UV-Responsive Screen-Printed Porous ZnO Nanostructures on Ofﬁce Paper for Sustainable and Foldable Electronics

: The fabrication of low-cost, ﬂexible, and recyclable electronic devices has been the focus of many research groups, particularly for integration in wearable technology and the Internet of Things (IoT). In this work, porous zinc oxide (ZnO) nanostructures are incorporated as a UV sensing material into the composition of a sustainable water-based screen-printable ink composed of carboxymethyl cellulose (CMC). The formulated ink is used to fabricate ﬂexible and foldable UV sensors on ubiquitous ofﬁce paper. The screen-printed CMC/ZnO UV sensors operate under low voltage ( ≤ 2 V) and reveal a stable response over several on/off cycles of UV light exposure. The devices reach a response current of 1.34 ± 0.15 mA and a rise and fall time of 8.2 ± 1.0 and 22.0 ± 2.3 s, respectively. The responsivity of the sensor is 432 ± 48 mA W − 1 , which is the highest value reported in the literature for ZnO-based UV sensors on paper substrates. The UV-responsive devices display impressive mechanical endurance under folding, showing a decrease in responsivity of only 21% after being folded 1000 times. Their low-voltage operation and extreme folding stability indicate a bright future for low-cost and sustainable ﬂexible electronics, showing potential for low-power wearable applications and smart packaging.


Introduction
Paper is a renewable raw material that has been used by humankind for thousands of years to share and store information. Over the last few decades, many researchers and industries have not been looking at paper for its conventional application of transmitting information, but as a substrate for the development of low-cost, flexible, and recyclable electronic devices that will in a close future find their place in our daily life, thus leading to a revolutionary and sustainable era of electronics [1][2][3][4][5].
Cellulose, known as the most abundant and renewable biopolymer on Earth, is the building block of paper, and therefore responsible for its beneficial features, such as lightweight, low thermal expansion, flexibility/foldability, biodegradability, and recyclability [5]. Paper is also compatible with roll-to-roll (R2R) manufacturing processes, and it is available in several textures, compositions, and coatings [4,6,7].
The idea of using ubiquitous paper substrates as a low-cost, flexible, and robust support for printed electronic applications has already been demonstrated on memory devices [8], chromogenic displays [9][10][11], photovoltaics cells [12], paper-based microfluidic devices [13], and transistors and integrated circuits [14][15][16][17][18][19][20][21]. Inspired by the pioneer work introduced in our research group by Fortunato et al. [22] in 2008, Grey et al. [21] also successfully explored the intrinsic dielectric/electrolytic properties of paper, which was sequent calcination at 700 • C. The synthesized porous ZnO nanostructures were employed in the formulation of a water-based screen-printable CMC/ZnO composite ink. To the best of our knowledge, this is the first work reporting the incorporation of ZnO nanostructures with a porous morphology in a water-based screen-printable ink. The fabricated ink was integrated into UV sensors built onto office paper. The photonic devices were tested under different applied voltages and UV light intensities, and their mechanical endurance was also investigated. Their UV responsiveness far exceeds the ones reported in the literature when using ZnO nanostructures together with cellulosic materials/substrates, reaching response currents in the mA range when operating under low voltage (2 V).

Synthesis and Characterization of Porous ZnO Nanostructures
Following the procedure reported in our previous work [67], porous ZnO nanostructures were synthesized by hydrothermal method assisted by microwave irradiation. Briefly, 0.05 M of zinc nitrate hexahydrate (Zn(NO 3 ) 2 ·6H 2 O, 98% from Sigma-Aldrich, St. Louis, MO, USA) was first dissolved in deionized water, followed by the addition of urea (CH 4 N 2 O, 99.0-100.5% from Sigma-Aldrich, St. Louis, MO, USA) to the aqueous solution in a molar ratio of zinc to urea of 1:5. The synthesis was carried out in a Discovery SP microwave (CEM, Matthews, NC, USA) at 140 • C for 15 min under a power of 100 W. After the synthesis, the resulting white precipitates were washed with deionized water followed by isopropanol and centrifuged at 4000 rpm for 5 min each. This washing process was repeated three times. The powders were dried in air at room temperature and then calcinated in a muffle furnace (Nabertherm GmbH, Lilienthal, Germany) at 700 • C for 2 h in air at a heating rate of 250 The morphology of the ZnO nanostructures was evaluated via scanning electron microscopy (SEM) using a Carl Zeiss AURIGA CrossBeam FIB-SEM workstation with an Oxford X-ray energy dispersive spectrometer (Carl Zeiss Microscopy GmbH, Oberkochen, Germany). The powder was spread on aluminum stubs using a double-sided carbon tape and coated with a thin iridium layer (<20 nm) using a Q300T D Quorum sputter coater (Quorum Technologies, East Sussex, UK).

