Synthesis and reactivity of [c/oso-1-CB9H9-1-N 2]: Functional group interconversion at the carbon vertex of the {c/oso-1-CB9} Cluster

Article abstract of DOI:10.1021/ic9007476

The dinitrogen derivative [c/oso-1-CB₉H₉-1-N2] (1) was prepared from amine [c/oso-1-CB₉H₉-1-NH₃] (2) and reacted with three types of nucleophiles: activated arenes (phenolate and aniline), divalent su

Full text of DOI:10.1021/ic9007476

Inorg. Chem. 2009, 48, 7313–7329 7313
DOI: 10.1021/ic9007476
Synthesis and Reactivity of [closo-1-CB9H9-1-N2]: Functional Group
Interconversion at the Carbon Vertex of the {closo-1-CB9} Cluster
Bryan Ringstrand, Piotr Kaszynski,* and Andreas Franken
Organic Materials Research Group Department of Chemistry, Vanderbilt University, Nashville,
Tennessee 37235
Received April 17, 2009
The dinitrogen derivative [closo-1-CB9H9-1-N2] (1) was prepared from amine [closo-1-CB9H9-1-NH3] (2) and reacted
with three types of nucleophiles: activated arenes (phenolate and aniline), divalent sulfur compounds (Me2S and
Me2NCHS), and pyridine, giving products of substitution at Ccage. The reaction of 1 with pyridine gave all four isomers
11a-11d, indicating the Gomberg-Bachmann mechanism, which involves radical anion [closo-1-CB9H9]-• (26). The
radical and also closed-shell electrophilic aromatic substitution mechanisms were probed with the aid of DFT and MP2
computational methods and compared to those of phenylation of pyridine. Overall, experimental results supported by
computational analysis suggest two mechanisms for the substitution of the N2 group in 1: (i) thermal heterolytic
cleavage of the Ccage-N bond and the formation of electrophilic carbonium ylide [closo-1-CB9H9] (19) and (ii) electron-
transfer-induced homolytic cleavage of the Ccage-N bond and the formation of 26. Decomposition of 1 in MeCN is
believed to proceed by the nonradical mechanism involving formation of the ylide 19 as the rate-determining step with
experimental activation parameters 4H‡ =38.4 ( 0.8 kcal mol-1 and 4S‡ =44.5(2.5 cal mol-1 K-1. The electron-
transfer-induced formation of 26 is consistent with the relatively high reduction potential of 1 (Epc = -0.54 V), which is
more cathodic than that of PhN2þ by 0.38 V. Transformations of the phenol 8a and the Me2NCHS adduct 10 were
demonstrated by O-methylation of the former and hydrolysis of 10 followed by S-alkylative cyclization. Direct products
and their derivatives were investigated by UV-vis spectroscopy and analyzed with the ZINDO computational method.
Introduction
general for the apical position of 10-vertex closo-boranes.11
Indeed, scant indirect evidence suggests that [closo-1-CB9H9-
1-N2] (1), formed by diazotization of amine [closo-1-CB9H9-
1-NH3] (2), is also reasonably stable and was implied as an
intermediate in a reaction with Me2S.12 Therefore, in analogy
to the dinitrogen derivativesof [closo-B10H10]2-, compound 1
and its derivatives have the potential of becoming syntheti-
cally valuable intermediates in the introduction of substituents
at the Ccage position of the {closo-1-CB9} cage.
Dinitrogen derivatives of the [closo-B10H10]2- cluster are
isolable stable1-6 and important intermediates in the prepa-
ration of a variety of nitrogen,1-4,6-8 sulfur,4,6,8,9 oxygen,4,10
and even carbon4 derivatives. Our calculations showed that
the moderate stability and hence synthetic usefulness of these
intermediates originate from the electronic interaction be-
tween the cluster and the N2 group, which appears to be
Our interest7,8,13-15 in highly polar and ionic molecular
materials, including liquid crystals, focused our attention on
dinitrogen derivatives I, 10-substituted derivatives of 1, as poten-
tial intermediates in the synthesis of zwitterionic and anionic
azo compounds of the general structures II and III, respectively
(Figure 1). In this context, we recently developed15,16 synthetic
access to isomerically pure [closo-1-CB9H8-1-NH2-10-I]- as the
*To whom correspondence should be addressed. Phone: (615) 322-3458.
Fax: (615) 343-1234. E-mail: piotr.kaszynski@vanderbilt.edu.
(1) Hertler, W. R.; Knoth, W. H.; Muetterties, E. L. J. Am. Chem. Soc.
1964, 86, 5434–5439.
(2) Hertler, W. R.; Knoth, W. H.; Muetterties, E. L. Inorg. Chem. 1965, 4,
288–293.
(3) Knoth, W. H.; Hertler, W. R.; Muetterties, E. L. Inorg. Chem. 1965, 4,
280–287.
(4) Knoth, W. H. J. Am. Chem. Soc. 1966, 88, 935–939.
(5) Leyden, R. N.; Hawthorne, M. F. Inorg. Chem. 1975, 14, 2444–2446.
(6) Balinski, A.; Januszko, A.; Harvey, J. E.; Brady, E.; Kaszynski, P.;
Young, V. G., Jr. Manuscript in preparation.
(12) Jelinek, T.; Stibr, B.; Plesek, J.; Thornton-Pett, M.; Kennedy, J. D.
J. Chem. Soc., Dalton Trans. 1997, 4231–4236.
(7) Kaszynski, P.; Huang, J.; Jenkins, G. S.; Bairamov, K. A.; Lipiak, D.
(13) Kaszynski, P.; Lipiak, D. In Materials for Optical Limiting; Crane, R.,
Lewis, K., Stryland, E. V., Khoshnevisan, M., Eds.; MRS: Boston, 1995; Vol. 374,
pp 341-347.
Mol. Cryst. Liq. Cryst. 1995, 260, 315–331.
(8) Kaszynski, P.; Douglass, A. G. J. Organomet. Chem. 1999, 581, 28–38.
(9) Komura, M.; Nakai, H.; Shiro, M. J. Chem. Soc., Dalton Trans. 1987,
1953–1956.
(14) Fendrich, W.; Harvey, J. E.; Kaszynski, P. Inorg. Chem. 1999, 38,
408–410.
(10) Bragin, V. I.; Sivaev, I. B.; Bregadze, V. I.; Votinova, N. A.
J. Organomet. Chem. 2005, 690, 2847–2849.
(11) Kaszynski, P.; Pakhomov, S.; Young, V. G., Jr. Collect. Czech. Chem.
Commun. 2002, 67, 1061–1083.
(15) Ringstrand, B.; Kaszynski, P. J. Mater. Chem. 2009, 19, 4805–4812.
(16) Ringstrand, B.; Balinski, A.; Franken, A.; Kaszynski, P. Inorg.
Chem. 2005, 44, 9561–9566.
r
2009 American Chemical Society
Published on Web 07/08/2009
pubs.acs.org/IC

7314 Inorganic Chemistry, Vol. 48, No. 15, 2009
Ringstrand et al.
Scheme 2
Figure 1. Generalstructuresfor the polarand ionicderivativesII and III
and their precursor I. Qþ represents an onium fragment such as quinu-
clidinium, pyridinium, or sulfonium.
Scheme 1
solutions in MeCN are stable for 24 h at ambient tem-
perature in the dark with little (about 10%) decomposi-
tion (vide infra). In contrast, dinitrogen 1 completely
decomposes in CH2Cl2 and benzene solutions into a
mostly intractable mixture of presumably polymeric
and degradation products after 24 h in the dark at
ambient temperatures. NMR spectra of 1 display a sig-
nificant solvent dependence. Thus, the signals of 11B(10)
key starting material for the preparation of II and III. In this
precursor, the amino group is expected to serve as a synthetic
handle for the introduction of a sulfonium fragment, a pyridin-1-
yl group, and an azo group through the dinitrogen derivative I.
Therefore, we desired to gain an understanding of the properties
and reactivity of 1 as a model for I and a step toward the
development of synthetic access to II and III.
Here, we report a reliable method for the preparation of
amine 2, which upon diazotization gives [closo-1-CB9H9-1-
N2] (1). Then, we describe the thermal stability and reactivity
of dinitrogen derivative 1 with several nucleophilic reagents.
In particular, we explored reactions that lead to derivatives
with the protected mercapto functionality, diazocoupling to
aromatic compounds, and reactions with pyridine. The
mechanism of the reaction of 1 with pyridine was investigated
experimentally and analyzed with the aid of DFT and MP2
computational methods. The dinitrogen derivative 1 and
several products derived from it were investigated with
UV-vis spectroscopic methods, and the experimental results
were compared to the ZINDO computational data. Finally,
the transmission of the Ccage substituent electronic effects
through the {closo-1-CB9} cage was briefly investigated and
compared to that of {closo-1-CB11} and the benzene ring.
1
and H(10) nuclei are shifted by about 6 and 1.2 ppm,
respectively, downfield in benzene-d6 relative to those
measured in CD3CN solutions.
Although amine 2 has been reported in the literature,12
its previous preparation was complicated by the forma-
tion of the closely related [closo-1-CB11H11-1-NH3] as a
byproduct. Also, this method does not allow for intro-
duction of a substituent at the B(10) position. Therefore,
we decided to develop reliable access to 2 starting from the
readily accessible carboxylic acid 3.16 Our initial attempts
at direct conversion of carboxylic acid 3 to amine 2 using
several methods such as the classical Schmidt17 and
Curtius reactions,18 azidotrimethylsilane,19 azidotri-
methylsilane and NaN3,20 diphenylphosphoryl azide,21
and hydroxylamine sulfonic acid22 were unsatisfactory
and plagued by poor yields, irreproducibility, low reac-
tivity with acid 3[NEt4], and numerous unidentified
side products, as evident from 11B NMR. Eventually,
we focused on the Curtius reaction and carbamate 4 as the
isolable intermediate, which could be purified and serve
as a convenient storage for amine 2. Conditions for each
synthetic step were optimized, and the reactions were
followed by monitoring the chemical shifts of the B(10)
nucleus in 11B NMR spectra and the IR vibrations of
the carbonyl group (see the Supporting Information).
Thus, the required carbamate 4a[NEt4] was prepared in
four steps and about 50% overall yield based on car-
boxylic acid 3[NEt4] using a modified Curtius rearrange-
ment (Scheme 2).
Results
Synthesis of [closo-1-CB9H9-1-N2] (1). The preparation
of [closo-1-CB9H9-1-N2] (1) was accomplished by diazo-
tization of [closo-1-CB9H9-1-NH3] (2) in a 50% aqueous
AcOH with 1.1 equiv of NaNO2 (Scheme 1). The dinitro-
gen derivative 1 was formed as a white precipitate and
conveniently isolated by filtration in yields>80%. Dia-
zotization of contaminated amine 2, for example, with
carboxylic acid 3, also was satisfactory, albeit with lower
yields of 1, since the ionic impurity tends to stay in
solution. The ratio and amounts of AcOH and H2O were
optimized to maximize the material recovery.
The dinitrogen derivative 1 is a white microcrystalline
substance easily soluble in common organic solvents,
except for alkanes. It is stable in the solid state at 0 °C
for at least 18 months without traces of decomposition.
When heated, solid 1 rapidly decomposes at 87 °C. Its
The first step of the preparation of carbamate 4a[NEt4]
was the typically quantitative transformation of carboxylic
(17) Banthorpe, D. V. In The Chemistry of the Azido Group; Patai, Ed.;
Wiley: New York, 1971; p 405.
(18) Banthorpe, D. V. In The Chemistry of the Azido Group; Patai, Ed.;
Wiley: New York, 1971; p 397.
(19) Krow, G.; Damodaran, K. M.; Michener, E.; Miller, S. I.; Dalton,
D. R. Synth. Commun. 1976, 6, 261–267.
(20) Warren, J. D.; Press, J. B. Synth. Commun. 1980, 10, 107–110.
(21) Shioiri, T.; Ninomiya, K.; Yamada, S. J. Am. Chem. Soc. 1972, 94,
6203–6205.
(22) Wallace, R. G.; Barker, J. M.; Wood, M. L. Synthesis 1990, 12, 1143–
1144.

