Proving Scalability of an Organic Semiconductor To Print a TFT-Active Matrix Using a Roll-to-Roll Gravure.

Organic semiconductor-based thin-film transistors' (TFTs) charge-carrier mobility has been enhanced up to 25 cm2/V s through the improvement of fabrication methods and greater understanding of the microstructure charge-transport mechanism. To expand the practical feasibility of organic semiconductor-based TFTs, their electrical properties should be easily accessed from the fully printed devices through a scalable printing method, such as a roll-to-roll (R2R) gravure. In this study, four commercially available organic semiconductors were separately formulated into gravure inks. They were then employed in the R2R gravure system (silver ink for printing gate and drain-source electrodes and BaTiO3 ink for printing dielectric layers) for printing 20 × 20 TFT-active matrix with the resolution of 10 pixels per inch on poly(ethylene terephthalate) (PET) foils to attain electrical properties of organic semiconductors a practical printing method. Electrical characteristics (mobility, on-off current ratio, threshold voltage, and transconductance) of the R2R gravure-printed 20 × 20 TFT-active matrices fabricated with organic semiconducting ink were analyzed statistically, and the results showed more than 98% device yield and 50 % electrical variations in the R2R gravure TFT-active matrices along the PET web.

Introduction

During the past two decades, printed electronics has been developed for manufacturing novel flexible and large electronic devices with low functionality, such as passive radio frequency identification tags, digital signage, e-papers, and wireless sensors.[1−3] However, to date, there are no commercial products in the category of fully printed thin-film transistor (TFT)-based electronic devices because Si-based devices prevail in the competition to commercialize TFT-based devices. To have competitiveness over Si-based devices, printed TFT-based electronic devices should prove their superior scalability to mass-produce flexible large-area devices through a roll-to-roll (R2R) printing method as a typical advanced manufacturing system[4−7] because the flexible large-area TFT-based devices are difficult to manufacture using current photolithography and vacuum deposition technologies, which are generally used in Si technology.

An R2R gravure has been recently demonstrated as a potential advanced manufacturing technology to fabricate TFT-active matrix (TFT-AM)-based tactile sensors on poly(ethylene terephthalate) (PET) roll using silver-nanoparticle-based conducting ink, BaTiO3-nanoparticle-based dielectric ink, and carbon-nanotube-based semiconducting ink.[8,9] Those inks were all formulated to meet the continuous R2R gravure printing with a printing speed of 6 m/min using a thermal curing chamber with length of 1 m under 150 °C, and R2R-printed TFTs were operated under a reasonable direct current (DC) voltage (<20 V). To meet the printing speed of 6 m/min, all inks should be cured in 5 s at 150 °C in a 1 m long drying chamber. The chamber length can be shorter or longer depending on the curing time of the employed inks. Furthermore, because all employed solvents were Environmental Protection Agency-approved for printing, the reported R2R gravure system (R2R gravure machine, PET film, and electronic inks) can be quickly adapted in a practical production line to optimize the yield of TFT-based electronic devices.

