overall decay of 14. In fact, “fission” was
recognized as the common fast “bottle-neck” step: All over-
all rates which involve ensuing faster reactions of 13 are lev-
eled to a common time domain of several minutes. The K
alkoxide 22 decays most readily, compared with 5 and 14,
but still with a very low overall rate that appears to hide
a similar kinetic system as above with fast “fission” and
faster “return” and “dimerization” steps. The equilibrium
component tBu-COK (23) is more reactive than tBu-COLi
(13) in deprotonating the acyloin 1; this property and the
somewhat complicate PhCH2K technique as required for
preparing the precursor 22 impair the practicability of 23.
Considering the various kinetic effects and reaction condi-
tions reported in this work, a rate sequencing as controlled
by the metal cations during the proceeding decay reactions
appears to be K >Li @Mg for the “fission” step. For all
three cations, a step rate gradation of “return”>“dimeriza-
tion”>“fission” was inferred.
Although the present method of generating polar acylme-
tals through alkoxide fission is restricted to the salts of acy-
loin 1,[23] this singular system provided novel knowledge
about reactivity patterns of this type of short-lived[42] inter-
mediates. This may have become possible because tBu-
COLi and tBu-COK are “domesticated” through frequently
repeated return events, as mediated by the maintained con-
tact to O=CtBu2 in tBu-COK & O=CtBu2 or by the fast
recapturing the free pivaloyl species with free O=CtBu2, so
that the overall “dimerizing” decay reactions were strongly
impeded. As a simpler alternative mechanism, the direct
(SE2-type) attack of the electrophilic reagents on the labile
Summing up, the kinetic assessments of 23ACTHNURGTNE(NUG & or +)7 were
aggravated by the adverse kinetic properties of this system
which tends to disturbances through trapping reactions by
both surplus PhCH2K and the parent acyloin 1 at room tem-
perature. The “return” and “dimerization” steps are sub-
stantially faster than the “fission” step (generating 23), so
that “fission” is a common kinetic “bottle-neck” that levels
the overall rates of deuterium incorporation and of the
other perturbations of the pre-equilibrium to a common
time domain of a few minutes. The rate sequencing of elec-
trophiles for trapping 23 appears to be tBuCH=O (20) >
protonation by acyloin 1 >adamantan-2-one >O=CtBu2
(7). The results of this section support the mechanistic pro-
posal of Scheme 7, with the rate sequencing of “return” >
“dimerization” >“fission” during a proceeding decay of the
K alkoxides 22.
Conclusion
The acyloin 1 was considered[39] for some time as an anti-
knock additive to Diesel fuels, which may suggest a propensi-
ty toward radical reactions. However, the presently investi-
gated metal alkoxides of 1 do not decay by a radical path-
way (not even in a perceptible side-reaction) which would
generate the thermodynamically stable ketyl radical-anion
of O=CtBu2 along with the kinetically unstable pivaloyl
radical (tBu-CO·): The evolution of carbon monoxide ex-
pected from the fast[40,41] decay of tBu-CO· was never ob-
served here. Instead, our pivaloylmetals are the first acylme-
tals that were generated through the heterolytic thermal
cleavage of metal alkoxides. These short-lived acylmetals
are formed not only as the free particles tBu-COM (separat-
alkoxides tBuC(=O)-CACHTUNTGRENUNG(-OM)tBu2 causing a concerted tBu-
COM transfer (no intermediates?) would not only be hard
to believe but must also be dismissed: The sterically quite
dissimilar ketone reagents [from dicyclopropyl ketone to
O=CACHTUNTRGENNUG(tBu)-CCAHTUNTGRNE(NUGN CD3)3] should then react with widely differing
rates, whereas their rates were found to be leveled to the
rate of the common initial “bottle-neck” fission that does
not involve the ketones.
Chem. Eur. J. 2012, 00, 0 – 0
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
&
7
&
ÞÞ
These are not the final page numbers!

Rudolf Knorr et al.
7.34 ppm (m, 10H); 13C NMR (CDCl3, 100.6 MHz): d=27.6 (q, CMe3),
45.9 (s, CMe3), 88.2 (s, COH), 127.6 and 128.2 (2 d, 4 o-C and 4 m-C),
127.8 (d, 2 p-C), 143.2 (s, 2 i-C), 215.0 ppm (s, C=O), assigned through
HSQC and off-resonance {1H} decoupling; IR (KBr): n˜ =3491 (sharp O-
H), 1687 (C=O), 1447, 1061, 696 cmꢀ1; elemental analysis calcd (%) for
C18H20O2 (268.4): C 80.56, H 7.51; found: C 80.89, H 7.63.
