On the Photo‐Carriers Dynamic Regulation by Piezo‐Phototronic Effect in Metal‐Oxide–Semiconductor Tunneling Junction

The piezo‐phototronic effect can modulate the dynamics of photo‐generated carriers by utilizing external‐strain–induced piezoelectric charges (piezo‐charges). Most of the current researches focus on the modulation effect on the optoelectronic properties of p–n junctions. There are only a few studies focusing on the metal‐oxide–semiconductor (MOS) structures. In this work, a PEDOT:PSS/Al2O3/n‐ZnO tri‐layer MOS tunneling junction is fabricated and the piezo‐phototronic effect on its photo‐carriers dynamic and performances is systematically investigated. The photoresponsivity to 365 nm laser illumination is reduced because of the negative piezo‐charges generated at the Al2O3/n‐ZnO interface. Meanwhile but unexpectedly, the similar phenomenon utilizing the positive piezo‐charges at the Al2O3/n‐ZnO interface is observed. The in‐depth working mechanisms are carefully investigated by analyzing the effect of piezo‐charges on the energy band diagram. Strain induced piezo‐charges could feasibly adjust the intermediate oxide layer's barrier height and width so that the tunneling effect can be efficiently modulated by the piezo‐phototronic effect, leading to the effective control over the MOS tunneling junction's photo‐carriers dynamic and corresponding photoresponses. This work provides an in‐depth understanding for the piezo‐phototronic effect on the MOS tunneling junction and provides guidance for the subsequent researches on the coupling of piezo‐phototronic effect and tunneling effect.


Introduction
The piezo-phototronic effect is a coupling effect among the semiconductor, photoexcitation, and piezoelectric properties. The piezo-phototronic effect can modulate and control the gen-such as dual-channel structure [23] and sandwiched structure. [24] These studies have focused on the regulation of the barrier height. In addition, fast-response piezoelectric tunnel junctions that can simultaneously modulate the barrier height and width, and broadband photodetector that can modulate the tunneling mechanism have been studied by coupling the piezoelectric effect and the tunneling effect. [25][26][27][28] However, these studies have focused on the use of a single piezo-charge to tune the performance of piezoelectric tunnel junctions. The effects of other types of piezo-charges on the tunneling properties of metal-oxide-semiconductor (MOS) structure have not been systematically analyzed. Meanwhile, the published work has shown that the piezotronic effect and the piezo-phototronic effect present different modulation on the tunneling current for different MOS structure. Therefore, the coupling of piezophototronic effect and tunneling effect on device performance still needs to be further investigated. This work will give scientific and technical support for the performance modulation of piezoelectric device, and expand the practical application of quantum tunneling effect in mechanical sensors.
In this work, PEDOT:PSS/Al 2 O 3 /n-ZnO tri-layer MOS structure on rigid ITO/glass substrate and flexible ITO/PET substrate are prepared, and the effect of positive and negative piezocharges on the tunneling characteristics of MOS structures at forward and reverse bias is systematically investigated based on the coupling between piezo-phototronic effect and Fowler-Nordheim tunneling mechanism. Interestingly but unexpectedly, both positive and negative piezo-charges at the interface decrease the photocurrent, photoresponsivity, and photosensitivity at reverse and forward bias. For ITO/glass substrate, the photoresponsivity R is reduced from 0.499 A W −1 under free strain to 0.036 A W −1 under a −0.69% compressive strain at reverse bias and is reduced from 2.56 A W −1 under free strain to 0.30 A W −1 under a −0.69% compressive strain at forward bias when the power density is 30 µW cm −2 . For ITO/PET substrate, the photoresponsivity R of reverse bias and forward bias are reduced by 61.1% and 65.8%, respectively, at a tensile strain of +0.69% under the power densities of 150 µW cm −2 . These cases further demonstrate the influence of piezo-phototronic effect and tunneling effect on the photo-carriers dynamic as well as the photoresponses of the MOS structure. The fundamental working mechanism of how the piezo-phototronic effect influences the Fowler-Nordheim tunneling effect is carefully and systematically investigated by analyzing their energy band diagrams under compressive and tensile strain at reverse and forward bias. In addition, we propose a new structure that can enhance the motion of photo-carriers by analyzing the energy band diagram. This work provides an in-depth understanding for the piezo-phototronic effect on the MOS tunneling junction. It also guides the coupling between piezo-phototronic effect and tunneling effect in preparation of potential novel optoelectronic devices. Figure 1a illustrates the 3D structure of PEDOT:PSS/Al 2 O 3 /n-ZnO tri-layer MOS structure based on ITO/glass substrate.

