Literature DB >> 35571849

All-Optical Reconfigurable Electronic Memory in a Graphene/SrTiO3 Heterostructure.

Liyun Qin1, Qinliang Li2, Shiteng Wu1, Jianyu Wang1, Zhendong Wang1, Li Wang1, Qisheng Wang1.   

Abstract

Direct optical data coding in an electronic device is meaningful for photonic technology. Herein, we report electronic memory in a graphene/SrTiO3 heterostructure, which presents the all-optical logic operation (encoding and decoding). The underlying physics have been elucidated in which the synergistic effect of surface localization with interface band bending is responsible for optical encoding and decoding in the electronic memory device of the graphene/SrTiO3 heterostructure. Further, we demonstrate a robust retention and synaptic-like processing of optical signals, which may lead to significant applications in neuromorphic imaging sensors.
© 2022 The Authors. Published by American Chemical Society.

Entities:  

Year:  2022        PMID: 35571849      PMCID: PMC9096928          DOI: 10.1021/acsomega.2c00938

Source DB:  PubMed          Journal:  ACS Omega        ISSN: 2470-1343


Introduction

Optical data processing,[1−3] featuring a high bandwidth and low power dissipation, is central to next-generation information technology such as big data, cloud computing, and artificial intelligence. Meanwhile, electrical data processing based on silicon field-effect transistors is still the dominating information operation principle in modern communication and memory technology because of high controllability of electrons and mature manufacture of integrated electrical circuits.[4−6] However, so far, optical information processing is always incompatible with integrated electrical circuits. The combination of optical transmission and data processing strategies with electrical chip memory technology is promising to break through the bottleneck of the state-of-the-art information industry. 2D material-based optoelectronic random memory (ORM) has recently renewed the researchers’ interest due to its intriguing electronic properties.[7,8] 2D materials like graphene,[9] MoS2, black phosphorus, and boron nitride with atomic-scale thin thickness strongly interact with external perturbations such as an electrical field,[10−14] surface adsorbates, and interface impurities.[15,16] Together with strong light–matter interactions,[17] 2D material-based ORM devices present high endurance and reproducibility,[8] multibit memory,[18,19] and CMOS compatibility.[20] However, the reported ORM devices are generally operated through optical writing and electrical erasing, which have not yet completely applied photons to control electronic memory.[21−24] The all-optical manipulation of electronic memory devices will enable the integration of photonics with electronic circuits. Herein, we propose all-optically manipulated ORM in a graphene/SrTiO3 heterostructure where the data is fully written and erased via light pulses. Optical information is stored by electrical methods, which connect the optical data manipulation with electrical information memory together. The behind mechanism has been elucidated via controlling the surface absorbates and photon energy. Lastly, the stability and multibit states have been demonstrated, indicating a neuromorphic imaging sensor application.[25,26]

Experimental Methods

Materials and Characterization

The all-optical manipulation of electronic memory was performed on two-terminal graphene on strontium titanate (SrTiO3). SrTiO3 with an area of 1 × 1 cm was bought from HF-Kejing Company. Monolayer graphene, which was transferred into the SrTiO3 substrate, was purchased from 6C Company. As shown in Figure S1, the composition was confirmed by Raman spectra (DXR3 laser confocal PL & Raman microscopy spectrophotometer). The excitation wavelength was 532 nm with a maximum laser power of 10 mW.

Optoelectronic Measurements

For characterizing the optoelectronic properties, the copper electrodes were fabricated using an electron beam evaporation system with a hard mask. Electrical transport was measured by a Keithley 2400 source meter at room temperature in a homemade probe station. TTL mode laser diodes with wavelengths of 405, 520, 658, 780, 850, and 980 nm were applied to irradiate the samples. A signal generator RIGOL LXI 2 was applied to adjust the laser output frequency.