Formulation of the CMC/ZnO Composite Ink
The produced porous ZnO nanostructures were blended with a concentration of 10 wt% into a previously prepared solution of 3 wt% of CMC (Mw ≈ 250,000, Sigma-Aldrich, St. Louis, MO, USA) dissolved in deionized water. The mixture was slowly stirred at 200 rpm for 4 h to obtain a well dispersed and homogeneous white viscous solution. The ink was stored in a refrigerator at 3 • C until being used.

Fabrication and Characterization of the CMC/ZnO UV Sensors on Office Paper
As illustrated in Figure 1, carbon electrodes were screen-printed on office paper (80 g m −2 , The Navigator Company, Portugal) using a conductive carbon paste (CRSN2644 C INK, Sun Chemical, USA), and a screen mold made of polyester with the following conditions: mesh model, 120-34; mesh count, 305 mesh/inch; aperture, 45 µm; thread diameter, 34 µm; opening, 30.5%; fabric thickness, 52-57 µm. The planar electrodes were screen-printed with an interdigital architecture composed of seven pairs of interdigital fingers with a width of 200 µm and length of 7.5 mm spaced from each other with a gap of 200 µm. The carbon electrodes were dried at 100 • C for 15 min in air.

Electrical Characterization of the Screen-Printed UV Sensors on Office Paper
The sensor devices were electrically analyzed in air at room temperature using a microprobe station (Everbeing Int'l Corp., Taiwan) connected to a semiconductor parameter analyzer (Agilent 4155C, Keysight Technologies, Santa Rosa, CA, USA).
Chronoamperometry measurements were carried out with a constant applied bias voltage of 2 V. The screen-printed devices were irradiated with UV light at 365 nm using a fiber-coupled UV LED (4.1 mW, M365F1, Thorlabs Inc., Newton, NJ, USA) controlled by a DC4100 Controller from Thorlabs (Thorlabs Inc., New Jersey, USA) with a pulse train generator (Pulse Pal, Sanworks LLC, Rochester, NY, USA) programmed to turn on and off the UV LED at different intervals of time. The sensors were placed 2 cm from the light source and they were tested under a UV intensity of 8.66 mW cm −2 . The sensors were exposed to 60 s of UV light followed by 90 s in the dark during several on/off cycles. The UV sensors were also tested under different applied bias voltages (0.25, 0.5, 1, and 2 V), and different UV light intensities (0.30, 0.91, 2.02, 4.27, and 8.66 mW cm −2 ). The UV light intensity was measured by a Suss MicroTec UV-Optometer (SÜ SS MICROTEC SE, Garching, Germany).
To test the mechanical stability, a sensor was manually folded several times (100, 200, 500, and 1000 times) and flattened again before measuring its photoresponse to UV light under an applied voltage of 1 V and UV light intensity of 8.66 mW cm −2 .

Characterization of Porous ZnO Nanostructures and CMC/ZnO Screen-Printed Films
As reported in our previous work [67], porous ZnO nanostructures were synthesized through a facile and fast hydrothermal method assisted by microwave irradiation, followed by a washing step to remove remnants of reagents from the synthesis and a calcination step in a furnace to thermally convert the resulting LZHC precursor into ZnO. As demonstrated in that work [67], a calcination process at 700 °C yields porous ZnO nanostructures with high crystallinity and larger pore size when compared to lower calcination temperatures, while ensuring a complete conversion of LZHC into ZnO. Therefore, this calcination temperature was selected to prepare the ZnO nanostructures in this work. Figure 2a shows the SEM images of the produced ZnO powders for different magnifications. The porous ZnO nanostructures consist of many serrate-like porous nanoplates, which are tightly arranged into flower-like 3D hierarchical microstructures. The porous nanoplates are thin (<100 nm thickness), elongated (maximum length: <3.5 µ m), and exhibit a highly porous surface with a pore size in the range of tens of nanometers which is formed during the calcination step.

