Nonvolatile organic field-effect transistor memory from pyrene-fused azaindacene regioisomers

Article abstract of DOI:10.1039/d1tc00560j

Nonvolatile memory devices based on organic materials are promising for a new generation of portable or flexible electronics. Herein, two pyrene-fused azaindacene configurational isomers, syn-B2IPIO and anti-B2IPIO, were designed and synthesized, and they are axially symmetric and centrally symmetric, respectively. These two regioisomers were applied as charge trapping elements (CTEs) for the electret layers in organic field-effect transistor nonvolatile memory (OFET-NVM) devices. Although the two regioisomers have nearly identical molecular structures, absorption spectra, and HOMO/LUMO energy levels, they exhibit distinctly different charge-trapping capabilities. The OFET-NVM devices based on anti-B2IPIO exhibit a moderate memory window of about 25 V. In contrast, the devices based on syn-B2IPIO show a more than two-fold wider memory window (～59 V). The different charge trapping behaviors of this pair of isomers are attributed to different intermolecular interactions associated with molecular symmetry, resulting in different molecular packing behaviors and film morphologies. The results reveal that controlling the molecular symmetry could be a new and efficient strategy for the design of organic CTEs for high-performance memory devices. This journal is

Full text of DOI:10.1039/d1tc00560j

Journal of
Materials Chemistry C
View Article Online
View Journal
PAPER
Nonvolatile organic field-effect transistor memory
from pyrene-fused azaindacene regioisomers†
Cite this: DOI: 10.1039/d1tc00560j
a
a
a
a
a
Duan-Wu Liu, Yamin Zhang,* Xiang-Yang Li, Qi Xiao, Wen-Jing Sun,
a
ab
Xiangfeng Shao
and Hao-Li Zhang
*
Nonvolatile memory devices based on organic materials are promising for a new generation of portable
or flexible electronics. Herein, two pyrene-fused azaindacene configurational isomers, syn-B2IPIO and
anti-B2IPIO, were designed and synthesized, and they are axially symmetric and centrally symmetric,
respectively. These two regioisomers were applied as charge trapping elements (CTEs) for the electret
layers in organic field-effect transistor nonvolatile memory (OFET-NVM) devices. Although the two
regioisomers have nearly identical molecular structures, absorption spectra, and HOMO/LUMO energy
levels, they exhibit distinctly different charge-trapping capabilities. The OFET-NVM devices based on
anti-B2IPIO exhibit a moderate memory window of about 25 V. In contrast, the devices based on syn-
B2IPIO show a more than two-fold wider memory window (B59 V). The different charge trapping
behaviors of this pair of isomers are attributed to different intermolecular interactions associated with
molecular symmetry, resulting in different molecular packing behaviors and film morphologies. The
results reveal that controlling the molecular symmetry could be a new and efficient strategy for the
design of organic CTEs for high-performance memory devices.
Received 4th February 2021,
Accepted 23rd April 2021
DOI: 10.1039/d1tc00560j
rsc.li/materials-c
have been carried out to gain insight into the structure–
property relationship of the organic charge trapping element
Introduction
Memory devices based on organic materials, such as organic (CTE) in electret memory devices, so that a good understanding
1
memristors, are showing promise for use in a new generation of the material system is still insufficient. Therefore, the design
of portable or flexible electronics. Organic field-effect transistor and development of high-performance electret materials are of
nonvolatile memory (OFET-NVM) devices have attracted great great importance for the further development of suitable mate-
9
attention recently because of their simple structure, non- rial systems for practical applications.
destructive readout and high memory performance. Great
2
,3
Research on the dielectric layer of the OFET-NVM is of
progress has been made with three main types of OFET-NVMs, critical importance for high-performance devices and circuits.
4
namely ferroelectric OFET memories, floating-gate OFET For example, Hu and co-workers used 2D mica as a single
5
6
memories, and polymer electret OFET memories. Out of the crystal electret and provided a model system to study the
1
0
three types, the OFET-NVMs based on polymer electrets can be fundamental working mechanism behind the devices. Small
manufactured via flexible and low-cost processing, and are aromatic molecules are also ideal CTE materials for the mecha-
7
therefore attractive for practical applications. Considerable nistic study, because of their well-defined structure and tunable
1
1
research has been conducted on the fabrication and structure electronic properties. A variety of solution-processable small
optimization of the electret OFET-NVMs device to achieve a molecules have been developed as CTE materials for OFET
8
12
high performance. However, relatively few comparative studies memory applications, such as fullerene, perylene diimide
2
derivatives, and so on. Several research groups, including us,
have devoted much effort to exploring the structure–property
correlation of small aromatic molecule CTE materials. Lee and
a
State Key Laboratory of Applied Organic Chemistry, Key Laboratory of Special
Function Materials and Structure Design, College of Chemistry and Chemical
co-workers reported that a large steric hindrance could produce
more charge trapping sites, which can enhance the charge
trapping capability of the memory device. Meanwhile, other
Engineering, Lanzhou University, Lanzhou 730000, P. R. China.
