Bipolar Photoresponse in Graphene/GaN Heterostructure and its Secure Function in Free‐Space Optical Communication

The free‐space optical communication is regarded as a promising technique for next generation networks. However, all the data are exposed in free space with high risk of being attacked or eavesdropping by unauthorized parties. Here, a bipolar photodetector based on the graphene/GaN heterojunction is demonstrated. The polarity of the UV light photocurrent is opposite to that of the red light photocurrent, which results from the interface state filling effect and hot carrier injection into graphene, verified by Kelvin probe force microscope measurement results. Four stable photocurrent levels are demonstrated with only a single graphene/GaN heterojunction via the photocurrent polarity control, which is employed for a secure capability in conventional optical communication by setting visible and UV light as secret and key information, respectively. The technique provides a new strategy to design photodetectors for information encryption technology.


Introduction
With rapid demands for anytime and anywhere communications, the free-space optical (FSO) communication is regarded as a DOI: 10.1002/aelm.2023002433][4][5] However, all the data are exposed in the long and wide free space for conventional FSO communication systems, with high risk of being attacked or eavesdropping by unauthorized party. [6,7]uantum secure communication is an absolutely secure technology. [8]However, quantum key generation technology can't meet the practical requirements of FSO communication. [2,9]It is extremely desirable to realize all-optical logic encryption mechanism and corresponding decryption hardware in real time for future secure FSO communication systems.
Existing efforts for obtaining encrypted and key information carried by dual lights involved two separate photodiodes with different response spectra, but complex configuration, complicated data processing and power consumption are inherent in these systems.All-in-one bipolar photodetector, which can control or discriminate the direction of photocarrier transport in real time, [10] is highly desired to be developed with the capability of differentiating encrypted information from key information in real time.However, limited by the lattice mismatch, or deposition temperature, it is hard to integrate narrow-and wide-bandgap hetero semiconductor materials in a single chip with high sensitivity.
[17][18] Recently, wavelength-induced bipolar photodetectors are developed, whose photocurrent transport direction is opposite with duallight incidence, such as back-to-back p + -i-n-p-p + self-powered perovskite photodetector, [19] showing potential applications in multifunctional logic gates.However, such structure is a lowband and a high-band photodetector back-to-back connected in series.Besides the signal cross-talk between these two photodetectors, the bipolar photocurrents can be obtained only when the two lights with different wavelength are incident from front and back side, respectively.Such structure cannot work when lights are incident from the same side, which is inconvenient in real applications.
In this work, an all-in-one bipolar photodetector based on graphene/unintentionally doped (UID) GaN heterojunction is proposed.The polarity alternates with UV and red light illumination incident from the same side, and their logic relationship of current level among UV light, red light, UV plus red light together and dark is kept constant with 100% accuracy due to two different working mechanism of interface state filling effect and hot carrier injection into graphene inside a single photodetection structure, verified by the Kelvin probe force microscope (KPFM) measurement results.With the prototypes of the bipolar photodetector, we demonstrate a secure FSO communication by setting visible and UV light as secret and key information, respectively.The technique provides a new strategy to design photodetectors for information encryption technology.