Device Fabrication and Structure of PEDOT:PSS/Al 2 O 3 /n-ZnO MOS Structure on ITO/Glass Substrate
First, a 100 nm thick Al:ZnO (AZO) seed layer is deposited on top of the ITO/glass substrate. Second, ZnO nanowire (NW) arrays with an average diameter of 70-100 nm and a length of 2 µm are grown vertically on the surface of the ITO/glass substrate by a low-temperature hydrothermal method. [29][30][31] Third, the Al 2 O 3 film with a thickness of 10 nm is deposited as the oxide layer on the surface of the ZnO NW arrays by atomic layer deposition (ALD). Finally, highly conductive PEDOT:PSS solution is spin-coated as the metal layer on top of the Al 2 O 3 film. Meanwhile, the PEDOT:PSS film locates at the top surface as the top metal contact electrode and the ITO film on the glass substrate serves as the bottom metal contact electrode on the semiconductor side. The above as-described fabrication processes of the MOS structure are shown in Figure 1b. The detailed fabrication processes and measurement could be found in the Experimental Section and Supporting Information. The scanning electron microscope (SEM) image of the PEDOT:PSS/ Al 2 O 3 /n-ZnO MOS structure is seen in Figure 1c1,c2. The cross section clearly shows the device structure and dense ZnO NW arrays. The continuous PEDOT:PSS film on the surface indicates that the PEDOT:PSS film and the ZnO NW arrays form a close contact. Empty between ZnO NWs filled with PEDOT:PSS are also shown in Figure 1c2.
Prior to the piezo-phototronic modification of I-V characteristics, we first study the electrical transport of the PEDOT:PSS/ Al 2 O 3 /n-ZnO MOS structure on ITO/glass without strain. Figure 1d illustrates the I-V characteristics of the MOS structure under dark condition and 365 nm laser illumination with different power densities ranging from 30 µW cm −2 to 3 mW cm −2 . As the illumination power densities increases, the current of the MOS structure under forward and reverse bias also increases. This is mainly because ZnO NW arrays absorbs most of the UV light and generates a large number of photogenerated electrons and holes, that is, the photo-carriers. Photogenerated carriers move under an electric field and contribute to the current. It is worth noting that the ideal Al 2 O 3 as an insulating layer is non-conductive. However, when the oxide layer is very thin, the quantum tunneling effect provides a pathway for carrier transport. Therefore, the photogenerated electrons generated on the side of ZnO pass through the oxide layer by tunneling and are absorbed by the electrode on the side of PEDOT:PSS when 0.5 V is applied to the side of PEDOT:PSS. The photogenerated holes are absorbed by the grounded ITO electrode. Under forward bias of 0.5 V, the current rises by 2692% from 8.38 µA in the dark field to 233.97 µA at power densities of 3 mW cm −2 . When −0.5 V is applied to the side of PEDOT:PSS, the photogenerated holes generated on the side of ZnO pass through the oxide layer by tunneling and be absorbed by the grounded electrode on the side of PEDOT:PSS. The photogenerated electrons are absorbed by the ITO electrode. Under reverse bias of −0.5 V, the current rises by 571% from 3.51 µA in the dark field to 23.55 µA at power densities of 3 mW cm −2 . This is due to the superposition of photogenerated and dark currents. The greater power densities generate more photogenerated carriers and therefore larger currents. The experimental results demonstrate the absorption of 365 nm UV light by ZnO NW and the tunneling effect that the MOS structure has outstanding photoresponsive properties. [25,32] www.advmatinterfaces.de

Piezo-Phototronic Effect on Performances of PEDOT:PSS/ Al 2 O 3 /n-ZnO MOS Structure on ITO/Glass Substrate
