Effect of Light and Electrostatic Gate at Oxide Interface LaFeO3–SrTiO3 at Room Temperature

Electrostatic gating and light illumination are two widely used stimuli for semiconductor devices. Two‐dimensional electron gas (2DEG) at oxide heterostructures has shown potential in optoelectronics due to its high optical response and gate tuning property. The appearance of high photoconductivity and persistent photoconductivity at the oxide heterostructure of two insulators LaFeO3 and SrTiO3 is shown here. The photoconductivity has been further tuned using positive and negative back gating. A large change in conductivity has been achieved under the simultaneous application of light and electrostatic gating. A few measurement protocols that manifest possible applications of this interface as memory and switching devices are implemented.


Introduction
Transition metal oxides (TMOs) are potential candidates for newgeneration electronic devices due to their wide range of functionalities from metal-insulator transition to solar photocatalytic water splitting. [1,2] They are more stable than semiconductors in an ambient atmosphere. The discovery of high mobile electron gas at the interface of two wide band gap insulators LaAlO 3 (5.2 eV) (LAO) and SrTiO 3 (3.2 eV) (STO) in 2004 brought new possibilities for TMOs. [3] The LAO/STO interface became important not only in the field of fundamental physics but also for device applications. STO-based interfaces exhibit emergent phenomena like photoconductivity, superconductivity, magnetism, [4] the coexistence of superconductivity and magnetism, [5] Shubnikov de-Hass oscillations, [6] tunable spin-orbit coupling, [7] and so on. It DOI: 10.1002/apxr.202200087 has been observed that these functionalities can be tuned using external stimuli such as electric field, light, magnetic field, heat, and mechanical stress. With this tuning property, these oxide interfaces open avenues for device applications in different areas. The LAO/STO interface has already been explored for devices like field-effective transistors (FET), nanoscale photo-detector, THz generation, and nonvolatile memory.
Photoconductivity is an optoelectronic property concentrated on the enhancement (decrement) of the electrical conductivity (resistance) of material due to the absorption of electromagnetic radiation. The phenomenon of enhancing electrical conductivity as a function of light is called photoconductivity. Photoconductivity is mainly divided into two categories: a) persistent photoconductivity (PPC) and b) transient photoconductivity (TPC). The PPC effect is an electrical conductivity enhancement phenomenon where the electrical conductivity of material increases when it absorbs electromagnetic radiation and the enhancement in conductivity remains for a considerably long time even after the removal of the light source. [8] On the other hand in TPC, the material regains its initial conductivity as soon as the light source is removed. The PPC effect has been already observed for many semiconductor heterostructures and quantum dots, for example, AlGaN/GaN, [9] and graphene/GaSe nanosheet, [10] but due to low operational temperature, their applications are limited. Unlike conventional semiconductors, oxide heterostructures like LaTiO 3 -STO, [11] LaVO 3 -STO, [12] EuO-KTaO 3 , [13] La 0.3 Sr 0.7 Al 0.65 Ta 0.35 O 3 -STO, [14] and LaVO 3 -KTaO 3 [15] have shown high PPC even at room temperature which opens up new opportunities in optoelectronic application such as storage device, switching device, transistor, solar cell, and many more.
In this work, we report both light and electrostatic gate tuning effects of electrical resistivity for two-dimensional electron gas (2DEG) formed at LFO-STO heterostructure at room temperature. The LFO-STO interface shows the highest photoconductivity and PPC effect at room temperature till reported for other heterostructures. We have shown separate positive and negative gate tuning effects at the interface first, then the combined effect has been shown for the same. The gating enhanced the light effect at the heterostructure. We have also shown the switching effect by the combined effect of light and gate for the LFO-STO heterostructure.