Results and Discussion
Figure 1a illustrates the schematic configuration of the proposed bipolar hybrid graphene/GaN photodetector, in which graphene, grown on the Cu foil by chemical vapor deposition (CVD) method, is placed on top of the GaN surface by wettransferring process (Experimental Section). [20]Source (S) and drain (D) electrodes are utilized for photocurrent measurement under bias with UV and red light illumination.Figure 1b shows the scanning electron microscope (SEM) image of graphene on the Cu foil.The obvious wrinkles are attributed to thermal expansion coefficient discrepancy between Cu and graphene. [21]The presence of metal steps indicates that graphene covers the Cu surface completely. [22]Raman spectrum results (Figure 1c) show that the G peak and 2D peak of the transferred graphene are located at 1587 and 2677 cm −1 , respectively.A few blue-shift of the G peak, compared with undoped graphene (1580 cm −1 ), indicates the transferred graphene is lightly p-type doped with the Fermi level below the Dirac point perhaps due to charge transferinduced doping effect. [23,24]The p-type doping property is also confirmed by the Hall measurements with 3.45 × 10 12 cm −2 hole concentration at room temperature, which estimates the Fermilevel of graphene 0.22 eV below the Dirac point (Figure S4, Supporting Information). [25]D peak (≈1350 cm −1 ), which is related to disorder or defects, is almost invisible during the Raman spectrum mapping measurement within the range of 50 × 50 μm 2 (inset of Figure 1c), indicating the high quality of the graphene. [22]n addition, the ratio of I G /I 2D is ≈0.5 and the FWHM of the 2D peak is ≈30 cm −1 , in consistence with the single-layer and continuous features of the graphene sample (Figure S1b, Supporting Information).[22,24] Figure 2a presents transient S-D photocurrent ( I photo = I light − I dark ) of the hybrid graphene/UID GaN photodetector when under the illumination of 650 nm, 261 nm laser light individually and then illumination simultaneously in sequence, which demonstrates obvious dual polarities.Compared with the positive photocurrent under the red light illumination I Red , the photocurrent under the UV light illumination I UV demonstrates obvious negative photoresponse, which indicates the photoresponse polarity alternates from positive or negative with red or UV light being switched on for a graphene/GaN heterojunction.However, under UV plus red light illumination together (U+R for short), the photocurrent I U + R increase obviously compared with I UV .I U + R even can be positive current with the UV light incident power is lower enough, but always lower than I Red .While withdrawing the red light, the photocurrent returns to its original current level.The measured S-D current level from high to low as a sequence is I Red , dark current, I U + R , and I UV , whose logic relationship is stable in descending order.
The photoresponse change (I U + R − I UV ) under different UV and red light power were then measured, respectively.When keeping UV light power as constant, the photocurrent change increases in a step-like manner with red light power varying from 13 μW to 2.6 mW in Figure 2b.Similarly, when keeping the red light power as constant, the photocurrent change also increases with the UV light power from 0.09 nW to 600 μW in Figure 2c.It is obvious that the U+R photocurrent can be affected by red and UV light power independently.
Responsivity R UV (I UV /P UV ) decreases with the UV light power, as shown in Figure 2d.The maximum responsivity of 111 A W −1 with current gain of 530 and specific detectivity of ≈9.79 ×10 8 Jones (Figure S8, Supporting Information) is achieved under the UV light power of 0.09 nW at bias of 1 V.The prolonged lifetime of photoexcited carriers owing to the separation in space results to the ultrahigh responsivity and gain as well. [26]The responsivity R U + R = (|I U + R − I UV |/P Red ) is defined to quantitatively illustrate the red light modulation ability of UV light response.Contrary to the descending trend of R UV , when UV plus red light illuminating simultaneously, R U + R behaves an upward trend from 1.7 × 10 −4 to 2.4 × 10 −3 A W −1 while keeping red light power at 2.6 mW at bias of 1 V with gain from 3.2 × 10 −4 to 4.7 × 10 −3 and specific detectivity from ≈1.49 ×10 3 to ≈2.15 ×10 4 Jones (Figure S8, Supporting Information), which is resulted from the positive relationship of Fermi level change in graphene with UV light in-cident power.Keeping UV light power of 600 μW at bias of −1 to 1 V, R U + R increases with bias approximately linearly, while decreases with the incident red light power (Figure 2e).The highest responsivity of ≈190 A W −1 is obtained at 1 V under the incident red light power of ≈1.3 nW.