E-mail: zym@lzu.edu.cn, haoli.zhang@lzu.edu.cn
b
13
Tianjin Key Laboratory of Molecular Optoelectronic Sciences,
Department of Chemistry, Tianjin University, Collaborative Innovation Center of
Chemical Science and Engineering (Tianjin), Tianjin 300072, P. R. China
Electronic supplementary information (ESI) available. CCDC 2045205. For ESI
research indicated that molecules with small dipole moments
are more likely to exhibit high charge trapping behavior in an
†
1
4
organic memoristor. Recently, Huang and co-workers sug-
gested that small molecules with non-planar topologies with
and crystallographic data in CIF or other electronic format see DOI: 10.1039/
d1tc00560j
This journal is © The Royal Society of Chemistry 2021
J. Mater. Chem. C

View Article Online
Paper
Journal of Materials Chemistry C
separated highest occupied molecular orbitals (HOMO) and
lowest unoccupied molecular orbitals (LUMO) can provide
separate trapping sites for holes and electrons, which may
1
5
inhibit charge leakage and promote charge trapping.
Configurational isomers refer to organic compounds with
the same chemical formula but different molecular configura-
tions, which are ideal model systems for the study of structure–
property correlation. It is known that many configurational
isomers could exhibit significantly different physical proper-
ties, including solubility, dipole moment and intermolecular
1
6,17
interaction.
Importantly, the different molecular symme-
tries represented by the regioisomers play a key role in solid-
state packing. It has been reported that a small variation in
molecular symmetry can lead to strong changes in film mor-
phology and charge transfer behavior in OFET devices and
1
8,19
OLEDs.
However, no systematic effort has been made to
evaluate the effects of molecular symmetry in the electret
OFET-NVM, so that their roles in organic memory devices
remain unclear.
Scheme 1 The synthetic route to syn-B2IPIO and anti-B2IPIO.
Herein, the design and synthesis of a pair of pyrene-fused
azaindacene regioisomers, namely syn-B2IPIO and anti-B2IPIO,
are reported. The two molecules have almost identical mole- would help to gain insights into the role of molecular symmetry
cular structure, except that the syn-B2IPIO shows axial symme- in the organic memory devices.
try and the anti-B2IPIO shows central symmetry. The behavior
The synthesis of the target molecules was based on a regiose-
of these two regioisomers as CTEs for the electret OFET-NVM lective modification of pyrene. Previous work by Zhang and
have been carefully compared. It was found that the molecular co-workers has revealed that the 4,5,9,10-position (K-region) sub-
symmetry can have a strong impact on charge-trapping ability, stituted derivatives of pyrene, such as 9,10-dibromo-2,7-di-tert-
22
resulting in different performance of the memory devices. butylpyrene-4,5-dione (3), can be obtained. Using the reduction
These results help to further clarify the role of molecular reaction with sodium hydrosulfite (Na ), two alkyl chains
symmetry in organic memory devices, and establish general were added to the side of compound 3 to increase the solubility.
S O
2 2 4
guidelines for designing high-efficiency CTEs.
Then
2,7-di-tert-butyl-9,10-bis(hexyloxy)pyrene-4,5-dicarboxylic
acid (6) was obtained through cyano substitution and hydrolysis.
The target compounds syn-B2IPIO and anti-B2IPIO were obtained
by a dehydration condensation reaction using compound 6 and
1
,2,4,5-benzenetetramine tetrahydrochloride. The two isomers
Results and discussion
Molecular design and synthesis
were separated by column chromatography. The final yields of
the anti-B2IPIO and syn-B2IPIO were about 10% and 40%, respec-
Scheme 1 shows the structures and synthetic routes of the two tively. More detailed synthetic processes are provided in the ESI.†
configurational isomers that were designed for this study. Each These compounds exhibited good thermal stability with decom-
molecule has two pyrene units fused by an s-indacene-like position temperatures over 350 1C under the N atmosphere (see
2
bridge. The pyrene unit was selected because of its high Fig. S4, ESI†), which ensured their stability during the fabrication
photoluminescence (PL) efficiency and electron-rich planar of the OFET-NVW devices.
2
0,21
skeleton.
Pyrene-fused pyrazaacene (PPA) and 9,10-imide-
The first thing that was noticed was that these two isomers
pyrene-fused pyrazaacene (IPPA) derivatives have been pre- exhibit very different solubilities, which suggests that there
viously reported as n-type doping components for use in high were different intermolecular interactions. The syn-B2IPIO can
1
4
performance memory devices. Here, the rigidization of 2,6- be well dissolved in most common organic solvents, which is
bis(pyren-4-yl)-s-indacene fragments ensure the planarity of the very advantageous for the preparation of devices by the solution
p-system. The introduction of electron-withdrawing ketones method. In comparison, the anti-B2IPIO exhibited a lower
and multiple N atoms ensures a lower LUMO energy level, solubility than the syn-B2IPIO in most solvents, suggesting a
and the hexyloxy chain connected to the two pyrene units can stronger intermolecular interaction so that it has a stronger
improve the processability and crystallinity of the material. The tendency to aggregate in solutions. Fortunately, the single
molecular configuration was controlled by the configuration of crystal of the syn-B2IPIO was obtained by slowly evaporating
the azaindacene bridge, and the two configurational isomers the mixed solvent of trichloromethane (CHCl
) and methanol
3
were named syn-B2IPIO and anti-B2IPIO. The syn-B2IPIO was (CH OH). The single crystal identification of syn-B2IPIO by
3
axially symmetric and the anti-B2IPIO was centrally symmetric. X-ray crystallographic analysis showed that it adopted a one-
It was hoped that the comparative study of these regioisomers dimensional packing mode (shown in Fig. S5, ESI†). Each of
J. Mater. Chem. C
This journal is © The Royal Society of Chemistry 2021

View Article Online
Journal of Materials Chemistry C
Paper
these columns presented a different group of the A–B–A type and anti-B2IPIO were estimated to be 2.29 and 2.35 eV, respec-
anti-symmetrical sandwich stacking shape. The crystal struc- tively. The cyclic voltammetry (CV) curves of the two com-
ture showed that the tert-butyl group in syn-B2IPIO hindered pounds are shown in Fig. S3 (ESI†) and the detailed
the p–p stacking in the solid state, which explained its low electrochemical parameters are listed in Table S1 (ESI†). It
tendency to aggregate in solution.