The piezo-phototronic effect is then further introduced to modulate the performance of PEDOT:PSS/Al 2 O 3 /n-ZnO MOS structure. As schematically illustrated in the Supporting Information ( Figure S7, Supporting Information), external compressive strains are applied onto the PEDOT:PSS/Al 2 O 3 /n-ZnO MOS structure by pressing the back of the devices through a piece of glass. The applied compressive strain along the caxis of ZnO NW has an effect on the photo-carriers transport and the potential barrier at the interface, which is due to the piezoelectric effect in ZnO NW. [33] A 365 nm laser with six different power densities varying from 30 µW cm −2 to 3 mW cm −2 is then introduced and illuminated onto the surface of ZnO NW arrays under different externally applied compressive strains varying from 0.00% (free state) to −0.69% to illustrate the piezo-phototronic effect at both reverse and forward biases. The photocurrent (I ph = I light − I dark ) as the function of the compressive strains is calculated and plotted in Figures 2a,b, respectively. In Figure 2a, at reverse bias, the photocurrent is reduced by 92.7% from 7.52 µA under free strain to 0.55 µA under −0.69% compressive strain when the power density is 30 µW cm −2 . The photocurrents are also reduced by 90.9%, 84.7%, 80.0%, 74.5%, and 71.5% at a −0.69% compression strain for power densities of 200, 500, 1000, 2000, and 3000 µW cm −2 , respectively. Unexpectedly, as illustrated in Figure 2b, at forward bias, the photocurrent is still reduced by 88.3% from 38.6 µA under free strain to 4.5 µA under −0.69% compressive strain when the power density is 30 µW cm −2 . Similarly, the photocurrents are reduced by 83.9%, 79.5%, 70.5%, 61.2%, and 54.9% at a −0.69% compression strain for power densities of 200, 500, 1000, 2000, and 3000 µW cm −2 , respectively. They have the same downward trend, and the decrease is smaller as the power density increases. This is because much photogenerated carriers at high power density will shield the piezo-charges at the interface, thereby reducing the modulation of the piezo-phototronic effect. These experimental results demonstrate the modulation of photo-carriers dynamic as well as the photoresponse of MOS tunneling junction by the piezophototronic effect. [32,34,35] The photoresponsivity R as a crucial parameter to describe the photocurrent changes at each power density and externally applied compressive strain is also calculated and plotted in Figures 2c,d, respectively. The photoresponsivity R is defined as R = I ph,ε /P inc , where I ph,ε is the photocurrent under compressive strain and P inc is the optical power of the incident laser illumination. In Figure 2c,d, it is obvious that the photoresponsivity R decreases under both reverse and forward bias as the power density increases and the external compression strain increases. In Figure 2c, at reverse bias, the photoresponsivity R is reduced from 0.499 A W −1 under free strain to 0.036 A W −1 under −0.69% compressive strain when the power density is 30 µW cm −2 . In Figure 2d, at forward bias, the photoresponsivity

www.advmatinterfaces.de
R is reduced from 2.56 A W −1 under free strain to 0.30 A W −1 under −0.69% compressive strain when the power density is 30 µW cm −2 . The higher the power density, the smaller the change in the photoresponsivity R. These results also indicate that the modulation of the piezo-phototronic effect of ZnO NW on the photogenerated carriers dynamic starts to weaken at larger power densities. To illustrate the piezo-phototronic effect on the photoresponsivity R more quantitatively, the relative change of photoresponsivity ΔR/R 0 (ΔR = R ε − R 0 , R ε is the R under a certain compressive strain ε, and R 0 is the R under strain-free condition) under each compressive strain and optical power density is calculated and summarized in the Supporting Information ( Figure S1, Supporting Information). It is clear that photoresponsivity R is almost linearly related to the compressive strain at each light intensity when the compressive strain is greater than −0.17% for both positive and negative bias. The results show that the photoresponsivity R of reverse bias and forward bias are reduced by 92.6% and 88.2%, respectively, at compression strain of −0.69% under the power densities of 30 µW cm −2 .
