On the mechanism of the conversion of methanol to 2,2,3-trimethylbutane (triptane) over zinc iodide

Article abstract of DOI:10.1021/jo0617823

Methanol is converted to a mixture of hydrocarbons by reaction with zinc iodide at 200 °C with one highly branched alkane, 2,2,3-trimethylbutane (triptane), being obtained in surprisingly high selectivity. Mechanistic studies implicate a two-stage process, the first involving heterogeneously catalyzed formation of a carbon-carbon-bonded species, probably ethylene, that undergoes homogeneously catalyzed sequential cationic methylation to higher hydrocarbons. The first stage can be bypassed by addition of olefins, higher alcohols, or arenes, which act as initiators. Rationales for the particular activity of zinc iodide and for the selectivity to triptane are proposed.

Full text of DOI:10.1021/jo0617823

On the Mechanism of the Conversion of Methanol to
,2,3-Trimethylbutane (Triptane) over Zinc Iodide
2
†
†,‡
†
§
John E. Bercaw, Paula L. Diaconescu, Robert H. Grubbs, Richard D. Kay,
§
,†
†
†,|
Sarah Kitching, Jay A. Labinger,* Xingwei Li, Parisa Mehrkhodavandi,
§
§
†
George E. Morris, Glenn J. Sunley, and Patrick Vagner
Arnold and Mabel Beckman Laboratories of Chemical Synthesis, California Institute of Technology,
Pasadena, California, 91125 and BP Chemicals Limited, Hull Research & Technology Centre, Kingston
Upon Hull, North Humberside HU1 28DS, England
jal@its.caltech.edu
ReceiVed August 29, 2006
Methanol is converted to a mixture of hydrocarbons by reaction with zinc iodide at 200 °C with one
highly branched alkane, 2,2,3-trimethylbutane (triptane), being obtained in surprisingly high selectivity.
Mechanistic studies implicate a two-stage process, the first involving heterogeneously catalyzed formation
of a carbon-carbon-bonded species, probably ethylene, that undergoes homogeneously catalyzed sequential
cationic methylation to higher hydrocarbons. The first stage can be bypassed by addition of olefins,
higher alcohols, or arenes, which act as initiators. Rationales for the particular activity of zinc iodide and
for the selectivity to triptane are proposed.
Introduction
polyphosphoric acid proceeds at considerably lower tempera-
tures (190-200 °C) to give a complex, alkane-rich mix of
hydrocarbons, of which C7 comprises the largest fraction (27%),
primarily mono- and dibranched isomers.
Against this background, a 1978 discovery stands out: at 200
With the forecast depletion of oil reserves, methanol is likely
to play an increasingly important role for both energy and
petrochemical feedstocks.1 One approach that has received
considerable attention over the years is the dehydrative conver-
sion of methanol to hydrocarbons over acidic catalysts, which
has a very long history (dating back to 1880! ). Different classes
of hydrocarbons can be obtained (usually not very selectively)
depending on the nature of the catalyst and the temperature
4
°
C methanol reacts with zinc iodidesnot normally considered
to have strongly acidic properties, unlike polyphosphoric acids
to give an alkane-rich hydrocarbon mix with distribution again
peaking at C7. However, in this case one highly branched alkane,
2
2
,2,3-trimethylbutane (common name triptane), was by far the
3
(typically 200-450 °C). The best known are the methanol-to-
major product obtained in overall yields up to 20% (moles
carbon basis). Triptane was reported to constitute as much as
one-half of the gasoline-range fraction. Similar chemistry was
found for zinc bromide at slightly higher temperatures but not
gasoline (MTG) and methanol-to-olefins (MTO) processes,
which operate at temperatures above 300 °C to produce
aromatic-rich and light olefin-rich products respectively; selec-
tivity is ascribed to the shape-selective zeolitic and alumino-
phosphate catalysts. In contrast, reaction of methanol with
5
for zinc chloride. Earlier patents report conversion of methanol
to hydrocarbons over molten ZnCl2 at higher temperatures
†
(325-425 °C) with 20-25% selectivity to isobutane along with
Caltech.
‡
substantial amounts of hexamethylbenzene (HMB) and coke.6
Current address: Department of Chemistry & Biochemistry, University of
California, Los Angeles, Los Angeles CA 90095.
These observations are of significant interest. On the practical
side, such a highly branched alkane, with an octane number of
§
BP.
|
Current address: Department of Chemistry, University of British Columbia,
Vancouver, BC, Canada, V6T 1Z1.
(
1) Olah, G. A. Angew. Chem., Int. Ed. Engl. 2005, 44, 2636-2639.
(4) Pearson, D. E. J. Chem. Soc., Chem. Commun. 1974, 397.
(5) Kim, L.; Wald, M. M.; Brandenburger, S. G. J. Org. Chem. 1978,
43, 3432-3433. Also see U.S. Patents 4,059,646, 4,059,647, 4,126,642,
4,126,643, 4,151,214, 4,166,189, and 4,249,031.
(6) Grosse, A. V.; Snyder, J. C. U.S. Patent 2,492,984, 1950. Bell, W.
K.; Chang, C. D. U.S. Patent 3,969,427, 1976.
(2) LeBel, J. A.; Greene, W. H. Am. Chem. J. 1880, 2, 20 (cited in ref
3
a).
(
3) a) Chang, C. D. Catal. ReV.sSci. Eng. 1983, 25, 1-118. (b) Olah,
G. A.; Moln a´ r, AÄ . Hydrocarbon Chemistry, 2nd ed.; John Wiley& Sons:
Hoboken, NJ, 2003; pp 117-122 and references therein.
1
0.1021/jo0617823 CCC: $33.50 © 2006 American Chemical Society
Published on Web 10/19/2006
J. Org. Chem. 2006, 71, 8907-8917
8907

Bercaw et al.
growth, and the properties of ZnI2 that account for its particular
ability to effect this reaction. We hope that these findings can
help point the way toward an eventual full understanding of
this fascinating system.
1
12, would be a valuable fuel component if it could be produced
Results and Discussion
cheaply and efficiently. Several selective routes to triptane have
been previously reported, but all involve precursors and/or
reagents that would be prohibitively expensive for large-scale
fuel production. However, they are perhaps even more intriguing
from a fundamental viewpoint: what mechanism(s) could
account for such high selectivity to a single product out of the
7
Toward a Reproducible Experimental Protocol. In a
“
typical” experiment, a mixture of MeOH and ZnI2, in molar
ratios ranging from 2:1 to 3:1, is heated in a closed vessel at
00 °C for several hours. After cooling to room temperature,
2
three phases are observed: an upper hydrocarbon layer, a lower
aqueous layer, and some solid. Analysis of the organic layer
by a variety of techniques (GC, GC/MS, H and C NMR)
reveals the presence of a very large number of components
(literally) hundreds that are formed in detectable quantities?
This issue takes on added importance in light of the fact that
1
13
there is to date no consensus mechanistic explanation for any
of the methanol-to-hydrocarbon conversions alluded to above.
