β-Cyanoethyl anion: Lusus naturae

Article abstract of DOI:10.1021/ja953796o

A stable El(cb) intermediate, β-cyanoethyl anion (1β), has been synthesized in the gas phase at room temperature under thermal conditions via the fluoride-induced desilylation of 3-(trimethylsilyl)propionitrile. The reactivity and thermodynamic properties of this ion are reported. The cyano group is found to lower the proton affinity of 1β by 29 ± 6 kcal/mol, which represents a particularly large substituent effect. High-level ab initio and density functional calculations have been carried out on 1β and several related species. The computational results are compared to each other, and their accuracy is evaluated.

Full text of DOI:10.1021/ja953796o

4
462
J. Am. Chem. Soc. 1996, 118, 4462-4468
â-Cyanoethyl Anion: Lusus Naturae
Grant N. Merrill, Gregg D. Dahlke, and Steven R. Kass*
Contribution from the Department of Chemistry, UniVersity of Minnesota,
Minneapolis, Minnesota 55455
X
ReceiVed NoVember 13, 1995
Abstract: A stable E1cb intermediate, â-cyanoethyl anion (1â), has been synthesized in the gas phase at room
temperature under thermal conditions via the fluoride-induced desilylation of 3-(trimethylsilyl)propionitrile. The
reactivity and thermodynamic properties of this ion are reported. The cyano group is found to lower the proton
affinity of 1â by 29 ( 6 kcal/mol, which represents a particularly large substituent effect. High-level ab initio and
density functional calculations have been carried out on 1â and several related species. The computational results
are compared to each other, and their accuracy is evaluated.
-
Elimination reactions are common processes which have been
An E1cb intermediate of the form CH2CH2X, where X is a
leaving group, is a â-substituted alkyl anion. The related
unsubstituted species are difficult to prepare in the gas phase
because they are extremely strong bases and tend to be unstable
with respect to their corresponding radicals. For example, ethyl,
n-propyl, isopropyl, and tert-butyl anions are all unbound.
These ions can be stabilized relative to their corresponding
radicals and conjugate acids, however, by incorporating an
electron-withdrawing substituent. Leaving groups invariably are
electron withdrawing, so the electron binding energy of a
substituted anion will be larger, its basicity will be less, and if
there is a large enough barrier to elimination, then a stable ion
will result. We now report a rare instance of a stable
â-substituted alkyl anion in the gas phase which can undergo
extensively examined.1-3 Mechanistic investigations indicate
that a range of pathways from E1cb to E1 can take place, as
originally described by Cram2 and Bunnett
c
2a,b
in what is
3
commonly referred to as the variable E2 transition state model.
Reactive intermediates are formed in both of these limiting
cases, and in the latter instance they (carbenium ions) can be
generated as long-lived species in non-nucleophilic and nonbasic
media.4 In the E1cb pathway, carbanions with a â leaving group
are produced. These ions usually are unstable, and this makes
them very difficult to study. The transient nature of these
species could be an intrinsic limitation, but it may be just an
environmental effect. Computations and gas-phase experiments
are of interest in this regard. The latter approach is particularly
promising since elusive species in solution can often be
generated in the gas phase. For example, we have made
1
0-12
elimination (i.e., an E1cb intermediate).
-
5
- 6
thiomethyl anion ( CH2SH), diazirinyl anion (c-CHN2 ),
Experimental Section
-
7
dehydrophenoxide (C6H3O ), and a cyclopropenyl anion (c-
C3H2R ) in the gas phase, whereas these ions are either
unknown or highly elusive in condensed media.
-
8
The gas-phase experiments reported in this work were carried out
with a variable temperature flowing afterglow apparatus which has been
described previously.5
,13
Briefly, ions are generated by electron
X
Abstract published in AdVance ACS Abstracts, April 15, 1996.
ionization and are carried down a 1 m long tube by a rapidly moving
stream of helium buffer gas ( Vj He ) 8-10000 cm/s, P ∼ 0.4 Torr).
Neutral reagents can be added to the system at numerous points along
the 1 m long reaction region so that a sequence of ion/molecule reactions
can be carried out. The charged products are subsequently mass filtered
and detected with a triple quadrupole-conversion dynode/electron
multiplier setup.
(
1) (a) Gandler, J. R. The Chemistry of Double-bonded Functional
Groups; Patai, S., Ed.; Wiley: New York, 1989; Vol. 2, Part I. (b) Reichardt,
C. SolVents and SolVent Effects in Organic Chemistry, 2nd ed.; VCH: New
York, 1988. (c) Bartsch, R. A.; Z a´ vada, J. Chem. ReV. 1980, 80, 453. (d)
Alekserov, M. A.; Yufit, S. S.; Kucherov, V. F. Russ. Chem. ReV. 1978,
4
7, 134. (e) Saunders, W. H. Acc. Chem. Res. 1976, 9, 19. (f) Bartsch, R.
A. Acc. Chem. Res. 1975, 8, 239. (g) Saunders, W. H.; Cockerill, A. F.
Mechanisms of Elimination Reactions; Wiley: New York, 1973. (h)
Bordwell, F. G. Acc. Chem. Res. 1972, 5, 374. (i) Wolfe, S. Acc. Chem.
Res. 1972, 5, 102.
-
-
-
Amide (NH
via electron ionization upon addition of ammonia (NH
(N O) and methane (CH , 1:2), and nitrogen trifluoride (NF
respectively. Propionitrile (CH CH CN, Aldrich) was deprotonated
2
), hydroxide (OH ), and fluoride (F ) were generated
), nitrous oxide
),
3
(
2) (a) Bartsch, R. A.; Bunnett, J. F. J. Am. Chem. Soc. 1968, 90, 408.
b) Bunnett, J. F. Angew. Chem., Int. Ed. Engl. 1962, 1, 225. (c) Cram, D.
J.; Greene, F. D.; DePuy, C. H. J. Am. Chem. Soc. 1956, 78, 790.