In addition, another important parameter, the sensitivity of the MOS structure is also derived, calculated, and plotted in Figure 2e,f at each power intensity and different compression strains. Sensitivity is defined as (I light − I dark )/I dark (I light is the device current under illumination and I dark is the device current under dark field). It is obvious that the sensitivity decreases with the increase of the applied compressive strain and the decrease of the power density. Specifically, in Figure 2e, at reverse bias, the sensitivity is reduced by 98.1% from 2.14 under free strain to 0.04 under −0.69% compressive strain when the power density is 30 µW cm −2 . While in Figure 2f, at forward bias, the sensitivity is reduced by 98.6% from 4.16 under free strain to 0.06 under −0.69% compressive strain when the power density is 30 µW cm −2 . When the power density is increased to 3 mW cm −2 , they are only reduced by 93.5% and 95.0% under reverse bias and forward bias conditions, respectively, showing a little decrease in the reduction amplitude due to the externally applied compressive strain.
To give a deeper and farsighted understanding of the piezophototronic effect on the photo-carriers dynamic in the MOS structure, we have modified the ALD parameters to obtain oxide layers with different thicknesses, ranging from 3 to 15 nm. The I-V characteristics and photoresponse of the MOS structure with different Al 2 O 3 thicknesses are also investigated under different compressive strains and power densities. The results are presented in the Supporting Information. In Figure S2, Supporting Information, the I-V characteristics of MOS structures with different Al 2 O 3 thicknesses are summarized and plotted. As shown in Figure S2a Figure S2c, Supporting Information, and Figure 2a,b, although these two devices all have the similar modulation trend at different power densities and compressive strains, the device with 15 nm thick oxide layer has a poor resolution for illumination power density below 500 µW cm −2 . Therefore, 10 nm thick oxide layer as the optimum choice is used to study the piezo-phototronic effect of MOS structure because of its higher resolution for UV light illumination.
The fundamental working mechanism of how the piezophototronic effect influences the Fowler-Nordheim tunneling effect is carefully and systematically investigated by analyzing their different band diagrams under compressive strain at forward and reverse bias, as illustrated in   www.advmatinterfaces.de raises the interface barrier. Both the barrier height and the barrier width of photogenerated electrons tunneling on the side of ZnO increase and meanwhile the negative piezo-charge would repel the photogenerated electrons at forward bias. So that the total number of photogenerated electrons that can pass through the barrier decreases. In addition, the Coulomb attraction of negative piezo-charges to photogenerated holes and the hole potential well formed at the interface of Al 2 O 3 and ZnO will also restrict the movement of photogenerated holes to the cathode. However, although the transport of both photogenerated electrons and photogenerated holes is weakened, the tunneling conduction mechanism of photogenerated electrons is still the dominant conduction mechanism. Therefore, the Coulomb repulsion of electrons by negative piezo-charges and the increased barrier width and height weaken the photoresponse.
