Reversible Vapochromic Luminescence Accompanied by Planar Half-Chair Conformational Change of a Propeller-Shaped Boron β-Diketiminate Complex

Article abstract of DOI:10.1002/chem.202101107

Leakage of volatile organic compounds (VOCs) is one of the most severe industrial problems, because it can cause environmental pollution, global warming, fire, and explosion. Hence, the visualization of leakage is an essential technology to detect it at a

Full text of DOI:10.1002/chem.202101107

Accepted Article
Accepted Article
Title: Reversible Vapochromic Luminescence Accompanied by Planar–
Half-Chair Conformational Change of a Propeller-Shaped Boron
β-Diketiminate Complex
Authors: Shunichiro Ito, Misao Yaegashi, Kazuo Tanaka, and Yoshiki
Chujo
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: Chem. Eur. J. 10.1002/chem.202101107
Link to VoR: https://doi.org/10.1002/chem.202101107
01/2020

10.1002/chem.202101107
Chemistry - A European Journal
Manuscript for Chemistry—A European Journal
Reversible Vapochromic Luminescence
Accompanied by Planar–Half-Chair Conformational
Change of a Propeller-Shaped Boron β-Diketiminate
Complex
Dr. Shunichiro Ito, Misao Yaegashi, Prof. Dr. Kazuo Tanaka* and Prof. Dr. Yoshiki Chujo
Department of Polymer Chemistry, Graduate School of Engineering, Kyoto University, Katsura,
Nishikyo-ku, Kyoto 615-8510, Japan
E-mail: tanaka@poly.synchem.kyoto-u.ac.jp
Keywords: Boron Complex; Crystallization-Induced Emission; Fluorescence; Vapochromism;
Crystal Transition
This article is protected by copyright. All rights reserved.

10.1002/chem.202101107
Chemistry - A European Journal
Manuscript for Chemistry—A European Journal
Abstract: Leakage of volatile organic compounds (VOCs) is one of the most severe industrial
accidents because it can cause environmental pollution, global warming, fire, and explosion.
Hence the visualization of leakage is an essential technology to detect it at an early stage.
Molecular crystals whose fluorescence color can be changed by the exposure to VOCs could
potentially serve as the sensing materials for realizing rapid and facile VOC detection. These
materials, however, usually require harsh conditions, such as heating or a vacuum, to recover their
initial phases for reuse. Therefore, it remains a challenge to obtain completely reversible sensing
systems without such energy-consuming recycling processes. Here we report the reversible color
change of fluorescence from the crystals of a propeller-shaped boron β-diketiminate complex. The
complex was crystallized in distinct crystalline phases having different luminescent colors.
Importantly, these phases were interconverted very rapidly (time constant < 60 sec) and repeatedly
upon exposure to the vapors of the appropriate VOCs. The small energy differences between
conformers of the complex could lead to this pseudopolymorphic behavior. This finding could be
applied for the development of further eco-friendly reversible sensing materials based on four-
coordinated boron complexes.
2
This article is protected by copyright. All rights reserved.

10.1002/chem.202101107
Chemistry - A European Journal
Manuscript for Chemistry—A European Journal
Introduction
Sensing,[1–4] separation,[5–7] and storage[8–11] technologies of gaseous materials are of paramount
importance for sustainable development. Porous materials (PMs) have been promising candidates
for these usages because they absorb guest molecules into their voids derived from their long-
ranged ordered structures and expel them by external stimuli like heating.[12–16] The physical
properties of PMs depend drastically on the existence and absence, size, and electronic nature of
guests. PMs which exhibit luminescent chromism in response to exposure to vapors of volatile
organic compounds (VOCs), especially, are attractive materials for detecting a small amount of
leakage of toxic or explosive gases and for security ink.[17–31] However, these compounds possess
potential instability toward the air, moisture, and/or other harsh chemical conditions, and therefore
a growing number of studies have recently reported chemically and physically stable PMs.[32,33]
Nonporous molecular crystals (NMCs), on the other hand, have no such functional voids in their
packing patterns because of the absence of the polymeric or supramolecular structures derived
from continuous valence, coordination, or hydrogen bonds. Nevertheless, various kinds of NMCs
change their crystal structures by exposure to guest molecules. For instance, Lewis basic guests
often increase the coordination numbers of transition metal complexes in NMCs, lead crystal–
crystal transition, and alter the color, luminescence, or magnetic properties.[34–39] Although such
specific affinities to guest compounds are useful for constructing stimuli-responsive systems, they
often require harsh conditions to recover their initial states, for example, heating, drying in a
vacuum, and both. Thus, the development of sensing materials towards VOCs without such
affinities is still of importance for achieving stable and environment-friendly systems.
In this context, we show that conformational degrees of freedom are able to play a pivotal role
in constructing stimuli-responsive NMCs. External stimuli can induce the interconversion between
3
This article is protected by copyright. All rights reserved.