Experimental Section
Typical procedures: Using tBu-COMgX, see compound 9b; using tBu-
COLi, see compounds 8, 16, 17a, or 17b; using tBu-COK, see compound
9a. All H and 13C NMR chemical shifts are referenced to internal Me4Si.
1
1-(2’-Hydroxyadamant-2ꢀ-yl)-2,2-dimethylpropan-1-one (8) from 1, cata-
lyzed by methyllithium: A catalytic amount of methyllithium (0.01 mL,
0.02 mmol) in Et2O was added to an NMR tube containing the acyloin
1 (39 mg, 0.17 mmol) in THF (0.70 mL) and [D12]cyclohexane (0.03 mL)
at ꢀ708C. Adamantan-2-one (31 mg, 0.20 mmol) was added at ambient
temperature and consumed the acyloin 1 and its Li alkoxides[11] 14 within
16 min, at which time the product started to crystallize from the THF so-
lution. After vigorous shaking with Et2O and water, the Et2O layer was
washed until neutral, dried over Na2SO4, filtered, and concentrated
under 60 Torr to leave a solid mixture (75 mg) containing adamantan-2-
one, the product 8, and O=CtBu2 (7) in a 19:81:75 ratio together with
some THF and a trace of Et2O. The crude material from different runs
was recrystallized first from methanol, then from hexane: Colorless nee-
dles, m.p. 134–1358C; 1H NMR (CDCl3, 400 MHz): d=1.29 (sharps, 9H;
CMe3), 1.60 (d, 2J=12.5 Hz, 2H; two equiv protons of 2 CH2 groups),
1.69 (unresolved, 3Jꢁ2.5 Hz, 2H; 1’-/3’-H), 1.75 (partially hidden d, 2J
ꢁ13 Hz, ca. 2H; two equiv protons of 2 CH2 groups), 1.77 (s, 1H; OH),
1.79 (m, 3Jꢁ2.5 Hz, 2H; enantiotopic CH2), 1.95 (d, 2J=12.5 Hz, 2H;
2 equiv protons of 2 CH2 groups), 2.20 (d, 2J=13 Hz, 2H; 2 equiv protons
of 2 CH2 groups), 2.25 ppm (unresolved, 2H; 2 tert-CH); 13C NMR
(CDCl3, 100.6 MHz): d=26.7 and 26.8 (2 tert-CH), 29.3 (CMe3), 32.8 (2
CH2), 34.9 (2 CH2+2 CH), 37.7 (1 CH2), 45.2 (CMe3), 83.0 ppm (C-2’);
IR (KBr): n˜ =3474 (sharp O-H), 2916, 1683 (sharp C=O) cmꢀ1; elemental
analysis calcd (%) for C15H24O2 (236.35): C 76.23, H 10.23; found: C
76.15, H 10.02.
4-Hydroxy-2,2,5,5-tetramethyl-4-phenylhexan-3-one (9b) from
1 with
methylmagnesium chloride: Pivalophenone (0.040 mL, 0.24 mmol, pre-
pared from benzonitrile with tBuLi in Et2O, dH =1.35 ppm in CDCl3,
1.307 ppm in THF) was added to an NMR tube containing the two Mg
alkoxides[11] 5 (85:15, 0.12 mmol, prepared from 1 with H3C-MgCl) in
THF (0.6 mL) and [D12]cyclohexane (0.030 mL). The sealed tube was
heated at 608C for 2 h, but 1H NMR signals of the product could be de-
tected only after methanol (0.020 mL) had been added. The acidified
mixture was dissolved in Et2O and the Et2O layer was extracted with
NaOH (1m), washed until neutral, dried over Na2SO4, concentrated in
vacuo, and evacuated in a desiccator for 3 min to give the non-acidic
product fraction containing 9b and pivaloin[33] (3) in a 91:9 ratio. Pivaloin
