Bronsted acid-catalyzed skeletal rearrangements in polycyclic conjugated boracycles: A thermal route to a ladder diborole

Article abstract of DOI:10.1039/c4sc01201a

The dibenzodiboracyclobutylidene 1 was recently shown to undergo photochemical rearrangement to the more thermodynamically stable dibenzodiborapentalene 2. In order to discover a more efficient thermal route for this isomerization, the redox chemistry of both 1 and 2 was explored. Radical anions can be formed and characterized by EPR spectroscopy by one electron reduction using potassium graphite. The [Cp*2Co]⁺ salt of the radical anion of 2 was characterized by X-ray crystallography. The dipotassium salts 1K₂ and 2K₂ can be prepared by using two equivalents of potassium naphthalenide; both were fully characterized. It was found that smooth rearrangement of 1K₂ to 2K₂ is mediated by weak Bronsted acids such as tert-butanol or 2,6-dimethylphenol; a mechanism based on similar rearrangements in the isoelectronic neutral all-carbon framework is proposed. the Partner Organisations 2014.

Full text of DOI:10.1039/c4sc01201a

Chemical
Science
View Article Online
View Journal | View Issue
EDGE ARTICLE
Bronsted acid-catalyzed skeletal rearrangements in
polycyclic conjugated boracycles: a thermal route
to a ladder diborole†
Cite this: Chem. Sci., 2014, 5, 3189
*
Juan F. Araneda, Warren E. Piers, Michael J. Sgro and Masood Parvez
The dibenzodiboracyclobutylidene 1 was recently shown to undergo photochemical rearrangement to the
more thermodynamically stable dibenzodiborapentalene 2. In order to discover a more efficient thermal
route for this isomerization, the redox chemistry of both 1 and 2 was explored. Radical anions can be
formed and characterized by EPR spectroscopy by one electron reduction using potassium graphite. The
[Cp*2Co]+ salt of the radical anion of 2 was characterized by X-ray crystallography. The dipotassium salts
1K2 and 2K2 can be prepared by using two equivalents of potassium naphthalenide; both were fully
characterized. It was found that smooth rearrangement of 1K2 to 2K2 is mediated by weak Bronsted
acids such as tert-butanol or 2,6-dimethylphenol; a mechanism based on similar rearrangements in the
isoelectronic neutral all-carbon framework is proposed.
Received 25th April 2014
Accepted 22nd May 2014
DOI: 10.1039/c4sc01201a
www.rsc.org/chemicalscience
borole ring on the photophysical properties of the molecule
when introduced into an extended p system.25,26 Recently, we
Introduction
reported the synthesis of a ladder diborole (2) as a new member
of the heteroacene family.27 Compound 2 was prepared by a
light-driven photoisomerization of 1, which is the kinetic
product in the KC8 reduction of a 2-boryl substituted diaryl
acetylene precursor.27 While small quantities of pure 2 can be
obtained in this way, attempts to garner workable amounts of 2
on a preparative scale using this procedure were hampered by
difficulties in driving the reaction to completion while avoiding
photochemical decomposition of the product 2 because of the
prolonged irradiation required.
Therefore, we focused our attention on the discovery of a
thermal pathway for the isomerization of 1 to 2. It has been
reported that the carbon analogue of compound 1 (1,10-
di(benzocyclobutylidene), II, Scheme 1) undergoes a facile
rearrangement in acetic anhydride containing sulfuric acid to
In the last two decades, conjugated polycyclic hydrocarbons
have garnered considerable interest in the area of organic
electronics.1 Amongst these organic semiconductors, penta-
cene's high hole mobility has made it a standard in the eld.1–3
However, its poor solubility in common organic solvents and its
reduced light and oxygen stability has limited its applicability.4
Incorporation of main group elements into these conjugated
polycyclic acene hydrocarbons can modify their photophysical
properties and provide enhanced stability.5,6 In particular,
molecules containing boron are promising candidates for use
as organic light emitting diodes,7–9 sensors10 and eld effect
transistors11,12 because the empty p orbital can participate in the
delocalization of electrons.13–15 Furthermore, the exchange of
carbon for the more electropositive boron atom can also alter
the electronic characteristics of a compound signicantly. For
instance, 9-boraanthracene and higher boraacenes show a
considerably smaller HOMO–LUMO gap compared to the all-
carbon analogues.16,17 Among boron containing heterocycles,
boroles have attracted signicant attention in the past few
years,18–24 and their incorporation into heteroacene frameworks
is an emerging area of research. For example, Yamaguchi and
co-workers have prepared a series of heteroacene-fused ladder
boroles (I, Fig. 1) and provided insight into the effect of the
Department of Chemistry, University of Calgary, 2500 University Drive NW, Calgary,
Alberta, T2N 1N4, Canada. E-mail: wpiers@ucalgary.ca; Tel: +1-403-220-5746
† Electronic supplementary information (ESI) available: Additional experimental
and spectroscopic details and the full citation for ref. 32. CCDC 997321–997324.