Article
Inorganic Chemistry, Vol. 48, No. 15, 2009 7315
acid 3[NEt4] to carbonyl chloride 5[NEt4] using (COCl)2.
The crude chloride 5[NEt4] was then converted to acyl
azide 6[NEt4] using Me3SiN3 in the presence of ZnCl2
according to a general literature procedure.23 11B NMR
spectra typically showed 85% conversion to the azide
6[NEt4], with the remaining chloride 5[NEt4] being hy-
drolyzed back to acid 3[NEt4]. A minor peak at 25.1 ppm
in 11B NMR spectra of the crude reaction mixture was also
present, which was later attributed to isocyanate 7[NEt4].
The premature rearrangement of azide 6[NEt4] to 7[NEt4]
was most likely promoted by the Lewis acid catalyst rather
than thermal induction.24
Scheme 3
Attempts at improving this process through manipula-
tion with reaction times, catalyst loading, and thorough
drying of the catalyst were unsuccessful. The IR spectrum
of azide 6[NEt4] revealed vibrations characteristic for the
azido and carbonyl groups and is similar to that reported
for benzoyl azide.25
Crude azide 6[NEt4] was rearranged to isocyanate
7[NEt4] by refluxing in anhydrous CH3CN. The comple-
teness of the transformation was evident from 11B NMR
spectra in which the B(10) chemical shift of the former at
36.4 ppm was replaced with a peak at 25.1 ppm in the
isocyanate. The product also showed an intense peak at
2262 cm-1 in the IR spectrum, which is consistent with
the νNCO vibration reported for phenyl isocyanate in
the solid state (2255 cm-1).26 On the basis of the 11B
NMR spectrum, the purity of the crude isocyanate 7
[NEt4] was estimated as about 90%.
Scheme 4
Initially, we envisioned conversion of isocyante 7[NEt4]
directly to amine 2, but this approach was not fruitful.
Isocyanate 7[NEt4] was insoluble in aqueous HCl even at
elevated temperatures. Attempted hydrolysis of 7[NEt4]
in a 1:1 mixture of 18% HCl and CH3OH at 80 °C gave a
mixture of amine 2 and methyl carbamate 4b[H3O] upon
extraction from acidified aqueous solutions, according to
11B NMR analysis. The carbamate 4b[H3O] was cleanly
converted to amine 2 under basic conditions, which
prompted us to investigate the carbamates as possible
intermediates that could easily be purified and converted
to amine 2 in high yields. However, attempts to sepa-
rate the methyl carbamate 4b[NEt4] from carboxylic
acid 3[NEt4] by chromatography proved difficult due
to its low solubility and also an insufficient difference
in mobility on silica gel. Therefore, we focused on t-Bu
carbamate 4a[NEt4], which could be converted to the
amine under mildly acidic conditions.
A reaction of isocyanate 7[NEt4] with anhydrous
t-butanol gave carbamate 4a[NEt4] in a high yield. Initial
attempts at chromatographic separation on a SiO2 col-
umn using CH3CN/CH2Cl2 (1:4) as the eluent showed
that pure carbamate 4a[NEt4] could be isolated in low
yields (∼20%). Mass balance was achieved with more
polar fractions, which contained mostly amine 2, appar-
ently arising from deprotection of the BOC group on the
SiO2 support. The decomposition of 4a[NEt4] was mini-
mized by buffering the solvent system with 1% NEt3. The
yield of carbamate 4a[NEt4] rose to about 50%, but
deprotection on the column was still observed. Treatment
of carbamate 4a[NEt4] with HCl in CH3OH solution
followed by evaporation of the volatiles gave pure 2 in
quantitative yield.
Reactivity of [closo-1-CB9H9-1-N2]. Dinitrogen deriva-
tive 1 was reacted with three types of nucleophiles:
activated arenes, divalent sulfur compounds, and pyri-
dine. The reagents were chosen to demonstrate the for-
mation of functionalizable azo derivatives 8 and the
formation of inner sulfonium salts directly (9a) or
through a masked mercaptan (10), and to investigate
the formation of pyridin-1-yl derivative 11a (Scheme 3).
A reaction of 1 with stoichiometric amounts of tetra-
ethylammonium phenolate showed that both mono (8a)
and double (12) diazocoupling products were formed in
a ratio of up to 5:1 (Scheme 4). Attempts to separate
8a[NEt4] from 12[2NEt4] were unsuccessful, both by
recrystallization (EtOH/H2O mixtures) and chromatog-
raphy (CH3CN/CH2Cl2 mixtures). The formation of the
double substitution product 12 was minimized by using a
5-fold excess of the phenolate. In contrast, no double
diazocoupling was observed in reactions of 1 with stoi-
chiometric amounts of aniline.
Purification of azo derivatives 8a and 8b was accom-
plished by chromatographic separation of the acid extract
followed by precipitation as the NMe4þ salt, which gave
reasonably pure products in yields of 50% and 80%,
respectively. The use of NMe4þ as the counterion gave
more crystalline products compared to those with NEt4þ.
(23) Prakash, G. K. S.; Iyer, P. S.; Arvanaghi, M.; Olah, G. A. J. Org.
Chem. 1983, 48, 3358–3359.
(24) Rawal, V. H.; Zhong, H. M. Tetrahedron Lett. 1994, 35, 4947–4950.
(25) Laszlo, P.; Polla, E. Tetrahedron Lett. 1984, 25, 3701–3704.
(26) Paraskewas, S. M.; Danopoulos, A. A. Synthesis 1983, 638–340.

7316 Inorganic Chemistry, Vol. 48, No. 15, 2009
Ringstrand et al.
Scheme 5
Scheme 6
The azophenol 8a[NMe4] was easily methylated under
basic conditions to give the methoxy derivative 13[NMe4]
in 95% yield (Scheme 5).
Dinitrogen derivative 1 was completely reacted with
neat Me2S at ambient temperature over a period of 24 h,
giving adduct 9a, which was isolated in 40% yield. 11B
NMR analysis of the crude reaction mixture revealed four
signals: three doublets of product 9a and an unrelated
singlet at 20.1 ppm. The singlet appears to be associated
with the film of white material, presumably boric acid,
which was poorly soluble in CD3CN and well-soluble in
CD3OD (18.8 ppm). Previously, 9a was prepared in 68%
yield based on amine 2 without isolation of the inter-
mediate 1.12
A 24 h reaction of 1 with N,N-dimethylthioformamide
at ambient temperature gave 80% yield of crude 10
after washing with toluene. 11B NMR analysis of the
crude product revealed the presence of about 10% of the
parent anion [closo-1-CB9H10]- (14), and also a low
intensity singlet at 20.1 ppm. Attempts to purify the
zwitterion 10 further were unsuccessful due to its partial
decomposition observed during recrystallization from
CH3CN/toluene mixtures or 1,2-dichloroethane. Purifi-
cation on silica gel was avoided, since 10 appeared to be
hydrolytically unstable. Alkylation of crude 10 with 1,5-
dibromopentane in the presence of a base followed by
chromatographic separation and recrystallization gave
pure 9b in 50% yield based on 1 (Scheme 6). This
represents a slightly higher yield than that for a single-
step preparation of 9a directly from 1.
Reactions of 1 with pyridine were most interesting.
When 1 was reacted in neat pyridine at ambient tempera-
ture for several hours, 11B NMR analysis revealed a com-
plex mixture of at least four products exhibiting charac-
teristic doublets in the region of 25-40 ppm (Table 1).
Surprisingly, the major component of the reaction mix-
ture, about 50%, was identified as the 2-isomer 11b[H].
A doublet at δ 29.6 ppm with 20% intensity was assigned
to the parent anion 14 on the basis of a comparison with
literature data.27 This was further supported by a com-
parison with an authentic sample of 14 with pyridinium as
the counterion (14[PyrH]) to maintain consistency with
the reaction conditions.28
mixture of two other isomers, 11c[H] and 11d[H], which
were identified on the basis of 1D H and 2D H-1H
correlation spectroscopy (COSY) NMR analysis.28
The addition of hot pyridine to the dinitrogen derivative
1 had a different outcome. 11B NMR analysis of the crude
reaction mixture revealed four products [δ B(10): 35.0,
33.1, 31.9, 29.6]. Chromatographic separation gave 11a in
17% isolated yield as the least polar fraction. The more
polar fraction contained the remaining three isomers and
also the parent anion 14 in about a 4:4:2:1 ratio (Table 1).
To provide more information about the possible me-
chanism for the formation of these products, reactions of
1 were run in dilute solutions of pyridine at ambient
temperature for 18 h. Thus, a reaction of 1 with a 4%
solution of pyridine in CH2Cl2 gave [closo-1-CB9H9-1-
Cl]- (15)29 as the major product along with 11a, 11b, and
14 in an approximate ratio of 8:3:1:3, respectively, ac-
cording to the 11B NMR analysis of the crude reaction
mixture (Table 1). A similar reaction of 1 with a 4%
solution of pyridine in benzene gave [closo-1-CB9H9-1-
Ph]- (16)30 as the major product. Isomers 11a and 11b
were formed as minor products, in 13% and 17% yields,
respectively, and no signals belonging to 11c or 11d were
found in the 11B NMR spectrum of the crude mixture
(Table 1). For comparison, decomposition of 1 in pure
benzene-d6 at ambient temperature gave a number of
products, among which 16-d5 was identified by MS. A
solution of 1 in CD2Cl2 gave only a broad featureless
absorption band in a range -10 to -40 ppm and a broad
band around 20 ppm in 11B NMR after 24 h at ambient
temperature, which suggests complete destruction of the
{closo-1-CB9} cluster. Dinitrogen derivative 1 in a 4%
solution of pyridine in CD3CN was stable at 0 °C for 3 h,
whereas at 45 °C, 45% of 1 reacted in 75 min, giving about
30% 17 (Scheme 7) and 4% 11a in addition to unidenti-
fied signals at 16.7, 25.3, and 10.0 ppm (in decreasing
intensity order) in the 1H-decoupled spectrum.
1
1
The crude mixture was separated on silica gel giving the
2-isomer 11b[H] in 40% isolated yield as the first fraction.
The doublet at δ 35.7 ppm observed in the crude mixture
and originally assigned to the B(10) nucleus of 11b was
shifted downfield by about 2 ppm in the isolated sample.
This shift presumably results from transformation of
the pyridinium salt of 11b[-] to the zwitterionic structure
11b[H] of the isolated product. Such downfield shifts were
observed for the other two isomers 11c and 11d, but they
were less significant. The more polar fraction contained a
Reaction with MeCN and Kinetic Data for 1. Decom-
position of dinitrogen derivative 1 in dried CH3CN at
elevated temperatures (∼50 °C) gave a mixture of zwit-
terion 17 and acetamide 18 in about a 1:1 ratio as the only
1
products (Scheme 7). H NMR revealed two partially
overlapping quartets centered at 5.91 and 5.52 ppm
attributed to 17 and 18, respectively. Adduct 17 appears
to be the primary product, which partially undergoes
hydration to the acetamide 18 with residual H2O in the
(29) Ringstrand, B.; Bateman, D.; Shoemaker, R. K.; Janousek, Z.
Collect. Czech. Chem. Commun. 2009, 74, 419–431.
(30) Jelinek, T.; Kilner, C. A.; Thornton-Pett, M.; Kennedy, J. D. Chem.
Commun. 2001, 1790–1791.
(27) Franken, A.; Kilner, C. A.; Thornton-Pett, M.; Kennedy, J. D. Inorg.
Chem. Commun. 2002, 5, 581–584.
(28) For details, see the Supporting Information.