and residual pivalophenone vanished within 5 d in an evacuated desicca-
tor, leaving crude 9b which was crystallized twice from hexane: Transpar-
ent blocks (28 mg, 93%), m.p. 104–1058C; 1H NMR (CDCl3, 400 MHz):
d=1.02 (s, 9H; 3 ꢃCH3-6), 1.06 (s, 9H; 3ꢃCH3-1),[15] 2.54 (OH), 7.26 (tt,
3Jꢁ7 Hz, 4J=1.5 Hz, 1H; p-H), 7.31 (tm, 3Jꢁ7.5 Hz, 2H; 2 m-H),
7.39 ppm (dm, 3Jꢁ7.5 Hz, 2H; 2 o-H); 13C NMR (CDCl3, 100.6 MHz):
d=26.23 (qsept, 1J=125.8 Hz, 3J=4.8 Hz, 3ꢃC-6), 27.80 (qsept, 1J=
127.0 Hz, 3J=4.9 Hz, 3ꢃC-1), 40.16 (oct, broad, 2Jꢁ3.4 Hz, C-5), 46.13
(>oct, sharp, 2J=3.8 Hz, C-2), 88.64 (unresolved, COH), 126.60 (dt, 1J=
159.0 Hz, 3J=6.7 Hz, 2 o-C), 127.19 (dt, 1J=160.5 Hz, 3J=7.4 Hz, p-C),
127.43 (dd, 1J=159.5 Hz, 3J=7.2 Hz, 2 m-C), 138.23 (td, 3J=7 Hz, 3J=
3.6 Hz to OH, i-C), 214.93 ppm (>oct, 3J=3.5 Hz, C=O), assigned
through HSQC and through HMBC with the 3J interactions C=O$1-
CH3$C-1 and C-OH$6-CH3$C-6 in addition to the 2J correlations 1-
CH3$C-2 and 6-CH3$C-5; IR (ATR on diamond): n˜ =3537 (sharp O-
H), 2956, 2925, 2870, 1685 (C=O),[15] 1483, 1363, 1077, 729, 698 cmꢀ1; ele-
mental analysis calcd (%) for C16H24O2 (248.4): C 77.38, H 9.74; found: C
77.32, H 9.69.
1-Hydroxy-3,3-dimethyl-1,1-diphenylbutan-2-one (9a) from 1 with ben-
zylpotassium: A dry NMR tube (5 mm) was kept under a blanket of
streaming dry argon gas while being loaded with potassium tert-butoxide
(45 mg, 0.40 mmol) and anhydrous toluene (0.12 mL, 1.1 mmol) and
during the following operations. The tube was cooled to ꢀ708C, cyclo-
pentane (0.30 mL) and then n-butyllithium (0.41 mmol) in hexanes
(0.20 mL) were introduced without shaking, and the sealed tube was vio-
lently shaken at ambient temperature for 2.5 min. The red precipitate[34]
was left settling at room temperature for one hour, whereafter the super-
natant was withdrawn by syringe and discarded. Cyclopentane (0.5 mL)
was injected into the tube, which was resealed and shaken briefly to
whirl up the precipitate. Upon deposition for 10–20 min, the same wash-
ing procedure with cyclopentane was repeated three times. The red pre-
cipitate remaining after the last withdrawal was covered with anhydrous
THF (0.50 mL) at ꢀ708C with subsequent extensive shaking of the
sealed tube for complete dissolution at room temperature. This dark red
solution of benzylpotassium[35] was titrated at ꢀ708C to decolorization by
adding small batches of the acyloin 1 (total 71 mg, 0.31 mmol, 78% yield
of PhCH2K), then diluted with further anhydrous THF (0.20 mL),
[D12]cyclohexane (0.030 mL), and a trace of TMS. A 1H NMR spectrum
taken after 40 min at 208C revealed the K alkoxide[11] 22 (0.23 mmol)
with dH =1.01 and 1.32 ppm (18:9) and already some O=CtBu2
(0.081 mmol); the potassium salt of the enediolate 2 remained invisible.
