close

Вход

Забыли?

вход по аккаунту

?

j.mattod.2018.07.018

код для вставкиСкачать
d
Volume xxx, Number xx
d
xxxx 2017
RESEARCH
Channel-restricted Meniscus Self-assembly
for Uniformly Aligned Growth of SingleCrystal Arrays of Organic Semiconductors
Wei Deng 1,†, Xiujuan Zhang 1,†, Huanli Dong 2,†, Jiansheng Jie 1,⇑, Xiuzhen Xu 1,
Jie Liu 2, Le He 1, Lai Xu 1, Wenping Hu 2,3,⇑, Xiaohong Zhang 1,⇑
1
Institute of Functional Nano & Soft Materials (FUNSOM), Jiangsu Key Laboratory for Carbon-Based Functional Materials & Devices, Soochow
University, Suzhou, Jiangsu 215123, China
2
Beijing National Laboratory for Molecular Science, Key Laboratory of Organic Solids, Institute of Chemistry, Chinese Academy of Sciences, Beijing 100190,
China
3
Tianjin Key Laboratory of Molecular Optoelectronic Science, Department of Chemistry, School of Science, Tianjin University & Collaborative
Innovation Center of Chemical Science and Engineering, Tianjin 300072, China
Organic semiconductor single-crystal (OSSC)-based field-effect transistors (FETs) with high mobility
and small device-to-device variation enable OSSCs to be adapted for practical applications. Research
attention has recently been focused on developing simple ways of fabricating large-area OSSC arrays by
means of solution-coating techniques. However, the lack of control of the meniscus front, where the
nucleation and growth of organic crystals occur, leads to inconsistent crystal alignment and
consequently induces large variation in device performance. Here, we propose a universal strategy,
termed the channel-restricted meniscus self-assembly (CRMS) method to fabricate ultrahigh-mobility,
uniform OSSC arrays. The microscale photoresist channels used in this method produce a confinement
effect to reduce the size of the meniscus, enabling the homogeneous nucleation of OSSCs at the
meniscus front. Meanwhile, the dip-coating process ensures consistent molecular packing in the OSSCs
and thus guarantees their highly uniform electrical properties. Using 2,6-diphenylanthracene as an
example, wafer-scale (>2 inch) OSSC arrays with very small size variations (10%) are successfully
prepared, which is very difficult to achieve by using the previously reported methods. As a result, fieldeffect transistors (FETs) based on the OSSC arrays show a high average hole mobility of up to 30.3 cm2 
V1 s1 with good uniformity among devices. This method is general for the growth of various OSSC
arrays, facilitating the applications of OSSCs in large-area, high-performance organic electronic
devices.
Keywords: Channel-restricted meniscus self-assembly; Organic semiconductor; Single-crystal array; High mobility;
Field-effect transistors (FETs)
Introduction
⇑ Corresponding authors.
E-mail addresses: Jie, J. (jsjie@suda.edu.cn), Hu, W. (huwp@tju.edu.cn), Zhang, X.
(xiaohong_zhang@suda.edu.cn).
†
The authors contributed equally to this work.
The well-organized molecular packing and exceptionally high
chemical purity in organic semiconductor single crystals (OSSCs)
dictate their physical and electronic properties [1–9]. Assembling
OSSCs into aligned structure has been the aim of research efforts,
which is essential for their practical applications [10]. Recently,
1369-7021/Ó 2018 Elsevier Ltd. All rights reserved. https://doi.org/10.1016/j.mattod.2018.07.018
1
Please cite this article in press as: W. Deng et al., Materials Today (2018), https://doi.org/10.1016/j.mattod.2018.07.018
RESEARCH: Original Research
Materials Today
RESEARCH
RESEARCH: Original Research
many solution-coating techniques, such as drop casting, dip
coating, and solution shearing, have demonstrated the assembly
of aligned OSSCs with high efficiency [11–13]. However, the
OSSC arrays assembled from these methods exhibit in general
non-uniform crystal morphologies, qualities, and growth orientations, leading to great variations in electronic properties
[14,15]. This largely hinders their practical application in devices.
In these methods, a unidirectional force is usually utilized to control the motion of the meniscus, where organic nuclei are formed
due to the highest solution evaporation rate. However, the areas
of the meniscus fronts in these techniques (width of several millimeters to several centimeters or larger) are much larger than the
size of OSSC nuclei (several tens to several hundreds of nanometers) [16], multiple nucleations occur at the meniscus front.
The location, density, and size of the nuclei are highly sensitive
to fluctuations in the processing conditions, resulting in heterogeneous nucleation behavior and polymorphs molecular stacking. Therefore, crystal arrays with consistent structures and
highly uniform electronic properties are difficult to achieve. In
this regard, precise control/confinement of the meniscus is
essential to obtain uniform OSSC arrays.
