Resonance in Formamide and Its Chalcogen Replacement

Analogues: A Natural Population Analysis/Natural Resonance

Theory Viewpoint
Eric D. Glendening* and John A. Hrabal II
Contribution from the Department of Chemistry, Indiana State UniVersity,
Terre Haute, Indiana 47809
ReceiVed January 9, 1997. ReVised Manuscript ReceiVed October 14, 1997
X
Abstract: The influence of resonance on the structure and rotation barrier of formamide and its S, Se, and Te
replacements analogues is examined using the natural bond orbital methods. Calculations are performed at the RHF,
B3LYP, and MP2 levels of theory with 6-31+G* basis sets and effective core potentials. At the MP2 level, the
rotation barriers increase with the increasing size of the chalcogen, from 17.2 kcal mol
-1
for formamide to 21.0 kcal
mol
-1
for telluroformamide. Natural population analysis and natural resonance theory (NRT) reveal shifts in the
charge density that are consistent with the strong resonance stabilization of the equilibrium, planar geometries. NRT
provides a simple, quantitative description of the amides as a resonance hybrid consisting primarily of two contributing
structures, the parent Lewis structure and a secondary dipolar form. Amide resonance effects strengthen from
formamide to telluroformamide as the weight of the dipolar form increases. Polarizability appears to contribute
importantly, allowing the chalcogens to accommodate more charge density than anticipated on the basis of
electronegativity.
I. Introduction
Amide structure and reactivity is conventionally interpreted
within the framework of resonance theory.
1,2
Formamide, the
simplest amide, is poorly represented by the single Lewis
structure I (X ) O). Its planar geometry, large rotation barrier
(18-19 kcal mol
-1
), and red-shifted carbonyl stretching fre-
quency can be readily understood in terms of the strong
resonance mixing of the dipolar form II. In contrast to typical
pyramidal amine geometries, the planar amino group is stabilized
by the resonance interaction of the p-type N lone pair with the
carbonyl system. The resulting CN double-bond character
gives rise to the large rotation barrier, while the loss of CO
double-bond character leads to a low CO stretching frequency.
Amide resonance has been extensively discussed for decades
and is now firmly embedded in standard textbooks of organic
chemistry.
3
The traditional amide resonance model has, however, been
challenged in recent years, primarily by Wiberg, Laidig, and
co-workers.
4-11
Their criticisms largely stem from computa-
tional studies of formamide in its planar (1a) and twisted (2a)
geometries. The twisted geometry should be only weakly
stabilized by resonance interactions as the N lone pair lies in
the frame, antiperiplanar to the
CO
bond. Resonance theory
suggests that the charge density of 1a should reflect, through
the contribution of II, a transfer of (negative) charge from N to
O relative to that of 2a. An analysis of calculated charge
densities using Baders atoms in molecules (AIM) method
12
reveals, however, the transfer of charge from C to N with
minimal change in the charge at O. Amide resonance further
predicts the elongation of the CN bond and contraction of the
CO bond as the planar geometry rotates into the twisted form,
1a f2a. While these bond length changes are indeed observed
in the calculations, the CN elongation (0.08 ) is considerably
more pronounced than the CO contraction (0.01 ). Thus, it
appears that the CN bond gains significant double-bond
character, whereas the CO remains essentially a double bond.
The resonance model might lead one to expect more similar
changes in bond length on rotation.
Thioformamide (1b) also presents an apparent difficulty for
the traditional resonance model.
9,10
A weaker contribution from
the dipolar structure II (X )S) and, hence, lower rotation barrier
are anticipated on the basis of the relative electronegativities
of O and S. However, ab initio calculations suggest that
* Corresponding author. E-mail: ericg@chem.indstate.edu.
X
Abstract published in AdVance ACS Abstracts, December 15, 1997.
(1) Pauling, L. The Nature of the Chemical Bond, 3rd ed.; Cornell
University: Ithaca, NY, 1960.
(2) Wheland, G. W. Resonance in Organic Chemistry; Wiley: New York,
1955.
(3) See, e.g.: (a) Carey, F. A. Organic Chemistry, 3rd ed.; McGraw-
Hill: New York, 1996; p 809. (b) McMurry, J. Organic Chemistry, 4th
ed.; Brooks/Cole: New York, 1996; p 942. (c) Solomons, T. W. G. Organic
Chemistry, 6th ed.; Wiley: New York, 1996; p 906.
(4) Wiberg, K. B.; Laidig, K. E. J. Am. Chem. Soc. 1987, 109, 5935.