Benzophenone (23 mg, 0.13 mmol, 0.4 equiv) was added at ꢀ708C to give
a green solution at 238C containing residual alkoxide 22 (0.11 mmol after
6 min, 0.10 mmol after 26 min). The golden colored, clear solution was
4-Hydroxy-2,2,5-trimethyl-4-isopropylhexan-3-one (16) from 1 with n-bu-
tyllithium: A suspension of the acyloin 1 (228 mg, 1.00 mmol) in pentane
(5 mL) was stirred at ꢀ408C during the addition of nBuLi (1.00 mmol) in
hexane (0.41 mL). After 10 min, dry di-isopropyl ketone (0.151 mL,
1.10 mmol) was injected and the mixture was warmed up to room tem-
perature, then diluted with water and shaken with pentane (50 mL). The
pentane phase was washed with brine until neutral, dried over Na2SO4,
and concentrated, affording 16 as the only product together with O=
CtBu2 (7) and a rest of di-isopropyl ketone. M.p. 79–808C (twice from
pentane); 1H NMR (CCl4, 80 MHz): d=0.87 and 0.91 (2 d, 3J=6.8 Hz,
12H; 4ꢃ5-CH3), 1.21 (s, 9H; CMe3), 1.63 (broad s, OH), 2.07 ppm (sept,
3J=6.8 Hz, 2H; 2ꢃ5-H); 13C NMR (CDCl3, 20 MHz): d=17.6 and 18.6
(4ꢃ5-CH3), 28.1 (CMe3), 35.5 (2ꢃC-5), 45.5 (CMe3), 89.4 (C-4),
219.2 ppm (C-3), assigned through off-resonance {1H} decoupling; IR
(KBr): n˜ =3496 (sharp O-H), 2968, 1676 (s, C=O), 1468, 1150, 1009 cmꢀ1
;
elemental analysis calcd (%) for C12H24O2 (200.32): C 71.95, H 12.08;
found: C 72.21, H 11.94.
1,1-Dicyclopropyl-1-hydroxy-3,3-dimethylbutan-2-one (17a) from 1 with
methyllithium: A solution of the acyloin 1 (500 mg, 2.19 mmol) in anhy-
drous THF (3 mL) was stirred at ꢀ408C under dry argon gas and titrated
with methyllithium (2.19 mmol) in Et2O (1.46 mL). After 10 min, the
generated Li alkoxides[11] 14 were treated with dicyclopropyl ketone
(0.260 mL, 2.18 mmol) in THF (1 mL) and warmed up to room tempera-
ture. The mixture was diluted with water and shaken with Et2O (3ꢃ
10 mL). The combined Et2O extracts were washed with water (3ꢃ5 mL),
dried over Na2SO4, and concentrated to give a 1:1 mixture (680 mg) of
17a and O=CtBu2 (7). Distillation at 40–608C (bath temp.)/0.022 Torr
removed 7 at the expense of a diminishing yield (128 mg, 30%) of 17a as
treated with
a second portion of benzophenone (33 mg, 0.18 mmol,
1.8 equiv) and remained dark green during the 1H NMR measurement
that confirmed the complete consumption of residual alkoxide 22 within
4 min at 238C. The product alkoxide showed strongly broadened
1H NMR signals (and those of residual benzophenone were extremely
broad) which sharpened on acidification with F3C-CO2H. After workup
with Et2O and NaOH (1m) and drying the non-acidic fraction at 4 mbar
for 19 h,[33] the crude product (82 mg) contained the adduct 9a
(0.24 mmol, 77%), Ph2C=O (0.07 mmol), and PhCH2-CH(OH)-tBu[11]
(26, 0.03 mmol). The collected crude material from several runs was crys-
tallized from boiling petroleum ether: Colorless needles, m.p. 108–1098C;
1H NMR (CDCl3, 400 MHz): d=1.23 (s, 9H; CMe3), 3.16 (s, OH), 7.27–
1
a colorless oil. H NMR (CDCl3, 400 MHz): d=0.17 and 0.68 (m, 4H; 2ꢃ
2’-H and 2ꢃ3’-H, all trans to 1’-H), 0.25 and 0.42 (m, 4H; 2ꢃcis-2ꢄ-H and
2ꢃcis-3’-H), 1.27 (tt, 2H; 2ꢃ1’-H), 1.38 (s, 9H; CMe3), 3.65 ppm (s, OH),
d values assigned through comparison with 17b and in proportion to the
&
8
&
www.chemeurj.org
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 0000, 00, 0 – 0
ÝÝ
These are not the final page numbers!

Pivaloylmetals as Equilibrium Components
FULL PAPER
following 1 H,1H coupling constants as extracted by a crude computation-
[9] G. J. Abruscato, T. T. Tidwell, J. Org. Chem. 1972, 37, 4151–4156,
compound 19 therein.
[10] G. Bçhrer, R. Knorr, P. Bçhrer, B. Schubert, Liebigs Ann. 1997,
193–202, compound 19 therein.
[11] See the Supporting Information.