Methylation/alkylation of intermediate olefinic species via
carbocationic intermediates may reasonably be invoked to
account for growth of the first-formed light hydrocarbons given
the acidic nature of all the catalysts that effect such reactions;
however, how might the first carbon-carbon bond form? Most
proposals center on some form of [-O-CH2:] center produced
by deprotonating a surface-bound methanol or methoxy group
(
several hundred by GC; see Figure 1 for a typical chromato-
gram) with the largest being triptane. Some 2,3,3-trimethylbut-
-ene (triptene) is also present; its GC signal is not always well
separated from that of triptane, but its presence is readily seen
1
1
3
in the C NMR spectra (Figure 2). There is always much more
triptane than triptene. The next most prominent single compo-
nent is hexamethylbenzene (HMB); there are also significant
amounts of branched lighter alkanes isobutane, 2-methylbutane,
and 2,3-dimethylbutane (signals identified in Figure 2). The
combined yield of “triptyls” (triptane plus triptene), on the basis
of moles carbon charged as methanol, is generally found to be
around 17 ( 3%.
(or perhaps the trimethyloxonium ion), which reacts with a
“
methyl cation equivalent”, i.e., some [CH3-O-X] species
activated for nucleophilic attack by coordination to the catalyst
surface. The plausibility of such a deprotonated intermediate
under acidic conditions has been questioned.3
However, it can be difficult to achieve such “typical” results
as reproducibility can be affected by several factors: the source
of the reagents used (less-pure samples of ZnI2 sometimes work
better); the nature and size of the reaction vessels (autoclaves,
glass tubes); the presence of small amounts of various organic
additives; and, most disconcertingly, whether or not the mixture
is agitated during reaction. The latter can have a dramatic
effect: in two otherwise identical reactions carried out in glass
tubes, one with stirring (or shaking) and the other quiescent,
the former is found to give no triptyls at allsindeed, no visible
hydrocarbon layer formsswhile the latter gives the standard
results summarized above. This contrast in behavior is quite
reproducible. On the other hand, in the presence of small
amounts of additives such as higher alcohols, olefins, and
aromatics, stirred and unstirred reactions behave identically.
In order to understand this highly unusual behavior and ensure
that reliable, reproducible baseline data can be obtained, we
developed a standard experimental protocol as follows. A 790
mg amount of MeOH and 2.4 g of ZnI2 (3.3:1 molar ratio) are
loaded into a thick-walled glass tube equipped with a Teflon
valve, and the mixture is agitated and/or gently warmed until
complete dissolution has occurred. The tube is then immersed
up to the Teflon valve in a 200 °C oil bath behind a blast shield,
where it is maintained for a desired period of time, then removed
and allowed to cool to room temperature. At this point the
mixture remains a completely homogeneous solution with no
trace of any visible hydrocarbon layer; analysis of the mixture
indicates that no reaction has occurred, other than partial
dehydration of methanol to dimethyl ether (DME) along with
formation of small amounts of MeI, even for heating times as
long as 24 h.
Haw put forth a quite different model wherein C-C bond
formation takes place via methylation of aromatic intermediates
in a “carbon pool” and there is no “first” C-C bond formation,
at least not from methanol: organic impurities, always present
at some levels, are responsible for the initial formation (during
an induction period) of the carbon pool, which then serves as a
“scaffold” on which subsequent chemistry takes place. Experi-
mental support for this proposal was obtained for reactions over
HZSM-5 and HSAPO-34.8
Kim et al. argued that a carbocation-based route could not
account for the selective formation of triptane over ZnI2 and
proposed instead a carbene-based route with a carbenoid
organozinc intermediate analogous to the Simmons-Smith
5,9
reagent. Such a route may offer a more attractive explanation
for the first C-C bond formationsvia coupling of two
carbenessand might explain the selectivity for triptane, at least
in some aspects (see below). On the other hand, it seems difficult
to reconcile stability of any organozinc species with the protic
medium in which the reaction takes place.
We report here the results of some recent studies. While we
are not yet in a position to formulate a complete detailed
mechanism, we have been able to shed some light on a number
of questions. These include the existence of a first C-C bond-
forming step, the carbocation-based route for hydrocarbon
(7) See, for example: (a) Ford, T. A.; Jacobson, H. W.; McGrew, F. C.
J. Am. Chem. Soc. 1948, 70, 3793-3795. (b) Howard, T. R.; Lee, J. B.;
Grubbs, R. H. J. Am. Chem. Soc. 1980, 102, 6876-6878. (c) Reetz, M. T.;
Westermann, J.; Steinbach, R. Angew. Chem., Int. Ed. Engl. 1980, 19, 900-
9
01.
(8) a) Haw, J. F.; Song, W.; Marcus, D. M.; Nicholas, J. B. Acc. Chem.
Res. 2003, 36, 317-326. (b) Olsbye, U.; Bjørgen, M.; Svelle, S.; Lillerud,
K.-P.; Kolboe, S. Catal. Today 2005, 106, 108-111.
If, on the other hand, an identically charged tube is directly
immersed in the oil bath, without trying to dissolve all the ZnI2,
separation of an organic layer can be observed after an hour or
less. After cooling two liquid phases along with a substantial
(9) The 1978 paper promises “supporting evidence for carbenoid and
organozinc intermediates and a detailed mechanism consistent with the high
triptane yield” in “subsequent publications”; however, no such publications
have appeared.
8908 J. Org. Chem., Vol. 71, No. 23, 2006

ConVersion of Methanol to 2,2,3-Trimethylbutane
FIGURE 1. GC trace of typical reaction product mixture. Labeled peaks: 1, dimethyl ether; 2, acetone (added as solvent); 3, chloroform (used to
extract organic layer); 4, triptene and triptane (not resolved in standard analysis); 5, cyclohexane (added as internal reference standard); 6,
hexamethylbenzene.
FIGURE 2. Part of the 13C NMR spectrum of a typical product mixture, showing the sp carbon signals for triptane, triptene, HMB, and several
3
minor components.
amount of solid are present, and the organic layer contains
substantial amounts of triptyls, as above. The solid is predomi-
nantly unchanged ZnI2, as demonstrated by powder-pattern
XRD.
Role of Additives. Unreactive homogeneous (predissolved)
mixtures can be “turned on” by the addition of a small amount
of a higher alcohol. In the presence of as little as 1% (by weight,
relative to MeOH) added isopropanol or tert-butanol, reaction
J. Org. Chem, Vol. 71, No. 23, 2006 8909

Bercaw et al.
FIGURE 3. Yield of triptyls as a function of time for standard reactions with 1.5 wt % added i-PrOH.
FIGURE 4. Yield of triptyls as a function of wt % added i-PrOH for standard reactions carried out for 2 h.
proceeds as in the non-predissolved case, although no solid
appears during reaction or after cooling at its conclusion.
Formation of triptyls as a function of time (determined by a
series of parallel experiments as we have no convenient way to
monitor progress in situ) is shown for i-PrOH as additive in
Figure 3. At an intermediate additive level (1.5%) triptyls are
detectable at an early stage, with little induction period, and
build to a maximum level over approximately 3 h; no detectable
change takes place if heating is continued beyond that point,
up to 24 h. The maximum yield, around 70 mg, corresponds to
about 20% on the basis of MeOH charged or 18% on the basis
of total carbon (including i-PrOH).
A “complete” analysis of the hydrocarbon layer obtained from
the reaction described above, using a standard refinery GC
routine (“PIANO”, for paraffin/isoparaffin/aromatic/naphthene/
olefin), indicated that the distribution as a function of carbon
number peaks at C7 with lesser amounts of C8-11 and much
less of the lighter fractions. Isoparaffins dominate, as can be
seen by NMR as well (see above); amounts of olefins,
naphthenes, and aromatics are considerably smaller, and hardly
any normal paraffins are found. This distribution should not be
considered quantitatively accurate, however, as many of the
assignments, which rely solely on retention time, are probably
not reliable, especially for the heavier components that elute at
longer times. In particular, hexamethylbenzene, known by NMR
to be a major component, was not detected in the PIANO
analysis at all; probably it eluted after the run was terminated.