Article
Inorganic Chemistry, Vol. 48, No. 15, 2009 7317
Table 1. Product Distributions in Reaction of 1 with Pyridine
condition\products
11a
11b[H]
11c[H]
11d[H]
14
Neat Pyridine
25 °Ca
90 °Cc
traces
17%b
40%b
17%c
10%c
6%c
20%c
18%c
9%c
5%c
Solutions of Pyridine
17%e
7%e
0%e
0%e
20%e
d
4% in CH2Cl2
4% in benzenef
13%e
33.1 ppm
17%e
37.2 ppm
0%e
33.9 ppm
0%e
35.9 ppm
5%e
29.6 ppm
11B NMR for B(10), δg
a About 75% of total mass recovery after separation on SiO2. b Yield of isolated product. c Yield based on 11B NMR of the mixture corrected for its
mass. d [closo-1-CB9H9-1-Cl]- (15) constituted 56% of the mixture. e Ratio of low field signals for the crude reaction mixture. f [closo-1-CB9H9-1-Ph] (16)
constituted 65% of the crude reaction mixture. g Chemical shifts of the isolated products or mixtures.
Scheme 7
Figure 2. The MP2/6-31G(d,p) derived contour for the LUMO (E=
-1.57 eV) of ylide 19.
solvent. Exposure of the crude reaction mixture to moisture
quickly and quantitatively converted 17 to the acetamide
18, which was isolated and characterized by NMR and
MS techniques. The clean transformation of 1 to the two
related products and sufficient separation of the signals in
the 1H NMR spectrum made it possible to analyze
kinetics of this reaction. It is believed that the rate-
determining step of the process is the heterolytic cleavage
of the C-N bond in 1.
The kinetics of decomposition of the dinitrogen deri-
vative 1 were investigated in dried CD3CN as a function
of the temperature and were followed by 1H NMR
spectroscopy. Intensities of the disappearing B(10)-H
signal at 7.21 ppm, which is part of a quartet centered at
6.72 ppm belonging to 1, and the growing peak at 5.82 ppm,
part of the pseudo quintet (partially overlapping quartets of
17-d3 and 18-d3), were used to calculate the ratio of 1 to the
sum of products 17-d3 and 18-d3. Standard kinetic analysis
using the Erying equation found 4H‡=38.4 ( 0.8 kcal
mol-1, 4S‡=44.5 ( 2.5 cal mol-1 K-1, and 4G‡=25.1 (
0.8 kcal mol-1 at 298 K.28
Mechanistic Studies. Computational Analysis. To
shed more light on properties of 1 and its reactions with
nucleophiles, we conducted quantum-mechanical calcu-
lations initially at the B3LYP/6-31G(d,p) level of theory
to establish conformational minima and to obtain ther-
modynamic corrections. The resulting structures were
used to calculate the self-consistent field (SCF) energies
at the MP2/6-31G(d,p) level of theory. Since 1 and other
molecules involved are highly polar or ionic, the SCRF
solvation model was used to compute their SCF energies
at the MP2/6-31G(d,p) or MP2/6-31þG(d,p) level of
theory in an appropriate dielectric medium (pyridine or
MeCN). Computational results are shown in Figures 2-6
and Tables 2-7, and full numerical data are provided in
the Supporting Information.28
Generation and Structure of the Carbonium Ylide 19.
The dinitrogen derivative 1 undergoes a heterolytic clea-
vage of the Ccage-N bond, and the process is calculated to
be endothermic by 33.2 kcal/mol in the gas phase at the
MP2//DFT level of theory with the 6-31G(d,p) basis set.
This value is practically the same as that obtained using
the MP2//MP2 method with the same basis set (Table 2).
Since the resulting carbonium ylide 19 (μ=4.0 D) is less
polar than 1 (μ=6.1 D), the endotherm increases to
35.4 kcal/mol in pyridine (ε=13.3) and to 36.4 kcal/mol
in the MeCN (ε = 36.6) dielectric medium. The MP2//
DFT and MP2//MP2 calculations using diffuse functions
do not change the results significantly, although they give
lower ΔH values up to 2 kcal/mol for the latter method. In
contrast to the MP2 results, the enthalpy change in the
reaction calculated by the DFT method was smaller by
about 10 kcal/mol. This difference is related to the
inadequate treatment of electron correlation by the
DFT methods and demonstrates the importance of dis-
persive forces in the stabilization of 1.
A transition state search with the DFT method for the
loss of N2 by 1 located a transition structure, 1-TS, in

7318 Inorganic Chemistry, Vol. 48, No. 15, 2009
Ringstrand et al.
Figure 3. Electrophilic substitution of pyridine. Energy change relative to ylide 19 and pyridine (MP2/6-31G(d,p)//B3LYP/6-31G(d,p), ε=13.3). For the
reaction with radical ion 26, the intermediates 28 are analogous to 21 with the delocalized unpaired spin instead of the positive charge.
highly reactive electrophile. In the simplest case, it rea-
cts with Lewis bases in a two-electron process, giving
C-substitution products such as 9, 11a, and 20. The overall
two-step nucleophilic substitution reactions (SN1) of 1
are highly exothermic, with the enthalpy change exceeding
-30 kcal/mol and typically about -60 kcal/mol in a vacuum
(Table 4). Analysis of the results for the four isomeric
pyridine adducts 11a[H]-11d[H] demonstrates that the
most thermodynamically stable isomer is 11b[H], while the
Figure 4. Structures of the edge adducts of 19 to pyridine.
˚
which the Ccage-N bond is elongated by about 0.8 A (dC-N
=
least stable is the N-isomer 11a (11b[H]>11d[H]>11c[H]>
11a) in a vacuum. The compounds have substantial dipole
moments, up to 3 times that for the starting dinitrogen
derivative 1 (6.1 D), and a significant solvent effect on the
reaction’s exotherm can be expected in polar media. Indeed,
isodensity polarizable continuum model (IPCM) calcula-
tions in the dielectric medium increase the stability of all
isomers, but the largest gain is observed for the C(3)-isomer
11c[H]. Consequently, theΔHfor the formation of 11c[H] in
pyridine solutions is about 21 kcal/mol larger than in a
vacuum, which makes it the most exothermic and exoergic
process among the formations of all regioisomers (Table 4).
The exotherm for the formations of other regioisomers are
smaller, and the resulting order of thermodynamic stability
follows: 11c[H]>11b[H]>11d[H]>11a. Interestingly, in
both sets of calculations, the N-isomer 11a is predicted to be
the least thermodynamically stable.
MP2//DFT calculations demonstrated that the dipole
moment of the zwitterionic compounds significantly de-
pends on the solvent polarity (Table 4). For instance, with
increasing polarity of the medium, the dipole moment of 1
increases (μ=6.1 D, ε=1; μ=8.1 D, ε=13.3; μ=8.4 D,
ε=36.6) due to increasing negative charge at the 10 posi-
tion from þ0.062 for B(10) and þ0.0013 for H(10) (ε=
13.3) to þ0.051 for B(10) and -0.0002 for H(10) (ε =
36.6). This is consistent with the observed solvent
effect on NMR spectra and higher nuclear shielding
of these atoms in CD3CN as compared to benzene-d6
solutions (vide supra).
˚
2.13 A). The DFT calculated activation parameters,
ΔH‡ = 25.4 kcal/mol and ΔS‡ = 12.5 cal/(mol K), fall
short of the experimental data (vide supra). Single-point
calculations for 1-TS at the MP2//DFT level of theory
gave ΔH‡=30.4 kcal/mol in the pyridine dielectric me-
dium, which is lower by 6 kcal/mol than the enthalpy
change for the complete reaction. In addition, the calcu-
lations closely approximated an energy plateau between
˚
˚
˚
dC-N=2.13 A (ΔESCF=33.95 kcal/mol) and dC-N=2.70 A
(ΔESCF=34.45 kcal/mol), whereas ΔESCF for dC-N=3.5 A
is 36.6 kcal/mol and for the complete reaction is
37.8 kcal/mol.
MP2-levelcalculationsdemonstratedthatthecarbonium
ylide 19, a member of a hypercloso family, is a zwitterionic
species. The carbon exocyclic (radial) orbital is practically a
pure p orbital and partially populated (-0.34 e) at the
expense of the boron atoms in the lower belt (Table 3). The
lowest unoccupied molecular orbital (LUMO) of 19 is
primarily localized on the Ccage atom with a contribution
from the lower belt of boron atoms (Figure 2).
The loss of electron density from the carbon exocyclic
orbital markedly affects the geometry of the {closo-1-
CB9} cage (Table 3). Thus, the departure of N2 from 1
results in a less pyramidalized apical carbon atom, con-
traction of the Ccage-B interatomic distances, and sig-
nificant expansion of the interbelt B-B bonds in the ylide
19 as compared to 1.
Electronic and geometrical features of 19 are similar to
those found for its recently reported and extensively
discussed 12-vertex B-undecamethyl analogue, the [clo-
so-1-CB11Me11] ylide.31
Reaction of Ylide 19 with Nucleophiles. The resulting
electron-deficient intermediate, carbonium ylide 19, is a
Calculations in the dielectric medium also demon-
strated that two of the protonated derivatives, 11b[H]
and 11d[H], are significantly stronger acids than pyridi-
nium, and proton transfer to pyridine is moderately
exothermic (Table 5). For the C(3) isomer 11c[H], the
process is practically thermoneutral. The resulting anions
11[-] have very similar energies in the pyridine dielectric
medium, with the C(4) isomer 11d[-] slightly more stable
(31) Vyakaranam, K.; Korbe, S.; Divisova, H.; Michl, J. J. Am. Chem.
Soc. 2004, 126, 15795–15801.

Article
Inorganic Chemistry, Vol. 48, No. 15, 2009 7319
Figure 5. Formation of radical ion 26. Enthalpy change relative to 1 and pyridine reactants calculated with MP2/6-31G(d,p)//B3LYP/6-31G(d,p) for the
pyridine solution (ε=13.3).
Table 2. Enthalpy Change in Heterolytic Dissociation of 1a
Table 3. Calculated Structural Parameters for Selected Derivatives of {closo-1-
CB9}
þ
X=H
X=N2
X=•
X=þ
method
ΔH, kcal/mol
14
1
25
19
Gas Phase
Geometrya,b
B3LYP/6-31G(d,p)
MP2/6-31G(d,p)//B3LYP/6-31G(d,p)
MP2/6-31G(d,p)
24.6
33.2
33.3
˚
C(1)-B(2)/A
1.602
1.836
1.802
1.700
108.3
1.624
1.885
1.785
1.711
110.2
1.585
1.849
1.809
1.698
111.2
1.532
1.928
1.829
1.678
125.8
˚
B(2)-B(3)/A
B(2)-B(6)/A
B(6)-B(10)/A
B-C-B/deg
˚
ICPM model (ε = 36.6)
˚
MP2/6-31G(d,p)//B3LYP/6-31G(d,p)
MP2/6-31þG(d,p)//B3LYP/6-31G(d,p)
MP2/6-31þG(d,p)//MP2/6-31G(d,p)
36.4
35.0
34.4
Natural Atomic Chargec
C(1)
B(2)
B(6)
B(10)
-0.77
-0.52
0.08
-0.19
-0.06
-0.36
-0.01
-0.18
-0.20
-0.34
0.07
0.11
a B3LYP/6-31G(d,p) thermodynamic corrections. ΔS =þ43.6 cal/
0.03
-0.19
-0.17
molK.
-0.32
than the C(2) isomer 11b[-] and more stable by 1.7 kcal/
mol than 11c[-].
Bondingc
Formation of Pyridine Derivatives 11. The formation of
four regioisomeric products, 11a-11d, may involve either
a two-electron or a one-electron process. Initially, we
focused on the former and considered electrophilic reac-
tions of the carbonium ylide 19 with pyridine.
C(1) exo
sp2.03 (1.94) sp3.37 (1.96) sp3.97 (0.95) sp99.8 (0.67)
a Optimized at the MP2/6-31G(d,p) level of theory at the C4v point
b
˚
˚
group symmetry. Ccage-N distance, 1.337 A; N-N distance, 1.149 A.
c NBO analysis. Exocage orbital occupancy in parentheses.
renders the process irreversible. The computational result
favoring the C(3) isomer 21c is consistent with the estab-
lished regioselectivity of electrophilic substitution in pyr-
idine32 but is not consistent with the experimental
observation of formation of the C(2) regioisomer 11b as
the major product.
The formation of the N(1) isomer (11a) can be envi-
sioned as a simple one-step collapse of a Lewis acid (ylide
19) and a Lewis base (pyridine) to form the Ccage-N bond
(Figure 3). Our attempts to locate the transition state for
the reaction using the DFT method were unsuccessful,
which indicates that the process may be practically bar-
rierless. In contrast, the formation of the C-substituted
isomers 11b-11d involves a two-step process, with the
initial formation of arenium zwitterions 21b-21d foll-
owed by a proton transfer. The calculated energy for
the arenium ions is much higher, by about 60 kcal/mol,
than that for the direct adduct 11a. Computational results
indicate that the order of stability for zwitterions 21
follows 21c > 21b > 21d in both the gas phase and the
dielectric medium; the latter stabilizes the highly polar
intermediates by 10-13 kcal/mol relative to the gas
phase. Proton transfer from 21 to pyridine and the
formation of anions 11[-] are highly exothermic, which
Full structure optimizations at the MP2/6-31G(d,p)
level of theory gave a different geometry for the inter-
mediate complex of ylide 19 with pyridine (Figure 4).
Instead of a Wheland-type adduct, in which the {1-closo-
CB9} cluster is connected to the arenium fragment by a
˚
single bond about 1.60 A long (21, Figure 3), the MP2-
optimized geometry shows adduction to an edge (CdC
bond) of the aromatic ring forming a three-center, two-
electron bond in 22bc and 22cd (Figure 4). Attempts at
(32) Eicher, T.; Hauptmann, S. In The Chemistry of Heterocycles; Georg
Thieme Verlag: New York, 1995; pp 273-276 and references therein.

