Spray-lithography of hybrid graphene-perovskite paper-based photodetectors for sustainable electronics

Paper is an ideal substrate for the development of flexible and environmentally sustainable ubiquitous electronic systems. When combined with nanomaterial-based devices, it can be harnessed for various Internet-of-Things applications, ranging from wearable electronics to smart packaging. However, paper remains a challenging substrate for electronics due to its rough and porous nature. In addition, the absence of established fabrication methods is impeding its utilization in wearable applications. Unlike other paper-based electronics with added layers, in this study, we present a scalable spray-lithography on a commercial paper substrate. We present a non-vacuum spray-lithography of chemical vapor deposition (CVD) single-layer graphene (SLG), carbon nanotubes (CNTs) and perovskite quantum dots (QDs) on a paper substrate. This approach combines the advantages of two large-area techniques: CVD and spray-coating. The first technique allows for the growth of SLG, while the second enables the spray coating of a mask to pattern CVD SLG, electrodes (CNTs), and photoactive (QDs) layers. We harness the advantages of perovskite QDs in photodetection, leveraging their strong absorption coefficients. Integrating them with the graphene enhances the photoconductive gain mechanism, leading to high external responsivity. The presented device shows high external responsivity of ∼520 A W−1 at 405 nm at <1 V bias due to the photoconductive gain mechanism. The prepared paper-based photodetectors (PDs) achieve an external responsivity of 520 A W−1 under 405 nm illumination at <1 V operating voltage. To the best of our knowledge, our devices have the highest external responsivity among paper-based PDs. By fabricating arrays of PDs on a paper substrate in the air, this work highlights the potential of this scalable approach for enabling ubiquitous electronics on paper.


