Bonding along a linear B...N...B triad

Article abstract of DOI:10.1021/jo040199o

The synthesis of tris(2,6-dihydroxyphenyl)amine diborate, 4, is reported. This compound contains a linear B...N...B array for which a symmetrical three-center two-electron (3c-2e) bond is possible. The X-ray crystal structure of 4 shows that 3c-2e bonding

Full text of DOI:10.1021/jo040199o

Bon d in g a lon g a Lin ea r B‚‚‚N‚‚‚B Tr ia d
Peter D. Livant,* D. J ohn D. Northcott, Yiping Shen, and Thomas R. Webb
Department of Chemistry, Auburn University, Auburn, Alabama 36849-5312
livanpd@auburn.edu
Received May 17, 2004
The synthesis of tris(2,6-dihydroxyphenyl)amine diborate, 4, is reported. This compound contains
a linear B‚‚‚N‚‚‚B array for which a symmetrical three-center two-electron (3c-2e) bond is possible.
The X-ray crystal structure of 4 shows that 3c-2e bonding is, in fact, absent. Rather, the B-N-B
array of 4 is unsymmetrical, having a 2c-2e B-N dative bond with the remaining boron
pyramidalized outward and bonded to the oxygen of THF, i.e., 4‚THF. In THF solution, 4 displays
temperature-dependent 13C NMR spectra from which a ∆Gq of 11.6 kcal/mol at 262 K may be
calculated. The dynamic process observed in solution corresponds to a bond-switching equilibrium
in which the B-N bond oscillates between the two borons (“bell clapper”). Ab initio calculations
indicate that the most likely pathway for the bond switch does not involve a 3c-2e B‚‚‚N‚‚‚B bond,
but rather occurs by nucleophilic attack of THF on the datively bonded boron to generate 4‚(THF)2,
lacking any B-N interactions, followed by loss of one THF. The B-N-B system of 4 sans the
perturbing effect of solvent was also investigated computationally. The form of 4 containing a
3c-2e bond is found to be a transition state in the solvent-free bond-switch reaction of 4, lying 2.66
kcal/mol above 4. The stability of three-center bonds to in-line distortion (viz., X‚‚‚Y‚‚‚X f
X-Y‚‚‚‚‚‚‚‚‚X) is discussed from the point of view of the second-order J ahn-Teller effect.
The syntheses decades ago of the noble gas halides1
provided the impetus for extending and modifying then-
current ideas about bonding. The conceptual advance
called for was provided by Pimentel,2 and Hach and
Rundle,3 who proposed the three-center four-electron
(3c-4e) bond to explain bonding in interhalogen anions
(e.g., ICl2-) as well as in the noble gas halides.4 The usual
orbital diagram of a 3c-4e bond is shown in Figure 1 for
a linear X-Y-X triad.
The 3c-4e bond concept has a history of 50 years of
success in rationalizing and predicting the chemistry of
F IGURE 1. An MO picture of a 3c-4e bond in the X-Y-X
so-called hypervalent5 systems, yet it is still, for some,
array. In parentheses, for comparison, is a diagram of the
an object of skeptical scrutiny.6 Until fairly recently, all
known hypervalent systems had central atoms which
analogous π-allyl system.
were nonmetals found in the third row of the periodic
table and beyondse.g., SF6, PhI(OAc)2, etc.swith the
exception of the bifluoride ion, FHF-, a unique case of
hypervalency in the first row.7 Chemists had come to
believe hypervalency in elements of the second row was
highly improbable. However, the seminal work of Martin
and later Akiba, and others, has recently provided
startling examples of hypervalency in the second-row
elements, to wit, in boron,8 carbon,9 and fluorine.10 An
isolable example of nitrogen hypervalency (10-N-5) has
* To whom correspondence should be addressed.
(1) Laszlo, P.; Schrobilgen, G. J . Angew. Chem., Int. Ed. Engl. 1988,
27, 479-489.
(2) Pimentel, G. C. J . Chem. Phys. 1951, 19, 446-448.
(3) Hach, R. J .; Rundle, R. E. J . Am. Chem. Soc. 1951, 73, 4321-
4324.
(4) Rundle, R. E. J . Am. Chem. Soc. 1963, 85, 112-113.
(5) We adopt here the Musher definition of this term. (a) Musher,
J . I. Angew. Chem., Int. Ed. Engl. 1969, 8, 54-68. (b) Schleyer, P. v.
R. Chem. Eng. News 1984, May 28, 4. (c) Martin, J . C. Ibid. 1984,
May 28, 4. (d) Harcourt, R. D. Ibid. 1985, J an 21, 3.
(6) (a) Kutzelnigg, W. Angew. Chem., Int. Ed. Engl. 1984, 23, 272-
295. (b) Mayer, I. J . Mol. Struct.: THEOCHEM 1987, 149, 81-89. (c)
Mayer, I. Ibid. 1989, 186, 43-52. (d) Kar, T.; Marcos, E. S. Chem. Phys.
Lett. 1992, 192, 14-20. (e) Ha¨ser, M. J . Am. Chem. Soc. 1996, 118,
7311-7325. (f) Harcourt, R. D. J . Phys. Chem. A 1999, 103, 4293-
4297. (g) Sannigrahi, A. B.; Kar, T. J . Mol. Struct.: THEOCHEM 2000,
496, 1-17. (h) Molina, J . M.; Dobado, J . A. Theor. Chem. Acc. 2001,
105, 328-337. (i) Ponec, R.; Roithova´, J . Ibid. 383-392. (j) Papoian,
G.; Hoffmann, R. J . Am. Chem. Soc. 2001, 123, 6600-6608. (k) Ienco,
A.; Hoffmann, R.; Papoian, G. Ibid. 2317-2325. (l) Papoian, G.;
Hoffmann, R. J . Solid State Chem. 1998, 139, 8-21.
(7) Chemistry of Hypervalent Compounds; Akiba, K.-y., Ed.; Wiley-
VCH: New York, 1999.