can be seen that the HOMO energy levels of these two com-
pounds were nearly identical, and the LUMO energy level of the
anti-B2IPIO was about 0.1 eV lower.
Spectroscopic and electrochemical properties
The optical properties of syn-B2IPIO and anti-B2IPIO were
studied by ultraviolet-visible (UV-vis) spectroscopy and PL
Theoretical calculations
All the calculations were based on the density functional theory
spectroscopy in solution and solid film states. In the CHCl
3
(DFT) using the [B3LYP/6-31G(d)] basis set to explore the effect
solution, syn-B2IPIO and anti-B2IPIO showed similar absorp-
tion peaks (Fig. 1a). The absorption peaks at about 300–350 nm
were attributed to the p–p* transition from the pyrene unit of
the compounds, whereas the absorption peaks after 350 nm
were attributed to the n–p* transition of the heteroatom con-
taining an unsaturated part. It is worth noting that the absorp-
tion spectrum of the anti-B2IPIO showed a red-shift in both the
solution and film states when compared with the syn-B2IPIO,
suggesting that the anti-B2IPIO had a stronger intermolecular
p–p interaction than syn-B2IPIO. Under the excitation wave-
length of 430 nm, the syn-B2IPIO and anti-B2IPIO showed
emission peaks at 660 and 665 nm in the CHCl3 solution,
respectively (Fig. 1b). With an increased concentration in solu-
tions, syn-B2IPIO showed a smaller peak shift than that of anti-
B2IPIO (Fig. S1, ESI†). In the solid state, the emission spectra
syn-B2IPIO and anti-B2IPIO showed red-shifting by 34 nm and
of frontier molecular orbitals and energy levels on charge
transport. As shown in Fig. 2a, the LUMO orbitals of the two
molecules were concentrated on the imidazole unit of the
conjugated skeleton, whereas the HOMO orbitals were deloca-
lized on the entire molecular backbone. As suggested by Huang
and co-workers, separation of HOMO and LUMO in the mole-
cules can provide ideal charge capture sites and this can
enhance the charge-storing ability of memory devices. In
addition, the low-lying LUMO energy level of the two com-
pounds helped to reduce the electron injection barrier from the
1
5
2
3
6,13-bis(triisopropylsilylethynyl)pentacene (TIPS-pentacene) in
the active layer to the electret layer. Conversely, the holes
can also overcome the HOMO energy barrier between the
TIPS-pentacene and the charge trapping molecule under
the negative gate bias and then the electron injection from
the TIPS-pentacene into the molecules occurred.
39 nm, respectively, when comparing with that measured in
The electrostatic potentials (ESP) were also calculated. As
shown in Fig. 2b, the negative charges were mainly located on
the nitrogen atoms in the imidazole unit and the carbonyl
groups due to their electron-withdrawing property. The central
symmetrical structure of anti-B2IPIO had a total dipole moment
of 0 Debye. The carbonyl groups of syn-B2IPIO were located on
the same side, which gave it a dipole moment direction mainly
along the Y-axis, and the total dipole moment was 1.6 Debye.
The detailed calculation information for the dipole moments of
syn-B2IPIO and anti-B2IPIO is listed in Table S2 (ESI†). It was
reported that the dissipation of trapped charges was commonly
caused by dipole interactions, and a small dipole moment
could help the retention of the charge trapped in the electret
solutions, as shown in Fig. 1b. More detailed data from the
absorption and the emission spectra, as well as the spectra in
other polar solvents are provided in Table S1 and Fig. S2 (ESI†).
In both absorption and emission spectra, the thin films of the
two molecules exhibited red shifts when compared with that of
the solutions, which indicated that there were stronger inter-
molecular interactions in the solid state. However, it was noted
that the anti-B2IPIO exhibited a larger red-shift than syn-B2IPIO
in both the absorption and the emission spectra, suggesting
that the anti-B2IPIO had a stronger tendency to aggregate and
2
4
form a closer packing than syn-B2IPIO in the solid-state.
The HOMO energy levels of the syn-B2IPIO and anti-B2IPIO
were estimated to be ꢀ5.66 and ꢀ5.62 eV, and their LUMO
energy levels were ꢀ3.37 and ꢀ3.27 eV, respectively. As a result,
the corresponding electrochemical band gaps of syn-B2IPIO
Fig. 1 (a) The UV-vis spectra of syn-B2IPIO and anti-B2IPIO in CHCl
solutions and in thin films. (b) The PL spectra of syn-B2IPIO and anti-
B2IPIO in CHCl solutions and in solid phases (excitation wavelength:
30 nm).