10.1002/chem.202101107
Chemistry - A European Journal
Manuscript for Chemistry—A European Journal
some conformers when slight energy barriers separate them. As a result, the crystals of one of the
conformers could transform into another one. For example, dimethoxy-substituted difluoroboron
avobenzone exhibits polymorphs resulted from the existence of the syn–anti rotational conformers
originating from the different mutual orientations of the two methoxy groups.[40] Some studies have
reported similar conformational flexibility of other NMCs, e.g., other difluoroboron
avobenzones,[41,42] a platinum complex of 9-phenyl-9-arsafluorene,[43] and a gallium β-diketiminate
complex.[44] However, there is a limited number of reports on completely reversible vapochromism
of luminescent NMCs without the harsh conditions in their recycling process,[43] probably because
tight crystal structures could hamper the flips between conformers.
The studies mentioned above suggest that the achievement of stimuli-responsive NMCs would
need the following two characteristics. First, the conformers of a molecule should be separated by
small potential energies to offer a possibility of (pseudo-)polymorphism. Second, the packing
structures of NMCs should be relatively sparse so that the flips between the conformers would not
be prohibited. Here, four-coordinated complexes of the group 13 elements possess the degree of
freedom of out-of-plane bending of the boron atom from the central coordination ring (Figure 1a).
[40,45]
In addition to this degree of freedom, these complexes often exhibit luminescent chromism
in response to external stimuli, such as mechanical grinding,[40,41,46–50] heating[51,52] and VOCs.[44] β-
Diketiminate complexes would be advantageous concerning their efficient solid-state emission
including aggregation- and crystallization-induced emission,[44,53–56] the designability of ligands,
and chemical stability.[57–60] In this report, we hypothesized that such flexible and emissive β-
diketiminate boron complexes might afford stimuli-responsive luminescent materials with low-
energy recycling ability. A synthesized propeller-shaped boron complex showed aggregation- and
crystallization-induced emission properties. Importantly, the complex crystallized into some
4
This article is protected by copyright. All rights reserved.

10.1002/chem.202101107
Chemistry - A European Journal
Manuscript for Chemistry—A European Journal
distinct phases depending on the solvents for crystallization: One phase contained only the
complex, while the others were composed of the complex and solvent molecules. These pseudo-
polymorphs exhibited different emission colors and repeatedly interconverted each other by
treating with the vapor of VOCs at room temperature.
5
This article is protected by copyright. All rights reserved.

10.1002/chem.202101107
Chemistry - A European Journal
Manuscript for Chemistry—A European Journal
Results and Discussion
To realize a stimuli-responsive luminescent system consisted of boron β-diketiminates, we
designed a propeller-like pentaphenyl boron complex (PPB, Figure 1b). The five phenyl groups
were introduced into the central chelation ring because bulky substituents often induce sparse
packing structures in the crystalline states.[61]
Single-point energy calculations of PPB in the gas phase were performed for some conformers
with Gaussian 16 Rev. B package[62] at the M06/6-31+G(d,p) level of theory.[63] A potential energy
surface of PPB along with the bent angle (θ) was calculated with partial optimizations, varying θ
from 160º to 180º, and the other degrees of freedom were optimized (Figure 1c). The energy
difference between the planar conformer (θ = 180°) and the most stable half-chair conformation
(θ = 172°) is 0.59 kJ mol–1. This value is significantly smaller than the rotational barrier of the
methoxy group of anisole (13–19 kJ mol–1 in a solution)[64–66] or the ring-flip one of cyclohexane
(43 kJ mol–1 in a solution).[67–69] Such a small value indicates that these conformers can be thermally
accessible at room temperature.
6
This article is protected by copyright. All rights reserved.

10.1002/chem.202101107
Chemistry - A European Journal
Manuscript for Chemistry—A European Journal
Figure 1. (a) Schematic illustration of the conformational change of boron complexes. The
magenta sphere represents a boron atom. (b) A chemical structure of a β-diketiminate complex
PPB investigated in this study. (c) Calculated potential energy surface along the bent angle θ
depicted in the inset. The energy is relative to the value at θ = 172°.
Scheme 1 shows the synthetic method of PPB. The imine derivative 3 was prepared by
condensation of aniline 1 and benzyl phenyl ketone 2, according to the literature.[70] Deprotonation
of 3 by tert-butyllithium, followed by the reaction with imidoyl chloride 4, yielded a ligand LH.
tert-Butyllithium gave the best result as a base for the preparation of the aza-enolate. Less bulky
alkyllithium reagents, such as methyllithium, n- and sec-butyllithium, lead to nucleophilic addition
to 3 as a side reaction. Lithium diisopropylamide, on the other hand, gave almost no product. The
reaction between LH and trifluoroborane diethyl etherate in the presence of triethylamine afforded
the complex PPB. The pure product was yielded by recrystallization from the mixed solution of
hexane and ethyl acetate. The 11B{1H} spectrum exhibited a sharp triplet signal at 1.29 ppm, which
is a distinctive feature of four-coordinated difluoroboron complexes.[54,55] The C{1H} spectra
13
showed the characteristic peak assigned to the α-carbon of the β-diketimine moiety at 109.4 ppm.
7
This article is protected by copyright. All rights reserved.

10.1002/chem.202101107
Chemistry - A European Journal
Manuscript for Chemistry—A European Journal
Scheme 1. Synthetic scheme of a pentaphenyl boron β-diketiminate complex PPB
The absorption spectrum of PPB in the chloroform solution (1.0 × 10–5 M) is shown in Figure 2.
The compound showed an absorption band at 370 nm (Table 1), which is comparable to the value
of the tetraphenyl boron complex (376 nm).[54,55] This electronic transition should originate from
the symmetrically allowed π‒π* nature of the first singlet excited state S1 of the compound.[56] This
boron complex exhibited typical aggregation-induced emission (AIE) behavior (Figure 2). The
absolute photoluminescence (PL) quantum yields of PPB in its dilute chloroform, THF and
hexane/chloroform (99/1, vol/vol) solutions (1.0 × 10–5 M) were less than the limit of detection (<
0.01). The absorption and emission peaks of these dilute solutions were not dependent highly on
the solvents. Figure 2 shows the emission spectra of the solution samples with different solvent
compositions (1.0 × 10–5 M in mixed solvents of THF and water; the fraction of water = 1–99 %).
The dependence of integrated emission intensity ratio A/A0 on the fraction of water (fw) is listed in
Table 2, where A0 is the integrated intensity at fw = 1 vol%. The A/A0 values significantly increased
over 80 vol% of the water content, while the values in the low proportion of water were almost
unity. The solutions containing over 80 vol% of water were turbid, and the absolute PL quantum
yields (ΦPL’s) of these water-rich suspensions were about 0.03. These results indicate that the
formation of aggregates induces the emission enhancement of PPB. In the solution states, intra-
and intermolecular motions and vibrations could non-radiatively consume the excitation energy at
room temperature. These motions and vibrations, on the other hand, should be suppressed in the
8
This article is protected by copyright. All rights reserved.