(5) Wiberg, K. B.; Breneman, C. M. J. Am. Chem. Soc. 1992, 114, 831.
(6) Wiberg, K. B.; Hadad, C. M.; Rablen, P. R.; Cioslowski, J. J. Am.
Chem. Soc. 1992, 114, 8644.
(7) Wiberg, K. B.; Rablen, P. R. J. Am. Chem. Soc. 1993, 115, 9234.
(8) Laidig, K. E.; Cameron, L. M. Can. J. Chem. 1993, 71, 872.
(9) Wiberg, K. B.; Rablen, P. R. J. Am. Chem. Soc. 1995, 117, 2201.
(10) Laidig, K. E.; Cameron, L. M. J. Am. Chem. Soc. 1996, 118, 1737.
(11) Also see: Knight, E. T.; Allen, L. C. J. Am. Chem. Soc. 1995, 117,
4401.
(12) Bader, R. F. W. Atoms in Molecules: A Quantum Theory; Clarendon
Press: Oxford, U.K., 1990.
12940 J. Am. Chem. Soc. 1997, 119, 12940-12946
S0002-7863(97)00074-7 CCC: $14.00 1997 Ameri can Chemi cal Soci ety
thioformamide has, in fact, a larger barrier than formamide.
Furthermore, AIM again reveals the significant transfer of charge
from C to N in the planar geometry (rather than from N to S).
Laidig and Cameron
10
thus argued that resonance is relatively
unimportant in thioformamide. Wiberg and Rablen,
9
however,
found significant transfer of density from N to S in electron
density difference maps. They concluded that the amide
resonance picture is more appropriate for describing the thioa-
mides than their O-based analogues.
In the present work, we reexamine the role of resonance
interactions in formamide, thioformamide, and their Se and Te
replacement analogues. Whereas previous studies
4-10
focused
largely on the AIM analysis of the calculated charge densities,
we present here an alternative analysis of amide resonance based
on Weinholds natural bond orbital (NBO) methods.
13-16
In
contrast to the AIM results, NBO provides a simple, quantitative
description of the amides that is consistent with the conventional
resonance-based picture. Particular attention is focused on the
natural population analysis (NPA)
14
and natural resonance theory
(NRT)
15
treatment of the calculated charge densities.
II. Methods
Electronic structure calculations were carried out using the Gaussian
94
17
and GAMESS
18
programs. Geometries were fully optimized at
the RHF, MP2, and B3LYP levels of theory.
19,20
Standard 6-31+G*
basis sets were employed for H, C, N, O, and S together with the
effective core potentials (ECPs) and valence basis sets for Se and Te
of Bergner et al.
21
The latter are [2s3p] contractions of (4s5p) primitive
sets, augmented by standard d-type polarization functions [R
d(Se) )
0.315, R
d(Te) ) 0.237]
18
and by diffuse s- and p-type functions [Rs-
(Se) ) 0.050224, R
p(Se) ) 0.010550, Rs(Te) ) 0.041359, Rp(Te) )
0.009158].
Resonance effects in the amides were investigated using NRT,
15
which describes calculated charge densities in terms of resonance
structures, weights, and bond orders. Briefly, the NRT formalism
expands the one-electron, reduced density operator ,
as a resonance hybrid of density operators {
R},
where each R corresponds to the idealized resonance structure
wavefunction
R. The resonance weights {wR} are variationally
optimized to give the best description of , subject to the constraints
The natural bond order between any two atoms A and B is given by
the expression
where b
AB
(R)
is the integer number of bonds between the two atoms in
resonance structure R.
For closed-shell calculations, the density operators {
R} of eq 2 are
constructed from doubly occupied NBOs.
13
Covalent-ionic resonance
interactions of, for example, the form A T B
are treated by single polar covalent structures (of connectivity A)
comprised of polarized NBOs. By permitting ionic character to enter
the resonance hybrid through the polar covalence of the contributing
structures, NRT avoids rather lengthy expansions and instead describes
the molecular charge density using only a handful of structures. NRT
typically generated 15-20 candidate structures for the amides studied
here, but only a few of these retained any significant weight following
optimization. Due to the rather strong delocalizing interactions in the
planar amides, a two-reference NRT treatment (based on both structures
I and II) was performed for all geometries. Additional details of the
NRT implementation are provided elsewhere.
15,16
III. Geometries and Rotation Barriers
Three stationary points were calculated for each of the amides.