3
Fig. 2 (a) The HOMO and LUMO energies calculated with DFT at the
B3LYP/6-31G(d) level, and (b) the ESP and dipole moments of syn-B2IPIO
and anti-B2IPIO.
3
4
This journal is © The Royal Society of Chemistry 2021
J. Mater. Chem. C

View Article Online
Paper
Journal of Materials Chemistry C
layer which improved the memory performance of the spots were found (Fig. S8b, ESI†). However, obvious small black
1
4,25
device.
The total dipole moments of syn-B2IPIO and anti- spots can be observed in the TEM image of the PS film doped
B2IPIO were both small, and therefore can avoid the occurrence with 0.5% anti-B2IPIO (Fig. S8c, ESI†). The blend films doped
of dissipation of trapped charges, which enabled them to with 1% or 3% anti-B2IPIO showed very dense aggregation
become effective doping materials for OFET memories.
spots, which indicated that anti-B2IPIO did not have good
dispersibility in PS. Atomic force microscopy (AFM) was also
used to study the film morphology of the electret layer. Fig. 3c
Thin film morphologies
The film morphology of the organic layer had a significant shows that the thin film of PS blended with the syn-B2IPIO
2
6
impact on the performance of organic electronic devices.
A
(syn-B2IPIO@PS) presented a smooth, uniform, and hydropho-
smooth interface was an important factor for the further growth bic surface. No phase segregation of the mixed film occurred
and crystallization of the atop semiconductor layer, which during the spin coating process, which was mainly due to the
2
7
facilitated charge injection. Herein, transmission electron high solubility and weak aggregation behaviors of syn-B2IPIO.
microscopy (TEM) images were used to study the films of syn- The root-mean-square (RMS) roughness of the electret blend
B2IPIO and anti-B2IPIO doped polystyrene (PS) with different film was 0.36 nm on 5 mm ꢁ 5 mm areas, which was close to that
ratios, where the dark spots, shown in Fig. 3b, represented the of pure PS film (Fig. S7b, ESI†), and this indicated that low
aggregation of anti-B2IPIO. Compared with the TEM image of concentration doping has little effect on the electret layer
pure PS shown in Fig. S7a (ESI†), syn-B2IPIO showed almost no surface morphology. The PS anti-B2IPIO blend (anti-
aggregation behavior at 0.5% and 1% blending ratios. When B2IPIO@PS) film showed the highest roughness of 0.60 nm
the blending ratio increased to 3%, a small number of dark (Fig. 3d), which was slightly higher than that of syn-B2IPIO@PS,
which indicated that the anti-B2IPIO may have a certain degree
of aggregation in the film. Water contact angles of the syn-
B2IPIO@PS and anti-B2IPIO@PS films were 96.861 and 95.991,
respectively, as shown in the insets of Fig. 3c and d. The
hydrophobic surfaces ensure face-on growth of TIPS-
pentacene on the electret surface.
Optical microscopy and fluorescence microscopy studies
showed that the blend film of syn-B2IPIO@PS presented a smooth
and uniform surface (Fig. S6a and c, ESI†). However, there were
microscopic particles distributed in the anti-B2IPIO@PS blend
film, which emitted red light when excited (Fig. S6b and d, ESI†),
indicating that significant aggregation had taken place. It has
been reported that the phase segregation in the blend film affects
28
the charge trapping ability of the doped electret layers. The
lower solubility and higher crystallinity property of the anti-B2IPIO
made it aggregate more easily during the spin coating process,
which resulted in phase segregation. In contrast, the syn-
B2IPIO@PS blend films show a weak phase segregation tendency,
which is favorable for charge storage.
Film morphologies of the TIPS-pentacene deposited on the
blend films by vacuum thermal evaporation were also studied
by AFM. The TIPS-pentacene crystal domain grown on the
blend film was about 0.2–0.3 mm (Fig. 3e and f). The small
domain size may be ascribed to the fact that the dielectric
surface hindered the lateral diffusion of TIPS-pentacene and
inhibited the continuity of the TIPS-pentacene arrangement. In
addition, X-ray diffraction (XRD) was also used to study the
crystal morphology of the TIPS-pentacene layer. The XRD
pattern of the TIPS-pentacene film showed a series of peaks,
corresponding to the reflections of (001) and (11–5) (see
Fig. S11, ESI†), which indicated that the TIPS-pentacene on
Fig. 3 TEM images of (a) 1% syn-B2IPIO@PS and (b) 1% anti-B2IPIO@PS different electret layers had a high crystallinity in the same
blend films. AFM topographic images of (c) 1% syn-B2IPIO@PS and (d) 1%
crystal phase.
anti-B2IPIO@PS blend films on SiO
2
/Si substrates, in 5 mm ꢁ 5 mm areas.
The insets of (c) and (d) are the contact angles of the B2IPIO@PS film
surfaces (side view of a drop of water). AFM topographic images of
TIPS-pentacene crystallized on the surface of (e) syn-B2IPIO@PS and (f)
anti-B2IPIO@PS, in 5 mm ꢁ 5 mm areas.
