graphene oxide nanocomposites for improving data ...

Journal of

Materials Chemistry C View Article Online

Published on 22 January 2018. Downloaded by National Chung Hsing University on 06/03/2018 14:21:19.

PAPER

Cite this: J. Mater. Chem. C, 2018, 6, 2026

View Journal | View Issue

Oxidized carbon quantum dot–graphene oxide nanocomposites for improving data retention of resistive switching memory† Meng Qi,a Liang Bai,b Haiyang Xu, *a Zhongqiang Wang,*a Zhenhui Kang, Xiaoning Zhao,a Weizhen Liu,a Jiangang Ma a and Yichun Liua

b

Data retention of nano-sized conducting filaments is a critical reliability issue in the pursuit of low-power graphene oxide-based resistive switching (RS) memory devices. Herein, an improvement in the low resistance state retention is demonstrated in fabricated oxidized carbon quantum dot (OCQD)–graphene oxide nanocomposites. Reliable RS characteristics with good retention properties were achieved instead of volatile switching, even with a relatively low compliance current of 100 mA. More epoxy groups were Received 23rd October 2017, Accepted 21st January 2018 DOI: 10.1039/c7tc04829g

introduced as the concentration of embedded OCQDs was increased, resulting in a larger high resistance state, a higher set voltage, and deeper trapping levels. The dependence of the set switching time on the temperature acts as experimental verification that the oxygen migration energy barrier Ea was improved from 0.37 to 0.78 eV after embedding the OCQDs, which explains the enhancement of the low resistance

rsc.li/materials-c

state retention based on a filamentary model.

Introduction Resistive random access memory (RRAM), which is a promising technology for next-generation nonvolatile storage, is attracting growing interest owing to its simple structure, fast switching speed, high-density integration, and process compatibility with CMOS.1–3 Additionally, low power consumption is another essential performance aspect of RRAM for practical applications.4–6 The formation and dissolution of conducting filaments (CFs) across the switching layer widely accounts for the operation mechanism of RRAM devices.1,4,6 To achieve a low power consumption, nano-sized CFs formed under a low compliance current (CC) are generally required to reduce the operation current.7,8 However, the main obstacle is that small-sized CFs suffer from significant reliability issues, e.g., issues with data retention time.8–10 As reported by Ielmini et al., the retention time of CFs proportionally decreases with the filament size. Spontaneous retention loss in nano-sized CFs usually occurs for reasons such as the lateral diffusion of atoms or ions, mechanical stress, or recombination of defects and ions.10–12 a

Key Laboratory for UV Light-Emitting Materials and Technology (Northeast Normal University), Ministry of Education, 5268 Renmin Street, Changchun, China. E-mail: [email protected], [email protected] b Jiangsu Key Laboratory for Carbon-based Functional Materials and Devices, Institute of Functional Nano and Soft Materials (FUNSOM), Soochow University, Suzhou, China † Electronic supplementary information (ESI) available. See DOI: 10.1039/ c7tc04829g

2026 | J. Mater. Chem. C, 2018, 6, 2026--2033

Furthermore, the low defect concentration of CFs and the low activation energy for ion diffusion are considered to be the main intrinsic reasons behind the degradation of the low-resistance state (LRS).8,9,12 Therefore, many efforts have been made to enhance the robustness of LRS retention by increasing the ion migration barrier (Ea) and the ion density in single filaments.8,12 To date, a large variety of materials have been found that display the resistive switching (RS) phenomenon, such as transition metal oxides, chalcogenides, perovskite oxides, ultrathin two-dimensional layered materials, and graphene-based materials.13–16 Graphene oxide (GO) with its unique physical and chemical properties is attracting significant attention for use in RRAM owing to its reliable RS operation and potential for integration with other graphene-based electronics.16–18 In particular, the layered structure of GO could potentially enable transferrable and free-standing RRAM devices, as reported previously.16,18 In GO films hybridization of graphite-like sp2 bonds and diamond-like sp3 bonds occurs, which has been demonstrated using Raman spectroscopy and TEM analyses;16 the stoichiometry of sp2 to sp3 bonds determines the electrical properties of the GO films. Thus, the formation of CFs in pure GO films has generally been attributed to the migration of oxygen functional groups (e.g., carbonyl (CQO), epoxide (C–O–C), and hydroxyl (C–OH) groups) driven by electric fields and the transformation from sp3 to sp2 bonds on the nanoscale.16,19,20 Even though Jeong et al. reported good retention properties of GO RRAM under a small CC of 20 mA,21 a number of literature studies only report on the operation of GO RRAM under a CC