Results and Discussion
To study the light and electrostatic gate tuning properties, we have grown LFO thin film of different thicknesses on top of a TiO 2 terminated STO (001) single crystal. The details of substrate preparation and sample fabrication are given in the Experimental Section. The interface became conducting after the critical thickness of 4 monolayers (ML) (i.e., 4 unit cells of LFO). The bandgap scenario of LaFeO 3 -SrTiO 3 (LFO-STO) interface is different from LaAlO 3 -SrTiO 3 interface. The band gap of LAO (≈5.6 eV) is larger than STO (≈3.2 eV) whereas the bandgap of LFO (2.2 eV) is smaller than STO. The band gap of LFO is rather similar to LaVO 3 (≈1.1 eV). In this respect, the formation of 2DEG at the interface of LFO-STO can most likely be explained using a similar mechanism as that of LVO-STO, where the interface becomes conducting due to the polar discontinuity doping process. [16] In addition, it is worth mentioning that it has recently been reported that the conducting interfaces of LFO-STO might also be related to "dynamic layer rearrangement", [17] oxygen off-stoichiometry or cation intermixing. [18] Figure 1a shows the sheet resistance as a function of temperature. The resistance decreases as the temperature decreases confirming its metallic nature. The inset of Figure 1a shows the measurement schematic of the LFO-STO interface. The upturn in the resistance might be attributed to many possible reasons such as impurities/defects present in the system, the Kondo effect, and weak localization. We have done Kondo fitting and appropriate fitting could not be achieved, in addition, the signature of weak localization in magnetoresistance has not been observed. The resistance decreases with an increase in the thickness of the LFO film as shown in Figure 1a. This behavior is in line with the polar catastrophe scenario: 4 monolayers of LFO is the critical thickness to make the interface conducting above which the diverging potential due to polar discontinuity is enough to create an electronic reconstruction that makes the interface conducting, below that the interface is insulating. Once the interface became conducting there is no significant change (Figure 1a); 6 and 10 mL are very similar) in the interface resistance as a function of the film thickness. Hall measurement is done to calculate the carrier density of the system. The carrier density calculated from Hall measurement is 4.42× 10 12 , 1.63× 10 13 , and 9.65× 10 13 cm −2 for 4 , 6 , and 10 mL, respectively. The I-V characteristics have been shown in Figure 1b   sponse of percentage change in resistance is recorded in dark and light illumination at room temperature. The back gate is applied to see the effect of electrostatic field tuning. All the lightgate experiment performed in this paper was done on a 4 mL sample.

Light Illumination Effect
For light illumination response, we have taken three different wavelength laser sources having wavelengths of 630 (red), 532 (green), and 405 nm (blue). The percentage change in resistance is shown as a function of time in Figure 2a for different wavelength values (405 (blue), 532 (green), and 630 nm (red)) at a constant power of 70 mW. The percentage change in resistance is calculated as, where R(t) is the resistance as a function of time and R(0) is the resistance in dark. The dark resistance is constant for all systems. In Figure 2a, the change in resistance is measured as follows: first, the sample is kept in dark for two days to remove any light effect, and then, the resistance is measured for 5 min. After that, the light is illuminated for 5 min on the sample and the resistance is measured during the illumination, the light is removed and again the resistance is measured and the process is repeated. The power of the laser light was kept constant for all measurements which are at 70 mW. From Figure 2a, we can see that the resistance decreases on light illumination for all wavelengths. The decrease in resistance signifies the generation of excited electrons on light illumination. The decrease in resistance increases with lowering the wavelength value (or increasing the energy of light). The R% is ≈ 30%, ≈ 55%, and ≈ 80% for 630 (red laser), 532 (green laser), and 405 nm (blue laser), respectively. Similar measurement has been carried out with bare STO as well as on the STO after the same etching treatment. In both cases, the resistance was beyond the measurement limit of the device, and no photoconductivity is detected; hence we can say that the photoconductivity and the PPC are coming from the interface only, not from the off-stoichiometry arising from the etching treatment. [19] The R% change in red and green laser light is small in comparison to the blue laser. The energy of the red laser and green laser is 1.96 and 2.3 eV, respectively. This energy is comparable with the band gap of LFO, which is ≈2.3 eV. This indicates that using red and green lasers we are exciting only the valence electrons of LFO to the quantum well formed at the interface of the LFO/STO heterostructure. But in the case of blue light having energy ≈ 3.06 eV (greater than both red and green laser), we are exciting the valance edge electrons of STO (band gap ≈3.2 eV) as well with the LFO giving more carrier density at the interface resulting in a larger change in resistance of the system. The interesting part here is that in all three cases, the system does not come to its initial resistance value after switching off the laser light. This is a clear signature of PPC in our system. The effect remains in the system even after 40 h after switching off the laser light as shown in the inset of Figure 2b. This change in resistance to light illumination at room temperature is the highest value at room temperature till reported as per our knowledge. The R% for blue light as a function of time is shown in Figure 2b for 900 s. To understand the physics better, we have done laser powerdependent measurements. The conclusive results are shown in Figure 2c,d, and the change in resistance and PPC are shown as a function of laser power. The curve comprises two parts: the first part is up to 10 mW where the increase in R% and PPC is very fast and the second part start from 10 to 70 mW, where the increase in R% and PPC is slower but still monotonic increase is observed for both. This monotonic increase of R% (or increase in photocurrent on increasing the laser power) is previously reported for many systems like InGaAs quantum dots, [20] Bi 2 Se 3 , [21] Graphene/GaSe, [10] and so on. The laser power dependency is explained as I ph = P , where I ph is photocurrent corresponding to laser power P. is the exponent, when = 1, the effect is purely photoconduction but for < 1 or > 1 the response is not purely photoconductive. When light is illuminated on semiconducting materials where the energy of the light is higher than the band gap of the material, then electrons are promoted to the conduction band from the valency band giving rise to a reduction of the resistance. On the other hand, when light is illuminated on a semiconductor junction, where an internal electric field is already present and the electron-hole separation happens through a diffusive process is known as the photovoltaic effect. In the LFO-STO interface, the presence of an internal electric field is already reported. [22] It is also evident from Figure 2c that for lower laser intensity is equal to one and hence the process is dominated by the photoelectric effect but for higher laser intensity is less than one and hence dominated by the photovoltaic effect. It is worth mentioning that the actual process could be far more complex.