10.1002/chem.202101107
Chemistry - A European Journal
Manuscript for Chemistry—A European Journal
aggregated states. Consequently, this complex is allowed to emit stronger in the aggregated state
than in the solution state.[54–56] Notably, the A/A0 decreased at the 99 vol% of the water content, but
still higher than those under the conditions with a low concentration of water (< 80 vol%). This
decrease may originate from the precipitation of the large aggregates.
Figure 2. An absorption spectrum of PPB in chloroform solution (1.0 × 10–5 M, black solid) and
photoluminescence spectra of PPB in mixed solvents of THF and water (1.0 × 10–5 M, colored
solid) with different water contents (fw = 1–99 vol%).
9
This article is protected by copyright. All rights reserved.

10.1002/chem.202101107
Chemistry - A European Journal
Manuscript for Chemistry—A European Journal
Table 1. Photophysical properties of PPB in solution states[a]
Solvent
Condition
λabs / nm
370
εmax / 104 M–1 cm–1
λFL / nm
ΦPL
CHCl3
r.t.
r.t.
2.74
3.05
464
465
< 0.01
< 0.01
THF
371
hexane/CHCl3
(99/1, vol/vol)
r.t.
374
3.37
463
< 0.01
r.t.
372
3.34
463
432
< 0.01
1.0
2-MeTHF
77 K
n.d.[b]
n.d. [b]
[a] In dilute solutions (1.0 × 10–5 M) at r.t. and 77 K. [b] Not determined.
Table 2. Dependency of emission properties towards the fraction of water
fw (%)[a]
1
10
20
30
40
50
60
70
80
90
99
λem / nm 467 464 464
464 462 463 463 462 471
1.05 1.09 1.12 1.28 1.28 6.28
479
8.42
0.04
480
A/A0[b]
1
0.95 1.03
2.98
0.03
[c]
ΦPL
< 0.01
0.04
[a] Volume fraction of water of mixed solvents of THF/water. [b] Integrated intensity (A) of
photoluminescence normalized to the integrated intensity at fw = 1% (A0). [c] Absolute
photoluminescence quantum yield.
The electronic nature in the excited state of a single molecule of the complex was investigated
with photophysical measurements for a dilute 2-methyltetrahydrofuran (2-MeTHF) solution (1.0
× 10–5 M) at room temperature and 77 K. The absorption band (λmax = 372 nm) was almost identical
to those in the other solvents. The solution was not emissive at room temperature (ΦPL < 0.01) as
well as in the other solvents but was emissive at 77 K (λPL = 432 nm, ΦPL = 1.0; Table 1). The
long-lifetime phosphorescence spectrum was detected at 521 nm (τPhos = 0.64 sec) when the
10
This article is protected by copyright. All rights reserved.

10.1002/chem.202101107
Chemistry - A European Journal
Manuscript for Chemistry—A European Journal
spectrum was recorded 1.0 msec after the excitation in order to eliminate background fluorescence
interference. The intensity of the phosphorescence, however, was too weak to be distinguished
from the fluorescence in the steady-state spectrum. Consequently, the quenching of the S1 state
through intersystem crossing from the singlet state to triplet states should be a minor process.
Distinct crystal phases were obtained by changing solvents for recrystallization. The yellow
crystals obtained from hexane/EtOAc exhibited weak green emission (Phase G) at room
temperature under UV irradiation (365 nm), while the colorless crystals obtained from
hexane/dichloromethane (DCM) showed intense blue emission (Phase B, Figure 3a). Powder X-
ray diffraction (PXRD) profiles were different from each other (Figure 3b). 1H NMR spectrum of
Phase B showed the peak at 5.30 ppm assigned to DCM, meanwhile that of Phase G was identical
to that of the pure PPB powder. The integration ratio of the peak of DCM against the peaks of
PPB in the 1H NMR signals indicates that Phase B contains the complex and DCM in the molar
ratio of 1:1. In addition, a profile of the thermogravimetric analysis of Phase B showed 11.4% of
weight loss at 67.3 °C (Figure 3c). This experimental value is in good agreement with the estimated
value for the dissociation of DCM from the 1:1 mixture of PPB and DCM (14.6%). Consequently,
it was revealed that PPB exhibits pseudopolymorphism.
11
This article is protected by copyright. All rights reserved.