At reverse bias, the photogenerated holes on the side of ZnO need to pass through the oxide layer to the side of PEDOT:PSS and are collected by PEDOT:PSS, while the photogenerated electrons move along the direction of the electric field to the interface between ZnO and the bottom ITO electrode, and are collected by ITO. Although the barrier height and barrier width of photogenerated hole tunneling on the side of ZnO are reduced, a deeper hole potential well is formed at the interface of Al 2 O 3 and ZnO due to the upward bending of the energy band by negative piezo-charges. As a result, the photogenerated holes must first escape from the hole potential well before tunneling into the side of PEDOT:PSS. The movement of holes in the direction of the electric field is therefore restricted and the holes attracted by the negative piezo-charges are also accumulated in the hole potential well. As a result, some of the photogenerated holes contribute to photoresponse through Fowler-Nordheim tunneling while the other part of the holes moves along the interface of Al 2 O 3 and ZnO, which is similar to the 2D electron gas (2DEG) formed at the interface of the heterojunction. [36,37] Consequently, the total number of tunneling holes is significantly reduced, and thus the photoresponse is reduced. As the compressive strain increases, there are more negative piezo-charges at the interface of Al 2 O 3 and ZnO. As a result, the hole potential well becomes deeper and the photoresponse continues to decline. In addition, the Coulomb repulsion of photogenerated electrons by negative piezo-charges has little effect on the photocurrent because it is the tunneling transport of photogenerated holes dominantly determines the photocurrent. Therefore, the deeper hole potential well formed by the negative piezo-charges and the Coulomb attraction to the holes are the main mechanisms for weakening the photoresponse. In a brief summary, the negative piezo-charges induced by compressive strain significantly modulate the energy band of ZnO and also barrier of Fowler-Nordheim tunneling effect, which in turn have a significant impact on their photo-carriers dynamic and corresponding photoresponse performances.

Device Fabrication and Piezo-Phototronic Effect on Performances of PEDOT:PSS/Al 2 O 3 /n-ZnO MOS Structure on Flexible ITO/PET Substrate
Only compressive strain can be applied to the ITO/glass substrate, not tensile strain. In order to systematically investigate the modulation of tensile and compressive strain on the tunneling effect, PEDOT:PSS/Al 2 O 3 /n-ZnO MOS structure based on flexible ITO/PET substrate is fabricated. The device structure and fabrication processes are shown in Figure 4a,b. In Figures S9 and S10, Supporting Information, the 2D structure shows how strain is applied to ZnO through the flexible substrate. The fabrication processes are identical to the ITO/glass substrate. In Figure 4a, one end of the ITO/PET substrate is fixed. Tensile and compressive strain can be applied to the ZnO NW by moving the metal rod clamped to the other end of the ITO/PET. In Figure 4c, the compressive strain applied to ITO/ PET is a tensile strain for ZnO NW. The positive piezo-charge is generated at the Al 2 O 3 /ZnO interface. In contrast, the tensile strain applied to ITO/PET is a compressive strain on ZnO NW. The negative piezo-charge is produced at the Al 2 O 3 /ZnO interface.
As shown in Figure 4d, the optoelectronic properties of the PEDOT:PSS/Al 2 O 3 /n-ZnO MOS structure on ITO/PET substrate without strains are first investigated. The current rises by 402% from 1.33 µA in the dark field to 6.67 µA under power densities of 20 mW cm −2 at reverse bias and the current rises by 183% from 6.62 µA in the dark field to 18.74 µA under power densities of 20 mW cm −2 at forward bias. The modulation of I-V characteristics by the photoelectric effect is the same as for ITO/glass substrate in the absence of strain.
Then the piezo-phototronic effect is further introduced to modulate the performance of PEDOT:PSS/Al 2 O 3 /n-ZnO MOS structure on flexible ITO/PET substrate, as shown in Figure 5.
The result indicates that all the variations in photocurrent, photoresponsivity, and sensitivity of the flexible ITO/PET substrate under compressive strains are almost the same as that observed in the device fabricated on rigid ITO/glass substrate. These results further confirm and validate the experimental results observed for devices on ITO/glass substrate. Therefore, we only focus on analyzing the piezo-phototronic effect on the performances of PEDOT:PSS/Al 2 O 3 /n-ZnO MOS structure when tensile strain is applied to ZnO NW.