7320 Inorganic Chemistry, Vol. 48, No. 15, 2009
Ringstrand et al.
Table 4. Enthalpy Change in Substitution Reactions of 1a
An alternative mechanism for the formation of
11b-11d may involve a radical process initiated by the
addition of pyridine to the electrophilic dinitrogen
group of 1. Indeed, both DFT- and MP2-level calcula-
tions located the pyridine adduct 23 on the potential
energy surface (PES). The formation of the highly polar
azene 23 (μ=23.2 D) was found to be endothermic by
14.3 kcal/mol or endoergic by 24.8 kcal/mol at the MP2//
DFT level of theory in the pyridine dielectric medium
(ε=13.3).
According to this mechanism, azene 23 undergoes
fragmentation with a homolytic cleavage of the N-
pyridine bond, leading to the radical ion pair 24 and 25
(Figure 5). The process is highly endothermic in the
gas phase (ΔH =133.0 kcal/mol), but it becomes more
ΔH/kcal mol-1
dipole/D
reagent product
X
ε=1b ε=13.3c ε=1b ε=13.3c
d
d
Me2S
pyridine 11a
9a
Me2Sþ
-45.5
12.8
1-C5H5Nþ
-59.2 -67.2
15.1 18.6
15.7 19.1
19.2 24.2
19.2 24.2
14.0 17.4e
11b[H] 2-C5H4NHþ -67.9 -76.4
11c[H] 3-C5H4NHþ -59.9 -81.4
11d[H] 4-C5H4NHþ -61.0 -75.4
favorable in the pyridine dielectric medium (ΔG298
=
MeCN 17
NMe3 20
MeCNþ
-30.7 -34.7e
þ
d
d
þ44.5 kcal/mol). The diazenyl radical 24 easily loses
molecular nitrogen (ΔG298=-24.8 kcal/mol) and forms
radical ion 26. The dissociation of the Ccage-N bond and
the direct formation of radical anion 26 is also possible,
although the second ion, the pyridinediazenyl radical
(27), was not located on the potential energy surface
with the DFT method. The calculated overall enthalpy
change for the formation of the ion pair and N2 from 1 is
þ145.9 kcal/mol in the gas phase, but in pyridine (ε =
13.3), this energy is lowered to ΔH = 56.4 kcal/mol or
ΔG298=44.4 kcal/mol (Figure 5).
NMe3
-60.0
13.3
a MP2/6-31G(d,p)//B3LYP/6-31G(d,p) level with DFT thermody-
namic corrections. b Vacuum calculations. c IPCM solvation model.
d Not calculated. e Calculated for ε=36.6.
Table 5. Calculated Free Energy Change in Deprotonation Reactions of 11a
NBO analysis indicates that the apical carbon atom
in radical ion 26 has a significant negative charge density
(-0.36) and the radical is expected to be moderately
nucleophilic (Table 3). Therefore, it may react with a
molecule of neutral pyridine (solvent), which leads to
regioisomeric radical anion adducts 28 (analogous to
zwitterions 21 in Figure 3). Results show that, in the gas
phase, the C(2) intermediate 28b, which leads to the
experimentally observed major product 11b, is least fa-
vored, and its regioisomers are more stable by about
3-4 kcal/mol (stability order: 28c>28a, 28d>28b, Table 6).
The IPCM calculations demonstrated diminished exo-
therms of the adduct formation in the dielectric medium,
with the heat of formation for the C(2) adduct 28b
being least affected. Consequently, the order of stability of
the intermediates is changed in pyridine; the N(1) adduct
28a, leading to the unobserved 11a, becomes the least
favored, whereas 28b is less stable than 28c by only 1.4
kcal/mol (order of stability: 28c>28b>28d>28a, Table 6).
The resulting adducts 28b-28d undergo a highly exo-
thermic (about 115 kcal/mol) process, leading to pro-
ducts 11b[-]-11d[-] and involving either a hydrogen atom
transfer or a sequence of electrons followed by proton
transfer with pyridinium radical 25. A reaction of radical
anion 26 with pyridine radical cation 25 is predicted to be
highly exothermic and lead directly to 11a.
b
c
d
ΔG298 kcal mol-1
-5.9
þ0.5
-7.3
a MP2/6-31G(d,p)//B3LYP/6-31G(d,p) calculations with DFT ther-
modynamic corrections. IPCM solvation model ε=13.3.
obtaining equilibrium geometry for the analogous struc-
ture 22ab, with the Ccage
C
N two-electron bond,
3 3 3 3 3 3
were unsuccessful, and the optimization gave the struc-
ture of 11a instead. Geometry reoptimization at the DFT
level of theory starting at the 22bc geometry resulted in
the structure 21b. The Ccage-Cring distances in 22bc
(adduct to the C(2)-C(3) edge) and 22cd (adduct to the
˚
C(3)-C(4) edge) are about 1.7 A. Subsequent deprotona-
tion of 22 leads to the formation of the product 11. MP2
results demonstrated that both complexes have similar
energies (ΔESCF=2.0 kcal/mol), which are comparable to
those of the Wheland complexes 21.
Structures similar to those of 22 were deduced from
experimental data for intermediates involved in reactions
of [hypercloso-1-CB11Me11] ylide with aromatic hydro-
carbons.33
Even though transition state structures for the forma-
tion of 21 and 22 were not located and their energies were
not calculated, it seems unlikely that 21 and 22 would
be formed preferentially to the direct collapse of the
reactants into the N(1) isomer 11a. Indeed, a reaction
of [hypercloso-1-CB11Me11] ylide with pyridine gives the
corresponding N(1) isomer as the only isolated product.33
Our initial attempts to locate transition state structures
for either type of adduct, radicals 28 or zwitterions 21,
were unsuccessful and, due to computational intensity,
were not pursued further.
Phenylation of Pyridine. To provide a better support
for the proposed radical process of formation of 11, we
also analyzed an analogous reaction between the benze-
nediazonium ion and pyridine (Figure 6 and Table 6).
According to the accepted mechanism, the formation of
(33) Vyakaranam, K.; Havlas, Z.; Michl, J. J. Am. Chem. Soc. 2007, 129,
4172–4174.

Article
Inorganic Chemistry, Vol. 48, No. 15, 2009 7321
Table 6. Calculated Enthalpy Change for Reactions with Pyridinea
reaction intermediate
N(1)
C(2)
C(3)
C(4)
d
R*
*
adduct
21
ε
a
b
c
þ
•
1b
-92.5
-102.6
-32.0
-44.4
-36.4
-49.4
-26.9
-37.2
19
26
13.3c
28
33
1b
-28.0
-22.9
-24.9
-23.4
-29.2
-24.8
-28.0
-23.1
13.3c
C6H5• 32
•
1b
-28.4
-26.0
-26.0
-25.3
-27.6
-28.4
-25.8
-28.0
13.3c
a MP2/6-31G(d,p)//B3LYP/6-31G(d,p) level with DFT thermodynamic corrections. b Vacuum calculations. c IPCM solvation model.
N-phenylazopyridinium (29) is the first step in the free
radical phenylation of pyridine, which leads to three
regioisomeric C-phenylpyridines (30).34-36 DFT calcula-
tions located the N-diazene 29 on the potential energy
surface, and subsequent MP2//DFT calculations revea-
led that its formation is moderately endothermic (ΔH=
þ2.9 kcal/mol, ΔG298=þ14.5 kcal/mol) in the pyridine
dielectric medium (Figure 6). Homolytic cleavage of the
N-N bond leading to the formation of phenyldiazenyl
radical (31) and radical cation 25 is significantly endoer-
gic (ΔG298=þ51.7 kcal/mol). This is partially compen-
sated by the favorable decomposition of the diazenyl
radical 31 and the formation of phenyl radical (32) and
N2 (ΔH = -11.6 kcal/mol, ΔG298= -21.5 kcal/mol).
Nevertheless, the overall process of formation of the
phenyl radical is endothermic (Figure 6).
Figure 6. Formation of phenyl radical 32. Energy change relative to
initial reactants calculated with MP2/6-31G(d,p)//B3LYP/6-31G(d,p) for
the pyridine solution (ε=13.3).
The addition of the resulting phenyl radical (32) to
pyridine and the formation of the intermediates 33 is
moderately exothermic with the highest exotherm calcu-
lated for the N-adduct 33a (stability order: 33a>33c>
33b>33d) in the gas phase (Table 6). The same calcula-
tions in the dielectric medium decrease the reaction
exotherm for 33a and 33b and increase the exotherm for
33c and 33d. Consequently, the order of thermodynamic
stability of the four intermediates is altered, giving the
preference for the C(3) isomer 33c (stability order: 33c>
33d>33a>33b) in pyridine. Subsequent H• transfer from
33 to 25 is highly exothermic (ΔH ≈ -110 kcal/mol) and
gives the isomeric pyridines 30. The calculated order of
their thermodynamic stability (30b > 30d ∼ 30c) is con-
sistent with experimentally established trends in heats of
formation.37
Table 7. Energy Change in Dissociation of 34 in MeOHa
ε
ΔH/kcal mol-1
ΔG302/kcal mol-1
1b
67.9
24.5
56.1
12.6
30.9c,d
a MP2/6-31G(d,p)//B3LYP/6-31G(d,p) level with DFT thermody-
namic corrections. b Vacuum calculations. c IPCM solvation model.
d Value interpolated from data in ref 39.
by computing of the energy change for dissociation of
Finally, we assessed the accuracy of the IPCM method
in predicting the formation of ion pairs, such as 25 and 26,
azosulfone 34 and the formation of ion pair PhN2þ and
PhSO2 in a dielectric medium. The experimentally38
-
established association constant Kassoc = 1.44 ꢀ 105 L/
þ
-
mol for the ion pair PhN2 and PhSO2 in MeOH at
29 °C allows for the calculation of ΔG302=þ7.1 kcal/mol
(34) Abramovitch, R. A.; Saha, J. G. Tetrahedron 1965, 21, 3297–3303.
(35) Zollinger, H. Acc. Chem. Res. 1973, 6, 335–341.
(36) Galli, C. Chem. Rev. 1988, 88, 765–792.
(37) Ribeiro da Silva, M. A. V.; Matos, M. A. R.; Rio, C. A.; Morais,
V. M. F.; Wang, J.; Nichols, G.; Chickos, J. S. J. Phys. Chem. A 2000, 104,
1774–1778.
(38) Ritchie, C. D.; Saltiel, J. D.; Lewis, E. S. J. Am. Chem. Soc. 1961, 83,
4601–4605.