10.1002/chem.202101107
Chemistry - A European Journal
Manuscript for Chemistry—A European Journal
Figure 3. (a) Preparation methods of Phases G and B. Pictures of Phases G and B were taken under
natural light (left panels) and UV irradiation (right panels). (b) Powder X-ray diffraction patterns
for Phases G and B. (c) Thermogravimetric analysis profiles for Phases G and B.
A crop of crystals of Phase B was grown from a hexane/DCM solution and suitable for a single-
crystal X-ray diffraction (SCXRD) analysis. Conversely, Phase G appeared only as a crystalline
powder when the compound was recrystallized from hexane or hexane/EtOAc solutions.
Therefore, the PXRD data of Phases G and B were analyzed, and subsequently their crystal
structures were refined by using the Rietveld method (Figure 4). Both phases belong to the
monoclinic space group P21/n (No. 14). The asymmetric units of Phases G and B contain one PPB
molecule, and one PPB and one DCM molecules, respectively. Importantly, the complex in Phase
G shows a half-chair conformation (θ ~ 165°), while the structure in Phase B is a planar
conformation (θ ~ 180°). As shown in Figure 1, the relative energies of the optimized structures
with fixed bent angles 165° and 180° are ca. 0.9 and 0.6 kJ mol–1, respectively. Hence these
12
This article is protected by copyright. All rights reserved.

10.1002/chem.202101107
Chemistry - A European Journal
Manuscript for Chemistry—A European Journal
conformers would be thermally redistributable at room temperature when there is no external
potential.
Figure 4. Rietveld plots for (a) Phase G and (d) Phase B. Red crosses, green tics, black line, and
blue line represent the observed data, the positions of Bragg reflections, the calculated data, and
the differences between the observed and calculated data. A single molecular structure (b, e) and
the bent angles (c, f) for Phase G and Phase B, respectively.
To elucidate the electronic nature of the complex in Phases G and B, diffuse reflectance and PL
spectra were recorded (Figure 5). Both diffuse reflectance spectra of Phases G and B show an
electronic transition band at 367 nm (3.38 eV), which is comparable to that of the dilute solution
(370 nm; Table 4). An additional transition band was observed at around 440 nm (2.82 eV) only
in the spectrum of Phase G. This sideband is responsible for the yellow color of the powder of
Phase G. This fact indicates that the lower electronic state is formed in Phase G. The PL spectrum
of Phase G was also detected in the lower energy region (490 nm) than that of Phase B (443 nm).
According to Kasha’s rule, the red-shifted emission band should be attributed from the additional
13
This article is protected by copyright. All rights reserved.

10.1002/chem.202101107
Chemistry - A European Journal
Manuscript for Chemistry—A European Journal
band which appears in Phase G. Indeed, the fluorescence at 490 nm was observed when the sample
of Phase G was excited at 440 nm as well at 367 nm.
Table 4. Photophysical properties of Phases G and B[a]
Phase
λref / nm
367, 440
367
λFL / nm
490
ΦPL
Phase G
Phase B
0.03
0.66
443
[a] Photoluminescence spectra were recorded under excitation at 370 nm.
Figure 5. Normalized diffuse reflectance (upper panel) and normalized photoluminescence (lower
panel) spectra of Phases G and B. Diffuse reflectance spectra were converted with a Kubelka–
Munk (KM.) function. Gray dashed line shows an absorption spectrum of PPB in a chloroform
solution (1.0 × 10–5 M).
14
This article is protected by copyright. All rights reserved.

10.1002/chem.202101107
Chemistry - A European Journal
Manuscript for Chemistry—A European Journal
To obtain further insight into the electronic properties of PPB, a series of density functional
theory (DFT) and time-dependent DFT (TD-DFT) calculations were carried out. The levels of
theory were the (TD-)M06/6-31+G(d,p) and (TD-)M06/6-311++G(d,p) for geometry
optimizations and single-point energy calculations, respectively. The optimized structure of the
complex in a vacuum at the S0 state for a single molecule shows a half-chair conformation (θ =
171.96°). As shown in Table 5, the transition energy corresponding to the S0–S1 electronic
transition was estimated to be 3.48 eV (357 nm), which is comparable to the experimental value
of the absorption maximum 3.35 eV (λabs = 370 nm). Highest occupied (HO) and lowest
unoccupied (LU) molecular orbitals (MOs) for the optimized geometry are shown in Figure 6. The
S1 state was assigned to the symmetrically allowed HOMO–LUMO transition, which has π‒π*
nature.[56]
The effects of the bent angle θ on the properties of the electronic transition of a single molecule
were evaluated. The parameters for the corresponding electronic transitions were calculated by
using the optimized structures with the fixed θ (= 160° and 180°), as shown in Table 5. The
transition energies varied in the small range of 3.44–3.54 eV (361–351 nm). These results indicate
that the low-energy absorption band observed in Phase G (θ = 165.11°) would not be reproducible
in a regime of a single molecule.
15
This article is protected by copyright. All rights reserved.

10.1002/chem.202101107
Chemistry - A European Journal
Manuscript for Chemistry—A European Journal
Table 5. Calculated parameters of the S0–S1 electronic transitions for single molecules with some
geometries
θ / deg[a]
160
Energy / eV[b] λ / nm
f[c]
Composition
Coefficient[d]
0.70326
3.44
3.48
3.54
361
357
351
0.4150
0.4181
0.4145
HOMO → LUMO
HOMO → LUMO
HOMO → LUMO
172[e]
180
0.70312
0.70280
[a] Bent angle. [b] Transition energy. [c] Oscillator strength. [d] Expansion coefficient for the
electronic transitions. [e] An optimized geometry in the singlet ground state.
Figure 6. HOMO and LUMO distributions of the optimized structure of PPB.
Next, we calculate transition energies considering intermolecular interactions between the
nearest neighbor dimers (NNDs) in the crystal structures of Phases G and B in order to elucidate
the origin of the lower energy absorption band. Typical intermolecular interactions that lead to the
large energy change of electronic transitions are H- and J-type excitonic couplings and
intermolecular charge transfer.[71,72] For a system composed of a dimer, the excitonic couplings are
expressed as the interactions between the electronic transition dipole moments (μel) of the two
molecules (Figure 7a). When the two dipoles couple each other, two delocalized excited states are
generated. The strength of the coupling is evaluated with a value of Coulomb coupling JC:
16
This article is protected by copyright. All rights reserved.

