Breaking the Built‐In Electric Field Barrier in p–n Heterojunction for Self‐Powered, Wavelength Distinguishable Photoelectrochemical Photodetectors: Toward Low Power Consumption and Secure Underwater Wireless Sensor Network

Self‐powered, light wavelength distinguishable photodetectors (PDs) are appealing components to build a robust, secure, and low energy consumption underwater wireless sensor network (UWSN). However, achieving such devices is extremely difficult even today. In this context, the first self‐powered, light wavelength distinguishable PDs with photoelectrochemical (PEC) principles and using tunnel junction (TJ) to overcome the technical hurdles for self‐powered, light wavelength distinguishable PEC‐PDs with p–n junction working electrode is reported. For such devices, a single photoelectrode is used, that is, one photoelectrode is able to distinguish different light wavelengths without using any external electrical power, and they are able to distinguish light wavelengths in both the ultraviolet (UV) and blue wavelength ranges. High responsivities reaching mA/W range and ultrafast response time with less than 10 ms are achieved in self‐powered operation mode. Moreover, such devices are able to operate not only in acidic but also in NaCl electrolyte, making them potentially attractive for applications in ocean environment. In the end, it is demonstrated that leveraging such PEC‐PDs, excellent data security can be achieved in the data communication mimicking that in an UWSN in ocean environment. This study not only represents a breakthrough in PDs, but also significantly advances the development of UWSNs, especially for ocean environment.


Introduction
3][4] Self-powered, wavelength-distinguishable PDs are appealing components to build the needed UWSNs, as the self-powered operation can significantly reduce the energy consumption and maintenance cost in operating UWSNs, whereas the wavelength distinguishability allows complex data encryption when transmitting it using light.7][8][9][10][11][12][13][14][15][16][17][18][19] However, achieving self-powered, wavelengthdistinguishable PEC-PDs remains a challenge.At this point, it should be noted that, a photoelectrode with photoresponse under zero potential versus a reference electrode in a three-electrode configuration does not necessarily make it a self-powered PEC-PD, due to the potential applied by the potentiostat between the photoelectrode and counter electrode.
Semiconductor p-n heterojunction is an emerging route to self-powered, wavelength-distinguishable PEC-PDs. [5,6,11,13]One basic principle of using p-n heterojunction for wavelength distinguishable photodetection is to utilize the polarity of the photocurrent to sense different incident light wavelengths, based on different types of chemical reactions occurring on the semiconductorelectrolyte interface under the light illumination.For n-type semiconductors, due to the upward band bending, the hole transfer process leads to oxygen evolution reaction (OER) and consequently positive photocurrent (Figure 1a), [20][21][22][23] whereas the hydrogen evolution reaction (HER) occurring on the p-type semiconductor-electrolyte interface due to the downward band bending leads to negative photocurrent (Figure 1b). [22,24,25]As such, in a nutshell, if choosing semiconductor p-n heterojunction with proper bandgap energies, wavelength distinguishable light detection can be made possible by sensing the polarity of the photocurrent.
Yet, there have not been any devices demonstrated.One critical issue is related to the built-in electric field at the junction.For example, in the working electrode configuration as shown in Figure 1c (n-down, p-up), due to the built-in electric field at the junction, HER is more difficult to happen (Note: it is a 2-electrode PEC cell), limiting the negative photocurrent.This is because the built-in electric field could have a photovoltaic (PV) effect similar to a solar cell, [26][27][28][29] i.e., upon light illumination on the p-side (whereas n-side remains transparent), the built-in electric field tends to separate the photogenerated electrons and holes near the p-n junction and sweep photogenerated electrons to n-side and photogenerated holes to p-side, making it difficult for HER to occur on the p-side and holes transport to the substrate.Although this could lead to abnormal photocurrent which might be favorable for certain applications, [30] this illustrates the generic difficulty of generating negative photocurrent in the working electrode with the n-down, p-up configuration in a 2-electrode PEC cell, which further limits the achievement of wavelength distinguishable light detection.This also means, in order to achieve wavelength-distinguishable light detection, improving the HER process on the working electrode is necessary.Vice versa, for the p-down, n-up working electrode configuration, the built-in electric field would hinder OER process and positive current generation.This analysis illustrates the difficulty in achieving selfpowered, wavelength-distinguishable PEC-PDs using p-n junction working electrode, which is originated from the built-in electric field at the junction.
In this work, we demonstrate that by integrating TJ into semiconductor nanowire p-n heterojunction, the adverse effect of the built-in electric field can be greatly reduced, which leads to the achievement of the first self-powered, wavelength distinguishable PEC-PDs using a single photoelectrode, i.e., one photoelectrode is able to detect two different light wavelengths without using any external electrical power.In specific, by using a working electrode made with n-GaN/p-InGaN p-n heterojunction nanowires in which an n ++ -GaN/InGaN/p ++ -GaN TJ is embedded, self-powered, wavelength distinguishable PEC-PDs in the visible (405 nm) and UV (302 nm) are obtained, with high responsivities in the mA/W range.Moreover, the electrolyte for such self-powered, wavelength-distinguishable PEC-PDs can be either acidic or NaCl, making them potentially suited for optical wireless communication in the ocean environment.Enhanced data transmission security leveraging such wavelength distinguishable, self-powered PEC-PDs is further demonstrated in the end in the data transmission process mimicking that occurs in an UWSN.