Memory device characteristics
As TIPS-pentacene was used as the active layer of the device,
all the devices showed the characteristic curve of p-type
J. Mater. Chem. C
This journal is © The Royal Society of Chemistry 2021

View Article Online
Journal of Materials Chemistry C
Paper
This result showed that the memory parameter values were
mainly determined by the intrinsic properties of the dopant.
Table S6 (ESI†) compares the memory windows of the OFET-
memory devices prepared in this work with those reported in
3
3–39
the literature,
which indicated that syn-B2IPIO was one of
the best materials reported to data.
The shift of DVth under different programming voltages
(VProg) was compared to further determine the trapping beha-
Fig. 4 Characteristic transfer curves of memory devices with (a) 1% syn- vior. When the positive gate bias voltages of 60, 80, 100, and
B2IPIO@PS blend film and (b) 1% anti-B2IPIO@PS blend film as electret 120 V were applied, the transfer curve of the syn-B2IPIO based
layers. VDS is ꢀ30 V.
device shifted with the increase of the gate bias, showing
multibit memory windows of 15.8, 31.6, 45.1 and 55.2 V
(
Fig. S17a, ESI†). When the negative gate voltage is greater than
transistors, as shown in Fig. S14 (ESI†). The OFET mobility of
2
ꢀ1 ꢀ1
ꢀ60 V, the syn-B2IPIO based memory device also undergoes a
similar negative movement (Fig. S17b, ESI†), which indicates
the devices based on syn-B2IPIO can perform multibit memory
functions in one unit to achieve high memory density. The
shifts of DVth with different VProg are plot in Fig. S17e (ESI†). All
the devices based on the syn-B2IPIO show an approximate
linear fit with DVth as a function of VProg, which was consistent
the devices was maintained at about 0.05 cm V
was comparable to the TIPS-pentacene mobility, obtained using
s , which
2
9
the same evaporation method, reported in the literature.
Fig. 4 shows the positive and negative shifts of the transfer
curves doped with the syn-B2IPIO and anti-B2IPIO when the
drain–source voltage was ꢀ30 V. After applying a positive gate
bias (V = 100 V, 100 ms), the transfer curves shifted in a
G
30
with the Fowler–Nordheim tunneling model. Moreover, the
positive direction, which was the positive ‘‘programming’’
process. In order to further verify whether memory devices
based on the syn-B2IPIO and anti-B2IPIO could be effectively
erased, the corresponding reverse gate bias was applied after
the trial programming. It was worth noting that when the
negative gate bias voltage was greater than ꢀ80 V, all the
devices were completely erased, which showed that the device
exhibited hole trapping capabilities. Subsequently, a negative
slope of the syn-B2IPIO was larger at a low voltage, indicating
that the electronic storage capacity was stronger than that of
the anti-B2IPIO.
The following equation:
Ci
Dn ¼ DVth ꢁ e
(1)
gate bias (V
G
= ꢀ100 V, 100 ms) was applied to the devices, and
is used to calculate charge trapping density (Dn), where e is the
it was set to the negative ‘‘programming’’ process. The memory
window (DVth) is defined as the difference between the Vth of
the programming and erasing states. Both syn-B2IPIO and anti-
B2IPIO exhibited charge storage capacity, and the total memory
windows obtained were higher than that of pure PS memory
devices (Fig. S15, ESI†), which are approximately 59 and 25 V,
respectively, in electron (100 V) and hole (ꢀ100 V) capture
modes. In the programming process, the programming time
of the syn-B2IPIO-doped device was gradually changed, and the
DVth did not change significantly (Fig. S18a, ESI†). This phe-
nomenon showed that syn-B2IPIO provided a strong capture
site, thus exhibiting a fast capture speed. In addition, OFET
memory devices based on PS films blended with 0.5% and 3%
of syn-B2IPIO and anti-B2IPIO as charge trapping layers were
also fabricated to investigate the change of the memory win-
dows (see Fig. S16, ESI†). The transistor and memory charac-
teristics of various wt% B2IPIO are shown in Table S5 (ESI†).
Although the memory windows changed under different dop-
ꢀ
19
element charge (1.602 ꢁ 10
C), C is the total capacitance per
i
unit area (calculated from the equation: C = 1/C + 1/CSiO2).
i
M
The Dn of the memory devices indicated that when the positive
gate bias (100 V) and negative gate bias (ꢀ100 V) were applied,
the devices showed electron and hole trapping capabilities,
respectively. More importantly, the syn-B2IPIO had a high
electron capture density, and thus a exhibits better charge
memory performance. Table 1 summarizes the characteristics
of memory devices based on the 1% doped syn-B2IPIO and
ing concentrations, devices based on syn-B2IPIO always showed Fig. 5 Data retention of the memory devices with (a) 1% syn-B2IPIO@PS
better charge trapping capabilities than those of anti-B2IPIO. blend film and (b) 1% anti-B2IPIO@PS blend film as the electret layer.