For continuous development of the expansion of the R2R gravure system to organic semiconductor (OSC)-based TFTs, the R2R gravure system, including the PET substrate and the conducting and dielectric inks, should be evaluated comparatively with organic semiconductor inks to prove the scalability of the R2R gravure system. Because the semiconducting layer in printed TFTs is very vulnerable to the condition of the printed layers, the selection of semiconductors will often lead to failure or poorly printed TFT devices through amplifying the surface properties of the printed dielectric layer and mismatched work function of the printed drain–source electrodes.[10] On the other hand, the same semiconductors usually show good electrical properties using a dielectric layer of thermally grown SiO2 or metal oxides, which can be grown by atomic layer deposition with a nanometer thickness.[11,12] Unlike SiO2 on Si wafer or metal-oxide-based dielectric layers on glass or plastic films, the surface roughness and polarity of the printed dielectric layers using BaTiO3 nanoparticle-based dielectric ink will seriously degrade the mobility of charge carriers at the interface between the printed dielectric and organic semiconducting layers.[13,14] Because our R2R gravure-printed dielectric layers usually give roughness in the range of 90–200 nm with a thickness of 1–3 μm,[8,9] a semiconductor with a low carrier mobility (<1 cm2/V s) would not show good electrical characteristics in the printed TFTs under a reasonable DC power (<20 V). Of course, there were some reports about sheet-to-sheet (S2S) and R2R gravure-printed dielectric layers with very smooth morphology and lower thickness,[15,16] but the reported gravure-printing methods are all lab scales with a pure polymer dielectric ink so that the device yield was not an issue. However, the R2R gravure system in this study is a pilot scale to really test the scalability so that the device yield is our utmost important issue. To provide the practical device yield, we needed to optimize the thickness to about 2 μm because the thinner is the printed dielectric layer, the lower is the device yield. Furthermore, the work function of R2R gravure-printed drain–source electrodes should be well matched to the Fermi level of the semiconductor to avoid serious contact resistance in the printed TFT devices.[17,18] Therefore, evaluating organic semiconductor-based inks with previously reported conducting and dielectric inks will be very valuable for not only proving the scalability of the R2R gravure system but also developing practical organic semiconducting inks for use in the R2R gravure system, where the PET substrate, silver-nanoparticle-based conducting ink, and BaTiO3-nanoparticle-based dielectric ink are kept the same.

In this study, commercially available organic semiconductors were tested to examine the scalability of the R2R gravure system, in which 250 mm wide PET roll, silver-nanoparticle-based conducting ink, and BaTiO3-nanoparticle-based dielectric ink were employed to fabricate TFT-active matrix. The reason for choosing the TFT-active matrix (TFT-AM) to prove the scalability of the R2R gravure system with organic semiconductor is the advantage of R2R gravure-printed TFT-AM over conventional photolithography method in terms of costs of manufacturing and expandability of the device size for developing digital signages and sensor arrays. Before testing the organic semiconductor with the R2R gravure system, all selected organic semiconductors were formulated into semiconducting inks and then printed using a convenient S2S gravure on the printed TFT template, where the gate and the dielectric and drain–source electrodes were all printed using R2R gravure (Scheme ). After printing the organic semiconducting inks on the printed TFT template, the organic semiconductor ink, which showed the gate effect, was selected for further fabrication into 20 × 20 TFT-AM with 10 pixels per inch (PPI) resolution using the fully R2R gravure-printing process. The resulting organic semiconductor-based TFT-AMs in the PET roll were evaluated by characterizing the electrical properties, device stability, and device yields to prove the scalability of the organic semiconductor in fully printing TFT-AMs by the R2R gravure system.

Scheme 1

Descriptive Sequences of Proving Scalability for the R2R Gravure System with Organic Semiconductor (OSC)-Based Inks

Results and Discussion

Four different organic semiconductors (poly(3-hexylthiophene-2,5-diyl) (P3HT), 6,13-bis((triethylsilyl)ethynyl)pentacene (TIPS-pentacene), poly(3,3‴-didodecylquaterthiophene) (PQT-12), and poly[2,5-bis(3-tetradecylthiophen-2-yl)thieno[3,2-b]thiophene] (PBTTT)) were formulated into gravure inks. Table shows the list of formulated organic semiconductor-based inks with their contact angles on the printed dielectric layer, ink stability, and device characteristics (mobility and on–off current ratio from the printed TFT template). Of the formulated inks, PBTTT showed a reasonable gate effect from the printed TFT template (Figure S1). This gate effect may originated from PBTTT’s liquid crystalline characteristics due to its long flexible alkyl chain on the head unit of thiophene to render better interface property.[19,20] This morphological characteristic would maintain the proximity of the conjugated units to each other for hopping carriers even on the rough surface of printed dielectric layers. On the other hand, for the other three organic semiconductors, the ordered domain of organic semiconductor would not be formed due to their molecular structures and the rough surface morphology of dielectric layer so that the charge accumulation may not be good enough to show the gate effect.[21] Because PBTTT-based semiconducting ink with 1-octanol showed better electrical performance than that of the terpineol-based one (Figure S1), the 1-octanol-based PBTTT ink was selected to further print 20 × 20 TFT-AM with 10 PPI resolution using the R2R gravure system.

