Journal of Materials Science & Technology  2020 , 38 (0): 237-243 https://doi.org/10.1016/j.jmst.2019.07.042

Research Article

Direct patterning of reduced graphene oxide/graphene oxide memristive heterostructures by electron-beam irradiation

Olesya O. Kapitanovaab, Evgeny V. Emelinc, Sergey G. Dorofeeva, Pavel V. Evdokimova, Gennady N. Panincd, Youngmin Leee, Sejoon Leeef*

aLomonosov Moscow State University, Leninskie gory, 119991 Moscow, Russia
bMoscow Institute of Physics and Technology, 9 Institutskiy per., Dolgoprudny, Moscow Region, 141701, Russia
cInstitute of Microelectronics Technology and High-Purity Materials, Russian Academy of Sciences, Chernogolovka, Moscow district, 142432, Russia
dNano Information Technology Academy, Dongguk University-Seoul, Seoul 04620, Korea
eQuantum-Functional Semiconductor Research Center, Dongguk University-Seoul, Seoul 04620, Korea
fDepartment of Semiconductor Science, Dongguk University-Seoul, Seoul 04620, Korea

Corresponding authors:   ∗Corresponding author at: Quantum-Functional Semiconductor Research Center, Dongguk University-Seoul, Seoul 04620, Republic of Korea.E-mail address: sejoon@dongguk.edu (S. Lee).

Received: 2019-06-13

Revised:  2019-07-12

Accepted:  2019-07-23

Online:  2020-02-01

Copyright:  2020 Editorial board of Journal of Materials Science & Technology Copyright reserved, Editorial board of Journal of Materials Science & Technology

More

Abstract

Memristive heterostructures, composed of reduced graphene oxide with different degree of reduction, were demonstrated through a simple method of ‘direct electron-beam writing’ on graphene oxide. Irradiation with an electron beam at various doses and accelerating voltages made it possible to define high- and low-conductivity graphene-oxide areas. The electron beam-reduced graphene oxide/graphene oxide heterostructure clearly exhibited a nonlinear behavior and a well-controlled resistive switching characteristic at a low operating-voltage range (< 1 V). The proposed memristive heterostructures are promising for highly-efficient digital storage and information process as well as for analogous neuromorphic computations.

Keywords: Electron beam irradiation ; Reduced graphene oxide ; Graphene oxide ; Memristive heterostructure

0

PDF (1758KB) Metadata Metrics Related articles

Cite this article Export EndNote Ris Bibtex

Olesya O. Kapitanova, Evgeny V. Emelin, Sergey G. Dorofeev, Pavel V. Evdokimov, Gennady N. Panin, Youngmin Lee, Sejoon Lee. Direct patterning of reduced graphene oxide/graphene oxide memristive heterostructures by electron-beam irradiation[J]. Journal of Materials Science & Technology, 2020, 38(0): 237-243 https://doi.org/10.1016/j.jmst.2019.07.042