7322 Inorganic Chemistry, Vol. 48, No. 15, 2009
Ringstrand et al.
Figure 7. Electronic absorption spectra for selected compounds recorded in CH3CN.
Figure 9. DFT-generated HOMO and LUMO contours for 11a.
Figure 8. DFT-generated MO contours for 1. For clarity, only one part
of each degenerated MO is shown.
Figure 10. ZINDO-generated HOMO and LUMOþ1 contours for 8b.
for the dissociation of 34. Our MP2//DFT calculations in
the gas phase demonstrated a high endotherm (ΔH =
þ67.9 kcal/mol) for the dissociation process of 34
(Table 7). In the dielectric medium, however, the dissocia-
tion process of 34 becomes significantly more favorable,
reaching a value ofΔG302=þ12.6 kcal/mol in MeOH. This
represents a fairly good agreement between the theory
and experimental results and indicates that MP2//DFT
calculations underestimated the solvation of the ions by
5.5 kcal/mol. Consequently, the stability of other ions in a
dielectric medium is most likely underestimated, and their
formation is more favorable than calculated.
Electronic Absorption Spectroscopy. Spectroscopic ana-
lysis of four derivatives, 1, 8, 9a, 11a, and trimethylamine
zwitterion 20 provides a convenient means to study
electronic interactions between the {closo-1-CB9} cage
and the substituent at the C(1) position.
Among the five substituents, NMe3 in 20 is the simplest
and does not possess easily available electrons or acces-
sible orbitals. Therefore, the electronic absorption spec-
trum of 20 closely resembles that of the parent [closo-1-
CB9H10]- anion (14) with the absorption tailing to about
250 nm and the maximum below 200 nm (Figure 7a). The
presence of an electron pair in the SMe2 substituent of
9a resulted in the increased intensity of the band at
about 200, the appearance of a shoulder band at about
225 nm, and also a broad low-intensity band at 272 nm.
These spectral features are consistent with the transitions
observed at about 230 and 200 nm for trialkylsulfonium
salts.40 It was also noted that the UV spectra of sulfonium
salts are affected by the presence of molecular oxygen,
which gives rise to the appearance of charge-transfer (CT)
bands.40
The dinitrogen, N2, and pyridine groups are π substit-
uents, and their orbitals can interact with the electronic
manifold of the {closo-1-CB9} cage. Thus, the dinitrogen
derivative 1 exhibits a single absorption band at 250 nm
(Figure 7a). ZINDO//MP2 calculations revealed that 1
exhibits two π f π* transitions, at 280 nm (f=0.49) and a
weaker one at 242 nm (f=0.12) in a vacuum. The former
involves the double-degenerated highest occupied mole-
cular orbital (HOMO), localized mostly on the cage,
and the LUMO, localized primarily on the N2 group
(Figure 8). The higher energy transition is due to intrac-
age excitations (HOMO f LUMOþ1). The same calcu-
lations conducted for 1 in the dielectric medium of the
solvent (ε=36.6) revealed a substantial negative solvato-
chromic effect of the cage-to-substituent excitation and
practically no solvent effect of the intracage excitation. As
a consequence, the two transitions merge into one, which
is calculated at 247 nm (f=0.56) and observed at 250 nm.
ꢀ
(39) Le Fevre, R. J. W. Trans. Faraday Soc. 1938, 34, 1127–1132.
(40) Ohkubo, K.; Yamabe, T. J. Org. Chem. 1971, 36, 3149–3155.

Article
Inorganic Chemistry, Vol. 48, No. 15, 2009 7323
Figure 11. Relative1H(a)and absolute 11B (b)NMR chemicalshiftsmeasuredin CD3CN fortheantipodal position of [closo-1-CB9H9-1-X]- and[closo-1-
CB11H11-1-X]- plotted against the substituent σΡ values. The σp parameters of the -NdN-Ph and -NHCOOMe groups were used for the substituents in
13 and 4a, respectively. Best fit lines: (a) y=0.73x, r2=0.990; (b) {closo-1-CB9} y=12.7x þ 25.1, r2=0.86; {closo-1-CB11} y=5.3x - 10.0, r2=0.67.
Pyridine is the most complex substitiuent among the
five groups; it has both the low-lying LUMO and also its
own π f π* transitions in the UV region. Consequently,
the pyridine derivative 11a is expected to exhibit three
types electronic transitions: cage-ring, intraring, and in-
tracage. The first two types apparently overlap and
form a complex structure in the range of 240-300 nm
(Figure 7a). The maximum of the broad, intense under-
lying band is estimated at 269 nm and attributed to the
cage-to-ring CT transition. The upper fine structure has
two maxima at 268 and 274 nm and represents the
pyridine B-band. Results of ZINDO//MP2 calculation
for 11a in a dielectric medium (ε = 36.6) support this
assignment. The CT transition is calculated at 260 nm
(f=0.32) and involves the HOMO f LUMO excitation
(Figure 9) with a small contribution from the intracage
excitation (HOMO f LUMOþ4). The pyridine B-band
appears at 272 nm (f=0.08) and involves HOMO-2 f
LUMO excitation. The same calculations for 11a in
a vacuum give the CT transition at 361 nm (f = 0.18),
which demonstrates a large negative solvatochromic
effect (-1.34 eV) in this polar derivative (Table 4). The
pyridine B-band experiences a smaller and positive solva-
tochromic shift (þ0.37 eV).
HOMO, localized on the organic fragment, mainly to the
LUMO þ 1 and also to the LUMO þ 4 localized on the
CB9H9-NdN fragment (Figure 10).
In comparison to the azobenzene analogs, the electro-
nic transitions of the {closo-1-CB9} derivatives 8a and 8b
are blue-shifted by 24-34 nm. For instance, absorption
bands of 4-hydroxyazobenzene are recorded at 344 and
425 nm in CH3CN.42 In 4-aminoazobenzene, both ab-
sorption bands overlap to form a broad band43 with a
maximum at 382 nm in CH3CN.44
NMR Studies. The availability of several diverse deri-
vatives of the [closo-1-CB9H10]- cluster provided an op-
portunity to analyze the transmission of the substitutent
effect through the cage and compare the results to those
for analogous benzene and [closo-1-CB11H12]- deriva-
tives. Thus, the relative NMR chemical shifts (Δδ) for
the antipodal B(10)-H were plotted versus the substituent
σp values,45 and the results are shown in Figure 11.
1
A Hammett plot of Δδ versus Hammett σp for H
NMR B(10)-H chemical shifts showed a good linear fit
for several substituents (Figure 11a). The correlation
factor r2 is 0.94 for all substituents, and it increases to
0.99 if the two outlaying data points (X = NH3 and
X=Cl) are removed. This change has little effect on the
slope (F) of the linear fit, which slightly increases from
0.70 ( 0.05 to 0.73 ( 0.02 for the smaller data set.
Interestingly, the slope for the B(10)-H chemical shift is
practically the same as that obtained for the C(4)-H
chemical shift in monosubstituted benzene derivatives (F
=0.70 ( 0.09) after correction for anisotropic magnetic
shielding.28,46
Electronic spectra of the azo derivatives 8a and 8b
exhibit absorption bands typical for azobenzenes.41 The
intense band of the π f π* transition in azophenol 8a
appears at 318 nm (log ε=4.30) and in azoaniline 8b at
348 nm (Figure 7b), while the forbidden n f π* transition
is located at about 400 nm in both derivatives.
The Hammett plot of B(10) and B(12)47 11B NMR
chemical shifts for {closo-1-CB9} and {closo-1-CB11},
ZINDO//DFT calculations in the MeCN dielectric
medium (ε=36.6) demonstrated a low-intensity transi-
tion at 445 nm (f<0.001), which involves excitation from
HOMO-2 and HOMO-4, localized on the nitrogen
atom lone pairs and the cage, to the π symmetry orbitals
LUMO þ 1 and LUMO þ 4, which are delocalized over
the molecule with the main component from the {closo-1-
CB9} cage. The more intense transition is calculated at
329 nm (f=0.82), which involves the excitation from the
(42) Kojima, M.; Nebashi, S.; Ogawa, K.; Kurita, N. J. Phys. Org. Chem.
2005, 18, 994–1000.
(43) Korolev, B. A.; Titova, S. P.; Ufimtsev, V. N. J. Org. Chem. USSR
1971, 7, 1228–1231.
(44) Uno, B.; Matsuhisa, Y.; Kano, K.; Kubota, T. Chem. Pharm. Bull.
1984, 32, 1691–1698.
(45) Hansch, C.; Leo, A.; Taft, R. W. Chem. Rev. 1991, 91, 165–195.
(46) Figeys, H. P.; Flammang, R. Mol. Phys. 1967, 12, 581–587.
(47) Jelinek, T.; Plesek, J.; Hermanek, S.; Stibr, B. Collect. Czech. Chem.
Commun. 1986, 51, 819–829.
ꢁ
(41) Jaffe, H. H. Theory and Applications of Ultraviolet Spectroscopy;
Wiley: New York, 1962; pp 276-286.

7324 Inorganic Chemistry, Vol. 48, No. 15, 2009
Ringstrand et al.
Figure 12. Proposed mechanism for the formation of pyridine derivatives 11.
respectively, showed that {closo-1-CB9} (F=12.7 ( 1.8)
appears to be significantly more sensitive than {closo-1-
CB11} (F=5.3 ( 1.3) to the nature of the C(1) substituent
(Figure 11b).
since the process is favored kinetically (anticipated low acti-
vation barrier) and has a large exotherm (ΔH=-102.6 kcal/
mol). Support for this expectation is provided by the reported
smooth transformations of similar dinitrogen derivatives
of the [closo-B10H10]2- cluster with pyridine,2-4,6-8 and also
the isolation of [closo-1-CB11Me11-1-C5H5N] as the only
product of a reaction of the [hypercloso-1-CB11Me11] ylide
with pyridine.33 Instead, the observed distribution of pro-
ducts in the reaction of 1 with pyridine suggests a radi-
cal pathway, which is similar to the known Gomberg-
Bachmann free-radical arylation reaction.49 In both reac-
tions, all C-substituted pyridine isomers are formed with
the dominant proportions of the C(2) isomer.34,50 The radi-
cal mechanism for the formation of 11 is supported also
by computational results, which demonstrate that the for-
mation of radical ion 26 from 1 is as feasible as the genera-
tion of Ph• (32) from PhN2þ, considering the difference in the
treatment of ion pairs and nonionic molecules by the IPCM
model.
From the mechanistic point of view, reaction of 1 in neat
pyridine can proceed by either a heterolytic or homolytic
pathway (Figure 12). The first and also rate-determining step
in the former mechanism is a unimolecular first-order reac-
tion, whereas the rate-limiting step in the radical pathway is
pseudo-first-order formation of diazenyl radical 24, since
[pyridine] ≈ const.[1]. At ambient temperature, the radical
process appears to be significantly faster (kc , Keqkr),51 and
the resulting radical anion 26 attacks the dominant neutral
pyridine (solvent) following pseudo-first-order kinetics. This
reaction is faster than the energetically more favorable
(Table 4) ion recombination reaction of 26 and the pyridi-
nium radical 25 (a second-order process), which, presumably,
leads directly to the N(1) adduct 11a. The attack of 26 on
pyridine is calculated to be moderately exoergic (ΔG298
between -11.3 kcal/mol for 28a and -13.7 kcal/mol for
28c), and with a sufficiently low activation barrier, the
process may be reversible, leading to the thermodynamic
intermediate 28. The subsequent electron transfer (ET)
from 28a and hydrogen transfer (HT) from 28b-28d to
2552 and the formation of 11a and 11b[-]-11d[-], respectively,
are highly exothermic (ΔH>100 kcal/mol in the dielectric
medium) and considered irreversible. Therefore, the forma-
tion of 11b as the major component of the mixture may
arise either from the kinetic preference for the formation of
intermediate 28b (if the barrier is sufficiently high and the HT
Discussion
Experiments demonstrated that the dinitrogen derivative 1
is stable in the solid state but decomposes in solutions. In
MeCN, the process is slow, and the estimated half-life of 1 is
t1/2=91 h at 25 °C and t1/2=2 s at 90 °C. The consistency
of the measured high enthalpy of activation ΔH‡ of 38.4 (
0.8 kcal/mol with the calculated enthalpy change, ΔH, and
the formation of a single product (17) support the heterolytic
mechanism for the decomposition of 1 and the formation of
the reactive intermediate 19 in MeCN. In addition, the
unusually high experimental entropy of activation, ΔS‡
(44.5 ( 2.5 cal/molK), is similar to the calculated ΔS value
of 43.6 cal/mol K. This suggests that the dissociation process
involves stretching the Ccage-N bond to the edge of the van
der Waals contact between the atoms. Alternatively, solvent
molecules are involved in the transition state, and the high
value of ΔS‡ reflects differential solvation of the ground and
transition states. The proposed heterolytic unimolecular
decomposition reaction of 1 is similar to that described for
PhN2þ in polar aprotic solvents.48 Interestingly, ΔG298‡ for 1
is higher by about 1.5 kcal/mol than that reported for
PhN2
.
þ 48 Also, the heterolysis of 1 haþs a significantly higher
ΔS‡ than that measured for PhN2 in MeCN (18.3 cal/
mol K) or in MeNO2 (30.9 cal/mol K).48
A change of solvent significantly accelerates the transfor-
mation of 1, which is then usually fully consumed in less than
20 h at ambient temperature. The observed increase of the
reaction rate can be due to the lower activation energy for the
decomposition of 1 in less polar solvents. Calculations
showed that the ΔH for the formation of ylide 19 is lower
by 1 kcal/mol in the pyridine dielectric medium and 3.2 kcal/
mol in a vacuum relative to that in MeCN (Table 2) due to
a greater dielectric stabilization of 1 than the products.
Consequently, heterolysis of 1 can be nearly 20 times faster
in pyridine and nearly 50 times faster in a vacuum than in
MeCN, assuming the same ΔS‡ and that the difference in ΔH
of the reaction reflects the difference in ΔH‡. Often, however,
ΔS‡ changes with the solvent (cf. ΔS‡ for PhN2þ). Thus, for
heterolysis of 1 in less polar solvents, the entropy of activa-
tion may be smaller, which will partially compensate the
change in ΔH‡, and consequently less impact on the reaction
rate will be observed. Alternatively, a change of mechanism
can be responsible for the significantly different reaction rate.
The reaction of ylide 19 with pyridine is expected to give
the N(1) adduct 11a directly as the only or dominant product,
(49) (a) Dermer, O. C.; Edmison, M. T. Chem. Rev. 1957, 57, 77–122.
(b) Bachmann, W. E.; Hoffman, R. A. Org. React. 1944, 2, 224–261.
(50) Beadle, J. R.; Korzeniowski, S. H.; Rosenberg, D. E.; Garcia-Slanga,
B. J.; Gokel, G. W. J. Org. Chem. 1984, 49, 1594–1603.
(51) From the steady-state approximation considering that the rate of
decomposition of adduct 23 is much slower than the reverse reaction.
(52) Since [26] ≈ [25], radical anion 26 competes with 25 for the hydrogen
from 28, which may explain the observed low yield formation of 14.
(48) Ishida, K.; Kobori, N.; Kobayashi, M.; Minato, H. Bull. Chem. Soc.
Jpn. 1970, 43, 285–286.