In this study, we report a strategy that combines microscale
photoresist (PR) channels with the dip-coating process, termed
channel-restricted meniscus self-assembly (CRMS), for the
wafer-scale growth of 2,6-diphenylanthracene (DPA) single-crystal arrays with consistent crystal morphologies, qualities, and
growth orientations. In this method, the microscale PR channels
provide a micro-confinement effect to control the width and
shape of the meniscus, enabling the homogeneous nucleation
of OSSCs at the meniscus front. Additionally, the dip-coating
process guides the unidirectional motion of the meniscus to
achieve the uniform growth of organic crystals with consistent
molecular stacking. The resulting DPA single-crystal arrays possess highly uniform crystal quality and electronic properties,
and FETs based on the OSSC arrays showed a high average carrier
mobility of 30.3 cm2 V1 s1, 80% of the maximum mobility
(39.3 cm2 V1 s1), and a small device-to-device variance of only
5.36 cm2 V1 s1. This simple method for the wafer-scale assembly of highly aligned OSSCs with homogeneous crystal and electrical properties opens opportunities for practical applications of
OSSCs in integrated electronic and optoelectronic devices.
Result and discussion
Conventional solution-coating techniques, involve the use of an
oriented force to align the growth of OSSCs [17–19]. Nevertheless, there are several challenges in controlling aligned growth
of OSSCs in solution-coating methods, a major one is the precaution of non-uniform crystal orientations during the growth process. For example, dip-coating technique features large meniscus
front (usually at milli/centimeter-size), bringing about heterogeneous nucleation behavior and polymorphs molecular stacking
during the crystal growth process (Fig. 1a). This often leads to
the dendritic growth with an inconsistency of crystallographic
alignment (Fig. 1b), which hinders efficient charge transport
due to anisotropic charge transport at the basal plane of OSSCs.
To address this issue, we proposed a CRMS strategy that involves
patterning of the substrate with microscale PR channels to
Materials Today
d
Volume xxx, Number xx
d
xxxx 2018
restrict the meniscus to reduce the size of meniscus front
(Fig. 1c). The typical PR channels have the width of 5 lm and
depth of 800 nm (Fig. S1, Supporting Information). When the
PR channels are immersed into organic solution, a contact line
in the shape of a meniscus profile between two adjacent PR
stripes will be formed due to surface tension effects. Note that
the meniscus is restricted within the PR channel due to the
height of the PR stripes and the size of meniscus front diminishes
to hundreds of nanometers. During the dip-coating process,
some of the solution will be pinned beside the lateral surfaces
of the PR stripes because of the existing friction/viscous force
(Fv) between them. This will decrease the meniscus profile radius
of the contact line, as illustrated in Fig. 1c. Computational fluid
dynamics (CFD) simulations of the solution around the PR channels (Figs. 1d and S2, Supporting Information) validate our
design concept, revealing that the microscale PR channels are
effective in restricting the meniscus and reducing the size of
meniscus front.
The restricted meniscus with small-size front offers an opportunity to rationally control evaporation and convective flow of
the solution in the PR channels. As such the homogeneous
nucleation can occur at the meniscus front, enabling the highly
aligned growth of OSSC array with uniform morphology and orientation. The optical micrograph of confined meniscus profile, as
shown in Fig. 1e, can be theoretically described by a parabolic
equation h = A(R r)2 [20], where A is a constant, R is the distance of the edge of the meniscus from the bulk solution (R =
3.7 lm in this experiment), r is the horizontal distance of the
meniscus surface from the bulk solution, and h is the height of
the meniscus at r. We fit the experimentally measured outline
of the meniscus curve with the parabolic equation to obtain A
= 0.057 (the dark dashed line in Fig. 1e). With solvent evaporation, outward convective flow is triggered. The velocity (v) of
the outward convective flow aroused by solvent evaporation
can be determined by the Navier–Stokes equation [21]:
Z r
1
dh
dr
ð1Þ
vðr; tÞ ¼
J e ðr; tÞr þ qr
qrh 0
dt
where Je is the evaporative flux of the solvent, t is the time, and q
is the density of the solution. As the volume of the bulk solution
is sufficient for completing the dip-coating process, the shape of
the meniscus remains relatively unchanged over time, and thus,
the dh/dt term can be neglected. According to the nonequilibrium one-sided (NEOS) model, (i) the mass flux Je obeys
the Hertz–Knudsen law, (ii) the temperature and the pressure satisfy the Clausius–Clapeyron relation, and (iii) reaction-limited
evaporation occurs at the interface at non-equilibrium [22]. Thus,
v can be deduced by:
!