Figure 4 shows triptyl yield as a function of the amount of
added i-PrOH after 2 h reaction. The yield increases steadily
with additive amount up to about 3% and then levels off.
Reactions with lower additive levels continue to form triptyls,
up to the same maximum yield as above, if allowed to react for
longer times (except, as noted above, for the zero-additive point).
Also, the (qualitative) composition of the organic layer looks
very similar for all these experiments, and reactions with lower
triptyl yields contain substantial amounts of unreacted MeOH/
DME. Hence, it appears the additive effect has to do with the
overall rate of conversion to hydrocarbons rather than any
significant alteration of the mechanism of conversion.
Furthermore, the selectivity for triptyls is established from
the earliest appearance of hydrocarbons, as opposed to initial
formation of lighter hydrocarbons that grow as reaction pro-
ceeds. In confirmation of this conclusion, two experiments under
standard conditions, with 1.5% i-PrOH added, were carried out
for 2 and 3 h, and the amounts of unreacted MeOH and DME
in the organic and aqueous layers were determined by NMR.
The results were as follows: triptyl yield (based on MeOH
charged), 12% and 19%, respectively; unconverted MeOH
(including DME), 37% and 5%, respectively. The triptyl yield
on the basis of conVerted MeOH is the same (19-20%) within
experimental accuracy.
Fairly good mass balances can be obtained. A mixture
containing 770 mg of MeOH, 40 mg of i-PrOH, and 2.27 g of
ZnI2 was heated for 3 h following the standard protocol. In
addition to the two liquid layers, about 90 mg of volatiles were
produced, estimated by weighing the tube before and after
opening. These were shown to consist primarily of light
hydrocarbons, particularly isobutane, isobutene, and propene,
along with some methyl iodide. The organic and aqueous layers
were separated and found to weigh 315 mg and 2.68 g,
respectively; the aqueous layer contained about 60 mg of MeOH
and 35 mg of DME. Subtracting the latter and the initial weight
10
of ZnI2 gives 310 mg of water. Correcting for the amount of
unconverted MeOH/DME, the reaction should have produced
(10) This does not take into account iodide lost as MeI, for which we as
yet do not have good quantitation, so the actual yield of water is somewhat
higher.
8910 J. Org. Chem., Vol. 71, No. 23, 2006

ConVersion of Methanol to 2,2,3-Trimethylbutane
EtOH does not initiate reaction following the standard protocol
but does under some other conditions (different reactor types/
sizes); running the standard reaction under an ethylene atmo-
sphere does initiate. We believe this is a consequence of the
lower solubility of ethylene compared to propene or isobutene;
whether there is enough dissolved ethylene to effectively initiate
will then depend on factors such as headspace volume.
From these observations we infer that the unpromoted
conversion of methanol to hydrocarbon involves two separate
stages: an initiation stage, involving formation of one (or more)
reactive C-C-bonded species, probably ethylene, and a growth
stage, involving elaboration of these species by successive
methylation, leading ultimately to triptyls and other hydrocar-
bons. The stirring/predissolution effects strongly suggest that
the first stage is not only relatively slow, taking place during a
moderately lengthy induction period, but also catalyzed het-
erogeneously, as it does not take place at all in the absence of
undissolved solid. The second stage, in contrast, proceeds
equally well whether or not there is any solid phase. The
initiation stage can be completely bypassed by addition of a
suitable additive, either an unsaturated compound (olefin, arene)
or an alcohol, which presumably functions as a facile precursor
to an olefin.
FIGURE 5. Yield of triptyls as a function of wt % added t-BuOH for
standard reactions carried out for 2 h. The dotted line shows the
theoretical maximum yield if triptane were formed solely by alkylating
t-BuOH.
about 315 mg of hydrocarbon and 380 mg of water, in
reasonable agreement with observation.
The role of i-PrOH may be probed with the use of 13C-labeled
reagents. (Use of deuterium labeling was found to be uninfor-
mative as H/D exchange among all positions of all products is
1
2
rapid and complete under reaction conditions.) When ( CH3)2
1
3
CHOH is used as promoter in the reaction of unlabeled MeOH,
13
relative intensities in the C NMR spectrum show that the label
is found in the internal (Cb,c) positions of triptane to a far greater
extent than in the methyl (Ca,d) groups (eq 2). The gas-phase
components from the converse experimentsreaction of CH3-
OH promoted with unlabeled i-PrOHsinclude (by NMR) MeI
The observation that less-pure ZnI2 samples are often more
effective catalysts might be due to the presence of insoluble
impurities that catalyze initiation. Another possibility is that the
impurities themselves are sources of initiator, as in Haw’s
proposed mechanism for MTO. For example, some samples
appear to be contaminated with alkylammonium salts, which
could generate olefins via Hoffmann elimination.
1
3
(completely labeled), large amounts of propene (mostly unla-
beled), and smaller amounts of isobutene, which is almost
completely labeled in the terminal positions but only partially
(
∼85%) labeled in the central position. These findings suggest
At least some of the triptane appears to form by direct
sequential buildup on the initiator, as suggested by the experi-
ment with labeled i-PrOH: carbon atoms that are part of the
olefinic backbone of the (presumed) intermediate propene
preferentially end up in the backbone (internal) positions of
triptane. The fact that triptane is produced in (molar) excess of
the amount of added initiator implies that other processes are
going on as well. This issue is addressed in more detail below.
that the alcohol is readily dehydrated under these acidic
conditions to the olefin, which serves as the actual promoter.
Under the particular conditions of these experiments (relatively
little promoter added), some of the isobutene comes from direct
methylation of propene, but most of it is built up entirely out
of MeOH. The fact that little 12C remains in terminal positions
further suggests that they are able to exchange with MeOH-
derived methyl groups (see below and Supporting Information
for more detailed discussion).
It is of interest to compare this picture to Haw’s proposal for
MTO. Our findings strongly suggest that formation of C-C-
bonded species from methanol alone is difficultsmuch more
so than the growth stepssbut does take place under the right
conditions. One could argue that even in the absence of
deliberately added initiator there are impurities present that are
responsible for initiating growth. However, these would have
to be species that only function effectively as initiators in the
presence of solids, whereas the reaction works perfectly well
with a wide variety of deliberately added initiators in the absence
of solid. Such an interpretation seems to us quite improbable.
The behavior with t-BuOH as additive is similar, although
smaller amounts are able to produce the same degree of
acceleration (Figure 5). In several of these experiments the
number of moles of triptyls obtained is substantially higher than
the number of moles of t-BuOH added, which demonstrates that
these additives act as true initiators, as opposed to being simply
frameworks on which triptyls are assembled.
Other additives have been shown to function as initiators.
Olefins ranging from light, such as ethylene and propylene, up
to some heavier than C7, such as di-isobutylene (2,4,4-
trimethylpent-2-ene), act similarly to the above-described al-
cohols as do aromatics, up to and including hexamethylbenzene,
a reaction product. Benzene itself is a less effective initiator:
under standard conditions with benzene as initiator the only new
hydrocarbon observed after 3 h is toluene. After more prolonged
reaction triptyls begin to form, along with more highly meth-
ylated benzenes. Paraffins appear to be completely ineffective.