Results and Discussion
PEC-PDs wherein the working electrode is made n-GaN/p-InGaN p-n heterojunction nanowires, without the TJ, are described first.The detailed growth condition for the heterojunction nanowires can be found in the Experimental Section.A scanning electron microscopy (SEM) image of such heterojunction nanowires is shown in Figure 2a, with the nanowire schematic shown in the inset.The SEM image was taken at a wafer-tilting angle of 45°.The heterojunction nanowires are further examined by transmission electron microscopy (TEM).Shown in Figure 2b are the dark field scanning-TEM (STEM) image of a heterojunction nanowire and the corresponding elemental mapping of Ga and In.The n-GaN and p-InGaN segments are clearly seen, confirming the formation of an axial heterojunction.Detailed highresolution TEM studies further confirm that such heterojunction nanowires have excellent crystalline quality in both segments.High-resolution TEM results are described in the Supporting Information.The room-temperature (RT) photoluminescence (PL) spectra of the InGaN and GaN nanowire segments are shown in Figure 2c.The PL peak ≈530 nm suggests that the InGaN nanowire segment has an In content of ≈27-30 mol.% using the Vegard's law. [31,32]This is consistent with the In molar fraction analysis using X-ray photoelectron spectroscopy (XPS).The details are described in the Supporting Information.In addition, the incorporation of Mg is also confirmed by XPS experiments.The PL peak ≈364 nm is attributed to the light emission from the GaN nanowire segment.As such, for both InGaN and GaN nanowire segments, their band edges straddle the water redox potentials, [31,32] and can in principle support both HER and OER reactions (Figure 2d).In this study, we focus on HER for InGaN and OER for GaN, due to the respective p-type and n-type doping.In the meantime, the proper band energies allow photon absorption when 405 and 302 nm, the two light wavelengths used in this study, are applied.33][34][35] The photoresponse of such PEC-PDs is further studied in a 2-electrode configuration (details are in the Experimental Section).The electrolyte is 0.5 m H 2 SO 4 and the counter electrode is Pt. Figure 2e shows the time-dependent photocurrent under the 302 and 405 nm light illumination at 0 V.Under the 302 nm light illumination, both n-GaN and p-InGaN segments absorb the light.For the n-GaN segment, the OER reaction can lead to a positive current, whereas for the p-InGaN segment, the HER reaction can lead to a negative current.However, due to the pres-ence of the built-in electric field (as explained earlier), HER will not be favorable, which in consequence leads to a net positive photocurrent.Under the 405 nm light illumination, the light absorption occurs only in the p-InGaN.In this case, in principle, negative photocurrent is expected; nonetheless, due to the builtin electric field, such a negative photocurrent is hindered, ultimately leading to a near-zero photocurrent.The existence of the photocurrent spike when the light turns on could be related to a transient current, and the mechanism is being investigated.Since the photoresponse increases with the light excitation, [5,8,10] and the above photoresponse is measured under the highest excitation, this means that self-powered wavelength distinguishable photodetection is not achieved, limited by the difficulty in obtaining negative photocurrent due to the unfavorable built-in electric field.
In the following study, we show that by incorporating the TJ into the same p-n heterojunction photoelectrode, self-powered, wavelength distinguishable photodetections can be realized in 2-electrode configuration, i.e., self-powered, wavelength distinguishable PEC-PDs can be achieved.The structure of the TJ InGaN nanowire photoelectrode is schematically shown in Figure 3a, which consists of the same p-InGaN and n-GaN nanowire segments but with an n ++ -GaN/InGaN/p ++ -GaN TJ incorporated.Figure 3b shows the dark field STEM image of a single TJ InGaN nanowire (Figure 3b-i  is clearly seen, and the n-GaN and p-InGaN segments are also clearly seen, confirmation the formation of the desired structure.Additional TEM studies on the TJ can be found in the Supporting Information. The time-dependent photocurrent of such PEC-PDs, made with Pt counter electrode in H 2 SO 4 electrolyte, under the UV (302 nm) and blue (405 nm) light illumination, is shown in Figure 3d.It is seen that both negative (under 405 nm light illumination) and positive photocurrents (under 302 nm light illumination) are measured at 0 V, i.e., self-powered, wavelength distinguishable photodetection is achieved, thanks to obtaining the negative photocurrent under the 405 nm light illumination (which is not the case for devices without the TJ).The mechanism