Flexible electronics is emerging as a new technology that can enable applications, including wearables, medical devices, solar cells, and displays. This technology offers solutions to many of today’s challenges, such as energy efficiency and sustainability^[1–4]. Photodetectors (PDs) have a wide range of uses in the fields of imaging, telecommunications, defense sensing, and ozone-hole monitoring as an essential branch of flexible optoelectronics. Due to their versatile properties, they are indispensable for a variety of cutting-edge applications. To develop wearable PDs, it is necessary to optimize key components such as substrates, electrodes, and light-absorption materials in order to fulfill the criteria of efficiency, light weight, flexibility, and stability^[5–8]. Researchers have long focused on establishing wearable and flexible opto/electronic devices on nonbiodegradable polydimethylsiloxane (PDMS) and polyethylene terephthalate (PET) substrates in order to maximize their durability and performance^[9–12]. On the other hand, the features of paper substrates such as light weight, easy availability, biodegradability, and nontoxicity make them a desirable choice for future intelligent electronics. With their cost-effectiveness and ability to produce ultralarge-area devices, paper provides an ecofriendly solution for commercial applications^[13–16]. Recently, paper-based flexible ZnS/MoS2 PDs have been proven to endure up to 102 to 103 of bending cycles without loss of stability, with a remarkable photoconductive responsivity of 17.8 µA/W^[17]. WSe2 nanodots have enabled the large-scale development of flexible PDs on paper substrates, and these have presented remarkable photoresponse abilities with a detectivity of 5.86×108 Jones, responsivity of 796.18 mA/W, and response time of 0.68 s^[18].

Gallium oxide (Ga2O3) has grown increasingly popular in the past few decades due to its ultrawide bandgap semiconductor capabilities^[19–25]. Its bandgap of around 4.9 eV fits nicely within the solar-blind spectrum of 200–280 nm, thus enabling PDs to detect light in this invisible ultraviolet (UV) range—unfathomable to the human eye^[26–29]. These detectors can facilitate clear imaging of an object even in direct sunlight, thus allowing for more precise sensing and monitoring. Solar-blind PDs are fundamental for a variety of applications, such as night vision, biological/chemical sensing, measuring air pollution, and detecting UV light. Ga2O3-based PDs are one kind of the intrigue semiconductors for solar-blind photodetection, providing use in devices for ozone hole detection and UV communication^[26,30,31]. So far, flexible Ga2O3-based PDs are still rarely reported, which restricts their usability in more demanding environmental conditions such as wearable, flexible, and portable applications^[7,32–34]. Herein, amorphous Ga2O3 (a-Ga2O3) thin films on ordinary printing-used paper were obtained by magnetron sputtering at room temperature. X-ray photoelectron spectroscopy (XPS) and UV-VIS absorbance spectrum measurements confirmed that the obtained Ga2O3 films had a stoichiometric ratio and a bandgap of 4.8 eV. Its solar-blind UV detection performance was tested through optical-coupled electrical measurements. The performance test results show that the paper-based film exhibits obvious solar-blind photoelectric response effect. Upon 254 nm illumination at 1000μW/cm2, the PDs exhibited an impressive responsivity of 66 mA/W, external quantum efficiency (EQE) of 20%, and photo-to-dark current ratio (PDCR) of 1×105. Further, to demonstrate the devices’ compatibility under complicated deformation status when integrated on variant flexible electronic devices, we also conducted a flexing resilience test showing that the paper-based flexible PD retains good detectivity performance under and after various twisting angles and bending radii.

High purity Ga2O3 ceramic (99.99%) with a 2-inch (1 inch = 2.54 cm) diameter was employed as the target (KJMTI Co., Ltd.), while commercial A4 paper was used as substrates. 500 nm thick a-Ga2O3 thin films were grown on the paper substrates through magnetron sputtering at room temperature with the following parameters: argon gas pressure of 0.5 Pa and sputtering power of 70 W. Au/Ti (30 nm/20 nm) interdigital electrodes were fabricated on the a-Ga2O3 film’s surface using a hard mask and sputtering techniques. The schematic device structure diagram is shown in Fig. 1(a). X-ray diffraction (XRD) was utilized to study the amorphous nature of the as-grown films by means of a Bruker D8 Discover. XPS analysis was conducted via a Thermo ESCALAB 250xi. A Keithley 4200-SCS equipped with a probe station was used for photoresponse characteristics.