The rise and decay time are also affected by the red light illumination.Under the UV light illumination, the rise and decay time are 5.0 and 5.3 s, respectively.However, when UV plus red light illuminating together, they change to 1.1 and 4.8 s, respectively (Figure S10, Supporting Information).And the rise and decay time decrease with red light power while keeping UV light power as constant (Figure 2f).[29] When UV plus red light incidence together, the improved transient time is resulted from the carrier filling of the interface states excited by red light.The slower decay time than rise time is attributed to the slow carrier detrapping process from the interface state, which is often described as the persistent photoconductance phenomenon. [30]Figure 3a shows the surface potential (SP) mapping results of graphene side in graphene/GaN heterostructure by KPFM measurement under dark, UV light, red light, and UV plus red light illumination, respectively.33][34] Figure 3b is the extracted SP from the dotted lines marked in Figure 3a.Negative photoresponse to UV light illumination for graphene/GaN heterojunction is resulted from the absorption by GaN (Figure 3c).After the photogenerated electronhole pairs separate under the action of graphene/GaN heterojunction electric field, electrons will transfer to graphene layer.The electron injection into the graphene decreases its Fermi level and then the current decreases under UV light illumination, which results to the negative photoresponse.The rise result of SP verifies the explanation of the negative photoresponse phenomenon under UV light illumination (Figure S11, Supporting Information).
The SP descends a little bit when red light is added compared with that under the UV light illumination.We attribute the descending to the hot carrier injection at graphene/GaN interface.Under the red light illumination, electrons in GaN layer can be excited and then trapped by the interface states at the graphene/GaN interface.If the trapped electrons are large enough, the potential barrier in GaN will be lowered (Figure 3d).The photogenerated holes will re-inject into graphene in order to realize re-equilibrium of electrochemical potential between graphene and GaN.The hole injection then reduces the Fermi level in graphene compared with that under UV light illumination.Compared with that under dark environment, the difference of Fermi level in graphene from that under UV plus red light illumination is smaller than from that under UV light illumination, which results to I UV is obviously lower than I U + R (Figure 2a).
On the other words, because of the large number of trapped state at the graphene/GaN interface, trapped electrons will cause the electrochemical potential in GaN side decreasing from equilibrium.For the coupled system of graphene and GaN, the change in the electrochemical potential of the GaN will lead to a change in the electrochemical potential of graphene via the Fermi level alignment tendency correspondingly.When the electrochemical potential of the graphene/GaN system comes to re-equilibrium, holes can then be injected into graphene by diffusion process from the GaN, which results in increased hole density in graphene.The trapped photogenerated electrons at the interface excited by red light first tune the electrochemical potential and then tune the carrier density in graphene, which leads to an amplified photocurrent.Due to the absorption of graphene itself is limited (≈7.5% under 261 nm), [35,36] which cannot generate such large apparent photocurrent (≈μA).Interface state filling effect is the origin of the photoresponse to the red light illumination, which results to bipolar characteristic for graphene/GaN heterojunction.Figure S12 (Supporting Information) demonstrates the bipolar performance by utilizing 532 nm/UV and 850 nm/UV light source combination, respectively.Realization of bipolar photoresponse characteristic verifies our proposed interface state filling effect by hot carriers.
Based on the prototype of the bipolar photodetector, we demonstrate a secure FSO communication by setting visible and UV light as secret and key information, respectively.Figure 4a illustrates the time-dependent current response of the bipolar photodetector from dark to red light, UV plus red light, UV light and then dark in sequence."0" indicates light off and "1" indicates light on, which illustrate the switch on/off sequence on the uppermost of the diagram.Four photocurrent levels are assigned by a two-bit binary number, namely "00", "01", "10", "11", corresponding to "UV on", "U+R on", "dark state", and "Red on", respectively.The truth table for these four states is depicted in Figure 4b, which reflects the logic functions.The comparative relationship of these four photocurrent levels will not change with the input light power due to the stable bipolar characteristic of such photodetectors, which indicates that these four states can be utilized for logic coding in secure optical communication.
With the specifically designed photodetector that features bipolar photoresponse, the encrypted and key signals can be distinguished from the detected current in real time according to the one-to-one correspondence of the truth table.For example, as shown in Figure 4c, a sequence of photocurrent levels of "01" "11" "10" "00" "01" "00" is obtained by the bipolar photodetector.According to the truth table, it can be concluded that the red light coding sequence should be "110 010", while the UV light coding sequence should be "100 111", as illustrated in Figure 4d.If a two-verifier protocol is assumed that only when the UV light signal is "1", the red light signal is true while the others are interfering terms.Hence the true information of "1010" can then be successfully decrypted from the original red light coded signals which involve both true and disturbing information.Based on the above discussion, Figure 4e graphically summarizes the framework of the information encryption scheme.
Such a scheme for FSO communication has the benefit of confidentiality.Both red and UV light signals can be intercepted or stolen arbitrarily by multiple eavesdroppers positioned between the information sender and receiver utilizing a regular wide spectrum photodetector, but it is impossible to differentiate the significant information from the two lights individually, implying a confidential and secure communication.