2
4
3
(
3
2
with amide or hydroxide to afford R-cyanoethyl anion (1R) while the
â ion (1â) was produced by fluoride-induced desilylation of 3-(tri-
methylsilyl)propionitrile (i.e., the DePuy reaction ). Liquid samples
(
3) (a) March, J. AdVanced Organic Chemistry, 4th ed.; John Wiley and
Sons: New York, 1992. (b) Lowry, T. H.; Richardson, K. S. Mechanism
and Theory in Organic Chemistry, 3rd ed.; Harper and Row: New York,
14
1
987.
4) (a) Olah, G. A. Carbonium Ions; Olah, G. A., Schleyer, P. v. R.,
(
(10) We have previously prepared the conjugate base of 1-cyanotricyclo-
[4.1.0.02 ]heptane at the 3-position (a â-substituted bicyclobutyl anion
derivative), â-methoxycyclopropyl anion, and â-methoxyvinyl anion.
Unpublished results, but for some details, see: (a) Chou, P. K. Ph.D. Thesis,
University of Minnesota, 1992. (b) Baschky, M. C. Ph.D. Thesis, University
of Minnesota, 1994.
,7
Eds.; John Wiley and Sons: New York, 1976. (b) Olah, G. A.; Prakash, G.
K. S.; Sommer, J. Superacids; John Wiley and Sons: New York, 1985. (c)
Vogel, P. Carbocation Chemistry; Elsevier: New York, 1985. (d) Saunders,
M.; Jimenez-Vazquez, H. G. Chem. ReV. 1991, 91, 375.
(5) Kass, S. R.; Guo, H., Dahlke, G. D. J. Am. Soc. Mass Spectrom.
1
990, 1, 366.
(11) Graul and Squires have reported the formation of CF3CH2CH2- and
its proton affinity via collision-induced dissociation of the corresponding
(
(
6) Kroeker, R. L.; Kass, S. R. J. Am. Chem. Soc. 1990, 112, 9024.
7) Dahlke, G. D.; Kass, S. R. Int. J. Mass Spectrom. Ion Processes 1992,
-
carboxylate (CF3CH2CH2CO2 ). (a) Graul, S., T.; Squires, R. R. J. Am.
1
17, 633.
Chem. Soc. 1990, 112, 2506. (b) Graul, S., T.; Squires, R. R. J. Am. Chem.
Soc. 1990, 112, 2517.
(12) More recently, Grabowski et al. has prepared â-furanide, the
conjugate base of furan at the â-position. Submitted for publication.
(13) Ahmad, M. R.; Dahlke, G. D.; Kass, S. R. J. Am. Chem. Soc. 1996,
118, 1398.
(
(
8) Sachs, R. K.; Kass, S. R. J. Am. Chem. Soc. 1994, 116, 783.
9) (a) Graul, S. T.; Squires, R. R. J. Am. Chem. Soc. 1990, 112, 2506.
(b) DePuy, C. H.; Gronert, S.; Barlow, S. E.; Bierbaum, V. M.; Damrauer,
R. J. Am. Chem. Soc. 1989, 111, 1968. (c) Schleyer, P. v. R.; Spitznagel,
G. W.; Chandrasekhar, J. Tetrahedron Lett. 1986, 27, 4411.
S0002-7863(95)03796-6 CCC: $12.00 © 1996 American Chemical Society

â-Cyanoethyl Anion
J. Am. Chem. Soc., Vol. 118, No. 18, 1996 4463
Table 1. Computed MP2/6-31+G(d) Structures for Propionitrile,
R-Cyanoethyl Anion (1R), â-Cyanoethyl Anion (1â), and the Latter
a
Ion’s Elimination Transition Structure (1TS)
structural param
propionitrile
1R
1â
1TS
C
C
C
C
C
C
â
â
â
-H
-H
-H
1
2
3
â
1.093
1.093
1.093
1.533
1.095
1.468
1.182
109.8
110.8
110.8
112.0
110.7
178.9
180.0
60.2
1.100
1.111
1.100
1.516
1.091
1.398
1.205
110.5
115.8
110.7
117.3
114.9
177.0
-169.4
69.2
1.102
1.102
1.527
1.100
1.497
1.188
1.092
1.092
1.431
1.091
1.806
1.197
R
R
R
-C
-H
-C
4
N
CtN
H
H
H
1
2
3
-C
-C
-C
-C
â
â
â
-C
R
R
R
-C
-C
109.9
109.9
117.4
110.9
175.8
116.4
116.4
117.3
115.6
157.2
C
â
R
-C
H
4
-C
R
-C
â
N-C-C
R
H
H
H
H
1
2
3
4
-C
-C
-C
-C
â
â
â
-C
-C
-C
R
R
R
-C
-C
-C
-C
59.8
-59.8
121.3
68.6
-68.6
114.8
-60.2
120.7
-51.7
141.8
R
-C
â
a
Distances are in angstroms, and angles are in degrees.
Figure 1. Calculated MP2/6-31+G(d) structures for CH
â, and 1TS.
3 2
CH CN, 1r,
were obtained from the following sources: D
2 3
O (Isotec), CH OD
1
(
Aldrich), CH CH OD (Aldrich), and (CH COD (Aldrich); they were
3
2
3 3
)
used as supplied. Noncondensible impurities, however, were removed
the curvature of the potential energy surface and the zero-point energies
(ZPE). The latter quantities were scaled throughout by 0.9135 (HF)
and 0.9646 (MP2), as previously recommended.19 The resulting lowest
energy minima and their connecting transition states were subsequently
examined at the MP2(UMP2)/aug-cc-pVTZ//MP2(UMP2)/aug-cc-
pVDZ20 and DFT/aug-cc-pVTZ//DFT/aug-cc-pVDZ (DFT function-
als: B-VWN5, B-LYP, and Becke3-LYP)2
1-24
levels of theory (Table
2
and supporting information). Force constants were again calculated
for each optimized geometry, but the DFT ZPEs were used without
employing any empirical scaling factor.25
The density functionals employed in this study are representative
of the range of those commonly used. These functionals, B-VWN5,
B-LYP, and Becke3-LYP, differ in their respective exchange and
correlation components.2
1-24
The first two methods make use of
Becke’s nonlocal correction to Slater’s local exchange functional. They
differ from each other in that the former’s correlation functional is
entirely local in nature, whereas the latter has a nonlocal correction
added to the local functional. Becke3-LYP, on the other hand, is a
“hybrid” method with both Hartree-Fock and density functional (local
with a nonlocal correction) exchange as well as a correlation term.