It would be harder, at this point, to say whether the growth
stage involves (in part or exclusively) a carbon pool of
substituted aromatics as opposed to buildup of increasingly
substituted olefins. The observations with benzene as initiator
suggest that Haw’s mechanism can serve as a source for light
olefins as no triptane formation is observed until after methylated
benzenes have begun to appear. A labeling experiment is also
1
3
suggestive: reaction of CH3OH with unlabeled hexamethyl-
benzene as initiator affords triptane with a substantial amount
1
2
of C in all positions. However, the recovered hexamethyl-
1
3
benzene contains a good deal of C in methyl positions,
indicating that those positions exchange with MeOH, so any
conclusion from this experiment is open to question. Additional
J. Org. Chem, Vol. 71, No. 23, 2006 8911

Bercaw et al.
experiments bearing on this point will be described in the
following section.
SCHEME 1
Attempts To Test a Carbene-Based Mechanism. The
proposed carbene-based route that could explain selectivity for
triptane was not spelled out in detail,5 but it seems likely that
any such mechanism would involve successive olefin cyclo-
propanation and ring-opening steps. In such a scheme the
conclusion of a sequence leading to the triptane framework
would be cyclopropanation of 2,3-dimethylbut-2-ene (1) to give
,9
1
,1,2,2-tetramethylcyclopropane (2), which would ring open at
one of the less-substituted C-C bonds; this hypothetical
sequence is represented by eq 3. Possible precedent may be
found in the Cu-catalyzed hydrogenolysis of 2, which under
certain conditions (admittedly, quite different from those of the
present study) gives triptane nearly quantitatively.11
Carbene or Carbocation? A Further Labeling Experi-
ment. The facile rearrangement of label in Scheme 1 strongly
implicates a carbocation-based mechanism; it is possible,
however, that the rearrangement and growth process involve
two distinct mechanisms. We can distinguish between the two
alternatives, carbene-based and carbocation-based, by examining
the labeling pattern of triptane synthesized from ordinary MeOH
and dilabeled EtOH. Although EtOH is not a reliable initiator
as discussed above, the reaction can be carried out by not
To test whether this sequence is at all plausible, we added 1
and 2, respectively, to ZnI2/MeOH reaction mixtures. In both
cases reaction was observed even at 160 °C (at which temper-
ature conversion of MeOH to hydrocarbons does not normally
proceed at all). Cyclopropane 2 readily ring opens to a mixture
of triptane and triptene along with much smaller amounts of a
number of additional components (by NMR). Similarly, 1 reacts
under these conditions to give triptyls. At early stages the latter
consist primarily of triptene; the triptane/triptene ratio gradually
increases over time. Side products, all in lower amounts, include
isomerized olefin (2,3-dimethylbut-1-ene) and oligomers. When
the reaction is carried out using 13CH3OH, label is found in all
methyl/methylene positions of triptane and triptene but not the
internal positions; GC/MS shows that each triptyl contains
exactly one 13CH3 group (eq 4).
1
3
predissolving the ZnI2; C NMR of the product mixture
obtained is shown in Figure 6.
There are two key informative features. First, as in the
previous experiment with labeled i-PrOH, the signals corre-
sponding to internal carbons (b, c) are considerably enhanced
relative to those for the methyl carbons (a, d). Second, for all
four carbons the signals appear as approximate 1:3:1 “triplets”,
attributable to superimposed singlet and doublet signals (cor-
1
3
13
responding to C with no or one adjacent C, respectively) in
about 60:40 ratio. The signals for the internal carbons exhibit
weak outer peaks as well, indicating an additional triplet signal
1
3
arising from two adjacent C nuclei.
The complete analysis of the spectral results and their
mechanistic implications is complex and, hence, provided in
the Supporting Information. Briefly summarized, the carbene
mechanism is not viable as the symmetry of the initial
cyclopropane intermediate precludes any outcome where internal
positions are more enriched in label than methyls. A carbocation
route, in contrast, can satisfy the observations, both qualitatively
and quantitatively, if (and only if) the growth sequences
methylation of olefins by a methyl cation equivalentsis at least
partly reversible, that is, that the resulting carbocation can lose
a methyl cation to revert to the precursor olefin in competition
with the normal growth step of losing a proton to give the next
higher olefin. That such a process does in fact operate is
demonstrated by the labeling pattern found for isobutene derived
from labeled methanol plus unlabeled isopropanol (see above).
These results are necessary, but not sufficient, to show that
2
is an intermediate in the ZnI2-catalyzed conversion of MeOH
to triptyls. Equation 3 suggests a further diagnostic test: label
introduced from the cyclopropanating species should wind up
exclusively in the tert-butyl group (Ca), not at Cd, but the
1
3
reaction of 1 with CH3OH puts label in all terminal positions
eq 4). If the ring opening of 2 labeled at the 3-position is
(
With that feature included, both the enrichment of label at
the internal positions and the relative intensities of the apparent
multiplets for each signal can be accurately modeled, as shown
in the Supporting Information; the latter agreement is illustrated
in Figure 7. The proposed mechanism is spelled out in the last
section below along with a proposed explanation of how it can
account for selectivity to triptane.
regiospecific as shown, that would rule out this mechanism for
the methylation of 1 to triptyls.
However, such a conclusion would require that we know label
does not scramble among the terminal positions under reaction
13
conditions. In fact, it does: a C label in the methylene/adjacent
methyl positions of triptene undergoes complete statistical
scrambling to all peripheral positions under the above conditions
Hydrogen Transfer. Formation of alkanes, such as triptane,
by dehydrative condensation of MeOH () CH2 + H2O) requires
that an equivalent of H2 be supplied at some point with
corresponding formation of hydrogen-deficient species. As noted
above, the latter are primarily aromatics, especially HMB. How
does this process occur?
(
160 °C, 3 h) (Scheme 1). Hence, the labeling experiment cannot
be used to determine whether the carbene/cyclopropane mech-
anism is viable.
(11) Bart o´ k, M.; P a´ link o´ , I.; Moln a´ r, AÄ . J. Chem. Soc., Chem. Commun.
1
987, 953-954.
8912 J. Org. Chem., Vol. 71, No. 23, 2006

ConVersion of Methanol to 2,2,3-Trimethylbutane
FIGURE 6. Part of 13C NMR spectrum showing triptane signals for reaction of unlabeled MeOH with 13
C H OH. (Position labeling shown in eq
2 5
2
.)
FIGURE 8. Correlation between yields of triptyls and HMB for
standard reactions carried out to varying degrees of completion.
up as an arene, which would be further methylated to HMB.
Under standard reaction conditions, the growth of HMB
correlates closely with that of triptane (Figure 8), supporting
1
2
the proposed link. The molar ratio of HMB:triptane is
considerably lower than 1:3, the ratio required for all the excess
hydrogen in triptane (not even considering other alkanes) to be
accounted for by HMB (eq 5). Presumably the deficit is
explained by formation of other aromatics, of which a number
are detected, even though no single one is obtained in high yield.
FIGURE 7. Observed and simulated 13C NMR patterns for the c and
d positions of triptane formed from unlabeled MeOH and doubly labeled
EtOH. (See Supporting Information for details of simulation.)