This journal is © The Royal Society of Chemistry 2018

View Article Online


Paper

Journal of Materials Chemistry C

greater than 1 mA.22–24 Thus, as with metal oxide-based RRAM, the volatility of the LRS is a general problem that could impact the reliability of GO-based RRAM when decreasing the CF size (i.e., reducing the CC) to achieve low power consumption.8,9,19 Therefore solutions that improve LRS retention in GO-based RRAM should be considered urgently; however, few studies have focused on this issue to date. There are still almost no effective approaches for adjusting the electrical characteristics to enhance the retention of CFs in GO films, such as increasing the Ea of oxygen migration. The solution may involve adjusting the composition of the oxygen functional groups because these can have diverse Ea values. For instance, the Ea values of C–O–C and C–OH in the GO films are 0.81 and 0.35 eV, respectively.25 Carbon quantum dots (CQDs),26,27 a novel emerging material, possess tunable sp2 and sp3 hybridization depending on the degree of oxidation or reduction, which provide a perfect platform to modulate the Ea for oxygen migration, e.g., via the fabrication of CQD–GO nanocomposites. In this work, oxidized CQD (OCQD)–GO nanocomposites were prepared by embedding OCQDs into GO films; with these nanocomposities, robust LRSs with long retention times (4104 s) could be obtained even under a low CC (100 mA). An increase in the oxygen migration energy barrier Ea was responsible for the improvement in the retention time in the memory cells, which was experimentally verified by investigating the effect of the temperature and the embedded OCQD concentration on the switching time in the memory devices.

using the modified Hummers method. Spin-coating produced GO films with a thickness of 30 nm, while OCQD–GO nanocomposites produced thicker films (60 nm) because of the insertion of the OCQDs (for details, see Fig. S1, ESI†). We also fabricated Al/GO/ITO structures with a GO thickness of 60 nm for comparison, and there was no obvious change in the RS characteristics except for the forming voltages (for details, see Fig. S2, ESI†). OCQDs with a concentration of 0.45 mg ml1 were synthesized using an electrochemical etching method developed in previous work and a subsequent nitric acid treatment to ensure complete surface oxidation.26,27 The concentrations of embedded OCQDs were 4.5, 8.5, 18.5, and 31.5 wt% in our experiments. Finally, the Al top electrodes (TEs) were deposited by thermal evaporation through a shadow mask with a diameter of 400 mm. Furthermore, CQD–GO nanocomposites using unoxidized CQDs (UCQDs) were also fabricated in parallel (for details, see Fig. S3 and S4, ESI†). Electrical measurements were conducted using an Agilent B1500A semiconductor analyzer and a cryogenic probe station (Lake Shore, TTPX) under vacuum. These instruments can characterize the devices in a voltage sweep module for DC measurements; for pulse operation to probe the switching time, we used the Keysight Technologies Waveform Generator/Fast Measurement Unit (WGFMU) module (for details, see Fig. S5, ESI†). We defined the positive voltage as the current flowing from TEs to BEs.

Experimental

The morphologies of GO films and OCQDs were characterized using transmission electron microscopy (TEM), as shown in Fig. 1(a) and (b). The representative TEM images depict the crumpled and layered morphological structures of the GO sheets, while the as-synthesized OCQDs appear to be uniform and monodisperse. An enlarged TEM image was also provided to further help characterize the morphology of the OCQDs, as shown in the inset of Fig. 1(b); examining the diameters of the OCQDs yielded that they were in the range of 2–6 nm (for details, see Fig. S6, ESI†).

Memory cells based on GO films and OCQD–GO nanocomposites were fabricated with Al/GO/indium tin oxide (ITO) and Al/OCQD–GO/ITO structures, respectively. GO solutions (1 mg ml1) with and without embedded OCQDs were spin-coated onto ITO bottom electrodes under ambient conditions. The GO solution was purchased from Nanoon NanoMaterials Technology Co., Ltd China; it had been prepared from natural graphite flakes

Results and discussion

Fig. 1 TEM images of (a) a GO nanosheet, (b) the OCQDs, and (c) the OCQD–GO nanocomposite; the insets of (b) and (c) show enlarged images. Panels (d) and (e) show the XPS C 1s core-level spectrum of the GO film and the OCQDs, respectively. (f) FTIR spectra of the OCQDs and UCQDs. (g) Raman spectra of the OCQD–GO nanocomposites with different OCQD concentrations.