XRD measurements were carried out on the paper-based a-Ga2O3 film and compared to those obtained from sapphire substrates deposited at room temperature. As shown in Fig. 1(b), apart from the diffraction patterns from the paper substrates (mainly cellulose, along with other clay minerals, calcium carbonate, and titanium dioxide used for surface treatment), no diffraction peaks of the Ga2O3 film were detected. When the same deposition parameters were applied to the sapphire substrate, the film was still noncrystalline, with only the substrate signals detected in the XRD characterization. XPS is an analytical technique used to measure the binding energy of electrons emitted when the sample is bombarded with X rays, enabling it to identify the atomic composition, electronic structure, and chemical bonding state of the sample. All peak positions were calibrated using the C 1s peak at 284.8 eV. Figure 2(a) illustrates the survey for XPS spectra of deposited a-Ga2O3 thin film. The data are shown in the binding energy range of 0–1200 eV for the as-grown a-Ga2O3 film. No other elements peaks except for Ga, O, and C elements were observed within the sensitivity range of the technique in our sample. The Ga 2p spectrum [Fig. 2(b)] can be deconvoluted into two peaks with Ga 2p1/2 located at 1145.5 eV and Ga 2p3/2 located at 1118.5 eV. The separation energy value between these two peaks is about 27 eV, which is consistent with the binding energy (∼26.9eV) of the Ga 2p^[35]. As shown in Fig. 2(c), the O 1s core level peak of a-Ga2O3 thin film is asymmetric; it can be divided into two components via the Gaussian fitting method, which is located at 530.7 and 531.4 eV, respectively. The peak at 530.7 eV is assigned to lattice oxygen ions, and the peak at 531.4 eV is assigned to the oxygen ions in the oxygen vacancies region. The peak ratio of I531.7/(I530.4+I531.7) is about 0.59, implying the existence of abundant oxygen vacancies in the film^[36]. The oxygen vacancy in as-deposited a-Ga2O3 film can be attributed to the low temperature deposition process. As low substrate temperature (room temperature) has restricted the incorporation of oxygen into films, the as-deposited films have a great oxygen deficiency. To reveal the bandgap of a-Ga2O3, a UV-VIS absorbance spectrum is measured by comparing the intensity of light of different wavelengths that passes through the sample with the incident intensity for each wavelength. To obtain the absorbance spectrum, transparent substrates must be used for thin film deposition, with sapphire chosen for our experiment. Figure 2(d) shows a significant absorption edge at a wavelength of ∼260nm for the a-Ga2O3 thin film prepared on double-sided polished Al2O3 substrate under the same conditions as for the paper deposition. As a direct bandgap semiconductor, the bandgap of the a-Ga2O3 thin film is calculated by extrapolating the linear region of the plot of (αhν)2 versus hν and taking the intercept on the hν axis. The estimated bandgap is about 4.8 eV, as shown in the inset of Fig. 2(d), which is smaller than our previous results due to the oxygen deficiency.

Au/Ti interdigital electrodes are deposited using magnetron sputtering to form a metal–semiconductor–metal structure PD. The inset of Fig. 3(a) shows a picture of the fabricated device and also plots the dark current versus the voltage curve. This linear relationship between dark current and voltage confirms that an ohmic contact has been successfully obtained, as illustrated vividly. The dark current is approximately 23 pA at 10 V, which is extremely low. Figure 3(b) shows the I-V characteristics of the device under both dark and 254 nm light illumination, with a sequence of light intensities. The curves of current under different light intensities are differentiated by different colors. When the light intensity is below 100μW/cm2, a small increase in light intensity can cause a large increase in current. For example, compared to in dark and under 100μW/cm2 illumination, the current has an obvious improvement of up to 4 orders of magnitude. With the highest light intensity of 1500μW/cm2, the current under 10 V can reach 3×10−6A. We have fitted the data from Fig. 3(b) to explore the relationship between current and light intensity, under 10 V bias. As illustrated in Fig. 3(c), the current increases with light intensity, showing that the device has good detectivity, even when light intensity is below 1μW/cm2. To further confirm the solar-blind band selectivity, the device’s spectral response was studied. UV light with longer wavelengths corresponds to a lower excitation energy. As shown in Fig. 3(d), maximum responsivity is observed around 258 nm, which is consistent with the bandgap of a-Ga2O3 films (4.8 eV). This demonstrates that the prepared PDs have good spectral selectivity.