Z r
1
r
dr
ð2Þ
vðrÞ ¼
qrh 0 0:057ðR rÞ2 þ K þ W
where K is the non-equilibrium parameter and W is the thermal
effect. According to Eq. (2), a plot of v as a function of r can be
extracted, as shown in Fig. 1f. It is seen that the outward convective flow is sharply accelerated at r 3.0 lm, and v reaches a maximum at the meniscus edge R (measured to be 3.7 lm), implying
that most of the solute will preferentially flow to the meniscus
front. Thus, organic molecules will first reach supersaturation
2
Please cite this article in press as: W. Deng et al., Materials Today (2018), https://doi.org/10.1016/j.mattod.2018.07.018
d
Volume xxx, Number xx
d
RESEARCH
xxxx 2018
RESEARCH: Original Research
Materials Today
FIGURE 1
Channel-restricted meniscus self-assembly strategy. (a) Schematic illustration of the dip-coating growth process of OSSC arrays. (b) Cross-polarized OM of the
DPA crystal arrays coated from its DCB solution. These crystals show different intensities of cross-polarized light, indicating an inconsistency of
crystallographic alignment. (c) Schematic illustration of the CRMS growth process of OSSC arrays. (d) CFD simulation of the formation of a confined meniscus
within the microscale PR channels. (e) OM image of the confined meniscus profile between two PR stripes (left) and a schematic diagram of the meniscus
(right). (f) Plot of the velocity v of the outward convective flow as a function of the distance r between the meniscus surface and the bulk solution.
and precipitate out as nuclei at the meniscus front along the lateral side of the PR stripes (Fig. 1c), which become the nucleation
sites as well as the initial growth sites of the OSSCs. With the
micro-PR channel as a template, the meniscus front can be effectively confined along the lateral sides of the PR stripes, and the
meniscus area is largely reduced. This ensures the exclusive nucleation and growth of OSSCs along the lateral sides of the PR stripes
to avoid the heterogeneous nucleation that usually occurs in conventional solution-processed methods. In the subsequent dipcoating process, the operating tensile force (Ft) for substrate liftoff will pin the solution meniscus edge at the lateral surface of
the PR stripes. With the outward flow persistently transporting
solute to the edge of the meniscus, the continuous epitaxial
growth of crystals along the PR sidewalls will occur, leading to
the formation of ultralong crystals (Fig. 1c).
Using organic semiconductor DPA as an example, we first
implement the CRMS strategy in wafer-scale growth of OSSC
arrays with high mobility and uniformity (Fig. 2a). In a typical
process, the PR patterned substrate was dipped into a DPA/1,2dichlorobenzene (DCB) solution (4 mg/mL) at 120 °C and
removed at a controlled speed of 150 lm/s. The DPA crystals
grew along both sides of the PR stripes within the channels,
forming highly aligned, parallel crystal arrays on the substrate.
Fig. 2b displays a photograph of the DPA crystal arrays on a
3
Please cite this article in press as: W. Deng et al., Materials Today (2018), https://doi.org/10.1016/j.mattod.2018.07.018
RESEARCH
Materials Today
d
Volume xxx, Number xx
d
xxxx 2018
RESEARCH: Original Research
FIGURE 2
Wafer-scale growth of DPA crystal arrays. (a) Schematic illustration of the wafer-scale growth of DPA crystal arrays by the CRMS method. The inset image
depicts the molecular structure of DPA. (b) Photograph of the DPA crystal arrays grown on a 2-inch SiO2/Si wafer. (c) Typical cross-polarized OM image of the
crystal arrays on a SiO2/Si wafer substrate. (d) SEM image of the crystal arrays on a SiO2/Si substrate. The inset shows the cross-sectional SEM image of the
crystal array. (e) AFM image of a crystal, indicating that the crystal is deposited at the side of the PR stripe. Histograms of width (f) and thickness (g) of the DPA
crystals on a 2-inch SiO2/Si wafer.