Article
Inorganic Chemistry, Vol. 48, No. 15, 2009 7325
process is fast) or from the thermodynamic stability
of 28b (if the barrier is sufficiently low and the HT process
is slow). The latter implies that the calculations underesti-
mated the stability of 28b by about 2 kcal/mol relative to 28c,
which is not unreasonable. In fact, all C-substituted inter-
mediates have comparable calculated energies within 2 kcal/
mol in the pyridine dielectric medium. It needs to be remem-
bered that the dielectric medium significantly destabilizes 28c
and has the least effect on the energy of formation of 28b
(Table 6). Also, 28a is the thermodynamically least favorable
intermediate, which is in agreement with its practical absence
among the reaction products. A similar discrepancy between
the calculated thermodynamic stability of the radical inter-
mediates 33 and the distribution of the products 30 is found
for the phenylation of pyridine (Table 6). Again, this may
indicate that product formation is governed by kinetics
(irreversible processes), or the relative energies are calculated
with an error of (2 kcal/mol.
as the counterion in the crystal structure of the isolated
chloride 15.53
Results of reactions of 1 in dilute solutions of pyridine in
MeCN are also consistent with the dual mechanism and
involvement of the two reactive intermediates. In addition to
17 and 11a, products expected for trapping of the ylide 19, the
reaction mixture contained several unidentified signals,
which are presumably related to reactions of radical ion 26
with the solvent and pyridine. Interestingly, the rate of
decomposition of 1 in the MeCN/pyridine solution is ap-
proximately the same as the rate in pure MeCN.
The mechanism of pyridine-induced radical reactions of 1
is similar to the accepted mechanism of free-radical arylation
of pyridine with benzenediazonium salts.34-36,54 In both
cases, a stable adduct of pyridine to the N2 group was located
on the PES (compounds 23 and 29). The relatively high
energy of adducts 23 and 29 suggests that they are transient
species formed at low concentrations, which give rise to the
radical ion pairs. Further studies of reactions of arenediazo-
nium salts indicated that the radical/cationic pathway corre-
lates with the nucleophilicity of the solvent36,54 and does
not necessarily involve a stable transient adduct but rather
involves an intermolecular electron transfer (outer-sphere
ET).36 Thus, pyridine has one of the highest nucleophilicity
parameters, whereas acetonitrile has one of the lowest
nucleophilicity parameters.54,55 Therefore, the former may
induce radical generation, while the latter will favor cationic
processes. This is indeed observed experimentally for the
decomposition of both 1 (present work) and benzenediazo-
nium salts56-60 for which the corresponding acetamides are
the main products in MeCN solutions (see Scheme 7).
The propensiþty of a nucleophile to induce radical reactions
of 1 and PhN2 can be assessed from the energetics of ET
between the diazonium and the donor calculated from
ionization potential (Ip) and electron affinity (Ea) for a gas-
At higher temperatures, the rate of formation of the ylide
19 becomes competitive with that of generation of the radical
anion 26 (kc ∼ Keqkr), and consequently, the N(1) isomer 11a
is formed in substantial amounts along with its C-isomers
(Table 1).
In dilute solutions of pyridine in benzene, both mechan-
isms, the heterolytic and the homolytic decomposition of 1,
may operate at the same time, as is indicated by the formation
of N(1) and C(2) isomers (11a and 11b, respectively) in
comparable amounts (Table 1). The lower dielectric permit-
tivity of benzene may increase the rate of the formation of 19,
while the lower concentration of pyridine in solution could
decrease the rate of formation of radical anion 26. Conse-
quently, both intermediates, 19 and 26, can be formed at
similar rates. The radical anion is expected to react easily with
both pyridine and benzene, leading to 11b and [closo-1-
CB9H9-1-Ph]- (16), respectively. Calculations support this
expectation and demonstrate that both addition reactions of
26 to benzene and pyridine at the C(2) position have compar-
able exotherms ΔH of -24.9 and -23.4 kcal/mol, respec-
tively, in the pyridine dielectric medium. The ylide 19 is
expected to attack pyridine’s N atom, leading to 11a and
benzene to form 16. The formation of the latter presumably
follows the mechanism proposed for the electrophilic sub-
stitution of aromatic hydrocarbons with the [hypercloso-1-
CB11Me11] ylide33 and involves an three-center, two-electron
bond intermediate analogous to 22 (Figure 4).
In methylene chloride solutions, the situation is similar,
and transformation of 1 may follow both mechanistic path-
ways involving both reactive intermediates. Theradical anion
26 is expected to react with pyridine, giving 11b, and to
abstract hydrogen from the solvent, giving the observed
parent cluster 14. It may also transfer a Cl atom to form
the chloride [closo-1-CB9H9-1-Cl]- (15). Calculations indi-
cate that abstractions of a chlorine atom and a hydrogen
from CH2Cl2 are both exothermic processes (ΔG298=-19.3
and -18.6 kcal/mol, respectively) in a dielectric medium
(ε=13.3) and more favorable energetically than addition to
the C(2) of pyridine (ΔG298=-12.3 kcal/mol). The ylide 19
formed in an unassisted thermolysis of 1 gives the N(1)
adduct 11a and may also abstract Cl- from the solvent,
forming the chloride [closo-1-CB9H9-1-Cl]- (15). A subse-
quent reaction of the resulting þCH2Cl with pyridinium
leads to N-(chloromethyl)pyridinium, which was observed
phase reaction, or from oxidation (Eox) and reduction (Ered
)
potentials in solutions (Table 8). Calculations demonstrate
that, while the Ea for the two electrophiles, 1 and PhN2þ, are
different in the gas phase by over 3.5 eV, they are practically
the same inMeCNsolutions. The dielectricmedium stabilizes
charged molecules, and consequently, the addition of an
electron to 1 becomes more favorable, while the reduction
of PhN2þ is less exothermic in MeCN.
The calculated Ip values for the five reagents compare well
with the experimental data61 with a systematic underestima-
tion by about 0.3 eV. In the MeCN dielectric medium, the
calculated Ip values are lowered by 2-3 eV relative to the gas
phase. A comparison of the calculated ionization values to
the experimental electrochemical data (Table 8) can be made
only at a qualitative level, since all investigated processes
(53) Carr, M. J.; Kaszynski, P.; Franken, A.; Kennedy, J. D. Manuscript
in preparation.
(54) Szele, I.; Zollinger, H. Helv. Chim. Acta 1978, 61, 1721–1729.
(55) Maria, P.-C.; Gal, J.-F. J. Phys. Chem. 1985, 89, 1296–1304.
(56) Kice, J. L.; Gabrielsen, R. S. J. Org. Chem. 1970, 35, 1004–1015.
(57) Makarova, L. G.; Nesmeyanov, A. N. Bull Akad. Sci. USSR, Div.
Chem. Sci. 1954, 887–891.
(58) Kobayashi, M.; Minato, H.; Yamada, E.; Kobori, N. Bull. Chem.
Soc. Jpn. 1970, 43, 215–219.
(59) Kobayashi, M.; Minato, H.; Kobori, N. Bull. Chem. Soc. Jpn. 1970,
43, 219–223.
(60) Kobori, N.; Kobayashi, M.; Minato, H. Bull. Chem. Soc. Jpn. 1970,
43, 223–225.
(61) NIST Standard Reference Database Number 69 (http://webbook.
nist.gov/chemistry/) and references therein.