J. Mater. Chem. C, 2018, 6, 2026--2033 | 2027

View Article Online



Fig. 1(c) shows a TEM image of an OCQD–GO nanocomposite. It can be observed that the OCQDs were successfully inserted into the GO nanosheets. X-ray photoelectron spectroscopy (XPS) was performed to investigate the composition of the oxygen functional groups in the GO film and the OCQDs, as shown in Fig. 1(d) and (e). The C 1s spectra acquired from the GO films show three peaks at 285.3, 286.5, and 287.8 eV, which are assigned to the C–C, C–O, and CQO groups, respectively.28 The C–O peak in the XPS spectra, consisting of C–OH and C–O–C species, seems to originate from a small proportion of the GO film, which indicates a limited number of C–O groups. In contrast, a noticeably higher C–O peak intensity can be observed in the OCQDs, as illustrated in Fig. 1(e). This confirms the existence of more C–O groups in the OCQDs than that in GO. As shown in Fig. 1(f), the Fourier transform infrared (FTIR) spectra for UCQDs and OCQDs exhibit an absorption peak related to the –OH group (O–H bonds) at 3400 cm1, a CQO bond peak at 1650 cm1, and a C–O–C bond stretching peak at 1100 cm1.27,29 It can be seen that the amount of C–O–C bonds clearly increases in the OCQDs after the oxidative treatment, while the amount of C–OH bonds decreases. This may be caused by the nitric acid oxidation process, in which the oxygen atoms tend to break the CQC bonds and form sp3 bonds with carbon. More importantly, the FTIR results verify that the abundance of the C–O groups is caused by the presence of C–O–C species rather than C–OH bonds. Furthermore, Raman spectroscopy was applied to identify the structure of the OCQD–GO nanocomposites as the embedded OCQD concentration was increased, as illustrated in Fig. 1(g). Three peaks can be seen for the pure GO film (i.e., 0 wt%) at about 1350, 1580, and 2710 cm1, which are the D, G, and 2D Raman bands, respectively. As the concentration of the embedded OCQDs was increased from 0 to 31.5%, there was no obvious change in the D and G bands. Interestingly, the 2D band intensity weakened and the band finally

Paper

disappeared at high concentrations, as shown in Fig. 1(g). Since the 2D band is related to the two-phonon scattering process, the intensity of this peak reflects the film’s layered structure;30 thus, the disappearance of the 2D band indicates that the layered symmetry of the GO film is affected by the introduction of the OCQDs in the nanocomposites. The RS characteristics and the LRS retention of the Al/GO/ITO memory device without the embedded OCQDs are shown in Fig. 2. In this figure, typical bipolar RS behavior can be obtained under a high CC of 5 mA after the formation process (not shown here). The device could be switched between the high resistance state (HRS) and the LRS by applying a positive or negative bias on the Al TE using a set and reset voltage (Vset and Vreset) of 0.7 and 0.7 V, respectively. However, volatile switching behavior started to appear with a probability of 14.5% when the CC was decreased to 500 mA to reduce the operation current; specifically, the retention loss of the LRS occurred during the voltage sweep back (Fig. 2(a)). Similar phenomena were also reported in RRAM devices based on GO films or other materials,4,19 which was usually attributed to the instability of CFs with relatively small sizes. Fig. 2(b) shows the retention time of the LRS under various CCs; retention failure generally occurred for 104 s when the CC was lower than 1 mA. In addition, as illustrated in Fig. 2(b) and (c), the retention time clearly decreased while the LRS increased as the CC was reduced. This result verifies a close correlation between the retention characteristics and the CF size, as reported in the literature.8,31,32 Importantly, volatile switching can be suppressed by embedding OCQDs into the GO-based RRAM while still obtaining reliable RS characteristics, as illustrated in Fig. 2(d). An Al/OCQD–GO/ITO memory device with an OCQD concentration of 8.5% was fabricated to analyze its characteristics. Compared with the volatile RS in Fig. 2(a), the stable RS transition between the HRS and LRS can be clearly observed under a CC of 500 mA and even

Fig. 2 Panels (a) and (b) show the RS behavior and retention characteristics, respectively, of the Al/GO/ITO device under the different CCs. (c) The dependence of LRS and retention time for various CCs. Measured I–V curves (d) and retention characteristics (e) of the Al/OCQD–GO/ITO devices under different CCs. (f) The evolution of the HRS/LRS with 100 consecutive RS cycles under a CC of 100 mA. The resistance values were read at 0.1 V.