Electrical Characterization of the Screen-Printed UV Sensors on Office Paper
The sensor devices were electrically analyzed in air at room temperature using a microprobe station (Everbeing Int'l Corp., Taiwan) connected to a semiconductor parameter analyzer (Agilent 4155C, Keysight Technologies, Santa Rosa, CA, USA).
Chronoamperometry measurements were carried out with a constant applied bias voltage of 2 V. The screen-printed devices were irradiated with UV light at 365 nm using a fiber-coupled UV LED (4.1 mW, M365F1, Thorlabs Inc., Newton, NJ, USA) controlled by a DC4100 Controller from Thorlabs (Thorlabs Inc., New Jersey, USA) with a pulse train generator (Pulse Pal, Sanworks LLC, Rochester, NY, USA) programmed to turn on and off the UV LED at different intervals of time. The sensors were placed 2 cm from the light source and they were tested under a UV intensity of 8.66 mW cm −2 . The sensors were exposed to 60 s of UV light followed by 90 s in the dark during several on/off cycles.
To test the mechanical stability, a sensor was manually folded several times (100, 200, 500, and 1000 times) and flattened again before measuring its photoresponse to UV light under an applied voltage of 1 V and UV light intensity of 8.66 mW cm −2 .

Characterization of Porous ZnO Nanostructures and CMC/ZnO Screen-Printed Films
As reported in our previous work [67], porous ZnO nanostructures were synthesized through a facile and fast hydrothermal method assisted by microwave irradiation, followed by a washing step to remove remnants of reagents from the synthesis and a calcination step in a furnace to thermally convert the resulting LZHC precursor into ZnO. As demonstrated in that work [67], a calcination process at 700 • C yields porous ZnO nanostructures with high crystallinity and larger pore size when compared to lower calcination temperatures, while ensuring a complete conversion of LZHC into ZnO. Therefore, this calcination temperature was selected to prepare the ZnO nanostructures in this work. Figure 2a shows the SEM images of the produced ZnO powders for different magnifications. The porous ZnO nanostructures consist of many serrate-like porous nanoplates, which are tightly arranged into flower-like 3D hierarchical microstructures. The porous nanoplates are thin (<100 nm thickness), elongated (maximum length: <3.5 µm), and exhibit a highly porous surface with a pore size in the range of tens of nanometers which is formed during the calcination step.
network is formed with a single screen-printing step at room temperature, showing no signs of cracks nor peeling. Higher magnifications of the screen-printed films reveal a dense and compact network of stacked and randomly oriented porous ZnO nanoplates. The elongated size of the porous ZnO nanostructures promotes a percolation network at low particle loading (10 wt%), thus reducing the amount of functional material needed to fabricate the semiconducting ink. Still, some micro-sized voids and agglomerates are visible, which contribute to their rough surface.