Results and Discussion

Photoresponse Properties

The output characteristic curves of graphene under illumination of 520 and 405 nm lasers are shown in Figure a,b. The laser is focused on a channel with powers of 30 and 5 mW for 520 and 405 nm, respectively. The channel length is around 2 mm. Interestingly, the graphene device presents positive response at 520 nm illumination where the current increases while negative response at 405 nm illumination where the current decreases. The photoresponse (Figure S2) of 658, 780, 850, and 980 nm all shows positive response. As shown in Figure c, the photocurrent Iph = Ilight – Idark changes from −39.5 to 40.0, 32.5, 10.5, 17.6, and 37.1 μA as the light wavelength increases from 405 to 520, 658, 780, 850, and 980 nm, respectively. The laser powers of 405, 520, 658, 780, 850, and 980 nm are 5, 30, 40, 40, 40, and 40 mW, respectively.
Figure 1

Laser wavelength-modulated photoresponse in the graphene/SrTiO3 heterostructure. The 520 and 405 nm lasers respectively induce (a) positive and (b) negative photoresponse. (c) Photocurrent Iph changes from negative to positive as the laser wavelength increases from 405 to 520, 658, 780, 850, and 980 nm. The Iph is extracted at a voltage bias of 1 V.

Laser wavelength-modulated photoresponse in the graphene/SrTiO3 heterostructure. The 520 and 405 nm lasers respectively induce (a) positive and (b) negative photoresponse. (c) Photocurrent Iph changes from negative to positive as the laser wavelength increases from 405 to 520, 658, 780, 850, and 980 nm. The Iph is extracted at a voltage bias of 1 V.

Reconfigurable Optoelectronic Memory

We next demonstrate the all-optical manipulation of electrical memory of graphene based on persistent positive and negative photocurrent effects. As shown in Figure a, a one-second 520 nm laser pulse induces a spontaneous increase in current. The Ioff and Ion respectively represent the dark current and current with light on. Then, the current exponentially decays to a stable current value (Ipersistent). The persistent positive photoresponse can also be seen with 658 and 850 nm laser illumination (Figure S3). The negative persistent photocurrent can be realized by irradiation of a one-second 405 nm laser pulse as shown in Figure b. The current sharply drops and comes to a stable value with the application of a 405 nm laser pulse. Therefore, an all-optical control of data writing and erasing of electrical memory can be realized as shown in Figure c. The continuous 520 nm laser pulses followed with 405 nm laser pulses constitute the reproducible encoding and decoding function, respectively. The readout current is obtained under a voltage bias of 1 V in all measurements.
Figure 2

All-optical reconfigurable electronic memory. (a) A 520 nm laser pulse induces the persistent photocurrent. (b) Erasing operation of a 405 nm laser pulse. The pulse duration is 1 s with powers of 30 and 5 mW for 520 and 405 nm lasers, respectively. (c) Reconfigurable electronic memory of graphene with optical encoding (520 nm laser pulses) and decoding (405 nm laser pulses).

All-optical reconfigurable electronic memory. (a) A 520 nm laser pulse induces the persistent photocurrent. (b) Erasing operation of a 405 nm laser pulse. The pulse duration is 1 s with powers of 30 and 5 mW for 520 and 405 nm lasers, respectively. (c) Reconfigurable electronic memory of graphene with optical encoding (520 nm laser pulses) and decoding (405 nm laser pulses).

Effect of Surface Absorbents

In order to illuminate the underlying physics of persistent photocurrent effects in graphene, we first annealed the graphene device in a furnace at 400 °C with a duration of 20 min. The tube was fed with a 150 sccm inert argon atmosphere. As shown in Figure a, both 658 nm laser-induced positive and 405 nm laser-induced negative photoresponse disappear after annealing. The current–voltage curve with light on repeats that under dark conditions. Therefore, the surface absorbents should be the origin of persistent photoconduction effects in graphene. This would be further confirmed by the experiment of the same devices, which were exposed to ambient for 24 days. As shown in Figure b, the positive and negative photoresponse is completely recovered. The device again shows persistent and erasable photocurrent as shown in Figure c. The 405 and 520 nm lasers with a pulse width of 1 s and a frequency of 1 Hz were applied to illuminate the samples.
Figure 3

Effect of surface absorbents on the memory effect. (a) After annealing, the positive and negative photoresponse disappears. (b,c) After exposure of the samples to ambient for 24 days, the devices recover their original states. (b) Low (wavelength, 658 nm)-/high (wavelength, 405 nm)-energy photons induce the positive/negative photoresponse. (c) Low (wavelength, 520 nm)-/high (wavelength, 405 nm)-energy photons arouse positive/negative persistent photoresponse.