These correspond to the planar equilibrium geometries 1 and
the two twisted saddle point geometries 2 and 3. The latter,
with the N lone pair antiperiplanar to the CH bond, are
somewhat less stable than the geometries of 2 (by 2.4, 1.6, 1.7,
and 1.1 kcal mol
-1
for X ) O, S, Se, and Te, respectively, at
the MP2 level). The optimized CX and CN bond lengths for 1
and 2 are listed in Table 1, together with the calculated rotation
barriers. We also optimized at each level of theory several
additional geometries along the 1 f2 torsional profile. A ghost
center was attached to C such that it makes constrained dihedral
angles of ( with the two amino H atoms. In the planar
geometry 1 (the 0 rotamer), the C-ghost line of centers is
normal to the plane of the molecule. Rotation of the ghost about
C in 15 increments with full reoptimization of the angle and
all other internal coordinates gave a series of five intermediate
rotamers along the minimum energy path up to the saddle point
2 (the 90 rotamer). Figure 1 shows the resulting torsional
profiles for the amides at the MP2 level.
Chalcogen replacement, from O to Te, increases the amide
rotation barrier somewhat, from 17.2 kcal mol
-1
in formamide
to 21.0 kcal mol
-1
in telluroformamide. A similar increase was
noted by Wiberg and Rablen
9
and Laidig and Cameron
10
in their
studies of thioformamide. As the qualitative details of our
(13) (a) Foster, J. P.; Weinhold, F. J. Am. Chem. Soc. 1980, 102, 7211.
(b) Reed, A. E.; Curtiss, L. A.; Weinhold, F. Chem. ReV. 1988, 88, 899.
(14) NPA: Reed, A. E.; Weinstock, R. B.; Weinhold, F. J. Chem. Phys.
1985, 83, 735. NPA is implemented in L607 (NBO 3.1) of the Gaussian
program (ref 17).
(15) NRT: (a) Glendening, E. D. Ph.D. Thesis, University of Wisconsin,
Madison, WI, 1991. (b) Glendening, E. D.; Weinhold, F. J. Comput. Chem.,
submitted. NRT is implemented in version 4.0 of the NBO program (ref
16).
(16) (a) Glendening, E. D.; Badenhoop, J. K; Reed, A. E.; Carpenter, J.
E.; Weinhold, F. NBO 4.0; Theoretical Chemistry Institute: Madison, WI,
1996. (b) Weinhold, F. NBO 4.0 Program Manual; Theoretical Chemistry
Institute: Madison, WI, 1996.
(17) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Gill, P. M. W.;
Johnson, B. G.; Robb, M. A.; Cheeseman, J. R.; Keith, T. A.; Petersson,
G. A.; Montgomery, J. A.; Raghavachari, K.; Al-Laham, M. A.; Zakrzewski,
V. G.; Ortiz, J. V.; Foresman, J. B.; Cioslowski, J.; Stefanov, B. B.;
Nanayakkara, A.; Challacombe, M.; Peng, C. Y.; Ayala, P. Y.; Chen, W.;
Wong, M. W.; Andres, J. L.; Replogle, E. S.; Gomperts, R.; Martin, R. L.;
Fox, D. J.; Binkley, J. S.; Defrees, D. J.; Baker, J.; Stewart, J. P.; Head-
Gordon, M.; Gonzalez, C.; Pople, J. A. Gaussian 94, ReV C.3; Gaussian,
Inc.: Pittsburgh, PA, 1995.
(18) Schmidt, M. W.; Baldridge, K. K.; Boatz, J. A.; Elbert, S. T.;
Gordon, M. S.; Jensen, J. H.; Koseki, S.; Matsunaga, N.; Nguyen, K. A.;
Su, S.; Windus, T. L.; Dupuis, M.; Montgomery, J. A., Jr. J. Comput. Chem.
1993, 14, 1347.
(19) For a comprehensive review of the ab initio methods and all-electron
basis sets employed in this work, see: Hehre, W. J.; Radom, L.; Schleyer,
P. v. R.; Pople, J. A. Ab Initio Molecular Orbital Theory; Wiley: New
York, 1986.
(20) B3LYP: (a) Becke, A. D. J. Chem. Phys. 1993, 98, 5648. (b) Lee,
C.; Yang, W.; Parr, R. G. Phys. ReV. B 1988, 37, 785.
(21) Bergner, A.; Dolg, M.; Kuechle, W.; Stoll, H.; Preuss, H. Mol. Phys.
1993, 80, 1431.