2028 | J. Mater. Chem. C, 2018, 6, 2026--2033


View Article Online


Paper

when the CC was decreased to 100 mA (Fig. 2(d)). It is interesting that a larger Vset was required to set the Al/OCQD–GO/ITO device, which may be related to the embedding effect and will be discussed later. Fig. 2(e) shows good retention of the HRS and LRS with different CCs (5 mA, 1 mA, 500 mA, and 100 mA) in the Al/OCQD–GO/ITO device, even at 85 1C. Although smaller CF sizes are expected when decreasing the CC, all those resistance states were retained for 104 s without any degradation, indicating their good nonvolatility. Furthermore, reproducible read/write characteristics with an acceptable ratio of B50 were achieved under a CC of 100 mA by recording 100 consecutive RS cycles for the Al/OCQD–GO/ITO device, as shown in Fig. 2(f). There was no error except for the visible fluctuation during switching processes; thus, such a device demonstrates the feasibility of low-power consumption RRAM. To summarize, all the above-mentioned results experimentally demonstrated than an improved LRS robustness can be achieved by embedding OCQDs in GO-based RRAM. However, the main role of these OCQDs in improving LRS retention has to be investigated in depth. To better understand the impact of OCQDs on the RS characteristics, Al/OCQD–GO/ITO devices with various OCQD concentrations (4.5, 8.5, 18.5, and 31.5 wt%) were fabricated and compared. As shown in Fig. 3(a) and (b), the HRS clearly increased from 2 104 to 2 106 O when increasing OCQD concentrations from 0 to 31.5%, while the LRS remained constant at B500 O. Conversely, the HRS decreased when the concentration of UCQDs was increased, which narrowed the switching window (for details, see Fig. S3, ESI†). This is the main reason why the UCQDs were not selected for the current work. The increasing trend can also be found for Vset ranging from 0.7 to 2.8 V. Although a larger HRS is beneficial for enlarging the switching window, an overlarge Vset augments the switching power and reduces the stability of the devices. Thus, a moderate OCQD concentration of 8.5% was chosen for the

Fig. 3 Panels (a) and (b) show the dependence of resistance states and Vset, respectively, for an increasing OCQD concentration in the Al/OCQD– GO/ITO device. (c) Typical I–V curves for the HRS and LRS of the memory devices with and without embedded OCQDs on a log–log scale. (d) Correlation between the fitting STFL at high voltages and the concentration of embedded OCQDs.



Al/OCQD–GO/ITO memory cells in this study. The concentration of OCQDs can affect the conduction mechanism of the Al/OCQD–GO/ITO devices, as shown in Fig. 3(c) and (d). Fig. 3(c) illustrates a log–log plot of the I–V curves of devices with and without embedded OCQDs. The LRSs of both these two devices showed a similar Ohmic behavior, with an I–V slope of B1, while the HRSs showed a dependence on the OCQD concentration. For the Al/GO/ITO device, the HRS conduction can be divided into two regions: ohmic conduction in the low voltage region (o0.3 V) and almost linear conduction in the high voltage region (40.3 V) with a slope of B1.3, which can be approximately linked to Ohm’s law and trap filled-limited (TFL) behavior (I p Vn) in a space charge-limited conduction model.33 Note that a weak TFL behavior is probably present in this case, since the I–V slope (STFL) of B1.3 is obviously smaller than that of STFL Z 2, which indicates a strong TFL behavior.34,35 According to previous literature results, the small STFL may be related to the shallow trapping level in the GO film.36 Interestingly, for the devices with embedded OCQDs, the STFL increased and saturated between B1.3 and B2.3, when the OCQD concentration increased to B10%, as shown in Fig. 3(c) and (d). This type of behavior is in agreement with trap-filling, which means trapping centers with deeper levels were induced by the OCQDs.36,37 In summary, the impact of OCQDs embedded in GO-based RRAM devices is an increase in HRS, Vset, and STFL. Experimental data and theoretical calculations imply that charges are trapped in the discrete graphene (sp2 bond) region surrounded by dielectric oxygen groups (e.g., epoxy and hydroxyl groups).38–40 Since the OCQDs have similar chemical structures and charge trapping effects, embedding them in GO can strongly impact the electrical properties of previously pristine GO films: (i) embedding them introduced more epoxy groups (Fig. 1(f)), which increases the migration energy barrier Ea compared with the hydroxyl group-dominated GO films;25 and (ii) it can lead to a higher degree of oxidation around the graphene trapping centers, thus lowering the conductivity of the film and deepening the charge trapping level.37 Thus, the above analysis can account for the increased band gap for a higher HRS, an increase in STFL, and a larger Vset. The increase in Ea in particular may be directly related to the improvement in the LRS retention; we will present an experimental verification of this later in this paper. Before explaining how the OCQDs affected the LRS retention, it is necessary to clarify the switching mechanism of OCQD–GO nanocomposite-based RRAM devices. The local conductivity distribution of an Al/OCQD–GO/ITO cell was investigated using conductive-atomic force microscope (C-AFM) to probe its switching mechanism, as shown in Fig. 4(a) and (b). The scanning area was 1 1 mm2 and the current distribution topography was measured at a read voltage of 10 mV after set and reset processes. In the set process, a bias voltage of 5 V was applied on the Al top electrode through a conductive diamond AFM tip, setting the device to an LRS. Then, the top electrode was removed by using the diamond tip as a scalpel to physical scrape the sample.41 It can be observed in Fig. 4(a) that the LRS presented a highly conductive path in a local region with a diameter of B200 nm. For the reset process, the diamond AFM

J. Mater. Chem. C, 2018, 6, 2026--2033 | 2029

View Article Online



Fig. 4 The local conductivity distribution of the Al/OCQD–GO/ITO cell in (a) the LRS and (b) HRS as measured by C-AFM. Photographs of the Al/OCQD–GO/Al device with a planar structure (c) before and (d) after the forming process. (e) The I–V curve of the Al/OCQD–GO/Al device with a planar structure during the forming process. (f) Raman spectra of an as-prepared cell and of the one undergoing the forming process.