1. Introduction

Graphene, a two-dimensional carbon crystal with the hexagonal honeycomb lattice structure, has great potential for high-performance electronics due to its fascinating electrical, thermal, and mechanical properties. In recent years, graphene field-effect transistor (GFETs) have been demonstrated on various device schemes for electronics [[1], [2], [3], [4], [5], [6]], optoelectronics [[7], [8], [9]], and bioelectronics [[10], [11], [12], [13], [14], [15], [16]] applications. However, the absence of the energy bandgap in graphene restricts the tangible application of the GFETs because the gapless nature of graphene inevitably results in a relatively large off-state current in most GFETs. As a family group of graphene-based material systems, graphene oxide (GO) has emerged to huddle up such a critical issue. GO, consisting of graphene bound to oxygen in the form of carboxyl, hydroxyl, carbonyl or epoxy groups, possesses a wide bandgap energy (Eg) up to 6 eV, and the magnitude of Eg can be controlled by the reduction of GO (i.e., formation of reduced-GO (rGO)). The semiconductor type of rGO with Eg = 0-4 eV [17,18] (depending upon the C/O ratio) is normally produced by chemical and thermal reduction of GO [19,20]. In addition, the high-energy electron irradiation (10 MeV) [21,22] and the laser irradiation [23] methods have been proposed as the effective ways to produce the fine structure of rGO. These techniques allows an easy control of electrical conductivity in rGO, and can be used for nonchemical reduction with no use of any reducing agent that is undesirable due to residual contamination. Owing to the tunability of Eg as well as electrical conductivity, rGO is promising for use in several types of nanoelectronic device schemes; for example, FETs [24,25], nonvolatile memories [26,27], and memristors [28,29]. Among various device applications of rGO, the memristive switching devices are currently of vast interest to demonstrate an artificial synaptic system for future information processing [[30], [31], [32]]. To increase the scalability and to construct more sophisticated device architectures, the partial and selective formation of high-quality GO, rGO, and their heterostructures are essential.

Motivated by all the above, we investigated the simple route to form the fine patterns of electron beam-reduced graphene oxide (EB-rGO) by ‘direct electron-beam writing’ onto the GO film that had been spin-coated on the SiO2/Si substrate. In this work, we demonstrate the memristive heterostructure of EB-rGO/GO through the low-energy (3-10 keV) electron-beam irradiation technique. This unique method of ‘direct electron-beam writing’ enables us to form the highly-conductive p-type EB-rGO (C/O > 10) stripe patterns in parallel with the dielectric GO (2 < C/O < 4) areas. The lateral heterostructures of EB-rGO/GO exhibit a nonlinear behavior and a well-controlled resistance-switching characteristic in an electric field at low operation-voltage ranges (< 1 V). Herein, we thoroughly examine the properties of EB-rGO and discuss its formation kinetics. Additionally, the electrical characteristics of the EB-rGO/GO memristive heterostructure are also represented.

2. Experimental

Fig. 1 schematically displays the experimental procedures for the fabrication of the EB-rGO/GO memristive heterostructure. To construct the EB-rGO/GO memristor, as a primary task, a high-quality thin layer of GO should be spin-coated onto the pre-patterned Pt/Ti/SiO2/Si substrate (Fig. 1(a) and (b)). Therefore, we firstly synthesized GO by using the modified Hummers’ method [19,20]. A long oxidation time was combined with a highly effective method to purify the reaction products. With this method, a mixture of concentrated H2SO4/H3PO4 (360:40 mL) was added to a mixture of graphite flakes (3.0 g) and KMnO4 (18.0 g). The reaction mixture was stirred for 12 h at 50 °C and then it was added to 500 ml of 5 wt% H2SO4 aqueous solution for 20 min with stirring. The resultant mixture was further stirred for 1 h. For workup, the mixture was sifted through two metal standard testing sieves with diameters of 320 μm and 30 μm. The filtrate was centrifuged (9000 rpm for 20 min), and the supernatant was decanted away. The remaining solid material was then washed in succession with 400 mL of 30% HCl and 1000 mL of water. The material remaining after this extended multiple-wash process was freeze-dried and used to prepare a GO suspension. The resultant products of GO flakes have a thickness in the range of 0.6-0.9 nm and a width of 0.5-3 μm (see Supplementary data). The formation of flakes and their dimensions have been described in detail previously [21,28,29,33,34]. The structures were prepared by spin-coating of GO-suspension in water (2.5 mg/ml) on a SiO2/Si substrate with Pt/Ti electrodes pre-fabricated by electron-beam deposition, at v1 = 500 rpm for t1 = 10 s and v2 = 4500 rpm for t2 = 45 s (Fig. 1(b)). The scanning of the electron beam in the form of stripes was used to control the local electron-beam reduction of GO (Fig. 1(c)). Doses of irradiation with an electron beam were varied from 30 to 3000 mA s/cm2 by changing both the accelerating voltage (i.e., Vacc = 3 and 10 keV) and the beam current (i.e., Ibc = 150 pA-40 mA). The main purpose of choosing two different Vacc was to find out the optimal Vacc condition for effective reduction of GO (i.e., whether higher Vacc or lower Vacc). Through the electron-beam writing process, we eventually obtained the lateral array of the EB-rGO/GO heterostructures (Fig. 1(d)). To the electrodes, the bias voltage could be applied in the range from 0 to 50 V. We here note that another type of rGO was also prepared as a reference sample by using the conventional thermal annealing method at 360 °C in N2/H2 (90%/10%).