) N

(1,2,...,N) *(1,2,...,N) d2...dN (1)

R
w
R

R
(2)
w
R
g 0,

R
w
R
b
AB
(R)
(3)
b
AB
)

R
w
R
b
AB
(R)
(4)
Resonance in Formamide and Its Analogues J. Am. Chem. Soc., Vol. 119, No. 52, 1997 12941
calculations are essentially identical at the various levels of
theory employed, we will primarily focus on the MP2 results.
Rotation strongly influences the amide geometries, particu-
larly the CN bond length and, to a lesser degree, the CX bond
length. The rotation of the planar geometries 1 into the twisted
forms 2 is accompanied by a significant elongation of the CN
bond. For example, the CN bond of formamide lengthens by
0.079 , from 1.362 in 1 to 1.441 in 2. Similar, although
somewhat larger, elongations are calculated for the CN bonds
of the other amides (0.086, 0.088, and 0.091 for S, Se, and
Te, respectively). At the same time, the CX bond contracts
slightly (by 0.009, 0.015, 0.017, and 0.020 for O, S, Se, and
Te, respectively). Clearly, the influence of rotation on bond
length is more significant for the CN bond than for the CX.
As noted previously for formamide and thioformamide,
4-10
the disparity of these bond elongations/contractions conflicts
somewhat with the resonance description of the amides.
However, two electronic effects are involved in the amide
rotations that can account, in part, for the differences. First,
conjugative interactions involving the N lone pair and the CX
system can shorten or lengthen a bond by increasing or
decreasing its double-bond character. This is the delocalizing
effect treated by resonance theory. Second, a change of
hybridization will tend to shorten or lengthen a bond as
p-character is lost from or gained by its bonding hybrids.
Conjugative interactions alone should influence the CX bond
length as C and X are roughly sp
2
hybridized in both 1 and 2.
But conjugation and hybridization both influence the CN bond
length, the latter as N rehybridizes from sp
2
to sp
3
during the 1
f 2 rotation.
Constrained reoptimizations of the twisted amides were
performed to determine the influence of rehybridization on CN
bond length. Twisted geometries 4 were optimized in which
the amino group was constrained to have a locally planar
configuration (hence, sp
2
hybridized N), while the N lone pair
was rotated 90 out of conjugation with the CX system. The
difference between the optimized CN bond lengths of 4 and 2
reflects the influence of rehybridization alone as the N lone pair
remains localized in both geometries. The MP2-optimized CN
bond lengths of 4 (1.416, 1.410, 1.406, and 1.406 for O, S,
Se, and Te, respectively) are 0.025-0.030 shorter than those
of 2. Thus, of the 0.079 contraction calculated for formamide,
0.025 results from the rehybridization of N while the
remaining 0.054 arises from conjugative stabilization and
other effects (Vide infra). In general, roughly one-third of the
elongation of the CN bond during the 1 f2 rotation should be
attributed to rehybridization.
IV. Population Analysis
Amide resonance suggests that the charge density of the
planar geometry 1 should reflect the transfer of electron density
from N to the chalcogen relative to that of the twisted geometry
2. Whereas 2 is well represented by a single Lewis structure
having no formal charges, 1 includes a significant contribution
of the dipolar form II exhibiting formal positive and negative
charges on the N and chalcogen centers, respectively. We
consider here the influence of rotation (2 f 1) on the natural
charges of NPA. Figure 2 shows the torsional dependence of
the MP2 charges relative to the values calculated for the twisted
geometries. The trends reflected in the RHF and B3LYP
Table 1. Optimized Structural Parameters and Rotation Barriers in
the Amides NH2C(dX)H
planar (1) twisted (2)
X rCX () rCN () rCX () rCN ()
Erot
(kcal mol
-1
)
RHF/6-31+G*
O 1.195 1.348 1.184 1.426 16.1
S 1.641 1.325 1.605 1.423 21.0
Se 1.790 1.318 1.742 1.420 22.8
Te 2.027 1.309 1.957 1.420 26.9
B3LYP/6-31+G*
O 1.220 1.362 1.208 1.441 18.4
S 1.649 1.347 1.627 1.436 22.0
Se 1.793 1.342 1.766 1.431 22.7
Te 2.004 1.338 1.974 1.429 24.6
MP2/6-31+G*
O 1.230 1.362 1.221 1.441 17.2
S 1.636 1.352 1.621 1.438 19.5
Se 1.782 1.347 1.765 1.435 20.4
Te 1.992 1.344 1.972 1.435 21.0
Figure 1. MP2 torsional profiles of formamide and its S, Se, and Te
replacement analogues. The 0 and 90 rotamers correspond, respec-
tively, to the geometries 1 and 2.