Introduction
The utilization of paper as a substrate presents a range of essential characteristics, encompassing sustainability, environmental friendliness, ubiquity, low cost, flexibility, biodegradability, recyclability, and deformability [1].Bendable substrates are of great interest due to their integration with non-flat surfaces that can allow for new device architectures.Original content from this work may be used under the terms of the Creative Commons Attribution 4.0 licence.Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
Traditional photodetector (PD) substrates such as silicon wafers, typically demand complex fabrication techniques and associated high energy costs.Thus, the use of paper is attractive, offering affordability for mass production alongside a scalable approach for commercialisation [2].This is further reflected through the ease of fabrication, where unlike traditional electronics that require complex equipment and elevated temperatures, paper offers compatibility among a variety of printing techniques and inks for simple architectures [3,4].Furthermore, the paper-based substrate exhibits exceptional thermal stability in comparison to certain flexible plastic foils.The paper provides an excellent platform for the integration of two-dimensional (2D) and layered materials [5].Paper is utilized as a versatile medium in various Internet-of-Things applications [6], including wearable electronics, smart packaging, and sensor-based systems such as humidity sensors and pressure sensors [2,6,7].However, paper is still a challenging substrate for electronics and has been rarely used without the inclusion of additional coating or laminating layers.The rough and porous nature, low stability, and endurance (primarily due to insufficient resistance to heat and humidity), in addition to the absence of established fabrication methods, are impeding its utilization in wearable applications [8,9].
The printing of patterns on the paper substrate can be accomplished through various techniques, including screen printing, gravure printing, flexography, and inkjet printing [10].Among these methods, inkjet printing stands out as one of the most promising due to its advantageous attributes, such as direct patterning without the need for masks and highresolution capabilities [11].Nevertheless, the development of printed electronics, has primarily been constrained by the performance characteristics of the materials utilized as the constituents of these devices.The inherent nature of hopping transport, within the printed 2D layers significantly limits their electrical performance.Adhesion of the 2D inks to flexible substrates is a major challenge in device fabrication.Moreover, the use of polymeric binders is restricted since binders limit 2D material ink functionalities [10,12].
Conti et al demonstrated the development of a field-effect transistor on a paper substrate through the integration of inkjet printing and chemical vapor deposition (CVD) [5].This innovative approach employed CVD deposited layered materials to address the limitations associated with solutionprocessed materials, specifically those derived from liquidphase exfoliation techniques of 2D materials.The fabrication process involved an initial step of optical lithography for patterning the CVD layer, followed by wet transferring and inkjet printing [5].However, the reliance on complex and expensive fabrication tools, such as optical lithography, underscores the need for more accessible and cost-effective techniques to promote the commercialization and sustainability of paper-based electronics.In a recent study, Akhavan et al [13] demonstrated inkjet-lithography for the fabrication of PDs utilizing CVD single-layer graphene (SLG) and black phosphorus ink on a Si/SiO 2 substrate.This approach resulted in the creation of PDs exhibiting responsivity of 337 A W −1 at 488 nm.Notably, all fabrication processes were exclusively conducted via inkjet printing technology [13].However, it is essential to acknowledge that the limitations of this technique include the relatively slower inkjet print speed and the associated higher operational costs, particularly concerning issues such as nozzle clogging that may occur with certain ink formulations [12].
Spray coating is a mature and cost-effective technology that offers numerous advantages over different coating techniques.It provides uniform coverage, fast application, and versatility for various substrates, including complex shapes and irregular surfaces, reducing material waste, and enabling precise control of layer thickness [14].Overall, spray coating stands out for its efficiency, adaptability, and consistent and controlled coating [14].Unlike spin-coating or doctor-blading methods, spray coating is not limited by the shape of the substrate.Additionally, the precursor solution is atomized into microsized droplets during spraying, so the deposition of the subsequent layer does not dissolve the previous ones.This allows for fabrication of heterostructures made of different layers, and uniform film thicknesses of hundreds of micrometers or even millimeters to be achieved [15].
Paper-based PDs are lightweight, flexible, and costeffective in nature, making them accessible for diverse applications.Paper-based PDs exhibit limitations in terms of responsivities, response time, and operating voltage due to several factors.The inherent properties of paper, such as its rough surface and light-scattering nature, can affect the efficiency of light absorption and conversion, resulting in lower responsivities compared to traditional PDs [8].Deng et al [16] present an alternative all sprayed-processable Perovskite/ MXene PD with responsivity of 0.049 A W −1 under a 10 V bias at 450 nm [16].The high operating voltage of paperbased PDs requires higher electrical fields for optimal device performance.Thus, benefits of a lower power consumption can be observed in situations where continuous monitoring is required such as environmental sensors or long-term medical assessments.Li et al [17], demonstrate the highest responsivity achieved by a paper-based PD, using MAPbBr 3 and achieving a maximum responsivity of 1.3 A W −1 under a 1 V bias at 365 nm [17].However, much higher responsivity (10 9 A W −1 ) at 0.5 V under 598 nm illumination on other substrates such as Si/SiO 2 are reported [18], necessitating the overcoming of the limitations of paper with a new device design structure.
Motivated by the benefits of solution-processed quantum dots (QDs) in photodetection [19][20][21] and their integration with graphene to enhance the photoconductive gain mechanism [22], this study introduces a novel class of paper-based PDs employing CVD SLG and QDs.Unlike other paperbased electronics that use additional, coating/laminating layers, we have developed a technology based on scalable spray coating on a commercial paper substrate.
Our results demonstrate high external responsivity (R ext ) of 520 A W −1 at 405 nm in the visible range, and low operating voltage (<1 V), the highest reported R ext to date for paper-based PDs, to the best of our knowledge.These findings highlight the potential of CVD layered-based materials for achieving high-performance photodetection in paperbased systems.

Dispersion of MWCNTs
20 mg of multi-walled carbon nanotubes (MWCNTs) and 10 mg of surfactant (Triton X-100) were added to 10 ml of a dimethylformamide (DMF) and then sonicated for 6 h at 50 °C.The process was followed by centrifugation of the suspension via Thermo Scientific Heraeus Multifuge X1 (2272 ×g) at 4000 rpm for 30 min to remove nondispersible materials.