Table 1 Parameters of OFET memory devices
2
ꢀ1 ꢀ1
ꢀ2
12
ꢀ2
Compound
m [cm V
s
]
Vth [V]
Capacitance [nF cm
]
ION/IOFF
Memory window [V]
Dn [10 cm ]
5
1
1
% syn-B2IPIO
% anti-B2IPIO
0.058 (0.043)
0.061 (0.036)
ꢀ1.3 (ꢀ3.6)
ꢀ3.9 (ꢀ6.8)
9.58
8.87
4104
410
59.2 (53.4)
25.5 (22.3)
3.54 (3.19)
1.41 (1.23)
This journal is © The Royal Society of Chemistry 2021
J. Mater. Chem. C

View Article Online
Paper
Journal of Materials Chemistry C
Fig. 6 (a) Schematic diagrams of the configurations of OFET-NVM devices based on syn-B2IPIO (left) and anti-B2IPIO (right), and (b) an illustration of
charge transfer in the OFET-NVM device.
anti-B2IPIO, where each average value was obtained from more was still difficult to totally avoid aggregation in the blend film.
than 20 working devices.
The relatively weaker intermolecular p–p interaction of syn-
B2IPIO avoided molecular aggregation in the blend films and
that provided more charge trapping sites and improved the
charge trapping density. As a result, the OFET-NVM based on
the syn-B2IPIO showed a much higher memory performance
than the devices based on anti-B2IPIO.
Reversibility and retention behaviors
The memory characteristics of the devices were evaluated by the
retention time tests. Retention time is a key indicator of
the stability of charge storage, and it reflects the time that
the charge remains in the charge trap layer. Fig. 5 shows that
all the devices exhibited good retention characteristics. The IDS Conclusions
4
could be maintained for at least 10 s with a VDS of ꢀ30 V. The
In summary, two pyrene-fused azaindacene configurational
on/off ratio gradually decreased with time, which was mainly
due to the increase in the cut-off current. Due to the high
mobility of the memory devices, the on/off ratio of the syn-
isomers, syn-B2IPIO and anti-B2IPIO, were designed and
synthesized via an effective synthetic method. These two iso-
mers have same molecular formulae and similar absorption
spectra and HOMO/LUMO energy levels. However, when used
as CTEs in OFET-NVM devices, they exhibit very different
charge-trapping capabilities. OFET-NVM devices based on syn-
B2IPIO show a much wider memory window (B59 V) and
higher charge trapping density than those based on anti-
B2IPIO. In this system, axisymmetric syn-B2IPIO exhibits high
solubility and moderate intermolecular interactions and, con-
sequently, much superior performance as a CTE material when
compared with centrosymmetric anti-B2IPIO. This investiga-
tion reveals that molecular symmetry can play a crucial role in
the charge capture behaviors of CTEs in OFET-NVMs.
Therefore, careful structural design and the optimization of
molecular symmetry are essential for future molecular design
for high-performance organic memory devices.
5
B2IPIO based memory devices was maintained at 10 during
the test time range. Write–read–erase–read (WRER) cycles of
the memory devices showed stable switching behaviors of more
than 100 times, indicating that these devices can maintain
good reversibility, as shown in Fig. S19 (ESI†). These results
illustrate the excellent data retention performance of memory
devices doped with syn-B2IPIO and anti-B2IPIO.
It is worth noting that the rational design of small molecule
CTEs for improved the charge storage density and endurance is
still a challenge. Several factors have been discussed in pre-
vious reports in the literature that may affect the performance
of memory devices, such as molecular energy level, dipole
moments, intermolecular interactions, and molecular
3
1,32
packing.
Herein, particular attention was paid to the effects
of molecular symmetry. In this research it was demonstrated
that the syn-B2IPIO and anti-B2IPIO, which have nearly iden-
tical structure but different molecular symmetry, exhibited
distinctly different charge storage performances. This was
attributed to the observed molecular symmetry effects of the
Experimental
Chemicals and materials
symmetry-induced differences in the intermolecular p–p inter- Unless otherwise stated, all solvents and chemicals were
action, which was reflected in their different solubility and obtained from commercial sources (TCI America or J&K Che-
1
13
aggregation behaviors. During the solution-based device fabri- micals) without further purification. The H-NMR and C-NMR
cation, a higher solubility helped to produce ideal film mor- spectra were recorded on a Bruker AVANCE III 400 MHz/
phology that facilitated the charge injection from the organic 600 MHz instrument using CDCl3 as the solvent and the
semiconductor layer into the electret layer (Fig. 6a and b). measurements were carried out at room temperature. Mass
Although efforts were made to improve the solubility of anti- spectrometry (MS) data were obtained with a Bruker Esquire
B2IPIO in solution and optimize the film casting conditions, it 6000 mass spectrometer with an ESQ6K operator. The thermal
J. Mater. Chem. C
This journal is © The Royal Society of Chemistry 2021

View Article Online
Journal of Materials Chemistry C
Paper
properties of the compounds were examined by thermogravi- 22005128), the 111 Project, and the Fundamental Research Funds
metric analysis (TGA) on a Linseis STA PT 1600. The heating for the Central Universities (lzujbky-2020-44).
ꢀ1
rate was 10 1C min under a flowing N atmosphere. Single
2
crystal X-ray diffraction measurements were conducted on a
Bruker D8 VENTURE Kappa Duo PHOTON II CPAD diffract-
ometer with graphite-monochromatized Cu Ka radiation
Notes and references
2
(l = 1.54184 Å) under a flowing N atmosphere.
1
Y. Liu, Y. Yang, D. Shi, M. Xiao, L. Jiang, J. Tian, G. Zhang,
Z. Liu, X. Zhang and D. Zhang, Adv. Mater., 2019,
Spectroscopic and electrochemical measurements
31, 1902576.