Fig. 1.   Schematic illustrations for the fabrication of the lateral EB-rGO/GO/EB-rGO memristive heterostructure: (a) patterning of Pt/Ti electrodes on the SiO2/Si substrate, (b) spin-coating of the GO film onto the Pt pre-patterned SiO2/Si substrate, (c) electron-beam writing onto GO for the formation of local EB-rGO regions, and (d) memristive heterostructure in the form of laterally-arrayed EB-rGO/GO/EB-rGO stripes.

The current‒voltage (I-V) characteristics of the fabricated structures were measured using a Keithley 4200 SCS semiconductor parameter analyzer in the voltage-sweeping mode at room temperature. The I-V measurements were carried out in an ultrahigh vacuum chamber (base pressure 3 × 10-10 mbar) immediately after irradiation with an electron beam. To characterize charge-carrier transport properties, Hall-effect measurements were carried out using an Ecopia-HMS-5300 system. Micro Raman spectra of the graphene oxide films were obtained with a Renishaw Raman microscope at a 633 nm (1.96 eV) excitation wavelength. The spectral resolution was $\widetilde{1}$ cm-1. Raman D and G bands were fitted with Lorentzian functions. High-resolution scanning electron microscopy (SEM, Philips XL-30) and atomic force microscopy (AFM, N-TEGRA NT-MDT) were used to analyze the topography of structures.

3. Results and discussion

A square pattern of the GO film was obtained using spin-coating of GO onto the pre-patterned Pt/Ti/SiO2/Si substrate and lift-off processes (Fig. 2(a)). From the AFM analysis, the thickness of the patterned GO film was confirmed to be $\widetilde{2}$0 nm (Fig. 2(b) and (c)), which corresponds to $\widetilde{2}$5 layers of GO. Electron-beam irradiation of the GO film at Vacc of 3 kV and Ibc of 0.5 nA (for 15 min) leads to a bright secondary electron contrast. As shown in the inset of Fig. 2(d), stripe patterns are created after electron-beam irradiation along the horizontal direction between Pt electrodes. The electron beam-irradiated GO region became conductive, and the electrical conductivity showed to depend on the radiation dose. Fig. 2(d) represents the I-V curves of electron beam-irradiated GO with different doses. An increase in the irradiation dose leads to an increase in electrical conductivity of electron beam-irradiated GO by several orders of magnitude (see also Table 1). This indicates the effective removal of oxygen groups from the electron beam-irradiated GO region and its reduction.

Fig. 2.   (a) Top-view SEM image of the GO film (dark square) coated on the Pt (bright bands) pre-patterned Si/SiO2 substrate, (b) AFM image of the GO film; (c) AFM histogram for the GO film, and (d) I-V curves of the EB-rGO stripes fabricated by electron-beam irradiation with various doses (D = 36-216 mA s/cm2). The inset of (d) displays the SEM image of the sample, representing the EB-rGO stripes formed by electron-beam irradiation.

Table 1   Sheet resistance of as-prepared GO and electron beam-reduced rGO films formed at various electron-beam irradiation conditions.