tip scanned the 1 1 mm2 area with a bias of 5 V, which resulted in the transition from the LRS to the HRS. The HRS generally showed a very low conductance under negative voltages as seen in Fig. 4(b). Thus, C-AFM measurements could be used to verify the conducting filamentary mechanism of our devices. However, the nature of CFs in GO-based RRAM devices is still controversial. Two main models have been proposed to explain the role of CFs in switching. One theory is that the CFs are involved in the redox reaction of the electrodes (e.g., Al) at the interface between GO and metals.21 Another model proposes that the CFs inside the GO film transform from insulating sp3 to conducting sp2 bonds based on the detachment of oxygen groups under high electric fields.16,20 To experimentally investigate the CF composition, an Al/OCQD–GO/Al device with a planar structure was fabricated and studied, as illustrated in Fig. 4(c) and (d). Fig. 4(e) shows that a forming process was initially employed to switch the device from the HRS to the LRS under a very high bias (B17 V), indicating the formation of CFs. Although it is impossible to obtain repeatable RS results during this operation owing to the permanent destruction of the CPs, it can provide distinguishable HRSs and LRSs (i.e., without and with CFs) for chemical composition analysis. Fig. 4(c) and (d) show the photographs of the plane device before and after the forming process, revealing an obvious change in the region of interest (black square) between these two electrodes. Micro-Raman spectroscopy using a 488 nm laser was applied to identify the a-C structure in the square region of Fig. 4(d). As shown in Fig. 4(f), the Raman spectra of the as-deposited films and after forming show broad D and G bands. More importantly, the density ratio (ID/IG) also clearly increased from 0.96 to 1.13 after forming, validating the increase in the amount of sp2 bonds during the CF formation.16 In other words, the formation and rupture of the sp2 bond-based CFs are responsible for the switching mechanism of Al/OCQD–GO/ITO devices.

2030 | J. Mater. Chem. C, 2018, 6, 2026--2033

Paper

Fig. 5 (a) Typical response profiles of the tset measurement. (b) Values of tset as a function of temperature and (c) the migration energy barrier Ea for the Al/GO/ITO and Al/OCQD–GO/ITO devices. In panel (d), the Arrhenius fitting plot shows that the Ea was increased for the set processes after embedding of the OCQDs in the GO.

Similar to the results in previous studies,8,9 the Ea increase induced by embedding OCQDs in GO may be the main reason that the LRS robustness is increased. Thus, we verified the specific values of Ea in this study, as shown in Fig. 5. Fig. 5(a) shows an example of the measured set switching time (tset) for the RRAM devices,42 in which the set process occurred through applying a square pulse (2.5 V/4 ms). The delay time between the input and output signals is tset, as illustrated in Fig. 5(a). The correlation between tset and the temperature was further investigated to calculate the migration energy barrier Ea in the GO and OCQD–GO nanocomposite-based RRAM devices. According to the thermally activated process of the filamentary model, the local temperature TA can be estimated by solving the Fourier heat flow equation,43 TA ¼ T0 þ Rth

V2 R

(1)

where T0 is the ambient temperature and Rth is the equivalent thermal resistance. The sample was measured as T0 was increased from 300 to 360 K. As shown in Fig. 5(b), both the tset of the GO and OCQD–GO nanocomposite-based RRAM devices decreased with increasing T0. The value of tset generally follows the temperature-accelerated Arrhenius law,44 Ea

tset ¼ t0 ekB TA

(2)

where t0 is a pre-exponential time constant, Ea is the activation energy for the physical set process, and kB is the Boltzmann constant. Thus, Ea can be extracted from the experimental data in Fig. 5(c) through fitting eqn (1) and (2), which yields Ea values of 0.37 and 0.71 eV for the GO and OCQD–GO nanocompositebased RRAM, respectively. The Ea values of the devices were measured and compared for all OCQD concentrations, as shown in Fig. 5(d), which illustrate that Ea clearly increases with increasing OCQD concentration (for details, see Fig. S7, ESI†).


View Article Online


Paper

Fig. 6 A schematic diagram of the oxygen functional group migration barrier Ea for (a) the GO film and (b) the OCQDs.