Characterization of Porous ZnO Nanostructures as UV Sensors
As depicted in Figure 3a, the UV sensing mechanism is based on the adsorption and desorption of oxygen-rich molecules on the surface of ZnO, and has been previously explained in several works [48,54,75,76]. In the dark, oxygen-rich molecules adsorb on the porous ZnO nanostructures due to the oxygen vacancies that exist in the ZnO surface, which are known to influence ZnO electrical conductivity [77]. By adsorbing on the surface of ZnO, the oxygen-rich molecules capture an electron from the conduction band, creating a depletion region on the nanostructures' surface that results in an increase of the electrical resistance of the device. Electron-hole pairs are created when the ZnO layer is exposed to UV light with energy higher than the ZnO bandgap. The photogenerated holes migrate to the nanostructures' surface, recombining with the previously captured electrons and, therefore, releasing the adsorbed oxygen molecules from the surface. The generated electrons from the absorption of UV light are now unpaired and available for conduction, leading to a decrease of the depletion region and a subsequent increase of the photocurrent in the UV sensor. Figure 3b shows the photoelectrical response of the fabricated UV sensors over time for several UV on/off cycles under a bias voltage of 2 V. The photoresponse curve in this figure corresponds to the average response of three tested sensors. The tested sensors were exposed to UV light at 365 nm with a light intensity of 8.66 mW cm −2 for 60 s followed by 90 s in the dark. The sensors show a stable photoresponse after successive on/off UV cycles, reaching a maximum average photocurrent (Iph) of 1.45 ± 0.23 mA under UV light On the other hand, the flower-like morphology is no longer visible on the screenprinted CMC/ZnO films (Figure 2b). The mechanical stress induced during the preparation of the inks and the screen-printing process promotes the disassembling of the 2D nanostructures, thus destroying their arrangement in the form of flowers. Despite all the challenges related to office paper surface roughness and absorption, a continuous CMC/ZnO composite film with good adhesion to the hydrogen-bonded cellulose fiber network is formed with a single screen-printing step at room temperature, showing no signs of cracks nor peeling.
Higher magnifications of the screen-printed films reveal a dense and compact network of stacked and randomly oriented porous ZnO nanoplates. The elongated size of the porous ZnO nanostructures promotes a percolation network at low particle loading (10 wt%), thus reducing the amount of functional material needed to fabricate the semiconducting ink. Still, some micro-sized voids and agglomerates are visible, which contribute to their rough surface.