SamplesVacc (kV)Ibc (nA)D (mA s/cm2)Vb (V)R (Ω/sq.)
GO---0>1010
EB-rGO353605 × 106
722 × 106
1088 × 105
1446 × 105
1805 × 105
2164 × 105
100.151002 × 109
2504 × 108
3503 × 108
5001 × 108
150503 × 103

New window

To find out the optimal condition for electron beam-induced reduction of GO, we investigated the variations of sheet resistance for EB-rGO upon varying both Vacc and Ibc. As can be confirmed from Fig. 3(a) and (b), GO is reduced more efficiently at lower Vacc. For example, when Vacc =3 kV, the sheet resistance of EB-rGO decreases down to $\widetilde{4}$× 105 Ω/sq. (Fig. 3(a)). This value is 5 orders of magnitude lower than that of the as-deposited highly-resistive GO film (1010 Ω/sq.). When Vacc =10 kV, however, EB-rGO exhibits a relatively high sheet-resistance ($\widetilde{1}$08 Ω/sq.) even though the dose is increased up to 500 mA s/cm2. At the same dose condition, the sheet-resistance of EB-rGO prepared by electron-beam irradiation at Vacc =3 kV and Ibc =5 nA is 3 orders of magnitude lower than that of EB-rGO obtained by electron-beam irradiation at Vacc =10 kV and Ibc =0.15 nA (see also Table 1).

Fig. 3.   (a) Sheet resistance of EB-rGO as a function of the electron-beam dose supplied at Vacc =3 kV and Ibc =5 nA, (b) dependence of the electron-beam dose on sheet resistance of EB-rGO prepared at Vacc =10 kV and Ibc =0.15 nA, (с) IV curves characteristics of EB-rGO fabricated by electron-beam irradiation with a dose of 150 mA s/cm2 at Vacc =10 kV and Ibc = 150 pA with and without bias voltages, and (d) Raman spectra of pristine GO and electron beam-induced EB-rGO.

Based on the above, one can conjecture that the electron-beam irradiation at higher Ibc and with a lower Vacc yields a higher reduction efficiency for the formation of EB-rGO. A lower penetration depth (R) of electrons at 3 kV results in a higher density of e-h pairs and more efficient energy transfer to a thin layer of GO on the surface. The value of R as a function of the electron-beam energy can be calculated using the Kanaya and Okayama expression [35]:

R=$\frac{0.0276×E^{1.67}×A}{Z^{0.0889}×ρ}$

where E (keV) is the energy of the electron beam, A (g/mole) is the molar mass of the substance, ρ (g/cm3) is the density of the substance, and Z is the atomic number of the substance. Therefore, the ratio of R10keV/R3keV could be simply estimated as:

$\frac{R_{10keV}}{R{3keV}}=\frac{10^{1.67}}{3^{1.67}}=7.5$.

An electron beam at Vacc of 3 kV generates a maximum electron-hole density in a volume that is 7.5 times closer to the surface of the heterostructure, compared to the case of Vacc =10 kV. This creates more charge carriers in electron beam-irradiated GO during the formation of EB-rGO at higher Ibc and lower Vacc. High-energy primary electrons participate in the reduction process by absorption of their energy and generating e-h pairs with energies close to the forbidden band of GO (4-6 eV). The thermal energy from possible heating of the sample during electron-beam irradiation with higher Ibc may also partially contribute to the GO reduction process. However, electron-stimulated reduction at low Vacc is a more efficient process and can be used to accurately control the conductivity of GO. One electron with an energy of 3-10 keV is capable of generating about 103e-h pairs in GO for its efficient electron-beam reduction. This suggests that electron-beam irradiation is more efficient than other techniques for the formation of fine structures of patterned rGO. For instance, only one e-h pair could be generated in rGO by irradiating one UV photon (3-4 eV) during laser annealing of GO; hence, the laser is usually used for local annealing of GO [23].