Figure 2. Torsional dependence of the MP2 natural charges of
formamide and its S, Se, and Te replacement analogues. The charges
are displayed relative to the absolute MP2 values of the twisted
geometries 2 (the 90 rotamers).
12942 J. Am. Chem. Soc., Vol. 119, No. 52, 1997 Glendening and Hrabal
charges are qualitatively identical to those at the MP2 level, so
we only focus on the latter.
22
The torsional dependence of the natural charges is largely
consistent with the amide resonance model. Rotating the twisted
geometries into the planar forms is principally accompanied by
charge transfer from N to the chalcogen. For example, in
formamide, N loses 0.105e while O gains 0.088e. The
remaining charge density tends to collect at C (0.033e).
Somewhat larger charge transfers are observed for the heavier
chalcogens with greater involvement of the C atom. Thus, in
telluroformamide, we find the transfer of 0.273e to Te, primarily
from N (0.173e) and C (0.062e). Similar effects are calculated
for the S and Se analogues.
It is perhaps surprising that charge transfer in the amides is
stronger for the heavier chalcogens than for the lighter ones.
Strong charge transfer implies a large contribution of the dipolar
form II, which exhibits a formal negative charge on the
chalcogen. The decreasing electronegativity of the chalcogens
down the periodic table might lead one to anticipate diminished
resonance stabilization and charge transfer. Nevertheless, the
stronger charge transfer suggested by NPA is consistent with
the increased rotation barriers and decreased CN bond lengths
of the seleno- and telluroformamides. Clearly, it appears that
the heavier chalcogens accommodate electron density more
readily than expected on the basis of electronegativity. This
property of the heavier chalcogens likely stems from their larger
polarizabilities relative to the lighter atoms.
9,10
In contrast to the natural charges, the AIM charges of Figure
3 reveal charge transfer that is largely inconsistent with the
traditional amide resonance picture. The principal charge
transfer during rotation (2 f 1) is from C to N.
9,10
Clearly,
NPA and AIM give distinctly differing representations of the
amide charge densities. What are we to conclude from this?
Atomic charges cannot be uniquely defined, so neither NPA
nor AIM should be considered to give the correct charges.
We may inquire, however, which method yields the more
appealing description of rotation in the amides. Here, it would
seem that NPA is preferred as the natural charge shifts are
essentially consistent with those anticipated from conventional
resonance theory and chemical intuition. But perhaps the more
compelling argument is that an orbital-based (Hilbert space)
analysis like NPA supports the amide resonance model, an
orbital-based description of electronic structure. AIM, a
coordinate-based (real space) analysis, gives a largely contrasting
view.
V. Natural Resonance Theory
NRT describes planar formamide (Table 2) as roughly a 2:1
mixture of two resonance structures, the parent Lewis structure
I and secondary dipolar form II, respectively. At the RHF level,
NRT calculates a set of 13 candidate structures, of which only
the leading four are listed in Table 2. As expected, the structure
of highest weight (65.40%) is the Lewis form I. NRT
recognizes a strong n
N
f *
CO
interaction in the parent Lewis
structure. This interaction is formally equivalent to the mixing
of the dipolar form II (28.54%) in the resonance hybrid as it
lends formamide CN double-bond character while tending to
break the CO bond. The Lewis and dipolar forms together
account for 94% of the resonance hybrid. Two additional
structures, III and IV (X )O) contribute about 2% each. These
structures exhibit CO triple bonds and arise, respectively, from
the delocalizing interactions of the in-plane p-type lone pair on
O with the vicinal CN and CH antibonds (n
O
f *
CN
, n
O
f
*
CH
). All other resonance forms have weights of less than 1%.
The B3LYP and MP2 level treatments of formamide yield a
slightly more delocalized description of formamide. Electron
correlation tends to weaken the contribution of the parent
structure while increasing the weights of several of the secondary
forms. Thus, we find that the Lewis and dipolar forms account
for only 87% of the resonance hybrid at the MP2 level, down
from 96% at RHF. Nevertheless, NRTs compact description
of formamide remains qualitatively unchanged: formamide is
principally a resonance hybrid of two contributing structures.
Natural bond orders reveal a significant exchange of bonding
interaction in planar formamide, the CN bond gaining double-
bond character at the expense of the carbonyl. At the RHF
level, the CN bond (b
CN
) 1.292) has roughly 29% double-
bond character while the CO double bond has about 26% single-
bond character (b
CO
) 1.744). Electron correlation tends to
increase the CN double-bond character. At the MP2 level, NRT
calculates a CN bond order of 1.340, about 5% larger than the
RHF value.
