836
C. Guo, J. Persons and G. S. Harbison
respectively; the average of these two yields a G298 of
stabilization of the D
putations correctly predict that D
than C , the agreement between computed and experimental
3
conformer by solvation. While com-
4
.0 š 0.4 kJ/mol for the reaction.
3
has a lower free energy
2
free energy – 1.6 kJ/mol vs 4.0 kJ/mol – is not particularly
good. (Note that the earlier energetic computations, in con-
DISCUSSION
1
cert with the increased configurational entropy of C , would
2
The existence in normal (achiral) solution of two sets of H
1
3
3
actually imply that it, and not D , is the stable conformer!)
It is plausible that the D form might have a 2 kJ/mol or
3
so greater van der Waals energy of interaction with the
solvent, and this could account for the computational under-
estimation of the free energy difference between the two
forms.
The above discussion, of course, assumes that the con-
former has reached equilibrium at room temperature. In
order to assure this, we attempted to obtain spectra upon
rapid dissolution of the crystalline material. Although we
were able to acquire data within a minute of dissolution of
the material, unaltered proportions of the two conformers
were detected; nor was any change in the relative intensities
seen over several days at room temperature. This suggests
that equilibrium is rapidly achieved in solution at room tem-
perature (although a possible alternative is that the crystal
and C resonances with the expected multiplicities, and the
twofold further splitting of these resonances in the presence
of the chiral shift reagents (since there are no enantiotopic
pairs of protons in the molecule), are powerful evidence in
favor of the predicted coexistence of two diastereoisomeric
pairs of enantiomers, exchanging slowly on the NMR time
scale, in solution at room temperature. Further support lies
in a comparison of experimental and computed chemical
shifts, which are in remarkably good agreement, especially
considering no corrections for vibration are being done.
Theory reproduces the experimental geminal J couplings
within 1%, and duplicates in magnitude the very small
experimental difference between diastereotopic protons in
3 2
the D conformer. For the C , while it is impossible
to make an absolute assignment of the three distinct
1
3
1
2
methylenes in the structure to the three C H spin clusters
structure, which accommodates both D
3
enantiomers with
in the experimental data, the close agreement between
the three sets of three measured parameters, and their
computed values, makes our tentative assignment highly
persuasive.
In every case, theory predicts the equatorial proton
shift to lie downfield from the axial. There are no clear
correlations between proton or carbon chemical shifts
and bond lengths or angles, although the methylene
with the largest carbon downfield shift and the largest
difference between axial and equatorial proton shifts is
5
0% occupancy in an achiral lattice, can also accommodate
the C
2
structure). It is notable that in DMSO above 40 °C,
the C
2
peaks disappear, apparently as a result of exchange
with the D
and H1eq signals of the C
tant from the D resonances, in fact, appears to be a result
3
form. The somewhat larger width of the C
1
2
conformer, which are most dis-
3
of exchange on a time scale of hundreds of milliseconds,
meaning that the exchange is not quite in the slow limit at
4
room temperature. Calculations by Wierzbicki et al., which
the methylene adjacent to the ‘unique’ peroxide in the C
2
are unfortunately given without much computational detail,
structure – i.e. that with the opposite O–O torsion angle
from the other two. An examination of the effect of the
chiral shift reagent indicates that the largest shift-reagent-
induced difference between enantiomeric proton signals is
in every case for the equatorial proton, and the largest
differences are encountered for the ‘unique’ methylene.
Significantly, there seems to be a correlation between the
shift of this proton and the angle between the C–H vector
and the N–N axis, with the unique methylene having
the ‘most equatorial’ proton – the one with the largest
angle – showing the largest chiral shift difference. This
suggests that differential enantiomeric interaction between
the shift reagent and HMTD is predominantly along the
molecular equator. This agrees with the molecular structure;
give barriers of the order of 60–70 kJ/mol for the C
2 2
–C
and C –D exchange reactions, reasonable for a process that
2
3
occurs on this timescale at room temperature, but more
detailed kinetic studies are clearly needed.
One major disagreement of the present work with ear-
lier computational studies lies in the origin of the nitrogen
6
planarity. Wierzbicki and Cioffi, on the basis of a com-
puted CNC of 117.7
°
for tris-(hydroperoxymethylene)amine
(THPMA), concluded the planarity of the nitrogen in HMTD
was a result of steric strain, since the nitrogen in THPMA is
more pyramidal. Since the nitrogen in THPMA is nowhere
near as pyramidal as that of trimethylamine, this conclusion
is questionable in any case; but it is clear that calculations at
our higher level of theory, and for a chemically somewhat
superior acyclic model compound, result in a considerably
more planar nitrogen than computed in the earlier work.
In fact, the displacement of the nitrogen from the plane of
the directly bonded carbons in TMPMA is only a quarter of
that computed in trimethylamine. To a first approximation,
this is a planar, not a pyramidal nitrogen, and therefore
we conclude that the major cause of nitrogen planarization
in HMTD is electronic, not steric. This is not a particularly
surprising observation; the difference in energies between
planar and pyramidal amines is not large even in ammonia,
and electron withdrawing substituents tend to stabilize the
‘
end-on’, the molecule is essentially achiral, while the band
of peroxides around the equator is responsible for the
chirality.
The energy difference between the two conformers,
5
.9 kJ/mol, is significantly greater than that computed
by previous workers at a lower B3LYP/6-31CG(d) level
0.3 kcal/mol or 1.3 kJ/mol). Inclusion of vibrational ener-
(
gies and entropies significantly reduces the difference
between the two conformers, as does the R ln 3 entropic
contribution from the three equivalent C structures, which
2
is inherent in the ‘symmetry number’ of the gas-phase rota-
tional entropy. PCM computations predict some differential
2
sp hybridized planar conformation.
Copyright 2006 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2006; 44: 832–837
DOI: 10.1002/mrc