For ESI and crystallographic data in CIF or other electronic format see DOI:
10.1039/c4sc01201a
Fig. 1 Fused-boroles and related compounds.
Chem. Sci., 2014, 5, 3189–3196 | 3189
This journal is © The Royal Society of Chemistry 2014

View Article Online
Chemical Science
Edge Article
moisture sensitive dark purple crystals by slow diffusion of
hexanes into a THF solution of the product. Likewise, addition
of one equivalent of potassium graphite to a THF solution of 2
affords air and moisture sensitive deep blue crystals of 2K aer
crystallization using the same solvent mixture. Despite a myriad
of attempts to obtain X-ray quality crystals of 1K and 2K, it has
not been possible so far, due mainly to the poor thermal
stability of the compounds over the time frame necessary to
form quality crystals. Nonetheless, radical anion formation was
conrmed by electron paramagnetic resonance (EPR) spec-
troscopy, which in both cases, showed broad featureless reso-
nances at giso ¼ 2.00279 for 1K and giso ¼ 2.00248 for 2K. These
giso values are in the range expected for organic p radicals31,32
and no hyperne coupling could be resolved in either case, even
at low temperatures. However, in the case of 1K, clear hyperne
coupling was observed in the EPR spectrum when the radical
was generated in situ by addition of 1 to a THF solution of the
dianion species 1K2 (see below). An analogous comproportio-
nation experiment using 2 and 2K2 failed to reveal a resolvable
hyperne splitting pattern in this case.
Due to the complicated nature of the EPR spectrum of 1K
and the lack of information from the spectrum of 2K the
structures of both compounds were modelled by DFT using the
UB3PW91 functional with the 6-31+G(d) basis set.33 The hyper-
ne coupling constants (hfcc) were calculated using the basis
set EPR-II,34 and were used as the initial values for the simula-
tion of the EPR spectrum using the program PEST Winsim.30
The computed hyperne coupling constants are in reasonable
agreement with those obtained from simulation of the experi-
mental spectrum (Table 1), except for a(11B).35,36 The hyperne
coupling constants a(11B) in 1K and 2K of 1.08 G and 0.55 G,
respectively, are lower than those observed in other organic
Scheme 1 Proton catalysed isomerization of 1,10-bi(benzocyclobu-
tylidene) (II).
5,10-dihydroindeno[2,1-a]indene, III (ref. 28) via a sequence of
1,2-carbenium ion shis (Scheme 1). Accordingly, we attempted
to catalyse the analogous isomerization of compound 1 using
weaker acids (both Bronsted and Lewis), to no avail. In fact,
when water was added to 1, preferential ring opening of the
boracycle took place, forming (E)-1,2-bis(hydroxy(tri-isopropyl)
boryl)-diphenylethene (IV, Fig. 1).27 Given the propensity of B–C
bonds to undergo protic cleavage, this outcome is perhaps not
too surprising.
Since compound 1 is electron decient relative to the all-
carbon analogue II, we postulated that its dianion, 1K2, might
undergo acid catalysed isomerization as observed for the all-
carbon system; subsequent oxidation of compound 2K2 could
be an alternative way to prepare compound 2. In this article we
discuss the redox chemistry of neutral boron heterocycles 1 and
2, and the successful isomerization of the dipotassium salt of
the dianion of 1 (1K2) to dibenzodiboratapentalene (2K2) as
catalysed by weak Bronsted acids.
Results and discussion
To assess the propensity of compounds 1 and 2 to undergo
reductions, cyclic voltammetry experiments were performed in
THF using Bu4NPF6 as the supporting electrolyte; the potentials
were referenced against the ferrocene/ferrocenium ion couple.
Compounds 1 and 2 are both reduced by one and two electrons
in sequential one-electron processes (see ESI†). Compound 1
exhibits irreversible reduction waves; the rst reduction
potential was observed at ꢀ2.47 V and the second at ꢀ2.89 V.
Conversely, ladder diborole 2 shows a reversible reduction wave
at ꢀ1.51 V followed by a second, irreversible, reduction wave at
ꢀ2.42 V.29 Both reduction potentials of 2 are considerably less
negative than the ones observed for heteroacene-fused boroles I
(E1/2 ¼ ꢀ1.89 V to ꢀ2.25 V for the rst reduction potential and
Epc ¼ ꢀ2.78 V to ꢀ3.04 V for the second reduction potential).25
These results suggest that it is possible to reduce both
compounds by one and/or two electrons, but powerful reducing
agents, such as alkali metals, are required—especially when
reducing compound 1.