(8) (a) Lee, D. Y.; Martin, J . C. J . Am. Chem. Soc. 1984, 106, 5745-
5746. (b) Yamashita, M.; Yamamoto, Y.; Akiba, K.-y.; Nagase, S.
Angew. Chem., Int. Ed. 2000, 39, 4055-4058.
(9) (a) Forbus, T. R., J r.; Martin, J . C. J . Am. Chem. Soc. 1979, 101,
5057-5059. (b) Forbus, T. R., J r.; Martin, J . C. Heteroat. Chem. 1993,
4, 113-128. (c) Forbus, T. R., J r.; Martin, J . C. Ibid. 1993, 4, 129-
136. (d) Forbus, T. R., J r.; Martin, J . C. Ibid. 1993, 4, 137-143. (e)
Akiba, K.-y.; Yamashita, M.; Yamamoto, Y.; Nagase, S. J . Am. Chem.
Soc. 1999, 121, 10644-10645.
10.1021/jo040199o CCC: $27.50 © 2004 American Chemical Society
Published on Web 09/01/2004
6564
J . Org. Chem. 2004, 69, 6564-6571

Bonding along a Linear B‚‚‚N‚‚‚B Triad
SCHEME 1
not yet been achieved, although attempts have been
reported sporadically since 1916.11 Transient 9-N-4 spe-
cies have been observed by ESR12 and by neutralization-
reionization mass spectrometry.13 Theorists have pon-
dered hypothetical hypervalent nitrogen species14 and
other second-row hypervalent systems.6d,15
If one is to succeed in preparing a 10-N-5 species, it is
reasonable to guess that (i) the geometry about 10-N-5
nitrogen will be trigonal bipyramidal and (ii) the five
ligands around nitrogen will have to be held in place.
Indeed, Christe et al. have pointed out that fitting five
fluorines about a central nitrogen may be impossible.16
Therefore, we sought to design a suitable scaffolding to
hold in place a possibly recalcitrant three-center bond
having nitrogen as the central element.
In Figure 1, the node structure of the orbitals involved
in a three-center bond can be seen to correspond with
that of an ordinary allyl system, in which three p orbitals
interact in a π fashion. Because of this, the three-center
bond is sometimes termed a “σ-allyl” system. Baldridge,
Leahy, and Siegel (BLS)17 calculated that a proposed all-
carbon σ-allyl cation, 1b, with a symmetrical 3c-2e bond,
was not a global minimum, but rather a transition
structure connecting ordinary carbocations 1a and 1c.
cc-pvdz) calculations. BLS noted that in cases which
lacked the polycyclic framework of 1 (e.g., [CMe3‚‚‚
CMe3‚‚‚CMe3]+) the symmetrical 3c-2e structure analo-
gous to 1b was much more unfavorable energetically, or
was a higher order stationary point. Therefore, in their
words, 1 “comes a long way toward the desired effect” of
stabilizing the delocalized σ-allyl cation (i.e., the 3c-2e
bond).
Reactions of 2 with nucleophiles would lead to 3, in
which a pentacoordinate boron (10-B-5) might be present.
This system was explored in detail by Mu¨ller and Bu¨rgi.18
Our interest in hypercoordinate (if not hypervalent)
nitrogen led us to mate 3 with 1 and generate thereby
target structure 4. This structure would accommodate,
Transition structure 1b was found to be 14.1 kcal/mol
higher in energy than 1a /1c using DFT (B3PW91/
(10) (a) Cahill, P. A.; Dykstra, C. E.; Martin, J . C. J . Am. Chem.
Soc. 1985, 107, 6359. (b) Ault, B. S.; Andrews, L. Inorg. Chem. 1977,
16, 2024. (c) Hunt, R. D.; Thompson, C.; Hassanzadeh, P.; Andrews,
L. Inorg. Chem. 1994, 33, 388-391. (d) Tuinman, A. A.; Gakh, A. A.;
Hinde, R. J .; Compton, R. N. J . Am. Chem. Soc. 1999, 121, 8397-
8398.
(11) (a) Schlenk, W.; Holtz, J . Chem. Ber. 1916, 49, 603. (b) Schlenk,
W.; Holtz, J . Ibid. 1917, 50, 274, 276. (c) Wittig, G.; Wetterling, M.
Liebigs Ann. Chem. 1947, 557, 193. (d) Hellwinkel, D.; Seifert, H. Ibid.
1972, 762, 29. (e) Olah, G. A.; Donovan, D. J .; Shen, J .; Klopman, G.
J . Am. Chem. Soc. 1975, 97, 3559. (f) J ohnson, R. W.; Holm, E. R. Ibid.
1977, 99, 8077. (g) Christe, K. O.; Wilson, W. W.; Schrobilgen, G. J .;
Chirakal, R. V.; Olah, G. A. Inorg. Chem. 1988, 27, 789-790.
(12) (a) Nishikida, K.; Williams, F. J . Am. Chem. Soc. 1975, 97, 7166.
(b) Hasegawa, A.; Hudson, R. L.; Kikuchi, O.; Nishikida, K.; Williams,
F. Ibid. 1981, 103, 3436.
and perhaps enforce, a trigonal bipyramidal geometry at
nitrogen. At the very least, it would ensure that the two
borons and nitrogen would be collinear. A symmetrical
B-N-B triad in 4 would be evidence of a 3c-2e bond,
analogous to that in 1b. If that were the case, the
conversion of the 3c-2e bond of 4 to a 3c-4e system would
be worth exploring as a way of synthesizing a 10-N-5
species. Although this goal has not yet been reached, we
report herein our synthesis of 4 and studies of the nature
of bonding along its B-N-B axis.
(13) (a) Shaffer, S. A.; Sad´ılek, M.; Turecek, F. J . Org. Chem. 1996,
61, 5234-5245. (b) Fuke, K.; Takasu, R. Bull. Chem. Soc. J pn. 1995,
68, 3309-3318. (c) Nguyen, V. Q.; Sad´ılek, M.; Ferrier, J .; Frank, A.