The photocurrent at reverse and forward bias under each tensile strain and illumination power density are concluded and plotted in Figure 5a,b. In Figure 5a, at reverse bias, the photocurrent is reduced by 61.1% from 0.90 µA under free state to 0.35 µA under +0.69% tensile strain when the power density is 150 µW cm −2 . Meanwhile in Figure 5b, at forward bias, the photocurrent is also reduced by 65.8% from 2.78 µA under free strain to 0.95 µA under +0.69% tensile strain when the power density is 150 µW cm −2 . The photoresponsivity R at reverse and forward bias under each tensile strain and power density are shown in Figure 5c,d. Similarly, in Figure 5c, at reverse bias, the photoresponsivity R is reduced from 0.024 A W −1 under free strain to 0.009 A W −1 under +0.69% tensile strain when the power density is 150 µW cm −2 . And in Figure 5d, at forward bias, the photoresponsivity R is also reduced from 0.074 A W −1 under free strain to 0.025 A W −1 under +0.69% tensile strain when the power density is 150 µW cm −2 . The relative change of photoresponsivity ΔR/R 0 under each tensile strain and optical power density is calculated and plotted in the Supporting Information ( Figure S4, Supporting Information). The results show that photoresponsivity R of reverse bias and forward bias are reduced by 61.1% and 65.8%, respectively, at tensile strain of www.advmatinterfaces.de +0.69% under the power densities of 150 µW cm −2 . The linearity of the photoresponsivity with respect to the compressive and tensile strains fluctuates when the compressive and tensile strains are greater than 0.17%. This should be attributed to the indirect application of compressive and tensile strains to the ZnO NW by directly bending the flexible ITO/PET substrate. However, the overall trend is not affected. The sensitivity at reverse and forward bias under each tensile strain and power density are shown in Figure 5e,f. At reverse bias, the sensitivity is reduced by 58.8% from 0.68 under free state to 0.28 under +0.69% tensile strain when the power density is 150 µW cm −2 . While at forward bias, the sensitivity is reduced by 71.4% from 0.42 under free strain to 0.12 under +0.69% tensile strain when the power density is 150 µW cm −2 . When the power density is increased to 3 mW cm −2 , they were only reduced by 38.0% and 36.6% under reverse bias and positive bias conditions, respectively. The results also indicate that the piezo-phototronic effect originated from the strain induced piezo-charges is reduced due to the charge screening effect by the photogenerated carriers.
The fundamental working mechanisms of how the piezophototronic effect influences the Fowler-Nordheim tunneling effect under tensile strain is carefully and systematically investigated by analyzing different band diagrams at forward and reverse bias under tensile strain, as shown in Figure 6. The positive piezo-charges are generated at the interface of Al 2 O 3 and ZnO when the tensile strain is applied along the c-axis of ZnO. The positive piezo-charge lowers the interface barrier. As a result, as shown in Figure 6a, at forward bias, although the barrier height and width of photogenerated electron tunneling on the side of ZnO are reduced, the deeper electron potential well is formed at the interface of Al 2 O 3 and ZnO due to the downward bending of the energy band, which hinders the electron collection through tunneling. Further, the electrons attracted by the positive piezo-charges are also accumulated in this electron potential well. Consequently, the photogenerated electrons must first escape from the electron potential well before tunneling into the side of PEDOT:PSS. The movement of electrons in the direction of the electric field is thus strongly restricted. Therefore, only a part of the photogenerated electrons contribute to the photoresponse through Fowler-Nordheim tunneling while the other part of the electrons would move along the interface of Al 2 O 3 and ZnO as similar to 2DEG. In addition, the Coulomb repulsion of photogenerated holes by positive piezo-charge has little effect on the photocurrent because it is the tunneling transport of photogenerated electrons dominantly determines the photocurrent. The weakened tunneling transport of photogenerated electrons reduces the photocurrent. Therefore, the deeper electron potential well formed by positive piezo-charges and the Coulomb attraction for photogenerated electrons weaken the photoresponse. Here the flexible ITO/PET is used as the substrate. Other manufacturing processes are the same as ITO/glass substrate. The blue material represents the metal frame. The bottom of the PET is fixed, and by moving the metal rod at the top, pressure and tension can be applied to the PET substrate to deform the ZnO NWs. c) Device structure when pressure and tension are applied to a PET substrate. When pressure is applied to PET, which is equivalent to applying tension along the c-axis of the ZnO NWs, positive piezo-charges are generated at the Al 2 O 3 /n-ZnO interface. When tension is applied to PET, which is equivalent to applying pressure along the c-axis of ZnO NWs, negative piezo-charges are generated at the Al 2 O 3 /n-ZnO interface. d) The output current-voltage (I-V) characteristics of 365 nm UV illumination under different power densities. With the increase of power densities, both the forward output current and the reverse output current increase.