[12] M. S. Newman, A. Arkell, T. Fukunaga, J. Am. Chem. Soc. 1960, 82,
2498–2501.
al simulation of the observed 1H NMR multiplets (error limits ꢂ0.5 Hz):
3
3
3J(1’/cis-2’)ꢁ J(1’/cis-3’)=8.3 Hz, 3J(1’/trans-2’)ꢁ J(1’/trans-3’)=5.3 Hz,
2J(cis-2’/trans-2’)ꢁ J(cis-3’/trans-3’)=ꢀ4.3 Hz, 3J(cis-2’/cis-3’)ꢁ J(trans-
2
3
3
2ꢄ/trans-3’)=9.4 Hz,
3J(cis-2ꢄ/trans-3’)ꢁ J(cis-3’/trans-2’)=6.4 Hz;
13C NMR (CDCl3, 100.6 MHz): d=ꢀ0.30 and 1.59 (2+2ꢃCH2), 16.97
(2ꢃCH), 28.67 (CMe3), 43.34 (CMe3), 82.67 (COH), 217.48 ppm (C=O),
assigned through comparison with 17b; IR (film): n˜ =3424 (O-H), 3092
(w), 3009, 2973, 2873, 1678 (C=O), 1482, 1367, 1008 cmꢀ1; elemental anal-
ysis calcd (%) for C12H20O2 (196.29): C 73.43, H 10.27; found: C 73.09, H
10.28.
[13] a) 1H NMR: A. Furuhata, M. Hirano, I. Fujimoto, M. Matsui, Agri-
cultural Biol. Chem. 1987, 51, 1633–1640, compound 28 therein;
b) 13C NMR: A. Ejchart, Org. Magn. Reson. 1977, 9, 351–354.
[14] The t1/2 values were based on initial 0.2–0.5m NMR concentrations
of the reacting compounds. More precise rate measurements could
not be achieved because of the intricate kinetic properties of the
present systems.
1,1-Bis(1ꢀ-methylcyclopropyl)-1-hydroxy-3,3-dimethylbutan-2-one (17b)
from 1 with methyllithium: The acyloin 1 (100 mg, 0.44 mmol) in anhy-
drous THF (2.2 mL) was stirred at ꢀ408C under dry argon gas during
the dropwise addition of methyllithium in Et2O till the sharp endpoint of
CH4 evolution (collected in a gas burette). After 10 min, bis(a-methylcy-
clopropyl) ketone[27] (73 mg, 0.53 mmol) was added by pipette and stirred
for 5 min. The mixture was warmed up to room temperature, diluted
with water and extracted with pentane (50 mL). The extract was washed
with brine (3ꢃ15 mL), dried over Na2SO4, and concentrated to yield
a mixture of O=CtBu2 (7), product 17b, and pivaloin (3), but no acyloin
1. Repeated crystallizations from petroleum ether furnished pure 17b:
M.p. 87–888C; 1H NMR[43] (CDCl3, 400 MHz): d=0.239 and 0.245 (dm,
4H; BC spectral system with additional 2J and 4J, 2ꢃcis-2’-H and 2ꢃcis-
3’-H relative to 1’-CH3), 0.617 (dm, 2H; 2ꢃtrans-2ꢄ-H), 0.904 (dm, 2H; 2
ꢃ trans-3’-H), 1.17 (m, 6H; 2ꢃ1’-CH3), 1.22 (s, 9H; CMe3), 1.63 (s, OH)
ppm, with the 1 H,1H coupling constants 4J(1’-CH3/trans-2’)=0.719 Hz,
4J(1’-CH3/trans-3’)=0.768 Hz, 2J(cis-2ꢄ/trans-2ꢄ)=ꢀ4.463 Hz, 2J(cis-3’/
trans-3’)=ꢀ4.475 Hz, 3J(cis-2ꢄ/cis-3’)=9.350 Hz, 3J(trans-2ꢄ/trans-3’)=
[15] 1H NMR and IR data of the chiral component 9b are equal to those
reported for a “recrystallized” substance to which the achiral consti-
tution Ph-C(=O)-CACTHNUTRGNEUNG(-OH)tBu2 had been assigned by D. Seyferth,
R. C. Hui, W.-L. Wang, J. Org. Chem. 1993, 58, 5843–5845, on
p.5845.
[16] “&” symbolizes a not yet established type of bonding between tBu-
COM and O=CtBu2 (7), perhaps a coordinative bond or an electro-
static attraction between the oxygen atom of 7 and the metal cation.