Synthesis of CsPbBr 3 perovskite QDs
The synthesis procedure for CsPbBr 3 perovskite QDs followed a straightforward and rapid one-step injection method.It commences by dissolving Cs 2 CO 3 in propionic acid (PrAc) to form a Cs + propionate complex.This complex is then diluted in a solvent mixture containing IPA, hexane (HEX), and butylamine (BuAm) all at room temperature.Notably, this dissolution reaction is exothermic, eliminating the need for external heating.Additionally, this method doesn't require degassing of the precursors.Simultaneously, another solution is prepared at ambient conditions, involving the dissolution of PbBr 2 in a similar chemical mixture (also at room temperature).This solution is subsequently injected into the first one.Within just 10 s after injection, the QDs initiate nucleation and rapidly reach their maximum size.Following this stage, the QDs are isolated through centrifugation and redispersed in toluene, making them ready for direct use in device fabrication.

Characterisations
The Lambda 750 S UV-vis spectrometers (Perkin Elmer) were used to conduct measurements of optical absorption spectra at room temperature.Fourier-transform infrared spectroscopy (FTIR) analysis to assess the functional groups was done using Perkin Elmer Spectrum Two FT-IR Spectrometer.Oxford Instruments X-Max 80 (SDD) EDS system was utilised to check SEM images.X-ray diffraction was conducted using the Aeris Benchtop XRD System from PANalytical.Steady-state and time-resolved photoluminescence (TRPL) spectra were acquired on a Time-correlated single photon counting (TCSPC) system (LifeSpec-ps) by Edinburgh Instruments, employing an excitation wavelength of 405 nm at −15 °C.Sample preparations for measurements were done through spin-coating on various substrates using the Ossila spin coater.SLG was etched using plasma etching via Diener electronic GmbH & Co. Raman spectra were gathered using the inVia Raman microscope (Renishaw) at room temperature, utilising an excitation wavelength of 514.5 nm and a 100x objective lens.Performance tests for the PD were carried out using the B1500A semiconductor device analyzer (Keysight).The sheet resistance of different layers of CVD SLG was assessed using the four-point probe (Ossila).Contact angle and surface tension measurements were conducted using Theta Optical Tensiometers (Biolin Scientific) in Sessile drop mode.The thickness of spray-coated QDs was measured using a DektakXT ® stylus profilometer.