One possibility would be direct transfer from MeOH to olefin,
giving alkane plus formaldehyde; the latter could be a precursor
to aromatics via condensation. Triptene is partially reduced to
triptane by heating with MeOH/ZnI2 even at 160 °C but only if
the mixture is unstirred and not predissolved. A fully predis-
solved reaction mixture shows no hydrogenation of added
triptene to triptane over 3 h at 200 °C. On the other hand,
triptane is formed by ring opening of 2 or methylation of 1
under the same conditions (see above), implying that intermedi-
ate hydrocarbons are the immediate source of the requisite
hydrogen.
33CH OH f 3C H + C (CH ) + 33H O
(5)
3
7
16
6
3 6
2
As a test, a standard reaction mixture containing added
triptene and cyclohexa-1,4-diene, a model arene precursor, was
heated at 150 °C for 2 h. Both triptane and benzene were formed
as major products along with a small amount of methylated
(12) The ratio of HMB/triptyls does vary under different reaction
conditions, especially in reactions starting with DME carried out at lower
temperatures (see below). We believe this reflects increased yields of less-
substituted benzenes at the expense of HMB but do not yet have sufficiently
complete analytical data to confirm that.
Stoichiometrically, then, this amounts to transfer of H2 from
one olefinic hydrocarbon to another, resulting in an alkane and
a multiply unsaturated species; the latter would eventually end
J. Org. Chem, Vol. 71, No. 23, 2006 8913

Bercaw et al.
benzenes. No hydrogenation of triptene was observed under
these conditions in the absence of the diene.
better described as a low-temperature molten salt: the melting
point of the tetraaquo salt has been reported to be -7 °C.13
If water formation is indeed the limiting factor, then it ought
to be possible to continue reaction by removing it. This is in
fact the case. Another multicycle DME reaction like the one
described above was stopped after the fourth cycle, and all
volatiles were removed by vacuum transfer, leaving an off-white
solid. This was treated with a fresh charge of DME and MeI,
and the next cycle produced the full expected amount of triptyls.
No reaction took place if DME but no MeI was added. This
pattern can be continued indefinitely: up to 18 CH2/Zn have
been converted over a single initial charge of ZnI2, as long as
water (along with all other volatiles) is periodically removed
and lost MeI replaced, with no apparent falloff in reactivity or
triptyl yield.
On the Role of ZnI2. In a standard run the ratio of MeOH:
ZnI2 is about 3.3:1; since the maximum yield of triptane is
around 20%, less than one MeOH per Zn is actually converted
to triptane. On the other hand, the MeOH (and its initial
dehydration product DME) can be essentially entirely converted
to hydrocarbon under these conditions, which suggests at least
the possibility that ZnI2 acts as a catalyst. Kim et al. reported
that ZnI2 can be recovered and reused, which again suggests it
5
is a catalyst, but provided no details. We carried out a number
of experiments which bear on this question.
If the starting ratio of MeOH:ZnI2 is increased, similar
behavior and triptyl yields are maintained only up to around
4:1. Beyond that yields fall off rapidly; in most such experiments
no hydrocarbon formation at all is observed. This suggests that
the first, initiation stage of the reaction is inhibited by too much
MeOH. Since partial dehydration of MeOH to DME is rapid at
any concentration, it seems likely that this is a water effect,
which could be offset by using DME instead of MeOH as feed.
A mixture of pure DME with dry ZnI2 is not reactive, but in
the presence of small amounts of water and MeI (typically
around 10% and 30%, respectively, relative to DME), reaction
proceeds readily. At 160 °C (where MeOH does not react at
all, even with initiator), DME reactions proceed at rates close
to those for MeOH at 200 °C. With addition of initiator (i-
PrOH or di-isobutylene) such reactions proceed even faster and
can be carried out at temperatures as low as 140 °C to attain
ultimate triptyl yields around 20%. This comparison strongly
suggests that water inhibits the initiation stage, which is
completely shut down if too much water is present.
Mechanistic Considerations: Why Triptane? Why Zinc
Iodide? While the results discussed above do not establish any
mechanism unequivocally, everything seems to us to suggest
that the growth stage of the reaction involves an acid-catalyzed
carbocation-based mechanism that takes place homogeneously
in a very concentrated aqueous/methanolic solution of ZnI2. This
could involve Brønsted acidity, Lewis acidity, or both. Further-
more, ZnI2 is quite soluble in alcohols, and concentrated aqueous
ZnI2 solutions readily dissolve substantial quantities of organic
compoundssa saturated solution can take up an equal weight
1
4
(or more) of diethyl ether at ambient temperatures sso that as
long as the ZnI2 is completely dissolved before heating, it should
stay that way throughout the reaction.
How can we account for the remarkable selectivity to triptane
within such a mechanistic framework? It should first be noted
that selectivity for triptane cannot be explained on any ther-
modynamic grounds. From published thermochemical data15 we
calculate that triptane would comprise only 6% of the total C7
alkane fraction at equilibrium (at 500 K). Furthermore, the
values of ∆G° calculated for the reactions of triptane plus
methanol to give 2,2,3-trimethylpentane plus water and the
corresponding conversion of 2,3-dimethylbutane plus methanol
to triptane plus water are identical (-13.4 kcal/mol at 200 °C).
Thus, there is no thermodynamic preference for triptane, relative
either to other C7 alkanes or to highly branched alkanes of
different carbon number.
As the conversion of MeOH/DME to hydrocarbon proceeds,
of course, still more water is formed: what is the effect on the
growth stage? To test that a mixture of DME and ZnI2 in 1.1:1
molar ratio, additionally containing small amounts of water, MeI,
and di-isobutylene, was heated at 160 °C for 4 h. The reactor
was cooled, a small sample removed for analysis, and another
charge of DME added; this was repeated for a total of five
cycles. Triptyl yields around 25% based on DME + MeI
(corresponding to 17% based on total carbon charged, including
the initiator) were found for each of the first three cycles, the
yield for the fourth cycle fell off to around 16%, and the fifth
cycle produced no additional triptyls.
A similar experiment was carried out using DME for the first
charge but MeOH for the second, and the full amount of
additional triptyls was produced in the second cycle, although
MeOH is normally unreactive at 160 °C. In contrast, when an
analogous experiment was performed using 3:1 MeOH:ZnI2 at
A different, quasi-thermodynamic explanation might be
offered: that the distribution of carbocationic intermediates
governs that of the alkane products, with the triptyl carbocation
being the most stable. However, this also is not the case:
1
Brouwer has shown by H NMR that the triptyl cation is in
fact not more stable than other C7H15 cations, comprising only
about one-fourth of the total tertiary heptyl cations at equilibrium
1
6
at -20 °C.
200 °C, the second cycle produced little or no additional triptyls.
Why then do we not see the equilibrium distribution of
products that might be anticipated once carbocations are
invoked? It is important to keep in mind that the conditions
here are far from superacidic: the pH of 5 M ZnI2 is around
These results indicate that water plays a limiting role in the
growth stage in addition to its inhibitory role in the initiation
stage. The latter explains why DME can be made to react at
temperatures considerably lower than can MeOH: much less
water is present at the outset. In subsequent reaction cycles
initiator is already present, so that MeOH reacts perfectly well
at the lower temperature. As for the former, it appears that about
four molecules of feed per Zn, corresponding to four (starting
with MeOH) or eight (starting with DME) CH2/Zn, can be
converted to hydrocarbons. In other words, reaction ceases when
four H2O/Zn have been generated, which suggests that the end
point is formation of the fully hydrated salt, [Zn(H2O)4]I2. The
1
7
1
. We would hence not expect carbocations to be at all long-
(
13) Aylett, B. J. In ComprehensiVe Inorganic Chemistry; Trotman-
Dickenson, A. F., Ed.; Pergamon Press: Oxford, 1973; Vol. 3, p 215.