Here, it should be also noted that the electric-field applied to electrodes strongly affects the reduction efficiency during the electron-beam irradiation process. Compared to the case of zero bias ($\widetilde{1}$0-9 Ω-1), the electrical conductance of EB-rGO is increased by 6 orders of magnitude (up to $\widetilde{3}$ × 10-3 Ω-1) when the bias voltage (Vb) of 50 V is applied to Pt electrodes during electron-beam induced reduction (Fig. 3(c)). This confirms an electron-stimulated origin of the GO reduction. The electric-field applied to the structure separates the e-h pairs, and prevents their recombination; thereby, most of the e-h pairs could contribute to an electron-stimulated reduction process with no thermal-heating. A significant increase in conductivity of EB-rGO prepared when Vb = 50 V indicates that the e-h pairs having an energy of several electron volts play a key role for electron-stimulated reduction of GO. All of the above imply that the electron-beam reduction process can be more efficient than conventional thermal annealing at high temperatures because electron-beam irradiation allows direct writing of local EB-rGO patterns at room temperature by utilizing a fine electron beam.

Electron beam-induced reduction of GO could be confirmed by Raman spectroscopy measurements. Fig. 3(d) shows Raman spectra of GO films before and after electron-beam irradiation with different beam currents. The intensity ratio of the G peak to the D peak (i.e., IG/ID) in Raman spectra corresponds to 1.1, 1.1, and 1.3 for the samples of as-prepared GO, EB-rGO (Ibc =5 nA), and EB-rGO (Ibc =40 mA), respectively. For the EB-rGO sample prepared by electron-beam irradiation with Ibc of 5 nA, the intensity ratio of IG/ID is comparable to that of pristine GO. In the case of EB-rGO prepared under higher Ibc (=40 mA), however, the D peak intensity is considerably decreased. As a result, the intensity ratio of IG/ID is increased up to 1.3 after electron beam irradiation with Ibc of 40 mA. The magnitude of IG/ID is well known to correlate with the size of sp2 clusters in sp3- and sp2-bound carbon networks. According to Mattevi et al. [36], the increase in IG/ID is attributed to the increased sp2 cluster size in the sp3 matrix because sp2 carbon-carbon bonds are restored by de-oxidation during thermal reduction of GO. Therefore, we can infer that effective reduction of GO to EB-rGO took place by electron-beam irradiation with higher Ibc.

To further elucidate the increase in sp2 carbon-carbon bonds, we examined electrochemical bonding states of GO before and after electron-beam irradiation. Fig. 4(a) and (b) displays the XPS spectra of C 1s for pristine GO and EB-rGO that had been prepared by electron-beam reduction with Ibc of 40 mA, respectively. Both GO and EB-rGO samples clearly reveal carbon- and oxygen-related C 1s core levels, and the XPS spectra of C 1s could be deconvoluted by 5 peaks of sp2 (284.6 eV), sp3 (285.5 eV), C—O—C with C—OH (287.5 eV), C═O (288.8 eV), and O═C—O (289.5 eV) bonds. For pristine GO, carbons are predominantly bonded with oxygens although sp3- and sp2-hybridized carbon bonds are formed (Fig. 4(a)). After electron-beam irradiation with Ibc of 40 mA, however, the portions of sp2-hybridized carbons are dramatically increased whereas the fractions of oxygen-related bonds are significantly decreased (Fig. 4(b)). This corroborates that electron-beam irradiation particularly with higher Ibc could promote effective reduction of GO to EB-rGO.

Fig. 4.   XPS spectra of the C 1s core level for (a) pristine GO and (b) EB-rGO prepared by electron-beam irradiation with Ibc of 40 mA.