Moreover, the components which make up this method were para-
metrically fitted to an empirical data set.
CDCl ): δ 2.30 (t, 2H, J ) 8.2 Hz), 0.90 (t, 2H, J ) 8.2 Hz), 0.60 (s,
3
13
(
17) For references dealing with the 6-31+G(d) basis set and Møller-
9
(
H). C NMR (75 MHz, CDCl
q).
All of the reported computations were carried out using Gaussian
3
): δ 121.3 (s), 12.7 (t), 11.9 (t), -2.1
Plesset perturbation theory, see: (a) Hehre, W. J.; Ditchfield, R.; Pople, J.
A. J. Chem. Phys. 1972, 56, 2257. (b) Hariharan, P. C.; Pople, J. A. Theor.
Chim. Acta 1973, 28, 213. (c) Gordon, M. S. Chem. Phys. Lett. 1980, 76,
63. (d) Frisch, M. J.; Pople, J. A.; Binkley, J. S. J. Chem. Phys. 1984, 80,
265. (e) Clark, T.; Chandrasekhar, J.; Spitznagel, G. W.; Schleyer, P. v.
R. J. Comput. Chem. 1983, 4, 294. (f) Møller, C.; Plesset, M. S. Phys. ReV.
1934, 46, 618. (g) Frisch, M. J.; Head-Gordon, M.; Pople, J. A. Chem.
Phys. Lett. 1990, 166, 281.
1
3
1
6
9
2/DFT on a UNIX-based workstation or a Cray supercomputer at
the Minnesota Supercomputer Institute. Systematic isomeric and
-
•
conformational analyses were carried out on C
and transition structures converting one species into another at the MP2-
UMP2)/6-31+G(d)//RHF(UHF)/6-31+G(d) and MP2(UMP2)/6-31+G-
3 5 3 4 3 4
H N, C H N , C H N ,
(18) Alternative structures to 1â were considered, and a few of them
(
were examined computationally. In particular, -CH CH NC, CH CH CdN ,
-
(
d)//MP2(UMP2)/6-31+G(d) levels of theory (Table 1, Figure 1, and
2
2
2
2
1
7,18
supporting information).
Force constants were also determined at
-
CH2CH2 Ch dN, and CH2CH2NdC were found to be 16.4, 22.5, 30.2, and
these levels of theory for each stationary point in order to determine
9
4.5 kcal/mol less stable than 1â at the MP2/6-31+G(d)//HF/6-31+G(d)
level. Isomerization to one of these anions, therefore, is unlikely. For further
information on these ions, structures, and absolute energies, see the
supporting information.
(
14) (a) DePuy, C. H.; Bierbaum, V. M.; Flippin, L. A.; Grabowski, J.
J.; King, G. K.; Schmitt, R. J.; Sullivan, S. A. J. Am. Chem. Soc. 1980,
02, 5012. (b) DePuy, C. H.; Bierbaum, V. M.; Damrauer, R.; Soderquist,
1
(19) Pople, J. A.; Scott, A. P.; Wong, M. W.; Radom, L. Isr. J. Chem.
1993, 33, 345.
(20) (a) Dunning, T. H., Jr. J. Chem. Phys. 1989, 90, 1007. (b) Kendall,
R. A.; Dunning, T. H., Jr.; Harrison, R. J. J. Chem. Phys. 1992, 96, 6796.
(21) Becke, A. D. Phys. ReV. A 1988, 38, 3098.
J. A. J. Am. Chem. Soc. 1985, 107, 3385. (c) O’Hair, R. A. J.; Gronert, S.;
DePuy, C. H.; Bowie, J. H. J. Am. Chem. Soc. 1989, 111, 3105.
(
15) Fleming, I.; Goldhill, J. J. Chem. Soc., Perkin Trans. 1 1980, 1493.
(16) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Gill, P. M. W.;
Johnson, B. G.; Wong, M. W.; Foresman, J. R.; Robb, M. A.; Head-Gordon,
M.; Replogle, E. S.; Gomperts, R.; Andres, J. L.; Raghavachari, K.; Binkley,
J. S.; Gonzales, C. A.; Martin, R. C.; Fox, D. J.; Defrees, D. J.; Baker, J.;
Stewart, J. J. P.; Pople, J. A. GAUSSIAN-92/DFT; Gaussian Inc.: Pittsburgh,
PA, 1993.
(22) Vosko, S. H.; Wilk, L.; Nusair, M. Can. J. Phys. 1980, 58, 1200.
(23) Lee, C.; Yang, W.; Parr, R. G. Phys. ReV. B 1988, 37, 785.
(24) Becke, A. D. J. Chem. Phys. 1993, 98, 5648.
(25) Johnson, B. G.; Gill, P. M. W.; Pople, J. A. J. Chem. Phys. 1993,
98, 5612.