The UV-vis and PL spectroscopy of syn-B2IPIO and anti-B2IPIO
2
3
4
W. Li, F. Guo, H. Ling, P. Zhang, M. Yi, L. Wang, D. Wu,
L. Xie and W. Huang, Adv. Sci., 2017, 4, 1700007.
Y. J. Jeong, D.-J. Yun, S. H. Noh, C. E. Park and J. Jang,
ACS Nano, 2018, 12, 7701–7709.
Q. Zhao, H. Wang, Z. Ni, J. Liu, Y. Zhen, X. Zhang, L. Jiang,
R. Li, H. Dong and W. Hu, Adv. Mater., 2017, 29, 1701907.
C. Yan, J. Wen, P. Lin and Z. Sun, Small, 2019, 15, 1804156.
Y.-C. Chiu, T.-Y. Chen, C.-C. Chueh, H.-Y. Chang,
K. Sugiyama, Y.-J. Sheng, A. Hirao and W.-C. Chen,
J. Mater. Chem. C, 2014, 2, 1436–1446.
J. Aimi, P.-H. Wang, C.-C. Shih, C.-F. Huang, T. Nakanishi,
M. Takeuchi, H.-Y. Hsueh and W.-C. Chen, J. Mater. Chem.
C, 2018, 6, 2724–2732.
A. Cui, Z. Liu, H. Dong, F. Yang, Y. Zhen, W. Li, J. Li, C. Gu,
X. Zhang, R. Li and W. Hu, Adv. Mater., 2016, 28, 8227–8233.
3
in CHCl solutions and films were measured using a Persee T6
UV-vis spectrometer (Purkinje General, China) and a FLS920
fluorescence spectrometer (Edinburgh Instruments, UK). The
frontier molecular orbital energy levels of the compounds were
estimated using CV. The oxidation potentials and reduction
potentials were measured in anhydrous CH
2
Cl
2
solution
5
6
(1.0 mM), in which 0.1 M tetrabutylammonium hexafluoropho-
6
sphate (TBAPF ) was the supporting electrolyte, a glassy carbon
electrode was the working electrode, a platinum wire was the
counter electrode, and an Ag/AgCl electrode was the reference
electrode. The extent of aggregation of B2IPIO was visualized
using a ThermoScientific Talos F200S transmission electron
microscope (TEM) operated at an acceleration voltage of
7
8
200 kV.
Fabrication of OFET-NVM devices
9 W. Hu, H. Zhang, K. Salaita and H. Sirringhaus, SmartMat,
020, 1, e1014.
0 X. Zhang, Y. He, R. Li, H. Dong and W. Hu, Adv. Mater.,
016, 28, 3755–3760.
1 S. Miao, H. Li, Q. Xu, N. Li, J. Zheng, R. Sun, J. Lu and
C. M. Li, J. Mater. Chem., 2012, 22, 16582–16589.
2 Y. Zhou, S.-T. Han, Y. Yan, L.-B. Huang, L. Zhou, J. Huang
and V. A. L. Roy, Sci. Rep., 2013, 3, 3093.
3 M.-H. Jung, K. H. Song, K. C. Ko, J. Y. Lee and H. Lee,
J. Mater. Chem., 2010, 20, 8016–8020.
14 Z.-H. Wu, W.-J. Sun, H.-H. Tian, Z.-F. Yu, R.-X. Guo, X. Shao
and H.-L. Zhang, Adv. Electron. Mater., 2019, 5, 1800598.
15 Y. Yu, L.-Y. Bian, J.-G. Chen, Q.-H. Ma, Y.-X. Li, H.-F. Ling,
Q.-Y. Feng, L.-H. Xie, M.-D. Yi and W. Huang, Adv. Sci., 2018,
2
The OFET-NVM devices were constructed using the bottom-gate
and top-contact structure. Highly doped n-type Si (100) wafers
with 300 nm thick SiO2 were used as the substrate. PS was
dissolved in toluene at a concentration of 4 mg/mL and stirred
overnight. The syn-B2IPIO and anti-B2IPIO compounds were
dissolved in the solution at a concentration of 1 wt%. Subse-
quently, the filtered, mixed solution was spin-coated on the
1
1
1
1
2
SiO
2
wafer at 3000 rpm for 30 s, and then annealed at 80 1C for
30 min. Tips-pentacene was thermally deposited to the thick-
ꢀ
1
ness of 25 nm with a deposition rate of 0.1 nm s under
ꢀ4
vacuum (3.0 ꢁ 10 Pa). A layer of 50 nm Ag was then deposited
ꢀ4
onto the substrate by thermal evaporation (3.0 ꢁ 10
Pa)
through a shadow mask. The channel width (W) and the
channel length (L) were 800 mm and 60 mm, respectively. The
cross-sectional scanning electron microscope (SEM) images
indicated the clearly layered structure of the memory devices,
as shown in Fig. S12 (ESI†). The transfer and output character-
istics of the device were tested in a glove box using a Keithley
Instruments, 4200-SCS parameter analyser.
5, 1800747.
1
6 M. Romain, D. Tondelier, J. C. Vanel, B. Geffroy, O. Jeannin,
J. Rault-Berthelot, R. M ´e tivier and C. Poriel, Angew. Chem.,
Int. Ed., 2013, 52, 14147–14151.