of obtaining the negative photocurrent under the 405 nm light illumination can be understood by the schematic shown in Figure 3f.At the p ++ -GaN/p-InGaN interface, the build-in electric field is in the opposite direction comparing to that at the n-GaN/p-InGaN interface (it is noted that for the n-GaN/n ++ -GaN interface of the TJ, the electric field is also in the opposite direction); as such, the PV effect is now essentially favorable for HER, i.e., the built-in electric field will sweep photogenerated electrons to p-InGaN (or non-doped InGaN) and photogenerated holes to the p ++ -GaN.Moreover, the large potential barrier at the p ++ -GaN/p-InGaN interface will block the transport of photogenerated electrons to the substrate, which also enhances HER.Ultimately, the negative photocurrent under the 405 nm light illumination is obtained.
Comparison of devices with and without the TJ suggests that TJ can effectively overcome the negative role of the built-in electric field in semiconductor p-n junction in achieving selfpowered, wavelength distinguishable PEC-PDs.In fact, phenomenally, self-powered, wavelength distinguishable light detection is even achievable in PEC-PDs made with non-doped InGaN nanowire working electrode when TJ is incorporated (Figure 3e), although non-doped InGaN nanowires have a weakly n-type surface, which is confirmed by XPS experiments (More details can be found in the Supporting Information).As such, it suggests that comparing to the downward surface band bending, TJ may play a greater role in driving HER, consequently negative photocurrent, and eventually self-powered, wavelength distinguishable photodetection.
More detailed photoresponse of such self-powered, wavelength-distinguishable PEC-PDs is shown in Figure 4. Figure 4a-c shows the results under the 302 nm light illumination, including the time-dependent photocurrent, extracted photocurrent density and responsivity versus the light excitation, and the response and the recovery time versus the light excitation.Figure 4d,e shows similar plots but for the 405 nm light illumination.It is seen that for both incident wavelengths, the photocurrent density increases as the excitation increases and nearly stable photocurrent can be obtained.It should be noted that, the decrease of the photocurrent when the light is on does not reflect the device degradation, as the photocurrent in each cycle can be repeated.In fact, we have performed 1 h long test, and devices did not show any degradation.The details can be found in the Supporting Information.The mechanism of this light on transient behavior is being investigated.For 302 nm, a responsivity of ≈4 mA W −1 is achieved, whereas for 405 nm, a responsivity of ≈0.5 mA W −1 is obtained.Strategies such as increasing p-InGaN segment length, increasing In content, surface passivation or using platinum nanoparticles could potentially improve the responsivity for the blue light detection.These responsivities are comparable to the previously reported self-powered, wavelength-distinguishable PEC-PDs in the UV and blue range with two photoelectrodes. [7]ith respect to the response time (t res ), it is defined as the time needed for the photocurrent to increase from 10% to 90% of its maximum value; while for the recovery time (t rec ), it is defined as the period needed for the photocurrent to decrease from 90% to 10% of its maximum value.It is seen that, for the 302 nm illumination, t res of 35 ms and t rec of 355 ms can be obtained; whereas the for the 405 nm illumination, both t res and t rec are limited by the instrumentation, which is 10 ms, suggesting that the actual t res and t rec are less than 10 ms.The large t rec under the 302 nm light illumination could be related to hole accumulation on the surface.
The device performance in this case is shown in Figure 5. Overall, similar performance compared to using H 2 SO 4 electrolyte is obtained, except that the responsivity is slightly reduced: for the 302 nm light illumination, the responsivity is ≈3 mA W −1 ; and for the 405 nm light illumination, the responsivity is up to ≈0.2 mA W −1 .The reduced responsivity could be attributed to several factors, such as the conductivity of the electrolyte, redox species concentration, redox potential, and so on.Nonetheless, the response and recovery time have remained identical compared to using H 2 SO 4 electrolyte, i.e., for the 405 nm light illumination, the device remains ultrafast and the photoresponse is mainly limited by the resolution of the instrumentation; and for the 302 nm light illumination, t res and t rec are ≈25 and 605 ms, respectively.