14) Guempel, O. Bull. Soc. Chim. Belg. 1929, 38, 443-477.
(
(15) Rossini, F. D.; Pitzer, K. S.; Arnett, R. L.; Braun, R. M.; Pimentel,
G. C. Selected Values of Physical and Thermodynamic Properties of
Hydrocarbons and Related Compounds; Carnegie Press: Pittsburgh, 1953;
pp 467-469, 751-753.
(
16) Brouwer, D. M. Recl. TraV. Chim. Pays-Bas 1968, 87, 210-224.
“
aqueous layer” that separates out on cooling is thus perhaps
(17) Reiff, F. Z. Anorg. Allg. Chem. 1932, 208, 321-347.
8914 J. Org. Chem., Vol. 71, No. 23, 2006

ConVersion of Methanol to 2,2,3-Trimethylbutane
SCHEME 2
SCHEME 3
and RH the hydride donor; for simplicity it starts at but-2-ene
and leaves out (an extremely large number of) alternate
pathways. (The reversibility of methylation, which as discussed
earlier is required to explain labeling results but has no
consequence otherwise, is not shown.) The observed preference
for a highly branched product such as triptane is a consequence
of both methylation and proton loss taking place so as to afford
the most highly substituted carbocation and olefin, respectively,
at every stage.
lived, and the reaction would need not to be under thermody-
namic control, as it obviously is not.18 We do not yet have a
detailed model that fully accounts for high selectivity to triptane
but offer the following sketch that may serve as the nucleus of
such an explanatory schema.
Of course, this does not yet explain the selectivity for a C7
species relative to lighter and heavier branched alkanes. Note
though that in this scheme each successive olefin in the sequence
is more substituted than its predecessors, until we reach triptene,
which is the first that is less substituted. We propose that this
fact is of key importance, as a consequence of the following
trends.
First, we need to have one or more light olefins; these (or
their precursors, such as alcohols) can be deliberately added or
generated directly from MeOH during the initiation stage. The
latter appears to take place heterogenously at surface sites; we
cannot say any more about its nature at this point. When
additives such as i-PrOH or t-BuOH are used, the amount of
triptyls formed exceeds the amount of additive. That implies
that there must be a mechanism for continuous regeneration of
light olefins as the reaction proceeds. There are many possibili-
ties, including cracking of intermediate or heavier hydrocarbons;
the fact that di-isobutylene serves as initiator implies such
processes are possible. On the other hand, the observations using
benzene as initiator (see above) support Haw’s carbon pool
(1) Alkanes and olefins interconvert (but not so readily as to
equilibrate everything) as the reaction proceeds, as indicated in
Scheme 2. The more substituted olefin will be favored in any
such reaction; for example, the interconversion shown in Scheme
-1 19
3
is exothermic by about 1.1 kcal mol . Such thermodynamic
preferences will reflect kinetic preferences for the corresponding
elementary steps. As noted above, triptane is the first alkane
produced by this sequence that must revert to a less substituted
olefin than the lower hydrocarbons from which it is generated.
Hence, it will be more likely to accumulate as the alkane than
its precursors: while intermediates such as 2-methylbut-2-ene
and 2,3-dimethylbut-2-ene can get temporarily trapped as
alkanes along the way, they will be more likely than triptane to
revert to highly substituted olefins by hydrogen transfer and
become available for further growth.
8
proposal as an attractive possibility. Aromatics generated as a
consequence of hydrogen transfer would be converted to highly
methylated species, such as HMB, that could function as nuclei
for side chain C-C bond formation, followed by splitting out
light olefin and a dealkylated aromatic that would rapidly re-
methylate.
Next, we need a methylating speciessa methyl cation
equivalent. Most probably this is generated by coordination of
MeX to a Zn2 center, where MeX might be MeOH, DME, or
even MeI. The limiting role of water on conversion suggests
(2) The rate of olefin methylation generally increases with
+
the degree of substitution of the olefin as it becomes more
electron rich. For example, Mayr has shown that of the series
isobutene, 2-methyl-2-butene, 2,3-dimethyl-2-butene, and triptene,
triptene is the least reactive olefin toward the carbon-centered
2+
that to be thus active Zn must have fewer than four waters
coordinated. (It also suggests that MeX is activated by coordina-
tion rather than protonation.) Reaction of the methylating species
with olefin produces a homologous carbocation, which can either
lose a proton to give the next higher olefin or accept the
equivalent of hydride from a suitable donor (ultimately an arene
precursor to satisfy stoichiometry in hydrogen). The ubiquitous
participation of proton transfers is consistent with the observa-
tion of complete H/D scrambling in all species.
+
electrophile (p-MeOC H )(C H )CH . It must be noted that this
6
4
6
5
trend is by no means universal: Mayr found that 2,3-dimethyl-
2-butene is apparently less reactive than 2-methyl-2-butene
(
there is some ambiguity here owing to possible reversibility),
20
although it is more reactive than triptene. In another study
-methyl-2-butene was shown to be slightly less reactive than
2
21
isobutene in acid-catalyzed hydration (it was more reactive,
by an order of magnitude, in Mayr’s study). Nonetheless, it
seems reasonable to postulate that triptene would be less
A representative sequence of reactions leading up to triptane
+
is shown in Scheme 2, where XMe is the methylating species
(
18) By way of contrast, in the reaction of methanol with polyphosphoric
4
acid at 190 °C the three largest C7 components identified were just those
that are calculated from the thermochemical data to be thermodynamically
most stable (2,3-dimethylpentane, 2-methylhexane, and 3-methylhexane),
although the distribution does not precisely match the thermodynamic
predictions.
(19) Data from Handbook of Chemistry and Physics, 82nd ed.; CRC
Press: Boca Raton, FL, 2001, pp 5-26s5-36.
(20) Mayr, H.; Schneider, R.; Irrgang, B.; Schaded, Ch. J. Am. Chem.
Soc. 1990, 112, 4454-4459.
(21) Knittel, P.; Tidwell, T. T. J. Am. Chem. Soc. 1977, 99, 3408-3414.
J. Org. Chem, Vol. 71, No. 23, 2006 8915

Bercaw et al.
susceptible (kinetically) to further methylation and consequent
growth beyond C7 than the lighter olefins from which it arises.
was charged with 2.4 g of zinc iodide and a stir bar, closed,
evacuated, and refilled with argon. (If reactions are carried out in
the presence of air, products are intensely colored owing to
formation of some iodine, but there does not appear to be any
significant effect on the nature or amounts of organic products.)
The stopcock was replaced with a septum, and 1.00 mL (790 mg)
of methanol plus 50.0 µL of isopropanol, previously sparged with
argon, were injected. The septum was replaced with the stopcock,
and the reaction mixture was stirred until all solids dissolved. The
pressure tube was placed into an oil bath, behind a blast shield
(CAUTION), and stirred at 200 °C for the desired period of time,
usually 2-3 h. (It is important that the tube be immersed in the
bath as close as possible to the bottom of the stopcock. If a
significant portion of the tube is above the bath, results become
much less reproducible, presumably as a consequence of distillation
of reactants/products to the cooler regions. Reactions carried out
under different conditions are also much less reproducible than the
above protocol.)