Electrostatic Gating Effect
To study the electrostatic gating effect, a back gate is applied to the interface from the substrate side and the change in resistance is measured as a function of time. Both negative and positive gate is applied to the interface without light illumination. The protocol for measurement is similar to light illumination. The leakage current is as small as ≈ 10 pA, suggesting the system is ideal for measuring back gating. From Figure 3a, we can see that the resistance is decreasing with positive gating and it reaches ≈ 20% for 150 V. The decrease in resistance is monotonic in nature. The decrease in resistance is because more and more carriers are accumulating at the interface of LFO-STO resulting in a decrease in resistance with positive electrostatic gating. In the case of negative gating, the resistance is increasing as a function of gate voltage. The maximum increase in resistance is observed at -150 V (our measurement limit) is ≈ 60%. The increase in resistance on applying negative gate voltage is because the carriers (or oxygen vacancy) migrate inside the substrate and create polarization (capacitive effect) at the interface. This polarization increases as a function of gate voltage. Note that the resistance change in both cases is not going to its original value after removing the gate voltage. This shows the presence of defects states and impurities in the system. This persistent resistance remains for a long time, which is why before every measurement the sample is kept in dark for some time to remove any persistence remaining in the sample. One more important point here is to note that the interface is more sensitive towards light illumination than gating which makes it important in the field of radiation detector device application. The high sensitivity towards electromagnetic radiation is because of the spontaneous polarization occurring at the interface of LFO-STO reported by Nakamura et al. in 2016. [22]

Combined Electrostatic Light and Gate Effect
To see the combined effect of light and electrostatic gate, we have done measurements using different wavelength laser sources with both positive and negative back gating. The experiment is done as: first, the samples were kept in dark to achieve the original resistance value. Second, light and the back gate are applied simultaneously and recorded the temporal variation of change in resistance. Figure 3c shows the combined light and gating effect for different laser sources (blue for 405 , green for 532 , and red for 630 nm laser). It is important to note that the change in resistance is higher for a red light when a simultaneous light gate is applied to the system, in comparison to green and blue lights. This observation is different from what is reported for the LAO-STO interface. A clear understanding of the same is still lacking but intuitively we may say that the band gap of LFO is around ≈2.2 eV and the energy of red light (≈1.95 eV) is just a little smaller than that. Hence, a band banding induced by the application of an external gating is detected easily by red light and manifested in the change in resistance under the simultaneous application of light and gate. On the other hand, the energy of both green and blue is already higher than the band gap of LFO hence the simultaneous application of the light gate might be less effective. But this observation demands detailed theoretical modeling of this interface. The PPC effect sustain for a very long time (relaxation time 1 ≈4357 and 1 ≈69753 s, respectively, for green light at 10 mW)'' due to slow recombination of carriers at the interface.
In contrast to the positive gate+light illumination, for negative gate+light illumination, we observe an unusual increase of 90%, 100%, and ≈ 400% in resistance for 405 , 532 , and 630 nm laser source, respectively, shown in Figure 3d. This is quite opposite of what was observed for earlier reported interfaces. [15,23] The reason for this could understand as when we apply the negative gate the charge carriers start migrating deep in the substrate but at the same time, light illumination generates more carriers at the interface. The competition between these two phenomena gives interesting results at the interface. For low-wavelength laser sources, the light effect dominates the gating effect, preventing further increases in resistance. For high-wavelength laser sources, the energy per photon reduces which results in the gating effect dominating over the photoconductive effect. Further study needs here to understand this unusual increase in resistance. As shown in Figure 3d, for the 630 nm laser source the increase in resistance reaches up to 400%. This is a very high change compared to what was observed for the crystalline LaAlO 3 -SrTiO 3 interface at -200 V and 532 nm laser source(≈ 93%). [23] This unusual increase in resistance demands more study on this interface. Figure 3e shows the PPC effect that arises in the system in different experiments. The LFO-STO system show PPC when the only light is illuminated, only the positive gate (negative gate) is given, and the positive gate and light are illuminated simultaneously. This change in resistance and the PPC remains in the system for many cycles of measurements. The PPC is coming from the defects and impurities present in the system. This phenomenon finds application in storage devices. In Figure 3f, a negative gate of -150 V along with a 70 mW laser (405 nm) tunes the resistance of 2DEG formed at the interface of LFO-STO. Because of the light application, the increase in resistance is slower compared to when only the gate is applied to the interface. The change in resistance increase from ≈ 60% to ≈ 90%. When we switch off both the light and gate simultaneously the resistance comes to its initial value. The PPC is removed with the application of both light and negative gate. This is the same for many cycles (two cycles have been shown in this paper). This kind of resistance switch can be used for switching applications.
To see the carrier density effect we have performed light illumination experiments on 6 and 10 mL samples, having high carrier density. Figure 4a shows the percentage change in resistance on light illumination as a function of carrier density for different thicknesses of LFO thin films. We can see from Figure 4a that the photoconductivity decreases for high carrier concentration and becomes saturated. It is worth mentioning that to identify if the change in resistance arises solely from the carrier density change or mobility change or a combination of both, the Hall measurement should be performed under the application of light, gate, and light-gate. Because recently it has been reported that such stimuli can change the mobility of the system giving rise to a change in resistance. [27] But this is beyond the scope of the present paper.The R% is almost similar for both 6 and 10 mL samples. We have also compared our result with other STO and KTO-based interfaces reported for 405 nm laser light at 300 K (room temperature) shown in Figure 4b. The magnitude of PPC decreases monotonously with an increase of the product of carrier density and corresponding mobility (nx ) as reported for semiconductor heterojunctions. [28,29] Figure 4b shows that the photoconductivity effect is highest for the LFO-STO interface among other oxide interfaces reported earlier. The interesting part in both figures is that photoconductivity is inversely proportional to the carrier density. This effect can be understood as, for higher carrier density other fluctuations like thermal fluctuation, electron-phonon coupling, and electron-electron scattering came into the picture reducing the photoresponse of the material.