J .; Turecek, F. J . Phys. Chem. A 1997, 97, 6621-6627. (d) Nguyen, V.
Q.; Sadilek, M.; Ferrier, J .; Frank, A. J .; Turecek, F. J . Phys. Chem.
1997, 101, 3789-3799.
(14) (a) Bettinger, H. F.; Schleyer, P. v. R.; Schaefer, H. F., III. J .
Am. Chem. Soc. 1998, 120, 11439-11448. (b) Boldyrev, A. I.; Simons,
J . J . Chem. Phys. 1992, 97, 6621-6627. (c) Ewig, C. S.; Van Wazer, J .
R. Ibid. 1990, 112, 109-114. (d) Ewig, C. S.; Van Wazer, J . R. Chem.
Eng. News 1990, April 2, p. 3. (e) Ewig, C. S.; Van Wazer, J . R. J . Am.
Chem. Soc. 1989, 111, 4172-4178. (f) Morosi, G.; Simonetta, M. Chem.
Phys. Lett. 1977, 47, 396-398. (g) Murrel, J . N.; Scollary, C. E. J . Chem.
Soc., Dalton Trans. 1976, 818-822. (h) Keil, F.; Kutzelnigg, W. Ibid.
1975, 97, 3623-3632.
(15) (a) Harcourt, R. D. Int. J . Quantum Chem. 1996, 60, 553-566.
(b) Vetter, R.; Zu¨licke, L. Chem. Phys. 1986, 101, 201-209. (c) Dedieu,
A.; Veillard, A. J . Am. Chem. Soc. 1972, 94, 6730-6738.
(16) Christe, K. O.; Wilson, W. W.; Schrobilgen, G. J .; Chirakal, R.
V.; Olah, G. A. Inorg. Chem. 1988, 27, 789-790.
(17) Baldridge, K. K.; Leahy, J .; Siegel, J . S. Tetrahedron Lett. 1999,
40, 3503-3506.
Resu lts a n d Discu ssion
The outline of our synthesis of 4 is shown in Scheme
1. The Ullmann reaction of 2,6-dimethoxyaniline with
2,6-dimethoxyiodobenzene (eq 1) was reported by J ackson
et al.19 to give 5 in 9% yield. We modified the procedure
somewhat and obtained a 14% yield of 5. We found, as
J ackson et al. had reported, that prolonged reaction times
resulted in the formation of cyclized product 7. Extending
that result, we found that prolonged heating of 7 under
the reaction conditions gave more highly cyclized prod-
ucts, which we have identified spectroscopically as 8 and
(18) (a) Mu¨ller, E.; Bu¨rgi, H.-B. Helv. Chim. Acta 1987, 70, 499-
510. (b) Ibid. 511-519. (c) Ibid. 520-533. (d) Ibid. 1063-1069.
(19) Stoudt, S. J .; Gopalan, P.; Kahr, B.; J ackson, J . E. Struct. Chem.
1994, 5, 335-340.
J . Org. Chem, Vol. 69, No. 20, 2004 6565

Livant et al.
TABLE 1. 13C NMR Ch em ica l Sh ifts (p p m ) of 6, 10, a n d
4a
C1
C2, C6
157.2
151.9, 159.5
154.3, 160.0
C3, C5
108.2
106.0, 111.9
108.9
C4
hexaphenol 6
monoborate 10
diborate 4
123.6
126.0
124.6
124.8
131.0
129.4
a
THF-d8 solvent; the aromatic carbon bearing N is defined as
C1.
9. Conditions which reproducibly lead to useful yields of
the intriguing structures 8 and 9 have eluded us,
F IGURE 2. ORTEP plot of 4. Nitrogen is depicted in green,
boron in yellow, oxygen in orange, and carbon in gray.
Hydrogens are depicted by open circles.
unfortunately. Curiously, the material to which we assign
the structure 9 on the basis of its FAB mass spectrum
(base peak ) m/z 287 ) MW of 9) gave no peaks in its
1H NMR spectrum.
(as in 6), but C2 and C6 were nonequivalent, indicating
that 4 lacked top/bottom symmetry, as in 10sa small
mystery.
Finally, it was found that reaction of 6 with BH3‚THF
gave the desired diborate 4 (eq 3). Chemical ionization
mass spectrometry of 4 using NH3 as the ionization gas
gave a cluster of three large peaks at m/z 391, 392, and
393. This corresponds to the mass of 4 plus two ammonia
molecules and a proton, and the observed relative inten-
sities of m/z 391, 392, and 393 (0.50, 1.00, and 0.23) were
in agreement with those calculated for a compound
containing two borons (0.50, 1.08, and 0.21).
The X-ray crystal structure of 4 is shown in Figure 2.
A symmetrical three-center two-electron B‚‚‚N‚‚‚B bond
is clearly absent from 4. Rather, the molecule has
acquired a molecule of THF (actually THF-d8) at one
boron, leaving the other boron to participate in an
ordinary two-center two-electron B-N dative bond. Re-
markably, the geometry of the “bottom” portion of 4 (i.e.,
that part containing the B-N bond) is virtually identical
to that of 2 (see Table 2).
Demethylation of 5 with AlCl3 (eq 2) was based on the
procedure used by Frye et al.20 in their synthesis of tris-
(2-hydroxyphenyl)amine. The product amine hexaphenol
6 crystallized as 6‚(EtOAc)2 from EtOAc/hexane. The
X-ray crystal structure of 6‚(EtOAc)2 revealed the geom-
etry about nitrogen was perfectly planar (sum of C-N-C
angles 360.0°) as was the case with the parent hexa-
methoxyamine 5.19 Both compounds have 2-fold rather
than 3-fold symmetry with the crystallographic C2 axis
coincident with one N-C bond. In 5, the differences
between the unique phenyl ring and the other two are
slight (twist angles 62.0°, 61.0°, and 61.0°), but in
6‚(EtOAc)2 the differences are more pronounced (twist
angles 70.3(1)°, 61.7(2)°, and 61.7(2)°). We were encour-
aged that the nitrogen of 6 was planar since a trigonal
bipyramidal nitrogen (either 3c-2e or 3c-4e) would require
the equatorial substituents to lie in a plane containing
the nitrogen.