Accordingly, addition of one equivalent of potassium
graphite to a THF solution of 1 resulted in an immediate colour
change of the solution from bright yellow to dark violet
(Scheme 2). Compound 1K was crystallized as extremely air and Scheme 2 Synthesis of radial anions 1K, 2K and 2Co.
3190 | Chem. Sci., 2014, 5, 3189–3196
This journal is © The Royal Society of Chemistry 2014

View Article Online
Edge Article
Chemical Science
Table 1 Calculated and simulated |hfcc| of 1K and 2K (in gauss)
boron are essentially perpendicular to the ligand framework,
hyperne coupling to the iso-propyl methyne protons, HV,
which are located above and below the boron center, is present.
Yamaguchi has demonstrated the interaction of these protons
with a planar p system and has used this feature to assess the
antiaromaticity in thiophene-fused ladder boroles.40
In an effort to structurally characterize the radical anions in
1K or 2K, exchange of the potassium counterion for the bis-
(triphenylphosphine)iminium (PPN+) cation via metathesis was
attempted. In both cases, decomposition was the outcome of
these experiments. However, since the rst reduction potential
of compound 2 is considerably lower than that of 1, it was
possible to carry out one electron reduction of 2 using the
milder reducing reagent bis-(pentamethylcyclopentadienyl)-
1K
2K
Element
# Nuclei
Calc.a
Sim.b
Calc.
Sim.
0.55
*
cobalt(II) (Cp2Co) in THF or CH2Cl2.41 Addition of one equiva-
11Bc
1Hz
1Hy
1Hx
1Hy
1Hv
2
2
2
2
2
4
3.75
2.54
0.89
1.41
2.12
0.70
1.08
3.72
0.89
1.80
1.81
0.70
1.87
0.89
0.30
1.10
0.01
0.49
*
0.98 lent of Cp2Co to a dichloromethane solution of 2 immediately
0.26
1.21
0.01
0.41
produced the characteristic deep blue colour observed in 2K. In
the freezer at ꢀ30 ꢁC the solvent was slowly removed by vapour
diffusion into toluene and deep blue X-ray quality crystals were
deposited aer several days. The molecular structure of 2Co is
shown in Fig. 3. When compared to its neutral counterpart,27
compound 2Co showed a lengthening of the C1–C10 bond
a
Geometries were optimized using UB3PW91/6-31+G(d) and the hfcc
b
were calculated using UB3PW91/EPR-II.
Simulated using PEST
Winsim30 to >99% correlation. c 10B was not included in the simulation.
˚
˚
(1.410(3) A for 2Co vs. 1.367(5) A for 2) and a shortening of the
˚
˚
B1–C1 bond (1.524(4) A for 2Co vs. 1.571(4) A for 2). These bond
length changes are also consistent with the nature of the SOMO
boron containing p-radicals (usually between 3.5 and 6 G),31,37–39 as revealed by the DFT computations (Fig. 2). The SOMO
perhaps an indication of the larger degree of delocalization of contains a node in the C1–C10 bond and is p bonding along the
the unpaired electron throughout the p system (Fig. 2). Inter- B1–C1 vector. The largest deviation from the mean plane
estingly, although the tri-iso-propylphenyl (Tipp) groups on dened by the core B2C6 atoms (fused borole rings) amounts to
˚
0.032(3) A, showing that the p system in 2Co remains planar
aer reduction. This is in contrast to the planarized triarylbor-
ane radical anion reported by Yamaguchi37 and the dibenzo(a,e)
pentalene radial anion reported by Saito,42 which both adopt a
shallow bowl-shaped conformation aer reduction with potas-
sium, and reects the high degree of delocalization of the
unpaired electron throughout the B2C14 framework of the
ladder diborole.
Fig. 3 Thermal ellipsoid (30%) diagram of the molecular structure of
2Co. Selected bond lengths [A˚], angles [ꢁ], and dihedral angles [ꢁ]: C1–
Fig. 2 Top: overlap of the experimental (black) and simulated (red)
B1 1.524(4), C1–C10 1.410(3), C2–B1 1.586(4), C2–C3 1.420(3), C3–C10
EPR spectrum of 1K0 (left) and 2K0 (right) in a THF solution at 298 K.
1.470(4); C2–B1–C1 102.3(2); C2–B1–C8–C13–68.1(4), C4–C3–C10–
C1 179.4(3). Hydrogen atoms have been removed for clarity.
Simulations performed in PEST Winsim30 to >99% correlation. Bottom:
SOMO (UB3PW91/6-31+G(d)) of 1K0 (left) and 2K0 (right).