(22) Natural charges for the planar and saddle point geometries of
formamide and thioformamide were previously reported by Wiberg and
Rablen (ref 9) at the MP2(full)/6-31+G* level but were not discussed in
detail.
Figure 3. Similar to Figure 2, for the AIM charges.
Table 2. Natural Resonance Weights and Bond Orders for Planar
Formamide
a
structure interaction
b
RHF B3LYP MP2
65.40 61.62 58.63
nN f*
CO
28.54 30.86 28.61
nO f*
CN
1.95 2.35 3.10
nO f*
CH
1.93 2.31 2.05
others 2.22 2.86 7.61
bond orders RHF B3LYP MP2
bCN 1.292 1.320 1.340
b
CO 1.744 1.725 1.717
a
All geometries are optimized at the respective RHF, B3LYP, and
MP2 levels of theory. Percentage weights are listed.
b
The orbital
interaction of the parent Lewis structure that gives rise to the secondary
forms.
Resonance in Formamide and Its Analogues J. Am. Chem. Soc., Vol. 119, No. 52, 1997 12943
Replacing the O of formamide by one of the heavier
chalcogens strengthens the principal n
N
f *
CX
resonance
interaction. Table 3 compares planar formamide with its
chalcogen replacement analogues at the MP2 level. The weight
of structure II increases from 28.61% in formamide to 29.23%,
31.01%, and 32.66% in the S, Se, and Te analogues, respec-
tively. As a result, CN double-bond character increases from
34.0% in formamide to 40.5% in telluroformamide. These
trends are fully consistent with the increasing rotation barriers
and decreasing CN bond lengths discussed in section III and
with the increasingly strong charge transfer from N to the
chalcogen reflected by the natural charges of section IV.
Twisting the amides results in a considerable reduction in
resonance stabilization. Figure 4 shows the torsional depen-
dence of the MP2 weights for the Lewis (circles) and dipolar
(squares) structures. There is a monotonic decrease in the
weight of the dipolar structure (and concomitant increase for
the Lewis structure) as the amide is rotated from the planar into
the twisted geometry. In formamide, the planar geometry has
a 28.61% contribution from the dipolar form. This contribution
decreases considerably to 7.79% in the twisted geometry as a
strong interaction (n
N
f *
CO
) is exchanged for a substan-
tially weaker one (n
N
f *
CO
).
A simple resonance treatment would suggest that the dipolar
contribution II to the resonance hybrid should diminish with
the decreasing electronegativity of the chalcogen.
9,10
In fact,
the opposite trend is reflected by the NRT weights of Table 3.
To determine the origin of this effect, we examined the character
of the CX NBO of the parent Lewis structure and, in particular,
how the polar covalent character of this orbital is influenced
by the presence of a conjugating N lone pair. Bonding and
antibonding NBOs {
AB
, *
AB
} are, respectively, in-phase and
out-of-phase superpositions of two orthonormal atomic hybrids
{h
A
, h
B
}
with coefficients {c
A
, c
B
} chosen to maximize the occupancy
of
AB
. The polar covalence of these orbitals can be judged
from the bond polarization (c
A
2
). A bond polarization of 50%
reflects a covalent bond, whereas polarizations of 100% and
0% describe fully ionic bonds, polarized completely toward
atoms A and B, respectively. Figure 5 compares the polariza-
tions of the
CX
and
CX
NBOs of the planar and twisted amides
(1 and 2) and aldehydes 5,
where X ) O, S, Se, and Te. The aldehydes are used to judge
the nominal bond polarizations of the CX bonds in the absence
of the amino group. The polarizations of Figure 5 are defined
such that values larger than 50% correspond to bond polarization
in the direction of the chalcogen.
The orientation of the N lone pair strongly influences the
polarization of the
CX
bond, particularly for the heavier
chalcogens. As shown in Figure 5, the
CX
bonds of 2 and 5,
which are not conjugated by an N lone pair, are essentially
polarized to the same degree. Twisted formamide and form-
aldehyde, for example, have
CO
bonds that are 66.5% and
65.6% polarized toward O. Twisting the N lone pair into a
conjugating position, as in 1, tends to polarize further the
CX
bond toward the chalcogen (e.g., 70.1% in planar formamide).
