Highly Sensitive, Ultra-Thin Dynamic Lateral Pressure Sensor Using Blue Laser Exposed Porous CNTs

Piezoresistive sensor is an essential component of wearable electronics that can detect resistance changes when pressure is applied. An effective approach to enhance its performance is to make micro-structured device. In this paper, porous carbon nanotubes (CNTs) are formed by blue laser (BL) exposure on CNTs layer, which increases its thickness ~4 times compared to the as-deposited layer. Then, the pressure sensor is fabricated by spin coating of styrene-ethylene-butylene-styrene (SEBS) elastomer on the porous CNTs layer. A 1.32 µm thick pressure sensor exhibits a high sensitivity of 6.54 x 10 6 kPa -1 , a wide sensing range of 278 Pa ~ 40 kPa, and fast response/recovery times of 900/760 µs, respectively. The stability of the pressure sensor is demonstrated by repeated loading and unloading of 20 kPa for 3600 cycles. The stretchable pressure sensor was also demonstrated using lateral CNT electrodes on SEBS surface, exhibiting stable pressure performance up to 20% stretching. Finally, a 32 x 32 active-matrix pressure sensor array is demonstrated consisting of amorphous InGaZnO 4 thin-film transistor (TFT) backplane for pressure mapping and real-time monitoring. The sensor array demonstrates dynamic area touch by pen writing with ~1 cm/s speed.


Results and Discussion
The schematic diagram of spray-coating of the CNTs and blue laser (BL) exposure on the CNTs layer to obtain porous layer are shown in Fig. 1a and 1b, respectively. The BL beam has a length of 520 µm and 20 µm width, as shown in Fig. 1c. The cross-sectional scanning electron microscopic (SEM) views of the CNTs before and after BL (laser energy density of 5.06 J cm -2 ) exposure are shown in Fig. 1d and 1e, respectively. The thickness of the as-sprayed CNTs layer is 330 nm, which is uniform over the substrate and increases to 1.32 μm upon BLA, which is almost 4 times by BL exposure. It is clearly observed that the surface morphology of CNTs layer changes significantly after BL exposure, where the top and bottom regions are very different, as shown in Fig. 1f and 1g, respectively. The morphology change of CNTs layer as a variation of incident BL energy density can be seen in Supplementary Fig. S1. With increasing the laser energy density up to 5.06 J cm -2 , the top region of the CNTs layer changes its morphology to very porous structure due to the absorption of the BL beam. The carbon-carbon bond in CNTs structure breaks under laser exposure, and the broken C atoms aggregate with the other C atoms. 57 Note that the sensitivity of the piezoresistive sensor depends on the microstructure of the porous CNTs. 44,58,59 The sheet resistance of BLA CNTs decreases with increasing BL exposure energy, as shown in Fig. 1h, from 955 ohm sq -1 to less than 490 ohm sq -1 when the BL energy density increases to 5.06 J cm -2 . The reduction in sheet resistance of the CNTs layer is mainly due to the evaporation of surfactants by the heat absorbed from BL energy and the sequential increase of electrical conductivity. 60 Cross-sectional and f, g top SEM views of the CNTs layer before and after BLA (5.06 J cm -2 ) respectively. h The sheet resistance of the porous CNTs as a function of BL energy density.
The porous CNTs film fabricated by BL exposure was coated with the highly elastic and stretchable SEBS layer to realize our pressure sensor. We considered SEBS as elastomeric matrix, of which the thickness could be controlled by changing the concentration diluted into toluene to achieve a highly sensitive, stretchable sensor. The detailed fabrication process of the porous CNT/SEBS pressure sensor is illustrated in Supplementary Fig. S2. The porous CNT/SEBS pressure sensor is simply fabricated by spincoating SEBS of 60 mg ml -1 on the porous CNTs layer of 2.5 x 2.5 cm 2 .  Fig. 2b and then transferred onto the lateral CNT electrodes on SEBS substrate. The fabrication process of stretchable pressure sensor was completed by the SEBS passivation, as shown in Fig. 2c. The detailed fabrication process of the stretchable pressure sensor can be seen in Supplementary Fig. S4. The schematic diagram for the operation of the pressure sensor is shown in Fig. 2d and 2e, with low and high pressures, respectively. Note that the top region of the CNT/SEBS film is more porous than the bottom CNT/SEBS layer due to more BL absorption at top region. Generally, the sensitivity of pressure sensor depends on the pressure range. At low-pressure region between 278 Pa to 537 Pa, the current starts to flow between stretchable electrodes through porous CNTs/SEBS sensor through near bottom CNT region, as shown using yellow line direction in Fig. 2d. High sensitivity at low pressures can be achieved by causing a large resistance change at the porous CNTs. However, at high pressure, the currents are contributed from both top and bottom regions of the CNT/SEBS film and then gradually being saturated, leading to reduced sensitivity at high pressure region, as shown in Fig. 2e. The sensitivity of the proposed pressure sensor, therefore, depends on the pressure range.