This journal is © The Royal Society of Chemistry 2014
Chem. Sci., 2014, 5, 3189–3196 | 3191

View Article Online
Chemical Science
Edge Article
Having demonstrated facile one electron reduction of both
compounds 1 and 2, we turned to preparing the dianionic
species that would be isoelectronic to the all-carbon frameworks
II and III shown in Scheme 1. A freshly prepared THF solution of
potassium naphthalenide (2 eq.) was slowly added into a solu-
tion of the respective starting materials 1 and 2 in THF at
ꢀ78 ꢁC. Aer stirring the solutions for an additional 2 hours at
room temperature, the solvent was evaporated and the naph-
thalene removed by sublimation under high vacuum.
Compound 1K2 was isolated as a dark brown solid aer crys-
tallization in THF and hexanes in 60% yield, and compound 2K2
as red crystals from the same solvent mixture in 55% yield
(Scheme 3).
The 1H NMR spectrum of compound 1K2 isolated in this way
is intriguing. All the signals are broad and it is not possible to
distinguish any ne structure due to coupling with adjacent
protons. Furthermore, the chemical shis of the aromatic
protons (except for the resonance belonging to the Tipp group)
are shied considerably upeld (between 5.5 and 4.2 ppm) from
where they are expected to resonate. Since the second reduction
potential of compound 1 is very negative, we surmised that 1K2
might be contaminated with small amounts of the radical anion
Fig. 4 Thermal ellipsoid (30%) diagram of the molecular structure of
1K2. Selected bond lengths [A˚], angles [ꢁ], and dihedral angles [ꢁ]: C1–
B1 1.504(3), C1–C10 1.437(4), C7–B1 1.629(4), C7–C2 1.437(3), C2–C1
1.492(3); C7–B1–C1 87.49(12); C1–B1–C8–C13 ꢀ56.2(4). Hydrogen
atoms and THF molecules have been removed for clarity.
1
1K, resulting in the broadness observed in the H NMR spec-
trum. EPR spectroscopy on isolated samples of 1K2 conrmed mean plane dened by the B2C6 core atoms amounts to 0.010(3)
˚
the presence of 1K; furthermore, augmentation of the signal A, showing that compound 1K2 remains planar aer dir-
was observed upon addition of small aliquots of 1 to the EPR eduction. The boron center adopts a distorted trigonal planar
sample (see Fig. S2 in the ESI†). In contrast to the broadening geometry, with a small C7–B1–C1 angle of 87.49(12)ꢁ due to the
observed in the 1H NMR spectrum, in the 13C NMR spectrum of constrained four membered ring. There are two potassium
1K2 minimal broadening was apparent, and all the resonances atoms situated on opposite sides of the p system, coordinating
are clearly identied. Even the resonances for the carbons to the bay regions on either side of the central C1–C10 bond. The
attached to the boron center appear as small, broad peaks, their distance between the potassium centers and the B1–C1–C10–B10
identity conrmed by 1H–13C Heteronuclear Multiple-Bond plane is 2.623 A. There are three molecules of THF coordinated
˚
Correlation (HMBC) spectroscopy. The 11B NMR spectrum to each potassium center (not shown in Fig. 4); however, only
shows a broad resonance at 38.7 ppm and it is comparable to one solvent molecule per metal remains coordinated when the
the one reported for 9-boraanthracene (39.7 ppm).16 This signal crystals are dried under vacuum based on 1H NMR spectroscopy
is considerably upeld shied compared to the resonance of 1 and elemental analysis.
(70.1 ppm), which is consistent with a more electron rich boron
center.
While the bond angles within the molecular core do not
change signicantly upon reduction, the bond distances in the
X-ray analysis of single crystals of 1K2 reveals the structure of diboratacyclobutylidene p system change signicantly. For
the dianionic species (Fig. 4). The largest deviation from the example, in neutral starting compound 1, the B1–C1 bond
distance is 1.581(5) A,27 which is indicative of a B–C single bond,
˚
but in compound 1K2 the B1–C1 bond distance is considerably
˚
shorter (1.504(3) A) suggesting double bond character. Similar
B]C values have been reported for 9-phenyl-9-borataan-
43
16
˚
˚
27
thracene (1.532(3) A) and 9-boraanthracene (1.507(4) A).
0
˚
Conversely, the C1–C1 distance in neutral 1 (1.330(7) A) is
consistent with a C]C double bond, but in 1K2 the distance is
2
˚
considerably longer (1.437(3) A) which is indicative of a Csp –
Csp2 single bond. These observations are again completely
consistent with the characteristics of the LUMO of 1 (ref. 27)
and the SOMO of 1K as revealed by computations.