Effect of surface absorbents on the memory effect. (a) After annealing, the positive and negative photoresponse disappears. (b,c) After exposure of the samples to ambient for 24 days, the devices recover their original states. (b) Low (wavelength, 658 nm)-/high (wavelength, 405 nm)-energy photons induce the positive/negative photoresponse. (c) Low (wavelength, 520 nm)-/high (wavelength, 405 nm)-energy photons arouse positive/negative persistent photoresponse.

Memory Mechanism and Performance

The above results point out that the surface absorbents are central to the reconfigurable all-optical processing on electrical memory of graphene. As shown in Figure a, the p-type doping of graphene originates from the presence of oxygen functional groups on the surface of graphene.[27,28] The Fermi level (EF) resides in the valence band. According to previous theoretical calculation of electronic density of states, oxygen brings out strongly localized and half-filled states at the Fermi level.[29] The photon with energy hν ≥ 2EF allows the transition of an electron from the valence to conduction band. The laser with a wavelength of 520–980 nm excites holes and electrons.[30] The electrons are captured by electron-trapping centers at the Fermi level. The localized electrons act as the negative gating effect, which accumulates holes in the channel and induces a persistent and positive photocurrent in p-type graphene. However, the short-wavelength laser (405 nm) with a large photon energy of 3.1 eV induces an electron transition from the valence band to the conduction band of SrTiO3.[31] As shown in Figure a, the energy alignment between graphene and SrTiO3 presents a band bending with a built-in field from SrTiO3 to graphene.[32] The photoexcited holes in SrTiO3 are injected to graphene. Then, the injected holes neutralize localized electrons, resulting in the reduction of current. Therefore, a 405 nm laser erases the persistent positive photocurrent. Also, graphene presents the persistent negative conductivity behavior.
Figure 4

Mechanism and performance of all-optical reconfigurable electronic memory. (a) Operation principle of optical writing and erasing; low-energy photon-induced electrons are localized, which induces positive and persistent photocurrent, and the photons with energy higher than that of SrTiO3 excite holes, which transfer into graphene through the interface built-in field and neutralize the localized electrons. (b) Devices show stable and endurable on and off states. (c) Neuromorphic function of potentiation and depression through applying illumination of continuous 520 and 405 nm laser pulses.

Mechanism and performance of all-optical reconfigurable electronic memory. (a) Operation principle of optical writing and erasing; low-energy photon-induced electrons are localized, which induces positive and persistent photocurrent, and the photons with energy higher than that of SrTiO3 excite holes, which transfer into graphene through the interface built-in field and neutralize the localized electrons. (b) Devices show stable and endurable on and off states. (c) Neuromorphic function of potentiation and depression through applying illumination of continuous 520 and 405 nm laser pulses. We finally evaluate the performance of our all-optical operated electrical memory of graphene. As shown in Figure b, the memory on and off states are highly stable within 24 cycles. In addition, the multistate writing and erasing operations are realized through irradiation of 520 and 405 nm laser pulses with a 1 Hz frequency. This indicates a pulse-tunable synaptic potentiation/depression function with neuromorphic computing application in the future. However, our electric memory has a low on/off ratio due to the semimetallic nature of graphene. To address the low on/off ratio in graphene, one may use the other alternative two-dimensional materials like MoS2 with semiconductive properties in the future. The writing and erasing speed of nanosecond scale is another obstacle for practical application. The slow response likely originates from the surface/interface disorders like charged surface states and impurities, substrate surface roughness, and substrate optical phonons. The response speed may be improved through packing of graphene with boron nitride or PMMA,[33] which will also improve the cyclability of the devices. Despite these, the all-optical controlled electrical memory of graphene fully utilizing the photon properties of low power dissipation would generate applications of optical data processing that does not require a large communication bandwidth.