After heating the tube was removed from the bath and allowed
to cool to ambient temperature. The stopcock was removed, and
1.00 mL of chloroform, containing a known amount of cyclohexane
as an internal standard, was pipetted into the reaction mixture
followed by 0.5 mL of water. The stopcock was replaced, the
mixture was shaken vigorously, and the organic layer was separated.
A small aliquot was diluted with acetone for GC analysis. In cases
of samples to be used for NMR analysis, chloroform-d was used
for the extraction.
22
Taken together, these considerations should significantly favor
triptane over all the other possible alkanesslighter, heavier,
differently branchedsthat could be generated by this sequence.
However, whether the entire speculative picture might in fact
be the basis for a plausible detailed mechanistic explanation
will probably need to be addressed by modeling both to test
the validity of the two above postulates and to determine whether
the kinetics they imply are capable of reproducing the observed
distributions. We hope to initiate such studies in the near future.
As for the special23 ability of ZnI2 to promote this reaction,
we believe its extremely high solubility in aqueous/alcoholic
2+
media, along with the Lewis-acidic character of the Zn centers
and/or the Brønsted acidity of the concentrated solutions, are
central. The further ability of concentrated ZnI2 solutions to
dissolve substantial quantities of organics (see above) may also
be important as the growth stage of the reaction appears to take
place homogeneously. Whether iodide plays an additional roles
for example, to convert MeOH into MeI which participates
directly at one or more stages of the reactionsis unclear at this
point. (MeI certainly is formed, but that could be a nonessential
side reaction.) The fact that some MeI must be added to reactions
of DME may suggest it does play an intimate role in the
initiation phase. As noted earlier, ZnBr2 also catalyzes triptane
formation, albeit at higher temperatures; this could be consistent
with either a parallel role for (less reactive) MeBr or an MeX-
independent pathway, with ZnBr2 less potent than ZnI2. Again,
modeling may be the best approach to distinguish between these
alternatives.
For reactions of dimethyl ether, all ingredients except DME were
loaded into the tube, which was then frozen in liquid nitrogen and
degassed, and the desired amount of DME was condensed in; the
tube was allowed to warm to room temperature and then heated as
usual. DME is only converted to triptane if small amounts of water
(5 wt% or more) are added to the reaction mixture along with ZnI
and promoter.
2
While we are still well short of complete understanding of
this interesting (and potentially useful) transformation, our
findings do provide a mechanistic framework that can account
for the unexpectedly selective formation of triptane from
methanol and may offer useful parallels for important processes
such as MTG and MTO, which mechanistically are not yet well
understood either.
2 2
For consecutive reactions, a tube containing DME, ZnI , H O,
MeI, di-isobutylene (promoter), and n-octane (internal standard)
in molar ratio 1.1:1:0.13:0.38:0.16:0.014 was prepared as described
above and heated for 4 h at 160 °C. The tube was allowed to cool;
a small sample was removed for GC analysis; a fresh charge of
DME was condensed in, and the cycle was repeated. In another
series, after four cycles all volatiles were removed to give an off-
white solid, to which was added a fresh charge of DME and MeI,
and the cycle was repeated.
Experimental Section
Synthesis of Triptane from 12CH
12
13
3
3 2
OH + ( CH ) CHOH. A
1
3
General. Zinc iodide, methanol, dimethyl ether, and other organic
2
standard reaction was carried out in the presence of added Me -
compounds used as initiators were reagent-grade commercial
CHOH (100 µL, 79.8 mg, 99% 13C), and the organic layer was
1
13
13
samples used without further purification. H and C NMR spectra
were obtained on a Varian 300 MHz instrument; GC analyses were
done on an HP model 6890N chromatograph equipped with a 10
m × 0.10 mm × 0.40 um DB-1 column.
extracted into CDCl
3
for analysis by C NMR. The triptane signals
are identified according to the labels shown in eq 2. The ap-
proximate intensity ratios for C :C :C :C
a
b
c
d
≈ 1.5:1:2.5:1, in contrast
to the values observed for an ordinary unlabeled reaction (with the
same spectrometer settings): 8.7:1:1.3:7. Quantitative interpretation
is not possible since in addition to the common problems associated
with reliable quantitation of 13C signal intensities there is no way
to assess the relative contributions of added label and natural-
abundance 13C. However, it is clear that much more label has been
Standard Reaction Protocols. Reactions were carried out in
thick-walled pressure tubes equipped with Teflon stopcocks (Ace
Glassware), rated up to 10 bar. In a typical experiment, the tube
(22) A referee has suggested this trend might have a steric component
b c
incorporated in the two internal positions, C and C , since their
as well. The results cited immediately above indicate that a steric effect
can compete with the positive electronic effect of increasing substitution;
perhaps some unfavorable factor specific to methylation of triptene, such
as the 1,3-interaction between the incoming methylating species and one
of the methyls of the adjacent tert-butyl group, could inhibit growth beyond
the C7 stage. We plan to examine this and related questions by means of
molecular modeling in the near future.
signals are enhanced relative to the methyl signals by 5-10-fold
compared to the unlabeled reaction.
13
Gaseous Products from CH
3
3 2
OH + (CH ) CHOH. A reaction
completely analogous to the preceding one but with the labels
reversed was carried out. After cooling the gaseous products were
condensed into a liquid-nitrogen-cooled tube and dissolved in CDCl
3
(23) However, not unique: several other metal iodides can also effect
1
conversion of methanol to triptane under similar conditions, most notably
InI3 (Kay, R. D.; Morris, G. E.; Sunley, J. G. PCT/GB2004/003578, 2005).
There is also a much older report of the homologation of 2-methylbut-2-
ene to higher olefins, including a substantial amount of triptene (but no
alkanes), by treatment with MeCl (or MeI) and stoichiometric amounts of
CaO above 200 °C (GB Patent 577,112, 1946). We will report on
mechanistic studies for InI3, and its similarities to/differences from ZnI2,
in a later publication.
for analysis by NMR. The H NMR shows strong signals for
13
(unlabeled) propene. In the C NMR, the region corresponding to
3
saturated (sp ) carbons is complex, but two signals stand out in the
olefinic region. The stronger one, centered around 111 ppm, consists
1
2
of a doublet ( JCC ≈ 71 Hz) of triplets ( JCC ≈ 3 Hz) along with a
2
much weaker triplet in the middle ( JCC ≈ 3 Hz); the relative total
intensities are about 85:15. A considerably weaker signal around
8916 J. Org. Chem., Vol. 71, No. 23, 2006

ConVersion of Methanol to 2,2,3-Trimethylbutane
1
1
43 ppm is a doublet (1JCC ≈ 72 Hz) of triplets ( JCC ≈ 41 Hz).
was placed in an ice batch, and a saturated aqueous solution of
NH Cl (100 mL) was slowly added. The organic layer was
separated, and the aqueous layer was further extracted with ether
The shifts and C-C coupling constants agree well with those
4
reported for isobutene,24 and the specific patterns are only consistent
13
13 13
13
12
with an 85:15 mixture of CH
2
d C( CH
3
)
2
and CH
2
d C
(2 × 20 mL). The combined organic fractions were washed with
1
3
(
3
CH )
2
.
4
brine and dried over MgSO . The ether solution was treated with
Mass Balance Determination. ZnI
2
(2.27 g, 7.11 mmol),
2
CaH (2.0 g, 48 mmol) and stirred at room temperature until
isopropanol (50.0 µL, 0.654 mmol), and methanol (0.772 g, 24.1
mmol) were reacted for 3 h as described above. The cooled tube
was weighed, then the stopcock removed to allow volatile products
to escape, and the tube re-weighed. A 90.4 mg amount was lost as
volatiles. Separate analysis of the volatile fraction identified propene
as the major component in addition to DME.