This result confirms the increase in Ea after embedding the OCQDs, which may be attributed to the introduction of more C–O–C groups. Fig. 6 schematically illustrates a comparison of the oxygen functional groups’ Ea in a GO film and in OCQDs. Generally, the migratable oxygen functional groups are C–OH and C–O–C groups rather than CQO groups since they have a much smaller energy density,45,46 with Ea values of 0.35 and 0.81 eV, respectively.25 The OCQDs contain a larger proportion of C–O–C groups than the GO film. Thus, a higher Ea can be expected for the OCQD–GO nanocomposites than for the GO film, in agreement with the results shown in Fig. 5(c). The above argument is also supported by the data in Fig. 3(b), where the Vset increases along with the OCQD concentration. A potential mechanism underpinning the data retention improvement in the OCQD–GO nanocomposites is explained in the following and is shown in Fig. 7. For the GO film, small


sp2 bond-based CFs (i.e., oxygen vacancies) are formed under a low CC (Fig. 7(a) and (b)). However, thermal perturbation can result in spontaneous diffusion or partial recombination of the oxygen vacancies and mobile oxygen ions, which usually results in the LRS being degraded or even in retention failure (Fig. 7(c) and (d)). According to a Monte Carlo simulation this recombination rate is determined by the oxygen migration barrier,9 and the retention time depends on Ea according to the Meyer–Neldel effect.8,47 Thus, a higher energy barrier is regarded as an effective method to improve LRS retention.8,9 Here, the proportion of C–O–C groups was enhanced in GO films by embedding OCQDs into them, leading to a higher migration barrier Ea and a larger Vset (Fig. 7(e) and (f)). The increase of Ea can help decrease the CF’s relaxation by decreasing the recombination effect (Fig. 7(g) and (h)), which results in an improvement in the data retention. All these results point to the rise in Ea in our samples being the result of the OCQDs embedded in the GO, which increases the LRS robustness in the GO-based RRAM devices.

Conclusions In summary, the volatile switching of small CFs was suppressed by embedding OCQDs into GO-based RRAM devices. The OCQD–GO nanocomposite-based memory devices exhibited stable bipolar switching and outstanding retention characteristics, which could potentially be applied in the field of nonvolatile memory devices. The switching mechanism in GO was found to be governed by the migration of oxygen functional groups in the switching layer of the OCQD–GO nanocomposite. By investigating the impact of the OCQD concentration and by studying the temperature dependence of tset, the enhancement of LRS retention could be attributed to an increase in the oxygen migration barrier Ea. Our findings provide one feasible approach to

Fig. 7 Panels (a)–(d) and (e)–(h) show schematic sketches of the retention characteristics for the Al/GO/ITO and Al/OCQD–GO/ITO devices, respectively.


J. Mater. Chem. C, 2018, 6, 2026--2033 | 2031

View Article Online


improve LRS retention and could be important for the design of future GO-based RRAM memory devices for low-power consumption applications.


Conflicts of interest There are no conflicts to declare.

Acknowledgements This work is supported by the NSFC for Excellent Young Scholars (No. 51422201), the NSFC Program (No. 51701037, 51732003, 51372035, 61774031, 61604037 and 61574031), the ‘‘111’’ Project (No. B13013), the Fund from Jilin Province (No. 20160101324JC and 20160520115JH) and the Postdoctoral Science Foundation (No. 2017M621189).

Notes and references 1 R. Waser and M. Aono, Nanoionics-Based Resistive Switching Memories, Nat. Mater., 2007, 6(11), 833–840. 2 A. C. Neto, F. Guinea, N. M. Peres, K. S. Novoselov and A. K. Geim, The electronic properties of graphene, Rev. Mod. Phys., 2009, 81(1), 109. 3 Y. Chen, B. Zhang, G. Liu, X. Zhuang and E. T. Kang, Graphene and its derivatives: switching ON and OFF, Chem. Soc. Rev., 2012, 41(13), 4688–4707. 4 S. G. Park, M. K. Yang, H. Ju, D. J. Seong, J. M. Lee and E. Kim, et al., A non-linear ReRAM cell with sub-1 mA ultralow operating current for high density vertical resistive memory (VRRAM), IEEE Electron Device Lett., 2012, 20(8), 1–20.8.4. 5 Y. Wu, S. Yu, B. Lee and P. Wong, Low-power TiN/Al2O3/Pt resistive switching device with sub-20 mA switching current and gradual resistance modulation, J. Appl. Phys., 2011, 110(9), 094104. 6 S. Gaba, F. Cai, J. Zhou and W. D. Lu, Ultralow sub-1-nA operating current resistive memory with intrinsic non-linear characteristics, IEEE Electron Device Lett., 2014, 35(12), 1239–1241. 7 J. H. Yoon, K. M. Kim, S. J. Song, J. Y. Seok, K. J. Yoon and D. E. Kwon, et al., Pt/Ta2O5/HfO2x/Ti resistive switching memory competing with multilevel NAND flash, Adv. Mater., 2015, 27(25), 3811–3816. 8 D. Ielmini, F. Nardi, C. Cagli and A. L. Lacaita, Size-dependent retention time in NiO-based resistive-switching memories, IEEE Electron Device Lett., 2010, 31(4), 353–355. 9 S. Yu, Y. Y. Chen, X. Guan, H. S. Philip Wong and A. K. Jorge, A Monte Carlo study of the low resistance state retention of HfOx based resistive switching memory, Appl. Phys. Lett., 2012, 100(4), 043507. 10 S. Choi, J. Lee, S. Kim and W. D. Lu, Retention failure analysis of metal-oxide based resistive memory, Appl. Phys. Lett., 2014, 105(11), 113510.