2-inch SiO2/Si wafer. The morphologies of the crystal arrays were
studied by cross-polarized optical microscopy (OM) and scanning electron microscopy (SEM), as shown in Fig. 2c and d,
respectively. The OM images taken from different positions on
the wafer show the high uniformity of the product (Fig. S3, Supporting Information). From the cross-sectional SEM image of the
DPA crystals (inset in Fig. 2d), it can be seen that the crystals were
tightly grown along the PR stripe with uniform thickness and
width. All images clearly indicated that the parallel crystals were
preferentially deposited along two sides of the PR stripes in the
channels. The atomic force microscopy (AFM) image (Fig. 2e)
and corresponding height distribution diagram (Fig. S4, Supporting Information) revealed a step between the PR and DPA crystal
due to their different thicknesses. The surface of the crystals was
very smooth with a surface roughness of 1 nm (Fig. S5, Supporting Information). More importantly, the resulting DPA crystal arrays could uniformly cover the entire SiO2/Si substrate with
a perfect periodic structure. Statistical analysis at different positions indicates very small size variations of the crystals, with a
width of 550 ± 50 nm and thickness of 200 ± 20 nm (Fig. 2f
and g). The crystal width can be readily tuned by adjusting the
pulling speed. As shown in Figs. S6 and 7, in Supporting Information, the width of the crystals was 1350 nm at 50 lm/s and
could be further decreased to 650 nm at 150 lm/s, 550 nm at
200 lm/s, and 415 nm at 250 lm/s. We note that, due to the
existence of PR stripes and empty regions between adjacent crystals, the surface coverage of DPA crystals on the substrate is estimated to be about 15–30%, depending on the crystal width. Our
CRMS method could be extended to other p-type and n-type
organic semiconductors (Figs. S8–S10, Supporting Information),
e.g., fluorinated 5,11-bis(triethylsilylethynyl)anthradithiophene
(Dif-TES-ADT),
2,7-dioctyl[1]benzothieno[3,2-b][1]benzothio
phene (C8-BTBT), and 6,13-bis((triisopropylsilyl)ethynyl)-5,7,12,
14-tetraazapentacene (TIPS-TAP). This result strongly suggests
that the CRMS method is capable of the large-scale fabrication
of uniform OSSC arrays.
We systematically examined the crystal quality of the crystal
arrays over a large area. The large-area cross-polarized OM images
show that the optical intensity of the crystal arrays has fourfold
symmetry, as expected for single crystals (Fig. S11, Supporting
Information). To further examine the molecular stacking order
and crystal structure of the DPA crystal arrays in a large region,
two-dimensional grazing incidence X-ray diffraction (2DGIXRD) was performed with a synchrotron X-ray source. In
Fig. S12, Supporting Information, we show a schematic illustration of the 2D-GIXRD experimental setup. The DPA crystal arrays
4
Please cite this article in press as: W. Deng et al., Materials Today (2018), https://doi.org/10.1016/j.mattod.2018.07.018
d
Volume xxx, Number xx
d
RESEARCH
xxxx 2018
produce a high intensity of (1 0 0) diffraction spots in the out-ofplane (qz) mode and simultaneously generate many diffraction
spots corresponding to the (0 0 1) and (0 1 1) reflections in the
in-plane (qxy) mode of 2D-GIXRD (Fig. 3a). Fig. 3b and c display
the 1D out-of-plane and in-plane profiles of the DPA crystal arrays.
The d-spacing of 17.9 Å, deduced from the out-of-plane X-ray
diffraction (XRD) pattern, corresponds to the a-axis length of
the crystallographic unit cell. Moreover, the high-order spots in
the qz and qxy directions reveal that the DPA crystal arrays are
essentially single crystalline within the X-ray irradiated region
(0.35 0.35 mm2), which is in a sharp contrast to common
polycrystalline films, which usually demonstrate diffuse intensities around the diffraction peaks along the Debye rings [23].
The DPA crystal arrays on the SiO2/Si substrate were further
transferred onto a Cu grid for transmission electron microscopy
(TEM). The detailed transfer process is described in the Method
section and illustrated in Fig. S13, Supporting Information. The
SEM image and corresponding OM image of the DPA crystal
arrays on the Cu grid are shown in Fig. 3d. The selective-area
electron diffraction (SAED) patterns acquired from nine different
regions in one DPA crystal display very ordered diffraction spots,
indicating the single-crystalline nature of the crystal (Fig. S14,
Supporting Information). The crystal growth direction is determined to be [0 0 1] from the SAED patterns. In addition, we also
examined the SAED patterns of the DPA crystals at other positions on the same substrate. Discrete diffraction points were
observed again, confirming that all the crystals were single crystalline with highly uniform crystallinity (Figs. 3e and S15, Supporting Information). The crystal structure of the DPA crystal
was further studied by high-resolution AFM, as shown in
Fig. 3f. It was observed that DPA molecules were packed in a rectangular lattice with b = 7.44 Å and c = 6.34 Å in two orthogonal
directions. The corresponding fast Fourier Transform (FFT) pattern is consistent with the SAED patterns obtained by TEM.
These experimental results reveal that the DPA crystals have an
orthorhombic crystal structure with lattice parameters of a =
17.9 Å, b = 7.44 Å, c = 6.34 Å, and a = 90°, b = 90°, c = 90°. Compared to the structure of large single crystals [24,25], the DPA
crystals have a more stretched molecular packing structure.