The 100 MHz 13C NMR spectrum of 4 in THF-d8 was
examined at temperatures between room temperature
and -90 °C. The single line at 108.9 ppm (room temper-
ature), due to C3 and C5, decoalesced as the temperature
was lowered, becoming eventually two singlets of equal
intensity separated by 513 Hz at -90 °C. The small
mystery mentioned above was thus resolved. A series of
spectra taken at 1 °C intervals near the coalescence
temperature (Tc) enabled Tc to be determined as 262 (
0.5 K. Therefore, ∆Gq at 262 K for the dynamic process
which equilibrates C3 and C5 is 11.6 ( 0.1 kcal/mol.
To help us understand the structure of 4 and its
behavior in solution, ab initio calculations were per-
formed on the structures shown in Chart 1. The choice
of a basis set was a matter of concern, since the systems
are unfortunately rather large. Performing calculations
on smaller model systems seemed unwise, because in
these systems there are bonding attractions or repulsions
within the B-N-B triad itself, and there are forces on
the B-N-B triad originating from the “outside”, i.e., from
the conformational strains engendered by the “scaffold-
Several methods were tried to convert 6 to the target
4. Reaction of 6 with tris(dimethylamino)borane produced
an intractable solid. Reaction of 6 with trimethyl borate
(THF-d8 solvent, 65 °C, 1.5 h) gave an equilibrium
mixture of 6, monoborate 10, and the desired diborate 4
as shown in eq 4. The progress of this reaction was
followed most conveniently by 13C NMR. The 13C spectra
of 6, 10, and 4 are summarized in Table 1. We noticed in
the 13C NMR spectrum of 4 that C3 and C5 were
equivalent, indicating that 4 had “top/bottom” symmetry
(20) Frye, C. L.; Vincent, G. A.; Hauschildt, G. L. J . Am. Chem. Soc.
1966, 88, 2727.
6566 J . Org. Chem., Vol. 69, No. 20, 2004

Bonding along a Linear B‚‚‚N‚‚‚B Triad
TABLE 2. Selected Str u ctu r a l P a r a m eter s of 4 a n d 2
param
4
2a
param
4
2a
B(1)-N (Å)
1.683(4)
1.460(3)b
116.2(2)c
114.9(2)
103.3(2)
101.2(2)
109.9(2)
114.6(2)
1.681(5)
1.470(18)
116.4(2)
114.4(3)
103.9(3)
101.1(2)
109.6(3)
115.7(3)
Ph twiste (deg)
5.8(3)
3.1(9)
av C-N (Å)
N-B(2) (Å)
2.959(4)
1.437(9)
1.554(4)
127.1(2)
115.5(7)
102.5(9)
179.6(2)
av C-N-C (deg)
av O-B(1)-O (deg)
av O-B(1)-N (deg)
av B(1)-N-C (deg)
av N-C-Cd (deg)
av C-C-Od (deg)
av B(2)-Of (Å)
B(2)-Og (Å)
av CAr-O-B(2) (deg)
av O-B(2)-Of (deg)
av O-B(2)-Og (deg)
B(1)-N-B(2) (deg)
a
b
Reference 18a. The average C-N lengths in 5 (ref 19) and 6 are 1.416(4) and 1.424(5) Å, respectively. c The average C-N-C angles
in 5 (ref 19) and 6 are 120.0(2)° and 120.7(4)°, respectively. The carbons are part of the N-B(1)-O-C-C five-membered ring. e In the
d
N-B(1)-O-C-C five-membered ring, the “Ph twist” is the torsion angle between N-B(1) and C-C. f Endocyclic. Exocyclic
g
TABLE 3. Geom etr y of A Ca lcu la ted by Va r iou s
Meth od s
CHART 1
B(1)-N B(2)-N
entry
method
(Å)
(Å)
1
2
3
4
5
HF/3-21G
B3LYP/3-21G*
LSDA/BP86/DN*
HF/6-31G(C,H);6-31+G*(B,N,O)
B3LYP/6-31G(C,H);6-31+G*(B,N,O)
2.332
1.843
1.789
1.785
1.766
2.336
2.654
2.652
2.603
2.635
B-N‚‚‚‚‚‚‚‚‚B geometry. The B3LYP calculation was
extremely slow, oddly, and we abandoned this method.
Calculations on B (Table 4) show that the addition of
THF to A is energetically favorable by 21 or 26 kcal/mol,
which is consistent with the fact that 4 crystallizes as a
THF adduct. The DFT calculation (entry 2) does a slightly
better job at reproducing the experimental geometry of
4‚THF than the HF/6-31G(C,H);6-31+G*(B,N,O) method
(entry 1). The NMR-detected degenerate equilibrium
which 4‚THF undergoes in solution, which leads to time-
averaged top/bottom symmetry, may be written in detail
two ways, viz., eqs 5 and 6. Equation 5 is a dissociative
pathway (loss of THF in the first step), and eq 6 is an
ing”. The interplay between these two types of forces
would be expected to be critical in describing the systems
accurately. Therefore, paring down or eliminating phenyl
rings, for example, was inadvisable, we thought. So,
facing the necessity of tackling the full systems, we
sought to economize as much as possible on the basis set.
HF/3-21G* and B3LYP/3-21G* calculations on 2 gave
B-N bond lengths of 1.823 and 1.775 Å, respectively,
which agreed poorly with the experimental B-N length
of 1.681(5) Å.18a Although a larger basis set was called
for, we felt the carbons and hydrogens did not need as
extensive a basis as borons, nitrogen, and oxygens.