Unlike the 1H NMR spectrum of 1K2, there is no broadening
of the signals in the spectrum of 2K2 and all of the coupling
between neighbouring hydrogen atoms is observed. Further-
more, all the resonances appear in the expected region of the
spectrum. The boron chemical shi appears at 32.1 ppm, very
Scheme 3 Synthesis of 1K2 and 2K2.
3192 | Chem. Sci., 2014, 5, 3189–3196
This journal is © The Royal Society of Chemistry 2014

View Article Online
Edge Article
Chemical Science
close to 1K2 (38.7 ppm) and is again considerably upeld shied
compared to the starting material 2 (68.6 ppm).27
X-ray crystal structure analysis of compound 2K2 reveals a
sandwich complex (Fig. 5), with the potassium atoms situated
just above and below of the center of the pentalene framework
in a centrosymmetric arrangement. The largest deviation from
0
Fig. 6 Compounds 2K2 and V used for NICS calculations.
˚
the mean B2C6 plane is only 0.010(3) A. The separation between
˚
the two potassium atoms measures 5.506 A, which is similar to
the value found in a B4N2C2 sandwich complex (5.52 A).44 In the
˚
X-ray structure there are two THF molecules per potassium than benzene itself (NICS(0) ¼ ꢀ9.7). Interestingly, dibenzo-
center (not shown in Fig. 5), but only one solvent molecule pentalene V possesses a stronger antiaromatic character on the
remains coordinated when the crystals are dried under vacuum. central rings (NICS(0) ¼ 10.4, NICS(1) ¼ 5.8), and a stronger
0
˚
Similar to compound 1K2, the C1–B1 (1.479(4) A) and C1–C1
aromatic character on the benzene rings (NICS(0) ¼ ꢀ2.6,
˚
(1.466(3) A) bond distances are indicative of a double and single NICS(1) ¼ ꢀ5.6). This suggests that dibenzopentalene has a
bond character respectively. As anticipated, the C1–C10 bond more localized p system than 2K2; nonetheless, it is clear that
distance in 2Co is of intermediate length between the C1–C10 these two isoelectronic p systems are closely related.
27
˚
double bond shown in 2 (1.367(5) A) and the single bond in
With dianions 1K2 and 2K2 in hand, we set out to test our
˚
2K2 (1.466(3) A). Likewise, the B1–C1 bond distance is between initial hypothesis that 1K2 might be thermally isomerized to the
that of the B1–C1 single bond for 2 (1.571(4) A)27 and the double more stable 2K2 using Bronsted acid catalysis. Upon dissolution
˚
˚
bond in 2K2 (1.479(4)) A. This clearly illustrates how changing of compound 1K2 in degassed “wet” THF-d8, the solution
the electronic structure of the molecule upon one and two immediately changed from dark brown to deep red; the 1H NMR
electron reductions impacts the physical structure of the p spectrum clearly shows formation of substantial quantities of
system in 2.
2K2, along with other hydrolysis products. The reaction was
Whereas the ladder diborole 2 is a 14 p aromatic system, therefore performed under more controlled conditions in dry
introduction of another electron pair should give the 16 p solvents using various measured amounts of 2,6-dimethylphenol
electron dianion 2K2 antiaromatic character. To evaluate this as the proton source. Aer optimization of the conditions, it was
notion, we carried out NICS calculations (B3LYP/6-311+G(d,p))45 determined that a 20% loading of 2,6-dimethylphenol at 75 ꢁC in
on the dianion of 2K2 and dibenzopentalene V (Fig. 6).46 The THF solution gave compound 2K2, which could be isolated in
central ve-membered rings possess antiaromatic character 84% yield aer stirring for 8 hours under these conditions.
(NICS(0) ¼ 6.1, NICS(1) ¼ 1.3) and the anking benzene rings However, measureable amounts of a contaminant were always
(NICS(0) ¼ ꢀ0.9, NICS(1) ¼ ꢀ3.9) are considerably less aromatic present and this species was determined to be the doubly
protonated neutral compound 3, which can be separately
synthesized by treating 2K2 with stoichiometric quantities of the
acid. Thus, when two equivalents of 2,6-dimethylphenol are
added to a THF solution of 2K2, the solution changes from red to
light yellow, forming the air and moisture stable 3 (Scheme 4).
Aer removal of the solvent and extraction/crystallization from
hexanes, this compound was isolated as a white solid in 72%
1
yield. The H NMR spectrum only shows formation of only one
isomer; DFT calculations (M062X/6-311+G(d,p)//B3LYP/6-
31G(d)+ZPE)47 show that the cis isomer is 17.6 kcal molꢀ1 more
stable than the trans isomer, and further characterization via
NMR spectroscopy and X-ray crystallography indicates that it is
the cis isomer that is formed exclusively (Fig. 7).