Due to the molecular lattice strain, the distance of p–p stacking
FIGURE 3
Crystal quality of DPA crystal arrays. (a) 2D-GIXRD pattern of DPA crystal arrays. (b) and (c) show the 1D out-of-plane and in-plane profiles, respectively, from
the 2D-GIXRD results in a. (d) SEM image of the DPA crystal arrays on a Cu grid (inset is the OM image of the crystal arrays) and (e) the corresponding SAED
patterns from different crystals marked in d. Each pattern shows the single-crystalline nature of the DPA crystals. (f) High-resolution AFM image of a DPA
crystal. The inset is the corresponding FFT pattern. Molecular packing structures of (g) a large DPA single crystal and (h) a quasi-1D DPA single crystal, as
derived from the given data. The d-spacing of the p–p stacking reduced from 3.12 Å for large single crystals to 3.01 Å for the quasi-1D single crystals, which
may be attributable to an oblique molecular packing structure with a tilted angle of the molecular crystal plane.
5
Please cite this article in press as: W. Deng et al., Materials Today (2018), https://doi.org/10.1016/j.mattod.2018.07.018
RESEARCH: Original Research
Materials Today
RESEARCH
RESEARCH: Original Research
decreases from 3.12 Å for the large DPA single crystal to 3.01 Å
for the quasi-1D DPA single crystal, as indicated in Fig. 3g and
h. In addition, the stretched molecular lattice of the quasi-1D
DPA crystals has a remarkably shortened p–p stacking distance,
which can potentially increase the charge-transfer integral [26].
The large-area uniform quasi-1D DPA single-crystal arrays with
reduced p–p stacking distance show great promise for highperformance electronic devices.
It should be noted that two important factors are essential for
the growth of the DPA single-crystal arrays. First, the PR stripes
are vital to guide the precise positioning of OSSC growth. If a
bare SiO2/Si wafer was used as the growth substrate with other
parameters kept constant, irregular needle-like nanoribbons were
obtained at uncontrollable locations (Fig. S16, Supporting Information). Second, the tensile fore (Ft) determines the formation of
quasi-1D OSSCs in the dip-coating process. Without the external
force, 2D microdisks were produced on the PR patterned substrate under the same conditions (Fig. S17, Supporting Information). Only when the two factors are simultaneously satisfied
can the DPA single-crystal arrays be obtained. These experimental results are consistent with theoretical calculations by the
Hartman theory, as shown in Fig. S18, Supporting Information.
DPA molecules tend to self-assemble into 2D crystals due to an
extensive network of CAH p interactions existing in the b–c
plane of the crystal [26]. Strong charge anisotropy will occur in
these crystals; that is, the 2D crystals possess remarkably different
electrical characteristics along different orientations. Therefore,
these randomly orientated 2D microdisk-based FETs would
unavoidably exhibit obvious device-to-device variation in performance, which is detrimental to their practical applications. However, as a shear strain applies in the b-c plane, corresponding to
the Ft in dip-coating process, 1D growth becomes more favorable
(Fig. S18, Supporting Information). DPA nuclei formed at initial
growth stage preferred to align along the longest axis, i.e., the
[0 0 1] direction, due to the higher shearing force along this
direction. The subsequent DPA crystal growth would then occur
along this growth orientation, while the growth along other
orientations was restricted. Due to the exclusive growth orientation, the electrical characteristics of the quasi-1D crystal arrays
are much more uniform than those of 2D crystals, thus guaranteeing the high uniformity and reproducibility of the device
performance.
We fabricated bottom-gate FETs based on DPA single-crystal
arrays on SiO2 (300 nm)/Si substrates by depositing MoOx/Au
(15 nm/100 nm) source-drain (S-D) electrodes through a shadow
mask. The insets of Fig. 4a show the OM images of the DPA single-crystal array-based FETs. Note that the channel length (L) and
full width (W) are 25 and 150 lm, respectively. The effective
channel width (Weff) is determined to be 26.4 lm by summing
the width of all the crystals in the channel, which is 17.6%
of the full device channel. The transfer and output characteristics
of the optimum FET are shown in Fig. 4b and c, respectively.
From the transfer characteristics, the field-effect mobility calculated from saturation region reached 39.3 cm2 V1 s1 (on the
basis of Weff), with an on/off ratio (Ion/Ioff) of 107, a threshold
voltage (VT) of 11.6 V and a subthreshold swing of 0.41 V
dec1. The FET also shows a high linear-regime mobility of 30
Materials Today
d
Volume xxx, Number xx
d
xxxx 2018
cm2 V1 s1 with a VT of 13 V, as shown in Fig. S19, Supporting
Information. We note that charge transport in the DPA singlecrystal arrays shows negligible gate voltage dependence, suggesting that the mobility value extracted from the traditional transistor equations is reliable (Fig. S20, Supporting Information).