Results and discussion
The fabrication process flow for QDs/SLG PD on paper substrate is outlined in figure 1. Canon A5 paper was chosen as the substrate to ensure the flexibility of the PDs.We carefully cut the entire sheet of paper into approximately 2 cm × 2 cm rectangular pieces.Afterwards, we proceeded to clean these cut paper pieces by dripping Milli-Q water, followed by acetone and isopropanol, and then drying them with nitrogen gas.A wet-transfer method was used to transfer the SLG from Cu foil to paper substrate.A poly(methyl methacrylate) (PMMA)/anisole (10 wt%) solution was spin coated at 3000 rpm, 20 s on top of the copper SLG, as a supporting layer.To etch the backside SLG on Cu foil, plasma etching was employed for 20 s at 30 W. After etching the backside graphene, the copper sheets were left to float on the APS solution in water with PMMA on the top.The resulting SLG/PMMA membrane were placed in water to clean the APS residuals and then transferred onto the paper substate, dried overnight and washed with acetone and IPA to remove PMMA.
To create the required pattern for spraying (Clarke, air brush kit), we designed a mesh screen with a side length of 0.81 mm as the frame of the mask, figure 1(a).When spraying CNTs, the procedure must be conducted on a hot plate.The solvent used in the CNT solution is DMF, with a boiling point of 151 °C.To ensure rapid solvent evaporation after spraying, allowing efficient adhesion of CNTs to the SLG while avoiding excessively high temperatures that could impact sample performance, we set the substrate temperature to 100 °C.The spraying distance was maintained at 20 cm, and the spraying duration for 10 s to create an optimal electrode.Subsequently, the sample was dried at 100 °C for 20 min.Then, 5 wt% polystyrene (PS) solution was dissolved in toluene solvent at 40 °C and spray-coated through the designed mask, figure 1(b).Afterwards, the SLG was etched using plasma etching (40 s at 30 W) to pattern the SLG while the SLG-based channel was protected by PS, figure 1(c).The sample was then rinsed with water, followed by acetone and IPA, to eliminate the protective PS layer on top of the channel.Subsequently, QDs were applied via spray coating (figure 1(d)) with a target thickness of approximately 500 nm (Supplementary Information 1).The substrate temperature was set to 100 °C, and a short-burst spraying technique was employed to control the thickness.The process involved spraying 30 times over short (2-3 s) durations to promote the formation of a consistent film.The sample dried rapidly after each spray coating.The schematic and image of completed QDs/SLG PDs in array format on paper substrate are shown in figure 1(e).
The presence and quality of SLG were further studied by Raman spectroscopy.Raman spectra were acquired at 514.5 nm using a Renishaw InVia with a 100X objective <0.1 mW. Figure 2 plots the Raman spectrum of the film as grown on Cu after the Cu PL removal [23].The 2D peak is a single Lorentzian with FWHM(2D)∼40 cm −1 , signature of SLG [23].
The position of the G peak, Pos(G), is ∼1591 cm −1 , with FWHM(G)∼17 cm −1 .The 2D peak position, Pos(2D), is ∼2688 cm −1 , with FWHM(2D)∼40 cm −1 , while the 2D to G peak intensity and area ratios, I(2D)/I(G) and A(2D)/A(G), are ∼3.4 and∼7.9.A Small D peak is observed at Pos(D)∼1351 cm −1 , with I(D)/I(G) ∼0.1 indicating presence of small defects [23] of grown SLG on Cu foil.In consideration of the significance of PS wettability on transferred SLG for patterning, the adhesion of PS was also tested.The PS ink was prepared by dissolving PS in toluene at 40 °C to obtain a 5 wt% solution.The contact angle measurement revealed a value of 25.49°and a surface tension of 235 mN m −1 for a drop of PS on SLG transferred onto the paper substrate (Supplementary Information 2).The wettability of our PS solution indicated its role as a protective layer that forms and good adherence to the SLG, safeguarding the underlying CVD SLG during plasma etching.The quality of SLG after spray-coated PS and etched CVD SLG was further investigated by Raman spectroscopy, The average sheet resistance of SLG on paper substrate, measured using a four-point probe method, was 9.36 kΩ sq −1 indicating a conductivity of 173.19 kS sq −1 with our paperbased substrate coated with SLG layer.Lee et al [26] reported sheet resistance of transferred undoped SLG on a flexible PET substrate for SLG is 2.1 kΩ sq −1 [26].The difference between our measured sheet resistance and the reported value could be attributed to the more roughness of our paper substrate than the PET counterpart.
CNT was dispersed in DMF (explained in detail in the experimental section), and spray-coated as electrodes due to high conductivity and solution processable properties.In figure 3, the graph depicts the optical absorbance (Abs) of dispersed CNTs in DMF.The concentration of dispersed CNT was determined through the Beer-Lambert Law, which relates Abs to concentration (c (g/L)), the extinction coefficient (ò ext (L/g.m)), and the cuvette length (L (m)).The solution was first diluted 150 times beforehand to prevent the saturation of the spectrometer's detector.In a previous study [27], the CNT ò ext at 300 nm was experimentally derived by measuring the slope of Abs per length versus CNT concentration, resulting in an ò ext value of approximately 5.625 × 10 -3 L/g.m.This estimation yielded a concentration of approximately c ∼ 0.63 g l −1 for our CNT ink, consistent with findings in prior research [27].The presence and quality of CNT was further studied by Raman spectroscopy, figure 3(b).Raman spectra were acquired at 514.5 nm using a Renishaw InVia with a 100 × objective < 0.5 mW.The three active modes for CNT are D, G, and 2D modes [28].The G band is the in-plane bond stretching mode of the C-C bonds in the hexagonal lattice.The 2D peak is related to the Raman scattering due to a vibrational mode characterized by the breathing of six carbons pertaining to a hexagon in the hexagonal lattice.In the presence of a defect, the first order component of