2032 | J. Mater. Chem. C, 2018, 6, 2026--2033

Paper

11 S. Ambrogio, S. Balatti, S. Choi and D. Ielmini, Impact of the mechanical stress on switching characteristics of electrochemical resistive memory, Adv. Mater., 2014, 26(23), 3885–3892. 12 T. Ninomiya, Z. Wei, S. Muraoka, R. Yasuhara, K. Katayama and T. Takagi, Conductive Filament Scaling of TaOx Bipolar ReRAM for Improving Data Retention Under Low Operation Current, IEEE Trans. Electron Devices, 2013, 60(4), 1384–1389. 13 A. A. Bessonov, M. N. Kirikova, D. I. Petukhov, M. Allen, ¨nen and M. J. Bailey, Layered memristive and T. Ryha memcapacitive switches for printable electronics, Nat. Mater., 2015, 14(2), 199. 14 C. Gu and J. S. Lee, Flexible hybrid organic-inorganic perovskite memory, ACS Nano, 2016, 10(5), 5413–5418. 15 Y. Li, S. Long, Q. Liu, H. Lv and M. Liu, Resistive Switching Performance Improvement via Modulating Nanoscale Conductive Filament, Involving the Application of TwoDimensional Layered Materials, Small, 2017, 13(35), 1604306. 16 S. K. Kim, J. Y. Kim, B. C. Jang, M. S. Cho, S. Y. Choi and J. Y. Lee, and Jeong, Conductive Graphitic Channel in Graphene Oxide-Based Memristive Devices, Adv. Funct. Mater., 2016, 26(41), 7406–7414. 17 D. A. Dikin, S. Stankovich, E. J. Zimney, R. D. Piner, G. H. Dommett, G. Evmenenko, R. S. Ruoff and D. A. Dikin, Preparation and characterization of graphene oxide paper, Nature, 2007, 448(7152), 457. 18 G. Eda, G. Fanchini and M. Chhowalla, Large-area ultrathin films of reduced graphene oxide as a transparent and flexible electronic material, Nat. Nanotechnol., 2008, 3(5), 270–274. 19 S. Pinto, R. Krishna, C. Dias, G. Pimentel, G. N. P. Oliveira and J. M. Teixeira, et al., Resistive switching and activitydependent modifications in Ni-doped graphene oxide thin films, Appl. Phys. Lett., 2012, 101(6), 063104. 20 L. H. Wang, W. Yang, Q. Q. Sun, P. Zhou, H. L. Lu and S. J. Ding, et al., The mechanism of the asymmetric SET and RESET speed of graphene oxide based flexible resistive switching memories, Appl. Phys. Lett., 2012, 100(6), 063509. 21 H. Y. Jeong, J. Y. Kim, J. W. Kim, J. O. Hwang, J. E. Kim and J. Y. Lee, et al., Graphene oxide thin films for flexible nonvolatile memory applications, Nano Lett., 2010, 10(11), 4381–4386. 22 D. Y. Yun and T. W. Kim, Nonvolatile memory devices based on Au/graphene oxide nanocomposites with bilateral multilevel characteristics, Carbon, 2015, 88, 26–32. 23 X. D. Zhuang, Y. Chen, G. Liu, P. P. Li, C. X. Zhu and Y. X. Li, et al., Conjugated-polymer-functionalized graphene oxide: Synthesis and nonvolatile rewritable memory effect, Adv. Mater., 2010, 22(15), 1731–1735. 24 F. Zhuge, B. Hu, C. He, X. Zhou, Z. Liu and R. W. Li, Mechanism of nonvolatile resistive switching in graphene oxide thin films, Carbon, 2011, 49(12), 3796–3802. 25 S. Kim, S. Zhou, Y. Hu, M. Acik, Y. J. Chaba, C. Berger and E. Riedo, Room-temperature metastability of multilayer graphene oxide films, Nat. Mater., 2012, 11(6), 544–549.