10.1002/chem.202101107
Chemistry - A European Journal
Manuscript for Chemistry—A European Journal
$
$
(
)
ꢁ"# 1 − 3 cos ꢂ
4ꢃꢄꢅ%
(1)
ꢀ! =
,
where φ is the angle formed by the two dipoles (0° < φ < 90°), ε is the dielectric constant, and R
is the distance between the dipoles. The two states split by 2|JC| consist of allowed “in-phase” and
forbidden “out-of-phase” states. H-aggregates maintain a “side-by-side” orientation where φ is
larger than the so-called “magic angle” φM = 54.7°. In this case, a Coulomb coupling is positive
(JC > 0). Conversely, J-aggregates maintains a “head-to-tail” orientation (0° < φ < φM) leading to
a negative Coulomb coupling (JC < 0). In Phases G and B, the angles φ between the two dipoles
of the NNDs (Figures 7b and 7c) are 78.4° and 56.3°, respectively. Therefore, both of these dimers
constitute H-aggregates where the lower-energy state is an optically dark “out-of-phase” state. The
coupling strengths were estimated by using TD-DFT calculations at the M06/6-311++G(d,p) level
of theory. The values corresponding to 2|JC| of Phases G and B are 0.07 and 0.04 eV, respectively
(Figures 7f and 7g, Table 6). These values were considerably small compared to the energy
difference between main and sidebands in the diffuse reflectance spectrum of Phase G (0.56 eV).
It is worth to note that such small interactions should originate from the large intermolecular
distances (Phase G: 6.49 Å; Phase B: 6.59 Å) likely due to the steric hindrance of the peripheral
aromatic rings. These results indicate that excitonic couplings between molecules separated by
longer distances than the NNDs should be much smaller. Consequently, the excitonic couplings
would play a limited role in the formation of the sideband.
17
This article is protected by copyright. All rights reserved.

10.1002/chem.202101107
Chemistry - A European Journal
Manuscript for Chemistry—A European Journal
Figure 7. (a) Schematic representation of excitonic coupling. (b, c) Packing structures of Phases
(b) G and (c) B. Red and blue vectors represent electronic-transition dipole moments of each
molecule. Cyan circles highlight nearest neighbor dimers (NNDs) for theoretical calculations. (d,
e) Molecular orbital distributions of the NNDs for Phases (d) G and (e) B. (f, g) Relative transition
energies obtained by considering intermolecular interactions between the NNDs for Phases (f) G
and (g) B. H and L denote HOMO and LUMO, respectively.
TD-DFT calculations for the NND in Phase G, on the other hand, revealed that the
intermolecular CT could form a low-energy state. The HOMO localized at a single molecule,
whereas the LUMO localized at the other (Figure 7d). The lowest excited state of the NND is
composed of the HOMO–LUMO transition, indicating that an electron on a complex would
transfer into the other one during a photoexcitation. This CT state was located 0.20 eV below the
S1 state of the monomer having the geometry in Phase G (Figure 7f, Table 6). The corresponding
oscillator strength of the CT state was estimated to be 0.0003, meaning that the strength of the
18
This article is protected by copyright. All rights reserved.

10.1002/chem.202101107
Chemistry - A European Journal
Manuscript for Chemistry—A European Journal
electronic transition attributed to this CT state should be quite low. Such a CT state was not
produced in the packing structures of Phase B (Figures 7e and 7g, Table 6). Hence, the sideband
observed in the diffuse reflectance spectrum of Phase G could be assigned to the transition to the
intermolecular CT state. The interactions of three or more molecules could result in the lower
energy state (440 nm; 2.82 eV) observed in the absorption spectrum of Phase G. According to
Kasha’s rule, the low PL quantum yield of Phase G is also derived from the small oscillator
strength of the CT state. The slight energy difference (40 meV) between the dimeric states in Phase
B could allow excitons to be activated thermally from the dark out-of-phase state to the emissive
in-phase state. It should be mentioned that the results of the calculations for Phase B were hardly
affected when a DCM molecule included in Phase B was considered.
19
This article is protected by copyright. All rights reserved.

10.1002/chem.202101107
Chemistry - A European Journal
Manuscript for Chemistry—A European Journal
Table 6. Results of DFT calculations for single molecules and NNDs in Phases G and B
Phase
Structure State
Monomer S1
E / eV[a] λ / nm
f[b]
Composition[c] Coeff. [d]
3.55
349
370
0.4007
0.0003
0.70254
0.70539
0.46898
H → L
S1 (CT) 3.35
S2
H → L
H–1 → L
3.50
354
347
0.0013
0.6707
(out-of-
phase)
G
0.51906
0.51759
–0.47229
H → L+1
H–1 → L
H → L+1
NND
S3
3.57
(in-
phase)
Monomer S1
3.77
3.74
329
331
0.4177
0.0006
0.70198
0.46502
0.49149
0.48128
0.49124
H → L
H–1 → L+1
H → L
S1
B
NND
H–1 → L
H → L+1
S2
3.78
328
0.7157
[a] Transition energy. [b] Oscillator strength. [c] Composition of the electronic states. H and L
denote HOMO and LUMO, respectively. [d] Expansion coefficient for the electronic transition.
Remarkably, Phases G and B can repeatedly interconvert by fuming the vapor of DCM or
hexane. The crystalline powder of Phase G was transformed into Phase B upon exposure to DCM
vapor for 15 h at room temperature. The powder of Phase B, on the other hand, was converted to
Phase G spontaneously but very slowly without the DCM vapor. (Phase B remained after a year.)
This phase transition from Phase B to Phase G was accelerated upon exposure to the hexane vapor.
This means that Phase B is a stable phase only under the DCM vapor, and is a metastable without
it. The hexane vapor can accelerate this spontaneous process probably because the hexane
20
This article is protected by copyright. All rights reserved.