7326 Inorganic Chemistry, Vol. 48, No. 15, 2009
Ringstrand et al.
Table 8. Calculated Adiabatic Electron Affinity (Ea) and Ionization Potential (Ip)
and Experimental Electrochemical Reduction/Oxidation Potentials for Selected
Compounds
Table 9. Calculated Free Energy Change ΔG298 for the Electron Transfer Process
between 1 and Selected Solventsa
•
R-N2þ þ :Nu f R-N2 þ þ• Nu
calculateda
experimentalb
Epaox/V Epcred/V
ε=1c
ε=13.3d ε=36.6d
R
[closo-1-CB9H9]- (1)
ε=1b ε=13.3c ε=36.6c ε=1b ε=13.3c ε=36.6c
Ph-
compound
Nu
electron affinity/eV
1
C6H5N2
2.11
5.84
3.69
3.86
3.76
3.76
-0.54
-0.16e
þ
Me2S
Me2NCHS 132.3
138.8
47.4
49.6
69.2
106.0
125.7
43.0
45.8
65.3
101.4
127.5
53.7
47.2
69.6
115.8
142.2
44.4
46.6
66.3
103.0
122.8
43.9
46.7
66.2
102.3
128.4
ionization potential/eV
Me2S
8.29
(8.69)f
7.94
5.91
5.93
6.82
8.40
9.28
5.77
1.78g
pyridine
CH2Cl2
MeCN
154.7
200.9
227.3
Me2NCHS
pyridine
CH2Cl2
MeCN
5.83
6.72
8.27
9.42
1.25
(8.2)f
8.95
a MP2/6-31G(d,p)//B3LYP/6-31G(d,p) level with DFT thermody-
namic corrections. Energies in kcal/mol. b Vacuum calculations. c IPCM
solvation model.
lit. ∼1.5h
∼1.8i
∼ 2.4i
(9.26)f
10.93
(11.33)f
12.10
by 20 kcal/mol, or 1/3, than those for pyridine (Table 9). This
is consistent with a smaller ΔEp of 1.79 V (Table 8) for the
reaction of 1 and Me2NCHS and indicates that the formation
of 10 and possibly also 9a may involve a radical process,
which begins with outer-sphere ET (intermolecular) and
formation of the radical ion pair.66 According to the pro-
posed radical mechanism of reaction of 1 with Me2S, the
radical ion pair collapses to form the product 9a (Figure 13).
Since the concentration of the ions is small and the rate of this
bimolecular process is low, other processestake place, such as
a loss of proton and presumably an abstraction of H. Conse-
quently, the adduct 9a is isolated only in 40% yield, with the
rest of the precursor forming polymeric/decomposition pro-
ducts. It is expected that the reaction of 19 with Me2S would
give the adduct 9a in a much higher yield.
The reaction with Me2NCHS is much more efficient. The
thionocarbonyls are known to react smoothly with C-cen-
tered radicals, often in a chain process, in which the weak
SdC bond is eliminated.67 Thus, it can be postulated that
radical ion 26 preferentially attacks the sulfur atom, giving
rise to the C-centered transient radical 35 stabilized by both
the N and S atoms (Figure 14). Subsequent ET to the
Me2NCHS radical cation gives product 10. Overall, the
preparation of dialkylsufonium derivatives such as 9 is most
efficiently done in a two-step process using 10 as the inter-
mediate.
(12.20)f
a MP2/6-31G(d,p)//B3LYP/6-31G(d,p) level with DFT enthalpic
corrections. b Peak potential recorded in MeCN (0.05 M Bu4NPF6) vs
SCE. c Vacuum calculations. d IPCM solvation model. e Lit. -0.17
(MeCN, SCE) ref 63. f Experimental data taken from ref 61. g Lit.62
þ1.71 V vs SCE. h Recorded in neat pyridine 64. i Solvent electrochemi-
cal limit; ref 65.
are irreversible in solution and the observed current peak
potentials do not represent thermodynamic processes. The
calculated ionization potentials Ip generally follow the trend
in experimental anodic peak potentials Epaox for all reagents,
with the exception of Me2S (Table 8). In agreement with
calculations, the lowest oxidation potential is observed for
the thioformamide and the highest for MeCN. The experi-
mental Epaox for Me2S is higher by over 0.5 V than predicted,
presumably due to complex chemical behavior of the radical
cation.62
A comparison of calculated electron affinities with the
electrochemical data shows that the reduction of 1 is more
cathodic by 0.39 V than that of PhN2þ. The discrepancy
between calculated solution Ea values and observed Epc
values (Table 8) is presumably due to the difference in
ion solvation (cation vs anion) and solvent reorganization
energy.
Calculated data in Table 9 demonstrate that ET between
The intermediacy of ylide 19 in the formation of 9a or 10
cannot be excluded, and clearly more detailed studies are
necessary for a better understanding of the mechanism and
intermediates involved in transformations of 1.
pyridine and PhN2þ (outer-sphere process) and the formation
of a radical ion pair are endothermic in pyridine (ΔG298
=
66.3 kcal/mol) and practically the same in MeCN solu-
tion (Table 9), which corresponds to the difference in the
measured current peak potentials of ΔEp ≈ 1.7 V (Table 8).
The same analysis for 1 and pyridine gives a similar value
for the ET process in MeCN (ΔG298 = 65.3 kcal/mol) but
Solvents with high a Ip (>10 eV) and Epa, such as MeCN
and CH2Cl2, are not expected to transfer an electron to 1 and
induce its radical transformations. Therefore, decomposition
of 1 in these solvents is likely to involve ylide 19, as postulated
for thermolysis in MeCN. The products of reactions of 1 in
CH2Cl2 and also benzene are largely intractable, which may
be a result of their extensive decomposition caused by an
unsolvated counterion, either þCH2Cl (in reaction with
CH2Cl2) or Hþ (in reaction with benzene). The yields of 15
and 16 may be improved by using a non-nucleophilic base
with a high Ip in the reaction medium.
a larger experimental electrochemical potential (ΔEp
2.0 V).
≈
Calculations suggest that the ET process between 1 and
Me2S and also Me2NCHS is more favorable, and it should be
faster than that with pyridine (Table 9). Both sulfur reagents
have markedly lower Ip values relative to pyridine, and
consequently ΔG298 values for electron transfer to 1 are lower
Overall, slow decomposition of 1 through ylide 19 can be
expected in solvents such as MeNO2 with a high Ip, for which
(62) Cottrell, P. T.; Mann, C. K. J. Electrochem. Soc. 1969, 116, 1499–
1503.
(63) Andrieux, C. P.; Pinson, J. J. Am. Chem. Soc. 2003, 125, 14801–
14806.
(64) Turner, W. R.; Elving, P. J. Anal. Chem. 1965, 37, 467–469.
(65) Baizer, M. M.; Lund, H. Organic Electrochemistry, 2nd ed.; Marcel
Dekker: New York, 1983 and references therein.
(66) No stable adduct of Me2S to 1 nor to PhN2þ was located on the PES
with the DFT method.
(67) Barton, D. H. R.; Crich, D.; Motherwell, W. B. Tetrahedron 1985, 41,
3901–3924.

Article
Inorganic Chemistry, Vol. 48, No. 15, 2009 7327
Figure 14. Proposed mechanism for the formation of 10.
Figure 13. Proposed mechanism for the formation of 9a.
Summary and Conclusions
adducts to 19 are relatively stable zwitterions (do not produce
aggressive electrophiles), and have been demonstrated to
The present studies describe an alternative and reliable
method for the preparation of amine 2 and dinitrogen
derivative 1 and provide good understanding of the proper-
ties of the latter and its usefulness as a synthetic intermediate
in the preparation of more complex derivatives, including
liquid crystals. The method for the formation of amino
functionality through the COOH group permits the intro-
duction of an antipodal substituent at the stage of [closo-2-
CB9H9-2-COOH]-.16 Such compounds are being currently
investigated in our laboratory.
In many respects, dinitrogen derivative 1 displays behavior
similar to that of benzenediazonium salt; it has similar
stability, undergoes diazo coupling, reacts with sulfur nu-
cleophiles, with pyridine gives C-substituted products, and
undergoes heterolysis in MeCN, forming N-substituted acet-
amide. A detailed analysis of the experimental data and
computational results revealed two possible mechanisms
for the transformations of 1: the closed-shell zwitterionic
pathway through ylide 19 and an open-shell radical pathway
involving radical anion 26. The latter is analogous to the
nucleophile-induced decomposition of PhN2þ and operates
for reagents and solvents, which are good nucleophiles and
electron donors (low Ip and low E1/2ox). In the absence of
electron donors (reagents with high Ip and E1/2ox), dinitrogen
1 undergoes heterolytic cleavage of N2, and the resulting ylide
19 gives products of electrophilic addition. The properties
and reactivity, such as the preferential formation of the
N-adduct with pyridine, of 19 are very similar to those
observed before for the [hypercloso-1-CB11Me11] ylide.
Clearly, more kinetic experiments need to be done for a
better understanding of the reactivity of 1 and to provide
more evidence for the involvement of ylide 19 and radical
anion 26 in transformations of 1.
Experimental results demonstrated that 1 serves as a
convenient precursor for the preparation of dialkylsulfonium
derivatives such as 9b; the two-step process through 10, a
masked thiol generatedin situ, is more efficient due to a higher
yield of the first step. Also, the diazocoupling of 1, leading to
azophenols such as 8, is an efficient process, which has already
been exploited for the preparation of ionic liquid crystals.15 In
contrast, the preparation of N-substituted pyridine 11a,
which is of interest for the investigation of molecular materi-
als, is a low-yield process of little synthetic value at present.
Spectroscopic investigation indicates significant electronic
interactions between the {closo-1-CB9} cage and π substitu-
ents, which indicates a possibility of control of photophysical
properties in specifically designed molecular materials.
promote the heterolytic unimolecular decomposition of
þ 48
On the other hand, reagents with a high nucleo-
PhN2
.
philicity and low Ip, such as amines and sulfur compounds,
are likely to induce radical transformations of 1.
Analysis of electronic absorption spectra and chemical
shifts revealed significant similarities between the [closo-1-
CB9H10]- and benzene. Thus, electronic absorption spectra
demonstrated substantial electronic communication between
the {closo-1-CB9} cage and π substituents. This is consistent
with our previous experimental and theoretical studies for the
derivatives of [B10H10]2- and C2B8H10.6-8,11,68-70 In the
absence of π susbtituents, the compounds are practically
UV-transparent. Since the cage is negatively charged, the
electronic excitations of zwitterionic derivatives such as 1 and
11a involve the cage-to-substituent transitions, which exhibit
strong solvatochromism6,8 and may find applications as
NLO chromophores.
The Hammett analysis of NMR chemical shifts demon-
strated a similar ability to transmit electronic effects through
the [closo-1-CB9H10]- cage and benzene ring, which is sig-
nificantly higher than that found for the [closo-1-CB11H12]-
derivatives. The stronger antipodal effect71,72 observed in
the {closo-1-CB9} cluster relative to the {closo-1-CB11} may
be related to the smaller size and higher electron density for
the former as compared to the latter. The only other Ham-
mett correlation previously reported for boron clusters was
for 13C NMR chemical shifts of C-arylated p-carborane
derivatives.73 A comparison28 of the slope (F values) re-
ported73 for 1-aryl-p-carboranes with that for 13C NMR
chemical shifts of 4-substituted biphenyls demonstrates
that the transmission of a substituent effect through the
12-vertex carborane cage is about 70% as effective as through
the benzene ring.28 This finding is consistent with results
presented in Figure 11.
(68) Kaszynski, P. In Anisotropic Organic Materials;Approaches to Polar
Order; Glaser, R., Kaszynski, P., Eds.; ACS Symposium Series: Washington, DC,
2001; Vol. 798, pp 68-82.
(69) Pakhomov, S.; Kaszynski, P.; Young, V. G., Jr. Inorg. Chem. 2000,
39, 2243–2245.
(70) Kaszynski, P.; Kulikiewicz, K. K.; Januszko, A.; Douglass, A. G.;
Tilford, R. W.; Pakhomov, S.; Patel, M. K.; Radziszewski, G. J.; Young,
V. G., Jr. Manuscript in preparation.
(71) (a) Hermanek, S.; Plesek, J.; Gregor, V.; Stirb, B. Chem. Commun.
1977, 561–563. (b) Hermanek, S. Chem. Rev. 1992, 92, 325–362.
::
(72) Buhl, M.; Schleyer, P. v. R.; Havlas, Z.; Hnyk, D.; Hermanek, S.
Inorg. Chem. 1991, 30, 3107–3111.
(73) Fox, M. A.; MacBride, H. J. A.; Peace, R. J.; Wade, K. J. Chem. Soc.,
Dalton Trans. 1998, 401–411.