Table 1

Summary of Organic Semiconductor-Based Ink Formulation and Their Electrical Characteristics on TFT Template

organic semiconductor	solvent	viscosity (Cp)	contact angle (deg)	on–off current ratio	mobility (cm2/V s)	reported best on–off ratio	reported best mobility (cm2/V s)
PBTTT	1-octanol	8.5	<10	104.1	2.2 × 10–3	103 [22]	0.1[22]
	terpineol	44.5	<10	103.8	1.1 × 10–3
TIPS-pentacene	m-xylene	0.8	<10	not observable	not observable	108 [23]	0.7–0.8[23]
	cresol	16.2	<10	not observable	not observable
P3HT	chlorobenzene	1.3	<10	not observable	not observable	104 [24]	0.04[24]
	toluene	0.9	<10	not observable	not observable
PQT-12	butyl carbitol	6.5	<10	not observable	not observable	106 [25]	0.1[25]

The topology and cross-sectional images of each R2R gravure-printed layer are summarized in Table with the average values of surface roughness and thickness. The surface roughness of the printed dielectric layer was 98.5 nm on average. This roughness is the highest value of dielectric layers reported for PBTTT-based TFTs. The thicknesses of gate electrodes, dielectric layers, and drain–source electrodes were in the normal ranges of previously reported values under the same printing condition.[8,9] Unlike the sudden height drop at the channel edge of the drain–source electrodes fabricated by vacuum deposition or photolithography, the R2R gravure-printed drain–source electrodes had an upslope of 30° starting from the channel edge (see the cross-sectional focused ion beam–scanning electron microscopy (SEM) images in Table ). This is the most characteristic feature of the R2R gravure-printed TFTs, which was not observed in fabricated TFTs using vacuum deposition and photolithography. However, the self-assembled monolayer, crucial for attaining good PBTTT-based TFT performance,[26,28] could not be printed by the R2R gravure due to the practical reason of maintaining the concept of R2R-based advanced manufacturing system. Under these practical circumstances, the printed PBTTT layers would not be very effective in transporting charge carriers in the channel because large trap sites will be generated at the interface between PBTTT and BaTiO3.[29,30] Furthermore, after PBTTT was R2R gravure-printed on the printed dielectric layers in 20 × 20 TFT-AM arrays with 10 PPI resolution (Figure ), the grain sizes of ordered PBTTT, characterized using AFM, were very small and almost looked like amorphous films (Figure d) due to the rough dielectric layers.[31,32] This was why the observed charge-carrier mobility at saturation was in the range of 1–5 × 10–4 cm2/V s with a printed dielectric capacitance of 7 nF/cm2 and a channel length of 80 μm.

Table 2

Summary of the Topology of Each R2R Gravure-Printed Layer with the Average Surface Roughness (Root Mean Square, RMS)

Figure 1

Optical images of R2R gravure-printed 20 × 20 TFT-active matrixes with 10 PPI resolutions: roll image (a), enlarged pixel images (b), single pixel image (c), and AFM image of R2R gravure-printed PBTTT on the channel (d).