Therefore, we used a 6-31G basis on carbons and hydro-
gens, and 6-31+G* on borons, nitrogen, and oxygens. We
will denote this 6-31G(C,H);6-31+G*(B,N,O). A B3LYP/
6-31G(C,H);6-31+G*(B,N,O) calculation on 2 gave a B-N
bond length of 1.743 Å. In addition, we employed the
LSDA/BP86/DN* method.
Table 3 shows the results of calculations performed on
A. Of the four calculations, only HF/3-21G (entry 1)
predicted a symmetrical B-N-B array. In the other
cases, the symmetrical starting geometry desymmetrized,
but only after many steps of geometry optimization. For
example (entry 3), the starting symmetrical geometry
changed very slightly in each of a very large number of
optimization steps and then suddenly careened lower in
energy by 4.1 kcal/mol to the final unsymmetrical
J . Org. Chem, Vol. 69, No. 20, 2004 6567

Livant et al.
TABLE 4. Ab In itio Ca lcu la tion s on B-F
Erel
(kcal/mol)
B(1)-N
(Å)
B(2)-N
(Å)
entry
species
method
1
2
3
4
5
6
7
8
B
B
HF/6-31G(C,H);6-31+G*(B,N,O)
LSDA/BP86/DN*
-21.25a
-26.21a
1.739
1.717
1.683(4)
2.589
2.599
1.745
2.596
2.261
2.911
2.965
2.959(4)
2.589
2.599
2.908
2.596
2.262
4‚THF
exptlb
C
C
D
E
F
HF/6-31G(C,H);6-31+G*(B,N,O)
LSDA/BP86/DN*
HF/6-31G(C,H);6-31+G*(B,N,O)
HF/6-31G(C,H);6-31+G*(B,N,O)
HF/6-31G(C,H);6-31+G*(B,N,O)
+4.49c
+0.08c
-27.35d
-8.22e
+2.66f
a
b
d
Relative to A + THF. This work. c Relative to B + THF. Relative to A + NH3. e Relative to D + NH3. f Relative to A; species F has
one negative eigenvalue in the force constant matrix.
associative pathway (bonding to THF in the first step).
For clarity, the equations omit the phenyls and most of
the oxygens, i.e., the scaffolding.
) B(2)-N ) 2.2887 Å, phenyl twist 0°) 4.8 kcal/mol above
A, but it was a second-order saddle point.
The calculation of a species such as C is formidable.
To make the job slightly more tractable, we replaced the
THF molecules in B and C with NH3 molecules and did
calculations on D and E. The addition of NH3 to species
A (Table 4, entry 6) was calculated to be favorable by
27.4 kcal/mol, as compared to 21.3 kcal/mol for addition
of THF (entry 1). In contrast to the addition of a second
THF to B, which is unfavorable by 4.5 kcal/mol (entry
4), addition of a second NH3 to D to give E is favorable
by 8.2 kcal/mol. This is consistent with the chemical
ionization mass spectrometry results in which the major
peak corresponds to an adduct of 4 with two NH3
molecules plus H+, i.e., protonated E.
To summarize, 4 in the presence of THF does not
engage in 3c-2e bonding. The 4‚THF adduct is fluxional;
the most likely pathway for degenerate equilibration does
not involve 3c-2e bonding. Calculations suggest that, even
if the complication of THF adduct formation were re-
moved, 4 with a 3c-2e bond (F ) would spontaneously
distort to A.
The distortion is reminiscent of that found by BLS,
whose all-carbon 3c-2e bond (1b) also was calculated to
spontaneously distort.17 Transition structure 1b relaxes
by 14.1 kcal/mol, while transition structure F relaxes by
2.7 kcal/mol. Part of this difference is certainly due to
differences in the basis set and other factors relating to
the method of calculation, and the fact that 1 lacks
phenyls, but some of the difference must reflect the
change from an all-carbon system to a B-N-B system.
Is the latter really more resistant than the former to
distortion of the symmetrical triad? What factors affect
the tendency of a linear X‚‚‚Y‚‚‚X system (a 3c-2e system)
to distort to a linear X-Y‚‚‚‚‚‚‚‚‚X system? Would the
same factors apply to in-line distortions of a 3c-4e
system? Are 3c-4e systems intrinsically more resistant
to in-line distortion than 3c-2e systems?
In eq 5, the first step is the loss of THF from B, which
is uphill in energy by 21 or 26 kcal/mol (Table 4, entries
1 and 2). This gives an estimate of part of the barrier to
equilibration by the dissociative pathway. The Hammond
postulate may be invoked for the highly endothermic
dissociative step to argue that transition state TS1 will
be close in energy to that of A + 2THF. An estimate of
the energy of TS2 comes from the behavior of A during
geometry optimization: the starting geometry (with a
symmetrical B-N-B triad), which is a good approxima-
tion for TS2, was about 4 kcal/mol higher in energy than
the optimized B-N‚‚‚‚‚‚‚‚‚B form. Thus, the dissociative
path is calculated to have an overall barrier of at least
21 or 26 kcal/mol.
In eq 6, C lies 4.5 or 0.1 kcal/mol higher in energy than
B + THF (Table 4, entries 4 and 5). Given the magnitude
of this energy difference, it would be reasonable to assert
that the barrier to equilibration by the associative path
would be much lower than the minimum 21-26 kcal/mol
calculated for the dissociative path. Therefore, the as-
sociative path, which includes no 3c-2e B-N-B bonding,
seems more likely than the dissociative path, which does
include 3c-2e bonding in TS2.
The generation of 4 in the presence of a Lewis basic
solvent such as THF obviously affected the behavior of
4. We have been unsuccessful in attempts to generate 4
in the absence of a Lewis base. Therefore, it was of great
interest to quantify by calculation the energetics of the
boxed part of eq 5 (a “bell clapper” mechanism21 of sorts).
Species F , having a three-center two-electron bond, was
confirmed as a transition structure (one imaginary
frequency) lying 2.66 kcal/mol higher in energy than
species A (eq 7). The B-N distances were 2.2612 and
These questions may be addressed in terms of the
second-order J ahn-Teller (SOJ T) effect.22 Briefly, in this
approach, a driving force for distortion of a symmetrical
molecule exists if the direct product representation ψHOMO
X ψLUMO of the undistorted molecule is the same as that
of the symmetry coordinate which generates the distor-
tion.