4
464 J. Am. Chem. Soc., Vol. 118, No. 18, 1996
Merrill et al.
a
Table 2. Calculated Energies for Propionitrile, R- and â-Cyanoethyl Anion (1R and 1â), R- and â-Cyanoethyl Radical (2R and 2â), the E1cb
Transition Structure (1TS), and Related Species
MP2/aug-
species
B-VWN5b
B-LYPb
Becke3-LYPb
MP2c
MP2/6-31+G(d)d
cc-pVTZd
G2+ (MP2)
G2+
CH3CH2CN
-172.919 86 -171.989 89 -172.061 93 -171.434 74
-171.439 68
-170.841 40
-170.805 30
-170.792 81
-171.650 22 -171.682 23 -171.745 19
-171.053 80 -171.085 15 -171.145 42
-171.019 78 -171.050 76 -171.111 87
-170.990 21 -171.028 89 -171.095 13
(0.890; 0.762) (0.888; 0.761) (0.888; 0.761)
-170.988 82 -171.019 67 -171.080 89
(0.763; 0.750) (0.762; 0.750) (0.762; 0.750)
-171.009 27 -171.039 70 -171.102 84
-
CH3CH CN (1R) -172.321 99 -171.402 74 -171.469 30 -170.837 99
-
CH2CH2CN (1â) -172.286 04 -171.366 68 -171.431 98 -170.800 58
e
CH3 C4 HCN (2R)
-172.264 14 -171.359 11 -171.422 75 -170.794 17
(
0.759; 0.750) (0.758; 0.750) (0.766; 0.750) (0.913; 0.766) (0.888; 0.761)
-170.793 05
0.754; 0.750) (0.753; 0.750) (0.754; 0.750) (0.762; 0.750) (0.762; 0.750)
•
e
CH2CH2CN (2â) -172.242 05 -171.336 29 -171.401 62 -170.788 16
(
f
E1cb TS (1TS)
-172.283 90 -171.363 27 -171.425 47 -170.792 10
-170.797 89
(469i)
(391i)
(393i)
(450i)
(511i)
(467i)
(465i)
(465i)
C2H4
CN
CH2dCHCN
-79.014 40
-78.526 50
-78.572 88
-78.240 54
-78.241 22
-92.599 04
-170.259 68
-78.354 47
-78.374 80
-78.414 52
-
-93.311 91
-92.871 78
-92.890 47
-92.595 49
-92.688 35
-92.703 27
-92.728 56
-171.672 92 -170.793 19 -170.848 15 -170.252 94
-170.444 48 -170.476 44 -170.533 48
a
b
In hartrees. DFT/aug-cc-pVTZ//DFT/aug-cc-pVDZ (e.g., BVWN5/aug-cc-pVTZ//BVWN5/aug-cc-pVTZ). Unscaled zero-point energies are
c
d
included. MP2/6-31+G(d)//HF/6-31+G(d) + (0.9135)ZPE. Energies and structures were obtained with the indicated basis set. Scaled (0.9646)
ZPEs are included. Spin contamination (〈S 〉) in the wavefunction before and after projection is given in parentheses. The imaginary frequencies
have been scaled (0.8929 (HF) and 0.9427 (MP2)) and are in cm
e
2
f
-
1
.
Several minima and transition structures pertinent to the experimental
or amide (eq 2). In addition to the M - 1 ion (m/z 54), a small
results were investigated at the G2+ and G2(MP2)+ levels of
2
6-29
-
theory.
These calculations effectively correspond to QCISD(T)/
CH CH CN + B f CH Ch HCN
(2)
3
2
3
6
-311+G(3df,2p)//MP2/6-31+G(d,p) + ZPE energies and represent the
highest levels of theory employed in the present study. Calculations
1
R
of this type generally reproduce experimental energetics to within 2-3
-
-
-
2
6,27
B ) OH or NH
2
kcal/mol.
-
amount of CN and a M + (M - 1) cluster (m/z 109) are also
formed. The R-anion (1R) was readily distinguished from its
less stable â-isomer (1â) on the basis of the following
observations: (1) Nitrous oxide does not react with 1R but
Results and Discussion
-(Trimethylsilyl)propionitrile (1), synthesized from aceto-
3
nitrile and (chloromethyl)trimethylsilane, reacts with fluoride
ion at room temperature in our flowing afterglow device to
afford cyanide (m/z 26) as the predominate product ion. Smaller
amounts of the M - TMS (m/z 54), M - 1 (m/z 126), and M
-
-
affords characteristic products, HN2O (m/z 45) and CH2dC CN
m/z 52), with 1â (eqs 3 and 4). The product ions also are in
31
(
-
30
+
CN (m/z 153) ions are also produced (eq 1). The relative
CH Ch HCN + N O f no reaction
(3)
3
2
abundance of the m/z 54 ion (1â) is only about 5%, but the
absolute intensity is sufficient to characterize this species and
assign its structure (signal-to-noise ratios of g100-1000:1 were
obtained in a single scan).
-
-
CH CH CN + N O
8 HN O
(4a)
2
2
2
2
∼
∼
60%
40%8
CH d Ch CN
(4b)
2
-
-
TMSCH CH CN + F
86%8 CN
(1a)
(1b)
(1c)
2
1
2
keeping with the structure of 1â and the previously reported
behavior of N2O. (2) Sulfur dioxide reacts with 1R to give
HSO2 (m/z 65) while 1â affords SO2 (m/z 64) and HSO2 in
∼
32
-
-
-
-
5%8
CH CH CN
approximately a 3:1 ratio, respectively (eqs 5 and 6). These
2
2
1â
-
CH Ch HCN + SO f HSO
2
(5)
3
2
6%8
TMSCH Ch HCN
2
-
-
CH CH CN + SO 75%8 SO2
(6a)
2
2
2
∼
∼
-
TMSCH CH CN‚CN (1d)
-
2
2
2
%8
25%8 HSO2
(6b)
R-Cyanoethyl anion (1R), the M - 1 ion of propionitrile,
results indicate that the electron affinity of the R-cyanoethyl
was produced by deprotonation of propionitrile with hydroxide
33
radical (2R) is greater than 25.5 kcal/mol (1.11 eV), and the
(
26) For a description of the G2 procedure, see: Curtiss, L. A.;
Raghavachari, K.; Trucks, G. W.; Pople, J. A. J. Chem. Phys. 1991, 94,
221.
27) For a description of the G2(MP2) method, see: Curtiss, L. A.;
Raghavachari, K.; Pople, J. A. J. Chem. Phys. 1993, 98, 1293.
28) For a discussion of the G2+ approach, see: Gronert, S. J. Am. Chem.
Soc. 1993, 115, 10258.