In the end, we demonstrate that such self-powered, wavelength-distinguishable PEC-PDs can allow a significantly enhanced data transmission security compared to using conventional, wavelength-indistinguishable photodetectors in an UWSN. Figure 6a depicts the concept of an UWSN in ocean environment, which consists of a vast number of nodes working collaboratively to monitor, detect, and track various events and objects in the underwater environment. [4]As will be shown below, integrating the present wavelength distinguishable PEC-PDs in nodes can significantly enhance the data transmission security in the UWSN.To illustrate the enhanced data transmission security, the intended characters "McGill" are transmitted with the setup shown in Figure 6b, to mimic the data transmission in an UWSN.First, random extra characters, such as "y", "n", and "z" are inserted within the word "McGil" at arbitrary positions for encryption purpose, resulting in an altered word "McynGizll".This encrypted word is then transformed into binary code following the extended ASCII scheme, which is further used to drive an arbitrary wave generator (AWG) to generate the corresponding voltage signals.
The voltage signals are in the end sent to a laser controller to modulate the blue laser diode (405 nm) to transmit the encrypted word (i.e., "McynGizll").An On-Off Keying method is used for data transmission with the blue light.
While the encrypted word is transmitted by the blue light, UV light is turned on when the random characters for encryption purpose are being transmitted.As such, for the intended receivers with the present wavelength distinguishable PEC-PDs, they could retrieve the intended word by ignoring signals related to UV light.On the other hand, for eavesdroppers who use wavelength indistinguishable PEC-PDs, as they do not know which signals to drop, they could not retrieve the intended message.
Figure 6c shows the measured photocurrent from the present PEC-PDs when "McynGizll" is transmitted by the blue light (signaled by the negative photocurrent) and the UV light is turned on when extra characters (i.e., "y", "n", and "z") are being transmitted (signaled by the positive photocurrent).As the UV light induces a higher magnitude of photocurrent compared to the blue light, the inserted characters are associated with positive photocurrent.The initial higher negative photocurrent related to "G" and first "l" could be related to the transient when the UV light is off and blue light is on.As such, intended receivers with the present PEC-PDs can retrieve the intended word "McGill" by ignoring signals related to positive photocurrent.On the other hand, eavesdroppers who use wavelength indistinguishable PEC-PDs will not know which signal to drop as only monopolar photocurrent can be measured and thus could not intercept the message.
It is noted, though, even with the present PEC-PDs, one may not decode character "y" due to the interruption caused by UV light; however, one with such PEC-PDs can simply ignore signal related to positive photocurrent and retrieve the intended characters.As such, it provides a simple way to secure the data transmission.It should also be noted that, the example shown here is to illustrate the benefits of wavelength distinguishable PEC-PDs in improving data transmission security compared to their wavelength indistinguishable counterparts; and for advanced applications, it might need to integrate with additional encryption.For example, one may argue that the word "McGill" could still be discernible, due to the inherent sequence of the characters remains.To mitigate this, one can simply add additional encryption prior to inserting random characters, such as Base64 encryption.In this case, "McGill" will be converted to "TWNHaWxs", and after inserting the random characters, the encrypted word will be "TacWNHTaWxs".This will make it difficult for an interceptor to intercept the intended message.Lastly, it is worth mentioning that, although in the present study, the blue light is used to transmit the intended message and the UV light is used for security purpose, it works vice versa.Further combining encryptions (such as inserting random characters and Base64 as mentioned above), it would make the interceptors nearly impossible to inter- from the present wavelength distinguishable PEC-PDs at 0 V, i.e., self-powered.The shaded regions denote the photocurrent measured.The binary bits shown in the bottom are from decoding the photocurrent."0" means no negative photocurrent and "1" means negative photocurrent.The correlated characters are labeled.The period when the UV light is on is also marked.Note that the signal is not decoded when UV light is on (denoted as "-").As can be seen, the intended word "McGill" is successfully decoded from the photocurrent.cept the intended message; and thus, this study provides a simple and efficient way to transmit data highly securely in an UWSN leveraging the present PEC-PDs.