The responsivity (R), detectivity (D*), PDCR, and EQE are the four most important parameters used for the quantitative evaluation of detector performance^[28]. These parameters were investigated under various light intensities at 10 V; the results are given in Fig. 4. R denotes the ratio of the photocurrent flowing in the PD to the incident optical power, which can be calculated using Eq. (1), R=Iph−IdarkP·S,(1)where Iph and Idark are photocurrent and dark current, P is light intensity, and S is the effective illumination area. In this work, R can reach a peak of 66 mA/W under 1000μW/cm2 254 nm illumination. The D*, which indicates the capacity of the detector to detect the smallest signal, is calculated using Eq. (2), D*=RS12(2eIdark)12,(2)where R, S, and Idark are given above, and e is the charge of the electron. The maximum D* achieves 3×1012 Jones under 1000μW/cm2 254 nm illumination. PDCR is mathematically denoted as Iph/Idark, and is generally used to describe the anti-interference performance under background noise. From the figure it can be observed that the PDCR curve exhibits excellent linearity. The peak value can be attained to 1×105 under 1000μW/cm2 254 nm illumination. EQE, as a critical parameter for quantifying the behavior of PDs, can be evaluated from Eq. (3), EQE=IphP·S×hceλ×100%,(3)where c is the speed of light, h is Planck’s constant, e is the elementary charge, and λ is the wavelength of the incident light. As shown in Fig. 4(b), an EQE of 20% is achieved.

The excitation and migration of carriers takes time, meaning that the signal current from the PD does not reach the maximum value immediately upon incident light irradiation. The photocurrent of the device generally refers to the current after the incident light has stabilized. Usually, the faster the response speed of a PD, the better its performance is. The response speed of the PD can be quantified to evaluate the performance of the device response to transient irradiation generated by the power switching process. Usually, the rising edge τr and falling edge τd of the instantaneous current of the incident optical switch are used to characterize the response speed of the PD. In order to obtain the response speed of the detector, an exponential function fitting method was used to fit the change curve of the device’s response current during the incident light irradiation switching on/off process. This enabled the estimation of the response times τr and τd, respectively, I=I0+A1e−tτ1+A2e−tτ2,(4)where I0 is the photocurrent in steady state, A1 and A2 are constants, t is the time scale, and τ is the relaxation time constant. For our paper-based PD, the calculated response time values are shown in Fig. 5, where varying response speeds were achieved under different light intensities. Taking τr1 and τd1 as an identifiable signal, the response time of less than 1 s was achieved under different levels of illumination.

Flexible electronics generally requires devices to remain structurally intact and operational even after being subjected to twisting, bending, and stretching. For the paper-based devices, the stretching behaviors are always constrained by the mechanical properties of the paper used, which is generally unstretchable. Considering the actual working environment for the paper-based electronics, we conducted I-V measurements of the a-Ga2O3 flexible PD at various twisting angles under fixed illumination and bending radii. A twisting fixture was designed to twist the paper device at different angles. For the bending test, the paper device was attached to cylinders of different radii. In Figs. 6(a) and 6(b), it is shown that neither the twisting nor bending deformation had an obvious effect on performance. We did notice that in Fig. 6(b), there was a slight decrease in current when a smaller bending radius was applied. This could be attributed to the change in light illumination conditions, as the light source we used was incident from a fixed position, and the bending changed the incidence angle. For devices being subjected to continuous mechanical disturbance, they should also be able to sustain their performance without any significant degradation after cycles of bending. In Figs. 6(c) and 6(d), the time-dependent photocurrent with different bendings applied and the intensity of the device’s photocurrent as a function of the number of bending cycles at a bending radius of 8 mm were measured. The results demonstrate good retention ability after 1000 cycles of bending.

In conclusion, we demonstrated a flexible solar-blind PD by depositing a-Ga2O3 on paper. Experimental results revealed that the a-Ga2O3 has a direct bandgap of 4.8 eV and shows good spectral selectivity within the solar-blind region. The responsivity, detectivity, PDCR, EQE, and response speed were characterized to verify the device performance of the paper-based PD. Furthermore, the flexible device was able to sustain its performance after cycles of mechanical disturbance. Our experiments showcase that the wide bandgap semiconductor Ga2O3 can be integrated into paper-based electronics, providing solutions to many of the challenges faced by flexible electronics.