Conclusion
In summary, we reported an all-in-one bipolar graphene/GaN photodetector to demonstrate four stable 100% accuracy photocurrent levels that can build advanced photodetectors for secure FSO communication.Two types of working mechanisms of photodetection are inside a single photodetector, which are interface state filling effect for red light photocurrent and hot carrier injection into graphene for UV light photocurrent, respectively.These two individual working mechanisms cause photocurrent with opposite polarity, which is distinguishable.Thus the output state of "00" "01" "10" or "11" is determined by the combination of "on" or "off" state of incident light, which allows information decryption by a single photodetector.In the free and wide space, any unauthorized party can eavesdrop red or UV light signals, which are not given practical significance.The encrypted data (real information) tried to be transferred can be "read" by such a bipolar photodetector under specific instructions (or encoding principle).This development can be applied to secure optical communication based on the photocurrent polarity.

Experimental Section
CVD Growth of Graphene: The preparation of graphene consists of four stages: heating, annealing, growing, and cooling process.First, the polished Cu foil was loaded into a quartz tube furnace, which was then heated up to 1030 °C with 10 sccm H 2 .Cu foil was then annealed at 1030 °C for 30 min to recognize Cu atoms on the Cu surface.Both two processes were under 10 sccm H 2 .After annealing, CH 4 was introduced as carbon source to grow graphene on Cu substrate.Under the dual catalysis of H 2 and Cu foil, graphene was grown on Cu surface by catalytic decomposition of CH 4 with 1 sccm.Then, CH 4 was turned off, while H 2 continued flowing until the furnace cooled to room temperature.
Characterizations of the Graphene and GaN: SEM (FEI Nano-SEM Nova 400) was used to characterize the structure of monolayer graphene.Raman measurements were conducted using a Witec alpha 300R Confocal Raman Spectrometer.The excitation laser had wavelength  = 532 nm for measurement of as-grown single layer graphene after it had been transferred to the 285 nm SiO 2 /Si substrate.The used laser power was 10 mW.For the single spectra, the Raman configuration was 0.5 s cycling for 10 times.For the large area scanning, the scanning duration was 0.1 s and the step was 0.3 μm.The quality of GaN was characterized by XRD measurements.Hall measurements were performed using the Van der Pauw Hall test to evaluate the doping type, carrier concentration and carrier mobility of the transferred graphene.
Fabrication of the Photodetector: The GaN substrate studied in this paper was grown on sapphire.After cleaning in acetone and ethanol solution, standard photolithography and etching processes were performed to determine the contact pattern.Then Ti/Al/Ni/Au (15 nm/200 nm/15 nm/50 nm) electrode was prepared by magnetron sputtering method, and rapid thermal annealing at 450 °C for 30 s in N 2 .The electrode size was 200 × 100 μm 2 , and the photosensitive area was about 0.02 mm 2 .Monolayer graphene films were grown on Cu foil by chemical vapor deposition.The as-grown graphene was further coated with polymethyl methacrylate (PMMA) and then floated on ammonium persulfate solution for about 6 h to ensure the complete etching of Cu foil.Subsequently, the graphene/PMMA was rinsing to deionized water to remove the residual solvents and then transferred to GaN substrate.The PMMA on the graphene was soaked in acetone for about 4 h to dissolve, and gently washed away with ethanol, and deionized water.Finally, after heating at 60 °C for 20 min to evaporate the water vapor, the device was finished.
Measurements of the Photodetectors: Photoelectric measurement was carried out in a dark room using a semiconductor parameter analyzer (Keithley 4200) under different power illumination at room temperature.A series of neutral density filters were used to attenuate the laser intensity at 261, 532, and 650 nm.These devices were wired to a printed circuit board for the measurements.The SP of graphene and GaN was tested under environmental conditions using an Asylum Cypher S AFM (Oxford Instruments-Asylum Research, Santa Barbara, USA) with an excitation laser of 532 nm.