www.advmatinterfaces.de
As shown in Figure 6b, at reverse bias, despite the hole potential well is reduced, both the barrier height and width of the hole tunneling increase due to the downward bending of the barrier and the positive piezo-charges also repel the hole away from the interface. Therefore, the total number of holes tunneling through the interface decreases. At the same time, the electronic potential well formed by positive piezo-charges and the Coulomb attraction to photogenerated electrons also weakens the transport of photogenerated electrons. In a brief summary, although the transport of photogenerated electrons and photogenerated holes is weakened, the tunneling conduction mechanism of photogenerated holes is still the most important conduction mechanism. Therefore, the Coulomb repulsion due to the positive piezo-charges and the increased tunneling barrier act together to reduce the photo-carriers dynamic and hence the photoresponse.
However, how to enhance the movement of photo-carriers? By further analyzing the energy band diagrams in these four cases, we found that an important reason for the constraint on the increase of photocurrent is the formation of carrier potential wells on the ZnO side when the tunneling barrier height and width are reduced, which limits the movement of photocarriers in the electric field direction. For example, in Figure 6a, when the positive piezo-charge is generated at the interface between Al 2 O 3 and ZnO under positive bias, both the barrier height and the barrier width of electron tunneling are reduced. However, the positive piezo-charge also causes the energy band to bend downward to form the deeper electron potential well, which reduces the number of electrons contributing to the photocurrent. In Figure 3b, though the negative piezo-charge reduces the hole tunneling barrier height and width at negative bias, it also bends the energy band upward to form the deeper hole potential well, which limits the number of holes that can contribute to the photocurrent. Because photo-carriers and piezo-charges are generated on the same side, it should be not possible to enhance photo-carriers transport while weakening the potential well by modulating the energy band.
Considering the above constraints, by carefully analyzing the energy band diagram, we could enhance the photoresponse by achieving spatial separation of piezo-charges and photocarriers, that is, utilizing the piezoelectric material to lower the interfacial barrier and the semiconductor material to generate photo-carriers. As illustrated in Figure 7a, under positive bias, the height and width of the hole tunneling potential barrier are reduced by applying a suitable strain to the n-type piezoelectric material when negative piezo-charges are generated at the interface. Meanwhile, a p-type semiconductor material acts as a light-sensitive material to produce photo-carriers. This achieves the spatial separation of the piezo-charges and the photo-carriers. As a result, the photo-carriers generated in the semiconductor side can contribute to photocurrent by tunneling to the electrode without being confined by the hole potential well produced at the interface between n-type piezoelectric semiconductor and insulating layer. Similarly, as illustrated in Figure 7b, the positive piezo-charges at the interface reduces the height and width of the electron-tunneling barrier under negative bias.  Figure S4, Supporting Information. e,f) The reverse and forward sensitivity as a function of compressive and tensile strains under different power densities. The wavelength of UV illumination is 365 nm, the forward bias voltage is 0.5 V and the reverse bias voltage is −0.5 V.

www.advmatinterfaces.de
The photogenerated electrons generated in the semiconductor can tunnel to the electrode to generate photocurrent. The improved scheme is expected to overcome the limitation of the piezo-charges on the motion of photo-carriers dynamic along the electric field direction while reducing the tunneling barrier. Moreover, the tunneling photocurrent can be augmented by enhancing the bias voltage, thereby strengthening the electric field in the potential barrier region, since the electric field profoundly impacts the width of the triangular potential barrier.