2.2616 Å. The “phenyl twist” in F was found to be 8.2°;
that is, F possessed very nearly C3 symmetry. Another
stationary point was found, having D3h symmetry (B(1)-N
The general case of interest here has been treated
previously.23 The symmetries of the orbitals shown in
(22) Nakatsuji, H.; Koga, T. In The Force Concept in Chemistry; Deb,
B. M., Ed.; Van Nostrand Reinhold: New York, 1981; Chapter 3, pp
144-146.
(21) Martin, J . C.; Basalay, R. J . J . Am. Chem. Soc. 1973, 95, 2572-
2578.
6568 J . Org. Chem., Vol. 69, No. 20, 2004

Bonding along a Linear B‚‚‚N‚‚‚B Triad
atom (Figure 4, b). The systems most resistant to in-line
distortion are (i) the two-electron case with termini less
electronegative than the central atom (Figure 4, a) and
(ii) the four-electron case with termini more electroneg-
ative than the central atom (Figure 3, b).
Both 1b and F are 3c-2e systems. In these cases, SOJ T
arguments posit higher resistance to in-line distortion
will result from increasing the difference in electronega-
tivity between the central element and the terminal
element, with the central element more electronegative
than the termini. In 1b there is no central-terminal
electronegativity difference (all-carbon case), while for F
the central nitrogen is more electronegative than the
terminal borons. This should increase the resistance of
F (relative to 1b) to in-line distortion. Comparing our
results (∆E(F f A) ) -2.66 kcal/mol) and those of BLS
(∆E(1b f 1a ) ) -14.1 kcal/mol) yields a “trend” consis-
tent with the SOJ T argument, bearing in mind that the
two results are not rigorously comparable, as mentioned
before.
F IGURE 3. Three-center bond X-Y-X with X more electro-
negative than Y.
Support for the SOJ T approach to the problem of in-
line distortion in three-center bonds may be found in
comparing the X-ray crystal structures of 11,24 12,25 and
13.8b These are 3c-4e cases in which resistance to in-line
F IGURE 4. Three-center bond X-Y-X with X less electro-
negative than Y.
Figure 1 are ψ1, σu+; ψ2, σg+; and ψ3, σu+. The symmetry
coordinate leading to the in-line distortion (X‚‚‚Y‚‚‚X f
+
X-Y‚‚‚‚‚‚‚‚‚X, i.e., the asymmetric stretching mode) is σu
.
In the 3c-4e case, the HOMO-LUMO (ψ2 X ψ3) direct
product is σu+, which is of the same symmetry as the
asymmetric stretching mode. Thus, SOJ T arguments
predict that X-Y-X 3c-4e bonds prefer to be unsym-
metrical. In the 3c-2e case, the HOMO-LUMO direct
distortion should be increased by making the terminal
elements increasingly more electronegative than the
central element boron. In fact, 11 and 12 are “distorted”
as shown, while 13, with the most electronegative
terminal element of the series, is symmetrical (Pauling
electronegativities,26 P, 2.1; N, 3.0; O, 3.5; p orbital
energies27 (eV), P, -9.52; N, -13.83; O, -16.76).
Three-center bonds which look symmetrical “on paper”
often exhibit some distortion in their X-ray structures,
e.g., 1428 and 15.29 Although structures such as these
+
product is once again σu (however, ψ1 X ψ2 instead of
ψ2 X ψ3), so one reaches the same conclusion: X-Y-X
3c-2e bonds prefer to be unsymmetrical. For all sym-
metrical X-Y-X three-center bonds, 3c-2e or 3c-4e, there
is a driving force toward the unsymmetrical triad,
X-Y‚‚‚‚‚‚‚‚‚X.
The magnitude of the driving force is inversely pro-
22
portional to EHOMO - ELUMO
.
The orbital diagram of
Figure 1 applies only if atom X is identical to atom Y. If,
as is commonly the case, the terminal atoms X are more
electronegative than the central atom Y, and therefore
the p orbital of X is lower in energy than the p orbital of
Y, the diagram will look like Figure 3. The energy gap
“a” is the HOMO-LUMO separation in the 3c-2e case,
and the energy gap “b” is the HOMO-LUMO separation
in the 3c-4e case. If, on the other hand, the central atom
is more electronegative (i.e., has a lower energy p orbital)
than the terminal atoms, the orbital diagram will look
like Figure 4, with a and b having the same meanings.
It is apparent on inspection of Figures 3 and 4 that
the smallest HOMO-LUMO gaps, and therefore the
largest driving forces for in-line distortion, are for (i) the
two-electron case with termini more electronegative than
the central atom (Figure 3, a) and (ii) the four-electron
case with termini less electronegative than the central
(24) Yamashita, M.; Watanabe, K.; Yamamoto, Y.; Akiba, K.-y.
Chem. Lett. 2001, 1104-1105.
(25) Yamashita, M.; Kamura, K.; Yamamoto, Y.; Akiba, K.-y.
Chem.sEur. J . 2002, 8, 2976-2979.
(26) Pauling, L. The Chemical Bond; Cornell University: Ithaca,
NY, 1967; p 64.
(27) Mann, J . B.; Meek, T. L.; Allen, L. C. J . Am. Chem. Soc. 2000,
122, 2780-2783.
(28) Paul, I. C.; Martin, J . C.; Perozzi, E. F. J . Am. Chem. Soc. 1972,
94, 5010-5017.
(23) (a) Burdett, J . K. Molecular Shapes. Theoretical Models of
Inorganic Stereochemistry; Wiley: New York, 1980; p 86 ff. (b) Borden,
W. T. Theor. Chim. Acta 1986, 69, 171-174.
(29) Baenziger, N. C.; Buckles, R. E.; Maner, R. J .; Simpson, T. D.