Despite such a remarkable increase in sp2-hybridized carbon bonds in EB-rGO, a relatively large portion of oxygen-related bonds still exists. The residual oxygen molecules act as acceptors in rGO; therefore, the presence of oxygen molecules confers the p-type doping effect in rGO [37]. In addition, the carrier type and conductivity of rGO strongly depend on the degree of reduction, which can be varied by controlling external energies (e.g., thermal annealing temperature, laser energy density, electron beam dose) supplied during GO reduction. Tu et al. [38] reported that n-type conductivity could be achieved when performing thermal reduction at 300-450 °C and 800-1000 °C, while highly conductive p-type rGO could be obtained only at annealing temperatures between 450 and 800 °C. Bhaumik and Narayan [23] observed only a type conversion from p-type to n-type when the laser power of 0.6-1.0 J/cm2 was supplied during laser-assisted reduction of GO using an ArF excimer laser source (λ =193 nm). Kwon et al. [22] observed p-type conductivity from rGO prepared by electron-beam irradiation with a dose of 50 kGy. As listed in Table 2, in our case, the EB-rGO sample showed p-type conductivity with the hole concentration of 5 × 1016 cm-3, whereas thermally-reduced GO at 360 °C (i.e., reference sample) exhibited n-type conductivity with the electron density of 5 × 1017 cm-3. Thus, we believe that electron-beam reduction belongs to one of the effective fabrication methods to produce p-type rGO.

Table 2   Hall-effect measurement results of as-prepared GO, electron beam-reduced rGO, thermally-reduced GO films.

SamplesCarrier concentration
(cm-3)
Conductivity
(S/cm)
As-prepared GO<1010<10-8
Thermally-reduced GO
(N2/H2, 360 °C)
-5 × 10170.4
EB-rGO
(Ibc =1 nA, Vacc =10 keV, D =200 mA s/cm2)
5 × 10168 × 10-2

New window

Electron-beam reduction of GO can be explained in terms of the radical formation/deformation mechanism, which is stimulated via energy-transferring of electron beam-induced hot electrons inside the local area of GO. For reduction of GO, first of all, several electron volts are required to break and remove oxygen groups. Since the electron beam has a higher energy than other beam sources (e.g., γ-ray) [22,39], the irradiated electron-beam energetically excites the electronic subsystem in GO, and results in generation of a large number of hot electrons in GO. The energy from hot electrons can be resonantly absorbed by the functional groups in the local regions of GO. As a result, the C—O and/or C—H bonds could be effectively broken because of their weaker bonding strengths than C—C bonds. This causes the formation of highly-reactive O$\dot{a}$nd H$\dot{r}$adicals inside the disjunct GO boundaries. The radicals eventually transfer into H2O, H2, and O2 through recombination with hydrogen and oxygen atoms, and the rest charges uncompensated during the radical formation/deformation process help restoring the sp2 carbon bonds so that GO could be reduced to EB-rGO.

Well-controlled electron-beam irradiation allows ‘direct writing’ of EB-rGO patterns along with rest GO areas. This enables us to easily fabricate memristive heterostructures by local patterning of EB-rGO on GO with no use of a complicated mask pattern process and high temperatures. The lower inset of Fig. 5(a) displays an SEM image of the EB-rGO/GO/EB-rGO heterostructure obtained through ‘direct writing’ of EB-rGO stripes using a 3 kV electron-beam with the dose of 150 mA s/cm2. A brighter contrast of the secondary electron emission from EB-rGO (i.e., horizontal stripes) indicates local reduction of GO, and represents a successful formation of the lateral EB-rGO/GO/EB-rGO heterostructure. When bias voltages are applied to the EB-rGO/GO/EB-rGO structure, the device shows nonlinear soft resistive-switching with a weak hysteresis (Fig. 5(a)). The forming process at 20 V leads to the decrease in conductivity by several orders of magnitude, and gives rise to a noticeable I-V non-linearity of the heterostructure (Fig. 5(b)). The device reveals a strong bipolar hysteresis characteristic with resistive-switching behaviors from the high-resistive-state (HRS) (i.e., (1.2 ± 0.1) × 1011 Ω) to the low-resistive-state (LRS) (i.e., (6.7 ± 0.4) × 108 Ω) at switching voltages of 0.8-0.9 V. In addition, the device shows a good reproducibility of both HRS and LRS upon repeating multiple switching cycles.