29) In contrast to the literature procedure which calls for MP2(full)/6-
â-cyanoethyl radical (2â) has a smaller electron affinity than
this value. In accord with this conclusion, the literature value
7
(
(31) It is worth noting that, given the large background signal at m/z 26,
-
the formation of CN would go undetected in all of the transformations of
(
1â that were studied.
(32) (a) Kass, S. R.; Filley, J.; Van Doren, J. M.; DePuy, C. H. J. Am.
Chem. Soc. 1986, 108, 2849. (b) Bierbaum, V. M.; DePuy, C. H.; Schmitt,
R. H. J. Am. Chem. Soc. 1977, 99, 5800.
(
3
1+G(d,p) geometries for anions, MP2(full)/6-31G(d,p) geometries for
neutrals and scaled HF frequencies, we used MP2(FC)/6-31+G(d,p)
optimizations for all structures and scaled (0.9646) MP2 frequencies. Our
G2+ energies were then obtained as follows: E(G2+) ) E(MP4(SDTQ)/
(33) Celotta, R. J.; Bennett, R. A.; Hall, J. L. J. Chem. Phys. 1974, 60,
1740.
(34) All thermodynamic data, unless otherwise noted, comes from Lias
et al. (Lias, S. G.; Bartmess, J. E.; Liebman, J. F.; Holmes, J. L.; Levin, R.
D.; Mallard, W. G. J. Phys. Chem. Ref. Data 1988, 17, Suppl. 1) or the
slightly updated form available on a personal computer: NIST NegatiVe
Ion Energetics Database (Version 3.00, 1993); NIST Standard Reference
Database 19B.
6
3
-311+G (2df,p)) + (E(QCISD(T)/6-311+G(d,p)) - E(MP4(SDTQ)/6-
11+G(d,p))) + (E(MP2/6-31+G(3df,2p)) - E(MP2/6-311+G(2df,p))) +
E(HLC) + scaled MP2 ZPE.
(30) Additional ions at m/z 46, 80, 89, 91, 93, 116, 130, and 146 were
also observed.

â-Cyanoethyl Anion
J. Am. Chem. Soc., Vol. 118, No. 18, 1996 4465
Table 3. Summary of Proton Transfer and Hydrogen-Deuterium
a
Exchange Data for 1R and 1â
∆
H°acid
a
ref acid
CH COD
CH OD
OD
(kcal/mol)
1R
1â
+
(
3 3
)
374.6b
377.4b
383.5
D transfer
1 H/D exchange
1 H/D exchange D transfer + 2 H/D
CH
CH
3
2
+
3
exchangesc
H transfer (small)
+
H
D
2
O
O
390.7
392.0
396.3
d
2
1 H/D exchange 2 H/D exchanges
+
(
CH
3
)
2
NH
no H transfer
a
The acidity data comes from refs 34 and 36. b Value for protic
acid. c Incorporation of the second deuterium only occurs to a small
extent. The possible presence of DO (m/z 18) was obscured by the
large F (m/z 19) signal.
d
-
-
for the electron affinity of 2R is 28.6 ( 2.3 kcal/mol (1.24 (
3
2
0
.10 eV). (3) Deuterium oxide reacts with 1R by exchanging
one hydrogen for a deuterium while 1â undergoes two H/D
exchanges (eqs 7 and 8). On the basis of these results it is
D O
2
CH Ch HCN
8 CH Ch DCN
(7)
3
3
D O
2
-
-
CH CH CN
8 [CH DCH CN‚OD ] ff CH D Ch DCN
2
2
2
2
2
(8)
apparent that 1R and 1â have different structures, and with the
aid of ab initio calculations (vide infra) we assign the â-cy-
anoethyl anion structure to 1â.1
8,35
The reactivities of the R- and â-cyanoethyl anions (1R and
â) with a series of acids of known strength were examined in
1
order to establish their proton affinities. By observing the
occurrence or nonoccurrence of proton transfer it was possible
to bracket their proton affinities. The results from these
experiments are summarized in Table 3 and indicate that PA-
3
4,36
(1R) ) 376 ( 2 kcal/mol and PA(1â) ) 391 ( 5 kcal/mol.
The former value is in excellent agreement with the literature
proton affinity of 375.1 ( 2.1 kcal/mol and our computed ab
initio acidity (376.4 kcal/mol (G2+), Table 4). The latter
quantity is also in reasonable agreement with our calculated
value (397.4 kcal/mol (G2+)) although the computations suggest
that the â-acidity is toward the higher end of our range.
The large difference in the acidity of the R and â sites in
propionitrile, 15 ( 5 kcal/mol, accounts for the different
hydrogen-deuterium (H/D) exchange behavior of 1R and 1â.
The former anion reacts with a deuterated acid such as deuterium
oxide to replace the unique hydrogen at the charged site with a
deuterium via a deuteron transfer-proton abstraction sequence
(
eq 7).37 The â-cyanoethyl anion also undergoes an initial
deuteron transfer, to the â-position in this case (eq 8), but the
subsequent proton abstraction now occurs at the R site. This
acid-catalyzed isomerization results in the formation of a
-
monodeuterated R-cyanoethyl anion, CH2DCH CN, which still
has an exchangeable hydrogen at the R-position. Consequently,
two H/D exchanges can take place for the â-anion.
Ab initio and density functional theory (DFT) calculations
have been carried out on 1R, 1â, and a number of related
1
8
species. Optimizations were performed at the Hartree-Fock
35) A cluster ion between CN- and ethylene (C2H4‚CN or C2H4‚NC )
-
-
(
is more stable than 1â, but can be excluded on the basis of the observed
reactivity of the M - TMS ion, the measured thermodynamic properties,
and the reported computational results.
(
36) Barlow, S. E.; Dang, T. T.; Bierbaum, V. M. J. Am. Chem. Soc.
990, 112, 6832.
37) For excellent descriptions of the gas-phase exchange process, see:
a) Grabowski, J. J.; DePuy, C. H.; Van Doren, J. V.; Bierbaum, V. M. J.