Conclusion
In conclusion, we have reported that the conducting interface of LFO-STO produces the highest photo conductivity (≈ 80%) and persistence photocurrent (≈ 60%) among so far reported perovskite oxide interfaces at room temperature. Our observation suggests a correlation between the product of carrier density and mobility with photoconductivity, photoconductivity increases with decreasing the product. Enhanced resistance change has been observed under the simultaneous application of red . Percentage change in resistance as a function of carrier density a) for different thicknesses of LFO thin film. b) For different interfaces (the red patch represents the STO-based interface and the blue patch represents the KTO-based interface). [12,13,15,[23][24][25][26] .
light and electrostatic gating, most likely because the red laser has energy near the bandgap of LFO (1.95 eV) whose modulation become visible under the simultaneous application of light and electrostatic gating. We have proposed protocols to use this interface as a resistive switch or memory device. Further logic could be implemented to achieve proper functionality of this interface at room temperature. This conducting interface may not only open up a pathway toward optoelectronic applications like holographic devices, radiation detectors, switching/memory devices, solar cells, etc but also be important from the point of view of fundamental material science due to the presence of iron band character in the conduction electron. This may give an additional degree of freedom to control the charge carrier transport in the system through an applied magnetic field. In short further detailed theoretical and experimental attention should be given to this interface that may open up a new way toward room temperature oxide electronics.

Experimental Section
Materials and Method: Thin films of LFO were grown on TiO 2 terminated STO (001) single crystal substrates. For TiO 2 termination, the hot water etching method is used. [30] The pristine STO (001) single crystals were annealed at 900 0 C temperature for 2 h in the presence of air. The substrates were heated in groups of two where one substrate is kept upside down on another substrate to avoid vacancies. [30] The high-temperature annealing aggregates SrO particles on the surface of the STO substrate, which was further removed using deionized water at 60 0 C in an ultrasonic bath. After etching high-temperature annealing is done at 1000 0 C for 2 h to get a step-terrace-like structure. The thin film of LFO was grown using a pulsed laser deposition (PLD) system. LFO target was prepared using a conventional solid-state reaction. The LFO target is ablated using a KrF excimer laser at a frequency of 2 Hz. For deposition, the substrate is heated at 760 0 C using an IR laser. The deposition is done at 1.5 J cm -2 laser fluence and oxygen partial pressure of 3× 10 −7 Torr. The substrate surface quality and thickness of LFO film are monitored using the reflection high energy diffraction (RHEED) technique.
Measurements: The transport measurements are done using a physical property measurement system (PPMS, Quantum Design 14T). For Hall measurements, the magnetic field was varied between +5 to -5 T. To characterize the photoconductivity and as an output voltage source, Keithley 2450 source meter was used. As a light source, solid-state lasers by Shanghai Dream Lasers Tech. Co., Ltd. of different wavelengths of 405 , 532 , and 630 nm were used.