Additionally, the behavior of the OSSC array-based FETs is highly
consistent with the physical model of FET operation (the Shockley model, that is, a FET with a linear |Ids|0.5 vs Vg). Therefore, the
mobilities in this case can, in principle, be reliably extracted from
the slope of the red line in Fig. 4b [27]. Note that the mobility is
nearly 1.7-fold larger than that of self-assembled DPA microdisks
[25]. Such high mobility value can be attributed to the uniaxial
alignment of the crystals and their lattice-strained packing motif
providing efficient charge transport, as revealed by the crosspolarized OM, SAED, and 2D-GIXRD characterizations. Moreover, the small hysteresis of the transfer characteristics
(Fig. 4b), along with the good on–off cycle stability (Fig. 4d), indicates a low shallow trap density at the semiconductor–dielectric
interface as well as the robust molecular packing of the OSSC
arrays under the bias stress. Significantly, the FETs exhibit very
high device uniformity. A total of 63 devices (7 9 FET arrays)
on the same substrate were measured (Fig. 4e) (the corresponding
device characteristics are shown in Figs. S21–S24, Supporting
Information). Fig. 4f displays a 2D plot of the mobility values
of the FETs, and the color intensity represents the measured performance. It is obvious that all the devices in the arrays can successfully operate, and a majority of the devices show similar
performance, revealing an average mobility of 30.3 cm2 V1 s1.
Furthermore, the standard deviation of mobility of 63 FETs is
only 5.36 cm2 V1 s1, showing a small property variation in
DPA single-crystal arrays. The distributions of Ion/Ioff, VT, and
SS of the 63 FETs are also shown in Fig. S25, Supporting Information, revealing an average Ion/Ioff, VT, and SS of 107, 12.5 V,
and 0.34 V dec1, respectively. Table S1 in Supporting Information lists the average mobility and standard deviation of mobility
of our DPA single-crystal array-based FETs and other typical
ultrahigh-mobility (maximum mobility >20 cm2 V1 s1) FETs,
including both single-crystal-based and thin-film-based FETs
reported in the literature. We note that the average mobility
and standard deviation of mobility of the DPA single-crystal
array-based FETs are superior to the values of other organic semiconductor based high-mobility FETs [14,15,28,29]. The outstanding device uniformity of the OSSC array-based FETs is attributed
to the following reasons: (i) the PR template guarantees uniform
crystal growth, leading to the large-area production of DPA crystals with the same geometries, dimensions, and crystal qualities
and (ii) the Ft in the dip-coating process ensures the homogeneous crystal structure of the crystals and thus the high uniformity of the electrical properties of the crystals. Considering
that one of the most important applications of the highmobility DPA single-crystal arrays is their use in backplane transistors for next-generation displays, we integrated the DPA single-crystal array-based transistor and a green organic lightemitting diode (OLED) on the same glass substrate, forming a
pixel circuit for AMOLEDs (Fig. 4g). The optical images in
Fig. 4h show that the green OLED could be effectively switched
between the ON and OFF state using the OSSC array-based FET.
6
Please cite this article in press as: W. Deng et al., Materials Today (2018), https://doi.org/10.1016/j.mattod.2018.07.018
d
Volume xxx, Number xx
d
RESEARCH
xxxx 2018
RESEARCH: Original Research
Materials Today
FIGURE 4
DPA single-crystal array-based FETs for AMOLEDs. (a) Schematic representation of the FET configuration and cross-polarized OM images of the DPA crystal
array-based FETs. (b and c) The transfer and output characteristics of an optimum FET based on the DPA crystal array. (d) Bias stress stability of the highmobility DPA crystal array-based FET. (e) Histogram of the mobility (l) of 63 DPA crystal array-based FETs on the same substrate. (f) Distribution of the l
measured from 7 9 array FETs. (g) Configuration for a pixel circuit of AMOLED that combines a DPA crystal array-based FET and a green OLED. (h)
Photographs showing OLED switching between the ON and OFF state by the crystal array-based FET.
Finally, Dif-TES-ADT, C8-BTBT, and TIPS-TAP crystal arraybased FETs were also fabricated. The typical transfer characteristics of these crystal array-based FETs are shown in Fig. S26a, Supporting Information. The maximum mobilities of the Dif-TESADT, C8-BTBT, and TIPS-TAP crystal arrays were measured to
be 7.37, 10.9, and 1.38 cm2 V1 s1, respectively, which are on
par with or even higher than their single-crystalline counterparts
[18,30,31]. Meanwhile, the average mobilities of the Dif-TESADT, C8-BTBT, and TIPS-TAP crystal arrays are approximately
5.81, 8.67, and 0.98 cm2 V1 s1, respectively (Fig. S26b, Supporting Information). All the OSSC array-based FETs fabricated
by the CRMS method show superior device performance, suggesting that the CRMS method has general applicability to the
growth of OSSC arrays. We note that the mobility values in this
work were obtained on the basis of Weff for the FETs, which actually represent the intrinsic transport properties of the organic single crystals. Nevertheless, the mobility value will decrease to 5.3
cm2 V1 s1 if the full W is taken into account during the mobil-
ity estimation. Therefore, for practical device applications, the
surface coverage of the organic crystals needs to be further
improved. Hopefully, this issue could be lessened by reducing
the width of PR stripes and controlling the growth of crystals
with suitable width in the channel. On the other hand, further
improvement of the device performance could be possible by
optimizing the device structures. A limiting factor for the device
performance is the use of SiO2 dielectric. It has been demonstrated that the SiO2 dielectric with electrochemical active
AOH groups can introduce electronic trapping [32], thus leading
to degradation of device performance.