The reaction flask was fitted with a septum, and the organic phase
was carefully transferred via cannula into a round-bottom flask
4
previously charged with MgSO . The organic fraction was dried,
and a weighed portion (173.4 mg) was analyzed by GC. The
analysis showed triptane and triptene as the major components,
combining for 31.2 mg (23% w/w) of the organic layer. The next
most abundant compound was hexamethylbenzene at 2.0% w/w.
From previous experiments, analyzed as described above, the total
yield of triptyls from such a reaction should be 67.6 mg; hence,
the actual weight of the organic layer was estimated to be 297 mg.
The bottom layer of the reaction mixture, after removal of the
hydrogen evolution stopped. The mixture was filtered and the
13
solvent removed under reduced pressure to give (CH
3
)
3
CC(CH
3
)( -
13
3
CH )OH ( C-labeled triptanol, 4.3 g, 63%) as a colorless liquid.
A 250 mL round-bottom flask equipped with a magnetic stirrer
and an addition funnel was charged with the 13C-labeled triptanol
(4.3 g, 37 mmol), 1-methylimidazole (10 mL, 125 mmol), and
methylene chloride (40 mL). The mixture was cooled in an ice
bath, and a solution of methylsulfonyl chloride (4.4 mL, 56 mmol)
in methylene chloride (10 mL) was added dropwise. After the
addition was complete, the addition funnel was replaced with a
reflux condenser, and the reaction was stirred at reflux overnight.
An aqueous HCl solution (1.0 N, 100 mL) was added to the reaction
mixture, and the organic layer was separated, washed with a
saturated solution of NaHCO
3
followed by brine, and dried over
MgSO
4
. Distillation provided 13C-labeled triptene as a clear liquid,
shown by NMR to be a 1:1 mixture of isotopomers (CH
3
)
3
CC-
13
13
1
(CH
3
)( CH
2
) and (CH
3
)
3
CC( CH
3
)(CH
2
). H NMR: δ 4.74 (d,
1
13
3
organic layer, weighed 2.68 g, corresponding to 0.41 g of liquids
1H, JC-H ) 150 Hz, C(CH
Hz, C( CH
3
)( CHH′)), 4.74 (d, 1H, JC-H ) 7.5
1
13
1
13
after subtracting the original weight of ZnI
2
. H NMR analysis of
3
)(CHH′)), 4.66 (d, 1H, JC-H ) 150 Hz, C(CH
3
)( -
3
13
this fraction, using added nitromethane as an internal standard,
identified methanol (14% w/w), DME (9% w/w), and water as the
major components. Neglecting the amount of DME lost with the
volatiles, the conversion of the reaction is calculated to be 86%.
Methylenation of 2,3-Dimethylbut-2-ene. Reaction mixtures of
CHH′)), 4.66 (d, 1H, JC-H ) 7.5 Hz, C( CH
3
)(CHH′)), 1.77 (d,
1
13
3
3H, JC-H ) 126 Hz, C( CH
3
)(CH
)), 1.08 (s, 18H, (CH
A pressure tube was charged with ZnI
2
)), 1.77 (d, 3H, JC-H ) 6 Hz,
1
3
C(CH
3
)( CH
2
3 3
) C)
2
(294 mg, 0.92 mmol),
methanol (130 mg, 4.1 mmol), and 13C-labeled triptene (50 µL,
0.36 mmol). The tube was sealed with a stopcock and placed into
an oil bath for 180 min at 160 °C. The vessel was then allowed to
cool to room temperature, and the reaction mixture was further
2
methanol (2.4 g), 2,3-dimethylbut-2-ene (1.07 g), and ZnI (7.2 g)
were heated in glass pressure tubes at 160 °C for varying periods
of time, and the organic product layer was analyzed by GC in the
usual manner. After 3 h approximately 60% of the starting olefin
had been converted, increasing to 90% by 17 h. The main products
identified were triptyls, comprising 20-30%, and isomerized olefin
6
diluted with water (0.2 mL) and benzene-d (0.8 mL). NMR analysis
of the organic layer showed that the 13C label was statistically
distributed among the four methyl and one methylene positions.
Hydrogenation of Triptene by Cyclohexa-1,4-diene. A pressure
2,3-dimethylbut-1-ene, around 5-10%. Substantial amounts of
heavier products were apparent in the GC trace. The triptane:triptene
ratio was initially very low, around 1:10, but increased over time.
A similar experiment was performed on a smaller scale, using
2
tube was charged with ZnI (2.40 g), MeOH (1.0 mL, 791 mg),
1,4-cyclohexadiene (42.4 mg, 0.529 mmol), and triptene (52.9 mg,
0.538 mmol). The mixture was stirred to dissolve all solids, and
the tube was heated at 150 °C for 2.5 h. After cooling to room
temperature, the organic layer was analyzed by GC in the usual
manner, using cyclohexane as an internal standard. Triptene was
almost completely consumed with formation of triptane and benzene
in yields (based on 1,4-cyclohexadiene as a limiting reagent) of
92% and 72%, respectively. The GC chromatograph also indicated
the presence of some higher boiling components, probably methyla-
ted benzenes. Under these conditions no reduction of triptene to
triptane is observed in the absence of cyclohexadiene nor is there
13
3
CH OH, and analyzed by both NMR and GC-MS, demonstrating
that label appeared in all peripheral positions (methyls, methylene)
of triptane and triptene but not in any internal positions and that
13
each product molecule contained exactly one C atom, within the
accuracy of the GC-MS analysis.
Synthesis and Isomerization of Regiospecifically 13C-Labeled
Triptene. A 500 mL three-neck round-bottom flask equipped with
a magnetic stirrer, a reflux condenser, and an addition funnel was
charged with finely divided magnesium turnings (2.4 g, 99 mmol)
and flushed with argon. Dry diethyl ether (10 mL) was placed over
the magnesium, and the addition funnel was charged with 13C-
labeled methyl iodide (10 g, 70 mmol) and ether (20 mL). The
solution was added dropwise to the magnesium suspension and
further diluted with ether fractions (2 × 10 mL) 10 min apart. After
the addition was complete, the reaction was refluxed for 15 min.
The mixture was then cooled in an ice bath, and a solution of
pinacolone (7.5 mL, 60 mmol) in ether (20 mL) was added dropwise
via the addition funnel. After the addition was complete, the reaction
was stirred for additional 30 min at room temperature. The mixture
2
any formation of triptyls from methanol and ZnI alone.
Acknowledgment. Work at Caltech was supported by BP
2
through the MC program. We thank Steve Cook, Gail Heath,
and Matthew Deaves for assistance in performing some
preliminary experiments on this system, Eugene Zaluzek for
obtaining the PIANO analysis, and Philip Howard, Enrique
Iglesia, Tom Knox, and Theo Fleisch for useful discussions.
Supporting Information Available: Full analysis and discus-
13
sion of the experiment involving di- C-labeled ethanol. This ma-
terial is available free of charge via the Internet at http://pubs.acs.org.
(24) Kalinowski, H.-O.; Berger, S.; Braun, S. Carbon-13 NMR Spec-
troscopy; John Wiley & Sons: Chichester, 1988.
JO0617823
J. Org. Chem, Vol. 71, No. 23, 2006 8917