the hexagon-breathing mode is activated combined with an elastic scattering of a photo-excited electron by the defect, as a double-resonance Raman peak and it is called the D band [28].The position of the G peak, Pos(G), is ∼1593 cm −1 , with FWHM(G) ∼76 cm −1 .The 2D peak position, Pos(2D), is ∼2701 cm −1 , with FWHM(2D) ∼106 cm −1 , while the 2D to G peak intensity and area ratios, I(2D)/I(G) and A(2D)/A(G), are ∼0.32 and ∼0.46, respectively.D peak was observed at 1353 cm −1 with I(D)/I(G) ∼1.02.FTIR was used to assess the functional groups in the CNTs.As shown in figure 3(c), (O)-H bonds were observed near 3542 cm −1 , C-H bonds near 2933 cm −1 , C=O bonds near 1659 cm −1 , O-C=O bonds near 1386 cm −1 , and C-O bonds near 1092 cm −1 .In particular, the characteristic peak of Triton X-100 was observed near 651 cm −1 (C=C) [29].When dissolving CNTs with Triton X-100, there were some extra peaks or spectral changes that can occur since Triton X-100 has specific functional groups, such as alkenyl groups (C=C), which can produce additional characteristic peaks in the FTIR spectrum.At the same time, additional peaks of its same characteristic groups as CNT can superimpose with the spectrum of CNT, resulting in different characteristic peaks observed in the FTIR spectrum.We measured the sheet resistance of the CNT ink coated on the paper substrate using a four-point probe setup, and the resultant sheet resistance was found to be 20.04Ohm sq −1 .Scanning electron microscopy (SEM) images were acquired to check the topography of the spray-coated CNT on paper substrate (Supplementary Information 3).The SEM image revealed the degree of dispersion and aggregation of coated CNTs on the paper substrate.To assess electrode feasibility, the wettability of CNT ink on transferred SLG over a paper substrate, was investigated.Optical contact angle and surface tension measurements (Attension) were employed for characterisation.The contact angle measurement was found to be 29.51°andsurface tension to be 224.3mN m −1 for a drop of CNT ink on SLG transferred onto the paper substrate, allowing for swift drying with minimal spread outside the restricted area (Supplementary Information 2).
Stable CsPbBr 3 perovskite QDs were synthesized using a modified recipe from [30], which is explained in detail in the experimental section.The synthesis of CsPbBr 3 perovskite QDs can be observed by a rapid and notable colour change reaction, which signifies its swift formation.Typically, within the first ten seconds of the reactants' contact, the reaction initiates, resulting in a gradual transition to a fluorescent yellow solution.The reaction reaches its completion in just two minutes, with no further product formation observed.The crystallinity and crystal structure of the CsPbBr 3 perovskite QDs domain can be confirmed by the characteristic peaks of the x-ray diffraction (Supplementary Information 4).
The photoluminescence (PL) and optical absorption spectra were collected from purified CsPbBr 3 perovskite QDs.These measurements were performed on samples dispersed in toluene and on solid films deposited via spin-coating on glass, as depicted in figure 4(a).The QDs in the solution exhibited a PL peak centred at approximately 516 nm, with FWHM ∼21 nm like that of cubic 8.5 nm nanocrystals [31,32].In contrast, the thin film displayed a redshift (∼ 4 nm) in the optical absorption edge, transitioning from 516 to 520 nm, as determined from the band edge.This shift may be attributed to several factors, such as surface defects, strain effects, and variations in size and shape [33,34].Notably, the PL observed in the film consistently aligned with the absorption edge, maintaining a consistent Stokes shift of approximately 6 nm and a FWHM like that observed in the solution.The Tauc plot (figure S5a) method was employed on the UV-vis spectra to assess the bandgap of the QDs, resulting in values of 2.18 eV and 2.22 eV for QDs in solution and on glass, respectively.
The PL spectra of QDs and QDs/SLG are shown in figure S5b, both show a PL peak ∼520 nm arising from the QDs band gap.The PL intensity (integrated area under PL curve) of QDs/SLG on glass was quenched ∼33% compared to QDs on glass, figure S5b.This can be assigned to charge carrier transfer between QDs and SLG [35]. Figure 4(b) shows the TRPL spectra of CsPbBr 3 QDs and CsPbBr 3 QDs/SLG both being on glass substrates to investigate the excited-state dynamics of the QDs/SLG.The CsPbBr 3 QDs on a glass substrate exhibited a biexponential decay pattern, consistent with prior literature [36], featuring an average fluorescence decay time of 3.4 ns.In contrast, the QDs/SLG structure displayed a notably shorter average fluorescence decay time of 1.6 ns, as depicted in figure 4(b).Our findings indicate that SLG, in this context, exhibits quenching effects like those reported with perovskite on SLG [37,38].We believe that the PL quenching and lifetime shortening in TRPL for CsPbBr 3 QDs/SLG, point toward a predominant hole transfer mechanism from CsPbBr 3 QDs to SLG.The deposition of QDs on SLG appears to enhance charge transfer by facilitating π-π electron interactions between the QDs and the sp 2 -hybridized SLG layer.In essence, the observed PL quenching suggests the occurrence of rapid charge transfer within the QDs/SLG superstructure [39].Furthermore, a contact angle of 10.88°and a surface tension of 329.3 mN m −1 was measured for a drop of QDs on SLG (Supplementary Information 2).This small contact angle suggests a strong affinity of the SLG surface towards QDs.The wettability characteristics of QDs provide valuable insights into the formation of a compact film through their strong interaction with SLG.
The photoelectrical performance of the paper-based PDs was evaluated using Thorlabs laser diode, which emitted light at a wavelength of 405 nm.The laser beam, with a beam spot size of 5 mm 2 in diameter, was employed as the excitation light and covered the entire SLG channel.The PD responsivity is a key parameter of PDs and can be defined either as external [40]:

/
where I light and I dark are the currents of the PD under illumination and in dark conditions, respectively.A PD refers to the PD area, while A opt represents the laser spot size.The scaling factor A PD /A opt considers that only a fraction of the optical power reaches the photoactive area of PD.P opt denotes the incident optical power, and P abs refers to the absorbed optical power.In our measurements, the light spot had a diameter of 5 mm, corresponding to an area of 19.64 mm 2 .The channel area was fabricated using a pattern size of 0.81 mm, yielding an area of 0.66 mm 2 .It is worth highlighting that not all incoming photons are entirely absorbed by the PD, which causes the optical power output (P opt ) to surpass the absorbed power (P abs ).Consequently, the internal responsivity (R int ) is higher than the R ext .Nonetheless, R ext offers a more comprehensive overview of PD performance as it accounts for various factors, including light reflection, reabsorption, material quality, etc [40].
Figure 5(a) depicts the current in the channel (I) as a function of voltage in dark and under illumination at a wavelength of 405 nm with 0.51 mW optical power.We observed a linear relationship between the current through the device and the applied voltages, indicating a strong connection between the channel and electrodes.This linearity can be attributed to the Ohmic nature of the QDs/SLG/CNT junction.The Ohmic behaviour of the PD is advantageous as it facilitates the generation of high photocurrent.This behaviour allows charges to flow easily from the conduction band of SLG to the metal contacts, enhancing the device's performance.Upon illumination, we noticed an increase in the current, indicating hole (h) dominated carrier transport.This suggests the transfer of holes from QDs to SLG.The mechanism was elucidated based on band diagrams (inset image of figure 5(a)), which confirmed the p-type nature of SLG through Raman shift analysis, as showed in figure 2.
The movement of carriers described above, resulted in the generation of a local electric field at the SLG and QD interface, with the field direction pointing towards SLG.This electric field induced an upward band bending of QDs towards SLG, further influencing the behaviour of the system.Under light illumination, the top layer (QDs) absorbs the photons from incoming light, leading to the generation of electron-hole (e-h) pairs.Due to the presence of the local electric field, photogenerated electrons or holes transfer to SLG and subsequently move toward the electrodes.
This phenomenon contributes to either an increase or decrease in the current, depending on the direction of charge flow.During illumination of our paper-based PD, light was absorbed by QDs, and photogenerated holes (h) were transferred from the valence band of the QDs to lower energy states in SLG.This transfer resulted in the accumulation of photogenerated electrons (e) as uncompensated charges in the system.The photogenerated electrons, being trapped in QDs, function as an additional negative gate when applied to the SLG channel.This alters the electric field at the junction between QDs and SLG. Figure 5(b) plots the photocurrent as a function of V ds .This is defined as: where I light is the current under illumination and I dark is current in dark conditions.When V ds exceeds 1 V, the drift velocity (ν d ) of free carriers [40]: where ν sat represents the saturation velocity of carriers in the channel, μ is the mobility of SLG, and E is the applied electric field to SLG, increases linearly until it reaches saturation, primarily due to carrier scattering with optical phonons.Consequently, all measurements were conducted with V ds 1 V to maintain the device's operation within the linear (Ohmic) regime, thus eliminating the nonlinear dependence of ν d on V ds , figure 5(b).To derive R ext , we measured I photo at different optical powers via an attenuator ranging from 510 to 0.32 μW, figure 5(c).Figure 5(c) illustrates an increase in R ext , starting from 8.87 A W −1 , and reaching approximately 520 A W −1 at wavelength of 405 nm when V ds is at approximately 1 V with an optical power of 0.32 μW.At an optical power of approximately 0.32 μW the number of photogenerated carriers decreases, resulting in an increase in the built-in field at the SLG and QDs interface.This increase explains the enhancement of R ext at lower optical powers.I photo decreases with decreasing optical power (R ext increases with decreasing optical power), consistent with the findings reported by [13,41].This change in R ext can be explained by the shielding of the built-in electric field.When light is illuminated into the photosensitive layer of the device, the generated electron-hole pairs are separated by the built-in electric field created at the interface between the SLG and QDs, causing holes to transfer into the SLG and increase its hole concentration, resulting in higher conductivity.At high light intensities, more electrons at QDs/ SLG interface recombine with holes, resulting in a lower responsivity.However, as the light intensity decreases, fewer electrons at the interface are recombined, reflecting a higher responsivity.The device exhibited an R ext of approximately 3.63 A W −1 at 1 V with a wavelength of 520 nm (Supplementary Information 6), consistent with the absorption spectrum of QDs in figure 4(a).We measure the temporal response decay in our PDs and get a response time ∼16.7 s (Supplementary Information 7).
Normalized detectivity (D * (cm.Hz 1/2 /W or Jones)) relates the performance of PDs in terms of R ext to the photoactive area of the PD, allowing the comparison of PDs with different active areas.The noise current in the shot noise limit is defined as I n = (2.q.I dark ) 1/2 .Thus D * on the shot noise limit can be expressed as: where B is the bandwidth and A is the photoactive area of the PD.D * is calculated as ∼3.6 × 10 12 Jones for our PD, which is 10 times greater than the highest reported detectivity of any current paper-based PD [42].Reference Flexibility is a characteristic of great desirability for PDs with wearable devices being a leading application for this technology.Therefore, the performance of our PD was tested over the course of a 500 bending cycle test to assess its effectiveness when subjected to cyclic strain.A bending radius of 4.5 mm was introduced, and a single point bending test was  conducted.Both the dark current and current under 0.51 mW light were recorded with the device at rest (i.e.prior to bending), to calculate the photocurrent at rest (I photo(Rest) ).Following this, the device was then bent at a series of intervals up to 500 bending cycles.Figure 7(a) plots I photo(Bend) /I photo(Rest) as a function of bending cycles.It can be observed that the device demonstrated a decay in performance with increasing bends, such that after 500 bending cycles, I photo(Bend) /I photo(Rest) drops ∼15% of initial cycle.SEM images of only the paper substrate, both before and after bending at a bending radius of 4.5 mm, revealed the formation of cracks, which could contribute to the observed drop in the photocurrent of our photodetectors (Supplementary Information 8).Additionally, this behaviour, where the device response initially experiences a downward drift for the first several cycles and subsequently maintains a stabilized trend, is also observed in other sensors [57][58][59].This could be attributed to the construction of some new conductive networks and the subsequent formation of an equilibrium state.In our QDs/SLG PDs, it was worth mentioning across all bending cycles a photocurrent was observed and device retained its ohmic contact (linear photocurrent versus voltage), indicating the device remained functional throughout all tests, figure 7(b).We believed improvements in this metric could come from an increase in the number of SLG layers used, as this would provide a more durable structure and allow current to travel through other SLG layers should initial layers fail.This would result in a high constant photocurrent bend/photocurrent ratio and further justify the use case of this fabrication technique in the industry.