Characterization of Porous ZnO Nanostructures as UV Sensors
As depicted in Figure 3a, the UV sensing mechanism is based on the adsorption and desorption of oxygen-rich molecules on the surface of ZnO, and has been previously explained in several works [48,54,75,76]. In the dark, oxygen-rich molecules adsorb on the porous ZnO nanostructures due to the oxygen vacancies that exist in the ZnO surface, which are known to influence ZnO electrical conductivity [77]. By adsorbing on the surface of ZnO, the oxygen-rich molecules capture an electron from the conduction band, creating a depletion region on the nanostructures' surface that results in an increase of the electrical resistance of the device. Electron-hole pairs are created when the ZnO layer is exposed to UV light with energy higher than the ZnO bandgap. The photogenerated holes migrate to the nanostructures' surface, recombining with the previously captured electrons and, therefore, releasing the adsorbed oxygen molecules from the surface. The generated electrons from the absorption of UV light are now unpaired and available for conduction, leading to a decrease of the depletion region and a subsequent increase of the photocurrent in the UV sensor.
found to be 432 ± 48 mA W −1 . As displayed in Figure 3c, the switching response of the UV sensors was also determined to evaluate how fast they respond during (on-state) and after (off-state) exposure to UV light. Therefore, the rise time (trise) was determined by the time required to shift from 10 to 90% of the maximum photocurrent, while the fall time (tfall) corresponds to the time that it takes to decrease from 90 to 10% when the UV light is turned off [76,78,79]. These parameters were determined to be 8.2 ± 1.0 and 22.0 ± 2.3 s for trise and tfall, respectively. The promising results obtained for the produced ZnO UV sensors can be attributed to the enhanced adsorption capacity for oxygen-rich molecules on the surface of the porous ZnO nanostructures due to their large specific surface area as well as the improvement of electron transport properties in ZnO as a result of both its high crystallinity [67] and plane contacts between the porous nanoplates [63]. Overall, the resistance of the sensing layer is strongly affected by the inter-nanostructure barrier at the contacts, meaning that the photoresponse of the sensor is mainly determined by the contacts between the ZnO nanostructures. In this case, the elongated nanoplate-shape of the porous ZnO nanostructures results not only in a more direct percolation path for electron transport due to the reduced nanoplates' thickness (<100 nm) but also in plane-to-plane contacts between the particles. This means that the area of the contact plane is much larger when  The responsivity (R) of the sensors can be determined by the following expression: where P UV is the output power of the UV LED (3.1 mW for a UV light intensity of 8.66 mW cm −2 and an irradiated area of 36 mm 2 ). The average responsivity of the tested sensors was found to be 432 ± 48 mA W −1 . As displayed in Figure 3c, the switching response of the UV sensors was also determined to evaluate how fast they respond during (on-state) and after (off-state) exposure to UV light. Therefore, the rise time (t rise ) was determined by the time required to shift from 10 to 90% of the maximum photocurrent, while the fall time (t fall ) corresponds to the time that it takes to decrease from 90 to 10% when the UV light is turned off [76,78,79]. These parameters were determined to be 8.2 ± 1.0 and 22.0 ± 2.3 s for t rise and t fall , respectively. The promising results obtained for the produced ZnO UV sensors can be attributed to the enhanced adsorption capacity for oxygen-rich molecules on the surface of the porous ZnO nanostructures due to their large specific surface area as well as the improvement of electron transport properties in ZnO as a result of both its high crystallinity [67] and plane contacts between the porous nanoplates [63]. Overall, the resistance of the sensing Chemosensors 2021, 9,192 7 of 14 layer is strongly affected by the inter-nanostructure barrier at the contacts, meaning that the photoresponse of the sensor is mainly determined by the contacts between the ZnO nanostructures. In this case, the elongated nanoplate-shape of the porous ZnO nanostructures results not only in a more direct percolation path for electron transport due to the reduced nanoplates' thickness (<100 nm) but also in plane-to-plane contacts between the particles. This means that the area of the contact plane is much larger when compared to other nanostructures' morphologies (such as nanowires [80] for example) thus leading to enhanced sensing properties. Table 1 shows a comparison between our results and the current state-of-the-art values in the field of UV sensors based either on ZnO nanostructures grown/deposited on paper substrates or ZnO/cellulose composites. Although these types of ZnO UV sensors have only been reported since the beginning of the last decade, most of the first published works do not provide a detailed characterization of the sensors. From Table 1, it is clear that the CMC/ZnO UV sensors fabricated in this work show the highest response current and responsivity. For instance, Figueira et al. [54] fabricated screen-printed ZnO UV sensors on cork substrates using an ethyl cellulose/ZnO commercial nanoparticles composite ink. Despite the similar methods involved in the fabrication process, these devices could only reach 20 µA under UV light, which is 100 times smaller than the response current obtained in this work. Besides, contrary to our work, a high loading content of small-sized (<100 nm) commercial ZnO nanoparticles (40 wt%) were needed to form a continuous and functional composite matrix on the irregular and punctually defective cork surface. Beyond the listed works displayed in Table 1, the screen-printed CMC/ZnO sensors on office paper also exhibit a comparable or even superior performance than some of recently published ZnO-based UV sensors comprising expensive materials and/or complex, time-consuming and energy-intensive processing methods [75,76,[81][82][83][84][85]. Table 1. Comparison of typical measurement parameters (V bias and UV light intensity) and respective UV sensor characteristics (t rise , t fall , ∆I, and R) obtained for devices based either on ZnO nanostructures grown/deposited on paper substrates or ZnO/cellulose composites.  