Figure 1 .
Figure 1.Schematic structure and characterization of a bipolar graphene/GaN photodetector.a) Three-dimensional schematic configuration of the proposed bipolar hybrid graphene/GaN photodetector and its electrical and optical measurement sketches with UV and red light illumination.b) SEM image of graphene on a Cu foil, where the presence of steps on the Cu surface indicates that graphene is continuous.c) Raman spectra of graphene transferred on SiO 2 /Si substrate, which indicates a monolayer graphene film.The inset show its mapping results of the D, 2D, G peak, and I G /I 2D of graphene film, respectively.

Figure 2 .
Figure 2. Bipolar photoresponse based on hybrid graphene/UID GaN photodetector.a) The transient photocurrent measurement results under dark, red light (650 nm, 2.6 mW), UV light (261 nm, 600 μW), and UV plus red light together at 1 V, which demonstrates obvious dual polarities.The uppermost of the figure illustrates the switch on/off sequence of the red and UV light.b) Photoresponse change (I U + R − I UV ) increases with the red light power of 1.3 μW, 0.13 mW, 1.3 mW, and 2.6 mW while keeping UV light power as 600 μW at 1 V. c) Photoresponse change increases with the UV light power of 0.09 nW, 30 μW, 180 μW and 600 μW while keeping the red light power as 2.6 mW at 1 V. d) Dependence of UV light responsivity (R UV ) and UV light gain (G UV ) under UV light illumination, R U + R and G U + R (keeping red light power of 2.6 mW) under UV plus red light illumination at 1 V, which demonstrates opposite trend.e) Dependence of R U + R on bias voltage at different red light power while keeping UV light power of 600 μW at 1 V. f) The rise and decay time as a function of red light power while keeping UV light power of 600 μW at 1 V.The inset shows the schematic representation of the carrier transport process.

Figure 3 .
Figure 3. Bipolar photoresponse induced by interface hot carrier mechanism.a) The surface potential (SP) mapping of graphene side in graphene/GaN heterostructure in dark, under UV light, red light, and UV plus red light illumination, respectively.b) Line profiles of the SP in dark, under UV light, red light, UV plus red light illumination extracted in the position marked by the dotted lines in (a).Illustration of energy band diagram of graphene/GaN heterostructure under UV light illumination (c) and under UV plus red light illumination simultaneously (d).

Figure 4 .
Figure 4. Scheme for secure optical communication based on the bipolar photodetector.a) The time-dependent current responses of the designed bipolar photodetector.The photocurrent levels are shown from bottom to top as "00", "01", "10", "11", corresponding to UV light, UV plus red light, dark, red light states, respectively.The uppermost of the figure shows the sequence of light switches, with "0" indicating off and "1" indicating on.b) The truth table shows the one-to-one logical relationship between photocurrent levels and optical switching states.c) The bipolar photocurrent in response to the encrypted and key signals.d) The encrypted and key signals distinguished from the bipolar photocurrent according to the truth table, and the decrypted signals further according to the two-verifier protocol.e) The framework of the information encryption scheme.