J . Am. Chem. Soc. 1969, 91, 5749-5755.
J . Org. Chem, Vol. 69, No. 20, 2004 6569

Livant et al.
Anal. Calcd for C16H19NO4: C, 66.42; H, 6.62; N, 4.84. Found:
C, 66.33; H, 6.59; N, 4.77.) Additional 2,6-dimethoxyiodoben-
zene (8.800 g, 33.32 mmol) was added and the mixture refluxed
24 h longer. The TLC spot corresponding to 5 (Rf 0.4) was now
the major spot. Silica gel column chromatography (hexanes/
EtOAc, 5:1) gave a yellow solid, which was recrystallized from
CH2Cl2/hexane to give 1.845 g (14.0% yield) of 5: mp 190-1
°C (lit.19 mp 190-1 °C); 15N NMR (41 MHz, acetone-d6) 57.2
(NH3 reference).
have not been discussed in terms of SOJ T distortion, the
effect seems applicable in such cases. A confounding effect
is the possibility of distortions caused by crystal packing
forces on weak 3c-4e bonds.
Con clu sion s
The synthesis of diborate 4 made available for study a
compound having a collinear BNB array and, thereby,
the possibility of a 3-center 2-electron B‚‚‚N‚‚‚B bond. The
fact that the immediate and penultimate precursors of 4
both have planar nitrogens seemed to increase the
likelihood that 4 would also exhibit a planar (i.e., trigonal
bipyramidal) nitrogen, and engage in 3c-2e bonding. The
X-ray crystal structure revealed, however, no evidence
for 3c-2e bonding (see Figure 2). A terminal boron was
bonded externally to the oxygen of THF rather than
internally to nitrogen. In THF solution, the nitrogen of
4 engages in a degenerate bond-switch equilibrium
involving bonding of boron to THF. Calculations sug-
gested that a bond-switch mechanism involving a 3c-2e
B‚‚‚N‚‚‚B transition structure is unlikely. We conclude
that it is unwise to employ a solvent capable of competing
with the central element in bonding to the terminal
atoms of a potential three-center bond. Although we tried
a variety of ways to avoid THF (and any other Lewis basic
solvents), we were unsuccessful in those attempts.
Calculations on “gas-phase” 4 showed that the form
containing a symmetrical 3c-2e B‚‚‚N‚‚‚B array is a
transition structure connected to, and 2.66 kcal/mol
higher in energy than, the unsymmetrical 2c-2e
B-N‚‚‚‚‚‚‚‚‚B form.
Reasoning that it would be helpful for those designing
new three-center (and more extended) systems to have
some qualitative theoretical tool with which to predict
or rationalize the occurrence of symmetrical versus
unsymmetrical arrays, the SOJ T approach to this prob-
lem was presented and applied to a few examples. The
3c-2e arrays (X‚‚‚Y‚‚‚X) most resistant to in-line distortion
are those having central elements of high electronega-
tivity and terminal elements of low electronegativity. The
3c-4e arrays most resistant to in-line distortions are those
having central elements of low electronegativity and
terminal elements of high electronegativity.
Cycliza tion of 1,9-Dim eth oxy-10-(2,6-d im eth oxyp h en -
yl)p h en oxa zin e, 7. A mixture of 50 mg of 7 (0.13 mmol),19
180 mg of copper powder (2.83 mmol), 760 mg of K2CO3 (5.50
mmol), and 36 mg of 18-crown-6 (0.14 mmol) in 3 mL of
o-dichlorobenzene was heated at reflux under a nitrogen
atmosphere for 10 days, at which time TLC (EtOAc/hexane,
1:3 (v/v)) showed two new spots at higher Rf than that of 7.
After silica gel chromatography, two fractions, A and B, were
obtained. Data for fraction A ()9): 3 mg; FABMS m/z 288 (27,
M + 1), 287 (100, M), 273 (22). 1H NMR showed no discernible
peaks, in either CDCl3 or DMSO-d6 solvent. Data for fraction
B ()8): 8 mg; FABMS m/z 334 (31, M + 1), 333 (100, M), 287
1
(23); H NMR (250 MHz, DMSO-d6) δ 6.93 (apparent t, J )
8.2 Hz, 2H), 6.89 (dd, J ) 8.9, 7.5 Hz, 1H), 6.75 (dd, J ) 8.4,
1.3 Hz, 2H), 6.68 (d, J ) 8.8 Hz, 1H), 6.68 (d, J ) 7.6 Hz, 1 H),
6.60 (dd, J ) 8.1, 1.3 Hz, 2H), 3.59 (s, 6H); 13C NMR (63 MHz,
DMSO-d6) δ 150.7 (quat), 148.4 (quat), 146.3 (quat), 124.2
(CH), 124.0 (quat), 123.1 (CH), 120.1 (quat), 111.3 (CH), 108.9
(CH), 107.8 (CH), 55.7 (CH3).
Tr is(2,6-d ih yd r oxyp h en yl)a m in e, 6. To a stirred solution
of 400 mg (0.944 mmol) of 5 in 30 mL of toluene, under rapid
nitrogen purge, was added quickly 765 mg (5.74 mmol) of
aluminum trichloride via one neck of the three-necked flask.
The reaction mixture was quickly brought to reflux using an
oil bath. The oil bath was turned off, and the flask and the oil
bath were cooled together for 1.5 h. To the resulting green
mixture was added 5 mL water, and stirring was continued
for 15 min, at which time the mixture had become purple and
deposited a lavender precipitate. This solid was collected by
filtration, affording 300 mg (94% yield) of a material which
was pure by 1H and 13C NMR spectroscopy, albeit lavender. It
was dissolved in 50 mL of boiling EtOAc, cooled, and extracted
with 3 × 20 mL of 0.95 M NaOH. The basic extract was
brought to pH 6 with 3 M HCl. The precipitate which resulted
was taken up in 3 × 20 mL of EtOAc, and dried over Na2SO4.