Conclusion
In summary, self-powered, UV, and blue wavelength distinguishable PEC-PDs in both acidic and NaCl electrolytes are achieved in this study, with responsivities reaching mA/W range in both electrolytes.Moreover, an ultrafast response time, with less than 10 ms for the 405 nm blue light, and 20-30 ms for the 302 nm UV light is achieved.This study represents a breakthrough in the development of self-powered, wavelength distinguishable PEC-PDs.As further shown in this study, such devices can enhance data transmission security in an UWSN, due to the wavelength distinguishability induced extra data encryption capability, which could make the UWSN more resilient to both passive and active attacks.Further given the self-powered nature, as well as the al-lowed data transmission using blue light, which is the most transmissive wavelength in the ocean environment, this study significantly advances the development of low energy and maintenance costs and secure UWSNs, especially for the ocean environment.

Experimental Section
Molecular Beam Epitaxy: All the nanowire samples in the present study were grown on 3-inch n-Si (111) substrates using molecular beam epitaxy (MBE) in nitrogen-rich conditions.To achieve p-type and n-type electrical doping, Mg and Si were used, respectively.The estimated n-type and p-type doping concentrations were ≈10 19 cm −3 . [22,25,39,40]The n-GaN nanowire segment was grown at a substrate temperature of 730 °C, with a Ga beam equivalent pressure (BEP) of ≈7.7 × 10 −8 Torr and a nitrogen flow rate of 1.5 sccm.The substrate temperature is estimated by the Si (111) surface reconstruction during the heating process.For the p-InGaN nanowire segment, the substrate temperature was ≈130 °C lower, while the Ga BEP was 2.4 × 10 −8 Torr, and the In BEP was ≈2.2 × 10 −8 Torr.For the n ++ -GaN/InGaN/p ++ -GaN TJ, the substrate temperature was 610 C.