10.1002/chem.202101107
Chemistry - A European Journal
Manuscript for Chemistry—A European Journal
molecules could enhance the molecular mobility of PPB and DCM molecules on the surface of
the crystals of Phase B.
The dynamics of the phase transitions between Phase B and Phase G were evaluated by recording
the PL spectra of the samples at different treatment time (Figure S3). The time constant of the
transition from Phase G to Phase B was estimated to be 28 sec. On the other hand, the kinetics of
the transition from Phase B to Phase G involve two distinct time constants, 53 sec and 4 hours.
The fast component in each transition may be attributed to the solvation–recrystallization process.
These very rapid transitions between the two phases enable us to detect the VOC gases in real time.
The slow component in the transition from Phase B to Phase G might be originated from the
permeation process of hexane molecules into the intact parts buried in the outer layer in which the
B-to-G phase transition has been accomplished. This deacceleration effect may like the passivation
of metals. The time constant of this slow component seemed to depend strongly on the particle
size of the powder. In addition, it is worth to note that PPB was deliquescent with DCM but not
with hexane. This might mean that the solvation of the surface molecules by DCM is much more
favorable than by hexane. Therefore, thin layer of the concentrated solution of PPB could be
formed on the surfaces of the crystals under the DCM vapor. Such a solvated layer could accelerate
the phase transition.
The reversibility of this interconversion was monitored with PL spectroscopy and PXRD (Figure
8). The PL maxima and the PXRD patterns were identical to those of the corresponding phases in
each cycle. Typical nonporous molecular crystals (NMCs) need to dry in a vacuum or upon heating
or immerse to the liquids of solvents in order to recycle the materials.[43] In contrast, the PPB
crystals can be recycled only using the vapors of VOCs under ambient condition. It should be
mentioned that the weak emission band was observed in the shorter wavelength region (400–450
21
This article is protected by copyright. All rights reserved.

10.1002/chem.202101107
Chemistry - A European Journal
Manuscript for Chemistry—A European Journal
nm) after the treatments with hexane even for 48 h. This result indicates that the small amount of
Phase B would remain in the hexane-fumed sample. Because of the deliquescence of PPB with
DCM, sometimes the size of the crystals of PPB changes during the G-to-B transitions. Hence, the
fractions of the remaining Phase B were not identical between the cycles.
Figure 8. (a) Photoluminescence spectra of Phases G (green) and B (blue) in each cycle.
Crystalline samples were excited at the absorption maxima of a chloroform solution (λabs = 370
nm). Inset represents the wavelengths at the luminescence maxima (λmax) for each cycle. (b)
Powder X-ray diffraction profiles for each cycle.
To investigate effects of the kind of VOCs on the vapochromic behavior, we treated the crystals
of Phase G with the vapor of 16 kinds of VOCs for 15 h. As shown in Figure 9, the relatively small
and aprotic six kinds of VOCs (Group I) induced the luminescence-color change from green to
blue, the other ten kinds of VOCs induce no change (Group II). PL spectra of the crystals, including
22
This article is protected by copyright. All rights reserved.

10.1002/chem.202101107
Chemistry - A European Journal
Manuscript for Chemistry—A European Journal
the group I VOCs except for DMF, were observed at around 440–465 nm (Figure 9b). These
maximum PL wavelengths were slightly different, probably due to the differences of polarity and
movability of the molecules in the phases, or maybe due to different packing structures. For the
phase exposed to the DMF vapor, the PL spectra showed a maximum peak at 500 nm and a
shoulder peak at around 430 nm. This result means that the phase transition induced by the DMF
vapor should be not completed under this condition. Indeed, after the exposure of the DMF vapor
for three days at room temperature, the peak at 500 nm disappeared completely, and a peak at 434
1
nm was observed. The H NMR spectra of the blue-emissive crystal phases showed the peaks
attributed to those of the corresponding VOCs in group I. The molecular ratios between the
1
complex and the VOCs were calculated to be 1:1 by using the integration ratios of the H NMR
signals. The blue-emissive phases prepared with the Group I VOCs exhibited completely different
PXRD patterns compared to Phase G (Figure 9c). Therefore, these group I VOCs cause the crystal‒
crystal transitions from Phase G to the blue-emissive phases composed of PPB and the VOC
molecules in the 1:1 molar ratio, like DCM. The original packing of Phase G, which would lead
to the green-emissive CT state, might be eliminated by the guest solvent molecules. Surprisingly,
the vapor of DMSO induced the phase transition, although DMSO has a quite low vapor pressure
(0.08 kPa). Thus, high vapor pressures of VOCs were not necessary for the induction of the phase
transition.
In contrast with Group I, the ten VOCs in Group II lead to phase transitions from the blue-
emissive phases to Phase G (Figure 9d). PXRD and 1H NMR revealed that Phase G was recovered
from all blue-emissive phases containing Group I VOCs after exposure of Group II VOCs for 15
h. The PL peaks attributed to the blue-emissive phases were suppressed entirely, and the
1
characteristic peak of Phase G was observed from all the samples fumed by Group II VOCs. H
23
This article is protected by copyright. All rights reserved.