This effect is enhanced somewhat for the more polarizable
chalcogens. Whereas the change in bond polarization in
Table 3. MP2 Natural Resonance Weights and Bond Orders for
Planar Formamide and its S, Se, and Te Replacement Analogues
a
structure X ) O X ) S X ) Se X ) Te
58.63 58.61 57.13 56.09
28.61 29.23 31.01 32.66
3.10 2.80 2.52 2.23
2.05 1.49 1.23 0.80
others 7.61 7.87 8.11 8.22
bond orders X ) O X ) S X ) Se X ) Te
bCN 1.340 1.355 1.381 1.405
bCX 1.717 1.694 1.664 1.636
a
All geometries are optimized at the MP2 level. Percentage weights
are listed.
Figure 4. Torsional dependence of the MP2 natural weights of the
Lewis (I) and dipolar (II) resonance structures for formamide and its
S, Se, and Te replacement analogues.
Figure 5. Torsional dependence of the MP2 polarizations of the CX
and CX bonds for formamide and thioformamide.

AB
) c
A
h
A
+ c
B
h
B
(5a)
*
AB
) c
B
h
A
- c
A
h
B
(5b)
12944 J. Am. Chem. Soc., Vol. 119, No. 52, 1997 Glendening and Hrabal
formamide is about 5% (from 66.5% to 70.1%), the change in
the Te analogue is nearly 15%, from 49.8% in twisted
telluroformamide to 64.1% in the planar form. (The
CX
bonds
of 1, 2, and 5 are relatively insensitive to the presence or
orientation of the N lone pair.) Polarizing the
CX
bond of 1
toward the chalcogen reverse polarizes the *
CX
antibond
toward C, making the latter a better acceptor for resonance
interactions with the N lone pair (n
N
f*
CX
). Thus, the higher
polarizability of the heavier chalcogens facilitates repolarization
of the *
CX
antibond, leading to stronger resonance stabiliza-
tion. Wiberg and Rablen
9
noted similar effects in a FMO
analysis of formamide and thioformamide.
The lengthening of the CN bond as an amide is rotated from
the planar to the twisted geometry is consistent with the loss of
double-bond character reflected in the natural bond orders.
Figure 6 shows the approximate linear relationship between the
MP2-optimized CN bond lengths and natural bond orders for
formamide and its chalcogen replacement analogues. For each
of the amides, we show seven data points corresponding to the
seven rotamer geometries optimized along the 1 f 2 torsional
profile (cf. section III). Regression analysis gives a best fit line
with a y-intercept of 1.467 (at b
CN
) 1) and slope of -0.317
. The intercept is, in fact, essentially identical to the nominal
single-bond length of methylamine (1.465 at MP2/6-31+G*),
although the latter has an sp
3
hybridized C compared to the sp
2
hybridized one of 2.
The CX bond lengths of the amides also exhibit an ap-
proximate linear correlation with natural bond order. Figure 7
shows the MP2-optimized CX bond lengths as a function of
bond order for formamide and its S, Se, and Te analogues.
Regression analysis of the data was performed using a linear
expression of the form
where m is the CX bond length dependence on bond order and
R
2
is the y-intercept at b
CX
) 2 (the extrapolated, idealized
double-bond length). Regression parameters are given in Table
4, together with the MP2-optimized CX bond lengths of the
aldehydes 5. The optimal R
2
values differ only marginally (by
0.009 or less) from the ideal CX double-bond lengths of the
aldehydes.
The disparity between the slopes of Figures 6 and 7 is closely
related to the differences in the CN and CX bond length
variations discussed in section III. For example, whereas we
find the CN bond length to contract 0.317 per unit bond order
(Figure 6), the CO bond contracts only 0.047 (Figure 7 and
Table 4). These values reflect the relatively large variation in
CN bond length compared to the small variation for CO in
formamide. As discussed in section III, a portion (perhaps one-
third) of the difference arises from the rehybridization of N
during rotation. There are, however, several additional factors
that could contribute to the discrepancy. First, as shown in
Figure 5, the
CX
bonds of the planar amides 1 are more ionic
(polar) than those of the twisted forms 2. The greater ionic
character of CX in the planar geometries tends to strengthen
and shorten these bonds despite the loss of double-bond
character in II. Second, vibrational stretching frequencies
suggest that double bonds are typically stiffer than single bonds.
For example, the CO double bonds of amides have stretching
frequencies near 1700 cm
-1
, compared to 1180-1360 cm
-1
for
the CN single bonds of amines.
