208
N.C. Smythe et al. / Inorganic Chemistry Communications 61 (2015) 207–209
Table 2
3
Negative of the fluoride affinity of AlR calculated at the density functional theory level.
FPD benchmark values in parentheses.
R
FA (kcal/mol)
R
FA (kcal/mol)
H
F
Cl
Br
I
84.8 (90.3)
112.0 (113.9)
116.3 (118.8)
110.4 (119.4)
110.7 (120.6)
79.5
O–C(CF
3
)
3
129.5
121.4
114.2
111.4
91.3
O-C
O-C
O-C
6
6
6
Cl
5
H
H
2
Cl
3
5
3 7
O-i-C H
CH
3
O-t-C
O-Et
4
H
9
87.8
90.7
C
2
H
5
81.5
O–Me
89.4
O-C6F5
128.7
further work was conducted using PCy3 despite the expected higher
binding energy (more electron-rich phosphine). PCy3-(Cl-3) shows
slow C–H activation of CH
the 31P NMR spectrum. This was verified using CD
information).
Testing for CO
solution or mixture of PCy3 and Lewis acid to ca. 1 atm. of CO
2
Cl
2
, with the appearance of [HPCy3] + in
2
Cl (see supporting
2
3
Fig. 1. X-ray crystal structure of PPh -(Cl-3) with hydrogen atoms omitted and 50% ther-
mal ellipsoids.
2
reactivity was accomplished by exposing a degassed
13
2
and
then examining the 13C{ H} and P{ H} NMR spectra. In contrast to
Al(OiPr) , Al(OtBu) , and Al(OPh) , F-5, Cl-3, and Cl-5 all bind CO
2
1
31
1
bonding interactions [25]. Attempts to quantify the Lewis Acidity trend
via the [31]P NMR shift of OPEt were ambiguous due lack of clear 1:1
OPEt -AlOR adduct formation.
3
3
3
13
3
31
3
3
3
with diagnostic P– C coupling when paired with PCy (see Fig. 2 as
Density functional theory (B3LYP/DZVP2) [28–30] and G3MP2 [31]
an example).
electronic structure calculations of the bond dissociation free energies
The Al systems structurally characterized by Stephen et al. show a
2:1:1 Al:CO :P ratio [10,12]. In contrast, reports of known B-based
Lewis acid FLP-CO complexes predominantly have a 1:1:1 B:CO :P
ratio [1]. This difference likely arises from the decreased Lewis acidity
and oxophilicity of B relative to Al. This would also be consistent with
known pre-organized P-Al complexes lacking electron-withdrawing
of AlX /PR
3 3
3 3
(X = H,F,Cl,Br,CH ; R = H, CH ) adducts (Table 1) are not
2
consistent with a clear trend along the halogens (which agrees with
the varied rankings available in the literature), but it is clear that more
2
2
electron-rich phosphines bind more tightly (PH
more electron-rich alanes bind more loosely (AlH
X = F, Cl, Br).
3
vs. PMe
3
) and that
vs. AlX
3
vs. AlMe
3
3
,
groups binding CO
2
through only one aluminum center. While we
adduct, the solu-
:P adducts are con-
sistent with a mixture consisting of mostly the 1:1:1 complex when
The negativity of the fluoride affinity (FA = ΔH(298) for the reaction
were unable to obtain a solid-state structure of a CO
2
−
−
13
1
31
1
A + F → AF ) of a compound has been shown to be a good measure of
its Lewis acidity [34,35]. The fluoride affinities of the types of AlR
tion C{ H} and P{ H} NMR spectra of our Al:CO
2
3
13
compounds being discussed have been calculated at the density
functional theory B3LYP/DZVP2++//B3LYP/DZVP2 (for Br and I,
the aug-cc-pVDZ-PP basis sets were used) level with the values for
CO
pected that the CO
two Lewis acids are coordinated to the CO
2
is added to a 1:2 PCy
carbon atom would have less electron density when
moiety rather than one. This
3
:Cl-3 mixture (Est. 5–9% bis). It would be ex-
2
2
AlR
3
, R = H, F, Cl, Br, I, benchmarked at the CCSD(T) FPD level
is consistent with the observed shift in the NMR spectra where the car-
bon atom is effectively deshielded. Taking this view, this Al ester is more
(
Table 2). The computational details are given in the Supporting In-
formation. The CCSD(T) FPD results show that the DFT B3LYP results
are semi-quantitative. The aluminum esters have FA's that range
from moderately strong ~90 kcal/mol (O-tBu, O-i-Pr) to very strong
6 5 3 3
akin to Lewis acids such as B(C F ) than AlX in terms of FLP reactivity.