The selected R2R gravure-printed 20 × 20 TFT-AM (Figure ) was spin-coated using CYTOP (CTX-SP2) (Asahi Glass Co., Japan) to retard the degradation of PBTTT under ambient condition and light of the prove station for characterizing the TFTs.[33] Without the CYTOP coating, all TFT pixels in the 20 × 20 TFT-AM quickly degraded and lost their electrical properties (Figure S2). This device instability was unexpected based on the previously reported stability under ambient condition because PBTTT was designed to have a highest occupied molecular orbital (HOMO) level of −5.1 eV to resist oxidation under moisture and UV–vis light.[34−36] After the CYTOP coating, the electrical properties of each TFT in the TFT-AM were not changed noticeably, but after a week, they degraded slowly and were lost completely (Figure S3). The major reason for the observed unexpected instability of the printed PBTTT may have originated from the solvent, that is, 1-octanol, which can donate electrons from the oxygen molecule of 1-octanol to PBTTT to reduce the HOMO level of PBTTT.[37] In fact, from the UV–vis spectroscopy study, the maximum absorption at 551 nm of PBTTT in 1-octanol shows a red shift compared to PBTTT in toluene (509 nm) (Figure S4). On the basis of the attained UV–vis absorption spectroscopy and electrical oxidation–reduction data using cyclic voltammetry (Figure S5), the calculated HOMO level of PBTTT ink was about −5.04 eV, that is, in a very vulnerable range to moisture and UV–vis light.[38] From this study, the importance of solvent selection to formulate electronic ink was proven again, as shown from previously reported results.[27]

For statistical analysis, we first selected five of 20 × 20 TFT-AM along 10 m of the printed web and then 10 TFT pixels per TFT-AM sample were selected to characterize their electrical properties (Figure S6). On the basis of the attained electrical properties from each of the five TFT-AM, the average device yield was approximately 90%. The relationship between the electrical properties of each TFT-AM along 10 m of the R2R gravure-printed web is shown in Figure on the basis of the calculated average values from each TFT-AM. The average values of mobility, threshold voltage, and transconductance were in the range of ±50% deviations, whereas the average on–off current was in the range of ±10% deviations. Furthermore, the hysteresis in the transfer output characteristics was about 0.78 V on average with less than ±32% deviations (Figure S7). On the basis of these calculated statistical data, the scalability of the R2R gravure system was easily extracted and in the range of ±50% of device deviations with the device yield of 90%. This result was also consistent with previously reported fully R2R gravure-printed carbon-nanotube-based TFT-AM with 10 PPI resolution and proved that the reported R2R gravure system, including silver-nanoparticle-based ink and BaTiO3-nanoparticle-based ink, is indeed scalable for organic semiconducting inks to manufacture TFT-AM with ±50% device deviations and 90% device yields. The device yields were related to the selected semiconducting ink because the failures of pixel TFTs were not due to a short, but an open device. Those results indirectly indicated that inhomogeneous transfer of organic semiconducting ink from the gravure cylinder to the PET web was the most important issue to improve the device yield to more than 99%. Furthermore, the deviation of on–off current ratio from TFTs in the TFT-AM along 10 m length of the web was in the range of ±10% deviations, whereas other electrical properties were in the range of ±50% deviations. Those results implied that the on–off current ratio of the devices, printed by the R2R gravure system, strongly relied on the quality of the semiconducting ink.[39] In addition, as aforementioned, the trap sites generated at the interface of gravure-printed PBTTT layer and BaTiO3 dielectric layer will degrade the transporting charge carriers at the channel. Therefore, we carried out the calculation for the trap density of TFTs along the printed 10 m length using the equation , where ci is the capacitance of the gate dielectrics and Ntr denotes the trap density, contributed from both the bulk traps and interface traps.[40] The calculated results are shown Figure S8. Due to the large trap sites, our R2R-printed device showed the modest mobility.

Figure 2

Calculated statistical data based on attained electrical properties from 10 TFTs per TFT-active matrix. Image of selected pixel point per TFT-active matrix (a), average mobility (b), average on–off current ratio (c), average threshold voltage (d), and average transconductance (e) from the 10 selected TFTs in a single TFT-active matrix at the first 2 m, second 2 m, third 2 m, fourth 2 m, and fifth 2 m along the 10 m length of the R2R gravure-printed web.