7328 Inorganic Chemistry, Vol. 48, No. 15, 2009
Ringstrand et al.
Overall, results demonstrate that dinitrogen 1 is an attrac-
tive compound for further mechanistic investigation and
synthetic applications.
Peak potentials were referenced by adding small amounts
of ferrocene to the solution, and the E-chem scan for each
compound was referenced to the ferrocene couple, which was
assumed to be 0.31 V versus SCE (MeCN, 0.2 M LiClO4).80
Preparation of [closo-1-CB9H9-1-N2] (1). A suspension of
[closo-1-CB9H9-1-NHBoc]-NEt4 (4a[NEt4], 0.21 g, 0.57 mmol)
þ
Computational Details
in a 1:3 mixture of concentrated HCl in CH3OH (10 mL) was
gently heated until all of the solid was dissolved, and stirring was
continued at room temperature for 18 h. Water (15 mL) was added,
and CH3OH was removed in vacuo. Concentrated HCl (3 mL) was
added, and [closo-1-CB9H9-1-NH3] (2) was extracted into Et2O
(3 ꢀ 15 mL). The organic layers were combined, dried (Na2SO4),
and evaporated in vacuo to give 0.120 g (166% yield) of crude
Quantum-mechanical calculations were carried out with
the B3LYP74,75 and MP2(fc)76 methods with the 6-31G(d,p)
basis set using the Gaussian 98 computational package.77
Geometry optimizations were undertaken using appropriate
symmetry constraints and tight convergence limits. Vibra-
tional frequencies were obtained with the B3LYP/6-31G(d,p)
method and were used to characterize the nature of the
stationary points and to obtain thermodynamic parameters.
Zero-point energy corrections were scaled by 0.9806.78
A population analysis for single-point calculations (MP2//
DFT) was performed using the Density keyword. For MP2-
level calculations of open-shell species, the spin-projected
energies were used for comparative studies. The IPCM solva-
tion model79 was used with default parameters.
1
amine 2 as a transparent glassy solid. H NMR: δ 0.61 (q, J =
145 Hz, 4H), 1.74 (q, J= 156 Hz, 4H), 5.46 (q, J= 156 Hz, 1H),
8.55 (s) [lit data:12 1H NMR: δ 0.81 (4H), 2.34 (4H), 5.53 (1H),
10.15 (3H)]. 13C NMR: δ 66.6. 11B NMR: δ -25.5 (d, J=141 Hz,
4B), -16.8 (d, J=152 Hz, 4B), 29.1 (d, J=159 Hz, 1B) [lit data:12
11B NMR: δ -25.8 (4B), -16.8 (4B), 29.2 (1B)]. IR: 3182 and 3162
(N-H) cm-1
.
Crude amine 2 was dissolved in a 1:1 mixture of AcOH/H2O
(1 mL), and an aqueous solution of NaNO2 (0.043 g, 0.63 mmol)
in H2O (2 mL) was added dropwise (∼1 drop/sec) at 0 °C. The
reaction temperature was maintained for 30 min. H2O (2 mL)
was added, and a precipitate was filtered and dried, giving 0.066 g
(80% yield) of [closo-1-CB9H9-1-N2] (1) as a white crystalline
solid: dec ∼87 °C (ΔH=148 kJ/mol, DSC). 1H NMR (CD3CN):
δ 0.96 (q, J=150 Hz, 4H), 2.51 (q, J=167 Hz, 4H), 6.65 (q, J=
164 Hz, 1H). 1H NMR (benzene-d6): δ 1.10-3.20 (m, 8H), 7.87
(q, J=170 Hz, 1H). 11B NMR (CD3CN): δ -22.1 (d, J=149 Hz,
4B), -8.6 (d, J=167 Hz, 4B), 51.3 (d, J=167 Hz, 1B). 11B NMR
(benzene-d6): δ -20.6 (d, J=147 Hz, 4B); -6.8 (d, J=163 Hz,
4B), 57.0 (d, J=179 Hz, 1B). IR: 2250 (N2) cm-1. UV (CH3CN):
Experimental Details
Reagents and solvents were obtained commercially. Sol-
vents were dried and deoxygenated before use, and reagents
were used as supplied. ZnCl2 was dried by heating at ∼100 °C
underavacuum. ReactionswerecarriedoutunderdryArand
subsequent manipulations conducted in the air. NMR spec-
tra were obtained at 128.4 MHz (11B), 100.6 MHz (13C), and
400.1 MHz (1H) in CD3CN unless otherwise specified. H
1
NMR and 13C NMR spectra were referenced to the solvent.
11B NMR chemical shifts are relative to the resonance of
an external boric acid sample in CH3OH that was set to
18.1 ppm. IR spectra were recorded in the solid state using an
AT-IR accessory.
λ
max 250 nm (log ε = 3.82).
þ
Preparation of [closo-1-CB9H9-1-NHBoc]-þ NEt4 (4a). A
UV spectra were recorded in UV-grade CH3CN. All com-
pounds were at a concentration of 0.7-5ꢀ10-5 M. Extinction
coefficients were obtained by fitting the maximum absor-
bance against concentration in agreement with Beer’s law.
Electrochemicalanalysis wasconducted indryMeCNwith
Bu4NPF6 (0.05 M) as the supporting electrolyte and using a
freshly polishedglassy carbon working electrode with the Ag/
AgCl reference electrode. The scanning rate was 1 V/s.
Following literature recommendations,63 the concentrations
suspension of [closo-1-CB9H9-1-COOH]-NEt4 (3[NEt4], 1.14 g,
3.88 mmol) [11B{1H} NMR: δ -23.6 (4B), -15.4 (4B), 34.2 (1B);
IR 1678 (CdO) cm-1] in anhydrous CH2Cl2 (10 mL) was treated
with 2 M (COCl)2 in CH2Cl2 (2.2 mL, 4.3 mmol). Vigorous
bubbling of CO and CO2 gases was observed, followed by the
dissolution of the substrate and the formation of a slight yellow
solution. The solution was stirred for 75 min at room temperature,
filtered to remove insoluble particulates, and the solvent removed
in vacuo to give 1.20 g of crude [closo-1-CB9H9-1-COCl]-NEt4þ
(5[NEt4]) as a white solid. 11B {1H} NMR: δ -23.1 (4B), -13.5
þ
of compounds 1 and PhN2 were set at 0.6 mM, and the
reported data were taken from the first scan. Continuous
scanning of the electrochemical window (þ0.7 to -1.0 V)
resulted in shifting of the observed cathodic peak to
more negative potentials due to electrode passivation by
the electrografting of Ph• or [closo-1-CB9H9] radicals.
(4B), 38.5 (1B). IR: 1776 (CdO) cm-1
.
Crude [closo-1-CB9H9-1-COCl]-NEt4þ (5[NEt4], 1.20 g, 3.85
mmol) was dissolved in anhydrous CH2Cl2 (10 mL) and added
via syringe to solid anhydrous ZnCl2 (0.052 g, 0.38 mmol) under
a N2 atmosphere. The reaction mixture was cooled to 0 °C, and
Me3SiN3 (0.55 mL, 4.2 mmol) was added. The reaction mixture
was stirred at 0 °C for an additional 30 min, after which it was
warmed to room temperature and stirred for 4 h. The reaction
mixture was poured into ice-cold H2O (50 mL) and extracted
with CH2Cl2 (3ꢀ30 mL). The organic layers were combined,
dried (MgSO4), and filtered, and the solvent was removed
in vacuo, giving 1.11 g of [closo-1-CB9H9-1-CON3]-NEt4þ
(6[NEt4]) as a white crystalline solid. 11B{1H} NMR: δ -23.3
(4B), -14.8 (4B), 36.4 (1B). IR 2360, 2339 and 2142 (N3), 1691
(CdO) cm-1. The crude product was contaminated with about
15% carboxylic acid 3[NEt4].
(74) Becke, A. D. J. Chem. Phys. 1993, 98, 5648–5652.
(75) Lee, C.; Yang, W.; Parr, R. G. Phys. Rev. B 1988, 37, 785–789.
(76) (a) Moller, C.; Plesset, M. S. Phys. Rev. 1934, 46, 618–622.
e
(b) Head-Gordon, M.; Pople, J. A.; Frisch, M. J. Chem. Phys. Lett. 1988, 153,
503–506.
(77) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb,
M. A.; Cheeseman, J. R.; Zakrzewski, V. G.; Montgomery, J. A., Jr.;
Stratmann, R. E.; Burant, J. C.; Dapprich, S.; Millam, J. M.; Daniels,
A. D.; Kudin, K. N.; Strain, M. C.; Farkas, O.; Tomasi, J.; Barone, V.; Cossi,
M.; Cammi, R.; Mennucci, B.; Pomelli, C.; Adamo, C.; Clifford, S.;
Ochterski, J.; Petersson, G. A.; Ayala, P. Y.; Cui, Q.; Morokuma, K.;
Malick, D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.;
Cioslowski, J.; Ortiz, J. V.; Baboul, A. G.; Stefanov, B. B.; Liu, G.;
Liashenko, A.; Piskorz, P.; Komaromi, I.; Gomperts, R.; Martin, R. L.;
Fox, D. J.; Keith, T.; Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.;
Challacombe, M.; Gill, P. M. W.; Johnson, B.; Chen, W.; Wong, M. W.;
Andres, J. L.; Gonzalez, C.; Head-Gordon, M.; Replogle, E. S.; Pople, J. A.
Gaussian 98, revision A.9; Gaussian, Inc.: Pittsburgh, PA, 1998.
Crude [closo-1-CB9H9-1-CON3]-NEt4þ (6[NEt4], 1.11 g, 3.49
mmol) was dissolved in anhydrous CH3CN (15 mL) and re-
fluxed for 1 h. The reaction was cooled to room tmeperature, the
solvent removed, and the residue dried in vacuo, giving 1.11 g
of crude [closo-1-CB9H9-1-NCO]-NEt4þ (7[NEt4]) as a slight
(78) Scott, A. P.; Radom, L. J. Phys. Chem. 1996, 100, 16502–16513.
(79) Foresman, J. B.; Keith, T. A.; Wiberg, K. B.; Snoonian, J.; Frisch,
M. J. J. Phys. Chem. 1996, 100, 16098–16104.
(80) Bard, A. J.; Faulkner, L. R. Electrochemical Methods: Fundamentals
and Applications, 2nd ed.; Wiley: Hoboken, NJ, 2001.

Article
Inorganic Chemistry, Vol. 48, No. 15, 2009 7329
yellow solid. 11B{1H} NMR: δ -26.1 (4B), -16.0 (4B), 25.1
(1B). IR: 2262 (NdCdO).
(CdO) cm-1. Anal. Calcd for C14H39B9N2O2: C, 46.10; H,
10.78; N, 7.38. Found: C, 46.26; H, 10.76; N, 7.53.
A solution of anhydrous tert-butanol (15 mL), anhydrous
CH3CN (5 mL), and crude [closo-1-CB9H9-1-NCO]-NEt4þ
(7[NEt4], 1.04 g, 3.58 mmol) was stirred at 90 °C for 2 h, after
which solvents were removed, leaving 1.20 g of crude [closo-1-
CB9H9-1-NHBoc]-NEt4þ (4a[NEt4]) as a yellow solid. The
crude solid was dissolved in CH2Cl2 and passed through a silica
gel plug buffered with 1% NEt3 in CH2Cl2. Elution with a
buffered CH3CN/CH2Cl2 solution (1% NEt3, 20% CH3CN,
79% CH2Cl2) afforded 0.68 g (48% yield based on starting acid
3[NEt4]) of pure [closo-1-CB9H9-1-NHBoc]-NEt4þ (4a[NEt4])
Subsequent fractions produceþd 0.56 g of a 30:70 mixture of
[closo-1-CB9H9-1-COOH]-NEt4
(3[NEt4]) and [closo-1-
CB9H9-1-NH2]-NEt4 (2[NEt4]). 1H NMR: δ 1.23 (t, J =
7.2 Hz, 12H), 3.19 (q, J=8.0 Hz, 8H), 7.31 (s, 2H). 11B {1H}
NMR: δ -26.1 (4B), -17.1 (4B), 24.1 (1B), [lit.12 data for
þ
1
(2[NHEt3]): H NMR: δ 0.91 (4H), 1.69 (4H), 5.40 (1H), 8.52
(2H). 11B NMR: δ -25.8 (4B), -17.2 (4B), 26.0 (1B)].
Acknowledgment. Financial support for this work was
received from the National Science Foundation (DMR-
0111657 and DMR-0606317). We thank Mr. Andrzej Balinski
for his technical assistance with the isolation of 11b[H].
1
as a light yellow solid: mp 140 °C. H NMR: δ 0.45 (q, J =
143 Hz, 4H), 1.19 (t, J=7.3 Hz, 12H), 1.49 (s, 9H), 1.67 (q, J=
156 Hz, 4H), 3.14 (q, J=7.3 Hz, 8H), 5.07 (q, J=151 Hz, 1H),
7.12 (s, 1H). 13C NMR: δ 7.7 (NþCH2CH3), 28.6 (C(CH3)3),
47.3 (C(CH3)3), 53.0 (NþCH2CH3), 156.0 (CdO). Carbon
associated with the {closo-1-CB9} cluster was not observed.
Minor signals at 9.0 and 53.5 ppm were attributed to excess
NEt4þBr-. 11B NMR: δ -26.2 (d, J=135 Hz, 4B), -17.2 (d, J=
152 Hz, 4B), 23.8 (d, J=158 Hz, 1B). IR: 3505 (N-H), 1737
Supporting Information Available: Experimental procedures
for the reactions of 1 and analytical data for products, COSY
spectra for 11c[H] and 11d[H], kinetic data for the decomposi-
tion of 1, Hammett correlation data, and details of computa-
tional results. This material is available free of charge via the
Internet at http://pubs.acs.org.