1
(
(
Am. Chem. Soc. 1985, 107, 7384. (b) Nibbering, N. M. M. AdV. Phys. Org.
Chem. 1988, 24, 1 and references therein.

4
466 J. Am. Chem. Soc., Vol. 118, No. 18, 1996
Merrill et al.
a
(
HF) level using the 6-31+G(d) basis set and at the MP2 (frozen
Table 5. Calculated Electron Affinities for R- and â-Cyanoethyl
Radicals (2R and 2â) Using Several Approachesb
core) level with both the 6-31+G(d) and aug-cc-pVDZ basis
sets.
and Becke3-LYP) were also used with the latter basis set.
1
7,20
Three diverse DFT functionals (B-VWN5, B-LYP,
compd level of theory eq 9 eq 10 eq 11 eq 12
av
2
1-24
2
R
B-VWN5
B-LYP
1.57
1.19
1.27
1.57
1.19
1.27
1.47
1.44
1.34
1.24 ( 0.09
1.20
0.83
The resulting geometries are generally in good agreement with
each other (e0.02 Å and e2°) except for the CtN bond length,
which elongates by approximately 0.05 Å in going from the
HF to the MP2 level. The effect of the basis set, MP2/6-31+G-
Becke3-LYP
MP2c
1.19 1.74
1.53 1.53 1.26
1.37 1.38 1.27
G2+(MP2)
G2+
expt
B-VWN5
B-LYP
(d) vs MP2/aug-cc-pVDZ, also is small (<0.015 Å and <2°)
except for the CR-CN distance in the transition structure for
cyanide expulsion (1TS), which increases by 0.028 Å with the
larger basis set. Therefore, only the MP2/6-31+G(d) geometries
are given in Figure 1 and Table 1 and are discussed below (all
of the structural data can be found in the supporting informa-
tion).
2
â
1.20
0.83
Becke3-LYP
0.83
0.34 0.89
0.83
0.39 0.54
MP2c
G2+(MP2)
G2+
0.85 0.85 0.58 0.56 0.71
0.84 0.86 0.74 0.71 0.79
a
In electronvolts. b Scaled and unscaled ZPEs are included in the
Deprotonation of propionitrile at the R-position results in a
.070 Å shortening of the CR-CN bond and a 0.023 Å
ab initio and DFT results, respectively. Experimental values for the
0
•
•
3 2 3
, CH CN, and CH CH(CN) , see refs 34 and
c
•
electron affinities of CH
1, were used in eqs 10-12, respectively. MP2/6-31+G(d)//HF/6-
31+G(d) energies.
lengthening of the CtN distance as expected for a delocalized
carbanion. The CR-Câ distance also contracts somewhat (0.017
Å) in accord with a formal hybridization change from sp to
4
3
for electron correlation.39 The latter requirement is a result of
the fact that the number of electron pairs differs in an anion
and its corresponding radical. As a result, high levels of theory
are needed to calculate electron affinities directly. A compu-
tationally less demanding approach which has met with success
is to make use of isogyric reactions, transformations in which
the number of paired and unpaired electrons are preserved.40
Both methods were utilized in calculating the electron affinities
of 2R and 2â as illustrated for the former compound in eqs
2
sp at the R-carbon (CR).
-
CH s Ch HsCtN T CH sCHdCdN
(1R)
3
3
The â-anion (1â) flattens out slightly at Câ (2.2°)38 and has a
longer CR-CN bond than its conjugate acid (0.029 Å) in accord
with a species stabilized by negative hyperconjugation. These
trends are further accentuated in the elimination transition
structure; the CR-CN distance is 0.309 Å longer, the â-carbon
4
1
9
-11 (Table 5).
The latter quantity was also calculated
-
-
CH sCH sCN T CH dCH CN
(1â)
2
2
2
2
-
CH Ch HCN ∆
8 CH C˙ HCN + e
(9)
3
3
H°rxn ) EA(2R)
is 18.9° flatter, and the CR-Câ bond length is 0.096 Å shorter
than in 1â due to the formation of carbon-carbon double bond
character. It is also interesting that the N-C-C angle deviates
from linearity by 22.9° and is 15.8° smaller than in 1â.
Presumably, this orientation reduces the electrostatic repulsion
between the incipient nitrile (CN ) and the negatively charged
â-carbon.
1R
2R
•
CH Ch HCN + CH
8
3
3
H°rxn ) EA(2R) - EA(CH3•)
∆
1R
CH C4 HCN + CH3- (10)
-
3
2
R
The proton affinities of 1R and 1â were calculated at several
levels of theory, and all of them, with the exception of the
B-LYP result, are in good accord with the experimental values
and each other (Table 4). These results, in combination with
•
CH Ch HCN + CH CN
8
3
2
•
∆H°rxn ) EA(2R) - EA( CH CN)
2
1
R
-
CH C4 HCN + CH CN (11)
3
2
9
b
the experimental acidity of ethane (∆H°acid ) 420 kcal/mol),
2R
reveal that the CtN group decreases the proton affinity of 1R
and 1â by 45 (experiment, 44 (G2+)) and 29 (experiment, 23
-
CH CH CN + CH C4 HCN ∆H°rxn ) EA(2â) - EA(2R)8
2
2
3
(
G2+)) kcal/mol, respectively.35 The bigger interaction at the
1â
2
R
R-position is consistent with 1R being resonance stabilized,
while the latter value indicates that â-substituent effects can be
considerable. It is worth adding that cyclization of 1â to a three-
or four-membered-ring isomer is energetically quite unfavorable
and this is not responsible for the large â-effect.18 A preliminary
analysis of the observed stabilization in terms of the field effect,
resonance, polarization, and the inductive effect indicates that
the first two contributors are the dominant factors.
•
CH CH CN + CH Ch HCN (12)
2
2
3
2
â
1R
making use of the known electron affinity of 2R (eq 12).34 For
both compounds the G2+ energies span a narrow range (e0.15
eV) and the average values, 1.34 and 0.79 eV, respectively,
were employed. Our computed results are in good agreement
with the measured electron affinity of the R-radical (1.24 (
Another commonly used measure of the stability of an anion
is its electron binding energy. This quantity, more typically
referred to as the electron affinity (EA), is the enthalpic
difference between an anion and its corresponding radical.