The bridgehead hydrogens alpha to boron appear as a singlet
at 4.16 ppm and the downeld 11B chemical shi of 77.1 is
consistent with a three coordinate borane moiety. The patterns
observed for the iso-propyl groups on the Tipp group in the 1H
NMR suggest that the planarity of the system has been
breached, since six sets of signals for the iso-propyl methyl
protons and three multiplets for the iso-propyl methyne protons
are now apparent. One of these methyne resonances is signi-
Fig. 5 Thermal ellipsoid (30%) diagram of the molecular structure of
cantly upeld shied compared to the other two (1.29 ppm vs.
2.93 and 2.65 ppm), due to the situation of one of the iso-propyl
2K2. Selected bond lengths [A˚], angles [ꢁ], and dihedral angles [ꢁ]: C1–
B1 1.479(4), C1–C10 1.466(3), C3–B1 1.586(4), C2–C3 1.408(3), C2–C10
groups under the anking benzene rings (see Fig. 7). The methyl
1.479(3); C1–B1–C3 102.2(2); C3–B1–C8–C13 ꢀ67.7(3), C7–C2–C1–
C10 179.5(2). Hydrogen atoms and THF molecules have been removed
for clarity.
signals of this iso-propyl group are also shielded due to the
inuence of the benzene ring.
This journal is © The Royal Society of Chemistry 2014
Chem. Sci., 2014, 5, 3189–3196 | 3193

View Article Online
Chemical Science
Edge Article
Scheme 5 Proposed mechanism for the rearrangement of 1K2.
Attempts to doubly deprotonate compound 3 using potas-
sium 2,6-dimethyl phenoxide to regenerate 2K2 did not succeed,
indicating protonation by phenol is irreversible. However,
addition of potassium tert-butoxide to 3 did generate 2K2
essentially quantitatively, and this salt was stable in the pres-
Scheme 4 Isomerization of 1K2 to 2K2, synthesis of 3 and generation
of 2 via oxidation of 2K2.
t
ence of BuOH. Thus, the more weakly acidic alcohol was used
in excess to convert 1K2 to 2K2 on a preparative scale, avoiding
contamination by 3 and providing 2K2 in excellent yield. As
shown in Scheme 4, the dianion can be oxidized using silver
triate to furnish 2 in 65% yield upon work up, successfully
completing a thermal synthesis of 2 from 1 as planned.
This type of isomerization is new for a boron containing
heterocycle, but we have proposed an analogous mechanism to
the one proposed for the carbon analogue (II). However, instead
of a carbocation as the key species in the rearrangement, the
Lewis acidic boron center triggers the bond migrations that
occur upon protonation of the framework (Scheme 5). Inter-
mediate IA is formed aer protonation of one of the benzylic
carbons in 1K2 by the alcohol. Nucleophilic attack on the borane
by the electron rich carbanion of the other bor-
abenzocyclobutene induces ring opening to generate IB, posi-
tioning the carbanion for attack on the other boron center to
expand the remaining four membered ring, giving IC, which is
monoprotonated 2K2. When the acid is a phenol, this species
can be irreversibly protonated to give 3; thus we surmise that IC
is itself acidic enough to protonate 1K2 or phenoxide to induce
further ring expansion rearrangements, or the reaction would
stop at this point in the phenol catalysed processes. For the
tBuOH mediated process, this is not an issue, since the alcohol
is not acidic enough to protonate 2K2. Attempts to generate IC
through protonation of 2K2 with one equivalent of a strong acid
such as HNTf2 were not successful, producing only half an
equivalent of 3; this indicates that IC is a very reactive species,
both as an acid and a base.
Fig. 7 Thermal ellipsoid (30%) diagram of the molecular structure of 3.
Selected bond lengths [A˚], angles [ꢁ], and dihedral angles [ꢁ]: C1–B1
1.589(7), C1–C10 1.551(6), C3–B1 1.548(7), C2–C3 1.388(6), C2–C10
1.508(6); C1–B1–C3 105.4(4), B1–C1–C9 115.4(4); C2–C10–C1–C9
131.2(4), C3–B1–C8–C13 95.7(6). Hydrogen atoms (except for the
central protons) have been removed for clarity.