Conclusion
In conclusion, we have developed a universal CRMS method that
combines photolithography and dip-coating techniques to fabricate highly uniform, aligned OSSC arrays. More specifically, the
photolithographic PR template confines the crystal growth along
7
Please cite this article in press as: W. Deng et al., Materials Today (2018), https://doi.org/10.1016/j.mattod.2018.07.018
RESEARCH
RESEARCH: Original Research
the lateral sides of the PR stripes, leading to the large-area production of DPA crystals with the same geometries. Meanwhile,
the operating tensile force for substrate lift-off in dip-coating process ensures the homogeneous molecular packing of the crystals
and thus guarantees their highly uniform electrical properties.
Benefiting from the highly uniform crystal and singlecrystalline nature, FETs based on the DPA crystal arrays show a
high average hole mobility of up to 30.3 cm2 V1 s1 with good
uniformity among devices. The great potential of the OSSC
array-based FETs in the driving circuit for AMOLEDs has further
been demonstrated. This method provides a possible way for the
realization of highly uniform OSSC arrays with high intrinsic
mobility.
Experimental
Fluidic dynamics simulation of the formation of a confined meniscus:
The formation of a meniscus in the PR channels was modeled
using CFD, which uses numerical methods to solve problems
that involve fluid flows.
Materials: DPA and TIPS-TAP were synthesized following the
procedure reported by Jie Liu et al. and Qian Miao et al., respectively [24,33]. Dif-TES-ADT and C8-BTBT were purchased from
Sigma–Aldrich and used without further purification. A DPA
solution was prepared by dissolving 40 mg of material in DCB
solvent (10 mL, HPLC, Sigma–Aldrich) at 120 °C, while Dif-TESADT, C8-BTBT, and TIPS-TAP solutions were prepared by dissolving 30 mg of material in dichloromethane solvent (10 mL, HPLC,
Sigma–Aldrich). A negative PR (SU-8, Gersteltec Sarl, 1040) was
used in the photolithography patterning process.
Preparation of pre-patterned substrates: Highly doped n-type
SiO2/Si (resistivity <0.02 X cm, with a 300 nm thermally grown
SiO2 gate dielectric layer) substrates were used in this research.
The substrates were first soaked in 90 °C H2SO4 for 2 h and then
ultrasonically cleaned several times in acetone, isopropanol, and
deionized water in sequence for 10 min. After drying with a
stream of nitrogen, the substrates were further treated by an oxygen plasma cleaner (PVA, Ion 40) at 100 W for 300 s. Photolithography on the SiO2/Si substrates was performed with a
mask aligner (Karl Suss, MJB4) using SU-8 PR according to the following steps. First, the PR was coated on the substrate at 3500 r/
min for 35 s, and then, the substrate was baked at 120 °C for 5
min. Second, photolithography on the PR was performed by
exposing the substrate to UV irradiation for 1.8 s using a mask.
The PR was developed in developer for 20 s to obtain the PR patterns. Finally, the patterned SU-8 PR was cured by exposure to
UV irradiation for 30 s.
Growth of OSSC arrays: In a typical growth process, the prepatterned PR substrate was immersed in the DPA/DCB solution
(4 mg/mL) at 120 °C, and then, the substrate was lifted-off by a
stepped-motor at a steady speed. In addition to DPA crystal
arrays, Dif-TES-ADT, TIPS-PEN and TIPS-TAP quais-1D singlecrystal arrays were also prepared by immersing the prepatterned PR substrates in their dichloromethane solutions and
lifting off at 80 lm/s, 300 lm/s, 60 lm/s, and 80 lm/s,
respectively.
Characterizations of the aligned OSSC arrays: The crystal arrays
were characterized using a cross-polarized OM (Leica
Materials Today
d
Volume xxx, Number xx
d
xxxx 2018
DM4000M). SEM (Carl Zeiss, Supra 55) and AFM (Veeco, MultiMode V) were used to characterize the morphologies, and TEM
(FEI, Tecnai G2 F20) was used to characterize the crystalline quality of the aligned crystal arrays. The 2D-GIXRD measurements
were performed on the BL14B1 beam line (energy = 10 keV) at
the Shanghai Synchrotron Radiation Facility. The sample-todetector distance (SDD) was 340 mm, and the diffraction patterns were mostly acquired for 90 s with a two-dimensional
charge-coupled device (2DCCD) detector. The data were
distortion-corrected (theta-dependent image distortion introduced by planar detector surface) before performing quantitative
analysis on the images. Numerical integration of the diffraction
peak areas was performed with the software Fit2D.