Unfortunately, this thermodynamic property is more difficult
to calculate accurately than a proton affinity because of spin
contamination in unrestricted wavefunctions of open-shell
systems (e.g., UHF or UMP2) and a greater need to account
(
39) Baker, J.; Nobes, R. H.; Radom, L. J. Comput. Chem. 1986, 7, 349.
(40) (a) Pople, J. A.; Schleyer, P. v. R.; Kaneti, J.; Spitznagel, G. W.
Chem. Phys. Lett. 1988, 145, 359. (b) Hehre, W. J.; Radom, L.; Schleyer,
P. v. R.; Pople, J. A. Ab Initio Molecular Orbital Theory; John Wiley and
Sons: New York, 1986.
•
•
(
41) The G2+ and G2+(MP2) electron affinities of CH3 and CH2CN
•
were calculated directly and are in good accord with experiment: EA(CH3 )
) 0.065 (G2+), 0.082 (G2+(MP2)), and 0.080 ( 0.030 eV (experiment);
•
EA( CH2CN) ) 1.64 (G2+), 1.81 (G2+(MP2)), and 1.543 ( 0.014 eV
(
experiment). For the experimental data, see: (a) Ellison, G. B.; Engelking,
•
(38) The pyramidalization angle at Câ was defined as the acute angle
P. C.; Lineberger, W. C. J. Am. Chem. Soc. 1978, 100, 2556 (CH3 ). (b)
formed by the line going through Câ and CR and the line bisecting the
H-Câ-H angle.
Moran, S.; Ellis, H. B., Jr.; DeFrees, D. J.; McLean, A. D.; Ellison, G. B.
J. Am. Chem. Soc. 1987, 109, 5996 ( CH2CN).
•

â-Cyanoethyl Anion
.09 eV), and they account for why 1â transfers an electron to
sulfur dioxide (EA ) 1.11 eV) while 1R does not; the former
reaction is exothermic whereas the latter one is endothermic. It
is also worth noting that B-LYP and Becke3-LYP do a very
good job on the electron affinities but at a fraction of the
computational time required for the G2+ results.
J. Am. Chem. Soc., Vol. 118, No. 18, 1996 4467
0
discrepancy in the proton affinity of 1â but also because cyanide
33
•
is not well described at the G2+ level (e.g., EA(CN ) ) 4.03
46
eV (G2+) and 3.74 ( 0.13 eV (experiment)). The elimination
q
barrier (∆H ) is computed to be 5.7 kcal/mol at our best level
of theory, and this value accounts for the difficulty in generating
1â; elimination is facile but not so facile as to preclude the
formation of 1â. In this regard, it is worth noting that anions
with electron binding energies of approximately 6 kcal/mol or
more can be readily generated in helium at 298 K in the reaction
region of a flowing afterglow device whereas species with
The hydride affinity (HA) of acrylonitrile at the R- and
â-positions (eq 13) provides another measure of the stability of
1
R and 1â. These energies were calculated at several levels of
-
47
CH dCHCN + H
8 CH Ch HCN
(13a)
smaller electron affinities cannot.
2
3
-∆H°rxn ) HA(R)
A few specific comments about the density functional
calculations carried out in this work are warranted. First, the
DFT structures are in good accord with one another and their
ab initio MP2/6-31+G(d) counterparts. Not surprisingly, the
biggest differences are found in the elimination transition
structures (1TS). Both B-VWN5 and B-LYP have shorter CR-
CN and longer CR-Câ bond lengths (0.03-0.06 and 0.02-
-
-∆H°rxn ) HA(â)8
CH CH CN (13b)
2 2
theory (Table 4), although in each case the exact energy of H-,
0.52775 hartrees, was used. The reason for this is that the
basis sets which were employed, as is usually the case, are more
flexible for heavy atoms than for hydrogen; consequently,
hydride is badly described computationally. Smith, Pople,
Curtiss, and Radom circumvented this problem by defining the
-
0.03 Å, respectively) than the Becke3-LYP and MP2 structures,
indicating that the former methods lead to an earlier and looser
48
transition state.
This is consistent with the fact that the
-
exact energy of H as its G2 (and thus G2+) value, and we
imaginary frequencies corresponding to the reaction coordinate
have adopted this proposal.4
2,43
At the G2+ level, HA(R) )
-1
for the B-VWN5 and B-LYP eliminations are about 70 cm
5
2.8 kcal/mol, which is in reasonable accord with the experi-
smaller than for the MP2 or Becke3-LYP structures. Second,
Becke3-LYP gives the best overall energetic data. In particular,
34
mental value of 57.0 ( 2.2 kcal/mol. The â-hydride affinity,
HA(â) ) 31.8 kcal/mol, is in poor agreement with experiment
41 ( 5 kcal/mol), but this is largely a reflection of the 6.4
the average unsigned error relative to the G2+ energies is only
(
49
1
.7 kcal/mol, and all of the values are within 2.3 kcal/mol
44
kcal/mol discrepancy in the acidity. In any case, these results
suggest that hydride transfer from 1R to nitrous oxide is slightly
endothermic (+3 kcal/mol) whereas it is exothermic with 1â
except for the proton affinity of 1R. This quantity is too small
by 4.5 and 3.2 ( 2.1 kcal/mol relative to the G2+ and the
experimental results, respectively. Third, the B-VWN5 results
are the poorest of the three with an average error of 4.3 kcal/
mol, but this method does appear to give the most reliable proton
4
5
-
(
-14 kcal/mol); this accounts for the formation of HN2O in
the latter case.