Fig. 5.   (a) IV characteristics (with no forming-voltage stress) of the lateral EB-rGO/GO/EB-rGO memristive heterostructure obtained by electron-beam irradiation with D =200 mA s/cm2 at Vacc =3 keV. The upper and the lower insets of (a) show the bias setup for the IV measurement of the fabricated memristor and the top-view SEM image of the fabricated memristor in the form of laterally-arrayed EB-rGO/GO/EB-rGO stripes. (b) Dependence of the hysteretic resistive-switching characteristics on the number of switching cycles for the EB-rGO/GO/EB-rGO memristive heterostructure after applying the forming-voltage stress (20 V, 15 min).

We ascribe the observed hysteretic resistive-switching characteristics in our EB-rGO/GO/EB-rGO heterostructure to the redox process in the GO insulator, which is laterally sandwiched in between conductive EB-rGO stripes. Similar resistive switching behaviors were also observed in the lateral Al/GO/Al geometry, where GO with a thickness of 50 nm was formed by using electromigration of oxygen through the lateral redox process [26,40], The resistance change in GO can be controlled by the formation of conductive graphene filaments due to the drift of oxygen groups inside the local GO area [[26], [27], [28],41], to which the high external electric-field is applied via conductive EB-rGO stripes. The mechanism of the high electric field-induced local resistance change in such a structure is associated with a cluster model of GO, which is indicative of sp2 graphene clusters in an oxygen-enriched sp3 matrix [24] and their reversible rearrangement under the electric-field [40,41], Rearrangement of sp2 and sp3 domains occurs through the redox process, and causes resistive-switching in GO. We, here, notice that the size of the lateral heterostructure (l: 2.5 μm, w: 3 μm, t: 20 nm) influences the magnitude of the forming voltage, but does not have a strong influence on the resistive-switching process. The switching voltage depends on the potential barrier at the interface between GO and electrode materials, and the magnitude of bias voltages for hysteretic resistive-switching can be reduced by using highly-conductive p-type GO [40]. All of the above suggest that electron-beam reduction of GO can be a powerful solution for an advanced fabrication method to form rGO/GO complex memristive nanosystems with neuromorphic architecture on various substrates at room temperature.

4. Conclusion

Memristive EB-rGO/GO/EB-rGO heterostructures demonstrated by using ‘direct electron-beam writing’ of GO, which had been spin-coated onto the Pt pre-patterned SiO2/Si substrate. Electron-beam irradiation resulted in electron-stimulated reduction of GO (i.e., formation of EB-rGO), and the conductivity of EB-rGO could be controlled by changing the irradiation dose of the electron-beam. The laterally-arrayed two-dimensional memristive heterostructure was created by local reduction of GO through ‘direct electron-beam writing’ of EB-rGO stripes. The device clearly showed stable resistive-switching with an HRS/LRS ratio by two orders of magnitude at low operating-voltage ranges of 0.8-0.9 V. The results provide a new avenue for the non-thermal and non-lithographic fabrication process without masking (i.e., local electron-beam reduction of GO), particularly, in search of the prospective low-dimensional electronic device schemes such as brain-like low-power memristive and neuromorphic computation systems.

Acknowledgments

This research was supported financially by the Russian Foundation of Basic Research (Nos. 19-29-03050 and 18-29-19120) and the National Research Foundation of Korea (Nos.2016R1A6A1A03012877, 2017R1D1A1B03035102 and 2017R1A2B4004281). The authors thank to Dr. V. Lebedev for AFM measurements and to the Lomonosov Moscow State University for providing some experimental equipments for the research.

Appendix A. Supplementary data

Supplementary material related to this article can be found, in the online version, at doi:https://doi.org/10.1016/j.jmst.2019.07.042.


/