Conclusion
We demonstrated high-performance paper-based PDs composed of CVD SLG and CsPbBr 3 perovskite QDs via spraylithography under ambient conditions.We achieved a maximum R ext of 520 A W −1 at 405 nm with <1 V operating voltage.Our device fabrication approach has proven suitable for the development of scalable electronics on a paper substrate.This work showcases the significant potential of non-vacuum fabrication of high-performance active devices for next-generation electronics on paper, enabling cost-efficient, environmentally friendly, and sustainable practical applications.

Figure 1 .
Figure 1.The QDs/SLG PD fabrication steps on paper substrate.(a) CVD SLG was transferred on paper substrate, followed by overnight drying and PMMA removal using acetone/IPA.Then, CNT ink was spray-coated through the mask to make electrodes.The sample was placed on a hot plate at ∼100 °C for ∼20 min.The image shows the spray-coated CNT ink after ∼20 min annealing.(b) PS was spraycoated on SLG.The image shows spray-coated PS as mask on SLG.(c) SLG was then etched via Plasma Etcher.The image shows the PS ink on SLG during plasma etching.(d) PS was removed by rinsing with water and then QDs are spray-coated through the mask.The image shows patterned SLG after removal of PS ink with water and spray-coated QDs.(e) Schematic and image of the completed array QDs/SLG PDs on the paper substrate.