â-Cyanoethyl anion is a marginally stable E1cb intermediate
which can readily expel cyanide to form ethylene (eq 14a). This
2
affinities. Fourth, the 〈S 〉 values for the DFT wavefunctions
are appreciably smaller (less spin contamination) than their
conventional ab initio counterparts, and the electron affinities
are accurately reproduced by both functionals which make use
of a nonlocal correction to the correlation functional. Finally,
-
-
CH CH CN
8 CH dCH + CN (14a)
2
2
2
2
∆
H°rxn ) ∆H°elim(â)
q
-
all of the DFT methods predict elimination barriers (∆H ) which
CH dCH + CN
8 CH Ch HCN (14b)
2
2
3
∆
H°rxn ) -∆H°elim(R)
are smaller than those at the G2+ level, although the Becke3-
LYP value is only 1.6 kcal/mol lower than the G2+ result. The
consistent underestimation of barrier heights by density func-
tional calculations appears to be a general shortcoming and has
-
CH CH CN
8 CH Ch HCN (14c)
2
2
3
∆
H°rxn ) ∆H°elim(â) - ∆H°elim(R)
5
0,51
been noted previously.
transformation is exothermic, -8 ( 5 kcal/mol, and is well
reproduced at every level of theory that was examined (Table
Conclusions
4
) by comparing the relative energies of 1R and 1â (eq 14c)
The fluoride-induced desilylation of 3-(trimethylsilyl)propi-
onitrile leads predominantly to the formation of cyanide. A
small but significant amount (5%) of â-cyanoethyl anion (1â)
and using the experimentally derived value for ∆Helim(R) (-8.6
(
1
3
4,45
3 kcal/mol).
A direct calculation of the elimination (eq
4a) provides poor results not only because of the 6.4 kcal/mol
(
42) Smith, B. J.; Pople, J. A.; Curtiss, L. A.; Radom, L. Aust. J. Chem.
992, 45, 285.
43) The B-LYP and Becke3-LYP results are less sensitive to this
problem. For example, HA(R or â) increases by 0.6 (B-LYP) and 5.0
Becke3-LYP) kcal/mol when the exact energy for hydride is used.
44) If one uses the calculated â-acidity of ethyl cyanide and experimental
heats of formation for all of the other quantities, then the following results
(46) The electron affinity is 4.18 and 3.97 eV at the G2+(MP2) and G2
-
1
levels, respectively. Additional calculations reveal that CN is too stable
(
at the G2 (G2+) level. This problem has recently been addressed; see:
Curtiss, L. A.; Raghavachari, K.; Pople, J. A. J. Chem. Phys. 1995, 103,
4192.
(47) Dahlke, G. D.; Kass, S. R. J. Am. Chem. Soc. 1991, 113, 5566.
(48) The further developed transition structures are also closer to planarity
(i.e., they are flatter).
(
(
(
(
in kcal/mol) for HA(â) are obtained: 34.7 (G2+), 35.8 (G2+(MP2)), 36.5
MP2/aug-cc-pVTZ), 34.0 (MP2/6-31+G(d)), 34.2 (MP2/6-31+G(d)//HF/
(49) In calculating the average error, the DFT hydride affinities with
-
6
-31+G(d)), 36.8 (Becke3-LYP), 41.0 (B-LYP), and 34.2 (B-VWN5).
the calculated energies for H were used (i.e., the parenthetical values in
(
45) The following heats of formation (∆Hf°298), in kcal/mol, were
Table 4). The direct calculation of ∆H°elim(1â) (eq 14a and the values not
in parentheses in Table 4) was also used in this comparison. In any case,
it is clear that the agreement between B3-LYP, G2+, and the experimental
data is good.
-
used: 19.6 (N2O), 44.0 (CH2dCHCN), 21.7 (1R), 38 (1â), and 0 (HN2O ).
All of these energies, with the exception of ∆Hf°298(HN2O ), are available
-
or can be derived using the data given in ref 34 and our experimental
measurement for the proton affinity of 1â. The missing quantity was
obtained by averaging the temperature-corrected (298 K) G2+ results for
(50) Merrill, G. N.; Kass, S. R. Unpublished results.
(51) (a) Gronert, S.; Merrill, G. N.; Kass, S. R. J. Org. Chem. 1995, 60,
488. (b) Latajka, Z.; Bouteiller, Y.; Scheiner, S. Chem. Phys. Lett. 1995,
234, 159. (c) Johnson, B. G.; Gonzales, C. A.; Gill, P. M. W.; Pople, J. A.
Chem. Phys. Lett. 1994, 221, 100. (d) Fan, L.; Ziegler, T. J. Am. Chem.
Soc. 1992, 114, 10890.
-
-
the following two transformations: (1) HN2O f H + N2O and (2) 1R
-
+
N2O f HN2O + CH2dCHCN. Note, G2+(N2O) ) -184.43698,
-
G2+(HN2O ) ) -185.04717, and the temperature corrections are as
-
follows: 2.68 (N2O), 2.54 (HN2O ), 2.30 (1R), 1.84 (CH2dCHCN).

4
468 J. Am. Chem. Soc., Vol. 118, No. 18, 1996
Merrill et al.
also is produced, and this latter ion is a stable E1cb intermediate.
Some of the reactions and thermodynamic properties of 1â were
explored experimentally and computationally, and an extremely
large (29 ( 6 kcal/mol) â-substituent effect was noted.
Preliminary analysis of this system suggests that the stabilization
is principally due to resonance and field effects. Given these
results, it should be possible to generate additional â-substituted
administered by the American Chemical Society, and the
Minnesota Supercomputer Institute is gratefully acknowledged.
Supporting Information Available: Calculated structures
(xyz coordinates) and energies for all of the computed species
in this work (23 pages). This material is contained in many
libraries on microfiche, immediately follows this article in the
microfilm version of the journal, can be ordered from the ACS,
and can be downloaded from the Internet; see any current
masthead page for ordering information and Internet access
instructions.
-
alkyl anions ( CH2CH2X) in the gas phase, and probably even
-
more remotely substituted carbanions (e.g., CH2CH2CH2X).
Work along these lines is currently in progress and will be
reported in due course.
Acknowledgment. Support from the National Science
Foundation, the donors of the Petroleum Research Fund,
JA953796O