Organometallics
Article
chlorobenzene o-CH bonds, are not included in this
correlation, as the steric effects of the substituents in these
substrates, Me and Cl, appear to be overestimated in our DFT
calculations of the reaction 5 Gibbs energy of activation (vide
infra). The 2.3 kcal/mol correction allows us to attain the root-
mean-square deviation (RMS) of the DFT-calculated corrected
values from the experimental values of 0.5 kcal/mol, which is
small, in comparison to some typical accuracies attainable
currently in DFT calculations of organometallic reactions.18 In
turn, the data points for p-xylene and chlorobenzene o-CH
bonds deviate from the solid line on the plot in Figure 4a by
4.2 and 5.7 kcal/mol, respectively, both being greater than
three RMS values. Interestingly, a more “compact” fluorine
substituent does not exhibit such issues. The use of a
dispersion-corrected DFT functional PBE-D319 for single-
point calculations corresponding to the level of theory PBE-
D3/LACVP**//PBE/LACVP** results only in a minor
With the exception of thiophene α-CH bonds and N,N-
dimethylaniline m-CH bonds, the remaining data in Figure 4a
fall into a relatively narrow range ΔG353⧧(exp) − 2.3 RMS
(RMS = 0.5 kcal/mol); this range is shown in Figure 4a with
two dashed lines with slope 1 each. The point corresponding
to N,N-dimethylaniline may be slightly off of the trend as a
consequence of the ability of this compound to become
involved in an acid−base equilibrium with TFE, so that a
sizable part of N,N-dimethylaniline is converted to the
corresponding cation, which has a different reactivity that is
not accounted for in our DFT modeling. Qualitatively, the
cation derived from N,N-dimethylaniline should be more
electron poor than the aniline itself and, on the basis of the
overall reactivity trend mentioned earlier, be less reactive.
Turning now to the DFT-calculated Gibbs energies of
activation of reaction 5 in TFE solutions, ΔG298⧧(DFT/TFE),
and holding in mind small solvation effects calculated for
reaction 5, as noted earlier, we can obtain an expected good
match of the calculated values and the experimental values,
ΔG353⧧(exp), when the former values are corrected (increased)
by 2.3 kcal/mol (Figure 4b; the solid line with slope 1). The
root-mean-square deviation of the DFT-calculated corrected
values from the experimental values is 0.7 kcal/mol (Figure
4b). Once again, a significant deviation from the solid line in
Figure 4b of the calculated values ΔG298⧧(DFT/TFE) for p-
xylene and chlorobenzene o-CH bonds, by 4.7 and 5.1 kcal/
mol, respectively, may suggest that the DFT model used here
overestimates the steric effects of the Me and Cl fragment in
these substrates in our calculations of the Gibbs energy of
activation of reaction 5.
substituent X, the negative sign is used in front of the
logarithmic function:
σXM = −log(kc(X)/kc(H))
The corresponding σX values are given in Table 2. Except
(6)
M
M
NMe2 group, the series of σX constants in Table 2 is
M
Table 2. Hammett-Type σX Constants for Metal-Assisted
C−H Bond Activation Reactions Based on Rate Constant kc
Data from Table 1
substituent X
OMe
OH
H
Cl
CO2H
0.43
F
NO2
σXM, meta
σXM, para
−0.10
−0.03
0
0
0.33
0.40
0.68
0.72
1.29
0.88
qualitatively similar to that of σX values. An unexpectedly large
positive value was calculated for the NMe2 group, 0.34 (not
shown in Table 2), which may be related to the group’s partial
protonation in TFE solutions, and therefore, this value is not
recommended for general use.
2.7. Regioselectivity of H/D Exchange (Reaction 1).
Notably, the initial regioselectivity of the H/D exchange
observed in reaction 1 typically is low, as follows from an
inspection of the rate constants kc given in Table 1. For
example, for chlorobenzene the kc,o (ortho-CH bonds), kc,m
(meta-CH bonds), and kc,p (para-CH bonds) values fall in a
narrow range of 2.2−3.2 M−1 h−1. For fluorobenzene the
reaction rate constant for o-CH bonds, kc,o = 0.7 M−1 h−1, is
only 2 times lower than the value for its m-CH bonds, kc,m
=
1.3 M−1 h−1, or p-CH bonds, kc,p = 1.43 M−1 h−1, which, in
turn, are very similar.
There are a few notable exceptions from this “low
selectivity” picture. In particular, nitrobenzene o-CH bonds
are virtually unreactive under our reaction conditions,
presumably due to both the electronic and steric effects of
this group, whereas the p-CH bonds, kc,p = 0.9 M−1 h−1, are
somewhat more reactive than the meta-CH bonds, kc,m = 0.35
M−1 h−1.
Two other substrates demonstrating notable positional
selectivity in their CH bond deuteration (1) are N,N-
dimethylaniline and indole. The relatively low reactivity of
m-CH bonds of N,N-dimethylaniline makes it possible to carry
out a selective deuteration of the CH bonds in the ortho and
para positions of this substrate after 27 h of the reaction
(Figure 5). An exclusive deuteration of the indole β-CH bonds
with a virtually statistical 88% degree of deuterium
incorporation was observed in our experiments for indole
(17) after 45 min of reaction.
To illustrate qualitatively some possible reasons behind the
observed (often low) regioselectivity of complex 1 in the H/D
exchange reaction (1) that would be consistent with the
proposed reaction mechanism (Scheme 2), we considered the
following two cases, one involving deuteration of α- vs β-CH
bonds of thiophene (TP) and another involving deuteration of
the m- vs p-CH bonds of nitrobenzene (NB). Experimentally,
the deuteration of thiophene α-CH bonds and the deuteration
of nitrobenzene p-CH bonds is slightly selective (Table 1 and
Chart 2). For thiophene the difference in the reactivity of α-
and β-CH bonds is reproduced qualitatively in our DFT
calculations of the Gibbs energy of activation of reaction 5; for
nitrobenzene the DFT predicts virtually no difference (Table 1
and Figure 4).
2.6. Novel Hammett Constants σXM for Metal-Assisted
C−H Bond Activation Reactions. The inability to describe
substituent effects in the organometallic reaction (5) using
Hammett σX constants and the somewhat limited accuracy of
modern DFT calculations, as demonstrated in Figure 4 and
shown by other researchers,18 prompted us to introduce a new
empirical set of Hammett constants, σXM, for metal-assisted
CH bond activation reactions. Such constants may be useful in
analyzing the reactivity of various metal complexes and
M
substrates in CH activation reactions. Our σX constants are
defined by eq 6. Since reaction 5 is decelerated by electron-
withdrawing groups, for the sake of consistency of the sign of
M
σX values and the sign of classical σX values for the same
H
Organometallics XXXX, XXX, XXX−XXX