[17] The possibility of an accumulation of NMR-silent intermediates (in
analogy with the NMR-inactive enediolate 2) was dismissed through
the in situ observation of an always constant sum of the concentra-
tions of starting alkoxides and the product O=CtBu2.
[18] S. Winstein, E. Clippinger, A. H. Fainberg, R. Heck, G. C. Robinson,
J. Am. Chem. Soc. 1956, 78, 328–335.
[19] The kinetic system 5!6&7!6<M+ >7!2 as implied in Scheme 2
is similar to the SN1 ionization sequence analyzed in Ref. [18]: “Ex-
ternal return” of a foreign component may occur by unification of
the free components or by a ligand exchange reaction at the com-
plex 6&7.
9.780 Hz,
3J(cis-2ꢄ/trans-3’)=5.853 Hz,
3J(cis-3’/trans-2ꢄ)=5.859 Hz;
13C NMR (CDCl3, 100.6 MHz): d=10.79 and 11.44 (2+2ꢃCH2), 22.17
(2ꢃC-1’), 22.75 (2ꢃ1’-CH3), 27.73 (CMe3), 45.33 (CMe3), 85.00 (COH),
216.02 ppm (C=O); IR (KBr): n˜ =3555 (sharp O-H), 3006, 1681 (C=O),
1485, 1381, 1101, 1083, 1022, 761 cmꢀ1; elemental analysis calcd (%) for
C14H24O2 (224.34): C 74.95, H 10.78; found: C 75.25, H 10.64.
[20] See Ref. [22] for the preparation and deuterium analysis of [D9]7.
[21] See Ref. [22] for the deuterium analyses by 1H and 13C NMR of the
products presented here.
Sterically Congested Molecules, Part 25. Part 24: Ref. [22].
[22] R. Knorr, D. S. Stephenson, Chem. Eur. J. 2012, DOI:10.1002/
chem.201102865, accompanying paper.
[23] Similar fairly fast tBu migrations preclude extensions of the present
cleavage method to the generation of other acylmetals.
[24] Named in analogy with the common-ion rate depression shown in
Ref. [18].
[25] L. C. Bateman, M. G. Church, E. D. Hughes, C. K. Ingold, N. A.
Taher, J. Chem. Soc. 1940, 979–1011.
[26] See Tables IV and V of Ref. [18] for the observation that steady rate
constants were obtained only after ca. 40% conversion.
[27] G. Bçhrer, R. Knorr, P. Bçhrer, Chem. Ber. 1990, 123, 2161–2166.
[28] This partitioning factor derives from the quotient of (overall decay
rate)=(fissionꢃdimerization/return) rates, divided by the rate of
the fission step.
[29] H. Ahlbrecht, G. Schneider, Tetrahedron 1986, 42, 4729–4741.
[30] R. R. Fraser, T. S. Mansour, S. Savard, J. Org. Chem. 1985, 50,
3232–3254.
Acknowledgements
Financial support of this project by the Deutsche Forschungsgemein-
schaft is gratefully acknowledged. We thank the Chemistry Department
of the LMU Mꢁnchen for supporting the final stage of this work.
[1] D. Seyferth, R. M. Weinstein, R. C. Hui, W.-L. Wang, C. M. Archer,
J. Org. Chem. 1992, 57, 5620–5629.
[2] For a survey of the literature for the years 1973–1990, see Ref. [1].
[3] For the earlier literature beginning in 1919, see: a) M. Schlosser,
Angew. Chem. 1964, 76, 124–143, on p.138; Angew. Chem. Int. Ed.
Engl. 1964, 3, 287–306; b) C. Hilf, F. Bosold, K. Harms, J. C. W.
Lohrenz, M. Marsch, M. Schiemeczek, G. Boche, Chem. Ber./Recueil
1997, 130, 1201–1212.
[4] N. S. Nudelman, F. Doctorovich, Tetrahedron 1994, 50, 4651–4666,
and refs. cited therein.
[5] T. Hiiro, Y. Morita, T. Inoue, N. Kambe, A. Ogawa, I. Ryu, N.
Sonoda, J. Am. Chem. Soc. 1990, 112, 455–457.
[31] H. Hoberg, V. Gçtz, A. Milchereit, J. Organomet. Chem. 1976, 118,
C6-C7.