1
1
7 J. L. Marshall, N. J. O’Neal, L. N. Zakharov and M. M. Haley,
J. Org. Chem., 2016, 81, 3674–3680.
8 C. Zhang, Y. Zang, E. Gann, C. R. McNeill, X. Zhu, C.-A. Di
and D. Zhu, J. Am. Chem. Soc., 2014, 136, 16176–16184.
Conflicts of interest
19 D. Zhang, X. Song, M. Cai, H. Kaji and L. Duan, Adv. Mater.,
018, 30, 1705406.
2
There are no conflicts to declare.
2
0 F. Moggia, C. Videlot-Ackermann, J. Ackermann, P. Raynal,
H. Brisset and F. Fages, J. Mater. Chem., 2006, 16, 2380.
1 H. Wang, D. Zhang, X. Guo, L. Zhu, Z. Shuai and D. Zhu,
Chem. Commun., 2004, 670–671.
2
Acknowledgements
This work was supported by the National Key R&D Program of 22 Z. H. Wu, Z. T. Huang, R. X. Guo, C. L. Sun, L. C. Chen,
China (2017YFA0204903), the National Natural Science Founda-
tion of China (NSFC, 51733004, 22075117, 22073038, 21602093,
B. Sun, Z. F. Shi, X. Shao, H. Li and H. L. Zhang, Angew.
Chem., Int. Ed., 2017, 56, 13031–13035.
This journal is © The Royal Society of Chemistry 2021
J. Mater. Chem. C

View Article Online
Paper
Journal of Materials Chemistry C
23 D. Shi, Z. Liu, J. Ma, Z. Zhao, L. Tan, G. Lin, J. Tian, 32 K. Schmidt, C. J. Tassone, J. R. Niskala, A. T. Yiu, O. P. Lee,
X. Zhang, G. Zhang and D. Zhang, Adv. Funct. Mater.,
020, 30, 1910235.
T. M. Weiss, C. Wang, J. M. J. Fr ´e chet, P. M. Beaujuge and
M. F. Toney, Adv. Mater., 2014, 26, 300–305.
2
2
2
4 J. Lin, W. Li, Z. Yu, M. Yi, H. Ling, L. Xie, S. Li and 33 L. Song, Y. Wang, Q. Gao, Y. Guo, Q. Wang, J. Qian, S. Jiang,
W. Huang, J. Mater. Chem. C, 2014, 2, 3738–3743.
B. Wu, X. Wang, Y. Shi, Y. Zheng and Y. Li, ACS Appl. Mater.
Interfaces, 2017, 9, 18127–18133.
5 C. Sun, Z. Lin, W. Xu, L. Xie, H. Ling, M. Chen, J. Wang,
Y. Wei, M. Yi and W. Huang, J. Phys. Chem. C, 2015, 119, 34 L. A. Frolova, A. A. Rezvanova, V. Z. Shirinian, A. G. Lvov,
18014–18021.
A. V. Kulikov, M. M. Krayushkin and P. A. Troshin,
2
2
2
2
3
3
6 Y.-C. Chiu, T.-Y. Chen, Y. Chen, T. Satoh, T. Kakuchi and
Adv. Electron. Mater., 2016, 2, 1500219.
W.-C. Chen, ACS Appl. Mater. Interfaces, 2014, 6, 12780–12788. 35 K.-J. Baeg, Y.-Y. Noh, H. Sirringhaus and D.-Y. Kim,
7 C. Zheng, T. Tong, Y. Hu, Y. Gu, H. Wu, D. Wu, H. Meng, Adv. Funct. Mater., 2010, 20, 224–230.
M. Yi, J. Ma, D. Gao and W. Huang, Small, 2018, 14, 1800756. 36 Y. Wang, Y. Yang, P. Ding, Q. Wei, X. Gao, S. Wang, C. Liu,
8 W. Li, F. Guo, H. Ling, H. Liu, M. Yi, P. Zhang, W. Wang,
L. Xie and W. Huang, Small, 2018, 14, 1701437.
9 S. K. Park, T. N. Jackson, J. E. Anthony and D. A. Mourey, 37 S.-J. Kim and J.-S. Lee, Nano Lett., 2010, 10,
Appl. Phys. Lett., 2007, 91, 063514. 2884–2890.
0 Y. Guo, C.-a. Di, S. Ye, X. Sun, J. Zheng, Y. Wen, W. Wu, 38 Y. Park, K.-J. Baeg and C. Kim, ACS Appl. Mater. Interfaces,
G. Yu and Y. Liu, Adv. Mater., 2009, 21, 1954–1959. 2019, 11, 8327–8336.
1 R. Noriega, J. Rivnay, K. Vandewal, F. P. V. Koch, 39 M. Mart ´ı nez-Abad ´ı a, G. Antonicelli, E. Zuccatti, A. Atxabal,
A. Li, J. Yin, Y. Xia and Z. Liu, Adv. Electron. Mater., 2019,
5, 1800865.
N. Stingelin, P. Smith, M. F. Toney and A. Salleo,
M. Melle-Franco, L. E. Hueso and A. Mateo-Alonso,
Nat. Mater., 2013, 12, 1038–1044.
Org. Lett., 2017, 19, 1718–1721.
J. Mater. Chem. C
This journal is © The Royal Society of Chemistry 2021