Fabrication process and electrical measurements of the OSSC arraybased FETs: Source-drain electrodes were patterned by placing a
shadow mask that consisted of flimsy metal wires on top of the
crystal arrays and thermally evaporating (Kurt J. Lesker,
NANO36) 15 nm/100 nm MoOx/Au on top. The evaporation rate
was maintained at 0.45 Å s1. The channel length and effective
width of the mask were 25 lm and 26.4 lm, respectively. Electrical measurements were performed under air environment using a
Keithley 4200-SCS semiconductor analyzer.
Acknowledgments
W. Deng, X. J. Zhang and H. L. Dong contributed equally to
this work. This work was supported by the National Natural
Science Foundation of China (Grant Nos. 51622306, 21673151,
51672180, 51633006, 51725304), the Strategic Priority Research
Program of the Chinese Academy of Sciences (XDB12030300).
The authors thank the Collaborative Innovation Center of Suzhou Nano Science and Technology, Soochow University and
Beamline BL14B1 (Shanghai Synchrotron Radiation Facility) for
providing beam time. The authors also thank Prof. Xinran Wang
and Dr. Yuhan Zhang for assistance with the high-resolution
AFM measurements.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at https://doi.org/10.1016/j.mattod.2018.07.
018.
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
B. Fraboni et al., Adv. Funct. Mater. 26 (14) (2016) 2229.
J. Wu et al., Adv. Mater. 29 (14) (2017) 1606101.
V.C. Sundar et al., Science 303 (5664) (2004) 1644.
M.A. Reyes-Martinez et al., Nat. Commun. 6 (2015) 6948.
Y. Wu et al., Adv. Mater. 28 (11) (2016) 2266.
L. Jiang et al., Adv. Mater. 20 (14) (2008) 2735.
C. Reese, Z.N. Bao, Mater. Today 10 (3) (2007) 20.
A. Yamamura et al., Sci. Adv. 4 (2018) 5758.
J. Mei et al., J. Am. Chem. Soc. 135 (18) (2013) 6724.
B. Su et al., Chem. Soc. Rev. 41 (23) (2012) 7832.
H.Y. Li et al., J. Am. Chem. Soc. 134 (5) (2012) 2760.
W. Deng et al., Adv. Mater. 27 (45) (2015) 7305.
K.S. Park et al., Adv. Funct. Mater. 23 (38) (2013) 4776.
Y.B. Yuan et al., Nat. Commun. 5 (2014) 3005.
H. Minemawari et al., Nature 475 (7356) (2011) 364.
H. Yang, J.H. Horst, New Perspectives on Mineral Nucleation and Growth,
Springer, Cham, 2017.
[17] R. Janneck et al., Adv. Mater. 28 (36) (2016) 8007.
[18] J. Chang et al., Adv. Mater. 25 (44) (2013) 6442.
8
Please cite this article in press as: W. Deng et al., Materials Today (2018), https://doi.org/10.1016/j.mattod.2018.07.018
Materials Today
Volume xxx, Number xx
d
RESEARCH
xxxx 2018
S. Galindo et al., Adv. Funct. Mater. 27 (25) (2017) 1700526.
X.D. Gu et al., Nat. Commun. 9 (2018) 534.
J. Xu et al., Phys. Rev. Lett. 96 (6) (2006) 066104.
J.P. Burelbachi et al., J. Fluid Mech. 195 (1988) 463.
A. Kim et al., Adv. Mater. 25 (43) (2013) 6219.
J. Liu et al., Chem. Commun. 51 (59) (2015) 11777.
J. Liu et al., Nat. Commun. 6 (2015) 10032.
[26]
[27]
[28]
[29]
[30]
[31]
[32]
G. Giri et al., Nature 480 (7378) (2011) 504.
H.H. Choi et al., Nat. Mater. 17 (1) (2017) 2.
D.Y. Ji et al., J. Am. Chem. Soc. 139 (7) (2017) 2734.
H. Tseng et al., Adv. Mater. 26 (19) (2014) 2993.
O.D. Jurchescu et al., Chem. Mater. 20 (21) (2008) 6733.
P.J. Diemer et al., Appl. Phys. Lett. 107 (10) (2015) 103303.
D.Q. Liu et al., Angew. Chem. Int. Ed. 52 (24) (2013) 6222.
RESEARCH: Original Research
[19]
[20]
[21]
[22]
[23]
[24]
[25]
d
9
Please cite this article in press as: W. Deng et al., Materials Today (2018), https://doi.org/10.1016/j.mattod.2018.07.018
Документ
Категория
Без категории
Просмотров
2
Размер файла
2 590 Кб
Теги
018, 2018, mattod
1/--страниц
Пожаловаться на содержимое документа