10.1002/chem.202101107
Chemistry - A European Journal
Manuscript for Chemistry—A European Journal
NMR spectra and PXRD profiles of the powders were absolutely identical to those of Phase G,
indicating that Group II molecules accelerate the exclusion of the guest molecules from the
crystalline phase containing the Group I VOCs.
There might be two key factors distinguishing Group I and Group II: solubility of PPB and
size/shape of the solvent molecules. All the molecules in Group I are good solvents for PPB. On
the other hand, hexane, pentane, methanol, ethanol, diethyl ether, and cyclohexane are not. The
molecules of poor solvents could not effectively solvate the PPB molecules and might not be able
to stabilize crystalline states involving the solvent molecules. Even if the solvents are good
solvents for PPB (ethyl acetate, THF, chlorobenzene, and toluene), stable crystalline phases could
not be formed because of the mismatch between the size/shape of the solvent molecules and the
structural features of the solvent sites in available crystal packing structures of PPB.[44]
24
This article is protected by copyright. All rights reserved.

10.1002/chem.202101107
Chemistry - A European Journal
Manuscript for Chemistry—A European Journal
Figure 9. (a) Reversible vapochromic behavior of PPB. (b) Photoluminescence spectra of blue-
emissive phases including the group I molecules. A solid black line shows a photoluminescence
spectrum of Phase G. (c) Powder X-ray diffraction profiles of the blue-emissive phases and Phase
G (solid black line). (d) Photoluminescence spectra of the powders recovered by the group II
compounds from Phase B.
25
This article is protected by copyright. All rights reserved.

10.1002/chem.202101107
Chemistry - A European Journal
Manuscript for Chemistry—A European Journal
Experimental Section
1
Characterization Methods: H (400 MHz) and 13C{1H} (100 MHz) NMR spectra were
recorded on the JEOL JNM-AL400 spectrometer. In 1H and 13C{1H} NMR spectra,
tetramethylsilane (TMS) and/or residual solvent peaks was used as an internal standard. High-
resolution mass (HRMS) spectrometry was performed at the Technical Support Office
(Department of Synthetic Chemistry and Biological Chemistry, Graduate School of Engineering,
Kyoto University), and the HRMS spectra were obtained on a Thermo Fisher Scientific
EXACTIVE spectrometer for electrospray ionization (ESI). Elemental analysis was performed at
the Microanalytical Center of Kyoto University.
Materials: All reactions were performed under argon atmosphere using modified Schlenk line
techniques unless otherwise noted. Analytical thin-layer chromatography (TLC) was performed
with silica gel 60 Merck F254 plates. Column chromatography was performed with Wakogel C-
200 SiO2. All chemicals were purchased from commercial sources (FUJIFILM Wako Pure
Chemical Corporation, Tokyo Chemical Industry Co., Ltd., Sigma-Aldrich Co. LLC. or Kanto
Chemical Co., Inc.) and used as received. Tetrahydrofuran (THF), diethyl ether (Et2O), and
triethylamine (Et3N) were purified using a two-column solid-state purification system (Glass
Contour Solvent System, Joerg Meyer, Irvine, CA). Compounds 3[70] and 4[54] were synthesized
according to the previous reports.
Synthesis of LH: To a solution of 3 (0.2 g, 0.74 mmol, 1 equiv.) in THF (6 mL) was added tert-
butyllithium (0.8 mL, 1.22 mmol, 1.52 M in n-pentane, 1.6 equiv.) dropwise at –78 °C over 1 h.
The reaction mixture was stirred at –78 °C for 0.5 h, and then 6 (0.50 g, 2.3 mmol) in THF (6 mL)
was added to the mixture at –78 °C. After stirred the solution at –78 °C for 1.5 h, and then at
ambient temperature for 18 h under an argon atmosphere, the solution was quenched with saturated
26
This article is protected by copyright. All rights reserved.