When 400 TFTs of the selected 20 × 20 TFT-AM were fully characterized (Figure ), the device yield was 98% with an average mobility of 2 ×10–4 cm2/V s, average on–off current ratio of 102.4, average threshold voltage of 4.1 V, and normalized average transconductance of 2.4 ×10–6 S/mm. The electrical properties of 400 TFTs from the single TFT-AM were also in the range of total variations (±50%) along 10 m of the web. The device properties about the actual I–V characteristics and the current density of the characterized 400 TFTs are shown in Figure S9. The calculated average saturation current density was 4.52 × 10–9 A/mm. Due to the rough and thick dielectric layer in our gravure-printing system, the attained device mobility dramatically reduced from usually reported values to 2 × 10–4 cm2/V s. The same phenomenon was also observed from the previous study about single-walled carbon-nanotube (SWCNT)-based flexible TFTs (SWCNT-TFT),[8] fabricated gate, and dielectric and drain–source electrodes using a hybrid method, such as vacuum deposition with printed SWCNT layers to yield the mobility of SWCNT-TFT in the range of 10–160 cm2/V s.[41,42] However, the mobility of R2R gravure-printed SWCNT-TFTs (150 000 TFTs) using our gravure system, optimized to print SWCNT as a semiconducting layer, was in the range of 0.05–0.3 cm2/V s along 150 m of PET roll with the device yield of 100 % due to the rough and thick dielectric layers. On the basis of those results, it would not be hard to speculate that the scalability in device yield can be kept to the other semiconductors, whereas the mobility of the devices would be varied due to the interaction between dielectric layers and semiconducting layers. As expected, for the organic semiconductor, the R2R gravure system can keep the device yield in the pilot scale of R2R gravure system, but the mobility was reduced dramatically as well. Therefore, on the basis of the results from both fully characterized 400 TFTs in a TFT-AM and 10 selected TFTs per TFT-AM, it was reasonable to extract the scalability of the R2R gravure system for printing PBTTT-based TFT-AM with 10 PPI resolution. The extracted scalability of the R2R gravure system for manufacturing PBTTT-based TFT-AM with 10 PPI resolution is summarized in Table . Furthermore, when the TFT-AM was bended with a radius of 2.1 cm, the variations of mobility, on–off current ratio, threshold voltage, and transconductance were all in the range of ±10% (Figure ). Although this variation range was slightly larger than the previously reported values,[43,44] the devices were in a stable range under the bending strain and stress because the larger variation range of the attained electrical properties may not have originated from the strain or stress of bending devices but more likely generated by the degradation of PBTTT under strong light exposure from the prove station due to the lower HOMO level of printed PBTTT. In addition, for the practical consideration about the R2R gravure-printed PBTTT-based TFT-AM, the cutoff frequency was measured on the basis of an inverter circuit with a load resistor (4 GΩ) (Figure S10). Due to high parasitic capacitances and low mobility of the printed TFT device, the low cutoff frequency of 700 mHz was observed, whereas the calculated average cutoff frequency was 27.5 ± 1.7 Hz using the equation ft = gm/2π(CGS + CGD), where CGS and CGD are parasitic capacitances.

Figure 3

Calculated statistical data based on attained electrical properties from 400 TFTs of selected 20 × 20 TFT-active matrix. Transfer characteristics of 400 TFTs in 20 × 20 TFT-active matrix (a), calculated mobility (b), on–off current ratio (c), threshold voltages (d), and transconductances (e).

Table 3

Summary of Scalability in R2R Gravure Printing for PBTTT-Based TFT-Active Matrix with 10 PPI Resolutions

parts, system, and conditions	specification
R2R gravure
PET web	thickness: 100 μm
	contact angle with water: 53°
	thermal expansion coefficient: MD: 0.5%, TD: 0.1%
	water permeation: 0.2–0.4%
web tension	5 kgf
printing speed	6 m/min
conducting ink for gate electrodes	viscosity: 500 Cp
	surface tension: 44 mN/m
dielectric ink for dielectric layers	viscosity: 120 Cp
	surface tension: 33 mN/m
semiconducting ink	viscosity: 8.5 Cp
	surface tension: 28 mN/m
conducting ink for drain–source electrodes	viscosity: 1200 Cp
	surface tension: 42 mN/m
device yield	98%
TFT mobility	average value: 2.01 × 10–4 cm2/V s
TFT on–off current	average value: 102.4
threshold voltage variation (Vth)	average value: 4.1 V
TFT hysteresis	average value: 2.2 V

Figure 4

Variation of transfer output characteristics of 20 × 20 TFT-active matrixes before and after bending by a radius of 2.1 cm.