Figure 4a,b depict the average photocurrent of the UV sensors under different applied voltages (0.25, 0.5, 1, and 2 V) and the corresponding response current (∆I) and responsivity. The bias voltage plays an important role in the electron-hole pair separation, recombination, and transportation processes and, therefore, it determines the overall photoresponse of the UV sensor [78]. The response current of the sensor increased almost linearly from 0.14 ± 0.02 to 1.34 ± 0.15 mA when the applied voltage is varied from 0.25 to 2 V, respectively. Typically, ZnO nanostructures present high recombination rates, low carrier mobilities, and high resistivities [77,78], which means that applied voltages above 1 V are usually needed to obtain a good response to UV light [78]. In this work, the sensors presented a very good responsivity even for low applied voltages (<2 V), reaching a value of 43.9 ± 6.8 mA W −1 for 0.25 V. The low-voltage operation of the fabricated screen-printed sensors is compatible with the emerging concept of IoT and demonstrates that these devices can be powered by portable power supplies, such as thin-film batteries or solar cells [5,12,18,95]. To the best of our knowledge, this is the first report showing low-powered and low-cost ZnO-based UV sensors on paper substrates with a response current in the mA range, as previously evidenced by Table 1.  Figure 4a,b depict the average photocurrent of the UV sensors under different applied voltages (0.25, 0.5, 1, and 2 V) and the corresponding response current (ΔI) and responsivity. The bias voltage plays an important role in the electron-hole pair separation, recombination, and transportation processes and, therefore, it determines the overall photoresponse of the UV sensor [78]. The response current of the sensor increased almost linearly from 0.14 ± 0.02 to 1.34 ± 0.15 mA when the applied voltage is varied from 0.25 to 2 V, respectively. Typically, ZnO nanostructures present high recombination rates, low carrier mobilities, and high resistivities [77,78], which means that applied voltages above 1 V are usually needed to obtain a good response to UV light [78]. In this work, the sensors presented a very good responsivity even for low applied voltages (<2 V), reaching a value of 43.9 ± 6.8 mA W −1 for 0.25 V. The low-voltage operation of the fabricated screen-printed sensors is compatible with the emerging concept of IoT and demonstrates that these devices can be powered by portable power supplies, such as thin-film batteries or solar cells [5,12,18,95]. To the best of our knowledge, this is the first report showing low-powered and low-cost ZnO-based UV sensors on paper substrates with a response current in the mA range, as previously evidenced by Table 1.  Figure 5a,b show the average photocurrent for different UV light intensities and the calculated response current and responsivity for an applied voltage of 1 V. In Figure 5b, the sensors' response current increases with UV light intensity, shifting from 0.17 ± 0.02 mA for an intensity of 0.30 mW cm −2 to 0.52 ± 0.08 mA for an intensity of 8.66 mW cm −2 . The UV light intensity is proportional to the number of photons irradiating the porous ZnO nanostructures for a certain UV wavelength. As such, when the UV intensity increases, the number of incident photons is higher and, consequently, more carriers are generated, and the response current is increased. On the other hand, the sensor's responsivity decreases with increasing UV irradiance. This is expected since the response current remains in the same order of magnitude (10 −1 mA) while the incident UV light power increases from 0.30 to 8.66 mW cm −2 . As a result, the calculated responsivity (mA W −1 ) decreases with increasing UV light intensity. More importantly, the obtained results indicate that the produced UV sensors can respond to much weaker UV light, making these sensors suitable for the detection of low levels of UV light.   Figure 5b, the sensors' response current increases with UV light intensity, shifting from 0.17 ± 0.02 mA for an intensity of 0.30 mW cm −2 to 0.52 ± 0.08 mA for an intensity of 8.66 mW cm −2 . The UV light intensity is proportional to the number of photons irradiating the porous ZnO nanostructures for a certain UV wavelength. As such, when the UV intensity increases, the number of incident photons is higher and, consequently, more carriers are generated, and the response current is increased. On the other hand, the sensor's responsivity decreases with increasing UV irradiance. This is expected since the response current remains in the same order of magnitude (10 −1 mA) while the incident UV light power increases from 0.30 to 8.66 mW cm −2 . As a result, the calculated responsivity (mA W −1 ) decreases with increasing UV light intensity. More importantly, the obtained results indicate that the produced UV sensors can respond to much weaker UV light, making these sensors suitable for the detection of low levels of UV light.