Conclusion
In conclusion, the PEDOT:PSS/Al 2 O 3 /n-ZnO MOS structure on rigid ITO/glass and flexible ITO/PET substrates are fabricated. We have systematically investigated the piezo-phototronic effect on photo-carriers dynamic and corresponding photoresponse performances of the MOS structure to 365 nm laser illumination. The experimental results show a unique phenomenon that both compressive strain and tensile strain would reduce the photo-carriers dynamic as well as the photoresponse of the MOS structure at forward and reverse bias. The fundamental working mechanism of the piezo-phototronic effect on the photo-carriers dynamic of the MOS structure is carefully investigated by comparing the variations on energy band diagrams originated from the positive and the negative piezo-charges at the Al 2 O 3 /n-ZnO interface under reverse and forward bias, respectively. This work mainly analyzes the following four cases. 1) Under compressive strain and forward bias, the increased electron tunneling barrier height as well as width and the repulsion of negative piezo-charges to electrons work together to reduce the photoresponse of the MOS structure.
2) Under compressive strain and reverse bias, the deeper hole potential well located near the interface limits the movement of holes in the direction of the electric field and also the resulting hole accumulation further reduces the photoresponse. 3) Under tensile strain and forward bias, the deeper electron potential well produced near the interface restricts the movement of electrons in the direction of the electric field and still the accumulation of electrons further reduces the photoresponse. 4) Under tensile strain and reverse bias, the photoresponse decreases due to the elevated hole tunneling barrier and the repulsion of holes by positive piezo-charges. In addition, by analyzing the energy band diagram, the new structures that can enhance the   www.advmatinterfaces.de photoresponse are proposed. This work not only promotes the in-depth understanding in the device physics of the coupling between piezo-phototronic effect and tunneling effect, but also provides a reference for the research of piezo-phototronic effect in MOS tunneling junction based photodetectors.

Experimental Section
Device Fabrication Process: For PEDOT:PSS/Al 2 O 3 /n-ZnO MOS structure on ITO/glass. First, ITO/glass with a size of 1 cm × 1 cm was cleaned with acetone, anhydrous ethanol, and deionized water each for 5 min, and then Kapton tape was used to stick one half as the bottom electrode and the other half as the active device area. The AZO seed layer was deposited on top of ITO/glass by radio frequency magnetron sputtering at the power of 100 W and chamber pressure of 1.0 Pa for 17 min, with the thickness of 100 nm. The AZO seed layer coated ITO/ glass was then placed into the mixed nutrient solution (50 mm Zinc Nitrate and 25 mm Hexamethylenetetramine) for ZnO NW arrays growth via a low-temperature hydrothermal method in a mechanical convection oven at 80 °C for 3 h. After cooling down the whole system, the sample was washed by distilled water and dried by high purity N 2 gas. Second, the Al 2 O 3 layer of 10 nm thickness was coated on the samples in ALD chamber. Finally, a thin layer of highly conductive PEDOT:PSS (Sigma-Aldrich) was spin-coated on the Al 2 O 3 layer as the top electrode at 3000 rpm, followed by heating at 120 °C for 5 min, and then the tape on the surface of the ITO/glass was torn off and the copper wires were bonded to the silver electrodes to lead the bottom and top electrode. ITO/PET based device fabrication is the same as ITO/glass. The only difference is that the fabricated device is packaged with PDMS after the lead ends.
Characterization and Measurement: Detailed microscopic structures of the MOS structure were characterized by FE-SEM (GeminiSEM 500). The electric signals of the devices were measured and recorded by a Dual-Channel System Source Meter Unit instrument (Agilent B2902A). The optical input stimuli were provided by a semiconductor laser (wavelength of 365 nm, CLASS IIIb LASER PRODUCT). A continuously variable filter (OMMB-NDFC50, Zolix) was used to control the optical power.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.  Figure 7. a) Band diagram of the proposed structure when negative piezo-charges are generated at the interface between the insulating layer and the piezoelectric material at forward bias. b) Band diagram of the proposed structure when positive piezo-charges are generated at the interface between the insulating layer and the piezoelectric material at reverse bias.