where I0 is the initial current, ΔI is the current change during the pressure loading, and P is applied pressure. The porous CNT/SEBS sensor shows a broad sensing range (278 Pa~40 kPa) with very high sensitivity of 1.13 × 10 5 kPa -1 in low-pressure region 278~537 Pa, 6.54 × 10 6 kPa -1 in medium pressure region 2~4 kPa, 6.63 × 10 5 kPa -1 in high-pressure region 5.7~38.9 kPa, respectively. High sensitivity can be achieved due to the extremely low initial current and highly sensitive current under applied external pressure. The comparison of pressure sensing performance without and with BLA on CNTs can be seen at the pressures of 2.5, 5.0, and 10 kPa, respectively, in Supplementary Fig. S6a and S6b. The sensor performance is also affected by the thickness of SEBS, as shown in Supplementary Fig. S7. When the thickness of SEBS layer is 2 μm, the sensor current is very low (less than 0.3 μA) at the high pressure of 150 kPa, but there is no response when the thickness of over 3 μm.
We compare the sensor performances and structures reported in the literatures shown in Table 1 together with our proposed device. Note that the thickness of our sensor is 1.32 µm which is extremely thin compared to the others. Thinner sensors have many advantages, such as easy adoption in mobile electronic systems including displays.    respectively, using a stretchability measurement machine. 61 Fig. 4c shows the current change of the pressure sensor with and without strain under dynamic pressure. Note that the stretchability of pressure sensor could be achieved by fabricating the pressure sensor on rigid and flexible polyimide (PI) island transferred onto SEBS substrate. Fig. 4d shows the relative current changes of the pressure sensor plotted as a function of strain. It demonstrates that the stretchable pressure sensor can be operated well at 20 % strain. The stretchable pressure sensor was mounted to the joint of a finger to mimic the motions of the human body as shown in Fig. 4e and 4f for the real-time current measurement as shown in Fig. 4g. Fig.   4h shows the current changes in the stretchable pressure sensor under different bending angles. The results show that the current increases and reaches to the maximum at 60°. A bending angle above 60° leads to a reduction of sensing current between CNT electrodes, which might be due to the reduction of sensor thickness of CNT/SEBS film. pressure sensor, makes contact through the two vias with the patterned electrodes metal layer in a pixel. Fig. 5c shows a schematic of a pixel circuit, which consists of three switching TFTs, one 1 pF capacitor, and one pressure sensor. The optical image of active-matrix (AM) sensor array can be seen in Fig. 5d, and the one-pixel dimension is 626 x 625 μm 2 . Fig. 5c shows a schematic of a pixel circuit, which consists of three switching TFTs, one 1 pF capacitor, and one pressure sensor. The optical image of active-matrix (AM) sensor array can be seen in Fig. 5d, and the one-pixel dimension is 626 x 625 μm 2 , and the contact area between the metal electrode and sensor is 30 x 400 μm 2 . Fig. 5e and f show transfer and output characteristics of a-IGZO TFT (W/L = 20/6 μm) in the pixel circuit. The threshold voltage (Vth), fieldeffect mobility (μfe), and subthreshold swing (SS) of the a-IGZO TFT are -0.3 V, 18.6 cm 2 /Vs, 0.53 V/dec, respectively. All the TFTs were used as switching units for charging, readout, and reset periods. Fig. 5g shows the timing diagram of the pixel circuit. The concept of readout circuit is to read the charges stored at the q node for the resistance change during the pressure being applied. The operation of the pixel, such as pre-charging, readout, and reset of the pixel circuit, is shown in Supplementary Fig. S8. In the precharging state, the gate signal from the n-1 th stage is applied to T1, as shown in Supplementary Fig. S8a.
Then, T1 is turned on, and the q node could be charged depending on the resistance value. The CNT/SEBS sensor can demonstrate a wide sensing range from 278 Pa to 40 kPa, corresponding current change from 2.5 nA to 20 μA as shown in Fig. 3a. Note that the off-state current of the sensor is less than 10 pA, as shown in Supplementary Fig. S5. When T2 and T3 are turned off, and capacitor is serially connected to the q node, charged voltage at q node (Vq) could be obtained from the following equation : where Vq, VDD, t, R, and C are charged voltage at q node, input voltage, gate pulse width, the resistance of pressure sensor, and capacitance of 1 pF, respectively. The Vq is decided by the n-1 th scan signal time t and the resistance of pressure sensor R. We set all scan signal time to be 20 μs for detecting the low-  Fig. S8b. Then, the n+1 th scan signal is applied to turn on T3, and the q node is discharged to VSS, as shown in Supplementary Fig S8c. The simulation results using Smart Spice are shown in Fig.  5h when the scan signal of 10 V is applied to gate for 20 μs. The charges in the q node depend on the resistance of pressure sensor.