Materials
To evaluate the mechanical stability, a single CMC/ZnO UV sensor was manually folded successive times and then flattened again prior to being subjected to a UV light intensity of 8.66 mW cm −2 under an applied voltage of 1 V. The photoresponse was recorded after folding the sensor 100, 200, 500, and 1000 consecutive times. Figure 6a shows the normalized response current curves for the initial flat position and after being folded Chemosensors 2021, 9,192 9 of 14 a different number of times. The response current reached by the sensor after 60 s of UV exposure decreased with the number of folds and its responsivity decreased by 1.5, 9, 20, and 21% after the sensor has been folded 100, 200, 500, and 1000 times, respectively, as depicted in Figure 6b. As illustrated by the inset in Figure 6b, when the ZnO sensor was folded in half, a folding crease appeared across the sensor's layers and paper substrate, which was intensified as the number of folds increased. An SEM image of the crease folding area after the sensor had been folded 1000 times is shown in the left inset of Figure 6b. Here, some cracks are visible in the CMC/ZnO layer deposited on paper. These cracks and their length increased in both the electrodes and ZnO photoactive layer with the number of folds, consequently leading to a decrease in the UV sensor's responsivity, since the path for the electrical carriers was broken in several places. Nonetheless, the produced ZnO sensors are capable of detecting UV light with very good responsivity even after being folded in half 1000 times, thus demonstrating their extremely high mechanical resistance. These results are of great significance since they validate the robustness of the fabricated ZnO UV sensors, therefore showing their potential to be used in wearable devices and smart packaging. To evaluate the mechanical stability, a single CMC/ZnO UV sensor was manually folded successive times and then flattened again prior to being subjected to a UV light intensity of 8.66 mW cm −2 under an applied voltage of 1 V. The photoresponse was recorded after folding the sensor 100, 200, 500, and 1000 consecutive times. Figure 6a shows the normalized response current curves for the initial flat position and after being folded a different number of times. The response current reached by the sensor after 60 s of UV exposure decreased with the number of folds and its responsivity decreased by 1.5, 9, 20, and 21% after the sensor has been folded 100, 200, 500, and 1000 times, respectively, as depicted in Figure 6b. As illustrated by the inset in Figure 6b, when the ZnO sensor was folded in half, a folding crease appeared across the sensor's layers and paper substrate, which was intensified as the number of folds increased. An SEM image of the crease folding area after the sensor had been folded 1000 times is shown in the left inset of Figure 6b. Here, some cracks are visible in the CMC/ZnO layer deposited on paper. These cracks and their length increased in both the electrodes and ZnO photoactive layer with the number of folds, consequently leading to a decrease in the UV sensor's responsivity, since the path for the electrical carriers was broken in several places. Nonetheless, the produced ZnO sensors are capable of detecting UV light with very good responsivity even after being folded in half 1000 times, thus demonstrating their extremely high mechanical resistance. These results are of great significance since they validate the robustness of the fabricated ZnO UV sensors, therefore showing their potential to be used in wearable devices and smart packaging. To evaluate the mechanical stability, a single CMC/ZnO UV sensor was manually folded successive times and then flattened again prior to being subjected to a UV light intensity of 8.66 mW cm −2 under an applied voltage of 1 V. The photoresponse was recorded after folding the sensor 100, 200, 500, and 1000 consecutive times. Figure 6a shows the normalized response current curves for the initial flat position and after being folded a different number of times. The response current reached by the sensor after 60 s of UV exposure decreased with the number of folds and its responsivity decreased by 1.5, 9, 20, and 21% after the sensor has been folded 100, 200, 500, and 1000 times, respectively, as depicted in Figure 6b. As illustrated by the inset in Figure 6b, when the ZnO sensor was folded in half, a folding crease appeared across the sensor's layers and paper substrate, which was intensified as the number of folds increased. An SEM image of the crease folding area after the sensor had been folded 1000 times is shown in the left inset of Figure 6b. Here, some cracks are visible in the CMC/ZnO layer deposited on paper. These cracks and their length increased in both the electrodes and ZnO photoactive layer with the number of folds, consequently leading to a decrease in the UV sensor's responsivity, since the path for the electrical carriers was broken in several places. Nonetheless, the produced ZnO sensors are capable of detecting UV light with very good responsivity even after being folded in half 1000 times, thus demonstrating their extremely high mechanical resistance. These results are of great significance since they validate the robustness of the fabricated ZnO UV sensors, therefore showing their potential to be used in wearable devices and smart packaging.

Conclusions
In summary, we demonstrated for the first time the use of porous ZnO nanostructures synthesized by microwave hydrothermal method to fabricate a water-based screenprintable ink where CMC was used as a binder. The CMC/ZnO composite ink was successfully screen-printed on office paper to be used as the photoactive layer in UV sensors.
The fabricated ZnO UV sensors showed promising results for application in lowpower and flexible paper-based electronics. The tested sensors revealed a stable and good responsivity of 432 ± 48 mA W −1 with a response current of 1.34 ± 0.15 mA for an applied voltage of 2 V, which are the highest values reported in the literature for UV sensors based either on ZnO nanostructures grown/deposited on paper or ZnO/cellulose composites. Furthermore, the ZnO sensors showed good responsivities even under low UV light intensity (0.30 mW cm −2 ) and ultra-low applied voltage (0.25 V). The produced UV sensors were also tested after being folded up to 1000 consecutive times with a decrease of only 21% in the final responsivity value, hence proving their potential for wearable technologies and smart packaging.