Conclusions

In conclusion, four different organic semiconductors (P3HT, TIPS-pentacene, PQT-12, and PBTTT) were purchased and formulated into gravure inks using various solvents without any additives and then the printability was tested by using S2S gravure on the printed TFT templates to screen out the appropriate semiconducting ink to run R2R gravure. PBTTT and 1-octanol-based semiconducting ink showed the best performance from the preliminary test due to PBTTT’s liquid crystalline property. Therefore, the PBTTT-based ink was employed with the R2R gravure system, which included sliver-nanoparticle-based conducting ink and BaTiO3-nanoparticle-based dielectric ink to fully print PBTTT-based 20 × 20 TFT-AM with 10 PPI resolution along 10 m of PET web. The attained printed TFTs in TFT-AM show the low mobility originated from the rough and thick printed dielectric layers that hindered the crystal formation of PBTTT. Furthermore, unstable device properties would be generated by low-positioned HOMO level of the ink. However, along 10 m of PET web, 147 TFT-AM (58 800 TFTs), the average device yield show more than 90% with the scalable R2R gravure-printing method. The attained values of the electrical properties from the TFTs along the 10 m length of the web were statistically analyzed to prove that scalability of PBTTT-based TFT-AM with ±50% device variations can be attained using the R2R gravure system (Table ). In other words, by simply employing PBTTT ink and R2R gravure system, TFT-AM with 10 PPI resolutions can be R2R-manufactured with a device yield of 90 % and ±50% variations of electrical properties under the reported R2R gravure system. To improve the printed device performance, we need to develop a better dielectric ink to print thinner and smoother dielectric layers. Also, finding an amorphous semiconductor will be very important to attain reliable printed devices.

Experimental Section

All organic semiconductor-based inks were formulated to attain a minimum viscosity of 1 Cp for use in R2P and R2R gravures (Table ). We purchased poly(3-hexylthiophene-2,5-diyl) (P3HT) and 6,13-bis((triethylsilyl)ethynyl)pentacene (TIPS-pentacene) from Sigma-Aldrich, poly(3,3‴-didodecylquaterthiophene) (PQT-12) from Solaris Chem Inc., and poly[2,5-bis(3-tetradecylthiophen-2-yl)thieno[3,2-b]thiophene] (PBTTT-C14) from Sigma-Aldrich. To meet the wetting and drying times on the printed gate dielectric layers and drain–source electrodes, terpineol (Sigma-Aldrich) and 1-octanol (Junsei Chemical Co. Ltd., Japan) were selected as a vehicle for the ink formulation, which can dissolve these four semiconductors as the vehicles for the ink formulation.

The formulated organic semiconducting ink was kept in a brown bottle and tested for R2R gravure printing. In this work, R2R gravure with two color units was manufactured by i-Pen Co., Ltd., Korea, and used under optimized printing conditions for printing each layers, as summarized in Table . First, gate electrodes were printed with a printing speed of 6 m/min and a web tension of 5 kgf using silver ink (PG-007; Paru Co., Ltd., Korea) and then dielectric layer was continuously printed with the same printing speed and web tension using dielectric ink (PD-100; Paru Co., Ltd., Korea). Second, the resulting web was rewound and then drain–source electrodes were printed with an overlay printing registration accuracy of ±20 μm with a printing speed of 6 m/min.[8] Third, after printing the drain–source electrodes, the PET web was rewound to print the PBTTT layers with a printing speed of 6 m/min and a web tension accuracy of ±0.3 kgf. Finally, the fully R2R gravure-printed PBTTT-based 20 × 20 TFT-AM with 10 PPI resolution (Figure ) was characterized every 2 m along the 10 m length of the printed PET web using semiconductor analyzer (KEITHLEY 4200) under ambient condition.