As the aryl substituent is adjusted going to Cl-5 (Est. ca. 1:1 bis:mono)
and then F5 (Est. 23–26% mono), this balance changes and the mixture
becomes more biased towards bis coordination. This is a trend that
tracks with what one may predict based on trends in substituent
electron-withdrawing ability. The stronger the Lewis acid, the greater
~
120–130 kcal/mol (O-C
6 5 3 3
Cl , O–C(CF ) ). The latter are stronger
Lewis acids than SbF [34]. The calculations were done with the Gauss-
5
ian and MOLPRO program systems [36,37],
Similar to PPh3-Al(OC(CF3))3 [26], we did not observe splitting of
the average number of acids there are per CO
Regardless of the ratio, addition of NH BH
5, Cl-5, Cl-3) systems resulted in complete conversion of PCy
2
.
2
H2 with PPh3-F-5, but the PPh3 adduct did bind CO . However, these
3
3
to PCy
3
/CO
2
/Al(OR)
3
(F-
early experiments displayed indications of aryl C–H activation and so
3
to
+
31
1
31
13
1
[
HPCy
NMR spectra show a methoxy group (presumably coordinated to
Al(OR) ) along with formate in the case of F-5 and Cl-3. Addition of
2
H O to these mixtures releases MeOH, but also results in the decompo-
3
]
as determined by P{ H} and P NMR spectra. C{ H}
Table 1
3
B3LYP/GZVP2 and G3MP2 gas-phase dissociation energies (kcal/mol) of AlX
adducts .a
3
/PR
3
a
sition of the aluminum esters.
Aluminum esters derived from halogenated aromatic alcohols were
synthesized and tested as Lewis acids for the FLP-mediated reduction of
Lewis acid–base adduct
B3LYP (ΔG(298 K))
G3MP2 (ΔG(298 K))
AlH
AlH
AlMe
AlMe
3
3
PH
PMe
3
−0.2
7.1
−4.8
1.2
11.6
26.7
6.2
21.0
4.5
4.8a
17.3
−0.1
10.6
9.8
25.5
8.8
27.0
8.8
3
3
CO
gard when paired with PCy
due to the need to quench the resulting complex with H
release MeOH. It may be possible to use stoichiometric quantities of
O to preserve these esters [26], however, the FLP would still need to
be regenerated from the resulting H O adduct. These results reinforce
2
using NH
3
BH
3
. While F-5, Cl-5, and Cl-3 were competent in this re-
, the reaction could not be made catalytic
O in order to
3
PH
3
PMe
3
3
AlF
AlF
3
PH
PMe
3
2
3
3
AlCl
AlCl
3
PH
PMe
3
H
2
3
3
2
AlBr
AlBr
3
PH
PMe
3
3
3
18.7
27.8
the need to develop systems that can release MeOH under mild
thermolysis rather than form Al (and B) oxides and hydroxides which
require much more energy to regenerate into usable FLPs. This is a lim-
a
A highly accurate calculated value [32] for ΔH298 obtained at the Feller–Peterson–
Dixon (FPD) level [33] is 14.7 kcal/mol as compared to the G3MP2 value of 13.2 kcal/mol
and the B3LYP value of 10.8 kcal/mol.
2
itation that even catalytic FLP-mediated CO reduction to MeOH suffers