10.1002/chem.202101107
Chemistry - A European Journal
Manuscript for Chemistry—A European Journal
aqueous NH4Cl and stirred at 0 °C. The solution was washed with brine and dried over anhydrous
sodium sulfate. After the solvent was removed by a rotary evaporator, the product was purified by
silica gel column chromatography two times with hexane/ethyl acetate = 9/1 as an eluent (Rf =
0.33). The obtained product was dried in vacuum to give crude LH as a yellow solid (90 mg, 26
%). The crude product was used as such without further purification. 1H NMR (400 MHz; CDCl3)
δ = 7.92–7.90 (m, 2H, Ar), 7.4–6.4 (m, 23H Ar), 6.00 (s, 1H, NH) ppm. HRMS (ESI) for [M+H]+:
Calcd., 451.2169; found, 451.2163.
Synthesis of PPB: To a solution of crude LH (0.086 g, 0.19mmol, 1 equiv.) in toluene (2.5 mL)
was added BF3·OEt2 (0.42 mL, 0.48 g, 3.42 mmol) at ambient temperature. After the solution was
stirred at 50 °C for 23 h under an argon atmosphere, the reaction was stopped by cooled to ambient
temperature. The reaction mixture was washed with water and brine and then dried over anhydrous
sodium sulfate. After the solvent was removed by a rotary evaporator, the product was purified by
silica gel column chromatography (eluted with hexane/ethyl acetate = 9/1, Rf = 0.14). The obtained
product was recrystallized from a mixed solvent of hexane/ethyl acetate (5/1). The precipitate
collected by filtration was dried in vacuum to give pure PPB as a yellow crystal (0.018 g, 19%).
1H NMR (400 MHz; CDCl3) δ = 7.19 (d, J = 7.7 Hz, 4H), 7.12 (t, J = 7.7 Hz, 4H), 7.04 (t, J = 7.2
Hz, 2H), 6.91 (m, 10H), 6.77 (m, 3H), 6.71 (m, 2H). 13C{1H} NMR (101 MHz; CD2Cl2) δ = 165.8
(C=N), 141.9, 137.5, 135.6, 133.4, 129.7, 128.6, 128.1, 127.8, 127.2, 127.1, 126.5, 125.6, 109.4
(N=C–C=C–N) ppm. 11B{1H} NMR (128.4 MHz; CDCl3) δ = 1.29 (t, 2J = 31.2 Hz) ppm. HRMS
(ESI) for [M+H]+: Calcd., 521.1971; found, 521.1965.
Photophysical Measurements: UV−vis absorption spectra were recorded on a SHIMADZU
UV–3600 spectrophotometer. Fluorescence and phosphorescence emission spectra and
phosphorescence decay were measured with a HORIBA JOBIN YVON Fluorolog-3
27
This article is protected by copyright. All rights reserved.

10.1002/chem.202101107
Chemistry - A European Journal
Manuscript for Chemistry—A European Journal
spectrofluorometer. Absolute photoluminescence quantum yields were measured with a
Hamamatsu Photonics Quantaurus-QY Plus C13534-01. Photoluminescence (PL) lifetimes were
measured by a Horiba FluoroCube spectrofluorometer system with an Oxford Optistat DN for
temperature control and a UV diode laser (NanoLED 375 nm).
Single-Crystal X-ray Diffraction: Data were collected using a Rigaku R-AXIS RAPID-F with
graphite-monochromated Mo Kα radiation diffractometer and an imaging plate. Data were
collected at 93 K. Equivalent reflections were merged, and a symmetry-related absorption
correction was carried out with the program ABSCOR.[73] The structures were solved with
SHELXT 2014[74] and refined on F2 with SHELXL[75] on Yadokari-XG[76] or ShelXle.[77] The
program ORTEP-3[78] was used to generate the X-ray structural diagram.
Powder X-Ray Diffraction Analysis: The analyzed PXRD data were measured using Rigaku
SmartLab with a Ge (111) Johannson monochromator (Cu Kα1 radiation λ = 1.54059 Å, 40 kV
and 30 mA, transmittance mode), a D/tex Ultra detector, and a capillary rotation attachment. The
analyses were carried out with the powder X-ray diffraction analysis software PDXL2. For all the
compounds, indexing programs DICVOL[79] or ITO[80] were used for the determination of cell
dimensions. The diffraction profiles were decomposed with the whole powder pattern fitting
(WPPF) method. Following these data, corresponding space groups were determined based on
systematic extinctions. After additional WPPF methods, the initial phases of Phases G and B were
obtained by direct space method using the DFT-optimized structures of PPB at the M06/6-
311++G(d,p) level of theory. These obtained structures were refined using the Rietveld method
under restrained conditions for bond lengths and angles obtained from the Cambridge Structural
Database System (CSDS) via Mogul Server Ver. 1.0. Restraint stiffness, sres (defined by Eq. (4)),
was determined as σNorm (defined by Eq. (8)) becomes a proximate value to one. We determined
28
This article is protected by copyright. All rights reserved.

10.1002/chem.202101107
Chemistry - A European Journal
Manuscript for Chemistry—A European Journal
converged structure, confirming residual sum of squares, Rwp, was less than 10%, and goodness-
of-fit indicator, S, sufficiently closed to unity.
Fitting parameters for the Rietveld refinement were defined as follows:
()*( )
∑
ꢆ
ꢆ
ꢈ
ꢇ' − ꢇ'
'
(1)
(2)
ꢅ& =
∑ | |
ꢇ'
'
$
()*( )
' ꢊ' ꢋꢇ' − ꢇ' ꢈ ꢌ
∑
ꢉ
ꢅ+&
=
' ꢊ' ꢇ'$
∑
ꢎ − ꢈ
∑
(3)
ꢅ =
,
ꢍ
' ꢊ' ꢇ'$
ꢅ+&
(4)
(5)
ꢏ =
ꢅ,
/
0
(
)
ꢅ = ꢅ+& + ꢐ-,. ꢅ-,. + ꢅ-,.
2
!
$
ꢓ1' − ꢓ'
ꢔ'/
ꢅ-/,. = ꢑ ꢒ
ꢕ
ꢕ
(6)
(7)
(8)
'
3
"
$
ꢖ1' − ꢖ'
ꢔ'0
ꢅ-0,. = ꢑ ꢒ
4
ꢅ-/,. + ꢅ-0,.
ꢗ + ꢎ
ꢍ
ꢔ56-7
=
,
where yi is a diffraction intensity, yical(p) is a calculated intensity, p is a parameter for the least-
squares method, wi is a weight, N is the number of data points, Md is the number of restraints for
bond length, and Na is the number of restraints for bond angles. In addition, d0i and a0j are mean
bond lengths and angles of similar structures, respectively. σid and σja are standard deviations of
mean bond lengths and angles, respectively.
29
This article is protected by copyright. All rights reserved.