Figure 2 .
Figure 2. 514.5 nm Raman spectra of CVD SLG on Cu and after patterning and PS removal on paper substrate.

Figure 3 .
Figure 3. (a) Absorbance of dispersed MWCNT in DMF.The dispersions were diluted to avoid detector saturation.Inset image shows dispersed CNT ink.(b) Raman spectra at 514.5 nm of MWCNT.(c) The FTIR of MWCNT ink.

Figure 5 .
Figure 5. (a) Current as a function of bias under dark and illumination, 405 nm and 0.51 mW.The inset image, is the schematic band diagram of QDs/SLG interface, showing the QDs conduction band and valence band, generation of e/h pairs and transfer of h from QDs to SLG.(b) Photocurrent as a function of bias for different illumination powers.(c) R ext and photocurrent as a function of optical power density.

[ 42 ]
reports a selfbiased PD that uses Ga 2 O 3 and exhibits a detectivity of 1.42 × 10 11 Jones with a responsivity of 3.1 × 10 -3 A W −1 under a 254 nm illumination.[17]reports a R ext of 1.3 A W −1 with a detectivity of 7.7 × 10 10 Jones under a 1 V bias and 365 nm wavelength composed of MAPbBr 3 .Thus, the device presented in this paper is superior through both detectivity and responsivity metrics, demonstrating great promise for this novel architecture and fabrication method.Figures6(a), (b) shows different paper-based PDs from recent published work and their corresponding parameters: operating voltage, R ext , operating wavelength, D * , and flexibility [16, 17, 43-56].Compared with other PDs, the responsivity of our work (520 A W −1 ) at only a 1 V bias shows great potential in flexible PD industry.

Figure 6 .
Figure 6.A comparison of different paper-based PDs regarding (a) operating voltage, R ext , operational wavelength, and bending cycles, and (b) operating voltage and D * .

Figure 7 .
Figure 7. (a) I photo(Bend) normalized to that measured on flat PD (I photo(Rest) ), as a function of bending radius.Inset image is the QDs/ SLG PD array on paper substrate.Scale bar is 1 cm.(b) I photo(Bend) normalized to that on flat PD (I photo(Rest) ) as a function of bending cycle.