The gate driver was integrated into the sensor array. The gate shift register (GSR) was designed and used to apply the gate pulse to the pixels, as shown in Supplementary Fig. S9. Three scan signals are connected to one pixel, as shown in Supplementary Fig. S9a. Two sequential scan lines are applied to a particular pixel. The nth scan line is divided into two parallel lines at the nth stage and connected to the particular pixel and is also connected to the n-1 th and n+1 th pixels. The timing diagram and optical image of GSR are shown respectively in Supplementary Fig. S9b and S9c. The detailed operation of GSR appears in our previous work. 62 The GSR output signals are working through the last stage, 32 nd , as can be seen in Supplementary Fig. S9d. The pressure mapping was tested using the capital letters "A", "D", "R", and "C". The 32 x 32 sensor array shows the clear pressure response without crosstalk, as shown in Fig. 5i-l. The VDD, VGS, and t used for the sensor operation are 2.84 V, 10 V, and 20 μs, respectively. The letters are softly pressed under 300 Pa to detect the characters in the low-pressure sensing region. The output signals of the pressed region are 1.32 V, which is similar to the simulation value of 1.33 V.
We also demonstrated the dynamic pressure distribution for real-time monitoring. The experimental set for displaying dynamic sensing on the screen is shown in Supplementary Fig. S10. The circuit schematic for the integrated AM sensor array can be seen in Supplementary Fig. S10a. The photographs of the customized board and zig with full integration are shown in Supplementary Fig. S10b and S10c. Using the driving board, the pressure detection can be seen in the range as shown in Video VS1, demonstrating the sensing Alphabet (A, D, R, C) writing on the sensor array respectively. Table 2 shows the summary of AM pressure sensor arrays reported in the literatures. Previous studies on the AM pressure sensors use the sensors such as commercial products (Pressure-sensitive rubber, PSR), [63][64][65][66][67][68][69][70] and conductive filler/polymer composites, 2 and pressure-sensitive TFT. 4,71 The key advantage of the sensor is ultra-thin (1 μm) so that can be applied to large-area electronic skin. The real-time writing on the AM sensor array can be seen on the PC screen.

Methods
Fabrication of porous CNT/SEBS film. The process flow of CNTs coating and BL exposure on the CNTs layer to make porous CNTs can be seen in Supplementary Fig. S2. A CNTs/graphene oxide (GO) layer was spray-coated on the carrier glass using a mixture of CNT/GO solution using the mixing ratio of CNT/GO = 1/8 for detaching the PI substrate from the carrier glass. The very thin CNT/GO layer was soft-baked at 130 °C for 15 min in air and then hard-baked at 290 °C for 2 h in a vacuum oven. The PI layer was then spin-coated on glass and then soft-baked at ~ 140 °C for 30 min in a hot plate and then cured at 450 ℃ for 2 h in N2 atmosphere. A buffer layer of SiNx/SiO2 was deposited on the PI substrate at the substrate temperature of 420 ℃. 72,73 The surface of the SiO2 layer was treated with UV/O3 for 300 s for uniform coating of CNTs by spray on the carrier glass and annealed at 290 °C, for 2 h in a vacuum.
The CNTs layer was exposed by a line beam of a BL (beam size: 520 µm x 20 µm, laser energy density 5.06 J cm -2 ) as shown in Supplementary Fig. S1c. The 60 mg ml -1 SEBS solution diluted in toluene was spin-coated and cured on BLA CNT at 120 °C for 10 minutes.
Fabrication of a-IGZO TFT backplane. On the buffer layer, 30 nm a-IGZO was deposited by reactive sputtering using a polycrystalline IGZO target (InO3:Ga2O3:ZnO = 1:1:1 mol %). Then, a 100 nm-thick SiO2 was deposited on the top of the a-IGZO by PECVD as a gate insulator (GI) without breaking vacuum. A 100 nm Mo was deposited by sputtering and patterned as the top gate electrode. GI layer was etched by a self-aligned process with the gate pattern. Then, a 300 nm-thick SiO2 layer was deposited as the interlayer by PECVD followed by the formation of via holes, and a 200 nm-thick Mo layer was deposited and patterned for the S/D electrodes. Then, SiNx and SiO2 double passivation layers were deposited through PECVD at 200 ℃. Finally, the devices were annealed at 300 ℃ in a vacuum for 1 h.