3
One might anticipate,
therefore, that conjugation in the amides would more strongly
influence the CN bond length than the CX bond. Third, the
contributions of the triply-bonded structures III and IV may
further stiffen the CX bonds. These structures account for
roughly 5% of the resonance hybrid for planar formamide. Thus,
through a combination of ionic and multiple bond character and
rehybridization, it seems reasonable that the CN bond lengths
of the amides are more strongly influenced by rotation than the
CX bonds.
Finally, we find that the dipolar structure II is principally
responsible for the planar configuration of the amino group.
To determine the influence of this structure, we deleted its
contribution from the resonance hybrid and reoptimized the
planar amide geometry 1. This was accomplished by zeroing
the n
N
f *
CX
element of the NBO Fock matrix and construct-
ing a localized wavefunction
loc
from the resulting eigen-
vectors. Geometry optimization was then performed to mini-
mize the energy of
loc
. Details of these calculations are given
in Table 5. The resonance energies RE (the energy differences
between the ground state and localized amides in their respective
optimized geometries) are a measure of the extra stability of
Figure 6. Approximate linear correlation of the MP2-optimized CN
bond lengths with natural bond order for formamide and its S, Se, and
Te replacement analogues.
Figure 7. Similar to Figure 6, for the CX bond lengths of formamide
and its S, Se, and Te replacement analogues. Regression parameters
are given in Table 4.
Table 4. Regression Parameters of the MP2 CX Bond Length/
Bond Order Correlation
a
X m () R2 () RCX
b
()
O -0.047 1.216 1.225
S -0.059 1.618 1.619
Se -0.059 1.762 1.760
Te -0.076 1.963 1.961
a
See text eq 7.
b
MP2-optimized CX bond lengths in formaldehyde,
H2CdO, and its S, Se, and Te replacement analogues.
R
CX
) R
2
+ m(b
CX
- 2) (6)
Resonance in Formamide and Its Analogues J. Am. Chem. Soc., Vol. 119, No. 52, 1997 12945
the planar amide geometries resulting from the dipolar contribu-
tion. As anticipated from our NRT results, RE increases
monotonically down the periodic table, from 25.1 kcal mol
-1
in formamide to 57.1 kcal mol
-1
in telluroformamide.
These resonance energies are considerably larger than the
corresponding rotation barriers (cf. Table 1). That is, the
localized amides in the planar geometries 1 are less stable than
the twisted geometries 2. Indeed, vibrational frequencies
analysis of the localized amides revealed that the planar
geometries are unstable with respect to an out-of-plane distortion
at N. Full geometry reoptimizations with no symmetry con-
straints allowed the localized amides to revert to the twisted
geometries 2. Thus, structure II (or alternatively the n
N
f
*
CX
interaction) is responsible for the planar amide geometry.
In the absence of its contribution, the amides would likely
exhibit pyramidal geometries rather than planar ones.
VI. Summary
We have examined the rotation barriers of formamide and
its S, Se, and Te replacement analogues using the NBO methods.
NPA reveals the apparent transfer of electrons from N to the
chalcogen in the planar, equilibrium amide geometries. NRT
represents the planar amides as resonance hybrids consisting
principally of two contributing structures, the parent Lewis form
I and a secondary dipolar form II. In formamide, the sizable
contribution of the dipolar form accounts for the planar amino
geometry, large rotation barrier, and relatively short CN bond
length. The weight of the dipolar form increases monotonically
from formamide to telluroformamide in accord with the increas-
ing rotation barrier and decreasing CN bond length. The larger
polarizabilities of the heavier chalcogens allow these atoms to
accommodate more charge density than anticipated on the basis
of electronegativity. In short, the NBO methods reveal torsional
behavior for formamide and its chalcogen replacement analogues
that is largely consistent with the conventional amide resonance
model.
Acknowledgment. This research was supported in part by
a faculty startup grant from the Camille and Henry Dreyfus
Foundation.
Supporting Information Available: A listing of the opti-
mized amide geometries (in Gaussian 94 format) and raw
energies (22 pages). See any current masthead page for ordering
and Internet access instructions.
JA970074J
Table 5. Resonance Energies (RE) and Reoptimized Bond Lengths
of the Localized Amides
a
X RE (kcal mol
-1
) rCX () rCN ()
O 25.1 1.170 1.497
S 48.4 1.580 1.497
Se 51.4 1.712 1.498
Te 57.1 1.923 1.505
a
RHF/6-31+G* values.
12946 J. Am. Chem. Soc., Vol. 119, No. 52, 1997 Glendening and Hrabal