Conclusions

In summary, we report an all-optically operated electronic memory device that encodes and decodes signals through light pulses. This changes the work modes of conventional optoelectronic memory in which light writes the signals while electricity erases information. The all-optical reconfigurable electronic memory of graphene originates from surface absorbates that generate the localization states of electrons. The all-optical coding of electronic memory provides a strategy for full integration of optical and electrical information technology, which may lead to fascinating applications in optical data memory and artificial visual systems.
  25 in total

1.  All-optical switching for 10-Gb/s packet data by using an ultralow-power optical bistability of photonic-crystal nanocavities.

Authors:  Kengo Nozaki; Amedee Lacraz; Akihiko Shinya; Shinji Matsuo; Tomonari Sato; Koji Takeda; Eiichi Kuramochi; Masaya Notomi
Journal:  Opt Express       Date:  2015-11-16       Impact factor: 3.894

2.  Optoelectronic memory using two-dimensional materials.

Authors:  Sidong Lei; Fangfang Wen; Bo Li; Qizhong Wang; Yihan Huang; Yongji Gong; Yongmin He; Pei Dong; James Bellah; Antony George; Liehui Ge; Jun Lou; Naomi J Halas; Robert Vajtai; Pulickel M Ajayan
Journal:  Nano Lett       Date:  2014-12-17       Impact factor: 11.189

3.  Toward a silicon-based quantum computer.

Authors:  Lars R Schreiber; Hendrik Bluhm
Journal:  Science       Date:  2018-01-26       Impact factor: 47.728

4.  Optoelectronic resistive random access memory for neuromorphic vision sensors.

Authors:  Feichi Zhou; Zheng Zhou; Jiewei Chen; Tsz Hin Choy; Jingli Wang; Ning Zhang; Ziyuan Lin; Shimeng Yu; Jinfeng Kang; H-S Philip Wong; Yang Chai
Journal:  Nat Nanotechnol       Date:  2019-07-15       Impact factor: 39.213

5.  Graphene photonics, plasmonics, and broadband optoelectronic devices.

Authors:  Qiaoliang Bao; Kian Ping Loh
Journal:  ACS Nano       Date:  2012-05-02       Impact factor: 15.881

6.  Graphene-MoS2 hybrid structures for multifunctional photoresponsive memory devices.

Authors:  Kallol Roy; Medini Padmanabhan; Srijit Goswami; T Phanindra Sai; Gopalakrishnan Ramalingam; Srinivasan Raghavan; Arindam Ghosh
Journal:  Nat Nanotechnol       Date:  2013-10-20       Impact factor: 39.213

Review 7.  Commercialisation of CMOS integrated circuit technology in multi-electrode arrays for neuroscience and cell-based biosensors.

Authors:  Anthony H D Graham; Jon Robbins; Chris R Bowen; John Taylor
Journal:  Sensors (Basel)       Date:  2011-05-04       Impact factor: 3.576

8.  Nonvolatile infrared memory in MoS2/PbS van der Waals heterostructures.

Authors:  Qisheng Wang; Yao Wen; Kaiming Cai; Ruiqing Cheng; Lei Yin; Yu Zhang; Jie Li; Zhenxing Wang; Feng Wang; Fengmei Wang; Tofik Ahmed Shifa; Chao Jiang; Hyunsoo Yang; Jun He
Journal:  Sci Adv       Date:  2018-04-20       Impact factor: 14.136

Review 9.  2D Materials Based Optoelectronic Memory: Convergence of Electronic Memory and Optical Sensor.

Authors:  Feichi Zhou; Jiewei Chen; Xiaoming Tao; Xinran Wang; Yang Chai
Journal:  Research (Wash D C)       Date:  2019-08-22

10.  Logic-in-memory based on an atomically thin semiconductor.

Authors:  Guilherme Migliato Marega; Yanfei Zhao; Ahmet Avsar; Zhenyu Wang; Mukesh Tripathi; Aleksandra Radenovic; Andras Kis
Journal:  Nature       Date:  2020-11-04       Impact factor: 49.962

View more

北京卡尤迪生物科技股份有限公司 © 2022-2023.