The X-ray structural analysis (Fig. 7) conrms that the poly-
cyclic ring framework is signicantly bent as a result of the
addition of the two hydrogen atoms. C1 adopts a distorted
tetrahedral geometry, with B1–C1–C9, C9–C1–C10, C10–C1–B1
angles of 115.2(4)ꢁ, 105.4(3)ꢁ and 105.5(4)ꢁ respectively, and the
torsion angle C2–C10–C1–C9 is 131.2(4)ꢁ. The C1–C10 bond
Conclusions
˚
distance is considerably longer (1.551(6) A) than for 2 (1.367(5)
3
3
˚
˚
A) and 2K2 (1.466(3) A), reecting the Csp –Csp single bond We have described the synthesis, characterization and reactivity
character.
studies of 1K2, which was synthesized by direct reduction of 1
with potassium naphthalenide. In the course of these studies,
3194 | Chem. Sci., 2014, 5, 3189–3196
This journal is © The Royal Society of Chemistry 2014

View Article Online
Edge Article
Chemical Science
we found that the persistent radical anions 1K, 2K and 2Co, 15 A. Lorbach, A. Hubner and M. Wagner, Dalton Trans., 2012,
formed upon one electron reduction, could be generated and 41, 6048–6063.
characterized. In the case of 2Co, an X-ray structural determi- 16 T. K. Wood, W. E. Piers, B. A. Keay and M. Parvez, Angew.
nation was possible, allowing for a detailed analysis of the bond Chem., 2009, 121, 4069–4072.
length changes in the conjugated framework upon sequential 17 T. K. Wood, W. E. Piers, B. A. Keay and M. Parvez, Chem.–Eur.
one and two electron reduction. Dianion 1K2 undergoes a J., 2010, 16, 12199–12206.
Bronsted acid mediated isomerization to form 2K2, conrming 18 H. Braunschweig and T. Kupfer, Chem. Commun., 2011, 47,
our hypothesis. tert-Butanol is the best reagent for inducing this 10903–10914.
skeletal rearrangement since more acidic phenols can doubly 19 T. B. Tai, V. T. T. Huong and M. T. Nguyen, Chem. Commun.,
protonate 2K2 to form the compound 3 as an air and moisture 2013, 49, 11548–11550.
stable white solid. Compound 2K2 can be oxidised using silver 20 H. Braunschweig, C.-W. Chiu, K. Radacki and P. Brenner,
triate, providing a more reliable and efficient preparative route
to the ladder diborole complex 2 from 1.
Chem. Commun., 2010, 46, 916–918.
21 T. Araki, A. Fukazawa and S. Yamaguchi, Angew. Chem., Int.
Ed., 2012, 51, 5484–5487.
22 A. Y. Houghton, V. A. Karttunen, W. E. Piers and
H. M. Tuononen, Chem. Commun., 2014, 50, 1295–1298.
Acknowledgements
Funding for the experimental work described was provided by 23 C. Fan, W. E. Piers and M. Parvez, Angew. Chem., Int. Ed.,
the Natural Sciences and Engineering Research Council of
2009, 48, 2955–2958.
Canada in the form of a Discovery Grant to W.E.P. W.E.P. also 24 C. Fan, L. G. Mercier, W. E. Piers, H. M. Tuononen and
thanks the Canada Research Chair secretariat for a Tier I CRC M. Parvez, J. Am. Chem. Soc., 2010, 132, 9604–9606.
(2013–2020). J. F. A. thanks Alberta Ingenuity—Technology 25 A. Iida, A. Sekioka and S. Yamaguchi, Chem. Sci., 2012, 3,
Futures for scholarship support.
1461–1466.
26 A. Wakamiya, K. Mishima, K. Ekawa and S. Yamaguchi,
Chem. Commun., 2008, 579–581.
27 J. F. Araneda, B. Neue, W. E. Piers and M. Parvez, Angew.
Chem., Int. Ed., 2012, 51, 8546–8550.
Notes and references
1 J. E. Anthony, Angew. Chem., Int. Ed., 2008, 47, 452–483.
2 R. Ruiz, D. Choudhary, B. Nickel, T. Toccoli, K.-C. Chang, 28 J. W. Barton and M. K. Shepherd, J. Chem. Soc., Perkin Trans.
A. C. Mayer, P. Clancy, J. M. Blakely, R. L. Headrick,
1, 1987, 1561.
S. Iannotta and G. G. Malliaras, Chem. Mater., 2004, 16, 29 Based on control experiments using potassium salts of the
4497–4508.
radical anions and dianions discussed below, the
irreversibility of these reduction waves is likely due to the
reactivity of the reduction products with the supporting
electrolyte.
3 M. A. Wolak, J. Delcamp, C. A. Landis, P. A. Lane, J. Anthony
and Z. Kafa, Adv. Funct. Mater., 2006, 16, 1943–1949.
4 K. P. Goetz, Z. Li, J. W. Ward, C. Bougher, J. Rivnay, J. Smith,
B. R. Conrad, S. R. Parkin, T. D. Anthopoulos, A. Salleo, 30 D. R. Duling, J. Magn. Reson., Ser. B, 1994, 104, 105–110.
J. E. Anthony and O. D. Jurchescu, Adv. Mater., 2011, 23, 31 T. K. Wood, W. E. Piers, B. A. Keay and M. Parvez, Chem.
3698–3703.