Figure 1 .
Figure 1.a) Schematic of the positive photocurrent generation in a PEC-PD wherein the working electrode is made with n-type material.The chemical reaction closes the circuit, and a representative closed circuit is shown.b) Schematic of the negative photocurrent generation in a PEC-PD wherein the working electrode is made with p-type material.A representative closed circuit is also shown.c) Schematic of the negative photocurrent quenching in a PEC-PD wherein the working electrode is made with p-n heterojunction with the n-down, p-up configuration, due to the presence of the built-in electric field at the junction.Note that the light absorption only occurs in p-region.A representative closed circuit is also shown.The photogenerated electrons, holes, OER, and HER reactions in an acidic solution also are shown.

Figure 2 .
Figure 2. a) SEM image of n-GaN/p-InGaN heterojunction nanowires.Inset: nanowire schematic.b) TEM studies of the heterojunction nanowires, with i-iii) showing the dark field STEM image, the corresponding In signal mapping, and the corresponding Ga signal mapping.The nanowire growth direction is also labeled.c) RT PL spectra of the InGaN and GaN nanowire segments used in this study.d) Schematic of the energy band diagram of the InGaN and GaN nanowire segments used in this study, together with other semiconductor materials.e) Photoresponse of PEC-PDs made with such heterojunction nanowire working electrode, measured in a 2-electrode configuration under 0 V, i.e., self-powered.The excitation density was 1.6 and 7.5 mW cm −2 for 302 and 405 nm, respectively.
), as well as the Ga and In elemental mapping (Figure 3bii-iv), whereas the Ga and In signals along the direction as indicated in Figure 3b-ii are shown in Figure 3c.It is seen that the InGaN layer in the TJ

Figure 3 .
Figure 3. a) Schematic of the TJ InGaN nanowire photoelectrode.Inset: TJ details including n ++ -GaN/InGaN/p ++ -GaN b) Dark field STEM image and the elemental mapping of a TJ InGaN nanowire, with i) showing the dark field image, ii) dark field imaging overlapped with Ga, In, and N signals, iii) Ga signal, and iv) In signal.The nanowire growth direction is also labeled in (i).c) Ga and In signals along the direction as shown by the arrow in Figure 2b-ii.d,e) Photoresponse of PEC-PDs made with the p-doped InGaN TJ nanowire working electrode and the non-doped InGaN TJ nanowire working electrode, respectively, under the 302 and 405 nm light illumination, measured in a 2-electrode configuration.The excitation power density was 1.6 and 7.5 mW cm −2 for 302 and 405 nm, respectively.(f) Schematic of the negative photocurrent generation in the PEC-PDs wherein the working electrode is made with TJ InGaN nanowires, together with the energy band diagram, highlighting the p ++ -GaN/p-InGaN interface.It is noted that replacing p-InGaN with non-doped InGaN would increase the potential barrier.A representative closed circuit is also shown.

Figure 4 .
Figure 4. Photoresponse of PEC-PDs made with the TJ InGaN nanowire working electrode and Pt counter electrode in H 2 SO 4 electrolyte.No potential is applied to the device, i.e., self-powered.a-c) Time-dependent photocurrent density under different light excitation power densities, the extracted photocurrent density and responsivity versus the excitation, and response and recovery time versus the excitation for the 302 nm light illumination, respectively.d,e) similar plots but for the 405 nm light illumination.

Figure 5 .
Figure 5. Photoresponse of PEC-PDs made with the TJ InGaN nanowire working electrode and Pt counter electrode in NaCl electrolyte.During the tests, no potential is applied to the device.a-c) Time-dependent photocurrent density under different light excitation power densities, the extracted photocurrent density and responsivity versus the power density, and the response and recovery time versus the power density under the 302 nm light illumination, respectively.d-f) similar plots but for the 405 nm light illumination.

Figure 6 .
Figure 6.a) Schematic of a representative UWSN with nodes including submarines, ships, and PEC-PDs.b) Schematic of the experiment setup to mimic the data transmission in UWSN.The light signals may come from the submarine, whereas the PEC-PDs can be the nodes.c) Photocurrent measuredfrom the present wavelength distinguishable PEC-PDs at 0 V, i.e., self-powered.The shaded regions denote the photocurrent measured.The binary bits shown in the bottom are from decoding the photocurrent."0" means no negative photocurrent and "1" means negative photocurrent.The correlated characters are labeled.The period when the UV light is on is also marked.Note that the signal is not decoded when UV light is on (denoted as "-").As can be seen, the intended word "McGill" is successfully decoded from the photocurrent.