View Article Online


Paper

26 C. Zhu, Y. Fu, C. Li, Y. Liu, L. Hu, J. Liu and Z. Kang, et al., Carbon Dots as Fillers Inducing Healing/Self-Healing and Anticorrosion Properties in Polymers, Adv. Mater., 2017, 29(32), 1701399. 27 H. Li, Z. Kang, Y. Liu and S. T. Lee, Carbon nanodots: synthesis, properties and applications, J. Mater. Chem., 2012, 22(46), 24230–24253. 28 K. Qu, J. Wang, J. Ren and X. Qu, Carbon Dots Prepared by Hydrothermal Treatment of Dopamine as an Effective Fluorescent Sensing Platform for the Label-Free Detection of Iron (III) Ions and Dopamine, Chem. – Eur. J., 2013, 19(22), 7243–7249. 29 Z. Yang, M. Xu, Y. Liu, F. He, F. Gao and Y. Su, et al., Nitrogen-doped, carbon-rich, highly photoluminescent carbon dots from ammonium citrate, Nanoscale, 2014, 6(3), 1890–1895. 30 S. Sahoo, G. Khurana, S. K. Barik, S. Dussan, D. Barrionuevo and R. S. Katiyar, In situ Raman studies of electrically reduced graphene oxide and its field-emission properties, J. Phys. Chem. C, 2013, 117(10), 5485–5491. 31 X. Zhao, H. Xu, Z. Wang, L. Zhang, J. Ma and Y. Liu, Nonvolatile/volatile behaviors and quantized conductance observed in resistive switching memory based on amorphous carbon, Carbon, 2015, 91, 38–44. 32 Y. Bai, H. Wu, R. Wu, Y. Zhang, N. Deng and Z. Yu, et al., Study of multi-level characteristics for 3D vertical resistive switching memory, Sci. Rep., 2014, 4, 5780. 33 R. L. Weisfield, Space-charge-limited currents: Refinements in analysis and applications to a-Si1xGex: H alloys, J. Appl. Phys., 1983, 54(11), 6401–6416. 34 J. Won Seo, S. J. Baik, S. J. Kang, Y. H. Hong, J. H. Yang and L. Fang, et al., Evidence of Al induced conducting filament formation in Al/amorphous silicon/Al resistive switching memory device, Appl. Phys. Lett., 2010, 96(5), 053504. 35 D. I. Son, T. W. Kim, J. H. Shim, J. H. Jung, D. U. Lee and J. M. Lee, et al., Flexible organic bistable devices based on graphene embedded in an insulating poly (methyl methacrylate) polymer layer, Nano Lett., 2010, 10(7), 2441–2447. 36 M. Yi, Y. Cao, H. Ling, Z. Du, L. Wang and T. Yang, et al., Temperature dependence of resistive switching behaviors in resistive random access memory based on graphene oxide film, Nanotechnology, 2014, 25(18), 185202.



37 Y. Shen, S. Yang, P. Zhou, Q. Sun, P. Wang and L. Wan, et al., Evolution of the band-gap and optical properties of graphene oxide with controllable reduction level, Carbon, 2013, 62, 157–164. 38 I. Y. Lee, E. S. Kannan and G. H. Kim, Capacitance-voltage and current-voltage characteristics of graphite oxide thin films patterned by ultraviolet photolithography, Appl. Phys. Lett., 2009, 95(26), 343. 39 S. Wang, J. Pu, D. S. Chan, B. J. Cho and K. P. Loh, Wide memory window in graphene oxide charge storage nodes, Appl. Phys. Lett., 2010, 96(14), 143109. 40 A. D. Yu, C. L. Liu and W. C. Chen, Supramolecular block copolymers: graphene oxide composites for memory device applications, Chem. Commun., 2012, 48(3), 383–385. 41 K. X. Shi, H. Y. Xu, Z. Q. Wang, X. N. Zhao, W. Z. Liu and Y. C. Liu, et al., Improved performance of Ta2O5x resistive switching memory by Gd-doping: Ultralow power operation, good data retention, and multilevel storage, Appl. Phys. Lett., 2017, 111(22), 223505. 42 Z. Q. Wang, H. Y. Xu, L. Zhang, X. H. Li, J. G. Ma and X. T. Zhang, et al., Performance improvement of resistive switching memory achieved by enhancing local-electricfield near electromigrated Ag-nanoclusters, Nanoscale, 2013, 5(10), 4490–4494. 43 F. Nardi, S. Balatti, S. Larentis, D. C. Gilmer and D. Ielmini, Complementary switching in oxide-based bipolar resistiveswitching random memory, IEEE Trans. Electron Devices, 2013, 60(1), 70–77. 44 M. Boniardi and D. Ielmini, Physical origin of the resistance drift exponent in amorphous phase change materials, Appl. Phys. Lett., 2011, 98(24), 243506. 45 C. Arcoumanis, C. Bae, R. Crookes and E. Kinoshita, The potential of di-methyl ether (DME) as an alternative fuel for compression-ignition engines: A review, Fuel, 2008, 87(7), 1014–1030. 46 R. Sanderson, Chemical bonds and bonds energy, Elsevier, 2012. 47 D. Lelmini and M. Boniardi, Common signature of manybody thermal excitation in structural relaxation and crystallization of chalcogenide glasses, Appl. Phys. Lett., 2009, 94(9), 091906.

J. Mater. Chem. C, 2018, 6, 2026--2033 | 2033