After decantation from the drying agent, the volume was
reduced by half and stored overnight in the freezer, affording
155 mg (48% yield) of light pink crystals: mp 308-9 °C
1
(discolors at 275 °C); H NMR (250 MHz, THF-d8) δ 8.12, (s,
6H), 6.69 (t, J ) 8.1 Hz, 3H), 6.22 (d, J ) 8.1 Hz, 6H); 13C
NMR (63 MHz, THF-d8) δ 157.3, 125.9, 123.5, 108.2; EIMS
m/z 341 (M+), 244, 215, 198; IR (KBr) 3271, 2322, 1661, 1590,
1511, 1464, 1330, 1234, 1220, 1011, 791, 733 cm-1
.
Exp er im en ta l Section
Tr is(2,6-d im eth oxyp h en yl)a m in e, 5.19 Under a nitrogen
atmosphere, a mixture of 4.760 g (31.07 mmol) of freshly
distilled 2,6-dimethoxyaniline (see the Supporting Informa-
tion), 13.20 g (49.98 mmol) of 2,6-dimethoxyiodobenzene (see
the Supporting Information), 36.40 g (263.4 mmol) of an-
hydrous powdered K2CO3, 8.474 g (133.4 mmol) of electrolytic
copper, 1.842 g (6.969 mmol) of anhydrous 18-crown-6, and
60 mL of o-dichlorobenzene was brought to reflux. The progress
of the reaction was monitored by silica gel TLC (5:1 hexanes/
EtOAc). 2,6-Dimethoxyaniline and a reaction intermediate,
bis(2,6-dimethoxyphenyl)amine, have nearly identical Rf val-
ues (0.7) but may be distinguished by the color generated by
exposing the plate to iodine vapors. After 24 h, 2,6-dimethoxy-
iodobenzene and 2,6-dimethoxyaniline had both been con-
sumed. The major species present was bis(2,6-dimethoxy-
phenyl)amine. (If desired, this product may be collected by
silica gel column chromatography (hexanes/EtOAc, 5:1): mp
122-124 °C; 1H NMR (250 MHz, CDCl3) δ 6.75, (t, J ) 8.1
Hz, 1H), 6.45 (d, J ) 8.1 Hz, 2H), 5.47 (s, 1H), 3.61 (s, 12H);
13C NMR (63 MHz, CDCl3) δ 151.5, 123.4, 119.9, 104.7, 57.0.
A crystal, 0.44 × 0.44 × 0.06 mm, was chosen for X-ray
crystallography, λ ) 0.71073 Å: monoclinic, a ) 14.092(5) Å,
b ) 8.590(2) Å, c ) 21.856(6) Å, â ) 94.64(2)°, Z ) 4, space
group C2/c. A total of 2433 reflections were collected (0 e h e
+16, 0 e k e +10, -26 e l e +25; 2.78° e θ e 25.05°), 2332
of which were independent reflections (Rint ) 0.0213). Solution
and refinement by full-matrix least squares on F2, data-to-
parameter ratio 13.2, gave a goodness of fit of 1.028, R1 )
0.0658 and wR2 ) 0.1486 (I > 2σ(I)), and R1 ) 0.1205 and
wR2 ) 0.1917 (all data). Crystalline 6 includes two molecules
of ethyl acetate per molecule of 6.
Tr is(2,6-d ih yd r oxyp h en yl)a m in e Dibor a te, 4. To a solu-
tion of 10 mg (0.029 mmol) of 6 in 0.5 mL of THF-d8 in a 5
mm NMR tube under nitrogen was added by syringe 0.0058
mL (0.058 mmol) of 1.0 M BH3‚THF. Gas was evolved over a
period of 20 min, with warming to 36 °C in a water bath during
the last 5 min. The reaction appeared quantitative by 1H
NMR: 1H NMR (400 MHz, THF-d8) δ 6.69 (t, J ) 8.1 Hz, 3H),
6.22 (d, J ) 8.1 Hz, 6H); 13C NMR spectral data given in Table
1; 11B NMR (128 MHz, THF) δ 17.9. A sample of 4 prepared
6570 J . Org. Chem., Vol. 69, No. 20, 2004

Bonding along a Linear B‚‚‚N‚‚‚B Triad
as described spontaneously deposited crystals, which proved
suitable for X-ray crystallography.
thanks are due to Dr. Dave Young for helpful discus-
sions about the calculations and Prof. Robert A. Don-
nelly for unique insights with regard to the SOJ T
approach. The support of the Auburn University Chem-
istry Department is gratefully acknowledged.
A crystal, 0.60 × 0.32 × 0.10 mm, was chosen for X-ray
crytstallography, λ ) 0.71073 Å: monoclinic, a ) 10.108(5) Å,
b ) 14.080(6) Å, c ) 13.864(4) Å, â ) 92.89(4)°, Z ) 4, space
group P21/n. A total of 3700 reflections were collected (-5 e h
e +12, -10 e k e +16, -16 e l e +16; 2.06° e θ e 25.05°),
3491 of which were independent reflections (Rint ) 0.0156).
Solution and refinement by full-matrix least squares on F2,
data-to-parameter ratio 12.1, gave a goodness of fit of 1.016,
R1 ) 0.0507 and wR2 ) 0.1164 (I > 2σ(I)), and R1 ) 0.0816
and wR2 ) 0.1367 (all data). Crystalline 4 includes one
molecule of THF-d8 per molecule of 4.
Su p p or tin g In for m a tion Ava ila ble: Experimental pro-
cedures for 2,6-dimethoxybenzamide, 2,6-dimethoxyaniline,
and 2,6-dimethoxyiodobenzene, detailed results of calculations,
spectroscopic data for 4-6, 8, and 9, and X-ray crystallographic
data, including ORTEP plots for 4 and 6 (PDF, CIF). This
material is available free of charge via the Internet at
http://pubs.acs.org.
Ack n ow led gm en t. We thank the Alabama Super-
computer Center for generous allocations of time. Many
J O040199O
J . Org. Chem, Vol. 69, No. 20, 2004 6571