[32] For NMR data, see the Supporting Information of: M. R. Crimmin,
A. G. M. Barrett, M. S. Hill, P. A. Procopiou, Org. Lett. 2007, 9,
331–333.
[33] Yields of isolated pivaloin (3) are sometimes low (<1= of 7) because
2
[6] C. S. Shiner, A. H. Berks, A. M. Fisher, J. Am. Chem. Soc. 1988, 110,
957–958.
of losses through volatility in vacuo and/or perceptible solubility in
water.
[7] Detailed experimental evidence of the structures of these largely
ionic intermediates is not known, but impressions of the structural
possibilities may be gathered from the following refs: a) E. Kauf-
mann, P. von R. Schleyer, S. Gronert, A. Streitwieser, M. Halpern, J.
Am. Chem. Soc. 1987, 109, 2553–2559; b) M. Tacke, Chem. Ber.
1995, 128, 1051–1053; c) K. Ohno, S. Maeda, J. Phys. Chem. A 2006,
110, 8933–8941.
ˇ
[34] a) L. Lochmann, J. Pospisil, D. Lꢅm, Tetrahedron Lett. 1966, 7, 257–
262; b) M. Schlosser, J. Organomet. Chem. 1967, 8, 9–16; c) com-
pare: M. Schlosser, Organometallics in Synthesis 2nd ed., Wiley, Chi-
chester, 2002, p. 273.
[35] 1H NMR in THF at ꢀ308C: G. Vanermen, S. Toppet, M. van Beylen,
P. Geerlings, J. Chem. Soc. Perkin Trans. 2 1986, 699–705, Table 4
therein.
[8] G. F. Hennion, C. F. Raley, J. Am. Chem. Soc. 1948, 70, 865–866.
Chem. Eur. J. 2012, 00, 0 – 0
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
&
9
&
ÞÞ
These are not the final page numbers!

Rudolf Knorr et al.
[36] a) 1H NMR and conformation: O. Takahashi, K. Saito, Y. Kohno, H.
Suezawa, S. Ishihara, M. Nishio, Eur. J. Org. Chem. 2004, 2398–
2403; b) 13C and 1H NMR: S. Zushi, Y. Kodama, Y. Fukuda, K.
Nishihata, M. Nishio, M. Hirota, J. Uzawa, Bull. Chem. Soc. Jpn.
1981, 54, 2113–2119.
[40] H. Schuh, E. J. Hamilton, H. Paul, H. Fischer, Helv. Chim. Acta
1974, 57, 2011–2024.
[41] Y. P. Tsentalovich, H. Fischer, J. Chem. Soc. Perkin Trans. 2 1994,
729–733, Table 2 therein.
[42] For comparison, benzoyllithium (Ph-COLi) was reported[5] as being
short-lived even at ꢀ1058C.
[37] a) J. S. Lomas, S. Briand, D. Fain, J. Org. Chem. 1991, 56, 166–175,
compound 3e therein; b) Deviating data: M. Sasaki, J. Collin, H. B.
Kagan, New J. Chem. 1992, 16, 89–94, compound 5 therein.
[38] pKa =40.7 for PhCH2K in equilibrium with toluene in THF at
ꢀ788C according to p. 4739 of Ref. [29]
[43] We thank Dr. David S. Stephenson for the computational 1H NMR
simulation for 17b with his program DAVINX.
Received: September 13, 2011
Revised: December 1, 2011
Published online: && &&, 0000
[39] G. Mendenhall, H. T. Chen, US Patent 4595784, Chem. Abstr.,
1986, 105, P82043q.
&
10
&
www.chemeurj.org
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 0000, 00, 0 – 0
ÝÝ
These are not the final page numbers!

Pivaloylmetals as Equilibrium Components
FULL PAPER
Reactive Intermediates
R. Knorr,* G. Bçhrer, B. Schubert,
P. Bçhrer ...................... &&&& — &&&&
Pivaloylmetals (tBu-COM: M=Li,
MgX, K) as Equilibrium Components
NMR-invisible intermediates pivaloyl-
magnesium, -lithium, or -potassium,
have been generated through the
of the parent tri-tert-butyl acyloin.
Their interception by protonation or
by ketones reveals that a common ini-
tial fission step limits the overall rate
(see scheme).
ꢀ
novel, reversible heterolytic C C bond
cleavage of the overcrowded alkoxides
Chem. Eur. J. 2012, 00, 0 – 0
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
&
11
&
ÞÞ
These are not the final page numbers!