Commun., 2009, 5147–5149, DOI: 10.1039/B911982E.
32 C. D. Martin, M. Soleilhavoup and G. Bertrand, Chem. Sci.,
2013, 4, 3020–3030.
5 J. E. Anthony, Chem. Rev., 2006, 106, 5028–5048.
6 A. R. Murphy and J. M. J. Frechet, Chem. Rev., 2007, 107,
´
1066–1096.
33 The theoretical calculations were performed using the
Gaussian 09 program. See the ESI† for a full citation.
34 V. Barone, in Recent Advances in Density Functional Methods,
ed. D. P. Chong, World Scientic Pub Co Inc, Singapore,
1996, DOI: 10.1142/9789812830586_0008.
7 T. Noda and Y. Shirota, J. Am. Chem. Soc., 1998, 120, 9714–
9715.
8 W.-L. Jia, D.-R. Bai, T. McCormick, Q.-D. Liu, M. Motala,
R.-Y. Wang, C. Seward, Y. Tao and S. Wang, Chem.–Eur. J.,
2004, 10, 994–1006.
ˆ
35 J. C. Walton, M. M. Brahmi, L. Fensterbank, E. Lacote,
9 A. Shuto, T. Kushida, T. Fukushima, H. Kaji and
S. Yamaguchi, Org. Lett., 2013, 15, 6234–6237.
M. Malacria, Q. Chu, S.-H. Ueng, A. Solovyev and
D. P. Curran, J. Am. Chem. Soc., 2010, 132, 2350–2358.
10 C. R. Wade, A. E. J. Broomsgrove, S. Aldridge and 36 J. C. Walton, M. M. Brahmi, J. Monot, L. Fensterbank,
¨
ˆ
F. P. Gabbaı, Chem. Rev., 2010, 110, 3958–3984.
M. Malacria, D. P. Curran and E. Lacote, J. Am. Chem. Soc.,
11 X.-Y. Wang, H.-R. Lin, T. Lei, D.-C. Yang, F.-D. Zhuang,
2011, 133, 10312–10321.
J.-Y. Wang, S.-C. Yuan and J. Pei, Angew. Chem., Int. Ed., 37 T. Kushida and S. Yamaguchi, Organometallics, 2013, 32,
2013, 52, 3117–3120.
6654–6657.
12 X.-Y. Wang, F.-D. Zhuang, R.-B. Wang, X.-C. Wang, X.-Y. Cao, 38 H. Braunschweig, V. Dyakonov, J. O. C. Jimenez-Halla,
J.-Y. Wang and J. Pei, J. Am. Chem. Soc., 2014, 136, 3764–
3767.
K. Kra, I. Krummenacher, K. Radacki, A. Sperlich and
J. Wahler, Angew. Chem., Int. Ed., 2012, 51, 2977–2980.
¨
13 C. D. Entwistle and T. B. Marder, Chem. Mater., 2004, 16, 39 R. Bertermann, H. Braunschweig, R. D. Dewhurst, C. Horl,
4574–4585.
T. Kramer and I. Krummenacher, Angew. Chem., Int. Ed.,
2014, 53, 5473.
¨
14 F. Jakle, Chem. Rev., 2010, 110, 3985–4022.
This journal is © The Royal Society of Chemistry 2014
Chem. Sci., 2014, 5, 3189–3196 | 3195

View Article Online
Chemical Science
Edge Article
40 A. Iida and S. Yamaguchi, J. Am. Chem. Soc., 2011, 133, 6952– 44 H. V. Ly, H. M. Tuononen, M. Parvez and R. Roesler, Angew.
6955. Chem., Int. Ed., 2008, 47, 361–364.
41 N. G. Connelly and W. E. Geiger, Chem. Rev., 1996, 96, 877– 45 Z. Chen, C. S. Wannere, C. Corminboeuf, R. Puchta and
910. P. v. R. Schleyer, Chem. Rev., 2005, 105, 3842–3888.
42 M. Saito, Y. Hashimoto, T. Tajima, K. Ishimura, S. Nagase 46 M. Saito, Symmetry, 2010, 2, 950–969.
and M. Minoura, Chem.–Asian J., 2012, 7, 480–483.
43 R. A. Lee, R. J. Lachicotte and G. C. Bazan, J. Am. Chem. Soc.,
1998, 120, 6037–6046.
47 Y. Zhao and D. Truhlar, Theor. Chem. Acc., 2008, 120, 215–
241.
3196 | Chem. Sci., 2014, 5, 3189–3196
This journal is © The Royal Society of Chemistry 2014